From 52e14895a12a4e5dd013a9be11e78b1c0054ba26 Mon Sep 17 00:00:00 2001
From: Tomasz Kalinowski <kalinowskit@gmail.com>
Date: Thu, 25 Sep 2025 11:26:30 -0400
Subject: [PATCH 01/81] redocument

---
 DESCRIPTION                                  |   2 +-
 man/figures/op-gelu-plot-1.png               | Bin 6996 -> 16638 bytes
 man/figures/op-hard-sigmoid-plot-1.png       | Bin 7293 -> 17921 bytes
 man/figures/op-leaky-relu-plot-1.png         | Bin 7242 -> 17394 bytes
 man/figures/op-relu-plot-1.png               | Bin 8250 -> 18241 bytes
 man/figures/op-relu6-plot-1.png              | Bin 8073 -> 17089 bytes
 man/figures/op-softplus-plot-1.png           | Bin 9302 -> 21578 bytes
 man/figures/op-softsign-plot-1.png           | Bin 9079 -> 20855 bytes
 man/layer_tfsm.Rd                            |   4 ++--
 man/metric_mean_absolute_percentage_error.Rd |   4 ++--
 man/multi-assign.Rd                          |   2 +-
 man/op_erf.Rd                                |   2 +-
 man/op_gelu.Rd                               |   2 +-
 man/plot.keras.src.models.model.Model.Rd     |  12 +++++++++---
 14 files changed, 17 insertions(+), 11 deletions(-)

diff --git a/DESCRIPTION b/DESCRIPTION
index 5fc222534..d08d661f9 100644
--- a/DESCRIPTION
+++ b/DESCRIPTION
@@ -53,5 +53,5 @@ Suggests:
     rstudioapi,
     R6,
     jpeg
-RoxygenNote: 7.3.2
+RoxygenNote: 7.3.3
 VignetteBuilder: knitr
diff --git a/man/figures/op-gelu-plot-1.png b/man/figures/op-gelu-plot-1.png
index d34697042776579594b2859afa21cfe37cfd6331..45e2d00b7b68d6939555ecc9f0231d5f5f14a0e5 100644
GIT binary patch
literal 16638
zcmeIaXH-+$7d?u4)vKbYfC?z6fb^m$0ck2tI#Q*J^cGs^O+^rtB1jKKq=^V92uKhR
zkxs<WJ18|&=^fq*LGk|nZ;bcpy-)9q;kcTU?6c3_tIRdm+}B@GUh?Q+>cb=?BuAxg
ziz$(i>{%xK`FAgT!Xj<T4F5S~b6dlngyhIc!k;~jJ)Y|%B<D$_#BQoM$4>T`Xsfif
z9{42`a9-}_iQ_NszALzkYkyT4fuIhk6bn>0D7&i^_G}N!Me#lU?H2Wc9F@IUmF?H4
z1369@C~!YIN|}6YA8CF4+C%LmrQ`|wZ}qwjTcV!(zVKV5YfR0(YoH3pnHs|LIFF<p
zfw`FJf8R&)?=f#$k{dQ>=1E8%mA^bla{TN8A(99E+@pI){&{kfjO08YsVfOd+5i7l
zRVB=W94UEuVaJK6MpY6L8)JR_A9%c(nHft(8P@VUoscVPX{x8E$VnY@oP;F4HB+k%
z<1oIuFfL<j!$UJ4@Sal#<33x%B&~jjJy62Hz#vI2Bb-xA6OYFqJ9bQ3{R01CtFeXE
zi3v+LH@B1{A!_pS#|-K_J2mmQo;pl*^W%>N1;q1Ys%5l|hiBcgi;j+#jy!*m@Ece7
z5Jef<ghySElgG1C7<jU|l8^lS`3iN&&Uv9lJ)HCAF^bL)IXSO>Mdq8gypfK4$?$fR
zzsd9}h1xtk{?9mp5=k&lT}=(kZEMpnFGommfw!4I_>cPT6+^_{`FPm)>XoAsNpfNc
zihJS>RIg}46ivx-NB5tZkG+67ama`FM33TwQD2_J>SX~Esb!|yzk~;gqUcW^wz`N(
zyfJ&z`lH(*I&efeFyL~8{o0dXUd1h08ur1x^rB81EQ3QA{|p*R=IYV?&99>B#nAwx
z1k&YF+qE}Gk7^Z~PYt%V?OXN85vpNcz#GA9-ZwHzfoUf_H{+(ZYV=1kry*v=5i6zB
z{EAj%WBbWhoyNO8-Zd2HE$|i3*htLic`g6?ama`L?z{fMnisCc!U)7tnfJj<%NMU*
zTiD!Kdm4FR;aPLDiZZk4*y)*>nN&sJWG>Wo9dq+HKaFm0{QMRtpqZ)m?S=W(m=u}F
zh|Qv8xrl(1P3JZ~=%$T|%m%I?USSqB&tz|{&2>Dv#Czt0aNe@DtdvxtV)d^d<p+;a
z8(yV8dO+0CG5=7B^YSl!2?^irai@BU;8D$7KL-Zn5=o`>28V}*-Io<-t|WVg#Ky+P
z$8(NI8Qi}=j76)jPE&)qX=}&cp=M3G)49wr{WZ+j+FB8T?0km|mrTpLs^hv?M5g^h
zo@)+G5x!m#y%pq^C)-BR+uQ5!?~mrS3VowZ^5C$HloS=8O+PM4QfABDg^P>J%gf8&
zL-zd;`iM2-!Jdur?9$EEakZtc4%LN#Aw2=-X4mfF7K2*eA?#jSWp>!rv6-m*hyH<U
z|9+`?&Wl6mtAt(iGVOa#eRI%4xsQ7#;Hh$w(ny)q9;s#I<>Z|7{BB}wd{|@;RUq$z
zYq44erRVq4mW3fE-}Q=JukrI2>Xmw4iOmSJDW@=@;N|9~?z(d(kQYCJ37ORP^z@t?
z#=e+DrCiW;e8t_(%}98@Z8?Q{SH6`+?<bcei9ka>b-eyxwkgTa&wu&y<%7r0KA&eF
zdT_IW$|vKV1bx%JtDRz~@;kD!^~*B@a=*T@gMt6H;a{_`O`=bZ1=0#>Hs*$r9S|~c
zZY4NEdQ~oxZ)R@p$dMzt@3<=xpZ<@nKhyKtP|a>DT^UA)#;0`4&d|~2`i@F4P{6a4
zZ6%UySLgA$CJhmz%al)GCOf7sS7FuuE>R|u(_ySRgmCM>qozK}$;pY})NviGKH(qK
zuwQN0*;Et_4F|(mlr62SQuz3|{&GW=uW<rS-(vZBP@{@dlEX)EM@=ZQ9#I9#dgH)j
z<;0hlm&GI`-l(v$9Q$_SuMxw#8br1x`uqET{3u&`r}TL@1f;spyB3$0dO17uRT)Sh
z{GS0=+~=@}>5{c9G@7QbT>2k2*Kfyr{2t#z`yp#DaBx^HjI}7I%KmV;k*@#U{?C#e
zOH54Mvu95@oBA;E&hY=~%+4tY1hV_1?SOAu8dGFa%|7pV9x=ql#P640p>t*z76OBU
zhV7U;?)^3FrM1P$46OoWT?7Z4Co{i(+g~sLLLQ|&<2^VugpIjo(GV|0DH2xy?$w__
zmpTtCy^52h1GBMZk&^#wNyC(slo}&Av@Xd;ea>LuzkvA>@I{q7DTEMoE@D_xsCbO)
zWBHpJ8$)G;u9QFBon_-k6kf^B&aTvRgD--CiF9{qq|4jd5Ld2*8VFqxMV0?;nR7}o
zf2*ccxpLp5{H^pn@T05Lfp9_K!5lYBNlO#p<73wG83dzj8_&(not&H`<dXe`fq8ij
z{3W*F4FsEi$-HtN^W%3IPc;Nr7^n)4oZ>JUdPphsADcKd)mQ4(|M?-@pW~E7-o@Gb
z>TIvGV^7%UzCN$TiH=jJj76BRC3}>;e;aiykIA<)q`t@=jFd`f?C~BOi65*{xazCm
zvnP5eYD}-LuB~Nj<VwWY4f?|K{E%X|5+QwvYnFO?1k(zQHK)_-NZI|BX;QfZC2}o0
za?F}ir8NJ~5l>N3;U_1pOiY+`{=3MhyGuOQ>Z5t+JeL!j8EX#xWhJMVS63y(#nV+-
z(`7=*U(tMH*51!U9ep4P2PR~4PQbCgtgNiOTy)IU@b5V);1^cG5q|x$agh!shZyjC
z@(P)nd9_VVO<1g4eCXig|MAuvnwp)E_Jhs|TvGV&6!(I)H>D|LYZvBN{%3|XmF<Bv
z1S1DH!D1q1ck92lP=01^+sq>}&#e3v+OT9I>x$XB#bA=P+M%)Bztil=VC3xVtZ_s9
zxwB{4z5Zu32HGBWjU2<TekbJR<*&5S^Zd!qN<qlb(9nGAK4JYY|A_zn#<e&(MMXs$
z8=DmAp9g7vIx+9&cPU<se_)`Ty!=#GLH5OkYSKS`Z<vIBl%~R}8eIQpVvPaFd}L2&
zmQIoHQ5x2}|FgM=BoI5Hnv7QE9c}cVH+^vr#p5*9eTe*07#_PHd+yIpnB%Y;YfEIb
zg3B{ibZj~yZ~tR{>EZ}j>;xgVe4qW9*ElngMh~`)mVvS-mfyb8^s3`{s~p~LF!l3G
z1jE_8`3rVg>r4^*w_`E{i5F<}`1G`h+oENP#hnN9xWLQfQ$rbKIe`iyAzR~!6KFId
zD>HLUG24Osv`h`#bILzyC7HbEtxi$CNkcru^E;SR`>VM}={ER^w+8p|b1vvPDDqUE
zX%d!}77}uqHy%ekBJ&XF7H4!hWY=hVm6w;-JP+_*a(I0FRDb_V0;>6*s+-;L?OT39
z0pLnAMkXdMjhsP9TGFRqk>ZnRjw~rS>a+_LztHNriCep$tyj9a789)N#x^#wFdnHE
zd3C73y016wIoNr3k&{Ap+sNSH*QzRX|FeLAfX9#jsi=FtDh>#QB@95%*RNs#dKyn(
zkK?4L7bO6X6KBr~nl&=yX<A)epzyq4_8|2Nbs*Ok>#_M@&lrKSwbE;8S+N^qPlKLa
z*KNPn_<lliW^qwoR<`?t8Tr=UJvS8L>5H?o*WH)jNgHAIJc^8rge0=Dx{$*6(NZKp
zwa9U)%f(PXL)4&mRygWXg9xYe=PUxQR93jx7MPuFefRENWY%k~8@lqX9`kR^Cz&!u
zdhPpC6Yd_=;#Xj&&s4K8GGcU>NQjPRad*a=62f7gX69$AjrRssXV0C}%C}5!!oSVT
z4Cj=jn(gB@Y8#BRl`|B{uw=HpckgO`5=8H(#p4p$Z3hn?wC>39Y2HV6xcsQ{4j8Bi
zLaJqI4gp3U9*%hN;)QZO+FO}`B9Kuh0q-vS?Bx!Rhyb(3uqda#``LFk_FutqU!I5A
zj}pW$5I_b6ALh`JBWPg{On5jY+a|Wm|MTzEkop>j1;5YIE1kmQHGW-qMbm$18*j6i
zqXx#uY0jP7m=_g8e17p-y^QPSHn=@Z)iVK}29M*=%6DI-`iVQb3r<c?X&_u&Tyl-E
zod7a&6WFV*s;819A((6%weqJ?<C&io^b!F!c6P3dFZ_@Xm}asbJrg6&lN}=^E!~-K
z)ukY^G5=-6UlQ>r1Ek(7LK@dcUA7;Vm8m=3H?p?~e!885Q*Yz2Eg9;tFl!6r>dTSG
zdq@;6>g<nvaEpNcS(H!%BO?gw-cJ|34&jdedc5s`6qhg^`Bqwh<Y=y$s|;pe#O#SD
zVukx+h{D3cxse94fB$8kIzkxii}wD=0ODX0+hM?VpfcR1Y2WTz-9XLtmw9_1BnzSy
z!aUBtiP|Sf#Buxmkn0;8<>loZ?9wsDgty(3^+?8hdoZ&p92SsmOhUreFxom|-H33T
z_(mF%2#F+IU!Taui$7EJ{6`Ns-`cKTVjNJp`sKcW{oB%&M~HW>jA!&W63dPlYgAH;
zc_f#f=db><M)%8qCTLcIlT%S?PD}7KH!)F<;ga|>LF`qG)a~1R5G4R^%B(C9=D<^Y
zn3L@=AvdK)OCSL7hJ=K$hQA5Z3*McG0m^=?nMF&h>)5&LVVhcnWg@1g+-FGA1EPwG
z-`{HF*3Rn?#<Tx)nDZcUybf_hpXZttpb)44Ra2*L?M{RZQ&V5BUF^EhoUUBA8o0X{
zF5-wEKYj=c3%9nm8f{1r&5e_jQ%y~6yI$R%D3v&(04&mdWfnG~Zl+i@LoFq?_YWYu
znlx(M`N8bYojXHox?s+bfx*3-v|pi<B!gRFs_^}xHAfiRvB$=kQ@xCmXpvW+je=J`
zdHmS2ucWwJ@?~rn_uJ7zVksFTj>?7S-Sj@dFCY-P_V4bD@sf!2^mH0t%jNkd1vb0m
z1pmS7?YF<VZRJ-tMlUfiFfcQ}euVDWT{!Y!BzWQGNP^embYJPCs$C<J6U5AHC*Q1P
zTf+A*c~aIN@gbn-rm!S*vD*@+sHngc_v+`9i?luP%BeV0OH0dU^Snl%+iXHYJ!NG#
z@vIDziE_od32|{&*4Ejm3tY8zb!Z{C#ol7K!n2v@3s|ah$-ffHW}*SsNFbh_=Az}d
zOMdZ!JTmD5xQFZ_YZMo87tJIPQc_X^juU*I>wMY!AlVbpYtc5YYU*|%KTEi@(@z$J
zGdun<4DtYEmTvJXBn<#bc7m6{IBWjNew1>9xF}_#l?@Hxr7rHlf7!bxD%XDG+a<Y}
ztLDvq)A}d=jP~Lw5{hA2Sy@HJ=M<7)&)rd-s_(>@4XFbp(3dECpsv2x{Q6E%P*8Vp
zz^>WKqvl3&Nw8!=Gzj^p-t>RmK9#)Zp92TJMO_wsRa8rOOsYL4I}7o3{lyUP-@k|O
z>#;fywbAZA!;O-Xk`pIRG@hgT6I!nNj2c5JXa#?uqC%zJ027b?qWX@MIJc*gi0g64
z&Gi+i4_?f(?+*48CkKUK`+I{F&$Czmcy78SY7RUTFh(#HkHi=`VJC=zs#XH=@Zm#!
zef?9XLbLgHS7v__8nzB-0jT8bY&icIyE1Ww0YT5k9>E2Yejwx2`4}<W{KkIHT#BhJ
zi~o4GYi85S6I*Rxbs6jML&QL-!iwp@I9L=pPEubrJ>P)$lSDL<xek=a4fLHYP}gbG
zcriQXCPX{El2r3K$Z;#K)qbd<OI`x?{>3K}mmk<Nae&(5y=S%4nmF#hyD`QQ-tYzO
zEhFPKUKG{jQLt${S<vT&*0NelY>PbG{?2Bz%`F(4UTiz`<7C#E1}GiVo*y~Q$-u-!
zLwX=i-Xw<W3J(v^66tz|hK5nMsGRT2oQ7;OyOk2??@z^VXQ-<iAIpTrVz2r1{ragp
znk*Z^pogQO6>?T8P_1KSO|_dHZAr55<PoSmqpWKE{j;M?PHVO0Gk<GckA)V;$<E?m
zHDGT;Lqq8*Z=}OnHMF$Q{YTW*)q&e`rcx?}qy@->K+O<7+(s!-H%6>*S-HBd28(Pu
z4_1eCcWZaZkByF+nwXf>dYLf>?op9Qx`RpLsSHryC9^m;Iooo0x+J?uJzM*il=pBj
zHo@haoeqsaT9T1Q(D3ABTvXIYv!+Wd(nEWRiP#G@Ff??{vi%+SYMrBm@iAQB-vDM1
z7g&q;_Ob9yW?!N%X66mPTp4xQVZ-fO*FljLXCJ*MSL9rBA8c1qsr5z3!|_Tib73a!
z4)hx=?B<KJec$W0@Kvl;WF=l#yL;v9-SXaz6Uc%vmo~Y3_xd00KVm4*qx{vWhxvW5
zt`Wyrn8`<dX?D0j=bJv+RRD44R%4pVse=gFJv$ko;}uPuJR#vectAq+O?9vXY!g9D
z4jr%1K1WUmQD2izr$g>K->euSm`(&<uK-@GnThq-AM<BXK=Bw<2K4ve+h1^Q_Xk(R
z5N%mHGSbpx07Q~aSg2-3Q-8f|kinqtjl_F1ee1jJ9BX!`CC`yOcxvOov9Yc%hqjo*
z_7d@u(MSEUI<C_7v5f3w0Z+52&7;bpu1M!J-F4n}FQd0sT}WE2Tku%lYvfUb#z~RQ
z<$>X;mk;$_&Ivy|OohvN<4hJiD4o%wsiAR5WR!=8$0y+d#GL)OK$N%VbZ_xdi2JQ?
zqs6MYQXX|5@Q%M+nLqh{)p;Ym;#-4fj!t9&l|jIe(DfqymY`>Hk_8oCN*IWXBb9_+
zUtj09AD-yUlX*1|Njoz-Tu4_LM7F8!mNx%Y(uK?YPOF}+zj<3wk~jB|-eF?Em2&D~
z2Y^}q`~p-{C(o76zZ)NRSuyd(pw1Q#&JVa(o-xqynu?u~&!D~YekCW;IDTh?k!^D3
z=3_wQF)75eSCrMkc=09+-*OBhj-<*>d`O&2+p@c!b8j*!t0bdAe#|&U%)$6yqDktY
z$f>Evrl+mV%uWV#ilOFS)n+fOT)c0rH|ZPQbXxcIQbkTw1Ew(gD_yw8jcw>6P0`a0
zB`G{^-vQEo&%xQ7s5!y%4bjB-Ot)_}3GRc7<!lz1xYdsyOAT$6sZwjruD0j5`CP`Q
z$W%G>r7jw;$OEv?HWFTw+V}gc8xnM{C5j#`!fYW7mV7pPRr9k&ZY~wPw$#?ve*5-o
zLm@HVDv2Sgg3j4HIwHIBIr_@g!~l1&q#+uN=405Ciggk`Q+mX2B<m!Vo!B~8f5yO<
zT`nlo!`6D^V8CVI_D&tPvMbfEy~q97*B5HzkhVRA=RM1t?b%irUWmk=$y=!xnmocE
zc$3^~w#X)%EX}ge;CU{eJu!9;()IwdNVV=OK_Zc7gEefwCZR{v=3lhB+dGWpM1@%O
z2J^I{teamV9SDZD^(jszy+**GG%El4Ybi?h5k0`T1gfG02b8$R`}-9$163~Po}aEQ
zJnClamQtIza&OD};Mz#ay35Q@tbi1?DrMw#xvqRWr?2^FB_d*+4Me_5N-8ccwz9G+
zJR3+4SqA5X(Nwc;PgE~IcBSp&KZ4utKDhiAy|y&HwLaGX3<0GGg_CnDc5XHs8Ptv%
z@_BOpZvL55+#mR9y%Y}<tu+W22sPB*yHV0cPxo-azI<PlR%NshSlfF4{(B{zUaslD
zwYWT%&5r1cM2>V8qd)eu;@S1MEQ_`*V7w^&`nXk1!?Tcr->a<qc@qBJMBSOrTH`bh
ze5<gk_qB0n{!fv6_w75)q4_n1f#p2L<&XpRz!YyOciF+yJ28b5hlOfB1_I8@oBKE!
zr5-w?=5hjhqt-?z@_QECA6X7&)>-Pe5vWs*8684#0w@rY@4#J3CkSg4-$q*^!ZWw@
z2ip6KN4?QZE#*aik;jOkf(`aM$EZ#s;LM$3R|j22sXYi;A%z~pcX}^RhW|^9&l+W;
zmpC{k*QQHFTxJ=nLX#hi4%1n96dUNdJF$TCMPy1!x_2M$Vw2gk?L36<3knLJKmU22
zZctJX<AQVsG_O@q?G*ZE(DKP-c#8LSur?NV>>=<;wc%_)n006TJXVC<ocWff9QH=^
z%=R8&pNb(iH#aXWd|x?vJQ(?OmC9l<q3D+6!i;I>a-5GkXXa7A_H&X;!roV)m;|=c
zxX?OkbtqDgor9yRz`CKBf#rCq{u>CKe9O~LsqtAtcEi716|Rt8qbbsz9~Uw5y56h*
z21N`P>`Pf_V4^f~jcv0fFA^4xdLE;fzuC?{zxL~nM~rI0y;IkEqh=G{@m4v;vCTgD
zp!jdEzBFVKc0E+C!?+q5S@3m-vD<e-Df~*yRepHPtA)2vnMH50_Xoy&KzZ-a$3>pp
zM&-F@adNu4z3VG;8|&)~tWt-;ZEk5@;t+5PY@qoy^(*X9mQzz>kabLf>C7id(E^84
zULmyy9k-d@p>~H{6uvQ7c(wYIrwK&D{oz_`lOoNA9XWwj&DU}qt@0)WUD^`8@~X&>
zOCQ?yFKmtk0vc$*bzUo5dd^BU)a^^Tp~w^N(X`}#x^#ri?efLX#O2elj(Y0n7sYE8
zladl$j6aKva@9>l(B4UJ4!6uGayr93%HbryFT+DjgxD+Mh(m`C0d&OSa7yYG=LkuP
z%pjAkXH6gX6I;5zz`&-Cwp*KMHOe@qTU1iDHMwceiR+0|F^B97U4;l1%pp=z(nE(r
za?Yk7?{iQ#6}PVc<c>4$^jGH5-4cjsC_Kr3?~OzIL#J13S;BME`>kU-DAC<Se3%l9
zyl-FtEuN8YaYnD73V`P)QWkG|ei?{a&|4cRC|}K6l`r7jFw<zt3%%5I-okeHUxchb
z(B&zm`mE>ApJ!*69dl*4weFjKi_JA7i_XOpzt9^N)*z%+dOvQizp<Cmo^w%aI9F)(
z;K<e*Y28X>**0mC8-~Mz-BKfvmbKFtRIfm|V!Qu6<LoLsU;a#&3tQ_qT_;3EfBU;N
zKP;1^n_%z8JzOm=&!(@rju`M|$a_W?R;#P4p?y%R240$kewDBOMn@nxp~+?4KEeLn
zmBrbZJQK+Pyq~y-|BiiPv(R}KZO6`Vzmy21&QXQs)<L9r@ti0hka>?UXJi*AB|1%{
z1VS^@5wd_}Fl~4yxIi%Sd4-Ly$+rD)d9-4|@R;OdO$`isovO&aS_$OF&VGCvzr@e)
zvbDKEFBue;%fNE4?Khvm&W|}nMn*<T`hk0^s%@7|U=%|D_f!IeL6>zYhrN^%p&B4k
z12TTdy?ggwym-+tOY)*0Qd&x<K%rE-3~|(s$Pu!0VhU}3Jo51o2k<?}B?WO;*HCuN
zp=HFnR5#<PDkFL<Piry$-MaRh5Wss*y=K0*w~UB;Um=cwM!>&hWRaz`{tz|eXMRXz
z@5vn~@|`Z8J-^bMA{0_g=W)%Z+xLmH{jbKG+Zp;4RnPh@@&fJOT7cz&U|@zEot)H@
zw?l9I)I@M8%KsDf)$bvIF$W&t$8)&yyWMn`vYpuKbmORq5&KjSfdBxu0XX3>@Lrhq
zJu4lS%RkCQw&wA)Uq$TF=U5w8em^t_$!ppEy~xQL0CkFV=w5=8puG`mr6p7Yt`0?4
z%x$W!=XVN>IA(RB=N$X9Ew<)pl-oQk46|L&tT2}45;7$Abi6}S6CNJu_^bb>5roPH
zc;_Fsp+hvH2e>0nF1Z~xsK*#>T?ptAbo0Gz9NjiXPr4dP+y^~BB(z(un>W8^Rei&J
z7s8Vy!smcTiX(YMX3R@5miwO#o<bFaHK9|EjgBrfs5)lv<*5<kCzAT{HBBmNDTs6R
z{_TcEj+i{tHjnQnM}oSO*2>4OlyCdOT!#+Aez<-G(kchXe)m)SV3a-wJ%({+#gdlQ
zQ-YW6hQ?HJ2}{*|f;meck|;OUkkvKS^&V3^D;?3Df-}5UKeuZrY#I&=JrRWj;f=C`
zbj&&-41T5j&WSg(1&VDI%Ke1&w%kk;IL`?zP+u`x@uH{MdWT1?Y;iPi6q+ej#T;vi
ziycQ%7tgl^Jo=F(u<k<MV13dDJx?@i;T{L_lP6CeKmJlFJuK_?NkVMb#bv0-CTvY4
zUpYhicHL{avWg~cLtrseZbpO3%W_41SbbWdlXD)cPV}iqqr;1nT|kiIdx{Dghi>o3
zVdvgn4?YviZSXmYzi-24;_G`lUa{cGX)lX@$+$YS#9$xhWSe<?b38IU)538bHMzRP
zwQ`W|He1E(z7h|&gj<di?Y2u(IzU+f13^b;{fO(0TIOnh***eS?1M(5x4Gi=b@%Nm
z%sdYRgc5Z3AYfkgab<7~2@Sn(Vv^=e#rTQ&ht3{8?o$zw&US?j?}Xzz^O0TY4T{r7
zOtgcPGpZ@c!O!qx!gf{k9%WB_KhW!4wWlk=XDwFQ=DBhdSu=OxQsv{Wqej2as?!(O
z*Vn@%Q!<`wYih<cGFCslg<e@%0g9iN*D{Jgw+Ij}Ej+9A6@ZH$Kl1q>(`{C+Wo?<b
zZi{U5ci+75-iD5{-D~03scf#b^Y?~c1$Uq8eY!j~cVXHEW%Kr}?Ww^s=Y}ua`35+5
zvSY_|KRLgLk&hr`rDRnHCdEs8&$!JmPg1LLoH3}+!Zr(?cAM_Y8UK*z6E~O^NkarK
z3USwmrlxw6rN}|UacYPIA)8fj@`Ih##^U$ZhS)sUXM4!CNS_aN{{{xm&dakybp%BR
z|4049d|$wL|Lq$FTm@vTR;K1P809ROt~1jKLQ7R_g7@fwa)^^o6YW0+2a{18q4dB!
zLB|`O?%!_ZxzytX?7BS)#Xbe*8w%8{H%;n;)4(mA_$U`AUjVcjw0O1k*3sV2{@Gzp
z_7)V{4b`5Ym!J>~iu$6bxkFs|bsDRwwJ!B}ZPhR(5Xfz9?j6qLL>?B3N%>g5_x+bM
z{E*Byh#jj-sMHzu3th4@GF?SZ)Bfa4hlSbKhx>2sP>uf2tt2ER*Or%6-pnsuB>N*A
z@prbm#K;&Y?BNRBse*|2=pVVsBV%eG|2AV-q9fCskKps2c2!xY?CskhJvUszU=ubN
zUeO%h=`xjdl0tL-&>@!}z;FMV*8S8w3?ZGLT;_&fFezYA_}MQb1f>eGA$d^=ec{3d
z=wq_4yAXse#76R$4=yXSMoGAKsDHkcLX+>e0_MIEHRJJAEZ{LeKfi5Wdq06F2WtEN
z2baxt7fHz~_{pgi!m{kN&%c`T@bNiU22fROu)<V#+Ej<a%)(?J@G8U$YGmsQIXNx;
zEFpm<*%@r_2P$f6>N98VemQ*}lSo2rXI@_7U7GA_d!w46uIrLQAiG8P?DR7)TnTLC
z9_I~|80apvhhE~{*~km<J7PcX>)S5))Fe$lPFqJuW)n*I->*5`m0ETW__~L|!3{E{
zC*k>iNWLT+%KCc2CT5{ID|?{nfwCj8hs4`E(=+Y_57C3A1jblqrNR&Sct_9>WriMa
z%U+rXu?6+1zlP&)VkWn@<l*9)21$u9s1jI8X?NdtxP7>w&r<+C77!>>@)X+L&|C{1
z1i}gA0}1Z4QuxySe+|jg#C$N<!hWbW9Ejxn_wQLY3J6QJ^U9w(>uYK{KA1HDtyeb>
zWccshsL#MwJUi{?;-aCU5soi9{ztM@Ps$TGdpU{s8a1J+@m$5H<#^J=Om^e}^{=tw
z&rSgyny!>`g{z0~JEDdoN(K!CrQ6~Jm9V=rf63{;rpVg#`d@o9d!QUacLRa)Y~Qn8
zBhEGDf#Rh)gkIE_l%=nLes`On@g^lDdAPf?v)8U{T_G-py{#?F_$IHtA!oG2{%B8h
zO-&6U_VL-=D-6EDgh~9vxTp3RtEhx!XJ<>Nur8hWYnskzQ2w0e(sdt+b7Cbh;}ig*
z2oiv3enV>QC}%VjXN3NA|IJY`#EwLaK({L+=Rh3=q|Z3q-5*SFPojq5YgiY-+ek^@
z63Fr2%CGuCo};?5`Pj#gAG<A0Edb~H^T{nRRHDknza}3Hdu(K+YDjT+Ph@kk5V3&_
zL5;Tt*5Upd>@m;L61e;QyCNsveGg&IL~&2lJv7Xy$vp4ebyUB@ja`er)NX8QYHDM{
zU9}GcXW&uWZ<uaO-5srohW?CNMho=se>%(lHI^nC6aawQ`rWoaB}RZtEyEDwfK89#
zwW?iuw;N83U88-)WZz()KYtEiuoGe5wGr7gta6%s@oX*I!g$+=r{V6jQkyPiGj&n;
z?go-e5oWim#5x7R=m#nTp)lvDDBF!n_b*yWNJz}O=E&*)wM>avkQX|Qe*Y1!uDeZ#
zGd~x`;Pmta99lOGMNf<%FFW43kzii&scUGubjx$NKKit<hE2I&**)$vq#zgsonh+Y
zE7yfF_<K03>Xi^k*LO11G9j^U=k~a`xE!yJH!A5T$;oqF7tE#VUh5XSg6@rF-B)Jm
zR(2bxPNHV@IzN39oN8K1&}ZvO+)71$?;G!#GZ5*x>i}I|i7^wep>TE9(_8VC&@}?E
z&$cd=guDAot=c@V^B+A!BO_2^{O&pfJ3^4YT|^+%JFetgcFeYFSwl-l#_6zk+4Jyl
zql=|K8jHV=FF;=~A|m3FlERPE`?tdVYE7?7JI$wE$tNA3pZnS_F2v5}c-Go_`s>%P
za%kQe7MZ<3YC57&AbC`Rra3HHgP5lx3yV=#^Y{Y%tFOR1Az9>hGad!i=a(%(UV3`(
z=*3{e;!mHTXPDA{v(B)~K{-XQJ*F*=#d7AvrdQNnic~>m4ECBu>!+n&H>HM46Ch_y
z?Z~h*9sF^=A)VrP=Vzg=v^jQLpX0a1^yVHKW)2H{-F5zTGgIUELPtl(;^LxG!flQH
zTFC9Tv=~*pLpc@H&|+o0ksMmQuJfi^T3YD}k980fino;w#X(WS5@%K|kxqzAK;kLH
z@?9t|B=`s<Fkplkr1PlA$>%Cg>s8OudvtS)?;%PgUfm`Hi6_U-Q3F5{OcUKr7Gemc
zrvU-6(3JM^k$uL!>j~9;!_Z0fT$?Na86*(iyY@R<Km|bc<jIS8xlpbfe-G98(PhpU
zv{9ga9TH^O^#wyD&-wHBzy$5$Lm(CHA*Qg9Tr4!Fp{JhbFy5LI_nriPu#*rkU_x%(
zzFiHZixJS?1gj@@%0n`-TboOz9GV}Vk7D=A5L}2Df5hTJLg;})&ClbTjQG?;Oe9hC
zf;&Q0KjcTv0;CQUjB&BCAPU^}qMh*)u@e&$(EfF;4P#MCc+I!#V1}kV*RQ+I4%S=|
z4Enh9d3{BN`0~;TLr9D+^P{*BdP(4<F5oq!CH{`ON(EFD6cnJi>6)v1N`MPHQ*7h~
zMMivRXd{$ac=<cjglA<DB}ifjLHCuRNT^Xlrj}F0m@s!R|1j>zUT(JyL(GMs(DTW~
z-akmsb$5;o<~$-IUeGv&{*%(#PoR4F9eGJ{auw`4nE8#RzU(_mz7z+!*mk$}HqKaD
zx)wyq5M}E(s)u&NxY$!(??;3SxT!+ALihg+=%NE{;UDEstV=g|fx`V~cYi+?;73MI
zQ&LteMN#g)@MCK<LAeIg12HBaZSK$ijMib6*9kEblx7KSdt3<cWk-8h#p~_u4fw>|
z+PY!<!KYow^P^=L1O~8%!iLqyIVA@`$@m+h9PbJ9!I{p_%xHYH(XSw5+`s#>HfHt!
zf+01Sf^u;8R`j+6RVkxI`t-%j`g&QAd=To1iY^f6YLvLIfDG0p{HxAQl))nr!TyZ%
z1@H6oL2hRRn2?}ryMSTI16`WW<Hzmo?KbI>P+xCDOrXwG-eC=@f~W#3JvUZo-fZ6?
zfIlF)ltOqxj{<xx3U#&0H5s}=6rji>NC!nx%7-E8{90UebaWiP8%(j|(ookaP4PA)
zh|rxrjezV>vT%D3yfsmZD7DgAkHu~*m2MS-s&I;UYX`Hzf|vrz)Ya7;p4B332@x%a
z+I?^Vp&e|C$NEf_QZxPbLJ^akkrfXc+c>mjwDT>a#r|hBih*ZbB?QVXBO_zy-VXdT
zGPeiqBx|@iLF~}-_kbzZbI?oz)dI*kB0taqZ~Qw4Q@uiu42_IHFV(mvlu`AKN(hEw
z>aSO><=wY$-xE@*MkssI$3ymlr4d1WXyb>KfnX76+{;FB-}Hqw+)bi>$juRGwt`53
zpmt0nK97h1lLvW07bzJ?op-<DuzkRA0s({&BcswSG-y2P>MR<Qr63MqVi|2gbo(3O
z`ytO9AtSSHecSPb&Xd_h2#BNKb}D6SHP!xpV5}trWUC;b`)eh|^Id*?@nMvSfWCu7
zic*c<Nv9q2ASlE5Foi%wfm9c2YyH~kC7nwIhbG1&mqaX7OVAZZg4lC-p*;uKk>7~l
zB_ErdoD7`mDB8LNwlK?%a^|n8&c8y3g@s)Z^@X1MOEAsah_B~h-3GOBG;_H;O-<MM
zZ)H23=<!Q_i3Mqwf<gn-EYNCY(K+D-&wp<4ePlDwsrB`goc6@q$v4-ptG&-l*DQ)l
zTk_&erZ(a$B=DIG&}S1C-t;CDzWd49Hj+~(j8pO?;n$$#KDRU}G8F0hqJ-atjZvWQ
zWt$nTZU@Z*a<f46N#=6h`4#-ouB4a#WRIb-anPB|KSj1y6hML=mL<-LdPP%<4;-ZO
z=$&>WYg?JMspc*yT{@d#mC@!|+_1bNf9Ra3r*z!tluEts;3Tvh@4o-(qtJ_SsO#y`
z0WsUqkdXWGue>&0wv~$SiK4kVIRO#|NMu%CULGme>>S}r*pPK~b%oDTQc`fH^$iVP
zi{el;aImvK6?h475RU?5Xl-5e(R%~ZOn^XT_o{1ZY;A4tK&LUYwWUQhQ_aN4s3l9M
zivXhzKyw)mXh5YkHHCze2QB8%bu_iGK!=B`g1AO$llIig7r>L1m8so*$WvU$Xa&zu
zP<*iJ;zb||mY1C?#zRKMA^%;amIK+He=|Mq*+DYe=C>M6U%q@18)I3&)i}-?j$^iZ
zaR2t*#)74WB(kgx_2SwFcg?xA?wg_m?p#@;209T*RvkHvz==Y(ip>}tfd&;QE*d$;
zfF8(Q`6!W8Q&(pYfJb58zGbhBQp<qOJje38{?h#kQgU)Fpfbz^&_mC`!GS<TURJZI
zQvf{pMm>9KaBy&J`<j^0&QOE92zB+gpHqi0PSf%2&sL+NqM(MB5EHv<(JG7AOipRS
zH#Rmt)@wE~G~{mD40-#X{y3ZKv+D-0M1;6MS`}w^qAc&U6+TV5%p`Pq>SrO}6Menu
zQPory>r=^_bd_Jeywc1wR}gVubSN~t8UtC1fss)NL@q2Wc<{F;WOSe^bYEY7ubZu&
z(E?pPV4Sm*(}IeM+`JDQH3#_oE3C~t512ZF1QO)8?Cf9rykv=4wdP%MW7cG|5JCjA
zKBq8}-6z#JsBu?)eCFX&BSL6Q<;0BF#oZ%0$)pn?r=FY~hvIfHn>w%(U?m9V1)Qd0
zMZ7#glnt!~z;V~in@u5r0C-0eP@#+c<xR<_fq@$nxedSoEcUuhU#@IyY62N<7^k!*
z95WFR=<4m&1$C$DyP?WX%};qo>D3>COLAoDbm=}!FgDasfH7sheOqtwJyJGB37|6|
zM@?YjJy&a>l<Mf}>H`j=Gmjt%IZQ^@P+$KR?!$LLUInv41FmPeYtKK5!a%ENW3kH`
zm>ZAD&b%YS0QaTzzgy&K{y6AS7C}Sm+V2esXss^`4GRS!X$H@vtgOtXgQ1|IDS~wc
zGo_(HgBZ}&nQ(nt?6jz*T(g6A#@x&d6`VXVH66_ri8#)e=z5r#nxYL*%E!cwsbSLN
zkVZk<eOS1@wie?&^K)fxq`<1napc=M6O$1{uX7S)N?KZ4NP2ea+5u>UHJnob+rJMB
z=~H$66^;IR5ZZiwhJ30LBU}FM8=+^ZQ{>38%ng<R3U9E4pn!n-Z{Ls|IYs~(dYuOX
z4;-bwt!I4#@ToC2hEQ^T|Ni}Fehoe5!~xJV_`|!hsi*#!m{^}Gf<p>ZMFE$Kb-f@4
zKtcQD$&)^aCN(wbFfHp(GZN6BOey)=I=%pZSp#*T3Tt?9umULDq?bXU_=V6#xD$kw
zR>)K<APDI))LvYYyB`3$T+kXXg3Rx!zkf}2^=mlC0wIndsDuhgcgt%Gg#pI_<g*25
zjsTo!<rqqtd0Ill0Xm8V@!^mnASQs_OCUk(FDD$k0ZOE+)XPgnMWxIiiqOtvsTa^+
zX@PTPSZrokX)%z#g+S*G)kQ)wM#sc-W@;6HBu-phJbY8(ATE%bpjac+_g|xZiwM`+
z^9!Pfg=G~KN+B5mtb#`+Jb!Mpwdql;o2`@%<w%BV#^RUIts0nWyr6yn6{&BVu3M&2
zT?7nGE8u9kvZ;N%=ihzXyrHka!GljuFWR1hXbCX|F6_5&#sNU9fOCMcw!OaN-F7Y$
zxS@wc6&Q_$g$2YDxbX7&hpnWK5>4kn=gRz+KT?VRIG0LjzJDsGAfTSV{<Q;iOt+<G
zyBU;l800rvARaW^sr!E)c=<!C5l-U#MgiL=5q^WQ-{&}f^GDl9SD3kTDT}?mMa9Iv
zg6=b%Gx@H1mNwK>KNb`@UVyJcW1qm=QgmKDLkcJFmY;lvgC$V#33+rmk;7p?9U2-M
z|6^I8TRM3zCY7sX`TqLy3@CS1pe9|I=$Hd)ig@f#x5V8AZ0Xr44nnz_sgdh@nNSD%
zj4S%}A5kD7`TT!-Fq5ZC*KC!Yo&9m2R?V3V75}e&XAfJQT%loGFLH|@?P4Pw@B=Sn
z4VNY;%A=z-Qlu#X@1FGmn$SQ^?NrE80G!BzCj!8QgP`*3GQ1=tKmVUjev)rhilhvt
zG7fcKSEpRq447C>aGc;@yP=vf+2!qe?VU2_?B?tX&dEOeizP})y73+{*Pmj}H!{X|
z6_0<lic@~p>)ngwv_$U@=tx(XS2b@}KI?jatA@4m`9}xY3{L&ivtgWK)R=aDkLy7?
z9@pr{v#k=GJO}zDoAwutfBR+A6?JUiaIDlPC+2tR>Iq3yj~3KF|7!E<wH9$M>tO2e
zfAXl;T)<W9)lGaTLl|AbeZ{6dGFvCNJcOSa{xday$l_p2ey&x&y>uFiCA3;d5T7^f
zX<{EBk#u>hA^SSNYnI!6MGSNNZ1y*EJ^S`n3UA!&(t;elz!J9UBQ1IlKehL(<yCDy
z*N5XczLFNbnZ1m|5xV$S$)RqH!yKn+x`OT?=E6C}PXJPmq+41TOyNujGCMyZ8a8W%
zKDIis(Hc%=Agfc3RSkc^(VX(^^2U0!NY~;d{`LL(FPZ8I5|Q#-smd7ULwB1eH!4c6
zckon3y)nne<QI1E-_kAQj+oP$Pb<3a>ebJ%O#k{Ru&&WlV_&2!s3j%m(=Sc+v2@8N
zaK(=5SDdNy`epE|+dfXjy-On*c~pUxN!`)$y*jiH#g9f(cLcw<`ZMseG)Gna0$xL|
zBI?4>d~Aovr`65R61O<zucnm?i!@rl_j1z;*^G!dT9F|0%gEi%cDl@h>Oi<rG2+Pw
z4?&Yg$KaJG81;=gzeXQhqh@W94<iK(`;^;U4HO$z!>M>ub+ji%SWNX={AqXtGC5iw
zm6;kg2zxeH2OQ9E8h?w{5@k|Y<8=G=6x*cvFW<2Myrg$(yqg!cL@iu_Rp;5Gud8$_
zhbm*4oPH9nP1j_X50{;9*0T9}+l!W~$2-ptMq1OW#mSyDV2O%!oK^$*QFrD+D%VOo
zYj=US<)2%^MY_Dqyllu4Bjy4w9~p_cW*7P*PonY+4+B%mU4bH%6K^F!CbB9ZaLI9s
zGFs5ke!_nxK5zfoIoYj|+RsNP!kaTg5_4O1ZkIY4N_}-IHDb{TUl&`?4Ueo~D;9~v
z74_#`S{$G<m>Q)rIH1LLgfjPi!TXZ~2}-JY@&?}5H#Qz_4y)__z~BL!Wm>13s-C=~
zx-i<!+&?OXt%_F-kYWnjz^(Q;?nS=1b}4JiatwQ7&2!PCM-8vZU~ngMJekBH^x@^y
zg7+%vnU<4Q?BeT^YmC1HUCjoo4iwO(>NvU}O=ct#g^|dD*1p!*tM6v|>K;o69#Ehg
zxX`EzTG2^5rx_XAEANYp8{Qo0X+3**T2u7{{+@fzVh<*9^|Gke)&^6WXI@`e%@>{c
z4N)gUrSg}UC!UMsb4(`~l|g*Xtm7S}a!Pf~74@2{+x=dE*_GUhbLTkNHiE<Lo8+^L
z{O7r1YD+pE==Ypeo-%2pI(xM)Bh&kV*HhUHoTj?NyUmec-1~=;U7dTkJ}af<ylx#m
znyAHoN{w*boSgpV0g?yz|35og&de|_JO6IR;wPNN?v{Bb8KmpNp@0u1VI}dVeR*6U
zxJ+lDytLa!WV#_-_}N^aOLlrnBjX;G2f#M6>g;=f$LR!3@Gq<MxO$oU()IEWUwiBK
zITmJ5A?G&VWIP>9Yd^JK-^RwGbL584y~MqW6b6-C0>L!q43oXo=p+;e3FYIPWQzYx
za^<GQxto;FRJ-=Auos79qxeZEAHkj5Qx%21MO~8rph#TxuE7EQD8m02!0OkXL5h|y
zW){<D!Z;7T-1{82?^AV@?fCjp1B2iL$5<;-{-p_S+T}^AnYGf9`xSN15Y|={weSC}
eqx&DX4%YCT;3jiwJp^+kkrI~|%er;%(f<Wjxn_F+

literal 6996
zcmd5>c{tSHzfZQJvXm`bCCQ#WWD6nNV8)1yB7`g<%UC9RA<;tFWg8zm$!?OZ>|s>4
znMg9WNyab?_cQu^f6smHbD#Uqy?@+4=6TNNytmhTIp@63na2<_gX1ixSq>dKblk{L
z&+^bAx-Hu82t7EHvdLz8=+I#@#007je$vs=9X@=Ro}Qk8fq{{ck%@_knVI>>kt0Ws
z9%W%+VP$1KcI?>k<Ht{&IKjro#?H>p!NI}F$$9eR$y29JadB~Rb93|X@bL2T^6~MV
zK7E>>pI<;gKu}OnNJvOnSorMOv**s8JAeMXsHo_L3m3%1#4cXEC@wB8At50pB_%B_
zEh8f%D=T~H(j_@LIeB?`1qB5~MMWhgC1qu06&01smoKZTs;a4}sjI7BxpGBALqk(j
zQ%g%rTU%R4N9XF*tGc?ndU|^L`uf+dT{AE+Ff=qYGBUb;{kpNSv5AR^si~=%nVGq{
zIRpZ+u&}VSw6wCaf<mD%7|hz*8V-ls*x1~-al_Wu*3QoE=FOY-_Vx$_!ok7e)~#E&
zZ{K!wbaZlZx^w4_v$L~{i;JtP>)pF|-Q3*V-Q7JrJnr4Q=jrL`<>lq=?d{{^bN~K*
zUteE8KR<te{|65q1Ox;G1_lNN1qBBOBaz6EkdV;O(6F$u@bGXH3ia^e!-$B8$jHd3
zsHo`Z=$M$8*x1;(xVT4;9zA~iI6gi;At50#G4aWhCrL?3PoF+bPEJlqNqP3{+4JYm
zU%Ytn^5x6a)YP=Jv{$cQy?*^VJv}`mBO^02Gb<}AJ3BikCnq;IH!m;m&6_v*`S}F}
z1%-u$MMXu$#l>hex}>B80D#ic(z3F$^78VEii*n0%BrfW>gwv6nwr|$+P81t*45R$
zd-txszP_QMp|P>Csi~>Cxw)mKrM0!St*s4%!MuO}zP-Ku!-o$Y9UYyWon2jB-QC?i
zJv|>ke*E<5Q*Uo?UteE;fB(S1z~JEE(9qEE@bKr)pGQVUMn^}-#>TK%?D+WjmoHzw
ze*OCG+qa2{iOI>ysi~>y>FJr785|CW$K$_$|2{iAJ2yA?<HwKr`T3tee=aO6{QC84
zadDAAAS^8{EiW&xtgNiAuCA@E5sAd#zkjc<umAb;XJcc7L?Uf&Zj#C5t*x!??QIH$
zva_?ZySq!JQup@u_V@S0#Cn6l(m(2Bc*F0|Ay!V>k8Wf-eD}~HUV0-v9cWPIN>*?v
zw|;uhYr#Ngfo?2cpg)}1b!?jBT04iYuc0=Bg`VQWpvW@{371$NeYwcOBJ_y)oJ<^<
zLtBqd{~9kH2bYD2@5S#d5kBX2t{yR(X`%L9@HRB0g|ev>7v4l5Q#Y8O5mpz<TQDuS
zjhz5ET!H?sQEip_p$~NJ9xRbxPBB!9A3<2Lw-lWTGSwlbN8*(j|6dZi42U~^hi?D0
zX5!kNq%WdxD)2rkwN--Hlz0-1(u<Aa^e~d<mV9Yb)-u(Hh<e0`ssFGeBLBm<Otj9X
z3-P!sobss;F&4>e6lqI@v|fT}&_jpfY>8)#Agn?F4-??}kgyF7&mUE7=Mi+7%tdNS
zp!e2&o%*^4!ozRm1v5g23>dL0>i7Kls-4Kc+J16DG~DhmpsvH7w1iI%*Z3Poj)hm}
zGk!4sEh8IpKh-g~(9xjYXxT=;x-)6zqF?QeML$!!ia&y(*`Jk%E-7bE4qLV;>+F{%
z1~n2bgFf~lGDRH_;v#MKjy)1Aq5;eA&ApUEe0@h#Gj^G9-^;j*H!fN4ZSwtzPx+R#
z6Io#K;<nnp%y)EK=-fiL>z~&y-tBW}1TWlxq^qT<#6Z@Zu_Imvc)ABHsFczHY0&eC
zVZYkOm;>)+pjxiSM)jnFKnYiDT5(kd=sLsut=ISX>+pHV+X665v6IrI;Q@mI7Ds%n
zFH(n(@E=FGVy_=eGGlvc-`Ocy(Sp~lL#oug_02>&=*;U5ap%;PvB)?=Of{U^>TO&9
zm&2N*-qcwR2d#W8T;s{)`paQO?bT^f`a8A=lX6n=<^rvlAYOIa>mOKYBN$Hz&#(UT
zVIYHF1IK5cKAHBkwxpSkEu|nDl|w8J#>Eu<j+M<f+*gwIoe6Z!*aOAi(R(GXo@8(q
zDw|kGAQw*Hnx5GbwK@BT`A+^d4W+95r;o6DVSxiBR|oCQ><+4<`zpy=*VW{|j16zW
zi@N?%avO%CvMjWs8QHL`9q2uUzC48s;791mwL7Z*-3$j&?FUfXIx2s`b{9}aHo3`Z
z&UxHdh<zH~jT4CnnK6Z7NSuTJXmtaSs|RD|J)3SpHK)I|`vm5#@Q=>ZrQs7!UNrZJ
zx<(SHUx`QE+jU7F6fIk)|52h^ft>sonp^7smV;u50TF-kT2&(B{xV!q*j7y`FQGZ%
zXF}Q;Db=Gz0xH`thh@8Q?kx|P9fP)2k#`yg{t%!ztlK{p&qH$*tbJtN+v&|;)bnKB
zcy-`T9e@f-{ROT&upmk-S~LEdX}L7~@4$)61}gb?M;GAXwWbga;YL8~mpiD_sJ;GH
zMlR*f?~%)F>f0y6_5RqVD|ve%=&*lM016cqKoeTS_r(@0aC2uvZN3i6t4QkfZP;up
ze=j^U1g<9M;ug#5^(UO_0R`HQJ+T8DHE`5E;4tid7Ah>{6*u@Q`L{~AIXQ0x#HPR4
z+#7uQK5K!%R96oFN99>`_S4RMigQ#}F-Lvm#7@we*qy)7u?cc3atm&q(qP`8vU*1D
zF@F?iQ^CqO@>_-cPmP?F^9MH+2!ixDPXW>z@O4PdiH#O9vH-stJyzjj1x0Krc2Jv?
z{#|h?QG*s7dn>J_W6+`8w}dZWhCrSr{h@)jfMU?b+A~!r0w$8ccgmXj`E9I&9vUHb
zv;SWT;ivpBYaZBwlT<=|cRkPu;qenh`-3`Rowes@%ZecIL1ENVHpRUYz?YaBMcIFW
z4E(wGB_os-tAH+{tQ663Ubusj$!DSA6tkYb=l&2BgoZsRK5WGbtz_IMqchcLNLN%}
zbUzXVy@QbQ7vxMKtxWegUOng*qLtmdjlEdUC<5OLD{4>u4aKXZ-jjV(V&&X4o&4EZ
z_pmIW^F@@|Q+;C006c4w=epE2aGhuO#rBm=<{_Etps4M@@fAqxM2oC^1X_V+PZe53
zyucm<g0ZANw+i;iMZV<7)E?ub*&Cng&e1*vVlf~sdD57;x)<!(rE<Zrd3UJ?PLxH1
zVisZg$j1Y=XvB>RrfCk(!LrmazFoBP{4j{+4Mj)zTSY8^YksLs2U-Zq4Goi+cSsP8
zC{7^9FRg<K()tWqU_`Q}r(rR@BH!W0iq6)d!>+mc>#CoCw8l@Z4?Sd|ZF1fpnmn@V
zkHSdY@T1zK(3_=gOHT(UqYEe>bh3(24_1d1>NDWlq`gmk{5Ed5cJlJ`(QVAkgxs$0
z9~Q9OjIqrjl044{=Hhax4mNWcIK{W2%5pl_(g^X?4}GgwX=h5-`gIumUdq?%_2~-b
zkUQxa1M6(a#ST(<)c*9f{9jHTA2o{eL{V;$nys_gLYq)2J9)_{$?)emd~r4+<iW=c
zEi?H_@SJO~qUIJ>#K$~<V%h?(&WU}stHpy5O7oHRuJc5?B*zV(%aRLhv=Ec}1?O5!
z4~JS0u{&d%wIrFbl3h-+40btS9K!@dM(wpYAR4;SFs*1U>p_$EfEH){w|*A%3^UGw
zYeS@grEPZG*mjrDIqtOi2pUV+z6`YWkXI8Jiz#X98f}z^7Mm*`(>*LUjKIQN0flUE
z<P@tJ$&Viq<Is3n<0cz)?KN<0#?$=A_f_xRoyRB#4&WH}&wU7W)JiX&F&K*5B-nS?
z^Q}Zn=#oO0VGfzr;+9e$^C^T9SJxa>yX6KUfZY3t{G5131~=IHxyE?urQ^#yq~<$-
zCoA!cAvUmoYjEqwE`JP7{vHgsBU$aJ1B6?b&r1q%gQh~9zLgsV^QyF?rid#~`|I9@
zf{lSPP`TP*S7qY^<cvHRFS;(0jcHiB6hTe{_b-0zQ05t1DQ#?85{Bqtw>Mj6w^FY}
zhYiWwn63G7C3*6>A{XNu6{l#zHW7x}=l{^<W1D&w3E;3dHqPIzG9d9iI0B0$X!x|^
z<3wlf^X#T8d1z*2shXhW!G5Kf@&J)t`yS>HUITGPXS11;(vD_Q@0}qgNC#$%Liqq5
zmp$|kjno~?Z#N9DG&r%~R8HyL)JMcKmQY}IKjJ~%+cu@)y$ad!-<v3ptVYa!%C?^L
zio{e2<@JL2jj_1j74?qoC3_jjgP@i{F?AU|j_tr+@-I~pG$MIf_Bq98HBy#9k}cO<
zk@7Jce@IxVf=~$&hy*Yrr~!1OHH;j<MN3pQBrNr7AHbml>}|aI+<O@bfUptKna$64
zAFf93kM^eHwKN<KT(bys2FUAR?6)z2>MP@;T@{oV!uBNqMYyrnbyx$yY5i$;Q||)v
zv?|WQfAuWAd(PBi3FY(J)<y0D!cs*?gMZ{lU2x%<063Y=@K1Wl0H$&vA;IP56|XHw
zL@%V(qhd#K`{|<07LMo3u3uH%*>yJ!=umM90kgVRYLg6THE$w9f?d01xzQkW@aQ%F
zGA$rxh{c3=I|y&%_A`~P|FB|0&xj^mi~Wt3c}n;rB)t8LNd(^60ny0m2A(=Eli^Z)
zU$B#=PF_sF!l(^NL5w1mnkebxbrMwB{6=ct@ZwWr>7W&;)}AP)>scNJAN9qQw4)#(
zcFO-S@bp|gihYFMiWQ-N4QU&$09O`$7?y0egt4gta82&UQ<s!kz(tg`?W=v4ZAE5l
zz|*`H|CUwxC@rgFrEMYE(O~szQu8E4l+T*d!n*K^z>_dl`pOt|uQ6Z*|JCg(SpH&w
z`l(nlel-$<4zO(u4EZZxviCBs;1sJR39jN1Xxi+Oc0yq5^$u7RIne}3%M(W8V4NLF
z#fba@?wvmtIzs*Hy%%Pg^|WlkS{?)TAh89K9#K$g+n$*6e!U0zuTo%dre2hbru6u<
zhXC(*4>IFfC#IkEFi`JY8K*w+snec_DG3v_-iyO>H!kciv#p)(w)n{siq1BFLWpb3
z5Sd~Gt7mGWdU9#7cH*4pM*F;L*)#4I82~q7Wm|m)mg7gry&r}0+%{PJkv{i=5mk2Y
z)b3OngY>B5{A86+Dxy?(RT~v(K+@8bBOF~??BCz87|{BLiXrGfDqj_=v@ukXL+ti>
z>0dKyPRK>lNz~{K`l+#xuMDo^uH5v)t1QU)sV5P(rT94m*e=;Bzi`az8uXcr+QBM7
zVANf8!D0@cLxJxSS+IhUIcnn^?(Ivtlr*u+W+F0&p+nkMBrfr|%pwjRFuzBJwi-UC
zTtPEdwh`Ehp_ffXWSBq}OVUY%Z~C>H>}XBS7X{Wa(=iNVk#Gbla7al26l4Tk&!8z`
zf+tV<>5`h{LXEy^TxP1nbHS;bWt}MSsG0>Z&ahc6G(thyo-csybS`(x1h7_Lv5EqZ
zDmU!HxRxlS6|phH9>A8QD;@L_w%=t+a?l|ilPKNWqRQqXvuxy~mW32!ZHX+UZG9E7
zvfx?#0EYRMH0Pj8qOh+C%pUg!Pt}$hASci~-#Zt{1cNzb-5C3&jz;j$@bUz(fxR)p
z0^2i^(IN%P>cg}`ET4PlAx+UH8Ug`4{UFs8i@8*|NCx3P1GFY)6u+}WE1AvBS(}-q
zKxZvT^Cd1%Xpn%t>eoNH(Fo@2qj!&`yMz4*7nB+ee|-V;MFT5ravPPFhddsnJ#G)o
zUzHA;hdMYa#a#qjr%b|Hw)f{FSjnNir0+AS*{pSVm=!fEq0AOU)CI-taQ3;N5f8lf
zFD?raL1%0M*^f0|aRk0GOX~3aLDz9FP33o@KbW2u1aCz94LYznDxe5g;>#llFJ15p
zx5@9B+}kA!C~K+;ey1U=j4UO46W$G_xyYDLE5ZS6_wu+~)PVZtQ4SDJA_3F#*+mpu
zDWkEH^K}>`NdOMc>S^aeT0Eh7rOJ*KTiR>T(Xko<UPIt|q^!Vi71lZ=#*C!fh`d??
z&iUuKS3C=VYU(=bdj9U+?Eqw?%rsCMPr3=n*GJDV7tb@{0Eda*u_{$Dogty$Ku&a8
zlqh0zR!VgJdROUY18$4NETD{5yIAeC^EBw&(luN=+}?Y-qLN@5s8*xRi1qtSMu&IL
zfo?H^2=M5=yF2p5rqIj4!kcn}m^yBY0e~luY|5ho>7#6kpW)P~N_B6rf7L^fc1ERR
z^)n&}*tsM!=z%(#QTPFt$KO{$>NMc`sEOcBEzT&<MPk2>sV@sPfR|1Vcvw!QJr?X^
z-;`67R8+lbT_x|dr)<MH#yY%6IfXI*<~WU7MzNvLrPbvk3ty28ziO~Q1ZAzMDW`fy
zt4Z3VW8gTdvY~<zE2($8D4|xtKbJeyAEp&lvSSPOA$MSy2-H3+SU!Tt3btluM+mDX
z0655-6mUa_*e$Sd=T)8(Mr`Sa7vh7H!(f;E85(AREQCm-v(Kgj!kZB`(rCo-aR;sd
zw$426(DN{yU+pa#F&-G=leT3A;1M<cuPbqPbfmYA0DFU~A3&h7ze>Wx*nZnW!Zsc|
z^iFMxbAncV!Mk2K8;y!CX}dW8C=Mk5Xx#3o3c-Urgn<dE@8*dtIk3X?j7iQb<Y0gi
zYmliv!_v&04H_<h?lT5DsxpJ(7fbf;#(rUbNDvwLmG)sZf(;DyS!^X|V{Vf;=&?T(
zb-8ZXPa3WHVM<<hS!fG%D#d~t5*0c|i{=~3?wGEoyAH{sP-3aHsu_P0G_FT-LJ(oB
zV4`9PJUegspu(mB6bijLJxJ3b8K{LV)zZR6hwWi*S*hoS3hQ@20fQ+I_SIq7q2}=c
z!NnEuTJf=TuWt}hQK3kOHAs5mHe)`J3>xo*VH7+w62WZHdt%r6zMd%r6-dXcl^t(p
z9ZBZgJzF|=lLW=6hrKMY1z%V`M*lIp7x1BRj#po@nBuv(bO&pxLsB6lU^ihcY8j0(
zI!ah?!-StzHYeJmw8~dKJ<FVR`@roTv&Jcj8^7E5rcZQn(72})gYuewXX<F=Y4dXA
zo#`}B8fj&8u1JS!FcUl}GAEqANq8&T$3w01@Eh$>dw0rC33-#dep;f^RzIaB>E*)9
z;BTcE9DcIje7C-=Jh*GpN|0Rl-iU<lQd;isn`{VJi7!m>MP%%abvDPcH5GZpoe)$<
zgm>UB3?dr5%sV*Z^srj@=gI5w;&~J$j7OG611`o@4qgz%I>S~tiMhEkQjqd3OWj}S
zQozKs774ePeN0orB^jTa-|t(5-CImFk`3<{N(^B|3|fS>;73(@aG^%-m7f)?Ly#JZ
z^Gk`_zPNK}ALx)MFP1gMp~hwWRi8X^TE)BcsRS39T$W{K_S24m3#4xY3l=TAB})p;
zPqOMqM`v<e`-v#EB%!fLwq^Rv4oNYx`I^T4?c3Fy88tf#wOza5MTlXCZm^BKx{|t`
z21ohZ-P*w;jtfs{lVtMpDMg)a!F#bAk{3dkxBR<inrC~tKK-(|2-dh!4tC99i!8*a
zVs@k56f-NFKe6BXNrw))#u*QIU!TbgLEb@VB7gSo7GXTqxiu%6=fZN*!x{6CX%JZJ
z<0uc88TBf^xTLgAe#<^=mfC{`uUTgKsJx?fcwKdAyS9+IiMqz%kj9*09S}(+g&40x
zIwOT<G~Z?s2nP3pgJ(S22Jt_>=N-I<K&+^7Ja>ZeiB{p#Pk++h3IlqLI1{LZE3)Pr
zC>{o`7M0kA>r&%Jj?yJvzH!D<$|3JM9Z6@lwByePF)_#<i*(f{X6q0mPU#Yl9YOFa
zMG99V9IK@s*A;NvX<c=!h9gpJ%CZZX9VGlR)5k(g`lPtiw&~1b`pp}cjF7~KO#kg8
z>kIZs+2`bv@mW?gvwbtOlA+1$fQ0B0aYFLfO55|APRFC!cawkCNX+m18ZG|%Q}`mD
zx|$^dH%qp>?vCO77}A~ed>7ACV7wzYv+TmljIutLKQ=NRTTQsa_wrZ?70{nm2*@m8
z#uURU-U`q8q}-O;k9uLLP;)$gOHaJ8V6!iICs9!BTYFEltkjEbC#lPG_TmpN99K>R
zT-FCiKN8Y*U6*-39SiSPUHEZDWi0uR(-^nHiOn5hBfsk{Un_rdp(4u^;?^n7yd&Aw
zMdx?8w5s$Io~PH@NEh2U?_i+4C3kl7pWH6fgj>ozras#RjYa1pgE1IC88urFNwq>+
zS51$RxM-cjQD{&XzdS0jdcZwJ;joKmMr!@XCcKUd#@*t4kT%`_!87&JE9@t<SM#N1
z#CZ+|2ZJkfPfFl%(~s{&;-CJu@2d@rnT`X;ON-A01K6IiL~3=5l@IMIFKeDZ337LB
yOS-y7A8;=PZCw+YS)H#6NzwaXyqmA@KlU||?0qF%4*o}Q$VlHzuj1;h$o~MBAmG3N

diff --git a/man/figures/op-hard-sigmoid-plot-1.png b/man/figures/op-hard-sigmoid-plot-1.png
index fb272c26ed07ae70add47bb97a29a375cb7271b1..8ca5306bdae80255f6a7e4c6523a874342f74c5f 100644
GIT binary patch
literal 17921
zcmeIabyU>d7e0!CAc%knNJ}XtNDhr6AyU#Qh|)21mqAL2bPFRO-8Gc7(jy&1h``X@
zb<dz*^}TD|_50`E|L&~yzOKcGIiGXR-e>P;Kl?cZK2?y$!y(7Pz`($hc`TuffpK9U
z{U7#4@Dt`IrcB_Ut2U3d>@hHKZ=nCX@Ez%OjDc|*Lq_7Es!QSu()j_A2@TddE`=e>
z?MhPUc;#NX_MWAcex8+)2Bt;XyBwC!_bLgbsnjf529>fdN$BL!4G1nhASAxW^vL+)
zRhiWj`aHF_@78r!H-#sA^iAYelT&8e>?eht@6Na;U&6p}iwh6|vmhWL2mj3WqAp-C
z`AgwrJP_b<!@y`yMO?yodOd_1<CUH1HU>uIBN!ION6P>ISO47sMNI>PR4Evt`^`5U
z9UVmO*nVHc<H$oY3JaeuD`DqXqCc9;N&#0wlJEP9H;BVHn3xFu_ghk_PXk178nRSb
zqMk7BGD*<Sxauu)Vqm;_bsu*38UjyVlZgm{hue#~`pDrC5lOp*L(|2#5|DQ3OlPZ6
zN1hESbK}>_#<QqZM8?q9u#9UKbGM0(hcPg6iRus*s}gt%R2h`QM!I1nj=C$IWs+43
z-@IE_$8F2Dq9+@V1eFGtaB*?-^77>6<y)u&3O6@5DJdzNsePw3nN(nio5aK;qoeOc
z2n-hNC|Frpi?HafbEo2etUE;d=}tCfnm}AnpVEsZejh9F>_;5e5iuQEe|pNq#>V#9
zdK(iHv)}naXb6`h0&gcoFJ|l`0$0))YLKDK)Xzcnj)Pg<&e(X{Q*vr?k!6?QTB=kC
z`3w980!^mA#>O|k-le;{yD#t`%DsWO^<_jDyExE{ZyXMXiwG}O4##b^xGsDno!O=C
zPAP;mDxjJuDAOq73!mOeqaeYaKnX;|6V=rZ3ma=}7m0-?^{}yV7TIObAr{}>n{Rc<
zGGN|E;9dW-w-VLQ-wj1Zu4j<k^n+kve5X4Ou5*Y_NKG0|iM2bbR4mhAs}}G&-k+#-
z*LUNDKq^dnl2g16v$XTNySwwX%d-3VHn+BjaeNaJ5+qvbd;QAF%0fayP)!7G^dcg)
z#$EAg_>X$=OiWCwboEqK4ZptiIy>2`fFhMyq6P;CwetosC?q8%)z#Ix;yT}B?X0h3
zU%$>iLnHg};p>u;k_UJirDUE;qxwTz&vf{lc)eRF?Ck86m6i3}IN8~|J3HC<`D5o@
z)0NGPj2O#8?kXk^gU`r_jG^#FD?=kAwF13o+XmRR;qmd>)lOCje_E7ZV4#eGLg?GK
zZyT<|o!2H_wuK(-??*;PYHDh(3tPVmIX?E_=H{lLpm;>x|Lt3r*U6#L@k^yF)#zyP
z^bskif&>%_m7B|^udiQ(brEAWJSj=U@ay~GVND^S!{`@}N$KnwZ3@r^$#}iyIJ34!
zLr!kK&mQKE#E_{RLi=>z`#rdT%lgNlAlWc#k!$vq^PFA<6w~#=YWi9NHqDTsplu?$
zp#0Q8#F;&8i5i<t*S)dGmXWYcD(*`4a~9x>JUl#L^%!m6{c$sVM8!L%n(mH^qIP``
zzQ5cW74sI>VuX;w;lki4s!<&MrS{25F*Y{2T+E}&=TBTn99!kc_q)R5@hEuR2FNMQ
zr`|B5=WD<oJ6d8AR$cABw~E9RX}17Bf13-R=4I3-vhRDRFjcg8<qHaIYwgi$wXLmf
zzrHuCmfs<ly0P{KyY<b&MZY_2k!G*dAYNWxbbfq-nvRzk3<l6kRzayjs-4&Jm0*4W
zsUg?Z!Cq)OqL1`VhXHwGV`D=@Lrcrt2S*%2ts2PZQi`|EJv!qfA|xzKQByM(coq7g
z1mwYuUr%fhUEN7SO9MGWRrV&dPN(GrV8{TdDOm1BNyx&&!qQNInAb7?hDZ#48pZ9a
z=MJy(5q0PG?dJ!Zv&z_34u8KPE&&;<u-;o=M}1Gn9YLyJ{_~9p7l(-G=yJ2ZI}1L1
zf&tK13=$9?gH*X6h>(H;YZsTwekk87)Ro8L=kbYw0%?hamuZz4>*?uza(ayOcbw<v
z<G60k?eDwLxzpn!@ZS5N(Ay<*AHK1>8yXPsUfUG*_U}-{NXv<uDzzN-8`u7wszh3X
zKRyj~vhG<t*vCKPm;6BB^=!eT_((`dK1oS}fbe_C;6Yc0e8OnC)mZ*BA>~ImF8+P=
z2GR8v3*TvVoVlz6woMDa@PPo?>*LGCdIPyGyf#07ccWJd>(4~B!clivSh78jcCE(B
z8HE^VZ~ys12NiO>-uuj-#_PS;zl&hgcWB!3c<|s$Jg-G(SC{WCV%tAk*tsbQ2@ema
z5^}Ocr1=v3XJV9yu9Ks^z2y=5+5b#zk`InBFH^}>cvb;TD*D$TofRR_!91Oz5)(!6
znf7D2{Kr-1bR;3bosTx!XhMwAySDzC-6#wi%u|YZ=x~XF7M^+)*C<*9u2t*im?G*K
z53y$7e(&?wv}`XOB0hY$K}2M~Jj{9Nnhf^eg%nFbKo}bvb05f7%gD&^Bl~xi;2|L)
z;mP5S#b9on!GGpO0ZRa0*Vf$J+|oj3`B?nHpLx`YqSpxyk;PCx@5H}GI5<SjC0N&y
z@t8jC%aF$t_}2>hL`VikMk)b220p&QFaK`nmKq{NJqHgPTeIG)w#fe<NAA>+gy?&p
z?0kIqmYmPhbf5O>g}(+jcd%a?%!5gXeJ316{k8TEjG|0?WoCwmm{?N!Uvu||LDh3L
zIqiNv#>dAm`p;Tr5nWvgkeQhoGn{|Dxknn(BThj?G%9%Mk6UgXpp3x@1+Rtnetcc|
zpLJ+C*w3_v^kpj2h<a!)6JEXeXR=vA(0=o>f`YZtG7BTFxMBJme-E(=DU$6O++n=Z
zc4ujb)SX}auf?@BHHATTTOKKq4y7b?XS*o=#|QwEh>6uZE)62c|6P6k;bP8olfMe(
z-ya@w73fykO*dj{Ym5rx|DKXVOqP01rhEbk85wg-M+<e(6an&9eEI%D1SO;O{v`@9
zz4+c|b7_)66W(Vhivv0J8`D!ap9{k|AHe4A*)6Z#8q7=5n9Lf>H@U;Ms)wx2(3y2x
z^V*UY^+iq$3=A|jiUWWZdMNkE{|6U`ex1j3qc5{cMhzInp|GODM4*<i2C?NTZ5)c<
zv&^}Yb1>7%cg<|0A&zh(PVxYv&TqXNRGC3?xPHYS#G%sYEaj|=7cZuzr5SRKEYVz7
zz>|P9DxWzX%boNzW`>dT|BNUwagHe4U++`j&4@1Z_vCtf+IB#>IJXU=uDC^95XrrG
zVOP66?XpUU2xTI90Q9oxD+}Iv_jwLO3ju}GAsX*lq->o=qb+5MJtX~eu|4CW`YIhb
zcTq-qaEFEO>9P@b6$ejl0Kj)yU(alGYDN-v7`F9!2BoU2X|cGe*LLTcTW`U19}yPy
z$=VbKMJm3vjjyjSKoLA5@tLZwuCCXwUw;uMZnGT)5ap$d7oFB7PL?;BLZnmx@UXC8
zUH#&d(P5RFo7+qs!0)3k&0)V>w!FNO<&E^PR4Q=n%g(5~SMEuXFyUr9?*Ax3Y2QEp
zA3PW7_(d6^G4}=&SM=a#1c+#&?vkX)r;{khD!h^qz1FR?=ASjfv$~Ca<jB#1Qjdf%
zNoi>lBcp@ELmXaq47X=BF1Cm?Nv!LX1n#$=o4$B<KV2S4!Q(X%zN~Q;Ia2LxEhJK6
z(o^TK@Iea3%+9VLBNIN=3VeNSqL%gWzR$$hZ{NZj!M~OE^PxBW9vaefaIEzyiajea
z?vkxfSI$Zn^<?oS#DRw=Cgwh6lAvj|;5$7%vYsZ-*W^C5`*0d8N>AT}K!hKzy`&Lw
zlb9uq;z-Tv1j&lmdgAykzNpKBW7LzVgap}>y`qDI1Ii|d`2C;3$r_}@KWB+aRBHDL
za)71f;{pQ%E3796*ET8Dk%bKeA$oB)wg&t~$g9<hvkli!rY0|~^mA5v7bf*SS`&((
zF^ITy+>>;>)?hLa&+hFw^Hx58ccBsDc2JOlf6A7c-!s?eoI;XSMy@tJQ~yGIQq7tS
zBFZAkx<8GFM=FQ#3FVsaPFh-8(y4XZQTEn&i<@!l_Wox3haAo}iyq;+=QT%S9ZvFA
zEJNmqwTN{Vd_lJTdu`Qv#PIyM7t?I9Fs42i5uA3p$B!!ku6akys<(5+DUG7Y|Mr~*
zlkJRw#mdPO-qX6xmKa>?#K)?tIS(hJihxtUST-UdCdS0Vl9rOHn-d9uYTn)tp7^x;
zbJnX(sywSY&C!3(QFpoX;785&Pg!pUq2S5N0(?&m4m6s&<sPT+bzt=1!K;LXF^dUU
z1E$Kg`@Ne_j9SYky@kBFc}ohLo7=PErE~I;wdP~BCksB?F&Y;iONfg%f+_B<jNRRR
z@WA7WTja;|9Fr8zWY25%_m!sgAUvECT*Nq%PoMUlo*Yf?E8wj<Hem=qd4E3_o*K9j
zEnw1}Xx5jZmZ!ZFveukUK{7Qx<x^Dn>NV_zz!R-oW^n+&d+x7K4;Q`c*fh?C8(c8n
z_CaH`;37F+@r~9fzMdxVt&a#IKM)pLHRp1HgLK9!MW1IOTmmFPYopeElhU}>F)lb)
zPoh8sfWL0rsABKS;Nbo)+k$iCs&qx|=Q;xd0xAw~g1Y2?<EhQfr+8D}yt!T8XD<O^
zd*JZs7OqJ{BB3$)>`MuwWMNk>4i4-7&q~%u55eOv!O8al-r|hrnzcJf6tHh>YU<d~
zLVxd9c^I7DIW;vE$87?ytc;u~0Tz0N^1Dyc{YC)R&co8k)GloM!$+5$M!?SSKr|HP
zhZ*~8S>HKS*kk;Cd}>ynB+S!-Js=k`00yT!;WFurCA@i4waRYRZVhC@KkH?PlWLKv
z=>)~>j~_n>>7`T+@1oyE_Un7ITw7aPkHp2>E56UUqJRDiz&@_(B;z*jjOWdvJI87y
zx8bzGz9r8UHsh`^f4H}`zcC}+gHIz`^P^fERJNVJ%3QPPxR@A-^XhmMtHzBHiocup
zoz4Dmdtqz7$Jx$~t{C#q_W5w*6*FkgfwUwqCwB#pT*BP{pV^ghZEbFnaOl;%Xu)}N
zgs!1^+!_JWlQ1{(+8r^u{D^|;pRpLY@n|1scgDuvl9laveF^IW@*bGNZ%F<)vvYiW
zyrAxog^ewEKba6&hym()V4`q^xYw^=<C3yFI5;rWnfx<eGZ)_U<4>KB{I9EJtD1dz
zbwPZY@~@|SdTI=XLNzsqH)lIAN5%dwR+*`<r-y@qp}D(T;o}KCuu&RCqb5J@97BW|
zw){4i_vwL&v~(Ld1GvuEzdRzhu&{S@bhN);b@??mD+fnUL&HOmA{EBDKp`1(Pltqp
zf)%C2LoTnR6b`EXkJ5xm>=i}D#W&n<|3Et;NXP*?#?PNmfmGy+i-swM7~F2ZUi#kb
zmS2DowXjR!lPDGthHe}__-p3?HDge}nwpyG>Ly+Y^#y(w0L4Z#7DesbdwckJc)W63
zlD`c662zlN4)mW;C{s{NQ;WFWnzQ*k(5<>oke$}oZJN%p?YTFj{of<5uC9(%J@>xh
z*nh_fMpioyffEYis-G~^-`(?L0~IVc7nd^X96uX0u#(quu^zOpV3A?o*;TXPwN$Qu
zmu?_!)O-^Mq)^PAkh+f_U-?Q|f8{zkK8|L8W^QF=726~N3xWbN13rEwL@yK7@NI1{
zf24F?H9OG##LqGDpwqo#O>}N?H~v0vulI`za?^e5Mi3kTo+y1*?KF`8tXii6>L)M`
zAa~)EuT=fc%tle@$wUfG$=ANF!}<cVvUjDdwqjnS!l(YJtm^ecuQsd9F#!}0G-&rY
z-rs;bF8!F9@fCfV^=kkzhULs}YTHeeL<#13{+Q*|b4gKJK5By2;MCvoW9I%<+WFz1
zn{LXtbw#-)SS3kYh~v<h_H0?wQA+&NWU&izn-+gH|JQ4x;`B3%i{=Ifw7X1~!JH#O
zDbk;#(es7XD!EO;p~=n-26+U@QG?kar^+9RUq{|kHetv;t5jLq!yo3eel;s#^!BbC
zLITCr*VhM2OCw>^yp2-Q!&i)b@mdcC$>J719tuLdr0!aHTA&^=Jt^dA`(v@F(Ak2%
zgqcG3z9dm<$2IW5$0ubiMS%3c#+2@DQQR$EFlm$aY0F=wjaN<#ew4|bwS0Qp!vY<i
zOyfYkVBP!i;nD7juaJb+ZMeZ}Jv!LCE$g+dsgo#MQh)bh8}c(TzMjpciRu1TTv@?{
zAkGIb->N})2=hBjOG`s2An(dRjM8*$x>ze^J!t8+Y!r5-9B%zgKBqJVE?}V9h$)bk
zP~4DB5s$5r0cvjv37?yXpk4<CJgv1K2(NA#AU_?OJu{jc%zEh=@#a&C#+i+zRWE{!
z7I-S*=gxaN<t}b+ZmzCZi?%AzXMv*=byNx$aR2%t_fW*$mDdV|D14aI?lD&@1cKZt
zE9YrN{~{3y2SsoT<lK_tk#W({yDJrwPiWD_S<@R*@(TAbvauhT6GtnVB3SqZlN<8p
z#;Kj%Wlj;hB3C?0IX~&#<C4A9j~S=vDoyHcFBtQa9(WOfk(;}_EW`OA7s-0Iti3dR
zKhsbpuhlm|Clphv+j>%#Pj&)%Lgdt|kUxnN>f?sD?|HjO{DE!Usu%`V6Ey<7yp905
zy}-XbyxAChis)K$9sNFG^y6uXE#8rxe8WMoNXkJHw9ZUJrG<7)5uZ1p5^k+|DJZx3
zYgpUa+EVoVoSQP=jwFZ(#YUnY`!>LS->?4mQmJ~k&1QbaHfHYC@h2K;c6mz3hY=f7
z26diW7q|y$_S0mmqBXMD^y&sIzx#>u-h-}GKC^G!O~qr0vq`kTnv83rYwDg~PwsY*
z<VPd%Lb>F5PfyRLrl$1WXCfdZ)D$inA5_W4uv6;~Bg#6`_TA&cRG^F^UtL^`Wpq?x
zT0`28?&WGVIVC&eugbFXWewyTovzhqEDo;R)C(1XW$X1t=d<?!jPMLQy2^H@6;KJ~
z7DJ)^*JXg4kf8MMM-{I5Wa=9hEW%SfjM<Yi4l1DJNgv5ip;b^RjvOwxmf}h2UCYWM
zqfG6=`Aeg&r@eWB1C!o*%J|yug%D3MJwiY`sK((_3lCS>XZDMq=c8VW?TmbA)Qw##
z84BUFc;+_S#5*S_#d@8pqsV&OxIxOgIbHm-zKweQ>dDp|3FFzPWzZnvTEPM5k40qY
zvLc+zHF<AqD+n+vb6M12BT)KeW6;sfhnMd1SrlhGx51-+?hz^KmAkD~!>m}#UrqF9
zDfJ*i+)no{awy91pO#8hIPo&On-T+rLYbCOK}bwIkow3UEMMJ2L;U$&IH5cCi|_R%
zuLH-X2Ij3(^l0wq+9&K@%FGvWoNpX;M#@ZHc$ff(m>$>Aj}+*QRROZm$;}wxuZ8c7
zpeO;^;=>2JhLB9f6u_=(oYRyM5%j)*uCxV!7iV>T6<IwAio;6VnNLp@vM5-h78m+5
z0iTKr`T5h(SDg0nzFx`3-lR9tYH^RN0adMX>U%Bi&Gr{NvL_nG#zrrheea7OE}=jo
zproQocm1Hy3#w|(*nSPPllWh!0%eynpiADs=w#ECXXoaW#j<*OdggB>57oc_Dwj3V
zZ`-zcd?r6~_T4iuPiom^8@c|@kF5s&`Dy+Kt<h0!nQ4E&`=8J7VNK#O^DMe(7OyRo
zDt&xbKmuavl8RXCtn0C*j1h=l=8R<KWF^TRcHQ={QxEr>H=J|g6?PBZ#JhU%If=-z
zgS7~7hs4ZVU?bdjm&4}(E;9pk;)Q~*6G1fkNB8F1P&D{SH6Qd;i`*H$dY^c#n_PAD
z<V&eLDt(+c@EM2ZjF{*!->su^Ly0+E&)=ed-G@0ovRW=reQ9N-7K#+Kx6cx5eFy?J
z=L2s}ycBoi1yoV0w?OH7u9eO)YB6d#qh$A#Co|~f!JSD)jhCTA$*vv^E_>dh*5)cz
z>Oo#9J{qpQ1`XB%6Fb?xyjdpad3@^*NW3c}C7I2?FjQ{>!UCF`RNf-6y?H`m@*|W|
z$aCd1DoU+FLEjlx6QSB??l``7O75+H@T0>QqQJnCJ@8pVBW8Zx#s(f)&<V<ulkGmm
zhK7dbHu}qW)YrQ)YRPH^1r}t6_Sy!|AnnDuLT)eiIGpe1R2k2{OOiKN)LV77G)#cA
zYCD_iwY797a&*fo)@@4{H|>WU9rlvOPJ>z`5(f4`$Z=5>23SuXbn$gaD(A}MC{|`A
zt(+1@F=9j6MvR;&+ZoEXdeh<EpqkFtcb&1;d1Y_1ds43Jyn(PNV0;K))hk;u^$@20
zbs#iAJ?1!B=a~SwvSt^+mHA;FRy;8MQ88-Ws9B(G5E4I7sw7+!{%JFwbY<w%3vN|e
zjULgL!bhs6G_4mvExB24l=0~k;J+w^oa8EL3H{X|Y`rO;JC=nmpV^rxfA<Zl;EsxB
zr?wj9b;>wiq(Re@Kl9rA^GdbK%%f~iRsbw~$Q*c$Ls5VzCiX+Y57*ER`QupyIuq^8
zd3O=>zSvTNwsSy|!Gk8@j`pkp31>4y2sls@6tA0AJ3Q5j)W~sh44yq(!pJ$N*VPI`
z#9ReHFtWDu)2C0dv9Xnv{HgvDfKLTqIQ&q5S_v_niK1#+X}%S@0U%xkS9I2Z`N_%2
zQ2sNEVDB}6<vv1WA=Zeh1(=vMmn{-TaPEmkodxx$ZXk8bC-4!w38SNjfj})H0g=$}
zwkI<0Bto%lfvkz|su#P$=GG~6zoOcf4VQ!rJ(G0wOu)HqZEYnte%uVomDgSl_JWqH
z2TmpQ9)eY;F^B|pD>>I0x+{R>p3uzr!}M*OhJYP<?FIAR)a)Cw9`t#=-G=|>`7hTk
z0Odv9R<{ry1s_kRi_aGY2l`!t=1(yQoxNE8b+`Srw=p!I#9Z&tD{77UKcrc8d=xPh
z;|lmd&|i7=3gc~vG&bmF;BFglXE&$d-_>V3t(^3m*DEF8wc4CZjTAwu-2&-oGoEX}
zyeurNr!|Bk<YbK;Ab&lV$dB7HXG(9r1f<KxjE;<u9XU|nQfilcs+~9Is-4Gq!azU;
z7G9Ix2|O#A-?kCNq%VJrW)$qa(#|($+!Grg7q6uO?QQbjqjE~=QFd}^yea?Wo(k!x
zgrsC#Y_ClbfAam7i?!ZaB*?aqP1Bukb1D@nKfi5puO8P8Soiiyib)>jG+ly-d|X3#
ze>y)X+Qz1)C{P+x@R&Z|-@tyI1U!<X24S(ynCW~n?&8!O38l{8<#S@kgUe}8Tb-?}
zb~<Nld%8_O3B2`^hXpJ+O@D46M*{?HXBrwBrB-^tL$iTYc4xqR_{<eG4D;4;lhGbe
z6=T!Q{z~&HzOwSWO*RV~fKIMia%I|7w{BH}U=<gfZ1S8AnC_G)LmSg7mdjQ=Xw*J|
zFAu7g=(MG-v@~TEMD0CNFa`6}tHf<gD6}EzEFwp;X%!nb`dmemoEGOUhE7l{vkmmu
zsovl-dOGYamVNS9EDNR9ON!Sh`>CqueZA<Dm|H&o_>$k9o*38J_9*-LZYn^nj)TMK
zeh%6TRGAE;qic#ayL8~Z(D}jG8U9d}j#+YURtjBqwL2h~-05jo%^3h}^jMvzkUS`K
z*<)Wc;D&fT>B<Ndik?5SZa<BwD=j21nn5@>Ar?DcfeJXZ7G=Egqh+mjPY)1=%~W9o
zze@z}BhfZn-8oGqzY9#HR{zrih+}~bLPWU&S{RhS9GZm>mZz}+_lx^PM~c4Q)8VK{
zh+M(*Y-)VZKjfHydzikg0n0cS-Hrl%Se7W3+qZACu)M1S6h;m>pF^mFHX*U)mHK(7
z)H_@LWGTy#6_Jlmatqd+%_>#1Eo~ZZ%A5CpE^^uUAro;&iEI>ANQ5j*oB&A2aih5f
znu@rP7o>RVk)ih7bgJLCuNo=rj%j~b3@|*8fal?M3aCvH(lmi01Q&o)p#x3aTbP=Z
zwz^Dfsz@gJHPp<VKGCX}?6PLkA|ucQB7c(3#l&RV<cC)ZdS>VdQV}{)({%TB^rca_
z|IM+wYR_jYLP||X9Nlp)h4d$8m&w=?G3_lSAfVpf-r331D#;kWL`Wip2pN7v5OIpB
zVl40R;)9gP<39MHV-7>d0<XzgI*FC}uuaYJ29cE-L@53)<nr?YJSuiTi*L<!0g4CA
zwgEsq+$XiJA%SHU$U|t99KXU`|L2zp<b0HEwv6+Q(Je*piz?~?P^ORP^y*-))+6V*
z^9hT={BaCGT!D$e<HWpAGgG{tIEVCr`s4ysTIQBI@>Vs7@}L;>+@JQR9550E70E9S
zCq#*ro&7ov&b#2?{*w=bTq&RsLs!q}IvECmU%h%2G{^GQ^Ws-yN}~^i!D0Hv3AHYb
zSJ{K|A&%enM~TK~wk^g#RWyCNDvQ|J@fJ7+VrO)oMvIToB}7_+G^AiY%1x1i$Ui>B
zV*B9En<lzL$oZr>`(n9GLc_ws09Cie@m(WB6ubw(mx5$rPK9r_?ZHA7-_@dYEmWO*
z4TW7|ttUOc1n*$gp<aaYof%g?Z@<OJz%UNFn@a`tUJH)x?QF#7M*z{aX&=M>Ot`KP
zW)UN~vByP|IsU}${l<Fdhw=>e5Q=G?vEqjgFCA{4vw5=M<j5P<ftDc&%4j>=$Mdf6
z$s2dIp_8B7_6s~+^*akESC>x-YpKRFhEW~|voR^2yIP>*wBsF=!KLq;gz}r+mMTCb
zz~O37hMgH!CS|gI<K=24YDogH<E^ux&5@Or1v)6KJI-0mTfE?eV7dhcYgL<dVBsYW
z=YGqcpjuM@FsUf)I<i>17FsUppu*?EUrDIMbQ6G*qrEln0SXk;)(*fcBwt;Uy{9ol
z=;pIV6J_Ov9XAxT>#}5F3ZCwK&U~b1E*wFjq22B6n>#zi?zbPj{qp4t6bTO@<Ca=}
zje#v|ou;>G@ibuv$#EuZrn)8i;@a0?gMpADe)~B&b@joeYGF*HIzmu)KHz{=EJ94Q
z20Zn1!|Ia{8CZ{K6md76S(1TYS9;2HliQ>_I3U2~WGuM0v-k_TB9Vl&zl~Uu6I}lu
zL!#bPft(bSyBJ0!;c!V}|L6#G745fC2vE#@A=pl%=rGXfvoC+gc~kH<JiHEcAm)Dl
zj7;yKaEwDAO-!T8p^&-VeAfi_9A6q-VglsjL|yh2ZW9F;7Z<Gw%NnC-kg7;f1@Zl&
z!uMZ(PdtIxtDYGOGWuddUvF&TfuFTBmfcO{)R*A_bSNMzxlVb>tmBN@x-nJI0uMKC
zx1E&mtsfNrt*3|blkG_Hp5ueD@~me^hjQlrj@w3n?EbXe97qHRd6-lP%gj+oX=lY7
z+zfOW<2A$I8of^5`*~XE5_#qXNA9T27v1{yv9udcnr|f$iaXG2*;NTMFc|JE4j7#U
za7G`1E_vaDGZO5wCABI7`{mb-aG8h?lM3TQ>ymb^US72NqnQu!R~_lK0BwnlgOd%~
zS|egFf{p6#0m3}Gj{?mZG&(N`RlKRG_~fdPXh=SYfQ4sH>c4Gu{Lo%jr0kE9iU*uc
z)RWl&Li&kC6)6p40W?kFwGI8jXL?ZJ%pW>a`HFrB>seJ!?qE$^HE-1;a~jH^L)d5}
zXYvir7)5IE`iEe(fIXiDgy!V#+g3H>gDE`-8`Fu9dS3BujeLOf?6XzEW^O9t%Cn9|
zbB9gM;ukzqWIg0KR6472T}U<#F=t+;S}ZqTeln7E5lt``8IeDH_)y<tqeaJ;PY^vK
zi0J_$mpK<8hiSMS{xuYKQcw$%gXKVF^`<)c3E@%?Zky9@;T%_vs9Bxz^qU?+y3~Zf
zbqXvAnT4sfy|27Sq`Tk-1qb~@JP?p%4J?AW#>|vRsTx=P4*CNN^5IX;nwL3r)amRm
zN4TX>S3dQezDqklc5nFp84@1Nc=seBck#W;)Fo*ytoI}mkPjbXVk(|Dys-o9U#qgQ
zBzof}Vv4xzak0Wk@UA+#QU#5^jiD4cPH(D!xw(7qX^0Nsx)d%-DH{XY`o@hPKYyy$
zg;B`H5jULkS6e`B<q@BhH2L!cpR0a1Fu2<3jPqNyM!fn8r-k=%cC>ATyJA>Zdd1Mo
zfe;7Kh5lw({P69SD0h4*C1Xr<HqZ-zW|Ucatfz>*Yu@v6UCqEE{)S4Sw>O_JnHkda
z_udb&uvS4N$pDH!@M}j0?$xWz&ZaQ>6E0SuFv8V~MOb^qYQg;z{IxeX+rn_MJDXW~
zy-eEsoyq5tb%ng`e^|Uo1Rt>-0C7?IKY;veZ%fOZf}NdRJ5L>55G%o?@R+a7@KOML
zfkuO6WsZ`&Gt0|t;dVY_g@uJSwzh<xx5TZ1c4v5OEJ!qX$dW1QiNbxOO7~f@C*8jH
zg6N&Da_S#5EG!~(hsHIhd&JJhKi?G<73JksgCRbvW`pYci48pj0ud6@`CEe5n7;&V
zHy}&_`8;dNi(1T!&DZWS@@bZ8RaMn_1s)uH?RN!Ei#p-krWW3OlD)scy(O*@a&w34
z?t+*L!@un|h7k@HK*L|TcFSfPr6k?<j(z!qUEC)Wxya6B+uZ!T{k4J&NWVOD*=DKE
z5H#JGgvm%fbd@^!-lAH@@{=s@r)M-$((Nl8fa7d<gGs!$nONGv;0y36I*ME|-Cp$P
zv=_q(kW+<2KChUY<lk$TSFF;|v~Nzkx0F#{4qLKg*1gYI4rqM}iUdIG-xx_S>|l6&
z-b-_V&8IA%vQ6pETU)no=by6eAV|hf&E|$>WVQ9NS00QsQ|CY>A`=oQ$jRM)ekD{E
ziZqThJg?7<&bZ1G`@h9l^PYUE*Vvga@g%DDqd$?-Mph<EBj31zchY%#p8?&=$jC_e
zXev(!L(s2OwGKkbC;O#@CDPTZ_?^jjRCTDe5{j5rM7h<>V;ws%F$sHAFRi=#(S=J_
z2*xuncQ7cRXZzI6L1p{?^w)5o1TvOtbLu>Iyk+f*55Xm<E&0oPyPjI=t`g1Vk*~~7
z%oWttoScie=YzJAg_*NQS7sQijmy&3F8BsOC4g$H7BroKYOA<9N;?jSFPbi%%tnp&
zqrwh7-TKRqJ}T^322nFm3xxNXG;OTRSZG*o7N{9cD9ZdQh_lrT^c;`ujJCe(>s?KA
z0Z{giV)A%_z8DaN-IjDyS^Qw1Gk1cxdw2*xRWk$#>bQ?6#c@w7>HgvpJMrOYe{*ZP
zo_&lP@rbVy$XZQ8F6+;uSk!z_n;%$IZl4phW^UhL{l|PK)oKTjPvL5-<Ml~={y)av
zhx_`!rO-TYIiQM-I(&w!9NKS&UctvQd8G%E<+{R=Phl0->dke^whiBJV*YIPgWeu3
z8C%x?Uv~-j+BMyB%eV(b$_tUql%N3!$ZCzrDqDS9Mh^*Cd%2o!&afaaWjWWzY8`WO
znfYMu$n(WiI?gL*a)pbn@2^t}qkjDO4s<rJcR9kjFn!TAdk5+syj#@HKzrd~NGBYp
z!QZFvY7X?86zM1kDp#GcnIoBb&~G|Dl*(xz$hL0@D`3=n?51qF@B@+m64%*cF}fPp
z58W9fkMm1O(FcOBk5o@^jBrAdP*TgZ0W$T7ttyk~P-=?9m`Auhqvb7O`Fh`=f);Cv
z_V8Y3tBBiRYWKYWxCA;1z_LB-g$3wkTmT3ccbmL+u65K*C-D)-<k`nMVn8EqPV7Y-
z$?HRer#p-z_t?3BF5`i*@m#iAPAH8S`}S)B<kPnsD5<lK;h&%GKAWV`Ihz#2?|{SZ
zI`z~<<?^k62_^=J_7%{Yv`4YBs?i-{AAP?KPrag!tohmr%R0)k68doZOjF0feP~cW
zhi~WOnlWZ0Fmb9UP(GGf46S}o4+3o>HAw%L>;%#{z73rfh(WGKUZGT%`6v_h#A9XG
zuAp5Dao3v<ucc8uG%_~M*RN0B7qQaAMBwq>3WMrxFg;XL{|`M9EqC*^JMuoXGnNyL
z3jrI^LXG{3(^Zo0<@ESx`!gv&DjX`1Dfj{Bo~u)lPg^alj(mD4=UmWsQASZwA1H$^
zsGrMM(60cJ+m{9xF5PXli+uI!)oe$Mjt<54s3qtzU+IvA@?R1*c*m>hNbj>UR^jgA
za?2R#ssNZIM+^RF;k4}c?pXnpxv{bFtNZix5)j2Sqx;6ga`gA_-xn6vjXX;WfPQ}c
zOYb%aq+1lc=KXb^?oE`d0LuUG!BcJR(YdY!$#b&&?<k+;n#RV2=;*|o&l3Osa(l(3
z*I2QUEC&ZiD*iSQp!{!?W<a9W8g-Bmp0jlQXq{CNsF<%NngN*`pa;zdSP*AAfHS^w
zE<%!T0i1soAZk`d%ZOG9&P%jkYA@o*_;~osw(FNB`{vYuoBVHCcQrE_wrYL}0wt2a
zgRTd7J6bA8Tx8I6g@dC2aK<=ATXvw!2xK)t!kA#^zCP6em|Kp+^A7a+n*-4Y;Fa9l
zc~_u23JAcqx3@2L&j-!T^BL2^nNz-g{TfOo1k{Lr%TQKq;0(XCZ0*x%+H`L<>g#_8
z3pLG19Wbx0tpzQ@m&SLhyhHx%$yPa9EeDgv2lTJ95CeLwUshLPy2-|K_~y+;x+8=y
zAHdcO#gHG5qnM50;3R@(YP#~mcM3k4Rl|{C^d1tn^Ejl+8ybH4^xJoS&j{KT&s*!h
z2c4*KZE<EK_`3!e5ZVF>VOK2Y+pjK{-~8Sw(6X<>F(Lnc-nu*ARv%Obn|hx1K!JWe
z&=7}{hdubc-#)0h5Fmf_1i2ybT*-EBWwg#C=K*@Tpno|CxVW37_@D*!%b;CLo?Bgz
zOIX(}M@yMUHn4$G;rGtz(K9nId<~}u=z?C9`Zf9#0~bIWB40PoPd#U6Z%@C(I7I2Z
z9sApzwpds-@+m<l_?H0j_fJ)_L3s;0h-|lSKlu3f73f0UI?l|rY!uW#pFcCvpUdBV
zuTW|dC_S#=Q!h_X(~ke$3<(HFmFt!<Xb%7jfb#m!0$J@dQq^-na|dEIYWU(sqzcZT
zeRb6X0O+$4jGV{x_V)jV^2w~Ucd>fh+sUXCI?;(Z-HM+Z#K#>>`8JPd>+wk3VXG5p
z^3YV3n3&k5VCmc7RShl<^d2w`{WG`cr1h7`)aSg{5)dUp`f^tQk)K)R?a`;Gv5@e%
z5)KE-Ika}amb33^&H%tuK;Ibt&CrkR1E6xAHcd<<gK~6Ts-UWD&Hp$I<{hYA%(K~>
zMOCYab(!jG8kG8sMgk(3nwkpG^KXeGunD%Tj^<O<^!Nk;5T`q<&Wd>-oh5S5W7p<U
zBk<a6H45}Z?B{|(rQXm=|I4Z9+3b5xk!uC9aHhlUuHYA;B}-NO=8u2$7wMn6tYqzW
zrSORc0BN8j=<5S%V2B{P8;CxI7jybe4qh|L7kJE274ph-*>WANrnZ&u$R5lk&{y#&
zED<6-b_O(v?#O-CudkzRJAWTqp7%x2EiP}qIqK+czQ^vHES;1QG2Pfa^{>3@QeyM&
zapI`}mfnY3oJ^_%2|G}kpDt<S>&nQ<3481pPWv<XXhg{nJ#%;o^k};eeA55u(VDfT
z3u-G?B53uGH@62O4uO8zy5Amj!MC>^lII7^9f4;3GK57VpDm_ClE$x>0?>>@XL&%5
z0s=JftPm;sL+n6H)ofK6oF)Q?rt$F_yV-U$RK*LCdTC;E7u_tPMUjfX0DP1uEcp?<
zKiyi2Bc`KFo2A~<<zZG_>jy?zsW9qiIqLVC!lZ8lRX)I$Hsm2v>@h*W(ovC-g_Ovz
zE)%gIHvdt;Z$2R-BMS`;?f0fRLLp!=b)RcI4mN>U2h?w*ICXD5rRwDl0nQ#D*Ild+
z>q57e>ImN$tUAKJRogE<*8LIE-+3{;vmJqi$@e@w8H4MP|D5uZJn_Dj6f>IgBpnAE
zd$`PkRQ_2q&?UXQKla1XCk(nju0kt9?z}VUaj2Kt**ej!9N{T)$I>!~WX(H9mk!);
z2<=jX{E{Scp*hK4$<8Oe_Gl%>J;CzSqUm|ey!YqN%6y@2CR)p})>t~YA{U!p*bH<5
zQX0@=r7jyUZ2cK7qch%24%EkbdxT72K&Slk=J%|(l6iy4gZ?G&T)N}4?LFtxDAJY5
z0CNlCUHf_SvLE;fP+umLQGzB{KfjpOx({{oAE{VSk&gIYf6_q$y<-<}7NA*)KP2<i
z(@TbCSKCZC0ug8LC_ze;4ClFUQQ{;RTpzhF1S1uNRhZXLeO<-RRrEco8Ozc-5p<*|
zU*t8|4UL@Br$Qb)#5;ug_w?67+4X8%K>h{nWuORUDyW_NpC9W%H1)2|eo|#Q0ve(w
ze6e1^B&?5DJXOrYi%J&n0Fq#_Kg+?{_4j(w`JVwU;JDZiGQ8OS)Fp8H1+diUg*7xa
zps~Rtp#KMCd6?KK)9!=I(t8K*&?<p*!At-Y<ZOu9cLrx?XYf7#-%(152AL@2%q}3{
z3i==~@SD+r0-Zr6_OuF>vyP9qdSEaZa(uj*ItXaXe?{U|lyP^W08qaGl@OaDIq1W@
zLkp3aLZLTq+yL!Sh}W@8Xo%d~moPcB2neWn@=O|VeCYt6dUkpO*Dro=W@8x{2$&oo
z{gSe901554JWg#-mV^rr+|c<gVq$?M0QDuhJHI?!)T3uW+c!0Qe)|YW-lQhA3dK{w
z#S||RKnDM3NPUN`7E*n=6wwncQ_u#@Xed&;0~jHocazd`*c!902k<SFjC(G9Nb4&P
zKi~NkCJBfS!hr;wW3(1;ed^)K<cmLV0}8s}4Ak6Kk5~%lOTVC?pi}FjKQA=h_E922
zCjL^%0TL_ylpQzj3b+=6CO?7lO|bjTFQJ^*>A}24?Q7rvZ48<;6--O`C6<c@dgi~2
zK#xg)Y;J2KB_Ro=6xifl`_IyimrWY%syt4Q4+;PX)BLjrAXekvMHvG@8sJqxE6M9*
z?C(RMFAElg2Zn`lFf)^JzG$YH{Pt&OfCjG9rRNP0%z+-aJ5VAP*Tns2o--j(Kxc4T
zPYB<=oB8iUAo}Zk3`+p5F!Qoe=Ly%$;5Cy!`v$CJ9Bk46M8LQtZ2UliJ|fom=bb>k
z7RxCG0b1G=F>fz200aMHdZ2Pb2P+koKEQs|4rKma2(T2{kd}>}-o%fVAaGabu$;#5
z&ntncIxEu>Kw`(lCE*totXskQYh&~lFu{RT76{D9`E8yHh<5y$4tfhTyWaqO_vq21
z4j0!*(ZACHg0o;o_*HD|SctV=)!9zTzs4t%bU@d{Iyz&ZCXRH+N1q4^BhV;9i{n0d
zlO+DHs2zT{?DL;ZU66+S(&v?yl;nI>_`5<dPX=lv`tT_zD1Z)W!IanEhhX_K2W<R8
zLhd8QM#SJA%HK!8{t<Q#;C55J<w!9O=v(}q3lLL&2!Z}B6zSYw{@F8NraN6FQ$Xd`
z+uI9ztI9(E*-yE6L`Z$TH>m&e^6q6({XM<FOTk5%z|c@9d;6t{TA>Edzt2FsAYxR@
z2GqkYF8mF>QvUy0d6p#lHj?50-R3fqzwaYm8Tt1cAcDj?biV=OIeL2fKwaweu>TCU
zAOQhT$H2e<h|a~A6aPE|t|>={=mLtj#lgHWa3ksUZqL8wmcub=0C$qWeV4W{8q!sQ
zzncZrw=%S7dlRsiBL1GdYg{l5QNdb^zI<39M41~ml8aT^IgGDX%8SQHlN&3J_PX^`
z9N$Lp0dmLO3m$t&XFPHq)M#1GM`>mzCL0?Yf^7n2a1|4d^4q_tn}wL=1N-v=UxE_o
zuDg;xPuDg@zRe6iniTryY9u>V>B`W(dfz?AwA`T}cap^9<T`^UKX79y6~Fn<a=;bC
z0gd{x_e<{k+sk))Xv}~5tGFGNg$c$_I$t!}N(UDcEG#UFjJJWqO;1l>i7o#*JF6r%
zFgyJFR;7{{Y-pCRvm1rkgEE>UF=p`<Kl$F_Ayf2$n;W*qdXZ2AX*pKj)zkz7g~Ann
z65I|Y5IE6VW^7Ry#7fZ2PkOI`svwMNq^Tq{C`mv+F(%oA6S_8ieAS&ETlnIm7tMmW
z3bL|erDhQqKtBF+xE`S8+jBl^9DHFT#GcDxFXy>&W#w`cvt3IMpLbMoJ~P+E@p5=e
z#0m$MYhXaY{oy#~Kk@bXdo{=%h2%#kBK~#j<OR%1S!E>szL2zBI5-<YnVFd}F)<CH
zUmNXif}8@>ESPA4e-+ti)=409Mc9$5drNV`Ok3C*vU)F_R7F`ip3f>9VG4k$lamu@
z<LK7|-SH>otWxbf@W1l%49A(nouz806_z6!*%rt+5*hU*-5`|l2e`f%NPDK*G74<`
zSXP$E^VS>aO%f7NE2*`gnd0K$fFOWrUdfd_k2$)tM4W!&C^A`Q^ipF>qlWhG-Mfs8
zv+1O2ENV%Dj$lmJWDS-m;1~sym2NxP=>B^h5M%(a!k_@&)!Z!M?C_SV6!dNQ`1q`k
z`)*gJs<N!Ctl-(R5kiVfN|KV202KX5F*i2{ejOk@x*0-KdNHxd@92**-3($9FsA4C
zCewJ`w2cf6L6|%*90&}6Xgt7|gGGx;<(Ks6%+u1#OSElWqxg#1Hrq2YT#VbizkHPz
z&ic~5)Md4OI7ukqDPyBu)0QKa>lR5g44v8m|MVrQ$R`loplo_>Ab<n1A(}jEpg8f?
zXsxqhGA{^D7hUR46|EWs_x5ywB0!f+_Sf}7bfTVW*F<Zf!L2tm4(MP7H^8O!AG32Y
z3mdd!Xu53Q0OJ8+0s0Pd#Mh<<-`2(xcLCsyR`0aH<uGtF{I{f>?SCL+cfx%wJtIR~
zM<)ne9cE-?1iZ#$P&tSCas|D6$Hm2E4(>nz<R29kwKMHcqmUx{Z25J-qujBZDjg#u
zuDi=4$;rv!plSk@@w=cPB5<+#WojqT8Ci{%nt?_R`q5aq6}V~r0^C*lf4Zs;uyUgu
z`hz;#Ff(p#uBRkvH~gv3wzEg(A(9Kb3$)UO%IQzI$c*s2-LycVl0@YRiSr`EHQE#g
zH33<bJL3+%y%+KMJ$Qt7`xT5Y{Qsvr@u>QBMYi^Fx%7}Ak-Q<3o5(Y($<$^Ztunq#
z(%WWFWtOMkXLi@vZTDs!8RH#YTN!E&AG{jQauW6yO982(W;Ws6{KdFGI;)9v*3e=(
z(MrqG&Ec%;om4^{7Qz!B0|PSw{dWPJ?K;?l*dqhL=6+`P-&tBm3@#mPM$%jobWXIT
z3oFi9Ee-Qd9QS%WT(vE2tG~NI?m(APeCSZhJQ->Z;iBWLW)guy*Z3m5<c8jfFkHxr
zfJyO?6|?)E&U-|^!}`I4Umo^Z?OVpfUb3j+M}$6`m5N=)B{zaxK@sMx?qElG_^e!g
z`C<5{x>GfwiVgP^b634B(<Y4hQ;O4!Y|*3-9kll>Jhx><d$wwJzm+?)Iv*V7OD#XV
zDtRz8RQoK?O3>#{RtxnzCIBB4l4=mnNjZ~0uk=qy$XN9m`kZ}Mj|N<EM)#_kQH!K%
z5`5itz6WX6S04ilo{<|r+O}|+U6za*etYAUOj6xzBB2pG%C>LI$Z^X^$rjb*DjVI6
zmRT<>K{*-~-%XvTwd5cfXD{V=l=EuWm8KBYo@H1HV&-n{7EkdjBsnn{&e9~cU3W;$
z(zbGvicSy8$qSz0?`|F$sA8fod0$6jIw#<2^pxW+S({l(GM>(+Y=}}5xw3!&vv4)j
zZLcr!BP*7shZ=`Yn#X!?*hG(%<(5M{@;Ymvdk8D8Mh~w|oUkPW@4I@-uL)a7CxV3>
zI8{30)R*|vG(2TPXY4twl2*9)3TyZBpV17YSu}px8sDm*nHA8JwQ{?sI9lBcaWO%B
z{VdYZ#cYY?F%ekRcY3RA#YVQH??srTgu@9I@g@Dlq<C}(DRJ}D`nT2h0ht43MXaRk
z6E=%7PL0v&2?{3PIf>c6c{)G@1OyJC+I4<2krm~^)?$`9>1PO5=|^tLyh-z5$t_(~
z{H(GK_LOaAC0r@mRh6Sk>hx`Z^N>N#BddCfa}n<BDL6FRL5+MpW~-L_cBC32CTqpM
z`kv0TYf8tX=|UXsE!0udh<#@Z%{v1}hh0lKW9&7@^{x>rykJl(HaWw@vb-`(DzPU<
z3Q3~zV+$$zrXSQjYBYCRz7f$3Ku8@`FNTk}J3{&An-?0-@XBVm6_aX^i^%eu>=Cnl
z&ijh8ZfEOjDHVm_p5hx$+r>rhHXfA@+r-vUU1M{DmljqEGx#fzFDEG9)H=%V&>FhU
z(f01`GI9CMc4%IC<-s+J?H-<n-080zcia)B(aGM&eP{5_cJdV@Y`r%F7g^qNZ`E8;
z$9bxxqF%=&)p)Z~{p`3aZt&K69EW43$(d5ikxOSD(cTf8V11yANFPAG0U}ofzsW5x
zx|9_tKJ%X9&o6ucT|${ccR6rl?j&hf_H7f%8;?1B-G_!Mp0Q_(*sW^NRM>u8w+>gi
z$&$NFm3w+*;b1v6kU`R-ui0*RM*&Yr`PmOvgY&<^A$Y?O1H*;!{}guqU;C>aZa?Rv
z!_EwKiv4EzXQOy&<VI~^63es~R58d!UL`-#qpzOO7q%aoWOU}coTpUJT@zR}{OICm
z44^E35q9B$o9a;X4!LWF0#eAAD0D4Fmx!D0zA^a2A?YfK>lTOIBj3CeQzjebmWwb{
zvG1^8q)%@V?Ve>jb_-YT*%b}Hyzu!pkJ~F{rW?_hQ@l)u3OW{7uW38W-rKdoV7VMZ
z{UB`fd;0>`*uA8_vIh4K(TF^piR5fz_s1B-uY#$kj66<}TNV3KdU%wEl$lTJ52gJ|
vG&;<qU*O|@togM?=jrPIn?Gc7hBfFfJS<lGst^1%6$}|k1&Ms|=WqTm==sv%

literal 7293
zcmdUUc{r5s+qW(2$i7#IY>D|=qU^gC3}!6JmJljymJws$BN7!^!?-CEO3D~nl59nT
zk{MB{n~bH~G8&%S@An+P=l$pT=Y9Wr|G1CizOL&yuk&-B_jN9x>&`^kTJdsAax*b8
z@t(0ZcVJ>--r9dSS>c)d4L%ztCYEK?d9(#QWM*b&VPRooV>@u*06RN72L}fyCnpyd
z7dJOI4-XFlf#BukJ$UdSA0MB9fPkQ&ppcM|u&}U*h={1DsF;|TxVX54goKoo)S*L%
z4j(>z<j4_eX=xc58ChA`qeqX*%gY}-c1%G*K}ktTSy@>{MMYIr^~8x2YHDig>gq@&
z@}GbH(a_M))YR0{($dz}K6&z_j*gD5uI{N*r%s<ft*57_udi=lU|?uyXk=t$Y;0^|
zVq$7)YG!6;Zf<U2VPR=$X=P<)ZEbz#%$c)i&z?JX?)>@lHa0f4wzhV5b|@6e-rnB9
z!NJkd5sgM;Fc>E%Cue787Z(>-SJw*{F1We5UA%bF-QC^8!^6|l)62`t+uQrnrAt0O
zK9?_F_VxAk^Yiof_rG%GN<ctBU|?WSP|($@SFc^WcK!PG;NakpkdV;O(6F$u@bK`6
zh=?0EZbU{#Mny$MM@PrR#9*=5*x1;cH*dzp#l^?R<8ZiygoMPz#H6I8<mBX(l$6xe
z)U>p;^z`(MjEv08%&e@e?Ck8EoSa*?Zsq3Y=H=zxzI{7CKfj=$;Le>pg@uK8@7^sc
zDk?56zIX55{rmSHJa|x2Qc_x4`taeyva+)B^74v`ipt8$s;a8$>gq?29@W&;)YjJ4
z)z#tg`1<;K000^q8X6lLA3uKl<jIq!rX~V`@bu}^=H}*>mX_Am)@RS2J%9eZt*!0F
zix=(f?Jr-xeD&&8M@I*dNbKzFB#}s6U0ttVzwYkt?&<02?d|RB>wEL&O@Dv?+qZ8A
z1_s`}dp9^ZI5adgJUl!yGBP?k`u_d<4<9~^jg5_ukAM95kxV8}OiWBpPEJitefsq2
z^XJdg)6-wRe3_Y<p-?DPDs^^t_UqTLb8~a^^Yh=nef$3X`;Q+#78VwM{`~pt*DnwR
z7Z(?omX?11{=K}sOrz0OR#sM5SJ&3o*4Nkn{P_bx(8k8b=H}+s*4Fm+Hl0r2+1dH~
z_b-FN*xlXT+uIxFpMMLVc&-p@*Kj5#9>M*Gd2A;BFB6l5_8D^%bad(OhcP#WF)!2R
zk!F_d9+n+POM1F4zP-zqdYjGUWGKIhi6FmKBIi-0?m;CLH3?q9*KrbxM`ctFR<KlM
zSQ=|(DrH}7eyhSi@t*6S*r=DguYMhKD+iWO*O36ng;Lr!IWGbLJK_p-7^q(n&V5%h
z3VK({n7TPsJ?sh57O&?-4rDtE|9>`kO5E}^Nbv4UaINqz^JMN(QDN!<3B;`yOoF!}
zQ;TsAFv61N-oYvE!~uBNqoAUcfhxM7`qbenu>R{K`r6RQu+M`V?-EoIUv0n5_4AJ2
zXm{VP7xG%>$Jp^IlG{`^aVxFq?Ph2L?NYQfX>8BiEa|{zZirmiYRPn8O>NzcpjW&<
z4%$N>S55<)-&Ehe^WY<k0cDb%cC*5dTV^Q&*YBwrsz2ZL8!<PkrmUbvQR+|bs8%CH
z#V!`sttrj+0vBr|2Ol6eC~pVvt=2}COl$qsN3GrpxdyE-=?}Y^5KEQ4C432`3E5*E
z{Y;8ZH{J|gAAA!$K$VPfB@KMMp8g=sxm!&7@~6cmpZp8lLg%(-+JE=W+s7FMc$V<Y
zN1#0kM%F*spU<u0J99DP=hH$@1GsjJmWg;%oeHpe?&i^NH2TMTs5o0^2874@oW~p9
zxbq3=VK@7NHs+ZJ=qNOUd+Q0mkaa{B4vh4HQj0t*NHo;~%}^SzilStnu!r85J21vL
zl)F*TScqCp>89|$pDj>~4E+a4v_WD8fQwSVhyxyiw#P!n3;;?MfdEM7pA$<16!{&X
zY=8T-O0cFv=2I32QX7vw*WHbepW<uv#$#ESF{BlH5Q!$T3Co`{{e#D5cOx)Hg%{wr
z!zqpow>-_1N-%XHI?IO=o&&VWpZA0)ZqEtuBo~__gH(Q*+l<UJ@|2P*u&)+rK#onn
z>dRwB7UHa->u8Agpdi!mKWTv!$14bcQh;=Tr6ZeRG+b;BjG$3udL?+^$CY5K%p;g!
zdlx(y#Qysd-@bG;-T8;PKpG%cBsX%pYBLd5-B=GVFXt*CFm^H-pl7MnzQw46o9V5G
z`1g&HZ0PixE&B;OiOm$hyS-3_fT0Z}N*z|hqKOvXyg-x0isQeHMjO$#@h~=)W>UGF
zMIO+m)MyK7v+|is!yXIi0Lx>UWe^zkLp!A^|B_`UzuMVFz;F~pUVsZ7EP^nw93mqq
z+U~VDnt06H5MDo&!UY4@y#U>BVo8O4HdpsL#QD;5rv3FY&zZP`Ch#oRp+<U+P8W=k
z<>|VDRrbHk_T|3!(;hml>jbqdu)@Ac^-gce`x#dOp5P4MG&tJh+shCxXh1=h`nyGl
zVgVz)WqSH(|6c8S=X+i}mNmZP(`uED<KOrB(g%rQ*U!93*q@FA9OD0py5`yv;=uUc
z_jz?n<#DU;Pj)oXFpqd``EzqE9Gtd6tx-N30n>VE*Nw@G%?x#x49fY{<2WiEn4Hde
zRe+bkG+Ftsg${r7?H%r3#CL{-*Khp3<H^5h4`GKcEbTb6lR1<xh`nUTH1$09?(p{#
z9PeA`o1k8#%-Hou*!6jX-@9dt?Afz<)ZRI+Y8qRIZqDyYVZ@jeBMCXaAJ0_Gok2nJ
zd?&bW7Fs~Zup9s$n>5@LGuow<pRGQ+^hLkcwGJ_tpJfGAq7}&7n6zqoXivENb?106
z1vZk7^6PMpI>%0K<G#pA9yg3&1q@UEPDhU>B)#qHZ~B9qy3PNR8AB~xdo&KU`Jxy<
zB17wc1<ZWD&NGt$1izyiU3qaL8Pv8LNorhzzOjm6sBD$yc}-o+JjWMr)PC^Sz*G3q
zaXz)`L4WTH13D}iYDUL&%xHoj+Q8by@k2Fu<F+H@=&Z#!glS@LuZCA7(FVpY0Zs2Y
zbb&HsOEPuO`==zD$bI$2olS(eG1=;{(}U<-mvlh!wjHz?Xa2;DJUB{!1taD{#}#_l
z42VTcRDoQV9T~lO5d7x5Fi2V}W0u3y@!e_)ARH!5$rpigf&<O~@VFlKX1$DA6Hiy_
z&We4BcwPYuIxt>5vvfm28tW3PM}!iM$(rdRijx15(+5a9;q%w<*ceyM{7pm^3uaW@
z>0^u1kQ1CjQ~^@9uJ(B!6x6&n5ZB?)e+-YUB~*PYAWYjsY+?9#M9fB;LIwE!GQ@ya
z;VhCy=<Z#qvLY|aB!m4|Gg{&FZ|P(+Z@|igf<AgL2+74Bf&;Y^bR^|=@Bto%MK%it
zto$gb(A&>E512HGAX8T#-zUNfpZ+nSXfm;2>_j5fYQCV0tO4~pVp?t%K$@2dcdf#A
zg16#N>^8)FF{_hXnqfPD*`pv}fgaDldT|Ti#R7fxs8agbbLr;{$mPqvwtu?uyFGM`
z=ZCe7&<5l@tsCYvGhuR2(BD0^N=cv;EY|OhWlvzswi6F^n7%)j9EcADjP71`g5^BM
z13hTBKl=&Hgc1V*>22b+syDjn#5JnX$e%qgW?@IhoNI?a(^<lnq4KL|Htx%{l0Xs2
zQ*b5acn&B=SM`Ln$f##%+#;WrYZ5*_j#0;vcV8MP1gY^kn}~!}vAwab)SU)xTIMr3
zMX?9cKmqv6Ts;@W#nny+Pw#d9CZ^{^hX3klL?0vb)d&)iz@|OMYsCz^G&0|hbzduy
z%A{qR0=UIc*nY*ea^F~kubnr3)I0L=a?Moaf_ZiC1JClq;Slm&8ynl&IQFETp%Hi?
z^*YLOKW)GF<Eo<1&G~bD7j2YSQzsf+_W*8fTaO#kL|>XnkV)I}Gzk}D6m<Azpvqn~
z<%r_^OTNrAKO;p}Q)*NIBg+$5*c$=YHYRpkNDr6QD7QYj20sZGS`@J-M!j{StD3;8
z>aFo%tB%eje!u>sa|r${{O&E>T3O<;kB7;@r5cwmOhsjU`(HCe%QD?enL~UGP-C)P
z@{$<D-}#Z{Z_GuG4gS<ua4Z9C+Sz@{WEp$+C!@xLi%i-;z<5(gTf<^)?C+<m^;^Q%
z@A*B{$gl#qtspC`v!*XsINEcuwOf*5@aDfekKRl-5mel;jKT+=YOIZ(&dZo3nq~f^
zT|*6K_O4&qMx~T6iZK`-J=&#_)JHw}tI<i<byM=^YGO9>sx{|~3<=Ep(xre+MB`HJ
zy2#Gp@2BnV7Xz^u+MZ9F>R30?A0)nVZAr0@Q>2#5K98E1>ATY+b3E2y+xV^rr0%y`
zmYB+x?MVqUwWLcI+4J6YhthXi>@ecwAdZF6dwZHQr~!<pW(Z$w5fYeum%CjdsnQOX
zm-olA7YVAykJS8rQlA=#{{m2PK3YN|mXNcizP;=-xG??|wt+=x^|+cLg7stiLKz%w
z*<7M~7Sq2*KbmTz{m3+$*o;=HeO%KcE5Av2<`Q`lq0jFQO9c<Cm>qixws?FqU>)QV
zr3Q=qKGk&DiP5Q4_L55`2b}l0u-7gMRE`0Znc}s2RY7kwNs)<+3c!?abJHZIn94pC
za86HCXOAfh!<AduNXK{VOX?#aYO_-Jdx-$tZA9MO@aK6ldhcZBCHx}*cVR^Rc%j!S
z?vJ~`D_vVTii{ZhDGX^{fldv!hJD9D4PE+rlCr3rcD5cHCEat!>^+Qv8f@S@^07f|
zBs%31z7W8HHWW=l=F4h&&-#(*$foDnZ+bL<_<k?W4?K^sqVj0sE1EhI4nxB^H+)1s
z3bMPrUY2;xJf7#S>{|V9e~GubdI89H1v4miN70BgJpyPt%e)_QldYeMgB)METSbLl
z{e?D)Wri)Zp3w`)ooM}n`-;P-y{n}McbTi-2Ah;Vf3!co6%14s)OW7Y=R>?<Qyu<l
zhuiWUveZUvkd2x%HW62kxpesVS<+TxBH|0qLDT7GyCm5haB+*H4^@x$ynmBrZI9O*
z?b#nEY*ypI=%KBYB@Rz{bZBBoG2%;2oZa;l`rP4qde%3VDNg|ubP2ic4qJX09e0At
z%Yj%=hP#pq@OW%_8V`?5DY$8mxN8aVseLzKJ!sx(nkw6jH{3hfMUa6#u_U5`$Tx?4
zi6dR9WEexw#ma8}Gw8i899)5T6rP~GQw{;64Mq>^k^r6XRBl0Kw5K}O0wcZ6Eh6uL
zkN+G-(VR7}*m}1dq29JGmkfqmv8HL8ij$2_k3P}7q<a?LT2k5d40)5-=v(~6{gngn
zU1P}6X{C%=C#jFdX{?RDNe&ig6`x@9MoZ870F~Vhomx*bZ{I~Tb~9lX-quztsnP)|
z;@-xxAA(P!&)P?+;zu2J+UN*!6G;fE3d>QenhEJnh42XuAr~v8my(>%kfWxmpPgvz
z%nHQM0_E0G=5kW6=R9)_cJfl<S@CIEkLgXz><Br2OX%|lmCZ5do}0h6zo}GWaQXW7
zuL^)XYo4&X?*Ti{fmrrZE>V(UdysRo|8i|VJtOlL?~F_{I61>AUYiWZ*Y-!->hMoR
zHs>ifehk(4kEC4ZcY*pc)m|VC;DehUveURsufGgPlZrMGFN9R%CmOzfiNLPN6XqUP
z(A6`K`&bFoX3GGy&(F>c1@|y+jO=`N39BIhlOn_dfZC$Cwhd-akf{*|Z1bUgb|O0$
zHtPR(!HeM#P_@OLLzOO{3_+J`ei%&%D0{~O_T4D(Eqpz9N#dEjJC0%Bm5FKZs?xtp
zDtwxA+78pf5?raW?8+wQmm&iMcU5c&62<(uuxQ-mLWln>=TR}wvQO<X=UZ$gdgU-f
z)pYH&$xEmed?)k6{}8p_CSy~eQ>mRzZAe@KO$EN}y@va&<l6u#-2Gczp-kB7YNC<<
zzcOPLXmlSTF3Jw4M)^(DZ-LU0oq3(M>~l7DixZJ<KiQcvp7*XO!KVo)Q~CYb<0mss
zud)^WO`JBcYl$iqpcM}Gey*&`%{8N$2a1}Rf%s^^Q(FY7stVWhFly|A{$qR^cO88_
z(|Zx+0L>Ty41Y2TK0)>Ie=I4QA^flBta|^vsJ0ri@#&BEkH=pDJRzMf0w<i2HbGs#
zc)S;oCj+Z*3ZR&N4pQQ&rQ4UlY2KbK+vwj_;9<3-N)c$ZrRU{ho?_3qc4s6o^-eG!
zDVWo=e)r-f<%=$t;c1c1(?HWBNtHs->IYoddN@U?0q1@jP3t-GikyLHB`^M@6#)N}
z;YF;W1&4QEyNBQm_AGhbe`xRib-Xd$Tb!Ut=pLMIQW!RaXd`S0%~T}dZz4`EKdT{&
z#b8Dgkbrce)gO6oeEh0v7xE6=++B5!GbtFLNdXjjxVh-vgukKnBWFgp6?4ydI0+D3
zp<0;B=-lE5r0J~i?!p5YlGjNu@dGf=bHm`qI$)Mj8`=pH+pc~!IM9Fe8Q4<Jgs6ra
zFRyA4_siynCmwr8*BzqBy2aSNer?@I#~B<Dx8wrKx=xa%7u|WqEO~%3DT-|G&dYpD
z>jwH{89Sny9TvX>=KZPcj8uS;**e7ySGaf7&4$Qh7{~7?cq8_a+cN|{%h5x;RV%=D
zl;#y3nzHT@Wz62##V(`*%-2rhcUO23a5Gu*rXHqg;>#U7)=XTOz3%V??;KcO)tQkf
zC4p;2K?<R@`^By|mfKY3A-FKif$-S}@D$wK^x1uFN-pWm1gt)kSZI*_iJao1`=+16
zkn-GNK0B^!2Xj1SyZLIQLw)@6xdd<fMA&O6D0XMOELZ2eOGE{Dt#@WqhNjG!Vh-Sb
zV7drO`DRdJ&{ILGSbBXt&5O8Lp^;{hjR2_R?*+LHT8|5+#$DkuXYyg3zkTb`{XQwT
zg(Q*SyI^@Py`LW@9)m%oH1eJsFQ&r+ph`T3KS1T*a@x_jddJeo*r;-|jyf6f{A|zI
zPZc~D?DLj$ofvn5S<;x{min2HbYb8HC@bA;QqLOFtJ<ga2~}x}!L;qR@>u`a88s_a
z5n^6wdGdS_ve>Rx#ZX<ec5BYGOT9@c#AjoWKj1)uP`DVYQ}pvvUt#pBe6BqsPlq)F
zW)N1Hl&A@1D+r@ZS}99PFjwpRKRPGJTu?ez3W?y9$OZp%TVtxDYYt8Cu^h#s<CGq=
zh()Zn^3BX5$j-IKqO}L}v-3QE?Ga>UBa$`IdvA|+A(5B=hieHc=7*(#ZM>|AWHWYh
zP+92#*H4QB9Dop9`f{rL%~qUe^;_AH2(`1^WT`nOGEU<V5`hU8+y$DfPxr-gcOPv9
z{V(_UR)RJ??hst@44TczB`IzD74L^0bESX_#23#L=#M74>NdcY&PA^p9PW1BFLEm#
zv0nw8pz-1WLZl_RB((utzrM-H2K+5i<m#W*7AUfv+wxulJU&mxJrv!_fZwdi#T%#l
z7oXIaSd*!wqtUFG(VPr`8sa~50&Sql0(5E#SV$YN0K1)eU~m<9U%(FXymjrxEm%|+
zE=QiTp>1Qbr67&*;)5a{>u2uBaNg9~MZmjw0i=6BoA+#M%#mcft;T^mc)8bsu`vH2
zJ{Oi&B8zRBe=w6R1p6QJx+cOi@?_kDBQUof%`pblmt8N@4xhbc4D2l~^s2y6=J4^u
z=3$3{G6}eMD7@NN+@uC~A4{hn-rZMg&rsa__1J=}DVG8A2+5T`5UDihC+BUOMrp$m
z|DZ|OtX(X;iz)ty1z#RNU^jh*zzl2Rgmdoj6*$Z#n9(>xfwiy6Z2ZPheF;LR61)%R
zj$#`fN*dt*W)+Syx8!+@(L0wvFaZsD2^JfQUq24Zr8n$eJlhUMRmmT8W@K~_O6;K_
zT{nm}%t<;7E2zfcRJbgHMV@;G6j(xXowo?Au+Vgq)|pCrRL1e*^dxxqjpH^C7SIJE
zM!&3?;H3Z$NRE{EVjtE1=mL~|BopS8{~i^w2or0j;%m;d03?V*NDs#O4<q)bz-k)%
zIn3UMkKcwVS<~m`R>u^*<dhHtu~kynT=7BUJB7>Vb<v}0o1V~0_YqbRD(9q=(RQ6y
zutl|R|HO&y{Nms+o8OU@RRJ${9j;!I9v_T(f#<g4n(fUx?qZ2C6>i!YIxN*yED&|&
z9s<z0`QxC#eWkK?AJz?xxdhi}v~<!}&8Vvt`<I~D9{*+j{LH!gdM9COyn<!v|M~A%
z-QOa*8E>aN^x^BwI^t9TFMJ7#$Fd#VzcPYvLOu+z&R*fTcR;e}3m39+nn(1HSM2I|
zBy$>cV2Fix+BaH|k6rWx3!;HFyG<UEt=KhJ|9Xy#5bZ~ZKG#i%W>GDZJe5`6$$`Ae
z<RyA#stoCWwaRZmu#0ed3CF^*{5|#oJGR#GEmHlPkS@ll1LQ4l#{4hW)2U{wz9FN5
zkBJ*=-)4~ZRK-?FzDOs7u9$76zq#a|%B{?&`J4arXAX95-S88;tz`EqZ!q0A$vkmn
zknc-WBlAwS=R?GX?@grjc$PYI`-^DX(%SElY1U)L3*T8+^d=I1_mwZd<XIY2B31-Q
z=_Q$;uqJCchu>Cu9}_^JK4o80qv&moYu@&J15Pv;hc%CB4<BaTjFQC1yvFq>O}YL}
zGGYX52i?)Sr`{u0D^kZeC3f`uc=LFm^y+ZMZWyPIx((}2u6EzD)!!-66Kb~t^7(|E
zX4ZED@5%peh_)G|u?e&5s?X9~SP$ibU*qGQBbpiy$iEx+OIFlomFJG|h`$`&8w9Iw
z6qiQQ5<W>%<MluC9FbZ0ZdNcSpqS8ioLjU>$)%&pYc(tB|8^kl<?Z=c;j1I1g&?46
zn4shPYf&!RuG^NdCG%=JN_QakU}RF?<Oid>FsC?>&DX5_WZ0GxP7qQxG{zJMi`A@@
wr9AXnWiIP<Ti=LI-&DQI_W$&|VH+2$CH}eOMs_m%SB2?}g{}D$Q?I1|0zf8D6#xJL

diff --git a/man/figures/op-leaky-relu-plot-1.png b/man/figures/op-leaky-relu-plot-1.png
index e66d27a632d252c8feeb3c572a887146c7cb34fb..385f87f8a1c58ad4b3ed3f0ccf68e52e5ef698ae 100644
GIT binary patch
literal 17394
zcmd_SbySpH_%4jC2x5RVhyn^Ipu~WPC@C!<B`F;;Al;})C?HDX5DL;I4MV92NK5B{
zDBT@W=bjmZ_xF8koxjfc=geBJcfCB%>}NlF-*Mg7eeHRES5AVMkcyCifPh%?j;JC5
z!M=6uzoYx%FDz0fm*79gtnR4W5D*+cjs3T;7301|KyZ#gQuOvc$EX>MldH_yuOstw
z1{YpeofYm-+*H|AY*%y<)wm~}wJL5Zd7eSfua_*lUGa+|xw4cU(-{`Y_QhP4Q=6x6
zn+Ttg4yrXkMJ{dXqjec`T$}m12FI3mewNGCMlIbal5g4=c4=z46ZD0E;FhMn897X5
z@$=gw1Se?@3lcp3&N;r1;GoZK5`uGlM9u^RpB{WWKydVgCmq2pD~e?Tf~S@L&);fu
zx8!l0q^#`rN|#PY=6-_5Tmk|mh(Xu3mKKGV(S6--2P-^wGqp<4wi9vo3<Lx>EP8Sh
zoR&uO&AY<%$9Q?iDfMXu?6-!4<=ZpQ>s*g+>gnkj7|@;(9u5r)8yy;YXHS+B|1Gh_
zq@k|P+Rje7<fN0Ku`yS*R5*KvJ;TOzr^WZK0utiQk=j`ZssJ-ph7x~&qI~JQ`@MJr
z#N)NIR7(hZMn<upR4p&Fhjab|b;_ytBxg>C+|9VpgTFC4D(cc!yq$xCYRN&KA^#Yz
zXgP@?|C?8?$ii*6-pR(dB)3hpXA0ZH+(wtZnTed6PGWqY38Q?Gmu}fsQhKH7ieG5c
z<?a==71`G`&M#tn;WVMedGtF1&loOI#QAsMnB%AkPhzaZhgc(8zH(Cp@Seileay9I
zm{)G!P_ti<-8+33o?orbu@rBjT@_Cg59nK7*IjMy@+#F7{WIvT)!UZ-yQVvJ2U3`;
zb56+52<XWduZ^Z}T0i$#s#kLHSQL4|xHd$lnxe>b@#4k&v{$rVGM+6a!E&}D{ACL4
z=&MnF-5SXL#F6-4WdrZ(KfVg^E1|>C1;^}m)qMHlpYy?WW3keQggT%)Dk|#zdq%c?
z$K~-BdXKGl&Xm@cH=LKhQq{e3`PrGF(RtT9ow~TVIFL?Iv5cq~ar(@e73Z>`BeL@H
zI{9YGLqbfJIgGWJ9XxJjbe+}Wy5r6EAR3)jWVkiel^Z9~MLE9`za_LjKUAr7*82qQ
zb={8yl0h=eN-8R0WXS#!2M$3&t(9xT14TAtFL^>3)~Yi!3e(>asUxhdt!FVMN!aPU
zPfScqNMLdL2A`FwF7h3v^78V9g@vFS+(C?%u0aka4>33Url%0;8f`N>ua6U}#&%yX
zvaV~HuXTTra=XSMFj?ij`HS!0zYh-&gC%}dW+%AyJeuFes40qHXnR>{Na&jN@YiqO
z?rONnWb2QSArzA^l6T-!)G}feR`U94WrEwLbl|#u+kBj?6Y6mdsUKK|_>IJfLRh|!
zA8(~yr&rdaEi3agZe)((LFKvpoYR@$ovJ|+(x?$Tmkrd`O0M6Ief^qY>!1O5FE>?y
zbD8$~ZA)$I_=dN!{5GQ(V_o%>g=)#?Lhc)?*<me9uY*|UqFhvw^+=r$y?gS<Y(f$+
zw0=$|xJLD;lC&%S8GL+v>fPSPeSSMegphfTol-w%RkJ@Omv|tZ7Hm#2$-R3N%TWyI
zQ^&l(9OtdK))%I_vPh_TWjf<YNe&Ad`yxe!-^q$1G+||A-f){XzTsJ<jbr5eXW&D6
zlkF+)?(W*!+W+GUOR;{<OH8Sg&G@%>*Jdh)_k4+kp-h|p^l82PcL9`pp2D;|krexq
zR-gdh%6dAAe2<e!{TOeFfByXWT-h4kn&6Vx?Pkh2>eqKo&a}v8Pptepc7|DPXhtDy
z?=uQmloAgfIilfYqVFB<Y4gwVCPWdg4vxcLU$?infAT!CgvpSx68|^e)$x^I&(B<B
z*C~(cIx*tC=Sf$wO8{%cYs>U(+?$K~R|&^4!pOt%EzEJKxL*+c=mnjZn*`?Od*AKf
zJK}2tBO|%!>vlakhD?K@L3<||1W%HS;QTb2`2>Q4k<IS@7e2y?pG@+ek0;Kr&JD)K
z#N0Ib_BLov>H%N;7*R{WolDqDudZ=%88<{MRFUh<vr0<t*+kOT^@X1z!oqdo9L;Hr
zObqBD|If;t@vluesURG&_;+`AgH7%`aCkmdV(*ZnUaf6yVC-!P(&39&{xf9h-8*-_
zhOj6-<j%V)^50me^r6gUjf%*-aL^rPDM(mJ`Ojny-I10?@tB9n`tTa<ou4i&AeT|Y
zCDqI*8YY%wEyTv8nNKMLu;B?d(>*Reo`*_yu-t7V`M>ikZf$KfYKX{oT+rK)QaruK
zY?^tH{d`x2sUglrDHIVJabF`Aj|Za-JAdH<EiG-;QHzE>yV1x61C2ICmzcDiFP8rA
zKu3v)qIk?zV4p_VB_(~4eDPLf4JHOu{^AhlW8Qa|X=!TyTAsMz@ZVS6b=+Eiq^1_Z
z$z~Dr?4P|f3RQ!zx&5+5AP|Ny<v8l5OPYtcsPGm;o$Kp+)WKnGpv;9$DeyBGMcagJ
zQ&ZFG>Z%#}`K-J=ZW0m_T;zr<rVsHjEiJ9r>uY^fUkJO_ui}c$@1Hz3md57#3gvs^
zLOj79|D>5}HC^5BQT#T~i107ICpYDqcV!Ziu`W;dDAhWX{<ZC3N88{xZ>Y|mJxfcQ
zA+At&8+IAzUU1<IDQTYWr~N|}&DgX^sQSrY9BZGF{OF#1^YEl3<E~6CG7bsTzk6IO
zju6Ff20pOB|L@zk?>IR->5f1YCFI6`i^?63f_PiH+$tR{;GmL5=k?&5jrX`XBWW)q
z1?K(x_vvcc=b4$sXZF6|r}65+L66!XCd0ZAg+fNA|IX?PdSeu2Y1W?HlC3XcH>E)F
z*G?2$WW~j+-|(0pbWfCZzvqjTDO~gRaJVV4ae!?<&aWg{(F<#9Iv;eO<mTq~j$D%6
z;c{kKYTEFVRBsik&|P<{+_I$Z%F1+DOYSJW%_Pyoc9-<;+P*)7d5A&ht7O~<M^`L4
zOZL@6oC6d1rZr37hwi}5c^>LwPTrA*@SXDV3*^-|gBG_nA%*#SefjbwJ^eB+=NTC#
zk|7OnpJZ+Ce(FGap#pEned|ljLK)bcM|4<3T-<xo@N3zlk-E^1G}Y!P{x+$}?EY6g
z$6k4S>nmHVZBDY6b;^)x+~>C%RR&4CxN)vxOT7DeefUpa;*9sTYc>dEzvJv%#BMPK
z(cx%0V=rm|TsOuu-qS_$Ub}V;a%(XHA_4As4ZCUB{OV6H<ldRC{_^EZ6~(|2?@zj~
zEQ!n+S$2cXV*J-g*eS_M`^@{%P92>En{k+wSfNYR8l9e=p25Mv4%6MADsB;+Y**EB
zvx6xgI({};>&|WQmo$`ARQF5lE%@y-yGitC*c>9al%`)8wNwrs9avL~v}8B?HZ<~m
z|GRIA^BhOyrKIjxxNi^F2J5U_TRmYxmnte{<_~rqLC^F&dX;adu=tU%b!tUUPVO^D
zR>Rk?C*3g&@hyo?9RVNf**u><4ZEUYWNaLcLZJrJ4tgbRUH^5BxKx7s?pE@PcOM=<
zX;FLm*(8r;H5Da`;_b-Nsj%oPFe8t<2Dviw1VZK*<jwtqoS`8hc@HYRySvp^<H9|U
zlQg4;ekWD9$m6^`K2jf!@^SgZafF->Z2ZhR>$)T<E)XfiqjtCFJRo(Nw<QK_2pxCi
z=R5S*TfF>{JL^9K1E>*iM6eO;;~AXeDV2_k2fc5Jm&j7f{!yF#a{RF4=|$G=(>NFh
z7mX^tiLYoBfXS@{SNOtJVH`FjJd0suW_AZ89%dORY<u7a-UY>8AQ>eW!L|z?Mg|AB
zaaDL>$D3y56T(mNbsWlWsx#lTEm1j5B@B1<_jqS8{pD_(D9dsu>x9POlNbVdJS4ap
zM=cq&fB*h44xQQY^(|FO2K2s|XT^l<6!;nXk2Bx6QR=q23X^P@JO1yB`{Gb2gr)l$
z#vu$FE&tBE*V$J8S|k9IGV2kkBR)^%G)}?8_G=y@Ka3l(tur147Vj|8ngG*0<N@*N
z`!Sr`nzpqB@aNT9{%3xD0mz1TvW||9?!TXsGm-ZFbNnl88tg^d<^sY%3^>M11~YqY
z=>2FRhApeC|0D)rhuV$bQCRqMpwJGV`_{U!yPLf|??||({-4N{S2dn^2XG5Kzg9f!
zsox7+x;FJ|X?#J(=Da%HQ}pQPXODv|#PiQ^ku0xxe6%;;{M4yaUL+O`{}!U}1rl)X
zty>>cmD66JpHjjX3~W?CRf#ZQ%ZR*;aU*zDw_nSq?a2xXLjSh7_pYtY%4A1!a`K~}
z)nq?&eUXnZ;v#=yhA4tL*t8+yDw}HN<pGN^>=kM>Av+$UU@uFDU7qO5(!uA--<H+}
z;03-gjVVEJa(?J`x%Jm-GKA1qBV}&iz70v^N^S0b?1u0@BGa_S8_iT8@+^mTKR689
zZYs>+;=hT6N+7~HbRb@^O9J-#yT$iyZ1p)Qwzjt+W0)JOwn}2HFlRAMSNjq!#dgwQ
zE^8K9eS3S`ZnB+aBlF){r3D}Xj%|7Du6aNz(Olg#rZH~Jtu2*I&6F0CqUn>^b<9`&
z<f?!xvNE{P$BrL&0dp&U#%uI%ie%!5v8HGN>tWy6So#{b^I*OBdam;<hJwiZ?c2BD
z2@8by4O4L>yL<!(^b+&M5V3BnJ>TAmz@jud6#g4gSq$;$(Idd*>rluR&0+WRw*kbA
zSL<>D63x;md})baket*s-Hp5W8D2z33OLMe&Q<7?Iot8uV_hBZeD@`7ZNQ!e-8U`1
z_@25pL;r6`?{Q`T6`Y(_GrjrN&v<uf_O}pQ=`*Q*vc{YEeGywO<EE&gN}qbt-UzI3
z;@>V+0s()9ilPB3E<-`WzfP2bw1s#Lxt_623Uv|tg?kl*DtOmZ!i*ciJT8d)-~HbF
z@4fC_Me~Dw0$y}=b0gzB8%JG5X|p85N~1!Fj}iHC)HWu$W*w}CCnRp56?DHzlXz&5
zbI%9JVFpmeW0TLD%hF2t1xl1n5qfQ2G8-hBMN*-x{QUeWT3*w5lQ0uQ09K^nF$>Z2
z_&hs18`4I-am3BJ{ry@P#(cMyaW`!2qP>&3c>vj?6q3xjU-K=-^j9h_4Ww}3qiE>(
z=$Gebz}_~$dy<d}oR6b+qRgbZuFm0D^@YLm#EHn2S!v}#nvM!2e_*PG&@m^rpX<*h
z6wB%Ri*1a}kM|W?RzTqi%tKR?+?_igJ_jr>FKZMcQ?TiIVqyY{-+pRINy$G&F#Ji1
zZh3K}{?n&V^<h_(7T-F|_Frvw?%4gf)c1<#R@)uLo!qb^KR5ViyeCv+kD9B#naKP?
z!;?SN^HsL;y~H4~#X{xV;{u0E&Kg?67`3&v0`}9d(liNP^}u^T;M|Gb-IACQPOty^
z`htjvNU>0CkIX8U^JX>u!P3RXz8SZt<;Ao8ZzZzCk#DA#24!eFU8d&*g~qkTXDiuG
z;N#^zUNpb^mZPg{>7vqr>-DVp+ZjGvQ{yvc#Jf8WH${%H3zY2GNjy^sf9J$Ap49A<
z$Hp-yc-rh6F2(+ar@*l6fu}&SIM`ijO|ZK<g0B<7w+LybC`XwPU>+(v-#Tx-rcRu$
zM@;j67!ul5AI(=-(pLIWWcHC*38<08pfqHnT=eAJ+}y}WLc>sus8cO*C9zUJK6{b*
zA)zR=Um9&>5b?fbby@mLJT;VtR+S-844K{=_GZCVlQ~9>RK+AA^gj?XL+Xg%Rc+{V
z?8w)V{Y7=5Y)}kuo%`zxgrW!uNH;cPO(k>XY*Pw^|DsAAEe47}C@QfPw1e;Ge=%|7
zB{ESa@-z)iy58Q<R!4=Qx#7sk#AG=C{Sy?s^S%;@ji{Ef_tiRR`^{z|fwj7hDV}t#
zcOf~-J)HP5z0RTXo{^D}iOD!{V?}et6jqh(!8yoXv|&dbM}=UBiE4s<*&DKi5L6ek
zUq2sBOiRXX)IHU~Ij?gNw~;J5wp^yI?-zd7xHvo4WiQ2@r^DRf8Ak-r<YYZ;-fd`a
zGMyaRvP{-*y;vpYQDOM`6|E)v&X4WuPh-=<Udbid*xR4o!Es#cN?K4c#kWL+hE{F+
z?{w*S$VyAQj72-F&lYqy6hn5T)Ne2zBB^9k^X9i_SJo&cYyXkalER|?by<f1KN$(z
z;K)dRh}@VG2b+H5W2XX<F;x9AWHf5C@08mdv^%#qr}ZjhYd#-1`(^an=)IOUk<z4B
zkG%2BgtaK*46{rG-=nxth|`Xh?QJxvs%;swU3JFdUji<}_d5MuGpan??JX3KOmQAR
zl7?d@X48;*`ucX)F(02C;GfMM{xCUeYR@yMZT~iHPW7ixzKdaqLF%@``i{+ksgfTu
z<*)Ez&@~>_(9jUcWqe;5yNjdh)$N1!nd*f?S(QHV18Lz~^{ezR8C$o-)eP{Vth`0m
z%E}5DpN+*4r<qseq6qo=0J^r-?O4H!60rkiRz~Yj&g@P_a%R|+$On$5C`1$EW4N_H
zc7jWbi-(CQJ{nrOGN7GC$*GRa31}2H>8J;|c@@O-(zr}@r^s9`WTC-Va9QKbf`V?n
zd8RI_(?g164!+2pd^@rBfa%ng$s5661suyd%L}xsJ1{!azFe2?|9RBn3uHBv3=on5
z+=RNSTAu@RuoE9F+ZeeU>y(mTVsGA-tc`KDGYbn>L}~+~?)5s8edLS|)pdv4LYK>(
z_TlZ)y7<9lxQ;ud{(*sk($pV-fw$A7!WU5b5||t7m4zj%=^M_bdN1UDs-%cl7HBzp
z<hfT*bwqctwuYU<QAyHtnD+?@0C?G0SzqWY_k$^1Zu-V!XlA!<CDlcel&layFrSX4
zZ*qsR3p)>6k-?jHH9o5E!x(!Occg#wwaAdH2Jr_SSiVF3vohlK>(>x@08fsn1=09s
zp|dHL`GwroKCM`?CqRu)6V29stY2geb6Lz2Z%|B5mG|!7?}a!{&Y|5<$jE=vqcWuG
zEeDU)RD62Hh=E`^9ZyEEs%N6zq2{RKPpP=sFt*6vz56X)J-06GN<H>FI(AD>M;1=Z
z=5NZNS~ucWcj=@Lc`!jj#WH4ov>228-qr-^wTU(#7fI_|woB<r*PK0f0xPc?)0lVK
z*0doB58=E+y>xsO{Pe+t2fU|xBe3hCf*IbLk9{FhzPlz6Tb=!I8$mujJ)CT!UT_>|
z+KoTN5Ko>wX-kqXe8>64TPoBOJ9!fq@d0}tsZ$4#5vQ>;cvN{$RP<R|n)%#7*+|+8
zfqu>v4maof12*xF0s52<<o%=?6h#lDprF1yNvuRdcOyc`V+R1eK?d&LtjE82Jltmk
z*`zAUHaZgv=Z)rtZ{Z4wvpncygoJMdT?(LVa%HOB*V0(Q%OuqMUWFm<_Z=+go-!AQ
zV(Ss-l}RQCWeC<zCw~51|0&^KVVFKxXdbF5AkU$C<9*QwywiCFBI#~8egG~3bsUSU
zZiqOar0%sG_x8!US6k;&_&WU2mp%o-LQ(IPWZ#uguO~S8eA(3&5I_V}j@_Cj`r<dx
z*?h_fZKBvko3?;(cKV17^ckCnJT|*8Pq-vy7GK(FwYrH<4C-I06YXbuM<ynKr%>Hs
z3p#|1aDKxu<ZSI%j$!NBwcDS?(w8aW*S>zR^wE2(-(QNLPE=G<s{H(Uv^h3QM$-D6
zJJxC(9`+{gWbbAKN+X0mDApBcyDfbR*S7Gj3ia|d#<c;wo+E+m$+a5#98`h4EA}PU
zPXB7fjG0@G(LnJ#LG#Ob#@Bs{s?+AOyT)c|0({<(omRnpk3-Wg*JezPgCh&jD4RN&
zgPPqEcf7Vv=5DHz=gpGS4*V=_?}Vd`WYbEH-OP9VjJMP8%7~Pd6hIh$P?Hz2&SJx!
z(-GnOnG>$|g_f1}q134guLCN7-#pHI?%c;ts4nY@iCyHxb<kt&heAD~a+p{3er&fN
z@o-9x-&FEH|9kAm$_RixSt=QCGJv`JqKZ+wS!Uf_`do;nX=iADEp}(q&+(w}JG=`l
z`5_e*6+c&2S}d<#k_L`UWirE+?%6rRffCEq=@IYEL_$0WZs0&KEG=miS-k^pWf5yn
zvJ4}%Iz{KGjy3t=qed(>=3!q!1~4eJ*KO}6F!KLP)nP5NK$bu?CnMwkLiv%;KZq=j
zb&~U5d!YR&S4$&Q{J=`Lr6^q_lMXuY7vydj*ak;Ml_D`{DC6_}evteH)UMaC=6)Z_
z<5z9i3Iadrt&8X$;IA=gyS@U8G<&fe@bEmEnQlbqm=g&N-v^IyZ!((QZQ`U^>2R*g
zaUQ4ebr?TdQeJ*{v7W;V2si)&XN{NTqH7wQ#ztI9l2%eS72{*d+d4&~<eI`L^!-g=
z@Q>Ps7s$IO*`)Cb3K3zJ`$U$?)dpVol!<$UpC>D^X7j=~6h_CZo0}~m1MP3V{Uea2
zfMS_y4UG{Z(%SgBpy3g<eLcBsbp7d0*1b&3MA@qZfk5U-zpLSPo6O57+k$M5?$SV1
zzh-~d?$g{J|EtV~{lyfqZ&Vjgq!Zz=a+?Wvrh0B76xSBLc?u<GDIrQN){fgKvRChX
zu;^R0tO!I;uA3DOU{L%zYP*9fjY0!j?FT3_Y4}vQ3@{B0f-YqOrfI9WIp)>rrJ504
z%2n7-(n)G+m)#vVz!p$ngoGG){l(3H@UzNJU_d>#ufq#8uHKiUY$+0n0WNE^(Bn(f
z7T-s`=uRIL^LlNQ+_Sgil165y`GJ~-zIEyTvh)g%n&mO?a*hDU)hQjux#@<m45O&p
z?I|JRu1c4VaT<H^SX!Pqv$c64nGJU+czlvB6M=1ge7rA+F;XTGc(;eT2dJ6)=NA_E
zp@}$Ys9V|9g}8ZBJIF^oDCVWUf7j}ajBAR-Sar<q;l{+}JQ)NDLUQ-}lXp*zPuZ|n
zCL~zeYK<&6Nzl9UO;)bPQ2m^-^j?XuuaKxop9>PYK|Z6Jf2r03kb|}LB0$&|DxOd0
zy;)u#9US#0riM=0McFruAJ2>(H`=z0pyIbSFflQKf^MX{Sj<*mO0rWqS$koQB^G7t
zxKcy==GRQ4VM9~GSEZP%qu(y^Eq;5kGb^cj#*Vo<!YyE};%1SPRb=TkVr>?Vfotxi
zG@6Whg&Q^W9zYS<wZ50+aOE!|@U5Wf*K0M?4B|a8?K0&d)MOf7Zrl@6D@_vWwAq$*
zwnF`>Pm6g!)iOPG=k9Y)m6_?n_qjR?YbI6L@OTiZgrlT9b~eT`Gc!NOHgPQ?WYmeA
zea<Sm_e^jvmv!3Zw8;jDHw2VjOBZwMS5Y<r_;lV11x+Q(4&XUdGW<E5`HLvhh0w1%
zY{ZMI<7L0fZ=SWyR4c2OaG!fnM@|g3=-WFw(^pssHV5sXoC@{sX01P30Luu)lJh+b
z=#8#<))Ds|W3RvvROHU>;Mo09wDYW=&SO&(Lth9Z%eiy9z~Tc1F)ap++~GKsM@G4`
z`G$?<^H#pFPazywsC6o*5n9j%P{_l~Mt7le`$@@jplaJ`H!+~Apz(UQ<7iD|mB#rC
z2X0J>GyceTn(EAei3x4UQ0w1>QHd6eF={6)+oEG@Hs!V_7UTOXVx{8i=Lug#r&v8G
zcmaKvGu&+-7}ZcgjE4>%&eSgJrNI{C?>_#xnBhHGnN-7jw9i4;G;lG?hjwH8`=~;+
z^ZG_X^@QutcXgOvdLs~Cg_eWm;FIz@aSKlHOp+ML5Bjaqc`9x$U0Gw8lG{J7XpK0R
zB^?SGEM1Lo45SY}0+~hhJ@1**r^|PDHfw_~3N1~k*@--<dRv&aAmydhTUNNzL`*$O
z+wW0-$<WX!F=^_VOUwtE`3Sw&N81u>4Hh9#7h5NIR_1n%+Qw1WO$0y&bnNgbUkiQJ
zLLL0(#8SRacR6x9aUc1w^0hZVmu?cDg%mFu!K<2azXKfE7)Sq;s)^6pYrB!t7Bx}m
zU*??D8!V#cS4MGX{z$idP(TPM=e#&nv;$j1KJ<EedNG?NFw0L%9@S>9G#*=D=L-E-
zHtjQ3bD}RMx(;FcI{OW&4xM=`8(hl`{Qms>{6*E?8_(rz<L!@`<<Km*dUk(v7eU-`
zO^{L<C<!R_zcBD}vY6>wx`bFMM*%0!fK8VCILe`sDDks8U}<xIKd0D}kyMo5kdiwR
zz0f*xfJ?_g?`4M3>|o1-xpp4%s=(-tkB~&Wol(b*9lM{w%gFftaTZ`j!pmJP7d7RP
zErl*jn-y~3c?|njOkPvP-i;{S-3QBGmDtDa!^@_Uerf$z5U%GAc=e@hU%Sw1#mb`z
z=4uLk#ioE66iNBAnMgmUG9R|7i$dCf$wEJk{aoHv$lZjOM+>zw<U1P#5P8M{UaK-=
zC%YV11Xq9cN5@YQ#_-#CDkIKkV|cT)%cv+RDQRi9KWBY`uf^EJX{Qc&ltU?)%l0Td
z_!=#1yJ=@08ueYt?;`YX8mhShKIgXZ$}+b_T-Aqks~mCW#PybdUZ&)Ul-pVI^7WgW
zC%Rjesq>QSDD;`N?!T$d_Afd98TA>r7A7d^aCUG5%Xsks!lf$go+-5nymfDkjYa7V
zRbBAHHgF2MM?=|E<Kp60XZlpXW=Y^&%Ux+PUUtJhA;f8}JS8WbI;@S{7knhQM<zb`
zVJ7S=DYTjZvVD#<JBNPvcAn!|qd>XUZElQRvX>Y3U5k0iuiS6P6OLzhF2JMr4Qw@1
z0v3L5k?#aW&-nQG$B!S+yAdK}6duQZu*UX$dvp^OZt%7ze$t%>yRL9Ai`Hss8aVk^
zeDTG*SF_Y{0SZVz998y5v5_6h%b`<FLrqP6^5k9rW~`J0->Zz2HKbyCaK_?|2#T1R
zmy`$F6#c9DK$&n|zC4X_7|c*gl3Td>a@_mrSZs?ihW{*?-SlE?F;aP|V7xO)DdQT|
zXuI2lwYS(Z$5ZM=QN3*oZP3SnZ9=6}H30ZN?&G5WKuA$rp;OzwrM$6VrCnb?wRP>K
zQTo{;3q5~R&QTqrK$f%5LkmrDq^c?qpe&zu5zyTRmduaqSLsi-#MtTbr5}^pWZaPO
z1^a~xatFz44!aGUnmfXA5F6A~21ZMrmb{556-(@Af~F5PV}q0b%{io=b3F)W);Ec6
zE0v36btXy#CKy+ojKS6>dW{y<RaLY??ykU0Gxd|GVD9Q(9hWX#Lb^O0%8Si1zT<yY
zlDxvDI&P`V`@1c!=Kf;V$4x;5RC1bX7M>LL-e0eJOuNEXn!fPit?k83GFD})E2?uD
zOc#+`d~v2LP=OXVCHr3Hd6CD2^^P<Po7w(i=z6$}gsMuFv!%mK&&NAVl2X~3^)Go+
z1=t)i?qSP$guQp_As0LQEcA<VjT(RCAh5SuI~h(aD(mR=tTn~**{!(SvoxbK`YP^M
zQBGM9`eCi`hm0o&+e0o64$J<cM>=jR%C&%)@orIT{Y^SBu)$*t^WivM7H`JT8@Dyh
zeoyh>ip>XRH8~WkUQe**`I&bN3fYI<A0|7}mU<bm>+-sVnj?d%Fk7#?M<e{DsKkt*
z>yAo)P#x;2s?KR4`_Ti~^|Y`6XC7S}^w=Gb59FIYlZL$XVe7%et&Yv;oh9T<Ma4pB
z>PpjxCKlBn#DwHVX&)wG<2sCxNF?-mfqQoZ`1?0bC-)`Sy=&Cb+1}J}?aa48ow$d*
zM=meKMw5;{^;|ihA2@|m7sN}q7VBLd{FD)-YDCVP-urY88Z=udl_62=mbRu!h0asz
zk?w5!sc=yS+_|%D6v&dM7{RHjsd~3&&^&#(q0m)Sh&t(nS({90>r%9qa_Y3`LEjcz
zh(C(NcDdQ=xrg<cEJlZi4<0`3zByCq`gb{5VnwS{^tMlwO|ogcpuoLN>nIf6H=z+0
zuI()OGFrcb_>dzVPD`LiY!P$Q?H90j*e04`uMo(n=DAdGo%FO??%_md6fnkMSy-ZU
z?j%g@BZv^NR?yKG1z!Y@GTe5DP)Vwb@8C=pIPI0WK&1KO*JK7%b$7EH)6>&my?QlQ
zv8z3M0FpI<F-{@#{Dp-aiq`_VE1(^Fy?G1!Akfu7b>D8b)SayliHzpDZSwvaVF(7l
zJN=>vYv@ZdFfa)1tjAyT*%Ky|d$*R`<ZE8iPT>eoh(_mjDa%+kzOMQnM~{`}*6k#8
z4X&9E2Q`lc1qX9zm##N(Ff%h#5gnEcVw9`{9o(Zw*#!X&4Kkjez7GvOU|RW_2Z3W~
zbTsjgKnsJ>s`aGfw`OEucmSOw=3vQ4&+B}AAVo3brZQhY5N>%vP_XP?iX!w)aedwN
zi@uAoE#@@Sw#J1r^jtej(VHYUMs)k>`I|kA8T9=zMUO623qz)cJ~gA{Vb9x1ZR$hf
zQ2#?|c8^13HUI$LeG%vv^9cyBIei0y4YVyyD1>~i=k@E?!`M`tq;@`z^u1!NZETcX
zxklK{Nnzl6wp2W5F*d(HVn?klX(}NlT*cn*a?i`Bxc0GOnZk!>wvL?vw@8sT)7{z7
zS*HR9K~KT}{?<XU7cgI=EYk`t*T1-&d3)%{e(d%$_1&-@z9S{|;<HlBgAQV1!VGkF
zT94&DS;G?v*9ebOP^e($6T7A^%0zrH?_xJRvBn5-;_+w6f;e0F`pK%^s2fh!ckcMI
z9v}yY{Jqy9rc010Na%060y#>`7se!tApP#mg0n_Z#1<^Bx3_odX&lP?V>tUQ<f8Z<
zeSUU~_2qrI{kIuf7Z!W$?)dxr-}XFQ>m2eFYiYl=Fmcp$czI8@%NlaygaXBHqsC%s
za}=1<&ze9;O{K0rmx%rixR_6Q=1hFo)R7Z3m)drYLv4HLw}9m&CPc>B`B$E48@7v6
z)8t18*M6&s49sj`X2BQbRRr;Q46m0sFM-O=&USKgI&hfid->Rp+k1_=%1{~C|0Mja
zs<E=<;^T7!T?}O4pCx@kx8P>{FrNz>lant=29=hTVcA)1(gmFgU=X4f*_H&L2x9~V
zfFkTV;e>s_Z`tHai<i_0<Y+^r3K@PK@e0bJ4kH7D_nw|2pbaXX4&Z<{KZh4LiP$QU
znwpvdk-t3>t^>INn1erVdn}w66aYqzZ<Grj_JE4&#6Q248c0-0v$MCKZA&5vcOHhB
z?f)$x@<kGk;n_NsqSjH+0pVf_EWhGs-U9~?B+19lj*eER(czwhpB;-N3Oc+Hk^#Ss
zss+;Qebq+48;<k3J6ju@t1~|}@YnFd&-dLHhYufyxY}2TYyST+jw*ATwOI@(3g$Y}
z$k+bI+!ohB`Ge)dM6ukp_x`9BoMxuV#0G-XZ-x<TO}{5*hJw_-xV${wn&2n1@^#-{
zySO25W@e_Qre<rK=Ze4d+bmwRj{-J4%dR;-ImxOc2-N1EEmd6St=L}eV^vPIo$Xie
z{2z0R348Mf?Yd!&XZZf;Ti?I3IDGJ6H?Cviy&tdYd(UZs94s%ppKn$`f#2po){m-4
zc>f+6FMZ!Y(Dx6g4SQMq+Ss_FS@f-^M?Dh%+}}fe=Pg8He$}yM*^Ym^o&7(iL?JCF
zr=4pg3)E52f8LSMgbNy|EpcMJw0rHWHmfnK{8uWFM9_1(SA^f>-*cN8XJ%oUsOP94
z1&IDHoCl2+Ow7$yJnYE<GMG&Vzr%PPm|8(GRBzP*)P<&P;-UYU+P!!bNSpvdeyOfb
zRo-hB!;!q$&J?WU1AHa*J7&li6oUk1v=Q|Oss5f;**NIg1OT&y>i)+uBcKwlEp4p^
zdFjlHb?kicZmEY{f_Hcb6**M29qbZk!NTueL%$|cM>lB7vZ!{ot`gFzsAF|`3FD&x
z;i2V8Hi{$oF5Lxqc=7gTkB=7BJys>Oze`UJlD%oRL!*pxt3oM|@z9A(Zm6vx$3D7s
z>v1xEKEw3Okw=(Us9n;-2E8j0bVigi)tZ>H<GmktyaVQl)ls@Z+w)SOnGWJH@5}@J
z4HV=cg1Z1d{`KbgQj@8^nM^;Y`JJR+%Kg$k=F>D<b*diuV>R_ECu9Szt9_K*hWFRd
z=musRgk=8W`u6tr!op9<kVPdWgJmwNSOsyiV!{hh5VuuNv?Ucj?78MR{~gK&5P+m;
z2c}a)a{=1Rso@N3;@}dI*9`A*#6KohufUiwN)8ZI*}5oa(d~|v%%wfZml9m4N83mr
zE+rS5Xy}a*LvJwh@$vEVJ3%^+N*k6)YU39W=<D;_R5}WE9t0*jIyy7ePwK+L!lI&I
zI4ZV+qN6boLqIdxS6?q(${YKDn<_Ua=izeWs8@J&bR$LXC<?PzBgMW68Ywd(9nudD
zy7l+{H1f<9+u9lx>g1$f3fWv6(Es|2zC>d#*7?!IiZjT)hmN>`OpM|TCjzksLWTPo
zs{n<GsknjW+L$ew+lh2`m>VFAZ`@B2AYP`;vGo=)#@l&g+%Vx|D(c#m8Zk|`-K|~P
ztr=^*7wpZMw#AvAo~jN=rApZ^`O&0jIXU}u?1F=Wa*Uf}zJC3>xi(j^^J_}P=Y$s8
zHYFO9yL;ZEtPxu(5|ZB6R!>!IoApH5tB&fX3ft;8HpL2Mb(rbv>1k?eLQ)Xg-CDu|
z{l{5_Exdl9nK|Tma6gES(`D`LS3#P4;rw~I*qfO5lELoVzYZNb6bh1F&e;5Syne)R
zFKDQ+v_Qct)BXw%owSbGp4^K(Jo#X(p#KI*OS=iTSnu<dLUrrEK_3m`hci$tQ&Ue3
z4GqoAfY_)GV9Qw-OO-xmiNCik=H%pn_?G_EsYg(BK=xRf8x&$>{L<4?JnT}C%4_%(
z;w}UqEYO|$${iFG)G#D-|Neal+TgfY&2zcW2XEdHY<R}SsIO8%?*ZQno<IKs@Iy>Y
z47Aq3_tG?rb7TM)UVuD~mE-UOfT;fR1-eRp<Q$N&asVPg9eG7LwcSrh{hBZG*%P3?
z<3N-eQpXMa;e`tq)~0(P>ANhCpCTn~&ofmz^m3C#6!E)b&B72|Yr8mH>+9<aYJL|d
zr;YXXhrM|zdQ~s1X{NkRVvZibk===o%g$~OAEv9Mm01l52Q$mMy1GKs<=rhlCC=Z2
zVOM(VS^y9>;_wj@g98<SCa|b1fQsiwjxqb+!A8CO#eYKjgw1<N!>^ceKYwR?(*q=F
zkk7_67|4IWg|KaKY+ha-wDiQlEoy2UKqjtJ;STDWNKa4Y-Q#fB0k<lkuv7_=xp4F5
z&GYB=pv~o<tzG6U8+)@1BD8f_HLThLe4~J~n&@!l>kATj_h*3NG4Dw2fMX{R!$YoU
zsMHb`12u>nZ)AKFm|)<S1AKkqRDqn&3EBbIrN$2*I8i(Ce{O<@uG%I(0-Y5s9Za#O
zuWxOpZ!Nd^Cbs;%mmUD23Ae_lo+F`O^z+9L&>G7}@hKD{wdm<%vBg8_E(i2T4ny-n
za(bMZhlhuiRVuAx0Bx5BCgOR7L_5de`AC`yw#@|q@9{;3tfzqUlRM1f<Kv}6Sx3M)
zO<Llpc+8TuLQcctij>wjp8xcAOeSKS`QpVfzyRRb?I}vI{EG_<l+Yv-%DSI`6S)08
zcA^o8+}T=c1}BA2NJt3C4BY2G9RXVEgKGyXPE~gX??yf2%2t}1Q7UNx0ReV4HcP9k
z+U2fJpHdF}=cc>1!O_v%Q&|<_qM}jI8HB_Jy8<#5c6N5JjmvxQRrkPlS{|M-rUONM
zHMONb4S$Z9-`TJsnNN6O58l-pu4g@ElDrNQX#;~{u#)<4j=7l`S~@!I1gTKHxmUlH
z@Hi(*XGL#}S<r#9gsZ{~%V=xLnf+%hsyI}M!`z$7NC4F_f-b){HxcHY{&Ng_Z<5JK
z(*>7+<`C>F#4M=gQo!h_c^|&YqWCT4rzQj~wh|w@wxOZF&{79xR^><D2k05h337)3
zKYy*8=6~*GMi<*nslbi!SZf;_tTWQk0E?in9=J}l_s-xs)^3AqK%Ho&b<o{(Ei&s!
z-QHXq%rTVqA|x}ouKAzE@prKee*HSzVNN?k#Rb{|kcT(d*BbyVwzRzN+VT5yP5`$$
zj$#NrMajxqXTrTY*<oH6!U8e>yCMTQUap6)6jXxJbmLp#FpmNZ3P&KozSh^*!Tu3g
z@UBM3@t>Rhab4?CD`p0Ucd%Np+aSWRvHCDsLG|~w(w*OQ$9<MSfXE6qMmfG^d7{l2
z&^stW6y6H6?_K@$?`gW8pRC}t4;W;;L||u%QW8WsP`~9>DG1|qkJ#f=08*L*u)X4W
zKq(LlVOQ-Z+fAWu10wRgps^uRlD)PjNQu!bcjZ5S{`~dpy)7**eSI#Uo*n`qaj)do
z&Hp?e0xEcWt6Ba<`Iv{_-aZKo46LZAaNOmH+v7HY;3<MGtMFVvXW%k$D9IM;XFwEY
zT}<SEUz>AWdU$JN2}ogPW}w5W;HhvIm@TVDLHhncG9~goa}*1Uj&}S0@d+3m&?wxd
ztq|{wA#{4VcBFFdnIcDed;2{bJ}meLKtlpg28<S#bw589Ayd0&zkiYQxvWC826tTn
zA{_6oaO?4PXLHBOko|jx^mqw;1{j%nes57x5u`Fm0dO)1YIO3^+&w2Ru<3MHN=gb=
z2Ph&*Y3ZqfvQ0R5)EXz|7dML+96g70HtDI2OJ!PM!F*P4g}n$Nl|c(-WpR;p^~9+^
zXX4IXA762Hv$e%Kw%2iT=&bgD`@jJ$MOY%&=Knl+k;GY~Kj`LxFztq*%kDKPO5M++
zcUPS4F)#cf$2Df%@0ePFFan1&Aa}70is0MDaD49E?pp^F-UrR@StQ!AKlnW6mJ8(R
z<dv%Hm3CMxuSLqm+@OREzCQnegM$O2GH9%wK`3R*hnLv^kwjW7F_<!{J9sm+u0k_w
zM`==d>5?uDWZ?f#-*Wjgv5w9l^ze{<1v4G0Iv{eI1DO;CtQh2e<cmM6x?v;k$jXN5
zH>>0;BuJ5&UlN%m77!4C|8^3)xVV5^uA}~yi%3!%pcru+|Dj{^RZP*l_LVs&*~A$j
z6>b<l0JyO<)*Q}QYjDlb(6G$vCwRTrt{ywKYZ@6DNl1ThBcLXbUIWXsfPJcoQ05>(
zue!W!iya)BcGcC@Q^OTLdS_)>g2IoHB!E%fd(8UUd!{54R^>+=+IiVfD1(`Kcz84^
zFD#A4dMG+}Hp2U`Y`qN<Mq!Wg(F>j38Ow$zHlRHiK_d%C-@bu_xjx$_fj-LwU1HiT
z(KpMs*(fF%%Bp<Yy^f2jUI7j?PNbu=M_D)!4`Sof&ce|GZmJGAFbM$IEd1?3IF=S5
z9xayyXF2Pen!*qV3z1SJ&H*+Vs}|-*w}LIpwB;XsGd3_t)_QsxRDIl3aEh>3`71jD
z(N#L{&75Ev(R$C_eBhy1V9KR`hfpq}72v!k9CG?|GU_+p^2MLbJohV|dgCr27w+Fj
zf=Iu^8J54>zvm#<U-ISmL8QL~3HG>^_kL=D__g_YtawIVX!|`JIz<CcvbQgYj{ZJ)
z8)oX~=LeI0DU4zcW<elS*T>dIM@E!W6ca!jV@4j71tE@fv8ui5?Yc;#C^;$V@<I*0
zjYsVB{E5ztzO0I!RNN8QQ8>?20n98%`MNNie$97UGOCWq%F3#<OBS6!BTqo^@&EU%
zF|~%a>Gt&W^fMieS_&Jha|;|k!YI+5s@^{GxtWBg9~Y&X8aPm0CfwL_MIhV+<uLwS
z5tC$HLxU~;L~UfO$E_ir*k&)>N#_NQ9s&a2|J~7Mxh?A5_Mjy`rkbiYt@cf>93<`5
z)gPikiDvaT51_kP;_dmjC>4LWKj*Wpf2X~MaN)hoHhD||WCdD6^yo;9canko+^CNC
zB}Rs<j6Z3~h-0gIcX4$v-Atx-kl3>GQB%r|{Nt}BW_!E@``q#-^qh@8uRWpd4IGSR
z?$&ZhbQvVEoXVYdT*QpampBmi#Imb-7$rX2B{mbZ+IW6My|$%oz+__6Ek|-D$iTo?
zN8s(d=E;K#gG(Elr>%@jw>utB8fs2Hqp4LcAyps#vR>BjxcK}b5UiCeRav2KHuVRB
zd&L$Eiq_T5vls>ixg~=ZX3+y=E*YC5$vX{;O-&{$+wos*;R|isjobAOTU*`=GjYQ~
zHbYAYR{oN_<0WQyOGuT$<OD>voOe4y_j`ApT4j(r9PWLfn#fH1#ntL*HkFk0+NRT!
ztzuG}GwUm1W$4fjP9xO>muBZj67plIS|LQ!W~)c3R^PCJRHVPGRMR>q$E(!(!R5{6
zxW}=5he$HdtzSr9^ts)vK3hfRj+)|lGngAyw411@eO&V@I^c|c!z)rA{o2|SbE^xa
zDH?5_LQ2w3D%KsG)e%}%Qe?S1GA+*$`)S{t&^g_j5#@cgtup@jn3xLnvWg&syqvAX
z*um;mZOYzWvG*PcVj>G;r(GpUnoQQTq!6Yw=63E$BeBP~-A!^Yqy08l)bfx2lxo#_
z`I0YMPj6MZk7Qh>gjB!vCl^Y1J8|X_eVIj{L`T*22d@}chkr^iF-Ja=&Em>(NA!7o
z?o5ATUJ?CG^y8y*2{W<0v%_y2UEYLVPq)9qdU|U2t=X#_ddBLSr_8go1Ox~CvA+vY
zQ&V%XzY8;XQ-7l}TxK(@A*N*Z+xC$f-lp9}s}~eWPn`ni27IwO{&fh_;PO~qHG{=q
z?>(~yjz<*@<E-ryH}o5t+S2sw%~G}X{I4pR)m65}g!UhtdVRGz;_5<83oV-Cj@wvY
zu1{E*y4J_5&Hi0oZs>rr<?)h3xk7!`tTV{z)C~=J<5aECMbWJi<uHu2;QqulPRpsL
z9fx1@q8IcVhJP3&J!r~olGAf#>j-g~i8syKUZJl}^OI<5j9#4D2{aQ(y&o#WJ!^5j
z9fjWgFk4}XmSpj9FY7GzVA~g7D^;r#Na?U5<g#Pgn#OL{b)LG!I(jRyHz|TbU))$C
zpe)!gEjQkGFi(e|IX>KHXxhJ1WO-;T-1ej;Db>&L(gD}_@Ij50-RrVe$q1{udEWCG
z3E{Kb^Mfygl?Ev>V&n@WDP0JO=9yvkUfWl6t{oMl4}Z?Dz1dws80%!Lsd)^sH}8Z}
zPpLZGZyeu6T^^M`PiQ1$-g!CS{A6%-Q-%?$Am0l`;(t|3Jkh)-=j90xo5hFgy=ls#
zzg{LjvKZzXUdbWZ{hE5!s@M2}hV#+&!mrEM-Ao+<TX%PO!ZO+t4eg$I=%%!Eho*ID
zkDp!g;==yK!t>YX4iMZb|G)Oj3u<3)l$(65LH!Ky&yFezrC-l12}!b-Bw%`cn?(1W
z4C+ZVvN5`7Ca`pBN?3)ebPze7jJ^8yHKXJV6+ycE?8YI?VY(FcDKyfgLB3L_n(|cl
zH6rK7V{LOJCc|#F>$QVM<&$siuMjXjJwx_*>~(A4w(ZAzR}H%yjB`6O550W`uNoub
z@o?Z-<v`zw%gOmX1}kKqy(0u6M-K}=_KM<?e(I1PBb2Wcy+-ceksdLx90xItK#G7;
z@MG@#W$Fs8*OP@QKi@sLjM(U}6=EweRml(OHQYx^j{6A<I7st<>L)8^clXI&)oI@9
UZvPCwn?WEcCMTLHto!tT0seX!X8-^I

literal 7242
zcmd6Mc|4T;*S~C$T}0NqB#JDlWErySRtyHoGIWzQ5m|;A`@SVbvW2o-V;8c_D6&K-
z8bg-RsAOyn#xNP4Yuw-8@2}^-=lSFLV_vW8x<2P~&igE%^Eu~CGSu9J=a|?rCMG7H
z>!yZQOiV|17(Wgc@JrS<-%TbaW-9as%m{ova^wgzGcyYd3o9!t8yg!tJNwb2M>#k+
zI5|0w9XrOw#l_9d&BMdP%gf8h$H&jlFCZWwC@6UR_;Dd2Az@+R6DLlHh=_=aii(Me
ziHnO%NJvOZN=ivdNlQ!r_19mgPoF+>=8UYY?Afzt<>ch#<>k+vJ9qy4c?AUpMMcF6
z7cM9%DJd%}tEi}`s;XYRcu`GF?b4-7moHzwa^;G;y86|tS2Z*=G&MD~w6wIfwIL9Q
zj*gD5uI{yK*Yx!C^!4=(3=9kn4ULS9jE#*=OiWBoO|M_S{`cR1o0*y2xN+m=&70=t
z<`xzfP$<;W($dPx%G%l*27}qy*x1_I+S%FNx^>Im-v0LO+YSy6j*gB_PEO9w&Tu%~
z#l^+d)%DJuJOBLikDHsDySw|{yLUZ2JUl%;y}Z1<y}f;We0+U<@7=rS=jV6-{(XOc
z|A2si2M-<u1_lNN1wDNDFgQ3kBqSs>G&C$M41qv|hlfW*L?Dq!6bcm?85tE76&)QN
z6B82~8ygoF7at#=kdTm=n3$B5l$@NLl9G~|nwplDmY$xTk&%&^nfd6^qpYl~$B!Rp
zXJ<co^5p5$r#U$}xw*N|o;}OU%X|L(d47I=K|w)bVPR2GQE_qcix)3SN=iygOUug2
zUcP+!>eZ|A^74v`ipt8$s;Vjg0MKZ3b#--3O-*fWZCzbmeSLjHLqlU@V^dSp>({TF
zo10r&T3TCM-@JL#*4EbE-rmvC(b?IF!C<ghY*$y;+qZAuy?fW)-QCmE^Zxz&-rnB6
zzP|qc{(*sk4<9}Z4h{|t4SoFhad>!mWMpJ?boA4wPoF=39vd4QA0Pkn<qHmno0you
z<MES|lT%YuU%!5xo}QkWnVFrPotv8@5D4@0^WVOGBNB-V3k%=BfB*60$IqWX7Z(?Q
z{ra`Ev_vA2mY0`TR#sM5SJ&3o*4Nj6|Nc!TlQ%XtHa9o7wzen~%J%j)l}g>&+1cIQ
zrO{}A{`}e7+oRLz`}_L`2L~jgw>=p8oc^ZvflN$Xf{fphv02m}6O-ufbwfQ^Xu)b>
zScH&GOuw4e+Ms;<)gJD&{*_9to&?Kx@s8K78X0QD8xonAj~KG^@f(`BK}}BF6%r6O
z;^+JRn<FVI!64CJJ~2sNK0)i$HB<e8!XUhK{@M(k^svj00+4Jl1gf>Psk058_A2s2
zZfS>=@T()T$qx|BZ%;xyOJ9T<>5)@nh{9~W3dd6HgoNtk<()16Z!X*riu99x^o*)d
zeIC^d;fn5jf8b_M=1N<6*f6*wBPhi2QBchGp{2_$uc|DJ@paq1yax-Jrv6qgIiKn`
z)R=pe990@~eS2B*AJv3b7Kv9DF##1a4&=9I-4vvNMRq_ginKIft6k5j+FPuAy8i`2
z(;12+SglqZq|<g+HA)oz{Q}FHK#q}D&czzoU87jZl<u5(irb*wq?;nVA5Fdl8Wq=M
z*Pmoz<jsoSR054<%yPCxa{hPvQ|+&GsB|q2sEqe|c&}c(ygJWAkqPUFJ;V2D@70W%
zc?q~>=z4(V^Mhw^lb<ef6x>>4%NdpmuUQ)4XzC5YmWQS4e4QY4inbHql?YJTS{hm-
z2cn|NXk)Q(Z!MkE3F*R_WeC4p8KfbiLFUbxKargc!~obR&yXek%c~NZMq$2c7Z2S#
zrY*!0gx_g2_L5GjXg0$V`2#@l#ubgT4&-avTKbv4SU@AdMsJ?M<v!G`Q9JKIhC<Y{
zU<bc1Vl1apj+Rd8Tno8%$fO&My_yqBxRm7EYe1Qav!lljQTp8#q>|jHuMHlXN=YQO
zaaJIn|N6nO@7V&9T<y!6teVij9;dmH&L1w9FgW;~_|G${`#jAV#QDBqw@5_+WCScG
zl+#L|`8$KVK#x5q^><18DKkJ!UQ%4~KHT=%4r#f!TOa;`0XiK^vexD9Vt85AS3=5_
z<zEtOdVIz=BV{;469gcAb3ew61C(P}?`rt|=Ari!U`V|i&hIh*d7&ehPp`{@v7ud@
zerNJulj+-xo<{5*DrH0GVE;U1r>!#dRrN;xq2!y^IFX=_9$(E4E%O}MpQHFPFsbzS
z|8r9P$)TP5fYD54x0!!Q#L%co_6df-XRgDcKW`qYTdqfGQ>`w|tz)=_9VZvrhh^xp
z@DlR2q2bWVU#j*Uxh;VIJnMvUjWR5j_IO*QcKBZw#*~QvFc8D>?Z(_^*sCve{nUu$
zp>k(m(3(^=sYkd{4|%Y{bg#ySFldZJsRC<GhcxA{w|w<Esm?Je1an?JN)Rg=_1ktI
z!zxTDkcc6z5?9Z$2VZpBdwKUr>GR8I=?~_77k?{t)g^_WQ@qqD;{HKI<Oj7-;A+F5
ze)MrIQEN9B<#B4Ef_XZjI6KhZyyul#t2xgDpAdK$FZFSuyPNXh1AWIQb$j(-;NRsc
zm|PW4TX}lNnSw6E>r+O^KQ9?{B=4y22dy`Br}SGw<m>}>M`nz*qYhpPJ>WM1;Ys26
zXk#VQN!HP%RTpwltkxhS7>sNk&y8c;uOF&J9I2aJyI^+OfgF`+WFb0d!|)TYc`utH
zAMSXlBV`}k!5pba0mxso&j#l(+?BUb=)J-LF5af>YX##Ym~iM`1&h3|PK=$#84rO<
zItL*4yPiZ3S*iNN57;H=-p%7a%q__>PUn|(6G)ueXau~D&QpAiLi$ZqE0<wF@t8NI
z3m};)i{6$2R0^$yTmi+L@BCd5#m@*<na5LYr5_fdkZ3uZ_2sY^OH8mj_Jcj$Q>vhw
zaVGPLwI!qcFpGIT+#f`Qh$brz{kEVjCWgV7Z>YrY`z8yB*qowSnn<#7+1~6fJzVM0
zll*Kz(G0yo!4~O~cnm1ci)|vT35>|K`0-~=XEp$YV-3{Z(5~)c2B=poMn%Nr7zaOo
zJ^ar(pBD&bzA9`Zm(dm!)0)?O&Hn`>cG@EhgdwoHSPvL{O7_N0C?+J24E#1_Xz_4T
z#2BRIm9j*-;}^c!f&zbhz+d2a$}sHw@^nwK6o7ws98k94P(W19p|$rDWE&U?Cf2uF
zgdRe@Msep9kQM=`r!y}!$TJcS?d-~uNSL^Z4g5MIC~w)oPmKPq<gL>2fDyYP>U*8z
z4WERem@p~iP^0H~FG1Dc9Zyiw)oHl@>sVzGZy70M+hc0*6s~mu-!kL4nh(ciR(ndn
zv~jfYI!k>_$z~*A&HziBR|KwtTh~vF%k)XPPp$Mv$OM{RaoB4<+Ps;6>D~c_aSt@z
zK=pU~kL!}TZnLd9VSi^fF)o9kXRxl=>sup|IO$tX^!^#XqyP!rxT^-Byd3oW0=M#f
zXlf#%>5?-;m)k9TE<8Ox6jmi$WPq>v%bTPq66XC&;S;)14BP6p@VN2Ck~kgF2wu<}
z&Uc7^4@v7YHhSw=5ILJ}QDn15n2xal6~~<^MhzZe-HI^e?$GRWc!1Zm$w~O%z!5-w
zkdF2Z<<BAJo*kJgLJLQ3ny-6F5FDY9!g_$9ou>Rz6W0ZY_eqiNFJ^=2yQ{GB31(Xp
zQn*Noz3lZM{>nMgg#l_ubyUqxaICo(;Y`r{-wC9pI7@n;(42}wCJD)fQr{xUM)Xsl
z5Xp8@Dsgl(!JfuE(zx8-Iy;!ljT3kMn0N)k??84E#Nx1?Y6S`>feHf~dY{0at1W;E
zc)Rf<LR|&q`3V+z>TISii0D3Wu%!~Ptu0wuRBTy1NriIvc6|yYmmC|8d@wstZd>^P
z%1R;OGcO6=Eg>oEAUC_<)Ks6oIKmay%AyP`iAz(AAA`F;l;c5m&gKIa-M?#!OLN~J
z(iA91)0)fD<L$#>#ZVTzi!cil3ksX$Cita_R;U%Sc5BQ_8~i-lUyPro`AY!Gn8ysf
zWo5KX>+}SBRO63ipEr+75SlMZn+rUgT2W5{zC9LcYU(*h5ES1zabMK67NQ#S%Ay&l
z5<dV}8wnL1LCg84lLG%amY!%|LCarem~g6X(Ry0?@v9x=Lw{KVN{~LR3lWp2;$s7U
zHk{9m8}|Yb)}^i6mHT^z2PZH!3jN`YVz5#3u@gt7p49Ag$axsq>C9y_u#Fp1yp9hu
z6tK0q{n|EIn4wlh18KBZv?DJ9VhikBIG|Gr6A2)($3MSBGPD&P$PW8Wc0;N)IzJiS
zyktdBxqi1t5d!E^`>g8<<{8*8RnEQsb-@N{A6|&IHem)lT$9vFJ?P2BI*HWQg%X<o
z7c^?B=Vwnf3`zW9*PNV@amWk2oP70@cNw9v3SNE}*zX0dl<dwGhR4Z(Sgp~tkh7^A
zO$+zzi$c;tGhPNpGmig-2v-WZizJ^{P1>Yp#)Ap^SQTmsrs@i&z%L+y8cdWEwywk;
zx%8`N$hPw$0N;34siI(W08LQdq3U7od80NHL!H6e?1x9AVe;H<oUXtu!m)jSneu0P
zZ-wgh+}5}tlM9!)P+KQa#DpgCjo$Rsh-e-O@tS!ObD>5pPPi_K1V=eKK5I6i7@_<3
zX?&a)P6#+S+S7Hjago)Yun4fSYe}|ox1TYa;FJls%sEah1qM<4R~w4`Ag7g9J+WDs
z1XbaKh*1T^$k3P9m@e6topBFbBzNJb%AL+xy~jsA_pKD$%?Nc8#XTJ5L=zM=klEi|
ztf9AkD7+BAq4heEG!;g#D<8#r#ytfHCwKmVhs@qI$J)L05=JNGmiWUg7GSxlX{`bt
z-GgN`G~RoP>ZBU0aMFY#w1*7Qg0Q7G@30(PCZ=TBM}V;yZ|6}zXI#KzmPxYM=Nok(
z>%BlA!7Iz9y;V_dq6!;%*{RfTc{DU+;r0uJ(_YBHUx*gsIEfLP55sovi!I!HZ{M(g
z&mODbZ9qXt)SQ4!-+Q0euu-`W^|rDadpiLJJcf54TOjhJfxCsMl@|G8*@1KY0T|4h
zA0st)a(!TgPQ}*PoJ<9W=dw*^!wv<tmh^RAPyBb=*F;bXX9UJg0OCX5d|nv#(y3L7
zae&@+cHvR;VmKd@A%)!G#Xn;uG-yUpZ6$r+GSIqCG?D7VZ$^z<d&c>S=x)@Lii@2N
zN_=4V3B8$}Xxhg$9>RF7B?bwuDPpBACOczOrNe9~fs2=D@BiHo_CO`I>0M<`Wr#mY
zsPC2$5Nsgg*r|cwsRx+=YDD*W1W6D#9X9pgKCti7T22^MvcweTUZ9=<?@Uby8?z%z
ze?om%7yre1L7aLNmyGB)7j^hRk{OJ^q?G0oHE6hN2$q;qFnW`+paB@YRM-KM3I((G
zdFa%S`f=+|MlD201<jr-h~4QWP3ENT?m7J15+>ZyowZ=hGOELvqWijoRGy<Q$*@p!
z)UO`hu90`P5qpI5&433H%se#jB7{-M<Yl1iqQ7RSjajW*rSteUWuog{ENJlkSd#A8
z+IPm@!I3<3i|Atx<k(Wpnb}jMmM2dn!NiH#{_gWc81#hJL;nI#F5I}``e~fcbKTqs
z`Z?`EH1;(4&A2rOiv!ssTlwC$w86*&S%o^9+yGi{;u}~@iiR5xN$NgQuz!eK$03)>
z+3}V8cF~&WQNvHG-;jnDG%U=UPqq;6PukG?=fv9{3)|7RdR=YXjVTIsVwVHAl<oXw
z(KR#-vU@gv{4#*&9WB((-P9I*1>WS5gG=WOnHF&lf>&0D>(QW(WGbs^5P@2@B`KWv
zHCJ42FBd(hiFGQ48b_+=lyqeQD;6$*AV4ndf?`C@B`;u|AAy}q=TWdzX-vw$MqypE
zqi-L{YK<mgOafNCDk|Q_`Sx}=Q}~oTCI`5HQB9Z`-82~SC^}>pX#_T4;2}P2z)%Fs
z^5K??NSwO{vs`fb+W}`^Y9Uuo=g-{NyjHr654|5bbo?0%ggGenMJ(?~Oy&rU0~VPR
z(sovqNc#^@ez4(uo>EDf=ulX3*2kRqGQ@pL37JNy57@)lq4%S0((H?%5FQGw>xa~Y
zV&$BfP@C!2hVc3wTxMjk1{WabUq@1ZRqvrknJbas?;N+D2d{l6G_66}-%XFg>Su%0
z{Nb*Me1WL*FDdI_M4XZhT@<Tz&j8dI2Hva@_E3!a8~*IYEul9InZQ2Q;%XCE!l?tO
zS<^XCB$rT7P%+q*&BX)K4H;StnZ+c0#?R3h(yWDEe?&!v3&&JSNFnK<bMD#+u-LqU
zi`1SlJpo$}-dNx&EQ--{Q<!rHET&xK8GCa{_%!t#He5&kjA(0$GnVM|iuT=t=Pm=S
z6`jOs3<_6YxC=H+ovE<mt+sE9pz9!jMZTvt7Z9Snc*k)6%hycMH^vl$E|n}c-0?R?
zlp*O$*&s{d)>H<=QrfAzhJ3i?d0w1y_@pC<#C+uJldBN6anR!mFkR354-BR}KxBBz
z6sTdkTYnpzVO3@v-=In%$a0{Lre2Z9(9p!TvqV~D2`to?YzXTLN9@6)x&pwpoeV*?
zR~m_(r@$^R*O3yE!q2E~XMjk0-JQl4h>(EOdf!`LHGvX80#xd(EJOK6_7pCUq`{38
z5*J-Fw#b^^U_K{+i6WX6wUNgv&HSES*FNWLjog>)h<6XuqsaX6_-^y_p0wuqpm;p&
zVC6=DJvd_+dQO!`*P^BEKkJyH-vDDb>J-Csc-L-|_Aaq1eYl(2^`mSao<AL5`#XcA
z2LDyKGjl5Iz%NvfqE&ZKD=27e7{9{l^o<9XTVatRx%<vY)s?483W~9PiuK&~$sdr0
z@k!u{k(+BHnd@5`5FWEfs|VzbO4LEqEDal>552|V421}*qp?9fag(ZEMt8b{e#eo#
z-q=z!QrUxx8qC%&pHn{NKsHCC)~ny6xSCkwEu%@C1u*1?uMG_gIwgf#JUBJtMV!Zl
zmE5b#J{2CW(v8CFNhEql(Teu3bO}{gMesOoqp|Y$f>z>q-0yUYMw1GE?hGM2ARUAy
z`>WBuz0;KlLdH_znlr-uhNv{?5+^{}Ia@5Z{cPiWUiTZ2{@cKw4&-j+;+A3om`E?!
zA#M7;#JvpRnPC<MJPeI-BGsMh+gZa;?{wc_!cBkp7C9F@t&0jmoM`;ZfqbG0yRrEp
zpeHB4-BJ`bnxqZrb}4Fn`y81^&xj$N%rc}T{@hxDD334M4<!2ba@o+^()oIM;O@VU
z#;^h&@4axv;XbB`OVhoDDn~(OD@lCXc+99-zeGXQ78nssxLl;BYVa46=;J?E+S1<~
z0nb|#pxYD*U6ktrAg&`h(@Ek-0FfbVMaLWk_|$>{+i!7O<Wf7ip~@#EKMibXz(e(n
z)C6BAxc`(Qrg?(JtZ(n;0s+!A6oqeE4O3YM^(Z*BO5g$F`7&7XGzyQF(X#92@^(IA
z{l!J|q9t9S7}p56&gZb-f?1=*cfd(bHYrs{LRgL5G~jnQ-->X<Nh7N)G8cJO;tz%D
zXk+hX%5sNk*+tYN25cilaU!*1Qhz>`AInfV#4rvk)~E{^)i|orb11XU1lJ{%-sNFm
zRC$#jEO<F_i^>+s`Fm{*YNcRR*C<SYQFK49KAhf6<gqw%IAi01`Jv8i&pp+T^weW%
z6o#W|Ga-)!%iW(r#E%{l9Kl8EcJ!}qW{O3AM;uBa_;XsioU;;++#Y@xFwW+zF9Z@=
zkT2)*KV2Biu}##YG(V~B(hdGdA`XZ%Tbe@m+Fph9o6p`3?=Bd(Yk4il4dAi7ZE^D8
zROk;3-Wcq-@}~K<SrW5ZK=sbNtRYpf2b=~eSz#jaX-X_2g%b{JxWI)+q$XaRV0(;h
zM8H?kat5`k)81j~7i`yW-AT&jk0-SS{7qT<Oj|HyfPskX@BoW_Rb{lna0B)+>)6*f
ztIrR4zvj*ia|yeiEuhG&)bn0?=Eo}U#bx_p_*YMjxh?BKt6LvmpPaV(OX=^;;^S8v
z0B}yCAb+_L&vB}^?#oGshNy#Zc0fHUTY28GPVH`WsLS!j!|7f@#1W6I5mA^$zZ!sN
z0dmC-=UZ8;%?Pl;h4B?IIz)sY7<-3~`+t+pDOlIHaw&GtSz<2H^Q?-Q=dIKs)&!@3
zvMf1TS5tS3M``FkH-y5<@P^U$`aRngLZU5h$#3g5p`Eg>@<mL&fw4qRXs7G{e_co?
z$~8DuKS`HjZi-8ge0D}<rKXiN0p}V{7D=AI@$3Z@9J#7=4T$&;8es>W5?6a!G_L)(
zQ|7DCh=w<I%@eah?t86wO0mB94IS59FTtyIOw7d%2nl(e=Df<6*}0L3fWO60S=!)H
zx5m~_p0d<m6gHeun~0+K)@6unyLaNv#*d+Lluw_>)Reo=$e+G`T=S3vyRY(o&=$%S
zZu~jy=JZDRDa+IZ@7r^)^h@JL%#Qy_yX+D|-k<XcG(??5J5+Ff$b;M-KVSA9V<F@1
zOe}7=y`qTvOu^ShcvYYMIm8j?#dbY%fM1~8t4`MD;63~_sn8%`bL^7q=q>A&{bUQT
zxxL29pVPf@HbI@mi`#MECPP!p?{(=xpS=y^MD>aJ&sRFnP}hNNcN|AD(2fYiead4X
zQAiR&5ie)K`(9Tv-9fS3rCz|#%8}xZW^Epu&mu;1|1S*b;XOpdVZ2|>cvHQA5Ik1b
zbt*>B+@jkY8gYT(Ozp5XAAJWtOU@>0LRcL1$juR-w@LR4ied|2?1&9c`#L?I-#W9x
t+KX55dGm9<{>zg6B**{Y--3PVD%5=6DXtwp@ShQ;>qh2=wfZhG{{?iPi4y<-

diff --git a/man/figures/op-relu-plot-1.png b/man/figures/op-relu-plot-1.png
index d246197c26637e35b034f238a7ffca0eb6d46ec5..d867ae184c9e2e76ef838daca506d88af54a8589 100644
GIT binary patch
literal 18241
zcmeIabx@V<+cvsTL==z^=~56>N<u&yr5lt^=|)1jB&0z?5Ge%|B&9)0x<LelB@NQu
z4f|aF-uInvW`DZpn?3u#y=K(s(Z#y&`?}6Jj^jM86|5*PfpeAYDguGPk&+ZuMj+4*
zP`@xQz~9`JHoF6VTzM|3;ebG35ukpd{TTB)K_F-lQlcVHU6NME-1(m#o?qVZi&B!|
ztE3zJ^j4YRhEANOl4MApw(1D|gmFmp_nN`MIE!uODlL4yJj(@Qt;*QZM5K;&^xI&P
zcpNdwma(q&$i1rLv9;BfgonDlLT(HEU0s%@t(}?s9&;3Tqa;P)HeN`@P{JQa#357&
zpJ~M%1cERy?K0v$51{}8(fnZn4RPn17%oEC?v5J*5%Re2B0`bk|9@-v3S#|JVIdm_
zN6ZQ#0<oS@QnFKI+-YQNd_2>0J9mMDojp1_TJZ4q^&&q6qFO=dWW%qvw)WE}6YtSV
z{~H1#c;e#XHa0d!nFxeSs@LIc{8^Pl*WVW#!(u30QL&S8T7{c`H9F0kC*(7qbwiLB
zk_BR{m((Y1J+HEQ#CT?3$T|E6S!Om3<BHkmoXPU=Q%CP%=iJ4s%BR!_gudePej2er
z7_wwDGxBj|yqrYm>Hhl-^Thz>y`HmvTdI`s00D$>-O@g(=lbbL_1zq9ri0?yE55H)
z-&TwoKCH<&2+pdbrbhUn4-O8xy1GVE3(@moieL-(_0xOvC%aor9P)*ac~e|}!LNVW
z+f$dvp(XI`C<0L$BzlV|bar($IWBH)auSb_P}C{yjCxEm{9tEd>IQiYQMk^CfI)1C
zZ*JChdt)Fa?Y!qAb*Y@*R?jyCVnY)jAAhDLoX=sdtGMnL63MLoMSp4FGo3h@&!;!4
z33a#k;uPE;8;iYotQE<b{6X+pgIP0o_gsH}|IExxDxdn(rx|&9V|r3(h_k|^q$El{
zr<2X0H;ig?v$IOI3#n05+3e4<JN(P*>bnv+NU>WSF4rB-C1xt5)(+{z=HkS68W$t?
z_V&Jh{aWR`{`1EV3oEOqN=i4m-yHPFlVG>NPo?tl*#B<(`juCC3-()APA)t#aU_aX
zLOzM7(4?Ep)O6nOGHy##lXj(jz^MMI-cG|;a<=%jvKk}z@U8_{`JYc-?YUn-_`Jj;
zCN4MbB(k)$ynXwlkB`r%PoLm^MI|I~sRfh%{G>AHV%dn0#|``A#>1Z+D(BAexYkSl
zH)K9xH@A=@-uq8a8@v>LpPkL?yf)g@)D#vLM$T;$MlF=GqJ2?m=bnScM9Y=+#MqTE
zZ1(!rbc9bdm(8fUj*gC&*2wqocqAkwl$4%R4VU=%_=+6xn}m}GWZed)<!Xs$o?|>f
zyw3;=!!|cx+}YV_k6{=Z929WfGWwaTfsN7F)a1WnhR&vM+ahhr^voQ)>oFPvyCjj@
zR#Q`x03Dx%L|sMY#p<up>gsA(i|FKKXLX~Ch!fh|w{Oq)rtWWTv2$?9$jVZ1+a$ev
z_fFk&TEe4T?E)gGE-sFUo4Y(H2$LwZ)V%-GmoHynBih>9{Qc33(oe52x*>e(SQ!~R
zdwOg-V(${O>KYXznVFe0<&#6g!zmwF5F-1|Am<R@CY+al=0!&nh>3~0yB{|O;tmZA
zto$mqTp2Djnm@wIh5S4Fl&4j`b?70OI)Ox*+t}#p=}{997(9C>;*>v+s)lC~$NTGQ
zs;c6mqAR17kmt%$Qs%G#V`F1Es(H8ger=;_sJDcaRPcunA0+8C3k}3nRewS#l$4ZQ
zqZX8P%8&mRMunJ6-gK#k{mstK&dA7!6goTj{^2p~0xXu%yxT*lzTc`AS65%Wco7~F
zvbDXP$Z7tgKT`qT;mWToDvFNm)4i5Xj_`4AoVABY&HwNLpNa}1-@mM^Y-MEyi<rgC
z!Xlb-bry9y>-4OwtPBjM$9t>VEc?a1f_!$<P5xJ|H(AV_INbk=hH$xj`5Hx5PEJf*
z92Y0&&eqniLPH4=k(VSkpHKAEq3BoV_4M?>rmxqX3-a>v!h4W$iLYK^*xK3_rPo|j
zg^SqcKzZ9)=)<Mr|Nir*yuAF{(o#`L31Khun{zB^Aseb4ot<$1GSbo>o}P(`iH(ho
z0gtA>JZo(0*)@Uo;!{37HHA+=5S^IlzWC#|gM)*XmKIs~-k9@5RPu5+>OpLO{`^S~
zTW!RiAUn5$EH2QkIr`lmv$A5PqM`yfLOLFH3E@)w`}gmYlam{as-mKzvI*?Q+jHF(
z78V43Tj+mol#r0HY1W>YN&U<7sRqdCwTU`CF|p=v-+WfIZU5XV#^&axhK9!K>Z+HQ
z*QGzNl$4aSvoqKmEs`-vad~HFX9)=jCnqiuyw*tS-mWfAHa7SvQg$OuWZyUu^vKCK
z7M7O8%$ftk!!Czg#xgRUurYIgUQ>;MF_f#6`i>V78`6S;f(!~Ngu$XzR8*>}s<6<8
zEn)cQ3X%esQOhR1ySwY==7uhk0qbjQY;Ive#pfier<b~-y^etjf%wqyaAqbZ)b;oF
z)|QvGzB=AbmQTI^>;kfKoD}uMtoYZj$E2jJt*`5;tD6}al~@ekBPS=napMN00xTN`
zD%@m}_?*|Kn?tv@wh9XiA$S7<0${5lxaj$YCSLa8AiizGeZVsr8W>1ROH)u%#>B)V
zCnc?{uFn4c-4?Ou{2rAcX`es0cXi1sDMhu;td3Rfw?)yyI$CkLL&-LhI8OYNA5-h=
z>+|#R(b3A<+PUxF<55sp85k7VP6+ZQ%R80SUEx3_b%n#cys$7LT)fJDR@&VBv#+mj
zVc}*&09Jd??*22>+GJ*BWu>LvqNSy^v|MgzXb8T};N;{4!BOOJN{b4qiI16?nHd?{
zwH^ubnN?L)#p=1XwTGWRe0cV&#2oHM0(Ccap`oE8#io%jUy3|`{v0AjK>;_iH8eOl
zDmof6TF93au#gCzsG6FZo!w^ngqI~Oq5W(}41)s3<;#qB?nr*uxTkjs3K8NF4W2qI
zv85$97gt<-JgZ)<HVs~qkoQ@Pg5XWQU+zADt!N%SeypRd4ZkPInn+7uP3`5re}7|h
zQ%OyY;lJPF!k+?VPXy0>e@#w7VdCqH<#!J(b8~YcRlM}YaS?txj*gD@_FLVF+~%C`
zAeoAb9gK}XMnqhLJtv1srmtA#dF=G)(Id!W=y{=`GLDWV<>lOo-MSDS5TWF-=hznJ
z=1}BpurX+vne)oa<LmXGs;bi8zTMu|_L_3FrhWkffl&aR4?0N3$B%YA6u7vNFP49@
zCCFY2e4MV7O<tIG*+(QhGt(4VWDJnz(Fa96y@{Vc<9g>WV_^Im9)?d5NFT6ehUAWa
z_YRi!tMv$%fItltUo7+sPoDI@e@_E70b<Y;%FH(!6-~{13=C>|dbZ}~32)xKiH$Wf
zFfg#TmUGJA$AtwZ37ejt?n&ljprwsTOw7s6^_c5Q$jxQ)^z<xp*u_FU6E74X%qv$a
zUo6K}Rq-Xu6Jw%talthOq@<*bB$kOGeZRfm?pF{(4iA4fE18_szkK;JeGGD7AT}w9
zsdDJGn*eJ4o|{`*LOX(9^6=q9axSa4Lf*AszaG`N?`M_Rojru$A`$g}^dTMfqD}2J
zEmW6?r1Cu4F;`cAo0-Xg>^s7T8i0XM70GSV^|n{&d~|sDAg7=f*1+D*?g|!Gddp&v
z-X(})Kg`J1>zL?BBobB-pw6QY&->EEWa5}^@~NFSBo-p>tDhXqKy<#OxW>TD+}+ic
zTU&d!G??>rO!=_x3u*^GR8<{A|42wkfD!~?voVa6qpPcHU~rICyF%Eh#48Z`ANJM#
zwJ{D(&W!%W9JPXourN*zj`aRTwE|t0r%&nlH0x04ez&x=G+EFS693o0095nzj*bo!
z69`jjWo2a}335zS_1i5}X9BqJ^QS5;ZTivCkxrG9imGb9UL6W!pu;x3uYjO|0OsSX
zDmLvcD=oF-NrJvzYB}U}w3rDWLqkVc^QCz4nM1tUBsB>F<KvGXKVbcm1>E;GXIcR&
zLtw%^5D^g>&6Dw5RNP6MOROiio+Bk;F$y5&etxZsTEHUrSI`xi$nzf|WtV-<UYs7i
zfNu0s_&XB0`1^NkR8&-Is*u}`S&;)ymRD+DXiA2a9k!K&kjry}{@dQm<~T7P1>8mo
zHnkNtg{^hO_9vgFbS@wkgrL9IIW8)zsJwwi&5(+L?19F2>(;Hdo(J!bYE0bMKTUie
zjV*L-zH_%h^sPJF=npwPrB{k2b~-EjFT9G)-gPjJ??gmK;$mXDtWDb1opQZc8aV60
zMhtu(9K6lSS^z~WCnqr>!S(#?w5X^kBZDqG<>QvhQO&|fmBBZMhuqXv%qnXY$DLyg
z9?V|2Y;vjt+PZyg24`Q?Y^{rmQ7`wk$+_;t<)x(~ZZ?GCBMFJtIA$##iXPufSmNPi
zFQ%KWKtjVZ*8kjULJ-!3{fs|Wd&_tKfuNCNnP(Gk5zT6Gvq~sEmoeR08GXBfnYR!>
z0$X*Y+%}cNG$JKM*U4!QS^*^`<>0`;&wL#jCl-R8bARon#!G?2A9Jflk8WQ7q(FLi
zMv$3*$6{>f;!v&nCyoWLf(ze#Djnu~pblgD>a!=<+1Wvpfd&VC&1n8s{kQ5H(#MLA
zz4Kd&3naRa?f0wET%RIHN)9}{R8$7btVg=KyCJ@1Wo6@7^#q;Qw2K^~F1qU?d@KM0
zL5@S8%>DF<h?=_Bq}bZp8Zas;HU_e<?Io1eY-G8uo`(m(%M&P<WMpIn1O&rYPS2m;
zM^;Muq26F_Z4I4BMNu&*BxGrAZM@pG7*P6fp<#QE7+KOX1|lmRYGAX0UF9eaITr~z
zxdRV{h6WN40@TSU%5Ygo`u7ortr2pmf>2LRp^N9|=R^GjumiBG@f8j&A6WtlPb~pF
zt+1aB$<4Kb60EJA1Z?p0=g&DgjBqcnAQu%y0bXcmB&4U$K)eIz_#~Hzjg1YJLd1z>
z3xKB2)e7zW(9ljubI490869sJL;oPbOu0mHC(f;BfE}I!lY&=9L4o(y@Ai$2jgXM5
z5MQtX?HwJ_l;Iwz8}^0NdiClRfE0IV-)?Rxf}Zc{_07!8UMI^ip%6&4934TSdhV}Z
zzh1w_yL>qiO0q8vBy+^#2{p7w8YIMiOE~%dcCQe0wVp(7R(AFu9UamN3X$zb(Ufx9
ze?Ej6@MUWR<<+ZKhw`BwtZI6C)?{Zdt&LT+_uLkTDuuYxGHZ{tDyylfDK6$n?JX}U
zN#cEx{ikI$VL`2=Q~4j-BpyccU$hB&-rKh%SXfv^>A}QdHV9$m=H_N=WKr|4Uz$=<
zcWG$efBtL^*X`@;gPy{`_s&J6@1jq}Jr))sOmqO(uoiN%vR$wpb~7ypn=^1TPN*FU
zE`k4FTC(6uq`Yxsc5?Fj(2(;~LxAh{^V6^K!tfElQ9Up&PF_JFmEYA~M`sMWy-D}G
zX8rpRoox~H5vWgTci5hjlaZ0}rMYnNVor88JUMiI=Z#5`B8T9+&_;wCVq;^W|BrWe
zc6N0!v9Z~WR@gxcfW<bNCqO}KA37*?b`B1qK|%V3>WN&|SFx~ia&kh$!ek{SY53k9
zG@zD7N>w#3Ie9Gh?h_W>>Nb;NElo}NWZoO(<a`zbcVKB)QP+wd{rVy&BSTb4sc(Gz
zY-6hN!Gi~|6J%Y)XI|u}OEnFDeZiR8JHLIHoSY0m6p-c3n>P<AR>&dZgxN>yyz7dJ
z*l*o3nD0r3`Uyq9*rXdWMYq)A2C}b-5_Kn!|ITb!uJ`~_l0;BN664`1LII79-PvrV
zrsj2odIzf%x}qJ81`TgIzq}l%<AZzmI(mC;e`H80qzVGjncZ}O?q?(sgAF~#r((vw
z5(v-G_wSgvxBzA!0U3a*6o5s{`QSl&1U)G#G}i#E+S}Wg*-rG#+6x`86klhMmwkI5
zjeLHBN}L-(^Ya=7k0QV-W@c+MvxV8&Qb=B3-wRfsJs42yaj>+qBCDWqes;VLUHtLm
z$Entp=4NIadwcYJ@965`ahufS<hr2?QuDjK&&va_J32UMY;GRR{AT?sJc&Vebv67=
zX+}oV&`^10<wJgc?K_0m$hpGcQBmmKprf&omX3~&n|uGqM;Tahs1uxIt5xf@w0!Zu
zQK8BMfI8;Q8)E>i^bUI~NNC5NTfe28oXWDZ?;-o>a8Z=T4xmX=YHFQ9b4XB7(A_7W
ztj4P~<m7Oe&-e9FS7EqwC#Z9l1#}UJVdwSnG1$qFkdT)z(U6sNH&D+Z@8RJAMcCcF
zDm7J41oe6e$hfXf2=xZkQt5sLw8YB`?d6IukJpKdiOHwn;NZx}-q$xDL%4rNEw}i;
z35}dDswf#WwYJU!DZ3Tb1`H420}|=l^_K0xe?Z6|ssNj^o?by_rUafSqya?rr%!W*
zhOIlhyO5bkR8HC9lap%!hBh#mUR@o7Wrb$Un+((zy~sg|9~I|JAZuh~W*#3O6B7}U
zk&@<VmUgtXhy#}a(y?MjfQwp^snJRY&;nxP<0Hevad2>eAr18PJ@h={L{?@YfH%?T
zDJUpF6M=MwkWNoGw6`w;LIdR~DTxBvmxX{&p|iBG09oS6)2F{TT>y&<*#DMvcjrs&
zPT_q4n{(#^>OI^0`}^D5GEj|JSX31iF|S?YVq=45M4c?pz(*#Ag6niuf2T5Hln{!D
zgtW9Y*fCIRa`W<_+%d=}4I`25M(og=?m`)-k%Q7LE!{EO8NcX=+N#jzqq#&|Xmijx
zpdUy>EFsJR976~h>gxlc0u|*~e}Ci5CID!V7}_Gl@4kiiJOuG1Bs8?{bWaPw8>nC@
z{I2XgJnAAMXr`Y%)=`C;02BQJ+yZpVmX;RCFt{cby6^UUulH6vgONl!4Z!TzG@P6R
z&CRztIb++6K4oNV9C}>5c#)c#8WOoET?r3jPV(E#CWs!GzQa|{wr*}mKl8MPay1N%
zjm4Z;EE+%q!O;9WiMfM?Li>v4wiEJ_l6UUhc^wnuU}qN`5dpx`q!=j^&vKiO%oxg+
zzVOG7A1f;>*+|0RV&mhw7=B97SYu=9XlWliadwa$^vSLFzoPF8ycbe5%z$-&0cg-f
zg&i=S@Wv?@7nfYE@`!dLdmaih9K+vUoCwCF@rh_><S&Y|QZ@C&y*JG+&G?2b_ZYLh
zs|AZP($A?W5KZC=3caBB!fpWt0GO4R$DG=Wi;Z1USa_3<ta;bHFO({I;&&9^i<#`Z
ztMN`1IzM>5GC!uvC?+p#Oy)e8Al%*EZ@lO;EGZ`kMS2X7&)({YLK4qSJWN?RxuHC*
zOsJSnoFU8m-YmOu6LX6^schtY{=`bWd&Lf&C9kBe`!8ATGjmN`{>>fg_(qSd0aEpU
z0OzYGPcKSl{rI8$@L}cAksFj=5Fa2DJ&*S^w6tQ9LyrJ!nqS9p>`XsjuZ>xjz;MeG
z(HIoG-rA3uwz@Y+K9(Nuv9Tml18oAjj>Bbc+i`6*wV}FTwD;6a&CSVt&hp~o^xWKP
z8XAY7egcvK763?{5QSl5RjCAG%SVk2zKoAX{@&&$TWIM?l=a@Q6j`6Pjghs}K=Y}e
ztoOSNt*6{(%--2Kvw!iz%Wo|$pMW>ry&Jq@=GTCz?q}xs$+eKF!1Y_VmoYo(*)Xjb
zwV?e;Njlb?#kE$Gdp`A3nW<hypj7mk)z{Z^+l;n#chA~GBvcM7KfI4&CKH2g7~S3>
zMSpQyA3-Az9hczxb)=P(ynN3q9P+u|)Pnqcuk~?(k`tf$aTUaZ#G?=20Y&unMfJ{u
zOpq=SDG0@aiK(l<KL&0I6C(NuKeCma!=!6I)!W9(>SRz&4>qr|qQXBQV9K%S;^Gal
zi9wgTjFP3kC+5vj)N)C=pXs>2d`9@(0zNNH)7I0&#m){D9PesCC=s(}k<k@oB}V*w
zNNxJP?Lb?<M?LdTWg>qhWqn|>n2e7bx;{hf(XutVaahfE#{=PF2_hk+wmx7Hz^tby
zCq1b`-p6~^R-Y+cyzd^Jdn~%_Hj3r!X^$7=(4Wtnngpzt1`H8+ZM+E6aoWj<AsQD#
z^Wk1zTN89#Py|lr;c<d1baLOY1)GC|<Hd_z&<tNwT7N&3>C?Kbt)8<peP6;Y>&TJ3
zcadwEjGXs*nGC{(@qeX|T-@Bf@M>?bblbhhaw4#CraXF2nlp`WexzW!!a7ap+MeLF
z$GY8u8@C9yk}7G$`D4U}36xyWlp-P`xE?$ZaNEfO1_ZnnWLeD=Le|{x=G5Wl$2RS<
zmUS|-lr@n&nYBrLo+{IqpOk(PxQrMl3vFg+W7DX1DTG2TOfw9W5)|B_0zCz%e37Mq
zv$&sMaJty-1m{l=SFSKcDfZ7Ehw*hX)XXJuf-DQTX#B?y15?v-D`XMi+-reP6czFC
z@t>gt-V%|e+Z_9~EhiOEOxfK8&wOX^7qUE<YY1QJd5MdGh|Y!~PhYx(X>W?>$w6U3
z0SMdg0KICah_VuT?w8^eClRQW=~V3+eoI*|_PVC?w9x%*MQ4SI{Rf&)My1#3*Yfgm
zVVZ@N6%cMZfFShtQVY02H^L|-7U$+n?pCwqDD_GmY`fEKkUTs5+~(?+F$~0b!`bOE
ztIpS_Dk`(EbLKzj7#J8pKmj=%h^+<ed)xK94m!jN18!&T19Lq$MHSXh4^n;Y)*{?o
zw1^P8`A|r7b#;}LKwjeJ;Mj(bGd4B`EyR&mUP}H`T}fGjpmSiUclxLH3ypWLCtNS(
z6b~XhfcJcK_O4CQ5jZNo7#R0{7Hx7C2zi+mvSI2u5tI6Ti-9ZvTW03w03tukO6|u_
zQ|$&jAG{zEI&L`3`27UGHHZ9H2I13Ixs=`D0GTfn)ses3uU<hA=z|IklbpW(eo*Xn
zVIu*0L92pYZ11_m+OzQSRXF#dK~+2PNV_DElVz0JBcDWr>DzA9(%cMW=-}W$T|*-(
z8B5n`Px`|b1-hf<n|JS$npcx!Aqtx2w*iMkJv9Md20;$0qQ~JDaMI{xEX?WKBaBsN
z>sTLg&n)lgsK(JV7G#vNRVIcxB51Av!-Qh@KMd1*^4m-3$vr*t7=9v7oHRl4xeK##
z5?s!rXZyA_d9#bH+BG&y>~=lU2$ujTtj(>hfbk2o%0Y^U%J#C`1Z1U{WGt`!S1+;U
zl=|)W_YRxmw>3oz&gFseYHDe@ySv{YB!p|Q=+&-FPCh@_m;xjO_vy7Coko21_-tV0
zjE1{LSoQH^eA6m4pZVRry&`ocmiVe?jTnZ8hSvgV85seT=rHHa3q@X}bF&loem400
zDD}6H5$`86oBjK7w`nO6LFNAsjP+!(ne;T|ZKeR5kth(On5Zbb*^V1PeEt3XD=K(k
zKJ(<s6VN`39P0H>J}!)CrS4Qa60+XrleN1eNQi|v2N@1tT~In9s>jAW`#+(Y8q`S|
zK4SiL)^zenDKb{>_8DnnjVoxv6n2%`#KiG~IchmsS@%i80D=$!1)HB&NaPA%G25uF
z)1|1+KNK>$1UEzQcY-^OBT%@#<tnS!d75f-qKV`-FKo}UF~3oYyHTi)I0*ea?Zwia
zZpI)MsI46{n@%dde9dOU`E!VKhsE@$^<wxiK)T*P^GX#hEph^a>Ho|t@tSs*huD&d
zwIJ8^Hup!azi5=M&~Tr_xoKp&)*f`qV?F<QS`d}$2f;x>B+!ojBl+>zj#sy}Nis4r
zf?_})Bjv>DQb`<ATb`C{b7a7wk}x4eb9;AcC7_>tU2W%@BRc|5;qgJk*?)ThKzqD)
z?b?4xbs~skpzC4d-~hsxb7FD%NF1W8I`o`?fwSU^iX<MoNYK{~!-irBg~YJ#Hwa-o
zZI<|MlaY}TkV&==JpclHev-Mgu<#l(XFeb;o@(8xL+><DY=)Ej6X9jPg|NgIK^vz|
zl=2oB2-rW!ABhMGQZ|Doke-v0lZ%Ur07)`2F||ct6l^y<swt>RF5DSOWW2U@wX*cW
zyXd%-HHoWct_Yf-ii!$mtul2jEylZdFQKCYVOt$8q~+yxgAg~8@Edadu;5<gVUqOD
z`&eSWy;wq?o)#Ywl=F|t<pu_8n9V|{6BXH^_lSZHYrfa;!GW`!b;TOyH=6RlQ29Bi
z5#uY(#cK4&ww13FzgT`OOyPR6?95C4<oQ)}9}y-%h?<(b0s=s|+5xvJr0^#^uq5)U
zzt6@7Fs>RIYFbd@+8m50>U$K}5M^<dhgIO)<Uc5Xy_V~-1NqJg;qwJDdLD$If|OMI
z;r4t@bv1~wfIeEo$unV~o0o^$J*{CXTtq?tj~~->b42*~^#+Ap*1rgeh*Z_oh)GEk
zUcVNGSkdCqB86eozuABN2udT+od<vo$qFn2FmfO+b^hni$WlvUmJ?<Tlt`@h>wi!F
zbDMxVeCtF}#($9i$uSu2OwNKN0P!bwt1sjRBQ)*^iic?*KfXBFcm~CR&;2m0HvLsh
zbi<V3lB%Mqyq>-}zt>wrwx*SXx+_*`mq{pG_;rPag<)2wto);u+S_B{`-l8|78usT
zk56n)hv=$P&gT{!e?Fg0``U6jApQcy&Td^$9svRR&4728NeJEC-C<ggo|Xn68OR+I
zGxO@w(pRt3BkPf}C`z2O2x0|`z5d_=FSby|vmR}!5E{%!1F?nXXGbwfHzr7?qU#WT
z7vnoY;QspRmAH$`0ql<LXhjqh{`mMn<~IaO0XIB*Z3pu>y5{emI`6VPm^O5{_{;%|
zg$(Py+3(p|7<5@p)FyY%dhH=ip)I|IDVwIQeT~ar2U4||2d(OQXG#dBFJ9@`dEdHZ
zR3&ZPHo~vz<;BaeD=<L;9ty$0&dz>zy6$Z#^HJ3OnPb0}sxng7WwBq1A&{M9r)O?%
zPY^jpd;am5Rs_#pb%Hoq<DSAJSp*3l+~wlpBFqW+sd`67j*)TXM4`L8LwYgr<h8^h
zUO}4?v5C5@)B}?3j2~vvvOp^R-wb4%;-sbar}w0fFn9YzO+EX1xz`0&VxK%C>q_8~
zraHDpWEK2f)Ywp{QQt|6BiAb^5%O?8;2g7~c06;++}Mn@-XwM+^9(Aw{GJPjM6X}J
z9vvM8mX-ht>#tuw%u3d4k0VyJ;pbiKHll578e(xL^0o7%Kqp?9mMG9EIpovoVWHH%
zMd9-mC5{;zgB}voo&+@yIt55@(aFhq+1b)goCle@!!P>X&&MCnsO<G>rqEzkZZZm`
zPJFAH;QLmO&=<bVz%V~QU!YzHx}^a)0>FZx`sB&u&L@q)Dajl<vXVA}+)ec4$)xen
z`;^FbQ*)y@JnRGw5lc{Kr|m1T0NzwQpBk%j-d~?6NJ}%&*B6F-71CX>O=zcL5D+Zp
zKDnbG-81}Q^}+4^uQ+uf+i$}vaS$&60+<>bmzI|wA09$yhnEY`-s59H0ppRx#7oK{
z#hwMR4=1Wzb4otQ$Ok-Z6xMRZywZ-)*LQ?*vZJGssVO-b8I_P1z#WjgsHszLKHi8?
zz%SsD+r*odqE2Kj)v!2Y)1>n0__q>+Gfo~hj)w67J^^IQ5Ybz>xVX@Qz=pzR*fM!t
znGp1Fl7Xe=<=vefOkW_O-5ni21_t<@R^Hau3L+~fJ#S52ebSI!vOW{~731tB+GK%&
zp&@7jpruyV)Eplj#j)r}XZAD5zC|rf?~%6{E5QKajUlHK*S(~1y<RU*UDF>Dt09f8
z%a%FQC2tWKLBYXN3JM^Vu)?UCon5ok;!SuscpE}sev9mz6d?CW^L|_88SJi^nLYP*
zs&~(<OvuX2q>3*!vv;$KXfI)jQ9kec=%}`m(&lJAl(#(qp0@fx1OFlPOr9>Er^TmC
z#<@7`2&&b2_|=&1F;TJ(#Z*4q3PuVs<ezPnmLEe3{V@u;0G0wCbcTg>UlK+>13!O)
zV+ZJM>mOQZ*BQqQg%#vRXkq)d!sB6->^8yPAtHNhKd<&P>YARkRu6^~v1mi=G|XSK
z8LzI)&tE%bKqrv}KZ}*MHB3UlyJ2CG{qki!FwkMEJ*W%NOivBIjfcG<J!~%v9a31(
z(X#w{c#lOEQ}~v*x3`c`9SF$K7%yE4fJr5c@d0Y~E$(nnkkEbiu_=f(Ajyqv(|k>l
zDAz^C`x+M;aXtfp84V2$R7MyXfHnd_4Kfbv*+sHDhLih!xcy~E@q0c(_YnNoZ`=Sl
zZDeGWk(M^Sz7DF~(0@rzmrjHHJ$Deoaj*<<{myvS{zX)XLcQtcp5Wx<l>LL%zDI&8
z=qIVIJ(?om4&w`7KR<wqFlK-{FDn}sOPEh1Va#*s!Z-cC!9f*m?UU8=36x%^T?q|0
z23R}X8;{KGS(xSDzw>X=$<GUHi6n%78J*WN;9b5*DDdt5KO`p-Opw16_5Ln7>HC0|
zDs1;}$q5?|#INRmN=_s+pkFI$fW;`BoEuOm4@E>^U@Od|-pbMkHh$>aC>9DrT}cBF
zA;c7ThzPC*fPMnfu!o?~{rlfrT3)S~-G9HKMd?$d{N#z8oLsr}2#6!j&}Ux?>swfm
z!OZ86K18Aj+M%K<cq5XMv|-@#@uM+_yvoYaAl;UhK0sFLzlXIImV-GQNcw>6+q=7A
zQY+{%Ckv3`PpbQ%gSvRt4MrA{l5Jpq8XX%0fl-*|OJN~28!2h&TYO}%QF`fBnB1mX
zSy`EJYCeDd#n10zOuLnZg|xhU%YzII2GC`*0N60Iu|;*8%tMzP{_^bX<Zx+e2^53I
zmO~Nn)d=7Hr^F&m221;K!=)N(l*ifOJVdRY=dQ|=Co$k&P**3Bm*9U3g5RBh)|o5l
zzTo)*{Ss*K2gM4rzO>siae!;v{J9O$vSt8UKs5e-&I6+r;DAslKvi#!ll~S$<#P+f
zQ(Q{k*n|X9YU=YwTp`e=8)4+AqjL(*5+jL=(Wn)}|6g^E|J$gX>@_yL=hEMljOf@{
z@SOR0puzwxw7ICX6yEuOojtptV4~Dg&DePMXm=UNt^pcooWVo?tw}h8bX!)&buI8Z
z5mA}b>MsSM)7TX=8Y0Lsf`b38_7D*ea0yu{tPRkNZUp!VNFtXmh?tsY#>6}UFM*fW
z8H_Ga7eWU=O?#R5`5AO%Kwp(rRSNGo+uPfrBD3DQ<-4LyBaXUC?%%WmNYWDk;4q?-
zkO%~!F|bkNi~dWvii1OOT%0PH<TR;L{B8*NWc;f2JWhJ(0CfW<2>{Vn78bHTeS*i)
zEOK}`3`kQl8*D~NJa)b`nDjBE#KdY!N>ej4|51hU_#hU8|3%@Oz*Hf;^?QCET9AAS
zKg>Xw1qD-p<^1}EGX9WY7p1@Cc?4HU`3EI$B1C;iNm&_`^Luyhh>D55aCQcshV1UP
z{^JF_z!tFj1`(a{uW2w&6IiB&qbc~Dtd4ddQx?IX=?EkOY-=z`C@rPrvLZqD346du
zghmPwU%qzbEk?#f7_kx)gCohv!UA$b+KC0v5>i+)48Sy?Z<yoHv_^7}gvljxk&=)=
zX!^n4tij$yzwDg-T2Ya!Tcar^7ErIBt6f=?m32o*$or3x>7rd=AC8X^CKS}GSAPr*
z34tvWv~756eEbloAAqOYBQU1H%1uKmEhF<El(B-k?L}ActYG+o-wP(suMrTlL3~k?
zO$LR4Fdjd8L_t9Twku5E3a?Y%Pum-Da4XX+fU97PDRh3~U}!i~?Ya#Hti0S@*mJ0G
z@XYKm5T;p!M(MON{Nerk=V1D@9LgUY8UkfzZM-HC9*GSY37+NO2`b$cSX>V<A5>IU
zQa-ddY6`-){F!U#<kZvA5fU6+_~i=?ADJWpgk+N%v>z}(z}N)ljTbIlfMyHV#iyhc
z@Hlu5;XnsL7gX>MZb|bD#!h%oz~hvgpAWMIkjj8_gWG_ZMf)wcEfr+g7GzkHlDPP*
zk`hONbkMxOdVw08SpH<f@Ox%p@W_etH9u<Y(*JIF#iAnsGoAr#{_m5+ZP4VPG(nvP
zXEQX-elRxj5lh~KI@$zkeQ08$7Pve^n_+jKfTIjp5A^9rk1oLSDno&y(S>jUJK-nU
z1pCctF|Ne1A3yBDU^hHGtduSeREX{dTx)vvZ<G+{9A$+xFakp?cwKwvot>P(MK&}v
z6zK0i>F{L(7Yc1t^Xc(EphntTx182SNl8gRXJl}~F08GsspP1lu0al)qz(9hgp4di
zG!6Wg5N5DjqB1heFnu<fzYxq#BQ~KMcaoRdOLYF_sZH_5zM*IJa80I7NeIUDCR!79
zWGfaXCJ)$AA&kmw#_lmQj|~n&dxaU%d*3tGqw#55rf_}&>ON!YlhpFysMcADTfQ_w
zwH-UJv7CP$olc?A3ESEh<m6aBeH#1u^CGxb;HP1^KoYJ4uUSz#+Krv#{6xX=^X4)y
z(*5nBSN8sGBTng8IvMPGa8vKqZN36W0q8X@Yx*!a2N%^b!(D`Vh4$LcV*dk9&W6Uu
zUeMWLmteW8tNE=y`|qbssO>bAWl~BnSrnm*;3;>^1^ULFeswUFrWNNxc!STiv%MWz
z>&X}2dAd^E3np+j7+;p~SF@hhro|6FNYy)J{pmGGd5!*6jLHq!Ye{7(<sDvX-=^uO
zEN!#W)3>LM9V@hL;3>{OULtY%-+FV=%APj351exxJu{IYIL#!a(ktH-+IPG1NZRgf
zs$@g#Mr&kdiM|G|pRTE?Da7plgf}(#dC16)tUBc$KlTGJAoM6+jLDcnW$Fblua)Wf
z)-xWObdxgLy}5{SClP9$>~m^r;V?+TFe-r~7}x+WgPs9*6&DvbtbApiI5<=IJr3oo
z>-WFCSJl-8zptjU^7QiZ$ml2q6%`BN)j%9_ao9()r_2a*etv$i!q_`FfS7Fr+7h_y
z()((U*T7|tHVEm}*7r|qFWTTFAtvItBusC5d-cFnEh^dsZU`>habUTx;sVcnz*Tbv
zN)>Dzn4kcL1DYR$00xl}JneuDvaUeS_66}6w936bCqUNlZZHGD6c~Uu+{FVZ)%U;u
za}16#03&P?MexYM{02?|RC^p20p^0%17+XAVH=_z%4}>vAI>+Qt<BAcu3H(BbXyx6
zUy6$Cot&sROt1xSUOENm!2S1sCtp4tP>)D>z=R79TzYzXzz0x_pmTrhU%XB)3!<tb
z1tjl9I`}sMfw75+x-Vb8x3!rKW~+eF@!q|AK&k-8pbqO`zXtz{PPL1YuC5^b8z*PU
zmoJJE67vw+key473mEz||HE`3k$HG|=@}UK9zH~lj6euOj%UFz{rfu9R8X)8Q3Pgu
zyv99};^B|+@g9?6@Yn+*Cn1EcaN095G11%03da^IDqxy?lZR)xsmW-f*3*$U8Q|<4
zn1g>`UmLBwj)y0l-UsbaSyMCn{d;{_#phFxKmgKzc|$<>9Bv*S0UBmeSs4ox6Cec`
z*D%Di)A12w-@4=G;}r`|?C5BFa8&aqLn-PpD6ECk8sKn(vjK3!82E$&!a^|N{)?lx
zVK(~NB<f!D{xt~qM+29Za2*~dI#>`P?t!X;_yar$+@x??;rw@u0x&nfYavwc_55GG
zg3b)TvnO-`@)Jz8)z_b#nMu#jmxr}+M6Hc)>kL3qeDp$4ZehIvX6I`Zw?Hdf(Y{ae
zOdNOPOiV%o20Ihgt}){2Fv5Ju#|MWD{v+06po~Q0#bC~3prZp(p$!IZ6%}wK1O^Dj
z$d!*W@zL<ucY*%6Lt=u*q(%Qe&<@bbNU_CXjstW64s9U&^j|kDU`%>i*xJ_F&4_Di
zk_9$yZfvNR+fc@|16JlF^ZMz-31Fm}``^SQ>)9K$jd5_JLsZy|lz~5a9pWA)xB~;6
z;DuIGi-+ybf+n>w9vc_e(cTW7AN)$-T?chC;q6<{g+UR);6puCBuK?5yINb@@42~q
zjEt!d9l+l&^T*wefZg7o`sbh+1y`zZg-&;3z~d1vj&<!C`2E0l2$TsbRmCg9po?&o
zv(G@Snwmtp5|d+Mik~&U0@V*JRufguxa9L9vQT`^@INRn{CTy_Z)-69gfU;LpeGp-
z(Hu}9Cr+OUXoqKixAHbF28nq5`53B520TnazTshEpg#FOno1K3jbqYy<iz45G}W!{
zQ@8#%Zi2@ROMfB<=STvtQFe57g+)f5PX$tgBk&nmabQ#pXL6vKom_&{_wEN9GU%|2
zjz{1R1ou05CKnu!Kqv!^0SeFtF)UqOdS2ctBsz!}pgDuN7Kk4gcv&DEP_}lmZ`)92
z4*1lW;_~vWEiH4RqLe&6Pr<<k+wf<e%Hk6bglRGv1TFCDfg1(fxn+5I;F^GWcb-nw
z9(bt1gz-%Vo_t&b4h6x;yeCEA_+ZlzOyDIYoX~Di#ytnLYB(}DIi_9bH8wJ$MT6()
z<iz{%p|Oz>;P^-?fp;s~hys67fp1mFI`4A`XRz%73uTad2j@7z=R=H-575*{4Pbq>
zAIQaU@C>XzFa?K_1ZLg%H*X-WrkLK)^uPnTXu(y%a7uB_wBGj;7&F083TJAdfSS4D
zIiR9DZ*`=+H-SR}s1F~X`{r~rjH1Dc1PJvuAMrPG)WiHl861*$MDWmK+QULap@o1w
zzp$w2)_=bNJo^RN*}uxH6Cqb&dL10x4lWs>*kF|P;fBInt>NWWTTrkLND=zy^vnzx
zA7PoHlKIsGa{ZPF0W}UvvM2bp!C?b1%+kzEQc_aC7zwfn)Mgh(*g|g0zdKehZK!^!
zA((stX|BOU6yG`a<(VIxNDOR@dh-UNBOlz?$UX$>Aj;&62ueOKu%^QcbTcS>U!lN2
zjsNxQk5kPIVT+Yc?lS~Q_&3igVjK;^t^f`Pz-w+{!R6$@4#Tg|yr2H!ML$vw6KJo0
z1_Zhh|7IyJ+^B@rRz$sCgNYr^kby^x@!maf?V{$qkou5q&W~r)z?=auDqcr3@LMIz
zuP-e{zj?E|ybLFlh=N7YMI^zqC3@@b-OnImz^HO<%?7+KaCF9$+jUdH)%6hm^XB7M
z@Wf#82ZN5#+7yx)iXt%N-n9_<Ob}bEm-1groNSNQxR-zUFa<7L=qvC7Z-U8v0JLNo
zwOMFq5omXh@+a)mAEhaKvwXI3`E^_D{5hdRi|;JUc~KwfxWbSU+zx8{tgK)f0WL%E
zcJz9E(K{&<VwpW@Rj5_G<METp+vBvg`(sAN+5R%|`UcYT%x+s~?Hq}7UcvY~Du^Fk
z$^Q|liwg^pU{nPvUw3u{Z$%S6+4b(*f0JDH0B1pSi#70pDp~f`qK1gLug7(|r*3vG
zs_IUYkHozlJpm8FPtwc1GkW%{etsJ~;u7IxAfBY>=ALX!J%;m5VAm%Lr~d=6!Nm$m
zF)%Oy&nbAckPc>A(15{k2!&HHRS1refv*w=Cmcc(#2DCi81!sGC8iY*GqbXy;IdK&
zkwKD<47Nv0D-pmA5E$^J04ZvTq}GFre*O%eAaFcuC@HadogBdDL<%0kQJo8}O+t}5
zIj~?_rrqyuz?>^OdJ8<F90^@sE~OWCs=T~8%JzG3O*Kbj3R1?qkE{{`kQe&JR6FZ7
zb|Sk@daEr>m9_Ld8+Udbe`0}2`3y3QThL5iKEluMv5ieW{0_1Q*p=7EYmS?V^`5G!
zX=!L&LB9Z=34B^{kAtZ|klOO{^5A^M=j?3FViP$q4TA9~u^R;CZh-CJ>nax7-r544
z+xv7^4UQfHF^7W;P(iy>h4f$o_V}^k$-ySNNLQ-RIdlq85nv$G-MtEqe86f@p1en;
z;8qF>SOXhHsDj(JJ6Y@e0|WQr^FeF+{{1;1)AdEu3;OSOk3)2&oiOvNYHIc?0||B=
zo$xt-S$g$$B|Y43<`563<Wq%nbm6#1R)^5kQ<$fz_+9QBwY`RW=`$;tn=^@$8~}6z
z#|j97;ZICoGaH*veP$p<P_P@d!O?@g`peX?eQ@ybBvJ@w64^O9fMZ~vfT_U=FrZ3s
zb_j3;=qE6q0_7Hb2@t1b<m7?aq~O$qbAx(1I(hHkZ?q=gxpxn;-)p1(avLqaD_FGq
zpy9jkSw%^Le!P`m;DKPfwPfo#*msSLc2{t$Vtt~1KFuJF_S|7<fkkaQIDLNQnULoY
zAD{QxiSvZ#ev2(6zGlU9pwDTgrMn*<2R?oJlwUvq;&vBy4vwsjLtKLG3|s}n!|D<e
zs6$2&u#np6;;7J6R#gQsx*vp3ySF~ktA8KNzV{89FolGK0EEH-I*j6Bbh800;!ijl
z3N~XfIDs?;2ZlJ{R9=%2JEUD}7-_NBsT&+V08Do8?%h>nabUeZ1;5Mddi@oc0D<lR
za+N4-;&pOz$Z-!iWdgDz;H<jcrOPf_=SM?zaC~q=!TD^<IcuEbnM2=fRz-<k#UPHg
zzz@e)yid-rkEl3_8;)0VdCO~SYa1G-!?6<bx|qf(baeEM!xQMu@Z?U8jsWXGh60-&
z#1q(qFasX>(qy~|MBG3iFh)?{3-IwHh+bx#zluGHVG;<>5BCpe4&bP;I~OGwPhCKV
zg42{KBW;^2`LLT;e4!R`lI6kathU!Td!79FM;gFim@;;POr`eNcqvRHOeNJ0FjdgF
zBNpat{E2&romAGk6MOK}1GYQ0o4;7JEdK68m6hLG(My*sp{`^;Zm_TmQxhPBnWkb2
z|IfZGfZ-W26{(=d!H}M}C$!!26R&H5+guDH<RvLOMt&=V(^$Q|%TtH00v{C3^z}F5
zCbXYCp=D*AP>{m-t<C)#6h{#g6ME#`CgPOTp~mfPHy1Bm1&<VJ5P|;oB?WhMcR1hq
zuToR$!+UZ6>*F_Fz3$Z3Y#|>^waMSC-8iYH`mfKM-ab*Rh9itV(;#;v;1Iw+zw1C=
zTd-pHY**m3^=s0PdoKsPaWM}b#w**FXLl88@|t3^h(Ch)q8;Cni?YUgaou-%&!uG^
z9EGlRH}b1bSZAO9pnu1##d-LcDd_BKL6X|_&bbduSn4};$mQ#)Ib-Li?nBJy0S-gS
ztlm#OoCa4E{MPrQNaeM6RJ5;Q_nfSnk8n^59ITA4`|n<{G}X9#@^pWsRL7)6kTlQ5
z<m>``BEirczLLCc^K1dP2d}k%*Zo$k<T`5iEpK^R)3v&OW+)lO+;Rh>vHfBuqHKMx
z#;azDK}1D{L8;@@ei#-Jn{tn$vewHvh&tv>sW97n=u|g4TuJS0sIKONNh^Di<@r`g
z|DglZVCU>8nN5+RRvZoHp~jr}YTe1xfCn`jN@8#LAHOCgAaUW1(CHU*qAI-U7JdE%
ziM@S^S{mPf=T5QMbu480*9ni}`ilig@dWd8AN2cVB9qK(Jes2yY0pn=mn~feGB}kF
zk!bHi?#*mAV%*d?Y!cT~xm36G1L?n&*&R@ygH23vP?@|UIW2N<pp>2zvQ&ICr=%#`
zzBT<77dFG0zgmw>u*0_EbhrAmRfY3-mGRrrNBUc7#vD&Gbhnsq`M!=mnz4+M<l*tQ
z5wjk(jJJ7!d4+WMVd^{CBQ>6p&hL}7uD%OpDqPx*E77Lajn9iXLeQcXdeyK*@NPY#
zYx+9r{T<fD(%=Vw>>W3yV_GoSt==+!)xnQ1Qmwwje$PU>JCZEGKMq@DN5IC6luaqY
z_dL!OOMtoV%|yW>tyxp7;9a@o!+Y?>3TcKsa=bI;;y2kmo(r;_%V>(pRdSRU>WH>U
z;WxCrqW>C_-Ow3_>-Q%1s_sKu39EO31Y)e|rB<O6MMEAJtvf~&m;;L$kLF~f7r(w~
zC*?Kmc1b`BWf0FGanNz<_1t?gaY~c(rW)trgvMP0QClSC+Uwv5ivx|@n2)6UHfDnP
zIP6D$GCje2@Y^VEi8cA5|B9kP#=6&c<!bx!kYP^T+aYJ&ghcMP{6#C<tnQSpME(r&
zn=%g?aPzy+M^vuQ7v>&o=d}v<L{zgc<#f(9wq3hTWmGa`!|`CtX;}40YUJZ<37e^i
z^zVk=4UF@v!7|&5dkwVKb>d!n=99&bj^Q|hm_lc9fi|<1N^-H$;!59wE6fSf);GtN
z%^y0GFUZCitGadgZRtHM_`dyxW>(s!G<UDTIJ6|t@cA}HwY{zwe_~(}IJgkKlX@Bd
zXXWYt-PfLMT&P!&pwMYt-^WMqQz0q&bto%^i;O~?p#d+)dPT~@ozMT}_L7JKk~H<b
zF5^ghk&(oOgUcDBxA2jqh$aQ+z7Ll=ZOy`1EsvQ_r+*cy<G-Xs@d$1ASFMDMd%yIy
z_GQzfk}pkC*FU;6X)KzT6sqxP_EpBgNV1VB+Cd<w46J=UoBQ3RAsxJ&f<NZ@bXk#h
zh%-<_zF~d68euKZe&Ci_7}x(nPMN`@R|Ts*0(}D)bdU{xB}-R91<rOw%L+mf$Dy&q
z*{GrqDPJj&g@`j2n8+cTQKvfozxo20bKKl3H<(E=c(&nNXb@6j@}l{V^nL#?Enj~4

literal 8250
zcmdUUXH*nHvo=AJEFg&FU?eJ8a#RtKsASM339Be^Sy*yjkSHRcL;)pd$r)BaL0K|L
z9+14Si{!X>FudP)&i!}quY1nSIaAYJ-8Eg^Pt{ZNQde7z<}BM;A|fIh^?P?75)l#a
zo!*qB;G5VTIxQk1l4D&>eN|8-CMG5!At5CtJ#*#^85tQlIXMLd1tldV6&2OlvuCNP
zsUZ*u4Gj$~EiD}#-MMq;=;`Sh7#Pl<KhMa>c;UhYCMG6kW@Z)^7FJePHa0ePc6JU9
z4o*%^E-o%^Ztja0FJ8KI>GI{vSFT*SdiCnHYu9*qczAhv`S|$w`S}F|1Ox>IuV24@
z<HijkAt7O5VG$7#QBhGbF)?v*aR~{Dn>TMtN=iyeNl8mf%gD&c%F4>g$;r#hD<~)^
zDk>@|DJd%}tEi~lx^?UJ?b~<m+)-6ky?ghrnwr|Zd-v4U)$iZGuc4u#si~=@rKPQ{
zt)ru(tE>Cq!GnhnAL{Ap>Feto7#J8D8X6fH85<j$n3$NFnwpuJJ$m%$@#Du(DAe5C
z+`_`b($dn(%F5c>`pJ_gHa0f4wze=B?CH~|&z?QAv$K2t{P~L)FYN8@9UL4S9UYyV
zoSdDVU0ht?aJZ|hD*}OVb8~ZdcSj<T9v&W^o}OM_Uf$l`K0ZFazP^5be*XUc0RaJl
zfq_9mLBYYnAt50s6e=_{G%PIa<;$1h;o%Vx5s{IRQBhIR(a|w6F|S^|ij9qZ{rYuW
zT-=*CZ{p+Q6A}^<6BFOQef#d+yQHL~<mBX(l$6xe)U>p;^z`)i@84%+WMpP$W@Tl4
z`0(N5$B)_B**Q5mxw*M{d3pKy`2__9g@uJhMMcHM#U&*rrKP21Wo6~%<)1!%s;H=_
ztgNi6s;aK8uBoZ1t*x!AtE;cCZ)j*}Y;0_5YHDt7ZfR+0ZEZ!P(VstmZf|ez=;-L|
z?ELcO%h#`8ySlo%ySu-A`}Y0&_nw{}3<lHN+uPUI*WceiFfcGUI5;#kG(0>!GBPqc
zI{M?skFl|_@$vDAiHXU{$*HNS>FMd2nHc~8W@l&T=H`C>{5d~Azp${dxVX5qw6wgu
zyt1;gy1I(RV%OHz*4Ni@I2<02-`Lpr_3PK>=H}Mc*7o-H&d$#6?(W{+-v0jn!NI}d
z;o;HI(ed%|$;k<UKq#%_)CLa#mD4>FI1v#w!|6@jKk0i!M8q+ten&+gnX)!XZTtq&
z)Z*-}v#1Y~OW^Nch!W9)G(;NNqF4h|g8R-fs(ywvoD<0ADYuWPmT+A144G4i7l9z?
zBeeaK!m_JMuUakL>JER($|M!Qx~Bm9e3j)b&}hCuIK+P1k-*6jc6Yy3TV%iUnAY|W
z8&t;<zmgV_&-g!a83z+KSYDU@^lWP5CBxnA|5(><w)glOE&?e<jgF&2pFjKW3Xj8I
zh+3Nn1Fl6O_L3@()7Q83n+*^P>>>>x%4oS|MGGc$(-aU1`{WSeXN73@f@cR2W!0w=
zKS9>7d1(3*t<u($0rMWdZ&DUWs%I&3%fe~nW>amlM<1a=dN2x_<^8DB?w3M`lTO|q
z*+-a-Diug&<Hk#Q2lmEkmf^`YC^0trjST)x-i?K7<Bp#Q@BJa30Uq%h;n{WsZ~M38
zQt`ZYt)pX#6ji-<)6CBKcT)4^vSbSmYTNA#(TP)g@u%anmhrz@Ta>8Fod;h!dadUr
zCcSlR!M~=jTe+;fuwqzUMtG3EK5??f0u^9FdJ4a_N^!3mio&AVZ7clbP>EsRLQ0>$
zuAV&}uAYsPORfgs{<z&9iTVrZ@j7o+znHjJbv;;Hh~per?yn(!U*|44+iGhhySt0v
z5laz*U$mJMJzn!xog)AJhqd!g5XbS%Ta3<K<!^%}!KKuyXQlkwMb(%&?$C;tykTQ~
zWste1`V$#?yqv^_&ftxxyS7}iNCS;{ZUn6GB`w0fL=;tR=@HxeN_x8Q=$iL=h$DKI
zT+N&`t9a6$Wy-B6nkA{QJ92Vx{Vv4u6PMllmd?$PC6rsW6*AdCdQDP!jfL7ww4tc(
zV<q)3A;BTD(MZ5P-W@o3nXH!Qa^pn@UNYZBzNPX8>AJ~SG+?j)+NTA$Te`*0g&xme
zqgOdhD4M7ooTX)%lB~U{FT}O!t(p9ZNdk?R%yZa#cb}*^{RS=ouz!%(Vfk|9QEf+;
z5X%%3+57_Eh;;6!LTn&lAG`e3yPY4y@x0yw$?jSa;kH5AOR%!%hB$gLpl93SeB7ID
z5WJw&@O&m0)BT{*(7g>WnKNbMr`XajIK112uUg=hGSHtv+0EZxt+zy)eY)D%CRK<F
z+oE+t<E!q%i1yY~b>zlq(Bp5<9L63zqO9JM^cIIWn)T<H_jyYSNWO2}d|3^e(w%av
zhCc7D{r=-xUeFbJ211&4{*x~`aW4UT^?1qad$*efV)~=q+woOT#xEHIl22P;jBcQG
zb(Tc=ap1m3^jHXB|E1ICZVgxPP2N0g1YoZg?N{qrGGb+J-C1viG%Hu6$)C&FW9M4~
z*ZJ=Dncex_i*uxQUe-vn2e?|OS6k)1bTxZ-h@+r(QHaFa#Zdk5da#Y2-HcrK#qp-$
z4RDcSe~Nu}OxTz)%s;e7jysvb-kxX8<{ocdbnp!V>~H^~^)Eyezk#+%fjcNe?6W9d
zkT3J-)L|&{V|s&*y#eCI+-Yz}1s*x^(LQif(waMU3J0AG%%7Z2)*aNks5-f;pk*?~
zAiLX&|79d~09a0D2FMnezm^CE=7ohBW*qV}(R9OAfq+ek3)E;gs)J3~6Ldv6v-kC@
zzN%Y;=eO2FfcfGvW;>~<%EQ2gN$InG&1+gLP!<{VZbD$ZIIC%k>``MceV-ETu@8Y{
zlZFL)j&)sIuHs!{&OjsgD*ikVuTH6%d0GB<#5s|MixSO0^I$(>Dv>>=8pffxBk?!#
zdK3nalN@Cv3tu+!PMVzashlE0k<iS$TPK&%P77a%b`|h%mP>MM@tY&l6>r{r`%H;N
ze+&0j_UJ3jynK4-%AiByDIWNTsO@O#;3=on2x-{#q3LFQvPGjK^W<^S1|m(J1<YM(
z-TmSe8E;Num}<%OZX&!e%*hBMY@K+s$*j$2abKiX7?mh`+!$_AQapG6?F2Kr`>h$j
z@?--o20kbR5jI2BP97Qjh)NQ;yW4>;uM|==+K+|zGTEL=^9FS8Rd%}~4N|DYf?J{f
z-|2wK0thY|uvw{-<vH?K)X~G>PgtPYmt8E9vvbr$2SA^{`*rk7QTd2B5*Gv5<T)4X
zy<c>D`{)aj3?i%%I&4@|!m)gN!&_>t6<_|O%U09NOQ!xpVJrT%>!iBqCJ$EAs_YCZ
zF@ph-zaF_#*wyvXS}qu{;gcjQKTf&JTxLRw?$%w(k05$iU{VwS%)<h5b_4l4JYXg_
zKzH-PFI`XiS%gfV6kvh6NqW=bkEh;UH)_SZ#FZ@JoEaYSKaOfS3lSEl)8AiVY?hwq
zOjxx*zLTlF;2SwyCvxC#qrn2Twc_E-4Sda7I09}jh6|ZtdI{}UOQDqL?ym33_YAd{
zE`0aB1`!tIXJ^l7YDo=>7imag<nzkSjI4!A8=pmYhp9Z4x7KpaE!eglv_P_cjR;14
z-j9A^e!Azy7^I{R*jUC>?8E3G!rTQ{P&}(*)>6*pEU3h&P&PjNhWDkqfVd%%h8K+5
z^)$K~hngrR3uJB0UkHR(aG(;Yac#%vrD7h6%F*01V}bGrbdo+A7xUj?y?z6nA@&Kk
zQ%&qiZTO8g{I61l%TK4v<LeV#sxG1uDXLLj?bc~Ad2=sRAi~rLG)A9fuWWxMk`*FM
z@sTa`T0VT9jW+Q!-sRkz<g7<??|WvQ)WN1~-pva+^6;8R4CwBbKB2L8I%S>Jz(CGc
ze19Sf6LT82KDpy&8-EMFfAKM?nh!KOf8O;0MA&0n=x~!b0SdSoK!g!WwRxE&Qu>0+
z=8j(kKv|<-&>MXaeQyzI0!p%^E$%q)uI+Lb=qv+PZ3TWQrJx!}wnRSDNcUCU{E<6K
z<jEYGLxk2C{{w{IC+l5wERgC2e}QnKBog>oc%Pl=D8oEnjel3e8p%KIMrl|>!`AZi
z{%g?IhfL1I?}Y-^ek0*E?4072M|5pTao&Ydz(>K(=0}Yq&&x_;W7<U<T#IyqR`&Q$
z$e~P&Jl-Cv<KrUo8bjwN5~qV8Ohy@{D8wEvC(#>sP(NKy822Vc!E&bHjXS~5V-;i4
z=S@Q0A3Gqgtn{SX*f)I(1%Aj>GWAykaXj=eO+)MnA!c-`dV7oZ6Rgomqh94)(ND|D
zY>3(?e5dlsh5;XjOMBAFW9K$$_8jE})rD>!*|O3(%s=-T1{~`l1MUwYWrVTQ8Rx2c
zM0mwCl@jJKlSioIPX`^Jyd;ls%=YGR3#8<0ccLdn@cDT#=z`dw9)Br1z+8piu^A9;
zkhr-NlPlJdPAlCxeMzK&aX$36b6EM|GcGz68E=|NI)VjqAv!cgTsep$J@h?elTgoG
zEx}{V#)LIBUR2(if29fctgAl(?*S12=%RkS@!`%V#EPuzNHkT40`KzKkMX4reaf6U
zARen-^LE*c2zPXNXwHfjk7uTGk}_!Br+99bu_Je~;U}V~`P>GA(VcH7UT2TG2|eNp
zF-5k27{}9X-!m1oMxp{^LOz#-rFjNUXX#Q;mlj|KxhR&8v90)T`iHmHvdi4e@!l-q
z!ix-EUUexs)nX_Z4{UGh;S3%;G`iF!27c7$G#}h$^dOjfsQLj~`-S;sZ|+djJ<=H8
zu93dObLUX7JDVD5o(adsJv}DRsp9ePzWaG?daZzy^Vl^KL&(^srlRccUOO@@A(dLm
z*qBb&iLTM{0gYDve$-=~eQcq-;dk}E(u&{+WFIqy*-5#*T5iOW7Vr<YI|;G5P3n3^
zrs{A|$$$Cl(GN-r#U?=*?Q}x?>rA5r8=}2LY9~HBU?`@~lodVTW-?cQ%^Q&xbFMR!
z>7@Uo1~XlqYVT=FcC!&Zb_nt4CD^i&bz%uSU81px`&v5L3(mf2S#Eul=KXQQT`Xxl
z=n03DeC%DV_={>;CvwpJ7mP)q>$Fy9;qlxu6!!EBPY2sz3b!`Ly}-^To;9Bg>eb$7
z0o^ire6EzEN9U#*1$u$pVD4XMPW5h#42c31TOLWC0nwmG*UwiNcHk@S{A`ib(uuLk
zh*sNH^J(X-1FbQYZ*HJFM|f$`nZ<OHL)GlwG^0JIVA6uGkmy02-I}0{qb8HPT!|0R
z&`71h4IEDJqH=)Vv9r7JSA3AN(^_7+A=OfW;)8A(m13xz);FzhcXnJXW1fC50Vq<X
zLYhThoqk^3s&@E0DNERE|IiOiC)iEv`tM~q?p$0D9g~-`2JUwehV1|tHtuYg9LF3Q
zbqsh9P@?sil1YB}YG*|R%kzB1SE5-`9@B#_e`nQpDlzJ7(|byenFJa|h|Um+4vtJO
za<k{Gr>-5)d6dzu3=;Cw#SRx^Zp&1k3f~P;5U9lOrN8i~_?=((A0{z)ZB3<mC3b-3
zCH={>ySj%c3*R%J*t3C}_y4Bb_s4J{rNMxz+n;=!{wHa|4NGKfyQZuz*){PqN<t`R
zBmJ<;+9phK7xw~08*&}UZ<V{;DtQXTL;|Yu<z@6M)Dx=GVncXUlWI2Nvrh{n0Xf6>
z*o<(og&umKcd*SbXGyGPe1`qhN1S4R6Y-DR+wf@cK$zMS&6!>Ued-<h>vF~==ccS)
zVIUbiTKj$V@D2w<_nV=B9OLf@g~;ScHr=!A)bcyBCppznGgO0U4cl}$FwD;kw&|=$
zh&ess#l3nJF=zzNgv<>f;l?m$vfq)Bt;AF5%$Sta=$)qUH*heq3jdYv+Kxk4LhEPz
z@MQlOnMY!LRjqE8a&8F0a%=!|H1dY~X;CClv?(;9<h;ojFOsA3G{O`~z<W6<UG@ik
zK9%3KK{QeYnK_0eB!Ko-_H3y&a)M?x7^v?MhqCW08I`FHw(Zti9n9g5pPi2NgNwex
z@kiHV4OMo3(>?3Qwna(Dmm7sWEs)PA*aVnkJQTCFD~S0N3z}K$B@z~-WJDV-{0WFu
z`$MeaNC8ICH+CEexjpgE?{E;KBQEhtsnvF<q+Xk)WPw8dM#Rq?0a^GT_Kt0MCeF3Y
z$45k44N%?ED;yzy#b7=;_4;#Z4gSgWq&)OW3dt}q?XeK6jgcSl*7!{gKXLWk^%!Y+
zke--jr^*?K*juU1Z>V=C9Ij6;nDsA8I&au2q}b=X|464R-3qVKbHZmSw(m=v<mRmZ
z<rLQlBkbkj5z`wsb8VvXHvb|jvj;>wZQ|xS<eAqMlw*1i5q?rGJzRmjDVo_h@c5pO
zP;kWdDajBn+*&r0NRJAqnc1DiGg=@mzw-yP$A<ryZ4}%+gJRY%a<P0!iq&THq6Pa3
z*c;<ReGRxiTG0gpA(d<A3Z2*`l*0to-Oiw?`G1p)-bEQZW)ySpKa$a&HV0-oDI5fs
zZD>h*)7>%(53n!HhfTjzX9MO0wcxTqw(9FRnfua-ooP55a+3whx1F|H$@?f`OQK}_
z3o}Icjo$N1T5?|kriQrof`AZ)Q?7A+eX+>21y7>x!a6I}jdUs)h06y3A<lovhTyN?
zWFsWDf%M|@)Pk$O$wpu<#2IpQHvH-)NH(amRQ&G5i=Mc)$dpm+)1a7d`ARc}^3|s^
zc||jcG@Sbr{z6BsD}n*CSDEXFY1!B9Uyi~-asNMr<3aP)XSC?eKZN7f)^Eb$VI)&h
zaCp}`DiQ3A0zCf1IIJ*zY^X)EO5s5Vw&%Mcqwu$_c<*l3rfSAJ@^KktENQTR{31dj
zR0e_(|1RvYVzOhcDhDgN#*$|Kn?6K60fP}Zs=+_P0FKIpS{!P-;+4&dJcB#$6<sAF
zU-E|++ym-e7DVLJ!=WY##53kXI~;2*c=@`*jY)<kS&SN1W-JKs`0y{LfaX%5H@_4y
zzIRAl`t8NVC`$(^KLfX{?E6?1kUDyN`o|5*7N2sX7Nd+5CVu2kUfE#*^V23q*)uPE
zDRIyXUbNIA@;b!@)@^i4n-berbXLTuMfE$hF9x=@_?WzCHNe)tQDW5EI2iEglOEyk
zh0AGXdlV@WTOy$xre;8s%ZHddJ?;Zs5DfJI+72WMwP5D}VBhHvN%-I{3bnLEa=ppb
zL%{T52A83-RkvBvDE<~t?qOv>Mj=o67(ss#UI}W30JYUE1759+a}A%YFqRiki;os=
z8B2UM+h`hv8;1fif9oWSUXipG$l`gy1Fw`|ef3ADWC2q~k?)NNH&6K$JsRXngoa~(
z8t#dju3r91XIvEw$S{d7D_?l??Hs6TiF8m6;hTM9BT5B&<`M*xmmKh_Fr`<}6e}Z(
z0%ZQ;cRN=Q32mFw(}Pdn*%_{)gmrgxRaxPSVG0yvh3G|^qtv-+BD-;{x&<<*4%v~k
zq-(SqlP6J1TM~J&kJT`_fi@qX{0t{q1%C1fGq5o%%H2N()4X9o&M3+ElgyAcGOnhz
zZ=@ynycguYaVYoJ$K{y3%?&k$mX)xKuXW!>wX85IRIB)Of+bc5Z}C$ELeM28Q%n=S
zBNe7Vc@1Qlgu|y7l#g@lp>+(*O-^I>196YOrhe@u@#e(Do0-;M#Fam69EIx!1GTe{
zYy4&x9tb>3BhR=2!NhiLWRHlOk>6VfVcG&2ef0+ghr^qx-&AB`^dBI%rv-^YFf{+T
z$ayc?E)jX&6;yiO8GPI?BiJ_!q|iU;{hSs=4#9-HpWv8hqd+8#!h8LJS_@AxbLTE`
zHLU1T3!X!IOtpTGc<CBKr8U|Dx%0O!qp`#WfQYv9uP_6CGc^iU^tUntN^~gL>R%ZB
zxK%9m<N^xzT-`Io(RiAE8Eig}f@%CI&VcR`f?yv00nYAu`#oEhwA6pV`D(Zw0C^Ag
z+aKWk<zci=1;O0^f53SlBF}CxRysp>oG*XBohw)og6XKbUD{^fOT-UVFTcW)78OQK
zgrIt;gn3XeB2Ue|%L;g&8%_Z+mq$b%E^{)!(g=<4iL#<=JA;Bb{DDr;Q52wN*$UNq
zH%<i=Tyo2j1_jIi$3b+kDj>XQrkYYe8&jdpv%<83iLQWT7394!G`-r67%l$hUzMh8
z(c&yx9R3eEKY?6=94#LA2RX0*rmT_rC~0zEGS%{9ni$~SvtI58Hjzr4PTJrkXuRgn
zzw7}V2prd0{$~m?oBU_4AQ~`54zI&`C6!lP0}UWY<&-#Vq0x1!q=@o^*B@jcI_A);
zEtC2oSViOUv#6Y+f0Y_=APV#V)4x&;s7s5=DfnBfxrLQ6Zo|9IL#EBnipl;4&TNQ#
z4(Hgn?g4OH-GJ!CEhNfsUq37>m-!8xX{L@uw*&50;W{MI_=>@AIC<MKI`a5Ycxfo`
z%mn!avvnqB#s{2)3zgHp_2lG5a1ci`n4O~s9&B}$Lx>12Ws5UJV3u0m;jXc_BF_zq
zB&Zw}zmm(N1*<g-f5g_`xQXpKEve|d`~PrLIso7|V@@wP`Xci*VifK~iq>2BQ&b64
zAXgKSHwd96PUdc_Hv)|Z9h$ki@)l2eAg|Rp8su~at;ZNkIxeB05DvBqrUaKf#lb^u
z0RrMAdDE!8si(<xWA<Z2qR^wBI;%QhgHUYyxCxKgX@4<*D$!yumUMj8f8t<1MG-~_
zUYQf{C`6AGt`S^wat?~@;14l|0Q(}j^!PZU`g#BNEyB9nj@!<px!5tL=`f3?*`a*Z
zcC2@Ds_9wKhaJCl1ZGX7_SFvQb%R25sP;DjsW&j4RGX#@gT33&rEn*4&s!+D+gaMb
z^xalqY`Vfzi>hh-NHf%smn6FqWeL4KgbuBx*=Cb;)_=94rIovLqk8%Em=ZAG8>=X$
z0H*z5&CGCMi`g3YC+}e$SrIw!x1)(|p4)DYq~w$maz%byl{@4{XgbSJN)2u76&$7x
z)A@GV|9VpGPVUH3N*J1p&hydP)pC9_ctYLmy}33I;j+lH;`Cjb8#qjTATj;WxAH3m
zyBYQZf`4n6Jvqgi9K)^n)kkv%PZngIB1JPM>6&BJIn!KM$XemLlw3ULtrklAlC5HR
zz9h(<&y99c^4`P4lA|XhJy-A2l}VJ*`sF2d5B^ZO?dfaXbm^H<MDxDkZuQ10mg!{U
zyHQ7d?qEqXfFI^&MjUoW6x4g;XMYtx*Z~?z%m--L7ls_~W5U+E>x!4kv%9;P3D@(E
zl^($IpSUiM(w6xdI%-Z=Tk6=3>)oz6OZUS!$a4p`ZHH^};o#)pINVwrDlsCYm25k$
zh`tBQg2$fQH|>%0@rpiP&13cEn3HUFiY=s;)nUwAlRLPKc*zrJXEkliRq}s7OaAlM
z<UbGL&3JRjaOJO1=KXC*vEvYR5xl|n`0~r^ZOL<oKQGlNCazi|Z;f=~gNtJTs>j^u
z^L2_SURNJl^lFXS*^xyb@X5%jnz0yjc))>Da|r~fLjK#;Zn9?|Gqp=F?YmKkD!B0d
znJi*OBCVu-pObShE$t)=^8k6f11?1r*{D6ZS2S<iUgdf6DL=e=l>#p(i<eV~Q%ldn
zi>IWEamr4TT`W+N60F+5%cL<mw5=|+Y;I|C`Skb>Hb0TwYRV&;*MV1Nz6)eVHNHRO
z)rr^y2T)K5QcEAev)d;7VqIsKzlW9hI1UMOv|`5OV`#l~Ff)BHZr|yG60XDTUavq+
z!Z(F>^fePd1@fbU0i6lqso?59vl&NE@)*S&LI(^&_GP_uDBM;kjiP3g+dI`HX{0Xz
zx6CSkkb;uhk+x2$3kN?j1q<yheetOCP!tW?xVr`6^3g`4hcX(`V=0ueg5%IN#bqJ$
zzvdmQaTB^Z|JQ%!j;tv<l!Im~M)fEu(MrR)+7ma`v6q5_0l#f~oMLr*n#w(v6Dmxe
zmCi*z2%(_lk>}BEEGyW~UBHe!d2jl=FZZDZk3_Wm)g4BXHhl2^>^kskJm<z^3h8tM
Q|A7*zt7_jVzGV^kUyRXSLI3~&

diff --git a/man/figures/op-relu6-plot-1.png b/man/figures/op-relu6-plot-1.png
index cec4334e6edfcf5d477133aec34f7c2d68436fa1..0672bfbc98ae08a8e2b343f89151b102157e72bf 100644
GIT binary patch
literal 17089
zcmeIabySsI^e(yq1q2C6MH&fdL<DI`1u2n~5TvBLOGOZnR=OJ$kPaz9q`MoWySwh(
zp7XmW?zm^1GsZdhue-;m-}kZie&6-3HP@Wa^UP=AuP85#gGGjgKp=3QJ(YNYK%g8W
z|6roRPwvZ@Fu}iWSU=UULm;pTkbhA6COyv)2x`PLi6<|e<2NRawIYU5(YGT)ec13H
z83v#+zI-WRolK2Igz-sT(#DD<pcpqkFCfiyGEb8<zlxZ&Me#eI;sh!xZA)m4c=#Dj
zO-=ye4aRE0E4*`C+t#?O%AFnOsV0KJrfUd9bfBC744{Q1fD++7tGJIq-1JMuK)mK7
z5<(yvQ<hN>w{a!$5cSqft_Xy$crO})ll=exsGp;EK}19ZAtj}}z;%Qu+neIi#Kc4`
zb~Tc_1;&z8BqWZz^R1SXRlT^V2*tR|(>0E}ckjx|%4(QbClU+M{QmVzZ+)b&L>Yk?
z$c*E)YO|bEPy3&qbhRt_XjE@kHb%y!)(A#sX6B5H46Cb`s;cZ997{XSsPDWRXE8sD
zkp$9{1ZuEG&ZES2=I7@AwXOX6^{b4GjG>|7ATRAnPI~%FC8b{rwjaf4;yOtJn|$zB
zhJ3_Q4Cn#^0u~k)sK%e>&A;tU{a9A^Xd|RrKA`pi{r-`MB&8jCY)#b<BfahE!U73b
z3+-mMXo2jn{NHe=s1Wl_R&^JSiybj-ZEeQgiT)C_kK)(}{L`<W9=>@dP$&JOpf8D)
zBoLq0CB<I3FX*LL-uzMPm8|YA!3w&Wkiy<#=VNmD=~|Dh=P@tZljhXIJ$gxX6Gq+s
za@T2#Un>k0l9Jfhsxo+mo$)!F4cqNZ?ZVC=`xZ$Eg6z}r@v+%(erJ-%)y%IC%9+nc
zOWrEa{Rtu@B<$?$gujD=f?&lC4i0o{T$`5m>}+jcs;IQLw>Rk*1_a31*c1i_%UfFJ
z`};qWlJfodv8g?ZgN=<1zJmNaJe-x4m4J{?PEHQ~`z8*KZmoMxT3SR*%xJb!T0?^v
zu3w{3F$*(uNlD4>-X6})n?=RN@ZtXce)zibix;eHY`WFXN`Wq!@(E;QWUEU{<$8_i
zjEq@Bd9Reu=M8m=RYJrX<%}-5%Hte&zgjo#ym~KvV6(pd(E2d(z~gVQ_Wk2yY!2O8
z!Ti{Rwc+V12ic(*6j74Yw6vO<8ag^U4|jK3Iy&R7`1>b!C@6$HPWQ*ltwJaS4o~)$
z&@ga2+S*pfN?8J%9`N#(|M)@eW4|%R!@|<u+>DH&H_gE{4r@dEEB$F{X&HC@0|Kya
z-u&$EAM?Pl?E?m$FV5Y_*x1#@Mb#HC^erqpm-cvgcr47#v9PfxCnw?QeDSC#si;I8
zRtMIG3oy{phg=gsefsp|1G=PygtMEQw6t{p#k`?#n(^A56(fdwUkSt$wr;lN*Rj)C
z3}DjyeLoz7O243>71lpAq$(@BR>L^_T|@ch%X{tB2x^vam%@ExTiYUEUn%&zs>;>W
zG`Fb8R!Ryzs=c$dRj+u|#l=NMMWv~!$^6kPYJ4IRk|Oo*@$ux8AxEbNW`>3f0|PG|
z92{1p$aqY@#K(Wn%sgBjWT6a+`ueqL)_Qk$_qV={o}Qkh<VST@c6N3KpZSFa8@~9C
zj_2^1$jHphOoqv)Y>O^uBC2X?QM=L{kvvLI2F~Rdhjh3-wRyiq?KlO$At^YsSonO&
z7G%C8N#eQjCsQu&kx37=#_wwg4Tv&3YwHTzg@ClQH&35_ii{lU%aA>Bzijx3U0PZS
zi<VPRAYwVeM@Sg)4h`dx)zrl3=;+YDS5s5Q+x|>pVc|c2{tz-e@9FQ4<1vdYD?3b;
z41{2dWP6pFkwHvE^m}P<sWUF?+cyD+)tBPpje3o~8tUrK&dw!o`~Up<^{cnHQ01E{
zB&o&4#mmb}Utiz;Ou7Ev-WU-t;iPWeB7<iei#KlEu(7wVhH-xQAnxgTX)F@2lqN-g
z?_SGt#ML&X@0e;LL9NqpVaxG+ygrKTGx6zEi(yZ3kDqZYUK_rssHhJfJP;ETGYBI<
z7%)G4xPN}KS6Em`OibK2Xy@Q?keW)XqB3}dYd>pS3F~G4C=Q>Nz@LviSd@BZ$2mOB
z@Sb1AmoHzQ>ad^WWM)FFrKYC7P*u&$&Q{mZxIg;o{rmS%@6r1F7&T8(Y?-yb*WCIg
zDvE`L1=c7xH<yHzbl8=WjxG(Rgk7`bY0=o;+cfErL;)wunp(=5@R<H+>MSmYT^$5<
z$e*CQSvfhQbzWX(W@`xoPE$2*QJ3$fKD?V-MiE`yocyV#qEc$RAS3qlXmhf&qvQRD
z5A@>MXnYm75C)q&bIlY2j$d#$*4HV8TqU>^#6Lj6j4>N4d8?%rZ|@9WfyIB((chDd
zyqdX%1vWNz6o;-Sya5C1isfP9<P3GpM!#mZkDz`fDJjXKmfz9R@)RN?FmP=kr+>8A
z_{3d!LkkDf2tFZB8`d&BT(14oF^XNA-(}ap&@dw?2uHIv7^<_VFQtg5xv44t>w3iX
z>(|lHK8Rc%&mB#BWl`k*deV#Lebdm$i0satXMXO&b;&b?!xgA8qR1bQj*R$x{0PhN
z%r_+^Wn-)~?QRZALzg-tltwsd3tp?pS8QV9<?+rO47(3^a0dB2ssscCki(cs0->B0
z=++4vcaaMS2&_zsBzGa-XSk<kFi~ljH9wJ^_o^Ji%yD(V-!b186Zx=&&z?PleepFB
zMZj{R0v@?4jEwa^fQBdo8Lwr0Xz2F-igK&<+ECu)Pp7Qd?AZn~Aw-LWp5E-ITXg3a
z7lVU?u%95g)tL@5kEM_?bu6PKA4*C|RsH;#m6i3gqJoH&bZe~C44#P`*C^6{#ty@K
zjMr&%!fvVSt%(V~m@h5`+3nlT=f^hib#~;}BMl7<qGMv7X(r<Leb*>HKX-?LAra8P
z1L35kgakGUtdC-a>d(3G;1QD}?=v%NYG~wX6lX$&CMUmc`*QySI!-D@?vq9s>p;<~
za?70TY^@6G*{!;(#k~_xPfysuCE`3L(Y<J*7W~#T^#Q~zrw41KL`3OXS-b`<_$7i4
z|K1y`cCk}eza7#{=R?kKpA9v%wzd|&X^(t253Z`HtQ=7yNaNEymy?^T`toIu+;u=^
z**s9R>FMbK352K27s5`PeCeHW7m+PjY}C0rU3c~6zVdJV9f*6lqMX33e~<A=B9-k`
z8M%Pt_wV1$8+~!)<>g<MnHQFnq$gkl5LA>^&C^uT(sJMVD~an@48?VM*`%a8z8?9E
z!96`a8M09_`C5<ke}7Ud&;e}Gi+kM%dEv4j>fg(^FutXvC;?FA!g~sFn(Vl|vNBoc
zrMt_=&AkT|%+=M^p6=bXA>>=E+Ew=REn$%F05<SiPK;F8&}~R;bHLKfyn+kk+<jyY
zcuGo23epMey_?wBXP0MNZ<pSC=}F-srcGxXK2lOrhMSG0XJp_|3XY7A$2l?`<H7FP
z&I1%iUt=3nSZLE8$wo*(uq^ey90oM=st`tL#r8r;>0LWz3o$IV`!~n~1zMFa`VPu1
zCuI~Ac)7XLD6lR+@VFw58BpIPLjv_W-|<UGNQjM%efRDcB#O6w@10)57<i&0BKl#Q
z2smw4R93F6t(_eolk-{Ex3qk6WOv4Zkto_ddGZ861%xIGjR8DvjqAbg;o(qkZ%P98
z)Vn({sS|L4D*F{$mky)ipH5qzTT?<+Vafc+I3!Q%_PRJNwU3UDHqqA?<HB>t2NZ5<
zdAh#>E9bVqtfZqeb+|DuD<c!jV<sg)RE-B4PEkB3Cnrlj;l+y=u(#kETcgEGwH{|C
ze(&+>Kg0KFX=wDeXMWYV9Zf-<2obq#F)9Y^MO(wh_isuDenXV<nve9<xH)li7vCUc
zfNe5a?Ggdg%ZVJ91OQ4vZ5iS$EZ^qm4P0$zl#!uB2y1!k)~!B02??LlQl2Kk?HeO^
zun|05c8eWUp2xG0LrF+Tmbwy(fBg7SRFq1ACHy@&P$&9qf2`Ck#+=UVOuqB@_!WAx
zC%@O_Ic#PP)}b&u>G*^MM)`OXGc#EMBKI#HQI^-jxHSa|6!34+vVX!_ol5QhK|k(z
zO<}e7kKUikVOotk0mMKR0Bp;t>XYT(G#HW}E=9oS&%aBV7vk3$2O9$VUc6I(E^+@S
z>H2`C<5^quA3?DItBy5cTZhkkWPTwOL0`kdcD835j}o8D$$fRuK^&Ke`{A<bRI7`N
zf56wbwA@@;a$4!j`1Jv!zw0`ciqkj7hK>^QxY`TWdrkR9XCDa7Mk7!QKd`Sg)+I$p
zJD=<>{JT`sK{#TWnVnq~_A@Tzw7`G>rp0&Nxq+#qhSS1srvVAU9|ymE3)7&h-XHb(
zB3x}-(^svK76N<!cQ5j_Q^Sb%ui_fqB}E9gj*hwv=WExU?_$$PFcDGmS-<X#<tY*W
zmDN(cOw8&Pz3RYhsk=S>?3q7DU!qlFuaixC_R7%3Rhf=g#@)?F=!#nvRaH7w_Bqla
zl&Y$#M_bcWBA$ZzR&+}SI{W6z!z*=FmtOY>Ffsi?Nz%kNIV_BGPj^lP5iJ>CS=1oS
zm%%<ODte);yw~=Xr9$ZY`L{Vb@=+~z8lT%G4Pokt0|j+;b$xw(XJ<Yd309J`lM}#e
zfO2}ft|z_YeuFY2`|;z)le05f%3}5JvyFbbm9`&EX^&|kxgEbSH8r)OdabOko}HgB
z53x^!+^zRHI0|4#0-d=f*4N*U%p2uax??Fbd)CCL@FfU+i?I?LJ3D1HH8o{rY+T$&
zJUkf?X?VbAlNd)2OkcmAnV*N0anLPt1!?eR`JEh&ANFX7BvLXmQ+<5`a`K6Ct7*wV
zl8fy|JlK$#6j-%Hu<eAmE-%g|s~jS5e>$u=!Y&PPnfV5pZ)UHfvlFI`l|tBczont!
zy*}kJGY~uPZh*zQTT?X$`}?rJ2}nsHzf|>SDfHuB*Tq7P1`=h7<Ax?+)&HE4q1tCD
zBth*cvHKO((1(L?NAk8h&x^4#3%BK-lxfd1HUWY0;bBrX-8$s0mGgCKV4n=b$i991
zHa<26|L+P#6hdGaa6o-lwd=v^#zj7CySH!O677lN{QK5exs}&!!}S%;hYyi)jr{p*
zV^+9Z+#?I%EEsr{>+9?1Knx%(AoG5Peb0|vl5F_@&P1V~+1c3@717OeL4d6n+P<=q
zSl)Trjl6QBjI3;PFa^w-4#1xjMLt-I{eAlqcMl(AI{h&+qHb(#+?uF<{rX3G*+_C>
z_h?6^>>J31RflO9AI_S$tpqBoPnpg2?}u=)?j#XiMMdg6mSZKYot;6J_p<MNhez|f
z&%|`|&VzJhX^Va6a&hVao7@Ol=ZFTbA(v|ga5g0WwZ4qNprFn`QjTB>fy}%-<on1h
zI;EznT4vB14#^qNB%xSWl1QETXb})12u>d4aGj@t1(B1J-(yq2T$%jIYsvkh09ivk
z+=Z_?<9Pe@G&EwM=0chag6y|l3w0Kj>U4dy*e5hRJlvm<F_`>uI6U8Dc$Dfb$kSzI
zWdO<ndL9D**xtU)XHC7KsAvaM>)Z?g769i@#|<RdgXuABk92W!gIJlqCOwI$_hM#Y
zIocQ}B_?h%D!xGz92`t1^_kCUlYB#KTaFW=?J7Mzy~K1-rf=5=-^1ggsOUje*vvR`
zyduD56cPn*1vZ82LFO8d`uWZk8Glz0fBP^BAy<2I^Yw-HDBzT&tm>kWWm%9jT<0S;
z3q-ZCp#jM4hxhOK9oH*7&JG#mVtcx-Ph)Yzf(`@d!^XmrkLR<2I3Quu%+AeCU5G(}
ztNY;75@A_QRX?_x5(JC_9ANO9T9U9wEKCRmJTWzLk_d2ca8^Q_^lkXa!=6fT5e$&P
zRN;#~O_K`3!omVJP2ULh1@^#-O{8Lh9>nPR+OV$I25c<26if*W4v;V2n18b4Oo17K
z8qhG{E|IJA;_&c3c%<=CGbb3?VrLxW=~6TpGc^hd3ZUgT@O?}gNS*-Ov|-txIBI^l
zBP1%$ND>J2pB~oS#@d?c{(VR|S#emVw;*KxDP=emflWm}euPdK`QQ^!P~^enrl(8r
z5)aVAoISKheOEs|?gnY%n`++Q9cO-iekeqcLQDw(4d9GHv_aXvML{uHY^)IWln4;%
z_RJG)ZSAqj<VOH&dXlfBqg##^Wk4MIgg%R4sH{M~W(4p1{<=y);ApC<My4;^pn=>7
zIpev!d}b`BDOCVv@{>1(>ZtET`GHc@SK;izW+W>CHW+l5mt91p4wy5jBGVp62~}16
z5OVEIL^T!ED_s%hdXsfrh-wyCAz=5QOgL>#0bK-!P<M4H0G~W#BJvnaC>6F73tYVn
zLYZN+v9W;}c64+EU;~j1A&rZPSx=#V=h>n3B<5VyxBzPi>8FhDPr1w=1#W(A-mSVi
zJf2Xei|n97h&sqdvOcz+@h&XnqDhBrx4F3q)j&Pj_}(aKjaXd0_!*XTSJjyLVdjZt
zrXX#_qw@A(M^Z<;ftTj#dpnUL?#J7?D&MBu9<Z^=x%waw?t#_Sg3sDz)nC10ymv2o
z`A!adaGF}lww6nd_()sfoM$y*#f6#`8}W5X93c;nPKDaGX5?$>|1P7X#L2Tg!@~+s
zpN38mKY|yqG|PLTVO9Kk0dGd;mLj6GJ(K2#FMPh5#*E|#PeIkyZw;r1DFC&px4$1K
ziEekI5R$nmQWHA37_r7p+qAv2&hltiwLLK!nL`upjo~(aWHI*Js2B>s>dXwMUc)s3
zf#-F%zvXkhtAAm&QEpqHOHs#ys9u1S<#xFK*!|cVb^xe6r0iN12CW2RDWbmC4Md_J
z_Lh>ofI}4mLHEIzmTtzS;D;UP=g1Ta#ZmMFvQ}nf&{LS2nJq0YzUV+J@=e7su_mXW
z2qAw=>%xkUDJCv1V81NNOVC2Zu6(m+=6t_j0k{T`K?Vi}<gVg($PEt0g=Z0Ga`pB#
z-1*xCBo<1j2Y^jTPF;yY-!d}+X(FZLWE9X2?z6E41q9IgeEIs-#o77z^fYz(4ibgp
zw15W1C>MLr1&GG_+#Dg$QC@-p6c{I3Dv*06B^nBV9Di-U;-DZeb%)P-=5JFVy>?P!
zqUZ7UEWm6?eh>V3T)ojHG&D4Tn*)^tb_%TjF0bY90}C=#<TFzfVSx$;Fl}RF<2vfQ
zfPgmO;jo+e-)G$QM%RP2fm{hCaqsBpYOh-aM3aWbMgZ{q?_p~2EvIWYR#%xRs;a7{
zsvH8L?#RM(qyHKi;k$qT8UJ$`8P~JJ7xm^)&rR@wht^bAKjh&VFSEEyOzZ%o+*|5~
zQ94|c#H)7)C<BG>(SrxELT+WCTb&(lK-vYuCL=&PaEo;k!Knp6%+F66G??MMSERRZ
z130>U`}XmnVZgf`$l<Rm8vXF_DEJ3}$>0NOdZDFNZ8cpBg`}_Rx&sCzFJD=C`Ld#-
zaNNA_-@_v!R+><-p!D1{wy^LxSXEisULE`fA-w+=ab3L%<#h#ABMN?dlcDb#uv2SY
z_7)=;71GP2-lL<!wHSVSU0r&kpcNGrH8eD^>r^k#&)ct$a6t-%WF)VW)QerTrOoGg
zs&B9!@#Q<MBb|rxH{40Bv%{@H&AY{3-mdfl0&ahQ`=1?c0cT=iV!937H8K)J<<tc1
z)iHsE1gE3!k9OrvIVEd-DPQ+EkBk=i|E`n5lDN6rdQQ{7503D{A?wS$ReKL{@VhCH
z)DMr!(#S{(v$v-Q6AO#qdHXfYArQW*L-}4Rq0YOtM`Kcx<w+$D%W{}B5}_yEmlR&p
z7PHg1gl^uaDNq1mU-YGbSg53Qt4UC_UBGPr4GFGAfp%!vednzxnQ(#!MEe`(OKA%z
zt6>1fDEO>JnTTY3f!|3N=+rDrp)%)ge9k*HuKkd3mAR&&;)5?HlSrjIV>Ed&T#?Hk
zf{vK}GchqSH8llVeqZ{tuC6XE_Tl#SKu4zJ<guFxOJ$Gae+pYyUtD%Y_?qf!)G_wk
z4CQ+<*#GjNM_K(!$SD6E0NbBGQnIovyu3#vg>OI>E)hqepi&P|dT0|<rLVM$xRSK{
z&3uThP!KZ_7Lvw^vKsOE^JhrpzJ7kPn-?`kow0yRf~iCj9odu9G(QCDXlt&&ohvCa
z{ri=)fNFiy*24MBQu*-rLg3qUG;cD0fB%)?f+S?s0LV}QNObb4sL}z?sWdOQ>%|>6
zkLHSC{@g_mlK5(3vSd$lD+Br<dxRq07R7(T-aQBMy`)i}Zgej!<wa2JU%l#6vUv2m
zDTgO9z~_@pfGk%bMm7e&)2937`3YdeTVg;EVVny~OVR>FihEsT7gPSiN1D!Oco)CA
z)E74PQYD&NhvnBN?PxRhiY(~ztal%JSG!%FyM#~*{rP++Ejc+kJDbnu?{A2eKHTdA
z)yn&Xj$7`|?t5f)TQvl~(Szc6xRwHobp^dHY2)53JhC;N$v;tD<>Md`ow_((E7;lD
zK|{aUGdL&&x=V_<AE@_l{ZM14gjRhsdQIymL@+vt=Y{Lsj^3I&a>;6Uf0q<k`xC1r
zbAp)pW?*0-AtePGqWpvPo)mFK#hx;Y@#Rd^Dq;6bmZ<5HH{F>OYkb%U_hJZGem=gZ
zegKl=1)SKM1gY`h4w>^ig}OtDUTn1V3L#xb8<$WED71Y!L?T{F-KRjf)7#nYHHT2i
z`n9&Uo}Qi(?Rg72XDx>WIV5?hX(&`KH|z^@Tk=rUm@XYIG<K$D?Z=X6dN*1iRzCpt
zVq#+Yu35?qz=WUQ6*3TT>l6wsCO6}i#fgJp*}se<%sO$Dy$13l;VDwAV*E~)-L4~>
zEe=n%ge$O6z1_d%=8hG<kp^WgM>&g@mR7Uevc_zf9dy01O79{kqkSn&4#F4LC4SYb
zizv9_$j-(MYnU4vl3SutAN>iT($&$a1yM-I?a;b%0=7`f!n<nTw500vyg=z;>DsAB
z7S00Io-I^!QxfOr6G@ED5#6J6Kkgz<4grG!tMK+lh<(b;Tm(I$x~67wV!~tz^>iwx
zcg@Ybma?io_~87IMQ$zg`)@D(zkAPpx9VJklY?mUyb;@mKy3^yE!in@b8=Kb0t-a5
zov&$R(J<|tVBm0yd(L<hF`xyigcV3X_?Q(yLhlI+Pa&lkrcms`;Isl)XA{Ym4@a9^
zO%@vhR*EbW+E}8&pz6sd@DD;qqrH6%P(7#<*f(y#jLwUdep4&rlgTx`%IaX3tF3Zt
z{3ZQW1_TZ2W?Ng^^Ye2cznUPngBSynnRcaZ)5u7)BRky7Xa%eDtSfr*D36guer#sl
zdDGHA$uRB6ixe^c8w3ep@Kp|LQkbKacE<Mh_R#qfIb$}V7WP_Wy8eM7=<AJNJS-^l
zDtdY;>a3AK{SEe)djUG6-n}v3wluWdL>zNPF7%e`O|9+ZBC0oss)`EzC<^xWlfmI-
z;)(#7E2SaxKC8hdaY9z>VReKH^u2(LB?-FF-qESBPJzY(XmDCNwNh3*uMfK)h9g9A
zZd`kRUpezL`QyI&`sC_rEp2T8`Wq$U@5o)UmXA8HJ`uVvmsdQxWbu79yuU4Db9pi6
zl~56p)o*e9sf@!j?xa12lr^GDeDQL9xa=10BFgb2g(P7LU=~2i_#IY>2nmY~+ex4u
z3Q(rDT6wC<XmL@6>YPn`kc`)Mp@ulW=5XzB%Yx+YOrf+!_B@}5Z`A5uBa7g<#ufPi
z_wQKz`toSNP$1ROFa54{-qDBT+xIgV@e~q+WrSz$0dKae`+0>}{Z52kEXzxsxL%`s
zX2ZE$36(zke?_cn=&D<`q=Q1Lw|5uXrGv>`wpyd3Z^y4=P^&<b0a}%6!XQcEV|@hO
zQIL}T5%$$dkL~l(J;HC1m!1rdRGiKWbsyrz50XEwu_DZftreHlxUTS_nbyZT?toJ0
zg~@#v*+(JGxdfkn`?$|fV*x1E)YRM((tvIjzPq)<jbs{V#BOQ#fOL#1JyY+NOX@M(
zv|`yGj?ShrFkjy%7WOXTQrqh&iO9-wBTp&qJQwf#e|P}~@5<lpo*Mmd&Y&Vk&NVS?
zxM(i+iSz7N##|g};={X-E(pg5`ud7WOFg!p0r7)c5j&4~O-m9uSoo%SW@cukKZ{Td
zs3o+Be(UQ^lUApVVYwJ~gzd4BQmONIWY8zcjqoRUp1x^4nq{;cawf}L|M9o)*R1PX
zn{L~y8d(*k{h@_s7L1Ehr)qkVWt3V;8JPfRn?jcfL`D$jDz!F?4BJioP%D)!q`Lp!
zXC_)C_4gmEQ}`6Phb^&g_0}L?DT*ub=Q;1tdfolpD_(Ov1SZYB2M=s}QzZcl6&Dtc
z78%L{34@NJJV47dQ_1cPLD$dqbAG=n=k|3MmvnJ7e(KGQbv%(4cr%Ra9j{Sj&^qmP
z#WwnN5Y*Fr?VqP7CxvxyRna?MUI?U^a0~4By%?X!Uc%yUWegU4814MGuTQ=L71g`?
z7S_i;G!i~?dEi5Z7}&zuW1njU9mn6FQK)r{jaZQc2_m*LM$8?Rl)`{2(Ma4Q0BUsS
z_U#_r>$~vT#BcN!bq-GUG(nt*ZDM4J+t>i;HRyGDVm$Ig55yD|#d<@dS+ra#9gUpf
zA?LlBt)Gr7Lo&Kw=^j51JiEwNp|h~|yN{@}Sslnhc9gz+VSQEB@#l{bMD*d|A?&&M
zT%9RFp1sYvMMtlr@|y$Jui3pPFmK#=4AcP1;lSV^^px(BkoZRWDP(&dv@uXs;)|)A
zZ`NdpOO4FN3n0dT5{GabcQG+D6NpJcv>&dIuCJT_W0|7z9w%Yyso&jf3M3wUMn&iX
zV+YNE0A%)<m>4KU@r^X)h=Y)r7!7mt@AEbq8lxNI<-lR(firNgzrLRB<#};13=PY%
z@$tQra{#~`>OT;q4?ZY>wwx?{%U=S?zN4a|Vq-Ny2ZpXEw3^TMSHuNKTX;a*UyXoX
z5P)KxpN>KJ;gOMxYiogP<MqZseif6Th+0DV4{wIPC$tADowg_^pHg4dx=R(j-u$ci
zu74Ae?YgD8r(YPYN{}rr%$Y{IG(E+Y;q;2mfl@lgYV428L}Z+cz8Def<b<Yv28K7Y
zlhY9h53-bX?kBq-l~oY)xvQ2glGF+FjlFScN}ujBw=?;!8c9&>)jLp|Hn2r`*VH@W
z4T=f0<dm(gtwDeSZbxGP48n>^5f74VK<#eQ9v5z%VEnXN1U?kA=J>j{&9cXYuO9DX
zxF;Z%K`}Vm+2P{h*#PhWEEyy~uhW$ejD={`1*d&M;RK9y^ffQ|^Pekp4RCY+SrtU6
zAo<D84wx7EKi}c*0<aF+G6)}IDaoGMo_;xz9<Ic}tD_6QexNuSF=yoPhFm#hRv&ej
zAD!FyF-0G3uf2WOhLJo23RDDxT)3c7p}M7|CA3YT?fX*v$%SIxvZTi7g!~kx&pg9K
zH@y#2sm&3?{E7%KCXLp+gAkBonD-A3z_)Oh$MoKvTGxYI&>!sSo_HwbEk7=sd$e_1
znhp!$-cVsPm-#%V=uI;&G_65Wn}VJV6BAU}Y${><Y6C;fks+115M8fujr{EKf!eG2
z$AoTVbBi;GeIg<v2?+@h@y||AbF;HYMn|DK5Dk~V<9O*|Y_6o9ycc?5n9zJxWHB79
zqFkj$v1^BWlO;722U5j6)Mc<XZ~?R8wVDE>$Kb#~B!@0L^egI(RYa<HPk;5xsht;l
z_)rGoi><G_+4rvx-=BB2##4K~iefSX*(v_9ouH7=4H}E_G7rEA(8Pio2h74X?eueE
z6Kw<Y>??&&60{8stuGKI_wL;TK|$6Js1$IOv}lag1XFHToh2GHRCJfmEjYJFdS7Pj
z8?N^{m=T^)u6|w-M9_m8Q*ATn3wy}mHX-4MHG)R9oLX+xyzw>`X(#ln%i-O|QR09q
zIe{R4Tf^t1JXoU$Z+tNZ`S|6{%|z%T!5Fu<x1pu>`0-<CsU5WX1y{}9<_pGiB-dTi
zKqSY43IvKNG}-DdPrpH)feF#5cILtk%-QC#7IH<Le2tGk?^h7X)2+J%8i;{QUg34+
z36#BOce@lw;_g~Y7#;KlAMo(d`>1Pb$|nk*K$Euw!UH<lp}JtF7#kbwODQXJ1d!Dv
zsHgzi;k+8)4ruPDH-?6WA{lug+5vF@rKl&!QZQYxx2|K8>~Btv1B2}C4TCuFvbGXM
zJa&Ng5{T7#AYg&M1I=!?qfKacfDXS(4>@4I7qT`krQmI1;!+^EK<9}-*KBSED*f;B
zl-jlKF26tB3SUz$8eIqe3s^iDE?$6qj;BgWg+;q{0<O>pone?!d;`(zG61?ctZj|U
zo-aQ1y^5fX&%wa~`==J!-FA8T@}+^56?C_Ap_PDn0~6C^IG+O4EnVa@cD=uj4SKEX
z;rhz@y8G4T1xR17fT_a`n~^tM0CgRZI?&vxn^!<BAb){K*C%)VfD9U$^X-rywY9X;
z#nse?K}=NYK)dM;6UqT8j6<h76m|;Ig>-)r1sz)OEm1Iw(o$03At*^so&ih@OA^be
zhYEwK(+i>$4F-P_^sr@hb-kd)2SS5cZ)y`{eURlLTu*U;xVzAh1cF4brK5A%&QZsz
zTU%XS?cU&X!^Gv-5jhvPsHnP^_MkSK8W{~iq{F;i$H0KOeU2O#DLe}3E`54%V`$vK
zxc(r=6$t1GNRNzPSXdZ{_9@XAx6uUNpv;4)@e$<nA9{_jf0UJ#L!i^~_xFvl6z?0W
znaeCX({&_gFA6i#5;FeOZ85b4{QQzUnpga|YaN~1$H2&_D~6l#@#D(u?5EWD&Mq#n
zu6NYF`)JafW!$wnJn@##)6V>^B|J3MVcL_Q6)sz#Iyw5#cJ@T8uw$A^^wM7<l}6n6
zGljq-K|w(V1}R<Ly6kLBHoQ{FMQ&H6w#%=KnwFL2PRS@gj}Hq(_9@DC&$6#Ss_w4k
zq@$aIcG8xRzlU=DQ|KU+gKMU|oY4nJ3v{la4IO-&>%AlUYs?tZ8qNztRqQOwQ-D>m
z8{b7pCG3+5n;7>Q1G{TDxwv)!F>VRP#Kz{PrP0NuBL=`|Aj>Yz`fG0I8Y)VYU_Ivp
z4p;ezGPx~_Y3o6cln9!TA8b+<rzh$CA;oc28;3C}cR{IJ<*mlU#B6D43Dr<X@B`_h
zfDD?HRFISN4D+$={3ioPJVh30y0O;$`t>c<Yv1BLZuES4P2`mwYx8+2huN~?8k>en
z!g%}XB+4pCG#`PzfuU*bY^$!jtqnpRY8&Z>Vm-No^#D%6N+?C4u$hjr%3B*u8lQ}Q
z2Tzl~$Bb7Ko^O5|w_aYL%)ipongDlzW|{dY<n*QPB-n;Euv<ZkNE;Zm*`JU?pRVB^
zDzvrFOFKdj)zvR_hF&}r!~Wr6MjszPzXHutQzS*OLrI(IP)Ec<D8)T|OZ(PE>=R@&
zKm{vOsEg2o>PZJvkCqlIML-i@={2NO?ag3eVG;J!9>~7fCq<^4SXj1BPFWOK(dd$R
zqRJo~fG7`naASFS1T=CvxlRDE$l!%6mAnYe6-XUJ!^1~MN5;jYEUMozF)?2k?_?2}
z-pSXwA=(KR3wG_Q^wd-_OdBUBs)r_u2gScoYAF_R5%%*-OKU4DNWVvFULG4spd{_B
z$(Qv{VbAbCeE1M_u=GwHpab}rKP=ITysd~46s#oRl`{s@CJ1@{At6)`3|b&(f{<W@
z^ruiTDkMVEmvVJIoo|hJRbp}rAHN4U9c+#mNE}4w(EVU%w}h4~Ij-?wt_t`EOk6@K
zKnYaDCm?`~PXZP`A)y)o=}=4m#;b=?fTF2Q=q~}DQUDI8H<P+KSXt9=--yPM#8Vs|
zF@NOO7&axwC?X=l%c};hXiMhToX~bq++HiF!zJhI0pYN57P@N4hI0uaN(7|0)i0vd
z>OWrpf-d2R`60C6{c!F&!VVTrQV<~JyaUp!Jezja00^GU%mMh|b}WW280ctFgpi)#
zY-J@SrP<lp@MZv~&~<=x4qB<bGD<PB8zZO1UJi1kBNoy*ESg|CS#<ztk3Ix)9)O9u
zv(29hb=PL@>aCzB%Lm=&(%+n@gcf68PY<XVBOsz~NYuN7-pc<%U440N%?yMWGqbz5
zZgF{@JO8Kj{RDE;Ko}Gmq)?ropYIde*Y|>yl+?+|32-?m7AHRdH%kAVn?rVNV5rcv
zfsP8w<kKEe=z*+$N=?;=Hl2NRLPF49da~uepVie}{q`pW#p4DIxbPq!7XJ8=Oo2sq
z4W@G7BN)ygp+Z3bbpyJ|P@p`4p@V<bk=+CzRNBx=$R$8Tzh!4XkLFAPQ5u-Y1(aNy
zN||hKaO|{9KoAZU=vo;YOJmyD+U_14$US{((wBZ?L#sy-d?VC$04vO3!`Yh{8y_wA
z(gnCAr@&rm(E?I3H#e8jh}4nIhCb_4nwUUqn%WLTo;whGfMT9Me-7QYl<aI1Lqkac
zqKDKFy`xn~Wqj}(Xxkw5_UEdwKq&%&1%xvk>_|(50TURJI61AG0*GOTAg~~bLmA=?
z&h_Dk20>^F0M2}k;y=*4Zqh#hfdaN4`~%gRjGRaa%w&d!cgV<O{QiR@h&KAbw`knl
z+*T~4FlnIlZVu!SHVO7{LQHtyl#Axn>q&kBO>|2bj<a(`V4y5uF3{Z3F(mP*S>WNO
zVC|E-LxO@XHh*q8u8-(};sy^R0S^=NDIlN<2)=>Ay{LA!`}gl~zODxxjr=wiv=tuM
zfZhTk5_^;^rjm*ZfXMxi*rcHMqeDxRnvhSOavlQ!3IkLt$RjNHn3yyWC}7Z8hX>wJ
z>=8v4Sq6It2XO0x90jNs%rj7ihbwH}09<${1|d)j83$hDqobuJy<(t+AUo=WPd>n0
zWgC)m=nw@q(a_L@&sMwafp)Y2DTSB#p#iiAvJJtH;QVt45~CY5qyt<KkOoSh>*|UK
z32E~`mzCw_<V=pnK)_~hhZqCPCpj3zF)LyDKoD~G@IXE#4m@Qk5TsS;zXR}1dSnp`
zA^G*|psv@&$2&qj@JBQcXe=l+Y^<#DP`4rZd=#4n2Zqf{dPOx1?$e_0=}|7IHV0Sl
zGE85;{hmU7b`9Y$2i+BbDpxC6Ntj3+tfHg@w7VT1!er^aul%7T$8&6Lev{KFx03+P
zsvu{pD-OYC-K)of6Ek#-0aTDwAca9C1V&foV0MT8`mq$Wf57t7mr`1K0Eq>lL};i-
zFUv!UvNb6)Lbt8zQUzYeCC{BT_kS;F+fh?fgICiL4Rs(A`rSI!&NlVlDB-gpZgX;=
zxG3@pydfTUy<8?5#3Xe=HCpRUgNh?S!mb6aiG8r+K<ZC?Z0GC9<Xy$T>f0~89dhWf
zr;4x#mjlub2F65r*dNcH**H4ZTF<_R=KO{hf^QW)`4M!dz#<0jza0qd8tYjx9UUFm
zY`weGjL0RTW?=X;I$8;^+6P}asm^g@3`i{4oxHsugPJ}8hd4+&_SxY5j*FA!!mA%g
zekoZwOCC@b0oKRxSug@YSj+R=s`s+Kam`!vHFVThR#xWc=a-kGzkK;Uw*&il7flim
ztZ5S!w{G7CKUaf(VWO}{6^vc$5HXJh8?_dklC8i1A*zrF`qL*&=*WUi8*=eBE|E|@
zuYX`5SlIvWorKRq?*NS9&v^;z-C!`&U=37JRh99B&Mi5vs>G98$aJ;-urM&Ua2FX&
zh!|{Ac5tVG<p8l&KY%t92l(u8YuXD5wEfSa69oM&C~g0IYdoX`1P_clZbMU8K|vu!
z5xB%92#3gl2m^pO&4#=R_!O*FTI^Bye`ja8K%4-_IXDCU(+8Ge4?*uK6y(cv;F1*@
zs4a|8B5*v8CY}DlVz53r3r;b5Qh7tqUlivpcyia)FfcHnaVhL|>E04bqn?a$0zM3H
zGH^uPqnU@&4_=SbA3vhGjpae6g`plJ;Hz?=pRjUs+kwadM#w-CHgHymwttv_o$Ou;
z%|du;0JcTytYR&&N}xp0)iCmTSN|dts#k08=y=G@jkH!u0N}Tz^j3hgF77-KqrZRu
z20wi3HkcNdT|<qbopx->N&+5U@M8B*PuG={*h78+=8{f<HIoJ_pqm9K9NMK^TwHLD
z0{9LEBrYh1)3ti6aJN33`r#O14_83lNRQ6k9~>|;RaEem2)2Mdq+Si*Uji`39WH$z
z13uFMq|wH|Nxv!wAq<s&n;0G=owPh`=ASu6mC~EzT`I;gmLR3Vvu7)SbHm{e9!Qrq
z{&z#|K%w4tfWsKvMjg@63MCMO+d=VXAM-~DBLgr4fDU;3Gt<(bQhWne17QTt^B9Qs
zZsgPFczb(8FAlQbEipiK7Z(>w`-&W|gD7WYGqSQ|0X^x17az7lM1(j*Ulg)JE5hu7
z<79&fkp!6G_yx4~wM$LE&S#%M^uC3U=jK8&+IE2)c}#n9ass*pSo-)}_d$I!1&#_*
zx{Lr(J?XE{lrww4cHn&md`!^Tp=|>Sa+Ogh8F=}}DiO6vb955Y*9-1N&{zq?PQkPW
zo?|Mn^IO>ain>r<w_PeLE5joheUKnyYz(3)-twL|2XbKGM9Bc3@5(kTjFFKM(H_Db
zJfQVUU}|!ksecEG33SUK8(Z{e2KoA;!%opf7M&y;r~q(7td9-IUOsrv;B?8quAg@S
zaH``k%gf6^&do=D{Oal12A&Ol`|ZO+uzk#|y<S0|Q3f<nXWUH*Mk5@YARxbB))H|$
z1YebR0l?bhFB215XlQ6FJq-=C&`#H<^d^H@b(aAT2^{aB50Ru*0`UaMAb{#)P2?ng
zfd(1S(xdg!ez+q@RbWN#15@e0JHiww2{^7V4;N4cHi4DW9bBT|Nfw6vJeGo}C4(uX
zLLOBb85!B$w#N0_1BwG%Lm`^;9gHXzbiA2Hur3k6`GQ$%Od4<oh@5XX!cKrKhcs)|
z`$4MNbrxI+{g-sWXMiN(-gtI)cH<Q`Y4bKpO1(hZp&tn8NwkyGB}f?Z2oH~nq+~`(
z$;)TYa#B;@w=aOZ8qNt-KEn@0o=RgSk!2#Xt(<6UlaZ8c+TFEnmCgNs9&cJ9xrYbp
ztcuE410+=dWI{DgwA<p92xSnQ?V|vIV`3l;0Hh^q9v2l-VU*MjBPsQ2zU=X$%rGKV
z`SIl;PF$HC8K>>bX;m`T(Q4?ifSKN1UcL*`3pkD>X~X9691nEg6kY~VUqo;a_}}ZK
z^<edQ9N6S~W^;feQ+D5@(*7)*REGDX-~m1I^b&NV;1~e7j9*-oA`U*T2kIQVo7KT;
zPe|jCZ5rigbrFAEFupfYTantm`OQn}yWLQ@<nY$y&*6@a@6eS8Tni}?&dz`z4^+ww
z5Z56%ksc^)?8oG|aF8oIyQi<Ot-G6-le49(3+xzj3JO4Jq11tK8)?;sTK0e9%u}fJ
z<=*oIex8R92fDkhAmQUt3Ytt-mENq)ML0ZJG}~7>x*CfV*_mj{+iS)-wBJ}6Xl$~m
z|B=&v^l@v|_U$W^*=AiN8L=7Es<h1j#iQ5?&LuoH>4ElcHXIdd`twIV=D`9)A6#5k
zPR_y3j^+OS!m=_jK>}R@1rGEFa0x-X6H;7ANC+5IRv{8$w*Ytg-+b`tNup~QBQ32y
zfLADN(BuchZRtJ+syAv(_p*i-WmZv1$!bkG0d^Y;rr(}_oiUv9C<QLy*7y6jgp$%u
zu@!SnE%+>-G1vge0QVQzR>*M~IXDzpNk~aZK(G4L^!M*yeX*xHI+T8mtsNcnJI?TV
zpjALZzzNWo1pXG_9AJpQbLWm0J1--nU#HR1!U9xvVp7uLQS*1Ait6eu`i0Pegz5<f
zQ^@+D${@A1yKe`wmEf#YY%JOCK%ii_RoHx1e*WgL=N`)ZC-otp91QfBwO2<;Ma~z8
zLT7lCBjsl8t4YoKria>DlEZ<uj&C5-o?aYHi=3H3_Z+MZkfK48JYI;>1-y*ZlT1wb
z%ttt(nTUrM0vnDGUi|!Jrh{!z6TucSJw4siV+~sYTyLNNEbX0u-K)2+PqWJYKAbj!
z-Xok~g2DvH@tWaGg1-I(=*<K6hLa><U4>UwMa4slu>mk>jf^;hq6In))XTo1A)OMF
zFu<JvSAly08a{Tv1l$^$r8EEEonG<`s=YwyLIG8RHy(@~`yD9`K2Xv*kGb&Um~fAe
z)n<Lo^7al->B^8*izvoicX#Ks9W>G7<KrP=c^>k<(brea*E#{H3@MqI_$w4O$VQNp
z*rQ&ls6YTy6Zm7&z=<=+Rea>FHl$GNVYMLRK#2oybNHa%e|6?*KGZy*(}++^g7%hQ
z<1Nf<;c0nrE{fjI`oB5w^d8bkNLZK>m{m#HH1BC_poAtsjp$AgP;)i=JEp=YJ^7<_
z%vs)YPJ`5n-q6a*3rsis9<m5Tr4L~y;(zI=ZhqpVvxK<#8=!*VK)7m{4)OOt2@g!>
zQ4fzZ?~$je+dgq818D#>2qaf%(Dj6T1c2dl_&{r^<ub5YEIhod1ToLqy_1LT!b!f}
zOa1dpdjN0#U(XFU1|~M<7&nIuasB7d!!2UZ-Gy(8RKKuQ;2XO?PYO3|T-tkk>@F<l
zDNN-{3Wtz6-B-{NNc!|YJeeF3MObc{^n+UH)vq4Wpo1$$86Am4+#r^Ljpaje#y1n6
z!NAk@7x%l+Nwv&_nTK0c%i|<e9d)@yTOSjHpD#WHIjB#JkTBJ0Hg3<xc2w~B1!^F6
zm)zQ;IsJvJW@4vEf$<QLFVP3qOZY!8#|eENHnw8x^~E@hH}=EHo3?9Uz|~D1mX@~P
zS9hVK+vw{SqA>O${u|+fQTv#RzMzxNC_Xoxn}y-Xs$EU-KuGm+%kH|@Rkqd7dCv?u
zb{KS=6SJTYM*NVj&40I;u6FGwZ<6E96XgSI4L7uDx=PwY_2{xrm8{`1HQKS5c`Udn
zZFgs5{T(6MrsSah5M_+AQ!9<!Rhtfqpv^8-EwN9w%{sUlyjJP2<vH49WR!-OZ%h|j
zP7XB=>#Fv2(eeJiUdWqRzP6CwO6y}W(w{L}v*yGsa#iGuHXV-_zO<<Ky0`eiLI~Zt
z(yQQV+AEGzfJN<<uJrCTX*Gp4Iis>qs=}2t&!4%5*}5+E{!XoS8~3xf);v|XzwvbU
z$vy*%w33=aQA^uUZ~y)FHugV%e|v3ro>043QcA6VCeD~XJ~P$-7@aZH&{UkjzJ^vu
z%iz-#E(}5(&R%z3#qmx?zf+N<#IJ1StIba7jauP6aW`c{G3K)sd+Qs*LI*9>IAeDf
zyw%UA%Rl1NGU(r0ylyo+6SvnipYC#7Y5#BEchVa-GW5$LElIg(1_O1T42ueIut-b9
zGU3y@M<jGxWj5A(wY#G{@zm6BkfoUojTDPC{FYyJCERT`n8GH&$|f!GkZciS?@?L)
zJ9(Fv@;T8=MW2`$1=OCYDO`y6`}qcB1X^`Pg~cuENdV8?iLdP0Bdn```_?M%%1k2s
zhqB$r0D)O9<^<nYOamH%j~n`ms2)%2?5aoq=<Pgfoo|_*B)j$E;QaE94!17axdy$@
z4>E`433DCwmpj%94mOSk$J}vL<}M}Tt+(_V{!qc>(b|qLHTWVBsIQQ}7ocD8Gt+6e
z@)oy88*_I`+BbK(%}wE6TfJ<w%Ngc}o4n~7Q8ttfT$Bt}w1){iTGp9S%o~ckPkx#H
zUZ=$E$LRDN$~)c<{$BXg1E=@R)i2(q_Ey<F&ci=+P31=x@|Gfv?W<Q_UIBpwbX_{t
zYT3iB7Z^XcdiG!!`B+hERpOsezsh;*q`+&IfPPXCAA)Z%CFJ#p`K#H&`ig$WyN_{4
zuo_&uqk^B=;P(cE*4I7#|MC|JB*%u|{pQ}hE{{(j#=L2<zc+pOR8)YVo)@t6%d{~)
z?>D=BF=x6r60PD23)}K^vDr7CBBs6Jy!=~ZM5xHuojN|83M0#%YUk~RIlAYZbnj1Z
z0$<Y0(2LNl(x$TEc2s(WyTOW>zD7vwjdzRm4|yGPU#qaoz5IAb5^pQe#E!lDq8HvN
zS)LVK>K9u5*;2?gX+}9_0Bmf^yortxZ(gUL_$GqN>%{-y)lA~cTgdw(fM0A8k}~`e
zzQw!MBk@7<so|fCI1PS^??FUwn^CjxgCic%2ny=|v0v11iSafDlVG*55q|vy0`W{z
LULs%Y_51$|wI&=p

literal 8073
zcmdUUcQoA3_csxuTb&?w5h7Yd&+5I45TdQ#!$w=HN3=zdC_(ftiIV7K@xh87(QAm+
z%VO2lpUvm@{PUdedCv2k^ZV;}&dfRY&g;JK%)K*r=FYrhpXzH+k>4dJARwU9(o}s$
zKtPE5TS<xVE$>fg^au!u2%hR1tKnHfLc&|OZV?d?-M)RBn3$M^gyhbhJEWwfWMpLI
z<m40-6qJ;dR8&;d)YLRIGynjAmX?-|j*gz5o`HdZk&%&!iHVt+nT3Ugm6i4G-MegT
zZ0zjp92^|??%m_$<OBkNAP|U)i;J6^n}>(z{{8#Byu5sTd=DNx;OFOm`0$~CfPkQ&
zppcM|u&}U*h{&Tyk3>a9A3uI9CMG5>E-oP<At@;-B_$;-EiEG>^W@1BFc>T=D=Q}_
zCoeCrprD|rsHmi*q^zv0qN1Xzs;Z`@rmn88p`oFvsi~!<rLC>4qobp%tE;D{r?0PX
zU|{g{=~F{P!)MQ)85tQF8yi1={@ldG#MIRE#fukaW@hH*<`xzfmX?-QR#p%Q#M;{0
z#>U3h*4EC>&febM!NI}N(b37t$=TW2#l^+d)z!_-&E4Jo<;#~I9v+^ao=_<C)vH%t
zUS8hb-abA)zP`SGets|*%-`Q14u=N>1Ox^K1_cEL2M33QgoK8MhJ}T_e*HQ;Jp9d@
zHxUsLk&%&6QBl#+(J?VG2m~TFHa0FUE<Qf~?c28r2?>dbiAhOG@7}$8|Neb)a&k&a
zN@{9qT3XtN4<FLg(=#$MGBYzje*Bn~m6e^Hos*N3o12@LmzSTPUr<m`SXfw8R8(AC
zTvAd}T3T9GR#skK{^`@F&!0bk`SPWrqN1|0vZ|`8y1M%7*RS8ceXFUdsjaQ8tE;Q8
zuWx8*Xl!h3YHDh3Zf<F5`TqSo5{YbWZEb67Yj1D==bwK%IyyQ#JG;8NP$*P)clVDU
zKYDt4dV72O`uh6&`v(RF1_uX+hK7Fr{5d>4{Oi}Rk&%(n(b2K7vGMWoiHV8H$;qjy
zsp;wInVFf{+1a_dx%v6|-@kt^EG#T8F8=xRXK85(jYcmoFR!eutgf!Et*x!EuWxK@
zY;JCDZEbCDZ}05vU@(~7-QB&tz5V@tEEaokaBz5dcyx4he0+Rza&mflio@Z~&d$!y
z&o3@6E-x>yuCA`HuWxQ{&|j@W@G(I4O4H1TfPjMTZzUZ49ePPXz%HVts$>k$#^jJ$
zGMMJ}WM=O0)~4hnsDsR~qz~5sv;g9ktP3VA0daSEd4}lSB*T4j^`akugjlTPJuhCY
z<kkiYA4C+GZ}-@(#tkVVZpgrxR=a2-K}z-8<@pr$b4a^L-sR!9&j$;`=h3@QAQL{o
z*IMi@cL;o6!_~w8CzSWV>mPynwvrzn-Ve3PyRZ0t<L3k3&~VI+(9>)sz%Vah_`!dc
z14O&uls79%XH|WTcFX`WSwM8B)~CkD+lTs|bZVMuJx+6NsUl*gS!Nz>>4{z1>OOuM
zR%^1H+-<Gd;@ffR-%itF=}(N(Ss8nJlg@7K?d$F9w$poMJlgwBwDF(5I`l}V`n<xt
zooDxQ0}bifAkwc79m1F1UAnej=2!Lo7P`c%V%;C=L8^JOSEjSUu+6wuu=mGtpY(8R
zlHYJ@SV6|-HK?yK#y{0L-&un*$!6GPPkON}n(!Ff&cAwrTXq`KVOq?OtZUNHR4dxL
z|2;ZmU$5b~ni9@b6%;lMp0n`unHhIGbLVVH5=Zs0skb;ITaXjH=^jFjjq_VIvwjO~
zdx3l!>b(A%VfJm?chf6{kD&^7r*-x5gCElc<=rGUYh&L*XH7+#)0_+ak(FVN|J1@d
zF8%g~PP5mqW9N=w_HyTE>gpFfjRqc4T^Rb)fZRj5wT7$-$=@L_Vw&>*$>oUtUe=g1
z;aR0|HD|CiC*}^R!3lUmayVq?z#J9d?Tp`Sv<p01ZtzSB^@wn#;%m>?Ix-o{^-yj_
zcAB4h-`mPKK?hNlN@a|`vRkGVISso1L$551yP^Aax`sqx<I)A*87A!-rY@7nqoTm&
zwi4EO{|{z@F>u#lvCgU5K6gEc2YPG1#dW|*D69QI`|!h%xr94rr(Y<zUW?6ns<g7H
zTUf}vNgO&;=VxTPE}1m9@cl;Sa4ij-6a&|y)S7FYKb2aYje&a{?1fv-<dj<80aWWv
z3a<6HvaCLN4fjYsWv&WVFmI*?LW+(uue(U{zVz4AU_Y+x7Uw5oYbh9Z2+{SLYDc99
z?%%$r?c74w_ZnPET_)^m!w9kF4#O9gpGZw=C=gRz#Nm#Smp&EAR*Fd8GcIX+ho-a<
zp#wtf-4$QsVuLC6qQVeG<jJG+8|B`@(^SlDbbU{zlBFmK{~Q_0od)5|VE>PnGL!!E
z271KQy(nv3B&)gwwa0r##1#95A7}A}JS3h{5xKT6z_Vg3RT2~quYA9lRUNMFPlPou
zl1>#^CHzZ%w8!nBI}0e%c8SsTKbGnqwP*MlEa-Cqszuwjazpax?yg>h!xIVwK*g%n
z!3uzCp*U{bUBe2v243Rml#KuJgGCyY``fTc=!M2K?60t%p{PPH-wsiC^%Naqibaes
z?L8yW#0D8!LXALS;UPMlTuupD`pBwq@A%8V<c44ByX|YYqubbz=+s?|Xbb1~Uujf?
zR99eNwSZxX2x1CgKSk?pW?i6d4X6g`#=uncKaDb@+!4TQhpHRm=jU(WRPRN{G^t}g
zjp3VPMlruzBLW3*=kz{6h}itE_M3pgLOv)T4WODe8pBu2wGyq}s)TfU#4)R9bkr^7
zq=<Av46OGP`01|!Au8;oS7(#Sxo*cKSo0-x6HCaBrC-%wYbeXEPOM%`>&;B%We{QS
z+P=7Fs=oGlk`v_~nGs}ZR%cRw4TNy(qBSOYbDvf!Af`y8qRrUu=s&Gi2tYA2m<dO~
zMR!P#_oPJ)x~cESCk1H!go7_ZTY<kB0U~cAM59)akjjC2+2n38!~1vdnGcQJoiYa&
z9Sb#R`j)1bsx_Kw9O~ob`a#!GkCpp5zN9-Oij9Dtc}v3{QGGXSRqei460e{mJ51+Y
z==YluSI*pVj-B7re``Lq8Nd^N>K}R?<&wOKGyAtYFu#(DUzlmF25aILXRfrSx1@z&
zG<+$TMcPtI>C72D)DeIrnNCqgh-$Us#{{S*<mLF&-za8`p+`&nGnivJ<KmTC`~cq`
zN_L%QN!3*DmWO9xNk`;wWhy-L00tcPqZvX*g~n`wM#Ly1=N@>Kpz+~n4jPovdKa|m
z;BfBYDLHoB3>+iC=JaV^0A*Ary+06dag~}aGkX_ckoDcD(D!%QCkzML)rju$__yqp
z>%>JAy#Laj+|>I=`GN?_$f^5bm8EHzwUVr3;_*^HbM5mzX0&TcZj)erZObV-0GKPR
z&~qDpJXgKI?K7b$`+Rr%=lg@Dhq_Akgearc@25@PU%*khpxlS0wo<2d4~{6Y#{A>V
zbdQa&GsKkySmWl#UICDeV3a)=QP!`26{*xA^~skC%MkGA^z}QAqVgj=EtA^MXb(th
z1Lbl#<IYNV#{!xd(EN697xd%KtII~EP(~}S0`s9Lk|!Z5vT{tkJG14c0n#B70D}a&
zc5>w^4Z|%C^e3zixiu00F*SDlfg}qVKWW)G7&XRi|EIu8DKqs+4BR|*IBA1HPVH)`
z{<)&8&@*RQ!MjB#hhh?nviJLmn(T}nuWi2I>0Q);#EYCWRs87dNO1}84s$;F(O~O(
zu~{Fgz*?0-h%yyhe2(*aXp5$-ngqT14jJ*TLOPBRm`k2kB*Mixs)INCHQa=?Dd*iF
zjt^SK0x*ag<*>ajh;292CBTX|73bj!Lir|00R|C}9<t_j70lmB9s5W8-0qLrM_OPm
ziBgGWY>STAE!3Ee;{hgln8YXnn1Xvbpa!rOVJ?MjGKgwmrkJMGd9lM5!f$ZrxRI}i
zy&K~FAf9;6Gxnq|pWz8BLhSL#)nn5_*IX1mD{7Im5DdFI53{PywDZkLCo2Q=E#*Gb
z?Kz^a2nK(SrQ|%enDgw>Xl0?p(S9iS+}_$C=^fl-9Un}AR(sXI{>AFb`5856pd-gC
z)OsKaU`;c}bxFBAk>l`_Tsp}#D+lYZS-MG1{TvAK2((((z8kun$(_vUlKw3MMu{-Z
z+6?0>7T+MjumGwhPoMB5oGxmDiBVsM7(zIclJ-XQ39)Xqyb15_beOvKyoT3r_84#=
zGkyg1Zxdj*k`E|M<>yqvBzXBot?rf2xd(oM_YkJ(gP>f(5rt8GB_PDAKj^J_rtDi7
zIX$47ci`+SLSv_dBjOHvVg-}ed@!8k((@-AE^%hVX1Z2c`Y*t<D0TP~rb3Hh!N(?-
zzdYhFUb|fl98HEWP2%c`*0J8+{gr4)hWgSIci-}C?$)g6!*KX<1~SihNTrAn2$7T2
zTkY4d9{E>BL#=vzh`uF#EYUF521|p|k0HhSNzY3Ba^GkacS@^^24e41MC9G<IsYE5
zigjg0yB}ul-r|2-AVHi!tc<EA&rS<TnO8V>T!GWBa{;WgMAE(QLn$#VNdT@;`{Sum
zTH3vd7~HJjQq1q{yUO4n+>s$YP_c>jSU0>bpBjqOixl-RxCdh=D(Ht5NaL78t)3EF
z&wrgiHU~;^DJxgcz{L>{(SNL9@K+?q)=%%8!W`^C7zPPa*&K`Cf+)R}U@ZVtKlBwD
zsFoRTQ^X@I;rX(|&Gyw_MQbn5-wF`Rx%8APW#AYu%YysF4TV&YGK)X|^q*bV*^dKN
zFLSpPa@&ANM}$<F=8YKtR&y5Y?cGSkjrhy%zGrqwBzNeB-qPQX6!cHTB184SQLJy|
zxAyd)H=IM)8`%dq5By|14er!><wE5#3~QvVT1MW_cj$8g)++J}X7DPfs#z>)8`xiG
zN5CJ{7Hu`wSG1{F_XdDvP8PP>zxurwAsqkL8`3$bG@sZy`wVVX<lain^!VkaxYrn$
zGLrS`@lU6_*Z7`glu=R-KtfGWT%3BLc{I+gp9wamRu=n2+=Ly3E;31x&mui5Z7u%*
zYWt|<<C(X4ZZOO5#;tnEC{_KFIv0~e9f$9U(<>!U`W$i_!%|P{YYd%l6t~CzeC#Ep
za><>T_{SZHZ|`_uvC4lUaDU|m=d7q3w=S}c4)3^N3HL6h;a{;}RugtxH2&J@V{$k8
zuf{J;BT`?D)`b|HDI4PuN#`+POcRe$rX$B-WqD2=u$q-BqJ2g1vm>H<g*r^Px?Tk=
zx`+)8V>oT%M43JV^q7Jf^SXU>Ee8PBYSU!Ou+m@phRm7z>lN9{D#7<lZuT+vCUjJm
z-34g5+OD`OJiJ1l{z>Mpys`Ppl1hx_)!))cIvMF?quR#LXFI2@NBmE@$vanjv+;YV
zar}nD%QZ~zEgT0U*NF@rU#x`A|DtH@liw}+^D}-z2Sz2jRQfG`%=ZD=y1Vr60EbGD
zZm?4NVxt_u!4KgvMqg}01L`q<{=XCCV15j<J%Ke*-GOm1RVV6{J*eJerudP&h2Sf6
z-lf%nx%%G|SnB<gFNIw8>==d*r1X2n`m@Ps5)-L2R{>Kei?(nMbbzpfYS5h_+5mea
zs>oPqdS9%<ehYrua_z(Zd<gTqcmpGMC#7%S3+rSnk@ld+JJTm}o~wAU@ePbpfK>K#
z`?ui!LLIbuH2j%giR30ZhQ$QH(>6U2Ey9mjCvDTwjH*^&cx6Ii+m5Aoj#c0chbQJN
zU3jO6AJC1>lG1bH%1EjkJ?2b|2LJ}WlPp%f882t~E4Xwo?S*G1DNrdAAr#u_d?Wg{
z9A5tGC}as0TI+$r!si}no&5{7Y+O6Sg%vH3v>C%$#?kc<Pz}7Ip*{GgF%?GnMG^VN
za;)s<(V`MK$`1$;eaL&R2D1#t2UXx}cy|Uk;@v<?3_kLRPy*_=!*mto)A0!4;kD%?
zD>sNwwQdlJ(}Y|~UoeIB*M<jipQaR=J3%gyu923b?wFBJ8l?2qLG5o-hxhSLIYiU&
z!8KGbRl#Z@r1Uq=*SnT7kSMM<F;>w-iI=IcxN}<Gg?LW0`jYta53I~L9gAe5FnI&8
zdMn-q>%YD^Ys4VhLVO|{2b*v&LeF!Fw_SV6nQhVS!M7oHf~gGGlrHGdS<x-Ejm>Oq
z>2$dL+xTG$IqNp=7g_ksDl+z_^0apN%2~wglzE298t0{4mXDpy5S@S6iSiHIHud%?
zZ#rYlU<w=a9~v?E?&9T(;f{<Admnc4rS4n#hfF4ggs^^|5I+2FiQk?kF9$Oen&R9t
zU9neZMK)uy;^t}@xMM03J&yTTIe*3TzI_m|Kb7?Y2jW+Nf?Uwfjy0-&unre=eRx`W
zI=LWLBk^bGgZ94t>FdO<2KN~-N}wa)CVu6ul9AD`7X(=H{SjDI#n=jw_=F<zmhK;@
z>kgBYLeVUq*<tTX>)={PKW9alzK-)RTujJN0(UQm!_Ulq^Y;Z6OLAs^CPX`~uM3wH
z+vR@{S=d)Z?hLM$y&IxG4Cbc-LSRgX>FzYT#T?6Ic*K&jMJvuO=~p85HXaK7xWUNM
zlm?6ABgA?idgiSz7uKDlEU5q@Vj36TjuV2jqPe7~fx)=vmd#aH?O;6FXw>AbxXtGS
z5CNBTu#wYj)k*l&NdO^`{trizk!?jGdjJuUAGl`bwy7IV49l(8a5I@Fkpc=d#T+UG
z01*L;^;ac=_~l5bh^)xo{*fkE<k|xuKn)C}N=XG*v}=K*2pj;`XS`DJpBT&-!a@}D
zP0=KkZ6~En$=#emSN_5lr2Gxj$x2v8&`h!K3i+()ZIBt#+HI*Vy=Jg`eedyz%6wP7
z0TJJs{W<#8zvyM&o>xyzb=t?+VRq+2mEx6GC(*9c38DxXrIIGo;;uy%4`r8zoW?g!
ze`KxxwmtOWd%@72yiST<Od<p@obylh@x4(Y!wY=&TFg0N4KY&H+$4!N6Q7~KDGJq8
z{^dH$fBVfLeCE+s6@86V6pS)l5&6<`V|YrjpFi>zw#G{=Q9vExtS`IS7LjT&dF}FD
z8wimU4yG%n$S7Jh&CVpkdViDiJ$~3lh}VQ?mi+TBKGGEAC_e&-&^Mleu{_fd`ArZ+
z_-$zRAsrEh<xM!e8Rz>e)*fp;q7Pz0q{j-yUW2+Y>(kU|M?0nd3F{u03pvyJaCpF@
z)I+tgy_yOt)PS<uBkS+xl(OT(01=v{vw7B>EA}xJq~&G)g+uOA&@8_y0oMEJ6Mg8u
zwTU7)ngs}<9$lwR=D;6s5Ad|E)RNlIQ8iXXI#CL)SfI`xL5N<p!7(ZSobG|2YbYYw
z*~w@UC>Uj&X&ix&C;o*bdi!&l;3!Qz!wsYJUL2OsEh;BLm)M_s8_4N`qowh{dzz-l
ze*`fs=-2Sg3XWHKFXO)zfqi5FBAK0Pa^TyE^+jwxyg<mq^bivahb>Jt_%jts$I(UD
zlx}AgG~<RxK$p{5G9U9Y^6^%srE)eOtdY%%qHbZI$!yoIYM$HR%_73m$t0JHT{_n0
z<P6Xv()+*V`}~k2gu$9W+(DO!qVH4<;-X=cSVg2bUA33vDl^`;zkt7}X<#;f-QtIn
zFuKcMTc}Ww(@6q^&<TVT&KWdogQJ8Lkqm8nR@!l$7@IAS4iG~ADW<6=ln}$xK!AOw
z83EUM0mw`14I7`Ena;w`I(eCTfa=|y{Vl3+xYk?!jInj0S<uVdXpeR4qNc|kySpgU
zTi6CMKM_-JA;>tCQxS;~yiOKi!(eMT!r}b$`(V|?jjEAETQ&ex%uyiuijp4|Aglpg
zfb|zC$hFMwG2wgWGP|5ec|LB1t&3k2%@nY3`ju!aklZ~Sa-Es3Yb;{6?%3HE3R&`?
z&+|b4n`}+T9VMMxc~pvo&4w-emw7oRv<|A;h5KS1q2|`tb@5WX#d|r-uaaPt79AZq
z36`uZzX}!?ckF(=!XGLgvX9ugh^=Z|dcJZX8WZG<n!N+@iAn<zw&9zte1JGjLI~a@
zFBxR^yQsX;L|ZTb%6&*hPq!dgl+y_VF7SH$#$L<q96xR)Lo`Nlwck=+3sD6}GXNKu
zYIwa$c(P%X{EA2+cJ0;vkc&X25Dq+`P06QqmXTIxuStmZSQ=%wj_A<uv3yR5mHf(U
zx2fB%Q)i?}gzh%lvC!_tF5t!RjF0baYa-X4cO<CzbtmSk`TFe?8AM}jM=g2iW=#Z)
z+*J_?Z=j^6)NLqY?W(+mEzFd>JR0%Law&U9j*2hQh4QQv<Z#X^A>ma#SVr{rsw*S(
zh@W@GoCm}5EF3=alG3_W_ml|3fM?#zvUC38-#W~~cTIv;d+bXL8i>Y6DhGmDBgz%l
z0lWxO2(bb}*weZ^J9S*v-%+CCc~1D0L$~m2$`=UXnQ6N^kD5Mfh2KLkcURMDMSyn;
zbCmIqTnr)JWX&w9;3#%Q<O5w(x1fc54i)^8WZ=b)sMXo`5}@MkyrbX0+)?>ggvGNO
zaZG)=Z8a;pLh$$A<=+=Q?O12jON5HgztdRT??>a(^Ntw79KQL>6J<Cn>OhEA0|aCV
zEJfj!-9p9NSZ@@FE%*G=2hjjTvb%^PZY4`i-*A}G14J?gztTtL^^EC*nDI<ga1~Ep
zA%_Y+SRdhU?t6O$|3!DTbK_rkT$_}-h_G*(T{CKm51-BQuiQdM)fn$BZgx@N2Sb9Q
zaM00aZ7%+<0*-#Eh*V6j@?KTB$kv#=3lIT}ea(`AJ;^I)1VVrUeDvN^H?A_J=~|ao
z?&+f2i1v&Cp2Js@9Xs>c+4I@+L@_ngJ`@kb4r-`5q%(3-hAO3dD|^2{3Q91C3rAYN
zFpMo0^+^s`8mA`t?S0+tPfWuf#_6E!qeYcQ_B}s%?|R)xOTyfCR&aEu%PYOr<>nJj
zw2hr3eSLMx7ISxQ8!)%}OT2Z!_WiE$O#9uh&tmMaCWQUs6pVC0LTUCM3@Z*N%HQkA
z9a->d3!x&qm3(!vN%y2XFe+dJ3}=<?M7H>8j~^czP_x)*j*g#XJ!?x+5Dq+bW9~z<
zJ*U>>hnloHOabUU*Y>Dcd`m5Cd)=l2pvdhUcGk>!&{e`&mAhp_%S&kUkfYI;)#{K9
z`B4BTlw~s)yLDiaW!X?azu^5PzO;e!ThQ#Rnk2@~gUbxr<22TO5@_-_O}6`Zwmom;
zlQN9@>6W!CU!h|8)Ml8v>-X5)PdBM4`?hu2Q-k@^Mz-ydVX4Qne9a+l-&sBLnuZUv
z_e2|WN*nTO{=Pu}m)>AcqsGXNSWHgRxYak$@EOj;7|+<u88~X-3KHuzwz-J#5dE`i
zn6Ti13;lmQE*F+ISbte83VQNm!I<l+ye)XD=50w(ztD#%I)j?;4L&fB+0SMJYyZbn
zu^z<zz>VH?eNxZDTL2PJ5-{LY`LP}bTV-mSEl&8+aJn0&KVLN=*`L!h#Y7`cf)XSD
zzmzy{w0}#?B-(%~ZB?f3<^`m@vS{de32I7h)AXV!L}o4quKpi2-7vU3Sg#<YdoJA&
zIl332dg0w5QUtd#P?RnD&(g~qDjc?WlkZ#<9@o0IxY*GFQvdWLEkQq3UnF#IfT!){
z=^-B(KlJ=KqDpMVb%bucR!xm2&CPOVLv`k;VJ9RL)jo5=TD5KE8R^j6KUPRglV@6P
zOCP7sM_$(GxHoWOctbaJTytC5Q%=A%-cK6)W{RS~nekr*n@5s6A!gmqlxy-Zc7_&K
zuSe&)o1{?3Srh3ehEH0lf~;F`j+N$>)mwqCT&FklV&sGSPbcm)Z#)HtE%P$Odq#V2
zdQ4x`T>scR0rGsWw-lNL@uj@GRCpk9R#rF?JNhpgCjzZ$l(qegM{R4DW8><2`Jm2h
z_yKB{HQMeCJj`7Q5NC72a;Y|F%=L@zwmeG^V9R-s1G!x2Ndy14|8(KBO>QTX4`b?#
z%1KeCcH`9$fCf17qhc^)PnJZxGC1O+;w85-oeSSnMntt;yG2S-ome>;{)&4LxTY<K
x0l3}5rs-}QyyY+*(J$co&;M2bpDi~uQj(gdCGxi=@&6MEwAA!ftCVft{4b{-xugI9

diff --git a/man/figures/op-softplus-plot-1.png b/man/figures/op-softplus-plot-1.png
index 2a0b841f4bd224a1ca15f6d94dcd08b6dcfb90dc..69d87ea65f24a01339363ef5d711f576278ae13c 100644
GIT binary patch
literal 21578
zcmeIaWmJ`27%jT}L@)>g0TED9K~X}bK~PXYL_j2@QBp!mIz`2x6hu%;8U#UUq(w@)
zJ5;)*n>)AXo_l}XbH=!1+`o5^&;2Rv{l;3)dY(DwGuQi4URIJ~Fa2Hufk1IX>Y4(9
zuw#YzXZL^j$ysTmGx*;g3n^7=0%6}l;-4K&y{;Ps!g0clYgZNRLnnJ3G-_UMlg-L4
zpEZ6fCHR)%^wXzj#$Hoj&&`w?eaI-ls}gxe^2|#qJ1OZ`(q|sLoZgoh^Qhmhc4N!P
z%}Kw=ODW?0*y2XC<nTQME%x;f*~=rBBJM>t8j%nPT%5{jI|+p1VzI`Z1TF#U%LGDz
z`2YXc|99h{%PLfyotT*Du(5KX^DKej%B8HV?Be1=N$LC8-h%Yp+S8{`U%YtnY#*&Z
zBMD(Za&CS;gx9QV<?kP|-z7(qNB4LH^O^T$nRQpn?jX#b++3ST-MKu_@xLGb+!DbF
z&n~s_h(K+%-kz%Om+xmJe^-1ExkEscnn2hx*#D|<M7O1+RyjyPNU~b$1^J(kS$vAr
zE=C8mEf>13&ySVfB^AMJzvdmdE|)}}_SNyivawbR^M#b{qBmU{_D<Pf6XzFHSOYI~
z>?DvU?C5V?&c9h7kzb;)>GzIxw}(v!iTFP8>7G!Dt}CLVh+65JHv|0rjZI9dD8=?a
zkzbe$7?!P{7e45(+Pt`Kma!yPU*Pv#KG49sc$kzByw6AC5+7f7cDA3756^`Qe`jZp
z3pEy}KD_%>IkZ4aPe1AG)}5dRqhI$JGJMN_{#?5KNj+@mlB{n2IW7WWLk?s0_Vw!=
z=T)nIPL<DUu_-AWoSbs^W3C#hDgWA)_e6WUM_E%tQLVSB_9r{PgqoVnK)>z0rKRP)
zd-vS(E}T1O^83fjoq^PZ0a2&5g|R}Hb=U2opMmFa2f>A->tE*f-mKnt(`}RS?c28`
zyLN5cMu&u`TU*alQIHa5O9zLBF1u~@badRkdiB>p$&-s0FY@s4eE<G^FfOXU_vOcr
zAN~EC`X4gz7%N`CzU!<5<|EthnCQdqtlHMr#hzSiOg+9~V`a5>@7`~2+akFoq=lo2
zhpVo>>15aDFYvr2WAnLBuWyGORwbR7xVYU+-{<&v$w+w(4UO^f@ujJ5S~@y@=M@X9
zClt<&>0i>bV%9gUPP_d0@ABz4q7g%-9Gqquk1le`=I<cHy$%Ry?&@M=W{!%AvRa<*
zMI;Um>YY7%HqDwsqH$kRu=eiQ!aLQyGc%E%*^)a5chb_**x1?GIXP2PQbeqWWps1~
zrn<8+%-3x>Jm`|WK794nx}Sbi{XZ4?orE~dL-Eg_^t7~m51L|Ly?Rw%Uf!8*xNq;?
zPl<_d7=4bDPU)Oa8dR{ZDsf;VxR)%Ax1M8UWb`;*7b<W^L1AvRF*+vZ6e}xhn)N2B
zI4_~t%|9gM=)r^cO-zy^BQ2MwHTUe<bNlw~Z;lmIo8ek}2v@f0>FKdhZ0Cn>SzB8l
zJsM)to}#3vxU#<9|INm0(#Wm&8Wmx7^WdRFxMUuOxxv}l*^&APK5lM%M@LdJGC!uv
zAO5dTX%)GKg@l|teq6`Y)Ol{O3e&{IZ*fIKBXm;3?eR+{!hq0JXZlF#bE;z^&ZDEF
z4NXmJ8>{ny=TtH>Gpp(r?anw7+-Lqy_vZcn{X2xue7e7=zrR01EV3s2a$ZuBN$)os
zJ;`+;B$49f%G%m69upcw{>6*ZGBS^BYy#iB5pbA0)tOsJ5W;cH-$g{s{Qj+=pm6ZO
zfeUuiJ%?CjAKBXetgL(;Hc>6VKp^zT#m86F)SN$lytuLPwybPQZf+<3DJr@>TpNND
zT}*yRAQaxx(&}3nYd(GYbgFg{2RnOjQxk42pgY?fw--KXr0p3+Mp)s!apQ(ugourW
zMP+rhdb&YvuJuSvOw8!`kJX}*aZ<v-Q3fH~<fteQA0M;fn)kN0w(Cn%H*emY$ZMeJ
z{Z1U?OhG}x-+%saYJQ`pqqCpt;$aIkHZr=au5My(E@zvuz3UKeyB)E_c-irhs_OgD
zP>n~A9vK?O`})?sJj7aAx5!DnDea-n_2up?v+CN~-gHAbGqb71iS`HF^rE7oow-|#
zuP_NockkX^Sy>TwS>wKVQ9)hZsqEF^$;nAtPOVcy?M1{YwUPJk9StQV<&l3cef=TK
z7%wj`3k!=cJ<qF2qisqu-5rE2%}0+Gs)IOvyuCLzH(BMvO$`nCj9bol8aez&yx5|-
z*;$On9Tk=KhK9<zIuWbE7hk^2Cab37Vl#;syUl$5{P_zPz7`bpHa7mng_`%}fBpK^
z($<zmsJ;C=aopU-#>b5@0x~i&jYbb2KE$9a-?@{Xojo!*cuJ_NomgKtqA~OjjE&n`
zTCPb-PK=J8Vqs}(C%zD9p_4}agruVccef10_WVfwA(F4wBeIN)VG*ut4<<XlE;ux<
zZtu8`Dex0<+e%JIklKGuMy4i!RZdS&Pc}@D$E=Gx&D!PYzi2pr=FFLsC&e{2d+X{_
zkVWL={Jgwws;PaL)aWNC-ZU*TTv<hhkdTm}o?gPIPbuoz4_D`gn3<XX`|m#ll>srv
zIdd%kNZ8x+Q&GuTSbXvGlar9}%+Jq{k7q^nk}2~M2yX4Itx8&2OmuX?0Ricbi{n@>
znI`S@M~_-tTE1bNa{4%m35ZHbdHDBt8IHVb*DgsvCL2r32WDosWM$t41+fZs+5H=w
zXX)wb(b0dNkWpD#Sz+E&G;*Wj;%pa2Z(}&qqFq=giHBmobg8+l>{(dY=GNxAd4J*I
z{rlyWlr;10FW{1nvEZUbkrWK|^=q*($C_ft#>SA^kbLm2H?k&2oX#Q#98^X|Mp2O(
z<>=Y7GUV3Yo}N!hNr$PaD}9$Dh!^d8h=#^?v?22SB|}Fir-~mx9@*KAjEs=&+V%Cz
zm)DF_q9wje1oz{+qoSivva(i`mg-LT<R&S97IfRn&(2={a<7zblr3<Z1cOaz?CSay
z9eoM8^4qs>O-)UQ&nahQWqCe*dR`;v^`wy~0paeRgVDo+c5!ybd@`Rrc~e^2pe{5Q
z!Eiyh9492Zf)Lujy1M$O!Y{tj2%#=3CDoB`=$Du{=70JoVpUg1l8PX%dF9F#tT;6_
zOhU-}_nmU0n~EA5H90d+rl@c+<PR_;+2;MO^TQOhwC(sU*0GVHVUWLn6avLI;}|Jo
zOkB;>bP@>{QKEbQeu5&~Gcgp8T&HD@qeqW+HoDRw%5+9i_;Ho5e0(sKSs5AJe0)J~
z-XO!=Gc^44@CUMmxFQx+ntnCS`MXzQ6$0PBeaET2wYBMFXJ@x_=gzd~O&yuVPrC@S
zi}!VPdAPZ4XZkMNjJF`9v$C@}wenZKKilW+TJmIzt8r)Xex<v2E6d81Zr_gl{CN`~
zD$i~P<%V7J+uKPEQ6@}1-77IMv5k$5j*bq0e}9y`{DOjGqOQXK76!#`<YRJy2oaaL
zzkhGsyxE;=%}PtFqOKlNSm<hQZhq|=NoOM|31-4G6e)&|j_yl(x|5TWsN2>CYMX<D
z18yuixhA((0vFTJ)YHSj#1xM61P2>Ec#s$u$HK;@CNE!{%r1xU(UFQ&NK8^mMX7PI
zwERm9u-@K~kPx6lGc&U)UtV_vBG*%ZikDtqsjBG)efbV#R7}aKseG3%<=IYkdF<L(
z`9D8JO--GVk%44U7s6LxU+*U!Yiw)`NZ=o4qH|$Nb}wPT9P1Xfb^dpmS56K;1%*$2
zglI*1xom_;*rdizO5(Kd(bUvL?nBt;+f2MmO`Y*Ra;c@YwXCeny!YF6+l(F0qsZJ(
zM1K6J)0V6X$T89Oxw)mqd2w8+f2f9T^GQ(=DHeS3#)ps)V(zD=j!8+G{{34lQog07
z<qY*t`Tvnk$#)G84@*c$eE$5|dblR>^JkQrGiT3se$^%G%q87}S<<n>?|giG@F5Wq
z5v<Irsw%#RJzSSAsTVp6U=H@|xh^-Akd!<VL$P&eXYp%A)B#mh2V|$8Kc$_V)`NqC
z&)$hMGB(Ej&ytHZHjisNs2@qF3(f4<aFn(A`E}R*=(6k2VuEtFyPjr|xj$Y6-q6gm
z&BU@?STOC$xxm4}!Na2@BjY`(@x-{;ZXu(bEKeokLFd>YcO*xXs=0AyPs++TYruGE
zHs@FVJpM9EHs6CFNdmT$9VlDtymLT*6YZ%Hyk;LFB97t4THV`C<!m+P>_bR|q8raY
zrC(HCu?aJ4Ch6F{V`utl-wN%v=D5AvrB`RZeBq*{4JzurtE|kz$tl|KeuujN>iyim
z*U#!|Ru&eT!^GG1Yu7A@In+#^!oD&6mG_a=UqeY<X4Mi|hT9){SFR3=^IZshqP$(0
zplP(DUo@gC!?^X6Qj+rKl!TB_9zfaSTjT_pNYt#pK9SR>^^t%91`X@O1O5HSr=}Q<
z+2t^J=~bqejbrqg&k2ZqiPjIzO>&(4aQgIzL*oah-<iE?7-k%eCv%T9p6M&tM{_<{
zz}i4e%mWiY-kO9wUVxRIX8lHIfN${p$rd?bPVV};^_sJnUFObYh22TjcDZ|Dvm@GP
z?M@EaM&}Nlgr7fuB0UOU-H+h|C<j(?TAprgY&;=EyN7f{{>=LBst0WTF8zMWy^iuW
z^5KUGp2^EI{j{{SGhLZLA$Xbm`u@z*(-TM18(}7^bJ#&J2>9^fo863-gTv}q-3p0o
z*MMivYZvCHr{9s6M|K?~=FN0W9m0t2U%tPN#VjQyWo-N@JDaaFSLz9nkwG)EMq}dz
zt3fGI(W2?;hX}OK>e;@GLhmLuVoEX}65Jh70n9toVu1}tBHY47-CTj$goK<>6qA$B
zcQ#f&#c_r$$6FEwtcQP;mAR}fJg9!h4g|(`;lhQB7k>{iHjP+CkM3}Hz*_$B;RD;5
zGvEdBpFiKacI}$FdUtub=cL9dBCrWS`95?0yhg57Y*Nw$YRP$a_IQ8?YLSCW+L1WI
zk-_2N5AWZbb*Aap1Yf|so>NJwsjQT`c@sB%T!?<vqK$&UwUdg<Zyzm(Oqk#zG6S;F
z=IpO01qGWvNBQ{q`73=TM357?o`A&6%~=6&@B8<Xlao7f;sg+QnswtY<azh6w-Z}1
z79h98bXMv~fgn1)o0)?{_Qs8ulST{wp<sB|*4N7^D=(FK(SjDG8`e{jlee_Bg$r26
zd3%$0HjeHhW)EUsMlP(Vuiu{iRf-J#?VI4ed%s-Pm)?YV`RZW&NZ*Bpg}r@CjAfgN
zwgQ)RTf{9Oap*<;y`8zE`;edMlpZ`7s|yoCj^jLYCdaHh3l}HixNwG>TM_9T7qkmF
zN=Gp=G7=@Ct*MFr!u=mr0jwq_Ccs$?)YPYrAHQPDz2<?_Nd$r@wX|^G|FK(1sa42s
z`V?NQtbRxNd?z&?kr9t0C1g99kdR=%I;U@MzXHC9q986VPRpSoWy|46ob&QSw6p*K
zhYlTz{`gT|QL&||>7s7AcVo09=iM*vxSrj_;|K$n1NoiW{k^p{69FS4vW1I!PRYnD
zL>oqg*5oXgE?wg1?<g;qxOI!2oBI(!pNEIXAr_hALbRL2Pk1BeJ{+kFD=8^a&ocd*
zm9@0I%*n|~ML|J9Ma3#a|35wi)Xxe+hVSKISs=~cxzkL<ti9iOfO^xS8#!eb2*pxb
zh0gE8!fMOPG7V}!oRsu^`|jP=Ts0>y9LrI&lz2g<<>hg}P5JquVPTq*lJWWZJ<ZK`
zBqg6uYAjF^Cq44ZmoJ|_jUyWrZOnQ3`T2ePH~`#?;6f1*x8-(3Q6ZGSj)+K2Na$#8
z*45E@9UAKV>-#PNf!t_GKmRbV9YoAReq2as{m08g`uh4<DM(#Fe2a^V{IPTAh{4`x
zgN4^$xY_nu-PhMwx59U7d>mAoosDfTIr%9edfj3m@KMc&52tJ1Un=|Y!`a!H`tV^P
zK|zH(cTO@hpE`ZI%D0>j1^3Df2;!gLzriA!nwv3da$$m}fosm6m$BvWeB4SwXvhTq
z1o)VroBQ<X6Ifw!SC=3=`z_$-Neyozm}^K~T3Q0N%*)LMnC;E6jQ;%Dto8#B>N^&Q
zo}})nU|g~*u#Pu_z+7*h9q`_P0|yk@l*GmVAn5f*?sEZKf4qZl8yFbizAn3N*jrkD
zMIrH#ILd2E*O^O4g)kp|8yg#2RJ4t?<niQ*jg1WjIeAWQZb3moRb81q@oxOo)YR~k
z)5MrP^)@gNch=S24bCbRDPQSZZg0UwLl~v>k;udP@}lJg_e|BuWoBnD(5;}vC1q?Z
zmtdA3Rn*pgj*1eKlQRc|7#bP^4vCRJGe2B=QiwL12+ez6dV2#mOpcCP*xJfUOM{yN
zch<i-DOp)pmhk1<4#HYfUtb>(5hEjt&Gg{l+u-0Fos$2MPHb(-<NEg?RvsZOUjrk`
z&0R%~IH#Qa!pG+`!r9pP;Qsx{SUXG4ogy^6L7mpjwR$EI{r>&?#H1tuSR^arykgu;
ziQ{$>=_P>W7BeybaG-*)vqN7|x_b37=7$37>?5s8ibhps<=q^MUjVQeA{utJYx|$R
zeM^mjCBnpy`w9ySRaI4|r>DWvw6aV)yE09J&Z{z?JzIq&X3OnKjM{Je85o>_M{$-D
zCn8bo0sia5gq9%$;8IBM_7VuIU0q!$L>vIB6RAbx6BGS~F5N$VT+`N$c=F`Qq)`n4
z_kAiUIT;9v;gW&Oty?}pK|#^cx_ALL=jRJ|HU<+Br_}J|BvFWfScqH<0OjsZKy;$+
z;p}PA7xsYqxa;NS=7O;Q9v&{VnNUWxTb&;X3=Bjl%Gq)Z6J?9aM<6`O$;sEfXdW0D
z3A=4Mqc*g*w*I3&_t`Ld$8RYrDmpqkiiu%vyzp%wpW0ug&#_d#gl&H%#?*(|Sqn0<
z=KxrM{soSUue`jHkY|WW!^nMJqBbdj%7r8PG6<mT%r(o23J3@|I65kvp;lHV|0qhg
z_2fUo9&isttdf#aUS8fFkGol>s<&?4l9H+fpvPK#3OrBPgJp_~MPSv1U!E8l0r~!7
z@e5&Ce>++NBGMk>(kFTL>=|Yo%sM-h_$qK(et7k22@xIFm4zbo+)He6(<>|XDk>`A
z$_UGYByZlm^Soq6N{NvrRRUB7pHx<+0VoO&*8=EEOniWFfqID1Zxc;^C}`3*MDtFO
zCw;FsJ>R`1`@MZH$O)nn&d&LXiD#*(>X8w?mzGxe9BoI0A3TT|60_xwTg;9!=X98~
zEBPZ`*{smk_=i=>H%7_rjPv&U*9zLY56yD5Y$mDBY=jSw3wG_kw}Wyc@;`C`+yV+~
z_QT$9)+2TGqT8mXrozrEvye7F4w1EJEAthqs|fYXXxX3n^zqXt{&Pe3^ai78J5RSN
zIoK^HW%PAsb}`Rt&SZ;Tms4GMJVeUc?3gh0waDn^p4~o&j|rI8^e}GaPKl1k69}XY
zZEaci%U)3&<t0DL3vf``(y{<n3>?9G@glPjeZYG4wbT)-byn$Zeadq)!DFOt%@kSf
zirH@SvJuLM@9X}R*^(|E%&MH~X4bDW=)3UxSMrU544Xw3zA{>}!yQ}V0Q1M8gFsv9
z&v?-Mbk8AQrps6p-;7(?w6(Q0It@N<#^yXbJ9(|Wd0)}f;miitqOb96L;htiS2#Qd
z7|sfBgls#s&Wl`jdzJXvlYb;WRO~Mg@7hJ@WRmAE=IVM$Z*gk1HE;d(MJd0uzUQC3
zxY|ZexIzWA3<7lV(j|z1Ov3gSD2luG(Ll#QNm*SCY3Qu*G0V+SyitE)&c#Vi>D7pG
z;Mm0UTy?ec%%0`RLk4X5emV0gPi%T*%Hm>vG%<OA$e-T4XB8C8zb#s1L3i|M(YPV9
zI_ly^-5J6aCJ3Vd0l<hzq|hR5ZEUVfN}fA+E+EWAYP7vS*}uB})Tj4-U$j%to5@D|
z_B)?F(JVjhMm<-g6PYY*!(1hA=8+xslRrF2`O;wzC00u}X{YA&L9rjq)7$TWX}*Kc
zV?Fr?1OV`AD=07@KmHL)tmDGyz<^F#G{qimx)vuBEl)=++0pR4M(4@-kPBC{Y3noE
z->qy(8ywP}{(C=iD19X+`S+3ZZ>l-Z{ro>TN^9zGSR{|^@bm=sgFuz`<x42PB{MX(
z)2Cw-5=K!_^+xE~)(*S0pRDc_l5PugU2x1j&)I0QnpCglwd<$c6D@)IO{!x})huV?
zvlL{fL`imGO{uDmjf@0^g}GqZ0l$tMJ$i`bjlaK`ZOVxl?Z$MuuTH@n=jOXpJbf;9
z_pp{7A5uEkC$^8!4<*WQsw)!(M%Z>T7<Chn9J%oB-43EM=*#<bBerO>{<3}l1p|+o
ze%`ssC^qZzbY9vXp<|(qmlDPxCF>g+R##UiB_%z7_6&t{1tKj`rp4j>{SS6FdcH1_
zS)gy8TN)E+_g1+R!Q<dDNpj@I3AM%N&^%}i;>$(P9#*%W{Gp&Rb&ybaRav=heR*c*
zfB(H>S4ZA$Yi>^0t2!8JdZ}_tp(mQQOWTicdv?%P>RQIgt>vb1ns@h@4+a=;w)9O-
zB?$UHPkxPNN{UtiJrfgTjdh?|<RA!7g=kVh4#MDCt=$?u@2p^al(B_Z@bw#XK_r}<
z@<6{Kaxg(%o}3i6{C&;FW)YlWWpx#FuX45NLR;JS;G<v7onD6je(_|2oIE?+o4wSS
z2`cZsdy%-&qer_T`7m8})YI2L4Z=uI|FhO-xt=`B^}5{pWRc|?eW`ml&27FEsPcC+
zoBb>e6WaWls28)7P_76j4fV^-%?))F_1WV%;1QrmRUOZ@MfKiCMY8sv9WPcJJ4%ke
z9DLAbo?QDZ<_<OUMMqx0U#CtUCvbfP*oNTk){|oisRe=ykSKa!0u}>=&RoxgEupnv
zCv%d1ntV(07Z3kwbc|L%(se^#dgwc=AE8*q*wj={SNG2C+mO{{jf_SyeG!+PXc!qk
zfBbmGHsi#n{&7$7&YBMs8BL>mx|_JC+HT}8sud30-5j^E=BnFS+zuT;UrQ_e$&*sZ
zD<~cqeYTS)pS)wey*>ZClXdC^8yDNz_*YUd=f1Wi%}}{rvMIl-bueuQ;m@AkyLVGj
z(H=agATQ5v`TGZ;Jk&H$3n|3y)IM|fD1`^1Z!P^vu1-wF?y0O%IJKKCP^XM%KVhIB
z1&-y^DMe*v%xXI1c+>!G?QM`MOmWj%tKCSL5B1h|MRjdSNnMHfo3F2?nNKnjxZc*)
z$)X$h>eXI|K*+!#7&RYws=K>GA|fItjl|W8qDC$*q?&KLg-ubIA7cpR6uf<vCWxBU
zeH)5Wc|!vifN}BTohV|MBSp5%=;$Z=Sgl`AY13Ss{O^wqrQrF>m(s>7BUGC^3CDM%
z#1Z9*wKZiWC9{q%B^@31(99_*?Sn3_7Uo3H#BTPDXVlf#|L(JpN;$AHT_5}(B%kDD
zQ>afQS7k!^FEBHE9Df0Uu|8BFU{YiD8RO=ZO=<3S@g*_a#V3C$Y38>}<9!K)W?=pL
z+S=Ck_T|~x*RNl%td2xLNn*YA;hHVS@fRQ%8WLA)d#~zU+d&|;LH`MW=Fp)4@be}k
zZnV@OfTAXOgD^}QUD@w;ZY?XWe3~=mw<`Y%iM{1nN5V?f%^;560;ixSPyRsaLr%o$
zKo1I>S1)$v4)D1>FyALomB*;PrrA@SO1n-!P$@uI5RX*=$s<aBCT+>$pjLv9Dp8MW
zYGkfl0S}uEDViL0IX7OE{2x0NLA#-&BOx}nwXKcj*s&r=ZpFpLb#-;fln|ueFj8#I
z4!OvzSG*;oQkeVn&C};iSW3zz?N?MIJ%lT6lD-U`1y0!z!(YCvLKsjn3PQ`vG^}Tm
zmDOPK@?8-13s<8z7QA)unYE(w_1buEDcXPs&Ag;R16B)!25o2tadC0M!FR>Pz5`Jj
z)`dPWG|WcUNV6`{UMOPP^s+A6_;lcd_3g?%w{!ld7r^Vg*nX0$NjJZ!e);44>P@$B
z=)FAzw}a3j0|R3}ePTO%c5HAk!~XAYh)fzq-js_Wqj}ZRAqGZ?3R`X}rOezaCM=mu
z=H9#9MZLVd(5{<XT#S`J1L?mP8UwJwIW{)=g+~ef*Xlnf3pqtpm*m}{JX%Sq8=0B!
zB|QEi)@Js6xdUP4kel1~op{+BQc|;j|3Z8Ji*^P2tdAc*K7XD+O41GmrM|o8cb0eM
zVXF~`UAuCH#wWTR1w8$_qAdK5@8;;esTvgY_x1imgwY$2f<fHCPf^LCasny~3k(1F
z@xvledaWgOv{OHbiLzJl=7;TN6@4`qSL0`2=JfXN+efvFor<d^Gb?L+Y|MvIC<|a3
zY9d->6&0+^%-_+D!+h(zbvCLPh&ml~D;*nHe49P9_1e1MrM>q4Jv4JJi1p|Vkq{hz
zjg37pFtD9yW2g6d2Z=T?kjG{15mF-rk+kTNpiTWKIVwTRwX*n7xz9miH@AaFRXcX^
zb)5MVa~EsQ<9J$nI<PJX1*jPc)8^*p1MaR@htE~k?b;sYyq-G6@VR$>?ZdiZnC3U#
z*~Z6&202LdK)>jlV8!BFZ{FMnN1yCWufkWVwZ}Jxq8I4J^ybKK<R73RtT{pY1d+UR
z=MGL{W@^fzkz;|*L`#bzI<~mP?YW?rTXVeX2Ur6~DnwEQGbQHgId_;w2WA+&xL;02
z*M)p-{@{UtK}|4<9psDo*;ze9!+tC;ABqSb>Gmy!g(b=OV-&6nDW6VHTK)MVO0^g|
z&&UlzAg*@vW;LcaLewq%pVB=xFmRlcQ(Io%y}GNRsEB%VkiyWTU60gJiM=bAE|Zc_
zgvM4wb8~ZBn=D!^^70L6Z|&KW`)JHRFHfkmaiD)u)X#NX?WV`*N@(EPr;lk*Zr-&W
z4o!@9_>xXW*90XSO@fCeCPMeVKfPpFhnPT@WEP7gXJCi;sN4JdYT|Q$1It;{pUWw|
z9+){(U6T-&&o{ZW<#1Adi6E|wJ}78*_0ON)|6ag|oKOQFjVWMhsf?*C6y(yHruqoK
z*UG;q%y)JFwUOt=5!_Q89C89&yPm5PJY!H-`Y<%lpR)=GsLIRF&J9&V`frZA1wQQj
z*)?P(+-_IkeESi<?d|1as~ev+iarW%Yt|np+e!E~i>5F{r$X0FHDzU6)X2-1FO!mz
zHa0e5x$8+jNo^&DG^?Mh<EFvYZu6ptoaD2&7<g1^j!{r@ogx|x8X8^)2DU=xFD|~K
zruHp1b_f^4z>pKeUJ9IP8gRZ&J^4a7Pu=y0ulUWoD|$RqREwUf$$al4)fr9v^l9&&
zJs`0~koZA0va+&XzeaoP@VRrc)w-uP$!utL3~4U+f1t^C)V$kUb^5%J?(mN{{I_Uy
zM#rY7wZQknFFS>XhB7cP^z`<2cXgo&XEoVz0YR3%+Ei_2a!*E{J}Is3;PvrlFQ@D&
z;Unsv6qk!%&#x^`pouXyI=Z0wA}1{knuir6V`*t=+_QOe>ga=rO+Jn^^0TI6pDDBW
z_8F?j+;YgVj@K3tVPw#G0lIF#z2$QEZZNb}@cn}a52C?9^G{)0G1wGk*s%F-(i)(u
z+$<y}cS<hZRo9H;R9MvhH;w5I<TU&QaZdnCbj>hL=s3s48Gva5Uk(4G#w$#lufp|L
zc29qr`!M}!@OAvfu8Qlb3CBd=D&*!=?j|F|7#SJuM$g304?^1F+?;<<kjkAquxqGY
zxk5m|IxU=Yq}0D;!eJKM=*zfK{ZQkr{Y0k5#NExc7XoxZ*WO4zs7=6<tw}1Js%fX7
z<DESD=O4QEm}%KgKDu;6C~B{_^v^_Qg(&Cqi{tC1{W4(}-UWTkbWMrzJ;>;J?gW7=
z6RaBI7F3_um>8K5-l>+vJH%FK@D<OcvgNB^=Uzzp%&aTRKlIh<x9jE9kmU{+HhgiK
zl5nRx-(f!SPCODP2BD*~6Hr%6P3_ekwj=a?i)hlWY^+{-CU#p<abal*&4t>U8gR~E
z=xQP7R{E0mQ$eS2v!5Hhfd(;}c@V~d=zvf_&xs;M?to~?X;Z=g9}^Q3nuoxA;9ijE
zAT27g*}x{CH{!9q7snvEx}Te#ou7XNhlVus?%g{iH((xaUS1KqX$FjC5m8UVqSMr5
z<m0n5Fo?^~-vDHR0P*g;>hH0!mlHu3c5EE@k8oQ<!_>$qI4H;t9Vo!#o}QkpEM6#V
z8N${pbJzN}E)11Cd7J+!<({rye9}Zs&~M!n{5(DN2l-zzxgY5S@^NyarJ-@Ow@=p0
zYlA2wB{d2lfh#>0y<N?d$=&kk!}q--3CXWM$Q^mZ!O|ROM$Qp=kfg-p1DTE(kXCH$
zby?Ym(AR@FwGGg$(AQu5`}aMcITP-pV$|(+I_-$-s?tgBktmvH&rk-GgLKk$Dpsja
zW;?71A0fnvM>i5ve35H`QDYRqJp;dm4!jgN609Cr702S-8iN9PYwyNOuWpZpT{lW;
zlNk>q?Lr!HFR?(E127*&2ty8m^3tVrs6BGw!tcWv4UM_{T$7GI)bH&)%Rx3%Xcl*X
z<f@5-U40|ZO%ooJ=QwI6rs?r<$CcT8T3TAijzyqT<Fd7}3U>(c_7k<?>Z)abSk0d<
zCi^0SS?m{WsE&o;;$|58-alwt=+~U=uG+naurms*8`Fc#3xx+x5C|8Lj=*k)zdx&b
zf3wIf%(AY3c_45qh?eKPFm?F$*4q~UyQ$h0rmC^-Mbd;*Z0Lv<xn6!IhM*C!{QWd7
zZ7%Ls2#PlpON5>1-EAI*>;}3EQGcwJ$CEsMQgYnud~Y4+JG@FY(%qmYYB2ORd;4!w
zW5BSru->bl9Nx#j6q;FH>9CSqc?D5ME6)~^k`V|c0IR6ek||a_>@0UJTp63kv==fw
z2mhS&ku#u*_;kFa|Mt80w}l+cJEk`0C7ef&iF}@WT-p@&m67>R1*PFhRPiphq=|?b
zZi-*we^y4j=-;^6_pGh0fs&(Dr>ff7-M#K|l`v3`4jm+^v(!H!U&GG_c?-f8gfvuc
zQLUn=nc4y!C$qt-h2Io^a`$)~7ZfGgig{pX>G1RMS=0VS^5mxO>B56T_lH-khJRH*
ze|N*uMl>t*k*vVtwC92CdDm2}g7NK_yq%q$IQ~`&17Y?kTj2P>!0WKE0?R+;ps2`e
z=zV0z#YxD@`c4{g9VddVJddBN6W;4;;Q&{OffKY0yB*uboedigk(8YM+Ok?w!x~@f
zL(`jGFe$v0G#!K0otU_qeROJa5_43@M@9gww0CfT!XjcntJ9vM2~!Q$8fXVHd9;KN
z)FSd1u7<r|s%4or7uIS$vie<nwde*y_>I#h-7d$!Gi&)?@ayjg^|8&2%IOnzKgo_a
zVup}E&|4V#^C#f#+aEC2n09<YMSmT(k5sy7%OFJ0vhTjn-(*V;t2;j&x<!U*9@Va$
zZn@uN=Qr>6otfXdqW0S2=m6^v)4pS~_JrA5j5j<YD#>?YlhQzc4$}^?`z~@xh0goC
z+gHP`UvX&P%#PcYszs}8XSJugE?B_w)CiTcMB&SC>MUEu6+7KimF4BXK(oMqfk8oX
z$ns!FSmM53Uft=2RGp1Fgw0Y3PwSW-4R(g09Xc2BI}vp90M4NPjtmbkjWx$ZQd?PD
zo16RH-%&#zlun=%BTB4#dKf~$GrS)5<XnwzWcf#j=<cp4?f}io&TjAM$U*lFMFJXJ
zR%Rx8Xe=xS=#(HG5Zmu`2f?V3P7fVBrmdvJV>MU_=7QV@F;?dN#b=o80Vtky2QepE
z=KXbLWrvR(fxoH-4Ws=(N}iA*2A{wnRJ;LC%1nRJHm>==NeR>gcI`r8xPgf6Pz^3A
z97D$nEl{Ku*y4VVjEFj~e1R(yWlu`##pB02FbJ>VVF-qCD+9p}@n1(r$KKu^<^_3q
zdH8{j9Xm$+jE>FD;u;9-u!BYalde7|C)Y)YZleK(nYnX@8n9siEr#L@*j^SE7h79e
z(5`?W4TWA+P0bNWF7{@SKO+U{VaW1vhtWTajU{rZJ$qh47&t_70X{3l`|qCBNK&+3
znx*CBQd3jqZr$SH<#k0y8>$Y1RU`jluOP6U4q8^lq0o}yP^m62Z)<G)H8@C3L$d^J
zzApH}lSz#$|DoM^$Z>mX6Oj`T5Ws%^yr7T}k`X#15Z6mfORIcIPobIQz6Uk<fr-f>
z5=(R+!Ha+<Z67_lcKte74yd7&JBmyv2wxc5n^95BTwE3qp|NI=@iDN;$#j#RXEsW9
zas`6_0}McL!%SUvS^}8m2dGq6pXtdB0QIFuCr;+{$&<Rc8H5n(wg~($kQm_+Me2!@
z4U@LzK2Gt9$=v~^yAPICfI$o(G#5)NtMSQ6ZZ581UNid6Mt2#=`@<mElqeTFc07hk
z!zAk3^Zh$Be6Lmhr}aku^S>uaReTDb9k4t!VMoTsa^1FF*#f_T`@*~iF*7ZCM-`-b
zu9wiG(d>rA_Ta$-sBD-u2UNn8loT9h?p_p`4k^eH$TBeAIZjS^z-sI2;5lTxWI*1T
z>&}5ju8uu7eQjr_07w}#^KZaIh{W2tR;Qnd?V+S(5u$I{g}EqAN=Vp`=KlD&8af*&
zG|+;|tE!G1I|jCY%a;4&KD1f5vLK(J06u+60+OU({SJ3df8@yRGt?Qv8$@f1ZhA(B
zF#7grqQh;34g)gsE)tTO+FC4#M<iGf#ozMs!0f=X6JuhiA>Z!b50GuMxweQ_Q~)Ez
z=x&Vg?c0fpP<g@s+0`<n&<#eH6}UlMT(>dmx}M}tV!v2N5n}-R3Lrg7A<jcXL-Ri@
zT{+UHH;D6e8~sONj96x-4F(5=6O;{Z8k`zk)doh~v^X2ggd7}bGPbw0fIBl^-4DYe
z1chHCBk18aFm4>8a{o*0-T*EDE1*#Tw~WoBN64lq#_2C!?8bC65obXdQ-C7P$jFGk
zVX&~n9I!J)L9~HRr&;d^1J>dS4i689aR5P%9vUlV0}hD7LUij~VS&b72|h#D^CWya
ztqlzjBXMaxUv-~DpTr_Y)U&`bgn7_LPm`_+ogzO50eB%z4GgkyS<Ufsk`fXhf`exP
znlXyBSj7_l@81g`{Xx)$6$A0^P0wSD#esoB?0ka>Rfimr(bmQb|0<L%bXh^qdJx$i
zUv+6=6T{hv{@QYjp`V9u-)@9^0{%!R2Zu1+9ugXc;$K6YzCj30F}hOZ34C8BQXWo?
zVNe(q6<E-I&CJyQ>(YF*BVKqhoXjwJFDxv)<Iw#1{ks(=9wx;U)pSzYu;;0dqsTtm
zLGyup0lWljN)U(U8U!skm61pxao@D%)*(kEHsnE9k(89Ye!UFL>Q)FZ`de7bWF#a5
zX}q(6V9Sp_gTNpYBPj)6)c=XUp_U%qzrUfc&sX=$>>^U{M{QWpy3+K$!0bW=tTCy)
zrX90m&8(1QoilbjA3}(1g2i>BCkb4EmSA8Y_*)VN3N2BX>@fLnka@4fBlCi7!gv7o
zWr9zjb^7f2bF?Xs($ix)=*Tf0jxCLik@4|-Ku1{0ASo^`n*b4TU*p=b-2Yr{+}RL~
zlOUDmefyU7^(#8s;sEyORiYu??XW^gym3Z)AMl3O=H@?RW7zHR7hO;I%sh_EhVoy<
z;7DPVN9(@XOc0S$XQzX`J&e+l-gJ%4%}-(}ocVA|E;nRlX$~C7b=?%)x9=u^KbB2o
z<S&2(v_+A2y}=)@Xy3noAD9EC2M$nns9}g#ocYL+s*)1oc@)uqY<Pq36gtd%qJ5l|
z1&{3|Vc~qF>w*HtpoFK}@S<Gdgo6MLA|z1M)@HO7(Qbplv>bK|&0QT+;thGBX<%hx
z@$~s~?L1pEv|gcJ;vi^dQSU{eE$%=XY}jRPKJ)tZevjj5f<i5a>0@Yc5c$*crS!@S
zW_!@Y)Rdf@Tt-UjZuUb)9|-`6dNfMa)zwoq^CWO8Vwd>!%-=s1aK8Tf^$XEOz4tU5
z+oz8okDocSyPetd7}0=S^{~Hi>enxrM4=gE=~ek3<$HKu^IOKJPtrt7x>4~dQvE1%
zr=u<<e@A-j+V9VeCl>X#g&cG4xRm>nm#&`bN;Vv-SmcaHk0yQm*wETKG(2qZ@L?hH
zI%>{qMvCByVp<nD-;g?CuRNt5hKT8oq}ucspQ1n<7HBFobHh3>4#-hmY1hiPFG2dH
zr+3^~vE2L23I!U=2pPj+ei&shxWYa4q2v4v37;9ot;-@_ULQL-vlT9FuTC=*9X`3e
z=!<lRb@Xqqg%_Bcj{kq;S#ay2gtxV|p=I`2Epr?!1Of|e&k=$)Y>4GV@|A0HW)B|5
zdmk7AjnX-iX!BiM)NFf;jOT7AVSj}$W1ib~VOEwcOqTKS@nK<b*H0vW{*2sSV^GZM
zNV)6tt}^u+zgLr&Zin#P&<pOmz<IgVrkqUUUCZt55yp5xC;gvqi+aH&Rc_zDVDVrF
z=`9#9tgX|7g4%%t;e#2QoJ5YE{QcV_N%1=s->i_M2t|8dUBIQDZ1q)|h5aO6#U24)
z+j+ctdXI7IBtMU26A|fyu{tl$>HE_?$OX`WU0q#K!Hpc~c0aLjJiV{z`&#p_3At_8
zogoS3g##+<j$LX_xt6gkbpFs+AaW6v6=)t9T!bbnty%B4z)6kbP$^-*@US2%!@s5S
zZ{NK6i=2yi2jBi#RYgP_FsmYaK^#9VM9;;Ve~Cs%9-)iv3X<~j7Jtfp-n@AOwuPh*
z{vc6ZT3jMS!IglULyR%f(%LPIHo^>|diO5Ss%dB1Gn}yop-vu-&gR0xqmw72LA3D+
z$c?a>rl@6NS?<IiE-5&YK~4k#g)w~)cE082G78yxV+^5KAxH<YAfEiFuAYDy2t^X<
z_4MwjZ{Hk|`;l-aCp$iVFaF-o1#XKL3ihHP5#1~G%+1Le92~T;u)xKXb#w&dKpQyF
zPLAWp2awuvDy{5?VL3S~o)kxM1)Cci$l~CR;{Uy3(kVu+Ds5{6QEBVxaW3~hf@{Y}
z!6TF-c8r{o`f@P=M#tWK2Wwf`I;>6nS~*pV=rG2oaPi4^zj{nQ;!*_GC4CJ540Bpq
zniXVLEISkfe}4rOAc*@&PvX1lfV|Jt*482=VYdMsL`a}G42+s^%EIh^faqdjBtLP+
z(S1iFfIixYd!I!mC3QD918UA=$I$7U?{f0;z#@Wh(hQ_<uH$H489#V%_S`wAf#MzT
z9YQ38Uk~m;L4JO$TTceugJ00=*8|hj94sv56&2agF@c59@x%rfm<P!WyNU<k7~Iey
z8i~NMeSB&w-<-r&2eig;ztC#Hz|bO;uoa|xg^?BQz;-lD{u&2&e@jbDWaK?Fv&&F}
z&@I6V_dLVky=&zSY_5NSngK%NVq*cfD36H<!`m$u*@MkCfrq#Ab_mS!Q4KtISeiud
zw;Lb|&X@IS&LS*!ZWz~WZP9f}`!b+==`lA_;07I<R{#JcSD`w~ec--<p;J`k77#$K
zP`!Tr8HOVUW)D|)1PZ%))?GEVw$|3pt}YlH1q20=Dqz!Ut*P<G=UV=K4glH8(lRq6
z1KTKW!UW&hIn|p7gUBLaEierXpBsiSQU9#3SCN<Bf9TM}<RttLIk3$DjiKWLKX8>S
znfsHkgzZ}ZvW|{x0M01<;3$akHS}Z9A4fTj9C|E^;3(|J3Gq91H8l-QO=wkUQa~P^
z>CJ<EJkMd?Agy`F25Hh!_nDUNZft`&hK(^;XxI{l@kmHGgHQ(#;N#;Xehvg}_A($S
z;3@2^z>+xL<kZwH8JTb(98_|U1>)y+y<!@03V#0{1LBP)80req>2u>oE-aCThK6vb
z5P{$rVG9*+*35|e1KiDsaS$3D_82<Dy1Iv$m=HBf^Yx<9Lyswc>6|al#+H@3+FE$%
z(G0!&MdxvJbo7T0YN+@9hHe32lmq(^BK^k&1Xi13iPv`h`gIgOK%hT^gTjsr36CF_
zV2Ckxv-*hh)WfJ0FuB5xiIj0GOb`|~UD!jx8RKQctnBPqg&2!PfkK@F0dp{(P>^uA
zzw`6!tE*_*v~_hkVoM%U7Q$4IjATU;2XzYy4qm|}L;Diq=ZAG1FJuXB9v)Ce*hxJn
z9}$REewGtW%XI8dLH7vCAmm6`PEdeu+^9nO!Tu5O^Z>^F1KOCttssou400sAnLv{b
z4HifiSjdpMac@HC1*82{Nx6G=#)9c`u_!h)^kzH=$jAUMv_Zo?J$)W<0firJ?_-RN
z?B~wiw7ue<3O=xshE5fb@U393UvOPw_ZF0XNL?f_-Pz5cNxf5ziEg(tA3+-{_~HHg
zV$qEtaT8!$dU_f79+*6+R;ZwjF^s~*5g-MU+L{LThD8E<_|T$5et3`HLk7S-;NPZZ
zX8$1R6?gL|FPR3YpuocR4Xy%00`!jESpZS6wxRKWLV)j7bZb980JgAFi%W7{8zcIN
zfpcz1OCz7u2Vc;In+o^@Q;oz4IY`ftWQ7R`9F1T!P<Tw*WP-U$dV32J5-I@(&|N7l
zJ;~3{&%?6_F3P{Q*LxozSEz`KAokP2`wWZ;-4`$bSrQimO&WR#2>;dUo#LgwOr9C=
z+`%dbu!*{fIwEoXI$E%5s;UShVhG*6i$fol)PCF;IeKsOm9K9&pShN#WX0p1WGI;j
zX=vWTK?-UG4>HOa`j6KoBp?h0FdiBx!cHtt>=}wfZimk;Au$pAJfP44vSV30B2T+S
zOcwxPp`S(kd<8ZQir2wXhMWZya3fM46#yG_AzTveS<t1iy9HK`rt0c_>{K;ilH!#R
zHqcmcTN`7AB8r@br3iZ<!XM-3vNC)1#tpa@tJ~XE#KoUX#t=jgzGCuxiUEP%bSpxn
zAT3P+DrJ4W`HvR|Aj=~2T3`=Xie_Gd;-^=yq@lb+PUyV1!<CErGFL8f8l9FZLBm*C
zQDT66Oa?38z-n7fbbpAt)>K!+oZL}UQv;Za@VB*P$o!85mgKnBrl$Ky>@di{VGsK(
z00uUiba!+RoBZ&sqa{zhS0C@l@d$>0W>3!ID_{eMAhKv@nZEb%5JU4pS2r4<D>-=*
z^a=(}Z2kxf4Mn6G)Q9)Bv}B;@0L#Klns#z0`EgvbA7ThrbmTHf@WB$%JbZkq>e;fW
zqu9C#GK+Q~JjGE__aRAO98o5qm*5Mhgm{YGQO8QrJj`+1-a`8nY75SVc?2{;dW(ws
zjr2xzx}H7_5{O-(kZ(|c%l(c;1O^UGPuG=}ird<{m$-||6(@l3Lt^<eF#+1&ja@F)
z)p_t1vITw!2snB6Y+^zJwqHm@H!3PAT^1Hb0~bpY(l;F3#2sLDlw1k&3&m16h<1VF
zVFrc`6pRla=ppdP%F42{f5sB}l#pOyWrh7~#DW0?2X2P_E)KH;S3zLR%wVX+ZmJOw
zyvnUT1S$6@vTqWx3aFAHA3m&s#v*NiyW@ql096B}0(l-t2p5E??kRAx18#x^bZcu1
zcm2@RG~aR2*yX30&u#+Y&>-=-04|$QIxs+7Wu@|9)oU1Bf3$3X{IKH^qo$=*S5Z-c
z{Zg8mX6EK@(6PXR4Gj!XW4k&!P{>U%=>V;$h`<SeJFvRJ_6uZzzDo!4A^c28<W*bm
z2~s%zK_pI8SPV}-;5f)12q#c4SR7hBGkn-91<!m+bhJ2ZG40xJus_3)3>czO;AnI0
z+SBmxKEOHLKlXbv?%!WrR@REzkL^-*b$v}u%#h$<O2Y7C_okJb3_<X5)bnHR(y{ob
zLShT3FFe;c6~YDPQ^@jQcW}5Pks-d8&00;2jp3a_yAq-D^w~2Q#Nf^cjQA8Ezq~L8
zr}eY6w2P<^cmXTIrG<!vX#xj@rXCv~@AcvZG#la$#=N|BBnJo^7%(K3a%5u6LVL36
zNkFJ2H!8x~_nPb4VBTtK=md1PwUuN4YG)_-X=8Uc^b+alMsOI^j7B3QV+;*S3^rZ?
zfJ3Blu>%PlTknGUvK$x^k{2EQ1gFIMnz_2h^lRB1?AZxegy|K2I<o^?Gf~d)zFs&o
zFfckjZJ?_=aZ@Z-VUNeC6N$(1wTFNW`Z$$BB2I;Lk38#Qhr>5Fi|@bI&xk(%x@$*a
z9ssj^VoP727F%G~G6l#toU;81)tG^vo*KM2G{7@~pFv2tqpc0$t+%7268F>81g9&}
z*WKNX71Y|-CmY?Urmnudfh`m1Gt=<rL+!-iAPRsK5JN6<Ez5rp2zcE|RNVVaK}jk0
zpDdjIpg9h0Bh22~`ufo$NAwL0UcGn$Jt{sY#~uU^vw>s^IT{7Z*4i37^Wx!?hBq7r
zF_=n`qF^TmVuUWB{8<e<KCz)d^!|?*&;?N#pyXgj-<w4~3imn+=|;PEm-poy+#m2k
z{eM0pVUpaYv{s!wj@<A_Fe`TFLQ+bFv-kY@WNc(IsVTJD?lzxDQu)1ovnbVJU3z1A
zW_@~VWX8>Wsp#^+Op%X7^rDhOzlyEx5*C-J+Z24>6I=CEtttyHsbnsx6fT^XFcPq0
zK)uIExb)Be{kyrPfPB)5O#$tnx#@kLJc&+E4~U2;N=$qy7Ku%fu-^o|eT&SF5+eKl
z;%Olvb#PN4`x&e%XIIxP^ePcf60jLHHeTKT1C9phvoPm3H#GqY9TT?yg3yI49b$ua
zjzvjd9~9(yFd^)u>wp&*i_Nem_)SQNka0@_R!(7iOQFhD2v?|Zkb0o_H$WQ3W)lDh
zOfI^9s0Ek-@N{T9&=i5l5VS6LHbB9!lCfJ8x5mi8pqza70F)q=X<#{YQPcsmv2o*T
z#Eun}o!Scp<^}IKt;RIImFWef#8h%o8as)X<z+mt2-1+tUmC2rU-Ec#4WS^7B3M-g
zh20H(XaFZ-vkP|&=7PI~c8gUGP><Av@CSE*poMCHRgN)*c#T4Y2Twq*L3zgCq4uCE
z;2lX7N{yrsJ$59EAU+CT=3vw%qW{j#VRtIFW2AvS!x9hr0fzww6Yqd4c!y<-`p}_s
z{}gN)ZGRjKjT3BNTI$W7`JRhe1geFzEL#2yN)O1pi}u9C#5$7(dPz7WvL)65kPZZ4
zkb4Xi=;74Vl#BRfpja|8vOsAFN0aLpSY?8t&hP+l0uewV1<e6ULS)m?)6>w>9;Ktp
zcUe~!7mvoefIN%qM9AapSln0yxL{OXz$B;_D5p?|=a-j7fy#mWaWF6vlmqP4p<>`4
z7#wWj^T&oG5wJ1fJDdewyr@Pa<RaoqO%=x#0*YgbG02!}upn%*%D4FC;ptf(DgGS&
zC0F=9P?|vGU%q?^t_94uu$hZC_7!Xpla}5)L^*qlbR+ER<@c^ZM{OlU7X*z8CA8JW
zUEai{{E{#++^#76<i}CLy*bZ_jVU$jFa7+?O-(o2Ss@`L!ZnAw2965c1z9rif3!E7
zniQ~O9gH4e3aAQm02{<-`1Mf?;U|DDR`c^GR2JzQH&%<5rY|QXp7Yzel3VrcHC@i0
zUq@;_NwvOv(8_i=)969#;`=g{>+GMujbA*RxLY<}GrL<g`<tzrpwdKg!X-fy8=7xr
zpKKmjx)4a8N$+EmcPP3RK-eLcv%j)eG^H#h_H9aQOvTIy6-!TU&j!EsTy2#A>kZUc
zBr3dPq2V210Av!6g=kCF*8V5m9So?VK5*bfpfu5wj6G#yl~{}35`Vy^%zJXaLWd+`
zol~di$UPCPm@T}6TSEWGI0$fXq@b{}un<L_rO8fggo7QV)ri|T(-BT0-VihV5R8Pg
zU_k+WD5QIoZ)fKX^w5Y8WPoCdlz<$B?&OPr6F^s(R2na6UuQ4AKabhNYOk#|!#@$A
zsQo5JM)rTpXi+S+T~~*YO0dVBnwAz@gwmig3ko7@sHL`t)DB~_W1n%R=d)*fJoX<v
z7^RTt3_`oPsrrs$@7}cTDZwz`5r;5Y!%VeIH8V-c=hCsT<XPI;(Z(0ueRtTfkDX@U
z33Qxqga+Z!BGaRz#78Zhp@!|YVeAoxi3u4CI=Q%*7<Ov`9ig;?6`Ar9TeNsw0|lu#
z05aU!zIc`jH$BkShc|D&qM!ir#t0YJFBVHKc!vmeo24)ISgUx1%`l%ADxR-F*=R9d
zyi8^z7$(EAUL~}+w8R@yYn<6-%1d?RB4iIRnqDLBwDa9SAG^h{ZIEB65hZT-gNF|d
zRF^<<FTPJMEYwO=R8~>3W@uk-hUpw~Ge-~~FYl}Bmj_?|xc{|d%GOe=`(buAh|H;j
z>7A3)(|z!i%Affm3cfqYp8~WWtAOSGf;Om14@8f;B8&2Q>HV)S4*jO_D*I&-EW`3n
zAIWXd$^j;=3QMFk)LTS4y-yr=OXYu7WJ5j#LItLbQ5Zu{&pw=X`)c?I;OgCX;GqH4
zL%?XbEgdcrQc|rPixX$gi~?Q?h7q?+seN8TQNR(=BvoV!)OJ}6yvDq`)NSq%Zb(;d
z%6l`m3GH}LWjyoa{HarV_2HKRP{1fa3d><m``kW3+^K>+Bsg=5c9AQHC(^+!?D~B5
zN)TBLDH{onho1OMmi~icXvCLt0SHUSrp3hQb!9w&sHB-`+=|W)LId6BYaVjLgq72a
zml`+rS+_8%Uto`XwBh9yCo0g>-h697{b=>Ri_Nmx>~hu4PIL9^!$Yt*rgv_PC#6z4
zV~<ulGBuvKLdn3-1q}~q8I>$PHnuiY;1cQtH+Kds)78~p_kxjKu{jML3qUvnJ-zel
zS(7L&;5kIz1FzKjOi>XeZK#lz%8uCOfiT3_{eohE9<bM|SC0h4%I-rP|3?sfj*0f^
zaMgtHabCXc0;-MO3*MkonI<a8n^smtON+?HY$>282;*`q#J)rH^lWhgrCc;#r{RTy
z9twbiVY&J4{DFR&w-6a2e1g~gsHniU=5xTfm~j@CVUP^R0ueNWn__NYR*5CY*!VtO
zIrb#rStAgXPzixd>uYK>zUW*9P=quIIDnoO%H=n>jDYlDG=jK?PADBYsb!p2QtO9W
zgItZi-rgw-{uqU`_460s%V_Kw#j^vXPW`4&)!@LB4@US?GBcH86<+E(hmQ#O21v)!
z`$$B}WB->WLwu_-G}jC@vnyBJ!7kwXc6oG`u=lfO-jWYrADw{J;FFw{CsdANFQSLW
z6aDTKRh~=aQJp$SmAUlQvN>9rBTR51zG8TpPOssRn3c6P6cSamR(pDqu}WT#Pfr8y
z9a6bkrg`r7PuA+R-={9t&;K?!YMFT_v=*E7C)HP}?Pgv$ts|3UXTC7`@_X%5PDweK
zo`%TO=%_qfpmLH@Pt{3|*rnF_O|Si8YaFLjek^H9_v}9PaO=Q1YCo?$#~)rVk}Akc
zUx*&@JRsU}Q-fOR4E67>^Y!!S#~{ZtRQwfARY)}DJz8Ojq`*+oIc4knhk^3B28W%s
zbwHIKSU8Xo29715YRKxD*olBDe&3T%vEaH)voMt})!W@%vkC5bOhyf&Z#DT2E(WaS
zur5;lNm&CAUd1AzTI4vUB;`P$eO@I`ApAUm$BUeK{*jFEk--BG{;=BZ_`mz%=FdX=
zr+;$5+`tm3<~C5fOPqIiovb!UKTEyWQ03RR29cfO>*O`REI{lnU$deWUo|vGrdTDu
zntXHw`!<E`rbE+{EHsTG9!I*6dM+Iap2(nz!vll<*B2n^(2zcvYPvltj*rrH8S3DB
zxw7l>g#~GNkN{l0ok}?z8TG~g=V@)ap3mVv2tVU-M8Z`n!W97$Jnv*^Kc2GkR=i;+
zL7wve{-MT@k9tvlUt0*v108q&)#)peN^Mo%=3UMa@AHm3pDFZyx?fDNNZ`YUs;b}0
zJ|V|@3JZ6Pf`uGqo*iLL|I~S#bJS2I<-7Z+Tja<Fe~=85n$VX9N5|%cADnqb>jg;z
zVMhXVw^i@uyXp+ZJUcs7P<OsgX+b%B?f8&}OP*58g9@jSX^WmUy~RuOc3Q^C^&5Zo
z)w6_~iC1q>ErcE$`c!pNK0bb~Whv*GoWsrM*S5Witdkd;4{sgXQfOh{?J@P;BwC{I
zUF%TIK~0kF)$x!0o29=eFKWouFBZv?zmPSt-7RLmS=hf;)73s;VS9^X*X4N8lu%)6
z-_ax2iv!-A`yJ%j7xr%aY!&-Jc*uovpZ&~k25pM!eX;c3V~piL>2tsR>Di(ZS#kOl
znNr=km_r`Pv#rqCoX)I7e!{<)*1(sIoBGlALd0dk2J5w<pH<@b8g6c|)@Wzg_t|gf
zZL6nNbtL~?K0x)Q&~-eno8@`>xb=|&o<JoYRw`$O#y*AMw2SAhy7fcSZ(33KvP7#6
zPq50SNjq;i3;M8}V_}^Zs~-zG$4YhB>#J`_kjQ22#?xFsD-6dYmh{&It{;DTZ|&he
zk3WshJ$u{yYXWw$ki9W3r4`{@TYgR+@||T%LRQM|tz1XVvTS)g`=3zF=zaU2<`q2K
z&%D3nv1Q5fmr;Qy!^~CU*Bkcn>MA+p-g=$ja*{;Mm2N?ypK!~oaq{>H1u@4E8Jd~f
zBGiTNj*#TV7DnZ#1brENsb!!y%H1zi(JNjW?jQP@PiXhD5pT`%4d<QCE2inUj$icT
zuZc6D(iW}X@}%-(E2q6|XraO4kg(hF!3)=F^|c+cIuXIbC3D&P0t<dc$7=2xu{!)K
z)%&$b#+=6P=<s}l3BA9z=Bv}WH>LliiF}y2E+waW`{jDVC_z|x&k1jpj<>hX!)vn&
zqn4><R7W0qt3wX$yq^2IZ?1P=gEeWHzW?)AAGaTbTl1Pbt86!Qg&Xy@Klc^=t6Z)q
zY!O>>`<&$UjeD*$-&ViH?UNLt+GtKx6%W;XYx3Gxb#rb04j-q}X8EIus?GAn=TSS~
z=-Ne&kcTIoqRP}3zF%OaXTJF3N11M+(-w8tIS%J7{d*zWEBU(uJYLUZxj2TTY>s#(
zo3Qv*gg4hSPRvUc{#-q<)UCa}K0m3Jec{r=r3#DSoLn-;IzP+W!AM?Yo-0pPV*VGa
z^MAQy|G)oGw83($chu(jPp3U6G@s?^Do>ru63RQ}zV7~!Y~}HZdtTw6`?7}qkUDi1
z{^-4I9nEBCZF_^T{P+M#Nhc2>Cww|dh?!QCK_sf?+lI`VF{8w_SrgBXWTO#Nlb)=@
z^vWYc+pA~WIx`79wB(9}wUZe`aS?Va!6G-;J?(7S3Cnw~QZ-PiPYd1Xv=$X9(*5z~
zho7`ls3v8x%hiUR8&}%fXHGlwTCRuGdnR}MrJujo^h@n|`py4xsDh)(a)KI^-)OUW
zQ||jkC-&{A*POS=J>ATUhC#b#x5fS?w>x+Q*8d~V{eK@c!rOJuTpIZ@x8P!J<W<6r
M>$2BU#B@CV4?T{N#{d8T

literal 9302
zcmd6NS3I0=@UL$5)mzl436T&jdXL_TwxWgT-Rf--5hQvSM6~F=vw|SVDp{S?yH!_r
zHov?7=W{;i=A4Uj@m_pq=9zb9-g)Mk_cKqdj+P1uAp;=>1_p`hYehW_49v@a4*(lY
z`Eo&~iGhKQp`&4-goc=ym{?d?*x1-OI5@btxOjMY`1tq$0DypifRK=oh=_=on3#lw
zgp`z&jEoEj1d@}JQ&3P)Qc_Y;QBhM<)6me+($dn=(b3b>GcYhPGBQ4T^oWUxiJ6(1
zg@uKcm6eT+jh&tS@#DuF92}gSoLpR7+}zweJUqO-yicAydHVDzA0Hn-KmW65&jbVn
zo<D#7;>8O=K|vuQp_eaT3JVL1h=_=aii(MeiHnQBdi6>|LPAnfQc6lnT3T90Mn+av
zR!&Y%US3{7K|xVbQAtTjSy@>{MdkJD*Q%<jYHDig>gpOA8k(A#T3TA#+S)ogI=Z^L
zdU|^L`uYY228M=)Mn*=)#>Q{nyfHB`F*P+cGcz+cH@C2`0D(Z3mX>edzO}NlvbMIi
zv9YnWwY9Ued-v|0y}iAIgTwpx?;RZ-ot&JUot<4=TwGmU-Q3*V-Q7JrJUl%;y}Z1<
zy}f;We0+U<{rvpEV6eZxe?UM$U|?WSP*8Aia7ai<XlQ6ySXg*?ctk`*WMm`+0*Q)>
zijIzsiHZ5};X`a}?8lEE<Kp5zefkt1AD@trkeHbG`Sa&5U%n(IB_$^(r=+B$rlzK)
zrKP8*XJlk#W@ct(Wo2h)=j7z%=H}+*<>lw+fBpKkprD|zu&}78=-an%#l^+nzke?&
zDJd;2Eh{T4FE6jCsQB^YM`dMYRaI4Wb#+ZmO>J#$U0q#$eSJeiLt|rOQ&UrOb8|~e
zOKWRuTU%RudwWMmM`veeS65eecXv-uPj7E86bkL@>-+ihXMcbHz`(%Z;NY)czlMf}
zhKGkoMn-=B{yjQ6IyN>oK0f~E&mR~JHZd_VIXO8sH8njw{rB%*I2=ARGc!9oJ2y8s
zKR>^)u&}texU{siyu7@!vVuS$R##Wo*4Eb7*EcpcHa9o7wzjsnw|91Sc6WF8_V)Jo
z_YV#Z4i68Hj*gH><ni(G$;rv->FL?o+4=eT#l-~*g}S`Fyt=x&zP`S>xw*Z)y}P@+
zzrTNYcv!8m2uAM#0{7P@UKkie6#pK~iN%mx3=CEbRYf@iaL&OZzPY}AcaO(WV4&O0
zQZqFSV>pfT!;k0nebE@C(P0sQ!>9^ZOwlJgD*c8PhPtp-#Yg@8?2e13%|1<}3m@uE
z%zJbR2y&9Ww6u~=AFj)~xVXo!LN7btvJNHF0G{g+wc%r&g&~P?5%h#mAu5PH4`5se
zm{-6msUQcAj`;tQ!4ACN;rcpo@}!KgyLM3UYZq9QmngsT)&+CpS{D57e;BO??M@qR
z1wt!?p@<RtxuC_>r+a18b;h#bMrE-eYk1+7;dorp%`XrpVwwSg{Bd$EQ?6-O9fow=
zID}2$g>pYyvU9p)!JqlJM40`6<;AQYB0U{QYH7U+TofIcy=>U027=;$<eA)<t@jH&
zV+*P7YIE<;*|@2AX#H9fDy3DZ_ZaKRl>L|mdtj{fa1hJGEh?b)i`oK>&MKLNFO;Og
zVU3nn7loZe4jW#k`Mk?4vwVbc#Co(rn@&#pfZSF2&)P4oOw@k0r)_r|c91@S%ZrMp
zbK8GrKK9u4fLv%hxzisDoJZxZTaqHC`*oIk3KlEyS)AS3j**7PCjv1@N9#~Q26FOW
zc~eNe7O#@?&3*apF>M0Dba}9gU#81iJ@rP(h&eH$ae-*b?*8IgTx;l%IVl?cz1rRp
z`_d;hJHpnHrXKsoB>GUe3meUu&k(z?K3wJ%kFN?tnjR=q4X;IKesT&TfxsQ-x72hu
zsNKZjvS1hAD5sx{`oUHCITfrOG!?54^fET)>?$A|Kz08!Vg=#7m^Z&dh%CvVPPP47
zYtITCEnDGo-~(Y7;EnMEHa)#(H3Vt8mkY2Uxo{EgQa)@1f}AYkS`+^S)<!t7bu_i{
zPQ;O1?pc*aRG<-I$FUcjPkTum*H{6}fByYK|89b;DUr&84AD5H9Eohy<In?3LEu_M
z`2DvppSi$Lgk)jJIaM8f5w#!5a}Mx`an=r_a`*WMyn<%})GU~YMi_R%mBUi+7L<!L
z9686cxgdV{(kn!lG6wA&hw`>h*YfOlG{@ATvkLsvi|VAj%!ak&!erV0IBc|cBx@-H
z3+j~L<JrJA<No`Gi!1_ZS^~A|=Y(|0OrPljLGzN7$N}$sivfT5nH3s-dJ@F1+Ti`v
zdGk4-+U2$!*d{Gm-P*_6k!Iqw!X_Svm*aLQ2B@|O+4EO>vyw^w%uVJPkF~>h$B)T^
zY2dKpQnytb?SlUxQ&Ry0zuEHb7a%A*CHDF(htg>CwB|~ItwY45hN1YF$;Hn~j0)<M
zET^PrkSWACQUC<St!MbV8x_*e*E}LZG!D+Xj^mEpDMJiILwO)@u6n<sX7!?WL#tPG
zP^b9jf~91Zg5b$oCK5!WQq;IP%&0m$iD-s2203RvsL`OZZ`(z=h5&-hY~S5S{mJjF
zWk#osVpWkk{M&BoX)QDQlk{PCmxm}MbA)Tkzz&IkAc^(64}WTBA6?Do{lZ2xLih4$
z$njzPkKE4j5sh;3wo=v4omHCqX56q4jlIQ~+pdz!Uoi$XSvy2#7Ea%CmLK<eQ39%Q
zn~wRp3GSDK8OB{$J2oqj0r_hZ3xB(ZXjO+JkXc92=MZMv7Ll8ubLhYg1WJA8%(J)?
z!C0<Dqqxzy@w|)RDWjlfv}lz^#FIL#^N>D&p(JR!NEZNFF+dM5M(v8q^v1Yf4#pm?
z<+a_8Y<~P#1vo=acUoSXznP#sIsn9w#r-Cl%)SdK&;O5yQ!Q{3$cQ675GVmfm~|<l
zy2@nj+VTZv*G<P|@6D_+^g79YC{M$qGFam8KDmSUf*P9osl5l}F9<I`8Hb;Q^Y_0<
zEC*k@sfykH4Q&?wC89PplGcd@t>~#ae;Tm<!Nz#CoAWsEyqME&+hJ}=Ib8uvUD_!W
z>hStzQEh>uD-muhI;sK>8Sn|L`l-p*f%lX=YK&2GBS$fFULH(BbF%K!;g-E}GrT|p
zb$Wg;#asTUeET3e3lAY|WS0~|g(@{b8kLzTfa8U|u5b^>FLX03<zYYw{LOvpHY{=J
zX-bDM1a4qF`fA)iPC3`;WD^Kdu%OXkY<YE=Q>r&KuK*?yMoOA;ygY6#+CO0Hh(4LU
zqbuldRa<6;I(^HMzJOt#>Rgf(8;2no-cRp~b|o)jzvJ!3g1}!l&N>p}mynwk@B%^n
z7W*p`YfnsDVtXinAnZlz{d4i35Alh6!uo=&9eBet;acAHYfGGUco4X3r+xx!LnuRm
zFggwB>F8bx76xD>90;tVUs2-D!7$1B$Ahre5dGOR>xl|(x>shlUOxj=zaMLV6DLx5
zu!EBxj%3)RKVk^=YWB}j!GWXUyAiB2$!R!2-#m#h3DhZrz)Ib+puH1Njg1gi_e|O&
z9P2x#TY6JP4+IsN$H~oia76R|G>$+r*aZeqm$x2hLe+qvWb-pY8cyR}W?85n5EN-}
z!pu>ea5Qw43~5CBATIhyRWz|CKz_g!&HUZW;(=!<Jm>Nn?Y)-<!$y}l7)8u!Da45l
zb^7G#=3QT3YRGWW%<`TP0;e-Bbe;OjupmMI1Op-5Z`dVQh`I~3;^4<e2uCgFHnPz>
zA95z8MIsqiHF@OWwDUu=FW-m*K|-%Xt#AdZmDwj~Ajm8>8-t@_{Q-Lo4i`B<H9chF
zM77YUzWlWr5b6Z+OwSz$?rN+T3!&r9)zVj=#kX(@`Ei8UlLc!6uLkW7IVhJ@Dbmq;
zzFz2e28I5?6+{Dux2;t*%al}S(Z7Y+I<C#coeis*LQJq#EvTRs-C-z)z!s14I>ziF
ze8lw9e1Zu#kl{|EwL%{3gZi*1%JyP|llI|;M3?|t(Qc)(q^xM#!p!M!0;oO=T4p<Q
zU!bW=k3u>=Gs|D|BIdN<2($oHFP+?2th)NB@5x<5fuK&#MR`B&<LIPQoDl&zaO1^r
zCp-z|aIAb{9F5%F=X5jJr+6q~FPyjhg>D8+-aX42tlfL<Y$J5V+VN`7QYYch-&q4b
z62$bIrsCIPWSUE={c}K2aOxQkDwi<>$XRUd@f$I6oW~5M6h0L-Xsxv*=3K0es(!P7
zdf;a%@#)SCW;Jdp+C>bdG}zq}MB>dUcylY@)<7)z$9SG8xs7k(#Jk<T9G8$fT14E!
zyE_;=>-izvo0TiHR%}O~mVF%41}UVguO}_FMMF**wHyt%UWDEWizr?Y0<?dp@fkc?
zclL)AZq(i}=3748F!-x%8?l~>0;X%EE$d*7dEo-8j|HcGt8S6EX?j;%N7Y2xt4PXj
zT``{|06>WW&dLtW_50LFdhK5n31$zj)NyHgjnYj>;n{aE<%;#C&Y+bQhKYy+XLoc(
z0%-p{0PbAQ;JD)WS7#=LAJ4a*fA1nC%)@4vY6%*fkc!8LAQhd5mA8l0FnzdtXh1)^
zKal!q^fU@WxL0&bTaT2zi&=c^duT%J7LBhy?z9+1BZn(a#8+~comypv#(&&Vq#<u)
z9?tO+L<P_+iR@2mDg?ypJd(I+$kRJxP^IF;O*=b2VkQ<F`qWMH_1M!oUqJ(m$@FVc
zp`3@ov*m%iENLS=i1FM<;Cxh0HltpdB>rFg_fMEtgNK+Q#zx}R0ki@zj$Niy?q0fk
zhVq#+qOK!ewvM_nPWXV#n$m%4O;r8@1&NO@<0mIOK=n5jM5R*XNIt%PAdNF({??=@
z@H|txDwVAxNj^?R<DTvB1F+E^t*YOO=u8J6J`4nj3$(MocTIXxhEAz){#y*%Q<Kme
zzg~0<jI|0zTawSS^L2>KE_W@-)Q;>{+AYZtsY#TG)Jc~`U(e?5&)9j+W$(b$DEUa!
zb9SF@M*o_qt<bQ4wtPaZN;EEyTqSo9MLe2;rl+JlvgD+L{9{2%D2%jW(}IL733oVy
z>DPYz9+kw1RG@LnQ8yIU9d<2*0ECXy{WFiC(Qnhu_Enx<Z)VK{<M<tl&ittY`1WEa
zhX02mo81#J>L#R;kNf>=Oz2vZu;)w;Zmfj0fyc8;U#<Ad?#di?+jBM#FVFrOh~GU9
zS|M{a!^PMjfTZX;rjg|>LFggnKYcl%>$!w83sjzdK~lZdonC1BHrJT3i##|-`Ov>d
zy`I9^aCEjNs(awCkF~f(KRP&rLnU~IR0E&8k_)hApr_vr3wATj|EUUU+V%)sP`-Q6
zf$h{pxtXATR5%(X$Doh&7AoZfVu>CzO65@zK)e;vQCsUpv&-euUn=NYk~==rr-pC$
z=(eaPD$xN!z5!!@{_IffB9%>&IrUKgg1y1Lepu_nF{gpJk7H<z@q|v1#zN)83@EjW
zP&Av3sR^088xlxHHlf31W1-otLmOzgSgbMwRqt{<Hq5R1m%3sc=E~bcWF}7}A2*GI
z?UpK;p*HC@`@N$I`o3kzls8ZrrQr@Iame7l?YcsB{OZ)o?j{*0Ng9%Ce1CWsVM-6R
zN%I9rsgsc!kCN@kKq;gRw=9?c_Wg^-ZhZGcie5^7*+jDSjVLHH8pvGaS7?Hn-HM8=
zE<OA4h9x&SZ8g}B7?7xYD48<o*YMRg(XqcKDwN{pT=Zf7jsTsclFIL87F4z+yVBVy
zWB7xbF%xxnzPkSls<V-pztUOd5U6T((mJTod68*MP3j**!xi%PXW~8+gZ!Zxo+z0a
zc(2VqAiwrs`VQKjt3gOLDU%=Wtw+&&IN3G7EZA%<kdhCO*l!aXWlPrf++c%y09<{q
zyot&b6V-}A(isFCZ8Gh&ab`esiK)Xz6CDf%X%=_-Uwi}4Q~JL>eQe4fH*-eQN%Rfz
zyXJ~Kmj>VOHat)Uf0?Uy{h#!iym`Ic)Usg2wF{6T5=}6WBrQl+k$f*M>Kvk&f75#n
zUH>?EDV_f-;%K_A$m0hwB6}^gN;^Qg(Cr03_aH6vIF~uFi#~MlYnI`$GVPAI1f)c>
z%Ez;o@L%D8PTE-|ZT<alF1|eJ9RSd#<Za$;;;1nURTC|C!-ABkmW|IoInz2%91g}p
z>>QWWEb82E*h@E%gt@5@OOB(<o#sxEPt3!r7L<JsFO2DhgRcHfSx8{)V<k881l;`(
za`-!SMoW~#pULcw9?eV6kVlddsc2`_J`*uJA8)_H|0i*jRnF33!s{*h1Vw<xk+e*k
zxeBGIeDAR^{-81LkG?-YpUPrGjMr&mvdQoIgsRFf#FOH_rRxm&OL0PsIWC56mhZpk
z8FbmXp8~oej+df7ra`BpTk3faAOVr24&KJNQYzJl*OD4XK!~OMd6Uc$v`FOvwGG=+
z581OEXA47%sE%fcQ1rf_x+uL;C{u*0mp1v$5m@flj^z31z3E{V&e?qN3rsM*EQif1
z`L4k2&p83}`-Y`@1&&ULS{qN*aTE|#^>j0&Okht$rhvW!qSjbN_>N?8H8>ZQb||+}
zscPV~PVh3{{;;fEs+>9!S-z1TOkQAV1d~En>&jgokTuku$}9Fzj<v(F@^=7>;GCJp
zpwot}9N3V-lUy}1fk_=j=fm1TNSz@;*j%Wzx1Co+3Ng-=iEwo%!x11aU1d_t#zh!%
z3$ZNHZ*m&EQ*k(v1E=~^@-aUEnbe8`6UbOQ$_>Ed1al*;jQ&Z7a?)YQ@;WJ|4NtKF
zaqsbCa)@zinT`6lZF#r0!Al%y#V^S!R`IjNk5OnQ0mL{yI~TTssn8L*6Qt~e4Fm~`
zynF0zEgjQDJ1ByMFkB@Q5L)8fDaJ%glL4v;kNn}m&DuGeUU`Rd_m~Jn#80s^&!j|{
zN!!*X7Q{Gwx3)=j@K4~YrXPmrzYE!J6XyPc_N}p!zyx%cvYQ%RIB6YwNO#_<p2*tq
zgbL3!AQlw2&0NF=&~6Q?-3U7DSFiORNB7|+SyOul*I}(pEvRkgZ)kR7jIO|+P9IWl
z7u!+*q-4f2NyI{)rL)9o#59UkvLC9g$!rFk3dUaa`W1nkDnC=1SYTF0`v^k|8;r_Z
z98Gzy7a1)>;gQG`wfqGMH@ns9;GMjp7XWSEFK44~b@#$Amqt}MStV1eGUIt<OK)u2
zwmxG+^|v#c!TO%VkJ-?gu@Sr+{sESfGgEuS|G_f-jfP{_(f%2I?{PL-NQp+-AKx}x
zrTDOm1K0HkWD0!%lES`HKV1pUW`UG^LK>YXb7GF{Fe_jJw8?ke@94BJ)u5=5XXR+s
zb$oXYEB?J5fHuDx101M6M<G?NM$J*BD_LJC9-{1v{R*de)VsstKvJ~FoFAdh1B>J|
zr&Gw7GZs0p8nH~w-&p6xDwtG^H8775?kt(?T9lKI&L;w39v0apQB3}Z6rfEB#i|*G
zr1~`%{q6+!Ky*?{MW0pjYUx9SUcS+~RzY9_mmGMalQN#(yoZ;mE&ot13kRW^E<r}J
z$GrSra?$H02{c)gfwul~bhP*w9bRI9cGC>VqoSoaZtDEs&${H$$7uC66)lQrg9)kE
zLTJo|GB*wZr^#Bdq}&iglbw<{oY{_4<p|Nrh@i>4MlCM`=2kJP0`^sxsQ}vLoUrWW
z&G<&cW&924NTjVP7md9Ki}}6+>JgnRn4*(v`lqcc(Ms8IQEk2)n8Jl*B(Sx-Iy39Q
z_c05ETRZQUi}coY@|gE{H~|5|a_Zq;@4|W@s`?@zU#xj%zWT$V+;g@^_ws`ebu-a_
zZj@C7p9Ot!A6=F~J=ZH@D$kY|ll1#N66q4xMI8FxL~8jtwC{b&BKr;XZBla=S}8dK
z$?w!{zvyaAfM~TV(i8A(b1exZ{RD{dyRE!v-e&vspclHtcDg&4;j6Nphy4_a@Vnz)
zp$|YHhLOmdAZI@umnwsiQGs>#km~#UXhF!6P$up`@p@>Sr5$O@K1F@>q4@L2qcv5J
zG3S}X+pbNCojhy_fOa$#V@p4>#MlF^6d$0C*KM5L_tApg_>UAmDKHNw@J+~M+@OTZ
zw?I-!R!I+u!rP;&_|jDam=u5+m?t*0SKrYH4YNcqrUhuv1TlHMfeBj1U(1srVn^tu
zSAON+(n_8$jA@~_T1sfI&k-zS%nU8e4bax`5$HRX9!Wsy?WHS@U?XBpI#?wv7%n6p
zs23wwa1jLlqgtjSdh34?bZCr+hy@Rwj*pP?X!Cvz1m3brlAADwkOk#P8*7S|x>7*6
z-*$hA@q1^FpNQ{GR)_<r?$`_QGVUU5%gNG4u?~kJht-)|EF@+6M9uG(3`ns7)eYew
zN<C~NPG94jbm#~!BGt$C*Uqo<{H=4I?}5M#Sum%Ck2{d$Y`47RFmRCv3z2%7)y#Ki
zu~6kxhJFPfpgPAppRf0NG9;&an_1BW2;wbLh~5hLMASSzbBIRGLE|4r1h{@reX}=4
z8=xlUdPKjr@ugHkO{hD$0v~Z)v6+)7+a9xjActaAlm#z<yhSPe6Zm$1w#uP6{t?)X
zQ7L1I%r=z?M}&z?kU`b9Dc0Zb1t{W8)leKiWx+Y$x&4Bre(O&;5-dgnaS+G6j_|?b
zCIfe|@I$!|=$6lUf4H3<-c=&EE!4)%MhIQ==;atfd@*pBS46S?lmkB^y{@lDiKzK`
z$|#^U%7ORCG9a@{3et@Gph+nL0`%!Mc)Os^T^B-lHm8i@unj}LIe4*>L0dF4YXslP
zgHWLxt;k-i_n?t|x+**wb`&QdQ8O(>XMWb<R*WR5HYEtioa|k&(jUAClUjUvhwqEt
zUBr&;xtoMi#TC)>PTTs98)3-N_Fgx)#<JmXfm#b~6wi7%a`crwe8aYCi6sWF&0+BF
zii|bLOE9O%pV^R7L&AG1?Z)SvHE8lMYR4oo&`KQnmd6r%NrMz0u>1{+^5H6m&ll8m
zM0Xc)Xz~jek&P^m@e9h@DUNlyF(WAoq^Xkw_cyhhs5EsSu<@OG&Z?mJzC|KGE)Y2!
z*m;c#U&pHAUx>HGplfUTk!H7umJwVkU~}(GaJN#oozeyeu~GJDvZ)9`Qoa4M@9ttO
z>liNrEO$AuL@c!PXS1&4TXS$GwuEwhzH~JZlI|lim9uyfmi}BB`Bg!s-g1XI<_WMX
z+;8OIhr*mK!Q~5!1T*m(<+MpDpdIV2%+kSrxlNYs)8icm+UEYO#kRd=Z*4_3z=N)x
z>(^Mho3VmCIm*?wG#^jbHLvi)FFd#iCqo+U`5pQ!RPEukv7mso?JIa&Bg&&KrjTy3
z4{mO;2p?9o^s}8$LDpQYetHd;%N_N3<%F^iWWtN-uDAEi=&Y)>8{rM$wZNLRsF%9I
zX}!dM(?ov^IfNO6`6KdWX1ar{Gb+mR@<0;WBRK6VtZ=IY%EctGxs`dKlhe4*w6}IC
z#T-l5rvDC036U4ACd-t<Pm#u{7ZY}~JKlfQ8}@exP-~D_Xk<IVt=Iy1<h2cxRy3EZ
zVkOnT;X^aQXN}8t6>SzNL+`?t^~>X=HO@gTn%wfv71ZMr$Ix6!Jw1b<AD3@nwS}vX
z2gz7u%snqJ3)W&1m|g{J|0+<xCu5ombzi=R1_u)Q9W1s=$b{Zp@S8Q=W~H}>9PAoC
z4FR1@cJzB~Y7b>RcUgz@_T8Kv&{Hx$;#g@HGn;3e-Og+loO<zRdeWv*Sj7AJ_Ao<b
z{<}=R!@TX>-G<UW!giK}5{WZPPZdT|`^Kgfeu2Jg-AOD+U;22%Ra?LDU;U@H)eb<=
zoJ>b@cwQGs%IF~UIsh*K@7l#Fbo$=pra94gr~d4G?(e;$@k0gwNSI-06lMj}$Ok*9
zpp%W!P5%3X2f+zZXQ#J6;9c?YRYeJA>SKPlX;@4kMluBzMn*sVSUL0TFKzE#-Z6a3
z;F`a1C-vS62x!4?IrGVIL?xfoj93`qHtjKOeWXa*Puo7Lia&gQOd-?-4gH@p92o2E
zZl=k|fERQza@M*jCOwYKY?ZwCfcG_!`(wayk(5OHRjt2k6&7u1^B4#gA?6XG-NBQ#
zJT3G`^VbE^Hf|wO%;M+bolT^5WZ|tu95$GpOxsvAp`>SC^ySxf4y_4^n&uYTl6I93
z?H~5YS0@JjJJ!qxol8=4|9Zc%ssFadw`Hq~;yKnye;IS8rN4+DgGW|7<|tfPoExiE
z(&H~mN7G4)o-&ciO#cwM3sSXu2q2Xs(V9FjJ_^WYIpA*R#>3@oE1nUkn|)imXkB90
z)&H&{|KObj39Yg+VYI|(?cJ<noKlxvct2^JdA5?zAU-~4Ysrkg*;doITlYv|Pf+&8
zlX>D2a%kxx^h-i!lM|nB)`76V%XynT`_6_ns%Gh=&m1U&oBg@-W>c5i(MF5u7b8|b
zytJDV+uQNSm8==<8AetcV4{xGy*9swC*!;HUf>{*&&>87#Cq<7hEA9FH1luzb#2rr
z+<eEjQ&wO8<d^45@Mc_1S^bP?>9HffQQ7h*6?)X|r@DfJ8+h(Zzb4jqjK77u&oX=Z
z$X-Ijn_+uSlaJ7kY5OJY#iDqh@Q74$?qSNRWD3(7Vr!3lh<QbD&ZiKrxP)_$S}4dn
zW-ee367|y-sQMs)I7iZLi~W5v?NENNXk>ZU({yn_d~U)gg{{7YT({#yy*#<YeXSrC
z6dSAHy-a7nHd$xv{QG`YfcCB_>iZsN^WL=dRQO=?k#pXRD~^-r9(~)S(*q}SkxUOq
zd1QXJQ<#t68s@dl{I?w~Ct)Q&ET=c~mlp*W=6S&x{z3BvE(1+j-;xW3v_vygZD&TI
z^RQSq-uH})uL|r|n`0khA4GN~RG&(Z!2V49{AJL60m4vUmW3lF){Znx&FJQQC~#3s
zxsARL8zn6=igU4-o!7fiYOk~-x7%vlNMvbCTv2U&Y-(9?W>jlyRAzB(V{E(!{$q*9
zW^1Hef4>{OOWQo)|II)8L(@gWS#w|GRmM!_T-Izub5l)2LtRrtV?+J618k@+IW;oA
zs8U}=1)tl#sUo`se|#i{mWS{(4zM#@>6a)Mtdf|MA@bA<rtK8MKjXJM6}d;v&}~AZ
zRC*)bWd;8F?*BCX)ujzdOt!>+c&}OM_V$je(J`Z_PoW$-LXq4|^+WsW8a+JTQvgR-
zks;{Yq#^d({}GbumJ&OP-v$rExnD?5yRBpGCAUh)BEq@C|F;w7FEk-?i);7XGe0<n
zFFsN=Il$Oc%KOWhm~^G2?h~a|5fvA~fz$<#$^07@?~>}RZ3dc$hnI94#TttqJC+!4
z7HR8UUTRrR=e7O8EMRB!G;&^FY^LHjPn>{}_fI}Og>fIK2hY#xSAS66zJl9U@~o*n
zg&T!5^O)w$!N{j4pB}@wg>Q4Z8ZCWIb<W=YBp4bUw)J!T`yK49cz&i9bQT+9X=T^A
zvsRWqICUR-_|)I5=!+MrcQdZxJgbh#HEn4&C^k{LEdTh%9QdS(%SnlTFp>QX9;)A{
z&s1FWxV*%8wMg6X(y<788m2#FsykqG_f*&>GD}8FJS8EaH1EIF2<~0M6Z8cE<iEQB
zu%l<+Ra0C9ntIe^R>=+H^d(*u1_a!)1YZs@{l6I1KlG}-D7QNL7Dz}Pfn<4owO9Xm
zn^``U^;U&ebCmgyROA!6OYNYdojloMG%bed)7a#HG_ohJm@WEC0s)5TWg6PDdWq-*
hvq1j;;tIiaXotk<b&%2-FWPquRV6LOD*3k&{|nMRX~+No

diff --git a/man/figures/op-softsign-plot-1.png b/man/figures/op-softsign-plot-1.png
index eac6f369030bbd4ecbf9bf44e9154b895c28c725..2822105546a2fc30dc7cff7358060152923b2727 100644
GIT binary patch
literal 20855
zcmeIahd<W;|2BLoNkbH6k5Vc!LiUQNBr2rLvRC%rQ7JQoWUELSA$uiR$x4Z2WoJkB
zxQ_GlxvuMXU-x}~ANTJMxX*`-&hvba*Xucs=kYwA=c6mi=P7nG>?RNh6c;a?yG9_8
zEEE5cZ^QqbR5U$-AG>TW+_WbU_8cJoBl*(fwN4<g5iXuPtL73n*<+-wW>!r$D^Rny
z=P=#dhj)GMJq&nw%kM~vK2^YMrw&(EiN$xGfE~N{+(<e?_Uhro1Dv-r4jc(~P#pcb
zc6W46$LP}cf%i5W(yqo~_D!dBD`VGwj!Mx;C8`hzKC>y|%(wv3J@S+UwxKHv1cL0o
z|Nqzj-a3RRJNfXcs;mE9o-@c*B@jFe)HuJ@);_4ZMgA*x>1~Sl!dTl>cVSagQ^US(
zgjsH#BAc^%dU~p=s#JYt{s$$`Ql+M*UW(>9CccC4zVOMDCr68{Lqy#Ogv0(RI|)>F
zC;sbK9lO&7nJ1=mTZY#jS?}Rw{Bhz`jFjesLr#~pP4-2kI@FAl5ZumG(Mk$8c6a=-
ziOruVlfV93F?_G?r8{PejXQUy?l`wr!xVcWFbhXMahF!|_{xW_ACXJHe)?~2syQ`h
zMZHmQ>&s3olF@A4MIg8d-)(7Ar8}~)IAYvv!aLuUd51JU<Ps@?;L9<)U-Gj1oI<8;
z$MY|fQ=c8b9Xv+A-cj;pQX{Nt={GGYYc$OQ@p2@qaELW-a*vnxqkr-rzIZqq8EnaU
z>AL!*z6RdgS!Q~v?_2f}S|=PEUNgSV$Z9gziB-`ZwC|c?N+9R3hD`=#lv@>z`zwTc
z3_V-1x&J%yRH^Io>}?yHSrgvW7cchG^PlS<ClKzD-;`Hrx7qbDA@jxZXdqW`P>`XK
zk+l0<w4RrlsVSAEAb~I-`(rvX<+T{Q+Sczqcowgba=&`>X7RgsYCbtQ#=g{g62d*E
z02YEw{VuQH59UWk)AJ<0XfDTZ2G)1VJ^Fqq{T3x||Lg&ewswE(ilqxR4VH|9G9R16
zDy3QFwTF3mYPS*I3%4DkHIV;0%9HSWf3J0Gf7LeES#APhy_iSSu{!Q|yhuzy)K0>f
zn@)aNxU4~I$(R~}FzbH9ar^)2d>6X6kq~4=Y($uNqGMxYyT5$7cH_p4D_5>)X~jlF
zRO{y~{QVn!>gJInM`W#JX(S2n6MxOGWM*d8)YKRm8R7d(UuAYi#^>hdtT%7CPM$nW
zo|2n;O-U)EqC!_z*0*ukPJ~Ha{hP0^Z$rfK=%}dQ6B8|QqVvu1;(`n@jaOm*t#52a
z;!;vlo;`aO7Z;b2k&&C5J2^S|`t|GIzke4L6m)c4X9+!@*1qB)bze3tENo?Vz%M6f
zE>_qYFXz!Id1pJQU~N6~?TtoXxx4dFec1i`_x1Jl$;il5Q<WMT8&%cRn9i7g_1j6A
zdO23%m;1T4?LM<_aQO7}iz+HDpFf{EdGg}r%Nw%;ayM_%$17*2r-wa$d@HX@f+gF!
zuL4Ko;NXy#mxl*g_m;J_w{vlE8RXp1=80|`{$eV4=1l&lPoL-ojL)4vpRHZYf8vDc
z+Tv6~LW29!^we_ok9-T<FZH2ALMF{|b)l>i<KspqCJ}3kR~HCuM_nXzy}Z1xU%&3;
zv{>pq^XH$3rk0k4g$2&2rKOdw_WIc~O)IOc%*;Fb`YENQrT+ePQQTTs&)Q%{>=+j(
zr}?d0k7HtrZ3aHn*I#gVUwixZZMtfjl*e*wef^2!$IokOKC8Ml>mqUT<VnuRPq+9^
z&?we9^KfxdQ&YRoj|7K@UsF?i5)hzhZvNZ3&y$vxR@i<-$xfOcyS>1NMR@(1^URqu
z$BrEn7Z(>5Eyh+A6hu`IP?4TZzH;u|xf`br#Vc#5s&eu23JVG<UA}xqNQj4rCn_q6
zh2QXkvGK_7Z*Tm3eVt}{J%4{qXZ2pU9c_+hVPTo*$P{<{{q?o_o6ny=TlJK@Oi%CX
z=-^;yZ|~@+y~T$OIwvo`XZP+g2e)h24g}Vec&<71Rd{-OcmxFphq6eTTUe|$@>YI+
zcBT=J4<C;3@>>7;_z0IeSQmPHiCV60yUbF}u${e~9hNmbB7*wBf!%wE5Bj|da&p`K
zs;*qS#>vBzAaqAlLE%Asd=E@yF9pT?pFbZzex%#0U~exF(eNZXIy*D-TU*;nUS3vF
zJ44uXdb*jp`RnlT1F=oUe0178g(W4LYHEU_qBqsm&qzpUYiX^mugpI`A1dYg=S5`X
zw-^BvCMKr!jg7R+u})i?o)i?p>+2q5er=~cw>CH2ec7ge7PIs4cujP0A3PX|;0#Z>
zq^S7p>C+%4Q9Im{p?3i3{g#<BZ=0q3+tm-geEGtqk@c9Flu#iFr+xZ#dG+ta*RNk`
zd3C?SS*~AKR8%}bdR0$P+UV1h+1c4duhsFfw&w`{!2tmbe7AgFy)so+ZoapR+Skud
zBUAG)R<WqGlvT>Z<=boZ-rnB%(H6>GyH1}u<GHzB{Q9-Ik<qb8#h2I8ErnuRMgs!_
zqa!2f>FBUXSo?`zHGx7xLWd6@e(>M{{3GnilOY?C%a<?XDv^$M@7~Q!y@2&+U|_&y
zZLI!97E`}^^(wZMn)>;b&2PqKR#sNi)6;POzOlS@QiTPA%)Wj8^RSwG_wM1I%3SB+
zda23DcWd->Ze+l&at!LCA|tsq-;MQ_yTi{ZC@2t%8l$*N;RA>)DhU$B)_qdW&P%X_
zY`scdJ-w|z!;Qaw{Zit{!)7Tea!N|-V6();+**^)BBtQ@r%#{89aL^C#ugR1T)s>Z
z+caWV>>;%^-BWt%5ZBS8FH{nxL`6mIM?PP@a>d5ZZgskJ#?a7^jFLs^dD~MZj;_Rx
zO5c>7K0NTIg@y6<&d$!Ep`rEl^$?4vYMi*`Fxlk241%w+vO*dR?TWR<;&rvPGoC-E
zW@WWDHjWAp*S~#xuD@!#nVA_&==aXf3u*1D*RJ6VR=s8EnVC}q1E<ZtyoeWf&bpDK
z4?nmRbxPEEI_~b>vznR=epR>l=!C6%pM-_E{{5|vr@wvu`tIGkuiw7GAJ+a(q@|=B
z;p9|YG-1E-w1k9kcdY29AnsU5XmPwf4g2)>?_V`e$JzcWdV2bhkPsXyPSETNLeiMS
zYSLN1@bK_dl|+_9hdONi!(P5Lvb4+y2~ma@#Kmdr=rBhr4%-#)C3~Kgr7S1M&dq&U
zOH2Il;j>73FJ9bh8FgOgOnUL+C=U;!SSI{BElt>A><Sjs(a|wiI<)d`IwN6>o`I2(
zXo54n<v2SM=S4+D{+Q1&oii34F0QVgTN_Ks$;q%FMnv;GlNPV#ft`rolJ1L^xZl6O
zzvblQ;Ms{HYIRxJUqC}hANa5K)XSxwYm2i3)rr#H&4&D1nwp)KPIW&PVGg^fIj|)c
zG&D3+R8(M;2m_NXN#e3?+uv7S)6{fBh(ek<L__l=DoXLf1yc(P6GKBL371*8)1gC$
zVC6W0R-sk7%N#=U8mw{jbBz5=ucTOfvwn`(+LSOicP6|ErbBdfMj^}2e2eJ(d`B0T
zqI>s9BNd5n%+0kkGJ1*d^Z2mb!otD>O4i<-p!mtcUW24DlK#fom38E~*49==VJn)0
z2OX6%80qQr-@QAf`R)V{Pe|jiCN9+7y}Pp$38>}iX)J%H!C6CXZM^FzHWxd)G}C8o
zWaK#gQ&di_`USS+9_jx5VH;~p7G`Go1qBE{?eE{q$jQm+=xpfmn38<P{Yzg{S3k(a
z<bf4MN={AXz&q}4-&s&t7#I-n$3?=IZP}yZ-8%;ZgXf44uCB$tzOuJ(C&PDCRjFYQ
z^D7c8p$&!|9UWfFv&q=^!IU=v0iXJ-{IF_u!Hj8HS-lpcx_sv;2`_e0Qc~{T?e=Hr
zBEoV+!(<WV32yE?^^XtZZIO|YgSElf!GgRzdl{McFD^z{^;KMc-iGAR{^g5t)3d)5
z*_90R^fh&L&OhHOCSQ3#!DxKvj-=PR3lbc1UT&XWwhtV$<=~-1y+yWzpCY-K3q7x$
zc!Ic$9s3UCl9ZHGQsPz<aBy>T6B{n&w(trdq^!(CQSsBqN7NhZ>jjowxp%sSBqhs#
z{P+=ZTm@E~+cgPqd=?vvReTf}7_l+p#*UP^uA!@|iwlJTVx{0dhy{6hd4<+}EcEm@
zuU*Tv9jtx-{=MTw2eY_SMtQl;*|P(<qzI17VmAFg&!0aJ4Gqo6I2GO4lV_q*m$`)}
z7$n^U18a_~-608JN&3!p>J%^;#ooP$vvziNuCA`QYCsNyye?u0v>Oh<K@t=7939Wx
zm!13jml^h2Y(I)+SwvbIZAqMWDSQ0*@k!NGgs7)6F@LAJqbG$)6Ok_$t|64-71G|D
z9!^dLPoH)IWP~z_!d1ebJ^PIi<h8zBR3yR@T6c>N$pe@mQNs1ZfDK&jOM83wk01EX
z&CNYEHFaF+sn2b9VU!nRIz`&kFgB!YO>J$Z<SU$9T%MkuOVd3tWHWtzJNV14UAxT8
z#+R2}I3gNu@dX41A|wG=_Sv?lhlDhDWNMkTBwS8=F6p*+J$u_2H;xd>EPjZJitXe{
zQT>k(0T*#+$B!Lzb#p_O!RAzWu3f)$DX{7mveaZ}jsa|ld2O!3b>4V>q^Y^N8Lu*|
z+_lVr!e?L!K?8ds;rizsT&XQZ9{$36>J*cR%`Id;|LPbK+h5&<)&LGjmicaf&GDO1
zY&mxDVg!e%xVTE3$bJC<N#+0+8B;)6<mNarM^1KjIa%2c4GkxyrKLGI6y)WNa2+iP
zk_hT>t#V{4@6Gu{>f_3=Ecvr%k=$0+)?#B~9zA^ccw=9Gw@ll1^5h2($N+8k(DJBh
zXaEv~ut>VS_aRwXSpmSsEnT{J(e=;JqeqV#h8v^TU}zm3PVn+emm1aITn}azkD8pE
zok|`fxu<yN&eS97<5@bT9i5%8Qd5Dij6Xj`68?fev$OSX+;ExuRWnfSUznedqa$Yk
z;v`6Ux&g7_RE8!y&z{BgJgTj&ZBJ9tD{)9qPCl!lvD8!Qytuf?OsRd0JjGy7LsZlu
zSJ#z~AJ6Uc&jX^1<kHx^At3BSycH6mSkLh<#}Kf8|29`uZO2{%sxB`tr>P`f*U*4X
z?q^qkPoF$-;$1<(r^d#u)ro9lV`Ia#2U&hFqOJo64!{mH)YJ@(jSn$0PW1P`zr~k@
z<gTOBG+;yYPlxd@aGYJc9>KJUi;9#xS>O5%C5EKBqvKxy6&Q9_c6MT7VkNF_b#+z1
zq}itUXFxyzR#+)c#1`<!+}u1CfbgB^{w~+Mf-dLw$h1@s+c{7Dc$t~0qo#H)`ATz~
zsDNo}5`16r;>Cjp4{qQ>@dH8Oy1Kfg^R%{|9lvHpG|`>-kRAXkG_th!)6#|p2BIF<
zDzJ>n%CbU9lq}~DPz1a@JUrYIFAgiTU}R*>o!_=XykZ3li;2alZjFo^RfBWaR8>(9
zFf%co;^szaAZ*#$+S@B-_T?x;uoH6j(2${;gvULs5<Mp;=V!Q1Z~0m^rL+f1*zSTm
zfM>v5%F4=u3{ADQsfcvLpCXax_wC!~QqQ^0f}pZf1R)RE_0_AfwWS$Ub@6d=$oV5J
zi7ILBqfN1J*3!AbI%Ge~&bK77in1mr+<dI2s{JXka)?Hej~|oo-i>JV191)u?JAKF
z5uxhlR^sdX0javItjul7Ab-FbwpfylV1XUN#Z}eSorfu)eCp{bFD`cNDRC5J2(Hne
z>dHGVBy=4W=;VvmOyYaeiQtv)v~P|6vE#@6@86G?@+_&S=&i1<uBy7rMlO}e%D_<K
zv0|5NAWP{pF5xphHrbVjYKxWMus(!Yyhu9>u=W1^nlSeB5)u-?6a_E<<0b}%U<D_o
zF?uX`PjmROODGQz{qUF>8w04e5Ed>3tZW;1pr?;P`VC?@6EGRn8gc74SrTJ<cDA#v
z?W3TejLb}6<BEz3tUa7#bj}4fWKbIfAa+obrGf^_TiS1j2;7WS>I9?$ybooS-b5+{
zbl6X7JN)S&PShC9Yj9V>V~0`?!C?<f0;m|(XXdL{_4V~Gv;B9G7Eqrf%j)R#Aafxp
z*Ef?(9t2z-8k}>%UXF~6OnlE)QBr~d#7lcu!0=G2AtCGQAJJslqzD=#xo}-ar`%)Z
z4`2omFZKa33=4`90<lT%>eV?^t4Mr?CifC3*xd->#YII#x2W>nfxy+?*4Eb10e1(A
z8=iA1wCa&<tgb|WX9HG7B6XIPl?7qa6f10DZf^Cn2*z~d`gI}OUmsBk6BA_o{(DGZ
zK7z=wBXcgQY04Col)?f60)m3t1(rAA>%b4u(X@v;-xmT%vz<ph3MVra1i;Y*z^JVB
z28Oz<q(nrbG&CRebL#hzf7niIKytwb_GwTTa*~S+0aX6{d6j2^W*;B!A0Hogp6R6^
zCub99CTAqxzs*={av&{l?u`t1UvH(i-u3GpO-*<3S`;uac*C^qxBd1@%0y!eH7h(;
z5cZ>a^?Exxh=>BA7N&ZqyFjPHBWBWpEshT!cMYYBlao_PNy**2cS%S{a1dY~EIeGl
z&vS{nYg|TzXu$x+)o>RC@e!<$`{Lw3C14J=7zaE}w>P)_ZYyx7Ll`Osc!y1~Hfq6F
zuU;W49zAv}GAhao_!rK06t^NkBPm1Mm1E!!n<y{$1OmXi!}QJvGkmM`_C|WjeSY^b
z0+7Rf>?R^DD=RCk21f?=fN3#Hd33h4gie~Wt;6!SMMp#kSah7)vEx2ij51(lT&JzA
zE#f0BkM@<TR}bTq(lnAj@0rCM>btwU+uQ8{xUqkrOsY|)q10&o(|QJn^8MD^oACHC
z6+L}dZ?D(L=NK4d^phtK0|FAn90MjzTebQhoFtIE&dx5($q9>!O1*NTs;UY_*4K1Z
z<Ylc257EU_emvL#9|BT2($yj?@_V+PFh4(10t`pD(rW{hOG5}*IXUvJ50WWLc7!+f
zBS$(d3$S1%C95scTS`tmV>*0}+u2UQF#*3FZEV=c^$`rWR$9Ds^YZ2vG{)TMkeALW
z5g(rsq+BF=L9BRfu8g8!J}D{L^Z9cSE_nhdob(r}N)!W(e7D|bY8CdDxmMre;};M>
zhJU5D{@gE%V>VG#_x2O()_Df0vsC9SdV9)4q7;5q2OSris)+n{YRrE7<54UtY78J!
zSdsMR^5B=3mzq9*err$vShnAO;pb%RPUU@kg~CyHLc67!&Uk&9mNaoZ#yMDCpmOum
z<uVbi{g*z=EN5uuzkc^_p|H=(eQl9DqCo<IJ2bQf;ibNLN4eT}ouJ{P{_^!%l%f@u
z<iTtYPuG4HWZv@sn@4-W`Q|nsp8YH=AZNUgAprS#d9wlSP)@UmJ1NP@`QdgiIrcal
zmRHbK)>c~Z-tNK~adFUgMaRj_CoBBPldn1Nj0Zm4hjG9w3UYIwiP#?LDo$D=GL@!f
z-~H1<`0{!WwLe|YbH8vhv_Dc>Jl{jL^VY=NwQbUpG8z$2o)A+K9(A{ooWUy<hb2W*
zFMp*lh%-_C7V_aI@xe4-cIYI}S({!dmydSkzh}PCPEg9!ySTX_pp6U-NqDY0plF8|
zFD@*A8{ohNXiNl%x;=ZEQWAo%N5fmXIm_d<E<CYKgZpoolgYw?hEWM2ej&4+4`C{H
zU$O#7zjEa<s^YG;Bn|e7<{MV4$!+Yc#l}+F$*ad3tyrWy=s?<Z1<uXQEpyYAWFA*Z
z_#7kP&qfY#1Qdg!lu62?NUze%aMZUqBqJrwtcdhn%k|a3nn(o&1^8WJd1q_ud*i$v
zj1>JA4Gj%2QcsDD*RR1uv7e@t+4*I$WWV}1`nhO1`<pj!va>&L(N(*CorjzI;H^_W
z+24T8<KiZVhe7!d0%kRS`ZV2{Q{CPkF{w?^$t%mvJkdN;H@v7?7tGJkM|b*%{)2gr
zP%SgZ8(O<${i+8(CrE060TL1_0CV{H>1h=8$n2~_mULZhgpyn5TAKb8<V*ZCo)WVT
zi!2F0>i7Mc*0YU2Rd>xu+PqHJaeVsKe1mEma5?xDq^@J+M+5~yadpA}KYqN2Yangg
zA>VSb!luA!xLjl@xl%eZKJU%&>#*LXiQ~lgl3}fpVOS~82nqsW)dy~$-E~)`=gR+`
z>92`oV}c8_qDq<LlaTNLfqQg>WLffV=$P4$Kw<OXJeEjfYm=-uStlC$3p*l>D+Bgc
zXjHN-7rjVQp=5JGtuZh#(ACui0ts0Ve}eI;upK1pDkf}Rnw2v*uiSAr`^eG_#puc)
z@(MZ|soCpF3bCmXda?^D&ySIMIT5Pt9UN%(?|&2!Ackxe%m^A&@I!TV&Bu@MsaZXp
z+_G@aHx_xC-&47iceK{<^YeeKseuP5Q{BtY|BE0fjC5z&#h+_WvM#L@zI~kwXbqRc
zN{P_u<=sLf<BS4yS;N86(ZIl9Wqqsn(4V0O>N|Jt*s+s%Ja{K=KynWh%NImJlcr~|
ztPFT1c59ITl!GvWlt(!jQX~byW0b?l?YiJsz~-Q8z(yp#f5_6$HMimc5P|Z}S%ST_
zrDX;M0l0I+(QV62fCzdgj~_QdLCu;-T58x34$K|SxT~KMiK*hiKjaN7rL=D7jxj3)
zaXnPw?83rC>SAyf)h{)N5}WXW_pBe7B;RYKTseVi6GR5825W0;;WKA?3aweYk_nrH
z0G56Wq%Bs~#J6wx4h81h4S(v&Gl_^N?^hw7rxxt{lgP+yU19}Yf~5EC83WFeD@U?^
z8ia#K4v@u@Cp6B^&M>+iJ9eP%0VIlG-u>^}Z>XzBf<LH}6JxHVmn(dL_0&h5hujB(
zgV`Ts5>mmC!nWl-bdLxg!KgP8^?Lw!U%otc<j9A<zQoCM1n=GQl(T0}{Zl|EO1jq?
z@)ufmeE^V9CY?<q-uh%sAT5vp&2FQ`$u79DVI9db<J0Fmm-&(1ckW|)o0AhG<++OL
zGE2Ky*6D~(<=y>~@1u#VAoyO8;pDzWZhwD+S|vhfrxQMvzIdJ%#8JfK$780hzHE36
zj0cr+{pB6A--+k+X)qK*Oc54#W^!7OX)p|r-QAyk`!`vCqa^@wN5)Oq!@&Fs;?Mf#
z+Drp3RhalT<Jus4IsbhqLl2#$LNtTD86MxZ%m`r4D`eCE9w;Lsep`PcK!wL$<Z|=|
zgkr&pEcaPEZ*Pqs&_DHFIx;+*|IUwhKr48`jy;EdR(cbcX^R{tv-NeU&)XZc8gBLg
z=|F!!ALvK)6~eJ2Gq=xC&Z>Zfa(E|kmI}s-LIzv_aruwq+O}vCTVNbpE;qn%zIt^*
zpT)<U94oI<4470@G&DRM869oiU4ZNtRCVj|KL-q9=}$!caP8{VmGQJhn26<_J19KP
zV~xE@@!lOAmjUX&97XrjgB}=}KrRZh#fOM;)+Jd&iG3e{o12=NIDwj)+GCjAo?AHR
z470RXS#hzbnAkJLmmuUnc6HGg*2q!<VJ^D^9ilz=;#xZC*&t@|>lZJkqj&O8NZ+=6
z9_V3q2@RU_jm}im)N9L=`+cToR01k8IoAqJuI}F-I;c0QHh9{swR?PG!r4+rh3XvQ
zI5<&M!Np;O&El3+E6rZ%3Y*_ATE2B|GiPl&`+N8Lr?Ss+{Ib>hRk`_Zn44CAJu++M
z0#(%V?c2AO78Om+Rah?Eb8~~hniRYCVr-*lMEY^e(ClEmWC>@=)|SC%3m#W$*NqP&
zD;ky|2{%L@cMKmUxxj&(j;anufGoMWu@=F+v%k^f+Cn>}e^I@2g~X+Us%M@ScibBJ
z^ZL`FApNAJa72>GNc4l>)2_WJIygLjj&}$1^#9TVxKF0XfB(X`!m2l<_A&9MBj4+I
zwsSNzFHh<uP!R~5k;?22Zd(<>X_>*PGrJqj2Gl;f7Y7U36dg+zpqk)AM>bV?<Fb08
z<RM9~xs~7$jy~?edBNy7*C&?3H@3065`0*GQ;z&BR&U96jraV>GFWEWoAhPv>|vEX
znx#^7GR|}0^fb|jym3QBTpS%K;_U=FGPD(*&tfV5RbJf^mTj5;k$n31(Y_gtV1Y*z
zZaLn2S#k*FDz|R+BM1Tw!qIfgTn5I*=;-M$rWtTjyM-%DXD{4lWGU-lSnexkc(z_|
zR52$os3&;4m|{T@4o<Ao_wPSU4n}D|90EuU$&VO9w<Rtt9c!75yZ-I$cInu#v^PO4
zFE8GyCzrGxrbxTgPVyK;CF)jB2{dmkEiI3prhKolY<N!ktaOj(>LzE?+@abZe(%UA
z)S?G9M$oD-O65?fp?SsK>E6mvnIk)+4@!fMZftT=iRzqz!5~5j>QhC858nBUwo9k`
z=&15f@p26Po^scvmE6tAcrsjZ`-2d{#N45uJb9ZbhWAO?nhNf8gNg?t1>qPG9-iyH
z<%K|s@ahqR-3lnGbs4=g7+)9`MwV2HyDWB@(+B;Is0wbEr&?BJp-wt~z2L#KRHx0l
zZiACb7e#l4@$A{r+Eh<$uQd{536~?!YE-mz%hMcz3U$qZmAPl9EFQF!+PCL89wW$5
zesyK`{8_A;$wkfSw3qfcJ0(%45Fb@%&&aU+%uUnO6>@_%vvV!a-kCSV%tR;ZTjhSa
zx_@iNJN2#MlNYD$w-Z{a9TQ3(EiPSdHofe$w{K&MH`jRc*_4G}DmvMB#STig*A~A0
z5*>7c&~w?;bPR>GD3da3k)^3_;al$sxzr>c!O}N$8wPE+yl)>nd*y^0=fP4Mx$JrH
z^#F{(?W(e}M}T75+UBbKcA`4ArgBSArZwoEKE=v*yf(@E-1GLa6ua$B8as{hBf_YB
zm{0<N8!@f(eCy`s29C#TagrYCu0kO#g_d5}x*%%P+rvPhapKu;uH1y>>}SS#$A^BH
zH|OL_&A1vGlCsHpdX@vP*U5oGC@B#;Fi-HvSC(%HDP<D-xKZ>bCMx{N(-*&F+H1~7
zU)o?$GaDA0s}2;l<mcb^xa!N7Kj2cBy^Zs_i1aa7V8oRQ<+PNv*Nc&E^oOo4R4ST8
zpL=lZ(TPvO<-AFpoVPXH#?D;NxDgf?CxM(q!6*b5%Fr&(MKUz>-?q#U&nl5pIiB%m
za9cTv_`CdXH!_cn&uH!y+Btc0?e?|maK>7fdYT0WFh$?L7lO7aD_ehmcYD#^J(3!u
z>cTW;5ly%0FWH5w^6!?OHGCIze&U3{Bd5XW_L$C&y>84z(>5|fh+gMPr18|U7r5Ur
zRG4qKtQ2{eUF`uH##C`Iw)kTYx?e*0@j@k_H$So9a}E37KyB*+uT}Moo3w#EohiQ&
zK$27fHm^Ani>a5AC$AC+GE|$U^p{mfHs<%=)b@?=cK5GV5-*h7FTXV%(|<?a)=QeF
zFj36vN!Z=0hi!)h;)Gp)D!v)q+UO%C_;|ljZJTf85!K47c-B>)S(ZB|Eo>I0I>efx
z?V3FF#yawsv1@mTQ-xZu^M{nC(omOmJDbsG#q2k-_bCZ<&%Pzofnv!t{(8?_JxMns
zrxQlOXD9OgRQ`V7EO3hX={#3b*k=><%IjMSqw=5odHGy3HN})08|jmAI@}t;3l)0O
zws#j2H`NJ*Wiqkgv6El;GXlano#Z>eb>3Cwcy+~P*mZ}*h=$(7mwe~Orka`(TQ+9@
zRlzX6g*GL#{b9oGKZ*IDPbzpi<tu&`@phAmDx4ZBKNC`T-gAn<?b+h5V9mjd^Dn!)
zV<uj6De^C8I=<}6KH@j?jh9@CXN7lgJGs(X;-jiHS#Iuz&Dc0xDf`(SmtVtLnv%eK
z@^RM;ayDw&_X}m-S9$2!qg(UUYtJ^l@t?&W+(%CSay4R){S>bL?ThoDOHVC+(C!gR
z7KaK1@GLtki$%;~OhtM_5atHtN=)Vxp4S$J(`{!IS!Z;%dgq?xk>w|y@M9zQHND(f
zDsiW)bdTQFy_5e>9Zu33g3>P^lI_h2lHIl<U{b+kFf%jTsgii)!yFu2s}iXL+5=*n
zr+FVQi+H8A8pU_ScnnTa{vXnv^^{pa^8Wt*7TuuGPz`nUje@HLDQZy0)6<)CHlT0O
zA3obA*Yfvbn@X3r_3qKg#PeNWo36k1Wj1dY)ZUh2!gX?U#o;mf!WkKpCcHv|f^WMz
zx#&qRDJy5aeyymgnm70sKT*Q|yg=BhJubf`@hAVm{qTarm$YX?_tE^Ps*d1m<K)!)
z_ICCCJ+#Cq9T&$$L*uYCtwW5`<+#JiFMk!+6r{VgM3%<3{p~{q<Qc@l_}b$yUFCn4
zWYy-GCNq9%YXjJB2{_0FSp={I+UCn^=P#2ih$PZ%)Jl@MXPq<F5F|^?j9B;F85q<f
zG++^!5IcOw!nLUBhhy&%F2`dR$+BM-2(AtLZExR}vY{Vx`HB2@IblJ!g)z{SVF>@-
z%P+?A<3n_AK4u!zxm|4c>w9RQ^nDaBg!*qQyHwrwXi%TCxT*LnFzC<q)Rc~gSKU0%
zReE)9Y&q9@*>V1)-$Q7X7n8mnptrX0o@q@k{fk@4_M@7C;eBcr-mFZyJ!&HppIDrs
zZF&rUweL&NoJq48bF7Hq2xKEq;pS~M`tqQ?U`~4PbI)(|HYTSCgmjqkaq;IK8GfGU
z8>vaFzCIlknf6bz+&32~ud+TD54Bb68L`XZ?fLK4K9XsFHr<@Dy)<I9WomRMv)Nt(
z&r&=5w*Jwoe*8N3Tj6PMm)W;o`c3z(1%3oO>uQ}@c;i2h!1KC!yJNtUhLO^nfdaZ^
zpE4{BIL+xp=u&2bh0#g($(ldj5=7y4@l1@w9T9L6OP4e@CroO$Eij>%R=NJCG1sV3
zIZmYQ=g;*eZ<3^L>U9AF61R}rS}^<;Xak_!{8V3m<K|6tVu{$g@4zzg1&@)hcy8tz
zf_o)ByO%|xC^`ACh)8Ec!y}Y@TLJGA>G!ny4ULR!diA+7GsjO(S`t$Wb$r0Tu_7^e
zb+6{C-4`-{lAFB+gNq%XotJ+TuZm4fObp3j!(|y2sE&@#zwgrMY4y)4ia5RIxogI|
zC8&HQHsQA3Y2QAz>&>dy!#-L?u$3L+(MB25-P(G~)buGH;o*P&F|aG%MQ`ZT)h^1Z
z%Eu{E(R5Wk-@6vDpY-fo>V2$1mNP#b2s&uP-MfE({P%B=&LK>q{6~-8Q><xrZyg+X
zpy8u(CH*V->~I*E)(6ua5XGTd{9pRG|I4o%bFX8gJ}l49E-wT)T~z1Z9++-$=HtnG
z^`xPLq=(x4X3edwiG51|aGd7Nn>XlFc0D3g1UIw0=~m^}td`_0_FX)g$>~My<dRTP
zF(0{vCiaO}u2h<qLhf)!OQG&>-_#++LeC3ZM@*MOV%FwS{lx-iAM)R|M+~1t0$ZmH
z?FiPgJ>s><n;MiAy0cYYR!$Dh&bM~MioR?h*3t3JuG}o~+HeCM;ITbkVPP`G*R8t2
z_|Fz&x$01@5=+&a^kHXPJ|CXQ_;^VWVxj(E2BAYE!bah*>0>uJuCotP&J;y{dw-nV
zx9=%)$;5?ELo0O6el}?EDPOp-?DF*S<Jq-4r%2r1DXR*{WfZP7S${C5kUGr4)o7lX
zZZ^XIB2vduZkO*Z&G?I=;&M;!Jtb@dg*`PoTIf8ZD<F^uP8DsGo^to-$t#4uf56NH
z%clMYtIF3a=Z4$<u*JPnGrs`IgQ=;KQczS>2g(xqz3U8nWK-t<C`QG-W@yul3SU0*
zyeu;P)fxRKx~x|-3P}dezq|%k0}sehXM-Ji+`OjRx7$xrzEqqT-}l&*ThU83bHmf7
z-bt1Ii6lWr1$SF)t&+K<DzYHfcQHbqinJ|?c8uSw+N^t-^+8*S6blrMTFAUMxdNZ+
z#5$dc^`=s=XviREpOJlu2pu3MBWynFWzTV;etm^`?@%p%-nH{w78V1$j<J=2(}gY&
ztqO<>8dg``atDc!G$299;oBqiO!we53;I~CXN(qqS?5&Scekj^ownXC8_+V!d*dw|
z3UcHM)TCg#p;|d2PV>I-t80>S7=M?Cu!U`HphisO9VL10>iu#w=Pq0zYdwR8DJY2_
zb5yhFGkt1kh!?TdN8^N8K^SiuDkz-0J^Qr5uT*F^;Uvgo*qAg6gbQf<Mm*iwA6xCU
z(ArHw&p!kXf|Ziks?5y1jMCY9-}f8Ig^U8mpTS{zua2hy_Ex$tHMg{&NgYJoz5^)N
zKfb&ajYix|PiftN4b++t&=9EtZdB?rS5DAadJ$yzqNVeX;Fk@Zs!mIxu9F1P9jd~u
zZ@|<QmzF{jd0kBnlCwWZpEMGT^!N5DEVQ~KncV5C*Z=|kJxeDrBqS;#Vt#RvgSZ<0
zDgM2Xe471uX9iA;+5e43c7^i{3qUF<i9BU_^uvclIbxlhP;79T;@vB<ZE`bE4z1(x
zWo7U^$O6FdqaAV{m5wzftwHh<G-2~|b2kgEbf5%@Zj2YUMtAbwfdCe_d^BuO+9FKI
zyhUaHkBGJP5_p8tq9S12AT-p_@&jK(#YQ8!pdi80QdM<({L4#p+CqYYlvGseSQgkm
zURh{m1VTk~8?E$bY~*Bqw#%~v|ETG9%UmHo3-Kbf0bUZ|1FLj}|9!{!*x1Lqy6<SN
zGm-7w=^Qjoa_>NthOG5RLw<-O_&rw_pvqHMQ+s>k)aN!oD!Dc}UL$jJ25RaEP`zAS
zZ$N=KF%d{Ue5F_rW7;<gnhzp9RH-e^%?gcvt<?`%+~U!&r(hOafD{3Cfe!TRH*d7w
z-XcM%K}yc<Hl_oH1zNg}Y&~8Mj#R*s(b3U~iAzo=2vx*o1#QxX5+T$dI}hE>M+t%s
z0Xq=*;DG`<S@!n!bpvQ+ojy+~OKHE*T8faXQ|bh^qN}s>>-X>H<>mirX<Z@s-rp~|
zzym4=i-Zf*xO(-pfB<AHzxyhsiI)QHe))xS=k7g!&VFAuxqTeG6Z-N*bMxOuC8E?L
z!<!4G#>0o?C_R1IjGN>Brz#Zyha5a)s}=-`oF`8@65SZ2umlU`C95Q1nKn7=!`$3D
zH*YqK&WrEsg3hoh_1c^Y;U25kr{uz<D<?o@Vw2F=Y(lDhXBrk8%ZSS<*)}qktkFN`
zf*udR8u$zlz{qcC7K8PX00CUG4W@G+G#LoC`wt#mOaF>CA$BThIC%+Y&89*n%*=c_
z`3kvjXNlvR`}cQ5H@<8$g~yKV-j?`$^q|5DnPm0=HR;)RBRxG6re-fBKf9!~Pix(5
z@|6-MoTVZ~cOC*&a72&8!?DyyP9KobIN;J|pJ&Y?La$T9{xGU&s$F}gp}E_>V{|lf
zb-X+66h->bp49~c>DiapKonzl#P@;T#|}fA1zNDxsKn~`g9xTWhfAe~mU9ma`m)v1
z3A8(`vv#!Tvh^74N-Eq9y^NKW5WJ3$kMHzpbXOszr6Def^7n(1|Fmz^7Q8DZRb!MG
zLId|pl50Ydgz6PM_1Uy`@AbJl=tcc0S#xikLY8u1uF8I%*5;B;slt5QxSTun^oq&S
z<7+HcT;5LRE2(d*ItJ$&1;~E@wxS?}B=P9cqh2e+v`|1B=QZrx(fX29D#avd{&V_|
zt54CSj*T@)_<p437<qH|&d?iK8XG#Rzr47q1f>#NZt+R65G9Auh+n%F{_z4qCxa{1
z%2Yq#w-~-SqPE;?ex)%E7F!C1p6@4fpyGpJLoEe>08x;zkWgw`8d5u0;iOAp-|~Jg
z$UPrUjpMF5DoL({LK_x<S{uBjH1vfam*IQm?o0aKjAM7|OKu0G4jiD8)Vdh<XGw4O
ztRs{%#ZUo&SAW|zIW{)-cjCLW+rovLH)DXXbb`Uk%d0yoyxROmE8(%sCnB<(M>{w;
zsII>CC!1tlMNu&zE$#SOs=Z_eUtU~9%NIKF6K-SoOJ0n12mIn$)2>KdE=bBv4*n26
z{@lp)cHu|I5x((Cx>ksu8a{pUTGU5`i&A<oeeQW1n&-k$l!O#N`}9c>Og6mbvCORe
z|7VM4^X7M=-O$r}E)rFQW@hP0>~bn~!*<XqL2!E7pmr~sy2#vE3e-WEP<YHt;X}HO
z*X=*lsr#L4KNOCRkBeG%@_~Q1_@0FZJk(OnU0o~aR#x^+lO-)tx%}6-M%u4Dt!{EL
z<~OV|RA7<CcxF4%AnNW$HwWVL3lOI69(zfuyC67K!<b)tGgL9JP~f||#uLL?tqH|E
zJ$bUeH+obaOwO6_g^{-4KghTSP(D~%f(cVqjd=Q$4qkkLeOc7)pEi!ew06azFq*lt
z+_1|=4X@3|#aE7fJ~P!wv)q*&ji{cRlXF!=19C#>IV(ZLL(`#eVBkF2dDcmg@WtVO
zkAE=wq8wG#(7br{H~anLQa)^GFY;&=e1`bBuTPqdjSU)qQPCCX<vMHjy5(gh934Hl
zBF<B+9m1%8mJd9Vfw{RN)xE&LOXwXVGaWj0!>2~5wQ<1w1bHaJkEG`+BAwIv@*E|L
z1RMD;H1ImR$dcX!CLDQv+I)zxeu$#)Cpz}kA3i`{MYp#Wo#(}wK5rx=AhS?L;gCsF
z!qJI8DH;jI-i47&*9EEz4?*H$*3NPHaufCiVgv|lE1~6pLeFKJ2*J#~hu|RxDHs~+
zP})I+yguO1x{pP|1q#kTf9$mVcCbNgDa$4;?Tt2gT?jLyplK_R>gr&qG{^DdYpbiu
zG5q&&+MaC))PsnYJbD!#=-~qZpt6Kw13Drw(<~x3Y{VAg4y0L2L|rKPkuWjufPw|&
z*iALHuTWis5Jiu^RR>ZTx3?HpKnMoT!D5JTaEzc*#T!AmA|i2!Gsp=0ruef=VzD8Z
z(9zQBez>2VL+L3bPSd}BrGQyI7D=e|PiGuU#|!~83ky-g1^heI5CM5XU0oeiaA;yY
zB3}9)g!>mkK|zo{g8(1`DM<FQsA<aaZV->8$OqQrs-}|0l1DKaqV>+S4MC8JnHk@$
z3=GiJ*DZH*9ID@c>C&a#b`l5ir_cRaQX#Q`C=gwk7*Ts;H8q<1vbK(nXnXbf?$`s$
zwLy7jD>Go0<gU;<ahrY_^m)Mn!<IosV{@R|IzjUZ3$5rR=3})Rx%?t}I#2=fL!A$C
z19Zg_kT65dhX?`zi60ro5piZ2;rm#VL6DTlP=IStli-LzVD7jLXkf2rX{Td;1&5s8
z0|+Mofi&I@Relho@HFBH(j5A(P&P0@vVg8lJuWhnc#$&D{GmVKPr(QY{TZ}_&WEz3
zzj}4*X6{EQ_Pe`dmFrw~^k-v*O84yDi$uZd=->dAt$bR0aB#47Psv}r8x4!C_4yX4
zWyVMVnlfQ5Sz3i5RksXHOhV~S|H1}4h?QdkNJ;4wKmUQyU8KamP^;(4>gt1g_YiBf
z%Uo<b-`++$2wcp-oC}Oz8p9IAl8uDKjf_Up;k-7_l0U0gXPK)#h%4MXgoQZm%NH*Y
zXYeP+DRS=6znAuZbE9?2MkEgK7lK^4q7?24T}$Hlh4*F|B&;wp<pc?$V9`mzCxH|j
zpbKyXl`3Si&>|!KQZkFZbDH`Afm_|jk2Ewi%PT7bkB$)Zh?}K=(_xd13=RM7NkO_S
z4Cy4qR#>hNA7n9Q_mghN>_OZfH>BFA6`-F)w;wVqJ6H`=;=}=jJmbr#_EBy_I1lXG
zb(+NOD2?Rf?DF!Sf8@vS8ydzT-{iWUF*`R$FX^`M^t1s=iPgm^_rY3<cM|V4(C2D^
zzzTf<&Lc;z0l7g91p*FQ-9B7MRn=h>b%2GCNYd=SKlzj(Nrx#hGgwYsd_4LZP}O>P
zcnD!Y1N9%4k|<vgcbWtg$4g*Va*f;9FP(5>t3wvVK4VcZXHy%(T#5*Tv=8wOuXYjN
zkt2{bwN+OeKyRhQ5i$9KAPIb4qUq(e36%n%wv~y=aW1Y>SRbfoNGrGR+?SzSt_jr{
zx(uA0pW52O@Fv-9@{~GhIXSk_{6IV9A(dEl3*vQL9Q4tDpeQ=79QQCZRIk)26T?pq
z4(~8)({-EV-cForn}0PF8JJBe$5_a{d*@KpB6b4J0EEJkF>6Iso?=J@(G}X{Ysk}+
zWdzwB@|3=q1;Qyr?S|?BejDoR`C>k!4Nbu$!pF}Kg{1%+TR-CTcCx*#-@ZYf93LOg
z+~rDgk1BwL^nUe#gW5XSH8gn8SVFHQH8ax_TjR_2;ll?Gl>{i{53#U7>OO!X=vw-+
z87UR{db_VRW{0MbInhl*&}J4irAlnkE3}G-)Dr!3tg?s4I-F(PVHFZ^#N^l~PXuj$
zoy%q4PA>41Z$}@h=}d^cpo&XuNyMIED~wwb5F?3`Pq==xULoR!Zg*m0f|5z36@tjt
z*2k0egb$SLZp%UtI!5v8NkKjcr7M)1$uD2ZBRr|7%wS+~KQ*<5si_EJkE3ILLIO)*
zjTkiZ4i18vEJ+qgV*`C9(0yQAp)?1@0ll?CT07<##0={nW2ne$V--r}$)Ck}FJ657
z`7=)8Ib!fQJQO_eVevCEyol-9Pr}20jf|iLKRhymEh&KQ;q<(`yny2n5rHRwP}sCB
z1#5^^nEyw@0AV9;^wy>AK0FUZ-In=*`H_`TRba#fmJYlHHA{YeK13f7**sSWHotSn
z6NwhO38Yf!hkkT-f6us?ce|Plp))rRqV^MR%jcoFf>@Y^j&6CnN8HwS4zmm}Y7-+P
zRNUtvcwlF*{`m1lMTL}zhzo!r(>@bOr0^Z8FXZ!=?Q;fXG92rB?p0pX(P0K_yScvd
z>zBSS8#EomLqph<`0#Mq<SW|R+6Wjf5(y6;)Ix&scVPj^9Z~+p%a@d_QkEtr{x=zX
zrcs-Dvq?%y!s6ht2ts%dZmHC13On}Y>sO5CJPHjhEGP(h_>idmL>C6B4tn0!A3s({
zniwG#LVMppltKndN&hW#Q_~{HiR-s+-Ga#XAKCv+$mF2+#^xu)#JFPK4yn-p-aQoe
zx#i`Oe0*1AW%VJVa-RNaW@*U;AOLX&=0yPQ>%CYBv!XPf*|ne%*x3=hHZeMdn1wY%
zgB{zqu{;<`1SyXm#fUpkLk@?0i@q}y!gYrHh;mTc`tM_2#ow5Jf-K(~!bM;Z_O(n5
zkIJ?!i9zp&9D@7_D(l3F6I2Hes;jEL$;xtqoHj5J$cyo`{<(j}S)q0AD}P>M5Cg(F
z0G|@aiJv_^aGRT_pbqYaaE&<9`Mj;(OO8NPMU-O%<+ir=PoP%xe&+_Ncar%X#@JC(
z5+4vPCRkvEB-_boXldKW9ky@x8yho){uo#C$E8d@@DO;8o*zHNFnfYshaMFpRyhR)
zwVD@wSV-l0<G`##G;#=x4WJ5?<`563wFB$|Tre;*6Xl`!&wc*DuNagn(68AIc^%9e
z$lDXDHy|lg%TSAAR0Hrr^~#l3>FIN`vu|^AG1g{SDL`1(KjC&5ntY5}5Qj#f`-OfA
z&<!OJaKQ27$7g=zPk=l?$V8!o$`O)e3GdBym=I`)KRwQU$U8`oH#9X9z&v5iAvV5#
z<;ty`8-Qmh=eJtCHw8SFZF2jV(4jlA%90NG63PP<>zD;YyBlg6SRzI~DFPq15D5GC
z6F(QAxENwiqFca>kct8JARq+F1d0>~^#PGOB9Ba3dyLWujs+y;HQRsI-Q68?4@fkq
z+=2b*85tpQ|6W^5MmfKoiUFrif+hpEglgyT;Q?@TxE)|d4Am+;2OI%@Q;HRug31w+
zOBuAptt&T-Ei4$RsQ7S7NTXn9d0mqW3l67G=lJ*#An!-Kr(_g5b@*^{d3iaqR9_!a
zFASmuuoaxZUNXPQ!aamOx_}Dz0lQ*|hLAe9a0b-S0DTZ@si~{i81ly`8Do5+K8zhb
zZ|G?w6B1xOGLUpySlHX=A^d|(f;1mZQ5>Jx{$gb0I(F<JiaPxq;xtijFT`1E80*~U
zud1h)otkQFWOU}v4?%c4N5sX&H8KK;Y+D4B%3Qd6h@Wx&Nl9$HyrGIOJHCGxi1`fH
z0J~?^m1_i{6{ZDt?cU8VBxGe~hW)X%vnv5m0ZAk&A%S7Cew1n^%}n0j-p7u?OUf`>
zVPrIn`HP)9{Smx@5Md~YwK(0<czYd2w!v*<lj`i(2yT=#<?n5ae<I?*fgv!%cX$GZ
z-2|~>LPA&x)a%1`h)V+e{D44MuS#ggmY0{IfP$z316MOpB=+@rBXPmfFf}!XRP5m3
z0Gxs#456{O6LXQnc=($)6Nm=Lr#J-f$&(kKx8YhJ>|-`JG!%r_U<&Wn*`qSq_fxiv
zp}?Q`XO;+~5~yvFmhmSb3RK`)T8#0^(T^XST3K~``GWo<#&toF8u8HqK(IZ;<SPg_
zNX;Py1gma?A`R9`oEFW=0Y_yiREm8>_g_#@?fm&s03TQ;WY~c6DApD*<EVJ)64=$_
zkX>VD=pfg1Fc~;Ero_WG!U!^jGUF$nCFoZ_pm_Q6B>>go;Yq9vXfli`wKO$xpE~vV
z(<e@(HcrkX<SD7C3fkIb<>m2>!~5v@k3w;P3L0?~`zh+am=PbZ<Ka;T0bH;G=C`rE
zA%TIh60UZJhA%J%fjRGjf<+iP&ZLFAW@H2*2i5~FBAz-~xi(#j0ZBRrhT57M6fx1!
z(e}mK$fgJr00NA3bWd?$I=Tyris6592=6OLfmNrcv(nQ~KC}+mzp(|=8?-BqflwYZ
zNk&Ex$>T-<T9IfW2myQ8IDCM#&31nOec7w(>fD&>78VW-3DGUGxsB_)e_tM^7+3>d
z+zs#z-jCVt#pxb*Kp$)fhE=<O4^hn^ej@+E*q{WX-M`<=+8W+6Vj}{N2gG`nk)d|w
z%2<(i{@`DryWRK0<Kj>?WdIf7hpFi+^j8;mk`W~3vQN0dm9P#-0q|zfMhK($d;+x2
zPHX6bGQ7oyasU$F*k{ki97-^dib*oS=LAW2V{2>7g_xU|EMcw|;T}ygs4x~6o$!Gq
z&)dlMF+EL40NDWISU_C76ef&@+tANqdq5a``U8fq6bua5|KT}{i;DpuP*<bRx{I9r
zshHzy%;5vGVN|U0R+TT1I3*ol64Io!v>uF1N~!`fGR!UDqXW>~`HoLjfR27mmDDFe
z(KO8pL6*or!wj^vE+t=CCVsL(^0jp4SQ|ueB#WxfP7(0x=>5UcY@M9;?b;Os^S~f~
zK)`OlDlFdD_IBjiSRf^2b_@yG!m(kqagy#GRaFBJ&BO37Uc3lS5fMOw1$5PGOnqZ^
z3?UwO0*H@ji)Zl~q!WAw0n$UI&46!3g_pTGXRrc3-~c={90P+(!hC$K0NCHZLxum_
zMJlmh`VwJGfzNTi27&YMe2X-4An?XBwud2W4=_{#7snvK0u^auiv}j_Po2UD$uF#e
zj<z;NH+wLoi$IA`<At6KXzx2ei{{V-K$f^;ZH>`#jFJrsy#y1+5fKv+A(z%J?&OC6
zHaNe+v@gvd83TU1BNgwMn$~~%v|711EqUVge9+*sdsL&{E}pkO^)(NV93i774N{}_
zJ4^BK+~xLL2PW<-hzDOb<%<wHW2?z^n)U^AUC4ps*RC2fSX$Oz670QA!%Izl=c^Ja
z_4WtGv<Gw=J{dWyY`s@|e!;S7TC~DL+MC-XN|dstY;CY->E~WD;?P5ru|04b)(%((
zi7^&N%)szTg4t(hJ^cdKoY=la#OCIXKW(6J`}P282UO8SRt^yl_ym*-#IukXp9nZW
zN7vcWF)}!aMGsOKM0@8lM?0pL$oA4<`XUTkE=;pw#1ROk2}Ec^_%Uo2FOH05h5w<9
z&AdJVpn;)EBw1P-nxWBATJ(f5ih@}T=yAcKQBxy7vSw;+gYr*c?2eF@jSVCWWPVkq
zf@GwmuU`MQE2fdp>9XACkB|I7_9Z`j>QEq6N=eBNOCjsdw}(%jOowp;cc!Ee)p3}_
zKv%`m$_mxC3#et3IH0n?YX|^mmG(%X&cZ0hSwFI4==DB$;G~nbo8ZQ7(h6qFYU)Qm
zjF8AY0pvhlMf|W4cs0IXzYb0UU^z>CrYtWx8P|q!WAS)pzpA7YpO6=~Z`+1WIzSw#
zFsK9RFm=bsC=RL@2vBj4Pj^_{S<G(fsj0oQ>Jf`=LbNdEj{#o!Y;2D}hl%JK{f4F{
z3QP#XcOE``2(~OZm}1{Pp%^7i+I<9RDN#RNa^g@V9`HbLIet}CbabyA<q1g?W_a#4
z8WYBO9qI(1+M@f12^X5#1y&bppZ&GY>T?<!qU{Kr3edButYYU+YdY!T^7P`|$7zxz
z4SVlaah2>i{^@Uv7k7j<eWs_{KQvh4P?0|Ew-lJyxKxp!Ey#5*V<(}|&b?JUBwNRE
zQ$x$bOL*b8&gC`Ns-m>>SwD>%pEbwu5c;oM={?u*@F*9bbgDZO`>naaX<(}9@HT>t
zaAC_>dAPBp@bc!x=gRArKRSm<2F^;Eci&Gfc`p9C;76*)MhxSt?LMW&-od?8;^lwF
zbYj0(mT$~1vZJZ&r+49Nzq@gn<7`{oX^991!p~-QlO8(R@cEJBqn_5Wzb7Yciac(X
zkr0I2IZfizqKxk;YyTadauDz!6mgq58f9qwocvQBdhVXR-7!K@jB%mNY-kBrW}SZm
zs;NADmdmqXozPAvgPR04=8fJ74t#9NM6$9*U~upm2GM(a2wU4X$UM<-#?aRpMpUzC
zBAiLzNw89#rgW>z5oUQ-GWqf41ZC-J$W6DgshF@+?c06Ci>1fPXLR2>Iu8Y{c}=L6
zZM+QAm{HjkG<53}p(yrzl$FO$39*S}qnR3xz4X5ZA8n8jvSwQu%IA_J{W<6^xDsY1
zRCM@Wx~#aIlqDZ1YW-?x?`E1!AcX4CPnc`Z55@h}vpsu?;Ig~$8401mbydQxM%wD@
z1y*UwnS%tk)1f;(`~0KVtE$&XeRP9gG7=s$ouwp%x6hPYl^F5I5IYuDKV2ebKmrVN
zm`c9wo&vw8sgbO=kY<7}o@g$NN^|u6Y;EnInVCsdq*0GtQL6(K!)F?p>fHBgOHBCp
zYdlT*R1d*th}`Xeo`QXhM8=L(l5j!!Ptww2MN>2DU%u6Lfq%a?7S8`^=Yj%=j)vxi
z<H1#VZ#RX1zETr6FR<Wqw;H0Ypew_>Ly~q98usIZXheNk2uIJiZvXcbak@M|SLZ}|
zJ<R_8rgioI)APVu`c3b<DR!;p`nZPY`Yv>jsARQQU)gn)rZMYCyHLiIuz8CF7k|Dj
zS#U@7?`ZEX%8_7JhR^rL*62iqQyYiE90_k|3l4>FyHyUBCU%GNG<;ef?wjB~eqGvj
z=4-B4T2#&W{rS%$UvedbX|+nVe-HgAZq(lD<#RWjezP-{!FheN>a=|69fuKj9Pq{W
zp{p+kMe>SWHop!e>)u`2GA%8e`XC;<ba`WNZlAjP)LO!2x&=$e(x_v<&vi$2wNYxs
zeJ|O&VH>xss%hx;?K_XUqfpi}I-%R&M58;-jPGVo6L(W)*=4ezC}U_hGT0&^Hu)*{
z$)MWQ)%NW-A4$oQZ-r4SbW^fQG{*HBiQ0Ea4|PjgM@40MtjWlgcKjWC;5JZS)y~dg
zNX8{gw<;+jHPCtYIkWV-MfpzlG%ezx{d!x4&p7{JkGQs~yV4zZz47}<#cdtNvTAl&
z&5GQTcdu21<4jsGbl?7);xBEb==H^OZX>_soH8$c*{KouvWt$TEx+EJJdnK9FsC3_
z#CFHO$33~PwxmL0xw&`MNxj%$Y`JXzfTFPV)%#bTY*k!W*O*!V`8qSr$x_VRaaZ=M
ztE#MY*RFUa@kwS@R#Gvme)v*XnKK^qx=u;+`>XA*-HZ_R)sjn|(rVYlhS|3}giBl_
zEjKUpxUONML2-CsDP2~ZbwPI^Y{dFh6VLvcWB2lE+@u5#D}9*Z7fzn{F!0pzQ*{YE
zn#OP1xiJu3xVG;St5s)D=j4IZ-ma%_=s&Pm|1f_x^_q|K!#B%~AG{JrUX_l^S--p|
zU}qI#6@ES-X(^z^eRH^srMEx+Ntm10c!faoRk^Ysq0{LUbY}#b8x4lXA07xlZE{=T
zyM;;5pSVOX{z^A_W!A$?X`zXKF4u<+<#FD-vif4{=s-=+nM{FsUp0%r2`V!&alhX<
zUKRV5p(Pa)X0PfHfg9VP-1Et6`g``U<jS7&UX%XATS{)O-KJ7Sy4%wv!fg7=v`1Bo
zbGX+!?u;*e8QFDLllCR0?xQ?CI^osgI^AUl%ZXHBZu`dhwJ-w>K0B)f-SG22c80$5
zdOMw+n08Pye7R>!<*)um1@{bVo_x?V!IfT9J6-a__wt4Zb-Le37pzivvGI;hztvsg
z9wig2T%9fFKBKpM)Oa{pMKH(4o<gBHcuc_c1%(?yhPEPX+y9&I{|E2=CkFlR0P){=
zkxjfcWkuAMnrpl&d7GEzO3=+W%1-7w2W<9`yY-9R9dEDos90Y4J1VA<&8azEAH)Ck
z{6x5qRuV9}BZbeV!}>(zXS21PALFg!E9?g+Zy28|ir>E=vou+nsQgKkZ*4YAzPda=
zYe&&BB{iSvR|P+U-F0qvIKPoHZ!bC2e+VxXp3J{==<85M-<smjl9YpqBJX&179As%
zw25D&>|1UsHLKwbdE~twX%bm^YPp-1C-c%aD-w%^_xWmC8(}J650B4r$Uh889rWkE
zJ??y-KA&gLviCn>Ey{}UZ=C#ZEW>{chyS&pTO>bIzRw6!*(u_K3<($ImCt3#-tzq)
D@h%gt

literal 9079
zcmdUVXH-+o*Dp#jQUjs)A|OQ&8!a>e5kZP{sfu)@NDEyd5D)~BE*KO<niQ!b5Huhl
zEz&z7^qx>cLd_ffKkr@7d)Hm}%l&lMnzPoPJu`dnnKOINZ_hsQMg}?zwA{256ch})
zceNf+P@E(Fc>t7TPRaq(eF}>6M@IUl+T`l_^XDlkDXFNasHv%GXlO26xBvhEE?&Gy
zM@L6bPtU->z{tqR#KZ&y0-2eaSy)&=AP_4nD;padJ3BiE2L~r7Cl?nNH#avA4-YRd
zFCQNtKR>^KfPkQ&ppcM|u&}U*h{&Z&mqbNH#l*zK#l<gQzAPajAt@;-B_$;-Eq&$6
z6&V>BSy|buSFg&+$z8j4?fUiWH*VaxdGqG2Tesxp<rNeZ6crVfl$4Z}mBC=Jii(P=
zs;Zirn!38WhK9!N+qX3}HMO*~w6(SG+_|HpqjUG}U0q$>d-v|?>FMd~>)*eB-@w4Y
z(9qDx$jI2(_`!n*CMG7Prlw|QX6EMR78VwkmX=mlRu3OOw6?akv9YnWwY9Uev$wZ*
zaBz6^=+WcHj~yKypFDZu<mBY+?Cj#=;_B+^=H}+^?(X5?;pyr5^yyPCFE4LzZyz5Y
zUtiy6&z|}D`T6_%2LuEJ1_lNN1wkN?;NakpkdV;O(6F$u@bK`6h=|C@$mh?Wzj*N?
zDk>^EIvNUv#>B+L#>T>6u(-Im`1tsjFJC4kBqSy#zIyd4DJdyAIr;VL*C{C}Z{EB~
zO-)TpOG{5r&&bHg%*@Qn%F52pe*5-qPEJm4ZtlBx@8EEFUS8h&_wPS^_>iBUUr<o+
z@#DwB!op9VJ{1)e6&Dwml$4Z~mX?*3eg6Epyu7@kqN1|0vZ|`;%a<?J)zvjMHMO<1
zb#-<1_4N%64PU>0ZES4(_U&6!Q&V$ub4yDL0)c34ZT<fJ`;Q+#kVs@(TU&d3`_G?0
zJ32Z#J3G6&x_<ro)!p6Q)6>)2+uPUI*WcfdLZN>D{yi`-FgQ3kG&D3kJUlWoGCDds
zHa0dsK0YxqF*!LoH8nLoJv}osGdnwrMx*EE<}et{{QUgF!ouR>;?mO6^78V^%F62M
z>e|}c`uh6D#>VF6=GNBM_VzXwi^bt^J3BkOySsaPd;9zQcs%~#;DA6N5Q)UY!^5Ma
zqvPY_larIv)6=uFGZKlUo+|an87}(Vweq8&pkw*-oEw~pIH8~rq|nvUFooo9%>>(k
zT$*sc{R^R!dk;r}^;KO6>f4r@EJ#E=ZMVCIHtTv4Te<Q!TW0B3GmpCZ$<m*bqlNXw
z<pFB8Pd(}?%FC_m=Ima44^tI~y5FSA`qjG#?+p(p#i?dP{8UaP#Zt55xz1HRi4CE6
zP6O>^{r{r8!pS$BV=QxfB6Tu)=0Oqk3su7&QHY6Tud1LzE?h)9(jpySrr?6C_opvn
z6ki1rVSX)n*Eg1L)@TG#LRW4~ViK-1jolpK>ysLoq4?7S@<Ve=)}3R90SF&*g)vQh
zI9+;f`cL0v<%GT1MLjL_s@HQ|?j4d<%eO$=k;&8bK!PU3z;Qhw?9^^1R1jPw$9C4W
zXBc5^T`DSdD@^jc68pjQNO-x~>!;&d-!G3(Ia$M9h<VhXqj6dJd}rO+k^d2>d{VtV
zvsvdauMRONAQd2P8IexOW~{=zB?lwb$mT+x!6X35|0<}U`^gb!%Wdmcls)EZewLS@
zD+L*_6~?C}rCo`8J73tdl_w|St~`5VQKTCFn+m#;{7DauQ1kW=fx_-OT&ZOXp&R_%
zIo`{BpOOFVWYG6z#hcd;z<39M_pTTG)D<)S(Ma6FPscjPVdF+8FV^49|B2-joAK{+
zZs*Fv>Vvlri2jQl=~HESM<&2o)0_6=O^=@#Mh1C^^H|3pQtVkgbvbTLn|3ks*TxOc
zS=`nG9W9sKiRq;c>#J09g?l1%*?4-?BG;2OAtpm%j-<g9ImE&F7v-ri!!Y%FZ)eEk
zh~dTeW@pwdLBV13XDKT*XbVamDi^4>cK2<QQyPWjYOdYe-aF$j3}q=X_hXMTTj-8r
zr<AQHX)M{t8TPs^Cg{E{{CUY)N(X9ABZCiZb1}n#msS%I+Jl+z#3vc5GZnuAsL#Ks
ztMOBR(yHO-k!Qt#%%Q-iq_-hkq~E=}^32}SQ_Xg^^-^HqZ*D9DqKDF1?~c}!?(|Yu
ztH%y0J)9dFvc=Nx_<G+x4LeEQ6~9uV7_x&vfOv$!oj2cEnzJr?x?!EiP{D~7UJv!P
zare;Nl{0qA;qvhu+VqC^y6`fAr~|;W`UW;j<(KEvelCph5f6S8$XLYUZ1Be!7}^wd
zS6Rk;fK(lc{6S$FnBnVQb_;6rSf(|Qd}-(Tp3Ekx@-3|luWdQ%p)rAG;_L2RAUt%~
z1<kdNb-wbNuG<!N`LUj;Npz8(lrW9X$L>Yr?)R$jmYByp9y(G2ZW!o&w9U261a*j4
zV}qkBG(92nkaa^c_)KA9^7)qPO@v33So_D7kfdj8PFTrWmTypY)6v}WB0`KHILO8}
zebI3{%k@jC&RRnE;ftJbkCRcDLEowwrg||Zj85D|FRvIgVm~Z@efOszSiCmZS~{q%
zRxgN4G`@$H8w57HCc0R-`t^LWG1An-TKvTgzk#!aqdYhd09g>^XVzR(*CKqo&{Q&{
zly=_FDgIipf12dDm8&C1ZqugnTeT5<6fSN{VqM5*x9_0m9$gN5?PZ@&wN)<PO=0vr
zFA7X{7g5_)2LVFh+nJen6crR0{UoqF7JSuK)xus5jT<kw2MpLFB=lOFP}^g%xMG`J
zeosdWVaSzo)L>lSql7mhuYz(NmSkAO^p$W8%m^@4XMWe;P`>2w;Pp8Z;Zo5d;RU<>
zewGP3m0k4dj3r1Nve487?j7_^@o`QH^R)b6Q<6xRK52c(IPR3#x@xEn5i!9&2oA9t
zr7GHgsed>54M!70?R7%$^x%_CMZsp{Q^j9Dnp%ex`Znt4yO56w_zVNDH!<=?`sXpP
zR-54LG9b}=BNP+05%JzR1Yh>W%^T||LXkSe(Je%L@WG91ni)#sEsygzct5-gokawW
zy|~E_@a7HTpi+VqwWOsqy}w*HXyS%_6A9Q-2hMV?yU={S1Jf6Hpb+r>>(%}VC8uM`
zVG#|;!Ay?8>RU{szp1gYxjy8xMY48UQkKI-sEhT&k%KY3XR7VlyrUFvbhAmTmWmM%
z09EbCoD|F`z;TeFGis?EoJ<1Y!O0pB^O^1{J!by<diQBn-8{r2=<o`LFL2Ti3_<rc
zWiPYro|jb~r;db%<hrOs4tVKz*L`Jg_*&vbEmqXMoM5=k!_YU84$+a=okuRG;2C+j
zka1R!AXv{!i|T10ud>_`?5;h=<4z>qCl4T;k1p4gT>v3WQtd#^i{4_)E?_{`U@Fgh
z=aq@H8T7Q*cbgQ$2AGgFbLN)aKm32Ok)CK1-G^)ZsgT#PeDX1xKSc??eyK}iMyNp{
z0P5>lxCxVhKiU?wRrT>y=Y}s0)a9Y?e4>U|!%F=0rB0*dK`sF5vvRc7qncskmx?K;
ziiE*aE+E`^eKCb6$qL(80SKeSU$%XY)Bgq+dUJ_+iZ9wGsDQsv3JBNeyGX}z6S5f2
zBi)n~k|v`8S)T3=Ox$%30cOE1VIgT1WX^s_xQ$-M*kfbKo)zV5=aJV7ePf?S9ouW|
z6E&tIlt@efRNBB$0{4Y)8MM2`?}RlNpe|j>+t&+Q8!n@Hp&`IGg5c|cqFs#bhwQXR
zv)U>bb%AiMmu0?V+)wO3=EI|J08r55n9-JkLXb*<%zHTW55riXcstLGk3~M89|{Jb
zo;O8EnHPp-WL0zF1zFW0M>4;fJ&x{YgM29Pf~MrU-HF?JcgA?QTk*<!D&$aLUwH56
z4(-^bZGPuMCTP({m}xtrn<YQlEdYkAWjH*u-2C#Hxy=}*!3-@b5%R(O$m`TL&uUk`
zPTm;%)5gCzUh!%y2A(}#6a*NeK5tp%-}<<cq^WYj2?&Sh>b1YCI~9A+AFTv1?BPio
z7!Uw~5b7!y0)X%&S=z`~lguU)3E9c<a+1zSQsK(w>%2&dFm0Q>6eEY**gmj^a##nt
z{m`d|^s!F)Jtqtoy+Tq4T3dYJZv)o0<;;k)beseRgw=1;&?A`)>A0@$2hM8``*B9a
z<1+43ePCbrnu{c=EX4P{<bP3TtM(>(71pYLR~FdrCOgA$!lbs~*VktZ3X>Rn8A#+b
z4(jore_i!3h?3MM3xAd^Sbs%ZCH4^~#GFJ}1GRtQ+uB|ZCl*Q?d#)KPy2w1Z$6!*q
zut8X#Z6n~+7J-q82Y0_EhA>k83rrZug-M!JxtA?3PXmV?IbX)81O?c8u+@gX=#;fK
zclS!lUNupUMWROt&Nonz!{{8(^ze0eq+#*ORn{#W(@Kb%%V?j6a}#O=6GZ&9_*Ctn
zI!-MqgfV)6(LOi$YemVnt6lsyO$1DB|K;-X;#IPaq(F$q)P>K*w*K-s<&%t2J@eVE
zm=OB50Y-tr1KP7c*r>HFuWxYL^WCez?8s@WT=`4(!bf+4C@b6Kr9n-Gp`z4>usE)y
zvzm~At#74u$5Zyrsbd&Fb5f(!T!4vk^?{oYRgzGDtsP;Wz&q^t`y7VhOC?@_kfHFO
zim!<7p2=Acb8+W;N+iz{-@JxpN$-4k6eGa<x&50Vk%hpld7%ta6mGcgvMft`j)`%y
z26^voPyZgZPK|KCfMM``boJ3XYE7tF=;7l?;kZISqdz_5zZdpGeoFD4?BA6X1Us#d
ztc80RF=_BXi#&UqL*@0po~2y3jl!AUZ1?bb`<N`016s6xk2GoxKno#a#h(NL=EJ;)
zO@32jB$VIVDCkA7<5~SRAnWUaBX2u8ll^IsCiszdUHby~i*rDDexH?2@)HCsg!V`f
zti4OqC#6KaF0mSo8wL-;Y2WSw@VvDeke94XfO!N9o^_v0s~XH6`VPvFRb0T7C^h(e
z_+|UWlIafwH0*o#mY}~`a9N20BXKAcvi5e|(8%xy6~It;68m;#|Jtz0)pJOb{BZet
z65FujRSGig-ZG!iU29nAN{NvOZX*5MsLUdkus#(8-^^p^3Pq(ECPbe93z^BNnHFjC
z+omv)$Db0<+WQ>GDqZ0z#<;UM*$ae65l54`q^^z7rK`~)P2SIVTLjYJITHo>&V)uR
zD#7&jh-krzU-xlC)ObNQ#3QpX0tXyq_<le35>p!*wP99#!cUed2#RprL>4r}T@aTq
zG-%e0aN>^@kNbu;GW^bZ4dVf~soc-;TeMhqf}9BwyiW|awx-g|W67{vrnwd^(5|aZ
zRhTXWE=r+8HxJ%Kxlz`w;K^9|E+RK{OjIK*ZnFD?u2tq%{VLP2=G7P*Ncz0a{Ckr6
zJ*hpJljUlf`7%pT>-Y$05n1VdDuGttidKMdAXZfR?jJT3cRB#FamRTlD8JS>_Sr!g
zkTt2=@R$Y9NiPJ7a2q_j{z|I{8(IyU5pKPPMWUwf0tt%*qvRjJ2WsoszLXQSOcFCg
zp?^Oo7AenqAd_&$i)YmrIu^U%##Txy*^>clOe?iz76sz!rR%$5(3JpBZ8s+X%54#;
z)2|wECwj1(Jd>`NhMRM|=l`-fqzld@X^?*0jksquq=OwizOeQhIUab1iFDU=b3(+B
zF3Z)wMkeW1L}&V4+Cm|8a!tWnDm0E=F&3_cy@k)V?BFp15M96P9s=ufoSHHx!>{X_
z#$r&sW_KPr!DCRRL(`_-I)4XX2aZTb^;xftEVErOnX)2r9ndoJiqJRWD$2)}TIYsY
z4E_wU$lRC7+*^?a|H9>L>GblfRUsa;#H0ta!v!ny7Evb3gQ#^`Y3WG3%lw6aBBls-
zEOL*oIk$~98}~>YIcfBei#(ldTqibc=2L;r%m{&>kvyR{YU`0g-1RMm1YGL>7OK5b
z_UVmJc~gZs;y3=+S@m~)rfl5nmG&}?&>~b%D}9G$zVVQQuoQax$lve2KK_7Z7E-Ql
z<$b;WH)@j!vx=K)_RtOASDF7>qu_lPp|$^Nn+6_3+5*P-(dtt);+W8+=A~U^>%OX%
zZBvE&*4aT*AcPH3Sc@Kpx|97X-UhO@Q<tpu4qO#cxvkPn+v;1fEOx9zcC$Er;=Y&j
z%tNQZ-R~kOwZo@ROER*=EzS=QM&#6gzN11D-F5m3tL^mbTYZZP^#l#mPnCsa4=;01
z?LSS>R$c#{=2%kbc7%TOz1%eP-vQ8|vmPuW1EJ$ky|V2pF!DX280Nr|jT^dhR@Ovj
ze}^j%X>Fs42HeO$)WIn+!<x?q$#yWEE%ppUl7?&_lKM|g&>}t)ex34*NSE{(K`(}l
zGJ=rhX5)gu8}GqdRTuBt|K3t9h{5Twjr;%XNbefSY{mAslSxwKmA;fWl4ouDWmUTF
zkLwMnE>~!V(MARtuT@_lo5a55xUJM)fFq<~;S<IA&q}Wn{&oS=wz@|zJ5$%8+~$-f
zJEsHUbTCp6?T5qU_{hgbet54f48nR=Uidag5nI3?ZM=vqtLvOLZrWu_M#_Ie&8ny*
z2;|8VIO|fyI*l!}bujVEOl5&wJ|${01?*>{v{HK#U?vTS{=Z8KL*x!FxNjqSh}C@A
zkQiH6kh!K(JDrw7M@kr$v;-!uU0}N*Rt`R_pO+zf#GT#hrJA<@{Io0ifDjAo3+wA^
z(j$BH)n}MKYxGMCy36IgIk-Kwf$3cC*bw^I6gq*)kSi!AiFZoB@^Atc?q0Jt@qhVf
zl(euoQo67T$9b3@3_>XSjjrg;9|myJOyF|(+jeW5a)Nrtr{h8Zp9WN8N;OX|qV#GF
zKdb}EyZ@-jD1eIW7P+@$BotywrTVNpEYKP%RC{x*r82dP{mNf-13}~^i#93#=2L9r
zSL4D>w*5k$(Ljv6HLBoiN^}T)hXzrGm#e&S!i;PWxl*xT3EutG%!p@AjK<lzO~^Sh
zU>Vz<<Hn<KSy4FK#ve`c%e8`GD~Pm};*Pz|TE!~>bjj4qkSIt1+q#@31x6WfwIOaY
z-J`BTLnjzkgeV@K9(MY`A59HJm@qAJT*P->As-Ltpx9|t%j0w@Bc>lT|6Ea^vp1q2
zn^hL@NAUp>puUSb@f~s-#>qcu0qFSa<hQN!i#(7Tiftb&&kXX2*HodA5)2dP*B%44
zkN(rt`zYAWSru)N=}C!EE<B5w40ilPi46Me7X0wQYgn_2noRruir$b#B!tjrs1q4y
z?vG60XeaL?Q6NHd1y(!Tk^djM&zSsKWhM20a3f2q*QxvZV25X2!ty*u+284Oz3-Tt
z{PAQ5pi^3Wd1j4xbyTQ51jBUlvCQpx@x?_B3{b3kxbwzSaI5NzRS6)1^2#oIl&dLO
z_Dvu{gwFn(&nE{FvN%c%-oq}Hm*wW7->_rV1>}Z&&~8D8q@+NNi+k2ZQYKvLLRKI#
zYZN5vI6|I0(-w{^6139Rca|Dx8x*Yz^BzUW59Pqd8bT8TAvMQZ$Sib45^8kDVGt7!
z=K#!et0mWav}r+l#G;HEj(vht8SOzXT=K?eeB^hfrFKGnyYz6%cTz7XS^|bPdJu_S
zqL1MBBrr{49JOYy48?0A-Dem+@v2tfcg5Ypj^|2TuHrtQVTZ)y9F-?$o_1sEbnNL7
z&qL^M$>Eg{e-1Zt=cjw&^YFxm<2CEQQ1lxi%ZsS&b|}2UvZOlfS4Dd;T_+zRJ3!2i
zY~ll*Any{JP&Q1T>rM4?9}`Vo-Ru#`Bw9S@Ci$idV}o7k{TAGJLf|a?R#`_oxiu)q
z7GHFe^o?mAZ=y_t<Z$7y@F-P&Ymwjd!!&ZZtS1UwaDeTdTQFnDyRF<j_u+t}E@7L-
zgrW88+wr#!a_hI&6)o>5huxi$B2qpePmNpk^u@CCY!P{JdA4w!?|uer?uzDn`^#s!
z62&HvB+)ntm=Ws7ts~U>NZm&1=(abhzs@bTRlpvo2AP%+tt1RJJYEhsxurtLb|yAj
z$K&M~bvcYtKa`x2q3eyP8*M9R9~vjRP;|B`lt{G&zM`umMeQFci&q>3u4Bg58d0om
z0{tqoQ*77OUUha(s?Z#-=iS0k?Hp@4F=F#3n1N?pouSY|KC#)ArIFGA4PvD2%hM{-
z2sBk<#fa%KJ@CxFryl6tEo>X~0p505@C20C7~3-nJpS|^-liaU;@J9A;Ibz#abhmI
z_<|fGfHeMDC{NmqUE#`Y6<CToF|s#Kx=jD;&LggM#YAdkgJP<m;#vQNA(~AcmAEhJ
zkn)5SJGlbW_|c<4ja?Abd*uDJ>A;PntpyRXb}-PcpmDJBM#w_tv!AlX;&S|eO1UuC
zL&INA;kHU7I~*NUJkC(%lkSFhQ3X6X1n3xrYcH+coc#S<xU5=iIF%CwRl}GvPy35w
z_9Dpk$!xZH8{Ai9k(^kSXo7~YEFe`*bfWC|85kb*MDN=)B#d2UB6U0;pK56#=$U0T
z-`tb^+*T%T-A)@JgcKvpkubyGd>GNDHk{hhHM3lx=*}-3sP76VSDcbZkQVU>HZO&(
zufg<a+_6nGU*)^%EHSzj!c9Y^=N`licip>sDcFLNg;VbjaO#WtZSwo9vOZTrS%m}Q
zRii`LZ{X<d7q&fWma1Gg@&{+ZtvGF%4}E08-BrKUk>^8Z#`Qx2ANXn5=-XPc!ldF$
za}8OqiQj{aP*IG57}=7=C9e6ru%3FSm}zwt`rBDCt@1LL#LRcK_iIs6n>xEhN5HpO
zGvcQ@jqUFDGb-Zc)jNBWNg>>|S7|HH1lv_Ryo1oY1B?ZQjm?PLau>&;-bd>0ne){}
z_PX}TwvX)UiaRwJQ*p_&7lBK3geM30_gdE$9@r&^@b6av>|5W%rYv&^CZNG+>{rE-
zN!vF}I-Jl}OELeldTUOO5&8L`lrrpb@d}1!7KSdV_8H&IE_x{d+<VtGQ^!4KTl<a3
zcLrlIRMrS6zI_v7*?Y#nZLr5}Y}n2PW+XX$X@CvOO37UDb=O|CK2ynEmiIq}Rs9&l
zjb-$F*7|Z9Rc*JeScY_QuYzsd``n+o$|Bo_e#pN(_j*%K=t`*Wox{6%x$0OPH?iT=
z>V%H8N98SUs8NTma*Nk{FhuJZZPP+GM{T2gLK{igGL`knnW(QxcvB{PrvuuVu4o0d
zXxrviA%YNYs5u8e7`UwPWjp3W;j{GW-y&^8&h1VM29K^0nGsHgR#oYaUtJST(^<1q
z__K4jTRSFyiMWZ=Scu$uy#{M-Zv5WUF)Q+bSvJ*0Cr$58dx^OfV%0rm=Uj8h22>YT
ziMEyDP*Rv;@aN+$%VMcjTCE#DmyQ&@bsn|9rC{^oEu#rT<kHaEHE+7m5jr~CPIfWr
zRId~SmvZy;(~|VegsC{}#7Q^!Q3!^lx@Yd2^{Q*2df46AwSP5k<OP<dB8@@vAlFl1
z&LQuFrvJjAvBN!#z)6+cliGyecQ_Z#IA8OF?06N7A8=h$5>{+e>m0LP1<Ea0k8>Di
z=t3>V!j)PT#)k&`f7Tc2R4B+8j9QGUAp})Ll*ArpNXJ*zFl+ap*n2Xi$OAgt@=l*v
zMaHDo#9aM+tgzAe;^(2<^^>uqiXD4iOBp>GpFC&rfLfe^b@)Ys$ZGL|GX$SbS($u)
z$FIJ+0n|Qj;LpjEZpME47LtILY31IbJ;dd|7R@NQqE&I;>5v~X;0*MNX+2q}Un?9q
zSASq$ud>OCh<{XFTpXA9o_SvJT+iv~kqkh#?_!=<?h}4zZwyFrCsmC;k1^B+U>_NR
zE!Caq=2ww^=UHCR#MF?r02ULLQg%;!a{F%M=SyGg4b5Vu?brrU*T;l@k4{b~yUhM*
zie;jCC`1>zl}za^d%2%>=L^T=IDqdR()ML8ER;=bpI5B%P4g1r!%8Q7s@Z}{AJCVW
zzQ2QA0j)ZCxL~J3kNTxzlc~;}dU5yV4Ywj)23F5#<5P2Dy1GX@!6RXDI8j2x4@s*u
zu<B={-FtH$2%hHpU@z8IF<F$pz!sv*{j2utcZFSY)jo%rWAp+vtJlnl;S&6}&hxzZ
z{X^*chlW#2m<1SCwo7Z$p<_l}v0uiRB_3>Loc1=CFP&pR%3u`j7v=CVII*t0rqF);
z_hG0vvd6E3OX5Vn#5;+5X+$SKF%FPSC&6mSXo+bVaWqUeR8c#K81vTsJ>I4IWpMHR
z$EwREIZfs2u3K!m@)1K4MCE>Nkr)Aj*;!qC`Wan&st8RHUh-v#)iqli2|gZ+ERb@?
zhS?AL1RJ_YYn}@{@@VLr(ygK@)#MR{MAnGa#%3NBd14<Ew?rQ~yPHU_heHMSaa^)K
zqW!GR&R;RW1<KeyS^9*rXE1v55r@4$W%K3N)A}u~4I7@4@~+W`4jXeq23u;5TjH%q
zlpkv%esO#0Mq^`5AxzJ8^ZTw7mhT8(kD;R!<J?-BF2`c$17ElrERX|}=M*9s?7Fh`
zIWoZg_g$3yFLJE*KkuK&kw$GB9lv{|annDEv9K=2E!hhn|HgF#jQ@Sb<Q`-v{0BDw
zuWxx|1q-?U$*yw@JJNE2tjGkD46Y1iA~VPRU=r-eu?+Wk6NVfBCp8^r12Oux<$G29
zbK~EPkxRX5{_>u-ZbZ}P0kv(i8?$|OdVjNiM{4UHlz21~IX4(V{3Lw(H`9k*;w6i8
zMtTqV3H&9<(13}oj+`_73j+So;KsIpa;Tvg0*$!p0?E2&BiORlBm=dKTRZ6%zJc6$
z>EtD0nTN*<fj@0%LiERAxU32$jGV0i((#*WGzOGyO*`rm9!}F%6VCj=GVd3ji=IjU
z;{P!->xjv$LtrCsOLCsPE@2D!NlwH&Rn;jp6@WYq4Nsd;M#N9yGbkW*@-P|}zI?6b
zacDjr0b6sRi`+~ys8VhbLWjp9Ij94N!qy(jY49F6JO2zk+wLDkKKx8>2mfjBh!~#o
zK<s}5M-_)Ak#*$O)7R-Ye;=IxBSEF*96J0?n+3zb$+(>3T3;M!ZG}e%R#S7U=Z^{A
z--GfZhyif;)Y;*n$KFj+U%*gMp@%bqW+!LDzE`&5=j5v4BkvGO#Bt);D>TP#piWEH
zL*vdbnpD1cHP3MN^JV`nW&r12M7iFatzTa32*dgBQH{#f{#;+Bni}H4&A8kE?*zp4
zIE<>E1ZNjjp#$pKU4s0hw$@4_36D^Gp9I1GZzU`T|3$0DpJQDj5rAxPGj49G7FZWb
zyoe+%=x~gw8KE5XfoET!;9(Fa8<f<k@nuMN_&OOCc@ND+x<;SJ13RcNQ@#9mYqdPJ
zRAQL_9kTx4Xp;p6^8W!89Dn}>_#c5TdDN+ZL$GnIYg+#BpFX<U23i%j9isjVc|{Iu

diff --git a/man/layer_tfsm.Rd b/man/layer_tfsm.Rd
index 32002781c..497de1425 100644
--- a/man/layer_tfsm.Rd
+++ b/man/layer_tfsm.Rd
@@ -59,8 +59,8 @@ model |> export_savedmodel("path/to/artifact")
 ## Output Type:
 ##   TensorSpec(shape=(None, 10), dtype=tf.float32, name=None)
 ## Captures:
-##   131061000739280: TensorSpec(shape=(), dtype=tf.resource, name=None)
-##   131061000735824: TensorSpec(shape=(), dtype=tf.resource, name=None)
+##   125092403276944: TensorSpec(shape=(), dtype=tf.resource, name=None)
+##   125092403280208: TensorSpec(shape=(), dtype=tf.resource, name=None)
 
 }\if{html}{\out{</div>}}
 
diff --git a/man/metric_mean_absolute_percentage_error.Rd b/man/metric_mean_absolute_percentage_error.Rd
index e50f040ac..930e47b4f 100644
--- a/man/metric_mean_absolute_percentage_error.Rd
+++ b/man/metric_mean_absolute_percentage_error.Rd
@@ -48,7 +48,7 @@ m$update_state(rbind(c(0, 1), c(0, 0)), rbind(c(1, 1), c(0, 0)))
 m$result()
 }\if{html}{\out{</div>}}
 
-\if{html}{\out{<div class="sourceCode">}}\preformatted{## tf.Tensor(250000000.0, shape=(), dtype=float32)
+\if{html}{\out{<div class="sourceCode">}}\preformatted{## tf.Tensor(2.5e+08, shape=(), dtype=float32)
 
 }\if{html}{\out{</div>}}
 
@@ -58,7 +58,7 @@ m$update_state(rbind(c(0, 1), c(0, 0)), rbind(c(1, 1), c(0, 0)),
 m$result()
 }\if{html}{\out{</div>}}
 
-\if{html}{\out{<div class="sourceCode">}}\preformatted{## tf.Tensor(500000000.0, shape=(), dtype=float32)
+\if{html}{\out{<div class="sourceCode">}}\preformatted{## tf.Tensor(5e+08, shape=(), dtype=float32)
 
 }\if{html}{\out{</div>}}
 
diff --git a/man/multi-assign.Rd b/man/multi-assign.Rd
index a39d27caa..cec8911f0 100644
--- a/man/multi-assign.Rd
+++ b/man/multi-assign.Rd
@@ -8,7 +8,7 @@ x \%<-\% value
 }
 \value{
 The right-hand-side argument, \code{value}, invisibly. This called
-primarily for it's side-effect of assigning symbols in the current frame.
+primarily for its side-effect of assigning symbols in the current frame.
 }
 \description{
 See \code{\link[zeallot]{\%<-\%}} for more details.
diff --git a/man/op_erf.Rd b/man/op_erf.Rd
index 600d886a5..f20512b68 100644
--- a/man/op_erf.Rd
+++ b/man/op_erf.Rd
@@ -20,7 +20,7 @@ Computes the error function of \code{x}, element-wise.
 op_erf(x)
 }\if{html}{\out{</div>}}
 
-\if{html}{\out{<div class="sourceCode">}}\preformatted{## tf.Tensor([-0.99997777 -0.9953222  -0.84270084  0.          0.84270084], shape=(5), dtype=float32)
+\if{html}{\out{<div class="sourceCode">}}\preformatted{## tf.Tensor([-0.999978   -0.99532235 -0.84270084  0.          0.84270084], shape=(5), dtype=float32)
 
 }\if{html}{\out{</div>}}
 
diff --git a/man/op_gelu.Rd b/man/op_gelu.Rd
index 256552bf7..9f6232110 100644
--- a/man/op_gelu.Rd
+++ b/man/op_gelu.Rd
@@ -27,7 +27,7 @@ where \code{P(X) ~ N(0, 1)}.
 op_gelu(x)
 }\if{html}{\out{</div>}}
 
-\if{html}{\out{<div class="sourceCode">}}\preformatted{## tf.Tensor([-0.15880796  0.          0.841192  ], shape=(3), dtype=float32)
+\if{html}{\out{<div class="sourceCode">}}\preformatted{## tf.Tensor([-0.158808  0.        0.841192], shape=(3), dtype=float32)
 
 }\if{html}{\out{</div>}}
 
diff --git a/man/plot.keras.src.models.model.Model.Rd b/man/plot.keras.src.models.model.Model.Rd
index 67073f551..830ec7893 100644
--- a/man/plot.keras.src.models.model.Model.Rd
+++ b/man/plot.keras.src.models.model.Model.Rd
@@ -57,22 +57,23 @@ have an \code{activation} property).}
 model is drawn on the default graphics device. Otherwise, a file is saved.}
 }
 \value{
-Nothing, called for it's side effects.
+Nothing, called for it side effects.
 }
 \description{
 Plot a Keras model
 }
 \section{Raises}{
  ValueError: if \code{plot(model)} is called before the model is
-built, unless a \verb{input_shape = } argument was supplied to
+built, unless an \verb{input_shape = } argument was supplied to
 \code{keras_model_sequential()}.
 }
 
 \section{Requirements}{
 
 This function requires pydot and graphviz.
+
 \code{pydot} is by default installed by \code{install_keras()}, but if you installed
-keras by other means, you can install \code{pydot} directly with :
+Keras by other means, you can install \code{pydot} directly with:
 
 \if{html}{\out{<div class="sourceCode r">}}\preformatted{reticulate::py_install("pydot", pip = TRUE)
 }\if{html}{\out{</div>}}
@@ -86,6 +87,11 @@ For example, on Ubuntu/Debian you can install with
 \if{html}{\out{<div class="sourceCode sh">}}\preformatted{sudo apt install graphviz
 }\if{html}{\out{</div>}}
 
+On macOS you can install graphviz using \code{brew}:
+
+\if{html}{\out{<div class="sourceCode sh">}}\preformatted{brew install graphviz
+}\if{html}{\out{</div>}}
+
 In a conda environment, you can install graphviz with:
 
 \if{html}{\out{<div class="sourceCode r">}}\preformatted{reticulate::conda_install(packages = "graphviz")

From 28ba55cae8b741a51a8dbe0a425c22a114926d07 Mon Sep 17 00:00:00 2001
From: Tomasz Kalinowski <kalinowskit@gmail.com>
Date: Thu, 25 Sep 2025 11:30:57 -0400
Subject: [PATCH 02/81] fix freeze_weights example, workaournd efficientnetb0
 bug

---
 R/freeze.R            | 7 ++++---
 man/freeze_weights.Rd | 7 ++++---
 man/layer_tfsm.Rd     | 4 ++--
 3 files changed, 10 insertions(+), 8 deletions(-)

diff --git a/R/freeze.R b/R/freeze.R
index 64d8cde72..44e6b8a1c 100644
--- a/R/freeze.R
+++ b/R/freeze.R
@@ -74,10 +74,11 @@
 #' conv_base
 #'
 #' # Freeze only layers of a certain type, e.g, BatchNorm layers
-#' batch_norm_layer_class_name <- class(layer_batch_normalization())[1]
-#' is_batch_norm_layer <- function(x) inherits(x, batch_norm_layer_class_name)
+#' # batch_norm_layer_class_name <- class(layer_batch_normalization())[1]
+#' # is_batch_norm_layer <- function(x) inherits(x, batch_norm_layer_class_name)
+#' is_batch_norm_layer <- function(x) inherits(x, keras$layers$BatchNormalization)
 #'
-#' model <- application_efficientnet_b0()
+#' model <- application_efficientnet_v2b0()
 #' freeze_weights(model, which = is_batch_norm_layer)
 #' # print(model)
 #'
diff --git a/man/freeze_weights.Rd b/man/freeze_weights.Rd
index aafd1ca1d..2308be67e 100644
--- a/man/freeze_weights.Rd
+++ b/man/freeze_weights.Rd
@@ -437,10 +437,11 @@ conv_base
 
 \if{html}{\out{<div class="sourceCode r">}}\preformatted{
 # Freeze only layers of a certain type, e.g, BatchNorm layers
-batch_norm_layer_class_name <- class(layer_batch_normalization())[1]
-is_batch_norm_layer <- function(x) inherits(x, batch_norm_layer_class_name)
+# batch_norm_layer_class_name <- class(layer_batch_normalization())[1]
+# is_batch_norm_layer <- function(x) inherits(x, batch_norm_layer_class_name)
+is_batch_norm_layer <- function(x) inherits(x, keras$layers$BatchNormalization)
 
-model <- application_efficientnet_b0()
+model <- application_efficientnet_v2b0()
 freeze_weights(model, which = is_batch_norm_layer)
 # print(model)
 
diff --git a/man/layer_tfsm.Rd b/man/layer_tfsm.Rd
index 497de1425..d97239208 100644
--- a/man/layer_tfsm.Rd
+++ b/man/layer_tfsm.Rd
@@ -59,8 +59,8 @@ model |> export_savedmodel("path/to/artifact")
 ## Output Type:
 ##   TensorSpec(shape=(None, 10), dtype=tf.float32, name=None)
 ## Captures:
-##   125092403276944: TensorSpec(shape=(), dtype=tf.resource, name=None)
-##   125092403280208: TensorSpec(shape=(), dtype=tf.resource, name=None)
+##   137054704163664: TensorSpec(shape=(), dtype=tf.resource, name=None)
+##   137054704159440: TensorSpec(shape=(), dtype=tf.resource, name=None)
 
 }\if{html}{\out{</div>}}
 

From b0f80c39727b4ec30537ccb0fcc5a8ca434c9de6 Mon Sep 17 00:00:00 2001
From: Tomasz Kalinowski <kalinowskit@gmail.com>
Date: Thu, 25 Sep 2025 11:44:03 -0400
Subject: [PATCH 03/81] `activation_elu()`: document alpha slope

---
 .tether/man/activation_elu.txt | 1 +
 R/activations.R                | 2 +-
 2 files changed, 2 insertions(+), 1 deletion(-)

diff --git a/.tether/man/activation_elu.txt b/.tether/man/activation_elu.txt
index d2be33c3f..13a23b8be 100644
--- a/.tether/man/activation_elu.txt
+++ b/.tether/man/activation_elu.txt
@@ -19,6 +19,7 @@ and the information that is propagated to the next layer.
 
 Args:
     x: Input tensor.
+    alpha: A scalar, slope of positive section. Defaults to `1.0`.
 
 Reference:
 
diff --git a/R/activations.R b/R/activations.R
index e674a2cf7..117da9ee1 100644
--- a/R/activations.R
+++ b/R/activations.R
@@ -25,7 +25,7 @@
 #' Input tensor.
 #'
 #' @param alpha
-#' Numeric. See description for details.
+#' A scalar, slope of positive section. Defaults to `1.0`.
 #'
 #' @returns A tensor, the result from applying the activation to the input tensor `x`.
 #' @export

From 45c335eab51d0c859df3a4165183ecdcd98819f9 Mon Sep 17 00:00:00 2001
From: Tomasz Kalinowski <kalinowskit@gmail.com>
Date: Thu, 25 Sep 2025 11:48:34 -0400
Subject: [PATCH 04/81] `callback_lambda()` docs: `logs` is keyword

---
 .tether/man/callback_lambda.txt | 5 +++--
 R/callbacks.R                   | 7 +++----
 2 files changed, 6 insertions(+), 6 deletions(-)

diff --git a/.tether/man/callback_lambda.txt b/.tether/man/callback_lambda.txt
index 44cdc12f8..030982b12 100644
--- a/.tether/man/callback_lambda.txt
+++ b/.tether/man/callback_lambda.txt
@@ -13,8 +13,8 @@ class LambdaCallback(keras.src.callbacks.callback.Callback)
  |        `epoch`, `logs`
  |      - `on_train_begin` and `on_train_end` expect one positional argument:
  |        `logs`
- |      - `on_train_batch_begin` and `on_train_batch_end` expect two positional
- |        arguments: `batch`, `logs`
+ |      - `on_train_batch_begin` and `on_train_batch_end` expect a positional
+ |        argument `batch` and a keyword argument `logs`
  |      - See `Callback` class definition for the full list of functions and their
  |        expected arguments.
  |
@@ -79,3 +79,4 @@ class LambdaCallback(keras.src.callbacks.callback.Callback)
  |  )
  |      Initialize self.  See help(type(self)) for accurate signature.
  |
+
diff --git a/R/callbacks.R b/R/callbacks.R
index dbd3e69ac..31a02a223 100644
--- a/R/callbacks.R
+++ b/R/callbacks.R
@@ -251,8 +251,8 @@ function (monitor = "val_loss", min_delta = 0L, patience = 0L,
 #'   `epoch`, `logs`
 #' - `on_train_begin` and `on_train_end` expect one positional argument:
 #'   `logs`
-#' - `on_train_batch_begin` and `on_train_batch_end` expect two positional
-#'   arguments: `batch`, `logs`
+#' - `on_train_batch_begin` and `on_train_batch_end` expect a positional
+#'   argument `batch` and a named argument `logs`
 #' - See `Callback` class definition for the full list of functions and their
 #'   expected arguments.
 #'
@@ -326,7 +326,7 @@ function (monitor = "val_loss", min_delta = 0L, patience = 0L,
 #' Any function in [`Callback()`] that you want to override by
 #' passing `function_name = function`. For example,
 #' `callback_lambda(.., on_train_end = train_end_fn)`. The custom function
-#' needs to have same arguments as the ones defined in [`Callback()`].
+#' needs to have the same arguments as the ones defined in [`Callback()`].
 #'
 #' @inherit callback_backup_and_restore return
 #' @export
@@ -995,4 +995,3 @@ normalize_callbacks_with_metrics <- function(view_metrics, initial_epoch, callba
 
   callbacks
 }
-

From 157df0edb81483b0f4f9024454463d4370dc12e5 Mon Sep 17 00:00:00 2001
From: Tomasz Kalinowski <kalinowskit@gmail.com>
Date: Thu, 25 Sep 2025 11:52:23 -0400
Subject: [PATCH 05/81] internal tether change: `EarlyStopping` uses
 `MonitorCallback`

---
 .tether/man/callback_early_stopping.txt | 3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)

diff --git a/.tether/man/callback_early_stopping.txt b/.tether/man/callback_early_stopping.txt
index f4e735529..286c062af 100644
--- a/.tether/man/callback_early_stopping.txt
+++ b/.tether/man/callback_early_stopping.txt
@@ -1,6 +1,6 @@
 Help on class EarlyStopping in module keras.src.callbacks.early_stopping:
 
-class EarlyStopping(keras.src.callbacks.callback.Callback)
+class EarlyStopping(keras.src.callbacks.monitor_callback.MonitorCallback)
  |  EarlyStopping(monitor='val_loss', min_delta=0, patience=0, verbose=0, mode='auto', baseline=None, restore_best_weights=False, start_from_epoch=0)
  |
  |  Stop training when a monitored metric has stopped improving.
@@ -60,6 +60,7 @@ class EarlyStopping(keras.src.callbacks.callback.Callback)
  |
  |  Method resolution order:
  |      EarlyStopping
+ |      keras.src.callbacks.monitor_callback.MonitorCallback
  |      keras.src.callbacks.callback.Callback
  |      builtins.object
  |

From 41baa8b183ee2bba081f2dbe577f88a9ee9c0e21 Mon Sep 17 00:00:00 2001
From: Tomasz Kalinowski <kalinowskit@gmail.com>
Date: Thu, 25 Sep 2025 11:54:53 -0400
Subject: [PATCH 06/81] `callback_model_checkpoint()`: clarify auto direction

---
 .tether/man/callback_model_checkpoint.txt | 8 ++++----
 R/callbacks.R                             | 5 ++---
 2 files changed, 6 insertions(+), 7 deletions(-)

diff --git a/.tether/man/callback_model_checkpoint.txt b/.tether/man/callback_model_checkpoint.txt
index c79cb11f0..13bbc3b37 100644
--- a/.tether/man/callback_model_checkpoint.txt
+++ b/.tether/man/callback_model_checkpoint.txt
@@ -1,6 +1,6 @@
 Help on class ModelCheckpoint in module keras.src.callbacks.model_checkpoint:
 
-class ModelCheckpoint(keras.src.callbacks.callback.Callback)
+class ModelCheckpoint(keras.src.callbacks.monitor_callback.MonitorCallback)
  |  ModelCheckpoint(filepath, monitor='val_loss', verbose=0, save_best_only=False, save_weights_only=False, mode='auto', save_freq='epoch', initial_value_threshold=None)
  |
  |  Callback to save the Keras model or model weights at some frequency.
@@ -95,9 +95,8 @@ class ModelCheckpoint(keras.src.callbacks.callback.Callback)
  |          decision to overwrite the current save file is made based on either
  |          the maximization or the minimization of the monitored quantity.
  |          For `val_acc`, this should be `"max"`, for `val_loss` this should be
- |          `"min"`, etc. In `"auto"` mode, the mode is set to `"max"` if the
- |          quantities monitored are `"acc"` or start with `"fmeasure"` and are
- |          set to `"min"` for the rest of the quantities.
+ |          `"min"`, etc. In `"auto"` mode, the direction is automatically
+ |          inferred from the name of the monitored quantity.
  |      save_weights_only: if `True`, then only the model's weights will be
  |          saved (`model.save_weights(filepath)`), else the full model is
  |          saved (`model.save(filepath)`).
@@ -116,6 +115,7 @@ class ModelCheckpoint(keras.src.callbacks.callback.Callback)
  |
  |  Method resolution order:
  |      ModelCheckpoint
+ |      keras.src.callbacks.monitor_callback.MonitorCallback
  |      keras.src.callbacks.callback.Callback
  |      builtins.object
  |
diff --git a/R/callbacks.R b/R/callbacks.R
index 31a02a223..98e722deb 100644
--- a/R/callbacks.R
+++ b/R/callbacks.R
@@ -534,9 +534,8 @@ function (schedule, verbose = 0L)
 #' decision to overwrite the current save file is made based on either
 #' the maximization or the minimization of the monitored quantity.
 #' For `val_acc`, this should be `"max"`, for `val_loss` this should be
-#' `"min"`, etc. In `"auto"` mode, the mode is set to `"max"` if the
-#' quantities monitored are `"acc"` or start with `"fmeasure"` and are
-#' set to `"min"` for the rest of the quantities.
+#' `"min"`, etc. In `"auto"` mode, the direction is automatically inferred from
+#' the name of the monitored quantity.
 #'
 #' @param save_weights_only
 #' if TRUE, then only the model's weights will be saved

From 5eb32281ebe6c872dee8618318561e76d8abe825 Mon Sep 17 00:00:00 2001
From: Tomasz Kalinowski <kalinowskit@gmail.com>
Date: Thu, 25 Sep 2025 11:56:06 -0400
Subject: [PATCH 07/81] internal tether change: `ReduceLROnPlateau` uses
 `MonitorCallback`

---
 .tether/man/callback_reduce_lr_on_plateau.txt | 3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)

diff --git a/.tether/man/callback_reduce_lr_on_plateau.txt b/.tether/man/callback_reduce_lr_on_plateau.txt
index 33451653d..ab2d51828 100644
--- a/.tether/man/callback_reduce_lr_on_plateau.txt
+++ b/.tether/man/callback_reduce_lr_on_plateau.txt
@@ -1,6 +1,6 @@
 Help on class ReduceLROnPlateau in module keras.src.callbacks.reduce_lr_on_plateau:
 
-class ReduceLROnPlateau(keras.src.callbacks.callback.Callback)
+class ReduceLROnPlateau(keras.src.callbacks.monitor_callback.MonitorCallback)
  |  ReduceLROnPlateau(monitor='val_loss', factor=0.1, patience=10, verbose=0, mode='auto', min_delta=0.0001, cooldown=0, min_lr=0.0, **kwargs)
  |
  |  Reduce learning rate when a metric has stopped improving.
@@ -39,6 +39,7 @@ class ReduceLROnPlateau(keras.src.callbacks.callback.Callback)
  |
  |  Method resolution order:
  |      ReduceLROnPlateau
+ |      keras.src.callbacks.monitor_callback.MonitorCallback
  |      keras.src.callbacks.callback.Callback
  |      builtins.object
  |

From ca152778a578e1cab6add2b87cfa13d50c0170a3 Mon Sep 17 00:00:00 2001
From: Tomasz Kalinowski <kalinowskit@gmail.com>
Date: Thu, 25 Sep 2025 12:49:15 -0400
Subject: [PATCH 08/81] `Layer`:  `add_weight()` gains
 `overwrite_with_gradient`; new `symbolic_call()` method

---
 .tether/man/Layer.txt | 12 +++++++++++-
 R/Layer.R             | 17 +++++++++++------
 2 files changed, 22 insertions(+), 7 deletions(-)

diff --git a/.tether/man/Layer.txt b/.tether/man/Layer.txt
index c0622237c..eee664282 100644
--- a/.tether/man/Layer.txt
+++ b/.tether/man/Layer.txt
@@ -1,6 +1,6 @@
 Help on class Layer in module keras.src.layers.layer:
 
-class Layer(keras.src.backend.tensorflow.layer.TFLayer, keras.src.ops.operation.Operation, keras.src.saving.keras_saveable.KerasSaveable)
+class Layer(keras.src.backend.tensorflow.layer.TFLayer, keras.src.ops.operation.Operation)
  |  Layer(*args, **kwargs)
  |
  |  This is the class from which all layers inherit.
@@ -243,6 +243,7 @@ class Layer(keras.src.backend.tensorflow.layer.TFLayer, keras.src.ops.operation.
  |    regularizer=None,
  |    constraint=None,
  |    aggregation='none',
+ |    overwrite_with_gradient=False,
  |    name=None
  |  )
  |      Add a weight variable to the layer.
@@ -274,6 +275,9 @@ class Layer(keras.src.backend.tensorflow.layer.TFLayer, keras.src.ops.operation.
  |              the type of multi-replica aggregation to be used for this
  |              variable when writing custom data parallel training loops.
  |              Defaults to `"none"`.
+ |          overwrite_with_gradient: Boolean, whether to overwrite the variable
+ |              with the computed gradient. This is useful for float8 training.
+ |              Defaults to `False`.
  |          name: String name of the variable. Useful for debugging purposes.
  |
  |  build(self, input_shape)
@@ -448,6 +452,12 @@ class Layer(keras.src.backend.tensorflow.layer.TFLayer, keras.src.ops.operation.
  |          ref_var.assign(value)
  |      ```
  |
+ |  symbolic_call(
+ |    self,
+ |    *args,
+ |    **kwargs
+ |  )
+ |
  |  ----------------------------------------------------------------------
  |  Static methods defined here:
  |
diff --git a/R/Layer.R b/R/Layer.R
index 28cbb9d41..889dc6056 100644
--- a/R/Layer.R
+++ b/R/Layer.R
@@ -234,6 +234,7 @@
 #'              regularizer = NULL,
 #'              constraint = NULL,
 #'              aggregation = 'none',
+#'              overwrite_with_gradient = FALSE,
 #'              name = NULL)
 #'   ```
 #'   Add a weight variable to the layer.
@@ -268,10 +269,12 @@
 #'       or string name of a built-in constraint.
 #'       Defaults to `NULL`.
 #'   * `aggregation`: Optional string, one of `NULL`, `"none"`, `"mean"`,
-#'      `"sum"` or `"only_first_replica"`. Annotates the variable with
-#'      the type of multi-replica aggregation to be used for this
-#'      variable when writing custom data parallel training loops.
-#'      Defaults to `"none"`.
+#'       `"sum"` or `"only_first_replica"`. Annotates the variable with
+#'       the type of multi-replica aggregation to be used for this
+#'       variable when writing custom data parallel training loops.
+#'       Defaults to `"none"`.
+#'   * `overwrite_with_gradient`: Boolean, whether to overwrite the variable with
+#'       the computed gradient. Useful for float8 training. Defaults to `FALSE`.
 #'   * `name`: String name of the variable. Useful for debugging purposes.
 #'
 #'   Returns:
@@ -327,6 +330,10 @@
 #'   ```
 #'
 #' * ```r
+#'   symbolic_call(...)
+#'   ```
+#'
+#' * ```r
 #'   count_params()
 #'   ```
 #'   Count the total number of scalars composing the weights.
@@ -692,5 +699,3 @@ function(classname,
 # ' @param .composing Bare Keras Layers (`layer_*` functions) conventionally
 # have `object` as the first argument, which allows users to instantiate
 # (`initialize`) and `call` one motion.
-
-

From f30b4528ad0b3f9569faa562a66e9d490eb364f4 Mon Sep 17 00:00:00 2001
From: Tomasz Kalinowski <kalinowskit@gmail.com>
Date: Thu, 25 Sep 2025 12:51:20 -0400
Subject: [PATCH 09/81] `layer_rescaling()`: doc per-channel scale/offset

---
 .tether/man/layer_rescaling.txt | 12 ++++++++++--
 R/layers-preprocessing.R        |  8 ++++++--
 2 files changed, 16 insertions(+), 4 deletions(-)

diff --git a/.tether/man/layer_rescaling.txt b/.tether/man/layer_rescaling.txt
index 5f3809581..140f33d12 100644
--- a/.tether/man/layer_rescaling.txt
+++ b/.tether/man/layer_rescaling.txt
@@ -24,8 +24,16 @@ class Rescaling(keras.src.layers.preprocessing.tf_data_layer.TFDataLayer)
  |  (independently of which backend you're using).
  |
  |  Args:
- |      scale: Float, the scale to apply to the inputs.
- |      offset: Float, the offset to apply to the inputs.
+ |      scale: Float, int, list, tuple or np.ndarray.
+ |          The scale to apply to the inputs.
+ |          If scalar, the same scale will be applied to
+ |          all features or channels of input. If a list, tuple or
+ |          1D array, the scaling is applied per channel.
+ |      offset: Float, int, list/tuple or numpy ndarray.
+ |          The offset to apply to the inputs.
+ |          If scalar, the same scale will be applied to
+ |          all features or channels of input. If a list, tuple or
+ |          1D array, the scaling is applied per channel.
  |      **kwargs: Base layer keyword arguments, such as `name` and `dtype`.
  |
  |  Method resolution order:
diff --git a/R/layers-preprocessing.R b/R/layers-preprocessing.R
index 7ffdefb3f..77d5313bc 100644
--- a/R/layers-preprocessing.R
+++ b/R/layers-preprocessing.R
@@ -1681,10 +1681,14 @@ function (object, height_factor, width_factor = NULL, fill_mode = "reflect",
 #' (independently of which backend you're using).
 #'
 #' @param scale
-#' Float, the scale to apply to the inputs.
+#' Numeric scalar, vector, or array. The scale to apply to the inputs. If
+#' scalar, the same scale is applied to every feature or channel; if a vector
+#' or array, scaling is applied per channel.
 #'
 #' @param offset
-#' Float, the offset to apply to the inputs.
+#' Numeric scalar, vector, or array. The offset to apply to the inputs. If
+#' scalar, the same offset is applied to every feature or channel; if a vector
+#' or array, the shift is applied per channel.
 #'
 #' @param ...
 #' Base layer keyword arguments, such as `name` and `dtype`.

From 30926e13b94660b626d650b0f024950fb651d73f Mon Sep 17 00:00:00 2001
From: Tomasz Kalinowski <kalinowskit@gmail.com>
Date: Thu, 25 Sep 2025 12:53:21 -0400
Subject: [PATCH 10/81] `layer_dense()`: add `lora_alpha`

---
 .tether/man/layer_dense.txt | 11 +++++++++--
 R/layers-core.R             |  8 ++++++++
 2 files changed, 17 insertions(+), 2 deletions(-)

diff --git a/.tether/man/layer_dense.txt b/.tether/man/layer_dense.txt
index b5d9b7b12..812851bc2 100644
--- a/.tether/man/layer_dense.txt
+++ b/.tether/man/layer_dense.txt
@@ -1,7 +1,7 @@
 Help on class Dense in module keras.src.layers.core.dense:
 
 class Dense(keras.src.layers.layer.Layer)
- |  Dense(units, activation=None, use_bias=True, kernel_initializer='glorot_uniform', bias_initializer='zeros', kernel_regularizer=None, bias_regularizer=None, activity_regularizer=None, kernel_constraint=None, bias_constraint=None, lora_rank=None, **kwargs)
+ |  Dense(units, activation=None, use_bias=True, kernel_initializer='glorot_uniform', bias_initializer='zeros', kernel_regularizer=None, bias_regularizer=None, activity_regularizer=None, kernel_constraint=None, bias_constraint=None, lora_rank=None, lora_alpha=None, **kwargs)
  |
  |  Just your regular densely-connected NN layer.
  |
@@ -46,6 +46,11 @@ class Dense(keras.src.layers.layer.Layer)
  |          computation cost of fine-tuning large dense layers.
  |          You can also enable LoRA on an existing
  |          `Dense` layer by calling `layer.enable_lora(rank)`.
+ |      lora_alpha: Optional integer. If set, this parameter scales the
+ |          low-rank adaptation delta (computed as the product of two lower-rank
+ |          trainable matrices) during the forward pass. The delta is scaled by
+ |          `lora_alpha / lora_rank`, allowing you to fine-tune the strength of
+ |          the LoRA adjustment independently of `lora_rank`.
  |
  |  Input shape:
  |      N-D tensor with shape: `(batch_size, ..., input_dim)`.
@@ -83,6 +88,7 @@ class Dense(keras.src.layers.layer.Layer)
  |    kernel_constraint=None,
  |    bias_constraint=None,
  |    lora_rank=None,
+ |    lora_alpha=None,
  |    **kwargs
  |  )
  |      Initialize self.  See help(type(self)) for accurate signature.
@@ -100,6 +106,7 @@ class Dense(keras.src.layers.layer.Layer)
  |  enable_lora(
  |    self,
  |    rank,
+ |    lora_alpha=None,
  |    a_initializer='he_uniform',
  |    b_initializer='zeros'
  |  )
@@ -127,7 +134,7 @@ class Dense(keras.src.layers.layer.Layer)
  |
  |  quantized_build(
  |    self,
- |    input_shape,
+ |    kernel_shape,
  |    mode
  |  )
  |
diff --git a/R/layers-core.R b/R/layers-core.R
index 231c271e7..dcde7fe96 100644
--- a/R/layers-core.R
+++ b/R/layers-core.R
@@ -35,6 +35,7 @@
 #' - ```r
 #'   enable_lora(
 #'     rank,
+#'     lora_alpha = NULL,
 #'     a_initializer = 'he_uniform',
 #'     b_initializer = 'zeros'
 #'   )
@@ -94,6 +95,11 @@
 #' You can also enable LoRA on an existing
 #' `Dense` layer by calling `layer$enable_lora(rank)`.
 #'
+#' @param lora_alpha
+#' Optional integer. Scales the low-rank adaptation delta during the forward
+#' pass. The delta is scaled by `lora_alpha / lora_rank`, letting you tune the
+#' LoRA adjustment strength independently of `lora_rank`.
+#'
 #' @param object
 #' Object to compose the layer with. A tensor, array, or sequential model.
 #'
@@ -119,9 +125,11 @@ function (object, units, activation = NULL, use_bias = TRUE,
     kernel_initializer = "glorot_uniform", bias_initializer = "zeros",
     kernel_regularizer = NULL, bias_regularizer = NULL, activity_regularizer = NULL,
     kernel_constraint = NULL, bias_constraint = NULL, lora_rank = NULL,
+    lora_alpha = NULL,
     ...)
 {
     args <- capture_args(list(units = as_integer, lora_rank = as_integer,
+        lora_alpha = as_integer,
         input_shape = normalize_shape, batch_size = as_integer,
         batch_input_shape = normalize_shape), ignore = "object")
     create_layer(keras$layers$Dense, object, args)

From 05d512faf9b914b52f2587bc352303ad56846de4 Mon Sep 17 00:00:00 2001
From: Tomasz Kalinowski <kalinowskit@gmail.com>
Date: Thu, 25 Sep 2025 12:55:59 -0400
Subject: [PATCH 11/81] `layer_embedding()`: add `lora_alpha` doc

---
 .tether/man/layer_embedding.txt | 13 +++++++++++--
 R/layers-core.R                 | 18 +++++++++++++++---
 2 files changed, 26 insertions(+), 5 deletions(-)

diff --git a/.tether/man/layer_embedding.txt b/.tether/man/layer_embedding.txt
index 82820af0d..00f8568d9 100644
--- a/.tether/man/layer_embedding.txt
+++ b/.tether/man/layer_embedding.txt
@@ -1,7 +1,7 @@
 Help on class Embedding in module keras.src.layers.core.embedding:
 
 class Embedding(keras.src.layers.layer.Layer)
- |  Embedding(input_dim, output_dim, embeddings_initializer='uniform', embeddings_regularizer=None, embeddings_constraint=None, mask_zero=False, weights=None, lora_rank=None, **kwargs)
+ |  Embedding(input_dim, output_dim, embeddings_initializer='uniform', embeddings_regularizer=None, embeddings_constraint=None, mask_zero=False, weights=None, lora_rank=None, lora_alpha=None, **kwargs)
  |
  |  Turns nonnegative integers (indexes) into dense vectors of fixed size.
  |
@@ -55,6 +55,11 @@ class Embedding(keras.src.layers.layer.Layer)
  |          computation cost of fine-tuning large embedding layers.
  |          You can also enable LoRA on an existing
  |          `Embedding` layer by calling `layer.enable_lora(rank)`.
+ |      lora_alpha: Optional integer. If set, this parameter scales the
+ |          low-rank adaptation delta (computed as the product of two lower-rank
+ |          trainable matrices) during the forward pass. The delta is scaled by
+ |          `lora_alpha / lora_rank`, allowing you to fine-tune the strength of
+ |          the LoRA adjustment independently of `lora_rank`.
  |
  |  Input shape:
  |      2D tensor with shape: `(batch_size, input_length)`.
@@ -85,6 +90,7 @@ class Embedding(keras.src.layers.layer.Layer)
  |    mask_zero=False,
  |    weights=None,
  |    lora_rank=None,
+ |    lora_alpha=None,
  |    **kwargs
  |  )
  |      Initialize self.  See help(type(self)) for accurate signature.
@@ -101,9 +107,12 @@ class Embedding(keras.src.layers.layer.Layer)
  |
  |  compute_output_shape(self, input_shape)
  |
+ |  compute_output_spec(self, inputs)
+ |
  |  enable_lora(
  |    self,
  |    rank,
+ |    lora_alpha=None,
  |    a_initializer='he_uniform',
  |    b_initializer='zeros'
  |  )
@@ -131,7 +140,7 @@ class Embedding(keras.src.layers.layer.Layer)
  |
  |  quantized_build(
  |    self,
- |    input_shape,
+ |    embeddings_shape,
  |    mode
  |  )
  |
diff --git a/R/layers-core.R b/R/layers-core.R
index dcde7fe96..140b06630 100644
--- a/R/layers-core.R
+++ b/R/layers-core.R
@@ -332,12 +332,17 @@ function (object, equation, output_shape, activation = NULL,
 #' - ```r
 #'   enable_lora(
 #'     rank,
+#'     lora_alpha = NULL,
 #'     a_initializer = 'he_uniform',
 #'     b_initializer = 'zeros'
 #'   )
 #'   ```
 #'
 #' - ```r
+#'   compute_output_spec(...)
+#'   ```
+#'
+#' - ```r
 #'   quantize(mode, type_check = TRUE)
 #'   ```
 #'
@@ -399,6 +404,11 @@ function (object, equation, output_shape, activation = NULL,
 #' You can also enable LoRA on an existing
 #' `Embedding` layer instance by calling `layer$enable_lora(rank)`.
 #'
+#' @param lora_alpha
+#' Optional integer. Scales the low-rank adaptation delta during the forward
+#' pass. The delta is scaled by `lora_alpha / lora_rank`, letting you tune the
+#' LoRA adjustment strength independently of `lora_rank`.
+#'
 #' @param object
 #' Object to compose the layer with. A tensor, array, or sequential model.
 #'
@@ -416,11 +426,13 @@ function (object, equation, output_shape, activation = NULL,
 layer_embedding <-
 function (object, input_dim, output_dim, embeddings_initializer = "uniform",
     embeddings_regularizer = NULL, embeddings_constraint = NULL,
-    mask_zero = FALSE, weights = NULL, lora_rank = NULL, ...)
+    mask_zero = FALSE, weights = NULL, lora_rank = NULL, lora_alpha = NULL,
+    ...)
 {
     args <- capture_args(list(input_dim = as_integer, output_dim = as_integer,
-        input_shape = normalize_shape, batch_size = as_integer,
-        batch_input_shape = normalize_shape, input_length = as_integer),
+        lora_rank = as_integer, lora_alpha = as_integer, input_shape = normalize_shape,
+        batch_size = as_integer, batch_input_shape = normalize_shape,
+        input_length = as_integer),
         ignore = "object")
     create_layer(keras$layers$Embedding, object, args)
 }

From 53e0d0f3f69f1e4439e70335cb0f5f2b64f3854f Mon Sep 17 00:00:00 2001
From: Tomasz Kalinowski <kalinowskit@gmail.com>
Date: Thu, 25 Sep 2025 12:59:19 -0400
Subject: [PATCH 12/81] `layer_layer_normalization()`: drop `rms_scaling`

---
 .tether/man/layer_layer_normalization.txt |  9 +--------
 R/layers-normalization.R                  | 10 +---------
 2 files changed, 2 insertions(+), 17 deletions(-)

diff --git a/.tether/man/layer_layer_normalization.txt b/.tether/man/layer_layer_normalization.txt
index e8ec5fb04..f941f9d93 100644
--- a/.tether/man/layer_layer_normalization.txt
+++ b/.tether/man/layer_layer_normalization.txt
@@ -1,7 +1,7 @@
 Help on class LayerNormalization in module keras.src.layers.normalization.layer_normalization:
 
 class LayerNormalization(keras.src.layers.layer.Layer)
- |  LayerNormalization(axis=-1, epsilon=0.001, center=True, scale=True, rms_scaling=False, beta_initializer='zeros', gamma_initializer='ones', beta_regularizer=None, gamma_regularizer=None, beta_constraint=None, gamma_constraint=None, **kwargs)
+ |  LayerNormalization(axis=-1, epsilon=0.001, center=True, scale=True, beta_initializer='zeros', gamma_initializer='ones', beta_regularizer=None, gamma_regularizer=None, beta_constraint=None, gamma_constraint=None, **kwargs)
  |
  |  Layer normalization layer (Ba et al., 2016).
  |
@@ -77,12 +77,6 @@ class LayerNormalization(keras.src.layers.layer.Layer)
  |          When the next layer is linear (also e.g. `nn.relu`), this can be
  |          disabled since the scaling will be done by the next layer.
  |          Defaults to `True`.
- |      rms_scaling: If True, `center` and `scale` are ignored, and the
- |          inputs are scaled by `gamma` and the inverse square root
- |          of the square of all inputs. This is an approximate and faster
- |          approach that avoids ever computing the mean of the input. Note that
- |          this *isn't* equivalent to the computation that the
- |          `keras.layers.RMSNormalization` layer performs.
  |      beta_initializer: Initializer for the beta weight. Defaults to zeros.
  |      gamma_initializer: Initializer for the gamma weight. Defaults to ones.
  |      beta_regularizer: Optional regularizer for the beta weight.
@@ -119,7 +113,6 @@ class LayerNormalization(keras.src.layers.layer.Layer)
  |    epsilon=0.001,
  |    center=True,
  |    scale=True,
- |    rms_scaling=False,
  |    beta_initializer='zeros',
  |    gamma_initializer='ones',
  |    beta_regularizer=None,
diff --git a/R/layers-normalization.R b/R/layers-normalization.R
index 4ccd03d0b..26bdee3e5 100644
--- a/R/layers-normalization.R
+++ b/R/layers-normalization.R
@@ -349,14 +349,6 @@ function (object, groups = 32L, axis = -1L, epsilon = 0.001,
 #' disabled since the scaling will be done by the next layer.
 #' Defaults to `TRUE`.
 #'
-#' @param rms_scaling
-#' If `TRUE`, `center` and `scale` are ignored, and the
-#' inputs are scaled by `gamma` and the inverse square root
-#' of the square of all inputs. This is an approximate and faster
-#' approach that avoids ever computing the mean of the input. Note that
-#' this *isn't* equivalent to the computation that the
-#' `layer_rms_normalization` layer performs.
-#'
 #' @param beta_initializer
 #' Initializer for the beta weight. Defaults to zeros.
 #'
@@ -395,7 +387,7 @@ function (object, groups = 32L, axis = -1L, epsilon = 0.001,
 #' @tether keras.layers.LayerNormalization
 layer_layer_normalization <-
 function (object, axis = -1L, epsilon = 0.001, center = TRUE,
-    scale = TRUE, rms_scaling = FALSE, beta_initializer = "zeros",
+    scale = TRUE, beta_initializer = "zeros",
     gamma_initializer = "ones", beta_regularizer = NULL, gamma_regularizer = NULL,
     beta_constraint = NULL, gamma_constraint = NULL, ...)
 {

From a7beb9299ee38fb03e4bac5e9b1ad0a5b3e489fa Mon Sep 17 00:00:00 2001
From: Tomasz Kalinowski <kalinowskit@gmail.com>
Date: Thu, 25 Sep 2025 13:24:08 -0400
Subject: [PATCH 13/81] `layer_average_pooling_2d()`: fix `padding="same"`
 shape

---
 .tether/man/layer_average_pooling_2d.txt | 2 +-
 R/layers-pooling.R                       | 2 +-
 2 files changed, 2 insertions(+), 2 deletions(-)

diff --git a/.tether/man/layer_average_pooling_2d.txt b/.tether/man/layer_average_pooling_2d.txt
index f1cf610f4..9ba3b156a 100644
--- a/.tether/man/layer_average_pooling_2d.txt
+++ b/.tether/man/layer_average_pooling_2d.txt
@@ -16,7 +16,7 @@ class AveragePooling2D(keras.src.layers.pooling.base_pooling.BasePooling)
  |  (when `input_shape >= pool_size`)
  |
  |  The resulting output shape when using the `"same"` padding option is:
- |  `output_shape = math.floor((input_shape - 1) / strides) + 1`
+ |  `output_shape = input_shape`
  |
  |  Args:
  |      pool_size: int or tuple of 2 integers, factors by which to downscale
diff --git a/R/layers-pooling.R b/R/layers-pooling.R
index a4ed58a72..0b36113a7 100644
--- a/R/layers-pooling.R
+++ b/R/layers-pooling.R
@@ -122,7 +122,7 @@ function (object, pool_size, strides = NULL, padding = "valid",
 #' (when `input_shape >= pool_size`)
 #'
 #' The resulting output shape when using the `"same"` padding option is:
-#' `output_shape = math.floor((input_shape - 1) / strides) + 1`
+#' `output_shape = input_shape`
 #'
 #' # Input Shape
 #' - If `data_format="channels_last"`:

From 9dfb9054080ae7fae932634c9c3b4eedabf74c87 Mon Sep 17 00:00:00 2001
From: Tomasz Kalinowski <kalinowskit@gmail.com>
Date: Thu, 25 Sep 2025 13:27:19 -0400
Subject: [PATCH 14/81] internal tether change: drop `Discretization.build()`

---
 .tether/man/layer_discretization.txt | 2 --
 1 file changed, 2 deletions(-)

diff --git a/.tether/man/layer_discretization.txt b/.tether/man/layer_discretization.txt
index 5bc0d285f..b0420b914 100644
--- a/.tether/man/layer_discretization.txt
+++ b/.tether/man/layer_discretization.txt
@@ -132,8 +132,6 @@ class Discretization(keras.src.layers.preprocessing.tf_data_layer.TFDataLayer)
  |              repeating dataset, you must specify the `steps` argument. This
  |              argument is not supported with array inputs or list inputs.
  |
- |  build(self, input_shape=None)
- |
  |  call(self, inputs)
  |
  |  compute_output_spec(self, inputs)

From 56ecd08afb07a201a96b5795841cd0b1d92074e4 Mon Sep 17 00:00:00 2001
From: Tomasz Kalinowski <kalinowskit@gmail.com>
Date: Thu, 25 Sep 2025 13:29:27 -0400
Subject: [PATCH 15/81] `layer_rnn()` example: drop `self$built`

---
 .tether/man/layer_rnn.txt | 1 -
 R/layers-rnn.R            | 1 -
 2 files changed, 2 deletions(-)

diff --git a/.tether/man/layer_rnn.txt b/.tether/man/layer_rnn.txt
index b61ea0711..d86c3c0c9 100644
--- a/.tether/man/layer_rnn.txt
+++ b/.tether/man/layer_rnn.txt
@@ -140,7 +140,6 @@ class RNN(keras.src.layers.layer.Layer)
  |              shape=(self.units, self.units),
  |              initializer='uniform',
  |              name='recurrent_kernel')
- |          self.built = True
  |
  |      def call(self, inputs, states):
  |          prev_output = states[0]
diff --git a/R/layers-rnn.R b/R/layers-rnn.R
index b1c283e4b..97a0d7790 100644
--- a/R/layers-rnn.R
+++ b/R/layers-rnn.R
@@ -1442,7 +1442,6 @@ function (units, activation = "tanh", recurrent_activation = "sigmoid",
 #'       initializer = 'uniform',
 #'       name = 'recurrent_kernel'
 #'     )
-#'     self$built <- TRUE
 #'   },
 #'
 #'   call = function(inputs, states) {

From a73c0f3fb422d67d488aabc1c60a1f3e4ac0f9bb Mon Sep 17 00:00:00 2001
From: Tomasz Kalinowski <kalinowskit@gmail.com>
Date: Thu, 25 Sep 2025 13:31:49 -0400
Subject: [PATCH 16/81] `loss_dice()`: use class-based example

---
 .tether/man/loss_dice.txt | 4 ++--
 R/losses.R                | 6 ++++--
 2 files changed, 6 insertions(+), 4 deletions(-)

diff --git a/.tether/man/loss_dice.txt b/.tether/man/loss_dice.txt
index 9ee7231ca..1aad5efd1 100644
--- a/.tether/man/loss_dice.txt
+++ b/.tether/man/loss_dice.txt
@@ -38,12 +38,12 @@ class Dice(LossFunctionWrapper)
  |  >>> y_pred = [[[[0.0], [1.0]], [[0.0], [1.0]]],
  |  ...           [[[0.4], [0.0]], [[0.0], [0.9]]]]
  |  >>> axis = (1, 2, 3)
- |  >>> loss = keras.losses.dice(y_true, y_pred, axis=axis)
+ |  >>> loss = keras.losses.Dice(axis=axis, reduction=None)(y_true, y_pred)
  |  >>> assert loss.shape == (2,)
  |  >>> loss
  |  array([0.5, 0.75757575], shape=(2,), dtype=float32)
  |
- |  >>> loss = keras.losses.dice(y_true, y_pred)
+ |  >>> loss = keras.losses.Dice()(y_true, y_pred)
  |  >>> assert loss.shape == ()
  |  >>> loss
  |  array(0.6164384, shape=(), dtype=float32)
diff --git a/R/losses.R b/R/losses.R
index 2c6754049..66f61d81a 100644
--- a/R/losses.R
+++ b/R/losses.R
@@ -820,12 +820,14 @@ function (y_true, y_pred, axis = -1L, ..., reduction = "sum_over_batch_size",
 #'                   1,   0, 1, 0.9), dim = c(2, 2, 2, 1))
 #'
 #' axis <- c(2, 3, 4)
-#' loss <- loss_dice(y_true, y_pred, axis = axis)
+#' loss_fn <- loss_dice(axis = axis, reduction = NULL)
+#' loss <- loss_fn(y_true, y_pred)
 #' stopifnot(shape(loss) == shape(2))
 #' loss
 #'
 #'
-#' loss = loss_dice(y_true, y_pred)
+#' loss_fn <- loss_dice()
+#' loss <- loss_fn(y_true, y_pred)
 #' stopifnot(shape(loss) == shape())
 #' loss
 #' ```

From eea32a517abd8f51e68aa4cf0d829f544682ca8a Mon Sep 17 00:00:00 2001
From: Tomasz Kalinowski <kalinowskit@gmail.com>
Date: Thu, 25 Sep 2025 13:40:52 -0400
Subject: [PATCH 17/81] `loss_sparse_categorical_crossentropy()`: update
 examples, add `axis`

---
 .../loss_sparse_categorical_crossentropy.txt  |  9 ++++++---
 R/losses.R                                    | 19 ++++++++-----------
 2 files changed, 14 insertions(+), 14 deletions(-)

diff --git a/.tether/man/loss_sparse_categorical_crossentropy.txt b/.tether/man/loss_sparse_categorical_crossentropy.txt
index 4f84895f1..c4b397d8c 100644
--- a/.tether/man/loss_sparse_categorical_crossentropy.txt
+++ b/.tether/man/loss_sparse_categorical_crossentropy.txt
@@ -1,7 +1,7 @@
 Help on class SparseCategoricalCrossentropy in module keras.src.losses.losses:
 
 class SparseCategoricalCrossentropy(LossFunctionWrapper)
- |  SparseCategoricalCrossentropy(from_logits=False, ignore_class=None, reduction='sum_over_batch_size', name='sparse_categorical_crossentropy', dtype=None)
+ |  SparseCategoricalCrossentropy(from_logits=False, ignore_class=None, reduction='sum_over_batch_size', axis=-1, name='sparse_categorical_crossentropy', dtype=None)
  |
  |  Computes the crossentropy loss between the labels and predictions.
  |
@@ -28,6 +28,8 @@ class SparseCategoricalCrossentropy(LossFunctionWrapper)
  |          sample size, and `"mean_with_sample_weight"` sums the loss and
  |          divides by the sum of the sample weights. `"none"` and `None`
  |          perform no aggregation. Defaults to `"sum_over_batch_size"`.
+ |      axis: The axis along which to compute crossentropy (the features
+ |          axis). Defaults to `-1`.
  |      name: Optional name for the loss instance.
  |      dtype: The dtype of the loss's computations. Defaults to `None`, which
  |          means using `keras.backend.floatx()`. `keras.backend.floatx()` is a
@@ -37,8 +39,8 @@ class SparseCategoricalCrossentropy(LossFunctionWrapper)
  |
  |  Examples:
  |
- |  >>> y_true = [1, 2]
- |  >>> y_pred = [[0.05, 0.95, 0], [0.1, 0.8, 0.1]]
+ |  >>> y_true = np.array([1, 2])
+ |  >>> y_pred = np.array([[0.05, 0.95, 0], [0.1, 0.8, 0.1]])
  |  >>> # Using 'auto'/'sum_over_batch_size' reduction type.
  |  >>> scce = keras.losses.SparseCategoricalCrossentropy()
  |  >>> scce(y_true, y_pred)
@@ -81,6 +83,7 @@ class SparseCategoricalCrossentropy(LossFunctionWrapper)
  |    from_logits=False,
  |    ignore_class=None,
  |    reduction='sum_over_batch_size',
+ |    axis=-1,
  |    name='sparse_categorical_crossentropy',
  |    dtype=None
  |  )
diff --git a/R/losses.R b/R/losses.R
index 66f61d81a..e8570e984 100644
--- a/R/losses.R
+++ b/R/losses.R
@@ -1541,37 +1541,34 @@ function (y_true, y_pred, ..., reduction = "sum_over_batch_size",
 #'
 #' # Examples
 #' ```{r}
-#' y_true <- c(1, 2)
-#' y_pred <- rbind(c(0.05, 0.95, 0), c(0.1, 0.8, 0.1))
+#' y_true <- op_array(c(1L, 2L))
+#' y_pred <- op_array(rbind(c(0.05, 0.95, 0), c(0.1, 0.8, 0.1)))
 #' loss <- loss_sparse_categorical_crossentropy(y_true, y_pred)
 #' loss
 #' ```
 #' ```{r}
-#' y_true <- c(1, 2)
-#' y_pred <- rbind(c(0.05, 0.95, 0), c(0.1, 0.8, 0.1))
+#' y_true <- op_array(c(1L, 2L))
+#' y_pred <- op_array(rbind(c(0.05, 0.95, 0), c(0.1, 0.8, 0.1)))
 #' # Using 'auto'/'sum_over_batch_size' reduction type.
 #' scce <- loss_sparse_categorical_crossentropy()
-#' scce(op_array(y_true), op_array(y_pred))
-#' # 1.177
+#' scce(y_true, y_pred)
 #' ```
 #'
 #' ```{r}
 #' # Calling with 'sample_weight'.
-#' scce(op_array(y_true), op_array(y_pred), sample_weight = op_array(c(0.3, 0.7)))
+#' scce(y_true, y_pred, sample_weight = op_array(c(0.3, 0.7)))
 #' ```
 #'
 #' ```{r}
 #' # Using 'sum' reduction type.
 #' scce <- loss_sparse_categorical_crossentropy(reduction="sum")
-#' scce(op_array(y_true), op_array(y_pred))
-#' # 2.354
+#' scce(y_true, y_pred)
 #' ```
 #'
 #' ```{r}
 #' # Using 'none' reduction type.
 #' scce <- loss_sparse_categorical_crossentropy(reduction=NULL)
-#' scce(op_array(y_true), op_array(y_pred))
-#' # array([0.0513, 2.303], dtype=float32)
+#' scce(y_true, y_pred)
 #' ```
 #'
 #' Usage with the `compile()` API:

From 30c97ee7848ddb85a25b9f16f3391104a0c5e7a3 Mon Sep 17 00:00:00 2001
From: Tomasz Kalinowski <kalinowskit@gmail.com>
Date: Thu, 25 Sep 2025 13:55:20 -0400
Subject: [PATCH 18/81] `layer_conv*_transpose()`: add `output_padding` arg

---
 .tether/man/layer_conv_1d_transpose.txt | 14 ++++++--
 .tether/man/layer_conv_2d_transpose.txt | 20 +++++++++--
 .tether/man/layer_conv_3d_transpose.txt | 19 +++++++++--
 R/layers-convolutional.R                | 45 ++++++++++++++++---------
 4 files changed, 74 insertions(+), 24 deletions(-)

diff --git a/.tether/man/layer_conv_1d_transpose.txt b/.tether/man/layer_conv_1d_transpose.txt
index e6f51c257..6b5266904 100644
--- a/.tether/man/layer_conv_1d_transpose.txt
+++ b/.tether/man/layer_conv_1d_transpose.txt
@@ -1,7 +1,7 @@
 Help on class Conv1DTranspose in module keras.src.layers.convolutional.conv1d_transpose:
 
 class Conv1DTranspose(keras.src.layers.convolutional.base_conv_transpose.BaseConvTranspose)
- |  Conv1DTranspose(filters, kernel_size, strides=1, padding='valid', data_format=None, dilation_rate=1, activation=None, use_bias=True, kernel_initializer='glorot_uniform', bias_initializer='zeros', kernel_regularizer=None, bias_regularizer=None, activity_regularizer=None, kernel_constraint=None, bias_constraint=None, **kwargs)
+ |  Conv1DTranspose(filters, kernel_size, strides=1, padding='valid', output_padding=None, data_format=None, dilation_rate=1, activation=None, use_bias=True, kernel_initializer='glorot_uniform', bias_initializer='zeros', kernel_regularizer=None, bias_regularizer=None, activity_regularizer=None, kernel_constraint=None, bias_constraint=None, **kwargs)
  |
  |  1D transposed convolution layer.
  |
@@ -23,6 +23,10 @@ class Conv1DTranspose(keras.src.layers.convolutional.base_conv_transpose.BaseCon
  |          `"valid"` means no padding. `"same"` results in padding evenly to
  |          the left/right or up/down of the input such that output has the same
  |          height/width dimension as the input.
+ |      output_padding: An integer tuple/list of 1 integer specifying the
+ |          amount of padding along the time dimension of the output tensor.
+ |          The amount of output padding must be lower than the stride.
+ |          If set to `None` (default), the output shape is inferred.
  |      data_format: string, either `"channels_last"` or `"channels_first"`.
  |          The ordering of the dimensions in the inputs. `"channels_last"`
  |          corresponds to inputs with shape `(batch, steps, features)`
@@ -30,8 +34,11 @@ class Conv1DTranspose(keras.src.layers.convolutional.base_conv_transpose.BaseCon
  |          `(batch, features, steps)`. It defaults to the `image_data_format`
  |          value found in your Keras config file at `~/.keras/keras.json`.
  |          If you never set it, then it will be `"channels_last"`.
- |      dilation_rate: int or tuple/list of 1 integers, specifying the dilation
- |          rate to use for dilated transposed convolution.
+ |      dilation_rate: An integer tuple/list of 1 integer, specifying
+ |          the dilation rate to use for dilated convolution.
+ |          Currently, specifying a `dilation_rate` value != 1 is
+ |          incompatible with specifying a stride value != 1.
+ |          Also dilation rate larger than 1 is not currently supported.
  |      activation: Activation function. If `None`, no activation is applied.
  |      use_bias: bool, if `True`, bias will be added to the output.
  |      kernel_initializer: Initializer for the convolution kernel. If `None`,
@@ -104,6 +111,7 @@ class Conv1DTranspose(keras.src.layers.convolutional.base_conv_transpose.BaseCon
  |    kernel_size,
  |    strides=1,
  |    padding='valid',
+ |    output_padding=None,
  |    data_format=None,
  |    dilation_rate=1,
  |    activation=None,
diff --git a/.tether/man/layer_conv_2d_transpose.txt b/.tether/man/layer_conv_2d_transpose.txt
index b540c565c..07e4151f9 100644
--- a/.tether/man/layer_conv_2d_transpose.txt
+++ b/.tether/man/layer_conv_2d_transpose.txt
@@ -1,7 +1,7 @@
 Help on class Conv2DTranspose in module keras.src.layers.convolutional.conv2d_transpose:
 
 class Conv2DTranspose(keras.src.layers.convolutional.base_conv_transpose.BaseConvTranspose)
- |  Conv2DTranspose(filters, kernel_size, strides=(1, 1), padding='valid', data_format=None, dilation_rate=(1, 1), activation=None, use_bias=True, kernel_initializer='glorot_uniform', bias_initializer='zeros', kernel_regularizer=None, bias_regularizer=None, activity_regularizer=None, kernel_constraint=None, bias_constraint=None, **kwargs)
+ |  Conv2DTranspose(filters, kernel_size, strides=(1, 1), padding='valid', output_padding=None, data_format=None, dilation_rate=(1, 1), activation=None, use_bias=True, kernel_initializer='glorot_uniform', bias_initializer='zeros', kernel_regularizer=None, bias_regularizer=None, activity_regularizer=None, kernel_constraint=None, bias_constraint=None, **kwargs)
  |
  |  2D transposed convolution layer.
  |
@@ -23,6 +23,14 @@ class Conv2DTranspose(keras.src.layers.convolutional.base_conv_transpose.BaseCon
  |          `"valid"` means no padding. `"same"` results in padding evenly to
  |          the left/right or up/down of the input. When `padding="same"` and
  |          `strides=1`, the output has the same size as the input.
+ |      output_padding: An integer or tuple/list of 2 integers,
+ |          specifying the amount of padding along the height and width
+ |          of the output tensor.
+ |          Can be a single integer to specify the same value for all
+ |          spatial dimensions.
+ |          The amount of output padding along a given dimension must be
+ |          lower than the stride along that same dimension.
+ |          If set to `None` (default), the output shape is inferred.
  |      data_format: string, either `"channels_last"` or `"channels_first"`.
  |          The ordering of the dimensions in the inputs. `"channels_last"`
  |          corresponds to inputs with shape
@@ -32,8 +40,13 @@ class Conv2DTranspose(keras.src.layers.convolutional.base_conv_transpose.BaseCon
  |          `image_data_format` value found in your Keras config file at
  |          `~/.keras/keras.json`. If you never set it, then it will be
  |          `"channels_last"`.
- |      dilation_rate: int or tuple/list of 1 integers, specifying the dilation
- |          rate to use for dilated transposed convolution.
+ |       dilation_rate: An integer or tuple/list of 2 integers,
+ |          specifying the dilation rate for
+ |          all spatial dimensions for dilated convolution.
+ |          Specifying different dilation rates
+ |          for different dimensions is not supported.
+ |          Currently, specifying any `dilation_rate` value != 1 is
+ |          incompatible with specifying any stride value != 1.
  |      activation: Activation function. If `None`, no activation is applied.
  |      use_bias: bool, if `True`, bias will be added to the output.
  |      kernel_initializer: Initializer for the convolution kernel. If `None`,
@@ -106,6 +119,7 @@ class Conv2DTranspose(keras.src.layers.convolutional.base_conv_transpose.BaseCon
  |    kernel_size,
  |    strides=(1, 1),
  |    padding='valid',
+ |    output_padding=None,
  |    data_format=None,
  |    dilation_rate=(1, 1),
  |    activation=None,
diff --git a/.tether/man/layer_conv_3d_transpose.txt b/.tether/man/layer_conv_3d_transpose.txt
index 35f84966f..9cdb0d464 100644
--- a/.tether/man/layer_conv_3d_transpose.txt
+++ b/.tether/man/layer_conv_3d_transpose.txt
@@ -1,7 +1,7 @@
 Help on class Conv3DTranspose in module keras.src.layers.convolutional.conv3d_transpose:
 
 class Conv3DTranspose(keras.src.layers.convolutional.base_conv_transpose.BaseConvTranspose)
- |  Conv3DTranspose(filters, kernel_size, strides=(1, 1, 1), padding='valid', data_format=None, dilation_rate=(1, 1, 1), activation=None, use_bias=True, kernel_initializer='glorot_uniform', bias_initializer='zeros', kernel_regularizer=None, bias_regularizer=None, activity_regularizer=None, kernel_constraint=None, bias_constraint=None, **kwargs)
+ |  Conv3DTranspose(filters, kernel_size, strides=(1, 1, 1), padding='valid', data_format=None, output_padding=None, dilation_rate=(1, 1, 1), activation=None, use_bias=True, kernel_initializer='glorot_uniform', bias_initializer='zeros', kernel_regularizer=None, bias_regularizer=None, activity_regularizer=None, kernel_constraint=None, bias_constraint=None, **kwargs)
  |
  |  3D transposed convolution layer.
  |
@@ -23,6 +23,14 @@ class Conv3DTranspose(keras.src.layers.convolutional.base_conv_transpose.BaseCon
  |          `"valid"` means no padding. `"same"` results in padding evenly to
  |          the left/right or up/down of the input. When `padding="same"` and
  |          `strides=1`, the output has the same size as the input.
+ |       output_padding: An integer or tuple/list of 3 integers,
+ |          specifying the amount of padding along the depth, height, and
+ |          width.
+ |          Can be a single integer to specify the same value for all
+ |          spatial dimensions.
+ |          The amount of output padding along a given dimension must be
+ |          lower than the stride along that same dimension.
+ |          If set to `None` (default), the output shape is inferred.
  |      data_format: string, either `"channels_last"` or `"channels_first"`.
  |          The ordering of the dimensions in the inputs. `"channels_last"`
  |          corresponds to inputs with shape
@@ -32,8 +40,12 @@ class Conv3DTranspose(keras.src.layers.convolutional.base_conv_transpose.BaseCon
  |          It defaults to the `image_data_format` value found in your Keras
  |          config file at `~/.keras/keras.json`. If you never set it, then it
  |          will be `"channels_last"`.
- |      dilation_rate: int or tuple/list of 1 integers, specifying the dilation
- |          rate to use for dilated transposed convolution.
+ |      dilation_rate: an integer or tuple/list of 3 integers, specifying
+ |          the dilation rate to use for dilated convolution.
+ |          Can be a single integer to specify the same value for
+ |          all spatial dimensions.
+ |          Currently, specifying any `dilation_rate` value != 1 is
+ |          incompatible with specifying any stride value != 1.
  |      activation: Activation function. If `None`, no activation is applied.
  |      use_bias: bool, if `True`, bias will be added to the output.
  |      kernel_initializer: Initializer for the convolution kernel. If `None`,
@@ -112,6 +124,7 @@ class Conv3DTranspose(keras.src.layers.convolutional.base_conv_transpose.BaseCon
  |    strides=(1, 1, 1),
  |    padding='valid',
  |    data_format=None,
+ |    output_padding=None,
  |    dilation_rate=(1, 1, 1),
  |    activation=None,
  |    use_bias=True,
diff --git a/R/layers-convolutional.R b/R/layers-convolutional.R
index 64ff5410f..f419e51f0 100644
--- a/R/layers-convolutional.R
+++ b/R/layers-convolutional.R
@@ -204,6 +204,10 @@ function (object, filters, kernel_size, strides = 1L, padding = "valid",
 #' the left/right or up/down of the input such that output has the same
 #' height/width dimension as the input.
 #'
+#' @param output_padding
+#' Scalar integer. Amount of padding to add to the output length. Must be less
+#' than the stride. When `NULL` (default) the output size is inferred.
+#'
 #' @param data_format
 #' string, either `"channels_last"` or `"channels_first"`.
 #' The ordering of the dimensions in the inputs. `"channels_last"`
@@ -214,8 +218,9 @@ function (object, filters, kernel_size, strides = 1L, padding = "valid",
 #' If you never set it, then it will be `"channels_last"`.
 #'
 #' @param dilation_rate
-#' int or list of 1 integers, specifying the dilation
-#' rate to use for dilated transposed convolution.
+#' Scalar integer. Specifies the dilation rate. Values other
+#' than 1 currently require `strides = 1` and rates greater than 1 are not
+#' supported.
 #'
 #' @param activation
 #' Activation function. If `NULL`, no activation is applied.
@@ -267,13 +272,14 @@ function (object, filters, kernel_size, strides = 1L, padding = "valid",
 #' @tether keras.layers.Conv1DTranspose
 layer_conv_1d_transpose <-
 function (object, filters, kernel_size, strides = 1L, padding = "valid",
-    data_format = NULL, dilation_rate = 1L, activation = NULL,
+    output_padding = NULL, data_format = NULL, dilation_rate = 1L, activation = NULL,
     use_bias = TRUE, kernel_initializer = "glorot_uniform", bias_initializer = "zeros",
     kernel_regularizer = NULL, bias_regularizer = NULL, activity_regularizer = NULL,
     kernel_constraint = NULL, bias_constraint = NULL, ...)
 {
     args <- capture_args(list(filters = as_integer, kernel_size = as_integer_tuple,
-        strides = as_integer_tuple, dilation_rate = as_integer_tuple,
+        strides = as_integer_tuple, output_padding = as_integer_tuple,
+        dilation_rate = as_integer_tuple,
         input_shape = normalize_shape, batch_size = as_integer,
         batch_input_shape = normalize_shape), ignore = "object")
     create_layer(keras$layers$Conv1DTranspose, object, args)
@@ -342,6 +348,11 @@ function (object, filters, kernel_size, strides = 1L, padding = "valid",
 #' the left/right or up/down of the input. When `padding="same"` and
 #' `strides=1`, the output has the same size as the input.
 #'
+#' @param output_padding
+#' Scalar integer or vector of two integers. Amount of padding to add to the output
+#' depth, height, and width. Each element must be smaller than the corresponding
+#' stride. When `NULL` (default) the output size is inferred.
+#'
 #' @param data_format
 #' string, either `"channels_last"` or `"channels_first"`.
 #' The ordering of the dimensions in the inputs. `"channels_last"`
@@ -501,8 +512,9 @@ function (object, filters, kernel_size, strides = list(1L, 1L),
 #' `"channels_last"`.
 #'
 #' @param dilation_rate
-#' int or list of 1 integers, specifying the dilation
-#' rate to use for dilated transposed convolution.
+#' Scalar integer or vector of 2 integers specifying the dilation rate. Values
+#' other than 1 require `strides = 1`; different rates per dimension are not
+#' supported.
 #'
 #' @param activation
 #' Activation function. If `NULL`, no activation is applied.
@@ -554,14 +566,15 @@ function (object, filters, kernel_size, strides = list(1L, 1L),
 #' @tether keras.layers.Conv2DTranspose
 layer_conv_2d_transpose <-
 function (object, filters, kernel_size, strides = list(1L, 1L),
-    padding = "valid", data_format = NULL, dilation_rate = list(
-        1L, 1L), activation = NULL, use_bias = TRUE, kernel_initializer = "glorot_uniform",
+    padding = "valid", output_padding = NULL, data_format = NULL,
+    dilation_rate = list(1L, 1L), activation = NULL, use_bias = TRUE, kernel_initializer = "glorot_uniform",
     bias_initializer = "zeros", kernel_regularizer = NULL, bias_regularizer = NULL,
     activity_regularizer = NULL, kernel_constraint = NULL, bias_constraint = NULL,
     ...)
 {
     args <- capture_args(list(filters = as_integer, kernel_size = as_integer_tuple,
-        strides = as_integer_tuple, dilation_rate = as_integer_tuple,
+        strides = as_integer_tuple, output_padding = as_integer_tuple,
+        dilation_rate = as_integer_tuple,
         input_shape = normalize_shape, batch_size = as_integer,
         batch_input_shape = normalize_shape), ignore = "object")
     create_layer(keras$layers$Conv2DTranspose, object, args)
@@ -639,7 +652,7 @@ function (object, filters, kernel_size, strides = list(1L, 1L),
 #' will be `"channels_last"`.
 #'
 #' @param dilation_rate
-#' int or list of 3 integers, specifying the dilation
+#' int or vector of 3 ints, specifying the dilation
 #' rate to use for dilated convolution.
 #'
 #' @param groups
@@ -790,8 +803,9 @@ function (object, filters, kernel_size, strides = list(1L, 1L,
 #' will be `"channels_last"`.
 #'
 #' @param dilation_rate
-#' int or list of 1 integers, specifying the dilation
-#' rate to use for dilated transposed convolution.
+#' Scalar integer or vector of 3 integers specifying the dilation rate. Values
+#' other than 1 require `strides = 1`; different rates per dimension are not
+#' supported.
 #'
 #' @param activation
 #' Activation function. If `NULL`, no activation is applied.
@@ -843,14 +857,15 @@ function (object, filters, kernel_size, strides = list(1L, 1L,
 #' @tether keras.layers.Conv3DTranspose
 layer_conv_3d_transpose <-
 function (object, filters, kernel_size, strides = list(1L, 1L,
-    1L), padding = "valid", data_format = NULL, dilation_rate = list(
-    1L, 1L, 1L), activation = NULL, use_bias = TRUE, kernel_initializer = "glorot_uniform",
+    1L), padding = "valid", output_padding = NULL, data_format = NULL,
+    dilation_rate = list(1L, 1L, 1L), activation = NULL, use_bias = TRUE, kernel_initializer = "glorot_uniform",
     bias_initializer = "zeros", kernel_regularizer = NULL, bias_regularizer = NULL,
     activity_regularizer = NULL, kernel_constraint = NULL, bias_constraint = NULL,
     ...)
 {
     args <- capture_args(list(filters = as_integer, kernel_size = as_integer_tuple,
-        strides = as_integer_tuple, dilation_rate = as_integer_tuple,
+        strides = as_integer_tuple, output_padding = as_integer_tuple,
+        dilation_rate = as_integer_tuple,
         input_shape = normalize_shape, batch_size = as_integer,
         batch_input_shape = normalize_shape), ignore = "object")
     create_layer(keras$layers$Conv3DTranspose, object, args)

From 30bf4222b82bb2714628166a4581822f4c0022e4 Mon Sep 17 00:00:00 2001
From: Tomasz Kalinowski <kalinowskit@gmail.com>
Date: Thu, 25 Sep 2025 14:00:07 -0400
Subject: [PATCH 19/81] `op_conv_transpose()`: default `strides=1`

---
 .tether/man/op_conv_transpose.txt | 3 ++-
 R/ops.R                           | 4 ++--
 2 files changed, 4 insertions(+), 3 deletions(-)

diff --git a/.tether/man/op_conv_transpose.txt b/.tether/man/op_conv_transpose.txt
index dab311e05..2e8863baa 100644
--- a/.tether/man/op_conv_transpose.txt
+++ b/.tether/man/op_conv_transpose.txt
@@ -2,7 +2,7 @@ __signature__
 keras.ops.conv_transpose(
   inputs,
   kernel,
-  strides,
+  strides=1,
   padding='valid',
   output_padding=None,
   data_format=None,
@@ -51,3 +51,4 @@ Args:
 
 Returns:
     A tensor of rank N+2, the result of the conv operation.
+
diff --git a/R/ops.R b/R/ops.R
index bec8d9854..0c6c68776 100644
--- a/R/ops.R
+++ b/R/ops.R
@@ -1784,7 +1784,7 @@ function (target, output, from_logits = FALSE, axis = -1L)
 #' int or int tuple/list of `len(inputs_spatial_shape)`,
 #' specifying the strides of the convolution along each spatial
 #' dimension. If `strides` is int, then every spatial dimension shares
-#' the same `strides`.
+#' the same `strides`. Defaults to `1`.
 #'
 #' @param padding
 #' string, either `"valid"` or `"same"`. `"valid"` means no
@@ -1886,7 +1886,7 @@ function (inputs, kernel, strides = 1L, padding = "valid", data_format = NULL,
 #  + <https://www.tensorflow.org/api_docs/python/tf/keras/ops/conv_transpose>
 #' @tether keras.ops.conv_transpose
 op_conv_transpose <-
-function (inputs, kernel, strides, padding = "valid", output_padding = NULL,
+function (inputs, kernel, strides = 1L, padding = "valid", output_padding = NULL,
     data_format = NULL, dilation_rate = 1L)
 {
     args <- capture_args(list(strides = as_integer, output_padding = as_integer,

From 53e93d60111a5176c7c3d400659de420d8bdd049 Mon Sep 17 00:00:00 2001
From: Tomasz Kalinowski <kalinowskit@gmail.com>
Date: Thu, 25 Sep 2025 14:07:47 -0400
Subject: [PATCH 20/81] internal tether changes to `op_vectorized_map()`

---
 .tether/man/op_vectorized_map.txt | 7 ++++---
 1 file changed, 4 insertions(+), 3 deletions(-)

diff --git a/.tether/man/op_vectorized_map.txt b/.tether/man/op_vectorized_map.txt
index 3af8b4df0..da1de681c 100644
--- a/.tether/man/op_vectorized_map.txt
+++ b/.tether/man/op_vectorized_map.txt
@@ -7,18 +7,18 @@ Schematically, `vectorized_map` implements the following,
 in the case of a single tensor input `elements`:
 
 ```python
-def vectorized_map(function, elements)
+def vectorized_map(function, elements):
     outputs = []
     for e in elements:
         outputs.append(function(e))
-    return stack(outputs)
+    return np.stack(outputs)
 ```
 
 In the case of an iterable of tensors `elements`,
 it implements the following:
 
 ```python
-def vectorized_map(function, elements)
+def vectorized_map(function, elements):
     batch_size = elements[0].shape[0]
     outputs = []
     for index in range(batch_size):
@@ -28,3 +28,4 @@ def vectorized_map(function, elements)
 
 In this case, `function` is expected to take as input
 a single list of tensor arguments.
+

From 0d943c4946b5384d509eb7903946ca124f73924f Mon Sep 17 00:00:00 2001
From: Tomasz Kalinowski <kalinowskit@gmail.com>
Date: Thu, 25 Sep 2025 14:12:09 -0400
Subject: [PATCH 21/81] `op_dot_product_attention()`: add
 `attn_logits_soft_cap` arg

---
 .tether/man/op_dot_product_attention.txt | 6 +++++-
 R/ops-nn.R                               | 6 +++++-
 2 files changed, 10 insertions(+), 2 deletions(-)

diff --git a/.tether/man/op_dot_product_attention.txt b/.tether/man/op_dot_product_attention.txt
index 52de124bf..14d58473d 100644
--- a/.tether/man/op_dot_product_attention.txt
+++ b/.tether/man/op_dot_product_attention.txt
@@ -7,7 +7,8 @@ keras.ops.dot_product_attention(
   mask=None,
   scale=None,
   is_causal=False,
-  flash_attention=None
+  flash_attention=None,
+  attn_logits_soft_cap=None
 )
 __doc__
 Scaled dot product attention function.
@@ -47,6 +48,9 @@ Args:
         attempt to use flash attention if the required conditions are met.
         Typically, the inputs must be in float16 and bfloat16 dtype and the
         input layout requirements may vary depending on the backend.
+    attn_logits_soft_cap: The value limit for maximum value of the
+        attention logits before the softmax function is applied. This is
+        only supported in JAX TPU backend. Defaults to None.
 
 Returns:
     An array of the attention output with the same shape of `query`.
diff --git a/R/ops-nn.R b/R/ops-nn.R
index bb7db4f21..36768bf0a 100644
--- a/R/ops-nn.R
+++ b/R/ops-nn.R
@@ -214,13 +214,17 @@ ops$psnr(x1, x2, max_val)
 #' Typically, the inputs must be in float16 and bfloat16 dtype and the
 #' input layout requirements may vary depending on the backend.
 #'
+#' @param attn_logits_soft_cap
+#' Optional numeric cap on the attention logits before softmax. Only supported
+#' on the JAX TPU backend.
+#'
 #' @export
 #' @tether keras.ops.dot_product_attention
 #' @family nn ops
 #' @family ops
 op_dot_product_attention <-
 function (query, key, value, bias = NULL, mask = NULL, scale = NULL,
-          is_causal = FALSE, flash_attention = NULL)
+          is_causal = FALSE, flash_attention = NULL, attn_logits_soft_cap = NULL)
 {
   args <- capture_args()
   do.call(ops$dot_product_attention, args)

From 4398732e8d15ecc3ec40217d217eeffe93c9b840 Mon Sep 17 00:00:00 2001
From: Tomasz Kalinowski <kalinowskit@gmail.com>
Date: Thu, 25 Sep 2025 14:15:28 -0400
Subject: [PATCH 22/81] `op_einsum()`: mention extra kwargs

---
 .tether/man/op_einsum.txt | 6 +++++-
 R/ops.R                   | 4 ++--
 2 files changed, 7 insertions(+), 3 deletions(-)

diff --git a/.tether/man/op_einsum.txt b/.tether/man/op_einsum.txt
index 9591c8a54..90b9221d5 100644
--- a/.tether/man/op_einsum.txt
+++ b/.tether/man/op_einsum.txt
@@ -1,5 +1,9 @@
 __signature__
-keras.ops.einsum(subscripts, *operands)
+keras.ops.einsum(
+  subscripts,
+  *operands,
+  **kwargs
+)
 __doc__
 Evaluates the Einstein summation convention on the operands.
 
diff --git a/R/ops.R b/R/ops.R
index 0c6c68776..956504615 100644
--- a/R/ops.R
+++ b/R/ops.R
@@ -4668,14 +4668,14 @@ ops$dot(x1, x2)
 #' output form.
 #'
 #' @param ...
-#' The operands to compute the Einstein sum of.
+#' The operands to compute the Einstein sum of as unnamed arguments.
+#' Additional named arguments are forwarded to the underlying backend.
 #'
 #' @export
 #' @family numpy ops
 #' @family ops
 #' @seealso
 #' + <https://keras.io/api/ops/numpy#einsum-function>
-#  + <https://www.tensorflow.org/api_docs/python/tf/keras/ops/einsum>
 #' @tether keras.ops.einsum
 op_einsum <-
 function (subscripts, ...)

From e59d45a3eb6e6fac9231ec38e939804b4d8f50d9 Mon Sep 17 00:00:00 2001
From: Tomasz Kalinowski <kalinowskit@gmail.com>
Date: Thu, 25 Sep 2025 14:18:41 -0400
Subject: [PATCH 23/81] `op_rms_normalization()`: default `scale` to `NULL`

---
 .tether/man/op_rms_normalization.txt | 17 ++++++++---------
 R/ops-nn.R                           | 17 ++++++++---------
 2 files changed, 16 insertions(+), 18 deletions(-)

diff --git a/.tether/man/op_rms_normalization.txt b/.tether/man/op_rms_normalization.txt
index 321afb838..d300b9e3a 100644
--- a/.tether/man/op_rms_normalization.txt
+++ b/.tether/man/op_rms_normalization.txt
@@ -1,7 +1,7 @@
 __signature__
 keras.ops.rms_normalization(
   x,
-  scale=1,
+  scale=None,
   axis=-1,
   epsilon=None
 )
@@ -18,20 +18,19 @@ It is defined as `rms_normalization(x) = x * rsqrt(mean(square(x))) * scale`
 
 Args:
     x: Input tensor.
-    axis: The axis or axes along which to perform normalization.
-        Default to -1.
     scale: Optional scaling factor for the normalization.
-    epsilon: A lower bound value for the norm.
-        Defaults to `backend.epsilon()`.
+    axis: The axis or axes along which to perform normalization. Defaults
+        to `-1`.
+    epsilon: A lower bound value for the norm. Defaults to
+        `backend.epsilon()`.
 
 Returns:
     The normalized array.
 
 Example:
 
->>> x = np.random.rand(1, 10)
->>> x_norm = keras.ops.rms_normalization(x, (10,))
->>> print(x_norm)
+>>> x = keras.random.normal((1, 10))
+>>> keras.ops.rms_normalization(x)
 array([[0.69384296, 0.94444374, 0.16551171, 0.05749961, 1.11008865,
-    0.52475186, 1.57686807, 1.69893307, 1.27292764, 0.30819128]])
+        0.52475186, 1.57686807, 1.69893307, 1.27292764, 0.30819128]])
 
diff --git a/R/ops-nn.R b/R/ops-nn.R
index 36768bf0a..dbac3510b 100644
--- a/R/ops-nn.R
+++ b/R/ops-nn.R
@@ -639,10 +639,10 @@ op_polar <-
 #'
 #' # Examples
 #'
-#' ```python
-#' x <- random_uniform(c(1, 10))
-#' x_norm <- op_rms_normalization(x, scale = 10)
-#' x_norm
+#' ```{r}
+#' x <- random_normal(c(1, 10))
+#' op_rms_normalization(x)
+#' op_rms_normalization(x, scale = 10)
 #' ```
 #'
 #' @returns
@@ -651,13 +651,12 @@ op_polar <-
 #' @param x
 #' Input tensor.
 #'
-#' @param axis
-#' The axis or axes along which to perform normalization.
-#' Default to -1.
-#'
 #' @param scale
 #' Optional scaling factor for the normalization.
 #'
+#' @param axis
+#' The axis or axes along which to perform normalization. Defaults to `-1`.
+#'
 #' @param epsilon
 #' A lower bound value for the norm.
 #' Defaults to `config_epsilon()`.
@@ -667,7 +666,7 @@ op_polar <-
 #' @family nn ops
 #' @family ops
 op_rms_normalization <-
-function (x, scale = 1L, axis = -1L, epsilon = NULL)
+function (x, scale = NULL, axis = -1L, epsilon = NULL)
 {
     args <- capture_args(list(axis = as_axis))
     do.call(keras$ops$rms_normalization, args)

From 866246292a63a088a05155af5023452c2e7ba883 Mon Sep 17 00:00:00 2001
From: Tomasz Kalinowski <kalinowskit@gmail.com>
Date: Thu, 25 Sep 2025 14:21:27 -0400
Subject: [PATCH 24/81] `layer_einsum_dense()`: add `lora_alpha` arg

---
 .tether/man/layer_einsum_dense.txt | 11 +++++++++--
 R/layers-core.R                    | 11 ++++++++---
 2 files changed, 17 insertions(+), 5 deletions(-)

diff --git a/.tether/man/layer_einsum_dense.txt b/.tether/man/layer_einsum_dense.txt
index 178c0c90b..1347c4be0 100644
--- a/.tether/man/layer_einsum_dense.txt
+++ b/.tether/man/layer_einsum_dense.txt
@@ -1,7 +1,7 @@
 Help on class EinsumDense in module keras.src.layers.core.einsum_dense:
 
 class EinsumDense(keras.src.layers.layer.Layer)
- |  EinsumDense(equation, output_shape, activation=None, bias_axes=None, kernel_initializer='glorot_uniform', bias_initializer='zeros', kernel_regularizer=None, bias_regularizer=None, kernel_constraint=None, bias_constraint=None, lora_rank=None, **kwargs)
+ |  EinsumDense(equation, output_shape, activation=None, bias_axes=None, kernel_initializer='glorot_uniform', bias_initializer='zeros', kernel_regularizer=None, bias_regularizer=None, kernel_constraint=None, bias_constraint=None, lora_rank=None, lora_alpha=None, **kwargs)
  |
  |  A layer that uses `einsum` as the backing computation.
  |
@@ -43,6 +43,11 @@ class EinsumDense(keras.src.layers.layer.Layer)
  |          computation cost of fine-tuning large dense layers.
  |          You can also enable LoRA on an existing
  |          `EinsumDense` layer by calling `layer.enable_lora(rank)`.
+ |       lora_alpha: Optional integer. If set, this parameter scales the
+ |          low-rank adaptation delta (computed as the product of two lower-rank
+ |          trainable matrices) during the forward pass. The delta is scaled by
+ |          `lora_alpha / lora_rank`, allowing you to fine-tune the strength of
+ |          the LoRA adjustment independently of `lora_rank`.
  |      **kwargs: Base layer keyword arguments, such as `name` and `dtype`.
  |
  |  Examples:
@@ -122,6 +127,7 @@ class EinsumDense(keras.src.layers.layer.Layer)
  |    kernel_constraint=None,
  |    bias_constraint=None,
  |    lora_rank=None,
+ |    lora_alpha=None,
  |    **kwargs
  |  )
  |      Initialize self.  See help(type(self)) for accurate signature.
@@ -139,6 +145,7 @@ class EinsumDense(keras.src.layers.layer.Layer)
  |  enable_lora(
  |    self,
  |    rank,
+ |    lora_alpha=None,
  |    a_initializer='he_uniform',
  |    b_initializer='zeros'
  |  )
@@ -166,7 +173,7 @@ class EinsumDense(keras.src.layers.layer.Layer)
  |
  |  quantized_build(
  |    self,
- |    input_shape,
+ |    kernel_shape,
  |    mode
  |  )
  |
diff --git a/R/layers-core.R b/R/layers-core.R
index 140b06630..1cb2928b6 100644
--- a/R/layers-core.R
+++ b/R/layers-core.R
@@ -269,6 +269,11 @@ function (object, units, activation = NULL, use_bias = TRUE,
 #' You can also enable LoRA on an existing
 #' `EinsumDense` layer by calling `layer$enable_lora(rank)`.
 #'
+#' @param lora_alpha
+#' Optional integer. Scales the low-rank adaptation delta during the forward
+#' pass. The delta is scaled by `lora_alpha / lora_rank`, letting you tune the
+#' LoRA adjustment strength independently of `lora_rank`.
+#'
 #' @param ...
 #' Base layer keyword arguments, such as `name` and `dtype`.
 #'
@@ -288,10 +293,10 @@ function (object, equation, output_shape, activation = NULL,
     bias_axes = NULL, kernel_initializer = "glorot_uniform",
     bias_initializer = "zeros", kernel_regularizer = NULL, bias_regularizer = NULL,
     kernel_constraint = NULL, bias_constraint = NULL, lora_rank = NULL,
-    ...)
+    lora_alpha = NULL, ...)
 {
-    args <- capture_args(list(lora_rank = as_integer, input_shape = normalize_shape,
-        batch_size = as_integer, batch_input_shape = normalize_shape,
+    args <- capture_args(list(lora_rank = as_integer, lora_alpha = as_integer,
+        input_shape = normalize_shape, batch_size = as_integer, batch_input_shape = normalize_shape,
         output_shape = normalize_shape), ignore = "object")
     create_layer(keras$layers$EinsumDense, object, args)
 }

From e64bdf7fa5da3692753ea239616cb423630c0e9f Mon Sep 17 00:00:00 2001
From: Tomasz Kalinowski <kalinowskit@gmail.com>
Date: Thu, 25 Sep 2025 14:23:40 -0400
Subject: [PATCH 25/81] `layer_torch_module_wrapper()`: add `output_shape` arg

---
 .tether/man/layer_torch_module_wrapper.txt | 7 ++++++-
 R/layers-backend-wrappers.R                | 8 ++++++--
 2 files changed, 12 insertions(+), 3 deletions(-)

diff --git a/.tether/man/layer_torch_module_wrapper.txt b/.tether/man/layer_torch_module_wrapper.txt
index b63b0377f..9d50c1956 100644
--- a/.tether/man/layer_torch_module_wrapper.txt
+++ b/.tether/man/layer_torch_module_wrapper.txt
@@ -1,7 +1,7 @@
 Help on class TorchModuleWrapper in module keras.src.utils.torch_utils:
 
 class TorchModuleWrapper(keras.src.layers.layer.Layer)
- |  TorchModuleWrapper(module, name=None, **kwargs)
+ |  TorchModuleWrapper(module, name=None, output_shape=None, **kwargs)
  |
  |  Torch module wrapper layer.
  |
@@ -17,6 +17,8 @@ class TorchModuleWrapper(keras.src.layers.layer.Layer)
  |          instance, then its parameters must be initialized before
  |          passing the instance to `TorchModuleWrapper` (e.g. by calling
  |          it once).
+ |      output_shape :The shape of the output of this layer. It helps Keras
+ |          perform automatic shape inference.
  |      name: The name of the layer (string).
  |
  |  Example:
@@ -88,6 +90,7 @@ class TorchModuleWrapper(keras.src.layers.layer.Layer)
  |    self,
  |    module,
  |    name=None,
+ |    output_shape=None,
  |    **kwargs
  |  )
  |      Initialize self.  See help(type(self)) for accurate signature.
@@ -99,6 +102,8 @@ class TorchModuleWrapper(keras.src.layers.layer.Layer)
  |    **kwargs
  |  )
  |
+ |  compute_output_shape(self, input_shape)
+ |
  |  get_config(self)
  |      Returns the config of the object.
  |
diff --git a/R/layers-backend-wrappers.R b/R/layers-backend-wrappers.R
index 32375b099..76e4e660e 100644
--- a/R/layers-backend-wrappers.R
+++ b/R/layers-backend-wrappers.R
@@ -76,6 +76,9 @@
 #' passing the instance to `layer_torch_module_wrapper` (e.g. by calling
 #' it once).
 #'
+#' @param output_shape
+#' Shape of the output from this layer. Helps Keras infer shapes.
+#'
 #' @param name
 #' The name of the layer (string).
 #'
@@ -91,10 +94,11 @@
 #' @family layers
 #' @tether keras.layers.TorchModuleWrapper
 layer_torch_module_wrapper <-
-function (object, module, name = NULL, ...)
+function (object, module, output_shape = NULL, name = NULL, ...)
 {
     args <- capture_args(list(input_shape = normalize_shape,
-        batch_size = as_integer, batch_input_shape = normalize_shape),
+        batch_size = as_integer, batch_input_shape = normalize_shape,
+        output_shape = normalize_shape),
         ignore = "object")
     create_layer(keras$layers$TorchModuleWrapper, object, args)
 }

From 9247e2e5b52319ec20bb17c557a96e459b575e83 Mon Sep 17 00:00:00 2001
From: Tomasz Kalinowski <kalinowskit@gmail.com>
Date: Thu, 25 Sep 2025 14:25:56 -0400
Subject: [PATCH 26/81]  `optimizer_lion()` docs: grammar tweak

---
 .tether/man/optimizer_lion.txt | 4 ++--
 R/optimizers.R                 | 4 ++--
 2 files changed, 4 insertions(+), 4 deletions(-)

diff --git a/.tether/man/optimizer_lion.txt b/.tether/man/optimizer_lion.txt
index 987adadbd..fa0dc9138 100644
--- a/.tether/man/optimizer_lion.txt
+++ b/.tether/man/optimizer_lion.txt
@@ -7,13 +7,13 @@ class Lion(keras.src.optimizers.optimizer.Optimizer)
  |
  |  The Lion optimizer is a stochastic-gradient-descent method that uses the
  |  sign operator to control the magnitude of the update, unlike other adaptive
- |  optimizers such as Adam that rely on second-order moments. This make
+ |  optimizers such as Adam that rely on second-order moments. This makes
  |  Lion more memory-efficient as it only keeps track of the momentum. According
  |  to the authors (see reference), its performance gain over Adam grows with
  |  the batch size. Because the update of Lion is produced through the sign
  |  operation, resulting in a larger norm, a suitable learning rate for Lion is
  |  typically 3-10x smaller than that for AdamW. The weight decay for Lion
- |  should be in turn 3-10x larger than that for AdamW to maintain a
+ |  should in turn be 3-10x larger than that for AdamW to maintain a
  |  similar strength (lr * wd).
  |
  |  Args:
diff --git a/R/optimizers.R b/R/optimizers.R
index f10ed7250..f480e2992 100644
--- a/R/optimizers.R
+++ b/R/optimizers.R
@@ -1044,13 +1044,13 @@ function (learning_rate = 0.001, beta_1 = 0.9, beta_2 = 0.999,
 #' @description
 #' The Lion optimizer is a stochastic-gradient-descent method that uses the
 #' sign operator to control the magnitude of the update, unlike other adaptive
-#' optimizers such as Adam that rely on second-order moments. This make
+#' optimizers such as Adam that rely on second-order moments. This makes
 #' Lion more memory-efficient as it only keeps track of the momentum. According
 #' to the authors (see reference), its performance gain over Adam grows with
 #' the batch size. Because the update of Lion is produced through the sign
 #' operation, resulting in a larger norm, a suitable learning rate for Lion is
 #' typically 3-10x smaller than that for AdamW. The weight decay for Lion
-#' should be in turn 3-10x larger than that for AdamW to maintain a
+#' should in turn be 3-10x larger than that for AdamW to maintain a
 #' similar strength (lr * wd).
 #'
 #' # References

From ba8400d2b701b9a1db976c90c6a554073befcb10 Mon Sep 17 00:00:00 2001
From: Tomasz Kalinowski <kalinowskit@gmail.com>
Date: Thu, 25 Sep 2025 14:27:51 -0400
Subject: [PATCH 27/81] internal tether change: `to_categorical` example
 reshape

---
 .tether/man/to_categorical.txt | 3 +--
 1 file changed, 1 insertion(+), 2 deletions(-)

diff --git a/.tether/man/to_categorical.txt b/.tether/man/to_categorical.txt
index ca6dbad25..bb45b450c 100644
--- a/.tether/man/to_categorical.txt
+++ b/.tether/man/to_categorical.txt
@@ -27,8 +27,7 @@ Example:
 >>> b = np.array([.9, .04, .03, .03,
 ...               .3, .45, .15, .13,
 ...               .04, .01, .94, .05,
-...               .12, .21, .5, .17],
-...               shape=[4, 4])
+...               .12, .21, .5, .17]).reshape(4,4)
 >>> loss = keras.ops.categorical_crossentropy(a, b)
 >>> print(np.around(loss, 5))
 [0.10536 0.82807 0.1011  1.77196]

From 50281075e34b93508b8ddefcd41171da631f531e Mon Sep 17 00:00:00 2001
From: Tomasz Kalinowski <kalinowskit@gmail.com>
Date: Thu, 25 Sep 2025 14:50:08 -0400
Subject: [PATCH 28/81] `load_model_weights()`: document sharded manifests

---
 .tether/man/load_model_weights.txt | 45 ++++++++++++++++++++----------
 R/model-persistence.R              | 20 ++++++++++---
 2 files changed, 47 insertions(+), 18 deletions(-)

diff --git a/.tether/man/load_model_weights.txt b/.tether/man/load_model_weights.txt
index 9a57d484a..724a79df6 100644
--- a/.tether/man/load_model_weights.txt
+++ b/.tether/man/load_model_weights.txt
@@ -6,27 +6,44 @@ keras.Model.load_weights(
   **kwargs
 )
 __doc__
-Load weights from a file saved via `save_weights()`.
+Load the weights from a single file or sharded files.
 
-Weights are loaded based on the network's
-topology. This means the architecture should be the same as when the
-weights were saved. Note that layers that don't have weights are not
-taken into account in the topological ordering, so adding or removing
-layers is fine as long as they don't have weights.
+Weights are loaded based on the network's topology. This means the
+architecture should be the same as when the weights were saved. Note
+that layers that don't have weights are not taken into account in the
+topological ordering, so adding or removing layers is fine as long as
+they don't have weights.
 
 **Partial weight loading**
 
 If you have modified your model, for instance by adding a new layer
-(with weights) or by changing the shape of the weights of a layer,
-you can choose to ignore errors and continue loading
-by setting `skip_mismatch=True`. In this case any layer with
-mismatching weights will be skipped. A warning will be displayed
-for each skipped layer.
+(with weights) or by changing the shape of the weights of a layer, you
+can choose to ignore errors and continue loading by setting
+`skip_mismatch=True`. In this case any layer with mismatching weights
+will be skipped. A warning will be displayed for each skipped layer.
+
+**Sharding**
+
+When loading sharded weights, it is important to specify `filepath` that
+ends with `*.weights.json` which is used as the configuration file.
+Additionally, the sharded files `*_xxxxx.weights.h5` must be in the same
+directory as the configuration file.
 
 Args:
-    filepath: String, path to the weights file to load.
-        It can either be a `.weights.h5` file
-        or a legacy `.h5` weights file.
+    filepath: `str` or `pathlib.Path` object. Path where the weights
+        will be saved.  When sharding, the filepath must end in
+        `.weights.json`.
     skip_mismatch: Boolean, whether to skip loading of layers where
         there is a mismatch in the number of weights, or a mismatch in
         the shape of the weights.
+
+Example:
+
+```python
+# Load the weights in a single file.
+model.load_weights("model.weights.h5")
+
+# Load the weights in sharded files.
+model.load_weights("model.weights.json")
+```
+
diff --git a/R/model-persistence.R b/R/model-persistence.R
index 370728f8c..802c5c218 100644
--- a/R/model-persistence.R
+++ b/R/model-persistence.R
@@ -196,7 +196,7 @@ function (model, filepath, overwrite = FALSE)
 }
 
 
-#' Load weights from a file saved via `save_model_weights()`.
+#' Load the weights from a single file or sharded files.
 #'
 #' @description
 #' Weights are loaded based on the network's
@@ -205,6 +205,12 @@ function (model, filepath, overwrite = FALSE)
 #' taken into account in the topological ordering, so adding or removing
 #' layers is fine as long as they don't have weights.
 #'
+#' **Sharding**
+#'
+#' When loading sharded weights, specify a `filepath` ending in
+#' `".weights.json"` (the configuration file), with the corresponding shard files
+#' (`*_xxxxx.weights.h5`) located alongside it.
+#'
 #' **Partial weight loading**
 #'
 #' If you have modified your model, for instance by adding a new layer
@@ -214,10 +220,16 @@ function (model, filepath, overwrite = FALSE)
 #' mismatching weights will be skipped. A warning will be displayed
 #' for each skipped layer.
 #'
+#' # Examples
+#' ```r
+#' model |> load_model_weights("model.weights.h5")
+#' model |> load_model_weights("model.weights.json")
+#' ```
+#'
 #' @param filepath
-#' String, path to the weights file to load.
-#' It can either be a `.weights.h5` file
-#' or a legacy `.h5` weights file.
+#' Path or path-like object to the weights. Accepts `.weights.h5`, legacy `.h5`,
+#' or sharded weights through a `.weights.json` manifest sitting alongside the
+#' shard files (`*_xxxxx.weights.h5`).
 #'
 #' @param skip_mismatch
 #' Boolean, whether to skip loading of layers where

From 060028e56b8d7594f86cc7b08c9143d56ca87502 Mon Sep 17 00:00:00 2001
From: Tomasz Kalinowski <kalinowskit@gmail.com>
Date: Thu, 25 Sep 2025 15:01:05 -0400
Subject: [PATCH 29/81] `metric_binary_focal_crossentropy()`: align docs

---
 .../man/metric_binary_focal_crossentropy.txt    | 17 +++++++++++++++--
 R/metrics.R                                     | 11 +++++++++--
 2 files changed, 24 insertions(+), 4 deletions(-)

diff --git a/.tether/man/metric_binary_focal_crossentropy.txt b/.tether/man/metric_binary_focal_crossentropy.txt
index ad3ddb52d..0aa9546df 100644
--- a/.tether/man/metric_binary_focal_crossentropy.txt
+++ b/.tether/man/metric_binary_focal_crossentropy.txt
@@ -53,8 +53,21 @@ Example:
 
 >>> y_true = [[0, 1], [0, 0]]
 >>> y_pred = [[0.6, 0.4], [0.4, 0.6]]
->>> loss = keras.losses.binary_focal_crossentropy(
+>>> # In this instance, the first sample in the second batch is the
+>>> # 'easier' example.
+>>> focal_loss = keras.losses.binary_focal_crossentropy(
 ...        y_true, y_pred, gamma=2)
 >>> assert loss.shape == (2,)
->>> loss
+>>> focal_loss
 array([0.330, 0.206], dtype=float32)
+>>> # Compare with binary_crossentropy
+>>> bce_loss = keras.losses.binary_focal_crossentropy(
+...        y_true, y_pred)
+>>> bce_loss
+array([0.916, 0.714], dtype=float32)
+>>> # Binary focal crossentropy loss attributes more importance to the
+>>> # harder example which results in a higher loss for the first batch
+>>> # when normalized by binary cross entropy loss
+>>> focal_loss/bce_loss
+array([0.360, 0.289]
+
diff --git a/R/metrics.R b/R/metrics.R
index afdd5fb05..1697ed363 100644
--- a/R/metrics.R
+++ b/R/metrics.R
@@ -23,8 +23,15 @@
 #' ```{r}
 #' y_true <- rbind(c(0, 1), c(0, 0))
 #' y_pred <- rbind(c(0.6, 0.4), c(0.4, 0.6))
-#' loss <- loss_binary_focal_crossentropy(y_true, y_pred, gamma=2)
-#' loss
+#' focal_loss <- loss_binary_focal_crossentropy(y_true, y_pred, gamma = 2)
+#' focal_loss
+#'
+#' # Compare with binary crossentropy.
+#' # Binary focal crossentropy emphasises harder examples, yielding a larger
+#' # relative loss where the model struggles.
+#' bce_loss <- loss_binary_crossentropy(y_true, y_pred)
+#' cbind(focal_loss, bce_loss,
+#'       ratio = focal_loss / bce_loss)
 #' ```
 #'
 #' @returns

From 97886e72b585e305a30c6b77a1a4d88df206b857 Mon Sep 17 00:00:00 2001
From: Tomasz Kalinowski <kalinowskit@gmail.com>
Date: Thu, 25 Sep 2025 15:03:01 -0400
Subject: [PATCH 30/81] internal tether change: update SensitivityAtSpecificity
 example

---
 .tether/man/metric_sensitivity_at_specificity.txt | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/.tether/man/metric_sensitivity_at_specificity.txt b/.tether/man/metric_sensitivity_at_specificity.txt
index bc9735e85..c3c3dd8fd 100644
--- a/.tether/man/metric_sensitivity_at_specificity.txt
+++ b/.tether/man/metric_sensitivity_at_specificity.txt
@@ -56,7 +56,7 @@ class SensitivityAtSpecificity(SensitivitySpecificityBase)
  |  model.compile(
  |      optimizer='sgd',
  |      loss='binary_crossentropy',
- |      metrics=[keras.metrics.SensitivityAtSpecificity()])
+ |      metrics=[keras.metrics.SensitivityAtSpecificity(specificity=0.5)])
  |  ```
  |
  |  Method resolution order:

From 3af08ae27ec30e7f9d08deefe9e801027baf2d6b Mon Sep 17 00:00:00 2001
From: Tomasz Kalinowski <kalinowskit@gmail.com>
Date: Thu, 25 Sep 2025 15:05:38 -0400
Subject: [PATCH 31/81] internal tether change: update SpecificityAtSensitivity
 example

---
 .tether/man/metric_specificity_at_sensitivity.txt | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/.tether/man/metric_specificity_at_sensitivity.txt b/.tether/man/metric_specificity_at_sensitivity.txt
index d320e059f..6e21b36c9 100644
--- a/.tether/man/metric_specificity_at_sensitivity.txt
+++ b/.tether/man/metric_specificity_at_sensitivity.txt
@@ -56,7 +56,7 @@ class SpecificityAtSensitivity(SensitivitySpecificityBase)
  |  model.compile(
  |      optimizer='sgd',
  |      loss='binary_crossentropy',
- |      metrics=[keras.metrics.SpecificityAtSensitivity()])
+ |      metrics=[keras.metrics.SpecificityAtSensitivity(sensitivity=0.3)])
  |  ```
  |
  |  Method resolution order:

From 39eb0d47df08281a1a3a057200a9d017658b32c9 Mon Sep 17 00:00:00 2001
From: Tomasz Kalinowski <kalinowskit@gmail.com>
Date: Thu, 25 Sep 2025 15:57:08 -0400
Subject: [PATCH 32/81] `save_model_weights()` gains arg `max_shard_size`.

---
 .tether/man/save_model_weights.txt | 59 ++++++++++++++++++++++++++----
 R/model-persistence.R              | 54 ++++++++++++++++++++++-----
 2 files changed, 96 insertions(+), 17 deletions(-)

diff --git a/.tether/man/save_model_weights.txt b/.tether/man/save_model_weights.txt
index 2820c7a22..fafd3596e 100644
--- a/.tether/man/save_model_weights.txt
+++ b/.tether/man/save_model_weights.txt
@@ -2,14 +2,59 @@ __signature__
 keras.Model.save_weights(
   self,
   filepath,
-  overwrite=True
+  overwrite=True,
+  max_shard_size=None
 )
 __doc__
-Saves all layer weights to a `.weights.h5` file.
+Saves all weights to a single file or sharded files.
+
+By default, the weights will be saved in a single `.weights.h5` file.
+If sharding is enabled (`max_shard_size` is not `None`), the weights
+will be saved in multiple files, each with a size at most
+`max_shard_size` (in GB). Additionally, a configuration file
+`.weights.json` will contain the metadata for the sharded files.
+
+The saved sharded files contain:
+
+- `*.weights.json`: The configuration file containing 'metadata' and
+    'weight_map'.
+- `*_xxxxxx.weights.h5`: The sharded files containing only the
+    weights.
 
 Args:
-    filepath: `str` or `pathlib.Path` object.
-        Path where to save the model. Must end in `.weights.h5`.
-    overwrite: Whether we should overwrite any existing model
-        at the target location, or instead ask the user
-        via an interactive prompt.
+    filepath: `str` or `pathlib.Path` object. Path where the weights
+        will be saved.  When sharding, the filepath must end in
+        `.weights.json`. If `.weights.h5` is provided, it will be
+        overridden.
+    overwrite: Whether to overwrite any existing weights at the target
+        location or instead ask the user via an interactive prompt.
+    max_shard_size: `int` or `float`. Maximum size in GB for each
+        sharded file. If `None`, no sharding will be done. Defaults to
+        `None`.
+
+Example:
+
+```python
+# Instantiate a EfficientNetV2L model with about 454MB of weights.
+model = keras.applications.EfficientNetV2L(weights=None)
+
+# Save the weights in a single file.
+model.save_weights("model.weights.h5")
+
+# Save the weights in sharded files. Use `max_shard_size=0.25` means
+# each sharded file will be at most ~250MB.
+model.save_weights("model.weights.json", max_shard_size=0.25)
+
+# Load the weights in a new model with the same architecture.
+loaded_model = keras.applications.EfficientNetV2L(weights=None)
+loaded_model.load_weights("model.weights.h5")
+x = keras.random.uniform((1, 480, 480, 3))
+assert np.allclose(model.predict(x), loaded_model.predict(x))
+
+# Load the sharded weights in a new model with the same architecture.
+loaded_model = keras.applications.EfficientNetV2L(weights=None)
+loaded_model.load_weights("model.weights.json")
+x = keras.random.uniform((1, 480, 480, 3))
+assert np.allclose(model.predict(x), loaded_model.predict(x))
+```
+
diff --git a/R/model-persistence.R b/R/model-persistence.R
index 802c5c218..fc4a7a25a 100644
--- a/R/model-persistence.R
+++ b/R/model-persistence.R
@@ -166,18 +166,48 @@ function (model, custom_objects = NULL, compile = TRUE, safe_mode = TRUE)
 }
 
 
-#' Saves all layer weights to a `.weights.h5` file.
+#' Saves all weights to a single file or sharded files.
 #'
-#' @param model A keras Model object
+#' @description
+#' By default, the weights are saved in a single `.weights.h5` file. Enable
+#' sharding via `max_shard_size` to split weights across multiple files (in GB)
+#' and produce a `.weights.json` manifest that tracks shard metadata.
+#'
+#' The saved sharded files contain:
+#'
+#' * `*.weights.json`: configuration file containing `metadata` and
+#'   `weight_map` entries.
+#' * `*_xxxxxx.weights.h5`: weight shards limited by `max_shard_size`.
+#'
+#' ```{r}
+#' model <-
+#'   keras_model_sequential(input_shape = 2) |>
+#'   layer_dense(4)
+#'
+#' path_h5 <- tempfile(fileext = ".weights.h5")
+#' path_json <- tempfile(fileext = ".weights.json")
+#'
+#' model |> save_model_weights(path_h5)
+#' model |> save_model_weights(path_json, max_shard_size = 0.01)
+#'
+#' model |> load_model_weights(path_h5)
+#' model |> load_model_weights(path_json)
+#' ```
+#'
+#' @param model A keras Model object.
 #'
 #' @param filepath
-#' string.
-#' Path where to save the model. Must end in `.weights.h5`.
+#' Path where the weights will be saved. Accepts `.weights.h5`, or when sharding
+#' is enabled, a `.weights.json` manifest path. If `.weights.h5` is provided
+#' while sharding, the filename will be overridden to end in `.weights.json`.
 #'
 #' @param overwrite
-#' Whether we should overwrite any existing model
-#' at the target location, or instead ask the user
-#' via an interactive prompt.
+#' Whether to overwrite any existing weights at the target location, or instead
+#' ask the user via an interactive prompt.
+#'
+#' @param max_shard_size
+#' Numeric size in GB for each sharded file. Use `NULL` to disable sharding.
+#'
 #'
 #' @returns This is called primarily for side effects. `model` is returned,
 #'   invisibly, to enable usage with the pipe.
@@ -186,12 +216,16 @@ function (model, custom_objects = NULL, compile = TRUE, safe_mode = TRUE)
 #' @tether keras.Model.save_weights
 #' @seealso
 #' + <https://keras.io/api/models/model_saving_apis/weights_saving_and_loading#saveweights-method>
-#  + <https://www.tensorflow.org/api_docs/python/tf/keras/Model/save_weights>
 save_model_weights <-
-function (model, filepath, overwrite = FALSE)
+function (model, filepath, overwrite = FALSE, max_shard_size = NULL)
 {
     overwrite <- confirm_overwrite(filepath, overwrite)
-    keras$Model$save_weights(model, filepath, overwrite = overwrite)
+    keras$Model$save_weights(
+      model,
+      filepath,
+      overwrite = overwrite,
+      max_shard_size = max_shard_size
+    )
     invisible(model)
 }
 

From 75c7a50e5b061a1d0e2d8d58eff68030bd6e671d Mon Sep 17 00:00:00 2001
From: Tomasz Kalinowski <kalinowskit@gmail.com>
Date: Thu, 25 Sep 2025 16:03:49 -0400
Subject: [PATCH 33/81] internal tether change: update applications module path

---
 .tether/man/keras.applications.txt | 34 +++++++++++++++---------------
 1 file changed, 17 insertions(+), 17 deletions(-)

diff --git a/.tether/man/keras.applications.txt b/.tether/man/keras.applications.txt
index 9f9c401ce..bb8eb26c9 100644
--- a/.tether/man/keras.applications.txt
+++ b/.tether/man/keras.applications.txt
@@ -1,4 +1,4 @@
-convnext: Module(keras.api.applications.convnext)
+convnext: Module(keras.applications.convnext)
 ConvNeXtBase(
   include_top=True,
   include_preprocessing=True,
@@ -54,7 +54,7 @@ ConvNeXtXLarge(
   classifier_activation='softmax',
   name='convnext_xlarge'
 )
-densenet: Module(keras.api.applications.densenet)
+densenet: Module(keras.applications.densenet)
 DenseNet121(
   include_top=True,
   weights='imagenet',
@@ -85,8 +85,8 @@ DenseNet201(
   classifier_activation='softmax',
   name='densenet201'
 )
-efficientnet: Module(keras.api.applications.efficientnet)
-efficientnet_v2: Module(keras.api.applications.efficientnet_v2)
+efficientnet: Module(keras.applications.efficientnet)
+efficientnet_v2: Module(keras.applications.efficientnet_v2)
 EfficientNetB0(
   include_top=True,
   weights='imagenet',
@@ -244,9 +244,9 @@ EfficientNetV2S(
   include_preprocessing=True,
   name='efficientnetv2-s'
 )
-imagenet_utils: Module(keras.api.applications.imagenet_utils)
-inception_resnet_v2: Module(keras.api.applications.inception_resnet_v2)
-inception_v3: Module(keras.api.applications.inception_v3)
+imagenet_utils: Module(keras.applications.imagenet_utils)
+inception_resnet_v2: Module(keras.applications.inception_resnet_v2)
+inception_v3: Module(keras.applications.inception_v3)
 InceptionResNetV2(
   include_top=True,
   weights='imagenet',
@@ -267,7 +267,7 @@ InceptionV3(
   classifier_activation='softmax',
   name='inception_v3'
 )
-mobilenet: Module(keras.api.applications.mobilenet)
+mobilenet: Module(keras.applications.mobilenet)
 MobileNet(
   input_shape=None,
   alpha=1.0,
@@ -281,8 +281,8 @@ MobileNet(
   classifier_activation='softmax',
   name=None
 )
-mobilenet_v2: Module(keras.api.applications.mobilenet_v2)
-mobilenet_v3: Module(keras.api.applications.mobilenet_v3)
+mobilenet_v2: Module(keras.applications.mobilenet_v2)
+mobilenet_v3: Module(keras.applications.mobilenet_v3)
 MobileNetV2(
   input_shape=None,
   alpha=1.0,
@@ -322,7 +322,7 @@ MobileNetV3Small(
   include_preprocessing=True,
   name='MobileNetV3Small'
 )
-nasnet: Module(keras.api.applications.nasnet)
+nasnet: Module(keras.applications.nasnet)
 NASNetLarge(
   input_shape=None,
   include_top=True,
@@ -343,8 +343,8 @@ NASNetMobile(
   classifier_activation='softmax',
   name='nasnet_mobile'
 )
-resnet: Module(keras.api.applications.resnet)
-resnet_v2: Module(keras.api.applications.resnet_v2)
+resnet: Module(keras.applications.resnet)
+resnet_v2: Module(keras.applications.resnet_v2)
 ResNet101(
   include_top=True,
   weights='imagenet',
@@ -385,7 +385,7 @@ ResNet152V2(
   classifier_activation='softmax',
   name='resnet152v2'
 )
-resnet50: Module(keras.api.applications.resnet50)
+resnet50: Module(keras.applications.resnet50)
 ResNet50(
   include_top=True,
   weights='imagenet',
@@ -406,7 +406,7 @@ ResNet50V2(
   classifier_activation='softmax',
   name='resnet50v2'
 )
-vgg16: Module(keras.api.applications.vgg16)
+vgg16: Module(keras.applications.vgg16)
 VGG16(
   include_top=True,
   weights='imagenet',
@@ -417,7 +417,7 @@ VGG16(
   classifier_activation='softmax',
   name='vgg16'
 )
-vgg19: Module(keras.api.applications.vgg19)
+vgg19: Module(keras.applications.vgg19)
 VGG19(
   include_top=True,
   weights='imagenet',
@@ -428,7 +428,7 @@ VGG19(
   classifier_activation='softmax',
   name='vgg19'
 )
-xception: Module(keras.api.applications.xception)
+xception: Module(keras.applications.xception)
 Xception(
   include_top=True,
   weights='imagenet',

From f9fb925b5a14851a436fde0bbab776fd26f43121 Mon Sep 17 00:00:00 2001
From: Tomasz Kalinowski <kalinowskit@gmail.com>
Date: Thu, 25 Sep 2025 16:08:18 -0400
Subject: [PATCH 34/81] internal tether changes

---
 .tether/man/keras.datasets.txt | 16 ++++++++--------
 1 file changed, 8 insertions(+), 8 deletions(-)

diff --git a/.tether/man/keras.datasets.txt b/.tether/man/keras.datasets.txt
index f771e1b50..79bb80818 100644
--- a/.tether/man/keras.datasets.txt
+++ b/.tether/man/keras.datasets.txt
@@ -1,9 +1,9 @@
-boston_housing: Module(keras.api.datasets.boston_housing)
-california_housing: Module(keras.api.datasets.california_housing)
-cifar10: Module(keras.api.datasets.cifar10)
-cifar100: Module(keras.api.datasets.cifar100)
-fashion_mnist: Module(keras.api.datasets.fashion_mnist)
-imdb: Module(keras.api.datasets.imdb)
-mnist: Module(keras.api.datasets.mnist)
-reuters: Module(keras.api.datasets.reuters)
+boston_housing: Module(keras.datasets.boston_housing)
+california_housing: Module(keras.datasets.california_housing)
+cifar10: Module(keras.datasets.cifar10)
+cifar100: Module(keras.datasets.cifar100)
+fashion_mnist: Module(keras.datasets.fashion_mnist)
+imdb: Module(keras.datasets.imdb)
+mnist: Module(keras.datasets.mnist)
+reuters: Module(keras.datasets.reuters)
 

From 953e5cc67c96945f061ae4bef30e323c4ec17706 Mon Sep 17 00:00:00 2001
From: Tomasz Kalinowski <kalinowskit@gmail.com>
Date: Thu, 25 Sep 2025 16:15:27 -0400
Subject: [PATCH 35/81] internal tether changes

---
 .tether/man/keras_model.txt | 122 +++++++++++++++++++++++++++---------
 1 file changed, 91 insertions(+), 31 deletions(-)

diff --git a/.tether/man/keras_model.txt b/.tether/man/keras_model.txt
index a288c1b05..6d7a8f60c 100644
--- a/.tether/man/keras_model.txt
+++ b/.tether/man/keras_model.txt
@@ -303,31 +303,47 @@ class Model(keras.src.backend.tensorflow.trainer.TensorFlowTrainer, keras.src.tr
  |    skip_mismatch=False,
  |    **kwargs
  |  )
- |      Load weights from a file saved via `save_weights()`.
+ |      Load the weights from a single file or sharded files.
  |
- |      Weights are loaded based on the network's
- |      topology. This means the architecture should be the same as when the
- |      weights were saved. Note that layers that don't have weights are not
- |      taken into account in the topological ordering, so adding or removing
- |      layers is fine as long as they don't have weights.
+ |      Weights are loaded based on the network's topology. This means the
+ |      architecture should be the same as when the weights were saved. Note
+ |      that layers that don't have weights are not taken into account in the
+ |      topological ordering, so adding or removing layers is fine as long as
+ |      they don't have weights.
  |
  |      **Partial weight loading**
  |
  |      If you have modified your model, for instance by adding a new layer
- |      (with weights) or by changing the shape of the weights of a layer,
- |      you can choose to ignore errors and continue loading
- |      by setting `skip_mismatch=True`. In this case any layer with
- |      mismatching weights will be skipped. A warning will be displayed
- |      for each skipped layer.
+ |      (with weights) or by changing the shape of the weights of a layer, you
+ |      can choose to ignore errors and continue loading by setting
+ |      `skip_mismatch=True`. In this case any layer with mismatching weights
+ |      will be skipped. A warning will be displayed for each skipped layer.
+ |
+ |      **Sharding**
+ |
+ |      When loading sharded weights, it is important to specify `filepath` that
+ |      ends with `*.weights.json` which is used as the configuration file.
+ |      Additionally, the sharded files `*_xxxxx.weights.h5` must be in the same
+ |      directory as the configuration file.
  |
  |      Args:
- |          filepath: String, path to the weights file to load.
- |              It can either be a `.weights.h5` file
- |              or a legacy `.h5` weights file.
+ |          filepath: `str` or `pathlib.Path` object. Path where the weights
+ |              will be saved.  When sharding, the filepath must end in
+ |              `.weights.json`.
  |          skip_mismatch: Boolean, whether to skip loading of layers where
  |              there is a mismatch in the number of weights, or a mismatch in
  |              the shape of the weights.
  |
+ |      Example:
+ |
+ |      ```python
+ |      # Load the weights in a single file.
+ |      model.load_weights("model.weights.h5")
+ |
+ |      # Load the weights in sharded files.
+ |      model.load_weights("model.weights.json")
+ |      ```
+ |
  |  quantize(
  |    self,
  |    mode,
@@ -352,6 +368,16 @@ class Model(keras.src.backend.tensorflow.trainer.TensorFlowTrainer, keras.src.tr
  |  )
  |      Saves a model as a `.keras` file.
  |
+ |      Note that `model.save()` is an alias for `keras.saving.save_model()`.
+ |
+ |      The saved `.keras` file contains:
+ |
+ |      - The model's configuration (architecture)
+ |      - The model's weights
+ |      - The model's optimizer's state (if any)
+ |
+ |      Thus models can be reinstantiated in the exact same state.
+ |
  |      Args:
  |          filepath: `str` or `pathlib.Path` object.
  |              The path where to save the model. Must end in `.keras`
@@ -380,29 +406,63 @@ class Model(keras.src.backend.tensorflow.trainer.TensorFlowTrainer, keras.src.tr
  |      assert np.allclose(model.predict(x), loaded_model.predict(x))
  |      ```
  |
- |      Note that `model.save()` is an alias for `keras.saving.save_model()`.
- |
- |      The saved `.keras` file contains:
- |
- |      - The model's configuration (architecture)
- |      - The model's weights
- |      - The model's optimizer's state (if any)
- |
- |      Thus models can be reinstantiated in the exact same state.
- |
  |  save_weights(
  |    self,
  |    filepath,
- |    overwrite=True
+ |    overwrite=True,
+ |    max_shard_size=None
  |  )
- |      Saves all layer weights to a `.weights.h5` file.
+ |      Saves all weights to a single file or sharded files.
+ |
+ |      By default, the weights will be saved in a single `.weights.h5` file.
+ |      If sharding is enabled (`max_shard_size` is not `None`), the weights
+ |      will be saved in multiple files, each with a size at most
+ |      `max_shard_size` (in GB). Additionally, a configuration file
+ |      `.weights.json` will contain the metadata for the sharded files.
+ |
+ |      The saved sharded files contain:
+ |
+ |      - `*.weights.json`: The configuration file containing 'metadata' and
+ |          'weight_map'.
+ |      - `*_xxxxxx.weights.h5`: The sharded files containing only the
+ |          weights.
  |
  |      Args:
- |          filepath: `str` or `pathlib.Path` object.
- |              Path where to save the model. Must end in `.weights.h5`.
- |          overwrite: Whether we should overwrite any existing model
- |              at the target location, or instead ask the user
- |              via an interactive prompt.
+ |          filepath: `str` or `pathlib.Path` object. Path where the weights
+ |              will be saved.  When sharding, the filepath must end in
+ |              `.weights.json`. If `.weights.h5` is provided, it will be
+ |              overridden.
+ |          overwrite: Whether to overwrite any existing weights at the target
+ |              location or instead ask the user via an interactive prompt.
+ |          max_shard_size: `int` or `float`. Maximum size in GB for each
+ |              sharded file. If `None`, no sharding will be done. Defaults to
+ |              `None`.
+ |
+ |      Example:
+ |
+ |      ```python
+ |      # Instantiate a EfficientNetV2L model with about 454MB of weights.
+ |      model = keras.applications.EfficientNetV2L(weights=None)
+ |
+ |      # Save the weights in a single file.
+ |      model.save_weights("model.weights.h5")
+ |
+ |      # Save the weights in sharded files. Use `max_shard_size=0.25` means
+ |      # each sharded file will be at most ~250MB.
+ |      model.save_weights("model.weights.json", max_shard_size=0.25)
+ |
+ |      # Load the weights in a new model with the same architecture.
+ |      loaded_model = keras.applications.EfficientNetV2L(weights=None)
+ |      loaded_model.load_weights("model.weights.h5")
+ |      x = keras.random.uniform((1, 480, 480, 3))
+ |      assert np.allclose(model.predict(x), loaded_model.predict(x))
+ |
+ |      # Load the sharded weights in a new model with the same architecture.
+ |      loaded_model = keras.applications.EfficientNetV2L(weights=None)
+ |      loaded_model.load_weights("model.weights.json")
+ |      x = keras.random.uniform((1, 480, 480, 3))
+ |      assert np.allclose(model.predict(x), loaded_model.predict(x))
+ |      ```
  |
  |  set_state_tree(self, state_tree)
  |      Assigns values to variables of the model.

From 60a050c7b0f71e2e5a1b513e27a730077fe1e4d5 Mon Sep 17 00:00:00 2001
From: Tomasz Kalinowski <kalinowskit@gmail.com>
Date: Thu, 25 Sep 2025 16:19:02 -0400
Subject: [PATCH 36/81] `keras_variable()` gains arg synchronization

---
 .tether/man/keras_variable.txt | 12 ++++++++++--
 R/variable.R                   | 11 ++++++++++-
 2 files changed, 20 insertions(+), 3 deletions(-)

diff --git a/.tether/man/keras_variable.txt b/.tether/man/keras_variable.txt
index e5bde6b92..75d0602a7 100644
--- a/.tether/man/keras_variable.txt
+++ b/.tether/man/keras_variable.txt
@@ -8,7 +8,9 @@ class Variable(builtins.object)
  |    trainable=True,
  |    autocast=True,
  |    aggregation='none',
- |    name=None
+ |    synchronization='auto',
+ |    name=None,
+ |    **kwargs
  |  )
  |
  |  Represents a backend-agnostic variable in Keras.
@@ -49,6 +51,7 @@ class Variable(builtins.object)
  |      value: The current value of the variable (NumPy array or tensor).
  |      name: The name of the variable (string).
  |      path: The path of the variable within the Keras model or layer (string).
+ |      kwargs: Additional backend-specific keyword arguments.
  |
  |  Examples:
  |
@@ -120,7 +123,9 @@ class Variable(builtins.object)
  |    trainable=True,
  |    autocast=True,
  |    aggregation='none',
- |    name=None
+ |    synchronization='auto',
+ |    name=None,
+ |    **kwargs
  |  )
  |      Initialize self.  See help(type(self)) for accurate signature.
  |
@@ -215,6 +220,9 @@ class Variable(builtins.object)
  |  shape
  |      The shape of the variable.
  |
+ |  synchronization
+ |      The strategy for synchronizing this variable.
+ |
  |  value
  |      The current value of the variable (numpy array or backend tensor).
  |
diff --git a/R/variable.R b/R/variable.R
index 4afca3e00..a914ba79b 100644
--- a/R/variable.R
+++ b/R/variable.R
@@ -14,6 +14,7 @@
 #' - `trainable`: Whether the variable is trainable (boolean).
 #' - `autocast`: Whether the variable supports autocasting (boolean).
 #' - `aggregation`: How a distributed variable will be aggregated (string).
+#' - `synchronization`: Strategy for synchronizing the variable across devices (string).
 #' - `value`: The current value of the variable (NumPy array or tensor).
 #' - `name`: The name of the variable (string).
 #' - `path`: The path of the variable within the Keras model or layer (string).
@@ -88,15 +89,23 @@
 #' to be taken into account by downstream backends or users. Defaults
 #' to `"none"`.
 #'
+#' @param synchronization
+#' Optional string specifying how distributed values should be synchronized.
+#' Defaults to `"auto"`.
+#'
 #' @param name
 #' Optional. A unique name for the variable. Automatically generated
 #' if not set.
 #'
+#' @param ...
+#' Additional backend-specific keyword arguments forwarded to `keras$Variable()`.
+#'
 #' @export
 #' @tether keras.src.backend.common.variables.Variable
 keras_variable <-
 function (initializer, shape = NULL, dtype = NULL, trainable = TRUE,
-          autocast = TRUE, aggregation = "none", name = NULL)
+          autocast = TRUE, aggregation = "none", synchronization = "auto",
+          name = NULL, ...)
 {
   args <- capture_args(list(shape = normalize_shape))
   do.call(keras$Variable, args)

From 07bf72fd1afd5fa67174a6a51806aeac29a11d9b Mon Sep 17 00:00:00 2001
From: Tomasz Kalinowski <kalinowskit@gmail.com>
Date: Thu, 25 Sep 2025 16:26:13 -0400
Subject: [PATCH 37/81] internal tether change: allow Sequential call kwargs

---
 .tether/man/keras_model_sequential.txt | 6 ++++--
 1 file changed, 4 insertions(+), 2 deletions(-)

diff --git a/.tether/man/keras_model_sequential.txt b/.tether/man/keras_model_sequential.txt
index c13d33388..2d91bcd38 100644
--- a/.tether/man/keras_model_sequential.txt
+++ b/.tether/man/keras_model_sequential.txt
@@ -87,7 +87,8 @@ class Sequential(keras.src.models.model.Model)
  |    self,
  |    inputs,
  |    training=None,
- |    mask=None
+ |    mask=None,
+ |    **kwargs
  |  )
  |
  |  compute_output_shape(self, input_shape)
@@ -96,7 +97,8 @@ class Sequential(keras.src.models.model.Model)
  |    self,
  |    inputs,
  |    training=None,
- |    mask=None
+ |    mask=None,
+ |    **kwargs
  |  )
  |
  |  get_config(self)

From 0ff476b4890043261f0b2c7ba4296838ccf89494 Mon Sep 17 00:00:00 2001
From: Tomasz Kalinowski <kalinowskit@gmail.com>
Date: Thu, 25 Sep 2025 16:33:52 -0400
Subject: [PATCH 38/81] redocument

---
 man/Layer.Rd                                |  7 ++++
 man/activation_elu.Rd                       |  2 +-
 man/callback_lambda.Rd                      |  6 +--
 man/callback_model_checkpoint.Rd            |  5 +--
 man/keras_variable.Rd                       | 10 ++++-
 man/layer_average_pooling_2d.Rd             |  2 +-
 man/layer_conv_1d_transpose.Rd              |  9 ++++-
 man/layer_conv_2d.Rd                        |  4 ++
 man/layer_conv_2d_transpose.Rd              |  6 ++-
 man/layer_conv_3d.Rd                        |  2 +-
 man/layer_conv_3d_transpose.Rd              |  6 ++-
 man/layer_dense.Rd                          |  6 +++
 man/layer_einsum_dense.Rd                   |  5 +++
 man/layer_embedding.Rd                      | 10 +++++
 man/layer_layer_normalization.Rd            |  8 ----
 man/layer_rescaling.Rd                      |  8 +++-
 man/layer_rnn.Rd                            |  1 -
 man/layer_tfsm.Rd                           |  4 +-
 man/layer_torch_module_wrapper.Rd           | 10 ++++-
 man/load_model_weights.Rd                   | 20 ++++++++--
 man/loss_dice.Rd                            |  6 ++-
 man/loss_sparse_categorical_crossentropy.Rd | 27 +++++--------
 man/metric_binary_focal_crossentropy.Rd     | 15 ++++---
 man/op_conv.Rd                              |  2 +-
 man/op_conv_transpose.Rd                    |  2 +-
 man/op_dot_product_attention.Rd             |  6 ++-
 man/op_einsum.Rd                            |  3 +-
 man/op_rms_normalization.Rd                 | 34 +++++++++++++---
 man/optimizer_lion.Rd                       |  4 +-
 man/save_model_weights.Rd                   | 43 ++++++++++++++++-----
 30 files changed, 192 insertions(+), 81 deletions(-)

diff --git a/man/Layer.Rd b/man/Layer.Rd
index 588e926fb..000fba8df 100644
--- a/man/Layer.Rd
+++ b/man/Layer.Rd
@@ -300,6 +300,7 @@ Alias of \code{add_weight()}.
            regularizer = NULL,
            constraint = NULL,
            aggregation = 'none',
+           overwrite_with_gradient = FALSE,
            name = NULL)
 }\if{html}{\out{</div>}}
 
@@ -340,6 +341,8 @@ Defaults to \code{NULL}.
 the type of multi-replica aggregation to be used for this
 variable when writing custom data parallel training loops.
 Defaults to \code{"none"}.
+\item \code{overwrite_with_gradient}: Boolean, whether to overwrite the variable with
+the computed gradient. Useful for float8 training. Defaults to \code{FALSE}.
 \item \code{name}: String name of the variable. Useful for debugging purposes.
 }
 
@@ -404,6 +407,10 @@ See description above
 }\if{html}{\out{</div>}}
 \item
 
+\if{html}{\out{<div class="sourceCode r">}}\preformatted{symbolic_call(...)
+}\if{html}{\out{</div>}}
+\item
+
 \if{html}{\out{<div class="sourceCode r">}}\preformatted{count_params()
 }\if{html}{\out{</div>}}
 
diff --git a/man/activation_elu.Rd b/man/activation_elu.Rd
index 8d5e89f06..dcfdbe8f1 100644
--- a/man/activation_elu.Rd
+++ b/man/activation_elu.Rd
@@ -9,7 +9,7 @@ activation_elu(x, alpha = 1)
 \arguments{
 \item{x}{Input tensor.}
 
-\item{alpha}{Numeric. See description for details.}
+\item{alpha}{A scalar, slope of positive section. Defaults to \code{1.0}.}
 }
 \value{
 A tensor, the result from applying the activation to the input tensor \code{x}.
diff --git a/man/callback_lambda.Rd b/man/callback_lambda.Rd
index 5421d3aeb..8d63df555 100644
--- a/man/callback_lambda.Rd
+++ b/man/callback_lambda.Rd
@@ -30,7 +30,7 @@ callback_lambda(
 \item{...}{Any function in \code{\link[=Callback]{Callback()}} that you want to override by
 passing \verb{function_name = function}. For example,
 \code{callback_lambda(.., on_train_end = train_end_fn)}. The custom function
-needs to have same arguments as the ones defined in \code{\link[=Callback]{Callback()}}.}
+needs to have the same arguments as the ones defined in \code{\link[=Callback]{Callback()}}.}
 }
 \value{
 A \code{Callback} instance that can be passed to \code{\link[=fit.keras.src.models.model.Model]{fit.keras.src.models.model.Model()}}.
@@ -44,8 +44,8 @@ Note that the callbacks expects positional arguments, as:
 \code{epoch}, \code{logs}
 \item \code{on_train_begin} and \code{on_train_end} expect one positional argument:
 \code{logs}
-\item \code{on_train_batch_begin} and \code{on_train_batch_end} expect two positional
-arguments: \code{batch}, \code{logs}
+\item \code{on_train_batch_begin} and \code{on_train_batch_end} expect a positional
+argument \code{batch} and a named argument \code{logs}
 \item See \code{Callback} class definition for the full list of functions and their
 expected arguments.
 }
diff --git a/man/callback_model_checkpoint.Rd b/man/callback_model_checkpoint.Rd
index 43a4ede43..0bce70c6d 100644
--- a/man/callback_model_checkpoint.Rd
+++ b/man/callback_model_checkpoint.Rd
@@ -62,9 +62,8 @@ overwritten by each new better model.}
 decision to overwrite the current save file is made based on either
 the maximization or the minimization of the monitored quantity.
 For \code{val_acc}, this should be \code{"max"}, for \code{val_loss} this should be
-\code{"min"}, etc. In \code{"auto"} mode, the mode is set to \code{"max"} if the
-quantities monitored are \code{"acc"} or start with \code{"fmeasure"} and are
-set to \code{"min"} for the rest of the quantities.}
+\code{"min"}, etc. In \code{"auto"} mode, the direction is automatically inferred from
+the name of the monitored quantity.}
 
 \item{save_freq}{\code{"epoch"} or integer. When using \code{"epoch"}, the callback
 saves the model after each epoch. When using integer, the callback
diff --git a/man/keras_variable.Rd b/man/keras_variable.Rd
index fbac11d52..729980535 100644
--- a/man/keras_variable.Rd
+++ b/man/keras_variable.Rd
@@ -11,7 +11,9 @@ keras_variable(
   trainable = TRUE,
   autocast = TRUE,
   aggregation = "none",
-  name = NULL
+  synchronization = "auto",
+  name = NULL,
+  ...
 )
 }
 \arguments{
@@ -38,8 +40,13 @@ variable will be aggregated. This serves as a semantic annotation,
 to be taken into account by downstream backends or users. Defaults
 to \code{"none"}.}
 
+\item{synchronization}{Optional string specifying how distributed values should be synchronized.
+Defaults to \code{"auto"}.}
+
 \item{name}{Optional. A unique name for the variable. Automatically generated
 if not set.}
+
+\item{...}{Additional backend-specific keyword arguments forwarded to \code{keras$Variable()}.}
 }
 \description{
 A \code{Variable} acts as a container for state. It holds a tensor value and can
@@ -55,6 +62,7 @@ a piece of computation to turn it into a stateless function.
 \item \code{trainable}: Whether the variable is trainable (boolean).
 \item \code{autocast}: Whether the variable supports autocasting (boolean).
 \item \code{aggregation}: How a distributed variable will be aggregated (string).
+\item \code{synchronization}: Strategy for synchronizing the variable across devices (string).
 \item \code{value}: The current value of the variable (NumPy array or tensor).
 \item \code{name}: The name of the variable (string).
 \item \code{path}: The path of the variable within the Keras model or layer (string).
diff --git a/man/layer_average_pooling_2d.Rd b/man/layer_average_pooling_2d.Rd
index 0a2f57b76..166361f04 100644
--- a/man/layer_average_pooling_2d.Rd
+++ b/man/layer_average_pooling_2d.Rd
@@ -66,7 +66,7 @@ shape (number of rows or columns) of:
 (when \code{input_shape >= pool_size})
 
 The resulting output shape when using the \code{"same"} padding option is:
-\code{output_shape = math.floor((input_shape - 1) / strides) + 1}
+\code{output_shape = input_shape}
 }
 \section{Input Shape}{
 \itemize{
diff --git a/man/layer_conv_1d_transpose.Rd b/man/layer_conv_1d_transpose.Rd
index ad5e0135a..08db7168c 100644
--- a/man/layer_conv_1d_transpose.Rd
+++ b/man/layer_conv_1d_transpose.Rd
@@ -10,6 +10,7 @@ layer_conv_1d_transpose(
   kernel_size,
   strides = 1L,
   padding = "valid",
+  output_padding = NULL,
   data_format = NULL,
   dilation_rate = 1L,
   activation = NULL,
@@ -42,6 +43,9 @@ of the transposed convolution. \code{strides > 1} is incompatible with
 the left/right or up/down of the input such that output has the same
 height/width dimension as the input.}
 
+\item{output_padding}{Scalar integer. Amount of padding to add to the output length. Must be less
+than the stride. When \code{NULL} (default) the output size is inferred.}
+
 \item{data_format}{string, either \code{"channels_last"} or \code{"channels_first"}.
 The ordering of the dimensions in the inputs. \code{"channels_last"}
 corresponds to inputs with shape \verb{(batch, steps, features)}
@@ -50,8 +54,9 @@ while \code{"channels_first"} corresponds to inputs with shape
 value found in your Keras config file at \verb{~/.keras/keras.json}.
 If you never set it, then it will be \code{"channels_last"}.}
 
-\item{dilation_rate}{int or list of 1 integers, specifying the dilation
-rate to use for dilated transposed convolution.}
+\item{dilation_rate}{Scalar integer. Specifies the dilation rate. Values other
+than 1 currently require \code{strides = 1} and rates greater than 1 are not
+supported.}
 
 \item{activation}{Activation function. If \code{NULL}, no activation is applied.}
 
diff --git a/man/layer_conv_2d.Rd b/man/layer_conv_2d.Rd
index 938dc563d..9ac6c189f 100644
--- a/man/layer_conv_2d.Rd
+++ b/man/layer_conv_2d.Rd
@@ -89,6 +89,10 @@ are not safe to use when doing asynchronous distributed training.}
 bias after being updated by an \code{Optimizer}.}
 
 \item{...}{For forward/backward compatability.}
+
+\item{output_padding}{Scalar integer or vector of two integers. Amount of padding to add to the output
+depth, height, and width. Each element must be smaller than the corresponding
+stride. When \code{NULL} (default) the output size is inferred.}
 }
 \value{
 A 4D tensor representing \code{activation(conv2d(inputs, kernel) + bias)}.
diff --git a/man/layer_conv_2d_transpose.Rd b/man/layer_conv_2d_transpose.Rd
index 1c06d7291..ef3584120 100644
--- a/man/layer_conv_2d_transpose.Rd
+++ b/man/layer_conv_2d_transpose.Rd
@@ -10,6 +10,7 @@ layer_conv_2d_transpose(
   kernel_size,
   strides = list(1L, 1L),
   padding = "valid",
+  output_padding = NULL,
   data_format = NULL,
   dilation_rate = list(1L, 1L),
   activation = NULL,
@@ -52,8 +53,9 @@ while \code{"channels_first"} corresponds to inputs with shape
 \verb{~/.keras/keras.json}. If you never set it, then it will be
 \code{"channels_last"}.}
 
-\item{dilation_rate}{int or list of 1 integers, specifying the dilation
-rate to use for dilated transposed convolution.}
+\item{dilation_rate}{Scalar integer or vector of 2 integers specifying the dilation rate. Values
+other than 1 require \code{strides = 1}; different rates per dimension are not
+supported.}
 
 \item{activation}{Activation function. If \code{NULL}, no activation is applied.}
 
diff --git a/man/layer_conv_3d.Rd b/man/layer_conv_3d.Rd
index b7682d9bb..d657cbcbf 100644
--- a/man/layer_conv_3d.Rd
+++ b/man/layer_conv_3d.Rd
@@ -53,7 +53,7 @@ It defaults to the \code{image_data_format} value found in your Keras
 config file at \verb{~/.keras/keras.json}. If you never set it, then it
 will be \code{"channels_last"}.}
 
-\item{dilation_rate}{int or list of 3 integers, specifying the dilation
+\item{dilation_rate}{int or vector of 3 ints, specifying the dilation
 rate to use for dilated convolution.}
 
 \item{groups}{A positive int specifying the number of groups in which the
diff --git a/man/layer_conv_3d_transpose.Rd b/man/layer_conv_3d_transpose.Rd
index 36fc55800..af696e205 100644
--- a/man/layer_conv_3d_transpose.Rd
+++ b/man/layer_conv_3d_transpose.Rd
@@ -10,6 +10,7 @@ layer_conv_3d_transpose(
   kernel_size,
   strides = list(1L, 1L, 1L),
   padding = "valid",
+  output_padding = NULL,
   data_format = NULL,
   dilation_rate = list(1L, 1L, 1L),
   activation = NULL,
@@ -52,8 +53,9 @@ It defaults to the \code{image_data_format} value found in your Keras
 config file at \verb{~/.keras/keras.json}. If you never set it, then it
 will be \code{"channels_last"}.}
 
-\item{dilation_rate}{int or list of 1 integers, specifying the dilation
-rate to use for dilated transposed convolution.}
+\item{dilation_rate}{Scalar integer or vector of 3 integers specifying the dilation rate. Values
+other than 1 require \code{strides = 1}; different rates per dimension are not
+supported.}
 
 \item{activation}{Activation function. If \code{NULL}, no activation is applied.}
 
diff --git a/man/layer_dense.Rd b/man/layer_dense.Rd
index 428a31619..dc72eadb1 100644
--- a/man/layer_dense.Rd
+++ b/man/layer_dense.Rd
@@ -17,6 +17,7 @@ layer_dense(
   kernel_constraint = NULL,
   bias_constraint = NULL,
   lora_rank = NULL,
+  lora_alpha = NULL,
   ...
 )
 }
@@ -58,6 +59,10 @@ computation cost of fine-tuning large dense layers.
 You can also enable LoRA on an existing
 \code{Dense} layer by calling \code{layer$enable_lora(rank)}.}
 
+\item{lora_alpha}{Optional integer. Scales the low-rank adaptation delta during the forward
+pass. The delta is scaled by \code{lora_alpha / lora_rank}, letting you tune the
+LoRA adjustment strength independently of \code{lora_rank}.}
+
 \item{...}{For forward/backward compatability.}
 }
 \value{
@@ -108,6 +113,7 @@ the output would have shape \verb{(batch_size, units)}.
 
 \if{html}{\out{<div class="sourceCode r">}}\preformatted{enable_lora(
   rank,
+  lora_alpha = NULL,
   a_initializer = 'he_uniform',
   b_initializer = 'zeros'
 )
diff --git a/man/layer_einsum_dense.Rd b/man/layer_einsum_dense.Rd
index 50111f9c3..39975114d 100644
--- a/man/layer_einsum_dense.Rd
+++ b/man/layer_einsum_dense.Rd
@@ -17,6 +17,7 @@ layer_einsum_dense(
   kernel_constraint = NULL,
   bias_constraint = NULL,
   lora_rank = NULL,
+  lora_alpha = NULL,
   ...
 )
 }
@@ -69,6 +70,10 @@ computation cost of fine-tuning large dense layers.
 You can also enable LoRA on an existing
 \code{EinsumDense} layer by calling \code{layer$enable_lora(rank)}.}
 
+\item{lora_alpha}{Optional integer. Scales the low-rank adaptation delta during the forward
+pass. The delta is scaled by \code{lora_alpha / lora_rank}, letting you tune the
+LoRA adjustment strength independently of \code{lora_rank}.}
+
 \item{...}{Base layer keyword arguments, such as \code{name} and \code{dtype}.}
 }
 \value{
diff --git a/man/layer_embedding.Rd b/man/layer_embedding.Rd
index 04634eba4..3a865ffd8 100644
--- a/man/layer_embedding.Rd
+++ b/man/layer_embedding.Rd
@@ -14,6 +14,7 @@ layer_embedding(
   mask_zero = FALSE,
   weights = NULL,
   lora_rank = NULL,
+  lora_alpha = NULL,
   ...
 )
 }
@@ -58,6 +59,10 @@ computation cost of fine-tuning large embedding layers.
 You can also enable LoRA on an existing
 \code{Embedding} layer instance by calling \code{layer$enable_lora(rank)}.}
 
+\item{lora_alpha}{Optional integer. Scales the low-rank adaptation delta during the forward
+pass. The delta is scaled by \code{lora_alpha / lora_rank}, letting you tune the
+LoRA adjustment strength independently of \code{lora_rank}.}
+
 \item{...}{For forward/backward compatability.}
 }
 \value{
@@ -111,12 +116,17 @@ dim(output_array)    # (32, 10, 64)
 
 \if{html}{\out{<div class="sourceCode r">}}\preformatted{enable_lora(
   rank,
+  lora_alpha = NULL,
   a_initializer = 'he_uniform',
   b_initializer = 'zeros'
 )
 }\if{html}{\out{</div>}}
 \item
 
+\if{html}{\out{<div class="sourceCode r">}}\preformatted{compute_output_spec(...)
+}\if{html}{\out{</div>}}
+\item
+
 \if{html}{\out{<div class="sourceCode r">}}\preformatted{quantize(mode, type_check = TRUE)
 }\if{html}{\out{</div>}}
 \item
diff --git a/man/layer_layer_normalization.Rd b/man/layer_layer_normalization.Rd
index cc53e41ab..365db89d8 100644
--- a/man/layer_layer_normalization.Rd
+++ b/man/layer_layer_normalization.Rd
@@ -10,7 +10,6 @@ layer_layer_normalization(
   epsilon = 0.001,
   center = TRUE,
   scale = TRUE,
-  rms_scaling = FALSE,
   beta_initializer = "zeros",
   gamma_initializer = "ones",
   beta_regularizer = NULL,
@@ -39,13 +38,6 @@ When the next layer is linear (also e.g. \code{layer_activation_relu()}), this c
 disabled since the scaling will be done by the next layer.
 Defaults to \code{TRUE}.}
 
-\item{rms_scaling}{If \code{TRUE}, \code{center} and \code{scale} are ignored, and the
-inputs are scaled by \code{gamma} and the inverse square root
-of the square of all inputs. This is an approximate and faster
-approach that avoids ever computing the mean of the input. Note that
-this \emph{isn't} equivalent to the computation that the
-\code{layer_rms_normalization} layer performs.}
-
 \item{beta_initializer}{Initializer for the beta weight. Defaults to zeros.}
 
 \item{gamma_initializer}{Initializer for the gamma weight. Defaults to ones.}
diff --git a/man/layer_rescaling.Rd b/man/layer_rescaling.Rd
index 9dd0327ee..d41c9b83e 100644
--- a/man/layer_rescaling.Rd
+++ b/man/layer_rescaling.Rd
@@ -9,9 +9,13 @@ layer_rescaling(object, scale, offset = 0, ...)
 \arguments{
 \item{object}{Object to compose the layer with. A tensor, array, or sequential model.}
 
-\item{scale}{Float, the scale to apply to the inputs.}
+\item{scale}{Numeric scalar, vector, or array. The scale to apply to the inputs. If
+scalar, the same scale is applied to every feature or channel; if a vector
+or array, scaling is applied per channel.}
 
-\item{offset}{Float, the offset to apply to the inputs.}
+\item{offset}{Numeric scalar, vector, or array. The offset to apply to the inputs. If
+scalar, the same offset is applied to every feature or channel; if a vector
+or array, the shift is applied per channel.}
 
 \item{...}{Base layer keyword arguments, such as \code{name} and \code{dtype}.}
 }
diff --git a/man/layer_rnn.Rd b/man/layer_rnn.Rd
index cb144433d..1bd25ee44 100644
--- a/man/layer_rnn.Rd
+++ b/man/layer_rnn.Rd
@@ -184,7 +184,6 @@ First, let's define a RNN Cell, as a layer subclass.
       initializer = 'uniform',
       name = 'recurrent_kernel'
     )
-    self$built <- TRUE
   \},
 
   call = function(inputs, states) \{
diff --git a/man/layer_tfsm.Rd b/man/layer_tfsm.Rd
index d97239208..b969d813e 100644
--- a/man/layer_tfsm.Rd
+++ b/man/layer_tfsm.Rd
@@ -59,8 +59,8 @@ model |> export_savedmodel("path/to/artifact")
 ## Output Type:
 ##   TensorSpec(shape=(None, 10), dtype=tf.float32, name=None)
 ## Captures:
-##   137054704163664: TensorSpec(shape=(), dtype=tf.resource, name=None)
-##   137054704159440: TensorSpec(shape=(), dtype=tf.resource, name=None)
+##   127996268992208: TensorSpec(shape=(), dtype=tf.resource, name=None)
+##   127996269001424: TensorSpec(shape=(), dtype=tf.resource, name=None)
 
 }\if{html}{\out{</div>}}
 
diff --git a/man/layer_torch_module_wrapper.Rd b/man/layer_torch_module_wrapper.Rd
index 543c70768..0be44bc55 100644
--- a/man/layer_torch_module_wrapper.Rd
+++ b/man/layer_torch_module_wrapper.Rd
@@ -4,7 +4,13 @@
 \alias{layer_torch_module_wrapper}
 \title{Torch module wrapper layer.}
 \usage{
-layer_torch_module_wrapper(object, module, name = NULL, ...)
+layer_torch_module_wrapper(
+  object,
+  module,
+  output_shape = NULL,
+  name = NULL,
+  ...
+)
 }
 \arguments{
 \item{object}{Object to compose the layer with. A tensor, array, or sequential model.}
@@ -14,6 +20,8 @@ instance, then its parameters must be initialized before
 passing the instance to \code{layer_torch_module_wrapper} (e.g. by calling
 it once).}
 
+\item{output_shape}{Shape of the output from this layer. Helps Keras infer shapes.}
+
 \item{name}{The name of the layer (string).}
 
 \item{...}{For forward/backward compatability.}
diff --git a/man/load_model_weights.Rd b/man/load_model_weights.Rd
index d8954755e..ae9b1e655 100644
--- a/man/load_model_weights.Rd
+++ b/man/load_model_weights.Rd
@@ -2,16 +2,16 @@
 % Please edit documentation in R/model-persistence.R
 \name{load_model_weights}
 \alias{load_model_weights}
-\title{Load weights from a file saved via \code{save_model_weights()}.}
+\title{Load the weights from a single file or sharded files.}
 \usage{
 load_model_weights(model, filepath, skip_mismatch = FALSE, ...)
 }
 \arguments{
 \item{model}{A keras model.}
 
-\item{filepath}{String, path to the weights file to load.
-It can either be a \code{.weights.h5} file
-or a legacy \code{.h5} weights file.}
+\item{filepath}{Path or path-like object to the weights. Accepts \code{.weights.h5}, legacy \code{.h5},
+or sharded weights through a \code{.weights.json} manifest sitting alongside the
+shard files (\verb{*_xxxxx.weights.h5}).}
 
 \item{skip_mismatch}{Boolean, whether to skip loading of layers where
 there is a mismatch in the number of weights, or a mismatch in
@@ -30,6 +30,12 @@ weights were saved. Note that layers that don't have weights are not
 taken into account in the topological ordering, so adding or removing
 layers is fine as long as they don't have weights.
 
+\strong{Sharding}
+
+When loading sharded weights, specify a \code{filepath} ending in
+\code{".weights.json"} (the configuration file), with the corresponding shard files
+(\verb{*_xxxxx.weights.h5}) located alongside it.
+
 \strong{Partial weight loading}
 
 If you have modified your model, for instance by adding a new layer
@@ -39,6 +45,12 @@ by setting \code{skip_mismatch=TRUE}. In this case any layer with
 mismatching weights will be skipped. A warning will be displayed
 for each skipped layer.
 }
+\section{Examples}{
+\if{html}{\out{<div class="sourceCode r">}}\preformatted{model |> load_model_weights("model.weights.h5")
+model |> load_model_weights("model.weights.json")
+}\if{html}{\out{</div>}}
+}
+
 \seealso{
 \itemize{
 \item \url{https://keras.io/api/models/model_saving_apis/weights_saving_and_loading#loadweights-method}
diff --git a/man/loss_dice.Rd b/man/loss_dice.Rd
index e4044e473..4620b21ec 100644
--- a/man/loss_dice.Rd
+++ b/man/loss_dice.Rd
@@ -63,7 +63,8 @@ y_pred <- array(c(0, 0.4, 0,   0,
                   1,   0, 1, 0.9), dim = c(2, 2, 2, 1))
 
 axis <- c(2, 3, 4)
-loss <- loss_dice(y_true, y_pred, axis = axis)
+loss_fn <- loss_dice(axis = axis, reduction = NULL)
+loss <- loss_fn(y_true, y_pred)
 stopifnot(shape(loss) == shape(2))
 loss
 }\if{html}{\out{</div>}}
@@ -72,7 +73,8 @@ loss
 
 }\if{html}{\out{</div>}}
 
-\if{html}{\out{<div class="sourceCode r">}}\preformatted{loss = loss_dice(y_true, y_pred)
+\if{html}{\out{<div class="sourceCode r">}}\preformatted{loss_fn <- loss_dice()
+loss <- loss_fn(y_true, y_pred)
 stopifnot(shape(loss) == shape())
 loss
 }\if{html}{\out{</div>}}
diff --git a/man/loss_sparse_categorical_crossentropy.Rd b/man/loss_sparse_categorical_crossentropy.Rd
index 8d1fe5f96..bd3678086 100644
--- a/man/loss_sparse_categorical_crossentropy.Rd
+++ b/man/loss_sparse_categorical_crossentropy.Rd
@@ -69,32 +69,29 @@ In the snippet below, there is a single floating point value per example for
 is \verb{[batch_size, num_classes]}.
 }
 \section{Examples}{
-\if{html}{\out{<div class="sourceCode r">}}\preformatted{y_true <- c(1, 2)
-y_pred <- rbind(c(0.05, 0.95, 0), c(0.1, 0.8, 0.1))
+\if{html}{\out{<div class="sourceCode r">}}\preformatted{y_true <- op_array(c(1L, 2L))
+y_pred <- op_array(rbind(c(0.05, 0.95, 0), c(0.1, 0.8, 0.1)))
 loss <- loss_sparse_categorical_crossentropy(y_true, y_pred)
 loss
 }\if{html}{\out{</div>}}
 
-\if{html}{\out{<div class="sourceCode">}}\preformatted{## tf.Tensor([0.05129344 2.3025851 ], shape=(2), dtype=float32)
+\if{html}{\out{<div class="sourceCode">}}\preformatted{## tf.Tensor([0.05129339 2.30258509], shape=(2), dtype=float64)
 
 }\if{html}{\out{</div>}}
 
-\if{html}{\out{<div class="sourceCode r">}}\preformatted{y_true <- c(1, 2)
-y_pred <- rbind(c(0.05, 0.95, 0), c(0.1, 0.8, 0.1))
+\if{html}{\out{<div class="sourceCode r">}}\preformatted{y_true <- op_array(c(1L, 2L))
+y_pred <- op_array(rbind(c(0.05, 0.95, 0), c(0.1, 0.8, 0.1)))
 # Using 'auto'/'sum_over_batch_size' reduction type.
 scce <- loss_sparse_categorical_crossentropy()
-scce(op_array(y_true), op_array(y_pred))
+scce(y_true, y_pred)
 }\if{html}{\out{</div>}}
 
 \if{html}{\out{<div class="sourceCode">}}\preformatted{## tf.Tensor(1.1769392, shape=(), dtype=float32)
 
 }\if{html}{\out{</div>}}
 
-\if{html}{\out{<div class="sourceCode r">}}\preformatted{# 1.177
-}\if{html}{\out{</div>}}
-
 \if{html}{\out{<div class="sourceCode r">}}\preformatted{# Calling with 'sample_weight'.
-scce(op_array(y_true), op_array(y_pred), sample_weight = op_array(c(0.3, 0.7)))
+scce(y_true, y_pred, sample_weight = op_array(c(0.3, 0.7)))
 }\if{html}{\out{</div>}}
 
 \if{html}{\out{<div class="sourceCode">}}\preformatted{## tf.Tensor(0.8135988, shape=(), dtype=float32)
@@ -103,28 +100,22 @@ scce(op_array(y_true), op_array(y_pred), sample_weight = op_array(c(0.3, 0.7)))
 
 \if{html}{\out{<div class="sourceCode r">}}\preformatted{# Using 'sum' reduction type.
 scce <- loss_sparse_categorical_crossentropy(reduction="sum")
-scce(op_array(y_true), op_array(y_pred))
+scce(y_true, y_pred)
 }\if{html}{\out{</div>}}
 
 \if{html}{\out{<div class="sourceCode">}}\preformatted{## tf.Tensor(2.3538785, shape=(), dtype=float32)
 
 }\if{html}{\out{</div>}}
 
-\if{html}{\out{<div class="sourceCode r">}}\preformatted{# 2.354
-}\if{html}{\out{</div>}}
-
 \if{html}{\out{<div class="sourceCode r">}}\preformatted{# Using 'none' reduction type.
 scce <- loss_sparse_categorical_crossentropy(reduction=NULL)
-scce(op_array(y_true), op_array(y_pred))
+scce(y_true, y_pred)
 }\if{html}{\out{</div>}}
 
 \if{html}{\out{<div class="sourceCode">}}\preformatted{## tf.Tensor([0.05129344 2.3025851 ], shape=(2), dtype=float32)
 
 }\if{html}{\out{</div>}}
 
-\if{html}{\out{<div class="sourceCode r">}}\preformatted{# array([0.0513, 2.303], dtype=float32)
-}\if{html}{\out{</div>}}
-
 Usage with the \code{compile()} API:
 
 \if{html}{\out{<div class="sourceCode r">}}\preformatted{model \%>\% compile(optimizer = 'sgd',
diff --git a/man/metric_binary_focal_crossentropy.Rd b/man/metric_binary_focal_crossentropy.Rd
index cbe191bea..70437522e 100644
--- a/man/metric_binary_focal_crossentropy.Rd
+++ b/man/metric_binary_focal_crossentropy.Rd
@@ -61,14 +61,17 @@ weight balancing factor for the binary classes 0 and 1 as follows:
 where \code{alpha} is a float in the range of \verb{[0, 1]}.
 }
 \section{Examples}{
-\if{html}{\out{<div class="sourceCode r">}}\preformatted{y_true <- rbind(c(0, 1), c(0, 0))
+\if{html}{\out{<div class="sourceCode {r}">}}\preformatted{y_true <- rbind(c(0, 1), c(0, 0))
 y_pred <- rbind(c(0.6, 0.4), c(0.4, 0.6))
-loss <- loss_binary_focal_crossentropy(y_true, y_pred, gamma=2)
-loss
-}\if{html}{\out{</div>}}
-
-\if{html}{\out{<div class="sourceCode">}}\preformatted{## tf.Tensor([0.32986468 0.20579839], shape=(2), dtype=float32)
+focal_loss <- loss_binary_focal_crossentropy(y_true, y_pred, gamma = 2)
+focal_loss
 
+# Compare with binary crossentropy.
+# Binary focal crossentropy emphasises harder examples, yielding a larger
+# relative loss where the model struggles.
+bce_loss <- loss_binary_crossentropy(y_true, y_pred)
+cbind(focal_loss, bce_loss,
+      ratio = focal_loss / bce_loss)
 }\if{html}{\out{</div>}}
 }
 
diff --git a/man/op_conv.Rd b/man/op_conv.Rd
index 98c280cbc..e52c91370 100644
--- a/man/op_conv.Rd
+++ b/man/op_conv.Rd
@@ -28,7 +28,7 @@ op_conv(
 \item{strides}{int or int tuple/list of \code{len(inputs_spatial_shape)},
 specifying the strides of the convolution along each spatial
 dimension. If \code{strides} is int, then every spatial dimension shares
-the same \code{strides}.}
+the same \code{strides}. Defaults to \code{1}.}
 
 \item{padding}{string, either \code{"valid"} or \code{"same"}. \code{"valid"} means no
 padding is applied, and \code{"same"} results in padding evenly to the
diff --git a/man/op_conv_transpose.Rd b/man/op_conv_transpose.Rd
index 3e0e0878b..9195fdc8c 100644
--- a/man/op_conv_transpose.Rd
+++ b/man/op_conv_transpose.Rd
@@ -7,7 +7,7 @@
 op_conv_transpose(
   inputs,
   kernel,
-  strides,
+  strides = 1L,
   padding = "valid",
   output_padding = NULL,
   data_format = NULL,
diff --git a/man/op_dot_product_attention.Rd b/man/op_dot_product_attention.Rd
index 280f79d46..5d02417a1 100644
--- a/man/op_dot_product_attention.Rd
+++ b/man/op_dot_product_attention.Rd
@@ -12,7 +12,8 @@ op_dot_product_attention(
   mask = NULL,
   scale = NULL,
   is_causal = FALSE,
-  flash_attention = NULL
+  flash_attention = NULL,
+  attn_logits_soft_cap = NULL
 )
 }
 \arguments{
@@ -43,6 +44,9 @@ to \code{1.0 / sqrt(H)}.}
 attempt to use flash attention if the required conditions are met.
 Typically, the inputs must be in float16 and bfloat16 dtype and the
 input layout requirements may vary depending on the backend.}
+
+\item{attn_logits_soft_cap}{Optional numeric cap on the attention logits before softmax. Only supported
+on the JAX TPU backend.}
 }
 \value{
 An array of the attention output with the same shape of \code{query}.
diff --git a/man/op_einsum.Rd b/man/op_einsum.Rd
index b38881f7d..4fb6cb228 100644
--- a/man/op_einsum.Rd
+++ b/man/op_einsum.Rd
@@ -13,7 +13,8 @@ summation) calculation is performed unless the explicit indicator
 \verb{->} is included as well as subscript labels of the precise
 output form.}
 
-\item{...}{The operands to compute the Einstein sum of.}
+\item{...}{The operands to compute the Einstein sum of as unnamed arguments.
+Additional named arguments are forwarded to the underlying backend.}
 }
 \value{
 The calculation based on the Einstein summation convention.
diff --git a/man/op_rms_normalization.Rd b/man/op_rms_normalization.Rd
index 0155baa17..73e5a7b3b 100644
--- a/man/op_rms_normalization.Rd
+++ b/man/op_rms_normalization.Rd
@@ -4,15 +4,14 @@
 \alias{op_rms_normalization}
 \title{Performs Root Mean Square (RMS) normalization on \code{x}.}
 \usage{
-op_rms_normalization(x, scale = 1L, axis = -1L, epsilon = NULL)
+op_rms_normalization(x, scale = NULL, axis = -1L, epsilon = NULL)
 }
 \arguments{
 \item{x}{Input tensor.}
 
 \item{scale}{Optional scaling factor for the normalization.}
 
-\item{axis}{The axis or axes along which to perform normalization.
-Default to -1.}
+\item{axis}{The axis or axes along which to perform normalization. Defaults to \code{-1}.}
 
 \item{epsilon}{A lower bound value for the norm.
 Defaults to \code{config_epsilon()}.}
@@ -30,10 +29,33 @@ The operation is different from LayerNormalization with RMS scaling.
 It is defined as \code{rms_normalization(x) = x * rsqrt(mean(square(x))) * scale}
 }
 \section{Examples}{
-\if{html}{\out{<div class="sourceCode python">}}\preformatted{x <- random_uniform(c(1, 10))
-x_norm <- op_rms_normalization(x, scale = 10)
-x_norm
+\if{html}{\out{<div class="sourceCode r">}}\preformatted{x <- random_normal(c(1, 10))
+op_rms_normalization(x)
 }\if{html}{\out{</div>}}
+
+\if{html}{\out{<div class="sourceCode">}}\preformatted{## tf.Tensor(
+## [[-1.1701705   0.9987454   0.27809578 -0.18452731  1.5731977   1.8934134
+##    0.10656697  1.0256069   0.6185074  -0.12678868]], shape=(1, 10), dtype=float32)
+
+}\if{html}{\out{</div>}}
+
+\if{html}{\out{<div class="sourceCode r">}}\preformatted{op_rms_normalization(x, scale = 10)
+}\if{html}{\out{</div>}}
+
+\if{html}{\out{<div class="sourceCode">}}\preformatted{## tf.Tensor(
+## [[-11.701706    9.987453    2.7809577  -1.845273   15.731977   18.934134
+##     1.0656698  10.256069    6.185074   -1.2678868]], shape=(1, 10), dtype=float32)
+
+}\if{html}{\out{</div>}}
+
+[-1.1701705   0.9987454   0.27809578 -0.18452731  1.5731977   1.8934134
+\subsection{0.10656697  1.0256069   0.6185074  -0.12678868]: R:-1.1701705\%20\%20\%200.9987454\%20\%20\%200.27809578\%20-0.18452731\%20\%201.5731977\%20\%20\%201.8934134\%0A##\%20\%20\%20\%200.10656697\%20\%201.0256069\%20\%20\%200.6185074\%20\%20-0.12678868}{
+
+[-11.701706    9.987453    2.7809577  -1.845273   15.731977   18.934134
+}
+
+\subsection{1.0656698  10.256069    6.185074   -1.2678868]: R:-11.701706\%20\%20\%20\%209.987453\%20\%20\%20\%202.7809577\%20\%20-1.845273\%20\%20\%2015.731977\%20\%20\%2018.934134\%0A##\%20\%20\%20\%20\%201.0656698\%20\%2010.256069\%20\%20\%20\%206.185074\%20\%20\%20-1.2678868}{
+}
 }
 
 \seealso{
diff --git a/man/optimizer_lion.Rd b/man/optimizer_lion.Rd
index e78b7f126..876064338 100644
--- a/man/optimizer_lion.Rd
+++ b/man/optimizer_lion.Rd
@@ -102,13 +102,13 @@ an \code{Optimizer} instance
 \description{
 The Lion optimizer is a stochastic-gradient-descent method that uses the
 sign operator to control the magnitude of the update, unlike other adaptive
-optimizers such as Adam that rely on second-order moments. This make
+optimizers such as Adam that rely on second-order moments. This makes
 Lion more memory-efficient as it only keeps track of the momentum. According
 to the authors (see reference), its performance gain over Adam grows with
 the batch size. Because the update of Lion is produced through the sign
 operation, resulting in a larger norm, a suitable learning rate for Lion is
 typically 3-10x smaller than that for AdamW. The weight decay for Lion
-should be in turn 3-10x larger than that for AdamW to maintain a
+should in turn be 3-10x larger than that for AdamW to maintain a
 similar strength (lr * wd).
 }
 \section{References}{
diff --git a/man/save_model_weights.Rd b/man/save_model_weights.Rd
index b35ee7849..33dddb421 100644
--- a/man/save_model_weights.Rd
+++ b/man/save_model_weights.Rd
@@ -2,26 +2,51 @@
 % Please edit documentation in R/model-persistence.R
 \name{save_model_weights}
 \alias{save_model_weights}
-\title{Saves all layer weights to a \code{.weights.h5} file.}
+\title{Saves all weights to a single file or sharded files.}
 \usage{
-save_model_weights(model, filepath, overwrite = FALSE)
+save_model_weights(model, filepath, overwrite = FALSE, max_shard_size = NULL)
 }
 \arguments{
-\item{model}{A keras Model object}
+\item{model}{A keras Model object.}
 
-\item{filepath}{string.
-Path where to save the model. Must end in \code{.weights.h5}.}
+\item{filepath}{Path where the weights will be saved. Accepts \code{.weights.h5}, or when sharding
+is enabled, a \code{.weights.json} manifest path. If \code{.weights.h5} is provided
+while sharding, the filename will be overridden to end in \code{.weights.json}.}
 
-\item{overwrite}{Whether we should overwrite any existing model
-at the target location, or instead ask the user
-via an interactive prompt.}
+\item{overwrite}{Whether to overwrite any existing weights at the target location, or instead
+ask the user via an interactive prompt.}
+
+\item{max_shard_size}{Numeric size in GB for each sharded file. Use \code{NULL} to disable sharding.}
 }
 \value{
 This is called primarily for side effects. \code{model} is returned,
 invisibly, to enable usage with the pipe.
 }
 \description{
-Saves all layer weights to a \code{.weights.h5} file.
+By default, the weights are saved in a single \code{.weights.h5} file. Enable
+sharding via \code{max_shard_size} to split weights across multiple files (in GB)
+and produce a \code{.weights.json} manifest that tracks shard metadata.
+
+The saved sharded files contain:
+\itemize{
+\item \verb{*.weights.json}: configuration file containing \code{metadata} and
+\code{weight_map} entries.
+\item \verb{*_xxxxxx.weights.h5}: weight shards limited by \code{max_shard_size}.
+}
+
+\if{html}{\out{<div class="sourceCode r">}}\preformatted{model <-
+  keras_model_sequential(input_shape = 2) |>
+  layer_dense(4)
+
+path_h5 <- tempfile(fileext = ".weights.h5")
+path_json <- tempfile(fileext = ".weights.json")
+
+model |> save_model_weights(path_h5)
+model |> save_model_weights(path_json, max_shard_size = 0.01)
+
+model |> load_model_weights(path_h5)
+model |> load_model_weights(path_json)
+}\if{html}{\out{</div>}}
 }
 \seealso{
 \itemize{

From f332596a4183f8a1d8b79f9d0eadf48409ac8cc6 Mon Sep 17 00:00:00 2001
From: Tomasz Kalinowski <kalinowskit@gmail.com>
Date: Fri, 26 Sep 2025 08:28:52 -0400
Subject: [PATCH 39/81] add `op_sparse_sigmoid()` and
 `activation_sparse_sigmoid()`

---
 .tether/man/keras.activations.txt |  1 +
 .tether/man/keras.ops.nn.txt      |  1 +
 .tether/man/keras.ops.txt         |  1 +
 R/activations.R                   | 24 ++++++++++++++++++++++++
 R/ops-nn.R                        | 29 +++++++++++++++++++++++++++++
 5 files changed, 56 insertions(+)

diff --git a/.tether/man/keras.activations.txt b/.tether/man/keras.activations.txt
index 48851e264..064021d14 100644
--- a/.tether/man/keras.activations.txt
+++ b/.tether/man/keras.activations.txt
@@ -31,6 +31,7 @@ softmax(x, axis=-1)
 softplus(x)
 softsign(x)
 sparse_plus(x)
+sparse_sigmoid(x)
 sparsemax(x, axis=-1)
 squareplus(x, b=4)
 swish(x)
diff --git a/.tether/man/keras.ops.nn.txt b/.tether/man/keras.ops.nn.txt
index a5655066d..59f5d89ee 100644
--- a/.tether/man/keras.ops.nn.txt
+++ b/.tether/man/keras.ops.nn.txt
@@ -159,6 +159,7 @@ sparse_categorical_crossentropy(
   axis=-1
 )
 sparse_plus(x)
+sparse_sigmoid(x)
 sparsemax(x, axis=-1)
 squareplus(x, b=4)
 swish(x)
diff --git a/.tether/man/keras.ops.txt b/.tether/man/keras.ops.txt
index 71440537f..a2738b3ce 100644
--- a/.tether/man/keras.ops.txt
+++ b/.tether/man/keras.ops.txt
@@ -608,6 +608,7 @@ sparse_categorical_crossentropy(
   axis=-1
 )
 sparse_plus(x)
+sparse_sigmoid(x)
 sparsemax(x, axis=-1)
 split(
   x,
diff --git a/R/activations.R b/R/activations.R
index 117da9ee1..cc91b6124 100644
--- a/R/activations.R
+++ b/R/activations.R
@@ -793,6 +793,30 @@ function (x)
     keras$activations$sparse_plus(x)
 }
 
+#' Sparse sigmoid activation function.
+#'
+#' @description
+#' It is defined as
+#'
+#' `f(x) = 0` for `x <= -1`,
+#' `f(x) = 0.5 * (x + 1)` for `-1 < x < 1`,
+#' `f(x) = 1` for `x >= 1`.
+#'
+#' # Reference
+#' - [M. Blondel, A. F. T. Martins, V. Niculae, 2019](https://arxiv.org/pdf/1901.02324)
+#'
+#' @param x
+#' Input tensor.
+#'
+#' @family activations
+#' @inherit activation_elu return
+#' @export
+#' @tether keras.activations.sparse_sigmoid
+activation_sparse_sigmoid <-
+function (x) {
+  keras$activations$sparse_sigmoid(x)
+}
+
 #' Sparsemax activation function.
 #'
 #' @description
diff --git a/R/ops-nn.R b/R/ops-nn.R
index dbac3510b..4bf672632 100644
--- a/R/ops-nn.R
+++ b/R/ops-nn.R
@@ -485,6 +485,35 @@ op_sparse_plus <-
 function (x)
 ops$sparse_plus(x)
 
+#' Sparse sigmoid activation function.
+#'
+#' @description
+#' It is defined as
+#'
+#' `f(x) = 0` for `x <= -1`,
+#' `f(x) = 0.5 * (x + 1)` for `-1 < x < 1`,
+#' `f(x) = 1` for `x >= 1`.
+#'
+#' # Examples
+#' ```{r}
+#' x <- op_array(c(-1.0, 0.0, 1.0))
+#' op_sparse_sigmoid(x)
+#' ```
+#'
+#' @returns
+#' A tensor with the same shape as `x`.
+#'
+#' @param x
+#' Input tensor.
+#'
+#' @export
+#' @tether keras.ops.sparse_sigmoid
+#' @family nn ops
+#' @family ops
+op_sparse_sigmoid <-
+function (x)
+ops$sparse_sigmoid(x)
+
 #' Sparsemax activation function.
 #'
 #' @description

From 147e48e8d523ad140598d676f217aed1881697fe Mon Sep 17 00:00:00 2001
From: Tomasz Kalinowski <kalinowskit@gmail.com>
Date: Fri, 26 Sep 2025 08:29:12 -0400
Subject: [PATCH 40/81] add `tools/make-r-wrapper.R` script

---
 tools/make-r-wrapper.R | 17 +++++++++++++++++
 1 file changed, 17 insertions(+)
 create mode 100755 tools/make-r-wrapper.R

diff --git a/tools/make-r-wrapper.R b/tools/make-r-wrapper.R
new file mode 100755
index 000000000..076cd0028
--- /dev/null
+++ b/tools/make-r-wrapper.R
@@ -0,0 +1,17 @@
+#!/usr/bin/env Rscript
+
+if (basename(getwd()) != "keras3") {
+  stop("must be called from the keras3 package directory")
+}
+Sys.setenv("TF_CPP_MIN_LOG_LEVEL" = "3", UV_NO_PROGRESS = 1)
+# sink(nullfile())
+suppressWarnings(suppressMessages(
+  envir::attach_source("tools/utils.R")
+))
+# sink()
+
+# Example usage:
+# tools/make-r-wrapper.R keras.ops.add
+for (py_expr in commandArgs(TRUE)) {
+  writeLines(mk_export(py_expr)$dump)
+}

From 8d0e5ab5051dc5c411c06ccb6a520203c8bf6d18 Mon Sep 17 00:00:00 2001
From: Tomasz Kalinowski <kalinowskit@gmail.com>
Date: Fri, 26 Sep 2025 08:36:21 -0400
Subject: [PATCH 41/81] add `op_angle()`

---
 .tether/man/keras.ops.numpy.txt |  1 +
 .tether/man/keras.ops.txt       |  1 +
 R/ops.R                         | 26 ++++++++++++++++++++++++++
 3 files changed, 28 insertions(+)

diff --git a/.tether/man/keras.ops.numpy.txt b/.tether/man/keras.ops.numpy.txt
index be45c6fe8..7cfb6c0db 100644
--- a/.tether/man/keras.ops.numpy.txt
+++ b/.tether/man/keras.ops.numpy.txt
@@ -16,6 +16,7 @@ amin(
   axis=None,
   keepdims=False
 )
+angle(x)
 any(
   x,
   axis=None,
diff --git a/.tether/man/keras.ops.txt b/.tether/man/keras.ops.txt
index a2738b3ce..5161581e7 100644
--- a/.tether/man/keras.ops.txt
+++ b/.tether/man/keras.ops.txt
@@ -16,6 +16,7 @@ amin(
   axis=None,
   keepdims=False
 )
+angle(x)
 any(
   x,
   axis=None,
diff --git a/R/ops.R b/R/ops.R
index 956504615..d6745e32a 100644
--- a/R/ops.R
+++ b/R/ops.R
@@ -4026,6 +4026,32 @@ function (x)
 ops$conjugate(x)
 
 
+#' Element-wise angle of a complex tensor.
+#'
+#' @description
+#' Returns the phase angle (in radians) of each element in `x`.
+#'
+#' # Examples
+#' ```{r}
+#' x <- op_convert_to_tensor(matrix(c(1 + 3i, 2 - 5i, 4 - 3i, 3 + 2i), nrow = 2))
+#' op_angle(x)
+#' ```
+#'
+#' @returns
+#' Tensor with the same shape as `x`, containing element-wise angles.
+#'
+#' @param x
+#' Input tensor. Can be real or complex.
+#'
+#' @export
+#' @family numpy ops
+#' @family ops
+#' @tether keras.ops.angle
+op_angle <-
+function (x)
+ops$angle(x)
+
+
 #' Returns a copy of `x`.
 #'
 #' @returns

From a49f307c22c8bcb41dff410ffe5f721eeb2e0ce4 Mon Sep 17 00:00:00 2001
From: Tomasz Kalinowski <kalinowskit@gmail.com>
Date: Fri, 26 Sep 2025 08:41:37 -0400
Subject: [PATCH 42/81] add `Arg()` method for tensors that calls `op_angle()`

---
 R/s3-methods.R | 9 ++++++++-
 1 file changed, 8 insertions(+), 1 deletion(-)

diff --git a/R/s3-methods.R b/R/s3-methods.R
index 64b8c5fd2..af8cfebee 100644
--- a/R/s3-methods.R
+++ b/R/s3-methods.R
@@ -11,6 +11,14 @@
 }
 
 
+#' @export
+Arg.keras.src.backend.common.keras_tensor.KerasTensor <- function(z) {
+  op_angle(z)
+}
+#' @export
+Arg.keras.src.backend.Tensor <- Arg.keras.src.backend.common.keras_tensor.KerasTensor
+
+
 
 #' @export
 as.array.keras.src.backend.common.variables.KerasVariable <- function(x, ...) {
@@ -102,4 +110,3 @@ py_to_r__keras.src.utils.tracking.TrackedSet <- function(x) import("builtins")$l
 # }
 # rm(list = c("generic", "cls"))
 
-

From 73a17e23c677fcbffb4f9a9bb0cbdded57cacfec Mon Sep 17 00:00:00 2001
From: Tomasz Kalinowski <kalinowskit@gmail.com>
Date: Fri, 26 Sep 2025 08:59:45 -0400
Subject: [PATCH 43/81] add `op_bartlett()`

---
 .tether/man/keras.ops.numpy.txt |  1 +
 .tether/man/keras.ops.txt       |  1 +
 R/ops-numpy.R                   | 25 +++++++++++++++++++++++++
 3 files changed, 27 insertions(+)

diff --git a/.tether/man/keras.ops.numpy.txt b/.tether/man/keras.ops.numpy.txt
index 7cfb6c0db..83356c856 100644
--- a/.tether/man/keras.ops.numpy.txt
+++ b/.tether/man/keras.ops.numpy.txt
@@ -62,6 +62,7 @@ average(
   axis=None,
   weights=None
 )
+bartlett(x)
 bincount(
   x,
   weights=None,
diff --git a/.tether/man/keras.ops.txt b/.tether/man/keras.ops.txt
index 5161581e7..336bfd242 100644
--- a/.tether/man/keras.ops.txt
+++ b/.tether/man/keras.ops.txt
@@ -75,6 +75,7 @@ average_pool(
   padding='valid',
   data_format=None
 )
+bartlett(x)
 batch_normalization(
   x,
   mean,
diff --git a/R/ops-numpy.R b/R/ops-numpy.R
index e8c47c92e..749ee11ea 100644
--- a/R/ops-numpy.R
+++ b/R/ops-numpy.R
@@ -71,6 +71,31 @@ function (x, kth, axis = -1L, zero_indexed = FALSE)
 }
 
 
+#' Bartlett window function.
+#'
+#' @description
+#' Generates a Bartlett (triangular) window of length `x`.
+#'
+#' # Examples
+#' ```{r}
+#' op_bartlett(5)
+#' ```
+#'
+#' @returns
+#' A 1D tensor containing the window values.
+#'
+#' @param x
+#' Length of the window. Must be a positive integer.
+#'
+#' @export
+#' @family numpy ops
+#' @family ops
+#' @tether keras.ops.bartlett
+op_bartlett <-
+function (x)
+ops$bartlett(as_integer(x))
+
+
 #' Compute the bit-wise AND of two arrays element-wise.
 #'
 #' @description

From 69c831edfa6c406f0a375a9f9919337575b797d4 Mon Sep 17 00:00:00 2001
From: Tomasz Kalinowski <kalinowskit@gmail.com>
Date: Fri, 26 Sep 2025 09:00:20 -0400
Subject: [PATCH 44/81] add `op_blackman()`

---
 .tether/man/keras.ops.numpy.txt |  1 +
 .tether/man/keras.ops.txt       |  1 +
 R/ops-numpy.R                   | 25 +++++++++++++++++++++++++
 3 files changed, 27 insertions(+)

diff --git a/.tether/man/keras.ops.numpy.txt b/.tether/man/keras.ops.numpy.txt
index 83356c856..c7943c9bd 100644
--- a/.tether/man/keras.ops.numpy.txt
+++ b/.tether/man/keras.ops.numpy.txt
@@ -76,6 +76,7 @@ bitwise_not(x)
 bitwise_or(x, y)
 bitwise_right_shift(x, y)
 bitwise_xor(x, y)
+blackman(x)
 broadcast_to(x, shape)
 ceil(x)
 clip(
diff --git a/.tether/man/keras.ops.txt b/.tether/man/keras.ops.txt
index 336bfd242..3fb08a954 100644
--- a/.tether/man/keras.ops.txt
+++ b/.tether/man/keras.ops.txt
@@ -103,6 +103,7 @@ bitwise_not(x)
 bitwise_or(x, y)
 bitwise_right_shift(x, y)
 bitwise_xor(x, y)
+blackman(x)
 broadcast_to(x, shape)
 cast(x, dtype)
 categorical_crossentropy(
diff --git a/R/ops-numpy.R b/R/ops-numpy.R
index 749ee11ea..01222f2d6 100644
--- a/R/ops-numpy.R
+++ b/R/ops-numpy.R
@@ -96,6 +96,31 @@ function (x)
 ops$bartlett(as_integer(x))
 
 
+#' Blackman window function.
+#'
+#' @description
+#' Generates a Blackman window of length `x`.
+#'
+#' # Examples
+#' ```{r}
+#' op_blackman(5)
+#' ```
+#'
+#' @returns
+#' A 1D tensor containing the window values.
+#'
+#' @param x
+#' Length of the window. Must be a positive integer.
+#'
+#' @export
+#' @family numpy ops
+#' @family ops
+#' @tether keras.ops.blackman
+op_blackman <-
+function (x)
+ops$blackman(as_integer(x))
+
+
 #' Compute the bit-wise AND of two arrays element-wise.
 #'
 #' @description

From 2b6f3dfb7badfc876a4a5d67b8eee04530366ed2 Mon Sep 17 00:00:00 2001
From: Tomasz Kalinowski <kalinowskit@gmail.com>
Date: Fri, 26 Sep 2025 09:06:08 -0400
Subject: [PATCH 45/81] add `op_hamming()`

---
 .tether/man/keras.ops.numpy.txt |  1 +
 .tether/man/keras.ops.txt       |  1 +
 R/ops-numpy.R                   | 25 +++++++++++++++++++++++++
 3 files changed, 27 insertions(+)

diff --git a/.tether/man/keras.ops.numpy.txt b/.tether/man/keras.ops.numpy.txt
index c7943c9bd..0d131345c 100644
--- a/.tether/man/keras.ops.numpy.txt
+++ b/.tether/man/keras.ops.numpy.txt
@@ -160,6 +160,7 @@ full_like(
 get_item(x, key)
 greater(x1, x2)
 greater_equal(x1, x2)
+hamming(x)
 histogram(
   x,
   bins=10,
diff --git a/.tether/man/keras.ops.txt b/.tether/man/keras.ops.txt
index 3fb08a954..27f9b7323 100644
--- a/.tether/man/keras.ops.txt
+++ b/.tether/man/keras.ops.txt
@@ -282,6 +282,7 @@ get_item(x, key)
 glu(x, axis=-1)
 greater(x1, x2)
 greater_equal(x1, x2)
+hamming(x)
 hard_shrink(x, threshold=0.5)
 hard_sigmoid(x)
 hard_silu(x)
diff --git a/R/ops-numpy.R b/R/ops-numpy.R
index 01222f2d6..fff1f3e90 100644
--- a/R/ops-numpy.R
+++ b/R/ops-numpy.R
@@ -121,6 +121,31 @@ function (x)
 ops$blackman(as_integer(x))
 
 
+#' Hamming window function.
+#'
+#' @description
+#' Generates a Hamming window of length `x`.
+#'
+#' # Examples
+#' ```{r}
+#' op_hamming(5)
+#' ```
+#'
+#' @returns
+#' A 1D tensor containing the window values.
+#'
+#' @param x
+#' Length of the window. Must be a positive integer.
+#'
+#' @export
+#' @family numpy ops
+#' @family ops
+#' @tether keras.ops.hamming
+op_hamming <-
+function (x)
+ops$hamming(as_integer(x))
+
+
 #' Compute the bit-wise AND of two arrays element-wise.
 #'
 #' @description

From 0383f949f0c66a78e88521587049026485588bef Mon Sep 17 00:00:00 2001
From: Tomasz Kalinowski <kalinowskit@gmail.com>
Date: Fri, 26 Sep 2025 09:09:33 -0400
Subject: [PATCH 46/81] add `op_hanning()`

---
 .tether/man/keras.ops.numpy.txt |  1 +
 .tether/man/keras.ops.txt       |  1 +
 R/ops-numpy.R                   | 25 +++++++++++++++++++++++++
 3 files changed, 27 insertions(+)

diff --git a/.tether/man/keras.ops.numpy.txt b/.tether/man/keras.ops.numpy.txt
index 0d131345c..9cad39204 100644
--- a/.tether/man/keras.ops.numpy.txt
+++ b/.tether/man/keras.ops.numpy.txt
@@ -161,6 +161,7 @@ get_item(x, key)
 greater(x1, x2)
 greater_equal(x1, x2)
 hamming(x)
+hanning(x)
 histogram(
   x,
   bins=10,
diff --git a/.tether/man/keras.ops.txt b/.tether/man/keras.ops.txt
index 27f9b7323..e7ffd1e8f 100644
--- a/.tether/man/keras.ops.txt
+++ b/.tether/man/keras.ops.txt
@@ -283,6 +283,7 @@ glu(x, axis=-1)
 greater(x1, x2)
 greater_equal(x1, x2)
 hamming(x)
+hanning(x)
 hard_shrink(x, threshold=0.5)
 hard_sigmoid(x)
 hard_silu(x)
diff --git a/R/ops-numpy.R b/R/ops-numpy.R
index fff1f3e90..b443ac883 100644
--- a/R/ops-numpy.R
+++ b/R/ops-numpy.R
@@ -146,6 +146,31 @@ function (x)
 ops$hamming(as_integer(x))
 
 
+#' Hanning window function.
+#'
+#' @description
+#' Generates a Hanning window of length `x`.
+#'
+#' # Examples
+#' ```{r}
+#' op_hanning(5)
+#' ```
+#'
+#' @returns
+#' A 1D tensor containing the window values.
+#'
+#' @param x
+#' Length of the window. Must be a positive integer.
+#'
+#' @export
+#' @family numpy ops
+#' @family ops
+#' @tether keras.ops.hanning
+op_hanning <-
+function (x)
+ops$hanning(as_integer(x))
+
+
 #' Compute the bit-wise AND of two arrays element-wise.
 #'
 #' @description

From 0d6cf91bf095b99bafda7fb02aa66e6f08e96f11 Mon Sep 17 00:00:00 2001
From: Tomasz Kalinowski <kalinowskit@gmail.com>
Date: Fri, 26 Sep 2025 09:12:36 -0400
Subject: [PATCH 47/81] add `op_heaviside()`

---
 .tether/man/keras.ops.numpy.txt |  1 +
 .tether/man/keras.ops.txt       |  1 +
 R/ops-numpy.R                   | 29 +++++++++++++++++++++++++++++
 3 files changed, 31 insertions(+)

diff --git a/.tether/man/keras.ops.numpy.txt b/.tether/man/keras.ops.numpy.txt
index 9cad39204..6ec3288f9 100644
--- a/.tether/man/keras.ops.numpy.txt
+++ b/.tether/man/keras.ops.numpy.txt
@@ -162,6 +162,7 @@ greater(x1, x2)
 greater_equal(x1, x2)
 hamming(x)
 hanning(x)
+heaviside(x1, x2)
 histogram(
   x,
   bins=10,
diff --git a/.tether/man/keras.ops.txt b/.tether/man/keras.ops.txt
index e7ffd1e8f..fa612bb5b 100644
--- a/.tether/man/keras.ops.txt
+++ b/.tether/man/keras.ops.txt
@@ -289,6 +289,7 @@ hard_sigmoid(x)
 hard_silu(x)
 hard_swish(x)
 hard_tanh(x)
+heaviside(x1, x2)
 histogram(
   x,
   bins=10,
diff --git a/R/ops-numpy.R b/R/ops-numpy.R
index b443ac883..8519c9e55 100644
--- a/R/ops-numpy.R
+++ b/R/ops-numpy.R
@@ -171,6 +171,35 @@ function (x)
 ops$hanning(as_integer(x))
 
 
+#' Heaviside step function.
+#'
+#' @description
+#' Computes the Heaviside step function element-wise.
+#'
+#' # Examples
+#' ```{r}
+#' x1 <- op_array(c(-2, 0, 3))
+#' op_heaviside(x1, 0.5)
+#' ```
+#'
+#' @returns
+#' A tensor broadcast from `x1` and `x2` containing `0`, `1`, or `x2`.
+#'
+#' @param x1
+#' Tensor input.
+#'
+#' @param x2
+#' Value to use when `x1 == 0`.
+#'
+#' @export
+#' @family numpy ops
+#' @family ops
+#' @tether keras.ops.heaviside
+op_heaviside <-
+function (x1, x2)
+ops$heaviside(x1, x2)
+
+
 #' Compute the bit-wise AND of two arrays element-wise.
 #'
 #' @description

From 90e6078e9a377c47dfb150818f468278c5dc9415 Mon Sep 17 00:00:00 2001
From: Tomasz Kalinowski <kalinowskit@gmail.com>
Date: Fri, 26 Sep 2025 09:28:32 -0400
Subject: [PATCH 48/81] expand docs for window functions

---
 R/ops-numpy.R | 15 ++++++++++-----
 1 file changed, 10 insertions(+), 5 deletions(-)

diff --git a/R/ops-numpy.R b/R/ops-numpy.R
index 8519c9e55..a5dceb59b 100644
--- a/R/ops-numpy.R
+++ b/R/ops-numpy.R
@@ -74,7 +74,7 @@ function (x, kth, axis = -1L, zero_indexed = FALSE)
 #' Bartlett window function.
 #'
 #' @description
-#' Generates a Bartlett (triangular) window of length `x`.
+#' The Bartlett window is a triangular window that rises then falls linearly.
 #'
 #' # Examples
 #' ```{r}
@@ -99,7 +99,7 @@ ops$bartlett(as_integer(x))
 #' Blackman window function.
 #'
 #' @description
-#' Generates a Blackman window of length `x`.
+#' The Blackman window is a taper formed by using a weighted cosine.
 #'
 #' # Examples
 #' ```{r}
@@ -124,7 +124,8 @@ ops$blackman(as_integer(x))
 #' Hamming window function.
 #'
 #' @description
-#' Generates a Hamming window of length `x`.
+#' The Hamming window is defined as:
+#' `w[n] = 0.54 - 0.46 * cos(2 * pi * n / (N - 1))` for `0 <= n <= N - 1`.
 #'
 #' # Examples
 #' ```{r}
@@ -149,7 +150,8 @@ ops$hamming(as_integer(x))
 #' Hanning window function.
 #'
 #' @description
-#' Generates a Hanning window of length `x`.
+#' The Hanning window is defined as:
+#' `w[n] = 0.5 - 0.5 * cos(2 * pi * n / (N - 1))` for `0 <= n <= N - 1`.
 #'
 #' # Examples
 #' ```{r}
@@ -174,7 +176,10 @@ ops$hanning(as_integer(x))
 #' Heaviside step function.
 #'
 #' @description
-#' Computes the Heaviside step function element-wise.
+#' The Heaviside step function is defined as:
+#' `heaviside(x1, x2) = 0` if `x1 < 0`,
+#' `heaviside(x1, x2) = 1` if `x1 > 0`, and
+#' `heaviside(x1, x2) = x2` if `x1 == 0`.
 #'
 #' # Examples
 #' ```{r}

From 5793c85d84a5ecc6355dea247aed55a0af610646 Mon Sep 17 00:00:00 2001
From: Tomasz Kalinowski <kalinowskit@gmail.com>
Date: Fri, 26 Sep 2025 09:29:27 -0400
Subject: [PATCH 49/81] add `op_kaiser()`

---
 .tether/man/keras.ops.numpy.txt |  1 +
 .tether/man/keras.ops.txt       |  1 +
 R/ops-numpy.R                   | 30 ++++++++++++++++++++++++++++++
 3 files changed, 32 insertions(+)

diff --git a/.tether/man/keras.ops.numpy.txt b/.tether/man/keras.ops.numpy.txt
index 6ec3288f9..7a55f1288 100644
--- a/.tether/man/keras.ops.numpy.txt
+++ b/.tether/man/keras.ops.numpy.txt
@@ -182,6 +182,7 @@ isclose(
 isfinite(x)
 isinf(x)
 isnan(x)
+kaiser(x, beta)
 left_shift(x, y)
 less(x1, x2)
 less_equal(x1, x2)
diff --git a/.tether/man/keras.ops.txt b/.tether/man/keras.ops.txt
index fa612bb5b..4d14ff513 100644
--- a/.tether/man/keras.ops.txt
+++ b/.tether/man/keras.ops.txt
@@ -328,6 +328,7 @@ istft(
   window='hann',
   center=True
 )
+kaiser(x, beta)
 leaky_relu(x, negative_slope=0.2)
 left_shift(x, y)
 less(x1, x2)
diff --git a/R/ops-numpy.R b/R/ops-numpy.R
index a5dceb59b..e44196397 100644
--- a/R/ops-numpy.R
+++ b/R/ops-numpy.R
@@ -205,6 +205,36 @@ function (x1, x2)
 ops$heaviside(x1, x2)
 
 
+#' Kaiser window function.
+#'
+#' @description
+#' The Kaiser window is defined as:
+#' `w[n] = I0(beta * sqrt(1 - (2 * n / (N - 1) - 1)^2)) / I0(beta)` where
+#' `I0` is the modified zeroth-order Bessel function of the first kind.
+#'
+#' # Examples
+#' ```{r}
+#' op_kaiser(5, beta = 14)
+#' ```
+#'
+#' @returns
+#' A 1D tensor containing the window values.
+#'
+#' @param x
+#' Length of the window. Must be a positive integer.
+#'
+#' @param beta
+#' Shape parameter for the window.
+#'
+#' @export
+#' @family numpy ops
+#' @family ops
+#' @tether keras.ops.kaiser
+op_kaiser <-
+function (x, beta)
+ops$kaiser(as_integer(x), beta)
+
+
 #' Compute the bit-wise AND of two arrays element-wise.
 #'
 #' @description

From 5bd398709739c290aec01230a095ec02f14f150f Mon Sep 17 00:00:00 2001
From: Tomasz Kalinowski <kalinowskit@gmail.com>
Date: Fri, 26 Sep 2025 09:39:29 -0400
Subject: [PATCH 50/81] add `op_cbrt` `op_corrcoef` `op_deg2rad`

---
 .tether/man/keras.ops.numpy.txt |  3 ++
 .tether/man/keras.ops.txt       |  3 ++
 R/ops-numpy.R                   | 77 +++++++++++++++++++++++++++++++++
 tools/make-r-wrapper.R          |  2 +-
 4 files changed, 84 insertions(+), 1 deletion(-)

diff --git a/.tether/man/keras.ops.numpy.txt b/.tether/man/keras.ops.numpy.txt
index 7a55f1288..f2a7ad91c 100644
--- a/.tether/man/keras.ops.numpy.txt
+++ b/.tether/man/keras.ops.numpy.txt
@@ -78,6 +78,7 @@ bitwise_right_shift(x, y)
 bitwise_xor(x, y)
 blackman(x)
 broadcast_to(x, shape)
+cbrt(x)
 ceil(x)
 clip(
   x,
@@ -95,6 +96,7 @@ correlate(
 )
 cos(x)
 cosh(x)
+corrcoef(x)
 count_nonzero(x, axis=None)
 cross(
   x1,
@@ -117,6 +119,7 @@ cumsum(
 diag(x, k=0)
 diagflat(x, k=0)
 diagonal(
+deg2rad(x)
   x,
   offset=0,
   axis1=0,
diff --git a/.tether/man/keras.ops.txt b/.tether/man/keras.ops.txt
index 4d14ff513..3603f3ef9 100644
--- a/.tether/man/keras.ops.txt
+++ b/.tether/man/keras.ops.txt
@@ -112,6 +112,7 @@ categorical_crossentropy(
   from_logits=False,
   axis=-1
 )
+cbrt(x)
 ceil(x)
 celu(x, alpha=1.0)
 cholesky(x)
@@ -153,6 +154,7 @@ convert_to_tensor(
   ragged=None
 )
 copy(x)
+corrcoef(x)
 correlate(
   x1,
   x2,
@@ -196,6 +198,7 @@ cumsum(
   dtype=None
 )
 custom_gradient(f)
+deg2rad(x)
 depthwise_conv(
   inputs,
   kernel,
diff --git a/R/ops-numpy.R b/R/ops-numpy.R
index e44196397..bcc5d581c 100644
--- a/R/ops-numpy.R
+++ b/R/ops-numpy.R
@@ -71,6 +71,83 @@ function (x, kth, axis = -1L, zero_indexed = FALSE)
 }
 
 
+#' Compute the Pearson correlation coefficient matrix.
+#'
+#' @returns
+#' A tensor of shape `(N, N)` representing the correlation matrix.
+#'
+#' @param x
+#' A 2D tensor of shape `(N, D)`, where `N` is the number of variables
+#' and `D` is the number of observations.
+#'
+#' # Examples
+#' ```{r}
+#' x <- op_array(matrix(c(1, 2, 3,
+#'                        2, 3, 4), nrow = 2, byrow = TRUE))
+#' op_corrcoef(x)
+#' ```
+#'
+#' @export
+#' @family numpy ops
+#' @family ops
+#' @tether keras.ops.corrcoef
+op_corrcoef <-
+function (x)
+ops$corrcoef(x)
+
+
+#' Computes the cube root of the input tensor, element-wise.
+#'
+#' @description
+#' Returns the real-valued cube root of `x`, handling negative inputs in the
+#' real domain.
+#'
+#' @returns
+#' A tensor containing the cube root of each element in `x`.
+#'
+#' @param x
+#' Input tensor.
+#'
+#' # Examples
+#' ```{r}
+#' op_cbrt(c(-8, 0, 8))
+#' ```
+#'
+#' @export
+#' @family numpy ops
+#' @family ops
+#' @tether keras.ops.cbrt
+op_cbrt <-
+function (x)
+ops$cbrt(x)
+
+
+#' Convert angles from degrees to radians.
+#'
+#' @description
+#' The conversion is defined as:
+#' `rad = deg * (pi / 180)`.
+#'
+#' # Examples
+#' ```{r}
+#' op_deg2rad(c(0, 90, 180))
+#' ```
+#'
+#' @returns
+#' A tensor containing angles converted to radians.
+#'
+#' @param x
+#' Input tensor of angles in degrees.
+#'
+#' @export
+#' @family numpy ops
+#' @family ops
+#' @tether keras.ops.deg2rad
+op_deg2rad <-
+function (x)
+ops$deg2rad(x)
+
+
 #' Bartlett window function.
 #'
 #' @description
diff --git a/tools/make-r-wrapper.R b/tools/make-r-wrapper.R
index 076cd0028..a864ee453 100755
--- a/tools/make-r-wrapper.R
+++ b/tools/make-r-wrapper.R
@@ -13,5 +13,5 @@ suppressWarnings(suppressMessages(
 # Example usage:
 # tools/make-r-wrapper.R keras.ops.add
 for (py_expr in commandArgs(TRUE)) {
-  writeLines(mk_export(py_expr)$dump)
+  writeLines(c("", mk_export(py_expr)$dump, ""))
 }

From 5c934630999b5b5df8382f1751643f1c0cb7528b Mon Sep 17 00:00:00 2001
From: Tomasz Kalinowski <kalinowskit@gmail.com>
Date: Fri, 26 Sep 2025 09:45:55 -0400
Subject: [PATCH 51/81] add `op_view_as_real()` and `op_view_as_complex()`

---
 .tether/man/keras.ops.txt |  2 ++
 R/ops.R                   | 56 +++++++++++++++++++++++++++++++++++++++
 2 files changed, 58 insertions(+)

diff --git a/.tether/man/keras.ops.txt b/.tether/man/keras.ops.txt
index 3603f3ef9..ce1e5f2f3 100644
--- a/.tether/man/keras.ops.txt
+++ b/.tether/man/keras.ops.txt
@@ -745,4 +745,6 @@ while_loop(
 )
 zeros(shape, dtype=None)
 zeros_like(x, dtype=None)
+view_as_complex(x)
+view_as_real(x)
 
diff --git a/R/ops.R b/R/ops.R
index d6745e32a..657b08f40 100644
--- a/R/ops.R
+++ b/R/ops.R
@@ -4052,6 +4052,62 @@ function (x)
 ops$angle(x)
 
 
+#' Convert a real tensor with two channels into a complex tensor.
+#'
+#' @description
+#' Expects a real-valued tensor whose last dimension has size `2`, holding the
+#' real and imaginary parts. Returns the corresponding complex tensor with the
+#' last dimension removed.
+#'
+#' # Examples
+#' ```{r}
+#' x <- op_array(matrix(c(1, 2, 3, 4), nrow = 2, byrow = TRUE))
+#' op_view_as_complex(x)
+#' ```
+#'
+#' @returns
+#' A complex tensor with shape `op_shape(x)[-length(op_shape(x))]`.
+#'
+#' @param x
+#' Real-valued tensor whose trailing dimension encodes the complex components.
+#'
+#' @export
+#' @family numpy ops
+#' @family ops
+#' @tether keras.ops.view_as_complex
+op_view_as_complex <-
+function (x)
+ops$view_as_complex(x)
+
+
+#' Convert a complex tensor into a stacked real representation.
+#'
+#' @description
+#' Produces a real-valued tensor where the last dimension gathers the real and
+#' imaginary parts of the complex input.
+#'
+#' # Examples
+#' ```{r}
+#' x <- op_array(matrix(c(1, 2, 3, 4), nrow = 2, byrow = TRUE))
+#' z <- op_view_as_complex(x)
+#' op_view_as_real(z)
+#' ```
+#'
+#' @returns
+#' A real tensor with shape `c(op_shape(x), 2)` containing real and imaginary parts.
+#'
+#' @param x
+#' Complex-valued tensor to be converted.
+#'
+#' @export
+#' @family numpy ops
+#' @family ops
+#' @tether keras.ops.view_as_real
+op_view_as_real <-
+function (x)
+ops$view_as_real(x)
+
+
 #' Returns a copy of `x`.
 #'
 #' @returns

From d233cae1faccd29e31f252aec831a6fa99939b3b Mon Sep 17 00:00:00 2001
From: Tomasz Kalinowski <kalinowskit@gmail.com>
Date: Fri, 26 Sep 2025 09:59:43 -0400
Subject: [PATCH 52/81] add `op_layer_normalization()`

---
 .tether/man/keras.ops.nn.txt |  8 ++++++
 .tether/man/keras.ops.txt    |  8 ++++++
 R/ops-nn.R                   | 49 ++++++++++++++++++++++++++++++++++++
 3 files changed, 65 insertions(+)

diff --git a/.tether/man/keras.ops.nn.txt b/.tether/man/keras.ops.nn.txt
index 59f5d89ee..04b363aa0 100644
--- a/.tether/man/keras.ops.nn.txt
+++ b/.tether/man/keras.ops.nn.txt
@@ -85,6 +85,14 @@ hard_sigmoid(x)
 hard_silu(x)
 hard_swish(x)
 hard_tanh(x)
+layer_normalization(
+  x,
+  gamma=None,
+  beta=None,
+  axis=-1,
+  epsilon=None,
+  **kwargs
+)
 leaky_relu(x, negative_slope=0.2)
 log_sigmoid(x)
 log_softmax(x, axis=-1)
diff --git a/.tether/man/keras.ops.txt b/.tether/man/keras.ops.txt
index ce1e5f2f3..29045d079 100644
--- a/.tether/man/keras.ops.txt
+++ b/.tether/man/keras.ops.txt
@@ -332,6 +332,14 @@ istft(
   center=True
 )
 kaiser(x, beta)
+layer_normalization(
+  x,
+  gamma=None,
+  beta=None,
+  axis=-1,
+  epsilon=None,
+  **kwargs
+)
 leaky_relu(x, negative_slope=0.2)
 left_shift(x, y)
 less(x1, x2)
diff --git a/R/ops-nn.R b/R/ops-nn.R
index 4bf672632..3b9542f2f 100644
--- a/R/ops-nn.R
+++ b/R/ops-nn.R
@@ -111,6 +111,55 @@ function (x, axis = -1L, order = 2L, epsilon = NULL)
 }
 
 
+#' Layer normalization (Ba et al., 2016).
+#'
+#' @description
+#' Normalizes activations in `x` for each example independently by centering to
+#' mean 0 and scaling to unit variance along the specified `axis`.
+#'
+#' # Examples
+#' ```{r}
+#' x <- op_arange(5, dtype = "float32")
+#' op_layer_normalization(x)
+#' ```
+#'
+#' @returns
+#' Tensor with the same shape as `x` containing the normalized values.
+#'
+#' @param x
+#' Input tensor.
+#'
+#' @param gamma
+#' Optional scaling factor applied to the normalized output.
+#'
+#' @param beta
+#' Optional offset added to the normalized output.
+#'
+#' @param axis
+#' Axis or axes along which to compute statistics. Defaults to `-1`.
+#'
+#' @param epsilon
+#' Small constant added to the variance for numerical stability.
+#'
+#' @param ...
+#' For forward/backward compatibility.
+#'
+#' @export
+#' @family nn ops
+#' @family ops
+#' @tether keras.ops.layer_normalization
+op_layer_normalization <-
+function (x, gamma = NULL, beta = NULL, axis = -1L, epsilon = NULL, ...)
+{
+    args <- capture_args(list(
+      axis = as_axis,
+      gamma = as_array,
+      beta = as_array
+    ))
+    do.call(ops$layer_normalization, args)
+}
+
+
 #' Peak Signal-to-Noise Ratio (PSNR) function.
 #'
 #' @description

From 811b4265e196e5349112ea6907e42fe5abdcb68f Mon Sep 17 00:00:00 2001
From: Tomasz Kalinowski <kalinowskit@gmail.com>
Date: Fri, 26 Sep 2025 10:02:20 -0400
Subject: [PATCH 53/81] internal tether changes: modules moved '.api.' -> '.'

---
 .tether/man/keras.legacy.txt        |  2 +-
 .tether/man/keras.optimizers.txt    |  4 +--
 .tether/man/keras.preprocessing.txt |  4 +--
 .tether/man/keras.txt               | 55 +++++++++++++++--------------
 .tether/man/keras.utils.txt         |  4 +--
 5 files changed, 35 insertions(+), 34 deletions(-)

diff --git a/.tether/man/keras.legacy.txt b/.tether/man/keras.legacy.txt
index 13f97b309..43e1bbaf2 100644
--- a/.tether/man/keras.legacy.txt
+++ b/.tether/man/keras.legacy.txt
@@ -1,2 +1,2 @@
-saving: Module(keras.api.legacy.saving)
+saving: Module(keras.legacy.saving)
 
diff --git a/.tether/man/keras.optimizers.txt b/.tether/man/keras.optimizers.txt
index 047b7c298..6b8c30ea9 100644
--- a/.tether/man/keras.optimizers.txt
+++ b/.tether/man/keras.optimizers.txt
@@ -141,7 +141,7 @@ Lamb(
   name='lamb',
   **kwargs
 )
-legacy: Module(keras.api.optimizers.legacy)
+legacy: Module(keras.optimizers.legacy)
 Lion(
   learning_rate=0.001,
   beta_1=0.9,
@@ -200,7 +200,7 @@ RMSprop(
   name='rmsprop',
   **kwargs
 )
-schedules: Module(keras.api.optimizers.schedules)
+schedules: Module(keras.optimizers.schedules)
 serialize(optimizer)
 SGD(
   learning_rate=0.01,
diff --git a/.tether/man/keras.preprocessing.txt b/.tether/man/keras.preprocessing.txt
index c7da2c534..4ba34405b 100644
--- a/.tether/man/keras.preprocessing.txt
+++ b/.tether/man/keras.preprocessing.txt
@@ -1,4 +1,4 @@
-image: Module(keras.api.preprocessing.image)
+image: Module(keras.preprocessing.image)
 image_dataset_from_directory(
   directory,
   labels='inferred',
@@ -18,7 +18,7 @@ image_dataset_from_directory(
   data_format=None,
   verbose=True
 )
-sequence: Module(keras.api.preprocessing.sequence)
+sequence: Module(keras.preprocessing.sequence)
 text_dataset_from_directory(
   directory,
   labels='inferred',
diff --git a/.tether/man/keras.txt b/.tether/man/keras.txt
index 19675df83..d38f8bbbe 100644
--- a/.tether/man/keras.txt
+++ b/.tether/man/keras.txt
@@ -1,15 +1,15 @@
-activations: Module(keras.api.activations)
-applications: Module(keras.api.applications)
-backend: Module(keras.api.backend)
-callbacks: Module(keras.api.callbacks)
-config: Module(keras.api.config)
-constraints: Module(keras.api.constraints)
-datasets: Module(keras.api.datasets)
+activations: Module(keras.activations)
+applications: Module(keras.applications)
+backend: Module(keras.backend)
+callbacks: Module(keras.callbacks)
+config: Module(keras.config)
+constraints: Module(keras.constraints)
+datasets: Module(keras.datasets)
 device(device_name)
-distribution: Module(keras.api.distribution)
-dtype_policies: Module(keras.api.dtype_policies)
+distribution: Module(keras.distribution)
+dtype_policies: Module(keras.dtype_policies)
 DTypePolicy(name=None)
-export: Module(keras.api.export)
+export: Module(keras.export)
 FloatDTypePolicy(name=None)
 Function(
   inputs,
@@ -17,7 +17,7 @@ Function(
   name=None
 )
 Initializer()
-initializers: Module(keras.api.initializers)
+initializers: Module(keras.initializers)
 Input(
   shape=None,
   batch_size=None,
@@ -49,46 +49,47 @@ KerasTensor(
   name=None
 )
 Layer(*args, **kwargs)
-layers: Module(keras.api.layers)
-legacy: Module(keras.api.legacy)
+layers: Module(keras.layers)
+legacy: Module(keras.legacy)
 Loss(
   name=None,
   reduction='sum_over_batch_size',
   dtype=None
 )
-losses: Module(keras.api.losses)
+losses: Module(keras.losses)
 Metric(dtype=None, name=None)
-metrics: Module(keras.api.metrics)
-mixed_precision: Module(keras.api.mixed_precision)
+metrics: Module(keras.metrics)
+mixed_precision: Module(keras.mixed_precision)
 Model(*args, **kwargs)
-models: Module(keras.api.models)
+models: Module(keras.models)
 name_scope(name, **kwargs)
 Operation(*args, **kwargs)
 ops: Module(keras.ops)
 Optimizer(*args, **kwargs)
-optimizers: Module(keras.api.optimizers)
-preprocessing: Module(keras.api.preprocessing)
+optimizers: Module(keras.optimizers)
+preprocessing: Module(keras.preprocessing)
 Quantizer(output_dtype='int8')
-quantizers: Module(keras.api.quantizers)
-random: Module(keras.api.random)
+quantizers: Module(keras.quantizers)
+random: Module(keras.random)
 Regularizer()
-regularizers: Module(keras.api.regularizers)
+regularizers: Module(keras.regularizers)
 remat(f)
 RematScope(
   mode='full',
   output_size_threshold=1024,
   layer_names=None
 )
-saving: Module(keras.api.saving)
+saving: Module(keras.saving)
 Sequential(*args, **kwargs)
+src: Module(keras.src)
 StatelessScope(
   state_mapping=None,
   collect_losses=False,
   initialize_variables=True
 )
 SymbolicScope()
-tree: Module(keras.api.tree)
-utils: Module(keras.api.utils)
+tree: Module(keras.tree)
+utils: Module(keras.utils)
 Variable(
   initializer,
   shape=None,
@@ -99,6 +100,6 @@ Variable(
   name=None
 )
 version()
-visualization: Module(keras.api.visualization)
-wrappers: Module(keras.api.wrappers)
+visualization: Module(keras.visualization)
+wrappers: Module(keras.wrappers)
 
diff --git a/.tether/man/keras.utils.txt b/.tether/man/keras.utils.txt
index c54a975cb..d164c61d6 100644
--- a/.tether/man/keras.utils.txt
+++ b/.tether/man/keras.utils.txt
@@ -20,7 +20,7 @@ audio_dataset_from_directory(
   follow_links=False,
   verbose=True
 )
-bounding_boxes: Module(keras.api.utils.bounding_boxes)
+bounding_boxes: Module(keras.utils.bounding_boxes)
 clear_session(free_memory=True)
 Config(**kwargs)
 custom_object_scope(custom_objects)
@@ -89,7 +89,7 @@ img_to_array(
 )
 is_interactive_logging_enabled()
 is_keras_tensor(x)
-legacy: Module(keras.api.utils.legacy)
+legacy: Module(keras.utils.legacy)
 load_img(
   path,
   color_mode='rgb',

From fc81ba403cb730d2d9d4d3de407947e3207b0dd0 Mon Sep 17 00:00:00 2001
From: Tomasz Kalinowski <kalinowskit@gmail.com>
Date: Fri, 26 Sep 2025 11:05:19 -0400
Subject: [PATCH 54/81] add `loss_categorical_generalized_cross_entropy()`

---
 .tether/man/keras.losses.txt | 11 +++++
 R/losses.R                   | 79 ++++++++++++++++++++++++++++++++++++
 2 files changed, 90 insertions(+)

diff --git a/.tether/man/keras.losses.txt b/.tether/man/keras.losses.txt
index b3c44ae29..c1733bcb8 100644
--- a/.tether/man/keras.losses.txt
+++ b/.tether/man/keras.losses.txt
@@ -50,6 +50,11 @@ categorical_focal_crossentropy(
   label_smoothing=0.0,
   axis=-1
 )
+categorical_generalized_cross_entropy(
+  y_true,
+  y_pred,
+  q
+)
 categorical_hinge(y_true, y_pred)
 CategoricalCrossentropy(
   from_logits=False,
@@ -69,6 +74,12 @@ CategoricalFocalCrossentropy(
   name='categorical_focal_crossentropy',
   dtype=None
 )
+CategoricalGeneralizedCrossEntropy(
+  q=0.5,
+  reduction='sum_over_batch_size',
+  name='categorical_generalized_cross_entropy',
+  dtype=None
+)
 CategoricalHinge(
   reduction='sum_over_batch_size',
   name='categorical_hinge',
diff --git a/R/losses.R b/R/losses.R
index e8570e984..136640812 100644
--- a/R/losses.R
+++ b/R/losses.R
@@ -647,6 +647,85 @@ function (y_true, y_pred, alpha = 0.25, gamma = 2,
 }
 
 
+
+#' Computes the generalized cross entropy loss.
+#'
+#' @description
+#' The generalized cross entropy (GCE) loss offers robustness to noisy labels by
+#' interpolating between categorical cross entropy (`q -> 0`) and mean absolute
+#' error (`q -> 1`). For a true-class probability `p` and noise parameter `q`,
+#' the loss is `loss = (1 - p^q) / q`.
+#'
+#' # References
+#' - Zhang & Sabuncu (2018), "Generalized Cross Entropy Loss for Training Deep
+#'   Neural Networks with Noisy Labels"
+#'
+#' # Examples
+#' ```{r}
+#' y_true <- c(0L, 1L, 0L, 1L)
+#' y_pred <- rbind(
+#'   c(0.7, 0.3),
+#'   c(0.2, 0.8),
+#'   c(0.6, 0.4),
+#'   c(0.4, 0.6)
+#' )
+#' gce <- loss_categorical_generalized_cross_entropy(q = 0.7)
+#' gce(y_true, y_pred)
+#' ```
+#'
+#' @returns
+#' Generalized cross entropy loss value(s).
+#'
+#' @param q
+#' Float in `(0, 1)`. Controls the transition between cross entropy and mean
+#' absolute error. Defaults to `0.5`.
+#'
+#' - As `q` approaches `0`: behaves like categorical cross entropy.
+#' - As `q` approaches `1`: behaves like mean absolute error.
+#'
+#' @param reduction
+#' Type of reduction to apply to the loss. In almost all cases
+#' this should be `"sum_over_batch_size"`. Supported options are
+#' `"sum"`, `"sum_over_batch_size"`, `"mean"`,
+#' `"mean_with_sample_weight"` or `NULL`. `"sum"` sums the loss,
+#' `"sum_over_batch_size"` and `"mean"` sum the loss and divide by the
+#' sample size, and `"mean_with_sample_weight"` sums the loss and
+#' divides by the sum of the sample weights. `"none"` and `NULL`
+#' perform no aggregation. Defaults to `"sum_over_batch_size"`.
+#'
+#' @param name
+#' Optional name for the loss instance.
+#'
+#' @param dtype
+#' Dtype used for loss computations. Defaults to `config_floatx()` (the global
+#' float type).
+#'
+#' @param y_true
+#' Integer class indices with shape `(batch_size)` or `(batch_size, 1)`.
+#'
+#' @param y_pred
+#' Predicted class probabilities with shape `(batch_size, num_classes)`.
+#'
+#' @param ...
+#' For forward/backward compatibility.
+#'
+#' @export
+#' @family losses
+#' @tether keras.losses.CategoricalGeneralizedCrossEntropy
+loss_categorical_generalized_cross_entropy <-
+function (y_true, y_pred, q = 0.5, ..., reduction = "sum_over_batch_size",
+          name = "categorical_generalized_cross_entropy", dtype = NULL)
+{
+    args <- capture_args(list(
+      y_true = as_py_array,
+      y_pred = as_py_array
+    ))
+    callable <- if (missing(y_true) && missing(y_pred))
+        keras$losses$CategoricalGeneralizedCrossEntropy
+    else keras$losses$categorical_generalized_cross_entropy
+    do.call(callable, args)
+}
+
 #' Computes the categorical hinge loss between `y_true` & `y_pred`.
 #'
 #' @description

From f8cb76934f02d66c5b2e01ae51818374a624af53 Mon Sep 17 00:00:00 2001
From: Tomasz Kalinowski <kalinowskit@gmail.com>
Date: Fri, 26 Sep 2025 11:19:07 -0400
Subject: [PATCH 55/81] redocument

---
 NAMESPACE                                     |  18 +
 man/Loss.Rd                                   |   1 +
 man/activation_celu.Rd                        |   1 +
 man/activation_elu.Rd                         |   1 +
 man/activation_exponential.Rd                 |   1 +
 man/activation_gelu.Rd                        |   1 +
 man/activation_glu.Rd                         |   1 +
 man/activation_hard_shrink.Rd                 |   1 +
 man/activation_hard_sigmoid.Rd                |   1 +
 man/activation_hard_tanh.Rd                   |   1 +
 man/activation_leaky_relu.Rd                  |   1 +
 man/activation_linear.Rd                      |   1 +
 man/activation_log_sigmoid.Rd                 |   1 +
 man/activation_log_softmax.Rd                 |   1 +
 man/activation_mish.Rd                        |   1 +
 man/activation_relu.Rd                        |   1 +
 man/activation_relu6.Rd                       |   1 +
 man/activation_selu.Rd                        |   1 +
 man/activation_sigmoid.Rd                     |   1 +
 man/activation_silu.Rd                        |   1 +
 man/activation_soft_shrink.Rd                 |   1 +
 man/activation_softmax.Rd                     |   1 +
 man/activation_softplus.Rd                    |   1 +
 man/activation_softsign.Rd                    |   1 +
 man/activation_sparse_plus.Rd                 |   1 +
 man/activation_sparse_sigmoid.Rd              |  59 +++
 man/activation_sparsemax.Rd                   |   1 +
 man/activation_squareplus.Rd                  |   1 +
 man/activation_tanh.Rd                        |   1 +
 man/activation_tanh_shrink.Rd                 |   1 +
 man/activation_threshold.Rd                   |   1 +
 man/layer_tfsm.Rd                             |   4 +-
 man/loss_binary_crossentropy.Rd               |   1 +
 man/loss_binary_focal_crossentropy.Rd         |   1 +
 man/loss_categorical_crossentropy.Rd          |   1 +
 man/loss_categorical_focal_crossentropy.Rd    |   1 +
 ...s_categorical_generalized_cross_entropy.Rd | 119 +++++
 man/loss_categorical_hinge.Rd                 |   1 +
 man/loss_circle.Rd                            |   1 +
 man/loss_cosine_similarity.Rd                 |   1 +
 man/loss_ctc.Rd                               |   1 +
 man/loss_dice.Rd                              |   1 +
 man/loss_hinge.Rd                             |   1 +
 man/loss_huber.Rd                             |   1 +
 man/loss_kl_divergence.Rd                     |   1 +
 man/loss_log_cosh.Rd                          |   1 +
 man/loss_mean_absolute_error.Rd               |   1 +
 man/loss_mean_absolute_percentage_error.Rd    |   1 +
 man/loss_mean_squared_error.Rd                |   1 +
 man/loss_mean_squared_logarithmic_error.Rd    |   1 +
 man/loss_poisson.Rd                           |   1 +
 man/loss_sparse_categorical_crossentropy.Rd   |   1 +
 man/loss_squared_hinge.Rd                     |   1 +
 man/loss_tversky.Rd                           |   1 +
 man/metric_binary_crossentropy.Rd             |   1 +
 man/metric_binary_focal_crossentropy.Rd       |   1 +
 man/metric_categorical_crossentropy.Rd        |   1 +
 man/metric_categorical_focal_crossentropy.Rd  |   1 +
 man/metric_categorical_hinge.Rd               |   1 +
 man/metric_hinge.Rd                           |   1 +
 man/metric_huber.Rd                           |   1 +
 man/metric_kl_divergence.Rd                   |   1 +
 man/metric_log_cosh.Rd                        |   1 +
 man/metric_mean_absolute_error.Rd             |   1 +
 man/metric_mean_absolute_percentage_error.Rd  |   1 +
 man/metric_mean_squared_error.Rd              |   1 +
 man/metric_mean_squared_logarithmic_error.Rd  |   1 +
 man/metric_poisson.Rd                         |   1 +
 man/metric_sparse_categorical_crossentropy.Rd |   1 +
 man/metric_squared_hinge.Rd                   |   1 +
 man/op_abs.Rd                                 |  26 +
 man/op_add.Rd                                 |  26 +
 man/op_all.Rd                                 |  26 +
 man/op_angle.Rd                               | 484 +++++++++++++++++
 man/op_any.Rd                                 |  26 +
 man/op_append.Rd                              |  26 +
 man/op_arange.Rd                              |  26 +
 man/op_arccos.Rd                              |  26 +
 man/op_arccosh.Rd                             |  26 +
 man/op_arcsin.Rd                              |  26 +
 man/op_arcsinh.Rd                             |  26 +
 man/op_arctan.Rd                              |  26 +
 man/op_arctan2.Rd                             |  26 +
 man/op_arctanh.Rd                             |  26 +
 man/op_argmax.Rd                              |  26 +
 man/op_argmin.Rd                              |  26 +
 man/op_argpartition.Rd                        |  26 +
 man/op_argsort.Rd                             |  26 +
 man/op_array.Rd                               |  26 +
 man/op_associative_scan.Rd                    |  14 +
 man/op_average.Rd                             |  26 +
 man/op_average_pool.Rd                        |  16 +
 man/op_bartlett.Rd                            | 481 +++++++++++++++++
 man/op_batch_normalization.Rd                 |  16 +
 man/op_binary_crossentropy.Rd                 |  16 +
 man/op_bincount.Rd                            |  26 +
 man/op_bitwise_and.Rd                         |  26 +
 man/op_bitwise_invert.Rd                      |  26 +
 man/op_bitwise_left_shift.Rd                  |  26 +
 man/op_bitwise_not.Rd                         |  26 +
 man/op_bitwise_or.Rd                          |  26 +
 man/op_bitwise_right_shift.Rd                 |  26 +
 man/op_bitwise_xor.Rd                         |  26 +
 man/op_blackman.Rd                            | 487 ++++++++++++++++++
 man/op_broadcast_to.Rd                        |  26 +
 man/op_cast.Rd                                |  14 +
 man/op_categorical_crossentropy.Rd            |  16 +
 man/op_cbrt.Rd                                | 480 +++++++++++++++++
 man/op_ceil.Rd                                |  26 +
 man/op_celu.Rd                                |  16 +
 man/op_cholesky.Rd                            |  14 +
 man/op_clip.Rd                                |  26 +
 man/op_concatenate.Rd                         |  26 +
 man/op_cond.Rd                                |  14 +
 man/op_conj.Rd                                |  26 +
 man/op_conv.Rd                                |  16 +
 man/op_conv_transpose.Rd                      |  16 +
 man/op_convert_to_numpy.Rd                    |  14 +
 man/op_convert_to_tensor.Rd                   |  14 +
 man/op_copy.Rd                                |  26 +
 man/op_corrcoef.Rd                            | 484 +++++++++++++++++
 man/op_correlate.Rd                           |  26 +
 man/op_cos.Rd                                 |  26 +
 man/op_cosh.Rd                                |  26 +
 man/op_count_nonzero.Rd                       |  26 +
 man/op_cross.Rd                               |  26 +
 man/op_ctc_decode.Rd                          |  26 +
 man/op_ctc_loss.Rd                            |  16 +
 man/op_cumprod.Rd                             |  26 +
 man/op_cumsum.Rd                              |  26 +
 man/op_custom_gradient.Rd                     |  14 +
 man/op_deg2rad.Rd                             | 482 +++++++++++++++++
 man/op_depthwise_conv.Rd                      |  16 +
 man/op_det.Rd                                 |  14 +
 man/op_diag.Rd                                |  26 +
 man/op_diagflat.Rd                            |  26 +
 man/op_diagonal.Rd                            |  26 +
 man/op_diff.Rd                                |  26 +
 man/op_digitize.Rd                            |  26 +
 man/op_divide.Rd                              |  26 +
 man/op_divide_no_nan.Rd                       |  26 +
 man/op_dot.Rd                                 |  26 +
 man/op_dot_product_attention.Rd               |  16 +
 man/op_dtype.Rd                               |  14 +
 man/op_eig.Rd                                 |  14 +
 man/op_eigh.Rd                                |  14 +
 man/op_einsum.Rd                              |  26 +
 man/op_elu.Rd                                 |  16 +
 man/op_empty.Rd                               |  26 +
 man/op_equal.Rd                               |  26 +
 man/op_erf.Rd                                 |  14 +
 man/op_erfinv.Rd                              |  14 +
 man/op_exp.Rd                                 |  26 +
 man/op_exp2.Rd                                |  26 +
 man/op_expand_dims.Rd                         |  26 +
 man/op_expm1.Rd                               |  26 +
 man/op_extract_sequences.Rd                   |  14 +
 man/op_eye.Rd                                 |  26 +
 man/op_fft.Rd                                 |  14 +
 man/op_fft2.Rd                                |  14 +
 man/op_flip.Rd                                |  26 +
 man/op_floor.Rd                               |  26 +
 man/op_floor_divide.Rd                        |  26 +
 man/op_fori_loop.Rd                           |  14 +
 man/op_full.Rd                                |  26 +
 man/op_full_like.Rd                           |  26 +
 man/op_gelu.Rd                                |  16 +
 man/op_get_item.Rd                            |  26 +
 man/op_glu.Rd                                 |  16 +
 man/op_greater.Rd                             |  26 +
 man/op_greater_equal.Rd                       |  26 +
 man/op_hamming.Rd                             | 482 +++++++++++++++++
 man/op_hanning.Rd                             | 482 +++++++++++++++++
 man/op_hard_shrink.Rd                         |  16 +
 man/op_hard_sigmoid.Rd                        |  16 +
 man/op_hard_silu.Rd                           |  16 +
 man/op_hard_tanh.Rd                           |  16 +
 man/op_heaviside.Rd                           | 487 ++++++++++++++++++
 man/op_histogram.Rd                           |  26 +
 man/op_hstack.Rd                              |  26 +
 man/op_identity.Rd                            |  26 +
 man/op_ifft2.Rd                               |  14 +
 man/op_imag.Rd                                |  26 +
 man/op_image_affine_transform.Rd              |  14 +
 man/op_image_crop.Rd                          |  14 +
 man/op_image_extract_patches.Rd               |  14 +
 man/op_image_gaussian_blur.Rd                 |  14 +
 man/op_image_hsv_to_rgb.Rd                    |  14 +
 man/op_image_map_coordinates.Rd               |  14 +
 man/op_image_pad.Rd                           |  14 +
 man/op_image_perspective_transform.Rd         |  14 +
 man/op_image_resize.Rd                        |  14 +
 man/op_image_rgb_to_grayscale.Rd              |  14 +
 man/op_image_rgb_to_hsv.Rd                    |  14 +
 man/op_in_top_k.Rd                            |  14 +
 man/op_inner.Rd                               |  26 +
 man/op_inv.Rd                                 |  14 +
 man/op_irfft.Rd                               |  14 +
 man/op_is_tensor.Rd                           |  14 +
 man/op_isclose.Rd                             |  26 +
 man/op_isfinite.Rd                            |  26 +
 man/op_isinf.Rd                               |  26 +
 man/op_isnan.Rd                               |  26 +
 man/op_istft.Rd                               |  14 +
 man/op_kaiser.Rd                              | 485 +++++++++++++++++
 man/op_layer_normalization.Rd                 | 377 ++++++++++++++
 man/op_leaky_relu.Rd                          |  16 +
 man/op_left_shift.Rd                          |  26 +
 man/op_less.Rd                                |  26 +
 man/op_less_equal.Rd                          |  26 +
 man/op_linspace.Rd                            |  26 +
 man/op_log.Rd                                 |  26 +
 man/op_log10.Rd                               |  26 +
 man/op_log1p.Rd                               |  26 +
 man/op_log2.Rd                                |  26 +
 man/op_log_sigmoid.Rd                         |  16 +
 man/op_log_softmax.Rd                         |  16 +
 man/op_logaddexp.Rd                           |  26 +
 man/op_logdet.Rd                              |  26 +
 man/op_logical_and.Rd                         |  26 +
 man/op_logical_not.Rd                         |  26 +
 man/op_logical_or.Rd                          |  26 +
 man/op_logical_xor.Rd                         |  26 +
 man/op_logspace.Rd                            |  26 +
 man/op_logsumexp.Rd                           |  14 +
 man/op_lstsq.Rd                               |  26 +
 man/op_lu_factor.Rd                           |  14 +
 man/op_map.Rd                                 |  14 +
 man/op_matmul.Rd                              |  26 +
 man/op_max.Rd                                 |  26 +
 man/op_max_pool.Rd                            |  16 +
 man/op_maximum.Rd                             |  26 +
 man/op_mean.Rd                                |  26 +
 man/op_median.Rd                              |  26 +
 man/op_meshgrid.Rd                            |  26 +
 man/op_min.Rd                                 |  26 +
 man/op_minimum.Rd                             |  26 +
 man/op_mod.Rd                                 |  26 +
 man/op_moments.Rd                             |  16 +
 man/op_moveaxis.Rd                            |  26 +
 man/op_multi_hot.Rd                           |  16 +
 man/op_multiply.Rd                            |  26 +
 man/op_nan_to_num.Rd                          |  26 +
 man/op_ndim.Rd                                |  26 +
 man/op_negative.Rd                            |  26 +
 man/op_nonzero.Rd                             |  26 +
 man/op_norm.Rd                                |  14 +
 man/op_normalize.Rd                           |  16 +
 man/op_not_equal.Rd                           |  26 +
 man/op_one_hot.Rd                             |  16 +
 man/op_ones.Rd                                |  26 +
 man/op_ones_like.Rd                           |  26 +
 man/op_outer.Rd                               |  26 +
 man/op_pad.Rd                                 |  26 +
 man/op_polar.Rd                               |  16 +
 man/op_power.Rd                               |  26 +
 man/op_prod.Rd                                |  26 +
 man/op_psnr.Rd                                |  16 +
 man/op_qr.Rd                                  |  14 +
 man/op_quantile.Rd                            |  26 +
 man/op_ravel.Rd                               |  26 +
 man/op_real.Rd                                |  26 +
 man/op_rearrange.Rd                           |  14 +
 man/op_reciprocal.Rd                          |  26 +
 man/op_relu.Rd                                |  16 +
 man/op_relu6.Rd                               |  16 +
 man/op_repeat.Rd                              |  26 +
 man/op_reshape.Rd                             |  26 +
 man/op_rfft.Rd                                |  14 +
 man/op_right_shift.Rd                         |  26 +
 man/op_rms_normalization.Rd                   |  16 +
 man/op_roll.Rd                                |  26 +
 man/op_rot90.Rd                               |  26 +
 man/op_round.Rd                               |  26 +
 man/op_rsqrt.Rd                               |  14 +
 man/op_saturate_cast.Rd                       |  26 +
 man/op_scan.Rd                                |  14 +
 man/op_scatter.Rd                             |  14 +
 man/op_scatter_update.Rd                      |  14 +
 man/op_searchsorted.Rd                        |  14 +
 man/op_segment_max.Rd                         |  14 +
 man/op_segment_sum.Rd                         |  14 +
 man/op_select.Rd                              |  26 +
 man/op_selu.Rd                                |  16 +
 man/op_separable_conv.Rd                      |  16 +
 man/op_shape.Rd                               |  14 +
 man/op_sigmoid.Rd                             |  16 +
 man/op_sign.Rd                                |  26 +
 man/op_signbit.Rd                             |  26 +
 man/op_silu.Rd                                |  16 +
 man/op_sin.Rd                                 |  26 +
 man/op_sinh.Rd                                |  26 +
 man/op_size.Rd                                |  26 +
 man/op_slice.Rd                               |  14 +
 man/op_slice_update.Rd                        |  14 +
 man/op_slogdet.Rd                             |  14 +
 man/op_soft_shrink.Rd                         |  16 +
 man/op_softmax.Rd                             |  16 +
 man/op_softplus.Rd                            |  16 +
 man/op_softsign.Rd                            |  16 +
 man/op_solve.Rd                               |  14 +
 man/op_solve_triangular.Rd                    |  14 +
 man/op_sort.Rd                                |  26 +
 man/op_sparse_categorical_crossentropy.Rd     |  16 +
 man/op_sparse_plus.Rd                         |  16 +
 man/op_sparse_sigmoid.Rd                      | 363 +++++++++++++
 man/op_sparsemax.Rd                           |  16 +
 man/op_split.Rd                               |  26 +
 man/op_sqrt.Rd                                |  26 +
 man/op_square.Rd                              |  26 +
 man/op_squareplus.Rd                          |  16 +
 man/op_squeeze.Rd                             |  26 +
 man/op_stack.Rd                               |  26 +
 man/op_std.Rd                                 |  26 +
 man/op_stft.Rd                                |  14 +
 man/op_stop_gradient.Rd                       |  14 +
 man/op_subset.Rd                              |  42 ++
 man/op_subtract.Rd                            |  26 +
 man/op_sum.Rd                                 |  26 +
 man/op_svd.Rd                                 |  14 +
 man/op_swapaxes.Rd                            |  26 +
 man/op_switch.Rd                              |  14 +
 man/op_take.Rd                                |  26 +
 man/op_take_along_axis.Rd                     |  26 +
 man/op_tan.Rd                                 |  26 +
 man/op_tanh.Rd                                |  26 +
 man/op_tanh_shrink.Rd                         |  16 +
 man/op_tensordot.Rd                           |  26 +
 man/op_threshold.Rd                           |  16 +
 man/op_tile.Rd                                |  26 +
 man/op_top_k.Rd                               |  14 +
 man/op_trace.Rd                               |  26 +
 man/op_transpose.Rd                           |  26 +
 man/op_tri.Rd                                 |  26 +
 man/op_tril.Rd                                |  26 +
 man/op_triu.Rd                                |  26 +
 man/op_trunc.Rd                               |  26 +
 man/op_unravel_index.Rd                       |  16 +
 man/op_unstack.Rd                             |  14 +
 man/op_var.Rd                                 |  26 +
 man/op_vdot.Rd                                |  26 +
 man/op_vectorize.Rd                           |  26 +
 man/op_vectorized_map.Rd                      |  14 +
 man/op_view_as_complex.Rd                     | 484 +++++++++++++++++
 man/op_view_as_real.Rd                        | 486 +++++++++++++++++
 man/op_vstack.Rd                              |  26 +
 man/op_where.Rd                               |  26 +
 man/op_while_loop.Rd                          |  14 +
 man/op_zeros.Rd                               |  26 +
 man/op_zeros_like.Rd                          |  26 +
 350 files changed, 12546 insertions(+), 2 deletions(-)
 create mode 100644 man/activation_sparse_sigmoid.Rd
 create mode 100644 man/loss_categorical_generalized_cross_entropy.Rd
 create mode 100644 man/op_angle.Rd
 create mode 100644 man/op_bartlett.Rd
 create mode 100644 man/op_blackman.Rd
 create mode 100644 man/op_cbrt.Rd
 create mode 100644 man/op_corrcoef.Rd
 create mode 100644 man/op_deg2rad.Rd
 create mode 100644 man/op_hamming.Rd
 create mode 100644 man/op_hanning.Rd
 create mode 100644 man/op_heaviside.Rd
 create mode 100644 man/op_kaiser.Rd
 create mode 100644 man/op_layer_normalization.Rd
 create mode 100644 man/op_sparse_sigmoid.Rd
 create mode 100644 man/op_view_as_complex.Rd
 create mode 100644 man/op_view_as_real.Rd

diff --git a/NAMESPACE b/NAMESPACE
index 9d4f3ec7a..e0cf8efc9 100644
--- a/NAMESPACE
+++ b/NAMESPACE
@@ -8,6 +8,8 @@ S3method("==",keras.src.backend.common.keras_tensor.KerasTensor)
 S3method("==",keras_shape)
 S3method("[",keras_shape)
 S3method("[[",python_builtin_super_getter)
+S3method(Arg,keras.src.backend.Tensor)
+S3method(Arg,keras.src.backend.common.keras_tensor.KerasTensor)
 S3method(Summary,keras_shape)
 S3method(as.array,jax.Array)
 S3method(as.array,jaxlib._jax.ArrayImpl)
@@ -102,6 +104,7 @@ export(activation_softmax)
 export(activation_softplus)
 export(activation_softsign)
 export(activation_sparse_plus)
+export(activation_sparse_sigmoid)
 export(activation_sparsemax)
 export(activation_squareplus)
 export(activation_tanh)
@@ -415,6 +418,7 @@ export(loss_binary_crossentropy)
 export(loss_binary_focal_crossentropy)
 export(loss_categorical_crossentropy)
 export(loss_categorical_focal_crossentropy)
+export(loss_categorical_generalized_cross_entropy)
 export(loss_categorical_hinge)
 export(loss_circle)
 export(loss_cosine_similarity)
@@ -493,6 +497,7 @@ export(np_array)
 export(op_abs)
 export(op_add)
 export(op_all)
+export(op_angle)
 export(op_any)
 export(op_append)
 export(op_arange)
@@ -511,6 +516,7 @@ export(op_array)
 export(op_associative_scan)
 export(op_average)
 export(op_average_pool)
+export(op_bartlett)
 export(op_batch_normalization)
 export(op_binary_crossentropy)
 export(op_bincount)
@@ -521,9 +527,11 @@ export(op_bitwise_not)
 export(op_bitwise_or)
 export(op_bitwise_right_shift)
 export(op_bitwise_xor)
+export(op_blackman)
 export(op_broadcast_to)
 export(op_cast)
 export(op_categorical_crossentropy)
+export(op_cbrt)
 export(op_ceil)
 export(op_celu)
 export(op_cholesky)
@@ -537,6 +545,7 @@ export(op_convert_to_array)
 export(op_convert_to_numpy)
 export(op_convert_to_tensor)
 export(op_copy)
+export(op_corrcoef)
 export(op_correlate)
 export(op_cos)
 export(op_cosh)
@@ -547,6 +556,7 @@ export(op_ctc_loss)
 export(op_cumprod)
 export(op_cumsum)
 export(op_custom_gradient)
+export(op_deg2rad)
 export(op_depthwise_conv)
 export(op_det)
 export(op_diag)
@@ -586,11 +596,14 @@ export(op_get_item)
 export(op_glu)
 export(op_greater)
 export(op_greater_equal)
+export(op_hamming)
+export(op_hanning)
 export(op_hard_shrink)
 export(op_hard_sigmoid)
 export(op_hard_silu)
 export(op_hard_swish)
 export(op_hard_tanh)
+export(op_heaviside)
 export(op_histogram)
 export(op_hstack)
 export(op_identity)
@@ -617,6 +630,8 @@ export(op_isfinite)
 export(op_isinf)
 export(op_isnan)
 export(op_istft)
+export(op_kaiser)
+export(op_layer_normalization)
 export(op_leaky_relu)
 export(op_left_shift)
 export(op_less)
@@ -718,6 +733,7 @@ export(op_solve_triangular)
 export(op_sort)
 export(op_sparse_categorical_crossentropy)
 export(op_sparse_plus)
+export(op_sparse_sigmoid)
 export(op_sparsemax)
 export(op_split)
 export(op_sqrt)
@@ -756,6 +772,8 @@ export(op_var)
 export(op_vdot)
 export(op_vectorize)
 export(op_vectorized_map)
+export(op_view_as_complex)
+export(op_view_as_real)
 export(op_vstack)
 export(op_where)
 export(op_while_loop)
diff --git a/man/Loss.Rd b/man/Loss.Rd
index bba72b0c2..30e8a6416 100644
--- a/man/Loss.Rd
+++ b/man/Loss.Rd
@@ -160,6 +160,7 @@ Other losses: \cr
 \code{\link{loss_binary_focal_crossentropy}()} \cr
 \code{\link{loss_categorical_crossentropy}()} \cr
 \code{\link{loss_categorical_focal_crossentropy}()} \cr
+\code{\link{loss_categorical_generalized_cross_entropy}()} \cr
 \code{\link{loss_categorical_hinge}()} \cr
 \code{\link{loss_circle}()} \cr
 \code{\link{loss_cosine_similarity}()} \cr
diff --git a/man/activation_celu.Rd b/man/activation_celu.Rd
index d17f2f026..90efc95ee 100644
--- a/man/activation_celu.Rd
+++ b/man/activation_celu.Rd
@@ -51,6 +51,7 @@ Other activations: \cr
 \code{\link{activation_softplus}()} \cr
 \code{\link{activation_softsign}()} \cr
 \code{\link{activation_sparse_plus}()} \cr
+\code{\link{activation_sparse_sigmoid}()} \cr
 \code{\link{activation_sparsemax}()} \cr
 \code{\link{activation_squareplus}()} \cr
 \code{\link{activation_tanh}()} \cr
diff --git a/man/activation_elu.Rd b/man/activation_elu.Rd
index dcfdbe8f1..69443ffcb 100644
--- a/man/activation_elu.Rd
+++ b/man/activation_elu.Rd
@@ -64,6 +64,7 @@ Other activations: \cr
 \code{\link{activation_softplus}()} \cr
 \code{\link{activation_softsign}()} \cr
 \code{\link{activation_sparse_plus}()} \cr
+\code{\link{activation_sparse_sigmoid}()} \cr
 \code{\link{activation_sparsemax}()} \cr
 \code{\link{activation_squareplus}()} \cr
 \code{\link{activation_tanh}()} \cr
diff --git a/man/activation_exponential.Rd b/man/activation_exponential.Rd
index eb93e0381..b1673e614 100644
--- a/man/activation_exponential.Rd
+++ b/man/activation_exponential.Rd
@@ -43,6 +43,7 @@ Other activations: \cr
 \code{\link{activation_softplus}()} \cr
 \code{\link{activation_softsign}()} \cr
 \code{\link{activation_sparse_plus}()} \cr
+\code{\link{activation_sparse_sigmoid}()} \cr
 \code{\link{activation_sparsemax}()} \cr
 \code{\link{activation_squareplus}()} \cr
 \code{\link{activation_tanh}()} \cr
diff --git a/man/activation_gelu.Rd b/man/activation_gelu.Rd
index 018ef3c98..9cf3474c8 100644
--- a/man/activation_gelu.Rd
+++ b/man/activation_gelu.Rd
@@ -57,6 +57,7 @@ Other activations: \cr
 \code{\link{activation_softplus}()} \cr
 \code{\link{activation_softsign}()} \cr
 \code{\link{activation_sparse_plus}()} \cr
+\code{\link{activation_sparse_sigmoid}()} \cr
 \code{\link{activation_sparsemax}()} \cr
 \code{\link{activation_squareplus}()} \cr
 \code{\link{activation_tanh}()} \cr
diff --git a/man/activation_glu.Rd b/man/activation_glu.Rd
index 2f225e2f3..45714db63 100644
--- a/man/activation_glu.Rd
+++ b/man/activation_glu.Rd
@@ -51,6 +51,7 @@ Other activations: \cr
 \code{\link{activation_softplus}()} \cr
 \code{\link{activation_softsign}()} \cr
 \code{\link{activation_sparse_plus}()} \cr
+\code{\link{activation_sparse_sigmoid}()} \cr
 \code{\link{activation_sparsemax}()} \cr
 \code{\link{activation_squareplus}()} \cr
 \code{\link{activation_tanh}()} \cr
diff --git a/man/activation_hard_shrink.Rd b/man/activation_hard_shrink.Rd
index cdce0a17d..06e8f7d8e 100644
--- a/man/activation_hard_shrink.Rd
+++ b/man/activation_hard_shrink.Rd
@@ -44,6 +44,7 @@ Other activations: \cr
 \code{\link{activation_softplus}()} \cr
 \code{\link{activation_softsign}()} \cr
 \code{\link{activation_sparse_plus}()} \cr
+\code{\link{activation_sparse_sigmoid}()} \cr
 \code{\link{activation_sparsemax}()} \cr
 \code{\link{activation_squareplus}()} \cr
 \code{\link{activation_tanh}()} \cr
diff --git a/man/activation_hard_sigmoid.Rd b/man/activation_hard_sigmoid.Rd
index ad4f0b4eb..25986f9a6 100644
--- a/man/activation_hard_sigmoid.Rd
+++ b/man/activation_hard_sigmoid.Rd
@@ -57,6 +57,7 @@ Other activations: \cr
 \code{\link{activation_softplus}()} \cr
 \code{\link{activation_softsign}()} \cr
 \code{\link{activation_sparse_plus}()} \cr
+\code{\link{activation_sparse_sigmoid}()} \cr
 \code{\link{activation_sparsemax}()} \cr
 \code{\link{activation_squareplus}()} \cr
 \code{\link{activation_tanh}()} \cr
diff --git a/man/activation_hard_tanh.Rd b/man/activation_hard_tanh.Rd
index 49868d4c4..423eef7ae 100644
--- a/man/activation_hard_tanh.Rd
+++ b/man/activation_hard_tanh.Rd
@@ -42,6 +42,7 @@ Other activations: \cr
 \code{\link{activation_softplus}()} \cr
 \code{\link{activation_softsign}()} \cr
 \code{\link{activation_sparse_plus}()} \cr
+\code{\link{activation_sparse_sigmoid}()} \cr
 \code{\link{activation_sparsemax}()} \cr
 \code{\link{activation_squareplus}()} \cr
 \code{\link{activation_tanh}()} \cr
diff --git a/man/activation_leaky_relu.Rd b/man/activation_leaky_relu.Rd
index fa2481fa0..0a7ebfcdb 100644
--- a/man/activation_leaky_relu.Rd
+++ b/man/activation_leaky_relu.Rd
@@ -46,6 +46,7 @@ Other activations: \cr
 \code{\link{activation_softplus}()} \cr
 \code{\link{activation_softsign}()} \cr
 \code{\link{activation_sparse_plus}()} \cr
+\code{\link{activation_sparse_sigmoid}()} \cr
 \code{\link{activation_sparsemax}()} \cr
 \code{\link{activation_squareplus}()} \cr
 \code{\link{activation_tanh}()} \cr
diff --git a/man/activation_linear.Rd b/man/activation_linear.Rd
index 03bf86925..5055e68b1 100644
--- a/man/activation_linear.Rd
+++ b/man/activation_linear.Rd
@@ -44,6 +44,7 @@ Other activations: \cr
 \code{\link{activation_softplus}()} \cr
 \code{\link{activation_softsign}()} \cr
 \code{\link{activation_sparse_plus}()} \cr
+\code{\link{activation_sparse_sigmoid}()} \cr
 \code{\link{activation_sparsemax}()} \cr
 \code{\link{activation_squareplus}()} \cr
 \code{\link{activation_tanh}()} \cr
diff --git a/man/activation_log_sigmoid.Rd b/man/activation_log_sigmoid.Rd
index cbc8b6a88..7115cc2cd 100644
--- a/man/activation_log_sigmoid.Rd
+++ b/man/activation_log_sigmoid.Rd
@@ -39,6 +39,7 @@ Other activations: \cr
 \code{\link{activation_softplus}()} \cr
 \code{\link{activation_softsign}()} \cr
 \code{\link{activation_sparse_plus}()} \cr
+\code{\link{activation_sparse_sigmoid}()} \cr
 \code{\link{activation_sparsemax}()} \cr
 \code{\link{activation_squareplus}()} \cr
 \code{\link{activation_tanh}()} \cr
diff --git a/man/activation_log_softmax.Rd b/man/activation_log_softmax.Rd
index b82963864..7e63db98f 100644
--- a/man/activation_log_softmax.Rd
+++ b/man/activation_log_softmax.Rd
@@ -47,6 +47,7 @@ Other activations: \cr
 \code{\link{activation_softplus}()} \cr
 \code{\link{activation_softsign}()} \cr
 \code{\link{activation_sparse_plus}()} \cr
+\code{\link{activation_sparse_sigmoid}()} \cr
 \code{\link{activation_sparsemax}()} \cr
 \code{\link{activation_squareplus}()} \cr
 \code{\link{activation_tanh}()} \cr
diff --git a/man/activation_mish.Rd b/man/activation_mish.Rd
index 08e63d148..67ef71e76 100644
--- a/man/activation_mish.Rd
+++ b/man/activation_mish.Rd
@@ -55,6 +55,7 @@ Other activations: \cr
 \code{\link{activation_softplus}()} \cr
 \code{\link{activation_softsign}()} \cr
 \code{\link{activation_sparse_plus}()} \cr
+\code{\link{activation_sparse_sigmoid}()} \cr
 \code{\link{activation_sparsemax}()} \cr
 \code{\link{activation_squareplus}()} \cr
 \code{\link{activation_tanh}()} \cr
diff --git a/man/activation_relu.Rd b/man/activation_relu.Rd
index 9cb858f7e..3b2878501 100644
--- a/man/activation_relu.Rd
+++ b/man/activation_relu.Rd
@@ -88,6 +88,7 @@ Other activations: \cr
 \code{\link{activation_softplus}()} \cr
 \code{\link{activation_softsign}()} \cr
 \code{\link{activation_sparse_plus}()} \cr
+\code{\link{activation_sparse_sigmoid}()} \cr
 \code{\link{activation_sparsemax}()} \cr
 \code{\link{activation_squareplus}()} \cr
 \code{\link{activation_tanh}()} \cr
diff --git a/man/activation_relu6.Rd b/man/activation_relu6.Rd
index 082c4eb81..59191ccd0 100644
--- a/man/activation_relu6.Rd
+++ b/man/activation_relu6.Rd
@@ -43,6 +43,7 @@ Other activations: \cr
 \code{\link{activation_softplus}()} \cr
 \code{\link{activation_softsign}()} \cr
 \code{\link{activation_sparse_plus}()} \cr
+\code{\link{activation_sparse_sigmoid}()} \cr
 \code{\link{activation_sparsemax}()} \cr
 \code{\link{activation_squareplus}()} \cr
 \code{\link{activation_tanh}()} \cr
diff --git a/man/activation_selu.Rd b/man/activation_selu.Rd
index 37e64d62a..dc88f750c 100644
--- a/man/activation_selu.Rd
+++ b/man/activation_selu.Rd
@@ -76,6 +76,7 @@ Other activations: \cr
 \code{\link{activation_softplus}()} \cr
 \code{\link{activation_softsign}()} \cr
 \code{\link{activation_sparse_plus}()} \cr
+\code{\link{activation_sparse_sigmoid}()} \cr
 \code{\link{activation_sparsemax}()} \cr
 \code{\link{activation_squareplus}()} \cr
 \code{\link{activation_tanh}()} \cr
diff --git a/man/activation_sigmoid.Rd b/man/activation_sigmoid.Rd
index 1a783324e..7a981d429 100644
--- a/man/activation_sigmoid.Rd
+++ b/man/activation_sigmoid.Rd
@@ -51,6 +51,7 @@ Other activations: \cr
 \code{\link{activation_softplus}()} \cr
 \code{\link{activation_softsign}()} \cr
 \code{\link{activation_sparse_plus}()} \cr
+\code{\link{activation_sparse_sigmoid}()} \cr
 \code{\link{activation_sparsemax}()} \cr
 \code{\link{activation_squareplus}()} \cr
 \code{\link{activation_tanh}()} \cr
diff --git a/man/activation_silu.Rd b/man/activation_silu.Rd
index f76a51377..fafde4ebd 100644
--- a/man/activation_silu.Rd
+++ b/man/activation_silu.Rd
@@ -53,6 +53,7 @@ Other activations: \cr
 \code{\link{activation_softplus}()} \cr
 \code{\link{activation_softsign}()} \cr
 \code{\link{activation_sparse_plus}()} \cr
+\code{\link{activation_sparse_sigmoid}()} \cr
 \code{\link{activation_sparsemax}()} \cr
 \code{\link{activation_squareplus}()} \cr
 \code{\link{activation_tanh}()} \cr
diff --git a/man/activation_soft_shrink.Rd b/man/activation_soft_shrink.Rd
index 449377cba..5d1156009 100644
--- a/man/activation_soft_shrink.Rd
+++ b/man/activation_soft_shrink.Rd
@@ -45,6 +45,7 @@ Other activations: \cr
 \code{\link{activation_softplus}()} \cr
 \code{\link{activation_softsign}()} \cr
 \code{\link{activation_sparse_plus}()} \cr
+\code{\link{activation_sparse_sigmoid}()} \cr
 \code{\link{activation_sparsemax}()} \cr
 \code{\link{activation_squareplus}()} \cr
 \code{\link{activation_tanh}()} \cr
diff --git a/man/activation_softmax.Rd b/man/activation_softmax.Rd
index 50f8f0721..3449f3bd2 100644
--- a/man/activation_softmax.Rd
+++ b/man/activation_softmax.Rd
@@ -58,6 +58,7 @@ Other activations: \cr
 \code{\link{activation_softplus}()} \cr
 \code{\link{activation_softsign}()} \cr
 \code{\link{activation_sparse_plus}()} \cr
+\code{\link{activation_sparse_sigmoid}()} \cr
 \code{\link{activation_sparsemax}()} \cr
 \code{\link{activation_squareplus}()} \cr
 \code{\link{activation_tanh}()} \cr
diff --git a/man/activation_softplus.Rd b/man/activation_softplus.Rd
index 393007af3..6fb1be008 100644
--- a/man/activation_softplus.Rd
+++ b/man/activation_softplus.Rd
@@ -43,6 +43,7 @@ Other activations: \cr
 \code{\link{activation_softmax}()} \cr
 \code{\link{activation_softsign}()} \cr
 \code{\link{activation_sparse_plus}()} \cr
+\code{\link{activation_sparse_sigmoid}()} \cr
 \code{\link{activation_sparsemax}()} \cr
 \code{\link{activation_squareplus}()} \cr
 \code{\link{activation_tanh}()} \cr
diff --git a/man/activation_softsign.Rd b/man/activation_softsign.Rd
index 5199e72db..910e25e26 100644
--- a/man/activation_softsign.Rd
+++ b/man/activation_softsign.Rd
@@ -43,6 +43,7 @@ Other activations: \cr
 \code{\link{activation_softmax}()} \cr
 \code{\link{activation_softplus}()} \cr
 \code{\link{activation_sparse_plus}()} \cr
+\code{\link{activation_sparse_sigmoid}()} \cr
 \code{\link{activation_sparsemax}()} \cr
 \code{\link{activation_squareplus}()} \cr
 \code{\link{activation_tanh}()} \cr
diff --git a/man/activation_sparse_plus.Rd b/man/activation_sparse_plus.Rd
index d704a7401..abde842d0 100644
--- a/man/activation_sparse_plus.Rd
+++ b/man/activation_sparse_plus.Rd
@@ -43,6 +43,7 @@ Other activations: \cr
 \code{\link{activation_softmax}()} \cr
 \code{\link{activation_softplus}()} \cr
 \code{\link{activation_softsign}()} \cr
+\code{\link{activation_sparse_sigmoid}()} \cr
 \code{\link{activation_sparsemax}()} \cr
 \code{\link{activation_squareplus}()} \cr
 \code{\link{activation_tanh}()} \cr
diff --git a/man/activation_sparse_sigmoid.Rd b/man/activation_sparse_sigmoid.Rd
new file mode 100644
index 000000000..465847082
--- /dev/null
+++ b/man/activation_sparse_sigmoid.Rd
@@ -0,0 +1,59 @@
+% Generated by roxygen2: do not edit by hand
+% Please edit documentation in R/activations.R
+\name{activation_sparse_sigmoid}
+\alias{activation_sparse_sigmoid}
+\title{Sparse sigmoid activation function.}
+\usage{
+activation_sparse_sigmoid(x)
+}
+\arguments{
+\item{x}{Input tensor.}
+}
+\value{
+A tensor, the result from applying the activation to the input tensor \code{x}.
+}
+\description{
+It is defined as
+
+\code{f(x) = 0} for \code{x <= -1},
+\code{f(x) = 0.5 * (x + 1)} for \verb{-1 < x < 1},
+\code{f(x) = 1} for \code{x >= 1}.
+}
+\section{Reference}{
+\itemize{
+\item \href{https://arxiv.org/pdf/1901.02324}{M. Blondel, A. F. T. Martins, V. Niculae, 2019}
+}
+}
+
+\seealso{
+Other activations: \cr
+\code{\link{activation_celu}()} \cr
+\code{\link{activation_elu}()} \cr
+\code{\link{activation_exponential}()} \cr
+\code{\link{activation_gelu}()} \cr
+\code{\link{activation_glu}()} \cr
+\code{\link{activation_hard_shrink}()} \cr
+\code{\link{activation_hard_sigmoid}()} \cr
+\code{\link{activation_hard_tanh}()} \cr
+\code{\link{activation_leaky_relu}()} \cr
+\code{\link{activation_linear}()} \cr
+\code{\link{activation_log_sigmoid}()} \cr
+\code{\link{activation_log_softmax}()} \cr
+\code{\link{activation_mish}()} \cr
+\code{\link{activation_relu}()} \cr
+\code{\link{activation_relu6}()} \cr
+\code{\link{activation_selu}()} \cr
+\code{\link{activation_sigmoid}()} \cr
+\code{\link{activation_silu}()} \cr
+\code{\link{activation_soft_shrink}()} \cr
+\code{\link{activation_softmax}()} \cr
+\code{\link{activation_softplus}()} \cr
+\code{\link{activation_softsign}()} \cr
+\code{\link{activation_sparse_plus}()} \cr
+\code{\link{activation_sparsemax}()} \cr
+\code{\link{activation_squareplus}()} \cr
+\code{\link{activation_tanh}()} \cr
+\code{\link{activation_tanh_shrink}()} \cr
+\code{\link{activation_threshold}()} \cr
+}
+\concept{activations}
diff --git a/man/activation_sparsemax.Rd b/man/activation_sparsemax.Rd
index f0fab02e6..a34c26368 100644
--- a/man/activation_sparsemax.Rd
+++ b/man/activation_sparsemax.Rd
@@ -52,6 +52,7 @@ Other activations: \cr
 \code{\link{activation_softplus}()} \cr
 \code{\link{activation_softsign}()} \cr
 \code{\link{activation_sparse_plus}()} \cr
+\code{\link{activation_sparse_sigmoid}()} \cr
 \code{\link{activation_squareplus}()} \cr
 \code{\link{activation_tanh}()} \cr
 \code{\link{activation_tanh_shrink}()} \cr
diff --git a/man/activation_squareplus.Rd b/man/activation_squareplus.Rd
index b2dbb896d..fe74a8ac7 100644
--- a/man/activation_squareplus.Rd
+++ b/man/activation_squareplus.Rd
@@ -52,6 +52,7 @@ Other activations: \cr
 \code{\link{activation_softplus}()} \cr
 \code{\link{activation_softsign}()} \cr
 \code{\link{activation_sparse_plus}()} \cr
+\code{\link{activation_sparse_sigmoid}()} \cr
 \code{\link{activation_sparsemax}()} \cr
 \code{\link{activation_tanh}()} \cr
 \code{\link{activation_tanh_shrink}()} \cr
diff --git a/man/activation_tanh.Rd b/man/activation_tanh.Rd
index 0869c3cd1..9d2749aa1 100644
--- a/man/activation_tanh.Rd
+++ b/man/activation_tanh.Rd
@@ -46,6 +46,7 @@ Other activations: \cr
 \code{\link{activation_softplus}()} \cr
 \code{\link{activation_softsign}()} \cr
 \code{\link{activation_sparse_plus}()} \cr
+\code{\link{activation_sparse_sigmoid}()} \cr
 \code{\link{activation_sparsemax}()} \cr
 \code{\link{activation_squareplus}()} \cr
 \code{\link{activation_tanh_shrink}()} \cr
diff --git a/man/activation_tanh_shrink.Rd b/man/activation_tanh_shrink.Rd
index cc17d06ed..b043ab2f1 100644
--- a/man/activation_tanh_shrink.Rd
+++ b/man/activation_tanh_shrink.Rd
@@ -42,6 +42,7 @@ Other activations: \cr
 \code{\link{activation_softplus}()} \cr
 \code{\link{activation_softsign}()} \cr
 \code{\link{activation_sparse_plus}()} \cr
+\code{\link{activation_sparse_sigmoid}()} \cr
 \code{\link{activation_sparsemax}()} \cr
 \code{\link{activation_squareplus}()} \cr
 \code{\link{activation_tanh}()} \cr
diff --git a/man/activation_threshold.Rd b/man/activation_threshold.Rd
index 419f4a576..65f5aa71a 100644
--- a/man/activation_threshold.Rd
+++ b/man/activation_threshold.Rd
@@ -47,6 +47,7 @@ Other activations: \cr
 \code{\link{activation_softplus}()} \cr
 \code{\link{activation_softsign}()} \cr
 \code{\link{activation_sparse_plus}()} \cr
+\code{\link{activation_sparse_sigmoid}()} \cr
 \code{\link{activation_sparsemax}()} \cr
 \code{\link{activation_squareplus}()} \cr
 \code{\link{activation_tanh}()} \cr
diff --git a/man/layer_tfsm.Rd b/man/layer_tfsm.Rd
index b969d813e..03c090911 100644
--- a/man/layer_tfsm.Rd
+++ b/man/layer_tfsm.Rd
@@ -59,8 +59,8 @@ model |> export_savedmodel("path/to/artifact")
 ## Output Type:
 ##   TensorSpec(shape=(None, 10), dtype=tf.float32, name=None)
 ## Captures:
-##   127996268992208: TensorSpec(shape=(), dtype=tf.resource, name=None)
-##   127996269001424: TensorSpec(shape=(), dtype=tf.resource, name=None)
+##   123789562403856: TensorSpec(shape=(), dtype=tf.resource, name=None)
+##   123789562409424: TensorSpec(shape=(), dtype=tf.resource, name=None)
 
 }\if{html}{\out{</div>}}
 
diff --git a/man/loss_binary_crossentropy.Rd b/man/loss_binary_crossentropy.Rd
index 8ff0d0154..977489d42 100644
--- a/man/loss_binary_crossentropy.Rd
+++ b/man/loss_binary_crossentropy.Rd
@@ -168,6 +168,7 @@ Other losses: \cr
 \code{\link{loss_binary_focal_crossentropy}()} \cr
 \code{\link{loss_categorical_crossentropy}()} \cr
 \code{\link{loss_categorical_focal_crossentropy}()} \cr
+\code{\link{loss_categorical_generalized_cross_entropy}()} \cr
 \code{\link{loss_categorical_hinge}()} \cr
 \code{\link{loss_circle}()} \cr
 \code{\link{loss_cosine_similarity}()} \cr
diff --git a/man/loss_binary_focal_crossentropy.Rd b/man/loss_binary_focal_crossentropy.Rd
index 3357fdac3..f115445e6 100644
--- a/man/loss_binary_focal_crossentropy.Rd
+++ b/man/loss_binary_focal_crossentropy.Rd
@@ -245,6 +245,7 @@ Other losses: \cr
 \code{\link{loss_binary_crossentropy}()} \cr
 \code{\link{loss_categorical_crossentropy}()} \cr
 \code{\link{loss_categorical_focal_crossentropy}()} \cr
+\code{\link{loss_categorical_generalized_cross_entropy}()} \cr
 \code{\link{loss_categorical_hinge}()} \cr
 \code{\link{loss_circle}()} \cr
 \code{\link{loss_cosine_similarity}()} \cr
diff --git a/man/loss_categorical_crossentropy.Rd b/man/loss_categorical_crossentropy.Rd
index 5520571e4..bfe0e2d7a 100644
--- a/man/loss_categorical_crossentropy.Rd
+++ b/man/loss_categorical_crossentropy.Rd
@@ -129,6 +129,7 @@ Other losses: \cr
 \code{\link{loss_binary_crossentropy}()} \cr
 \code{\link{loss_binary_focal_crossentropy}()} \cr
 \code{\link{loss_categorical_focal_crossentropy}()} \cr
+\code{\link{loss_categorical_generalized_cross_entropy}()} \cr
 \code{\link{loss_categorical_hinge}()} \cr
 \code{\link{loss_circle}()} \cr
 \code{\link{loss_cosine_similarity}()} \cr
diff --git a/man/loss_categorical_focal_crossentropy.Rd b/man/loss_categorical_focal_crossentropy.Rd
index 771ec256f..81c4ce8e9 100644
--- a/man/loss_categorical_focal_crossentropy.Rd
+++ b/man/loss_categorical_focal_crossentropy.Rd
@@ -172,6 +172,7 @@ Other losses: \cr
 \code{\link{loss_binary_crossentropy}()} \cr
 \code{\link{loss_binary_focal_crossentropy}()} \cr
 \code{\link{loss_categorical_crossentropy}()} \cr
+\code{\link{loss_categorical_generalized_cross_entropy}()} \cr
 \code{\link{loss_categorical_hinge}()} \cr
 \code{\link{loss_circle}()} \cr
 \code{\link{loss_cosine_similarity}()} \cr
diff --git a/man/loss_categorical_generalized_cross_entropy.Rd b/man/loss_categorical_generalized_cross_entropy.Rd
new file mode 100644
index 000000000..feaff93a0
--- /dev/null
+++ b/man/loss_categorical_generalized_cross_entropy.Rd
@@ -0,0 +1,119 @@
+% Generated by roxygen2: do not edit by hand
+% Please edit documentation in R/losses.R
+\name{loss_categorical_generalized_cross_entropy}
+\alias{loss_categorical_generalized_cross_entropy}
+\title{Computes the generalized cross entropy loss.}
+\usage{
+loss_categorical_generalized_cross_entropy(
+  y_true,
+  y_pred,
+  q = 0.5,
+  ...,
+  reduction = "sum_over_batch_size",
+  name = "categorical_generalized_cross_entropy",
+  dtype = NULL
+)
+}
+\arguments{
+\item{y_true}{Integer class indices with shape \code{(batch_size)} or \verb{(batch_size, 1)}.}
+
+\item{y_pred}{Predicted class probabilities with shape \verb{(batch_size, num_classes)}.}
+
+\item{q}{Float in \verb{(0, 1)}. Controls the transition between cross entropy and mean
+absolute error. Defaults to \code{0.5}.
+\itemize{
+\item As \code{q} approaches \code{0}: behaves like categorical cross entropy.
+\item As \code{q} approaches \code{1}: behaves like mean absolute error.
+}}
+
+\item{...}{For forward/backward compatibility.}
+
+\item{reduction}{Type of reduction to apply to the loss. In almost all cases
+this should be \code{"sum_over_batch_size"}. Supported options are
+\code{"sum"}, \code{"sum_over_batch_size"}, \code{"mean"},
+\code{"mean_with_sample_weight"} or \code{NULL}. \code{"sum"} sums the loss,
+\code{"sum_over_batch_size"} and \code{"mean"} sum the loss and divide by the
+sample size, and \code{"mean_with_sample_weight"} sums the loss and
+divides by the sum of the sample weights. \code{"none"} and \code{NULL}
+perform no aggregation. Defaults to \code{"sum_over_batch_size"}.}
+
+\item{name}{Optional name for the loss instance.}
+
+\item{dtype}{Dtype used for loss computations. Defaults to \code{config_floatx()} (the global
+float type).}
+}
+\value{
+Generalized cross entropy loss value(s).
+}
+\description{
+The generalized cross entropy (GCE) loss offers robustness to noisy labels by
+interpolating between categorical cross entropy (\code{q -> 0}) and mean absolute
+error (\code{q -> 1}). For a true-class probability \code{p} and noise parameter \code{q},
+the loss is \code{loss = (1 - p^q) / q}.
+}
+\section{References}{
+\itemize{
+\item Zhang & Sabuncu (2018), "Generalized Cross Entropy Loss for Training Deep
+Neural Networks with Noisy Labels"
+}
+}
+
+\section{Examples}{
+\if{html}{\out{<div class="sourceCode r">}}\preformatted{y_true <- c(0L, 1L, 0L, 1L)
+y_pred <- rbind(
+  c(0.7, 0.3),
+  c(0.2, 0.8),
+  c(0.6, 0.4),
+  c(0.4, 0.6)
+)
+gce <- loss_categorical_generalized_cross_entropy(q = 0.7)
+gce(y_true, y_pred)
+}\if{html}{\out{</div>}}
+
+\if{html}{\out{<div class="sourceCode">}}\preformatted{## tf.Tensor(0.34529287, shape=(), dtype=float32)
+
+}\if{html}{\out{</div>}}
+}
+
+\seealso{
+Other losses: \cr
+\code{\link{Loss}()} \cr
+\code{\link{loss_binary_crossentropy}()} \cr
+\code{\link{loss_binary_focal_crossentropy}()} \cr
+\code{\link{loss_categorical_crossentropy}()} \cr
+\code{\link{loss_categorical_focal_crossentropy}()} \cr
+\code{\link{loss_categorical_hinge}()} \cr
+\code{\link{loss_circle}()} \cr
+\code{\link{loss_cosine_similarity}()} \cr
+\code{\link{loss_ctc}()} \cr
+\code{\link{loss_dice}()} \cr
+\code{\link{loss_hinge}()} \cr
+\code{\link{loss_huber}()} \cr
+\code{\link{loss_kl_divergence}()} \cr
+\code{\link{loss_log_cosh}()} \cr
+\code{\link{loss_mean_absolute_error}()} \cr
+\code{\link{loss_mean_absolute_percentage_error}()} \cr
+\code{\link{loss_mean_squared_error}()} \cr
+\code{\link{loss_mean_squared_logarithmic_error}()} \cr
+\code{\link{loss_poisson}()} \cr
+\code{\link{loss_sparse_categorical_crossentropy}()} \cr
+\code{\link{loss_squared_hinge}()} \cr
+\code{\link{loss_tversky}()} \cr
+\code{\link{metric_binary_crossentropy}()} \cr
+\code{\link{metric_binary_focal_crossentropy}()} \cr
+\code{\link{metric_categorical_crossentropy}()} \cr
+\code{\link{metric_categorical_focal_crossentropy}()} \cr
+\code{\link{metric_categorical_hinge}()} \cr
+\code{\link{metric_hinge}()} \cr
+\code{\link{metric_huber}()} \cr
+\code{\link{metric_kl_divergence}()} \cr
+\code{\link{metric_log_cosh}()} \cr
+\code{\link{metric_mean_absolute_error}()} \cr
+\code{\link{metric_mean_absolute_percentage_error}()} \cr
+\code{\link{metric_mean_squared_error}()} \cr
+\code{\link{metric_mean_squared_logarithmic_error}()} \cr
+\code{\link{metric_poisson}()} \cr
+\code{\link{metric_sparse_categorical_crossentropy}()} \cr
+\code{\link{metric_squared_hinge}()} \cr
+}
+\concept{losses}
diff --git a/man/loss_categorical_hinge.Rd b/man/loss_categorical_hinge.Rd
index ec97a98c2..d5d6d81d3 100644
--- a/man/loss_categorical_hinge.Rd
+++ b/man/loss_categorical_hinge.Rd
@@ -68,6 +68,7 @@ Other losses: \cr
 \code{\link{loss_binary_focal_crossentropy}()} \cr
 \code{\link{loss_categorical_crossentropy}()} \cr
 \code{\link{loss_categorical_focal_crossentropy}()} \cr
+\code{\link{loss_categorical_generalized_cross_entropy}()} \cr
 \code{\link{loss_circle}()} \cr
 \code{\link{loss_cosine_similarity}()} \cr
 \code{\link{loss_ctc}()} \cr
diff --git a/man/loss_circle.Rd b/man/loss_circle.Rd
index 3dab46e2d..092b8bb11 100644
--- a/man/loss_circle.Rd
+++ b/man/loss_circle.Rd
@@ -102,6 +102,7 @@ Other losses: \cr
 \code{\link{loss_binary_focal_crossentropy}()} \cr
 \code{\link{loss_categorical_crossentropy}()} \cr
 \code{\link{loss_categorical_focal_crossentropy}()} \cr
+\code{\link{loss_categorical_generalized_cross_entropy}()} \cr
 \code{\link{loss_categorical_hinge}()} \cr
 \code{\link{loss_cosine_similarity}()} \cr
 \code{\link{loss_ctc}()} \cr
diff --git a/man/loss_cosine_similarity.Rd b/man/loss_cosine_similarity.Rd
index 7f3376671..63f000032 100644
--- a/man/loss_cosine_similarity.Rd
+++ b/man/loss_cosine_similarity.Rd
@@ -81,6 +81,7 @@ Other losses: \cr
 \code{\link{loss_binary_focal_crossentropy}()} \cr
 \code{\link{loss_categorical_crossentropy}()} \cr
 \code{\link{loss_categorical_focal_crossentropy}()} \cr
+\code{\link{loss_categorical_generalized_cross_entropy}()} \cr
 \code{\link{loss_categorical_hinge}()} \cr
 \code{\link{loss_circle}()} \cr
 \code{\link{loss_ctc}()} \cr
diff --git a/man/loss_ctc.Rd b/man/loss_ctc.Rd
index ef827189e..63221ab05 100644
--- a/man/loss_ctc.Rd
+++ b/man/loss_ctc.Rd
@@ -54,6 +54,7 @@ Other losses: \cr
 \code{\link{loss_binary_focal_crossentropy}()} \cr
 \code{\link{loss_categorical_crossentropy}()} \cr
 \code{\link{loss_categorical_focal_crossentropy}()} \cr
+\code{\link{loss_categorical_generalized_cross_entropy}()} \cr
 \code{\link{loss_categorical_hinge}()} \cr
 \code{\link{loss_circle}()} \cr
 \code{\link{loss_cosine_similarity}()} \cr
diff --git a/man/loss_dice.Rd b/man/loss_dice.Rd
index 4620b21ec..d2e061b2b 100644
--- a/man/loss_dice.Rd
+++ b/man/loss_dice.Rd
@@ -91,6 +91,7 @@ Other losses: \cr
 \code{\link{loss_binary_focal_crossentropy}()} \cr
 \code{\link{loss_categorical_crossentropy}()} \cr
 \code{\link{loss_categorical_focal_crossentropy}()} \cr
+\code{\link{loss_categorical_generalized_cross_entropy}()} \cr
 \code{\link{loss_categorical_hinge}()} \cr
 \code{\link{loss_circle}()} \cr
 \code{\link{loss_cosine_similarity}()} \cr
diff --git a/man/loss_hinge.Rd b/man/loss_hinge.Rd
index eda42851c..263704a69 100644
--- a/man/loss_hinge.Rd
+++ b/man/loss_hinge.Rd
@@ -74,6 +74,7 @@ Other losses: \cr
 \code{\link{loss_binary_focal_crossentropy}()} \cr
 \code{\link{loss_categorical_crossentropy}()} \cr
 \code{\link{loss_categorical_focal_crossentropy}()} \cr
+\code{\link{loss_categorical_generalized_cross_entropy}()} \cr
 \code{\link{loss_categorical_hinge}()} \cr
 \code{\link{loss_circle}()} \cr
 \code{\link{loss_cosine_similarity}()} \cr
diff --git a/man/loss_huber.Rd b/man/loss_huber.Rd
index a60f5104f..327e7faf5 100644
--- a/man/loss_huber.Rd
+++ b/man/loss_huber.Rd
@@ -72,6 +72,7 @@ Other losses: \cr
 \code{\link{loss_binary_focal_crossentropy}()} \cr
 \code{\link{loss_categorical_crossentropy}()} \cr
 \code{\link{loss_categorical_focal_crossentropy}()} \cr
+\code{\link{loss_categorical_generalized_cross_entropy}()} \cr
 \code{\link{loss_categorical_hinge}()} \cr
 \code{\link{loss_circle}()} \cr
 \code{\link{loss_cosine_similarity}()} \cr
diff --git a/man/loss_kl_divergence.Rd b/man/loss_kl_divergence.Rd
index 9bc471e56..043afd3ab 100644
--- a/man/loss_kl_divergence.Rd
+++ b/man/loss_kl_divergence.Rd
@@ -73,6 +73,7 @@ Other losses: \cr
 \code{\link{loss_binary_focal_crossentropy}()} \cr
 \code{\link{loss_categorical_crossentropy}()} \cr
 \code{\link{loss_categorical_focal_crossentropy}()} \cr
+\code{\link{loss_categorical_generalized_cross_entropy}()} \cr
 \code{\link{loss_categorical_hinge}()} \cr
 \code{\link{loss_circle}()} \cr
 \code{\link{loss_cosine_similarity}()} \cr
diff --git a/man/loss_log_cosh.Rd b/man/loss_log_cosh.Rd
index e69e9e7cc..9d8110d72 100644
--- a/man/loss_log_cosh.Rd
+++ b/man/loss_log_cosh.Rd
@@ -65,6 +65,7 @@ Other losses: \cr
 \code{\link{loss_binary_focal_crossentropy}()} \cr
 \code{\link{loss_categorical_crossentropy}()} \cr
 \code{\link{loss_categorical_focal_crossentropy}()} \cr
+\code{\link{loss_categorical_generalized_cross_entropy}()} \cr
 \code{\link{loss_categorical_hinge}()} \cr
 \code{\link{loss_circle}()} \cr
 \code{\link{loss_cosine_similarity}()} \cr
diff --git a/man/loss_mean_absolute_error.Rd b/man/loss_mean_absolute_error.Rd
index 7d8c0ec72..911e3b163 100644
--- a/man/loss_mean_absolute_error.Rd
+++ b/man/loss_mean_absolute_error.Rd
@@ -64,6 +64,7 @@ Other losses: \cr
 \code{\link{loss_binary_focal_crossentropy}()} \cr
 \code{\link{loss_categorical_crossentropy}()} \cr
 \code{\link{loss_categorical_focal_crossentropy}()} \cr
+\code{\link{loss_categorical_generalized_cross_entropy}()} \cr
 \code{\link{loss_categorical_hinge}()} \cr
 \code{\link{loss_circle}()} \cr
 \code{\link{loss_cosine_similarity}()} \cr
diff --git a/man/loss_mean_absolute_percentage_error.Rd b/man/loss_mean_absolute_percentage_error.Rd
index 60531319f..dd3e4c0b8 100644
--- a/man/loss_mean_absolute_percentage_error.Rd
+++ b/man/loss_mean_absolute_percentage_error.Rd
@@ -69,6 +69,7 @@ Other losses: \cr
 \code{\link{loss_binary_focal_crossentropy}()} \cr
 \code{\link{loss_categorical_crossentropy}()} \cr
 \code{\link{loss_categorical_focal_crossentropy}()} \cr
+\code{\link{loss_categorical_generalized_cross_entropy}()} \cr
 \code{\link{loss_categorical_hinge}()} \cr
 \code{\link{loss_circle}()} \cr
 \code{\link{loss_cosine_similarity}()} \cr
diff --git a/man/loss_mean_squared_error.Rd b/man/loss_mean_squared_error.Rd
index 2f53ffb18..88f5c5ccc 100644
--- a/man/loss_mean_squared_error.Rd
+++ b/man/loss_mean_squared_error.Rd
@@ -64,6 +64,7 @@ Other losses: \cr
 \code{\link{loss_binary_focal_crossentropy}()} \cr
 \code{\link{loss_categorical_crossentropy}()} \cr
 \code{\link{loss_categorical_focal_crossentropy}()} \cr
+\code{\link{loss_categorical_generalized_cross_entropy}()} \cr
 \code{\link{loss_categorical_hinge}()} \cr
 \code{\link{loss_circle}()} \cr
 \code{\link{loss_cosine_similarity}()} \cr
diff --git a/man/loss_mean_squared_logarithmic_error.Rd b/man/loss_mean_squared_logarithmic_error.Rd
index 60246fdbe..423b49c57 100644
--- a/man/loss_mean_squared_logarithmic_error.Rd
+++ b/man/loss_mean_squared_logarithmic_error.Rd
@@ -68,6 +68,7 @@ Other losses: \cr
 \code{\link{loss_binary_focal_crossentropy}()} \cr
 \code{\link{loss_categorical_crossentropy}()} \cr
 \code{\link{loss_categorical_focal_crossentropy}()} \cr
+\code{\link{loss_categorical_generalized_cross_entropy}()} \cr
 \code{\link{loss_categorical_hinge}()} \cr
 \code{\link{loss_circle}()} \cr
 \code{\link{loss_cosine_similarity}()} \cr
diff --git a/man/loss_poisson.Rd b/man/loss_poisson.Rd
index 2bfa6f922..81a436336 100644
--- a/man/loss_poisson.Rd
+++ b/man/loss_poisson.Rd
@@ -69,6 +69,7 @@ Other losses: \cr
 \code{\link{loss_binary_focal_crossentropy}()} \cr
 \code{\link{loss_categorical_crossentropy}()} \cr
 \code{\link{loss_categorical_focal_crossentropy}()} \cr
+\code{\link{loss_categorical_generalized_cross_entropy}()} \cr
 \code{\link{loss_categorical_hinge}()} \cr
 \code{\link{loss_circle}()} \cr
 \code{\link{loss_cosine_similarity}()} \cr
diff --git a/man/loss_sparse_categorical_crossentropy.Rd b/man/loss_sparse_categorical_crossentropy.Rd
index bd3678086..cd308020a 100644
--- a/man/loss_sparse_categorical_crossentropy.Rd
+++ b/man/loss_sparse_categorical_crossentropy.Rd
@@ -134,6 +134,7 @@ Other losses: \cr
 \code{\link{loss_binary_focal_crossentropy}()} \cr
 \code{\link{loss_categorical_crossentropy}()} \cr
 \code{\link{loss_categorical_focal_crossentropy}()} \cr
+\code{\link{loss_categorical_generalized_cross_entropy}()} \cr
 \code{\link{loss_categorical_hinge}()} \cr
 \code{\link{loss_circle}()} \cr
 \code{\link{loss_cosine_similarity}()} \cr
diff --git a/man/loss_squared_hinge.Rd b/man/loss_squared_hinge.Rd
index 900f118ea..b5d45543e 100644
--- a/man/loss_squared_hinge.Rd
+++ b/man/loss_squared_hinge.Rd
@@ -69,6 +69,7 @@ Other losses: \cr
 \code{\link{loss_binary_focal_crossentropy}()} \cr
 \code{\link{loss_categorical_crossentropy}()} \cr
 \code{\link{loss_categorical_focal_crossentropy}()} \cr
+\code{\link{loss_categorical_generalized_cross_entropy}()} \cr
 \code{\link{loss_categorical_hinge}()} \cr
 \code{\link{loss_circle}()} \cr
 \code{\link{loss_cosine_similarity}()} \cr
diff --git a/man/loss_tversky.Rd b/man/loss_tversky.Rd
index 90e6b4cd9..581e9d4e8 100644
--- a/man/loss_tversky.Rd
+++ b/man/loss_tversky.Rd
@@ -77,6 +77,7 @@ Other losses: \cr
 \code{\link{loss_binary_focal_crossentropy}()} \cr
 \code{\link{loss_categorical_crossentropy}()} \cr
 \code{\link{loss_categorical_focal_crossentropy}()} \cr
+\code{\link{loss_categorical_generalized_cross_entropy}()} \cr
 \code{\link{loss_categorical_hinge}()} \cr
 \code{\link{loss_circle}()} \cr
 \code{\link{loss_cosine_similarity}()} \cr
diff --git a/man/metric_binary_crossentropy.Rd b/man/metric_binary_crossentropy.Rd
index 319ec923a..f54977036 100644
--- a/man/metric_binary_crossentropy.Rd
+++ b/man/metric_binary_crossentropy.Rd
@@ -91,6 +91,7 @@ Other losses: \cr
 \code{\link{loss_binary_focal_crossentropy}()} \cr
 \code{\link{loss_categorical_crossentropy}()} \cr
 \code{\link{loss_categorical_focal_crossentropy}()} \cr
+\code{\link{loss_categorical_generalized_cross_entropy}()} \cr
 \code{\link{loss_categorical_hinge}()} \cr
 \code{\link{loss_circle}()} \cr
 \code{\link{loss_cosine_similarity}()} \cr
diff --git a/man/metric_binary_focal_crossentropy.Rd b/man/metric_binary_focal_crossentropy.Rd
index 70437522e..b4a486291 100644
--- a/man/metric_binary_focal_crossentropy.Rd
+++ b/man/metric_binary_focal_crossentropy.Rd
@@ -82,6 +82,7 @@ Other losses: \cr
 \code{\link{loss_binary_focal_crossentropy}()} \cr
 \code{\link{loss_categorical_crossentropy}()} \cr
 \code{\link{loss_categorical_focal_crossentropy}()} \cr
+\code{\link{loss_categorical_generalized_cross_entropy}()} \cr
 \code{\link{loss_categorical_hinge}()} \cr
 \code{\link{loss_circle}()} \cr
 \code{\link{loss_cosine_similarity}()} \cr
diff --git a/man/metric_categorical_crossentropy.Rd b/man/metric_categorical_crossentropy.Rd
index abc3c0b67..eb6be3e53 100644
--- a/man/metric_categorical_crossentropy.Rd
+++ b/man/metric_categorical_crossentropy.Rd
@@ -107,6 +107,7 @@ Other losses: \cr
 \code{\link{loss_binary_focal_crossentropy}()} \cr
 \code{\link{loss_categorical_crossentropy}()} \cr
 \code{\link{loss_categorical_focal_crossentropy}()} \cr
+\code{\link{loss_categorical_generalized_cross_entropy}()} \cr
 \code{\link{loss_categorical_hinge}()} \cr
 \code{\link{loss_circle}()} \cr
 \code{\link{loss_cosine_similarity}()} \cr
diff --git a/man/metric_categorical_focal_crossentropy.Rd b/man/metric_categorical_focal_crossentropy.Rd
index daa6255fb..4f75ab1d2 100644
--- a/man/metric_categorical_focal_crossentropy.Rd
+++ b/man/metric_categorical_focal_crossentropy.Rd
@@ -65,6 +65,7 @@ Other losses: \cr
 \code{\link{loss_binary_focal_crossentropy}()} \cr
 \code{\link{loss_categorical_crossentropy}()} \cr
 \code{\link{loss_categorical_focal_crossentropy}()} \cr
+\code{\link{loss_categorical_generalized_cross_entropy}()} \cr
 \code{\link{loss_categorical_hinge}()} \cr
 \code{\link{loss_circle}()} \cr
 \code{\link{loss_cosine_similarity}()} \cr
diff --git a/man/metric_categorical_hinge.Rd b/man/metric_categorical_hinge.Rd
index cb9084826..d4a3849c9 100644
--- a/man/metric_categorical_hinge.Rd
+++ b/man/metric_categorical_hinge.Rd
@@ -70,6 +70,7 @@ Other losses: \cr
 \code{\link{loss_binary_focal_crossentropy}()} \cr
 \code{\link{loss_categorical_crossentropy}()} \cr
 \code{\link{loss_categorical_focal_crossentropy}()} \cr
+\code{\link{loss_categorical_generalized_cross_entropy}()} \cr
 \code{\link{loss_categorical_hinge}()} \cr
 \code{\link{loss_circle}()} \cr
 \code{\link{loss_cosine_similarity}()} \cr
diff --git a/man/metric_hinge.Rd b/man/metric_hinge.Rd
index e1cdc81d5..91a5eb17d 100644
--- a/man/metric_hinge.Rd
+++ b/man/metric_hinge.Rd
@@ -69,6 +69,7 @@ Other losses: \cr
 \code{\link{loss_binary_focal_crossentropy}()} \cr
 \code{\link{loss_categorical_crossentropy}()} \cr
 \code{\link{loss_categorical_focal_crossentropy}()} \cr
+\code{\link{loss_categorical_generalized_cross_entropy}()} \cr
 \code{\link{loss_categorical_hinge}()} \cr
 \code{\link{loss_circle}()} \cr
 \code{\link{loss_cosine_similarity}()} \cr
diff --git a/man/metric_huber.Rd b/man/metric_huber.Rd
index 02c299731..c759ebee4 100644
--- a/man/metric_huber.Rd
+++ b/man/metric_huber.Rd
@@ -46,6 +46,7 @@ Other losses: \cr
 \code{\link{loss_binary_focal_crossentropy}()} \cr
 \code{\link{loss_categorical_crossentropy}()} \cr
 \code{\link{loss_categorical_focal_crossentropy}()} \cr
+\code{\link{loss_categorical_generalized_cross_entropy}()} \cr
 \code{\link{loss_categorical_hinge}()} \cr
 \code{\link{loss_circle}()} \cr
 \code{\link{loss_cosine_similarity}()} \cr
diff --git a/man/metric_kl_divergence.Rd b/man/metric_kl_divergence.Rd
index a0cd02cfd..6883856bd 100644
--- a/man/metric_kl_divergence.Rd
+++ b/man/metric_kl_divergence.Rd
@@ -75,6 +75,7 @@ Other losses: \cr
 \code{\link{loss_binary_focal_crossentropy}()} \cr
 \code{\link{loss_categorical_crossentropy}()} \cr
 \code{\link{loss_categorical_focal_crossentropy}()} \cr
+\code{\link{loss_categorical_generalized_cross_entropy}()} \cr
 \code{\link{loss_categorical_hinge}()} \cr
 \code{\link{loss_circle}()} \cr
 \code{\link{loss_cosine_similarity}()} \cr
diff --git a/man/metric_log_cosh.Rd b/man/metric_log_cosh.Rd
index 845873832..7c92ad839 100644
--- a/man/metric_log_cosh.Rd
+++ b/man/metric_log_cosh.Rd
@@ -44,6 +44,7 @@ Other losses: \cr
 \code{\link{loss_binary_focal_crossentropy}()} \cr
 \code{\link{loss_categorical_crossentropy}()} \cr
 \code{\link{loss_categorical_focal_crossentropy}()} \cr
+\code{\link{loss_categorical_generalized_cross_entropy}()} \cr
 \code{\link{loss_categorical_hinge}()} \cr
 \code{\link{loss_circle}()} \cr
 \code{\link{loss_cosine_similarity}()} \cr
diff --git a/man/metric_mean_absolute_error.Rd b/man/metric_mean_absolute_error.Rd
index 0b1d6c3d7..43843b51f 100644
--- a/man/metric_mean_absolute_error.Rd
+++ b/man/metric_mean_absolute_error.Rd
@@ -78,6 +78,7 @@ Other losses: \cr
 \code{\link{loss_binary_focal_crossentropy}()} \cr
 \code{\link{loss_categorical_crossentropy}()} \cr
 \code{\link{loss_categorical_focal_crossentropy}()} \cr
+\code{\link{loss_categorical_generalized_cross_entropy}()} \cr
 \code{\link{loss_categorical_hinge}()} \cr
 \code{\link{loss_circle}()} \cr
 \code{\link{loss_cosine_similarity}()} \cr
diff --git a/man/metric_mean_absolute_percentage_error.Rd b/man/metric_mean_absolute_percentage_error.Rd
index 930e47b4f..2550bbba6 100644
--- a/man/metric_mean_absolute_percentage_error.Rd
+++ b/man/metric_mean_absolute_percentage_error.Rd
@@ -82,6 +82,7 @@ Other losses: \cr
 \code{\link{loss_binary_focal_crossentropy}()} \cr
 \code{\link{loss_categorical_crossentropy}()} \cr
 \code{\link{loss_categorical_focal_crossentropy}()} \cr
+\code{\link{loss_categorical_generalized_cross_entropy}()} \cr
 \code{\link{loss_categorical_hinge}()} \cr
 \code{\link{loss_circle}()} \cr
 \code{\link{loss_cosine_similarity}()} \cr
diff --git a/man/metric_mean_squared_error.Rd b/man/metric_mean_squared_error.Rd
index ee43f89a4..f8ae1c986 100644
--- a/man/metric_mean_squared_error.Rd
+++ b/man/metric_mean_squared_error.Rd
@@ -58,6 +58,7 @@ Other losses: \cr
 \code{\link{loss_binary_focal_crossentropy}()} \cr
 \code{\link{loss_categorical_crossentropy}()} \cr
 \code{\link{loss_categorical_focal_crossentropy}()} \cr
+\code{\link{loss_categorical_generalized_cross_entropy}()} \cr
 \code{\link{loss_categorical_hinge}()} \cr
 \code{\link{loss_circle}()} \cr
 \code{\link{loss_cosine_similarity}()} \cr
diff --git a/man/metric_mean_squared_logarithmic_error.Rd b/man/metric_mean_squared_logarithmic_error.Rd
index ffbe876d2..4ffd42c06 100644
--- a/man/metric_mean_squared_logarithmic_error.Rd
+++ b/man/metric_mean_squared_logarithmic_error.Rd
@@ -82,6 +82,7 @@ Other losses: \cr
 \code{\link{loss_binary_focal_crossentropy}()} \cr
 \code{\link{loss_categorical_crossentropy}()} \cr
 \code{\link{loss_categorical_focal_crossentropy}()} \cr
+\code{\link{loss_categorical_generalized_cross_entropy}()} \cr
 \code{\link{loss_categorical_hinge}()} \cr
 \code{\link{loss_circle}()} \cr
 \code{\link{loss_cosine_similarity}()} \cr
diff --git a/man/metric_poisson.Rd b/man/metric_poisson.Rd
index 774051935..761dee0c1 100644
--- a/man/metric_poisson.Rd
+++ b/man/metric_poisson.Rd
@@ -73,6 +73,7 @@ Other losses: \cr
 \code{\link{loss_binary_focal_crossentropy}()} \cr
 \code{\link{loss_categorical_crossentropy}()} \cr
 \code{\link{loss_categorical_focal_crossentropy}()} \cr
+\code{\link{loss_categorical_generalized_cross_entropy}()} \cr
 \code{\link{loss_categorical_hinge}()} \cr
 \code{\link{loss_circle}()} \cr
 \code{\link{loss_cosine_similarity}()} \cr
diff --git a/man/metric_sparse_categorical_crossentropy.Rd b/man/metric_sparse_categorical_crossentropy.Rd
index 2835e672d..b2f3e2cc4 100644
--- a/man/metric_sparse_categorical_crossentropy.Rd
+++ b/man/metric_sparse_categorical_crossentropy.Rd
@@ -100,6 +100,7 @@ Other losses: \cr
 \code{\link{loss_binary_focal_crossentropy}()} \cr
 \code{\link{loss_categorical_crossentropy}()} \cr
 \code{\link{loss_categorical_focal_crossentropy}()} \cr
+\code{\link{loss_categorical_generalized_cross_entropy}()} \cr
 \code{\link{loss_categorical_hinge}()} \cr
 \code{\link{loss_circle}()} \cr
 \code{\link{loss_cosine_similarity}()} \cr
diff --git a/man/metric_squared_hinge.Rd b/man/metric_squared_hinge.Rd
index ddd1ddce7..5b841e9da 100644
--- a/man/metric_squared_hinge.Rd
+++ b/man/metric_squared_hinge.Rd
@@ -69,6 +69,7 @@ Other losses: \cr
 \code{\link{loss_binary_focal_crossentropy}()} \cr
 \code{\link{loss_categorical_crossentropy}()} \cr
 \code{\link{loss_categorical_focal_crossentropy}()} \cr
+\code{\link{loss_categorical_generalized_cross_entropy}()} \cr
 \code{\link{loss_categorical_hinge}()} \cr
 \code{\link{loss_circle}()} \cr
 \code{\link{loss_cosine_similarity}()} \cr
diff --git a/man/op_abs.Rd b/man/op_abs.Rd
index c42f5417e..fbc8361f0 100644
--- a/man/op_abs.Rd
+++ b/man/op_abs.Rd
@@ -29,6 +29,7 @@ op_abs(x)
 Other numpy ops: \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -45,6 +46,7 @@ Other numpy ops: \cr
 \code{\link{op_argsort}()} \cr
 \code{\link{op_array}()} \cr
 \code{\link{op_average}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_bincount}()} \cr
 \code{\link{op_bitwise_and}()} \cr
 \code{\link{op_bitwise_invert}()} \cr
@@ -53,12 +55,15 @@ Other numpy ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_clip}()} \cr
 \code{\link{op_concatenate}()} \cr
 \code{\link{op_conj}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -67,6 +72,7 @@ Other numpy ops: \cr
 \code{\link{op_ctc_decode}()} \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_diag}()} \cr
 \code{\link{op_diagflat}()} \cr
 \code{\link{op_diagonal}()} \cr
@@ -91,6 +97,9 @@ Other numpy ops: \cr
 \code{\link{op_get_item}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -100,6 +109,7 @@ Other numpy ops: \cr
 \code{\link{op_isfinite}()} \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
+\code{\link{op_kaiser}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
 \code{\link{op_less_equal}()} \cr
@@ -180,6 +190,8 @@ Other numpy ops: \cr
 \code{\link{op_var}()} \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_zeros}()} \cr
@@ -188,6 +200,7 @@ Other numpy ops: \cr
 Other ops: \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -206,6 +219,7 @@ Other ops: \cr
 \code{\link{op_associative_scan}()} \cr
 \code{\link{op_average}()} \cr
 \code{\link{op_average_pool}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_batch_normalization}()} \cr
 \code{\link{op_binary_crossentropy}()} \cr
 \code{\link{op_bincount}()} \cr
@@ -216,9 +230,11 @@ Other ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
 \code{\link{op_cast}()} \cr
 \code{\link{op_categorical_crossentropy}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_celu}()} \cr
 \code{\link{op_cholesky}()} \cr
@@ -231,6 +247,7 @@ Other ops: \cr
 \code{\link{op_convert_to_numpy}()} \cr
 \code{\link{op_convert_to_tensor}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -241,6 +258,7 @@ Other ops: \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
 \code{\link{op_custom_gradient}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_depthwise_conv}()} \cr
 \code{\link{op_det}()} \cr
 \code{\link{op_diag}()} \cr
@@ -280,10 +298,13 @@ Other ops: \cr
 \code{\link{op_glu}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
 \code{\link{op_hard_shrink}()} \cr
 \code{\link{op_hard_sigmoid}()} \cr
 \code{\link{op_hard_silu}()} \cr
 \code{\link{op_hard_tanh}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -310,6 +331,8 @@ Other ops: \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
 \code{\link{op_istft}()} \cr
+\code{\link{op_kaiser}()} \cr
+\code{\link{op_layer_normalization}()} \cr
 \code{\link{op_leaky_relu}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
@@ -409,6 +432,7 @@ Other ops: \cr
 \code{\link{op_sort}()} \cr
 \code{\link{op_sparse_categorical_crossentropy}()} \cr
 \code{\link{op_sparse_plus}()} \cr
+\code{\link{op_sparse_sigmoid}()} \cr
 \code{\link{op_sparsemax}()} \cr
 \code{\link{op_split}()} \cr
 \code{\link{op_sqrt}()} \cr
@@ -446,6 +470,8 @@ Other ops: \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
 \code{\link{op_vectorized_map}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_while_loop}()} \cr
diff --git a/man/op_add.Rd b/man/op_add.Rd
index a7a2a649e..3eb4d005b 100644
--- a/man/op_add.Rd
+++ b/man/op_add.Rd
@@ -74,6 +74,7 @@ op_add(x, x)
 Other numpy ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -90,6 +91,7 @@ Other numpy ops: \cr
 \code{\link{op_argsort}()} \cr
 \code{\link{op_array}()} \cr
 \code{\link{op_average}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_bincount}()} \cr
 \code{\link{op_bitwise_and}()} \cr
 \code{\link{op_bitwise_invert}()} \cr
@@ -98,12 +100,15 @@ Other numpy ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_clip}()} \cr
 \code{\link{op_concatenate}()} \cr
 \code{\link{op_conj}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -112,6 +117,7 @@ Other numpy ops: \cr
 \code{\link{op_ctc_decode}()} \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_diag}()} \cr
 \code{\link{op_diagflat}()} \cr
 \code{\link{op_diagonal}()} \cr
@@ -136,6 +142,9 @@ Other numpy ops: \cr
 \code{\link{op_get_item}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -145,6 +154,7 @@ Other numpy ops: \cr
 \code{\link{op_isfinite}()} \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
+\code{\link{op_kaiser}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
 \code{\link{op_less_equal}()} \cr
@@ -225,6 +235,8 @@ Other numpy ops: \cr
 \code{\link{op_var}()} \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_zeros}()} \cr
@@ -233,6 +245,7 @@ Other numpy ops: \cr
 Other ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -251,6 +264,7 @@ Other ops: \cr
 \code{\link{op_associative_scan}()} \cr
 \code{\link{op_average}()} \cr
 \code{\link{op_average_pool}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_batch_normalization}()} \cr
 \code{\link{op_binary_crossentropy}()} \cr
 \code{\link{op_bincount}()} \cr
@@ -261,9 +275,11 @@ Other ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
 \code{\link{op_cast}()} \cr
 \code{\link{op_categorical_crossentropy}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_celu}()} \cr
 \code{\link{op_cholesky}()} \cr
@@ -276,6 +292,7 @@ Other ops: \cr
 \code{\link{op_convert_to_numpy}()} \cr
 \code{\link{op_convert_to_tensor}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -286,6 +303,7 @@ Other ops: \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
 \code{\link{op_custom_gradient}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_depthwise_conv}()} \cr
 \code{\link{op_det}()} \cr
 \code{\link{op_diag}()} \cr
@@ -325,10 +343,13 @@ Other ops: \cr
 \code{\link{op_glu}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
 \code{\link{op_hard_shrink}()} \cr
 \code{\link{op_hard_sigmoid}()} \cr
 \code{\link{op_hard_silu}()} \cr
 \code{\link{op_hard_tanh}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -355,6 +376,8 @@ Other ops: \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
 \code{\link{op_istft}()} \cr
+\code{\link{op_kaiser}()} \cr
+\code{\link{op_layer_normalization}()} \cr
 \code{\link{op_leaky_relu}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
@@ -454,6 +477,7 @@ Other ops: \cr
 \code{\link{op_sort}()} \cr
 \code{\link{op_sparse_categorical_crossentropy}()} \cr
 \code{\link{op_sparse_plus}()} \cr
+\code{\link{op_sparse_sigmoid}()} \cr
 \code{\link{op_sparsemax}()} \cr
 \code{\link{op_split}()} \cr
 \code{\link{op_sqrt}()} \cr
@@ -491,6 +515,8 @@ Other ops: \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
 \code{\link{op_vectorized_map}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_while_loop}()} \cr
diff --git a/man/op_all.Rd b/man/op_all.Rd
index de3cd1d44..6a6ece307 100644
--- a/man/op_all.Rd
+++ b/man/op_all.Rd
@@ -69,6 +69,7 @@ op_all(x)
 Other numpy ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -85,6 +86,7 @@ Other numpy ops: \cr
 \code{\link{op_argsort}()} \cr
 \code{\link{op_array}()} \cr
 \code{\link{op_average}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_bincount}()} \cr
 \code{\link{op_bitwise_and}()} \cr
 \code{\link{op_bitwise_invert}()} \cr
@@ -93,12 +95,15 @@ Other numpy ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_clip}()} \cr
 \code{\link{op_concatenate}()} \cr
 \code{\link{op_conj}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -107,6 +112,7 @@ Other numpy ops: \cr
 \code{\link{op_ctc_decode}()} \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_diag}()} \cr
 \code{\link{op_diagflat}()} \cr
 \code{\link{op_diagonal}()} \cr
@@ -131,6 +137,9 @@ Other numpy ops: \cr
 \code{\link{op_get_item}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -140,6 +149,7 @@ Other numpy ops: \cr
 \code{\link{op_isfinite}()} \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
+\code{\link{op_kaiser}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
 \code{\link{op_less_equal}()} \cr
@@ -220,6 +230,8 @@ Other numpy ops: \cr
 \code{\link{op_var}()} \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_zeros}()} \cr
@@ -228,6 +240,7 @@ Other numpy ops: \cr
 Other ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -246,6 +259,7 @@ Other ops: \cr
 \code{\link{op_associative_scan}()} \cr
 \code{\link{op_average}()} \cr
 \code{\link{op_average_pool}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_batch_normalization}()} \cr
 \code{\link{op_binary_crossentropy}()} \cr
 \code{\link{op_bincount}()} \cr
@@ -256,9 +270,11 @@ Other ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
 \code{\link{op_cast}()} \cr
 \code{\link{op_categorical_crossentropy}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_celu}()} \cr
 \code{\link{op_cholesky}()} \cr
@@ -271,6 +287,7 @@ Other ops: \cr
 \code{\link{op_convert_to_numpy}()} \cr
 \code{\link{op_convert_to_tensor}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -281,6 +298,7 @@ Other ops: \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
 \code{\link{op_custom_gradient}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_depthwise_conv}()} \cr
 \code{\link{op_det}()} \cr
 \code{\link{op_diag}()} \cr
@@ -320,10 +338,13 @@ Other ops: \cr
 \code{\link{op_glu}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
 \code{\link{op_hard_shrink}()} \cr
 \code{\link{op_hard_sigmoid}()} \cr
 \code{\link{op_hard_silu}()} \cr
 \code{\link{op_hard_tanh}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -350,6 +371,8 @@ Other ops: \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
 \code{\link{op_istft}()} \cr
+\code{\link{op_kaiser}()} \cr
+\code{\link{op_layer_normalization}()} \cr
 \code{\link{op_leaky_relu}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
@@ -449,6 +472,7 @@ Other ops: \cr
 \code{\link{op_sort}()} \cr
 \code{\link{op_sparse_categorical_crossentropy}()} \cr
 \code{\link{op_sparse_plus}()} \cr
+\code{\link{op_sparse_sigmoid}()} \cr
 \code{\link{op_sparsemax}()} \cr
 \code{\link{op_split}()} \cr
 \code{\link{op_sqrt}()} \cr
@@ -486,6 +510,8 @@ Other ops: \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
 \code{\link{op_vectorized_map}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_while_loop}()} \cr
diff --git a/man/op_angle.Rd b/man/op_angle.Rd
new file mode 100644
index 000000000..a0e4d2c33
--- /dev/null
+++ b/man/op_angle.Rd
@@ -0,0 +1,484 @@
+% Generated by roxygen2: do not edit by hand
+% Please edit documentation in R/ops.R
+\name{op_angle}
+\alias{op_angle}
+\title{Element-wise angle of a complex tensor.}
+\usage{
+op_angle(x)
+}
+\arguments{
+\item{x}{Input tensor. Can be real or complex.}
+}
+\value{
+Tensor with the same shape as \code{x}, containing element-wise angles.
+}
+\description{
+Returns the phase angle (in radians) of each element in \code{x}.
+}
+\section{Examples}{
+\if{html}{\out{<div class="sourceCode r">}}\preformatted{x <- op_convert_to_tensor(matrix(c(1 + 3i, 2 - 5i, 4 - 3i, 3 + 2i), nrow = 2))
+op_angle(x)
+}\if{html}{\out{</div>}}
+
+\if{html}{\out{<div class="sourceCode">}}\preformatted{## tf.Tensor(
+## [[ 1.24904577 -0.64350111]
+##  [-1.19028995  0.5880026 ]], shape=(2, 2), dtype=float64)
+
+}\if{html}{\out{</div>}}
+}
+
+\seealso{
+Other numpy ops: \cr
+\code{\link{op_abs}()} \cr
+\code{\link{op_add}()} \cr
+\code{\link{op_all}()} \cr
+\code{\link{op_any}()} \cr
+\code{\link{op_append}()} \cr
+\code{\link{op_arange}()} \cr
+\code{\link{op_arccos}()} \cr
+\code{\link{op_arccosh}()} \cr
+\code{\link{op_arcsin}()} \cr
+\code{\link{op_arcsinh}()} \cr
+\code{\link{op_arctan}()} \cr
+\code{\link{op_arctan2}()} \cr
+\code{\link{op_arctanh}()} \cr
+\code{\link{op_argmax}()} \cr
+\code{\link{op_argmin}()} \cr
+\code{\link{op_argpartition}()} \cr
+\code{\link{op_argsort}()} \cr
+\code{\link{op_array}()} \cr
+\code{\link{op_average}()} \cr
+\code{\link{op_bartlett}()} \cr
+\code{\link{op_bincount}()} \cr
+\code{\link{op_bitwise_and}()} \cr
+\code{\link{op_bitwise_invert}()} \cr
+\code{\link{op_bitwise_left_shift}()} \cr
+\code{\link{op_bitwise_not}()} \cr
+\code{\link{op_bitwise_or}()} \cr
+\code{\link{op_bitwise_right_shift}()} \cr
+\code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
+\code{\link{op_broadcast_to}()} \cr
+\code{\link{op_cbrt}()} \cr
+\code{\link{op_ceil}()} \cr
+\code{\link{op_clip}()} \cr
+\code{\link{op_concatenate}()} \cr
+\code{\link{op_conj}()} \cr
+\code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
+\code{\link{op_correlate}()} \cr
+\code{\link{op_cos}()} \cr
+\code{\link{op_cosh}()} \cr
+\code{\link{op_count_nonzero}()} \cr
+\code{\link{op_cross}()} \cr
+\code{\link{op_ctc_decode}()} \cr
+\code{\link{op_cumprod}()} \cr
+\code{\link{op_cumsum}()} \cr
+\code{\link{op_deg2rad}()} \cr
+\code{\link{op_diag}()} \cr
+\code{\link{op_diagflat}()} \cr
+\code{\link{op_diagonal}()} \cr
+\code{\link{op_diff}()} \cr
+\code{\link{op_digitize}()} \cr
+\code{\link{op_divide}()} \cr
+\code{\link{op_divide_no_nan}()} \cr
+\code{\link{op_dot}()} \cr
+\code{\link{op_einsum}()} \cr
+\code{\link{op_empty}()} \cr
+\code{\link{op_equal}()} \cr
+\code{\link{op_exp}()} \cr
+\code{\link{op_exp2}()} \cr
+\code{\link{op_expand_dims}()} \cr
+\code{\link{op_expm1}()} \cr
+\code{\link{op_eye}()} \cr
+\code{\link{op_flip}()} \cr
+\code{\link{op_floor}()} \cr
+\code{\link{op_floor_divide}()} \cr
+\code{\link{op_full}()} \cr
+\code{\link{op_full_like}()} \cr
+\code{\link{op_get_item}()} \cr
+\code{\link{op_greater}()} \cr
+\code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
+\code{\link{op_heaviside}()} \cr
+\code{\link{op_histogram}()} \cr
+\code{\link{op_hstack}()} \cr
+\code{\link{op_identity}()} \cr
+\code{\link{op_imag}()} \cr
+\code{\link{op_inner}()} \cr
+\code{\link{op_isclose}()} \cr
+\code{\link{op_isfinite}()} \cr
+\code{\link{op_isinf}()} \cr
+\code{\link{op_isnan}()} \cr
+\code{\link{op_kaiser}()} \cr
+\code{\link{op_left_shift}()} \cr
+\code{\link{op_less}()} \cr
+\code{\link{op_less_equal}()} \cr
+\code{\link{op_linspace}()} \cr
+\code{\link{op_log}()} \cr
+\code{\link{op_log10}()} \cr
+\code{\link{op_log1p}()} \cr
+\code{\link{op_log2}()} \cr
+\code{\link{op_logaddexp}()} \cr
+\code{\link{op_logdet}()} \cr
+\code{\link{op_logical_and}()} \cr
+\code{\link{op_logical_not}()} \cr
+\code{\link{op_logical_or}()} \cr
+\code{\link{op_logical_xor}()} \cr
+\code{\link{op_logspace}()} \cr
+\code{\link{op_lstsq}()} \cr
+\code{\link{op_matmul}()} \cr
+\code{\link{op_max}()} \cr
+\code{\link{op_maximum}()} \cr
+\code{\link{op_mean}()} \cr
+\code{\link{op_median}()} \cr
+\code{\link{op_meshgrid}()} \cr
+\code{\link{op_min}()} \cr
+\code{\link{op_minimum}()} \cr
+\code{\link{op_mod}()} \cr
+\code{\link{op_moveaxis}()} \cr
+\code{\link{op_multiply}()} \cr
+\code{\link{op_nan_to_num}()} \cr
+\code{\link{op_ndim}()} \cr
+\code{\link{op_negative}()} \cr
+\code{\link{op_nonzero}()} \cr
+\code{\link{op_not_equal}()} \cr
+\code{\link{op_ones}()} \cr
+\code{\link{op_ones_like}()} \cr
+\code{\link{op_outer}()} \cr
+\code{\link{op_pad}()} \cr
+\code{\link{op_power}()} \cr
+\code{\link{op_prod}()} \cr
+\code{\link{op_quantile}()} \cr
+\code{\link{op_ravel}()} \cr
+\code{\link{op_real}()} \cr
+\code{\link{op_reciprocal}()} \cr
+\code{\link{op_repeat}()} \cr
+\code{\link{op_reshape}()} \cr
+\code{\link{op_right_shift}()} \cr
+\code{\link{op_roll}()} \cr
+\code{\link{op_rot90}()} \cr
+\code{\link{op_round}()} \cr
+\code{\link{op_saturate_cast}()} \cr
+\code{\link{op_select}()} \cr
+\code{\link{op_sign}()} \cr
+\code{\link{op_signbit}()} \cr
+\code{\link{op_sin}()} \cr
+\code{\link{op_sinh}()} \cr
+\code{\link{op_size}()} \cr
+\code{\link{op_sort}()} \cr
+\code{\link{op_split}()} \cr
+\code{\link{op_sqrt}()} \cr
+\code{\link{op_square}()} \cr
+\code{\link{op_squeeze}()} \cr
+\code{\link{op_stack}()} \cr
+\code{\link{op_std}()} \cr
+\code{\link{op_subtract}()} \cr
+\code{\link{op_sum}()} \cr
+\code{\link{op_swapaxes}()} \cr
+\code{\link{op_take}()} \cr
+\code{\link{op_take_along_axis}()} \cr
+\code{\link{op_tan}()} \cr
+\code{\link{op_tanh}()} \cr
+\code{\link{op_tensordot}()} \cr
+\code{\link{op_tile}()} \cr
+\code{\link{op_trace}()} \cr
+\code{\link{op_transpose}()} \cr
+\code{\link{op_tri}()} \cr
+\code{\link{op_tril}()} \cr
+\code{\link{op_triu}()} \cr
+\code{\link{op_trunc}()} \cr
+\code{\link{op_var}()} \cr
+\code{\link{op_vdot}()} \cr
+\code{\link{op_vectorize}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
+\code{\link{op_vstack}()} \cr
+\code{\link{op_where}()} \cr
+\code{\link{op_zeros}()} \cr
+\code{\link{op_zeros_like}()} \cr
+
+Other ops: \cr
+\code{\link{op_abs}()} \cr
+\code{\link{op_add}()} \cr
+\code{\link{op_all}()} \cr
+\code{\link{op_any}()} \cr
+\code{\link{op_append}()} \cr
+\code{\link{op_arange}()} \cr
+\code{\link{op_arccos}()} \cr
+\code{\link{op_arccosh}()} \cr
+\code{\link{op_arcsin}()} \cr
+\code{\link{op_arcsinh}()} \cr
+\code{\link{op_arctan}()} \cr
+\code{\link{op_arctan2}()} \cr
+\code{\link{op_arctanh}()} \cr
+\code{\link{op_argmax}()} \cr
+\code{\link{op_argmin}()} \cr
+\code{\link{op_argpartition}()} \cr
+\code{\link{op_argsort}()} \cr
+\code{\link{op_array}()} \cr
+\code{\link{op_associative_scan}()} \cr
+\code{\link{op_average}()} \cr
+\code{\link{op_average_pool}()} \cr
+\code{\link{op_bartlett}()} \cr
+\code{\link{op_batch_normalization}()} \cr
+\code{\link{op_binary_crossentropy}()} \cr
+\code{\link{op_bincount}()} \cr
+\code{\link{op_bitwise_and}()} \cr
+\code{\link{op_bitwise_invert}()} \cr
+\code{\link{op_bitwise_left_shift}()} \cr
+\code{\link{op_bitwise_not}()} \cr
+\code{\link{op_bitwise_or}()} \cr
+\code{\link{op_bitwise_right_shift}()} \cr
+\code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
+\code{\link{op_broadcast_to}()} \cr
+\code{\link{op_cast}()} \cr
+\code{\link{op_categorical_crossentropy}()} \cr
+\code{\link{op_cbrt}()} \cr
+\code{\link{op_ceil}()} \cr
+\code{\link{op_celu}()} \cr
+\code{\link{op_cholesky}()} \cr
+\code{\link{op_clip}()} \cr
+\code{\link{op_concatenate}()} \cr
+\code{\link{op_cond}()} \cr
+\code{\link{op_conj}()} \cr
+\code{\link{op_conv}()} \cr
+\code{\link{op_conv_transpose}()} \cr
+\code{\link{op_convert_to_numpy}()} \cr
+\code{\link{op_convert_to_tensor}()} \cr
+\code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
+\code{\link{op_correlate}()} \cr
+\code{\link{op_cos}()} \cr
+\code{\link{op_cosh}()} \cr
+\code{\link{op_count_nonzero}()} \cr
+\code{\link{op_cross}()} \cr
+\code{\link{op_ctc_decode}()} \cr
+\code{\link{op_ctc_loss}()} \cr
+\code{\link{op_cumprod}()} \cr
+\code{\link{op_cumsum}()} \cr
+\code{\link{op_custom_gradient}()} \cr
+\code{\link{op_deg2rad}()} \cr
+\code{\link{op_depthwise_conv}()} \cr
+\code{\link{op_det}()} \cr
+\code{\link{op_diag}()} \cr
+\code{\link{op_diagflat}()} \cr
+\code{\link{op_diagonal}()} \cr
+\code{\link{op_diff}()} \cr
+\code{\link{op_digitize}()} \cr
+\code{\link{op_divide}()} \cr
+\code{\link{op_divide_no_nan}()} \cr
+\code{\link{op_dot}()} \cr
+\code{\link{op_dot_product_attention}()} \cr
+\code{\link{op_dtype}()} \cr
+\code{\link{op_eig}()} \cr
+\code{\link{op_eigh}()} \cr
+\code{\link{op_einsum}()} \cr
+\code{\link{op_elu}()} \cr
+\code{\link{op_empty}()} \cr
+\code{\link{op_equal}()} \cr
+\code{\link{op_erf}()} \cr
+\code{\link{op_erfinv}()} \cr
+\code{\link{op_exp}()} \cr
+\code{\link{op_exp2}()} \cr
+\code{\link{op_expand_dims}()} \cr
+\code{\link{op_expm1}()} \cr
+\code{\link{op_extract_sequences}()} \cr
+\code{\link{op_eye}()} \cr
+\code{\link{op_fft}()} \cr
+\code{\link{op_fft2}()} \cr
+\code{\link{op_flip}()} \cr
+\code{\link{op_floor}()} \cr
+\code{\link{op_floor_divide}()} \cr
+\code{\link{op_fori_loop}()} \cr
+\code{\link{op_full}()} \cr
+\code{\link{op_full_like}()} \cr
+\code{\link{op_gelu}()} \cr
+\code{\link{op_get_item}()} \cr
+\code{\link{op_glu}()} \cr
+\code{\link{op_greater}()} \cr
+\code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
+\code{\link{op_hard_shrink}()} \cr
+\code{\link{op_hard_sigmoid}()} \cr
+\code{\link{op_hard_silu}()} \cr
+\code{\link{op_hard_tanh}()} \cr
+\code{\link{op_heaviside}()} \cr
+\code{\link{op_histogram}()} \cr
+\code{\link{op_hstack}()} \cr
+\code{\link{op_identity}()} \cr
+\code{\link{op_ifft2}()} \cr
+\code{\link{op_imag}()} \cr
+\code{\link{op_image_affine_transform}()} \cr
+\code{\link{op_image_crop}()} \cr
+\code{\link{op_image_extract_patches}()} \cr
+\code{\link{op_image_gaussian_blur}()} \cr
+\code{\link{op_image_hsv_to_rgb}()} \cr
+\code{\link{op_image_map_coordinates}()} \cr
+\code{\link{op_image_pad}()} \cr
+\code{\link{op_image_perspective_transform}()} \cr
+\code{\link{op_image_resize}()} \cr
+\code{\link{op_image_rgb_to_grayscale}()} \cr
+\code{\link{op_image_rgb_to_hsv}()} \cr
+\code{\link{op_in_top_k}()} \cr
+\code{\link{op_inner}()} \cr
+\code{\link{op_inv}()} \cr
+\code{\link{op_irfft}()} \cr
+\code{\link{op_is_tensor}()} \cr
+\code{\link{op_isclose}()} \cr
+\code{\link{op_isfinite}()} \cr
+\code{\link{op_isinf}()} \cr
+\code{\link{op_isnan}()} \cr
+\code{\link{op_istft}()} \cr
+\code{\link{op_kaiser}()} \cr
+\code{\link{op_layer_normalization}()} \cr
+\code{\link{op_leaky_relu}()} \cr
+\code{\link{op_left_shift}()} \cr
+\code{\link{op_less}()} \cr
+\code{\link{op_less_equal}()} \cr
+\code{\link{op_linspace}()} \cr
+\code{\link{op_log}()} \cr
+\code{\link{op_log10}()} \cr
+\code{\link{op_log1p}()} \cr
+\code{\link{op_log2}()} \cr
+\code{\link{op_log_sigmoid}()} \cr
+\code{\link{op_log_softmax}()} \cr
+\code{\link{op_logaddexp}()} \cr
+\code{\link{op_logdet}()} \cr
+\code{\link{op_logical_and}()} \cr
+\code{\link{op_logical_not}()} \cr
+\code{\link{op_logical_or}()} \cr
+\code{\link{op_logical_xor}()} \cr
+\code{\link{op_logspace}()} \cr
+\code{\link{op_logsumexp}()} \cr
+\code{\link{op_lstsq}()} \cr
+\code{\link{op_lu_factor}()} \cr
+\code{\link{op_map}()} \cr
+\code{\link{op_matmul}()} \cr
+\code{\link{op_max}()} \cr
+\code{\link{op_max_pool}()} \cr
+\code{\link{op_maximum}()} \cr
+\code{\link{op_mean}()} \cr
+\code{\link{op_median}()} \cr
+\code{\link{op_meshgrid}()} \cr
+\code{\link{op_min}()} \cr
+\code{\link{op_minimum}()} \cr
+\code{\link{op_mod}()} \cr
+\code{\link{op_moments}()} \cr
+\code{\link{op_moveaxis}()} \cr
+\code{\link{op_multi_hot}()} \cr
+\code{\link{op_multiply}()} \cr
+\code{\link{op_nan_to_num}()} \cr
+\code{\link{op_ndim}()} \cr
+\code{\link{op_negative}()} \cr
+\code{\link{op_nonzero}()} \cr
+\code{\link{op_norm}()} \cr
+\code{\link{op_normalize}()} \cr
+\code{\link{op_not_equal}()} \cr
+\code{\link{op_one_hot}()} \cr
+\code{\link{op_ones}()} \cr
+\code{\link{op_ones_like}()} \cr
+\code{\link{op_outer}()} \cr
+\code{\link{op_pad}()} \cr
+\code{\link{op_polar}()} \cr
+\code{\link{op_power}()} \cr
+\code{\link{op_prod}()} \cr
+\code{\link{op_psnr}()} \cr
+\code{\link{op_qr}()} \cr
+\code{\link{op_quantile}()} \cr
+\code{\link{op_ravel}()} \cr
+\code{\link{op_real}()} \cr
+\code{\link{op_rearrange}()} \cr
+\code{\link{op_reciprocal}()} \cr
+\code{\link{op_relu}()} \cr
+\code{\link{op_relu6}()} \cr
+\code{\link{op_repeat}()} \cr
+\code{\link{op_reshape}()} \cr
+\code{\link{op_rfft}()} \cr
+\code{\link{op_right_shift}()} \cr
+\code{\link{op_rms_normalization}()} \cr
+\code{\link{op_roll}()} \cr
+\code{\link{op_rot90}()} \cr
+\code{\link{op_round}()} \cr
+\code{\link{op_rsqrt}()} \cr
+\code{\link{op_saturate_cast}()} \cr
+\code{\link{op_scan}()} \cr
+\code{\link{op_scatter}()} \cr
+\code{\link{op_scatter_update}()} \cr
+\code{\link{op_searchsorted}()} \cr
+\code{\link{op_segment_max}()} \cr
+\code{\link{op_segment_sum}()} \cr
+\code{\link{op_select}()} \cr
+\code{\link{op_selu}()} \cr
+\code{\link{op_separable_conv}()} \cr
+\code{\link{op_shape}()} \cr
+\code{\link{op_sigmoid}()} \cr
+\code{\link{op_sign}()} \cr
+\code{\link{op_signbit}()} \cr
+\code{\link{op_silu}()} \cr
+\code{\link{op_sin}()} \cr
+\code{\link{op_sinh}()} \cr
+\code{\link{op_size}()} \cr
+\code{\link{op_slice}()} \cr
+\code{\link{op_slice_update}()} \cr
+\code{\link{op_slogdet}()} \cr
+\code{\link{op_soft_shrink}()} \cr
+\code{\link{op_softmax}()} \cr
+\code{\link{op_softplus}()} \cr
+\code{\link{op_softsign}()} \cr
+\code{\link{op_solve}()} \cr
+\code{\link{op_solve_triangular}()} \cr
+\code{\link{op_sort}()} \cr
+\code{\link{op_sparse_categorical_crossentropy}()} \cr
+\code{\link{op_sparse_plus}()} \cr
+\code{\link{op_sparse_sigmoid}()} \cr
+\code{\link{op_sparsemax}()} \cr
+\code{\link{op_split}()} \cr
+\code{\link{op_sqrt}()} \cr
+\code{\link{op_square}()} \cr
+\code{\link{op_squareplus}()} \cr
+\code{\link{op_squeeze}()} \cr
+\code{\link{op_stack}()} \cr
+\code{\link{op_std}()} \cr
+\code{\link{op_stft}()} \cr
+\code{\link{op_stop_gradient}()} \cr
+\code{\link{op_subset}()} \cr
+\code{\link{op_subtract}()} \cr
+\code{\link{op_sum}()} \cr
+\code{\link{op_svd}()} \cr
+\code{\link{op_swapaxes}()} \cr
+\code{\link{op_switch}()} \cr
+\code{\link{op_take}()} \cr
+\code{\link{op_take_along_axis}()} \cr
+\code{\link{op_tan}()} \cr
+\code{\link{op_tanh}()} \cr
+\code{\link{op_tanh_shrink}()} \cr
+\code{\link{op_tensordot}()} \cr
+\code{\link{op_threshold}()} \cr
+\code{\link{op_tile}()} \cr
+\code{\link{op_top_k}()} \cr
+\code{\link{op_trace}()} \cr
+\code{\link{op_transpose}()} \cr
+\code{\link{op_tri}()} \cr
+\code{\link{op_tril}()} \cr
+\code{\link{op_triu}()} \cr
+\code{\link{op_trunc}()} \cr
+\code{\link{op_unravel_index}()} \cr
+\code{\link{op_unstack}()} \cr
+\code{\link{op_var}()} \cr
+\code{\link{op_vdot}()} \cr
+\code{\link{op_vectorize}()} \cr
+\code{\link{op_vectorized_map}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
+\code{\link{op_vstack}()} \cr
+\code{\link{op_where}()} \cr
+\code{\link{op_while_loop}()} \cr
+\code{\link{op_zeros}()} \cr
+\code{\link{op_zeros_like}()} \cr
+}
+\concept{numpy ops}
+\concept{ops}
diff --git a/man/op_any.Rd b/man/op_any.Rd
index b38e36f76..3588dbe7d 100644
--- a/man/op_any.Rd
+++ b/man/op_any.Rd
@@ -92,6 +92,7 @@ Other numpy ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
 \code{\link{op_arccos}()} \cr
@@ -107,6 +108,7 @@ Other numpy ops: \cr
 \code{\link{op_argsort}()} \cr
 \code{\link{op_array}()} \cr
 \code{\link{op_average}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_bincount}()} \cr
 \code{\link{op_bitwise_and}()} \cr
 \code{\link{op_bitwise_invert}()} \cr
@@ -115,12 +117,15 @@ Other numpy ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_clip}()} \cr
 \code{\link{op_concatenate}()} \cr
 \code{\link{op_conj}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -129,6 +134,7 @@ Other numpy ops: \cr
 \code{\link{op_ctc_decode}()} \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_diag}()} \cr
 \code{\link{op_diagflat}()} \cr
 \code{\link{op_diagonal}()} \cr
@@ -153,6 +159,9 @@ Other numpy ops: \cr
 \code{\link{op_get_item}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -162,6 +171,7 @@ Other numpy ops: \cr
 \code{\link{op_isfinite}()} \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
+\code{\link{op_kaiser}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
 \code{\link{op_less_equal}()} \cr
@@ -242,6 +252,8 @@ Other numpy ops: \cr
 \code{\link{op_var}()} \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_zeros}()} \cr
@@ -251,6 +263,7 @@ Other ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
 \code{\link{op_arccos}()} \cr
@@ -268,6 +281,7 @@ Other ops: \cr
 \code{\link{op_associative_scan}()} \cr
 \code{\link{op_average}()} \cr
 \code{\link{op_average_pool}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_batch_normalization}()} \cr
 \code{\link{op_binary_crossentropy}()} \cr
 \code{\link{op_bincount}()} \cr
@@ -278,9 +292,11 @@ Other ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
 \code{\link{op_cast}()} \cr
 \code{\link{op_categorical_crossentropy}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_celu}()} \cr
 \code{\link{op_cholesky}()} \cr
@@ -293,6 +309,7 @@ Other ops: \cr
 \code{\link{op_convert_to_numpy}()} \cr
 \code{\link{op_convert_to_tensor}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -303,6 +320,7 @@ Other ops: \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
 \code{\link{op_custom_gradient}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_depthwise_conv}()} \cr
 \code{\link{op_det}()} \cr
 \code{\link{op_diag}()} \cr
@@ -342,10 +360,13 @@ Other ops: \cr
 \code{\link{op_glu}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
 \code{\link{op_hard_shrink}()} \cr
 \code{\link{op_hard_sigmoid}()} \cr
 \code{\link{op_hard_silu}()} \cr
 \code{\link{op_hard_tanh}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -372,6 +393,8 @@ Other ops: \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
 \code{\link{op_istft}()} \cr
+\code{\link{op_kaiser}()} \cr
+\code{\link{op_layer_normalization}()} \cr
 \code{\link{op_leaky_relu}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
@@ -471,6 +494,7 @@ Other ops: \cr
 \code{\link{op_sort}()} \cr
 \code{\link{op_sparse_categorical_crossentropy}()} \cr
 \code{\link{op_sparse_plus}()} \cr
+\code{\link{op_sparse_sigmoid}()} \cr
 \code{\link{op_sparsemax}()} \cr
 \code{\link{op_split}()} \cr
 \code{\link{op_sqrt}()} \cr
@@ -508,6 +532,8 @@ Other ops: \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
 \code{\link{op_vectorized_map}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_while_loop}()} \cr
diff --git a/man/op_append.Rd b/man/op_append.Rd
index 2f4d15ca4..8de5ef750 100644
--- a/man/op_append.Rd
+++ b/man/op_append.Rd
@@ -63,6 +63,7 @@ Other numpy ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_arange}()} \cr
 \code{\link{op_arccos}()} \cr
@@ -78,6 +79,7 @@ Other numpy ops: \cr
 \code{\link{op_argsort}()} \cr
 \code{\link{op_array}()} \cr
 \code{\link{op_average}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_bincount}()} \cr
 \code{\link{op_bitwise_and}()} \cr
 \code{\link{op_bitwise_invert}()} \cr
@@ -86,12 +88,15 @@ Other numpy ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_clip}()} \cr
 \code{\link{op_concatenate}()} \cr
 \code{\link{op_conj}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -100,6 +105,7 @@ Other numpy ops: \cr
 \code{\link{op_ctc_decode}()} \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_diag}()} \cr
 \code{\link{op_diagflat}()} \cr
 \code{\link{op_diagonal}()} \cr
@@ -124,6 +130,9 @@ Other numpy ops: \cr
 \code{\link{op_get_item}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -133,6 +142,7 @@ Other numpy ops: \cr
 \code{\link{op_isfinite}()} \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
+\code{\link{op_kaiser}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
 \code{\link{op_less_equal}()} \cr
@@ -213,6 +223,8 @@ Other numpy ops: \cr
 \code{\link{op_var}()} \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_zeros}()} \cr
@@ -222,6 +234,7 @@ Other ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_arange}()} \cr
 \code{\link{op_arccos}()} \cr
@@ -239,6 +252,7 @@ Other ops: \cr
 \code{\link{op_associative_scan}()} \cr
 \code{\link{op_average}()} \cr
 \code{\link{op_average_pool}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_batch_normalization}()} \cr
 \code{\link{op_binary_crossentropy}()} \cr
 \code{\link{op_bincount}()} \cr
@@ -249,9 +263,11 @@ Other ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
 \code{\link{op_cast}()} \cr
 \code{\link{op_categorical_crossentropy}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_celu}()} \cr
 \code{\link{op_cholesky}()} \cr
@@ -264,6 +280,7 @@ Other ops: \cr
 \code{\link{op_convert_to_numpy}()} \cr
 \code{\link{op_convert_to_tensor}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -274,6 +291,7 @@ Other ops: \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
 \code{\link{op_custom_gradient}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_depthwise_conv}()} \cr
 \code{\link{op_det}()} \cr
 \code{\link{op_diag}()} \cr
@@ -313,10 +331,13 @@ Other ops: \cr
 \code{\link{op_glu}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
 \code{\link{op_hard_shrink}()} \cr
 \code{\link{op_hard_sigmoid}()} \cr
 \code{\link{op_hard_silu}()} \cr
 \code{\link{op_hard_tanh}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -343,6 +364,8 @@ Other ops: \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
 \code{\link{op_istft}()} \cr
+\code{\link{op_kaiser}()} \cr
+\code{\link{op_layer_normalization}()} \cr
 \code{\link{op_leaky_relu}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
@@ -442,6 +465,7 @@ Other ops: \cr
 \code{\link{op_sort}()} \cr
 \code{\link{op_sparse_categorical_crossentropy}()} \cr
 \code{\link{op_sparse_plus}()} \cr
+\code{\link{op_sparse_sigmoid}()} \cr
 \code{\link{op_sparsemax}()} \cr
 \code{\link{op_split}()} \cr
 \code{\link{op_sqrt}()} \cr
@@ -479,6 +503,8 @@ Other ops: \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
 \code{\link{op_vectorized_map}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_while_loop}()} \cr
diff --git a/man/op_arange.Rd b/man/op_arange.Rd
index 10cfd1be6..6c06b2f2a 100644
--- a/man/op_arange.Rd
+++ b/man/op_arange.Rd
@@ -89,6 +89,7 @@ Other numpy ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arccos}()} \cr
@@ -104,6 +105,7 @@ Other numpy ops: \cr
 \code{\link{op_argsort}()} \cr
 \code{\link{op_array}()} \cr
 \code{\link{op_average}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_bincount}()} \cr
 \code{\link{op_bitwise_and}()} \cr
 \code{\link{op_bitwise_invert}()} \cr
@@ -112,12 +114,15 @@ Other numpy ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_clip}()} \cr
 \code{\link{op_concatenate}()} \cr
 \code{\link{op_conj}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -126,6 +131,7 @@ Other numpy ops: \cr
 \code{\link{op_ctc_decode}()} \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_diag}()} \cr
 \code{\link{op_diagflat}()} \cr
 \code{\link{op_diagonal}()} \cr
@@ -150,6 +156,9 @@ Other numpy ops: \cr
 \code{\link{op_get_item}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -159,6 +168,7 @@ Other numpy ops: \cr
 \code{\link{op_isfinite}()} \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
+\code{\link{op_kaiser}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
 \code{\link{op_less_equal}()} \cr
@@ -239,6 +249,8 @@ Other numpy ops: \cr
 \code{\link{op_var}()} \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_zeros}()} \cr
@@ -248,6 +260,7 @@ Other ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arccos}()} \cr
@@ -265,6 +278,7 @@ Other ops: \cr
 \code{\link{op_associative_scan}()} \cr
 \code{\link{op_average}()} \cr
 \code{\link{op_average_pool}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_batch_normalization}()} \cr
 \code{\link{op_binary_crossentropy}()} \cr
 \code{\link{op_bincount}()} \cr
@@ -275,9 +289,11 @@ Other ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
 \code{\link{op_cast}()} \cr
 \code{\link{op_categorical_crossentropy}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_celu}()} \cr
 \code{\link{op_cholesky}()} \cr
@@ -290,6 +306,7 @@ Other ops: \cr
 \code{\link{op_convert_to_numpy}()} \cr
 \code{\link{op_convert_to_tensor}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -300,6 +317,7 @@ Other ops: \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
 \code{\link{op_custom_gradient}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_depthwise_conv}()} \cr
 \code{\link{op_det}()} \cr
 \code{\link{op_diag}()} \cr
@@ -339,10 +357,13 @@ Other ops: \cr
 \code{\link{op_glu}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
 \code{\link{op_hard_shrink}()} \cr
 \code{\link{op_hard_sigmoid}()} \cr
 \code{\link{op_hard_silu}()} \cr
 \code{\link{op_hard_tanh}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -369,6 +390,8 @@ Other ops: \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
 \code{\link{op_istft}()} \cr
+\code{\link{op_kaiser}()} \cr
+\code{\link{op_layer_normalization}()} \cr
 \code{\link{op_leaky_relu}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
@@ -468,6 +491,7 @@ Other ops: \cr
 \code{\link{op_sort}()} \cr
 \code{\link{op_sparse_categorical_crossentropy}()} \cr
 \code{\link{op_sparse_plus}()} \cr
+\code{\link{op_sparse_sigmoid}()} \cr
 \code{\link{op_sparsemax}()} \cr
 \code{\link{op_split}()} \cr
 \code{\link{op_sqrt}()} \cr
@@ -505,6 +529,8 @@ Other ops: \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
 \code{\link{op_vectorized_map}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_while_loop}()} \cr
diff --git a/man/op_arccos.Rd b/man/op_arccos.Rd
index 4ea079a57..40f9bea71 100644
--- a/man/op_arccos.Rd
+++ b/man/op_arccos.Rd
@@ -35,6 +35,7 @@ Other numpy ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -50,6 +51,7 @@ Other numpy ops: \cr
 \code{\link{op_argsort}()} \cr
 \code{\link{op_array}()} \cr
 \code{\link{op_average}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_bincount}()} \cr
 \code{\link{op_bitwise_and}()} \cr
 \code{\link{op_bitwise_invert}()} \cr
@@ -58,12 +60,15 @@ Other numpy ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_clip}()} \cr
 \code{\link{op_concatenate}()} \cr
 \code{\link{op_conj}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -72,6 +77,7 @@ Other numpy ops: \cr
 \code{\link{op_ctc_decode}()} \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_diag}()} \cr
 \code{\link{op_diagflat}()} \cr
 \code{\link{op_diagonal}()} \cr
@@ -96,6 +102,9 @@ Other numpy ops: \cr
 \code{\link{op_get_item}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -105,6 +114,7 @@ Other numpy ops: \cr
 \code{\link{op_isfinite}()} \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
+\code{\link{op_kaiser}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
 \code{\link{op_less_equal}()} \cr
@@ -185,6 +195,8 @@ Other numpy ops: \cr
 \code{\link{op_var}()} \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_zeros}()} \cr
@@ -194,6 +206,7 @@ Other ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -211,6 +224,7 @@ Other ops: \cr
 \code{\link{op_associative_scan}()} \cr
 \code{\link{op_average}()} \cr
 \code{\link{op_average_pool}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_batch_normalization}()} \cr
 \code{\link{op_binary_crossentropy}()} \cr
 \code{\link{op_bincount}()} \cr
@@ -221,9 +235,11 @@ Other ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
 \code{\link{op_cast}()} \cr
 \code{\link{op_categorical_crossentropy}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_celu}()} \cr
 \code{\link{op_cholesky}()} \cr
@@ -236,6 +252,7 @@ Other ops: \cr
 \code{\link{op_convert_to_numpy}()} \cr
 \code{\link{op_convert_to_tensor}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -246,6 +263,7 @@ Other ops: \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
 \code{\link{op_custom_gradient}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_depthwise_conv}()} \cr
 \code{\link{op_det}()} \cr
 \code{\link{op_diag}()} \cr
@@ -285,10 +303,13 @@ Other ops: \cr
 \code{\link{op_glu}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
 \code{\link{op_hard_shrink}()} \cr
 \code{\link{op_hard_sigmoid}()} \cr
 \code{\link{op_hard_silu}()} \cr
 \code{\link{op_hard_tanh}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -315,6 +336,8 @@ Other ops: \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
 \code{\link{op_istft}()} \cr
+\code{\link{op_kaiser}()} \cr
+\code{\link{op_layer_normalization}()} \cr
 \code{\link{op_leaky_relu}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
@@ -414,6 +437,7 @@ Other ops: \cr
 \code{\link{op_sort}()} \cr
 \code{\link{op_sparse_categorical_crossentropy}()} \cr
 \code{\link{op_sparse_plus}()} \cr
+\code{\link{op_sparse_sigmoid}()} \cr
 \code{\link{op_sparsemax}()} \cr
 \code{\link{op_split}()} \cr
 \code{\link{op_sqrt}()} \cr
@@ -451,6 +475,8 @@ Other ops: \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
 \code{\link{op_vectorized_map}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_while_loop}()} \cr
diff --git a/man/op_arccosh.Rd b/man/op_arccosh.Rd
index c29c14cab..d49de31a8 100644
--- a/man/op_arccosh.Rd
+++ b/man/op_arccosh.Rd
@@ -30,6 +30,7 @@ Other numpy ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -45,6 +46,7 @@ Other numpy ops: \cr
 \code{\link{op_argsort}()} \cr
 \code{\link{op_array}()} \cr
 \code{\link{op_average}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_bincount}()} \cr
 \code{\link{op_bitwise_and}()} \cr
 \code{\link{op_bitwise_invert}()} \cr
@@ -53,12 +55,15 @@ Other numpy ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_clip}()} \cr
 \code{\link{op_concatenate}()} \cr
 \code{\link{op_conj}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -67,6 +72,7 @@ Other numpy ops: \cr
 \code{\link{op_ctc_decode}()} \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_diag}()} \cr
 \code{\link{op_diagflat}()} \cr
 \code{\link{op_diagonal}()} \cr
@@ -91,6 +97,9 @@ Other numpy ops: \cr
 \code{\link{op_get_item}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -100,6 +109,7 @@ Other numpy ops: \cr
 \code{\link{op_isfinite}()} \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
+\code{\link{op_kaiser}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
 \code{\link{op_less_equal}()} \cr
@@ -180,6 +190,8 @@ Other numpy ops: \cr
 \code{\link{op_var}()} \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_zeros}()} \cr
@@ -189,6 +201,7 @@ Other ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -206,6 +219,7 @@ Other ops: \cr
 \code{\link{op_associative_scan}()} \cr
 \code{\link{op_average}()} \cr
 \code{\link{op_average_pool}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_batch_normalization}()} \cr
 \code{\link{op_binary_crossentropy}()} \cr
 \code{\link{op_bincount}()} \cr
@@ -216,9 +230,11 @@ Other ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
 \code{\link{op_cast}()} \cr
 \code{\link{op_categorical_crossentropy}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_celu}()} \cr
 \code{\link{op_cholesky}()} \cr
@@ -231,6 +247,7 @@ Other ops: \cr
 \code{\link{op_convert_to_numpy}()} \cr
 \code{\link{op_convert_to_tensor}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -241,6 +258,7 @@ Other ops: \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
 \code{\link{op_custom_gradient}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_depthwise_conv}()} \cr
 \code{\link{op_det}()} \cr
 \code{\link{op_diag}()} \cr
@@ -280,10 +298,13 @@ Other ops: \cr
 \code{\link{op_glu}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
 \code{\link{op_hard_shrink}()} \cr
 \code{\link{op_hard_sigmoid}()} \cr
 \code{\link{op_hard_silu}()} \cr
 \code{\link{op_hard_tanh}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -310,6 +331,8 @@ Other ops: \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
 \code{\link{op_istft}()} \cr
+\code{\link{op_kaiser}()} \cr
+\code{\link{op_layer_normalization}()} \cr
 \code{\link{op_leaky_relu}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
@@ -409,6 +432,7 @@ Other ops: \cr
 \code{\link{op_sort}()} \cr
 \code{\link{op_sparse_categorical_crossentropy}()} \cr
 \code{\link{op_sparse_plus}()} \cr
+\code{\link{op_sparse_sigmoid}()} \cr
 \code{\link{op_sparsemax}()} \cr
 \code{\link{op_split}()} \cr
 \code{\link{op_sqrt}()} \cr
@@ -446,6 +470,8 @@ Other ops: \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
 \code{\link{op_vectorized_map}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_while_loop}()} \cr
diff --git a/man/op_arcsin.Rd b/man/op_arcsin.Rd
index 4d2ae319b..a7bc57f72 100644
--- a/man/op_arcsin.Rd
+++ b/man/op_arcsin.Rd
@@ -35,6 +35,7 @@ Other numpy ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -50,6 +51,7 @@ Other numpy ops: \cr
 \code{\link{op_argsort}()} \cr
 \code{\link{op_array}()} \cr
 \code{\link{op_average}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_bincount}()} \cr
 \code{\link{op_bitwise_and}()} \cr
 \code{\link{op_bitwise_invert}()} \cr
@@ -58,12 +60,15 @@ Other numpy ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_clip}()} \cr
 \code{\link{op_concatenate}()} \cr
 \code{\link{op_conj}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -72,6 +77,7 @@ Other numpy ops: \cr
 \code{\link{op_ctc_decode}()} \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_diag}()} \cr
 \code{\link{op_diagflat}()} \cr
 \code{\link{op_diagonal}()} \cr
@@ -96,6 +102,9 @@ Other numpy ops: \cr
 \code{\link{op_get_item}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -105,6 +114,7 @@ Other numpy ops: \cr
 \code{\link{op_isfinite}()} \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
+\code{\link{op_kaiser}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
 \code{\link{op_less_equal}()} \cr
@@ -185,6 +195,8 @@ Other numpy ops: \cr
 \code{\link{op_var}()} \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_zeros}()} \cr
@@ -194,6 +206,7 @@ Other ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -211,6 +224,7 @@ Other ops: \cr
 \code{\link{op_associative_scan}()} \cr
 \code{\link{op_average}()} \cr
 \code{\link{op_average_pool}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_batch_normalization}()} \cr
 \code{\link{op_binary_crossentropy}()} \cr
 \code{\link{op_bincount}()} \cr
@@ -221,9 +235,11 @@ Other ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
 \code{\link{op_cast}()} \cr
 \code{\link{op_categorical_crossentropy}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_celu}()} \cr
 \code{\link{op_cholesky}()} \cr
@@ -236,6 +252,7 @@ Other ops: \cr
 \code{\link{op_convert_to_numpy}()} \cr
 \code{\link{op_convert_to_tensor}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -246,6 +263,7 @@ Other ops: \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
 \code{\link{op_custom_gradient}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_depthwise_conv}()} \cr
 \code{\link{op_det}()} \cr
 \code{\link{op_diag}()} \cr
@@ -285,10 +303,13 @@ Other ops: \cr
 \code{\link{op_glu}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
 \code{\link{op_hard_shrink}()} \cr
 \code{\link{op_hard_sigmoid}()} \cr
 \code{\link{op_hard_silu}()} \cr
 \code{\link{op_hard_tanh}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -315,6 +336,8 @@ Other ops: \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
 \code{\link{op_istft}()} \cr
+\code{\link{op_kaiser}()} \cr
+\code{\link{op_layer_normalization}()} \cr
 \code{\link{op_leaky_relu}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
@@ -414,6 +437,7 @@ Other ops: \cr
 \code{\link{op_sort}()} \cr
 \code{\link{op_sparse_categorical_crossentropy}()} \cr
 \code{\link{op_sparse_plus}()} \cr
+\code{\link{op_sparse_sigmoid}()} \cr
 \code{\link{op_sparsemax}()} \cr
 \code{\link{op_split}()} \cr
 \code{\link{op_sqrt}()} \cr
@@ -451,6 +475,8 @@ Other ops: \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
 \code{\link{op_vectorized_map}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_while_loop}()} \cr
diff --git a/man/op_arcsinh.Rd b/man/op_arcsinh.Rd
index eede89acb..c1d0e1ee2 100644
--- a/man/op_arcsinh.Rd
+++ b/man/op_arcsinh.Rd
@@ -34,6 +34,7 @@ Other numpy ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -49,6 +50,7 @@ Other numpy ops: \cr
 \code{\link{op_argsort}()} \cr
 \code{\link{op_array}()} \cr
 \code{\link{op_average}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_bincount}()} \cr
 \code{\link{op_bitwise_and}()} \cr
 \code{\link{op_bitwise_invert}()} \cr
@@ -57,12 +59,15 @@ Other numpy ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_clip}()} \cr
 \code{\link{op_concatenate}()} \cr
 \code{\link{op_conj}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -71,6 +76,7 @@ Other numpy ops: \cr
 \code{\link{op_ctc_decode}()} \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_diag}()} \cr
 \code{\link{op_diagflat}()} \cr
 \code{\link{op_diagonal}()} \cr
@@ -95,6 +101,9 @@ Other numpy ops: \cr
 \code{\link{op_get_item}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -104,6 +113,7 @@ Other numpy ops: \cr
 \code{\link{op_isfinite}()} \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
+\code{\link{op_kaiser}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
 \code{\link{op_less_equal}()} \cr
@@ -184,6 +194,8 @@ Other numpy ops: \cr
 \code{\link{op_var}()} \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_zeros}()} \cr
@@ -193,6 +205,7 @@ Other ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -210,6 +223,7 @@ Other ops: \cr
 \code{\link{op_associative_scan}()} \cr
 \code{\link{op_average}()} \cr
 \code{\link{op_average_pool}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_batch_normalization}()} \cr
 \code{\link{op_binary_crossentropy}()} \cr
 \code{\link{op_bincount}()} \cr
@@ -220,9 +234,11 @@ Other ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
 \code{\link{op_cast}()} \cr
 \code{\link{op_categorical_crossentropy}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_celu}()} \cr
 \code{\link{op_cholesky}()} \cr
@@ -235,6 +251,7 @@ Other ops: \cr
 \code{\link{op_convert_to_numpy}()} \cr
 \code{\link{op_convert_to_tensor}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -245,6 +262,7 @@ Other ops: \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
 \code{\link{op_custom_gradient}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_depthwise_conv}()} \cr
 \code{\link{op_det}()} \cr
 \code{\link{op_diag}()} \cr
@@ -284,10 +302,13 @@ Other ops: \cr
 \code{\link{op_glu}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
 \code{\link{op_hard_shrink}()} \cr
 \code{\link{op_hard_sigmoid}()} \cr
 \code{\link{op_hard_silu}()} \cr
 \code{\link{op_hard_tanh}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -314,6 +335,8 @@ Other ops: \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
 \code{\link{op_istft}()} \cr
+\code{\link{op_kaiser}()} \cr
+\code{\link{op_layer_normalization}()} \cr
 \code{\link{op_leaky_relu}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
@@ -413,6 +436,7 @@ Other ops: \cr
 \code{\link{op_sort}()} \cr
 \code{\link{op_sparse_categorical_crossentropy}()} \cr
 \code{\link{op_sparse_plus}()} \cr
+\code{\link{op_sparse_sigmoid}()} \cr
 \code{\link{op_sparsemax}()} \cr
 \code{\link{op_split}()} \cr
 \code{\link{op_sqrt}()} \cr
@@ -450,6 +474,8 @@ Other ops: \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
 \code{\link{op_vectorized_map}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_while_loop}()} \cr
diff --git a/man/op_arctan.Rd b/man/op_arctan.Rd
index 6f2235df7..8867b253a 100644
--- a/man/op_arctan.Rd
+++ b/man/op_arctan.Rd
@@ -35,6 +35,7 @@ Other numpy ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -50,6 +51,7 @@ Other numpy ops: \cr
 \code{\link{op_argsort}()} \cr
 \code{\link{op_array}()} \cr
 \code{\link{op_average}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_bincount}()} \cr
 \code{\link{op_bitwise_and}()} \cr
 \code{\link{op_bitwise_invert}()} \cr
@@ -58,12 +60,15 @@ Other numpy ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_clip}()} \cr
 \code{\link{op_concatenate}()} \cr
 \code{\link{op_conj}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -72,6 +77,7 @@ Other numpy ops: \cr
 \code{\link{op_ctc_decode}()} \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_diag}()} \cr
 \code{\link{op_diagflat}()} \cr
 \code{\link{op_diagonal}()} \cr
@@ -96,6 +102,9 @@ Other numpy ops: \cr
 \code{\link{op_get_item}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -105,6 +114,7 @@ Other numpy ops: \cr
 \code{\link{op_isfinite}()} \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
+\code{\link{op_kaiser}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
 \code{\link{op_less_equal}()} \cr
@@ -185,6 +195,8 @@ Other numpy ops: \cr
 \code{\link{op_var}()} \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_zeros}()} \cr
@@ -194,6 +206,7 @@ Other ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -211,6 +224,7 @@ Other ops: \cr
 \code{\link{op_associative_scan}()} \cr
 \code{\link{op_average}()} \cr
 \code{\link{op_average_pool}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_batch_normalization}()} \cr
 \code{\link{op_binary_crossentropy}()} \cr
 \code{\link{op_bincount}()} \cr
@@ -221,9 +235,11 @@ Other ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
 \code{\link{op_cast}()} \cr
 \code{\link{op_categorical_crossentropy}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_celu}()} \cr
 \code{\link{op_cholesky}()} \cr
@@ -236,6 +252,7 @@ Other ops: \cr
 \code{\link{op_convert_to_numpy}()} \cr
 \code{\link{op_convert_to_tensor}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -246,6 +263,7 @@ Other ops: \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
 \code{\link{op_custom_gradient}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_depthwise_conv}()} \cr
 \code{\link{op_det}()} \cr
 \code{\link{op_diag}()} \cr
@@ -285,10 +303,13 @@ Other ops: \cr
 \code{\link{op_glu}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
 \code{\link{op_hard_shrink}()} \cr
 \code{\link{op_hard_sigmoid}()} \cr
 \code{\link{op_hard_silu}()} \cr
 \code{\link{op_hard_tanh}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -315,6 +336,8 @@ Other ops: \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
 \code{\link{op_istft}()} \cr
+\code{\link{op_kaiser}()} \cr
+\code{\link{op_layer_normalization}()} \cr
 \code{\link{op_leaky_relu}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
@@ -414,6 +437,7 @@ Other ops: \cr
 \code{\link{op_sort}()} \cr
 \code{\link{op_sparse_categorical_crossentropy}()} \cr
 \code{\link{op_sparse_plus}()} \cr
+\code{\link{op_sparse_sigmoid}()} \cr
 \code{\link{op_sparsemax}()} \cr
 \code{\link{op_split}()} \cr
 \code{\link{op_sqrt}()} \cr
@@ -451,6 +475,8 @@ Other ops: \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
 \code{\link{op_vectorized_map}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_while_loop}()} \cr
diff --git a/man/op_arctan2.Rd b/man/op_arctan2.Rd
index 47e3b0bd8..ac69c1f0a 100644
--- a/man/op_arctan2.Rd
+++ b/man/op_arctan2.Rd
@@ -68,6 +68,7 @@ Other numpy ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -83,6 +84,7 @@ Other numpy ops: \cr
 \code{\link{op_argsort}()} \cr
 \code{\link{op_array}()} \cr
 \code{\link{op_average}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_bincount}()} \cr
 \code{\link{op_bitwise_and}()} \cr
 \code{\link{op_bitwise_invert}()} \cr
@@ -91,12 +93,15 @@ Other numpy ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_clip}()} \cr
 \code{\link{op_concatenate}()} \cr
 \code{\link{op_conj}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -105,6 +110,7 @@ Other numpy ops: \cr
 \code{\link{op_ctc_decode}()} \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_diag}()} \cr
 \code{\link{op_diagflat}()} \cr
 \code{\link{op_diagonal}()} \cr
@@ -129,6 +135,9 @@ Other numpy ops: \cr
 \code{\link{op_get_item}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -138,6 +147,7 @@ Other numpy ops: \cr
 \code{\link{op_isfinite}()} \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
+\code{\link{op_kaiser}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
 \code{\link{op_less_equal}()} \cr
@@ -218,6 +228,8 @@ Other numpy ops: \cr
 \code{\link{op_var}()} \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_zeros}()} \cr
@@ -227,6 +239,7 @@ Other ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -244,6 +257,7 @@ Other ops: \cr
 \code{\link{op_associative_scan}()} \cr
 \code{\link{op_average}()} \cr
 \code{\link{op_average_pool}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_batch_normalization}()} \cr
 \code{\link{op_binary_crossentropy}()} \cr
 \code{\link{op_bincount}()} \cr
@@ -254,9 +268,11 @@ Other ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
 \code{\link{op_cast}()} \cr
 \code{\link{op_categorical_crossentropy}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_celu}()} \cr
 \code{\link{op_cholesky}()} \cr
@@ -269,6 +285,7 @@ Other ops: \cr
 \code{\link{op_convert_to_numpy}()} \cr
 \code{\link{op_convert_to_tensor}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -279,6 +296,7 @@ Other ops: \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
 \code{\link{op_custom_gradient}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_depthwise_conv}()} \cr
 \code{\link{op_det}()} \cr
 \code{\link{op_diag}()} \cr
@@ -318,10 +336,13 @@ Other ops: \cr
 \code{\link{op_glu}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
 \code{\link{op_hard_shrink}()} \cr
 \code{\link{op_hard_sigmoid}()} \cr
 \code{\link{op_hard_silu}()} \cr
 \code{\link{op_hard_tanh}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -348,6 +369,8 @@ Other ops: \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
 \code{\link{op_istft}()} \cr
+\code{\link{op_kaiser}()} \cr
+\code{\link{op_layer_normalization}()} \cr
 \code{\link{op_leaky_relu}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
@@ -447,6 +470,7 @@ Other ops: \cr
 \code{\link{op_sort}()} \cr
 \code{\link{op_sparse_categorical_crossentropy}()} \cr
 \code{\link{op_sparse_plus}()} \cr
+\code{\link{op_sparse_sigmoid}()} \cr
 \code{\link{op_sparsemax}()} \cr
 \code{\link{op_split}()} \cr
 \code{\link{op_sqrt}()} \cr
@@ -484,6 +508,8 @@ Other ops: \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
 \code{\link{op_vectorized_map}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_while_loop}()} \cr
diff --git a/man/op_arctanh.Rd b/man/op_arctanh.Rd
index cc3dc64d6..38b0602c2 100644
--- a/man/op_arctanh.Rd
+++ b/man/op_arctanh.Rd
@@ -24,6 +24,7 @@ Other numpy ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -39,6 +40,7 @@ Other numpy ops: \cr
 \code{\link{op_argsort}()} \cr
 \code{\link{op_array}()} \cr
 \code{\link{op_average}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_bincount}()} \cr
 \code{\link{op_bitwise_and}()} \cr
 \code{\link{op_bitwise_invert}()} \cr
@@ -47,12 +49,15 @@ Other numpy ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_clip}()} \cr
 \code{\link{op_concatenate}()} \cr
 \code{\link{op_conj}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -61,6 +66,7 @@ Other numpy ops: \cr
 \code{\link{op_ctc_decode}()} \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_diag}()} \cr
 \code{\link{op_diagflat}()} \cr
 \code{\link{op_diagonal}()} \cr
@@ -85,6 +91,9 @@ Other numpy ops: \cr
 \code{\link{op_get_item}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -94,6 +103,7 @@ Other numpy ops: \cr
 \code{\link{op_isfinite}()} \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
+\code{\link{op_kaiser}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
 \code{\link{op_less_equal}()} \cr
@@ -174,6 +184,8 @@ Other numpy ops: \cr
 \code{\link{op_var}()} \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_zeros}()} \cr
@@ -183,6 +195,7 @@ Other ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -200,6 +213,7 @@ Other ops: \cr
 \code{\link{op_associative_scan}()} \cr
 \code{\link{op_average}()} \cr
 \code{\link{op_average_pool}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_batch_normalization}()} \cr
 \code{\link{op_binary_crossentropy}()} \cr
 \code{\link{op_bincount}()} \cr
@@ -210,9 +224,11 @@ Other ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
 \code{\link{op_cast}()} \cr
 \code{\link{op_categorical_crossentropy}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_celu}()} \cr
 \code{\link{op_cholesky}()} \cr
@@ -225,6 +241,7 @@ Other ops: \cr
 \code{\link{op_convert_to_numpy}()} \cr
 \code{\link{op_convert_to_tensor}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -235,6 +252,7 @@ Other ops: \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
 \code{\link{op_custom_gradient}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_depthwise_conv}()} \cr
 \code{\link{op_det}()} \cr
 \code{\link{op_diag}()} \cr
@@ -274,10 +292,13 @@ Other ops: \cr
 \code{\link{op_glu}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
 \code{\link{op_hard_shrink}()} \cr
 \code{\link{op_hard_sigmoid}()} \cr
 \code{\link{op_hard_silu}()} \cr
 \code{\link{op_hard_tanh}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -304,6 +325,8 @@ Other ops: \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
 \code{\link{op_istft}()} \cr
+\code{\link{op_kaiser}()} \cr
+\code{\link{op_layer_normalization}()} \cr
 \code{\link{op_leaky_relu}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
@@ -403,6 +426,7 @@ Other ops: \cr
 \code{\link{op_sort}()} \cr
 \code{\link{op_sparse_categorical_crossentropy}()} \cr
 \code{\link{op_sparse_plus}()} \cr
+\code{\link{op_sparse_sigmoid}()} \cr
 \code{\link{op_sparsemax}()} \cr
 \code{\link{op_split}()} \cr
 \code{\link{op_sqrt}()} \cr
@@ -440,6 +464,8 @@ Other ops: \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
 \code{\link{op_vectorized_map}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_while_loop}()} \cr
diff --git a/man/op_argmax.Rd b/man/op_argmax.Rd
index 1bf0bf2ba..957c97625 100644
--- a/man/op_argmax.Rd
+++ b/man/op_argmax.Rd
@@ -74,6 +74,7 @@ Other numpy ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -89,6 +90,7 @@ Other numpy ops: \cr
 \code{\link{op_argsort}()} \cr
 \code{\link{op_array}()} \cr
 \code{\link{op_average}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_bincount}()} \cr
 \code{\link{op_bitwise_and}()} \cr
 \code{\link{op_bitwise_invert}()} \cr
@@ -97,12 +99,15 @@ Other numpy ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_clip}()} \cr
 \code{\link{op_concatenate}()} \cr
 \code{\link{op_conj}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -111,6 +116,7 @@ Other numpy ops: \cr
 \code{\link{op_ctc_decode}()} \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_diag}()} \cr
 \code{\link{op_diagflat}()} \cr
 \code{\link{op_diagonal}()} \cr
@@ -135,6 +141,9 @@ Other numpy ops: \cr
 \code{\link{op_get_item}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -144,6 +153,7 @@ Other numpy ops: \cr
 \code{\link{op_isfinite}()} \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
+\code{\link{op_kaiser}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
 \code{\link{op_less_equal}()} \cr
@@ -224,6 +234,8 @@ Other numpy ops: \cr
 \code{\link{op_var}()} \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_zeros}()} \cr
@@ -233,6 +245,7 @@ Other ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -250,6 +263,7 @@ Other ops: \cr
 \code{\link{op_associative_scan}()} \cr
 \code{\link{op_average}()} \cr
 \code{\link{op_average_pool}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_batch_normalization}()} \cr
 \code{\link{op_binary_crossentropy}()} \cr
 \code{\link{op_bincount}()} \cr
@@ -260,9 +274,11 @@ Other ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
 \code{\link{op_cast}()} \cr
 \code{\link{op_categorical_crossentropy}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_celu}()} \cr
 \code{\link{op_cholesky}()} \cr
@@ -275,6 +291,7 @@ Other ops: \cr
 \code{\link{op_convert_to_numpy}()} \cr
 \code{\link{op_convert_to_tensor}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -285,6 +302,7 @@ Other ops: \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
 \code{\link{op_custom_gradient}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_depthwise_conv}()} \cr
 \code{\link{op_det}()} \cr
 \code{\link{op_diag}()} \cr
@@ -324,10 +342,13 @@ Other ops: \cr
 \code{\link{op_glu}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
 \code{\link{op_hard_shrink}()} \cr
 \code{\link{op_hard_sigmoid}()} \cr
 \code{\link{op_hard_silu}()} \cr
 \code{\link{op_hard_tanh}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -354,6 +375,8 @@ Other ops: \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
 \code{\link{op_istft}()} \cr
+\code{\link{op_kaiser}()} \cr
+\code{\link{op_layer_normalization}()} \cr
 \code{\link{op_leaky_relu}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
@@ -453,6 +476,7 @@ Other ops: \cr
 \code{\link{op_sort}()} \cr
 \code{\link{op_sparse_categorical_crossentropy}()} \cr
 \code{\link{op_sparse_plus}()} \cr
+\code{\link{op_sparse_sigmoid}()} \cr
 \code{\link{op_sparsemax}()} \cr
 \code{\link{op_split}()} \cr
 \code{\link{op_sqrt}()} \cr
@@ -490,6 +514,8 @@ Other ops: \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
 \code{\link{op_vectorized_map}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_while_loop}()} \cr
diff --git a/man/op_argmin.Rd b/man/op_argmin.Rd
index f49123999..bc379ecdc 100644
--- a/man/op_argmin.Rd
+++ b/man/op_argmin.Rd
@@ -72,6 +72,7 @@ Other numpy ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -87,6 +88,7 @@ Other numpy ops: \cr
 \code{\link{op_argsort}()} \cr
 \code{\link{op_array}()} \cr
 \code{\link{op_average}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_bincount}()} \cr
 \code{\link{op_bitwise_and}()} \cr
 \code{\link{op_bitwise_invert}()} \cr
@@ -95,12 +97,15 @@ Other numpy ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_clip}()} \cr
 \code{\link{op_concatenate}()} \cr
 \code{\link{op_conj}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -109,6 +114,7 @@ Other numpy ops: \cr
 \code{\link{op_ctc_decode}()} \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_diag}()} \cr
 \code{\link{op_diagflat}()} \cr
 \code{\link{op_diagonal}()} \cr
@@ -133,6 +139,9 @@ Other numpy ops: \cr
 \code{\link{op_get_item}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -142,6 +151,7 @@ Other numpy ops: \cr
 \code{\link{op_isfinite}()} \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
+\code{\link{op_kaiser}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
 \code{\link{op_less_equal}()} \cr
@@ -222,6 +232,8 @@ Other numpy ops: \cr
 \code{\link{op_var}()} \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_zeros}()} \cr
@@ -231,6 +243,7 @@ Other ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -248,6 +261,7 @@ Other ops: \cr
 \code{\link{op_associative_scan}()} \cr
 \code{\link{op_average}()} \cr
 \code{\link{op_average_pool}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_batch_normalization}()} \cr
 \code{\link{op_binary_crossentropy}()} \cr
 \code{\link{op_bincount}()} \cr
@@ -258,9 +272,11 @@ Other ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
 \code{\link{op_cast}()} \cr
 \code{\link{op_categorical_crossentropy}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_celu}()} \cr
 \code{\link{op_cholesky}()} \cr
@@ -273,6 +289,7 @@ Other ops: \cr
 \code{\link{op_convert_to_numpy}()} \cr
 \code{\link{op_convert_to_tensor}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -283,6 +300,7 @@ Other ops: \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
 \code{\link{op_custom_gradient}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_depthwise_conv}()} \cr
 \code{\link{op_det}()} \cr
 \code{\link{op_diag}()} \cr
@@ -322,10 +340,13 @@ Other ops: \cr
 \code{\link{op_glu}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
 \code{\link{op_hard_shrink}()} \cr
 \code{\link{op_hard_sigmoid}()} \cr
 \code{\link{op_hard_silu}()} \cr
 \code{\link{op_hard_tanh}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -352,6 +373,8 @@ Other ops: \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
 \code{\link{op_istft}()} \cr
+\code{\link{op_kaiser}()} \cr
+\code{\link{op_layer_normalization}()} \cr
 \code{\link{op_leaky_relu}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
@@ -451,6 +474,7 @@ Other ops: \cr
 \code{\link{op_sort}()} \cr
 \code{\link{op_sparse_categorical_crossentropy}()} \cr
 \code{\link{op_sparse_plus}()} \cr
+\code{\link{op_sparse_sigmoid}()} \cr
 \code{\link{op_sparsemax}()} \cr
 \code{\link{op_split}()} \cr
 \code{\link{op_sqrt}()} \cr
@@ -488,6 +512,8 @@ Other ops: \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
 \code{\link{op_vectorized_map}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_while_loop}()} \cr
diff --git a/man/op_argpartition.Rd b/man/op_argpartition.Rd
index 87429bc57..8fb9e5b4b 100644
--- a/man/op_argpartition.Rd
+++ b/man/op_argpartition.Rd
@@ -58,6 +58,7 @@ Other numpy ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -73,6 +74,7 @@ Other numpy ops: \cr
 \code{\link{op_argsort}()} \cr
 \code{\link{op_array}()} \cr
 \code{\link{op_average}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_bincount}()} \cr
 \code{\link{op_bitwise_and}()} \cr
 \code{\link{op_bitwise_invert}()} \cr
@@ -81,12 +83,15 @@ Other numpy ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_clip}()} \cr
 \code{\link{op_concatenate}()} \cr
 \code{\link{op_conj}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -95,6 +100,7 @@ Other numpy ops: \cr
 \code{\link{op_ctc_decode}()} \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_diag}()} \cr
 \code{\link{op_diagflat}()} \cr
 \code{\link{op_diagonal}()} \cr
@@ -119,6 +125,9 @@ Other numpy ops: \cr
 \code{\link{op_get_item}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -128,6 +137,7 @@ Other numpy ops: \cr
 \code{\link{op_isfinite}()} \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
+\code{\link{op_kaiser}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
 \code{\link{op_less_equal}()} \cr
@@ -208,6 +218,8 @@ Other numpy ops: \cr
 \code{\link{op_var}()} \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_zeros}()} \cr
@@ -217,6 +229,7 @@ Other ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -234,6 +247,7 @@ Other ops: \cr
 \code{\link{op_associative_scan}()} \cr
 \code{\link{op_average}()} \cr
 \code{\link{op_average_pool}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_batch_normalization}()} \cr
 \code{\link{op_binary_crossentropy}()} \cr
 \code{\link{op_bincount}()} \cr
@@ -244,9 +258,11 @@ Other ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
 \code{\link{op_cast}()} \cr
 \code{\link{op_categorical_crossentropy}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_celu}()} \cr
 \code{\link{op_cholesky}()} \cr
@@ -259,6 +275,7 @@ Other ops: \cr
 \code{\link{op_convert_to_numpy}()} \cr
 \code{\link{op_convert_to_tensor}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -269,6 +286,7 @@ Other ops: \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
 \code{\link{op_custom_gradient}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_depthwise_conv}()} \cr
 \code{\link{op_det}()} \cr
 \code{\link{op_diag}()} \cr
@@ -308,10 +326,13 @@ Other ops: \cr
 \code{\link{op_glu}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
 \code{\link{op_hard_shrink}()} \cr
 \code{\link{op_hard_sigmoid}()} \cr
 \code{\link{op_hard_silu}()} \cr
 \code{\link{op_hard_tanh}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -338,6 +359,8 @@ Other ops: \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
 \code{\link{op_istft}()} \cr
+\code{\link{op_kaiser}()} \cr
+\code{\link{op_layer_normalization}()} \cr
 \code{\link{op_leaky_relu}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
@@ -437,6 +460,7 @@ Other ops: \cr
 \code{\link{op_sort}()} \cr
 \code{\link{op_sparse_categorical_crossentropy}()} \cr
 \code{\link{op_sparse_plus}()} \cr
+\code{\link{op_sparse_sigmoid}()} \cr
 \code{\link{op_sparsemax}()} \cr
 \code{\link{op_split}()} \cr
 \code{\link{op_sqrt}()} \cr
@@ -474,6 +498,8 @@ Other ops: \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
 \code{\link{op_vectorized_map}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_while_loop}()} \cr
diff --git a/man/op_argsort.Rd b/man/op_argsort.Rd
index d04f77aa9..67e4be3a3 100644
--- a/man/op_argsort.Rd
+++ b/man/op_argsort.Rd
@@ -105,6 +105,7 @@ Other numpy ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -120,6 +121,7 @@ Other numpy ops: \cr
 \code{\link{op_argpartition}()} \cr
 \code{\link{op_array}()} \cr
 \code{\link{op_average}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_bincount}()} \cr
 \code{\link{op_bitwise_and}()} \cr
 \code{\link{op_bitwise_invert}()} \cr
@@ -128,12 +130,15 @@ Other numpy ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_clip}()} \cr
 \code{\link{op_concatenate}()} \cr
 \code{\link{op_conj}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -142,6 +147,7 @@ Other numpy ops: \cr
 \code{\link{op_ctc_decode}()} \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_diag}()} \cr
 \code{\link{op_diagflat}()} \cr
 \code{\link{op_diagonal}()} \cr
@@ -166,6 +172,9 @@ Other numpy ops: \cr
 \code{\link{op_get_item}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -175,6 +184,7 @@ Other numpy ops: \cr
 \code{\link{op_isfinite}()} \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
+\code{\link{op_kaiser}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
 \code{\link{op_less_equal}()} \cr
@@ -255,6 +265,8 @@ Other numpy ops: \cr
 \code{\link{op_var}()} \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_zeros}()} \cr
@@ -264,6 +276,7 @@ Other ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -281,6 +294,7 @@ Other ops: \cr
 \code{\link{op_associative_scan}()} \cr
 \code{\link{op_average}()} \cr
 \code{\link{op_average_pool}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_batch_normalization}()} \cr
 \code{\link{op_binary_crossentropy}()} \cr
 \code{\link{op_bincount}()} \cr
@@ -291,9 +305,11 @@ Other ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
 \code{\link{op_cast}()} \cr
 \code{\link{op_categorical_crossentropy}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_celu}()} \cr
 \code{\link{op_cholesky}()} \cr
@@ -306,6 +322,7 @@ Other ops: \cr
 \code{\link{op_convert_to_numpy}()} \cr
 \code{\link{op_convert_to_tensor}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -316,6 +333,7 @@ Other ops: \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
 \code{\link{op_custom_gradient}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_depthwise_conv}()} \cr
 \code{\link{op_det}()} \cr
 \code{\link{op_diag}()} \cr
@@ -355,10 +373,13 @@ Other ops: \cr
 \code{\link{op_glu}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
 \code{\link{op_hard_shrink}()} \cr
 \code{\link{op_hard_sigmoid}()} \cr
 \code{\link{op_hard_silu}()} \cr
 \code{\link{op_hard_tanh}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -385,6 +406,8 @@ Other ops: \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
 \code{\link{op_istft}()} \cr
+\code{\link{op_kaiser}()} \cr
+\code{\link{op_layer_normalization}()} \cr
 \code{\link{op_leaky_relu}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
@@ -484,6 +507,7 @@ Other ops: \cr
 \code{\link{op_sort}()} \cr
 \code{\link{op_sparse_categorical_crossentropy}()} \cr
 \code{\link{op_sparse_plus}()} \cr
+\code{\link{op_sparse_sigmoid}()} \cr
 \code{\link{op_sparsemax}()} \cr
 \code{\link{op_split}()} \cr
 \code{\link{op_sqrt}()} \cr
@@ -521,6 +545,8 @@ Other ops: \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
 \code{\link{op_vectorized_map}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_while_loop}()} \cr
diff --git a/man/op_array.Rd b/man/op_array.Rd
index 07d87d495..4480d14bb 100644
--- a/man/op_array.Rd
+++ b/man/op_array.Rd
@@ -49,6 +49,7 @@ Other numpy ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -64,6 +65,7 @@ Other numpy ops: \cr
 \code{\link{op_argpartition}()} \cr
 \code{\link{op_argsort}()} \cr
 \code{\link{op_average}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_bincount}()} \cr
 \code{\link{op_bitwise_and}()} \cr
 \code{\link{op_bitwise_invert}()} \cr
@@ -72,12 +74,15 @@ Other numpy ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_clip}()} \cr
 \code{\link{op_concatenate}()} \cr
 \code{\link{op_conj}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -86,6 +91,7 @@ Other numpy ops: \cr
 \code{\link{op_ctc_decode}()} \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_diag}()} \cr
 \code{\link{op_diagflat}()} \cr
 \code{\link{op_diagonal}()} \cr
@@ -110,6 +116,9 @@ Other numpy ops: \cr
 \code{\link{op_get_item}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -119,6 +128,7 @@ Other numpy ops: \cr
 \code{\link{op_isfinite}()} \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
+\code{\link{op_kaiser}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
 \code{\link{op_less_equal}()} \cr
@@ -199,6 +209,8 @@ Other numpy ops: \cr
 \code{\link{op_var}()} \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_zeros}()} \cr
@@ -208,6 +220,7 @@ Other ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -225,6 +238,7 @@ Other ops: \cr
 \code{\link{op_associative_scan}()} \cr
 \code{\link{op_average}()} \cr
 \code{\link{op_average_pool}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_batch_normalization}()} \cr
 \code{\link{op_binary_crossentropy}()} \cr
 \code{\link{op_bincount}()} \cr
@@ -235,9 +249,11 @@ Other ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
 \code{\link{op_cast}()} \cr
 \code{\link{op_categorical_crossentropy}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_celu}()} \cr
 \code{\link{op_cholesky}()} \cr
@@ -250,6 +266,7 @@ Other ops: \cr
 \code{\link{op_convert_to_numpy}()} \cr
 \code{\link{op_convert_to_tensor}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -260,6 +277,7 @@ Other ops: \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
 \code{\link{op_custom_gradient}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_depthwise_conv}()} \cr
 \code{\link{op_det}()} \cr
 \code{\link{op_diag}()} \cr
@@ -299,10 +317,13 @@ Other ops: \cr
 \code{\link{op_glu}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
 \code{\link{op_hard_shrink}()} \cr
 \code{\link{op_hard_sigmoid}()} \cr
 \code{\link{op_hard_silu}()} \cr
 \code{\link{op_hard_tanh}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -329,6 +350,8 @@ Other ops: \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
 \code{\link{op_istft}()} \cr
+\code{\link{op_kaiser}()} \cr
+\code{\link{op_layer_normalization}()} \cr
 \code{\link{op_leaky_relu}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
@@ -428,6 +451,7 @@ Other ops: \cr
 \code{\link{op_sort}()} \cr
 \code{\link{op_sparse_categorical_crossentropy}()} \cr
 \code{\link{op_sparse_plus}()} \cr
+\code{\link{op_sparse_sigmoid}()} \cr
 \code{\link{op_sparsemax}()} \cr
 \code{\link{op_split}()} \cr
 \code{\link{op_sqrt}()} \cr
@@ -465,6 +489,8 @@ Other ops: \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
 \code{\link{op_vectorized_map}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_while_loop}()} \cr
diff --git a/man/op_associative_scan.Rd b/man/op_associative_scan.Rd
index 4d2f48820..0b44a3762 100644
--- a/man/op_associative_scan.Rd
+++ b/man/op_associative_scan.Rd
@@ -123,6 +123,7 @@ Other ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -140,6 +141,7 @@ Other ops: \cr
 \code{\link{op_array}()} \cr
 \code{\link{op_average}()} \cr
 \code{\link{op_average_pool}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_batch_normalization}()} \cr
 \code{\link{op_binary_crossentropy}()} \cr
 \code{\link{op_bincount}()} \cr
@@ -150,9 +152,11 @@ Other ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
 \code{\link{op_cast}()} \cr
 \code{\link{op_categorical_crossentropy}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_celu}()} \cr
 \code{\link{op_cholesky}()} \cr
@@ -165,6 +169,7 @@ Other ops: \cr
 \code{\link{op_convert_to_numpy}()} \cr
 \code{\link{op_convert_to_tensor}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -175,6 +180,7 @@ Other ops: \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
 \code{\link{op_custom_gradient}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_depthwise_conv}()} \cr
 \code{\link{op_det}()} \cr
 \code{\link{op_diag}()} \cr
@@ -214,10 +220,13 @@ Other ops: \cr
 \code{\link{op_glu}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
 \code{\link{op_hard_shrink}()} \cr
 \code{\link{op_hard_sigmoid}()} \cr
 \code{\link{op_hard_silu}()} \cr
 \code{\link{op_hard_tanh}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -244,6 +253,8 @@ Other ops: \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
 \code{\link{op_istft}()} \cr
+\code{\link{op_kaiser}()} \cr
+\code{\link{op_layer_normalization}()} \cr
 \code{\link{op_leaky_relu}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
@@ -343,6 +354,7 @@ Other ops: \cr
 \code{\link{op_sort}()} \cr
 \code{\link{op_sparse_categorical_crossentropy}()} \cr
 \code{\link{op_sparse_plus}()} \cr
+\code{\link{op_sparse_sigmoid}()} \cr
 \code{\link{op_sparsemax}()} \cr
 \code{\link{op_split}()} \cr
 \code{\link{op_sqrt}()} \cr
@@ -380,6 +392,8 @@ Other ops: \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
 \code{\link{op_vectorized_map}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_while_loop}()} \cr
diff --git a/man/op_average.Rd b/man/op_average.Rd
index 6bd233a19..805e68db0 100644
--- a/man/op_average.Rd
+++ b/man/op_average.Rd
@@ -99,6 +99,7 @@ Other numpy ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -114,6 +115,7 @@ Other numpy ops: \cr
 \code{\link{op_argpartition}()} \cr
 \code{\link{op_argsort}()} \cr
 \code{\link{op_array}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_bincount}()} \cr
 \code{\link{op_bitwise_and}()} \cr
 \code{\link{op_bitwise_invert}()} \cr
@@ -122,12 +124,15 @@ Other numpy ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_clip}()} \cr
 \code{\link{op_concatenate}()} \cr
 \code{\link{op_conj}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -136,6 +141,7 @@ Other numpy ops: \cr
 \code{\link{op_ctc_decode}()} \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_diag}()} \cr
 \code{\link{op_diagflat}()} \cr
 \code{\link{op_diagonal}()} \cr
@@ -160,6 +166,9 @@ Other numpy ops: \cr
 \code{\link{op_get_item}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -169,6 +178,7 @@ Other numpy ops: \cr
 \code{\link{op_isfinite}()} \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
+\code{\link{op_kaiser}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
 \code{\link{op_less_equal}()} \cr
@@ -249,6 +259,8 @@ Other numpy ops: \cr
 \code{\link{op_var}()} \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_zeros}()} \cr
@@ -258,6 +270,7 @@ Other ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -275,6 +288,7 @@ Other ops: \cr
 \code{\link{op_array}()} \cr
 \code{\link{op_associative_scan}()} \cr
 \code{\link{op_average_pool}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_batch_normalization}()} \cr
 \code{\link{op_binary_crossentropy}()} \cr
 \code{\link{op_bincount}()} \cr
@@ -285,9 +299,11 @@ Other ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
 \code{\link{op_cast}()} \cr
 \code{\link{op_categorical_crossentropy}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_celu}()} \cr
 \code{\link{op_cholesky}()} \cr
@@ -300,6 +316,7 @@ Other ops: \cr
 \code{\link{op_convert_to_numpy}()} \cr
 \code{\link{op_convert_to_tensor}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -310,6 +327,7 @@ Other ops: \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
 \code{\link{op_custom_gradient}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_depthwise_conv}()} \cr
 \code{\link{op_det}()} \cr
 \code{\link{op_diag}()} \cr
@@ -349,10 +367,13 @@ Other ops: \cr
 \code{\link{op_glu}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
 \code{\link{op_hard_shrink}()} \cr
 \code{\link{op_hard_sigmoid}()} \cr
 \code{\link{op_hard_silu}()} \cr
 \code{\link{op_hard_tanh}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -379,6 +400,8 @@ Other ops: \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
 \code{\link{op_istft}()} \cr
+\code{\link{op_kaiser}()} \cr
+\code{\link{op_layer_normalization}()} \cr
 \code{\link{op_leaky_relu}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
@@ -478,6 +501,7 @@ Other ops: \cr
 \code{\link{op_sort}()} \cr
 \code{\link{op_sparse_categorical_crossentropy}()} \cr
 \code{\link{op_sparse_plus}()} \cr
+\code{\link{op_sparse_sigmoid}()} \cr
 \code{\link{op_sparsemax}()} \cr
 \code{\link{op_split}()} \cr
 \code{\link{op_sqrt}()} \cr
@@ -515,6 +539,8 @@ Other ops: \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
 \code{\link{op_vectorized_map}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_while_loop}()} \cr
diff --git a/man/op_average_pool.Rd b/man/op_average_pool.Rd
index fbe9e7a17..f3bf8bc56 100644
--- a/man/op_average_pool.Rd
+++ b/man/op_average_pool.Rd
@@ -71,6 +71,7 @@ Other nn ops: \cr
 \code{\link{op_hard_sigmoid}()} \cr
 \code{\link{op_hard_silu}()} \cr
 \code{\link{op_hard_tanh}()} \cr
+\code{\link{op_layer_normalization}()} \cr
 \code{\link{op_leaky_relu}()} \cr
 \code{\link{op_log_sigmoid}()} \cr
 \code{\link{op_log_softmax}()} \cr
@@ -94,6 +95,7 @@ Other nn ops: \cr
 \code{\link{op_softsign}()} \cr
 \code{\link{op_sparse_categorical_crossentropy}()} \cr
 \code{\link{op_sparse_plus}()} \cr
+\code{\link{op_sparse_sigmoid}()} \cr
 \code{\link{op_sparsemax}()} \cr
 \code{\link{op_squareplus}()} \cr
 \code{\link{op_tanh_shrink}()} \cr
@@ -104,6 +106,7 @@ Other ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -121,6 +124,7 @@ Other ops: \cr
 \code{\link{op_array}()} \cr
 \code{\link{op_associative_scan}()} \cr
 \code{\link{op_average}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_batch_normalization}()} \cr
 \code{\link{op_binary_crossentropy}()} \cr
 \code{\link{op_bincount}()} \cr
@@ -131,9 +135,11 @@ Other ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
 \code{\link{op_cast}()} \cr
 \code{\link{op_categorical_crossentropy}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_celu}()} \cr
 \code{\link{op_cholesky}()} \cr
@@ -146,6 +152,7 @@ Other ops: \cr
 \code{\link{op_convert_to_numpy}()} \cr
 \code{\link{op_convert_to_tensor}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -156,6 +163,7 @@ Other ops: \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
 \code{\link{op_custom_gradient}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_depthwise_conv}()} \cr
 \code{\link{op_det}()} \cr
 \code{\link{op_diag}()} \cr
@@ -195,10 +203,13 @@ Other ops: \cr
 \code{\link{op_glu}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
 \code{\link{op_hard_shrink}()} \cr
 \code{\link{op_hard_sigmoid}()} \cr
 \code{\link{op_hard_silu}()} \cr
 \code{\link{op_hard_tanh}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -225,6 +236,8 @@ Other ops: \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
 \code{\link{op_istft}()} \cr
+\code{\link{op_kaiser}()} \cr
+\code{\link{op_layer_normalization}()} \cr
 \code{\link{op_leaky_relu}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
@@ -324,6 +337,7 @@ Other ops: \cr
 \code{\link{op_sort}()} \cr
 \code{\link{op_sparse_categorical_crossentropy}()} \cr
 \code{\link{op_sparse_plus}()} \cr
+\code{\link{op_sparse_sigmoid}()} \cr
 \code{\link{op_sparsemax}()} \cr
 \code{\link{op_split}()} \cr
 \code{\link{op_sqrt}()} \cr
@@ -361,6 +375,8 @@ Other ops: \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
 \code{\link{op_vectorized_map}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_while_loop}()} \cr
diff --git a/man/op_bartlett.Rd b/man/op_bartlett.Rd
new file mode 100644
index 000000000..19b277fc8
--- /dev/null
+++ b/man/op_bartlett.Rd
@@ -0,0 +1,481 @@
+% Generated by roxygen2: do not edit by hand
+% Please edit documentation in R/ops-numpy.R
+\name{op_bartlett}
+\alias{op_bartlett}
+\title{Bartlett window function.}
+\usage{
+op_bartlett(x)
+}
+\arguments{
+\item{x}{Length of the window. Must be a positive integer.}
+}
+\value{
+A 1D tensor containing the window values.
+}
+\description{
+The Bartlett window is a triangular window that rises then falls linearly.
+}
+\section{Examples}{
+\if{html}{\out{<div class="sourceCode r">}}\preformatted{op_bartlett(5)
+}\if{html}{\out{</div>}}
+
+\if{html}{\out{<div class="sourceCode">}}\preformatted{## tf.Tensor([0.  0.5 1.  0.5 0. ], shape=(5), dtype=float32)
+
+}\if{html}{\out{</div>}}
+}
+
+\seealso{
+Other numpy ops: \cr
+\code{\link{op_abs}()} \cr
+\code{\link{op_add}()} \cr
+\code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
+\code{\link{op_any}()} \cr
+\code{\link{op_append}()} \cr
+\code{\link{op_arange}()} \cr
+\code{\link{op_arccos}()} \cr
+\code{\link{op_arccosh}()} \cr
+\code{\link{op_arcsin}()} \cr
+\code{\link{op_arcsinh}()} \cr
+\code{\link{op_arctan}()} \cr
+\code{\link{op_arctan2}()} \cr
+\code{\link{op_arctanh}()} \cr
+\code{\link{op_argmax}()} \cr
+\code{\link{op_argmin}()} \cr
+\code{\link{op_argpartition}()} \cr
+\code{\link{op_argsort}()} \cr
+\code{\link{op_array}()} \cr
+\code{\link{op_average}()} \cr
+\code{\link{op_bincount}()} \cr
+\code{\link{op_bitwise_and}()} \cr
+\code{\link{op_bitwise_invert}()} \cr
+\code{\link{op_bitwise_left_shift}()} \cr
+\code{\link{op_bitwise_not}()} \cr
+\code{\link{op_bitwise_or}()} \cr
+\code{\link{op_bitwise_right_shift}()} \cr
+\code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
+\code{\link{op_broadcast_to}()} \cr
+\code{\link{op_cbrt}()} \cr
+\code{\link{op_ceil}()} \cr
+\code{\link{op_clip}()} \cr
+\code{\link{op_concatenate}()} \cr
+\code{\link{op_conj}()} \cr
+\code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
+\code{\link{op_correlate}()} \cr
+\code{\link{op_cos}()} \cr
+\code{\link{op_cosh}()} \cr
+\code{\link{op_count_nonzero}()} \cr
+\code{\link{op_cross}()} \cr
+\code{\link{op_ctc_decode}()} \cr
+\code{\link{op_cumprod}()} \cr
+\code{\link{op_cumsum}()} \cr
+\code{\link{op_deg2rad}()} \cr
+\code{\link{op_diag}()} \cr
+\code{\link{op_diagflat}()} \cr
+\code{\link{op_diagonal}()} \cr
+\code{\link{op_diff}()} \cr
+\code{\link{op_digitize}()} \cr
+\code{\link{op_divide}()} \cr
+\code{\link{op_divide_no_nan}()} \cr
+\code{\link{op_dot}()} \cr
+\code{\link{op_einsum}()} \cr
+\code{\link{op_empty}()} \cr
+\code{\link{op_equal}()} \cr
+\code{\link{op_exp}()} \cr
+\code{\link{op_exp2}()} \cr
+\code{\link{op_expand_dims}()} \cr
+\code{\link{op_expm1}()} \cr
+\code{\link{op_eye}()} \cr
+\code{\link{op_flip}()} \cr
+\code{\link{op_floor}()} \cr
+\code{\link{op_floor_divide}()} \cr
+\code{\link{op_full}()} \cr
+\code{\link{op_full_like}()} \cr
+\code{\link{op_get_item}()} \cr
+\code{\link{op_greater}()} \cr
+\code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
+\code{\link{op_heaviside}()} \cr
+\code{\link{op_histogram}()} \cr
+\code{\link{op_hstack}()} \cr
+\code{\link{op_identity}()} \cr
+\code{\link{op_imag}()} \cr
+\code{\link{op_inner}()} \cr
+\code{\link{op_isclose}()} \cr
+\code{\link{op_isfinite}()} \cr
+\code{\link{op_isinf}()} \cr
+\code{\link{op_isnan}()} \cr
+\code{\link{op_kaiser}()} \cr
+\code{\link{op_left_shift}()} \cr
+\code{\link{op_less}()} \cr
+\code{\link{op_less_equal}()} \cr
+\code{\link{op_linspace}()} \cr
+\code{\link{op_log}()} \cr
+\code{\link{op_log10}()} \cr
+\code{\link{op_log1p}()} \cr
+\code{\link{op_log2}()} \cr
+\code{\link{op_logaddexp}()} \cr
+\code{\link{op_logdet}()} \cr
+\code{\link{op_logical_and}()} \cr
+\code{\link{op_logical_not}()} \cr
+\code{\link{op_logical_or}()} \cr
+\code{\link{op_logical_xor}()} \cr
+\code{\link{op_logspace}()} \cr
+\code{\link{op_lstsq}()} \cr
+\code{\link{op_matmul}()} \cr
+\code{\link{op_max}()} \cr
+\code{\link{op_maximum}()} \cr
+\code{\link{op_mean}()} \cr
+\code{\link{op_median}()} \cr
+\code{\link{op_meshgrid}()} \cr
+\code{\link{op_min}()} \cr
+\code{\link{op_minimum}()} \cr
+\code{\link{op_mod}()} \cr
+\code{\link{op_moveaxis}()} \cr
+\code{\link{op_multiply}()} \cr
+\code{\link{op_nan_to_num}()} \cr
+\code{\link{op_ndim}()} \cr
+\code{\link{op_negative}()} \cr
+\code{\link{op_nonzero}()} \cr
+\code{\link{op_not_equal}()} \cr
+\code{\link{op_ones}()} \cr
+\code{\link{op_ones_like}()} \cr
+\code{\link{op_outer}()} \cr
+\code{\link{op_pad}()} \cr
+\code{\link{op_power}()} \cr
+\code{\link{op_prod}()} \cr
+\code{\link{op_quantile}()} \cr
+\code{\link{op_ravel}()} \cr
+\code{\link{op_real}()} \cr
+\code{\link{op_reciprocal}()} \cr
+\code{\link{op_repeat}()} \cr
+\code{\link{op_reshape}()} \cr
+\code{\link{op_right_shift}()} \cr
+\code{\link{op_roll}()} \cr
+\code{\link{op_rot90}()} \cr
+\code{\link{op_round}()} \cr
+\code{\link{op_saturate_cast}()} \cr
+\code{\link{op_select}()} \cr
+\code{\link{op_sign}()} \cr
+\code{\link{op_signbit}()} \cr
+\code{\link{op_sin}()} \cr
+\code{\link{op_sinh}()} \cr
+\code{\link{op_size}()} \cr
+\code{\link{op_sort}()} \cr
+\code{\link{op_split}()} \cr
+\code{\link{op_sqrt}()} \cr
+\code{\link{op_square}()} \cr
+\code{\link{op_squeeze}()} \cr
+\code{\link{op_stack}()} \cr
+\code{\link{op_std}()} \cr
+\code{\link{op_subtract}()} \cr
+\code{\link{op_sum}()} \cr
+\code{\link{op_swapaxes}()} \cr
+\code{\link{op_take}()} \cr
+\code{\link{op_take_along_axis}()} \cr
+\code{\link{op_tan}()} \cr
+\code{\link{op_tanh}()} \cr
+\code{\link{op_tensordot}()} \cr
+\code{\link{op_tile}()} \cr
+\code{\link{op_trace}()} \cr
+\code{\link{op_transpose}()} \cr
+\code{\link{op_tri}()} \cr
+\code{\link{op_tril}()} \cr
+\code{\link{op_triu}()} \cr
+\code{\link{op_trunc}()} \cr
+\code{\link{op_var}()} \cr
+\code{\link{op_vdot}()} \cr
+\code{\link{op_vectorize}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
+\code{\link{op_vstack}()} \cr
+\code{\link{op_where}()} \cr
+\code{\link{op_zeros}()} \cr
+\code{\link{op_zeros_like}()} \cr
+
+Other ops: \cr
+\code{\link{op_abs}()} \cr
+\code{\link{op_add}()} \cr
+\code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
+\code{\link{op_any}()} \cr
+\code{\link{op_append}()} \cr
+\code{\link{op_arange}()} \cr
+\code{\link{op_arccos}()} \cr
+\code{\link{op_arccosh}()} \cr
+\code{\link{op_arcsin}()} \cr
+\code{\link{op_arcsinh}()} \cr
+\code{\link{op_arctan}()} \cr
+\code{\link{op_arctan2}()} \cr
+\code{\link{op_arctanh}()} \cr
+\code{\link{op_argmax}()} \cr
+\code{\link{op_argmin}()} \cr
+\code{\link{op_argpartition}()} \cr
+\code{\link{op_argsort}()} \cr
+\code{\link{op_array}()} \cr
+\code{\link{op_associative_scan}()} \cr
+\code{\link{op_average}()} \cr
+\code{\link{op_average_pool}()} \cr
+\code{\link{op_batch_normalization}()} \cr
+\code{\link{op_binary_crossentropy}()} \cr
+\code{\link{op_bincount}()} \cr
+\code{\link{op_bitwise_and}()} \cr
+\code{\link{op_bitwise_invert}()} \cr
+\code{\link{op_bitwise_left_shift}()} \cr
+\code{\link{op_bitwise_not}()} \cr
+\code{\link{op_bitwise_or}()} \cr
+\code{\link{op_bitwise_right_shift}()} \cr
+\code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
+\code{\link{op_broadcast_to}()} \cr
+\code{\link{op_cast}()} \cr
+\code{\link{op_categorical_crossentropy}()} \cr
+\code{\link{op_cbrt}()} \cr
+\code{\link{op_ceil}()} \cr
+\code{\link{op_celu}()} \cr
+\code{\link{op_cholesky}()} \cr
+\code{\link{op_clip}()} \cr
+\code{\link{op_concatenate}()} \cr
+\code{\link{op_cond}()} \cr
+\code{\link{op_conj}()} \cr
+\code{\link{op_conv}()} \cr
+\code{\link{op_conv_transpose}()} \cr
+\code{\link{op_convert_to_numpy}()} \cr
+\code{\link{op_convert_to_tensor}()} \cr
+\code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
+\code{\link{op_correlate}()} \cr
+\code{\link{op_cos}()} \cr
+\code{\link{op_cosh}()} \cr
+\code{\link{op_count_nonzero}()} \cr
+\code{\link{op_cross}()} \cr
+\code{\link{op_ctc_decode}()} \cr
+\code{\link{op_ctc_loss}()} \cr
+\code{\link{op_cumprod}()} \cr
+\code{\link{op_cumsum}()} \cr
+\code{\link{op_custom_gradient}()} \cr
+\code{\link{op_deg2rad}()} \cr
+\code{\link{op_depthwise_conv}()} \cr
+\code{\link{op_det}()} \cr
+\code{\link{op_diag}()} \cr
+\code{\link{op_diagflat}()} \cr
+\code{\link{op_diagonal}()} \cr
+\code{\link{op_diff}()} \cr
+\code{\link{op_digitize}()} \cr
+\code{\link{op_divide}()} \cr
+\code{\link{op_divide_no_nan}()} \cr
+\code{\link{op_dot}()} \cr
+\code{\link{op_dot_product_attention}()} \cr
+\code{\link{op_dtype}()} \cr
+\code{\link{op_eig}()} \cr
+\code{\link{op_eigh}()} \cr
+\code{\link{op_einsum}()} \cr
+\code{\link{op_elu}()} \cr
+\code{\link{op_empty}()} \cr
+\code{\link{op_equal}()} \cr
+\code{\link{op_erf}()} \cr
+\code{\link{op_erfinv}()} \cr
+\code{\link{op_exp}()} \cr
+\code{\link{op_exp2}()} \cr
+\code{\link{op_expand_dims}()} \cr
+\code{\link{op_expm1}()} \cr
+\code{\link{op_extract_sequences}()} \cr
+\code{\link{op_eye}()} \cr
+\code{\link{op_fft}()} \cr
+\code{\link{op_fft2}()} \cr
+\code{\link{op_flip}()} \cr
+\code{\link{op_floor}()} \cr
+\code{\link{op_floor_divide}()} \cr
+\code{\link{op_fori_loop}()} \cr
+\code{\link{op_full}()} \cr
+\code{\link{op_full_like}()} \cr
+\code{\link{op_gelu}()} \cr
+\code{\link{op_get_item}()} \cr
+\code{\link{op_glu}()} \cr
+\code{\link{op_greater}()} \cr
+\code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
+\code{\link{op_hard_shrink}()} \cr
+\code{\link{op_hard_sigmoid}()} \cr
+\code{\link{op_hard_silu}()} \cr
+\code{\link{op_hard_tanh}()} \cr
+\code{\link{op_heaviside}()} \cr
+\code{\link{op_histogram}()} \cr
+\code{\link{op_hstack}()} \cr
+\code{\link{op_identity}()} \cr
+\code{\link{op_ifft2}()} \cr
+\code{\link{op_imag}()} \cr
+\code{\link{op_image_affine_transform}()} \cr
+\code{\link{op_image_crop}()} \cr
+\code{\link{op_image_extract_patches}()} \cr
+\code{\link{op_image_gaussian_blur}()} \cr
+\code{\link{op_image_hsv_to_rgb}()} \cr
+\code{\link{op_image_map_coordinates}()} \cr
+\code{\link{op_image_pad}()} \cr
+\code{\link{op_image_perspective_transform}()} \cr
+\code{\link{op_image_resize}()} \cr
+\code{\link{op_image_rgb_to_grayscale}()} \cr
+\code{\link{op_image_rgb_to_hsv}()} \cr
+\code{\link{op_in_top_k}()} \cr
+\code{\link{op_inner}()} \cr
+\code{\link{op_inv}()} \cr
+\code{\link{op_irfft}()} \cr
+\code{\link{op_is_tensor}()} \cr
+\code{\link{op_isclose}()} \cr
+\code{\link{op_isfinite}()} \cr
+\code{\link{op_isinf}()} \cr
+\code{\link{op_isnan}()} \cr
+\code{\link{op_istft}()} \cr
+\code{\link{op_kaiser}()} \cr
+\code{\link{op_layer_normalization}()} \cr
+\code{\link{op_leaky_relu}()} \cr
+\code{\link{op_left_shift}()} \cr
+\code{\link{op_less}()} \cr
+\code{\link{op_less_equal}()} \cr
+\code{\link{op_linspace}()} \cr
+\code{\link{op_log}()} \cr
+\code{\link{op_log10}()} \cr
+\code{\link{op_log1p}()} \cr
+\code{\link{op_log2}()} \cr
+\code{\link{op_log_sigmoid}()} \cr
+\code{\link{op_log_softmax}()} \cr
+\code{\link{op_logaddexp}()} \cr
+\code{\link{op_logdet}()} \cr
+\code{\link{op_logical_and}()} \cr
+\code{\link{op_logical_not}()} \cr
+\code{\link{op_logical_or}()} \cr
+\code{\link{op_logical_xor}()} \cr
+\code{\link{op_logspace}()} \cr
+\code{\link{op_logsumexp}()} \cr
+\code{\link{op_lstsq}()} \cr
+\code{\link{op_lu_factor}()} \cr
+\code{\link{op_map}()} \cr
+\code{\link{op_matmul}()} \cr
+\code{\link{op_max}()} \cr
+\code{\link{op_max_pool}()} \cr
+\code{\link{op_maximum}()} \cr
+\code{\link{op_mean}()} \cr
+\code{\link{op_median}()} \cr
+\code{\link{op_meshgrid}()} \cr
+\code{\link{op_min}()} \cr
+\code{\link{op_minimum}()} \cr
+\code{\link{op_mod}()} \cr
+\code{\link{op_moments}()} \cr
+\code{\link{op_moveaxis}()} \cr
+\code{\link{op_multi_hot}()} \cr
+\code{\link{op_multiply}()} \cr
+\code{\link{op_nan_to_num}()} \cr
+\code{\link{op_ndim}()} \cr
+\code{\link{op_negative}()} \cr
+\code{\link{op_nonzero}()} \cr
+\code{\link{op_norm}()} \cr
+\code{\link{op_normalize}()} \cr
+\code{\link{op_not_equal}()} \cr
+\code{\link{op_one_hot}()} \cr
+\code{\link{op_ones}()} \cr
+\code{\link{op_ones_like}()} \cr
+\code{\link{op_outer}()} \cr
+\code{\link{op_pad}()} \cr
+\code{\link{op_polar}()} \cr
+\code{\link{op_power}()} \cr
+\code{\link{op_prod}()} \cr
+\code{\link{op_psnr}()} \cr
+\code{\link{op_qr}()} \cr
+\code{\link{op_quantile}()} \cr
+\code{\link{op_ravel}()} \cr
+\code{\link{op_real}()} \cr
+\code{\link{op_rearrange}()} \cr
+\code{\link{op_reciprocal}()} \cr
+\code{\link{op_relu}()} \cr
+\code{\link{op_relu6}()} \cr
+\code{\link{op_repeat}()} \cr
+\code{\link{op_reshape}()} \cr
+\code{\link{op_rfft}()} \cr
+\code{\link{op_right_shift}()} \cr
+\code{\link{op_rms_normalization}()} \cr
+\code{\link{op_roll}()} \cr
+\code{\link{op_rot90}()} \cr
+\code{\link{op_round}()} \cr
+\code{\link{op_rsqrt}()} \cr
+\code{\link{op_saturate_cast}()} \cr
+\code{\link{op_scan}()} \cr
+\code{\link{op_scatter}()} \cr
+\code{\link{op_scatter_update}()} \cr
+\code{\link{op_searchsorted}()} \cr
+\code{\link{op_segment_max}()} \cr
+\code{\link{op_segment_sum}()} \cr
+\code{\link{op_select}()} \cr
+\code{\link{op_selu}()} \cr
+\code{\link{op_separable_conv}()} \cr
+\code{\link{op_shape}()} \cr
+\code{\link{op_sigmoid}()} \cr
+\code{\link{op_sign}()} \cr
+\code{\link{op_signbit}()} \cr
+\code{\link{op_silu}()} \cr
+\code{\link{op_sin}()} \cr
+\code{\link{op_sinh}()} \cr
+\code{\link{op_size}()} \cr
+\code{\link{op_slice}()} \cr
+\code{\link{op_slice_update}()} \cr
+\code{\link{op_slogdet}()} \cr
+\code{\link{op_soft_shrink}()} \cr
+\code{\link{op_softmax}()} \cr
+\code{\link{op_softplus}()} \cr
+\code{\link{op_softsign}()} \cr
+\code{\link{op_solve}()} \cr
+\code{\link{op_solve_triangular}()} \cr
+\code{\link{op_sort}()} \cr
+\code{\link{op_sparse_categorical_crossentropy}()} \cr
+\code{\link{op_sparse_plus}()} \cr
+\code{\link{op_sparse_sigmoid}()} \cr
+\code{\link{op_sparsemax}()} \cr
+\code{\link{op_split}()} \cr
+\code{\link{op_sqrt}()} \cr
+\code{\link{op_square}()} \cr
+\code{\link{op_squareplus}()} \cr
+\code{\link{op_squeeze}()} \cr
+\code{\link{op_stack}()} \cr
+\code{\link{op_std}()} \cr
+\code{\link{op_stft}()} \cr
+\code{\link{op_stop_gradient}()} \cr
+\code{\link{op_subset}()} \cr
+\code{\link{op_subtract}()} \cr
+\code{\link{op_sum}()} \cr
+\code{\link{op_svd}()} \cr
+\code{\link{op_swapaxes}()} \cr
+\code{\link{op_switch}()} \cr
+\code{\link{op_take}()} \cr
+\code{\link{op_take_along_axis}()} \cr
+\code{\link{op_tan}()} \cr
+\code{\link{op_tanh}()} \cr
+\code{\link{op_tanh_shrink}()} \cr
+\code{\link{op_tensordot}()} \cr
+\code{\link{op_threshold}()} \cr
+\code{\link{op_tile}()} \cr
+\code{\link{op_top_k}()} \cr
+\code{\link{op_trace}()} \cr
+\code{\link{op_transpose}()} \cr
+\code{\link{op_tri}()} \cr
+\code{\link{op_tril}()} \cr
+\code{\link{op_triu}()} \cr
+\code{\link{op_trunc}()} \cr
+\code{\link{op_unravel_index}()} \cr
+\code{\link{op_unstack}()} \cr
+\code{\link{op_var}()} \cr
+\code{\link{op_vdot}()} \cr
+\code{\link{op_vectorize}()} \cr
+\code{\link{op_vectorized_map}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
+\code{\link{op_vstack}()} \cr
+\code{\link{op_where}()} \cr
+\code{\link{op_while_loop}()} \cr
+\code{\link{op_zeros}()} \cr
+\code{\link{op_zeros_like}()} \cr
+}
+\concept{numpy ops}
+\concept{ops}
diff --git a/man/op_batch_normalization.Rd b/man/op_batch_normalization.Rd
index 671e14be6..a30beee10 100644
--- a/man/op_batch_normalization.Rd
+++ b/man/op_batch_normalization.Rd
@@ -85,6 +85,7 @@ Other nn ops: \cr
 \code{\link{op_hard_sigmoid}()} \cr
 \code{\link{op_hard_silu}()} \cr
 \code{\link{op_hard_tanh}()} \cr
+\code{\link{op_layer_normalization}()} \cr
 \code{\link{op_leaky_relu}()} \cr
 \code{\link{op_log_sigmoid}()} \cr
 \code{\link{op_log_softmax}()} \cr
@@ -108,6 +109,7 @@ Other nn ops: \cr
 \code{\link{op_softsign}()} \cr
 \code{\link{op_sparse_categorical_crossentropy}()} \cr
 \code{\link{op_sparse_plus}()} \cr
+\code{\link{op_sparse_sigmoid}()} \cr
 \code{\link{op_sparsemax}()} \cr
 \code{\link{op_squareplus}()} \cr
 \code{\link{op_tanh_shrink}()} \cr
@@ -118,6 +120,7 @@ Other ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -136,6 +139,7 @@ Other ops: \cr
 \code{\link{op_associative_scan}()} \cr
 \code{\link{op_average}()} \cr
 \code{\link{op_average_pool}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_binary_crossentropy}()} \cr
 \code{\link{op_bincount}()} \cr
 \code{\link{op_bitwise_and}()} \cr
@@ -145,9 +149,11 @@ Other ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
 \code{\link{op_cast}()} \cr
 \code{\link{op_categorical_crossentropy}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_celu}()} \cr
 \code{\link{op_cholesky}()} \cr
@@ -160,6 +166,7 @@ Other ops: \cr
 \code{\link{op_convert_to_numpy}()} \cr
 \code{\link{op_convert_to_tensor}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -170,6 +177,7 @@ Other ops: \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
 \code{\link{op_custom_gradient}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_depthwise_conv}()} \cr
 \code{\link{op_det}()} \cr
 \code{\link{op_diag}()} \cr
@@ -209,10 +217,13 @@ Other ops: \cr
 \code{\link{op_glu}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
 \code{\link{op_hard_shrink}()} \cr
 \code{\link{op_hard_sigmoid}()} \cr
 \code{\link{op_hard_silu}()} \cr
 \code{\link{op_hard_tanh}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -239,6 +250,8 @@ Other ops: \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
 \code{\link{op_istft}()} \cr
+\code{\link{op_kaiser}()} \cr
+\code{\link{op_layer_normalization}()} \cr
 \code{\link{op_leaky_relu}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
@@ -338,6 +351,7 @@ Other ops: \cr
 \code{\link{op_sort}()} \cr
 \code{\link{op_sparse_categorical_crossentropy}()} \cr
 \code{\link{op_sparse_plus}()} \cr
+\code{\link{op_sparse_sigmoid}()} \cr
 \code{\link{op_sparsemax}()} \cr
 \code{\link{op_split}()} \cr
 \code{\link{op_sqrt}()} \cr
@@ -375,6 +389,8 @@ Other ops: \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
 \code{\link{op_vectorized_map}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_while_loop}()} \cr
diff --git a/man/op_binary_crossentropy.Rd b/man/op_binary_crossentropy.Rd
index 73fa63ded..a86179e47 100644
--- a/man/op_binary_crossentropy.Rd
+++ b/man/op_binary_crossentropy.Rd
@@ -63,6 +63,7 @@ Other nn ops: \cr
 \code{\link{op_hard_sigmoid}()} \cr
 \code{\link{op_hard_silu}()} \cr
 \code{\link{op_hard_tanh}()} \cr
+\code{\link{op_layer_normalization}()} \cr
 \code{\link{op_leaky_relu}()} \cr
 \code{\link{op_log_sigmoid}()} \cr
 \code{\link{op_log_softmax}()} \cr
@@ -86,6 +87,7 @@ Other nn ops: \cr
 \code{\link{op_softsign}()} \cr
 \code{\link{op_sparse_categorical_crossentropy}()} \cr
 \code{\link{op_sparse_plus}()} \cr
+\code{\link{op_sparse_sigmoid}()} \cr
 \code{\link{op_sparsemax}()} \cr
 \code{\link{op_squareplus}()} \cr
 \code{\link{op_tanh_shrink}()} \cr
@@ -96,6 +98,7 @@ Other ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -114,6 +117,7 @@ Other ops: \cr
 \code{\link{op_associative_scan}()} \cr
 \code{\link{op_average}()} \cr
 \code{\link{op_average_pool}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_batch_normalization}()} \cr
 \code{\link{op_bincount}()} \cr
 \code{\link{op_bitwise_and}()} \cr
@@ -123,9 +127,11 @@ Other ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
 \code{\link{op_cast}()} \cr
 \code{\link{op_categorical_crossentropy}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_celu}()} \cr
 \code{\link{op_cholesky}()} \cr
@@ -138,6 +144,7 @@ Other ops: \cr
 \code{\link{op_convert_to_numpy}()} \cr
 \code{\link{op_convert_to_tensor}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -148,6 +155,7 @@ Other ops: \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
 \code{\link{op_custom_gradient}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_depthwise_conv}()} \cr
 \code{\link{op_det}()} \cr
 \code{\link{op_diag}()} \cr
@@ -187,10 +195,13 @@ Other ops: \cr
 \code{\link{op_glu}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
 \code{\link{op_hard_shrink}()} \cr
 \code{\link{op_hard_sigmoid}()} \cr
 \code{\link{op_hard_silu}()} \cr
 \code{\link{op_hard_tanh}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -217,6 +228,8 @@ Other ops: \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
 \code{\link{op_istft}()} \cr
+\code{\link{op_kaiser}()} \cr
+\code{\link{op_layer_normalization}()} \cr
 \code{\link{op_leaky_relu}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
@@ -316,6 +329,7 @@ Other ops: \cr
 \code{\link{op_sort}()} \cr
 \code{\link{op_sparse_categorical_crossentropy}()} \cr
 \code{\link{op_sparse_plus}()} \cr
+\code{\link{op_sparse_sigmoid}()} \cr
 \code{\link{op_sparsemax}()} \cr
 \code{\link{op_split}()} \cr
 \code{\link{op_sqrt}()} \cr
@@ -353,6 +367,8 @@ Other ops: \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
 \code{\link{op_vectorized_map}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_while_loop}()} \cr
diff --git a/man/op_bincount.Rd b/man/op_bincount.Rd
index 24da5147e..c706c704d 100644
--- a/man/op_bincount.Rd
+++ b/man/op_bincount.Rd
@@ -80,6 +80,7 @@ Other numpy ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -96,6 +97,7 @@ Other numpy ops: \cr
 \code{\link{op_argsort}()} \cr
 \code{\link{op_array}()} \cr
 \code{\link{op_average}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_bitwise_and}()} \cr
 \code{\link{op_bitwise_invert}()} \cr
 \code{\link{op_bitwise_left_shift}()} \cr
@@ -103,12 +105,15 @@ Other numpy ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_clip}()} \cr
 \code{\link{op_concatenate}()} \cr
 \code{\link{op_conj}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -117,6 +122,7 @@ Other numpy ops: \cr
 \code{\link{op_ctc_decode}()} \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_diag}()} \cr
 \code{\link{op_diagflat}()} \cr
 \code{\link{op_diagonal}()} \cr
@@ -141,6 +147,9 @@ Other numpy ops: \cr
 \code{\link{op_get_item}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -150,6 +159,7 @@ Other numpy ops: \cr
 \code{\link{op_isfinite}()} \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
+\code{\link{op_kaiser}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
 \code{\link{op_less_equal}()} \cr
@@ -230,6 +240,8 @@ Other numpy ops: \cr
 \code{\link{op_var}()} \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_zeros}()} \cr
@@ -239,6 +251,7 @@ Other ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -257,6 +270,7 @@ Other ops: \cr
 \code{\link{op_associative_scan}()} \cr
 \code{\link{op_average}()} \cr
 \code{\link{op_average_pool}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_batch_normalization}()} \cr
 \code{\link{op_binary_crossentropy}()} \cr
 \code{\link{op_bitwise_and}()} \cr
@@ -266,9 +280,11 @@ Other ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
 \code{\link{op_cast}()} \cr
 \code{\link{op_categorical_crossentropy}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_celu}()} \cr
 \code{\link{op_cholesky}()} \cr
@@ -281,6 +297,7 @@ Other ops: \cr
 \code{\link{op_convert_to_numpy}()} \cr
 \code{\link{op_convert_to_tensor}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -291,6 +308,7 @@ Other ops: \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
 \code{\link{op_custom_gradient}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_depthwise_conv}()} \cr
 \code{\link{op_det}()} \cr
 \code{\link{op_diag}()} \cr
@@ -330,10 +348,13 @@ Other ops: \cr
 \code{\link{op_glu}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
 \code{\link{op_hard_shrink}()} \cr
 \code{\link{op_hard_sigmoid}()} \cr
 \code{\link{op_hard_silu}()} \cr
 \code{\link{op_hard_tanh}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -360,6 +381,8 @@ Other ops: \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
 \code{\link{op_istft}()} \cr
+\code{\link{op_kaiser}()} \cr
+\code{\link{op_layer_normalization}()} \cr
 \code{\link{op_leaky_relu}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
@@ -459,6 +482,7 @@ Other ops: \cr
 \code{\link{op_sort}()} \cr
 \code{\link{op_sparse_categorical_crossentropy}()} \cr
 \code{\link{op_sparse_plus}()} \cr
+\code{\link{op_sparse_sigmoid}()} \cr
 \code{\link{op_sparsemax}()} \cr
 \code{\link{op_split}()} \cr
 \code{\link{op_sqrt}()} \cr
@@ -496,6 +520,8 @@ Other ops: \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
 \code{\link{op_vectorized_map}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_while_loop}()} \cr
diff --git a/man/op_bitwise_and.Rd b/man/op_bitwise_and.Rd
index 26f621f2e..d91f6768f 100644
--- a/man/op_bitwise_and.Rd
+++ b/man/op_bitwise_and.Rd
@@ -24,6 +24,7 @@ Other numpy ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -40,6 +41,7 @@ Other numpy ops: \cr
 \code{\link{op_argsort}()} \cr
 \code{\link{op_array}()} \cr
 \code{\link{op_average}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_bincount}()} \cr
 \code{\link{op_bitwise_invert}()} \cr
 \code{\link{op_bitwise_left_shift}()} \cr
@@ -47,12 +49,15 @@ Other numpy ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_clip}()} \cr
 \code{\link{op_concatenate}()} \cr
 \code{\link{op_conj}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -61,6 +66,7 @@ Other numpy ops: \cr
 \code{\link{op_ctc_decode}()} \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_diag}()} \cr
 \code{\link{op_diagflat}()} \cr
 \code{\link{op_diagonal}()} \cr
@@ -85,6 +91,9 @@ Other numpy ops: \cr
 \code{\link{op_get_item}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -94,6 +103,7 @@ Other numpy ops: \cr
 \code{\link{op_isfinite}()} \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
+\code{\link{op_kaiser}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
 \code{\link{op_less_equal}()} \cr
@@ -174,6 +184,8 @@ Other numpy ops: \cr
 \code{\link{op_var}()} \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_zeros}()} \cr
@@ -183,6 +195,7 @@ Other ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -201,6 +214,7 @@ Other ops: \cr
 \code{\link{op_associative_scan}()} \cr
 \code{\link{op_average}()} \cr
 \code{\link{op_average_pool}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_batch_normalization}()} \cr
 \code{\link{op_binary_crossentropy}()} \cr
 \code{\link{op_bincount}()} \cr
@@ -210,9 +224,11 @@ Other ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
 \code{\link{op_cast}()} \cr
 \code{\link{op_categorical_crossentropy}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_celu}()} \cr
 \code{\link{op_cholesky}()} \cr
@@ -225,6 +241,7 @@ Other ops: \cr
 \code{\link{op_convert_to_numpy}()} \cr
 \code{\link{op_convert_to_tensor}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -235,6 +252,7 @@ Other ops: \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
 \code{\link{op_custom_gradient}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_depthwise_conv}()} \cr
 \code{\link{op_det}()} \cr
 \code{\link{op_diag}()} \cr
@@ -274,10 +292,13 @@ Other ops: \cr
 \code{\link{op_glu}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
 \code{\link{op_hard_shrink}()} \cr
 \code{\link{op_hard_sigmoid}()} \cr
 \code{\link{op_hard_silu}()} \cr
 \code{\link{op_hard_tanh}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -304,6 +325,8 @@ Other ops: \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
 \code{\link{op_istft}()} \cr
+\code{\link{op_kaiser}()} \cr
+\code{\link{op_layer_normalization}()} \cr
 \code{\link{op_leaky_relu}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
@@ -403,6 +426,7 @@ Other ops: \cr
 \code{\link{op_sort}()} \cr
 \code{\link{op_sparse_categorical_crossentropy}()} \cr
 \code{\link{op_sparse_plus}()} \cr
+\code{\link{op_sparse_sigmoid}()} \cr
 \code{\link{op_sparsemax}()} \cr
 \code{\link{op_split}()} \cr
 \code{\link{op_sqrt}()} \cr
@@ -440,6 +464,8 @@ Other ops: \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
 \code{\link{op_vectorized_map}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_while_loop}()} \cr
diff --git a/man/op_bitwise_invert.Rd b/man/op_bitwise_invert.Rd
index 9dc73fd09..e8d195040 100644
--- a/man/op_bitwise_invert.Rd
+++ b/man/op_bitwise_invert.Rd
@@ -22,6 +22,7 @@ Other numpy ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -38,6 +39,7 @@ Other numpy ops: \cr
 \code{\link{op_argsort}()} \cr
 \code{\link{op_array}()} \cr
 \code{\link{op_average}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_bincount}()} \cr
 \code{\link{op_bitwise_and}()} \cr
 \code{\link{op_bitwise_left_shift}()} \cr
@@ -45,12 +47,15 @@ Other numpy ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_clip}()} \cr
 \code{\link{op_concatenate}()} \cr
 \code{\link{op_conj}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -59,6 +64,7 @@ Other numpy ops: \cr
 \code{\link{op_ctc_decode}()} \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_diag}()} \cr
 \code{\link{op_diagflat}()} \cr
 \code{\link{op_diagonal}()} \cr
@@ -83,6 +89,9 @@ Other numpy ops: \cr
 \code{\link{op_get_item}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -92,6 +101,7 @@ Other numpy ops: \cr
 \code{\link{op_isfinite}()} \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
+\code{\link{op_kaiser}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
 \code{\link{op_less_equal}()} \cr
@@ -172,6 +182,8 @@ Other numpy ops: \cr
 \code{\link{op_var}()} \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_zeros}()} \cr
@@ -181,6 +193,7 @@ Other ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -199,6 +212,7 @@ Other ops: \cr
 \code{\link{op_associative_scan}()} \cr
 \code{\link{op_average}()} \cr
 \code{\link{op_average_pool}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_batch_normalization}()} \cr
 \code{\link{op_binary_crossentropy}()} \cr
 \code{\link{op_bincount}()} \cr
@@ -208,9 +222,11 @@ Other ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
 \code{\link{op_cast}()} \cr
 \code{\link{op_categorical_crossentropy}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_celu}()} \cr
 \code{\link{op_cholesky}()} \cr
@@ -223,6 +239,7 @@ Other ops: \cr
 \code{\link{op_convert_to_numpy}()} \cr
 \code{\link{op_convert_to_tensor}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -233,6 +250,7 @@ Other ops: \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
 \code{\link{op_custom_gradient}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_depthwise_conv}()} \cr
 \code{\link{op_det}()} \cr
 \code{\link{op_diag}()} \cr
@@ -272,10 +290,13 @@ Other ops: \cr
 \code{\link{op_glu}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
 \code{\link{op_hard_shrink}()} \cr
 \code{\link{op_hard_sigmoid}()} \cr
 \code{\link{op_hard_silu}()} \cr
 \code{\link{op_hard_tanh}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -302,6 +323,8 @@ Other ops: \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
 \code{\link{op_istft}()} \cr
+\code{\link{op_kaiser}()} \cr
+\code{\link{op_layer_normalization}()} \cr
 \code{\link{op_leaky_relu}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
@@ -401,6 +424,7 @@ Other ops: \cr
 \code{\link{op_sort}()} \cr
 \code{\link{op_sparse_categorical_crossentropy}()} \cr
 \code{\link{op_sparse_plus}()} \cr
+\code{\link{op_sparse_sigmoid}()} \cr
 \code{\link{op_sparsemax}()} \cr
 \code{\link{op_split}()} \cr
 \code{\link{op_sqrt}()} \cr
@@ -438,6 +462,8 @@ Other ops: \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
 \code{\link{op_vectorized_map}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_while_loop}()} \cr
diff --git a/man/op_bitwise_left_shift.Rd b/man/op_bitwise_left_shift.Rd
index 0af4280ae..1ebdea586 100644
--- a/man/op_bitwise_left_shift.Rd
+++ b/man/op_bitwise_left_shift.Rd
@@ -24,6 +24,7 @@ Other numpy ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -40,6 +41,7 @@ Other numpy ops: \cr
 \code{\link{op_argsort}()} \cr
 \code{\link{op_array}()} \cr
 \code{\link{op_average}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_bincount}()} \cr
 \code{\link{op_bitwise_and}()} \cr
 \code{\link{op_bitwise_invert}()} \cr
@@ -47,12 +49,15 @@ Other numpy ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_clip}()} \cr
 \code{\link{op_concatenate}()} \cr
 \code{\link{op_conj}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -61,6 +66,7 @@ Other numpy ops: \cr
 \code{\link{op_ctc_decode}()} \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_diag}()} \cr
 \code{\link{op_diagflat}()} \cr
 \code{\link{op_diagonal}()} \cr
@@ -85,6 +91,9 @@ Other numpy ops: \cr
 \code{\link{op_get_item}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -94,6 +103,7 @@ Other numpy ops: \cr
 \code{\link{op_isfinite}()} \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
+\code{\link{op_kaiser}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
 \code{\link{op_less_equal}()} \cr
@@ -174,6 +184,8 @@ Other numpy ops: \cr
 \code{\link{op_var}()} \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_zeros}()} \cr
@@ -183,6 +195,7 @@ Other ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -201,6 +214,7 @@ Other ops: \cr
 \code{\link{op_associative_scan}()} \cr
 \code{\link{op_average}()} \cr
 \code{\link{op_average_pool}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_batch_normalization}()} \cr
 \code{\link{op_binary_crossentropy}()} \cr
 \code{\link{op_bincount}()} \cr
@@ -210,9 +224,11 @@ Other ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
 \code{\link{op_cast}()} \cr
 \code{\link{op_categorical_crossentropy}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_celu}()} \cr
 \code{\link{op_cholesky}()} \cr
@@ -225,6 +241,7 @@ Other ops: \cr
 \code{\link{op_convert_to_numpy}()} \cr
 \code{\link{op_convert_to_tensor}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -235,6 +252,7 @@ Other ops: \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
 \code{\link{op_custom_gradient}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_depthwise_conv}()} \cr
 \code{\link{op_det}()} \cr
 \code{\link{op_diag}()} \cr
@@ -274,10 +292,13 @@ Other ops: \cr
 \code{\link{op_glu}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
 \code{\link{op_hard_shrink}()} \cr
 \code{\link{op_hard_sigmoid}()} \cr
 \code{\link{op_hard_silu}()} \cr
 \code{\link{op_hard_tanh}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -304,6 +325,8 @@ Other ops: \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
 \code{\link{op_istft}()} \cr
+\code{\link{op_kaiser}()} \cr
+\code{\link{op_layer_normalization}()} \cr
 \code{\link{op_leaky_relu}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
@@ -403,6 +426,7 @@ Other ops: \cr
 \code{\link{op_sort}()} \cr
 \code{\link{op_sparse_categorical_crossentropy}()} \cr
 \code{\link{op_sparse_plus}()} \cr
+\code{\link{op_sparse_sigmoid}()} \cr
 \code{\link{op_sparsemax}()} \cr
 \code{\link{op_split}()} \cr
 \code{\link{op_sqrt}()} \cr
@@ -440,6 +464,8 @@ Other ops: \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
 \code{\link{op_vectorized_map}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_while_loop}()} \cr
diff --git a/man/op_bitwise_not.Rd b/man/op_bitwise_not.Rd
index 61c678019..c5a40a4fd 100644
--- a/man/op_bitwise_not.Rd
+++ b/man/op_bitwise_not.Rd
@@ -22,6 +22,7 @@ Other numpy ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -38,6 +39,7 @@ Other numpy ops: \cr
 \code{\link{op_argsort}()} \cr
 \code{\link{op_array}()} \cr
 \code{\link{op_average}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_bincount}()} \cr
 \code{\link{op_bitwise_and}()} \cr
 \code{\link{op_bitwise_invert}()} \cr
@@ -45,12 +47,15 @@ Other numpy ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_clip}()} \cr
 \code{\link{op_concatenate}()} \cr
 \code{\link{op_conj}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -59,6 +64,7 @@ Other numpy ops: \cr
 \code{\link{op_ctc_decode}()} \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_diag}()} \cr
 \code{\link{op_diagflat}()} \cr
 \code{\link{op_diagonal}()} \cr
@@ -83,6 +89,9 @@ Other numpy ops: \cr
 \code{\link{op_get_item}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -92,6 +101,7 @@ Other numpy ops: \cr
 \code{\link{op_isfinite}()} \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
+\code{\link{op_kaiser}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
 \code{\link{op_less_equal}()} \cr
@@ -172,6 +182,8 @@ Other numpy ops: \cr
 \code{\link{op_var}()} \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_zeros}()} \cr
@@ -181,6 +193,7 @@ Other ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -199,6 +212,7 @@ Other ops: \cr
 \code{\link{op_associative_scan}()} \cr
 \code{\link{op_average}()} \cr
 \code{\link{op_average_pool}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_batch_normalization}()} \cr
 \code{\link{op_binary_crossentropy}()} \cr
 \code{\link{op_bincount}()} \cr
@@ -208,9 +222,11 @@ Other ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
 \code{\link{op_cast}()} \cr
 \code{\link{op_categorical_crossentropy}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_celu}()} \cr
 \code{\link{op_cholesky}()} \cr
@@ -223,6 +239,7 @@ Other ops: \cr
 \code{\link{op_convert_to_numpy}()} \cr
 \code{\link{op_convert_to_tensor}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -233,6 +250,7 @@ Other ops: \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
 \code{\link{op_custom_gradient}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_depthwise_conv}()} \cr
 \code{\link{op_det}()} \cr
 \code{\link{op_diag}()} \cr
@@ -272,10 +290,13 @@ Other ops: \cr
 \code{\link{op_glu}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
 \code{\link{op_hard_shrink}()} \cr
 \code{\link{op_hard_sigmoid}()} \cr
 \code{\link{op_hard_silu}()} \cr
 \code{\link{op_hard_tanh}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -302,6 +323,8 @@ Other ops: \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
 \code{\link{op_istft}()} \cr
+\code{\link{op_kaiser}()} \cr
+\code{\link{op_layer_normalization}()} \cr
 \code{\link{op_leaky_relu}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
@@ -401,6 +424,7 @@ Other ops: \cr
 \code{\link{op_sort}()} \cr
 \code{\link{op_sparse_categorical_crossentropy}()} \cr
 \code{\link{op_sparse_plus}()} \cr
+\code{\link{op_sparse_sigmoid}()} \cr
 \code{\link{op_sparsemax}()} \cr
 \code{\link{op_split}()} \cr
 \code{\link{op_sqrt}()} \cr
@@ -438,6 +462,8 @@ Other ops: \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
 \code{\link{op_vectorized_map}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_while_loop}()} \cr
diff --git a/man/op_bitwise_or.Rd b/man/op_bitwise_or.Rd
index 0f2c46d9c..0b6f70e3e 100644
--- a/man/op_bitwise_or.Rd
+++ b/man/op_bitwise_or.Rd
@@ -24,6 +24,7 @@ Other numpy ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -40,6 +41,7 @@ Other numpy ops: \cr
 \code{\link{op_argsort}()} \cr
 \code{\link{op_array}()} \cr
 \code{\link{op_average}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_bincount}()} \cr
 \code{\link{op_bitwise_and}()} \cr
 \code{\link{op_bitwise_invert}()} \cr
@@ -47,12 +49,15 @@ Other numpy ops: \cr
 \code{\link{op_bitwise_not}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_clip}()} \cr
 \code{\link{op_concatenate}()} \cr
 \code{\link{op_conj}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -61,6 +66,7 @@ Other numpy ops: \cr
 \code{\link{op_ctc_decode}()} \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_diag}()} \cr
 \code{\link{op_diagflat}()} \cr
 \code{\link{op_diagonal}()} \cr
@@ -85,6 +91,9 @@ Other numpy ops: \cr
 \code{\link{op_get_item}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -94,6 +103,7 @@ Other numpy ops: \cr
 \code{\link{op_isfinite}()} \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
+\code{\link{op_kaiser}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
 \code{\link{op_less_equal}()} \cr
@@ -174,6 +184,8 @@ Other numpy ops: \cr
 \code{\link{op_var}()} \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_zeros}()} \cr
@@ -183,6 +195,7 @@ Other ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -201,6 +214,7 @@ Other ops: \cr
 \code{\link{op_associative_scan}()} \cr
 \code{\link{op_average}()} \cr
 \code{\link{op_average_pool}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_batch_normalization}()} \cr
 \code{\link{op_binary_crossentropy}()} \cr
 \code{\link{op_bincount}()} \cr
@@ -210,9 +224,11 @@ Other ops: \cr
 \code{\link{op_bitwise_not}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
 \code{\link{op_cast}()} \cr
 \code{\link{op_categorical_crossentropy}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_celu}()} \cr
 \code{\link{op_cholesky}()} \cr
@@ -225,6 +241,7 @@ Other ops: \cr
 \code{\link{op_convert_to_numpy}()} \cr
 \code{\link{op_convert_to_tensor}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -235,6 +252,7 @@ Other ops: \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
 \code{\link{op_custom_gradient}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_depthwise_conv}()} \cr
 \code{\link{op_det}()} \cr
 \code{\link{op_diag}()} \cr
@@ -274,10 +292,13 @@ Other ops: \cr
 \code{\link{op_glu}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
 \code{\link{op_hard_shrink}()} \cr
 \code{\link{op_hard_sigmoid}()} \cr
 \code{\link{op_hard_silu}()} \cr
 \code{\link{op_hard_tanh}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -304,6 +325,8 @@ Other ops: \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
 \code{\link{op_istft}()} \cr
+\code{\link{op_kaiser}()} \cr
+\code{\link{op_layer_normalization}()} \cr
 \code{\link{op_leaky_relu}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
@@ -403,6 +426,7 @@ Other ops: \cr
 \code{\link{op_sort}()} \cr
 \code{\link{op_sparse_categorical_crossentropy}()} \cr
 \code{\link{op_sparse_plus}()} \cr
+\code{\link{op_sparse_sigmoid}()} \cr
 \code{\link{op_sparsemax}()} \cr
 \code{\link{op_split}()} \cr
 \code{\link{op_sqrt}()} \cr
@@ -440,6 +464,8 @@ Other ops: \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
 \code{\link{op_vectorized_map}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_while_loop}()} \cr
diff --git a/man/op_bitwise_right_shift.Rd b/man/op_bitwise_right_shift.Rd
index 2a6dba4a4..11070bd12 100644
--- a/man/op_bitwise_right_shift.Rd
+++ b/man/op_bitwise_right_shift.Rd
@@ -24,6 +24,7 @@ Other numpy ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -40,6 +41,7 @@ Other numpy ops: \cr
 \code{\link{op_argsort}()} \cr
 \code{\link{op_array}()} \cr
 \code{\link{op_average}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_bincount}()} \cr
 \code{\link{op_bitwise_and}()} \cr
 \code{\link{op_bitwise_invert}()} \cr
@@ -47,12 +49,15 @@ Other numpy ops: \cr
 \code{\link{op_bitwise_not}()} \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_clip}()} \cr
 \code{\link{op_concatenate}()} \cr
 \code{\link{op_conj}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -61,6 +66,7 @@ Other numpy ops: \cr
 \code{\link{op_ctc_decode}()} \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_diag}()} \cr
 \code{\link{op_diagflat}()} \cr
 \code{\link{op_diagonal}()} \cr
@@ -85,6 +91,9 @@ Other numpy ops: \cr
 \code{\link{op_get_item}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -94,6 +103,7 @@ Other numpy ops: \cr
 \code{\link{op_isfinite}()} \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
+\code{\link{op_kaiser}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
 \code{\link{op_less_equal}()} \cr
@@ -174,6 +184,8 @@ Other numpy ops: \cr
 \code{\link{op_var}()} \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_zeros}()} \cr
@@ -183,6 +195,7 @@ Other ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -201,6 +214,7 @@ Other ops: \cr
 \code{\link{op_associative_scan}()} \cr
 \code{\link{op_average}()} \cr
 \code{\link{op_average_pool}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_batch_normalization}()} \cr
 \code{\link{op_binary_crossentropy}()} \cr
 \code{\link{op_bincount}()} \cr
@@ -210,9 +224,11 @@ Other ops: \cr
 \code{\link{op_bitwise_not}()} \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
 \code{\link{op_cast}()} \cr
 \code{\link{op_categorical_crossentropy}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_celu}()} \cr
 \code{\link{op_cholesky}()} \cr
@@ -225,6 +241,7 @@ Other ops: \cr
 \code{\link{op_convert_to_numpy}()} \cr
 \code{\link{op_convert_to_tensor}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -235,6 +252,7 @@ Other ops: \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
 \code{\link{op_custom_gradient}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_depthwise_conv}()} \cr
 \code{\link{op_det}()} \cr
 \code{\link{op_diag}()} \cr
@@ -274,10 +292,13 @@ Other ops: \cr
 \code{\link{op_glu}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
 \code{\link{op_hard_shrink}()} \cr
 \code{\link{op_hard_sigmoid}()} \cr
 \code{\link{op_hard_silu}()} \cr
 \code{\link{op_hard_tanh}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -304,6 +325,8 @@ Other ops: \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
 \code{\link{op_istft}()} \cr
+\code{\link{op_kaiser}()} \cr
+\code{\link{op_layer_normalization}()} \cr
 \code{\link{op_leaky_relu}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
@@ -403,6 +426,7 @@ Other ops: \cr
 \code{\link{op_sort}()} \cr
 \code{\link{op_sparse_categorical_crossentropy}()} \cr
 \code{\link{op_sparse_plus}()} \cr
+\code{\link{op_sparse_sigmoid}()} \cr
 \code{\link{op_sparsemax}()} \cr
 \code{\link{op_split}()} \cr
 \code{\link{op_sqrt}()} \cr
@@ -440,6 +464,8 @@ Other ops: \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
 \code{\link{op_vectorized_map}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_while_loop}()} \cr
diff --git a/man/op_bitwise_xor.Rd b/man/op_bitwise_xor.Rd
index 65450e9cf..a4d0bd68b 100644
--- a/man/op_bitwise_xor.Rd
+++ b/man/op_bitwise_xor.Rd
@@ -24,6 +24,7 @@ Other numpy ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -40,6 +41,7 @@ Other numpy ops: \cr
 \code{\link{op_argsort}()} \cr
 \code{\link{op_array}()} \cr
 \code{\link{op_average}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_bincount}()} \cr
 \code{\link{op_bitwise_and}()} \cr
 \code{\link{op_bitwise_invert}()} \cr
@@ -47,12 +49,15 @@ Other numpy ops: \cr
 \code{\link{op_bitwise_not}()} \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_clip}()} \cr
 \code{\link{op_concatenate}()} \cr
 \code{\link{op_conj}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -61,6 +66,7 @@ Other numpy ops: \cr
 \code{\link{op_ctc_decode}()} \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_diag}()} \cr
 \code{\link{op_diagflat}()} \cr
 \code{\link{op_diagonal}()} \cr
@@ -85,6 +91,9 @@ Other numpy ops: \cr
 \code{\link{op_get_item}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -94,6 +103,7 @@ Other numpy ops: \cr
 \code{\link{op_isfinite}()} \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
+\code{\link{op_kaiser}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
 \code{\link{op_less_equal}()} \cr
@@ -174,6 +184,8 @@ Other numpy ops: \cr
 \code{\link{op_var}()} \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_zeros}()} \cr
@@ -183,6 +195,7 @@ Other ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -201,6 +214,7 @@ Other ops: \cr
 \code{\link{op_associative_scan}()} \cr
 \code{\link{op_average}()} \cr
 \code{\link{op_average_pool}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_batch_normalization}()} \cr
 \code{\link{op_binary_crossentropy}()} \cr
 \code{\link{op_bincount}()} \cr
@@ -210,9 +224,11 @@ Other ops: \cr
 \code{\link{op_bitwise_not}()} \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
 \code{\link{op_cast}()} \cr
 \code{\link{op_categorical_crossentropy}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_celu}()} \cr
 \code{\link{op_cholesky}()} \cr
@@ -225,6 +241,7 @@ Other ops: \cr
 \code{\link{op_convert_to_numpy}()} \cr
 \code{\link{op_convert_to_tensor}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -235,6 +252,7 @@ Other ops: \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
 \code{\link{op_custom_gradient}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_depthwise_conv}()} \cr
 \code{\link{op_det}()} \cr
 \code{\link{op_diag}()} \cr
@@ -274,10 +292,13 @@ Other ops: \cr
 \code{\link{op_glu}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
 \code{\link{op_hard_shrink}()} \cr
 \code{\link{op_hard_sigmoid}()} \cr
 \code{\link{op_hard_silu}()} \cr
 \code{\link{op_hard_tanh}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -304,6 +325,8 @@ Other ops: \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
 \code{\link{op_istft}()} \cr
+\code{\link{op_kaiser}()} \cr
+\code{\link{op_layer_normalization}()} \cr
 \code{\link{op_leaky_relu}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
@@ -403,6 +426,7 @@ Other ops: \cr
 \code{\link{op_sort}()} \cr
 \code{\link{op_sparse_categorical_crossentropy}()} \cr
 \code{\link{op_sparse_plus}()} \cr
+\code{\link{op_sparse_sigmoid}()} \cr
 \code{\link{op_sparsemax}()} \cr
 \code{\link{op_split}()} \cr
 \code{\link{op_sqrt}()} \cr
@@ -440,6 +464,8 @@ Other ops: \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
 \code{\link{op_vectorized_map}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_while_loop}()} \cr
diff --git a/man/op_blackman.Rd b/man/op_blackman.Rd
new file mode 100644
index 000000000..c8a253f3f
--- /dev/null
+++ b/man/op_blackman.Rd
@@ -0,0 +1,487 @@
+% Generated by roxygen2: do not edit by hand
+% Please edit documentation in R/ops-numpy.R
+\name{op_blackman}
+\alias{op_blackman}
+\title{Blackman window function.}
+\usage{
+op_blackman(x)
+}
+\arguments{
+\item{x}{Length of the window. Must be a positive integer.}
+}
+\value{
+A 1D tensor containing the window values.
+}
+\description{
+The Blackman window is a taper formed by using a weighted cosine.
+}
+\section{Examples}{
+\if{html}{\out{<div class="sourceCode r">}}\preformatted{op_blackman(5)
+}\if{html}{\out{</div>}}
+
+\if{html}{\out{<div class="sourceCode">}}\preformatted{## tf.Tensor(
+## [-1.4901161e-08  3.4000003e-01  9.9999994e-01  3.3999997e-01
+##  -1.4901161e-08], shape=(5), dtype=float32)
+
+}\if{html}{\out{</div>}}
+
+[-1.4901161e-08  3.4000003e-01  9.9999994e-01  3.3999997e-01
+\subsection{-1.4901161e-08]: R:-1.4901161e-08\%20\%203.4000003e-01\%20\%209.9999994e-01\%20\%203.3999997e-01\%0A##\%20\%20-1.4901161e-08}{
+}
+}
+
+\seealso{
+Other numpy ops: \cr
+\code{\link{op_abs}()} \cr
+\code{\link{op_add}()} \cr
+\code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
+\code{\link{op_any}()} \cr
+\code{\link{op_append}()} \cr
+\code{\link{op_arange}()} \cr
+\code{\link{op_arccos}()} \cr
+\code{\link{op_arccosh}()} \cr
+\code{\link{op_arcsin}()} \cr
+\code{\link{op_arcsinh}()} \cr
+\code{\link{op_arctan}()} \cr
+\code{\link{op_arctan2}()} \cr
+\code{\link{op_arctanh}()} \cr
+\code{\link{op_argmax}()} \cr
+\code{\link{op_argmin}()} \cr
+\code{\link{op_argpartition}()} \cr
+\code{\link{op_argsort}()} \cr
+\code{\link{op_array}()} \cr
+\code{\link{op_average}()} \cr
+\code{\link{op_bartlett}()} \cr
+\code{\link{op_bincount}()} \cr
+\code{\link{op_bitwise_and}()} \cr
+\code{\link{op_bitwise_invert}()} \cr
+\code{\link{op_bitwise_left_shift}()} \cr
+\code{\link{op_bitwise_not}()} \cr
+\code{\link{op_bitwise_or}()} \cr
+\code{\link{op_bitwise_right_shift}()} \cr
+\code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_broadcast_to}()} \cr
+\code{\link{op_cbrt}()} \cr
+\code{\link{op_ceil}()} \cr
+\code{\link{op_clip}()} \cr
+\code{\link{op_concatenate}()} \cr
+\code{\link{op_conj}()} \cr
+\code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
+\code{\link{op_correlate}()} \cr
+\code{\link{op_cos}()} \cr
+\code{\link{op_cosh}()} \cr
+\code{\link{op_count_nonzero}()} \cr
+\code{\link{op_cross}()} \cr
+\code{\link{op_ctc_decode}()} \cr
+\code{\link{op_cumprod}()} \cr
+\code{\link{op_cumsum}()} \cr
+\code{\link{op_deg2rad}()} \cr
+\code{\link{op_diag}()} \cr
+\code{\link{op_diagflat}()} \cr
+\code{\link{op_diagonal}()} \cr
+\code{\link{op_diff}()} \cr
+\code{\link{op_digitize}()} \cr
+\code{\link{op_divide}()} \cr
+\code{\link{op_divide_no_nan}()} \cr
+\code{\link{op_dot}()} \cr
+\code{\link{op_einsum}()} \cr
+\code{\link{op_empty}()} \cr
+\code{\link{op_equal}()} \cr
+\code{\link{op_exp}()} \cr
+\code{\link{op_exp2}()} \cr
+\code{\link{op_expand_dims}()} \cr
+\code{\link{op_expm1}()} \cr
+\code{\link{op_eye}()} \cr
+\code{\link{op_flip}()} \cr
+\code{\link{op_floor}()} \cr
+\code{\link{op_floor_divide}()} \cr
+\code{\link{op_full}()} \cr
+\code{\link{op_full_like}()} \cr
+\code{\link{op_get_item}()} \cr
+\code{\link{op_greater}()} \cr
+\code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
+\code{\link{op_heaviside}()} \cr
+\code{\link{op_histogram}()} \cr
+\code{\link{op_hstack}()} \cr
+\code{\link{op_identity}()} \cr
+\code{\link{op_imag}()} \cr
+\code{\link{op_inner}()} \cr
+\code{\link{op_isclose}()} \cr
+\code{\link{op_isfinite}()} \cr
+\code{\link{op_isinf}()} \cr
+\code{\link{op_isnan}()} \cr
+\code{\link{op_kaiser}()} \cr
+\code{\link{op_left_shift}()} \cr
+\code{\link{op_less}()} \cr
+\code{\link{op_less_equal}()} \cr
+\code{\link{op_linspace}()} \cr
+\code{\link{op_log}()} \cr
+\code{\link{op_log10}()} \cr
+\code{\link{op_log1p}()} \cr
+\code{\link{op_log2}()} \cr
+\code{\link{op_logaddexp}()} \cr
+\code{\link{op_logdet}()} \cr
+\code{\link{op_logical_and}()} \cr
+\code{\link{op_logical_not}()} \cr
+\code{\link{op_logical_or}()} \cr
+\code{\link{op_logical_xor}()} \cr
+\code{\link{op_logspace}()} \cr
+\code{\link{op_lstsq}()} \cr
+\code{\link{op_matmul}()} \cr
+\code{\link{op_max}()} \cr
+\code{\link{op_maximum}()} \cr
+\code{\link{op_mean}()} \cr
+\code{\link{op_median}()} \cr
+\code{\link{op_meshgrid}()} \cr
+\code{\link{op_min}()} \cr
+\code{\link{op_minimum}()} \cr
+\code{\link{op_mod}()} \cr
+\code{\link{op_moveaxis}()} \cr
+\code{\link{op_multiply}()} \cr
+\code{\link{op_nan_to_num}()} \cr
+\code{\link{op_ndim}()} \cr
+\code{\link{op_negative}()} \cr
+\code{\link{op_nonzero}()} \cr
+\code{\link{op_not_equal}()} \cr
+\code{\link{op_ones}()} \cr
+\code{\link{op_ones_like}()} \cr
+\code{\link{op_outer}()} \cr
+\code{\link{op_pad}()} \cr
+\code{\link{op_power}()} \cr
+\code{\link{op_prod}()} \cr
+\code{\link{op_quantile}()} \cr
+\code{\link{op_ravel}()} \cr
+\code{\link{op_real}()} \cr
+\code{\link{op_reciprocal}()} \cr
+\code{\link{op_repeat}()} \cr
+\code{\link{op_reshape}()} \cr
+\code{\link{op_right_shift}()} \cr
+\code{\link{op_roll}()} \cr
+\code{\link{op_rot90}()} \cr
+\code{\link{op_round}()} \cr
+\code{\link{op_saturate_cast}()} \cr
+\code{\link{op_select}()} \cr
+\code{\link{op_sign}()} \cr
+\code{\link{op_signbit}()} \cr
+\code{\link{op_sin}()} \cr
+\code{\link{op_sinh}()} \cr
+\code{\link{op_size}()} \cr
+\code{\link{op_sort}()} \cr
+\code{\link{op_split}()} \cr
+\code{\link{op_sqrt}()} \cr
+\code{\link{op_square}()} \cr
+\code{\link{op_squeeze}()} \cr
+\code{\link{op_stack}()} \cr
+\code{\link{op_std}()} \cr
+\code{\link{op_subtract}()} \cr
+\code{\link{op_sum}()} \cr
+\code{\link{op_swapaxes}()} \cr
+\code{\link{op_take}()} \cr
+\code{\link{op_take_along_axis}()} \cr
+\code{\link{op_tan}()} \cr
+\code{\link{op_tanh}()} \cr
+\code{\link{op_tensordot}()} \cr
+\code{\link{op_tile}()} \cr
+\code{\link{op_trace}()} \cr
+\code{\link{op_transpose}()} \cr
+\code{\link{op_tri}()} \cr
+\code{\link{op_tril}()} \cr
+\code{\link{op_triu}()} \cr
+\code{\link{op_trunc}()} \cr
+\code{\link{op_var}()} \cr
+\code{\link{op_vdot}()} \cr
+\code{\link{op_vectorize}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
+\code{\link{op_vstack}()} \cr
+\code{\link{op_where}()} \cr
+\code{\link{op_zeros}()} \cr
+\code{\link{op_zeros_like}()} \cr
+
+Other ops: \cr
+\code{\link{op_abs}()} \cr
+\code{\link{op_add}()} \cr
+\code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
+\code{\link{op_any}()} \cr
+\code{\link{op_append}()} \cr
+\code{\link{op_arange}()} \cr
+\code{\link{op_arccos}()} \cr
+\code{\link{op_arccosh}()} \cr
+\code{\link{op_arcsin}()} \cr
+\code{\link{op_arcsinh}()} \cr
+\code{\link{op_arctan}()} \cr
+\code{\link{op_arctan2}()} \cr
+\code{\link{op_arctanh}()} \cr
+\code{\link{op_argmax}()} \cr
+\code{\link{op_argmin}()} \cr
+\code{\link{op_argpartition}()} \cr
+\code{\link{op_argsort}()} \cr
+\code{\link{op_array}()} \cr
+\code{\link{op_associative_scan}()} \cr
+\code{\link{op_average}()} \cr
+\code{\link{op_average_pool}()} \cr
+\code{\link{op_bartlett}()} \cr
+\code{\link{op_batch_normalization}()} \cr
+\code{\link{op_binary_crossentropy}()} \cr
+\code{\link{op_bincount}()} \cr
+\code{\link{op_bitwise_and}()} \cr
+\code{\link{op_bitwise_invert}()} \cr
+\code{\link{op_bitwise_left_shift}()} \cr
+\code{\link{op_bitwise_not}()} \cr
+\code{\link{op_bitwise_or}()} \cr
+\code{\link{op_bitwise_right_shift}()} \cr
+\code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_broadcast_to}()} \cr
+\code{\link{op_cast}()} \cr
+\code{\link{op_categorical_crossentropy}()} \cr
+\code{\link{op_cbrt}()} \cr
+\code{\link{op_ceil}()} \cr
+\code{\link{op_celu}()} \cr
+\code{\link{op_cholesky}()} \cr
+\code{\link{op_clip}()} \cr
+\code{\link{op_concatenate}()} \cr
+\code{\link{op_cond}()} \cr
+\code{\link{op_conj}()} \cr
+\code{\link{op_conv}()} \cr
+\code{\link{op_conv_transpose}()} \cr
+\code{\link{op_convert_to_numpy}()} \cr
+\code{\link{op_convert_to_tensor}()} \cr
+\code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
+\code{\link{op_correlate}()} \cr
+\code{\link{op_cos}()} \cr
+\code{\link{op_cosh}()} \cr
+\code{\link{op_count_nonzero}()} \cr
+\code{\link{op_cross}()} \cr
+\code{\link{op_ctc_decode}()} \cr
+\code{\link{op_ctc_loss}()} \cr
+\code{\link{op_cumprod}()} \cr
+\code{\link{op_cumsum}()} \cr
+\code{\link{op_custom_gradient}()} \cr
+\code{\link{op_deg2rad}()} \cr
+\code{\link{op_depthwise_conv}()} \cr
+\code{\link{op_det}()} \cr
+\code{\link{op_diag}()} \cr
+\code{\link{op_diagflat}()} \cr
+\code{\link{op_diagonal}()} \cr
+\code{\link{op_diff}()} \cr
+\code{\link{op_digitize}()} \cr
+\code{\link{op_divide}()} \cr
+\code{\link{op_divide_no_nan}()} \cr
+\code{\link{op_dot}()} \cr
+\code{\link{op_dot_product_attention}()} \cr
+\code{\link{op_dtype}()} \cr
+\code{\link{op_eig}()} \cr
+\code{\link{op_eigh}()} \cr
+\code{\link{op_einsum}()} \cr
+\code{\link{op_elu}()} \cr
+\code{\link{op_empty}()} \cr
+\code{\link{op_equal}()} \cr
+\code{\link{op_erf}()} \cr
+\code{\link{op_erfinv}()} \cr
+\code{\link{op_exp}()} \cr
+\code{\link{op_exp2}()} \cr
+\code{\link{op_expand_dims}()} \cr
+\code{\link{op_expm1}()} \cr
+\code{\link{op_extract_sequences}()} \cr
+\code{\link{op_eye}()} \cr
+\code{\link{op_fft}()} \cr
+\code{\link{op_fft2}()} \cr
+\code{\link{op_flip}()} \cr
+\code{\link{op_floor}()} \cr
+\code{\link{op_floor_divide}()} \cr
+\code{\link{op_fori_loop}()} \cr
+\code{\link{op_full}()} \cr
+\code{\link{op_full_like}()} \cr
+\code{\link{op_gelu}()} \cr
+\code{\link{op_get_item}()} \cr
+\code{\link{op_glu}()} \cr
+\code{\link{op_greater}()} \cr
+\code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
+\code{\link{op_hard_shrink}()} \cr
+\code{\link{op_hard_sigmoid}()} \cr
+\code{\link{op_hard_silu}()} \cr
+\code{\link{op_hard_tanh}()} \cr
+\code{\link{op_heaviside}()} \cr
+\code{\link{op_histogram}()} \cr
+\code{\link{op_hstack}()} \cr
+\code{\link{op_identity}()} \cr
+\code{\link{op_ifft2}()} \cr
+\code{\link{op_imag}()} \cr
+\code{\link{op_image_affine_transform}()} \cr
+\code{\link{op_image_crop}()} \cr
+\code{\link{op_image_extract_patches}()} \cr
+\code{\link{op_image_gaussian_blur}()} \cr
+\code{\link{op_image_hsv_to_rgb}()} \cr
+\code{\link{op_image_map_coordinates}()} \cr
+\code{\link{op_image_pad}()} \cr
+\code{\link{op_image_perspective_transform}()} \cr
+\code{\link{op_image_resize}()} \cr
+\code{\link{op_image_rgb_to_grayscale}()} \cr
+\code{\link{op_image_rgb_to_hsv}()} \cr
+\code{\link{op_in_top_k}()} \cr
+\code{\link{op_inner}()} \cr
+\code{\link{op_inv}()} \cr
+\code{\link{op_irfft}()} \cr
+\code{\link{op_is_tensor}()} \cr
+\code{\link{op_isclose}()} \cr
+\code{\link{op_isfinite}()} \cr
+\code{\link{op_isinf}()} \cr
+\code{\link{op_isnan}()} \cr
+\code{\link{op_istft}()} \cr
+\code{\link{op_kaiser}()} \cr
+\code{\link{op_layer_normalization}()} \cr
+\code{\link{op_leaky_relu}()} \cr
+\code{\link{op_left_shift}()} \cr
+\code{\link{op_less}()} \cr
+\code{\link{op_less_equal}()} \cr
+\code{\link{op_linspace}()} \cr
+\code{\link{op_log}()} \cr
+\code{\link{op_log10}()} \cr
+\code{\link{op_log1p}()} \cr
+\code{\link{op_log2}()} \cr
+\code{\link{op_log_sigmoid}()} \cr
+\code{\link{op_log_softmax}()} \cr
+\code{\link{op_logaddexp}()} \cr
+\code{\link{op_logdet}()} \cr
+\code{\link{op_logical_and}()} \cr
+\code{\link{op_logical_not}()} \cr
+\code{\link{op_logical_or}()} \cr
+\code{\link{op_logical_xor}()} \cr
+\code{\link{op_logspace}()} \cr
+\code{\link{op_logsumexp}()} \cr
+\code{\link{op_lstsq}()} \cr
+\code{\link{op_lu_factor}()} \cr
+\code{\link{op_map}()} \cr
+\code{\link{op_matmul}()} \cr
+\code{\link{op_max}()} \cr
+\code{\link{op_max_pool}()} \cr
+\code{\link{op_maximum}()} \cr
+\code{\link{op_mean}()} \cr
+\code{\link{op_median}()} \cr
+\code{\link{op_meshgrid}()} \cr
+\code{\link{op_min}()} \cr
+\code{\link{op_minimum}()} \cr
+\code{\link{op_mod}()} \cr
+\code{\link{op_moments}()} \cr
+\code{\link{op_moveaxis}()} \cr
+\code{\link{op_multi_hot}()} \cr
+\code{\link{op_multiply}()} \cr
+\code{\link{op_nan_to_num}()} \cr
+\code{\link{op_ndim}()} \cr
+\code{\link{op_negative}()} \cr
+\code{\link{op_nonzero}()} \cr
+\code{\link{op_norm}()} \cr
+\code{\link{op_normalize}()} \cr
+\code{\link{op_not_equal}()} \cr
+\code{\link{op_one_hot}()} \cr
+\code{\link{op_ones}()} \cr
+\code{\link{op_ones_like}()} \cr
+\code{\link{op_outer}()} \cr
+\code{\link{op_pad}()} \cr
+\code{\link{op_polar}()} \cr
+\code{\link{op_power}()} \cr
+\code{\link{op_prod}()} \cr
+\code{\link{op_psnr}()} \cr
+\code{\link{op_qr}()} \cr
+\code{\link{op_quantile}()} \cr
+\code{\link{op_ravel}()} \cr
+\code{\link{op_real}()} \cr
+\code{\link{op_rearrange}()} \cr
+\code{\link{op_reciprocal}()} \cr
+\code{\link{op_relu}()} \cr
+\code{\link{op_relu6}()} \cr
+\code{\link{op_repeat}()} \cr
+\code{\link{op_reshape}()} \cr
+\code{\link{op_rfft}()} \cr
+\code{\link{op_right_shift}()} \cr
+\code{\link{op_rms_normalization}()} \cr
+\code{\link{op_roll}()} \cr
+\code{\link{op_rot90}()} \cr
+\code{\link{op_round}()} \cr
+\code{\link{op_rsqrt}()} \cr
+\code{\link{op_saturate_cast}()} \cr
+\code{\link{op_scan}()} \cr
+\code{\link{op_scatter}()} \cr
+\code{\link{op_scatter_update}()} \cr
+\code{\link{op_searchsorted}()} \cr
+\code{\link{op_segment_max}()} \cr
+\code{\link{op_segment_sum}()} \cr
+\code{\link{op_select}()} \cr
+\code{\link{op_selu}()} \cr
+\code{\link{op_separable_conv}()} \cr
+\code{\link{op_shape}()} \cr
+\code{\link{op_sigmoid}()} \cr
+\code{\link{op_sign}()} \cr
+\code{\link{op_signbit}()} \cr
+\code{\link{op_silu}()} \cr
+\code{\link{op_sin}()} \cr
+\code{\link{op_sinh}()} \cr
+\code{\link{op_size}()} \cr
+\code{\link{op_slice}()} \cr
+\code{\link{op_slice_update}()} \cr
+\code{\link{op_slogdet}()} \cr
+\code{\link{op_soft_shrink}()} \cr
+\code{\link{op_softmax}()} \cr
+\code{\link{op_softplus}()} \cr
+\code{\link{op_softsign}()} \cr
+\code{\link{op_solve}()} \cr
+\code{\link{op_solve_triangular}()} \cr
+\code{\link{op_sort}()} \cr
+\code{\link{op_sparse_categorical_crossentropy}()} \cr
+\code{\link{op_sparse_plus}()} \cr
+\code{\link{op_sparse_sigmoid}()} \cr
+\code{\link{op_sparsemax}()} \cr
+\code{\link{op_split}()} \cr
+\code{\link{op_sqrt}()} \cr
+\code{\link{op_square}()} \cr
+\code{\link{op_squareplus}()} \cr
+\code{\link{op_squeeze}()} \cr
+\code{\link{op_stack}()} \cr
+\code{\link{op_std}()} \cr
+\code{\link{op_stft}()} \cr
+\code{\link{op_stop_gradient}()} \cr
+\code{\link{op_subset}()} \cr
+\code{\link{op_subtract}()} \cr
+\code{\link{op_sum}()} \cr
+\code{\link{op_svd}()} \cr
+\code{\link{op_swapaxes}()} \cr
+\code{\link{op_switch}()} \cr
+\code{\link{op_take}()} \cr
+\code{\link{op_take_along_axis}()} \cr
+\code{\link{op_tan}()} \cr
+\code{\link{op_tanh}()} \cr
+\code{\link{op_tanh_shrink}()} \cr
+\code{\link{op_tensordot}()} \cr
+\code{\link{op_threshold}()} \cr
+\code{\link{op_tile}()} \cr
+\code{\link{op_top_k}()} \cr
+\code{\link{op_trace}()} \cr
+\code{\link{op_transpose}()} \cr
+\code{\link{op_tri}()} \cr
+\code{\link{op_tril}()} \cr
+\code{\link{op_triu}()} \cr
+\code{\link{op_trunc}()} \cr
+\code{\link{op_unravel_index}()} \cr
+\code{\link{op_unstack}()} \cr
+\code{\link{op_var}()} \cr
+\code{\link{op_vdot}()} \cr
+\code{\link{op_vectorize}()} \cr
+\code{\link{op_vectorized_map}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
+\code{\link{op_vstack}()} \cr
+\code{\link{op_where}()} \cr
+\code{\link{op_while_loop}()} \cr
+\code{\link{op_zeros}()} \cr
+\code{\link{op_zeros_like}()} \cr
+}
+\concept{numpy ops}
+\concept{ops}
diff --git a/man/op_broadcast_to.Rd b/man/op_broadcast_to.Rd
index 795f51d35..86874ce3e 100644
--- a/man/op_broadcast_to.Rd
+++ b/man/op_broadcast_to.Rd
@@ -39,6 +39,7 @@ Other numpy ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -55,6 +56,7 @@ Other numpy ops: \cr
 \code{\link{op_argsort}()} \cr
 \code{\link{op_array}()} \cr
 \code{\link{op_average}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_bincount}()} \cr
 \code{\link{op_bitwise_and}()} \cr
 \code{\link{op_bitwise_invert}()} \cr
@@ -63,11 +65,14 @@ Other numpy ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_clip}()} \cr
 \code{\link{op_concatenate}()} \cr
 \code{\link{op_conj}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -76,6 +81,7 @@ Other numpy ops: \cr
 \code{\link{op_ctc_decode}()} \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_diag}()} \cr
 \code{\link{op_diagflat}()} \cr
 \code{\link{op_diagonal}()} \cr
@@ -100,6 +106,9 @@ Other numpy ops: \cr
 \code{\link{op_get_item}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -109,6 +118,7 @@ Other numpy ops: \cr
 \code{\link{op_isfinite}()} \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
+\code{\link{op_kaiser}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
 \code{\link{op_less_equal}()} \cr
@@ -189,6 +199,8 @@ Other numpy ops: \cr
 \code{\link{op_var}()} \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_zeros}()} \cr
@@ -198,6 +210,7 @@ Other ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -216,6 +229,7 @@ Other ops: \cr
 \code{\link{op_associative_scan}()} \cr
 \code{\link{op_average}()} \cr
 \code{\link{op_average_pool}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_batch_normalization}()} \cr
 \code{\link{op_binary_crossentropy}()} \cr
 \code{\link{op_bincount}()} \cr
@@ -226,8 +240,10 @@ Other ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_cast}()} \cr
 \code{\link{op_categorical_crossentropy}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_celu}()} \cr
 \code{\link{op_cholesky}()} \cr
@@ -240,6 +256,7 @@ Other ops: \cr
 \code{\link{op_convert_to_numpy}()} \cr
 \code{\link{op_convert_to_tensor}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -250,6 +267,7 @@ Other ops: \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
 \code{\link{op_custom_gradient}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_depthwise_conv}()} \cr
 \code{\link{op_det}()} \cr
 \code{\link{op_diag}()} \cr
@@ -289,10 +307,13 @@ Other ops: \cr
 \code{\link{op_glu}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
 \code{\link{op_hard_shrink}()} \cr
 \code{\link{op_hard_sigmoid}()} \cr
 \code{\link{op_hard_silu}()} \cr
 \code{\link{op_hard_tanh}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -319,6 +340,8 @@ Other ops: \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
 \code{\link{op_istft}()} \cr
+\code{\link{op_kaiser}()} \cr
+\code{\link{op_layer_normalization}()} \cr
 \code{\link{op_leaky_relu}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
@@ -418,6 +441,7 @@ Other ops: \cr
 \code{\link{op_sort}()} \cr
 \code{\link{op_sparse_categorical_crossentropy}()} \cr
 \code{\link{op_sparse_plus}()} \cr
+\code{\link{op_sparse_sigmoid}()} \cr
 \code{\link{op_sparsemax}()} \cr
 \code{\link{op_split}()} \cr
 \code{\link{op_sqrt}()} \cr
@@ -455,6 +479,8 @@ Other ops: \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
 \code{\link{op_vectorized_map}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_while_loop}()} \cr
diff --git a/man/op_cast.Rd b/man/op_cast.Rd
index 0abb46c0e..4fc9cb49a 100644
--- a/man/op_cast.Rd
+++ b/man/op_cast.Rd
@@ -67,6 +67,7 @@ Other ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -85,6 +86,7 @@ Other ops: \cr
 \code{\link{op_associative_scan}()} \cr
 \code{\link{op_average}()} \cr
 \code{\link{op_average_pool}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_batch_normalization}()} \cr
 \code{\link{op_binary_crossentropy}()} \cr
 \code{\link{op_bincount}()} \cr
@@ -95,8 +97,10 @@ Other ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
 \code{\link{op_categorical_crossentropy}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_celu}()} \cr
 \code{\link{op_cholesky}()} \cr
@@ -109,6 +113,7 @@ Other ops: \cr
 \code{\link{op_convert_to_numpy}()} \cr
 \code{\link{op_convert_to_tensor}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -119,6 +124,7 @@ Other ops: \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
 \code{\link{op_custom_gradient}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_depthwise_conv}()} \cr
 \code{\link{op_det}()} \cr
 \code{\link{op_diag}()} \cr
@@ -158,10 +164,13 @@ Other ops: \cr
 \code{\link{op_glu}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
 \code{\link{op_hard_shrink}()} \cr
 \code{\link{op_hard_sigmoid}()} \cr
 \code{\link{op_hard_silu}()} \cr
 \code{\link{op_hard_tanh}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -188,6 +197,8 @@ Other ops: \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
 \code{\link{op_istft}()} \cr
+\code{\link{op_kaiser}()} \cr
+\code{\link{op_layer_normalization}()} \cr
 \code{\link{op_leaky_relu}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
@@ -287,6 +298,7 @@ Other ops: \cr
 \code{\link{op_sort}()} \cr
 \code{\link{op_sparse_categorical_crossentropy}()} \cr
 \code{\link{op_sparse_plus}()} \cr
+\code{\link{op_sparse_sigmoid}()} \cr
 \code{\link{op_sparsemax}()} \cr
 \code{\link{op_split}()} \cr
 \code{\link{op_sqrt}()} \cr
@@ -324,6 +336,8 @@ Other ops: \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
 \code{\link{op_vectorized_map}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_while_loop}()} \cr
diff --git a/man/op_categorical_crossentropy.Rd b/man/op_categorical_crossentropy.Rd
index 00432aade..5b7e04707 100644
--- a/man/op_categorical_crossentropy.Rd
+++ b/man/op_categorical_crossentropy.Rd
@@ -73,6 +73,7 @@ Other nn ops: \cr
 \code{\link{op_hard_sigmoid}()} \cr
 \code{\link{op_hard_silu}()} \cr
 \code{\link{op_hard_tanh}()} \cr
+\code{\link{op_layer_normalization}()} \cr
 \code{\link{op_leaky_relu}()} \cr
 \code{\link{op_log_sigmoid}()} \cr
 \code{\link{op_log_softmax}()} \cr
@@ -96,6 +97,7 @@ Other nn ops: \cr
 \code{\link{op_softsign}()} \cr
 \code{\link{op_sparse_categorical_crossentropy}()} \cr
 \code{\link{op_sparse_plus}()} \cr
+\code{\link{op_sparse_sigmoid}()} \cr
 \code{\link{op_sparsemax}()} \cr
 \code{\link{op_squareplus}()} \cr
 \code{\link{op_tanh_shrink}()} \cr
@@ -106,6 +108,7 @@ Other ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -124,6 +127,7 @@ Other ops: \cr
 \code{\link{op_associative_scan}()} \cr
 \code{\link{op_average}()} \cr
 \code{\link{op_average_pool}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_batch_normalization}()} \cr
 \code{\link{op_binary_crossentropy}()} \cr
 \code{\link{op_bincount}()} \cr
@@ -134,8 +138,10 @@ Other ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
 \code{\link{op_cast}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_celu}()} \cr
 \code{\link{op_cholesky}()} \cr
@@ -148,6 +154,7 @@ Other ops: \cr
 \code{\link{op_convert_to_numpy}()} \cr
 \code{\link{op_convert_to_tensor}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -158,6 +165,7 @@ Other ops: \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
 \code{\link{op_custom_gradient}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_depthwise_conv}()} \cr
 \code{\link{op_det}()} \cr
 \code{\link{op_diag}()} \cr
@@ -197,10 +205,13 @@ Other ops: \cr
 \code{\link{op_glu}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
 \code{\link{op_hard_shrink}()} \cr
 \code{\link{op_hard_sigmoid}()} \cr
 \code{\link{op_hard_silu}()} \cr
 \code{\link{op_hard_tanh}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -227,6 +238,8 @@ Other ops: \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
 \code{\link{op_istft}()} \cr
+\code{\link{op_kaiser}()} \cr
+\code{\link{op_layer_normalization}()} \cr
 \code{\link{op_leaky_relu}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
@@ -326,6 +339,7 @@ Other ops: \cr
 \code{\link{op_sort}()} \cr
 \code{\link{op_sparse_categorical_crossentropy}()} \cr
 \code{\link{op_sparse_plus}()} \cr
+\code{\link{op_sparse_sigmoid}()} \cr
 \code{\link{op_sparsemax}()} \cr
 \code{\link{op_split}()} \cr
 \code{\link{op_sqrt}()} \cr
@@ -363,6 +377,8 @@ Other ops: \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
 \code{\link{op_vectorized_map}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_while_loop}()} \cr
diff --git a/man/op_cbrt.Rd b/man/op_cbrt.Rd
new file mode 100644
index 000000000..a023a3679
--- /dev/null
+++ b/man/op_cbrt.Rd
@@ -0,0 +1,480 @@
+% Generated by roxygen2: do not edit by hand
+% Please edit documentation in R/ops-numpy.R
+\name{op_cbrt}
+\alias{op_cbrt}
+\title{Computes the cube root of the input tensor, element-wise.}
+\usage{
+op_cbrt(x)
+}
+\arguments{
+\item{x}{Input tensor.Examples
+
+\if{html}{\out{<div class="sourceCode r">}}\preformatted{op_cbrt(c(-8, 0, 8))
+}\if{html}{\out{</div>}}
+
+\if{html}{\out{<div class="sourceCode">}}\preformatted{## tf.Tensor([-2.  0.  2.], shape=(3), dtype=float32)
+
+}\if{html}{\out{</div>}}}
+}
+\value{
+A tensor containing the cube root of each element in \code{x}.
+}
+\description{
+Returns the real-valued cube root of \code{x}, handling negative inputs in the
+real domain.
+}
+\seealso{
+Other numpy ops: \cr
+\code{\link{op_abs}()} \cr
+\code{\link{op_add}()} \cr
+\code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
+\code{\link{op_any}()} \cr
+\code{\link{op_append}()} \cr
+\code{\link{op_arange}()} \cr
+\code{\link{op_arccos}()} \cr
+\code{\link{op_arccosh}()} \cr
+\code{\link{op_arcsin}()} \cr
+\code{\link{op_arcsinh}()} \cr
+\code{\link{op_arctan}()} \cr
+\code{\link{op_arctan2}()} \cr
+\code{\link{op_arctanh}()} \cr
+\code{\link{op_argmax}()} \cr
+\code{\link{op_argmin}()} \cr
+\code{\link{op_argpartition}()} \cr
+\code{\link{op_argsort}()} \cr
+\code{\link{op_array}()} \cr
+\code{\link{op_average}()} \cr
+\code{\link{op_bartlett}()} \cr
+\code{\link{op_bincount}()} \cr
+\code{\link{op_bitwise_and}()} \cr
+\code{\link{op_bitwise_invert}()} \cr
+\code{\link{op_bitwise_left_shift}()} \cr
+\code{\link{op_bitwise_not}()} \cr
+\code{\link{op_bitwise_or}()} \cr
+\code{\link{op_bitwise_right_shift}()} \cr
+\code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
+\code{\link{op_broadcast_to}()} \cr
+\code{\link{op_ceil}()} \cr
+\code{\link{op_clip}()} \cr
+\code{\link{op_concatenate}()} \cr
+\code{\link{op_conj}()} \cr
+\code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
+\code{\link{op_correlate}()} \cr
+\code{\link{op_cos}()} \cr
+\code{\link{op_cosh}()} \cr
+\code{\link{op_count_nonzero}()} \cr
+\code{\link{op_cross}()} \cr
+\code{\link{op_ctc_decode}()} \cr
+\code{\link{op_cumprod}()} \cr
+\code{\link{op_cumsum}()} \cr
+\code{\link{op_deg2rad}()} \cr
+\code{\link{op_diag}()} \cr
+\code{\link{op_diagflat}()} \cr
+\code{\link{op_diagonal}()} \cr
+\code{\link{op_diff}()} \cr
+\code{\link{op_digitize}()} \cr
+\code{\link{op_divide}()} \cr
+\code{\link{op_divide_no_nan}()} \cr
+\code{\link{op_dot}()} \cr
+\code{\link{op_einsum}()} \cr
+\code{\link{op_empty}()} \cr
+\code{\link{op_equal}()} \cr
+\code{\link{op_exp}()} \cr
+\code{\link{op_exp2}()} \cr
+\code{\link{op_expand_dims}()} \cr
+\code{\link{op_expm1}()} \cr
+\code{\link{op_eye}()} \cr
+\code{\link{op_flip}()} \cr
+\code{\link{op_floor}()} \cr
+\code{\link{op_floor_divide}()} \cr
+\code{\link{op_full}()} \cr
+\code{\link{op_full_like}()} \cr
+\code{\link{op_get_item}()} \cr
+\code{\link{op_greater}()} \cr
+\code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
+\code{\link{op_heaviside}()} \cr
+\code{\link{op_histogram}()} \cr
+\code{\link{op_hstack}()} \cr
+\code{\link{op_identity}()} \cr
+\code{\link{op_imag}()} \cr
+\code{\link{op_inner}()} \cr
+\code{\link{op_isclose}()} \cr
+\code{\link{op_isfinite}()} \cr
+\code{\link{op_isinf}()} \cr
+\code{\link{op_isnan}()} \cr
+\code{\link{op_kaiser}()} \cr
+\code{\link{op_left_shift}()} \cr
+\code{\link{op_less}()} \cr
+\code{\link{op_less_equal}()} \cr
+\code{\link{op_linspace}()} \cr
+\code{\link{op_log}()} \cr
+\code{\link{op_log10}()} \cr
+\code{\link{op_log1p}()} \cr
+\code{\link{op_log2}()} \cr
+\code{\link{op_logaddexp}()} \cr
+\code{\link{op_logdet}()} \cr
+\code{\link{op_logical_and}()} \cr
+\code{\link{op_logical_not}()} \cr
+\code{\link{op_logical_or}()} \cr
+\code{\link{op_logical_xor}()} \cr
+\code{\link{op_logspace}()} \cr
+\code{\link{op_lstsq}()} \cr
+\code{\link{op_matmul}()} \cr
+\code{\link{op_max}()} \cr
+\code{\link{op_maximum}()} \cr
+\code{\link{op_mean}()} \cr
+\code{\link{op_median}()} \cr
+\code{\link{op_meshgrid}()} \cr
+\code{\link{op_min}()} \cr
+\code{\link{op_minimum}()} \cr
+\code{\link{op_mod}()} \cr
+\code{\link{op_moveaxis}()} \cr
+\code{\link{op_multiply}()} \cr
+\code{\link{op_nan_to_num}()} \cr
+\code{\link{op_ndim}()} \cr
+\code{\link{op_negative}()} \cr
+\code{\link{op_nonzero}()} \cr
+\code{\link{op_not_equal}()} \cr
+\code{\link{op_ones}()} \cr
+\code{\link{op_ones_like}()} \cr
+\code{\link{op_outer}()} \cr
+\code{\link{op_pad}()} \cr
+\code{\link{op_power}()} \cr
+\code{\link{op_prod}()} \cr
+\code{\link{op_quantile}()} \cr
+\code{\link{op_ravel}()} \cr
+\code{\link{op_real}()} \cr
+\code{\link{op_reciprocal}()} \cr
+\code{\link{op_repeat}()} \cr
+\code{\link{op_reshape}()} \cr
+\code{\link{op_right_shift}()} \cr
+\code{\link{op_roll}()} \cr
+\code{\link{op_rot90}()} \cr
+\code{\link{op_round}()} \cr
+\code{\link{op_saturate_cast}()} \cr
+\code{\link{op_select}()} \cr
+\code{\link{op_sign}()} \cr
+\code{\link{op_signbit}()} \cr
+\code{\link{op_sin}()} \cr
+\code{\link{op_sinh}()} \cr
+\code{\link{op_size}()} \cr
+\code{\link{op_sort}()} \cr
+\code{\link{op_split}()} \cr
+\code{\link{op_sqrt}()} \cr
+\code{\link{op_square}()} \cr
+\code{\link{op_squeeze}()} \cr
+\code{\link{op_stack}()} \cr
+\code{\link{op_std}()} \cr
+\code{\link{op_subtract}()} \cr
+\code{\link{op_sum}()} \cr
+\code{\link{op_swapaxes}()} \cr
+\code{\link{op_take}()} \cr
+\code{\link{op_take_along_axis}()} \cr
+\code{\link{op_tan}()} \cr
+\code{\link{op_tanh}()} \cr
+\code{\link{op_tensordot}()} \cr
+\code{\link{op_tile}()} \cr
+\code{\link{op_trace}()} \cr
+\code{\link{op_transpose}()} \cr
+\code{\link{op_tri}()} \cr
+\code{\link{op_tril}()} \cr
+\code{\link{op_triu}()} \cr
+\code{\link{op_trunc}()} \cr
+\code{\link{op_var}()} \cr
+\code{\link{op_vdot}()} \cr
+\code{\link{op_vectorize}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
+\code{\link{op_vstack}()} \cr
+\code{\link{op_where}()} \cr
+\code{\link{op_zeros}()} \cr
+\code{\link{op_zeros_like}()} \cr
+
+Other ops: \cr
+\code{\link{op_abs}()} \cr
+\code{\link{op_add}()} \cr
+\code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
+\code{\link{op_any}()} \cr
+\code{\link{op_append}()} \cr
+\code{\link{op_arange}()} \cr
+\code{\link{op_arccos}()} \cr
+\code{\link{op_arccosh}()} \cr
+\code{\link{op_arcsin}()} \cr
+\code{\link{op_arcsinh}()} \cr
+\code{\link{op_arctan}()} \cr
+\code{\link{op_arctan2}()} \cr
+\code{\link{op_arctanh}()} \cr
+\code{\link{op_argmax}()} \cr
+\code{\link{op_argmin}()} \cr
+\code{\link{op_argpartition}()} \cr
+\code{\link{op_argsort}()} \cr
+\code{\link{op_array}()} \cr
+\code{\link{op_associative_scan}()} \cr
+\code{\link{op_average}()} \cr
+\code{\link{op_average_pool}()} \cr
+\code{\link{op_bartlett}()} \cr
+\code{\link{op_batch_normalization}()} \cr
+\code{\link{op_binary_crossentropy}()} \cr
+\code{\link{op_bincount}()} \cr
+\code{\link{op_bitwise_and}()} \cr
+\code{\link{op_bitwise_invert}()} \cr
+\code{\link{op_bitwise_left_shift}()} \cr
+\code{\link{op_bitwise_not}()} \cr
+\code{\link{op_bitwise_or}()} \cr
+\code{\link{op_bitwise_right_shift}()} \cr
+\code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
+\code{\link{op_broadcast_to}()} \cr
+\code{\link{op_cast}()} \cr
+\code{\link{op_categorical_crossentropy}()} \cr
+\code{\link{op_ceil}()} \cr
+\code{\link{op_celu}()} \cr
+\code{\link{op_cholesky}()} \cr
+\code{\link{op_clip}()} \cr
+\code{\link{op_concatenate}()} \cr
+\code{\link{op_cond}()} \cr
+\code{\link{op_conj}()} \cr
+\code{\link{op_conv}()} \cr
+\code{\link{op_conv_transpose}()} \cr
+\code{\link{op_convert_to_numpy}()} \cr
+\code{\link{op_convert_to_tensor}()} \cr
+\code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
+\code{\link{op_correlate}()} \cr
+\code{\link{op_cos}()} \cr
+\code{\link{op_cosh}()} \cr
+\code{\link{op_count_nonzero}()} \cr
+\code{\link{op_cross}()} \cr
+\code{\link{op_ctc_decode}()} \cr
+\code{\link{op_ctc_loss}()} \cr
+\code{\link{op_cumprod}()} \cr
+\code{\link{op_cumsum}()} \cr
+\code{\link{op_custom_gradient}()} \cr
+\code{\link{op_deg2rad}()} \cr
+\code{\link{op_depthwise_conv}()} \cr
+\code{\link{op_det}()} \cr
+\code{\link{op_diag}()} \cr
+\code{\link{op_diagflat}()} \cr
+\code{\link{op_diagonal}()} \cr
+\code{\link{op_diff}()} \cr
+\code{\link{op_digitize}()} \cr
+\code{\link{op_divide}()} \cr
+\code{\link{op_divide_no_nan}()} \cr
+\code{\link{op_dot}()} \cr
+\code{\link{op_dot_product_attention}()} \cr
+\code{\link{op_dtype}()} \cr
+\code{\link{op_eig}()} \cr
+\code{\link{op_eigh}()} \cr
+\code{\link{op_einsum}()} \cr
+\code{\link{op_elu}()} \cr
+\code{\link{op_empty}()} \cr
+\code{\link{op_equal}()} \cr
+\code{\link{op_erf}()} \cr
+\code{\link{op_erfinv}()} \cr
+\code{\link{op_exp}()} \cr
+\code{\link{op_exp2}()} \cr
+\code{\link{op_expand_dims}()} \cr
+\code{\link{op_expm1}()} \cr
+\code{\link{op_extract_sequences}()} \cr
+\code{\link{op_eye}()} \cr
+\code{\link{op_fft}()} \cr
+\code{\link{op_fft2}()} \cr
+\code{\link{op_flip}()} \cr
+\code{\link{op_floor}()} \cr
+\code{\link{op_floor_divide}()} \cr
+\code{\link{op_fori_loop}()} \cr
+\code{\link{op_full}()} \cr
+\code{\link{op_full_like}()} \cr
+\code{\link{op_gelu}()} \cr
+\code{\link{op_get_item}()} \cr
+\code{\link{op_glu}()} \cr
+\code{\link{op_greater}()} \cr
+\code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
+\code{\link{op_hard_shrink}()} \cr
+\code{\link{op_hard_sigmoid}()} \cr
+\code{\link{op_hard_silu}()} \cr
+\code{\link{op_hard_tanh}()} \cr
+\code{\link{op_heaviside}()} \cr
+\code{\link{op_histogram}()} \cr
+\code{\link{op_hstack}()} \cr
+\code{\link{op_identity}()} \cr
+\code{\link{op_ifft2}()} \cr
+\code{\link{op_imag}()} \cr
+\code{\link{op_image_affine_transform}()} \cr
+\code{\link{op_image_crop}()} \cr
+\code{\link{op_image_extract_patches}()} \cr
+\code{\link{op_image_gaussian_blur}()} \cr
+\code{\link{op_image_hsv_to_rgb}()} \cr
+\code{\link{op_image_map_coordinates}()} \cr
+\code{\link{op_image_pad}()} \cr
+\code{\link{op_image_perspective_transform}()} \cr
+\code{\link{op_image_resize}()} \cr
+\code{\link{op_image_rgb_to_grayscale}()} \cr
+\code{\link{op_image_rgb_to_hsv}()} \cr
+\code{\link{op_in_top_k}()} \cr
+\code{\link{op_inner}()} \cr
+\code{\link{op_inv}()} \cr
+\code{\link{op_irfft}()} \cr
+\code{\link{op_is_tensor}()} \cr
+\code{\link{op_isclose}()} \cr
+\code{\link{op_isfinite}()} \cr
+\code{\link{op_isinf}()} \cr
+\code{\link{op_isnan}()} \cr
+\code{\link{op_istft}()} \cr
+\code{\link{op_kaiser}()} \cr
+\code{\link{op_layer_normalization}()} \cr
+\code{\link{op_leaky_relu}()} \cr
+\code{\link{op_left_shift}()} \cr
+\code{\link{op_less}()} \cr
+\code{\link{op_less_equal}()} \cr
+\code{\link{op_linspace}()} \cr
+\code{\link{op_log}()} \cr
+\code{\link{op_log10}()} \cr
+\code{\link{op_log1p}()} \cr
+\code{\link{op_log2}()} \cr
+\code{\link{op_log_sigmoid}()} \cr
+\code{\link{op_log_softmax}()} \cr
+\code{\link{op_logaddexp}()} \cr
+\code{\link{op_logdet}()} \cr
+\code{\link{op_logical_and}()} \cr
+\code{\link{op_logical_not}()} \cr
+\code{\link{op_logical_or}()} \cr
+\code{\link{op_logical_xor}()} \cr
+\code{\link{op_logspace}()} \cr
+\code{\link{op_logsumexp}()} \cr
+\code{\link{op_lstsq}()} \cr
+\code{\link{op_lu_factor}()} \cr
+\code{\link{op_map}()} \cr
+\code{\link{op_matmul}()} \cr
+\code{\link{op_max}()} \cr
+\code{\link{op_max_pool}()} \cr
+\code{\link{op_maximum}()} \cr
+\code{\link{op_mean}()} \cr
+\code{\link{op_median}()} \cr
+\code{\link{op_meshgrid}()} \cr
+\code{\link{op_min}()} \cr
+\code{\link{op_minimum}()} \cr
+\code{\link{op_mod}()} \cr
+\code{\link{op_moments}()} \cr
+\code{\link{op_moveaxis}()} \cr
+\code{\link{op_multi_hot}()} \cr
+\code{\link{op_multiply}()} \cr
+\code{\link{op_nan_to_num}()} \cr
+\code{\link{op_ndim}()} \cr
+\code{\link{op_negative}()} \cr
+\code{\link{op_nonzero}()} \cr
+\code{\link{op_norm}()} \cr
+\code{\link{op_normalize}()} \cr
+\code{\link{op_not_equal}()} \cr
+\code{\link{op_one_hot}()} \cr
+\code{\link{op_ones}()} \cr
+\code{\link{op_ones_like}()} \cr
+\code{\link{op_outer}()} \cr
+\code{\link{op_pad}()} \cr
+\code{\link{op_polar}()} \cr
+\code{\link{op_power}()} \cr
+\code{\link{op_prod}()} \cr
+\code{\link{op_psnr}()} \cr
+\code{\link{op_qr}()} \cr
+\code{\link{op_quantile}()} \cr
+\code{\link{op_ravel}()} \cr
+\code{\link{op_real}()} \cr
+\code{\link{op_rearrange}()} \cr
+\code{\link{op_reciprocal}()} \cr
+\code{\link{op_relu}()} \cr
+\code{\link{op_relu6}()} \cr
+\code{\link{op_repeat}()} \cr
+\code{\link{op_reshape}()} \cr
+\code{\link{op_rfft}()} \cr
+\code{\link{op_right_shift}()} \cr
+\code{\link{op_rms_normalization}()} \cr
+\code{\link{op_roll}()} \cr
+\code{\link{op_rot90}()} \cr
+\code{\link{op_round}()} \cr
+\code{\link{op_rsqrt}()} \cr
+\code{\link{op_saturate_cast}()} \cr
+\code{\link{op_scan}()} \cr
+\code{\link{op_scatter}()} \cr
+\code{\link{op_scatter_update}()} \cr
+\code{\link{op_searchsorted}()} \cr
+\code{\link{op_segment_max}()} \cr
+\code{\link{op_segment_sum}()} \cr
+\code{\link{op_select}()} \cr
+\code{\link{op_selu}()} \cr
+\code{\link{op_separable_conv}()} \cr
+\code{\link{op_shape}()} \cr
+\code{\link{op_sigmoid}()} \cr
+\code{\link{op_sign}()} \cr
+\code{\link{op_signbit}()} \cr
+\code{\link{op_silu}()} \cr
+\code{\link{op_sin}()} \cr
+\code{\link{op_sinh}()} \cr
+\code{\link{op_size}()} \cr
+\code{\link{op_slice}()} \cr
+\code{\link{op_slice_update}()} \cr
+\code{\link{op_slogdet}()} \cr
+\code{\link{op_soft_shrink}()} \cr
+\code{\link{op_softmax}()} \cr
+\code{\link{op_softplus}()} \cr
+\code{\link{op_softsign}()} \cr
+\code{\link{op_solve}()} \cr
+\code{\link{op_solve_triangular}()} \cr
+\code{\link{op_sort}()} \cr
+\code{\link{op_sparse_categorical_crossentropy}()} \cr
+\code{\link{op_sparse_plus}()} \cr
+\code{\link{op_sparse_sigmoid}()} \cr
+\code{\link{op_sparsemax}()} \cr
+\code{\link{op_split}()} \cr
+\code{\link{op_sqrt}()} \cr
+\code{\link{op_square}()} \cr
+\code{\link{op_squareplus}()} \cr
+\code{\link{op_squeeze}()} \cr
+\code{\link{op_stack}()} \cr
+\code{\link{op_std}()} \cr
+\code{\link{op_stft}()} \cr
+\code{\link{op_stop_gradient}()} \cr
+\code{\link{op_subset}()} \cr
+\code{\link{op_subtract}()} \cr
+\code{\link{op_sum}()} \cr
+\code{\link{op_svd}()} \cr
+\code{\link{op_swapaxes}()} \cr
+\code{\link{op_switch}()} \cr
+\code{\link{op_take}()} \cr
+\code{\link{op_take_along_axis}()} \cr
+\code{\link{op_tan}()} \cr
+\code{\link{op_tanh}()} \cr
+\code{\link{op_tanh_shrink}()} \cr
+\code{\link{op_tensordot}()} \cr
+\code{\link{op_threshold}()} \cr
+\code{\link{op_tile}()} \cr
+\code{\link{op_top_k}()} \cr
+\code{\link{op_trace}()} \cr
+\code{\link{op_transpose}()} \cr
+\code{\link{op_tri}()} \cr
+\code{\link{op_tril}()} \cr
+\code{\link{op_triu}()} \cr
+\code{\link{op_trunc}()} \cr
+\code{\link{op_unravel_index}()} \cr
+\code{\link{op_unstack}()} \cr
+\code{\link{op_var}()} \cr
+\code{\link{op_vdot}()} \cr
+\code{\link{op_vectorize}()} \cr
+\code{\link{op_vectorized_map}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
+\code{\link{op_vstack}()} \cr
+\code{\link{op_where}()} \cr
+\code{\link{op_while_loop}()} \cr
+\code{\link{op_zeros}()} \cr
+\code{\link{op_zeros_like}()} \cr
+}
+\concept{numpy ops}
+\concept{ops}
diff --git a/man/op_ceil.Rd b/man/op_ceil.Rd
index f5a0d9054..fc15e20d3 100644
--- a/man/op_ceil.Rd
+++ b/man/op_ceil.Rd
@@ -25,6 +25,7 @@ Other numpy ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -41,6 +42,7 @@ Other numpy ops: \cr
 \code{\link{op_argsort}()} \cr
 \code{\link{op_array}()} \cr
 \code{\link{op_average}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_bincount}()} \cr
 \code{\link{op_bitwise_and}()} \cr
 \code{\link{op_bitwise_invert}()} \cr
@@ -49,11 +51,14 @@ Other numpy ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_clip}()} \cr
 \code{\link{op_concatenate}()} \cr
 \code{\link{op_conj}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -62,6 +67,7 @@ Other numpy ops: \cr
 \code{\link{op_ctc_decode}()} \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_diag}()} \cr
 \code{\link{op_diagflat}()} \cr
 \code{\link{op_diagonal}()} \cr
@@ -86,6 +92,9 @@ Other numpy ops: \cr
 \code{\link{op_get_item}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -95,6 +104,7 @@ Other numpy ops: \cr
 \code{\link{op_isfinite}()} \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
+\code{\link{op_kaiser}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
 \code{\link{op_less_equal}()} \cr
@@ -175,6 +185,8 @@ Other numpy ops: \cr
 \code{\link{op_var}()} \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_zeros}()} \cr
@@ -184,6 +196,7 @@ Other ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -202,6 +215,7 @@ Other ops: \cr
 \code{\link{op_associative_scan}()} \cr
 \code{\link{op_average}()} \cr
 \code{\link{op_average_pool}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_batch_normalization}()} \cr
 \code{\link{op_binary_crossentropy}()} \cr
 \code{\link{op_bincount}()} \cr
@@ -212,9 +226,11 @@ Other ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
 \code{\link{op_cast}()} \cr
 \code{\link{op_categorical_crossentropy}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_celu}()} \cr
 \code{\link{op_cholesky}()} \cr
 \code{\link{op_clip}()} \cr
@@ -226,6 +242,7 @@ Other ops: \cr
 \code{\link{op_convert_to_numpy}()} \cr
 \code{\link{op_convert_to_tensor}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -236,6 +253,7 @@ Other ops: \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
 \code{\link{op_custom_gradient}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_depthwise_conv}()} \cr
 \code{\link{op_det}()} \cr
 \code{\link{op_diag}()} \cr
@@ -275,10 +293,13 @@ Other ops: \cr
 \code{\link{op_glu}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
 \code{\link{op_hard_shrink}()} \cr
 \code{\link{op_hard_sigmoid}()} \cr
 \code{\link{op_hard_silu}()} \cr
 \code{\link{op_hard_tanh}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -305,6 +326,8 @@ Other ops: \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
 \code{\link{op_istft}()} \cr
+\code{\link{op_kaiser}()} \cr
+\code{\link{op_layer_normalization}()} \cr
 \code{\link{op_leaky_relu}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
@@ -404,6 +427,7 @@ Other ops: \cr
 \code{\link{op_sort}()} \cr
 \code{\link{op_sparse_categorical_crossentropy}()} \cr
 \code{\link{op_sparse_plus}()} \cr
+\code{\link{op_sparse_sigmoid}()} \cr
 \code{\link{op_sparsemax}()} \cr
 \code{\link{op_split}()} \cr
 \code{\link{op_sqrt}()} \cr
@@ -441,6 +465,8 @@ Other ops: \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
 \code{\link{op_vectorized_map}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_while_loop}()} \cr
diff --git a/man/op_celu.Rd b/man/op_celu.Rd
index ffbd94e36..94c5e5d67 100644
--- a/man/op_celu.Rd
+++ b/man/op_celu.Rd
@@ -47,6 +47,7 @@ Other nn ops: \cr
 \code{\link{op_hard_sigmoid}()} \cr
 \code{\link{op_hard_silu}()} \cr
 \code{\link{op_hard_tanh}()} \cr
+\code{\link{op_layer_normalization}()} \cr
 \code{\link{op_leaky_relu}()} \cr
 \code{\link{op_log_sigmoid}()} \cr
 \code{\link{op_log_softmax}()} \cr
@@ -70,6 +71,7 @@ Other nn ops: \cr
 \code{\link{op_softsign}()} \cr
 \code{\link{op_sparse_categorical_crossentropy}()} \cr
 \code{\link{op_sparse_plus}()} \cr
+\code{\link{op_sparse_sigmoid}()} \cr
 \code{\link{op_sparsemax}()} \cr
 \code{\link{op_squareplus}()} \cr
 \code{\link{op_tanh_shrink}()} \cr
@@ -80,6 +82,7 @@ Other ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -98,6 +101,7 @@ Other ops: \cr
 \code{\link{op_associative_scan}()} \cr
 \code{\link{op_average}()} \cr
 \code{\link{op_average_pool}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_batch_normalization}()} \cr
 \code{\link{op_binary_crossentropy}()} \cr
 \code{\link{op_bincount}()} \cr
@@ -108,9 +112,11 @@ Other ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
 \code{\link{op_cast}()} \cr
 \code{\link{op_categorical_crossentropy}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_cholesky}()} \cr
 \code{\link{op_clip}()} \cr
@@ -122,6 +128,7 @@ Other ops: \cr
 \code{\link{op_convert_to_numpy}()} \cr
 \code{\link{op_convert_to_tensor}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -132,6 +139,7 @@ Other ops: \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
 \code{\link{op_custom_gradient}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_depthwise_conv}()} \cr
 \code{\link{op_det}()} \cr
 \code{\link{op_diag}()} \cr
@@ -171,10 +179,13 @@ Other ops: \cr
 \code{\link{op_glu}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
 \code{\link{op_hard_shrink}()} \cr
 \code{\link{op_hard_sigmoid}()} \cr
 \code{\link{op_hard_silu}()} \cr
 \code{\link{op_hard_tanh}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -201,6 +212,8 @@ Other ops: \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
 \code{\link{op_istft}()} \cr
+\code{\link{op_kaiser}()} \cr
+\code{\link{op_layer_normalization}()} \cr
 \code{\link{op_leaky_relu}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
@@ -300,6 +313,7 @@ Other ops: \cr
 \code{\link{op_sort}()} \cr
 \code{\link{op_sparse_categorical_crossentropy}()} \cr
 \code{\link{op_sparse_plus}()} \cr
+\code{\link{op_sparse_sigmoid}()} \cr
 \code{\link{op_sparsemax}()} \cr
 \code{\link{op_split}()} \cr
 \code{\link{op_sqrt}()} \cr
@@ -337,6 +351,8 @@ Other ops: \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
 \code{\link{op_vectorized_map}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_while_loop}()} \cr
diff --git a/man/op_cholesky.Rd b/man/op_cholesky.Rd
index fbe05de83..3f47299bc 100644
--- a/man/op_cholesky.Rd
+++ b/man/op_cholesky.Rd
@@ -33,6 +33,7 @@ Other ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -51,6 +52,7 @@ Other ops: \cr
 \code{\link{op_associative_scan}()} \cr
 \code{\link{op_average}()} \cr
 \code{\link{op_average_pool}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_batch_normalization}()} \cr
 \code{\link{op_binary_crossentropy}()} \cr
 \code{\link{op_bincount}()} \cr
@@ -61,9 +63,11 @@ Other ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
 \code{\link{op_cast}()} \cr
 \code{\link{op_categorical_crossentropy}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_celu}()} \cr
 \code{\link{op_clip}()} \cr
@@ -75,6 +79,7 @@ Other ops: \cr
 \code{\link{op_convert_to_numpy}()} \cr
 \code{\link{op_convert_to_tensor}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -85,6 +90,7 @@ Other ops: \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
 \code{\link{op_custom_gradient}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_depthwise_conv}()} \cr
 \code{\link{op_det}()} \cr
 \code{\link{op_diag}()} \cr
@@ -124,10 +130,13 @@ Other ops: \cr
 \code{\link{op_glu}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
 \code{\link{op_hard_shrink}()} \cr
 \code{\link{op_hard_sigmoid}()} \cr
 \code{\link{op_hard_silu}()} \cr
 \code{\link{op_hard_tanh}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -154,6 +163,8 @@ Other ops: \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
 \code{\link{op_istft}()} \cr
+\code{\link{op_kaiser}()} \cr
+\code{\link{op_layer_normalization}()} \cr
 \code{\link{op_leaky_relu}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
@@ -253,6 +264,7 @@ Other ops: \cr
 \code{\link{op_sort}()} \cr
 \code{\link{op_sparse_categorical_crossentropy}()} \cr
 \code{\link{op_sparse_plus}()} \cr
+\code{\link{op_sparse_sigmoid}()} \cr
 \code{\link{op_sparsemax}()} \cr
 \code{\link{op_split}()} \cr
 \code{\link{op_sqrt}()} \cr
@@ -290,6 +302,8 @@ Other ops: \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
 \code{\link{op_vectorized_map}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_while_loop}()} \cr
diff --git a/man/op_clip.Rd b/man/op_clip.Rd
index 918aa5378..847b54706 100644
--- a/man/op_clip.Rd
+++ b/man/op_clip.Rd
@@ -30,6 +30,7 @@ Other numpy ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -46,6 +47,7 @@ Other numpy ops: \cr
 \code{\link{op_argsort}()} \cr
 \code{\link{op_array}()} \cr
 \code{\link{op_average}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_bincount}()} \cr
 \code{\link{op_bitwise_and}()} \cr
 \code{\link{op_bitwise_invert}()} \cr
@@ -54,11 +56,14 @@ Other numpy ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_concatenate}()} \cr
 \code{\link{op_conj}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -67,6 +72,7 @@ Other numpy ops: \cr
 \code{\link{op_ctc_decode}()} \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_diag}()} \cr
 \code{\link{op_diagflat}()} \cr
 \code{\link{op_diagonal}()} \cr
@@ -91,6 +97,9 @@ Other numpy ops: \cr
 \code{\link{op_get_item}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -100,6 +109,7 @@ Other numpy ops: \cr
 \code{\link{op_isfinite}()} \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
+\code{\link{op_kaiser}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
 \code{\link{op_less_equal}()} \cr
@@ -180,6 +190,8 @@ Other numpy ops: \cr
 \code{\link{op_var}()} \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_zeros}()} \cr
@@ -189,6 +201,7 @@ Other ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -207,6 +220,7 @@ Other ops: \cr
 \code{\link{op_associative_scan}()} \cr
 \code{\link{op_average}()} \cr
 \code{\link{op_average_pool}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_batch_normalization}()} \cr
 \code{\link{op_binary_crossentropy}()} \cr
 \code{\link{op_bincount}()} \cr
@@ -217,9 +231,11 @@ Other ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
 \code{\link{op_cast}()} \cr
 \code{\link{op_categorical_crossentropy}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_celu}()} \cr
 \code{\link{op_cholesky}()} \cr
@@ -231,6 +247,7 @@ Other ops: \cr
 \code{\link{op_convert_to_numpy}()} \cr
 \code{\link{op_convert_to_tensor}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -241,6 +258,7 @@ Other ops: \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
 \code{\link{op_custom_gradient}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_depthwise_conv}()} \cr
 \code{\link{op_det}()} \cr
 \code{\link{op_diag}()} \cr
@@ -280,10 +298,13 @@ Other ops: \cr
 \code{\link{op_glu}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
 \code{\link{op_hard_shrink}()} \cr
 \code{\link{op_hard_sigmoid}()} \cr
 \code{\link{op_hard_silu}()} \cr
 \code{\link{op_hard_tanh}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -310,6 +331,8 @@ Other ops: \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
 \code{\link{op_istft}()} \cr
+\code{\link{op_kaiser}()} \cr
+\code{\link{op_layer_normalization}()} \cr
 \code{\link{op_leaky_relu}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
@@ -409,6 +432,7 @@ Other ops: \cr
 \code{\link{op_sort}()} \cr
 \code{\link{op_sparse_categorical_crossentropy}()} \cr
 \code{\link{op_sparse_plus}()} \cr
+\code{\link{op_sparse_sigmoid}()} \cr
 \code{\link{op_sparsemax}()} \cr
 \code{\link{op_split}()} \cr
 \code{\link{op_sqrt}()} \cr
@@ -446,6 +470,8 @@ Other ops: \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
 \code{\link{op_vectorized_map}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_while_loop}()} \cr
diff --git a/man/op_concatenate.Rd b/man/op_concatenate.Rd
index 23f416d3f..0bb067dba 100644
--- a/man/op_concatenate.Rd
+++ b/man/op_concatenate.Rd
@@ -26,6 +26,7 @@ Other numpy ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -42,6 +43,7 @@ Other numpy ops: \cr
 \code{\link{op_argsort}()} \cr
 \code{\link{op_array}()} \cr
 \code{\link{op_average}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_bincount}()} \cr
 \code{\link{op_bitwise_and}()} \cr
 \code{\link{op_bitwise_invert}()} \cr
@@ -50,11 +52,14 @@ Other numpy ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_clip}()} \cr
 \code{\link{op_conj}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -63,6 +68,7 @@ Other numpy ops: \cr
 \code{\link{op_ctc_decode}()} \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_diag}()} \cr
 \code{\link{op_diagflat}()} \cr
 \code{\link{op_diagonal}()} \cr
@@ -87,6 +93,9 @@ Other numpy ops: \cr
 \code{\link{op_get_item}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -96,6 +105,7 @@ Other numpy ops: \cr
 \code{\link{op_isfinite}()} \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
+\code{\link{op_kaiser}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
 \code{\link{op_less_equal}()} \cr
@@ -176,6 +186,8 @@ Other numpy ops: \cr
 \code{\link{op_var}()} \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_zeros}()} \cr
@@ -185,6 +197,7 @@ Other ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -203,6 +216,7 @@ Other ops: \cr
 \code{\link{op_associative_scan}()} \cr
 \code{\link{op_average}()} \cr
 \code{\link{op_average_pool}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_batch_normalization}()} \cr
 \code{\link{op_binary_crossentropy}()} \cr
 \code{\link{op_bincount}()} \cr
@@ -213,9 +227,11 @@ Other ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
 \code{\link{op_cast}()} \cr
 \code{\link{op_categorical_crossentropy}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_celu}()} \cr
 \code{\link{op_cholesky}()} \cr
@@ -227,6 +243,7 @@ Other ops: \cr
 \code{\link{op_convert_to_numpy}()} \cr
 \code{\link{op_convert_to_tensor}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -237,6 +254,7 @@ Other ops: \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
 \code{\link{op_custom_gradient}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_depthwise_conv}()} \cr
 \code{\link{op_det}()} \cr
 \code{\link{op_diag}()} \cr
@@ -276,10 +294,13 @@ Other ops: \cr
 \code{\link{op_glu}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
 \code{\link{op_hard_shrink}()} \cr
 \code{\link{op_hard_sigmoid}()} \cr
 \code{\link{op_hard_silu}()} \cr
 \code{\link{op_hard_tanh}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -306,6 +327,8 @@ Other ops: \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
 \code{\link{op_istft}()} \cr
+\code{\link{op_kaiser}()} \cr
+\code{\link{op_layer_normalization}()} \cr
 \code{\link{op_leaky_relu}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
@@ -405,6 +428,7 @@ Other ops: \cr
 \code{\link{op_sort}()} \cr
 \code{\link{op_sparse_categorical_crossentropy}()} \cr
 \code{\link{op_sparse_plus}()} \cr
+\code{\link{op_sparse_sigmoid}()} \cr
 \code{\link{op_sparsemax}()} \cr
 \code{\link{op_split}()} \cr
 \code{\link{op_sqrt}()} \cr
@@ -442,6 +466,8 @@ Other ops: \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
 \code{\link{op_vectorized_map}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_while_loop}()} \cr
diff --git a/man/op_cond.Rd b/man/op_cond.Rd
index 1f9ce3c9d..5513d5fec 100644
--- a/man/op_cond.Rd
+++ b/man/op_cond.Rd
@@ -109,6 +109,7 @@ Other ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -127,6 +128,7 @@ Other ops: \cr
 \code{\link{op_associative_scan}()} \cr
 \code{\link{op_average}()} \cr
 \code{\link{op_average_pool}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_batch_normalization}()} \cr
 \code{\link{op_binary_crossentropy}()} \cr
 \code{\link{op_bincount}()} \cr
@@ -137,9 +139,11 @@ Other ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
 \code{\link{op_cast}()} \cr
 \code{\link{op_categorical_crossentropy}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_celu}()} \cr
 \code{\link{op_cholesky}()} \cr
@@ -151,6 +155,7 @@ Other ops: \cr
 \code{\link{op_convert_to_numpy}()} \cr
 \code{\link{op_convert_to_tensor}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -161,6 +166,7 @@ Other ops: \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
 \code{\link{op_custom_gradient}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_depthwise_conv}()} \cr
 \code{\link{op_det}()} \cr
 \code{\link{op_diag}()} \cr
@@ -200,10 +206,13 @@ Other ops: \cr
 \code{\link{op_glu}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
 \code{\link{op_hard_shrink}()} \cr
 \code{\link{op_hard_sigmoid}()} \cr
 \code{\link{op_hard_silu}()} \cr
 \code{\link{op_hard_tanh}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -230,6 +239,8 @@ Other ops: \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
 \code{\link{op_istft}()} \cr
+\code{\link{op_kaiser}()} \cr
+\code{\link{op_layer_normalization}()} \cr
 \code{\link{op_leaky_relu}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
@@ -329,6 +340,7 @@ Other ops: \cr
 \code{\link{op_sort}()} \cr
 \code{\link{op_sparse_categorical_crossentropy}()} \cr
 \code{\link{op_sparse_plus}()} \cr
+\code{\link{op_sparse_sigmoid}()} \cr
 \code{\link{op_sparsemax}()} \cr
 \code{\link{op_split}()} \cr
 \code{\link{op_sqrt}()} \cr
@@ -366,6 +378,8 @@ Other ops: \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
 \code{\link{op_vectorized_map}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_while_loop}()} \cr
diff --git a/man/op_conj.Rd b/man/op_conj.Rd
index 7909b15d7..04ed2d193 100644
--- a/man/op_conj.Rd
+++ b/man/op_conj.Rd
@@ -25,6 +25,7 @@ Other numpy ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -41,6 +42,7 @@ Other numpy ops: \cr
 \code{\link{op_argsort}()} \cr
 \code{\link{op_array}()} \cr
 \code{\link{op_average}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_bincount}()} \cr
 \code{\link{op_bitwise_and}()} \cr
 \code{\link{op_bitwise_invert}()} \cr
@@ -49,11 +51,14 @@ Other numpy ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_clip}()} \cr
 \code{\link{op_concatenate}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -62,6 +67,7 @@ Other numpy ops: \cr
 \code{\link{op_ctc_decode}()} \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_diag}()} \cr
 \code{\link{op_diagflat}()} \cr
 \code{\link{op_diagonal}()} \cr
@@ -86,6 +92,9 @@ Other numpy ops: \cr
 \code{\link{op_get_item}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -95,6 +104,7 @@ Other numpy ops: \cr
 \code{\link{op_isfinite}()} \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
+\code{\link{op_kaiser}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
 \code{\link{op_less_equal}()} \cr
@@ -175,6 +185,8 @@ Other numpy ops: \cr
 \code{\link{op_var}()} \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_zeros}()} \cr
@@ -184,6 +196,7 @@ Other ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -202,6 +215,7 @@ Other ops: \cr
 \code{\link{op_associative_scan}()} \cr
 \code{\link{op_average}()} \cr
 \code{\link{op_average_pool}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_batch_normalization}()} \cr
 \code{\link{op_binary_crossentropy}()} \cr
 \code{\link{op_bincount}()} \cr
@@ -212,9 +226,11 @@ Other ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
 \code{\link{op_cast}()} \cr
 \code{\link{op_categorical_crossentropy}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_celu}()} \cr
 \code{\link{op_cholesky}()} \cr
@@ -226,6 +242,7 @@ Other ops: \cr
 \code{\link{op_convert_to_numpy}()} \cr
 \code{\link{op_convert_to_tensor}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -236,6 +253,7 @@ Other ops: \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
 \code{\link{op_custom_gradient}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_depthwise_conv}()} \cr
 \code{\link{op_det}()} \cr
 \code{\link{op_diag}()} \cr
@@ -275,10 +293,13 @@ Other ops: \cr
 \code{\link{op_glu}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
 \code{\link{op_hard_shrink}()} \cr
 \code{\link{op_hard_sigmoid}()} \cr
 \code{\link{op_hard_silu}()} \cr
 \code{\link{op_hard_tanh}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -305,6 +326,8 @@ Other ops: \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
 \code{\link{op_istft}()} \cr
+\code{\link{op_kaiser}()} \cr
+\code{\link{op_layer_normalization}()} \cr
 \code{\link{op_leaky_relu}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
@@ -404,6 +427,7 @@ Other ops: \cr
 \code{\link{op_sort}()} \cr
 \code{\link{op_sparse_categorical_crossentropy}()} \cr
 \code{\link{op_sparse_plus}()} \cr
+\code{\link{op_sparse_sigmoid}()} \cr
 \code{\link{op_sparsemax}()} \cr
 \code{\link{op_split}()} \cr
 \code{\link{op_sqrt}()} \cr
@@ -441,6 +465,8 @@ Other ops: \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
 \code{\link{op_vectorized_map}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_while_loop}()} \cr
diff --git a/man/op_conv.Rd b/man/op_conv.Rd
index e52c91370..43f3e4cf2 100644
--- a/man/op_conv.Rd
+++ b/man/op_conv.Rd
@@ -75,6 +75,7 @@ Other nn ops: \cr
 \code{\link{op_hard_sigmoid}()} \cr
 \code{\link{op_hard_silu}()} \cr
 \code{\link{op_hard_tanh}()} \cr
+\code{\link{op_layer_normalization}()} \cr
 \code{\link{op_leaky_relu}()} \cr
 \code{\link{op_log_sigmoid}()} \cr
 \code{\link{op_log_softmax}()} \cr
@@ -98,6 +99,7 @@ Other nn ops: \cr
 \code{\link{op_softsign}()} \cr
 \code{\link{op_sparse_categorical_crossentropy}()} \cr
 \code{\link{op_sparse_plus}()} \cr
+\code{\link{op_sparse_sigmoid}()} \cr
 \code{\link{op_sparsemax}()} \cr
 \code{\link{op_squareplus}()} \cr
 \code{\link{op_tanh_shrink}()} \cr
@@ -108,6 +110,7 @@ Other ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -126,6 +129,7 @@ Other ops: \cr
 \code{\link{op_associative_scan}()} \cr
 \code{\link{op_average}()} \cr
 \code{\link{op_average_pool}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_batch_normalization}()} \cr
 \code{\link{op_binary_crossentropy}()} \cr
 \code{\link{op_bincount}()} \cr
@@ -136,9 +140,11 @@ Other ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
 \code{\link{op_cast}()} \cr
 \code{\link{op_categorical_crossentropy}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_celu}()} \cr
 \code{\link{op_cholesky}()} \cr
@@ -150,6 +156,7 @@ Other ops: \cr
 \code{\link{op_convert_to_numpy}()} \cr
 \code{\link{op_convert_to_tensor}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -160,6 +167,7 @@ Other ops: \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
 \code{\link{op_custom_gradient}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_depthwise_conv}()} \cr
 \code{\link{op_det}()} \cr
 \code{\link{op_diag}()} \cr
@@ -199,10 +207,13 @@ Other ops: \cr
 \code{\link{op_glu}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
 \code{\link{op_hard_shrink}()} \cr
 \code{\link{op_hard_sigmoid}()} \cr
 \code{\link{op_hard_silu}()} \cr
 \code{\link{op_hard_tanh}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -229,6 +240,8 @@ Other ops: \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
 \code{\link{op_istft}()} \cr
+\code{\link{op_kaiser}()} \cr
+\code{\link{op_layer_normalization}()} \cr
 \code{\link{op_leaky_relu}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
@@ -328,6 +341,7 @@ Other ops: \cr
 \code{\link{op_sort}()} \cr
 \code{\link{op_sparse_categorical_crossentropy}()} \cr
 \code{\link{op_sparse_plus}()} \cr
+\code{\link{op_sparse_sigmoid}()} \cr
 \code{\link{op_sparsemax}()} \cr
 \code{\link{op_split}()} \cr
 \code{\link{op_sqrt}()} \cr
@@ -365,6 +379,8 @@ Other ops: \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
 \code{\link{op_vectorized_map}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_while_loop}()} \cr
diff --git a/man/op_conv_transpose.Rd b/man/op_conv_transpose.Rd
index 9195fdc8c..f4857b754 100644
--- a/man/op_conv_transpose.Rd
+++ b/man/op_conv_transpose.Rd
@@ -84,6 +84,7 @@ Other nn ops: \cr
 \code{\link{op_hard_sigmoid}()} \cr
 \code{\link{op_hard_silu}()} \cr
 \code{\link{op_hard_tanh}()} \cr
+\code{\link{op_layer_normalization}()} \cr
 \code{\link{op_leaky_relu}()} \cr
 \code{\link{op_log_sigmoid}()} \cr
 \code{\link{op_log_softmax}()} \cr
@@ -107,6 +108,7 @@ Other nn ops: \cr
 \code{\link{op_softsign}()} \cr
 \code{\link{op_sparse_categorical_crossentropy}()} \cr
 \code{\link{op_sparse_plus}()} \cr
+\code{\link{op_sparse_sigmoid}()} \cr
 \code{\link{op_sparsemax}()} \cr
 \code{\link{op_squareplus}()} \cr
 \code{\link{op_tanh_shrink}()} \cr
@@ -117,6 +119,7 @@ Other ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -135,6 +138,7 @@ Other ops: \cr
 \code{\link{op_associative_scan}()} \cr
 \code{\link{op_average}()} \cr
 \code{\link{op_average_pool}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_batch_normalization}()} \cr
 \code{\link{op_binary_crossentropy}()} \cr
 \code{\link{op_bincount}()} \cr
@@ -145,9 +149,11 @@ Other ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
 \code{\link{op_cast}()} \cr
 \code{\link{op_categorical_crossentropy}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_celu}()} \cr
 \code{\link{op_cholesky}()} \cr
@@ -159,6 +165,7 @@ Other ops: \cr
 \code{\link{op_convert_to_numpy}()} \cr
 \code{\link{op_convert_to_tensor}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -169,6 +176,7 @@ Other ops: \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
 \code{\link{op_custom_gradient}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_depthwise_conv}()} \cr
 \code{\link{op_det}()} \cr
 \code{\link{op_diag}()} \cr
@@ -208,10 +216,13 @@ Other ops: \cr
 \code{\link{op_glu}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
 \code{\link{op_hard_shrink}()} \cr
 \code{\link{op_hard_sigmoid}()} \cr
 \code{\link{op_hard_silu}()} \cr
 \code{\link{op_hard_tanh}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -238,6 +249,8 @@ Other ops: \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
 \code{\link{op_istft}()} \cr
+\code{\link{op_kaiser}()} \cr
+\code{\link{op_layer_normalization}()} \cr
 \code{\link{op_leaky_relu}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
@@ -337,6 +350,7 @@ Other ops: \cr
 \code{\link{op_sort}()} \cr
 \code{\link{op_sparse_categorical_crossentropy}()} \cr
 \code{\link{op_sparse_plus}()} \cr
+\code{\link{op_sparse_sigmoid}()} \cr
 \code{\link{op_sparsemax}()} \cr
 \code{\link{op_split}()} \cr
 \code{\link{op_sqrt}()} \cr
@@ -374,6 +388,8 @@ Other ops: \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
 \code{\link{op_vectorized_map}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_while_loop}()} \cr
diff --git a/man/op_convert_to_numpy.Rd b/man/op_convert_to_numpy.Rd
index afa8bebf2..f382a63f3 100644
--- a/man/op_convert_to_numpy.Rd
+++ b/man/op_convert_to_numpy.Rd
@@ -52,6 +52,7 @@ Other ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -70,6 +71,7 @@ Other ops: \cr
 \code{\link{op_associative_scan}()} \cr
 \code{\link{op_average}()} \cr
 \code{\link{op_average_pool}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_batch_normalization}()} \cr
 \code{\link{op_binary_crossentropy}()} \cr
 \code{\link{op_bincount}()} \cr
@@ -80,9 +82,11 @@ Other ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
 \code{\link{op_cast}()} \cr
 \code{\link{op_categorical_crossentropy}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_celu}()} \cr
 \code{\link{op_cholesky}()} \cr
@@ -94,6 +98,7 @@ Other ops: \cr
 \code{\link{op_conv_transpose}()} \cr
 \code{\link{op_convert_to_tensor}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -104,6 +109,7 @@ Other ops: \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
 \code{\link{op_custom_gradient}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_depthwise_conv}()} \cr
 \code{\link{op_det}()} \cr
 \code{\link{op_diag}()} \cr
@@ -143,10 +149,13 @@ Other ops: \cr
 \code{\link{op_glu}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
 \code{\link{op_hard_shrink}()} \cr
 \code{\link{op_hard_sigmoid}()} \cr
 \code{\link{op_hard_silu}()} \cr
 \code{\link{op_hard_tanh}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -173,6 +182,8 @@ Other ops: \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
 \code{\link{op_istft}()} \cr
+\code{\link{op_kaiser}()} \cr
+\code{\link{op_layer_normalization}()} \cr
 \code{\link{op_leaky_relu}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
@@ -272,6 +283,7 @@ Other ops: \cr
 \code{\link{op_sort}()} \cr
 \code{\link{op_sparse_categorical_crossentropy}()} \cr
 \code{\link{op_sparse_plus}()} \cr
+\code{\link{op_sparse_sigmoid}()} \cr
 \code{\link{op_sparsemax}()} \cr
 \code{\link{op_split}()} \cr
 \code{\link{op_sqrt}()} \cr
@@ -309,6 +321,8 @@ Other ops: \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
 \code{\link{op_vectorized_map}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_while_loop}()} \cr
diff --git a/man/op_convert_to_tensor.Rd b/man/op_convert_to_tensor.Rd
index 9c950deb5..582a07e5a 100644
--- a/man/op_convert_to_tensor.Rd
+++ b/man/op_convert_to_tensor.Rd
@@ -79,6 +79,7 @@ Other ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -97,6 +98,7 @@ Other ops: \cr
 \code{\link{op_associative_scan}()} \cr
 \code{\link{op_average}()} \cr
 \code{\link{op_average_pool}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_batch_normalization}()} \cr
 \code{\link{op_binary_crossentropy}()} \cr
 \code{\link{op_bincount}()} \cr
@@ -107,9 +109,11 @@ Other ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
 \code{\link{op_cast}()} \cr
 \code{\link{op_categorical_crossentropy}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_celu}()} \cr
 \code{\link{op_cholesky}()} \cr
@@ -121,6 +125,7 @@ Other ops: \cr
 \code{\link{op_conv_transpose}()} \cr
 \code{\link{op_convert_to_numpy}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -131,6 +136,7 @@ Other ops: \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
 \code{\link{op_custom_gradient}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_depthwise_conv}()} \cr
 \code{\link{op_det}()} \cr
 \code{\link{op_diag}()} \cr
@@ -170,10 +176,13 @@ Other ops: \cr
 \code{\link{op_glu}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
 \code{\link{op_hard_shrink}()} \cr
 \code{\link{op_hard_sigmoid}()} \cr
 \code{\link{op_hard_silu}()} \cr
 \code{\link{op_hard_tanh}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -200,6 +209,8 @@ Other ops: \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
 \code{\link{op_istft}()} \cr
+\code{\link{op_kaiser}()} \cr
+\code{\link{op_layer_normalization}()} \cr
 \code{\link{op_leaky_relu}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
@@ -299,6 +310,7 @@ Other ops: \cr
 \code{\link{op_sort}()} \cr
 \code{\link{op_sparse_categorical_crossentropy}()} \cr
 \code{\link{op_sparse_plus}()} \cr
+\code{\link{op_sparse_sigmoid}()} \cr
 \code{\link{op_sparsemax}()} \cr
 \code{\link{op_split}()} \cr
 \code{\link{op_sqrt}()} \cr
@@ -336,6 +348,8 @@ Other ops: \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
 \code{\link{op_vectorized_map}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_while_loop}()} \cr
diff --git a/man/op_copy.Rd b/man/op_copy.Rd
index 822e119dd..c36ef712e 100644
--- a/man/op_copy.Rd
+++ b/man/op_copy.Rd
@@ -24,6 +24,7 @@ Other numpy ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -40,6 +41,7 @@ Other numpy ops: \cr
 \code{\link{op_argsort}()} \cr
 \code{\link{op_array}()} \cr
 \code{\link{op_average}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_bincount}()} \cr
 \code{\link{op_bitwise_and}()} \cr
 \code{\link{op_bitwise_invert}()} \cr
@@ -48,11 +50,14 @@ Other numpy ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_clip}()} \cr
 \code{\link{op_concatenate}()} \cr
 \code{\link{op_conj}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -61,6 +66,7 @@ Other numpy ops: \cr
 \code{\link{op_ctc_decode}()} \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_diag}()} \cr
 \code{\link{op_diagflat}()} \cr
 \code{\link{op_diagonal}()} \cr
@@ -85,6 +91,9 @@ Other numpy ops: \cr
 \code{\link{op_get_item}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -94,6 +103,7 @@ Other numpy ops: \cr
 \code{\link{op_isfinite}()} \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
+\code{\link{op_kaiser}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
 \code{\link{op_less_equal}()} \cr
@@ -174,6 +184,8 @@ Other numpy ops: \cr
 \code{\link{op_var}()} \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_zeros}()} \cr
@@ -183,6 +195,7 @@ Other ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -201,6 +214,7 @@ Other ops: \cr
 \code{\link{op_associative_scan}()} \cr
 \code{\link{op_average}()} \cr
 \code{\link{op_average_pool}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_batch_normalization}()} \cr
 \code{\link{op_binary_crossentropy}()} \cr
 \code{\link{op_bincount}()} \cr
@@ -211,9 +225,11 @@ Other ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
 \code{\link{op_cast}()} \cr
 \code{\link{op_categorical_crossentropy}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_celu}()} \cr
 \code{\link{op_cholesky}()} \cr
@@ -225,6 +241,7 @@ Other ops: \cr
 \code{\link{op_conv_transpose}()} \cr
 \code{\link{op_convert_to_numpy}()} \cr
 \code{\link{op_convert_to_tensor}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -235,6 +252,7 @@ Other ops: \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
 \code{\link{op_custom_gradient}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_depthwise_conv}()} \cr
 \code{\link{op_det}()} \cr
 \code{\link{op_diag}()} \cr
@@ -274,10 +292,13 @@ Other ops: \cr
 \code{\link{op_glu}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
 \code{\link{op_hard_shrink}()} \cr
 \code{\link{op_hard_sigmoid}()} \cr
 \code{\link{op_hard_silu}()} \cr
 \code{\link{op_hard_tanh}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -304,6 +325,8 @@ Other ops: \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
 \code{\link{op_istft}()} \cr
+\code{\link{op_kaiser}()} \cr
+\code{\link{op_layer_normalization}()} \cr
 \code{\link{op_leaky_relu}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
@@ -403,6 +426,7 @@ Other ops: \cr
 \code{\link{op_sort}()} \cr
 \code{\link{op_sparse_categorical_crossentropy}()} \cr
 \code{\link{op_sparse_plus}()} \cr
+\code{\link{op_sparse_sigmoid}()} \cr
 \code{\link{op_sparsemax}()} \cr
 \code{\link{op_split}()} \cr
 \code{\link{op_sqrt}()} \cr
@@ -440,6 +464,8 @@ Other ops: \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
 \code{\link{op_vectorized_map}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_while_loop}()} \cr
diff --git a/man/op_corrcoef.Rd b/man/op_corrcoef.Rd
new file mode 100644
index 000000000..f7c7f4326
--- /dev/null
+++ b/man/op_corrcoef.Rd
@@ -0,0 +1,484 @@
+% Generated by roxygen2: do not edit by hand
+% Please edit documentation in R/ops-numpy.R
+\name{op_corrcoef}
+\alias{op_corrcoef}
+\title{Compute the Pearson correlation coefficient matrix.}
+\usage{
+op_corrcoef(x)
+}
+\arguments{
+\item{x}{A 2D tensor of shape \verb{(N, D)}, where \code{N} is the number of variables
+and \code{D} is the number of observations.Examples
+
+\if{html}{\out{<div class="sourceCode r">}}\preformatted{x <- op_array(matrix(c(1, 2, 3,
+                       2, 3, 4), nrow = 2, byrow = TRUE))
+op_corrcoef(x)
+}\if{html}{\out{</div>}}
+
+\if{html}{\out{<div class="sourceCode">}}\preformatted{## tf.Tensor(
+## [[1. 1.]
+##  [1. 1.]], shape=(2, 2), dtype=float64)
+
+}\if{html}{\out{</div>}}}
+}
+\value{
+A tensor of shape \verb{(N, N)} representing the correlation matrix.
+}
+\description{
+Compute the Pearson correlation coefficient matrix.
+}
+\seealso{
+Other numpy ops: \cr
+\code{\link{op_abs}()} \cr
+\code{\link{op_add}()} \cr
+\code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
+\code{\link{op_any}()} \cr
+\code{\link{op_append}()} \cr
+\code{\link{op_arange}()} \cr
+\code{\link{op_arccos}()} \cr
+\code{\link{op_arccosh}()} \cr
+\code{\link{op_arcsin}()} \cr
+\code{\link{op_arcsinh}()} \cr
+\code{\link{op_arctan}()} \cr
+\code{\link{op_arctan2}()} \cr
+\code{\link{op_arctanh}()} \cr
+\code{\link{op_argmax}()} \cr
+\code{\link{op_argmin}()} \cr
+\code{\link{op_argpartition}()} \cr
+\code{\link{op_argsort}()} \cr
+\code{\link{op_array}()} \cr
+\code{\link{op_average}()} \cr
+\code{\link{op_bartlett}()} \cr
+\code{\link{op_bincount}()} \cr
+\code{\link{op_bitwise_and}()} \cr
+\code{\link{op_bitwise_invert}()} \cr
+\code{\link{op_bitwise_left_shift}()} \cr
+\code{\link{op_bitwise_not}()} \cr
+\code{\link{op_bitwise_or}()} \cr
+\code{\link{op_bitwise_right_shift}()} \cr
+\code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
+\code{\link{op_broadcast_to}()} \cr
+\code{\link{op_cbrt}()} \cr
+\code{\link{op_ceil}()} \cr
+\code{\link{op_clip}()} \cr
+\code{\link{op_concatenate}()} \cr
+\code{\link{op_conj}()} \cr
+\code{\link{op_copy}()} \cr
+\code{\link{op_correlate}()} \cr
+\code{\link{op_cos}()} \cr
+\code{\link{op_cosh}()} \cr
+\code{\link{op_count_nonzero}()} \cr
+\code{\link{op_cross}()} \cr
+\code{\link{op_ctc_decode}()} \cr
+\code{\link{op_cumprod}()} \cr
+\code{\link{op_cumsum}()} \cr
+\code{\link{op_deg2rad}()} \cr
+\code{\link{op_diag}()} \cr
+\code{\link{op_diagflat}()} \cr
+\code{\link{op_diagonal}()} \cr
+\code{\link{op_diff}()} \cr
+\code{\link{op_digitize}()} \cr
+\code{\link{op_divide}()} \cr
+\code{\link{op_divide_no_nan}()} \cr
+\code{\link{op_dot}()} \cr
+\code{\link{op_einsum}()} \cr
+\code{\link{op_empty}()} \cr
+\code{\link{op_equal}()} \cr
+\code{\link{op_exp}()} \cr
+\code{\link{op_exp2}()} \cr
+\code{\link{op_expand_dims}()} \cr
+\code{\link{op_expm1}()} \cr
+\code{\link{op_eye}()} \cr
+\code{\link{op_flip}()} \cr
+\code{\link{op_floor}()} \cr
+\code{\link{op_floor_divide}()} \cr
+\code{\link{op_full}()} \cr
+\code{\link{op_full_like}()} \cr
+\code{\link{op_get_item}()} \cr
+\code{\link{op_greater}()} \cr
+\code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
+\code{\link{op_heaviside}()} \cr
+\code{\link{op_histogram}()} \cr
+\code{\link{op_hstack}()} \cr
+\code{\link{op_identity}()} \cr
+\code{\link{op_imag}()} \cr
+\code{\link{op_inner}()} \cr
+\code{\link{op_isclose}()} \cr
+\code{\link{op_isfinite}()} \cr
+\code{\link{op_isinf}()} \cr
+\code{\link{op_isnan}()} \cr
+\code{\link{op_kaiser}()} \cr
+\code{\link{op_left_shift}()} \cr
+\code{\link{op_less}()} \cr
+\code{\link{op_less_equal}()} \cr
+\code{\link{op_linspace}()} \cr
+\code{\link{op_log}()} \cr
+\code{\link{op_log10}()} \cr
+\code{\link{op_log1p}()} \cr
+\code{\link{op_log2}()} \cr
+\code{\link{op_logaddexp}()} \cr
+\code{\link{op_logdet}()} \cr
+\code{\link{op_logical_and}()} \cr
+\code{\link{op_logical_not}()} \cr
+\code{\link{op_logical_or}()} \cr
+\code{\link{op_logical_xor}()} \cr
+\code{\link{op_logspace}()} \cr
+\code{\link{op_lstsq}()} \cr
+\code{\link{op_matmul}()} \cr
+\code{\link{op_max}()} \cr
+\code{\link{op_maximum}()} \cr
+\code{\link{op_mean}()} \cr
+\code{\link{op_median}()} \cr
+\code{\link{op_meshgrid}()} \cr
+\code{\link{op_min}()} \cr
+\code{\link{op_minimum}()} \cr
+\code{\link{op_mod}()} \cr
+\code{\link{op_moveaxis}()} \cr
+\code{\link{op_multiply}()} \cr
+\code{\link{op_nan_to_num}()} \cr
+\code{\link{op_ndim}()} \cr
+\code{\link{op_negative}()} \cr
+\code{\link{op_nonzero}()} \cr
+\code{\link{op_not_equal}()} \cr
+\code{\link{op_ones}()} \cr
+\code{\link{op_ones_like}()} \cr
+\code{\link{op_outer}()} \cr
+\code{\link{op_pad}()} \cr
+\code{\link{op_power}()} \cr
+\code{\link{op_prod}()} \cr
+\code{\link{op_quantile}()} \cr
+\code{\link{op_ravel}()} \cr
+\code{\link{op_real}()} \cr
+\code{\link{op_reciprocal}()} \cr
+\code{\link{op_repeat}()} \cr
+\code{\link{op_reshape}()} \cr
+\code{\link{op_right_shift}()} \cr
+\code{\link{op_roll}()} \cr
+\code{\link{op_rot90}()} \cr
+\code{\link{op_round}()} \cr
+\code{\link{op_saturate_cast}()} \cr
+\code{\link{op_select}()} \cr
+\code{\link{op_sign}()} \cr
+\code{\link{op_signbit}()} \cr
+\code{\link{op_sin}()} \cr
+\code{\link{op_sinh}()} \cr
+\code{\link{op_size}()} \cr
+\code{\link{op_sort}()} \cr
+\code{\link{op_split}()} \cr
+\code{\link{op_sqrt}()} \cr
+\code{\link{op_square}()} \cr
+\code{\link{op_squeeze}()} \cr
+\code{\link{op_stack}()} \cr
+\code{\link{op_std}()} \cr
+\code{\link{op_subtract}()} \cr
+\code{\link{op_sum}()} \cr
+\code{\link{op_swapaxes}()} \cr
+\code{\link{op_take}()} \cr
+\code{\link{op_take_along_axis}()} \cr
+\code{\link{op_tan}()} \cr
+\code{\link{op_tanh}()} \cr
+\code{\link{op_tensordot}()} \cr
+\code{\link{op_tile}()} \cr
+\code{\link{op_trace}()} \cr
+\code{\link{op_transpose}()} \cr
+\code{\link{op_tri}()} \cr
+\code{\link{op_tril}()} \cr
+\code{\link{op_triu}()} \cr
+\code{\link{op_trunc}()} \cr
+\code{\link{op_var}()} \cr
+\code{\link{op_vdot}()} \cr
+\code{\link{op_vectorize}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
+\code{\link{op_vstack}()} \cr
+\code{\link{op_where}()} \cr
+\code{\link{op_zeros}()} \cr
+\code{\link{op_zeros_like}()} \cr
+
+Other ops: \cr
+\code{\link{op_abs}()} \cr
+\code{\link{op_add}()} \cr
+\code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
+\code{\link{op_any}()} \cr
+\code{\link{op_append}()} \cr
+\code{\link{op_arange}()} \cr
+\code{\link{op_arccos}()} \cr
+\code{\link{op_arccosh}()} \cr
+\code{\link{op_arcsin}()} \cr
+\code{\link{op_arcsinh}()} \cr
+\code{\link{op_arctan}()} \cr
+\code{\link{op_arctan2}()} \cr
+\code{\link{op_arctanh}()} \cr
+\code{\link{op_argmax}()} \cr
+\code{\link{op_argmin}()} \cr
+\code{\link{op_argpartition}()} \cr
+\code{\link{op_argsort}()} \cr
+\code{\link{op_array}()} \cr
+\code{\link{op_associative_scan}()} \cr
+\code{\link{op_average}()} \cr
+\code{\link{op_average_pool}()} \cr
+\code{\link{op_bartlett}()} \cr
+\code{\link{op_batch_normalization}()} \cr
+\code{\link{op_binary_crossentropy}()} \cr
+\code{\link{op_bincount}()} \cr
+\code{\link{op_bitwise_and}()} \cr
+\code{\link{op_bitwise_invert}()} \cr
+\code{\link{op_bitwise_left_shift}()} \cr
+\code{\link{op_bitwise_not}()} \cr
+\code{\link{op_bitwise_or}()} \cr
+\code{\link{op_bitwise_right_shift}()} \cr
+\code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
+\code{\link{op_broadcast_to}()} \cr
+\code{\link{op_cast}()} \cr
+\code{\link{op_categorical_crossentropy}()} \cr
+\code{\link{op_cbrt}()} \cr
+\code{\link{op_ceil}()} \cr
+\code{\link{op_celu}()} \cr
+\code{\link{op_cholesky}()} \cr
+\code{\link{op_clip}()} \cr
+\code{\link{op_concatenate}()} \cr
+\code{\link{op_cond}()} \cr
+\code{\link{op_conj}()} \cr
+\code{\link{op_conv}()} \cr
+\code{\link{op_conv_transpose}()} \cr
+\code{\link{op_convert_to_numpy}()} \cr
+\code{\link{op_convert_to_tensor}()} \cr
+\code{\link{op_copy}()} \cr
+\code{\link{op_correlate}()} \cr
+\code{\link{op_cos}()} \cr
+\code{\link{op_cosh}()} \cr
+\code{\link{op_count_nonzero}()} \cr
+\code{\link{op_cross}()} \cr
+\code{\link{op_ctc_decode}()} \cr
+\code{\link{op_ctc_loss}()} \cr
+\code{\link{op_cumprod}()} \cr
+\code{\link{op_cumsum}()} \cr
+\code{\link{op_custom_gradient}()} \cr
+\code{\link{op_deg2rad}()} \cr
+\code{\link{op_depthwise_conv}()} \cr
+\code{\link{op_det}()} \cr
+\code{\link{op_diag}()} \cr
+\code{\link{op_diagflat}()} \cr
+\code{\link{op_diagonal}()} \cr
+\code{\link{op_diff}()} \cr
+\code{\link{op_digitize}()} \cr
+\code{\link{op_divide}()} \cr
+\code{\link{op_divide_no_nan}()} \cr
+\code{\link{op_dot}()} \cr
+\code{\link{op_dot_product_attention}()} \cr
+\code{\link{op_dtype}()} \cr
+\code{\link{op_eig}()} \cr
+\code{\link{op_eigh}()} \cr
+\code{\link{op_einsum}()} \cr
+\code{\link{op_elu}()} \cr
+\code{\link{op_empty}()} \cr
+\code{\link{op_equal}()} \cr
+\code{\link{op_erf}()} \cr
+\code{\link{op_erfinv}()} \cr
+\code{\link{op_exp}()} \cr
+\code{\link{op_exp2}()} \cr
+\code{\link{op_expand_dims}()} \cr
+\code{\link{op_expm1}()} \cr
+\code{\link{op_extract_sequences}()} \cr
+\code{\link{op_eye}()} \cr
+\code{\link{op_fft}()} \cr
+\code{\link{op_fft2}()} \cr
+\code{\link{op_flip}()} \cr
+\code{\link{op_floor}()} \cr
+\code{\link{op_floor_divide}()} \cr
+\code{\link{op_fori_loop}()} \cr
+\code{\link{op_full}()} \cr
+\code{\link{op_full_like}()} \cr
+\code{\link{op_gelu}()} \cr
+\code{\link{op_get_item}()} \cr
+\code{\link{op_glu}()} \cr
+\code{\link{op_greater}()} \cr
+\code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
+\code{\link{op_hard_shrink}()} \cr
+\code{\link{op_hard_sigmoid}()} \cr
+\code{\link{op_hard_silu}()} \cr
+\code{\link{op_hard_tanh}()} \cr
+\code{\link{op_heaviside}()} \cr
+\code{\link{op_histogram}()} \cr
+\code{\link{op_hstack}()} \cr
+\code{\link{op_identity}()} \cr
+\code{\link{op_ifft2}()} \cr
+\code{\link{op_imag}()} \cr
+\code{\link{op_image_affine_transform}()} \cr
+\code{\link{op_image_crop}()} \cr
+\code{\link{op_image_extract_patches}()} \cr
+\code{\link{op_image_gaussian_blur}()} \cr
+\code{\link{op_image_hsv_to_rgb}()} \cr
+\code{\link{op_image_map_coordinates}()} \cr
+\code{\link{op_image_pad}()} \cr
+\code{\link{op_image_perspective_transform}()} \cr
+\code{\link{op_image_resize}()} \cr
+\code{\link{op_image_rgb_to_grayscale}()} \cr
+\code{\link{op_image_rgb_to_hsv}()} \cr
+\code{\link{op_in_top_k}()} \cr
+\code{\link{op_inner}()} \cr
+\code{\link{op_inv}()} \cr
+\code{\link{op_irfft}()} \cr
+\code{\link{op_is_tensor}()} \cr
+\code{\link{op_isclose}()} \cr
+\code{\link{op_isfinite}()} \cr
+\code{\link{op_isinf}()} \cr
+\code{\link{op_isnan}()} \cr
+\code{\link{op_istft}()} \cr
+\code{\link{op_kaiser}()} \cr
+\code{\link{op_layer_normalization}()} \cr
+\code{\link{op_leaky_relu}()} \cr
+\code{\link{op_left_shift}()} \cr
+\code{\link{op_less}()} \cr
+\code{\link{op_less_equal}()} \cr
+\code{\link{op_linspace}()} \cr
+\code{\link{op_log}()} \cr
+\code{\link{op_log10}()} \cr
+\code{\link{op_log1p}()} \cr
+\code{\link{op_log2}()} \cr
+\code{\link{op_log_sigmoid}()} \cr
+\code{\link{op_log_softmax}()} \cr
+\code{\link{op_logaddexp}()} \cr
+\code{\link{op_logdet}()} \cr
+\code{\link{op_logical_and}()} \cr
+\code{\link{op_logical_not}()} \cr
+\code{\link{op_logical_or}()} \cr
+\code{\link{op_logical_xor}()} \cr
+\code{\link{op_logspace}()} \cr
+\code{\link{op_logsumexp}()} \cr
+\code{\link{op_lstsq}()} \cr
+\code{\link{op_lu_factor}()} \cr
+\code{\link{op_map}()} \cr
+\code{\link{op_matmul}()} \cr
+\code{\link{op_max}()} \cr
+\code{\link{op_max_pool}()} \cr
+\code{\link{op_maximum}()} \cr
+\code{\link{op_mean}()} \cr
+\code{\link{op_median}()} \cr
+\code{\link{op_meshgrid}()} \cr
+\code{\link{op_min}()} \cr
+\code{\link{op_minimum}()} \cr
+\code{\link{op_mod}()} \cr
+\code{\link{op_moments}()} \cr
+\code{\link{op_moveaxis}()} \cr
+\code{\link{op_multi_hot}()} \cr
+\code{\link{op_multiply}()} \cr
+\code{\link{op_nan_to_num}()} \cr
+\code{\link{op_ndim}()} \cr
+\code{\link{op_negative}()} \cr
+\code{\link{op_nonzero}()} \cr
+\code{\link{op_norm}()} \cr
+\code{\link{op_normalize}()} \cr
+\code{\link{op_not_equal}()} \cr
+\code{\link{op_one_hot}()} \cr
+\code{\link{op_ones}()} \cr
+\code{\link{op_ones_like}()} \cr
+\code{\link{op_outer}()} \cr
+\code{\link{op_pad}()} \cr
+\code{\link{op_polar}()} \cr
+\code{\link{op_power}()} \cr
+\code{\link{op_prod}()} \cr
+\code{\link{op_psnr}()} \cr
+\code{\link{op_qr}()} \cr
+\code{\link{op_quantile}()} \cr
+\code{\link{op_ravel}()} \cr
+\code{\link{op_real}()} \cr
+\code{\link{op_rearrange}()} \cr
+\code{\link{op_reciprocal}()} \cr
+\code{\link{op_relu}()} \cr
+\code{\link{op_relu6}()} \cr
+\code{\link{op_repeat}()} \cr
+\code{\link{op_reshape}()} \cr
+\code{\link{op_rfft}()} \cr
+\code{\link{op_right_shift}()} \cr
+\code{\link{op_rms_normalization}()} \cr
+\code{\link{op_roll}()} \cr
+\code{\link{op_rot90}()} \cr
+\code{\link{op_round}()} \cr
+\code{\link{op_rsqrt}()} \cr
+\code{\link{op_saturate_cast}()} \cr
+\code{\link{op_scan}()} \cr
+\code{\link{op_scatter}()} \cr
+\code{\link{op_scatter_update}()} \cr
+\code{\link{op_searchsorted}()} \cr
+\code{\link{op_segment_max}()} \cr
+\code{\link{op_segment_sum}()} \cr
+\code{\link{op_select}()} \cr
+\code{\link{op_selu}()} \cr
+\code{\link{op_separable_conv}()} \cr
+\code{\link{op_shape}()} \cr
+\code{\link{op_sigmoid}()} \cr
+\code{\link{op_sign}()} \cr
+\code{\link{op_signbit}()} \cr
+\code{\link{op_silu}()} \cr
+\code{\link{op_sin}()} \cr
+\code{\link{op_sinh}()} \cr
+\code{\link{op_size}()} \cr
+\code{\link{op_slice}()} \cr
+\code{\link{op_slice_update}()} \cr
+\code{\link{op_slogdet}()} \cr
+\code{\link{op_soft_shrink}()} \cr
+\code{\link{op_softmax}()} \cr
+\code{\link{op_softplus}()} \cr
+\code{\link{op_softsign}()} \cr
+\code{\link{op_solve}()} \cr
+\code{\link{op_solve_triangular}()} \cr
+\code{\link{op_sort}()} \cr
+\code{\link{op_sparse_categorical_crossentropy}()} \cr
+\code{\link{op_sparse_plus}()} \cr
+\code{\link{op_sparse_sigmoid}()} \cr
+\code{\link{op_sparsemax}()} \cr
+\code{\link{op_split}()} \cr
+\code{\link{op_sqrt}()} \cr
+\code{\link{op_square}()} \cr
+\code{\link{op_squareplus}()} \cr
+\code{\link{op_squeeze}()} \cr
+\code{\link{op_stack}()} \cr
+\code{\link{op_std}()} \cr
+\code{\link{op_stft}()} \cr
+\code{\link{op_stop_gradient}()} \cr
+\code{\link{op_subset}()} \cr
+\code{\link{op_subtract}()} \cr
+\code{\link{op_sum}()} \cr
+\code{\link{op_svd}()} \cr
+\code{\link{op_swapaxes}()} \cr
+\code{\link{op_switch}()} \cr
+\code{\link{op_take}()} \cr
+\code{\link{op_take_along_axis}()} \cr
+\code{\link{op_tan}()} \cr
+\code{\link{op_tanh}()} \cr
+\code{\link{op_tanh_shrink}()} \cr
+\code{\link{op_tensordot}()} \cr
+\code{\link{op_threshold}()} \cr
+\code{\link{op_tile}()} \cr
+\code{\link{op_top_k}()} \cr
+\code{\link{op_trace}()} \cr
+\code{\link{op_transpose}()} \cr
+\code{\link{op_tri}()} \cr
+\code{\link{op_tril}()} \cr
+\code{\link{op_triu}()} \cr
+\code{\link{op_trunc}()} \cr
+\code{\link{op_unravel_index}()} \cr
+\code{\link{op_unstack}()} \cr
+\code{\link{op_var}()} \cr
+\code{\link{op_vdot}()} \cr
+\code{\link{op_vectorize}()} \cr
+\code{\link{op_vectorized_map}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
+\code{\link{op_vstack}()} \cr
+\code{\link{op_where}()} \cr
+\code{\link{op_while_loop}()} \cr
+\code{\link{op_zeros}()} \cr
+\code{\link{op_zeros_like}()} \cr
+}
+\concept{numpy ops}
+\concept{ops}
diff --git a/man/op_correlate.Rd b/man/op_correlate.Rd
index 8dcee313a..a96518115 100644
--- a/man/op_correlate.Rd
+++ b/man/op_correlate.Rd
@@ -29,6 +29,7 @@ Other numpy ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -45,6 +46,7 @@ Other numpy ops: \cr
 \code{\link{op_argsort}()} \cr
 \code{\link{op_array}()} \cr
 \code{\link{op_average}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_bincount}()} \cr
 \code{\link{op_bitwise_and}()} \cr
 \code{\link{op_bitwise_invert}()} \cr
@@ -53,12 +55,15 @@ Other numpy ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_clip}()} \cr
 \code{\link{op_concatenate}()} \cr
 \code{\link{op_conj}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
 \code{\link{op_count_nonzero}()} \cr
@@ -66,6 +71,7 @@ Other numpy ops: \cr
 \code{\link{op_ctc_decode}()} \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_diag}()} \cr
 \code{\link{op_diagflat}()} \cr
 \code{\link{op_diagonal}()} \cr
@@ -90,6 +96,9 @@ Other numpy ops: \cr
 \code{\link{op_get_item}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -99,6 +108,7 @@ Other numpy ops: \cr
 \code{\link{op_isfinite}()} \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
+\code{\link{op_kaiser}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
 \code{\link{op_less_equal}()} \cr
@@ -179,6 +189,8 @@ Other numpy ops: \cr
 \code{\link{op_var}()} \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_zeros}()} \cr
@@ -188,6 +200,7 @@ Other ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -206,6 +219,7 @@ Other ops: \cr
 \code{\link{op_associative_scan}()} \cr
 \code{\link{op_average}()} \cr
 \code{\link{op_average_pool}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_batch_normalization}()} \cr
 \code{\link{op_binary_crossentropy}()} \cr
 \code{\link{op_bincount}()} \cr
@@ -216,9 +230,11 @@ Other ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
 \code{\link{op_cast}()} \cr
 \code{\link{op_categorical_crossentropy}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_celu}()} \cr
 \code{\link{op_cholesky}()} \cr
@@ -231,6 +247,7 @@ Other ops: \cr
 \code{\link{op_convert_to_numpy}()} \cr
 \code{\link{op_convert_to_tensor}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
 \code{\link{op_count_nonzero}()} \cr
@@ -240,6 +257,7 @@ Other ops: \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
 \code{\link{op_custom_gradient}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_depthwise_conv}()} \cr
 \code{\link{op_det}()} \cr
 \code{\link{op_diag}()} \cr
@@ -279,10 +297,13 @@ Other ops: \cr
 \code{\link{op_glu}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
 \code{\link{op_hard_shrink}()} \cr
 \code{\link{op_hard_sigmoid}()} \cr
 \code{\link{op_hard_silu}()} \cr
 \code{\link{op_hard_tanh}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -309,6 +330,8 @@ Other ops: \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
 \code{\link{op_istft}()} \cr
+\code{\link{op_kaiser}()} \cr
+\code{\link{op_layer_normalization}()} \cr
 \code{\link{op_leaky_relu}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
@@ -408,6 +431,7 @@ Other ops: \cr
 \code{\link{op_sort}()} \cr
 \code{\link{op_sparse_categorical_crossentropy}()} \cr
 \code{\link{op_sparse_plus}()} \cr
+\code{\link{op_sparse_sigmoid}()} \cr
 \code{\link{op_sparsemax}()} \cr
 \code{\link{op_split}()} \cr
 \code{\link{op_sqrt}()} \cr
@@ -445,6 +469,8 @@ Other ops: \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
 \code{\link{op_vectorized_map}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_while_loop}()} \cr
diff --git a/man/op_cos.Rd b/man/op_cos.Rd
index fcdb281c6..77af2d0fd 100644
--- a/man/op_cos.Rd
+++ b/man/op_cos.Rd
@@ -24,6 +24,7 @@ Other numpy ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -40,6 +41,7 @@ Other numpy ops: \cr
 \code{\link{op_argsort}()} \cr
 \code{\link{op_array}()} \cr
 \code{\link{op_average}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_bincount}()} \cr
 \code{\link{op_bitwise_and}()} \cr
 \code{\link{op_bitwise_invert}()} \cr
@@ -48,12 +50,15 @@ Other numpy ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_clip}()} \cr
 \code{\link{op_concatenate}()} \cr
 \code{\link{op_conj}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cosh}()} \cr
 \code{\link{op_count_nonzero}()} \cr
@@ -61,6 +66,7 @@ Other numpy ops: \cr
 \code{\link{op_ctc_decode}()} \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_diag}()} \cr
 \code{\link{op_diagflat}()} \cr
 \code{\link{op_diagonal}()} \cr
@@ -85,6 +91,9 @@ Other numpy ops: \cr
 \code{\link{op_get_item}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -94,6 +103,7 @@ Other numpy ops: \cr
 \code{\link{op_isfinite}()} \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
+\code{\link{op_kaiser}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
 \code{\link{op_less_equal}()} \cr
@@ -174,6 +184,8 @@ Other numpy ops: \cr
 \code{\link{op_var}()} \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_zeros}()} \cr
@@ -183,6 +195,7 @@ Other ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -201,6 +214,7 @@ Other ops: \cr
 \code{\link{op_associative_scan}()} \cr
 \code{\link{op_average}()} \cr
 \code{\link{op_average_pool}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_batch_normalization}()} \cr
 \code{\link{op_binary_crossentropy}()} \cr
 \code{\link{op_bincount}()} \cr
@@ -211,9 +225,11 @@ Other ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
 \code{\link{op_cast}()} \cr
 \code{\link{op_categorical_crossentropy}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_celu}()} \cr
 \code{\link{op_cholesky}()} \cr
@@ -226,6 +242,7 @@ Other ops: \cr
 \code{\link{op_convert_to_numpy}()} \cr
 \code{\link{op_convert_to_tensor}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cosh}()} \cr
 \code{\link{op_count_nonzero}()} \cr
@@ -235,6 +252,7 @@ Other ops: \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
 \code{\link{op_custom_gradient}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_depthwise_conv}()} \cr
 \code{\link{op_det}()} \cr
 \code{\link{op_diag}()} \cr
@@ -274,10 +292,13 @@ Other ops: \cr
 \code{\link{op_glu}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
 \code{\link{op_hard_shrink}()} \cr
 \code{\link{op_hard_sigmoid}()} \cr
 \code{\link{op_hard_silu}()} \cr
 \code{\link{op_hard_tanh}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -304,6 +325,8 @@ Other ops: \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
 \code{\link{op_istft}()} \cr
+\code{\link{op_kaiser}()} \cr
+\code{\link{op_layer_normalization}()} \cr
 \code{\link{op_leaky_relu}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
@@ -403,6 +426,7 @@ Other ops: \cr
 \code{\link{op_sort}()} \cr
 \code{\link{op_sparse_categorical_crossentropy}()} \cr
 \code{\link{op_sparse_plus}()} \cr
+\code{\link{op_sparse_sigmoid}()} \cr
 \code{\link{op_sparsemax}()} \cr
 \code{\link{op_split}()} \cr
 \code{\link{op_sqrt}()} \cr
@@ -440,6 +464,8 @@ Other ops: \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
 \code{\link{op_vectorized_map}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_while_loop}()} \cr
diff --git a/man/op_cosh.Rd b/man/op_cosh.Rd
index 93874e59d..dbdbf1ba5 100644
--- a/man/op_cosh.Rd
+++ b/man/op_cosh.Rd
@@ -24,6 +24,7 @@ Other numpy ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -40,6 +41,7 @@ Other numpy ops: \cr
 \code{\link{op_argsort}()} \cr
 \code{\link{op_array}()} \cr
 \code{\link{op_average}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_bincount}()} \cr
 \code{\link{op_bitwise_and}()} \cr
 \code{\link{op_bitwise_invert}()} \cr
@@ -48,12 +50,15 @@ Other numpy ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_clip}()} \cr
 \code{\link{op_concatenate}()} \cr
 \code{\link{op_conj}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_count_nonzero}()} \cr
@@ -61,6 +66,7 @@ Other numpy ops: \cr
 \code{\link{op_ctc_decode}()} \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_diag}()} \cr
 \code{\link{op_diagflat}()} \cr
 \code{\link{op_diagonal}()} \cr
@@ -85,6 +91,9 @@ Other numpy ops: \cr
 \code{\link{op_get_item}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -94,6 +103,7 @@ Other numpy ops: \cr
 \code{\link{op_isfinite}()} \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
+\code{\link{op_kaiser}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
 \code{\link{op_less_equal}()} \cr
@@ -174,6 +184,8 @@ Other numpy ops: \cr
 \code{\link{op_var}()} \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_zeros}()} \cr
@@ -183,6 +195,7 @@ Other ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -201,6 +214,7 @@ Other ops: \cr
 \code{\link{op_associative_scan}()} \cr
 \code{\link{op_average}()} \cr
 \code{\link{op_average_pool}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_batch_normalization}()} \cr
 \code{\link{op_binary_crossentropy}()} \cr
 \code{\link{op_bincount}()} \cr
@@ -211,9 +225,11 @@ Other ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
 \code{\link{op_cast}()} \cr
 \code{\link{op_categorical_crossentropy}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_celu}()} \cr
 \code{\link{op_cholesky}()} \cr
@@ -226,6 +242,7 @@ Other ops: \cr
 \code{\link{op_convert_to_numpy}()} \cr
 \code{\link{op_convert_to_tensor}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_count_nonzero}()} \cr
@@ -235,6 +252,7 @@ Other ops: \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
 \code{\link{op_custom_gradient}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_depthwise_conv}()} \cr
 \code{\link{op_det}()} \cr
 \code{\link{op_diag}()} \cr
@@ -274,10 +292,13 @@ Other ops: \cr
 \code{\link{op_glu}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
 \code{\link{op_hard_shrink}()} \cr
 \code{\link{op_hard_sigmoid}()} \cr
 \code{\link{op_hard_silu}()} \cr
 \code{\link{op_hard_tanh}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -304,6 +325,8 @@ Other ops: \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
 \code{\link{op_istft}()} \cr
+\code{\link{op_kaiser}()} \cr
+\code{\link{op_layer_normalization}()} \cr
 \code{\link{op_leaky_relu}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
@@ -403,6 +426,7 @@ Other ops: \cr
 \code{\link{op_sort}()} \cr
 \code{\link{op_sparse_categorical_crossentropy}()} \cr
 \code{\link{op_sparse_plus}()} \cr
+\code{\link{op_sparse_sigmoid}()} \cr
 \code{\link{op_sparsemax}()} \cr
 \code{\link{op_split}()} \cr
 \code{\link{op_sqrt}()} \cr
@@ -440,6 +464,8 @@ Other ops: \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
 \code{\link{op_vectorized_map}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_while_loop}()} \cr
diff --git a/man/op_count_nonzero.Rd b/man/op_count_nonzero.Rd
index 04d14a678..adde16c7b 100644
--- a/man/op_count_nonzero.Rd
+++ b/man/op_count_nonzero.Rd
@@ -52,6 +52,7 @@ Other numpy ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -68,6 +69,7 @@ Other numpy ops: \cr
 \code{\link{op_argsort}()} \cr
 \code{\link{op_array}()} \cr
 \code{\link{op_average}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_bincount}()} \cr
 \code{\link{op_bitwise_and}()} \cr
 \code{\link{op_bitwise_invert}()} \cr
@@ -76,12 +78,15 @@ Other numpy ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_clip}()} \cr
 \code{\link{op_concatenate}()} \cr
 \code{\link{op_conj}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -89,6 +94,7 @@ Other numpy ops: \cr
 \code{\link{op_ctc_decode}()} \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_diag}()} \cr
 \code{\link{op_diagflat}()} \cr
 \code{\link{op_diagonal}()} \cr
@@ -113,6 +119,9 @@ Other numpy ops: \cr
 \code{\link{op_get_item}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -122,6 +131,7 @@ Other numpy ops: \cr
 \code{\link{op_isfinite}()} \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
+\code{\link{op_kaiser}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
 \code{\link{op_less_equal}()} \cr
@@ -202,6 +212,8 @@ Other numpy ops: \cr
 \code{\link{op_var}()} \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_zeros}()} \cr
@@ -211,6 +223,7 @@ Other ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -229,6 +242,7 @@ Other ops: \cr
 \code{\link{op_associative_scan}()} \cr
 \code{\link{op_average}()} \cr
 \code{\link{op_average_pool}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_batch_normalization}()} \cr
 \code{\link{op_binary_crossentropy}()} \cr
 \code{\link{op_bincount}()} \cr
@@ -239,9 +253,11 @@ Other ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
 \code{\link{op_cast}()} \cr
 \code{\link{op_categorical_crossentropy}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_celu}()} \cr
 \code{\link{op_cholesky}()} \cr
@@ -254,6 +270,7 @@ Other ops: \cr
 \code{\link{op_convert_to_numpy}()} \cr
 \code{\link{op_convert_to_tensor}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -263,6 +280,7 @@ Other ops: \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
 \code{\link{op_custom_gradient}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_depthwise_conv}()} \cr
 \code{\link{op_det}()} \cr
 \code{\link{op_diag}()} \cr
@@ -302,10 +320,13 @@ Other ops: \cr
 \code{\link{op_glu}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
 \code{\link{op_hard_shrink}()} \cr
 \code{\link{op_hard_sigmoid}()} \cr
 \code{\link{op_hard_silu}()} \cr
 \code{\link{op_hard_tanh}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -332,6 +353,8 @@ Other ops: \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
 \code{\link{op_istft}()} \cr
+\code{\link{op_kaiser}()} \cr
+\code{\link{op_layer_normalization}()} \cr
 \code{\link{op_leaky_relu}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
@@ -431,6 +454,7 @@ Other ops: \cr
 \code{\link{op_sort}()} \cr
 \code{\link{op_sparse_categorical_crossentropy}()} \cr
 \code{\link{op_sparse_plus}()} \cr
+\code{\link{op_sparse_sigmoid}()} \cr
 \code{\link{op_sparsemax}()} \cr
 \code{\link{op_split}()} \cr
 \code{\link{op_sqrt}()} \cr
@@ -468,6 +492,8 @@ Other ops: \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
 \code{\link{op_vectorized_map}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_while_loop}()} \cr
diff --git a/man/op_cross.Rd b/man/op_cross.Rd
index 01495adf1..8b986eac5 100644
--- a/man/op_cross.Rd
+++ b/man/op_cross.Rd
@@ -53,6 +53,7 @@ Other numpy ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -69,6 +70,7 @@ Other numpy ops: \cr
 \code{\link{op_argsort}()} \cr
 \code{\link{op_array}()} \cr
 \code{\link{op_average}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_bincount}()} \cr
 \code{\link{op_bitwise_and}()} \cr
 \code{\link{op_bitwise_invert}()} \cr
@@ -77,12 +79,15 @@ Other numpy ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_clip}()} \cr
 \code{\link{op_concatenate}()} \cr
 \code{\link{op_conj}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -90,6 +95,7 @@ Other numpy ops: \cr
 \code{\link{op_ctc_decode}()} \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_diag}()} \cr
 \code{\link{op_diagflat}()} \cr
 \code{\link{op_diagonal}()} \cr
@@ -114,6 +120,9 @@ Other numpy ops: \cr
 \code{\link{op_get_item}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -123,6 +132,7 @@ Other numpy ops: \cr
 \code{\link{op_isfinite}()} \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
+\code{\link{op_kaiser}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
 \code{\link{op_less_equal}()} \cr
@@ -203,6 +213,8 @@ Other numpy ops: \cr
 \code{\link{op_var}()} \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_zeros}()} \cr
@@ -212,6 +224,7 @@ Other ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -230,6 +243,7 @@ Other ops: \cr
 \code{\link{op_associative_scan}()} \cr
 \code{\link{op_average}()} \cr
 \code{\link{op_average_pool}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_batch_normalization}()} \cr
 \code{\link{op_binary_crossentropy}()} \cr
 \code{\link{op_bincount}()} \cr
@@ -240,9 +254,11 @@ Other ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
 \code{\link{op_cast}()} \cr
 \code{\link{op_categorical_crossentropy}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_celu}()} \cr
 \code{\link{op_cholesky}()} \cr
@@ -255,6 +271,7 @@ Other ops: \cr
 \code{\link{op_convert_to_numpy}()} \cr
 \code{\link{op_convert_to_tensor}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -264,6 +281,7 @@ Other ops: \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
 \code{\link{op_custom_gradient}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_depthwise_conv}()} \cr
 \code{\link{op_det}()} \cr
 \code{\link{op_diag}()} \cr
@@ -303,10 +321,13 @@ Other ops: \cr
 \code{\link{op_glu}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
 \code{\link{op_hard_shrink}()} \cr
 \code{\link{op_hard_sigmoid}()} \cr
 \code{\link{op_hard_silu}()} \cr
 \code{\link{op_hard_tanh}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -333,6 +354,8 @@ Other ops: \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
 \code{\link{op_istft}()} \cr
+\code{\link{op_kaiser}()} \cr
+\code{\link{op_layer_normalization}()} \cr
 \code{\link{op_leaky_relu}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
@@ -432,6 +455,7 @@ Other ops: \cr
 \code{\link{op_sort}()} \cr
 \code{\link{op_sparse_categorical_crossentropy}()} \cr
 \code{\link{op_sparse_plus}()} \cr
+\code{\link{op_sparse_sigmoid}()} \cr
 \code{\link{op_sparsemax}()} \cr
 \code{\link{op_split}()} \cr
 \code{\link{op_sqrt}()} \cr
@@ -469,6 +493,8 @@ Other ops: \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
 \code{\link{op_vectorized_map}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_while_loop}()} \cr
diff --git a/man/op_ctc_decode.Rd b/man/op_ctc_decode.Rd
index 3f1038641..78b4156c5 100644
--- a/man/op_ctc_decode.Rd
+++ b/man/op_ctc_decode.Rd
@@ -60,6 +60,7 @@ Other numpy ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -76,6 +77,7 @@ Other numpy ops: \cr
 \code{\link{op_argsort}()} \cr
 \code{\link{op_array}()} \cr
 \code{\link{op_average}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_bincount}()} \cr
 \code{\link{op_bitwise_and}()} \cr
 \code{\link{op_bitwise_invert}()} \cr
@@ -84,12 +86,15 @@ Other numpy ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_clip}()} \cr
 \code{\link{op_concatenate}()} \cr
 \code{\link{op_conj}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -97,6 +102,7 @@ Other numpy ops: \cr
 \code{\link{op_cross}()} \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_diag}()} \cr
 \code{\link{op_diagflat}()} \cr
 \code{\link{op_diagonal}()} \cr
@@ -121,6 +127,9 @@ Other numpy ops: \cr
 \code{\link{op_get_item}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -130,6 +139,7 @@ Other numpy ops: \cr
 \code{\link{op_isfinite}()} \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
+\code{\link{op_kaiser}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
 \code{\link{op_less_equal}()} \cr
@@ -210,6 +220,8 @@ Other numpy ops: \cr
 \code{\link{op_var}()} \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_zeros}()} \cr
@@ -219,6 +231,7 @@ Other ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -237,6 +250,7 @@ Other ops: \cr
 \code{\link{op_associative_scan}()} \cr
 \code{\link{op_average}()} \cr
 \code{\link{op_average_pool}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_batch_normalization}()} \cr
 \code{\link{op_binary_crossentropy}()} \cr
 \code{\link{op_bincount}()} \cr
@@ -247,9 +261,11 @@ Other ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
 \code{\link{op_cast}()} \cr
 \code{\link{op_categorical_crossentropy}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_celu}()} \cr
 \code{\link{op_cholesky}()} \cr
@@ -262,6 +278,7 @@ Other ops: \cr
 \code{\link{op_convert_to_numpy}()} \cr
 \code{\link{op_convert_to_tensor}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -271,6 +288,7 @@ Other ops: \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
 \code{\link{op_custom_gradient}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_depthwise_conv}()} \cr
 \code{\link{op_det}()} \cr
 \code{\link{op_diag}()} \cr
@@ -310,10 +328,13 @@ Other ops: \cr
 \code{\link{op_glu}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
 \code{\link{op_hard_shrink}()} \cr
 \code{\link{op_hard_sigmoid}()} \cr
 \code{\link{op_hard_silu}()} \cr
 \code{\link{op_hard_tanh}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -340,6 +361,8 @@ Other ops: \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
 \code{\link{op_istft}()} \cr
+\code{\link{op_kaiser}()} \cr
+\code{\link{op_layer_normalization}()} \cr
 \code{\link{op_leaky_relu}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
@@ -439,6 +462,7 @@ Other ops: \cr
 \code{\link{op_sort}()} \cr
 \code{\link{op_sparse_categorical_crossentropy}()} \cr
 \code{\link{op_sparse_plus}()} \cr
+\code{\link{op_sparse_sigmoid}()} \cr
 \code{\link{op_sparsemax}()} \cr
 \code{\link{op_split}()} \cr
 \code{\link{op_sqrt}()} \cr
@@ -476,6 +500,8 @@ Other ops: \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
 \code{\link{op_vectorized_map}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_while_loop}()} \cr
diff --git a/man/op_ctc_loss.Rd b/man/op_ctc_loss.Rd
index ad178ae84..6039f3872 100644
--- a/man/op_ctc_loss.Rd
+++ b/man/op_ctc_loss.Rd
@@ -46,6 +46,7 @@ Other nn ops: \cr
 \code{\link{op_hard_sigmoid}()} \cr
 \code{\link{op_hard_silu}()} \cr
 \code{\link{op_hard_tanh}()} \cr
+\code{\link{op_layer_normalization}()} \cr
 \code{\link{op_leaky_relu}()} \cr
 \code{\link{op_log_sigmoid}()} \cr
 \code{\link{op_log_softmax}()} \cr
@@ -69,6 +70,7 @@ Other nn ops: \cr
 \code{\link{op_softsign}()} \cr
 \code{\link{op_sparse_categorical_crossentropy}()} \cr
 \code{\link{op_sparse_plus}()} \cr
+\code{\link{op_sparse_sigmoid}()} \cr
 \code{\link{op_sparsemax}()} \cr
 \code{\link{op_squareplus}()} \cr
 \code{\link{op_tanh_shrink}()} \cr
@@ -79,6 +81,7 @@ Other ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -97,6 +100,7 @@ Other ops: \cr
 \code{\link{op_associative_scan}()} \cr
 \code{\link{op_average}()} \cr
 \code{\link{op_average_pool}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_batch_normalization}()} \cr
 \code{\link{op_binary_crossentropy}()} \cr
 \code{\link{op_bincount}()} \cr
@@ -107,9 +111,11 @@ Other ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
 \code{\link{op_cast}()} \cr
 \code{\link{op_categorical_crossentropy}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_celu}()} \cr
 \code{\link{op_cholesky}()} \cr
@@ -122,6 +128,7 @@ Other ops: \cr
 \code{\link{op_convert_to_numpy}()} \cr
 \code{\link{op_convert_to_tensor}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -131,6 +138,7 @@ Other ops: \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
 \code{\link{op_custom_gradient}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_depthwise_conv}()} \cr
 \code{\link{op_det}()} \cr
 \code{\link{op_diag}()} \cr
@@ -170,10 +178,13 @@ Other ops: \cr
 \code{\link{op_glu}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
 \code{\link{op_hard_shrink}()} \cr
 \code{\link{op_hard_sigmoid}()} \cr
 \code{\link{op_hard_silu}()} \cr
 \code{\link{op_hard_tanh}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -200,6 +211,8 @@ Other ops: \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
 \code{\link{op_istft}()} \cr
+\code{\link{op_kaiser}()} \cr
+\code{\link{op_layer_normalization}()} \cr
 \code{\link{op_leaky_relu}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
@@ -299,6 +312,7 @@ Other ops: \cr
 \code{\link{op_sort}()} \cr
 \code{\link{op_sparse_categorical_crossentropy}()} \cr
 \code{\link{op_sparse_plus}()} \cr
+\code{\link{op_sparse_sigmoid}()} \cr
 \code{\link{op_sparsemax}()} \cr
 \code{\link{op_split}()} \cr
 \code{\link{op_sqrt}()} \cr
@@ -336,6 +350,8 @@ Other ops: \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
 \code{\link{op_vectorized_map}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_while_loop}()} \cr
diff --git a/man/op_cumprod.Rd b/man/op_cumprod.Rd
index 0ab754601..a221acb0d 100644
--- a/man/op_cumprod.Rd
+++ b/man/op_cumprod.Rd
@@ -29,6 +29,7 @@ Other numpy ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -45,6 +46,7 @@ Other numpy ops: \cr
 \code{\link{op_argsort}()} \cr
 \code{\link{op_array}()} \cr
 \code{\link{op_average}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_bincount}()} \cr
 \code{\link{op_bitwise_and}()} \cr
 \code{\link{op_bitwise_invert}()} \cr
@@ -53,12 +55,15 @@ Other numpy ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_clip}()} \cr
 \code{\link{op_concatenate}()} \cr
 \code{\link{op_conj}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -66,6 +71,7 @@ Other numpy ops: \cr
 \code{\link{op_cross}()} \cr
 \code{\link{op_ctc_decode}()} \cr
 \code{\link{op_cumsum}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_diag}()} \cr
 \code{\link{op_diagflat}()} \cr
 \code{\link{op_diagonal}()} \cr
@@ -90,6 +96,9 @@ Other numpy ops: \cr
 \code{\link{op_get_item}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -99,6 +108,7 @@ Other numpy ops: \cr
 \code{\link{op_isfinite}()} \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
+\code{\link{op_kaiser}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
 \code{\link{op_less_equal}()} \cr
@@ -179,6 +189,8 @@ Other numpy ops: \cr
 \code{\link{op_var}()} \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_zeros}()} \cr
@@ -188,6 +200,7 @@ Other ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -206,6 +219,7 @@ Other ops: \cr
 \code{\link{op_associative_scan}()} \cr
 \code{\link{op_average}()} \cr
 \code{\link{op_average_pool}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_batch_normalization}()} \cr
 \code{\link{op_binary_crossentropy}()} \cr
 \code{\link{op_bincount}()} \cr
@@ -216,9 +230,11 @@ Other ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
 \code{\link{op_cast}()} \cr
 \code{\link{op_categorical_crossentropy}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_celu}()} \cr
 \code{\link{op_cholesky}()} \cr
@@ -231,6 +247,7 @@ Other ops: \cr
 \code{\link{op_convert_to_numpy}()} \cr
 \code{\link{op_convert_to_tensor}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -240,6 +257,7 @@ Other ops: \cr
 \code{\link{op_ctc_loss}()} \cr
 \code{\link{op_cumsum}()} \cr
 \code{\link{op_custom_gradient}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_depthwise_conv}()} \cr
 \code{\link{op_det}()} \cr
 \code{\link{op_diag}()} \cr
@@ -279,10 +297,13 @@ Other ops: \cr
 \code{\link{op_glu}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
 \code{\link{op_hard_shrink}()} \cr
 \code{\link{op_hard_sigmoid}()} \cr
 \code{\link{op_hard_silu}()} \cr
 \code{\link{op_hard_tanh}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -309,6 +330,8 @@ Other ops: \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
 \code{\link{op_istft}()} \cr
+\code{\link{op_kaiser}()} \cr
+\code{\link{op_layer_normalization}()} \cr
 \code{\link{op_leaky_relu}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
@@ -408,6 +431,7 @@ Other ops: \cr
 \code{\link{op_sort}()} \cr
 \code{\link{op_sparse_categorical_crossentropy}()} \cr
 \code{\link{op_sparse_plus}()} \cr
+\code{\link{op_sparse_sigmoid}()} \cr
 \code{\link{op_sparsemax}()} \cr
 \code{\link{op_split}()} \cr
 \code{\link{op_sqrt}()} \cr
@@ -445,6 +469,8 @@ Other ops: \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
 \code{\link{op_vectorized_map}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_while_loop}()} \cr
diff --git a/man/op_cumsum.Rd b/man/op_cumsum.Rd
index 8b0cba897..40ca88e96 100644
--- a/man/op_cumsum.Rd
+++ b/man/op_cumsum.Rd
@@ -29,6 +29,7 @@ Other numpy ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -45,6 +46,7 @@ Other numpy ops: \cr
 \code{\link{op_argsort}()} \cr
 \code{\link{op_array}()} \cr
 \code{\link{op_average}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_bincount}()} \cr
 \code{\link{op_bitwise_and}()} \cr
 \code{\link{op_bitwise_invert}()} \cr
@@ -53,12 +55,15 @@ Other numpy ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_clip}()} \cr
 \code{\link{op_concatenate}()} \cr
 \code{\link{op_conj}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -66,6 +71,7 @@ Other numpy ops: \cr
 \code{\link{op_cross}()} \cr
 \code{\link{op_ctc_decode}()} \cr
 \code{\link{op_cumprod}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_diag}()} \cr
 \code{\link{op_diagflat}()} \cr
 \code{\link{op_diagonal}()} \cr
@@ -90,6 +96,9 @@ Other numpy ops: \cr
 \code{\link{op_get_item}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -99,6 +108,7 @@ Other numpy ops: \cr
 \code{\link{op_isfinite}()} \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
+\code{\link{op_kaiser}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
 \code{\link{op_less_equal}()} \cr
@@ -179,6 +189,8 @@ Other numpy ops: \cr
 \code{\link{op_var}()} \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_zeros}()} \cr
@@ -188,6 +200,7 @@ Other ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -206,6 +219,7 @@ Other ops: \cr
 \code{\link{op_associative_scan}()} \cr
 \code{\link{op_average}()} \cr
 \code{\link{op_average_pool}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_batch_normalization}()} \cr
 \code{\link{op_binary_crossentropy}()} \cr
 \code{\link{op_bincount}()} \cr
@@ -216,9 +230,11 @@ Other ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
 \code{\link{op_cast}()} \cr
 \code{\link{op_categorical_crossentropy}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_celu}()} \cr
 \code{\link{op_cholesky}()} \cr
@@ -231,6 +247,7 @@ Other ops: \cr
 \code{\link{op_convert_to_numpy}()} \cr
 \code{\link{op_convert_to_tensor}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -240,6 +257,7 @@ Other ops: \cr
 \code{\link{op_ctc_loss}()} \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_custom_gradient}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_depthwise_conv}()} \cr
 \code{\link{op_det}()} \cr
 \code{\link{op_diag}()} \cr
@@ -279,10 +297,13 @@ Other ops: \cr
 \code{\link{op_glu}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
 \code{\link{op_hard_shrink}()} \cr
 \code{\link{op_hard_sigmoid}()} \cr
 \code{\link{op_hard_silu}()} \cr
 \code{\link{op_hard_tanh}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -309,6 +330,8 @@ Other ops: \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
 \code{\link{op_istft}()} \cr
+\code{\link{op_kaiser}()} \cr
+\code{\link{op_layer_normalization}()} \cr
 \code{\link{op_leaky_relu}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
@@ -408,6 +431,7 @@ Other ops: \cr
 \code{\link{op_sort}()} \cr
 \code{\link{op_sparse_categorical_crossentropy}()} \cr
 \code{\link{op_sparse_plus}()} \cr
+\code{\link{op_sparse_sigmoid}()} \cr
 \code{\link{op_sparsemax}()} \cr
 \code{\link{op_split}()} \cr
 \code{\link{op_sqrt}()} \cr
@@ -445,6 +469,8 @@ Other ops: \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
 \code{\link{op_vectorized_map}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_while_loop}()} \cr
diff --git a/man/op_custom_gradient.Rd b/man/op_custom_gradient.Rd
index 7debd1978..c6eb9e758 100644
--- a/man/op_custom_gradient.Rd
+++ b/man/op_custom_gradient.Rd
@@ -108,6 +108,7 @@ Other ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -126,6 +127,7 @@ Other ops: \cr
 \code{\link{op_associative_scan}()} \cr
 \code{\link{op_average}()} \cr
 \code{\link{op_average_pool}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_batch_normalization}()} \cr
 \code{\link{op_binary_crossentropy}()} \cr
 \code{\link{op_bincount}()} \cr
@@ -136,9 +138,11 @@ Other ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
 \code{\link{op_cast}()} \cr
 \code{\link{op_categorical_crossentropy}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_celu}()} \cr
 \code{\link{op_cholesky}()} \cr
@@ -151,6 +155,7 @@ Other ops: \cr
 \code{\link{op_convert_to_numpy}()} \cr
 \code{\link{op_convert_to_tensor}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -160,6 +165,7 @@ Other ops: \cr
 \code{\link{op_ctc_loss}()} \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_depthwise_conv}()} \cr
 \code{\link{op_det}()} \cr
 \code{\link{op_diag}()} \cr
@@ -199,10 +205,13 @@ Other ops: \cr
 \code{\link{op_glu}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
 \code{\link{op_hard_shrink}()} \cr
 \code{\link{op_hard_sigmoid}()} \cr
 \code{\link{op_hard_silu}()} \cr
 \code{\link{op_hard_tanh}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -229,6 +238,8 @@ Other ops: \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
 \code{\link{op_istft}()} \cr
+\code{\link{op_kaiser}()} \cr
+\code{\link{op_layer_normalization}()} \cr
 \code{\link{op_leaky_relu}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
@@ -328,6 +339,7 @@ Other ops: \cr
 \code{\link{op_sort}()} \cr
 \code{\link{op_sparse_categorical_crossentropy}()} \cr
 \code{\link{op_sparse_plus}()} \cr
+\code{\link{op_sparse_sigmoid}()} \cr
 \code{\link{op_sparsemax}()} \cr
 \code{\link{op_split}()} \cr
 \code{\link{op_sqrt}()} \cr
@@ -365,6 +377,8 @@ Other ops: \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
 \code{\link{op_vectorized_map}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_while_loop}()} \cr
diff --git a/man/op_deg2rad.Rd b/man/op_deg2rad.Rd
new file mode 100644
index 000000000..9854952ce
--- /dev/null
+++ b/man/op_deg2rad.Rd
@@ -0,0 +1,482 @@
+% Generated by roxygen2: do not edit by hand
+% Please edit documentation in R/ops-numpy.R
+\name{op_deg2rad}
+\alias{op_deg2rad}
+\title{Convert angles from degrees to radians.}
+\usage{
+op_deg2rad(x)
+}
+\arguments{
+\item{x}{Input tensor of angles in degrees.}
+}
+\value{
+A tensor containing angles converted to radians.
+}
+\description{
+The conversion is defined as:
+\code{rad = deg * (pi / 180)}.
+}
+\section{Examples}{
+\if{html}{\out{<div class="sourceCode r">}}\preformatted{op_deg2rad(c(0, 90, 180))
+}\if{html}{\out{</div>}}
+
+\if{html}{\out{<div class="sourceCode">}}\preformatted{## tf.Tensor([0.        1.5707964 3.1415927], shape=(3), dtype=float32)
+
+}\if{html}{\out{</div>}}
+}
+
+\seealso{
+Other numpy ops: \cr
+\code{\link{op_abs}()} \cr
+\code{\link{op_add}()} \cr
+\code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
+\code{\link{op_any}()} \cr
+\code{\link{op_append}()} \cr
+\code{\link{op_arange}()} \cr
+\code{\link{op_arccos}()} \cr
+\code{\link{op_arccosh}()} \cr
+\code{\link{op_arcsin}()} \cr
+\code{\link{op_arcsinh}()} \cr
+\code{\link{op_arctan}()} \cr
+\code{\link{op_arctan2}()} \cr
+\code{\link{op_arctanh}()} \cr
+\code{\link{op_argmax}()} \cr
+\code{\link{op_argmin}()} \cr
+\code{\link{op_argpartition}()} \cr
+\code{\link{op_argsort}()} \cr
+\code{\link{op_array}()} \cr
+\code{\link{op_average}()} \cr
+\code{\link{op_bartlett}()} \cr
+\code{\link{op_bincount}()} \cr
+\code{\link{op_bitwise_and}()} \cr
+\code{\link{op_bitwise_invert}()} \cr
+\code{\link{op_bitwise_left_shift}()} \cr
+\code{\link{op_bitwise_not}()} \cr
+\code{\link{op_bitwise_or}()} \cr
+\code{\link{op_bitwise_right_shift}()} \cr
+\code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
+\code{\link{op_broadcast_to}()} \cr
+\code{\link{op_cbrt}()} \cr
+\code{\link{op_ceil}()} \cr
+\code{\link{op_clip}()} \cr
+\code{\link{op_concatenate}()} \cr
+\code{\link{op_conj}()} \cr
+\code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
+\code{\link{op_correlate}()} \cr
+\code{\link{op_cos}()} \cr
+\code{\link{op_cosh}()} \cr
+\code{\link{op_count_nonzero}()} \cr
+\code{\link{op_cross}()} \cr
+\code{\link{op_ctc_decode}()} \cr
+\code{\link{op_cumprod}()} \cr
+\code{\link{op_cumsum}()} \cr
+\code{\link{op_diag}()} \cr
+\code{\link{op_diagflat}()} \cr
+\code{\link{op_diagonal}()} \cr
+\code{\link{op_diff}()} \cr
+\code{\link{op_digitize}()} \cr
+\code{\link{op_divide}()} \cr
+\code{\link{op_divide_no_nan}()} \cr
+\code{\link{op_dot}()} \cr
+\code{\link{op_einsum}()} \cr
+\code{\link{op_empty}()} \cr
+\code{\link{op_equal}()} \cr
+\code{\link{op_exp}()} \cr
+\code{\link{op_exp2}()} \cr
+\code{\link{op_expand_dims}()} \cr
+\code{\link{op_expm1}()} \cr
+\code{\link{op_eye}()} \cr
+\code{\link{op_flip}()} \cr
+\code{\link{op_floor}()} \cr
+\code{\link{op_floor_divide}()} \cr
+\code{\link{op_full}()} \cr
+\code{\link{op_full_like}()} \cr
+\code{\link{op_get_item}()} \cr
+\code{\link{op_greater}()} \cr
+\code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
+\code{\link{op_heaviside}()} \cr
+\code{\link{op_histogram}()} \cr
+\code{\link{op_hstack}()} \cr
+\code{\link{op_identity}()} \cr
+\code{\link{op_imag}()} \cr
+\code{\link{op_inner}()} \cr
+\code{\link{op_isclose}()} \cr
+\code{\link{op_isfinite}()} \cr
+\code{\link{op_isinf}()} \cr
+\code{\link{op_isnan}()} \cr
+\code{\link{op_kaiser}()} \cr
+\code{\link{op_left_shift}()} \cr
+\code{\link{op_less}()} \cr
+\code{\link{op_less_equal}()} \cr
+\code{\link{op_linspace}()} \cr
+\code{\link{op_log}()} \cr
+\code{\link{op_log10}()} \cr
+\code{\link{op_log1p}()} \cr
+\code{\link{op_log2}()} \cr
+\code{\link{op_logaddexp}()} \cr
+\code{\link{op_logdet}()} \cr
+\code{\link{op_logical_and}()} \cr
+\code{\link{op_logical_not}()} \cr
+\code{\link{op_logical_or}()} \cr
+\code{\link{op_logical_xor}()} \cr
+\code{\link{op_logspace}()} \cr
+\code{\link{op_lstsq}()} \cr
+\code{\link{op_matmul}()} \cr
+\code{\link{op_max}()} \cr
+\code{\link{op_maximum}()} \cr
+\code{\link{op_mean}()} \cr
+\code{\link{op_median}()} \cr
+\code{\link{op_meshgrid}()} \cr
+\code{\link{op_min}()} \cr
+\code{\link{op_minimum}()} \cr
+\code{\link{op_mod}()} \cr
+\code{\link{op_moveaxis}()} \cr
+\code{\link{op_multiply}()} \cr
+\code{\link{op_nan_to_num}()} \cr
+\code{\link{op_ndim}()} \cr
+\code{\link{op_negative}()} \cr
+\code{\link{op_nonzero}()} \cr
+\code{\link{op_not_equal}()} \cr
+\code{\link{op_ones}()} \cr
+\code{\link{op_ones_like}()} \cr
+\code{\link{op_outer}()} \cr
+\code{\link{op_pad}()} \cr
+\code{\link{op_power}()} \cr
+\code{\link{op_prod}()} \cr
+\code{\link{op_quantile}()} \cr
+\code{\link{op_ravel}()} \cr
+\code{\link{op_real}()} \cr
+\code{\link{op_reciprocal}()} \cr
+\code{\link{op_repeat}()} \cr
+\code{\link{op_reshape}()} \cr
+\code{\link{op_right_shift}()} \cr
+\code{\link{op_roll}()} \cr
+\code{\link{op_rot90}()} \cr
+\code{\link{op_round}()} \cr
+\code{\link{op_saturate_cast}()} \cr
+\code{\link{op_select}()} \cr
+\code{\link{op_sign}()} \cr
+\code{\link{op_signbit}()} \cr
+\code{\link{op_sin}()} \cr
+\code{\link{op_sinh}()} \cr
+\code{\link{op_size}()} \cr
+\code{\link{op_sort}()} \cr
+\code{\link{op_split}()} \cr
+\code{\link{op_sqrt}()} \cr
+\code{\link{op_square}()} \cr
+\code{\link{op_squeeze}()} \cr
+\code{\link{op_stack}()} \cr
+\code{\link{op_std}()} \cr
+\code{\link{op_subtract}()} \cr
+\code{\link{op_sum}()} \cr
+\code{\link{op_swapaxes}()} \cr
+\code{\link{op_take}()} \cr
+\code{\link{op_take_along_axis}()} \cr
+\code{\link{op_tan}()} \cr
+\code{\link{op_tanh}()} \cr
+\code{\link{op_tensordot}()} \cr
+\code{\link{op_tile}()} \cr
+\code{\link{op_trace}()} \cr
+\code{\link{op_transpose}()} \cr
+\code{\link{op_tri}()} \cr
+\code{\link{op_tril}()} \cr
+\code{\link{op_triu}()} \cr
+\code{\link{op_trunc}()} \cr
+\code{\link{op_var}()} \cr
+\code{\link{op_vdot}()} \cr
+\code{\link{op_vectorize}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
+\code{\link{op_vstack}()} \cr
+\code{\link{op_where}()} \cr
+\code{\link{op_zeros}()} \cr
+\code{\link{op_zeros_like}()} \cr
+
+Other ops: \cr
+\code{\link{op_abs}()} \cr
+\code{\link{op_add}()} \cr
+\code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
+\code{\link{op_any}()} \cr
+\code{\link{op_append}()} \cr
+\code{\link{op_arange}()} \cr
+\code{\link{op_arccos}()} \cr
+\code{\link{op_arccosh}()} \cr
+\code{\link{op_arcsin}()} \cr
+\code{\link{op_arcsinh}()} \cr
+\code{\link{op_arctan}()} \cr
+\code{\link{op_arctan2}()} \cr
+\code{\link{op_arctanh}()} \cr
+\code{\link{op_argmax}()} \cr
+\code{\link{op_argmin}()} \cr
+\code{\link{op_argpartition}()} \cr
+\code{\link{op_argsort}()} \cr
+\code{\link{op_array}()} \cr
+\code{\link{op_associative_scan}()} \cr
+\code{\link{op_average}()} \cr
+\code{\link{op_average_pool}()} \cr
+\code{\link{op_bartlett}()} \cr
+\code{\link{op_batch_normalization}()} \cr
+\code{\link{op_binary_crossentropy}()} \cr
+\code{\link{op_bincount}()} \cr
+\code{\link{op_bitwise_and}()} \cr
+\code{\link{op_bitwise_invert}()} \cr
+\code{\link{op_bitwise_left_shift}()} \cr
+\code{\link{op_bitwise_not}()} \cr
+\code{\link{op_bitwise_or}()} \cr
+\code{\link{op_bitwise_right_shift}()} \cr
+\code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
+\code{\link{op_broadcast_to}()} \cr
+\code{\link{op_cast}()} \cr
+\code{\link{op_categorical_crossentropy}()} \cr
+\code{\link{op_cbrt}()} \cr
+\code{\link{op_ceil}()} \cr
+\code{\link{op_celu}()} \cr
+\code{\link{op_cholesky}()} \cr
+\code{\link{op_clip}()} \cr
+\code{\link{op_concatenate}()} \cr
+\code{\link{op_cond}()} \cr
+\code{\link{op_conj}()} \cr
+\code{\link{op_conv}()} \cr
+\code{\link{op_conv_transpose}()} \cr
+\code{\link{op_convert_to_numpy}()} \cr
+\code{\link{op_convert_to_tensor}()} \cr
+\code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
+\code{\link{op_correlate}()} \cr
+\code{\link{op_cos}()} \cr
+\code{\link{op_cosh}()} \cr
+\code{\link{op_count_nonzero}()} \cr
+\code{\link{op_cross}()} \cr
+\code{\link{op_ctc_decode}()} \cr
+\code{\link{op_ctc_loss}()} \cr
+\code{\link{op_cumprod}()} \cr
+\code{\link{op_cumsum}()} \cr
+\code{\link{op_custom_gradient}()} \cr
+\code{\link{op_depthwise_conv}()} \cr
+\code{\link{op_det}()} \cr
+\code{\link{op_diag}()} \cr
+\code{\link{op_diagflat}()} \cr
+\code{\link{op_diagonal}()} \cr
+\code{\link{op_diff}()} \cr
+\code{\link{op_digitize}()} \cr
+\code{\link{op_divide}()} \cr
+\code{\link{op_divide_no_nan}()} \cr
+\code{\link{op_dot}()} \cr
+\code{\link{op_dot_product_attention}()} \cr
+\code{\link{op_dtype}()} \cr
+\code{\link{op_eig}()} \cr
+\code{\link{op_eigh}()} \cr
+\code{\link{op_einsum}()} \cr
+\code{\link{op_elu}()} \cr
+\code{\link{op_empty}()} \cr
+\code{\link{op_equal}()} \cr
+\code{\link{op_erf}()} \cr
+\code{\link{op_erfinv}()} \cr
+\code{\link{op_exp}()} \cr
+\code{\link{op_exp2}()} \cr
+\code{\link{op_expand_dims}()} \cr
+\code{\link{op_expm1}()} \cr
+\code{\link{op_extract_sequences}()} \cr
+\code{\link{op_eye}()} \cr
+\code{\link{op_fft}()} \cr
+\code{\link{op_fft2}()} \cr
+\code{\link{op_flip}()} \cr
+\code{\link{op_floor}()} \cr
+\code{\link{op_floor_divide}()} \cr
+\code{\link{op_fori_loop}()} \cr
+\code{\link{op_full}()} \cr
+\code{\link{op_full_like}()} \cr
+\code{\link{op_gelu}()} \cr
+\code{\link{op_get_item}()} \cr
+\code{\link{op_glu}()} \cr
+\code{\link{op_greater}()} \cr
+\code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
+\code{\link{op_hard_shrink}()} \cr
+\code{\link{op_hard_sigmoid}()} \cr
+\code{\link{op_hard_silu}()} \cr
+\code{\link{op_hard_tanh}()} \cr
+\code{\link{op_heaviside}()} \cr
+\code{\link{op_histogram}()} \cr
+\code{\link{op_hstack}()} \cr
+\code{\link{op_identity}()} \cr
+\code{\link{op_ifft2}()} \cr
+\code{\link{op_imag}()} \cr
+\code{\link{op_image_affine_transform}()} \cr
+\code{\link{op_image_crop}()} \cr
+\code{\link{op_image_extract_patches}()} \cr
+\code{\link{op_image_gaussian_blur}()} \cr
+\code{\link{op_image_hsv_to_rgb}()} \cr
+\code{\link{op_image_map_coordinates}()} \cr
+\code{\link{op_image_pad}()} \cr
+\code{\link{op_image_perspective_transform}()} \cr
+\code{\link{op_image_resize}()} \cr
+\code{\link{op_image_rgb_to_grayscale}()} \cr
+\code{\link{op_image_rgb_to_hsv}()} \cr
+\code{\link{op_in_top_k}()} \cr
+\code{\link{op_inner}()} \cr
+\code{\link{op_inv}()} \cr
+\code{\link{op_irfft}()} \cr
+\code{\link{op_is_tensor}()} \cr
+\code{\link{op_isclose}()} \cr
+\code{\link{op_isfinite}()} \cr
+\code{\link{op_isinf}()} \cr
+\code{\link{op_isnan}()} \cr
+\code{\link{op_istft}()} \cr
+\code{\link{op_kaiser}()} \cr
+\code{\link{op_layer_normalization}()} \cr
+\code{\link{op_leaky_relu}()} \cr
+\code{\link{op_left_shift}()} \cr
+\code{\link{op_less}()} \cr
+\code{\link{op_less_equal}()} \cr
+\code{\link{op_linspace}()} \cr
+\code{\link{op_log}()} \cr
+\code{\link{op_log10}()} \cr
+\code{\link{op_log1p}()} \cr
+\code{\link{op_log2}()} \cr
+\code{\link{op_log_sigmoid}()} \cr
+\code{\link{op_log_softmax}()} \cr
+\code{\link{op_logaddexp}()} \cr
+\code{\link{op_logdet}()} \cr
+\code{\link{op_logical_and}()} \cr
+\code{\link{op_logical_not}()} \cr
+\code{\link{op_logical_or}()} \cr
+\code{\link{op_logical_xor}()} \cr
+\code{\link{op_logspace}()} \cr
+\code{\link{op_logsumexp}()} \cr
+\code{\link{op_lstsq}()} \cr
+\code{\link{op_lu_factor}()} \cr
+\code{\link{op_map}()} \cr
+\code{\link{op_matmul}()} \cr
+\code{\link{op_max}()} \cr
+\code{\link{op_max_pool}()} \cr
+\code{\link{op_maximum}()} \cr
+\code{\link{op_mean}()} \cr
+\code{\link{op_median}()} \cr
+\code{\link{op_meshgrid}()} \cr
+\code{\link{op_min}()} \cr
+\code{\link{op_minimum}()} \cr
+\code{\link{op_mod}()} \cr
+\code{\link{op_moments}()} \cr
+\code{\link{op_moveaxis}()} \cr
+\code{\link{op_multi_hot}()} \cr
+\code{\link{op_multiply}()} \cr
+\code{\link{op_nan_to_num}()} \cr
+\code{\link{op_ndim}()} \cr
+\code{\link{op_negative}()} \cr
+\code{\link{op_nonzero}()} \cr
+\code{\link{op_norm}()} \cr
+\code{\link{op_normalize}()} \cr
+\code{\link{op_not_equal}()} \cr
+\code{\link{op_one_hot}()} \cr
+\code{\link{op_ones}()} \cr
+\code{\link{op_ones_like}()} \cr
+\code{\link{op_outer}()} \cr
+\code{\link{op_pad}()} \cr
+\code{\link{op_polar}()} \cr
+\code{\link{op_power}()} \cr
+\code{\link{op_prod}()} \cr
+\code{\link{op_psnr}()} \cr
+\code{\link{op_qr}()} \cr
+\code{\link{op_quantile}()} \cr
+\code{\link{op_ravel}()} \cr
+\code{\link{op_real}()} \cr
+\code{\link{op_rearrange}()} \cr
+\code{\link{op_reciprocal}()} \cr
+\code{\link{op_relu}()} \cr
+\code{\link{op_relu6}()} \cr
+\code{\link{op_repeat}()} \cr
+\code{\link{op_reshape}()} \cr
+\code{\link{op_rfft}()} \cr
+\code{\link{op_right_shift}()} \cr
+\code{\link{op_rms_normalization}()} \cr
+\code{\link{op_roll}()} \cr
+\code{\link{op_rot90}()} \cr
+\code{\link{op_round}()} \cr
+\code{\link{op_rsqrt}()} \cr
+\code{\link{op_saturate_cast}()} \cr
+\code{\link{op_scan}()} \cr
+\code{\link{op_scatter}()} \cr
+\code{\link{op_scatter_update}()} \cr
+\code{\link{op_searchsorted}()} \cr
+\code{\link{op_segment_max}()} \cr
+\code{\link{op_segment_sum}()} \cr
+\code{\link{op_select}()} \cr
+\code{\link{op_selu}()} \cr
+\code{\link{op_separable_conv}()} \cr
+\code{\link{op_shape}()} \cr
+\code{\link{op_sigmoid}()} \cr
+\code{\link{op_sign}()} \cr
+\code{\link{op_signbit}()} \cr
+\code{\link{op_silu}()} \cr
+\code{\link{op_sin}()} \cr
+\code{\link{op_sinh}()} \cr
+\code{\link{op_size}()} \cr
+\code{\link{op_slice}()} \cr
+\code{\link{op_slice_update}()} \cr
+\code{\link{op_slogdet}()} \cr
+\code{\link{op_soft_shrink}()} \cr
+\code{\link{op_softmax}()} \cr
+\code{\link{op_softplus}()} \cr
+\code{\link{op_softsign}()} \cr
+\code{\link{op_solve}()} \cr
+\code{\link{op_solve_triangular}()} \cr
+\code{\link{op_sort}()} \cr
+\code{\link{op_sparse_categorical_crossentropy}()} \cr
+\code{\link{op_sparse_plus}()} \cr
+\code{\link{op_sparse_sigmoid}()} \cr
+\code{\link{op_sparsemax}()} \cr
+\code{\link{op_split}()} \cr
+\code{\link{op_sqrt}()} \cr
+\code{\link{op_square}()} \cr
+\code{\link{op_squareplus}()} \cr
+\code{\link{op_squeeze}()} \cr
+\code{\link{op_stack}()} \cr
+\code{\link{op_std}()} \cr
+\code{\link{op_stft}()} \cr
+\code{\link{op_stop_gradient}()} \cr
+\code{\link{op_subset}()} \cr
+\code{\link{op_subtract}()} \cr
+\code{\link{op_sum}()} \cr
+\code{\link{op_svd}()} \cr
+\code{\link{op_swapaxes}()} \cr
+\code{\link{op_switch}()} \cr
+\code{\link{op_take}()} \cr
+\code{\link{op_take_along_axis}()} \cr
+\code{\link{op_tan}()} \cr
+\code{\link{op_tanh}()} \cr
+\code{\link{op_tanh_shrink}()} \cr
+\code{\link{op_tensordot}()} \cr
+\code{\link{op_threshold}()} \cr
+\code{\link{op_tile}()} \cr
+\code{\link{op_top_k}()} \cr
+\code{\link{op_trace}()} \cr
+\code{\link{op_transpose}()} \cr
+\code{\link{op_tri}()} \cr
+\code{\link{op_tril}()} \cr
+\code{\link{op_triu}()} \cr
+\code{\link{op_trunc}()} \cr
+\code{\link{op_unravel_index}()} \cr
+\code{\link{op_unstack}()} \cr
+\code{\link{op_var}()} \cr
+\code{\link{op_vdot}()} \cr
+\code{\link{op_vectorize}()} \cr
+\code{\link{op_vectorized_map}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
+\code{\link{op_vstack}()} \cr
+\code{\link{op_where}()} \cr
+\code{\link{op_while_loop}()} \cr
+\code{\link{op_zeros}()} \cr
+\code{\link{op_zeros_like}()} \cr
+}
+\concept{numpy ops}
+\concept{ops}
diff --git a/man/op_depthwise_conv.Rd b/man/op_depthwise_conv.Rd
index 5189e3a33..b07e96d7c 100644
--- a/man/op_depthwise_conv.Rd
+++ b/man/op_depthwise_conv.Rd
@@ -75,6 +75,7 @@ Other nn ops: \cr
 \code{\link{op_hard_sigmoid}()} \cr
 \code{\link{op_hard_silu}()} \cr
 \code{\link{op_hard_tanh}()} \cr
+\code{\link{op_layer_normalization}()} \cr
 \code{\link{op_leaky_relu}()} \cr
 \code{\link{op_log_sigmoid}()} \cr
 \code{\link{op_log_softmax}()} \cr
@@ -98,6 +99,7 @@ Other nn ops: \cr
 \code{\link{op_softsign}()} \cr
 \code{\link{op_sparse_categorical_crossentropy}()} \cr
 \code{\link{op_sparse_plus}()} \cr
+\code{\link{op_sparse_sigmoid}()} \cr
 \code{\link{op_sparsemax}()} \cr
 \code{\link{op_squareplus}()} \cr
 \code{\link{op_tanh_shrink}()} \cr
@@ -108,6 +110,7 @@ Other ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -126,6 +129,7 @@ Other ops: \cr
 \code{\link{op_associative_scan}()} \cr
 \code{\link{op_average}()} \cr
 \code{\link{op_average_pool}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_batch_normalization}()} \cr
 \code{\link{op_binary_crossentropy}()} \cr
 \code{\link{op_bincount}()} \cr
@@ -136,9 +140,11 @@ Other ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
 \code{\link{op_cast}()} \cr
 \code{\link{op_categorical_crossentropy}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_celu}()} \cr
 \code{\link{op_cholesky}()} \cr
@@ -151,6 +157,7 @@ Other ops: \cr
 \code{\link{op_convert_to_numpy}()} \cr
 \code{\link{op_convert_to_tensor}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -161,6 +168,7 @@ Other ops: \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
 \code{\link{op_custom_gradient}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_det}()} \cr
 \code{\link{op_diag}()} \cr
 \code{\link{op_diagflat}()} \cr
@@ -199,10 +207,13 @@ Other ops: \cr
 \code{\link{op_glu}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
 \code{\link{op_hard_shrink}()} \cr
 \code{\link{op_hard_sigmoid}()} \cr
 \code{\link{op_hard_silu}()} \cr
 \code{\link{op_hard_tanh}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -229,6 +240,8 @@ Other ops: \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
 \code{\link{op_istft}()} \cr
+\code{\link{op_kaiser}()} \cr
+\code{\link{op_layer_normalization}()} \cr
 \code{\link{op_leaky_relu}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
@@ -328,6 +341,7 @@ Other ops: \cr
 \code{\link{op_sort}()} \cr
 \code{\link{op_sparse_categorical_crossentropy}()} \cr
 \code{\link{op_sparse_plus}()} \cr
+\code{\link{op_sparse_sigmoid}()} \cr
 \code{\link{op_sparsemax}()} \cr
 \code{\link{op_split}()} \cr
 \code{\link{op_sqrt}()} \cr
@@ -365,6 +379,8 @@ Other ops: \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
 \code{\link{op_vectorized_map}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_while_loop}()} \cr
diff --git a/man/op_det.Rd b/man/op_det.Rd
index 302d4513e..ea12db935 100644
--- a/man/op_det.Rd
+++ b/man/op_det.Rd
@@ -32,6 +32,7 @@ Other ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -50,6 +51,7 @@ Other ops: \cr
 \code{\link{op_associative_scan}()} \cr
 \code{\link{op_average}()} \cr
 \code{\link{op_average_pool}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_batch_normalization}()} \cr
 \code{\link{op_binary_crossentropy}()} \cr
 \code{\link{op_bincount}()} \cr
@@ -60,9 +62,11 @@ Other ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
 \code{\link{op_cast}()} \cr
 \code{\link{op_categorical_crossentropy}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_celu}()} \cr
 \code{\link{op_cholesky}()} \cr
@@ -75,6 +79,7 @@ Other ops: \cr
 \code{\link{op_convert_to_numpy}()} \cr
 \code{\link{op_convert_to_tensor}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -85,6 +90,7 @@ Other ops: \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
 \code{\link{op_custom_gradient}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_depthwise_conv}()} \cr
 \code{\link{op_diag}()} \cr
 \code{\link{op_diagflat}()} \cr
@@ -123,10 +129,13 @@ Other ops: \cr
 \code{\link{op_glu}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
 \code{\link{op_hard_shrink}()} \cr
 \code{\link{op_hard_sigmoid}()} \cr
 \code{\link{op_hard_silu}()} \cr
 \code{\link{op_hard_tanh}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -153,6 +162,8 @@ Other ops: \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
 \code{\link{op_istft}()} \cr
+\code{\link{op_kaiser}()} \cr
+\code{\link{op_layer_normalization}()} \cr
 \code{\link{op_leaky_relu}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
@@ -252,6 +263,7 @@ Other ops: \cr
 \code{\link{op_sort}()} \cr
 \code{\link{op_sparse_categorical_crossentropy}()} \cr
 \code{\link{op_sparse_plus}()} \cr
+\code{\link{op_sparse_sigmoid}()} \cr
 \code{\link{op_sparsemax}()} \cr
 \code{\link{op_split}()} \cr
 \code{\link{op_sqrt}()} \cr
@@ -289,6 +301,8 @@ Other ops: \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
 \code{\link{op_vectorized_map}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_while_loop}()} \cr
diff --git a/man/op_diag.Rd b/man/op_diag.Rd
index 35a236f91..bafb03cd2 100644
--- a/man/op_diag.Rd
+++ b/man/op_diag.Rd
@@ -73,6 +73,7 @@ Other numpy ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -89,6 +90,7 @@ Other numpy ops: \cr
 \code{\link{op_argsort}()} \cr
 \code{\link{op_array}()} \cr
 \code{\link{op_average}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_bincount}()} \cr
 \code{\link{op_bitwise_and}()} \cr
 \code{\link{op_bitwise_invert}()} \cr
@@ -97,12 +99,15 @@ Other numpy ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_clip}()} \cr
 \code{\link{op_concatenate}()} \cr
 \code{\link{op_conj}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -111,6 +116,7 @@ Other numpy ops: \cr
 \code{\link{op_ctc_decode}()} \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_diagflat}()} \cr
 \code{\link{op_diagonal}()} \cr
 \code{\link{op_diff}()} \cr
@@ -134,6 +140,9 @@ Other numpy ops: \cr
 \code{\link{op_get_item}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -143,6 +152,7 @@ Other numpy ops: \cr
 \code{\link{op_isfinite}()} \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
+\code{\link{op_kaiser}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
 \code{\link{op_less_equal}()} \cr
@@ -223,6 +233,8 @@ Other numpy ops: \cr
 \code{\link{op_var}()} \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_zeros}()} \cr
@@ -232,6 +244,7 @@ Other ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -250,6 +263,7 @@ Other ops: \cr
 \code{\link{op_associative_scan}()} \cr
 \code{\link{op_average}()} \cr
 \code{\link{op_average_pool}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_batch_normalization}()} \cr
 \code{\link{op_binary_crossentropy}()} \cr
 \code{\link{op_bincount}()} \cr
@@ -260,9 +274,11 @@ Other ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
 \code{\link{op_cast}()} \cr
 \code{\link{op_categorical_crossentropy}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_celu}()} \cr
 \code{\link{op_cholesky}()} \cr
@@ -275,6 +291,7 @@ Other ops: \cr
 \code{\link{op_convert_to_numpy}()} \cr
 \code{\link{op_convert_to_tensor}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -285,6 +302,7 @@ Other ops: \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
 \code{\link{op_custom_gradient}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_depthwise_conv}()} \cr
 \code{\link{op_det}()} \cr
 \code{\link{op_diagflat}()} \cr
@@ -323,10 +341,13 @@ Other ops: \cr
 \code{\link{op_glu}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
 \code{\link{op_hard_shrink}()} \cr
 \code{\link{op_hard_sigmoid}()} \cr
 \code{\link{op_hard_silu}()} \cr
 \code{\link{op_hard_tanh}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -353,6 +374,8 @@ Other ops: \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
 \code{\link{op_istft}()} \cr
+\code{\link{op_kaiser}()} \cr
+\code{\link{op_layer_normalization}()} \cr
 \code{\link{op_leaky_relu}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
@@ -452,6 +475,7 @@ Other ops: \cr
 \code{\link{op_sort}()} \cr
 \code{\link{op_sparse_categorical_crossentropy}()} \cr
 \code{\link{op_sparse_plus}()} \cr
+\code{\link{op_sparse_sigmoid}()} \cr
 \code{\link{op_sparsemax}()} \cr
 \code{\link{op_split}()} \cr
 \code{\link{op_sqrt}()} \cr
@@ -489,6 +513,8 @@ Other ops: \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
 \code{\link{op_vectorized_map}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_while_loop}()} \cr
diff --git a/man/op_diagflat.Rd b/man/op_diagflat.Rd
index 57d7ee3a4..ed0438dad 100644
--- a/man/op_diagflat.Rd
+++ b/man/op_diagflat.Rd
@@ -24,6 +24,7 @@ Other numpy ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -40,6 +41,7 @@ Other numpy ops: \cr
 \code{\link{op_argsort}()} \cr
 \code{\link{op_array}()} \cr
 \code{\link{op_average}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_bincount}()} \cr
 \code{\link{op_bitwise_and}()} \cr
 \code{\link{op_bitwise_invert}()} \cr
@@ -48,12 +50,15 @@ Other numpy ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_clip}()} \cr
 \code{\link{op_concatenate}()} \cr
 \code{\link{op_conj}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -62,6 +67,7 @@ Other numpy ops: \cr
 \code{\link{op_ctc_decode}()} \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_diag}()} \cr
 \code{\link{op_diagonal}()} \cr
 \code{\link{op_diff}()} \cr
@@ -85,6 +91,9 @@ Other numpy ops: \cr
 \code{\link{op_get_item}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -94,6 +103,7 @@ Other numpy ops: \cr
 \code{\link{op_isfinite}()} \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
+\code{\link{op_kaiser}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
 \code{\link{op_less_equal}()} \cr
@@ -174,6 +184,8 @@ Other numpy ops: \cr
 \code{\link{op_var}()} \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_zeros}()} \cr
@@ -183,6 +195,7 @@ Other ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -201,6 +214,7 @@ Other ops: \cr
 \code{\link{op_associative_scan}()} \cr
 \code{\link{op_average}()} \cr
 \code{\link{op_average_pool}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_batch_normalization}()} \cr
 \code{\link{op_binary_crossentropy}()} \cr
 \code{\link{op_bincount}()} \cr
@@ -211,9 +225,11 @@ Other ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
 \code{\link{op_cast}()} \cr
 \code{\link{op_categorical_crossentropy}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_celu}()} \cr
 \code{\link{op_cholesky}()} \cr
@@ -226,6 +242,7 @@ Other ops: \cr
 \code{\link{op_convert_to_numpy}()} \cr
 \code{\link{op_convert_to_tensor}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -236,6 +253,7 @@ Other ops: \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
 \code{\link{op_custom_gradient}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_depthwise_conv}()} \cr
 \code{\link{op_det}()} \cr
 \code{\link{op_diag}()} \cr
@@ -274,10 +292,13 @@ Other ops: \cr
 \code{\link{op_glu}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
 \code{\link{op_hard_shrink}()} \cr
 \code{\link{op_hard_sigmoid}()} \cr
 \code{\link{op_hard_silu}()} \cr
 \code{\link{op_hard_tanh}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -304,6 +325,8 @@ Other ops: \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
 \code{\link{op_istft}()} \cr
+\code{\link{op_kaiser}()} \cr
+\code{\link{op_layer_normalization}()} \cr
 \code{\link{op_leaky_relu}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
@@ -403,6 +426,7 @@ Other ops: \cr
 \code{\link{op_sort}()} \cr
 \code{\link{op_sparse_categorical_crossentropy}()} \cr
 \code{\link{op_sparse_plus}()} \cr
+\code{\link{op_sparse_sigmoid}()} \cr
 \code{\link{op_sparsemax}()} \cr
 \code{\link{op_split}()} \cr
 \code{\link{op_sqrt}()} \cr
@@ -440,6 +464,8 @@ Other ops: \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
 \code{\link{op_vectorized_map}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_while_loop}()} \cr
diff --git a/man/op_diagonal.Rd b/man/op_diagonal.Rd
index 563933490..5fe76b69a 100644
--- a/man/op_diagonal.Rd
+++ b/man/op_diagonal.Rd
@@ -108,6 +108,7 @@ Other numpy ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -124,6 +125,7 @@ Other numpy ops: \cr
 \code{\link{op_argsort}()} \cr
 \code{\link{op_array}()} \cr
 \code{\link{op_average}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_bincount}()} \cr
 \code{\link{op_bitwise_and}()} \cr
 \code{\link{op_bitwise_invert}()} \cr
@@ -132,12 +134,15 @@ Other numpy ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_clip}()} \cr
 \code{\link{op_concatenate}()} \cr
 \code{\link{op_conj}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -146,6 +151,7 @@ Other numpy ops: \cr
 \code{\link{op_ctc_decode}()} \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_diag}()} \cr
 \code{\link{op_diagflat}()} \cr
 \code{\link{op_diff}()} \cr
@@ -169,6 +175,9 @@ Other numpy ops: \cr
 \code{\link{op_get_item}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -178,6 +187,7 @@ Other numpy ops: \cr
 \code{\link{op_isfinite}()} \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
+\code{\link{op_kaiser}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
 \code{\link{op_less_equal}()} \cr
@@ -258,6 +268,8 @@ Other numpy ops: \cr
 \code{\link{op_var}()} \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_zeros}()} \cr
@@ -267,6 +279,7 @@ Other ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -285,6 +298,7 @@ Other ops: \cr
 \code{\link{op_associative_scan}()} \cr
 \code{\link{op_average}()} \cr
 \code{\link{op_average_pool}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_batch_normalization}()} \cr
 \code{\link{op_binary_crossentropy}()} \cr
 \code{\link{op_bincount}()} \cr
@@ -295,9 +309,11 @@ Other ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
 \code{\link{op_cast}()} \cr
 \code{\link{op_categorical_crossentropy}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_celu}()} \cr
 \code{\link{op_cholesky}()} \cr
@@ -310,6 +326,7 @@ Other ops: \cr
 \code{\link{op_convert_to_numpy}()} \cr
 \code{\link{op_convert_to_tensor}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -320,6 +337,7 @@ Other ops: \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
 \code{\link{op_custom_gradient}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_depthwise_conv}()} \cr
 \code{\link{op_det}()} \cr
 \code{\link{op_diag}()} \cr
@@ -358,10 +376,13 @@ Other ops: \cr
 \code{\link{op_glu}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
 \code{\link{op_hard_shrink}()} \cr
 \code{\link{op_hard_sigmoid}()} \cr
 \code{\link{op_hard_silu}()} \cr
 \code{\link{op_hard_tanh}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -388,6 +409,8 @@ Other ops: \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
 \code{\link{op_istft}()} \cr
+\code{\link{op_kaiser}()} \cr
+\code{\link{op_layer_normalization}()} \cr
 \code{\link{op_leaky_relu}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
@@ -487,6 +510,7 @@ Other ops: \cr
 \code{\link{op_sort}()} \cr
 \code{\link{op_sparse_categorical_crossentropy}()} \cr
 \code{\link{op_sparse_plus}()} \cr
+\code{\link{op_sparse_sigmoid}()} \cr
 \code{\link{op_sparsemax}()} \cr
 \code{\link{op_split}()} \cr
 \code{\link{op_sqrt}()} \cr
@@ -524,6 +548,8 @@ Other ops: \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
 \code{\link{op_vectorized_map}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_while_loop}()} \cr
diff --git a/man/op_diff.Rd b/man/op_diff.Rd
index d650a7f15..e770a3d3a 100644
--- a/man/op_diff.Rd
+++ b/man/op_diff.Rd
@@ -62,6 +62,7 @@ Other numpy ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -78,6 +79,7 @@ Other numpy ops: \cr
 \code{\link{op_argsort}()} \cr
 \code{\link{op_array}()} \cr
 \code{\link{op_average}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_bincount}()} \cr
 \code{\link{op_bitwise_and}()} \cr
 \code{\link{op_bitwise_invert}()} \cr
@@ -86,12 +88,15 @@ Other numpy ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_clip}()} \cr
 \code{\link{op_concatenate}()} \cr
 \code{\link{op_conj}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -100,6 +105,7 @@ Other numpy ops: \cr
 \code{\link{op_ctc_decode}()} \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_diag}()} \cr
 \code{\link{op_diagflat}()} \cr
 \code{\link{op_diagonal}()} \cr
@@ -123,6 +129,9 @@ Other numpy ops: \cr
 \code{\link{op_get_item}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -132,6 +141,7 @@ Other numpy ops: \cr
 \code{\link{op_isfinite}()} \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
+\code{\link{op_kaiser}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
 \code{\link{op_less_equal}()} \cr
@@ -212,6 +222,8 @@ Other numpy ops: \cr
 \code{\link{op_var}()} \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_zeros}()} \cr
@@ -221,6 +233,7 @@ Other ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -239,6 +252,7 @@ Other ops: \cr
 \code{\link{op_associative_scan}()} \cr
 \code{\link{op_average}()} \cr
 \code{\link{op_average_pool}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_batch_normalization}()} \cr
 \code{\link{op_binary_crossentropy}()} \cr
 \code{\link{op_bincount}()} \cr
@@ -249,9 +263,11 @@ Other ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
 \code{\link{op_cast}()} \cr
 \code{\link{op_categorical_crossentropy}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_celu}()} \cr
 \code{\link{op_cholesky}()} \cr
@@ -264,6 +280,7 @@ Other ops: \cr
 \code{\link{op_convert_to_numpy}()} \cr
 \code{\link{op_convert_to_tensor}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -274,6 +291,7 @@ Other ops: \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
 \code{\link{op_custom_gradient}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_depthwise_conv}()} \cr
 \code{\link{op_det}()} \cr
 \code{\link{op_diag}()} \cr
@@ -312,10 +330,13 @@ Other ops: \cr
 \code{\link{op_glu}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
 \code{\link{op_hard_shrink}()} \cr
 \code{\link{op_hard_sigmoid}()} \cr
 \code{\link{op_hard_silu}()} \cr
 \code{\link{op_hard_tanh}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -342,6 +363,8 @@ Other ops: \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
 \code{\link{op_istft}()} \cr
+\code{\link{op_kaiser}()} \cr
+\code{\link{op_layer_normalization}()} \cr
 \code{\link{op_leaky_relu}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
@@ -441,6 +464,7 @@ Other ops: \cr
 \code{\link{op_sort}()} \cr
 \code{\link{op_sparse_categorical_crossentropy}()} \cr
 \code{\link{op_sparse_plus}()} \cr
+\code{\link{op_sparse_sigmoid}()} \cr
 \code{\link{op_sparsemax}()} \cr
 \code{\link{op_split}()} \cr
 \code{\link{op_sqrt}()} \cr
@@ -478,6 +502,8 @@ Other ops: \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
 \code{\link{op_vectorized_map}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_while_loop}()} \cr
diff --git a/man/op_digitize.Rd b/man/op_digitize.Rd
index 6c82c1625..3834ae275 100644
--- a/man/op_digitize.Rd
+++ b/man/op_digitize.Rd
@@ -42,6 +42,7 @@ Other numpy ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -58,6 +59,7 @@ Other numpy ops: \cr
 \code{\link{op_argsort}()} \cr
 \code{\link{op_array}()} \cr
 \code{\link{op_average}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_bincount}()} \cr
 \code{\link{op_bitwise_and}()} \cr
 \code{\link{op_bitwise_invert}()} \cr
@@ -66,12 +68,15 @@ Other numpy ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_clip}()} \cr
 \code{\link{op_concatenate}()} \cr
 \code{\link{op_conj}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -80,6 +85,7 @@ Other numpy ops: \cr
 \code{\link{op_ctc_decode}()} \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_diag}()} \cr
 \code{\link{op_diagflat}()} \cr
 \code{\link{op_diagonal}()} \cr
@@ -103,6 +109,9 @@ Other numpy ops: \cr
 \code{\link{op_get_item}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -112,6 +121,7 @@ Other numpy ops: \cr
 \code{\link{op_isfinite}()} \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
+\code{\link{op_kaiser}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
 \code{\link{op_less_equal}()} \cr
@@ -192,6 +202,8 @@ Other numpy ops: \cr
 \code{\link{op_var}()} \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_zeros}()} \cr
@@ -201,6 +213,7 @@ Other ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -219,6 +232,7 @@ Other ops: \cr
 \code{\link{op_associative_scan}()} \cr
 \code{\link{op_average}()} \cr
 \code{\link{op_average_pool}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_batch_normalization}()} \cr
 \code{\link{op_binary_crossentropy}()} \cr
 \code{\link{op_bincount}()} \cr
@@ -229,9 +243,11 @@ Other ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
 \code{\link{op_cast}()} \cr
 \code{\link{op_categorical_crossentropy}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_celu}()} \cr
 \code{\link{op_cholesky}()} \cr
@@ -244,6 +260,7 @@ Other ops: \cr
 \code{\link{op_convert_to_numpy}()} \cr
 \code{\link{op_convert_to_tensor}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -254,6 +271,7 @@ Other ops: \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
 \code{\link{op_custom_gradient}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_depthwise_conv}()} \cr
 \code{\link{op_det}()} \cr
 \code{\link{op_diag}()} \cr
@@ -292,10 +310,13 @@ Other ops: \cr
 \code{\link{op_glu}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
 \code{\link{op_hard_shrink}()} \cr
 \code{\link{op_hard_sigmoid}()} \cr
 \code{\link{op_hard_silu}()} \cr
 \code{\link{op_hard_tanh}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -322,6 +343,8 @@ Other ops: \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
 \code{\link{op_istft}()} \cr
+\code{\link{op_kaiser}()} \cr
+\code{\link{op_layer_normalization}()} \cr
 \code{\link{op_leaky_relu}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
@@ -421,6 +444,7 @@ Other ops: \cr
 \code{\link{op_sort}()} \cr
 \code{\link{op_sparse_categorical_crossentropy}()} \cr
 \code{\link{op_sparse_plus}()} \cr
+\code{\link{op_sparse_sigmoid}()} \cr
 \code{\link{op_sparsemax}()} \cr
 \code{\link{op_split}()} \cr
 \code{\link{op_sqrt}()} \cr
@@ -458,6 +482,8 @@ Other ops: \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
 \code{\link{op_vectorized_map}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_while_loop}()} \cr
diff --git a/man/op_divide.Rd b/man/op_divide.Rd
index 4620ce237..adcc04282 100644
--- a/man/op_divide.Rd
+++ b/man/op_divide.Rd
@@ -56,6 +56,7 @@ Other numpy ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -72,6 +73,7 @@ Other numpy ops: \cr
 \code{\link{op_argsort}()} \cr
 \code{\link{op_array}()} \cr
 \code{\link{op_average}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_bincount}()} \cr
 \code{\link{op_bitwise_and}()} \cr
 \code{\link{op_bitwise_invert}()} \cr
@@ -80,12 +82,15 @@ Other numpy ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_clip}()} \cr
 \code{\link{op_concatenate}()} \cr
 \code{\link{op_conj}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -94,6 +99,7 @@ Other numpy ops: \cr
 \code{\link{op_ctc_decode}()} \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_diag}()} \cr
 \code{\link{op_diagflat}()} \cr
 \code{\link{op_diagonal}()} \cr
@@ -117,6 +123,9 @@ Other numpy ops: \cr
 \code{\link{op_get_item}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -126,6 +135,7 @@ Other numpy ops: \cr
 \code{\link{op_isfinite}()} \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
+\code{\link{op_kaiser}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
 \code{\link{op_less_equal}()} \cr
@@ -206,6 +216,8 @@ Other numpy ops: \cr
 \code{\link{op_var}()} \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_zeros}()} \cr
@@ -215,6 +227,7 @@ Other ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -233,6 +246,7 @@ Other ops: \cr
 \code{\link{op_associative_scan}()} \cr
 \code{\link{op_average}()} \cr
 \code{\link{op_average_pool}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_batch_normalization}()} \cr
 \code{\link{op_binary_crossentropy}()} \cr
 \code{\link{op_bincount}()} \cr
@@ -243,9 +257,11 @@ Other ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
 \code{\link{op_cast}()} \cr
 \code{\link{op_categorical_crossentropy}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_celu}()} \cr
 \code{\link{op_cholesky}()} \cr
@@ -258,6 +274,7 @@ Other ops: \cr
 \code{\link{op_convert_to_numpy}()} \cr
 \code{\link{op_convert_to_tensor}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -268,6 +285,7 @@ Other ops: \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
 \code{\link{op_custom_gradient}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_depthwise_conv}()} \cr
 \code{\link{op_det}()} \cr
 \code{\link{op_diag}()} \cr
@@ -306,10 +324,13 @@ Other ops: \cr
 \code{\link{op_glu}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
 \code{\link{op_hard_shrink}()} \cr
 \code{\link{op_hard_sigmoid}()} \cr
 \code{\link{op_hard_silu}()} \cr
 \code{\link{op_hard_tanh}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -336,6 +357,8 @@ Other ops: \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
 \code{\link{op_istft}()} \cr
+\code{\link{op_kaiser}()} \cr
+\code{\link{op_layer_normalization}()} \cr
 \code{\link{op_leaky_relu}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
@@ -435,6 +458,7 @@ Other ops: \cr
 \code{\link{op_sort}()} \cr
 \code{\link{op_sparse_categorical_crossentropy}()} \cr
 \code{\link{op_sparse_plus}()} \cr
+\code{\link{op_sparse_sigmoid}()} \cr
 \code{\link{op_sparsemax}()} \cr
 \code{\link{op_split}()} \cr
 \code{\link{op_sqrt}()} \cr
@@ -472,6 +496,8 @@ Other ops: \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
 \code{\link{op_vectorized_map}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_while_loop}()} \cr
diff --git a/man/op_divide_no_nan.Rd b/man/op_divide_no_nan.Rd
index 9aacfe0fa..b87b0923e 100644
--- a/man/op_divide_no_nan.Rd
+++ b/man/op_divide_no_nan.Rd
@@ -22,6 +22,7 @@ Other numpy ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -38,6 +39,7 @@ Other numpy ops: \cr
 \code{\link{op_argsort}()} \cr
 \code{\link{op_array}()} \cr
 \code{\link{op_average}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_bincount}()} \cr
 \code{\link{op_bitwise_and}()} \cr
 \code{\link{op_bitwise_invert}()} \cr
@@ -46,12 +48,15 @@ Other numpy ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_clip}()} \cr
 \code{\link{op_concatenate}()} \cr
 \code{\link{op_conj}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -60,6 +65,7 @@ Other numpy ops: \cr
 \code{\link{op_ctc_decode}()} \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_diag}()} \cr
 \code{\link{op_diagflat}()} \cr
 \code{\link{op_diagonal}()} \cr
@@ -83,6 +89,9 @@ Other numpy ops: \cr
 \code{\link{op_get_item}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -92,6 +101,7 @@ Other numpy ops: \cr
 \code{\link{op_isfinite}()} \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
+\code{\link{op_kaiser}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
 \code{\link{op_less_equal}()} \cr
@@ -172,6 +182,8 @@ Other numpy ops: \cr
 \code{\link{op_var}()} \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_zeros}()} \cr
@@ -181,6 +193,7 @@ Other ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -199,6 +212,7 @@ Other ops: \cr
 \code{\link{op_associative_scan}()} \cr
 \code{\link{op_average}()} \cr
 \code{\link{op_average_pool}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_batch_normalization}()} \cr
 \code{\link{op_binary_crossentropy}()} \cr
 \code{\link{op_bincount}()} \cr
@@ -209,9 +223,11 @@ Other ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
 \code{\link{op_cast}()} \cr
 \code{\link{op_categorical_crossentropy}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_celu}()} \cr
 \code{\link{op_cholesky}()} \cr
@@ -224,6 +240,7 @@ Other ops: \cr
 \code{\link{op_convert_to_numpy}()} \cr
 \code{\link{op_convert_to_tensor}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -234,6 +251,7 @@ Other ops: \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
 \code{\link{op_custom_gradient}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_depthwise_conv}()} \cr
 \code{\link{op_det}()} \cr
 \code{\link{op_diag}()} \cr
@@ -272,10 +290,13 @@ Other ops: \cr
 \code{\link{op_glu}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
 \code{\link{op_hard_shrink}()} \cr
 \code{\link{op_hard_sigmoid}()} \cr
 \code{\link{op_hard_silu}()} \cr
 \code{\link{op_hard_tanh}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -302,6 +323,8 @@ Other ops: \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
 \code{\link{op_istft}()} \cr
+\code{\link{op_kaiser}()} \cr
+\code{\link{op_layer_normalization}()} \cr
 \code{\link{op_leaky_relu}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
@@ -401,6 +424,7 @@ Other ops: \cr
 \code{\link{op_sort}()} \cr
 \code{\link{op_sparse_categorical_crossentropy}()} \cr
 \code{\link{op_sparse_plus}()} \cr
+\code{\link{op_sparse_sigmoid}()} \cr
 \code{\link{op_sparsemax}()} \cr
 \code{\link{op_split}()} \cr
 \code{\link{op_sqrt}()} \cr
@@ -438,6 +462,8 @@ Other ops: \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
 \code{\link{op_vectorized_map}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_while_loop}()} \cr
diff --git a/man/op_dot.Rd b/man/op_dot.Rd
index 15ac05575..fae5f346a 100644
--- a/man/op_dot.Rd
+++ b/man/op_dot.Rd
@@ -40,6 +40,7 @@ Other numpy ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -56,6 +57,7 @@ Other numpy ops: \cr
 \code{\link{op_argsort}()} \cr
 \code{\link{op_array}()} \cr
 \code{\link{op_average}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_bincount}()} \cr
 \code{\link{op_bitwise_and}()} \cr
 \code{\link{op_bitwise_invert}()} \cr
@@ -64,12 +66,15 @@ Other numpy ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_clip}()} \cr
 \code{\link{op_concatenate}()} \cr
 \code{\link{op_conj}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -78,6 +83,7 @@ Other numpy ops: \cr
 \code{\link{op_ctc_decode}()} \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_diag}()} \cr
 \code{\link{op_diagflat}()} \cr
 \code{\link{op_diagonal}()} \cr
@@ -101,6 +107,9 @@ Other numpy ops: \cr
 \code{\link{op_get_item}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -110,6 +119,7 @@ Other numpy ops: \cr
 \code{\link{op_isfinite}()} \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
+\code{\link{op_kaiser}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
 \code{\link{op_less_equal}()} \cr
@@ -190,6 +200,8 @@ Other numpy ops: \cr
 \code{\link{op_var}()} \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_zeros}()} \cr
@@ -199,6 +211,7 @@ Other ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -217,6 +230,7 @@ Other ops: \cr
 \code{\link{op_associative_scan}()} \cr
 \code{\link{op_average}()} \cr
 \code{\link{op_average_pool}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_batch_normalization}()} \cr
 \code{\link{op_binary_crossentropy}()} \cr
 \code{\link{op_bincount}()} \cr
@@ -227,9 +241,11 @@ Other ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
 \code{\link{op_cast}()} \cr
 \code{\link{op_categorical_crossentropy}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_celu}()} \cr
 \code{\link{op_cholesky}()} \cr
@@ -242,6 +258,7 @@ Other ops: \cr
 \code{\link{op_convert_to_numpy}()} \cr
 \code{\link{op_convert_to_tensor}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -252,6 +269,7 @@ Other ops: \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
 \code{\link{op_custom_gradient}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_depthwise_conv}()} \cr
 \code{\link{op_det}()} \cr
 \code{\link{op_diag}()} \cr
@@ -290,10 +308,13 @@ Other ops: \cr
 \code{\link{op_glu}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
 \code{\link{op_hard_shrink}()} \cr
 \code{\link{op_hard_sigmoid}()} \cr
 \code{\link{op_hard_silu}()} \cr
 \code{\link{op_hard_tanh}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -320,6 +341,8 @@ Other ops: \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
 \code{\link{op_istft}()} \cr
+\code{\link{op_kaiser}()} \cr
+\code{\link{op_layer_normalization}()} \cr
 \code{\link{op_leaky_relu}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
@@ -419,6 +442,7 @@ Other ops: \cr
 \code{\link{op_sort}()} \cr
 \code{\link{op_sparse_categorical_crossentropy}()} \cr
 \code{\link{op_sparse_plus}()} \cr
+\code{\link{op_sparse_sigmoid}()} \cr
 \code{\link{op_sparsemax}()} \cr
 \code{\link{op_split}()} \cr
 \code{\link{op_sqrt}()} \cr
@@ -456,6 +480,8 @@ Other ops: \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
 \code{\link{op_vectorized_map}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_while_loop}()} \cr
diff --git a/man/op_dot_product_attention.Rd b/man/op_dot_product_attention.Rd
index 5d02417a1..b6fdea637 100644
--- a/man/op_dot_product_attention.Rd
+++ b/man/op_dot_product_attention.Rd
@@ -98,6 +98,7 @@ Other nn ops: \cr
 \code{\link{op_hard_sigmoid}()} \cr
 \code{\link{op_hard_silu}()} \cr
 \code{\link{op_hard_tanh}()} \cr
+\code{\link{op_layer_normalization}()} \cr
 \code{\link{op_leaky_relu}()} \cr
 \code{\link{op_log_sigmoid}()} \cr
 \code{\link{op_log_softmax}()} \cr
@@ -121,6 +122,7 @@ Other nn ops: \cr
 \code{\link{op_softsign}()} \cr
 \code{\link{op_sparse_categorical_crossentropy}()} \cr
 \code{\link{op_sparse_plus}()} \cr
+\code{\link{op_sparse_sigmoid}()} \cr
 \code{\link{op_sparsemax}()} \cr
 \code{\link{op_squareplus}()} \cr
 \code{\link{op_tanh_shrink}()} \cr
@@ -131,6 +133,7 @@ Other ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -149,6 +152,7 @@ Other ops: \cr
 \code{\link{op_associative_scan}()} \cr
 \code{\link{op_average}()} \cr
 \code{\link{op_average_pool}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_batch_normalization}()} \cr
 \code{\link{op_binary_crossentropy}()} \cr
 \code{\link{op_bincount}()} \cr
@@ -159,9 +163,11 @@ Other ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
 \code{\link{op_cast}()} \cr
 \code{\link{op_categorical_crossentropy}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_celu}()} \cr
 \code{\link{op_cholesky}()} \cr
@@ -174,6 +180,7 @@ Other ops: \cr
 \code{\link{op_convert_to_numpy}()} \cr
 \code{\link{op_convert_to_tensor}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -184,6 +191,7 @@ Other ops: \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
 \code{\link{op_custom_gradient}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_depthwise_conv}()} \cr
 \code{\link{op_det}()} \cr
 \code{\link{op_diag}()} \cr
@@ -222,10 +230,13 @@ Other ops: \cr
 \code{\link{op_glu}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
 \code{\link{op_hard_shrink}()} \cr
 \code{\link{op_hard_sigmoid}()} \cr
 \code{\link{op_hard_silu}()} \cr
 \code{\link{op_hard_tanh}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -252,6 +263,8 @@ Other ops: \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
 \code{\link{op_istft}()} \cr
+\code{\link{op_kaiser}()} \cr
+\code{\link{op_layer_normalization}()} \cr
 \code{\link{op_leaky_relu}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
@@ -351,6 +364,7 @@ Other ops: \cr
 \code{\link{op_sort}()} \cr
 \code{\link{op_sparse_categorical_crossentropy}()} \cr
 \code{\link{op_sparse_plus}()} \cr
+\code{\link{op_sparse_sigmoid}()} \cr
 \code{\link{op_sparsemax}()} \cr
 \code{\link{op_split}()} \cr
 \code{\link{op_sqrt}()} \cr
@@ -388,6 +402,8 @@ Other ops: \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
 \code{\link{op_vectorized_map}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_while_loop}()} \cr
diff --git a/man/op_dtype.Rd b/man/op_dtype.Rd
index e6d3f94c2..8a8d46458 100644
--- a/man/op_dtype.Rd
+++ b/man/op_dtype.Rd
@@ -57,6 +57,7 @@ Other ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -75,6 +76,7 @@ Other ops: \cr
 \code{\link{op_associative_scan}()} \cr
 \code{\link{op_average}()} \cr
 \code{\link{op_average_pool}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_batch_normalization}()} \cr
 \code{\link{op_binary_crossentropy}()} \cr
 \code{\link{op_bincount}()} \cr
@@ -85,9 +87,11 @@ Other ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
 \code{\link{op_cast}()} \cr
 \code{\link{op_categorical_crossentropy}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_celu}()} \cr
 \code{\link{op_cholesky}()} \cr
@@ -100,6 +104,7 @@ Other ops: \cr
 \code{\link{op_convert_to_numpy}()} \cr
 \code{\link{op_convert_to_tensor}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -110,6 +115,7 @@ Other ops: \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
 \code{\link{op_custom_gradient}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_depthwise_conv}()} \cr
 \code{\link{op_det}()} \cr
 \code{\link{op_diag}()} \cr
@@ -148,10 +154,13 @@ Other ops: \cr
 \code{\link{op_glu}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
 \code{\link{op_hard_shrink}()} \cr
 \code{\link{op_hard_sigmoid}()} \cr
 \code{\link{op_hard_silu}()} \cr
 \code{\link{op_hard_tanh}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -178,6 +187,8 @@ Other ops: \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
 \code{\link{op_istft}()} \cr
+\code{\link{op_kaiser}()} \cr
+\code{\link{op_layer_normalization}()} \cr
 \code{\link{op_leaky_relu}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
@@ -277,6 +288,7 @@ Other ops: \cr
 \code{\link{op_sort}()} \cr
 \code{\link{op_sparse_categorical_crossentropy}()} \cr
 \code{\link{op_sparse_plus}()} \cr
+\code{\link{op_sparse_sigmoid}()} \cr
 \code{\link{op_sparsemax}()} \cr
 \code{\link{op_split}()} \cr
 \code{\link{op_sqrt}()} \cr
@@ -314,6 +326,8 @@ Other ops: \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
 \code{\link{op_vectorized_map}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_while_loop}()} \cr
diff --git a/man/op_eig.Rd b/man/op_eig.Rd
index ab61a6461..be85283b8 100644
--- a/man/op_eig.Rd
+++ b/man/op_eig.Rd
@@ -33,6 +33,7 @@ Other ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -51,6 +52,7 @@ Other ops: \cr
 \code{\link{op_associative_scan}()} \cr
 \code{\link{op_average}()} \cr
 \code{\link{op_average_pool}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_batch_normalization}()} \cr
 \code{\link{op_binary_crossentropy}()} \cr
 \code{\link{op_bincount}()} \cr
@@ -61,9 +63,11 @@ Other ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
 \code{\link{op_cast}()} \cr
 \code{\link{op_categorical_crossentropy}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_celu}()} \cr
 \code{\link{op_cholesky}()} \cr
@@ -76,6 +80,7 @@ Other ops: \cr
 \code{\link{op_convert_to_numpy}()} \cr
 \code{\link{op_convert_to_tensor}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -86,6 +91,7 @@ Other ops: \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
 \code{\link{op_custom_gradient}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_depthwise_conv}()} \cr
 \code{\link{op_det}()} \cr
 \code{\link{op_diag}()} \cr
@@ -124,10 +130,13 @@ Other ops: \cr
 \code{\link{op_glu}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
 \code{\link{op_hard_shrink}()} \cr
 \code{\link{op_hard_sigmoid}()} \cr
 \code{\link{op_hard_silu}()} \cr
 \code{\link{op_hard_tanh}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -154,6 +163,8 @@ Other ops: \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
 \code{\link{op_istft}()} \cr
+\code{\link{op_kaiser}()} \cr
+\code{\link{op_layer_normalization}()} \cr
 \code{\link{op_leaky_relu}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
@@ -253,6 +264,7 @@ Other ops: \cr
 \code{\link{op_sort}()} \cr
 \code{\link{op_sparse_categorical_crossentropy}()} \cr
 \code{\link{op_sparse_plus}()} \cr
+\code{\link{op_sparse_sigmoid}()} \cr
 \code{\link{op_sparsemax}()} \cr
 \code{\link{op_split}()} \cr
 \code{\link{op_sqrt}()} \cr
@@ -290,6 +302,8 @@ Other ops: \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
 \code{\link{op_vectorized_map}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_while_loop}()} \cr
diff --git a/man/op_eigh.Rd b/man/op_eigh.Rd
index 669de0ab7..c4974423d 100644
--- a/man/op_eigh.Rd
+++ b/man/op_eigh.Rd
@@ -33,6 +33,7 @@ Other ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -51,6 +52,7 @@ Other ops: \cr
 \code{\link{op_associative_scan}()} \cr
 \code{\link{op_average}()} \cr
 \code{\link{op_average_pool}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_batch_normalization}()} \cr
 \code{\link{op_binary_crossentropy}()} \cr
 \code{\link{op_bincount}()} \cr
@@ -61,9 +63,11 @@ Other ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
 \code{\link{op_cast}()} \cr
 \code{\link{op_categorical_crossentropy}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_celu}()} \cr
 \code{\link{op_cholesky}()} \cr
@@ -76,6 +80,7 @@ Other ops: \cr
 \code{\link{op_convert_to_numpy}()} \cr
 \code{\link{op_convert_to_tensor}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -86,6 +91,7 @@ Other ops: \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
 \code{\link{op_custom_gradient}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_depthwise_conv}()} \cr
 \code{\link{op_det}()} \cr
 \code{\link{op_diag}()} \cr
@@ -124,10 +130,13 @@ Other ops: \cr
 \code{\link{op_glu}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
 \code{\link{op_hard_shrink}()} \cr
 \code{\link{op_hard_sigmoid}()} \cr
 \code{\link{op_hard_silu}()} \cr
 \code{\link{op_hard_tanh}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -154,6 +163,8 @@ Other ops: \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
 \code{\link{op_istft}()} \cr
+\code{\link{op_kaiser}()} \cr
+\code{\link{op_layer_normalization}()} \cr
 \code{\link{op_leaky_relu}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
@@ -253,6 +264,7 @@ Other ops: \cr
 \code{\link{op_sort}()} \cr
 \code{\link{op_sparse_categorical_crossentropy}()} \cr
 \code{\link{op_sparse_plus}()} \cr
+\code{\link{op_sparse_sigmoid}()} \cr
 \code{\link{op_sparsemax}()} \cr
 \code{\link{op_split}()} \cr
 \code{\link{op_sqrt}()} \cr
@@ -290,6 +302,8 @@ Other ops: \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
 \code{\link{op_vectorized_map}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_while_loop}()} \cr
diff --git a/man/op_einsum.Rd b/man/op_einsum.Rd
index 4fb6cb228..760ab2123 100644
--- a/man/op_einsum.Rd
+++ b/man/op_einsum.Rd
@@ -124,6 +124,7 @@ Other numpy ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -140,6 +141,7 @@ Other numpy ops: \cr
 \code{\link{op_argsort}()} \cr
 \code{\link{op_array}()} \cr
 \code{\link{op_average}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_bincount}()} \cr
 \code{\link{op_bitwise_and}()} \cr
 \code{\link{op_bitwise_invert}()} \cr
@@ -148,12 +150,15 @@ Other numpy ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_clip}()} \cr
 \code{\link{op_concatenate}()} \cr
 \code{\link{op_conj}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -162,6 +167,7 @@ Other numpy ops: \cr
 \code{\link{op_ctc_decode}()} \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_diag}()} \cr
 \code{\link{op_diagflat}()} \cr
 \code{\link{op_diagonal}()} \cr
@@ -185,6 +191,9 @@ Other numpy ops: \cr
 \code{\link{op_get_item}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -194,6 +203,7 @@ Other numpy ops: \cr
 \code{\link{op_isfinite}()} \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
+\code{\link{op_kaiser}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
 \code{\link{op_less_equal}()} \cr
@@ -274,6 +284,8 @@ Other numpy ops: \cr
 \code{\link{op_var}()} \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_zeros}()} \cr
@@ -283,6 +295,7 @@ Other ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -301,6 +314,7 @@ Other ops: \cr
 \code{\link{op_associative_scan}()} \cr
 \code{\link{op_average}()} \cr
 \code{\link{op_average_pool}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_batch_normalization}()} \cr
 \code{\link{op_binary_crossentropy}()} \cr
 \code{\link{op_bincount}()} \cr
@@ -311,9 +325,11 @@ Other ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
 \code{\link{op_cast}()} \cr
 \code{\link{op_categorical_crossentropy}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_celu}()} \cr
 \code{\link{op_cholesky}()} \cr
@@ -326,6 +342,7 @@ Other ops: \cr
 \code{\link{op_convert_to_numpy}()} \cr
 \code{\link{op_convert_to_tensor}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -336,6 +353,7 @@ Other ops: \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
 \code{\link{op_custom_gradient}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_depthwise_conv}()} \cr
 \code{\link{op_det}()} \cr
 \code{\link{op_diag}()} \cr
@@ -374,10 +392,13 @@ Other ops: \cr
 \code{\link{op_glu}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
 \code{\link{op_hard_shrink}()} \cr
 \code{\link{op_hard_sigmoid}()} \cr
 \code{\link{op_hard_silu}()} \cr
 \code{\link{op_hard_tanh}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -404,6 +425,8 @@ Other ops: \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
 \code{\link{op_istft}()} \cr
+\code{\link{op_kaiser}()} \cr
+\code{\link{op_layer_normalization}()} \cr
 \code{\link{op_leaky_relu}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
@@ -503,6 +526,7 @@ Other ops: \cr
 \code{\link{op_sort}()} \cr
 \code{\link{op_sparse_categorical_crossentropy}()} \cr
 \code{\link{op_sparse_plus}()} \cr
+\code{\link{op_sparse_sigmoid}()} \cr
 \code{\link{op_sparsemax}()} \cr
 \code{\link{op_split}()} \cr
 \code{\link{op_sqrt}()} \cr
@@ -540,6 +564,8 @@ Other ops: \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
 \code{\link{op_vectorized_map}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_while_loop}()} \cr
diff --git a/man/op_elu.Rd b/man/op_elu.Rd
index 3429b38e2..2a282eec6 100644
--- a/man/op_elu.Rd
+++ b/man/op_elu.Rd
@@ -51,6 +51,7 @@ Other nn ops: \cr
 \code{\link{op_hard_sigmoid}()} \cr
 \code{\link{op_hard_silu}()} \cr
 \code{\link{op_hard_tanh}()} \cr
+\code{\link{op_layer_normalization}()} \cr
 \code{\link{op_leaky_relu}()} \cr
 \code{\link{op_log_sigmoid}()} \cr
 \code{\link{op_log_softmax}()} \cr
@@ -74,6 +75,7 @@ Other nn ops: \cr
 \code{\link{op_softsign}()} \cr
 \code{\link{op_sparse_categorical_crossentropy}()} \cr
 \code{\link{op_sparse_plus}()} \cr
+\code{\link{op_sparse_sigmoid}()} \cr
 \code{\link{op_sparsemax}()} \cr
 \code{\link{op_squareplus}()} \cr
 \code{\link{op_tanh_shrink}()} \cr
@@ -84,6 +86,7 @@ Other ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -102,6 +105,7 @@ Other ops: \cr
 \code{\link{op_associative_scan}()} \cr
 \code{\link{op_average}()} \cr
 \code{\link{op_average_pool}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_batch_normalization}()} \cr
 \code{\link{op_binary_crossentropy}()} \cr
 \code{\link{op_bincount}()} \cr
@@ -112,9 +116,11 @@ Other ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
 \code{\link{op_cast}()} \cr
 \code{\link{op_categorical_crossentropy}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_celu}()} \cr
 \code{\link{op_cholesky}()} \cr
@@ -127,6 +133,7 @@ Other ops: \cr
 \code{\link{op_convert_to_numpy}()} \cr
 \code{\link{op_convert_to_tensor}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -137,6 +144,7 @@ Other ops: \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
 \code{\link{op_custom_gradient}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_depthwise_conv}()} \cr
 \code{\link{op_det}()} \cr
 \code{\link{op_diag}()} \cr
@@ -175,10 +183,13 @@ Other ops: \cr
 \code{\link{op_glu}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
 \code{\link{op_hard_shrink}()} \cr
 \code{\link{op_hard_sigmoid}()} \cr
 \code{\link{op_hard_silu}()} \cr
 \code{\link{op_hard_tanh}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -205,6 +216,8 @@ Other ops: \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
 \code{\link{op_istft}()} \cr
+\code{\link{op_kaiser}()} \cr
+\code{\link{op_layer_normalization}()} \cr
 \code{\link{op_leaky_relu}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
@@ -304,6 +317,7 @@ Other ops: \cr
 \code{\link{op_sort}()} \cr
 \code{\link{op_sparse_categorical_crossentropy}()} \cr
 \code{\link{op_sparse_plus}()} \cr
+\code{\link{op_sparse_sigmoid}()} \cr
 \code{\link{op_sparsemax}()} \cr
 \code{\link{op_split}()} \cr
 \code{\link{op_sqrt}()} \cr
@@ -341,6 +355,8 @@ Other ops: \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
 \code{\link{op_vectorized_map}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_while_loop}()} \cr
diff --git a/man/op_empty.Rd b/man/op_empty.Rd
index 1978067d1..989090d72 100644
--- a/man/op_empty.Rd
+++ b/man/op_empty.Rd
@@ -26,6 +26,7 @@ Other numpy ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -42,6 +43,7 @@ Other numpy ops: \cr
 \code{\link{op_argsort}()} \cr
 \code{\link{op_array}()} \cr
 \code{\link{op_average}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_bincount}()} \cr
 \code{\link{op_bitwise_and}()} \cr
 \code{\link{op_bitwise_invert}()} \cr
@@ -50,12 +52,15 @@ Other numpy ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_clip}()} \cr
 \code{\link{op_concatenate}()} \cr
 \code{\link{op_conj}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -64,6 +69,7 @@ Other numpy ops: \cr
 \code{\link{op_ctc_decode}()} \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_diag}()} \cr
 \code{\link{op_diagflat}()} \cr
 \code{\link{op_diagonal}()} \cr
@@ -87,6 +93,9 @@ Other numpy ops: \cr
 \code{\link{op_get_item}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -96,6 +105,7 @@ Other numpy ops: \cr
 \code{\link{op_isfinite}()} \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
+\code{\link{op_kaiser}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
 \code{\link{op_less_equal}()} \cr
@@ -176,6 +186,8 @@ Other numpy ops: \cr
 \code{\link{op_var}()} \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_zeros}()} \cr
@@ -185,6 +197,7 @@ Other ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -203,6 +216,7 @@ Other ops: \cr
 \code{\link{op_associative_scan}()} \cr
 \code{\link{op_average}()} \cr
 \code{\link{op_average_pool}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_batch_normalization}()} \cr
 \code{\link{op_binary_crossentropy}()} \cr
 \code{\link{op_bincount}()} \cr
@@ -213,9 +227,11 @@ Other ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
 \code{\link{op_cast}()} \cr
 \code{\link{op_categorical_crossentropy}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_celu}()} \cr
 \code{\link{op_cholesky}()} \cr
@@ -228,6 +244,7 @@ Other ops: \cr
 \code{\link{op_convert_to_numpy}()} \cr
 \code{\link{op_convert_to_tensor}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -238,6 +255,7 @@ Other ops: \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
 \code{\link{op_custom_gradient}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_depthwise_conv}()} \cr
 \code{\link{op_det}()} \cr
 \code{\link{op_diag}()} \cr
@@ -276,10 +294,13 @@ Other ops: \cr
 \code{\link{op_glu}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
 \code{\link{op_hard_shrink}()} \cr
 \code{\link{op_hard_sigmoid}()} \cr
 \code{\link{op_hard_silu}()} \cr
 \code{\link{op_hard_tanh}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -306,6 +327,8 @@ Other ops: \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
 \code{\link{op_istft}()} \cr
+\code{\link{op_kaiser}()} \cr
+\code{\link{op_layer_normalization}()} \cr
 \code{\link{op_leaky_relu}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
@@ -405,6 +428,7 @@ Other ops: \cr
 \code{\link{op_sort}()} \cr
 \code{\link{op_sparse_categorical_crossentropy}()} \cr
 \code{\link{op_sparse_plus}()} \cr
+\code{\link{op_sparse_sigmoid}()} \cr
 \code{\link{op_sparsemax}()} \cr
 \code{\link{op_split}()} \cr
 \code{\link{op_sqrt}()} \cr
@@ -442,6 +466,8 @@ Other ops: \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
 \code{\link{op_vectorized_map}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_while_loop}()} \cr
diff --git a/man/op_equal.Rd b/man/op_equal.Rd
index e89f6f564..a734ef0e9 100644
--- a/man/op_equal.Rd
+++ b/man/op_equal.Rd
@@ -47,6 +47,7 @@ Other numpy ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -63,6 +64,7 @@ Other numpy ops: \cr
 \code{\link{op_argsort}()} \cr
 \code{\link{op_array}()} \cr
 \code{\link{op_average}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_bincount}()} \cr
 \code{\link{op_bitwise_and}()} \cr
 \code{\link{op_bitwise_invert}()} \cr
@@ -71,12 +73,15 @@ Other numpy ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_clip}()} \cr
 \code{\link{op_concatenate}()} \cr
 \code{\link{op_conj}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -85,6 +90,7 @@ Other numpy ops: \cr
 \code{\link{op_ctc_decode}()} \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_diag}()} \cr
 \code{\link{op_diagflat}()} \cr
 \code{\link{op_diagonal}()} \cr
@@ -108,6 +114,9 @@ Other numpy ops: \cr
 \code{\link{op_get_item}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -117,6 +126,7 @@ Other numpy ops: \cr
 \code{\link{op_isfinite}()} \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
+\code{\link{op_kaiser}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
 \code{\link{op_less_equal}()} \cr
@@ -197,6 +207,8 @@ Other numpy ops: \cr
 \code{\link{op_var}()} \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_zeros}()} \cr
@@ -206,6 +218,7 @@ Other ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -224,6 +237,7 @@ Other ops: \cr
 \code{\link{op_associative_scan}()} \cr
 \code{\link{op_average}()} \cr
 \code{\link{op_average_pool}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_batch_normalization}()} \cr
 \code{\link{op_binary_crossentropy}()} \cr
 \code{\link{op_bincount}()} \cr
@@ -234,9 +248,11 @@ Other ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
 \code{\link{op_cast}()} \cr
 \code{\link{op_categorical_crossentropy}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_celu}()} \cr
 \code{\link{op_cholesky}()} \cr
@@ -249,6 +265,7 @@ Other ops: \cr
 \code{\link{op_convert_to_numpy}()} \cr
 \code{\link{op_convert_to_tensor}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -259,6 +276,7 @@ Other ops: \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
 \code{\link{op_custom_gradient}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_depthwise_conv}()} \cr
 \code{\link{op_det}()} \cr
 \code{\link{op_diag}()} \cr
@@ -297,10 +315,13 @@ Other ops: \cr
 \code{\link{op_glu}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
 \code{\link{op_hard_shrink}()} \cr
 \code{\link{op_hard_sigmoid}()} \cr
 \code{\link{op_hard_silu}()} \cr
 \code{\link{op_hard_tanh}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -327,6 +348,8 @@ Other ops: \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
 \code{\link{op_istft}()} \cr
+\code{\link{op_kaiser}()} \cr
+\code{\link{op_layer_normalization}()} \cr
 \code{\link{op_leaky_relu}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
@@ -426,6 +449,7 @@ Other ops: \cr
 \code{\link{op_sort}()} \cr
 \code{\link{op_sparse_categorical_crossentropy}()} \cr
 \code{\link{op_sparse_plus}()} \cr
+\code{\link{op_sparse_sigmoid}()} \cr
 \code{\link{op_sparsemax}()} \cr
 \code{\link{op_split}()} \cr
 \code{\link{op_sqrt}()} \cr
@@ -463,6 +487,8 @@ Other ops: \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
 \code{\link{op_vectorized_map}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_while_loop}()} \cr
diff --git a/man/op_erf.Rd b/man/op_erf.Rd
index f20512b68..af02afbc7 100644
--- a/man/op_erf.Rd
+++ b/man/op_erf.Rd
@@ -52,6 +52,7 @@ Other ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -70,6 +71,7 @@ Other ops: \cr
 \code{\link{op_associative_scan}()} \cr
 \code{\link{op_average}()} \cr
 \code{\link{op_average_pool}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_batch_normalization}()} \cr
 \code{\link{op_binary_crossentropy}()} \cr
 \code{\link{op_bincount}()} \cr
@@ -80,9 +82,11 @@ Other ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
 \code{\link{op_cast}()} \cr
 \code{\link{op_categorical_crossentropy}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_celu}()} \cr
 \code{\link{op_cholesky}()} \cr
@@ -95,6 +99,7 @@ Other ops: \cr
 \code{\link{op_convert_to_numpy}()} \cr
 \code{\link{op_convert_to_tensor}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -105,6 +110,7 @@ Other ops: \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
 \code{\link{op_custom_gradient}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_depthwise_conv}()} \cr
 \code{\link{op_det}()} \cr
 \code{\link{op_diag}()} \cr
@@ -143,10 +149,13 @@ Other ops: \cr
 \code{\link{op_glu}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
 \code{\link{op_hard_shrink}()} \cr
 \code{\link{op_hard_sigmoid}()} \cr
 \code{\link{op_hard_silu}()} \cr
 \code{\link{op_hard_tanh}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -173,6 +182,8 @@ Other ops: \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
 \code{\link{op_istft}()} \cr
+\code{\link{op_kaiser}()} \cr
+\code{\link{op_layer_normalization}()} \cr
 \code{\link{op_leaky_relu}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
@@ -272,6 +283,7 @@ Other ops: \cr
 \code{\link{op_sort}()} \cr
 \code{\link{op_sparse_categorical_crossentropy}()} \cr
 \code{\link{op_sparse_plus}()} \cr
+\code{\link{op_sparse_sigmoid}()} \cr
 \code{\link{op_sparsemax}()} \cr
 \code{\link{op_split}()} \cr
 \code{\link{op_sqrt}()} \cr
@@ -309,6 +321,8 @@ Other ops: \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
 \code{\link{op_vectorized_map}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_while_loop}()} \cr
diff --git a/man/op_erfinv.Rd b/man/op_erfinv.Rd
index d4f10ef66..0c2a9781b 100644
--- a/man/op_erfinv.Rd
+++ b/man/op_erfinv.Rd
@@ -49,6 +49,7 @@ Other ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -67,6 +68,7 @@ Other ops: \cr
 \code{\link{op_associative_scan}()} \cr
 \code{\link{op_average}()} \cr
 \code{\link{op_average_pool}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_batch_normalization}()} \cr
 \code{\link{op_binary_crossentropy}()} \cr
 \code{\link{op_bincount}()} \cr
@@ -77,9 +79,11 @@ Other ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
 \code{\link{op_cast}()} \cr
 \code{\link{op_categorical_crossentropy}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_celu}()} \cr
 \code{\link{op_cholesky}()} \cr
@@ -92,6 +96,7 @@ Other ops: \cr
 \code{\link{op_convert_to_numpy}()} \cr
 \code{\link{op_convert_to_tensor}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -102,6 +107,7 @@ Other ops: \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
 \code{\link{op_custom_gradient}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_depthwise_conv}()} \cr
 \code{\link{op_det}()} \cr
 \code{\link{op_diag}()} \cr
@@ -140,10 +146,13 @@ Other ops: \cr
 \code{\link{op_glu}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
 \code{\link{op_hard_shrink}()} \cr
 \code{\link{op_hard_sigmoid}()} \cr
 \code{\link{op_hard_silu}()} \cr
 \code{\link{op_hard_tanh}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -170,6 +179,8 @@ Other ops: \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
 \code{\link{op_istft}()} \cr
+\code{\link{op_kaiser}()} \cr
+\code{\link{op_layer_normalization}()} \cr
 \code{\link{op_leaky_relu}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
@@ -269,6 +280,7 @@ Other ops: \cr
 \code{\link{op_sort}()} \cr
 \code{\link{op_sparse_categorical_crossentropy}()} \cr
 \code{\link{op_sparse_plus}()} \cr
+\code{\link{op_sparse_sigmoid}()} \cr
 \code{\link{op_sparsemax}()} \cr
 \code{\link{op_split}()} \cr
 \code{\link{op_sqrt}()} \cr
@@ -306,6 +318,8 @@ Other ops: \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
 \code{\link{op_vectorized_map}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_while_loop}()} \cr
diff --git a/man/op_exp.Rd b/man/op_exp.Rd
index 2714d8c46..14fb8d648 100644
--- a/man/op_exp.Rd
+++ b/man/op_exp.Rd
@@ -24,6 +24,7 @@ Other numpy ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -40,6 +41,7 @@ Other numpy ops: \cr
 \code{\link{op_argsort}()} \cr
 \code{\link{op_array}()} \cr
 \code{\link{op_average}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_bincount}()} \cr
 \code{\link{op_bitwise_and}()} \cr
 \code{\link{op_bitwise_invert}()} \cr
@@ -48,12 +50,15 @@ Other numpy ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_clip}()} \cr
 \code{\link{op_concatenate}()} \cr
 \code{\link{op_conj}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -62,6 +67,7 @@ Other numpy ops: \cr
 \code{\link{op_ctc_decode}()} \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_diag}()} \cr
 \code{\link{op_diagflat}()} \cr
 \code{\link{op_diagonal}()} \cr
@@ -85,6 +91,9 @@ Other numpy ops: \cr
 \code{\link{op_get_item}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -94,6 +103,7 @@ Other numpy ops: \cr
 \code{\link{op_isfinite}()} \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
+\code{\link{op_kaiser}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
 \code{\link{op_less_equal}()} \cr
@@ -174,6 +184,8 @@ Other numpy ops: \cr
 \code{\link{op_var}()} \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_zeros}()} \cr
@@ -183,6 +195,7 @@ Other ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -201,6 +214,7 @@ Other ops: \cr
 \code{\link{op_associative_scan}()} \cr
 \code{\link{op_average}()} \cr
 \code{\link{op_average_pool}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_batch_normalization}()} \cr
 \code{\link{op_binary_crossentropy}()} \cr
 \code{\link{op_bincount}()} \cr
@@ -211,9 +225,11 @@ Other ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
 \code{\link{op_cast}()} \cr
 \code{\link{op_categorical_crossentropy}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_celu}()} \cr
 \code{\link{op_cholesky}()} \cr
@@ -226,6 +242,7 @@ Other ops: \cr
 \code{\link{op_convert_to_numpy}()} \cr
 \code{\link{op_convert_to_tensor}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -236,6 +253,7 @@ Other ops: \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
 \code{\link{op_custom_gradient}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_depthwise_conv}()} \cr
 \code{\link{op_det}()} \cr
 \code{\link{op_diag}()} \cr
@@ -274,10 +292,13 @@ Other ops: \cr
 \code{\link{op_glu}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
 \code{\link{op_hard_shrink}()} \cr
 \code{\link{op_hard_sigmoid}()} \cr
 \code{\link{op_hard_silu}()} \cr
 \code{\link{op_hard_tanh}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -304,6 +325,8 @@ Other ops: \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
 \code{\link{op_istft}()} \cr
+\code{\link{op_kaiser}()} \cr
+\code{\link{op_layer_normalization}()} \cr
 \code{\link{op_leaky_relu}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
@@ -403,6 +426,7 @@ Other ops: \cr
 \code{\link{op_sort}()} \cr
 \code{\link{op_sparse_categorical_crossentropy}()} \cr
 \code{\link{op_sparse_plus}()} \cr
+\code{\link{op_sparse_sigmoid}()} \cr
 \code{\link{op_sparsemax}()} \cr
 \code{\link{op_split}()} \cr
 \code{\link{op_sqrt}()} \cr
@@ -440,6 +464,8 @@ Other ops: \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
 \code{\link{op_vectorized_map}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_while_loop}()} \cr
diff --git a/man/op_exp2.Rd b/man/op_exp2.Rd
index 2d3e68be9..851091fea 100644
--- a/man/op_exp2.Rd
+++ b/man/op_exp2.Rd
@@ -20,6 +20,7 @@ Other numpy ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -36,6 +37,7 @@ Other numpy ops: \cr
 \code{\link{op_argsort}()} \cr
 \code{\link{op_array}()} \cr
 \code{\link{op_average}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_bincount}()} \cr
 \code{\link{op_bitwise_and}()} \cr
 \code{\link{op_bitwise_invert}()} \cr
@@ -44,12 +46,15 @@ Other numpy ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_clip}()} \cr
 \code{\link{op_concatenate}()} \cr
 \code{\link{op_conj}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -58,6 +63,7 @@ Other numpy ops: \cr
 \code{\link{op_ctc_decode}()} \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_diag}()} \cr
 \code{\link{op_diagflat}()} \cr
 \code{\link{op_diagonal}()} \cr
@@ -81,6 +87,9 @@ Other numpy ops: \cr
 \code{\link{op_get_item}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -90,6 +99,7 @@ Other numpy ops: \cr
 \code{\link{op_isfinite}()} \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
+\code{\link{op_kaiser}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
 \code{\link{op_less_equal}()} \cr
@@ -170,6 +180,8 @@ Other numpy ops: \cr
 \code{\link{op_var}()} \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_zeros}()} \cr
@@ -179,6 +191,7 @@ Other ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -197,6 +210,7 @@ Other ops: \cr
 \code{\link{op_associative_scan}()} \cr
 \code{\link{op_average}()} \cr
 \code{\link{op_average_pool}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_batch_normalization}()} \cr
 \code{\link{op_binary_crossentropy}()} \cr
 \code{\link{op_bincount}()} \cr
@@ -207,9 +221,11 @@ Other ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
 \code{\link{op_cast}()} \cr
 \code{\link{op_categorical_crossentropy}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_celu}()} \cr
 \code{\link{op_cholesky}()} \cr
@@ -222,6 +238,7 @@ Other ops: \cr
 \code{\link{op_convert_to_numpy}()} \cr
 \code{\link{op_convert_to_tensor}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -232,6 +249,7 @@ Other ops: \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
 \code{\link{op_custom_gradient}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_depthwise_conv}()} \cr
 \code{\link{op_det}()} \cr
 \code{\link{op_diag}()} \cr
@@ -270,10 +288,13 @@ Other ops: \cr
 \code{\link{op_glu}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
 \code{\link{op_hard_shrink}()} \cr
 \code{\link{op_hard_sigmoid}()} \cr
 \code{\link{op_hard_silu}()} \cr
 \code{\link{op_hard_tanh}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -300,6 +321,8 @@ Other ops: \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
 \code{\link{op_istft}()} \cr
+\code{\link{op_kaiser}()} \cr
+\code{\link{op_layer_normalization}()} \cr
 \code{\link{op_leaky_relu}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
@@ -399,6 +422,7 @@ Other ops: \cr
 \code{\link{op_sort}()} \cr
 \code{\link{op_sparse_categorical_crossentropy}()} \cr
 \code{\link{op_sparse_plus}()} \cr
+\code{\link{op_sparse_sigmoid}()} \cr
 \code{\link{op_sparsemax}()} \cr
 \code{\link{op_split}()} \cr
 \code{\link{op_sqrt}()} \cr
@@ -436,6 +460,8 @@ Other ops: \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
 \code{\link{op_vectorized_map}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_while_loop}()} \cr
diff --git a/man/op_expand_dims.Rd b/man/op_expand_dims.Rd
index cccbdd4c0..f599e44f4 100644
--- a/man/op_expand_dims.Rd
+++ b/man/op_expand_dims.Rd
@@ -27,6 +27,7 @@ Other numpy ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -43,6 +44,7 @@ Other numpy ops: \cr
 \code{\link{op_argsort}()} \cr
 \code{\link{op_array}()} \cr
 \code{\link{op_average}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_bincount}()} \cr
 \code{\link{op_bitwise_and}()} \cr
 \code{\link{op_bitwise_invert}()} \cr
@@ -51,12 +53,15 @@ Other numpy ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_clip}()} \cr
 \code{\link{op_concatenate}()} \cr
 \code{\link{op_conj}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -65,6 +70,7 @@ Other numpy ops: \cr
 \code{\link{op_ctc_decode}()} \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_diag}()} \cr
 \code{\link{op_diagflat}()} \cr
 \code{\link{op_diagonal}()} \cr
@@ -88,6 +94,9 @@ Other numpy ops: \cr
 \code{\link{op_get_item}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -97,6 +106,7 @@ Other numpy ops: \cr
 \code{\link{op_isfinite}()} \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
+\code{\link{op_kaiser}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
 \code{\link{op_less_equal}()} \cr
@@ -177,6 +187,8 @@ Other numpy ops: \cr
 \code{\link{op_var}()} \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_zeros}()} \cr
@@ -186,6 +198,7 @@ Other ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -204,6 +217,7 @@ Other ops: \cr
 \code{\link{op_associative_scan}()} \cr
 \code{\link{op_average}()} \cr
 \code{\link{op_average_pool}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_batch_normalization}()} \cr
 \code{\link{op_binary_crossentropy}()} \cr
 \code{\link{op_bincount}()} \cr
@@ -214,9 +228,11 @@ Other ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
 \code{\link{op_cast}()} \cr
 \code{\link{op_categorical_crossentropy}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_celu}()} \cr
 \code{\link{op_cholesky}()} \cr
@@ -229,6 +245,7 @@ Other ops: \cr
 \code{\link{op_convert_to_numpy}()} \cr
 \code{\link{op_convert_to_tensor}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -239,6 +256,7 @@ Other ops: \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
 \code{\link{op_custom_gradient}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_depthwise_conv}()} \cr
 \code{\link{op_det}()} \cr
 \code{\link{op_diag}()} \cr
@@ -277,10 +295,13 @@ Other ops: \cr
 \code{\link{op_glu}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
 \code{\link{op_hard_shrink}()} \cr
 \code{\link{op_hard_sigmoid}()} \cr
 \code{\link{op_hard_silu}()} \cr
 \code{\link{op_hard_tanh}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -307,6 +328,8 @@ Other ops: \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
 \code{\link{op_istft}()} \cr
+\code{\link{op_kaiser}()} \cr
+\code{\link{op_layer_normalization}()} \cr
 \code{\link{op_leaky_relu}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
@@ -406,6 +429,7 @@ Other ops: \cr
 \code{\link{op_sort}()} \cr
 \code{\link{op_sparse_categorical_crossentropy}()} \cr
 \code{\link{op_sparse_plus}()} \cr
+\code{\link{op_sparse_sigmoid}()} \cr
 \code{\link{op_sparsemax}()} \cr
 \code{\link{op_split}()} \cr
 \code{\link{op_sqrt}()} \cr
@@ -443,6 +467,8 @@ Other ops: \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
 \code{\link{op_vectorized_map}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_while_loop}()} \cr
diff --git a/man/op_expm1.Rd b/man/op_expm1.Rd
index 031885c5e..347926f7d 100644
--- a/man/op_expm1.Rd
+++ b/man/op_expm1.Rd
@@ -24,6 +24,7 @@ Other numpy ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -40,6 +41,7 @@ Other numpy ops: \cr
 \code{\link{op_argsort}()} \cr
 \code{\link{op_array}()} \cr
 \code{\link{op_average}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_bincount}()} \cr
 \code{\link{op_bitwise_and}()} \cr
 \code{\link{op_bitwise_invert}()} \cr
@@ -48,12 +50,15 @@ Other numpy ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_clip}()} \cr
 \code{\link{op_concatenate}()} \cr
 \code{\link{op_conj}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -62,6 +67,7 @@ Other numpy ops: \cr
 \code{\link{op_ctc_decode}()} \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_diag}()} \cr
 \code{\link{op_diagflat}()} \cr
 \code{\link{op_diagonal}()} \cr
@@ -85,6 +91,9 @@ Other numpy ops: \cr
 \code{\link{op_get_item}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -94,6 +103,7 @@ Other numpy ops: \cr
 \code{\link{op_isfinite}()} \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
+\code{\link{op_kaiser}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
 \code{\link{op_less_equal}()} \cr
@@ -174,6 +184,8 @@ Other numpy ops: \cr
 \code{\link{op_var}()} \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_zeros}()} \cr
@@ -183,6 +195,7 @@ Other ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -201,6 +214,7 @@ Other ops: \cr
 \code{\link{op_associative_scan}()} \cr
 \code{\link{op_average}()} \cr
 \code{\link{op_average_pool}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_batch_normalization}()} \cr
 \code{\link{op_binary_crossentropy}()} \cr
 \code{\link{op_bincount}()} \cr
@@ -211,9 +225,11 @@ Other ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
 \code{\link{op_cast}()} \cr
 \code{\link{op_categorical_crossentropy}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_celu}()} \cr
 \code{\link{op_cholesky}()} \cr
@@ -226,6 +242,7 @@ Other ops: \cr
 \code{\link{op_convert_to_numpy}()} \cr
 \code{\link{op_convert_to_tensor}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -236,6 +253,7 @@ Other ops: \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
 \code{\link{op_custom_gradient}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_depthwise_conv}()} \cr
 \code{\link{op_det}()} \cr
 \code{\link{op_diag}()} \cr
@@ -274,10 +292,13 @@ Other ops: \cr
 \code{\link{op_glu}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
 \code{\link{op_hard_shrink}()} \cr
 \code{\link{op_hard_sigmoid}()} \cr
 \code{\link{op_hard_silu}()} \cr
 \code{\link{op_hard_tanh}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -304,6 +325,8 @@ Other ops: \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
 \code{\link{op_istft}()} \cr
+\code{\link{op_kaiser}()} \cr
+\code{\link{op_layer_normalization}()} \cr
 \code{\link{op_leaky_relu}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
@@ -403,6 +426,7 @@ Other ops: \cr
 \code{\link{op_sort}()} \cr
 \code{\link{op_sparse_categorical_crossentropy}()} \cr
 \code{\link{op_sparse_plus}()} \cr
+\code{\link{op_sparse_sigmoid}()} \cr
 \code{\link{op_sparsemax}()} \cr
 \code{\link{op_split}()} \cr
 \code{\link{op_sqrt}()} \cr
@@ -440,6 +464,8 @@ Other ops: \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
 \code{\link{op_vectorized_map}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_while_loop}()} \cr
diff --git a/man/op_extract_sequences.Rd b/man/op_extract_sequences.Rd
index f9fc5ed69..32e48032c 100644
--- a/man/op_extract_sequences.Rd
+++ b/man/op_extract_sequences.Rd
@@ -66,6 +66,7 @@ Other ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -84,6 +85,7 @@ Other ops: \cr
 \code{\link{op_associative_scan}()} \cr
 \code{\link{op_average}()} \cr
 \code{\link{op_average_pool}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_batch_normalization}()} \cr
 \code{\link{op_binary_crossentropy}()} \cr
 \code{\link{op_bincount}()} \cr
@@ -94,9 +96,11 @@ Other ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
 \code{\link{op_cast}()} \cr
 \code{\link{op_categorical_crossentropy}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_celu}()} \cr
 \code{\link{op_cholesky}()} \cr
@@ -109,6 +113,7 @@ Other ops: \cr
 \code{\link{op_convert_to_numpy}()} \cr
 \code{\link{op_convert_to_tensor}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -119,6 +124,7 @@ Other ops: \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
 \code{\link{op_custom_gradient}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_depthwise_conv}()} \cr
 \code{\link{op_det}()} \cr
 \code{\link{op_diag}()} \cr
@@ -157,10 +163,13 @@ Other ops: \cr
 \code{\link{op_glu}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
 \code{\link{op_hard_shrink}()} \cr
 \code{\link{op_hard_sigmoid}()} \cr
 \code{\link{op_hard_silu}()} \cr
 \code{\link{op_hard_tanh}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -187,6 +196,8 @@ Other ops: \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
 \code{\link{op_istft}()} \cr
+\code{\link{op_kaiser}()} \cr
+\code{\link{op_layer_normalization}()} \cr
 \code{\link{op_leaky_relu}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
@@ -286,6 +297,7 @@ Other ops: \cr
 \code{\link{op_sort}()} \cr
 \code{\link{op_sparse_categorical_crossentropy}()} \cr
 \code{\link{op_sparse_plus}()} \cr
+\code{\link{op_sparse_sigmoid}()} \cr
 \code{\link{op_sparsemax}()} \cr
 \code{\link{op_split}()} \cr
 \code{\link{op_sqrt}()} \cr
@@ -323,6 +335,8 @@ Other ops: \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
 \code{\link{op_vectorized_map}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_while_loop}()} \cr
diff --git a/man/op_eye.Rd b/man/op_eye.Rd
index 56daf45fa..a3023b5d9 100644
--- a/man/op_eye.Rd
+++ b/man/op_eye.Rd
@@ -32,6 +32,7 @@ Other numpy ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -48,6 +49,7 @@ Other numpy ops: \cr
 \code{\link{op_argsort}()} \cr
 \code{\link{op_array}()} \cr
 \code{\link{op_average}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_bincount}()} \cr
 \code{\link{op_bitwise_and}()} \cr
 \code{\link{op_bitwise_invert}()} \cr
@@ -56,12 +58,15 @@ Other numpy ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_clip}()} \cr
 \code{\link{op_concatenate}()} \cr
 \code{\link{op_conj}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -70,6 +75,7 @@ Other numpy ops: \cr
 \code{\link{op_ctc_decode}()} \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_diag}()} \cr
 \code{\link{op_diagflat}()} \cr
 \code{\link{op_diagonal}()} \cr
@@ -93,6 +99,9 @@ Other numpy ops: \cr
 \code{\link{op_get_item}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -102,6 +111,7 @@ Other numpy ops: \cr
 \code{\link{op_isfinite}()} \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
+\code{\link{op_kaiser}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
 \code{\link{op_less_equal}()} \cr
@@ -182,6 +192,8 @@ Other numpy ops: \cr
 \code{\link{op_var}()} \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_zeros}()} \cr
@@ -191,6 +203,7 @@ Other ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -209,6 +222,7 @@ Other ops: \cr
 \code{\link{op_associative_scan}()} \cr
 \code{\link{op_average}()} \cr
 \code{\link{op_average_pool}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_batch_normalization}()} \cr
 \code{\link{op_binary_crossentropy}()} \cr
 \code{\link{op_bincount}()} \cr
@@ -219,9 +233,11 @@ Other ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
 \code{\link{op_cast}()} \cr
 \code{\link{op_categorical_crossentropy}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_celu}()} \cr
 \code{\link{op_cholesky}()} \cr
@@ -234,6 +250,7 @@ Other ops: \cr
 \code{\link{op_convert_to_numpy}()} \cr
 \code{\link{op_convert_to_tensor}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -244,6 +261,7 @@ Other ops: \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
 \code{\link{op_custom_gradient}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_depthwise_conv}()} \cr
 \code{\link{op_det}()} \cr
 \code{\link{op_diag}()} \cr
@@ -282,10 +300,13 @@ Other ops: \cr
 \code{\link{op_glu}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
 \code{\link{op_hard_shrink}()} \cr
 \code{\link{op_hard_sigmoid}()} \cr
 \code{\link{op_hard_silu}()} \cr
 \code{\link{op_hard_tanh}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -312,6 +333,8 @@ Other ops: \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
 \code{\link{op_istft}()} \cr
+\code{\link{op_kaiser}()} \cr
+\code{\link{op_layer_normalization}()} \cr
 \code{\link{op_leaky_relu}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
@@ -411,6 +434,7 @@ Other ops: \cr
 \code{\link{op_sort}()} \cr
 \code{\link{op_sparse_categorical_crossentropy}()} \cr
 \code{\link{op_sparse_plus}()} \cr
+\code{\link{op_sparse_sigmoid}()} \cr
 \code{\link{op_sparsemax}()} \cr
 \code{\link{op_split}()} \cr
 \code{\link{op_sqrt}()} \cr
@@ -448,6 +472,8 @@ Other ops: \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
 \code{\link{op_vectorized_map}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_while_loop}()} \cr
diff --git a/man/op_fft.Rd b/man/op_fft.Rd
index 423527e31..12190e164 100644
--- a/man/op_fft.Rd
+++ b/man/op_fft.Rd
@@ -60,6 +60,7 @@ Other ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -78,6 +79,7 @@ Other ops: \cr
 \code{\link{op_associative_scan}()} \cr
 \code{\link{op_average}()} \cr
 \code{\link{op_average_pool}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_batch_normalization}()} \cr
 \code{\link{op_binary_crossentropy}()} \cr
 \code{\link{op_bincount}()} \cr
@@ -88,9 +90,11 @@ Other ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
 \code{\link{op_cast}()} \cr
 \code{\link{op_categorical_crossentropy}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_celu}()} \cr
 \code{\link{op_cholesky}()} \cr
@@ -103,6 +107,7 @@ Other ops: \cr
 \code{\link{op_convert_to_numpy}()} \cr
 \code{\link{op_convert_to_tensor}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -113,6 +118,7 @@ Other ops: \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
 \code{\link{op_custom_gradient}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_depthwise_conv}()} \cr
 \code{\link{op_det}()} \cr
 \code{\link{op_diag}()} \cr
@@ -151,10 +157,13 @@ Other ops: \cr
 \code{\link{op_glu}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
 \code{\link{op_hard_shrink}()} \cr
 \code{\link{op_hard_sigmoid}()} \cr
 \code{\link{op_hard_silu}()} \cr
 \code{\link{op_hard_tanh}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -181,6 +190,8 @@ Other ops: \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
 \code{\link{op_istft}()} \cr
+\code{\link{op_kaiser}()} \cr
+\code{\link{op_layer_normalization}()} \cr
 \code{\link{op_leaky_relu}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
@@ -280,6 +291,7 @@ Other ops: \cr
 \code{\link{op_sort}()} \cr
 \code{\link{op_sparse_categorical_crossentropy}()} \cr
 \code{\link{op_sparse_plus}()} \cr
+\code{\link{op_sparse_sigmoid}()} \cr
 \code{\link{op_sparsemax}()} \cr
 \code{\link{op_split}()} \cr
 \code{\link{op_sqrt}()} \cr
@@ -317,6 +329,8 @@ Other ops: \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
 \code{\link{op_vectorized_map}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_while_loop}()} \cr
diff --git a/man/op_fft2.Rd b/man/op_fft2.Rd
index dcf1f777f..eba81eaac 100644
--- a/man/op_fft2.Rd
+++ b/man/op_fft2.Rd
@@ -66,6 +66,7 @@ Other ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -84,6 +85,7 @@ Other ops: \cr
 \code{\link{op_associative_scan}()} \cr
 \code{\link{op_average}()} \cr
 \code{\link{op_average_pool}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_batch_normalization}()} \cr
 \code{\link{op_binary_crossentropy}()} \cr
 \code{\link{op_bincount}()} \cr
@@ -94,9 +96,11 @@ Other ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
 \code{\link{op_cast}()} \cr
 \code{\link{op_categorical_crossentropy}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_celu}()} \cr
 \code{\link{op_cholesky}()} \cr
@@ -109,6 +113,7 @@ Other ops: \cr
 \code{\link{op_convert_to_numpy}()} \cr
 \code{\link{op_convert_to_tensor}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -119,6 +124,7 @@ Other ops: \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
 \code{\link{op_custom_gradient}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_depthwise_conv}()} \cr
 \code{\link{op_det}()} \cr
 \code{\link{op_diag}()} \cr
@@ -157,10 +163,13 @@ Other ops: \cr
 \code{\link{op_glu}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
 \code{\link{op_hard_shrink}()} \cr
 \code{\link{op_hard_sigmoid}()} \cr
 \code{\link{op_hard_silu}()} \cr
 \code{\link{op_hard_tanh}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -187,6 +196,8 @@ Other ops: \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
 \code{\link{op_istft}()} \cr
+\code{\link{op_kaiser}()} \cr
+\code{\link{op_layer_normalization}()} \cr
 \code{\link{op_leaky_relu}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
@@ -286,6 +297,7 @@ Other ops: \cr
 \code{\link{op_sort}()} \cr
 \code{\link{op_sparse_categorical_crossentropy}()} \cr
 \code{\link{op_sparse_plus}()} \cr
+\code{\link{op_sparse_sigmoid}()} \cr
 \code{\link{op_sparsemax}()} \cr
 \code{\link{op_split}()} \cr
 \code{\link{op_sqrt}()} \cr
@@ -323,6 +335,8 @@ Other ops: \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
 \code{\link{op_vectorized_map}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_while_loop}()} \cr
diff --git a/man/op_flip.Rd b/man/op_flip.Rd
index fc6665a08..04fb6b82d 100644
--- a/man/op_flip.Rd
+++ b/man/op_flip.Rd
@@ -27,6 +27,7 @@ Other numpy ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -43,6 +44,7 @@ Other numpy ops: \cr
 \code{\link{op_argsort}()} \cr
 \code{\link{op_array}()} \cr
 \code{\link{op_average}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_bincount}()} \cr
 \code{\link{op_bitwise_and}()} \cr
 \code{\link{op_bitwise_invert}()} \cr
@@ -51,12 +53,15 @@ Other numpy ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_clip}()} \cr
 \code{\link{op_concatenate}()} \cr
 \code{\link{op_conj}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -65,6 +70,7 @@ Other numpy ops: \cr
 \code{\link{op_ctc_decode}()} \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_diag}()} \cr
 \code{\link{op_diagflat}()} \cr
 \code{\link{op_diagonal}()} \cr
@@ -88,6 +94,9 @@ Other numpy ops: \cr
 \code{\link{op_get_item}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -97,6 +106,7 @@ Other numpy ops: \cr
 \code{\link{op_isfinite}()} \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
+\code{\link{op_kaiser}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
 \code{\link{op_less_equal}()} \cr
@@ -177,6 +187,8 @@ Other numpy ops: \cr
 \code{\link{op_var}()} \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_zeros}()} \cr
@@ -186,6 +198,7 @@ Other ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -204,6 +217,7 @@ Other ops: \cr
 \code{\link{op_associative_scan}()} \cr
 \code{\link{op_average}()} \cr
 \code{\link{op_average_pool}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_batch_normalization}()} \cr
 \code{\link{op_binary_crossentropy}()} \cr
 \code{\link{op_bincount}()} \cr
@@ -214,9 +228,11 @@ Other ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
 \code{\link{op_cast}()} \cr
 \code{\link{op_categorical_crossentropy}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_celu}()} \cr
 \code{\link{op_cholesky}()} \cr
@@ -229,6 +245,7 @@ Other ops: \cr
 \code{\link{op_convert_to_numpy}()} \cr
 \code{\link{op_convert_to_tensor}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -239,6 +256,7 @@ Other ops: \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
 \code{\link{op_custom_gradient}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_depthwise_conv}()} \cr
 \code{\link{op_det}()} \cr
 \code{\link{op_diag}()} \cr
@@ -277,10 +295,13 @@ Other ops: \cr
 \code{\link{op_glu}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
 \code{\link{op_hard_shrink}()} \cr
 \code{\link{op_hard_sigmoid}()} \cr
 \code{\link{op_hard_silu}()} \cr
 \code{\link{op_hard_tanh}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -307,6 +328,8 @@ Other ops: \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
 \code{\link{op_istft}()} \cr
+\code{\link{op_kaiser}()} \cr
+\code{\link{op_layer_normalization}()} \cr
 \code{\link{op_leaky_relu}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
@@ -406,6 +429,7 @@ Other ops: \cr
 \code{\link{op_sort}()} \cr
 \code{\link{op_sparse_categorical_crossentropy}()} \cr
 \code{\link{op_sparse_plus}()} \cr
+\code{\link{op_sparse_sigmoid}()} \cr
 \code{\link{op_sparsemax}()} \cr
 \code{\link{op_split}()} \cr
 \code{\link{op_sqrt}()} \cr
@@ -443,6 +467,8 @@ Other ops: \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
 \code{\link{op_vectorized_map}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_while_loop}()} \cr
diff --git a/man/op_floor.Rd b/man/op_floor.Rd
index 1f2ebe24f..450e90846 100644
--- a/man/op_floor.Rd
+++ b/man/op_floor.Rd
@@ -24,6 +24,7 @@ Other numpy ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -40,6 +41,7 @@ Other numpy ops: \cr
 \code{\link{op_argsort}()} \cr
 \code{\link{op_array}()} \cr
 \code{\link{op_average}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_bincount}()} \cr
 \code{\link{op_bitwise_and}()} \cr
 \code{\link{op_bitwise_invert}()} \cr
@@ -48,12 +50,15 @@ Other numpy ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_clip}()} \cr
 \code{\link{op_concatenate}()} \cr
 \code{\link{op_conj}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -62,6 +67,7 @@ Other numpy ops: \cr
 \code{\link{op_ctc_decode}()} \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_diag}()} \cr
 \code{\link{op_diagflat}()} \cr
 \code{\link{op_diagonal}()} \cr
@@ -85,6 +91,9 @@ Other numpy ops: \cr
 \code{\link{op_get_item}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -94,6 +103,7 @@ Other numpy ops: \cr
 \code{\link{op_isfinite}()} \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
+\code{\link{op_kaiser}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
 \code{\link{op_less_equal}()} \cr
@@ -174,6 +184,8 @@ Other numpy ops: \cr
 \code{\link{op_var}()} \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_zeros}()} \cr
@@ -183,6 +195,7 @@ Other ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -201,6 +214,7 @@ Other ops: \cr
 \code{\link{op_associative_scan}()} \cr
 \code{\link{op_average}()} \cr
 \code{\link{op_average_pool}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_batch_normalization}()} \cr
 \code{\link{op_binary_crossentropy}()} \cr
 \code{\link{op_bincount}()} \cr
@@ -211,9 +225,11 @@ Other ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
 \code{\link{op_cast}()} \cr
 \code{\link{op_categorical_crossentropy}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_celu}()} \cr
 \code{\link{op_cholesky}()} \cr
@@ -226,6 +242,7 @@ Other ops: \cr
 \code{\link{op_convert_to_numpy}()} \cr
 \code{\link{op_convert_to_tensor}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -236,6 +253,7 @@ Other ops: \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
 \code{\link{op_custom_gradient}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_depthwise_conv}()} \cr
 \code{\link{op_det}()} \cr
 \code{\link{op_diag}()} \cr
@@ -274,10 +292,13 @@ Other ops: \cr
 \code{\link{op_glu}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
 \code{\link{op_hard_shrink}()} \cr
 \code{\link{op_hard_sigmoid}()} \cr
 \code{\link{op_hard_silu}()} \cr
 \code{\link{op_hard_tanh}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -304,6 +325,8 @@ Other ops: \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
 \code{\link{op_istft}()} \cr
+\code{\link{op_kaiser}()} \cr
+\code{\link{op_layer_normalization}()} \cr
 \code{\link{op_leaky_relu}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
@@ -403,6 +426,7 @@ Other ops: \cr
 \code{\link{op_sort}()} \cr
 \code{\link{op_sparse_categorical_crossentropy}()} \cr
 \code{\link{op_sparse_plus}()} \cr
+\code{\link{op_sparse_sigmoid}()} \cr
 \code{\link{op_sparsemax}()} \cr
 \code{\link{op_split}()} \cr
 \code{\link{op_sqrt}()} \cr
@@ -440,6 +464,8 @@ Other ops: \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
 \code{\link{op_vectorized_map}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_while_loop}()} \cr
diff --git a/man/op_floor_divide.Rd b/man/op_floor_divide.Rd
index 090ff3922..201e67577 100644
--- a/man/op_floor_divide.Rd
+++ b/man/op_floor_divide.Rd
@@ -43,6 +43,7 @@ Other numpy ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -59,6 +60,7 @@ Other numpy ops: \cr
 \code{\link{op_argsort}()} \cr
 \code{\link{op_array}()} \cr
 \code{\link{op_average}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_bincount}()} \cr
 \code{\link{op_bitwise_and}()} \cr
 \code{\link{op_bitwise_invert}()} \cr
@@ -67,12 +69,15 @@ Other numpy ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_clip}()} \cr
 \code{\link{op_concatenate}()} \cr
 \code{\link{op_conj}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -81,6 +86,7 @@ Other numpy ops: \cr
 \code{\link{op_ctc_decode}()} \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_diag}()} \cr
 \code{\link{op_diagflat}()} \cr
 \code{\link{op_diagonal}()} \cr
@@ -104,6 +110,9 @@ Other numpy ops: \cr
 \code{\link{op_get_item}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -113,6 +122,7 @@ Other numpy ops: \cr
 \code{\link{op_isfinite}()} \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
+\code{\link{op_kaiser}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
 \code{\link{op_less_equal}()} \cr
@@ -193,6 +203,8 @@ Other numpy ops: \cr
 \code{\link{op_var}()} \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_zeros}()} \cr
@@ -202,6 +214,7 @@ Other ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -220,6 +233,7 @@ Other ops: \cr
 \code{\link{op_associative_scan}()} \cr
 \code{\link{op_average}()} \cr
 \code{\link{op_average_pool}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_batch_normalization}()} \cr
 \code{\link{op_binary_crossentropy}()} \cr
 \code{\link{op_bincount}()} \cr
@@ -230,9 +244,11 @@ Other ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
 \code{\link{op_cast}()} \cr
 \code{\link{op_categorical_crossentropy}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_celu}()} \cr
 \code{\link{op_cholesky}()} \cr
@@ -245,6 +261,7 @@ Other ops: \cr
 \code{\link{op_convert_to_numpy}()} \cr
 \code{\link{op_convert_to_tensor}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -255,6 +272,7 @@ Other ops: \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
 \code{\link{op_custom_gradient}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_depthwise_conv}()} \cr
 \code{\link{op_det}()} \cr
 \code{\link{op_diag}()} \cr
@@ -293,10 +311,13 @@ Other ops: \cr
 \code{\link{op_glu}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
 \code{\link{op_hard_shrink}()} \cr
 \code{\link{op_hard_sigmoid}()} \cr
 \code{\link{op_hard_silu}()} \cr
 \code{\link{op_hard_tanh}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -323,6 +344,8 @@ Other ops: \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
 \code{\link{op_istft}()} \cr
+\code{\link{op_kaiser}()} \cr
+\code{\link{op_layer_normalization}()} \cr
 \code{\link{op_leaky_relu}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
@@ -422,6 +445,7 @@ Other ops: \cr
 \code{\link{op_sort}()} \cr
 \code{\link{op_sparse_categorical_crossentropy}()} \cr
 \code{\link{op_sparse_plus}()} \cr
+\code{\link{op_sparse_sigmoid}()} \cr
 \code{\link{op_sparsemax}()} \cr
 \code{\link{op_split}()} \cr
 \code{\link{op_sqrt}()} \cr
@@ -459,6 +483,8 @@ Other ops: \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
 \code{\link{op_vectorized_map}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_while_loop}()} \cr
diff --git a/man/op_fori_loop.Rd b/man/op_fori_loop.Rd
index 4e90095be..424755739 100644
--- a/man/op_fori_loop.Rd
+++ b/man/op_fori_loop.Rd
@@ -67,6 +67,7 @@ Other ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -85,6 +86,7 @@ Other ops: \cr
 \code{\link{op_associative_scan}()} \cr
 \code{\link{op_average}()} \cr
 \code{\link{op_average_pool}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_batch_normalization}()} \cr
 \code{\link{op_binary_crossentropy}()} \cr
 \code{\link{op_bincount}()} \cr
@@ -95,9 +97,11 @@ Other ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
 \code{\link{op_cast}()} \cr
 \code{\link{op_categorical_crossentropy}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_celu}()} \cr
 \code{\link{op_cholesky}()} \cr
@@ -110,6 +114,7 @@ Other ops: \cr
 \code{\link{op_convert_to_numpy}()} \cr
 \code{\link{op_convert_to_tensor}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -120,6 +125,7 @@ Other ops: \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
 \code{\link{op_custom_gradient}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_depthwise_conv}()} \cr
 \code{\link{op_det}()} \cr
 \code{\link{op_diag}()} \cr
@@ -158,10 +164,13 @@ Other ops: \cr
 \code{\link{op_glu}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
 \code{\link{op_hard_shrink}()} \cr
 \code{\link{op_hard_sigmoid}()} \cr
 \code{\link{op_hard_silu}()} \cr
 \code{\link{op_hard_tanh}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -188,6 +197,8 @@ Other ops: \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
 \code{\link{op_istft}()} \cr
+\code{\link{op_kaiser}()} \cr
+\code{\link{op_layer_normalization}()} \cr
 \code{\link{op_leaky_relu}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
@@ -287,6 +298,7 @@ Other ops: \cr
 \code{\link{op_sort}()} \cr
 \code{\link{op_sparse_categorical_crossentropy}()} \cr
 \code{\link{op_sparse_plus}()} \cr
+\code{\link{op_sparse_sigmoid}()} \cr
 \code{\link{op_sparsemax}()} \cr
 \code{\link{op_split}()} \cr
 \code{\link{op_sqrt}()} \cr
@@ -324,6 +336,8 @@ Other ops: \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
 \code{\link{op_vectorized_map}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_while_loop}()} \cr
diff --git a/man/op_full.Rd b/man/op_full.Rd
index 1bfdb92cb..fa49f43b4 100644
--- a/man/op_full.Rd
+++ b/man/op_full.Rd
@@ -28,6 +28,7 @@ Other numpy ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -44,6 +45,7 @@ Other numpy ops: \cr
 \code{\link{op_argsort}()} \cr
 \code{\link{op_array}()} \cr
 \code{\link{op_average}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_bincount}()} \cr
 \code{\link{op_bitwise_and}()} \cr
 \code{\link{op_bitwise_invert}()} \cr
@@ -52,12 +54,15 @@ Other numpy ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_clip}()} \cr
 \code{\link{op_concatenate}()} \cr
 \code{\link{op_conj}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -66,6 +71,7 @@ Other numpy ops: \cr
 \code{\link{op_ctc_decode}()} \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_diag}()} \cr
 \code{\link{op_diagflat}()} \cr
 \code{\link{op_diagonal}()} \cr
@@ -89,6 +95,9 @@ Other numpy ops: \cr
 \code{\link{op_get_item}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -98,6 +107,7 @@ Other numpy ops: \cr
 \code{\link{op_isfinite}()} \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
+\code{\link{op_kaiser}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
 \code{\link{op_less_equal}()} \cr
@@ -178,6 +188,8 @@ Other numpy ops: \cr
 \code{\link{op_var}()} \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_zeros}()} \cr
@@ -187,6 +199,7 @@ Other ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -205,6 +218,7 @@ Other ops: \cr
 \code{\link{op_associative_scan}()} \cr
 \code{\link{op_average}()} \cr
 \code{\link{op_average_pool}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_batch_normalization}()} \cr
 \code{\link{op_binary_crossentropy}()} \cr
 \code{\link{op_bincount}()} \cr
@@ -215,9 +229,11 @@ Other ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
 \code{\link{op_cast}()} \cr
 \code{\link{op_categorical_crossentropy}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_celu}()} \cr
 \code{\link{op_cholesky}()} \cr
@@ -230,6 +246,7 @@ Other ops: \cr
 \code{\link{op_convert_to_numpy}()} \cr
 \code{\link{op_convert_to_tensor}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -240,6 +257,7 @@ Other ops: \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
 \code{\link{op_custom_gradient}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_depthwise_conv}()} \cr
 \code{\link{op_det}()} \cr
 \code{\link{op_diag}()} \cr
@@ -278,10 +296,13 @@ Other ops: \cr
 \code{\link{op_glu}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
 \code{\link{op_hard_shrink}()} \cr
 \code{\link{op_hard_sigmoid}()} \cr
 \code{\link{op_hard_silu}()} \cr
 \code{\link{op_hard_tanh}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -308,6 +329,8 @@ Other ops: \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
 \code{\link{op_istft}()} \cr
+\code{\link{op_kaiser}()} \cr
+\code{\link{op_layer_normalization}()} \cr
 \code{\link{op_leaky_relu}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
@@ -407,6 +430,7 @@ Other ops: \cr
 \code{\link{op_sort}()} \cr
 \code{\link{op_sparse_categorical_crossentropy}()} \cr
 \code{\link{op_sparse_plus}()} \cr
+\code{\link{op_sparse_sigmoid}()} \cr
 \code{\link{op_sparsemax}()} \cr
 \code{\link{op_split}()} \cr
 \code{\link{op_sqrt}()} \cr
@@ -444,6 +468,8 @@ Other ops: \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
 \code{\link{op_vectorized_map}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_while_loop}()} \cr
diff --git a/man/op_full_like.Rd b/man/op_full_like.Rd
index 77e4994f2..559e33791 100644
--- a/man/op_full_like.Rd
+++ b/man/op_full_like.Rd
@@ -28,6 +28,7 @@ Other numpy ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -44,6 +45,7 @@ Other numpy ops: \cr
 \code{\link{op_argsort}()} \cr
 \code{\link{op_array}()} \cr
 \code{\link{op_average}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_bincount}()} \cr
 \code{\link{op_bitwise_and}()} \cr
 \code{\link{op_bitwise_invert}()} \cr
@@ -52,12 +54,15 @@ Other numpy ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_clip}()} \cr
 \code{\link{op_concatenate}()} \cr
 \code{\link{op_conj}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -66,6 +71,7 @@ Other numpy ops: \cr
 \code{\link{op_ctc_decode}()} \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_diag}()} \cr
 \code{\link{op_diagflat}()} \cr
 \code{\link{op_diagonal}()} \cr
@@ -89,6 +95,9 @@ Other numpy ops: \cr
 \code{\link{op_get_item}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -98,6 +107,7 @@ Other numpy ops: \cr
 \code{\link{op_isfinite}()} \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
+\code{\link{op_kaiser}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
 \code{\link{op_less_equal}()} \cr
@@ -178,6 +188,8 @@ Other numpy ops: \cr
 \code{\link{op_var}()} \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_zeros}()} \cr
@@ -187,6 +199,7 @@ Other ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -205,6 +218,7 @@ Other ops: \cr
 \code{\link{op_associative_scan}()} \cr
 \code{\link{op_average}()} \cr
 \code{\link{op_average_pool}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_batch_normalization}()} \cr
 \code{\link{op_binary_crossentropy}()} \cr
 \code{\link{op_bincount}()} \cr
@@ -215,9 +229,11 @@ Other ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
 \code{\link{op_cast}()} \cr
 \code{\link{op_categorical_crossentropy}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_celu}()} \cr
 \code{\link{op_cholesky}()} \cr
@@ -230,6 +246,7 @@ Other ops: \cr
 \code{\link{op_convert_to_numpy}()} \cr
 \code{\link{op_convert_to_tensor}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -240,6 +257,7 @@ Other ops: \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
 \code{\link{op_custom_gradient}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_depthwise_conv}()} \cr
 \code{\link{op_det}()} \cr
 \code{\link{op_diag}()} \cr
@@ -278,10 +296,13 @@ Other ops: \cr
 \code{\link{op_glu}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
 \code{\link{op_hard_shrink}()} \cr
 \code{\link{op_hard_sigmoid}()} \cr
 \code{\link{op_hard_silu}()} \cr
 \code{\link{op_hard_tanh}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -308,6 +329,8 @@ Other ops: \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
 \code{\link{op_istft}()} \cr
+\code{\link{op_kaiser}()} \cr
+\code{\link{op_layer_normalization}()} \cr
 \code{\link{op_leaky_relu}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
@@ -407,6 +430,7 @@ Other ops: \cr
 \code{\link{op_sort}()} \cr
 \code{\link{op_sparse_categorical_crossentropy}()} \cr
 \code{\link{op_sparse_plus}()} \cr
+\code{\link{op_sparse_sigmoid}()} \cr
 \code{\link{op_sparsemax}()} \cr
 \code{\link{op_split}()} \cr
 \code{\link{op_sqrt}()} \cr
@@ -444,6 +468,8 @@ Other ops: \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
 \code{\link{op_vectorized_map}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_while_loop}()} \cr
diff --git a/man/op_gelu.Rd b/man/op_gelu.Rd
index 9f6232110..96b24d856 100644
--- a/man/op_gelu.Rd
+++ b/man/op_gelu.Rd
@@ -69,6 +69,7 @@ Other nn ops: \cr
 \code{\link{op_hard_sigmoid}()} \cr
 \code{\link{op_hard_silu}()} \cr
 \code{\link{op_hard_tanh}()} \cr
+\code{\link{op_layer_normalization}()} \cr
 \code{\link{op_leaky_relu}()} \cr
 \code{\link{op_log_sigmoid}()} \cr
 \code{\link{op_log_softmax}()} \cr
@@ -92,6 +93,7 @@ Other nn ops: \cr
 \code{\link{op_softsign}()} \cr
 \code{\link{op_sparse_categorical_crossentropy}()} \cr
 \code{\link{op_sparse_plus}()} \cr
+\code{\link{op_sparse_sigmoid}()} \cr
 \code{\link{op_sparsemax}()} \cr
 \code{\link{op_squareplus}()} \cr
 \code{\link{op_tanh_shrink}()} \cr
@@ -102,6 +104,7 @@ Other ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -120,6 +123,7 @@ Other ops: \cr
 \code{\link{op_associative_scan}()} \cr
 \code{\link{op_average}()} \cr
 \code{\link{op_average_pool}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_batch_normalization}()} \cr
 \code{\link{op_binary_crossentropy}()} \cr
 \code{\link{op_bincount}()} \cr
@@ -130,9 +134,11 @@ Other ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
 \code{\link{op_cast}()} \cr
 \code{\link{op_categorical_crossentropy}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_celu}()} \cr
 \code{\link{op_cholesky}()} \cr
@@ -145,6 +151,7 @@ Other ops: \cr
 \code{\link{op_convert_to_numpy}()} \cr
 \code{\link{op_convert_to_tensor}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -155,6 +162,7 @@ Other ops: \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
 \code{\link{op_custom_gradient}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_depthwise_conv}()} \cr
 \code{\link{op_det}()} \cr
 \code{\link{op_diag}()} \cr
@@ -193,10 +201,13 @@ Other ops: \cr
 \code{\link{op_glu}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
 \code{\link{op_hard_shrink}()} \cr
 \code{\link{op_hard_sigmoid}()} \cr
 \code{\link{op_hard_silu}()} \cr
 \code{\link{op_hard_tanh}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -223,6 +234,8 @@ Other ops: \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
 \code{\link{op_istft}()} \cr
+\code{\link{op_kaiser}()} \cr
+\code{\link{op_layer_normalization}()} \cr
 \code{\link{op_leaky_relu}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
@@ -322,6 +335,7 @@ Other ops: \cr
 \code{\link{op_sort}()} \cr
 \code{\link{op_sparse_categorical_crossentropy}()} \cr
 \code{\link{op_sparse_plus}()} \cr
+\code{\link{op_sparse_sigmoid}()} \cr
 \code{\link{op_sparsemax}()} \cr
 \code{\link{op_split}()} \cr
 \code{\link{op_sqrt}()} \cr
@@ -359,6 +373,8 @@ Other ops: \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
 \code{\link{op_vectorized_map}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_while_loop}()} \cr
diff --git a/man/op_get_item.Rd b/man/op_get_item.Rd
index 7fb450caf..9fb2540b8 100644
--- a/man/op_get_item.Rd
+++ b/man/op_get_item.Rd
@@ -29,6 +29,7 @@ Other numpy ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -45,6 +46,7 @@ Other numpy ops: \cr
 \code{\link{op_argsort}()} \cr
 \code{\link{op_array}()} \cr
 \code{\link{op_average}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_bincount}()} \cr
 \code{\link{op_bitwise_and}()} \cr
 \code{\link{op_bitwise_invert}()} \cr
@@ -53,12 +55,15 @@ Other numpy ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_clip}()} \cr
 \code{\link{op_concatenate}()} \cr
 \code{\link{op_conj}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -67,6 +72,7 @@ Other numpy ops: \cr
 \code{\link{op_ctc_decode}()} \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_diag}()} \cr
 \code{\link{op_diagflat}()} \cr
 \code{\link{op_diagonal}()} \cr
@@ -90,6 +96,9 @@ Other numpy ops: \cr
 \code{\link{op_full_like}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -99,6 +108,7 @@ Other numpy ops: \cr
 \code{\link{op_isfinite}()} \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
+\code{\link{op_kaiser}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
 \code{\link{op_less_equal}()} \cr
@@ -179,6 +189,8 @@ Other numpy ops: \cr
 \code{\link{op_var}()} \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_zeros}()} \cr
@@ -188,6 +200,7 @@ Other ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -206,6 +219,7 @@ Other ops: \cr
 \code{\link{op_associative_scan}()} \cr
 \code{\link{op_average}()} \cr
 \code{\link{op_average_pool}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_batch_normalization}()} \cr
 \code{\link{op_binary_crossentropy}()} \cr
 \code{\link{op_bincount}()} \cr
@@ -216,9 +230,11 @@ Other ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
 \code{\link{op_cast}()} \cr
 \code{\link{op_categorical_crossentropy}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_celu}()} \cr
 \code{\link{op_cholesky}()} \cr
@@ -231,6 +247,7 @@ Other ops: \cr
 \code{\link{op_convert_to_numpy}()} \cr
 \code{\link{op_convert_to_tensor}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -241,6 +258,7 @@ Other ops: \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
 \code{\link{op_custom_gradient}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_depthwise_conv}()} \cr
 \code{\link{op_det}()} \cr
 \code{\link{op_diag}()} \cr
@@ -279,10 +297,13 @@ Other ops: \cr
 \code{\link{op_glu}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
 \code{\link{op_hard_shrink}()} \cr
 \code{\link{op_hard_sigmoid}()} \cr
 \code{\link{op_hard_silu}()} \cr
 \code{\link{op_hard_tanh}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -309,6 +330,8 @@ Other ops: \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
 \code{\link{op_istft}()} \cr
+\code{\link{op_kaiser}()} \cr
+\code{\link{op_layer_normalization}()} \cr
 \code{\link{op_leaky_relu}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
@@ -408,6 +431,7 @@ Other ops: \cr
 \code{\link{op_sort}()} \cr
 \code{\link{op_sparse_categorical_crossentropy}()} \cr
 \code{\link{op_sparse_plus}()} \cr
+\code{\link{op_sparse_sigmoid}()} \cr
 \code{\link{op_sparsemax}()} \cr
 \code{\link{op_split}()} \cr
 \code{\link{op_sqrt}()} \cr
@@ -445,6 +469,8 @@ Other ops: \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
 \code{\link{op_vectorized_map}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_while_loop}()} \cr
diff --git a/man/op_glu.Rd b/man/op_glu.Rd
index f04a27a53..0719f79f1 100644
--- a/man/op_glu.Rd
+++ b/man/op_glu.Rd
@@ -48,6 +48,7 @@ Other nn ops: \cr
 \code{\link{op_hard_sigmoid}()} \cr
 \code{\link{op_hard_silu}()} \cr
 \code{\link{op_hard_tanh}()} \cr
+\code{\link{op_layer_normalization}()} \cr
 \code{\link{op_leaky_relu}()} \cr
 \code{\link{op_log_sigmoid}()} \cr
 \code{\link{op_log_softmax}()} \cr
@@ -71,6 +72,7 @@ Other nn ops: \cr
 \code{\link{op_softsign}()} \cr
 \code{\link{op_sparse_categorical_crossentropy}()} \cr
 \code{\link{op_sparse_plus}()} \cr
+\code{\link{op_sparse_sigmoid}()} \cr
 \code{\link{op_sparsemax}()} \cr
 \code{\link{op_squareplus}()} \cr
 \code{\link{op_tanh_shrink}()} \cr
@@ -81,6 +83,7 @@ Other ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -99,6 +102,7 @@ Other ops: \cr
 \code{\link{op_associative_scan}()} \cr
 \code{\link{op_average}()} \cr
 \code{\link{op_average_pool}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_batch_normalization}()} \cr
 \code{\link{op_binary_crossentropy}()} \cr
 \code{\link{op_bincount}()} \cr
@@ -109,9 +113,11 @@ Other ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
 \code{\link{op_cast}()} \cr
 \code{\link{op_categorical_crossentropy}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_celu}()} \cr
 \code{\link{op_cholesky}()} \cr
@@ -124,6 +130,7 @@ Other ops: \cr
 \code{\link{op_convert_to_numpy}()} \cr
 \code{\link{op_convert_to_tensor}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -134,6 +141,7 @@ Other ops: \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
 \code{\link{op_custom_gradient}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_depthwise_conv}()} \cr
 \code{\link{op_det}()} \cr
 \code{\link{op_diag}()} \cr
@@ -172,10 +180,13 @@ Other ops: \cr
 \code{\link{op_get_item}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
 \code{\link{op_hard_shrink}()} \cr
 \code{\link{op_hard_sigmoid}()} \cr
 \code{\link{op_hard_silu}()} \cr
 \code{\link{op_hard_tanh}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -202,6 +213,8 @@ Other ops: \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
 \code{\link{op_istft}()} \cr
+\code{\link{op_kaiser}()} \cr
+\code{\link{op_layer_normalization}()} \cr
 \code{\link{op_leaky_relu}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
@@ -301,6 +314,7 @@ Other ops: \cr
 \code{\link{op_sort}()} \cr
 \code{\link{op_sparse_categorical_crossentropy}()} \cr
 \code{\link{op_sparse_plus}()} \cr
+\code{\link{op_sparse_sigmoid}()} \cr
 \code{\link{op_sparsemax}()} \cr
 \code{\link{op_split}()} \cr
 \code{\link{op_sqrt}()} \cr
@@ -338,6 +352,8 @@ Other ops: \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
 \code{\link{op_vectorized_map}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_while_loop}()} \cr
diff --git a/man/op_greater.Rd b/man/op_greater.Rd
index f82caa00a..497032132 100644
--- a/man/op_greater.Rd
+++ b/man/op_greater.Rd
@@ -47,6 +47,7 @@ Other numpy ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -63,6 +64,7 @@ Other numpy ops: \cr
 \code{\link{op_argsort}()} \cr
 \code{\link{op_array}()} \cr
 \code{\link{op_average}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_bincount}()} \cr
 \code{\link{op_bitwise_and}()} \cr
 \code{\link{op_bitwise_invert}()} \cr
@@ -71,12 +73,15 @@ Other numpy ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_clip}()} \cr
 \code{\link{op_concatenate}()} \cr
 \code{\link{op_conj}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -85,6 +90,7 @@ Other numpy ops: \cr
 \code{\link{op_ctc_decode}()} \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_diag}()} \cr
 \code{\link{op_diagflat}()} \cr
 \code{\link{op_diagonal}()} \cr
@@ -108,6 +114,9 @@ Other numpy ops: \cr
 \code{\link{op_full_like}()} \cr
 \code{\link{op_get_item}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -117,6 +126,7 @@ Other numpy ops: \cr
 \code{\link{op_isfinite}()} \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
+\code{\link{op_kaiser}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
 \code{\link{op_less_equal}()} \cr
@@ -197,6 +207,8 @@ Other numpy ops: \cr
 \code{\link{op_var}()} \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_zeros}()} \cr
@@ -206,6 +218,7 @@ Other ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -224,6 +237,7 @@ Other ops: \cr
 \code{\link{op_associative_scan}()} \cr
 \code{\link{op_average}()} \cr
 \code{\link{op_average_pool}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_batch_normalization}()} \cr
 \code{\link{op_binary_crossentropy}()} \cr
 \code{\link{op_bincount}()} \cr
@@ -234,9 +248,11 @@ Other ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
 \code{\link{op_cast}()} \cr
 \code{\link{op_categorical_crossentropy}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_celu}()} \cr
 \code{\link{op_cholesky}()} \cr
@@ -249,6 +265,7 @@ Other ops: \cr
 \code{\link{op_convert_to_numpy}()} \cr
 \code{\link{op_convert_to_tensor}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -259,6 +276,7 @@ Other ops: \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
 \code{\link{op_custom_gradient}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_depthwise_conv}()} \cr
 \code{\link{op_det}()} \cr
 \code{\link{op_diag}()} \cr
@@ -297,10 +315,13 @@ Other ops: \cr
 \code{\link{op_get_item}()} \cr
 \code{\link{op_glu}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
 \code{\link{op_hard_shrink}()} \cr
 \code{\link{op_hard_sigmoid}()} \cr
 \code{\link{op_hard_silu}()} \cr
 \code{\link{op_hard_tanh}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -327,6 +348,8 @@ Other ops: \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
 \code{\link{op_istft}()} \cr
+\code{\link{op_kaiser}()} \cr
+\code{\link{op_layer_normalization}()} \cr
 \code{\link{op_leaky_relu}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
@@ -426,6 +449,7 @@ Other ops: \cr
 \code{\link{op_sort}()} \cr
 \code{\link{op_sparse_categorical_crossentropy}()} \cr
 \code{\link{op_sparse_plus}()} \cr
+\code{\link{op_sparse_sigmoid}()} \cr
 \code{\link{op_sparsemax}()} \cr
 \code{\link{op_split}()} \cr
 \code{\link{op_sqrt}()} \cr
@@ -463,6 +487,8 @@ Other ops: \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
 \code{\link{op_vectorized_map}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_while_loop}()} \cr
diff --git a/man/op_greater_equal.Rd b/man/op_greater_equal.Rd
index 3a69f7170..d110940eb 100644
--- a/man/op_greater_equal.Rd
+++ b/man/op_greater_equal.Rd
@@ -47,6 +47,7 @@ Other numpy ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -63,6 +64,7 @@ Other numpy ops: \cr
 \code{\link{op_argsort}()} \cr
 \code{\link{op_array}()} \cr
 \code{\link{op_average}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_bincount}()} \cr
 \code{\link{op_bitwise_and}()} \cr
 \code{\link{op_bitwise_invert}()} \cr
@@ -71,12 +73,15 @@ Other numpy ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_clip}()} \cr
 \code{\link{op_concatenate}()} \cr
 \code{\link{op_conj}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -85,6 +90,7 @@ Other numpy ops: \cr
 \code{\link{op_ctc_decode}()} \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_diag}()} \cr
 \code{\link{op_diagflat}()} \cr
 \code{\link{op_diagonal}()} \cr
@@ -108,6 +114,9 @@ Other numpy ops: \cr
 \code{\link{op_full_like}()} \cr
 \code{\link{op_get_item}()} \cr
 \code{\link{op_greater}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -117,6 +126,7 @@ Other numpy ops: \cr
 \code{\link{op_isfinite}()} \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
+\code{\link{op_kaiser}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
 \code{\link{op_less_equal}()} \cr
@@ -197,6 +207,8 @@ Other numpy ops: \cr
 \code{\link{op_var}()} \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_zeros}()} \cr
@@ -206,6 +218,7 @@ Other ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -224,6 +237,7 @@ Other ops: \cr
 \code{\link{op_associative_scan}()} \cr
 \code{\link{op_average}()} \cr
 \code{\link{op_average_pool}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_batch_normalization}()} \cr
 \code{\link{op_binary_crossentropy}()} \cr
 \code{\link{op_bincount}()} \cr
@@ -234,9 +248,11 @@ Other ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
 \code{\link{op_cast}()} \cr
 \code{\link{op_categorical_crossentropy}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_celu}()} \cr
 \code{\link{op_cholesky}()} \cr
@@ -249,6 +265,7 @@ Other ops: \cr
 \code{\link{op_convert_to_numpy}()} \cr
 \code{\link{op_convert_to_tensor}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -259,6 +276,7 @@ Other ops: \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
 \code{\link{op_custom_gradient}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_depthwise_conv}()} \cr
 \code{\link{op_det}()} \cr
 \code{\link{op_diag}()} \cr
@@ -297,10 +315,13 @@ Other ops: \cr
 \code{\link{op_get_item}()} \cr
 \code{\link{op_glu}()} \cr
 \code{\link{op_greater}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
 \code{\link{op_hard_shrink}()} \cr
 \code{\link{op_hard_sigmoid}()} \cr
 \code{\link{op_hard_silu}()} \cr
 \code{\link{op_hard_tanh}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -327,6 +348,8 @@ Other ops: \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
 \code{\link{op_istft}()} \cr
+\code{\link{op_kaiser}()} \cr
+\code{\link{op_layer_normalization}()} \cr
 \code{\link{op_leaky_relu}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
@@ -426,6 +449,7 @@ Other ops: \cr
 \code{\link{op_sort}()} \cr
 \code{\link{op_sparse_categorical_crossentropy}()} \cr
 \code{\link{op_sparse_plus}()} \cr
+\code{\link{op_sparse_sigmoid}()} \cr
 \code{\link{op_sparsemax}()} \cr
 \code{\link{op_split}()} \cr
 \code{\link{op_sqrt}()} \cr
@@ -463,6 +487,8 @@ Other ops: \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
 \code{\link{op_vectorized_map}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_while_loop}()} \cr
diff --git a/man/op_hamming.Rd b/man/op_hamming.Rd
new file mode 100644
index 000000000..afd701c04
--- /dev/null
+++ b/man/op_hamming.Rd
@@ -0,0 +1,482 @@
+% Generated by roxygen2: do not edit by hand
+% Please edit documentation in R/ops-numpy.R
+\name{op_hamming}
+\alias{op_hamming}
+\title{Hamming window function.}
+\usage{
+op_hamming(x)
+}
+\arguments{
+\item{x}{Length of the window. Must be a positive integer.}
+}
+\value{
+A 1D tensor containing the window values.
+}
+\description{
+The Hamming window is defined as:
+\code{w[n] = 0.54 - 0.46 * cos(2 * pi * n / (N - 1))} for \verb{0 <= n <= N - 1}.
+}
+\section{Examples}{
+\if{html}{\out{<div class="sourceCode r">}}\preformatted{op_hamming(5)
+}\if{html}{\out{</div>}}
+
+\if{html}{\out{<div class="sourceCode">}}\preformatted{## tf.Tensor([0.08000001 0.54       1.         0.54       0.08000001], shape=(5), dtype=float32)
+
+}\if{html}{\out{</div>}}
+}
+
+\seealso{
+Other numpy ops: \cr
+\code{\link{op_abs}()} \cr
+\code{\link{op_add}()} \cr
+\code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
+\code{\link{op_any}()} \cr
+\code{\link{op_append}()} \cr
+\code{\link{op_arange}()} \cr
+\code{\link{op_arccos}()} \cr
+\code{\link{op_arccosh}()} \cr
+\code{\link{op_arcsin}()} \cr
+\code{\link{op_arcsinh}()} \cr
+\code{\link{op_arctan}()} \cr
+\code{\link{op_arctan2}()} \cr
+\code{\link{op_arctanh}()} \cr
+\code{\link{op_argmax}()} \cr
+\code{\link{op_argmin}()} \cr
+\code{\link{op_argpartition}()} \cr
+\code{\link{op_argsort}()} \cr
+\code{\link{op_array}()} \cr
+\code{\link{op_average}()} \cr
+\code{\link{op_bartlett}()} \cr
+\code{\link{op_bincount}()} \cr
+\code{\link{op_bitwise_and}()} \cr
+\code{\link{op_bitwise_invert}()} \cr
+\code{\link{op_bitwise_left_shift}()} \cr
+\code{\link{op_bitwise_not}()} \cr
+\code{\link{op_bitwise_or}()} \cr
+\code{\link{op_bitwise_right_shift}()} \cr
+\code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
+\code{\link{op_broadcast_to}()} \cr
+\code{\link{op_cbrt}()} \cr
+\code{\link{op_ceil}()} \cr
+\code{\link{op_clip}()} \cr
+\code{\link{op_concatenate}()} \cr
+\code{\link{op_conj}()} \cr
+\code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
+\code{\link{op_correlate}()} \cr
+\code{\link{op_cos}()} \cr
+\code{\link{op_cosh}()} \cr
+\code{\link{op_count_nonzero}()} \cr
+\code{\link{op_cross}()} \cr
+\code{\link{op_ctc_decode}()} \cr
+\code{\link{op_cumprod}()} \cr
+\code{\link{op_cumsum}()} \cr
+\code{\link{op_deg2rad}()} \cr
+\code{\link{op_diag}()} \cr
+\code{\link{op_diagflat}()} \cr
+\code{\link{op_diagonal}()} \cr
+\code{\link{op_diff}()} \cr
+\code{\link{op_digitize}()} \cr
+\code{\link{op_divide}()} \cr
+\code{\link{op_divide_no_nan}()} \cr
+\code{\link{op_dot}()} \cr
+\code{\link{op_einsum}()} \cr
+\code{\link{op_empty}()} \cr
+\code{\link{op_equal}()} \cr
+\code{\link{op_exp}()} \cr
+\code{\link{op_exp2}()} \cr
+\code{\link{op_expand_dims}()} \cr
+\code{\link{op_expm1}()} \cr
+\code{\link{op_eye}()} \cr
+\code{\link{op_flip}()} \cr
+\code{\link{op_floor}()} \cr
+\code{\link{op_floor_divide}()} \cr
+\code{\link{op_full}()} \cr
+\code{\link{op_full_like}()} \cr
+\code{\link{op_get_item}()} \cr
+\code{\link{op_greater}()} \cr
+\code{\link{op_greater_equal}()} \cr
+\code{\link{op_hanning}()} \cr
+\code{\link{op_heaviside}()} \cr
+\code{\link{op_histogram}()} \cr
+\code{\link{op_hstack}()} \cr
+\code{\link{op_identity}()} \cr
+\code{\link{op_imag}()} \cr
+\code{\link{op_inner}()} \cr
+\code{\link{op_isclose}()} \cr
+\code{\link{op_isfinite}()} \cr
+\code{\link{op_isinf}()} \cr
+\code{\link{op_isnan}()} \cr
+\code{\link{op_kaiser}()} \cr
+\code{\link{op_left_shift}()} \cr
+\code{\link{op_less}()} \cr
+\code{\link{op_less_equal}()} \cr
+\code{\link{op_linspace}()} \cr
+\code{\link{op_log}()} \cr
+\code{\link{op_log10}()} \cr
+\code{\link{op_log1p}()} \cr
+\code{\link{op_log2}()} \cr
+\code{\link{op_logaddexp}()} \cr
+\code{\link{op_logdet}()} \cr
+\code{\link{op_logical_and}()} \cr
+\code{\link{op_logical_not}()} \cr
+\code{\link{op_logical_or}()} \cr
+\code{\link{op_logical_xor}()} \cr
+\code{\link{op_logspace}()} \cr
+\code{\link{op_lstsq}()} \cr
+\code{\link{op_matmul}()} \cr
+\code{\link{op_max}()} \cr
+\code{\link{op_maximum}()} \cr
+\code{\link{op_mean}()} \cr
+\code{\link{op_median}()} \cr
+\code{\link{op_meshgrid}()} \cr
+\code{\link{op_min}()} \cr
+\code{\link{op_minimum}()} \cr
+\code{\link{op_mod}()} \cr
+\code{\link{op_moveaxis}()} \cr
+\code{\link{op_multiply}()} \cr
+\code{\link{op_nan_to_num}()} \cr
+\code{\link{op_ndim}()} \cr
+\code{\link{op_negative}()} \cr
+\code{\link{op_nonzero}()} \cr
+\code{\link{op_not_equal}()} \cr
+\code{\link{op_ones}()} \cr
+\code{\link{op_ones_like}()} \cr
+\code{\link{op_outer}()} \cr
+\code{\link{op_pad}()} \cr
+\code{\link{op_power}()} \cr
+\code{\link{op_prod}()} \cr
+\code{\link{op_quantile}()} \cr
+\code{\link{op_ravel}()} \cr
+\code{\link{op_real}()} \cr
+\code{\link{op_reciprocal}()} \cr
+\code{\link{op_repeat}()} \cr
+\code{\link{op_reshape}()} \cr
+\code{\link{op_right_shift}()} \cr
+\code{\link{op_roll}()} \cr
+\code{\link{op_rot90}()} \cr
+\code{\link{op_round}()} \cr
+\code{\link{op_saturate_cast}()} \cr
+\code{\link{op_select}()} \cr
+\code{\link{op_sign}()} \cr
+\code{\link{op_signbit}()} \cr
+\code{\link{op_sin}()} \cr
+\code{\link{op_sinh}()} \cr
+\code{\link{op_size}()} \cr
+\code{\link{op_sort}()} \cr
+\code{\link{op_split}()} \cr
+\code{\link{op_sqrt}()} \cr
+\code{\link{op_square}()} \cr
+\code{\link{op_squeeze}()} \cr
+\code{\link{op_stack}()} \cr
+\code{\link{op_std}()} \cr
+\code{\link{op_subtract}()} \cr
+\code{\link{op_sum}()} \cr
+\code{\link{op_swapaxes}()} \cr
+\code{\link{op_take}()} \cr
+\code{\link{op_take_along_axis}()} \cr
+\code{\link{op_tan}()} \cr
+\code{\link{op_tanh}()} \cr
+\code{\link{op_tensordot}()} \cr
+\code{\link{op_tile}()} \cr
+\code{\link{op_trace}()} \cr
+\code{\link{op_transpose}()} \cr
+\code{\link{op_tri}()} \cr
+\code{\link{op_tril}()} \cr
+\code{\link{op_triu}()} \cr
+\code{\link{op_trunc}()} \cr
+\code{\link{op_var}()} \cr
+\code{\link{op_vdot}()} \cr
+\code{\link{op_vectorize}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
+\code{\link{op_vstack}()} \cr
+\code{\link{op_where}()} \cr
+\code{\link{op_zeros}()} \cr
+\code{\link{op_zeros_like}()} \cr
+
+Other ops: \cr
+\code{\link{op_abs}()} \cr
+\code{\link{op_add}()} \cr
+\code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
+\code{\link{op_any}()} \cr
+\code{\link{op_append}()} \cr
+\code{\link{op_arange}()} \cr
+\code{\link{op_arccos}()} \cr
+\code{\link{op_arccosh}()} \cr
+\code{\link{op_arcsin}()} \cr
+\code{\link{op_arcsinh}()} \cr
+\code{\link{op_arctan}()} \cr
+\code{\link{op_arctan2}()} \cr
+\code{\link{op_arctanh}()} \cr
+\code{\link{op_argmax}()} \cr
+\code{\link{op_argmin}()} \cr
+\code{\link{op_argpartition}()} \cr
+\code{\link{op_argsort}()} \cr
+\code{\link{op_array}()} \cr
+\code{\link{op_associative_scan}()} \cr
+\code{\link{op_average}()} \cr
+\code{\link{op_average_pool}()} \cr
+\code{\link{op_bartlett}()} \cr
+\code{\link{op_batch_normalization}()} \cr
+\code{\link{op_binary_crossentropy}()} \cr
+\code{\link{op_bincount}()} \cr
+\code{\link{op_bitwise_and}()} \cr
+\code{\link{op_bitwise_invert}()} \cr
+\code{\link{op_bitwise_left_shift}()} \cr
+\code{\link{op_bitwise_not}()} \cr
+\code{\link{op_bitwise_or}()} \cr
+\code{\link{op_bitwise_right_shift}()} \cr
+\code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
+\code{\link{op_broadcast_to}()} \cr
+\code{\link{op_cast}()} \cr
+\code{\link{op_categorical_crossentropy}()} \cr
+\code{\link{op_cbrt}()} \cr
+\code{\link{op_ceil}()} \cr
+\code{\link{op_celu}()} \cr
+\code{\link{op_cholesky}()} \cr
+\code{\link{op_clip}()} \cr
+\code{\link{op_concatenate}()} \cr
+\code{\link{op_cond}()} \cr
+\code{\link{op_conj}()} \cr
+\code{\link{op_conv}()} \cr
+\code{\link{op_conv_transpose}()} \cr
+\code{\link{op_convert_to_numpy}()} \cr
+\code{\link{op_convert_to_tensor}()} \cr
+\code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
+\code{\link{op_correlate}()} \cr
+\code{\link{op_cos}()} \cr
+\code{\link{op_cosh}()} \cr
+\code{\link{op_count_nonzero}()} \cr
+\code{\link{op_cross}()} \cr
+\code{\link{op_ctc_decode}()} \cr
+\code{\link{op_ctc_loss}()} \cr
+\code{\link{op_cumprod}()} \cr
+\code{\link{op_cumsum}()} \cr
+\code{\link{op_custom_gradient}()} \cr
+\code{\link{op_deg2rad}()} \cr
+\code{\link{op_depthwise_conv}()} \cr
+\code{\link{op_det}()} \cr
+\code{\link{op_diag}()} \cr
+\code{\link{op_diagflat}()} \cr
+\code{\link{op_diagonal}()} \cr
+\code{\link{op_diff}()} \cr
+\code{\link{op_digitize}()} \cr
+\code{\link{op_divide}()} \cr
+\code{\link{op_divide_no_nan}()} \cr
+\code{\link{op_dot}()} \cr
+\code{\link{op_dot_product_attention}()} \cr
+\code{\link{op_dtype}()} \cr
+\code{\link{op_eig}()} \cr
+\code{\link{op_eigh}()} \cr
+\code{\link{op_einsum}()} \cr
+\code{\link{op_elu}()} \cr
+\code{\link{op_empty}()} \cr
+\code{\link{op_equal}()} \cr
+\code{\link{op_erf}()} \cr
+\code{\link{op_erfinv}()} \cr
+\code{\link{op_exp}()} \cr
+\code{\link{op_exp2}()} \cr
+\code{\link{op_expand_dims}()} \cr
+\code{\link{op_expm1}()} \cr
+\code{\link{op_extract_sequences}()} \cr
+\code{\link{op_eye}()} \cr
+\code{\link{op_fft}()} \cr
+\code{\link{op_fft2}()} \cr
+\code{\link{op_flip}()} \cr
+\code{\link{op_floor}()} \cr
+\code{\link{op_floor_divide}()} \cr
+\code{\link{op_fori_loop}()} \cr
+\code{\link{op_full}()} \cr
+\code{\link{op_full_like}()} \cr
+\code{\link{op_gelu}()} \cr
+\code{\link{op_get_item}()} \cr
+\code{\link{op_glu}()} \cr
+\code{\link{op_greater}()} \cr
+\code{\link{op_greater_equal}()} \cr
+\code{\link{op_hanning}()} \cr
+\code{\link{op_hard_shrink}()} \cr
+\code{\link{op_hard_sigmoid}()} \cr
+\code{\link{op_hard_silu}()} \cr
+\code{\link{op_hard_tanh}()} \cr
+\code{\link{op_heaviside}()} \cr
+\code{\link{op_histogram}()} \cr
+\code{\link{op_hstack}()} \cr
+\code{\link{op_identity}()} \cr
+\code{\link{op_ifft2}()} \cr
+\code{\link{op_imag}()} \cr
+\code{\link{op_image_affine_transform}()} \cr
+\code{\link{op_image_crop}()} \cr
+\code{\link{op_image_extract_patches}()} \cr
+\code{\link{op_image_gaussian_blur}()} \cr
+\code{\link{op_image_hsv_to_rgb}()} \cr
+\code{\link{op_image_map_coordinates}()} \cr
+\code{\link{op_image_pad}()} \cr
+\code{\link{op_image_perspective_transform}()} \cr
+\code{\link{op_image_resize}()} \cr
+\code{\link{op_image_rgb_to_grayscale}()} \cr
+\code{\link{op_image_rgb_to_hsv}()} \cr
+\code{\link{op_in_top_k}()} \cr
+\code{\link{op_inner}()} \cr
+\code{\link{op_inv}()} \cr
+\code{\link{op_irfft}()} \cr
+\code{\link{op_is_tensor}()} \cr
+\code{\link{op_isclose}()} \cr
+\code{\link{op_isfinite}()} \cr
+\code{\link{op_isinf}()} \cr
+\code{\link{op_isnan}()} \cr
+\code{\link{op_istft}()} \cr
+\code{\link{op_kaiser}()} \cr
+\code{\link{op_layer_normalization}()} \cr
+\code{\link{op_leaky_relu}()} \cr
+\code{\link{op_left_shift}()} \cr
+\code{\link{op_less}()} \cr
+\code{\link{op_less_equal}()} \cr
+\code{\link{op_linspace}()} \cr
+\code{\link{op_log}()} \cr
+\code{\link{op_log10}()} \cr
+\code{\link{op_log1p}()} \cr
+\code{\link{op_log2}()} \cr
+\code{\link{op_log_sigmoid}()} \cr
+\code{\link{op_log_softmax}()} \cr
+\code{\link{op_logaddexp}()} \cr
+\code{\link{op_logdet}()} \cr
+\code{\link{op_logical_and}()} \cr
+\code{\link{op_logical_not}()} \cr
+\code{\link{op_logical_or}()} \cr
+\code{\link{op_logical_xor}()} \cr
+\code{\link{op_logspace}()} \cr
+\code{\link{op_logsumexp}()} \cr
+\code{\link{op_lstsq}()} \cr
+\code{\link{op_lu_factor}()} \cr
+\code{\link{op_map}()} \cr
+\code{\link{op_matmul}()} \cr
+\code{\link{op_max}()} \cr
+\code{\link{op_max_pool}()} \cr
+\code{\link{op_maximum}()} \cr
+\code{\link{op_mean}()} \cr
+\code{\link{op_median}()} \cr
+\code{\link{op_meshgrid}()} \cr
+\code{\link{op_min}()} \cr
+\code{\link{op_minimum}()} \cr
+\code{\link{op_mod}()} \cr
+\code{\link{op_moments}()} \cr
+\code{\link{op_moveaxis}()} \cr
+\code{\link{op_multi_hot}()} \cr
+\code{\link{op_multiply}()} \cr
+\code{\link{op_nan_to_num}()} \cr
+\code{\link{op_ndim}()} \cr
+\code{\link{op_negative}()} \cr
+\code{\link{op_nonzero}()} \cr
+\code{\link{op_norm}()} \cr
+\code{\link{op_normalize}()} \cr
+\code{\link{op_not_equal}()} \cr
+\code{\link{op_one_hot}()} \cr
+\code{\link{op_ones}()} \cr
+\code{\link{op_ones_like}()} \cr
+\code{\link{op_outer}()} \cr
+\code{\link{op_pad}()} \cr
+\code{\link{op_polar}()} \cr
+\code{\link{op_power}()} \cr
+\code{\link{op_prod}()} \cr
+\code{\link{op_psnr}()} \cr
+\code{\link{op_qr}()} \cr
+\code{\link{op_quantile}()} \cr
+\code{\link{op_ravel}()} \cr
+\code{\link{op_real}()} \cr
+\code{\link{op_rearrange}()} \cr
+\code{\link{op_reciprocal}()} \cr
+\code{\link{op_relu}()} \cr
+\code{\link{op_relu6}()} \cr
+\code{\link{op_repeat}()} \cr
+\code{\link{op_reshape}()} \cr
+\code{\link{op_rfft}()} \cr
+\code{\link{op_right_shift}()} \cr
+\code{\link{op_rms_normalization}()} \cr
+\code{\link{op_roll}()} \cr
+\code{\link{op_rot90}()} \cr
+\code{\link{op_round}()} \cr
+\code{\link{op_rsqrt}()} \cr
+\code{\link{op_saturate_cast}()} \cr
+\code{\link{op_scan}()} \cr
+\code{\link{op_scatter}()} \cr
+\code{\link{op_scatter_update}()} \cr
+\code{\link{op_searchsorted}()} \cr
+\code{\link{op_segment_max}()} \cr
+\code{\link{op_segment_sum}()} \cr
+\code{\link{op_select}()} \cr
+\code{\link{op_selu}()} \cr
+\code{\link{op_separable_conv}()} \cr
+\code{\link{op_shape}()} \cr
+\code{\link{op_sigmoid}()} \cr
+\code{\link{op_sign}()} \cr
+\code{\link{op_signbit}()} \cr
+\code{\link{op_silu}()} \cr
+\code{\link{op_sin}()} \cr
+\code{\link{op_sinh}()} \cr
+\code{\link{op_size}()} \cr
+\code{\link{op_slice}()} \cr
+\code{\link{op_slice_update}()} \cr
+\code{\link{op_slogdet}()} \cr
+\code{\link{op_soft_shrink}()} \cr
+\code{\link{op_softmax}()} \cr
+\code{\link{op_softplus}()} \cr
+\code{\link{op_softsign}()} \cr
+\code{\link{op_solve}()} \cr
+\code{\link{op_solve_triangular}()} \cr
+\code{\link{op_sort}()} \cr
+\code{\link{op_sparse_categorical_crossentropy}()} \cr
+\code{\link{op_sparse_plus}()} \cr
+\code{\link{op_sparse_sigmoid}()} \cr
+\code{\link{op_sparsemax}()} \cr
+\code{\link{op_split}()} \cr
+\code{\link{op_sqrt}()} \cr
+\code{\link{op_square}()} \cr
+\code{\link{op_squareplus}()} \cr
+\code{\link{op_squeeze}()} \cr
+\code{\link{op_stack}()} \cr
+\code{\link{op_std}()} \cr
+\code{\link{op_stft}()} \cr
+\code{\link{op_stop_gradient}()} \cr
+\code{\link{op_subset}()} \cr
+\code{\link{op_subtract}()} \cr
+\code{\link{op_sum}()} \cr
+\code{\link{op_svd}()} \cr
+\code{\link{op_swapaxes}()} \cr
+\code{\link{op_switch}()} \cr
+\code{\link{op_take}()} \cr
+\code{\link{op_take_along_axis}()} \cr
+\code{\link{op_tan}()} \cr
+\code{\link{op_tanh}()} \cr
+\code{\link{op_tanh_shrink}()} \cr
+\code{\link{op_tensordot}()} \cr
+\code{\link{op_threshold}()} \cr
+\code{\link{op_tile}()} \cr
+\code{\link{op_top_k}()} \cr
+\code{\link{op_trace}()} \cr
+\code{\link{op_transpose}()} \cr
+\code{\link{op_tri}()} \cr
+\code{\link{op_tril}()} \cr
+\code{\link{op_triu}()} \cr
+\code{\link{op_trunc}()} \cr
+\code{\link{op_unravel_index}()} \cr
+\code{\link{op_unstack}()} \cr
+\code{\link{op_var}()} \cr
+\code{\link{op_vdot}()} \cr
+\code{\link{op_vectorize}()} \cr
+\code{\link{op_vectorized_map}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
+\code{\link{op_vstack}()} \cr
+\code{\link{op_where}()} \cr
+\code{\link{op_while_loop}()} \cr
+\code{\link{op_zeros}()} \cr
+\code{\link{op_zeros_like}()} \cr
+}
+\concept{numpy ops}
+\concept{ops}
diff --git a/man/op_hanning.Rd b/man/op_hanning.Rd
new file mode 100644
index 000000000..322ce5cfa
--- /dev/null
+++ b/man/op_hanning.Rd
@@ -0,0 +1,482 @@
+% Generated by roxygen2: do not edit by hand
+% Please edit documentation in R/ops-numpy.R
+\name{op_hanning}
+\alias{op_hanning}
+\title{Hanning window function.}
+\usage{
+op_hanning(x)
+}
+\arguments{
+\item{x}{Length of the window. Must be a positive integer.}
+}
+\value{
+A 1D tensor containing the window values.
+}
+\description{
+The Hanning window is defined as:
+\code{w[n] = 0.5 - 0.5 * cos(2 * pi * n / (N - 1))} for \verb{0 <= n <= N - 1}.
+}
+\section{Examples}{
+\if{html}{\out{<div class="sourceCode r">}}\preformatted{op_hanning(5)
+}\if{html}{\out{</div>}}
+
+\if{html}{\out{<div class="sourceCode">}}\preformatted{## tf.Tensor([0.  0.5 1.  0.5 0. ], shape=(5), dtype=float32)
+
+}\if{html}{\out{</div>}}
+}
+
+\seealso{
+Other numpy ops: \cr
+\code{\link{op_abs}()} \cr
+\code{\link{op_add}()} \cr
+\code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
+\code{\link{op_any}()} \cr
+\code{\link{op_append}()} \cr
+\code{\link{op_arange}()} \cr
+\code{\link{op_arccos}()} \cr
+\code{\link{op_arccosh}()} \cr
+\code{\link{op_arcsin}()} \cr
+\code{\link{op_arcsinh}()} \cr
+\code{\link{op_arctan}()} \cr
+\code{\link{op_arctan2}()} \cr
+\code{\link{op_arctanh}()} \cr
+\code{\link{op_argmax}()} \cr
+\code{\link{op_argmin}()} \cr
+\code{\link{op_argpartition}()} \cr
+\code{\link{op_argsort}()} \cr
+\code{\link{op_array}()} \cr
+\code{\link{op_average}()} \cr
+\code{\link{op_bartlett}()} \cr
+\code{\link{op_bincount}()} \cr
+\code{\link{op_bitwise_and}()} \cr
+\code{\link{op_bitwise_invert}()} \cr
+\code{\link{op_bitwise_left_shift}()} \cr
+\code{\link{op_bitwise_not}()} \cr
+\code{\link{op_bitwise_or}()} \cr
+\code{\link{op_bitwise_right_shift}()} \cr
+\code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
+\code{\link{op_broadcast_to}()} \cr
+\code{\link{op_cbrt}()} \cr
+\code{\link{op_ceil}()} \cr
+\code{\link{op_clip}()} \cr
+\code{\link{op_concatenate}()} \cr
+\code{\link{op_conj}()} \cr
+\code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
+\code{\link{op_correlate}()} \cr
+\code{\link{op_cos}()} \cr
+\code{\link{op_cosh}()} \cr
+\code{\link{op_count_nonzero}()} \cr
+\code{\link{op_cross}()} \cr
+\code{\link{op_ctc_decode}()} \cr
+\code{\link{op_cumprod}()} \cr
+\code{\link{op_cumsum}()} \cr
+\code{\link{op_deg2rad}()} \cr
+\code{\link{op_diag}()} \cr
+\code{\link{op_diagflat}()} \cr
+\code{\link{op_diagonal}()} \cr
+\code{\link{op_diff}()} \cr
+\code{\link{op_digitize}()} \cr
+\code{\link{op_divide}()} \cr
+\code{\link{op_divide_no_nan}()} \cr
+\code{\link{op_dot}()} \cr
+\code{\link{op_einsum}()} \cr
+\code{\link{op_empty}()} \cr
+\code{\link{op_equal}()} \cr
+\code{\link{op_exp}()} \cr
+\code{\link{op_exp2}()} \cr
+\code{\link{op_expand_dims}()} \cr
+\code{\link{op_expm1}()} \cr
+\code{\link{op_eye}()} \cr
+\code{\link{op_flip}()} \cr
+\code{\link{op_floor}()} \cr
+\code{\link{op_floor_divide}()} \cr
+\code{\link{op_full}()} \cr
+\code{\link{op_full_like}()} \cr
+\code{\link{op_get_item}()} \cr
+\code{\link{op_greater}()} \cr
+\code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_heaviside}()} \cr
+\code{\link{op_histogram}()} \cr
+\code{\link{op_hstack}()} \cr
+\code{\link{op_identity}()} \cr
+\code{\link{op_imag}()} \cr
+\code{\link{op_inner}()} \cr
+\code{\link{op_isclose}()} \cr
+\code{\link{op_isfinite}()} \cr
+\code{\link{op_isinf}()} \cr
+\code{\link{op_isnan}()} \cr
+\code{\link{op_kaiser}()} \cr
+\code{\link{op_left_shift}()} \cr
+\code{\link{op_less}()} \cr
+\code{\link{op_less_equal}()} \cr
+\code{\link{op_linspace}()} \cr
+\code{\link{op_log}()} \cr
+\code{\link{op_log10}()} \cr
+\code{\link{op_log1p}()} \cr
+\code{\link{op_log2}()} \cr
+\code{\link{op_logaddexp}()} \cr
+\code{\link{op_logdet}()} \cr
+\code{\link{op_logical_and}()} \cr
+\code{\link{op_logical_not}()} \cr
+\code{\link{op_logical_or}()} \cr
+\code{\link{op_logical_xor}()} \cr
+\code{\link{op_logspace}()} \cr
+\code{\link{op_lstsq}()} \cr
+\code{\link{op_matmul}()} \cr
+\code{\link{op_max}()} \cr
+\code{\link{op_maximum}()} \cr
+\code{\link{op_mean}()} \cr
+\code{\link{op_median}()} \cr
+\code{\link{op_meshgrid}()} \cr
+\code{\link{op_min}()} \cr
+\code{\link{op_minimum}()} \cr
+\code{\link{op_mod}()} \cr
+\code{\link{op_moveaxis}()} \cr
+\code{\link{op_multiply}()} \cr
+\code{\link{op_nan_to_num}()} \cr
+\code{\link{op_ndim}()} \cr
+\code{\link{op_negative}()} \cr
+\code{\link{op_nonzero}()} \cr
+\code{\link{op_not_equal}()} \cr
+\code{\link{op_ones}()} \cr
+\code{\link{op_ones_like}()} \cr
+\code{\link{op_outer}()} \cr
+\code{\link{op_pad}()} \cr
+\code{\link{op_power}()} \cr
+\code{\link{op_prod}()} \cr
+\code{\link{op_quantile}()} \cr
+\code{\link{op_ravel}()} \cr
+\code{\link{op_real}()} \cr
+\code{\link{op_reciprocal}()} \cr
+\code{\link{op_repeat}()} \cr
+\code{\link{op_reshape}()} \cr
+\code{\link{op_right_shift}()} \cr
+\code{\link{op_roll}()} \cr
+\code{\link{op_rot90}()} \cr
+\code{\link{op_round}()} \cr
+\code{\link{op_saturate_cast}()} \cr
+\code{\link{op_select}()} \cr
+\code{\link{op_sign}()} \cr
+\code{\link{op_signbit}()} \cr
+\code{\link{op_sin}()} \cr
+\code{\link{op_sinh}()} \cr
+\code{\link{op_size}()} \cr
+\code{\link{op_sort}()} \cr
+\code{\link{op_split}()} \cr
+\code{\link{op_sqrt}()} \cr
+\code{\link{op_square}()} \cr
+\code{\link{op_squeeze}()} \cr
+\code{\link{op_stack}()} \cr
+\code{\link{op_std}()} \cr
+\code{\link{op_subtract}()} \cr
+\code{\link{op_sum}()} \cr
+\code{\link{op_swapaxes}()} \cr
+\code{\link{op_take}()} \cr
+\code{\link{op_take_along_axis}()} \cr
+\code{\link{op_tan}()} \cr
+\code{\link{op_tanh}()} \cr
+\code{\link{op_tensordot}()} \cr
+\code{\link{op_tile}()} \cr
+\code{\link{op_trace}()} \cr
+\code{\link{op_transpose}()} \cr
+\code{\link{op_tri}()} \cr
+\code{\link{op_tril}()} \cr
+\code{\link{op_triu}()} \cr
+\code{\link{op_trunc}()} \cr
+\code{\link{op_var}()} \cr
+\code{\link{op_vdot}()} \cr
+\code{\link{op_vectorize}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
+\code{\link{op_vstack}()} \cr
+\code{\link{op_where}()} \cr
+\code{\link{op_zeros}()} \cr
+\code{\link{op_zeros_like}()} \cr
+
+Other ops: \cr
+\code{\link{op_abs}()} \cr
+\code{\link{op_add}()} \cr
+\code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
+\code{\link{op_any}()} \cr
+\code{\link{op_append}()} \cr
+\code{\link{op_arange}()} \cr
+\code{\link{op_arccos}()} \cr
+\code{\link{op_arccosh}()} \cr
+\code{\link{op_arcsin}()} \cr
+\code{\link{op_arcsinh}()} \cr
+\code{\link{op_arctan}()} \cr
+\code{\link{op_arctan2}()} \cr
+\code{\link{op_arctanh}()} \cr
+\code{\link{op_argmax}()} \cr
+\code{\link{op_argmin}()} \cr
+\code{\link{op_argpartition}()} \cr
+\code{\link{op_argsort}()} \cr
+\code{\link{op_array}()} \cr
+\code{\link{op_associative_scan}()} \cr
+\code{\link{op_average}()} \cr
+\code{\link{op_average_pool}()} \cr
+\code{\link{op_bartlett}()} \cr
+\code{\link{op_batch_normalization}()} \cr
+\code{\link{op_binary_crossentropy}()} \cr
+\code{\link{op_bincount}()} \cr
+\code{\link{op_bitwise_and}()} \cr
+\code{\link{op_bitwise_invert}()} \cr
+\code{\link{op_bitwise_left_shift}()} \cr
+\code{\link{op_bitwise_not}()} \cr
+\code{\link{op_bitwise_or}()} \cr
+\code{\link{op_bitwise_right_shift}()} \cr
+\code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
+\code{\link{op_broadcast_to}()} \cr
+\code{\link{op_cast}()} \cr
+\code{\link{op_categorical_crossentropy}()} \cr
+\code{\link{op_cbrt}()} \cr
+\code{\link{op_ceil}()} \cr
+\code{\link{op_celu}()} \cr
+\code{\link{op_cholesky}()} \cr
+\code{\link{op_clip}()} \cr
+\code{\link{op_concatenate}()} \cr
+\code{\link{op_cond}()} \cr
+\code{\link{op_conj}()} \cr
+\code{\link{op_conv}()} \cr
+\code{\link{op_conv_transpose}()} \cr
+\code{\link{op_convert_to_numpy}()} \cr
+\code{\link{op_convert_to_tensor}()} \cr
+\code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
+\code{\link{op_correlate}()} \cr
+\code{\link{op_cos}()} \cr
+\code{\link{op_cosh}()} \cr
+\code{\link{op_count_nonzero}()} \cr
+\code{\link{op_cross}()} \cr
+\code{\link{op_ctc_decode}()} \cr
+\code{\link{op_ctc_loss}()} \cr
+\code{\link{op_cumprod}()} \cr
+\code{\link{op_cumsum}()} \cr
+\code{\link{op_custom_gradient}()} \cr
+\code{\link{op_deg2rad}()} \cr
+\code{\link{op_depthwise_conv}()} \cr
+\code{\link{op_det}()} \cr
+\code{\link{op_diag}()} \cr
+\code{\link{op_diagflat}()} \cr
+\code{\link{op_diagonal}()} \cr
+\code{\link{op_diff}()} \cr
+\code{\link{op_digitize}()} \cr
+\code{\link{op_divide}()} \cr
+\code{\link{op_divide_no_nan}()} \cr
+\code{\link{op_dot}()} \cr
+\code{\link{op_dot_product_attention}()} \cr
+\code{\link{op_dtype}()} \cr
+\code{\link{op_eig}()} \cr
+\code{\link{op_eigh}()} \cr
+\code{\link{op_einsum}()} \cr
+\code{\link{op_elu}()} \cr
+\code{\link{op_empty}()} \cr
+\code{\link{op_equal}()} \cr
+\code{\link{op_erf}()} \cr
+\code{\link{op_erfinv}()} \cr
+\code{\link{op_exp}()} \cr
+\code{\link{op_exp2}()} \cr
+\code{\link{op_expand_dims}()} \cr
+\code{\link{op_expm1}()} \cr
+\code{\link{op_extract_sequences}()} \cr
+\code{\link{op_eye}()} \cr
+\code{\link{op_fft}()} \cr
+\code{\link{op_fft2}()} \cr
+\code{\link{op_flip}()} \cr
+\code{\link{op_floor}()} \cr
+\code{\link{op_floor_divide}()} \cr
+\code{\link{op_fori_loop}()} \cr
+\code{\link{op_full}()} \cr
+\code{\link{op_full_like}()} \cr
+\code{\link{op_gelu}()} \cr
+\code{\link{op_get_item}()} \cr
+\code{\link{op_glu}()} \cr
+\code{\link{op_greater}()} \cr
+\code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hard_shrink}()} \cr
+\code{\link{op_hard_sigmoid}()} \cr
+\code{\link{op_hard_silu}()} \cr
+\code{\link{op_hard_tanh}()} \cr
+\code{\link{op_heaviside}()} \cr
+\code{\link{op_histogram}()} \cr
+\code{\link{op_hstack}()} \cr
+\code{\link{op_identity}()} \cr
+\code{\link{op_ifft2}()} \cr
+\code{\link{op_imag}()} \cr
+\code{\link{op_image_affine_transform}()} \cr
+\code{\link{op_image_crop}()} \cr
+\code{\link{op_image_extract_patches}()} \cr
+\code{\link{op_image_gaussian_blur}()} \cr
+\code{\link{op_image_hsv_to_rgb}()} \cr
+\code{\link{op_image_map_coordinates}()} \cr
+\code{\link{op_image_pad}()} \cr
+\code{\link{op_image_perspective_transform}()} \cr
+\code{\link{op_image_resize}()} \cr
+\code{\link{op_image_rgb_to_grayscale}()} \cr
+\code{\link{op_image_rgb_to_hsv}()} \cr
+\code{\link{op_in_top_k}()} \cr
+\code{\link{op_inner}()} \cr
+\code{\link{op_inv}()} \cr
+\code{\link{op_irfft}()} \cr
+\code{\link{op_is_tensor}()} \cr
+\code{\link{op_isclose}()} \cr
+\code{\link{op_isfinite}()} \cr
+\code{\link{op_isinf}()} \cr
+\code{\link{op_isnan}()} \cr
+\code{\link{op_istft}()} \cr
+\code{\link{op_kaiser}()} \cr
+\code{\link{op_layer_normalization}()} \cr
+\code{\link{op_leaky_relu}()} \cr
+\code{\link{op_left_shift}()} \cr
+\code{\link{op_less}()} \cr
+\code{\link{op_less_equal}()} \cr
+\code{\link{op_linspace}()} \cr
+\code{\link{op_log}()} \cr
+\code{\link{op_log10}()} \cr
+\code{\link{op_log1p}()} \cr
+\code{\link{op_log2}()} \cr
+\code{\link{op_log_sigmoid}()} \cr
+\code{\link{op_log_softmax}()} \cr
+\code{\link{op_logaddexp}()} \cr
+\code{\link{op_logdet}()} \cr
+\code{\link{op_logical_and}()} \cr
+\code{\link{op_logical_not}()} \cr
+\code{\link{op_logical_or}()} \cr
+\code{\link{op_logical_xor}()} \cr
+\code{\link{op_logspace}()} \cr
+\code{\link{op_logsumexp}()} \cr
+\code{\link{op_lstsq}()} \cr
+\code{\link{op_lu_factor}()} \cr
+\code{\link{op_map}()} \cr
+\code{\link{op_matmul}()} \cr
+\code{\link{op_max}()} \cr
+\code{\link{op_max_pool}()} \cr
+\code{\link{op_maximum}()} \cr
+\code{\link{op_mean}()} \cr
+\code{\link{op_median}()} \cr
+\code{\link{op_meshgrid}()} \cr
+\code{\link{op_min}()} \cr
+\code{\link{op_minimum}()} \cr
+\code{\link{op_mod}()} \cr
+\code{\link{op_moments}()} \cr
+\code{\link{op_moveaxis}()} \cr
+\code{\link{op_multi_hot}()} \cr
+\code{\link{op_multiply}()} \cr
+\code{\link{op_nan_to_num}()} \cr
+\code{\link{op_ndim}()} \cr
+\code{\link{op_negative}()} \cr
+\code{\link{op_nonzero}()} \cr
+\code{\link{op_norm}()} \cr
+\code{\link{op_normalize}()} \cr
+\code{\link{op_not_equal}()} \cr
+\code{\link{op_one_hot}()} \cr
+\code{\link{op_ones}()} \cr
+\code{\link{op_ones_like}()} \cr
+\code{\link{op_outer}()} \cr
+\code{\link{op_pad}()} \cr
+\code{\link{op_polar}()} \cr
+\code{\link{op_power}()} \cr
+\code{\link{op_prod}()} \cr
+\code{\link{op_psnr}()} \cr
+\code{\link{op_qr}()} \cr
+\code{\link{op_quantile}()} \cr
+\code{\link{op_ravel}()} \cr
+\code{\link{op_real}()} \cr
+\code{\link{op_rearrange}()} \cr
+\code{\link{op_reciprocal}()} \cr
+\code{\link{op_relu}()} \cr
+\code{\link{op_relu6}()} \cr
+\code{\link{op_repeat}()} \cr
+\code{\link{op_reshape}()} \cr
+\code{\link{op_rfft}()} \cr
+\code{\link{op_right_shift}()} \cr
+\code{\link{op_rms_normalization}()} \cr
+\code{\link{op_roll}()} \cr
+\code{\link{op_rot90}()} \cr
+\code{\link{op_round}()} \cr
+\code{\link{op_rsqrt}()} \cr
+\code{\link{op_saturate_cast}()} \cr
+\code{\link{op_scan}()} \cr
+\code{\link{op_scatter}()} \cr
+\code{\link{op_scatter_update}()} \cr
+\code{\link{op_searchsorted}()} \cr
+\code{\link{op_segment_max}()} \cr
+\code{\link{op_segment_sum}()} \cr
+\code{\link{op_select}()} \cr
+\code{\link{op_selu}()} \cr
+\code{\link{op_separable_conv}()} \cr
+\code{\link{op_shape}()} \cr
+\code{\link{op_sigmoid}()} \cr
+\code{\link{op_sign}()} \cr
+\code{\link{op_signbit}()} \cr
+\code{\link{op_silu}()} \cr
+\code{\link{op_sin}()} \cr
+\code{\link{op_sinh}()} \cr
+\code{\link{op_size}()} \cr
+\code{\link{op_slice}()} \cr
+\code{\link{op_slice_update}()} \cr
+\code{\link{op_slogdet}()} \cr
+\code{\link{op_soft_shrink}()} \cr
+\code{\link{op_softmax}()} \cr
+\code{\link{op_softplus}()} \cr
+\code{\link{op_softsign}()} \cr
+\code{\link{op_solve}()} \cr
+\code{\link{op_solve_triangular}()} \cr
+\code{\link{op_sort}()} \cr
+\code{\link{op_sparse_categorical_crossentropy}()} \cr
+\code{\link{op_sparse_plus}()} \cr
+\code{\link{op_sparse_sigmoid}()} \cr
+\code{\link{op_sparsemax}()} \cr
+\code{\link{op_split}()} \cr
+\code{\link{op_sqrt}()} \cr
+\code{\link{op_square}()} \cr
+\code{\link{op_squareplus}()} \cr
+\code{\link{op_squeeze}()} \cr
+\code{\link{op_stack}()} \cr
+\code{\link{op_std}()} \cr
+\code{\link{op_stft}()} \cr
+\code{\link{op_stop_gradient}()} \cr
+\code{\link{op_subset}()} \cr
+\code{\link{op_subtract}()} \cr
+\code{\link{op_sum}()} \cr
+\code{\link{op_svd}()} \cr
+\code{\link{op_swapaxes}()} \cr
+\code{\link{op_switch}()} \cr
+\code{\link{op_take}()} \cr
+\code{\link{op_take_along_axis}()} \cr
+\code{\link{op_tan}()} \cr
+\code{\link{op_tanh}()} \cr
+\code{\link{op_tanh_shrink}()} \cr
+\code{\link{op_tensordot}()} \cr
+\code{\link{op_threshold}()} \cr
+\code{\link{op_tile}()} \cr
+\code{\link{op_top_k}()} \cr
+\code{\link{op_trace}()} \cr
+\code{\link{op_transpose}()} \cr
+\code{\link{op_tri}()} \cr
+\code{\link{op_tril}()} \cr
+\code{\link{op_triu}()} \cr
+\code{\link{op_trunc}()} \cr
+\code{\link{op_unravel_index}()} \cr
+\code{\link{op_unstack}()} \cr
+\code{\link{op_var}()} \cr
+\code{\link{op_vdot}()} \cr
+\code{\link{op_vectorize}()} \cr
+\code{\link{op_vectorized_map}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
+\code{\link{op_vstack}()} \cr
+\code{\link{op_where}()} \cr
+\code{\link{op_while_loop}()} \cr
+\code{\link{op_zeros}()} \cr
+\code{\link{op_zeros_like}()} \cr
+}
+\concept{numpy ops}
+\concept{ops}
diff --git a/man/op_hard_shrink.Rd b/man/op_hard_shrink.Rd
index 84946f2b2..f19dac134 100644
--- a/man/op_hard_shrink.Rd
+++ b/man/op_hard_shrink.Rd
@@ -48,6 +48,7 @@ Other nn ops: \cr
 \code{\link{op_hard_sigmoid}()} \cr
 \code{\link{op_hard_silu}()} \cr
 \code{\link{op_hard_tanh}()} \cr
+\code{\link{op_layer_normalization}()} \cr
 \code{\link{op_leaky_relu}()} \cr
 \code{\link{op_log_sigmoid}()} \cr
 \code{\link{op_log_softmax}()} \cr
@@ -71,6 +72,7 @@ Other nn ops: \cr
 \code{\link{op_softsign}()} \cr
 \code{\link{op_sparse_categorical_crossentropy}()} \cr
 \code{\link{op_sparse_plus}()} \cr
+\code{\link{op_sparse_sigmoid}()} \cr
 \code{\link{op_sparsemax}()} \cr
 \code{\link{op_squareplus}()} \cr
 \code{\link{op_tanh_shrink}()} \cr
@@ -81,6 +83,7 @@ Other ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -99,6 +102,7 @@ Other ops: \cr
 \code{\link{op_associative_scan}()} \cr
 \code{\link{op_average}()} \cr
 \code{\link{op_average_pool}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_batch_normalization}()} \cr
 \code{\link{op_binary_crossentropy}()} \cr
 \code{\link{op_bincount}()} \cr
@@ -109,9 +113,11 @@ Other ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
 \code{\link{op_cast}()} \cr
 \code{\link{op_categorical_crossentropy}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_celu}()} \cr
 \code{\link{op_cholesky}()} \cr
@@ -124,6 +130,7 @@ Other ops: \cr
 \code{\link{op_convert_to_numpy}()} \cr
 \code{\link{op_convert_to_tensor}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -134,6 +141,7 @@ Other ops: \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
 \code{\link{op_custom_gradient}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_depthwise_conv}()} \cr
 \code{\link{op_det}()} \cr
 \code{\link{op_diag}()} \cr
@@ -173,9 +181,12 @@ Other ops: \cr
 \code{\link{op_glu}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
 \code{\link{op_hard_sigmoid}()} \cr
 \code{\link{op_hard_silu}()} \cr
 \code{\link{op_hard_tanh}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -202,6 +213,8 @@ Other ops: \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
 \code{\link{op_istft}()} \cr
+\code{\link{op_kaiser}()} \cr
+\code{\link{op_layer_normalization}()} \cr
 \code{\link{op_leaky_relu}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
@@ -301,6 +314,7 @@ Other ops: \cr
 \code{\link{op_sort}()} \cr
 \code{\link{op_sparse_categorical_crossentropy}()} \cr
 \code{\link{op_sparse_plus}()} \cr
+\code{\link{op_sparse_sigmoid}()} \cr
 \code{\link{op_sparsemax}()} \cr
 \code{\link{op_split}()} \cr
 \code{\link{op_sqrt}()} \cr
@@ -338,6 +352,8 @@ Other ops: \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
 \code{\link{op_vectorized_map}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_while_loop}()} \cr
diff --git a/man/op_hard_sigmoid.Rd b/man/op_hard_sigmoid.Rd
index a39560228..4a7591cf4 100644
--- a/man/op_hard_sigmoid.Rd
+++ b/man/op_hard_sigmoid.Rd
@@ -56,6 +56,7 @@ Other nn ops: \cr
 \code{\link{op_hard_shrink}()} \cr
 \code{\link{op_hard_silu}()} \cr
 \code{\link{op_hard_tanh}()} \cr
+\code{\link{op_layer_normalization}()} \cr
 \code{\link{op_leaky_relu}()} \cr
 \code{\link{op_log_sigmoid}()} \cr
 \code{\link{op_log_softmax}()} \cr
@@ -79,6 +80,7 @@ Other nn ops: \cr
 \code{\link{op_softsign}()} \cr
 \code{\link{op_sparse_categorical_crossentropy}()} \cr
 \code{\link{op_sparse_plus}()} \cr
+\code{\link{op_sparse_sigmoid}()} \cr
 \code{\link{op_sparsemax}()} \cr
 \code{\link{op_squareplus}()} \cr
 \code{\link{op_tanh_shrink}()} \cr
@@ -89,6 +91,7 @@ Other ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -107,6 +110,7 @@ Other ops: \cr
 \code{\link{op_associative_scan}()} \cr
 \code{\link{op_average}()} \cr
 \code{\link{op_average_pool}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_batch_normalization}()} \cr
 \code{\link{op_binary_crossentropy}()} \cr
 \code{\link{op_bincount}()} \cr
@@ -117,9 +121,11 @@ Other ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
 \code{\link{op_cast}()} \cr
 \code{\link{op_categorical_crossentropy}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_celu}()} \cr
 \code{\link{op_cholesky}()} \cr
@@ -132,6 +138,7 @@ Other ops: \cr
 \code{\link{op_convert_to_numpy}()} \cr
 \code{\link{op_convert_to_tensor}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -142,6 +149,7 @@ Other ops: \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
 \code{\link{op_custom_gradient}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_depthwise_conv}()} \cr
 \code{\link{op_det}()} \cr
 \code{\link{op_diag}()} \cr
@@ -181,9 +189,12 @@ Other ops: \cr
 \code{\link{op_glu}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
 \code{\link{op_hard_shrink}()} \cr
 \code{\link{op_hard_silu}()} \cr
 \code{\link{op_hard_tanh}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -210,6 +221,8 @@ Other ops: \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
 \code{\link{op_istft}()} \cr
+\code{\link{op_kaiser}()} \cr
+\code{\link{op_layer_normalization}()} \cr
 \code{\link{op_leaky_relu}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
@@ -309,6 +322,7 @@ Other ops: \cr
 \code{\link{op_sort}()} \cr
 \code{\link{op_sparse_categorical_crossentropy}()} \cr
 \code{\link{op_sparse_plus}()} \cr
+\code{\link{op_sparse_sigmoid}()} \cr
 \code{\link{op_sparsemax}()} \cr
 \code{\link{op_split}()} \cr
 \code{\link{op_sqrt}()} \cr
@@ -346,6 +360,8 @@ Other ops: \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
 \code{\link{op_vectorized_map}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_while_loop}()} \cr
diff --git a/man/op_hard_silu.Rd b/man/op_hard_silu.Rd
index ed9768955..cbc3d5103 100644
--- a/man/op_hard_silu.Rd
+++ b/man/op_hard_silu.Rd
@@ -53,6 +53,7 @@ Other nn ops: \cr
 \code{\link{op_hard_shrink}()} \cr
 \code{\link{op_hard_sigmoid}()} \cr
 \code{\link{op_hard_tanh}()} \cr
+\code{\link{op_layer_normalization}()} \cr
 \code{\link{op_leaky_relu}()} \cr
 \code{\link{op_log_sigmoid}()} \cr
 \code{\link{op_log_softmax}()} \cr
@@ -76,6 +77,7 @@ Other nn ops: \cr
 \code{\link{op_softsign}()} \cr
 \code{\link{op_sparse_categorical_crossentropy}()} \cr
 \code{\link{op_sparse_plus}()} \cr
+\code{\link{op_sparse_sigmoid}()} \cr
 \code{\link{op_sparsemax}()} \cr
 \code{\link{op_squareplus}()} \cr
 \code{\link{op_tanh_shrink}()} \cr
@@ -86,6 +88,7 @@ Other ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -104,6 +107,7 @@ Other ops: \cr
 \code{\link{op_associative_scan}()} \cr
 \code{\link{op_average}()} \cr
 \code{\link{op_average_pool}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_batch_normalization}()} \cr
 \code{\link{op_binary_crossentropy}()} \cr
 \code{\link{op_bincount}()} \cr
@@ -114,9 +118,11 @@ Other ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
 \code{\link{op_cast}()} \cr
 \code{\link{op_categorical_crossentropy}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_celu}()} \cr
 \code{\link{op_cholesky}()} \cr
@@ -129,6 +135,7 @@ Other ops: \cr
 \code{\link{op_convert_to_numpy}()} \cr
 \code{\link{op_convert_to_tensor}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -139,6 +146,7 @@ Other ops: \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
 \code{\link{op_custom_gradient}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_depthwise_conv}()} \cr
 \code{\link{op_det}()} \cr
 \code{\link{op_diag}()} \cr
@@ -178,9 +186,12 @@ Other ops: \cr
 \code{\link{op_glu}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
 \code{\link{op_hard_shrink}()} \cr
 \code{\link{op_hard_sigmoid}()} \cr
 \code{\link{op_hard_tanh}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -207,6 +218,8 @@ Other ops: \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
 \code{\link{op_istft}()} \cr
+\code{\link{op_kaiser}()} \cr
+\code{\link{op_layer_normalization}()} \cr
 \code{\link{op_leaky_relu}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
@@ -306,6 +319,7 @@ Other ops: \cr
 \code{\link{op_sort}()} \cr
 \code{\link{op_sparse_categorical_crossentropy}()} \cr
 \code{\link{op_sparse_plus}()} \cr
+\code{\link{op_sparse_sigmoid}()} \cr
 \code{\link{op_sparsemax}()} \cr
 \code{\link{op_split}()} \cr
 \code{\link{op_sqrt}()} \cr
@@ -343,6 +357,8 @@ Other ops: \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
 \code{\link{op_vectorized_map}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_while_loop}()} \cr
diff --git a/man/op_hard_tanh.Rd b/man/op_hard_tanh.Rd
index 0ded77fd8..b4eee2682 100644
--- a/man/op_hard_tanh.Rd
+++ b/man/op_hard_tanh.Rd
@@ -46,6 +46,7 @@ Other nn ops: \cr
 \code{\link{op_hard_shrink}()} \cr
 \code{\link{op_hard_sigmoid}()} \cr
 \code{\link{op_hard_silu}()} \cr
+\code{\link{op_layer_normalization}()} \cr
 \code{\link{op_leaky_relu}()} \cr
 \code{\link{op_log_sigmoid}()} \cr
 \code{\link{op_log_softmax}()} \cr
@@ -69,6 +70,7 @@ Other nn ops: \cr
 \code{\link{op_softsign}()} \cr
 \code{\link{op_sparse_categorical_crossentropy}()} \cr
 \code{\link{op_sparse_plus}()} \cr
+\code{\link{op_sparse_sigmoid}()} \cr
 \code{\link{op_sparsemax}()} \cr
 \code{\link{op_squareplus}()} \cr
 \code{\link{op_tanh_shrink}()} \cr
@@ -79,6 +81,7 @@ Other ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -97,6 +100,7 @@ Other ops: \cr
 \code{\link{op_associative_scan}()} \cr
 \code{\link{op_average}()} \cr
 \code{\link{op_average_pool}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_batch_normalization}()} \cr
 \code{\link{op_binary_crossentropy}()} \cr
 \code{\link{op_bincount}()} \cr
@@ -107,9 +111,11 @@ Other ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
 \code{\link{op_cast}()} \cr
 \code{\link{op_categorical_crossentropy}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_celu}()} \cr
 \code{\link{op_cholesky}()} \cr
@@ -122,6 +128,7 @@ Other ops: \cr
 \code{\link{op_convert_to_numpy}()} \cr
 \code{\link{op_convert_to_tensor}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -132,6 +139,7 @@ Other ops: \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
 \code{\link{op_custom_gradient}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_depthwise_conv}()} \cr
 \code{\link{op_det}()} \cr
 \code{\link{op_diag}()} \cr
@@ -171,9 +179,12 @@ Other ops: \cr
 \code{\link{op_glu}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
 \code{\link{op_hard_shrink}()} \cr
 \code{\link{op_hard_sigmoid}()} \cr
 \code{\link{op_hard_silu}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -200,6 +211,8 @@ Other ops: \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
 \code{\link{op_istft}()} \cr
+\code{\link{op_kaiser}()} \cr
+\code{\link{op_layer_normalization}()} \cr
 \code{\link{op_leaky_relu}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
@@ -299,6 +312,7 @@ Other ops: \cr
 \code{\link{op_sort}()} \cr
 \code{\link{op_sparse_categorical_crossentropy}()} \cr
 \code{\link{op_sparse_plus}()} \cr
+\code{\link{op_sparse_sigmoid}()} \cr
 \code{\link{op_sparsemax}()} \cr
 \code{\link{op_split}()} \cr
 \code{\link{op_sqrt}()} \cr
@@ -336,6 +350,8 @@ Other ops: \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
 \code{\link{op_vectorized_map}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_while_loop}()} \cr
diff --git a/man/op_heaviside.Rd b/man/op_heaviside.Rd
new file mode 100644
index 000000000..9f045cd73
--- /dev/null
+++ b/man/op_heaviside.Rd
@@ -0,0 +1,487 @@
+% Generated by roxygen2: do not edit by hand
+% Please edit documentation in R/ops-numpy.R
+\name{op_heaviside}
+\alias{op_heaviside}
+\title{Heaviside step function.}
+\usage{
+op_heaviside(x1, x2)
+}
+\arguments{
+\item{x1}{Tensor input.}
+
+\item{x2}{Value to use when \code{x1 == 0}.}
+}
+\value{
+A tensor broadcast from \code{x1} and \code{x2} containing \code{0}, \code{1}, or \code{x2}.
+}
+\description{
+The Heaviside step function is defined as:
+\code{heaviside(x1, x2) = 0} if \code{x1 < 0},
+\code{heaviside(x1, x2) = 1} if \code{x1 > 0}, and
+\code{heaviside(x1, x2) = x2} if \code{x1 == 0}.
+}
+\section{Examples}{
+\if{html}{\out{<div class="sourceCode r">}}\preformatted{x1 <- op_array(c(-2, 0, 3))
+op_heaviside(x1, 0.5)
+}\if{html}{\out{</div>}}
+
+\if{html}{\out{<div class="sourceCode">}}\preformatted{## tf.Tensor([0.  0.5 1. ], shape=(3), dtype=float32)
+
+}\if{html}{\out{</div>}}
+}
+
+\seealso{
+Other numpy ops: \cr
+\code{\link{op_abs}()} \cr
+\code{\link{op_add}()} \cr
+\code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
+\code{\link{op_any}()} \cr
+\code{\link{op_append}()} \cr
+\code{\link{op_arange}()} \cr
+\code{\link{op_arccos}()} \cr
+\code{\link{op_arccosh}()} \cr
+\code{\link{op_arcsin}()} \cr
+\code{\link{op_arcsinh}()} \cr
+\code{\link{op_arctan}()} \cr
+\code{\link{op_arctan2}()} \cr
+\code{\link{op_arctanh}()} \cr
+\code{\link{op_argmax}()} \cr
+\code{\link{op_argmin}()} \cr
+\code{\link{op_argpartition}()} \cr
+\code{\link{op_argsort}()} \cr
+\code{\link{op_array}()} \cr
+\code{\link{op_average}()} \cr
+\code{\link{op_bartlett}()} \cr
+\code{\link{op_bincount}()} \cr
+\code{\link{op_bitwise_and}()} \cr
+\code{\link{op_bitwise_invert}()} \cr
+\code{\link{op_bitwise_left_shift}()} \cr
+\code{\link{op_bitwise_not}()} \cr
+\code{\link{op_bitwise_or}()} \cr
+\code{\link{op_bitwise_right_shift}()} \cr
+\code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
+\code{\link{op_broadcast_to}()} \cr
+\code{\link{op_cbrt}()} \cr
+\code{\link{op_ceil}()} \cr
+\code{\link{op_clip}()} \cr
+\code{\link{op_concatenate}()} \cr
+\code{\link{op_conj}()} \cr
+\code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
+\code{\link{op_correlate}()} \cr
+\code{\link{op_cos}()} \cr
+\code{\link{op_cosh}()} \cr
+\code{\link{op_count_nonzero}()} \cr
+\code{\link{op_cross}()} \cr
+\code{\link{op_ctc_decode}()} \cr
+\code{\link{op_cumprod}()} \cr
+\code{\link{op_cumsum}()} \cr
+\code{\link{op_deg2rad}()} \cr
+\code{\link{op_diag}()} \cr
+\code{\link{op_diagflat}()} \cr
+\code{\link{op_diagonal}()} \cr
+\code{\link{op_diff}()} \cr
+\code{\link{op_digitize}()} \cr
+\code{\link{op_divide}()} \cr
+\code{\link{op_divide_no_nan}()} \cr
+\code{\link{op_dot}()} \cr
+\code{\link{op_einsum}()} \cr
+\code{\link{op_empty}()} \cr
+\code{\link{op_equal}()} \cr
+\code{\link{op_exp}()} \cr
+\code{\link{op_exp2}()} \cr
+\code{\link{op_expand_dims}()} \cr
+\code{\link{op_expm1}()} \cr
+\code{\link{op_eye}()} \cr
+\code{\link{op_flip}()} \cr
+\code{\link{op_floor}()} \cr
+\code{\link{op_floor_divide}()} \cr
+\code{\link{op_full}()} \cr
+\code{\link{op_full_like}()} \cr
+\code{\link{op_get_item}()} \cr
+\code{\link{op_greater}()} \cr
+\code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
+\code{\link{op_histogram}()} \cr
+\code{\link{op_hstack}()} \cr
+\code{\link{op_identity}()} \cr
+\code{\link{op_imag}()} \cr
+\code{\link{op_inner}()} \cr
+\code{\link{op_isclose}()} \cr
+\code{\link{op_isfinite}()} \cr
+\code{\link{op_isinf}()} \cr
+\code{\link{op_isnan}()} \cr
+\code{\link{op_kaiser}()} \cr
+\code{\link{op_left_shift}()} \cr
+\code{\link{op_less}()} \cr
+\code{\link{op_less_equal}()} \cr
+\code{\link{op_linspace}()} \cr
+\code{\link{op_log}()} \cr
+\code{\link{op_log10}()} \cr
+\code{\link{op_log1p}()} \cr
+\code{\link{op_log2}()} \cr
+\code{\link{op_logaddexp}()} \cr
+\code{\link{op_logdet}()} \cr
+\code{\link{op_logical_and}()} \cr
+\code{\link{op_logical_not}()} \cr
+\code{\link{op_logical_or}()} \cr
+\code{\link{op_logical_xor}()} \cr
+\code{\link{op_logspace}()} \cr
+\code{\link{op_lstsq}()} \cr
+\code{\link{op_matmul}()} \cr
+\code{\link{op_max}()} \cr
+\code{\link{op_maximum}()} \cr
+\code{\link{op_mean}()} \cr
+\code{\link{op_median}()} \cr
+\code{\link{op_meshgrid}()} \cr
+\code{\link{op_min}()} \cr
+\code{\link{op_minimum}()} \cr
+\code{\link{op_mod}()} \cr
+\code{\link{op_moveaxis}()} \cr
+\code{\link{op_multiply}()} \cr
+\code{\link{op_nan_to_num}()} \cr
+\code{\link{op_ndim}()} \cr
+\code{\link{op_negative}()} \cr
+\code{\link{op_nonzero}()} \cr
+\code{\link{op_not_equal}()} \cr
+\code{\link{op_ones}()} \cr
+\code{\link{op_ones_like}()} \cr
+\code{\link{op_outer}()} \cr
+\code{\link{op_pad}()} \cr
+\code{\link{op_power}()} \cr
+\code{\link{op_prod}()} \cr
+\code{\link{op_quantile}()} \cr
+\code{\link{op_ravel}()} \cr
+\code{\link{op_real}()} \cr
+\code{\link{op_reciprocal}()} \cr
+\code{\link{op_repeat}()} \cr
+\code{\link{op_reshape}()} \cr
+\code{\link{op_right_shift}()} \cr
+\code{\link{op_roll}()} \cr
+\code{\link{op_rot90}()} \cr
+\code{\link{op_round}()} \cr
+\code{\link{op_saturate_cast}()} \cr
+\code{\link{op_select}()} \cr
+\code{\link{op_sign}()} \cr
+\code{\link{op_signbit}()} \cr
+\code{\link{op_sin}()} \cr
+\code{\link{op_sinh}()} \cr
+\code{\link{op_size}()} \cr
+\code{\link{op_sort}()} \cr
+\code{\link{op_split}()} \cr
+\code{\link{op_sqrt}()} \cr
+\code{\link{op_square}()} \cr
+\code{\link{op_squeeze}()} \cr
+\code{\link{op_stack}()} \cr
+\code{\link{op_std}()} \cr
+\code{\link{op_subtract}()} \cr
+\code{\link{op_sum}()} \cr
+\code{\link{op_swapaxes}()} \cr
+\code{\link{op_take}()} \cr
+\code{\link{op_take_along_axis}()} \cr
+\code{\link{op_tan}()} \cr
+\code{\link{op_tanh}()} \cr
+\code{\link{op_tensordot}()} \cr
+\code{\link{op_tile}()} \cr
+\code{\link{op_trace}()} \cr
+\code{\link{op_transpose}()} \cr
+\code{\link{op_tri}()} \cr
+\code{\link{op_tril}()} \cr
+\code{\link{op_triu}()} \cr
+\code{\link{op_trunc}()} \cr
+\code{\link{op_var}()} \cr
+\code{\link{op_vdot}()} \cr
+\code{\link{op_vectorize}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
+\code{\link{op_vstack}()} \cr
+\code{\link{op_where}()} \cr
+\code{\link{op_zeros}()} \cr
+\code{\link{op_zeros_like}()} \cr
+
+Other ops: \cr
+\code{\link{op_abs}()} \cr
+\code{\link{op_add}()} \cr
+\code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
+\code{\link{op_any}()} \cr
+\code{\link{op_append}()} \cr
+\code{\link{op_arange}()} \cr
+\code{\link{op_arccos}()} \cr
+\code{\link{op_arccosh}()} \cr
+\code{\link{op_arcsin}()} \cr
+\code{\link{op_arcsinh}()} \cr
+\code{\link{op_arctan}()} \cr
+\code{\link{op_arctan2}()} \cr
+\code{\link{op_arctanh}()} \cr
+\code{\link{op_argmax}()} \cr
+\code{\link{op_argmin}()} \cr
+\code{\link{op_argpartition}()} \cr
+\code{\link{op_argsort}()} \cr
+\code{\link{op_array}()} \cr
+\code{\link{op_associative_scan}()} \cr
+\code{\link{op_average}()} \cr
+\code{\link{op_average_pool}()} \cr
+\code{\link{op_bartlett}()} \cr
+\code{\link{op_batch_normalization}()} \cr
+\code{\link{op_binary_crossentropy}()} \cr
+\code{\link{op_bincount}()} \cr
+\code{\link{op_bitwise_and}()} \cr
+\code{\link{op_bitwise_invert}()} \cr
+\code{\link{op_bitwise_left_shift}()} \cr
+\code{\link{op_bitwise_not}()} \cr
+\code{\link{op_bitwise_or}()} \cr
+\code{\link{op_bitwise_right_shift}()} \cr
+\code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
+\code{\link{op_broadcast_to}()} \cr
+\code{\link{op_cast}()} \cr
+\code{\link{op_categorical_crossentropy}()} \cr
+\code{\link{op_cbrt}()} \cr
+\code{\link{op_ceil}()} \cr
+\code{\link{op_celu}()} \cr
+\code{\link{op_cholesky}()} \cr
+\code{\link{op_clip}()} \cr
+\code{\link{op_concatenate}()} \cr
+\code{\link{op_cond}()} \cr
+\code{\link{op_conj}()} \cr
+\code{\link{op_conv}()} \cr
+\code{\link{op_conv_transpose}()} \cr
+\code{\link{op_convert_to_numpy}()} \cr
+\code{\link{op_convert_to_tensor}()} \cr
+\code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
+\code{\link{op_correlate}()} \cr
+\code{\link{op_cos}()} \cr
+\code{\link{op_cosh}()} \cr
+\code{\link{op_count_nonzero}()} \cr
+\code{\link{op_cross}()} \cr
+\code{\link{op_ctc_decode}()} \cr
+\code{\link{op_ctc_loss}()} \cr
+\code{\link{op_cumprod}()} \cr
+\code{\link{op_cumsum}()} \cr
+\code{\link{op_custom_gradient}()} \cr
+\code{\link{op_deg2rad}()} \cr
+\code{\link{op_depthwise_conv}()} \cr
+\code{\link{op_det}()} \cr
+\code{\link{op_diag}()} \cr
+\code{\link{op_diagflat}()} \cr
+\code{\link{op_diagonal}()} \cr
+\code{\link{op_diff}()} \cr
+\code{\link{op_digitize}()} \cr
+\code{\link{op_divide}()} \cr
+\code{\link{op_divide_no_nan}()} \cr
+\code{\link{op_dot}()} \cr
+\code{\link{op_dot_product_attention}()} \cr
+\code{\link{op_dtype}()} \cr
+\code{\link{op_eig}()} \cr
+\code{\link{op_eigh}()} \cr
+\code{\link{op_einsum}()} \cr
+\code{\link{op_elu}()} \cr
+\code{\link{op_empty}()} \cr
+\code{\link{op_equal}()} \cr
+\code{\link{op_erf}()} \cr
+\code{\link{op_erfinv}()} \cr
+\code{\link{op_exp}()} \cr
+\code{\link{op_exp2}()} \cr
+\code{\link{op_expand_dims}()} \cr
+\code{\link{op_expm1}()} \cr
+\code{\link{op_extract_sequences}()} \cr
+\code{\link{op_eye}()} \cr
+\code{\link{op_fft}()} \cr
+\code{\link{op_fft2}()} \cr
+\code{\link{op_flip}()} \cr
+\code{\link{op_floor}()} \cr
+\code{\link{op_floor_divide}()} \cr
+\code{\link{op_fori_loop}()} \cr
+\code{\link{op_full}()} \cr
+\code{\link{op_full_like}()} \cr
+\code{\link{op_gelu}()} \cr
+\code{\link{op_get_item}()} \cr
+\code{\link{op_glu}()} \cr
+\code{\link{op_greater}()} \cr
+\code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
+\code{\link{op_hard_shrink}()} \cr
+\code{\link{op_hard_sigmoid}()} \cr
+\code{\link{op_hard_silu}()} \cr
+\code{\link{op_hard_tanh}()} \cr
+\code{\link{op_histogram}()} \cr
+\code{\link{op_hstack}()} \cr
+\code{\link{op_identity}()} \cr
+\code{\link{op_ifft2}()} \cr
+\code{\link{op_imag}()} \cr
+\code{\link{op_image_affine_transform}()} \cr
+\code{\link{op_image_crop}()} \cr
+\code{\link{op_image_extract_patches}()} \cr
+\code{\link{op_image_gaussian_blur}()} \cr
+\code{\link{op_image_hsv_to_rgb}()} \cr
+\code{\link{op_image_map_coordinates}()} \cr
+\code{\link{op_image_pad}()} \cr
+\code{\link{op_image_perspective_transform}()} \cr
+\code{\link{op_image_resize}()} \cr
+\code{\link{op_image_rgb_to_grayscale}()} \cr
+\code{\link{op_image_rgb_to_hsv}()} \cr
+\code{\link{op_in_top_k}()} \cr
+\code{\link{op_inner}()} \cr
+\code{\link{op_inv}()} \cr
+\code{\link{op_irfft}()} \cr
+\code{\link{op_is_tensor}()} \cr
+\code{\link{op_isclose}()} \cr
+\code{\link{op_isfinite}()} \cr
+\code{\link{op_isinf}()} \cr
+\code{\link{op_isnan}()} \cr
+\code{\link{op_istft}()} \cr
+\code{\link{op_kaiser}()} \cr
+\code{\link{op_layer_normalization}()} \cr
+\code{\link{op_leaky_relu}()} \cr
+\code{\link{op_left_shift}()} \cr
+\code{\link{op_less}()} \cr
+\code{\link{op_less_equal}()} \cr
+\code{\link{op_linspace}()} \cr
+\code{\link{op_log}()} \cr
+\code{\link{op_log10}()} \cr
+\code{\link{op_log1p}()} \cr
+\code{\link{op_log2}()} \cr
+\code{\link{op_log_sigmoid}()} \cr
+\code{\link{op_log_softmax}()} \cr
+\code{\link{op_logaddexp}()} \cr
+\code{\link{op_logdet}()} \cr
+\code{\link{op_logical_and}()} \cr
+\code{\link{op_logical_not}()} \cr
+\code{\link{op_logical_or}()} \cr
+\code{\link{op_logical_xor}()} \cr
+\code{\link{op_logspace}()} \cr
+\code{\link{op_logsumexp}()} \cr
+\code{\link{op_lstsq}()} \cr
+\code{\link{op_lu_factor}()} \cr
+\code{\link{op_map}()} \cr
+\code{\link{op_matmul}()} \cr
+\code{\link{op_max}()} \cr
+\code{\link{op_max_pool}()} \cr
+\code{\link{op_maximum}()} \cr
+\code{\link{op_mean}()} \cr
+\code{\link{op_median}()} \cr
+\code{\link{op_meshgrid}()} \cr
+\code{\link{op_min}()} \cr
+\code{\link{op_minimum}()} \cr
+\code{\link{op_mod}()} \cr
+\code{\link{op_moments}()} \cr
+\code{\link{op_moveaxis}()} \cr
+\code{\link{op_multi_hot}()} \cr
+\code{\link{op_multiply}()} \cr
+\code{\link{op_nan_to_num}()} \cr
+\code{\link{op_ndim}()} \cr
+\code{\link{op_negative}()} \cr
+\code{\link{op_nonzero}()} \cr
+\code{\link{op_norm}()} \cr
+\code{\link{op_normalize}()} \cr
+\code{\link{op_not_equal}()} \cr
+\code{\link{op_one_hot}()} \cr
+\code{\link{op_ones}()} \cr
+\code{\link{op_ones_like}()} \cr
+\code{\link{op_outer}()} \cr
+\code{\link{op_pad}()} \cr
+\code{\link{op_polar}()} \cr
+\code{\link{op_power}()} \cr
+\code{\link{op_prod}()} \cr
+\code{\link{op_psnr}()} \cr
+\code{\link{op_qr}()} \cr
+\code{\link{op_quantile}()} \cr
+\code{\link{op_ravel}()} \cr
+\code{\link{op_real}()} \cr
+\code{\link{op_rearrange}()} \cr
+\code{\link{op_reciprocal}()} \cr
+\code{\link{op_relu}()} \cr
+\code{\link{op_relu6}()} \cr
+\code{\link{op_repeat}()} \cr
+\code{\link{op_reshape}()} \cr
+\code{\link{op_rfft}()} \cr
+\code{\link{op_right_shift}()} \cr
+\code{\link{op_rms_normalization}()} \cr
+\code{\link{op_roll}()} \cr
+\code{\link{op_rot90}()} \cr
+\code{\link{op_round}()} \cr
+\code{\link{op_rsqrt}()} \cr
+\code{\link{op_saturate_cast}()} \cr
+\code{\link{op_scan}()} \cr
+\code{\link{op_scatter}()} \cr
+\code{\link{op_scatter_update}()} \cr
+\code{\link{op_searchsorted}()} \cr
+\code{\link{op_segment_max}()} \cr
+\code{\link{op_segment_sum}()} \cr
+\code{\link{op_select}()} \cr
+\code{\link{op_selu}()} \cr
+\code{\link{op_separable_conv}()} \cr
+\code{\link{op_shape}()} \cr
+\code{\link{op_sigmoid}()} \cr
+\code{\link{op_sign}()} \cr
+\code{\link{op_signbit}()} \cr
+\code{\link{op_silu}()} \cr
+\code{\link{op_sin}()} \cr
+\code{\link{op_sinh}()} \cr
+\code{\link{op_size}()} \cr
+\code{\link{op_slice}()} \cr
+\code{\link{op_slice_update}()} \cr
+\code{\link{op_slogdet}()} \cr
+\code{\link{op_soft_shrink}()} \cr
+\code{\link{op_softmax}()} \cr
+\code{\link{op_softplus}()} \cr
+\code{\link{op_softsign}()} \cr
+\code{\link{op_solve}()} \cr
+\code{\link{op_solve_triangular}()} \cr
+\code{\link{op_sort}()} \cr
+\code{\link{op_sparse_categorical_crossentropy}()} \cr
+\code{\link{op_sparse_plus}()} \cr
+\code{\link{op_sparse_sigmoid}()} \cr
+\code{\link{op_sparsemax}()} \cr
+\code{\link{op_split}()} \cr
+\code{\link{op_sqrt}()} \cr
+\code{\link{op_square}()} \cr
+\code{\link{op_squareplus}()} \cr
+\code{\link{op_squeeze}()} \cr
+\code{\link{op_stack}()} \cr
+\code{\link{op_std}()} \cr
+\code{\link{op_stft}()} \cr
+\code{\link{op_stop_gradient}()} \cr
+\code{\link{op_subset}()} \cr
+\code{\link{op_subtract}()} \cr
+\code{\link{op_sum}()} \cr
+\code{\link{op_svd}()} \cr
+\code{\link{op_swapaxes}()} \cr
+\code{\link{op_switch}()} \cr
+\code{\link{op_take}()} \cr
+\code{\link{op_take_along_axis}()} \cr
+\code{\link{op_tan}()} \cr
+\code{\link{op_tanh}()} \cr
+\code{\link{op_tanh_shrink}()} \cr
+\code{\link{op_tensordot}()} \cr
+\code{\link{op_threshold}()} \cr
+\code{\link{op_tile}()} \cr
+\code{\link{op_top_k}()} \cr
+\code{\link{op_trace}()} \cr
+\code{\link{op_transpose}()} \cr
+\code{\link{op_tri}()} \cr
+\code{\link{op_tril}()} \cr
+\code{\link{op_triu}()} \cr
+\code{\link{op_trunc}()} \cr
+\code{\link{op_unravel_index}()} \cr
+\code{\link{op_unstack}()} \cr
+\code{\link{op_var}()} \cr
+\code{\link{op_vdot}()} \cr
+\code{\link{op_vectorize}()} \cr
+\code{\link{op_vectorized_map}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
+\code{\link{op_vstack}()} \cr
+\code{\link{op_where}()} \cr
+\code{\link{op_while_loop}()} \cr
+\code{\link{op_zeros}()} \cr
+\code{\link{op_zeros_like}()} \cr
+}
+\concept{numpy ops}
+\concept{ops}
diff --git a/man/op_histogram.Rd b/man/op_histogram.Rd
index bd65af15e..7d7febc5f 100644
--- a/man/op_histogram.Rd
+++ b/man/op_histogram.Rd
@@ -51,6 +51,7 @@ Other numpy ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -67,6 +68,7 @@ Other numpy ops: \cr
 \code{\link{op_argsort}()} \cr
 \code{\link{op_array}()} \cr
 \code{\link{op_average}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_bincount}()} \cr
 \code{\link{op_bitwise_and}()} \cr
 \code{\link{op_bitwise_invert}()} \cr
@@ -75,12 +77,15 @@ Other numpy ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_clip}()} \cr
 \code{\link{op_concatenate}()} \cr
 \code{\link{op_conj}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -89,6 +94,7 @@ Other numpy ops: \cr
 \code{\link{op_ctc_decode}()} \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_diag}()} \cr
 \code{\link{op_diagflat}()} \cr
 \code{\link{op_diagonal}()} \cr
@@ -113,6 +119,9 @@ Other numpy ops: \cr
 \code{\link{op_get_item}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
 \code{\link{op_imag}()} \cr
@@ -121,6 +130,7 @@ Other numpy ops: \cr
 \code{\link{op_isfinite}()} \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
+\code{\link{op_kaiser}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
 \code{\link{op_less_equal}()} \cr
@@ -201,6 +211,8 @@ Other numpy ops: \cr
 \code{\link{op_var}()} \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_zeros}()} \cr
@@ -210,6 +222,7 @@ Other ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -228,6 +241,7 @@ Other ops: \cr
 \code{\link{op_associative_scan}()} \cr
 \code{\link{op_average}()} \cr
 \code{\link{op_average_pool}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_batch_normalization}()} \cr
 \code{\link{op_binary_crossentropy}()} \cr
 \code{\link{op_bincount}()} \cr
@@ -238,9 +252,11 @@ Other ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
 \code{\link{op_cast}()} \cr
 \code{\link{op_categorical_crossentropy}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_celu}()} \cr
 \code{\link{op_cholesky}()} \cr
@@ -253,6 +269,7 @@ Other ops: \cr
 \code{\link{op_convert_to_numpy}()} \cr
 \code{\link{op_convert_to_tensor}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -263,6 +280,7 @@ Other ops: \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
 \code{\link{op_custom_gradient}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_depthwise_conv}()} \cr
 \code{\link{op_det}()} \cr
 \code{\link{op_diag}()} \cr
@@ -302,10 +320,13 @@ Other ops: \cr
 \code{\link{op_glu}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
 \code{\link{op_hard_shrink}()} \cr
 \code{\link{op_hard_sigmoid}()} \cr
 \code{\link{op_hard_silu}()} \cr
 \code{\link{op_hard_tanh}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
 \code{\link{op_ifft2}()} \cr
@@ -331,6 +352,8 @@ Other ops: \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
 \code{\link{op_istft}()} \cr
+\code{\link{op_kaiser}()} \cr
+\code{\link{op_layer_normalization}()} \cr
 \code{\link{op_leaky_relu}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
@@ -430,6 +453,7 @@ Other ops: \cr
 \code{\link{op_sort}()} \cr
 \code{\link{op_sparse_categorical_crossentropy}()} \cr
 \code{\link{op_sparse_plus}()} \cr
+\code{\link{op_sparse_sigmoid}()} \cr
 \code{\link{op_sparsemax}()} \cr
 \code{\link{op_split}()} \cr
 \code{\link{op_sqrt}()} \cr
@@ -467,6 +491,8 @@ Other ops: \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
 \code{\link{op_vectorized_map}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_while_loop}()} \cr
diff --git a/man/op_hstack.Rd b/man/op_hstack.Rd
index e02fd35be..4b17e5563 100644
--- a/man/op_hstack.Rd
+++ b/man/op_hstack.Rd
@@ -25,6 +25,7 @@ Other numpy ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -41,6 +42,7 @@ Other numpy ops: \cr
 \code{\link{op_argsort}()} \cr
 \code{\link{op_array}()} \cr
 \code{\link{op_average}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_bincount}()} \cr
 \code{\link{op_bitwise_and}()} \cr
 \code{\link{op_bitwise_invert}()} \cr
@@ -49,12 +51,15 @@ Other numpy ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_clip}()} \cr
 \code{\link{op_concatenate}()} \cr
 \code{\link{op_conj}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -63,6 +68,7 @@ Other numpy ops: \cr
 \code{\link{op_ctc_decode}()} \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_diag}()} \cr
 \code{\link{op_diagflat}()} \cr
 \code{\link{op_diagonal}()} \cr
@@ -87,6 +93,9 @@ Other numpy ops: \cr
 \code{\link{op_get_item}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_identity}()} \cr
 \code{\link{op_imag}()} \cr
@@ -95,6 +104,7 @@ Other numpy ops: \cr
 \code{\link{op_isfinite}()} \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
+\code{\link{op_kaiser}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
 \code{\link{op_less_equal}()} \cr
@@ -175,6 +185,8 @@ Other numpy ops: \cr
 \code{\link{op_var}()} \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_zeros}()} \cr
@@ -184,6 +196,7 @@ Other ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -202,6 +215,7 @@ Other ops: \cr
 \code{\link{op_associative_scan}()} \cr
 \code{\link{op_average}()} \cr
 \code{\link{op_average_pool}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_batch_normalization}()} \cr
 \code{\link{op_binary_crossentropy}()} \cr
 \code{\link{op_bincount}()} \cr
@@ -212,9 +226,11 @@ Other ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
 \code{\link{op_cast}()} \cr
 \code{\link{op_categorical_crossentropy}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_celu}()} \cr
 \code{\link{op_cholesky}()} \cr
@@ -227,6 +243,7 @@ Other ops: \cr
 \code{\link{op_convert_to_numpy}()} \cr
 \code{\link{op_convert_to_tensor}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -237,6 +254,7 @@ Other ops: \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
 \code{\link{op_custom_gradient}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_depthwise_conv}()} \cr
 \code{\link{op_det}()} \cr
 \code{\link{op_diag}()} \cr
@@ -276,10 +294,13 @@ Other ops: \cr
 \code{\link{op_glu}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
 \code{\link{op_hard_shrink}()} \cr
 \code{\link{op_hard_sigmoid}()} \cr
 \code{\link{op_hard_silu}()} \cr
 \code{\link{op_hard_tanh}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_identity}()} \cr
 \code{\link{op_ifft2}()} \cr
@@ -305,6 +326,8 @@ Other ops: \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
 \code{\link{op_istft}()} \cr
+\code{\link{op_kaiser}()} \cr
+\code{\link{op_layer_normalization}()} \cr
 \code{\link{op_leaky_relu}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
@@ -404,6 +427,7 @@ Other ops: \cr
 \code{\link{op_sort}()} \cr
 \code{\link{op_sparse_categorical_crossentropy}()} \cr
 \code{\link{op_sparse_plus}()} \cr
+\code{\link{op_sparse_sigmoid}()} \cr
 \code{\link{op_sparsemax}()} \cr
 \code{\link{op_split}()} \cr
 \code{\link{op_sqrt}()} \cr
@@ -441,6 +465,8 @@ Other ops: \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
 \code{\link{op_vectorized_map}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_while_loop}()} \cr
diff --git a/man/op_identity.Rd b/man/op_identity.Rd
index 38a171be7..28cf61d25 100644
--- a/man/op_identity.Rd
+++ b/man/op_identity.Rd
@@ -27,6 +27,7 @@ Other numpy ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -43,6 +44,7 @@ Other numpy ops: \cr
 \code{\link{op_argsort}()} \cr
 \code{\link{op_array}()} \cr
 \code{\link{op_average}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_bincount}()} \cr
 \code{\link{op_bitwise_and}()} \cr
 \code{\link{op_bitwise_invert}()} \cr
@@ -51,12 +53,15 @@ Other numpy ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_clip}()} \cr
 \code{\link{op_concatenate}()} \cr
 \code{\link{op_conj}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -65,6 +70,7 @@ Other numpy ops: \cr
 \code{\link{op_ctc_decode}()} \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_diag}()} \cr
 \code{\link{op_diagflat}()} \cr
 \code{\link{op_diagonal}()} \cr
@@ -89,6 +95,9 @@ Other numpy ops: \cr
 \code{\link{op_get_item}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_imag}()} \cr
@@ -97,6 +106,7 @@ Other numpy ops: \cr
 \code{\link{op_isfinite}()} \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
+\code{\link{op_kaiser}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
 \code{\link{op_less_equal}()} \cr
@@ -177,6 +187,8 @@ Other numpy ops: \cr
 \code{\link{op_var}()} \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_zeros}()} \cr
@@ -186,6 +198,7 @@ Other ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -204,6 +217,7 @@ Other ops: \cr
 \code{\link{op_associative_scan}()} \cr
 \code{\link{op_average}()} \cr
 \code{\link{op_average_pool}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_batch_normalization}()} \cr
 \code{\link{op_binary_crossentropy}()} \cr
 \code{\link{op_bincount}()} \cr
@@ -214,9 +228,11 @@ Other ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
 \code{\link{op_cast}()} \cr
 \code{\link{op_categorical_crossentropy}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_celu}()} \cr
 \code{\link{op_cholesky}()} \cr
@@ -229,6 +245,7 @@ Other ops: \cr
 \code{\link{op_convert_to_numpy}()} \cr
 \code{\link{op_convert_to_tensor}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -239,6 +256,7 @@ Other ops: \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
 \code{\link{op_custom_gradient}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_depthwise_conv}()} \cr
 \code{\link{op_det}()} \cr
 \code{\link{op_diag}()} \cr
@@ -278,10 +296,13 @@ Other ops: \cr
 \code{\link{op_glu}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
 \code{\link{op_hard_shrink}()} \cr
 \code{\link{op_hard_sigmoid}()} \cr
 \code{\link{op_hard_silu}()} \cr
 \code{\link{op_hard_tanh}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_ifft2}()} \cr
@@ -307,6 +328,8 @@ Other ops: \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
 \code{\link{op_istft}()} \cr
+\code{\link{op_kaiser}()} \cr
+\code{\link{op_layer_normalization}()} \cr
 \code{\link{op_leaky_relu}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
@@ -406,6 +429,7 @@ Other ops: \cr
 \code{\link{op_sort}()} \cr
 \code{\link{op_sparse_categorical_crossentropy}()} \cr
 \code{\link{op_sparse_plus}()} \cr
+\code{\link{op_sparse_sigmoid}()} \cr
 \code{\link{op_sparsemax}()} \cr
 \code{\link{op_split}()} \cr
 \code{\link{op_sqrt}()} \cr
@@ -443,6 +467,8 @@ Other ops: \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
 \code{\link{op_vectorized_map}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_while_loop}()} \cr
diff --git a/man/op_ifft2.Rd b/man/op_ifft2.Rd
index 03aa9e9ed..62c2ad801 100644
--- a/man/op_ifft2.Rd
+++ b/man/op_ifft2.Rd
@@ -60,6 +60,7 @@ Other ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -78,6 +79,7 @@ Other ops: \cr
 \code{\link{op_associative_scan}()} \cr
 \code{\link{op_average}()} \cr
 \code{\link{op_average_pool}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_batch_normalization}()} \cr
 \code{\link{op_binary_crossentropy}()} \cr
 \code{\link{op_bincount}()} \cr
@@ -88,9 +90,11 @@ Other ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
 \code{\link{op_cast}()} \cr
 \code{\link{op_categorical_crossentropy}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_celu}()} \cr
 \code{\link{op_cholesky}()} \cr
@@ -103,6 +107,7 @@ Other ops: \cr
 \code{\link{op_convert_to_numpy}()} \cr
 \code{\link{op_convert_to_tensor}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -113,6 +118,7 @@ Other ops: \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
 \code{\link{op_custom_gradient}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_depthwise_conv}()} \cr
 \code{\link{op_det}()} \cr
 \code{\link{op_diag}()} \cr
@@ -152,10 +158,13 @@ Other ops: \cr
 \code{\link{op_glu}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
 \code{\link{op_hard_shrink}()} \cr
 \code{\link{op_hard_sigmoid}()} \cr
 \code{\link{op_hard_silu}()} \cr
 \code{\link{op_hard_tanh}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -181,6 +190,8 @@ Other ops: \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
 \code{\link{op_istft}()} \cr
+\code{\link{op_kaiser}()} \cr
+\code{\link{op_layer_normalization}()} \cr
 \code{\link{op_leaky_relu}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
@@ -280,6 +291,7 @@ Other ops: \cr
 \code{\link{op_sort}()} \cr
 \code{\link{op_sparse_categorical_crossentropy}()} \cr
 \code{\link{op_sparse_plus}()} \cr
+\code{\link{op_sparse_sigmoid}()} \cr
 \code{\link{op_sparsemax}()} \cr
 \code{\link{op_split}()} \cr
 \code{\link{op_sqrt}()} \cr
@@ -317,6 +329,8 @@ Other ops: \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
 \code{\link{op_vectorized_map}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_while_loop}()} \cr
diff --git a/man/op_imag.Rd b/man/op_imag.Rd
index 65856842f..bccd03a8b 100644
--- a/man/op_imag.Rd
+++ b/man/op_imag.Rd
@@ -24,6 +24,7 @@ Other numpy ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -40,6 +41,7 @@ Other numpy ops: \cr
 \code{\link{op_argsort}()} \cr
 \code{\link{op_array}()} \cr
 \code{\link{op_average}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_bincount}()} \cr
 \code{\link{op_bitwise_and}()} \cr
 \code{\link{op_bitwise_invert}()} \cr
@@ -48,12 +50,15 @@ Other numpy ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_clip}()} \cr
 \code{\link{op_concatenate}()} \cr
 \code{\link{op_conj}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -62,6 +67,7 @@ Other numpy ops: \cr
 \code{\link{op_ctc_decode}()} \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_diag}()} \cr
 \code{\link{op_diagflat}()} \cr
 \code{\link{op_diagonal}()} \cr
@@ -86,6 +92,9 @@ Other numpy ops: \cr
 \code{\link{op_get_item}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -94,6 +103,7 @@ Other numpy ops: \cr
 \code{\link{op_isfinite}()} \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
+\code{\link{op_kaiser}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
 \code{\link{op_less_equal}()} \cr
@@ -174,6 +184,8 @@ Other numpy ops: \cr
 \code{\link{op_var}()} \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_zeros}()} \cr
@@ -183,6 +195,7 @@ Other ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -201,6 +214,7 @@ Other ops: \cr
 \code{\link{op_associative_scan}()} \cr
 \code{\link{op_average}()} \cr
 \code{\link{op_average_pool}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_batch_normalization}()} \cr
 \code{\link{op_binary_crossentropy}()} \cr
 \code{\link{op_bincount}()} \cr
@@ -211,9 +225,11 @@ Other ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
 \code{\link{op_cast}()} \cr
 \code{\link{op_categorical_crossentropy}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_celu}()} \cr
 \code{\link{op_cholesky}()} \cr
@@ -226,6 +242,7 @@ Other ops: \cr
 \code{\link{op_convert_to_numpy}()} \cr
 \code{\link{op_convert_to_tensor}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -236,6 +253,7 @@ Other ops: \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
 \code{\link{op_custom_gradient}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_depthwise_conv}()} \cr
 \code{\link{op_det}()} \cr
 \code{\link{op_diag}()} \cr
@@ -275,10 +293,13 @@ Other ops: \cr
 \code{\link{op_glu}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
 \code{\link{op_hard_shrink}()} \cr
 \code{\link{op_hard_sigmoid}()} \cr
 \code{\link{op_hard_silu}()} \cr
 \code{\link{op_hard_tanh}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -304,6 +325,8 @@ Other ops: \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
 \code{\link{op_istft}()} \cr
+\code{\link{op_kaiser}()} \cr
+\code{\link{op_layer_normalization}()} \cr
 \code{\link{op_leaky_relu}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
@@ -403,6 +426,7 @@ Other ops: \cr
 \code{\link{op_sort}()} \cr
 \code{\link{op_sparse_categorical_crossentropy}()} \cr
 \code{\link{op_sparse_plus}()} \cr
+\code{\link{op_sparse_sigmoid}()} \cr
 \code{\link{op_sparsemax}()} \cr
 \code{\link{op_split}()} \cr
 \code{\link{op_sqrt}()} \cr
@@ -440,6 +464,8 @@ Other ops: \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
 \code{\link{op_vectorized_map}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_while_loop}()} \cr
diff --git a/man/op_image_affine_transform.Rd b/man/op_image_affine_transform.Rd
index a4a937853..e62108967 100644
--- a/man/op_image_affine_transform.Rd
+++ b/man/op_image_affine_transform.Rd
@@ -148,6 +148,7 @@ Other ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -166,6 +167,7 @@ Other ops: \cr
 \code{\link{op_associative_scan}()} \cr
 \code{\link{op_average}()} \cr
 \code{\link{op_average_pool}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_batch_normalization}()} \cr
 \code{\link{op_binary_crossentropy}()} \cr
 \code{\link{op_bincount}()} \cr
@@ -176,9 +178,11 @@ Other ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
 \code{\link{op_cast}()} \cr
 \code{\link{op_categorical_crossentropy}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_celu}()} \cr
 \code{\link{op_cholesky}()} \cr
@@ -191,6 +195,7 @@ Other ops: \cr
 \code{\link{op_convert_to_numpy}()} \cr
 \code{\link{op_convert_to_tensor}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -201,6 +206,7 @@ Other ops: \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
 \code{\link{op_custom_gradient}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_depthwise_conv}()} \cr
 \code{\link{op_det}()} \cr
 \code{\link{op_diag}()} \cr
@@ -240,10 +246,13 @@ Other ops: \cr
 \code{\link{op_glu}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
 \code{\link{op_hard_shrink}()} \cr
 \code{\link{op_hard_sigmoid}()} \cr
 \code{\link{op_hard_silu}()} \cr
 \code{\link{op_hard_tanh}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -269,6 +278,8 @@ Other ops: \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
 \code{\link{op_istft}()} \cr
+\code{\link{op_kaiser}()} \cr
+\code{\link{op_layer_normalization}()} \cr
 \code{\link{op_leaky_relu}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
@@ -368,6 +379,7 @@ Other ops: \cr
 \code{\link{op_sort}()} \cr
 \code{\link{op_sparse_categorical_crossentropy}()} \cr
 \code{\link{op_sparse_plus}()} \cr
+\code{\link{op_sparse_sigmoid}()} \cr
 \code{\link{op_sparsemax}()} \cr
 \code{\link{op_split}()} \cr
 \code{\link{op_sqrt}()} \cr
@@ -405,6 +417,8 @@ Other ops: \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
 \code{\link{op_vectorized_map}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_while_loop}()} \cr
diff --git a/man/op_image_crop.Rd b/man/op_image_crop.Rd
index 2a1640f6c..3a7c0908b 100644
--- a/man/op_image_crop.Rd
+++ b/man/op_image_crop.Rd
@@ -91,6 +91,7 @@ Other ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -109,6 +110,7 @@ Other ops: \cr
 \code{\link{op_associative_scan}()} \cr
 \code{\link{op_average}()} \cr
 \code{\link{op_average_pool}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_batch_normalization}()} \cr
 \code{\link{op_binary_crossentropy}()} \cr
 \code{\link{op_bincount}()} \cr
@@ -119,9 +121,11 @@ Other ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
 \code{\link{op_cast}()} \cr
 \code{\link{op_categorical_crossentropy}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_celu}()} \cr
 \code{\link{op_cholesky}()} \cr
@@ -134,6 +138,7 @@ Other ops: \cr
 \code{\link{op_convert_to_numpy}()} \cr
 \code{\link{op_convert_to_tensor}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -144,6 +149,7 @@ Other ops: \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
 \code{\link{op_custom_gradient}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_depthwise_conv}()} \cr
 \code{\link{op_det}()} \cr
 \code{\link{op_diag}()} \cr
@@ -183,10 +189,13 @@ Other ops: \cr
 \code{\link{op_glu}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
 \code{\link{op_hard_shrink}()} \cr
 \code{\link{op_hard_sigmoid}()} \cr
 \code{\link{op_hard_silu}()} \cr
 \code{\link{op_hard_tanh}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -212,6 +221,8 @@ Other ops: \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
 \code{\link{op_istft}()} \cr
+\code{\link{op_kaiser}()} \cr
+\code{\link{op_layer_normalization}()} \cr
 \code{\link{op_leaky_relu}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
@@ -311,6 +322,7 @@ Other ops: \cr
 \code{\link{op_sort}()} \cr
 \code{\link{op_sparse_categorical_crossentropy}()} \cr
 \code{\link{op_sparse_plus}()} \cr
+\code{\link{op_sparse_sigmoid}()} \cr
 \code{\link{op_sparsemax}()} \cr
 \code{\link{op_split}()} \cr
 \code{\link{op_sqrt}()} \cr
@@ -348,6 +360,8 @@ Other ops: \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
 \code{\link{op_vectorized_map}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_while_loop}()} \cr
diff --git a/man/op_image_extract_patches.Rd b/man/op_image_extract_patches.Rd
index 9ff9c97e9..5acf402de 100644
--- a/man/op_image_extract_patches.Rd
+++ b/man/op_image_extract_patches.Rd
@@ -104,6 +104,7 @@ Other ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -122,6 +123,7 @@ Other ops: \cr
 \code{\link{op_associative_scan}()} \cr
 \code{\link{op_average}()} \cr
 \code{\link{op_average_pool}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_batch_normalization}()} \cr
 \code{\link{op_binary_crossentropy}()} \cr
 \code{\link{op_bincount}()} \cr
@@ -132,9 +134,11 @@ Other ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
 \code{\link{op_cast}()} \cr
 \code{\link{op_categorical_crossentropy}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_celu}()} \cr
 \code{\link{op_cholesky}()} \cr
@@ -147,6 +151,7 @@ Other ops: \cr
 \code{\link{op_convert_to_numpy}()} \cr
 \code{\link{op_convert_to_tensor}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -157,6 +162,7 @@ Other ops: \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
 \code{\link{op_custom_gradient}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_depthwise_conv}()} \cr
 \code{\link{op_det}()} \cr
 \code{\link{op_diag}()} \cr
@@ -196,10 +202,13 @@ Other ops: \cr
 \code{\link{op_glu}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
 \code{\link{op_hard_shrink}()} \cr
 \code{\link{op_hard_sigmoid}()} \cr
 \code{\link{op_hard_silu}()} \cr
 \code{\link{op_hard_tanh}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -225,6 +234,8 @@ Other ops: \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
 \code{\link{op_istft}()} \cr
+\code{\link{op_kaiser}()} \cr
+\code{\link{op_layer_normalization}()} \cr
 \code{\link{op_leaky_relu}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
@@ -324,6 +335,7 @@ Other ops: \cr
 \code{\link{op_sort}()} \cr
 \code{\link{op_sparse_categorical_crossentropy}()} \cr
 \code{\link{op_sparse_plus}()} \cr
+\code{\link{op_sparse_sigmoid}()} \cr
 \code{\link{op_sparsemax}()} \cr
 \code{\link{op_split}()} \cr
 \code{\link{op_sqrt}()} \cr
@@ -361,6 +373,8 @@ Other ops: \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
 \code{\link{op_vectorized_map}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_while_loop}()} \cr
diff --git a/man/op_image_gaussian_blur.Rd b/man/op_image_gaussian_blur.Rd
index ec0907d76..5627e1045 100644
--- a/man/op_image_gaussian_blur.Rd
+++ b/man/op_image_gaussian_blur.Rd
@@ -97,6 +97,7 @@ Other ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -115,6 +116,7 @@ Other ops: \cr
 \code{\link{op_associative_scan}()} \cr
 \code{\link{op_average}()} \cr
 \code{\link{op_average_pool}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_batch_normalization}()} \cr
 \code{\link{op_binary_crossentropy}()} \cr
 \code{\link{op_bincount}()} \cr
@@ -125,9 +127,11 @@ Other ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
 \code{\link{op_cast}()} \cr
 \code{\link{op_categorical_crossentropy}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_celu}()} \cr
 \code{\link{op_cholesky}()} \cr
@@ -140,6 +144,7 @@ Other ops: \cr
 \code{\link{op_convert_to_numpy}()} \cr
 \code{\link{op_convert_to_tensor}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -150,6 +155,7 @@ Other ops: \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
 \code{\link{op_custom_gradient}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_depthwise_conv}()} \cr
 \code{\link{op_det}()} \cr
 \code{\link{op_diag}()} \cr
@@ -189,10 +195,13 @@ Other ops: \cr
 \code{\link{op_glu}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
 \code{\link{op_hard_shrink}()} \cr
 \code{\link{op_hard_sigmoid}()} \cr
 \code{\link{op_hard_silu}()} \cr
 \code{\link{op_hard_tanh}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -218,6 +227,8 @@ Other ops: \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
 \code{\link{op_istft}()} \cr
+\code{\link{op_kaiser}()} \cr
+\code{\link{op_layer_normalization}()} \cr
 \code{\link{op_leaky_relu}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
@@ -317,6 +328,7 @@ Other ops: \cr
 \code{\link{op_sort}()} \cr
 \code{\link{op_sparse_categorical_crossentropy}()} \cr
 \code{\link{op_sparse_plus}()} \cr
+\code{\link{op_sparse_sigmoid}()} \cr
 \code{\link{op_sparsemax}()} \cr
 \code{\link{op_split}()} \cr
 \code{\link{op_sqrt}()} \cr
@@ -354,6 +366,8 @@ Other ops: \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
 \code{\link{op_vectorized_map}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_while_loop}()} \cr
diff --git a/man/op_image_hsv_to_rgb.Rd b/man/op_image_hsv_to_rgb.Rd
index 2842bf8e3..4bf639094 100644
--- a/man/op_image_hsv_to_rgb.Rd
+++ b/man/op_image_hsv_to_rgb.Rd
@@ -87,6 +87,7 @@ Other ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -105,6 +106,7 @@ Other ops: \cr
 \code{\link{op_associative_scan}()} \cr
 \code{\link{op_average}()} \cr
 \code{\link{op_average_pool}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_batch_normalization}()} \cr
 \code{\link{op_binary_crossentropy}()} \cr
 \code{\link{op_bincount}()} \cr
@@ -115,9 +117,11 @@ Other ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
 \code{\link{op_cast}()} \cr
 \code{\link{op_categorical_crossentropy}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_celu}()} \cr
 \code{\link{op_cholesky}()} \cr
@@ -130,6 +134,7 @@ Other ops: \cr
 \code{\link{op_convert_to_numpy}()} \cr
 \code{\link{op_convert_to_tensor}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -140,6 +145,7 @@ Other ops: \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
 \code{\link{op_custom_gradient}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_depthwise_conv}()} \cr
 \code{\link{op_det}()} \cr
 \code{\link{op_diag}()} \cr
@@ -179,10 +185,13 @@ Other ops: \cr
 \code{\link{op_glu}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
 \code{\link{op_hard_shrink}()} \cr
 \code{\link{op_hard_sigmoid}()} \cr
 \code{\link{op_hard_silu}()} \cr
 \code{\link{op_hard_tanh}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -208,6 +217,8 @@ Other ops: \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
 \code{\link{op_istft}()} \cr
+\code{\link{op_kaiser}()} \cr
+\code{\link{op_layer_normalization}()} \cr
 \code{\link{op_leaky_relu}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
@@ -307,6 +318,7 @@ Other ops: \cr
 \code{\link{op_sort}()} \cr
 \code{\link{op_sparse_categorical_crossentropy}()} \cr
 \code{\link{op_sparse_plus}()} \cr
+\code{\link{op_sparse_sigmoid}()} \cr
 \code{\link{op_sparsemax}()} \cr
 \code{\link{op_split}()} \cr
 \code{\link{op_sqrt}()} \cr
@@ -344,6 +356,8 @@ Other ops: \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
 \code{\link{op_vectorized_map}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_while_loop}()} \cr
diff --git a/man/op_image_map_coordinates.Rd b/man/op_image_map_coordinates.Rd
index 4ccdf619d..1daee38e2 100644
--- a/man/op_image_map_coordinates.Rd
+++ b/man/op_image_map_coordinates.Rd
@@ -86,6 +86,7 @@ Other ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -104,6 +105,7 @@ Other ops: \cr
 \code{\link{op_associative_scan}()} \cr
 \code{\link{op_average}()} \cr
 \code{\link{op_average_pool}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_batch_normalization}()} \cr
 \code{\link{op_binary_crossentropy}()} \cr
 \code{\link{op_bincount}()} \cr
@@ -114,9 +116,11 @@ Other ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
 \code{\link{op_cast}()} \cr
 \code{\link{op_categorical_crossentropy}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_celu}()} \cr
 \code{\link{op_cholesky}()} \cr
@@ -129,6 +133,7 @@ Other ops: \cr
 \code{\link{op_convert_to_numpy}()} \cr
 \code{\link{op_convert_to_tensor}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -139,6 +144,7 @@ Other ops: \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
 \code{\link{op_custom_gradient}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_depthwise_conv}()} \cr
 \code{\link{op_det}()} \cr
 \code{\link{op_diag}()} \cr
@@ -178,10 +184,13 @@ Other ops: \cr
 \code{\link{op_glu}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
 \code{\link{op_hard_shrink}()} \cr
 \code{\link{op_hard_sigmoid}()} \cr
 \code{\link{op_hard_silu}()} \cr
 \code{\link{op_hard_tanh}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -207,6 +216,8 @@ Other ops: \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
 \code{\link{op_istft}()} \cr
+\code{\link{op_kaiser}()} \cr
+\code{\link{op_layer_normalization}()} \cr
 \code{\link{op_leaky_relu}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
@@ -306,6 +317,7 @@ Other ops: \cr
 \code{\link{op_sort}()} \cr
 \code{\link{op_sparse_categorical_crossentropy}()} \cr
 \code{\link{op_sparse_plus}()} \cr
+\code{\link{op_sparse_sigmoid}()} \cr
 \code{\link{op_sparsemax}()} \cr
 \code{\link{op_split}()} \cr
 \code{\link{op_sqrt}()} \cr
@@ -343,6 +355,8 @@ Other ops: \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
 \code{\link{op_vectorized_map}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_while_loop}()} \cr
diff --git a/man/op_image_pad.Rd b/man/op_image_pad.Rd
index 0d11666df..06c50d56f 100644
--- a/man/op_image_pad.Rd
+++ b/man/op_image_pad.Rd
@@ -102,6 +102,7 @@ Other ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -120,6 +121,7 @@ Other ops: \cr
 \code{\link{op_associative_scan}()} \cr
 \code{\link{op_average}()} \cr
 \code{\link{op_average_pool}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_batch_normalization}()} \cr
 \code{\link{op_binary_crossentropy}()} \cr
 \code{\link{op_bincount}()} \cr
@@ -130,9 +132,11 @@ Other ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
 \code{\link{op_cast}()} \cr
 \code{\link{op_categorical_crossentropy}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_celu}()} \cr
 \code{\link{op_cholesky}()} \cr
@@ -145,6 +149,7 @@ Other ops: \cr
 \code{\link{op_convert_to_numpy}()} \cr
 \code{\link{op_convert_to_tensor}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -155,6 +160,7 @@ Other ops: \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
 \code{\link{op_custom_gradient}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_depthwise_conv}()} \cr
 \code{\link{op_det}()} \cr
 \code{\link{op_diag}()} \cr
@@ -194,10 +200,13 @@ Other ops: \cr
 \code{\link{op_glu}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
 \code{\link{op_hard_shrink}()} \cr
 \code{\link{op_hard_sigmoid}()} \cr
 \code{\link{op_hard_silu}()} \cr
 \code{\link{op_hard_tanh}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -223,6 +232,8 @@ Other ops: \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
 \code{\link{op_istft}()} \cr
+\code{\link{op_kaiser}()} \cr
+\code{\link{op_layer_normalization}()} \cr
 \code{\link{op_leaky_relu}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
@@ -322,6 +333,7 @@ Other ops: \cr
 \code{\link{op_sort}()} \cr
 \code{\link{op_sparse_categorical_crossentropy}()} \cr
 \code{\link{op_sparse_plus}()} \cr
+\code{\link{op_sparse_sigmoid}()} \cr
 \code{\link{op_sparsemax}()} \cr
 \code{\link{op_split}()} \cr
 \code{\link{op_sqrt}()} \cr
@@ -359,6 +371,8 @@ Other ops: \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
 \code{\link{op_vectorized_map}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_while_loop}()} \cr
diff --git a/man/op_image_perspective_transform.Rd b/man/op_image_perspective_transform.Rd
index cd25e08ad..10b5d362a 100644
--- a/man/op_image_perspective_transform.Rd
+++ b/man/op_image_perspective_transform.Rd
@@ -131,6 +131,7 @@ Other ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -149,6 +150,7 @@ Other ops: \cr
 \code{\link{op_associative_scan}()} \cr
 \code{\link{op_average}()} \cr
 \code{\link{op_average_pool}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_batch_normalization}()} \cr
 \code{\link{op_binary_crossentropy}()} \cr
 \code{\link{op_bincount}()} \cr
@@ -159,9 +161,11 @@ Other ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
 \code{\link{op_cast}()} \cr
 \code{\link{op_categorical_crossentropy}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_celu}()} \cr
 \code{\link{op_cholesky}()} \cr
@@ -174,6 +178,7 @@ Other ops: \cr
 \code{\link{op_convert_to_numpy}()} \cr
 \code{\link{op_convert_to_tensor}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -184,6 +189,7 @@ Other ops: \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
 \code{\link{op_custom_gradient}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_depthwise_conv}()} \cr
 \code{\link{op_det}()} \cr
 \code{\link{op_diag}()} \cr
@@ -223,10 +229,13 @@ Other ops: \cr
 \code{\link{op_glu}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
 \code{\link{op_hard_shrink}()} \cr
 \code{\link{op_hard_sigmoid}()} \cr
 \code{\link{op_hard_silu}()} \cr
 \code{\link{op_hard_tanh}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -252,6 +261,8 @@ Other ops: \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
 \code{\link{op_istft}()} \cr
+\code{\link{op_kaiser}()} \cr
+\code{\link{op_layer_normalization}()} \cr
 \code{\link{op_leaky_relu}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
@@ -351,6 +362,7 @@ Other ops: \cr
 \code{\link{op_sort}()} \cr
 \code{\link{op_sparse_categorical_crossentropy}()} \cr
 \code{\link{op_sparse_plus}()} \cr
+\code{\link{op_sparse_sigmoid}()} \cr
 \code{\link{op_sparsemax}()} \cr
 \code{\link{op_split}()} \cr
 \code{\link{op_sqrt}()} \cr
@@ -388,6 +400,8 @@ Other ops: \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
 \code{\link{op_vectorized_map}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_while_loop}()} \cr
diff --git a/man/op_image_resize.Rd b/man/op_image_resize.Rd
index 1d59a3be2..768580cc8 100644
--- a/man/op_image_resize.Rd
+++ b/man/op_image_resize.Rd
@@ -128,6 +128,7 @@ Other ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -146,6 +147,7 @@ Other ops: \cr
 \code{\link{op_associative_scan}()} \cr
 \code{\link{op_average}()} \cr
 \code{\link{op_average_pool}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_batch_normalization}()} \cr
 \code{\link{op_binary_crossentropy}()} \cr
 \code{\link{op_bincount}()} \cr
@@ -156,9 +158,11 @@ Other ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
 \code{\link{op_cast}()} \cr
 \code{\link{op_categorical_crossentropy}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_celu}()} \cr
 \code{\link{op_cholesky}()} \cr
@@ -171,6 +175,7 @@ Other ops: \cr
 \code{\link{op_convert_to_numpy}()} \cr
 \code{\link{op_convert_to_tensor}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -181,6 +186,7 @@ Other ops: \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
 \code{\link{op_custom_gradient}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_depthwise_conv}()} \cr
 \code{\link{op_det}()} \cr
 \code{\link{op_diag}()} \cr
@@ -220,10 +226,13 @@ Other ops: \cr
 \code{\link{op_glu}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
 \code{\link{op_hard_shrink}()} \cr
 \code{\link{op_hard_sigmoid}()} \cr
 \code{\link{op_hard_silu}()} \cr
 \code{\link{op_hard_tanh}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -249,6 +258,8 @@ Other ops: \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
 \code{\link{op_istft}()} \cr
+\code{\link{op_kaiser}()} \cr
+\code{\link{op_layer_normalization}()} \cr
 \code{\link{op_leaky_relu}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
@@ -348,6 +359,7 @@ Other ops: \cr
 \code{\link{op_sort}()} \cr
 \code{\link{op_sparse_categorical_crossentropy}()} \cr
 \code{\link{op_sparse_plus}()} \cr
+\code{\link{op_sparse_sigmoid}()} \cr
 \code{\link{op_sparsemax}()} \cr
 \code{\link{op_split}()} \cr
 \code{\link{op_sqrt}()} \cr
@@ -385,6 +397,8 @@ Other ops: \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
 \code{\link{op_vectorized_map}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_while_loop}()} \cr
diff --git a/man/op_image_rgb_to_grayscale.Rd b/man/op_image_rgb_to_grayscale.Rd
index c28f3c5a2..55b43c158 100644
--- a/man/op_image_rgb_to_grayscale.Rd
+++ b/man/op_image_rgb_to_grayscale.Rd
@@ -87,6 +87,7 @@ Other ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -105,6 +106,7 @@ Other ops: \cr
 \code{\link{op_associative_scan}()} \cr
 \code{\link{op_average}()} \cr
 \code{\link{op_average_pool}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_batch_normalization}()} \cr
 \code{\link{op_binary_crossentropy}()} \cr
 \code{\link{op_bincount}()} \cr
@@ -115,9 +117,11 @@ Other ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
 \code{\link{op_cast}()} \cr
 \code{\link{op_categorical_crossentropy}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_celu}()} \cr
 \code{\link{op_cholesky}()} \cr
@@ -130,6 +134,7 @@ Other ops: \cr
 \code{\link{op_convert_to_numpy}()} \cr
 \code{\link{op_convert_to_tensor}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -140,6 +145,7 @@ Other ops: \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
 \code{\link{op_custom_gradient}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_depthwise_conv}()} \cr
 \code{\link{op_det}()} \cr
 \code{\link{op_diag}()} \cr
@@ -179,10 +185,13 @@ Other ops: \cr
 \code{\link{op_glu}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
 \code{\link{op_hard_shrink}()} \cr
 \code{\link{op_hard_sigmoid}()} \cr
 \code{\link{op_hard_silu}()} \cr
 \code{\link{op_hard_tanh}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -208,6 +217,8 @@ Other ops: \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
 \code{\link{op_istft}()} \cr
+\code{\link{op_kaiser}()} \cr
+\code{\link{op_layer_normalization}()} \cr
 \code{\link{op_leaky_relu}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
@@ -307,6 +318,7 @@ Other ops: \cr
 \code{\link{op_sort}()} \cr
 \code{\link{op_sparse_categorical_crossentropy}()} \cr
 \code{\link{op_sparse_plus}()} \cr
+\code{\link{op_sparse_sigmoid}()} \cr
 \code{\link{op_sparsemax}()} \cr
 \code{\link{op_split}()} \cr
 \code{\link{op_sqrt}()} \cr
@@ -344,6 +356,8 @@ Other ops: \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
 \code{\link{op_vectorized_map}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_while_loop}()} \cr
diff --git a/man/op_image_rgb_to_hsv.Rd b/man/op_image_rgb_to_hsv.Rd
index 9ce30134d..824ec9860 100644
--- a/man/op_image_rgb_to_hsv.Rd
+++ b/man/op_image_rgb_to_hsv.Rd
@@ -90,6 +90,7 @@ Other ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -108,6 +109,7 @@ Other ops: \cr
 \code{\link{op_associative_scan}()} \cr
 \code{\link{op_average}()} \cr
 \code{\link{op_average_pool}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_batch_normalization}()} \cr
 \code{\link{op_binary_crossentropy}()} \cr
 \code{\link{op_bincount}()} \cr
@@ -118,9 +120,11 @@ Other ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
 \code{\link{op_cast}()} \cr
 \code{\link{op_categorical_crossentropy}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_celu}()} \cr
 \code{\link{op_cholesky}()} \cr
@@ -133,6 +137,7 @@ Other ops: \cr
 \code{\link{op_convert_to_numpy}()} \cr
 \code{\link{op_convert_to_tensor}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -143,6 +148,7 @@ Other ops: \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
 \code{\link{op_custom_gradient}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_depthwise_conv}()} \cr
 \code{\link{op_det}()} \cr
 \code{\link{op_diag}()} \cr
@@ -182,10 +188,13 @@ Other ops: \cr
 \code{\link{op_glu}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
 \code{\link{op_hard_shrink}()} \cr
 \code{\link{op_hard_sigmoid}()} \cr
 \code{\link{op_hard_silu}()} \cr
 \code{\link{op_hard_tanh}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -211,6 +220,8 @@ Other ops: \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
 \code{\link{op_istft}()} \cr
+\code{\link{op_kaiser}()} \cr
+\code{\link{op_layer_normalization}()} \cr
 \code{\link{op_leaky_relu}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
@@ -310,6 +321,7 @@ Other ops: \cr
 \code{\link{op_sort}()} \cr
 \code{\link{op_sparse_categorical_crossentropy}()} \cr
 \code{\link{op_sparse_plus}()} \cr
+\code{\link{op_sparse_sigmoid}()} \cr
 \code{\link{op_sparsemax}()} \cr
 \code{\link{op_split}()} \cr
 \code{\link{op_sqrt}()} \cr
@@ -347,6 +359,8 @@ Other ops: \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
 \code{\link{op_vectorized_map}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_while_loop}()} \cr
diff --git a/man/op_in_top_k.Rd b/man/op_in_top_k.Rd
index 2106743b6..0e71a75f6 100644
--- a/man/op_in_top_k.Rd
+++ b/man/op_in_top_k.Rd
@@ -63,6 +63,7 @@ Other ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -81,6 +82,7 @@ Other ops: \cr
 \code{\link{op_associative_scan}()} \cr
 \code{\link{op_average}()} \cr
 \code{\link{op_average_pool}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_batch_normalization}()} \cr
 \code{\link{op_binary_crossentropy}()} \cr
 \code{\link{op_bincount}()} \cr
@@ -91,9 +93,11 @@ Other ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
 \code{\link{op_cast}()} \cr
 \code{\link{op_categorical_crossentropy}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_celu}()} \cr
 \code{\link{op_cholesky}()} \cr
@@ -106,6 +110,7 @@ Other ops: \cr
 \code{\link{op_convert_to_numpy}()} \cr
 \code{\link{op_convert_to_tensor}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -116,6 +121,7 @@ Other ops: \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
 \code{\link{op_custom_gradient}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_depthwise_conv}()} \cr
 \code{\link{op_det}()} \cr
 \code{\link{op_diag}()} \cr
@@ -155,10 +161,13 @@ Other ops: \cr
 \code{\link{op_glu}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
 \code{\link{op_hard_shrink}()} \cr
 \code{\link{op_hard_sigmoid}()} \cr
 \code{\link{op_hard_silu}()} \cr
 \code{\link{op_hard_tanh}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -184,6 +193,8 @@ Other ops: \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
 \code{\link{op_istft}()} \cr
+\code{\link{op_kaiser}()} \cr
+\code{\link{op_layer_normalization}()} \cr
 \code{\link{op_leaky_relu}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
@@ -283,6 +294,7 @@ Other ops: \cr
 \code{\link{op_sort}()} \cr
 \code{\link{op_sparse_categorical_crossentropy}()} \cr
 \code{\link{op_sparse_plus}()} \cr
+\code{\link{op_sparse_sigmoid}()} \cr
 \code{\link{op_sparsemax}()} \cr
 \code{\link{op_split}()} \cr
 \code{\link{op_sqrt}()} \cr
@@ -320,6 +332,8 @@ Other ops: \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
 \code{\link{op_vectorized_map}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_while_loop}()} \cr
diff --git a/man/op_inner.Rd b/man/op_inner.Rd
index e4dbc3381..2955d548c 100644
--- a/man/op_inner.Rd
+++ b/man/op_inner.Rd
@@ -31,6 +31,7 @@ Other numpy ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -47,6 +48,7 @@ Other numpy ops: \cr
 \code{\link{op_argsort}()} \cr
 \code{\link{op_array}()} \cr
 \code{\link{op_average}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_bincount}()} \cr
 \code{\link{op_bitwise_and}()} \cr
 \code{\link{op_bitwise_invert}()} \cr
@@ -55,12 +57,15 @@ Other numpy ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_clip}()} \cr
 \code{\link{op_concatenate}()} \cr
 \code{\link{op_conj}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -69,6 +74,7 @@ Other numpy ops: \cr
 \code{\link{op_ctc_decode}()} \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_diag}()} \cr
 \code{\link{op_diagflat}()} \cr
 \code{\link{op_diagonal}()} \cr
@@ -93,6 +99,9 @@ Other numpy ops: \cr
 \code{\link{op_get_item}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -101,6 +110,7 @@ Other numpy ops: \cr
 \code{\link{op_isfinite}()} \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
+\code{\link{op_kaiser}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
 \code{\link{op_less_equal}()} \cr
@@ -181,6 +191,8 @@ Other numpy ops: \cr
 \code{\link{op_var}()} \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_zeros}()} \cr
@@ -190,6 +202,7 @@ Other ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -208,6 +221,7 @@ Other ops: \cr
 \code{\link{op_associative_scan}()} \cr
 \code{\link{op_average}()} \cr
 \code{\link{op_average_pool}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_batch_normalization}()} \cr
 \code{\link{op_binary_crossentropy}()} \cr
 \code{\link{op_bincount}()} \cr
@@ -218,9 +232,11 @@ Other ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
 \code{\link{op_cast}()} \cr
 \code{\link{op_categorical_crossentropy}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_celu}()} \cr
 \code{\link{op_cholesky}()} \cr
@@ -233,6 +249,7 @@ Other ops: \cr
 \code{\link{op_convert_to_numpy}()} \cr
 \code{\link{op_convert_to_tensor}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -243,6 +260,7 @@ Other ops: \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
 \code{\link{op_custom_gradient}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_depthwise_conv}()} \cr
 \code{\link{op_det}()} \cr
 \code{\link{op_diag}()} \cr
@@ -282,10 +300,13 @@ Other ops: \cr
 \code{\link{op_glu}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
 \code{\link{op_hard_shrink}()} \cr
 \code{\link{op_hard_sigmoid}()} \cr
 \code{\link{op_hard_silu}()} \cr
 \code{\link{op_hard_tanh}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -311,6 +332,8 @@ Other ops: \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
 \code{\link{op_istft}()} \cr
+\code{\link{op_kaiser}()} \cr
+\code{\link{op_layer_normalization}()} \cr
 \code{\link{op_leaky_relu}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
@@ -410,6 +433,7 @@ Other ops: \cr
 \code{\link{op_sort}()} \cr
 \code{\link{op_sparse_categorical_crossentropy}()} \cr
 \code{\link{op_sparse_plus}()} \cr
+\code{\link{op_sparse_sigmoid}()} \cr
 \code{\link{op_sparsemax}()} \cr
 \code{\link{op_split}()} \cr
 \code{\link{op_sqrt}()} \cr
@@ -447,6 +471,8 @@ Other ops: \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
 \code{\link{op_vectorized_map}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_while_loop}()} \cr
diff --git a/man/op_inv.Rd b/man/op_inv.Rd
index d7eb7fc71..8f7473fe0 100644
--- a/man/op_inv.Rd
+++ b/man/op_inv.Rd
@@ -32,6 +32,7 @@ Other ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -50,6 +51,7 @@ Other ops: \cr
 \code{\link{op_associative_scan}()} \cr
 \code{\link{op_average}()} \cr
 \code{\link{op_average_pool}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_batch_normalization}()} \cr
 \code{\link{op_binary_crossentropy}()} \cr
 \code{\link{op_bincount}()} \cr
@@ -60,9 +62,11 @@ Other ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
 \code{\link{op_cast}()} \cr
 \code{\link{op_categorical_crossentropy}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_celu}()} \cr
 \code{\link{op_cholesky}()} \cr
@@ -75,6 +79,7 @@ Other ops: \cr
 \code{\link{op_convert_to_numpy}()} \cr
 \code{\link{op_convert_to_tensor}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -85,6 +90,7 @@ Other ops: \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
 \code{\link{op_custom_gradient}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_depthwise_conv}()} \cr
 \code{\link{op_det}()} \cr
 \code{\link{op_diag}()} \cr
@@ -124,10 +130,13 @@ Other ops: \cr
 \code{\link{op_glu}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
 \code{\link{op_hard_shrink}()} \cr
 \code{\link{op_hard_sigmoid}()} \cr
 \code{\link{op_hard_silu}()} \cr
 \code{\link{op_hard_tanh}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -153,6 +162,8 @@ Other ops: \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
 \code{\link{op_istft}()} \cr
+\code{\link{op_kaiser}()} \cr
+\code{\link{op_layer_normalization}()} \cr
 \code{\link{op_leaky_relu}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
@@ -252,6 +263,7 @@ Other ops: \cr
 \code{\link{op_sort}()} \cr
 \code{\link{op_sparse_categorical_crossentropy}()} \cr
 \code{\link{op_sparse_plus}()} \cr
+\code{\link{op_sparse_sigmoid}()} \cr
 \code{\link{op_sparsemax}()} \cr
 \code{\link{op_split}()} \cr
 \code{\link{op_sqrt}()} \cr
@@ -289,6 +301,8 @@ Other ops: \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
 \code{\link{op_vectorized_map}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_while_loop}()} \cr
diff --git a/man/op_irfft.Rd b/man/op_irfft.Rd
index 2892cc386..ba41b0243 100644
--- a/man/op_irfft.Rd
+++ b/man/op_irfft.Rd
@@ -81,6 +81,7 @@ Other ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -99,6 +100,7 @@ Other ops: \cr
 \code{\link{op_associative_scan}()} \cr
 \code{\link{op_average}()} \cr
 \code{\link{op_average_pool}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_batch_normalization}()} \cr
 \code{\link{op_binary_crossentropy}()} \cr
 \code{\link{op_bincount}()} \cr
@@ -109,9 +111,11 @@ Other ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
 \code{\link{op_cast}()} \cr
 \code{\link{op_categorical_crossentropy}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_celu}()} \cr
 \code{\link{op_cholesky}()} \cr
@@ -124,6 +128,7 @@ Other ops: \cr
 \code{\link{op_convert_to_numpy}()} \cr
 \code{\link{op_convert_to_tensor}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -134,6 +139,7 @@ Other ops: \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
 \code{\link{op_custom_gradient}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_depthwise_conv}()} \cr
 \code{\link{op_det}()} \cr
 \code{\link{op_diag}()} \cr
@@ -173,10 +179,13 @@ Other ops: \cr
 \code{\link{op_glu}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
 \code{\link{op_hard_shrink}()} \cr
 \code{\link{op_hard_sigmoid}()} \cr
 \code{\link{op_hard_silu}()} \cr
 \code{\link{op_hard_tanh}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -202,6 +211,8 @@ Other ops: \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
 \code{\link{op_istft}()} \cr
+\code{\link{op_kaiser}()} \cr
+\code{\link{op_layer_normalization}()} \cr
 \code{\link{op_leaky_relu}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
@@ -301,6 +312,7 @@ Other ops: \cr
 \code{\link{op_sort}()} \cr
 \code{\link{op_sparse_categorical_crossentropy}()} \cr
 \code{\link{op_sparse_plus}()} \cr
+\code{\link{op_sparse_sigmoid}()} \cr
 \code{\link{op_sparsemax}()} \cr
 \code{\link{op_split}()} \cr
 \code{\link{op_sqrt}()} \cr
@@ -338,6 +350,8 @@ Other ops: \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
 \code{\link{op_vectorized_map}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_while_loop}()} \cr
diff --git a/man/op_is_tensor.Rd b/man/op_is_tensor.Rd
index 9edd31488..e1e5ae058 100644
--- a/man/op_is_tensor.Rd
+++ b/man/op_is_tensor.Rd
@@ -50,6 +50,7 @@ Other ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -68,6 +69,7 @@ Other ops: \cr
 \code{\link{op_associative_scan}()} \cr
 \code{\link{op_average}()} \cr
 \code{\link{op_average_pool}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_batch_normalization}()} \cr
 \code{\link{op_binary_crossentropy}()} \cr
 \code{\link{op_bincount}()} \cr
@@ -78,9 +80,11 @@ Other ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
 \code{\link{op_cast}()} \cr
 \code{\link{op_categorical_crossentropy}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_celu}()} \cr
 \code{\link{op_cholesky}()} \cr
@@ -93,6 +97,7 @@ Other ops: \cr
 \code{\link{op_convert_to_numpy}()} \cr
 \code{\link{op_convert_to_tensor}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -103,6 +108,7 @@ Other ops: \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
 \code{\link{op_custom_gradient}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_depthwise_conv}()} \cr
 \code{\link{op_det}()} \cr
 \code{\link{op_diag}()} \cr
@@ -142,10 +148,13 @@ Other ops: \cr
 \code{\link{op_glu}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
 \code{\link{op_hard_shrink}()} \cr
 \code{\link{op_hard_sigmoid}()} \cr
 \code{\link{op_hard_silu}()} \cr
 \code{\link{op_hard_tanh}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -171,6 +180,8 @@ Other ops: \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
 \code{\link{op_istft}()} \cr
+\code{\link{op_kaiser}()} \cr
+\code{\link{op_layer_normalization}()} \cr
 \code{\link{op_leaky_relu}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
@@ -270,6 +281,7 @@ Other ops: \cr
 \code{\link{op_sort}()} \cr
 \code{\link{op_sparse_categorical_crossentropy}()} \cr
 \code{\link{op_sparse_plus}()} \cr
+\code{\link{op_sparse_sigmoid}()} \cr
 \code{\link{op_sparsemax}()} \cr
 \code{\link{op_split}()} \cr
 \code{\link{op_sqrt}()} \cr
@@ -307,6 +319,8 @@ Other ops: \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
 \code{\link{op_vectorized_map}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_while_loop}()} \cr
diff --git a/man/op_isclose.Rd b/man/op_isclose.Rd
index 265c5e09e..19133d157 100644
--- a/man/op_isclose.Rd
+++ b/man/op_isclose.Rd
@@ -32,6 +32,7 @@ Other numpy ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -48,6 +49,7 @@ Other numpy ops: \cr
 \code{\link{op_argsort}()} \cr
 \code{\link{op_array}()} \cr
 \code{\link{op_average}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_bincount}()} \cr
 \code{\link{op_bitwise_and}()} \cr
 \code{\link{op_bitwise_invert}()} \cr
@@ -56,12 +58,15 @@ Other numpy ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_clip}()} \cr
 \code{\link{op_concatenate}()} \cr
 \code{\link{op_conj}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -70,6 +75,7 @@ Other numpy ops: \cr
 \code{\link{op_ctc_decode}()} \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_diag}()} \cr
 \code{\link{op_diagflat}()} \cr
 \code{\link{op_diagonal}()} \cr
@@ -94,6 +100,9 @@ Other numpy ops: \cr
 \code{\link{op_get_item}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -102,6 +111,7 @@ Other numpy ops: \cr
 \code{\link{op_isfinite}()} \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
+\code{\link{op_kaiser}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
 \code{\link{op_less_equal}()} \cr
@@ -182,6 +192,8 @@ Other numpy ops: \cr
 \code{\link{op_var}()} \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_zeros}()} \cr
@@ -191,6 +203,7 @@ Other ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -209,6 +222,7 @@ Other ops: \cr
 \code{\link{op_associative_scan}()} \cr
 \code{\link{op_average}()} \cr
 \code{\link{op_average_pool}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_batch_normalization}()} \cr
 \code{\link{op_binary_crossentropy}()} \cr
 \code{\link{op_bincount}()} \cr
@@ -219,9 +233,11 @@ Other ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
 \code{\link{op_cast}()} \cr
 \code{\link{op_categorical_crossentropy}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_celu}()} \cr
 \code{\link{op_cholesky}()} \cr
@@ -234,6 +250,7 @@ Other ops: \cr
 \code{\link{op_convert_to_numpy}()} \cr
 \code{\link{op_convert_to_tensor}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -244,6 +261,7 @@ Other ops: \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
 \code{\link{op_custom_gradient}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_depthwise_conv}()} \cr
 \code{\link{op_det}()} \cr
 \code{\link{op_diag}()} \cr
@@ -283,10 +301,13 @@ Other ops: \cr
 \code{\link{op_glu}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
 \code{\link{op_hard_shrink}()} \cr
 \code{\link{op_hard_sigmoid}()} \cr
 \code{\link{op_hard_silu}()} \cr
 \code{\link{op_hard_tanh}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -312,6 +333,8 @@ Other ops: \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
 \code{\link{op_istft}()} \cr
+\code{\link{op_kaiser}()} \cr
+\code{\link{op_layer_normalization}()} \cr
 \code{\link{op_leaky_relu}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
@@ -411,6 +434,7 @@ Other ops: \cr
 \code{\link{op_sort}()} \cr
 \code{\link{op_sparse_categorical_crossentropy}()} \cr
 \code{\link{op_sparse_plus}()} \cr
+\code{\link{op_sparse_sigmoid}()} \cr
 \code{\link{op_sparsemax}()} \cr
 \code{\link{op_split}()} \cr
 \code{\link{op_sqrt}()} \cr
@@ -448,6 +472,8 @@ Other ops: \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
 \code{\link{op_vectorized_map}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_while_loop}()} \cr
diff --git a/man/op_isfinite.Rd b/man/op_isfinite.Rd
index a02dce81f..5ce039cce 100644
--- a/man/op_isfinite.Rd
+++ b/man/op_isfinite.Rd
@@ -26,6 +26,7 @@ Other numpy ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -42,6 +43,7 @@ Other numpy ops: \cr
 \code{\link{op_argsort}()} \cr
 \code{\link{op_array}()} \cr
 \code{\link{op_average}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_bincount}()} \cr
 \code{\link{op_bitwise_and}()} \cr
 \code{\link{op_bitwise_invert}()} \cr
@@ -50,12 +52,15 @@ Other numpy ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_clip}()} \cr
 \code{\link{op_concatenate}()} \cr
 \code{\link{op_conj}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -64,6 +69,7 @@ Other numpy ops: \cr
 \code{\link{op_ctc_decode}()} \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_diag}()} \cr
 \code{\link{op_diagflat}()} \cr
 \code{\link{op_diagonal}()} \cr
@@ -88,6 +94,9 @@ Other numpy ops: \cr
 \code{\link{op_get_item}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -96,6 +105,7 @@ Other numpy ops: \cr
 \code{\link{op_isclose}()} \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
+\code{\link{op_kaiser}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
 \code{\link{op_less_equal}()} \cr
@@ -176,6 +186,8 @@ Other numpy ops: \cr
 \code{\link{op_var}()} \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_zeros}()} \cr
@@ -185,6 +197,7 @@ Other ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -203,6 +216,7 @@ Other ops: \cr
 \code{\link{op_associative_scan}()} \cr
 \code{\link{op_average}()} \cr
 \code{\link{op_average_pool}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_batch_normalization}()} \cr
 \code{\link{op_binary_crossentropy}()} \cr
 \code{\link{op_bincount}()} \cr
@@ -213,9 +227,11 @@ Other ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
 \code{\link{op_cast}()} \cr
 \code{\link{op_categorical_crossentropy}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_celu}()} \cr
 \code{\link{op_cholesky}()} \cr
@@ -228,6 +244,7 @@ Other ops: \cr
 \code{\link{op_convert_to_numpy}()} \cr
 \code{\link{op_convert_to_tensor}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -238,6 +255,7 @@ Other ops: \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
 \code{\link{op_custom_gradient}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_depthwise_conv}()} \cr
 \code{\link{op_det}()} \cr
 \code{\link{op_diag}()} \cr
@@ -277,10 +295,13 @@ Other ops: \cr
 \code{\link{op_glu}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
 \code{\link{op_hard_shrink}()} \cr
 \code{\link{op_hard_sigmoid}()} \cr
 \code{\link{op_hard_silu}()} \cr
 \code{\link{op_hard_tanh}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -306,6 +327,8 @@ Other ops: \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
 \code{\link{op_istft}()} \cr
+\code{\link{op_kaiser}()} \cr
+\code{\link{op_layer_normalization}()} \cr
 \code{\link{op_leaky_relu}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
@@ -405,6 +428,7 @@ Other ops: \cr
 \code{\link{op_sort}()} \cr
 \code{\link{op_sparse_categorical_crossentropy}()} \cr
 \code{\link{op_sparse_plus}()} \cr
+\code{\link{op_sparse_sigmoid}()} \cr
 \code{\link{op_sparsemax}()} \cr
 \code{\link{op_split}()} \cr
 \code{\link{op_sqrt}()} \cr
@@ -442,6 +466,8 @@ Other ops: \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
 \code{\link{op_vectorized_map}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_while_loop}()} \cr
diff --git a/man/op_isinf.Rd b/man/op_isinf.Rd
index 4c95f2855..020aa02ee 100644
--- a/man/op_isinf.Rd
+++ b/man/op_isinf.Rd
@@ -24,6 +24,7 @@ Other numpy ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -40,6 +41,7 @@ Other numpy ops: \cr
 \code{\link{op_argsort}()} \cr
 \code{\link{op_array}()} \cr
 \code{\link{op_average}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_bincount}()} \cr
 \code{\link{op_bitwise_and}()} \cr
 \code{\link{op_bitwise_invert}()} \cr
@@ -48,12 +50,15 @@ Other numpy ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_clip}()} \cr
 \code{\link{op_concatenate}()} \cr
 \code{\link{op_conj}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -62,6 +67,7 @@ Other numpy ops: \cr
 \code{\link{op_ctc_decode}()} \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_diag}()} \cr
 \code{\link{op_diagflat}()} \cr
 \code{\link{op_diagonal}()} \cr
@@ -86,6 +92,9 @@ Other numpy ops: \cr
 \code{\link{op_get_item}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -94,6 +103,7 @@ Other numpy ops: \cr
 \code{\link{op_isclose}()} \cr
 \code{\link{op_isfinite}()} \cr
 \code{\link{op_isnan}()} \cr
+\code{\link{op_kaiser}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
 \code{\link{op_less_equal}()} \cr
@@ -174,6 +184,8 @@ Other numpy ops: \cr
 \code{\link{op_var}()} \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_zeros}()} \cr
@@ -183,6 +195,7 @@ Other ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -201,6 +214,7 @@ Other ops: \cr
 \code{\link{op_associative_scan}()} \cr
 \code{\link{op_average}()} \cr
 \code{\link{op_average_pool}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_batch_normalization}()} \cr
 \code{\link{op_binary_crossentropy}()} \cr
 \code{\link{op_bincount}()} \cr
@@ -211,9 +225,11 @@ Other ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
 \code{\link{op_cast}()} \cr
 \code{\link{op_categorical_crossentropy}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_celu}()} \cr
 \code{\link{op_cholesky}()} \cr
@@ -226,6 +242,7 @@ Other ops: \cr
 \code{\link{op_convert_to_numpy}()} \cr
 \code{\link{op_convert_to_tensor}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -236,6 +253,7 @@ Other ops: \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
 \code{\link{op_custom_gradient}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_depthwise_conv}()} \cr
 \code{\link{op_det}()} \cr
 \code{\link{op_diag}()} \cr
@@ -275,10 +293,13 @@ Other ops: \cr
 \code{\link{op_glu}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
 \code{\link{op_hard_shrink}()} \cr
 \code{\link{op_hard_sigmoid}()} \cr
 \code{\link{op_hard_silu}()} \cr
 \code{\link{op_hard_tanh}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -304,6 +325,8 @@ Other ops: \cr
 \code{\link{op_isfinite}()} \cr
 \code{\link{op_isnan}()} \cr
 \code{\link{op_istft}()} \cr
+\code{\link{op_kaiser}()} \cr
+\code{\link{op_layer_normalization}()} \cr
 \code{\link{op_leaky_relu}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
@@ -403,6 +426,7 @@ Other ops: \cr
 \code{\link{op_sort}()} \cr
 \code{\link{op_sparse_categorical_crossentropy}()} \cr
 \code{\link{op_sparse_plus}()} \cr
+\code{\link{op_sparse_sigmoid}()} \cr
 \code{\link{op_sparsemax}()} \cr
 \code{\link{op_split}()} \cr
 \code{\link{op_sqrt}()} \cr
@@ -440,6 +464,8 @@ Other ops: \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
 \code{\link{op_vectorized_map}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_while_loop}()} \cr
diff --git a/man/op_isnan.Rd b/man/op_isnan.Rd
index 13eddaaf9..2fd64bb20 100644
--- a/man/op_isnan.Rd
+++ b/man/op_isnan.Rd
@@ -24,6 +24,7 @@ Other numpy ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -40,6 +41,7 @@ Other numpy ops: \cr
 \code{\link{op_argsort}()} \cr
 \code{\link{op_array}()} \cr
 \code{\link{op_average}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_bincount}()} \cr
 \code{\link{op_bitwise_and}()} \cr
 \code{\link{op_bitwise_invert}()} \cr
@@ -48,12 +50,15 @@ Other numpy ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_clip}()} \cr
 \code{\link{op_concatenate}()} \cr
 \code{\link{op_conj}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -62,6 +67,7 @@ Other numpy ops: \cr
 \code{\link{op_ctc_decode}()} \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_diag}()} \cr
 \code{\link{op_diagflat}()} \cr
 \code{\link{op_diagonal}()} \cr
@@ -86,6 +92,9 @@ Other numpy ops: \cr
 \code{\link{op_get_item}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -94,6 +103,7 @@ Other numpy ops: \cr
 \code{\link{op_isclose}()} \cr
 \code{\link{op_isfinite}()} \cr
 \code{\link{op_isinf}()} \cr
+\code{\link{op_kaiser}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
 \code{\link{op_less_equal}()} \cr
@@ -174,6 +184,8 @@ Other numpy ops: \cr
 \code{\link{op_var}()} \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_zeros}()} \cr
@@ -183,6 +195,7 @@ Other ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -201,6 +214,7 @@ Other ops: \cr
 \code{\link{op_associative_scan}()} \cr
 \code{\link{op_average}()} \cr
 \code{\link{op_average_pool}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_batch_normalization}()} \cr
 \code{\link{op_binary_crossentropy}()} \cr
 \code{\link{op_bincount}()} \cr
@@ -211,9 +225,11 @@ Other ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
 \code{\link{op_cast}()} \cr
 \code{\link{op_categorical_crossentropy}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_celu}()} \cr
 \code{\link{op_cholesky}()} \cr
@@ -226,6 +242,7 @@ Other ops: \cr
 \code{\link{op_convert_to_numpy}()} \cr
 \code{\link{op_convert_to_tensor}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -236,6 +253,7 @@ Other ops: \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
 \code{\link{op_custom_gradient}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_depthwise_conv}()} \cr
 \code{\link{op_det}()} \cr
 \code{\link{op_diag}()} \cr
@@ -275,10 +293,13 @@ Other ops: \cr
 \code{\link{op_glu}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
 \code{\link{op_hard_shrink}()} \cr
 \code{\link{op_hard_sigmoid}()} \cr
 \code{\link{op_hard_silu}()} \cr
 \code{\link{op_hard_tanh}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -304,6 +325,8 @@ Other ops: \cr
 \code{\link{op_isfinite}()} \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_istft}()} \cr
+\code{\link{op_kaiser}()} \cr
+\code{\link{op_layer_normalization}()} \cr
 \code{\link{op_leaky_relu}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
@@ -403,6 +426,7 @@ Other ops: \cr
 \code{\link{op_sort}()} \cr
 \code{\link{op_sparse_categorical_crossentropy}()} \cr
 \code{\link{op_sparse_plus}()} \cr
+\code{\link{op_sparse_sigmoid}()} \cr
 \code{\link{op_sparsemax}()} \cr
 \code{\link{op_split}()} \cr
 \code{\link{op_sqrt}()} \cr
@@ -440,6 +464,8 @@ Other ops: \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
 \code{\link{op_vectorized_map}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_while_loop}()} \cr
diff --git a/man/op_istft.Rd b/man/op_istft.Rd
index 69e91cc4c..6241a50df 100644
--- a/man/op_istft.Rd
+++ b/man/op_istft.Rd
@@ -87,6 +87,7 @@ Other ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -105,6 +106,7 @@ Other ops: \cr
 \code{\link{op_associative_scan}()} \cr
 \code{\link{op_average}()} \cr
 \code{\link{op_average_pool}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_batch_normalization}()} \cr
 \code{\link{op_binary_crossentropy}()} \cr
 \code{\link{op_bincount}()} \cr
@@ -115,9 +117,11 @@ Other ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
 \code{\link{op_cast}()} \cr
 \code{\link{op_categorical_crossentropy}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_celu}()} \cr
 \code{\link{op_cholesky}()} \cr
@@ -130,6 +134,7 @@ Other ops: \cr
 \code{\link{op_convert_to_numpy}()} \cr
 \code{\link{op_convert_to_tensor}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -140,6 +145,7 @@ Other ops: \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
 \code{\link{op_custom_gradient}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_depthwise_conv}()} \cr
 \code{\link{op_det}()} \cr
 \code{\link{op_diag}()} \cr
@@ -179,10 +185,13 @@ Other ops: \cr
 \code{\link{op_glu}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
 \code{\link{op_hard_shrink}()} \cr
 \code{\link{op_hard_sigmoid}()} \cr
 \code{\link{op_hard_silu}()} \cr
 \code{\link{op_hard_tanh}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -208,6 +217,8 @@ Other ops: \cr
 \code{\link{op_isfinite}()} \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
+\code{\link{op_kaiser}()} \cr
+\code{\link{op_layer_normalization}()} \cr
 \code{\link{op_leaky_relu}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
@@ -307,6 +318,7 @@ Other ops: \cr
 \code{\link{op_sort}()} \cr
 \code{\link{op_sparse_categorical_crossentropy}()} \cr
 \code{\link{op_sparse_plus}()} \cr
+\code{\link{op_sparse_sigmoid}()} \cr
 \code{\link{op_sparsemax}()} \cr
 \code{\link{op_split}()} \cr
 \code{\link{op_sqrt}()} \cr
@@ -344,6 +356,8 @@ Other ops: \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
 \code{\link{op_vectorized_map}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_while_loop}()} \cr
diff --git a/man/op_kaiser.Rd b/man/op_kaiser.Rd
new file mode 100644
index 000000000..21644ccac
--- /dev/null
+++ b/man/op_kaiser.Rd
@@ -0,0 +1,485 @@
+% Generated by roxygen2: do not edit by hand
+% Please edit documentation in R/ops-numpy.R
+\name{op_kaiser}
+\alias{op_kaiser}
+\title{Kaiser window function.}
+\usage{
+op_kaiser(x, beta)
+}
+\arguments{
+\item{x}{Length of the window. Must be a positive integer.}
+
+\item{beta}{Shape parameter for the window.}
+}
+\value{
+A 1D tensor containing the window values.
+}
+\description{
+The Kaiser window is defined as:
+\code{w[n] = I0(beta * sqrt(1 - (2 * n / (N - 1) - 1)^2)) / I0(beta)} where
+\code{I0} is the modified zeroth-order Bessel function of the first kind.
+}
+\section{Examples}{
+\if{html}{\out{<div class="sourceCode r">}}\preformatted{op_kaiser(5, beta = 14)
+}\if{html}{\out{</div>}}
+
+\if{html}{\out{<div class="sourceCode">}}\preformatted{## tf.Tensor([7.7268669e-06 1.6493215e-01 1.0000000e+00 1.6493215e-01 7.7268669e-06], shape=(5), dtype=float32)
+
+}\if{html}{\out{</div>}}
+}
+
+\seealso{
+Other numpy ops: \cr
+\code{\link{op_abs}()} \cr
+\code{\link{op_add}()} \cr
+\code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
+\code{\link{op_any}()} \cr
+\code{\link{op_append}()} \cr
+\code{\link{op_arange}()} \cr
+\code{\link{op_arccos}()} \cr
+\code{\link{op_arccosh}()} \cr
+\code{\link{op_arcsin}()} \cr
+\code{\link{op_arcsinh}()} \cr
+\code{\link{op_arctan}()} \cr
+\code{\link{op_arctan2}()} \cr
+\code{\link{op_arctanh}()} \cr
+\code{\link{op_argmax}()} \cr
+\code{\link{op_argmin}()} \cr
+\code{\link{op_argpartition}()} \cr
+\code{\link{op_argsort}()} \cr
+\code{\link{op_array}()} \cr
+\code{\link{op_average}()} \cr
+\code{\link{op_bartlett}()} \cr
+\code{\link{op_bincount}()} \cr
+\code{\link{op_bitwise_and}()} \cr
+\code{\link{op_bitwise_invert}()} \cr
+\code{\link{op_bitwise_left_shift}()} \cr
+\code{\link{op_bitwise_not}()} \cr
+\code{\link{op_bitwise_or}()} \cr
+\code{\link{op_bitwise_right_shift}()} \cr
+\code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
+\code{\link{op_broadcast_to}()} \cr
+\code{\link{op_cbrt}()} \cr
+\code{\link{op_ceil}()} \cr
+\code{\link{op_clip}()} \cr
+\code{\link{op_concatenate}()} \cr
+\code{\link{op_conj}()} \cr
+\code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
+\code{\link{op_correlate}()} \cr
+\code{\link{op_cos}()} \cr
+\code{\link{op_cosh}()} \cr
+\code{\link{op_count_nonzero}()} \cr
+\code{\link{op_cross}()} \cr
+\code{\link{op_ctc_decode}()} \cr
+\code{\link{op_cumprod}()} \cr
+\code{\link{op_cumsum}()} \cr
+\code{\link{op_deg2rad}()} \cr
+\code{\link{op_diag}()} \cr
+\code{\link{op_diagflat}()} \cr
+\code{\link{op_diagonal}()} \cr
+\code{\link{op_diff}()} \cr
+\code{\link{op_digitize}()} \cr
+\code{\link{op_divide}()} \cr
+\code{\link{op_divide_no_nan}()} \cr
+\code{\link{op_dot}()} \cr
+\code{\link{op_einsum}()} \cr
+\code{\link{op_empty}()} \cr
+\code{\link{op_equal}()} \cr
+\code{\link{op_exp}()} \cr
+\code{\link{op_exp2}()} \cr
+\code{\link{op_expand_dims}()} \cr
+\code{\link{op_expm1}()} \cr
+\code{\link{op_eye}()} \cr
+\code{\link{op_flip}()} \cr
+\code{\link{op_floor}()} \cr
+\code{\link{op_floor_divide}()} \cr
+\code{\link{op_full}()} \cr
+\code{\link{op_full_like}()} \cr
+\code{\link{op_get_item}()} \cr
+\code{\link{op_greater}()} \cr
+\code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
+\code{\link{op_heaviside}()} \cr
+\code{\link{op_histogram}()} \cr
+\code{\link{op_hstack}()} \cr
+\code{\link{op_identity}()} \cr
+\code{\link{op_imag}()} \cr
+\code{\link{op_inner}()} \cr
+\code{\link{op_isclose}()} \cr
+\code{\link{op_isfinite}()} \cr
+\code{\link{op_isinf}()} \cr
+\code{\link{op_isnan}()} \cr
+\code{\link{op_left_shift}()} \cr
+\code{\link{op_less}()} \cr
+\code{\link{op_less_equal}()} \cr
+\code{\link{op_linspace}()} \cr
+\code{\link{op_log}()} \cr
+\code{\link{op_log10}()} \cr
+\code{\link{op_log1p}()} \cr
+\code{\link{op_log2}()} \cr
+\code{\link{op_logaddexp}()} \cr
+\code{\link{op_logdet}()} \cr
+\code{\link{op_logical_and}()} \cr
+\code{\link{op_logical_not}()} \cr
+\code{\link{op_logical_or}()} \cr
+\code{\link{op_logical_xor}()} \cr
+\code{\link{op_logspace}()} \cr
+\code{\link{op_lstsq}()} \cr
+\code{\link{op_matmul}()} \cr
+\code{\link{op_max}()} \cr
+\code{\link{op_maximum}()} \cr
+\code{\link{op_mean}()} \cr
+\code{\link{op_median}()} \cr
+\code{\link{op_meshgrid}()} \cr
+\code{\link{op_min}()} \cr
+\code{\link{op_minimum}()} \cr
+\code{\link{op_mod}()} \cr
+\code{\link{op_moveaxis}()} \cr
+\code{\link{op_multiply}()} \cr
+\code{\link{op_nan_to_num}()} \cr
+\code{\link{op_ndim}()} \cr
+\code{\link{op_negative}()} \cr
+\code{\link{op_nonzero}()} \cr
+\code{\link{op_not_equal}()} \cr
+\code{\link{op_ones}()} \cr
+\code{\link{op_ones_like}()} \cr
+\code{\link{op_outer}()} \cr
+\code{\link{op_pad}()} \cr
+\code{\link{op_power}()} \cr
+\code{\link{op_prod}()} \cr
+\code{\link{op_quantile}()} \cr
+\code{\link{op_ravel}()} \cr
+\code{\link{op_real}()} \cr
+\code{\link{op_reciprocal}()} \cr
+\code{\link{op_repeat}()} \cr
+\code{\link{op_reshape}()} \cr
+\code{\link{op_right_shift}()} \cr
+\code{\link{op_roll}()} \cr
+\code{\link{op_rot90}()} \cr
+\code{\link{op_round}()} \cr
+\code{\link{op_saturate_cast}()} \cr
+\code{\link{op_select}()} \cr
+\code{\link{op_sign}()} \cr
+\code{\link{op_signbit}()} \cr
+\code{\link{op_sin}()} \cr
+\code{\link{op_sinh}()} \cr
+\code{\link{op_size}()} \cr
+\code{\link{op_sort}()} \cr
+\code{\link{op_split}()} \cr
+\code{\link{op_sqrt}()} \cr
+\code{\link{op_square}()} \cr
+\code{\link{op_squeeze}()} \cr
+\code{\link{op_stack}()} \cr
+\code{\link{op_std}()} \cr
+\code{\link{op_subtract}()} \cr
+\code{\link{op_sum}()} \cr
+\code{\link{op_swapaxes}()} \cr
+\code{\link{op_take}()} \cr
+\code{\link{op_take_along_axis}()} \cr
+\code{\link{op_tan}()} \cr
+\code{\link{op_tanh}()} \cr
+\code{\link{op_tensordot}()} \cr
+\code{\link{op_tile}()} \cr
+\code{\link{op_trace}()} \cr
+\code{\link{op_transpose}()} \cr
+\code{\link{op_tri}()} \cr
+\code{\link{op_tril}()} \cr
+\code{\link{op_triu}()} \cr
+\code{\link{op_trunc}()} \cr
+\code{\link{op_var}()} \cr
+\code{\link{op_vdot}()} \cr
+\code{\link{op_vectorize}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
+\code{\link{op_vstack}()} \cr
+\code{\link{op_where}()} \cr
+\code{\link{op_zeros}()} \cr
+\code{\link{op_zeros_like}()} \cr
+
+Other ops: \cr
+\code{\link{op_abs}()} \cr
+\code{\link{op_add}()} \cr
+\code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
+\code{\link{op_any}()} \cr
+\code{\link{op_append}()} \cr
+\code{\link{op_arange}()} \cr
+\code{\link{op_arccos}()} \cr
+\code{\link{op_arccosh}()} \cr
+\code{\link{op_arcsin}()} \cr
+\code{\link{op_arcsinh}()} \cr
+\code{\link{op_arctan}()} \cr
+\code{\link{op_arctan2}()} \cr
+\code{\link{op_arctanh}()} \cr
+\code{\link{op_argmax}()} \cr
+\code{\link{op_argmin}()} \cr
+\code{\link{op_argpartition}()} \cr
+\code{\link{op_argsort}()} \cr
+\code{\link{op_array}()} \cr
+\code{\link{op_associative_scan}()} \cr
+\code{\link{op_average}()} \cr
+\code{\link{op_average_pool}()} \cr
+\code{\link{op_bartlett}()} \cr
+\code{\link{op_batch_normalization}()} \cr
+\code{\link{op_binary_crossentropy}()} \cr
+\code{\link{op_bincount}()} \cr
+\code{\link{op_bitwise_and}()} \cr
+\code{\link{op_bitwise_invert}()} \cr
+\code{\link{op_bitwise_left_shift}()} \cr
+\code{\link{op_bitwise_not}()} \cr
+\code{\link{op_bitwise_or}()} \cr
+\code{\link{op_bitwise_right_shift}()} \cr
+\code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
+\code{\link{op_broadcast_to}()} \cr
+\code{\link{op_cast}()} \cr
+\code{\link{op_categorical_crossentropy}()} \cr
+\code{\link{op_cbrt}()} \cr
+\code{\link{op_ceil}()} \cr
+\code{\link{op_celu}()} \cr
+\code{\link{op_cholesky}()} \cr
+\code{\link{op_clip}()} \cr
+\code{\link{op_concatenate}()} \cr
+\code{\link{op_cond}()} \cr
+\code{\link{op_conj}()} \cr
+\code{\link{op_conv}()} \cr
+\code{\link{op_conv_transpose}()} \cr
+\code{\link{op_convert_to_numpy}()} \cr
+\code{\link{op_convert_to_tensor}()} \cr
+\code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
+\code{\link{op_correlate}()} \cr
+\code{\link{op_cos}()} \cr
+\code{\link{op_cosh}()} \cr
+\code{\link{op_count_nonzero}()} \cr
+\code{\link{op_cross}()} \cr
+\code{\link{op_ctc_decode}()} \cr
+\code{\link{op_ctc_loss}()} \cr
+\code{\link{op_cumprod}()} \cr
+\code{\link{op_cumsum}()} \cr
+\code{\link{op_custom_gradient}()} \cr
+\code{\link{op_deg2rad}()} \cr
+\code{\link{op_depthwise_conv}()} \cr
+\code{\link{op_det}()} \cr
+\code{\link{op_diag}()} \cr
+\code{\link{op_diagflat}()} \cr
+\code{\link{op_diagonal}()} \cr
+\code{\link{op_diff}()} \cr
+\code{\link{op_digitize}()} \cr
+\code{\link{op_divide}()} \cr
+\code{\link{op_divide_no_nan}()} \cr
+\code{\link{op_dot}()} \cr
+\code{\link{op_dot_product_attention}()} \cr
+\code{\link{op_dtype}()} \cr
+\code{\link{op_eig}()} \cr
+\code{\link{op_eigh}()} \cr
+\code{\link{op_einsum}()} \cr
+\code{\link{op_elu}()} \cr
+\code{\link{op_empty}()} \cr
+\code{\link{op_equal}()} \cr
+\code{\link{op_erf}()} \cr
+\code{\link{op_erfinv}()} \cr
+\code{\link{op_exp}()} \cr
+\code{\link{op_exp2}()} \cr
+\code{\link{op_expand_dims}()} \cr
+\code{\link{op_expm1}()} \cr
+\code{\link{op_extract_sequences}()} \cr
+\code{\link{op_eye}()} \cr
+\code{\link{op_fft}()} \cr
+\code{\link{op_fft2}()} \cr
+\code{\link{op_flip}()} \cr
+\code{\link{op_floor}()} \cr
+\code{\link{op_floor_divide}()} \cr
+\code{\link{op_fori_loop}()} \cr
+\code{\link{op_full}()} \cr
+\code{\link{op_full_like}()} \cr
+\code{\link{op_gelu}()} \cr
+\code{\link{op_get_item}()} \cr
+\code{\link{op_glu}()} \cr
+\code{\link{op_greater}()} \cr
+\code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
+\code{\link{op_hard_shrink}()} \cr
+\code{\link{op_hard_sigmoid}()} \cr
+\code{\link{op_hard_silu}()} \cr
+\code{\link{op_hard_tanh}()} \cr
+\code{\link{op_heaviside}()} \cr
+\code{\link{op_histogram}()} \cr
+\code{\link{op_hstack}()} \cr
+\code{\link{op_identity}()} \cr
+\code{\link{op_ifft2}()} \cr
+\code{\link{op_imag}()} \cr
+\code{\link{op_image_affine_transform}()} \cr
+\code{\link{op_image_crop}()} \cr
+\code{\link{op_image_extract_patches}()} \cr
+\code{\link{op_image_gaussian_blur}()} \cr
+\code{\link{op_image_hsv_to_rgb}()} \cr
+\code{\link{op_image_map_coordinates}()} \cr
+\code{\link{op_image_pad}()} \cr
+\code{\link{op_image_perspective_transform}()} \cr
+\code{\link{op_image_resize}()} \cr
+\code{\link{op_image_rgb_to_grayscale}()} \cr
+\code{\link{op_image_rgb_to_hsv}()} \cr
+\code{\link{op_in_top_k}()} \cr
+\code{\link{op_inner}()} \cr
+\code{\link{op_inv}()} \cr
+\code{\link{op_irfft}()} \cr
+\code{\link{op_is_tensor}()} \cr
+\code{\link{op_isclose}()} \cr
+\code{\link{op_isfinite}()} \cr
+\code{\link{op_isinf}()} \cr
+\code{\link{op_isnan}()} \cr
+\code{\link{op_istft}()} \cr
+\code{\link{op_layer_normalization}()} \cr
+\code{\link{op_leaky_relu}()} \cr
+\code{\link{op_left_shift}()} \cr
+\code{\link{op_less}()} \cr
+\code{\link{op_less_equal}()} \cr
+\code{\link{op_linspace}()} \cr
+\code{\link{op_log}()} \cr
+\code{\link{op_log10}()} \cr
+\code{\link{op_log1p}()} \cr
+\code{\link{op_log2}()} \cr
+\code{\link{op_log_sigmoid}()} \cr
+\code{\link{op_log_softmax}()} \cr
+\code{\link{op_logaddexp}()} \cr
+\code{\link{op_logdet}()} \cr
+\code{\link{op_logical_and}()} \cr
+\code{\link{op_logical_not}()} \cr
+\code{\link{op_logical_or}()} \cr
+\code{\link{op_logical_xor}()} \cr
+\code{\link{op_logspace}()} \cr
+\code{\link{op_logsumexp}()} \cr
+\code{\link{op_lstsq}()} \cr
+\code{\link{op_lu_factor}()} \cr
+\code{\link{op_map}()} \cr
+\code{\link{op_matmul}()} \cr
+\code{\link{op_max}()} \cr
+\code{\link{op_max_pool}()} \cr
+\code{\link{op_maximum}()} \cr
+\code{\link{op_mean}()} \cr
+\code{\link{op_median}()} \cr
+\code{\link{op_meshgrid}()} \cr
+\code{\link{op_min}()} \cr
+\code{\link{op_minimum}()} \cr
+\code{\link{op_mod}()} \cr
+\code{\link{op_moments}()} \cr
+\code{\link{op_moveaxis}()} \cr
+\code{\link{op_multi_hot}()} \cr
+\code{\link{op_multiply}()} \cr
+\code{\link{op_nan_to_num}()} \cr
+\code{\link{op_ndim}()} \cr
+\code{\link{op_negative}()} \cr
+\code{\link{op_nonzero}()} \cr
+\code{\link{op_norm}()} \cr
+\code{\link{op_normalize}()} \cr
+\code{\link{op_not_equal}()} \cr
+\code{\link{op_one_hot}()} \cr
+\code{\link{op_ones}()} \cr
+\code{\link{op_ones_like}()} \cr
+\code{\link{op_outer}()} \cr
+\code{\link{op_pad}()} \cr
+\code{\link{op_polar}()} \cr
+\code{\link{op_power}()} \cr
+\code{\link{op_prod}()} \cr
+\code{\link{op_psnr}()} \cr
+\code{\link{op_qr}()} \cr
+\code{\link{op_quantile}()} \cr
+\code{\link{op_ravel}()} \cr
+\code{\link{op_real}()} \cr
+\code{\link{op_rearrange}()} \cr
+\code{\link{op_reciprocal}()} \cr
+\code{\link{op_relu}()} \cr
+\code{\link{op_relu6}()} \cr
+\code{\link{op_repeat}()} \cr
+\code{\link{op_reshape}()} \cr
+\code{\link{op_rfft}()} \cr
+\code{\link{op_right_shift}()} \cr
+\code{\link{op_rms_normalization}()} \cr
+\code{\link{op_roll}()} \cr
+\code{\link{op_rot90}()} \cr
+\code{\link{op_round}()} \cr
+\code{\link{op_rsqrt}()} \cr
+\code{\link{op_saturate_cast}()} \cr
+\code{\link{op_scan}()} \cr
+\code{\link{op_scatter}()} \cr
+\code{\link{op_scatter_update}()} \cr
+\code{\link{op_searchsorted}()} \cr
+\code{\link{op_segment_max}()} \cr
+\code{\link{op_segment_sum}()} \cr
+\code{\link{op_select}()} \cr
+\code{\link{op_selu}()} \cr
+\code{\link{op_separable_conv}()} \cr
+\code{\link{op_shape}()} \cr
+\code{\link{op_sigmoid}()} \cr
+\code{\link{op_sign}()} \cr
+\code{\link{op_signbit}()} \cr
+\code{\link{op_silu}()} \cr
+\code{\link{op_sin}()} \cr
+\code{\link{op_sinh}()} \cr
+\code{\link{op_size}()} \cr
+\code{\link{op_slice}()} \cr
+\code{\link{op_slice_update}()} \cr
+\code{\link{op_slogdet}()} \cr
+\code{\link{op_soft_shrink}()} \cr
+\code{\link{op_softmax}()} \cr
+\code{\link{op_softplus}()} \cr
+\code{\link{op_softsign}()} \cr
+\code{\link{op_solve}()} \cr
+\code{\link{op_solve_triangular}()} \cr
+\code{\link{op_sort}()} \cr
+\code{\link{op_sparse_categorical_crossentropy}()} \cr
+\code{\link{op_sparse_plus}()} \cr
+\code{\link{op_sparse_sigmoid}()} \cr
+\code{\link{op_sparsemax}()} \cr
+\code{\link{op_split}()} \cr
+\code{\link{op_sqrt}()} \cr
+\code{\link{op_square}()} \cr
+\code{\link{op_squareplus}()} \cr
+\code{\link{op_squeeze}()} \cr
+\code{\link{op_stack}()} \cr
+\code{\link{op_std}()} \cr
+\code{\link{op_stft}()} \cr
+\code{\link{op_stop_gradient}()} \cr
+\code{\link{op_subset}()} \cr
+\code{\link{op_subtract}()} \cr
+\code{\link{op_sum}()} \cr
+\code{\link{op_svd}()} \cr
+\code{\link{op_swapaxes}()} \cr
+\code{\link{op_switch}()} \cr
+\code{\link{op_take}()} \cr
+\code{\link{op_take_along_axis}()} \cr
+\code{\link{op_tan}()} \cr
+\code{\link{op_tanh}()} \cr
+\code{\link{op_tanh_shrink}()} \cr
+\code{\link{op_tensordot}()} \cr
+\code{\link{op_threshold}()} \cr
+\code{\link{op_tile}()} \cr
+\code{\link{op_top_k}()} \cr
+\code{\link{op_trace}()} \cr
+\code{\link{op_transpose}()} \cr
+\code{\link{op_tri}()} \cr
+\code{\link{op_tril}()} \cr
+\code{\link{op_triu}()} \cr
+\code{\link{op_trunc}()} \cr
+\code{\link{op_unravel_index}()} \cr
+\code{\link{op_unstack}()} \cr
+\code{\link{op_var}()} \cr
+\code{\link{op_vdot}()} \cr
+\code{\link{op_vectorize}()} \cr
+\code{\link{op_vectorized_map}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
+\code{\link{op_vstack}()} \cr
+\code{\link{op_where}()} \cr
+\code{\link{op_while_loop}()} \cr
+\code{\link{op_zeros}()} \cr
+\code{\link{op_zeros_like}()} \cr
+}
+\concept{numpy ops}
+\concept{ops}
diff --git a/man/op_layer_normalization.Rd b/man/op_layer_normalization.Rd
new file mode 100644
index 000000000..371beadd5
--- /dev/null
+++ b/man/op_layer_normalization.Rd
@@ -0,0 +1,377 @@
+% Generated by roxygen2: do not edit by hand
+% Please edit documentation in R/ops-nn.R
+\name{op_layer_normalization}
+\alias{op_layer_normalization}
+\title{Layer normalization (Ba et al., 2016).}
+\usage{
+op_layer_normalization(
+  x,
+  gamma = NULL,
+  beta = NULL,
+  axis = -1L,
+  epsilon = NULL,
+  ...
+)
+}
+\arguments{
+\item{x}{Input tensor.}
+
+\item{gamma}{Optional scaling factor applied to the normalized output.}
+
+\item{beta}{Optional offset added to the normalized output.}
+
+\item{axis}{Axis or axes along which to compute statistics. Defaults to \code{-1}.}
+
+\item{epsilon}{Small constant added to the variance for numerical stability.}
+
+\item{...}{For forward/backward compatibility.}
+}
+\value{
+Tensor with the same shape as \code{x} containing the normalized values.
+}
+\description{
+Normalizes activations in \code{x} for each example independently by centering to
+mean 0 and scaling to unit variance along the specified \code{axis}.
+}
+\section{Examples}{
+\if{html}{\out{<div class="sourceCode r">}}\preformatted{x <- op_arange(5, dtype = "float32")
+op_layer_normalization(x)
+}\if{html}{\out{</div>}}
+
+\if{html}{\out{<div class="sourceCode">}}\preformatted{## tf.Tensor([-1.4142134 -0.7071067  0.         0.7071068  1.4142137], shape=(5), dtype=float32)
+
+}\if{html}{\out{</div>}}
+}
+
+\seealso{
+Other nn ops: \cr
+\code{\link{op_average_pool}()} \cr
+\code{\link{op_batch_normalization}()} \cr
+\code{\link{op_binary_crossentropy}()} \cr
+\code{\link{op_categorical_crossentropy}()} \cr
+\code{\link{op_celu}()} \cr
+\code{\link{op_conv}()} \cr
+\code{\link{op_conv_transpose}()} \cr
+\code{\link{op_ctc_loss}()} \cr
+\code{\link{op_depthwise_conv}()} \cr
+\code{\link{op_dot_product_attention}()} \cr
+\code{\link{op_elu}()} \cr
+\code{\link{op_gelu}()} \cr
+\code{\link{op_glu}()} \cr
+\code{\link{op_hard_shrink}()} \cr
+\code{\link{op_hard_sigmoid}()} \cr
+\code{\link{op_hard_silu}()} \cr
+\code{\link{op_hard_tanh}()} \cr
+\code{\link{op_leaky_relu}()} \cr
+\code{\link{op_log_sigmoid}()} \cr
+\code{\link{op_log_softmax}()} \cr
+\code{\link{op_max_pool}()} \cr
+\code{\link{op_moments}()} \cr
+\code{\link{op_multi_hot}()} \cr
+\code{\link{op_normalize}()} \cr
+\code{\link{op_one_hot}()} \cr
+\code{\link{op_polar}()} \cr
+\code{\link{op_psnr}()} \cr
+\code{\link{op_relu}()} \cr
+\code{\link{op_relu6}()} \cr
+\code{\link{op_rms_normalization}()} \cr
+\code{\link{op_selu}()} \cr
+\code{\link{op_separable_conv}()} \cr
+\code{\link{op_sigmoid}()} \cr
+\code{\link{op_silu}()} \cr
+\code{\link{op_soft_shrink}()} \cr
+\code{\link{op_softmax}()} \cr
+\code{\link{op_softplus}()} \cr
+\code{\link{op_softsign}()} \cr
+\code{\link{op_sparse_categorical_crossentropy}()} \cr
+\code{\link{op_sparse_plus}()} \cr
+\code{\link{op_sparse_sigmoid}()} \cr
+\code{\link{op_sparsemax}()} \cr
+\code{\link{op_squareplus}()} \cr
+\code{\link{op_tanh_shrink}()} \cr
+\code{\link{op_threshold}()} \cr
+\code{\link{op_unravel_index}()} \cr
+
+Other ops: \cr
+\code{\link{op_abs}()} \cr
+\code{\link{op_add}()} \cr
+\code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
+\code{\link{op_any}()} \cr
+\code{\link{op_append}()} \cr
+\code{\link{op_arange}()} \cr
+\code{\link{op_arccos}()} \cr
+\code{\link{op_arccosh}()} \cr
+\code{\link{op_arcsin}()} \cr
+\code{\link{op_arcsinh}()} \cr
+\code{\link{op_arctan}()} \cr
+\code{\link{op_arctan2}()} \cr
+\code{\link{op_arctanh}()} \cr
+\code{\link{op_argmax}()} \cr
+\code{\link{op_argmin}()} \cr
+\code{\link{op_argpartition}()} \cr
+\code{\link{op_argsort}()} \cr
+\code{\link{op_array}()} \cr
+\code{\link{op_associative_scan}()} \cr
+\code{\link{op_average}()} \cr
+\code{\link{op_average_pool}()} \cr
+\code{\link{op_bartlett}()} \cr
+\code{\link{op_batch_normalization}()} \cr
+\code{\link{op_binary_crossentropy}()} \cr
+\code{\link{op_bincount}()} \cr
+\code{\link{op_bitwise_and}()} \cr
+\code{\link{op_bitwise_invert}()} \cr
+\code{\link{op_bitwise_left_shift}()} \cr
+\code{\link{op_bitwise_not}()} \cr
+\code{\link{op_bitwise_or}()} \cr
+\code{\link{op_bitwise_right_shift}()} \cr
+\code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
+\code{\link{op_broadcast_to}()} \cr
+\code{\link{op_cast}()} \cr
+\code{\link{op_categorical_crossentropy}()} \cr
+\code{\link{op_cbrt}()} \cr
+\code{\link{op_ceil}()} \cr
+\code{\link{op_celu}()} \cr
+\code{\link{op_cholesky}()} \cr
+\code{\link{op_clip}()} \cr
+\code{\link{op_concatenate}()} \cr
+\code{\link{op_cond}()} \cr
+\code{\link{op_conj}()} \cr
+\code{\link{op_conv}()} \cr
+\code{\link{op_conv_transpose}()} \cr
+\code{\link{op_convert_to_numpy}()} \cr
+\code{\link{op_convert_to_tensor}()} \cr
+\code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
+\code{\link{op_correlate}()} \cr
+\code{\link{op_cos}()} \cr
+\code{\link{op_cosh}()} \cr
+\code{\link{op_count_nonzero}()} \cr
+\code{\link{op_cross}()} \cr
+\code{\link{op_ctc_decode}()} \cr
+\code{\link{op_ctc_loss}()} \cr
+\code{\link{op_cumprod}()} \cr
+\code{\link{op_cumsum}()} \cr
+\code{\link{op_custom_gradient}()} \cr
+\code{\link{op_deg2rad}()} \cr
+\code{\link{op_depthwise_conv}()} \cr
+\code{\link{op_det}()} \cr
+\code{\link{op_diag}()} \cr
+\code{\link{op_diagflat}()} \cr
+\code{\link{op_diagonal}()} \cr
+\code{\link{op_diff}()} \cr
+\code{\link{op_digitize}()} \cr
+\code{\link{op_divide}()} \cr
+\code{\link{op_divide_no_nan}()} \cr
+\code{\link{op_dot}()} \cr
+\code{\link{op_dot_product_attention}()} \cr
+\code{\link{op_dtype}()} \cr
+\code{\link{op_eig}()} \cr
+\code{\link{op_eigh}()} \cr
+\code{\link{op_einsum}()} \cr
+\code{\link{op_elu}()} \cr
+\code{\link{op_empty}()} \cr
+\code{\link{op_equal}()} \cr
+\code{\link{op_erf}()} \cr
+\code{\link{op_erfinv}()} \cr
+\code{\link{op_exp}()} \cr
+\code{\link{op_exp2}()} \cr
+\code{\link{op_expand_dims}()} \cr
+\code{\link{op_expm1}()} \cr
+\code{\link{op_extract_sequences}()} \cr
+\code{\link{op_eye}()} \cr
+\code{\link{op_fft}()} \cr
+\code{\link{op_fft2}()} \cr
+\code{\link{op_flip}()} \cr
+\code{\link{op_floor}()} \cr
+\code{\link{op_floor_divide}()} \cr
+\code{\link{op_fori_loop}()} \cr
+\code{\link{op_full}()} \cr
+\code{\link{op_full_like}()} \cr
+\code{\link{op_gelu}()} \cr
+\code{\link{op_get_item}()} \cr
+\code{\link{op_glu}()} \cr
+\code{\link{op_greater}()} \cr
+\code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
+\code{\link{op_hard_shrink}()} \cr
+\code{\link{op_hard_sigmoid}()} \cr
+\code{\link{op_hard_silu}()} \cr
+\code{\link{op_hard_tanh}()} \cr
+\code{\link{op_heaviside}()} \cr
+\code{\link{op_histogram}()} \cr
+\code{\link{op_hstack}()} \cr
+\code{\link{op_identity}()} \cr
+\code{\link{op_ifft2}()} \cr
+\code{\link{op_imag}()} \cr
+\code{\link{op_image_affine_transform}()} \cr
+\code{\link{op_image_crop}()} \cr
+\code{\link{op_image_extract_patches}()} \cr
+\code{\link{op_image_gaussian_blur}()} \cr
+\code{\link{op_image_hsv_to_rgb}()} \cr
+\code{\link{op_image_map_coordinates}()} \cr
+\code{\link{op_image_pad}()} \cr
+\code{\link{op_image_perspective_transform}()} \cr
+\code{\link{op_image_resize}()} \cr
+\code{\link{op_image_rgb_to_grayscale}()} \cr
+\code{\link{op_image_rgb_to_hsv}()} \cr
+\code{\link{op_in_top_k}()} \cr
+\code{\link{op_inner}()} \cr
+\code{\link{op_inv}()} \cr
+\code{\link{op_irfft}()} \cr
+\code{\link{op_is_tensor}()} \cr
+\code{\link{op_isclose}()} \cr
+\code{\link{op_isfinite}()} \cr
+\code{\link{op_isinf}()} \cr
+\code{\link{op_isnan}()} \cr
+\code{\link{op_istft}()} \cr
+\code{\link{op_kaiser}()} \cr
+\code{\link{op_leaky_relu}()} \cr
+\code{\link{op_left_shift}()} \cr
+\code{\link{op_less}()} \cr
+\code{\link{op_less_equal}()} \cr
+\code{\link{op_linspace}()} \cr
+\code{\link{op_log}()} \cr
+\code{\link{op_log10}()} \cr
+\code{\link{op_log1p}()} \cr
+\code{\link{op_log2}()} \cr
+\code{\link{op_log_sigmoid}()} \cr
+\code{\link{op_log_softmax}()} \cr
+\code{\link{op_logaddexp}()} \cr
+\code{\link{op_logdet}()} \cr
+\code{\link{op_logical_and}()} \cr
+\code{\link{op_logical_not}()} \cr
+\code{\link{op_logical_or}()} \cr
+\code{\link{op_logical_xor}()} \cr
+\code{\link{op_logspace}()} \cr
+\code{\link{op_logsumexp}()} \cr
+\code{\link{op_lstsq}()} \cr
+\code{\link{op_lu_factor}()} \cr
+\code{\link{op_map}()} \cr
+\code{\link{op_matmul}()} \cr
+\code{\link{op_max}()} \cr
+\code{\link{op_max_pool}()} \cr
+\code{\link{op_maximum}()} \cr
+\code{\link{op_mean}()} \cr
+\code{\link{op_median}()} \cr
+\code{\link{op_meshgrid}()} \cr
+\code{\link{op_min}()} \cr
+\code{\link{op_minimum}()} \cr
+\code{\link{op_mod}()} \cr
+\code{\link{op_moments}()} \cr
+\code{\link{op_moveaxis}()} \cr
+\code{\link{op_multi_hot}()} \cr
+\code{\link{op_multiply}()} \cr
+\code{\link{op_nan_to_num}()} \cr
+\code{\link{op_ndim}()} \cr
+\code{\link{op_negative}()} \cr
+\code{\link{op_nonzero}()} \cr
+\code{\link{op_norm}()} \cr
+\code{\link{op_normalize}()} \cr
+\code{\link{op_not_equal}()} \cr
+\code{\link{op_one_hot}()} \cr
+\code{\link{op_ones}()} \cr
+\code{\link{op_ones_like}()} \cr
+\code{\link{op_outer}()} \cr
+\code{\link{op_pad}()} \cr
+\code{\link{op_polar}()} \cr
+\code{\link{op_power}()} \cr
+\code{\link{op_prod}()} \cr
+\code{\link{op_psnr}()} \cr
+\code{\link{op_qr}()} \cr
+\code{\link{op_quantile}()} \cr
+\code{\link{op_ravel}()} \cr
+\code{\link{op_real}()} \cr
+\code{\link{op_rearrange}()} \cr
+\code{\link{op_reciprocal}()} \cr
+\code{\link{op_relu}()} \cr
+\code{\link{op_relu6}()} \cr
+\code{\link{op_repeat}()} \cr
+\code{\link{op_reshape}()} \cr
+\code{\link{op_rfft}()} \cr
+\code{\link{op_right_shift}()} \cr
+\code{\link{op_rms_normalization}()} \cr
+\code{\link{op_roll}()} \cr
+\code{\link{op_rot90}()} \cr
+\code{\link{op_round}()} \cr
+\code{\link{op_rsqrt}()} \cr
+\code{\link{op_saturate_cast}()} \cr
+\code{\link{op_scan}()} \cr
+\code{\link{op_scatter}()} \cr
+\code{\link{op_scatter_update}()} \cr
+\code{\link{op_searchsorted}()} \cr
+\code{\link{op_segment_max}()} \cr
+\code{\link{op_segment_sum}()} \cr
+\code{\link{op_select}()} \cr
+\code{\link{op_selu}()} \cr
+\code{\link{op_separable_conv}()} \cr
+\code{\link{op_shape}()} \cr
+\code{\link{op_sigmoid}()} \cr
+\code{\link{op_sign}()} \cr
+\code{\link{op_signbit}()} \cr
+\code{\link{op_silu}()} \cr
+\code{\link{op_sin}()} \cr
+\code{\link{op_sinh}()} \cr
+\code{\link{op_size}()} \cr
+\code{\link{op_slice}()} \cr
+\code{\link{op_slice_update}()} \cr
+\code{\link{op_slogdet}()} \cr
+\code{\link{op_soft_shrink}()} \cr
+\code{\link{op_softmax}()} \cr
+\code{\link{op_softplus}()} \cr
+\code{\link{op_softsign}()} \cr
+\code{\link{op_solve}()} \cr
+\code{\link{op_solve_triangular}()} \cr
+\code{\link{op_sort}()} \cr
+\code{\link{op_sparse_categorical_crossentropy}()} \cr
+\code{\link{op_sparse_plus}()} \cr
+\code{\link{op_sparse_sigmoid}()} \cr
+\code{\link{op_sparsemax}()} \cr
+\code{\link{op_split}()} \cr
+\code{\link{op_sqrt}()} \cr
+\code{\link{op_square}()} \cr
+\code{\link{op_squareplus}()} \cr
+\code{\link{op_squeeze}()} \cr
+\code{\link{op_stack}()} \cr
+\code{\link{op_std}()} \cr
+\code{\link{op_stft}()} \cr
+\code{\link{op_stop_gradient}()} \cr
+\code{\link{op_subset}()} \cr
+\code{\link{op_subtract}()} \cr
+\code{\link{op_sum}()} \cr
+\code{\link{op_svd}()} \cr
+\code{\link{op_swapaxes}()} \cr
+\code{\link{op_switch}()} \cr
+\code{\link{op_take}()} \cr
+\code{\link{op_take_along_axis}()} \cr
+\code{\link{op_tan}()} \cr
+\code{\link{op_tanh}()} \cr
+\code{\link{op_tanh_shrink}()} \cr
+\code{\link{op_tensordot}()} \cr
+\code{\link{op_threshold}()} \cr
+\code{\link{op_tile}()} \cr
+\code{\link{op_top_k}()} \cr
+\code{\link{op_trace}()} \cr
+\code{\link{op_transpose}()} \cr
+\code{\link{op_tri}()} \cr
+\code{\link{op_tril}()} \cr
+\code{\link{op_triu}()} \cr
+\code{\link{op_trunc}()} \cr
+\code{\link{op_unravel_index}()} \cr
+\code{\link{op_unstack}()} \cr
+\code{\link{op_var}()} \cr
+\code{\link{op_vdot}()} \cr
+\code{\link{op_vectorize}()} \cr
+\code{\link{op_vectorized_map}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
+\code{\link{op_vstack}()} \cr
+\code{\link{op_where}()} \cr
+\code{\link{op_while_loop}()} \cr
+\code{\link{op_zeros}()} \cr
+\code{\link{op_zeros_like}()} \cr
+}
+\concept{nn ops}
+\concept{ops}
diff --git a/man/op_leaky_relu.Rd b/man/op_leaky_relu.Rd
index f1e23de15..64dae3415 100644
--- a/man/op_leaky_relu.Rd
+++ b/man/op_leaky_relu.Rd
@@ -63,6 +63,7 @@ Other nn ops: \cr
 \code{\link{op_hard_sigmoid}()} \cr
 \code{\link{op_hard_silu}()} \cr
 \code{\link{op_hard_tanh}()} \cr
+\code{\link{op_layer_normalization}()} \cr
 \code{\link{op_log_sigmoid}()} \cr
 \code{\link{op_log_softmax}()} \cr
 \code{\link{op_max_pool}()} \cr
@@ -85,6 +86,7 @@ Other nn ops: \cr
 \code{\link{op_softsign}()} \cr
 \code{\link{op_sparse_categorical_crossentropy}()} \cr
 \code{\link{op_sparse_plus}()} \cr
+\code{\link{op_sparse_sigmoid}()} \cr
 \code{\link{op_sparsemax}()} \cr
 \code{\link{op_squareplus}()} \cr
 \code{\link{op_tanh_shrink}()} \cr
@@ -95,6 +97,7 @@ Other ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -113,6 +116,7 @@ Other ops: \cr
 \code{\link{op_associative_scan}()} \cr
 \code{\link{op_average}()} \cr
 \code{\link{op_average_pool}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_batch_normalization}()} \cr
 \code{\link{op_binary_crossentropy}()} \cr
 \code{\link{op_bincount}()} \cr
@@ -123,9 +127,11 @@ Other ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
 \code{\link{op_cast}()} \cr
 \code{\link{op_categorical_crossentropy}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_celu}()} \cr
 \code{\link{op_cholesky}()} \cr
@@ -138,6 +144,7 @@ Other ops: \cr
 \code{\link{op_convert_to_numpy}()} \cr
 \code{\link{op_convert_to_tensor}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -148,6 +155,7 @@ Other ops: \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
 \code{\link{op_custom_gradient}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_depthwise_conv}()} \cr
 \code{\link{op_det}()} \cr
 \code{\link{op_diag}()} \cr
@@ -187,10 +195,13 @@ Other ops: \cr
 \code{\link{op_glu}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
 \code{\link{op_hard_shrink}()} \cr
 \code{\link{op_hard_sigmoid}()} \cr
 \code{\link{op_hard_silu}()} \cr
 \code{\link{op_hard_tanh}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -217,6 +228,8 @@ Other ops: \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
 \code{\link{op_istft}()} \cr
+\code{\link{op_kaiser}()} \cr
+\code{\link{op_layer_normalization}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
 \code{\link{op_less_equal}()} \cr
@@ -315,6 +328,7 @@ Other ops: \cr
 \code{\link{op_sort}()} \cr
 \code{\link{op_sparse_categorical_crossentropy}()} \cr
 \code{\link{op_sparse_plus}()} \cr
+\code{\link{op_sparse_sigmoid}()} \cr
 \code{\link{op_sparsemax}()} \cr
 \code{\link{op_split}()} \cr
 \code{\link{op_sqrt}()} \cr
@@ -352,6 +366,8 @@ Other ops: \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
 \code{\link{op_vectorized_map}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_while_loop}()} \cr
diff --git a/man/op_left_shift.Rd b/man/op_left_shift.Rd
index ea8dbcb66..bb5439f60 100644
--- a/man/op_left_shift.Rd
+++ b/man/op_left_shift.Rd
@@ -24,6 +24,7 @@ Other numpy ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -40,6 +41,7 @@ Other numpy ops: \cr
 \code{\link{op_argsort}()} \cr
 \code{\link{op_array}()} \cr
 \code{\link{op_average}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_bincount}()} \cr
 \code{\link{op_bitwise_and}()} \cr
 \code{\link{op_bitwise_invert}()} \cr
@@ -48,12 +50,15 @@ Other numpy ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_clip}()} \cr
 \code{\link{op_concatenate}()} \cr
 \code{\link{op_conj}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -62,6 +67,7 @@ Other numpy ops: \cr
 \code{\link{op_ctc_decode}()} \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_diag}()} \cr
 \code{\link{op_diagflat}()} \cr
 \code{\link{op_diagonal}()} \cr
@@ -86,6 +92,9 @@ Other numpy ops: \cr
 \code{\link{op_get_item}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -95,6 +104,7 @@ Other numpy ops: \cr
 \code{\link{op_isfinite}()} \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
+\code{\link{op_kaiser}()} \cr
 \code{\link{op_less}()} \cr
 \code{\link{op_less_equal}()} \cr
 \code{\link{op_linspace}()} \cr
@@ -174,6 +184,8 @@ Other numpy ops: \cr
 \code{\link{op_var}()} \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_zeros}()} \cr
@@ -183,6 +195,7 @@ Other ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -201,6 +214,7 @@ Other ops: \cr
 \code{\link{op_associative_scan}()} \cr
 \code{\link{op_average}()} \cr
 \code{\link{op_average_pool}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_batch_normalization}()} \cr
 \code{\link{op_binary_crossentropy}()} \cr
 \code{\link{op_bincount}()} \cr
@@ -211,9 +225,11 @@ Other ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
 \code{\link{op_cast}()} \cr
 \code{\link{op_categorical_crossentropy}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_celu}()} \cr
 \code{\link{op_cholesky}()} \cr
@@ -226,6 +242,7 @@ Other ops: \cr
 \code{\link{op_convert_to_numpy}()} \cr
 \code{\link{op_convert_to_tensor}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -236,6 +253,7 @@ Other ops: \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
 \code{\link{op_custom_gradient}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_depthwise_conv}()} \cr
 \code{\link{op_det}()} \cr
 \code{\link{op_diag}()} \cr
@@ -275,10 +293,13 @@ Other ops: \cr
 \code{\link{op_glu}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
 \code{\link{op_hard_shrink}()} \cr
 \code{\link{op_hard_sigmoid}()} \cr
 \code{\link{op_hard_silu}()} \cr
 \code{\link{op_hard_tanh}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -305,6 +326,8 @@ Other ops: \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
 \code{\link{op_istft}()} \cr
+\code{\link{op_kaiser}()} \cr
+\code{\link{op_layer_normalization}()} \cr
 \code{\link{op_leaky_relu}()} \cr
 \code{\link{op_less}()} \cr
 \code{\link{op_less_equal}()} \cr
@@ -403,6 +426,7 @@ Other ops: \cr
 \code{\link{op_sort}()} \cr
 \code{\link{op_sparse_categorical_crossentropy}()} \cr
 \code{\link{op_sparse_plus}()} \cr
+\code{\link{op_sparse_sigmoid}()} \cr
 \code{\link{op_sparsemax}()} \cr
 \code{\link{op_split}()} \cr
 \code{\link{op_sqrt}()} \cr
@@ -440,6 +464,8 @@ Other ops: \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
 \code{\link{op_vectorized_map}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_while_loop}()} \cr
diff --git a/man/op_less.Rd b/man/op_less.Rd
index 854b1e841..d2173084f 100644
--- a/man/op_less.Rd
+++ b/man/op_less.Rd
@@ -47,6 +47,7 @@ Other numpy ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -63,6 +64,7 @@ Other numpy ops: \cr
 \code{\link{op_argsort}()} \cr
 \code{\link{op_array}()} \cr
 \code{\link{op_average}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_bincount}()} \cr
 \code{\link{op_bitwise_and}()} \cr
 \code{\link{op_bitwise_invert}()} \cr
@@ -71,12 +73,15 @@ Other numpy ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_clip}()} \cr
 \code{\link{op_concatenate}()} \cr
 \code{\link{op_conj}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -85,6 +90,7 @@ Other numpy ops: \cr
 \code{\link{op_ctc_decode}()} \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_diag}()} \cr
 \code{\link{op_diagflat}()} \cr
 \code{\link{op_diagonal}()} \cr
@@ -109,6 +115,9 @@ Other numpy ops: \cr
 \code{\link{op_get_item}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -118,6 +127,7 @@ Other numpy ops: \cr
 \code{\link{op_isfinite}()} \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
+\code{\link{op_kaiser}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less_equal}()} \cr
 \code{\link{op_linspace}()} \cr
@@ -197,6 +207,8 @@ Other numpy ops: \cr
 \code{\link{op_var}()} \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_zeros}()} \cr
@@ -206,6 +218,7 @@ Other ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -224,6 +237,7 @@ Other ops: \cr
 \code{\link{op_associative_scan}()} \cr
 \code{\link{op_average}()} \cr
 \code{\link{op_average_pool}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_batch_normalization}()} \cr
 \code{\link{op_binary_crossentropy}()} \cr
 \code{\link{op_bincount}()} \cr
@@ -234,9 +248,11 @@ Other ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
 \code{\link{op_cast}()} \cr
 \code{\link{op_categorical_crossentropy}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_celu}()} \cr
 \code{\link{op_cholesky}()} \cr
@@ -249,6 +265,7 @@ Other ops: \cr
 \code{\link{op_convert_to_numpy}()} \cr
 \code{\link{op_convert_to_tensor}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -259,6 +276,7 @@ Other ops: \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
 \code{\link{op_custom_gradient}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_depthwise_conv}()} \cr
 \code{\link{op_det}()} \cr
 \code{\link{op_diag}()} \cr
@@ -298,10 +316,13 @@ Other ops: \cr
 \code{\link{op_glu}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
 \code{\link{op_hard_shrink}()} \cr
 \code{\link{op_hard_sigmoid}()} \cr
 \code{\link{op_hard_silu}()} \cr
 \code{\link{op_hard_tanh}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -328,6 +349,8 @@ Other ops: \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
 \code{\link{op_istft}()} \cr
+\code{\link{op_kaiser}()} \cr
+\code{\link{op_layer_normalization}()} \cr
 \code{\link{op_leaky_relu}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less_equal}()} \cr
@@ -426,6 +449,7 @@ Other ops: \cr
 \code{\link{op_sort}()} \cr
 \code{\link{op_sparse_categorical_crossentropy}()} \cr
 \code{\link{op_sparse_plus}()} \cr
+\code{\link{op_sparse_sigmoid}()} \cr
 \code{\link{op_sparsemax}()} \cr
 \code{\link{op_split}()} \cr
 \code{\link{op_sqrt}()} \cr
@@ -463,6 +487,8 @@ Other ops: \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
 \code{\link{op_vectorized_map}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_while_loop}()} \cr
diff --git a/man/op_less_equal.Rd b/man/op_less_equal.Rd
index 3bf0490f3..c95888e09 100644
--- a/man/op_less_equal.Rd
+++ b/man/op_less_equal.Rd
@@ -47,6 +47,7 @@ Other numpy ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -63,6 +64,7 @@ Other numpy ops: \cr
 \code{\link{op_argsort}()} \cr
 \code{\link{op_array}()} \cr
 \code{\link{op_average}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_bincount}()} \cr
 \code{\link{op_bitwise_and}()} \cr
 \code{\link{op_bitwise_invert}()} \cr
@@ -71,12 +73,15 @@ Other numpy ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_clip}()} \cr
 \code{\link{op_concatenate}()} \cr
 \code{\link{op_conj}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -85,6 +90,7 @@ Other numpy ops: \cr
 \code{\link{op_ctc_decode}()} \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_diag}()} \cr
 \code{\link{op_diagflat}()} \cr
 \code{\link{op_diagonal}()} \cr
@@ -109,6 +115,9 @@ Other numpy ops: \cr
 \code{\link{op_get_item}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -118,6 +127,7 @@ Other numpy ops: \cr
 \code{\link{op_isfinite}()} \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
+\code{\link{op_kaiser}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
 \code{\link{op_linspace}()} \cr
@@ -197,6 +207,8 @@ Other numpy ops: \cr
 \code{\link{op_var}()} \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_zeros}()} \cr
@@ -206,6 +218,7 @@ Other ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -224,6 +237,7 @@ Other ops: \cr
 \code{\link{op_associative_scan}()} \cr
 \code{\link{op_average}()} \cr
 \code{\link{op_average_pool}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_batch_normalization}()} \cr
 \code{\link{op_binary_crossentropy}()} \cr
 \code{\link{op_bincount}()} \cr
@@ -234,9 +248,11 @@ Other ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
 \code{\link{op_cast}()} \cr
 \code{\link{op_categorical_crossentropy}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_celu}()} \cr
 \code{\link{op_cholesky}()} \cr
@@ -249,6 +265,7 @@ Other ops: \cr
 \code{\link{op_convert_to_numpy}()} \cr
 \code{\link{op_convert_to_tensor}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -259,6 +276,7 @@ Other ops: \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
 \code{\link{op_custom_gradient}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_depthwise_conv}()} \cr
 \code{\link{op_det}()} \cr
 \code{\link{op_diag}()} \cr
@@ -298,10 +316,13 @@ Other ops: \cr
 \code{\link{op_glu}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
 \code{\link{op_hard_shrink}()} \cr
 \code{\link{op_hard_sigmoid}()} \cr
 \code{\link{op_hard_silu}()} \cr
 \code{\link{op_hard_tanh}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -328,6 +349,8 @@ Other ops: \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
 \code{\link{op_istft}()} \cr
+\code{\link{op_kaiser}()} \cr
+\code{\link{op_layer_normalization}()} \cr
 \code{\link{op_leaky_relu}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
@@ -426,6 +449,7 @@ Other ops: \cr
 \code{\link{op_sort}()} \cr
 \code{\link{op_sparse_categorical_crossentropy}()} \cr
 \code{\link{op_sparse_plus}()} \cr
+\code{\link{op_sparse_sigmoid}()} \cr
 \code{\link{op_sparsemax}()} \cr
 \code{\link{op_split}()} \cr
 \code{\link{op_sqrt}()} \cr
@@ -463,6 +487,8 @@ Other ops: \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
 \code{\link{op_vectorized_map}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_while_loop}()} \cr
diff --git a/man/op_linspace.Rd b/man/op_linspace.Rd
index db3a59dc8..06f5ffa5f 100644
--- a/man/op_linspace.Rd
+++ b/man/op_linspace.Rd
@@ -59,6 +59,7 @@ Other numpy ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -75,6 +76,7 @@ Other numpy ops: \cr
 \code{\link{op_argsort}()} \cr
 \code{\link{op_array}()} \cr
 \code{\link{op_average}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_bincount}()} \cr
 \code{\link{op_bitwise_and}()} \cr
 \code{\link{op_bitwise_invert}()} \cr
@@ -83,12 +85,15 @@ Other numpy ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_clip}()} \cr
 \code{\link{op_concatenate}()} \cr
 \code{\link{op_conj}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -97,6 +102,7 @@ Other numpy ops: \cr
 \code{\link{op_ctc_decode}()} \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_diag}()} \cr
 \code{\link{op_diagflat}()} \cr
 \code{\link{op_diagonal}()} \cr
@@ -121,6 +127,9 @@ Other numpy ops: \cr
 \code{\link{op_get_item}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -130,6 +139,7 @@ Other numpy ops: \cr
 \code{\link{op_isfinite}()} \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
+\code{\link{op_kaiser}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
 \code{\link{op_less_equal}()} \cr
@@ -209,6 +219,8 @@ Other numpy ops: \cr
 \code{\link{op_var}()} \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_zeros}()} \cr
@@ -218,6 +230,7 @@ Other ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -236,6 +249,7 @@ Other ops: \cr
 \code{\link{op_associative_scan}()} \cr
 \code{\link{op_average}()} \cr
 \code{\link{op_average_pool}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_batch_normalization}()} \cr
 \code{\link{op_binary_crossentropy}()} \cr
 \code{\link{op_bincount}()} \cr
@@ -246,9 +260,11 @@ Other ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
 \code{\link{op_cast}()} \cr
 \code{\link{op_categorical_crossentropy}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_celu}()} \cr
 \code{\link{op_cholesky}()} \cr
@@ -261,6 +277,7 @@ Other ops: \cr
 \code{\link{op_convert_to_numpy}()} \cr
 \code{\link{op_convert_to_tensor}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -271,6 +288,7 @@ Other ops: \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
 \code{\link{op_custom_gradient}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_depthwise_conv}()} \cr
 \code{\link{op_det}()} \cr
 \code{\link{op_diag}()} \cr
@@ -310,10 +328,13 @@ Other ops: \cr
 \code{\link{op_glu}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
 \code{\link{op_hard_shrink}()} \cr
 \code{\link{op_hard_sigmoid}()} \cr
 \code{\link{op_hard_silu}()} \cr
 \code{\link{op_hard_tanh}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -340,6 +361,8 @@ Other ops: \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
 \code{\link{op_istft}()} \cr
+\code{\link{op_kaiser}()} \cr
+\code{\link{op_layer_normalization}()} \cr
 \code{\link{op_leaky_relu}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
@@ -438,6 +461,7 @@ Other ops: \cr
 \code{\link{op_sort}()} \cr
 \code{\link{op_sparse_categorical_crossentropy}()} \cr
 \code{\link{op_sparse_plus}()} \cr
+\code{\link{op_sparse_sigmoid}()} \cr
 \code{\link{op_sparsemax}()} \cr
 \code{\link{op_split}()} \cr
 \code{\link{op_sqrt}()} \cr
@@ -475,6 +499,8 @@ Other ops: \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
 \code{\link{op_vectorized_map}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_while_loop}()} \cr
diff --git a/man/op_log.Rd b/man/op_log.Rd
index b8f02f7db..fd703ba1c 100644
--- a/man/op_log.Rd
+++ b/man/op_log.Rd
@@ -24,6 +24,7 @@ Other numpy ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -40,6 +41,7 @@ Other numpy ops: \cr
 \code{\link{op_argsort}()} \cr
 \code{\link{op_array}()} \cr
 \code{\link{op_average}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_bincount}()} \cr
 \code{\link{op_bitwise_and}()} \cr
 \code{\link{op_bitwise_invert}()} \cr
@@ -48,12 +50,15 @@ Other numpy ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_clip}()} \cr
 \code{\link{op_concatenate}()} \cr
 \code{\link{op_conj}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -62,6 +67,7 @@ Other numpy ops: \cr
 \code{\link{op_ctc_decode}()} \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_diag}()} \cr
 \code{\link{op_diagflat}()} \cr
 \code{\link{op_diagonal}()} \cr
@@ -86,6 +92,9 @@ Other numpy ops: \cr
 \code{\link{op_get_item}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -95,6 +104,7 @@ Other numpy ops: \cr
 \code{\link{op_isfinite}()} \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
+\code{\link{op_kaiser}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
 \code{\link{op_less_equal}()} \cr
@@ -174,6 +184,8 @@ Other numpy ops: \cr
 \code{\link{op_var}()} \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_zeros}()} \cr
@@ -183,6 +195,7 @@ Other ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -201,6 +214,7 @@ Other ops: \cr
 \code{\link{op_associative_scan}()} \cr
 \code{\link{op_average}()} \cr
 \code{\link{op_average_pool}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_batch_normalization}()} \cr
 \code{\link{op_binary_crossentropy}()} \cr
 \code{\link{op_bincount}()} \cr
@@ -211,9 +225,11 @@ Other ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
 \code{\link{op_cast}()} \cr
 \code{\link{op_categorical_crossentropy}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_celu}()} \cr
 \code{\link{op_cholesky}()} \cr
@@ -226,6 +242,7 @@ Other ops: \cr
 \code{\link{op_convert_to_numpy}()} \cr
 \code{\link{op_convert_to_tensor}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -236,6 +253,7 @@ Other ops: \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
 \code{\link{op_custom_gradient}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_depthwise_conv}()} \cr
 \code{\link{op_det}()} \cr
 \code{\link{op_diag}()} \cr
@@ -275,10 +293,13 @@ Other ops: \cr
 \code{\link{op_glu}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
 \code{\link{op_hard_shrink}()} \cr
 \code{\link{op_hard_sigmoid}()} \cr
 \code{\link{op_hard_silu}()} \cr
 \code{\link{op_hard_tanh}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -305,6 +326,8 @@ Other ops: \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
 \code{\link{op_istft}()} \cr
+\code{\link{op_kaiser}()} \cr
+\code{\link{op_layer_normalization}()} \cr
 \code{\link{op_leaky_relu}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
@@ -403,6 +426,7 @@ Other ops: \cr
 \code{\link{op_sort}()} \cr
 \code{\link{op_sparse_categorical_crossentropy}()} \cr
 \code{\link{op_sparse_plus}()} \cr
+\code{\link{op_sparse_sigmoid}()} \cr
 \code{\link{op_sparsemax}()} \cr
 \code{\link{op_split}()} \cr
 \code{\link{op_sqrt}()} \cr
@@ -440,6 +464,8 @@ Other ops: \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
 \code{\link{op_vectorized_map}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_while_loop}()} \cr
diff --git a/man/op_log10.Rd b/man/op_log10.Rd
index bfe7338af..cd5413d06 100644
--- a/man/op_log10.Rd
+++ b/man/op_log10.Rd
@@ -24,6 +24,7 @@ Other numpy ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -40,6 +41,7 @@ Other numpy ops: \cr
 \code{\link{op_argsort}()} \cr
 \code{\link{op_array}()} \cr
 \code{\link{op_average}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_bincount}()} \cr
 \code{\link{op_bitwise_and}()} \cr
 \code{\link{op_bitwise_invert}()} \cr
@@ -48,12 +50,15 @@ Other numpy ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_clip}()} \cr
 \code{\link{op_concatenate}()} \cr
 \code{\link{op_conj}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -62,6 +67,7 @@ Other numpy ops: \cr
 \code{\link{op_ctc_decode}()} \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_diag}()} \cr
 \code{\link{op_diagflat}()} \cr
 \code{\link{op_diagonal}()} \cr
@@ -86,6 +92,9 @@ Other numpy ops: \cr
 \code{\link{op_get_item}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -95,6 +104,7 @@ Other numpy ops: \cr
 \code{\link{op_isfinite}()} \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
+\code{\link{op_kaiser}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
 \code{\link{op_less_equal}()} \cr
@@ -174,6 +184,8 @@ Other numpy ops: \cr
 \code{\link{op_var}()} \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_zeros}()} \cr
@@ -183,6 +195,7 @@ Other ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -201,6 +214,7 @@ Other ops: \cr
 \code{\link{op_associative_scan}()} \cr
 \code{\link{op_average}()} \cr
 \code{\link{op_average_pool}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_batch_normalization}()} \cr
 \code{\link{op_binary_crossentropy}()} \cr
 \code{\link{op_bincount}()} \cr
@@ -211,9 +225,11 @@ Other ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
 \code{\link{op_cast}()} \cr
 \code{\link{op_categorical_crossentropy}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_celu}()} \cr
 \code{\link{op_cholesky}()} \cr
@@ -226,6 +242,7 @@ Other ops: \cr
 \code{\link{op_convert_to_numpy}()} \cr
 \code{\link{op_convert_to_tensor}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -236,6 +253,7 @@ Other ops: \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
 \code{\link{op_custom_gradient}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_depthwise_conv}()} \cr
 \code{\link{op_det}()} \cr
 \code{\link{op_diag}()} \cr
@@ -275,10 +293,13 @@ Other ops: \cr
 \code{\link{op_glu}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
 \code{\link{op_hard_shrink}()} \cr
 \code{\link{op_hard_sigmoid}()} \cr
 \code{\link{op_hard_silu}()} \cr
 \code{\link{op_hard_tanh}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -305,6 +326,8 @@ Other ops: \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
 \code{\link{op_istft}()} \cr
+\code{\link{op_kaiser}()} \cr
+\code{\link{op_layer_normalization}()} \cr
 \code{\link{op_leaky_relu}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
@@ -403,6 +426,7 @@ Other ops: \cr
 \code{\link{op_sort}()} \cr
 \code{\link{op_sparse_categorical_crossentropy}()} \cr
 \code{\link{op_sparse_plus}()} \cr
+\code{\link{op_sparse_sigmoid}()} \cr
 \code{\link{op_sparsemax}()} \cr
 \code{\link{op_split}()} \cr
 \code{\link{op_sqrt}()} \cr
@@ -440,6 +464,8 @@ Other ops: \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
 \code{\link{op_vectorized_map}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_while_loop}()} \cr
diff --git a/man/op_log1p.Rd b/man/op_log1p.Rd
index 6bc15cb9a..b65e58b1e 100644
--- a/man/op_log1p.Rd
+++ b/man/op_log1p.Rd
@@ -24,6 +24,7 @@ Other numpy ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -40,6 +41,7 @@ Other numpy ops: \cr
 \code{\link{op_argsort}()} \cr
 \code{\link{op_array}()} \cr
 \code{\link{op_average}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_bincount}()} \cr
 \code{\link{op_bitwise_and}()} \cr
 \code{\link{op_bitwise_invert}()} \cr
@@ -48,12 +50,15 @@ Other numpy ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_clip}()} \cr
 \code{\link{op_concatenate}()} \cr
 \code{\link{op_conj}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -62,6 +67,7 @@ Other numpy ops: \cr
 \code{\link{op_ctc_decode}()} \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_diag}()} \cr
 \code{\link{op_diagflat}()} \cr
 \code{\link{op_diagonal}()} \cr
@@ -86,6 +92,9 @@ Other numpy ops: \cr
 \code{\link{op_get_item}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -95,6 +104,7 @@ Other numpy ops: \cr
 \code{\link{op_isfinite}()} \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
+\code{\link{op_kaiser}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
 \code{\link{op_less_equal}()} \cr
@@ -174,6 +184,8 @@ Other numpy ops: \cr
 \code{\link{op_var}()} \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_zeros}()} \cr
@@ -183,6 +195,7 @@ Other ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -201,6 +214,7 @@ Other ops: \cr
 \code{\link{op_associative_scan}()} \cr
 \code{\link{op_average}()} \cr
 \code{\link{op_average_pool}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_batch_normalization}()} \cr
 \code{\link{op_binary_crossentropy}()} \cr
 \code{\link{op_bincount}()} \cr
@@ -211,9 +225,11 @@ Other ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
 \code{\link{op_cast}()} \cr
 \code{\link{op_categorical_crossentropy}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_celu}()} \cr
 \code{\link{op_cholesky}()} \cr
@@ -226,6 +242,7 @@ Other ops: \cr
 \code{\link{op_convert_to_numpy}()} \cr
 \code{\link{op_convert_to_tensor}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -236,6 +253,7 @@ Other ops: \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
 \code{\link{op_custom_gradient}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_depthwise_conv}()} \cr
 \code{\link{op_det}()} \cr
 \code{\link{op_diag}()} \cr
@@ -275,10 +293,13 @@ Other ops: \cr
 \code{\link{op_glu}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
 \code{\link{op_hard_shrink}()} \cr
 \code{\link{op_hard_sigmoid}()} \cr
 \code{\link{op_hard_silu}()} \cr
 \code{\link{op_hard_tanh}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -305,6 +326,8 @@ Other ops: \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
 \code{\link{op_istft}()} \cr
+\code{\link{op_kaiser}()} \cr
+\code{\link{op_layer_normalization}()} \cr
 \code{\link{op_leaky_relu}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
@@ -403,6 +426,7 @@ Other ops: \cr
 \code{\link{op_sort}()} \cr
 \code{\link{op_sparse_categorical_crossentropy}()} \cr
 \code{\link{op_sparse_plus}()} \cr
+\code{\link{op_sparse_sigmoid}()} \cr
 \code{\link{op_sparsemax}()} \cr
 \code{\link{op_split}()} \cr
 \code{\link{op_sqrt}()} \cr
@@ -440,6 +464,8 @@ Other ops: \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
 \code{\link{op_vectorized_map}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_while_loop}()} \cr
diff --git a/man/op_log2.Rd b/man/op_log2.Rd
index f3f4f7fbe..4b5ca2dcd 100644
--- a/man/op_log2.Rd
+++ b/man/op_log2.Rd
@@ -24,6 +24,7 @@ Other numpy ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -40,6 +41,7 @@ Other numpy ops: \cr
 \code{\link{op_argsort}()} \cr
 \code{\link{op_array}()} \cr
 \code{\link{op_average}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_bincount}()} \cr
 \code{\link{op_bitwise_and}()} \cr
 \code{\link{op_bitwise_invert}()} \cr
@@ -48,12 +50,15 @@ Other numpy ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_clip}()} \cr
 \code{\link{op_concatenate}()} \cr
 \code{\link{op_conj}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -62,6 +67,7 @@ Other numpy ops: \cr
 \code{\link{op_ctc_decode}()} \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_diag}()} \cr
 \code{\link{op_diagflat}()} \cr
 \code{\link{op_diagonal}()} \cr
@@ -86,6 +92,9 @@ Other numpy ops: \cr
 \code{\link{op_get_item}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -95,6 +104,7 @@ Other numpy ops: \cr
 \code{\link{op_isfinite}()} \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
+\code{\link{op_kaiser}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
 \code{\link{op_less_equal}()} \cr
@@ -174,6 +184,8 @@ Other numpy ops: \cr
 \code{\link{op_var}()} \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_zeros}()} \cr
@@ -183,6 +195,7 @@ Other ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -201,6 +214,7 @@ Other ops: \cr
 \code{\link{op_associative_scan}()} \cr
 \code{\link{op_average}()} \cr
 \code{\link{op_average_pool}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_batch_normalization}()} \cr
 \code{\link{op_binary_crossentropy}()} \cr
 \code{\link{op_bincount}()} \cr
@@ -211,9 +225,11 @@ Other ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
 \code{\link{op_cast}()} \cr
 \code{\link{op_categorical_crossentropy}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_celu}()} \cr
 \code{\link{op_cholesky}()} \cr
@@ -226,6 +242,7 @@ Other ops: \cr
 \code{\link{op_convert_to_numpy}()} \cr
 \code{\link{op_convert_to_tensor}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -236,6 +253,7 @@ Other ops: \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
 \code{\link{op_custom_gradient}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_depthwise_conv}()} \cr
 \code{\link{op_det}()} \cr
 \code{\link{op_diag}()} \cr
@@ -275,10 +293,13 @@ Other ops: \cr
 \code{\link{op_glu}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
 \code{\link{op_hard_shrink}()} \cr
 \code{\link{op_hard_sigmoid}()} \cr
 \code{\link{op_hard_silu}()} \cr
 \code{\link{op_hard_tanh}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -305,6 +326,8 @@ Other ops: \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
 \code{\link{op_istft}()} \cr
+\code{\link{op_kaiser}()} \cr
+\code{\link{op_layer_normalization}()} \cr
 \code{\link{op_leaky_relu}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
@@ -403,6 +426,7 @@ Other ops: \cr
 \code{\link{op_sort}()} \cr
 \code{\link{op_sparse_categorical_crossentropy}()} \cr
 \code{\link{op_sparse_plus}()} \cr
+\code{\link{op_sparse_sigmoid}()} \cr
 \code{\link{op_sparsemax}()} \cr
 \code{\link{op_split}()} \cr
 \code{\link{op_sqrt}()} \cr
@@ -440,6 +464,8 @@ Other ops: \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
 \code{\link{op_vectorized_map}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_while_loop}()} \cr
diff --git a/man/op_log_sigmoid.Rd b/man/op_log_sigmoid.Rd
index 6f826f433..804fd543e 100644
--- a/man/op_log_sigmoid.Rd
+++ b/man/op_log_sigmoid.Rd
@@ -48,6 +48,7 @@ Other nn ops: \cr
 \code{\link{op_hard_sigmoid}()} \cr
 \code{\link{op_hard_silu}()} \cr
 \code{\link{op_hard_tanh}()} \cr
+\code{\link{op_layer_normalization}()} \cr
 \code{\link{op_leaky_relu}()} \cr
 \code{\link{op_log_softmax}()} \cr
 \code{\link{op_max_pool}()} \cr
@@ -70,6 +71,7 @@ Other nn ops: \cr
 \code{\link{op_softsign}()} \cr
 \code{\link{op_sparse_categorical_crossentropy}()} \cr
 \code{\link{op_sparse_plus}()} \cr
+\code{\link{op_sparse_sigmoid}()} \cr
 \code{\link{op_sparsemax}()} \cr
 \code{\link{op_squareplus}()} \cr
 \code{\link{op_tanh_shrink}()} \cr
@@ -80,6 +82,7 @@ Other ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -98,6 +101,7 @@ Other ops: \cr
 \code{\link{op_associative_scan}()} \cr
 \code{\link{op_average}()} \cr
 \code{\link{op_average_pool}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_batch_normalization}()} \cr
 \code{\link{op_binary_crossentropy}()} \cr
 \code{\link{op_bincount}()} \cr
@@ -108,9 +112,11 @@ Other ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
 \code{\link{op_cast}()} \cr
 \code{\link{op_categorical_crossentropy}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_celu}()} \cr
 \code{\link{op_cholesky}()} \cr
@@ -123,6 +129,7 @@ Other ops: \cr
 \code{\link{op_convert_to_numpy}()} \cr
 \code{\link{op_convert_to_tensor}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -133,6 +140,7 @@ Other ops: \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
 \code{\link{op_custom_gradient}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_depthwise_conv}()} \cr
 \code{\link{op_det}()} \cr
 \code{\link{op_diag}()} \cr
@@ -172,10 +180,13 @@ Other ops: \cr
 \code{\link{op_glu}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
 \code{\link{op_hard_shrink}()} \cr
 \code{\link{op_hard_sigmoid}()} \cr
 \code{\link{op_hard_silu}()} \cr
 \code{\link{op_hard_tanh}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -202,6 +213,8 @@ Other ops: \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
 \code{\link{op_istft}()} \cr
+\code{\link{op_kaiser}()} \cr
+\code{\link{op_layer_normalization}()} \cr
 \code{\link{op_leaky_relu}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
@@ -300,6 +313,7 @@ Other ops: \cr
 \code{\link{op_sort}()} \cr
 \code{\link{op_sparse_categorical_crossentropy}()} \cr
 \code{\link{op_sparse_plus}()} \cr
+\code{\link{op_sparse_sigmoid}()} \cr
 \code{\link{op_sparsemax}()} \cr
 \code{\link{op_split}()} \cr
 \code{\link{op_sqrt}()} \cr
@@ -337,6 +351,8 @@ Other ops: \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
 \code{\link{op_vectorized_map}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_while_loop}()} \cr
diff --git a/man/op_log_softmax.Rd b/man/op_log_softmax.Rd
index 5c9febe7b..67a6c51a2 100644
--- a/man/op_log_softmax.Rd
+++ b/man/op_log_softmax.Rd
@@ -52,6 +52,7 @@ Other nn ops: \cr
 \code{\link{op_hard_sigmoid}()} \cr
 \code{\link{op_hard_silu}()} \cr
 \code{\link{op_hard_tanh}()} \cr
+\code{\link{op_layer_normalization}()} \cr
 \code{\link{op_leaky_relu}()} \cr
 \code{\link{op_log_sigmoid}()} \cr
 \code{\link{op_max_pool}()} \cr
@@ -74,6 +75,7 @@ Other nn ops: \cr
 \code{\link{op_softsign}()} \cr
 \code{\link{op_sparse_categorical_crossentropy}()} \cr
 \code{\link{op_sparse_plus}()} \cr
+\code{\link{op_sparse_sigmoid}()} \cr
 \code{\link{op_sparsemax}()} \cr
 \code{\link{op_squareplus}()} \cr
 \code{\link{op_tanh_shrink}()} \cr
@@ -84,6 +86,7 @@ Other ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -102,6 +105,7 @@ Other ops: \cr
 \code{\link{op_associative_scan}()} \cr
 \code{\link{op_average}()} \cr
 \code{\link{op_average_pool}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_batch_normalization}()} \cr
 \code{\link{op_binary_crossentropy}()} \cr
 \code{\link{op_bincount}()} \cr
@@ -112,9 +116,11 @@ Other ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
 \code{\link{op_cast}()} \cr
 \code{\link{op_categorical_crossentropy}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_celu}()} \cr
 \code{\link{op_cholesky}()} \cr
@@ -127,6 +133,7 @@ Other ops: \cr
 \code{\link{op_convert_to_numpy}()} \cr
 \code{\link{op_convert_to_tensor}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -137,6 +144,7 @@ Other ops: \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
 \code{\link{op_custom_gradient}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_depthwise_conv}()} \cr
 \code{\link{op_det}()} \cr
 \code{\link{op_diag}()} \cr
@@ -176,10 +184,13 @@ Other ops: \cr
 \code{\link{op_glu}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
 \code{\link{op_hard_shrink}()} \cr
 \code{\link{op_hard_sigmoid}()} \cr
 \code{\link{op_hard_silu}()} \cr
 \code{\link{op_hard_tanh}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -206,6 +217,8 @@ Other ops: \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
 \code{\link{op_istft}()} \cr
+\code{\link{op_kaiser}()} \cr
+\code{\link{op_layer_normalization}()} \cr
 \code{\link{op_leaky_relu}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
@@ -304,6 +317,7 @@ Other ops: \cr
 \code{\link{op_sort}()} \cr
 \code{\link{op_sparse_categorical_crossentropy}()} \cr
 \code{\link{op_sparse_plus}()} \cr
+\code{\link{op_sparse_sigmoid}()} \cr
 \code{\link{op_sparsemax}()} \cr
 \code{\link{op_split}()} \cr
 \code{\link{op_sqrt}()} \cr
@@ -341,6 +355,8 @@ Other ops: \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
 \code{\link{op_vectorized_map}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_while_loop}()} \cr
diff --git a/man/op_logaddexp.Rd b/man/op_logaddexp.Rd
index e9e7485cc..6295d34ab 100644
--- a/man/op_logaddexp.Rd
+++ b/man/op_logaddexp.Rd
@@ -27,6 +27,7 @@ Other numpy ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -43,6 +44,7 @@ Other numpy ops: \cr
 \code{\link{op_argsort}()} \cr
 \code{\link{op_array}()} \cr
 \code{\link{op_average}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_bincount}()} \cr
 \code{\link{op_bitwise_and}()} \cr
 \code{\link{op_bitwise_invert}()} \cr
@@ -51,12 +53,15 @@ Other numpy ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_clip}()} \cr
 \code{\link{op_concatenate}()} \cr
 \code{\link{op_conj}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -65,6 +70,7 @@ Other numpy ops: \cr
 \code{\link{op_ctc_decode}()} \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_diag}()} \cr
 \code{\link{op_diagflat}()} \cr
 \code{\link{op_diagonal}()} \cr
@@ -89,6 +95,9 @@ Other numpy ops: \cr
 \code{\link{op_get_item}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -98,6 +107,7 @@ Other numpy ops: \cr
 \code{\link{op_isfinite}()} \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
+\code{\link{op_kaiser}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
 \code{\link{op_less_equal}()} \cr
@@ -177,6 +187,8 @@ Other numpy ops: \cr
 \code{\link{op_var}()} \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_zeros}()} \cr
@@ -186,6 +198,7 @@ Other ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -204,6 +217,7 @@ Other ops: \cr
 \code{\link{op_associative_scan}()} \cr
 \code{\link{op_average}()} \cr
 \code{\link{op_average_pool}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_batch_normalization}()} \cr
 \code{\link{op_binary_crossentropy}()} \cr
 \code{\link{op_bincount}()} \cr
@@ -214,9 +228,11 @@ Other ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
 \code{\link{op_cast}()} \cr
 \code{\link{op_categorical_crossentropy}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_celu}()} \cr
 \code{\link{op_cholesky}()} \cr
@@ -229,6 +245,7 @@ Other ops: \cr
 \code{\link{op_convert_to_numpy}()} \cr
 \code{\link{op_convert_to_tensor}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -239,6 +256,7 @@ Other ops: \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
 \code{\link{op_custom_gradient}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_depthwise_conv}()} \cr
 \code{\link{op_det}()} \cr
 \code{\link{op_diag}()} \cr
@@ -278,10 +296,13 @@ Other ops: \cr
 \code{\link{op_glu}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
 \code{\link{op_hard_shrink}()} \cr
 \code{\link{op_hard_sigmoid}()} \cr
 \code{\link{op_hard_silu}()} \cr
 \code{\link{op_hard_tanh}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -308,6 +329,8 @@ Other ops: \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
 \code{\link{op_istft}()} \cr
+\code{\link{op_kaiser}()} \cr
+\code{\link{op_layer_normalization}()} \cr
 \code{\link{op_leaky_relu}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
@@ -406,6 +429,7 @@ Other ops: \cr
 \code{\link{op_sort}()} \cr
 \code{\link{op_sparse_categorical_crossentropy}()} \cr
 \code{\link{op_sparse_plus}()} \cr
+\code{\link{op_sparse_sigmoid}()} \cr
 \code{\link{op_sparsemax}()} \cr
 \code{\link{op_split}()} \cr
 \code{\link{op_sqrt}()} \cr
@@ -443,6 +467,8 @@ Other ops: \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
 \code{\link{op_vectorized_map}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_while_loop}()} \cr
diff --git a/man/op_logdet.Rd b/man/op_logdet.Rd
index 4d85ba94e..5817cbb60 100644
--- a/man/op_logdet.Rd
+++ b/man/op_logdet.Rd
@@ -20,6 +20,7 @@ Other numpy ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -36,6 +37,7 @@ Other numpy ops: \cr
 \code{\link{op_argsort}()} \cr
 \code{\link{op_array}()} \cr
 \code{\link{op_average}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_bincount}()} \cr
 \code{\link{op_bitwise_and}()} \cr
 \code{\link{op_bitwise_invert}()} \cr
@@ -44,12 +46,15 @@ Other numpy ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_clip}()} \cr
 \code{\link{op_concatenate}()} \cr
 \code{\link{op_conj}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -58,6 +63,7 @@ Other numpy ops: \cr
 \code{\link{op_ctc_decode}()} \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_diag}()} \cr
 \code{\link{op_diagflat}()} \cr
 \code{\link{op_diagonal}()} \cr
@@ -82,6 +88,9 @@ Other numpy ops: \cr
 \code{\link{op_get_item}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -91,6 +100,7 @@ Other numpy ops: \cr
 \code{\link{op_isfinite}()} \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
+\code{\link{op_kaiser}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
 \code{\link{op_less_equal}()} \cr
@@ -170,6 +180,8 @@ Other numpy ops: \cr
 \code{\link{op_var}()} \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_zeros}()} \cr
@@ -179,6 +191,7 @@ Other ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -197,6 +210,7 @@ Other ops: \cr
 \code{\link{op_associative_scan}()} \cr
 \code{\link{op_average}()} \cr
 \code{\link{op_average_pool}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_batch_normalization}()} \cr
 \code{\link{op_binary_crossentropy}()} \cr
 \code{\link{op_bincount}()} \cr
@@ -207,9 +221,11 @@ Other ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
 \code{\link{op_cast}()} \cr
 \code{\link{op_categorical_crossentropy}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_celu}()} \cr
 \code{\link{op_cholesky}()} \cr
@@ -222,6 +238,7 @@ Other ops: \cr
 \code{\link{op_convert_to_numpy}()} \cr
 \code{\link{op_convert_to_tensor}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -232,6 +249,7 @@ Other ops: \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
 \code{\link{op_custom_gradient}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_depthwise_conv}()} \cr
 \code{\link{op_det}()} \cr
 \code{\link{op_diag}()} \cr
@@ -271,10 +289,13 @@ Other ops: \cr
 \code{\link{op_glu}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
 \code{\link{op_hard_shrink}()} \cr
 \code{\link{op_hard_sigmoid}()} \cr
 \code{\link{op_hard_silu}()} \cr
 \code{\link{op_hard_tanh}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -301,6 +322,8 @@ Other ops: \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
 \code{\link{op_istft}()} \cr
+\code{\link{op_kaiser}()} \cr
+\code{\link{op_layer_normalization}()} \cr
 \code{\link{op_leaky_relu}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
@@ -399,6 +422,7 @@ Other ops: \cr
 \code{\link{op_sort}()} \cr
 \code{\link{op_sparse_categorical_crossentropy}()} \cr
 \code{\link{op_sparse_plus}()} \cr
+\code{\link{op_sparse_sigmoid}()} \cr
 \code{\link{op_sparsemax}()} \cr
 \code{\link{op_split}()} \cr
 \code{\link{op_sqrt}()} \cr
@@ -436,6 +460,8 @@ Other ops: \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
 \code{\link{op_vectorized_map}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_while_loop}()} \cr
diff --git a/man/op_logical_and.Rd b/man/op_logical_and.Rd
index d182f8b5c..6c942d06b 100644
--- a/man/op_logical_and.Rd
+++ b/man/op_logical_and.Rd
@@ -29,6 +29,7 @@ Other numpy ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -45,6 +46,7 @@ Other numpy ops: \cr
 \code{\link{op_argsort}()} \cr
 \code{\link{op_array}()} \cr
 \code{\link{op_average}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_bincount}()} \cr
 \code{\link{op_bitwise_and}()} \cr
 \code{\link{op_bitwise_invert}()} \cr
@@ -53,12 +55,15 @@ Other numpy ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_clip}()} \cr
 \code{\link{op_concatenate}()} \cr
 \code{\link{op_conj}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -67,6 +72,7 @@ Other numpy ops: \cr
 \code{\link{op_ctc_decode}()} \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_diag}()} \cr
 \code{\link{op_diagflat}()} \cr
 \code{\link{op_diagonal}()} \cr
@@ -91,6 +97,9 @@ Other numpy ops: \cr
 \code{\link{op_get_item}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -100,6 +109,7 @@ Other numpy ops: \cr
 \code{\link{op_isfinite}()} \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
+\code{\link{op_kaiser}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
 \code{\link{op_less_equal}()} \cr
@@ -179,6 +189,8 @@ Other numpy ops: \cr
 \code{\link{op_var}()} \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_zeros}()} \cr
@@ -188,6 +200,7 @@ Other ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -206,6 +219,7 @@ Other ops: \cr
 \code{\link{op_associative_scan}()} \cr
 \code{\link{op_average}()} \cr
 \code{\link{op_average_pool}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_batch_normalization}()} \cr
 \code{\link{op_binary_crossentropy}()} \cr
 \code{\link{op_bincount}()} \cr
@@ -216,9 +230,11 @@ Other ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
 \code{\link{op_cast}()} \cr
 \code{\link{op_categorical_crossentropy}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_celu}()} \cr
 \code{\link{op_cholesky}()} \cr
@@ -231,6 +247,7 @@ Other ops: \cr
 \code{\link{op_convert_to_numpy}()} \cr
 \code{\link{op_convert_to_tensor}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -241,6 +258,7 @@ Other ops: \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
 \code{\link{op_custom_gradient}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_depthwise_conv}()} \cr
 \code{\link{op_det}()} \cr
 \code{\link{op_diag}()} \cr
@@ -280,10 +298,13 @@ Other ops: \cr
 \code{\link{op_glu}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
 \code{\link{op_hard_shrink}()} \cr
 \code{\link{op_hard_sigmoid}()} \cr
 \code{\link{op_hard_silu}()} \cr
 \code{\link{op_hard_tanh}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -310,6 +331,8 @@ Other ops: \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
 \code{\link{op_istft}()} \cr
+\code{\link{op_kaiser}()} \cr
+\code{\link{op_layer_normalization}()} \cr
 \code{\link{op_leaky_relu}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
@@ -408,6 +431,7 @@ Other ops: \cr
 \code{\link{op_sort}()} \cr
 \code{\link{op_sparse_categorical_crossentropy}()} \cr
 \code{\link{op_sparse_plus}()} \cr
+\code{\link{op_sparse_sigmoid}()} \cr
 \code{\link{op_sparsemax}()} \cr
 \code{\link{op_split}()} \cr
 \code{\link{op_sqrt}()} \cr
@@ -445,6 +469,8 @@ Other ops: \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
 \code{\link{op_vectorized_map}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_while_loop}()} \cr
diff --git a/man/op_logical_not.Rd b/man/op_logical_not.Rd
index 64b2b2813..043eb0fbd 100644
--- a/man/op_logical_not.Rd
+++ b/man/op_logical_not.Rd
@@ -26,6 +26,7 @@ Other numpy ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -42,6 +43,7 @@ Other numpy ops: \cr
 \code{\link{op_argsort}()} \cr
 \code{\link{op_array}()} \cr
 \code{\link{op_average}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_bincount}()} \cr
 \code{\link{op_bitwise_and}()} \cr
 \code{\link{op_bitwise_invert}()} \cr
@@ -50,12 +52,15 @@ Other numpy ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_clip}()} \cr
 \code{\link{op_concatenate}()} \cr
 \code{\link{op_conj}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -64,6 +69,7 @@ Other numpy ops: \cr
 \code{\link{op_ctc_decode}()} \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_diag}()} \cr
 \code{\link{op_diagflat}()} \cr
 \code{\link{op_diagonal}()} \cr
@@ -88,6 +94,9 @@ Other numpy ops: \cr
 \code{\link{op_get_item}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -97,6 +106,7 @@ Other numpy ops: \cr
 \code{\link{op_isfinite}()} \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
+\code{\link{op_kaiser}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
 \code{\link{op_less_equal}()} \cr
@@ -176,6 +186,8 @@ Other numpy ops: \cr
 \code{\link{op_var}()} \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_zeros}()} \cr
@@ -185,6 +197,7 @@ Other ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -203,6 +216,7 @@ Other ops: \cr
 \code{\link{op_associative_scan}()} \cr
 \code{\link{op_average}()} \cr
 \code{\link{op_average_pool}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_batch_normalization}()} \cr
 \code{\link{op_binary_crossentropy}()} \cr
 \code{\link{op_bincount}()} \cr
@@ -213,9 +227,11 @@ Other ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
 \code{\link{op_cast}()} \cr
 \code{\link{op_categorical_crossentropy}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_celu}()} \cr
 \code{\link{op_cholesky}()} \cr
@@ -228,6 +244,7 @@ Other ops: \cr
 \code{\link{op_convert_to_numpy}()} \cr
 \code{\link{op_convert_to_tensor}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -238,6 +255,7 @@ Other ops: \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
 \code{\link{op_custom_gradient}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_depthwise_conv}()} \cr
 \code{\link{op_det}()} \cr
 \code{\link{op_diag}()} \cr
@@ -277,10 +295,13 @@ Other ops: \cr
 \code{\link{op_glu}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
 \code{\link{op_hard_shrink}()} \cr
 \code{\link{op_hard_sigmoid}()} \cr
 \code{\link{op_hard_silu}()} \cr
 \code{\link{op_hard_tanh}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -307,6 +328,8 @@ Other ops: \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
 \code{\link{op_istft}()} \cr
+\code{\link{op_kaiser}()} \cr
+\code{\link{op_layer_normalization}()} \cr
 \code{\link{op_leaky_relu}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
@@ -405,6 +428,7 @@ Other ops: \cr
 \code{\link{op_sort}()} \cr
 \code{\link{op_sparse_categorical_crossentropy}()} \cr
 \code{\link{op_sparse_plus}()} \cr
+\code{\link{op_sparse_sigmoid}()} \cr
 \code{\link{op_sparsemax}()} \cr
 \code{\link{op_split}()} \cr
 \code{\link{op_sqrt}()} \cr
@@ -442,6 +466,8 @@ Other ops: \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
 \code{\link{op_vectorized_map}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_while_loop}()} \cr
diff --git a/man/op_logical_or.Rd b/man/op_logical_or.Rd
index 007f79dcc..b4922c960 100644
--- a/man/op_logical_or.Rd
+++ b/man/op_logical_or.Rd
@@ -28,6 +28,7 @@ Other numpy ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -44,6 +45,7 @@ Other numpy ops: \cr
 \code{\link{op_argsort}()} \cr
 \code{\link{op_array}()} \cr
 \code{\link{op_average}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_bincount}()} \cr
 \code{\link{op_bitwise_and}()} \cr
 \code{\link{op_bitwise_invert}()} \cr
@@ -52,12 +54,15 @@ Other numpy ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_clip}()} \cr
 \code{\link{op_concatenate}()} \cr
 \code{\link{op_conj}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -66,6 +71,7 @@ Other numpy ops: \cr
 \code{\link{op_ctc_decode}()} \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_diag}()} \cr
 \code{\link{op_diagflat}()} \cr
 \code{\link{op_diagonal}()} \cr
@@ -90,6 +96,9 @@ Other numpy ops: \cr
 \code{\link{op_get_item}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -99,6 +108,7 @@ Other numpy ops: \cr
 \code{\link{op_isfinite}()} \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
+\code{\link{op_kaiser}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
 \code{\link{op_less_equal}()} \cr
@@ -178,6 +188,8 @@ Other numpy ops: \cr
 \code{\link{op_var}()} \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_zeros}()} \cr
@@ -187,6 +199,7 @@ Other ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -205,6 +218,7 @@ Other ops: \cr
 \code{\link{op_associative_scan}()} \cr
 \code{\link{op_average}()} \cr
 \code{\link{op_average_pool}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_batch_normalization}()} \cr
 \code{\link{op_binary_crossentropy}()} \cr
 \code{\link{op_bincount}()} \cr
@@ -215,9 +229,11 @@ Other ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
 \code{\link{op_cast}()} \cr
 \code{\link{op_categorical_crossentropy}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_celu}()} \cr
 \code{\link{op_cholesky}()} \cr
@@ -230,6 +246,7 @@ Other ops: \cr
 \code{\link{op_convert_to_numpy}()} \cr
 \code{\link{op_convert_to_tensor}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -240,6 +257,7 @@ Other ops: \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
 \code{\link{op_custom_gradient}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_depthwise_conv}()} \cr
 \code{\link{op_det}()} \cr
 \code{\link{op_diag}()} \cr
@@ -279,10 +297,13 @@ Other ops: \cr
 \code{\link{op_glu}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
 \code{\link{op_hard_shrink}()} \cr
 \code{\link{op_hard_sigmoid}()} \cr
 \code{\link{op_hard_silu}()} \cr
 \code{\link{op_hard_tanh}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -309,6 +330,8 @@ Other ops: \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
 \code{\link{op_istft}()} \cr
+\code{\link{op_kaiser}()} \cr
+\code{\link{op_layer_normalization}()} \cr
 \code{\link{op_leaky_relu}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
@@ -407,6 +430,7 @@ Other ops: \cr
 \code{\link{op_sort}()} \cr
 \code{\link{op_sparse_categorical_crossentropy}()} \cr
 \code{\link{op_sparse_plus}()} \cr
+\code{\link{op_sparse_sigmoid}()} \cr
 \code{\link{op_sparsemax}()} \cr
 \code{\link{op_split}()} \cr
 \code{\link{op_sqrt}()} \cr
@@ -444,6 +468,8 @@ Other ops: \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
 \code{\link{op_vectorized_map}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_while_loop}()} \cr
diff --git a/man/op_logical_xor.Rd b/man/op_logical_xor.Rd
index ecc2a15ba..c8f6679c5 100644
--- a/man/op_logical_xor.Rd
+++ b/man/op_logical_xor.Rd
@@ -22,6 +22,7 @@ Other numpy ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -38,6 +39,7 @@ Other numpy ops: \cr
 \code{\link{op_argsort}()} \cr
 \code{\link{op_array}()} \cr
 \code{\link{op_average}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_bincount}()} \cr
 \code{\link{op_bitwise_and}()} \cr
 \code{\link{op_bitwise_invert}()} \cr
@@ -46,12 +48,15 @@ Other numpy ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_clip}()} \cr
 \code{\link{op_concatenate}()} \cr
 \code{\link{op_conj}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -60,6 +65,7 @@ Other numpy ops: \cr
 \code{\link{op_ctc_decode}()} \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_diag}()} \cr
 \code{\link{op_diagflat}()} \cr
 \code{\link{op_diagonal}()} \cr
@@ -84,6 +90,9 @@ Other numpy ops: \cr
 \code{\link{op_get_item}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -93,6 +102,7 @@ Other numpy ops: \cr
 \code{\link{op_isfinite}()} \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
+\code{\link{op_kaiser}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
 \code{\link{op_less_equal}()} \cr
@@ -172,6 +182,8 @@ Other numpy ops: \cr
 \code{\link{op_var}()} \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_zeros}()} \cr
@@ -181,6 +193,7 @@ Other ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -199,6 +212,7 @@ Other ops: \cr
 \code{\link{op_associative_scan}()} \cr
 \code{\link{op_average}()} \cr
 \code{\link{op_average_pool}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_batch_normalization}()} \cr
 \code{\link{op_binary_crossentropy}()} \cr
 \code{\link{op_bincount}()} \cr
@@ -209,9 +223,11 @@ Other ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
 \code{\link{op_cast}()} \cr
 \code{\link{op_categorical_crossentropy}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_celu}()} \cr
 \code{\link{op_cholesky}()} \cr
@@ -224,6 +240,7 @@ Other ops: \cr
 \code{\link{op_convert_to_numpy}()} \cr
 \code{\link{op_convert_to_tensor}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -234,6 +251,7 @@ Other ops: \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
 \code{\link{op_custom_gradient}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_depthwise_conv}()} \cr
 \code{\link{op_det}()} \cr
 \code{\link{op_diag}()} \cr
@@ -273,10 +291,13 @@ Other ops: \cr
 \code{\link{op_glu}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
 \code{\link{op_hard_shrink}()} \cr
 \code{\link{op_hard_sigmoid}()} \cr
 \code{\link{op_hard_silu}()} \cr
 \code{\link{op_hard_tanh}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -303,6 +324,8 @@ Other ops: \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
 \code{\link{op_istft}()} \cr
+\code{\link{op_kaiser}()} \cr
+\code{\link{op_layer_normalization}()} \cr
 \code{\link{op_leaky_relu}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
@@ -401,6 +424,7 @@ Other ops: \cr
 \code{\link{op_sort}()} \cr
 \code{\link{op_sparse_categorical_crossentropy}()} \cr
 \code{\link{op_sparse_plus}()} \cr
+\code{\link{op_sparse_sigmoid}()} \cr
 \code{\link{op_sparsemax}()} \cr
 \code{\link{op_split}()} \cr
 \code{\link{op_sqrt}()} \cr
@@ -438,6 +462,8 @@ Other ops: \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
 \code{\link{op_vectorized_map}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_while_loop}()} \cr
diff --git a/man/op_logspace.Rd b/man/op_logspace.Rd
index 8222f0133..c3c0ae520 100644
--- a/man/op_logspace.Rd
+++ b/man/op_logspace.Rd
@@ -54,6 +54,7 @@ Other numpy ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -70,6 +71,7 @@ Other numpy ops: \cr
 \code{\link{op_argsort}()} \cr
 \code{\link{op_array}()} \cr
 \code{\link{op_average}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_bincount}()} \cr
 \code{\link{op_bitwise_and}()} \cr
 \code{\link{op_bitwise_invert}()} \cr
@@ -78,12 +80,15 @@ Other numpy ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_clip}()} \cr
 \code{\link{op_concatenate}()} \cr
 \code{\link{op_conj}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -92,6 +97,7 @@ Other numpy ops: \cr
 \code{\link{op_ctc_decode}()} \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_diag}()} \cr
 \code{\link{op_diagflat}()} \cr
 \code{\link{op_diagonal}()} \cr
@@ -116,6 +122,9 @@ Other numpy ops: \cr
 \code{\link{op_get_item}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -125,6 +134,7 @@ Other numpy ops: \cr
 \code{\link{op_isfinite}()} \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
+\code{\link{op_kaiser}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
 \code{\link{op_less_equal}()} \cr
@@ -204,6 +214,8 @@ Other numpy ops: \cr
 \code{\link{op_var}()} \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_zeros}()} \cr
@@ -213,6 +225,7 @@ Other ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -231,6 +244,7 @@ Other ops: \cr
 \code{\link{op_associative_scan}()} \cr
 \code{\link{op_average}()} \cr
 \code{\link{op_average_pool}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_batch_normalization}()} \cr
 \code{\link{op_binary_crossentropy}()} \cr
 \code{\link{op_bincount}()} \cr
@@ -241,9 +255,11 @@ Other ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
 \code{\link{op_cast}()} \cr
 \code{\link{op_categorical_crossentropy}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_celu}()} \cr
 \code{\link{op_cholesky}()} \cr
@@ -256,6 +272,7 @@ Other ops: \cr
 \code{\link{op_convert_to_numpy}()} \cr
 \code{\link{op_convert_to_tensor}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -266,6 +283,7 @@ Other ops: \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
 \code{\link{op_custom_gradient}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_depthwise_conv}()} \cr
 \code{\link{op_det}()} \cr
 \code{\link{op_diag}()} \cr
@@ -305,10 +323,13 @@ Other ops: \cr
 \code{\link{op_glu}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
 \code{\link{op_hard_shrink}()} \cr
 \code{\link{op_hard_sigmoid}()} \cr
 \code{\link{op_hard_silu}()} \cr
 \code{\link{op_hard_tanh}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -335,6 +356,8 @@ Other ops: \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
 \code{\link{op_istft}()} \cr
+\code{\link{op_kaiser}()} \cr
+\code{\link{op_layer_normalization}()} \cr
 \code{\link{op_leaky_relu}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
@@ -433,6 +456,7 @@ Other ops: \cr
 \code{\link{op_sort}()} \cr
 \code{\link{op_sparse_categorical_crossentropy}()} \cr
 \code{\link{op_sparse_plus}()} \cr
+\code{\link{op_sparse_sigmoid}()} \cr
 \code{\link{op_sparsemax}()} \cr
 \code{\link{op_split}()} \cr
 \code{\link{op_sqrt}()} \cr
@@ -470,6 +494,8 @@ Other ops: \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
 \code{\link{op_vectorized_map}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_while_loop}()} \cr
diff --git a/man/op_logsumexp.Rd b/man/op_logsumexp.Rd
index 9de9c833b..13179d5ca 100644
--- a/man/op_logsumexp.Rd
+++ b/man/op_logsumexp.Rd
@@ -61,6 +61,7 @@ Other ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -79,6 +80,7 @@ Other ops: \cr
 \code{\link{op_associative_scan}()} \cr
 \code{\link{op_average}()} \cr
 \code{\link{op_average_pool}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_batch_normalization}()} \cr
 \code{\link{op_binary_crossentropy}()} \cr
 \code{\link{op_bincount}()} \cr
@@ -89,9 +91,11 @@ Other ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
 \code{\link{op_cast}()} \cr
 \code{\link{op_categorical_crossentropy}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_celu}()} \cr
 \code{\link{op_cholesky}()} \cr
@@ -104,6 +108,7 @@ Other ops: \cr
 \code{\link{op_convert_to_numpy}()} \cr
 \code{\link{op_convert_to_tensor}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -114,6 +119,7 @@ Other ops: \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
 \code{\link{op_custom_gradient}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_depthwise_conv}()} \cr
 \code{\link{op_det}()} \cr
 \code{\link{op_diag}()} \cr
@@ -153,10 +159,13 @@ Other ops: \cr
 \code{\link{op_glu}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
 \code{\link{op_hard_shrink}()} \cr
 \code{\link{op_hard_sigmoid}()} \cr
 \code{\link{op_hard_silu}()} \cr
 \code{\link{op_hard_tanh}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -183,6 +192,8 @@ Other ops: \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
 \code{\link{op_istft}()} \cr
+\code{\link{op_kaiser}()} \cr
+\code{\link{op_layer_normalization}()} \cr
 \code{\link{op_leaky_relu}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
@@ -281,6 +292,7 @@ Other ops: \cr
 \code{\link{op_sort}()} \cr
 \code{\link{op_sparse_categorical_crossentropy}()} \cr
 \code{\link{op_sparse_plus}()} \cr
+\code{\link{op_sparse_sigmoid}()} \cr
 \code{\link{op_sparsemax}()} \cr
 \code{\link{op_split}()} \cr
 \code{\link{op_sqrt}()} \cr
@@ -318,6 +330,8 @@ Other ops: \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
 \code{\link{op_vectorized_map}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_while_loop}()} \cr
diff --git a/man/op_lstsq.Rd b/man/op_lstsq.Rd
index 8dec986a8..a920de4c9 100644
--- a/man/op_lstsq.Rd
+++ b/man/op_lstsq.Rd
@@ -61,6 +61,7 @@ Other numpy ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -77,6 +78,7 @@ Other numpy ops: \cr
 \code{\link{op_argsort}()} \cr
 \code{\link{op_array}()} \cr
 \code{\link{op_average}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_bincount}()} \cr
 \code{\link{op_bitwise_and}()} \cr
 \code{\link{op_bitwise_invert}()} \cr
@@ -85,12 +87,15 @@ Other numpy ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_clip}()} \cr
 \code{\link{op_concatenate}()} \cr
 \code{\link{op_conj}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -99,6 +104,7 @@ Other numpy ops: \cr
 \code{\link{op_ctc_decode}()} \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_diag}()} \cr
 \code{\link{op_diagflat}()} \cr
 \code{\link{op_diagonal}()} \cr
@@ -123,6 +129,9 @@ Other numpy ops: \cr
 \code{\link{op_get_item}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -132,6 +141,7 @@ Other numpy ops: \cr
 \code{\link{op_isfinite}()} \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
+\code{\link{op_kaiser}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
 \code{\link{op_less_equal}()} \cr
@@ -211,6 +221,8 @@ Other numpy ops: \cr
 \code{\link{op_var}()} \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_zeros}()} \cr
@@ -220,6 +232,7 @@ Other ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -238,6 +251,7 @@ Other ops: \cr
 \code{\link{op_associative_scan}()} \cr
 \code{\link{op_average}()} \cr
 \code{\link{op_average_pool}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_batch_normalization}()} \cr
 \code{\link{op_binary_crossentropy}()} \cr
 \code{\link{op_bincount}()} \cr
@@ -248,9 +262,11 @@ Other ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
 \code{\link{op_cast}()} \cr
 \code{\link{op_categorical_crossentropy}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_celu}()} \cr
 \code{\link{op_cholesky}()} \cr
@@ -263,6 +279,7 @@ Other ops: \cr
 \code{\link{op_convert_to_numpy}()} \cr
 \code{\link{op_convert_to_tensor}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -273,6 +290,7 @@ Other ops: \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
 \code{\link{op_custom_gradient}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_depthwise_conv}()} \cr
 \code{\link{op_det}()} \cr
 \code{\link{op_diag}()} \cr
@@ -312,10 +330,13 @@ Other ops: \cr
 \code{\link{op_glu}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
 \code{\link{op_hard_shrink}()} \cr
 \code{\link{op_hard_sigmoid}()} \cr
 \code{\link{op_hard_silu}()} \cr
 \code{\link{op_hard_tanh}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -342,6 +363,8 @@ Other ops: \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
 \code{\link{op_istft}()} \cr
+\code{\link{op_kaiser}()} \cr
+\code{\link{op_layer_normalization}()} \cr
 \code{\link{op_leaky_relu}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
@@ -440,6 +463,7 @@ Other ops: \cr
 \code{\link{op_sort}()} \cr
 \code{\link{op_sparse_categorical_crossentropy}()} \cr
 \code{\link{op_sparse_plus}()} \cr
+\code{\link{op_sparse_sigmoid}()} \cr
 \code{\link{op_sparsemax}()} \cr
 \code{\link{op_split}()} \cr
 \code{\link{op_sqrt}()} \cr
@@ -477,6 +501,8 @@ Other ops: \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
 \code{\link{op_vectorized_map}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_while_loop}()} \cr
diff --git a/man/op_lu_factor.Rd b/man/op_lu_factor.Rd
index 5d5fcc1b4..d121072e9 100644
--- a/man/op_lu_factor.Rd
+++ b/man/op_lu_factor.Rd
@@ -34,6 +34,7 @@ Other ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -52,6 +53,7 @@ Other ops: \cr
 \code{\link{op_associative_scan}()} \cr
 \code{\link{op_average}()} \cr
 \code{\link{op_average_pool}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_batch_normalization}()} \cr
 \code{\link{op_binary_crossentropy}()} \cr
 \code{\link{op_bincount}()} \cr
@@ -62,9 +64,11 @@ Other ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
 \code{\link{op_cast}()} \cr
 \code{\link{op_categorical_crossentropy}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_celu}()} \cr
 \code{\link{op_cholesky}()} \cr
@@ -77,6 +81,7 @@ Other ops: \cr
 \code{\link{op_convert_to_numpy}()} \cr
 \code{\link{op_convert_to_tensor}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -87,6 +92,7 @@ Other ops: \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
 \code{\link{op_custom_gradient}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_depthwise_conv}()} \cr
 \code{\link{op_det}()} \cr
 \code{\link{op_diag}()} \cr
@@ -126,10 +132,13 @@ Other ops: \cr
 \code{\link{op_glu}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
 \code{\link{op_hard_shrink}()} \cr
 \code{\link{op_hard_sigmoid}()} \cr
 \code{\link{op_hard_silu}()} \cr
 \code{\link{op_hard_tanh}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -156,6 +165,8 @@ Other ops: \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
 \code{\link{op_istft}()} \cr
+\code{\link{op_kaiser}()} \cr
+\code{\link{op_layer_normalization}()} \cr
 \code{\link{op_leaky_relu}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
@@ -254,6 +265,7 @@ Other ops: \cr
 \code{\link{op_sort}()} \cr
 \code{\link{op_sparse_categorical_crossentropy}()} \cr
 \code{\link{op_sparse_plus}()} \cr
+\code{\link{op_sparse_sigmoid}()} \cr
 \code{\link{op_sparsemax}()} \cr
 \code{\link{op_split}()} \cr
 \code{\link{op_sqrt}()} \cr
@@ -291,6 +303,8 @@ Other ops: \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
 \code{\link{op_vectorized_map}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_while_loop}()} \cr
diff --git a/man/op_map.Rd b/man/op_map.Rd
index 86fb1c888..70456bfe5 100644
--- a/man/op_map.Rd
+++ b/man/op_map.Rd
@@ -90,6 +90,7 @@ Other ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -108,6 +109,7 @@ Other ops: \cr
 \code{\link{op_associative_scan}()} \cr
 \code{\link{op_average}()} \cr
 \code{\link{op_average_pool}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_batch_normalization}()} \cr
 \code{\link{op_binary_crossentropy}()} \cr
 \code{\link{op_bincount}()} \cr
@@ -118,9 +120,11 @@ Other ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
 \code{\link{op_cast}()} \cr
 \code{\link{op_categorical_crossentropy}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_celu}()} \cr
 \code{\link{op_cholesky}()} \cr
@@ -133,6 +137,7 @@ Other ops: \cr
 \code{\link{op_convert_to_numpy}()} \cr
 \code{\link{op_convert_to_tensor}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -143,6 +148,7 @@ Other ops: \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
 \code{\link{op_custom_gradient}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_depthwise_conv}()} \cr
 \code{\link{op_det}()} \cr
 \code{\link{op_diag}()} \cr
@@ -182,10 +188,13 @@ Other ops: \cr
 \code{\link{op_glu}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
 \code{\link{op_hard_shrink}()} \cr
 \code{\link{op_hard_sigmoid}()} \cr
 \code{\link{op_hard_silu}()} \cr
 \code{\link{op_hard_tanh}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -212,6 +221,8 @@ Other ops: \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
 \code{\link{op_istft}()} \cr
+\code{\link{op_kaiser}()} \cr
+\code{\link{op_layer_normalization}()} \cr
 \code{\link{op_leaky_relu}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
@@ -310,6 +321,7 @@ Other ops: \cr
 \code{\link{op_sort}()} \cr
 \code{\link{op_sparse_categorical_crossentropy}()} \cr
 \code{\link{op_sparse_plus}()} \cr
+\code{\link{op_sparse_sigmoid}()} \cr
 \code{\link{op_sparsemax}()} \cr
 \code{\link{op_split}()} \cr
 \code{\link{op_sqrt}()} \cr
@@ -347,6 +359,8 @@ Other ops: \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
 \code{\link{op_vectorized_map}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_while_loop}()} \cr
diff --git a/man/op_matmul.Rd b/man/op_matmul.Rd
index 7281b89f0..24f450369 100644
--- a/man/op_matmul.Rd
+++ b/man/op_matmul.Rd
@@ -35,6 +35,7 @@ Other numpy ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -51,6 +52,7 @@ Other numpy ops: \cr
 \code{\link{op_argsort}()} \cr
 \code{\link{op_array}()} \cr
 \code{\link{op_average}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_bincount}()} \cr
 \code{\link{op_bitwise_and}()} \cr
 \code{\link{op_bitwise_invert}()} \cr
@@ -59,12 +61,15 @@ Other numpy ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_clip}()} \cr
 \code{\link{op_concatenate}()} \cr
 \code{\link{op_conj}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -73,6 +78,7 @@ Other numpy ops: \cr
 \code{\link{op_ctc_decode}()} \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_diag}()} \cr
 \code{\link{op_diagflat}()} \cr
 \code{\link{op_diagonal}()} \cr
@@ -97,6 +103,9 @@ Other numpy ops: \cr
 \code{\link{op_get_item}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -106,6 +115,7 @@ Other numpy ops: \cr
 \code{\link{op_isfinite}()} \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
+\code{\link{op_kaiser}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
 \code{\link{op_less_equal}()} \cr
@@ -185,6 +195,8 @@ Other numpy ops: \cr
 \code{\link{op_var}()} \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_zeros}()} \cr
@@ -194,6 +206,7 @@ Other ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -212,6 +225,7 @@ Other ops: \cr
 \code{\link{op_associative_scan}()} \cr
 \code{\link{op_average}()} \cr
 \code{\link{op_average_pool}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_batch_normalization}()} \cr
 \code{\link{op_binary_crossentropy}()} \cr
 \code{\link{op_bincount}()} \cr
@@ -222,9 +236,11 @@ Other ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
 \code{\link{op_cast}()} \cr
 \code{\link{op_categorical_crossentropy}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_celu}()} \cr
 \code{\link{op_cholesky}()} \cr
@@ -237,6 +253,7 @@ Other ops: \cr
 \code{\link{op_convert_to_numpy}()} \cr
 \code{\link{op_convert_to_tensor}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -247,6 +264,7 @@ Other ops: \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
 \code{\link{op_custom_gradient}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_depthwise_conv}()} \cr
 \code{\link{op_det}()} \cr
 \code{\link{op_diag}()} \cr
@@ -286,10 +304,13 @@ Other ops: \cr
 \code{\link{op_glu}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
 \code{\link{op_hard_shrink}()} \cr
 \code{\link{op_hard_sigmoid}()} \cr
 \code{\link{op_hard_silu}()} \cr
 \code{\link{op_hard_tanh}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -316,6 +337,8 @@ Other ops: \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
 \code{\link{op_istft}()} \cr
+\code{\link{op_kaiser}()} \cr
+\code{\link{op_layer_normalization}()} \cr
 \code{\link{op_leaky_relu}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
@@ -414,6 +437,7 @@ Other ops: \cr
 \code{\link{op_sort}()} \cr
 \code{\link{op_sparse_categorical_crossentropy}()} \cr
 \code{\link{op_sparse_plus}()} \cr
+\code{\link{op_sparse_sigmoid}()} \cr
 \code{\link{op_sparsemax}()} \cr
 \code{\link{op_split}()} \cr
 \code{\link{op_sqrt}()} \cr
@@ -451,6 +475,8 @@ Other ops: \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
 \code{\link{op_vectorized_map}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_while_loop}()} \cr
diff --git a/man/op_max.Rd b/man/op_max.Rd
index 6378d9056..996429527 100644
--- a/man/op_max.Rd
+++ b/man/op_max.Rd
@@ -65,6 +65,7 @@ Other numpy ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -81,6 +82,7 @@ Other numpy ops: \cr
 \code{\link{op_argsort}()} \cr
 \code{\link{op_array}()} \cr
 \code{\link{op_average}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_bincount}()} \cr
 \code{\link{op_bitwise_and}()} \cr
 \code{\link{op_bitwise_invert}()} \cr
@@ -89,12 +91,15 @@ Other numpy ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_clip}()} \cr
 \code{\link{op_concatenate}()} \cr
 \code{\link{op_conj}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -103,6 +108,7 @@ Other numpy ops: \cr
 \code{\link{op_ctc_decode}()} \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_diag}()} \cr
 \code{\link{op_diagflat}()} \cr
 \code{\link{op_diagonal}()} \cr
@@ -127,6 +133,9 @@ Other numpy ops: \cr
 \code{\link{op_get_item}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -136,6 +145,7 @@ Other numpy ops: \cr
 \code{\link{op_isfinite}()} \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
+\code{\link{op_kaiser}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
 \code{\link{op_less_equal}()} \cr
@@ -215,6 +225,8 @@ Other numpy ops: \cr
 \code{\link{op_var}()} \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_zeros}()} \cr
@@ -224,6 +236,7 @@ Other ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -242,6 +255,7 @@ Other ops: \cr
 \code{\link{op_associative_scan}()} \cr
 \code{\link{op_average}()} \cr
 \code{\link{op_average_pool}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_batch_normalization}()} \cr
 \code{\link{op_binary_crossentropy}()} \cr
 \code{\link{op_bincount}()} \cr
@@ -252,9 +266,11 @@ Other ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
 \code{\link{op_cast}()} \cr
 \code{\link{op_categorical_crossentropy}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_celu}()} \cr
 \code{\link{op_cholesky}()} \cr
@@ -267,6 +283,7 @@ Other ops: \cr
 \code{\link{op_convert_to_numpy}()} \cr
 \code{\link{op_convert_to_tensor}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -277,6 +294,7 @@ Other ops: \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
 \code{\link{op_custom_gradient}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_depthwise_conv}()} \cr
 \code{\link{op_det}()} \cr
 \code{\link{op_diag}()} \cr
@@ -316,10 +334,13 @@ Other ops: \cr
 \code{\link{op_glu}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
 \code{\link{op_hard_shrink}()} \cr
 \code{\link{op_hard_sigmoid}()} \cr
 \code{\link{op_hard_silu}()} \cr
 \code{\link{op_hard_tanh}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -346,6 +367,8 @@ Other ops: \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
 \code{\link{op_istft}()} \cr
+\code{\link{op_kaiser}()} \cr
+\code{\link{op_layer_normalization}()} \cr
 \code{\link{op_leaky_relu}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
@@ -444,6 +467,7 @@ Other ops: \cr
 \code{\link{op_sort}()} \cr
 \code{\link{op_sparse_categorical_crossentropy}()} \cr
 \code{\link{op_sparse_plus}()} \cr
+\code{\link{op_sparse_sigmoid}()} \cr
 \code{\link{op_sparsemax}()} \cr
 \code{\link{op_split}()} \cr
 \code{\link{op_sqrt}()} \cr
@@ -481,6 +505,8 @@ Other ops: \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
 \code{\link{op_vectorized_map}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_while_loop}()} \cr
diff --git a/man/op_max_pool.Rd b/man/op_max_pool.Rd
index 25a18efd7..475388d60 100644
--- a/man/op_max_pool.Rd
+++ b/man/op_max_pool.Rd
@@ -72,6 +72,7 @@ Other nn ops: \cr
 \code{\link{op_hard_sigmoid}()} \cr
 \code{\link{op_hard_silu}()} \cr
 \code{\link{op_hard_tanh}()} \cr
+\code{\link{op_layer_normalization}()} \cr
 \code{\link{op_leaky_relu}()} \cr
 \code{\link{op_log_sigmoid}()} \cr
 \code{\link{op_log_softmax}()} \cr
@@ -94,6 +95,7 @@ Other nn ops: \cr
 \code{\link{op_softsign}()} \cr
 \code{\link{op_sparse_categorical_crossentropy}()} \cr
 \code{\link{op_sparse_plus}()} \cr
+\code{\link{op_sparse_sigmoid}()} \cr
 \code{\link{op_sparsemax}()} \cr
 \code{\link{op_squareplus}()} \cr
 \code{\link{op_tanh_shrink}()} \cr
@@ -104,6 +106,7 @@ Other ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -122,6 +125,7 @@ Other ops: \cr
 \code{\link{op_associative_scan}()} \cr
 \code{\link{op_average}()} \cr
 \code{\link{op_average_pool}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_batch_normalization}()} \cr
 \code{\link{op_binary_crossentropy}()} \cr
 \code{\link{op_bincount}()} \cr
@@ -132,9 +136,11 @@ Other ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
 \code{\link{op_cast}()} \cr
 \code{\link{op_categorical_crossentropy}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_celu}()} \cr
 \code{\link{op_cholesky}()} \cr
@@ -147,6 +153,7 @@ Other ops: \cr
 \code{\link{op_convert_to_numpy}()} \cr
 \code{\link{op_convert_to_tensor}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -157,6 +164,7 @@ Other ops: \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
 \code{\link{op_custom_gradient}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_depthwise_conv}()} \cr
 \code{\link{op_det}()} \cr
 \code{\link{op_diag}()} \cr
@@ -196,10 +204,13 @@ Other ops: \cr
 \code{\link{op_glu}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
 \code{\link{op_hard_shrink}()} \cr
 \code{\link{op_hard_sigmoid}()} \cr
 \code{\link{op_hard_silu}()} \cr
 \code{\link{op_hard_tanh}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -226,6 +237,8 @@ Other ops: \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
 \code{\link{op_istft}()} \cr
+\code{\link{op_kaiser}()} \cr
+\code{\link{op_layer_normalization}()} \cr
 \code{\link{op_leaky_relu}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
@@ -324,6 +337,7 @@ Other ops: \cr
 \code{\link{op_sort}()} \cr
 \code{\link{op_sparse_categorical_crossentropy}()} \cr
 \code{\link{op_sparse_plus}()} \cr
+\code{\link{op_sparse_sigmoid}()} \cr
 \code{\link{op_sparsemax}()} \cr
 \code{\link{op_split}()} \cr
 \code{\link{op_sqrt}()} \cr
@@ -361,6 +375,8 @@ Other ops: \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
 \code{\link{op_vectorized_map}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_while_loop}()} \cr
diff --git a/man/op_maximum.Rd b/man/op_maximum.Rd
index 1eeda1dde..b1e16305d 100644
--- a/man/op_maximum.Rd
+++ b/man/op_maximum.Rd
@@ -29,6 +29,7 @@ Other numpy ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -45,6 +46,7 @@ Other numpy ops: \cr
 \code{\link{op_argsort}()} \cr
 \code{\link{op_array}()} \cr
 \code{\link{op_average}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_bincount}()} \cr
 \code{\link{op_bitwise_and}()} \cr
 \code{\link{op_bitwise_invert}()} \cr
@@ -53,12 +55,15 @@ Other numpy ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_clip}()} \cr
 \code{\link{op_concatenate}()} \cr
 \code{\link{op_conj}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -67,6 +72,7 @@ Other numpy ops: \cr
 \code{\link{op_ctc_decode}()} \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_diag}()} \cr
 \code{\link{op_diagflat}()} \cr
 \code{\link{op_diagonal}()} \cr
@@ -91,6 +97,9 @@ Other numpy ops: \cr
 \code{\link{op_get_item}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -100,6 +109,7 @@ Other numpy ops: \cr
 \code{\link{op_isfinite}()} \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
+\code{\link{op_kaiser}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
 \code{\link{op_less_equal}()} \cr
@@ -179,6 +189,8 @@ Other numpy ops: \cr
 \code{\link{op_var}()} \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_zeros}()} \cr
@@ -188,6 +200,7 @@ Other ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -206,6 +219,7 @@ Other ops: \cr
 \code{\link{op_associative_scan}()} \cr
 \code{\link{op_average}()} \cr
 \code{\link{op_average_pool}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_batch_normalization}()} \cr
 \code{\link{op_binary_crossentropy}()} \cr
 \code{\link{op_bincount}()} \cr
@@ -216,9 +230,11 @@ Other ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
 \code{\link{op_cast}()} \cr
 \code{\link{op_categorical_crossentropy}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_celu}()} \cr
 \code{\link{op_cholesky}()} \cr
@@ -231,6 +247,7 @@ Other ops: \cr
 \code{\link{op_convert_to_numpy}()} \cr
 \code{\link{op_convert_to_tensor}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -241,6 +258,7 @@ Other ops: \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
 \code{\link{op_custom_gradient}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_depthwise_conv}()} \cr
 \code{\link{op_det}()} \cr
 \code{\link{op_diag}()} \cr
@@ -280,10 +298,13 @@ Other ops: \cr
 \code{\link{op_glu}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
 \code{\link{op_hard_shrink}()} \cr
 \code{\link{op_hard_sigmoid}()} \cr
 \code{\link{op_hard_silu}()} \cr
 \code{\link{op_hard_tanh}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -310,6 +331,8 @@ Other ops: \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
 \code{\link{op_istft}()} \cr
+\code{\link{op_kaiser}()} \cr
+\code{\link{op_layer_normalization}()} \cr
 \code{\link{op_leaky_relu}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
@@ -408,6 +431,7 @@ Other ops: \cr
 \code{\link{op_sort}()} \cr
 \code{\link{op_sparse_categorical_crossentropy}()} \cr
 \code{\link{op_sparse_plus}()} \cr
+\code{\link{op_sparse_sigmoid}()} \cr
 \code{\link{op_sparsemax}()} \cr
 \code{\link{op_split}()} \cr
 \code{\link{op_sqrt}()} \cr
@@ -445,6 +469,8 @@ Other ops: \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
 \code{\link{op_vectorized_map}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_while_loop}()} \cr
diff --git a/man/op_mean.Rd b/man/op_mean.Rd
index 33d3f5ec5..704985600 100644
--- a/man/op_mean.Rd
+++ b/man/op_mean.Rd
@@ -30,6 +30,7 @@ Other numpy ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -46,6 +47,7 @@ Other numpy ops: \cr
 \code{\link{op_argsort}()} \cr
 \code{\link{op_array}()} \cr
 \code{\link{op_average}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_bincount}()} \cr
 \code{\link{op_bitwise_and}()} \cr
 \code{\link{op_bitwise_invert}()} \cr
@@ -54,12 +56,15 @@ Other numpy ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_clip}()} \cr
 \code{\link{op_concatenate}()} \cr
 \code{\link{op_conj}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -68,6 +73,7 @@ Other numpy ops: \cr
 \code{\link{op_ctc_decode}()} \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_diag}()} \cr
 \code{\link{op_diagflat}()} \cr
 \code{\link{op_diagonal}()} \cr
@@ -92,6 +98,9 @@ Other numpy ops: \cr
 \code{\link{op_get_item}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -101,6 +110,7 @@ Other numpy ops: \cr
 \code{\link{op_isfinite}()} \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
+\code{\link{op_kaiser}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
 \code{\link{op_less_equal}()} \cr
@@ -180,6 +190,8 @@ Other numpy ops: \cr
 \code{\link{op_var}()} \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_zeros}()} \cr
@@ -189,6 +201,7 @@ Other ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -207,6 +220,7 @@ Other ops: \cr
 \code{\link{op_associative_scan}()} \cr
 \code{\link{op_average}()} \cr
 \code{\link{op_average_pool}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_batch_normalization}()} \cr
 \code{\link{op_binary_crossentropy}()} \cr
 \code{\link{op_bincount}()} \cr
@@ -217,9 +231,11 @@ Other ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
 \code{\link{op_cast}()} \cr
 \code{\link{op_categorical_crossentropy}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_celu}()} \cr
 \code{\link{op_cholesky}()} \cr
@@ -232,6 +248,7 @@ Other ops: \cr
 \code{\link{op_convert_to_numpy}()} \cr
 \code{\link{op_convert_to_tensor}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -242,6 +259,7 @@ Other ops: \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
 \code{\link{op_custom_gradient}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_depthwise_conv}()} \cr
 \code{\link{op_det}()} \cr
 \code{\link{op_diag}()} \cr
@@ -281,10 +299,13 @@ Other ops: \cr
 \code{\link{op_glu}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
 \code{\link{op_hard_shrink}()} \cr
 \code{\link{op_hard_sigmoid}()} \cr
 \code{\link{op_hard_silu}()} \cr
 \code{\link{op_hard_tanh}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -311,6 +332,8 @@ Other ops: \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
 \code{\link{op_istft}()} \cr
+\code{\link{op_kaiser}()} \cr
+\code{\link{op_layer_normalization}()} \cr
 \code{\link{op_leaky_relu}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
@@ -409,6 +432,7 @@ Other ops: \cr
 \code{\link{op_sort}()} \cr
 \code{\link{op_sparse_categorical_crossentropy}()} \cr
 \code{\link{op_sparse_plus}()} \cr
+\code{\link{op_sparse_sigmoid}()} \cr
 \code{\link{op_sparsemax}()} \cr
 \code{\link{op_split}()} \cr
 \code{\link{op_sqrt}()} \cr
@@ -446,6 +470,8 @@ Other ops: \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
 \code{\link{op_vectorized_map}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_while_loop}()} \cr
diff --git a/man/op_median.Rd b/man/op_median.Rd
index 1ab5d3d77..798954d11 100644
--- a/man/op_median.Rd
+++ b/man/op_median.Rd
@@ -27,6 +27,7 @@ Other numpy ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -43,6 +44,7 @@ Other numpy ops: \cr
 \code{\link{op_argsort}()} \cr
 \code{\link{op_array}()} \cr
 \code{\link{op_average}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_bincount}()} \cr
 \code{\link{op_bitwise_and}()} \cr
 \code{\link{op_bitwise_invert}()} \cr
@@ -51,12 +53,15 @@ Other numpy ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_clip}()} \cr
 \code{\link{op_concatenate}()} \cr
 \code{\link{op_conj}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -65,6 +70,7 @@ Other numpy ops: \cr
 \code{\link{op_ctc_decode}()} \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_diag}()} \cr
 \code{\link{op_diagflat}()} \cr
 \code{\link{op_diagonal}()} \cr
@@ -89,6 +95,9 @@ Other numpy ops: \cr
 \code{\link{op_get_item}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -98,6 +107,7 @@ Other numpy ops: \cr
 \code{\link{op_isfinite}()} \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
+\code{\link{op_kaiser}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
 \code{\link{op_less_equal}()} \cr
@@ -177,6 +187,8 @@ Other numpy ops: \cr
 \code{\link{op_var}()} \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_zeros}()} \cr
@@ -186,6 +198,7 @@ Other ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -204,6 +217,7 @@ Other ops: \cr
 \code{\link{op_associative_scan}()} \cr
 \code{\link{op_average}()} \cr
 \code{\link{op_average_pool}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_batch_normalization}()} \cr
 \code{\link{op_binary_crossentropy}()} \cr
 \code{\link{op_bincount}()} \cr
@@ -214,9 +228,11 @@ Other ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
 \code{\link{op_cast}()} \cr
 \code{\link{op_categorical_crossentropy}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_celu}()} \cr
 \code{\link{op_cholesky}()} \cr
@@ -229,6 +245,7 @@ Other ops: \cr
 \code{\link{op_convert_to_numpy}()} \cr
 \code{\link{op_convert_to_tensor}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -239,6 +256,7 @@ Other ops: \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
 \code{\link{op_custom_gradient}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_depthwise_conv}()} \cr
 \code{\link{op_det}()} \cr
 \code{\link{op_diag}()} \cr
@@ -278,10 +296,13 @@ Other ops: \cr
 \code{\link{op_glu}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
 \code{\link{op_hard_shrink}()} \cr
 \code{\link{op_hard_sigmoid}()} \cr
 \code{\link{op_hard_silu}()} \cr
 \code{\link{op_hard_tanh}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -308,6 +329,8 @@ Other ops: \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
 \code{\link{op_istft}()} \cr
+\code{\link{op_kaiser}()} \cr
+\code{\link{op_layer_normalization}()} \cr
 \code{\link{op_leaky_relu}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
@@ -406,6 +429,7 @@ Other ops: \cr
 \code{\link{op_sort}()} \cr
 \code{\link{op_sparse_categorical_crossentropy}()} \cr
 \code{\link{op_sparse_plus}()} \cr
+\code{\link{op_sparse_sigmoid}()} \cr
 \code{\link{op_sparsemax}()} \cr
 \code{\link{op_split}()} \cr
 \code{\link{op_sqrt}()} \cr
@@ -443,6 +467,8 @@ Other ops: \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
 \code{\link{op_vectorized_map}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_while_loop}()} \cr
diff --git a/man/op_meshgrid.Rd b/man/op_meshgrid.Rd
index 66e353710..0d0f66884 100644
--- a/man/op_meshgrid.Rd
+++ b/man/op_meshgrid.Rd
@@ -65,6 +65,7 @@ Other numpy ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -81,6 +82,7 @@ Other numpy ops: \cr
 \code{\link{op_argsort}()} \cr
 \code{\link{op_array}()} \cr
 \code{\link{op_average}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_bincount}()} \cr
 \code{\link{op_bitwise_and}()} \cr
 \code{\link{op_bitwise_invert}()} \cr
@@ -89,12 +91,15 @@ Other numpy ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_clip}()} \cr
 \code{\link{op_concatenate}()} \cr
 \code{\link{op_conj}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -103,6 +108,7 @@ Other numpy ops: \cr
 \code{\link{op_ctc_decode}()} \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_diag}()} \cr
 \code{\link{op_diagflat}()} \cr
 \code{\link{op_diagonal}()} \cr
@@ -127,6 +133,9 @@ Other numpy ops: \cr
 \code{\link{op_get_item}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -136,6 +145,7 @@ Other numpy ops: \cr
 \code{\link{op_isfinite}()} \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
+\code{\link{op_kaiser}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
 \code{\link{op_less_equal}()} \cr
@@ -215,6 +225,8 @@ Other numpy ops: \cr
 \code{\link{op_var}()} \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_zeros}()} \cr
@@ -224,6 +236,7 @@ Other ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -242,6 +255,7 @@ Other ops: \cr
 \code{\link{op_associative_scan}()} \cr
 \code{\link{op_average}()} \cr
 \code{\link{op_average_pool}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_batch_normalization}()} \cr
 \code{\link{op_binary_crossentropy}()} \cr
 \code{\link{op_bincount}()} \cr
@@ -252,9 +266,11 @@ Other ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
 \code{\link{op_cast}()} \cr
 \code{\link{op_categorical_crossentropy}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_celu}()} \cr
 \code{\link{op_cholesky}()} \cr
@@ -267,6 +283,7 @@ Other ops: \cr
 \code{\link{op_convert_to_numpy}()} \cr
 \code{\link{op_convert_to_tensor}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -277,6 +294,7 @@ Other ops: \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
 \code{\link{op_custom_gradient}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_depthwise_conv}()} \cr
 \code{\link{op_det}()} \cr
 \code{\link{op_diag}()} \cr
@@ -316,10 +334,13 @@ Other ops: \cr
 \code{\link{op_glu}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
 \code{\link{op_hard_shrink}()} \cr
 \code{\link{op_hard_sigmoid}()} \cr
 \code{\link{op_hard_silu}()} \cr
 \code{\link{op_hard_tanh}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -346,6 +367,8 @@ Other ops: \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
 \code{\link{op_istft}()} \cr
+\code{\link{op_kaiser}()} \cr
+\code{\link{op_layer_normalization}()} \cr
 \code{\link{op_leaky_relu}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
@@ -444,6 +467,7 @@ Other ops: \cr
 \code{\link{op_sort}()} \cr
 \code{\link{op_sparse_categorical_crossentropy}()} \cr
 \code{\link{op_sparse_plus}()} \cr
+\code{\link{op_sparse_sigmoid}()} \cr
 \code{\link{op_sparsemax}()} \cr
 \code{\link{op_split}()} \cr
 \code{\link{op_sqrt}()} \cr
@@ -481,6 +505,8 @@ Other ops: \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
 \code{\link{op_vectorized_map}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_while_loop}()} \cr
diff --git a/man/op_min.Rd b/man/op_min.Rd
index 187ad08f9..ce844846a 100644
--- a/man/op_min.Rd
+++ b/man/op_min.Rd
@@ -65,6 +65,7 @@ Other numpy ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -81,6 +82,7 @@ Other numpy ops: \cr
 \code{\link{op_argsort}()} \cr
 \code{\link{op_array}()} \cr
 \code{\link{op_average}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_bincount}()} \cr
 \code{\link{op_bitwise_and}()} \cr
 \code{\link{op_bitwise_invert}()} \cr
@@ -89,12 +91,15 @@ Other numpy ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_clip}()} \cr
 \code{\link{op_concatenate}()} \cr
 \code{\link{op_conj}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -103,6 +108,7 @@ Other numpy ops: \cr
 \code{\link{op_ctc_decode}()} \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_diag}()} \cr
 \code{\link{op_diagflat}()} \cr
 \code{\link{op_diagonal}()} \cr
@@ -127,6 +133,9 @@ Other numpy ops: \cr
 \code{\link{op_get_item}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -136,6 +145,7 @@ Other numpy ops: \cr
 \code{\link{op_isfinite}()} \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
+\code{\link{op_kaiser}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
 \code{\link{op_less_equal}()} \cr
@@ -215,6 +225,8 @@ Other numpy ops: \cr
 \code{\link{op_var}()} \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_zeros}()} \cr
@@ -224,6 +236,7 @@ Other ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -242,6 +255,7 @@ Other ops: \cr
 \code{\link{op_associative_scan}()} \cr
 \code{\link{op_average}()} \cr
 \code{\link{op_average_pool}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_batch_normalization}()} \cr
 \code{\link{op_binary_crossentropy}()} \cr
 \code{\link{op_bincount}()} \cr
@@ -252,9 +266,11 @@ Other ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
 \code{\link{op_cast}()} \cr
 \code{\link{op_categorical_crossentropy}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_celu}()} \cr
 \code{\link{op_cholesky}()} \cr
@@ -267,6 +283,7 @@ Other ops: \cr
 \code{\link{op_convert_to_numpy}()} \cr
 \code{\link{op_convert_to_tensor}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -277,6 +294,7 @@ Other ops: \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
 \code{\link{op_custom_gradient}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_depthwise_conv}()} \cr
 \code{\link{op_det}()} \cr
 \code{\link{op_diag}()} \cr
@@ -316,10 +334,13 @@ Other ops: \cr
 \code{\link{op_glu}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
 \code{\link{op_hard_shrink}()} \cr
 \code{\link{op_hard_sigmoid}()} \cr
 \code{\link{op_hard_silu}()} \cr
 \code{\link{op_hard_tanh}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -346,6 +367,8 @@ Other ops: \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
 \code{\link{op_istft}()} \cr
+\code{\link{op_kaiser}()} \cr
+\code{\link{op_layer_normalization}()} \cr
 \code{\link{op_leaky_relu}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
@@ -444,6 +467,7 @@ Other ops: \cr
 \code{\link{op_sort}()} \cr
 \code{\link{op_sparse_categorical_crossentropy}()} \cr
 \code{\link{op_sparse_plus}()} \cr
+\code{\link{op_sparse_sigmoid}()} \cr
 \code{\link{op_sparsemax}()} \cr
 \code{\link{op_split}()} \cr
 \code{\link{op_sqrt}()} \cr
@@ -481,6 +505,8 @@ Other ops: \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
 \code{\link{op_vectorized_map}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_while_loop}()} \cr
diff --git a/man/op_minimum.Rd b/man/op_minimum.Rd
index fd528ed3b..b7818fba1 100644
--- a/man/op_minimum.Rd
+++ b/man/op_minimum.Rd
@@ -29,6 +29,7 @@ Other numpy ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -45,6 +46,7 @@ Other numpy ops: \cr
 \code{\link{op_argsort}()} \cr
 \code{\link{op_array}()} \cr
 \code{\link{op_average}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_bincount}()} \cr
 \code{\link{op_bitwise_and}()} \cr
 \code{\link{op_bitwise_invert}()} \cr
@@ -53,12 +55,15 @@ Other numpy ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_clip}()} \cr
 \code{\link{op_concatenate}()} \cr
 \code{\link{op_conj}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -67,6 +72,7 @@ Other numpy ops: \cr
 \code{\link{op_ctc_decode}()} \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_diag}()} \cr
 \code{\link{op_diagflat}()} \cr
 \code{\link{op_diagonal}()} \cr
@@ -91,6 +97,9 @@ Other numpy ops: \cr
 \code{\link{op_get_item}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -100,6 +109,7 @@ Other numpy ops: \cr
 \code{\link{op_isfinite}()} \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
+\code{\link{op_kaiser}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
 \code{\link{op_less_equal}()} \cr
@@ -179,6 +189,8 @@ Other numpy ops: \cr
 \code{\link{op_var}()} \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_zeros}()} \cr
@@ -188,6 +200,7 @@ Other ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -206,6 +219,7 @@ Other ops: \cr
 \code{\link{op_associative_scan}()} \cr
 \code{\link{op_average}()} \cr
 \code{\link{op_average_pool}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_batch_normalization}()} \cr
 \code{\link{op_binary_crossentropy}()} \cr
 \code{\link{op_bincount}()} \cr
@@ -216,9 +230,11 @@ Other ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
 \code{\link{op_cast}()} \cr
 \code{\link{op_categorical_crossentropy}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_celu}()} \cr
 \code{\link{op_cholesky}()} \cr
@@ -231,6 +247,7 @@ Other ops: \cr
 \code{\link{op_convert_to_numpy}()} \cr
 \code{\link{op_convert_to_tensor}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -241,6 +258,7 @@ Other ops: \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
 \code{\link{op_custom_gradient}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_depthwise_conv}()} \cr
 \code{\link{op_det}()} \cr
 \code{\link{op_diag}()} \cr
@@ -280,10 +298,13 @@ Other ops: \cr
 \code{\link{op_glu}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
 \code{\link{op_hard_shrink}()} \cr
 \code{\link{op_hard_sigmoid}()} \cr
 \code{\link{op_hard_silu}()} \cr
 \code{\link{op_hard_tanh}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -310,6 +331,8 @@ Other ops: \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
 \code{\link{op_istft}()} \cr
+\code{\link{op_kaiser}()} \cr
+\code{\link{op_layer_normalization}()} \cr
 \code{\link{op_leaky_relu}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
@@ -408,6 +431,7 @@ Other ops: \cr
 \code{\link{op_sort}()} \cr
 \code{\link{op_sparse_categorical_crossentropy}()} \cr
 \code{\link{op_sparse_plus}()} \cr
+\code{\link{op_sparse_sigmoid}()} \cr
 \code{\link{op_sparsemax}()} \cr
 \code{\link{op_split}()} \cr
 \code{\link{op_sqrt}()} \cr
@@ -445,6 +469,8 @@ Other ops: \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
 \code{\link{op_vectorized_map}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_while_loop}()} \cr
diff --git a/man/op_mod.Rd b/man/op_mod.Rd
index f5d8cd554..b42f4a844 100644
--- a/man/op_mod.Rd
+++ b/man/op_mod.Rd
@@ -47,6 +47,7 @@ Other numpy ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -63,6 +64,7 @@ Other numpy ops: \cr
 \code{\link{op_argsort}()} \cr
 \code{\link{op_array}()} \cr
 \code{\link{op_average}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_bincount}()} \cr
 \code{\link{op_bitwise_and}()} \cr
 \code{\link{op_bitwise_invert}()} \cr
@@ -71,12 +73,15 @@ Other numpy ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_clip}()} \cr
 \code{\link{op_concatenate}()} \cr
 \code{\link{op_conj}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -85,6 +90,7 @@ Other numpy ops: \cr
 \code{\link{op_ctc_decode}()} \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_diag}()} \cr
 \code{\link{op_diagflat}()} \cr
 \code{\link{op_diagonal}()} \cr
@@ -109,6 +115,9 @@ Other numpy ops: \cr
 \code{\link{op_get_item}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -118,6 +127,7 @@ Other numpy ops: \cr
 \code{\link{op_isfinite}()} \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
+\code{\link{op_kaiser}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
 \code{\link{op_less_equal}()} \cr
@@ -197,6 +207,8 @@ Other numpy ops: \cr
 \code{\link{op_var}()} \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_zeros}()} \cr
@@ -206,6 +218,7 @@ Other ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -224,6 +237,7 @@ Other ops: \cr
 \code{\link{op_associative_scan}()} \cr
 \code{\link{op_average}()} \cr
 \code{\link{op_average_pool}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_batch_normalization}()} \cr
 \code{\link{op_binary_crossentropy}()} \cr
 \code{\link{op_bincount}()} \cr
@@ -234,9 +248,11 @@ Other ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
 \code{\link{op_cast}()} \cr
 \code{\link{op_categorical_crossentropy}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_celu}()} \cr
 \code{\link{op_cholesky}()} \cr
@@ -249,6 +265,7 @@ Other ops: \cr
 \code{\link{op_convert_to_numpy}()} \cr
 \code{\link{op_convert_to_tensor}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -259,6 +276,7 @@ Other ops: \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
 \code{\link{op_custom_gradient}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_depthwise_conv}()} \cr
 \code{\link{op_det}()} \cr
 \code{\link{op_diag}()} \cr
@@ -298,10 +316,13 @@ Other ops: \cr
 \code{\link{op_glu}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
 \code{\link{op_hard_shrink}()} \cr
 \code{\link{op_hard_sigmoid}()} \cr
 \code{\link{op_hard_silu}()} \cr
 \code{\link{op_hard_tanh}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -328,6 +349,8 @@ Other ops: \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
 \code{\link{op_istft}()} \cr
+\code{\link{op_kaiser}()} \cr
+\code{\link{op_layer_normalization}()} \cr
 \code{\link{op_leaky_relu}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
@@ -426,6 +449,7 @@ Other ops: \cr
 \code{\link{op_sort}()} \cr
 \code{\link{op_sparse_categorical_crossentropy}()} \cr
 \code{\link{op_sparse_plus}()} \cr
+\code{\link{op_sparse_sigmoid}()} \cr
 \code{\link{op_sparsemax}()} \cr
 \code{\link{op_split}()} \cr
 \code{\link{op_sqrt}()} \cr
@@ -463,6 +487,8 @@ Other ops: \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
 \code{\link{op_vectorized_map}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_while_loop}()} \cr
diff --git a/man/op_moments.Rd b/man/op_moments.Rd
index 92c50086c..97d3055f7 100644
--- a/man/op_moments.Rd
+++ b/man/op_moments.Rd
@@ -61,6 +61,7 @@ Other nn ops: \cr
 \code{\link{op_hard_sigmoid}()} \cr
 \code{\link{op_hard_silu}()} \cr
 \code{\link{op_hard_tanh}()} \cr
+\code{\link{op_layer_normalization}()} \cr
 \code{\link{op_leaky_relu}()} \cr
 \code{\link{op_log_sigmoid}()} \cr
 \code{\link{op_log_softmax}()} \cr
@@ -83,6 +84,7 @@ Other nn ops: \cr
 \code{\link{op_softsign}()} \cr
 \code{\link{op_sparse_categorical_crossentropy}()} \cr
 \code{\link{op_sparse_plus}()} \cr
+\code{\link{op_sparse_sigmoid}()} \cr
 \code{\link{op_sparsemax}()} \cr
 \code{\link{op_squareplus}()} \cr
 \code{\link{op_tanh_shrink}()} \cr
@@ -93,6 +95,7 @@ Other ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -111,6 +114,7 @@ Other ops: \cr
 \code{\link{op_associative_scan}()} \cr
 \code{\link{op_average}()} \cr
 \code{\link{op_average_pool}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_batch_normalization}()} \cr
 \code{\link{op_binary_crossentropy}()} \cr
 \code{\link{op_bincount}()} \cr
@@ -121,9 +125,11 @@ Other ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
 \code{\link{op_cast}()} \cr
 \code{\link{op_categorical_crossentropy}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_celu}()} \cr
 \code{\link{op_cholesky}()} \cr
@@ -136,6 +142,7 @@ Other ops: \cr
 \code{\link{op_convert_to_numpy}()} \cr
 \code{\link{op_convert_to_tensor}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -146,6 +153,7 @@ Other ops: \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
 \code{\link{op_custom_gradient}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_depthwise_conv}()} \cr
 \code{\link{op_det}()} \cr
 \code{\link{op_diag}()} \cr
@@ -185,10 +193,13 @@ Other ops: \cr
 \code{\link{op_glu}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
 \code{\link{op_hard_shrink}()} \cr
 \code{\link{op_hard_sigmoid}()} \cr
 \code{\link{op_hard_silu}()} \cr
 \code{\link{op_hard_tanh}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -215,6 +226,8 @@ Other ops: \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
 \code{\link{op_istft}()} \cr
+\code{\link{op_kaiser}()} \cr
+\code{\link{op_layer_normalization}()} \cr
 \code{\link{op_leaky_relu}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
@@ -313,6 +326,7 @@ Other ops: \cr
 \code{\link{op_sort}()} \cr
 \code{\link{op_sparse_categorical_crossentropy}()} \cr
 \code{\link{op_sparse_plus}()} \cr
+\code{\link{op_sparse_sigmoid}()} \cr
 \code{\link{op_sparsemax}()} \cr
 \code{\link{op_split}()} \cr
 \code{\link{op_sqrt}()} \cr
@@ -350,6 +364,8 @@ Other ops: \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
 \code{\link{op_vectorized_map}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_while_loop}()} \cr
diff --git a/man/op_moveaxis.Rd b/man/op_moveaxis.Rd
index 999de17b6..af092a112 100644
--- a/man/op_moveaxis.Rd
+++ b/man/op_moveaxis.Rd
@@ -29,6 +29,7 @@ Other numpy ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -45,6 +46,7 @@ Other numpy ops: \cr
 \code{\link{op_argsort}()} \cr
 \code{\link{op_array}()} \cr
 \code{\link{op_average}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_bincount}()} \cr
 \code{\link{op_bitwise_and}()} \cr
 \code{\link{op_bitwise_invert}()} \cr
@@ -53,12 +55,15 @@ Other numpy ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_clip}()} \cr
 \code{\link{op_concatenate}()} \cr
 \code{\link{op_conj}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -67,6 +72,7 @@ Other numpy ops: \cr
 \code{\link{op_ctc_decode}()} \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_diag}()} \cr
 \code{\link{op_diagflat}()} \cr
 \code{\link{op_diagonal}()} \cr
@@ -91,6 +97,9 @@ Other numpy ops: \cr
 \code{\link{op_get_item}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -100,6 +109,7 @@ Other numpy ops: \cr
 \code{\link{op_isfinite}()} \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
+\code{\link{op_kaiser}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
 \code{\link{op_less_equal}()} \cr
@@ -179,6 +189,8 @@ Other numpy ops: \cr
 \code{\link{op_var}()} \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_zeros}()} \cr
@@ -188,6 +200,7 @@ Other ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -206,6 +219,7 @@ Other ops: \cr
 \code{\link{op_associative_scan}()} \cr
 \code{\link{op_average}()} \cr
 \code{\link{op_average_pool}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_batch_normalization}()} \cr
 \code{\link{op_binary_crossentropy}()} \cr
 \code{\link{op_bincount}()} \cr
@@ -216,9 +230,11 @@ Other ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
 \code{\link{op_cast}()} \cr
 \code{\link{op_categorical_crossentropy}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_celu}()} \cr
 \code{\link{op_cholesky}()} \cr
@@ -231,6 +247,7 @@ Other ops: \cr
 \code{\link{op_convert_to_numpy}()} \cr
 \code{\link{op_convert_to_tensor}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -241,6 +258,7 @@ Other ops: \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
 \code{\link{op_custom_gradient}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_depthwise_conv}()} \cr
 \code{\link{op_det}()} \cr
 \code{\link{op_diag}()} \cr
@@ -280,10 +298,13 @@ Other ops: \cr
 \code{\link{op_glu}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
 \code{\link{op_hard_shrink}()} \cr
 \code{\link{op_hard_sigmoid}()} \cr
 \code{\link{op_hard_silu}()} \cr
 \code{\link{op_hard_tanh}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -310,6 +331,8 @@ Other ops: \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
 \code{\link{op_istft}()} \cr
+\code{\link{op_kaiser}()} \cr
+\code{\link{op_layer_normalization}()} \cr
 \code{\link{op_leaky_relu}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
@@ -408,6 +431,7 @@ Other ops: \cr
 \code{\link{op_sort}()} \cr
 \code{\link{op_sparse_categorical_crossentropy}()} \cr
 \code{\link{op_sparse_plus}()} \cr
+\code{\link{op_sparse_sigmoid}()} \cr
 \code{\link{op_sparsemax}()} \cr
 \code{\link{op_split}()} \cr
 \code{\link{op_sqrt}()} \cr
@@ -445,6 +469,8 @@ Other ops: \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
 \code{\link{op_vectorized_map}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_while_loop}()} \cr
diff --git a/man/op_multi_hot.Rd b/man/op_multi_hot.Rd
index 1fc6d7f6c..7f984e773 100644
--- a/man/op_multi_hot.Rd
+++ b/man/op_multi_hot.Rd
@@ -65,6 +65,7 @@ Other nn ops: \cr
 \code{\link{op_hard_sigmoid}()} \cr
 \code{\link{op_hard_silu}()} \cr
 \code{\link{op_hard_tanh}()} \cr
+\code{\link{op_layer_normalization}()} \cr
 \code{\link{op_leaky_relu}()} \cr
 \code{\link{op_log_sigmoid}()} \cr
 \code{\link{op_log_softmax}()} \cr
@@ -87,6 +88,7 @@ Other nn ops: \cr
 \code{\link{op_softsign}()} \cr
 \code{\link{op_sparse_categorical_crossentropy}()} \cr
 \code{\link{op_sparse_plus}()} \cr
+\code{\link{op_sparse_sigmoid}()} \cr
 \code{\link{op_sparsemax}()} \cr
 \code{\link{op_squareplus}()} \cr
 \code{\link{op_tanh_shrink}()} \cr
@@ -97,6 +99,7 @@ Other ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -115,6 +118,7 @@ Other ops: \cr
 \code{\link{op_associative_scan}()} \cr
 \code{\link{op_average}()} \cr
 \code{\link{op_average_pool}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_batch_normalization}()} \cr
 \code{\link{op_binary_crossentropy}()} \cr
 \code{\link{op_bincount}()} \cr
@@ -125,9 +129,11 @@ Other ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
 \code{\link{op_cast}()} \cr
 \code{\link{op_categorical_crossentropy}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_celu}()} \cr
 \code{\link{op_cholesky}()} \cr
@@ -140,6 +146,7 @@ Other ops: \cr
 \code{\link{op_convert_to_numpy}()} \cr
 \code{\link{op_convert_to_tensor}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -150,6 +157,7 @@ Other ops: \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
 \code{\link{op_custom_gradient}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_depthwise_conv}()} \cr
 \code{\link{op_det}()} \cr
 \code{\link{op_diag}()} \cr
@@ -189,10 +197,13 @@ Other ops: \cr
 \code{\link{op_glu}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
 \code{\link{op_hard_shrink}()} \cr
 \code{\link{op_hard_sigmoid}()} \cr
 \code{\link{op_hard_silu}()} \cr
 \code{\link{op_hard_tanh}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -219,6 +230,8 @@ Other ops: \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
 \code{\link{op_istft}()} \cr
+\code{\link{op_kaiser}()} \cr
+\code{\link{op_layer_normalization}()} \cr
 \code{\link{op_leaky_relu}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
@@ -317,6 +330,7 @@ Other ops: \cr
 \code{\link{op_sort}()} \cr
 \code{\link{op_sparse_categorical_crossentropy}()} \cr
 \code{\link{op_sparse_plus}()} \cr
+\code{\link{op_sparse_sigmoid}()} \cr
 \code{\link{op_sparsemax}()} \cr
 \code{\link{op_split}()} \cr
 \code{\link{op_sqrt}()} \cr
@@ -354,6 +368,8 @@ Other ops: \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
 \code{\link{op_vectorized_map}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_while_loop}()} \cr
diff --git a/man/op_multiply.Rd b/man/op_multiply.Rd
index 7f1ca7b8b..bb0fc0f63 100644
--- a/man/op_multiply.Rd
+++ b/man/op_multiply.Rd
@@ -47,6 +47,7 @@ Other numpy ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -63,6 +64,7 @@ Other numpy ops: \cr
 \code{\link{op_argsort}()} \cr
 \code{\link{op_array}()} \cr
 \code{\link{op_average}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_bincount}()} \cr
 \code{\link{op_bitwise_and}()} \cr
 \code{\link{op_bitwise_invert}()} \cr
@@ -71,12 +73,15 @@ Other numpy ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_clip}()} \cr
 \code{\link{op_concatenate}()} \cr
 \code{\link{op_conj}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -85,6 +90,7 @@ Other numpy ops: \cr
 \code{\link{op_ctc_decode}()} \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_diag}()} \cr
 \code{\link{op_diagflat}()} \cr
 \code{\link{op_diagonal}()} \cr
@@ -109,6 +115,9 @@ Other numpy ops: \cr
 \code{\link{op_get_item}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -118,6 +127,7 @@ Other numpy ops: \cr
 \code{\link{op_isfinite}()} \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
+\code{\link{op_kaiser}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
 \code{\link{op_less_equal}()} \cr
@@ -197,6 +207,8 @@ Other numpy ops: \cr
 \code{\link{op_var}()} \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_zeros}()} \cr
@@ -206,6 +218,7 @@ Other ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -224,6 +237,7 @@ Other ops: \cr
 \code{\link{op_associative_scan}()} \cr
 \code{\link{op_average}()} \cr
 \code{\link{op_average_pool}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_batch_normalization}()} \cr
 \code{\link{op_binary_crossentropy}()} \cr
 \code{\link{op_bincount}()} \cr
@@ -234,9 +248,11 @@ Other ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
 \code{\link{op_cast}()} \cr
 \code{\link{op_categorical_crossentropy}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_celu}()} \cr
 \code{\link{op_cholesky}()} \cr
@@ -249,6 +265,7 @@ Other ops: \cr
 \code{\link{op_convert_to_numpy}()} \cr
 \code{\link{op_convert_to_tensor}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -259,6 +276,7 @@ Other ops: \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
 \code{\link{op_custom_gradient}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_depthwise_conv}()} \cr
 \code{\link{op_det}()} \cr
 \code{\link{op_diag}()} \cr
@@ -298,10 +316,13 @@ Other ops: \cr
 \code{\link{op_glu}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
 \code{\link{op_hard_shrink}()} \cr
 \code{\link{op_hard_sigmoid}()} \cr
 \code{\link{op_hard_silu}()} \cr
 \code{\link{op_hard_tanh}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -328,6 +349,8 @@ Other ops: \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
 \code{\link{op_istft}()} \cr
+\code{\link{op_kaiser}()} \cr
+\code{\link{op_layer_normalization}()} \cr
 \code{\link{op_leaky_relu}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
@@ -426,6 +449,7 @@ Other ops: \cr
 \code{\link{op_sort}()} \cr
 \code{\link{op_sparse_categorical_crossentropy}()} \cr
 \code{\link{op_sparse_plus}()} \cr
+\code{\link{op_sparse_sigmoid}()} \cr
 \code{\link{op_sparsemax}()} \cr
 \code{\link{op_split}()} \cr
 \code{\link{op_sqrt}()} \cr
@@ -463,6 +487,8 @@ Other ops: \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
 \code{\link{op_vectorized_map}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_while_loop}()} \cr
diff --git a/man/op_nan_to_num.Rd b/man/op_nan_to_num.Rd
index 60d983173..7159b37b0 100644
--- a/man/op_nan_to_num.Rd
+++ b/man/op_nan_to_num.Rd
@@ -53,6 +53,7 @@ Other numpy ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -69,6 +70,7 @@ Other numpy ops: \cr
 \code{\link{op_argsort}()} \cr
 \code{\link{op_array}()} \cr
 \code{\link{op_average}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_bincount}()} \cr
 \code{\link{op_bitwise_and}()} \cr
 \code{\link{op_bitwise_invert}()} \cr
@@ -77,12 +79,15 @@ Other numpy ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_clip}()} \cr
 \code{\link{op_concatenate}()} \cr
 \code{\link{op_conj}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -91,6 +96,7 @@ Other numpy ops: \cr
 \code{\link{op_ctc_decode}()} \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_diag}()} \cr
 \code{\link{op_diagflat}()} \cr
 \code{\link{op_diagonal}()} \cr
@@ -115,6 +121,9 @@ Other numpy ops: \cr
 \code{\link{op_get_item}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -124,6 +133,7 @@ Other numpy ops: \cr
 \code{\link{op_isfinite}()} \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
+\code{\link{op_kaiser}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
 \code{\link{op_less_equal}()} \cr
@@ -203,6 +213,8 @@ Other numpy ops: \cr
 \code{\link{op_var}()} \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_zeros}()} \cr
@@ -212,6 +224,7 @@ Other ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -230,6 +243,7 @@ Other ops: \cr
 \code{\link{op_associative_scan}()} \cr
 \code{\link{op_average}()} \cr
 \code{\link{op_average_pool}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_batch_normalization}()} \cr
 \code{\link{op_binary_crossentropy}()} \cr
 \code{\link{op_bincount}()} \cr
@@ -240,9 +254,11 @@ Other ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
 \code{\link{op_cast}()} \cr
 \code{\link{op_categorical_crossentropy}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_celu}()} \cr
 \code{\link{op_cholesky}()} \cr
@@ -255,6 +271,7 @@ Other ops: \cr
 \code{\link{op_convert_to_numpy}()} \cr
 \code{\link{op_convert_to_tensor}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -265,6 +282,7 @@ Other ops: \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
 \code{\link{op_custom_gradient}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_depthwise_conv}()} \cr
 \code{\link{op_det}()} \cr
 \code{\link{op_diag}()} \cr
@@ -304,10 +322,13 @@ Other ops: \cr
 \code{\link{op_glu}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
 \code{\link{op_hard_shrink}()} \cr
 \code{\link{op_hard_sigmoid}()} \cr
 \code{\link{op_hard_silu}()} \cr
 \code{\link{op_hard_tanh}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -334,6 +355,8 @@ Other ops: \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
 \code{\link{op_istft}()} \cr
+\code{\link{op_kaiser}()} \cr
+\code{\link{op_layer_normalization}()} \cr
 \code{\link{op_leaky_relu}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
@@ -432,6 +455,7 @@ Other ops: \cr
 \code{\link{op_sort}()} \cr
 \code{\link{op_sparse_categorical_crossentropy}()} \cr
 \code{\link{op_sparse_plus}()} \cr
+\code{\link{op_sparse_sigmoid}()} \cr
 \code{\link{op_sparsemax}()} \cr
 \code{\link{op_split}()} \cr
 \code{\link{op_sqrt}()} \cr
@@ -469,6 +493,8 @@ Other ops: \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
 \code{\link{op_vectorized_map}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_while_loop}()} \cr
diff --git a/man/op_ndim.Rd b/man/op_ndim.Rd
index 4aee18e67..543407315 100644
--- a/man/op_ndim.Rd
+++ b/man/op_ndim.Rd
@@ -24,6 +24,7 @@ Other numpy ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -40,6 +41,7 @@ Other numpy ops: \cr
 \code{\link{op_argsort}()} \cr
 \code{\link{op_array}()} \cr
 \code{\link{op_average}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_bincount}()} \cr
 \code{\link{op_bitwise_and}()} \cr
 \code{\link{op_bitwise_invert}()} \cr
@@ -48,12 +50,15 @@ Other numpy ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_clip}()} \cr
 \code{\link{op_concatenate}()} \cr
 \code{\link{op_conj}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -62,6 +67,7 @@ Other numpy ops: \cr
 \code{\link{op_ctc_decode}()} \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_diag}()} \cr
 \code{\link{op_diagflat}()} \cr
 \code{\link{op_diagonal}()} \cr
@@ -86,6 +92,9 @@ Other numpy ops: \cr
 \code{\link{op_get_item}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -95,6 +104,7 @@ Other numpy ops: \cr
 \code{\link{op_isfinite}()} \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
+\code{\link{op_kaiser}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
 \code{\link{op_less_equal}()} \cr
@@ -174,6 +184,8 @@ Other numpy ops: \cr
 \code{\link{op_var}()} \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_zeros}()} \cr
@@ -183,6 +195,7 @@ Other ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -201,6 +214,7 @@ Other ops: \cr
 \code{\link{op_associative_scan}()} \cr
 \code{\link{op_average}()} \cr
 \code{\link{op_average_pool}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_batch_normalization}()} \cr
 \code{\link{op_binary_crossentropy}()} \cr
 \code{\link{op_bincount}()} \cr
@@ -211,9 +225,11 @@ Other ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
 \code{\link{op_cast}()} \cr
 \code{\link{op_categorical_crossentropy}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_celu}()} \cr
 \code{\link{op_cholesky}()} \cr
@@ -226,6 +242,7 @@ Other ops: \cr
 \code{\link{op_convert_to_numpy}()} \cr
 \code{\link{op_convert_to_tensor}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -236,6 +253,7 @@ Other ops: \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
 \code{\link{op_custom_gradient}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_depthwise_conv}()} \cr
 \code{\link{op_det}()} \cr
 \code{\link{op_diag}()} \cr
@@ -275,10 +293,13 @@ Other ops: \cr
 \code{\link{op_glu}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
 \code{\link{op_hard_shrink}()} \cr
 \code{\link{op_hard_sigmoid}()} \cr
 \code{\link{op_hard_silu}()} \cr
 \code{\link{op_hard_tanh}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -305,6 +326,8 @@ Other ops: \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
 \code{\link{op_istft}()} \cr
+\code{\link{op_kaiser}()} \cr
+\code{\link{op_layer_normalization}()} \cr
 \code{\link{op_leaky_relu}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
@@ -403,6 +426,7 @@ Other ops: \cr
 \code{\link{op_sort}()} \cr
 \code{\link{op_sparse_categorical_crossentropy}()} \cr
 \code{\link{op_sparse_plus}()} \cr
+\code{\link{op_sparse_sigmoid}()} \cr
 \code{\link{op_sparsemax}()} \cr
 \code{\link{op_split}()} \cr
 \code{\link{op_sqrt}()} \cr
@@ -440,6 +464,8 @@ Other ops: \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
 \code{\link{op_vectorized_map}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_while_loop}()} \cr
diff --git a/man/op_negative.Rd b/man/op_negative.Rd
index b5e38362f..c4aadc5d5 100644
--- a/man/op_negative.Rd
+++ b/man/op_negative.Rd
@@ -45,6 +45,7 @@ Other numpy ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -61,6 +62,7 @@ Other numpy ops: \cr
 \code{\link{op_argsort}()} \cr
 \code{\link{op_array}()} \cr
 \code{\link{op_average}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_bincount}()} \cr
 \code{\link{op_bitwise_and}()} \cr
 \code{\link{op_bitwise_invert}()} \cr
@@ -69,12 +71,15 @@ Other numpy ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_clip}()} \cr
 \code{\link{op_concatenate}()} \cr
 \code{\link{op_conj}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -83,6 +88,7 @@ Other numpy ops: \cr
 \code{\link{op_ctc_decode}()} \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_diag}()} \cr
 \code{\link{op_diagflat}()} \cr
 \code{\link{op_diagonal}()} \cr
@@ -107,6 +113,9 @@ Other numpy ops: \cr
 \code{\link{op_get_item}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -116,6 +125,7 @@ Other numpy ops: \cr
 \code{\link{op_isfinite}()} \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
+\code{\link{op_kaiser}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
 \code{\link{op_less_equal}()} \cr
@@ -195,6 +205,8 @@ Other numpy ops: \cr
 \code{\link{op_var}()} \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_zeros}()} \cr
@@ -204,6 +216,7 @@ Other ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -222,6 +235,7 @@ Other ops: \cr
 \code{\link{op_associative_scan}()} \cr
 \code{\link{op_average}()} \cr
 \code{\link{op_average_pool}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_batch_normalization}()} \cr
 \code{\link{op_binary_crossentropy}()} \cr
 \code{\link{op_bincount}()} \cr
@@ -232,9 +246,11 @@ Other ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
 \code{\link{op_cast}()} \cr
 \code{\link{op_categorical_crossentropy}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_celu}()} \cr
 \code{\link{op_cholesky}()} \cr
@@ -247,6 +263,7 @@ Other ops: \cr
 \code{\link{op_convert_to_numpy}()} \cr
 \code{\link{op_convert_to_tensor}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -257,6 +274,7 @@ Other ops: \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
 \code{\link{op_custom_gradient}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_depthwise_conv}()} \cr
 \code{\link{op_det}()} \cr
 \code{\link{op_diag}()} \cr
@@ -296,10 +314,13 @@ Other ops: \cr
 \code{\link{op_glu}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
 \code{\link{op_hard_shrink}()} \cr
 \code{\link{op_hard_sigmoid}()} \cr
 \code{\link{op_hard_silu}()} \cr
 \code{\link{op_hard_tanh}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -326,6 +347,8 @@ Other ops: \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
 \code{\link{op_istft}()} \cr
+\code{\link{op_kaiser}()} \cr
+\code{\link{op_layer_normalization}()} \cr
 \code{\link{op_leaky_relu}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
@@ -424,6 +447,7 @@ Other ops: \cr
 \code{\link{op_sort}()} \cr
 \code{\link{op_sparse_categorical_crossentropy}()} \cr
 \code{\link{op_sparse_plus}()} \cr
+\code{\link{op_sparse_sigmoid}()} \cr
 \code{\link{op_sparsemax}()} \cr
 \code{\link{op_split}()} \cr
 \code{\link{op_sqrt}()} \cr
@@ -461,6 +485,8 @@ Other ops: \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
 \code{\link{op_vectorized_map}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_while_loop}()} \cr
diff --git a/man/op_nonzero.Rd b/man/op_nonzero.Rd
index 1b155dfc4..c97ce4ddf 100644
--- a/man/op_nonzero.Rd
+++ b/man/op_nonzero.Rd
@@ -95,6 +95,7 @@ Other numpy ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -111,6 +112,7 @@ Other numpy ops: \cr
 \code{\link{op_argsort}()} \cr
 \code{\link{op_array}()} \cr
 \code{\link{op_average}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_bincount}()} \cr
 \code{\link{op_bitwise_and}()} \cr
 \code{\link{op_bitwise_invert}()} \cr
@@ -119,12 +121,15 @@ Other numpy ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_clip}()} \cr
 \code{\link{op_concatenate}()} \cr
 \code{\link{op_conj}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -133,6 +138,7 @@ Other numpy ops: \cr
 \code{\link{op_ctc_decode}()} \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_diag}()} \cr
 \code{\link{op_diagflat}()} \cr
 \code{\link{op_diagonal}()} \cr
@@ -157,6 +163,9 @@ Other numpy ops: \cr
 \code{\link{op_get_item}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -166,6 +175,7 @@ Other numpy ops: \cr
 \code{\link{op_isfinite}()} \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
+\code{\link{op_kaiser}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
 \code{\link{op_less_equal}()} \cr
@@ -245,6 +255,8 @@ Other numpy ops: \cr
 \code{\link{op_var}()} \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_zeros}()} \cr
@@ -254,6 +266,7 @@ Other ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -272,6 +285,7 @@ Other ops: \cr
 \code{\link{op_associative_scan}()} \cr
 \code{\link{op_average}()} \cr
 \code{\link{op_average_pool}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_batch_normalization}()} \cr
 \code{\link{op_binary_crossentropy}()} \cr
 \code{\link{op_bincount}()} \cr
@@ -282,9 +296,11 @@ Other ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
 \code{\link{op_cast}()} \cr
 \code{\link{op_categorical_crossentropy}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_celu}()} \cr
 \code{\link{op_cholesky}()} \cr
@@ -297,6 +313,7 @@ Other ops: \cr
 \code{\link{op_convert_to_numpy}()} \cr
 \code{\link{op_convert_to_tensor}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -307,6 +324,7 @@ Other ops: \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
 \code{\link{op_custom_gradient}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_depthwise_conv}()} \cr
 \code{\link{op_det}()} \cr
 \code{\link{op_diag}()} \cr
@@ -346,10 +364,13 @@ Other ops: \cr
 \code{\link{op_glu}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
 \code{\link{op_hard_shrink}()} \cr
 \code{\link{op_hard_sigmoid}()} \cr
 \code{\link{op_hard_silu}()} \cr
 \code{\link{op_hard_tanh}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -376,6 +397,8 @@ Other ops: \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
 \code{\link{op_istft}()} \cr
+\code{\link{op_kaiser}()} \cr
+\code{\link{op_layer_normalization}()} \cr
 \code{\link{op_leaky_relu}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
@@ -474,6 +497,7 @@ Other ops: \cr
 \code{\link{op_sort}()} \cr
 \code{\link{op_sparse_categorical_crossentropy}()} \cr
 \code{\link{op_sparse_plus}()} \cr
+\code{\link{op_sparse_sigmoid}()} \cr
 \code{\link{op_sparsemax}()} \cr
 \code{\link{op_split}()} \cr
 \code{\link{op_sqrt}()} \cr
@@ -511,6 +535,8 @@ Other ops: \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
 \code{\link{op_vectorized_map}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_while_loop}()} \cr
diff --git a/man/op_norm.Rd b/man/op_norm.Rd
index ccdb4029d..b713d1a49 100644
--- a/man/op_norm.Rd
+++ b/man/op_norm.Rd
@@ -93,6 +93,7 @@ Other ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -111,6 +112,7 @@ Other ops: \cr
 \code{\link{op_associative_scan}()} \cr
 \code{\link{op_average}()} \cr
 \code{\link{op_average_pool}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_batch_normalization}()} \cr
 \code{\link{op_binary_crossentropy}()} \cr
 \code{\link{op_bincount}()} \cr
@@ -121,9 +123,11 @@ Other ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
 \code{\link{op_cast}()} \cr
 \code{\link{op_categorical_crossentropy}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_celu}()} \cr
 \code{\link{op_cholesky}()} \cr
@@ -136,6 +140,7 @@ Other ops: \cr
 \code{\link{op_convert_to_numpy}()} \cr
 \code{\link{op_convert_to_tensor}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -146,6 +151,7 @@ Other ops: \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
 \code{\link{op_custom_gradient}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_depthwise_conv}()} \cr
 \code{\link{op_det}()} \cr
 \code{\link{op_diag}()} \cr
@@ -185,10 +191,13 @@ Other ops: \cr
 \code{\link{op_glu}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
 \code{\link{op_hard_shrink}()} \cr
 \code{\link{op_hard_sigmoid}()} \cr
 \code{\link{op_hard_silu}()} \cr
 \code{\link{op_hard_tanh}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -215,6 +224,8 @@ Other ops: \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
 \code{\link{op_istft}()} \cr
+\code{\link{op_kaiser}()} \cr
+\code{\link{op_layer_normalization}()} \cr
 \code{\link{op_leaky_relu}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
@@ -313,6 +324,7 @@ Other ops: \cr
 \code{\link{op_sort}()} \cr
 \code{\link{op_sparse_categorical_crossentropy}()} \cr
 \code{\link{op_sparse_plus}()} \cr
+\code{\link{op_sparse_sigmoid}()} \cr
 \code{\link{op_sparsemax}()} \cr
 \code{\link{op_split}()} \cr
 \code{\link{op_sqrt}()} \cr
@@ -350,6 +362,8 @@ Other ops: \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
 \code{\link{op_vectorized_map}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_while_loop}()} \cr
diff --git a/man/op_normalize.Rd b/man/op_normalize.Rd
index ce7d3ac66..1fde1cc04 100644
--- a/man/op_normalize.Rd
+++ b/man/op_normalize.Rd
@@ -60,6 +60,7 @@ Other nn ops: \cr
 \code{\link{op_hard_sigmoid}()} \cr
 \code{\link{op_hard_silu}()} \cr
 \code{\link{op_hard_tanh}()} \cr
+\code{\link{op_layer_normalization}()} \cr
 \code{\link{op_leaky_relu}()} \cr
 \code{\link{op_log_sigmoid}()} \cr
 \code{\link{op_log_softmax}()} \cr
@@ -82,6 +83,7 @@ Other nn ops: \cr
 \code{\link{op_softsign}()} \cr
 \code{\link{op_sparse_categorical_crossentropy}()} \cr
 \code{\link{op_sparse_plus}()} \cr
+\code{\link{op_sparse_sigmoid}()} \cr
 \code{\link{op_sparsemax}()} \cr
 \code{\link{op_squareplus}()} \cr
 \code{\link{op_tanh_shrink}()} \cr
@@ -92,6 +94,7 @@ Other ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -110,6 +113,7 @@ Other ops: \cr
 \code{\link{op_associative_scan}()} \cr
 \code{\link{op_average}()} \cr
 \code{\link{op_average_pool}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_batch_normalization}()} \cr
 \code{\link{op_binary_crossentropy}()} \cr
 \code{\link{op_bincount}()} \cr
@@ -120,9 +124,11 @@ Other ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
 \code{\link{op_cast}()} \cr
 \code{\link{op_categorical_crossentropy}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_celu}()} \cr
 \code{\link{op_cholesky}()} \cr
@@ -135,6 +141,7 @@ Other ops: \cr
 \code{\link{op_convert_to_numpy}()} \cr
 \code{\link{op_convert_to_tensor}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -145,6 +152,7 @@ Other ops: \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
 \code{\link{op_custom_gradient}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_depthwise_conv}()} \cr
 \code{\link{op_det}()} \cr
 \code{\link{op_diag}()} \cr
@@ -184,10 +192,13 @@ Other ops: \cr
 \code{\link{op_glu}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
 \code{\link{op_hard_shrink}()} \cr
 \code{\link{op_hard_sigmoid}()} \cr
 \code{\link{op_hard_silu}()} \cr
 \code{\link{op_hard_tanh}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -214,6 +225,8 @@ Other ops: \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
 \code{\link{op_istft}()} \cr
+\code{\link{op_kaiser}()} \cr
+\code{\link{op_layer_normalization}()} \cr
 \code{\link{op_leaky_relu}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
@@ -312,6 +325,7 @@ Other ops: \cr
 \code{\link{op_sort}()} \cr
 \code{\link{op_sparse_categorical_crossentropy}()} \cr
 \code{\link{op_sparse_plus}()} \cr
+\code{\link{op_sparse_sigmoid}()} \cr
 \code{\link{op_sparsemax}()} \cr
 \code{\link{op_split}()} \cr
 \code{\link{op_sqrt}()} \cr
@@ -349,6 +363,8 @@ Other ops: \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
 \code{\link{op_vectorized_map}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_while_loop}()} \cr
diff --git a/man/op_not_equal.Rd b/man/op_not_equal.Rd
index 6245fd536..88a843e0e 100644
--- a/man/op_not_equal.Rd
+++ b/man/op_not_equal.Rd
@@ -47,6 +47,7 @@ Other numpy ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -63,6 +64,7 @@ Other numpy ops: \cr
 \code{\link{op_argsort}()} \cr
 \code{\link{op_array}()} \cr
 \code{\link{op_average}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_bincount}()} \cr
 \code{\link{op_bitwise_and}()} \cr
 \code{\link{op_bitwise_invert}()} \cr
@@ -71,12 +73,15 @@ Other numpy ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_clip}()} \cr
 \code{\link{op_concatenate}()} \cr
 \code{\link{op_conj}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -85,6 +90,7 @@ Other numpy ops: \cr
 \code{\link{op_ctc_decode}()} \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_diag}()} \cr
 \code{\link{op_diagflat}()} \cr
 \code{\link{op_diagonal}()} \cr
@@ -109,6 +115,9 @@ Other numpy ops: \cr
 \code{\link{op_get_item}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -118,6 +127,7 @@ Other numpy ops: \cr
 \code{\link{op_isfinite}()} \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
+\code{\link{op_kaiser}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
 \code{\link{op_less_equal}()} \cr
@@ -197,6 +207,8 @@ Other numpy ops: \cr
 \code{\link{op_var}()} \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_zeros}()} \cr
@@ -206,6 +218,7 @@ Other ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -224,6 +237,7 @@ Other ops: \cr
 \code{\link{op_associative_scan}()} \cr
 \code{\link{op_average}()} \cr
 \code{\link{op_average_pool}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_batch_normalization}()} \cr
 \code{\link{op_binary_crossentropy}()} \cr
 \code{\link{op_bincount}()} \cr
@@ -234,9 +248,11 @@ Other ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
 \code{\link{op_cast}()} \cr
 \code{\link{op_categorical_crossentropy}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_celu}()} \cr
 \code{\link{op_cholesky}()} \cr
@@ -249,6 +265,7 @@ Other ops: \cr
 \code{\link{op_convert_to_numpy}()} \cr
 \code{\link{op_convert_to_tensor}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -259,6 +276,7 @@ Other ops: \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
 \code{\link{op_custom_gradient}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_depthwise_conv}()} \cr
 \code{\link{op_det}()} \cr
 \code{\link{op_diag}()} \cr
@@ -298,10 +316,13 @@ Other ops: \cr
 \code{\link{op_glu}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
 \code{\link{op_hard_shrink}()} \cr
 \code{\link{op_hard_sigmoid}()} \cr
 \code{\link{op_hard_silu}()} \cr
 \code{\link{op_hard_tanh}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -328,6 +349,8 @@ Other ops: \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
 \code{\link{op_istft}()} \cr
+\code{\link{op_kaiser}()} \cr
+\code{\link{op_layer_normalization}()} \cr
 \code{\link{op_leaky_relu}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
@@ -426,6 +449,7 @@ Other ops: \cr
 \code{\link{op_sort}()} \cr
 \code{\link{op_sparse_categorical_crossentropy}()} \cr
 \code{\link{op_sparse_plus}()} \cr
+\code{\link{op_sparse_sigmoid}()} \cr
 \code{\link{op_sparsemax}()} \cr
 \code{\link{op_split}()} \cr
 \code{\link{op_sqrt}()} \cr
@@ -463,6 +487,8 @@ Other ops: \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
 \code{\link{op_vectorized_map}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_while_loop}()} \cr
diff --git a/man/op_one_hot.Rd b/man/op_one_hot.Rd
index f61cb1bb0..7380769ec 100644
--- a/man/op_one_hot.Rd
+++ b/man/op_one_hot.Rd
@@ -94,6 +94,7 @@ Other nn ops: \cr
 \code{\link{op_hard_sigmoid}()} \cr
 \code{\link{op_hard_silu}()} \cr
 \code{\link{op_hard_tanh}()} \cr
+\code{\link{op_layer_normalization}()} \cr
 \code{\link{op_leaky_relu}()} \cr
 \code{\link{op_log_sigmoid}()} \cr
 \code{\link{op_log_softmax}()} \cr
@@ -116,6 +117,7 @@ Other nn ops: \cr
 \code{\link{op_softsign}()} \cr
 \code{\link{op_sparse_categorical_crossentropy}()} \cr
 \code{\link{op_sparse_plus}()} \cr
+\code{\link{op_sparse_sigmoid}()} \cr
 \code{\link{op_sparsemax}()} \cr
 \code{\link{op_squareplus}()} \cr
 \code{\link{op_tanh_shrink}()} \cr
@@ -126,6 +128,7 @@ Other ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -144,6 +147,7 @@ Other ops: \cr
 \code{\link{op_associative_scan}()} \cr
 \code{\link{op_average}()} \cr
 \code{\link{op_average_pool}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_batch_normalization}()} \cr
 \code{\link{op_binary_crossentropy}()} \cr
 \code{\link{op_bincount}()} \cr
@@ -154,9 +158,11 @@ Other ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
 \code{\link{op_cast}()} \cr
 \code{\link{op_categorical_crossentropy}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_celu}()} \cr
 \code{\link{op_cholesky}()} \cr
@@ -169,6 +175,7 @@ Other ops: \cr
 \code{\link{op_convert_to_numpy}()} \cr
 \code{\link{op_convert_to_tensor}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -179,6 +186,7 @@ Other ops: \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
 \code{\link{op_custom_gradient}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_depthwise_conv}()} \cr
 \code{\link{op_det}()} \cr
 \code{\link{op_diag}()} \cr
@@ -218,10 +226,13 @@ Other ops: \cr
 \code{\link{op_glu}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
 \code{\link{op_hard_shrink}()} \cr
 \code{\link{op_hard_sigmoid}()} \cr
 \code{\link{op_hard_silu}()} \cr
 \code{\link{op_hard_tanh}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -248,6 +259,8 @@ Other ops: \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
 \code{\link{op_istft}()} \cr
+\code{\link{op_kaiser}()} \cr
+\code{\link{op_layer_normalization}()} \cr
 \code{\link{op_leaky_relu}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
@@ -346,6 +359,7 @@ Other ops: \cr
 \code{\link{op_sort}()} \cr
 \code{\link{op_sparse_categorical_crossentropy}()} \cr
 \code{\link{op_sparse_plus}()} \cr
+\code{\link{op_sparse_sigmoid}()} \cr
 \code{\link{op_sparsemax}()} \cr
 \code{\link{op_split}()} \cr
 \code{\link{op_sqrt}()} \cr
@@ -383,6 +397,8 @@ Other ops: \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
 \code{\link{op_vectorized_map}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_while_loop}()} \cr
diff --git a/man/op_ones.Rd b/man/op_ones.Rd
index 81fe08b51..6106b86df 100644
--- a/man/op_ones.Rd
+++ b/man/op_ones.Rd
@@ -26,6 +26,7 @@ Other numpy ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -42,6 +43,7 @@ Other numpy ops: \cr
 \code{\link{op_argsort}()} \cr
 \code{\link{op_array}()} \cr
 \code{\link{op_average}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_bincount}()} \cr
 \code{\link{op_bitwise_and}()} \cr
 \code{\link{op_bitwise_invert}()} \cr
@@ -50,12 +52,15 @@ Other numpy ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_clip}()} \cr
 \code{\link{op_concatenate}()} \cr
 \code{\link{op_conj}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -64,6 +69,7 @@ Other numpy ops: \cr
 \code{\link{op_ctc_decode}()} \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_diag}()} \cr
 \code{\link{op_diagflat}()} \cr
 \code{\link{op_diagonal}()} \cr
@@ -88,6 +94,9 @@ Other numpy ops: \cr
 \code{\link{op_get_item}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -97,6 +106,7 @@ Other numpy ops: \cr
 \code{\link{op_isfinite}()} \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
+\code{\link{op_kaiser}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
 \code{\link{op_less_equal}()} \cr
@@ -176,6 +186,8 @@ Other numpy ops: \cr
 \code{\link{op_var}()} \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_zeros}()} \cr
@@ -185,6 +197,7 @@ Other ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -203,6 +216,7 @@ Other ops: \cr
 \code{\link{op_associative_scan}()} \cr
 \code{\link{op_average}()} \cr
 \code{\link{op_average_pool}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_batch_normalization}()} \cr
 \code{\link{op_binary_crossentropy}()} \cr
 \code{\link{op_bincount}()} \cr
@@ -213,9 +227,11 @@ Other ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
 \code{\link{op_cast}()} \cr
 \code{\link{op_categorical_crossentropy}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_celu}()} \cr
 \code{\link{op_cholesky}()} \cr
@@ -228,6 +244,7 @@ Other ops: \cr
 \code{\link{op_convert_to_numpy}()} \cr
 \code{\link{op_convert_to_tensor}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -238,6 +255,7 @@ Other ops: \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
 \code{\link{op_custom_gradient}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_depthwise_conv}()} \cr
 \code{\link{op_det}()} \cr
 \code{\link{op_diag}()} \cr
@@ -277,10 +295,13 @@ Other ops: \cr
 \code{\link{op_glu}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
 \code{\link{op_hard_shrink}()} \cr
 \code{\link{op_hard_sigmoid}()} \cr
 \code{\link{op_hard_silu}()} \cr
 \code{\link{op_hard_tanh}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -307,6 +328,8 @@ Other ops: \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
 \code{\link{op_istft}()} \cr
+\code{\link{op_kaiser}()} \cr
+\code{\link{op_layer_normalization}()} \cr
 \code{\link{op_leaky_relu}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
@@ -405,6 +428,7 @@ Other ops: \cr
 \code{\link{op_sort}()} \cr
 \code{\link{op_sparse_categorical_crossentropy}()} \cr
 \code{\link{op_sparse_plus}()} \cr
+\code{\link{op_sparse_sigmoid}()} \cr
 \code{\link{op_sparsemax}()} \cr
 \code{\link{op_split}()} \cr
 \code{\link{op_sqrt}()} \cr
@@ -442,6 +466,8 @@ Other ops: \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
 \code{\link{op_vectorized_map}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_while_loop}()} \cr
diff --git a/man/op_ones_like.Rd b/man/op_ones_like.Rd
index 51eccb913..16300eddd 100644
--- a/man/op_ones_like.Rd
+++ b/man/op_ones_like.Rd
@@ -26,6 +26,7 @@ Other numpy ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -42,6 +43,7 @@ Other numpy ops: \cr
 \code{\link{op_argsort}()} \cr
 \code{\link{op_array}()} \cr
 \code{\link{op_average}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_bincount}()} \cr
 \code{\link{op_bitwise_and}()} \cr
 \code{\link{op_bitwise_invert}()} \cr
@@ -50,12 +52,15 @@ Other numpy ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_clip}()} \cr
 \code{\link{op_concatenate}()} \cr
 \code{\link{op_conj}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -64,6 +69,7 @@ Other numpy ops: \cr
 \code{\link{op_ctc_decode}()} \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_diag}()} \cr
 \code{\link{op_diagflat}()} \cr
 \code{\link{op_diagonal}()} \cr
@@ -88,6 +94,9 @@ Other numpy ops: \cr
 \code{\link{op_get_item}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -97,6 +106,7 @@ Other numpy ops: \cr
 \code{\link{op_isfinite}()} \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
+\code{\link{op_kaiser}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
 \code{\link{op_less_equal}()} \cr
@@ -176,6 +186,8 @@ Other numpy ops: \cr
 \code{\link{op_var}()} \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_zeros}()} \cr
@@ -185,6 +197,7 @@ Other ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -203,6 +216,7 @@ Other ops: \cr
 \code{\link{op_associative_scan}()} \cr
 \code{\link{op_average}()} \cr
 \code{\link{op_average_pool}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_batch_normalization}()} \cr
 \code{\link{op_binary_crossentropy}()} \cr
 \code{\link{op_bincount}()} \cr
@@ -213,9 +227,11 @@ Other ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
 \code{\link{op_cast}()} \cr
 \code{\link{op_categorical_crossentropy}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_celu}()} \cr
 \code{\link{op_cholesky}()} \cr
@@ -228,6 +244,7 @@ Other ops: \cr
 \code{\link{op_convert_to_numpy}()} \cr
 \code{\link{op_convert_to_tensor}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -238,6 +255,7 @@ Other ops: \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
 \code{\link{op_custom_gradient}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_depthwise_conv}()} \cr
 \code{\link{op_det}()} \cr
 \code{\link{op_diag}()} \cr
@@ -277,10 +295,13 @@ Other ops: \cr
 \code{\link{op_glu}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
 \code{\link{op_hard_shrink}()} \cr
 \code{\link{op_hard_sigmoid}()} \cr
 \code{\link{op_hard_silu}()} \cr
 \code{\link{op_hard_tanh}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -307,6 +328,8 @@ Other ops: \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
 \code{\link{op_istft}()} \cr
+\code{\link{op_kaiser}()} \cr
+\code{\link{op_layer_normalization}()} \cr
 \code{\link{op_leaky_relu}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
@@ -405,6 +428,7 @@ Other ops: \cr
 \code{\link{op_sort}()} \cr
 \code{\link{op_sparse_categorical_crossentropy}()} \cr
 \code{\link{op_sparse_plus}()} \cr
+\code{\link{op_sparse_sigmoid}()} \cr
 \code{\link{op_sparsemax}()} \cr
 \code{\link{op_split}()} \cr
 \code{\link{op_sqrt}()} \cr
@@ -442,6 +466,8 @@ Other ops: \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
 \code{\link{op_vectorized_map}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_while_loop}()} \cr
diff --git a/man/op_outer.Rd b/man/op_outer.Rd
index 2b54bab89..7e46784a2 100644
--- a/man/op_outer.Rd
+++ b/man/op_outer.Rd
@@ -29,6 +29,7 @@ Other numpy ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -45,6 +46,7 @@ Other numpy ops: \cr
 \code{\link{op_argsort}()} \cr
 \code{\link{op_array}()} \cr
 \code{\link{op_average}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_bincount}()} \cr
 \code{\link{op_bitwise_and}()} \cr
 \code{\link{op_bitwise_invert}()} \cr
@@ -53,12 +55,15 @@ Other numpy ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_clip}()} \cr
 \code{\link{op_concatenate}()} \cr
 \code{\link{op_conj}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -67,6 +72,7 @@ Other numpy ops: \cr
 \code{\link{op_ctc_decode}()} \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_diag}()} \cr
 \code{\link{op_diagflat}()} \cr
 \code{\link{op_diagonal}()} \cr
@@ -91,6 +97,9 @@ Other numpy ops: \cr
 \code{\link{op_get_item}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -100,6 +109,7 @@ Other numpy ops: \cr
 \code{\link{op_isfinite}()} \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
+\code{\link{op_kaiser}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
 \code{\link{op_less_equal}()} \cr
@@ -179,6 +189,8 @@ Other numpy ops: \cr
 \code{\link{op_var}()} \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_zeros}()} \cr
@@ -188,6 +200,7 @@ Other ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -206,6 +219,7 @@ Other ops: \cr
 \code{\link{op_associative_scan}()} \cr
 \code{\link{op_average}()} \cr
 \code{\link{op_average_pool}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_batch_normalization}()} \cr
 \code{\link{op_binary_crossentropy}()} \cr
 \code{\link{op_bincount}()} \cr
@@ -216,9 +230,11 @@ Other ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
 \code{\link{op_cast}()} \cr
 \code{\link{op_categorical_crossentropy}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_celu}()} \cr
 \code{\link{op_cholesky}()} \cr
@@ -231,6 +247,7 @@ Other ops: \cr
 \code{\link{op_convert_to_numpy}()} \cr
 \code{\link{op_convert_to_tensor}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -241,6 +258,7 @@ Other ops: \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
 \code{\link{op_custom_gradient}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_depthwise_conv}()} \cr
 \code{\link{op_det}()} \cr
 \code{\link{op_diag}()} \cr
@@ -280,10 +298,13 @@ Other ops: \cr
 \code{\link{op_glu}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
 \code{\link{op_hard_shrink}()} \cr
 \code{\link{op_hard_sigmoid}()} \cr
 \code{\link{op_hard_silu}()} \cr
 \code{\link{op_hard_tanh}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -310,6 +331,8 @@ Other ops: \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
 \code{\link{op_istft}()} \cr
+\code{\link{op_kaiser}()} \cr
+\code{\link{op_layer_normalization}()} \cr
 \code{\link{op_leaky_relu}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
@@ -408,6 +431,7 @@ Other ops: \cr
 \code{\link{op_sort}()} \cr
 \code{\link{op_sparse_categorical_crossentropy}()} \cr
 \code{\link{op_sparse_plus}()} \cr
+\code{\link{op_sparse_sigmoid}()} \cr
 \code{\link{op_sparsemax}()} \cr
 \code{\link{op_split}()} \cr
 \code{\link{op_sqrt}()} \cr
@@ -445,6 +469,8 @@ Other ops: \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
 \code{\link{op_vectorized_map}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_while_loop}()} \cr
diff --git a/man/op_pad.Rd b/man/op_pad.Rd
index a31743e9f..9da4a403b 100644
--- a/man/op_pad.Rd
+++ b/man/op_pad.Rd
@@ -51,6 +51,7 @@ Other numpy ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -67,6 +68,7 @@ Other numpy ops: \cr
 \code{\link{op_argsort}()} \cr
 \code{\link{op_array}()} \cr
 \code{\link{op_average}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_bincount}()} \cr
 \code{\link{op_bitwise_and}()} \cr
 \code{\link{op_bitwise_invert}()} \cr
@@ -75,12 +77,15 @@ Other numpy ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_clip}()} \cr
 \code{\link{op_concatenate}()} \cr
 \code{\link{op_conj}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -89,6 +94,7 @@ Other numpy ops: \cr
 \code{\link{op_ctc_decode}()} \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_diag}()} \cr
 \code{\link{op_diagflat}()} \cr
 \code{\link{op_diagonal}()} \cr
@@ -113,6 +119,9 @@ Other numpy ops: \cr
 \code{\link{op_get_item}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -122,6 +131,7 @@ Other numpy ops: \cr
 \code{\link{op_isfinite}()} \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
+\code{\link{op_kaiser}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
 \code{\link{op_less_equal}()} \cr
@@ -201,6 +211,8 @@ Other numpy ops: \cr
 \code{\link{op_var}()} \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_zeros}()} \cr
@@ -210,6 +222,7 @@ Other ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -228,6 +241,7 @@ Other ops: \cr
 \code{\link{op_associative_scan}()} \cr
 \code{\link{op_average}()} \cr
 \code{\link{op_average_pool}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_batch_normalization}()} \cr
 \code{\link{op_binary_crossentropy}()} \cr
 \code{\link{op_bincount}()} \cr
@@ -238,9 +252,11 @@ Other ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
 \code{\link{op_cast}()} \cr
 \code{\link{op_categorical_crossentropy}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_celu}()} \cr
 \code{\link{op_cholesky}()} \cr
@@ -253,6 +269,7 @@ Other ops: \cr
 \code{\link{op_convert_to_numpy}()} \cr
 \code{\link{op_convert_to_tensor}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -263,6 +280,7 @@ Other ops: \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
 \code{\link{op_custom_gradient}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_depthwise_conv}()} \cr
 \code{\link{op_det}()} \cr
 \code{\link{op_diag}()} \cr
@@ -302,10 +320,13 @@ Other ops: \cr
 \code{\link{op_glu}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
 \code{\link{op_hard_shrink}()} \cr
 \code{\link{op_hard_sigmoid}()} \cr
 \code{\link{op_hard_silu}()} \cr
 \code{\link{op_hard_tanh}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -332,6 +353,8 @@ Other ops: \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
 \code{\link{op_istft}()} \cr
+\code{\link{op_kaiser}()} \cr
+\code{\link{op_layer_normalization}()} \cr
 \code{\link{op_leaky_relu}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
@@ -430,6 +453,7 @@ Other ops: \cr
 \code{\link{op_sort}()} \cr
 \code{\link{op_sparse_categorical_crossentropy}()} \cr
 \code{\link{op_sparse_plus}()} \cr
+\code{\link{op_sparse_sigmoid}()} \cr
 \code{\link{op_sparsemax}()} \cr
 \code{\link{op_split}()} \cr
 \code{\link{op_sqrt}()} \cr
@@ -467,6 +491,8 @@ Other ops: \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
 \code{\link{op_vectorized_map}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_while_loop}()} \cr
diff --git a/man/op_polar.Rd b/man/op_polar.Rd
index a39b7a9a9..ccf49b5f9 100644
--- a/man/op_polar.Rd
+++ b/man/op_polar.Rd
@@ -67,6 +67,7 @@ Other nn ops: \cr
 \code{\link{op_hard_sigmoid}()} \cr
 \code{\link{op_hard_silu}()} \cr
 \code{\link{op_hard_tanh}()} \cr
+\code{\link{op_layer_normalization}()} \cr
 \code{\link{op_leaky_relu}()} \cr
 \code{\link{op_log_sigmoid}()} \cr
 \code{\link{op_log_softmax}()} \cr
@@ -89,6 +90,7 @@ Other nn ops: \cr
 \code{\link{op_softsign}()} \cr
 \code{\link{op_sparse_categorical_crossentropy}()} \cr
 \code{\link{op_sparse_plus}()} \cr
+\code{\link{op_sparse_sigmoid}()} \cr
 \code{\link{op_sparsemax}()} \cr
 \code{\link{op_squareplus}()} \cr
 \code{\link{op_tanh_shrink}()} \cr
@@ -99,6 +101,7 @@ Other ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -117,6 +120,7 @@ Other ops: \cr
 \code{\link{op_associative_scan}()} \cr
 \code{\link{op_average}()} \cr
 \code{\link{op_average_pool}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_batch_normalization}()} \cr
 \code{\link{op_binary_crossentropy}()} \cr
 \code{\link{op_bincount}()} \cr
@@ -127,9 +131,11 @@ Other ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
 \code{\link{op_cast}()} \cr
 \code{\link{op_categorical_crossentropy}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_celu}()} \cr
 \code{\link{op_cholesky}()} \cr
@@ -142,6 +148,7 @@ Other ops: \cr
 \code{\link{op_convert_to_numpy}()} \cr
 \code{\link{op_convert_to_tensor}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -152,6 +159,7 @@ Other ops: \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
 \code{\link{op_custom_gradient}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_depthwise_conv}()} \cr
 \code{\link{op_det}()} \cr
 \code{\link{op_diag}()} \cr
@@ -191,10 +199,13 @@ Other ops: \cr
 \code{\link{op_glu}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
 \code{\link{op_hard_shrink}()} \cr
 \code{\link{op_hard_sigmoid}()} \cr
 \code{\link{op_hard_silu}()} \cr
 \code{\link{op_hard_tanh}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -221,6 +232,8 @@ Other ops: \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
 \code{\link{op_istft}()} \cr
+\code{\link{op_kaiser}()} \cr
+\code{\link{op_layer_normalization}()} \cr
 \code{\link{op_leaky_relu}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
@@ -319,6 +332,7 @@ Other ops: \cr
 \code{\link{op_sort}()} \cr
 \code{\link{op_sparse_categorical_crossentropy}()} \cr
 \code{\link{op_sparse_plus}()} \cr
+\code{\link{op_sparse_sigmoid}()} \cr
 \code{\link{op_sparsemax}()} \cr
 \code{\link{op_split}()} \cr
 \code{\link{op_sqrt}()} \cr
@@ -356,6 +370,8 @@ Other ops: \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
 \code{\link{op_vectorized_map}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_while_loop}()} \cr
diff --git a/man/op_power.Rd b/man/op_power.Rd
index e1a81ddfc..5c96ad46f 100644
--- a/man/op_power.Rd
+++ b/man/op_power.Rd
@@ -47,6 +47,7 @@ Other numpy ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -63,6 +64,7 @@ Other numpy ops: \cr
 \code{\link{op_argsort}()} \cr
 \code{\link{op_array}()} \cr
 \code{\link{op_average}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_bincount}()} \cr
 \code{\link{op_bitwise_and}()} \cr
 \code{\link{op_bitwise_invert}()} \cr
@@ -71,12 +73,15 @@ Other numpy ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_clip}()} \cr
 \code{\link{op_concatenate}()} \cr
 \code{\link{op_conj}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -85,6 +90,7 @@ Other numpy ops: \cr
 \code{\link{op_ctc_decode}()} \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_diag}()} \cr
 \code{\link{op_diagflat}()} \cr
 \code{\link{op_diagonal}()} \cr
@@ -109,6 +115,9 @@ Other numpy ops: \cr
 \code{\link{op_get_item}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -118,6 +127,7 @@ Other numpy ops: \cr
 \code{\link{op_isfinite}()} \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
+\code{\link{op_kaiser}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
 \code{\link{op_less_equal}()} \cr
@@ -197,6 +207,8 @@ Other numpy ops: \cr
 \code{\link{op_var}()} \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_zeros}()} \cr
@@ -206,6 +218,7 @@ Other ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -224,6 +237,7 @@ Other ops: \cr
 \code{\link{op_associative_scan}()} \cr
 \code{\link{op_average}()} \cr
 \code{\link{op_average_pool}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_batch_normalization}()} \cr
 \code{\link{op_binary_crossentropy}()} \cr
 \code{\link{op_bincount}()} \cr
@@ -234,9 +248,11 @@ Other ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
 \code{\link{op_cast}()} \cr
 \code{\link{op_categorical_crossentropy}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_celu}()} \cr
 \code{\link{op_cholesky}()} \cr
@@ -249,6 +265,7 @@ Other ops: \cr
 \code{\link{op_convert_to_numpy}()} \cr
 \code{\link{op_convert_to_tensor}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -259,6 +276,7 @@ Other ops: \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
 \code{\link{op_custom_gradient}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_depthwise_conv}()} \cr
 \code{\link{op_det}()} \cr
 \code{\link{op_diag}()} \cr
@@ -298,10 +316,13 @@ Other ops: \cr
 \code{\link{op_glu}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
 \code{\link{op_hard_shrink}()} \cr
 \code{\link{op_hard_sigmoid}()} \cr
 \code{\link{op_hard_silu}()} \cr
 \code{\link{op_hard_tanh}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -328,6 +349,8 @@ Other ops: \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
 \code{\link{op_istft}()} \cr
+\code{\link{op_kaiser}()} \cr
+\code{\link{op_layer_normalization}()} \cr
 \code{\link{op_leaky_relu}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
@@ -426,6 +449,7 @@ Other ops: \cr
 \code{\link{op_sort}()} \cr
 \code{\link{op_sparse_categorical_crossentropy}()} \cr
 \code{\link{op_sparse_plus}()} \cr
+\code{\link{op_sparse_sigmoid}()} \cr
 \code{\link{op_sparsemax}()} \cr
 \code{\link{op_split}()} \cr
 \code{\link{op_sqrt}()} \cr
@@ -463,6 +487,8 @@ Other ops: \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
 \code{\link{op_vectorized_map}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_while_loop}()} \cr
diff --git a/man/op_prod.Rd b/man/op_prod.Rd
index b0b0e4070..6112e2999 100644
--- a/man/op_prod.Rd
+++ b/man/op_prod.Rd
@@ -33,6 +33,7 @@ Other numpy ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -49,6 +50,7 @@ Other numpy ops: \cr
 \code{\link{op_argsort}()} \cr
 \code{\link{op_array}()} \cr
 \code{\link{op_average}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_bincount}()} \cr
 \code{\link{op_bitwise_and}()} \cr
 \code{\link{op_bitwise_invert}()} \cr
@@ -57,12 +59,15 @@ Other numpy ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_clip}()} \cr
 \code{\link{op_concatenate}()} \cr
 \code{\link{op_conj}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -71,6 +76,7 @@ Other numpy ops: \cr
 \code{\link{op_ctc_decode}()} \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_diag}()} \cr
 \code{\link{op_diagflat}()} \cr
 \code{\link{op_diagonal}()} \cr
@@ -95,6 +101,9 @@ Other numpy ops: \cr
 \code{\link{op_get_item}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -104,6 +113,7 @@ Other numpy ops: \cr
 \code{\link{op_isfinite}()} \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
+\code{\link{op_kaiser}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
 \code{\link{op_less_equal}()} \cr
@@ -183,6 +193,8 @@ Other numpy ops: \cr
 \code{\link{op_var}()} \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_zeros}()} \cr
@@ -192,6 +204,7 @@ Other ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -210,6 +223,7 @@ Other ops: \cr
 \code{\link{op_associative_scan}()} \cr
 \code{\link{op_average}()} \cr
 \code{\link{op_average_pool}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_batch_normalization}()} \cr
 \code{\link{op_binary_crossentropy}()} \cr
 \code{\link{op_bincount}()} \cr
@@ -220,9 +234,11 @@ Other ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
 \code{\link{op_cast}()} \cr
 \code{\link{op_categorical_crossentropy}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_celu}()} \cr
 \code{\link{op_cholesky}()} \cr
@@ -235,6 +251,7 @@ Other ops: \cr
 \code{\link{op_convert_to_numpy}()} \cr
 \code{\link{op_convert_to_tensor}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -245,6 +262,7 @@ Other ops: \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
 \code{\link{op_custom_gradient}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_depthwise_conv}()} \cr
 \code{\link{op_det}()} \cr
 \code{\link{op_diag}()} \cr
@@ -284,10 +302,13 @@ Other ops: \cr
 \code{\link{op_glu}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
 \code{\link{op_hard_shrink}()} \cr
 \code{\link{op_hard_sigmoid}()} \cr
 \code{\link{op_hard_silu}()} \cr
 \code{\link{op_hard_tanh}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -314,6 +335,8 @@ Other ops: \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
 \code{\link{op_istft}()} \cr
+\code{\link{op_kaiser}()} \cr
+\code{\link{op_layer_normalization}()} \cr
 \code{\link{op_leaky_relu}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
@@ -412,6 +435,7 @@ Other ops: \cr
 \code{\link{op_sort}()} \cr
 \code{\link{op_sparse_categorical_crossentropy}()} \cr
 \code{\link{op_sparse_plus}()} \cr
+\code{\link{op_sparse_sigmoid}()} \cr
 \code{\link{op_sparsemax}()} \cr
 \code{\link{op_split}()} \cr
 \code{\link{op_sqrt}()} \cr
@@ -449,6 +473,8 @@ Other ops: \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
 \code{\link{op_vectorized_map}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_while_loop}()} \cr
diff --git a/man/op_psnr.Rd b/man/op_psnr.Rd
index c9cb65216..71dbd6d7f 100644
--- a/man/op_psnr.Rd
+++ b/man/op_psnr.Rd
@@ -54,6 +54,7 @@ Other nn ops: \cr
 \code{\link{op_hard_sigmoid}()} \cr
 \code{\link{op_hard_silu}()} \cr
 \code{\link{op_hard_tanh}()} \cr
+\code{\link{op_layer_normalization}()} \cr
 \code{\link{op_leaky_relu}()} \cr
 \code{\link{op_log_sigmoid}()} \cr
 \code{\link{op_log_softmax}()} \cr
@@ -76,6 +77,7 @@ Other nn ops: \cr
 \code{\link{op_softsign}()} \cr
 \code{\link{op_sparse_categorical_crossentropy}()} \cr
 \code{\link{op_sparse_plus}()} \cr
+\code{\link{op_sparse_sigmoid}()} \cr
 \code{\link{op_sparsemax}()} \cr
 \code{\link{op_squareplus}()} \cr
 \code{\link{op_tanh_shrink}()} \cr
@@ -86,6 +88,7 @@ Other ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -104,6 +107,7 @@ Other ops: \cr
 \code{\link{op_associative_scan}()} \cr
 \code{\link{op_average}()} \cr
 \code{\link{op_average_pool}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_batch_normalization}()} \cr
 \code{\link{op_binary_crossentropy}()} \cr
 \code{\link{op_bincount}()} \cr
@@ -114,9 +118,11 @@ Other ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
 \code{\link{op_cast}()} \cr
 \code{\link{op_categorical_crossentropy}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_celu}()} \cr
 \code{\link{op_cholesky}()} \cr
@@ -129,6 +135,7 @@ Other ops: \cr
 \code{\link{op_convert_to_numpy}()} \cr
 \code{\link{op_convert_to_tensor}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -139,6 +146,7 @@ Other ops: \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
 \code{\link{op_custom_gradient}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_depthwise_conv}()} \cr
 \code{\link{op_det}()} \cr
 \code{\link{op_diag}()} \cr
@@ -178,10 +186,13 @@ Other ops: \cr
 \code{\link{op_glu}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
 \code{\link{op_hard_shrink}()} \cr
 \code{\link{op_hard_sigmoid}()} \cr
 \code{\link{op_hard_silu}()} \cr
 \code{\link{op_hard_tanh}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -208,6 +219,8 @@ Other ops: \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
 \code{\link{op_istft}()} \cr
+\code{\link{op_kaiser}()} \cr
+\code{\link{op_layer_normalization}()} \cr
 \code{\link{op_leaky_relu}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
@@ -306,6 +319,7 @@ Other ops: \cr
 \code{\link{op_sort}()} \cr
 \code{\link{op_sparse_categorical_crossentropy}()} \cr
 \code{\link{op_sparse_plus}()} \cr
+\code{\link{op_sparse_sigmoid}()} \cr
 \code{\link{op_sparsemax}()} \cr
 \code{\link{op_split}()} \cr
 \code{\link{op_sqrt}()} \cr
@@ -343,6 +357,8 @@ Other ops: \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
 \code{\link{op_vectorized_map}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_while_loop}()} \cr
diff --git a/man/op_qr.Rd b/man/op_qr.Rd
index 0c1a4ae36..777e84495 100644
--- a/man/op_qr.Rd
+++ b/man/op_qr.Rd
@@ -73,6 +73,7 @@ Other ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -91,6 +92,7 @@ Other ops: \cr
 \code{\link{op_associative_scan}()} \cr
 \code{\link{op_average}()} \cr
 \code{\link{op_average_pool}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_batch_normalization}()} \cr
 \code{\link{op_binary_crossentropy}()} \cr
 \code{\link{op_bincount}()} \cr
@@ -101,9 +103,11 @@ Other ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
 \code{\link{op_cast}()} \cr
 \code{\link{op_categorical_crossentropy}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_celu}()} \cr
 \code{\link{op_cholesky}()} \cr
@@ -116,6 +120,7 @@ Other ops: \cr
 \code{\link{op_convert_to_numpy}()} \cr
 \code{\link{op_convert_to_tensor}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -126,6 +131,7 @@ Other ops: \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
 \code{\link{op_custom_gradient}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_depthwise_conv}()} \cr
 \code{\link{op_det}()} \cr
 \code{\link{op_diag}()} \cr
@@ -165,10 +171,13 @@ Other ops: \cr
 \code{\link{op_glu}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
 \code{\link{op_hard_shrink}()} \cr
 \code{\link{op_hard_sigmoid}()} \cr
 \code{\link{op_hard_silu}()} \cr
 \code{\link{op_hard_tanh}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -195,6 +204,8 @@ Other ops: \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
 \code{\link{op_istft}()} \cr
+\code{\link{op_kaiser}()} \cr
+\code{\link{op_layer_normalization}()} \cr
 \code{\link{op_leaky_relu}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
@@ -293,6 +304,7 @@ Other ops: \cr
 \code{\link{op_sort}()} \cr
 \code{\link{op_sparse_categorical_crossentropy}()} \cr
 \code{\link{op_sparse_plus}()} \cr
+\code{\link{op_sparse_sigmoid}()} \cr
 \code{\link{op_sparsemax}()} \cr
 \code{\link{op_split}()} \cr
 \code{\link{op_sqrt}()} \cr
@@ -330,6 +342,8 @@ Other ops: \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
 \code{\link{op_vectorized_map}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_while_loop}()} \cr
diff --git a/man/op_quantile.Rd b/man/op_quantile.Rd
index 58b5a56f9..e521bed0e 100644
--- a/man/op_quantile.Rd
+++ b/man/op_quantile.Rd
@@ -46,6 +46,7 @@ Other numpy ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -62,6 +63,7 @@ Other numpy ops: \cr
 \code{\link{op_argsort}()} \cr
 \code{\link{op_array}()} \cr
 \code{\link{op_average}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_bincount}()} \cr
 \code{\link{op_bitwise_and}()} \cr
 \code{\link{op_bitwise_invert}()} \cr
@@ -70,12 +72,15 @@ Other numpy ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_clip}()} \cr
 \code{\link{op_concatenate}()} \cr
 \code{\link{op_conj}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -84,6 +89,7 @@ Other numpy ops: \cr
 \code{\link{op_ctc_decode}()} \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_diag}()} \cr
 \code{\link{op_diagflat}()} \cr
 \code{\link{op_diagonal}()} \cr
@@ -108,6 +114,9 @@ Other numpy ops: \cr
 \code{\link{op_get_item}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -117,6 +126,7 @@ Other numpy ops: \cr
 \code{\link{op_isfinite}()} \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
+\code{\link{op_kaiser}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
 \code{\link{op_less_equal}()} \cr
@@ -196,6 +206,8 @@ Other numpy ops: \cr
 \code{\link{op_var}()} \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_zeros}()} \cr
@@ -205,6 +217,7 @@ Other ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -223,6 +236,7 @@ Other ops: \cr
 \code{\link{op_associative_scan}()} \cr
 \code{\link{op_average}()} \cr
 \code{\link{op_average_pool}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_batch_normalization}()} \cr
 \code{\link{op_binary_crossentropy}()} \cr
 \code{\link{op_bincount}()} \cr
@@ -233,9 +247,11 @@ Other ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
 \code{\link{op_cast}()} \cr
 \code{\link{op_categorical_crossentropy}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_celu}()} \cr
 \code{\link{op_cholesky}()} \cr
@@ -248,6 +264,7 @@ Other ops: \cr
 \code{\link{op_convert_to_numpy}()} \cr
 \code{\link{op_convert_to_tensor}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -258,6 +275,7 @@ Other ops: \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
 \code{\link{op_custom_gradient}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_depthwise_conv}()} \cr
 \code{\link{op_det}()} \cr
 \code{\link{op_diag}()} \cr
@@ -297,10 +315,13 @@ Other ops: \cr
 \code{\link{op_glu}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
 \code{\link{op_hard_shrink}()} \cr
 \code{\link{op_hard_sigmoid}()} \cr
 \code{\link{op_hard_silu}()} \cr
 \code{\link{op_hard_tanh}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -327,6 +348,8 @@ Other ops: \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
 \code{\link{op_istft}()} \cr
+\code{\link{op_kaiser}()} \cr
+\code{\link{op_layer_normalization}()} \cr
 \code{\link{op_leaky_relu}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
@@ -425,6 +448,7 @@ Other ops: \cr
 \code{\link{op_sort}()} \cr
 \code{\link{op_sparse_categorical_crossentropy}()} \cr
 \code{\link{op_sparse_plus}()} \cr
+\code{\link{op_sparse_sigmoid}()} \cr
 \code{\link{op_sparsemax}()} \cr
 \code{\link{op_split}()} \cr
 \code{\link{op_sqrt}()} \cr
@@ -462,6 +486,8 @@ Other ops: \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
 \code{\link{op_vectorized_map}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_while_loop}()} \cr
diff --git a/man/op_ravel.Rd b/man/op_ravel.Rd
index 88600c8f5..0a5f2c1b5 100644
--- a/man/op_ravel.Rd
+++ b/man/op_ravel.Rd
@@ -24,6 +24,7 @@ Other numpy ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -40,6 +41,7 @@ Other numpy ops: \cr
 \code{\link{op_argsort}()} \cr
 \code{\link{op_array}()} \cr
 \code{\link{op_average}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_bincount}()} \cr
 \code{\link{op_bitwise_and}()} \cr
 \code{\link{op_bitwise_invert}()} \cr
@@ -48,12 +50,15 @@ Other numpy ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_clip}()} \cr
 \code{\link{op_concatenate}()} \cr
 \code{\link{op_conj}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -62,6 +67,7 @@ Other numpy ops: \cr
 \code{\link{op_ctc_decode}()} \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_diag}()} \cr
 \code{\link{op_diagflat}()} \cr
 \code{\link{op_diagonal}()} \cr
@@ -86,6 +92,9 @@ Other numpy ops: \cr
 \code{\link{op_get_item}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -95,6 +104,7 @@ Other numpy ops: \cr
 \code{\link{op_isfinite}()} \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
+\code{\link{op_kaiser}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
 \code{\link{op_less_equal}()} \cr
@@ -174,6 +184,8 @@ Other numpy ops: \cr
 \code{\link{op_var}()} \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_zeros}()} \cr
@@ -183,6 +195,7 @@ Other ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -201,6 +214,7 @@ Other ops: \cr
 \code{\link{op_associative_scan}()} \cr
 \code{\link{op_average}()} \cr
 \code{\link{op_average_pool}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_batch_normalization}()} \cr
 \code{\link{op_binary_crossentropy}()} \cr
 \code{\link{op_bincount}()} \cr
@@ -211,9 +225,11 @@ Other ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
 \code{\link{op_cast}()} \cr
 \code{\link{op_categorical_crossentropy}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_celu}()} \cr
 \code{\link{op_cholesky}()} \cr
@@ -226,6 +242,7 @@ Other ops: \cr
 \code{\link{op_convert_to_numpy}()} \cr
 \code{\link{op_convert_to_tensor}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -236,6 +253,7 @@ Other ops: \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
 \code{\link{op_custom_gradient}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_depthwise_conv}()} \cr
 \code{\link{op_det}()} \cr
 \code{\link{op_diag}()} \cr
@@ -275,10 +293,13 @@ Other ops: \cr
 \code{\link{op_glu}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
 \code{\link{op_hard_shrink}()} \cr
 \code{\link{op_hard_sigmoid}()} \cr
 \code{\link{op_hard_silu}()} \cr
 \code{\link{op_hard_tanh}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -305,6 +326,8 @@ Other ops: \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
 \code{\link{op_istft}()} \cr
+\code{\link{op_kaiser}()} \cr
+\code{\link{op_layer_normalization}()} \cr
 \code{\link{op_leaky_relu}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
@@ -403,6 +426,7 @@ Other ops: \cr
 \code{\link{op_sort}()} \cr
 \code{\link{op_sparse_categorical_crossentropy}()} \cr
 \code{\link{op_sparse_plus}()} \cr
+\code{\link{op_sparse_sigmoid}()} \cr
 \code{\link{op_sparsemax}()} \cr
 \code{\link{op_split}()} \cr
 \code{\link{op_sqrt}()} \cr
@@ -440,6 +464,8 @@ Other ops: \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
 \code{\link{op_vectorized_map}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_while_loop}()} \cr
diff --git a/man/op_real.Rd b/man/op_real.Rd
index 4b9de1161..41ba00598 100644
--- a/man/op_real.Rd
+++ b/man/op_real.Rd
@@ -24,6 +24,7 @@ Other numpy ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -40,6 +41,7 @@ Other numpy ops: \cr
 \code{\link{op_argsort}()} \cr
 \code{\link{op_array}()} \cr
 \code{\link{op_average}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_bincount}()} \cr
 \code{\link{op_bitwise_and}()} \cr
 \code{\link{op_bitwise_invert}()} \cr
@@ -48,12 +50,15 @@ Other numpy ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_clip}()} \cr
 \code{\link{op_concatenate}()} \cr
 \code{\link{op_conj}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -62,6 +67,7 @@ Other numpy ops: \cr
 \code{\link{op_ctc_decode}()} \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_diag}()} \cr
 \code{\link{op_diagflat}()} \cr
 \code{\link{op_diagonal}()} \cr
@@ -86,6 +92,9 @@ Other numpy ops: \cr
 \code{\link{op_get_item}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -95,6 +104,7 @@ Other numpy ops: \cr
 \code{\link{op_isfinite}()} \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
+\code{\link{op_kaiser}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
 \code{\link{op_less_equal}()} \cr
@@ -174,6 +184,8 @@ Other numpy ops: \cr
 \code{\link{op_var}()} \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_zeros}()} \cr
@@ -183,6 +195,7 @@ Other ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -201,6 +214,7 @@ Other ops: \cr
 \code{\link{op_associative_scan}()} \cr
 \code{\link{op_average}()} \cr
 \code{\link{op_average_pool}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_batch_normalization}()} \cr
 \code{\link{op_binary_crossentropy}()} \cr
 \code{\link{op_bincount}()} \cr
@@ -211,9 +225,11 @@ Other ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
 \code{\link{op_cast}()} \cr
 \code{\link{op_categorical_crossentropy}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_celu}()} \cr
 \code{\link{op_cholesky}()} \cr
@@ -226,6 +242,7 @@ Other ops: \cr
 \code{\link{op_convert_to_numpy}()} \cr
 \code{\link{op_convert_to_tensor}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -236,6 +253,7 @@ Other ops: \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
 \code{\link{op_custom_gradient}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_depthwise_conv}()} \cr
 \code{\link{op_det}()} \cr
 \code{\link{op_diag}()} \cr
@@ -275,10 +293,13 @@ Other ops: \cr
 \code{\link{op_glu}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
 \code{\link{op_hard_shrink}()} \cr
 \code{\link{op_hard_sigmoid}()} \cr
 \code{\link{op_hard_silu}()} \cr
 \code{\link{op_hard_tanh}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -305,6 +326,8 @@ Other ops: \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
 \code{\link{op_istft}()} \cr
+\code{\link{op_kaiser}()} \cr
+\code{\link{op_layer_normalization}()} \cr
 \code{\link{op_leaky_relu}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
@@ -403,6 +426,7 @@ Other ops: \cr
 \code{\link{op_sort}()} \cr
 \code{\link{op_sparse_categorical_crossentropy}()} \cr
 \code{\link{op_sparse_plus}()} \cr
+\code{\link{op_sparse_sigmoid}()} \cr
 \code{\link{op_sparsemax}()} \cr
 \code{\link{op_split}()} \cr
 \code{\link{op_sqrt}()} \cr
@@ -440,6 +464,8 @@ Other ops: \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
 \code{\link{op_vectorized_map}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_while_loop}()} \cr
diff --git a/man/op_rearrange.Rd b/man/op_rearrange.Rd
index 7821b3763..f629e613d 100644
--- a/man/op_rearrange.Rd
+++ b/man/op_rearrange.Rd
@@ -109,6 +109,7 @@ Other ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -127,6 +128,7 @@ Other ops: \cr
 \code{\link{op_associative_scan}()} \cr
 \code{\link{op_average}()} \cr
 \code{\link{op_average_pool}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_batch_normalization}()} \cr
 \code{\link{op_binary_crossentropy}()} \cr
 \code{\link{op_bincount}()} \cr
@@ -137,9 +139,11 @@ Other ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
 \code{\link{op_cast}()} \cr
 \code{\link{op_categorical_crossentropy}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_celu}()} \cr
 \code{\link{op_cholesky}()} \cr
@@ -152,6 +156,7 @@ Other ops: \cr
 \code{\link{op_convert_to_numpy}()} \cr
 \code{\link{op_convert_to_tensor}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -162,6 +167,7 @@ Other ops: \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
 \code{\link{op_custom_gradient}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_depthwise_conv}()} \cr
 \code{\link{op_det}()} \cr
 \code{\link{op_diag}()} \cr
@@ -201,10 +207,13 @@ Other ops: \cr
 \code{\link{op_glu}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
 \code{\link{op_hard_shrink}()} \cr
 \code{\link{op_hard_sigmoid}()} \cr
 \code{\link{op_hard_silu}()} \cr
 \code{\link{op_hard_tanh}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -231,6 +240,8 @@ Other ops: \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
 \code{\link{op_istft}()} \cr
+\code{\link{op_kaiser}()} \cr
+\code{\link{op_layer_normalization}()} \cr
 \code{\link{op_leaky_relu}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
@@ -329,6 +340,7 @@ Other ops: \cr
 \code{\link{op_sort}()} \cr
 \code{\link{op_sparse_categorical_crossentropy}()} \cr
 \code{\link{op_sparse_plus}()} \cr
+\code{\link{op_sparse_sigmoid}()} \cr
 \code{\link{op_sparsemax}()} \cr
 \code{\link{op_split}()} \cr
 \code{\link{op_sqrt}()} \cr
@@ -366,6 +378,8 @@ Other ops: \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
 \code{\link{op_vectorized_map}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_while_loop}()} \cr
diff --git a/man/op_reciprocal.Rd b/man/op_reciprocal.Rd
index e8fdbd3e3..7522e8964 100644
--- a/man/op_reciprocal.Rd
+++ b/man/op_reciprocal.Rd
@@ -24,6 +24,7 @@ Other numpy ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -40,6 +41,7 @@ Other numpy ops: \cr
 \code{\link{op_argsort}()} \cr
 \code{\link{op_array}()} \cr
 \code{\link{op_average}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_bincount}()} \cr
 \code{\link{op_bitwise_and}()} \cr
 \code{\link{op_bitwise_invert}()} \cr
@@ -48,12 +50,15 @@ Other numpy ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_clip}()} \cr
 \code{\link{op_concatenate}()} \cr
 \code{\link{op_conj}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -62,6 +67,7 @@ Other numpy ops: \cr
 \code{\link{op_ctc_decode}()} \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_diag}()} \cr
 \code{\link{op_diagflat}()} \cr
 \code{\link{op_diagonal}()} \cr
@@ -86,6 +92,9 @@ Other numpy ops: \cr
 \code{\link{op_get_item}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -95,6 +104,7 @@ Other numpy ops: \cr
 \code{\link{op_isfinite}()} \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
+\code{\link{op_kaiser}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
 \code{\link{op_less_equal}()} \cr
@@ -174,6 +184,8 @@ Other numpy ops: \cr
 \code{\link{op_var}()} \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_zeros}()} \cr
@@ -183,6 +195,7 @@ Other ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -201,6 +214,7 @@ Other ops: \cr
 \code{\link{op_associative_scan}()} \cr
 \code{\link{op_average}()} \cr
 \code{\link{op_average_pool}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_batch_normalization}()} \cr
 \code{\link{op_binary_crossentropy}()} \cr
 \code{\link{op_bincount}()} \cr
@@ -211,9 +225,11 @@ Other ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
 \code{\link{op_cast}()} \cr
 \code{\link{op_categorical_crossentropy}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_celu}()} \cr
 \code{\link{op_cholesky}()} \cr
@@ -226,6 +242,7 @@ Other ops: \cr
 \code{\link{op_convert_to_numpy}()} \cr
 \code{\link{op_convert_to_tensor}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -236,6 +253,7 @@ Other ops: \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
 \code{\link{op_custom_gradient}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_depthwise_conv}()} \cr
 \code{\link{op_det}()} \cr
 \code{\link{op_diag}()} \cr
@@ -275,10 +293,13 @@ Other ops: \cr
 \code{\link{op_glu}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
 \code{\link{op_hard_shrink}()} \cr
 \code{\link{op_hard_sigmoid}()} \cr
 \code{\link{op_hard_silu}()} \cr
 \code{\link{op_hard_tanh}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -305,6 +326,8 @@ Other ops: \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
 \code{\link{op_istft}()} \cr
+\code{\link{op_kaiser}()} \cr
+\code{\link{op_layer_normalization}()} \cr
 \code{\link{op_leaky_relu}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
@@ -403,6 +426,7 @@ Other ops: \cr
 \code{\link{op_sort}()} \cr
 \code{\link{op_sparse_categorical_crossentropy}()} \cr
 \code{\link{op_sparse_plus}()} \cr
+\code{\link{op_sparse_sigmoid}()} \cr
 \code{\link{op_sparsemax}()} \cr
 \code{\link{op_split}()} \cr
 \code{\link{op_sqrt}()} \cr
@@ -440,6 +464,8 @@ Other ops: \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
 \code{\link{op_vectorized_map}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_while_loop}()} \cr
diff --git a/man/op_relu.Rd b/man/op_relu.Rd
index 4fd4c6f54..cec65b974 100644
--- a/man/op_relu.Rd
+++ b/man/op_relu.Rd
@@ -54,6 +54,7 @@ Other nn ops: \cr
 \code{\link{op_hard_sigmoid}()} \cr
 \code{\link{op_hard_silu}()} \cr
 \code{\link{op_hard_tanh}()} \cr
+\code{\link{op_layer_normalization}()} \cr
 \code{\link{op_leaky_relu}()} \cr
 \code{\link{op_log_sigmoid}()} \cr
 \code{\link{op_log_softmax}()} \cr
@@ -76,6 +77,7 @@ Other nn ops: \cr
 \code{\link{op_softsign}()} \cr
 \code{\link{op_sparse_categorical_crossentropy}()} \cr
 \code{\link{op_sparse_plus}()} \cr
+\code{\link{op_sparse_sigmoid}()} \cr
 \code{\link{op_sparsemax}()} \cr
 \code{\link{op_squareplus}()} \cr
 \code{\link{op_tanh_shrink}()} \cr
@@ -86,6 +88,7 @@ Other ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -104,6 +107,7 @@ Other ops: \cr
 \code{\link{op_associative_scan}()} \cr
 \code{\link{op_average}()} \cr
 \code{\link{op_average_pool}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_batch_normalization}()} \cr
 \code{\link{op_binary_crossentropy}()} \cr
 \code{\link{op_bincount}()} \cr
@@ -114,9 +118,11 @@ Other ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
 \code{\link{op_cast}()} \cr
 \code{\link{op_categorical_crossentropy}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_celu}()} \cr
 \code{\link{op_cholesky}()} \cr
@@ -129,6 +135,7 @@ Other ops: \cr
 \code{\link{op_convert_to_numpy}()} \cr
 \code{\link{op_convert_to_tensor}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -139,6 +146,7 @@ Other ops: \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
 \code{\link{op_custom_gradient}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_depthwise_conv}()} \cr
 \code{\link{op_det}()} \cr
 \code{\link{op_diag}()} \cr
@@ -178,10 +186,13 @@ Other ops: \cr
 \code{\link{op_glu}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
 \code{\link{op_hard_shrink}()} \cr
 \code{\link{op_hard_sigmoid}()} \cr
 \code{\link{op_hard_silu}()} \cr
 \code{\link{op_hard_tanh}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -208,6 +219,8 @@ Other ops: \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
 \code{\link{op_istft}()} \cr
+\code{\link{op_kaiser}()} \cr
+\code{\link{op_layer_normalization}()} \cr
 \code{\link{op_leaky_relu}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
@@ -306,6 +319,7 @@ Other ops: \cr
 \code{\link{op_sort}()} \cr
 \code{\link{op_sparse_categorical_crossentropy}()} \cr
 \code{\link{op_sparse_plus}()} \cr
+\code{\link{op_sparse_sigmoid}()} \cr
 \code{\link{op_sparsemax}()} \cr
 \code{\link{op_split}()} \cr
 \code{\link{op_sqrt}()} \cr
@@ -343,6 +357,8 @@ Other ops: \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
 \code{\link{op_vectorized_map}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_while_loop}()} \cr
diff --git a/man/op_relu6.Rd b/man/op_relu6.Rd
index 1e1a764d7..f8ad50f30 100644
--- a/man/op_relu6.Rd
+++ b/man/op_relu6.Rd
@@ -54,6 +54,7 @@ Other nn ops: \cr
 \code{\link{op_hard_sigmoid}()} \cr
 \code{\link{op_hard_silu}()} \cr
 \code{\link{op_hard_tanh}()} \cr
+\code{\link{op_layer_normalization}()} \cr
 \code{\link{op_leaky_relu}()} \cr
 \code{\link{op_log_sigmoid}()} \cr
 \code{\link{op_log_softmax}()} \cr
@@ -76,6 +77,7 @@ Other nn ops: \cr
 \code{\link{op_softsign}()} \cr
 \code{\link{op_sparse_categorical_crossentropy}()} \cr
 \code{\link{op_sparse_plus}()} \cr
+\code{\link{op_sparse_sigmoid}()} \cr
 \code{\link{op_sparsemax}()} \cr
 \code{\link{op_squareplus}()} \cr
 \code{\link{op_tanh_shrink}()} \cr
@@ -86,6 +88,7 @@ Other ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -104,6 +107,7 @@ Other ops: \cr
 \code{\link{op_associative_scan}()} \cr
 \code{\link{op_average}()} \cr
 \code{\link{op_average_pool}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_batch_normalization}()} \cr
 \code{\link{op_binary_crossentropy}()} \cr
 \code{\link{op_bincount}()} \cr
@@ -114,9 +118,11 @@ Other ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
 \code{\link{op_cast}()} \cr
 \code{\link{op_categorical_crossentropy}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_celu}()} \cr
 \code{\link{op_cholesky}()} \cr
@@ -129,6 +135,7 @@ Other ops: \cr
 \code{\link{op_convert_to_numpy}()} \cr
 \code{\link{op_convert_to_tensor}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -139,6 +146,7 @@ Other ops: \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
 \code{\link{op_custom_gradient}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_depthwise_conv}()} \cr
 \code{\link{op_det}()} \cr
 \code{\link{op_diag}()} \cr
@@ -178,10 +186,13 @@ Other ops: \cr
 \code{\link{op_glu}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
 \code{\link{op_hard_shrink}()} \cr
 \code{\link{op_hard_sigmoid}()} \cr
 \code{\link{op_hard_silu}()} \cr
 \code{\link{op_hard_tanh}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -208,6 +219,8 @@ Other ops: \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
 \code{\link{op_istft}()} \cr
+\code{\link{op_kaiser}()} \cr
+\code{\link{op_layer_normalization}()} \cr
 \code{\link{op_leaky_relu}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
@@ -306,6 +319,7 @@ Other ops: \cr
 \code{\link{op_sort}()} \cr
 \code{\link{op_sparse_categorical_crossentropy}()} \cr
 \code{\link{op_sparse_plus}()} \cr
+\code{\link{op_sparse_sigmoid}()} \cr
 \code{\link{op_sparsemax}()} \cr
 \code{\link{op_split}()} \cr
 \code{\link{op_sqrt}()} \cr
@@ -343,6 +357,8 @@ Other ops: \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
 \code{\link{op_vectorized_map}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_while_loop}()} \cr
diff --git a/man/op_repeat.Rd b/man/op_repeat.Rd
index dc91d01a0..3712918b1 100644
--- a/man/op_repeat.Rd
+++ b/man/op_repeat.Rd
@@ -29,6 +29,7 @@ Other numpy ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -45,6 +46,7 @@ Other numpy ops: \cr
 \code{\link{op_argsort}()} \cr
 \code{\link{op_array}()} \cr
 \code{\link{op_average}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_bincount}()} \cr
 \code{\link{op_bitwise_and}()} \cr
 \code{\link{op_bitwise_invert}()} \cr
@@ -53,12 +55,15 @@ Other numpy ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_clip}()} \cr
 \code{\link{op_concatenate}()} \cr
 \code{\link{op_conj}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -67,6 +72,7 @@ Other numpy ops: \cr
 \code{\link{op_ctc_decode}()} \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_diag}()} \cr
 \code{\link{op_diagflat}()} \cr
 \code{\link{op_diagonal}()} \cr
@@ -91,6 +97,9 @@ Other numpy ops: \cr
 \code{\link{op_get_item}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -100,6 +109,7 @@ Other numpy ops: \cr
 \code{\link{op_isfinite}()} \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
+\code{\link{op_kaiser}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
 \code{\link{op_less_equal}()} \cr
@@ -179,6 +189,8 @@ Other numpy ops: \cr
 \code{\link{op_var}()} \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_zeros}()} \cr
@@ -188,6 +200,7 @@ Other ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -206,6 +219,7 @@ Other ops: \cr
 \code{\link{op_associative_scan}()} \cr
 \code{\link{op_average}()} \cr
 \code{\link{op_average_pool}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_batch_normalization}()} \cr
 \code{\link{op_binary_crossentropy}()} \cr
 \code{\link{op_bincount}()} \cr
@@ -216,9 +230,11 @@ Other ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
 \code{\link{op_cast}()} \cr
 \code{\link{op_categorical_crossentropy}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_celu}()} \cr
 \code{\link{op_cholesky}()} \cr
@@ -231,6 +247,7 @@ Other ops: \cr
 \code{\link{op_convert_to_numpy}()} \cr
 \code{\link{op_convert_to_tensor}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -241,6 +258,7 @@ Other ops: \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
 \code{\link{op_custom_gradient}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_depthwise_conv}()} \cr
 \code{\link{op_det}()} \cr
 \code{\link{op_diag}()} \cr
@@ -280,10 +298,13 @@ Other ops: \cr
 \code{\link{op_glu}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
 \code{\link{op_hard_shrink}()} \cr
 \code{\link{op_hard_sigmoid}()} \cr
 \code{\link{op_hard_silu}()} \cr
 \code{\link{op_hard_tanh}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -310,6 +331,8 @@ Other ops: \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
 \code{\link{op_istft}()} \cr
+\code{\link{op_kaiser}()} \cr
+\code{\link{op_layer_normalization}()} \cr
 \code{\link{op_leaky_relu}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
@@ -408,6 +431,7 @@ Other ops: \cr
 \code{\link{op_sort}()} \cr
 \code{\link{op_sparse_categorical_crossentropy}()} \cr
 \code{\link{op_sparse_plus}()} \cr
+\code{\link{op_sparse_sigmoid}()} \cr
 \code{\link{op_sparsemax}()} \cr
 \code{\link{op_split}()} \cr
 \code{\link{op_sqrt}()} \cr
@@ -445,6 +469,8 @@ Other ops: \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
 \code{\link{op_vectorized_map}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_while_loop}()} \cr
diff --git a/man/op_reshape.Rd b/man/op_reshape.Rd
index 023b461af..b38d7d62f 100644
--- a/man/op_reshape.Rd
+++ b/man/op_reshape.Rd
@@ -28,6 +28,7 @@ Other numpy ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -44,6 +45,7 @@ Other numpy ops: \cr
 \code{\link{op_argsort}()} \cr
 \code{\link{op_array}()} \cr
 \code{\link{op_average}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_bincount}()} \cr
 \code{\link{op_bitwise_and}()} \cr
 \code{\link{op_bitwise_invert}()} \cr
@@ -52,12 +54,15 @@ Other numpy ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_clip}()} \cr
 \code{\link{op_concatenate}()} \cr
 \code{\link{op_conj}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -66,6 +71,7 @@ Other numpy ops: \cr
 \code{\link{op_ctc_decode}()} \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_diag}()} \cr
 \code{\link{op_diagflat}()} \cr
 \code{\link{op_diagonal}()} \cr
@@ -90,6 +96,9 @@ Other numpy ops: \cr
 \code{\link{op_get_item}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -99,6 +108,7 @@ Other numpy ops: \cr
 \code{\link{op_isfinite}()} \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
+\code{\link{op_kaiser}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
 \code{\link{op_less_equal}()} \cr
@@ -178,6 +188,8 @@ Other numpy ops: \cr
 \code{\link{op_var}()} \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_zeros}()} \cr
@@ -187,6 +199,7 @@ Other ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -205,6 +218,7 @@ Other ops: \cr
 \code{\link{op_associative_scan}()} \cr
 \code{\link{op_average}()} \cr
 \code{\link{op_average_pool}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_batch_normalization}()} \cr
 \code{\link{op_binary_crossentropy}()} \cr
 \code{\link{op_bincount}()} \cr
@@ -215,9 +229,11 @@ Other ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
 \code{\link{op_cast}()} \cr
 \code{\link{op_categorical_crossentropy}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_celu}()} \cr
 \code{\link{op_cholesky}()} \cr
@@ -230,6 +246,7 @@ Other ops: \cr
 \code{\link{op_convert_to_numpy}()} \cr
 \code{\link{op_convert_to_tensor}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -240,6 +257,7 @@ Other ops: \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
 \code{\link{op_custom_gradient}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_depthwise_conv}()} \cr
 \code{\link{op_det}()} \cr
 \code{\link{op_diag}()} \cr
@@ -279,10 +297,13 @@ Other ops: \cr
 \code{\link{op_glu}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
 \code{\link{op_hard_shrink}()} \cr
 \code{\link{op_hard_sigmoid}()} \cr
 \code{\link{op_hard_silu}()} \cr
 \code{\link{op_hard_tanh}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -309,6 +330,8 @@ Other ops: \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
 \code{\link{op_istft}()} \cr
+\code{\link{op_kaiser}()} \cr
+\code{\link{op_layer_normalization}()} \cr
 \code{\link{op_leaky_relu}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
@@ -407,6 +430,7 @@ Other ops: \cr
 \code{\link{op_sort}()} \cr
 \code{\link{op_sparse_categorical_crossentropy}()} \cr
 \code{\link{op_sparse_plus}()} \cr
+\code{\link{op_sparse_sigmoid}()} \cr
 \code{\link{op_sparsemax}()} \cr
 \code{\link{op_split}()} \cr
 \code{\link{op_sqrt}()} \cr
@@ -444,6 +468,8 @@ Other ops: \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
 \code{\link{op_vectorized_map}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_while_loop}()} \cr
diff --git a/man/op_rfft.Rd b/man/op_rfft.Rd
index 3f8db1384..219884397 100644
--- a/man/op_rfft.Rd
+++ b/man/op_rfft.Rd
@@ -83,6 +83,7 @@ Other ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -101,6 +102,7 @@ Other ops: \cr
 \code{\link{op_associative_scan}()} \cr
 \code{\link{op_average}()} \cr
 \code{\link{op_average_pool}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_batch_normalization}()} \cr
 \code{\link{op_binary_crossentropy}()} \cr
 \code{\link{op_bincount}()} \cr
@@ -111,9 +113,11 @@ Other ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
 \code{\link{op_cast}()} \cr
 \code{\link{op_categorical_crossentropy}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_celu}()} \cr
 \code{\link{op_cholesky}()} \cr
@@ -126,6 +130,7 @@ Other ops: \cr
 \code{\link{op_convert_to_numpy}()} \cr
 \code{\link{op_convert_to_tensor}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -136,6 +141,7 @@ Other ops: \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
 \code{\link{op_custom_gradient}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_depthwise_conv}()} \cr
 \code{\link{op_det}()} \cr
 \code{\link{op_diag}()} \cr
@@ -175,10 +181,13 @@ Other ops: \cr
 \code{\link{op_glu}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
 \code{\link{op_hard_shrink}()} \cr
 \code{\link{op_hard_sigmoid}()} \cr
 \code{\link{op_hard_silu}()} \cr
 \code{\link{op_hard_tanh}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -205,6 +214,8 @@ Other ops: \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
 \code{\link{op_istft}()} \cr
+\code{\link{op_kaiser}()} \cr
+\code{\link{op_layer_normalization}()} \cr
 \code{\link{op_leaky_relu}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
@@ -303,6 +314,7 @@ Other ops: \cr
 \code{\link{op_sort}()} \cr
 \code{\link{op_sparse_categorical_crossentropy}()} \cr
 \code{\link{op_sparse_plus}()} \cr
+\code{\link{op_sparse_sigmoid}()} \cr
 \code{\link{op_sparsemax}()} \cr
 \code{\link{op_split}()} \cr
 \code{\link{op_sqrt}()} \cr
@@ -340,6 +352,8 @@ Other ops: \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
 \code{\link{op_vectorized_map}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_while_loop}()} \cr
diff --git a/man/op_right_shift.Rd b/man/op_right_shift.Rd
index f88050a4a..13edaa5c6 100644
--- a/man/op_right_shift.Rd
+++ b/man/op_right_shift.Rd
@@ -24,6 +24,7 @@ Other numpy ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -40,6 +41,7 @@ Other numpy ops: \cr
 \code{\link{op_argsort}()} \cr
 \code{\link{op_array}()} \cr
 \code{\link{op_average}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_bincount}()} \cr
 \code{\link{op_bitwise_and}()} \cr
 \code{\link{op_bitwise_invert}()} \cr
@@ -48,12 +50,15 @@ Other numpy ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_clip}()} \cr
 \code{\link{op_concatenate}()} \cr
 \code{\link{op_conj}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -62,6 +67,7 @@ Other numpy ops: \cr
 \code{\link{op_ctc_decode}()} \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_diag}()} \cr
 \code{\link{op_diagflat}()} \cr
 \code{\link{op_diagonal}()} \cr
@@ -86,6 +92,9 @@ Other numpy ops: \cr
 \code{\link{op_get_item}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -95,6 +104,7 @@ Other numpy ops: \cr
 \code{\link{op_isfinite}()} \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
+\code{\link{op_kaiser}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
 \code{\link{op_less_equal}()} \cr
@@ -174,6 +184,8 @@ Other numpy ops: \cr
 \code{\link{op_var}()} \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_zeros}()} \cr
@@ -183,6 +195,7 @@ Other ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -201,6 +214,7 @@ Other ops: \cr
 \code{\link{op_associative_scan}()} \cr
 \code{\link{op_average}()} \cr
 \code{\link{op_average_pool}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_batch_normalization}()} \cr
 \code{\link{op_binary_crossentropy}()} \cr
 \code{\link{op_bincount}()} \cr
@@ -211,9 +225,11 @@ Other ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
 \code{\link{op_cast}()} \cr
 \code{\link{op_categorical_crossentropy}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_celu}()} \cr
 \code{\link{op_cholesky}()} \cr
@@ -226,6 +242,7 @@ Other ops: \cr
 \code{\link{op_convert_to_numpy}()} \cr
 \code{\link{op_convert_to_tensor}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -236,6 +253,7 @@ Other ops: \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
 \code{\link{op_custom_gradient}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_depthwise_conv}()} \cr
 \code{\link{op_det}()} \cr
 \code{\link{op_diag}()} \cr
@@ -275,10 +293,13 @@ Other ops: \cr
 \code{\link{op_glu}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
 \code{\link{op_hard_shrink}()} \cr
 \code{\link{op_hard_sigmoid}()} \cr
 \code{\link{op_hard_silu}()} \cr
 \code{\link{op_hard_tanh}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -305,6 +326,8 @@ Other ops: \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
 \code{\link{op_istft}()} \cr
+\code{\link{op_kaiser}()} \cr
+\code{\link{op_layer_normalization}()} \cr
 \code{\link{op_leaky_relu}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
@@ -403,6 +426,7 @@ Other ops: \cr
 \code{\link{op_sort}()} \cr
 \code{\link{op_sparse_categorical_crossentropy}()} \cr
 \code{\link{op_sparse_plus}()} \cr
+\code{\link{op_sparse_sigmoid}()} \cr
 \code{\link{op_sparsemax}()} \cr
 \code{\link{op_split}()} \cr
 \code{\link{op_sqrt}()} \cr
@@ -440,6 +464,8 @@ Other ops: \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
 \code{\link{op_vectorized_map}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_while_loop}()} \cr
diff --git a/man/op_rms_normalization.Rd b/man/op_rms_normalization.Rd
index 73e5a7b3b..f2e15a9b5 100644
--- a/man/op_rms_normalization.Rd
+++ b/man/op_rms_normalization.Rd
@@ -77,6 +77,7 @@ Other nn ops: \cr
 \code{\link{op_hard_sigmoid}()} \cr
 \code{\link{op_hard_silu}()} \cr
 \code{\link{op_hard_tanh}()} \cr
+\code{\link{op_layer_normalization}()} \cr
 \code{\link{op_leaky_relu}()} \cr
 \code{\link{op_log_sigmoid}()} \cr
 \code{\link{op_log_softmax}()} \cr
@@ -99,6 +100,7 @@ Other nn ops: \cr
 \code{\link{op_softsign}()} \cr
 \code{\link{op_sparse_categorical_crossentropy}()} \cr
 \code{\link{op_sparse_plus}()} \cr
+\code{\link{op_sparse_sigmoid}()} \cr
 \code{\link{op_sparsemax}()} \cr
 \code{\link{op_squareplus}()} \cr
 \code{\link{op_tanh_shrink}()} \cr
@@ -109,6 +111,7 @@ Other ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -127,6 +130,7 @@ Other ops: \cr
 \code{\link{op_associative_scan}()} \cr
 \code{\link{op_average}()} \cr
 \code{\link{op_average_pool}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_batch_normalization}()} \cr
 \code{\link{op_binary_crossentropy}()} \cr
 \code{\link{op_bincount}()} \cr
@@ -137,9 +141,11 @@ Other ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
 \code{\link{op_cast}()} \cr
 \code{\link{op_categorical_crossentropy}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_celu}()} \cr
 \code{\link{op_cholesky}()} \cr
@@ -152,6 +158,7 @@ Other ops: \cr
 \code{\link{op_convert_to_numpy}()} \cr
 \code{\link{op_convert_to_tensor}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -162,6 +169,7 @@ Other ops: \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
 \code{\link{op_custom_gradient}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_depthwise_conv}()} \cr
 \code{\link{op_det}()} \cr
 \code{\link{op_diag}()} \cr
@@ -201,10 +209,13 @@ Other ops: \cr
 \code{\link{op_glu}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
 \code{\link{op_hard_shrink}()} \cr
 \code{\link{op_hard_sigmoid}()} \cr
 \code{\link{op_hard_silu}()} \cr
 \code{\link{op_hard_tanh}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -231,6 +242,8 @@ Other ops: \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
 \code{\link{op_istft}()} \cr
+\code{\link{op_kaiser}()} \cr
+\code{\link{op_layer_normalization}()} \cr
 \code{\link{op_leaky_relu}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
@@ -329,6 +342,7 @@ Other ops: \cr
 \code{\link{op_sort}()} \cr
 \code{\link{op_sparse_categorical_crossentropy}()} \cr
 \code{\link{op_sparse_plus}()} \cr
+\code{\link{op_sparse_sigmoid}()} \cr
 \code{\link{op_sparsemax}()} \cr
 \code{\link{op_split}()} \cr
 \code{\link{op_sqrt}()} \cr
@@ -366,6 +380,8 @@ Other ops: \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
 \code{\link{op_vectorized_map}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_while_loop}()} \cr
diff --git a/man/op_roll.Rd b/man/op_roll.Rd
index f052f3fea..a13c7c701 100644
--- a/man/op_roll.Rd
+++ b/man/op_roll.Rd
@@ -30,6 +30,7 @@ Other numpy ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -46,6 +47,7 @@ Other numpy ops: \cr
 \code{\link{op_argsort}()} \cr
 \code{\link{op_array}()} \cr
 \code{\link{op_average}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_bincount}()} \cr
 \code{\link{op_bitwise_and}()} \cr
 \code{\link{op_bitwise_invert}()} \cr
@@ -54,12 +56,15 @@ Other numpy ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_clip}()} \cr
 \code{\link{op_concatenate}()} \cr
 \code{\link{op_conj}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -68,6 +73,7 @@ Other numpy ops: \cr
 \code{\link{op_ctc_decode}()} \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_diag}()} \cr
 \code{\link{op_diagflat}()} \cr
 \code{\link{op_diagonal}()} \cr
@@ -92,6 +98,9 @@ Other numpy ops: \cr
 \code{\link{op_get_item}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -101,6 +110,7 @@ Other numpy ops: \cr
 \code{\link{op_isfinite}()} \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
+\code{\link{op_kaiser}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
 \code{\link{op_less_equal}()} \cr
@@ -180,6 +190,8 @@ Other numpy ops: \cr
 \code{\link{op_var}()} \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_zeros}()} \cr
@@ -189,6 +201,7 @@ Other ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -207,6 +220,7 @@ Other ops: \cr
 \code{\link{op_associative_scan}()} \cr
 \code{\link{op_average}()} \cr
 \code{\link{op_average_pool}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_batch_normalization}()} \cr
 \code{\link{op_binary_crossentropy}()} \cr
 \code{\link{op_bincount}()} \cr
@@ -217,9 +231,11 @@ Other ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
 \code{\link{op_cast}()} \cr
 \code{\link{op_categorical_crossentropy}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_celu}()} \cr
 \code{\link{op_cholesky}()} \cr
@@ -232,6 +248,7 @@ Other ops: \cr
 \code{\link{op_convert_to_numpy}()} \cr
 \code{\link{op_convert_to_tensor}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -242,6 +259,7 @@ Other ops: \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
 \code{\link{op_custom_gradient}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_depthwise_conv}()} \cr
 \code{\link{op_det}()} \cr
 \code{\link{op_diag}()} \cr
@@ -281,10 +299,13 @@ Other ops: \cr
 \code{\link{op_glu}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
 \code{\link{op_hard_shrink}()} \cr
 \code{\link{op_hard_sigmoid}()} \cr
 \code{\link{op_hard_silu}()} \cr
 \code{\link{op_hard_tanh}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -311,6 +332,8 @@ Other ops: \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
 \code{\link{op_istft}()} \cr
+\code{\link{op_kaiser}()} \cr
+\code{\link{op_layer_normalization}()} \cr
 \code{\link{op_leaky_relu}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
@@ -409,6 +432,7 @@ Other ops: \cr
 \code{\link{op_sort}()} \cr
 \code{\link{op_sparse_categorical_crossentropy}()} \cr
 \code{\link{op_sparse_plus}()} \cr
+\code{\link{op_sparse_sigmoid}()} \cr
 \code{\link{op_sparsemax}()} \cr
 \code{\link{op_split}()} \cr
 \code{\link{op_sqrt}()} \cr
@@ -446,6 +470,8 @@ Other ops: \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
 \code{\link{op_vectorized_map}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_while_loop}()} \cr
diff --git a/man/op_rot90.Rd b/man/op_rot90.Rd
index 09e4ef09f..1429616a3 100644
--- a/man/op_rot90.Rd
+++ b/man/op_rot90.Rd
@@ -73,6 +73,7 @@ Other numpy ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -89,6 +90,7 @@ Other numpy ops: \cr
 \code{\link{op_argsort}()} \cr
 \code{\link{op_array}()} \cr
 \code{\link{op_average}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_bincount}()} \cr
 \code{\link{op_bitwise_and}()} \cr
 \code{\link{op_bitwise_invert}()} \cr
@@ -97,12 +99,15 @@ Other numpy ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_clip}()} \cr
 \code{\link{op_concatenate}()} \cr
 \code{\link{op_conj}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -111,6 +116,7 @@ Other numpy ops: \cr
 \code{\link{op_ctc_decode}()} \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_diag}()} \cr
 \code{\link{op_diagflat}()} \cr
 \code{\link{op_diagonal}()} \cr
@@ -135,6 +141,9 @@ Other numpy ops: \cr
 \code{\link{op_get_item}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -144,6 +153,7 @@ Other numpy ops: \cr
 \code{\link{op_isfinite}()} \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
+\code{\link{op_kaiser}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
 \code{\link{op_less_equal}()} \cr
@@ -223,6 +233,8 @@ Other numpy ops: \cr
 \code{\link{op_var}()} \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_zeros}()} \cr
@@ -232,6 +244,7 @@ Other ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -250,6 +263,7 @@ Other ops: \cr
 \code{\link{op_associative_scan}()} \cr
 \code{\link{op_average}()} \cr
 \code{\link{op_average_pool}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_batch_normalization}()} \cr
 \code{\link{op_binary_crossentropy}()} \cr
 \code{\link{op_bincount}()} \cr
@@ -260,9 +274,11 @@ Other ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
 \code{\link{op_cast}()} \cr
 \code{\link{op_categorical_crossentropy}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_celu}()} \cr
 \code{\link{op_cholesky}()} \cr
@@ -275,6 +291,7 @@ Other ops: \cr
 \code{\link{op_convert_to_numpy}()} \cr
 \code{\link{op_convert_to_tensor}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -285,6 +302,7 @@ Other ops: \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
 \code{\link{op_custom_gradient}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_depthwise_conv}()} \cr
 \code{\link{op_det}()} \cr
 \code{\link{op_diag}()} \cr
@@ -324,10 +342,13 @@ Other ops: \cr
 \code{\link{op_glu}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
 \code{\link{op_hard_shrink}()} \cr
 \code{\link{op_hard_sigmoid}()} \cr
 \code{\link{op_hard_silu}()} \cr
 \code{\link{op_hard_tanh}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -354,6 +375,8 @@ Other ops: \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
 \code{\link{op_istft}()} \cr
+\code{\link{op_kaiser}()} \cr
+\code{\link{op_layer_normalization}()} \cr
 \code{\link{op_leaky_relu}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
@@ -452,6 +475,7 @@ Other ops: \cr
 \code{\link{op_sort}()} \cr
 \code{\link{op_sparse_categorical_crossentropy}()} \cr
 \code{\link{op_sparse_plus}()} \cr
+\code{\link{op_sparse_sigmoid}()} \cr
 \code{\link{op_sparsemax}()} \cr
 \code{\link{op_split}()} \cr
 \code{\link{op_sqrt}()} \cr
@@ -489,6 +513,8 @@ Other ops: \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
 \code{\link{op_vectorized_map}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_while_loop}()} \cr
diff --git a/man/op_round.Rd b/man/op_round.Rd
index d03b17664..4f02b7fe7 100644
--- a/man/op_round.Rd
+++ b/man/op_round.Rd
@@ -26,6 +26,7 @@ Other numpy ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -42,6 +43,7 @@ Other numpy ops: \cr
 \code{\link{op_argsort}()} \cr
 \code{\link{op_array}()} \cr
 \code{\link{op_average}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_bincount}()} \cr
 \code{\link{op_bitwise_and}()} \cr
 \code{\link{op_bitwise_invert}()} \cr
@@ -50,12 +52,15 @@ Other numpy ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_clip}()} \cr
 \code{\link{op_concatenate}()} \cr
 \code{\link{op_conj}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -64,6 +69,7 @@ Other numpy ops: \cr
 \code{\link{op_ctc_decode}()} \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_diag}()} \cr
 \code{\link{op_diagflat}()} \cr
 \code{\link{op_diagonal}()} \cr
@@ -88,6 +94,9 @@ Other numpy ops: \cr
 \code{\link{op_get_item}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -97,6 +106,7 @@ Other numpy ops: \cr
 \code{\link{op_isfinite}()} \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
+\code{\link{op_kaiser}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
 \code{\link{op_less_equal}()} \cr
@@ -176,6 +186,8 @@ Other numpy ops: \cr
 \code{\link{op_var}()} \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_zeros}()} \cr
@@ -185,6 +197,7 @@ Other ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -203,6 +216,7 @@ Other ops: \cr
 \code{\link{op_associative_scan}()} \cr
 \code{\link{op_average}()} \cr
 \code{\link{op_average_pool}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_batch_normalization}()} \cr
 \code{\link{op_binary_crossentropy}()} \cr
 \code{\link{op_bincount}()} \cr
@@ -213,9 +227,11 @@ Other ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
 \code{\link{op_cast}()} \cr
 \code{\link{op_categorical_crossentropy}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_celu}()} \cr
 \code{\link{op_cholesky}()} \cr
@@ -228,6 +244,7 @@ Other ops: \cr
 \code{\link{op_convert_to_numpy}()} \cr
 \code{\link{op_convert_to_tensor}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -238,6 +255,7 @@ Other ops: \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
 \code{\link{op_custom_gradient}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_depthwise_conv}()} \cr
 \code{\link{op_det}()} \cr
 \code{\link{op_diag}()} \cr
@@ -277,10 +295,13 @@ Other ops: \cr
 \code{\link{op_glu}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
 \code{\link{op_hard_shrink}()} \cr
 \code{\link{op_hard_sigmoid}()} \cr
 \code{\link{op_hard_silu}()} \cr
 \code{\link{op_hard_tanh}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -307,6 +328,8 @@ Other ops: \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
 \code{\link{op_istft}()} \cr
+\code{\link{op_kaiser}()} \cr
+\code{\link{op_layer_normalization}()} \cr
 \code{\link{op_leaky_relu}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
@@ -405,6 +428,7 @@ Other ops: \cr
 \code{\link{op_sort}()} \cr
 \code{\link{op_sparse_categorical_crossentropy}()} \cr
 \code{\link{op_sparse_plus}()} \cr
+\code{\link{op_sparse_sigmoid}()} \cr
 \code{\link{op_sparsemax}()} \cr
 \code{\link{op_split}()} \cr
 \code{\link{op_sqrt}()} \cr
@@ -442,6 +466,8 @@ Other ops: \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
 \code{\link{op_vectorized_map}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_while_loop}()} \cr
diff --git a/man/op_rsqrt.Rd b/man/op_rsqrt.Rd
index 3478b0012..cf2e66286 100644
--- a/man/op_rsqrt.Rd
+++ b/man/op_rsqrt.Rd
@@ -56,6 +56,7 @@ Other ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -74,6 +75,7 @@ Other ops: \cr
 \code{\link{op_associative_scan}()} \cr
 \code{\link{op_average}()} \cr
 \code{\link{op_average_pool}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_batch_normalization}()} \cr
 \code{\link{op_binary_crossentropy}()} \cr
 \code{\link{op_bincount}()} \cr
@@ -84,9 +86,11 @@ Other ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
 \code{\link{op_cast}()} \cr
 \code{\link{op_categorical_crossentropy}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_celu}()} \cr
 \code{\link{op_cholesky}()} \cr
@@ -99,6 +103,7 @@ Other ops: \cr
 \code{\link{op_convert_to_numpy}()} \cr
 \code{\link{op_convert_to_tensor}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -109,6 +114,7 @@ Other ops: \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
 \code{\link{op_custom_gradient}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_depthwise_conv}()} \cr
 \code{\link{op_det}()} \cr
 \code{\link{op_diag}()} \cr
@@ -148,10 +154,13 @@ Other ops: \cr
 \code{\link{op_glu}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
 \code{\link{op_hard_shrink}()} \cr
 \code{\link{op_hard_sigmoid}()} \cr
 \code{\link{op_hard_silu}()} \cr
 \code{\link{op_hard_tanh}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -178,6 +187,8 @@ Other ops: \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
 \code{\link{op_istft}()} \cr
+\code{\link{op_kaiser}()} \cr
+\code{\link{op_layer_normalization}()} \cr
 \code{\link{op_leaky_relu}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
@@ -276,6 +287,7 @@ Other ops: \cr
 \code{\link{op_sort}()} \cr
 \code{\link{op_sparse_categorical_crossentropy}()} \cr
 \code{\link{op_sparse_plus}()} \cr
+\code{\link{op_sparse_sigmoid}()} \cr
 \code{\link{op_sparsemax}()} \cr
 \code{\link{op_split}()} \cr
 \code{\link{op_sqrt}()} \cr
@@ -313,6 +325,8 @@ Other ops: \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
 \code{\link{op_vectorized_map}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_while_loop}()} \cr
diff --git a/man/op_saturate_cast.Rd b/man/op_saturate_cast.Rd
index 02a68dcdd..7533eac02 100644
--- a/man/op_saturate_cast.Rd
+++ b/man/op_saturate_cast.Rd
@@ -75,6 +75,7 @@ Other numpy ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -91,6 +92,7 @@ Other numpy ops: \cr
 \code{\link{op_argsort}()} \cr
 \code{\link{op_array}()} \cr
 \code{\link{op_average}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_bincount}()} \cr
 \code{\link{op_bitwise_and}()} \cr
 \code{\link{op_bitwise_invert}()} \cr
@@ -99,12 +101,15 @@ Other numpy ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_clip}()} \cr
 \code{\link{op_concatenate}()} \cr
 \code{\link{op_conj}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -113,6 +118,7 @@ Other numpy ops: \cr
 \code{\link{op_ctc_decode}()} \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_diag}()} \cr
 \code{\link{op_diagflat}()} \cr
 \code{\link{op_diagonal}()} \cr
@@ -137,6 +143,9 @@ Other numpy ops: \cr
 \code{\link{op_get_item}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -146,6 +155,7 @@ Other numpy ops: \cr
 \code{\link{op_isfinite}()} \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
+\code{\link{op_kaiser}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
 \code{\link{op_less_equal}()} \cr
@@ -225,6 +235,8 @@ Other numpy ops: \cr
 \code{\link{op_var}()} \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_zeros}()} \cr
@@ -234,6 +246,7 @@ Other ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -252,6 +265,7 @@ Other ops: \cr
 \code{\link{op_associative_scan}()} \cr
 \code{\link{op_average}()} \cr
 \code{\link{op_average_pool}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_batch_normalization}()} \cr
 \code{\link{op_binary_crossentropy}()} \cr
 \code{\link{op_bincount}()} \cr
@@ -262,9 +276,11 @@ Other ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
 \code{\link{op_cast}()} \cr
 \code{\link{op_categorical_crossentropy}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_celu}()} \cr
 \code{\link{op_cholesky}()} \cr
@@ -277,6 +293,7 @@ Other ops: \cr
 \code{\link{op_convert_to_numpy}()} \cr
 \code{\link{op_convert_to_tensor}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -287,6 +304,7 @@ Other ops: \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
 \code{\link{op_custom_gradient}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_depthwise_conv}()} \cr
 \code{\link{op_det}()} \cr
 \code{\link{op_diag}()} \cr
@@ -326,10 +344,13 @@ Other ops: \cr
 \code{\link{op_glu}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
 \code{\link{op_hard_shrink}()} \cr
 \code{\link{op_hard_sigmoid}()} \cr
 \code{\link{op_hard_silu}()} \cr
 \code{\link{op_hard_tanh}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -356,6 +377,8 @@ Other ops: \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
 \code{\link{op_istft}()} \cr
+\code{\link{op_kaiser}()} \cr
+\code{\link{op_layer_normalization}()} \cr
 \code{\link{op_leaky_relu}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
@@ -454,6 +477,7 @@ Other ops: \cr
 \code{\link{op_sort}()} \cr
 \code{\link{op_sparse_categorical_crossentropy}()} \cr
 \code{\link{op_sparse_plus}()} \cr
+\code{\link{op_sparse_sigmoid}()} \cr
 \code{\link{op_sparsemax}()} \cr
 \code{\link{op_split}()} \cr
 \code{\link{op_sqrt}()} \cr
@@ -491,6 +515,8 @@ Other ops: \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
 \code{\link{op_vectorized_map}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_while_loop}()} \cr
diff --git a/man/op_scan.Rd b/man/op_scan.Rd
index b74fbfc3e..162cc5e92 100644
--- a/man/op_scan.Rd
+++ b/man/op_scan.Rd
@@ -119,6 +119,7 @@ Other ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -137,6 +138,7 @@ Other ops: \cr
 \code{\link{op_associative_scan}()} \cr
 \code{\link{op_average}()} \cr
 \code{\link{op_average_pool}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_batch_normalization}()} \cr
 \code{\link{op_binary_crossentropy}()} \cr
 \code{\link{op_bincount}()} \cr
@@ -147,9 +149,11 @@ Other ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
 \code{\link{op_cast}()} \cr
 \code{\link{op_categorical_crossentropy}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_celu}()} \cr
 \code{\link{op_cholesky}()} \cr
@@ -162,6 +166,7 @@ Other ops: \cr
 \code{\link{op_convert_to_numpy}()} \cr
 \code{\link{op_convert_to_tensor}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -172,6 +177,7 @@ Other ops: \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
 \code{\link{op_custom_gradient}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_depthwise_conv}()} \cr
 \code{\link{op_det}()} \cr
 \code{\link{op_diag}()} \cr
@@ -211,10 +217,13 @@ Other ops: \cr
 \code{\link{op_glu}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
 \code{\link{op_hard_shrink}()} \cr
 \code{\link{op_hard_sigmoid}()} \cr
 \code{\link{op_hard_silu}()} \cr
 \code{\link{op_hard_tanh}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -241,6 +250,8 @@ Other ops: \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
 \code{\link{op_istft}()} \cr
+\code{\link{op_kaiser}()} \cr
+\code{\link{op_layer_normalization}()} \cr
 \code{\link{op_leaky_relu}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
@@ -339,6 +350,7 @@ Other ops: \cr
 \code{\link{op_sort}()} \cr
 \code{\link{op_sparse_categorical_crossentropy}()} \cr
 \code{\link{op_sparse_plus}()} \cr
+\code{\link{op_sparse_sigmoid}()} \cr
 \code{\link{op_sparsemax}()} \cr
 \code{\link{op_split}()} \cr
 \code{\link{op_sqrt}()} \cr
@@ -376,6 +388,8 @@ Other ops: \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
 \code{\link{op_vectorized_map}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_while_loop}()} \cr
diff --git a/man/op_scatter.Rd b/man/op_scatter.Rd
index 8efe31e0d..caa8b0250 100644
--- a/man/op_scatter.Rd
+++ b/man/op_scatter.Rd
@@ -85,6 +85,7 @@ Other ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -103,6 +104,7 @@ Other ops: \cr
 \code{\link{op_associative_scan}()} \cr
 \code{\link{op_average}()} \cr
 \code{\link{op_average_pool}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_batch_normalization}()} \cr
 \code{\link{op_binary_crossentropy}()} \cr
 \code{\link{op_bincount}()} \cr
@@ -113,9 +115,11 @@ Other ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
 \code{\link{op_cast}()} \cr
 \code{\link{op_categorical_crossentropy}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_celu}()} \cr
 \code{\link{op_cholesky}()} \cr
@@ -128,6 +132,7 @@ Other ops: \cr
 \code{\link{op_convert_to_numpy}()} \cr
 \code{\link{op_convert_to_tensor}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -138,6 +143,7 @@ Other ops: \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
 \code{\link{op_custom_gradient}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_depthwise_conv}()} \cr
 \code{\link{op_det}()} \cr
 \code{\link{op_diag}()} \cr
@@ -177,10 +183,13 @@ Other ops: \cr
 \code{\link{op_glu}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
 \code{\link{op_hard_shrink}()} \cr
 \code{\link{op_hard_sigmoid}()} \cr
 \code{\link{op_hard_silu}()} \cr
 \code{\link{op_hard_tanh}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -207,6 +216,8 @@ Other ops: \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
 \code{\link{op_istft}()} \cr
+\code{\link{op_kaiser}()} \cr
+\code{\link{op_layer_normalization}()} \cr
 \code{\link{op_leaky_relu}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
@@ -305,6 +316,7 @@ Other ops: \cr
 \code{\link{op_sort}()} \cr
 \code{\link{op_sparse_categorical_crossentropy}()} \cr
 \code{\link{op_sparse_plus}()} \cr
+\code{\link{op_sparse_sigmoid}()} \cr
 \code{\link{op_sparsemax}()} \cr
 \code{\link{op_split}()} \cr
 \code{\link{op_sqrt}()} \cr
@@ -342,6 +354,8 @@ Other ops: \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
 \code{\link{op_vectorized_map}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_while_loop}()} \cr
diff --git a/man/op_scatter_update.Rd b/man/op_scatter_update.Rd
index 373a51d5d..43add5106 100644
--- a/man/op_scatter_update.Rd
+++ b/man/op_scatter_update.Rd
@@ -135,6 +135,7 @@ Other ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -153,6 +154,7 @@ Other ops: \cr
 \code{\link{op_associative_scan}()} \cr
 \code{\link{op_average}()} \cr
 \code{\link{op_average_pool}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_batch_normalization}()} \cr
 \code{\link{op_binary_crossentropy}()} \cr
 \code{\link{op_bincount}()} \cr
@@ -163,9 +165,11 @@ Other ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
 \code{\link{op_cast}()} \cr
 \code{\link{op_categorical_crossentropy}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_celu}()} \cr
 \code{\link{op_cholesky}()} \cr
@@ -178,6 +182,7 @@ Other ops: \cr
 \code{\link{op_convert_to_numpy}()} \cr
 \code{\link{op_convert_to_tensor}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -188,6 +193,7 @@ Other ops: \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
 \code{\link{op_custom_gradient}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_depthwise_conv}()} \cr
 \code{\link{op_det}()} \cr
 \code{\link{op_diag}()} \cr
@@ -227,10 +233,13 @@ Other ops: \cr
 \code{\link{op_glu}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
 \code{\link{op_hard_shrink}()} \cr
 \code{\link{op_hard_sigmoid}()} \cr
 \code{\link{op_hard_silu}()} \cr
 \code{\link{op_hard_tanh}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -257,6 +266,8 @@ Other ops: \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
 \code{\link{op_istft}()} \cr
+\code{\link{op_kaiser}()} \cr
+\code{\link{op_layer_normalization}()} \cr
 \code{\link{op_leaky_relu}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
@@ -355,6 +366,7 @@ Other ops: \cr
 \code{\link{op_sort}()} \cr
 \code{\link{op_sparse_categorical_crossentropy}()} \cr
 \code{\link{op_sparse_plus}()} \cr
+\code{\link{op_sparse_sigmoid}()} \cr
 \code{\link{op_sparsemax}()} \cr
 \code{\link{op_split}()} \cr
 \code{\link{op_sqrt}()} \cr
@@ -392,6 +404,8 @@ Other ops: \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
 \code{\link{op_vectorized_map}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_while_loop}()} \cr
diff --git a/man/op_searchsorted.Rd b/man/op_searchsorted.Rd
index 9b22f3101..7d0f9cf38 100644
--- a/man/op_searchsorted.Rd
+++ b/man/op_searchsorted.Rd
@@ -56,6 +56,7 @@ Other ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -74,6 +75,7 @@ Other ops: \cr
 \code{\link{op_associative_scan}()} \cr
 \code{\link{op_average}()} \cr
 \code{\link{op_average_pool}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_batch_normalization}()} \cr
 \code{\link{op_binary_crossentropy}()} \cr
 \code{\link{op_bincount}()} \cr
@@ -84,9 +86,11 @@ Other ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
 \code{\link{op_cast}()} \cr
 \code{\link{op_categorical_crossentropy}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_celu}()} \cr
 \code{\link{op_cholesky}()} \cr
@@ -99,6 +103,7 @@ Other ops: \cr
 \code{\link{op_convert_to_numpy}()} \cr
 \code{\link{op_convert_to_tensor}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -109,6 +114,7 @@ Other ops: \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
 \code{\link{op_custom_gradient}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_depthwise_conv}()} \cr
 \code{\link{op_det}()} \cr
 \code{\link{op_diag}()} \cr
@@ -148,10 +154,13 @@ Other ops: \cr
 \code{\link{op_glu}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
 \code{\link{op_hard_shrink}()} \cr
 \code{\link{op_hard_sigmoid}()} \cr
 \code{\link{op_hard_silu}()} \cr
 \code{\link{op_hard_tanh}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -178,6 +187,8 @@ Other ops: \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
 \code{\link{op_istft}()} \cr
+\code{\link{op_kaiser}()} \cr
+\code{\link{op_layer_normalization}()} \cr
 \code{\link{op_leaky_relu}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
@@ -276,6 +287,7 @@ Other ops: \cr
 \code{\link{op_sort}()} \cr
 \code{\link{op_sparse_categorical_crossentropy}()} \cr
 \code{\link{op_sparse_plus}()} \cr
+\code{\link{op_sparse_sigmoid}()} \cr
 \code{\link{op_sparsemax}()} \cr
 \code{\link{op_split}()} \cr
 \code{\link{op_sqrt}()} \cr
@@ -313,6 +325,8 @@ Other ops: \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
 \code{\link{op_vectorized_map}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_while_loop}()} \cr
diff --git a/man/op_segment_max.Rd b/man/op_segment_max.Rd
index a851afd47..2285f4bb9 100644
--- a/man/op_segment_max.Rd
+++ b/man/op_segment_max.Rd
@@ -71,6 +71,7 @@ Other ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -89,6 +90,7 @@ Other ops: \cr
 \code{\link{op_associative_scan}()} \cr
 \code{\link{op_average}()} \cr
 \code{\link{op_average_pool}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_batch_normalization}()} \cr
 \code{\link{op_binary_crossentropy}()} \cr
 \code{\link{op_bincount}()} \cr
@@ -99,9 +101,11 @@ Other ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
 \code{\link{op_cast}()} \cr
 \code{\link{op_categorical_crossentropy}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_celu}()} \cr
 \code{\link{op_cholesky}()} \cr
@@ -114,6 +118,7 @@ Other ops: \cr
 \code{\link{op_convert_to_numpy}()} \cr
 \code{\link{op_convert_to_tensor}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -124,6 +129,7 @@ Other ops: \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
 \code{\link{op_custom_gradient}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_depthwise_conv}()} \cr
 \code{\link{op_det}()} \cr
 \code{\link{op_diag}()} \cr
@@ -163,10 +169,13 @@ Other ops: \cr
 \code{\link{op_glu}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
 \code{\link{op_hard_shrink}()} \cr
 \code{\link{op_hard_sigmoid}()} \cr
 \code{\link{op_hard_silu}()} \cr
 \code{\link{op_hard_tanh}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -193,6 +202,8 @@ Other ops: \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
 \code{\link{op_istft}()} \cr
+\code{\link{op_kaiser}()} \cr
+\code{\link{op_layer_normalization}()} \cr
 \code{\link{op_leaky_relu}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
@@ -291,6 +302,7 @@ Other ops: \cr
 \code{\link{op_sort}()} \cr
 \code{\link{op_sparse_categorical_crossentropy}()} \cr
 \code{\link{op_sparse_plus}()} \cr
+\code{\link{op_sparse_sigmoid}()} \cr
 \code{\link{op_sparsemax}()} \cr
 \code{\link{op_split}()} \cr
 \code{\link{op_sqrt}()} \cr
@@ -328,6 +340,8 @@ Other ops: \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
 \code{\link{op_vectorized_map}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_while_loop}()} \cr
diff --git a/man/op_segment_sum.Rd b/man/op_segment_sum.Rd
index b0f9b63be..5de30c95e 100644
--- a/man/op_segment_sum.Rd
+++ b/man/op_segment_sum.Rd
@@ -67,6 +67,7 @@ Other ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -85,6 +86,7 @@ Other ops: \cr
 \code{\link{op_associative_scan}()} \cr
 \code{\link{op_average}()} \cr
 \code{\link{op_average_pool}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_batch_normalization}()} \cr
 \code{\link{op_binary_crossentropy}()} \cr
 \code{\link{op_bincount}()} \cr
@@ -95,9 +97,11 @@ Other ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
 \code{\link{op_cast}()} \cr
 \code{\link{op_categorical_crossentropy}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_celu}()} \cr
 \code{\link{op_cholesky}()} \cr
@@ -110,6 +114,7 @@ Other ops: \cr
 \code{\link{op_convert_to_numpy}()} \cr
 \code{\link{op_convert_to_tensor}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -120,6 +125,7 @@ Other ops: \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
 \code{\link{op_custom_gradient}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_depthwise_conv}()} \cr
 \code{\link{op_det}()} \cr
 \code{\link{op_diag}()} \cr
@@ -159,10 +165,13 @@ Other ops: \cr
 \code{\link{op_glu}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
 \code{\link{op_hard_shrink}()} \cr
 \code{\link{op_hard_sigmoid}()} \cr
 \code{\link{op_hard_silu}()} \cr
 \code{\link{op_hard_tanh}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -189,6 +198,8 @@ Other ops: \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
 \code{\link{op_istft}()} \cr
+\code{\link{op_kaiser}()} \cr
+\code{\link{op_layer_normalization}()} \cr
 \code{\link{op_leaky_relu}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
@@ -287,6 +298,7 @@ Other ops: \cr
 \code{\link{op_sort}()} \cr
 \code{\link{op_sparse_categorical_crossentropy}()} \cr
 \code{\link{op_sparse_plus}()} \cr
+\code{\link{op_sparse_sigmoid}()} \cr
 \code{\link{op_sparsemax}()} \cr
 \code{\link{op_split}()} \cr
 \code{\link{op_sqrt}()} \cr
@@ -324,6 +336,8 @@ Other ops: \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
 \code{\link{op_vectorized_map}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_while_loop}()} \cr
diff --git a/man/op_select.Rd b/man/op_select.Rd
index acdd63c86..b6700d5d0 100644
--- a/man/op_select.Rd
+++ b/man/op_select.Rd
@@ -46,6 +46,7 @@ Other numpy ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -62,6 +63,7 @@ Other numpy ops: \cr
 \code{\link{op_argsort}()} \cr
 \code{\link{op_array}()} \cr
 \code{\link{op_average}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_bincount}()} \cr
 \code{\link{op_bitwise_and}()} \cr
 \code{\link{op_bitwise_invert}()} \cr
@@ -70,12 +72,15 @@ Other numpy ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_clip}()} \cr
 \code{\link{op_concatenate}()} \cr
 \code{\link{op_conj}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -84,6 +89,7 @@ Other numpy ops: \cr
 \code{\link{op_ctc_decode}()} \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_diag}()} \cr
 \code{\link{op_diagflat}()} \cr
 \code{\link{op_diagonal}()} \cr
@@ -108,6 +114,9 @@ Other numpy ops: \cr
 \code{\link{op_get_item}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -117,6 +126,7 @@ Other numpy ops: \cr
 \code{\link{op_isfinite}()} \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
+\code{\link{op_kaiser}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
 \code{\link{op_less_equal}()} \cr
@@ -196,6 +206,8 @@ Other numpy ops: \cr
 \code{\link{op_var}()} \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_zeros}()} \cr
@@ -205,6 +217,7 @@ Other ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -223,6 +236,7 @@ Other ops: \cr
 \code{\link{op_associative_scan}()} \cr
 \code{\link{op_average}()} \cr
 \code{\link{op_average_pool}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_batch_normalization}()} \cr
 \code{\link{op_binary_crossentropy}()} \cr
 \code{\link{op_bincount}()} \cr
@@ -233,9 +247,11 @@ Other ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
 \code{\link{op_cast}()} \cr
 \code{\link{op_categorical_crossentropy}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_celu}()} \cr
 \code{\link{op_cholesky}()} \cr
@@ -248,6 +264,7 @@ Other ops: \cr
 \code{\link{op_convert_to_numpy}()} \cr
 \code{\link{op_convert_to_tensor}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -258,6 +275,7 @@ Other ops: \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
 \code{\link{op_custom_gradient}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_depthwise_conv}()} \cr
 \code{\link{op_det}()} \cr
 \code{\link{op_diag}()} \cr
@@ -297,10 +315,13 @@ Other ops: \cr
 \code{\link{op_glu}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
 \code{\link{op_hard_shrink}()} \cr
 \code{\link{op_hard_sigmoid}()} \cr
 \code{\link{op_hard_silu}()} \cr
 \code{\link{op_hard_tanh}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -327,6 +348,8 @@ Other ops: \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
 \code{\link{op_istft}()} \cr
+\code{\link{op_kaiser}()} \cr
+\code{\link{op_layer_normalization}()} \cr
 \code{\link{op_leaky_relu}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
@@ -425,6 +448,7 @@ Other ops: \cr
 \code{\link{op_sort}()} \cr
 \code{\link{op_sparse_categorical_crossentropy}()} \cr
 \code{\link{op_sparse_plus}()} \cr
+\code{\link{op_sparse_sigmoid}()} \cr
 \code{\link{op_sparsemax}()} \cr
 \code{\link{op_split}()} \cr
 \code{\link{op_sqrt}()} \cr
@@ -462,6 +486,8 @@ Other ops: \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
 \code{\link{op_vectorized_map}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_while_loop}()} \cr
diff --git a/man/op_selu.Rd b/man/op_selu.Rd
index ef7e4b66c..3dceb71f4 100644
--- a/man/op_selu.Rd
+++ b/man/op_selu.Rd
@@ -51,6 +51,7 @@ Other nn ops: \cr
 \code{\link{op_hard_sigmoid}()} \cr
 \code{\link{op_hard_silu}()} \cr
 \code{\link{op_hard_tanh}()} \cr
+\code{\link{op_layer_normalization}()} \cr
 \code{\link{op_leaky_relu}()} \cr
 \code{\link{op_log_sigmoid}()} \cr
 \code{\link{op_log_softmax}()} \cr
@@ -73,6 +74,7 @@ Other nn ops: \cr
 \code{\link{op_softsign}()} \cr
 \code{\link{op_sparse_categorical_crossentropy}()} \cr
 \code{\link{op_sparse_plus}()} \cr
+\code{\link{op_sparse_sigmoid}()} \cr
 \code{\link{op_sparsemax}()} \cr
 \code{\link{op_squareplus}()} \cr
 \code{\link{op_tanh_shrink}()} \cr
@@ -83,6 +85,7 @@ Other ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -101,6 +104,7 @@ Other ops: \cr
 \code{\link{op_associative_scan}()} \cr
 \code{\link{op_average}()} \cr
 \code{\link{op_average_pool}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_batch_normalization}()} \cr
 \code{\link{op_binary_crossentropy}()} \cr
 \code{\link{op_bincount}()} \cr
@@ -111,9 +115,11 @@ Other ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
 \code{\link{op_cast}()} \cr
 \code{\link{op_categorical_crossentropy}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_celu}()} \cr
 \code{\link{op_cholesky}()} \cr
@@ -126,6 +132,7 @@ Other ops: \cr
 \code{\link{op_convert_to_numpy}()} \cr
 \code{\link{op_convert_to_tensor}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -136,6 +143,7 @@ Other ops: \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
 \code{\link{op_custom_gradient}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_depthwise_conv}()} \cr
 \code{\link{op_det}()} \cr
 \code{\link{op_diag}()} \cr
@@ -175,10 +183,13 @@ Other ops: \cr
 \code{\link{op_glu}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
 \code{\link{op_hard_shrink}()} \cr
 \code{\link{op_hard_sigmoid}()} \cr
 \code{\link{op_hard_silu}()} \cr
 \code{\link{op_hard_tanh}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -205,6 +216,8 @@ Other ops: \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
 \code{\link{op_istft}()} \cr
+\code{\link{op_kaiser}()} \cr
+\code{\link{op_layer_normalization}()} \cr
 \code{\link{op_leaky_relu}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
@@ -303,6 +316,7 @@ Other ops: \cr
 \code{\link{op_sort}()} \cr
 \code{\link{op_sparse_categorical_crossentropy}()} \cr
 \code{\link{op_sparse_plus}()} \cr
+\code{\link{op_sparse_sigmoid}()} \cr
 \code{\link{op_sparsemax}()} \cr
 \code{\link{op_split}()} \cr
 \code{\link{op_sqrt}()} \cr
@@ -340,6 +354,8 @@ Other ops: \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
 \code{\link{op_vectorized_map}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_while_loop}()} \cr
diff --git a/man/op_separable_conv.Rd b/man/op_separable_conv.Rd
index f7830f70e..9234cb37b 100644
--- a/man/op_separable_conv.Rd
+++ b/man/op_separable_conv.Rd
@@ -81,6 +81,7 @@ Other nn ops: \cr
 \code{\link{op_hard_sigmoid}()} \cr
 \code{\link{op_hard_silu}()} \cr
 \code{\link{op_hard_tanh}()} \cr
+\code{\link{op_layer_normalization}()} \cr
 \code{\link{op_leaky_relu}()} \cr
 \code{\link{op_log_sigmoid}()} \cr
 \code{\link{op_log_softmax}()} \cr
@@ -103,6 +104,7 @@ Other nn ops: \cr
 \code{\link{op_softsign}()} \cr
 \code{\link{op_sparse_categorical_crossentropy}()} \cr
 \code{\link{op_sparse_plus}()} \cr
+\code{\link{op_sparse_sigmoid}()} \cr
 \code{\link{op_sparsemax}()} \cr
 \code{\link{op_squareplus}()} \cr
 \code{\link{op_tanh_shrink}()} \cr
@@ -113,6 +115,7 @@ Other ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -131,6 +134,7 @@ Other ops: \cr
 \code{\link{op_associative_scan}()} \cr
 \code{\link{op_average}()} \cr
 \code{\link{op_average_pool}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_batch_normalization}()} \cr
 \code{\link{op_binary_crossentropy}()} \cr
 \code{\link{op_bincount}()} \cr
@@ -141,9 +145,11 @@ Other ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
 \code{\link{op_cast}()} \cr
 \code{\link{op_categorical_crossentropy}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_celu}()} \cr
 \code{\link{op_cholesky}()} \cr
@@ -156,6 +162,7 @@ Other ops: \cr
 \code{\link{op_convert_to_numpy}()} \cr
 \code{\link{op_convert_to_tensor}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -166,6 +173,7 @@ Other ops: \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
 \code{\link{op_custom_gradient}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_depthwise_conv}()} \cr
 \code{\link{op_det}()} \cr
 \code{\link{op_diag}()} \cr
@@ -205,10 +213,13 @@ Other ops: \cr
 \code{\link{op_glu}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
 \code{\link{op_hard_shrink}()} \cr
 \code{\link{op_hard_sigmoid}()} \cr
 \code{\link{op_hard_silu}()} \cr
 \code{\link{op_hard_tanh}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -235,6 +246,8 @@ Other ops: \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
 \code{\link{op_istft}()} \cr
+\code{\link{op_kaiser}()} \cr
+\code{\link{op_layer_normalization}()} \cr
 \code{\link{op_leaky_relu}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
@@ -333,6 +346,7 @@ Other ops: \cr
 \code{\link{op_sort}()} \cr
 \code{\link{op_sparse_categorical_crossentropy}()} \cr
 \code{\link{op_sparse_plus}()} \cr
+\code{\link{op_sparse_sigmoid}()} \cr
 \code{\link{op_sparsemax}()} \cr
 \code{\link{op_split}()} \cr
 \code{\link{op_sqrt}()} \cr
@@ -370,6 +384,8 @@ Other ops: \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
 \code{\link{op_vectorized_map}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_while_loop}()} \cr
diff --git a/man/op_shape.Rd b/man/op_shape.Rd
index a19f20822..8f0369ed8 100644
--- a/man/op_shape.Rd
+++ b/man/op_shape.Rd
@@ -67,6 +67,7 @@ Other ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -85,6 +86,7 @@ Other ops: \cr
 \code{\link{op_associative_scan}()} \cr
 \code{\link{op_average}()} \cr
 \code{\link{op_average_pool}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_batch_normalization}()} \cr
 \code{\link{op_binary_crossentropy}()} \cr
 \code{\link{op_bincount}()} \cr
@@ -95,9 +97,11 @@ Other ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
 \code{\link{op_cast}()} \cr
 \code{\link{op_categorical_crossentropy}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_celu}()} \cr
 \code{\link{op_cholesky}()} \cr
@@ -110,6 +114,7 @@ Other ops: \cr
 \code{\link{op_convert_to_numpy}()} \cr
 \code{\link{op_convert_to_tensor}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -120,6 +125,7 @@ Other ops: \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
 \code{\link{op_custom_gradient}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_depthwise_conv}()} \cr
 \code{\link{op_det}()} \cr
 \code{\link{op_diag}()} \cr
@@ -159,10 +165,13 @@ Other ops: \cr
 \code{\link{op_glu}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
 \code{\link{op_hard_shrink}()} \cr
 \code{\link{op_hard_sigmoid}()} \cr
 \code{\link{op_hard_silu}()} \cr
 \code{\link{op_hard_tanh}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -189,6 +198,8 @@ Other ops: \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
 \code{\link{op_istft}()} \cr
+\code{\link{op_kaiser}()} \cr
+\code{\link{op_layer_normalization}()} \cr
 \code{\link{op_leaky_relu}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
@@ -287,6 +298,7 @@ Other ops: \cr
 \code{\link{op_sort}()} \cr
 \code{\link{op_sparse_categorical_crossentropy}()} \cr
 \code{\link{op_sparse_plus}()} \cr
+\code{\link{op_sparse_sigmoid}()} \cr
 \code{\link{op_sparsemax}()} \cr
 \code{\link{op_split}()} \cr
 \code{\link{op_sqrt}()} \cr
@@ -324,6 +336,8 @@ Other ops: \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
 \code{\link{op_vectorized_map}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_while_loop}()} \cr
diff --git a/man/op_sigmoid.Rd b/man/op_sigmoid.Rd
index 4758d180f..43722c1dd 100644
--- a/man/op_sigmoid.Rd
+++ b/man/op_sigmoid.Rd
@@ -48,6 +48,7 @@ Other nn ops: \cr
 \code{\link{op_hard_sigmoid}()} \cr
 \code{\link{op_hard_silu}()} \cr
 \code{\link{op_hard_tanh}()} \cr
+\code{\link{op_layer_normalization}()} \cr
 \code{\link{op_leaky_relu}()} \cr
 \code{\link{op_log_sigmoid}()} \cr
 \code{\link{op_log_softmax}()} \cr
@@ -70,6 +71,7 @@ Other nn ops: \cr
 \code{\link{op_softsign}()} \cr
 \code{\link{op_sparse_categorical_crossentropy}()} \cr
 \code{\link{op_sparse_plus}()} \cr
+\code{\link{op_sparse_sigmoid}()} \cr
 \code{\link{op_sparsemax}()} \cr
 \code{\link{op_squareplus}()} \cr
 \code{\link{op_tanh_shrink}()} \cr
@@ -80,6 +82,7 @@ Other ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -98,6 +101,7 @@ Other ops: \cr
 \code{\link{op_associative_scan}()} \cr
 \code{\link{op_average}()} \cr
 \code{\link{op_average_pool}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_batch_normalization}()} \cr
 \code{\link{op_binary_crossentropy}()} \cr
 \code{\link{op_bincount}()} \cr
@@ -108,9 +112,11 @@ Other ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
 \code{\link{op_cast}()} \cr
 \code{\link{op_categorical_crossentropy}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_celu}()} \cr
 \code{\link{op_cholesky}()} \cr
@@ -123,6 +129,7 @@ Other ops: \cr
 \code{\link{op_convert_to_numpy}()} \cr
 \code{\link{op_convert_to_tensor}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -133,6 +140,7 @@ Other ops: \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
 \code{\link{op_custom_gradient}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_depthwise_conv}()} \cr
 \code{\link{op_det}()} \cr
 \code{\link{op_diag}()} \cr
@@ -172,10 +180,13 @@ Other ops: \cr
 \code{\link{op_glu}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
 \code{\link{op_hard_shrink}()} \cr
 \code{\link{op_hard_sigmoid}()} \cr
 \code{\link{op_hard_silu}()} \cr
 \code{\link{op_hard_tanh}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -202,6 +213,8 @@ Other ops: \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
 \code{\link{op_istft}()} \cr
+\code{\link{op_kaiser}()} \cr
+\code{\link{op_layer_normalization}()} \cr
 \code{\link{op_leaky_relu}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
@@ -300,6 +313,7 @@ Other ops: \cr
 \code{\link{op_sort}()} \cr
 \code{\link{op_sparse_categorical_crossentropy}()} \cr
 \code{\link{op_sparse_plus}()} \cr
+\code{\link{op_sparse_sigmoid}()} \cr
 \code{\link{op_sparsemax}()} \cr
 \code{\link{op_split}()} \cr
 \code{\link{op_sqrt}()} \cr
@@ -337,6 +351,8 @@ Other ops: \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
 \code{\link{op_vectorized_map}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_while_loop}()} \cr
diff --git a/man/op_sign.Rd b/man/op_sign.Rd
index d9965af41..d8b81139d 100644
--- a/man/op_sign.Rd
+++ b/man/op_sign.Rd
@@ -24,6 +24,7 @@ Other numpy ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -40,6 +41,7 @@ Other numpy ops: \cr
 \code{\link{op_argsort}()} \cr
 \code{\link{op_array}()} \cr
 \code{\link{op_average}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_bincount}()} \cr
 \code{\link{op_bitwise_and}()} \cr
 \code{\link{op_bitwise_invert}()} \cr
@@ -48,12 +50,15 @@ Other numpy ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_clip}()} \cr
 \code{\link{op_concatenate}()} \cr
 \code{\link{op_conj}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -62,6 +67,7 @@ Other numpy ops: \cr
 \code{\link{op_ctc_decode}()} \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_diag}()} \cr
 \code{\link{op_diagflat}()} \cr
 \code{\link{op_diagonal}()} \cr
@@ -86,6 +92,9 @@ Other numpy ops: \cr
 \code{\link{op_get_item}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -95,6 +104,7 @@ Other numpy ops: \cr
 \code{\link{op_isfinite}()} \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
+\code{\link{op_kaiser}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
 \code{\link{op_less_equal}()} \cr
@@ -174,6 +184,8 @@ Other numpy ops: \cr
 \code{\link{op_var}()} \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_zeros}()} \cr
@@ -183,6 +195,7 @@ Other ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -201,6 +214,7 @@ Other ops: \cr
 \code{\link{op_associative_scan}()} \cr
 \code{\link{op_average}()} \cr
 \code{\link{op_average_pool}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_batch_normalization}()} \cr
 \code{\link{op_binary_crossentropy}()} \cr
 \code{\link{op_bincount}()} \cr
@@ -211,9 +225,11 @@ Other ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
 \code{\link{op_cast}()} \cr
 \code{\link{op_categorical_crossentropy}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_celu}()} \cr
 \code{\link{op_cholesky}()} \cr
@@ -226,6 +242,7 @@ Other ops: \cr
 \code{\link{op_convert_to_numpy}()} \cr
 \code{\link{op_convert_to_tensor}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -236,6 +253,7 @@ Other ops: \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
 \code{\link{op_custom_gradient}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_depthwise_conv}()} \cr
 \code{\link{op_det}()} \cr
 \code{\link{op_diag}()} \cr
@@ -275,10 +293,13 @@ Other ops: \cr
 \code{\link{op_glu}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
 \code{\link{op_hard_shrink}()} \cr
 \code{\link{op_hard_sigmoid}()} \cr
 \code{\link{op_hard_silu}()} \cr
 \code{\link{op_hard_tanh}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -305,6 +326,8 @@ Other ops: \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
 \code{\link{op_istft}()} \cr
+\code{\link{op_kaiser}()} \cr
+\code{\link{op_layer_normalization}()} \cr
 \code{\link{op_leaky_relu}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
@@ -403,6 +426,7 @@ Other ops: \cr
 \code{\link{op_sort}()} \cr
 \code{\link{op_sparse_categorical_crossentropy}()} \cr
 \code{\link{op_sparse_plus}()} \cr
+\code{\link{op_sparse_sigmoid}()} \cr
 \code{\link{op_sparsemax}()} \cr
 \code{\link{op_split}()} \cr
 \code{\link{op_sqrt}()} \cr
@@ -440,6 +464,8 @@ Other ops: \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
 \code{\link{op_vectorized_map}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_while_loop}()} \cr
diff --git a/man/op_signbit.Rd b/man/op_signbit.Rd
index 14370774a..afc1b4c1f 100644
--- a/man/op_signbit.Rd
+++ b/man/op_signbit.Rd
@@ -21,6 +21,7 @@ Other numpy ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -37,6 +38,7 @@ Other numpy ops: \cr
 \code{\link{op_argsort}()} \cr
 \code{\link{op_array}()} \cr
 \code{\link{op_average}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_bincount}()} \cr
 \code{\link{op_bitwise_and}()} \cr
 \code{\link{op_bitwise_invert}()} \cr
@@ -45,12 +47,15 @@ Other numpy ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_clip}()} \cr
 \code{\link{op_concatenate}()} \cr
 \code{\link{op_conj}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -59,6 +64,7 @@ Other numpy ops: \cr
 \code{\link{op_ctc_decode}()} \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_diag}()} \cr
 \code{\link{op_diagflat}()} \cr
 \code{\link{op_diagonal}()} \cr
@@ -83,6 +89,9 @@ Other numpy ops: \cr
 \code{\link{op_get_item}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -92,6 +101,7 @@ Other numpy ops: \cr
 \code{\link{op_isfinite}()} \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
+\code{\link{op_kaiser}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
 \code{\link{op_less_equal}()} \cr
@@ -171,6 +181,8 @@ Other numpy ops: \cr
 \code{\link{op_var}()} \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_zeros}()} \cr
@@ -180,6 +192,7 @@ Other ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -198,6 +211,7 @@ Other ops: \cr
 \code{\link{op_associative_scan}()} \cr
 \code{\link{op_average}()} \cr
 \code{\link{op_average_pool}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_batch_normalization}()} \cr
 \code{\link{op_binary_crossentropy}()} \cr
 \code{\link{op_bincount}()} \cr
@@ -208,9 +222,11 @@ Other ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
 \code{\link{op_cast}()} \cr
 \code{\link{op_categorical_crossentropy}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_celu}()} \cr
 \code{\link{op_cholesky}()} \cr
@@ -223,6 +239,7 @@ Other ops: \cr
 \code{\link{op_convert_to_numpy}()} \cr
 \code{\link{op_convert_to_tensor}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -233,6 +250,7 @@ Other ops: \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
 \code{\link{op_custom_gradient}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_depthwise_conv}()} \cr
 \code{\link{op_det}()} \cr
 \code{\link{op_diag}()} \cr
@@ -272,10 +290,13 @@ Other ops: \cr
 \code{\link{op_glu}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
 \code{\link{op_hard_shrink}()} \cr
 \code{\link{op_hard_sigmoid}()} \cr
 \code{\link{op_hard_silu}()} \cr
 \code{\link{op_hard_tanh}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -302,6 +323,8 @@ Other ops: \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
 \code{\link{op_istft}()} \cr
+\code{\link{op_kaiser}()} \cr
+\code{\link{op_layer_normalization}()} \cr
 \code{\link{op_leaky_relu}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
@@ -400,6 +423,7 @@ Other ops: \cr
 \code{\link{op_sort}()} \cr
 \code{\link{op_sparse_categorical_crossentropy}()} \cr
 \code{\link{op_sparse_plus}()} \cr
+\code{\link{op_sparse_sigmoid}()} \cr
 \code{\link{op_sparsemax}()} \cr
 \code{\link{op_split}()} \cr
 \code{\link{op_sqrt}()} \cr
@@ -437,6 +461,8 @@ Other ops: \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
 \code{\link{op_vectorized_map}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_while_loop}()} \cr
diff --git a/man/op_silu.Rd b/man/op_silu.Rd
index e78037d0b..427a0142c 100644
--- a/man/op_silu.Rd
+++ b/man/op_silu.Rd
@@ -56,6 +56,7 @@ Other nn ops: \cr
 \code{\link{op_hard_sigmoid}()} \cr
 \code{\link{op_hard_silu}()} \cr
 \code{\link{op_hard_tanh}()} \cr
+\code{\link{op_layer_normalization}()} \cr
 \code{\link{op_leaky_relu}()} \cr
 \code{\link{op_log_sigmoid}()} \cr
 \code{\link{op_log_softmax}()} \cr
@@ -78,6 +79,7 @@ Other nn ops: \cr
 \code{\link{op_softsign}()} \cr
 \code{\link{op_sparse_categorical_crossentropy}()} \cr
 \code{\link{op_sparse_plus}()} \cr
+\code{\link{op_sparse_sigmoid}()} \cr
 \code{\link{op_sparsemax}()} \cr
 \code{\link{op_squareplus}()} \cr
 \code{\link{op_tanh_shrink}()} \cr
@@ -88,6 +90,7 @@ Other ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -106,6 +109,7 @@ Other ops: \cr
 \code{\link{op_associative_scan}()} \cr
 \code{\link{op_average}()} \cr
 \code{\link{op_average_pool}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_batch_normalization}()} \cr
 \code{\link{op_binary_crossentropy}()} \cr
 \code{\link{op_bincount}()} \cr
@@ -116,9 +120,11 @@ Other ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
 \code{\link{op_cast}()} \cr
 \code{\link{op_categorical_crossentropy}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_celu}()} \cr
 \code{\link{op_cholesky}()} \cr
@@ -131,6 +137,7 @@ Other ops: \cr
 \code{\link{op_convert_to_numpy}()} \cr
 \code{\link{op_convert_to_tensor}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -141,6 +148,7 @@ Other ops: \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
 \code{\link{op_custom_gradient}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_depthwise_conv}()} \cr
 \code{\link{op_det}()} \cr
 \code{\link{op_diag}()} \cr
@@ -180,10 +188,13 @@ Other ops: \cr
 \code{\link{op_glu}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
 \code{\link{op_hard_shrink}()} \cr
 \code{\link{op_hard_sigmoid}()} \cr
 \code{\link{op_hard_silu}()} \cr
 \code{\link{op_hard_tanh}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -210,6 +221,8 @@ Other ops: \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
 \code{\link{op_istft}()} \cr
+\code{\link{op_kaiser}()} \cr
+\code{\link{op_layer_normalization}()} \cr
 \code{\link{op_leaky_relu}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
@@ -308,6 +321,7 @@ Other ops: \cr
 \code{\link{op_sort}()} \cr
 \code{\link{op_sparse_categorical_crossentropy}()} \cr
 \code{\link{op_sparse_plus}()} \cr
+\code{\link{op_sparse_sigmoid}()} \cr
 \code{\link{op_sparsemax}()} \cr
 \code{\link{op_split}()} \cr
 \code{\link{op_sqrt}()} \cr
@@ -345,6 +359,8 @@ Other ops: \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
 \code{\link{op_vectorized_map}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_while_loop}()} \cr
diff --git a/man/op_sin.Rd b/man/op_sin.Rd
index be1a303b4..1aaab4c45 100644
--- a/man/op_sin.Rd
+++ b/man/op_sin.Rd
@@ -24,6 +24,7 @@ Other numpy ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -40,6 +41,7 @@ Other numpy ops: \cr
 \code{\link{op_argsort}()} \cr
 \code{\link{op_array}()} \cr
 \code{\link{op_average}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_bincount}()} \cr
 \code{\link{op_bitwise_and}()} \cr
 \code{\link{op_bitwise_invert}()} \cr
@@ -48,12 +50,15 @@ Other numpy ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_clip}()} \cr
 \code{\link{op_concatenate}()} \cr
 \code{\link{op_conj}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -62,6 +67,7 @@ Other numpy ops: \cr
 \code{\link{op_ctc_decode}()} \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_diag}()} \cr
 \code{\link{op_diagflat}()} \cr
 \code{\link{op_diagonal}()} \cr
@@ -86,6 +92,9 @@ Other numpy ops: \cr
 \code{\link{op_get_item}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -95,6 +104,7 @@ Other numpy ops: \cr
 \code{\link{op_isfinite}()} \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
+\code{\link{op_kaiser}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
 \code{\link{op_less_equal}()} \cr
@@ -174,6 +184,8 @@ Other numpy ops: \cr
 \code{\link{op_var}()} \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_zeros}()} \cr
@@ -183,6 +195,7 @@ Other ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -201,6 +214,7 @@ Other ops: \cr
 \code{\link{op_associative_scan}()} \cr
 \code{\link{op_average}()} \cr
 \code{\link{op_average_pool}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_batch_normalization}()} \cr
 \code{\link{op_binary_crossentropy}()} \cr
 \code{\link{op_bincount}()} \cr
@@ -211,9 +225,11 @@ Other ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
 \code{\link{op_cast}()} \cr
 \code{\link{op_categorical_crossentropy}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_celu}()} \cr
 \code{\link{op_cholesky}()} \cr
@@ -226,6 +242,7 @@ Other ops: \cr
 \code{\link{op_convert_to_numpy}()} \cr
 \code{\link{op_convert_to_tensor}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -236,6 +253,7 @@ Other ops: \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
 \code{\link{op_custom_gradient}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_depthwise_conv}()} \cr
 \code{\link{op_det}()} \cr
 \code{\link{op_diag}()} \cr
@@ -275,10 +293,13 @@ Other ops: \cr
 \code{\link{op_glu}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
 \code{\link{op_hard_shrink}()} \cr
 \code{\link{op_hard_sigmoid}()} \cr
 \code{\link{op_hard_silu}()} \cr
 \code{\link{op_hard_tanh}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -305,6 +326,8 @@ Other ops: \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
 \code{\link{op_istft}()} \cr
+\code{\link{op_kaiser}()} \cr
+\code{\link{op_layer_normalization}()} \cr
 \code{\link{op_leaky_relu}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
@@ -403,6 +426,7 @@ Other ops: \cr
 \code{\link{op_sort}()} \cr
 \code{\link{op_sparse_categorical_crossentropy}()} \cr
 \code{\link{op_sparse_plus}()} \cr
+\code{\link{op_sparse_sigmoid}()} \cr
 \code{\link{op_sparsemax}()} \cr
 \code{\link{op_split}()} \cr
 \code{\link{op_sqrt}()} \cr
@@ -440,6 +464,8 @@ Other ops: \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
 \code{\link{op_vectorized_map}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_while_loop}()} \cr
diff --git a/man/op_sinh.Rd b/man/op_sinh.Rd
index b826d6d68..69c0c0709 100644
--- a/man/op_sinh.Rd
+++ b/man/op_sinh.Rd
@@ -24,6 +24,7 @@ Other numpy ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -40,6 +41,7 @@ Other numpy ops: \cr
 \code{\link{op_argsort}()} \cr
 \code{\link{op_array}()} \cr
 \code{\link{op_average}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_bincount}()} \cr
 \code{\link{op_bitwise_and}()} \cr
 \code{\link{op_bitwise_invert}()} \cr
@@ -48,12 +50,15 @@ Other numpy ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_clip}()} \cr
 \code{\link{op_concatenate}()} \cr
 \code{\link{op_conj}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -62,6 +67,7 @@ Other numpy ops: \cr
 \code{\link{op_ctc_decode}()} \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_diag}()} \cr
 \code{\link{op_diagflat}()} \cr
 \code{\link{op_diagonal}()} \cr
@@ -86,6 +92,9 @@ Other numpy ops: \cr
 \code{\link{op_get_item}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -95,6 +104,7 @@ Other numpy ops: \cr
 \code{\link{op_isfinite}()} \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
+\code{\link{op_kaiser}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
 \code{\link{op_less_equal}()} \cr
@@ -174,6 +184,8 @@ Other numpy ops: \cr
 \code{\link{op_var}()} \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_zeros}()} \cr
@@ -183,6 +195,7 @@ Other ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -201,6 +214,7 @@ Other ops: \cr
 \code{\link{op_associative_scan}()} \cr
 \code{\link{op_average}()} \cr
 \code{\link{op_average_pool}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_batch_normalization}()} \cr
 \code{\link{op_binary_crossentropy}()} \cr
 \code{\link{op_bincount}()} \cr
@@ -211,9 +225,11 @@ Other ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
 \code{\link{op_cast}()} \cr
 \code{\link{op_categorical_crossentropy}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_celu}()} \cr
 \code{\link{op_cholesky}()} \cr
@@ -226,6 +242,7 @@ Other ops: \cr
 \code{\link{op_convert_to_numpy}()} \cr
 \code{\link{op_convert_to_tensor}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -236,6 +253,7 @@ Other ops: \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
 \code{\link{op_custom_gradient}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_depthwise_conv}()} \cr
 \code{\link{op_det}()} \cr
 \code{\link{op_diag}()} \cr
@@ -275,10 +293,13 @@ Other ops: \cr
 \code{\link{op_glu}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
 \code{\link{op_hard_shrink}()} \cr
 \code{\link{op_hard_sigmoid}()} \cr
 \code{\link{op_hard_silu}()} \cr
 \code{\link{op_hard_tanh}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -305,6 +326,8 @@ Other ops: \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
 \code{\link{op_istft}()} \cr
+\code{\link{op_kaiser}()} \cr
+\code{\link{op_layer_normalization}()} \cr
 \code{\link{op_leaky_relu}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
@@ -403,6 +426,7 @@ Other ops: \cr
 \code{\link{op_sort}()} \cr
 \code{\link{op_sparse_categorical_crossentropy}()} \cr
 \code{\link{op_sparse_plus}()} \cr
+\code{\link{op_sparse_sigmoid}()} \cr
 \code{\link{op_sparsemax}()} \cr
 \code{\link{op_split}()} \cr
 \code{\link{op_sqrt}()} \cr
@@ -440,6 +464,8 @@ Other ops: \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
 \code{\link{op_vectorized_map}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_while_loop}()} \cr
diff --git a/man/op_size.Rd b/man/op_size.Rd
index a36604d69..0b37c4999 100644
--- a/man/op_size.Rd
+++ b/man/op_size.Rd
@@ -24,6 +24,7 @@ Other numpy ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -40,6 +41,7 @@ Other numpy ops: \cr
 \code{\link{op_argsort}()} \cr
 \code{\link{op_array}()} \cr
 \code{\link{op_average}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_bincount}()} \cr
 \code{\link{op_bitwise_and}()} \cr
 \code{\link{op_bitwise_invert}()} \cr
@@ -48,12 +50,15 @@ Other numpy ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_clip}()} \cr
 \code{\link{op_concatenate}()} \cr
 \code{\link{op_conj}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -62,6 +67,7 @@ Other numpy ops: \cr
 \code{\link{op_ctc_decode}()} \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_diag}()} \cr
 \code{\link{op_diagflat}()} \cr
 \code{\link{op_diagonal}()} \cr
@@ -86,6 +92,9 @@ Other numpy ops: \cr
 \code{\link{op_get_item}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -95,6 +104,7 @@ Other numpy ops: \cr
 \code{\link{op_isfinite}()} \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
+\code{\link{op_kaiser}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
 \code{\link{op_less_equal}()} \cr
@@ -174,6 +184,8 @@ Other numpy ops: \cr
 \code{\link{op_var}()} \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_zeros}()} \cr
@@ -183,6 +195,7 @@ Other ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -201,6 +214,7 @@ Other ops: \cr
 \code{\link{op_associative_scan}()} \cr
 \code{\link{op_average}()} \cr
 \code{\link{op_average_pool}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_batch_normalization}()} \cr
 \code{\link{op_binary_crossentropy}()} \cr
 \code{\link{op_bincount}()} \cr
@@ -211,9 +225,11 @@ Other ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
 \code{\link{op_cast}()} \cr
 \code{\link{op_categorical_crossentropy}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_celu}()} \cr
 \code{\link{op_cholesky}()} \cr
@@ -226,6 +242,7 @@ Other ops: \cr
 \code{\link{op_convert_to_numpy}()} \cr
 \code{\link{op_convert_to_tensor}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -236,6 +253,7 @@ Other ops: \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
 \code{\link{op_custom_gradient}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_depthwise_conv}()} \cr
 \code{\link{op_det}()} \cr
 \code{\link{op_diag}()} \cr
@@ -275,10 +293,13 @@ Other ops: \cr
 \code{\link{op_glu}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
 \code{\link{op_hard_shrink}()} \cr
 \code{\link{op_hard_sigmoid}()} \cr
 \code{\link{op_hard_silu}()} \cr
 \code{\link{op_hard_tanh}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -305,6 +326,8 @@ Other ops: \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
 \code{\link{op_istft}()} \cr
+\code{\link{op_kaiser}()} \cr
+\code{\link{op_layer_normalization}()} \cr
 \code{\link{op_leaky_relu}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
@@ -403,6 +426,7 @@ Other ops: \cr
 \code{\link{op_sort}()} \cr
 \code{\link{op_sparse_categorical_crossentropy}()} \cr
 \code{\link{op_sparse_plus}()} \cr
+\code{\link{op_sparse_sigmoid}()} \cr
 \code{\link{op_sparsemax}()} \cr
 \code{\link{op_split}()} \cr
 \code{\link{op_sqrt}()} \cr
@@ -440,6 +464,8 @@ Other ops: \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
 \code{\link{op_vectorized_map}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_while_loop}()} \cr
diff --git a/man/op_slice.Rd b/man/op_slice.Rd
index b5be4291d..9e0d80b7e 100644
--- a/man/op_slice.Rd
+++ b/man/op_slice.Rd
@@ -81,6 +81,7 @@ Other ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -99,6 +100,7 @@ Other ops: \cr
 \code{\link{op_associative_scan}()} \cr
 \code{\link{op_average}()} \cr
 \code{\link{op_average_pool}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_batch_normalization}()} \cr
 \code{\link{op_binary_crossentropy}()} \cr
 \code{\link{op_bincount}()} \cr
@@ -109,9 +111,11 @@ Other ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
 \code{\link{op_cast}()} \cr
 \code{\link{op_categorical_crossentropy}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_celu}()} \cr
 \code{\link{op_cholesky}()} \cr
@@ -124,6 +128,7 @@ Other ops: \cr
 \code{\link{op_convert_to_numpy}()} \cr
 \code{\link{op_convert_to_tensor}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -134,6 +139,7 @@ Other ops: \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
 \code{\link{op_custom_gradient}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_depthwise_conv}()} \cr
 \code{\link{op_det}()} \cr
 \code{\link{op_diag}()} \cr
@@ -173,10 +179,13 @@ Other ops: \cr
 \code{\link{op_glu}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
 \code{\link{op_hard_shrink}()} \cr
 \code{\link{op_hard_sigmoid}()} \cr
 \code{\link{op_hard_silu}()} \cr
 \code{\link{op_hard_tanh}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -203,6 +212,8 @@ Other ops: \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
 \code{\link{op_istft}()} \cr
+\code{\link{op_kaiser}()} \cr
+\code{\link{op_layer_normalization}()} \cr
 \code{\link{op_leaky_relu}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
@@ -301,6 +312,7 @@ Other ops: \cr
 \code{\link{op_sort}()} \cr
 \code{\link{op_sparse_categorical_crossentropy}()} \cr
 \code{\link{op_sparse_plus}()} \cr
+\code{\link{op_sparse_sigmoid}()} \cr
 \code{\link{op_sparsemax}()} \cr
 \code{\link{op_split}()} \cr
 \code{\link{op_sqrt}()} \cr
@@ -338,6 +350,8 @@ Other ops: \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
 \code{\link{op_vectorized_map}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_while_loop}()} \cr
diff --git a/man/op_slice_update.Rd b/man/op_slice_update.Rd
index 8af790ea2..4e74e0b41 100644
--- a/man/op_slice_update.Rd
+++ b/man/op_slice_update.Rd
@@ -78,6 +78,7 @@ Other ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -96,6 +97,7 @@ Other ops: \cr
 \code{\link{op_associative_scan}()} \cr
 \code{\link{op_average}()} \cr
 \code{\link{op_average_pool}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_batch_normalization}()} \cr
 \code{\link{op_binary_crossentropy}()} \cr
 \code{\link{op_bincount}()} \cr
@@ -106,9 +108,11 @@ Other ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
 \code{\link{op_cast}()} \cr
 \code{\link{op_categorical_crossentropy}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_celu}()} \cr
 \code{\link{op_cholesky}()} \cr
@@ -121,6 +125,7 @@ Other ops: \cr
 \code{\link{op_convert_to_numpy}()} \cr
 \code{\link{op_convert_to_tensor}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -131,6 +136,7 @@ Other ops: \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
 \code{\link{op_custom_gradient}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_depthwise_conv}()} \cr
 \code{\link{op_det}()} \cr
 \code{\link{op_diag}()} \cr
@@ -170,10 +176,13 @@ Other ops: \cr
 \code{\link{op_glu}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
 \code{\link{op_hard_shrink}()} \cr
 \code{\link{op_hard_sigmoid}()} \cr
 \code{\link{op_hard_silu}()} \cr
 \code{\link{op_hard_tanh}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -200,6 +209,8 @@ Other ops: \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
 \code{\link{op_istft}()} \cr
+\code{\link{op_kaiser}()} \cr
+\code{\link{op_layer_normalization}()} \cr
 \code{\link{op_leaky_relu}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
@@ -298,6 +309,7 @@ Other ops: \cr
 \code{\link{op_sort}()} \cr
 \code{\link{op_sparse_categorical_crossentropy}()} \cr
 \code{\link{op_sparse_plus}()} \cr
+\code{\link{op_sparse_sigmoid}()} \cr
 \code{\link{op_sparsemax}()} \cr
 \code{\link{op_split}()} \cr
 \code{\link{op_sqrt}()} \cr
@@ -335,6 +347,8 @@ Other ops: \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
 \code{\link{op_vectorized_map}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_while_loop}()} \cr
diff --git a/man/op_slogdet.Rd b/man/op_slogdet.Rd
index 684904eb6..291417d05 100644
--- a/man/op_slogdet.Rd
+++ b/man/op_slogdet.Rd
@@ -36,6 +36,7 @@ Other ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -54,6 +55,7 @@ Other ops: \cr
 \code{\link{op_associative_scan}()} \cr
 \code{\link{op_average}()} \cr
 \code{\link{op_average_pool}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_batch_normalization}()} \cr
 \code{\link{op_binary_crossentropy}()} \cr
 \code{\link{op_bincount}()} \cr
@@ -64,9 +66,11 @@ Other ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
 \code{\link{op_cast}()} \cr
 \code{\link{op_categorical_crossentropy}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_celu}()} \cr
 \code{\link{op_cholesky}()} \cr
@@ -79,6 +83,7 @@ Other ops: \cr
 \code{\link{op_convert_to_numpy}()} \cr
 \code{\link{op_convert_to_tensor}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -89,6 +94,7 @@ Other ops: \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
 \code{\link{op_custom_gradient}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_depthwise_conv}()} \cr
 \code{\link{op_det}()} \cr
 \code{\link{op_diag}()} \cr
@@ -128,10 +134,13 @@ Other ops: \cr
 \code{\link{op_glu}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
 \code{\link{op_hard_shrink}()} \cr
 \code{\link{op_hard_sigmoid}()} \cr
 \code{\link{op_hard_silu}()} \cr
 \code{\link{op_hard_tanh}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -158,6 +167,8 @@ Other ops: \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
 \code{\link{op_istft}()} \cr
+\code{\link{op_kaiser}()} \cr
+\code{\link{op_layer_normalization}()} \cr
 \code{\link{op_leaky_relu}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
@@ -256,6 +267,7 @@ Other ops: \cr
 \code{\link{op_sort}()} \cr
 \code{\link{op_sparse_categorical_crossentropy}()} \cr
 \code{\link{op_sparse_plus}()} \cr
+\code{\link{op_sparse_sigmoid}()} \cr
 \code{\link{op_sparsemax}()} \cr
 \code{\link{op_split}()} \cr
 \code{\link{op_sqrt}()} \cr
@@ -293,6 +305,8 @@ Other ops: \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
 \code{\link{op_vectorized_map}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_while_loop}()} \cr
diff --git a/man/op_soft_shrink.Rd b/man/op_soft_shrink.Rd
index 2aced2a32..bb2cdebb8 100644
--- a/man/op_soft_shrink.Rd
+++ b/man/op_soft_shrink.Rd
@@ -50,6 +50,7 @@ Other nn ops: \cr
 \code{\link{op_hard_sigmoid}()} \cr
 \code{\link{op_hard_silu}()} \cr
 \code{\link{op_hard_tanh}()} \cr
+\code{\link{op_layer_normalization}()} \cr
 \code{\link{op_leaky_relu}()} \cr
 \code{\link{op_log_sigmoid}()} \cr
 \code{\link{op_log_softmax}()} \cr
@@ -72,6 +73,7 @@ Other nn ops: \cr
 \code{\link{op_softsign}()} \cr
 \code{\link{op_sparse_categorical_crossentropy}()} \cr
 \code{\link{op_sparse_plus}()} \cr
+\code{\link{op_sparse_sigmoid}()} \cr
 \code{\link{op_sparsemax}()} \cr
 \code{\link{op_squareplus}()} \cr
 \code{\link{op_tanh_shrink}()} \cr
@@ -82,6 +84,7 @@ Other ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -100,6 +103,7 @@ Other ops: \cr
 \code{\link{op_associative_scan}()} \cr
 \code{\link{op_average}()} \cr
 \code{\link{op_average_pool}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_batch_normalization}()} \cr
 \code{\link{op_binary_crossentropy}()} \cr
 \code{\link{op_bincount}()} \cr
@@ -110,9 +114,11 @@ Other ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
 \code{\link{op_cast}()} \cr
 \code{\link{op_categorical_crossentropy}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_celu}()} \cr
 \code{\link{op_cholesky}()} \cr
@@ -125,6 +131,7 @@ Other ops: \cr
 \code{\link{op_convert_to_numpy}()} \cr
 \code{\link{op_convert_to_tensor}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -135,6 +142,7 @@ Other ops: \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
 \code{\link{op_custom_gradient}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_depthwise_conv}()} \cr
 \code{\link{op_det}()} \cr
 \code{\link{op_diag}()} \cr
@@ -174,10 +182,13 @@ Other ops: \cr
 \code{\link{op_glu}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
 \code{\link{op_hard_shrink}()} \cr
 \code{\link{op_hard_sigmoid}()} \cr
 \code{\link{op_hard_silu}()} \cr
 \code{\link{op_hard_tanh}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -204,6 +215,8 @@ Other ops: \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
 \code{\link{op_istft}()} \cr
+\code{\link{op_kaiser}()} \cr
+\code{\link{op_layer_normalization}()} \cr
 \code{\link{op_leaky_relu}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
@@ -302,6 +315,7 @@ Other ops: \cr
 \code{\link{op_sort}()} \cr
 \code{\link{op_sparse_categorical_crossentropy}()} \cr
 \code{\link{op_sparse_plus}()} \cr
+\code{\link{op_sparse_sigmoid}()} \cr
 \code{\link{op_sparsemax}()} \cr
 \code{\link{op_split}()} \cr
 \code{\link{op_sqrt}()} \cr
@@ -339,6 +353,8 @@ Other ops: \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
 \code{\link{op_vectorized_map}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_while_loop}()} \cr
diff --git a/man/op_softmax.Rd b/man/op_softmax.Rd
index 6cda944cc..44b810d2a 100644
--- a/man/op_softmax.Rd
+++ b/man/op_softmax.Rd
@@ -57,6 +57,7 @@ Other nn ops: \cr
 \code{\link{op_hard_sigmoid}()} \cr
 \code{\link{op_hard_silu}()} \cr
 \code{\link{op_hard_tanh}()} \cr
+\code{\link{op_layer_normalization}()} \cr
 \code{\link{op_leaky_relu}()} \cr
 \code{\link{op_log_sigmoid}()} \cr
 \code{\link{op_log_softmax}()} \cr
@@ -79,6 +80,7 @@ Other nn ops: \cr
 \code{\link{op_softsign}()} \cr
 \code{\link{op_sparse_categorical_crossentropy}()} \cr
 \code{\link{op_sparse_plus}()} \cr
+\code{\link{op_sparse_sigmoid}()} \cr
 \code{\link{op_sparsemax}()} \cr
 \code{\link{op_squareplus}()} \cr
 \code{\link{op_tanh_shrink}()} \cr
@@ -89,6 +91,7 @@ Other ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -107,6 +110,7 @@ Other ops: \cr
 \code{\link{op_associative_scan}()} \cr
 \code{\link{op_average}()} \cr
 \code{\link{op_average_pool}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_batch_normalization}()} \cr
 \code{\link{op_binary_crossentropy}()} \cr
 \code{\link{op_bincount}()} \cr
@@ -117,9 +121,11 @@ Other ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
 \code{\link{op_cast}()} \cr
 \code{\link{op_categorical_crossentropy}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_celu}()} \cr
 \code{\link{op_cholesky}()} \cr
@@ -132,6 +138,7 @@ Other ops: \cr
 \code{\link{op_convert_to_numpy}()} \cr
 \code{\link{op_convert_to_tensor}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -142,6 +149,7 @@ Other ops: \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
 \code{\link{op_custom_gradient}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_depthwise_conv}()} \cr
 \code{\link{op_det}()} \cr
 \code{\link{op_diag}()} \cr
@@ -181,10 +189,13 @@ Other ops: \cr
 \code{\link{op_glu}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
 \code{\link{op_hard_shrink}()} \cr
 \code{\link{op_hard_sigmoid}()} \cr
 \code{\link{op_hard_silu}()} \cr
 \code{\link{op_hard_tanh}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -211,6 +222,8 @@ Other ops: \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
 \code{\link{op_istft}()} \cr
+\code{\link{op_kaiser}()} \cr
+\code{\link{op_layer_normalization}()} \cr
 \code{\link{op_leaky_relu}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
@@ -309,6 +322,7 @@ Other ops: \cr
 \code{\link{op_sort}()} \cr
 \code{\link{op_sparse_categorical_crossentropy}()} \cr
 \code{\link{op_sparse_plus}()} \cr
+\code{\link{op_sparse_sigmoid}()} \cr
 \code{\link{op_sparsemax}()} \cr
 \code{\link{op_split}()} \cr
 \code{\link{op_sqrt}()} \cr
@@ -346,6 +360,8 @@ Other ops: \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
 \code{\link{op_vectorized_map}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_while_loop}()} \cr
diff --git a/man/op_softplus.Rd b/man/op_softplus.Rd
index fbe39ea84..a0a3acab2 100644
--- a/man/op_softplus.Rd
+++ b/man/op_softplus.Rd
@@ -55,6 +55,7 @@ Other nn ops: \cr
 \code{\link{op_hard_sigmoid}()} \cr
 \code{\link{op_hard_silu}()} \cr
 \code{\link{op_hard_tanh}()} \cr
+\code{\link{op_layer_normalization}()} \cr
 \code{\link{op_leaky_relu}()} \cr
 \code{\link{op_log_sigmoid}()} \cr
 \code{\link{op_log_softmax}()} \cr
@@ -77,6 +78,7 @@ Other nn ops: \cr
 \code{\link{op_softsign}()} \cr
 \code{\link{op_sparse_categorical_crossentropy}()} \cr
 \code{\link{op_sparse_plus}()} \cr
+\code{\link{op_sparse_sigmoid}()} \cr
 \code{\link{op_sparsemax}()} \cr
 \code{\link{op_squareplus}()} \cr
 \code{\link{op_tanh_shrink}()} \cr
@@ -87,6 +89,7 @@ Other ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -105,6 +108,7 @@ Other ops: \cr
 \code{\link{op_associative_scan}()} \cr
 \code{\link{op_average}()} \cr
 \code{\link{op_average_pool}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_batch_normalization}()} \cr
 \code{\link{op_binary_crossentropy}()} \cr
 \code{\link{op_bincount}()} \cr
@@ -115,9 +119,11 @@ Other ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
 \code{\link{op_cast}()} \cr
 \code{\link{op_categorical_crossentropy}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_celu}()} \cr
 \code{\link{op_cholesky}()} \cr
@@ -130,6 +136,7 @@ Other ops: \cr
 \code{\link{op_convert_to_numpy}()} \cr
 \code{\link{op_convert_to_tensor}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -140,6 +147,7 @@ Other ops: \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
 \code{\link{op_custom_gradient}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_depthwise_conv}()} \cr
 \code{\link{op_det}()} \cr
 \code{\link{op_diag}()} \cr
@@ -179,10 +187,13 @@ Other ops: \cr
 \code{\link{op_glu}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
 \code{\link{op_hard_shrink}()} \cr
 \code{\link{op_hard_sigmoid}()} \cr
 \code{\link{op_hard_silu}()} \cr
 \code{\link{op_hard_tanh}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -209,6 +220,8 @@ Other ops: \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
 \code{\link{op_istft}()} \cr
+\code{\link{op_kaiser}()} \cr
+\code{\link{op_layer_normalization}()} \cr
 \code{\link{op_leaky_relu}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
@@ -307,6 +320,7 @@ Other ops: \cr
 \code{\link{op_sort}()} \cr
 \code{\link{op_sparse_categorical_crossentropy}()} \cr
 \code{\link{op_sparse_plus}()} \cr
+\code{\link{op_sparse_sigmoid}()} \cr
 \code{\link{op_sparsemax}()} \cr
 \code{\link{op_split}()} \cr
 \code{\link{op_sqrt}()} \cr
@@ -344,6 +358,8 @@ Other ops: \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
 \code{\link{op_vectorized_map}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_while_loop}()} \cr
diff --git a/man/op_softsign.Rd b/man/op_softsign.Rd
index acbf6e00c..49d63c5f4 100644
--- a/man/op_softsign.Rd
+++ b/man/op_softsign.Rd
@@ -54,6 +54,7 @@ Other nn ops: \cr
 \code{\link{op_hard_sigmoid}()} \cr
 \code{\link{op_hard_silu}()} \cr
 \code{\link{op_hard_tanh}()} \cr
+\code{\link{op_layer_normalization}()} \cr
 \code{\link{op_leaky_relu}()} \cr
 \code{\link{op_log_sigmoid}()} \cr
 \code{\link{op_log_softmax}()} \cr
@@ -76,6 +77,7 @@ Other nn ops: \cr
 \code{\link{op_softplus}()} \cr
 \code{\link{op_sparse_categorical_crossentropy}()} \cr
 \code{\link{op_sparse_plus}()} \cr
+\code{\link{op_sparse_sigmoid}()} \cr
 \code{\link{op_sparsemax}()} \cr
 \code{\link{op_squareplus}()} \cr
 \code{\link{op_tanh_shrink}()} \cr
@@ -86,6 +88,7 @@ Other ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -104,6 +107,7 @@ Other ops: \cr
 \code{\link{op_associative_scan}()} \cr
 \code{\link{op_average}()} \cr
 \code{\link{op_average_pool}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_batch_normalization}()} \cr
 \code{\link{op_binary_crossentropy}()} \cr
 \code{\link{op_bincount}()} \cr
@@ -114,9 +118,11 @@ Other ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
 \code{\link{op_cast}()} \cr
 \code{\link{op_categorical_crossentropy}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_celu}()} \cr
 \code{\link{op_cholesky}()} \cr
@@ -129,6 +135,7 @@ Other ops: \cr
 \code{\link{op_convert_to_numpy}()} \cr
 \code{\link{op_convert_to_tensor}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -139,6 +146,7 @@ Other ops: \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
 \code{\link{op_custom_gradient}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_depthwise_conv}()} \cr
 \code{\link{op_det}()} \cr
 \code{\link{op_diag}()} \cr
@@ -178,10 +186,13 @@ Other ops: \cr
 \code{\link{op_glu}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
 \code{\link{op_hard_shrink}()} \cr
 \code{\link{op_hard_sigmoid}()} \cr
 \code{\link{op_hard_silu}()} \cr
 \code{\link{op_hard_tanh}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -208,6 +219,8 @@ Other ops: \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
 \code{\link{op_istft}()} \cr
+\code{\link{op_kaiser}()} \cr
+\code{\link{op_layer_normalization}()} \cr
 \code{\link{op_leaky_relu}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
@@ -306,6 +319,7 @@ Other ops: \cr
 \code{\link{op_sort}()} \cr
 \code{\link{op_sparse_categorical_crossentropy}()} \cr
 \code{\link{op_sparse_plus}()} \cr
+\code{\link{op_sparse_sigmoid}()} \cr
 \code{\link{op_sparsemax}()} \cr
 \code{\link{op_split}()} \cr
 \code{\link{op_sqrt}()} \cr
@@ -343,6 +357,8 @@ Other ops: \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
 \code{\link{op_vectorized_map}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_while_loop}()} \cr
diff --git a/man/op_solve.Rd b/man/op_solve.Rd
index 2edc021c7..c5160cc8e 100644
--- a/man/op_solve.Rd
+++ b/man/op_solve.Rd
@@ -56,6 +56,7 @@ Other ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -74,6 +75,7 @@ Other ops: \cr
 \code{\link{op_associative_scan}()} \cr
 \code{\link{op_average}()} \cr
 \code{\link{op_average_pool}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_batch_normalization}()} \cr
 \code{\link{op_binary_crossentropy}()} \cr
 \code{\link{op_bincount}()} \cr
@@ -84,9 +86,11 @@ Other ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
 \code{\link{op_cast}()} \cr
 \code{\link{op_categorical_crossentropy}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_celu}()} \cr
 \code{\link{op_cholesky}()} \cr
@@ -99,6 +103,7 @@ Other ops: \cr
 \code{\link{op_convert_to_numpy}()} \cr
 \code{\link{op_convert_to_tensor}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -109,6 +114,7 @@ Other ops: \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
 \code{\link{op_custom_gradient}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_depthwise_conv}()} \cr
 \code{\link{op_det}()} \cr
 \code{\link{op_diag}()} \cr
@@ -148,10 +154,13 @@ Other ops: \cr
 \code{\link{op_glu}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
 \code{\link{op_hard_shrink}()} \cr
 \code{\link{op_hard_sigmoid}()} \cr
 \code{\link{op_hard_silu}()} \cr
 \code{\link{op_hard_tanh}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -178,6 +187,8 @@ Other ops: \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
 \code{\link{op_istft}()} \cr
+\code{\link{op_kaiser}()} \cr
+\code{\link{op_layer_normalization}()} \cr
 \code{\link{op_leaky_relu}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
@@ -276,6 +287,7 @@ Other ops: \cr
 \code{\link{op_sort}()} \cr
 \code{\link{op_sparse_categorical_crossentropy}()} \cr
 \code{\link{op_sparse_plus}()} \cr
+\code{\link{op_sparse_sigmoid}()} \cr
 \code{\link{op_sparsemax}()} \cr
 \code{\link{op_split}()} \cr
 \code{\link{op_sqrt}()} \cr
@@ -313,6 +325,8 @@ Other ops: \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
 \code{\link{op_vectorized_map}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_while_loop}()} \cr
diff --git a/man/op_solve_triangular.Rd b/man/op_solve_triangular.Rd
index 0c420b7ce..00a7976fa 100644
--- a/man/op_solve_triangular.Rd
+++ b/man/op_solve_triangular.Rd
@@ -39,6 +39,7 @@ Other ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -57,6 +58,7 @@ Other ops: \cr
 \code{\link{op_associative_scan}()} \cr
 \code{\link{op_average}()} \cr
 \code{\link{op_average_pool}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_batch_normalization}()} \cr
 \code{\link{op_binary_crossentropy}()} \cr
 \code{\link{op_bincount}()} \cr
@@ -67,9 +69,11 @@ Other ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
 \code{\link{op_cast}()} \cr
 \code{\link{op_categorical_crossentropy}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_celu}()} \cr
 \code{\link{op_cholesky}()} \cr
@@ -82,6 +86,7 @@ Other ops: \cr
 \code{\link{op_convert_to_numpy}()} \cr
 \code{\link{op_convert_to_tensor}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -92,6 +97,7 @@ Other ops: \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
 \code{\link{op_custom_gradient}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_depthwise_conv}()} \cr
 \code{\link{op_det}()} \cr
 \code{\link{op_diag}()} \cr
@@ -131,10 +137,13 @@ Other ops: \cr
 \code{\link{op_glu}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
 \code{\link{op_hard_shrink}()} \cr
 \code{\link{op_hard_sigmoid}()} \cr
 \code{\link{op_hard_silu}()} \cr
 \code{\link{op_hard_tanh}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -161,6 +170,8 @@ Other ops: \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
 \code{\link{op_istft}()} \cr
+\code{\link{op_kaiser}()} \cr
+\code{\link{op_layer_normalization}()} \cr
 \code{\link{op_leaky_relu}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
@@ -259,6 +270,7 @@ Other ops: \cr
 \code{\link{op_sort}()} \cr
 \code{\link{op_sparse_categorical_crossentropy}()} \cr
 \code{\link{op_sparse_plus}()} \cr
+\code{\link{op_sparse_sigmoid}()} \cr
 \code{\link{op_sparsemax}()} \cr
 \code{\link{op_split}()} \cr
 \code{\link{op_sqrt}()} \cr
@@ -296,6 +308,8 @@ Other ops: \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
 \code{\link{op_vectorized_map}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_while_loop}()} \cr
diff --git a/man/op_sort.Rd b/man/op_sort.Rd
index 4d873b9a6..15b9e954d 100644
--- a/man/op_sort.Rd
+++ b/man/op_sort.Rd
@@ -27,6 +27,7 @@ Other numpy ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -43,6 +44,7 @@ Other numpy ops: \cr
 \code{\link{op_argsort}()} \cr
 \code{\link{op_array}()} \cr
 \code{\link{op_average}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_bincount}()} \cr
 \code{\link{op_bitwise_and}()} \cr
 \code{\link{op_bitwise_invert}()} \cr
@@ -51,12 +53,15 @@ Other numpy ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_clip}()} \cr
 \code{\link{op_concatenate}()} \cr
 \code{\link{op_conj}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -65,6 +70,7 @@ Other numpy ops: \cr
 \code{\link{op_ctc_decode}()} \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_diag}()} \cr
 \code{\link{op_diagflat}()} \cr
 \code{\link{op_diagonal}()} \cr
@@ -89,6 +95,9 @@ Other numpy ops: \cr
 \code{\link{op_get_item}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -98,6 +107,7 @@ Other numpy ops: \cr
 \code{\link{op_isfinite}()} \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
+\code{\link{op_kaiser}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
 \code{\link{op_less_equal}()} \cr
@@ -177,6 +187,8 @@ Other numpy ops: \cr
 \code{\link{op_var}()} \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_zeros}()} \cr
@@ -186,6 +198,7 @@ Other ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -204,6 +217,7 @@ Other ops: \cr
 \code{\link{op_associative_scan}()} \cr
 \code{\link{op_average}()} \cr
 \code{\link{op_average_pool}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_batch_normalization}()} \cr
 \code{\link{op_binary_crossentropy}()} \cr
 \code{\link{op_bincount}()} \cr
@@ -214,9 +228,11 @@ Other ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
 \code{\link{op_cast}()} \cr
 \code{\link{op_categorical_crossentropy}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_celu}()} \cr
 \code{\link{op_cholesky}()} \cr
@@ -229,6 +245,7 @@ Other ops: \cr
 \code{\link{op_convert_to_numpy}()} \cr
 \code{\link{op_convert_to_tensor}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -239,6 +256,7 @@ Other ops: \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
 \code{\link{op_custom_gradient}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_depthwise_conv}()} \cr
 \code{\link{op_det}()} \cr
 \code{\link{op_diag}()} \cr
@@ -278,10 +296,13 @@ Other ops: \cr
 \code{\link{op_glu}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
 \code{\link{op_hard_shrink}()} \cr
 \code{\link{op_hard_sigmoid}()} \cr
 \code{\link{op_hard_silu}()} \cr
 \code{\link{op_hard_tanh}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -308,6 +329,8 @@ Other ops: \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
 \code{\link{op_istft}()} \cr
+\code{\link{op_kaiser}()} \cr
+\code{\link{op_layer_normalization}()} \cr
 \code{\link{op_leaky_relu}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
@@ -406,6 +429,7 @@ Other ops: \cr
 \code{\link{op_solve_triangular}()} \cr
 \code{\link{op_sparse_categorical_crossentropy}()} \cr
 \code{\link{op_sparse_plus}()} \cr
+\code{\link{op_sparse_sigmoid}()} \cr
 \code{\link{op_sparsemax}()} \cr
 \code{\link{op_split}()} \cr
 \code{\link{op_sqrt}()} \cr
@@ -443,6 +467,8 @@ Other ops: \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
 \code{\link{op_vectorized_map}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_while_loop}()} \cr
diff --git a/man/op_sparse_categorical_crossentropy.Rd b/man/op_sparse_categorical_crossentropy.Rd
index aa90e042b..0500c5927 100644
--- a/man/op_sparse_categorical_crossentropy.Rd
+++ b/man/op_sparse_categorical_crossentropy.Rd
@@ -78,6 +78,7 @@ Other nn ops: \cr
 \code{\link{op_hard_sigmoid}()} \cr
 \code{\link{op_hard_silu}()} \cr
 \code{\link{op_hard_tanh}()} \cr
+\code{\link{op_layer_normalization}()} \cr
 \code{\link{op_leaky_relu}()} \cr
 \code{\link{op_log_sigmoid}()} \cr
 \code{\link{op_log_softmax}()} \cr
@@ -100,6 +101,7 @@ Other nn ops: \cr
 \code{\link{op_softplus}()} \cr
 \code{\link{op_softsign}()} \cr
 \code{\link{op_sparse_plus}()} \cr
+\code{\link{op_sparse_sigmoid}()} \cr
 \code{\link{op_sparsemax}()} \cr
 \code{\link{op_squareplus}()} \cr
 \code{\link{op_tanh_shrink}()} \cr
@@ -110,6 +112,7 @@ Other ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -128,6 +131,7 @@ Other ops: \cr
 \code{\link{op_associative_scan}()} \cr
 \code{\link{op_average}()} \cr
 \code{\link{op_average_pool}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_batch_normalization}()} \cr
 \code{\link{op_binary_crossentropy}()} \cr
 \code{\link{op_bincount}()} \cr
@@ -138,9 +142,11 @@ Other ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
 \code{\link{op_cast}()} \cr
 \code{\link{op_categorical_crossentropy}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_celu}()} \cr
 \code{\link{op_cholesky}()} \cr
@@ -153,6 +159,7 @@ Other ops: \cr
 \code{\link{op_convert_to_numpy}()} \cr
 \code{\link{op_convert_to_tensor}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -163,6 +170,7 @@ Other ops: \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
 \code{\link{op_custom_gradient}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_depthwise_conv}()} \cr
 \code{\link{op_det}()} \cr
 \code{\link{op_diag}()} \cr
@@ -202,10 +210,13 @@ Other ops: \cr
 \code{\link{op_glu}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
 \code{\link{op_hard_shrink}()} \cr
 \code{\link{op_hard_sigmoid}()} \cr
 \code{\link{op_hard_silu}()} \cr
 \code{\link{op_hard_tanh}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -232,6 +243,8 @@ Other ops: \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
 \code{\link{op_istft}()} \cr
+\code{\link{op_kaiser}()} \cr
+\code{\link{op_layer_normalization}()} \cr
 \code{\link{op_leaky_relu}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
@@ -330,6 +343,7 @@ Other ops: \cr
 \code{\link{op_solve_triangular}()} \cr
 \code{\link{op_sort}()} \cr
 \code{\link{op_sparse_plus}()} \cr
+\code{\link{op_sparse_sigmoid}()} \cr
 \code{\link{op_sparsemax}()} \cr
 \code{\link{op_split}()} \cr
 \code{\link{op_sqrt}()} \cr
@@ -367,6 +381,8 @@ Other ops: \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
 \code{\link{op_vectorized_map}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_while_loop}()} \cr
diff --git a/man/op_sparse_plus.Rd b/man/op_sparse_plus.Rd
index 82f6b7067..31a1f1e9d 100644
--- a/man/op_sparse_plus.Rd
+++ b/man/op_sparse_plus.Rd
@@ -48,6 +48,7 @@ Other nn ops: \cr
 \code{\link{op_hard_sigmoid}()} \cr
 \code{\link{op_hard_silu}()} \cr
 \code{\link{op_hard_tanh}()} \cr
+\code{\link{op_layer_normalization}()} \cr
 \code{\link{op_leaky_relu}()} \cr
 \code{\link{op_log_sigmoid}()} \cr
 \code{\link{op_log_softmax}()} \cr
@@ -70,6 +71,7 @@ Other nn ops: \cr
 \code{\link{op_softplus}()} \cr
 \code{\link{op_softsign}()} \cr
 \code{\link{op_sparse_categorical_crossentropy}()} \cr
+\code{\link{op_sparse_sigmoid}()} \cr
 \code{\link{op_sparsemax}()} \cr
 \code{\link{op_squareplus}()} \cr
 \code{\link{op_tanh_shrink}()} \cr
@@ -80,6 +82,7 @@ Other ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -98,6 +101,7 @@ Other ops: \cr
 \code{\link{op_associative_scan}()} \cr
 \code{\link{op_average}()} \cr
 \code{\link{op_average_pool}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_batch_normalization}()} \cr
 \code{\link{op_binary_crossentropy}()} \cr
 \code{\link{op_bincount}()} \cr
@@ -108,9 +112,11 @@ Other ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
 \code{\link{op_cast}()} \cr
 \code{\link{op_categorical_crossentropy}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_celu}()} \cr
 \code{\link{op_cholesky}()} \cr
@@ -123,6 +129,7 @@ Other ops: \cr
 \code{\link{op_convert_to_numpy}()} \cr
 \code{\link{op_convert_to_tensor}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -133,6 +140,7 @@ Other ops: \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
 \code{\link{op_custom_gradient}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_depthwise_conv}()} \cr
 \code{\link{op_det}()} \cr
 \code{\link{op_diag}()} \cr
@@ -172,10 +180,13 @@ Other ops: \cr
 \code{\link{op_glu}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
 \code{\link{op_hard_shrink}()} \cr
 \code{\link{op_hard_sigmoid}()} \cr
 \code{\link{op_hard_silu}()} \cr
 \code{\link{op_hard_tanh}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -202,6 +213,8 @@ Other ops: \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
 \code{\link{op_istft}()} \cr
+\code{\link{op_kaiser}()} \cr
+\code{\link{op_layer_normalization}()} \cr
 \code{\link{op_leaky_relu}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
@@ -300,6 +313,7 @@ Other ops: \cr
 \code{\link{op_solve_triangular}()} \cr
 \code{\link{op_sort}()} \cr
 \code{\link{op_sparse_categorical_crossentropy}()} \cr
+\code{\link{op_sparse_sigmoid}()} \cr
 \code{\link{op_sparsemax}()} \cr
 \code{\link{op_split}()} \cr
 \code{\link{op_sqrt}()} \cr
@@ -337,6 +351,8 @@ Other ops: \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
 \code{\link{op_vectorized_map}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_while_loop}()} \cr
diff --git a/man/op_sparse_sigmoid.Rd b/man/op_sparse_sigmoid.Rd
new file mode 100644
index 000000000..e209ecaf8
--- /dev/null
+++ b/man/op_sparse_sigmoid.Rd
@@ -0,0 +1,363 @@
+% Generated by roxygen2: do not edit by hand
+% Please edit documentation in R/ops-nn.R
+\name{op_sparse_sigmoid}
+\alias{op_sparse_sigmoid}
+\title{Sparse sigmoid activation function.}
+\usage{
+op_sparse_sigmoid(x)
+}
+\arguments{
+\item{x}{Input tensor.}
+}
+\value{
+A tensor with the same shape as \code{x}.
+}
+\description{
+It is defined as
+
+\code{f(x) = 0} for \code{x <= -1},
+\code{f(x) = 0.5 * (x + 1)} for \verb{-1 < x < 1},
+\code{f(x) = 1} for \code{x >= 1}.
+}
+\section{Examples}{
+\if{html}{\out{<div class="sourceCode r">}}\preformatted{x <- op_array(c(-1.0, 0.0, 1.0))
+op_sparse_sigmoid(x)
+}\if{html}{\out{</div>}}
+
+\if{html}{\out{<div class="sourceCode">}}\preformatted{## tf.Tensor([0.  0.5 1. ], shape=(3), dtype=float32)
+
+}\if{html}{\out{</div>}}
+}
+
+\seealso{
+Other nn ops: \cr
+\code{\link{op_average_pool}()} \cr
+\code{\link{op_batch_normalization}()} \cr
+\code{\link{op_binary_crossentropy}()} \cr
+\code{\link{op_categorical_crossentropy}()} \cr
+\code{\link{op_celu}()} \cr
+\code{\link{op_conv}()} \cr
+\code{\link{op_conv_transpose}()} \cr
+\code{\link{op_ctc_loss}()} \cr
+\code{\link{op_depthwise_conv}()} \cr
+\code{\link{op_dot_product_attention}()} \cr
+\code{\link{op_elu}()} \cr
+\code{\link{op_gelu}()} \cr
+\code{\link{op_glu}()} \cr
+\code{\link{op_hard_shrink}()} \cr
+\code{\link{op_hard_sigmoid}()} \cr
+\code{\link{op_hard_silu}()} \cr
+\code{\link{op_hard_tanh}()} \cr
+\code{\link{op_layer_normalization}()} \cr
+\code{\link{op_leaky_relu}()} \cr
+\code{\link{op_log_sigmoid}()} \cr
+\code{\link{op_log_softmax}()} \cr
+\code{\link{op_max_pool}()} \cr
+\code{\link{op_moments}()} \cr
+\code{\link{op_multi_hot}()} \cr
+\code{\link{op_normalize}()} \cr
+\code{\link{op_one_hot}()} \cr
+\code{\link{op_polar}()} \cr
+\code{\link{op_psnr}()} \cr
+\code{\link{op_relu}()} \cr
+\code{\link{op_relu6}()} \cr
+\code{\link{op_rms_normalization}()} \cr
+\code{\link{op_selu}()} \cr
+\code{\link{op_separable_conv}()} \cr
+\code{\link{op_sigmoid}()} \cr
+\code{\link{op_silu}()} \cr
+\code{\link{op_soft_shrink}()} \cr
+\code{\link{op_softmax}()} \cr
+\code{\link{op_softplus}()} \cr
+\code{\link{op_softsign}()} \cr
+\code{\link{op_sparse_categorical_crossentropy}()} \cr
+\code{\link{op_sparse_plus}()} \cr
+\code{\link{op_sparsemax}()} \cr
+\code{\link{op_squareplus}()} \cr
+\code{\link{op_tanh_shrink}()} \cr
+\code{\link{op_threshold}()} \cr
+\code{\link{op_unravel_index}()} \cr
+
+Other ops: \cr
+\code{\link{op_abs}()} \cr
+\code{\link{op_add}()} \cr
+\code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
+\code{\link{op_any}()} \cr
+\code{\link{op_append}()} \cr
+\code{\link{op_arange}()} \cr
+\code{\link{op_arccos}()} \cr
+\code{\link{op_arccosh}()} \cr
+\code{\link{op_arcsin}()} \cr
+\code{\link{op_arcsinh}()} \cr
+\code{\link{op_arctan}()} \cr
+\code{\link{op_arctan2}()} \cr
+\code{\link{op_arctanh}()} \cr
+\code{\link{op_argmax}()} \cr
+\code{\link{op_argmin}()} \cr
+\code{\link{op_argpartition}()} \cr
+\code{\link{op_argsort}()} \cr
+\code{\link{op_array}()} \cr
+\code{\link{op_associative_scan}()} \cr
+\code{\link{op_average}()} \cr
+\code{\link{op_average_pool}()} \cr
+\code{\link{op_bartlett}()} \cr
+\code{\link{op_batch_normalization}()} \cr
+\code{\link{op_binary_crossentropy}()} \cr
+\code{\link{op_bincount}()} \cr
+\code{\link{op_bitwise_and}()} \cr
+\code{\link{op_bitwise_invert}()} \cr
+\code{\link{op_bitwise_left_shift}()} \cr
+\code{\link{op_bitwise_not}()} \cr
+\code{\link{op_bitwise_or}()} \cr
+\code{\link{op_bitwise_right_shift}()} \cr
+\code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
+\code{\link{op_broadcast_to}()} \cr
+\code{\link{op_cast}()} \cr
+\code{\link{op_categorical_crossentropy}()} \cr
+\code{\link{op_cbrt}()} \cr
+\code{\link{op_ceil}()} \cr
+\code{\link{op_celu}()} \cr
+\code{\link{op_cholesky}()} \cr
+\code{\link{op_clip}()} \cr
+\code{\link{op_concatenate}()} \cr
+\code{\link{op_cond}()} \cr
+\code{\link{op_conj}()} \cr
+\code{\link{op_conv}()} \cr
+\code{\link{op_conv_transpose}()} \cr
+\code{\link{op_convert_to_numpy}()} \cr
+\code{\link{op_convert_to_tensor}()} \cr
+\code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
+\code{\link{op_correlate}()} \cr
+\code{\link{op_cos}()} \cr
+\code{\link{op_cosh}()} \cr
+\code{\link{op_count_nonzero}()} \cr
+\code{\link{op_cross}()} \cr
+\code{\link{op_ctc_decode}()} \cr
+\code{\link{op_ctc_loss}()} \cr
+\code{\link{op_cumprod}()} \cr
+\code{\link{op_cumsum}()} \cr
+\code{\link{op_custom_gradient}()} \cr
+\code{\link{op_deg2rad}()} \cr
+\code{\link{op_depthwise_conv}()} \cr
+\code{\link{op_det}()} \cr
+\code{\link{op_diag}()} \cr
+\code{\link{op_diagflat}()} \cr
+\code{\link{op_diagonal}()} \cr
+\code{\link{op_diff}()} \cr
+\code{\link{op_digitize}()} \cr
+\code{\link{op_divide}()} \cr
+\code{\link{op_divide_no_nan}()} \cr
+\code{\link{op_dot}()} \cr
+\code{\link{op_dot_product_attention}()} \cr
+\code{\link{op_dtype}()} \cr
+\code{\link{op_eig}()} \cr
+\code{\link{op_eigh}()} \cr
+\code{\link{op_einsum}()} \cr
+\code{\link{op_elu}()} \cr
+\code{\link{op_empty}()} \cr
+\code{\link{op_equal}()} \cr
+\code{\link{op_erf}()} \cr
+\code{\link{op_erfinv}()} \cr
+\code{\link{op_exp}()} \cr
+\code{\link{op_exp2}()} \cr
+\code{\link{op_expand_dims}()} \cr
+\code{\link{op_expm1}()} \cr
+\code{\link{op_extract_sequences}()} \cr
+\code{\link{op_eye}()} \cr
+\code{\link{op_fft}()} \cr
+\code{\link{op_fft2}()} \cr
+\code{\link{op_flip}()} \cr
+\code{\link{op_floor}()} \cr
+\code{\link{op_floor_divide}()} \cr
+\code{\link{op_fori_loop}()} \cr
+\code{\link{op_full}()} \cr
+\code{\link{op_full_like}()} \cr
+\code{\link{op_gelu}()} \cr
+\code{\link{op_get_item}()} \cr
+\code{\link{op_glu}()} \cr
+\code{\link{op_greater}()} \cr
+\code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
+\code{\link{op_hard_shrink}()} \cr
+\code{\link{op_hard_sigmoid}()} \cr
+\code{\link{op_hard_silu}()} \cr
+\code{\link{op_hard_tanh}()} \cr
+\code{\link{op_heaviside}()} \cr
+\code{\link{op_histogram}()} \cr
+\code{\link{op_hstack}()} \cr
+\code{\link{op_identity}()} \cr
+\code{\link{op_ifft2}()} \cr
+\code{\link{op_imag}()} \cr
+\code{\link{op_image_affine_transform}()} \cr
+\code{\link{op_image_crop}()} \cr
+\code{\link{op_image_extract_patches}()} \cr
+\code{\link{op_image_gaussian_blur}()} \cr
+\code{\link{op_image_hsv_to_rgb}()} \cr
+\code{\link{op_image_map_coordinates}()} \cr
+\code{\link{op_image_pad}()} \cr
+\code{\link{op_image_perspective_transform}()} \cr
+\code{\link{op_image_resize}()} \cr
+\code{\link{op_image_rgb_to_grayscale}()} \cr
+\code{\link{op_image_rgb_to_hsv}()} \cr
+\code{\link{op_in_top_k}()} \cr
+\code{\link{op_inner}()} \cr
+\code{\link{op_inv}()} \cr
+\code{\link{op_irfft}()} \cr
+\code{\link{op_is_tensor}()} \cr
+\code{\link{op_isclose}()} \cr
+\code{\link{op_isfinite}()} \cr
+\code{\link{op_isinf}()} \cr
+\code{\link{op_isnan}()} \cr
+\code{\link{op_istft}()} \cr
+\code{\link{op_kaiser}()} \cr
+\code{\link{op_layer_normalization}()} \cr
+\code{\link{op_leaky_relu}()} \cr
+\code{\link{op_left_shift}()} \cr
+\code{\link{op_less}()} \cr
+\code{\link{op_less_equal}()} \cr
+\code{\link{op_linspace}()} \cr
+\code{\link{op_log}()} \cr
+\code{\link{op_log10}()} \cr
+\code{\link{op_log1p}()} \cr
+\code{\link{op_log2}()} \cr
+\code{\link{op_log_sigmoid}()} \cr
+\code{\link{op_log_softmax}()} \cr
+\code{\link{op_logaddexp}()} \cr
+\code{\link{op_logdet}()} \cr
+\code{\link{op_logical_and}()} \cr
+\code{\link{op_logical_not}()} \cr
+\code{\link{op_logical_or}()} \cr
+\code{\link{op_logical_xor}()} \cr
+\code{\link{op_logspace}()} \cr
+\code{\link{op_logsumexp}()} \cr
+\code{\link{op_lstsq}()} \cr
+\code{\link{op_lu_factor}()} \cr
+\code{\link{op_map}()} \cr
+\code{\link{op_matmul}()} \cr
+\code{\link{op_max}()} \cr
+\code{\link{op_max_pool}()} \cr
+\code{\link{op_maximum}()} \cr
+\code{\link{op_mean}()} \cr
+\code{\link{op_median}()} \cr
+\code{\link{op_meshgrid}()} \cr
+\code{\link{op_min}()} \cr
+\code{\link{op_minimum}()} \cr
+\code{\link{op_mod}()} \cr
+\code{\link{op_moments}()} \cr
+\code{\link{op_moveaxis}()} \cr
+\code{\link{op_multi_hot}()} \cr
+\code{\link{op_multiply}()} \cr
+\code{\link{op_nan_to_num}()} \cr
+\code{\link{op_ndim}()} \cr
+\code{\link{op_negative}()} \cr
+\code{\link{op_nonzero}()} \cr
+\code{\link{op_norm}()} \cr
+\code{\link{op_normalize}()} \cr
+\code{\link{op_not_equal}()} \cr
+\code{\link{op_one_hot}()} \cr
+\code{\link{op_ones}()} \cr
+\code{\link{op_ones_like}()} \cr
+\code{\link{op_outer}()} \cr
+\code{\link{op_pad}()} \cr
+\code{\link{op_polar}()} \cr
+\code{\link{op_power}()} \cr
+\code{\link{op_prod}()} \cr
+\code{\link{op_psnr}()} \cr
+\code{\link{op_qr}()} \cr
+\code{\link{op_quantile}()} \cr
+\code{\link{op_ravel}()} \cr
+\code{\link{op_real}()} \cr
+\code{\link{op_rearrange}()} \cr
+\code{\link{op_reciprocal}()} \cr
+\code{\link{op_relu}()} \cr
+\code{\link{op_relu6}()} \cr
+\code{\link{op_repeat}()} \cr
+\code{\link{op_reshape}()} \cr
+\code{\link{op_rfft}()} \cr
+\code{\link{op_right_shift}()} \cr
+\code{\link{op_rms_normalization}()} \cr
+\code{\link{op_roll}()} \cr
+\code{\link{op_rot90}()} \cr
+\code{\link{op_round}()} \cr
+\code{\link{op_rsqrt}()} \cr
+\code{\link{op_saturate_cast}()} \cr
+\code{\link{op_scan}()} \cr
+\code{\link{op_scatter}()} \cr
+\code{\link{op_scatter_update}()} \cr
+\code{\link{op_searchsorted}()} \cr
+\code{\link{op_segment_max}()} \cr
+\code{\link{op_segment_sum}()} \cr
+\code{\link{op_select}()} \cr
+\code{\link{op_selu}()} \cr
+\code{\link{op_separable_conv}()} \cr
+\code{\link{op_shape}()} \cr
+\code{\link{op_sigmoid}()} \cr
+\code{\link{op_sign}()} \cr
+\code{\link{op_signbit}()} \cr
+\code{\link{op_silu}()} \cr
+\code{\link{op_sin}()} \cr
+\code{\link{op_sinh}()} \cr
+\code{\link{op_size}()} \cr
+\code{\link{op_slice}()} \cr
+\code{\link{op_slice_update}()} \cr
+\code{\link{op_slogdet}()} \cr
+\code{\link{op_soft_shrink}()} \cr
+\code{\link{op_softmax}()} \cr
+\code{\link{op_softplus}()} \cr
+\code{\link{op_softsign}()} \cr
+\code{\link{op_solve}()} \cr
+\code{\link{op_solve_triangular}()} \cr
+\code{\link{op_sort}()} \cr
+\code{\link{op_sparse_categorical_crossentropy}()} \cr
+\code{\link{op_sparse_plus}()} \cr
+\code{\link{op_sparsemax}()} \cr
+\code{\link{op_split}()} \cr
+\code{\link{op_sqrt}()} \cr
+\code{\link{op_square}()} \cr
+\code{\link{op_squareplus}()} \cr
+\code{\link{op_squeeze}()} \cr
+\code{\link{op_stack}()} \cr
+\code{\link{op_std}()} \cr
+\code{\link{op_stft}()} \cr
+\code{\link{op_stop_gradient}()} \cr
+\code{\link{op_subset}()} \cr
+\code{\link{op_subtract}()} \cr
+\code{\link{op_sum}()} \cr
+\code{\link{op_svd}()} \cr
+\code{\link{op_swapaxes}()} \cr
+\code{\link{op_switch}()} \cr
+\code{\link{op_take}()} \cr
+\code{\link{op_take_along_axis}()} \cr
+\code{\link{op_tan}()} \cr
+\code{\link{op_tanh}()} \cr
+\code{\link{op_tanh_shrink}()} \cr
+\code{\link{op_tensordot}()} \cr
+\code{\link{op_threshold}()} \cr
+\code{\link{op_tile}()} \cr
+\code{\link{op_top_k}()} \cr
+\code{\link{op_trace}()} \cr
+\code{\link{op_transpose}()} \cr
+\code{\link{op_tri}()} \cr
+\code{\link{op_tril}()} \cr
+\code{\link{op_triu}()} \cr
+\code{\link{op_trunc}()} \cr
+\code{\link{op_unravel_index}()} \cr
+\code{\link{op_unstack}()} \cr
+\code{\link{op_var}()} \cr
+\code{\link{op_vdot}()} \cr
+\code{\link{op_vectorize}()} \cr
+\code{\link{op_vectorized_map}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
+\code{\link{op_vstack}()} \cr
+\code{\link{op_where}()} \cr
+\code{\link{op_while_loop}()} \cr
+\code{\link{op_zeros}()} \cr
+\code{\link{op_zeros_like}()} \cr
+}
+\concept{nn ops}
+\concept{ops}
diff --git a/man/op_sparsemax.Rd b/man/op_sparsemax.Rd
index f37de1f1b..6e30bc6c4 100644
--- a/man/op_sparsemax.Rd
+++ b/man/op_sparsemax.Rd
@@ -50,6 +50,7 @@ Other nn ops: \cr
 \code{\link{op_hard_sigmoid}()} \cr
 \code{\link{op_hard_silu}()} \cr
 \code{\link{op_hard_tanh}()} \cr
+\code{\link{op_layer_normalization}()} \cr
 \code{\link{op_leaky_relu}()} \cr
 \code{\link{op_log_sigmoid}()} \cr
 \code{\link{op_log_softmax}()} \cr
@@ -73,6 +74,7 @@ Other nn ops: \cr
 \code{\link{op_softsign}()} \cr
 \code{\link{op_sparse_categorical_crossentropy}()} \cr
 \code{\link{op_sparse_plus}()} \cr
+\code{\link{op_sparse_sigmoid}()} \cr
 \code{\link{op_squareplus}()} \cr
 \code{\link{op_tanh_shrink}()} \cr
 \code{\link{op_threshold}()} \cr
@@ -82,6 +84,7 @@ Other ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -100,6 +103,7 @@ Other ops: \cr
 \code{\link{op_associative_scan}()} \cr
 \code{\link{op_average}()} \cr
 \code{\link{op_average_pool}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_batch_normalization}()} \cr
 \code{\link{op_binary_crossentropy}()} \cr
 \code{\link{op_bincount}()} \cr
@@ -110,9 +114,11 @@ Other ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
 \code{\link{op_cast}()} \cr
 \code{\link{op_categorical_crossentropy}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_celu}()} \cr
 \code{\link{op_cholesky}()} \cr
@@ -125,6 +131,7 @@ Other ops: \cr
 \code{\link{op_convert_to_numpy}()} \cr
 \code{\link{op_convert_to_tensor}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -135,6 +142,7 @@ Other ops: \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
 \code{\link{op_custom_gradient}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_depthwise_conv}()} \cr
 \code{\link{op_det}()} \cr
 \code{\link{op_diag}()} \cr
@@ -174,10 +182,13 @@ Other ops: \cr
 \code{\link{op_glu}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
 \code{\link{op_hard_shrink}()} \cr
 \code{\link{op_hard_sigmoid}()} \cr
 \code{\link{op_hard_silu}()} \cr
 \code{\link{op_hard_tanh}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -204,6 +215,8 @@ Other ops: \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
 \code{\link{op_istft}()} \cr
+\code{\link{op_kaiser}()} \cr
+\code{\link{op_layer_normalization}()} \cr
 \code{\link{op_leaky_relu}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
@@ -303,6 +316,7 @@ Other ops: \cr
 \code{\link{op_sort}()} \cr
 \code{\link{op_sparse_categorical_crossentropy}()} \cr
 \code{\link{op_sparse_plus}()} \cr
+\code{\link{op_sparse_sigmoid}()} \cr
 \code{\link{op_split}()} \cr
 \code{\link{op_sqrt}()} \cr
 \code{\link{op_square}()} \cr
@@ -339,6 +353,8 @@ Other ops: \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
 \code{\link{op_vectorized_map}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_while_loop}()} \cr
diff --git a/man/op_split.Rd b/man/op_split.Rd
index 66c4bbeda..cdb5efd1a 100644
--- a/man/op_split.Rd
+++ b/man/op_split.Rd
@@ -92,6 +92,7 @@ Other numpy ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -108,6 +109,7 @@ Other numpy ops: \cr
 \code{\link{op_argsort}()} \cr
 \code{\link{op_array}()} \cr
 \code{\link{op_average}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_bincount}()} \cr
 \code{\link{op_bitwise_and}()} \cr
 \code{\link{op_bitwise_invert}()} \cr
@@ -116,12 +118,15 @@ Other numpy ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_clip}()} \cr
 \code{\link{op_concatenate}()} \cr
 \code{\link{op_conj}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -130,6 +135,7 @@ Other numpy ops: \cr
 \code{\link{op_ctc_decode}()} \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_diag}()} \cr
 \code{\link{op_diagflat}()} \cr
 \code{\link{op_diagonal}()} \cr
@@ -154,6 +160,9 @@ Other numpy ops: \cr
 \code{\link{op_get_item}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -163,6 +172,7 @@ Other numpy ops: \cr
 \code{\link{op_isfinite}()} \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
+\code{\link{op_kaiser}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
 \code{\link{op_less_equal}()} \cr
@@ -242,6 +252,8 @@ Other numpy ops: \cr
 \code{\link{op_var}()} \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_zeros}()} \cr
@@ -251,6 +263,7 @@ Other ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -269,6 +282,7 @@ Other ops: \cr
 \code{\link{op_associative_scan}()} \cr
 \code{\link{op_average}()} \cr
 \code{\link{op_average_pool}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_batch_normalization}()} \cr
 \code{\link{op_binary_crossentropy}()} \cr
 \code{\link{op_bincount}()} \cr
@@ -279,9 +293,11 @@ Other ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
 \code{\link{op_cast}()} \cr
 \code{\link{op_categorical_crossentropy}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_celu}()} \cr
 \code{\link{op_cholesky}()} \cr
@@ -294,6 +310,7 @@ Other ops: \cr
 \code{\link{op_convert_to_numpy}()} \cr
 \code{\link{op_convert_to_tensor}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -304,6 +321,7 @@ Other ops: \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
 \code{\link{op_custom_gradient}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_depthwise_conv}()} \cr
 \code{\link{op_det}()} \cr
 \code{\link{op_diag}()} \cr
@@ -343,10 +361,13 @@ Other ops: \cr
 \code{\link{op_glu}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
 \code{\link{op_hard_shrink}()} \cr
 \code{\link{op_hard_sigmoid}()} \cr
 \code{\link{op_hard_silu}()} \cr
 \code{\link{op_hard_tanh}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -373,6 +394,8 @@ Other ops: \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
 \code{\link{op_istft}()} \cr
+\code{\link{op_kaiser}()} \cr
+\code{\link{op_layer_normalization}()} \cr
 \code{\link{op_leaky_relu}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
@@ -472,6 +495,7 @@ Other ops: \cr
 \code{\link{op_sort}()} \cr
 \code{\link{op_sparse_categorical_crossentropy}()} \cr
 \code{\link{op_sparse_plus}()} \cr
+\code{\link{op_sparse_sigmoid}()} \cr
 \code{\link{op_sparsemax}()} \cr
 \code{\link{op_sqrt}()} \cr
 \code{\link{op_square}()} \cr
@@ -508,6 +532,8 @@ Other ops: \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
 \code{\link{op_vectorized_map}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_while_loop}()} \cr
diff --git a/man/op_sqrt.Rd b/man/op_sqrt.Rd
index a92c3248e..4b613785b 100644
--- a/man/op_sqrt.Rd
+++ b/man/op_sqrt.Rd
@@ -24,6 +24,7 @@ Other numpy ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -40,6 +41,7 @@ Other numpy ops: \cr
 \code{\link{op_argsort}()} \cr
 \code{\link{op_array}()} \cr
 \code{\link{op_average}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_bincount}()} \cr
 \code{\link{op_bitwise_and}()} \cr
 \code{\link{op_bitwise_invert}()} \cr
@@ -48,12 +50,15 @@ Other numpy ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_clip}()} \cr
 \code{\link{op_concatenate}()} \cr
 \code{\link{op_conj}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -62,6 +67,7 @@ Other numpy ops: \cr
 \code{\link{op_ctc_decode}()} \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_diag}()} \cr
 \code{\link{op_diagflat}()} \cr
 \code{\link{op_diagonal}()} \cr
@@ -86,6 +92,9 @@ Other numpy ops: \cr
 \code{\link{op_get_item}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -95,6 +104,7 @@ Other numpy ops: \cr
 \code{\link{op_isfinite}()} \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
+\code{\link{op_kaiser}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
 \code{\link{op_less_equal}()} \cr
@@ -174,6 +184,8 @@ Other numpy ops: \cr
 \code{\link{op_var}()} \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_zeros}()} \cr
@@ -183,6 +195,7 @@ Other ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -201,6 +214,7 @@ Other ops: \cr
 \code{\link{op_associative_scan}()} \cr
 \code{\link{op_average}()} \cr
 \code{\link{op_average_pool}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_batch_normalization}()} \cr
 \code{\link{op_binary_crossentropy}()} \cr
 \code{\link{op_bincount}()} \cr
@@ -211,9 +225,11 @@ Other ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
 \code{\link{op_cast}()} \cr
 \code{\link{op_categorical_crossentropy}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_celu}()} \cr
 \code{\link{op_cholesky}()} \cr
@@ -226,6 +242,7 @@ Other ops: \cr
 \code{\link{op_convert_to_numpy}()} \cr
 \code{\link{op_convert_to_tensor}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -236,6 +253,7 @@ Other ops: \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
 \code{\link{op_custom_gradient}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_depthwise_conv}()} \cr
 \code{\link{op_det}()} \cr
 \code{\link{op_diag}()} \cr
@@ -275,10 +293,13 @@ Other ops: \cr
 \code{\link{op_glu}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
 \code{\link{op_hard_shrink}()} \cr
 \code{\link{op_hard_sigmoid}()} \cr
 \code{\link{op_hard_silu}()} \cr
 \code{\link{op_hard_tanh}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -305,6 +326,8 @@ Other ops: \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
 \code{\link{op_istft}()} \cr
+\code{\link{op_kaiser}()} \cr
+\code{\link{op_layer_normalization}()} \cr
 \code{\link{op_leaky_relu}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
@@ -404,6 +427,7 @@ Other ops: \cr
 \code{\link{op_sort}()} \cr
 \code{\link{op_sparse_categorical_crossentropy}()} \cr
 \code{\link{op_sparse_plus}()} \cr
+\code{\link{op_sparse_sigmoid}()} \cr
 \code{\link{op_sparsemax}()} \cr
 \code{\link{op_split}()} \cr
 \code{\link{op_square}()} \cr
@@ -440,6 +464,8 @@ Other ops: \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
 \code{\link{op_vectorized_map}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_while_loop}()} \cr
diff --git a/man/op_square.Rd b/man/op_square.Rd
index 3a82b091d..07fbb5642 100644
--- a/man/op_square.Rd
+++ b/man/op_square.Rd
@@ -24,6 +24,7 @@ Other numpy ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -40,6 +41,7 @@ Other numpy ops: \cr
 \code{\link{op_argsort}()} \cr
 \code{\link{op_array}()} \cr
 \code{\link{op_average}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_bincount}()} \cr
 \code{\link{op_bitwise_and}()} \cr
 \code{\link{op_bitwise_invert}()} \cr
@@ -48,12 +50,15 @@ Other numpy ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_clip}()} \cr
 \code{\link{op_concatenate}()} \cr
 \code{\link{op_conj}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -62,6 +67,7 @@ Other numpy ops: \cr
 \code{\link{op_ctc_decode}()} \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_diag}()} \cr
 \code{\link{op_diagflat}()} \cr
 \code{\link{op_diagonal}()} \cr
@@ -86,6 +92,9 @@ Other numpy ops: \cr
 \code{\link{op_get_item}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -95,6 +104,7 @@ Other numpy ops: \cr
 \code{\link{op_isfinite}()} \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
+\code{\link{op_kaiser}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
 \code{\link{op_less_equal}()} \cr
@@ -174,6 +184,8 @@ Other numpy ops: \cr
 \code{\link{op_var}()} \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_zeros}()} \cr
@@ -183,6 +195,7 @@ Other ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -201,6 +214,7 @@ Other ops: \cr
 \code{\link{op_associative_scan}()} \cr
 \code{\link{op_average}()} \cr
 \code{\link{op_average_pool}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_batch_normalization}()} \cr
 \code{\link{op_binary_crossentropy}()} \cr
 \code{\link{op_bincount}()} \cr
@@ -211,9 +225,11 @@ Other ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
 \code{\link{op_cast}()} \cr
 \code{\link{op_categorical_crossentropy}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_celu}()} \cr
 \code{\link{op_cholesky}()} \cr
@@ -226,6 +242,7 @@ Other ops: \cr
 \code{\link{op_convert_to_numpy}()} \cr
 \code{\link{op_convert_to_tensor}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -236,6 +253,7 @@ Other ops: \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
 \code{\link{op_custom_gradient}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_depthwise_conv}()} \cr
 \code{\link{op_det}()} \cr
 \code{\link{op_diag}()} \cr
@@ -275,10 +293,13 @@ Other ops: \cr
 \code{\link{op_glu}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
 \code{\link{op_hard_shrink}()} \cr
 \code{\link{op_hard_sigmoid}()} \cr
 \code{\link{op_hard_silu}()} \cr
 \code{\link{op_hard_tanh}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -305,6 +326,8 @@ Other ops: \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
 \code{\link{op_istft}()} \cr
+\code{\link{op_kaiser}()} \cr
+\code{\link{op_layer_normalization}()} \cr
 \code{\link{op_leaky_relu}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
@@ -404,6 +427,7 @@ Other ops: \cr
 \code{\link{op_sort}()} \cr
 \code{\link{op_sparse_categorical_crossentropy}()} \cr
 \code{\link{op_sparse_plus}()} \cr
+\code{\link{op_sparse_sigmoid}()} \cr
 \code{\link{op_sparsemax}()} \cr
 \code{\link{op_split}()} \cr
 \code{\link{op_sqrt}()} \cr
@@ -440,6 +464,8 @@ Other ops: \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
 \code{\link{op_vectorized_map}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_while_loop}()} \cr
diff --git a/man/op_squareplus.Rd b/man/op_squareplus.Rd
index b9add3e57..b3ae4bb1a 100644
--- a/man/op_squareplus.Rd
+++ b/man/op_squareplus.Rd
@@ -48,6 +48,7 @@ Other nn ops: \cr
 \code{\link{op_hard_sigmoid}()} \cr
 \code{\link{op_hard_silu}()} \cr
 \code{\link{op_hard_tanh}()} \cr
+\code{\link{op_layer_normalization}()} \cr
 \code{\link{op_leaky_relu}()} \cr
 \code{\link{op_log_sigmoid}()} \cr
 \code{\link{op_log_softmax}()} \cr
@@ -71,6 +72,7 @@ Other nn ops: \cr
 \code{\link{op_softsign}()} \cr
 \code{\link{op_sparse_categorical_crossentropy}()} \cr
 \code{\link{op_sparse_plus}()} \cr
+\code{\link{op_sparse_sigmoid}()} \cr
 \code{\link{op_sparsemax}()} \cr
 \code{\link{op_tanh_shrink}()} \cr
 \code{\link{op_threshold}()} \cr
@@ -80,6 +82,7 @@ Other ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -98,6 +101,7 @@ Other ops: \cr
 \code{\link{op_associative_scan}()} \cr
 \code{\link{op_average}()} \cr
 \code{\link{op_average_pool}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_batch_normalization}()} \cr
 \code{\link{op_binary_crossentropy}()} \cr
 \code{\link{op_bincount}()} \cr
@@ -108,9 +112,11 @@ Other ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
 \code{\link{op_cast}()} \cr
 \code{\link{op_categorical_crossentropy}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_celu}()} \cr
 \code{\link{op_cholesky}()} \cr
@@ -123,6 +129,7 @@ Other ops: \cr
 \code{\link{op_convert_to_numpy}()} \cr
 \code{\link{op_convert_to_tensor}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -133,6 +140,7 @@ Other ops: \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
 \code{\link{op_custom_gradient}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_depthwise_conv}()} \cr
 \code{\link{op_det}()} \cr
 \code{\link{op_diag}()} \cr
@@ -172,10 +180,13 @@ Other ops: \cr
 \code{\link{op_glu}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
 \code{\link{op_hard_shrink}()} \cr
 \code{\link{op_hard_sigmoid}()} \cr
 \code{\link{op_hard_silu}()} \cr
 \code{\link{op_hard_tanh}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -202,6 +213,8 @@ Other ops: \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
 \code{\link{op_istft}()} \cr
+\code{\link{op_kaiser}()} \cr
+\code{\link{op_layer_normalization}()} \cr
 \code{\link{op_leaky_relu}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
@@ -301,6 +314,7 @@ Other ops: \cr
 \code{\link{op_sort}()} \cr
 \code{\link{op_sparse_categorical_crossentropy}()} \cr
 \code{\link{op_sparse_plus}()} \cr
+\code{\link{op_sparse_sigmoid}()} \cr
 \code{\link{op_sparsemax}()} \cr
 \code{\link{op_split}()} \cr
 \code{\link{op_sqrt}()} \cr
@@ -337,6 +351,8 @@ Other ops: \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
 \code{\link{op_vectorized_map}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_while_loop}()} \cr
diff --git a/man/op_squeeze.Rd b/man/op_squeeze.Rd
index 533814893..98ff73f56 100644
--- a/man/op_squeeze.Rd
+++ b/man/op_squeeze.Rd
@@ -27,6 +27,7 @@ Other numpy ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -43,6 +44,7 @@ Other numpy ops: \cr
 \code{\link{op_argsort}()} \cr
 \code{\link{op_array}()} \cr
 \code{\link{op_average}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_bincount}()} \cr
 \code{\link{op_bitwise_and}()} \cr
 \code{\link{op_bitwise_invert}()} \cr
@@ -51,12 +53,15 @@ Other numpy ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_clip}()} \cr
 \code{\link{op_concatenate}()} \cr
 \code{\link{op_conj}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -65,6 +70,7 @@ Other numpy ops: \cr
 \code{\link{op_ctc_decode}()} \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_diag}()} \cr
 \code{\link{op_diagflat}()} \cr
 \code{\link{op_diagonal}()} \cr
@@ -89,6 +95,9 @@ Other numpy ops: \cr
 \code{\link{op_get_item}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -98,6 +107,7 @@ Other numpy ops: \cr
 \code{\link{op_isfinite}()} \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
+\code{\link{op_kaiser}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
 \code{\link{op_less_equal}()} \cr
@@ -177,6 +187,8 @@ Other numpy ops: \cr
 \code{\link{op_var}()} \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_zeros}()} \cr
@@ -186,6 +198,7 @@ Other ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -204,6 +217,7 @@ Other ops: \cr
 \code{\link{op_associative_scan}()} \cr
 \code{\link{op_average}()} \cr
 \code{\link{op_average_pool}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_batch_normalization}()} \cr
 \code{\link{op_binary_crossentropy}()} \cr
 \code{\link{op_bincount}()} \cr
@@ -214,9 +228,11 @@ Other ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
 \code{\link{op_cast}()} \cr
 \code{\link{op_categorical_crossentropy}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_celu}()} \cr
 \code{\link{op_cholesky}()} \cr
@@ -229,6 +245,7 @@ Other ops: \cr
 \code{\link{op_convert_to_numpy}()} \cr
 \code{\link{op_convert_to_tensor}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -239,6 +256,7 @@ Other ops: \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
 \code{\link{op_custom_gradient}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_depthwise_conv}()} \cr
 \code{\link{op_det}()} \cr
 \code{\link{op_diag}()} \cr
@@ -278,10 +296,13 @@ Other ops: \cr
 \code{\link{op_glu}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
 \code{\link{op_hard_shrink}()} \cr
 \code{\link{op_hard_sigmoid}()} \cr
 \code{\link{op_hard_silu}()} \cr
 \code{\link{op_hard_tanh}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -308,6 +329,8 @@ Other ops: \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
 \code{\link{op_istft}()} \cr
+\code{\link{op_kaiser}()} \cr
+\code{\link{op_layer_normalization}()} \cr
 \code{\link{op_leaky_relu}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
@@ -407,6 +430,7 @@ Other ops: \cr
 \code{\link{op_sort}()} \cr
 \code{\link{op_sparse_categorical_crossentropy}()} \cr
 \code{\link{op_sparse_plus}()} \cr
+\code{\link{op_sparse_sigmoid}()} \cr
 \code{\link{op_sparsemax}()} \cr
 \code{\link{op_split}()} \cr
 \code{\link{op_sqrt}()} \cr
@@ -443,6 +467,8 @@ Other ops: \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
 \code{\link{op_vectorized_map}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_while_loop}()} \cr
diff --git a/man/op_stack.Rd b/man/op_stack.Rd
index 0d8b7092c..a651a802c 100644
--- a/man/op_stack.Rd
+++ b/man/op_stack.Rd
@@ -27,6 +27,7 @@ Other numpy ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -43,6 +44,7 @@ Other numpy ops: \cr
 \code{\link{op_argsort}()} \cr
 \code{\link{op_array}()} \cr
 \code{\link{op_average}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_bincount}()} \cr
 \code{\link{op_bitwise_and}()} \cr
 \code{\link{op_bitwise_invert}()} \cr
@@ -51,12 +53,15 @@ Other numpy ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_clip}()} \cr
 \code{\link{op_concatenate}()} \cr
 \code{\link{op_conj}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -65,6 +70,7 @@ Other numpy ops: \cr
 \code{\link{op_ctc_decode}()} \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_diag}()} \cr
 \code{\link{op_diagflat}()} \cr
 \code{\link{op_diagonal}()} \cr
@@ -89,6 +95,9 @@ Other numpy ops: \cr
 \code{\link{op_get_item}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -98,6 +107,7 @@ Other numpy ops: \cr
 \code{\link{op_isfinite}()} \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
+\code{\link{op_kaiser}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
 \code{\link{op_less_equal}()} \cr
@@ -177,6 +187,8 @@ Other numpy ops: \cr
 \code{\link{op_var}()} \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_zeros}()} \cr
@@ -186,6 +198,7 @@ Other ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -204,6 +217,7 @@ Other ops: \cr
 \code{\link{op_associative_scan}()} \cr
 \code{\link{op_average}()} \cr
 \code{\link{op_average_pool}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_batch_normalization}()} \cr
 \code{\link{op_binary_crossentropy}()} \cr
 \code{\link{op_bincount}()} \cr
@@ -214,9 +228,11 @@ Other ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
 \code{\link{op_cast}()} \cr
 \code{\link{op_categorical_crossentropy}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_celu}()} \cr
 \code{\link{op_cholesky}()} \cr
@@ -229,6 +245,7 @@ Other ops: \cr
 \code{\link{op_convert_to_numpy}()} \cr
 \code{\link{op_convert_to_tensor}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -239,6 +256,7 @@ Other ops: \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
 \code{\link{op_custom_gradient}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_depthwise_conv}()} \cr
 \code{\link{op_det}()} \cr
 \code{\link{op_diag}()} \cr
@@ -278,10 +296,13 @@ Other ops: \cr
 \code{\link{op_glu}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
 \code{\link{op_hard_shrink}()} \cr
 \code{\link{op_hard_sigmoid}()} \cr
 \code{\link{op_hard_silu}()} \cr
 \code{\link{op_hard_tanh}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -308,6 +329,8 @@ Other ops: \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
 \code{\link{op_istft}()} \cr
+\code{\link{op_kaiser}()} \cr
+\code{\link{op_layer_normalization}()} \cr
 \code{\link{op_leaky_relu}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
@@ -407,6 +430,7 @@ Other ops: \cr
 \code{\link{op_sort}()} \cr
 \code{\link{op_sparse_categorical_crossentropy}()} \cr
 \code{\link{op_sparse_plus}()} \cr
+\code{\link{op_sparse_sigmoid}()} \cr
 \code{\link{op_sparsemax}()} \cr
 \code{\link{op_split}()} \cr
 \code{\link{op_sqrt}()} \cr
@@ -443,6 +467,8 @@ Other ops: \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
 \code{\link{op_vectorized_map}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_while_loop}()} \cr
diff --git a/man/op_std.Rd b/man/op_std.Rd
index 64b906b67..c54e320f7 100644
--- a/man/op_std.Rd
+++ b/man/op_std.Rd
@@ -31,6 +31,7 @@ Other numpy ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -47,6 +48,7 @@ Other numpy ops: \cr
 \code{\link{op_argsort}()} \cr
 \code{\link{op_array}()} \cr
 \code{\link{op_average}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_bincount}()} \cr
 \code{\link{op_bitwise_and}()} \cr
 \code{\link{op_bitwise_invert}()} \cr
@@ -55,12 +57,15 @@ Other numpy ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_clip}()} \cr
 \code{\link{op_concatenate}()} \cr
 \code{\link{op_conj}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -69,6 +74,7 @@ Other numpy ops: \cr
 \code{\link{op_ctc_decode}()} \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_diag}()} \cr
 \code{\link{op_diagflat}()} \cr
 \code{\link{op_diagonal}()} \cr
@@ -93,6 +99,9 @@ Other numpy ops: \cr
 \code{\link{op_get_item}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -102,6 +111,7 @@ Other numpy ops: \cr
 \code{\link{op_isfinite}()} \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
+\code{\link{op_kaiser}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
 \code{\link{op_less_equal}()} \cr
@@ -181,6 +191,8 @@ Other numpy ops: \cr
 \code{\link{op_var}()} \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_zeros}()} \cr
@@ -190,6 +202,7 @@ Other ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -208,6 +221,7 @@ Other ops: \cr
 \code{\link{op_associative_scan}()} \cr
 \code{\link{op_average}()} \cr
 \code{\link{op_average_pool}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_batch_normalization}()} \cr
 \code{\link{op_binary_crossentropy}()} \cr
 \code{\link{op_bincount}()} \cr
@@ -218,9 +232,11 @@ Other ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
 \code{\link{op_cast}()} \cr
 \code{\link{op_categorical_crossentropy}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_celu}()} \cr
 \code{\link{op_cholesky}()} \cr
@@ -233,6 +249,7 @@ Other ops: \cr
 \code{\link{op_convert_to_numpy}()} \cr
 \code{\link{op_convert_to_tensor}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -243,6 +260,7 @@ Other ops: \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
 \code{\link{op_custom_gradient}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_depthwise_conv}()} \cr
 \code{\link{op_det}()} \cr
 \code{\link{op_diag}()} \cr
@@ -282,10 +300,13 @@ Other ops: \cr
 \code{\link{op_glu}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
 \code{\link{op_hard_shrink}()} \cr
 \code{\link{op_hard_sigmoid}()} \cr
 \code{\link{op_hard_silu}()} \cr
 \code{\link{op_hard_tanh}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -312,6 +333,8 @@ Other ops: \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
 \code{\link{op_istft}()} \cr
+\code{\link{op_kaiser}()} \cr
+\code{\link{op_layer_normalization}()} \cr
 \code{\link{op_leaky_relu}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
@@ -411,6 +434,7 @@ Other ops: \cr
 \code{\link{op_sort}()} \cr
 \code{\link{op_sparse_categorical_crossentropy}()} \cr
 \code{\link{op_sparse_plus}()} \cr
+\code{\link{op_sparse_sigmoid}()} \cr
 \code{\link{op_sparsemax}()} \cr
 \code{\link{op_split}()} \cr
 \code{\link{op_sqrt}()} \cr
@@ -447,6 +471,8 @@ Other ops: \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
 \code{\link{op_vectorized_map}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_while_loop}()} \cr
diff --git a/man/op_stft.Rd b/man/op_stft.Rd
index f51e52784..4ac4adb72 100644
--- a/man/op_stft.Rd
+++ b/man/op_stft.Rd
@@ -90,6 +90,7 @@ Other ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -108,6 +109,7 @@ Other ops: \cr
 \code{\link{op_associative_scan}()} \cr
 \code{\link{op_average}()} \cr
 \code{\link{op_average_pool}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_batch_normalization}()} \cr
 \code{\link{op_binary_crossentropy}()} \cr
 \code{\link{op_bincount}()} \cr
@@ -118,9 +120,11 @@ Other ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
 \code{\link{op_cast}()} \cr
 \code{\link{op_categorical_crossentropy}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_celu}()} \cr
 \code{\link{op_cholesky}()} \cr
@@ -133,6 +137,7 @@ Other ops: \cr
 \code{\link{op_convert_to_numpy}()} \cr
 \code{\link{op_convert_to_tensor}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -143,6 +148,7 @@ Other ops: \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
 \code{\link{op_custom_gradient}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_depthwise_conv}()} \cr
 \code{\link{op_det}()} \cr
 \code{\link{op_diag}()} \cr
@@ -182,10 +188,13 @@ Other ops: \cr
 \code{\link{op_glu}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
 \code{\link{op_hard_shrink}()} \cr
 \code{\link{op_hard_sigmoid}()} \cr
 \code{\link{op_hard_silu}()} \cr
 \code{\link{op_hard_tanh}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -212,6 +221,8 @@ Other ops: \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
 \code{\link{op_istft}()} \cr
+\code{\link{op_kaiser}()} \cr
+\code{\link{op_layer_normalization}()} \cr
 \code{\link{op_leaky_relu}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
@@ -311,6 +322,7 @@ Other ops: \cr
 \code{\link{op_sort}()} \cr
 \code{\link{op_sparse_categorical_crossentropy}()} \cr
 \code{\link{op_sparse_plus}()} \cr
+\code{\link{op_sparse_sigmoid}()} \cr
 \code{\link{op_sparsemax}()} \cr
 \code{\link{op_split}()} \cr
 \code{\link{op_sqrt}()} \cr
@@ -347,6 +359,8 @@ Other ops: \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
 \code{\link{op_vectorized_map}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_while_loop}()} \cr
diff --git a/man/op_stop_gradient.Rd b/man/op_stop_gradient.Rd
index efec6c7b8..4d2749415 100644
--- a/man/op_stop_gradient.Rd
+++ b/man/op_stop_gradient.Rd
@@ -56,6 +56,7 @@ Other ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -74,6 +75,7 @@ Other ops: \cr
 \code{\link{op_associative_scan}()} \cr
 \code{\link{op_average}()} \cr
 \code{\link{op_average_pool}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_batch_normalization}()} \cr
 \code{\link{op_binary_crossentropy}()} \cr
 \code{\link{op_bincount}()} \cr
@@ -84,9 +86,11 @@ Other ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
 \code{\link{op_cast}()} \cr
 \code{\link{op_categorical_crossentropy}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_celu}()} \cr
 \code{\link{op_cholesky}()} \cr
@@ -99,6 +103,7 @@ Other ops: \cr
 \code{\link{op_convert_to_numpy}()} \cr
 \code{\link{op_convert_to_tensor}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -109,6 +114,7 @@ Other ops: \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
 \code{\link{op_custom_gradient}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_depthwise_conv}()} \cr
 \code{\link{op_det}()} \cr
 \code{\link{op_diag}()} \cr
@@ -148,10 +154,13 @@ Other ops: \cr
 \code{\link{op_glu}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
 \code{\link{op_hard_shrink}()} \cr
 \code{\link{op_hard_sigmoid}()} \cr
 \code{\link{op_hard_silu}()} \cr
 \code{\link{op_hard_tanh}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -178,6 +187,8 @@ Other ops: \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
 \code{\link{op_istft}()} \cr
+\code{\link{op_kaiser}()} \cr
+\code{\link{op_layer_normalization}()} \cr
 \code{\link{op_leaky_relu}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
@@ -277,6 +288,7 @@ Other ops: \cr
 \code{\link{op_sort}()} \cr
 \code{\link{op_sparse_categorical_crossentropy}()} \cr
 \code{\link{op_sparse_plus}()} \cr
+\code{\link{op_sparse_sigmoid}()} \cr
 \code{\link{op_sparsemax}()} \cr
 \code{\link{op_split}()} \cr
 \code{\link{op_sqrt}()} \cr
@@ -313,6 +325,8 @@ Other ops: \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
 \code{\link{op_vectorized_map}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_while_loop}()} \cr
diff --git a/man/op_subset.Rd b/man/op_subset.Rd
index e87ee30cd..513ba4539 100644
--- a/man/op_subset.Rd
+++ b/man/op_subset.Rd
@@ -467,6 +467,7 @@ Other ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -485,6 +486,7 @@ Other ops: \cr
 \code{\link{op_associative_scan}()} \cr
 \code{\link{op_average}()} \cr
 \code{\link{op_average_pool}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_batch_normalization}()} \cr
 \code{\link{op_binary_crossentropy}()} \cr
 \code{\link{op_bincount}()} \cr
@@ -495,9 +497,11 @@ Other ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
 \code{\link{op_cast}()} \cr
 \code{\link{op_categorical_crossentropy}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_celu}()} \cr
 \code{\link{op_cholesky}()} \cr
@@ -510,6 +514,7 @@ Other ops: \cr
 \code{\link{op_convert_to_numpy}()} \cr
 \code{\link{op_convert_to_tensor}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -520,6 +525,7 @@ Other ops: \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
 \code{\link{op_custom_gradient}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_depthwise_conv}()} \cr
 \code{\link{op_det}()} \cr
 \code{\link{op_diag}()} \cr
@@ -559,10 +565,13 @@ Other ops: \cr
 \code{\link{op_glu}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
 \code{\link{op_hard_shrink}()} \cr
 \code{\link{op_hard_sigmoid}()} \cr
 \code{\link{op_hard_silu}()} \cr
 \code{\link{op_hard_tanh}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -589,6 +598,8 @@ Other ops: \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
 \code{\link{op_istft}()} \cr
+\code{\link{op_kaiser}()} \cr
+\code{\link{op_layer_normalization}()} \cr
 \code{\link{op_leaky_relu}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
@@ -688,6 +699,7 @@ Other ops: \cr
 \code{\link{op_sort}()} \cr
 \code{\link{op_sparse_categorical_crossentropy}()} \cr
 \code{\link{op_sparse_plus}()} \cr
+\code{\link{op_sparse_sigmoid}()} \cr
 \code{\link{op_sparsemax}()} \cr
 \code{\link{op_split}()} \cr
 \code{\link{op_sqrt}()} \cr
@@ -724,6 +736,8 @@ Other ops: \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
 \code{\link{op_vectorized_map}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_while_loop}()} \cr
@@ -759,6 +773,7 @@ Other ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -777,6 +792,7 @@ Other ops: \cr
 \code{\link{op_associative_scan}()} \cr
 \code{\link{op_average}()} \cr
 \code{\link{op_average_pool}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_batch_normalization}()} \cr
 \code{\link{op_binary_crossentropy}()} \cr
 \code{\link{op_bincount}()} \cr
@@ -787,9 +803,11 @@ Other ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
 \code{\link{op_cast}()} \cr
 \code{\link{op_categorical_crossentropy}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_celu}()} \cr
 \code{\link{op_cholesky}()} \cr
@@ -802,6 +820,7 @@ Other ops: \cr
 \code{\link{op_convert_to_numpy}()} \cr
 \code{\link{op_convert_to_tensor}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -812,6 +831,7 @@ Other ops: \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
 \code{\link{op_custom_gradient}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_depthwise_conv}()} \cr
 \code{\link{op_det}()} \cr
 \code{\link{op_diag}()} \cr
@@ -851,10 +871,13 @@ Other ops: \cr
 \code{\link{op_glu}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
 \code{\link{op_hard_shrink}()} \cr
 \code{\link{op_hard_sigmoid}()} \cr
 \code{\link{op_hard_silu}()} \cr
 \code{\link{op_hard_tanh}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -881,6 +904,8 @@ Other ops: \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
 \code{\link{op_istft}()} \cr
+\code{\link{op_kaiser}()} \cr
+\code{\link{op_layer_normalization}()} \cr
 \code{\link{op_leaky_relu}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
@@ -980,6 +1005,7 @@ Other ops: \cr
 \code{\link{op_sort}()} \cr
 \code{\link{op_sparse_categorical_crossentropy}()} \cr
 \code{\link{op_sparse_plus}()} \cr
+\code{\link{op_sparse_sigmoid}()} \cr
 \code{\link{op_sparsemax}()} \cr
 \code{\link{op_split}()} \cr
 \code{\link{op_sqrt}()} \cr
@@ -1016,6 +1042,8 @@ Other ops: \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
 \code{\link{op_vectorized_map}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_while_loop}()} \cr
@@ -1051,6 +1079,7 @@ Other ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -1069,6 +1098,7 @@ Other ops: \cr
 \code{\link{op_associative_scan}()} \cr
 \code{\link{op_average}()} \cr
 \code{\link{op_average_pool}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_batch_normalization}()} \cr
 \code{\link{op_binary_crossentropy}()} \cr
 \code{\link{op_bincount}()} \cr
@@ -1079,9 +1109,11 @@ Other ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
 \code{\link{op_cast}()} \cr
 \code{\link{op_categorical_crossentropy}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_celu}()} \cr
 \code{\link{op_cholesky}()} \cr
@@ -1094,6 +1126,7 @@ Other ops: \cr
 \code{\link{op_convert_to_numpy}()} \cr
 \code{\link{op_convert_to_tensor}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -1104,6 +1137,7 @@ Other ops: \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
 \code{\link{op_custom_gradient}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_depthwise_conv}()} \cr
 \code{\link{op_det}()} \cr
 \code{\link{op_diag}()} \cr
@@ -1143,10 +1177,13 @@ Other ops: \cr
 \code{\link{op_glu}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
 \code{\link{op_hard_shrink}()} \cr
 \code{\link{op_hard_sigmoid}()} \cr
 \code{\link{op_hard_silu}()} \cr
 \code{\link{op_hard_tanh}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -1173,6 +1210,8 @@ Other ops: \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
 \code{\link{op_istft}()} \cr
+\code{\link{op_kaiser}()} \cr
+\code{\link{op_layer_normalization}()} \cr
 \code{\link{op_leaky_relu}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
@@ -1272,6 +1311,7 @@ Other ops: \cr
 \code{\link{op_sort}()} \cr
 \code{\link{op_sparse_categorical_crossentropy}()} \cr
 \code{\link{op_sparse_plus}()} \cr
+\code{\link{op_sparse_sigmoid}()} \cr
 \code{\link{op_sparsemax}()} \cr
 \code{\link{op_split}()} \cr
 \code{\link{op_sqrt}()} \cr
@@ -1308,6 +1348,8 @@ Other ops: \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
 \code{\link{op_vectorized_map}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_while_loop}()} \cr
diff --git a/man/op_subtract.Rd b/man/op_subtract.Rd
index c0e1af55c..b140b50e9 100644
--- a/man/op_subtract.Rd
+++ b/man/op_subtract.Rd
@@ -41,6 +41,7 @@ Other numpy ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -57,6 +58,7 @@ Other numpy ops: \cr
 \code{\link{op_argsort}()} \cr
 \code{\link{op_array}()} \cr
 \code{\link{op_average}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_bincount}()} \cr
 \code{\link{op_bitwise_and}()} \cr
 \code{\link{op_bitwise_invert}()} \cr
@@ -65,12 +67,15 @@ Other numpy ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_clip}()} \cr
 \code{\link{op_concatenate}()} \cr
 \code{\link{op_conj}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -79,6 +84,7 @@ Other numpy ops: \cr
 \code{\link{op_ctc_decode}()} \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_diag}()} \cr
 \code{\link{op_diagflat}()} \cr
 \code{\link{op_diagonal}()} \cr
@@ -103,6 +109,9 @@ Other numpy ops: \cr
 \code{\link{op_get_item}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -112,6 +121,7 @@ Other numpy ops: \cr
 \code{\link{op_isfinite}()} \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
+\code{\link{op_kaiser}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
 \code{\link{op_less_equal}()} \cr
@@ -191,6 +201,8 @@ Other numpy ops: \cr
 \code{\link{op_var}()} \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_zeros}()} \cr
@@ -200,6 +212,7 @@ Other ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -218,6 +231,7 @@ Other ops: \cr
 \code{\link{op_associative_scan}()} \cr
 \code{\link{op_average}()} \cr
 \code{\link{op_average_pool}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_batch_normalization}()} \cr
 \code{\link{op_binary_crossentropy}()} \cr
 \code{\link{op_bincount}()} \cr
@@ -228,9 +242,11 @@ Other ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
 \code{\link{op_cast}()} \cr
 \code{\link{op_categorical_crossentropy}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_celu}()} \cr
 \code{\link{op_cholesky}()} \cr
@@ -243,6 +259,7 @@ Other ops: \cr
 \code{\link{op_convert_to_numpy}()} \cr
 \code{\link{op_convert_to_tensor}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -253,6 +270,7 @@ Other ops: \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
 \code{\link{op_custom_gradient}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_depthwise_conv}()} \cr
 \code{\link{op_det}()} \cr
 \code{\link{op_diag}()} \cr
@@ -292,10 +310,13 @@ Other ops: \cr
 \code{\link{op_glu}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
 \code{\link{op_hard_shrink}()} \cr
 \code{\link{op_hard_sigmoid}()} \cr
 \code{\link{op_hard_silu}()} \cr
 \code{\link{op_hard_tanh}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -322,6 +343,8 @@ Other ops: \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
 \code{\link{op_istft}()} \cr
+\code{\link{op_kaiser}()} \cr
+\code{\link{op_layer_normalization}()} \cr
 \code{\link{op_leaky_relu}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
@@ -421,6 +444,7 @@ Other ops: \cr
 \code{\link{op_sort}()} \cr
 \code{\link{op_sparse_categorical_crossentropy}()} \cr
 \code{\link{op_sparse_plus}()} \cr
+\code{\link{op_sparse_sigmoid}()} \cr
 \code{\link{op_sparsemax}()} \cr
 \code{\link{op_split}()} \cr
 \code{\link{op_sqrt}()} \cr
@@ -457,6 +481,8 @@ Other ops: \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
 \code{\link{op_vectorized_map}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_while_loop}()} \cr
diff --git a/man/op_sum.Rd b/man/op_sum.Rd
index 1cc47bfe3..b53604689 100644
--- a/man/op_sum.Rd
+++ b/man/op_sum.Rd
@@ -30,6 +30,7 @@ Other numpy ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -46,6 +47,7 @@ Other numpy ops: \cr
 \code{\link{op_argsort}()} \cr
 \code{\link{op_array}()} \cr
 \code{\link{op_average}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_bincount}()} \cr
 \code{\link{op_bitwise_and}()} \cr
 \code{\link{op_bitwise_invert}()} \cr
@@ -54,12 +56,15 @@ Other numpy ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_clip}()} \cr
 \code{\link{op_concatenate}()} \cr
 \code{\link{op_conj}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -68,6 +73,7 @@ Other numpy ops: \cr
 \code{\link{op_ctc_decode}()} \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_diag}()} \cr
 \code{\link{op_diagflat}()} \cr
 \code{\link{op_diagonal}()} \cr
@@ -92,6 +98,9 @@ Other numpy ops: \cr
 \code{\link{op_get_item}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -101,6 +110,7 @@ Other numpy ops: \cr
 \code{\link{op_isfinite}()} \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
+\code{\link{op_kaiser}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
 \code{\link{op_less_equal}()} \cr
@@ -180,6 +190,8 @@ Other numpy ops: \cr
 \code{\link{op_var}()} \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_zeros}()} \cr
@@ -189,6 +201,7 @@ Other ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -207,6 +220,7 @@ Other ops: \cr
 \code{\link{op_associative_scan}()} \cr
 \code{\link{op_average}()} \cr
 \code{\link{op_average_pool}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_batch_normalization}()} \cr
 \code{\link{op_binary_crossentropy}()} \cr
 \code{\link{op_bincount}()} \cr
@@ -217,9 +231,11 @@ Other ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
 \code{\link{op_cast}()} \cr
 \code{\link{op_categorical_crossentropy}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_celu}()} \cr
 \code{\link{op_cholesky}()} \cr
@@ -232,6 +248,7 @@ Other ops: \cr
 \code{\link{op_convert_to_numpy}()} \cr
 \code{\link{op_convert_to_tensor}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -242,6 +259,7 @@ Other ops: \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
 \code{\link{op_custom_gradient}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_depthwise_conv}()} \cr
 \code{\link{op_det}()} \cr
 \code{\link{op_diag}()} \cr
@@ -281,10 +299,13 @@ Other ops: \cr
 \code{\link{op_glu}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
 \code{\link{op_hard_shrink}()} \cr
 \code{\link{op_hard_sigmoid}()} \cr
 \code{\link{op_hard_silu}()} \cr
 \code{\link{op_hard_tanh}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -311,6 +332,8 @@ Other ops: \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
 \code{\link{op_istft}()} \cr
+\code{\link{op_kaiser}()} \cr
+\code{\link{op_layer_normalization}()} \cr
 \code{\link{op_leaky_relu}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
@@ -410,6 +433,7 @@ Other ops: \cr
 \code{\link{op_sort}()} \cr
 \code{\link{op_sparse_categorical_crossentropy}()} \cr
 \code{\link{op_sparse_plus}()} \cr
+\code{\link{op_sparse_sigmoid}()} \cr
 \code{\link{op_sparsemax}()} \cr
 \code{\link{op_split}()} \cr
 \code{\link{op_sqrt}()} \cr
@@ -446,6 +470,8 @@ Other ops: \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
 \code{\link{op_vectorized_map}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_while_loop}()} \cr
diff --git a/man/op_svd.Rd b/man/op_svd.Rd
index ba8744dc3..aee5088c8 100644
--- a/man/op_svd.Rd
+++ b/man/op_svd.Rd
@@ -48,6 +48,7 @@ Other ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -66,6 +67,7 @@ Other ops: \cr
 \code{\link{op_associative_scan}()} \cr
 \code{\link{op_average}()} \cr
 \code{\link{op_average_pool}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_batch_normalization}()} \cr
 \code{\link{op_binary_crossentropy}()} \cr
 \code{\link{op_bincount}()} \cr
@@ -76,9 +78,11 @@ Other ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
 \code{\link{op_cast}()} \cr
 \code{\link{op_categorical_crossentropy}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_celu}()} \cr
 \code{\link{op_cholesky}()} \cr
@@ -91,6 +95,7 @@ Other ops: \cr
 \code{\link{op_convert_to_numpy}()} \cr
 \code{\link{op_convert_to_tensor}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -101,6 +106,7 @@ Other ops: \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
 \code{\link{op_custom_gradient}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_depthwise_conv}()} \cr
 \code{\link{op_det}()} \cr
 \code{\link{op_diag}()} \cr
@@ -140,10 +146,13 @@ Other ops: \cr
 \code{\link{op_glu}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
 \code{\link{op_hard_shrink}()} \cr
 \code{\link{op_hard_sigmoid}()} \cr
 \code{\link{op_hard_silu}()} \cr
 \code{\link{op_hard_tanh}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -170,6 +179,8 @@ Other ops: \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
 \code{\link{op_istft}()} \cr
+\code{\link{op_kaiser}()} \cr
+\code{\link{op_layer_normalization}()} \cr
 \code{\link{op_leaky_relu}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
@@ -269,6 +280,7 @@ Other ops: \cr
 \code{\link{op_sort}()} \cr
 \code{\link{op_sparse_categorical_crossentropy}()} \cr
 \code{\link{op_sparse_plus}()} \cr
+\code{\link{op_sparse_sigmoid}()} \cr
 \code{\link{op_sparsemax}()} \cr
 \code{\link{op_split}()} \cr
 \code{\link{op_sqrt}()} \cr
@@ -305,6 +317,8 @@ Other ops: \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
 \code{\link{op_vectorized_map}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_while_loop}()} \cr
diff --git a/man/op_swapaxes.Rd b/man/op_swapaxes.Rd
index ef6a234d8..258d5a2ff 100644
--- a/man/op_swapaxes.Rd
+++ b/man/op_swapaxes.Rd
@@ -28,6 +28,7 @@ Other numpy ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -44,6 +45,7 @@ Other numpy ops: \cr
 \code{\link{op_argsort}()} \cr
 \code{\link{op_array}()} \cr
 \code{\link{op_average}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_bincount}()} \cr
 \code{\link{op_bitwise_and}()} \cr
 \code{\link{op_bitwise_invert}()} \cr
@@ -52,12 +54,15 @@ Other numpy ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_clip}()} \cr
 \code{\link{op_concatenate}()} \cr
 \code{\link{op_conj}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -66,6 +71,7 @@ Other numpy ops: \cr
 \code{\link{op_ctc_decode}()} \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_diag}()} \cr
 \code{\link{op_diagflat}()} \cr
 \code{\link{op_diagonal}()} \cr
@@ -90,6 +96,9 @@ Other numpy ops: \cr
 \code{\link{op_get_item}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -99,6 +108,7 @@ Other numpy ops: \cr
 \code{\link{op_isfinite}()} \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
+\code{\link{op_kaiser}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
 \code{\link{op_less_equal}()} \cr
@@ -178,6 +188,8 @@ Other numpy ops: \cr
 \code{\link{op_var}()} \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_zeros}()} \cr
@@ -187,6 +199,7 @@ Other ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -205,6 +218,7 @@ Other ops: \cr
 \code{\link{op_associative_scan}()} \cr
 \code{\link{op_average}()} \cr
 \code{\link{op_average_pool}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_batch_normalization}()} \cr
 \code{\link{op_binary_crossentropy}()} \cr
 \code{\link{op_bincount}()} \cr
@@ -215,9 +229,11 @@ Other ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
 \code{\link{op_cast}()} \cr
 \code{\link{op_categorical_crossentropy}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_celu}()} \cr
 \code{\link{op_cholesky}()} \cr
@@ -230,6 +246,7 @@ Other ops: \cr
 \code{\link{op_convert_to_numpy}()} \cr
 \code{\link{op_convert_to_tensor}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -240,6 +257,7 @@ Other ops: \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
 \code{\link{op_custom_gradient}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_depthwise_conv}()} \cr
 \code{\link{op_det}()} \cr
 \code{\link{op_diag}()} \cr
@@ -279,10 +297,13 @@ Other ops: \cr
 \code{\link{op_glu}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
 \code{\link{op_hard_shrink}()} \cr
 \code{\link{op_hard_sigmoid}()} \cr
 \code{\link{op_hard_silu}()} \cr
 \code{\link{op_hard_tanh}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -309,6 +330,8 @@ Other ops: \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
 \code{\link{op_istft}()} \cr
+\code{\link{op_kaiser}()} \cr
+\code{\link{op_layer_normalization}()} \cr
 \code{\link{op_leaky_relu}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
@@ -408,6 +431,7 @@ Other ops: \cr
 \code{\link{op_sort}()} \cr
 \code{\link{op_sparse_categorical_crossentropy}()} \cr
 \code{\link{op_sparse_plus}()} \cr
+\code{\link{op_sparse_sigmoid}()} \cr
 \code{\link{op_sparsemax}()} \cr
 \code{\link{op_split}()} \cr
 \code{\link{op_sqrt}()} \cr
@@ -444,6 +468,8 @@ Other ops: \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
 \code{\link{op_vectorized_map}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_while_loop}()} \cr
diff --git a/man/op_switch.Rd b/man/op_switch.Rd
index 2f9e5aa74..aa7f11036 100644
--- a/man/op_switch.Rd
+++ b/man/op_switch.Rd
@@ -93,6 +93,7 @@ Other ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -111,6 +112,7 @@ Other ops: \cr
 \code{\link{op_associative_scan}()} \cr
 \code{\link{op_average}()} \cr
 \code{\link{op_average_pool}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_batch_normalization}()} \cr
 \code{\link{op_binary_crossentropy}()} \cr
 \code{\link{op_bincount}()} \cr
@@ -121,9 +123,11 @@ Other ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
 \code{\link{op_cast}()} \cr
 \code{\link{op_categorical_crossentropy}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_celu}()} \cr
 \code{\link{op_cholesky}()} \cr
@@ -136,6 +140,7 @@ Other ops: \cr
 \code{\link{op_convert_to_numpy}()} \cr
 \code{\link{op_convert_to_tensor}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -146,6 +151,7 @@ Other ops: \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
 \code{\link{op_custom_gradient}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_depthwise_conv}()} \cr
 \code{\link{op_det}()} \cr
 \code{\link{op_diag}()} \cr
@@ -185,10 +191,13 @@ Other ops: \cr
 \code{\link{op_glu}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
 \code{\link{op_hard_shrink}()} \cr
 \code{\link{op_hard_sigmoid}()} \cr
 \code{\link{op_hard_silu}()} \cr
 \code{\link{op_hard_tanh}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -215,6 +224,8 @@ Other ops: \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
 \code{\link{op_istft}()} \cr
+\code{\link{op_kaiser}()} \cr
+\code{\link{op_layer_normalization}()} \cr
 \code{\link{op_leaky_relu}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
@@ -314,6 +325,7 @@ Other ops: \cr
 \code{\link{op_sort}()} \cr
 \code{\link{op_sparse_categorical_crossentropy}()} \cr
 \code{\link{op_sparse_plus}()} \cr
+\code{\link{op_sparse_sigmoid}()} \cr
 \code{\link{op_sparsemax}()} \cr
 \code{\link{op_split}()} \cr
 \code{\link{op_sqrt}()} \cr
@@ -350,6 +362,8 @@ Other ops: \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
 \code{\link{op_vectorized_map}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_while_loop}()} \cr
diff --git a/man/op_take.Rd b/man/op_take.Rd
index 272f03cf4..fd232ff78 100644
--- a/man/op_take.Rd
+++ b/man/op_take.Rd
@@ -33,6 +33,7 @@ Other numpy ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -49,6 +50,7 @@ Other numpy ops: \cr
 \code{\link{op_argsort}()} \cr
 \code{\link{op_array}()} \cr
 \code{\link{op_average}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_bincount}()} \cr
 \code{\link{op_bitwise_and}()} \cr
 \code{\link{op_bitwise_invert}()} \cr
@@ -57,12 +59,15 @@ Other numpy ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_clip}()} \cr
 \code{\link{op_concatenate}()} \cr
 \code{\link{op_conj}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -71,6 +76,7 @@ Other numpy ops: \cr
 \code{\link{op_ctc_decode}()} \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_diag}()} \cr
 \code{\link{op_diagflat}()} \cr
 \code{\link{op_diagonal}()} \cr
@@ -95,6 +101,9 @@ Other numpy ops: \cr
 \code{\link{op_get_item}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -104,6 +113,7 @@ Other numpy ops: \cr
 \code{\link{op_isfinite}()} \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
+\code{\link{op_kaiser}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
 \code{\link{op_less_equal}()} \cr
@@ -183,6 +193,8 @@ Other numpy ops: \cr
 \code{\link{op_var}()} \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_zeros}()} \cr
@@ -192,6 +204,7 @@ Other ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -210,6 +223,7 @@ Other ops: \cr
 \code{\link{op_associative_scan}()} \cr
 \code{\link{op_average}()} \cr
 \code{\link{op_average_pool}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_batch_normalization}()} \cr
 \code{\link{op_binary_crossentropy}()} \cr
 \code{\link{op_bincount}()} \cr
@@ -220,9 +234,11 @@ Other ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
 \code{\link{op_cast}()} \cr
 \code{\link{op_categorical_crossentropy}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_celu}()} \cr
 \code{\link{op_cholesky}()} \cr
@@ -235,6 +251,7 @@ Other ops: \cr
 \code{\link{op_convert_to_numpy}()} \cr
 \code{\link{op_convert_to_tensor}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -245,6 +262,7 @@ Other ops: \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
 \code{\link{op_custom_gradient}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_depthwise_conv}()} \cr
 \code{\link{op_det}()} \cr
 \code{\link{op_diag}()} \cr
@@ -284,10 +302,13 @@ Other ops: \cr
 \code{\link{op_glu}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
 \code{\link{op_hard_shrink}()} \cr
 \code{\link{op_hard_sigmoid}()} \cr
 \code{\link{op_hard_silu}()} \cr
 \code{\link{op_hard_tanh}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -314,6 +335,8 @@ Other ops: \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
 \code{\link{op_istft}()} \cr
+\code{\link{op_kaiser}()} \cr
+\code{\link{op_layer_normalization}()} \cr
 \code{\link{op_leaky_relu}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
@@ -413,6 +436,7 @@ Other ops: \cr
 \code{\link{op_sort}()} \cr
 \code{\link{op_sparse_categorical_crossentropy}()} \cr
 \code{\link{op_sparse_plus}()} \cr
+\code{\link{op_sparse_sigmoid}()} \cr
 \code{\link{op_sparsemax}()} \cr
 \code{\link{op_split}()} \cr
 \code{\link{op_sqrt}()} \cr
@@ -449,6 +473,8 @@ Other ops: \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
 \code{\link{op_vectorized_map}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_while_loop}()} \cr
diff --git a/man/op_take_along_axis.Rd b/man/op_take_along_axis.Rd
index b51fa560c..979225afd 100644
--- a/man/op_take_along_axis.Rd
+++ b/man/op_take_along_axis.Rd
@@ -33,6 +33,7 @@ Other numpy ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -49,6 +50,7 @@ Other numpy ops: \cr
 \code{\link{op_argsort}()} \cr
 \code{\link{op_array}()} \cr
 \code{\link{op_average}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_bincount}()} \cr
 \code{\link{op_bitwise_and}()} \cr
 \code{\link{op_bitwise_invert}()} \cr
@@ -57,12 +59,15 @@ Other numpy ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_clip}()} \cr
 \code{\link{op_concatenate}()} \cr
 \code{\link{op_conj}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -71,6 +76,7 @@ Other numpy ops: \cr
 \code{\link{op_ctc_decode}()} \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_diag}()} \cr
 \code{\link{op_diagflat}()} \cr
 \code{\link{op_diagonal}()} \cr
@@ -95,6 +101,9 @@ Other numpy ops: \cr
 \code{\link{op_get_item}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -104,6 +113,7 @@ Other numpy ops: \cr
 \code{\link{op_isfinite}()} \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
+\code{\link{op_kaiser}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
 \code{\link{op_less_equal}()} \cr
@@ -183,6 +193,8 @@ Other numpy ops: \cr
 \code{\link{op_var}()} \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_zeros}()} \cr
@@ -192,6 +204,7 @@ Other ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -210,6 +223,7 @@ Other ops: \cr
 \code{\link{op_associative_scan}()} \cr
 \code{\link{op_average}()} \cr
 \code{\link{op_average_pool}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_batch_normalization}()} \cr
 \code{\link{op_binary_crossentropy}()} \cr
 \code{\link{op_bincount}()} \cr
@@ -220,9 +234,11 @@ Other ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
 \code{\link{op_cast}()} \cr
 \code{\link{op_categorical_crossentropy}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_celu}()} \cr
 \code{\link{op_cholesky}()} \cr
@@ -235,6 +251,7 @@ Other ops: \cr
 \code{\link{op_convert_to_numpy}()} \cr
 \code{\link{op_convert_to_tensor}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -245,6 +262,7 @@ Other ops: \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
 \code{\link{op_custom_gradient}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_depthwise_conv}()} \cr
 \code{\link{op_det}()} \cr
 \code{\link{op_diag}()} \cr
@@ -284,10 +302,13 @@ Other ops: \cr
 \code{\link{op_glu}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
 \code{\link{op_hard_shrink}()} \cr
 \code{\link{op_hard_sigmoid}()} \cr
 \code{\link{op_hard_silu}()} \cr
 \code{\link{op_hard_tanh}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -314,6 +335,8 @@ Other ops: \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
 \code{\link{op_istft}()} \cr
+\code{\link{op_kaiser}()} \cr
+\code{\link{op_layer_normalization}()} \cr
 \code{\link{op_leaky_relu}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
@@ -413,6 +436,7 @@ Other ops: \cr
 \code{\link{op_sort}()} \cr
 \code{\link{op_sparse_categorical_crossentropy}()} \cr
 \code{\link{op_sparse_plus}()} \cr
+\code{\link{op_sparse_sigmoid}()} \cr
 \code{\link{op_sparsemax}()} \cr
 \code{\link{op_split}()} \cr
 \code{\link{op_sqrt}()} \cr
@@ -449,6 +473,8 @@ Other ops: \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
 \code{\link{op_vectorized_map}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_while_loop}()} \cr
diff --git a/man/op_tan.Rd b/man/op_tan.Rd
index 9ba4bc4eb..78bf86602 100644
--- a/man/op_tan.Rd
+++ b/man/op_tan.Rd
@@ -24,6 +24,7 @@ Other numpy ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -40,6 +41,7 @@ Other numpy ops: \cr
 \code{\link{op_argsort}()} \cr
 \code{\link{op_array}()} \cr
 \code{\link{op_average}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_bincount}()} \cr
 \code{\link{op_bitwise_and}()} \cr
 \code{\link{op_bitwise_invert}()} \cr
@@ -48,12 +50,15 @@ Other numpy ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_clip}()} \cr
 \code{\link{op_concatenate}()} \cr
 \code{\link{op_conj}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -62,6 +67,7 @@ Other numpy ops: \cr
 \code{\link{op_ctc_decode}()} \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_diag}()} \cr
 \code{\link{op_diagflat}()} \cr
 \code{\link{op_diagonal}()} \cr
@@ -86,6 +92,9 @@ Other numpy ops: \cr
 \code{\link{op_get_item}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -95,6 +104,7 @@ Other numpy ops: \cr
 \code{\link{op_isfinite}()} \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
+\code{\link{op_kaiser}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
 \code{\link{op_less_equal}()} \cr
@@ -174,6 +184,8 @@ Other numpy ops: \cr
 \code{\link{op_var}()} \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_zeros}()} \cr
@@ -183,6 +195,7 @@ Other ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -201,6 +214,7 @@ Other ops: \cr
 \code{\link{op_associative_scan}()} \cr
 \code{\link{op_average}()} \cr
 \code{\link{op_average_pool}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_batch_normalization}()} \cr
 \code{\link{op_binary_crossentropy}()} \cr
 \code{\link{op_bincount}()} \cr
@@ -211,9 +225,11 @@ Other ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
 \code{\link{op_cast}()} \cr
 \code{\link{op_categorical_crossentropy}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_celu}()} \cr
 \code{\link{op_cholesky}()} \cr
@@ -226,6 +242,7 @@ Other ops: \cr
 \code{\link{op_convert_to_numpy}()} \cr
 \code{\link{op_convert_to_tensor}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -236,6 +253,7 @@ Other ops: \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
 \code{\link{op_custom_gradient}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_depthwise_conv}()} \cr
 \code{\link{op_det}()} \cr
 \code{\link{op_diag}()} \cr
@@ -275,10 +293,13 @@ Other ops: \cr
 \code{\link{op_glu}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
 \code{\link{op_hard_shrink}()} \cr
 \code{\link{op_hard_sigmoid}()} \cr
 \code{\link{op_hard_silu}()} \cr
 \code{\link{op_hard_tanh}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -305,6 +326,8 @@ Other ops: \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
 \code{\link{op_istft}()} \cr
+\code{\link{op_kaiser}()} \cr
+\code{\link{op_layer_normalization}()} \cr
 \code{\link{op_leaky_relu}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
@@ -404,6 +427,7 @@ Other ops: \cr
 \code{\link{op_sort}()} \cr
 \code{\link{op_sparse_categorical_crossentropy}()} \cr
 \code{\link{op_sparse_plus}()} \cr
+\code{\link{op_sparse_sigmoid}()} \cr
 \code{\link{op_sparsemax}()} \cr
 \code{\link{op_split}()} \cr
 \code{\link{op_sqrt}()} \cr
@@ -440,6 +464,8 @@ Other ops: \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
 \code{\link{op_vectorized_map}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_while_loop}()} \cr
diff --git a/man/op_tanh.Rd b/man/op_tanh.Rd
index d33dafca5..751faca50 100644
--- a/man/op_tanh.Rd
+++ b/man/op_tanh.Rd
@@ -24,6 +24,7 @@ Other numpy ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -40,6 +41,7 @@ Other numpy ops: \cr
 \code{\link{op_argsort}()} \cr
 \code{\link{op_array}()} \cr
 \code{\link{op_average}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_bincount}()} \cr
 \code{\link{op_bitwise_and}()} \cr
 \code{\link{op_bitwise_invert}()} \cr
@@ -48,12 +50,15 @@ Other numpy ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_clip}()} \cr
 \code{\link{op_concatenate}()} \cr
 \code{\link{op_conj}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -62,6 +67,7 @@ Other numpy ops: \cr
 \code{\link{op_ctc_decode}()} \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_diag}()} \cr
 \code{\link{op_diagflat}()} \cr
 \code{\link{op_diagonal}()} \cr
@@ -86,6 +92,9 @@ Other numpy ops: \cr
 \code{\link{op_get_item}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -95,6 +104,7 @@ Other numpy ops: \cr
 \code{\link{op_isfinite}()} \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
+\code{\link{op_kaiser}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
 \code{\link{op_less_equal}()} \cr
@@ -174,6 +184,8 @@ Other numpy ops: \cr
 \code{\link{op_var}()} \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_zeros}()} \cr
@@ -183,6 +195,7 @@ Other ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -201,6 +214,7 @@ Other ops: \cr
 \code{\link{op_associative_scan}()} \cr
 \code{\link{op_average}()} \cr
 \code{\link{op_average_pool}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_batch_normalization}()} \cr
 \code{\link{op_binary_crossentropy}()} \cr
 \code{\link{op_bincount}()} \cr
@@ -211,9 +225,11 @@ Other ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
 \code{\link{op_cast}()} \cr
 \code{\link{op_categorical_crossentropy}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_celu}()} \cr
 \code{\link{op_cholesky}()} \cr
@@ -226,6 +242,7 @@ Other ops: \cr
 \code{\link{op_convert_to_numpy}()} \cr
 \code{\link{op_convert_to_tensor}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -236,6 +253,7 @@ Other ops: \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
 \code{\link{op_custom_gradient}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_depthwise_conv}()} \cr
 \code{\link{op_det}()} \cr
 \code{\link{op_diag}()} \cr
@@ -275,10 +293,13 @@ Other ops: \cr
 \code{\link{op_glu}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
 \code{\link{op_hard_shrink}()} \cr
 \code{\link{op_hard_sigmoid}()} \cr
 \code{\link{op_hard_silu}()} \cr
 \code{\link{op_hard_tanh}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -305,6 +326,8 @@ Other ops: \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
 \code{\link{op_istft}()} \cr
+\code{\link{op_kaiser}()} \cr
+\code{\link{op_layer_normalization}()} \cr
 \code{\link{op_leaky_relu}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
@@ -404,6 +427,7 @@ Other ops: \cr
 \code{\link{op_sort}()} \cr
 \code{\link{op_sparse_categorical_crossentropy}()} \cr
 \code{\link{op_sparse_plus}()} \cr
+\code{\link{op_sparse_sigmoid}()} \cr
 \code{\link{op_sparsemax}()} \cr
 \code{\link{op_split}()} \cr
 \code{\link{op_sqrt}()} \cr
@@ -440,6 +464,8 @@ Other ops: \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
 \code{\link{op_vectorized_map}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_while_loop}()} \cr
diff --git a/man/op_tanh_shrink.Rd b/man/op_tanh_shrink.Rd
index 7be28824a..bf1fd5841 100644
--- a/man/op_tanh_shrink.Rd
+++ b/man/op_tanh_shrink.Rd
@@ -47,6 +47,7 @@ Other nn ops: \cr
 \code{\link{op_hard_sigmoid}()} \cr
 \code{\link{op_hard_silu}()} \cr
 \code{\link{op_hard_tanh}()} \cr
+\code{\link{op_layer_normalization}()} \cr
 \code{\link{op_leaky_relu}()} \cr
 \code{\link{op_log_sigmoid}()} \cr
 \code{\link{op_log_softmax}()} \cr
@@ -70,6 +71,7 @@ Other nn ops: \cr
 \code{\link{op_softsign}()} \cr
 \code{\link{op_sparse_categorical_crossentropy}()} \cr
 \code{\link{op_sparse_plus}()} \cr
+\code{\link{op_sparse_sigmoid}()} \cr
 \code{\link{op_sparsemax}()} \cr
 \code{\link{op_squareplus}()} \cr
 \code{\link{op_threshold}()} \cr
@@ -79,6 +81,7 @@ Other ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -97,6 +100,7 @@ Other ops: \cr
 \code{\link{op_associative_scan}()} \cr
 \code{\link{op_average}()} \cr
 \code{\link{op_average_pool}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_batch_normalization}()} \cr
 \code{\link{op_binary_crossentropy}()} \cr
 \code{\link{op_bincount}()} \cr
@@ -107,9 +111,11 @@ Other ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
 \code{\link{op_cast}()} \cr
 \code{\link{op_categorical_crossentropy}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_celu}()} \cr
 \code{\link{op_cholesky}()} \cr
@@ -122,6 +128,7 @@ Other ops: \cr
 \code{\link{op_convert_to_numpy}()} \cr
 \code{\link{op_convert_to_tensor}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -132,6 +139,7 @@ Other ops: \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
 \code{\link{op_custom_gradient}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_depthwise_conv}()} \cr
 \code{\link{op_det}()} \cr
 \code{\link{op_diag}()} \cr
@@ -171,10 +179,13 @@ Other ops: \cr
 \code{\link{op_glu}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
 \code{\link{op_hard_shrink}()} \cr
 \code{\link{op_hard_sigmoid}()} \cr
 \code{\link{op_hard_silu}()} \cr
 \code{\link{op_hard_tanh}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -201,6 +212,8 @@ Other ops: \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
 \code{\link{op_istft}()} \cr
+\code{\link{op_kaiser}()} \cr
+\code{\link{op_layer_normalization}()} \cr
 \code{\link{op_leaky_relu}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
@@ -300,6 +313,7 @@ Other ops: \cr
 \code{\link{op_sort}()} \cr
 \code{\link{op_sparse_categorical_crossentropy}()} \cr
 \code{\link{op_sparse_plus}()} \cr
+\code{\link{op_sparse_sigmoid}()} \cr
 \code{\link{op_sparsemax}()} \cr
 \code{\link{op_split}()} \cr
 \code{\link{op_sqrt}()} \cr
@@ -336,6 +350,8 @@ Other ops: \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
 \code{\link{op_vectorized_map}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_while_loop}()} \cr
diff --git a/man/op_tensordot.Rd b/man/op_tensordot.Rd
index 5eb601d32..0c139941a 100644
--- a/man/op_tensordot.Rd
+++ b/man/op_tensordot.Rd
@@ -35,6 +35,7 @@ Other numpy ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -51,6 +52,7 @@ Other numpy ops: \cr
 \code{\link{op_argsort}()} \cr
 \code{\link{op_array}()} \cr
 \code{\link{op_average}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_bincount}()} \cr
 \code{\link{op_bitwise_and}()} \cr
 \code{\link{op_bitwise_invert}()} \cr
@@ -59,12 +61,15 @@ Other numpy ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_clip}()} \cr
 \code{\link{op_concatenate}()} \cr
 \code{\link{op_conj}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -73,6 +78,7 @@ Other numpy ops: \cr
 \code{\link{op_ctc_decode}()} \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_diag}()} \cr
 \code{\link{op_diagflat}()} \cr
 \code{\link{op_diagonal}()} \cr
@@ -97,6 +103,9 @@ Other numpy ops: \cr
 \code{\link{op_get_item}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -106,6 +115,7 @@ Other numpy ops: \cr
 \code{\link{op_isfinite}()} \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
+\code{\link{op_kaiser}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
 \code{\link{op_less_equal}()} \cr
@@ -185,6 +195,8 @@ Other numpy ops: \cr
 \code{\link{op_var}()} \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_zeros}()} \cr
@@ -194,6 +206,7 @@ Other ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -212,6 +225,7 @@ Other ops: \cr
 \code{\link{op_associative_scan}()} \cr
 \code{\link{op_average}()} \cr
 \code{\link{op_average_pool}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_batch_normalization}()} \cr
 \code{\link{op_binary_crossentropy}()} \cr
 \code{\link{op_bincount}()} \cr
@@ -222,9 +236,11 @@ Other ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
 \code{\link{op_cast}()} \cr
 \code{\link{op_categorical_crossentropy}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_celu}()} \cr
 \code{\link{op_cholesky}()} \cr
@@ -237,6 +253,7 @@ Other ops: \cr
 \code{\link{op_convert_to_numpy}()} \cr
 \code{\link{op_convert_to_tensor}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -247,6 +264,7 @@ Other ops: \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
 \code{\link{op_custom_gradient}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_depthwise_conv}()} \cr
 \code{\link{op_det}()} \cr
 \code{\link{op_diag}()} \cr
@@ -286,10 +304,13 @@ Other ops: \cr
 \code{\link{op_glu}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
 \code{\link{op_hard_shrink}()} \cr
 \code{\link{op_hard_sigmoid}()} \cr
 \code{\link{op_hard_silu}()} \cr
 \code{\link{op_hard_tanh}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -316,6 +337,8 @@ Other ops: \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
 \code{\link{op_istft}()} \cr
+\code{\link{op_kaiser}()} \cr
+\code{\link{op_layer_normalization}()} \cr
 \code{\link{op_leaky_relu}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
@@ -415,6 +438,7 @@ Other ops: \cr
 \code{\link{op_sort}()} \cr
 \code{\link{op_sparse_categorical_crossentropy}()} \cr
 \code{\link{op_sparse_plus}()} \cr
+\code{\link{op_sparse_sigmoid}()} \cr
 \code{\link{op_sparsemax}()} \cr
 \code{\link{op_split}()} \cr
 \code{\link{op_sqrt}()} \cr
@@ -451,6 +475,8 @@ Other ops: \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
 \code{\link{op_vectorized_map}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_while_loop}()} \cr
diff --git a/man/op_threshold.Rd b/man/op_threshold.Rd
index 5a0b5378b..2422ed528 100644
--- a/man/op_threshold.Rd
+++ b/man/op_threshold.Rd
@@ -50,6 +50,7 @@ Other nn ops: \cr
 \code{\link{op_hard_sigmoid}()} \cr
 \code{\link{op_hard_silu}()} \cr
 \code{\link{op_hard_tanh}()} \cr
+\code{\link{op_layer_normalization}()} \cr
 \code{\link{op_leaky_relu}()} \cr
 \code{\link{op_log_sigmoid}()} \cr
 \code{\link{op_log_softmax}()} \cr
@@ -73,6 +74,7 @@ Other nn ops: \cr
 \code{\link{op_softsign}()} \cr
 \code{\link{op_sparse_categorical_crossentropy}()} \cr
 \code{\link{op_sparse_plus}()} \cr
+\code{\link{op_sparse_sigmoid}()} \cr
 \code{\link{op_sparsemax}()} \cr
 \code{\link{op_squareplus}()} \cr
 \code{\link{op_tanh_shrink}()} \cr
@@ -82,6 +84,7 @@ Other ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -100,6 +103,7 @@ Other ops: \cr
 \code{\link{op_associative_scan}()} \cr
 \code{\link{op_average}()} \cr
 \code{\link{op_average_pool}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_batch_normalization}()} \cr
 \code{\link{op_binary_crossentropy}()} \cr
 \code{\link{op_bincount}()} \cr
@@ -110,9 +114,11 @@ Other ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
 \code{\link{op_cast}()} \cr
 \code{\link{op_categorical_crossentropy}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_celu}()} \cr
 \code{\link{op_cholesky}()} \cr
@@ -125,6 +131,7 @@ Other ops: \cr
 \code{\link{op_convert_to_numpy}()} \cr
 \code{\link{op_convert_to_tensor}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -135,6 +142,7 @@ Other ops: \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
 \code{\link{op_custom_gradient}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_depthwise_conv}()} \cr
 \code{\link{op_det}()} \cr
 \code{\link{op_diag}()} \cr
@@ -174,10 +182,13 @@ Other ops: \cr
 \code{\link{op_glu}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
 \code{\link{op_hard_shrink}()} \cr
 \code{\link{op_hard_sigmoid}()} \cr
 \code{\link{op_hard_silu}()} \cr
 \code{\link{op_hard_tanh}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -204,6 +215,8 @@ Other ops: \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
 \code{\link{op_istft}()} \cr
+\code{\link{op_kaiser}()} \cr
+\code{\link{op_layer_normalization}()} \cr
 \code{\link{op_leaky_relu}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
@@ -303,6 +316,7 @@ Other ops: \cr
 \code{\link{op_sort}()} \cr
 \code{\link{op_sparse_categorical_crossentropy}()} \cr
 \code{\link{op_sparse_plus}()} \cr
+\code{\link{op_sparse_sigmoid}()} \cr
 \code{\link{op_sparsemax}()} \cr
 \code{\link{op_split}()} \cr
 \code{\link{op_sqrt}()} \cr
@@ -339,6 +353,8 @@ Other ops: \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
 \code{\link{op_vectorized_map}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_while_loop}()} \cr
diff --git a/man/op_tile.Rd b/man/op_tile.Rd
index f8084f9a4..f15d9fbc4 100644
--- a/man/op_tile.Rd
+++ b/man/op_tile.Rd
@@ -32,6 +32,7 @@ Other numpy ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -48,6 +49,7 @@ Other numpy ops: \cr
 \code{\link{op_argsort}()} \cr
 \code{\link{op_array}()} \cr
 \code{\link{op_average}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_bincount}()} \cr
 \code{\link{op_bitwise_and}()} \cr
 \code{\link{op_bitwise_invert}()} \cr
@@ -56,12 +58,15 @@ Other numpy ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_clip}()} \cr
 \code{\link{op_concatenate}()} \cr
 \code{\link{op_conj}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -70,6 +75,7 @@ Other numpy ops: \cr
 \code{\link{op_ctc_decode}()} \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_diag}()} \cr
 \code{\link{op_diagflat}()} \cr
 \code{\link{op_diagonal}()} \cr
@@ -94,6 +100,9 @@ Other numpy ops: \cr
 \code{\link{op_get_item}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -103,6 +112,7 @@ Other numpy ops: \cr
 \code{\link{op_isfinite}()} \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
+\code{\link{op_kaiser}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
 \code{\link{op_less_equal}()} \cr
@@ -182,6 +192,8 @@ Other numpy ops: \cr
 \code{\link{op_var}()} \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_zeros}()} \cr
@@ -191,6 +203,7 @@ Other ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -209,6 +222,7 @@ Other ops: \cr
 \code{\link{op_associative_scan}()} \cr
 \code{\link{op_average}()} \cr
 \code{\link{op_average_pool}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_batch_normalization}()} \cr
 \code{\link{op_binary_crossentropy}()} \cr
 \code{\link{op_bincount}()} \cr
@@ -219,9 +233,11 @@ Other ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
 \code{\link{op_cast}()} \cr
 \code{\link{op_categorical_crossentropy}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_celu}()} \cr
 \code{\link{op_cholesky}()} \cr
@@ -234,6 +250,7 @@ Other ops: \cr
 \code{\link{op_convert_to_numpy}()} \cr
 \code{\link{op_convert_to_tensor}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -244,6 +261,7 @@ Other ops: \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
 \code{\link{op_custom_gradient}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_depthwise_conv}()} \cr
 \code{\link{op_det}()} \cr
 \code{\link{op_diag}()} \cr
@@ -283,10 +301,13 @@ Other ops: \cr
 \code{\link{op_glu}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
 \code{\link{op_hard_shrink}()} \cr
 \code{\link{op_hard_sigmoid}()} \cr
 \code{\link{op_hard_silu}()} \cr
 \code{\link{op_hard_tanh}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -313,6 +334,8 @@ Other ops: \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
 \code{\link{op_istft}()} \cr
+\code{\link{op_kaiser}()} \cr
+\code{\link{op_layer_normalization}()} \cr
 \code{\link{op_leaky_relu}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
@@ -412,6 +435,7 @@ Other ops: \cr
 \code{\link{op_sort}()} \cr
 \code{\link{op_sparse_categorical_crossentropy}()} \cr
 \code{\link{op_sparse_plus}()} \cr
+\code{\link{op_sparse_sigmoid}()} \cr
 \code{\link{op_sparsemax}()} \cr
 \code{\link{op_split}()} \cr
 \code{\link{op_sqrt}()} \cr
@@ -448,6 +472,8 @@ Other ops: \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
 \code{\link{op_vectorized_map}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_while_loop}()} \cr
diff --git a/man/op_top_k.Rd b/man/op_top_k.Rd
index 74cffb9e9..e3a9f336f 100644
--- a/man/op_top_k.Rd
+++ b/man/op_top_k.Rd
@@ -83,6 +83,7 @@ Other ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -101,6 +102,7 @@ Other ops: \cr
 \code{\link{op_associative_scan}()} \cr
 \code{\link{op_average}()} \cr
 \code{\link{op_average_pool}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_batch_normalization}()} \cr
 \code{\link{op_binary_crossentropy}()} \cr
 \code{\link{op_bincount}()} \cr
@@ -111,9 +113,11 @@ Other ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
 \code{\link{op_cast}()} \cr
 \code{\link{op_categorical_crossentropy}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_celu}()} \cr
 \code{\link{op_cholesky}()} \cr
@@ -126,6 +130,7 @@ Other ops: \cr
 \code{\link{op_convert_to_numpy}()} \cr
 \code{\link{op_convert_to_tensor}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -136,6 +141,7 @@ Other ops: \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
 \code{\link{op_custom_gradient}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_depthwise_conv}()} \cr
 \code{\link{op_det}()} \cr
 \code{\link{op_diag}()} \cr
@@ -175,10 +181,13 @@ Other ops: \cr
 \code{\link{op_glu}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
 \code{\link{op_hard_shrink}()} \cr
 \code{\link{op_hard_sigmoid}()} \cr
 \code{\link{op_hard_silu}()} \cr
 \code{\link{op_hard_tanh}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -205,6 +214,8 @@ Other ops: \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
 \code{\link{op_istft}()} \cr
+\code{\link{op_kaiser}()} \cr
+\code{\link{op_layer_normalization}()} \cr
 \code{\link{op_leaky_relu}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
@@ -304,6 +315,7 @@ Other ops: \cr
 \code{\link{op_sort}()} \cr
 \code{\link{op_sparse_categorical_crossentropy}()} \cr
 \code{\link{op_sparse_plus}()} \cr
+\code{\link{op_sparse_sigmoid}()} \cr
 \code{\link{op_sparsemax}()} \cr
 \code{\link{op_split}()} \cr
 \code{\link{op_sqrt}()} \cr
@@ -340,6 +352,8 @@ Other ops: \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
 \code{\link{op_vectorized_map}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_while_loop}()} \cr
diff --git a/man/op_trace.Rd b/man/op_trace.Rd
index 02661d34d..b7aaf632a 100644
--- a/man/op_trace.Rd
+++ b/man/op_trace.Rd
@@ -43,6 +43,7 @@ Other numpy ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -59,6 +60,7 @@ Other numpy ops: \cr
 \code{\link{op_argsort}()} \cr
 \code{\link{op_array}()} \cr
 \code{\link{op_average}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_bincount}()} \cr
 \code{\link{op_bitwise_and}()} \cr
 \code{\link{op_bitwise_invert}()} \cr
@@ -67,12 +69,15 @@ Other numpy ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_clip}()} \cr
 \code{\link{op_concatenate}()} \cr
 \code{\link{op_conj}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -81,6 +86,7 @@ Other numpy ops: \cr
 \code{\link{op_ctc_decode}()} \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_diag}()} \cr
 \code{\link{op_diagflat}()} \cr
 \code{\link{op_diagonal}()} \cr
@@ -105,6 +111,9 @@ Other numpy ops: \cr
 \code{\link{op_get_item}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -114,6 +123,7 @@ Other numpy ops: \cr
 \code{\link{op_isfinite}()} \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
+\code{\link{op_kaiser}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
 \code{\link{op_less_equal}()} \cr
@@ -193,6 +203,8 @@ Other numpy ops: \cr
 \code{\link{op_var}()} \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_zeros}()} \cr
@@ -202,6 +214,7 @@ Other ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -220,6 +233,7 @@ Other ops: \cr
 \code{\link{op_associative_scan}()} \cr
 \code{\link{op_average}()} \cr
 \code{\link{op_average_pool}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_batch_normalization}()} \cr
 \code{\link{op_binary_crossentropy}()} \cr
 \code{\link{op_bincount}()} \cr
@@ -230,9 +244,11 @@ Other ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
 \code{\link{op_cast}()} \cr
 \code{\link{op_categorical_crossentropy}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_celu}()} \cr
 \code{\link{op_cholesky}()} \cr
@@ -245,6 +261,7 @@ Other ops: \cr
 \code{\link{op_convert_to_numpy}()} \cr
 \code{\link{op_convert_to_tensor}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -255,6 +272,7 @@ Other ops: \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
 \code{\link{op_custom_gradient}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_depthwise_conv}()} \cr
 \code{\link{op_det}()} \cr
 \code{\link{op_diag}()} \cr
@@ -294,10 +312,13 @@ Other ops: \cr
 \code{\link{op_glu}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
 \code{\link{op_hard_shrink}()} \cr
 \code{\link{op_hard_sigmoid}()} \cr
 \code{\link{op_hard_silu}()} \cr
 \code{\link{op_hard_tanh}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -324,6 +345,8 @@ Other ops: \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
 \code{\link{op_istft}()} \cr
+\code{\link{op_kaiser}()} \cr
+\code{\link{op_layer_normalization}()} \cr
 \code{\link{op_leaky_relu}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
@@ -423,6 +446,7 @@ Other ops: \cr
 \code{\link{op_sort}()} \cr
 \code{\link{op_sparse_categorical_crossentropy}()} \cr
 \code{\link{op_sparse_plus}()} \cr
+\code{\link{op_sparse_sigmoid}()} \cr
 \code{\link{op_sparsemax}()} \cr
 \code{\link{op_split}()} \cr
 \code{\link{op_sqrt}()} \cr
@@ -459,6 +483,8 @@ Other ops: \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
 \code{\link{op_vectorized_map}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_while_loop}()} \cr
diff --git a/man/op_transpose.Rd b/man/op_transpose.Rd
index ebc0b2bf8..86347fc11 100644
--- a/man/op_transpose.Rd
+++ b/man/op_transpose.Rd
@@ -27,6 +27,7 @@ Other numpy ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -43,6 +44,7 @@ Other numpy ops: \cr
 \code{\link{op_argsort}()} \cr
 \code{\link{op_array}()} \cr
 \code{\link{op_average}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_bincount}()} \cr
 \code{\link{op_bitwise_and}()} \cr
 \code{\link{op_bitwise_invert}()} \cr
@@ -51,12 +53,15 @@ Other numpy ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_clip}()} \cr
 \code{\link{op_concatenate}()} \cr
 \code{\link{op_conj}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -65,6 +70,7 @@ Other numpy ops: \cr
 \code{\link{op_ctc_decode}()} \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_diag}()} \cr
 \code{\link{op_diagflat}()} \cr
 \code{\link{op_diagonal}()} \cr
@@ -89,6 +95,9 @@ Other numpy ops: \cr
 \code{\link{op_get_item}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -98,6 +107,7 @@ Other numpy ops: \cr
 \code{\link{op_isfinite}()} \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
+\code{\link{op_kaiser}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
 \code{\link{op_less_equal}()} \cr
@@ -177,6 +187,8 @@ Other numpy ops: \cr
 \code{\link{op_var}()} \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_zeros}()} \cr
@@ -186,6 +198,7 @@ Other ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -204,6 +217,7 @@ Other ops: \cr
 \code{\link{op_associative_scan}()} \cr
 \code{\link{op_average}()} \cr
 \code{\link{op_average_pool}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_batch_normalization}()} \cr
 \code{\link{op_binary_crossentropy}()} \cr
 \code{\link{op_bincount}()} \cr
@@ -214,9 +228,11 @@ Other ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
 \code{\link{op_cast}()} \cr
 \code{\link{op_categorical_crossentropy}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_celu}()} \cr
 \code{\link{op_cholesky}()} \cr
@@ -229,6 +245,7 @@ Other ops: \cr
 \code{\link{op_convert_to_numpy}()} \cr
 \code{\link{op_convert_to_tensor}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -239,6 +256,7 @@ Other ops: \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
 \code{\link{op_custom_gradient}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_depthwise_conv}()} \cr
 \code{\link{op_det}()} \cr
 \code{\link{op_diag}()} \cr
@@ -278,10 +296,13 @@ Other ops: \cr
 \code{\link{op_glu}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
 \code{\link{op_hard_shrink}()} \cr
 \code{\link{op_hard_sigmoid}()} \cr
 \code{\link{op_hard_silu}()} \cr
 \code{\link{op_hard_tanh}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -308,6 +329,8 @@ Other ops: \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
 \code{\link{op_istft}()} \cr
+\code{\link{op_kaiser}()} \cr
+\code{\link{op_layer_normalization}()} \cr
 \code{\link{op_leaky_relu}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
@@ -407,6 +430,7 @@ Other ops: \cr
 \code{\link{op_sort}()} \cr
 \code{\link{op_sparse_categorical_crossentropy}()} \cr
 \code{\link{op_sparse_plus}()} \cr
+\code{\link{op_sparse_sigmoid}()} \cr
 \code{\link{op_sparsemax}()} \cr
 \code{\link{op_split}()} \cr
 \code{\link{op_sqrt}()} \cr
@@ -443,6 +467,8 @@ Other ops: \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
 \code{\link{op_vectorized_map}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_while_loop}()} \cr
diff --git a/man/op_tri.Rd b/man/op_tri.Rd
index 8aa284a56..2bccac0bf 100644
--- a/man/op_tri.Rd
+++ b/man/op_tri.Rd
@@ -33,6 +33,7 @@ Other numpy ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -49,6 +50,7 @@ Other numpy ops: \cr
 \code{\link{op_argsort}()} \cr
 \code{\link{op_array}()} \cr
 \code{\link{op_average}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_bincount}()} \cr
 \code{\link{op_bitwise_and}()} \cr
 \code{\link{op_bitwise_invert}()} \cr
@@ -57,12 +59,15 @@ Other numpy ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_clip}()} \cr
 \code{\link{op_concatenate}()} \cr
 \code{\link{op_conj}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -71,6 +76,7 @@ Other numpy ops: \cr
 \code{\link{op_ctc_decode}()} \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_diag}()} \cr
 \code{\link{op_diagflat}()} \cr
 \code{\link{op_diagonal}()} \cr
@@ -95,6 +101,9 @@ Other numpy ops: \cr
 \code{\link{op_get_item}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -104,6 +113,7 @@ Other numpy ops: \cr
 \code{\link{op_isfinite}()} \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
+\code{\link{op_kaiser}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
 \code{\link{op_less_equal}()} \cr
@@ -183,6 +193,8 @@ Other numpy ops: \cr
 \code{\link{op_var}()} \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_zeros}()} \cr
@@ -192,6 +204,7 @@ Other ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -210,6 +223,7 @@ Other ops: \cr
 \code{\link{op_associative_scan}()} \cr
 \code{\link{op_average}()} \cr
 \code{\link{op_average_pool}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_batch_normalization}()} \cr
 \code{\link{op_binary_crossentropy}()} \cr
 \code{\link{op_bincount}()} \cr
@@ -220,9 +234,11 @@ Other ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
 \code{\link{op_cast}()} \cr
 \code{\link{op_categorical_crossentropy}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_celu}()} \cr
 \code{\link{op_cholesky}()} \cr
@@ -235,6 +251,7 @@ Other ops: \cr
 \code{\link{op_convert_to_numpy}()} \cr
 \code{\link{op_convert_to_tensor}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -245,6 +262,7 @@ Other ops: \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
 \code{\link{op_custom_gradient}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_depthwise_conv}()} \cr
 \code{\link{op_det}()} \cr
 \code{\link{op_diag}()} \cr
@@ -284,10 +302,13 @@ Other ops: \cr
 \code{\link{op_glu}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
 \code{\link{op_hard_shrink}()} \cr
 \code{\link{op_hard_sigmoid}()} \cr
 \code{\link{op_hard_silu}()} \cr
 \code{\link{op_hard_tanh}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -314,6 +335,8 @@ Other ops: \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
 \code{\link{op_istft}()} \cr
+\code{\link{op_kaiser}()} \cr
+\code{\link{op_layer_normalization}()} \cr
 \code{\link{op_leaky_relu}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
@@ -413,6 +436,7 @@ Other ops: \cr
 \code{\link{op_sort}()} \cr
 \code{\link{op_sparse_categorical_crossentropy}()} \cr
 \code{\link{op_sparse_plus}()} \cr
+\code{\link{op_sparse_sigmoid}()} \cr
 \code{\link{op_sparsemax}()} \cr
 \code{\link{op_split}()} \cr
 \code{\link{op_sqrt}()} \cr
@@ -449,6 +473,8 @@ Other ops: \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
 \code{\link{op_vectorized_map}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_while_loop}()} \cr
diff --git a/man/op_tril.Rd b/man/op_tril.Rd
index 16b8e8fe1..86c74ce15 100644
--- a/man/op_tril.Rd
+++ b/man/op_tril.Rd
@@ -28,6 +28,7 @@ Other numpy ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -44,6 +45,7 @@ Other numpy ops: \cr
 \code{\link{op_argsort}()} \cr
 \code{\link{op_array}()} \cr
 \code{\link{op_average}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_bincount}()} \cr
 \code{\link{op_bitwise_and}()} \cr
 \code{\link{op_bitwise_invert}()} \cr
@@ -52,12 +54,15 @@ Other numpy ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_clip}()} \cr
 \code{\link{op_concatenate}()} \cr
 \code{\link{op_conj}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -66,6 +71,7 @@ Other numpy ops: \cr
 \code{\link{op_ctc_decode}()} \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_diag}()} \cr
 \code{\link{op_diagflat}()} \cr
 \code{\link{op_diagonal}()} \cr
@@ -90,6 +96,9 @@ Other numpy ops: \cr
 \code{\link{op_get_item}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -99,6 +108,7 @@ Other numpy ops: \cr
 \code{\link{op_isfinite}()} \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
+\code{\link{op_kaiser}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
 \code{\link{op_less_equal}()} \cr
@@ -178,6 +188,8 @@ Other numpy ops: \cr
 \code{\link{op_var}()} \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_zeros}()} \cr
@@ -187,6 +199,7 @@ Other ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -205,6 +218,7 @@ Other ops: \cr
 \code{\link{op_associative_scan}()} \cr
 \code{\link{op_average}()} \cr
 \code{\link{op_average_pool}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_batch_normalization}()} \cr
 \code{\link{op_binary_crossentropy}()} \cr
 \code{\link{op_bincount}()} \cr
@@ -215,9 +229,11 @@ Other ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
 \code{\link{op_cast}()} \cr
 \code{\link{op_categorical_crossentropy}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_celu}()} \cr
 \code{\link{op_cholesky}()} \cr
@@ -230,6 +246,7 @@ Other ops: \cr
 \code{\link{op_convert_to_numpy}()} \cr
 \code{\link{op_convert_to_tensor}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -240,6 +257,7 @@ Other ops: \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
 \code{\link{op_custom_gradient}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_depthwise_conv}()} \cr
 \code{\link{op_det}()} \cr
 \code{\link{op_diag}()} \cr
@@ -279,10 +297,13 @@ Other ops: \cr
 \code{\link{op_glu}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
 \code{\link{op_hard_shrink}()} \cr
 \code{\link{op_hard_sigmoid}()} \cr
 \code{\link{op_hard_silu}()} \cr
 \code{\link{op_hard_tanh}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -309,6 +330,8 @@ Other ops: \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
 \code{\link{op_istft}()} \cr
+\code{\link{op_kaiser}()} \cr
+\code{\link{op_layer_normalization}()} \cr
 \code{\link{op_leaky_relu}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
@@ -408,6 +431,7 @@ Other ops: \cr
 \code{\link{op_sort}()} \cr
 \code{\link{op_sparse_categorical_crossentropy}()} \cr
 \code{\link{op_sparse_plus}()} \cr
+\code{\link{op_sparse_sigmoid}()} \cr
 \code{\link{op_sparsemax}()} \cr
 \code{\link{op_split}()} \cr
 \code{\link{op_sqrt}()} \cr
@@ -444,6 +468,8 @@ Other ops: \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
 \code{\link{op_vectorized_map}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_while_loop}()} \cr
diff --git a/man/op_triu.Rd b/man/op_triu.Rd
index 94cd94d64..b18a94663 100644
--- a/man/op_triu.Rd
+++ b/man/op_triu.Rd
@@ -28,6 +28,7 @@ Other numpy ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -44,6 +45,7 @@ Other numpy ops: \cr
 \code{\link{op_argsort}()} \cr
 \code{\link{op_array}()} \cr
 \code{\link{op_average}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_bincount}()} \cr
 \code{\link{op_bitwise_and}()} \cr
 \code{\link{op_bitwise_invert}()} \cr
@@ -52,12 +54,15 @@ Other numpy ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_clip}()} \cr
 \code{\link{op_concatenate}()} \cr
 \code{\link{op_conj}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -66,6 +71,7 @@ Other numpy ops: \cr
 \code{\link{op_ctc_decode}()} \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_diag}()} \cr
 \code{\link{op_diagflat}()} \cr
 \code{\link{op_diagonal}()} \cr
@@ -90,6 +96,9 @@ Other numpy ops: \cr
 \code{\link{op_get_item}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -99,6 +108,7 @@ Other numpy ops: \cr
 \code{\link{op_isfinite}()} \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
+\code{\link{op_kaiser}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
 \code{\link{op_less_equal}()} \cr
@@ -178,6 +188,8 @@ Other numpy ops: \cr
 \code{\link{op_var}()} \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_zeros}()} \cr
@@ -187,6 +199,7 @@ Other ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -205,6 +218,7 @@ Other ops: \cr
 \code{\link{op_associative_scan}()} \cr
 \code{\link{op_average}()} \cr
 \code{\link{op_average_pool}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_batch_normalization}()} \cr
 \code{\link{op_binary_crossentropy}()} \cr
 \code{\link{op_bincount}()} \cr
@@ -215,9 +229,11 @@ Other ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
 \code{\link{op_cast}()} \cr
 \code{\link{op_categorical_crossentropy}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_celu}()} \cr
 \code{\link{op_cholesky}()} \cr
@@ -230,6 +246,7 @@ Other ops: \cr
 \code{\link{op_convert_to_numpy}()} \cr
 \code{\link{op_convert_to_tensor}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -240,6 +257,7 @@ Other ops: \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
 \code{\link{op_custom_gradient}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_depthwise_conv}()} \cr
 \code{\link{op_det}()} \cr
 \code{\link{op_diag}()} \cr
@@ -279,10 +297,13 @@ Other ops: \cr
 \code{\link{op_glu}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
 \code{\link{op_hard_shrink}()} \cr
 \code{\link{op_hard_sigmoid}()} \cr
 \code{\link{op_hard_silu}()} \cr
 \code{\link{op_hard_tanh}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -309,6 +330,8 @@ Other ops: \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
 \code{\link{op_istft}()} \cr
+\code{\link{op_kaiser}()} \cr
+\code{\link{op_layer_normalization}()} \cr
 \code{\link{op_leaky_relu}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
@@ -408,6 +431,7 @@ Other ops: \cr
 \code{\link{op_sort}()} \cr
 \code{\link{op_sparse_categorical_crossentropy}()} \cr
 \code{\link{op_sparse_plus}()} \cr
+\code{\link{op_sparse_sigmoid}()} \cr
 \code{\link{op_sparsemax}()} \cr
 \code{\link{op_split}()} \cr
 \code{\link{op_sqrt}()} \cr
@@ -444,6 +468,8 @@ Other ops: \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
 \code{\link{op_vectorized_map}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_while_loop}()} \cr
diff --git a/man/op_trunc.Rd b/man/op_trunc.Rd
index bc3b980a5..e98b10eae 100644
--- a/man/op_trunc.Rd
+++ b/man/op_trunc.Rd
@@ -32,6 +32,7 @@ Other numpy ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -48,6 +49,7 @@ Other numpy ops: \cr
 \code{\link{op_argsort}()} \cr
 \code{\link{op_array}()} \cr
 \code{\link{op_average}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_bincount}()} \cr
 \code{\link{op_bitwise_and}()} \cr
 \code{\link{op_bitwise_invert}()} \cr
@@ -56,12 +58,15 @@ Other numpy ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_clip}()} \cr
 \code{\link{op_concatenate}()} \cr
 \code{\link{op_conj}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -70,6 +75,7 @@ Other numpy ops: \cr
 \code{\link{op_ctc_decode}()} \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_diag}()} \cr
 \code{\link{op_diagflat}()} \cr
 \code{\link{op_diagonal}()} \cr
@@ -94,6 +100,9 @@ Other numpy ops: \cr
 \code{\link{op_get_item}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -103,6 +112,7 @@ Other numpy ops: \cr
 \code{\link{op_isfinite}()} \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
+\code{\link{op_kaiser}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
 \code{\link{op_less_equal}()} \cr
@@ -182,6 +192,8 @@ Other numpy ops: \cr
 \code{\link{op_var}()} \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_zeros}()} \cr
@@ -191,6 +203,7 @@ Other ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -209,6 +222,7 @@ Other ops: \cr
 \code{\link{op_associative_scan}()} \cr
 \code{\link{op_average}()} \cr
 \code{\link{op_average_pool}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_batch_normalization}()} \cr
 \code{\link{op_binary_crossentropy}()} \cr
 \code{\link{op_bincount}()} \cr
@@ -219,9 +233,11 @@ Other ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
 \code{\link{op_cast}()} \cr
 \code{\link{op_categorical_crossentropy}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_celu}()} \cr
 \code{\link{op_cholesky}()} \cr
@@ -234,6 +250,7 @@ Other ops: \cr
 \code{\link{op_convert_to_numpy}()} \cr
 \code{\link{op_convert_to_tensor}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -244,6 +261,7 @@ Other ops: \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
 \code{\link{op_custom_gradient}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_depthwise_conv}()} \cr
 \code{\link{op_det}()} \cr
 \code{\link{op_diag}()} \cr
@@ -283,10 +301,13 @@ Other ops: \cr
 \code{\link{op_glu}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
 \code{\link{op_hard_shrink}()} \cr
 \code{\link{op_hard_sigmoid}()} \cr
 \code{\link{op_hard_silu}()} \cr
 \code{\link{op_hard_tanh}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -313,6 +334,8 @@ Other ops: \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
 \code{\link{op_istft}()} \cr
+\code{\link{op_kaiser}()} \cr
+\code{\link{op_layer_normalization}()} \cr
 \code{\link{op_leaky_relu}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
@@ -412,6 +435,7 @@ Other ops: \cr
 \code{\link{op_sort}()} \cr
 \code{\link{op_sparse_categorical_crossentropy}()} \cr
 \code{\link{op_sparse_plus}()} \cr
+\code{\link{op_sparse_sigmoid}()} \cr
 \code{\link{op_sparsemax}()} \cr
 \code{\link{op_split}()} \cr
 \code{\link{op_sqrt}()} \cr
@@ -448,6 +472,8 @@ Other ops: \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
 \code{\link{op_vectorized_map}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_while_loop}()} \cr
diff --git a/man/op_unravel_index.Rd b/man/op_unravel_index.Rd
index 35d818e7e..7c64a1589 100644
--- a/man/op_unravel_index.Rd
+++ b/man/op_unravel_index.Rd
@@ -51,6 +51,7 @@ Other nn ops: \cr
 \code{\link{op_hard_sigmoid}()} \cr
 \code{\link{op_hard_silu}()} \cr
 \code{\link{op_hard_tanh}()} \cr
+\code{\link{op_layer_normalization}()} \cr
 \code{\link{op_leaky_relu}()} \cr
 \code{\link{op_log_sigmoid}()} \cr
 \code{\link{op_log_softmax}()} \cr
@@ -74,6 +75,7 @@ Other nn ops: \cr
 \code{\link{op_softsign}()} \cr
 \code{\link{op_sparse_categorical_crossentropy}()} \cr
 \code{\link{op_sparse_plus}()} \cr
+\code{\link{op_sparse_sigmoid}()} \cr
 \code{\link{op_sparsemax}()} \cr
 \code{\link{op_squareplus}()} \cr
 \code{\link{op_tanh_shrink}()} \cr
@@ -83,6 +85,7 @@ Other ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -101,6 +104,7 @@ Other ops: \cr
 \code{\link{op_associative_scan}()} \cr
 \code{\link{op_average}()} \cr
 \code{\link{op_average_pool}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_batch_normalization}()} \cr
 \code{\link{op_binary_crossentropy}()} \cr
 \code{\link{op_bincount}()} \cr
@@ -111,9 +115,11 @@ Other ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
 \code{\link{op_cast}()} \cr
 \code{\link{op_categorical_crossentropy}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_celu}()} \cr
 \code{\link{op_cholesky}()} \cr
@@ -126,6 +132,7 @@ Other ops: \cr
 \code{\link{op_convert_to_numpy}()} \cr
 \code{\link{op_convert_to_tensor}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -136,6 +143,7 @@ Other ops: \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
 \code{\link{op_custom_gradient}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_depthwise_conv}()} \cr
 \code{\link{op_det}()} \cr
 \code{\link{op_diag}()} \cr
@@ -175,10 +183,13 @@ Other ops: \cr
 \code{\link{op_glu}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
 \code{\link{op_hard_shrink}()} \cr
 \code{\link{op_hard_sigmoid}()} \cr
 \code{\link{op_hard_silu}()} \cr
 \code{\link{op_hard_tanh}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -205,6 +216,8 @@ Other ops: \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
 \code{\link{op_istft}()} \cr
+\code{\link{op_kaiser}()} \cr
+\code{\link{op_layer_normalization}()} \cr
 \code{\link{op_leaky_relu}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
@@ -304,6 +317,7 @@ Other ops: \cr
 \code{\link{op_sort}()} \cr
 \code{\link{op_sparse_categorical_crossentropy}()} \cr
 \code{\link{op_sparse_plus}()} \cr
+\code{\link{op_sparse_sigmoid}()} \cr
 \code{\link{op_sparsemax}()} \cr
 \code{\link{op_split}()} \cr
 \code{\link{op_sqrt}()} \cr
@@ -340,6 +354,8 @@ Other ops: \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
 \code{\link{op_vectorized_map}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_while_loop}()} \cr
diff --git a/man/op_unstack.Rd b/man/op_unstack.Rd
index fcbaac4ec..9ad7d6801 100644
--- a/man/op_unstack.Rd
+++ b/man/op_unstack.Rd
@@ -97,6 +97,7 @@ Other ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -115,6 +116,7 @@ Other ops: \cr
 \code{\link{op_associative_scan}()} \cr
 \code{\link{op_average}()} \cr
 \code{\link{op_average_pool}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_batch_normalization}()} \cr
 \code{\link{op_binary_crossentropy}()} \cr
 \code{\link{op_bincount}()} \cr
@@ -125,9 +127,11 @@ Other ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
 \code{\link{op_cast}()} \cr
 \code{\link{op_categorical_crossentropy}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_celu}()} \cr
 \code{\link{op_cholesky}()} \cr
@@ -140,6 +144,7 @@ Other ops: \cr
 \code{\link{op_convert_to_numpy}()} \cr
 \code{\link{op_convert_to_tensor}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -150,6 +155,7 @@ Other ops: \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
 \code{\link{op_custom_gradient}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_depthwise_conv}()} \cr
 \code{\link{op_det}()} \cr
 \code{\link{op_diag}()} \cr
@@ -189,10 +195,13 @@ Other ops: \cr
 \code{\link{op_glu}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
 \code{\link{op_hard_shrink}()} \cr
 \code{\link{op_hard_sigmoid}()} \cr
 \code{\link{op_hard_silu}()} \cr
 \code{\link{op_hard_tanh}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -219,6 +228,8 @@ Other ops: \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
 \code{\link{op_istft}()} \cr
+\code{\link{op_kaiser}()} \cr
+\code{\link{op_layer_normalization}()} \cr
 \code{\link{op_leaky_relu}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
@@ -318,6 +329,7 @@ Other ops: \cr
 \code{\link{op_sort}()} \cr
 \code{\link{op_sparse_categorical_crossentropy}()} \cr
 \code{\link{op_sparse_plus}()} \cr
+\code{\link{op_sparse_sigmoid}()} \cr
 \code{\link{op_sparsemax}()} \cr
 \code{\link{op_split}()} \cr
 \code{\link{op_sqrt}()} \cr
@@ -354,6 +366,8 @@ Other ops: \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
 \code{\link{op_vectorized_map}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_while_loop}()} \cr
diff --git a/man/op_var.Rd b/man/op_var.Rd
index 01263918a..ff32664ed 100644
--- a/man/op_var.Rd
+++ b/man/op_var.Rd
@@ -30,6 +30,7 @@ Other numpy ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -46,6 +47,7 @@ Other numpy ops: \cr
 \code{\link{op_argsort}()} \cr
 \code{\link{op_array}()} \cr
 \code{\link{op_average}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_bincount}()} \cr
 \code{\link{op_bitwise_and}()} \cr
 \code{\link{op_bitwise_invert}()} \cr
@@ -54,12 +56,15 @@ Other numpy ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_clip}()} \cr
 \code{\link{op_concatenate}()} \cr
 \code{\link{op_conj}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -68,6 +73,7 @@ Other numpy ops: \cr
 \code{\link{op_ctc_decode}()} \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_diag}()} \cr
 \code{\link{op_diagflat}()} \cr
 \code{\link{op_diagonal}()} \cr
@@ -92,6 +98,9 @@ Other numpy ops: \cr
 \code{\link{op_get_item}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -101,6 +110,7 @@ Other numpy ops: \cr
 \code{\link{op_isfinite}()} \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
+\code{\link{op_kaiser}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
 \code{\link{op_less_equal}()} \cr
@@ -180,6 +190,8 @@ Other numpy ops: \cr
 \code{\link{op_trunc}()} \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_zeros}()} \cr
@@ -189,6 +201,7 @@ Other ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -207,6 +220,7 @@ Other ops: \cr
 \code{\link{op_associative_scan}()} \cr
 \code{\link{op_average}()} \cr
 \code{\link{op_average_pool}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_batch_normalization}()} \cr
 \code{\link{op_binary_crossentropy}()} \cr
 \code{\link{op_bincount}()} \cr
@@ -217,9 +231,11 @@ Other ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
 \code{\link{op_cast}()} \cr
 \code{\link{op_categorical_crossentropy}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_celu}()} \cr
 \code{\link{op_cholesky}()} \cr
@@ -232,6 +248,7 @@ Other ops: \cr
 \code{\link{op_convert_to_numpy}()} \cr
 \code{\link{op_convert_to_tensor}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -242,6 +259,7 @@ Other ops: \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
 \code{\link{op_custom_gradient}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_depthwise_conv}()} \cr
 \code{\link{op_det}()} \cr
 \code{\link{op_diag}()} \cr
@@ -281,10 +299,13 @@ Other ops: \cr
 \code{\link{op_glu}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
 \code{\link{op_hard_shrink}()} \cr
 \code{\link{op_hard_sigmoid}()} \cr
 \code{\link{op_hard_silu}()} \cr
 \code{\link{op_hard_tanh}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -311,6 +332,8 @@ Other ops: \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
 \code{\link{op_istft}()} \cr
+\code{\link{op_kaiser}()} \cr
+\code{\link{op_layer_normalization}()} \cr
 \code{\link{op_leaky_relu}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
@@ -410,6 +433,7 @@ Other ops: \cr
 \code{\link{op_sort}()} \cr
 \code{\link{op_sparse_categorical_crossentropy}()} \cr
 \code{\link{op_sparse_plus}()} \cr
+\code{\link{op_sparse_sigmoid}()} \cr
 \code{\link{op_sparsemax}()} \cr
 \code{\link{op_split}()} \cr
 \code{\link{op_sqrt}()} \cr
@@ -446,6 +470,8 @@ Other ops: \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
 \code{\link{op_vectorized_map}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_while_loop}()} \cr
diff --git a/man/op_vdot.Rd b/man/op_vdot.Rd
index 34a8de9a6..5d18033c7 100644
--- a/man/op_vdot.Rd
+++ b/man/op_vdot.Rd
@@ -30,6 +30,7 @@ Other numpy ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -46,6 +47,7 @@ Other numpy ops: \cr
 \code{\link{op_argsort}()} \cr
 \code{\link{op_array}()} \cr
 \code{\link{op_average}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_bincount}()} \cr
 \code{\link{op_bitwise_and}()} \cr
 \code{\link{op_bitwise_invert}()} \cr
@@ -54,12 +56,15 @@ Other numpy ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_clip}()} \cr
 \code{\link{op_concatenate}()} \cr
 \code{\link{op_conj}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -68,6 +73,7 @@ Other numpy ops: \cr
 \code{\link{op_ctc_decode}()} \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_diag}()} \cr
 \code{\link{op_diagflat}()} \cr
 \code{\link{op_diagonal}()} \cr
@@ -92,6 +98,9 @@ Other numpy ops: \cr
 \code{\link{op_get_item}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -101,6 +110,7 @@ Other numpy ops: \cr
 \code{\link{op_isfinite}()} \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
+\code{\link{op_kaiser}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
 \code{\link{op_less_equal}()} \cr
@@ -180,6 +190,8 @@ Other numpy ops: \cr
 \code{\link{op_trunc}()} \cr
 \code{\link{op_var}()} \cr
 \code{\link{op_vectorize}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_zeros}()} \cr
@@ -189,6 +201,7 @@ Other ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -207,6 +220,7 @@ Other ops: \cr
 \code{\link{op_associative_scan}()} \cr
 \code{\link{op_average}()} \cr
 \code{\link{op_average_pool}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_batch_normalization}()} \cr
 \code{\link{op_binary_crossentropy}()} \cr
 \code{\link{op_bincount}()} \cr
@@ -217,9 +231,11 @@ Other ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
 \code{\link{op_cast}()} \cr
 \code{\link{op_categorical_crossentropy}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_celu}()} \cr
 \code{\link{op_cholesky}()} \cr
@@ -232,6 +248,7 @@ Other ops: \cr
 \code{\link{op_convert_to_numpy}()} \cr
 \code{\link{op_convert_to_tensor}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -242,6 +259,7 @@ Other ops: \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
 \code{\link{op_custom_gradient}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_depthwise_conv}()} \cr
 \code{\link{op_det}()} \cr
 \code{\link{op_diag}()} \cr
@@ -281,10 +299,13 @@ Other ops: \cr
 \code{\link{op_glu}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
 \code{\link{op_hard_shrink}()} \cr
 \code{\link{op_hard_sigmoid}()} \cr
 \code{\link{op_hard_silu}()} \cr
 \code{\link{op_hard_tanh}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -311,6 +332,8 @@ Other ops: \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
 \code{\link{op_istft}()} \cr
+\code{\link{op_kaiser}()} \cr
+\code{\link{op_layer_normalization}()} \cr
 \code{\link{op_leaky_relu}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
@@ -410,6 +433,7 @@ Other ops: \cr
 \code{\link{op_sort}()} \cr
 \code{\link{op_sparse_categorical_crossentropy}()} \cr
 \code{\link{op_sparse_plus}()} \cr
+\code{\link{op_sparse_sigmoid}()} \cr
 \code{\link{op_sparsemax}()} \cr
 \code{\link{op_split}()} \cr
 \code{\link{op_sqrt}()} \cr
@@ -446,6 +470,8 @@ Other ops: \cr
 \code{\link{op_var}()} \cr
 \code{\link{op_vectorize}()} \cr
 \code{\link{op_vectorized_map}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_while_loop}()} \cr
diff --git a/man/op_vectorize.Rd b/man/op_vectorize.Rd
index f1de487c2..46ce7f4ab 100644
--- a/man/op_vectorize.Rd
+++ b/man/op_vectorize.Rd
@@ -51,6 +51,7 @@ Other numpy ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -67,6 +68,7 @@ Other numpy ops: \cr
 \code{\link{op_argsort}()} \cr
 \code{\link{op_array}()} \cr
 \code{\link{op_average}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_bincount}()} \cr
 \code{\link{op_bitwise_and}()} \cr
 \code{\link{op_bitwise_invert}()} \cr
@@ -75,12 +77,15 @@ Other numpy ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_clip}()} \cr
 \code{\link{op_concatenate}()} \cr
 \code{\link{op_conj}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -89,6 +94,7 @@ Other numpy ops: \cr
 \code{\link{op_ctc_decode}()} \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_diag}()} \cr
 \code{\link{op_diagflat}()} \cr
 \code{\link{op_diagonal}()} \cr
@@ -113,6 +119,9 @@ Other numpy ops: \cr
 \code{\link{op_get_item}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -122,6 +131,7 @@ Other numpy ops: \cr
 \code{\link{op_isfinite}()} \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
+\code{\link{op_kaiser}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
 \code{\link{op_less_equal}()} \cr
@@ -201,6 +211,8 @@ Other numpy ops: \cr
 \code{\link{op_trunc}()} \cr
 \code{\link{op_var}()} \cr
 \code{\link{op_vdot}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_zeros}()} \cr
@@ -210,6 +222,7 @@ Other ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -228,6 +241,7 @@ Other ops: \cr
 \code{\link{op_associative_scan}()} \cr
 \code{\link{op_average}()} \cr
 \code{\link{op_average_pool}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_batch_normalization}()} \cr
 \code{\link{op_binary_crossentropy}()} \cr
 \code{\link{op_bincount}()} \cr
@@ -238,9 +252,11 @@ Other ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
 \code{\link{op_cast}()} \cr
 \code{\link{op_categorical_crossentropy}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_celu}()} \cr
 \code{\link{op_cholesky}()} \cr
@@ -253,6 +269,7 @@ Other ops: \cr
 \code{\link{op_convert_to_numpy}()} \cr
 \code{\link{op_convert_to_tensor}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -263,6 +280,7 @@ Other ops: \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
 \code{\link{op_custom_gradient}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_depthwise_conv}()} \cr
 \code{\link{op_det}()} \cr
 \code{\link{op_diag}()} \cr
@@ -302,10 +320,13 @@ Other ops: \cr
 \code{\link{op_glu}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
 \code{\link{op_hard_shrink}()} \cr
 \code{\link{op_hard_sigmoid}()} \cr
 \code{\link{op_hard_silu}()} \cr
 \code{\link{op_hard_tanh}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -332,6 +353,8 @@ Other ops: \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
 \code{\link{op_istft}()} \cr
+\code{\link{op_kaiser}()} \cr
+\code{\link{op_layer_normalization}()} \cr
 \code{\link{op_leaky_relu}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
@@ -431,6 +454,7 @@ Other ops: \cr
 \code{\link{op_sort}()} \cr
 \code{\link{op_sparse_categorical_crossentropy}()} \cr
 \code{\link{op_sparse_plus}()} \cr
+\code{\link{op_sparse_sigmoid}()} \cr
 \code{\link{op_sparsemax}()} \cr
 \code{\link{op_split}()} \cr
 \code{\link{op_sqrt}()} \cr
@@ -467,6 +491,8 @@ Other ops: \cr
 \code{\link{op_var}()} \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorized_map}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_while_loop}()} \cr
diff --git a/man/op_vectorized_map.Rd b/man/op_vectorized_map.Rd
index 7dcff5653..adc825ab8 100644
--- a/man/op_vectorized_map.Rd
+++ b/man/op_vectorized_map.Rd
@@ -256,6 +256,7 @@ Other ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -274,6 +275,7 @@ Other ops: \cr
 \code{\link{op_associative_scan}()} \cr
 \code{\link{op_average}()} \cr
 \code{\link{op_average_pool}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_batch_normalization}()} \cr
 \code{\link{op_binary_crossentropy}()} \cr
 \code{\link{op_bincount}()} \cr
@@ -284,9 +286,11 @@ Other ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
 \code{\link{op_cast}()} \cr
 \code{\link{op_categorical_crossentropy}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_celu}()} \cr
 \code{\link{op_cholesky}()} \cr
@@ -299,6 +303,7 @@ Other ops: \cr
 \code{\link{op_convert_to_numpy}()} \cr
 \code{\link{op_convert_to_tensor}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -309,6 +314,7 @@ Other ops: \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
 \code{\link{op_custom_gradient}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_depthwise_conv}()} \cr
 \code{\link{op_det}()} \cr
 \code{\link{op_diag}()} \cr
@@ -348,10 +354,13 @@ Other ops: \cr
 \code{\link{op_glu}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
 \code{\link{op_hard_shrink}()} \cr
 \code{\link{op_hard_sigmoid}()} \cr
 \code{\link{op_hard_silu}()} \cr
 \code{\link{op_hard_tanh}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -378,6 +387,8 @@ Other ops: \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
 \code{\link{op_istft}()} \cr
+\code{\link{op_kaiser}()} \cr
+\code{\link{op_layer_normalization}()} \cr
 \code{\link{op_leaky_relu}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
@@ -477,6 +488,7 @@ Other ops: \cr
 \code{\link{op_sort}()} \cr
 \code{\link{op_sparse_categorical_crossentropy}()} \cr
 \code{\link{op_sparse_plus}()} \cr
+\code{\link{op_sparse_sigmoid}()} \cr
 \code{\link{op_sparsemax}()} \cr
 \code{\link{op_split}()} \cr
 \code{\link{op_sqrt}()} \cr
@@ -513,6 +525,8 @@ Other ops: \cr
 \code{\link{op_var}()} \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_while_loop}()} \cr
diff --git a/man/op_view_as_complex.Rd b/man/op_view_as_complex.Rd
new file mode 100644
index 000000000..e798817f3
--- /dev/null
+++ b/man/op_view_as_complex.Rd
@@ -0,0 +1,484 @@
+% Generated by roxygen2: do not edit by hand
+% Please edit documentation in R/ops.R
+\name{op_view_as_complex}
+\alias{op_view_as_complex}
+\title{Convert a real tensor with two channels into a complex tensor.}
+\usage{
+op_view_as_complex(x)
+}
+\arguments{
+\item{x}{Real-valued tensor whose trailing dimension encodes the complex components.}
+}
+\value{
+A complex tensor with shape \code{op_shape(x)[-length(op_shape(x))]}.
+}
+\description{
+Expects a real-valued tensor whose last dimension has size \code{2}, holding the
+real and imaginary parts. Returns the corresponding complex tensor with the
+last dimension removed.
+}
+\section{Examples}{
+\if{html}{\out{<div class="sourceCode r">}}\preformatted{x <- op_array(matrix(c(1, 2, 3, 4), nrow = 2, byrow = TRUE))
+op_view_as_complex(x)
+}\if{html}{\out{</div>}}
+
+\if{html}{\out{<div class="sourceCode">}}\preformatted{## tf.Tensor([1.+2.j 3.+4.j], shape=(2), dtype=complex64)
+
+}\if{html}{\out{</div>}}
+}
+
+\seealso{
+Other numpy ops: \cr
+\code{\link{op_abs}()} \cr
+\code{\link{op_add}()} \cr
+\code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
+\code{\link{op_any}()} \cr
+\code{\link{op_append}()} \cr
+\code{\link{op_arange}()} \cr
+\code{\link{op_arccos}()} \cr
+\code{\link{op_arccosh}()} \cr
+\code{\link{op_arcsin}()} \cr
+\code{\link{op_arcsinh}()} \cr
+\code{\link{op_arctan}()} \cr
+\code{\link{op_arctan2}()} \cr
+\code{\link{op_arctanh}()} \cr
+\code{\link{op_argmax}()} \cr
+\code{\link{op_argmin}()} \cr
+\code{\link{op_argpartition}()} \cr
+\code{\link{op_argsort}()} \cr
+\code{\link{op_array}()} \cr
+\code{\link{op_average}()} \cr
+\code{\link{op_bartlett}()} \cr
+\code{\link{op_bincount}()} \cr
+\code{\link{op_bitwise_and}()} \cr
+\code{\link{op_bitwise_invert}()} \cr
+\code{\link{op_bitwise_left_shift}()} \cr
+\code{\link{op_bitwise_not}()} \cr
+\code{\link{op_bitwise_or}()} \cr
+\code{\link{op_bitwise_right_shift}()} \cr
+\code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
+\code{\link{op_broadcast_to}()} \cr
+\code{\link{op_cbrt}()} \cr
+\code{\link{op_ceil}()} \cr
+\code{\link{op_clip}()} \cr
+\code{\link{op_concatenate}()} \cr
+\code{\link{op_conj}()} \cr
+\code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
+\code{\link{op_correlate}()} \cr
+\code{\link{op_cos}()} \cr
+\code{\link{op_cosh}()} \cr
+\code{\link{op_count_nonzero}()} \cr
+\code{\link{op_cross}()} \cr
+\code{\link{op_ctc_decode}()} \cr
+\code{\link{op_cumprod}()} \cr
+\code{\link{op_cumsum}()} \cr
+\code{\link{op_deg2rad}()} \cr
+\code{\link{op_diag}()} \cr
+\code{\link{op_diagflat}()} \cr
+\code{\link{op_diagonal}()} \cr
+\code{\link{op_diff}()} \cr
+\code{\link{op_digitize}()} \cr
+\code{\link{op_divide}()} \cr
+\code{\link{op_divide_no_nan}()} \cr
+\code{\link{op_dot}()} \cr
+\code{\link{op_einsum}()} \cr
+\code{\link{op_empty}()} \cr
+\code{\link{op_equal}()} \cr
+\code{\link{op_exp}()} \cr
+\code{\link{op_exp2}()} \cr
+\code{\link{op_expand_dims}()} \cr
+\code{\link{op_expm1}()} \cr
+\code{\link{op_eye}()} \cr
+\code{\link{op_flip}()} \cr
+\code{\link{op_floor}()} \cr
+\code{\link{op_floor_divide}()} \cr
+\code{\link{op_full}()} \cr
+\code{\link{op_full_like}()} \cr
+\code{\link{op_get_item}()} \cr
+\code{\link{op_greater}()} \cr
+\code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
+\code{\link{op_heaviside}()} \cr
+\code{\link{op_histogram}()} \cr
+\code{\link{op_hstack}()} \cr
+\code{\link{op_identity}()} \cr
+\code{\link{op_imag}()} \cr
+\code{\link{op_inner}()} \cr
+\code{\link{op_isclose}()} \cr
+\code{\link{op_isfinite}()} \cr
+\code{\link{op_isinf}()} \cr
+\code{\link{op_isnan}()} \cr
+\code{\link{op_kaiser}()} \cr
+\code{\link{op_left_shift}()} \cr
+\code{\link{op_less}()} \cr
+\code{\link{op_less_equal}()} \cr
+\code{\link{op_linspace}()} \cr
+\code{\link{op_log}()} \cr
+\code{\link{op_log10}()} \cr
+\code{\link{op_log1p}()} \cr
+\code{\link{op_log2}()} \cr
+\code{\link{op_logaddexp}()} \cr
+\code{\link{op_logdet}()} \cr
+\code{\link{op_logical_and}()} \cr
+\code{\link{op_logical_not}()} \cr
+\code{\link{op_logical_or}()} \cr
+\code{\link{op_logical_xor}()} \cr
+\code{\link{op_logspace}()} \cr
+\code{\link{op_lstsq}()} \cr
+\code{\link{op_matmul}()} \cr
+\code{\link{op_max}()} \cr
+\code{\link{op_maximum}()} \cr
+\code{\link{op_mean}()} \cr
+\code{\link{op_median}()} \cr
+\code{\link{op_meshgrid}()} \cr
+\code{\link{op_min}()} \cr
+\code{\link{op_minimum}()} \cr
+\code{\link{op_mod}()} \cr
+\code{\link{op_moveaxis}()} \cr
+\code{\link{op_multiply}()} \cr
+\code{\link{op_nan_to_num}()} \cr
+\code{\link{op_ndim}()} \cr
+\code{\link{op_negative}()} \cr
+\code{\link{op_nonzero}()} \cr
+\code{\link{op_not_equal}()} \cr
+\code{\link{op_ones}()} \cr
+\code{\link{op_ones_like}()} \cr
+\code{\link{op_outer}()} \cr
+\code{\link{op_pad}()} \cr
+\code{\link{op_power}()} \cr
+\code{\link{op_prod}()} \cr
+\code{\link{op_quantile}()} \cr
+\code{\link{op_ravel}()} \cr
+\code{\link{op_real}()} \cr
+\code{\link{op_reciprocal}()} \cr
+\code{\link{op_repeat}()} \cr
+\code{\link{op_reshape}()} \cr
+\code{\link{op_right_shift}()} \cr
+\code{\link{op_roll}()} \cr
+\code{\link{op_rot90}()} \cr
+\code{\link{op_round}()} \cr
+\code{\link{op_saturate_cast}()} \cr
+\code{\link{op_select}()} \cr
+\code{\link{op_sign}()} \cr
+\code{\link{op_signbit}()} \cr
+\code{\link{op_sin}()} \cr
+\code{\link{op_sinh}()} \cr
+\code{\link{op_size}()} \cr
+\code{\link{op_sort}()} \cr
+\code{\link{op_split}()} \cr
+\code{\link{op_sqrt}()} \cr
+\code{\link{op_square}()} \cr
+\code{\link{op_squeeze}()} \cr
+\code{\link{op_stack}()} \cr
+\code{\link{op_std}()} \cr
+\code{\link{op_subtract}()} \cr
+\code{\link{op_sum}()} \cr
+\code{\link{op_swapaxes}()} \cr
+\code{\link{op_take}()} \cr
+\code{\link{op_take_along_axis}()} \cr
+\code{\link{op_tan}()} \cr
+\code{\link{op_tanh}()} \cr
+\code{\link{op_tensordot}()} \cr
+\code{\link{op_tile}()} \cr
+\code{\link{op_trace}()} \cr
+\code{\link{op_transpose}()} \cr
+\code{\link{op_tri}()} \cr
+\code{\link{op_tril}()} \cr
+\code{\link{op_triu}()} \cr
+\code{\link{op_trunc}()} \cr
+\code{\link{op_var}()} \cr
+\code{\link{op_vdot}()} \cr
+\code{\link{op_vectorize}()} \cr
+\code{\link{op_view_as_real}()} \cr
+\code{\link{op_vstack}()} \cr
+\code{\link{op_where}()} \cr
+\code{\link{op_zeros}()} \cr
+\code{\link{op_zeros_like}()} \cr
+
+Other ops: \cr
+\code{\link{op_abs}()} \cr
+\code{\link{op_add}()} \cr
+\code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
+\code{\link{op_any}()} \cr
+\code{\link{op_append}()} \cr
+\code{\link{op_arange}()} \cr
+\code{\link{op_arccos}()} \cr
+\code{\link{op_arccosh}()} \cr
+\code{\link{op_arcsin}()} \cr
+\code{\link{op_arcsinh}()} \cr
+\code{\link{op_arctan}()} \cr
+\code{\link{op_arctan2}()} \cr
+\code{\link{op_arctanh}()} \cr
+\code{\link{op_argmax}()} \cr
+\code{\link{op_argmin}()} \cr
+\code{\link{op_argpartition}()} \cr
+\code{\link{op_argsort}()} \cr
+\code{\link{op_array}()} \cr
+\code{\link{op_associative_scan}()} \cr
+\code{\link{op_average}()} \cr
+\code{\link{op_average_pool}()} \cr
+\code{\link{op_bartlett}()} \cr
+\code{\link{op_batch_normalization}()} \cr
+\code{\link{op_binary_crossentropy}()} \cr
+\code{\link{op_bincount}()} \cr
+\code{\link{op_bitwise_and}()} \cr
+\code{\link{op_bitwise_invert}()} \cr
+\code{\link{op_bitwise_left_shift}()} \cr
+\code{\link{op_bitwise_not}()} \cr
+\code{\link{op_bitwise_or}()} \cr
+\code{\link{op_bitwise_right_shift}()} \cr
+\code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
+\code{\link{op_broadcast_to}()} \cr
+\code{\link{op_cast}()} \cr
+\code{\link{op_categorical_crossentropy}()} \cr
+\code{\link{op_cbrt}()} \cr
+\code{\link{op_ceil}()} \cr
+\code{\link{op_celu}()} \cr
+\code{\link{op_cholesky}()} \cr
+\code{\link{op_clip}()} \cr
+\code{\link{op_concatenate}()} \cr
+\code{\link{op_cond}()} \cr
+\code{\link{op_conj}()} \cr
+\code{\link{op_conv}()} \cr
+\code{\link{op_conv_transpose}()} \cr
+\code{\link{op_convert_to_numpy}()} \cr
+\code{\link{op_convert_to_tensor}()} \cr
+\code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
+\code{\link{op_correlate}()} \cr
+\code{\link{op_cos}()} \cr
+\code{\link{op_cosh}()} \cr
+\code{\link{op_count_nonzero}()} \cr
+\code{\link{op_cross}()} \cr
+\code{\link{op_ctc_decode}()} \cr
+\code{\link{op_ctc_loss}()} \cr
+\code{\link{op_cumprod}()} \cr
+\code{\link{op_cumsum}()} \cr
+\code{\link{op_custom_gradient}()} \cr
+\code{\link{op_deg2rad}()} \cr
+\code{\link{op_depthwise_conv}()} \cr
+\code{\link{op_det}()} \cr
+\code{\link{op_diag}()} \cr
+\code{\link{op_diagflat}()} \cr
+\code{\link{op_diagonal}()} \cr
+\code{\link{op_diff}()} \cr
+\code{\link{op_digitize}()} \cr
+\code{\link{op_divide}()} \cr
+\code{\link{op_divide_no_nan}()} \cr
+\code{\link{op_dot}()} \cr
+\code{\link{op_dot_product_attention}()} \cr
+\code{\link{op_dtype}()} \cr
+\code{\link{op_eig}()} \cr
+\code{\link{op_eigh}()} \cr
+\code{\link{op_einsum}()} \cr
+\code{\link{op_elu}()} \cr
+\code{\link{op_empty}()} \cr
+\code{\link{op_equal}()} \cr
+\code{\link{op_erf}()} \cr
+\code{\link{op_erfinv}()} \cr
+\code{\link{op_exp}()} \cr
+\code{\link{op_exp2}()} \cr
+\code{\link{op_expand_dims}()} \cr
+\code{\link{op_expm1}()} \cr
+\code{\link{op_extract_sequences}()} \cr
+\code{\link{op_eye}()} \cr
+\code{\link{op_fft}()} \cr
+\code{\link{op_fft2}()} \cr
+\code{\link{op_flip}()} \cr
+\code{\link{op_floor}()} \cr
+\code{\link{op_floor_divide}()} \cr
+\code{\link{op_fori_loop}()} \cr
+\code{\link{op_full}()} \cr
+\code{\link{op_full_like}()} \cr
+\code{\link{op_gelu}()} \cr
+\code{\link{op_get_item}()} \cr
+\code{\link{op_glu}()} \cr
+\code{\link{op_greater}()} \cr
+\code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
+\code{\link{op_hard_shrink}()} \cr
+\code{\link{op_hard_sigmoid}()} \cr
+\code{\link{op_hard_silu}()} \cr
+\code{\link{op_hard_tanh}()} \cr
+\code{\link{op_heaviside}()} \cr
+\code{\link{op_histogram}()} \cr
+\code{\link{op_hstack}()} \cr
+\code{\link{op_identity}()} \cr
+\code{\link{op_ifft2}()} \cr
+\code{\link{op_imag}()} \cr
+\code{\link{op_image_affine_transform}()} \cr
+\code{\link{op_image_crop}()} \cr
+\code{\link{op_image_extract_patches}()} \cr
+\code{\link{op_image_gaussian_blur}()} \cr
+\code{\link{op_image_hsv_to_rgb}()} \cr
+\code{\link{op_image_map_coordinates}()} \cr
+\code{\link{op_image_pad}()} \cr
+\code{\link{op_image_perspective_transform}()} \cr
+\code{\link{op_image_resize}()} \cr
+\code{\link{op_image_rgb_to_grayscale}()} \cr
+\code{\link{op_image_rgb_to_hsv}()} \cr
+\code{\link{op_in_top_k}()} \cr
+\code{\link{op_inner}()} \cr
+\code{\link{op_inv}()} \cr
+\code{\link{op_irfft}()} \cr
+\code{\link{op_is_tensor}()} \cr
+\code{\link{op_isclose}()} \cr
+\code{\link{op_isfinite}()} \cr
+\code{\link{op_isinf}()} \cr
+\code{\link{op_isnan}()} \cr
+\code{\link{op_istft}()} \cr
+\code{\link{op_kaiser}()} \cr
+\code{\link{op_layer_normalization}()} \cr
+\code{\link{op_leaky_relu}()} \cr
+\code{\link{op_left_shift}()} \cr
+\code{\link{op_less}()} \cr
+\code{\link{op_less_equal}()} \cr
+\code{\link{op_linspace}()} \cr
+\code{\link{op_log}()} \cr
+\code{\link{op_log10}()} \cr
+\code{\link{op_log1p}()} \cr
+\code{\link{op_log2}()} \cr
+\code{\link{op_log_sigmoid}()} \cr
+\code{\link{op_log_softmax}()} \cr
+\code{\link{op_logaddexp}()} \cr
+\code{\link{op_logdet}()} \cr
+\code{\link{op_logical_and}()} \cr
+\code{\link{op_logical_not}()} \cr
+\code{\link{op_logical_or}()} \cr
+\code{\link{op_logical_xor}()} \cr
+\code{\link{op_logspace}()} \cr
+\code{\link{op_logsumexp}()} \cr
+\code{\link{op_lstsq}()} \cr
+\code{\link{op_lu_factor}()} \cr
+\code{\link{op_map}()} \cr
+\code{\link{op_matmul}()} \cr
+\code{\link{op_max}()} \cr
+\code{\link{op_max_pool}()} \cr
+\code{\link{op_maximum}()} \cr
+\code{\link{op_mean}()} \cr
+\code{\link{op_median}()} \cr
+\code{\link{op_meshgrid}()} \cr
+\code{\link{op_min}()} \cr
+\code{\link{op_minimum}()} \cr
+\code{\link{op_mod}()} \cr
+\code{\link{op_moments}()} \cr
+\code{\link{op_moveaxis}()} \cr
+\code{\link{op_multi_hot}()} \cr
+\code{\link{op_multiply}()} \cr
+\code{\link{op_nan_to_num}()} \cr
+\code{\link{op_ndim}()} \cr
+\code{\link{op_negative}()} \cr
+\code{\link{op_nonzero}()} \cr
+\code{\link{op_norm}()} \cr
+\code{\link{op_normalize}()} \cr
+\code{\link{op_not_equal}()} \cr
+\code{\link{op_one_hot}()} \cr
+\code{\link{op_ones}()} \cr
+\code{\link{op_ones_like}()} \cr
+\code{\link{op_outer}()} \cr
+\code{\link{op_pad}()} \cr
+\code{\link{op_polar}()} \cr
+\code{\link{op_power}()} \cr
+\code{\link{op_prod}()} \cr
+\code{\link{op_psnr}()} \cr
+\code{\link{op_qr}()} \cr
+\code{\link{op_quantile}()} \cr
+\code{\link{op_ravel}()} \cr
+\code{\link{op_real}()} \cr
+\code{\link{op_rearrange}()} \cr
+\code{\link{op_reciprocal}()} \cr
+\code{\link{op_relu}()} \cr
+\code{\link{op_relu6}()} \cr
+\code{\link{op_repeat}()} \cr
+\code{\link{op_reshape}()} \cr
+\code{\link{op_rfft}()} \cr
+\code{\link{op_right_shift}()} \cr
+\code{\link{op_rms_normalization}()} \cr
+\code{\link{op_roll}()} \cr
+\code{\link{op_rot90}()} \cr
+\code{\link{op_round}()} \cr
+\code{\link{op_rsqrt}()} \cr
+\code{\link{op_saturate_cast}()} \cr
+\code{\link{op_scan}()} \cr
+\code{\link{op_scatter}()} \cr
+\code{\link{op_scatter_update}()} \cr
+\code{\link{op_searchsorted}()} \cr
+\code{\link{op_segment_max}()} \cr
+\code{\link{op_segment_sum}()} \cr
+\code{\link{op_select}()} \cr
+\code{\link{op_selu}()} \cr
+\code{\link{op_separable_conv}()} \cr
+\code{\link{op_shape}()} \cr
+\code{\link{op_sigmoid}()} \cr
+\code{\link{op_sign}()} \cr
+\code{\link{op_signbit}()} \cr
+\code{\link{op_silu}()} \cr
+\code{\link{op_sin}()} \cr
+\code{\link{op_sinh}()} \cr
+\code{\link{op_size}()} \cr
+\code{\link{op_slice}()} \cr
+\code{\link{op_slice_update}()} \cr
+\code{\link{op_slogdet}()} \cr
+\code{\link{op_soft_shrink}()} \cr
+\code{\link{op_softmax}()} \cr
+\code{\link{op_softplus}()} \cr
+\code{\link{op_softsign}()} \cr
+\code{\link{op_solve}()} \cr
+\code{\link{op_solve_triangular}()} \cr
+\code{\link{op_sort}()} \cr
+\code{\link{op_sparse_categorical_crossentropy}()} \cr
+\code{\link{op_sparse_plus}()} \cr
+\code{\link{op_sparse_sigmoid}()} \cr
+\code{\link{op_sparsemax}()} \cr
+\code{\link{op_split}()} \cr
+\code{\link{op_sqrt}()} \cr
+\code{\link{op_square}()} \cr
+\code{\link{op_squareplus}()} \cr
+\code{\link{op_squeeze}()} \cr
+\code{\link{op_stack}()} \cr
+\code{\link{op_std}()} \cr
+\code{\link{op_stft}()} \cr
+\code{\link{op_stop_gradient}()} \cr
+\code{\link{op_subset}()} \cr
+\code{\link{op_subtract}()} \cr
+\code{\link{op_sum}()} \cr
+\code{\link{op_svd}()} \cr
+\code{\link{op_swapaxes}()} \cr
+\code{\link{op_switch}()} \cr
+\code{\link{op_take}()} \cr
+\code{\link{op_take_along_axis}()} \cr
+\code{\link{op_tan}()} \cr
+\code{\link{op_tanh}()} \cr
+\code{\link{op_tanh_shrink}()} \cr
+\code{\link{op_tensordot}()} \cr
+\code{\link{op_threshold}()} \cr
+\code{\link{op_tile}()} \cr
+\code{\link{op_top_k}()} \cr
+\code{\link{op_trace}()} \cr
+\code{\link{op_transpose}()} \cr
+\code{\link{op_tri}()} \cr
+\code{\link{op_tril}()} \cr
+\code{\link{op_triu}()} \cr
+\code{\link{op_trunc}()} \cr
+\code{\link{op_unravel_index}()} \cr
+\code{\link{op_unstack}()} \cr
+\code{\link{op_var}()} \cr
+\code{\link{op_vdot}()} \cr
+\code{\link{op_vectorize}()} \cr
+\code{\link{op_vectorized_map}()} \cr
+\code{\link{op_view_as_real}()} \cr
+\code{\link{op_vstack}()} \cr
+\code{\link{op_where}()} \cr
+\code{\link{op_while_loop}()} \cr
+\code{\link{op_zeros}()} \cr
+\code{\link{op_zeros_like}()} \cr
+}
+\concept{numpy ops}
+\concept{ops}
diff --git a/man/op_view_as_real.Rd b/man/op_view_as_real.Rd
new file mode 100644
index 000000000..c452b1206
--- /dev/null
+++ b/man/op_view_as_real.Rd
@@ -0,0 +1,486 @@
+% Generated by roxygen2: do not edit by hand
+% Please edit documentation in R/ops.R
+\name{op_view_as_real}
+\alias{op_view_as_real}
+\title{Convert a complex tensor into a stacked real representation.}
+\usage{
+op_view_as_real(x)
+}
+\arguments{
+\item{x}{Complex-valued tensor to be converted.}
+}
+\value{
+A real tensor with shape \code{c(op_shape(x), 2)} containing real and imaginary parts.
+}
+\description{
+Produces a real-valued tensor where the last dimension gathers the real and
+imaginary parts of the complex input.
+}
+\section{Examples}{
+\if{html}{\out{<div class="sourceCode r">}}\preformatted{x <- op_array(matrix(c(1, 2, 3, 4), nrow = 2, byrow = TRUE))
+z <- op_view_as_complex(x)
+op_view_as_real(z)
+}\if{html}{\out{</div>}}
+
+\if{html}{\out{<div class="sourceCode">}}\preformatted{## tf.Tensor(
+## [[1. 2.]
+##  [3. 4.]], shape=(2, 2), dtype=float32)
+
+}\if{html}{\out{</div>}}
+}
+
+\seealso{
+Other numpy ops: \cr
+\code{\link{op_abs}()} \cr
+\code{\link{op_add}()} \cr
+\code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
+\code{\link{op_any}()} \cr
+\code{\link{op_append}()} \cr
+\code{\link{op_arange}()} \cr
+\code{\link{op_arccos}()} \cr
+\code{\link{op_arccosh}()} \cr
+\code{\link{op_arcsin}()} \cr
+\code{\link{op_arcsinh}()} \cr
+\code{\link{op_arctan}()} \cr
+\code{\link{op_arctan2}()} \cr
+\code{\link{op_arctanh}()} \cr
+\code{\link{op_argmax}()} \cr
+\code{\link{op_argmin}()} \cr
+\code{\link{op_argpartition}()} \cr
+\code{\link{op_argsort}()} \cr
+\code{\link{op_array}()} \cr
+\code{\link{op_average}()} \cr
+\code{\link{op_bartlett}()} \cr
+\code{\link{op_bincount}()} \cr
+\code{\link{op_bitwise_and}()} \cr
+\code{\link{op_bitwise_invert}()} \cr
+\code{\link{op_bitwise_left_shift}()} \cr
+\code{\link{op_bitwise_not}()} \cr
+\code{\link{op_bitwise_or}()} \cr
+\code{\link{op_bitwise_right_shift}()} \cr
+\code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
+\code{\link{op_broadcast_to}()} \cr
+\code{\link{op_cbrt}()} \cr
+\code{\link{op_ceil}()} \cr
+\code{\link{op_clip}()} \cr
+\code{\link{op_concatenate}()} \cr
+\code{\link{op_conj}()} \cr
+\code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
+\code{\link{op_correlate}()} \cr
+\code{\link{op_cos}()} \cr
+\code{\link{op_cosh}()} \cr
+\code{\link{op_count_nonzero}()} \cr
+\code{\link{op_cross}()} \cr
+\code{\link{op_ctc_decode}()} \cr
+\code{\link{op_cumprod}()} \cr
+\code{\link{op_cumsum}()} \cr
+\code{\link{op_deg2rad}()} \cr
+\code{\link{op_diag}()} \cr
+\code{\link{op_diagflat}()} \cr
+\code{\link{op_diagonal}()} \cr
+\code{\link{op_diff}()} \cr
+\code{\link{op_digitize}()} \cr
+\code{\link{op_divide}()} \cr
+\code{\link{op_divide_no_nan}()} \cr
+\code{\link{op_dot}()} \cr
+\code{\link{op_einsum}()} \cr
+\code{\link{op_empty}()} \cr
+\code{\link{op_equal}()} \cr
+\code{\link{op_exp}()} \cr
+\code{\link{op_exp2}()} \cr
+\code{\link{op_expand_dims}()} \cr
+\code{\link{op_expm1}()} \cr
+\code{\link{op_eye}()} \cr
+\code{\link{op_flip}()} \cr
+\code{\link{op_floor}()} \cr
+\code{\link{op_floor_divide}()} \cr
+\code{\link{op_full}()} \cr
+\code{\link{op_full_like}()} \cr
+\code{\link{op_get_item}()} \cr
+\code{\link{op_greater}()} \cr
+\code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
+\code{\link{op_heaviside}()} \cr
+\code{\link{op_histogram}()} \cr
+\code{\link{op_hstack}()} \cr
+\code{\link{op_identity}()} \cr
+\code{\link{op_imag}()} \cr
+\code{\link{op_inner}()} \cr
+\code{\link{op_isclose}()} \cr
+\code{\link{op_isfinite}()} \cr
+\code{\link{op_isinf}()} \cr
+\code{\link{op_isnan}()} \cr
+\code{\link{op_kaiser}()} \cr
+\code{\link{op_left_shift}()} \cr
+\code{\link{op_less}()} \cr
+\code{\link{op_less_equal}()} \cr
+\code{\link{op_linspace}()} \cr
+\code{\link{op_log}()} \cr
+\code{\link{op_log10}()} \cr
+\code{\link{op_log1p}()} \cr
+\code{\link{op_log2}()} \cr
+\code{\link{op_logaddexp}()} \cr
+\code{\link{op_logdet}()} \cr
+\code{\link{op_logical_and}()} \cr
+\code{\link{op_logical_not}()} \cr
+\code{\link{op_logical_or}()} \cr
+\code{\link{op_logical_xor}()} \cr
+\code{\link{op_logspace}()} \cr
+\code{\link{op_lstsq}()} \cr
+\code{\link{op_matmul}()} \cr
+\code{\link{op_max}()} \cr
+\code{\link{op_maximum}()} \cr
+\code{\link{op_mean}()} \cr
+\code{\link{op_median}()} \cr
+\code{\link{op_meshgrid}()} \cr
+\code{\link{op_min}()} \cr
+\code{\link{op_minimum}()} \cr
+\code{\link{op_mod}()} \cr
+\code{\link{op_moveaxis}()} \cr
+\code{\link{op_multiply}()} \cr
+\code{\link{op_nan_to_num}()} \cr
+\code{\link{op_ndim}()} \cr
+\code{\link{op_negative}()} \cr
+\code{\link{op_nonzero}()} \cr
+\code{\link{op_not_equal}()} \cr
+\code{\link{op_ones}()} \cr
+\code{\link{op_ones_like}()} \cr
+\code{\link{op_outer}()} \cr
+\code{\link{op_pad}()} \cr
+\code{\link{op_power}()} \cr
+\code{\link{op_prod}()} \cr
+\code{\link{op_quantile}()} \cr
+\code{\link{op_ravel}()} \cr
+\code{\link{op_real}()} \cr
+\code{\link{op_reciprocal}()} \cr
+\code{\link{op_repeat}()} \cr
+\code{\link{op_reshape}()} \cr
+\code{\link{op_right_shift}()} \cr
+\code{\link{op_roll}()} \cr
+\code{\link{op_rot90}()} \cr
+\code{\link{op_round}()} \cr
+\code{\link{op_saturate_cast}()} \cr
+\code{\link{op_select}()} \cr
+\code{\link{op_sign}()} \cr
+\code{\link{op_signbit}()} \cr
+\code{\link{op_sin}()} \cr
+\code{\link{op_sinh}()} \cr
+\code{\link{op_size}()} \cr
+\code{\link{op_sort}()} \cr
+\code{\link{op_split}()} \cr
+\code{\link{op_sqrt}()} \cr
+\code{\link{op_square}()} \cr
+\code{\link{op_squeeze}()} \cr
+\code{\link{op_stack}()} \cr
+\code{\link{op_std}()} \cr
+\code{\link{op_subtract}()} \cr
+\code{\link{op_sum}()} \cr
+\code{\link{op_swapaxes}()} \cr
+\code{\link{op_take}()} \cr
+\code{\link{op_take_along_axis}()} \cr
+\code{\link{op_tan}()} \cr
+\code{\link{op_tanh}()} \cr
+\code{\link{op_tensordot}()} \cr
+\code{\link{op_tile}()} \cr
+\code{\link{op_trace}()} \cr
+\code{\link{op_transpose}()} \cr
+\code{\link{op_tri}()} \cr
+\code{\link{op_tril}()} \cr
+\code{\link{op_triu}()} \cr
+\code{\link{op_trunc}()} \cr
+\code{\link{op_var}()} \cr
+\code{\link{op_vdot}()} \cr
+\code{\link{op_vectorize}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_vstack}()} \cr
+\code{\link{op_where}()} \cr
+\code{\link{op_zeros}()} \cr
+\code{\link{op_zeros_like}()} \cr
+
+Other ops: \cr
+\code{\link{op_abs}()} \cr
+\code{\link{op_add}()} \cr
+\code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
+\code{\link{op_any}()} \cr
+\code{\link{op_append}()} \cr
+\code{\link{op_arange}()} \cr
+\code{\link{op_arccos}()} \cr
+\code{\link{op_arccosh}()} \cr
+\code{\link{op_arcsin}()} \cr
+\code{\link{op_arcsinh}()} \cr
+\code{\link{op_arctan}()} \cr
+\code{\link{op_arctan2}()} \cr
+\code{\link{op_arctanh}()} \cr
+\code{\link{op_argmax}()} \cr
+\code{\link{op_argmin}()} \cr
+\code{\link{op_argpartition}()} \cr
+\code{\link{op_argsort}()} \cr
+\code{\link{op_array}()} \cr
+\code{\link{op_associative_scan}()} \cr
+\code{\link{op_average}()} \cr
+\code{\link{op_average_pool}()} \cr
+\code{\link{op_bartlett}()} \cr
+\code{\link{op_batch_normalization}()} \cr
+\code{\link{op_binary_crossentropy}()} \cr
+\code{\link{op_bincount}()} \cr
+\code{\link{op_bitwise_and}()} \cr
+\code{\link{op_bitwise_invert}()} \cr
+\code{\link{op_bitwise_left_shift}()} \cr
+\code{\link{op_bitwise_not}()} \cr
+\code{\link{op_bitwise_or}()} \cr
+\code{\link{op_bitwise_right_shift}()} \cr
+\code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
+\code{\link{op_broadcast_to}()} \cr
+\code{\link{op_cast}()} \cr
+\code{\link{op_categorical_crossentropy}()} \cr
+\code{\link{op_cbrt}()} \cr
+\code{\link{op_ceil}()} \cr
+\code{\link{op_celu}()} \cr
+\code{\link{op_cholesky}()} \cr
+\code{\link{op_clip}()} \cr
+\code{\link{op_concatenate}()} \cr
+\code{\link{op_cond}()} \cr
+\code{\link{op_conj}()} \cr
+\code{\link{op_conv}()} \cr
+\code{\link{op_conv_transpose}()} \cr
+\code{\link{op_convert_to_numpy}()} \cr
+\code{\link{op_convert_to_tensor}()} \cr
+\code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
+\code{\link{op_correlate}()} \cr
+\code{\link{op_cos}()} \cr
+\code{\link{op_cosh}()} \cr
+\code{\link{op_count_nonzero}()} \cr
+\code{\link{op_cross}()} \cr
+\code{\link{op_ctc_decode}()} \cr
+\code{\link{op_ctc_loss}()} \cr
+\code{\link{op_cumprod}()} \cr
+\code{\link{op_cumsum}()} \cr
+\code{\link{op_custom_gradient}()} \cr
+\code{\link{op_deg2rad}()} \cr
+\code{\link{op_depthwise_conv}()} \cr
+\code{\link{op_det}()} \cr
+\code{\link{op_diag}()} \cr
+\code{\link{op_diagflat}()} \cr
+\code{\link{op_diagonal}()} \cr
+\code{\link{op_diff}()} \cr
+\code{\link{op_digitize}()} \cr
+\code{\link{op_divide}()} \cr
+\code{\link{op_divide_no_nan}()} \cr
+\code{\link{op_dot}()} \cr
+\code{\link{op_dot_product_attention}()} \cr
+\code{\link{op_dtype}()} \cr
+\code{\link{op_eig}()} \cr
+\code{\link{op_eigh}()} \cr
+\code{\link{op_einsum}()} \cr
+\code{\link{op_elu}()} \cr
+\code{\link{op_empty}()} \cr
+\code{\link{op_equal}()} \cr
+\code{\link{op_erf}()} \cr
+\code{\link{op_erfinv}()} \cr
+\code{\link{op_exp}()} \cr
+\code{\link{op_exp2}()} \cr
+\code{\link{op_expand_dims}()} \cr
+\code{\link{op_expm1}()} \cr
+\code{\link{op_extract_sequences}()} \cr
+\code{\link{op_eye}()} \cr
+\code{\link{op_fft}()} \cr
+\code{\link{op_fft2}()} \cr
+\code{\link{op_flip}()} \cr
+\code{\link{op_floor}()} \cr
+\code{\link{op_floor_divide}()} \cr
+\code{\link{op_fori_loop}()} \cr
+\code{\link{op_full}()} \cr
+\code{\link{op_full_like}()} \cr
+\code{\link{op_gelu}()} \cr
+\code{\link{op_get_item}()} \cr
+\code{\link{op_glu}()} \cr
+\code{\link{op_greater}()} \cr
+\code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
+\code{\link{op_hard_shrink}()} \cr
+\code{\link{op_hard_sigmoid}()} \cr
+\code{\link{op_hard_silu}()} \cr
+\code{\link{op_hard_tanh}()} \cr
+\code{\link{op_heaviside}()} \cr
+\code{\link{op_histogram}()} \cr
+\code{\link{op_hstack}()} \cr
+\code{\link{op_identity}()} \cr
+\code{\link{op_ifft2}()} \cr
+\code{\link{op_imag}()} \cr
+\code{\link{op_image_affine_transform}()} \cr
+\code{\link{op_image_crop}()} \cr
+\code{\link{op_image_extract_patches}()} \cr
+\code{\link{op_image_gaussian_blur}()} \cr
+\code{\link{op_image_hsv_to_rgb}()} \cr
+\code{\link{op_image_map_coordinates}()} \cr
+\code{\link{op_image_pad}()} \cr
+\code{\link{op_image_perspective_transform}()} \cr
+\code{\link{op_image_resize}()} \cr
+\code{\link{op_image_rgb_to_grayscale}()} \cr
+\code{\link{op_image_rgb_to_hsv}()} \cr
+\code{\link{op_in_top_k}()} \cr
+\code{\link{op_inner}()} \cr
+\code{\link{op_inv}()} \cr
+\code{\link{op_irfft}()} \cr
+\code{\link{op_is_tensor}()} \cr
+\code{\link{op_isclose}()} \cr
+\code{\link{op_isfinite}()} \cr
+\code{\link{op_isinf}()} \cr
+\code{\link{op_isnan}()} \cr
+\code{\link{op_istft}()} \cr
+\code{\link{op_kaiser}()} \cr
+\code{\link{op_layer_normalization}()} \cr
+\code{\link{op_leaky_relu}()} \cr
+\code{\link{op_left_shift}()} \cr
+\code{\link{op_less}()} \cr
+\code{\link{op_less_equal}()} \cr
+\code{\link{op_linspace}()} \cr
+\code{\link{op_log}()} \cr
+\code{\link{op_log10}()} \cr
+\code{\link{op_log1p}()} \cr
+\code{\link{op_log2}()} \cr
+\code{\link{op_log_sigmoid}()} \cr
+\code{\link{op_log_softmax}()} \cr
+\code{\link{op_logaddexp}()} \cr
+\code{\link{op_logdet}()} \cr
+\code{\link{op_logical_and}()} \cr
+\code{\link{op_logical_not}()} \cr
+\code{\link{op_logical_or}()} \cr
+\code{\link{op_logical_xor}()} \cr
+\code{\link{op_logspace}()} \cr
+\code{\link{op_logsumexp}()} \cr
+\code{\link{op_lstsq}()} \cr
+\code{\link{op_lu_factor}()} \cr
+\code{\link{op_map}()} \cr
+\code{\link{op_matmul}()} \cr
+\code{\link{op_max}()} \cr
+\code{\link{op_max_pool}()} \cr
+\code{\link{op_maximum}()} \cr
+\code{\link{op_mean}()} \cr
+\code{\link{op_median}()} \cr
+\code{\link{op_meshgrid}()} \cr
+\code{\link{op_min}()} \cr
+\code{\link{op_minimum}()} \cr
+\code{\link{op_mod}()} \cr
+\code{\link{op_moments}()} \cr
+\code{\link{op_moveaxis}()} \cr
+\code{\link{op_multi_hot}()} \cr
+\code{\link{op_multiply}()} \cr
+\code{\link{op_nan_to_num}()} \cr
+\code{\link{op_ndim}()} \cr
+\code{\link{op_negative}()} \cr
+\code{\link{op_nonzero}()} \cr
+\code{\link{op_norm}()} \cr
+\code{\link{op_normalize}()} \cr
+\code{\link{op_not_equal}()} \cr
+\code{\link{op_one_hot}()} \cr
+\code{\link{op_ones}()} \cr
+\code{\link{op_ones_like}()} \cr
+\code{\link{op_outer}()} \cr
+\code{\link{op_pad}()} \cr
+\code{\link{op_polar}()} \cr
+\code{\link{op_power}()} \cr
+\code{\link{op_prod}()} \cr
+\code{\link{op_psnr}()} \cr
+\code{\link{op_qr}()} \cr
+\code{\link{op_quantile}()} \cr
+\code{\link{op_ravel}()} \cr
+\code{\link{op_real}()} \cr
+\code{\link{op_rearrange}()} \cr
+\code{\link{op_reciprocal}()} \cr
+\code{\link{op_relu}()} \cr
+\code{\link{op_relu6}()} \cr
+\code{\link{op_repeat}()} \cr
+\code{\link{op_reshape}()} \cr
+\code{\link{op_rfft}()} \cr
+\code{\link{op_right_shift}()} \cr
+\code{\link{op_rms_normalization}()} \cr
+\code{\link{op_roll}()} \cr
+\code{\link{op_rot90}()} \cr
+\code{\link{op_round}()} \cr
+\code{\link{op_rsqrt}()} \cr
+\code{\link{op_saturate_cast}()} \cr
+\code{\link{op_scan}()} \cr
+\code{\link{op_scatter}()} \cr
+\code{\link{op_scatter_update}()} \cr
+\code{\link{op_searchsorted}()} \cr
+\code{\link{op_segment_max}()} \cr
+\code{\link{op_segment_sum}()} \cr
+\code{\link{op_select}()} \cr
+\code{\link{op_selu}()} \cr
+\code{\link{op_separable_conv}()} \cr
+\code{\link{op_shape}()} \cr
+\code{\link{op_sigmoid}()} \cr
+\code{\link{op_sign}()} \cr
+\code{\link{op_signbit}()} \cr
+\code{\link{op_silu}()} \cr
+\code{\link{op_sin}()} \cr
+\code{\link{op_sinh}()} \cr
+\code{\link{op_size}()} \cr
+\code{\link{op_slice}()} \cr
+\code{\link{op_slice_update}()} \cr
+\code{\link{op_slogdet}()} \cr
+\code{\link{op_soft_shrink}()} \cr
+\code{\link{op_softmax}()} \cr
+\code{\link{op_softplus}()} \cr
+\code{\link{op_softsign}()} \cr
+\code{\link{op_solve}()} \cr
+\code{\link{op_solve_triangular}()} \cr
+\code{\link{op_sort}()} \cr
+\code{\link{op_sparse_categorical_crossentropy}()} \cr
+\code{\link{op_sparse_plus}()} \cr
+\code{\link{op_sparse_sigmoid}()} \cr
+\code{\link{op_sparsemax}()} \cr
+\code{\link{op_split}()} \cr
+\code{\link{op_sqrt}()} \cr
+\code{\link{op_square}()} \cr
+\code{\link{op_squareplus}()} \cr
+\code{\link{op_squeeze}()} \cr
+\code{\link{op_stack}()} \cr
+\code{\link{op_std}()} \cr
+\code{\link{op_stft}()} \cr
+\code{\link{op_stop_gradient}()} \cr
+\code{\link{op_subset}()} \cr
+\code{\link{op_subtract}()} \cr
+\code{\link{op_sum}()} \cr
+\code{\link{op_svd}()} \cr
+\code{\link{op_swapaxes}()} \cr
+\code{\link{op_switch}()} \cr
+\code{\link{op_take}()} \cr
+\code{\link{op_take_along_axis}()} \cr
+\code{\link{op_tan}()} \cr
+\code{\link{op_tanh}()} \cr
+\code{\link{op_tanh_shrink}()} \cr
+\code{\link{op_tensordot}()} \cr
+\code{\link{op_threshold}()} \cr
+\code{\link{op_tile}()} \cr
+\code{\link{op_top_k}()} \cr
+\code{\link{op_trace}()} \cr
+\code{\link{op_transpose}()} \cr
+\code{\link{op_tri}()} \cr
+\code{\link{op_tril}()} \cr
+\code{\link{op_triu}()} \cr
+\code{\link{op_trunc}()} \cr
+\code{\link{op_unravel_index}()} \cr
+\code{\link{op_unstack}()} \cr
+\code{\link{op_var}()} \cr
+\code{\link{op_vdot}()} \cr
+\code{\link{op_vectorize}()} \cr
+\code{\link{op_vectorized_map}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_vstack}()} \cr
+\code{\link{op_where}()} \cr
+\code{\link{op_while_loop}()} \cr
+\code{\link{op_zeros}()} \cr
+\code{\link{op_zeros_like}()} \cr
+}
+\concept{numpy ops}
+\concept{ops}
diff --git a/man/op_vstack.Rd b/man/op_vstack.Rd
index c16e2ea36..aec21cfc7 100644
--- a/man/op_vstack.Rd
+++ b/man/op_vstack.Rd
@@ -24,6 +24,7 @@ Other numpy ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -40,6 +41,7 @@ Other numpy ops: \cr
 \code{\link{op_argsort}()} \cr
 \code{\link{op_array}()} \cr
 \code{\link{op_average}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_bincount}()} \cr
 \code{\link{op_bitwise_and}()} \cr
 \code{\link{op_bitwise_invert}()} \cr
@@ -48,12 +50,15 @@ Other numpy ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_clip}()} \cr
 \code{\link{op_concatenate}()} \cr
 \code{\link{op_conj}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -62,6 +67,7 @@ Other numpy ops: \cr
 \code{\link{op_ctc_decode}()} \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_diag}()} \cr
 \code{\link{op_diagflat}()} \cr
 \code{\link{op_diagonal}()} \cr
@@ -86,6 +92,9 @@ Other numpy ops: \cr
 \code{\link{op_get_item}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -95,6 +104,7 @@ Other numpy ops: \cr
 \code{\link{op_isfinite}()} \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
+\code{\link{op_kaiser}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
 \code{\link{op_less_equal}()} \cr
@@ -175,6 +185,8 @@ Other numpy ops: \cr
 \code{\link{op_var}()} \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_zeros}()} \cr
 \code{\link{op_zeros_like}()} \cr
@@ -183,6 +195,7 @@ Other ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -201,6 +214,7 @@ Other ops: \cr
 \code{\link{op_associative_scan}()} \cr
 \code{\link{op_average}()} \cr
 \code{\link{op_average_pool}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_batch_normalization}()} \cr
 \code{\link{op_binary_crossentropy}()} \cr
 \code{\link{op_bincount}()} \cr
@@ -211,9 +225,11 @@ Other ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
 \code{\link{op_cast}()} \cr
 \code{\link{op_categorical_crossentropy}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_celu}()} \cr
 \code{\link{op_cholesky}()} \cr
@@ -226,6 +242,7 @@ Other ops: \cr
 \code{\link{op_convert_to_numpy}()} \cr
 \code{\link{op_convert_to_tensor}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -236,6 +253,7 @@ Other ops: \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
 \code{\link{op_custom_gradient}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_depthwise_conv}()} \cr
 \code{\link{op_det}()} \cr
 \code{\link{op_diag}()} \cr
@@ -275,10 +293,13 @@ Other ops: \cr
 \code{\link{op_glu}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
 \code{\link{op_hard_shrink}()} \cr
 \code{\link{op_hard_sigmoid}()} \cr
 \code{\link{op_hard_silu}()} \cr
 \code{\link{op_hard_tanh}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -305,6 +326,8 @@ Other ops: \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
 \code{\link{op_istft}()} \cr
+\code{\link{op_kaiser}()} \cr
+\code{\link{op_layer_normalization}()} \cr
 \code{\link{op_leaky_relu}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
@@ -404,6 +427,7 @@ Other ops: \cr
 \code{\link{op_sort}()} \cr
 \code{\link{op_sparse_categorical_crossentropy}()} \cr
 \code{\link{op_sparse_plus}()} \cr
+\code{\link{op_sparse_sigmoid}()} \cr
 \code{\link{op_sparsemax}()} \cr
 \code{\link{op_split}()} \cr
 \code{\link{op_sqrt}()} \cr
@@ -441,6 +465,8 @@ Other ops: \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
 \code{\link{op_vectorized_map}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_while_loop}()} \cr
 \code{\link{op_zeros}()} \cr
diff --git a/man/op_where.Rd b/man/op_where.Rd
index 4da5af8a6..6d1aae696 100644
--- a/man/op_where.Rd
+++ b/man/op_where.Rd
@@ -29,6 +29,7 @@ Other numpy ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -45,6 +46,7 @@ Other numpy ops: \cr
 \code{\link{op_argsort}()} \cr
 \code{\link{op_array}()} \cr
 \code{\link{op_average}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_bincount}()} \cr
 \code{\link{op_bitwise_and}()} \cr
 \code{\link{op_bitwise_invert}()} \cr
@@ -53,12 +55,15 @@ Other numpy ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_clip}()} \cr
 \code{\link{op_concatenate}()} \cr
 \code{\link{op_conj}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -67,6 +72,7 @@ Other numpy ops: \cr
 \code{\link{op_ctc_decode}()} \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_diag}()} \cr
 \code{\link{op_diagflat}()} \cr
 \code{\link{op_diagonal}()} \cr
@@ -91,6 +97,9 @@ Other numpy ops: \cr
 \code{\link{op_get_item}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -100,6 +109,7 @@ Other numpy ops: \cr
 \code{\link{op_isfinite}()} \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
+\code{\link{op_kaiser}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
 \code{\link{op_less_equal}()} \cr
@@ -180,6 +190,8 @@ Other numpy ops: \cr
 \code{\link{op_var}()} \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_zeros}()} \cr
 \code{\link{op_zeros_like}()} \cr
@@ -188,6 +200,7 @@ Other ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -206,6 +219,7 @@ Other ops: \cr
 \code{\link{op_associative_scan}()} \cr
 \code{\link{op_average}()} \cr
 \code{\link{op_average_pool}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_batch_normalization}()} \cr
 \code{\link{op_binary_crossentropy}()} \cr
 \code{\link{op_bincount}()} \cr
@@ -216,9 +230,11 @@ Other ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
 \code{\link{op_cast}()} \cr
 \code{\link{op_categorical_crossentropy}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_celu}()} \cr
 \code{\link{op_cholesky}()} \cr
@@ -231,6 +247,7 @@ Other ops: \cr
 \code{\link{op_convert_to_numpy}()} \cr
 \code{\link{op_convert_to_tensor}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -241,6 +258,7 @@ Other ops: \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
 \code{\link{op_custom_gradient}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_depthwise_conv}()} \cr
 \code{\link{op_det}()} \cr
 \code{\link{op_diag}()} \cr
@@ -280,10 +298,13 @@ Other ops: \cr
 \code{\link{op_glu}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
 \code{\link{op_hard_shrink}()} \cr
 \code{\link{op_hard_sigmoid}()} \cr
 \code{\link{op_hard_silu}()} \cr
 \code{\link{op_hard_tanh}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -310,6 +331,8 @@ Other ops: \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
 \code{\link{op_istft}()} \cr
+\code{\link{op_kaiser}()} \cr
+\code{\link{op_layer_normalization}()} \cr
 \code{\link{op_leaky_relu}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
@@ -409,6 +432,7 @@ Other ops: \cr
 \code{\link{op_sort}()} \cr
 \code{\link{op_sparse_categorical_crossentropy}()} \cr
 \code{\link{op_sparse_plus}()} \cr
+\code{\link{op_sparse_sigmoid}()} \cr
 \code{\link{op_sparsemax}()} \cr
 \code{\link{op_split}()} \cr
 \code{\link{op_sqrt}()} \cr
@@ -446,6 +470,8 @@ Other ops: \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
 \code{\link{op_vectorized_map}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_while_loop}()} \cr
 \code{\link{op_zeros}()} \cr
diff --git a/man/op_while_loop.Rd b/man/op_while_loop.Rd
index eabd05ba1..a480b115d 100644
--- a/man/op_while_loop.Rd
+++ b/man/op_while_loop.Rd
@@ -98,6 +98,7 @@ Other ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -116,6 +117,7 @@ Other ops: \cr
 \code{\link{op_associative_scan}()} \cr
 \code{\link{op_average}()} \cr
 \code{\link{op_average_pool}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_batch_normalization}()} \cr
 \code{\link{op_binary_crossentropy}()} \cr
 \code{\link{op_bincount}()} \cr
@@ -126,9 +128,11 @@ Other ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
 \code{\link{op_cast}()} \cr
 \code{\link{op_categorical_crossentropy}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_celu}()} \cr
 \code{\link{op_cholesky}()} \cr
@@ -141,6 +145,7 @@ Other ops: \cr
 \code{\link{op_convert_to_numpy}()} \cr
 \code{\link{op_convert_to_tensor}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -151,6 +156,7 @@ Other ops: \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
 \code{\link{op_custom_gradient}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_depthwise_conv}()} \cr
 \code{\link{op_det}()} \cr
 \code{\link{op_diag}()} \cr
@@ -190,10 +196,13 @@ Other ops: \cr
 \code{\link{op_glu}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
 \code{\link{op_hard_shrink}()} \cr
 \code{\link{op_hard_sigmoid}()} \cr
 \code{\link{op_hard_silu}()} \cr
 \code{\link{op_hard_tanh}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -220,6 +229,8 @@ Other ops: \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
 \code{\link{op_istft}()} \cr
+\code{\link{op_kaiser}()} \cr
+\code{\link{op_layer_normalization}()} \cr
 \code{\link{op_leaky_relu}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
@@ -319,6 +330,7 @@ Other ops: \cr
 \code{\link{op_sort}()} \cr
 \code{\link{op_sparse_categorical_crossentropy}()} \cr
 \code{\link{op_sparse_plus}()} \cr
+\code{\link{op_sparse_sigmoid}()} \cr
 \code{\link{op_sparsemax}()} \cr
 \code{\link{op_split}()} \cr
 \code{\link{op_sqrt}()} \cr
@@ -356,6 +368,8 @@ Other ops: \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
 \code{\link{op_vectorized_map}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_zeros}()} \cr
diff --git a/man/op_zeros.Rd b/man/op_zeros.Rd
index b6a275d6c..851ad5806 100644
--- a/man/op_zeros.Rd
+++ b/man/op_zeros.Rd
@@ -26,6 +26,7 @@ Other numpy ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -42,6 +43,7 @@ Other numpy ops: \cr
 \code{\link{op_argsort}()} \cr
 \code{\link{op_array}()} \cr
 \code{\link{op_average}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_bincount}()} \cr
 \code{\link{op_bitwise_and}()} \cr
 \code{\link{op_bitwise_invert}()} \cr
@@ -50,12 +52,15 @@ Other numpy ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_clip}()} \cr
 \code{\link{op_concatenate}()} \cr
 \code{\link{op_conj}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -64,6 +69,7 @@ Other numpy ops: \cr
 \code{\link{op_ctc_decode}()} \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_diag}()} \cr
 \code{\link{op_diagflat}()} \cr
 \code{\link{op_diagonal}()} \cr
@@ -88,6 +94,9 @@ Other numpy ops: \cr
 \code{\link{op_get_item}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -97,6 +106,7 @@ Other numpy ops: \cr
 \code{\link{op_isfinite}()} \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
+\code{\link{op_kaiser}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
 \code{\link{op_less_equal}()} \cr
@@ -177,6 +187,8 @@ Other numpy ops: \cr
 \code{\link{op_var}()} \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_zeros_like}()} \cr
@@ -185,6 +197,7 @@ Other ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -203,6 +216,7 @@ Other ops: \cr
 \code{\link{op_associative_scan}()} \cr
 \code{\link{op_average}()} \cr
 \code{\link{op_average_pool}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_batch_normalization}()} \cr
 \code{\link{op_binary_crossentropy}()} \cr
 \code{\link{op_bincount}()} \cr
@@ -213,9 +227,11 @@ Other ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
 \code{\link{op_cast}()} \cr
 \code{\link{op_categorical_crossentropy}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_celu}()} \cr
 \code{\link{op_cholesky}()} \cr
@@ -228,6 +244,7 @@ Other ops: \cr
 \code{\link{op_convert_to_numpy}()} \cr
 \code{\link{op_convert_to_tensor}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -238,6 +255,7 @@ Other ops: \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
 \code{\link{op_custom_gradient}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_depthwise_conv}()} \cr
 \code{\link{op_det}()} \cr
 \code{\link{op_diag}()} \cr
@@ -277,10 +295,13 @@ Other ops: \cr
 \code{\link{op_glu}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
 \code{\link{op_hard_shrink}()} \cr
 \code{\link{op_hard_sigmoid}()} \cr
 \code{\link{op_hard_silu}()} \cr
 \code{\link{op_hard_tanh}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -307,6 +328,8 @@ Other ops: \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
 \code{\link{op_istft}()} \cr
+\code{\link{op_kaiser}()} \cr
+\code{\link{op_layer_normalization}()} \cr
 \code{\link{op_leaky_relu}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
@@ -406,6 +429,7 @@ Other ops: \cr
 \code{\link{op_sort}()} \cr
 \code{\link{op_sparse_categorical_crossentropy}()} \cr
 \code{\link{op_sparse_plus}()} \cr
+\code{\link{op_sparse_sigmoid}()} \cr
 \code{\link{op_sparsemax}()} \cr
 \code{\link{op_split}()} \cr
 \code{\link{op_sqrt}()} \cr
@@ -443,6 +467,8 @@ Other ops: \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
 \code{\link{op_vectorized_map}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_while_loop}()} \cr
diff --git a/man/op_zeros_like.Rd b/man/op_zeros_like.Rd
index 6cabe8940..d52b81434 100644
--- a/man/op_zeros_like.Rd
+++ b/man/op_zeros_like.Rd
@@ -26,6 +26,7 @@ Other numpy ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -42,6 +43,7 @@ Other numpy ops: \cr
 \code{\link{op_argsort}()} \cr
 \code{\link{op_array}()} \cr
 \code{\link{op_average}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_bincount}()} \cr
 \code{\link{op_bitwise_and}()} \cr
 \code{\link{op_bitwise_invert}()} \cr
@@ -50,12 +52,15 @@ Other numpy ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_clip}()} \cr
 \code{\link{op_concatenate}()} \cr
 \code{\link{op_conj}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -64,6 +69,7 @@ Other numpy ops: \cr
 \code{\link{op_ctc_decode}()} \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_diag}()} \cr
 \code{\link{op_diagflat}()} \cr
 \code{\link{op_diagonal}()} \cr
@@ -88,6 +94,9 @@ Other numpy ops: \cr
 \code{\link{op_get_item}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -97,6 +106,7 @@ Other numpy ops: \cr
 \code{\link{op_isfinite}()} \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
+\code{\link{op_kaiser}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
 \code{\link{op_less_equal}()} \cr
@@ -177,6 +187,8 @@ Other numpy ops: \cr
 \code{\link{op_var}()} \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_zeros}()} \cr
@@ -185,6 +197,7 @@ Other ops: \cr
 \code{\link{op_abs}()} \cr
 \code{\link{op_add}()} \cr
 \code{\link{op_all}()} \cr
+\code{\link{op_angle}()} \cr
 \code{\link{op_any}()} \cr
 \code{\link{op_append}()} \cr
 \code{\link{op_arange}()} \cr
@@ -203,6 +216,7 @@ Other ops: \cr
 \code{\link{op_associative_scan}()} \cr
 \code{\link{op_average}()} \cr
 \code{\link{op_average_pool}()} \cr
+\code{\link{op_bartlett}()} \cr
 \code{\link{op_batch_normalization}()} \cr
 \code{\link{op_binary_crossentropy}()} \cr
 \code{\link{op_bincount}()} \cr
@@ -213,9 +227,11 @@ Other ops: \cr
 \code{\link{op_bitwise_or}()} \cr
 \code{\link{op_bitwise_right_shift}()} \cr
 \code{\link{op_bitwise_xor}()} \cr
+\code{\link{op_blackman}()} \cr
 \code{\link{op_broadcast_to}()} \cr
 \code{\link{op_cast}()} \cr
 \code{\link{op_categorical_crossentropy}()} \cr
+\code{\link{op_cbrt}()} \cr
 \code{\link{op_ceil}()} \cr
 \code{\link{op_celu}()} \cr
 \code{\link{op_cholesky}()} \cr
@@ -228,6 +244,7 @@ Other ops: \cr
 \code{\link{op_convert_to_numpy}()} \cr
 \code{\link{op_convert_to_tensor}()} \cr
 \code{\link{op_copy}()} \cr
+\code{\link{op_corrcoef}()} \cr
 \code{\link{op_correlate}()} \cr
 \code{\link{op_cos}()} \cr
 \code{\link{op_cosh}()} \cr
@@ -238,6 +255,7 @@ Other ops: \cr
 \code{\link{op_cumprod}()} \cr
 \code{\link{op_cumsum}()} \cr
 \code{\link{op_custom_gradient}()} \cr
+\code{\link{op_deg2rad}()} \cr
 \code{\link{op_depthwise_conv}()} \cr
 \code{\link{op_det}()} \cr
 \code{\link{op_diag}()} \cr
@@ -277,10 +295,13 @@ Other ops: \cr
 \code{\link{op_glu}()} \cr
 \code{\link{op_greater}()} \cr
 \code{\link{op_greater_equal}()} \cr
+\code{\link{op_hamming}()} \cr
+\code{\link{op_hanning}()} \cr
 \code{\link{op_hard_shrink}()} \cr
 \code{\link{op_hard_sigmoid}()} \cr
 \code{\link{op_hard_silu}()} \cr
 \code{\link{op_hard_tanh}()} \cr
+\code{\link{op_heaviside}()} \cr
 \code{\link{op_histogram}()} \cr
 \code{\link{op_hstack}()} \cr
 \code{\link{op_identity}()} \cr
@@ -307,6 +328,8 @@ Other ops: \cr
 \code{\link{op_isinf}()} \cr
 \code{\link{op_isnan}()} \cr
 \code{\link{op_istft}()} \cr
+\code{\link{op_kaiser}()} \cr
+\code{\link{op_layer_normalization}()} \cr
 \code{\link{op_leaky_relu}()} \cr
 \code{\link{op_left_shift}()} \cr
 \code{\link{op_less}()} \cr
@@ -406,6 +429,7 @@ Other ops: \cr
 \code{\link{op_sort}()} \cr
 \code{\link{op_sparse_categorical_crossentropy}()} \cr
 \code{\link{op_sparse_plus}()} \cr
+\code{\link{op_sparse_sigmoid}()} \cr
 \code{\link{op_sparsemax}()} \cr
 \code{\link{op_split}()} \cr
 \code{\link{op_sqrt}()} \cr
@@ -443,6 +467,8 @@ Other ops: \cr
 \code{\link{op_vdot}()} \cr
 \code{\link{op_vectorize}()} \cr
 \code{\link{op_vectorized_map}()} \cr
+\code{\link{op_view_as_complex}()} \cr
+\code{\link{op_view_as_real}()} \cr
 \code{\link{op_vstack}()} \cr
 \code{\link{op_where}()} \cr
 \code{\link{op_while_loop}()} \cr

From 52a3823074e8191ee1dcecbb95baaf3d6926876d Mon Sep 17 00:00:00 2001
From: Tomasz Kalinowski <kalinowskit@gmail.com>
Date: Fri, 26 Sep 2025 12:52:37 -0400
Subject: [PATCH 56/81] add `op_image_elastic_transform()`

---
 .tether/man/keras.ops.image.txt | 10 ++++
 R/ops-image.R                   | 88 +++++++++++++++++++++++++++++++++
 2 files changed, 98 insertions(+)

diff --git a/.tether/man/keras.ops.image.txt b/.tether/man/keras.ops.image.txt
index 242330155..4864fdf94 100644
--- a/.tether/man/keras.ops.image.txt
+++ b/.tether/man/keras.ops.image.txt
@@ -16,6 +16,16 @@ crop_images(
   target_width=None,
   data_format=None
 )
+elastic_transform(
+  images,
+  alpha=20.0,
+  sigma=5.0,
+  interpolation='bilinear',
+  fill_mode='reflect',
+  fill_value=0.0,
+  seed=None,
+  data_format=None
+)
 extract_patches(
   images,
   size,
diff --git a/R/ops-image.R b/R/ops-image.R
index 00363bce8..3a52d4f50 100644
--- a/R/ops-image.R
+++ b/R/ops-image.R
@@ -475,6 +475,94 @@ function (images, top_cropping = NULL, left_cropping = NULL,
   do.call(ops$image$crop_images, args)
 }
 
+#' Applies elastic deformation to the image(s).
+#'
+#' @description
+#' Apply random elastic deformation to 3D or 4D image tensors.
+#'
+#' # Examples
+#' ```{r}
+#' x <- random_uniform(c(2, 64, 80, 3))  # batch of 2 RGB images
+#' y <- op_image_elastic_transform(x)
+#' op_shape(y)
+#' ```
+#'
+#' ```{r}
+#' x <- random_uniform(c(64, 80, 3))  # single RGB image
+#' y <- op_image_elastic_transform(x)
+#' op_shape(y)
+#' ```
+#'
+#' ```{r}
+#' x <- random_uniform(c(2, 3, 64, 80))  # batch of 2 RGB images
+#' y <- op_image_elastic_transform(
+#'   x,
+#'   data_format = "channels_first",
+#'   seed = 123
+#' )
+#' op_shape(y)
+#' ```
+#'
+#' @returns
+#' Transformed image or batch of images with elastic deformation.
+#'
+#' @param images
+#' Input image or batch of images. Must be 3D or 4D.
+#'
+#' @param alpha
+#' Scaling factor that controls the intensity of the deformation.
+#'
+#' @param sigma
+#' Standard deviation of the Gaussian filter used for
+#' smoothing the displacement fields.
+#'
+#' @param interpolation
+#' Interpolation method. Available methods are `"nearest"`,
+#' and `"bilinear"`. Defaults to `"bilinear"`.
+#'
+#' @param fill_mode
+#' Points outside the boundaries of the input are filled
+#' according to the given mode. Available methods are `"constant"`,
+#' `"nearest"`, `"wrap"` and `"reflect"`. Defaults to `"reflect"`.
+#' - `"reflect"`: `(d c b a | a b c d | d c b a)`
+#'     The input is extended by reflecting about the edge of the last
+#'     pixel.
+#' - `"constant"`: `(k k k k | a b c d | k k k k)`
+#'     The input is extended by filling all values beyond
+#'     the edge with the same constant value `k` specified by
+#'     `fill_value`.
+#' - `"wrap"`: `(a b c d | a b c d | a b c d)`
+#'     The input is extended by wrapping around to the opposite edge.
+#' - `"nearest"`: `(a a a a | a b c d | d d d d)`
+#'     The input is extended by the nearest pixel.
+#'
+#' @param fill_value
+#' Value used for points outside the boundaries of the input if
+#' `fill_mode="constant"`. Defaults to `0`.
+#'
+#' @param seed
+#' Optional integer seed for the random number generator.
+#'
+#' @param data_format
+#' A string specifying the data format of the input tensor.
+#' It can be either `"channels_last"` or `"channels_first"`.
+#' `"channels_last"` corresponds to inputs with shape
+#' `(batch, height, width, channels)`, while `"channels_first"`
+#' corresponds to inputs with shape `(batch, channels, height, width)`.
+#' If not specified, the value will default to
+#' `keras.config.image_data_format`.
+#'
+#' @export
+#' @tether keras.ops.image.elastic_transform
+op_image_elastic_transform <-
+function (images, alpha = 20, sigma = 5, interpolation = "bilinear",
+    fill_mode = "reflect", fill_value = 0, seed = NULL, data_format = NULL)
+{
+    args <- capture_args(list(seed = as_integer))
+    do.call(keras$ops$image$elastic_transform, args)
+}
+
+
 #' Convert RGB images to grayscale.
 #'
 #' @description

From 1a10cd49493d8c9bcb2c7013563345a2f08256ee Mon Sep 17 00:00:00 2001
From: Tomasz Kalinowski <kalinowskit@gmail.com>
Date: Fri, 26 Sep 2025 13:05:13 -0400
Subject: [PATCH 57/81] update `op_searchsorted()` docs

---
 .tether/man/keras.ops.numpy.txt |  5 +++++
 R/ops.R                         | 27 +++++++++++++++++----------
 2 files changed, 22 insertions(+), 10 deletions(-)

diff --git a/.tether/man/keras.ops.numpy.txt b/.tether/man/keras.ops.numpy.txt
index f2a7ad91c..74095b6b6 100644
--- a/.tether/man/keras.ops.numpy.txt
+++ b/.tether/man/keras.ops.numpy.txt
@@ -303,6 +303,11 @@ rot90(
   axes=(0, 1)
 )
 round(x, decimals=0)
+searchsorted(
+  sorted_sequence,
+  values,
+  side='left'
+)
 select(
   condlist,
   choicelist,
diff --git a/R/ops.R b/R/ops.R
index 657b08f40..6cc1f7045 100644
--- a/R/ops.R
+++ b/R/ops.R
@@ -361,35 +361,42 @@ function (inputs, indices, updates)
     do.call(ops$scatter_update, args)
 }
 
-#' Perform a binary search
+#' Perform a binary search.
 #'
 #' @description
-#' Perform a binary search, returning indices for insertion of `values`
-#' into `sorted_sequence` that maintain the sorting order.
+#' Return insertion indices that keep `values` in sorted order when placed
+#' into `sorted_sequence`.
+#'
+#' # Examples
+#' ```{r}
+#' sorted <- op_array(c(-1, 0, 2, 4))
+#' queries <- c(-2, 0, 3, 7)
+#' op_searchsorted(sorted, queries)
+#' op_searchsorted(sorted, queries, side = "right", zero_indexed = TRUE)
+#' ```
 #'
 #' @returns
-#' Tensor of insertion indices of same shape as `values`.
+#' Tensor of insertion indices with the same shape as `values`.
 #'
 #' @param sorted_sequence
-#' 1-D input tensor, sorted along the innermost
-#' dimension.
+#' 1-D input tensor that is sorted along its innermost dimension.
 #'
 #' @param values
 #' N-D tensor of query insertion values.
 #'
 #' @param side
-#' `'left'` or `'right'`, specifying the direction in which to insert
-#' for the equality case (tie-breaker).
+#' Either `"left"` or `"right"`, choosing which side to insert on ties.
 #'
 #' @param zero_indexed
-#' If `TRUE`, the returned indices are zero-based (`0` encodes to first
+#' If `TRUE`, the returned indices are zero-based (`0` encodes the first
 #' position); if `FALSE` (default), the returned indices are one-based (`1`
-#' encodes to first position).
+#' encodes the first position).
 #'
 #' @export
 #' @family core ops
 #' @family ops
 #' @tether keras.ops.searchsorted
+#' @tether keras.ops.numpy.searchsorted
 op_searchsorted <-
 function (sorted_sequence, values, side = "left", zero_indexed = FALSE) {
 

From dff5ed0f7a6ab2e8853f0dee9f1db6a15b25e8e1 Mon Sep 17 00:00:00 2001
From: Tomasz Kalinowski <kalinowskit@gmail.com>
Date: Fri, 26 Sep 2025 14:05:07 -0400
Subject: [PATCH 58/81] remove dup tether

---
 R/ops.R | 1 -
 1 file changed, 1 deletion(-)

diff --git a/R/ops.R b/R/ops.R
index 6cc1f7045..1b459c836 100644
--- a/R/ops.R
+++ b/R/ops.R
@@ -396,7 +396,6 @@ function (inputs, indices, updates)
 #' @family core ops
 #' @family ops
 #' @tether keras.ops.searchsorted
-#' @tether keras.ops.numpy.searchsorted
 op_searchsorted <-
 function (sorted_sequence, values, side = "left", zero_indexed = FALSE) {
 

From 0816b431635fc8e56c1c145e3d3daae2cb1941f5 Mon Sep 17 00:00:00 2001
From: Tomasz Kalinowski <kalinowskit@gmail.com>
Date: Fri, 26 Sep 2025 14:20:27 -0400
Subject: [PATCH 59/81] add `optimizer_muon()`

---
 .tether/man/keras.optimizers.txt |  26 +++++++
 R/optimizers.R                   | 115 +++++++++++++++++++++++++++++++
 2 files changed, 141 insertions(+)

diff --git a/.tether/man/keras.optimizers.txt b/.tether/man/keras.optimizers.txt
index 6b8c30ea9..557b2e339 100644
--- a/.tether/man/keras.optimizers.txt
+++ b/.tether/man/keras.optimizers.txt
@@ -164,6 +164,32 @@ LossScaleOptimizer(
   dynamic_growth_steps=2000,
   **kwargs
 )
+Muon(
+  learning_rate=0.001,
+  adam_beta_1=0.9,
+  adam_beta_2=0.999,
+  epsilon=1e-07,
+  weight_decay=0.1,
+  clipnorm=None,
+  clipvalue=None,
+  global_clipnorm=None,
+  use_ema=False,
+  ema_momentum=0.99,
+  ema_overwrite_frequency=None,
+  loss_scale_factor=None,
+  gradient_accumulation_steps=None,
+  name='muon',
+  exclude_layers=None,
+  exclude_embeddings=True,
+  muon_a=3.4445,
+  muon_b=-4.775,
+  muon_c=2.0315,
+  adam_lr_ratio=0.1,
+  momentum=0.95,
+  ns_steps=6,
+  nesterov=True,
+  **kwargs
+)
 Nadam(
   learning_rate=0.001,
   beta_1=0.9,
diff --git a/R/optimizers.R b/R/optimizers.R
index f480e2992..7847fc0b1 100644
--- a/R/optimizers.R
+++ b/R/optimizers.R
@@ -1270,6 +1270,121 @@ function (inner_optimizer, initial_scale = 32768, dynamic_growth_steps = 2000L,
     do.call(keras$optimizers$LossScaleOptimizer, args)
 }
 
+#' Optimizer that implements the Muon algorithm.
+#'
+#' @description
+#' Note that this optimizer should not be used in the following layers:
+#'
+#' 1. Embedding layer
+#' 2. Final output fully connected layer
+#' 3. Any {0,1}-D variables
+#'
+#' These should all be optimized using AdamW.
+#'
+#' The Muon optimizer can use both the Muon update step or the
+#' AdamW update step based on the following:
+#'
+#' - For any variable that isn't 2D, 3D or 4D, the AdamW step
+#'     will be used. This is not configurable.
+#' - If the argument `exclude_embeddings` (defaults to `TRUE`) is set
+#' to `TRUE`, the AdamW step will be used.
+#' - For any variablewith a name that matches an expression
+#'     listed in the argument `exclude_layers` (a list), the
+#'     AdamW step will be used.
+#' - Any other variable uses the Muon step.
+#'
+#' Typically, you only need to pass the name of your densely-connected
+#' output layer to `exclude_layers`, e.g.
+#' `exclude_layers=["output_dense"]`.
+#'
+#' # References
+#' - [Original implementation](https://github.com/KellerJordan/Muon)
+#' - [Liu et al, 2025](https://arxiv.org/abs/2502.16982)
+#'
+#' @param learning_rate
+#' A float,
+#' `LearningRateSchedule()` instance, or
+#' a callable that takes no arguments and returns the actual value to
+#' use. The learning rate. Defaults to `0.001`.
+#'
+#' @param adam_beta_1
+#' A float value or a constant float tensor, or a callable
+#' that takes no arguments and returns the actual value to use.
+#' The exponential decay rate for the 1st moment estimates. Defaults to
+#' `0.9`.
+#'
+#' @param adam_beta_2
+#' A float value or a constant float tensor, ora callable
+#' that takes no arguments and returns the actual value to use.
+#' The exponential decay rate for the 2nd moment estimates. Defaults to
+#' `0.999`.
+#'
+#' @param epsilon
+#' A small constant for numerical stability. This is
+#' "epsilon hat" in the Kingma and Ba paper
+#' (in the formula just before Section 2.1),
+#' not the epsilon in Algorithm 1 of the paper.
+#' It be used at Adamw.Defaults to `1e-7`.
+#'
+#' @param exclude_layers
+#' List of strings, keywords of layer names to exclude.
+#' All layers with keywords in their path will use adamw.
+#'
+#' @param exclude_embeddings
+#' Boolean value
+#' If `TRUE`, embedding layers will use adamw.
+#'
+#' @param muon_a
+#' Float, parameter a of the muon algorithm.
+#' It is recommended to use the default value.
+#'
+#' @param muon_b
+#' Float, parameter b of the muon algorithm.
+#' It is recommended to use the default value.
+#'
+#' @param muon_c
+#' Float, parameter c of the muon algorithm.
+#' It is recommended to use the default value.
+#'
+#' @param adam_lr_ratio
+#' Float, the ratio of the learning rate when
+#' using Adam to the main learning rate.
+#' it is recommended to set it to `0.1`.
+#'
+#' @param momentum
+#' Float, momentum used by internal SGD.
+#'
+#' @param ns_steps
+#' Integer, number of Newton-Schulz iterations to run.
+#'
+#' @param nesterov
+#' Boolean, whether to use Nesterov-style momentum.
+#'
+#' @param name
+#' String, name for the object
+#'
+#' @param ...
+#' For forward/backward compatibility.
+#'
+#' @export
+#' @tether keras.optimizers.Muon
+optimizer_muon <-
+function (learning_rate = 0.001, adam_beta_1 = 0.9, adam_beta_2 = 0.999,
+    epsilon = 1e-07, weight_decay = 0.1, clipnorm = NULL, clipvalue = NULL,
+    global_clipnorm = NULL, use_ema = FALSE, ema_momentum = 0.99,
+    ema_overwrite_frequency = NULL, loss_scale_factor = NULL,
+    gradient_accumulation_steps = NULL, name = "muon", exclude_layers = NULL,
+    exclude_embeddings = TRUE, muon_a = 3.4445, muon_b = -4.775,
+    muon_c = 2.0315, adam_lr_ratio = 0.1, momentum = 0.95, ns_steps = 6L,
+    nesterov = TRUE, ...)
+{
+    args <- capture_args(list(
+        ema_overwrite_frequency = as_integer,
+        gradient_accumulation_steps = as_integer,
+        ns_steps = as_integer))
+    do.call(keras$optimizers$Muon, args)
+}
+
 
 #' Optimizer that implements the Nadam algorithm.
 #'

From 4175115d25815f2eedbffb52069ead7609c56f11 Mon Sep 17 00:00:00 2001
From: Tomasz Kalinowski <kalinowskit@gmail.com>
Date: Fri, 26 Sep 2025 15:58:05 -0400
Subject: [PATCH 60/81] add `layer_random_elastic_transform()`

---
 .tether/man/keras.layers.txt | 22 ++++++++-
 R/layers-preprocessing.R     | 89 ++++++++++++++++++++++++++++++++++++
 R/optimizers.R               | 12 ++---
 3 files changed, 116 insertions(+), 7 deletions(-)

diff --git a/.tether/man/keras.layers.txt b/.tether/man/keras.layers.txt
index 91187291b..c6af1f082 100644
--- a/.tether/man/keras.layers.txt
+++ b/.tether/man/keras.layers.txt
@@ -153,6 +153,7 @@ Conv1DTranspose(
   kernel_size,
   strides=1,
   padding='valid',
+  output_padding=None,
   data_format=None,
   dilation_rate=1,
   activation=None,
@@ -190,6 +191,7 @@ Conv2DTranspose(
   kernel_size,
   strides=(1, 1),
   padding='valid',
+  output_padding=None,
   data_format=None,
   dilation_rate=(1, 1),
   activation=None,
@@ -228,6 +230,7 @@ Conv3DTranspose(
   strides=(1, 1, 1),
   padding='valid',
   data_format=None,
+  output_padding=None,
   dilation_rate=(1, 1, 1),
   activation=None,
   use_bias=True,
@@ -354,6 +357,7 @@ Convolution1DTranspose(
   kernel_size,
   strides=1,
   padding='valid',
+  output_padding=None,
   data_format=None,
   dilation_rate=1,
   activation=None,
@@ -391,6 +395,7 @@ Convolution2DTranspose(
   kernel_size,
   strides=(1, 1),
   padding='valid',
+  output_padding=None,
   data_format=None,
   dilation_rate=(1, 1),
   activation=None,
@@ -429,6 +434,7 @@ Convolution3DTranspose(
   strides=(1, 1, 1),
   padding='valid',
   data_format=None,
+  output_padding=None,
   dilation_rate=(1, 1, 1),
   activation=None,
   use_bias=True,
@@ -470,6 +476,7 @@ Dense(
   kernel_constraint=None,
   bias_constraint=None,
   lora_rank=None,
+  lora_alpha=None,
   **kwargs
 )
 DepthwiseConv1D(
@@ -546,6 +553,7 @@ EinsumDense(
   kernel_constraint=None,
   bias_constraint=None,
   lora_rank=None,
+  lora_alpha=None,
   **kwargs
 )
 ELU(alpha=1.0, **kwargs)
@@ -558,6 +566,7 @@ Embedding(
   mask_zero=False,
   weights=None,
   lora_rank=None,
+  lora_alpha=None,
   **kwargs
 )
 Equalization(
@@ -808,7 +817,6 @@ LayerNormalization(
   epsilon=0.001,
   center=True,
   scale=True,
-  rms_scaling=False,
   beta_initializer='zeros',
   gamma_initializer='ones',
   beta_regularizer=None,
@@ -1028,6 +1036,17 @@ RandomCrop(
   name=None,
   **kwargs
 )
+RandomElasticTransform(
+  factor=1.0,
+  scale=1.0,
+  interpolation='bilinear',
+  fill_mode='reflect',
+  fill_value=0.0,
+  value_range=(0, 255),
+  seed=None,
+  data_format=None,
+  **kwargs
+)
 RandomErasing(
   factor=1.0,
   scale=(0.02, 0.33),
@@ -1404,6 +1423,7 @@ TimeDistributed(layer, **kwargs)
 TorchModuleWrapper(
   module,
   name=None,
+  output_shape=None,
   **kwargs
 )
 UnitNormalization(axis=-1, **kwargs)
diff --git a/R/layers-preprocessing.R b/R/layers-preprocessing.R
index 77d5313bc..994b7fe57 100644
--- a/R/layers-preprocessing.R
+++ b/R/layers-preprocessing.R
@@ -2659,6 +2659,95 @@ function (object, factor = 1, scale = list(0.02, 0.33), fill_value = NULL,
     create_layer(keras$layers$RandomErasing, object, args)
 }
 
+
+#' A preprocessing layer that applies random elastic transformations.
+#'
+#' @description
+#' This layer distorts input images by applying elastic deformations,
+#' simulating a physically realistic transformation. The magnitude of the
+#' distortion is controlled by the `scale` parameter, while the `factor`
+#' determines the probability of applying the transformation.
+#'
+#' @param factor
+#' A single float or a tuple of two floats.
+#' `factor` controls the probability of applying the transformation.
+#' - `factor = 0.0` ensures no transformation is applied.
+#' - `factor = 1.0` means the transformation is always applied.
+#' - If a tuple `(min, max)` is provided, a probability value
+#'   is sampled between `min` and `max` for each image.
+#' - If a single float is provided, a probability is sampled
+#'   between `0.0` and the given float.
+#' Default is `1.0`.
+#'
+#' @param scale
+#' A float or a tuple of two floats defining the magnitude of
+#' the distortion applied.
+#' - If a tuple `(min, max)` is provided, a random scale value is
+#'   sampled within this range.
+#' - If a single float is provided, a random scale value is sampled
+#'   between `0.0` and the given float.
+#' Default is `1.0`.
+#'
+#' @param interpolation
+#' Interpolation mode. Supported values: `"nearest"`,
+#' `"bilinear"`.
+#'
+#' @param fill_mode
+#' Points outside the boundaries of the input are filled
+#' according to the given mode. Available methods are `"constant"`,
+#' `"nearest"`, `"wrap"` and `"reflect"`. Defaults to `"reflect"`.
+#' - `"reflect"`: `(d c b a | a b c d | d c b a)`
+#'     The input is extended by reflecting about the edge of the last
+#'     pixel.
+#' - `"constant"`: `(k k k k | a b c d | k k k k)`
+#'     The input is extended by filling all values beyond
+#'     the edge with the same constant value `k` specified by
+#'     `fill_value`.
+#' - `"wrap"`: `(a b c d | a b c d | a b c d)`
+#'     The input is extended by wrapping around to the opposite edge.
+#' - `"nearest"`: `(a a a a | a b c d | d d d d)`
+#'     The input is extended by the nearest pixel.
+#' When using the torch backend, `"reflect"` is redirected to
+#' `"mirror"` because torch does not support `"reflect"`.
+#' The torch backend also does not support `"wrap"`.
+#'
+#' @param fill_value
+#' A float representing the value to fill outside the boundaries when
+#' `fill_mode = "constant"`.
+#'
+#' @param value_range
+#' The range of values the incoming images will have.
+#' Represented as a two-number tuple written `[low, high]`. This is
+#' typically either `[0, 1]` or `[0, 255]` depending on how your
+#' preprocessing pipeline is set up.
+#'
+#' @param seed
+#' Integer. Used to create a random seed.
+#'
+#' @param object
+#' Object to compose the layer with. A tensor, array, or sequential model.
+#'
+#' @param ...
+#' For forward/backward compatability.
+#'
+#' @inheritParams layer_center_crop
+#'
+#' @export
+#' @tether keras.layers.RandomElasticTransform
+#' @family image preprocessing layers
+#' @family preprocessing layers
+#' @family layers
+layer_random_elastic_transform <-
+function (object, factor = 1, scale = 1, interpolation = "bilinear",
+    fill_mode = "reflect", fill_value = 0, value_range = list(0L, 255L),
+    seed = NULL, data_format = NULL, ...)
+{
+    args <- capture_args(list(seed = as_integer, input_shape = normalize_shape,
+        batch_size = as_integer, batch_input_shape = normalize_shape),
+        ignore = "object")
+    create_layer(keras$layers$RandomElasticTransform, object, args)
+}
+
 #' Applies random Gaussian blur to images for data augmentation.
 #'
 #' @description
diff --git a/R/optimizers.R b/R/optimizers.R
index 7847fc0b1..1b8a884df 100644
--- a/R/optimizers.R
+++ b/R/optimizers.R
@@ -1288,14 +1288,14 @@ function (inner_optimizer, initial_scale = 32768, dynamic_growth_steps = 2000L,
 #'     will be used. This is not configurable.
 #' - If the argument `exclude_embeddings` (defaults to `TRUE`) is set
 #' to `TRUE`, the AdamW step will be used.
-#' - For any variablewith a name that matches an expression
+#' - For any variable with a name that matches an expression
 #'     listed in the argument `exclude_layers` (a list), the
 #'     AdamW step will be used.
 #' - Any other variable uses the Muon step.
 #'
 #' Typically, you only need to pass the name of your densely-connected
 #' output layer to `exclude_layers`, e.g.
-#' `exclude_layers=["output_dense"]`.
+#' `exclude_layers = "output_dense"`.
 #'
 #' # References
 #' - [Original implementation](https://github.com/KellerJordan/Muon)
@@ -1324,15 +1324,15 @@ function (inner_optimizer, initial_scale = 32768, dynamic_growth_steps = 2000L,
 #' "epsilon hat" in the Kingma and Ba paper
 #' (in the formula just before Section 2.1),
 #' not the epsilon in Algorithm 1 of the paper.
-#' It be used at Adamw.Defaults to `1e-7`.
+#' It is used as in AdamW. Defaults to `1e-7`.
 #'
 #' @param exclude_layers
 #' List of strings, keywords of layer names to exclude.
-#' All layers with keywords in their path will use adamw.
+#' All layers with keywords in their path will use AdamW.
 #'
 #' @param exclude_embeddings
-#' Boolean value
-#' If `TRUE`, embedding layers will use adamw.
+#' Boolean value.
+#' If `TRUE`, embedding layers will use AdamW.
 #'
 #' @param muon_a
 #' Float, parameter a of the muon algorithm.

From aba497e07dcc97c6e97b0cc6829d4da2bd963aa0 Mon Sep 17 00:00:00 2001
From: Tomasz Kalinowski <kalinowskit@gmail.com>
Date: Fri, 26 Sep 2025 16:00:06 -0400
Subject: [PATCH 61/81] internal tether change

---
 .tether/man/keras.losses.txt | 1 +
 1 file changed, 1 insertion(+)

diff --git a/.tether/man/keras.losses.txt b/.tether/man/keras.losses.txt
index c1733bcb8..33dc78c49 100644
--- a/.tether/man/keras.losses.txt
+++ b/.tether/man/keras.losses.txt
@@ -208,6 +208,7 @@ SparseCategoricalCrossentropy(
   from_logits=False,
   ignore_class=None,
   reduction='sum_over_batch_size',
+  axis=-1,
   name='sparse_categorical_crossentropy',
   dtype=None
 )

From 98400042847c81a757202b756cc849e74b8e6333 Mon Sep 17 00:00:00 2001
From: Tomasz Kalinowski <kalinowskit@gmail.com>
Date: Fri, 26 Sep 2025 16:02:09 -0400
Subject: [PATCH 62/81] internal tether changes

---
 .tether/man/keras.ops.nn.txt | 7 ++++---
 1 file changed, 4 insertions(+), 3 deletions(-)

diff --git a/.tether/man/keras.ops.nn.txt b/.tether/man/keras.ops.nn.txt
index 04b363aa0..141ffedbc 100644
--- a/.tether/man/keras.ops.nn.txt
+++ b/.tether/man/keras.ops.nn.txt
@@ -37,7 +37,7 @@ conv(
 conv_transpose(
   inputs,
   kernel,
-  strides,
+  strides=1,
   padding='valid',
   output_padding=None,
   data_format=None,
@@ -75,7 +75,8 @@ dot_product_attention(
   mask=None,
   scale=None,
   is_causal=False,
-  flash_attention=None
+  flash_attention=None,
+  attn_logits_soft_cap=None
 )
 elu(x, alpha=1.0)
 gelu(x, approximate=True)
@@ -140,7 +141,7 @@ relu(x)
 relu6(x)
 rms_normalization(
   x,
-  scale=1,
+  scale=None,
   axis=-1,
   epsilon=None
 )

From 4bf0c91f32933b8e757d435a4c5d5ad736eaff07 Mon Sep 17 00:00:00 2001
From: Tomasz Kalinowski <kalinowskit@gmail.com>
Date: Fri, 26 Sep 2025 16:04:04 -0400
Subject: [PATCH 63/81] internal tether changes

---
 .tether/man/keras.ops.txt | 25 +++++++++++++++----------
 1 file changed, 15 insertions(+), 10 deletions(-)

diff --git a/.tether/man/keras.ops.txt b/.tether/man/keras.ops.txt
index 29045d079..2081aad40 100644
--- a/.tether/man/keras.ops.txt
+++ b/.tether/man/keras.ops.txt
@@ -140,7 +140,7 @@ conv(
 conv_transpose(
   inputs,
   kernel,
-  strides,
+  strides=1,
   padding='valid',
   output_padding=None,
   data_format=None,
@@ -233,12 +233,17 @@ dot_product_attention(
   mask=None,
   scale=None,
   is_causal=False,
-  flash_attention=None
+  flash_attention=None,
+  attn_logits_soft_cap=None
 )
 dtype(x)
 eig(x)
 eigh(x)
-einsum(subscripts, *operands)
+einsum(
+  subscripts,
+  *operands,
+  **kwargs
+)
 elu(x, alpha=1.0)
 empty(shape, dtype=None)
 equal(x1, x2)
@@ -302,7 +307,7 @@ hstack(xs)
 identity(n, dtype=None)
 ifft2(x)
 imag(x)
-image: Module(keras.api.ops.image)
+image: Module(keras.ops.image)
 in_top_k(
   targets,
   predictions,
@@ -344,7 +349,7 @@ leaky_relu(x, negative_slope=0.2)
 left_shift(x, y)
 less(x1, x2)
 less_equal(x1, x2)
-linalg: Module(keras.api.ops.linalg)
+linalg: Module(keras.ops.linalg)
 linspace(
   start,
   stop,
@@ -449,7 +454,7 @@ nan_to_num(
 )
 ndim(x)
 negative(x)
-nn: Module(keras.api.ops.nn)
+nn: Module(keras.ops.nn)
 nonzero(x)
 norm(
   x,
@@ -464,7 +469,7 @@ normalize(
   epsilon=None
 )
 not_equal(x1, x2)
-numpy: Module(keras.api.ops.numpy)
+numpy: Module(keras.ops.numpy)
 one_hot(
   x,
   num_classes,
@@ -522,7 +527,7 @@ rfft(x, fft_length=None)
 right_shift(x, y)
 rms_normalization(
   x,
-  scale=1,
+  scale=None,
   axis=-1,
   epsilon=None
 )
@@ -739,6 +744,8 @@ vectorize(
   signature=None
 )
 vectorized_map(function, elements)
+view_as_complex(x)
+view_as_real(x)
 vstack(xs)
 where(
   condition,
@@ -753,6 +760,4 @@ while_loop(
 )
 zeros(shape, dtype=None)
 zeros_like(x, dtype=None)
-view_as_complex(x)
-view_as_real(x)
 

From 9125798f617a07b028b685c84609771d23d3e616 Mon Sep 17 00:00:00 2001
From: Tomasz Kalinowski <kalinowskit@gmail.com>
Date: Fri, 26 Sep 2025 16:05:15 -0400
Subject: [PATCH 64/81] internal tether diff

---
 .tether/man/keras.saving.txt | 1 +
 1 file changed, 1 insertion(+)

diff --git a/.tether/man/keras.saving.txt b/.tether/man/keras.saving.txt
index 1c17c37ff..24be1e1c6 100644
--- a/.tether/man/keras.saving.txt
+++ b/.tether/man/keras.saving.txt
@@ -38,6 +38,7 @@ save_weights(
   model,
   filepath,
   overwrite=True,
+  max_shard_size=None,
   **kwargs
 )
 serialize_keras_object(obj)

From 7460fac18528ebcea414c6b79992c24bfada2301 Mon Sep 17 00:00:00 2001
From: Tomasz Kalinowski <kalinowskit@gmail.com>
Date: Fri, 26 Sep 2025 16:07:22 -0400
Subject: [PATCH 65/81] internal tether change

---
 .tether/man/keras.txt | 7 +++++--
 1 file changed, 5 insertions(+), 2 deletions(-)

diff --git a/.tether/man/keras.txt b/.tether/man/keras.txt
index d38f8bbbe..9fb6f9837 100644
--- a/.tether/man/keras.txt
+++ b/.tether/man/keras.txt
@@ -46,7 +46,8 @@ KerasTensor(
   sparse=False,
   ragged=False,
   record_history=True,
-  name=None
+  name=None,
+  **kwargs
 )
 Layer(*args, **kwargs)
 layers: Module(keras.layers)
@@ -97,7 +98,9 @@ Variable(
   trainable=True,
   autocast=True,
   aggregation='none',
-  name=None
+  synchronization='auto',
+  name=None,
+  **kwargs
 )
 version()
 visualization: Module(keras.visualization)

From 05497ca8da701da6c04acbb78e6e76100fb26b19 Mon Sep 17 00:00:00 2001
From: Tomasz Kalinowski <kalinowskit@gmail.com>
Date: Fri, 26 Sep 2025 16:09:41 -0400
Subject: [PATCH 66/81] redocument

---
 man/Layer.Rd                            |   1 +
 man/layer_activation.Rd                 |   1 +
 man/layer_activation_elu.Rd             |   1 +
 man/layer_activation_leaky_relu.Rd      |   1 +
 man/layer_activation_parametric_relu.Rd |   1 +
 man/layer_activation_relu.Rd            |   1 +
 man/layer_activation_softmax.Rd         |   1 +
 man/layer_activity_regularization.Rd    |   1 +
 man/layer_add.Rd                        |   1 +
 man/layer_additive_attention.Rd         |   1 +
 man/layer_alpha_dropout.Rd              |   1 +
 man/layer_attention.Rd                  |   1 +
 man/layer_aug_mix.Rd                    |   3 +
 man/layer_auto_contrast.Rd              |   3 +
 man/layer_average.Rd                    |   1 +
 man/layer_average_pooling_1d.Rd         |   1 +
 man/layer_average_pooling_2d.Rd         |   1 +
 man/layer_average_pooling_3d.Rd         |   1 +
 man/layer_batch_normalization.Rd        |   1 +
 man/layer_bidirectional.Rd              |   1 +
 man/layer_category_encoding.Rd          |   2 +
 man/layer_center_crop.Rd                |   3 +
 man/layer_concatenate.Rd                |   1 +
 man/layer_conv_1d.Rd                    |   1 +
 man/layer_conv_1d_transpose.Rd          |   1 +
 man/layer_conv_2d.Rd                    |   1 +
 man/layer_conv_2d_transpose.Rd          |   1 +
 man/layer_conv_3d.Rd                    |   1 +
 man/layer_conv_3d_transpose.Rd          |   1 +
 man/layer_conv_lstm_1d.Rd               |   1 +
 man/layer_conv_lstm_2d.Rd               |   1 +
 man/layer_conv_lstm_3d.Rd               |   1 +
 man/layer_cropping_1d.Rd                |   1 +
 man/layer_cropping_2d.Rd                |   1 +
 man/layer_cropping_3d.Rd                |   1 +
 man/layer_cut_mix.Rd                    |   3 +
 man/layer_dense.Rd                      |   1 +
 man/layer_depthwise_conv_1d.Rd          |   1 +
 man/layer_depthwise_conv_2d.Rd          |   1 +
 man/layer_discretization.Rd             |   2 +
 man/layer_dot.Rd                        |   1 +
 man/layer_dropout.Rd                    |   1 +
 man/layer_einsum_dense.Rd               |   1 +
 man/layer_embedding.Rd                  |   1 +
 man/layer_equalization.Rd               |   3 +
 man/layer_feature_space.Rd              |   2 +
 man/layer_flatten.Rd                    |   1 +
 man/layer_flax_module_wrapper.Rd        |   1 +
 man/layer_gaussian_dropout.Rd           |   1 +
 man/layer_gaussian_noise.Rd             |   1 +
 man/layer_global_average_pooling_1d.Rd  |   1 +
 man/layer_global_average_pooling_2d.Rd  |   1 +
 man/layer_global_average_pooling_3d.Rd  |   1 +
 man/layer_global_max_pooling_1d.Rd      |   1 +
 man/layer_global_max_pooling_2d.Rd      |   1 +
 man/layer_global_max_pooling_3d.Rd      |   1 +
 man/layer_group_normalization.Rd        |   1 +
 man/layer_group_query_attention.Rd      |   1 +
 man/layer_gru.Rd                        |   1 +
 man/layer_hashed_crossing.Rd            |   2 +
 man/layer_hashing.Rd                    |   2 +
 man/layer_identity.Rd                   |   1 +
 man/layer_integer_lookup.Rd             |   2 +
 man/layer_jax_model_wrapper.Rd          |   1 +
 man/layer_lambda.Rd                     |   1 +
 man/layer_layer_normalization.Rd        |   1 +
 man/layer_lstm.Rd                       |   1 +
 man/layer_masking.Rd                    |   1 +
 man/layer_max_num_bounding_boxes.Rd     |   3 +
 man/layer_max_pooling_1d.Rd             |   1 +
 man/layer_max_pooling_2d.Rd             |   1 +
 man/layer_max_pooling_3d.Rd             |   1 +
 man/layer_maximum.Rd                    |   1 +
 man/layer_mel_spectrogram.Rd            |   2 +
 man/layer_minimum.Rd                    |   1 +
 man/layer_mix_up.Rd                     |   3 +
 man/layer_multi_head_attention.Rd       |   1 +
 man/layer_multiply.Rd                   |   1 +
 man/layer_normalization.Rd              |   2 +
 man/layer_permute.Rd                    |   1 +
 man/layer_rand_augment.Rd               |   3 +
 man/layer_random_brightness.Rd          |   2 +
 man/layer_random_color_degeneration.Rd  |   3 +
 man/layer_random_color_jitter.Rd        |   3 +
 man/layer_random_contrast.Rd            |   2 +
 man/layer_random_crop.Rd                |   2 +
 man/layer_random_elastic_transform.Rd   | 300 ++++++++++++++++++++++++
 man/layer_random_erasing.Rd             |   3 +
 man/layer_random_flip.Rd                |   2 +
 man/layer_random_gaussian_blur.Rd       |   3 +
 man/layer_random_grayscale.Rd           |   3 +
 man/layer_random_hue.Rd                 |   3 +
 man/layer_random_invert.Rd              |   3 +
 man/layer_random_perspective.Rd         |   3 +
 man/layer_random_posterization.Rd       |   3 +
 man/layer_random_rotation.Rd            |   2 +
 man/layer_random_saturation.Rd          |   3 +
 man/layer_random_sharpness.Rd           |   3 +
 man/layer_random_shear.Rd               |   3 +
 man/layer_random_translation.Rd         |   2 +
 man/layer_random_zoom.Rd                |   2 +
 man/layer_repeat_vector.Rd              |   1 +
 man/layer_rescaling.Rd                  |   3 +
 man/layer_reshape.Rd                    |   1 +
 man/layer_resizing.Rd                   |   3 +
 man/layer_rms_normalization.Rd          |   1 +
 man/layer_rnn.Rd                        |   1 +
 man/layer_separable_conv_1d.Rd          |   1 +
 man/layer_separable_conv_2d.Rd          |   1 +
 man/layer_simple_rnn.Rd                 |   1 +
 man/layer_solarization.Rd               |   3 +
 man/layer_spatial_dropout_1d.Rd         |   1 +
 man/layer_spatial_dropout_2d.Rd         |   1 +
 man/layer_spatial_dropout_3d.Rd         |   1 +
 man/layer_spectral_normalization.Rd     |   1 +
 man/layer_stft_spectrogram.Rd           |   2 +
 man/layer_string_lookup.Rd              |   2 +
 man/layer_subtract.Rd                   |   1 +
 man/layer_text_vectorization.Rd         |   2 +
 man/layer_tfsm.Rd                       |   5 +-
 man/layer_time_distributed.Rd           |   1 +
 man/layer_torch_module_wrapper.Rd       |   1 +
 man/layer_unit_normalization.Rd         |   1 +
 man/layer_upsampling_1d.Rd              |   1 +
 man/layer_upsampling_2d.Rd              |   1 +
 man/layer_upsampling_3d.Rd              |   1 +
 man/layer_zero_padding_1d.Rd            |   1 +
 man/layer_zero_padding_2d.Rd            |   1 +
 man/layer_zero_padding_3d.Rd            |   1 +
 man/op_image_elastic_transform.Rd       |  97 ++++++++
 man/op_searchsorted.Rd                  |  26 +-
 man/optimizer_muon.Rd                   | 118 ++++++++++
 man/rnn_cell_gru.Rd                     |   1 +
 man/rnn_cell_lstm.Rd                    |   1 +
 man/rnn_cell_simple.Rd                  |   1 +
 man/rnn_cells_stack.Rd                  |   1 +
 136 files changed, 729 insertions(+), 12 deletions(-)
 create mode 100644 man/layer_random_elastic_transform.Rd
 create mode 100644 man/op_image_elastic_transform.Rd
 create mode 100644 man/optimizer_muon.Rd

diff --git a/man/Layer.Rd b/man/Layer.Rd
index 000fba8df..8ac448afb 100644
--- a/man/Layer.Rd
+++ b/man/Layer.Rd
@@ -783,6 +783,7 @@ Other layers: \cr
 \code{\link{layer_random_color_jitter}()} \cr
 \code{\link{layer_random_contrast}()} \cr
 \code{\link{layer_random_crop}()} \cr
+\code{\link{layer_random_elastic_transform}()} \cr
 \code{\link{layer_random_erasing}()} \cr
 \code{\link{layer_random_flip}()} \cr
 \code{\link{layer_random_gaussian_blur}()} \cr
diff --git a/man/layer_activation.Rd b/man/layer_activation.Rd
index dad3684fc..3adfcbf1b 100644
--- a/man/layer_activation.Rd
+++ b/man/layer_activation.Rd
@@ -153,6 +153,7 @@ Other layers: \cr
 \code{\link{layer_random_color_jitter}()} \cr
 \code{\link{layer_random_contrast}()} \cr
 \code{\link{layer_random_crop}()} \cr
+\code{\link{layer_random_elastic_transform}()} \cr
 \code{\link{layer_random_erasing}()} \cr
 \code{\link{layer_random_flip}()} \cr
 \code{\link{layer_random_gaussian_blur}()} \cr
diff --git a/man/layer_activation_elu.Rd b/man/layer_activation_elu.Rd
index 6f2752415..f13e72cdd 100644
--- a/man/layer_activation_elu.Rd
+++ b/man/layer_activation_elu.Rd
@@ -129,6 +129,7 @@ Other layers: \cr
 \code{\link{layer_random_color_jitter}()} \cr
 \code{\link{layer_random_contrast}()} \cr
 \code{\link{layer_random_crop}()} \cr
+\code{\link{layer_random_elastic_transform}()} \cr
 \code{\link{layer_random_erasing}()} \cr
 \code{\link{layer_random_flip}()} \cr
 \code{\link{layer_random_gaussian_blur}()} \cr
diff --git a/man/layer_activation_leaky_relu.Rd b/man/layer_activation_leaky_relu.Rd
index 77e3deba6..ab0252968 100644
--- a/man/layer_activation_leaky_relu.Rd
+++ b/man/layer_activation_leaky_relu.Rd
@@ -144,6 +144,7 @@ Other layers: \cr
 \code{\link{layer_random_color_jitter}()} \cr
 \code{\link{layer_random_contrast}()} \cr
 \code{\link{layer_random_crop}()} \cr
+\code{\link{layer_random_elastic_transform}()} \cr
 \code{\link{layer_random_erasing}()} \cr
 \code{\link{layer_random_flip}()} \cr
 \code{\link{layer_random_gaussian_blur}()} \cr
diff --git a/man/layer_activation_parametric_relu.Rd b/man/layer_activation_parametric_relu.Rd
index 768646de4..f7a56c5be 100644
--- a/man/layer_activation_parametric_relu.Rd
+++ b/man/layer_activation_parametric_relu.Rd
@@ -148,6 +148,7 @@ Other layers: \cr
 \code{\link{layer_random_color_jitter}()} \cr
 \code{\link{layer_random_contrast}()} \cr
 \code{\link{layer_random_crop}()} \cr
+\code{\link{layer_random_elastic_transform}()} \cr
 \code{\link{layer_random_erasing}()} \cr
 \code{\link{layer_random_flip}()} \cr
 \code{\link{layer_random_gaussian_blur}()} \cr
diff --git a/man/layer_activation_relu.Rd b/man/layer_activation_relu.Rd
index 6d544375e..708d854ea 100644
--- a/man/layer_activation_relu.Rd
+++ b/man/layer_activation_relu.Rd
@@ -163,6 +163,7 @@ Other layers: \cr
 \code{\link{layer_random_color_jitter}()} \cr
 \code{\link{layer_random_contrast}()} \cr
 \code{\link{layer_random_crop}()} \cr
+\code{\link{layer_random_elastic_transform}()} \cr
 \code{\link{layer_random_erasing}()} \cr
 \code{\link{layer_random_flip}()} \cr
 \code{\link{layer_random_gaussian_blur}()} \cr
diff --git a/man/layer_activation_softmax.Rd b/man/layer_activation_softmax.Rd
index 15eb93546..ba8d0aece 100644
--- a/man/layer_activation_softmax.Rd
+++ b/man/layer_activation_softmax.Rd
@@ -141,6 +141,7 @@ Other layers: \cr
 \code{\link{layer_random_color_jitter}()} \cr
 \code{\link{layer_random_contrast}()} \cr
 \code{\link{layer_random_crop}()} \cr
+\code{\link{layer_random_elastic_transform}()} \cr
 \code{\link{layer_random_erasing}()} \cr
 \code{\link{layer_random_flip}()} \cr
 \code{\link{layer_random_gaussian_blur}()} \cr
diff --git a/man/layer_activity_regularization.Rd b/man/layer_activity_regularization.Rd
index 8d7bfe0a4..c5de9c888 100644
--- a/man/layer_activity_regularization.Rd
+++ b/man/layer_activity_regularization.Rd
@@ -139,6 +139,7 @@ Other layers: \cr
 \code{\link{layer_random_color_jitter}()} \cr
 \code{\link{layer_random_contrast}()} \cr
 \code{\link{layer_random_crop}()} \cr
+\code{\link{layer_random_elastic_transform}()} \cr
 \code{\link{layer_random_erasing}()} \cr
 \code{\link{layer_random_flip}()} \cr
 \code{\link{layer_random_gaussian_blur}()} \cr
diff --git a/man/layer_add.Rd b/man/layer_add.Rd
index cf90982f4..7f18763b1 100644
--- a/man/layer_add.Rd
+++ b/man/layer_add.Rd
@@ -155,6 +155,7 @@ Other layers: \cr
 \code{\link{layer_random_color_jitter}()} \cr
 \code{\link{layer_random_contrast}()} \cr
 \code{\link{layer_random_crop}()} \cr
+\code{\link{layer_random_elastic_transform}()} \cr
 \code{\link{layer_random_erasing}()} \cr
 \code{\link{layer_random_flip}()} \cr
 \code{\link{layer_random_gaussian_blur}()} \cr
diff --git a/man/layer_additive_attention.Rd b/man/layer_additive_attention.Rd
index 2b874246f..383b2aa68 100644
--- a/man/layer_additive_attention.Rd
+++ b/man/layer_additive_attention.Rd
@@ -180,6 +180,7 @@ Other layers: \cr
 \code{\link{layer_random_color_jitter}()} \cr
 \code{\link{layer_random_contrast}()} \cr
 \code{\link{layer_random_crop}()} \cr
+\code{\link{layer_random_elastic_transform}()} \cr
 \code{\link{layer_random_erasing}()} \cr
 \code{\link{layer_random_flip}()} \cr
 \code{\link{layer_random_gaussian_blur}()} \cr
diff --git a/man/layer_alpha_dropout.Rd b/man/layer_alpha_dropout.Rd
index 396e923bb..48e415c6d 100644
--- a/man/layer_alpha_dropout.Rd
+++ b/man/layer_alpha_dropout.Rd
@@ -150,6 +150,7 @@ Other layers: \cr
 \code{\link{layer_random_color_jitter}()} \cr
 \code{\link{layer_random_contrast}()} \cr
 \code{\link{layer_random_crop}()} \cr
+\code{\link{layer_random_elastic_transform}()} \cr
 \code{\link{layer_random_erasing}()} \cr
 \code{\link{layer_random_flip}()} \cr
 \code{\link{layer_random_gaussian_blur}()} \cr
diff --git a/man/layer_attention.Rd b/man/layer_attention.Rd
index 2e5e800da..1e0b56b04 100644
--- a/man/layer_attention.Rd
+++ b/man/layer_attention.Rd
@@ -193,6 +193,7 @@ Other layers: \cr
 \code{\link{layer_random_color_jitter}()} \cr
 \code{\link{layer_random_contrast}()} \cr
 \code{\link{layer_random_crop}()} \cr
+\code{\link{layer_random_elastic_transform}()} \cr
 \code{\link{layer_random_erasing}()} \cr
 \code{\link{layer_random_flip}()} \cr
 \code{\link{layer_random_gaussian_blur}()} \cr
diff --git a/man/layer_aug_mix.Rd b/man/layer_aug_mix.Rd
index 3ace5c516..45295fad9 100644
--- a/man/layer_aug_mix.Rd
+++ b/man/layer_aug_mix.Rd
@@ -85,6 +85,7 @@ Other image preprocessing layers: \cr
 \code{\link{layer_rand_augment}()} \cr
 \code{\link{layer_random_color_degeneration}()} \cr
 \code{\link{layer_random_color_jitter}()} \cr
+\code{\link{layer_random_elastic_transform}()} \cr
 \code{\link{layer_random_erasing}()} \cr
 \code{\link{layer_random_gaussian_blur}()} \cr
 \code{\link{layer_random_grayscale}()} \cr
@@ -120,6 +121,7 @@ Other preprocessing layers: \cr
 \code{\link{layer_random_color_jitter}()} \cr
 \code{\link{layer_random_contrast}()} \cr
 \code{\link{layer_random_crop}()} \cr
+\code{\link{layer_random_elastic_transform}()} \cr
 \code{\link{layer_random_erasing}()} \cr
 \code{\link{layer_random_flip}()} \cr
 \code{\link{layer_random_gaussian_blur}()} \cr
@@ -227,6 +229,7 @@ Other layers: \cr
 \code{\link{layer_random_color_jitter}()} \cr
 \code{\link{layer_random_contrast}()} \cr
 \code{\link{layer_random_crop}()} \cr
+\code{\link{layer_random_elastic_transform}()} \cr
 \code{\link{layer_random_erasing}()} \cr
 \code{\link{layer_random_flip}()} \cr
 \code{\link{layer_random_gaussian_blur}()} \cr
diff --git a/man/layer_auto_contrast.Rd b/man/layer_auto_contrast.Rd
index 1e66b4faa..b390c33c5 100644
--- a/man/layer_auto_contrast.Rd
+++ b/man/layer_auto_contrast.Rd
@@ -36,6 +36,7 @@ Other image preprocessing layers: \cr
 \code{\link{layer_rand_augment}()} \cr
 \code{\link{layer_random_color_degeneration}()} \cr
 \code{\link{layer_random_color_jitter}()} \cr
+\code{\link{layer_random_elastic_transform}()} \cr
 \code{\link{layer_random_erasing}()} \cr
 \code{\link{layer_random_gaussian_blur}()} \cr
 \code{\link{layer_random_grayscale}()} \cr
@@ -71,6 +72,7 @@ Other preprocessing layers: \cr
 \code{\link{layer_random_color_jitter}()} \cr
 \code{\link{layer_random_contrast}()} \cr
 \code{\link{layer_random_crop}()} \cr
+\code{\link{layer_random_elastic_transform}()} \cr
 \code{\link{layer_random_erasing}()} \cr
 \code{\link{layer_random_flip}()} \cr
 \code{\link{layer_random_gaussian_blur}()} \cr
@@ -178,6 +180,7 @@ Other layers: \cr
 \code{\link{layer_random_color_jitter}()} \cr
 \code{\link{layer_random_contrast}()} \cr
 \code{\link{layer_random_crop}()} \cr
+\code{\link{layer_random_elastic_transform}()} \cr
 \code{\link{layer_random_erasing}()} \cr
 \code{\link{layer_random_flip}()} \cr
 \code{\link{layer_random_gaussian_blur}()} \cr
diff --git a/man/layer_average.Rd b/man/layer_average.Rd
index 244b4bb8b..d3d35898c 100644
--- a/man/layer_average.Rd
+++ b/man/layer_average.Rd
@@ -154,6 +154,7 @@ Other layers: \cr
 \code{\link{layer_random_color_jitter}()} \cr
 \code{\link{layer_random_contrast}()} \cr
 \code{\link{layer_random_crop}()} \cr
+\code{\link{layer_random_elastic_transform}()} \cr
 \code{\link{layer_random_erasing}()} \cr
 \code{\link{layer_random_flip}()} \cr
 \code{\link{layer_random_gaussian_blur}()} \cr
diff --git a/man/layer_average_pooling_1d.Rd b/man/layer_average_pooling_1d.Rd
index ac21cf6c1..5c89f40b5 100644
--- a/man/layer_average_pooling_1d.Rd
+++ b/man/layer_average_pooling_1d.Rd
@@ -236,6 +236,7 @@ Other layers: \cr
 \code{\link{layer_random_color_jitter}()} \cr
 \code{\link{layer_random_contrast}()} \cr
 \code{\link{layer_random_crop}()} \cr
+\code{\link{layer_random_elastic_transform}()} \cr
 \code{\link{layer_random_erasing}()} \cr
 \code{\link{layer_random_flip}()} \cr
 \code{\link{layer_random_gaussian_blur}()} \cr
diff --git a/man/layer_average_pooling_2d.Rd b/man/layer_average_pooling_2d.Rd
index 166361f04..7fb91f3a3 100644
--- a/man/layer_average_pooling_2d.Rd
+++ b/man/layer_average_pooling_2d.Rd
@@ -254,6 +254,7 @@ Other layers: \cr
 \code{\link{layer_random_color_jitter}()} \cr
 \code{\link{layer_random_contrast}()} \cr
 \code{\link{layer_random_crop}()} \cr
+\code{\link{layer_random_elastic_transform}()} \cr
 \code{\link{layer_random_erasing}()} \cr
 \code{\link{layer_random_flip}()} \cr
 \code{\link{layer_random_gaussian_blur}()} \cr
diff --git a/man/layer_average_pooling_3d.Rd b/man/layer_average_pooling_3d.Rd
index 09d1d8abb..bd2fb0639 100644
--- a/man/layer_average_pooling_3d.Rd
+++ b/man/layer_average_pooling_3d.Rd
@@ -201,6 +201,7 @@ Other layers: \cr
 \code{\link{layer_random_color_jitter}()} \cr
 \code{\link{layer_random_contrast}()} \cr
 \code{\link{layer_random_crop}()} \cr
+\code{\link{layer_random_elastic_transform}()} \cr
 \code{\link{layer_random_erasing}()} \cr
 \code{\link{layer_random_flip}()} \cr
 \code{\link{layer_random_gaussian_blur}()} \cr
diff --git a/man/layer_batch_normalization.Rd b/man/layer_batch_normalization.Rd
index b8f2b2eda..050d19cc4 100644
--- a/man/layer_batch_normalization.Rd
+++ b/man/layer_batch_normalization.Rd
@@ -267,6 +267,7 @@ Other layers: \cr
 \code{\link{layer_random_color_jitter}()} \cr
 \code{\link{layer_random_contrast}()} \cr
 \code{\link{layer_random_crop}()} \cr
+\code{\link{layer_random_elastic_transform}()} \cr
 \code{\link{layer_random_erasing}()} \cr
 \code{\link{layer_random_flip}()} \cr
 \code{\link{layer_random_gaussian_blur}()} \cr
diff --git a/man/layer_bidirectional.Rd b/man/layer_bidirectional.Rd
index 07b54f17d..6098ee187 100644
--- a/man/layer_bidirectional.Rd
+++ b/man/layer_bidirectional.Rd
@@ -218,6 +218,7 @@ Other layers: \cr
 \code{\link{layer_random_color_jitter}()} \cr
 \code{\link{layer_random_contrast}()} \cr
 \code{\link{layer_random_crop}()} \cr
+\code{\link{layer_random_elastic_transform}()} \cr
 \code{\link{layer_random_erasing}()} \cr
 \code{\link{layer_random_flip}()} \cr
 \code{\link{layer_random_gaussian_blur}()} \cr
diff --git a/man/layer_category_encoding.Rd b/man/layer_category_encoding.Rd
index c40263a91..ac0abb2c0 100644
--- a/man/layer_category_encoding.Rd
+++ b/man/layer_category_encoding.Rd
@@ -159,6 +159,7 @@ Other preprocessing layers: \cr
 \code{\link{layer_random_color_jitter}()} \cr
 \code{\link{layer_random_contrast}()} \cr
 \code{\link{layer_random_crop}()} \cr
+\code{\link{layer_random_elastic_transform}()} \cr
 \code{\link{layer_random_erasing}()} \cr
 \code{\link{layer_random_flip}()} \cr
 \code{\link{layer_random_gaussian_blur}()} \cr
@@ -266,6 +267,7 @@ Other layers: \cr
 \code{\link{layer_random_color_jitter}()} \cr
 \code{\link{layer_random_contrast}()} \cr
 \code{\link{layer_random_crop}()} \cr
+\code{\link{layer_random_elastic_transform}()} \cr
 \code{\link{layer_random_erasing}()} \cr
 \code{\link{layer_random_flip}()} \cr
 \code{\link{layer_random_gaussian_blur}()} \cr
diff --git a/man/layer_center_crop.Rd b/man/layer_center_crop.Rd
index f870bfe47..36210edc6 100644
--- a/man/layer_center_crop.Rd
+++ b/man/layer_center_crop.Rd
@@ -77,6 +77,7 @@ Other image preprocessing layers: \cr
 \code{\link{layer_rand_augment}()} \cr
 \code{\link{layer_random_color_degeneration}()} \cr
 \code{\link{layer_random_color_jitter}()} \cr
+\code{\link{layer_random_elastic_transform}()} \cr
 \code{\link{layer_random_erasing}()} \cr
 \code{\link{layer_random_gaussian_blur}()} \cr
 \code{\link{layer_random_grayscale}()} \cr
@@ -112,6 +113,7 @@ Other preprocessing layers: \cr
 \code{\link{layer_random_color_jitter}()} \cr
 \code{\link{layer_random_contrast}()} \cr
 \code{\link{layer_random_crop}()} \cr
+\code{\link{layer_random_elastic_transform}()} \cr
 \code{\link{layer_random_erasing}()} \cr
 \code{\link{layer_random_flip}()} \cr
 \code{\link{layer_random_gaussian_blur}()} \cr
@@ -219,6 +221,7 @@ Other layers: \cr
 \code{\link{layer_random_color_jitter}()} \cr
 \code{\link{layer_random_contrast}()} \cr
 \code{\link{layer_random_crop}()} \cr
+\code{\link{layer_random_elastic_transform}()} \cr
 \code{\link{layer_random_erasing}()} \cr
 \code{\link{layer_random_flip}()} \cr
 \code{\link{layer_random_gaussian_blur}()} \cr
diff --git a/man/layer_concatenate.Rd b/man/layer_concatenate.Rd
index cc7094b69..7e5fc889d 100644
--- a/man/layer_concatenate.Rd
+++ b/man/layer_concatenate.Rd
@@ -148,6 +148,7 @@ Other layers: \cr
 \code{\link{layer_random_color_jitter}()} \cr
 \code{\link{layer_random_contrast}()} \cr
 \code{\link{layer_random_crop}()} \cr
+\code{\link{layer_random_elastic_transform}()} \cr
 \code{\link{layer_random_erasing}()} \cr
 \code{\link{layer_random_flip}()} \cr
 \code{\link{layer_random_gaussian_blur}()} \cr
diff --git a/man/layer_conv_1d.Rd b/man/layer_conv_1d.Rd
index 8c9ae2e07..9680b7e45 100644
--- a/man/layer_conv_1d.Rd
+++ b/man/layer_conv_1d.Rd
@@ -239,6 +239,7 @@ Other layers: \cr
 \code{\link{layer_random_color_jitter}()} \cr
 \code{\link{layer_random_contrast}()} \cr
 \code{\link{layer_random_crop}()} \cr
+\code{\link{layer_random_elastic_transform}()} \cr
 \code{\link{layer_random_erasing}()} \cr
 \code{\link{layer_random_flip}()} \cr
 \code{\link{layer_random_gaussian_blur}()} \cr
diff --git a/man/layer_conv_1d_transpose.Rd b/man/layer_conv_1d_transpose.Rd
index 08db7168c..9e46e3002 100644
--- a/man/layer_conv_1d_transpose.Rd
+++ b/man/layer_conv_1d_transpose.Rd
@@ -239,6 +239,7 @@ Other layers: \cr
 \code{\link{layer_random_color_jitter}()} \cr
 \code{\link{layer_random_contrast}()} \cr
 \code{\link{layer_random_crop}()} \cr
+\code{\link{layer_random_elastic_transform}()} \cr
 \code{\link{layer_random_erasing}()} \cr
 \code{\link{layer_random_flip}()} \cr
 \code{\link{layer_random_gaussian_blur}()} \cr
diff --git a/man/layer_conv_2d.Rd b/man/layer_conv_2d.Rd
index 9ac6c189f..8e4e39d1c 100644
--- a/man/layer_conv_2d.Rd
+++ b/man/layer_conv_2d.Rd
@@ -248,6 +248,7 @@ Other layers: \cr
 \code{\link{layer_random_color_jitter}()} \cr
 \code{\link{layer_random_contrast}()} \cr
 \code{\link{layer_random_crop}()} \cr
+\code{\link{layer_random_elastic_transform}()} \cr
 \code{\link{layer_random_erasing}()} \cr
 \code{\link{layer_random_flip}()} \cr
 \code{\link{layer_random_gaussian_blur}()} \cr
diff --git a/man/layer_conv_2d_transpose.Rd b/man/layer_conv_2d_transpose.Rd
index ef3584120..f0663933b 100644
--- a/man/layer_conv_2d_transpose.Rd
+++ b/man/layer_conv_2d_transpose.Rd
@@ -241,6 +241,7 @@ Other layers: \cr
 \code{\link{layer_random_color_jitter}()} \cr
 \code{\link{layer_random_contrast}()} \cr
 \code{\link{layer_random_crop}()} \cr
+\code{\link{layer_random_elastic_transform}()} \cr
 \code{\link{layer_random_erasing}()} \cr
 \code{\link{layer_random_flip}()} \cr
 \code{\link{layer_random_gaussian_blur}()} \cr
diff --git a/man/layer_conv_3d.Rd b/man/layer_conv_3d.Rd
index d657cbcbf..686d4a3e2 100644
--- a/man/layer_conv_3d.Rd
+++ b/man/layer_conv_3d.Rd
@@ -239,6 +239,7 @@ Other layers: \cr
 \code{\link{layer_random_color_jitter}()} \cr
 \code{\link{layer_random_contrast}()} \cr
 \code{\link{layer_random_crop}()} \cr
+\code{\link{layer_random_elastic_transform}()} \cr
 \code{\link{layer_random_erasing}()} \cr
 \code{\link{layer_random_flip}()} \cr
 \code{\link{layer_random_gaussian_blur}()} \cr
diff --git a/man/layer_conv_3d_transpose.Rd b/man/layer_conv_3d_transpose.Rd
index af696e205..79e3cf54f 100644
--- a/man/layer_conv_3d_transpose.Rd
+++ b/man/layer_conv_3d_transpose.Rd
@@ -241,6 +241,7 @@ Other layers: \cr
 \code{\link{layer_random_color_jitter}()} \cr
 \code{\link{layer_random_contrast}()} \cr
 \code{\link{layer_random_crop}()} \cr
+\code{\link{layer_random_elastic_transform}()} \cr
 \code{\link{layer_random_erasing}()} \cr
 \code{\link{layer_random_flip}()} \cr
 \code{\link{layer_random_gaussian_blur}()} \cr
diff --git a/man/layer_conv_lstm_1d.Rd b/man/layer_conv_lstm_1d.Rd
index 44a41d05d..4c90267fa 100644
--- a/man/layer_conv_lstm_1d.Rd
+++ b/man/layer_conv_lstm_1d.Rd
@@ -304,6 +304,7 @@ Other layers: \cr
 \code{\link{layer_random_color_jitter}()} \cr
 \code{\link{layer_random_contrast}()} \cr
 \code{\link{layer_random_crop}()} \cr
+\code{\link{layer_random_elastic_transform}()} \cr
 \code{\link{layer_random_erasing}()} \cr
 \code{\link{layer_random_flip}()} \cr
 \code{\link{layer_random_gaussian_blur}()} \cr
diff --git a/man/layer_conv_lstm_2d.Rd b/man/layer_conv_lstm_2d.Rd
index 7ba9a14e7..8af7a4e49 100644
--- a/man/layer_conv_lstm_2d.Rd
+++ b/man/layer_conv_lstm_2d.Rd
@@ -304,6 +304,7 @@ Other layers: \cr
 \code{\link{layer_random_color_jitter}()} \cr
 \code{\link{layer_random_contrast}()} \cr
 \code{\link{layer_random_crop}()} \cr
+\code{\link{layer_random_elastic_transform}()} \cr
 \code{\link{layer_random_erasing}()} \cr
 \code{\link{layer_random_flip}()} \cr
 \code{\link{layer_random_gaussian_blur}()} \cr
diff --git a/man/layer_conv_lstm_3d.Rd b/man/layer_conv_lstm_3d.Rd
index 088458813..971b5cf15 100644
--- a/man/layer_conv_lstm_3d.Rd
+++ b/man/layer_conv_lstm_3d.Rd
@@ -303,6 +303,7 @@ Other layers: \cr
 \code{\link{layer_random_color_jitter}()} \cr
 \code{\link{layer_random_contrast}()} \cr
 \code{\link{layer_random_crop}()} \cr
+\code{\link{layer_random_elastic_transform}()} \cr
 \code{\link{layer_random_erasing}()} \cr
 \code{\link{layer_random_flip}()} \cr
 \code{\link{layer_random_gaussian_blur}()} \cr
diff --git a/man/layer_cropping_1d.Rd b/man/layer_cropping_1d.Rd
index be427660a..44b61b144 100644
--- a/man/layer_cropping_1d.Rd
+++ b/man/layer_cropping_1d.Rd
@@ -176,6 +176,7 @@ Other layers: \cr
 \code{\link{layer_random_color_jitter}()} \cr
 \code{\link{layer_random_contrast}()} \cr
 \code{\link{layer_random_crop}()} \cr
+\code{\link{layer_random_elastic_transform}()} \cr
 \code{\link{layer_random_erasing}()} \cr
 \code{\link{layer_random_flip}()} \cr
 \code{\link{layer_random_gaussian_blur}()} \cr
diff --git a/man/layer_cropping_2d.Rd b/man/layer_cropping_2d.Rd
index e83c88199..684e1a76a 100644
--- a/man/layer_cropping_2d.Rd
+++ b/man/layer_cropping_2d.Rd
@@ -188,6 +188,7 @@ Other layers: \cr
 \code{\link{layer_random_color_jitter}()} \cr
 \code{\link{layer_random_contrast}()} \cr
 \code{\link{layer_random_crop}()} \cr
+\code{\link{layer_random_elastic_transform}()} \cr
 \code{\link{layer_random_erasing}()} \cr
 \code{\link{layer_random_flip}()} \cr
 \code{\link{layer_random_gaussian_blur}()} \cr
diff --git a/man/layer_cropping_3d.Rd b/man/layer_cropping_3d.Rd
index 285f77f28..1c143675e 100644
--- a/man/layer_cropping_3d.Rd
+++ b/man/layer_cropping_3d.Rd
@@ -188,6 +188,7 @@ Other layers: \cr
 \code{\link{layer_random_color_jitter}()} \cr
 \code{\link{layer_random_contrast}()} \cr
 \code{\link{layer_random_crop}()} \cr
+\code{\link{layer_random_elastic_transform}()} \cr
 \code{\link{layer_random_erasing}()} \cr
 \code{\link{layer_random_flip}()} \cr
 \code{\link{layer_random_gaussian_blur}()} \cr
diff --git a/man/layer_cut_mix.Rd b/man/layer_cut_mix.Rd
index be3f5a6b8..2fe03a169 100644
--- a/man/layer_cut_mix.Rd
+++ b/man/layer_cut_mix.Rd
@@ -53,6 +53,7 @@ Other image preprocessing layers: \cr
 \code{\link{layer_rand_augment}()} \cr
 \code{\link{layer_random_color_degeneration}()} \cr
 \code{\link{layer_random_color_jitter}()} \cr
+\code{\link{layer_random_elastic_transform}()} \cr
 \code{\link{layer_random_erasing}()} \cr
 \code{\link{layer_random_gaussian_blur}()} \cr
 \code{\link{layer_random_grayscale}()} \cr
@@ -88,6 +89,7 @@ Other preprocessing layers: \cr
 \code{\link{layer_random_color_jitter}()} \cr
 \code{\link{layer_random_contrast}()} \cr
 \code{\link{layer_random_crop}()} \cr
+\code{\link{layer_random_elastic_transform}()} \cr
 \code{\link{layer_random_erasing}()} \cr
 \code{\link{layer_random_flip}()} \cr
 \code{\link{layer_random_gaussian_blur}()} \cr
@@ -195,6 +197,7 @@ Other layers: \cr
 \code{\link{layer_random_color_jitter}()} \cr
 \code{\link{layer_random_contrast}()} \cr
 \code{\link{layer_random_crop}()} \cr
+\code{\link{layer_random_elastic_transform}()} \cr
 \code{\link{layer_random_erasing}()} \cr
 \code{\link{layer_random_flip}()} \cr
 \code{\link{layer_random_gaussian_blur}()} \cr
diff --git a/man/layer_dense.Rd b/man/layer_dense.Rd
index dc72eadb1..ed7ebc480 100644
--- a/man/layer_dense.Rd
+++ b/man/layer_dense.Rd
@@ -229,6 +229,7 @@ Other layers: \cr
 \code{\link{layer_random_color_jitter}()} \cr
 \code{\link{layer_random_contrast}()} \cr
 \code{\link{layer_random_crop}()} \cr
+\code{\link{layer_random_elastic_transform}()} \cr
 \code{\link{layer_random_erasing}()} \cr
 \code{\link{layer_random_flip}()} \cr
 \code{\link{layer_random_gaussian_blur}()} \cr
diff --git a/man/layer_depthwise_conv_1d.Rd b/man/layer_depthwise_conv_1d.Rd
index b0d0edb44..c391dabbd 100644
--- a/man/layer_depthwise_conv_1d.Rd
+++ b/man/layer_depthwise_conv_1d.Rd
@@ -245,6 +245,7 @@ Other layers: \cr
 \code{\link{layer_random_color_jitter}()} \cr
 \code{\link{layer_random_contrast}()} \cr
 \code{\link{layer_random_crop}()} \cr
+\code{\link{layer_random_elastic_transform}()} \cr
 \code{\link{layer_random_erasing}()} \cr
 \code{\link{layer_random_flip}()} \cr
 \code{\link{layer_random_gaussian_blur}()} \cr
diff --git a/man/layer_depthwise_conv_2d.Rd b/man/layer_depthwise_conv_2d.Rd
index 09cace4a4..2c1df1096 100644
--- a/man/layer_depthwise_conv_2d.Rd
+++ b/man/layer_depthwise_conv_2d.Rd
@@ -246,6 +246,7 @@ Other layers: \cr
 \code{\link{layer_random_color_jitter}()} \cr
 \code{\link{layer_random_contrast}()} \cr
 \code{\link{layer_random_crop}()} \cr
+\code{\link{layer_random_elastic_transform}()} \cr
 \code{\link{layer_random_erasing}()} \cr
 \code{\link{layer_random_flip}()} \cr
 \code{\link{layer_random_gaussian_blur}()} \cr
diff --git a/man/layer_discretization.Rd b/man/layer_discretization.Rd
index 4538b5db8..915fc3c82 100644
--- a/man/layer_discretization.Rd
+++ b/man/layer_discretization.Rd
@@ -160,6 +160,7 @@ Other preprocessing layers: \cr
 \code{\link{layer_random_color_jitter}()} \cr
 \code{\link{layer_random_contrast}()} \cr
 \code{\link{layer_random_crop}()} \cr
+\code{\link{layer_random_elastic_transform}()} \cr
 \code{\link{layer_random_erasing}()} \cr
 \code{\link{layer_random_flip}()} \cr
 \code{\link{layer_random_gaussian_blur}()} \cr
@@ -267,6 +268,7 @@ Other layers: \cr
 \code{\link{layer_random_color_jitter}()} \cr
 \code{\link{layer_random_contrast}()} \cr
 \code{\link{layer_random_crop}()} \cr
+\code{\link{layer_random_elastic_transform}()} \cr
 \code{\link{layer_random_erasing}()} \cr
 \code{\link{layer_random_flip}()} \cr
 \code{\link{layer_random_gaussian_blur}()} \cr
diff --git a/man/layer_dot.Rd b/man/layer_dot.Rd
index a369160e8..3f7f91d41 100644
--- a/man/layer_dot.Rd
+++ b/man/layer_dot.Rd
@@ -176,6 +176,7 @@ Other layers: \cr
 \code{\link{layer_random_color_jitter}()} \cr
 \code{\link{layer_random_contrast}()} \cr
 \code{\link{layer_random_crop}()} \cr
+\code{\link{layer_random_elastic_transform}()} \cr
 \code{\link{layer_random_erasing}()} \cr
 \code{\link{layer_random_flip}()} \cr
 \code{\link{layer_random_gaussian_blur}()} \cr
diff --git a/man/layer_dropout.Rd b/man/layer_dropout.Rd
index 4936fdacf..70bc0713d 100644
--- a/man/layer_dropout.Rd
+++ b/man/layer_dropout.Rd
@@ -157,6 +157,7 @@ Other layers: \cr
 \code{\link{layer_random_color_jitter}()} \cr
 \code{\link{layer_random_contrast}()} \cr
 \code{\link{layer_random_crop}()} \cr
+\code{\link{layer_random_elastic_transform}()} \cr
 \code{\link{layer_random_erasing}()} \cr
 \code{\link{layer_random_flip}()} \cr
 \code{\link{layer_random_gaussian_blur}()} \cr
diff --git a/man/layer_einsum_dense.Rd b/man/layer_einsum_dense.Rd
index 39975114d..9a09ceab8 100644
--- a/man/layer_einsum_dense.Rd
+++ b/man/layer_einsum_dense.Rd
@@ -271,6 +271,7 @@ Other layers: \cr
 \code{\link{layer_random_color_jitter}()} \cr
 \code{\link{layer_random_contrast}()} \cr
 \code{\link{layer_random_crop}()} \cr
+\code{\link{layer_random_elastic_transform}()} \cr
 \code{\link{layer_random_erasing}()} \cr
 \code{\link{layer_random_flip}()} \cr
 \code{\link{layer_random_gaussian_blur}()} \cr
diff --git a/man/layer_embedding.Rd b/man/layer_embedding.Rd
index 3a865ffd8..f0540ce03 100644
--- a/man/layer_embedding.Rd
+++ b/man/layer_embedding.Rd
@@ -244,6 +244,7 @@ Other layers: \cr
 \code{\link{layer_random_color_jitter}()} \cr
 \code{\link{layer_random_contrast}()} \cr
 \code{\link{layer_random_crop}()} \cr
+\code{\link{layer_random_elastic_transform}()} \cr
 \code{\link{layer_random_erasing}()} \cr
 \code{\link{layer_random_flip}()} \cr
 \code{\link{layer_random_gaussian_blur}()} \cr
diff --git a/man/layer_equalization.Rd b/man/layer_equalization.Rd
index 954606d54..13c0da9a1 100644
--- a/man/layer_equalization.Rd
+++ b/man/layer_equalization.Rd
@@ -91,6 +91,7 @@ Other image preprocessing layers: \cr
 \code{\link{layer_rand_augment}()} \cr
 \code{\link{layer_random_color_degeneration}()} \cr
 \code{\link{layer_random_color_jitter}()} \cr
+\code{\link{layer_random_elastic_transform}()} \cr
 \code{\link{layer_random_erasing}()} \cr
 \code{\link{layer_random_gaussian_blur}()} \cr
 \code{\link{layer_random_grayscale}()} \cr
@@ -126,6 +127,7 @@ Other preprocessing layers: \cr
 \code{\link{layer_random_color_jitter}()} \cr
 \code{\link{layer_random_contrast}()} \cr
 \code{\link{layer_random_crop}()} \cr
+\code{\link{layer_random_elastic_transform}()} \cr
 \code{\link{layer_random_erasing}()} \cr
 \code{\link{layer_random_flip}()} \cr
 \code{\link{layer_random_gaussian_blur}()} \cr
@@ -233,6 +235,7 @@ Other layers: \cr
 \code{\link{layer_random_color_jitter}()} \cr
 \code{\link{layer_random_contrast}()} \cr
 \code{\link{layer_random_crop}()} \cr
+\code{\link{layer_random_elastic_transform}()} \cr
 \code{\link{layer_random_erasing}()} \cr
 \code{\link{layer_random_flip}()} \cr
 \code{\link{layer_random_gaussian_blur}()} \cr
diff --git a/man/layer_feature_space.Rd b/man/layer_feature_space.Rd
index 6f08b5712..5e0d15892 100644
--- a/man/layer_feature_space.Rd
+++ b/man/layer_feature_space.Rd
@@ -345,6 +345,7 @@ Other preprocessing layers: \cr
 \code{\link{layer_random_color_jitter}()} \cr
 \code{\link{layer_random_contrast}()} \cr
 \code{\link{layer_random_crop}()} \cr
+\code{\link{layer_random_elastic_transform}()} \cr
 \code{\link{layer_random_erasing}()} \cr
 \code{\link{layer_random_flip}()} \cr
 \code{\link{layer_random_gaussian_blur}()} \cr
@@ -452,6 +453,7 @@ Other layers: \cr
 \code{\link{layer_random_color_jitter}()} \cr
 \code{\link{layer_random_contrast}()} \cr
 \code{\link{layer_random_crop}()} \cr
+\code{\link{layer_random_elastic_transform}()} \cr
 \code{\link{layer_random_erasing}()} \cr
 \code{\link{layer_random_flip}()} \cr
 \code{\link{layer_random_gaussian_blur}()} \cr
diff --git a/man/layer_flatten.Rd b/man/layer_flatten.Rd
index 5bce1c156..a2f8223a1 100644
--- a/man/layer_flatten.Rd
+++ b/man/layer_flatten.Rd
@@ -155,6 +155,7 @@ Other layers: \cr
 \code{\link{layer_random_color_jitter}()} \cr
 \code{\link{layer_random_contrast}()} \cr
 \code{\link{layer_random_crop}()} \cr
+\code{\link{layer_random_elastic_transform}()} \cr
 \code{\link{layer_random_erasing}()} \cr
 \code{\link{layer_random_flip}()} \cr
 \code{\link{layer_random_gaussian_blur}()} \cr
diff --git a/man/layer_flax_module_wrapper.Rd b/man/layer_flax_module_wrapper.Rd
index a47799f63..9236c9091 100644
--- a/man/layer_flax_module_wrapper.Rd
+++ b/man/layer_flax_module_wrapper.Rd
@@ -229,6 +229,7 @@ Other layers: \cr
 \code{\link{layer_random_color_jitter}()} \cr
 \code{\link{layer_random_contrast}()} \cr
 \code{\link{layer_random_crop}()} \cr
+\code{\link{layer_random_elastic_transform}()} \cr
 \code{\link{layer_random_erasing}()} \cr
 \code{\link{layer_random_flip}()} \cr
 \code{\link{layer_random_gaussian_blur}()} \cr
diff --git a/man/layer_gaussian_dropout.Rd b/man/layer_gaussian_dropout.Rd
index f5be1bd14..eefc7c32b 100644
--- a/man/layer_gaussian_dropout.Rd
+++ b/man/layer_gaussian_dropout.Rd
@@ -139,6 +139,7 @@ Other layers: \cr
 \code{\link{layer_random_color_jitter}()} \cr
 \code{\link{layer_random_contrast}()} \cr
 \code{\link{layer_random_crop}()} \cr
+\code{\link{layer_random_elastic_transform}()} \cr
 \code{\link{layer_random_erasing}()} \cr
 \code{\link{layer_random_flip}()} \cr
 \code{\link{layer_random_gaussian_blur}()} \cr
diff --git a/man/layer_gaussian_noise.Rd b/man/layer_gaussian_noise.Rd
index e0163a298..3f2fc4d19 100644
--- a/man/layer_gaussian_noise.Rd
+++ b/man/layer_gaussian_noise.Rd
@@ -142,6 +142,7 @@ Other layers: \cr
 \code{\link{layer_random_color_jitter}()} \cr
 \code{\link{layer_random_contrast}()} \cr
 \code{\link{layer_random_crop}()} \cr
+\code{\link{layer_random_elastic_transform}()} \cr
 \code{\link{layer_random_erasing}()} \cr
 \code{\link{layer_random_flip}()} \cr
 \code{\link{layer_random_gaussian_blur}()} \cr
diff --git a/man/layer_global_average_pooling_1d.Rd b/man/layer_global_average_pooling_1d.Rd
index b2d9c2768..3eaaf61a6 100644
--- a/man/layer_global_average_pooling_1d.Rd
+++ b/man/layer_global_average_pooling_1d.Rd
@@ -192,6 +192,7 @@ Other layers: \cr
 \code{\link{layer_random_color_jitter}()} \cr
 \code{\link{layer_random_contrast}()} \cr
 \code{\link{layer_random_crop}()} \cr
+\code{\link{layer_random_elastic_transform}()} \cr
 \code{\link{layer_random_erasing}()} \cr
 \code{\link{layer_random_flip}()} \cr
 \code{\link{layer_random_gaussian_blur}()} \cr
diff --git a/man/layer_global_average_pooling_2d.Rd b/man/layer_global_average_pooling_2d.Rd
index a6aad835f..6c015633c 100644
--- a/man/layer_global_average_pooling_2d.Rd
+++ b/man/layer_global_average_pooling_2d.Rd
@@ -185,6 +185,7 @@ Other layers: \cr
 \code{\link{layer_random_color_jitter}()} \cr
 \code{\link{layer_random_contrast}()} \cr
 \code{\link{layer_random_crop}()} \cr
+\code{\link{layer_random_elastic_transform}()} \cr
 \code{\link{layer_random_erasing}()} \cr
 \code{\link{layer_random_flip}()} \cr
 \code{\link{layer_random_gaussian_blur}()} \cr
diff --git a/man/layer_global_average_pooling_3d.Rd b/man/layer_global_average_pooling_3d.Rd
index 40d735a4e..5c14a026a 100644
--- a/man/layer_global_average_pooling_3d.Rd
+++ b/man/layer_global_average_pooling_3d.Rd
@@ -186,6 +186,7 @@ Other layers: \cr
 \code{\link{layer_random_color_jitter}()} \cr
 \code{\link{layer_random_contrast}()} \cr
 \code{\link{layer_random_crop}()} \cr
+\code{\link{layer_random_elastic_transform}()} \cr
 \code{\link{layer_random_erasing}()} \cr
 \code{\link{layer_random_flip}()} \cr
 \code{\link{layer_random_gaussian_blur}()} \cr
diff --git a/man/layer_global_max_pooling_1d.Rd b/man/layer_global_max_pooling_1d.Rd
index d0198afbe..7d477ae4b 100644
--- a/man/layer_global_max_pooling_1d.Rd
+++ b/man/layer_global_max_pooling_1d.Rd
@@ -179,6 +179,7 @@ Other layers: \cr
 \code{\link{layer_random_color_jitter}()} \cr
 \code{\link{layer_random_contrast}()} \cr
 \code{\link{layer_random_crop}()} \cr
+\code{\link{layer_random_elastic_transform}()} \cr
 \code{\link{layer_random_erasing}()} \cr
 \code{\link{layer_random_flip}()} \cr
 \code{\link{layer_random_gaussian_blur}()} \cr
diff --git a/man/layer_global_max_pooling_2d.Rd b/man/layer_global_max_pooling_2d.Rd
index 8dc193930..60ef3f162 100644
--- a/man/layer_global_max_pooling_2d.Rd
+++ b/man/layer_global_max_pooling_2d.Rd
@@ -180,6 +180,7 @@ Other layers: \cr
 \code{\link{layer_random_color_jitter}()} \cr
 \code{\link{layer_random_contrast}()} \cr
 \code{\link{layer_random_crop}()} \cr
+\code{\link{layer_random_elastic_transform}()} \cr
 \code{\link{layer_random_erasing}()} \cr
 \code{\link{layer_random_flip}()} \cr
 \code{\link{layer_random_gaussian_blur}()} \cr
diff --git a/man/layer_global_max_pooling_3d.Rd b/man/layer_global_max_pooling_3d.Rd
index 4ba5d14f1..e0203f7c8 100644
--- a/man/layer_global_max_pooling_3d.Rd
+++ b/man/layer_global_max_pooling_3d.Rd
@@ -181,6 +181,7 @@ Other layers: \cr
 \code{\link{layer_random_color_jitter}()} \cr
 \code{\link{layer_random_contrast}()} \cr
 \code{\link{layer_random_crop}()} \cr
+\code{\link{layer_random_elastic_transform}()} \cr
 \code{\link{layer_random_erasing}()} \cr
 \code{\link{layer_random_flip}()} \cr
 \code{\link{layer_random_gaussian_blur}()} \cr
diff --git a/man/layer_group_normalization.Rd b/man/layer_group_normalization.Rd
index 933861f7b..0d33ff64c 100644
--- a/man/layer_group_normalization.Rd
+++ b/man/layer_group_normalization.Rd
@@ -204,6 +204,7 @@ Other layers: \cr
 \code{\link{layer_random_color_jitter}()} \cr
 \code{\link{layer_random_contrast}()} \cr
 \code{\link{layer_random_crop}()} \cr
+\code{\link{layer_random_elastic_transform}()} \cr
 \code{\link{layer_random_erasing}()} \cr
 \code{\link{layer_random_flip}()} \cr
 \code{\link{layer_random_gaussian_blur}()} \cr
diff --git a/man/layer_group_query_attention.Rd b/man/layer_group_query_attention.Rd
index 91f3e7816..525892c65 100644
--- a/man/layer_group_query_attention.Rd
+++ b/man/layer_group_query_attention.Rd
@@ -206,6 +206,7 @@ Other layers: \cr
 \code{\link{layer_random_color_jitter}()} \cr
 \code{\link{layer_random_contrast}()} \cr
 \code{\link{layer_random_crop}()} \cr
+\code{\link{layer_random_elastic_transform}()} \cr
 \code{\link{layer_random_erasing}()} \cr
 \code{\link{layer_random_flip}()} \cr
 \code{\link{layer_random_gaussian_blur}()} \cr
diff --git a/man/layer_gru.Rd b/man/layer_gru.Rd
index 5de983fb6..53be001f9 100644
--- a/man/layer_gru.Rd
+++ b/man/layer_gru.Rd
@@ -317,6 +317,7 @@ Other layers: \cr
 \code{\link{layer_random_color_jitter}()} \cr
 \code{\link{layer_random_contrast}()} \cr
 \code{\link{layer_random_crop}()} \cr
+\code{\link{layer_random_elastic_transform}()} \cr
 \code{\link{layer_random_erasing}()} \cr
 \code{\link{layer_random_flip}()} \cr
 \code{\link{layer_random_gaussian_blur}()} \cr
diff --git a/man/layer_hashed_crossing.Rd b/man/layer_hashed_crossing.Rd
index 073697dcc..260c3f42f 100644
--- a/man/layer_hashed_crossing.Rd
+++ b/man/layer_hashed_crossing.Rd
@@ -132,6 +132,7 @@ Other preprocessing layers: \cr
 \code{\link{layer_random_color_jitter}()} \cr
 \code{\link{layer_random_contrast}()} \cr
 \code{\link{layer_random_crop}()} \cr
+\code{\link{layer_random_elastic_transform}()} \cr
 \code{\link{layer_random_erasing}()} \cr
 \code{\link{layer_random_flip}()} \cr
 \code{\link{layer_random_gaussian_blur}()} \cr
@@ -239,6 +240,7 @@ Other layers: \cr
 \code{\link{layer_random_color_jitter}()} \cr
 \code{\link{layer_random_contrast}()} \cr
 \code{\link{layer_random_crop}()} \cr
+\code{\link{layer_random_elastic_transform}()} \cr
 \code{\link{layer_random_erasing}()} \cr
 \code{\link{layer_random_flip}()} \cr
 \code{\link{layer_random_gaussian_blur}()} \cr
diff --git a/man/layer_hashing.Rd b/man/layer_hashing.Rd
index 8080c39d8..8d9ee46bd 100644
--- a/man/layer_hashing.Rd
+++ b/man/layer_hashing.Rd
@@ -213,6 +213,7 @@ Other preprocessing layers: \cr
 \code{\link{layer_random_color_jitter}()} \cr
 \code{\link{layer_random_contrast}()} \cr
 \code{\link{layer_random_crop}()} \cr
+\code{\link{layer_random_elastic_transform}()} \cr
 \code{\link{layer_random_erasing}()} \cr
 \code{\link{layer_random_flip}()} \cr
 \code{\link{layer_random_gaussian_blur}()} \cr
@@ -320,6 +321,7 @@ Other layers: \cr
 \code{\link{layer_random_color_jitter}()} \cr
 \code{\link{layer_random_contrast}()} \cr
 \code{\link{layer_random_crop}()} \cr
+\code{\link{layer_random_elastic_transform}()} \cr
 \code{\link{layer_random_erasing}()} \cr
 \code{\link{layer_random_flip}()} \cr
 \code{\link{layer_random_gaussian_blur}()} \cr
diff --git a/man/layer_identity.Rd b/man/layer_identity.Rd
index 1b4fe2441..5d7b83e51 100644
--- a/man/layer_identity.Rd
+++ b/man/layer_identity.Rd
@@ -120,6 +120,7 @@ Other layers: \cr
 \code{\link{layer_random_color_jitter}()} \cr
 \code{\link{layer_random_contrast}()} \cr
 \code{\link{layer_random_crop}()} \cr
+\code{\link{layer_random_elastic_transform}()} \cr
 \code{\link{layer_random_erasing}()} \cr
 \code{\link{layer_random_flip}()} \cr
 \code{\link{layer_random_gaussian_blur}()} \cr
diff --git a/man/layer_integer_lookup.Rd b/man/layer_integer_lookup.Rd
index 7a810d372..0390ec513 100644
--- a/man/layer_integer_lookup.Rd
+++ b/man/layer_integer_lookup.Rd
@@ -439,6 +439,7 @@ Other preprocessing layers: \cr
 \code{\link{layer_random_color_jitter}()} \cr
 \code{\link{layer_random_contrast}()} \cr
 \code{\link{layer_random_crop}()} \cr
+\code{\link{layer_random_elastic_transform}()} \cr
 \code{\link{layer_random_erasing}()} \cr
 \code{\link{layer_random_flip}()} \cr
 \code{\link{layer_random_gaussian_blur}()} \cr
@@ -546,6 +547,7 @@ Other layers: \cr
 \code{\link{layer_random_color_jitter}()} \cr
 \code{\link{layer_random_contrast}()} \cr
 \code{\link{layer_random_crop}()} \cr
+\code{\link{layer_random_elastic_transform}()} \cr
 \code{\link{layer_random_erasing}()} \cr
 \code{\link{layer_random_flip}()} \cr
 \code{\link{layer_random_gaussian_blur}()} \cr
diff --git a/man/layer_jax_model_wrapper.Rd b/man/layer_jax_model_wrapper.Rd
index 9496c02ab..a52be755a 100644
--- a/man/layer_jax_model_wrapper.Rd
+++ b/man/layer_jax_model_wrapper.Rd
@@ -335,6 +335,7 @@ Other layers: \cr
 \code{\link{layer_random_color_jitter}()} \cr
 \code{\link{layer_random_contrast}()} \cr
 \code{\link{layer_random_crop}()} \cr
+\code{\link{layer_random_elastic_transform}()} \cr
 \code{\link{layer_random_erasing}()} \cr
 \code{\link{layer_random_flip}()} \cr
 \code{\link{layer_random_gaussian_blur}()} \cr
diff --git a/man/layer_lambda.Rd b/man/layer_lambda.Rd
index f6be69d3b..0a219ddf3 100644
--- a/man/layer_lambda.Rd
+++ b/man/layer_lambda.Rd
@@ -164,6 +164,7 @@ Other layers: \cr
 \code{\link{layer_random_color_jitter}()} \cr
 \code{\link{layer_random_contrast}()} \cr
 \code{\link{layer_random_crop}()} \cr
+\code{\link{layer_random_elastic_transform}()} \cr
 \code{\link{layer_random_erasing}()} \cr
 \code{\link{layer_random_flip}()} \cr
 \code{\link{layer_random_gaussian_blur}()} \cr
diff --git a/man/layer_layer_normalization.Rd b/man/layer_layer_normalization.Rd
index 365db89d8..e4e90f2bc 100644
--- a/man/layer_layer_normalization.Rd
+++ b/man/layer_layer_normalization.Rd
@@ -233,6 +233,7 @@ Other layers: \cr
 \code{\link{layer_random_color_jitter}()} \cr
 \code{\link{layer_random_contrast}()} \cr
 \code{\link{layer_random_crop}()} \cr
+\code{\link{layer_random_elastic_transform}()} \cr
 \code{\link{layer_random_erasing}()} \cr
 \code{\link{layer_random_flip}()} \cr
 \code{\link{layer_random_gaussian_blur}()} \cr
diff --git a/man/layer_lstm.Rd b/man/layer_lstm.Rd
index 5a4e578e3..d5c40f622 100644
--- a/man/layer_lstm.Rd
+++ b/man/layer_lstm.Rd
@@ -312,6 +312,7 @@ Other layers: \cr
 \code{\link{layer_random_color_jitter}()} \cr
 \code{\link{layer_random_contrast}()} \cr
 \code{\link{layer_random_crop}()} \cr
+\code{\link{layer_random_elastic_transform}()} \cr
 \code{\link{layer_random_erasing}()} \cr
 \code{\link{layer_random_flip}()} \cr
 \code{\link{layer_random_gaussian_blur}()} \cr
diff --git a/man/layer_masking.Rd b/man/layer_masking.Rd
index d6dfa05b5..0a5be6c88 100644
--- a/man/layer_masking.Rd
+++ b/man/layer_masking.Rd
@@ -160,6 +160,7 @@ Other layers: \cr
 \code{\link{layer_random_color_jitter}()} \cr
 \code{\link{layer_random_contrast}()} \cr
 \code{\link{layer_random_crop}()} \cr
+\code{\link{layer_random_elastic_transform}()} \cr
 \code{\link{layer_random_erasing}()} \cr
 \code{\link{layer_random_flip}()} \cr
 \code{\link{layer_random_gaussian_blur}()} \cr
diff --git a/man/layer_max_num_bounding_boxes.Rd b/man/layer_max_num_bounding_boxes.Rd
index 55c5586d1..a4ed0b067 100644
--- a/man/layer_max_num_bounding_boxes.Rd
+++ b/man/layer_max_num_bounding_boxes.Rd
@@ -41,6 +41,7 @@ Other image preprocessing layers: \cr
 \code{\link{layer_rand_augment}()} \cr
 \code{\link{layer_random_color_degeneration}()} \cr
 \code{\link{layer_random_color_jitter}()} \cr
+\code{\link{layer_random_elastic_transform}()} \cr
 \code{\link{layer_random_erasing}()} \cr
 \code{\link{layer_random_gaussian_blur}()} \cr
 \code{\link{layer_random_grayscale}()} \cr
@@ -76,6 +77,7 @@ Other preprocessing layers: \cr
 \code{\link{layer_random_color_jitter}()} \cr
 \code{\link{layer_random_contrast}()} \cr
 \code{\link{layer_random_crop}()} \cr
+\code{\link{layer_random_elastic_transform}()} \cr
 \code{\link{layer_random_erasing}()} \cr
 \code{\link{layer_random_flip}()} \cr
 \code{\link{layer_random_gaussian_blur}()} \cr
@@ -183,6 +185,7 @@ Other layers: \cr
 \code{\link{layer_random_color_jitter}()} \cr
 \code{\link{layer_random_contrast}()} \cr
 \code{\link{layer_random_crop}()} \cr
+\code{\link{layer_random_elastic_transform}()} \cr
 \code{\link{layer_random_erasing}()} \cr
 \code{\link{layer_random_flip}()} \cr
 \code{\link{layer_random_gaussian_blur}()} \cr
diff --git a/man/layer_max_pooling_1d.Rd b/man/layer_max_pooling_1d.Rd
index 30b9588f9..088bb5270 100644
--- a/man/layer_max_pooling_1d.Rd
+++ b/man/layer_max_pooling_1d.Rd
@@ -237,6 +237,7 @@ Other layers: \cr
 \code{\link{layer_random_color_jitter}()} \cr
 \code{\link{layer_random_contrast}()} \cr
 \code{\link{layer_random_crop}()} \cr
+\code{\link{layer_random_elastic_transform}()} \cr
 \code{\link{layer_random_erasing}()} \cr
 \code{\link{layer_random_flip}()} \cr
 \code{\link{layer_random_gaussian_blur}()} \cr
diff --git a/man/layer_max_pooling_2d.Rd b/man/layer_max_pooling_2d.Rd
index 9f3d88cfe..452bcbaec 100644
--- a/man/layer_max_pooling_2d.Rd
+++ b/man/layer_max_pooling_2d.Rd
@@ -257,6 +257,7 @@ Other layers: \cr
 \code{\link{layer_random_color_jitter}()} \cr
 \code{\link{layer_random_contrast}()} \cr
 \code{\link{layer_random_crop}()} \cr
+\code{\link{layer_random_elastic_transform}()} \cr
 \code{\link{layer_random_erasing}()} \cr
 \code{\link{layer_random_flip}()} \cr
 \code{\link{layer_random_gaussian_blur}()} \cr
diff --git a/man/layer_max_pooling_3d.Rd b/man/layer_max_pooling_3d.Rd
index 8b2ef1521..e61304edb 100644
--- a/man/layer_max_pooling_3d.Rd
+++ b/man/layer_max_pooling_3d.Rd
@@ -204,6 +204,7 @@ Other layers: \cr
 \code{\link{layer_random_color_jitter}()} \cr
 \code{\link{layer_random_contrast}()} \cr
 \code{\link{layer_random_crop}()} \cr
+\code{\link{layer_random_elastic_transform}()} \cr
 \code{\link{layer_random_erasing}()} \cr
 \code{\link{layer_random_flip}()} \cr
 \code{\link{layer_random_gaussian_blur}()} \cr
diff --git a/man/layer_maximum.Rd b/man/layer_maximum.Rd
index ca9636a89..ae275327b 100644
--- a/man/layer_maximum.Rd
+++ b/man/layer_maximum.Rd
@@ -146,6 +146,7 @@ Other layers: \cr
 \code{\link{layer_random_color_jitter}()} \cr
 \code{\link{layer_random_contrast}()} \cr
 \code{\link{layer_random_crop}()} \cr
+\code{\link{layer_random_elastic_transform}()} \cr
 \code{\link{layer_random_erasing}()} \cr
 \code{\link{layer_random_flip}()} \cr
 \code{\link{layer_random_gaussian_blur}()} \cr
diff --git a/man/layer_mel_spectrogram.Rd b/man/layer_mel_spectrogram.Rd
index a416f338d..37ee76a9d 100644
--- a/man/layer_mel_spectrogram.Rd
+++ b/man/layer_mel_spectrogram.Rd
@@ -153,6 +153,7 @@ Other preprocessing layers: \cr
 \code{\link{layer_random_color_jitter}()} \cr
 \code{\link{layer_random_contrast}()} \cr
 \code{\link{layer_random_crop}()} \cr
+\code{\link{layer_random_elastic_transform}()} \cr
 \code{\link{layer_random_erasing}()} \cr
 \code{\link{layer_random_flip}()} \cr
 \code{\link{layer_random_gaussian_blur}()} \cr
@@ -260,6 +261,7 @@ Other layers: \cr
 \code{\link{layer_random_color_jitter}()} \cr
 \code{\link{layer_random_contrast}()} \cr
 \code{\link{layer_random_crop}()} \cr
+\code{\link{layer_random_elastic_transform}()} \cr
 \code{\link{layer_random_erasing}()} \cr
 \code{\link{layer_random_flip}()} \cr
 \code{\link{layer_random_gaussian_blur}()} \cr
diff --git a/man/layer_minimum.Rd b/man/layer_minimum.Rd
index b4cd26757..6dfeb015e 100644
--- a/man/layer_minimum.Rd
+++ b/man/layer_minimum.Rd
@@ -146,6 +146,7 @@ Other layers: \cr
 \code{\link{layer_random_color_jitter}()} \cr
 \code{\link{layer_random_contrast}()} \cr
 \code{\link{layer_random_crop}()} \cr
+\code{\link{layer_random_elastic_transform}()} \cr
 \code{\link{layer_random_erasing}()} \cr
 \code{\link{layer_random_flip}()} \cr
 \code{\link{layer_random_gaussian_blur}()} \cr
diff --git a/man/layer_mix_up.Rd b/man/layer_mix_up.Rd
index 5b56d9fe9..e50b7f1e2 100644
--- a/man/layer_mix_up.Rd
+++ b/man/layer_mix_up.Rd
@@ -55,6 +55,7 @@ Other image preprocessing layers: \cr
 \code{\link{layer_rand_augment}()} \cr
 \code{\link{layer_random_color_degeneration}()} \cr
 \code{\link{layer_random_color_jitter}()} \cr
+\code{\link{layer_random_elastic_transform}()} \cr
 \code{\link{layer_random_erasing}()} \cr
 \code{\link{layer_random_gaussian_blur}()} \cr
 \code{\link{layer_random_grayscale}()} \cr
@@ -90,6 +91,7 @@ Other preprocessing layers: \cr
 \code{\link{layer_random_color_jitter}()} \cr
 \code{\link{layer_random_contrast}()} \cr
 \code{\link{layer_random_crop}()} \cr
+\code{\link{layer_random_elastic_transform}()} \cr
 \code{\link{layer_random_erasing}()} \cr
 \code{\link{layer_random_flip}()} \cr
 \code{\link{layer_random_gaussian_blur}()} \cr
@@ -197,6 +199,7 @@ Other layers: \cr
 \code{\link{layer_random_color_jitter}()} \cr
 \code{\link{layer_random_contrast}()} \cr
 \code{\link{layer_random_crop}()} \cr
+\code{\link{layer_random_elastic_transform}()} \cr
 \code{\link{layer_random_erasing}()} \cr
 \code{\link{layer_random_flip}()} \cr
 \code{\link{layer_random_gaussian_blur}()} \cr
diff --git a/man/layer_multi_head_attention.Rd b/man/layer_multi_head_attention.Rd
index 680e16998..f834beeec 100644
--- a/man/layer_multi_head_attention.Rd
+++ b/man/layer_multi_head_attention.Rd
@@ -254,6 +254,7 @@ Other layers: \cr
 \code{\link{layer_random_color_jitter}()} \cr
 \code{\link{layer_random_contrast}()} \cr
 \code{\link{layer_random_crop}()} \cr
+\code{\link{layer_random_elastic_transform}()} \cr
 \code{\link{layer_random_erasing}()} \cr
 \code{\link{layer_random_flip}()} \cr
 \code{\link{layer_random_gaussian_blur}()} \cr
diff --git a/man/layer_multiply.Rd b/man/layer_multiply.Rd
index 6c470f049..96ade0155 100644
--- a/man/layer_multiply.Rd
+++ b/man/layer_multiply.Rd
@@ -146,6 +146,7 @@ Other layers: \cr
 \code{\link{layer_random_color_jitter}()} \cr
 \code{\link{layer_random_contrast}()} \cr
 \code{\link{layer_random_crop}()} \cr
+\code{\link{layer_random_elastic_transform}()} \cr
 \code{\link{layer_random_erasing}()} \cr
 \code{\link{layer_random_flip}()} \cr
 \code{\link{layer_random_gaussian_blur}()} \cr
diff --git a/man/layer_normalization.Rd b/man/layer_normalization.Rd
index 1aa7bccca..e801aab3c 100644
--- a/man/layer_normalization.Rd
+++ b/man/layer_normalization.Rd
@@ -157,6 +157,7 @@ Other preprocessing layers: \cr
 \code{\link{layer_random_color_jitter}()} \cr
 \code{\link{layer_random_contrast}()} \cr
 \code{\link{layer_random_crop}()} \cr
+\code{\link{layer_random_elastic_transform}()} \cr
 \code{\link{layer_random_erasing}()} \cr
 \code{\link{layer_random_flip}()} \cr
 \code{\link{layer_random_gaussian_blur}()} \cr
@@ -264,6 +265,7 @@ Other layers: \cr
 \code{\link{layer_random_color_jitter}()} \cr
 \code{\link{layer_random_contrast}()} \cr
 \code{\link{layer_random_crop}()} \cr
+\code{\link{layer_random_elastic_transform}()} \cr
 \code{\link{layer_random_erasing}()} \cr
 \code{\link{layer_random_flip}()} \cr
 \code{\link{layer_random_gaussian_blur}()} \cr
diff --git a/man/layer_permute.Rd b/man/layer_permute.Rd
index 0347eafe0..01ec8a864 100644
--- a/man/layer_permute.Rd
+++ b/man/layer_permute.Rd
@@ -155,6 +155,7 @@ Other layers: \cr
 \code{\link{layer_random_color_jitter}()} \cr
 \code{\link{layer_random_contrast}()} \cr
 \code{\link{layer_random_crop}()} \cr
+\code{\link{layer_random_elastic_transform}()} \cr
 \code{\link{layer_random_erasing}()} \cr
 \code{\link{layer_random_flip}()} \cr
 \code{\link{layer_random_gaussian_blur}()} \cr
diff --git a/man/layer_rand_augment.Rd b/man/layer_rand_augment.Rd
index 5ef328eb9..cc9dea40d 100644
--- a/man/layer_rand_augment.Rd
+++ b/man/layer_rand_augment.Rd
@@ -64,6 +64,7 @@ Other image preprocessing layers: \cr
 \code{\link{layer_mix_up}()} \cr
 \code{\link{layer_random_color_degeneration}()} \cr
 \code{\link{layer_random_color_jitter}()} \cr
+\code{\link{layer_random_elastic_transform}()} \cr
 \code{\link{layer_random_erasing}()} \cr
 \code{\link{layer_random_gaussian_blur}()} \cr
 \code{\link{layer_random_grayscale}()} \cr
@@ -99,6 +100,7 @@ Other preprocessing layers: \cr
 \code{\link{layer_random_color_jitter}()} \cr
 \code{\link{layer_random_contrast}()} \cr
 \code{\link{layer_random_crop}()} \cr
+\code{\link{layer_random_elastic_transform}()} \cr
 \code{\link{layer_random_erasing}()} \cr
 \code{\link{layer_random_flip}()} \cr
 \code{\link{layer_random_gaussian_blur}()} \cr
@@ -206,6 +208,7 @@ Other layers: \cr
 \code{\link{layer_random_color_jitter}()} \cr
 \code{\link{layer_random_contrast}()} \cr
 \code{\link{layer_random_crop}()} \cr
+\code{\link{layer_random_elastic_transform}()} \cr
 \code{\link{layer_random_erasing}()} \cr
 \code{\link{layer_random_flip}()} \cr
 \code{\link{layer_random_gaussian_blur}()} \cr
diff --git a/man/layer_random_brightness.Rd b/man/layer_random_brightness.Rd
index 299d7cb84..5fba18465 100644
--- a/man/layer_random_brightness.Rd
+++ b/man/layer_random_brightness.Rd
@@ -124,6 +124,7 @@ Other preprocessing layers: \cr
 \code{\link{layer_random_color_jitter}()} \cr
 \code{\link{layer_random_contrast}()} \cr
 \code{\link{layer_random_crop}()} \cr
+\code{\link{layer_random_elastic_transform}()} \cr
 \code{\link{layer_random_erasing}()} \cr
 \code{\link{layer_random_flip}()} \cr
 \code{\link{layer_random_gaussian_blur}()} \cr
@@ -231,6 +232,7 @@ Other layers: \cr
 \code{\link{layer_random_color_jitter}()} \cr
 \code{\link{layer_random_contrast}()} \cr
 \code{\link{layer_random_crop}()} \cr
+\code{\link{layer_random_elastic_transform}()} \cr
 \code{\link{layer_random_erasing}()} \cr
 \code{\link{layer_random_flip}()} \cr
 \code{\link{layer_random_gaussian_blur}()} \cr
diff --git a/man/layer_random_color_degeneration.Rd b/man/layer_random_color_degeneration.Rd
index 60ada417a..2f831459e 100644
--- a/man/layer_random_color_degeneration.Rd
+++ b/man/layer_random_color_degeneration.Rd
@@ -59,6 +59,7 @@ Other image preprocessing layers: \cr
 \code{\link{layer_mix_up}()} \cr
 \code{\link{layer_rand_augment}()} \cr
 \code{\link{layer_random_color_jitter}()} \cr
+\code{\link{layer_random_elastic_transform}()} \cr
 \code{\link{layer_random_erasing}()} \cr
 \code{\link{layer_random_gaussian_blur}()} \cr
 \code{\link{layer_random_grayscale}()} \cr
@@ -94,6 +95,7 @@ Other preprocessing layers: \cr
 \code{\link{layer_random_color_jitter}()} \cr
 \code{\link{layer_random_contrast}()} \cr
 \code{\link{layer_random_crop}()} \cr
+\code{\link{layer_random_elastic_transform}()} \cr
 \code{\link{layer_random_erasing}()} \cr
 \code{\link{layer_random_flip}()} \cr
 \code{\link{layer_random_gaussian_blur}()} \cr
@@ -201,6 +203,7 @@ Other layers: \cr
 \code{\link{layer_random_color_jitter}()} \cr
 \code{\link{layer_random_contrast}()} \cr
 \code{\link{layer_random_crop}()} \cr
+\code{\link{layer_random_elastic_transform}()} \cr
 \code{\link{layer_random_erasing}()} \cr
 \code{\link{layer_random_flip}()} \cr
 \code{\link{layer_random_gaussian_blur}()} \cr
diff --git a/man/layer_random_color_jitter.Rd b/man/layer_random_color_jitter.Rd
index a0406e149..d1ce8e95b 100644
--- a/man/layer_random_color_jitter.Rd
+++ b/man/layer_random_color_jitter.Rd
@@ -88,6 +88,7 @@ Other image preprocessing layers: \cr
 \code{\link{layer_mix_up}()} \cr
 \code{\link{layer_rand_augment}()} \cr
 \code{\link{layer_random_color_degeneration}()} \cr
+\code{\link{layer_random_elastic_transform}()} \cr
 \code{\link{layer_random_erasing}()} \cr
 \code{\link{layer_random_gaussian_blur}()} \cr
 \code{\link{layer_random_grayscale}()} \cr
@@ -123,6 +124,7 @@ Other preprocessing layers: \cr
 \code{\link{layer_random_color_degeneration}()} \cr
 \code{\link{layer_random_contrast}()} \cr
 \code{\link{layer_random_crop}()} \cr
+\code{\link{layer_random_elastic_transform}()} \cr
 \code{\link{layer_random_erasing}()} \cr
 \code{\link{layer_random_flip}()} \cr
 \code{\link{layer_random_gaussian_blur}()} \cr
@@ -230,6 +232,7 @@ Other layers: \cr
 \code{\link{layer_random_color_degeneration}()} \cr
 \code{\link{layer_random_contrast}()} \cr
 \code{\link{layer_random_crop}()} \cr
+\code{\link{layer_random_elastic_transform}()} \cr
 \code{\link{layer_random_erasing}()} \cr
 \code{\link{layer_random_flip}()} \cr
 \code{\link{layer_random_gaussian_blur}()} \cr
diff --git a/man/layer_random_contrast.Rd b/man/layer_random_contrast.Rd
index 8721412be..418e57a7c 100644
--- a/man/layer_random_contrast.Rd
+++ b/man/layer_random_contrast.Rd
@@ -103,6 +103,7 @@ Other preprocessing layers: \cr
 \code{\link{layer_random_color_degeneration}()} \cr
 \code{\link{layer_random_color_jitter}()} \cr
 \code{\link{layer_random_crop}()} \cr
+\code{\link{layer_random_elastic_transform}()} \cr
 \code{\link{layer_random_erasing}()} \cr
 \code{\link{layer_random_flip}()} \cr
 \code{\link{layer_random_gaussian_blur}()} \cr
@@ -210,6 +211,7 @@ Other layers: \cr
 \code{\link{layer_random_color_degeneration}()} \cr
 \code{\link{layer_random_color_jitter}()} \cr
 \code{\link{layer_random_crop}()} \cr
+\code{\link{layer_random_elastic_transform}()} \cr
 \code{\link{layer_random_erasing}()} \cr
 \code{\link{layer_random_flip}()} \cr
 \code{\link{layer_random_gaussian_blur}()} \cr
diff --git a/man/layer_random_crop.Rd b/man/layer_random_crop.Rd
index 12fee33b7..be59bb552 100644
--- a/man/layer_random_crop.Rd
+++ b/man/layer_random_crop.Rd
@@ -110,6 +110,7 @@ Other preprocessing layers: \cr
 \code{\link{layer_random_color_degeneration}()} \cr
 \code{\link{layer_random_color_jitter}()} \cr
 \code{\link{layer_random_contrast}()} \cr
+\code{\link{layer_random_elastic_transform}()} \cr
 \code{\link{layer_random_erasing}()} \cr
 \code{\link{layer_random_flip}()} \cr
 \code{\link{layer_random_gaussian_blur}()} \cr
@@ -217,6 +218,7 @@ Other layers: \cr
 \code{\link{layer_random_color_degeneration}()} \cr
 \code{\link{layer_random_color_jitter}()} \cr
 \code{\link{layer_random_contrast}()} \cr
+\code{\link{layer_random_elastic_transform}()} \cr
 \code{\link{layer_random_erasing}()} \cr
 \code{\link{layer_random_flip}()} \cr
 \code{\link{layer_random_gaussian_blur}()} \cr
diff --git a/man/layer_random_elastic_transform.Rd b/man/layer_random_elastic_transform.Rd
new file mode 100644
index 000000000..093963774
--- /dev/null
+++ b/man/layer_random_elastic_transform.Rd
@@ -0,0 +1,300 @@
+% Generated by roxygen2: do not edit by hand
+% Please edit documentation in R/layers-preprocessing.R
+\name{layer_random_elastic_transform}
+\alias{layer_random_elastic_transform}
+\title{A preprocessing layer that applies random elastic transformations.}
+\usage{
+layer_random_elastic_transform(
+  object,
+  factor = 1,
+  scale = 1,
+  interpolation = "bilinear",
+  fill_mode = "reflect",
+  fill_value = 0,
+  value_range = list(0L, 255L),
+  seed = NULL,
+  data_format = NULL,
+  ...
+)
+}
+\arguments{
+\item{object}{Object to compose the layer with. A tensor, array, or sequential model.}
+
+\item{factor}{A single float or a tuple of two floats.
+\code{factor} controls the probability of applying the transformation.
+\itemize{
+\item \code{factor = 0.0} ensures no transformation is applied.
+\item \code{factor = 1.0} means the transformation is always applied.
+\item If a tuple \verb{(min, max)} is provided, a probability value
+is sampled between \code{min} and \code{max} for each image.
+\item If a single float is provided, a probability is sampled
+between \code{0.0} and the given float.
+Default is \code{1.0}.
+}}
+
+\item{scale}{A float or a tuple of two floats defining the magnitude of
+the distortion applied.
+\itemize{
+\item If a tuple \verb{(min, max)} is provided, a random scale value is
+sampled within this range.
+\item If a single float is provided, a random scale value is sampled
+between \code{0.0} and the given float.
+Default is \code{1.0}.
+}}
+
+\item{interpolation}{Interpolation mode. Supported values: \code{"nearest"},
+\code{"bilinear"}.}
+
+\item{fill_mode}{Points outside the boundaries of the input are filled
+according to the given mode. Available methods are \code{"constant"},
+\code{"nearest"}, \code{"wrap"} and \code{"reflect"}. Defaults to \code{"reflect"}.
+\itemize{
+\item \code{"reflect"}: \verb{(d c b a | a b c d | d c b a)}
+The input is extended by reflecting about the edge of the last
+pixel.
+\item \code{"constant"}: \verb{(k k k k | a b c d | k k k k)}
+The input is extended by filling all values beyond
+the edge with the same constant value \code{k} specified by
+\code{fill_value}.
+\item \code{"wrap"}: \verb{(a b c d | a b c d | a b c d)}
+The input is extended by wrapping around to the opposite edge.
+\item \code{"nearest"}: \verb{(a a a a | a b c d | d d d d)}
+The input is extended by the nearest pixel.
+When using the torch backend, \code{"reflect"} is redirected to
+\code{"mirror"} because torch does not support \code{"reflect"}.
+The torch backend also does not support \code{"wrap"}.
+}}
+
+\item{fill_value}{A float representing the value to fill outside the boundaries when
+\code{fill_mode = "constant"}.}
+
+\item{value_range}{The range of values the incoming images will have.
+Represented as a two-number tuple written \verb{[low, high]}. This is
+typically either \verb{[0, 1]} or \verb{[0, 255]} depending on how your
+preprocessing pipeline is set up.}
+
+\item{seed}{Integer. Used to create a random seed.}
+
+\item{data_format}{string, either \code{"channels_last"} or \code{"channels_first"}.
+The ordering of the dimensions in the inputs. \code{"channels_last"}
+corresponds to inputs with shape \verb{(batch, height, width, channels)}
+while \code{"channels_first"} corresponds to inputs with shape
+\verb{(batch, channels, height, width)}. It defaults to the
+\code{image_data_format} value found in your Keras config file at
+\verb{~/.keras/keras.json}. If you never set it, then it will be
+\code{"channels_last"}.}
+
+\item{...}{For forward/backward compatability.}
+}
+\description{
+This layer distorts input images by applying elastic deformations,
+simulating a physically realistic transformation. The magnitude of the
+distortion is controlled by the \code{scale} parameter, while the \code{factor}
+determines the probability of applying the transformation.
+}
+\seealso{
+Other image preprocessing layers: \cr
+\code{\link{layer_aug_mix}()} \cr
+\code{\link{layer_auto_contrast}()} \cr
+\code{\link{layer_center_crop}()} \cr
+\code{\link{layer_cut_mix}()} \cr
+\code{\link{layer_equalization}()} \cr
+\code{\link{layer_max_num_bounding_boxes}()} \cr
+\code{\link{layer_mix_up}()} \cr
+\code{\link{layer_rand_augment}()} \cr
+\code{\link{layer_random_color_degeneration}()} \cr
+\code{\link{layer_random_color_jitter}()} \cr
+\code{\link{layer_random_erasing}()} \cr
+\code{\link{layer_random_gaussian_blur}()} \cr
+\code{\link{layer_random_grayscale}()} \cr
+\code{\link{layer_random_hue}()} \cr
+\code{\link{layer_random_invert}()} \cr
+\code{\link{layer_random_perspective}()} \cr
+\code{\link{layer_random_posterization}()} \cr
+\code{\link{layer_random_saturation}()} \cr
+\code{\link{layer_random_sharpness}()} \cr
+\code{\link{layer_random_shear}()} \cr
+\code{\link{layer_rescaling}()} \cr
+\code{\link{layer_resizing}()} \cr
+\code{\link{layer_solarization}()} \cr
+
+Other preprocessing layers: \cr
+\code{\link{layer_aug_mix}()} \cr
+\code{\link{layer_auto_contrast}()} \cr
+\code{\link{layer_category_encoding}()} \cr
+\code{\link{layer_center_crop}()} \cr
+\code{\link{layer_cut_mix}()} \cr
+\code{\link{layer_discretization}()} \cr
+\code{\link{layer_equalization}()} \cr
+\code{\link{layer_feature_space}()} \cr
+\code{\link{layer_hashed_crossing}()} \cr
+\code{\link{layer_hashing}()} \cr
+\code{\link{layer_integer_lookup}()} \cr
+\code{\link{layer_max_num_bounding_boxes}()} \cr
+\code{\link{layer_mel_spectrogram}()} \cr
+\code{\link{layer_mix_up}()} \cr
+\code{\link{layer_normalization}()} \cr
+\code{\link{layer_rand_augment}()} \cr
+\code{\link{layer_random_brightness}()} \cr
+\code{\link{layer_random_color_degeneration}()} \cr
+\code{\link{layer_random_color_jitter}()} \cr
+\code{\link{layer_random_contrast}()} \cr
+\code{\link{layer_random_crop}()} \cr
+\code{\link{layer_random_erasing}()} \cr
+\code{\link{layer_random_flip}()} \cr
+\code{\link{layer_random_gaussian_blur}()} \cr
+\code{\link{layer_random_grayscale}()} \cr
+\code{\link{layer_random_hue}()} \cr
+\code{\link{layer_random_invert}()} \cr
+\code{\link{layer_random_perspective}()} \cr
+\code{\link{layer_random_posterization}()} \cr
+\code{\link{layer_random_rotation}()} \cr
+\code{\link{layer_random_saturation}()} \cr
+\code{\link{layer_random_sharpness}()} \cr
+\code{\link{layer_random_shear}()} \cr
+\code{\link{layer_random_translation}()} \cr
+\code{\link{layer_random_zoom}()} \cr
+\code{\link{layer_rescaling}()} \cr
+\code{\link{layer_resizing}()} \cr
+\code{\link{layer_solarization}()} \cr
+\code{\link{layer_stft_spectrogram}()} \cr
+\code{\link{layer_string_lookup}()} \cr
+\code{\link{layer_text_vectorization}()} \cr
+
+Other layers: \cr
+\code{\link{Layer}()} \cr
+\code{\link{layer_activation}()} \cr
+\code{\link{layer_activation_elu}()} \cr
+\code{\link{layer_activation_leaky_relu}()} \cr
+\code{\link{layer_activation_parametric_relu}()} \cr
+\code{\link{layer_activation_relu}()} \cr
+\code{\link{layer_activation_softmax}()} \cr
+\code{\link{layer_activity_regularization}()} \cr
+\code{\link{layer_add}()} \cr
+\code{\link{layer_additive_attention}()} \cr
+\code{\link{layer_alpha_dropout}()} \cr
+\code{\link{layer_attention}()} \cr
+\code{\link{layer_aug_mix}()} \cr
+\code{\link{layer_auto_contrast}()} \cr
+\code{\link{layer_average}()} \cr
+\code{\link{layer_average_pooling_1d}()} \cr
+\code{\link{layer_average_pooling_2d}()} \cr
+\code{\link{layer_average_pooling_3d}()} \cr
+\code{\link{layer_batch_normalization}()} \cr
+\code{\link{layer_bidirectional}()} \cr
+\code{\link{layer_category_encoding}()} \cr
+\code{\link{layer_center_crop}()} \cr
+\code{\link{layer_concatenate}()} \cr
+\code{\link{layer_conv_1d}()} \cr
+\code{\link{layer_conv_1d_transpose}()} \cr
+\code{\link{layer_conv_2d}()} \cr
+\code{\link{layer_conv_2d_transpose}()} \cr
+\code{\link{layer_conv_3d}()} \cr
+\code{\link{layer_conv_3d_transpose}()} \cr
+\code{\link{layer_conv_lstm_1d}()} \cr
+\code{\link{layer_conv_lstm_2d}()} \cr
+\code{\link{layer_conv_lstm_3d}()} \cr
+\code{\link{layer_cropping_1d}()} \cr
+\code{\link{layer_cropping_2d}()} \cr
+\code{\link{layer_cropping_3d}()} \cr
+\code{\link{layer_cut_mix}()} \cr
+\code{\link{layer_dense}()} \cr
+\code{\link{layer_depthwise_conv_1d}()} \cr
+\code{\link{layer_depthwise_conv_2d}()} \cr
+\code{\link{layer_discretization}()} \cr
+\code{\link{layer_dot}()} \cr
+\code{\link{layer_dropout}()} \cr
+\code{\link{layer_einsum_dense}()} \cr
+\code{\link{layer_embedding}()} \cr
+\code{\link{layer_equalization}()} \cr
+\code{\link{layer_feature_space}()} \cr
+\code{\link{layer_flatten}()} \cr
+\code{\link{layer_flax_module_wrapper}()} \cr
+\code{\link{layer_gaussian_dropout}()} \cr
+\code{\link{layer_gaussian_noise}()} \cr
+\code{\link{layer_global_average_pooling_1d}()} \cr
+\code{\link{layer_global_average_pooling_2d}()} \cr
+\code{\link{layer_global_average_pooling_3d}()} \cr
+\code{\link{layer_global_max_pooling_1d}()} \cr
+\code{\link{layer_global_max_pooling_2d}()} \cr
+\code{\link{layer_global_max_pooling_3d}()} \cr
+\code{\link{layer_group_normalization}()} \cr
+\code{\link{layer_group_query_attention}()} \cr
+\code{\link{layer_gru}()} \cr
+\code{\link{layer_hashed_crossing}()} \cr
+\code{\link{layer_hashing}()} \cr
+\code{\link{layer_identity}()} \cr
+\code{\link{layer_integer_lookup}()} \cr
+\code{\link{layer_jax_model_wrapper}()} \cr
+\code{\link{layer_lambda}()} \cr
+\code{\link{layer_layer_normalization}()} \cr
+\code{\link{layer_lstm}()} \cr
+\code{\link{layer_masking}()} \cr
+\code{\link{layer_max_num_bounding_boxes}()} \cr
+\code{\link{layer_max_pooling_1d}()} \cr
+\code{\link{layer_max_pooling_2d}()} \cr
+\code{\link{layer_max_pooling_3d}()} \cr
+\code{\link{layer_maximum}()} \cr
+\code{\link{layer_mel_spectrogram}()} \cr
+\code{\link{layer_minimum}()} \cr
+\code{\link{layer_mix_up}()} \cr
+\code{\link{layer_multi_head_attention}()} \cr
+\code{\link{layer_multiply}()} \cr
+\code{\link{layer_normalization}()} \cr
+\code{\link{layer_permute}()} \cr
+\code{\link{layer_rand_augment}()} \cr
+\code{\link{layer_random_brightness}()} \cr
+\code{\link{layer_random_color_degeneration}()} \cr
+\code{\link{layer_random_color_jitter}()} \cr
+\code{\link{layer_random_contrast}()} \cr
+\code{\link{layer_random_crop}()} \cr
+\code{\link{layer_random_erasing}()} \cr
+\code{\link{layer_random_flip}()} \cr
+\code{\link{layer_random_gaussian_blur}()} \cr
+\code{\link{layer_random_grayscale}()} \cr
+\code{\link{layer_random_hue}()} \cr
+\code{\link{layer_random_invert}()} \cr
+\code{\link{layer_random_perspective}()} \cr
+\code{\link{layer_random_posterization}()} \cr
+\code{\link{layer_random_rotation}()} \cr
+\code{\link{layer_random_saturation}()} \cr
+\code{\link{layer_random_sharpness}()} \cr
+\code{\link{layer_random_shear}()} \cr
+\code{\link{layer_random_translation}()} \cr
+\code{\link{layer_random_zoom}()} \cr
+\code{\link{layer_repeat_vector}()} \cr
+\code{\link{layer_rescaling}()} \cr
+\code{\link{layer_reshape}()} \cr
+\code{\link{layer_resizing}()} \cr
+\code{\link{layer_rms_normalization}()} \cr
+\code{\link{layer_rnn}()} \cr
+\code{\link{layer_separable_conv_1d}()} \cr
+\code{\link{layer_separable_conv_2d}()} \cr
+\code{\link{layer_simple_rnn}()} \cr
+\code{\link{layer_solarization}()} \cr
+\code{\link{layer_spatial_dropout_1d}()} \cr
+\code{\link{layer_spatial_dropout_2d}()} \cr
+\code{\link{layer_spatial_dropout_3d}()} \cr
+\code{\link{layer_spectral_normalization}()} \cr
+\code{\link{layer_stft_spectrogram}()} \cr
+\code{\link{layer_string_lookup}()} \cr
+\code{\link{layer_subtract}()} \cr
+\code{\link{layer_text_vectorization}()} \cr
+\code{\link{layer_tfsm}()} \cr
+\code{\link{layer_time_distributed}()} \cr
+\code{\link{layer_torch_module_wrapper}()} \cr
+\code{\link{layer_unit_normalization}()} \cr
+\code{\link{layer_upsampling_1d}()} \cr
+\code{\link{layer_upsampling_2d}()} \cr
+\code{\link{layer_upsampling_3d}()} \cr
+\code{\link{layer_zero_padding_1d}()} \cr
+\code{\link{layer_zero_padding_2d}()} \cr
+\code{\link{layer_zero_padding_3d}()} \cr
+\code{\link{rnn_cell_gru}()} \cr
+\code{\link{rnn_cell_lstm}()} \cr
+\code{\link{rnn_cell_simple}()} \cr
+\code{\link{rnn_cells_stack}()} \cr
+}
+\concept{image preprocessing layers}
+\concept{layers}
+\concept{preprocessing layers}
diff --git a/man/layer_random_erasing.Rd b/man/layer_random_erasing.Rd
index 57429dc36..5cdf9e274 100644
--- a/man/layer_random_erasing.Rd
+++ b/man/layer_random_erasing.Rd
@@ -80,6 +80,7 @@ Other image preprocessing layers: \cr
 \code{\link{layer_rand_augment}()} \cr
 \code{\link{layer_random_color_degeneration}()} \cr
 \code{\link{layer_random_color_jitter}()} \cr
+\code{\link{layer_random_elastic_transform}()} \cr
 \code{\link{layer_random_gaussian_blur}()} \cr
 \code{\link{layer_random_grayscale}()} \cr
 \code{\link{layer_random_hue}()} \cr
@@ -115,6 +116,7 @@ Other preprocessing layers: \cr
 \code{\link{layer_random_color_jitter}()} \cr
 \code{\link{layer_random_contrast}()} \cr
 \code{\link{layer_random_crop}()} \cr
+\code{\link{layer_random_elastic_transform}()} \cr
 \code{\link{layer_random_flip}()} \cr
 \code{\link{layer_random_gaussian_blur}()} \cr
 \code{\link{layer_random_grayscale}()} \cr
@@ -222,6 +224,7 @@ Other layers: \cr
 \code{\link{layer_random_color_jitter}()} \cr
 \code{\link{layer_random_contrast}()} \cr
 \code{\link{layer_random_crop}()} \cr
+\code{\link{layer_random_elastic_transform}()} \cr
 \code{\link{layer_random_flip}()} \cr
 \code{\link{layer_random_gaussian_blur}()} \cr
 \code{\link{layer_random_grayscale}()} \cr
diff --git a/man/layer_random_flip.Rd b/man/layer_random_flip.Rd
index eeea38de5..6f82f5d71 100644
--- a/man/layer_random_flip.Rd
+++ b/man/layer_random_flip.Rd
@@ -101,6 +101,7 @@ Other preprocessing layers: \cr
 \code{\link{layer_random_color_jitter}()} \cr
 \code{\link{layer_random_contrast}()} \cr
 \code{\link{layer_random_crop}()} \cr
+\code{\link{layer_random_elastic_transform}()} \cr
 \code{\link{layer_random_erasing}()} \cr
 \code{\link{layer_random_gaussian_blur}()} \cr
 \code{\link{layer_random_grayscale}()} \cr
@@ -208,6 +209,7 @@ Other layers: \cr
 \code{\link{layer_random_color_jitter}()} \cr
 \code{\link{layer_random_contrast}()} \cr
 \code{\link{layer_random_crop}()} \cr
+\code{\link{layer_random_elastic_transform}()} \cr
 \code{\link{layer_random_erasing}()} \cr
 \code{\link{layer_random_gaussian_blur}()} \cr
 \code{\link{layer_random_grayscale}()} \cr
diff --git a/man/layer_random_gaussian_blur.Rd b/man/layer_random_gaussian_blur.Rd
index b74671d17..b9eb28a52 100644
--- a/man/layer_random_gaussian_blur.Rd
+++ b/man/layer_random_gaussian_blur.Rd
@@ -69,6 +69,7 @@ Other image preprocessing layers: \cr
 \code{\link{layer_rand_augment}()} \cr
 \code{\link{layer_random_color_degeneration}()} \cr
 \code{\link{layer_random_color_jitter}()} \cr
+\code{\link{layer_random_elastic_transform}()} \cr
 \code{\link{layer_random_erasing}()} \cr
 \code{\link{layer_random_grayscale}()} \cr
 \code{\link{layer_random_hue}()} \cr
@@ -104,6 +105,7 @@ Other preprocessing layers: \cr
 \code{\link{layer_random_color_jitter}()} \cr
 \code{\link{layer_random_contrast}()} \cr
 \code{\link{layer_random_crop}()} \cr
+\code{\link{layer_random_elastic_transform}()} \cr
 \code{\link{layer_random_erasing}()} \cr
 \code{\link{layer_random_flip}()} \cr
 \code{\link{layer_random_grayscale}()} \cr
@@ -211,6 +213,7 @@ Other layers: \cr
 \code{\link{layer_random_color_jitter}()} \cr
 \code{\link{layer_random_contrast}()} \cr
 \code{\link{layer_random_crop}()} \cr
+\code{\link{layer_random_elastic_transform}()} \cr
 \code{\link{layer_random_erasing}()} \cr
 \code{\link{layer_random_flip}()} \cr
 \code{\link{layer_random_grayscale}()} \cr
diff --git a/man/layer_random_grayscale.Rd b/man/layer_random_grayscale.Rd
index 19461384e..1a46e0f75 100644
--- a/man/layer_random_grayscale.Rd
+++ b/man/layer_random_grayscale.Rd
@@ -69,6 +69,7 @@ Other image preprocessing layers: \cr
 \code{\link{layer_rand_augment}()} \cr
 \code{\link{layer_random_color_degeneration}()} \cr
 \code{\link{layer_random_color_jitter}()} \cr
+\code{\link{layer_random_elastic_transform}()} \cr
 \code{\link{layer_random_erasing}()} \cr
 \code{\link{layer_random_gaussian_blur}()} \cr
 \code{\link{layer_random_hue}()} \cr
@@ -104,6 +105,7 @@ Other preprocessing layers: \cr
 \code{\link{layer_random_color_jitter}()} \cr
 \code{\link{layer_random_contrast}()} \cr
 \code{\link{layer_random_crop}()} \cr
+\code{\link{layer_random_elastic_transform}()} \cr
 \code{\link{layer_random_erasing}()} \cr
 \code{\link{layer_random_flip}()} \cr
 \code{\link{layer_random_gaussian_blur}()} \cr
@@ -211,6 +213,7 @@ Other layers: \cr
 \code{\link{layer_random_color_jitter}()} \cr
 \code{\link{layer_random_contrast}()} \cr
 \code{\link{layer_random_crop}()} \cr
+\code{\link{layer_random_elastic_transform}()} \cr
 \code{\link{layer_random_erasing}()} \cr
 \code{\link{layer_random_flip}()} \cr
 \code{\link{layer_random_gaussian_blur}()} \cr
diff --git a/man/layer_random_hue.Rd b/man/layer_random_hue.Rd
index 3e837761d..f56d0507b 100644
--- a/man/layer_random_hue.Rd
+++ b/man/layer_random_hue.Rd
@@ -69,6 +69,7 @@ Other image preprocessing layers: \cr
 \code{\link{layer_rand_augment}()} \cr
 \code{\link{layer_random_color_degeneration}()} \cr
 \code{\link{layer_random_color_jitter}()} \cr
+\code{\link{layer_random_elastic_transform}()} \cr
 \code{\link{layer_random_erasing}()} \cr
 \code{\link{layer_random_gaussian_blur}()} \cr
 \code{\link{layer_random_grayscale}()} \cr
@@ -104,6 +105,7 @@ Other preprocessing layers: \cr
 \code{\link{layer_random_color_jitter}()} \cr
 \code{\link{layer_random_contrast}()} \cr
 \code{\link{layer_random_crop}()} \cr
+\code{\link{layer_random_elastic_transform}()} \cr
 \code{\link{layer_random_erasing}()} \cr
 \code{\link{layer_random_flip}()} \cr
 \code{\link{layer_random_gaussian_blur}()} \cr
@@ -211,6 +213,7 @@ Other layers: \cr
 \code{\link{layer_random_color_jitter}()} \cr
 \code{\link{layer_random_contrast}()} \cr
 \code{\link{layer_random_crop}()} \cr
+\code{\link{layer_random_elastic_transform}()} \cr
 \code{\link{layer_random_erasing}()} \cr
 \code{\link{layer_random_flip}()} \cr
 \code{\link{layer_random_gaussian_blur}()} \cr
diff --git a/man/layer_random_invert.Rd b/man/layer_random_invert.Rd
index 26f75f85e..65a88262d 100644
--- a/man/layer_random_invert.Rd
+++ b/man/layer_random_invert.Rd
@@ -61,6 +61,7 @@ Other image preprocessing layers: \cr
 \code{\link{layer_rand_augment}()} \cr
 \code{\link{layer_random_color_degeneration}()} \cr
 \code{\link{layer_random_color_jitter}()} \cr
+\code{\link{layer_random_elastic_transform}()} \cr
 \code{\link{layer_random_erasing}()} \cr
 \code{\link{layer_random_gaussian_blur}()} \cr
 \code{\link{layer_random_grayscale}()} \cr
@@ -96,6 +97,7 @@ Other preprocessing layers: \cr
 \code{\link{layer_random_color_jitter}()} \cr
 \code{\link{layer_random_contrast}()} \cr
 \code{\link{layer_random_crop}()} \cr
+\code{\link{layer_random_elastic_transform}()} \cr
 \code{\link{layer_random_erasing}()} \cr
 \code{\link{layer_random_flip}()} \cr
 \code{\link{layer_random_gaussian_blur}()} \cr
@@ -203,6 +205,7 @@ Other layers: \cr
 \code{\link{layer_random_color_jitter}()} \cr
 \code{\link{layer_random_contrast}()} \cr
 \code{\link{layer_random_crop}()} \cr
+\code{\link{layer_random_elastic_transform}()} \cr
 \code{\link{layer_random_erasing}()} \cr
 \code{\link{layer_random_flip}()} \cr
 \code{\link{layer_random_gaussian_blur}()} \cr
diff --git a/man/layer_random_perspective.Rd b/man/layer_random_perspective.Rd
index e93e1d3b3..1f539f71d 100644
--- a/man/layer_random_perspective.Rd
+++ b/man/layer_random_perspective.Rd
@@ -71,6 +71,7 @@ Other image preprocessing layers: \cr
 \code{\link{layer_rand_augment}()} \cr
 \code{\link{layer_random_color_degeneration}()} \cr
 \code{\link{layer_random_color_jitter}()} \cr
+\code{\link{layer_random_elastic_transform}()} \cr
 \code{\link{layer_random_erasing}()} \cr
 \code{\link{layer_random_gaussian_blur}()} \cr
 \code{\link{layer_random_grayscale}()} \cr
@@ -106,6 +107,7 @@ Other preprocessing layers: \cr
 \code{\link{layer_random_color_jitter}()} \cr
 \code{\link{layer_random_contrast}()} \cr
 \code{\link{layer_random_crop}()} \cr
+\code{\link{layer_random_elastic_transform}()} \cr
 \code{\link{layer_random_erasing}()} \cr
 \code{\link{layer_random_flip}()} \cr
 \code{\link{layer_random_gaussian_blur}()} \cr
@@ -213,6 +215,7 @@ Other layers: \cr
 \code{\link{layer_random_color_jitter}()} \cr
 \code{\link{layer_random_contrast}()} \cr
 \code{\link{layer_random_crop}()} \cr
+\code{\link{layer_random_elastic_transform}()} \cr
 \code{\link{layer_random_erasing}()} \cr
 \code{\link{layer_random_flip}()} \cr
 \code{\link{layer_random_gaussian_blur}()} \cr
diff --git a/man/layer_random_posterization.Rd b/man/layer_random_posterization.Rd
index ec6d20673..3690d491c 100644
--- a/man/layer_random_posterization.Rd
+++ b/man/layer_random_posterization.Rd
@@ -57,6 +57,7 @@ Other image preprocessing layers: \cr
 \code{\link{layer_rand_augment}()} \cr
 \code{\link{layer_random_color_degeneration}()} \cr
 \code{\link{layer_random_color_jitter}()} \cr
+\code{\link{layer_random_elastic_transform}()} \cr
 \code{\link{layer_random_erasing}()} \cr
 \code{\link{layer_random_gaussian_blur}()} \cr
 \code{\link{layer_random_grayscale}()} \cr
@@ -92,6 +93,7 @@ Other preprocessing layers: \cr
 \code{\link{layer_random_color_jitter}()} \cr
 \code{\link{layer_random_contrast}()} \cr
 \code{\link{layer_random_crop}()} \cr
+\code{\link{layer_random_elastic_transform}()} \cr
 \code{\link{layer_random_erasing}()} \cr
 \code{\link{layer_random_flip}()} \cr
 \code{\link{layer_random_gaussian_blur}()} \cr
@@ -199,6 +201,7 @@ Other layers: \cr
 \code{\link{layer_random_color_jitter}()} \cr
 \code{\link{layer_random_contrast}()} \cr
 \code{\link{layer_random_crop}()} \cr
+\code{\link{layer_random_elastic_transform}()} \cr
 \code{\link{layer_random_erasing}()} \cr
 \code{\link{layer_random_flip}()} \cr
 \code{\link{layer_random_gaussian_blur}()} \cr
diff --git a/man/layer_random_rotation.Rd b/man/layer_random_rotation.Rd
index ca0dad261..fac5f19cd 100644
--- a/man/layer_random_rotation.Rd
+++ b/man/layer_random_rotation.Rd
@@ -139,6 +139,7 @@ Other preprocessing layers: \cr
 \code{\link{layer_random_color_jitter}()} \cr
 \code{\link{layer_random_contrast}()} \cr
 \code{\link{layer_random_crop}()} \cr
+\code{\link{layer_random_elastic_transform}()} \cr
 \code{\link{layer_random_erasing}()} \cr
 \code{\link{layer_random_flip}()} \cr
 \code{\link{layer_random_gaussian_blur}()} \cr
@@ -246,6 +247,7 @@ Other layers: \cr
 \code{\link{layer_random_color_jitter}()} \cr
 \code{\link{layer_random_contrast}()} \cr
 \code{\link{layer_random_crop}()} \cr
+\code{\link{layer_random_elastic_transform}()} \cr
 \code{\link{layer_random_erasing}()} \cr
 \code{\link{layer_random_flip}()} \cr
 \code{\link{layer_random_gaussian_blur}()} \cr
diff --git a/man/layer_random_saturation.Rd b/man/layer_random_saturation.Rd
index cbb66f946..ece35d0bf 100644
--- a/man/layer_random_saturation.Rd
+++ b/man/layer_random_saturation.Rd
@@ -67,6 +67,7 @@ Other image preprocessing layers: \cr
 \code{\link{layer_rand_augment}()} \cr
 \code{\link{layer_random_color_degeneration}()} \cr
 \code{\link{layer_random_color_jitter}()} \cr
+\code{\link{layer_random_elastic_transform}()} \cr
 \code{\link{layer_random_erasing}()} \cr
 \code{\link{layer_random_gaussian_blur}()} \cr
 \code{\link{layer_random_grayscale}()} \cr
@@ -102,6 +103,7 @@ Other preprocessing layers: \cr
 \code{\link{layer_random_color_jitter}()} \cr
 \code{\link{layer_random_contrast}()} \cr
 \code{\link{layer_random_crop}()} \cr
+\code{\link{layer_random_elastic_transform}()} \cr
 \code{\link{layer_random_erasing}()} \cr
 \code{\link{layer_random_flip}()} \cr
 \code{\link{layer_random_gaussian_blur}()} \cr
@@ -209,6 +211,7 @@ Other layers: \cr
 \code{\link{layer_random_color_jitter}()} \cr
 \code{\link{layer_random_contrast}()} \cr
 \code{\link{layer_random_crop}()} \cr
+\code{\link{layer_random_elastic_transform}()} \cr
 \code{\link{layer_random_erasing}()} \cr
 \code{\link{layer_random_flip}()} \cr
 \code{\link{layer_random_gaussian_blur}()} \cr
diff --git a/man/layer_random_sharpness.Rd b/man/layer_random_sharpness.Rd
index ad3b746fc..dbba021e5 100644
--- a/man/layer_random_sharpness.Rd
+++ b/man/layer_random_sharpness.Rd
@@ -60,6 +60,7 @@ Other image preprocessing layers: \cr
 \code{\link{layer_rand_augment}()} \cr
 \code{\link{layer_random_color_degeneration}()} \cr
 \code{\link{layer_random_color_jitter}()} \cr
+\code{\link{layer_random_elastic_transform}()} \cr
 \code{\link{layer_random_erasing}()} \cr
 \code{\link{layer_random_gaussian_blur}()} \cr
 \code{\link{layer_random_grayscale}()} \cr
@@ -95,6 +96,7 @@ Other preprocessing layers: \cr
 \code{\link{layer_random_color_jitter}()} \cr
 \code{\link{layer_random_contrast}()} \cr
 \code{\link{layer_random_crop}()} \cr
+\code{\link{layer_random_elastic_transform}()} \cr
 \code{\link{layer_random_erasing}()} \cr
 \code{\link{layer_random_flip}()} \cr
 \code{\link{layer_random_gaussian_blur}()} \cr
@@ -202,6 +204,7 @@ Other layers: \cr
 \code{\link{layer_random_color_jitter}()} \cr
 \code{\link{layer_random_contrast}()} \cr
 \code{\link{layer_random_crop}()} \cr
+\code{\link{layer_random_elastic_transform}()} \cr
 \code{\link{layer_random_erasing}()} \cr
 \code{\link{layer_random_flip}()} \cr
 \code{\link{layer_random_gaussian_blur}()} \cr
diff --git a/man/layer_random_shear.Rd b/man/layer_random_shear.Rd
index e1da25a01..1eb8a3a31 100644
--- a/man/layer_random_shear.Rd
+++ b/man/layer_random_shear.Rd
@@ -91,6 +91,7 @@ Other image preprocessing layers: \cr
 \code{\link{layer_rand_augment}()} \cr
 \code{\link{layer_random_color_degeneration}()} \cr
 \code{\link{layer_random_color_jitter}()} \cr
+\code{\link{layer_random_elastic_transform}()} \cr
 \code{\link{layer_random_erasing}()} \cr
 \code{\link{layer_random_gaussian_blur}()} \cr
 \code{\link{layer_random_grayscale}()} \cr
@@ -126,6 +127,7 @@ Other preprocessing layers: \cr
 \code{\link{layer_random_color_jitter}()} \cr
 \code{\link{layer_random_contrast}()} \cr
 \code{\link{layer_random_crop}()} \cr
+\code{\link{layer_random_elastic_transform}()} \cr
 \code{\link{layer_random_erasing}()} \cr
 \code{\link{layer_random_flip}()} \cr
 \code{\link{layer_random_gaussian_blur}()} \cr
@@ -233,6 +235,7 @@ Other layers: \cr
 \code{\link{layer_random_color_jitter}()} \cr
 \code{\link{layer_random_contrast}()} \cr
 \code{\link{layer_random_crop}()} \cr
+\code{\link{layer_random_elastic_transform}()} \cr
 \code{\link{layer_random_erasing}()} \cr
 \code{\link{layer_random_flip}()} \cr
 \code{\link{layer_random_gaussian_blur}()} \cr
diff --git a/man/layer_random_translation.Rd b/man/layer_random_translation.Rd
index 5a2bff3f4..5936d9bdb 100644
--- a/man/layer_random_translation.Rd
+++ b/man/layer_random_translation.Rd
@@ -148,6 +148,7 @@ Other preprocessing layers: \cr
 \code{\link{layer_random_color_jitter}()} \cr
 \code{\link{layer_random_contrast}()} \cr
 \code{\link{layer_random_crop}()} \cr
+\code{\link{layer_random_elastic_transform}()} \cr
 \code{\link{layer_random_erasing}()} \cr
 \code{\link{layer_random_flip}()} \cr
 \code{\link{layer_random_gaussian_blur}()} \cr
@@ -255,6 +256,7 @@ Other layers: \cr
 \code{\link{layer_random_color_jitter}()} \cr
 \code{\link{layer_random_contrast}()} \cr
 \code{\link{layer_random_crop}()} \cr
+\code{\link{layer_random_elastic_transform}()} \cr
 \code{\link{layer_random_erasing}()} \cr
 \code{\link{layer_random_flip}()} \cr
 \code{\link{layer_random_gaussian_blur}()} \cr
diff --git a/man/layer_random_zoom.Rd b/man/layer_random_zoom.Rd
index d3adbf565..9086f8e2f 100644
--- a/man/layer_random_zoom.Rd
+++ b/man/layer_random_zoom.Rd
@@ -155,6 +155,7 @@ Other preprocessing layers: \cr
 \code{\link{layer_random_color_jitter}()} \cr
 \code{\link{layer_random_contrast}()} \cr
 \code{\link{layer_random_crop}()} \cr
+\code{\link{layer_random_elastic_transform}()} \cr
 \code{\link{layer_random_erasing}()} \cr
 \code{\link{layer_random_flip}()} \cr
 \code{\link{layer_random_gaussian_blur}()} \cr
@@ -262,6 +263,7 @@ Other layers: \cr
 \code{\link{layer_random_color_jitter}()} \cr
 \code{\link{layer_random_contrast}()} \cr
 \code{\link{layer_random_crop}()} \cr
+\code{\link{layer_random_elastic_transform}()} \cr
 \code{\link{layer_random_erasing}()} \cr
 \code{\link{layer_random_flip}()} \cr
 \code{\link{layer_random_gaussian_blur}()} \cr
diff --git a/man/layer_repeat_vector.Rd b/man/layer_repeat_vector.Rd
index dadff70b0..ac36e97e3 100644
--- a/man/layer_repeat_vector.Rd
+++ b/man/layer_repeat_vector.Rd
@@ -152,6 +152,7 @@ Other layers: \cr
 \code{\link{layer_random_color_jitter}()} \cr
 \code{\link{layer_random_contrast}()} \cr
 \code{\link{layer_random_crop}()} \cr
+\code{\link{layer_random_elastic_transform}()} \cr
 \code{\link{layer_random_erasing}()} \cr
 \code{\link{layer_random_flip}()} \cr
 \code{\link{layer_random_gaussian_blur}()} \cr
diff --git a/man/layer_rescaling.Rd b/man/layer_rescaling.Rd
index d41c9b83e..003f0425c 100644
--- a/man/layer_rescaling.Rd
+++ b/man/layer_rescaling.Rd
@@ -65,6 +65,7 @@ Other image preprocessing layers: \cr
 \code{\link{layer_rand_augment}()} \cr
 \code{\link{layer_random_color_degeneration}()} \cr
 \code{\link{layer_random_color_jitter}()} \cr
+\code{\link{layer_random_elastic_transform}()} \cr
 \code{\link{layer_random_erasing}()} \cr
 \code{\link{layer_random_gaussian_blur}()} \cr
 \code{\link{layer_random_grayscale}()} \cr
@@ -100,6 +101,7 @@ Other preprocessing layers: \cr
 \code{\link{layer_random_color_jitter}()} \cr
 \code{\link{layer_random_contrast}()} \cr
 \code{\link{layer_random_crop}()} \cr
+\code{\link{layer_random_elastic_transform}()} \cr
 \code{\link{layer_random_erasing}()} \cr
 \code{\link{layer_random_flip}()} \cr
 \code{\link{layer_random_gaussian_blur}()} \cr
@@ -207,6 +209,7 @@ Other layers: \cr
 \code{\link{layer_random_color_jitter}()} \cr
 \code{\link{layer_random_contrast}()} \cr
 \code{\link{layer_random_crop}()} \cr
+\code{\link{layer_random_elastic_transform}()} \cr
 \code{\link{layer_random_erasing}()} \cr
 \code{\link{layer_random_flip}()} \cr
 \code{\link{layer_random_gaussian_blur}()} \cr
diff --git a/man/layer_reshape.Rd b/man/layer_reshape.Rd
index 6340285e8..20961c6d2 100644
--- a/man/layer_reshape.Rd
+++ b/man/layer_reshape.Rd
@@ -165,6 +165,7 @@ Other layers: \cr
 \code{\link{layer_random_color_jitter}()} \cr
 \code{\link{layer_random_contrast}()} \cr
 \code{\link{layer_random_crop}()} \cr
+\code{\link{layer_random_elastic_transform}()} \cr
 \code{\link{layer_random_erasing}()} \cr
 \code{\link{layer_random_flip}()} \cr
 \code{\link{layer_random_gaussian_blur}()} \cr
diff --git a/man/layer_resizing.Rd b/man/layer_resizing.Rd
index a9ddba757..a74f0389d 100644
--- a/man/layer_resizing.Rd
+++ b/man/layer_resizing.Rd
@@ -111,6 +111,7 @@ Other image preprocessing layers: \cr
 \code{\link{layer_rand_augment}()} \cr
 \code{\link{layer_random_color_degeneration}()} \cr
 \code{\link{layer_random_color_jitter}()} \cr
+\code{\link{layer_random_elastic_transform}()} \cr
 \code{\link{layer_random_erasing}()} \cr
 \code{\link{layer_random_gaussian_blur}()} \cr
 \code{\link{layer_random_grayscale}()} \cr
@@ -146,6 +147,7 @@ Other preprocessing layers: \cr
 \code{\link{layer_random_color_jitter}()} \cr
 \code{\link{layer_random_contrast}()} \cr
 \code{\link{layer_random_crop}()} \cr
+\code{\link{layer_random_elastic_transform}()} \cr
 \code{\link{layer_random_erasing}()} \cr
 \code{\link{layer_random_flip}()} \cr
 \code{\link{layer_random_gaussian_blur}()} \cr
@@ -253,6 +255,7 @@ Other layers: \cr
 \code{\link{layer_random_color_jitter}()} \cr
 \code{\link{layer_random_contrast}()} \cr
 \code{\link{layer_random_crop}()} \cr
+\code{\link{layer_random_elastic_transform}()} \cr
 \code{\link{layer_random_erasing}()} \cr
 \code{\link{layer_random_flip}()} \cr
 \code{\link{layer_random_gaussian_blur}()} \cr
diff --git a/man/layer_rms_normalization.Rd b/man/layer_rms_normalization.Rd
index c7e7b9499..e6fef24e0 100644
--- a/man/layer_rms_normalization.Rd
+++ b/man/layer_rms_normalization.Rd
@@ -140,6 +140,7 @@ Other layers: \cr
 \code{\link{layer_random_color_jitter}()} \cr
 \code{\link{layer_random_contrast}()} \cr
 \code{\link{layer_random_crop}()} \cr
+\code{\link{layer_random_elastic_transform}()} \cr
 \code{\link{layer_random_erasing}()} \cr
 \code{\link{layer_random_flip}()} \cr
 \code{\link{layer_random_gaussian_blur}()} \cr
diff --git a/man/layer_rnn.Rd b/man/layer_rnn.Rd
index 1bd25ee44..e198966df 100644
--- a/man/layer_rnn.Rd
+++ b/man/layer_rnn.Rd
@@ -321,6 +321,7 @@ Other layers: \cr
 \code{\link{layer_random_color_jitter}()} \cr
 \code{\link{layer_random_contrast}()} \cr
 \code{\link{layer_random_crop}()} \cr
+\code{\link{layer_random_elastic_transform}()} \cr
 \code{\link{layer_random_erasing}()} \cr
 \code{\link{layer_random_flip}()} \cr
 \code{\link{layer_random_gaussian_blur}()} \cr
diff --git a/man/layer_separable_conv_1d.Rd b/man/layer_separable_conv_1d.Rd
index bacb0be27..0593dabf9 100644
--- a/man/layer_separable_conv_1d.Rd
+++ b/man/layer_separable_conv_1d.Rd
@@ -245,6 +245,7 @@ Other layers: \cr
 \code{\link{layer_random_color_jitter}()} \cr
 \code{\link{layer_random_contrast}()} \cr
 \code{\link{layer_random_crop}()} \cr
+\code{\link{layer_random_elastic_transform}()} \cr
 \code{\link{layer_random_erasing}()} \cr
 \code{\link{layer_random_flip}()} \cr
 \code{\link{layer_random_gaussian_blur}()} \cr
diff --git a/man/layer_separable_conv_2d.Rd b/man/layer_separable_conv_2d.Rd
index c70929184..317b450f1 100644
--- a/man/layer_separable_conv_2d.Rd
+++ b/man/layer_separable_conv_2d.Rd
@@ -246,6 +246,7 @@ Other layers: \cr
 \code{\link{layer_random_color_jitter}()} \cr
 \code{\link{layer_random_contrast}()} \cr
 \code{\link{layer_random_crop}()} \cr
+\code{\link{layer_random_elastic_transform}()} \cr
 \code{\link{layer_random_erasing}()} \cr
 \code{\link{layer_random_flip}()} \cr
 \code{\link{layer_random_gaussian_blur}()} \cr
diff --git a/man/layer_simple_rnn.Rd b/man/layer_simple_rnn.Rd
index db504506b..ec1233b3d 100644
--- a/man/layer_simple_rnn.Rd
+++ b/man/layer_simple_rnn.Rd
@@ -260,6 +260,7 @@ Other layers: \cr
 \code{\link{layer_random_color_jitter}()} \cr
 \code{\link{layer_random_contrast}()} \cr
 \code{\link{layer_random_crop}()} \cr
+\code{\link{layer_random_elastic_transform}()} \cr
 \code{\link{layer_random_erasing}()} \cr
 \code{\link{layer_random_flip}()} \cr
 \code{\link{layer_random_gaussian_blur}()} \cr
diff --git a/man/layer_solarization.Rd b/man/layer_solarization.Rd
index aacfb0d67..1750c5630 100644
--- a/man/layer_solarization.Rd
+++ b/man/layer_solarization.Rd
@@ -85,6 +85,7 @@ Other image preprocessing layers: \cr
 \code{\link{layer_rand_augment}()} \cr
 \code{\link{layer_random_color_degeneration}()} \cr
 \code{\link{layer_random_color_jitter}()} \cr
+\code{\link{layer_random_elastic_transform}()} \cr
 \code{\link{layer_random_erasing}()} \cr
 \code{\link{layer_random_gaussian_blur}()} \cr
 \code{\link{layer_random_grayscale}()} \cr
@@ -120,6 +121,7 @@ Other preprocessing layers: \cr
 \code{\link{layer_random_color_jitter}()} \cr
 \code{\link{layer_random_contrast}()} \cr
 \code{\link{layer_random_crop}()} \cr
+\code{\link{layer_random_elastic_transform}()} \cr
 \code{\link{layer_random_erasing}()} \cr
 \code{\link{layer_random_flip}()} \cr
 \code{\link{layer_random_gaussian_blur}()} \cr
@@ -227,6 +229,7 @@ Other layers: \cr
 \code{\link{layer_random_color_jitter}()} \cr
 \code{\link{layer_random_contrast}()} \cr
 \code{\link{layer_random_crop}()} \cr
+\code{\link{layer_random_elastic_transform}()} \cr
 \code{\link{layer_random_erasing}()} \cr
 \code{\link{layer_random_flip}()} \cr
 \code{\link{layer_random_gaussian_blur}()} \cr
diff --git a/man/layer_spatial_dropout_1d.Rd b/man/layer_spatial_dropout_1d.Rd
index a4bee7a5b..cea0ad451 100644
--- a/man/layer_spatial_dropout_1d.Rd
+++ b/man/layer_spatial_dropout_1d.Rd
@@ -165,6 +165,7 @@ Other layers: \cr
 \code{\link{layer_random_color_jitter}()} \cr
 \code{\link{layer_random_contrast}()} \cr
 \code{\link{layer_random_crop}()} \cr
+\code{\link{layer_random_elastic_transform}()} \cr
 \code{\link{layer_random_erasing}()} \cr
 \code{\link{layer_random_flip}()} \cr
 \code{\link{layer_random_gaussian_blur}()} \cr
diff --git a/man/layer_spatial_dropout_2d.Rd b/man/layer_spatial_dropout_2d.Rd
index ec8dc3846..06536fedd 100644
--- a/man/layer_spatial_dropout_2d.Rd
+++ b/man/layer_spatial_dropout_2d.Rd
@@ -182,6 +182,7 @@ Other layers: \cr
 \code{\link{layer_random_color_jitter}()} \cr
 \code{\link{layer_random_contrast}()} \cr
 \code{\link{layer_random_crop}()} \cr
+\code{\link{layer_random_elastic_transform}()} \cr
 \code{\link{layer_random_erasing}()} \cr
 \code{\link{layer_random_flip}()} \cr
 \code{\link{layer_random_gaussian_blur}()} \cr
diff --git a/man/layer_spatial_dropout_3d.Rd b/man/layer_spatial_dropout_3d.Rd
index 233c11df0..8b60c0465 100644
--- a/man/layer_spatial_dropout_3d.Rd
+++ b/man/layer_spatial_dropout_3d.Rd
@@ -182,6 +182,7 @@ Other layers: \cr
 \code{\link{layer_random_color_jitter}()} \cr
 \code{\link{layer_random_contrast}()} \cr
 \code{\link{layer_random_crop}()} \cr
+\code{\link{layer_random_elastic_transform}()} \cr
 \code{\link{layer_random_erasing}()} \cr
 \code{\link{layer_random_flip}()} \cr
 \code{\link{layer_random_gaussian_blur}()} \cr
diff --git a/man/layer_spectral_normalization.Rd b/man/layer_spectral_normalization.Rd
index 3122047e5..a89fd946c 100644
--- a/man/layer_spectral_normalization.Rd
+++ b/man/layer_spectral_normalization.Rd
@@ -161,6 +161,7 @@ Other layers: \cr
 \code{\link{layer_random_color_jitter}()} \cr
 \code{\link{layer_random_contrast}()} \cr
 \code{\link{layer_random_crop}()} \cr
+\code{\link{layer_random_elastic_transform}()} \cr
 \code{\link{layer_random_erasing}()} \cr
 \code{\link{layer_random_flip}()} \cr
 \code{\link{layer_random_gaussian_blur}()} \cr
diff --git a/man/layer_stft_spectrogram.Rd b/man/layer_stft_spectrogram.Rd
index d1e202eac..6e9548af9 100644
--- a/man/layer_stft_spectrogram.Rd
+++ b/man/layer_stft_spectrogram.Rd
@@ -203,6 +203,7 @@ Other preprocessing layers: \cr
 \code{\link{layer_random_color_jitter}()} \cr
 \code{\link{layer_random_contrast}()} \cr
 \code{\link{layer_random_crop}()} \cr
+\code{\link{layer_random_elastic_transform}()} \cr
 \code{\link{layer_random_erasing}()} \cr
 \code{\link{layer_random_flip}()} \cr
 \code{\link{layer_random_gaussian_blur}()} \cr
@@ -310,6 +311,7 @@ Other layers: \cr
 \code{\link{layer_random_color_jitter}()} \cr
 \code{\link{layer_random_contrast}()} \cr
 \code{\link{layer_random_crop}()} \cr
+\code{\link{layer_random_elastic_transform}()} \cr
 \code{\link{layer_random_erasing}()} \cr
 \code{\link{layer_random_flip}()} \cr
 \code{\link{layer_random_gaussian_blur}()} \cr
diff --git a/man/layer_string_lookup.Rd b/man/layer_string_lookup.Rd
index 9499cf0cd..1c9171c6a 100644
--- a/man/layer_string_lookup.Rd
+++ b/man/layer_string_lookup.Rd
@@ -424,6 +424,7 @@ Other preprocessing layers: \cr
 \code{\link{layer_random_color_jitter}()} \cr
 \code{\link{layer_random_contrast}()} \cr
 \code{\link{layer_random_crop}()} \cr
+\code{\link{layer_random_elastic_transform}()} \cr
 \code{\link{layer_random_erasing}()} \cr
 \code{\link{layer_random_flip}()} \cr
 \code{\link{layer_random_gaussian_blur}()} \cr
@@ -531,6 +532,7 @@ Other layers: \cr
 \code{\link{layer_random_color_jitter}()} \cr
 \code{\link{layer_random_contrast}()} \cr
 \code{\link{layer_random_crop}()} \cr
+\code{\link{layer_random_elastic_transform}()} \cr
 \code{\link{layer_random_erasing}()} \cr
 \code{\link{layer_random_flip}()} \cr
 \code{\link{layer_random_gaussian_blur}()} \cr
diff --git a/man/layer_subtract.Rd b/man/layer_subtract.Rd
index 787418628..591b79d0a 100644
--- a/man/layer_subtract.Rd
+++ b/man/layer_subtract.Rd
@@ -147,6 +147,7 @@ Other layers: \cr
 \code{\link{layer_random_color_jitter}()} \cr
 \code{\link{layer_random_contrast}()} \cr
 \code{\link{layer_random_crop}()} \cr
+\code{\link{layer_random_elastic_transform}()} \cr
 \code{\link{layer_random_erasing}()} \cr
 \code{\link{layer_random_flip}()} \cr
 \code{\link{layer_random_gaussian_blur}()} \cr
diff --git a/man/layer_text_vectorization.Rd b/man/layer_text_vectorization.Rd
index 754393cd1..ed8e92f08 100644
--- a/man/layer_text_vectorization.Rd
+++ b/man/layer_text_vectorization.Rd
@@ -293,6 +293,7 @@ Other preprocessing layers: \cr
 \code{\link{layer_random_color_jitter}()} \cr
 \code{\link{layer_random_contrast}()} \cr
 \code{\link{layer_random_crop}()} \cr
+\code{\link{layer_random_elastic_transform}()} \cr
 \code{\link{layer_random_erasing}()} \cr
 \code{\link{layer_random_flip}()} \cr
 \code{\link{layer_random_gaussian_blur}()} \cr
@@ -400,6 +401,7 @@ Other layers: \cr
 \code{\link{layer_random_color_jitter}()} \cr
 \code{\link{layer_random_contrast}()} \cr
 \code{\link{layer_random_crop}()} \cr
+\code{\link{layer_random_elastic_transform}()} \cr
 \code{\link{layer_random_erasing}()} \cr
 \code{\link{layer_random_flip}()} \cr
 \code{\link{layer_random_gaussian_blur}()} \cr
diff --git a/man/layer_tfsm.Rd b/man/layer_tfsm.Rd
index 03c090911..a27ae0273 100644
--- a/man/layer_tfsm.Rd
+++ b/man/layer_tfsm.Rd
@@ -59,8 +59,8 @@ model |> export_savedmodel("path/to/artifact")
 ## Output Type:
 ##   TensorSpec(shape=(None, 10), dtype=tf.float32, name=None)
 ## Captures:
-##   123789562403856: TensorSpec(shape=(), dtype=tf.resource, name=None)
-##   123789562409424: TensorSpec(shape=(), dtype=tf.resource, name=None)
+##   138000861911120: TensorSpec(shape=(), dtype=tf.resource, name=None)
+##   138000861919184: TensorSpec(shape=(), dtype=tf.resource, name=None)
 
 }\if{html}{\out{</div>}}
 
@@ -179,6 +179,7 @@ Other layers: \cr
 \code{\link{layer_random_color_jitter}()} \cr
 \code{\link{layer_random_contrast}()} \cr
 \code{\link{layer_random_crop}()} \cr
+\code{\link{layer_random_elastic_transform}()} \cr
 \code{\link{layer_random_erasing}()} \cr
 \code{\link{layer_random_flip}()} \cr
 \code{\link{layer_random_gaussian_blur}()} \cr
diff --git a/man/layer_time_distributed.Rd b/man/layer_time_distributed.Rd
index 268cefc92..5b8d4a226 100644
--- a/man/layer_time_distributed.Rd
+++ b/man/layer_time_distributed.Rd
@@ -167,6 +167,7 @@ Other layers: \cr
 \code{\link{layer_random_color_jitter}()} \cr
 \code{\link{layer_random_contrast}()} \cr
 \code{\link{layer_random_crop}()} \cr
+\code{\link{layer_random_elastic_transform}()} \cr
 \code{\link{layer_random_erasing}()} \cr
 \code{\link{layer_random_flip}()} \cr
 \code{\link{layer_random_gaussian_blur}()} \cr
diff --git a/man/layer_torch_module_wrapper.Rd b/man/layer_torch_module_wrapper.Rd
index 0be44bc55..f5b4ab134 100644
--- a/man/layer_torch_module_wrapper.Rd
+++ b/man/layer_torch_module_wrapper.Rd
@@ -197,6 +197,7 @@ Other layers: \cr
 \code{\link{layer_random_color_jitter}()} \cr
 \code{\link{layer_random_contrast}()} \cr
 \code{\link{layer_random_crop}()} \cr
+\code{\link{layer_random_elastic_transform}()} \cr
 \code{\link{layer_random_erasing}()} \cr
 \code{\link{layer_random_flip}()} \cr
 \code{\link{layer_random_gaussian_blur}()} \cr
diff --git a/man/layer_unit_normalization.Rd b/man/layer_unit_normalization.Rd
index 2fca58f16..a006fa6af 100644
--- a/man/layer_unit_normalization.Rd
+++ b/man/layer_unit_normalization.Rd
@@ -141,6 +141,7 @@ Other layers: \cr
 \code{\link{layer_random_color_jitter}()} \cr
 \code{\link{layer_random_contrast}()} \cr
 \code{\link{layer_random_crop}()} \cr
+\code{\link{layer_random_elastic_transform}()} \cr
 \code{\link{layer_random_erasing}()} \cr
 \code{\link{layer_random_flip}()} \cr
 \code{\link{layer_random_gaussian_blur}()} \cr
diff --git a/man/layer_upsampling_1d.Rd b/man/layer_upsampling_1d.Rd
index cf0d77e3b..cf4d76d98 100644
--- a/man/layer_upsampling_1d.Rd
+++ b/man/layer_upsampling_1d.Rd
@@ -174,6 +174,7 @@ Other layers: \cr
 \code{\link{layer_random_color_jitter}()} \cr
 \code{\link{layer_random_contrast}()} \cr
 \code{\link{layer_random_crop}()} \cr
+\code{\link{layer_random_elastic_transform}()} \cr
 \code{\link{layer_random_erasing}()} \cr
 \code{\link{layer_random_flip}()} \cr
 \code{\link{layer_random_gaussian_blur}()} \cr
diff --git a/man/layer_upsampling_2d.Rd b/man/layer_upsampling_2d.Rd
index a3f49e7f1..c6be5ae55 100644
--- a/man/layer_upsampling_2d.Rd
+++ b/man/layer_upsampling_2d.Rd
@@ -216,6 +216,7 @@ Other layers: \cr
 \code{\link{layer_random_color_jitter}()} \cr
 \code{\link{layer_random_contrast}()} \cr
 \code{\link{layer_random_crop}()} \cr
+\code{\link{layer_random_elastic_transform}()} \cr
 \code{\link{layer_random_erasing}()} \cr
 \code{\link{layer_random_flip}()} \cr
 \code{\link{layer_random_gaussian_blur}()} \cr
diff --git a/man/layer_upsampling_3d.Rd b/man/layer_upsampling_3d.Rd
index fd89bdc48..bc9f6f417 100644
--- a/man/layer_upsampling_3d.Rd
+++ b/man/layer_upsampling_3d.Rd
@@ -179,6 +179,7 @@ Other layers: \cr
 \code{\link{layer_random_color_jitter}()} \cr
 \code{\link{layer_random_contrast}()} \cr
 \code{\link{layer_random_crop}()} \cr
+\code{\link{layer_random_elastic_transform}()} \cr
 \code{\link{layer_random_erasing}()} \cr
 \code{\link{layer_random_flip}()} \cr
 \code{\link{layer_random_gaussian_blur}()} \cr
diff --git a/man/layer_zero_padding_1d.Rd b/man/layer_zero_padding_1d.Rd
index 57c569046..4a0742af0 100644
--- a/man/layer_zero_padding_1d.Rd
+++ b/man/layer_zero_padding_1d.Rd
@@ -206,6 +206,7 @@ Other layers: \cr
 \code{\link{layer_random_color_jitter}()} \cr
 \code{\link{layer_random_contrast}()} \cr
 \code{\link{layer_random_crop}()} \cr
+\code{\link{layer_random_elastic_transform}()} \cr
 \code{\link{layer_random_erasing}()} \cr
 \code{\link{layer_random_flip}()} \cr
 \code{\link{layer_random_gaussian_blur}()} \cr
diff --git a/man/layer_zero_padding_2d.Rd b/man/layer_zero_padding_2d.Rd
index 27e672949..50f30c0b0 100644
--- a/man/layer_zero_padding_2d.Rd
+++ b/man/layer_zero_padding_2d.Rd
@@ -207,6 +207,7 @@ Other layers: \cr
 \code{\link{layer_random_color_jitter}()} \cr
 \code{\link{layer_random_contrast}()} \cr
 \code{\link{layer_random_crop}()} \cr
+\code{\link{layer_random_elastic_transform}()} \cr
 \code{\link{layer_random_erasing}()} \cr
 \code{\link{layer_random_flip}()} \cr
 \code{\link{layer_random_gaussian_blur}()} \cr
diff --git a/man/layer_zero_padding_3d.Rd b/man/layer_zero_padding_3d.Rd
index 5bb918f59..d18f8a41a 100644
--- a/man/layer_zero_padding_3d.Rd
+++ b/man/layer_zero_padding_3d.Rd
@@ -201,6 +201,7 @@ Other layers: \cr
 \code{\link{layer_random_color_jitter}()} \cr
 \code{\link{layer_random_contrast}()} \cr
 \code{\link{layer_random_crop}()} \cr
+\code{\link{layer_random_elastic_transform}()} \cr
 \code{\link{layer_random_erasing}()} \cr
 \code{\link{layer_random_flip}()} \cr
 \code{\link{layer_random_gaussian_blur}()} \cr
diff --git a/man/op_image_elastic_transform.Rd b/man/op_image_elastic_transform.Rd
new file mode 100644
index 000000000..eab1535ca
--- /dev/null
+++ b/man/op_image_elastic_transform.Rd
@@ -0,0 +1,97 @@
+% Generated by roxygen2: do not edit by hand
+% Please edit documentation in R/ops-image.R
+\name{op_image_elastic_transform}
+\alias{op_image_elastic_transform}
+\title{Applies elastic deformation to the image(s).}
+\usage{
+op_image_elastic_transform(
+  images,
+  alpha = 20,
+  sigma = 5,
+  interpolation = "bilinear",
+  fill_mode = "reflect",
+  fill_value = 0,
+  seed = NULL,
+  data_format = NULL
+)
+}
+\arguments{
+\item{images}{Input image or batch of images. Must be 3D or 4D.}
+
+\item{alpha}{Scaling factor that controls the intensity of the deformation.}
+
+\item{sigma}{Standard deviation of the Gaussian filter used for
+smoothing the displacement fields.}
+
+\item{interpolation}{Interpolation method. Available methods are \code{"nearest"},
+and \code{"bilinear"}. Defaults to \code{"bilinear"}.}
+
+\item{fill_mode}{Points outside the boundaries of the input are filled
+according to the given mode. Available methods are \code{"constant"},
+\code{"nearest"}, \code{"wrap"} and \code{"reflect"}. Defaults to \code{"reflect"}.
+\itemize{
+\item \code{"reflect"}: \verb{(d c b a | a b c d | d c b a)}
+The input is extended by reflecting about the edge of the last
+pixel.
+\item \code{"constant"}: \verb{(k k k k | a b c d | k k k k)}
+The input is extended by filling all values beyond
+the edge with the same constant value \code{k} specified by
+\code{fill_value}.
+\item \code{"wrap"}: \verb{(a b c d | a b c d | a b c d)}
+The input is extended by wrapping around to the opposite edge.
+\item \code{"nearest"}: \verb{(a a a a | a b c d | d d d d)}
+The input is extended by the nearest pixel.
+}}
+
+\item{fill_value}{Value used for points outside the boundaries of the input if
+\code{fill_mode="constant"}. Defaults to \code{0}.}
+
+\item{seed}{Optional integer seed for the random number generator.}
+
+\item{data_format}{A string specifying the data format of the input tensor.
+It can be either \code{"channels_last"} or \code{"channels_first"}.
+\code{"channels_last"} corresponds to inputs with shape
+\verb{(batch, height, width, channels)}, while \code{"channels_first"}
+corresponds to inputs with shape \verb{(batch, channels, height, width)}.
+If not specified, the value will default to
+\code{keras.config.image_data_format}.}
+}
+\value{
+Transformed image or batch of images with elastic deformation.
+}
+\description{
+Apply random elastic deformation to 3D or 4D image tensors.
+}
+\section{Examples}{
+\if{html}{\out{<div class="sourceCode r">}}\preformatted{x <- random_uniform(c(2, 64, 80, 3))  # batch of 2 RGB images
+y <- op_image_elastic_transform(x)
+op_shape(y)
+}\if{html}{\out{</div>}}
+
+\if{html}{\out{<div class="sourceCode">}}\preformatted{## shape(2, 64, 80, 3)
+
+}\if{html}{\out{</div>}}
+
+\if{html}{\out{<div class="sourceCode r">}}\preformatted{x <- random_uniform(c(64, 80, 3))  # single RGB image
+y <- op_image_elastic_transform(x)
+op_shape(y)
+}\if{html}{\out{</div>}}
+
+\if{html}{\out{<div class="sourceCode">}}\preformatted{## shape(64, 80, 3)
+
+}\if{html}{\out{</div>}}
+
+\if{html}{\out{<div class="sourceCode r">}}\preformatted{x <- random_uniform(c(2, 3, 64, 80))  # batch of 2 RGB images
+y <- op_image_elastic_transform(
+  x,
+  data_format = "channels_first",
+  seed = 123
+)
+op_shape(y)
+}\if{html}{\out{</div>}}
+
+\if{html}{\out{<div class="sourceCode">}}\preformatted{## shape(2, 3, 64, 80)
+
+}\if{html}{\out{</div>}}
+}
+
diff --git a/man/op_searchsorted.Rd b/man/op_searchsorted.Rd
index 7d0f9cf38..e92933767 100644
--- a/man/op_searchsorted.Rd
+++ b/man/op_searchsorted.Rd
@@ -2,30 +2,36 @@
 % Please edit documentation in R/ops.R
 \name{op_searchsorted}
 \alias{op_searchsorted}
-\title{Perform a binary search}
+\title{Perform a binary search.}
 \usage{
 op_searchsorted(sorted_sequence, values, side = "left", zero_indexed = FALSE)
 }
 \arguments{
-\item{sorted_sequence}{1-D input tensor, sorted along the innermost
-dimension.}
+\item{sorted_sequence}{1-D input tensor that is sorted along its innermost dimension.}
 
 \item{values}{N-D tensor of query insertion values.}
 
-\item{side}{\code{'left'} or \code{'right'}, specifying the direction in which to insert
-for the equality case (tie-breaker).}
+\item{side}{Either \code{"left"} or \code{"right"}, choosing which side to insert on ties.}
 
-\item{zero_indexed}{If \code{TRUE}, the returned indices are zero-based (\code{0} encodes to first
+\item{zero_indexed}{If \code{TRUE}, the returned indices are zero-based (\code{0} encodes the first
 position); if \code{FALSE} (default), the returned indices are one-based (\code{1}
-encodes to first position).}
+encodes the first position).}
 }
 \value{
-Tensor of insertion indices of same shape as \code{values}.
+Tensor of insertion indices with the same shape as \code{values}.
 }
 \description{
-Perform a binary search, returning indices for insertion of \code{values}
-into \code{sorted_sequence} that maintain the sorting order.
+Return insertion indices that keep \code{values} in sorted order when placed
+into \code{sorted_sequence}.
 }
+\section{Examples}{
+\if{html}{\out{<div class="sourceCode {r}">}}\preformatted{sorted <- op_array(c(-1, 0, 2, 4))
+queries <- c(-2, 0, 3, 7)
+op_searchsorted(sorted, queries)
+op_searchsorted(sorted, queries, side = "right", zero_indexed = TRUE)
+}\if{html}{\out{</div>}}
+}
+
 \seealso{
 Other core ops: \cr
 \code{\link{op_associative_scan}()} \cr
diff --git a/man/optimizer_muon.Rd b/man/optimizer_muon.Rd
new file mode 100644
index 000000000..2f507f80c
--- /dev/null
+++ b/man/optimizer_muon.Rd
@@ -0,0 +1,118 @@
+% Generated by roxygen2: do not edit by hand
+% Please edit documentation in R/optimizers.R
+\name{optimizer_muon}
+\alias{optimizer_muon}
+\title{Optimizer that implements the Muon algorithm.}
+\usage{
+optimizer_muon(
+  learning_rate = 0.001,
+  adam_beta_1 = 0.9,
+  adam_beta_2 = 0.999,
+  epsilon = 1e-07,
+  weight_decay = 0.1,
+  clipnorm = NULL,
+  clipvalue = NULL,
+  global_clipnorm = NULL,
+  use_ema = FALSE,
+  ema_momentum = 0.99,
+  ema_overwrite_frequency = NULL,
+  loss_scale_factor = NULL,
+  gradient_accumulation_steps = NULL,
+  name = "muon",
+  exclude_layers = NULL,
+  exclude_embeddings = TRUE,
+  muon_a = 3.4445,
+  muon_b = -4.775,
+  muon_c = 2.0315,
+  adam_lr_ratio = 0.1,
+  momentum = 0.95,
+  ns_steps = 6L,
+  nesterov = TRUE,
+  ...
+)
+}
+\arguments{
+\item{learning_rate}{A float,
+\code{LearningRateSchedule()} instance, or
+a callable that takes no arguments and returns the actual value to
+use. The learning rate. Defaults to \code{0.001}.}
+
+\item{adam_beta_1}{A float value or a constant float tensor, or a callable
+that takes no arguments and returns the actual value to use.
+The exponential decay rate for the 1st moment estimates. Defaults to
+\code{0.9}.}
+
+\item{adam_beta_2}{A float value or a constant float tensor, ora callable
+that takes no arguments and returns the actual value to use.
+The exponential decay rate for the 2nd moment estimates. Defaults to
+\code{0.999}.}
+
+\item{epsilon}{A small constant for numerical stability. This is
+"epsilon hat" in the Kingma and Ba paper
+(in the formula just before Section 2.1),
+not the epsilon in Algorithm 1 of the paper.
+It is used as in AdamW. Defaults to \code{1e-7}.}
+
+\item{name}{String, name for the object}
+
+\item{exclude_layers}{List of strings, keywords of layer names to exclude.
+All layers with keywords in their path will use AdamW.}
+
+\item{exclude_embeddings}{Boolean value.
+If \code{TRUE}, embedding layers will use AdamW.}
+
+\item{muon_a}{Float, parameter a of the muon algorithm.
+It is recommended to use the default value.}
+
+\item{muon_b}{Float, parameter b of the muon algorithm.
+It is recommended to use the default value.}
+
+\item{muon_c}{Float, parameter c of the muon algorithm.
+It is recommended to use the default value.}
+
+\item{adam_lr_ratio}{Float, the ratio of the learning rate when
+using Adam to the main learning rate.
+it is recommended to set it to \code{0.1}.}
+
+\item{momentum}{Float, momentum used by internal SGD.}
+
+\item{ns_steps}{Integer, number of Newton-Schulz iterations to run.}
+
+\item{nesterov}{Boolean, whether to use Nesterov-style momentum.}
+
+\item{...}{For forward/backward compatibility.}
+}
+\description{
+Note that this optimizer should not be used in the following layers:
+\enumerate{
+\item Embedding layer
+\item Final output fully connected layer
+\item Any {0,1}-D variables
+}
+
+These should all be optimized using AdamW.
+
+The Muon optimizer can use both the Muon update step or the
+AdamW update step based on the following:
+\itemize{
+\item For any variable that isn't 2D, 3D or 4D, the AdamW step
+will be used. This is not configurable.
+\item If the argument \code{exclude_embeddings} (defaults to \code{TRUE}) is set
+to \code{TRUE}, the AdamW step will be used.
+\item For any variable with a name that matches an expression
+listed in the argument \code{exclude_layers} (a list), the
+AdamW step will be used.
+\item Any other variable uses the Muon step.
+}
+
+Typically, you only need to pass the name of your densely-connected
+output layer to \code{exclude_layers}, e.g.
+\code{exclude_layers = "output_dense"}.
+}
+\section{References}{
+\itemize{
+\item \href{https://github.com/KellerJordan/Muon}{Original implementation}
+\item \href{https://arxiv.org/abs/2502.16982}{Liu et al, 2025}
+}
+}
+
diff --git a/man/rnn_cell_gru.Rd b/man/rnn_cell_gru.Rd
index a4d49ea82..0910fe1c2 100644
--- a/man/rnn_cell_gru.Rd
+++ b/man/rnn_cell_gru.Rd
@@ -239,6 +239,7 @@ Other layers: \cr
 \code{\link{layer_random_color_jitter}()} \cr
 \code{\link{layer_random_contrast}()} \cr
 \code{\link{layer_random_crop}()} \cr
+\code{\link{layer_random_elastic_transform}()} \cr
 \code{\link{layer_random_erasing}()} \cr
 \code{\link{layer_random_flip}()} \cr
 \code{\link{layer_random_gaussian_blur}()} \cr
diff --git a/man/rnn_cell_lstm.Rd b/man/rnn_cell_lstm.Rd
index 206b4ead3..d1b341211 100644
--- a/man/rnn_cell_lstm.Rd
+++ b/man/rnn_cell_lstm.Rd
@@ -242,6 +242,7 @@ Other layers: \cr
 \code{\link{layer_random_color_jitter}()} \cr
 \code{\link{layer_random_contrast}()} \cr
 \code{\link{layer_random_crop}()} \cr
+\code{\link{layer_random_elastic_transform}()} \cr
 \code{\link{layer_random_erasing}()} \cr
 \code{\link{layer_random_flip}()} \cr
 \code{\link{layer_random_gaussian_blur}()} \cr
diff --git a/man/rnn_cell_simple.Rd b/man/rnn_cell_simple.Rd
index eda876cbe..c8accc414 100644
--- a/man/rnn_cell_simple.Rd
+++ b/man/rnn_cell_simple.Rd
@@ -215,6 +215,7 @@ Other layers: \cr
 \code{\link{layer_random_color_jitter}()} \cr
 \code{\link{layer_random_contrast}()} \cr
 \code{\link{layer_random_crop}()} \cr
+\code{\link{layer_random_elastic_transform}()} \cr
 \code{\link{layer_random_erasing}()} \cr
 \code{\link{layer_random_flip}()} \cr
 \code{\link{layer_random_gaussian_blur}()} \cr
diff --git a/man/rnn_cells_stack.Rd b/man/rnn_cells_stack.Rd
index 1f8344d6e..3c974a00a 100644
--- a/man/rnn_cells_stack.Rd
+++ b/man/rnn_cells_stack.Rd
@@ -140,6 +140,7 @@ Other layers: \cr
 \code{\link{layer_random_color_jitter}()} \cr
 \code{\link{layer_random_contrast}()} \cr
 \code{\link{layer_random_crop}()} \cr
+\code{\link{layer_random_elastic_transform}()} \cr
 \code{\link{layer_random_erasing}()} \cr
 \code{\link{layer_random_flip}()} \cr
 \code{\link{layer_random_gaussian_blur}()} \cr

From d55d8074a68b51ee7ef0d475a1eab790880bf7d3 Mon Sep 17 00:00:00 2001
From: Tomasz Kalinowski <kalinowskit@gmail.com>
Date: Fri, 26 Sep 2025 16:43:42 -0400
Subject: [PATCH 67/81] add `config_max_epochs()`, `config_set_max_epochs()`,
 `config_max_steps_per_epoch()`, `config_set_max_steps_per_epoch()`,
 `config_is_nnx_enabled()`

---
 .tether/man/keras.config.txt |  5 +++
 R/config.R                   | 79 ++++++++++++++++++++++++++++++++++++
 2 files changed, 84 insertions(+)

diff --git a/.tether/man/keras.config.txt b/.tether/man/keras.config.txt
index ffbe8f439..451ef0afc 100644
--- a/.tether/man/keras.config.txt
+++ b/.tether/man/keras.config.txt
@@ -12,10 +12,15 @@ floatx()
 image_data_format()
 is_flash_attention_enabled()
 is_interactive_logging_enabled()
+is_nnx_enabled()
 is_traceback_filtering_enabled()
+max_epochs()
+max_steps_per_epoch()
 set_backend(backend)
 set_dtype_policy(policy)
 set_epsilon(value)
 set_floatx(value)
 set_image_data_format(data_format)
+set_max_epochs(max_epochs)
+set_max_steps_per_epoch(max_steps_per_epoch)
 
diff --git a/R/config.R b/R/config.R
index 29bc5752e..5e6b6dc28 100644
--- a/R/config.R
+++ b/R/config.R
@@ -289,6 +289,70 @@ function (data_format)
 }
 
 
+#' Configure the default training loop limits.
+#'
+#' @description
+#' These helpers control the caps that built-in training loops obey when running
+#' `fit()`, `evaluate()`, or `predict()`. The values can also be provided via the
+#' `KERAS_MAX_EPOCHS` or `KERAS_MAX_STEPS_PER_EPOCH` environment variables to
+#' quickly constrain a run without modifying source code.
+#'
+#' @returns
+#' `config_max_epochs()` and `config_max_steps_per_epoch()` return the current
+#' integer limits (or `NULL` if the cap is unset). The setter variants return
+#' `NULL` invisibly and are called for side effects.
+#'
+#' @param max_epochs
+#' Integer upper bound for epochs processed by built-in training loops. Use
+#' `NULL` to remove the cap.
+#'
+#' @param max_steps_per_epoch
+#' Integer upper bound for steps processed per epoch by built-in training
+#' loops. Use `NULL` to remove the cap.
+#'
+#' @name config_training_limits
+#' @family config
+#' @rdname config_training_limits
+#' @export
+#' @tether keras.config.max_epochs
+config_max_epochs <-
+function ()
+{
+    args <- capture_args()
+    do.call(keras$config$max_epochs, args)
+}
+
+#' @rdname config_training_limits
+#' @export
+#' @tether keras.config.set_max_epochs
+config_set_max_epochs <-
+function (max_epochs)
+{
+    args <- capture_args(list(max_epochs = as_integer))
+    do.call(keras$config$set_max_epochs, args)
+}
+
+#' @rdname config_training_limits
+#' @export
+#' @tether keras.config.max_steps_per_epoch
+config_max_steps_per_epoch <-
+function ()
+{
+    args <- capture_args()
+    do.call(keras$config$max_steps_per_epoch, args)
+}
+
+#' @rdname config_training_limits
+#' @export
+#' @tether keras.config.set_max_steps_per_epoch
+config_set_max_steps_per_epoch <-
+function (max_steps_per_epoch)
+{
+    args <- capture_args(list(max_steps_per_epoch = as_integer))
+    do.call(keras$config$set_max_steps_per_epoch, args)
+}
+
+
 #' Disables safe mode globally, allowing deserialization of lambdas.
 #'
 #' @returns No return value, called for side effects.
@@ -378,6 +442,21 @@ function ()
 }
 
 
+#' Check whether NNX-specific features are enabled on the JAX backend.
+#'
+#' @returns
+#' Logical flag; `TRUE` if NNX backend features are enabled, `FALSE` otherwise.
+#'
+#' @export
+#' @family config
+#' @tether keras.config.is_nnx_enabled
+config_is_nnx_enabled <-
+function ()
+{
+    keras$config$is_nnx_enabled()
+}
+
+
 #' Turn off traceback filtering.
 #'
 #' @description

From 9708bb58064416ae9e52df570342bdbd8b730363 Mon Sep 17 00:00:00 2001
From: Tomasz Kalinowski <kalinowskit@gmail.com>
Date: Sat, 27 Sep 2025 08:00:05 -0400
Subject: [PATCH 68/81] redocument

---
 R/config.R                                   | 10 ++--
 man/config_backend.Rd                        |  2 +
 man/config_disable_flash_attention.Rd        |  2 +
 man/config_disable_interactive_logging.Rd    |  2 +
 man/config_disable_traceback_filtering.Rd    |  2 +
 man/config_dtype_policy.Rd                   |  2 +
 man/config_enable_flash_attention.Rd         |  2 +
 man/config_enable_interactive_logging.Rd     |  2 +
 man/config_enable_traceback_filtering.Rd     |  2 +
 man/config_enable_unsafe_deserialization.Rd  |  2 +
 man/config_epsilon.Rd                        |  2 +
 man/config_floatx.Rd                         |  2 +
 man/config_image_data_format.Rd              |  2 +
 man/config_is_interactive_logging_enabled.Rd |  2 +
 man/config_is_nnx_enabled.Rd                 | 38 +++++++++++++
 man/config_is_traceback_filtering_enabled.Rd |  2 +
 man/config_max_epochs.Rd                     | 59 ++++++++++++++++++++
 man/config_set_backend.Rd                    |  2 +
 man/config_set_dtype_policy.Rd               |  2 +
 man/config_set_epsilon.Rd                    |  2 +
 man/config_set_floatx.Rd                     |  2 +
 man/config_set_image_data_format.Rd          |  2 +
 man/layer_tfsm.Rd                            |  4 +-
 23 files changed, 142 insertions(+), 7 deletions(-)
 create mode 100644 man/config_is_nnx_enabled.Rd
 create mode 100644 man/config_max_epochs.Rd

diff --git a/R/config.R b/R/config.R
index 5e6b6dc28..a12dac02c 100644
--- a/R/config.R
+++ b/R/config.R
@@ -310,9 +310,9 @@ function (data_format)
 #' Integer upper bound for steps processed per epoch by built-in training
 #' loops. Use `NULL` to remove the cap.
 #'
-#' @name config_training_limits
+#' @name config_max_epochs
 #' @family config
-#' @rdname config_training_limits
+#' @rdname config_max_epochs
 #' @export
 #' @tether keras.config.max_epochs
 config_max_epochs <-
@@ -322,7 +322,7 @@ function ()
     do.call(keras$config$max_epochs, args)
 }
 
-#' @rdname config_training_limits
+#' @rdname config_max_epochs
 #' @export
 #' @tether keras.config.set_max_epochs
 config_set_max_epochs <-
@@ -332,7 +332,7 @@ function (max_epochs)
     do.call(keras$config$set_max_epochs, args)
 }
 
-#' @rdname config_training_limits
+#' @rdname config_max_epochs
 #' @export
 #' @tether keras.config.max_steps_per_epoch
 config_max_steps_per_epoch <-
@@ -342,7 +342,7 @@ function ()
     do.call(keras$config$max_steps_per_epoch, args)
 }
 
-#' @rdname config_training_limits
+#' @rdname config_max_epochs
 #' @export
 #' @tether keras.config.set_max_steps_per_epoch
 config_set_max_steps_per_epoch <-
diff --git a/man/config_backend.Rd b/man/config_backend.Rd
index db97fec3c..6571abd7a 100644
--- a/man/config_backend.Rd
+++ b/man/config_backend.Rd
@@ -55,7 +55,9 @@ Other config: \cr
 \code{\link{config_floatx}()} \cr
 \code{\link{config_image_data_format}()} \cr
 \code{\link{config_is_interactive_logging_enabled}()} \cr
+\code{\link{config_is_nnx_enabled}()} \cr
 \code{\link{config_is_traceback_filtering_enabled}()} \cr
+\code{\link{config_max_epochs}()} \cr
 \code{\link{config_set_backend}()} \cr
 \code{\link{config_set_dtype_policy}()} \cr
 \code{\link{config_set_epsilon}()} \cr
diff --git a/man/config_disable_flash_attention.Rd b/man/config_disable_flash_attention.Rd
index b5f501e87..3f4880a89 100644
--- a/man/config_disable_flash_attention.Rd
+++ b/man/config_disable_flash_attention.Rd
@@ -30,7 +30,9 @@ Other config: \cr
 \code{\link{config_floatx}()} \cr
 \code{\link{config_image_data_format}()} \cr
 \code{\link{config_is_interactive_logging_enabled}()} \cr
+\code{\link{config_is_nnx_enabled}()} \cr
 \code{\link{config_is_traceback_filtering_enabled}()} \cr
+\code{\link{config_max_epochs}()} \cr
 \code{\link{config_set_backend}()} \cr
 \code{\link{config_set_dtype_policy}()} \cr
 \code{\link{config_set_epsilon}()} \cr
diff --git a/man/config_disable_interactive_logging.Rd b/man/config_disable_interactive_logging.Rd
index 536f4e95e..5364b9492 100644
--- a/man/config_disable_interactive_logging.Rd
+++ b/man/config_disable_interactive_logging.Rd
@@ -58,7 +58,9 @@ Other config: \cr
 \code{\link{config_floatx}()} \cr
 \code{\link{config_image_data_format}()} \cr
 \code{\link{config_is_interactive_logging_enabled}()} \cr
+\code{\link{config_is_nnx_enabled}()} \cr
 \code{\link{config_is_traceback_filtering_enabled}()} \cr
+\code{\link{config_max_epochs}()} \cr
 \code{\link{config_set_backend}()} \cr
 \code{\link{config_set_dtype_policy}()} \cr
 \code{\link{config_set_epsilon}()} \cr
diff --git a/man/config_disable_traceback_filtering.Rd b/man/config_disable_traceback_filtering.Rd
index e5ca83c53..187d92bb9 100644
--- a/man/config_disable_traceback_filtering.Rd
+++ b/man/config_disable_traceback_filtering.Rd
@@ -68,7 +68,9 @@ Other config: \cr
 \code{\link{config_floatx}()} \cr
 \code{\link{config_image_data_format}()} \cr
 \code{\link{config_is_interactive_logging_enabled}()} \cr
+\code{\link{config_is_nnx_enabled}()} \cr
 \code{\link{config_is_traceback_filtering_enabled}()} \cr
+\code{\link{config_max_epochs}()} \cr
 \code{\link{config_set_backend}()} \cr
 \code{\link{config_set_dtype_policy}()} \cr
 \code{\link{config_set_epsilon}()} \cr
diff --git a/man/config_dtype_policy.Rd b/man/config_dtype_policy.Rd
index da2ede03f..e03221282 100644
--- a/man/config_dtype_policy.Rd
+++ b/man/config_dtype_policy.Rd
@@ -26,7 +26,9 @@ Other config: \cr
 \code{\link{config_floatx}()} \cr
 \code{\link{config_image_data_format}()} \cr
 \code{\link{config_is_interactive_logging_enabled}()} \cr
+\code{\link{config_is_nnx_enabled}()} \cr
 \code{\link{config_is_traceback_filtering_enabled}()} \cr
+\code{\link{config_max_epochs}()} \cr
 \code{\link{config_set_backend}()} \cr
 \code{\link{config_set_dtype_policy}()} \cr
 \code{\link{config_set_epsilon}()} \cr
diff --git a/man/config_enable_flash_attention.Rd b/man/config_enable_flash_attention.Rd
index a3049f454..6a317e232 100644
--- a/man/config_enable_flash_attention.Rd
+++ b/man/config_enable_flash_attention.Rd
@@ -35,7 +35,9 @@ Other config: \cr
 \code{\link{config_floatx}()} \cr
 \code{\link{config_image_data_format}()} \cr
 \code{\link{config_is_interactive_logging_enabled}()} \cr
+\code{\link{config_is_nnx_enabled}()} \cr
 \code{\link{config_is_traceback_filtering_enabled}()} \cr
+\code{\link{config_max_epochs}()} \cr
 \code{\link{config_set_backend}()} \cr
 \code{\link{config_set_dtype_policy}()} \cr
 \code{\link{config_set_epsilon}()} \cr
diff --git a/man/config_enable_interactive_logging.Rd b/man/config_enable_interactive_logging.Rd
index 75c784741..fd4480950 100644
--- a/man/config_enable_interactive_logging.Rd
+++ b/man/config_enable_interactive_logging.Rd
@@ -58,7 +58,9 @@ Other config: \cr
 \code{\link{config_floatx}()} \cr
 \code{\link{config_image_data_format}()} \cr
 \code{\link{config_is_interactive_logging_enabled}()} \cr
+\code{\link{config_is_nnx_enabled}()} \cr
 \code{\link{config_is_traceback_filtering_enabled}()} \cr
+\code{\link{config_max_epochs}()} \cr
 \code{\link{config_set_backend}()} \cr
 \code{\link{config_set_dtype_policy}()} \cr
 \code{\link{config_set_epsilon}()} \cr
diff --git a/man/config_enable_traceback_filtering.Rd b/man/config_enable_traceback_filtering.Rd
index 698f5687d..ab39b9845 100644
--- a/man/config_enable_traceback_filtering.Rd
+++ b/man/config_enable_traceback_filtering.Rd
@@ -68,7 +68,9 @@ Other config: \cr
 \code{\link{config_floatx}()} \cr
 \code{\link{config_image_data_format}()} \cr
 \code{\link{config_is_interactive_logging_enabled}()} \cr
+\code{\link{config_is_nnx_enabled}()} \cr
 \code{\link{config_is_traceback_filtering_enabled}()} \cr
+\code{\link{config_max_epochs}()} \cr
 \code{\link{config_set_backend}()} \cr
 \code{\link{config_set_dtype_policy}()} \cr
 \code{\link{config_set_epsilon}()} \cr
diff --git a/man/config_enable_unsafe_deserialization.Rd b/man/config_enable_unsafe_deserialization.Rd
index 81776932e..f47830c1f 100644
--- a/man/config_enable_unsafe_deserialization.Rd
+++ b/man/config_enable_unsafe_deserialization.Rd
@@ -26,7 +26,9 @@ Other config: \cr
 \code{\link{config_floatx}()} \cr
 \code{\link{config_image_data_format}()} \cr
 \code{\link{config_is_interactive_logging_enabled}()} \cr
+\code{\link{config_is_nnx_enabled}()} \cr
 \code{\link{config_is_traceback_filtering_enabled}()} \cr
+\code{\link{config_max_epochs}()} \cr
 \code{\link{config_set_backend}()} \cr
 \code{\link{config_set_dtype_policy}()} \cr
 \code{\link{config_set_epsilon}()} \cr
diff --git a/man/config_epsilon.Rd b/man/config_epsilon.Rd
index a60dcc5ee..09689f138 100644
--- a/man/config_epsilon.Rd
+++ b/man/config_epsilon.Rd
@@ -56,7 +56,9 @@ Other config: \cr
 \code{\link{config_floatx}()} \cr
 \code{\link{config_image_data_format}()} \cr
 \code{\link{config_is_interactive_logging_enabled}()} \cr
+\code{\link{config_is_nnx_enabled}()} \cr
 \code{\link{config_is_traceback_filtering_enabled}()} \cr
+\code{\link{config_max_epochs}()} \cr
 \code{\link{config_set_backend}()} \cr
 \code{\link{config_set_dtype_policy}()} \cr
 \code{\link{config_set_epsilon}()} \cr
diff --git a/man/config_floatx.Rd b/man/config_floatx.Rd
index f9a05b6c1..3c57df6a7 100644
--- a/man/config_floatx.Rd
+++ b/man/config_floatx.Rd
@@ -56,7 +56,9 @@ Other config: \cr
 \code{\link{config_epsilon}()} \cr
 \code{\link{config_image_data_format}()} \cr
 \code{\link{config_is_interactive_logging_enabled}()} \cr
+\code{\link{config_is_nnx_enabled}()} \cr
 \code{\link{config_is_traceback_filtering_enabled}()} \cr
+\code{\link{config_max_epochs}()} \cr
 \code{\link{config_set_backend}()} \cr
 \code{\link{config_set_dtype_policy}()} \cr
 \code{\link{config_set_epsilon}()} \cr
diff --git a/man/config_image_data_format.Rd b/man/config_image_data_format.Rd
index b6fac294e..43019c069 100644
--- a/man/config_image_data_format.Rd
+++ b/man/config_image_data_format.Rd
@@ -56,7 +56,9 @@ Other config: \cr
 \code{\link{config_epsilon}()} \cr
 \code{\link{config_floatx}()} \cr
 \code{\link{config_is_interactive_logging_enabled}()} \cr
+\code{\link{config_is_nnx_enabled}()} \cr
 \code{\link{config_is_traceback_filtering_enabled}()} \cr
+\code{\link{config_max_epochs}()} \cr
 \code{\link{config_set_backend}()} \cr
 \code{\link{config_set_dtype_policy}()} \cr
 \code{\link{config_set_epsilon}()} \cr
diff --git a/man/config_is_interactive_logging_enabled.Rd b/man/config_is_interactive_logging_enabled.Rd
index d9835b8e0..12285507c 100644
--- a/man/config_is_interactive_logging_enabled.Rd
+++ b/man/config_is_interactive_logging_enabled.Rd
@@ -59,7 +59,9 @@ Other config: \cr
 \code{\link{config_epsilon}()} \cr
 \code{\link{config_floatx}()} \cr
 \code{\link{config_image_data_format}()} \cr
+\code{\link{config_is_nnx_enabled}()} \cr
 \code{\link{config_is_traceback_filtering_enabled}()} \cr
+\code{\link{config_max_epochs}()} \cr
 \code{\link{config_set_backend}()} \cr
 \code{\link{config_set_dtype_policy}()} \cr
 \code{\link{config_set_epsilon}()} \cr
diff --git a/man/config_is_nnx_enabled.Rd b/man/config_is_nnx_enabled.Rd
new file mode 100644
index 000000000..112ccd395
--- /dev/null
+++ b/man/config_is_nnx_enabled.Rd
@@ -0,0 +1,38 @@
+% Generated by roxygen2: do not edit by hand
+% Please edit documentation in R/config.R
+\name{config_is_nnx_enabled}
+\alias{config_is_nnx_enabled}
+\title{Check whether NNX-specific features are enabled on the JAX backend.}
+\usage{
+config_is_nnx_enabled()
+}
+\value{
+Logical flag; \code{TRUE} if NNX backend features are enabled, \code{FALSE} otherwise.
+}
+\description{
+Check whether NNX-specific features are enabled on the JAX backend.
+}
+\seealso{
+Other config: \cr
+\code{\link{config_backend}()} \cr
+\code{\link{config_disable_flash_attention}()} \cr
+\code{\link{config_disable_interactive_logging}()} \cr
+\code{\link{config_disable_traceback_filtering}()} \cr
+\code{\link{config_dtype_policy}()} \cr
+\code{\link{config_enable_flash_attention}()} \cr
+\code{\link{config_enable_interactive_logging}()} \cr
+\code{\link{config_enable_traceback_filtering}()} \cr
+\code{\link{config_enable_unsafe_deserialization}()} \cr
+\code{\link{config_epsilon}()} \cr
+\code{\link{config_floatx}()} \cr
+\code{\link{config_image_data_format}()} \cr
+\code{\link{config_is_interactive_logging_enabled}()} \cr
+\code{\link{config_is_traceback_filtering_enabled}()} \cr
+\code{\link{config_max_epochs}()} \cr
+\code{\link{config_set_backend}()} \cr
+\code{\link{config_set_dtype_policy}()} \cr
+\code{\link{config_set_epsilon}()} \cr
+\code{\link{config_set_floatx}()} \cr
+\code{\link{config_set_image_data_format}()} \cr
+}
+\concept{config}
diff --git a/man/config_is_traceback_filtering_enabled.Rd b/man/config_is_traceback_filtering_enabled.Rd
index e762b0bb6..7368a9472 100644
--- a/man/config_is_traceback_filtering_enabled.Rd
+++ b/man/config_is_traceback_filtering_enabled.Rd
@@ -70,6 +70,8 @@ Other config: \cr
 \code{\link{config_floatx}()} \cr
 \code{\link{config_image_data_format}()} \cr
 \code{\link{config_is_interactive_logging_enabled}()} \cr
+\code{\link{config_is_nnx_enabled}()} \cr
+\code{\link{config_max_epochs}()} \cr
 \code{\link{config_set_backend}()} \cr
 \code{\link{config_set_dtype_policy}()} \cr
 \code{\link{config_set_epsilon}()} \cr
diff --git a/man/config_max_epochs.Rd b/man/config_max_epochs.Rd
new file mode 100644
index 000000000..6547dd360
--- /dev/null
+++ b/man/config_max_epochs.Rd
@@ -0,0 +1,59 @@
+% Generated by roxygen2: do not edit by hand
+% Please edit documentation in R/config.R
+\name{config_max_epochs}
+\alias{config_max_epochs}
+\alias{config_set_max_epochs}
+\alias{config_max_steps_per_epoch}
+\alias{config_set_max_steps_per_epoch}
+\title{Configure the default training loop limits.}
+\usage{
+config_max_epochs()
+
+config_set_max_epochs(max_epochs)
+
+config_max_steps_per_epoch()
+
+config_set_max_steps_per_epoch(max_steps_per_epoch)
+}
+\arguments{
+\item{max_epochs}{Integer upper bound for epochs processed by built-in training loops. Use
+\code{NULL} to remove the cap.}
+
+\item{max_steps_per_epoch}{Integer upper bound for steps processed per epoch by built-in training
+loops. Use \code{NULL} to remove the cap.}
+}
+\value{
+\code{config_max_epochs()} and \code{config_max_steps_per_epoch()} return the current
+integer limits (or \code{NULL} if the cap is unset). The setter variants return
+\code{NULL} invisibly and are called for side effects.
+}
+\description{
+These helpers control the caps that built-in training loops obey when running
+\code{fit()}, \code{evaluate()}, or \code{predict()}. The values can also be provided via the
+\code{KERAS_MAX_EPOCHS} or \code{KERAS_MAX_STEPS_PER_EPOCH} environment variables to
+quickly constrain a run without modifying source code.
+}
+\seealso{
+Other config: \cr
+\code{\link{config_backend}()} \cr
+\code{\link{config_disable_flash_attention}()} \cr
+\code{\link{config_disable_interactive_logging}()} \cr
+\code{\link{config_disable_traceback_filtering}()} \cr
+\code{\link{config_dtype_policy}()} \cr
+\code{\link{config_enable_flash_attention}()} \cr
+\code{\link{config_enable_interactive_logging}()} \cr
+\code{\link{config_enable_traceback_filtering}()} \cr
+\code{\link{config_enable_unsafe_deserialization}()} \cr
+\code{\link{config_epsilon}()} \cr
+\code{\link{config_floatx}()} \cr
+\code{\link{config_image_data_format}()} \cr
+\code{\link{config_is_interactive_logging_enabled}()} \cr
+\code{\link{config_is_nnx_enabled}()} \cr
+\code{\link{config_is_traceback_filtering_enabled}()} \cr
+\code{\link{config_set_backend}()} \cr
+\code{\link{config_set_dtype_policy}()} \cr
+\code{\link{config_set_epsilon}()} \cr
+\code{\link{config_set_floatx}()} \cr
+\code{\link{config_set_image_data_format}()} \cr
+}
+\concept{config}
diff --git a/man/config_set_backend.Rd b/man/config_set_backend.Rd
index 91044db01..c1323c2cd 100644
--- a/man/config_set_backend.Rd
+++ b/man/config_set_backend.Rd
@@ -49,7 +49,9 @@ Other config: \cr
 \code{\link{config_floatx}()} \cr
 \code{\link{config_image_data_format}()} \cr
 \code{\link{config_is_interactive_logging_enabled}()} \cr
+\code{\link{config_is_nnx_enabled}()} \cr
 \code{\link{config_is_traceback_filtering_enabled}()} \cr
+\code{\link{config_max_epochs}()} \cr
 \code{\link{config_set_dtype_policy}()} \cr
 \code{\link{config_set_epsilon}()} \cr
 \code{\link{config_set_floatx}()} \cr
diff --git a/man/config_set_dtype_policy.Rd b/man/config_set_dtype_policy.Rd
index a34f24080..a43eb55d1 100644
--- a/man/config_set_dtype_policy.Rd
+++ b/man/config_set_dtype_policy.Rd
@@ -35,7 +35,9 @@ Other config: \cr
 \code{\link{config_floatx}()} \cr
 \code{\link{config_image_data_format}()} \cr
 \code{\link{config_is_interactive_logging_enabled}()} \cr
+\code{\link{config_is_nnx_enabled}()} \cr
 \code{\link{config_is_traceback_filtering_enabled}()} \cr
+\code{\link{config_max_epochs}()} \cr
 \code{\link{config_set_backend}()} \cr
 \code{\link{config_set_epsilon}()} \cr
 \code{\link{config_set_floatx}()} \cr
diff --git a/man/config_set_epsilon.Rd b/man/config_set_epsilon.Rd
index d43c4e136..6930b4f38 100644
--- a/man/config_set_epsilon.Rd
+++ b/man/config_set_epsilon.Rd
@@ -72,7 +72,9 @@ Other config: \cr
 \code{\link{config_floatx}()} \cr
 \code{\link{config_image_data_format}()} \cr
 \code{\link{config_is_interactive_logging_enabled}()} \cr
+\code{\link{config_is_nnx_enabled}()} \cr
 \code{\link{config_is_traceback_filtering_enabled}()} \cr
+\code{\link{config_max_epochs}()} \cr
 \code{\link{config_set_backend}()} \cr
 \code{\link{config_set_dtype_policy}()} \cr
 \code{\link{config_set_floatx}()} \cr
diff --git a/man/config_set_floatx.Rd b/man/config_set_floatx.Rd
index 89d9864eb..6dfcae62c 100644
--- a/man/config_set_floatx.Rd
+++ b/man/config_set_floatx.Rd
@@ -84,7 +84,9 @@ Other config: \cr
 \code{\link{config_floatx}()} \cr
 \code{\link{config_image_data_format}()} \cr
 \code{\link{config_is_interactive_logging_enabled}()} \cr
+\code{\link{config_is_nnx_enabled}()} \cr
 \code{\link{config_is_traceback_filtering_enabled}()} \cr
+\code{\link{config_max_epochs}()} \cr
 \code{\link{config_set_backend}()} \cr
 \code{\link{config_set_dtype_policy}()} \cr
 \code{\link{config_set_epsilon}()} \cr
diff --git a/man/config_set_image_data_format.Rd b/man/config_set_image_data_format.Rd
index dc29a37f7..bf20fbf97 100644
--- a/man/config_set_image_data_format.Rd
+++ b/man/config_set_image_data_format.Rd
@@ -75,7 +75,9 @@ Other config: \cr
 \code{\link{config_floatx}()} \cr
 \code{\link{config_image_data_format}()} \cr
 \code{\link{config_is_interactive_logging_enabled}()} \cr
+\code{\link{config_is_nnx_enabled}()} \cr
 \code{\link{config_is_traceback_filtering_enabled}()} \cr
+\code{\link{config_max_epochs}()} \cr
 \code{\link{config_set_backend}()} \cr
 \code{\link{config_set_dtype_policy}()} \cr
 \code{\link{config_set_epsilon}()} \cr
diff --git a/man/layer_tfsm.Rd b/man/layer_tfsm.Rd
index a27ae0273..ff3c10712 100644
--- a/man/layer_tfsm.Rd
+++ b/man/layer_tfsm.Rd
@@ -59,8 +59,8 @@ model |> export_savedmodel("path/to/artifact")
 ## Output Type:
 ##   TensorSpec(shape=(None, 10), dtype=tf.float32, name=None)
 ## Captures:
-##   138000861911120: TensorSpec(shape=(), dtype=tf.resource, name=None)
-##   138000861919184: TensorSpec(shape=(), dtype=tf.resource, name=None)
+##   138303250719632: TensorSpec(shape=(), dtype=tf.resource, name=None)
+##   138303250721744: TensorSpec(shape=(), dtype=tf.resource, name=None)
 
 }\if{html}{\out{</div>}}
 

From cf36a216892f7e7346206c07752cdf7f385d1239 Mon Sep 17 00:00:00 2001
From: Tomasz Kalinowski <kalinowskit@gmail.com>
Date: Sat, 27 Sep 2025 08:00:24 -0400
Subject: [PATCH 69/81] internal tether changes: quantizers

---
 .tether/man/keras.quantizers.txt | 6 ++++++
 1 file changed, 6 insertions(+)

diff --git a/.tether/man/keras.quantizers.txt b/.tether/man/keras.quantizers.txt
index 0970068c1..c113bde91 100644
--- a/.tether/man/keras.quantizers.txt
+++ b/.tether/man/keras.quantizers.txt
@@ -29,6 +29,7 @@ fake_quant_with_min_max_vars(
   axis=None
 )
 get(identifier, **kwargs)
+pack_int4(arr, axis=0)
 quantize_and_dequantize(
   inputs,
   scale,
@@ -37,4 +38,9 @@ quantize_and_dequantize(
 )
 Quantizer(output_dtype='int8')
 serialize(initializer)
+unpack_int4(
+  packed,
+  orig_len,
+  axis=0
+)
 

From c687e5bb73ee351638da4dec6dda5d6503aa5fe8 Mon Sep 17 00:00:00 2001
From: Tomasz Kalinowski <kalinowskit@gmail.com>
Date: Sat, 27 Sep 2025 08:00:39 -0400
Subject: [PATCH 70/81] internal tether changes: distribution

---
 .tether/man/keras.distribution.txt | 1 +
 1 file changed, 1 insertion(+)

diff --git a/.tether/man/keras.distribution.txt b/.tether/man/keras.distribution.txt
index ee66b1b01..67af628b9 100644
--- a/.tether/man/keras.distribution.txt
+++ b/.tether/man/keras.distribution.txt
@@ -21,6 +21,7 @@ ModelParallel(
   *,
   layout_map=None,
   batch_dim_name=None,
+  auto_shard_dataset=True,
   **kwargs
 )
 set_distribution(value)

From 033c605f5483e342780c9fe8a1639bd1300adc5c Mon Sep 17 00:00:00 2001
From: Tomasz Kalinowski <kalinowskit@gmail.com>
Date: Sat, 27 Sep 2025 08:00:55 -0400
Subject: [PATCH 71/81] update NAMESPACE

---
 NAMESPACE | 8 ++++++++
 1 file changed, 8 insertions(+)

diff --git a/NAMESPACE b/NAMESPACE
index e0cf8efc9..a284e4f0e 100644
--- a/NAMESPACE
+++ b/NAMESPACE
@@ -190,12 +190,17 @@ export(config_floatx)
 export(config_image_data_format)
 export(config_is_flash_attention_enabled)
 export(config_is_interactive_logging_enabled)
+export(config_is_nnx_enabled)
 export(config_is_traceback_filtering_enabled)
+export(config_max_epochs)
+export(config_max_steps_per_epoch)
 export(config_set_backend)
 export(config_set_dtype_policy)
 export(config_set_epsilon)
 export(config_set_floatx)
 export(config_set_image_data_format)
+export(config_set_max_epochs)
+export(config_set_max_steps_per_epoch)
 export(constraint_maxnorm)
 export(constraint_minmaxnorm)
 export(constraint_nonneg)
@@ -363,6 +368,7 @@ export(layer_random_color_degeneration)
 export(layer_random_color_jitter)
 export(layer_random_contrast)
 export(layer_random_crop)
+export(layer_random_elastic_transform)
 export(layer_random_erasing)
 export(layer_random_flip)
 export(layer_random_gaussian_blur)
@@ -611,6 +617,7 @@ export(op_ifft2)
 export(op_imag)
 export(op_image_affine_transform)
 export(op_image_crop)
+export(op_image_elastic_transform)
 export(op_image_extract_patches)
 export(op_image_gaussian_blur)
 export(op_image_hsv_to_rgb)
@@ -789,6 +796,7 @@ export(optimizer_ftrl)
 export(optimizer_lamb)
 export(optimizer_lion)
 export(optimizer_loss_scale)
+export(optimizer_muon)
 export(optimizer_nadam)
 export(optimizer_rmsprop)
 export(optimizer_sgd)

From eb76f9472bfd292968c657343e633666125cfd72 Mon Sep 17 00:00:00 2001
From: Tomasz Kalinowski <kalinowskit@gmail.com>
Date: Sat, 27 Sep 2025 08:03:45 -0400
Subject: [PATCH 72/81] check in new tethers

---
 .tether/man/activation_sparse_sigmoid.txt     |  18 ++
 .tether/man/config_is_nnx_enabled.txt         |   9 +
 .tether/man/config_max_epochs.txt             |  13 ++
 .tether/man/config_max_steps_per_epoch.txt    |  13 ++
 .tether/man/config_set_max_epochs.txt         |  13 ++
 .../man/config_set_max_steps_per_epoch.txt    |  14 ++
 .tether/man/keras.ops.numpy.txt               |  10 +-
 .../man/layer_random_elastic_transform.txt    | 130 ++++++++++++++
 ..._categorical_generalized_cross_entropy.txt |  75 ++++++++
 .tether/man/op_angle.txt                      |  18 ++
 .tether/man/op_bartlett.txt                   |  17 ++
 .tether/man/op_blackman.txt                   |  18 ++
 .tether/man/op_cbrt.txt                       |  14 ++
 .tether/man/op_corrcoef.txt                   |  12 ++
 .tether/man/op_deg2rad.txt                    |  21 +++
 .tether/man/op_hamming.txt                    |  19 +++
 .tether/man/op_hanning.txt                    |  19 +++
 .tether/man/op_heaviside.txt                  |  21 +++
 .tether/man/op_image_elastic_transform.txt    |  66 ++++++++
 .tether/man/op_kaiser.txt                     |  22 +++
 .tether/man/op_layer_normalization.txt        |  35 ++++
 .tether/man/op_sparse_sigmoid.txt             |  23 +++
 .tether/man/op_view_as_complex.txt            |  25 +++
 .tether/man/op_view_as_real.txt               |  25 +++
 .tether/man/optimizer_muon.txt                | 160 ++++++++++++++++++
 25 files changed, 807 insertions(+), 3 deletions(-)
 create mode 100644 .tether/man/activation_sparse_sigmoid.txt
 create mode 100644 .tether/man/config_is_nnx_enabled.txt
 create mode 100644 .tether/man/config_max_epochs.txt
 create mode 100644 .tether/man/config_max_steps_per_epoch.txt
 create mode 100644 .tether/man/config_set_max_epochs.txt
 create mode 100644 .tether/man/config_set_max_steps_per_epoch.txt
 create mode 100644 .tether/man/layer_random_elastic_transform.txt
 create mode 100644 .tether/man/loss_categorical_generalized_cross_entropy.txt
 create mode 100644 .tether/man/op_angle.txt
 create mode 100644 .tether/man/op_bartlett.txt
 create mode 100644 .tether/man/op_blackman.txt
 create mode 100644 .tether/man/op_cbrt.txt
 create mode 100644 .tether/man/op_corrcoef.txt
 create mode 100644 .tether/man/op_deg2rad.txt
 create mode 100644 .tether/man/op_hamming.txt
 create mode 100644 .tether/man/op_hanning.txt
 create mode 100644 .tether/man/op_heaviside.txt
 create mode 100644 .tether/man/op_image_elastic_transform.txt
 create mode 100644 .tether/man/op_kaiser.txt
 create mode 100644 .tether/man/op_layer_normalization.txt
 create mode 100644 .tether/man/op_sparse_sigmoid.txt
 create mode 100644 .tether/man/op_view_as_complex.txt
 create mode 100644 .tether/man/op_view_as_real.txt
 create mode 100644 .tether/man/optimizer_muon.txt

diff --git a/.tether/man/activation_sparse_sigmoid.txt b/.tether/man/activation_sparse_sigmoid.txt
new file mode 100644
index 000000000..6de7b64d3
--- /dev/null
+++ b/.tether/man/activation_sparse_sigmoid.txt
@@ -0,0 +1,18 @@
+__signature__
+keras.activations.sparse_sigmoid(x)
+__doc__
+Sparse sigmoid activation function.
+
+It is defined as
+
+`f(x) = 0` for `x <= -1`,
+`f(x) = 0.5 * (x + 1)` for `-1 < x < 1`,
+`f(x) = 1` for `x >= 1`.
+
+Args:
+    x: Input tensor.
+
+Reference:
+
+- [M. Blondel, A. F. T. Martins, V. Niculae, 2019](https://arxiv.org/pdf/1901.02324)
+
diff --git a/.tether/man/config_is_nnx_enabled.txt b/.tether/man/config_is_nnx_enabled.txt
new file mode 100644
index 000000000..658046298
--- /dev/null
+++ b/.tether/man/config_is_nnx_enabled.txt
@@ -0,0 +1,9 @@
+__signature__
+keras.config.is_nnx_enabled()
+__doc__
+Checks whether NNX specific features are enabled for the JAX backend.
+
+Returns:
+    bool: `True` if NNX backend features are enabled, `False` otherwise.
+    Defaults to `False`.
+
diff --git a/.tether/man/config_max_epochs.txt b/.tether/man/config_max_epochs.txt
new file mode 100644
index 000000000..945af4d2e
--- /dev/null
+++ b/.tether/man/config_max_epochs.txt
@@ -0,0 +1,13 @@
+__signature__
+keras.config.max_epochs()
+__doc__
+Get the maximum number of epochs for any call to fit.
+
+Retrieves the limit on the number of epochs set by
+`keras.config.set_max_epochs` or the `KERAS_MAX_EPOCHS` environment
+variable.
+
+Returns:
+    The integer limit on the number of epochs or `None`, if no limit has
+    been set.
+
diff --git a/.tether/man/config_max_steps_per_epoch.txt b/.tether/man/config_max_steps_per_epoch.txt
new file mode 100644
index 000000000..de4b03408
--- /dev/null
+++ b/.tether/man/config_max_steps_per_epoch.txt
@@ -0,0 +1,13 @@
+__signature__
+keras.config.max_steps_per_epoch()
+__doc__
+Get the maximum number of steps for any call to fit/evaluate/predict.
+
+Retrieves the limit on the number of epochs set by
+`keras.config.set_max_steps_per_epoch` or the `KERAS_MAX_STEPS_PER_EPOCH`
+environment variable.
+
+Args:
+    max_epochs: The integer limit on the number of epochs or `None`. If
+        `None`, no limit is applied.
+
diff --git a/.tether/man/config_set_max_epochs.txt b/.tether/man/config_set_max_epochs.txt
new file mode 100644
index 000000000..0845c02a9
--- /dev/null
+++ b/.tether/man/config_set_max_epochs.txt
@@ -0,0 +1,13 @@
+__signature__
+keras.config.set_max_epochs(max_epochs)
+__doc__
+Limit the maximum number of epochs for any call to fit.
+
+This will cap the number of epochs for any training run using `model.fit()`.
+This is purely for debugging, and can also be set via the `KERAS_MAX_EPOCHS`
+environment variable to quickly run a script without modifying its source.
+
+Args:
+    max_epochs: The integer limit on the number of epochs or `None`. If
+        `None`, no limit is applied.
+
diff --git a/.tether/man/config_set_max_steps_per_epoch.txt b/.tether/man/config_set_max_steps_per_epoch.txt
new file mode 100644
index 000000000..f0b060b19
--- /dev/null
+++ b/.tether/man/config_set_max_steps_per_epoch.txt
@@ -0,0 +1,14 @@
+__signature__
+keras.config.set_max_steps_per_epoch(max_steps_per_epoch)
+__doc__
+Limit the maximum number of steps for any call to fit/evaluate/predict.
+
+This will cap the number of steps for single epoch of a call to `fit()`,
+`evaluate()`, or `predict()`. This is purely for debugging, and can also be
+set via the `KERAS_MAX_STEPS_PER_EPOCH` environment variable to quickly run
+a scrip without modifying its source.
+
+Args:
+    max_epochs: The integer limit on the number of epochs or `None`. If
+        `None`, no limit is applied.
+
diff --git a/.tether/man/keras.ops.numpy.txt b/.tether/man/keras.ops.numpy.txt
index 74095b6b6..b91190169 100644
--- a/.tether/man/keras.ops.numpy.txt
+++ b/.tether/man/keras.ops.numpy.txt
@@ -89,6 +89,7 @@ concatenate(xs, axis=0)
 conj(x)
 conjugate(x)
 copy(x)
+corrcoef(x)
 correlate(
   x1,
   x2,
@@ -96,7 +97,6 @@ correlate(
 )
 cos(x)
 cosh(x)
-corrcoef(x)
 count_nonzero(x, axis=None)
 cross(
   x1,
@@ -116,10 +116,10 @@ cumsum(
   axis=None,
   dtype=None
 )
+deg2rad(x)
 diag(x, k=0)
 diagflat(x, k=0)
 diagonal(
-deg2rad(x)
   x,
   offset=0,
   axis1=0,
@@ -134,7 +134,11 @@ digitize(x, bins)
 divide(x1, x2)
 divide_no_nan(x1, x2)
 dot(x1, x2)
-einsum(subscripts, *operands)
+einsum(
+  subscripts,
+  *operands,
+  **kwargs
+)
 empty(shape, dtype=None)
 equal(x1, x2)
 exp(x)
diff --git a/.tether/man/layer_random_elastic_transform.txt b/.tether/man/layer_random_elastic_transform.txt
new file mode 100644
index 000000000..ff1a2df5a
--- /dev/null
+++ b/.tether/man/layer_random_elastic_transform.txt
@@ -0,0 +1,130 @@
+Help on class RandomElasticTransform in module keras.src.layers.preprocessing.image_preprocessing.random_elastic_transform:
+
+class RandomElasticTransform(keras.src.layers.preprocessing.image_preprocessing.base_image_preprocessing_layer.BaseImagePreprocessingLayer)
+ |  RandomElasticTransform(factor=1.0, scale=1.0, interpolation='bilinear', fill_mode='reflect', fill_value=0.0, value_range=(0, 255), seed=None, data_format=None, **kwargs)
+ |
+ |  A preprocessing layer that applies random elastic transformations.
+ |
+ |  This layer distorts input images by applying elastic deformations,
+ |  simulating a physically realistic transformation. The magnitude of the
+ |  distortion is controlled by the `scale` parameter, while the `factor`
+ |  determines the probability of applying the transformation.
+ |
+ |  Args:
+ |      factor: A single float or a tuple of two floats.
+ |          `factor` controls the probability of applying the transformation.
+ |          - `factor=0.0` ensures no erasing is applied.
+ |          - `factor=1.0` means erasing is always applied.
+ |          - If a tuple `(min, max)` is provided, a probability value
+ |            is sampled between `min` and `max` for each image.
+ |          - If a single float is provided, a probability is sampled
+ |            between `0.0` and the given float.
+ |          Default is 1.0.
+ |      scale: A float or a tuple of two floats defining the magnitude of
+ |          the distortion applied.
+ |          - If a tuple `(min, max)` is provided, a random scale value is
+ |            sampled within this range.
+ |          - If a single float is provided, a random scale value is sampled
+ |            between `0.0` and the given float.
+ |          Default is 1.0.
+ |      interpolation: Interpolation mode. Supported values: `"nearest"`,
+ |          `"bilinear"`.
+ |      fill_mode: Points outside the boundaries of the input are filled
+ |          according to the given mode. Available methods are `"constant"`,
+ |          `"nearest"`, `"wrap"` and `"reflect"`. Defaults to `"constant"`.
+ |          - `"reflect"`: `(d c b a | a b c d | d c b a)`
+ |              The input is extended by reflecting about the edge of the last
+ |              pixel.
+ |          - `"constant"`: `(k k k k | a b c d | k k k k)`
+ |              The input is extended by filling all values beyond
+ |              the edge with the same constant value k specified by
+ |              `fill_value`.
+ |          - `"wrap"`: `(a b c d | a b c d | a b c d)`
+ |              The input is extended by wrapping around to the opposite edge.
+ |          - `"nearest"`: `(a a a a | a b c d | d d d d)`
+ |              The input is extended by the nearest pixel.
+ |          Note that when using torch backend, `"reflect"` is redirected to
+ |          `"mirror"` `(c d c b | a b c d | c b a b)` because torch does not
+ |          support `"reflect"`.
+ |          Note that torch backend does not support `"wrap"`.
+ |      fill_value: a float represents the value to be filled outside the
+ |          boundaries when `fill_mode="constant"`.
+ |      value_range: the range of values the incoming images will have.
+ |          Represented as a two-number tuple written `[low, high]`. This is
+ |          typically either `[0, 1]` or `[0, 255]` depending on how your
+ |          preprocessing pipeline is set up.
+ |      seed: Integer. Used to create a random seed.
+ |
+ |  Method resolution order:
+ |      RandomElasticTransform
+ |      keras.src.layers.preprocessing.image_preprocessing.base_image_preprocessing_layer.BaseImagePreprocessingLayer
+ |      keras.src.layers.preprocessing.tf_data_layer.TFDataLayer
+ |      keras.src.layers.layer.Layer
+ |      keras.src.backend.tensorflow.layer.TFLayer
+ |      keras.src.backend.tensorflow.trackable.KerasAutoTrackable
+ |      tensorflow.python.trackable.autotrackable.AutoTrackable
+ |      tensorflow.python.trackable.base.Trackable
+ |      keras.src.ops.operation.Operation
+ |      keras.src.saving.keras_saveable.KerasSaveable
+ |      builtins.object
+ |
+ |  Methods defined here:
+ |
+ |  __init__(
+ |    self,
+ |    factor=1.0,
+ |    scale=1.0,
+ |    interpolation='bilinear',
+ |    fill_mode='reflect',
+ |    fill_value=0.0,
+ |    value_range=(0, 255),
+ |    seed=None,
+ |    data_format=None,
+ |    **kwargs
+ |  )
+ |      Initialize self.  See help(type(self)) for accurate signature.
+ |
+ |  compute_output_shape(self, input_shape)
+ |
+ |  get_config(self)
+ |      Returns the config of the object.
+ |
+ |      An object config is a Python dictionary (serializable)
+ |      containing the information needed to re-instantiate it.
+ |
+ |  get_elastic_transform_params(
+ |    self,
+ |    height,
+ |    width,
+ |    factor
+ |  )
+ |
+ |  get_random_transformation(
+ |    self,
+ |    data,
+ |    training=True,
+ |    seed=None
+ |  )
+ |
+ |  transform_images(
+ |    self,
+ |    images,
+ |    transformation,
+ |    training=True
+ |  )
+ |
+ |  transform_labels(
+ |    self,
+ |    labels,
+ |    transformation,
+ |    training=True
+ |  )
+ |
+ |  transform_segmentation_masks(
+ |    self,
+ |    segmentation_masks,
+ |    transformation,
+ |    training=True
+ |  )
+ |
+
diff --git a/.tether/man/loss_categorical_generalized_cross_entropy.txt b/.tether/man/loss_categorical_generalized_cross_entropy.txt
new file mode 100644
index 000000000..57d2a8b85
--- /dev/null
+++ b/.tether/man/loss_categorical_generalized_cross_entropy.txt
@@ -0,0 +1,75 @@
+Help on class CategoricalGeneralizedCrossEntropy in module keras.src.losses.losses:
+
+class CategoricalGeneralizedCrossEntropy(LossFunctionWrapper)
+ |  CategoricalGeneralizedCrossEntropy(q=0.5, reduction='sum_over_batch_size', name='categorical_generalized_cross_entropy', dtype=None)
+ |
+ |  Computes the Generalized Cross Entropy loss between `y_true` & `y_pred`.
+ |
+ |  Generalized Cross Entropy (GCE) is a noise-robust loss function
+ |  that provides better robustness against noisy labels than
+ |  standard cross entropy.
+ |  It generalizes both cross entropy and mean absolute error through
+ |  the parameter q, where values closer to 1 make the loss more robust
+ |  to noisy labels.
+ |
+ |  Formula:
+ |  ```python
+ |  loss = (1 - p**q) / q
+ |  ```
+ |  where `p` is the predicted probability for the true class and `q`
+ |  is the noise parameter.
+ |
+ |  Args:
+ |      q: Float in range `(0, 1)`. It is the noise parameter.
+ |         Controls the behavior of the loss:
+ |          - As `q` approaches 0: Behaves more like cross entropy
+ |          - As `q` approaches 1: Behaves more like mean absolute error
+ |         Defaults to `0.5`
+ |      reduction: Type of reduction to apply to the loss. In almost all cases
+ |          this should be `"sum_over_batch_size"`. Supported options are
+ |          `"sum"`, `"sum_over_batch_size"`, `"mean"`,
+ |          `"mean_with_sample_weight"` or `None`. `"sum"` sums the loss,
+ |          `"sum_over_batch_size"` and `"mean"` sum the loss and divide by the
+ |          sample size, and `"mean_with_sample_weight"` sums the loss and
+ |          divides by the sum of the sample weights. `"none"` and `None`
+ |          perform no aggregation. Defaults to `"sum_over_batch_size"`.
+ |      name: Optional name for the loss instance.
+ |      dtype: The dtype of the loss's computations. Defaults to `None`, which
+ |          means using `keras.backend.floatx()`. `keras.backend.floatx()` is a
+ |          `"float32"` unless set to different value
+ |          (via `keras.backend.set_floatx()`). If a `keras.DTypePolicy` is
+ |          provided, then the `compute_dtype` will be utilized.
+ |
+ |  Example:
+ |  ```python
+ |  y_true = np.array([0, 1, 0, 1])
+ |  y_pred = np.array([[0.7, 0.3], [0.2, 0.8], [0.6, 0.4], [0.4, 0.6]])
+ |  keras.losses.CategoricalGeneralizedCrossEntropy()(y_true, y_pred)
+ |  ```
+ |
+ |  References:
+ |      - [Zhang, Sabuncu, 2018](https://arxiv.org/abs/1805.07836)
+ |        ("Generalized Cross Entropy Loss for Training
+ |          Deep Neural Networks with Noisy Labels")
+ |
+ |  Method resolution order:
+ |      CategoricalGeneralizedCrossEntropy
+ |      LossFunctionWrapper
+ |      keras.src.losses.loss.Loss
+ |      keras.src.saving.keras_saveable.KerasSaveable
+ |      builtins.object
+ |
+ |  Methods defined here:
+ |
+ |  __init__(
+ |    self,
+ |    q=0.5,
+ |    reduction='sum_over_batch_size',
+ |    name='categorical_generalized_cross_entropy',
+ |    dtype=None
+ |  )
+ |      Initialize self.  See help(type(self)) for accurate signature.
+ |
+ |  get_config(self)
+ |
+
diff --git a/.tether/man/op_angle.txt b/.tether/man/op_angle.txt
new file mode 100644
index 000000000..1a0b0114e
--- /dev/null
+++ b/.tether/man/op_angle.txt
@@ -0,0 +1,18 @@
+__signature__
+keras.ops.angle(x)
+__doc__
+Element-wise angle of a complex tensor.
+
+Arguments:
+    x: Input tensor. Can be real or complex.
+
+Returns:
+    Output tensor of same shape as x. containing the angle of each element
+    (in radians).
+
+Example:
+>>> x = keras.ops.convert_to_tensor([[1 + 3j, 2 - 5j], [4 - 3j, 3 + 2j]])
+>>> keras.ops.angle(x)
+array([[ 1.2490457, -1.19029  ],
+   [-0.6435011,  0.5880026]], dtype=float32)
+
diff --git a/.tether/man/op_bartlett.txt b/.tether/man/op_bartlett.txt
new file mode 100644
index 000000000..6c00ebeb1
--- /dev/null
+++ b/.tether/man/op_bartlett.txt
@@ -0,0 +1,17 @@
+__signature__
+keras.ops.bartlett(x)
+__doc__
+Bartlett window function.
+The Bartlett window is a triangular window that rises then falls linearly.
+
+Args:
+    x: Scalar or 1D Tensor. Window length.
+
+Returns:
+    A 1D tensor containing the Bartlett window values.
+
+Example:
+>>> x = keras.ops.convert_to_tensor(5)
+>>> keras.ops.bartlett(x)
+array([0. , 0.5, 1. , 0.5, 0. ], dtype=float32)
+
diff --git a/.tether/man/op_blackman.txt b/.tether/man/op_blackman.txt
new file mode 100644
index 000000000..50a5f5e4f
--- /dev/null
+++ b/.tether/man/op_blackman.txt
@@ -0,0 +1,18 @@
+__signature__
+keras.ops.blackman(x)
+__doc__
+Blackman window function.
+The Blackman window is a taper formed by using a weighted cosine.
+
+Args:
+    x: Scalar or 1D Tensor. Window length.
+
+Returns:
+    A 1D tensor containing the Blackman window values.
+
+Example:
+>>> x = keras.ops.convert_to_tensor(5)
+>>> keras.ops.blackman(x)
+array([-1.3877788e-17,  3.4000000e-01,  1.0000000e+00,  3.4000000e-01,
+       -1.3877788e-17], dtype=float32)
+
diff --git a/.tether/man/op_cbrt.txt b/.tether/man/op_cbrt.txt
new file mode 100644
index 000000000..d3e3c285a
--- /dev/null
+++ b/.tether/man/op_cbrt.txt
@@ -0,0 +1,14 @@
+__signature__
+keras.ops.cbrt(x)
+__doc__
+Computes the cube root of the input tensor, element-wise.
+
+This operation returns the real-valued cube root of `x`, handling
+negative numbers properly in the real domain.
+
+Args:
+    x: Input tensor.
+
+Returns:
+    A tensor containing the cube root of each element in `x`.
+
diff --git a/.tether/man/op_corrcoef.txt b/.tether/man/op_corrcoef.txt
new file mode 100644
index 000000000..bff558a2c
--- /dev/null
+++ b/.tether/man/op_corrcoef.txt
@@ -0,0 +1,12 @@
+__signature__
+keras.ops.corrcoef(x)
+__doc__
+Compute the Pearson correlation coefficient matrix.
+
+Args:
+    x: A 2D tensor of shape `(N, D)`, where N is the number of variables
+       and D is the number of observations.
+
+Returns:
+    A tensor of shape `(N, N)` representing the correlation matrix.
+
diff --git a/.tether/man/op_deg2rad.txt b/.tether/man/op_deg2rad.txt
new file mode 100644
index 000000000..c8d24cda5
--- /dev/null
+++ b/.tether/man/op_deg2rad.txt
@@ -0,0 +1,21 @@
+__signature__
+keras.ops.deg2rad(x)
+__doc__
+Convert angles from degrees to radians.
+
+The conversion is defined as:
+`rad = deg * (π / 180)`
+
+Args:
+    x: Input tensor of angles in degrees.
+
+Returns:
+    A tensor containing angles converted to radians.
+
+Examples:
+>>> from keras import ops
+>>> ops.deg2rad(180.0)
+3.141592653589793
+>>> ops.deg2rad([0.0, 90.0, 180.0])
+array([0., 1.57079633, 3.14159265])
+
diff --git a/.tether/man/op_hamming.txt b/.tether/man/op_hamming.txt
new file mode 100644
index 000000000..031efc106
--- /dev/null
+++ b/.tether/man/op_hamming.txt
@@ -0,0 +1,19 @@
+__signature__
+keras.ops.hamming(x)
+__doc__
+Hamming window function.
+
+The Hamming window is defined as:
+`w[n] = 0.54 - 0.46 * cos(2 * pi * n / (N - 1))` for `0 <= n <= N - 1`.
+
+Args:
+    x: Scalar or 1D Tensor. The window length.
+
+Returns:
+    A 1D tensor containing the Hamming window values.
+
+Example:
+>>> x = keras.ops.convert_to_tensor(5)
+>>> keras.ops.hamming(x)
+array([0.08, 0.54, 1.  , 0.54, 0.08], dtype=float32)
+
diff --git a/.tether/man/op_hanning.txt b/.tether/man/op_hanning.txt
new file mode 100644
index 000000000..33126574d
--- /dev/null
+++ b/.tether/man/op_hanning.txt
@@ -0,0 +1,19 @@
+__signature__
+keras.ops.hanning(x)
+__doc__
+Hanning window function.
+
+The Hanning window is defined as:
+`w[n] = 0.5 - 0.5 * cos(2 * pi * n / (N - 1))` for `0 <= n <= N - 1`.
+
+Args:
+    x: Scalar or 1D Tensor. The window length.
+
+Returns:
+    A 1D tensor containing the Hanning window values.
+
+Example:
+>>> x = keras.ops.convert_to_tensor(5)
+>>> keras.ops.hanning(x)
+array([0. , 0.5, 1. , 0.5, 0. ], dtype=float32)
+
diff --git a/.tether/man/op_heaviside.txt b/.tether/man/op_heaviside.txt
new file mode 100644
index 000000000..bafbe1c8b
--- /dev/null
+++ b/.tether/man/op_heaviside.txt
@@ -0,0 +1,21 @@
+__signature__
+keras.ops.heaviside(x1, x2)
+__doc__
+Heaviside step function.
+
+The Heaviside step function is defined as:
+`heaviside(x1, x2) = 0 if x1 < 0, 1 if x1 > 0, x2 if x1 == 0`
+
+Args:
+    x1: A tensor input.
+    x2: A scalar or tensor, the value to return when `x1 == 0`.
+
+Returns:
+    A tensor with a shape determined by broadcasting `x1` and `x2`.
+
+Example:
+>>> x1 = keras.ops.convert_to_tensor([-2.0, 0.0, 3.0])
+>>> x2 = 0.5
+>>> keras.ops.heaviside(x1, x2)
+array([0. , 0.5, 1. ], dtype=float32)
+
diff --git a/.tether/man/op_image_elastic_transform.txt b/.tether/man/op_image_elastic_transform.txt
new file mode 100644
index 000000000..6a9cd7675
--- /dev/null
+++ b/.tether/man/op_image_elastic_transform.txt
@@ -0,0 +1,66 @@
+__signature__
+keras.ops.image.elastic_transform(
+  images,
+  alpha=20.0,
+  sigma=5.0,
+  interpolation='bilinear',
+  fill_mode='reflect',
+  fill_value=0.0,
+  seed=None,
+  data_format=None
+)
+__doc__
+Applies elastic deformation to the image(s).
+
+Args:
+    images: Input image or batch of images. Must be 3D or 4D.
+    alpha: Scaling factor that controls the intensity of the deformation.
+    sigma: Standard deviation of the Gaussian filter used for
+        smoothing the displacement fields.
+    interpolation: Interpolation method. Available methods are `"nearest"`,
+        and `"bilinear"`. Defaults to `"bilinear"`.
+    fill_mode: Points outside the boundaries of the input are filled
+        according to the given mode. Available methods are `"constant"`,
+        `"nearest"`, `"wrap"` and `"reflect"`. Defaults to `"constant"`.
+        - `"reflect"`: `(d c b a | a b c d | d c b a)`
+            The input is extended by reflecting about the edge of the last
+            pixel.
+        - `"constant"`: `(k k k k | a b c d | k k k k)`
+            The input is extended by filling all values beyond
+            the edge with the same constant value k specified by
+            `fill_value`.
+        - `"wrap"`: `(a b c d | a b c d | a b c d)`
+            The input is extended by wrapping around to the opposite edge.
+        - `"nearest"`: `(a a a a | a b c d | d d d d)`
+            The input is extended by the nearest pixel.
+    fill_value: Value used for points outside the boundaries of the input if
+        `fill_mode="constant"`. Defaults to `0`.
+    data_format: A string specifying the data format of the input tensor.
+        It can be either `"channels_last"` or `"channels_first"`.
+        `"channels_last"` corresponds to inputs with shape
+        `(batch, height, width, channels)`, while `"channels_first"`
+        corresponds to inputs with shape `(batch, channels, height, width)`.
+        If not specified, the value will default to
+        `keras.config.image_data_format`.
+
+Returns:
+    Transformed image or batch of images with elastic deformation.
+
+Examples:
+
+>>> x = np.random.random((2, 64, 80, 3))  # batch of 2 RGB images
+>>> y = keras.ops.image.elastic_transform(x)
+>>> y.shape
+(2, 64, 80, 3)
+
+>>> x = np.random.random((64, 80, 3))  # single RGB image
+>>> y = keras.ops.image.elastic_transform(x)
+>>> y.shape
+(64, 80, 3)
+
+>>> x = np.random.random((2, 3, 64, 80))  # batch of 2 RGB images
+>>> y = keras.ops.image.elastic_transform(
+...     x, data_format="channels_first")
+>>> y.shape
+(2, 3, 64, 80)
+
diff --git a/.tether/man/op_kaiser.txt b/.tether/man/op_kaiser.txt
new file mode 100644
index 000000000..b49a7d3e4
--- /dev/null
+++ b/.tether/man/op_kaiser.txt
@@ -0,0 +1,22 @@
+__signature__
+keras.ops.kaiser(x, beta)
+__doc__
+Kaiser window function.
+
+The Kaiser window is defined as:
+`w[n] = I0(beta * sqrt(1 - (2n / (N - 1) - 1)^2)) / I0(beta)`
+where I0 is the modified zeroth-order Bessel function of the first kind.
+
+Args:
+    x: Scalar or 1D Tensor. The window length.
+    beta: Float. Shape parameter for the Kaiser window.
+
+Returns:
+    A 1D tensor containing the Kaiser window values.
+
+Example:
+>>> x = keras.ops.convert_to_tensor(5)
+>>> keras.ops.kaiser(x, beta=14.0)
+array([7.7268669e-06, 1.6493219e-01, 1.0000000e+00, 1.6493219e-01,
+   7.7268669e-06], dtype=float32)
+
diff --git a/.tether/man/op_layer_normalization.txt b/.tether/man/op_layer_normalization.txt
new file mode 100644
index 000000000..33bc93250
--- /dev/null
+++ b/.tether/man/op_layer_normalization.txt
@@ -0,0 +1,35 @@
+__signature__
+keras.ops.layer_normalization(
+  x,
+  gamma=None,
+  beta=None,
+  axis=-1,
+  epsilon=None,
+  **kwargs
+)
+__doc__
+Layer normalization layer (Ba et al., 2016).
+
+Normalize the activations of the previous layer for each given example in a
+batch independently, rather than across a batch like Batch Normalization.
+i.e. applies a transformation that maintains the mean activation within each
+example close to 0 and the activation standard deviation close to 1.
+
+Args:
+    x: Input tensor.
+    gamma: Optional scaling factor for the normalization.
+    beta: Optional add offset for the normalized tensor.
+    axis: The axis or axes along which to perform normalization. Default to
+        `-1`.
+    epsilon: A lower bound value for the norm.
+        Defaults to `backend.epsilon()`.
+
+Returns:
+    The normalized array.
+
+Example:
+
+>>> x = keras.ops.arange(5, dtype="float32")
+>>> keras.ops.layer_normalization(x)
+array([-1.4142135, -0.70710677, 0.0, 0.7071067, 1.4142135])
+
diff --git a/.tether/man/op_sparse_sigmoid.txt b/.tether/man/op_sparse_sigmoid.txt
new file mode 100644
index 000000000..06861c5a7
--- /dev/null
+++ b/.tether/man/op_sparse_sigmoid.txt
@@ -0,0 +1,23 @@
+__signature__
+keras.ops.sparse_sigmoid(x)
+__doc__
+Sparse sigmoid activation function.
+
+It is defined as
+
+`f(x) = 0` for `x <= -1`,
+`f(x) = 0.5 * (x + 1)` for `-1 < x < 1`,
+`f(x) = 1` for `x >= 1`.
+
+Args:
+    x: Input tensor.
+
+Returns:
+    A tensor with the same shape as `x`.
+
+Example:
+
+>>> x = keras.ops.convert_to_tensor([-6.0, 1.0, 0.0, 1.0, 6.0])
+>>> keras.ops.sparse_sigmoid(x)
+array([0. , 1. , 0.5, 1. , 1. ], dtype=float32)
+
diff --git a/.tether/man/op_view_as_complex.txt b/.tether/man/op_view_as_complex.txt
new file mode 100644
index 000000000..c195cbce8
--- /dev/null
+++ b/.tether/man/op_view_as_complex.txt
@@ -0,0 +1,25 @@
+__signature__
+keras.ops.view_as_complex(x)
+__doc__
+Converts a real tensor with shape `(..., 2)` to a complex tensor,
+where the last dimension represents the real and imaginary components
+of a complex tensor.
+
+Args:
+    x: A real tensor with last dimension of size 2.
+
+Returns:
+    A complex tensor with shape `x.shape[:-1]`.
+
+Example:
+
+```
+>>> import numpy as np
+>>> from keras import ops
+
+>>> real_imag = np.array([[1.0, 2.0], [3.0, 4.0]])
+>>> complex_tensor = ops.view_as_complex(real_imag)
+>>> complex_tensor
+array([1.+2.j, 3.+4.j])
+```
+
diff --git a/.tether/man/op_view_as_real.txt b/.tether/man/op_view_as_real.txt
new file mode 100644
index 000000000..9573606b5
--- /dev/null
+++ b/.tether/man/op_view_as_real.txt
@@ -0,0 +1,25 @@
+__signature__
+keras.ops.view_as_real(x)
+__doc__
+Converts a complex tensor to a real tensor with shape `(..., 2)`,
+where the last dimension represents the real and imaginary components.
+
+Args:
+    x: A complex tensor.
+
+Returns:
+    A real tensor where the last dimension contains the
+    real and imaginary parts.
+
+Example:
+```
+>>> import numpy as np
+>>> from keras import ops
+
+>>> complex_tensor = np.array([1 + 2j, 3 + 4j])
+>>> real = ops.view_as_real(complex_tensor)
+>>> real
+array([[1., 2.],
+       [3., 4.]])
+```
+
diff --git a/.tether/man/optimizer_muon.txt b/.tether/man/optimizer_muon.txt
new file mode 100644
index 000000000..47ed5a5c3
--- /dev/null
+++ b/.tether/man/optimizer_muon.txt
@@ -0,0 +1,160 @@
+Help on class Muon in module keras.src.optimizers.muon:
+
+class Muon(keras.src.optimizers.optimizer.Optimizer)
+ |  Muon(learning_rate=0.001, adam_beta_1=0.9, adam_beta_2=0.999, epsilon=1e-07, weight_decay=0.1, clipnorm=None, clipvalue=None, global_clipnorm=None, use_ema=False, ema_momentum=0.99, ema_overwrite_frequency=None, loss_scale_factor=None, gradient_accumulation_steps=None, name='muon', exclude_layers=None, exclude_embeddings=True, muon_a=3.4445, muon_b=-4.775, muon_c=2.0315, adam_lr_ratio=0.1, momentum=0.95, ns_steps=6, nesterov=True, **kwargs)
+ |
+ |  Optimizer that implements the Muon algorithm.
+ |
+ |  Note that this optimizer should not be used in the following layers:
+ |
+ |  1. Embedding layer
+ |  2. Final output fully connected layer
+ |  3. Any {0,1}-D variables
+ |
+ |  These should all be optimized using AdamW.
+ |
+ |  The Muon optimizer can use both the Muon update step or the
+ |  AdamW update step based on the following:
+ |
+ |  - For any variable that isn't 2D, 3D or 4D, the AdamW step
+ |      will be used. This is not configurable.
+ |  - If the argument `exclude_embeddings` (defaults to `True`) is set
+ |  to `True`, the AdamW step will be used.
+ |  - For any variablewith a name that matches an expression
+ |      listed in the argument `exclude_layers` (a list), the
+ |      AdamW step will be used.
+ |  - Any other variable uses the Muon step.
+ |
+ |  Typically, you only need to pass the name of your densely-connected
+ |  output layer to `exclude_layers`, e.g.
+ |  `exclude_layers=["output_dense"]`.
+ |
+ |  References:
+ |      - [Original implementation](https://github.com/KellerJordan/Muon)
+ |      - [Liu et al, 2025](https://arxiv.org/abs/2502.16982)
+ |
+ |  Args:
+ |      learning_rate: A float,
+ |          `keras.optimizers.schedules.LearningRateSchedule` instance, or
+ |          a callable that takes no arguments and returns the actual value to
+ |          use. The learning rate. Defaults to `0.001`.
+ |      adam_beta_1: A float value or a constant float tensor, or a callable
+ |          that takes no arguments and returns the actual value to use.
+ |          The exponential decay rate for the 1st moment estimates. Defaults to
+ |          `0.9`.
+ |      adam_beta_2: A float value or a constant float tensor, ora callable
+ |          that takes no arguments and returns the actual value to use.
+ |          The exponential decay rate for the 2nd moment estimates. Defaults to
+ |          `0.999`.
+ |      epsilon: A small constant for numerical stability. This is
+ |          "epsilon hat" in the Kingma and Ba paper
+ |          (in the formula just before Section 2.1),
+ |          not the epsilon in Algorithm 1 of the paper.
+ |          It be used at Adamw.Defaults to `1e-7`.
+ |      exclude_layers: List of strings, keywords of layer names to exclude.
+ |          All layers with keywords in their path will use adamw.
+ |      exclude_embeddings: Boolean value
+ |          If True, embedding layers will use adamw.
+ |      muon_a: Float, parameter a of the muon algorithm.
+ |          It is recommended to use the default value
+ |      muon_b: Float, parameter b of the muon algorithm.
+ |          It is recommended to use the default value
+ |      muon_c: Float, parameter c of the muon algorithm.
+ |          It is recommended to use the default value
+ |      adam_lr_ratio: Float, the ratio of the learning rate when
+ |              using Adam to the main learning rate.
+ |              it is recommended to set it to 0.1
+ |      momentum: Float, momentum used by internal SGD.
+ |      ns_steps: Integer, number of Newton-Schulz iterations to run.
+ |      nesterov: Boolean, whether to use Nesterov-style momentum
+ |      {{base_optimizer_keyword_args}}
+ |
+ |  Method resolution order:
+ |      Muon
+ |      keras.src.optimizers.optimizer.Optimizer
+ |      keras.src.backend.tensorflow.optimizer.TFOptimizer
+ |      keras.src.backend.tensorflow.trackable.KerasAutoTrackable
+ |      tensorflow.python.trackable.autotrackable.AutoTrackable
+ |      tensorflow.python.trackable.base.Trackable
+ |      keras.src.optimizers.base_optimizer.BaseOptimizer
+ |      keras.src.saving.keras_saveable.KerasSaveable
+ |      builtins.object
+ |
+ |  Methods defined here:
+ |
+ |  __init__(
+ |    self,
+ |    learning_rate=0.001,
+ |    adam_beta_1=0.9,
+ |    adam_beta_2=0.999,
+ |    epsilon=1e-07,
+ |    weight_decay=0.1,
+ |    clipnorm=None,
+ |    clipvalue=None,
+ |    global_clipnorm=None,
+ |    use_ema=False,
+ |    ema_momentum=0.99,
+ |    ema_overwrite_frequency=None,
+ |    loss_scale_factor=None,
+ |    gradient_accumulation_steps=None,
+ |    name='muon',
+ |    exclude_layers=None,
+ |    exclude_embeddings=True,
+ |    muon_a=3.4445,
+ |    muon_b=-4.775,
+ |    muon_c=2.0315,
+ |    adam_lr_ratio=0.1,
+ |    momentum=0.95,
+ |    ns_steps=6,
+ |    nesterov=True,
+ |    **kwargs
+ |  )
+ |      Initialize self.  See help(type(self)) for accurate signature.
+ |
+ |  build(self, var_list)
+ |      Initialize optimizer variables.
+ |
+ |      Adam optimizer has 3 types of variables: momentums, velocities and
+ |      velocity_hat (only set when amsgrad is applied),
+ |
+ |      Args:
+ |          var_list: list of model variables to build Adam variables on.
+ |
+ |  get_config(self)
+ |      Returns the config of the optimizer.
+ |
+ |      An optimizer config is a Python dictionary (serializable)
+ |      containing the configuration of an optimizer.
+ |      The same optimizer can be reinstantiated later
+ |      (without any saved state) from this configuration.
+ |
+ |      Subclass optimizer should override this method to include other
+ |      hyperparameters.
+ |
+ |      Returns:
+ |          Python dictionary.
+ |
+ |  transpose_last_axis(self, X)
+ |
+ |  update_step(
+ |    self,
+ |    gradient,
+ |    variable,
+ |    learning_rate
+ |  )
+ |
+ |  zeropower_via_newtonschulz5(
+ |    self,
+ |    x,
+ |    steps: int
+ |  )
+ |      We apply the Newton-Schulz iteration to compute matrix G.
+ |
+ |      We select a quintic iteration that maximizes the slope at zero. This
+ |      approach helps minimize steps, even if the iteration doesn't fully
+ |      converge across the interval. The result isn't exactly UV^T (from the
+ |      SVD of G), but rather an approximation like US'V^T. Despite this
+ |      approximation, model performance remains unaffected compared to using
+ |      the exact UV^T from the SVD.
+ |
+

From 01eb4b072e54294c037d9a402106f270186fc0cf Mon Sep 17 00:00:00 2001
From: Tomasz Kalinowski <kalinowskit@gmail.com>
Date: Sat, 27 Sep 2025 08:11:41 -0400
Subject: [PATCH 73/81] retether guides

---
 .../making_new_layers_and_models_via_subclassing.Rmd          | 2 +-
 .tether/vignettes-src/transfer_learning.Rmd                   | 2 +-
 .../writing_a_custom_training_loop_in_tensorflow.Rmd          | 2 +-
 tools/retether.R                                              | 2 +-
 .../making_new_layers_and_models_via_subclassing.Rmd          | 4 ++--
 vignettes-src/transfer_learning.Rmd                           | 2 +-
 6 files changed, 7 insertions(+), 7 deletions(-)

diff --git a/.tether/vignettes-src/making_new_layers_and_models_via_subclassing.Rmd b/.tether/vignettes-src/making_new_layers_and_models_via_subclassing.Rmd
index ddaf7aafd..9af53ff42 100644
--- a/.tether/vignettes-src/making_new_layers_and_models_via_subclassing.Rmd
+++ b/.tether/vignettes-src/making_new_layers_and_models_via_subclassing.Rmd
@@ -218,7 +218,7 @@ All layers you've seen so far in this guide work with all Keras backends.
 The `keras.ops` namespace gives you access to:
 
 - The NumPy API, e.g. `ops.matmul`, `ops.sum`, `ops.reshape`, `ops.stack`, etc.
-- Neural networks-specific APIs such as `ops.softmax`, `ops`.conv`, `ops.binary_crossentropy`, `ops.relu`, etc.
+- Neural networks-specific APIs such as `ops.softmax`, `ops.conv`, `ops.binary_crossentropy`, `ops.relu`, etc.
 
 You can also use backend-native APIs in your layers (such as `tf.nn` functions),
 but if you do this, then your layer will only be usable with the backend in question.
diff --git a/.tether/vignettes-src/transfer_learning.Rmd b/.tether/vignettes-src/transfer_learning.Rmd
index 9230064c8..ed90015ed 100644
--- a/.tether/vignettes-src/transfer_learning.Rmd
+++ b/.tether/vignettes-src/transfer_learning.Rmd
@@ -53,7 +53,7 @@ ImageNet dataset, and retraining it on the Kaggle "cats vs dogs" classification
  dataset.
 
 This is adapted from
-[Deep Learning with Python](https://www.manning.com/books/deep-learning-with-python)
+[Deep Learning with Python](https://deeplearningwithpython.io/)
 and the 2016 blog post
 ["building powerful image classification models using very little data"](https://blog.keras.io/building-powerful-image-classification-models-using-very-little-data.html).
 
diff --git a/.tether/vignettes-src/writing_a_custom_training_loop_in_tensorflow.Rmd b/.tether/vignettes-src/writing_a_custom_training_loop_in_tensorflow.Rmd
index bfc75688b..888f85c33 100644
--- a/.tether/vignettes-src/writing_a_custom_training_loop_in_tensorflow.Rmd
+++ b/.tether/vignettes-src/writing_a_custom_training_loop_in_tensorflow.Rmd
@@ -368,7 +368,7 @@ A GAN training loop looks like this:
 as real.
 
 For a much more detailed overview of how GANs works, see
-[Deep Learning with Python](https://www.manning.com/books/deep-learning-with-python).
+[Deep Learning with Python](https://deeplearningwithpython.io/).
 
 Let's implement this training loop. First, create the discriminator meant to classify
 fake vs real digits:
diff --git a/tools/retether.R b/tools/retether.R
index 74d2d162c..f35f57303 100755
--- a/tools/retether.R
+++ b/tools/retether.R
@@ -48,7 +48,7 @@ resolve_rmd_tether <- function(url) {
 }
 
 # resolve_rmd_tether <- NULL
-resolve_roxy_tether <- NULL
+# resolve_roxy_tether <- NULL
 
 
 # options(warn = 2)
diff --git a/vignettes-src/making_new_layers_and_models_via_subclassing.Rmd b/vignettes-src/making_new_layers_and_models_via_subclassing.Rmd
index 51035a393..b5296a708 100644
--- a/vignettes-src/making_new_layers_and_models_via_subclassing.Rmd
+++ b/vignettes-src/making_new_layers_and_models_via_subclassing.Rmd
@@ -510,7 +510,7 @@ layer_custom_dropout <- Layer(
   },
   call = function(inputs, training = NULL) {
     if (isTRUE(training))
-      return(random_dropout(inputs, rate = self$rate, 
+      return(random_dropout(inputs, rate = self$rate,
                             seed = self.seed_generator))
     inputs
   }
@@ -624,7 +624,7 @@ layer_sampling <- Layer(
     c(z_mean, z_log_var) %<-% inputs
     batch <- op_shape(z_mean)[[1]]
     dim <- op_shape(z_mean)[[2]]
-    epsilon <- random_normal(shape = c(batch, dim), 
+    epsilon <- random_normal(shape = c(batch, dim),
                              seed=self$seed_generator)
     z_mean + op_exp(0.5 * z_log_var) * epsilon
   }
diff --git a/vignettes-src/transfer_learning.Rmd b/vignettes-src/transfer_learning.Rmd
index 4a8144e76..3f13c2be9 100644
--- a/vignettes-src/transfer_learning.Rmd
+++ b/vignettes-src/transfer_learning.Rmd
@@ -49,7 +49,7 @@ ImageNet dataset, and retraining it on the Kaggle "cats vs dogs" classification
  dataset.
 
 This is adapted from
-[Deep Learning with Python](https://www.manning.com/books/deep-learning-with-python)
+[Deep Learning with R](https://www.manning.com/books/deep-learning-with-r-third-edition)
 and the 2016 blog post
 ["building powerful image classification models using very little data"](https://blog.keras.io/building-powerful-image-classification-models-using-very-little-data.html).
 

From 8f185ae06a26ac9ce0754cfc69cdb620b2d99484 Mon Sep 17 00:00:00 2001
From: Tomasz Kalinowski <kalinowskit@gmail.com>
Date: Sat, 27 Sep 2025 08:19:35 -0400
Subject: [PATCH 74/81] add missing family tag for optimizer_muon

---
 R/optimizers.R              |  2 ++
 man/layer_tfsm.Rd           |  4 ++--
 man/optimizer_adadelta.Rd   |  1 +
 man/optimizer_adafactor.Rd  |  1 +
 man/optimizer_adagrad.Rd    |  1 +
 man/optimizer_adam.Rd       |  1 +
 man/optimizer_adam_w.Rd     |  1 +
 man/optimizer_adamax.Rd     |  1 +
 man/optimizer_ftrl.Rd       |  1 +
 man/optimizer_lamb.Rd       |  1 +
 man/optimizer_lion.Rd       |  1 +
 man/optimizer_loss_scale.Rd |  1 +
 man/optimizer_muon.Rd       | 20 ++++++++++++++++++++
 man/optimizer_nadam.Rd      |  1 +
 man/optimizer_rmsprop.Rd    |  1 +
 man/optimizer_sgd.Rd        |  1 +
 16 files changed, 37 insertions(+), 2 deletions(-)

diff --git a/R/optimizers.R b/R/optimizers.R
index 1b8a884df..a3ddccb16 100644
--- a/R/optimizers.R
+++ b/R/optimizers.R
@@ -1367,6 +1367,8 @@ function (inner_optimizer, initial_scale = 32768, dynamic_growth_steps = 2000L,
 #' For forward/backward compatibility.
 #'
 #' @export
+#' @family optimizers
+#' @returns an `Optimizer` instance
 #' @tether keras.optimizers.Muon
 optimizer_muon <-
 function (learning_rate = 0.001, adam_beta_1 = 0.9, adam_beta_2 = 0.999,
diff --git a/man/layer_tfsm.Rd b/man/layer_tfsm.Rd
index ff3c10712..058847963 100644
--- a/man/layer_tfsm.Rd
+++ b/man/layer_tfsm.Rd
@@ -59,8 +59,8 @@ model |> export_savedmodel("path/to/artifact")
 ## Output Type:
 ##   TensorSpec(shape=(None, 10), dtype=tf.float32, name=None)
 ## Captures:
-##   138303250719632: TensorSpec(shape=(), dtype=tf.resource, name=None)
-##   138303250721744: TensorSpec(shape=(), dtype=tf.resource, name=None)
+##   131671756274576: TensorSpec(shape=(), dtype=tf.resource, name=None)
+##   131671756265936: TensorSpec(shape=(), dtype=tf.resource, name=None)
 
 }\if{html}{\out{</div>}}
 
diff --git a/man/optimizer_adadelta.Rd b/man/optimizer_adadelta.Rd
index abf3d6374..335ede692 100644
--- a/man/optimizer_adadelta.Rd
+++ b/man/optimizer_adadelta.Rd
@@ -134,6 +134,7 @@ Other optimizers: \cr
 \code{\link{optimizer_lamb}()} \cr
 \code{\link{optimizer_lion}()} \cr
 \code{\link{optimizer_loss_scale}()} \cr
+\code{\link{optimizer_muon}()} \cr
 \code{\link{optimizer_nadam}()} \cr
 \code{\link{optimizer_rmsprop}()} \cr
 \code{\link{optimizer_sgd}()} \cr
diff --git a/man/optimizer_adafactor.Rd b/man/optimizer_adafactor.Rd
index d3b64e06d..9de7ce24f 100644
--- a/man/optimizer_adafactor.Rd
+++ b/man/optimizer_adafactor.Rd
@@ -140,6 +140,7 @@ Other optimizers: \cr
 \code{\link{optimizer_lamb}()} \cr
 \code{\link{optimizer_lion}()} \cr
 \code{\link{optimizer_loss_scale}()} \cr
+\code{\link{optimizer_muon}()} \cr
 \code{\link{optimizer_nadam}()} \cr
 \code{\link{optimizer_rmsprop}()} \cr
 \code{\link{optimizer_sgd}()} \cr
diff --git a/man/optimizer_adagrad.Rd b/man/optimizer_adagrad.Rd
index 23a855cb7..51024b063 100644
--- a/man/optimizer_adagrad.Rd
+++ b/man/optimizer_adagrad.Rd
@@ -124,6 +124,7 @@ Other optimizers: \cr
 \code{\link{optimizer_lamb}()} \cr
 \code{\link{optimizer_lion}()} \cr
 \code{\link{optimizer_loss_scale}()} \cr
+\code{\link{optimizer_muon}()} \cr
 \code{\link{optimizer_nadam}()} \cr
 \code{\link{optimizer_rmsprop}()} \cr
 \code{\link{optimizer_sgd}()} \cr
diff --git a/man/optimizer_adam.Rd b/man/optimizer_adam.Rd
index 5ea027be9..9623dd83b 100644
--- a/man/optimizer_adam.Rd
+++ b/man/optimizer_adam.Rd
@@ -136,6 +136,7 @@ Other optimizers: \cr
 \code{\link{optimizer_lamb}()} \cr
 \code{\link{optimizer_lion}()} \cr
 \code{\link{optimizer_loss_scale}()} \cr
+\code{\link{optimizer_muon}()} \cr
 \code{\link{optimizer_nadam}()} \cr
 \code{\link{optimizer_rmsprop}()} \cr
 \code{\link{optimizer_sgd}()} \cr
diff --git a/man/optimizer_adam_w.Rd b/man/optimizer_adam_w.Rd
index 547addb87..17195a8a5 100644
--- a/man/optimizer_adam_w.Rd
+++ b/man/optimizer_adam_w.Rd
@@ -147,6 +147,7 @@ Other optimizers: \cr
 \code{\link{optimizer_lamb}()} \cr
 \code{\link{optimizer_lion}()} \cr
 \code{\link{optimizer_loss_scale}()} \cr
+\code{\link{optimizer_muon}()} \cr
 \code{\link{optimizer_nadam}()} \cr
 \code{\link{optimizer_rmsprop}()} \cr
 \code{\link{optimizer_sgd}()} \cr
diff --git a/man/optimizer_adamax.Rd b/man/optimizer_adamax.Rd
index a459da625..b9b09bb4d 100644
--- a/man/optimizer_adamax.Rd
+++ b/man/optimizer_adamax.Rd
@@ -145,6 +145,7 @@ Other optimizers: \cr
 \code{\link{optimizer_lamb}()} \cr
 \code{\link{optimizer_lion}()} \cr
 \code{\link{optimizer_loss_scale}()} \cr
+\code{\link{optimizer_muon}()} \cr
 \code{\link{optimizer_nadam}()} \cr
 \code{\link{optimizer_rmsprop}()} \cr
 \code{\link{optimizer_sgd}()} \cr
diff --git a/man/optimizer_ftrl.Rd b/man/optimizer_ftrl.Rd
index 8dd8e38ec..7ac47751e 100644
--- a/man/optimizer_ftrl.Rd
+++ b/man/optimizer_ftrl.Rd
@@ -174,6 +174,7 @@ Other optimizers: \cr
 \code{\link{optimizer_lamb}()} \cr
 \code{\link{optimizer_lion}()} \cr
 \code{\link{optimizer_loss_scale}()} \cr
+\code{\link{optimizer_muon}()} \cr
 \code{\link{optimizer_nadam}()} \cr
 \code{\link{optimizer_rmsprop}()} \cr
 \code{\link{optimizer_sgd}()} \cr
diff --git a/man/optimizer_lamb.Rd b/man/optimizer_lamb.Rd
index 8787e15e5..021a3361d 100644
--- a/man/optimizer_lamb.Rd
+++ b/man/optimizer_lamb.Rd
@@ -129,6 +129,7 @@ Other optimizers: \cr
 \code{\link{optimizer_ftrl}()} \cr
 \code{\link{optimizer_lion}()} \cr
 \code{\link{optimizer_loss_scale}()} \cr
+\code{\link{optimizer_muon}()} \cr
 \code{\link{optimizer_nadam}()} \cr
 \code{\link{optimizer_rmsprop}()} \cr
 \code{\link{optimizer_sgd}()} \cr
diff --git a/man/optimizer_lion.Rd b/man/optimizer_lion.Rd
index 876064338..236b99d3e 100644
--- a/man/optimizer_lion.Rd
+++ b/man/optimizer_lion.Rd
@@ -129,6 +129,7 @@ Other optimizers: \cr
 \code{\link{optimizer_ftrl}()} \cr
 \code{\link{optimizer_lamb}()} \cr
 \code{\link{optimizer_loss_scale}()} \cr
+\code{\link{optimizer_muon}()} \cr
 \code{\link{optimizer_nadam}()} \cr
 \code{\link{optimizer_rmsprop}()} \cr
 \code{\link{optimizer_sgd}()} \cr
diff --git a/man/optimizer_loss_scale.Rd b/man/optimizer_loss_scale.Rd
index aa001af7e..f9d57f93a 100644
--- a/man/optimizer_loss_scale.Rd
+++ b/man/optimizer_loss_scale.Rd
@@ -124,6 +124,7 @@ Other optimizers: \cr
 \code{\link{optimizer_ftrl}()} \cr
 \code{\link{optimizer_lamb}()} \cr
 \code{\link{optimizer_lion}()} \cr
+\code{\link{optimizer_muon}()} \cr
 \code{\link{optimizer_nadam}()} \cr
 \code{\link{optimizer_rmsprop}()} \cr
 \code{\link{optimizer_sgd}()} \cr
diff --git a/man/optimizer_muon.Rd b/man/optimizer_muon.Rd
index 2f507f80c..31fcab94f 100644
--- a/man/optimizer_muon.Rd
+++ b/man/optimizer_muon.Rd
@@ -82,6 +82,9 @@ it is recommended to set it to \code{0.1}.}
 
 \item{...}{For forward/backward compatibility.}
 }
+\value{
+an \code{Optimizer} instance
+}
 \description{
 Note that this optimizer should not be used in the following layers:
 \enumerate{
@@ -116,3 +119,20 @@ output layer to \code{exclude_layers}, e.g.
 }
 }
 
+\seealso{
+Other optimizers: \cr
+\code{\link{optimizer_adadelta}()} \cr
+\code{\link{optimizer_adafactor}()} \cr
+\code{\link{optimizer_adagrad}()} \cr
+\code{\link{optimizer_adam}()} \cr
+\code{\link{optimizer_adam_w}()} \cr
+\code{\link{optimizer_adamax}()} \cr
+\code{\link{optimizer_ftrl}()} \cr
+\code{\link{optimizer_lamb}()} \cr
+\code{\link{optimizer_lion}()} \cr
+\code{\link{optimizer_loss_scale}()} \cr
+\code{\link{optimizer_nadam}()} \cr
+\code{\link{optimizer_rmsprop}()} \cr
+\code{\link{optimizer_sgd}()} \cr
+}
+\concept{optimizers}
diff --git a/man/optimizer_nadam.Rd b/man/optimizer_nadam.Rd
index 94f974ac0..326d59c26 100644
--- a/man/optimizer_nadam.Rd
+++ b/man/optimizer_nadam.Rd
@@ -131,6 +131,7 @@ Other optimizers: \cr
 \code{\link{optimizer_lamb}()} \cr
 \code{\link{optimizer_lion}()} \cr
 \code{\link{optimizer_loss_scale}()} \cr
+\code{\link{optimizer_muon}()} \cr
 \code{\link{optimizer_rmsprop}()} \cr
 \code{\link{optimizer_sgd}()} \cr
 }
diff --git a/man/optimizer_rmsprop.Rd b/man/optimizer_rmsprop.Rd
index 10f96bbe4..513bc25a5 100644
--- a/man/optimizer_rmsprop.Rd
+++ b/man/optimizer_rmsprop.Rd
@@ -145,6 +145,7 @@ Other optimizers: \cr
 \code{\link{optimizer_lamb}()} \cr
 \code{\link{optimizer_lion}()} \cr
 \code{\link{optimizer_loss_scale}()} \cr
+\code{\link{optimizer_muon}()} \cr
 \code{\link{optimizer_nadam}()} \cr
 \code{\link{optimizer_sgd}()} \cr
 }
diff --git a/man/optimizer_sgd.Rd b/man/optimizer_sgd.Rd
index 6c4b0c9b6..3c8015a7c 100644
--- a/man/optimizer_sgd.Rd
+++ b/man/optimizer_sgd.Rd
@@ -130,6 +130,7 @@ Other optimizers: \cr
 \code{\link{optimizer_lamb}()} \cr
 \code{\link{optimizer_lion}()} \cr
 \code{\link{optimizer_loss_scale}()} \cr
+\code{\link{optimizer_muon}()} \cr
 \code{\link{optimizer_nadam}()} \cr
 \code{\link{optimizer_rmsprop}()} \cr
 }

From 159480f29dacb97d1e9995819d381fe6b316ee6b Mon Sep 17 00:00:00 2001
From: Tomasz Kalinowski <kalinowskit@gmail.com>
Date: Sat, 27 Sep 2025 08:39:05 -0400
Subject: [PATCH 75/81] add NEWS

---
 NEWS.md | 47 +++++++++++++++++++++++++++++++++++++++++++++++
 1 file changed, 47 insertions(+)

diff --git a/NEWS.md b/NEWS.md
index 5f32df531..ec7e290a7 100644
--- a/NEWS.md
+++ b/NEWS.md
@@ -9,6 +9,53 @@
 - Updated dependencies declared by `use_backend("jax", gpu=TRUE)`
   for compatability with `keras-hub`.
 
+- Added training loop configuration helpers:
+  `config_max_epochs()`, `config_set_max_epochs()`, `config_max_steps_per_epoch()`,
+  and `config_set_max_steps_per_epoch()`. The caps can also be set via the
+  `KERAS_MAX_EPOCHS` and `KERAS_MAX_STEPS_PER_EPOCH` environment variables.
+  Added `config_is_nnx_enabled()` to check whether JAX NNX features are enabled.
+
+- LoRA-enabled layers (`layer_dense()`, `layer_embedding()`, `layer_einsum_dense()`)
+  gain a `lora_alpha` argument to scale the adaptation delta independently of the
+  chosen rank.
+
+- `keras_variable()` now accepts a `synchronization` argument for distributed
+  strategies.
+
+- `Layer$add_weight()` gains an `overwrite_with_gradient` option and
+  layers now provide a `symbolic_call()` method.
+
+- Transposed convolution utilities now follow the latest Keras API:
+  `op_conv_transpose()` defaults `strides = 1` and the `layer_conv_*_transpose()`
+  layers expose `output_padding` for precise shape control.
+
+- `layer_torch_module_wrapper()` gains an `output_shape` argument to help Keras
+  infer shapes when wrapping PyTorch modules.
+
+- `save_model_weights()` adds a `max_shard_size` argument to split large weight
+  files into manageable shards.
+
+- Added elastic deformation utilities for images: `layer_random_elastic_transform()`
+  and the lower-level `op_image_elastic_transform()`.
+
+- Added `loss_categorical_generalized_cross_entropy()` for training with noisy
+  labels.
+
+- Added the Muon optimizer via `optimizer_muon()`.
+
+- Added complex-valued helpers: S3 `Arg()` methods for tensors, `op_angle()`,
+  and conversions `op_view_as_real()` / `op_view_as_complex()`.
+
+- Added signal window operations: `op_bartlett()`, `op_blackman()`,
+  `op_hamming()`, `op_hanning()`, and `op_kaiser()`.
+
+- Expanded numeric operations with `op_layer_normalization()`, `op_cbrt()`,
+  `op_corrcoef()`, `op_deg2rad()`, `op_heaviside()`, the new `op_sparse_sigmoid()`
+  plus matching `activation_sparse_sigmoid()`, and an `attn_logits_soft_cap`
+  argument for `op_dot_product_attention()`.
+
+- `layer_layer_normalization()` removes the `rms_scaling` argument.
+
 # keras3 1.4.0
 
 - New `op_subset()` and `x@r[...]` methods enable tensor subsetting

From ad1543713a820da72364421973140fe720ee6d0c Mon Sep 17 00:00:00 2001
From: Tomasz Kalinowski <kalinowskit@gmail.com>
Date: Sat, 27 Sep 2025 10:51:47 -0400
Subject: [PATCH 76/81] knit examples and guides

---
 .../examples/vision/mnist_siamese_graph.md    | 253 +++++++
 .../mnist_siamese_graph/unnamed-chunk-5-1.png | Bin 0 -> 32549 bytes
 vignettes/custom_train_step_in_tensorflow.Rmd |  12 +-
 .../distributed_training_with_tensorflow.Rmd  |  14 +-
 vignettes/distribution.Rmd                    |  20 +-
 ...l_machine_translation_with_transformer.Rmd |   2 +-
 .../nlp/text_classification_from_scratch.Rmd  |  18 +-
 .../imbalanced_classification.Rmd             |  72 +-
 ...data_classification_with_feature_space.Rmd |  68 +-
 .../timeseries_anomaly_detection.Rmd          | 153 ++---
 .../unnamed-chunk-10-1.png                    | Bin 17824 -> 19143 bytes
 .../unnamed-chunk-11-1.png                    | Bin 7110 -> 15315 bytes
 .../unnamed-chunk-12-1.png                    | Bin 31199 -> 33046 bytes
 .../unnamed-chunk-13-1.png                    | Bin 24995 -> 36351 bytes
 .../unnamed-chunk-13-2.png                    | Bin 5748 -> 13529 bytes
 .../unnamed-chunk-14-1.png                    | Bin 30523 -> 32775 bytes
 .../unnamed-chunk-4-1.png                     | Bin 33191 -> 42503 bytes
 .../unnamed-chunk-5-1.png                     | Bin 24266 -> 35821 bytes
 ...timeseries_classification_from_scratch.Rmd | 618 +++++++++++-------
 .../unnamed-chunk-12-1.png                    | Bin 53506 -> 65538 bytes
 .../unnamed-chunk-13-1.png                    | Bin 53428 -> 58558 bytes
 .../unnamed-chunk-3-1.png                     | Bin 53086 -> 54970 bytes
 .../unnamed-chunk-9-1.png                     | Bin 225954 -> 321147 bytes
 vignettes/examples/vision/autoencoder.Rmd     | 298 ++++-----
 .../vision/autoencoder/unnamed-chunk-5-1.png  | Bin 54370 -> 54089 bytes
 .../vision/autoencoder/unnamed-chunk-7-1.png  | Bin 103099 -> 102613 bytes
 vignettes/examples/vision/mnist_convnet.Rmd   |  34 +-
 vignettes/functional_api.Rmd                  |  18 +-
 .../functional_api/unnamed-chunk-10-1.png     | Bin 17031 -> 20488 bytes
 .../functional_api/unnamed-chunk-11-1.png     | Bin 66248 -> 92394 bytes
 .../functional_api/unnamed-chunk-20-1.png     | Bin 121898 -> 197539 bytes
 .../functional_api/unnamed-chunk-25-1.png     | Bin 270993 -> 400451 bytes
 vignettes/getting_started.Rmd                 |   8 +-
 .../getting_started/unnamed-chunk-12-1.png    | Bin 32193 -> 32138 bytes
 .../getting_started/unnamed-chunk-9-1.png     | Bin 19900 -> 24497 bytes
 vignettes/intro_to_keras_for_engineers.Rmd    |  22 +-
 ..._new_layers_and_models_via_subclassing.Rmd |   6 +-
 vignettes/training_with_built_in_methods.Rmd  |  34 +-
 .../unnamed-chunk-26-1.png                    | Bin 118726 -> 206413 bytes
 vignettes/transfer_learning.Rmd               |  12 +-
 .../transfer_learning/unnamed-chunk-13-1.png  | Bin 373610 -> 375562 bytes
 .../transfer_learning/unnamed-chunk-17-1.png  | Bin 331799 -> 333527 bytes
 ...g_a_custom_training_loop_in_tensorflow.Rmd |  18 +-
 43 files changed, 1027 insertions(+), 653 deletions(-)
 create mode 100644 vignettes-src/examples/vision/mnist_siamese_graph.md
 create mode 100644 vignettes-src/examples/vision/mnist_siamese_graph/unnamed-chunk-5-1.png

diff --git a/vignettes-src/examples/vision/mnist_siamese_graph.md b/vignettes-src/examples/vision/mnist_siamese_graph.md
new file mode 100644
index 000000000..341f4fed9
--- /dev/null
+++ b/vignettes-src/examples/vision/mnist_siamese_graph.md
@@ -0,0 +1,253 @@
+---
+title: Train a Siamese MLP on pairs of digits from the MNIST dataset.
+author: Mehdi, Ivo Kwee - https://github.com/ivokwee
+date-created: 2020/05/30
+last-modified: 2020/04/21
+domain: vision
+category: intermediate
+output: rmarkdown::html_vignette
+knit: ({source(here::here("tools/knit.R")); knit_vignette})
+tether: https://raw.githubusercontent.com/keras-team/keras-io/master/examples/vision/siamese_contrastive.py
+---
+
+## Introduction
+
+[Siamese Networks](https://en.wikipedia.org/wiki/Siamese_neural_network)
+are neural networks which share weights between two or more sister networks,
+each producing embedding vectors of its respective inputs.
+
+In supervised similarity learning, the networks are then trained to maximize the
+contrast (distance) between embeddings of inputs of different classes, while minimizing the distance between
+embeddings of similar classes, resulting in embedding spaces that reflect
+the class segmentation of the training inputs.
+
+This implementation loosely follows Hadsell-et-al.'06 [1] (see paper for mode
+details) but the euclidean distance is replaced by a subtraction
+layer and one fully-connect (FC) layer.
+
+[1] "Dimensionality Reduction by Learning an Invariant Mapping"
+     https://yann.lecun.com/exdb/publis/pdf/hadsell-chopra-lecun-06.pdf
+
+Gets to 98.11% test accuracy after 20 epochs. 3 seconds per epoch
+on a AMD Ryzen 7 PRO 4750U (CPU)
+
+
+``` r
+library(keras3)
+```
+
+
+
+``` r
+contrastive_loss <- function(y_true, y_pred) {
+    # Contrastive loss from Hadsell-et-al.'06
+    # https://yann.lecun.com/exdb/publis/pdf/hadsell-chopra-lecun-06.pdf
+    margin = 1
+    margin_square = op_square(op_maximum(margin - (y_pred), 0))
+    op_mean((1 - y_true) * op_square(y_pred) + (y_true) * margin_square)
+}
+```
+
+## Create pairs of images
+
+We will train the model to differentiate between digits of different classes. For
+example, digit `0` needs to be differentiated from the rest of the
+digits (`1` through `9`), digit `1` - from `0` and `2` through `9`, and so on.
+To carry this out, we will select N random images from class A (for example,
+for digit `0`) and pair them with N random images from another class B
+(for example, for digit `1`). Then, we can repeat this process for all classes
+of digits (until digit `9`). Once we have paired digit `0` with other digits,
+we can repeat this process for the remaining classes for the rest of the digits
+(from `1` until `9`).
+
+``` r
+create_pairs <- function(x, y) {
+    # Positive and negative pair creation.
+    # Alternates between positive and negative pairs.
+    digit_indices <- tapply(1:length(y), y, list)
+    y1 <- y
+    y2 <- sapply(y, function(a) sample(0:9,1,prob=0.1+0.8*(0:9==a)))
+    idx1 <- 1:nrow(x)
+    idx2 <- sapply(as.character(y2), function(a) sample(digit_indices[[a]],1))
+    is_same  <- 1*(y1==y2)
+    list(pair1 = x[idx1,], pair2 = x[idx2,], y = is_same)
+}
+
+compute_accuracy <- function(predictions, labels) {
+    # Compute classification accuracy with a fixed threshold on distances.
+    mean(labels[predictions > 0.5])
+}
+```
+
+
+
+``` r
+# the data, shuffled and split between train and test sets
+mnist   <- dataset_mnist()
+x_train <- mnist$train$x
+y_train <- mnist$train$y
+x_test  <- mnist$test$x
+y_test  <- mnist$test$y
+x_train <- array_reshape(x_train, c(nrow(x_train), 784))
+x_test  <- array_reshape(x_test, c(nrow(x_test), 784))
+x_train <- x_train / 255
+x_test  <- x_test / 255
+
+# create training+test positive and negative pairs
+tr <- create_pairs(x_train, y_train)
+te <- create_pairs(x_test,  y_test)
+
+names(tr)
+```
+
+```
+## [1] "pair1" "pair2" "y"
+```
+
+## Network definition
+
+
+``` r
+# input layers
+input_dim = 784
+input_1 <- layer_input(shape = c(input_dim))
+input_2 <- layer_input(shape = c(input_dim))
+
+# definition of the base network that will be shared
+base_network <- keras_model_sequential() %>%
+    layer_dense(units = 128, activation = 'relu') %>%
+    layer_dropout(rate = 0.1) %>%
+    layer_dense(units = 128, activation = 'relu') %>%
+    layer_dropout(rate = 0.1) %>%
+    layer_dense(units = 128, activation = 'relu')
+
+# because we re-use the same instance `base_network`, the weights of
+# the network will be shared across the two branches
+branch_1 <- base_network(input_1)
+branch_2 <- base_network(input_2)
+
+# merging layer
+out <- layer_subtract(list(branch_1, branch_2)) %>%
+    layer_dropout(rate = 0.1) %>%
+    layer_dense(units = 16, activation = 'relu') %>%
+    layer_dense(1, activation = "sigmoid")
+
+# create and compile model
+model <- keras_model(list(input_1, input_2), out)
+```
+
+## Train
+
+
+``` r
+model %>% compile(
+    optimizer = "rmsprop",
+    #loss = "binary_crossentropy",
+    loss = contrastive_loss,
+    metrics = metric_binary_accuracy
+)
+
+history <- model %>% fit(
+    list(tr$pair1, tr$pair2), tr$y,
+    batch_size = 128,
+    epochs = 20,
+    validation_data = list(
+        list(te$pair1, te$pair2),
+        te$y
+    )
+)
+```
+
+```
+## Epoch 1/20
+## 469/469 - 5s - 11ms/step - binary_accuracy: 0.7568 - loss: 0.1640 - val_binary_accuracy: 0.8796 - val_loss: 0.0981
+## Epoch 2/20
+## 469/469 - 1s - 2ms/step - binary_accuracy: 0.8892 - loss: 0.0870 - val_binary_accuracy: 0.9222 - val_loss: 0.0633
+## Epoch 3/20
+## 469/469 - 1s - 1ms/step - binary_accuracy: 0.9240 - loss: 0.0598 - val_binary_accuracy: 0.9398 - val_loss: 0.0490
+## Epoch 4/20
+## 469/469 - 1s - 1ms/step - binary_accuracy: 0.9433 - loss: 0.0452 - val_binary_accuracy: 0.9485 - val_loss: 0.0395
+## Epoch 5/20
+## 469/469 - 1s - 2ms/step - binary_accuracy: 0.9569 - loss: 0.0349 - val_binary_accuracy: 0.9609 - val_loss: 0.0314
+## Epoch 6/20
+## 469/469 - 1s - 1ms/step - binary_accuracy: 0.9643 - loss: 0.0287 - val_binary_accuracy: 0.9634 - val_loss: 0.0287
+## Epoch 7/20
+## 469/469 - 1s - 2ms/step - binary_accuracy: 0.9699 - loss: 0.0243 - val_binary_accuracy: 0.9652 - val_loss: 0.0276
+## Epoch 8/20
+## 469/469 - 1s - 2ms/step - binary_accuracy: 0.9745 - loss: 0.0210 - val_binary_accuracy: 0.9680 - val_loss: 0.0261
+## Epoch 9/20
+## 469/469 - 1s - 1ms/step - binary_accuracy: 0.9769 - loss: 0.0190 - val_binary_accuracy: 0.9681 - val_loss: 0.0260
+## Epoch 10/20
+## 469/469 - 1s - 1ms/step - binary_accuracy: 0.9785 - loss: 0.0174 - val_binary_accuracy: 0.9706 - val_loss: 0.0236
+## Epoch 11/20
+## 469/469 - 1s - 2ms/step - binary_accuracy: 0.9811 - loss: 0.0153 - val_binary_accuracy: 0.9722 - val_loss: 0.0229
+## Epoch 12/20
+## 469/469 - 1s - 1ms/step - binary_accuracy: 0.9824 - loss: 0.0144 - val_binary_accuracy: 0.9723 - val_loss: 0.0220
+## Epoch 13/20
+## 469/469 - 1s - 2ms/step - binary_accuracy: 0.9832 - loss: 0.0135 - val_binary_accuracy: 0.9739 - val_loss: 0.0212
+## Epoch 14/20
+## 469/469 - 1s - 1ms/step - binary_accuracy: 0.9845 - loss: 0.0127 - val_binary_accuracy: 0.9753 - val_loss: 0.0204
+## Epoch 15/20
+## 469/469 - 1s - 2ms/step - binary_accuracy: 0.9860 - loss: 0.0117 - val_binary_accuracy: 0.9732 - val_loss: 0.0209
+## Epoch 16/20
+## 469/469 - 1s - 2ms/step - binary_accuracy: 0.9871 - loss: 0.0108 - val_binary_accuracy: 0.9763 - val_loss: 0.0197
+## Epoch 17/20
+## 469/469 - 1s - 1ms/step - binary_accuracy: 0.9871 - loss: 0.0105 - val_binary_accuracy: 0.9737 - val_loss: 0.0203
+## Epoch 18/20
+## 469/469 - 1s - 1ms/step - binary_accuracy: 0.9877 - loss: 0.0101 - val_binary_accuracy: 0.9742 - val_loss: 0.0208
+## Epoch 19/20
+## 469/469 - 1s - 2ms/step - binary_accuracy: 0.9881 - loss: 0.0096 - val_binary_accuracy: 0.9765 - val_loss: 0.0201
+## Epoch 20/20
+## 469/469 - 1s - 1ms/step - binary_accuracy: 0.9895 - loss: 0.0086 - val_binary_accuracy: 0.9746 - val_loss: 0.0202
+```
+
+``` r
+plot(history)
+```
+
+![plot of chunk unnamed-chunk-5](mnist_siamese_graph/unnamed-chunk-5-1.png)
+
+## Evaluate
+
+
+``` r
+# compute final accuracy on training and test sets
+
+tr_pred <- predict(model, list(tr$pair1, tr$pair2))[,1]
+```
+
+```
+## 1875/1875 - 2s - 845us/step
+```
+
+``` r
+tr_acc  <- compute_accuracy(tr_pred, tr$y)
+te_pred <- predict(model, list(te$pair1, te$pair2))[,1]
+```
+
+```
+## 313/313 - 0s - 1ms/step
+```
+
+``` r
+te_acc  <- compute_accuracy(te_pred, te$y)
+
+sprintf('* Accuracy on training set: %0.2f%%', (100 * tr_acc))
+```
+
+```
+## [1] "* Accuracy on training set: 99.60%"
+```
+
+``` r
+sprintf('* Accuracy on test set: %0.2f%%', (100 * te_acc))
+```
+
+```
+## [1] "* Accuracy on test set: 97.81%"
+```
+
+## Plots
+
+
+
diff --git a/vignettes-src/examples/vision/mnist_siamese_graph/unnamed-chunk-5-1.png b/vignettes-src/examples/vision/mnist_siamese_graph/unnamed-chunk-5-1.png
new file mode 100644
index 0000000000000000000000000000000000000000..faa72326116002fb4e4dd1063609d129eccba984
GIT binary patch
literal 32549
zcmb@tWk6g}vn_ZC0h*w}T>`-^xI4k!J$P_;O(3`gYaD{RyIYVzfZz_n9fCW|Nxr#n
z=FWTf-~4DA=ssufvv=*PRjXEoDJe*zydro7004@#l(-52z})<OL0*9Wp_Tba1HK_T
zNNGC*020RE7tF-6-va=U0n*|kY98r_%kDm^>i58(ou$~caVpzMk-Zl~Hri$sgmC!b
zG&O4I=*mRg9_r$HWOPp#>aPuOY3RZSm9Sv6Yc%J*>+9<Bz4Pa}t}|CjVHlLi7!kRm
z#;$XH7B-jCsHykj<DJLI5>$jC<XBC3`LN*orf%}?Yf+jmM?pTe@<F-Ae#-ayelIPy
z_nXh%Q)seSZuTS>gTpnSWPETSqL9bd+X4w@T_i>Or{j%2N!ui5K*@Q(X8AOx^R%D7
zT^3p)<E(#Mx*5-cE(*CVl2AOtjV2x`{<<mC_@jBRAgIAryNBJjU_RA$c%<Z^6qdjW
zrpvfmDoz|?q&<WC$uralPo}sgshB4^-b@?1m4<tX4>fx$y*BnCb4Q;ksiyMF`lT>r
zjavar0?q^wDnbZ0!V7k6*2^Rng^Yh>??5VU_#>S|y$~#Lg62;2+%{o5Wdofe0LTcU
zK5(SbSKLgCipT=}F{J+6KU04F`sI2sbNhxxjqsGe(gorNbI>iHui|N-bqWp7`lC<!
zs2-{?uy6zjA904fFGOlt*DP<1@LOr~JjU^LbUZzJRrcyLnN*xwd2rzD-lbPa@0tTQ
zR-{Pbq&5Q%94rt}RaJF%=4AG)hh2#PHM{ikwX`h${4RiWDv%69mspaM({nDS1w}zb
zJk^(~@3WP?kdw<HKT!L$^wVKGa_P6Aeq_m6C2ptgo(u0anuLTzBYhR!FIg6N|77!Q
z54<Pzt}XvWl-!x=!1cs!YCFlCB!x$ASKnOx7@@h-;oDk4fNJ}RdP7^^Wy{bpfg0S9
zSEpl&{ARirbFLc&H=3>h3qh2bnc1AH2@9T#j0{#J%)>X5mm=sdyG>a={UF0N+1c5D
zm$a*ZT@(VxEq4K$&2PKKKCK}15+0eonR49`bLW9*V&<3yVUivk0Z}QKzP`TcmJ3$T
zv(1c?4*iFbq}*0_WuH9D@<2wfpckqd8c;GwY;0^Msd2RyD;wLvbZOzVl>k2_2Z!ru
zI!_#B6d_lI`^id@JvGy7Jtlzc7zQp*T$~yR7*^NNx|z4gn&wDKBO2|-%#<oKL`%n|
ze#8ht0Z&xy+a`WYtvK<BTXGADN+x2Mu5UOR<vN#tdeOX}+R&-If*$lOEiJvgywubZ
zbjq*t+8Bn%1g_KXZj4%G!}llc7wRn4v2qf)56OnbvAVb3FrEH9-G~r+_$Fa`6I^qa
z9V7G_7gH=9^{bf!FC_o_9wqYl+@YG&Yw0&kuTAN+86681ndR+-fuTnGUnJ6I5OUP}
zsNTsJZL@VxeTLD*0VW7`cads9JI4BwkZ!thJ=MUQuC+w<p0z}~aI`{LA_K%fvxfeR
zJ|;S8yssB)orQFsbVWr!x4Oy^p`vGe-+IxS(Imd^^r=@woZ7M3X*3_V@YoKQ7!C+I
zLn3Uj%BOls8OD7m9@n+ilH6@i4b5*GdK5F#*5)2y34#fz)$k4s4pV#EB>RI72I7bt
zI4mp3!k}KPqN+MPn$D}I)Ff8^e6?r~C-Ude$kF31Ky^oFl;hqjezM)?mCR+ovQTf$
zK`^q3YiKi__l}$B``$}sN(F=smw-vcoOtex??%)L)S(LKeJ0+jx^R|=<T;dK$fa9a
zf4pp~7`D{&v9Ys>WFfJ%0!H;*@(KlQ;jGAi=sw^Nm3q|c<R#QB^_9tY(fu%#_rIUN
zD(4CIv{*cLJn10C#+H~S0KO3+FdzZPhV+YdmXl+X1{q%`jK$vL`0p%_2|c%k7~$aH
zEZ8*6&1nUnUaY^N^5T-aQ122y2@YE+AM&E1p}E=3jd5~vf=3}>_Zfu!(1_VeOyt(T
zMuy7AefZsuS^>e4J9KhQVRw7|O?4HEUTJ|aqK6Rz{E$=Exf*(argAiFeX0|`nKte~
zxj|?4yMg44QA}hZ<BMEz^Z+xnM1Tl^U(pl0iUIqjyUDz7!*r?UGpF-idlW$#cLwLh
z$e!>0)ny(DxYRqN83#_yB;JSB2JK#+c(<_>k_~pt3uW4DV`wnxEUri_<C-v-?Tek1
z;}JPGXT!|6CrjormCyU7l?8QHuCyY2-FsxHA*lhqrf=NR?%mum6#2X;MSf2q)2j@5
zFLkxynDHo`s=d|wIZKDFjz|Vc&{R%*<uM%%i!jk(Ye|IBa;3NZi?)(ePNsk#?bvwM
zU$1of4m3N&&%YY&4WrL5RP=Oo7CylufTv;dn@FJbiTcB1Kv+lre<qf;?mwsWDBfwm
z+7To4G`Ht+bueqc)Z|Du{N0Yijc@MgyOBjYYSm&`FFYz}8yF$5>CtUJk>x#8lyeIb
zcxfSW@{0?|VsM{c5zs-tjSn9Clw%ZqbKBX9Y-<zheA+K=U_@eU;9_O^nZk9_|KqMk
zwNL@{Pj9KIB_$<iSvc*L{xchgDU_1(&|!(6*x!w*G?1!w@Jmy47|!fFnXJb{&7`yy
zQnx$Vc+dpNjbVgy;Q`T5CE6dV(e4&jIpc$L(|z9=i<X~9AGa6Zxr+8q?cat(e?cW#
zxbC=TGO%guNY?lY98yH!;79c*t$IH%j2;|F%gF36HP^0NFMnG=!W1|&EG{laxH_~X
zA)G9iaPGHxgH6@1Hkv$=my@16qK>+8kb!>iM6Sm8;5;6Ou4HYR7+%6bR#dEWbWG>y
z9+zjfV{^B)LoQE{$-tyPTvFKac-7a+RA2APn{k;6wK@6vhkDg_)`rvl*k|AuIhxA)
z_;o<*Nyq(h>q&*SEt#6hoU1C_VT<GWv8HR`B;u?Mnz-JZxx>g_cusng+E0CQWg7h|
zB0smomym2M!wi~?*iz6-rj^T~RxULPk{k;X^cu)41fSJQVpOkJ@6V?0q<>5;BX?$(
z>o!*C)YCCB*&WVR&mQv^dL+SB80-b>b8!SSdTkQc$bk#4*F9JE-D<wp&v!2@7e0}W
zj;X^@?=aubnXuV!R)z-98#?nuwGi*^SF23=e92CeypN)WOL_3QSixPL)V=k+w5rf;
zULCvt6dN;@OG|HW_4;xP7=_Dqqq0g#=@=OFYD~j)%1L-#J~}v5O|0pCQn~9z6KaxN
zmP_Y;eeKeYQYB#sXEWaFx$rsOte`D{mf?eJcuL$G6@&idp|Iwjb6TE+9g*C`X{%rG
z%H;|(dmokl#7&njYRkz-Wt*W6<WjTYGU<89DsJO=RKNB9-SsJ7Hi;mr-*Kbgv!hSn
zRrg|RYYT;x=oJMiOi24qziR{z<C)Lz5fqP762FJjfgLoGU?U16`~9ZK+FJ9Gvp>Q%
z<Fj}TyULddP-w@~-9gDl$fDcsjZ}<X>bhtH*@>5`N6~vRR}t*Ebk{qz^H2w1F4I?J
zIbTv`u|cgo`?K&g!LN@rx_$Rm){pNMg&tjYua+e^7Y|l_b`~wvDG<;|evXVBf#w%N
zd;c?q&3DxYiAoR_4z3uJnn|Z_@PK^YbpP8~@N(<g@zbwt3czOQ$L;-QaeowHD^h5J
zPQ^J418f2}(U&h@UcY{=?#EK3AGhpx=^C4plfz*>)1+6?;d>7*|8<FXJ#uc#_I}?9
zwPN?5*l#n!jmN8pOHWhzU+e(kF_y##O+lZ>yGsh82MC01FIu6p^XYEH>V<x{qVC1?
z!t?9_cXVz79iOyh?s$_PlQNdQ!IZ^Ik*0^l&!KNN&DnV7B{iN?eqv;{J1LR%{`$&7
zfG^!kVJSs39U?3RnGOL4U;-xhh<IIg<H-p=nF*~pe;BV25D^i{4C!Wx^hw(}Ff4Ww
z(7RgkK52K~XJG#b9Lyf;+=wkFCp$^!V9CW%!fKh#lMe$*RMNojoDa}EM%@Mm3I@t4
zSYYnqIzAA-*_)A}hC$^Oz?x6Bx6Z~P8+T8bua3obc6R92%EYbZTI^_*XjbTR%G@Xl
zecSm!e<k=z80dI*n-_q+yV%?K@jB?MRr~yl#4M`2ni^I((u^HXZ76nD6m59U{fmMZ
zrYQ-DCBwsdxHVEDf~sE?Wg^y68;APi&MMHNa^6+fyz3EIee{x}nq2RTkWeNP^l2LZ
zc)^bz*sY|jyylrB<g_`6K<Yylz7Bd`60hBCLR`+_8g=MM&cS1|b5dS@BN|fcNp`H!
zxa(1k9+R}23^CRa<qwHaZ5MT^ILH*`4^-_2<y2)oWRY%MRmG!h<4~xYfYYiuC*RZy
zba+#3i;B?D(54cAT(b}<q&L=@Q%s2k2$&|IEvnGDnxOg-ZN0*opprJSZJO+_QCkdM
z=6jC$a=J0#tpYW|Yd0{AmqcXnOWm!Gd1|;CA7)k6=)u8e?fa5f6+p{mEE`dI6qRjN
zuSRV?+>D%>GS0(!2Oc)(MOVMIFStIqni?82<GGTU7#MMJadvtWls~rO)V7^2{ft=o
zIZ4=c=9>KUVva?|T}Ntcpr*A9x~9!?aZ5s4X17j@^R1upf+@sD&_AIAV4?v0MjU^K
zoX^h2$XNRgpjjHPbL^F^&2m9rLxcs8h5zL5>F@@xA;JXw$QljK?u7wt8jcS%6jRv|
zD-I_egmUgJP{f#4nZk2|jXv`m?4HI-lH48$RlXSbBzt&0UPkJ7vsUtA_G%m#8=Her
z2&Q1{W$-7ol|^1xnaKkAQf2K~nxwp;+d(STo&1*^mJ@lz?pJUbs83uL*N2KF>8;+3
zhq>;(f_MHDZ+KH4_#4P7cL&K&eay?coM%yCE$|>Hm{G6U->AHbNuW3w3`Qr*tEch1
zoYu@bSzfHxJ+Tn5R-gU}ZTAlG;V1OE4nE=|wO#FSy*@UJQ=|HkAE&nOlsN2^LYdS|
z8PAZU{v_vh)f^$TG`~X#&@f!>cAMU=?`EW@M@JEq;P-d2SOqJyiI1^#f9`4)ci%=?
z(;^9llBYfE+|L)tqmzt3`SdSo*NTdXT_Ag(OV?j+l<1=AGMhfS31Y5eeprOA!28CK
z#c*l1thtw2qmT+dR9>%s%5?APzoZ}`-ugklwzg&%>2>oX$?&D<3gKB7qq!%*qKEG+
zmA{J9c71%k>Tl1Q0^i9s*~<{a)QjVb{h~%f(ME%lK7K@S%sk5i>AaEvV<QJEa*h3s
z4Ci|#lO~(Uuf35v?N4604Cx9xmoqQjho&fpskBSJZn8={sZ=S?Q~oe7t=Eg2B8ZYM
za40@sSgzD(8a|E)a+6^MZKvYrGRY)sWPKTy3-1T5LGIaHgZ8GXUu40E#wp+rb3EAL
zeLSjAGxngMsxqii9kTRtjoGfdP}hcHR|?y;{p!QCrlG#;!a)@0kKy6ru`&CZvJ}N6
z$+6*K@B5?1By%%Gn|M~+1@<h#nJ`RhR3h%Lad8u6+O^gveU2CPxFB+Vsgh$6{;H{N
z@?m~b&Q4)V!L4Kli$!nzHZsqF0w%fI_B!LFZ8b+CVrW3)HY6h@h37VrHqBLpc5Uri
z;56th4UMD=c5|MwG@V0pxE@p5;gLJmm)|`P8kYwlRdO`L!45W%UOX+%ATtyOlctNx
zOd}7`iHQlphd*Mq^Zdr&LqP=Rbpt5{6QE`%Gf~l=?_rpgM-%G&-D^R#AUfs<4(oUY
z^L00@SkLu>%>9r>^=j-mwPuBE;BAxfMt-u#JYFUgdbL`eyMuPPk&tK9`NVT_!l)Ug
zsAO4hF+5zGZVXlZiJ`wEaoVQJ|24@T>DA}7ksE3N9*SZfnrg(Ibnv|yIWz_VkWwU6
zVkYP3Ut0Dow+L;el$2r_h6L?0akR@a>+SNL(tezSjAm+|4ao|4UzKUAFY)$a68%H)
z9KpU9pCq8o;3BR)>b<|~8^sTf-S2N9;B@4$c@q+>&Xks%JhxVjST6LOu3nmzd}&1Y
zR{O1)b`HHR))-%(jq*=-Y4ylOXT{n)ASl?>Vs)CICwPZ~duFiaH4$OAqgU{FZCS^}
z%`k;z^o2nye7_<9a98=uRTXFsZ>9+*dR<~TX*}rY=@DUIe2%An-GZIG3mcT5V|@QC
zc(7k1QOty@9lt+?t=sULkPB`^h+;)A!JNBK6?K58(4(fOZ4Mni%O8bEC<T4+XUA3D
zOS15tbq}U!&A#Frk4(M}$KD56glu0)t=j#O3U_`C?^$^h>zOj`lcmF`(&v|nC5zdQ
z83vh$Yw;?@Y9nb}nm!w^gRJ=EF6y%woCjjM!p+3gh>ARFRewU>apLi+gh##z@JTu-
zns@qcE@bSS?#4Trp<uBkcoB?CE%zed6>mbgG-{`r8=JIHrO!!^B2e(>TLXdG({6Hi
zf{Xj>6MLb<?v-|LZw#^l$5r3!Wm5l#HK*p|$0u)_ObVAYqnyg+xx@2)SASNCF9g^Y
z#`9zQTp1aL#!E?egDirVKBI!A?Y~p$QHaP9={l;vO+cYciw(qH7cYV2)0zI}EPW2i
z_k}nNP&y{Hi@nK#UeB<kX3g1>8FC1@eYf~YOoPq**+FGza8MBc?Pk2Pva&jq+^x0B
z4A$R6NQv%MPu{DYvFjCk6RnSsE&W!-G3GWsR*gO`O(e!CQPuiiQu#7!#cIc=E2r-q
z={0>=w}gH>J+6kX&5j0%L*U@$F8p@ps!hZM(_qKiA#iX5nZ8#^hJMWr$nJMRUXS~%
zh0f|OJ@zPRkHn@dh8<Qi!W7aOD0f#n@_gkji8{nX`tIvLtt;EFYG`U0hL?IO3L)e|
zAz>>gHme=P@Ti}^DnSq*({Mo&rKg833K~nfSb)-rjr|CnvMYTj>QbpIS{!YZgLEFK
zklEB|cfoeOv~k-}=rb}yC<2)#N#1=_$bBF6p^x-eg}~gA(pP^^L_OU6v{w=SJib=y
z;^`E0-M%LLV`%IzFOg8K3df7$2t;BkhTx*n&s6_je0{vE+i^a+mPAb$Lqc->lMT5S
zt0fQat=TFz&W`FU@-^TWunQZz27B2(GW~ht^i9wMF_l+mt<%;}b!;}~VqIQ4ughxn
z`NM6xAw9z}$HN8F>8H5QM)GM5rMit8jL{M3Dj47es#;pfSjJK)h~9-p=TlF;s`BQ5
z6lao9yrWlWxCQ)@<TTg(J^^_U9Y`Fw8y_FCDG;hHvBCiX6CGp%jrQIrIX;?`+gC8x
z%0wiA=p|I(wrnOZzhmok$mm05RE6ZznorvFPumT90qDB*)z|nQllkA#MR#>Nq%TH)
zPCM_Muqq20{h5quKzR!Qb?bxc&x|}r-{h8j6=N)MRsP%)s?IL-um^>;I8Q`88b3{=
zOd}WG_=${Hno;2gz~B|3d6kN^7w~abr*Q!RwffYOV}i+BuSGdFBE%I(iih0>gr53g
zvs;BSo9W6gdr`E3+lLZECKun^OQxky%z|3ZOlHaDQj;fJjUoNdrI`G}*|K+?pL4$C
zZyg_H-@iOChzOU`K<XYq>)l2He(z3A&yVE3HA5FIlIf7TaI$Nm&9>EtvdE`dHkk8$
z%f=BN8cQ<a-GhAfLNt|cnI6T($#^_CpRcIWfIU9Q{02TF92{H@4J&f4#)TI}FH3?-
zLn>dj`j0)ACy*WB;o-Sna#nOY!?r$OpnWtHIi-e^FxQ3;P~bfx9FNU#(E&b8j&_j_
z16_#|2K(~^NRCqw^6&JC4iQ*c^`#yHRScyb$iERTyIAC0O_Ux+HmH3jbGT%!PyqC(
z_f@HkSamO(ZtfmPTPNUys?*%l-_^76?Nikq&#gV7K2$W<`a?W-!Zcn0av8iF0s;b*
zl&14FW`Yno7?XkMN2U^{DcSVPGy3~&A81_M3jphbkzaS6v$i%W0zkP}cQPB(IpE=V
zM?WIST%q-%<|k$SFH~Qt(p^4vbS#oWMM$&TaN*(@>HgZ;6c5~Vi&qz1#j;QAZrJ}r
z$?n*d#q&O95nL?R+^$;f56mLX?1?VM;GqD3K+Av)uPRMrxEZiLlC~o#@!^o=t@MPu
z#b{6R01VL2e>JHn)Hg6dS~qVv!Z>GL4<pPy(1C&fBvteY7IgZ^fShNF+nUe1%hx84
zfp6qn`T1OakSrA+N%Fs}X?Ue%rH8<8R8_mCUbU#(och7<lms;j*n=PVoZs91>T18?
z<{(yNw11pCA483Bt3n|xQDjWLJeQ^pQ1i&1(xPT3>olxpK;q<U5(fg#Q$(sA$96WC
zGoa10R+tb^!Jei<<-m7F<V*RbW#7#w8MLL!aOUw)V<HQQDntX(&t9z7?JoW|mKT3-
zDn@mdD^-x6;jy-{5Q#uOkX(_fh+S{db+zre<CR{pfeFHa4|=en7OU&fuO#X~rlw1M
zcfX~Az~VY0;O-#Md@4*%Nk3)at@Qc5N<m%F{^v3=l2<cUYfyK1NeFq@NO$RG-$=J&
zCjagJgFWhX^P(CT4nXj##636H3(m5VvDlz7T&O9jCf)V>$7Z`jMmyAF>~Mc|gu@mX
z;IO6nWviVPGPT_8R|zQTLIZ{S1vf5v+iu&e8fT%mW=F}|pOQ!Kn?UbR=-!>)n=RFi
zLTW)u0McNGV`tS5n?0{>J&i&pCedT2pO0HHSsRVE#jL}%I}<^jvo8X2RW&tLh8Ia_
z!9oTYse-fopFF3EltH-e9P_m~(@EM(klJ4By<5XB#?S4>XM_5WJN=2eZY~EiM+-f2
zmSN^wtY6JEH_S(w1bnf1MRyI6G208lV0?ZEwY8?|JToh8c)lI;uh6uy{nETRA4w`Z
zOxqQz)H@T5*bMhJCUd02WCCfq)xEQX<#k3Xb|-qPet4M)DGMGhk1*u+#nrgC6-bji
zjFJ8J_z0~#yN63qaWXUD96mc}hC`Aah9T4F^H<|4asP@6nEy<6Y<GTT{q*4#Qqg-8
zQvS$ws2_xU6(lr^H#be^+Cg0Yt?HX%)o}?!=YW+r9Us>uEk=N{5xKpD#LHT)+lD-u
z<sVu6@IiA3f@FjVSr?(`uhf!v1vqTxs+?=hiIY#?J2?EYX0tHp(9gRF{JcIEE<-6p
z!E+p+%<KEvUXU?BAdB2)9sr^%+Vw6meet-!^*{sH146D^AlJO#f1bllPVb&I4G)0f
zgtSi_UqH(-Of8w+zlphv)79;8J>KE9=q?5bGp}2V*T1Cm$5UG4Z$GvuQ;6~FddosN
z=#wx@B~XzXuUpZWyWRx(pkVsS&(UT;b^qu~;{%kXUn)2DIWg2oNJ<wv!bSuYwMPX%
zLe#w&g@b-|!cg5=4ZDP!jO`Fr^8vr#nd~mQzIhp=yLEVrY&3=nI4n&YhKHURB|yks
z5UK9=wKa`A>G&Gu;w~7@Eth$(@jeR(LA4lD6Ng2heN@sp-Y51!=w||Q7nwOg7B<e<
zPv$rn0>l!y(|1t<#}zQZH?{Mf%J0Tdu^uA#)q`3V@X4&e#WarFq^@ButE>XL0?F@U
z_TQ_!Z%`Y}a&l5+)RaATAXHnEkh}01<#aZ)hd58&rn8|IhLca`z9C`L@SaxXJo#+3
zg9A8R98yTkoQgz(bJb7YwO>DZH5_!f#47!2WA4_`wROKlOXjIRR95ZzmL9A!2`S52
zo<xv=Gk|wct|h#!X>!@TO=p$1s=%4yl0uIrP2+Li1{1Cm7hVCOuqsDJRTVi70$qP1
zju>>6fUJhyX{-9rF{6$|X33h)XWQ5~uwB2EdXp|UV)6?79v^B|TV1$4?siK<P)9QO
zX5&?3YYUru?_q>VShQGzy-8q)Yxu`7C4}&_pz2p&957#sK+WJ_mD)(~g2{tuo^)so
z(1-@!A*+_-B|IXXk%B~lg!0ASZKLfXr}x$2rCCAa3GvNg?U;dU<(CO%i_vm6b@YIw
zeepI7Y!OP!zHsbH_oD<Xl9@Med~R!m_``o$jX!i?(YV&zY}ia;m7M1-E#lv`kl00T
zaGSw8U`i;@i-Rt>iUFzPIk$<n@8A=9PmN8z@rM$gk8hN2Epx8-NV1udtbY;O_uP|y
zCTqLx<e^<`a6MkSJ!|*OIfyND4k_cr7JDKj=HQ<qlf?kUB_&B$U8!jKf6}$2$8Aan
zrA~KZMrflV)?e?-mfl9v<kv9Jw`$VmtIY#TRU|C%Tp;-qlMTJ&@>{~j!4b6~Y8v2f
zP~dUhpW14TUL^LNU_;Zk%_eH4BzN;DT07Ivm5ebMwItQ+a2h;IWFCE7FNo>M;_uJk
zK}0qo^$=25xA-m~3h_(?x5~FfA%j<dZip+32c_RZtG#9zp?0a!zEMjuYwVr0ldnj9
zlb?_8akwiDoXFoo!CPr#0qv{_b4<=M7{FpN8pG}?3SDKY^W8Y6sNnr!ZL{+ZG$~2w
z;!HM*fa67gkUX}c=@u`IZMibD^%PkoIaAWXf{DP7Gk~n@K3<E_sc3w@J}D_z5|V9o
z$3^7@ghogE?PRz!J@ZyJUDPkV!2jm`)lEcCj(cobMu^P)3d?_=8{PQC9c}9=pCc0e
zjFjYRJi@H6K)C(InStMUX)_#aR~35i7-BWo@{2!IHX4_P^e|m169(b(e-uJN`$yuJ
zwW{+ua;A-H<!8SvtuGvULA^GNRj`Tk&{%~BWFn%aVd{o-ZYejWMp=hVzabP&Up~2b
z|M%#W^2=UO`porU^6f;23JRQzUyZz=Sd&c<%!=phY{CMlB=A<tFdYxr=;xd*mri?y
zL)%x|6L#zcq=L{}Tr=HOi{9v>d0n-S?fCPOKXn}zhAfSjNzYbi*<2fa?AXk6@o_uO
zVa0?=#wWh@QsdYzGeC=_yzJ)y$T;oa_RV%)V!jFbs_C&HnN{oC9f|m;%X#=`t`V2_
z=bQsmCyn#Lq7hPMrm7D8oavpC`RLHutIV&=>p9eF{v1Per+kepZ~MND^Lj09CH}jQ
zO60Kq|Iz|*QU!sr(&g*P#pY-H>$<SItsR;7jB9lQnO^>>F}>%45ez~JDYd=E*pOSQ
z4!hOgWR>YqT1Jhdc^Kd-)3>N#fnE=s+zB|j)mp=4*U7>M4^jBNpVt(+jlO4%dMUDP
z2QZ`|J7!eUB&o6Wy~filrbeQXjWzsAB?hhGQgK*>%v6mRqdc!g2W9yWTi#XP-h|va
zyZg<>BQR~{qY#qH-hb>a)^X!#(2EN;GcVLR(h0tOC)i-|byU}4JDS*hEDIJ|e~llC
z^qL~*Au43jQ@k`0h0Th2n`H*Qpm8g;_@w8og;km$F4lL>9L+o>3)7hIBgYCaTlM@@
z+6+uUuCE0Te~)gV1ehkCvW$mZ5NEKSSFe>=1Q&=8YU}QdwX@tLY1EsKS~OD%B>%k$
zSJfdiVzq7Eik_<OTu0bknUhq}Tyra~Vg$PA%DT}kN*EEp{lT?Z-~{J$wUf79JFZ5B
zFJnHGVen(zrcug#h>e@Tu!q1fn*`iLwl4_ceYq;!0?^MCrw8Rl)zV;c<L^}ghr?9X
zi`i2lMd{P)DGl-;y{xs@x<}oDgJ;0ECvW1x8nwz#^wrhwW0C&`Er8bIqMkh6HM7KJ
zT7R7U<$BoU_H%uYOJ^W<A)@vyJ-d3mJCQ`B2FDb4u)m2{<j{54BUT$4VrG6)yRf9)
zQGR{5wR|&4Jr~LT0ndA^V?}@sbQG~6v<=3Sm3tIP$0^X~i(U&u{X5gX``#yuol3s8
zAn3{MLty94`<BR`UK<61Uk70UDeY3hn=_B){=>bC8^=l%%Max_k+}1;zT9jUmD4!i
zSt2mk7d%@T74?{=SFK^X<_iLeUhgf(Qv`j*^_-JbguxRP{x~7d@9-<fYJ%tKW>vDX
z(V2pAzPcgjI`7nkN`R=*;t}z><apwT>*AaD{MWm?(lH!7HETh&C9DP;342ZFJbVlq
z9OOut#tYtj!>|EXW2pUV3iZ6y27wZgTP`%+i{K744)2}u9llC$_N|{0@fOEFwMl)R
z>1PjGVv)}D#`Ql#SgIYn-@<Xe>V8?Y{GLMxJ^=Am+?f;oSt$+d-rgQ+NTAvL;9zQF
z>|2?v(NPcqD8Jjh_c|50n9js@om`d(Elyu;CkDQgxYvjnq`JCJi-Dnop>P+@AlN7Y
zr7H^LkaiDWf)T|CPc(G>ujJ&nAGI1i{1Va9NG}~ltXRi<s?P#`oXZXu^T^}g$AQWY
zKlW`e4k=>PupJ=4fq$op8~6+}_ue)Khxs<wkIQXr!DUfDx@4%UFv~ABHr9v*53Knu
zrV5aTO-L%{KWb=<VxBoq_6J(kMMv?KD|s!4y*Nu<HA{lHn$+ui-+c_m>SbXVesvsW
zi=A^r$_b#0y6{%Y+m_-B6^H!({lUxx48sue>2d^Zs4uP}<NBHoSN`G-b0B$)t1{+C
z{*Pz6fp%J{yC{*xpI<1|{vg1@5ntkR^Sj&ns*iBa-tNVzZ35xcs$_)C^1Ca(#{#7$
z93HG#zw?o9g_07ggMLOu9rp`^@(Gkt3dm<lX?44|5WvQ$<Lb8{P~~>;HZgJf_e18d
zgG<!E5}2D^@b^6zoCt|-0O<P2o+?+MJOAfkhSmU$VVD3pIJG^cUFy^e{jbm@{>SDm
zIm2*pu$RSjkIqPMy`78bXk6vYrsvh#V1K>b;Jot^-BuyVOr0D?mv*M175(DV$a?zw
z<yP?wKBl}nNJ~p3S6iCl1&ADBZi?)t9^nl+{Db5VKB;ILo9IR0k*m>r%;hv{2lgt+
z*=0O<wwf#J-~f0gom|`eP`%4FpNT(`qt6%9h0#SF8^K>*)0!;i=%0@9vyj~me0k0&
zw`TuS%@-Af(wg(LTqxf2tyvZph?@vq-;khkEZrfNe$U`LAW_-dU0IYw`vjhK3A<mI
z`eAI>e~u^f6A3QFLoMg4jY2D#V$~ZDwv%ekENaeax}0D4TQ;^u47#@3(M<b2bLf7h
zZm==wywBA2Ng(X4Amqv&*V51c>#p#D*?D>5-FRJS`!Hbk^s`r%Uac-=WP2#g!0BdW
zJ<&~Ob&NR&$Q4DY^faj~I#`P#Hby$58Q?kYB2R>5>-0`mq}4GFCy}_r08%qZ2%L6}
z%eS|N0&!7X_`x3DMTpwULVgC)^MuN49fV&bDU+&v*#K_7C6T~OL2n`u7_*LH2q5lV
zPRzD5J@510OJ|APJ8vn6`Ni(SkSX+a>KA&+zRcotuP7@kGaH6-5p$gMpDatrZK;)F
zF0Lx);=LA?!sj5s9-g+H5{^IKnlEdd25XqZTon-$mKK(mNH`2irS`0@(7B@$0X~>6
zM@jdRlD7wLZ-USPk0n>ttLqSEU;hVVt0XyUE>atg9kHz)xfwaoKKo0A{7Co#po7op
z0aSqyv6>D>p{-pXQ@Rmy1cfp#E<wSu2d%-FHBRZ~|GPUc0@nnQ@WLsLFEwOlM&O)M
zT19`qLb<x^(wu@fbfjRRkq{tqzW~PWJkdo5{XORBaM$^+Vrr8Teq?78$of^7MqXT5
z#WyNTNl7W~R7`T7gicsKJH*00-mf;bSzU9ky9{L;8LLDrVv_%L&4u(&TGze~nv+Z?
zhmN0Iw(n!h7X@#1?Q9LijKRdxy<LFqDa^v}Zm*D#IYWfO{He<TaBw?<YZAr~Ey6jg
zlCU>8KSjITIew_3M*Dg;5gaA@JdUKWd~a2&vlKJlV5_Sbj>y-DkwhaY9xfl}LyQ5+
zyFQ1=tBqTk^+qm4w4arc%h|J~2&Aa!w#J^Z9{=$J99)jJa@Fa;b(!V2`5!~6>r~WE
zP=;<OrS$l_rnd~DMOSAY0(af#4gHLnjUUfRTT_ZOn9F>2R;c{gQhoL&@^u?*q!koM
z_}t2dyaSg~&RC!3jKHx@r*qChbVlg0LT3^6<1sT?@NB&(?SsFx4_Hx8t>eQEjt-g<
zm}dhM-gmva78*wjuWA?uzDr+Xw!12T$n4STu_z6Nkh7~dmz*8>?&^NMi`+f~nmokJ
zx#RfA8l=Ry1E%`y`@yoFmc9K06M{MpL(I@hIXStY+z=Sx%;}_5^*qD8vgc^-MI5Ds
zL@Ne?$3Z%TJP(4_YFC2i8h=2VYeyxYns7cuACbgN_z~H6QumRw!0C7p25`8ukp)$5
zEw<n!2s0G8EI@IZfdF-@2~v}uiYc+?W3Yk1J5dO2m!jnIM|J|z%*Xzp*oPoo0NU?9
z*6HNue9+d&S9upc+PB56@B?=pqh5z`SjOPB%70Insu^_4@gRN_2mw}%stn@w_$<85
zLU6!91+ADr@$-2CmLbRfky~rbsE=w7G>R<T$&^^v$5)97IZ@_qcutKjBFGH>Mp`SF
zwjxFi0so}?>tbJh{&r)X?I8g9EDR~uKxP=WO0=%HMMGn&Pj7edxSk#R5__R1mKI!L
zGEi@Vqj<e`hmd(>df@xrY7O-m7SK#0!2+;e!2-K)W#Almkj-ENWYx1hki&&q))T*Z
zAafKU_!qJJ|27&E?kg~ylw?|NcO9*YHraV@w_ipY-*OWKIx$c0-7u4+C~_$E=Hbb)
zX!zwWumZ?m22sg>hecp4p~~2ZdG=Miu>IIb?8vK*a19gSBI!<K6Nvr_kA%ugY&9g6
z(f8tZy2XyxLNJy8{1R;zflIomPgYYSOek6Akh6qqu<4jzlqSF<c~^S#t9=7g${cUr
z@|3-S)f)nmoBe}#?;_K^zc-FAzZZ{FyT1at-bdo5C@fK$uO5On!7(3Gk%3<xzcw4;
zo9&&5ZnQ&DzxFe`U!cF!40uo9@bQRixSkw@JinpRAQT@inIZ;$+bn)MtM^C}2aN&;
zOo)J|pVA;lcGi~b%9n2<0wHJa6-NWQPF7F>;{&&E-ObpLY~jE#5)iR*3gK$a%s0eO
zRFegLy*A20P~T?)(9`O_b$81Qd<J92&ffNKFjunHBvf|wS{hzka+5QHNlHNt3O%ag
zi9vrQm-*I9r{k(Qp<kaQ^=^7W)$Ezl2SZ~_(rB5pq@528b?xG`J=SSZ@cxFU4CC>C
zjF1!$)A`=qU9?Fqa<x*e&1!&~t3U(`P;>6Q7X37QF6gChdLS~^aee!sNd28O`SIc_
zUn<8Q?^XPfAo9rdjc;q%W?T<Ne$d^&DR!(|lKPHb!Ld<Y*w0t>nyeb^vY7NXKY&5e
zK7`R{rg8ih!!fKU`w?8HRs19tF$f&X7*zP%RtP7t+GTpndM?vtAhR#$M{oYiC-zl|
zZR@4jzz?+(-#%#8ZvPuWa1+EVW#lEEvrqa52%2`cPGRl>4(+f-?cDTZ_ymN+dV8-a
zW&O-E*%+9Z#No&i%`<-%e*?scQ<CwDJKAlnWYoRgVFPOAz4dyK$+70ZoKT#4li}88
zq-&HSS77fTd&^t!`8RV%EE7fTN>ec!<rx4tjqYnZNB1`%%RIk-?6NjVO|nUUe!PYM
z@je(&XG%I!ZFQ)&i1YZ6wxA5$>A9LCTi*}h{vUx+rTTwB6kBU`rUMJ%EAa*$s_yG)
zmIsQ+GB!V8e#ODIN@XQUUTb9O{OO-@;gP0OaVL4_RVA54WKj4w_~<dJ6ipB^fk?a3
zb1r9=z2%_z+4|lzUpFW@`QXxH5YDq=mr3Mh@?6vJxEF+9KD-cLZ7MdICtdt3ie@I>
z&=R9{y=7se`=|R|2z}Yk+<<q#0{}=y6IIXK5DmPgqFUK!YZWZ03K$uBDY|);aa%G|
zVyX>A&pCWLMeeO)&^7aDWP=BZl@0D_qlEOHat{F(WHE^HLp=yBxuZ+%1ZeI@PyeR%
zr8&AQuM>Q?^}+&G00*Ln{lJWFFb^tjkvMUX;zt02Hh4n%V9=2PjIHT98m>i4Gj0c%
z#c!)Y(u|A!Y?er?H;&$mNnS5810*7FKy@D+0>W6Pz@oL{DZ!axD=lETBuAK~WCr6N
zuIZ+01aeFwO2r9!gsYE6xgP*P7&2~EKh%K}J{#CNpFO$vSF@Z8Qc%Sj0)sV&tsYjd
zL8|iq=;u0DWEC#p`_)8XisnsGULm~!fb}&FZ_M|jL^@gZ5)Z3<y3BP4i)GCk@&E?G
z%)6~*8gP#8pc2nFJSXQ3Gk-H(VC}q1$Q34FJ&dKPt(|E0QsgTrV{y8rfmhu9-{s#x
z?)XR66iVqId!FI5AbSRaxkFz2*~Rz&9sANP>YQPc_*E9qU0B}@=C?ME;?Fo9K6GA}
zdy}^;jT%;31V&k>e68JSQXf(jD6>qV45;ProWj=*8RpjC&4FI9nY|vrJ0G_2J9Ga;
znMAy#sn;p}-S*xupSLciNzdtHGJSUUk(XQ$x=LIKBJ>4IP{6OeFpuBgB&PG6A5J@C
z!e{A6s8ReW-L5sWm^dzp-7WB9nQU|~G`*I7Q@7wSJ)IyJ+QfKx*iifCl5Z@UJwB|)
z;1F4J+L@jm-3=z1J^c~{92$;7<MYr(pE17It|;nM*|=OcB9oSTyIcn{Z(nv&><pM(
zSKj;jL{<{43+}h9DUzr@S4Y13ycM=nOr}?f=eFEL`=m|`M7}rC*2Z^R#zPK{i^KRQ
z@efILT@DHkzGKQ;=;;;o!r<9VEB{fLK`!m?cRsO{747!0e<0}#-SQ1t?L+{+<Ir*q
zo`E{`4`9pzYMQyP4`X{J&0bZHV6645^uoa|iq912jp1xo^Joi)CQlgbX@T{{E09`f
zMIAY0`{3ErF0jsVoM}rHWX<+XzFem#!(B1X>X#I1ccl{~+y6D1ML*8bLb0#iDKj%y
zMg4HkTp4asrhVXuC&*4mgb>EOoUODFa;d7i8Wm6?<`Je$A|jw#E0O^}jSOM*ij6^C
z;jHjHOGk~0dI1YWRzEKq%6&X`#O<u1*+L@)fOq)<(Iw)L$3>5(nW%<&*pr8;R+Yd^
z$hc#D0bvMVQhx$eDH=CkmyQ@o6tZl`rFcQO;_qXBKf6%icez|GvoVxX)mo6p&N3CP
zH%7XI1XSeVs6|<;zX^JBwaeTkT;+H}IfRGJ=_PkR#OUQ}ld<rwAuf5S+1Ua{*d|JL
zj3VeM3jE&2!ZJPrP<=$_cfa}!J>u8Qj<WV13$l36j#y5dv0M^(FAUvE=BB07`MZEp
zFeD8_Th4|kN9X<B$xn)H+?J-g!;NSwmv>BQ_7vx+((5rqpgW?s83pg3&rjs4+G1%o
zDrWdBm>>TN@VY*JlfKvf@Jly?m$s9>FUqkn`gQ3SD|VGh%74Mt%!51HAW2D4_p3!m
z(^HCCf+himWf`3B1vn1?6krC|)SFV$fVg{cWO8-oWcBtuT^xU7<}~B((a$~&VsW=^
zR$3m2Yx&q7lgUWe)YODx5G(<Cj9BI$x9;7as+LyT=}59LXv;SU+-{ft+P!+t|0m=M
zHh@0+lMO_f7Y95$KSwZqCj>UEO|Cf|o0~Qy!C|N0|9f`n`fpk7=1uVfq08@GMJm4c
z{|jg`XOprCjU8(O&KG2W@FFU2dq>Ma4TCJLy4$nIU$6F?`>i2a=8pa2mnK(Ge_HGN
z%SYnWi!2CT6w99x>u<w|HdMWh#EL>l{-UwT1YcDzZy5C5W%6;G`&mPz4H%}{IE@%D
zfqEU=+Ia_hkwlsNf3UBq5u@xM>SJ&EBOTzj{0Bi5Mz~~F7!^_V--BoYR9@>&rvouB
z0_HH7d*1{pkx_Zg_~L7nVxF=xfPEgOIhw>PXn^rB)(7hWx*@PdNc9tRH)XiiNH>q>
zuhrdt5thAphz$hD`ex(BL=&Uo$b%k4KfLvlUSF;?ss<s8XwplyoSjJp%+fC<(-rUK
z|ML|I;kRDLe?V~kwmvUhfd-`XIx)DzGS$=^QuY9{hL7bg4B%jrFQ~^$qlsT=LvwQS
z2YLT3rXQsehCn`u83Mp#hTif%fYJ+8kkQg-GwfK+=>6S=!OoXb;JkStKt=yZ5rJlj
z7$lqirDzT{Dxde=OCXf7dKI2Lr~(JliyroaG9>tQp;&ln;lBahUyzRvzmA=Gr-+`0
z9lC0-6^$<1m{HkcSJZyv^WLs6(VJZw`fyQK#<~+VH1+VBL+1ocBZ6K;DC%6)`Hbz5
zKJ;--n3>KHa$SBBCjo><wFRt4$LF9Td#gpH9=38w?=BZu!3)|(aT&dIYD{b;PNX*m
z<h;1;#_ljHBDhHAB<}b*yf%YkqehRVlS=3U!N<Ax8P7@j-{`K;4oB628leWbFC;^u
zHm;;tc?P=hoHlbx1_n$J(HIf|AN1rx?r0NjsH#gh5!B2xlRDhM^*6UdxduI6KpGhY
ztYF{;tb+xAWODJ-S<zFk-(a{;w(9$KX!!N{j|5XLYRGqIK$ttzS&Inh2FR2R<RhGF
zGc<x>HK)y@Nd!{U(n=@yj$00xNMCN#Ld|INhp2Ex!A_x}4tf0GHP0{ipu&J?Kv@Pm
zQsmUjp%dl5Z251Y<V+%m+6&gC<Qp($Qv?46?ll0M{z+{|q_99OA6t0ybdpitYyxt4
zfBSO&kx+t=pQ?WwtTvr!=d<%>55Od+cTI@}9uziJYhfV+fErz5pK&aubT{6=1A_uD
zCL!_VzdEcwlc|-GXhE@GcDs{L<8<uiT5RZ0Z65O2C7!9GS?f?C{$?0sc6~py#;SnZ
zQuOvwPFuY0-BZcmt@7!mUjFqzP(w@4;<Ilfp6@@=Q|_%*XSmQc<wP>~fSq(N0rB~j
zv4idkLQtE}_J8QCTGHpDivRJi<qM_nC>PeaoDe+iTmfMupf!Q2w;=Ra4*zM!Gs*!g
zFF?rue;@1=^2yEHaDtL$Gi{k-Fh}hFrcw-YAZ`|KGj}AE)mQ*PsdO4OL~P!@_-}{Z
z>zCbjjr4f1HzHt#ZMddva<6Pg8T3+;r&DhP=h?1iXgf?&)^!nU7$S5LUj*osH_Y2K
zC4Q61SNR7Rmvj*|=;5Je@5W)>mdS#Yx_ZSyrvQVB3D&p%tB;IjoFYhRI?p64#@AX5
zTMQ`P+g#P1OdulE68O6aMOIFp3z@la>d3coW6UY6zOJ!x6jh~tiXoa{MML!C?WdEK
z`_1dgUl$!r$Jyz1l$VRo_r}w)K1GX;7xb0LG`AOhaMah;xd!%UJAAe=njO|kIWlXz
zr9ogL8u}f=$i3qdY!)B9LWW@V-c_@Z=@2kFVlQ}>@9~_L^>qEk+&lArjJd##`N>^w
zvLW6tc|4?ZzXHQQJs|w=4N8J=DjqCXGwd0OW)xs;SN$z-&$a-cLU*^aw{`62as`1h
z`F+^Wrss6WuP9yx`xF0trd?L6;wpkS6dnM8q1IUO38sqZ&Uh-mvxN8J<E$JhmWu?}
zAj<I>wI3lK7~B2imS;w6bve^tgT$EeL4_!Ox_8HYoTuAXrQvAml?QchsngC+@cZ<0
zf6c`oB4Cgj%l+X)l75D9xU;^Iqt)-U>pf4q1&WxH2-@N5<S&yU+qQlm%M85N2WEHN
z(;Y|jW8_ZEE?P&^grzP(Qtht{a3W6{Y&-j#t6K3;8Qr+wpYVFK60WEAYblyTS_VUJ
zy?KcQl#68LtkBi`)<1|8Ws5ZC6Y^z!`JUXbxr%-0u1)gbma+~aW;9u_i(|L_)C7S;
zX30sgbqwhlAEfEZg<6El42owD94}@5?*_ysul(*0G_<tJEbxB<y&NhHELl_dg@VlH
zfBjNg{i3Ejbtib|dA>bD_-iLD0cRC$ztQvuMt(1tb$6;BU<wS^<S@Vn93~mcgGj`&
zz9YtO{d-u95c~S5mt(7w(?A=C*q#IEQ%`Pa#wX8e7aOyx2I@{$N}|2rSEV;v^cAr4
z!2us!qP&;WY!)uOtukMB;l<rw+2iZ<(&NdevTIj)9#oqQ1XEE3hlF?^;e!VK?T_?q
zRC?|D<d>9e*FF?d|LTKvO4gjx$tvl1kW*rhX5e$_`7Yh;-DnEOVFj+~#@$*Eqkx{0
zk`gEcc{dE2e&Syb@9a*>evvlOTwZcKZ%eNbn!#HHNICdKBzSzA30$??dyu+j@%ol*
zKh<oIE;xu+QgNyI3Oq<X=j8s0x1M5?xM7)4tjRwb{D4U$gpRphMM-}@DVDiRm-^C!
zGdRBw&eMXESvwPLX97jcfOu6w5%>)%1eh5a)k~*A?N|)ji*u|2#~Oxn{>l%?$B*Ce
z-~SjBj^21MGBNwdSXCf^GO_=b2p*fS2Hk8FT`i7j`d&OgsNE_~dHRgR-Uwb7!F2sP
zYL4+H6q{12+dtx23|~ZrPy@HOw_xvyPC1BYR$p$vzFtkNv41TJ>17$3fh89QOX*9@
zA)jGD^}umq<$nOoDq{GonC&Jg$?J)fbF4<fke06Is5AJr)v3HwZOapNH&6?ZC2eia
ztw)gye@6DAY2oJBI~^c(e^#osS6umwS+InaMFtCeCetd{NN4FgZ=1K#_>XBRCYNaZ
zmAz?{`r@kCu0>?RP_>w(g`?1;Wz5x_eY8p1rql6qI>_e72c<FEuJI=LV3sJO`|diV
zd_eK|!K}UBwpHrE?+d!9;CejM2deEAi{STL>8`GxRyKqv0mD``p;+RA)rw={*RxD2
z7G0Z#3%nBwDRKSs9n|oZ{cnO^ASPDw1;*o4c-YRkjRz98_bbk}fQBrgHgE5<*7L>C
z?^dF8IvVetXyvS-?(Y(1u2I)9WC3o=TiK}6woEb{i_KF+M=z&I!uI@`u}Ba@Rc<I|
z1;|EysGV<f>AxV4!aRz*SeEt9^nbb??p-S)YubNn;C*vO%1O$*are2@g!TJ&7P6P0
z&bjT0@6ElXwNzmk8`ZUGgYTp2O`2k=;LRUA%_{dS+U+N<RbtDs_5P4l(S<Yjf@dG5
z99uDk=Po`>-5C}WlfOQvwAkO!i)XwP#^7M0`_Y*Y&}x2gddI^*!O;*d{@hhW5`ikn
zr*$nl7A4Q+$JbM3w2-;k=$DS)W-~P>k+MwklgA6gwEE+YB0|5-^J&TkGxE)9G=f3J
zem`z>=Y{HFh){>#ja#L1v-^xs81CQxV;2rux;v+wMlA;HW?bXXAO>x(<1a80i*8Nk
z+4@RY@kwre#{D{G<CB|b(e*<mwPKsYoszcVz}AYB`XY?`tWiDmE_+?5_keD7m%D0l
z-q~D$oclO)#ZlINZSTXsXVR98>U>~>xpC=_RF%~XjpP44zeD1Cp<Fnyv!>1;(B#}d
z+1ZfFbv@stKPJ@FdDgCnjo&m8W#dql+aw;$U7a`QLa(X%4h(O<<lDf8u=NDm%rPU%
z7lu`?x6z!eK>&b>Q!dbHcKhIr>sDaS<DMt5yVqON5_Rx6Upl5>KSs>|owMaj@tq$x
zkKB*{(gN`A!&A%smG3VZy>bXCmbDv0RbA0Oc+aFL^py(kT;-$FBm}+->V4sno$aXN
z@Ra@VYx8%T967BrzaP|ghW&H(4$<%g)!Vt~tk%1J^M1BYyFz)4p1Oe*-+Y_9^)N>c
zqOj<{k?oAL|JOI$G?`B8oV3O$cmQHY_yXvs0)QX#Fn};|{1OBZjjwDU$A(#4Nxh^O
z3rcwY8D4(tL-BWeHVd2X#LRtnzW7!-%Fz9Pf_=dM8wE7!WbHEI41+36ffAGZcP9O&
zW@fa+s9<)H0e)fuQbGhoodO;w{wBSvUI&c-%iH_v=;#!-j00ak{Njrt(ct}CrQ)j>
z9A9;v)vtBS$I&W7ipebUclZB$1ss{oUba{WNZvaiSPb2tb&L-aZTI?@$v#^c#Y~(8
z3%1R9sVi&l52*SLvFdBrnESA?K{e9#E;mRekKQdyA|L4$o~iK1EmI*9V8eX7`Z*%2
zaX4y3fJVg8^d-Asu_O}joiYAcXN{-xIa7N<6#mAS-yn3sP{taFA?+MOf2DD71KIf7
zur?{i>3^>~Rn#rdGh2);wCRS2;p_0DBzHTMkA!F?p9#cY`&C+SfR5wCSk{c0M4i(h
zI*RZ};S58ER!wc2-`e`B0(PS(zw}2`AUaZB=-Kbr%D1t>Pd~gCyA5a4L}jmz>nX2D
zIcRPR7FBNZB$!DI&rHl)5yzG*4IZ})ZxeCkowpt@*OqL}Ujd^gLVmt=<3+fc`{w$c
zOh-_`%3x%CR#2veh)GqXR-zk8L>>f??e7+c%05pEBzElQ7qb|>;TQ5;={0#nw>*7h
z^FA?z@7k~BLl)BaDSJUa(}jz}o7JEOeN4qRj=fercfq`^;s`&!y`Sk^`={7GD#qhz
z=bU^na-9N?ia#C(uk_RU8V_dKtxxT#`{Gu!NCd_v^78$9YthC84r-2{ulBfR@p&8T
zN4(b`+qn$=#&N?8hIrHQ7xU{yi2WsX_on6}WVt58#bgB^ja^ftUI0=%Xl8x&?CDF6
zO2?<)zXp?(KZK!+){D-Hj9gU;S_$3kj3*%6qdc$XQ0JT}5{=O@($OV7(~8S-RTP+*
zM;}dISJTEc{01(r)tP?miMbGnHHp9hqx$r<cZ$C2$14=8B$bcWzJmF3t>et7$4eVe
z21faPXRaGf-*NT1>Ut;-0uh1#tF^a`s_Kj0MmHdxBHbV&BHe;?gMhR&h$!9Nk}4ob
zN|$tZ9vUR1TUt`O56xZpd+-0=`;PH`yJMUW=Zvw>-fORz@yt1&IYRD4Xi7-%08!Lm
z7u%adOyt<nmDExDY;UZQ{4_%F^1{>n(Sis2M`XWzwcPjrgMFfYbpL{f2=%cp{Q}hm
z6=dPkZ6qYY%NM$DkY>rzXgy#CGxxq4O&~iKsGrL;!g?*ZEf!~+`k=5X*FK<ML@;HL
z*wD*ftE1bf9LMfWu5y~s1#7DMym{td7u!D{p2ydJt@W|fx!lQ!=G^o3IiE+G-Tl>+
zNkF@I_c2fXCBom;osbf>BUN*#on1Zd8D6m_sE_mD0x9c(8`|-hP@aO*VFA-3sbn^e
zrSDw_O!qjY>xK)yCKCBN9)aYA!;gXYEBs0`rlszw@uu2`Ji{qUxr|?ckD*xzqF4S1
zke@+mtxhq|t>Y=ieOkr$^mLjoI0`_P)tLx?+G}o45VP3&8Glba-DFT#(qr<un0ql>
zV?AcZO)2O?pSr4P4fo74^!C{B?q7JT5T8J&!yQq+fBR1By=>}58ZRdL--$HR)Afk(
z;9dovmrOS@tS~!MImDcu%8NrNeQTQS@<w~niGsSpA~k>LE;FHc!czde+$(i`4V$mA
zyp!t&WmD$F>KeuD%ila2r5-F%@P;hinjxBvWCmT?nDS8>lSm(i%gm*5_c3??oW%;V
zVU1Uq!PN&ZbMH_6mm{*Ul&{0#R!2Loc=&R$P^)x{B4hjXR}8U(K+PYw_3C<e%)8%F
zQ){QlrMEu3G7~ndH#!S=A`WE+vFw|TY?`-KXKy;EXABL7w*}_)>)-Lht`0wk)V`mh
z7Os0{YjxD+N*k*AgUI=6HP@5ek+=U7EfXPxzWwqj^&m+^CT)jZs$Z?;(oOAk)Si{j
z+;seY_`rPKlIX82IIpq!j3JlEMrxA72lk97g6lp_+VK>r-jaLpzQ8jr1Kw`R9J`SJ
z(JU5YIBGuKTDs_}p7xrlBt%+IKOtKT2<jGDljJ_7iA`-0efvjzTEUd{O@N7sgoYO1
zCC7-#Jy!(o_f57oNkfz7-^O3^!VgbbZ;Fb?WVdg>VTVUFX1+qoia@+4oYCC=01P3#
z29E#uMC=^xfIk+J2CtdIs`*Dn^Z&c|`+sz#Wf-&Nf4)Xe8pv77ssz|vtf_Vz2>=!7
z<)Y}HksuA-`Fg@BsWoEH5UyX;Dh>RIj>MqP2*#&T)t5^2SKTp^5-l&jDrvAZB`i;G
zWZf!xD<+9dZb}$K_sL&&Kk~J>$A0vyddGhEI}*`AC_2(EBw^z!gu02#aicdOVNu9&
zea;b75IZwN5Ee=Rz1(H9g0n5E_$S6MHnqAx$y#xLgY6Y(DnzARNoC4SHk+5)|0&IY
zYjW9c^L!_hv7n%U@yl19GUn%CxN-uMM!ZR<-n3Aw5j&{A_QrK<s0C36f11r(oy`TD
zOvW#n(?w)&;=0mNDS*Y9;TIxy?~}9I`KN9Q6oIck{mMT#lY;V#*u7ekWo>>|$@t;H
z`hjhFhmR)YL5ANxuFiGpY$cYuPz?f&i|C)yiLvE%h=E73KiLQ%(j&o-cqb4|E*t^c
zlO_5(4(uONMEcE%KzbJL<UFP{XVHC~|A4w-BwBe7TWINQZ@iHJ{E_V={t1+Nve3$I
zrGcz#P~gh)9}c*gKa=Ssivamh-_nEZy%x;mbW~_Y&?DN}k5PcF3s~W|kJJ4P{>DfT
z5Yg4obZC1_fq3dIUAjLNiS*tACKTx1c*1&*PD&Xe3;E4e^5L};5p(ir{JYMFcG%wl
zGxYUJw4bUEo68n(N@x!Jsi0K@Qiz6PPDWj*lmgt&hYl}5xcE6wLfcwr+B<L&BIo8u
z)3rSz1ku(?@SBl6fJ$_49v96x6ytjV2Fw!7SPJs{Vap6-3vqnt;0nH}4?xqu%-}k3
zZ*K%YkuubwC@D1L`kMNf^rv?rcV1OpN*6y4ap&jxcHo6qBf-}zgLO)~{OM<~(1qK}
zJo$IcL;4YlJ62X}_^acrnZ2))an5ApMvFf@->0MvWTmzq$;u6%jki)vKD$2lCD%M2
zfWO%KJdjv<)H9UUbg_m&jGP`T;_FxiH{nY2UH(?m*m6;wTH`^;{2acC3?VAz$)F+E
z)}VP5pVSHETYAQQZ*<}zB9vPw9@lp`5Va2|aKF>q@#?S+phLqOp)vb1iDK$E$A#nz
zUL&K1q#+|byr2yY>NF!r3CF=;KoouTOiS@o3=Rwvf;!_S7f~F#zINSR)X&%DO2D!N
zYZra4KrMJjd;4UG{wzk2q;%i2l<KS*+8NaP!+9O!ka)lnq*<u3NEX&TNu_4lTvo;o
zQ#c59`4kp5-Wb?lN~x{^J4o>t)qNnuk)#cxr+KqaNCy~3S7BxYUnA~dMfSrm<p}xm
z)Vlr<YM;CUCfXkj4GkLd%N{436a$(rHq{?+axynpvLsd1H8fOCUY7IH(@TOw6y;RZ
zX2mN=fkTN2`5hb_TwZ#3i<JC3NVK%H)z#GklaS9aI7}G+fIVvI>Yly{8{4`uij~;<
z;mIq#r@G6_7{!IVn<~XCbqtMA)md&tGta`sT#R~$Ibchwgi~v16j#zG%yDQ`1w~y7
zoDB=tyNXa#ODS8#lIe8&S`3#%BF{RgG5Ajnz5DBSev84Sj~`i&W$pWp5rVM2H4eM|
zBP!W<;~i-tHzr#}A)G;r`;Vz-9MgZ~xlAG;w$dA1#s8IPPue>-q!0c!c7>MYgYRM+
z?|F(m%AkCII>OF39+p;noIs*_O2>`1_!`a&KDA_NtNjvoo81cV(k%$@;X@TItq=)5
z-lXE39O-|8NQ^*Ag#ak?eOy4)MeVaPqTiaz?u@~c7*kbppZ1V(e`bFKwBmX|kyj_E
z(5)G-xmEWNKZwMU_|t7p)IH!4KvCD2ZqdCL0-LIsU(!ROwH`0ScOE2bh2mjww&9=1
z!&B9{BfC*pH$L5qnBgac%wo#!%bgRlF4klB5~ksY|C3Dw(Y++}z_OqVEBE|=mt{Ya
z6e|r1eom<h`7Ofq|J{ESJrjruWoZV;4g_p-2R9IC!TlcTiIp<Aj?>Y7F2K05OMN2F
z_~wBY5Jn41LM@(xgO#9C^fL{n2b3+tX2BiRjdE*N^WPKjoWzu080?LZ|NE2K1N~tn
zzg%TsBa)oDNEViV($B)NjpSsUsK7}c38E4_Ma!VU3lqbIQhl@_D9t-^di%5l2%_Jb
zkti^-v>(+GNQ#dRR-XYR33K#*q>Bh(7U$*YYwsWUnjB*xJ$|ggqUJT7-1SG;KXGNv
z=oSFP+F;KE$aq6@m>X6Igm?{prx-2`{8xJB$MLD<UNwLk6OD^0Ba7|LphD-?N&h_*
zBi9u{kt=DSd9nV-k9Poo(r<8*Q&6xx<HClKxmcU@%m12)jr_hd-(A6+tv=@bGQ?C{
zh2VwK5b>vti}x@*>rOLnRe5J-QMk)}e9Fm}#mV+xd};zl$tvSRc7hX9+K;vmIHJb6
zhT0}NmI5Lc64>`9b#1K|xXP1hi=PAGe`N}@Z2$c8OJ-g{K_=tCtVZEmZO#8Va#y$F
zRSyH&mK^HU^|jK+V3ha3y%>B-9PWY#c3#)}iPnmu5i_g%lNNt7Jn)m+7fV?Pb1FP2
zHw`49X=9$zz^Tz7y}^e$4F%PO5@@tH`e$N8$1mL)VFmJN=518tF69;uPt(;~zj=U=
zSHn(#Pj?vU!R7OVUN5)3zkpn))Ku^rCn^GzR!yxXDT;<nB|@16*SD~+xIHqRnVISB
z?FD29EVG{z$HgqEhhhjo`klO1gw;F-NOClc_l}@~UNNXiOG}5UE%cB1(zt0!+STji
zvjY8w2^3?oxYBN>l=MSwBA1olT|oW%i}~WyruqPyxW(%59;$!KsW}#vJ_Uc33993^
z<s_^+*WSA958v;63!Y{aSXARluUw_?H;bv;><@uFIr~}9`%hnLm0On;J$?KLnl5hf
zFZr+uw$Jk;j?H;XLm$@Xj#d$uws)29d$goll!VW_KfFMgF1;59S?F8a(>IO(_J>Z0
zt(dYtp+*Lhj5BeKO%JJ4$8cF-Oki+DAGC9<Dt2%(@Bs2TW}D6jS^{O!KQ&N~<Gb!j
z=VP@GHiHx*={r{&i*qsKjbo-Fs4a4-VbAm0WVg1@i;({9+v3)r8RRwnhRum0^*<w9
z*nCz59G&YAYa;GEb)6I@$q#(h395g6glMkMFbyIf&=Npid$E<foVLD<k09uBPN<U2
zab~E04x~JKd*2LaNMNIMkn)))1SAwEB{b-;UZsz+=*J&{8r$jze~Xg9#<-5x=IY-V
zO5!$;nzs>I(X?P^B)vfpovV20q>p~DFAbJ{HRNk}-RxG?)J(#5Yl%xJ_#bC&c=L|o
zid@^BdB;3QQ8O?qC_I0xU97_(j;3$%k$qb_!%0bMV#%c5v_d%^W0`-@`zUWolOS~?
zsV;hi9=?YOwXYkNsYT#56d|`4k>h6fZ<8|-p5A7vPK3O!(J9VfVR&`7{By(iVwXH9
zjFc<q@BDk5Op2fC8b-OGSWNqIvq88T%i`Gt-|Y0qXj>(b=A~@PXShm~zrR{nRumK@
zx|ZiM_i`P*NlhAkn)x_2?CZ%IEhrBLN(m&r|I!v=PHTusKOZAA{z=5b+|FT&qW60-
zO+sA!K-K^M7U}W$RWnvRJUsttu>}S6)SIl|?q$s6L`0J=PxM*6aM*Y&Og79|uKI~@
zZP#@;tzNw+&+)=^A`Bun$wF94Bb*>7^Y&n4U|=}gn%Lfn5lQ*b`kD#tl{HU%xFfKj
z;-6Coq-vHXbFAjLR?c)%>1XX0Qj>e^CWWll6c^R+VdKI$xFwO?%JQn3VoSCa!=+z<
zo!e=6D5)Jdyw>+Ky|&b#%zgDb_Rf1fBY%wA(|x2b7b}!#U&Qdcs;je@fq8&Kb9h14
zwVRT${x%Ex%cnK^6|GuKTM|;H%~+FH?PI=pJR}JXWVcJrnlm61SO2Xr?))_^FKK%+
zn~eEQjq8US*VEvx(m{j7$9yoA=eNt%@kBPiYyK><c{L5q2xL&Sb5b)@5u;tyj~seP
zIT!9klVLe5zJl<4uY&$96e&1d64LWrcDv*DzCJer$j6Hb?o;36$!%L$ZuBZ`%m&tZ
zusf(yJchEOAJyD^&*rXHDVc!<g}EJ+%;qkpdRWaKY9>W3_M5pq<0XR@ii!+2W{qx;
z-7j%}NIxo{a`beq>B9lRBGn`NFZ7>D1ovxy+_C5#ZSUdJuJ^@HDUrpFj_&~S&|#*&
znCUhm#8+Pa)j91_ZSsS2n>9Jk-{QP#d+}@d*cztl9{o&O`5B=Q)mdTX<m423_rrYT
zTjSg)F^@AYL7dIz9ETiUVk2tY%?jH9MJ_nhi)ET~73iFoYJ17P?$7B0LF~`oW6;=?
zM0ydi=}p`d$)TWFAo_!i(P$f5h#G1G;3>mJWP>8JbZ&Ziy=J?K+PL0jceo-=aRij}
zrz;gBLoJSma0Y6nCvs~qfKfdphlAegaV8ymgrcLNvue*tkxWqy^2coq3{nT)U!AW+
zu&tLSPCap6OhpPwF3CrJA_CuH<vs2?>+YjwNHq}Bo3AG%+sV1{{Qf*d|CP1yTaw^=
zRmIlkn39V!-w}lO7WhELZOXsfEULG31NL@B(adcGA8f51ByNkdk#o4V`20o+7_wk~
z`^na~1U-(vvwNDH5wfllWLYK}?Uc5^E1;CC9cr#e+Wz8klO6sffT5cMa@-g^-OH>$
z-W)nJ*8A0yd`%htRi+}MnP|3YA2oM<B(3?|5XHMUwes@^Swh*EJY;MYO_=TOCkB9V
z%cYTWV`SE<=J#(MN=rz@3^iG1@h$O7VN84P5<s{lJQJ8I*Oi3(!c)?A@x#tVI$G+b
zLkP)TNO-qP*WBgfSow<6^O1(Ul8oZpftMNt3t^!Nnr}E;8hATs7RcT=_E7f0E-@Nw
zII4NtClyCE9_wV^m$g0k{SFG<ciE;q)Ll%w3in=(0lHVG!u)V6>lZbzL+{Eywd%!M
z%X$;sa(vYj_518wG?%6U0U<gph>Gpm;<kg~&fc3zE+-P`ps^<UvLTcW7syTQw0P0F
zDpnt0CuW?iH)Sf~e4rRARgn8)<Smd^+nYfSLGlBk#Oz%`@^FKa?e1#%pbdWsK7+ak
zh)eirPnU-kLy^ywM=5BjpHwNYUcA+A7HoOb+>=E{yZ0QVY`dCK>#9r&JOL>w<d_HP
zsD5@|Z|I*5_CyvdO%)yWh0Y}b%gN~#f_#wSWVO6I8twQbqR1vfdWcx^m@o>oP+V}+
zP(r$RmAHB_W@kjg$`>*qxIi{R8~Az_;|sXmthhvubRa^j;g^0FetqJ~=RBM`dvQ3>
zpCQY`t!3cn5;Fsfjjc;k--JUUW<me8DzoAE@1)Vl!e>L3glpHgLd6<X;%L?1XfQnD
zBH{RN1u?{fGUpNucbYy8@>d7<Xknp1_A4HLnm0cufzieJGlDXS7*i4DFO}sxKD8&W
z?Pt1PwH?0085IY*ouJ0bs`?V>E$n+ERP+mDQyN}Benn=pv658(hi$dEV?7>3C{Ta)
zd*Y-vJ}EFB#fOoFAgs#ohMzto`uS?A`)cbh{<-N%Y0O8LOE35nTwSb13C34pgyjL%
ziO}sde`3cth>tolWPLm1VLojJ=41mcK56XB8km#k7{KuKo8eH&Dk4f4&6mpl`Gd${
z=ykZw(|ecjk8<H*W(AlNf+n&3l)$lhnn|XuVs|MGc}x=>GJ!BMtXIylWh=Fl#dx%V
zI#LHHLeLwM_bRGhSVM$8NBYQJti?~p?DBe<vFpj8zyWeny$O8URbPJruwFt#-l-%X
zjKA+HO;xqyXG;~P_QI~rcUL~qg>GxrdIW(e&o_S|aAEGqfw^k3XwOz~w+L~3E&X3~
zd)yWrl3+oX@BJ@Ew+QJ1hr7^5)8kFuK^emF#!zWd(E~r)_hYk%99XSd&ZjI%dZjg?
zvB2DcIwC|OLv;;-8=#{ViE;xdX!b?Cc5JDNXMgk(5)wk}BSc4fuLz(f6&0bDiYHG+
z<s?E>DubrC-Ct@E(XJ}{b8HqC6lnG^>51S3V+En}mFRo^amo#O+_4{lhUC{m6vu3R
z`&nI`U|tIw6|%pwePt@OJqfD+Sct;JY<ZJ1O3pg()HQ>HgHs#+c75L@myl4Gr)Ffl
zEzg*Fs`^ZSY~lw_dO<hO;xa}zpq6<E{=`kTF~|S#_V&RqpTqT|DBv)sHGHN$aMAl^
zru*sDUl?*%66N6fZXzChlbn$al##ZVSQqQ@fej$*XGn4VGFd6Nf9=bsPyAO5?U=Qz
zt0lF3I@<5e9mH+3+8Vt<tkH6+Bs@Ia`_Jd6fD3T2`U5Z=4s&^8R#M=?=|MFI?VWD+
zc5F@j{7YSJ%I%>6wQtzZHetZVjedp;Axa3#DQ%i?bLagAyFxn(71Aux(>|KmP#9Hv
z{~8{xEbN{t<aue3Q4{g>i`APO%JVgQ0$Kv~>nfRj2mSGhTUPx3ikMNRh?hsh%Mz@v
z1VhDJQJ1};Y;zlVe<>cKK#zbLdv}*{yqj6!pTI{<Al0}m-P2<whV4%S90*(GIKaKP
zpPz{mAQnaMy_gMLbXrLQsoI73x{-V5!Bl&4#eAVpT{^2ge5*dp&VzWEd%lIXpn!>~
zFY3qnp?grs{Aga$W8AZ?o870>!VgFIS{6#Ootj017%C%QyqUY^(msYgX8op9G*Aps
zI10dF@y|W6+myz8FmH%CKiekgH1;2$7Ggxza41FplT|@U_E8X%ls8|o<#Jy6)c&AP
zLB;Y~ykYZ_$r^%0Rgc*p7gEh-2uEC2ndF2Rud95K{;3n*xxCcPo^pqQPMHzlmj3+t
zGqD$qsW#ks#m}^U&g8mzK0aL?xB&PlH$h!Ta}_w~a)^dZ&-p0WUhbSZ^GjJ1KiB7Y
z0ClEma#af`Cjifd1<NC?^YHp;H4q2Uu!M2N0TcH;`84&T#Kbf|U4QU?{K_+RG+KS5
z;VUMCB`ud%TTB=e1p>mgb|GW6?qXOg%{CNk{53yhr^~a$EhW)17>tJbBnW=8ZhxN!
za-eXF<w8o=tKWm=<><YJ00=`n56G{Gr$X^QmMq(;h~7}0Pu8vLTiEQ+dg)#I`cc&9
z0x~VuCtpS<rNEK>%f6Y!5@r%l{}hWOX)b0#_f17lyfZ7hy}l<Bu-*DF>^pwfls=>l
z^a&AL#Mg|w;C$wjJW%lZ@{N50+Z0nq&1-mG`90<^NI770MFuRT2UwRx_#~Wc!e^-m
zuQ3ihq)x}O*S3NNzo?Ej^uaix*0L?Zx3*^<a3GX9NlpfqTpDYwW>od`c5V#gBBy3+
zJ?R7xA>q7|#@w<__YCxyQJ$=cy!*w{V(plTL(5P=3Rutrx$5Xsi*`3hww`y3E1-0Z
zzxt+kvxjfxt(&;!47qI&3v=`kT=%)@3-e!+eGXPCAL~rjl8D3c#7Rz?uTI94jr+Lk
zR&3l{)*l)AIA1Ed2F^6g-fISfcu%le-X+JyoO#kx3R)2mv|MXA-Lzt@3wXDUhp&`0
zG?w@wNF3@Rtu-*`DBX;VY;`oP;AE{m@bv6(cX<N;%jT+ps>oIHMxHFuo)8)A76dF0
zL&emWqzk^paXv_n2JVN!<sTL^LdEpvpVg=JPUo4!31XTIRJ>EEKV*M1srn@Q$04!#
z2E|D~fS8uxfPK?DHZn4@ay6+dpyd2?Cilay?<1?;nfzzIl=uumGmNdtUkZ9Jq+B#A
zXLsxC>wi=GdSo{(HQX}v9lw(Mc77ZYzsd%-;xnwv_h{(oq{3cqfLYjm?=c=HAM1M;
zD{h&d-zYzag|7Btjz>?qa^_uTt+BCvNrpy1)~Cu8l3$tQbKmo(sX}qIj8H$ZMTQ)i
zK<@+s;=%GKnZ`qyrC5N4W|{qXBhA;Rdw#V<=pI-V)jMTT)5w>}M$2dgTSv2`>busH
zpR-4fyA&e%L+M=|(z7XpmHQ0(Tsq<5kEtT;FXB6kH)0(?MGQtjGqy83iOp;|kIC$N
z=sd@BUnA=$k^5j%4X3I0RG@DDMBn%%NsI264=O-~P@;IxdYG^O`k<z^$J)n-UtyS~
zwXs^tZ`r0`QNv%_^0|UJDgN<og;Ko0Lve(nV*k{JLd+w1@)%58EH{Y|9B-48p6ZmS
z2C_~84OngZdFhii)?Hjjw(7bE@Sbvr#_BFyq)aAP9*its(LZ(4*3%VduMGu)s4DM!
z1WfvB*}l2n+R1)#biwPNdt&4=flP#iiI#%>Cs9dGjjkQ7wLP(6`Gxxjy@=O;M-=wP
zsG-j4#mek1f!NP@7@`2Gta^W(7;GG&1lt(tMW6Ngr8)>3NvYn9)oKx;YSs6W|9g45
zEdq!mKo&j{ZE2S@+p93ljp@fxAI1-_eYG^xn%A?hw!T@2bE*)mjIyPdX#T7KgAKF{
z@U|c{mESSKL|_nZ(@|T_G<u?FRmBhYDIZNG8M?jg0i>+lyu1KD1yM>~#$Q_ZoFN?*
zJ3;#%w93L4tDaGal!9$SYA)U<V)F8Rb&TaLEbpX!MwHKjXVx?<$y1I6=$-JT_HU|n
z_l$^|WC{v5+i?T)aoc)#8$iw7cs0Z(hR3b%n>7ieyeB64FZl6thzkqIYP1)Q)lp01
zL}JSSn?ItH1~b#&4Ck8>*1o-ZJXLCFsILC5{0#~Y#WYIL#TU?&;2CWA@`(6H!Iq~d
zysaO2U+o?7EV^fNjzb8dG6BXQq5zlgPwQiwTerg*6u5LyZAaqRB|$V|Zxw=%x!aEP
zVH|adcL>2HEGcYtfX^~7@^rzk@qFsfTM(N8nA)c{qv3Lp{pNB&b&xoLbJ^^8Ed$VP
z^m2|f6bw=FT_uIOpkx?F*v#pSstkOV%#n|8yg)|Y(<zkap&zI!|J@i`P=IMznZNt#
z9$*_~B6f|X9|mjzdBqQ<%A0TD+u^S`NlP}rpV-;Y;N)(XuUv%oGu=`WN7G@nmhA-U
zDjG8XsDF`aOH?njIE-Oh^=Zb+g`dJhs8LH&Qc^-9pkUiU^dAn+o3wkv7r_eLKox)q
zSIs7s<IvoPnX2^V;zMglvOYZWU?8*=TvhL{-RC;nVerpwZqodE*C8$8HuSN#WX|Mb
zLdjFk$Mn$$eB@v_bscr82jcC`&@%ZX5LXvrx`LnEzu=$`)r1nD&ulsAJ2Ttbm{yDD
zlZ~amszXLwhVMV*Et$%807<1Z)<+Y;0$+=CPP_rnUJpy{J86l9zS>%PVK2*{E4kJu
zUuA~36w|IL+&4N|KVT_8P>P^xUA7yaob2xEO4+d7SLr=|V?V>2iwTWl{}EYvdppjm
zb779%=z|T(B87Ma>&-KDafj{+@k4+(H?Od;g9<kAzA!(3<ljn}X-5b%GZ*d3H|SL<
z4jWd9Vt7!S6Rb~-H>(@04RfO`p@`qHb<Pzo)zw{mfDYc_#N9d#WEY`cF9l$eoIrj6
z|3mQ(T}~*!^`~X$p|etxKSGNZLwcjso0nN3SKqZKriJKNwyS0P2V@I60yPlt_K~`A
zZR%&7Z^EWST@mebPdKbW|EM8j3dAb5hyJi_x3RT+1ko;16xgDQ1`pT%^j^5gOP`lX
zDlzjm=3!49b`B0bCB%b@ldu+jBx1O<q(PSaaceqU5OBcXTzQ`P)|rN!><LCsdHE{j
zg(IAQBSys`bGPh{Od*Wz9@==+ILNgN6#YXikCsPdWS2*ON-Lt(SRR*npO%Z<Jr#a0
zVG;6Bvr@X}?su0y&q1OwrRR0srKhSvjw_JK{YYR~&zPAhP7WSZ=fd5CIB{PdKoG!|
zq~s1Q1)hXgYHN$v1WPSU@;uHeI|&!^#CS;s+Xl1{g5Y_T9;X&{Yc8A72w5dJgHI+V
zCfsY`9Av_;%vxBAJDifbGR=`u#w?S|7r)J+|8!`{tDIu;P!g3yyWqMXEg>QDlP(B~
z5w{MXu@b(SluS)a!K*FR{jHnYjc}?#U=%%jEURK(+wp>Ljc}l}ge4ZqjpKFTYlS6r
zHO1kzz<f1z4ey=upUsRcAt+Z(SGmf`{D*q*chCEdr;2aqtaR~{A}z0P+#>rNwSwN1
za3`TADAHQ>N31n~OhmVK0>D|U_H-FtjU`xY@qu^1Lg)%VRA+=OsS+5z9W*}+Pblg)
z`6D8?`G%QHdA<*SRGZ*3DeUG|H&zy{6yA4{lJ?iSx{Y2J4p}4oKG)}!#K_|}|L<L7
zJj~Zd?JxYlf>^9xT2M}d#W8v5S2SD%KW2rVZ5-|Tzh{Q@`8yFE{O^PRTHrO3cT%C0
zhl*LTd0s9!alL9n*DJBCxjDPoNJ}e!(6UEu)e7PgSeMc4o4Q3TanGdV6Nw&PoOiRN
zFerN{I=fm|4WXv@3&(n%4Q@^N3ij_i)w}ZRYe)UeFc(ZNp;VgMkMP<}qd|kp2H#&T
z^_0_6MnxkpwCfpFZ5%9jc*6GvdU|B#<PKpyKp#H)5|x@-azzG5LnhxD;ajfQ9zhbd
zw%!G98m5C|T9eDFKINZ;EhY$x(-6q+{R@j-Rbr)?iE?UXp>$9{o!{$-%3Y^;<V^R5
zSDPTpO?cw11#TqsfgIvu>IUqv>222+PhuaGfROg?G?5em)054S<{0>f$esH)dL%Zz
z{W9N!Hq>TA7^dfmpU`D=89->JvvBc@jg7&vsD>}%$k%6u1X;IKWOvPdV+O(PYMr1x
z*rki&{v%T7En1#}%c<H0gXJ?(GqZSBeK{|$ROUu*o#KKCr70ylv77xg#*dLWnZt5;
z6@7(EGYPHi>?g`;cUR7a{<l5f9*53Rxvg?Q=X+zypc{mM*>L)#4HP*1!gv^!PCTZH
z6~x$c>R$7gh4c%{l96??B!N%W0os$|!l>WC_$ME%2w-0C9uh`CbQQj9$Ih2PNLSZ9
zGJT<ZwA$}bE*<eA#ht2NRYfR<%0(e$9$(M?h$#ZGWxdpw+=-h{dnt>bHFzb{jTI2e
zYBR)`S~HWkvlGQVL{rfqtc%{~{%WZS*IO*?q9Y)&+sr|aUl|^7$J1WNp<cz^F%89B
zerKavnJvKmwyR)&Uv9rd)njaw(9!peV|V<Vn&CuBV%)~J2!+%Ys-dqHN7N3Ev^hfG
zUkMW#nP|J3Wc^@*=zo;czIZXy$xhe7E?SI*3Gv1)qm@@M;~JTgKJN&0VCGMzZoVj8
z?0a6K{*q1T%4(`a-`h(-ag0F9ybH7EukgE%0AS*G8hcA}qKH<OE)rCiJU3#sC(ElF
zsc<`8yr82>B>Y})r?qX|>P7XT9<^0`e{5PBb;|cQ-DUMu%H>5d&n<#7)%)b@Gsg=H
z^s`Qj1que49Ey4g+A6*=1D@bz=0|Pz4=J{9WJo#!9rgJ8%L19TD^@?cQ~QnBxmxsr
z+`{jeH)T7R^GG-W3zz}TUBJJ=%M<g(_(osDKIFJ5y7INyUbFu%pSkg}g`Pjau)4d*
zA4RdQ2k7S(;3%pwj^`JIsOIJ%h{AFZ{M061#xTY3rcS&4S;WuAE|^TuchL~k)Lcg^
z7jw+noF>U>ZbHnYyC*o1kq(CMO+O1Wks#eS!++FQVa#l4LV8;@uPVRjwi2Eq=&HjU
z_aW6i2b{1EJDPC6j&yi?x-AFRCBp-+;k2mO*y=x(%8jI5mW?holj<DKzG3*$+MgAG
z!!~{{$x}d?yhCX_U1kiXJRu=LA{fiO@waBt8~c4eh2?c!<GWPBU+d(#ukzK$Eu`{Q
zCdf1jHDj2hitoP_tEek;Ep{^mAFlhLb?`rDI@8h+b3nBdP_4D{S^6qy1$J2lQScj*
z?mqTi@uH?#?{2Cup`nJ}rxS^ZiOU@!iy&#T81Bkv3=%;>kOQzozfw}N8G9<w;)r|$
zIO#!73^;1P8-Ro&>PWX0KX`CMDr<tIrQS+FT+tO|W%?rdOoEY6iicH2aj)*(P_w?F
zue)0OcvlDx{5$7O&8&@OqX(#*Dg(_~0M%MJwaD#t06Pw#cdurWa^G6ha>;=Q4!@rE
zH5#qkHVr7()Y#T2me$xNs8$*qk856L!7_PIg4qvVpqo>>(N(MaiC^ynPk7jN<Qx}I
zS6>Apt!CUqbPl|?#w1@O@#0J&D`q5!7OT;r%k^HSOjvk>UJ^sMTf;v<KK@;TwIEFv
zn^()xRl?Be{$h*isbOS}xk4QCOtqy}i_Z;%QnK6GUmQi0o5kC++6xUH8nj317clt3
zv3*fdP0d;BN2C-1XN$#VSj3w9%T&VWn*0=B!7Tpl?!G){#PVRfUvlf=w68r>YR<9A
zj)BjAeUK0B)h^CZ`5xtby&FMXWllPf0-%}t=JP+HAXfq~n=?FllHwhLqEV=((U>!l
zUjW|-EJnrkaam<N*~P`i<{(@%_x6}7)PWW7<|Vpv=WE>2fS?59zLga<l@fSgOo)zF
zIv4?ZE<r+rDDesl36d7q^oHx-rqe=@lFa_i)m}0?6@*4x6J1D>fYy=r>)rkcGQ?{$
zSLd-NFQ&0yxAHYC3~Rp_CHT_@35JQE7w<kS;X;bt>p_wJjcP8={mhZVUl5*zC+57y
zxGyy5xVvEZ<wgQ6^-HB0@(FNpjTMh@k?oGRrf9j?>~-9E6E`>q%+1M2hwr2${GB&&
zJ?g{>kv1)GPm@JPgq#Z+j5~t;576=o0)sHmyenFN7X3_zAC0ApRP&ayx}{EWE5f6l
z-f5q0Hq14(8QFP;FcSnlV|jn%Fg-`1qu!cIM5ObsDsE0JG(4B1T*r+*IZL9}X1)Qw
z;2EJR0D5A9^xAWSfPwdxW@cvFa&e)=Y`384Qgl?*?d4R9w3JkePqa+d7aDbb3hczm
zJaZgHOZ5o>DewJgWt<N<4+;q@DnSv7WK~3^Wx)@Xkeqs@+v#*wozjBabA-ks>xC1x
ztfdt@(hhczt75JH?e2Ii-PX)(_;BV0=*+Q2ej=TvqC%iOd3;os$(VW*OUEX-9jf&n
zcU@NJRpWy!w@byC4IrPx!!nn@annUfOR6Mc^Ec<@tba7t^5(b|_)6>QTYGs?8h6fb
zfdCPOcKvnQ5=?6Dbpbt=zVldW0)@>prN?E`9d`UO8PRQn=BCQ}YCes{-!sRr1A*H2
zq@{KPk=FSOB4iWZ0BcbwBtzcH7|&OM?D8yp<a)RC8)@k1<d07NCeecG{MYHZmn}+)
zf~Sw9vI?HX^i}M<)>yjGwD}Z=PulIKv(Y*>y+j~sqz#p+Yvj-~c+`eCFe4yB@7Fs^
z*NDzKi3zwS7A`!{^@Ok>qrX#W@6X8jY?9fO717GyFcUW$rh~}=f$8}Wo28~oW(u5;
z87l^C7gmvY7AB)Uh3PFOv#nuS#dYbp=L18^n7GJ(hM1wf!-byMWLQKX8SHO;$0yYW
zx5GcyfKzg7+#EDa`oLD=-dpv@Dtbux%4)4I7A0W0_i5V6%`ZqUij0g5^jLT?n9D_h
ziHXTY<wKB~f*$pt^8GrSt_5cT)T*dVp_J%URCHZEWib0%%h&nM=HlHg9QYkvdq&)3
z`mDsz8`AZ0&b-7N>zRtKl|zuP-Km@U**NfcFt)F+ui<FO_am)@(f$JNUh+$yYd6E&
zysl7Uk%j}VG9BCSugQR<DWcG9X*ir1#rqNaCM=QAPEzCh7l)k8<ox3I76l237TpF@
z?@rRbI#7IP8Gkzw#$bV+!B9!|Hj9>*?5*zg4Q{rf&#hg{!uAmcujJ2`qrT_FH>lN-
zdt(^7>@bIw?r>`FlW#O=0MV5NZPp57M!xLc8?7Lfe|&uW{M>DkS_&8y)!G}<M#G-K
zU_}U6*nTTAohVHa9tKM_FVmtB^f22d!u+_0j?Vk_324jpV!3Hw2m^}o*zobHf-n6b
zHXOAw=H?E&Y@*SztfvTQW&zGKKdj<j<NAs1Pt;IbDL!U>y5b;i^4Ol6^Q-jev}iI;
zx53c#wycTCSO|i6URS}4cs|RC<l-V(rZ)H~A1)0671!Z=drVzUQwtxc6}mlNDOSzj
zLkJPu4-#8^TbD;Vir(?MQi?~HhO^<buEvC##zk+^9XIj%6NV@HGsQ_GYer-{=><R9
z26?A5@|MEjtfnsp-tM~XO!wUiP5(MMbq{fDxuC1ovbEq;%le%7YTU@`&u0iS0-cYZ
zYn1@TqGVp{-=xHUr5GWjh-UQZEchr474^fftSRoCHYE#75>KJ>sme2tx22EIJVvKw
z)u|1bgELSb^N{7Y$%Zpnu)hv?7?zVICDp>BE0Cd(yHoJnH0J83rDL8Mzjvxx^seYL
z{T845mWAL=6M2T`5Wyqbsk}q~TLlF;F7LTlJAr*Yi40Qt?4tL|Y0_hHD-?TMDUCE?
z?^mLq$K>>s%+?<$GNBo<Tc9A3goVvy+BLxi+wxaBla3c^DPDZXC%YURgjEvzCL|_q
z_1jS5`~)3{IyySIlvAIqV*XP>_lg=31r`r{3$xpVat#d)P}2w8=>%?b)x~CS-<Gzu
zcXO<E0;IK=daT{Mh_#{ijcvP3f~h8X#h*DzpCKP8t=iy7jct+e7I{<fB<79ChUYwz
zLCcbRBz559i6>>jJ~=-AikxF-oz*<(EO=`VX5f9QL7nDrsvX*;YG}ClYg8v;Zy)tP
zo8f;ArH(!xU!7nr;X~@P8+X_o?fo}7WkuR3$)=x3z1TMKdHEIdN_wnzjU@5?p1)r2
zp6YQ;iN=vg$I?F<LTVixJ+jf#A4p^wH-bh)r5;KKFnls%ACA7;PBC;kSZYh)`k)hv
zSyfeH3p<Sil0RDW;N$;3go72!ZXeKF<j&LCr8W=E^WCq?J2>&<*WJ59D0Rv=@pZ|P
zQJ<VZ@LxES88!2bjxO;)B}IpNgI>LVKD_lK4Ot$Ig+JskE45!P_QM~|UsgF)r5Yr!
zul2`25(izwnn9zv%kwb67Y|B2dV6}zj>b*r7ja2W{&}NdZRSO$mwtlSCu@8muqN8K
z*;FxXh#$*QdESs$zCTre5?|B_?<Gl`Nt3maj%xK}kDQZ?QlcQyxz2w_!t{wMIG>SC
zp-ku5@PJe{tdz7ZX`mHTogkkE8bV?-1LJ`xWU!_CDq8Iq@2ZS1c8E)r|DLyGz2F=2
zI^5}yh~7N_RY0J{MTDvcn1n00ug{_B-Ln8xTo=%XOux~k`Og0T87h`XsiqaGh6XOB
z009w^%k5_YFqGA1naj6r{&ks6R3i^{L;D+WVHGA;SZ)!hROs}dzX>f-npp_NjN(U7
zM1Ozq5|<+PGm4S+lKbGhIGzE)C!aCsOtnMkVS?L1E!dkwct?M7SfUMlcxX93f1!gi
zHVi801O=~VdTS##BkQVvSp69r3a1pD{Mim%W2}H?bdnPT@VTrB$T9DaXB!$Apky~{
zU1|#$a_0MU=tuu>M&hNawfWz6;yaGVS-CPATC1xIylcWrW1pA<Way|q_i*-bew&kw
z6VvF?o>DiLT5u%JV@y3~GZw}By1GjLi2W!TrL3Xi{rSOi46&c-M?9ua5cE=68wV0y
z`e=;charCUElVMuRp^fsoRobjBZTUx9Qx1b`zL8V|1RLLd^e~Ek~l*6vd4%WFs@|*
zivEojE|qtYu@a501HOBI!@h5>IG;7%?jL?Vw5RbDRry5!Py<0~nMd)^XJVL}&Qmmw
zCnA^mX)dg*lf83Ug*kKhNbG6a%}uVNug9f3b*WA8KY{t+em^1Mux}zP;SXlH3Z%HZ
zyMx@hg$dT7;dg@f{k>9L7rvRggN|A^S8!DM&#O5xoKz9;MF;{)j-3jf5qU;U^-b0K
z7&0s?3v+8%9~B#P-{U%T6IBz-TKWe0^>B`;Lo`&l8XC~>T*>9=sBH8@NRxU<WZZRW
zJYUWGA{6Mi|5QS@XDb2-O+Wa9t_ISc8eR956ciL>We*N%`4+v7lUJwD@%lRoeLZ#p
zE`(W+!hjBG`rHcQ&xh?D9i{c_WQ%H4(1OQy@de84ipEz#gM^hU)!QCSVjc-JeU!H*
zCL6yJ#^$v7dH`~8-y4<vzxst?0N4R%>t7B3za)E{fcp~v<TRzyciNgiKRgA6JXyBb
zU_{LQ4#|UTk<H-{1nRFDqUGS6CoH04RRDJPt@~cwybD5A<mBYo8wRYqR<F29N%8Wo
zPky~Q0Ua$3Ja>Vu;90y7v<ci8V@zUJtL@N=%lhl<YfxNNqTi^wFYZ(WppVn-sk?2%
zyV1tTeq5?OynBlL;j8?%0Mw@p3|sm^x7}YUw{aziE-Q43f$28Vh;akxqHsvqMfaIl
z75JS18wJ9kvTqo+jjG0AIVUkP&H8i|&cmjK(@iCF#%Bdwot+`%pBAz=Lj?N2u4}q8
zLgi1K6v~iQ)ba-VGC8$m^~Ar+LZZ%^1RCP%UR81%X`5`1NltEMgl^}r4EY<gb<oFH
zxn8{R$IXm5ksEj4-0)u3!uez@I-q-d6yh%T`K)8~(^D!c5+e0tdmX_Z)MhW&*nH&*
zG^@;dM}L+#!VRe%q3?ZOPiIG*P8)dKaN(Z0mflUxQGR=NB5S%RlSc9|G+;;=J<hd%
zO|ql_y~X2FzoSTu(7Yv!CuFijufa-T&O`#W(=YLK<?9bY!#crY<AI&}q(aoN&scHq
zFH8i7FXd(WoN*T>*m|0m@@0Hx#+LVHTyq^w*EdAwH>|WSL+#tu;AbJljvuUhs9ii!
zcP6e7BSL&+3QO{S>0YI2875xz`CDtWonwBX%~WggdA6`Z;T#jEi~mQPW-^e!x<Y=e
zu})$0#04rYXrjCFwQV~3BVus3<tOB}K=i6lQF&n@7S%2`wwIPy)j;~sBT0ysHMJO{
zpteJ*4S=_u<KA;fzPH<IL)X?Z$AzWvK94EK-iK<9O6KUsPd!Sa=p7MJp;r3>dSvDH
zlo1^I%)0^H+qK@@)GNokxuIskk`2OI)@N_gLaTZ%L|}qf3Y^R9PSR0+fXw!Q!nrXZ
zhm@5FV7D3FbtfgeZZzFrQ^YL`^u!3P6=^h+X0#VY5vqcv8R*}HR6=<?`O5PAn4OdP
zkG1qEU+Qb?w5)W@P;Z?2Y*NDwA1c*F;tkkLK~=#hWcdCPL&EJ~{6}p;Y)vg$sZ6ng
z@Mq@vap~c_Y~{Mj#zSWq6ZUqV<{_!*Z31)(<i1z?n4uN=>P&O?({uF_!}Q5fFsB5E
zm1sDQ6&oUCmoL<TYs+@vc2q<Cc`Jd8d%&*TjtX{{;f$)~1><e}`kC5cEF5VT(_<<N
zm1qA-87Fs<i|Hf6>*qC~{aWeT+ggrjne0UW?K)Q-!8}62IF^eww=7!IbC#7$NtE7)
z5xiDc=cjjTZXsrMtb9VwA8g?ZrlC`7Vj&NsyK_;;=*L3+|2yp=HvgTn|Bo0zj53+Z
z#N~}a9io>E97GMELZ<s&cYF-zi!mO4{J_t0p62c`;dO)6S{y#Bb!WKAXa}wDsmxW@
zrH9Yevj~M|hwD0@U4Fp0rd#VY(R?{<ko6QxfddJ$=5KQOo5=jkW@)-S;c3o~(e9nt
z{hLUsG3j|92@gi4PKh=koq9@pP;YtGFSTdqi8Ab#DYb+S^s}SjH_6Xd-g{|0DXYh1
zFm%7G)lmP<VcF@-W*skSz8yM+<7=M#U~n;&TG)P1j7goX<*wk_qWWv-H72phRksqW
z-qq>Fc8>SMm{1g2C&aVOZzA@0=lk`2=kAZVilKbh-G&v|{_WudjmI`u@TXj;U)ZyC
z^-E{cC`Vm<2yS+2>RqKC32-=Bt<Pz|!){NGV~AY`>NyBr2K+JTFPNmE@cRqz(H&C8
z8j%*5dey_EN{hd^!}w;|lltP}Rso}H{rG^$Z4V-J;%oPonCQ$de_jw2oOIg=3C2(j
z4zgAMaK?EGiPAMV@v;nxB&(5(WLFYGqMOcBjxfmVMcJ;JKqA=TFGD(b8S9pZv5y{i
zjy`+B$?f>tkqZ!hUYu{bRJ>@H7VYg7W<r8wWh#!c@KqnZC5xcS&xS-Tzfw4N(*K~Z
zDje)}Mugm4Zs5zcHGnWOu+2*bz}_+u*O(#)Z88a#wuzJM*@i2)`$ao1_i;3lb(+t6
z&1>Ye>NHzmTVJ%3MoX{8Vy5=LU%Sl@$o*__%k{X-w2T%AQS`d{)|kph_NW5zR=rfV
zM(aEMh$rw=ZyV);Jl|++biXDJBSmK3LEkm$@;FOq;cw%P@%#CbZkK?c%D=+aAhVSb
zX9ca@;@VR(@09~}E@-$RK!jO8X)TPD`>^}tuDjh1Xdvn{;H_0{l5W4T5lUWCzh6Ld
zefh<o5Jl+7_Gl>L%<aQ)NvSQ@wm!OeJj>21&0V)7oRW(lJ%r?81KRX`g8(Q*=pS<p
f$^StE!td<YCvU!LEH;6@_mG^Fl4Oavk^lb%w8=hv

literal 0
HcmV?d00001

diff --git a/vignettes/custom_train_step_in_tensorflow.Rmd b/vignettes/custom_train_step_in_tensorflow.Rmd
index 06e262d4d..96551c921 100644
--- a/vignettes/custom_train_step_in_tensorflow.Rmd
+++ b/vignettes/custom_train_step_in_tensorflow.Rmd
@@ -127,9 +127,9 @@ model |> fit(x, y, epochs = 3)
 
 ```
 ## Epoch 1/3
-## 32/32 - 1s - 23ms/step - mae: 1.4339 - loss: 3.2271
+## 32/32 - 1s - 20ms/step - mae: 1.4339 - loss: 3.2271
 ## Epoch 2/3
-## 32/32 - 0s - 2ms/step - mae: 1.3605 - loss: 2.9034
+## 32/32 - 0s - 3ms/step - mae: 1.3605 - loss: 2.9034
 ## Epoch 3/3
 ## 32/32 - 0s - 2ms/step - mae: 1.2960 - loss: 2.6272
 ```
@@ -212,7 +212,7 @@ model |> fit(x, y, epochs = 3)
 
 ```
 ## Epoch 1/3
-## 32/32 - 1s - 20ms/step - loss: 2.5170 - mae: 1.2923
+## 32/32 - 1s - 18ms/step - loss: 2.5170 - mae: 1.2923
 ## Epoch 2/3
 ## 32/32 - 0s - 2ms/step - loss: 2.2689 - mae: 1.2241
 ## Epoch 3/3
@@ -282,11 +282,11 @@ model |> fit(x, y, sample_weight = sw, epochs = 3)
 
 ```
 ## Epoch 1/3
-## 32/32 - 1s - 23ms/step - mae: 1.3434 - loss: 0.1681
+## 32/32 - 1s - 20ms/step - mae: 1.3434 - loss: 0.1681
 ## Epoch 2/3
 ## 32/32 - 0s - 2ms/step - mae: 1.3364 - loss: 0.1394
 ## Epoch 3/3
-## 32/32 - 0s - 4ms/step - mae: 1.3286 - loss: 0.1148
+## 32/32 - 0s - 2ms/step - mae: 1.3286 - loss: 0.1148
 ```
 
 ## Providing your own evaluation step
@@ -508,7 +508,7 @@ gan |> fit(
 ```
 
 ```
-## 100/100 - 5s - 53ms/step - d_loss: 0.0000e+00 - g_loss: 0.0000e+00
+## 100/100 - 5s - 51ms/step - d_loss: 0.0000e+00 - g_loss: 0.0000e+00
 ```
 
 The ideas behind deep learning are simple, so why should their implementation be painful?
diff --git a/vignettes/distributed_training_with_tensorflow.Rmd b/vignettes/distributed_training_with_tensorflow.Rmd
index b65330a68..18727cc51 100644
--- a/vignettes/distributed_training_with_tensorflow.Rmd
+++ b/vignettes/distributed_training_with_tensorflow.Rmd
@@ -196,18 +196,18 @@ with(strategy$scope(), {
 
 ```
 ## Epoch 1/2
-## 782/782 - 7s - 9ms/step - loss: nan - sparse_categorical_accuracy: nan - val_loss: nan - val_sparse_categorical_accuracy: nan
+## 782/782 - 6s - 7ms/step - loss: 0.2616 - sparse_categorical_accuracy: 0.9232 - val_loss: 0.1510 - val_sparse_categorical_accuracy: 0.9524
 ## Epoch 2/2
-## 782/782 - 5s - 7ms/step - loss: nan - sparse_categorical_accuracy: nan - val_loss: nan - val_sparse_categorical_accuracy: nan
-## 157/157 - 1s - 5ms/step - loss: nan - sparse_categorical_accuracy: nan
+## 782/782 - 4s - 5ms/step - loss: 0.1044 - sparse_categorical_accuracy: 0.9682 - val_loss: 0.1004 - val_sparse_categorical_accuracy: 0.9675
+## 157/157 - 0s - 3ms/step - loss: 0.1004 - sparse_categorical_accuracy: 0.9683
 ```
 
 ```
 ## $loss
-## [1] NaN
+## [1] 0.1004203
 ##
 ## $sparse_categorical_accuracy
-## [1] NaN
+## [1] 0.9683
 ```
 
 ## Using callbacks to ensure fault tolerance
@@ -277,7 +277,7 @@ run_training(epochs = 1)
 ```
 
 ```
-## 782/782 - 5s - 7ms/step - loss: nan - sparse_categorical_accuracy: nan - val_loss: nan - val_sparse_categorical_accuracy: nan
+## 782/782 - 5s - 6ms/step - loss: 0.2575 - sparse_categorical_accuracy: 0.9237 - val_loss: 0.1428 - val_sparse_categorical_accuracy: 0.9537
 ```
 
 ``` r
@@ -286,7 +286,7 @@ run_training(epochs = 1)
 ```
 
 ```
-## 782/782 - 6s - 7ms/step - loss: nan - sparse_categorical_accuracy: nan - val_loss: nan - val_sparse_categorical_accuracy: nan
+## 782/782 - 5s - 6ms/step - loss: 0.1011 - sparse_categorical_accuracy: 0.9693 - val_loss: 0.1185 - val_sparse_categorical_accuracy: 0.9602
 ```
 
 ## `tf$data` performance tips
diff --git a/vignettes/distribution.Rmd b/vignettes/distribution.Rmd
index 9cd65610b..9887c9dd6 100644
--- a/vignettes/distribution.Rmd
+++ b/vignettes/distribution.Rmd
@@ -184,11 +184,11 @@ model |> fit(dataset, epochs = 3)
 
 ```
 ## Epoch 1/3
-## 8/8 - 0s - 40ms/step - loss: 1.1536
+## 8/8 - 0s - 42ms/step - loss: 1.1575
 ## Epoch 2/3
-## 8/8 - 0s - 5ms/step - loss: 1.0540
+## 8/8 - 0s - 4ms/step - loss: 1.0557
 ## Epoch 3/3
-## 8/8 - 0s - 6ms/step - loss: 1.0072
+## 8/8 - 0s - 4ms/step - loss: 1.0179
 ```
 
 ``` r
@@ -196,12 +196,12 @@ model |> evaluate(dataset)
 ```
 
 ```
-## 8/8 - 0s - 9ms/step - loss: 0.9620
+## 8/8 - 0s - 9ms/step - loss: 0.9707
 ```
 
 ```
 ## $loss
-## [1] 0.9620273
+## [1] 0.9707004
 ```
 
 
@@ -356,11 +356,11 @@ model |> fit(dataset, epochs = 3)
 
 ```
 ## Epoch 1/3
-## 8/8 - 0s - 46ms/step - loss: 1.1676
+## 8/8 - 0s - 48ms/step - loss: 1.1454
 ## Epoch 2/3
-## 8/8 - 0s - 4ms/step - loss: 1.1134
+## 8/8 - 0s - 2ms/step - loss: 1.0527
 ## Epoch 3/3
-## 8/8 - 0s - 5ms/step - loss: 1.1034
+## 8/8 - 0s - 2ms/step - loss: 1.0115
 ```
 
 ``` r
@@ -368,12 +368,12 @@ model |> evaluate(dataset)
 ```
 
 ```
-## 8/8 - 0s - 8ms/step - loss: 1.0676
+## 8/8 - 0s - 10ms/step - loss: 0.9674
 ```
 
 ```
 ## $loss
-## [1] 1.067567
+## [1] 0.9673891
 ```
 
 
diff --git a/vignettes/examples/nlp/neural_machine_translation_with_transformer.Rmd b/vignettes/examples/nlp/neural_machine_translation_with_transformer.Rmd
index a95368fab..82d600edc 100644
--- a/vignettes/examples/nlp/neural_machine_translation_with_transformer.Rmd
+++ b/vignettes/examples/nlp/neural_machine_translation_with_transformer.Rmd
@@ -566,7 +566,7 @@ transformer |> fit(train_ds, epochs = epochs,
 ```
 
 ```
-## 1297/1297 - 61s - 47ms/step - accuracy: 0.1590 - loss: 3.8930 - val_accuracy: 0.1533 - val_loss: 3.6121
+## 1297/1297 - 55s - 42ms/step - accuracy: 0.1592 - loss: 3.8833 - val_accuracy: 0.1691 - val_loss: 3.1841
 ```
 
 ## Decoding test sentences
diff --git a/vignettes/examples/nlp/text_classification_from_scratch.Rmd b/vignettes/examples/nlp/text_classification_from_scratch.Rmd
index 000b781b4..4c163e038 100644
--- a/vignettes/examples/nlp/text_classification_from_scratch.Rmd
+++ b/vignettes/examples/nlp/text_classification_from_scratch.Rmd
@@ -401,11 +401,11 @@ model |> fit(train_ds, validation_data = val_ds, epochs = epochs)
 
 ```
 ## Epoch 1/3
-## 625/625 - 6s - 10ms/step - accuracy: 0.6931 - loss: 0.5276 - val_accuracy: 0.8588 - val_loss: 0.3292
+## 625/625 - 5s - 8ms/step - accuracy: 0.6964 - loss: 0.5261 - val_accuracy: 0.8668 - val_loss: 0.3185
 ## Epoch 2/3
-## 625/625 - 4s - 6ms/step - accuracy: 0.9050 - loss: 0.2397 - val_accuracy: 0.8762 - val_loss: 0.3251
+## 625/625 - 2s - 3ms/step - accuracy: 0.9074 - loss: 0.2365 - val_accuracy: 0.8754 - val_loss: 0.3245
 ## Epoch 3/3
-## 625/625 - 3s - 4ms/step - accuracy: 0.9579 - loss: 0.1206 - val_accuracy: 0.8586 - val_loss: 0.4119
+## 625/625 - 2s - 3ms/step - accuracy: 0.9551 - loss: 0.1220 - val_accuracy: 0.8284 - val_loss: 0.4862
 ```
 
 ## Evaluate the model on the test set
@@ -416,15 +416,15 @@ model |> evaluate(test_ds)
 ```
 
 ```
-## 782/782 - 1s - 2ms/step - accuracy: 0.8437 - loss: 0.4384
+## 782/782 - 1s - 2ms/step - accuracy: 0.8115 - loss: 0.5299
 ```
 
 ```
 ## $accuracy
-## [1] 0.84368
+## [1] 0.81152
 ##
 ## $loss
-## [1] 0.4383631
+## [1] 0.5299004
 ```
 
 ## Make an end-to-end model
@@ -454,13 +454,13 @@ end_to_end_model |> evaluate(raw_test_ds)
 ```
 
 ```
-## 782/782 - 3s - 4ms/step - accuracy: 0.8437 - loss: 0.4384
+## 782/782 - 3s - 4ms/step - accuracy: 0.8115 - loss: 0.5299
 ```
 
 ```
 ## $accuracy
-## [1] 0.84368
+## [1] 0.81152
 ##
 ## $loss
-## [1] 0.4383632
+## [1] 0.5299006
 ```
diff --git a/vignettes/examples/structured_data/imbalanced_classification.Rmd b/vignettes/examples/structured_data/imbalanced_classification.Rmd
index 3d8227af5..3a8493848 100644
--- a/vignettes/examples/structured_data/imbalanced_classification.Rmd
+++ b/vignettes/examples/structured_data/imbalanced_classification.Rmd
@@ -118,7 +118,7 @@ counts
 ```
 ##
 ##      0      1
-## 227451    395
+## 227464    382
 ```
 
 ``` r
@@ -127,7 +127,7 @@ cat(sprintf("Number of positive samples in training data: %i (%.2f%% of total)",
 ```
 
 ```
-## Number of positive samples in training data: 395 (0.17% of total)
+## Number of positive samples in training data: 382 (0.17% of total)
 ```
 
 ``` r
@@ -227,65 +227,65 @@ model |> fit(
 
 ```
 ## Epoch 1/30
-## 112/112 - 5s - 41ms/step - fn: 36.0000 - fp: 30458.0000 - loss: 2.3118e-06 - precision: 0.0116 - recall: 0.9089 - tn: 196993.0000 - tp: 359.0000 - val_fn: 14.0000 - val_fp: 714.0000 - val_loss: 0.0906 - val_precision: 0.1041 - val_recall: 0.8557 - val_tn: 56150.0000 - val_tp: 83.0000
+## 112/112 - 4s - 37ms/step - fn: 36.0000 - fp: 29717.0000 - loss: 2.3629e-06 - precision: 0.0115 - recall: 0.9058 - tn: 197747.0000 - tp: 346.0000 - val_fn: 12.0000 - val_fp: 1039.0000 - val_loss: 0.0984 - val_precision: 0.0862 - val_recall: 0.8909 - val_tn: 55812.0000 - val_tp: 98.0000
 ## Epoch 2/30
-## 112/112 - 1s - 11ms/step - fn: 34.0000 - fp: 5698.0000 - loss: 1.4544e-06 - precision: 0.0596 - recall: 0.9139 - tn: 221753.0000 - tp: 361.0000 - val_fn: 6.0000 - val_fp: 2982.0000 - val_loss: 0.1611 - val_precision: 0.0296 - val_recall: 0.9381 - val_tn: 53882.0000 - val_tp: 91.0000
+## 112/112 - 1s - 9ms/step - fn: 30.0000 - fp: 6694.0000 - loss: 1.3508e-06 - precision: 0.0500 - recall: 0.9215 - tn: 220770.0000 - tp: 352.0000 - val_fn: 8.0000 - val_fp: 2418.0000 - val_loss: 0.1390 - val_precision: 0.0405 - val_recall: 0.9273 - val_tn: 54433.0000 - val_tp: 102.0000
 ## Epoch 3/30
-## 112/112 - 0s - 2ms/step - fn: 30.0000 - fp: 7494.0000 - loss: 1.1267e-06 - precision: 0.0464 - recall: 0.9241 - tn: 219957.0000 - tp: 365.0000 - val_fn: 9.0000 - val_fp: 2262.0000 - val_loss: 0.1346 - val_precision: 0.0374 - val_recall: 0.9072 - val_tn: 54602.0000 - val_tp: 88.0000
+## 112/112 - 0s - 2ms/step - fn: 27.0000 - fp: 9781.0000 - loss: 1.3090e-06 - precision: 0.0350 - recall: 0.9293 - tn: 217683.0000 - tp: 355.0000 - val_fn: 10.0000 - val_fp: 984.0000 - val_loss: 0.1063 - val_precision: 0.0923 - val_recall: 0.9091 - val_tn: 55867.0000 - val_tp: 100.0000
 ## Epoch 4/30
-## 112/112 - 0s - 2ms/step - fn: 25.0000 - fp: 5976.0000 - loss: 9.9649e-07 - precision: 0.0583 - recall: 0.9367 - tn: 221475.0000 - tp: 370.0000 - val_fn: 9.0000 - val_fp: 1593.0000 - val_loss: 0.0781 - val_precision: 0.0523 - val_recall: 0.9072 - val_tn: 55271.0000 - val_tp: 88.0000
+## 112/112 - 0s - 2ms/step - fn: 32.0000 - fp: 6894.0000 - loss: 1.1099e-06 - precision: 0.0483 - recall: 0.9162 - tn: 220570.0000 - tp: 350.0000 - val_fn: 9.0000 - val_fp: 1557.0000 - val_loss: 0.0857 - val_precision: 0.0609 - val_recall: 0.9182 - val_tn: 55294.0000 - val_tp: 101.0000
 ## Epoch 5/30
-## 112/112 - 0s - 2ms/step - fn: 22.0000 - fp: 7713.0000 - loss: 9.5664e-07 - precision: 0.0461 - recall: 0.9443 - tn: 219738.0000 - tp: 373.0000 - val_fn: 10.0000 - val_fp: 613.0000 - val_loss: 0.0444 - val_precision: 0.1243 - val_recall: 0.8969 - val_tn: 56251.0000 - val_tp: 87.0000
+## 112/112 - 0s - 2ms/step - fn: 23.0000 - fp: 8892.0000 - loss: 9.2992e-07 - precision: 0.0388 - recall: 0.9398 - tn: 218572.0000 - tp: 359.0000 - val_fn: 7.0000 - val_fp: 1492.0000 - val_loss: 0.0665 - val_precision: 0.0646 - val_recall: 0.9364 - val_tn: 55359.0000 - val_tp: 103.0000
 ## Epoch 6/30
-## 112/112 - 0s - 2ms/step - fn: 18.0000 - fp: 8517.0000 - loss: 9.6051e-07 - precision: 0.0424 - recall: 0.9544 - tn: 218934.0000 - tp: 377.0000 - val_fn: 12.0000 - val_fp: 1033.0000 - val_loss: 0.0643 - val_precision: 0.0760 - val_recall: 0.8763 - val_tn: 55831.0000 - val_tp: 85.0000
+## 112/112 - 0s - 2ms/step - fn: 18.0000 - fp: 8209.0000 - loss: 7.7549e-07 - precision: 0.0425 - recall: 0.9529 - tn: 219255.0000 - tp: 364.0000 - val_fn: 7.0000 - val_fp: 2749.0000 - val_loss: 0.1010 - val_precision: 0.0361 - val_recall: 0.9364 - val_tn: 54102.0000 - val_tp: 103.0000
 ## Epoch 7/30
-## 112/112 - 0s - 2ms/step - fn: 17.0000 - fp: 6409.0000 - loss: 7.3375e-07 - precision: 0.0557 - recall: 0.9570 - tn: 221042.0000 - tp: 378.0000 - val_fn: 10.0000 - val_fp: 1549.0000 - val_loss: 0.0697 - val_precision: 0.0532 - val_recall: 0.8969 - val_tn: 55315.0000 - val_tp: 87.0000
+## 112/112 - 0s - 2ms/step - fn: 15.0000 - fp: 7189.0000 - loss: 6.9032e-07 - precision: 0.0486 - recall: 0.9607 - tn: 220275.0000 - tp: 367.0000 - val_fn: 12.0000 - val_fp: 934.0000 - val_loss: 0.0539 - val_precision: 0.0950 - val_recall: 0.8909 - val_tn: 55917.0000 - val_tp: 98.0000
 ## Epoch 8/30
-## 112/112 - 0s - 2ms/step - fn: 14.0000 - fp: 6080.0000 - loss: 6.5853e-07 - precision: 0.0590 - recall: 0.9646 - tn: 221371.0000 - tp: 381.0000 - val_fn: 10.0000 - val_fp: 2677.0000 - val_loss: 0.1178 - val_precision: 0.0315 - val_recall: 0.8969 - val_tn: 54187.0000 - val_tp: 87.0000
+## 112/112 - 0s - 2ms/step - fn: 18.0000 - fp: 7759.0000 - loss: 8.2432e-07 - precision: 0.0448 - recall: 0.9529 - tn: 219705.0000 - tp: 364.0000 - val_fn: 10.0000 - val_fp: 1481.0000 - val_loss: 0.0777 - val_precision: 0.0633 - val_recall: 0.9091 - val_tn: 55370.0000 - val_tp: 100.0000
 ## Epoch 9/30
-## 112/112 - 0s - 2ms/step - fn: 11.0000 - fp: 6592.0000 - loss: 5.9467e-07 - precision: 0.0550 - recall: 0.9722 - tn: 220859.0000 - tp: 384.0000 - val_fn: 9.0000 - val_fp: 2094.0000 - val_loss: 0.0806 - val_precision: 0.0403 - val_recall: 0.9072 - val_tn: 54770.0000 - val_tp: 88.0000
+## 112/112 - 0s - 2ms/step - fn: 16.0000 - fp: 7565.0000 - loss: 7.2956e-07 - precision: 0.0461 - recall: 0.9581 - tn: 219899.0000 - tp: 366.0000 - val_fn: 8.0000 - val_fp: 2075.0000 - val_loss: 0.0718 - val_precision: 0.0469 - val_recall: 0.9273 - val_tn: 54776.0000 - val_tp: 102.0000
 ## Epoch 10/30
-## 112/112 - 0s - 2ms/step - fn: 7.0000 - fp: 6334.0000 - loss: 5.3129e-07 - precision: 0.0577 - recall: 0.9823 - tn: 221117.0000 - tp: 388.0000 - val_fn: 10.0000 - val_fp: 850.0000 - val_loss: 0.0396 - val_precision: 0.0928 - val_recall: 0.8969 - val_tn: 56014.0000 - val_tp: 87.0000
+## 112/112 - 0s - 2ms/step - fn: 10.0000 - fp: 7374.0000 - loss: 5.9522e-07 - precision: 0.0480 - recall: 0.9738 - tn: 220090.0000 - tp: 372.0000 - val_fn: 11.0000 - val_fp: 995.0000 - val_loss: 0.0425 - val_precision: 0.0905 - val_recall: 0.9000 - val_tn: 55856.0000 - val_tp: 99.0000
 ## Epoch 11/30
-## 112/112 - 0s - 2ms/step - fn: 8.0000 - fp: 5922.0000 - loss: 5.0009e-07 - precision: 0.0613 - recall: 0.9797 - tn: 221529.0000 - tp: 387.0000 - val_fn: 12.0000 - val_fp: 1488.0000 - val_loss: 0.0506 - val_precision: 0.0540 - val_recall: 0.8763 - val_tn: 55376.0000 - val_tp: 85.0000
+## 112/112 - 0s - 2ms/step - fn: 10.0000 - fp: 6586.0000 - loss: 6.2194e-07 - precision: 0.0535 - recall: 0.9738 - tn: 220878.0000 - tp: 372.0000 - val_fn: 9.0000 - val_fp: 5097.0000 - val_loss: 0.2103 - val_precision: 0.0194 - val_recall: 0.9182 - val_tn: 51754.0000 - val_tp: 101.0000
 ## Epoch 12/30
-## 112/112 - 0s - 2ms/step - fn: 8.0000 - fp: 7243.0000 - loss: 5.7125e-07 - precision: 0.0507 - recall: 0.9797 - tn: 220208.0000 - tp: 387.0000 - val_fn: 11.0000 - val_fp: 1014.0000 - val_loss: 0.0414 - val_precision: 0.0782 - val_recall: 0.8866 - val_tn: 55850.0000 - val_tp: 86.0000
+## 112/112 - 0s - 2ms/step - fn: 6.0000 - fp: 8285.0000 - loss: 6.0991e-07 - precision: 0.0434 - recall: 0.9843 - tn: 219179.0000 - tp: 376.0000 - val_fn: 9.0000 - val_fp: 1926.0000 - val_loss: 0.0781 - val_precision: 0.0498 - val_recall: 0.9182 - val_tn: 54925.0000 - val_tp: 101.0000
 ## Epoch 13/30
-## 112/112 - 0s - 2ms/step - fn: 6.0000 - fp: 4141.0000 - loss: 3.7800e-07 - precision: 0.0859 - recall: 0.9848 - tn: 223310.0000 - tp: 389.0000 - val_fn: 8.0000 - val_fp: 2491.0000 - val_loss: 0.0986 - val_precision: 0.0345 - val_recall: 0.9175 - val_tn: 54373.0000 - val_tp: 89.0000
+## 112/112 - 0s - 2ms/step - fn: 7.0000 - fp: 6864.0000 - loss: 5.7087e-07 - precision: 0.0518 - recall: 0.9817 - tn: 220600.0000 - tp: 375.0000 - val_fn: 8.0000 - val_fp: 1900.0000 - val_loss: 0.0837 - val_precision: 0.0509 - val_recall: 0.9273 - val_tn: 54951.0000 - val_tp: 102.0000
 ## Epoch 14/30
-## 112/112 - 0s - 2ms/step - fn: 6.0000 - fp: 6118.0000 - loss: 5.4537e-07 - precision: 0.0598 - recall: 0.9848 - tn: 221333.0000 - tp: 389.0000 - val_fn: 9.0000 - val_fp: 2488.0000 - val_loss: 0.1019 - val_precision: 0.0342 - val_recall: 0.9072 - val_tn: 54376.0000 - val_tp: 88.0000
+## 112/112 - 0s - 2ms/step - fn: 7.0000 - fp: 7663.0000 - loss: 5.7664e-07 - precision: 0.0467 - recall: 0.9817 - tn: 219801.0000 - tp: 375.0000 - val_fn: 8.0000 - val_fp: 1540.0000 - val_loss: 0.0656 - val_precision: 0.0621 - val_recall: 0.9273 - val_tn: 55311.0000 - val_tp: 102.0000
 ## Epoch 15/30
-## 112/112 - 0s - 2ms/step - fn: 4.0000 - fp: 4646.0000 - loss: 4.0941e-07 - precision: 0.0776 - recall: 0.9899 - tn: 222805.0000 - tp: 391.0000 - val_fn: 8.0000 - val_fp: 2949.0000 - val_loss: 0.1263 - val_precision: 0.0293 - val_recall: 0.9175 - val_tn: 53915.0000 - val_tp: 89.0000
+## 112/112 - 0s - 2ms/step - fn: 6.0000 - fp: 4607.0000 - loss: 3.9004e-07 - precision: 0.0755 - recall: 0.9843 - tn: 222857.0000 - tp: 376.0000 - val_fn: 11.0000 - val_fp: 2546.0000 - val_loss: 0.0994 - val_precision: 0.0374 - val_recall: 0.9000 - val_tn: 54305.0000 - val_tp: 99.0000
 ## Epoch 16/30
-## 112/112 - 0s - 2ms/step - fn: 10.0000 - fp: 7978.0000 - loss: 8.1907e-07 - precision: 0.0460 - recall: 0.9747 - tn: 219473.0000 - tp: 385.0000 - val_fn: 8.0000 - val_fp: 3430.0000 - val_loss: 0.2915 - val_precision: 0.0253 - val_recall: 0.9175 - val_tn: 53434.0000 - val_tp: 89.0000
+## 112/112 - 0s - 2ms/step - fn: 6.0000 - fp: 6195.0000 - loss: 5.1274e-07 - precision: 0.0572 - recall: 0.9843 - tn: 221269.0000 - tp: 376.0000 - val_fn: 9.0000 - val_fp: 1652.0000 - val_loss: 0.0693 - val_precision: 0.0576 - val_recall: 0.9182 - val_tn: 55199.0000 - val_tp: 101.0000
 ## Epoch 17/30
-## 112/112 - 0s - 2ms/step - fn: 23.0000 - fp: 12673.0000 - loss: 3.1681e-06 - precision: 0.0285 - recall: 0.9418 - tn: 214778.0000 - tp: 372.0000 - val_fn: 12.0000 - val_fp: 3416.0000 - val_loss: 0.7861 - val_precision: 0.0243 - val_recall: 0.8763 - val_tn: 53448.0000 - val_tp: 85.0000
+## 112/112 - 0s - 2ms/step - fn: 6.0000 - fp: 5367.0000 - loss: 4.6364e-07 - precision: 0.0655 - recall: 0.9843 - tn: 222097.0000 - tp: 376.0000 - val_fn: 10.0000 - val_fp: 1378.0000 - val_loss: 0.0575 - val_precision: 0.0677 - val_recall: 0.9091 - val_tn: 55473.0000 - val_tp: 100.0000
 ## Epoch 18/30
-## 112/112 - 0s - 2ms/step - fn: 23.0000 - fp: 11244.0000 - loss: 2.3250e-06 - precision: 0.0320 - recall: 0.9418 - tn: 216207.0000 - tp: 372.0000 - val_fn: 11.0000 - val_fp: 1717.0000 - val_loss: 0.1895 - val_precision: 0.0477 - val_recall: 0.8866 - val_tn: 55147.0000 - val_tp: 86.0000
+## 112/112 - 0s - 2ms/step - fn: 2.0000 - fp: 3696.0000 - loss: 2.9266e-07 - precision: 0.0932 - recall: 0.9948 - tn: 223768.0000 - tp: 380.0000 - val_fn: 11.0000 - val_fp: 824.0000 - val_loss: 0.0375 - val_precision: 0.1073 - val_recall: 0.9000 - val_tn: 56027.0000 - val_tp: 99.0000
 ## Epoch 19/30
-## 112/112 - 0s - 2ms/step - fn: 16.0000 - fp: 8824.0000 - loss: 1.3502e-06 - precision: 0.0412 - recall: 0.9595 - tn: 218627.0000 - tp: 379.0000 - val_fn: 6.0000 - val_fp: 5161.0000 - val_loss: 0.1947 - val_precision: 0.0173 - val_recall: 0.9381 - val_tn: 51703.0000 - val_tp: 91.0000
+## 112/112 - 0s - 2ms/step - fn: 3.0000 - fp: 3638.0000 - loss: 3.6945e-07 - precision: 0.0943 - recall: 0.9921 - tn: 223826.0000 - tp: 379.0000 - val_fn: 4.0000 - val_fp: 4639.0000 - val_loss: 0.5475 - val_precision: 0.0223 - val_recall: 0.9636 - val_tn: 52212.0000 - val_tp: 106.0000
 ## Epoch 20/30
-## 112/112 - 0s - 2ms/step - fn: 10.0000 - fp: 8621.0000 - loss: 1.0916e-06 - precision: 0.0427 - recall: 0.9747 - tn: 218830.0000 - tp: 385.0000 - val_fn: 12.0000 - val_fp: 1107.0000 - val_loss: 0.1877 - val_precision: 0.0713 - val_recall: 0.8763 - val_tn: 55757.0000 - val_tp: 85.0000
+## 112/112 - 0s - 2ms/step - fn: 26.0000 - fp: 11743.0000 - loss: 4.2168e-06 - precision: 0.0294 - recall: 0.9319 - tn: 215721.0000 - tp: 356.0000 - val_fn: 104.0000 - val_fp: 395.0000 - val_loss: 0.0507 - val_precision: 0.0150 - val_recall: 0.0545 - val_tn: 56456.0000 - val_tp: 6.0000
 ## Epoch 21/30
-## 112/112 - 0s - 2ms/step - fn: 10.0000 - fp: 7096.0000 - loss: 1.0528e-06 - precision: 0.0515 - recall: 0.9747 - tn: 220355.0000 - tp: 385.0000 - val_fn: 11.0000 - val_fp: 1099.0000 - val_loss: 0.0539 - val_precision: 0.0726 - val_recall: 0.8866 - val_tn: 55765.0000 - val_tp: 86.0000
+## 112/112 - 0s - 2ms/step - fn: 41.0000 - fp: 8860.0000 - loss: 5.8042e-06 - precision: 0.0371 - recall: 0.8927 - tn: 218604.0000 - tp: 341.0000 - val_fn: 10.0000 - val_fp: 2791.0000 - val_loss: 0.2183 - val_precision: 0.0346 - val_recall: 0.9091 - val_tn: 54060.0000 - val_tp: 100.0000
 ## Epoch 22/30
-## 112/112 - 0s - 3ms/step - fn: 11.0000 - fp: 5263.0000 - loss: 5.9445e-07 - precision: 0.0680 - recall: 0.9722 - tn: 222188.0000 - tp: 384.0000 - val_fn: 10.0000 - val_fp: 1399.0000 - val_loss: 0.0632 - val_precision: 0.0585 - val_recall: 0.8969 - val_tn: 55465.0000 - val_tp: 87.0000
+## 112/112 - 0s - 2ms/step - fn: 15.0000 - fp: 9031.0000 - loss: 1.0670e-06 - precision: 0.0391 - recall: 0.9607 - tn: 218433.0000 - tp: 367.0000 - val_fn: 10.0000 - val_fp: 1275.0000 - val_loss: 0.0764 - val_precision: 0.0727 - val_recall: 0.9091 - val_tn: 55576.0000 - val_tp: 100.0000
 ## Epoch 23/30
-## 112/112 - 0s - 4ms/step - fn: 8.0000 - fp: 5412.0000 - loss: 6.1044e-07 - precision: 0.0667 - recall: 0.9797 - tn: 222039.0000 - tp: 387.0000 - val_fn: 12.0000 - val_fp: 873.0000 - val_loss: 0.0727 - val_precision: 0.0887 - val_recall: 0.8763 - val_tn: 55991.0000 - val_tp: 85.0000
+## 112/112 - 0s - 2ms/step - fn: 10.0000 - fp: 5141.0000 - loss: 6.9193e-07 - precision: 0.0675 - recall: 0.9738 - tn: 222323.0000 - tp: 372.0000 - val_fn: 11.0000 - val_fp: 1872.0000 - val_loss: 0.0824 - val_precision: 0.0502 - val_recall: 0.9000 - val_tn: 54979.0000 - val_tp: 99.0000
 ## Epoch 24/30
-## 112/112 - 0s - 3ms/step - fn: 5.0000 - fp: 3469.0000 - loss: 4.1501e-07 - precision: 0.1011 - recall: 0.9873 - tn: 223982.0000 - tp: 390.0000 - val_fn: 11.0000 - val_fp: 1369.0000 - val_loss: 0.0758 - val_precision: 0.0591 - val_recall: 0.8866 - val_tn: 55495.0000 - val_tp: 86.0000
+## 112/112 - 0s - 2ms/step - fn: 2.0000 - fp: 4232.0000 - loss: 4.2135e-07 - precision: 0.0824 - recall: 0.9948 - tn: 223232.0000 - tp: 380.0000 - val_fn: 11.0000 - val_fp: 976.0000 - val_loss: 0.0488 - val_precision: 0.0921 - val_recall: 0.9000 - val_tn: 55875.0000 - val_tp: 99.0000
 ## Epoch 25/30
-## 112/112 - 0s - 2ms/step - fn: 7.0000 - fp: 4489.0000 - loss: 6.9404e-07 - precision: 0.0796 - recall: 0.9823 - tn: 222962.0000 - tp: 388.0000 - val_fn: 11.0000 - val_fp: 1580.0000 - val_loss: 0.1499 - val_precision: 0.0516 - val_recall: 0.8866 - val_tn: 55284.0000 - val_tp: 86.0000
+## 112/112 - 0s - 2ms/step - fn: 3.0000 - fp: 4149.0000 - loss: 3.7232e-07 - precision: 0.0837 - recall: 0.9921 - tn: 223315.0000 - tp: 379.0000 - val_fn: 13.0000 - val_fp: 359.0000 - val_loss: 0.0229 - val_precision: 0.2127 - val_recall: 0.8818 - val_tn: 56492.0000 - val_tp: 97.0000
 ## Epoch 26/30
-## 112/112 - 0s - 2ms/step - fn: 4.0000 - fp: 3469.0000 - loss: 4.4795e-07 - precision: 0.1013 - recall: 0.9899 - tn: 223982.0000 - tp: 391.0000 - val_fn: 12.0000 - val_fp: 951.0000 - val_loss: 0.0502 - val_precision: 0.0820 - val_recall: 0.8763 - val_tn: 55913.0000 - val_tp: 85.0000
+## 112/112 - 0s - 2ms/step - fn: 3.0000 - fp: 3757.0000 - loss: 3.5368e-07 - precision: 0.0916 - recall: 0.9921 - tn: 223707.0000 - tp: 379.0000 - val_fn: 12.0000 - val_fp: 1111.0000 - val_loss: 0.0569 - val_precision: 0.0811 - val_recall: 0.8909 - val_tn: 55740.0000 - val_tp: 98.0000
 ## Epoch 27/30
-## 112/112 - 0s - 2ms/step - fn: 5.0000 - fp: 4854.0000 - loss: 7.1714e-07 - precision: 0.0744 - recall: 0.9873 - tn: 222597.0000 - tp: 390.0000 - val_fn: 13.0000 - val_fp: 1354.0000 - val_loss: 0.0818 - val_precision: 0.0584 - val_recall: 0.8660 - val_tn: 55510.0000 - val_tp: 84.0000
+## 112/112 - 0s - 2ms/step - fn: 2.0000 - fp: 3220.0000 - loss: 2.5512e-07 - precision: 0.1056 - recall: 0.9948 - tn: 224244.0000 - tp: 380.0000 - val_fn: 11.0000 - val_fp: 839.0000 - val_loss: 0.0396 - val_precision: 0.1055 - val_recall: 0.9000 - val_tn: 56012.0000 - val_tp: 99.0000
 ## Epoch 28/30
-## 112/112 - 0s - 3ms/step - fn: 6.0000 - fp: 4904.0000 - loss: 4.4107e-07 - precision: 0.0735 - recall: 0.9848 - tn: 222547.0000 - tp: 389.0000 - val_fn: 12.0000 - val_fp: 651.0000 - val_loss: 0.0420 - val_precision: 0.1155 - val_recall: 0.8763 - val_tn: 56213.0000 - val_tp: 85.0000
+## 112/112 - 0s - 2ms/step - fn: 0.0000e+00 - fp: 2287.0000 - loss: 1.8121e-07 - precision: 0.1431 - recall: 1.0000 - tn: 225177.0000 - tp: 382.0000 - val_fn: 12.0000 - val_fp: 425.0000 - val_loss: 0.0272 - val_precision: 0.1874 - val_recall: 0.8909 - val_tn: 56426.0000 - val_tp: 98.0000
 ## Epoch 29/30
-## 112/112 - 0s - 3ms/step - fn: 4.0000 - fp: 3909.0000 - loss: 4.2965e-07 - precision: 0.0909 - recall: 0.9899 - tn: 223542.0000 - tp: 391.0000 - val_fn: 12.0000 - val_fp: 817.0000 - val_loss: 0.0405 - val_precision: 0.0942 - val_recall: 0.8763 - val_tn: 56047.0000 - val_tp: 85.0000
+## 112/112 - 0s - 2ms/step - fn: 0.0000e+00 - fp: 1435.0000 - loss: 1.3034e-07 - precision: 0.2102 - recall: 1.0000 - tn: 226029.0000 - tp: 382.0000 - val_fn: 13.0000 - val_fp: 321.0000 - val_loss: 0.0250 - val_precision: 0.2321 - val_recall: 0.8818 - val_tn: 56530.0000 - val_tp: 97.0000
 ## Epoch 30/30
-## 112/112 - 0s - 2ms/step - fn: 2.0000 - fp: 2985.0000 - loss: 2.8492e-07 - precision: 0.1163 - recall: 0.9949 - tn: 224466.0000 - tp: 393.0000 - val_fn: 12.0000 - val_fp: 1040.0000 - val_loss: 0.0593 - val_precision: 0.0756 - val_recall: 0.8763 - val_tn: 55824.0000 - val_tp: 85.0000
+## 112/112 - 0s - 2ms/step - fn: 0.0000e+00 - fp: 1365.0000 - loss: 1.2132e-07 - precision: 0.2187 - recall: 1.0000 - tn: 226099.0000 - tp: 382.0000 - val_fn: 12.0000 - val_fp: 241.0000 - val_loss: 0.0183 - val_precision: 0.2891 - val_recall: 0.8909 - val_tn: 56610.0000 - val_tp: 98.0000
 ```
 
 
@@ -296,7 +296,7 @@ val_pred <- model %>%
 ```
 
 ```
-## 1781/1781 - 1s - 436us/step
+## 1781/1781 - 1s - 427us/step
 ```
 
 ``` r
@@ -305,7 +305,7 @@ cat(sprintf("Validation accuracy: %.2f", mean(pred_correct)))
 ```
 
 ```
-## Validation accuracy: 0.98
+## Validation accuracy: 1.00
 ```
 
 ``` r
@@ -323,12 +323,12 @@ At the end of training, out of
 are:
 
 - Correctly identifying
-  85 of them as
+  98 of them as
   fraudulent
 - Missing 12
   fraudulent transactions
 - At the cost of incorrectly flagging
-  1,040 legitimate
+  241 legitimate
   transactions
 
 In the real world, one would put an even higher weight on class 1,
diff --git a/vignettes/examples/structured_data/structured_data_classification_with_feature_space.Rmd b/vignettes/examples/structured_data/structured_data_classification_with_feature_space.Rmd
index 5e6f6945c..6d50cf468 100644
--- a/vignettes/examples/structured_data/structured_data_classification_with_feature_space.Rmd
+++ b/vignettes/examples/structured_data/structured_data_classification_with_feature_space.Rmd
@@ -201,20 +201,20 @@ cat("Target: "); str(y)
 
 ```
 ## Input: List of 13
-##  $ age     :<tf.Tensor: shape=(), dtype=int32, numpy=63>
-##  $ sex     :<tf.Tensor: shape=(), dtype=int32, numpy=0>
+##  $ age     :<tf.Tensor: shape=(), dtype=int32, numpy=68>
+##  $ sex     :<tf.Tensor: shape=(), dtype=int32, numpy=1>
 ##  $ cp      :<tf.Tensor: shape=(), dtype=int32, numpy=4>
-##  $ trestbps:<tf.Tensor: shape=(), dtype=int32, numpy=124>
-##  $ chol    :<tf.Tensor: shape=(), dtype=int32, numpy=197>
-##  $ fbs     :<tf.Tensor: shape=(), dtype=int32, numpy=0>
+##  $ trestbps:<tf.Tensor: shape=(), dtype=int32, numpy=144>
+##  $ chol    :<tf.Tensor: shape=(), dtype=int32, numpy=193>
+##  $ fbs     :<tf.Tensor: shape=(), dtype=int32, numpy=1>
 ##  $ restecg :<tf.Tensor: shape=(), dtype=int32, numpy=0>
-##  $ thalach :<tf.Tensor: shape=(), dtype=int32, numpy=136>
-##  $ exang   :<tf.Tensor: shape=(), dtype=int32, numpy=1>
-##  $ oldpeak :<tf.Tensor: shape=(), dtype=float32, numpy=0.0>
+##  $ thalach :<tf.Tensor: shape=(), dtype=int32, numpy=141>
+##  $ exang   :<tf.Tensor: shape=(), dtype=int32, numpy=0>
+##  $ oldpeak :<tf.Tensor: shape=(), dtype=float32, numpy=3.4000000953674316>
 ##  $ slope   :<tf.Tensor: shape=(), dtype=int32, numpy=2>
-##  $ ca      :<tf.Tensor: shape=(), dtype=int32, numpy=0>
-##  $ thal    :<tf.Tensor: shape=(), dtype=string, numpy=b'normal'>
-## Target: <tf.Tensor: shape=(), dtype=int32, numpy=0>
+##  $ ca      :<tf.Tensor: shape=(), dtype=int32, numpy=2>
+##  $ thal    :<tf.Tensor: shape=(), dtype=string, numpy=b'reversible'>
+## Target: <tf.Tensor: shape=(), dtype=int32, numpy=1>
 ```
 
 Let's batch the datasets:
@@ -377,7 +377,7 @@ preprocessed_x
 ##  ...
 ##  [0. 0. 0. ... 0. 0. 0.]
 ##  [0. 0. 0. ... 0. 0. 0.]
-##  [0. 0. 0. ... 1. 0. 0.]], shape=(32, 136), dtype=float32)
+##  [0. 0. 0. ... 0. 0. 0.]], shape=(32, 136), dtype=float32)
 ```
 
 ## Two ways to manage preprocessing: as part of the `tf.data` pipeline, or in the model itself
@@ -463,45 +463,45 @@ training_model |> fit(
 
 ```
 ## Epoch 1/20
-## 8/8 - 3s - 325ms/step - accuracy: 0.4481 - loss: 0.7302 - val_accuracy: 0.6000 - val_loss: 0.6799
+## 8/8 - 2s - 262ms/step - accuracy: 0.4108 - loss: 0.7398 - val_accuracy: 0.3667 - val_loss: 0.7218
 ## Epoch 2/20
-## 8/8 - 0s - 28ms/step - accuracy: 0.5851 - loss: 0.6797 - val_accuracy: 0.7000 - val_loss: 0.6329
+## 8/8 - 0s - 26ms/step - accuracy: 0.5934 - loss: 0.6798 - val_accuracy: 0.5333 - val_loss: 0.6670
 ## Epoch 3/20
-## 8/8 - 0s - 31ms/step - accuracy: 0.6141 - loss: 0.6458 - val_accuracy: 0.7333 - val_loss: 0.5951
+## 8/8 - 0s - 26ms/step - accuracy: 0.6473 - loss: 0.6384 - val_accuracy: 0.6333 - val_loss: 0.6207
 ## Epoch 4/20
-## 8/8 - 0s - 32ms/step - accuracy: 0.7220 - loss: 0.5843 - val_accuracy: 0.7667 - val_loss: 0.5644
+## 8/8 - 0s - 26ms/step - accuracy: 0.7012 - loss: 0.5829 - val_accuracy: 0.7167 - val_loss: 0.5800
 ## Epoch 5/20
-## 8/8 - 0s - 32ms/step - accuracy: 0.7552 - loss: 0.5539 - val_accuracy: 0.7667 - val_loss: 0.5367
+## 8/8 - 0s - 27ms/step - accuracy: 0.6846 - loss: 0.5780 - val_accuracy: 0.7500 - val_loss: 0.5445
 ## Epoch 6/20
-## 8/8 - 0s - 29ms/step - accuracy: 0.7884 - loss: 0.5248 - val_accuracy: 0.7667 - val_loss: 0.5146
+## 8/8 - 0s - 26ms/step - accuracy: 0.7718 - loss: 0.5595 - val_accuracy: 0.7667 - val_loss: 0.5134
 ## Epoch 7/20
-## 8/8 - 0s - 32ms/step - accuracy: 0.7469 - loss: 0.5084 - val_accuracy: 0.7667 - val_loss: 0.4960
+## 8/8 - 0s - 26ms/step - accuracy: 0.7801 - loss: 0.5166 - val_accuracy: 0.8167 - val_loss: 0.4868
 ## Epoch 8/20
-## 8/8 - 0s - 30ms/step - accuracy: 0.7718 - loss: 0.4921 - val_accuracy: 0.7833 - val_loss: 0.4804
+## 8/8 - 0s - 26ms/step - accuracy: 0.7718 - loss: 0.5053 - val_accuracy: 0.8333 - val_loss: 0.4626
 ## Epoch 9/20
-## 8/8 - 0s - 30ms/step - accuracy: 0.8216 - loss: 0.4617 - val_accuracy: 0.8000 - val_loss: 0.4668
+## 8/8 - 0s - 28ms/step - accuracy: 0.8091 - loss: 0.4774 - val_accuracy: 0.8333 - val_loss: 0.4401
 ## Epoch 10/20
-## 8/8 - 0s - 30ms/step - accuracy: 0.8008 - loss: 0.4550 - val_accuracy: 0.8333 - val_loss: 0.4551
+## 8/8 - 0s - 27ms/step - accuracy: 0.8050 - loss: 0.4682 - val_accuracy: 0.8500 - val_loss: 0.4197
 ## Epoch 11/20
-## 8/8 - 0s - 30ms/step - accuracy: 0.8133 - loss: 0.4299 - val_accuracy: 0.8500 - val_loss: 0.4447
+## 8/8 - 0s - 26ms/step - accuracy: 0.8133 - loss: 0.4541 - val_accuracy: 0.8667 - val_loss: 0.4011
 ## Epoch 12/20
-## 8/8 - 0s - 29ms/step - accuracy: 0.8506 - loss: 0.3980 - val_accuracy: 0.8500 - val_loss: 0.4372
+## 8/8 - 0s - 27ms/step - accuracy: 0.8216 - loss: 0.4360 - val_accuracy: 0.8667 - val_loss: 0.3843
 ## Epoch 13/20
-## 8/8 - 0s - 29ms/step - accuracy: 0.8008 - loss: 0.4105 - val_accuracy: 0.8500 - val_loss: 0.4308
+## 8/8 - 0s - 26ms/step - accuracy: 0.8174 - loss: 0.4081 - val_accuracy: 0.8667 - val_loss: 0.3684
 ## Epoch 14/20
-## 8/8 - 0s - 29ms/step - accuracy: 0.8548 - loss: 0.3820 - val_accuracy: 0.8500 - val_loss: 0.4254
+## 8/8 - 0s - 27ms/step - accuracy: 0.8423 - loss: 0.4034 - val_accuracy: 0.8667 - val_loss: 0.3531
 ## Epoch 15/20
-## 8/8 - 0s - 31ms/step - accuracy: 0.8589 - loss: 0.3669 - val_accuracy: 0.8500 - val_loss: 0.4205
+## 8/8 - 0s - 26ms/step - accuracy: 0.8174 - loss: 0.3956 - val_accuracy: 0.8833 - val_loss: 0.3397
 ## Epoch 16/20
-## 8/8 - 0s - 31ms/step - accuracy: 0.8257 - loss: 0.3786 - val_accuracy: 0.8333 - val_loss: 0.4142
+## 8/8 - 0s - 27ms/step - accuracy: 0.8423 - loss: 0.3952 - val_accuracy: 0.8833 - val_loss: 0.3274
 ## Epoch 17/20
-## 8/8 - 0s - 30ms/step - accuracy: 0.8548 - loss: 0.3524 - val_accuracy: 0.8333 - val_loss: 0.4105
+## 8/8 - 0s - 27ms/step - accuracy: 0.8548 - loss: 0.3445 - val_accuracy: 0.8833 - val_loss: 0.3169
 ## Epoch 18/20
-## 8/8 - 0s - 31ms/step - accuracy: 0.8631 - loss: 0.3530 - val_accuracy: 0.8333 - val_loss: 0.4078
+## 8/8 - 0s - 27ms/step - accuracy: 0.8382 - loss: 0.3609 - val_accuracy: 0.8833 - val_loss: 0.3080
 ## Epoch 19/20
-## 8/8 - 0s - 30ms/step - accuracy: 0.8548 - loss: 0.3440 - val_accuracy: 0.8333 - val_loss: 0.4044
+## 8/8 - 0s - 26ms/step - accuracy: 0.8714 - loss: 0.3412 - val_accuracy: 0.8833 - val_loss: 0.2986
 ## Epoch 20/20
-## 8/8 - 0s - 30ms/step - accuracy: 0.8838 - loss: 0.3303 - val_accuracy: 0.8333 - val_loss: 0.4021
+## 8/8 - 0s - 26ms/step - accuracy: 0.8506 - loss: 0.3541 - val_accuracy: 0.9000 - val_loss: 0.2905
 ```
 
 We quickly get to 80% validation accuracy.
@@ -534,7 +534,7 @@ predictions <- inference_model |> predict(input_dict)
 ```
 
 ```
-## 1/1 - 0s - 368ms/step
+## 1/1 - 0s - 298ms/step
 ```
 
 ``` r
@@ -545,6 +545,6 @@ glue::glue(r"---(
 ```
 
 ```
-## This particular patient had a 48.8% probability
+## This particular patient had a 48.6% probability
 ## of having a heart disease, as evaluated by our model.
 ```
diff --git a/vignettes/examples/timeseries/timeseries_anomaly_detection.Rmd b/vignettes/examples/timeseries/timeseries_anomaly_detection.Rmd
index 9b35aed2a..bfd820c0a 100644
--- a/vignettes/examples/timeseries/timeseries_anomaly_detection.Rmd
+++ b/vignettes/examples/timeseries/timeseries_anomaly_detection.Rmd
@@ -55,28 +55,9 @@ get_data <- function(url_suffix) {
 }
 
 df_small_noise   <- get_data("artificialNoAnomaly/art_daily_small_noise.csv")
-```
-
-```
-## Downloading data from https://raw.githubusercontent.com/numenta/NAB/master/data/artificialNoAnomaly/art_daily_small_noise.csv
-##
-     0/140187 ━━━━━━━━━━━━━━━━━━━━ 0s 0s/step
-122880/140187 ━━━━━━━━━━━━━━━━━━━━ 0s 0us/step
-140187/140187 ━━━━━━━━━━━━━━━━━━━━ 0s 0us/step
-```
-
-``` r
 df_daily_jumpsup <- get_data("artificialWithAnomaly/art_daily_jumpsup.csv")
 ```
 
-```
-## Downloading data from https://raw.githubusercontent.com/numenta/NAB/master/data/artificialWithAnomaly/art_daily_jumpsup.csv
-##
-     0/140548 ━━━━━━━━━━━━━━━━━━━━ 0s 0s/step
-114688/140548 ━━━━━━━━━━━━━━━━━━━━ 0s 0us/step
-140548/140548 ━━━━━━━━━━━━━━━━━━━━ 0s 0us/step
-```
-
 
 ## Quick look at the data
 
@@ -282,61 +263,61 @@ history = model |> fit(
 
 ```
 ## Epoch 1/50
-## 106/106 - 5s - 48ms/step - loss: 0.1863 - val_loss: 0.0316
+## 106/106 - 4s - 39ms/step - loss: 0.1863 - val_loss: 0.0316
 ## Epoch 2/50
 ## 106/106 - 0s - 2ms/step - loss: 0.0333 - val_loss: 0.0233
 ## Epoch 3/50
 ## 106/106 - 0s - 2ms/step - loss: 0.0248 - val_loss: 0.0201
 ## Epoch 4/50
-## 106/106 - 0s - 3ms/step - loss: 0.0209 - val_loss: 0.0187
+## 106/106 - 0s - 2ms/step - loss: 0.0209 - val_loss: 0.0186
 ## Epoch 5/50
-## 106/106 - 0s - 3ms/step - loss: 0.0179 - val_loss: 0.0146
+## 106/106 - 0s - 2ms/step - loss: 0.0179 - val_loss: 0.0146
 ## Epoch 6/50
-## 106/106 - 0s - 3ms/step - loss: 0.0150 - val_loss: 0.0113
+## 106/106 - 0s - 2ms/step - loss: 0.0150 - val_loss: 0.0112
 ## Epoch 7/50
-## 106/106 - 0s - 3ms/step - loss: 0.0127 - val_loss: 0.0094
+## 106/106 - 0s - 2ms/step - loss: 0.0127 - val_loss: 0.0094
 ## Epoch 8/50
-## 106/106 - 0s - 3ms/step - loss: 0.0109 - val_loss: 0.0088
+## 106/106 - 0s - 2ms/step - loss: 0.0109 - val_loss: 0.0088
 ## Epoch 9/50
-## 106/106 - 0s - 3ms/step - loss: 0.0096 - val_loss: 0.0084
+## 106/106 - 0s - 2ms/step - loss: 0.0096 - val_loss: 0.0083
 ## Epoch 10/50
-## 106/106 - 0s - 3ms/step - loss: 0.0086 - val_loss: 0.0070
+## 106/106 - 0s - 2ms/step - loss: 0.0086 - val_loss: 0.0069
 ## Epoch 11/50
-## 106/106 - 0s - 3ms/step - loss: 0.0078 - val_loss: 0.0061
+## 106/106 - 0s - 2ms/step - loss: 0.0078 - val_loss: 0.0061
 ## Epoch 12/50
-## 106/106 - 0s - 2ms/step - loss: 0.0073 - val_loss: 0.0058
+## 106/106 - 0s - 2ms/step - loss: 0.0073 - val_loss: 0.0057
 ## Epoch 13/50
-## 106/106 - 0s - 3ms/step - loss: 0.0068 - val_loss: 0.0053
+## 106/106 - 0s - 2ms/step - loss: 0.0068 - val_loss: 0.0053
 ## Epoch 14/50
-## 106/106 - 0s - 3ms/step - loss: 0.0064 - val_loss: 0.0050
+## 106/106 - 0s - 2ms/step - loss: 0.0064 - val_loss: 0.0050
 ## Epoch 15/50
-## 106/106 - 0s - 3ms/step - loss: 0.0061 - val_loss: 0.0046
+## 106/106 - 0s - 2ms/step - loss: 0.0061 - val_loss: 0.0046
 ## Epoch 16/50
-## 106/106 - 0s - 4ms/step - loss: 0.0058 - val_loss: 0.0040
+## 106/106 - 0s - 2ms/step - loss: 0.0058 - val_loss: 0.0040
 ## Epoch 17/50
-## 106/106 - 0s - 3ms/step - loss: 0.0056 - val_loss: 0.0041
+## 106/106 - 0s - 2ms/step - loss: 0.0056 - val_loss: 0.0041
 ## Epoch 18/50
-## 106/106 - 0s - 3ms/step - loss: 0.0053 - val_loss: 0.0036
+## 106/106 - 0s - 2ms/step - loss: 0.0053 - val_loss: 0.0035
 ## Epoch 19/50
-## 106/106 - 0s - 3ms/step - loss: 0.0050 - val_loss: 0.0038
+## 106/106 - 0s - 2ms/step - loss: 0.0050 - val_loss: 0.0037
 ## Epoch 20/50
-## 106/106 - 0s - 3ms/step - loss: 0.0048 - val_loss: 0.0032
+## 106/106 - 0s - 2ms/step - loss: 0.0048 - val_loss: 0.0032
 ## Epoch 21/50
-## 106/106 - 0s - 3ms/step - loss: 0.0046 - val_loss: 0.0032
+## 106/106 - 0s - 2ms/step - loss: 0.0046 - val_loss: 0.0031
 ## Epoch 22/50
-## 106/106 - 0s - 3ms/step - loss: 0.0044 - val_loss: 0.0032
+## 106/106 - 0s - 2ms/step - loss: 0.0044 - val_loss: 0.0032
 ## Epoch 23/50
-## 106/106 - 0s - 3ms/step - loss: 0.0042 - val_loss: 0.0033
+## 106/106 - 0s - 2ms/step - loss: 0.0042 - val_loss: 0.0033
 ## Epoch 24/50
-## 106/106 - 0s - 3ms/step - loss: 0.0041 - val_loss: 0.0031
+## 106/106 - 0s - 2ms/step - loss: 0.0041 - val_loss: 0.0031
 ## Epoch 25/50
-## 106/106 - 0s - 3ms/step - loss: 0.0039 - val_loss: 0.0033
+## 106/106 - 0s - 2ms/step - loss: 0.0039 - val_loss: 0.0033
 ## Epoch 26/50
-## 106/106 - 0s - 3ms/step - loss: 0.0037 - val_loss: 0.0031
+## 106/106 - 0s - 2ms/step - loss: 0.0037 - val_loss: 0.0030
 ## Epoch 27/50
-## 106/106 - 0s - 3ms/step - loss: 0.0035 - val_loss: 0.0032
+## 106/106 - 0s - 2ms/step - loss: 0.0035 - val_loss: 0.0032
 ## Epoch 28/50
-## 106/106 - 0s - 2ms/step - loss: 0.0034 - val_loss: 0.0026
+## 106/106 - 0s - 2ms/step - loss: 0.0034 - val_loss: 0.0025
 ## Epoch 29/50
 ## 106/106 - 0s - 2ms/step - loss: 0.0033 - val_loss: 0.0028
 ## Epoch 30/50
@@ -348,17 +329,17 @@ history = model |> fit(
 ## Epoch 33/50
 ## 106/106 - 0s - 2ms/step - loss: 0.0029 - val_loss: 0.0027
 ## Epoch 34/50
-## 106/106 - 0s - 3ms/step - loss: 0.0028 - val_loss: 0.0023
+## 106/106 - 0s - 2ms/step - loss: 0.0028 - val_loss: 0.0022
 ## Epoch 35/50
 ## 106/106 - 0s - 2ms/step - loss: 0.0027 - val_loss: 0.0024
 ## Epoch 36/50
-## 106/106 - 0s - 2ms/step - loss: 0.0027 - val_loss: 0.0025
+## 106/106 - 0s - 2ms/step - loss: 0.0026 - val_loss: 0.0025
 ## Epoch 37/50
-## 106/106 - 0s - 2ms/step - loss: 0.0026 - val_loss: 0.0026
+## 106/106 - 0s - 2ms/step - loss: 0.0026 - val_loss: 0.0027
 ## Epoch 38/50
-## 106/106 - 0s - 2ms/step - loss: 0.0026 - val_loss: 0.0024
+## 106/106 - 0s - 2ms/step - loss: 0.0025 - val_loss: 0.0024
 ## Epoch 39/50
-## 106/106 - 0s - 3ms/step - loss: 0.0025 - val_loss: 0.0023
+## 106/106 - 0s - 2ms/step - loss: 0.0025 - val_loss: 0.0024
 ```
 
 Let's plot training and validation loss to see how the training went.
@@ -391,7 +372,7 @@ x_train_pred <- model |> predict(x_train)
 ```
 
 ```
-## 118/118 - 1s - 5ms/step
+## 118/118 - 0s - 4ms/step
 ```
 
 ``` r
@@ -409,7 +390,7 @@ cat("Reconstruction error threshold: ", threshold, "\n")
 ```
 
 ```
-## Reconstruction error threshold:  0.03910884
+## Reconstruction error threshold:  0.03994837
 ```
 
 
@@ -478,41 +459,41 @@ cat("Indices of anomaly samples:", which(anomalies), "\n", fill = TRUE)
 ## 4 2014-04-01 00:15:00 -0.746
 ## 5 2014-04-01 00:20:00 -0.792
 ## 6 2014-04-01 00:25:00 -0.802
-## 118/118 - 0s - 1ms/step
+## 118/118 - 0s - 808us/step
 ## Number of anomaly samples: 483
-## Indices of anomaly samples: 216 217 218 219 220 221 396 398 507 793 794 795
-## 797 798 799 974 1658 1659 1944 1945 1946 1947 2042 2046 2050 2052 2054 2056
-## 2058 2060 2061 2062 2064 2065 2066 2068 2069 2070 2072 2073 2074 2076 2077
-## 2078 2080 2081 2082 2084 2085 2086 2088 2089 2090 2092 2093 2094 2096 2097
-## 2098 2100 2101 2102 2104 2108 2110 2112 2114 2116 2118 2120 2121 2122 2124
-## 2126 2141 2521 2522 2523 2525 2546 2698 2700 2702 2703 2704 2705 2706 2707
-## 2708 2709 2710 2711 2712 2713 2714 2715 2716 2717 2718 2719 2720 2721 2722
-## 2723 2724 2725 2726 2727 2728 2729 2730 2731 2732 2733 2734 2735 2736 2737
-## 2738 2739 2740 2741 2742 2743 2744 2745 2746 2747 2748 2749 2750 2751 2752
-## 2753 2754 2755 2756 2757 2758 2759 2760 2761 2762 2763 2764 2765 2766 2767
-## 2768 2769 2770 2771 2772 2773 2774 2775 2776 2777 2778 2779 2780 2781 2782
-## 2783 2784 2785 2786 2787 2788 2789 2790 2791 2792 2793 2794 2795 2796 2797
-## 2798 2799 2800 2801 2802 2803 2804 2805 2806 2807 2808 2809 2810 2811 2812
-## 2813 2814 2815 2816 2817 2818 2819 2820 2821 2822 2823 2824 2825 2826 2827
-## 2828 2829 2830 2831 2832 2833 2834 2835 2836 2837 2838 2839 2840 2841 2842
-## 2843 2844 2845 2846 2847 2848 2849 2850 2851 2852 2853 2854 2855 2856 2857
-## 2858 2859 2860 2861 2862 2863 2864 2865 2866 2867 2868 2869 2870 2871 2872
-## 2873 2874 2875 2876 2877 2878 2879 2880 2881 2882 2883 2884 2885 2886 2887
-## 2888 2889 2890 2891 2892 2893 2894 2895 2896 2897 2898 2899 2900 2901 2902
-## 2903 2904 2905 2906 2907 2908 2909 2910 2911 2912 2913 2914 2915 2916 2917
-## 2918 2919 2920 2921 2922 2923 2924 2925 2926 2927 2928 2929 2930 2931 2932
-## 2933 2934 2935 2936 2937 2938 2939 2940 2941 2942 2943 2944 2945 2946 2947
-## 2948 2949 2950 2951 2952 2953 2954 2955 2956 2957 2958 2959 2960 2961 2962
-## 2963 2964 2965 2966 2967 2968 2969 2970 2971 2972 2973 2974 2975 2976 2977
-## 2978 2979 2980 2981 2982 2983 2984 2985 2986 2987 2988 2989 2990 2991 2992
-## 2993 2994 2995 2996 2997 2998 2999 3000 3001 3002 3003 3004 3005 3006 3007
-## 3008 3009 3010 3011 3012 3013 3014 3015 3016 3017 3018 3019 3020 3021 3022
-## 3023 3024 3025 3026 3027 3028 3029 3030 3031 3032 3033 3034 3035 3036 3037
-## 3038 3039 3040 3041 3042 3043 3044 3045 3046 3047 3048 3049 3050 3051 3052
-## 3053 3054 3055 3056 3057 3058 3059 3060 3061 3062 3063 3064 3065 3066 3067
-## 3068 3069 3070 3071 3072 3073 3074 3075 3076 3077 3078 3079 3080 3081 3082
-## 3083 3084 3085 3086 3087 3088 3089 3090 3091 3092 3093 3094 3095 3096 3097
-## 3098 3099 3100 3101 3102
+## Indices of anomaly samples: 216 217 218 219 220 221 396 507 793 794 795 797
+## 799 861 974 1659 1944 1945 1946 1947 2016 2020 2036 2044 2046 2052 2053
+## 2054 2056 2057 2058 2060 2061 2062 2064 2065 2066 2068 2069 2070 2072 2073
+## 2074 2076 2077 2078 2080 2081 2082 2084 2085 2086 2088 2089 2090 2092 2093
+## 2094 2096 2097 2098 2100 2101 2102 2104 2105 2108 2112 2113 2116 2117 2120
+## 2121 2124 2126 2141 2521 2522 2523 2525 2546 2698 2700 2702 2703 2704 2705
+## 2706 2707 2708 2709 2710 2711 2712 2713 2714 2715 2716 2717 2718 2719 2720
+## 2721 2722 2723 2724 2725 2726 2727 2728 2729 2730 2731 2732 2733 2734 2735
+## 2736 2737 2738 2739 2740 2741 2742 2743 2744 2745 2746 2747 2748 2749 2750
+## 2751 2752 2753 2754 2755 2756 2757 2758 2759 2760 2761 2762 2763 2764 2765
+## 2766 2767 2768 2769 2770 2771 2772 2773 2774 2775 2776 2777 2778 2779 2780
+## 2781 2782 2783 2784 2785 2786 2787 2788 2789 2790 2791 2792 2793 2794 2795
+## 2796 2797 2798 2799 2800 2801 2802 2803 2804 2805 2806 2807 2808 2809 2810
+## 2811 2812 2813 2814 2815 2816 2817 2818 2819 2820 2821 2822 2823 2824 2825
+## 2826 2827 2828 2829 2830 2831 2832 2833 2834 2835 2836 2837 2838 2839 2840
+## 2841 2842 2843 2844 2845 2846 2847 2848 2849 2850 2851 2852 2853 2854 2855
+## 2856 2857 2858 2859 2860 2861 2862 2863 2864 2865 2866 2867 2868 2869 2870
+## 2871 2872 2873 2874 2875 2876 2877 2878 2879 2880 2881 2882 2883 2884 2885
+## 2886 2887 2888 2889 2890 2891 2892 2893 2894 2895 2896 2897 2898 2899 2900
+## 2901 2902 2903 2904 2905 2906 2907 2908 2909 2910 2911 2912 2913 2914 2915
+## 2916 2917 2918 2919 2920 2921 2922 2923 2924 2925 2926 2927 2928 2929 2930
+## 2931 2932 2933 2934 2935 2936 2937 2938 2939 2940 2941 2942 2943 2944 2945
+## 2946 2947 2948 2949 2950 2951 2952 2953 2954 2955 2956 2957 2958 2959 2960
+## 2961 2962 2963 2964 2965 2966 2967 2968 2969 2970 2971 2972 2973 2974 2975
+## 2976 2977 2978 2979 2980 2981 2982 2983 2984 2985 2986 2987 2988 2989 2990
+## 2991 2992 2993 2994 2995 2996 2997 2998 2999 3000 3001 3002 3003 3004 3005
+## 3006 3007 3008 3009 3010 3011 3012 3013 3014 3015 3016 3017 3018 3019 3020
+## 3021 3022 3023 3024 3025 3026 3027 3028 3029 3030 3031 3032 3033 3034 3035
+## 3036 3037 3038 3039 3040 3041 3042 3043 3044 3045 3046 3047 3048 3049 3050
+## 3051 3052 3053 3054 3055 3056 3057 3058 3059 3060 3061 3062 3063 3064 3065
+## 3066 3067 3068 3069 3070 3071 3072 3073 3074 3075 3076 3077 3078 3079 3080
+## 3081 3082 3083 3084 3085 3086 3087 3088 3089 3090 3091 3092 3093 3094 3095
+## 3096 3097 3098 3099 3100 3101
 ```
 
 
diff --git a/vignettes/examples/timeseries/timeseries_anomaly_detection/unnamed-chunk-10-1.png b/vignettes/examples/timeseries/timeseries_anomaly_detection/unnamed-chunk-10-1.png
index f79260b1cec5f7cc9b463d87aca320ee654e1a65..383ce9364e0550ccdea02ab0d955640182a06423 100644
GIT binary patch
literal 19143
zcmd?RWl$YmyDdr*2n2#_Ah-v22o{0|cXwF0yC>+vgF6HWZUKU`AOv@pg}b{ur{Vqf
zclNG*>(;&J$N6#2s!E`m*|U4jeq@aCJiS8XWyO&Z@eyHQV2~vxL=<6QUV>p@Ui?CM
z2^`7YdG!nU@y1R<!x07s3FGPCi_v+XI~W)e7)cQ!Ww*4wc~>{Ai5vJMS5Qr!nuNsX
zH(3EFS-7HT^abjlu&_Haby{6IEa61ql$FnB%E$^N6krv`=|n|fe+J-_WCf#1e6A!l
z*^N8BHF_0*w(_zuId<*V{mi|)f3H6^HvRe_nV*fExf21niz|{H9pDE{+Bf~zFfdC9
ze(zvlzLFz=fJ2ev`}!^)(^K-cbwuH0#*|ei8bcAoSsLfdT(M1QLjMji3#+K;FeWaj
zq%9x2oiQ%I)kvr>3NAL(W=zbc@ldl5qVMxav<qrmUfML%o|*t=tJl73!e*CQHO)5(
z=Si1^6m!37)qYimQeH`asv5OG`7}}8&C!6$YMnNt&&6>=nGHM}49q8f|Nnk>BuOt|
zV7UGM7ylq}|8~QqF9DhGqm!9K6|GX<mJSjO%pbUd;MTokw>-Wh-$hr8^Rez_Y9IAF
zk8=wM9q{OmRQTU++}+)&Mx#?nl4Uo)Io8n7@KA#VHm=wMuRkOtB-n6;lW=H=kv6^B
zXzGWM|5+KZ$$!+&nUmmzu6ByceL`&?YCs;>sn@i$wEvt*=f404E8^vK_uG8ztPHry
zlHWud(bKBu1OFEn_g|jk|EGNd|G^t;3@XAiLJ4k5q@1^@9WlcaWfaggOX<XmrI0-Q
z3<l#FF(&p|RPX3Wz^SN+Dqz71Jsnk6QNil4;@ky}Y9vI6<t(Tu%VPzybjtB-b3}Hn
zMpJ}jdTT3S`QUf0J0uTmNX>PLDy(f5<dOrcRu)DJSX$E8OE`1mQ!Sp*Us|b^#}ZXh
zkwv*o$*b39Bu=fs0qh{Qr@cp<CaP=Ij{C~|edGL~91BXY|GzHb3cqPgYzi3D;Ex|9
z5TwE-m&Q_6I_gZ#osWztaex&s{qNhv|Nq<mzwwNqKP3tAdwPv7QFK{dCh4ov6HF^J
zEjwK@?jNG@7JRli(#Zn<C<Epvp9uT2t7M%qnT49SGNkEbUhaX>D1M{P#p`s!10Frv
z!|ODYhqQDCFxLjra2_-jV3WuBziqKtc5zuDwcIRvk4np2JXRY=%{AtSfYqTFfyCbR
zPEzej1bERuu(981ottN1;b0J&x4v-(&Lu;XW@ghu58Jk)z{oKY{t`+UF8wkZZN_~q
zAhAvNak^33Txg+rVv(UYUl_j4*8QOQz`#`2gspSWrIAT%lzm{?4orsIT4s)KB%NRa
zK_VqH3`bXRcduUvRF+p!NiY#wT0ufm0Y2;b_E@6JMR+UQ>1|r8M^#an(b+WrgtyM5
zw&l8ZZ*8m&z>Iw3m0jU>csJM{iypkm15n-W=7mPkaX#Fw$3Nbw#JN~Yuean=NH$xE
zp}|VVaK8Zwp|Q)HR-Ak~ea%MQfrf4*9YZ1Omkjs@aK*Y0owwI9nI3f(?}gAr6xMzc
zb>azjxP@f9g@_yZprGF%iZCN7j2J3Je`bKwiYBjgmvCk%P46FY=8T9WA*J98M~tu1
zWkLWPB#{AdiYllTGwjxCN1(!!t5hY2G4(?>F}$ogHJ2RfrKvEfiL>~KqOxi3*dz1I
z@37kNjKYtVP}L&H`zm{kSN`!8S->5)oiyoj$^>Y}b-r=ShaN*3;!Cv~yoWNS*f4IL
znJZTMIc84>54e$fzfltVKtQe=H@0frqYYetyC{H%hB#bc#j<?UCB`{8Uw-CkY{f+v
zy+Tvn7Y>G%>Ro>oGXOk6*Zr}mpY^E5@imQ?+q)QbTM|IP-eUYj{Vnh)?qMn~?S`~w
z_LushAOi#ZnIH#8o%toHOd~w#7MR}|v!Mw1@m`4e6%33WZ0lPfilV~;K@mnAt^fbN
zKYU}!b46BR>3ok8%p}z?Z<8}Z6RPdiG|V2sLM0iq5Qv7((g|eU35$uw?}$Z3q1iMD
zeosdWi;8GMp;nW?(Mo!P2r)=;!4Dclqwr^QjOrqn#1Oo#d$yInb=K5@#JZ@k?Cv22
zDe>13q=Fz4gx1#9X8P@8w-VLj);}G#Su~+VMQQI3IAk2_wFNkh80_k`sbWK7qvVOA
zx|Z#Z#<*P?(=!d<qrm^`5{D>*JpzMxjvv#7goHeB6<haRv`STF1**rLwcpbK=I{sU
z|GY;#4!=_t)g`OB&FA`>Za}+w;}ZuP_fr=7dh>~n=NC)Pr&3U7^x?z(PI;i_exj58
zTMt~or0qu3dnTFkn=V&%EqI$U=XpD7%qO)n?<(0cQ_k>D_o}95Kztm$Y&s3QxtxM6
z2{nx;B?Ou&PG!xSTP&|&KIwFvG}m8lM9sHl_?|D<SPI^K)FMAA$#C1uKas&SZ8|mH
z>Fjzuw>H1bU#QQwp1CTgMrBb=HZk&;(qlLSz|@aOL+XS<8Dc*3!P5ECIL`BA(J&nJ
zcb+XZHpzcNt1e9%lQp%KU8uZ^ld&nc)F0fRkOgqk1*^DQjygGdG}`D-$vO$+#fKgp
zR0huTrY27db%}0`{6Nd34on*)z>^D-r+-SpZ)Vb6_K2)4RBaq$;=6PDTo-cNY_G`W
zL)?51&2mqAcJM*=aQpxSEhz3!6lAMkuWq}!pXbfms>6skCN&BseT52f+q~a?L+!Y~
zz5$sP)Z_QCQXkH+)lGp~p72y1c%6ru9ms;npYCrEBQ@g2p)dZHElN>w7FK-2z|k&e
zI+?%|w8w)!SE=J5(UV~f7ror`3g=<LRIpmwOvhQ3#<fsd<fYNKt%0hTd&{W<d&;xA
z634k~v7Fn=enA>WMigXZ4!pW6w`Y$4RcSmyA$A!cHnXs>JlQz_cBH8k)>eu!pK%Kr
z8QB@*PrrE^R1}m+07d^BzsQ=K7ecKz{tJZBuY&$<YonALZTx;iLIVGV!L=ux2O2H2
z7Juk-;;ON}<hp;gQG6bmk_b4ES=~$Y{+VOs*}vt@xpFfqwN8DDb>bJYCML5jsTX%=
z#HhT=F3l!@ec{)AN`nQ}kl~C$B`CzYg5t1v@seT5NPnMpzhMf7JlG@mDK}-XB=)V3
zCEz(W&R6z}mRaU+=^&s-UlJhZ$0nw>eF)7qPa&~8c8)A6O3AaI;MT@o<rgn<CB+7^
zn~DlQ5vt%Vw_3z1zm49?l!_mTIz7&FF5Q}ivgPMN-51+A7;=gx<_><Rs4(XhS?D)Y
zwdkkhx$llNqGc<!_DiyetnnutWA@+O<!+77svmy%;P$9^H{PbQ%G?RihJ2#3&@82{
zKT;a6sN3Y7aKZXG+1VZ0WPzH9HDuz(r{8x_Iyo4~5k5;ufSI&nOn7K$5d2#PbIns`
z_b)2mS-uGNZvh}=Stpmlks31dRq<=0DC*}icYilu<r*E1M&TZ+u6a8W6S%A&<!^w}
zAGNFS7K^o@y}7fOclGS#6?*WPbc5Pe{Bp(i$zhMkz`pLF6Z>mG*{>&YCmSO!9w98|
zdXLmLw8;Vnx(n`d8sIlmf}EB12R^xr<VKAeAz0#tIv^1Wbwx(v?^ZZP8UX8p@#2pk
z_`>QEyzsEmz3~pS#)<HD!I|s6`pRDr<pY)}-w;y3f6tiZE&x@7A<^^f|EyjdcY^KM
za=!)9iwV-EGI_M!-uNEi=9LHxSTYsJ=;gfdJh<OI!7VjEE9W`@Yo4}xQo6R-%Q-T>
zJ^{vzQ#S_{qp>Z8$Te$SEk8O}YL>ZsPR!-sh}hY$KhYve!x?vYnF>rP-J}}T3xn)M
zSVLt_fGFBT67hdD>5Y5G28xS|yQ*w5y{=1LSn$k{`Wpy-xQse5I|2q@)#;jyt!e3X
z-(UV6dGs_IZGJeMtMFY3aQFAYBM8<7()yP4Y<-06{Or$?lD(E8*}6h!sl=i|_1n$U
zAwk~_6hxe`>5u1giA*;h=NI0+)IoKlN9}k0->G%?j;7*jedl(;TlYt6`S78>-ke;Z
z72#alW=Nm>FCGZ{a$bn-9FfjI`-t`SpX@2!bYB`yh7L5BQ%h9WhVC^K0LTCf(IoR_
z+jU~r0_$!~jolvFi0m{m$+-pwb{^;D7rog?J#6q=sIljIEy#0pS<^Dqz~HqWMldUP
zvvEFrc!<x-<oO3Y7cM=z_V?VecA@GbwaZ1&eP4XnCPRL{o!V89jq7;dmMYC{J+v_r
zHN$J9CDOK)KA`#w?IY-*c5=RjLg)OqtFlWfF3Sf37??9^3W{`;jgym;SaC;;&%)p>
zeS68O8f&>lLPc=dv1Fa}j>80pq?qBo7;4HjbPSbvw()-L)8#f(ri}TNE;cnBR7<4S
zrpPASbNBVopQ+ST;n*^QxZg(@K7_=)Go3y|5a0Q*L$&Z;pEo&`FJQjn=Gi}fyg#JL
z=6bw8v|b!YRN0^0)O9;*h4<Er8ZoAbnUeEyvfmsMe4luBiAi3goS`+ppGhYZ_gQ#e
zlrtj6SW96$caD8-if!q0|JSZXa~`wP3;(`{a8I}6DF*UkJ;2COWYRyXWbkQVi4q^q
zGbBa8S>lxL<|)9UhbSBii9cwwpAJ`?r#kTMqJ$qgSLMC9M7de2w=ch7wwh(EZV0gT
zfFmY;y#l>I1{=hp=AP}orz4)JQ-Q3X<|+lZZ}DFg*VS>evtyxFa3|)ObEYy-<#nEN
zR)gDu)kQ=^wg7rsmNE7<SbZD@wW8qL00;>cwY(U$V)zS+4gZfI{E6-hYa>yz&ExQS
z>c3jGNCS%*#83ftrrEi&+$PVKxEW4Wr}cy0Hx^U<^|$g9A%!EiIOQhRDWY)u9;?vU
zaz;1qZL%mg0kM?T+FB5vF7&szw4a1X<N%ZQLp$#6YZ)^guE!@=&ktGdF`xf%ZxwO)
z9Y;<CD+NbLWW*W_o~flIBa`}kt15y8eV7aRLVwR;K4YlB?0P@-yFKhM_=@|(tbI1k
zPvQJd?AbS8_9thWo8`%$5ivUYJIy)P{XQ*5Z+?LVI$noAKZffushAdiA0D-at7Jy`
zdP|j)655wy0fwG=?e;;-g2SeDDq(8TW8*@SGOy0%6}8vnUzMzHr2x$S@#3aSzs8jQ
z+T&c|@ElqSA7#a#1O|sbs9%|0=_FFxu_V9)qZY36^+73Z37Luc%IS<jT&LQ-CYr$d
z<Rg3$L8QMor$hKHbAb!O{HZqs{*U}`D#Q+Ur@DuS2MDy0TTXj8wrtewns|6+e%(4I
z7xi<6NqnK=kAjJU$ynoKMyis>MGS6mSfexmA+7@H`)U_Fc2Otw9k_|4zFaN&qdpu{
z7MtD=UzHKe;>VRx>JcBhOWoBR`t;C5?eiKcw@`_Qv?t9TFk@C!&I=CdWMIxhOBMLg
z>4t6={X!{k&FD7huSkmvX7Bg=3BldoS5eo3Hjb32_1@~Y%$qeGbYIA=Q@LifbRDOV
z_uC$}SuoFQyGqP8H*8!g)ek=cLHO5C!AZpY2e#7rtG3n0Y<QwP>WqgWhT1prAm{1d
z)ia5noI^?JW#5~&!%%AyA6O2WiTlm(wGNWmANp1oiv2Wt^4tZDJt_rTb{{A^T^NvB
z9=xi93(kVYhv<_R3&bDC^XAKKQmdANrJ)zAIYt7twcA)V4s(RW2Ws!NK>Q!=-Q6<T
zH#bpb=e-8kyEY4I`9asa$ukG(qf5oe!sQc<I6x*)dn8a?yy0l0d`KH{ZeLT-X{^?$
z^_iVaTlqVTvETo><myrn%AXldI4WW3bJ>&_@;Atry&SqtiEM!0I(d|2hl%d~pqt`2
zaGI$*A-fZM*+xirBL1%9z$l}kLT;^<E_;D-bWp2{_M9`Rqh`N)p10aM)x4X7V(Z@P
z3k(cGk~d^j_HlOKMeuA`*q75<)}<>#I$6N`W?Ligl+S(g<I+t6G5x(Rbl#*r41STA
z=ix|3F1*1){qh`Y%0P**@F>Zg9R9K}`7lagC?-FF3PU8?9G{$&ru#!$#*N&LHM3ND
z1C1b~v1s5-bemxnjz=P3FyCUVs?)ApEi{hL)mk1OJmux(A0O`F^eUORDtA{iJ)Q24
z4yh$O785&N_P#I894#zD(cRZg5A0|~wCy9@rrK>a=Zm~AWrA*)O0TA&Qi#ePPQE@$
zWUS{6AANRyGkMJ8UiWS03JZkdPH9h<lgp&AZ~xV8#Z2ro;Y{tY5QKdfv8Rqg*NyvD
ztit+MlSXWey)y}5Q>KIqMA>CDp%jt=`=sCUi;C#`aKhFzaZ9l)=~Jbt0IEjP$BJ&R
zf-2ZiH*e$L`lm-gRwOhVFQSQWedvk3C+tLZR?Eix1TrsZAm!StpM{b$jM=&F;D&QV
zlHU`o<+E6LQFr(tdo9i|Kf{G+F<PNhVl2BgZrS}ym)?8-0lfU^Y83X<#Xso0==ab1
z+jX5blsy``gAvARrKuNFC~s#W9#R^K;?xn6$QZD#Sc@#x(EzdJs1V*Q@HM0SR`JoR
zZ+B&7Wq5e_bf0==(E22eiH4^8Ze^DO?rvlxG`pb4f;06MgjCNg1ylt!G&Fny!2r`t
zmQLII{Y)AE6`LF9iQd02amB{|H+4)K9aXG?_5<v3HrO_G0EdW(5$-`^7R}E7$6GX(
z>Q@@2Yj$Rym(Qy0SGgTW=Gaye=e^G&JCg$*iBBSGtOsGgh~d<M$j|*5NyhwzY9I>T
zRoxKvGbjdh(<mA{Psu_o0RVkvdd`<5iPk<i50+?y0$88`X~&ZpQ*zz|Bf3T~EI#fZ
zVOX%>v6Y@Z5_1I;%wpYz^Uta9=YmNg5<4^L;Bf|mbH*nfn3x>^rHwAs3dUUy&|~fA
zaRrk82q02=ZE15E5sQx>2Y{eh{TjVrgd+Ex;&9pu(A5F&{Ca9Z(@I}+D6L9I$24|f
zvh%D*yK}_pQ{^b^3<uhA0?6QMQr|CG;w)oeq1vBk&tvLIu7d<&^$)}bZs94aSa+%Z
zb3q~|xx2D(mJNq`FYVEN7f<5-1Eee?+~p|j-n<t6^fAV)KRf36SKb4-t+z?m0Jax*
zrFp{kL#Pa>^eQSGGOI<;!qpEwKvMmqvRY8IbL@5n?K08W$hU9m+L!X$Xdpqr3G2|K
zoY3KWT}ARNPf-HJ0H_rwp4UBt`r$*y1l{%)x|Dk%+UymEMM+RLO;M4<37WIa7@%TA
z4X7glf<=#ULwow9-v6s$b?I{jTr^EKzr!&4Nr~Qh1qi#<V&HI$RBFuV29wz5N2{D;
z0LO+NVh#aING(cur9}Z-Q&7?hS2bQp_fDo{?Wc===05``mi&_YL4`)9WQtsmrLMGq
zU8rz0=~H1eUQ&@U7{T)pKJTkt!s~baJ+z;l=xHk(7<3MwlH1@Yntw8^(9S{_cwB}%
z6+r)9=HFZQ2vZqC>A8Xld5c^bfKjIG!}>UX+d5RO-=qU1c1ijbGKeh3<BW>Td6(+s
zIx+(o^KH`3U&Crj0pSg0zS3;9@bmN^*yr@ABryQ%(H;geF`xOLR%wd!9VwBS2S`bQ
z{PP$vz8f=AMMOv50GNeu3|J)h$PfQya66GVSYq~$E5jTxX+E>9R(fYy@~Y%?`pHU{
z-6u=><c|tsP0eLA7x`(Up8vdR1Q1!_5DeL9rAw7z>eJqQ5)Rg<A{7vlu=<jcXS49*
zX*I*cq)8#*pSt(J^uKU~{{W-(fD_3IrSLWAOG=dc=5n8h{dgMIZ5-3!L`~-0rU{V(
zVC8G%R&Al3`AqQd5`Mgw0Q&j(nOO~OB~^J2&T7o#$x^=tA+>r1Z4th#vF9%=dcKEO
zf%*c%J97~ws;a6*`wNB7vwtE@qGL(<X~Pee&}k7n1nOx&zI|rS6y5RUdE@7zPu$0w
z_dE`bG--&ey*-rH8W{MP_|KvO^vk~N@O}k+G_`wu{K?nM?9lpELJA233d~OG?027>
z-Wy@V5y9_r61vmC(Vnz00j#slProc^yqF0vpRIrzCZ(tS|6w4+tk#83;d~&rAp2CK
zVE~pSx8wQ_-8{K5sZ`>5%nM+QoLLMFr=?JMq3>z~U}6RQ2pgwGbv71b9x?pS_Js`R
zciq$;sS`NN?7%&K7V+<ccNCI4p6}2LEmfDF&m=_HNRfct0#y)1IkV)(vobh<%frGt
zM)>&w8`3jt4s0v}AMB~%_t$Sr+!n?wa-OF;C-%E0Mqr6(=iCSO;xzbqFYa**^g(6W
z<iSyiw40d)IHrfa-N0wMy5&#D`S1txKiGhI$eB`7pFdq?hew!1wf76B&)tS?r)q0(
zRqEpH-2g+xgtS%k<86O=Bt>l$PT*0bJp+NpO9ikmn6@;K>dl^d+0CV1P;uE<<9lG?
zLoY{+KsC;Ny{{u$4l|BjDXg_C#0y=rRh7-Q9cPnRj5_Bynk;{#ePz9q-0_7&FWh}3
zL$%=qEc*o(n5atidw%;R61s%6G-5S2_q#wu73z<BalTjOP}>6sI)ztV3xdp9ilpHz
zQMT8cI*KY2Fn2hnQv07svy^o=7V08B&I(=7pCTHP8)dUWIC}K#{*xLSnvE6>2pMmr
z6|iK+07*barT#+@OQ#C|9afHc%4K^0tIninUG&n7J~VM0iV53l%z9n?vw;q+wSUfL
z!i<EhRiC|X-zIz5k@`!anjmwtlBUw}pR~jljns?&^{n_PC{QlTXJFYlcC*;Dvz6F2
zQHS{5p9eR3n-(1mbOEybL}xxMv70}%3seHPK#ot{S^|_X!ZSOML0Lc*?Z~H2WK~&e
zj6JHWO^(xXf?H$_sCU>CCsMp=mHzhTDWA|Y7>SfLjiX~=lPPnJjMG#Le-|aONWf4_
z(dyljW=8TuBO*HJO9%}M6aM`9ZO~UQCEbZ1qzA~BfreUJ(ZYtCHk*TL&s7AdHbKW7
zrp%ZXXX<~7%yMH7E5~j?N*0VTs-5-s_dhSo_G??#|EHj!q^SvHcOGchJ0sbAHX0g<
z!_Ob4t!jpc<t1Z)JTHLI{!Q`A&N#j6s!ec8i+OOh5-*$kmy@vd_9`m;DqP6JLeWbV
z-g*W|foQ5aBg!;Km3pqOcZh^W^?TPXSeUOcZzM?YXCoTe?Av@z>FEhQ4wq<z_%EiK
z7B50fW!QhtIp|%cT^~wkLg>yHW@o1&9C1g#(sk<_cR5x=lK=66*gq3$Shz*c)bMwL
ztO+K1^vS)K^Xf+yo{p=ftD>qa-grVx%p5?$;^vDdL`Tnh(y~5kg(*Qf{uJ)vyS4DP
z6<W0{t$fl5mQZ_3sxXdnFja}V{S;Bhg%LK8$VEiM2tQ&*-*h29$1@LvR<G8#{y{M$
zW}d9-3--?kZ(aij1Fm2zRJ07Gbs&n*L84iRS?}$3C|%F3I+9lOvv)%L^qcVv9q3Lk
z>j?j&7ob7|Nm1qe(&V4@Ao^XCNvi1zi|iysY_Y$B`ASTh#1u2#P&%DWSy<G7_>^ys
z2w+b;Vqfcs{tIg?aUpDk1wRF=D?-2Pm>*+6YIz9MZ>D&jz!#WcVq2CCd-gf7>;Qoz
z7elKSJC!=N$f7O}keC?X`9V>Cl1^cu&(<U(p8PY66Sj&NaCacbNsgIyJpTKjFJ2Hb
zwa|}#otake=UaBh^MeWkYX{H=nWFD~G*D)KiVDSO{XXO<gkmv23410ljGr;qxC#R3
zF(f?ux30*~yAyKrl;<d%9-x0wEF|1x1)@HgctTlEfr$dPN)HZB>QxyDne%n|&u$Mj
z0O&iT`ay@@CL-hb2~YPJo)4+~a|6K9#eR_aRy8ndb2yL@8}We~Fhc8ppg!LLl$Pi{
z1z)Y_%q`!=UZ)QVzrSpGB;;{e`zaaQ+t>H15?P3O&JO8~UxRa%%a#5oP9VdcEBhmD
zK&wnQy*N&*@+#8VYWVKY#$L>XPV;H(O$XvB`P>m*5&#1iIqLLvt1TwT&C^R?w^cKw
z5^JFU*%W#gG%q2B**Bn2YlS$c3Z#{c!rIm*=iNz_V$~jp3Co0;THBTrI}~(68pb!f
z3v709fC@Wm;t!U%H+fU9{cRTQNwhm>OX`XuUbOE0Ep*yKz-g5<Z&S{<8$z9-o|)t|
zb}0za-0`{HuYIK%WUv(f*$tMCNlA5C0f3)Nm_YUYN>4#y1itZNSSl85cKzt1QEA+>
zHIgGY#F08c>~m$n5O2u(t_&iSaT(tLiXSfx9_DUNr~wS$S(Sf}1<B72)P!y1N3x|J
zKFuQoKwE&oq=hNy0|pqiW56sfD#|}m%?5d79JoFy+Sdv~B`!1N@YAm|$9fk&6K&-@
z=D|+~0Dy)cL>awQ+1q3JQ3KuF?kC2F3lQYj>AvEmRI2!HxeY67>zkB@vlpM$>zdC2
z;54*j5Ctt34o11Zy`18YhXB5Y7=a(#S9sSX^5j=vEdlZgI7v{dcr=OGET`pD#so6t
zRiJ(urC}HrZ7D7~t)P$GrnU|M-+<<TGXBwD%b)EK=sqYX+?PC|h(ENjj{E3-e#l&N
z0G~)vajOGNPPteB{*>#Q`hicQ2xxynZ~PgO-hg4$>sQ{hI{9SMOR^B;jrX;;%Mke2
zHnTjSp=L}z?|jR;V^va$IA~~`p6}>02?vIOsz`Kk2|K{egG^?%_PcvcFNtj%RWvm<
z)z$Ip{st`Rw=vYBM>qIH)Nrxc5xfL8sbJx?BYCX7P2*(W$6C1xO+*uStBvMwvn~g}
zc`bY<Qf-^v#WfSV7<XM4y)P@i&I>64*m7?)9Zv~opvylWFfdk9yJN=@poe2|rFYvY
zWdNj5vb`E8HMYQ)@HlwP7uEUJ@5pX-8W&@w?d@p26b=rK0E_YP-0~FK$oP5cC{+`)
z+mZfa-z2-t7(218IT?5Q-{7&rrf}!oRvWzKTUD=iJV);LK)ujM@3aFbO~+7pA8Pre
z6cqt9@1LfTs4gSc<TD_H(D#M2`0Okf*#iWiRex!%Ml_s7BjMWoeQ=vWg}Q|R{v){%
z8a+-}%a@|YI=5hz<=e&1r+P|W$|WY!2mvj_fei+i&tukl2en`~-Br!7OZn{ar!Njw
zc{0T%3<t_FlmS80`xi(aVF0HoDuP;Fn45=*64*Dg-nRTrkm(#pG3;V&w(Z6RDi4>K
zk7XCx?vILzbz?VSU8@RMftj5n>=DI=U~sX=6|Go;NzTaDt^kpKI15m>k67KOl2qjI
zCaRO?d(GyaiMD<k(@a0d$!(^wd9j#I^S;*9wmDlmT3YEr{Z{N~V^eI-!6O+U=xG6Y
zRW5soY^g1S(^YFl#f$~P)M+cLCyEyElcryRM_p;wOk0?Y_HO;Q#)~$H@#ni&`=1G@
zf4Et`Un#^BLQ7g0GA{P3HM}YW`m4Y#9y^ST9?aDsW2^piq)jFFcjyXLASMwmAG5=6
zf0c}C;Dprc><p%L9@TvIBf@GAz+f<f6MxXI4MqyZ_c2;uwlHOm;Zz`IK5&%iQ<3Fl
z%NJg*fI4B9$Bf)dxvbnDx9pQy$N-+6BOc|m^bI~c`wr^M&*b0g)&()<tV**o=_s$R
z5CyDW3kZZ=M4sUb8DdB+xgkt4G>le+#tf=)U6;grb>_QKfi*Mb;@}VtZC^I@!r}zW
zTf1{CfN4xE`PJ(*y41wesm?hGUrg!AHkw}z9_TDuK>7aacRO$go9?}9;y3cFzxOKF
zY4*AfV`{#dcU`=8rD+_`iJGR20BpP8U}0<z;3)Dyl%>iBfNZNjJeT!M99TO0sQhPu
zH>80-2F(^NU@=Up!Fqv{(@`(7ap#`3<^Ez8ZivrW_~0A=qr2<LTF(PRkJ}gx>Bpe>
z_z#Yzv3b|7>j1M3W^dOQPzxWqGALdn<Zp1X=jF(Ty2JHYatU}I*;Qnx>4KL32HGOy
z57#T~RY4kUr+>8$VXM=3@^;3w77$M7-&M8`(ZK`$`>NGT3=>*qz0&uSg-lW8c!0c~
zmSA*ou*cZ~ADzQ|)L1S$0xZj&_N~rEUP6L3ye<Mf&QT1o0v|m=Akx1<*kA*y0R`r4
z7;t3S;pq{|9aMBf@43?|N$*#U+%YqJA0o0(iZ_{vQr)krr*x4-UMNlxDQ3P9`AM()
z$HDPs#T(#Jz4yBg2TYMJ49R>KXQnn0G&C1Cb91lS6Cu-SWHUtWiF&r+d9+Ct;(NAJ
zn|)=xl;L1d_W-I8m5MO|#KeA|1p`3JmDd9})^`f?gSq<hgiNoCPpTVSjwzfW#EEFW
zKh$OjdIrdlgWikgj<}AHubTC0jV*W-_gapPjfqDQ@mq`+02xll5Iu;u6EVx6WIB)j
z3h`AdKxpuZhz*w;{}Gck>+<g;MYkVYgBu%uMw6f^fJ?W=e7watLZe~7fqQ#<gIuj1
z2awfLDYg%CTfBdtMCF+-2Ht))hX$|7EmEt*-{^wyd|0CDhS>)|RUPf^+F#!Jw9i5+
z2~oru=AZ-OPhARK5!~-l0GB?uo8ZpP%>`8@XL_3O{zn_xKQ&nU@@MSV(}Li?d_PFI
z9nljPD@;_+_u(dz-*eV><|A+UClEayU9p~_SKI`KA**iUC>cNhJq8ywr=>yNH7A+k
z)P3Q@c`&gLfCUQ3(fXV5+0Eo9gNePexYfjh9~#XK!4xJuX+~_9)id88#6`w$KBwv=
z@E0}Q4Mz{A^XlDf{5*jCY*j;#u5_xXvFyq%2QvQ-Y+7%#glHw}>Q=?$8quZh?t9Vk
zt@4RC+o?p2w+95aO(TM{`qCQA;`0q5;bN<=Rsp%d%a>NxWC|v~((>!S!0nQ94Y9rb
zkCL$Lh#)mBcK*4>+wnC}gp3%00_d@ii|>4onD8U9iweXN4?R9HQSXQowv~B>v%;OV
zr}AXNbkDegcn8R=%>aUwOcL^_F<;4Am>0B)PooG~FZu}Fn<~Mm?P7X$*6)?FDJT>w
z9Q5|OCBs}z*L^*M-;>em)$1IC{(C@Rp`(izd;~+8_~wCny{r)P?Kc%HS7{&0O^B-*
z)K=lz9nt6%3DMIQI^x+E=?+ItC7Z#QBN|&?*0K{;Ge9ky=9Wc=>*pVFDL@erP>Fec
z(jV2cj@Vufz=gC=%(t!0-nRk+bU*CnhKGx-=n@7<))S`9B+)qI5+W8LuB|fUn=Xfr
zb5_<^@7L_O84d40OP3r!i1*;PJ2<W}2~b6K9ku{;)b3^L9(*Nz?1=iY|H$~{s_u02
z{x@CoAqq+n{I_LG->0-(tvHSvWZ#s=+P-tQH`))Vqw4e|CX1Eq6QCGMmnqt5IRO9<
z&|51gAtM7BAQ@GwGHY$JtU^V3&;1>oqS0?#W0f(Me@k7eRYQ>nUTK!q5j2aAX|V4*
zOPj;2-1FW)UhSfm%ed>MD_`5#m~6tXFn3%{zl@<U_hel=Y|d)nOCthAD)>aHYj`88
zi6RxoM9Zz=tcl60vY&gaMqMqu3Tu<>WUL{7kw+9<#YCHCYp24z8>IY>di4l=dZ!8T
z@ik9_e(!6ox$9vVDb{8LMdxhu;gT#n6yh)amw&W^F;p@sKV6Y=ho0IxK}0B*ek2r>
zlpMATbH=PHjh|vVyd160gk@G%2J+t5QV)?Ix^QpeV&kIXzki22)o1nni_`=TZBI|n
z<FBkbH&VXOvsGb)LzG(4(`7lAj9Ds2HDez_$s}XsbqcHHR8%w)ULEfZhLQK3TrY4|
z%O#kQtrQXYgxWXGzxMwYobeUtH|`wM%ibP%?rI~7>^z6TXIV;<-Bgm}cfLQLiXz||
zTyqjgo;FsCD%)*vZs2OR|4tGYB-#|8He>P!xb6EtnF=iVa`v9}+GQmr^tHDoNIAJg
z-r=<oqzcNlHZhytA@D$>^f)!b##hh%>Zyf@aGjg8mdCNm0mmW!&Gcz%dM8k7T+;*O
zfG>rx-Zb%td7zAzG-EZFM(y?tVPnf66OGbDsaCD-Sj|MR>w=4OYTW*8t(tEBH040*
zz)WOmHwb7D3P^NhJ)e#w<lQPX4FNYe@i*N~^;UhR+}`CC#~d-px0G!S;kl-+d_SmX
z-O(}=TDtOW3!=fqe=^*?LMHAGRkCfW&~NqYEyw0{Wiu-}+2;aguFQL9+mc$x(*}4e
zE;9edS1Bo}oJBPZbnB}|ARMb~uy%4<P1||RPU5CefiCtEGA6ILe+GzD^%R<mD{1mi
zJL<1|UyeE_boym}T3N*u{n6uW%Tn;+q>3j%M4@k*(f6BY6tFk&c=7vOPpZGi0a&JX
z_I}2)JU2hT)^^bcXbg%EPO?aG=^}LP=Sq^qvZ+X9Q2+IZH&4OncivCq-W%j%lL>5`
z4V`JSP=E3^+whQP(vQ7+PHzv#4GbT3EAvVnieqPTztbbVkzDW$Ko>KaLdXF^nF7Gm
z3-<8!pF$bB`FSeTDud76p&BTb&vNR0-+DZt&OS8ttuSZ{0)hZWG|lij`f`L8JLEQ_
zf`He>YA&0Sz)s^^(Y?0vouk6Bu;2Hh!hwTkag{W}Ua3jzzV01Q-&q+;4V%cjW!}zP
zq7kxepwx1Hx#+9sz8;?=7O`SX&km4_m#zKflCf0QUK>22k%NK?cDK~Tk&joa#E8~m
zNMRQr)86?dv25*RhHS0%NLX4qI26Y*>b%wS`fDY6LVf7FoORNoDrXjd*Ac!=NlA$+
zaHSv5(WuQR8FN)X41^xp)s=BW|A}hLsoA72Y=Su&nNhzIFAnA#PqZ87Wuv1-L`GQF
zO>!Y^pYy4>`gWk!fR#rjQ+)JE1JjUYU!@qhZen-t4b?BFpg)=~awwV9@`iRfACA)V
z?d3i|e6H6MKM(*_xO{OJE3OimUe7~_nzah<X#h~PJHk9mI;TMRJ6G?RIbEkP_Im;1
zsm8PEbvnSslzxnY+z^xS$muZNJvKHA2S2~{`_Xj3qT!sAEZZNRT(gF+(7Yuc>T5ZZ
zl2RjU$5Tf0!K}j9_Us^iAcdp2X%Ha1A(m=PP1hgE8dhUbZ5#J%>gwvo9BXanZ`LBX
z+Fxt*rPikLd%CQ4MigUNRYB=-kOcJr-FRFO0hIp6Z@5^k+`=@1khcbC<tN~>*}Fe{
z1MtFfzGB<OW;NOL21DqhghE-*^g~+=aV?L|V-@zJ1>J(OYKG;Tl-S`kjFho;;nZ#-
z&&h{}WdrZF&$F-^NQj6r<bNS_=0xbK>gwt|Nz_0=@3GU8oMEH%nmSK9S+QQK)O_X@
zDxZAAg<K=O-u2cMb*5M4mui54clY<d=eE<2f2uy%2{C_|wH@U8!24;><u+S1?Ct9|
z8LoNd>w_3-?Dbw#s$bvydy#`6t)}ztyq#uyRvfmO{JeKt2D`1oLZH(I=iLuc+3x|z
zblr~)doLR{!Yoj3ywK#HZwkOyTY^P5lh20-B2K$2nWpy;+7BOg&gNVf+)WSDps!*u
znrG@wPws3zFZMLt8<B7s(>N?u-Q2FEm>_W}DVpoJfV|PY+!j1GI%;8k!~l2`%)h<K
zN^pg2bcqZW9)8hzN@r6xiWx5kh4@B@eNJ7s((W{+6g}EkyuD1$tYOTIF>XD+QkpA@
z=8SR2<x+Zafrb9Jj=a3Zuj=@Tl;I}#qfqB@ZbqAd9WH^ZA14;poLYtw-II5i22@J1
z!{1|KVtQMf_S!BJ!|T3*Jvc%<Z^M{8F7_mePzw<y6>kQ^J7RS;?FN>}f=>^MgCBfo
z8c#%(^t_Hc%57!#N4YO@g`s+755?}M8}VEXK-dF*5fu@UPUEfsIu})Z#(-Bfnwpx%
z$CZ5!a^J$(Z4Vn=q-q|pOzHO}FqmJ?%Sp=EI`axZjx~3)UWq6Z5!G8CP3UFPJXH3}
z9?O8y(;czn|KO2}`AGo2<q*l|a4I8-<CqPNGvmzcA3B_7t>5a@Dkv#g&(OKO*e|{k
zn_C6JU5`DSJE2Olg(!MS*vYG_|L&0>8mbQ@{6K3xnyV$R-GqkmS*Z21{IDac;N6ai
zH`r`5B%3k<gIX@##bX)pi2KW|TS2^!N9y(W6|d_15}C@IiZrS~49O=R7SiKlmRon%
zr$$CbbVKP}whO@TtE;Pe9^1JA0RbOO?#`$50HoCxguME5tY0M`jei6GN0;`iv!r{3
z);+E}ST^+f`GfxceiXuw5-~qb4ippUjcdrrMoWZ<iTfWmZqT))Ra5S-^Vt>zf6z4U
zMM?*=&918sK7tTH{kpnhSq7v<7_~;T(>0~3+`ahNYh0D-VEaZR<Ju9X%Uv8%RxQ}r
z5Q@Sdzx%D2P}x$!H2h30r+<W~!c>@3!C+YXhY}5PMXfg?!LkWN5mI$4p>z^4-Nsbe
z{Us%Wn?C$`MvbfM7s;X~<#O)Aztka?R6We@J8>(EiM1i2NsV-t3pdE|Mp|RP^D})s
zE@L`2P|~U7GJw#u*_#U>RNo7zDI6Re5SDjFbEV?{a>(=z3`9gkK)}JO_VGXk2A?F!
z@~i#@Ic(LU9d^IU;T>gPZ_)Zr!^E1C2lZNSXs%LiCn0Z0#A<+;y)kP~IOHr7HJ`92
zS2lgCJG#&FQt>Z2N~5-WQQplF)f-jpu%a(TGX+N?EhvQea2iI^dFqn%i!Fx}^{4*9
zsB+xqVX2n`d!`vUy?xy_rn0;|dmZ&%u)+M}Rz>%oKWI`wTzKU8TsYxktXkbx0&Wyt
z?smQw(!Z4yS{Q!aFZ!d#jEGRhl1Lg?%X+QHzes76wCkQFsNs9;D4E=!jWs5T4FOwm
z&hFPPF8p4a&Wfd!d)QnB5}nGzaGp%6uJ65bVjqXs)e%5G>b|^dblJbT-pu56J@A@P
zQA^iat1#*U!cx7f1M+)VM`W~q4ge{m`$28%P(VzglP;zOTAbz@jYA$J@gxWa%M_9#
zYj*X0d>Il|JzFl(K=p(S0S+Wp){$abxuIDC2kJA3&Hb&g`nif@QJ-Gl_!3fXRa+Yp
zE`fj&%85rF5rp9X+iN7dv?795V4=ZbJq$7$<Z?(<QX~e&9GqS_JG)F|LV_4F^Q0TJ
z+}}-J9sexqU%hzn*nL!whj8_gfW;*7*`3TJBoJs3dx*$-1vAv_wtn!^_ZJmrr6j~9
zb!#&!s+`n4W;KP8OEN~hsVrfY!DuJXHL-4RBCSe>ji$%xoE{RSZ?1G#%AvUTxspgN
z4(6~O@uSDV5NS1p#wPP@-n%O0WJkp(itD;`yjty4OJhTWW7VR9+*nMyp8g%!$WE}{
z7)VvS_C7dv0~q&pI<G4yK;3~CVIuH3inK=nbS`+aMV{E_42W>m=3{wZAn>R}+ATib
zQ)N1{)t0UYwpG55KB5XUHP!^+yaoiIBq+o@WrP4~DFg%pM4;^NVq+v8W#-<cA><<y
zV#18Ly!=_3oRm>FA{6RHA#U}4YV+h<Jg$AgOif#`p3fm9em$~ve-`;2%Su2IcJIj>
zjls)q@%;-gjV!4-9i6hPVYK4bPa;_{#%KXRUQ84bjUEwkcPAtxpP`~cpg@ZjLW?vu
zDRb^ygS@qx=zU|<De?|k#28(auA&??IH#s<g@P0Xt>5p+&-j9fOsRqri#@VGxY**`
zCMU%=GBVIBg)c6mJDHD4%UZ7^J<+}0&F|1|QGqW?DpCSDfJmue#nYyMBEblVNaQ(^
z-zivAFAoLnj)q%fj=EWc*+{z`&Yh=XM<ghK_JN`z;(;`Gm4XS15WT~>(<=g{HUbVR
z1qw-w{*4F%E}yFtse!+34qv2#?{}fm<l-?SPB-<&wTnKtGkQKa;E!j-znr#5dvPam
zv9TkF1)H}Hs}&$OgGnqoy@@xeoYo!(^_wCJX{;t+0HJkVBP9=jAiGh54|O*?#eg)S
zvu@}e+;+SgC72?ew_D!wFsC&FwW<N0ab9k2U_e0HEVqe?2~}+248N>G+=s-MH*kLS
zuK_OinfX?R5OJf5xl<O0O-n(PSaYm<pR8B~0SWi--=3Y|`3JxOubc+CI>jQy1W&+}
zjEsxqvsk(6s<Iz?M<<G-YRx3Y$m0`5lLz|omGg=?ye4#R$=<%UjloKckDmA<vgXkF
zMbXTRh6RKT8}_NoAu9egT*&Jj32FKdUu;Gx(%8SD19d}sUamwk@`&(cDXEo+&ZX~Y
ziYjDOw4oMs7z{W<XwoS-2fwtCF<8j5<wX<JJKPb%@-$S|o%jtx$wreeLcW`1jYi?9
zRHxlzRU(Mue5lU<R`LCdB{^A6o_l_Qm{DPysA`|2f3z$PS7bP1WGER>{UnK{Wn)wN
zz);;+2_=b#4Zf(~V<jU~AghAf)DGRP#jY0ahg5vupUp_(W}#pY2TR1j#+q;RC%5d+
zt1T2nr+My0>hd4fZxXdyPzsTd0dDhn)ipaUXMw-o-s#Qid6Dc5wYs*xJ~2ILVCyz>
zqB)t1FaGGXW!C`eoCc6dt;soY_*Nu5af*Pq$NHifAV|43@u#n>tk&d}YgAEDQs$n_
zQD?Xr0verIjy>R}K--|@x<*$dVdMFv7AmoT*8s<q7mzbHo8SG<__t%hJQaG%uVb;;
z>^&rKySK7>Eq8WsJ*j1z{-Bhhj1^enF6DD~oS#`zQ33D(UlHvztub(S>fBhq+~RO9
zZ%5~7W-;;pSh0EIMK~ThZMZ{NnBapCI%O?IRAWxZT)teUp#~{gejbmC^Lj#QT$;4V
z+Hn9I)oXZ9F129SSH7`$NJ?uTF#WMRBvc4`vWrw@avjyNJG?TOxc_2DxjocyvNEWb
zHhdP7Ht07isiR?(7=zVt=_X>;0^`+dXBjNVIBk@w$lLvW_fFLBQ>Yaqla$T)v^2it
zYm5K@agmFPVqppwQ4#W{%ryI}7p4Nl$VJU+@7u5L@Ia)l4i1mU`^#cwp!ea0A2bRm
z)A{R{eOn~HRhO4P0tlUnsp%pB{((A!YkYlRU?8BSYAGs?nk+=>P(E!}s(7Sg)yCOu
zIgj9dQc{we$7HDB<NK2fUiJ;5k4{%XuQU#uZy8m3_nxroL64%vdvUF?meh|Oo`vyB
zW7<!BVnit)BQ$^iXj+XzCh~%sB*xPexmXN&96~NHoTPGoplRMtDUsXq+<X7(rD^<i
z)FMcq6#)Bz{{48hRBxe>Zf<C{nbmS={5WTP`0HMQIWCR~PXtB<tcZf2KaG8TUWkf(
z-wR<+CWBlmQ#+Vz){8Sb*LU9v#F@03m9%-*5#3<nmQ585APs>)Ak}1`K=iZ^FR}ak
zRpMedL!`q|InrDWtNXOkU{9~dZ}pq801_hLv^5Nri{!6~XmtT>srB1T;dEnw*mJMK
zv1LlT)<#E2sCB8;UszbUxZj~(d$!ircCrLozi7F!z7DxPXcQF{HR_7opDNdjuLs1)
z-O1A3;Vj|xC<I;L1!LQ$<DhvPU!WU1`0YDIlY<~+0w3TR$pZb<?_dvH8X3HAMpoq%
z<<Vq(Pq!|S==3mZ_{nqngrnpeUE<=;SLrZsFAte4OYW4Zo$qhXmGWgx{Ok_v>=X5T
z+}9_X?>gi38k}-E?UmhDdItzHye~A!94z~@3d1V^O1mEtPA0Z5CD(j+(#IH4FMFP(
zR;tP2dn;=(5z2ZwH`}6Tae0sre@)D7w^U_4TWE=oJ~G4ch|kK+_n~3#_Oqx?lbeFi
zW|?sXxTC}RBvQv`hb=)^=@c<+5(HBCLN}^xU^}V8@2Psgl5Z>Ps!orE@ulW(-ZFYb
zDu*NMiI2D4iLs=t<Qvg0G^04zCgHxd-p%XF!^Lv;NXHau_{m4_(*wdq-^crUUwYty
zkR;vHkhJ#yKlQ9gyC_%n2LfF~s6gO8e@I9JlC#%lx+|a(wV?{mK|C)Vxf0{!<6~kl
zaBy4>EQeg{$AOei>~++dDT}3pi+O(BMVxudVmMo4?MI?qe>H0fpnk0ytL^zl*LZ@;
zq9PBV*y>CDu4mPqglcC@POy~E6$m1iK-D6XXnhC#i^p+8L|j}ggV%I0txzc+=->q2
zn^3;417xj$H<+V<L>_^EzLgWDN(-8+vwytyG^)a+|I`Y|<A?@N-wSXJJUYmJ&FqIE
zO^C;NI^aR<q!q<o5SM+;Vpj<v%4dLkPXlzQg!*{jJC4=8@%O*<I$lXRDCeK<u^Vjf
zA$Q16uj+>r{?{f5Br%CH=ymnu%{Dj}L#=FpyRAk7^zi?fC^`>&jrREdGb{#X>}P8@
zS+T6o6CoFO770AU3UQ}p%^00Y9)t4ekb)xM6BQ4il4ri=$Ts;>W(;`GmLf?-#U%WA
zlG{T~L1^r^rcD}Pu_M=u#lVZ<mtWf?`NLVt@PlRdH4-ad3!$OyV<qW>Y-%mo(W3R4
z@#kE)6t;LUbtV`bch19sr*i$95_sX)GoFt8e8pS5Vg6fvSO?q*&Z-e`xlTN4U#gGJ
zB0$sP6;G4LB9Hb&n;c6@USU!1Q_n0=T`c?Pqsz3Ab4*z#g4J<!WY;pTln+Oi6!^=o
z;&(WD#fhT0gfduupndpg)G?9g+(U?)zxOwYJKkGhK?(ZwqT^FDh@Eo{PVHLB3qVHy
z1F8aIUslU5F7Wc~7s>RLb(^_;`BSd_UoIA=RC1C&6=`mh?LUh8015gN(6Sl6FE*cC
zx7@;plYjzowTt^fbZfa_d)r<`t*SwVW&(^F=c#K2FWR|%AjqjK4*a)5?}7}72vK$1
z@_?Z5756`XIm3TzsQ-^|ZTRO8e>+Nf7*l9M@A&Y)fv;$TYJ-0tjj@VQAiSim{dl}m
zgje>U*sig9<nm{^)c%Yy>!D!Q^#Qm@f$7mZpuhjWPNtaXui=bkY>LaR4B1^uR&?33
ziyR2FhZPYiS{1oyCGW6t*%5E@m5r5(BINt-II46^?ubpu<JBZRQcWexGSrhMEH1_O
z*klapdhhP(YGMq0_IhazM7Wo(!r0A6DIk`0H1xjfXSF^(Owb0mz$)*>gsk_-@#DCT
zd%EdmoHB3GCHU{!2#8Dg-3<q_&+}~fV@r@5BKR~M^)i|pw0j=h7xKPm@TbJxMP3fk
zo;=u6x3mm0600|mkmuPZibcv0%O(33&|7ZHb){ZPMXZ=Pwb+dR?dPXlj`&PDRaQ<t
z)c=cEr`gO^`fxBJ0{J$g%6TI$^Q6ap#qHZ<fzIhKa^O}ab@o{-#Z@28A=}3lmjd=V
z>rsN0BfF$zWT07y)gUjiS8aX(P`Eek*Z{dkG~PZOzmiHQrL&kOHk^R?+F06OD!DEm
zALV$Tsvg=c-8%-QRPD>BHWIg3o(b1d9nN~GSC?w3a+=+FUxv4ko3DoQZ*gGKLIq`g
ztqp==qmBA1h0Rd~*f*Bp8}=R<<_ld7ZVV}@_h_dU*g*YBEUs^@4F0_6@&yn~@wNMB
z!<gjmxbHOBI2ImLZEvKU`VQumn&t&Tf*HgyaWv6rDRDnC%$BSv^1_qsJd2{clM|_f
z1CiVw(?}_!+%6*HR!XX+)HWz;<bL?^Q3}wCkkGQwj;k(N9l@&)J#-*Nao8}TM}v5E
zF=EFhC@Sp7#B!-a2`&i4>H3+vwNEbM)Tkln)?$T0uOc7Eq53Y3Jx@JzX8Me3Z3o9`
zRInU2(2umETlYQt3CL8h(4$*_G|VI4UVTs}Gy1%?r$XG){He~|SRDN@^8@>44<<Fg
zpOq_Xt=4%>M5~!y<GhDj9@|vLq0f=Fw`IQmH*iu{DSo(!h#xQw{*5g5ywMuDoUGeH
zZkOl#l+!UGcoc3QB{hF28ctq(f`y2t2K2Gkv0P9_*Z%z<Ux++aE>@U!yA$}OXk2!*
zkbI8u$|rryFe8+A_IRBe<7$KLBCYQ3EIK#}hEA#WgBou?SmDDl8JjuDe&MwrQ^eC`
zjBFM8;jL+w#qVj5&`<9PKiYSBikrrFoxXU<v*|u)@Z!aR_cv(!lH^ar-S5fhJOl3q
z|1`E_V^n12m0zYa4?eo7Nmhmj-(2I0%T?8c^~5?NN+sBFIlsc7T{H%XgtKYf=RWA_
za2o{`DWz8#rFYuIUWa6*#!7hi=P8g?ruLJWe{H-NI&P&vN6I#Sxgu#ggmEE=nX?>D
z;L1R6@LLfy>@>44CS&=Fb*7PE_wgt~DfGbn=6wQuN`^%4T%+&uJIlhdcYX60K40pX
z(04XP^YjMOC*E==*O$hp3XmVEn6$@s2CIsfZ#-a`q#YVZ*(e_087XZXHOvs7J7_Pi
z67`L{JJQR{2z7mm9pqG_J>5Anm`f+mEYb=}E-mgMQ56XtvFnVu)$Qqdooe}EHv+4k
zOpsc4X99B#4rO&cHt6bn+d&{Y=g>xFVqbT$uEcqJ#Ao3yJgNS;{TuO2rmY5&(Ug_1
z?E}KjCUL)2fr_3P5f&_6)nX4g`g?A*tha_Ol1@xwB9`B4DG~CI=kBJmKH0=W3(y~B
zBPf!XD%X8$**!#4YYt}|6KO5iU{A^q47L+~<}caPj#rwW-a=pcHs?$SG}-?lCw=;A
z8SUpU=lBx{#M3#jX#PV{7OpE7j2kN~+t*wDw2QMuWM8tR7ktTy#7nSl=2q#omW=mz
z!OS308`Y-pk7e&4Ioz`D2cyK6`*V4UF@ys}jISZh4F?bM`ky;@rgGXv+WUhnC)Cbs
z-qAv#lFLJKo6KeFqa&L9VUoU=rB|Qce|d4luFmqR2{xz2JNkTsXaL+&usiEFj>`1F
z0cvV$F`E!qkrVanS}f7hbXcaX;@H=Su2(ZUt)8EUuv(W-icRw7nirkGOOHQ#OInQ7
z3y6tCmh}>5S_ZOi?vM7bu*uoap#NFb?|936ELKKLPu$0o_az2R6n6(PQg`nBjU#L+
zzgs1JM`Gy8Ys?Gp%i%(PqRO0#Cgy-H<!?~Qb?03++4RVn(%DP`i3?FSfw4<oNwIdh
zIj;g`>?N-@59jLUS)noJkQaPN8|;4qjM}^*ZU1A3`{u8U7Z(uRyDd;udqMS~7dyEG
z1ussqNKNTqxODMxdpE_$f`a-QVGqM^CT6_n>I(Jf=y2g|wHEG((C0j~IPIsP;Q8b3
z9l1FS{w-^B8aR)Co-b>vYp}8+9oSI5d9CL{VBv%*?#XX;a~Hd~YzwP(-(V9ZcWmoc
zJ_Dts^HH~T5<h8FX4L;q))t%?Tb^=xp_{azV1K_2+lLHbNc-5#pVAxsL;JeO>~5i*
zx4OK$l3AJ9H{Lsyd-T$uCl9sTUr+jer|W!7SjmEfyVnkGSC*T*^2L9Ly+%%kp`{hY
ziJ9eL4bNT_{{F_MzkinG0q+OvavG)|zE-*V*Q2{~6P0ZBi`QK#Og_D*tJQS&kFQpG
zC53%jnY$)*6y|=tT6)&`QcKUBEg6P_8xOK?y?Xu2wGFPT?o26=-}K^2*5yyUD;L&0
z+Ok~x>}m5QZHc$J4{QdeOSQu<dUZc*iM>3xPSpRK2g8A*4^pKQj6OJvrRnw^SgykG
vgMSSqDTkc_wlVa;=V&}|0-vCP%6XDMD>gi??(!i6pgIOmS3j3^P6<r_W^Qa)

literal 17824
zcmd^nbySq^+b4cOK#`J^mTp9l4uPSh98wxd3F%HH1cs9CkPvC<1_1%-E@24i?ykKD
zfA7A#=Y7xFvwPk@_Uv-bFfjAX^W4vMU-`K{*95;-l*W8W@(>9L2~$=EqKt&}PY@E)
zz3=G%fLC%hkiUbc2evYr4oFBCj}ec1!*K5_BqS;%S%{dbYsxm<#S`D;25oP**Gh@u
zBhF*x4>JBZFWtWnLd2k|uS+x5)|p(?h3YedY#$4GyUqvSXZ<j$0>P3*iP1xOL`Ul%
zMJz5eHM-4pcy%+0V}$-RN?}Fzz^&Qy+&6V9bwFgR-^|R}S;`<37c68?Y*YX|Azfj$
zKSDzKfs4-t{*iv&3x1QC&$e&b@~Ntg2}shZsS!vHf#;5dKGP9__FZGmmDOhwlF$^z
z+;JxFRp<K3@IAqZ&NelQ;#R&?Ma{TM#eML9mXrV7)V8ztct@Y^A-ierNY{Xzn3|gR
zY(*$%VA~fb*xZ+d%^zbMV~hKHanhUU%?|Z&9j~($3)>V$2GQO9m0~v{u%ee3euk(>
zNb~5hCrC&iXc$ew^1?F`?ja%Z`TifiA=AUUf=S8~O2%(*oSTryWoD6jA6!A{-N*JV
z=Xx`di@lBPv&uyqwlnuJCDH3_Cu(rz3yY8Ko;x$OXP*uHXmO$T#t?V+y76GJnvH_{
zgbu2zqj}#?JO<KwnAx;r+>Ix{ztv&@yZ)_&@J+8TmFLa*swIt~S;41?c-GC~oRGht
zl*@JcAcc#IyWcOk`z{q6KCfi?Q5&%XeMJ9t_xryRoj?)ts4wS=qrfJAZqM?8t56tu
zp2+B71(rDECov5*VzWNKDq}6kXGZrO7Yho)mxQXt7PKrdKQam}5W4b*;<s-*A0ASi
zx(Z!A3U_H_gJJ%j#6xK@etyoZeAsTpiQ#L&lf225JF?DA5+*=SG$0;lA_)zZ%0iq?
zRjqP*`<SeQe%?_46IB9vuA=FWbG?b?ikYpd+A;&U24{y7MKNmU;y&>sRK?lFS?oD*
znYM*ziSkjj=H%IQ^M;ZSfXMXzD|F+3C9D7YuaRjPil&Bd(x1Z5lguU{;69u!JI0VI
z<o2ANUMecPwY8N=<!goIM0ik|+e#u9xd68m@I)V0a-dZSF5ABeNJ#~qHb%i=DJf*T
zVWFj=p`ladW*$Nm)Xa>G#KP{5uMj^EwSYDLD1gHFxwyDEIU|D}<Kcn78hHg$`y~Rg
zm|%adH)|bM?HwFe`;x0odg3~MfgKtod(mCI1%o{dX%~l>me*!xwt8euSv4(&3t~|S
ztz8$i8_K?VwJ}w`I#cT)AtB*$v<7u@LHM150pB}4$^6yMyPTi@yrba~nY)viCNsZf
zA|;cNB0?GmrN|ElmkE~1_)yf3rsuiF+n`(xEE~VJ0gsd*V@-86wN3n3Vk77J#*e#p
zQ2a*tQ1^>hoR{b4z(-Nr^O1Rbdsp_4H2e7MQhvfDjY~_@YfK21iORlUmBB%gjpAc6
z6~j?gD>6IiSYb0&Y{kJf0@|ktO9a-iU}jtMaPiu+3!$c9Ay9n&t~TODhR}FM7^V*C
zVL5D2JqzE#)6-K+OG{i_TubYBs>`hArjT45E8>2KvZU43)xq1xvc6OaeJdBNzt$Z|
z!lY7*p=s^ubUkI6j8a+9O)ALhDcz_8aq<UXN5qd(JpJwK;bp9cTzvWY2Y8=2JHREe
zHT)L$T_2$$Gwj-RB(yEeo7s|K1$G(#+3*uT9AjkZ{)mibYG#(3o$VhGASo$nVr<--
z%v;hS)E0=XUu(Z?S<<MkrInPDGDFM>HY|y}rGB71k(r$x3myFj;b8#S1g)aCZ{KR6
z<^7x{EQecTV_}U8$j~MCIhWdaCZ)lx%|iF10}EQ1SXj77!crPK8?}3{JZUT|wfua2
zzqfDU=@c>o-~9vQKgJv}PGlq`Ta^C~-@qj{(lYxN%xPLZ4ZLK}^6=&{Q1ng=HMJ*3
z2p8uc6ohckeWvB(V$4i1YFH)}x)$0~YGr)MIFqcQZ@@hR3n-beOwEj9?Fr@dKCpek
zXI-rwMe9Agc&8zXxg+nmdG>_aVbr3QY?$j6Sg$d}aIUVdHn?)9w|SEL{<~M^(}Wk~
zZy;hpsKO7RhQQ^Dh}Xic>*4(a^OIIQ-mR0{&K<;}ZZ#xiyQC@6r+h}e3$|}q35)zx
z?Q3;R%M+uO*!swcSY84B*%SYNZ%^4;1a}*_Ty|zKY^tZMMsk&GE`;JS5n9WZII^x&
zW*j`ubc2FOrnoiASZ}X)eImlb$e$U!w6I{t<?KlFzF4oRsR;@SlBV>E;pA~USlZg!
zYQEmKAyOO`@ZNURLIZ-Uk;fon7xFx_m6DROv$Kn!5WYCwO;1X4zdGAr?2O>*K=|9|
z&u4vZuSZ8m&kt8`FE&eF&^vAaR@*dtjq&&D$-Fjce0G^F2I$z>l>GLzxW`*l4mVfl
za`EiK_DeF>*2N?1g-NyHvN|orR!B(iiXkA9cT;0Ld`O*Hq+LDhdDzDuzqwPtyE9#F
zbDX%D@RQ*0i6uTgI{y>GAw;2MXSbj9!ZPSxPaIp8bcFX(q;#+8VT&Ic+?p5oV*65A
zvd1;oC-ow*kPzg5{&5swcmCB=R>%k&dN(S7jvJYuuU_zil~sO|Di&muEe78%3<}t)
zr{7mk`}z4*Vkvi-mY<)Wn*+7}kB;Su5msQEVd*>a?5Hsd%!d!7X!C;OS$Ynll|;_3
zPv@%NPxJ8e^V8F>jBDF?ZIyN&931>vT%@6<#v<eU4kFS$WYnXzA)}4aychKE;dPb4
z$P;L=DYMtTPA@qpJ{IhVaZcQ8*Fh_S)k^;PSAr2%R+Pusox<;6#iQ-$c$6(0ov$R|
zc3><oA4)6h)C=6#pv8-B!oy<SdP5_lT*XwaH*XT6qZb<^IpZQDIXzCSB03#P-N628
z)k;kn_V@OZJu^VUrK6`eH8quzl*El3Sr>9#`+PcB6VkWXPV7*tbF7JD)VuB6Q?_-v
zy0@B_pP$dnFgN94w0FoN403K_JQOhy7$Au3tb1VoS6Qzb@t^hBwz2td`5RtzmUphB
zV{G1aVj5E@9D;+Q*wi|K2u6~U0jHl)%BOUo(1pe!E1o&KPI5GCibA6_=GmeRxq~Kd
zeoc5eA_sJAN<jF6mr#6f*wp2DaC*yr@lK^8^nsrWnwnZ;i7_zzKNhw@-xe&5p!m>U
zN4Iv*t`~*r>FIE529P}3mC`V8CP?G>fB5j>f|iM$z3L4sAM>U7Y*rcKydT=K%FdsV
z6=Ml?K&PC3fX8>OA`p!En5P*w>N965%tsCfL<fjTLF)1P^&pIZWg~%KL_~y*Ei7nw
zco=LchX(~nV8*{TwF`mF1q9REv&9I)kma?ZtlRSepN#~!`5zyrrly#fm_X!$LZMHd
zJULwHEAU9`3B?9mRyKNyLIbaJ5_CUWa~KewC^sAM*lpzH;o%V!tS@cTeXRJ*o|CUd
zhTPUSQoumNnfyCTJr-X}5cZgh<Wz}%Ga7}f!TG^5LjM=Tcm<rdCcl*;wmA2K@DOn#
z+xZ{In-hVbK0Wl^TaMS~v0z0kH?K+XBM6BUK38_T>2t1&n2O$~pa2+rsa{ao&>%!l
z|8+c%((4FnY`g?+gu^YXgu5F$g*b0lS63px!wN`|<mBWwxo$y7k7Ltbxpo-S{GKm#
z@Qa3s(6)&8;+0o*0#LyVp#Ey)v0k^pCfrt2!(u?HoN+TMfwq42v3GfSnX+l`>b-`e
zGjsRMwrKO!K~FQKv&HjZ4IQ@yI1#&!D{wd<8T|9dP%{o8O~n*Gpe=!cfj(yofrqOD
z5il4tE2~bqY1qhm46S)z(hIQYp#a{vxVTwxIansT?=z#0P~Z~uXGijEIEe@dFrPeM
zs$EU1KOWZ>bX>Fj^E0?d8?l2nZgH&n%%*rlEPj_Bx1I!%2g_z^$v?zh{THURVCP=~
zDSlVm=-b-v-CSRa-kdE0$5%b;Zu#TGJ>U@sy1Vm}*nqXq-QJw`a5Oi(pRUrfL1+)~
z2mD}9mX?-238h{TmY;sxQb^`$JQ|kcV<tsHI(rInE@WPT!C+A-bJWOJ!E45L=JMj=
zV@ehQgK}5CM;iqt^qD|+pKzJJVT}Xn@WzB8G50$h+<;G?)J5?uj!8#GMi4tChI4VU
zJ^eXXWzNuhxA8QNUH2lw=L%Wc1o#DwM@m8($E!S$ay`%e`EC;m$;QJRjhl;!=J8eB
zhTNlP#E&3HvmUz&X+r0}K@zrIHJv}8vRP75Fh@*8<N_3UZEX!14Qs~h6dr%7v!+BI
zOHM>o?YeKMrWS}rPC`!ZH5Gyae7Ml<)vBnlO(Px5P*}<JF#*BxK@W$%L*I+*i({J1
zX`AM2kb1g-<*^d>fB>X=EBsUCtNasD@2kt*W}D}UB=$S;7is64eY|rzjqb<qtKsyr
z!})3j9bu%WAjIgOD0|&EyQw2iIWNHCJJcYzT`3Wm@6~tuk=T3}1wc-DeR=8w0)cw5
zj@y(*H`sHR^<lafFUBsZzfW&BdYmdMDK$4pUf{sa4gWe@gqy{8Ab;HpS#qvN>?{zn
z$4+go&+Se8HJ?xBaK0k5Z+QsN{ZWHw#PmvOGkcxjv~w<VUceT>=9`5U={35y`ai(b
z=Rq9x8tectMbt;c-`x&Psgp#!+!s^#fp{`r)H`xUP>MEnupzv2#k-;ISQb(SPyoH$
zbXqs`24NzJ3K4%_d%@h!_7)UAgS|ugcB2swO!EW%hyM|9&`IIK0Ew&yvC-w}?&;y7
zoQ%x7c?q^R->Fzxdo3zK;8wp@JS!g%;YyTufjem&^B1XGy1AW$>LP!@%?*LG4BER2
zrGaeX(6XFzYTIX*_<YyrwtgYv{1;7xj(r-~@2%<T-2fuJcV~aJ>^Tdjasxi|h`-Dj
z7X!C~E2~{qFm`*Rx~xXEirY6`*0`~<;~m#_?XK^7`b8qJSg)}TC_2I|U2a$FyDr=>
zpCY56xLF9KbT>;7_3>6!RaI6VZo?KOBPL$6<)d(4`_j<R5RAv5-dAh8&<f<by1WdG
zm6o614ain8w_GJxF&(5YnUW#ng_?_w0(7!k!0=E1L^-`Ho#XU~V?F4-jjuDyMXc|G
zrBP?NZnL+Sxj7TAY}65OLq(e9Uu$akSXc@!tbih(0Ka~FJfZKwK^mK$uJ6?!2$m2o
zxU)8#gMr(ZD(C_{!*tbqj7N{IPNr>w89|h6`-BM;XtwJ8G~HpvB%3K`Re6&8OaAXl
zB;J1Gnh2qgQx4{C41lzyw3FhJUm(uvq*UJf6z)<$|AgKCV?DwE&R5eJGC>xFhJw<Y
zzzI?aV`JlAWjbC}cnpe<D1;J%8g5HG`q3yo2L;VQTn0fWL(DITk(+yTabM?P9dX^#
z2{l^T!g(^m!{oo0b$Xthe^dXQ@7*F%x(~1CuK|~c%iSxtF$T#emUDqKrAD2ZnVBWc
zHy7Yee1P=<C2jyRR!kKbODF>Q*z7hO+}Pux00(tlDSmc#>(d=L1XStlyQ88`V!edg
zy}s$q#B;GhG02w{{XiKNK)OF?+lD!z<DhWOq3n5mJi*oB>+5R`<f;iag_O77u+;Nt
zNP2A5?R+^t7ifaj<g;CMn=t;Y%*>{XjRLU0ctaxq-|AvWfAr+}Pm9X%CtNE7D@6Gu
zP5gf=zQJ{ixu+!*GppwZ4lSZT_PVtlBZY}YXb}}@iVj3qO|4*zgplwns6Rk~Vel*{
z?6%}JrFHeRbu}6qnkC50`b-ZG4@Ura2+D`Q<*ACgx+ymazZ=M5L1o%*I0ULka29<0
zqueTt1;d&*P@c!79*>`d6=v%6^$rRE8++MUv1CZ!D7KLXfiRz#A|t)z8p;ewl-YvE
z2gS6Zb#0F$j{1>Pd1M%_gyZY`Yg;{nV)$RL?HSUG;dmU;lO<yUR1G4>VM?9nqED^N
zHT*`h7O{Cq>|d6N%5ezT{G~tV=jWl_iP!c4TQeyCH^u^kMv*-#v4F&F9~5OKw9k6U
z;*4&ws@p1Vf-sKcWKPh|>R*AoD1aAv&_hvF4qhA0QN!oL4feXgi>&j(&fB0f1Ni0O
zD@!I*raay!p@kKy!ge}$pC%g9f+arc=pJ#{>6`@__`F5|)5@Vod2CyAR8Ni*A+Vy~
zm0t~a>;rJf#I39C*D~%FY@u-(8(Im{LxI%el#yk7w-QZVO7RH2*73fBS>2r-Sxodd
zg2zJXq6a2>dZyD0wul^=wI@G+{5Eqr>=hJWO|6!Mthm*c<EQ{^oepMrCZlzAa{6ga
z=OVx|8aB64^&E6@P@=n*3S|0-BHCHeeW$F72SADrs>lt;V=qFhd4tm`wiu`3rSBN;
z#YsPx5|**Dz*@T`N7eo-;Jhhg|L@L*o$m}EncV{(SfsMpke<490Et@JdrB_U90Av6
zd{x)zRUTH$9x2WI=f!5sU&&ZFwaXKROctTx=6Sf7liwXTuF(rop8E*N&fe+7PP&Km
zfw!nMp`GP1v(I2ko&McM_zaSSB%*vdNFtm>s7Z=>Z2J~GnGM}yoXMG8>F$tb%ba%B
ziU$OO?Rb%8*ra_x;Ud+wdLgtYq%x3=o5pNLEi(XYd>eHS(cBn{?j=-*7Jnrly3^mB
zaqZsh5xK?-9EA2FDQVep;>pf7POVF8%9h?OJV?up6C;v504p*(f8(Rq-KnT7mfPjQ
z%EchrTkrr&EudQ1T3(#`WAR7s>Z%!Oe8qPfqORX-TqPsy4wmInKMn;)fk67_M++pI
zLe^y21aYbl<5=zThVm(cO~_2Jiymu)n!dCQ1Tu|6rA2%W4GSUW_g@l%gR%=QB4>XT
zN^a~s-W^u(2^>Zpt+^2_@dJFygK_26pAU$W3;*nXRf%v#MMSXLC}MIJ#~^AowbKQh
zXF%H@!OmTjZ^f42$wm2hD^P#fekGXaBKn*d+;WngJVK^}uf8M{a3*H!Y2ZedITRq)
zo)Ig~dHXTOlg*6-MKhrUaf9!|UibQd5(ieA8pLUGiuHH#{)WX}^+SbQa9QhWgak9j
zK)pswe}WIpf8jOO&Im&A6#|Sa7&1|l9Em|O??o3zLHa!S^L20KXH+bDS;N)-{9L7X
z2wl=5!gpZv|L~HtlmVAXaIc;TTveHI<+s|6^~PRFVFl!FwXlo?ONI{T$JByR`OF@F
zj{+A6i9*#?fNboL)Asz8^*`;NP~{R4-}Z>uc&6r^iMy2$f&)iLe}41}mN`F{zOn`_
zq9jyUE&K5NrcADpo7os%=U-v|177s?`4HnsfXk%KYk9Z4`}l`+G833IVy{Oc%B~#l
zE)$2{?)gCc63Z}`{bxeiHN%})dIAlbxYv`uK(epOlBcZh4kT2ZAjZ!BpUYN@#?tTK
zZM$QbwtrXBWSXV9eFdOxSmaNJpZ+`I^|o<7itt8V5+3DEJ|MBE|D>c<*$lgX|31hq
z6lD=V3$?MqK`3G*p5(Yp(>~NwSxzKCgk2;iApjqC02w+YCLtjqG4Tljfzv`5Bqljo
z&}O#b=g*%{B|eS1qD18CU3U^kn<Tzyi3p8$$CxWIB3RzJzVx5<vP_jk_^ZvthSI$J
zSH~0`^cjpW)*fG~cc8TWb(sYG)tTEA4nPu!Q$pT`GQnD9DLRNX{3Jk0>@gt}uYiLB
zU{tL<H}jaM-@bkO{)QD|%#+N*%nU@JTkj0kF6?oVBIsiM?%g|pvgGgd6PJ(ePU|4R
z(|1W@jl~+2*Wu~0={}C_46onI9eIn`*25KWR-zi-ZXl}Hf)_z9$K7M|jIc~6*sECr
z77e$v2aDHxeO&-{R#l6w<$wF#<^9YLv`AVeDykn0Zb_9=Os1luqV;e^2mmk!21>?B
zK?1qk=mlCv>+9=dA(}Y;^Yil;tlvPcL`9WhF7mc?fw$&#2Or}A`LI7ogI<r%kF^Ju
zNgUQekM1CwpRZLk?v7=IXDg73>L-R@%Qrc=?K}-SZ>XK7<!X%z+-B9&PmHdX7A$PL
zySqcBXaF{BgX>=H>}pV{1yC7U*<k@0xj_I9pPijmRB%(jl9LOGh$uPkpp%Ou6L1U)
zDl03arluyL`DF9p=H8tvwYt!|vk60hp<|QrJrfiZq@$x_Wo2b%W~QMrnr(1<VNE9D
zc?QyJeUI%*P^g3Yt0QQ^VYsqQ27hde^?3BbUyY^ZZjV$6`^w~E;^EeGr6XLUB)|4G
z{pd}YEZI&s76!%$$Pc!<i;*=vEK5P%JsGQuc9$;r>rRQZw8P4t3}KRs`DRcEcH2AL
zg>E)bSPbC)cEPE4rlCI1RrqTJ=<QdzvJz4_kx6yw1O$Jro&^6eoxa0~&`2;cpILNK
z>obVJMvN;x3-EX9C)=j~@|}eD+dcUab^)R7qA=-^x?6ULuVE9bZw9Fy4K7oCRr#=T
z(FkykJs%C5<PA@=yTew}eMax>o(!BXAPh7u-=ToN0ug`h@#8xn8)Q8}Lh{6HpP`z3
z2hxK%aZcVxJRmsz$bH?5N$0HeSO5~W^;Wxd=n-fo2N+sE6B`U8mV{UhC^A=Bq=HW*
z@#2?b2f?F%Ch#fapDeoE-R25r`xfc&H|-!9KF80g=L&aw6#6B>OCofW)p1)Yv}?|%
z;LfRbyzUjDnYG1x;G2~8)ztJ(AW1Cv<-7jUuZPtjv?aHnh(`oKEaaABG)*V*MqYPh
zWzZJB`s<}N#W^okZG*V?EU2TJZ{t+eHo1YLxXMWIk|%LSPLwaC(x0mQd-b~)_YOxe
zNo64ipKdvm74Do$FF01I&ixzCcsQA}$)fw6RJle-Le7(In%w>bGl-ncBS&<OFk*bB
zg)3A)jx4hz6-=IH3;{V2Vg4AwA*e~mo6mpG&6re}E`v&9PDT#5{urlTPd1;q<l2UY
z4MZlbY>*>ZP855+x4o^I&ybacM|d4NEwahD9B$LfZUjF%wzXJfLj-^XkYNL~(A?a7
z%NNh<XM6kh01H*715Z6up2p&RoZkT&H;xu?4O-rQ5eKlBc3P&KO36G$i`jwo3SNg*
z)7GGQJ;oq>I^!Vc6jxd>F5<nh)tksg`r^~h1=Kg4vYe1fyPA8$7X+&IL38pSr@+Q?
zt*;`#_EBZNnDcBj?MvEUEFjMS(fcnKpi@KP`1Dl2&e0rCl2bc7lDA90=(Eq2-qx5j
zy7>sC=RVR0(}Mffob@{F&$XFv|E#Xs@^L?+!6ast>ipcv=2NNDMhvz@r{ag9^+Yio
z8F!)Lj7zIJZoH^*IZnGmo(8>cdY2xmIx;dkimXKsa)gzP`-EJ$Y+4m-f%UsOL(NL(
z1ETjq7EdY_6ciIxLPvd`JeIw_A_x??<i2;OJj9nOl$cX|6Tf)`Zd>k78%v<QyxTk$
zM((o?j(KT0_Pjcw*67(<$E+-*yZAjwr4TlP)cm?v0V2Of@2o`Kv2@GM0H_66Ah~|n
zKI7<E9Zr)KxxCCQdi@Qdr1SW&+I4iXmbxLZ{|$D+?A9Q`s<h~8zj_~e{JF2#2`2g(
z^$}r7O%#56`4z_2AS%)ajLJzv_Jd#1;U;>OyYniE3@m4PFM;jU2lkpepM*B7d)%$U
zMaznFS}mu#IlXt%gWt?!OZa>7x0;(%kiG8vJ`fYDOYQ(FXO7&{W!-T99@0yPQNxhh
ztf87OW~X!-dbrku>CSc5B92EVZJlMQ#dW*KQ)cOyVcUEHO(Ot&nsJ@V_V)YfZ=Ic=
zwyZzjsseJdRNI9B+!vhy(`6sZ&~vVb8$S;=)TQKjWqf`-XD`aHk?~lQcE9~sQ10(Q
zQplG$ez{i^P40smf*t6PlA8*n(NIkQb{t0(0SGu>K=-<TN?Bd%aE>4ai^ml`j7;}7
zXB5w<7vSZWmzP(O91)$1L+g&duY11p3Twu}t*hG&YmW*QT9@sggh>^ts2=Er^%*<^
zg~4OcF?1Ag643RUTMP^3T-kR$JC84$vI6bdoyvW>2fjZBM5{m5xCV?-q5JBUl&q!*
zq$);b3#^zIOx&)*`lG2^Z`0f?d;+$Qx47gU7O6=5V-vCSW)ai0Qmft{#iJn0<INiA
zAa6ctlY;{0z=lWs;OpbK=`jL1e{xY-4QD)o_6Z-c;{`bBZSDat^R9FpYLmIwFTe`n
z;sd=aJSiytzuHalJ#?~<8Qh4Cd<+9d&Nj+<iU<78n3c*qj5zV2*SAiR(87UUKVQF-
zqxD)U-rZ%q+>z~Yk1ec=oj`&{kI?3ejeNH{=j|lI;gic~&8PH>tAQg;WSoDL8h;TR
zg_d5-Np5ich>k3#n8{_o6u&*f5PxZK9q%MNJK=&nM=P>)ZT2TSfj|-$<zsrw0^6hg
zqwa<Gwi~sr=Kp|}TdS*V)K9gvm5>kulabFTZ|#kLHqQCT<<LPO!cjZU%q3~vWvQa3
z+ELldJ@L7x&F)6IETU6aDz2`sM$*G`cJVmlacVI=Eb%x#b0GhGIbtc<zO~%jLz8om
zVX2nuA;`k;cHkyw^<hI1@#Dt_09jlF7+q*ch$W00HOspdnO4@qJ*n)yoC+&0KsCoV
zVCOfNySe;K6*kyjXDv0oN9cY}JC~by*^CBnM#*!Q71E`HVB>IGJlw9$W@veNk_fO)
z+Z^CpKR=bgx?a8J7gz`kq<C{4A2^Jrb1d!{Ba816AFJ<0bzGdao}1Di0zdr~Nki$m
z*V3OV=tU}9=Uj8@XmC(Akmntk1&VYzy2kwuQZ2I@&%3(uW=aC~yV4Q?hXdFHHxi03
z?T<1?G|N7)En-+sSvp0ul*SSp+jQ^7bR8SJ{%zYQP3+lp9cY7l*Pl83aR9WFU;_$@
z(U!gUtEcFH2p)hssSxZ3A*g+ze%K1eq5Nc6-rl$XouHLXA+yh8gmo@*CdwmP)ngHQ
zLKN8hM-EJYN3v|cf9SyUXX~>QXijudmWT_##R(2FROm2)*I93-;&XVU<O^I@b3sMl
zr^kAAK)3M?LI~Nko>}}<(EYQbCu$6O1x=V#XJ~6XX)>ALC<MM;RSZ>Q125Kx4$7Y3
ze*|wf<g%Yr+pjrtqzJ*CPH!aL&L0{<@QH~NY|8PV0n>0|(FZv7I$!GdO+oy;R96t)
zIP63N9{pbXRsGD24lJB*SGFZQ=YD>-4f}*dXE?)-@OSFwz4mxL_X9}3-zHD!<!&AL
zCy(<VnF%bSXm>oLG!9WS$$+p@bl^A1z%q2}9G_UTWzYmNZgP7a45gjb<du>m)2=7L
z?Iwnfk_S?zXTB_!`Dds9u8KFQg+502^}Z4ozeKm66xR2Yr*O$)1#yw@li^LOVy<?B
z>pqK_Lz+m<?p`iNz4O+Yl+{t!N3XArq1+*Vpb_#uZZS7E_<ad4EXNB;PW!sM-4_Cg
z(!5TlvZP6?h<+Q4&fO9{PoA368Q3tvbY}z+^b;uR2AjMt*GF;(61j3cPIqRtTaPZL
zQjNIKj@Ipup5f9^KPRqIsJRu~+?|MY<Rin@b2(0xto3n<6?Pj518qUfE-L@Zu{kHv
z2fpqG=?65$Q;(1a>oO8HbNq(BQ$0<aRrpeFCJU7lfP}vi7I^2g@tc~3M!@~(9pG-2
z>{aPXNLcUw&RGz3BrK61h4Xz1TB5wkrMPM*8UTin1Q2O2KoZ-U+|o?v>u!3$E&3R@
zn8;t(WVxE`)h3879AZzK3ANNZg5D|Z)UxZ=aeLYgf5sjDQ4q{I0Bhd}Uc~fy32u}S
z+^G0Ou`VTmS2?dQyeBJr_V0PoOG_m{I4fIECQ{0qdvBNWuMVL4MVqG+U263$q%tl&
z<c6OAFu@vJpR!mD?%09zu)pq&XX-Fcp`2SE{ULI@M*M_I)=dCjibl!b@GkFoMf;&J
z#T=_d)X}NJ<5rpk_=@Jy)ADK0LqIwF!HMtSSHVkc>J%<4{C&5gW{oHF^?anhXe@cT
z<=k9&x_q@@!y!Ha0o-Xqm)9wnl-GI&u;OlJ{*gDm&Tf8q?_=RJ{hK2R`$aTW(B&qB
zO`-i>b8;)cF;BhZ_+7@7kv&=bZ8x3YV>#_QOC`eFhAr`;gH(xRz1bx+k;Jemlq(wj
z>H8yWkylb#Nlx8|=ber>8$%#Fp$3`hJZNa|t}YGZ4GeyMyKFh3$YAo$7KcX5hOLS7
z_?MYP$sCt~ZC<~vWb$VKq4v0XK4sRr{gIVDQqUOz8b+BGW_uv)6QZ_nAp;C6bd^KR
z`1y&0An)<3t4f=aK$=#C%r`^yM^igqe?Emh`Q4<_`L>P22G?-a^|}!|iZNeWc{x@t
zm@_>G!2W3<Db3KXwqBI|LZ?5NPr#cd<PBtk3yo@?>(s5k;`z2aZ8!9#tdo7+`DCX$
zW8j0>r3%sV@^t@hWYv@v0vm4)zyQVNh8}FPDA0OFm=?YJM)=@%ocp32<fClph5R-i
z1Vlstdfu~jQ&m#~l3sHZkPahpS|CtXle#U}x{Xdg)rFS2@H%EF*467y2Y-1#n_@d5
zj4FoX&p~zoJZPV(0~|xS>V4$IO?*)H(}_)~usoUw&;!T7u$(_UAjlmte@4t#t6ne-
zx=7nCEiJ$U8kbK2vPXp?w9C4htK+N-G>BtjV?io_V2|KX-xSwc-!XJZ7P~!)J|A_E
z+Pq6$IN(oJ%o_UDwxNgP56X80SnUpV5zAm<UGZp}R<xR9e>t3LD}W(<z_Cn#afs1f
zPfsf)4c{^=OBIn!EeQ$<Dew?8bS$KT0c+Sgl0B$tKl9fzTijM)LHOvuHM|TIKcds<
zc_O>8W%Qn_P-T-9_rktwP!NJxXjiKvkQyGyc8^0dj$`K0AncCxCsY`wjr7PcE82vg
zj&2#z-QO&#rVg5gh4hn5o=#K1J3a+$1U>v+=~HVtz%J40X?u!Xe~tW}h|KYQ`PkKY
z8TS<`d`%kwG3(Oj(<Mo3<{;cgzvSeti;0O@J$wktkEsch6VUC#i=;*84fy04oA5b$
z#>e$yC?my65~r-**`sUa(Ol^%pt9P86jF3N=i-tMCEhofo28Ks7XpNU<Ib33(&Bn^
zRs@FAqhIu+)Bx;0D#xX<)^~D2=RvN4XvX2e1!-kd)pUI8L7S^h8c;q5ZiknU^>Yjm
zq&d~cr?I?oFQ!s0!QhX%xOjoQYwIYgqQwIm+#F43nHOF;mVvlNK(QF>t0Jk^tjVQ4
zw-|vPeELD?kOewX8Q0A?xa4vnp`i`7sPR2xni(#Fo!3BrJJR5kFm&wkb(oO7sd#np
zL>P>M$9W~=OQ)Z&u+I!DI~&_I?e))_DsmZ|m{e_HeeX{W5dcX;0{wV1K%>K7Zl&P?
z8va#2@nGRg8@6+!;mOw2moHz&az!q;${n!CpF+<r3HCu3wz-kFPSvTg(siXDj)p09
zdaf@MxaJ7t8H$O~SgP=S-aZU89G?34NvjQ<-xnb#0+pL{llbiq8e{Iq;oZe};DP+_
zkCUhP<pVl}37fWnwy{&vAf+TM2RXWF(AT@_RMXH^d+~yWWi$K%o_FnYPHxVjv0R$~
zt#b|&!uGT>2K#GXn<tFJ%Q92PJzlr2BF+QOQNrS?6*pfxQABT@HpfA?`2FZN=!Bj}
zky84yiTMA>00DKj9n`!?-OY_(ZA?N&MpIMsIIje>P&kx?<Zy8%`T0-1F7}1X=dSAp
z>%M+pOPMa0i&^_Rt<EK7=0G!Zk+!elyfLeLePt9Ey?t)5v)XHGHCCqmN6_cu=h95P
zUgPCX{R>xXkbG&@O!}Hv7T0a$lOEce!?m=Ut(VsE{7^CTMJj7*5`I@RP;U%=$<8oP
zP5yje!?B-y)JJ=lB3d4mnVD8B11r+ZhEh~yj>N8b>7(}Ud8SJ1)bOx&f13AYT??{6
z#clIuiFT=>WLQ8y;Ow|s9qmo7<;cbJOmTr?(p{X>QnqBm=Xgv*|75+gLL&lntLi$~
z%Sf$jK8l<NDW+ye5Qb6+|MtjY9GuVi@nD&*!cAq_)2yF37$$UZV1G%Y>(PN#?=<;2
z`K!*9N1VYK58xF}^a5;4wAO0>c8BD0JGGiU^}-@7#3F^y&Jgt8vIII65}tkOKophS
zm|NZB30=l@lf^3xX|B}F8wK!s7Y3!8!#Hj0JP{_)gk0-QOm=}pQAeD)9p+bBx=f|c
zX@)eH6H=B7hNGa6KS0l%WPzxK9oFQg@HsiXHn2A^s1ftII9f0GT}?v`#i<JnSXo+9
z6(!zoC(=)uE`4uY%_(ZK+6xFhE4_(hL>Hh{zN=mJGb=x$Kjp>SR(}$6rF0Q5(1sN8
zy4X8}vw=F$->=zo<vG@k1$x0)9caI&DuOQb>3vjGb@)wkk5donh$AxEjcHj5*Uds<
zv~KU$9e4OF_N-0vq4rDXxak3(&GDkK0T;e}hZ&8<ju<{&R9VlfP$$42QH%BlByk&k
z*OMa4=qE#yYIB7`u8aTxwAdih_uTIbt-f%1*UWx6iZ*xq`SV^NUVt^c?szdm6o6#O
z&rMzd8i0jRB;YZCE+8Na&<k3jp!tm5`}fB=4pA&>{1?y(1PrOAfNB?R5~B$6>x&Ib
z%7M%6mN~_p2M?x)zHS$vkSje9|Js(gPE}`SSnM($6$K59I7WZ8X3D9*ZWhd&{9D(n
z=sa*6f1JntG~B1*I?&;Zv6&e|6=n}B`~JixDEra{U5ec~IW)_b3OIR#d~J<#697p~
z<9a6qLE14Dy_!$Kok(R7I{bl*Z8iD*%f*v&kRyS5<%WXtaTHj@aE=@>No}t$+ErGX
z4X$>VC)<edis5sy>|8JR>J^BCF0~TqNIz(-I4|Y#Y&a+0?#?z|Q+lKrF1iye-wK&#
zKfNWxU3OZl@tMSw^A@cCZJJwE8K1bc&lm;Cl%b=WFiH1bNqayZ_9>j+NV`!4kbUN2
zHCFU}kkwQjMZBKV&}3~?D-{+V1;}F08pNm?Lhs$+<4o3id8q5X@|A(G$A{Fy=~E#P
zv_p)?#mz&aLMi|Jb3cSq^NOPO{niZWr;Yn>JrsB1`r%&{lGO=T=3D$mc1U<V&MGx?
zOv>fRfFcx70K#6SL)w|P@dqDSt5sl)>bEg3As_+MY4UUf)E^ZkVS>kxl^4w^1f92z
z)<*zLzq~*h!Z*$MlhA?uJ0MPg+s6U4wKFDDYI}jyKy1pLsq$RCToq@V_~y$vGQe#P
z^lyrEcbpQ&UVinM?Pe=|<ZG(xXcmu05?pU7lH=r^6!ekul4n7zcpL54PrH*Re=_co
zlUIHBby#pSYkeihwjsHz?c*%v{q(6Zca$Orv4SQ)f5&4>!H~qDD&t${<C^+VLJZso
zs5G*}(!0H$8PdE^9IRy`0_P^fr3B7cGupAu2$Qrou?PKajdf<RaV7TUF00K-fa)wy
z`^uR6Yj$?FcN1fi!X9B)g<(5vgY-{mDInWyO_tJ7Q5gX91*J?Lg;Lxy0gF05K7J^<
zAP?aAv|d?PUmg_u{Zu;zDgOt<Vr1A<AOJPn)Ho;j4v7jFHO76iw&vvWUqP6M-qv%~
z_}H(Cb&0!rJf->^1VER{ThC`4&49PgbSgiWop$o$hgD#?xEe<-UpX~eH$`;*hDojN
zTQU37odrS%ZCQ%~JY9aX+4H674?{++a<B`ltDRt3?6QOB`QHqjNvGcL3`fy7S)|HA
zL!%&&W$S7<2Ng9>TqynU=6%0~$@N0t>&qw_npZl-IRPe0!nZ=}j@i7QTd39?RrI`b
z%<R1V(@l0}kq^BmRj>a^<w@TDh(@uOO~L3?JEvD?wKv}~7M}npgztZIC-?E#{rNd<
zU0)=X#%0!j1;}ooivhSg`E*4+%l5VTloErJ|3<9Yk^t!KLTy6h8@EJIB$fGK3npI%
z*0K?n>Us=s27mkh{d;C6H7nyUrASS;CtdcXv$o^!`d=g-hJMwFK2I5n$>k+Psb+eX
zcu1!^l<t0xuYKvXhO@p(ls^%$=GY<tzaZ#2P3m3QJMk8^N%OwmTk<b9jo8X{mABS#
z*81I=Uj5K7%8$9b&Be;YwiQL0fZC9qC~pH4epajfiN{fyWfWc)XR^x6bXCT+Gg_*K
zY|j&}kA6DllNh{9*&vd9o`eFW%^K2De&w7@`!I^im=ErSs;>a?osx$bmwDE1^^i`Z
zoUb42f6c-6nE$fIu8Sf|?e~#Zw_|-|SL9a5Q)bU*nLv2gW1b^I<P|DSs(xi(y6A1P
zi^>(|#rMrmUVANQW6jALYEEEY0`O%O<>ft>V>K+cx=Ox%I|Ce7GOW(V(^*mq3O$;d
zF$-bGwb$a}tpU+NpgHpfFd`LKV1`Y^_zoDK3{Rfigw#G1OfDe13P|lTbu-gE&+#rP
z9KqQB{osAuGSAz~va+$Ng^3vfb)nmn!)NMi`&+LBc^q1N8<~%TgDLDf`B;qZQ`G)g
z>~it2C&Q9PPD6ET?qX*O$Hc}KZqDL+!;o{6q&3K%B2O>iWbryM2=ue4QfVLlgPTJx
z(lEA?5l$W)baT4h+pG|W@-!$<UhP`#WH}du>tV<h>&jkes9CyaW@B^n)G|K4(&x_9
z)X2ytFVgc%7svY0o14bshA87iA#3~k#LQ=bjqAm;IlRwg0S_cjZLElyA|OCY8Z9s|
znkuRb0?7jm%I2G-ZAxM-oh8aKyDNVW8LH=+K|$+ITyu#|MPh>(e_%$Pxn3udFAb<<
z1)dWJtr`3-)nJ$<!Xt~(GxpjM#gGakQS&PGxtyN6xtucdec9GItG;Mq3N$|_UQfX8
z4@5_21|aB`p7?y;S7)PnDq!G@=e-u7ZCMB<gTmYjgomF&xI}4G>T%2IJ~vYa&j5Mn
zP0jqlJHY-~>Wbdq-v{FdPz0U#&O;_aL1($GIIYJh;J!}{U#ubn3JUti7JOpHHzak9
z2k8MA@kO_a=$MUO!xCbgoB<R{U!>XNmJKy$E6kMMb`nR?+Sm3*Esd_by77102jv}f
znxaC#2mQ&L(Xqf|evD^T^-k6T@3FW%oz%}!z~HR!vieLL%=uJQ^l2Rpu3PmkW|=2D
zgEY~w6poBp`Tg}{(=1Mz{h<6;`5P6}lClqDg@@0p-+u2@BY65U&Z+Rg10|C2^@*y&
za6x|a!R6`2W21X%bGeTro2fSoywISFJI$iLEia?Og8++n+u3ZHD5mngwC{XZUn4aS
zJvVoZqx)`gjJ~qURBbs7u!g7FrXv7c&*!6GKYe{sP!e~#t`wRlSlw7XFA>tNT)E!7
zRz9Hab#CHq7*Q374L>|vJQY3KJ4bPk$OoiPPHu*t9L5uCSunGbH`@oIk*nhto0P2b
zw9A`Dcc-u>1Ru0nyHp3q-|TA>r%89rMzr(J3_sXAE+}~kD+|2I%Ofo77Bo644O;#w
zpArHMV9@<lO<%I*1I2_rVA}!C-zuPOWntVo^t*hYd&1EAF;RU`J0Q6N=43l2UCyaX
z;~zj;<4%-kSF0?@*M;il*ZrBcnXs8Oz<O9`Hk=)VMgD53o0&N>*6na5=Y;&_)0PUG
zxke4=ZNLrAD|8L{R#S1*w%e4O@Ft%z;<Wr|V@&(0Iz0<SmA=PgcqC7P7B`YLIrx`>
z?dYQ1(fE&S<`q#re(skiqi=dpe^J2iTZS4`U$#nPM8W^uFCau#VSUAxRb-Zbd+ww2
z7{z9;eukYM275^uDAD=G0MNT56_sfDQm`!D?w?;qcV}n%o`sXMX#C^*Qow8Di>i{U
z^zde0CUtgUh0>|D8U|p~3b`KOOG@$+NXiQeY<x{F5P695{C9<&-`Jn(P6Ls$mE<Xx
zu#%xuK!L#f8}s}Oi~R;l*ke;uuR+ziy1E)c#(%J$n_kTMLR8cTFh3(BBWa{Uzj2x$
zEOp&RK56e5&W}xGi3dXri9xDTIj{zY)&At!REg7BB~4A<bdmjzbG-E0C*@4t%{Rv-
z$G(uaZ!0WU(Ny$pM!qT>w_kgm)W;-1+{|{j9n_A=tE@X_FZbFO;U1sc#iy&4p161b
z<Ltqgd<%i77aV8Vj4NBQ#^Pgd|9&AN(sXuJx}Tq4Jlm97_cNMyXk7)lPVcFPpsu`_
zon2L*l=sb&J?aB-+hm?uB??G3<Ez9GS<eQn<SeoF073F%Ynx`brC*UG(RUH<&alB>
z^Q&bMp``$QOaYz|FqQ$y`^8u5-xcf4x7RlE^1Z+vSduyXB)$af;P~`(pTj=h2|nD1
zxUsRZ+f(HX?Cky1M`M5teLZ)3tq+9`0cL+xbhM>B8yg!!Zw(C%9UUFRWC4F~GxQ19
zmh&Z0Kp{XXCQ$)ukG8uo1Bw8b<Bv{!WN`&0U*h#?qJvH4MH?OM<mDgH9PHTe0nmGI
zH7%>)VkojLUFZ-{sA?6}uDyHWfQshjU+lGEN07Y8mxsn2ZUzGS-t}NEWk6+C)Ss41
zXVK$P6*U*ulV4n>y}v9d#X{Q&mYVlkF?^|Fm|@*-O!^(pNAtMI2Z#yg=<;qqbl2x~
z+h}UuDhy=!TR&E5KQiJy-g-hz{4&(7S#7GqTt!x#7eKlKWM9TJci{DSkuJ)h>-l@C
z9DnE|u?hx8os@J)W;Hf(o`th>)%09?s1$R_UOdF;4RNXuAy#14=r(=1BsIONsy`~4
z+Ivoo&u^>>4Co!cZ}s<HykG0);C=ra;saC@k&cn%fOZK$62tpUL&C$;xXqP|c*g|>
z(gW^dCb`xX6F)Yk$kl+)&B(U1Qkqb0$HX^?>YF!bn<dT8o8xTM)PGtp^|n@(&H7W$
zH}!ApiffD=Z3h8US${EpCbZP+@+3HOGFK@bgVJHKBTV%6>JacrefnNZs5a|DV;<nq
z7sebdIg65vo3{AgAItrklOudK@8{w8=_xAwV7bRK<<EPNUzQYUdPr(^f%LjFf>Pg0
z87Khg<VL4~eCBup;n!Y-Ga)e1jY2Esv07~TLc5)Y`xt~mC;Q^Cu<h?g)1d#sarN!r
zZ@SuM&Ksa2c8eW3+1c_5&syqrl@~h-3kxw;&&yx;5&?Ua2;o@y^zG867nXVE+}A@a
zXgmL5=w0CqRntq5&8ghTsi~?it*ihST05Q_{BPnM21payBjT5k;NalI#KhU*B0D2>
z_3?m1)x6uz`yIp#9*a|MU@}ZV^h1jOe(U_`$Mk1T*x`&Yk%-^*F1rG8-zGMb?QH=R
z2<9d>fBD7Rk$XPpBf7CDCa{PxEdKUw&pVtexsQN=pmO%jrjXujQKLJ1K}E%A{19nj
zhr^J1SPz?olMbe$g!dMlxTd>JjktQcZ_H!Hy4sR_!3AJgn4Vx*l3l3Qi)ECteG9lE
zuUbB`@O&@B7uJPN6dX|e&01{2`Z5mG`m)Urx?tdgiU03&u@$9741vkrkR%CiPEL;7
zDTc1AQPCj)<16gLlzRaiy0^a{@Qof|V1&S6-M@Y%@Y?8wky>6=avFkzxCPJ1%Lo2#
zK0{}_sJ(dWU*j6nfgo7}`0I!NouTU1_B?*}WJZ{dF3dmunJ<;?v7C>$)F(i5nqN|7
z(9;Q6*wBaZk&?}-^EF_GkK<~EuXwF+Am|CG=>V|*Psh`_C$h@^XN!m$x8t)1UO6Fw
zST^bzWNd2t1kBCW!K<LGM@UHb1rXm0u#=II(Mjo}lRNU(HXome2sBIp`xhwL4TsNp
z2!i>95gQ*5Fy;C8YPxUVF2pdYfbv)xG9Yv)@9N5j+R2c#`5%UovYg1+pLINc{@ipR
zZF}DpUE)Yu^lDE6Fv3Bu)z;QlZ4HOD<XHp8HSdgV8)7b0N5}Ke&j*kC19)s^uZE;4
zac9+iRA0Y-2F7Lpa0-6FFtLHHXHG4l1E5VC8dCpju*rfv5JNHdaPW%+EYtNC_B5aU
z>C>nD{E0zV*Vo7H!l=1*ku=gLn-e8K&rHj~RE43f1s@1afSSB?%>~{D!D)GAr5~i{
zV9d<FWj}RAs~4y!VMstwuPI`ZY22;`44CiMEMo0i)3xGULjjf`AubN4F{;65!MKQ#
zZ9c|>2X8}xIimX-l)ib!HwL^eh>(Z@y4hw!+x+r3tibLNSWblKjqBRr(?{VY+!MPA
z*H>3t&Vsn0wFo#{MT>jr5~^*8y?{dZ!C**IpXnn-r+|O}0Dyu?K>aGKDd1B%`NoR0
zDlI^mx_9s1-Pt$inww9tfh^^elsf0nUOB)G!y_V;<m6~FD=I3^&RqYV84gxVC+iLt
z)|kGAIE!B6#oj_2WXVzBzSxg>RY2qd^zP|XFen0~2_)7zuUc)~6$MZMTqqbJhConI
zJ2jFvnLzH42!{Wrx3RFX)4>QJP_hHWSjj;*3w`th2~{{?w|`Dc+j+kX&62_-;YRmU
zq2u7N0nWy~QG|+W9<Ukcs2?d}f-YMK5r_B5l-Yh&K061;E*K&*3kCNp3<}}=d_!B$
z%vwS&0H249>bM~0Tq<P#(*gO6sYy`sflg1Q;yt9W|33Ts67PJX`DQFnCD-5IAN+ni
z<a-}ur*5+tIMmLcKZQ;vjb`N{8qQae0R#{n5s}XGKIv2sK(7|SO?Y}<MbXIxeap@^
z2Ga?kCtJT+To1mnVs9PkC5?G;?h?v7J)o%#R#RpA-sh4f6A`4mJ75^-X=WHHF9jHf
z<mN{7IS}=;k&zJ~2NaBss#24HQ2pT^vK7xXn5haU<5vdse_(0tAw=L*=im-XOH2LG
zuxG(wQ4lUog|{h!GG_YWD`L>3V9b3r)#dRJm@%jVdgpz8q3d&f43v-|6ND@uutYG*
z3267HX=z)4GBn-b=3s8VUSqfT2vq`LJzyjOL8W^)m;s1?)EhvbpwP%FrH4o)$gn3F
zOknt^E1JPW$<-V|=L6e9*xtPxb>!a2+5@3Q_dO+)jRJf_6Ho`a+Ay)QI@apUQlot!
zd^E?2%pHJBBgLv!@eNqmy5p?tV%V(1wGDvXGBRGqErTfx&tIVD|LKutrA6$>dd2j1
zfqD^%fa4laGL2##rXG7eJ-xXmug3VDdmrL3!S_=re6g8piV!-G0k7(^PxxVrOadVR
zF)st7Xaa*>hX)6ru*feajUt|ebpVFt(biOS&<?1ifxdBefMNg1Mh}<xo);b`TZ4nI
zi_+1MxCL~6{_~O)gOK%CSC<>;=u`mGz31IvDpI&X$di|(Ai9*X1_lI}lcOM={p-6w
ioDoX*pUsS)``i>o)e;r9A-;D2Nmf!3QY>!Z``-ZhBi4-o

diff --git a/vignettes/examples/timeseries/timeseries_anomaly_detection/unnamed-chunk-11-1.png b/vignettes/examples/timeseries/timeseries_anomaly_detection/unnamed-chunk-11-1.png
index 4c29cf710f3c08f4580446c8722460fb3b301bf1..c9c9d6216b743ce4bf0dcdad542e45a375cb576d 100644
GIT binary patch
literal 15315
zcmd732UJt-+ASJH1q1{YL5f7ADIi4zq$^6V(u))Y0#ZNeHBm%S5m35R=~4rP4j~lj
zy_e8?3jsn1HMxs#|6`wh_Fu-i<BV~~U1I=?6<F&n&pV$v=ku;FI$FxKS6HrqKp<LG
zmFK!3(1p(+5Lq$h1>j2726-{?gX%986L%1Z`uh17+2Azv6a=~hQhokZ-zNz(?QL}f
zPiouk<JKf(mf1!qFxz6ooNTY?$I_57-Tl>0&tSu8mh|f$UsD8!y3Ku78?0s;gXu5-
z{=pc?6qc@Eh|Br?AlKNz(INHB*bGwWXvqS}tQu=e1_DJtwcWf70yU86yFs9Pl#Mi?
zcR~MOpWYMR=jAmVQQ^aKpdXSRtcQOqwCBwYy*m9PS6W=KL95jD5ut`7#MdU9i|Y@s
zOnY=;ziW!<%!F{=+xrs!A$#DhF<Hs`ahrH2p#<F}xXXK|Yu)#_xK^(lIF1z=bo4##
zB!-kpe~RHZEwdq&tByJf^xX=Aot(BUUUOS6A~%ci+lxv@<NdOkXrIiKB0S08d5({;
zA(w~k&AeZ!w*Cn|QF>*Rh26g|%a@0<iM>QpBxm?63t_mf&AzHpTe^|a>)CZuB~`;v
zIKG89k;bllJcuBZlQoq6uGpITAz-t+%7$KHt;xGP0I7P()lS*YmDC!T?L;*~lztXC
zT;JzE#*Z<tQ>uK!pHu1e{ZV;#${6*J>Q&@*hmJIS5DLaTP;a_95l5p?4|7<rb<R7k
zmb7O>b`Eb6J68HNID);pp3ucCjjv?<d`ed2;7=&koIS%d#gua<#J%-C5HLVt7ZwSK
z$NNd1>XY)9eIzV>hVFRe@EW*dkQOH>OJOs(9d>7z2{Su;9Dqx=7zs3g8ER<1+ba||
z<mT6t>UjxaQbYf#9qCT-2rsm-rJL!5+`cht-t+ct-@d{Zr<t>)jdw8}Psq}wMbf?|
zf8Y9gc{rb>g5-l<jnwg_n<(m$k*Y{HwP)$28tq~x1u$6g=GNtTvp=r#4t48|TpHfN
zssHqjC}KuWsa{X;YlhAE9@P)rLM={TVNor`6^&M?pKQo^McNp7wS!BPOk#($wF*}S
zRJ~>l62+)?Jc`Funu=DU-K9)Mx7iI}au0KN##F=wXCLf`9%ni~Q_tgyF|s=OWs$te
zIHVG(e!@L}S#mLgY&()ZvU3{sYRI9UMODS?;2<12k~k`~ns+@idopvVhIdQ^9ZLhW
zD|u!@n|qiT$W$a7(6)n`6^SnzLwFw^h92xxa`D=E&o*TD<Q+|_E4VpS+%Zl)livHr
zACK8|cTh9uRW;A_il_|1I~opJ=$>3EtaEJldK=-L*z6yNuT7o)?NcH`wRE}h+l9qf
z!gh_1;%OzA#1Tw+caxb?!$vx59P@dvg2fl|VUf0y*6uI~)yH1uKkjvv&L0>RvT=8m
zkEA=An#78_22A&wN3tad)x7zRg=5X0b4xyohsv494)tm|&e>n@j<hbkmI!ajaqzqx
zxZ?`0oAGWK#)v4H^wtR%hAaK<`Zkq%bgy#jf)ZFGX8GF(@Ns@ihRQ>wv;M}X*TZFr
z*v1?A@p|fREAt{ZhZXc-2R9J~0k4rM8P9N5hBsk>f!|)l;>T7V4N`WjmzU>Gpg8*~
zf+ilwE=N37G>Ek8C$%u!n)E)FD*BpSoG*7&AXwWXOJuYoDcBOW?ezM3R4!W3gp|H$
zP*J?>3M`6x&o=oYr<9Sr$R6}XsV*wCT!C4EeKln9U~=RN%NayeA__cYE){6O@5pkn
zaob8LxL<#5q_zZ^<U7Bwm`t4fBHE!;n`iE?@6OPatAw&B*JQLdswMpvj0t^=M&}O4
zT<%UeR=P2rO|WobGtS;EbC=?<IYNCH-cS?Em!x`S^auTnTffC|L@@h$WRCYWX^!B5
zd-9IzeP;@?{vq0@hRZy(NN$ed<qMr?_0obNS^Sl?oI)#cb&09^zwUS3o1NY9dlAE>
z8m_rwIYiyu!{0cW^FU|m@E`q!^%~u5C%-6qg(KwC5S^uDl>>2YM)NGP*Tr0XaXOB5
z-vrtgkJnZo%utUsh~x_X9;Re6y6#a*?|^T1EZ~w3OYX4kii|=GWkk-mN|>{HupgZ%
zVh^x=mMHfX^#celu9T=<QEI=wj<S)8u)_*>DayA+x>+=C#TMzRD*RUWN<}*KzzpGs
zzkaM7F`n5&Nd9T1{scskJht<KP`H)Se3akDYKLh3Z2tBm@190h5pP0xVV1~@yw~9%
z%ILWLGV&>-f?lDECWWfE+u1RFGQEQLCSf+hRxM$}Gv9IRLY~3f@B2nxPSgWbKYRm+
zvJn&8rJZk2NHR%Wbu}8tZc`1tV(e2i9cShhl4bies55({z~nrx%Vm9|R4m{B_qY44
zk<R{v^`RUiYPi3hc5*fT$(cZ9<kHUh5c1lZSFY8Nn=E+UXOO#kK&TK(aW^bDb$k}W
zvaP!6Z84HxXIfU2aV;`5F`p}4n|Qdp@K99Veh4=8t|bIEDUt2_A-PBGIr~hYoQZ*7
z|KZb{Bav%z1%zP~O!l-xowmYg!2Ge>16ef6je9IIt4_altsLvELYyG(rJP+;;&@nI
z?f2<Ro>RpGsvkX9QS?nVNK@nD=G8dNU^1E(R(U)*VWwQR5%4R^sJ*Bsq$z1qkll34
z<bcP`Cyh-Ia<*MNNs2d^Uf86Ie-**}F~YoOmvu}Vqt_>5bMh4on`U2_S_@M&i^U7^
zP@O<rVN>N_S3g9dT;)=^)Mt`W6Nq($g=caz!Fk#ro$J+~N$VfyMTiOhURN7fc``;o
z>oZ7b&+IerP5)V$z?Rn3JzQLkuq;}N{A*pqXF$RbdX`XY^KH8H7#sbvEajQHq_5jQ
zM&D~aoHoa0`f--slTP`3DB2f|z{Fb;T-aXVC&F$1v0NQXi_CnM-wUWryj&o=L&01b
zu_XoOznOpHmNCeZCM{V48InxXtf!nIy^<u3U)oW498J5VbXY?dRDZ?Y`EkI^IOF6r
zF3v)ts+@A;WrF_Yo%Hc=Q1#rn+vw;`uqPkPyQLjU&f}t%E9x^);mJw9xAHXp=<EF8
zH>m?%!!6bTs8}ETE@E#1WHzTG4gy_iViyIzD+rT=e%-nF|Jzfq3KvjubaMz@PEJnZ
zSiV8NVV%$0O#(TnPa)EFVpJC7E5ymkdFRfrXCYU4@P~(oS8nnNysYZyF(+%#D{1={
zq#C>Iw%i%UE<2eC#czyQw}nq+a)W#rldLa;o;;<5fItkvR{wTd#m7po@cFj1R~=!-
zUr<nx2Ok|BU1(N0`ujyQaN~Qyl#nNs91Dw!AvDa_85nGINynS;sHiCC!RzmuzC4@w
z?+KxiExBa1H+H7Z7x(ohKSbBt!{cSK)qACk5>+G672zP}cGc(4pGUCE@);F5=~`C(
zHUEKv?kmIn*HJ|v2CIvrI}FR6G1N3PkM7sh)}k|+L6_K9<H<le^at@Ts@hB0SXup#
z_uoVv1C4bWqyxp@0WJrB{0MYKKvs4-6MC>U@t&Mw@_Yo?8oxCu#xC<{CVXR+=*^UK
zU(rb1CkOF7Zu@^hW$A8_xh)>!H<P72Be^;<1-tf2tlOM`VF-F}b=eJ=8S8;`#g$nR
zF^@GJ;O@T~{uz6FNI(3>1R9;lKzsQEaA+Qry<>VR%sH!<KhhQZK=eKjFE8wLw@rMV
z0%UP9r1ZR#LChe~C62%Q`!A<Ool5m3P7|eiIa(7n)SzO4Am*dim7c^|=!vtkvU1e%
z1j5c47!2bJjo(f)^>cMs0J46G>L26wFZcPEQ%Ca{XN|BbuU%80E%L&pH4~X%N}8m*
zEc_2xK<`eQuQV2*5T1kmKW1-_F~W^U$26@qP+(?1{w$A(^)6IkWq6{^5g-oQB?Q>J
z-S#)=l-M1QheisW)rf?0d@IG6#J5dZAH!rlcb1Yx8c+jF02*!rcH1_cYq(ZOitALM
zj4{hf!4f>`+$2>i_V>s@?b^VdybKaYt7~i~C#m25B#kWX`zu4^k)y`&B3@-+cznLK
z(Alah`Uw!ZJ(fZ+>U13fx>IeMoOWD_$~h<HK8~Y#5stX$I&%}|)o#nFFro-NUq4S`
z#^+had>y?Gy)g%~f!u?(LP=4c2M1pEK2Be?me&lLW>J~aX!<!x=ka2WyLa#2xzmtY
z?=X_5l`Iu=??n@^8rsba(Y6MXCx=Fa`-$`chdW1{gx8K@Z(in7o^&<1=UJkl8A@Rq
zU;PiZN>oKb0*wo+@tzcI`mU9a|J(~s^00Tv&kLi_Bta*F2<TyZafpBPz@Rz>=nkyp
zGzvvCS@tMdfO-*Q$wOKb!|mgv{E0K(#K}(;6s^7T2}k?06rf-9B{Ksy&-SOQCgEN}
zu0p}|P30D2ql4X!s&1mGPJ-Pt#jlzNv(z=emXwsh>-IX6B;5tj{7xG~+VW?(K6&$(
zFEuvodXs*NJ11zGUI68nC<(J#YrCGfYiFb{=TD(w!Vs13scN6gy8Od!VmH@P&M>-9
z|2}wHJU2&MT3Sph%*PW9^OGOF(Y3PrrFxbXXHlvu53<r^q=R^VEXoL~NHVLovbJCA
zG4pl}kXRXc^FBl_Z^YWVu*KiP&F7c~fQ;wB6z4q@f<0PYhLP-s*MpPk4Q5}uzfzih
zC@5ob5%iAN_$Zuvq*?h6L%v~OP1hx0h~`fIUJYGX5mrt*I1#6-#yD%3COENx7kuB#
zhy{V(1vKw93QXkc_7=w1Z6(=t3E~65v(GB?x52}b{XcK?G(`0#OH244IN^?C_>Ghz
z*qzUyAmgZ;K@&M0wX+4HpUP_XIy&KIyTML#xe5vjYH@<j(Av4*FEY*n#<8@w#WHW<
zw)`!#8Ufz2_jXHbYirxvF1lri#iP9^jdodm)p;*hy}L~8UVuPL)?0rr{2!;ArVqhZ
zZ<z#Nt_@^_z~OM!Gve8a`)V&{v@lV#<^ss}Zv3ma_wV0#MpuZzdU35`tWlAXPJ@y7
zz@z0jW@hFUe70s{h5bO&@fk=*;HoV-BX8GUB)589kA3;1nB(YpjWnneNP<?h?Ck7e
ztP<k)+2sO!)(5kIND#>l`S_8R`QabaLNXo3Am-wq-QC@Wg{Faf9lSjFOVo_xnP8CG
z6X9maHzjKa2Zz_M(^bv$M<z2<WnMjE=^0arwM~3Pen%c?qA1XV@v66rKw~sE+QM&l
zp~jw>*LoGY&em6KUw9&?1l)w?a`Pq}HIt>z6y;ZcRXj$m{I7Y1sPqB*fD=q50A23_
zLh}C|0DDEqk1UjwT0c@V-R0!e+yK&)L7skzsyfIm%<A$U5>W7k#wAT19UYnVz_a@M
zbrQ=i-Wj8W{JP@<+%+&DK;tWukVP4=BrPufCoJq0;p}TK1+o+bY2m6%clBxo5Y6Ag
z*Sgz93D{ZEQk3@b{<^iUeBiOi?m`QY_drY17e)7UJ0iI){V`%u=>lCVmE#vcrO$=Q
zKlbJs<m>9`^=y_qj=78%r{Pb?b{-27FI-8^)yv6_r=q4lnGXRwO|yUw;@>y5b#>*R
zC-W9*?`@~?;zph`AXoYSmnO?!yG%M(%wes6{yaDHCYw8=?|$<BCL7ySrK{yXf9F;m
zZ2CyuvkBa?94ZP*x%Qv>`Zw;2P(v`IbP1lUa+`6}MvI;m_LQJ^P2B?V7_;zrqUrkY
z2T>MO_n60_JDJ?DlrP=iod;MzppOwI!s^>(&J>!bev2h&LhqS)UWvyRu80V|nLuDh
z*iZl1d2}84L-o!THiY=WUJ7l#VWCAWRJk}CvKfaSb*tHLX2m<8klde~E`u6gce4wp
z%i}y}=cx%VjSQTptJUME98o^G*{K--Bl!Y89j@+;6|HfBkDDPi0s2OPu!?=jPth|~
zk!dRi9y*MvJ2~^Jh?C9*a(GMp?M_1@zg*~okZbwvlKZNZ&C4InS-eR5eP}6;U>!;S
zlPbr8Sx%rEpG<^VUC(y5<MqVaBAga^f@?a>EH&Wnr_HbAKQcQH+y?&Un~CtIU<!@z
zdb4(|;eien$1pn?T;taCUhA&<)G+Y8<)4=j$*^~`9X-<9>)^9l`tr>PYj#`p+`_3_
z!2kc4XtCUI$f);QEVmdrvm41bZxs_=SlBgdQ7%Z<i^7Inyiyh%94za9@aWCppUjlV
z$cHXdRZl-$p2+0?MgOqRtRti256YEB%71Mwru~uT0F}C_xB&Lxb;R)KjBJ_1L|$?m
zJh)A+UoAYPg?yu8TXMh0DBo8_=#eGJS1EsFw9w3Zvg{2Ho}A+9pFe-bGr52L)d^8H
zUbZNLZkE!ea!3BWl{z<P_(v>`pvvqbfSG|5Rp1f>XV&z4|0BV>9b+eM!zNIt4NRE2
zc7Qq(t=YTRbC^6ZQb+cmLTZ~~Ki9`SC(6M!kCz~1$JG4wDEWYsv@Zr=OZ~p5t`)wh
z`N8B4%(3qhqzK@h8j!{$-?b3eho8WqKlQt~wHfcc*}^;=8r<gzjr3$TeP!2HsdWLg
z^B7?zF5`2@=r{9siClqT9g3^Od%tW~TjTr*0l7+&V2QA#9}LWHkTEPe>g23``~jSi
z9Q5S96=QceW)^%r-PT<H_SR6YpGcf?6t}XrWyH(d9gGSFHM6AT+<MQE8kE+l=R<MD
z9JhVGYdZdOd)CN=n9gMZGSI!>-E_8C(oq|ABQw+GEuXCgjW#3Lm7@}~B5VC+gADX%
zmT9+>O`t0*kjAZkOq+Cx{0c^)%KQrGom~_jqx<QEqwQ*Hz4Py>5`$fh7;TX;rYe<C
zbQTEY16gSlXv6*-ubz#dBDM^9BFRIUXSV7y&b~UkPu}b?4_Z0)$^~d_5ddPDKC8XS
zj1Nqg@~(kEdt|u$QX!bcEBCNvqu2`R$HjO9P2_$5F|Kl!S{tj)juCf!)z-^d!I27Q
zbVI|Fs0H9B*Ox#iU`T8kyN~->e|XkZ;tk}#%6A}M)JnbGRKVZ$ZcN%lXC);iot~X(
zeuYAzZ{NPHUMC00UV}iqm>7#bJAR066^-N!rdJr9d3o0LJ~B+c10QGnSFJu&+<CGL
zMm%b(W50Fla5>J>6a@y|+v(mjG$1B(BLw#yExn3NGslzR2hP#<BG%Kz@0gAzm0jXG
zE3dN4Ja(RZ6w|q%)cohmjdPaD;A65yDXqVO+S2duc3g!kWls7U6&PfEW3_C3`MJ8b
zHjx71dkOTNx0Di+F>t3VAw6cJ^7nqRmGCWYG1k{0kU*b+OF1irZ!NKF;zdTu&|_aC
zwrW|?O7;F!%l=-}^Id^x4Bj#N^fiVYq*EL(f0V!4`DZ_dMAS|Va3=&+%F7sV5?13e
z4q?;vmE#SJHy$BK2~#_}%c_Dc0abV^(7QI7<=2kVge!23g#L$f`LJyi>s&ILp<HOp
zafd(DE4oO(x|Aml0RsKfY2RkUp-4Z<Uxaw2jzn*+5*1D=uEGUmXPzqZk%PXABCJMT
z)-21g$KOaqp*J%8-X{U(I3^kc)x!Bf7Z50=^_0!Me}kfCq{99Z2o!$_krY6fOy;s?
zl;@l9jmgz-is>o`fr@Y8>})MhosCKBAx&W}EnU%nsYcL*s*FbeW|^}V3N~LR7{dK>
z?uKNNfr=RrRu$AcLi2+Gt;E^Zash}?u;6j(CD(5?a$DHtY4;;!1PC<u^QWv87w+&N
z1FsbR#-!VVk|Vekex%Il2%ZWM7z1Pv2-F8;hY@bIYVdTiY}3vf=~H_$D{<hEVfjlE
z2xOdcl(tm^7TlV^<2T-0mb>WBaN`elkvStz<w?`u{l5b(k-<@XQ<1~0ve)mf{P>Vl
z^s$X~a{tK20PV-k=G=dtUbMX;mh^Ln9abMckgDKI$t|sPPzF}p>;Gx(ynO+vdqz7i
z&A9nJ?;GM{#?a*r;{69(AG2qok&X#Od`D#0rvfM|K*#O_!yhv9xnQWI5Pm#xwRCCX
zw|R2acO7K8G~f!*Kct3ehFy+{AnrFGErn<=N0&<&k=%aS^yFzDK~+GY?-c(-jDy(!
z0HJc86F469<Znvk1VJpMVPbbjdTU=7ve-_k*n&U>O#<<rCvC&#gL_l4jRveo2IY!>
z_Yblfe;W<ktDbG8Qa^ZkH6lC|#>Meb*_Iq+bqV6%J&8fIIzNssneRW=CK-8b5xX#{
zmY&5BYhY}YB7h)a@Z#<e-5E*cyaB5)e+FEIL|k%TDcUU*KR}+qWZ8oIQ;E>2pPz8E
zMa|~AADb6iLMxpn(hU<WR=iibnE|D`Oi4{`hJv11!p>}U``X)$%j{o#xJ>IbcvI({
z?TxxmX8uOoxa4QTOpF{H9Du&bGcHM2MVcTWfL`s~oooD%BI^&V?(s|_ATZj1bd%2c
z>)oPb>_26!?}AQw0rkvrPu#qh!djoDHi<B+Kw0~)w^_<yNd!pB#^ctF1$-k_2-1Kx
z=@bB-^w-LyB^c?!7H!zfxd<+<Y6~(MI?OOYp^I(+voUQj@gyAJf&f-J>S}0fR|3>d
z=R19OJ8hWKA_n8E`z<sS@y~0#yg2}6PGsP`d)Es);~WJA0pi^2J}_4QYKORYgWMLk
zr)y^ea6PdP%rrJ4xq1>M`lhF+0p)iI)JStj0(c_2($$rS(X|ZNRSXD_X{1?V0X~TP
z3Lti#{{5(Q1OlO%Kz4-&psfGL<p2NBmkI!jzN_;fo*w=Bb%%n2BFfLl=Md>n033+#
zD4-g*7ujfac#7*&j}>^M3zqWu4Y&+@w*f8izqezQ24a#*ymS9Pce7QC?FG<xxG`sZ
zd77EsRlnS?@X(CVTEKilMEU>pn^q!u@V%&LYWiD}gS&H2+PsN#_D2M$jE5+Px$7gF
z?C0i_QaSlh-_&~bXJL$%G9b{=mloH!LG$sN+O4W-y%WYbcE;ge#m-?T(gD-VtFUZ#
z6}A=Z1Cts6;r3`lu_F2E(fm+%QVM#Kx9h&q0lPN;i_VYwmqBwPjI-8;F}ELL&rp7&
z59Co=85_$1Q;*z~bZy!vZg8*%!p6r(2L?j{@oZVr6oq=Z0X?+a*<>O~{<Y`u4?#bY
zH@4<#CVp=-c5(YSS=u`Ya9+-JfQP#~!U}-w@#q-#j`caS3}-lt^GEtpLCJ%9{Xe{`
z`iu&IF_k12IGK5ig~hbwwIa0`u)cdtWa9BmaaYaer2}31uJbh~JEvY{KUjW$E5j6n
z_84EVbr0g3v^xn6V-a&0{{2Js?R1SNkXvIh03}419JW}xO=TfNT(dhB!p$U(F_@c7
z$?6(I*)tQvoHkS6;X%Hmlps(9VkQc;-<Tt41as|ljF}#8>a8XEFV?iMJ$6gA`vwB-
z+;U8^jx8rxM~R+w8RbX&@vHHwjZ<?hW!4fr*o7qBY8iAM2Y>n`4lv6)iwdU+K>rz|
zz)wm5$sOq(bDY2pH#+8<S=;!&JNLo$GKTc~bOcPXf3m-q?>Jd)?6(Vw<`5NipD49m
z!T;b?HbpT%={PrJ#FS5=PQmnU5I22ed;25=T#a%9xtoJVts6wJrUEP{D8?o3x<aY9
z5W7yT&pUv^vmoUQUKL&fa$m$w;J=BQ&m+Ae&#nP-JZ`!0zA~W1$BL2xt#t0<QO5xm
zs2v0%?lLu=S?`B;IJfGAp9(WcWvRtnIrjsl9zPxj{J?iq|M#n`DbAQd13(e~&BDsM
zx;l*w374r`Vh+*g-iQo!P_>nsii-VI)h^(wXlww6YB`_{-`V|1b2<kPSqP|t;r8w6
zOw}7v=@xL;(Siv8T>*t?Wd^v+W^SII<`qsqRYiDsAV+)409qDH{>yr7O4~Zs%pd8E
z7kLIBOB`-)YN|sb&*kNuB7OIfnSw$SU;^~GYKDLQTzvjirsu+|{(yde|NcEsOBAnm
zc1{i&kme1cb8Mn^{uhoxAXeY=y#W7})BW-^#`PivMNVdBJj3F`g3}=2#3@j&1Y`hR
z0*I&heI72ZzlZcA@MtB#$2^}BU7D*`SAH}B!1`DC|8gUN9_S<u4-W&vzT9<oCbMX0
zeI}4F4Lv&s#7lz|B_y#r;Aju<SGqR=oer3#rBL~IZkahb688O8y-AWhc;K1Sb-qJ9
z56BFk38TB2ECc4Zwq9ncT#Gt(MV%=LlRrsQ6h`X;27)?Z!01lZ2i5|Z3c5$e1S9+3
z!R+6?{3eX<*BeYs60e#r0-_jHOdDkKc41**W3<rGVz$;Bj!z_u=8=_^6%iT8{B4*I
zKs!J~zpqk4xPiDT^Q3-v-5do4N>Q3QTFDP{LqbCOQsq%VBgEIQw3}T-wA_|-TN<;3
ziNV3Levj_mYX|I>qc{sVcD!U!8e3<rYgz7K0D)u!!L!HK==2sCMEw4Kfe}1KHE?sZ
z5D0*KGVC%w^sKCo7T#~)?&5F=sC)zBPJCgS$NJ!kat;PaU4t|_J{n1nMSyJ#Jv}{{
z(6#=w-@oqxhK9_|Mw(v@4_U&1<v}12By#C)OYOJVu`@D?Lr;B7Vl_ATVjP-+sQ~8z
zx~ur>ue<SNOa4E9{*-W?kqJC5Q<asIn{lJ2rK2;-9|6+)!`$;`cgFBv6g_uQzGKbn
ze0EQdaAQU0qPf_~@_ZPH7;9EkTkD^zo2Bv9!onif?CurvAm%6ayj^Z_g65Ouj&{28
zK0CjFL}PIg@R_yW?#w;Kp-TiU>g=N+S8q#YjyPz3^+U>WbGHl)4P^?d>H^R8IaVVA
z8SjbK_3PJXd=|q1MG-#?w?LXAAe#IVQ3(l?C>ZGggH9ZNrl4^B#tqg-_Er{P8K0f+
z9M1|y?(BgYlk5Q8AdH&1{v5Nmwr&e!wF7LtB`T9x@(UNvO+8&pq+hwBkYHC4GYkg9
zL5PQo5s4D631a~7c@3R@%n)wXD|+qPwFhPuk5IJMOYJ{^nR%B!1vpnz)$S{vOiXcr
z<!gch-(v)NHM%qHJ=@aSdVGAGFno><Y;0R|b5>DBzn=V^OH(2f?n_qHNaYn)jn^?U
zF^$!DI^win0QMna8wjuK?^fMT4lqF7AaU&M>};&ij1VLWkgemr<u0ZNrZ&2C0QoAl
zQH|j<(AaPsEqF6omdmBegRk|*Cd`J~3g42Ek+CRlX8Xo03Lz2CR(cMPk6mW#r`@_*
z77MO*Pxcq$a(XSRaW*P1Yo0PjwifD#89r^6KXI}T6dLlLE!9eT{9`$d%heStfYN@A
znJiaTRpn{<NAajTNpdK&9tgd<jT`S2ZEb8$_qv2u@G_FS`v9xE#;Ft!P6h~(cdUGB
zN=m)c#6z6jJ}|8w0GR!Ki~D-nGXA1~?bg`X=!cxiWWRA^aeI4vZOvYH7MR4!S?K1*
zhQo|TEY7mp&DPX3LG?w>5a1x+6tbAgB+fOG1BM$k)!WIbAU{7Js41yV4-Aw5VdsFS
z#&o<Xp9DaJh2}C&6?o)5qoS$M?kNR<|KK5aG6@A>y24zO|K(S}ZS_v}d#1`4WnnQb
z(ub*V6mhbxt?jh;{Sj`}j=4J(dc6LQj9laE{QP|E1|4W`7@&8eqM`sR)Yy<e+aKx@
zLOInd15<;yORctmr#`gpY<N$uxnTrEXW)q!fJd4efKj)9a3GHIIE}M}z8T4j@l`6a
zL=gcc3~hw7p4)rdwb&XjjKjH+sHLeXhI*+*#iedpoFgq&aZvWP)|-Z(xYS!h?9Nou
z<3BxXqsKaVB>+)zzDlEBtEqMNr#)+n0H@YMvNeA0t){~A^s`|!&y|(!j%Jh<6>stz
zc4x-_{K<UdhPB);zy&TdEj*&i93MCK_xHzTXag~a+uao*iAu`I)M@;DxY7fJbPh1Z
zJc`L$OgC@ZT4)0S0JO_8x2o!Kxv4XeX%irkK<4Qo1@spE+`3u{OnBevs-B)6FeAXl
z@Q8>wu3VKLcX{yb?JC;Z+Bo^t=H_Pg7(QF#f$hY%0sA_CLXofN`uh5pFJHLTqJ?vL
zxwvkHwWceECJbk!r@u5e@A1B~Jyorj{+tSjT1Gf2QC?$YG^}tkiF&Q1q!bVk09dJ-
z8^C`9K?KZLCj0H%n5``*VhjyXKE&g2L-vnWG!r6dfApY_>&T^fcv^vg3$U6+`i}ue
zfR)`?w>$0D($doYz6b98ZC2L&0sC(3LfGfe_jEJc5p-KXS8*a%k63`#bhl}!ogZJ1
zc<#!{S!C({gmnV79w7IZvW;h{$3a`4O-#^kZOPeEQBjo&0Hc~X4Al03Q~{Jux)bl4
zlx7|^gn#)W0w$+3>O7561x7?grQOYn6^K-$$5K+0K!Bg>G2XnnytrsR(+$)#yX7w_
zd|q0z!JP>La{var62;d&YC58LQ3sQbE4UKLPzr>|R^n9*k@=JWC-0Pp>WVKn1+fPz
zw5FO*&g|a+94-#Xt`*~5eD^}VY)<mQ@1fs*|MbERo_|)J1&a%A8+Im~7sqs@_%9;p
z73{4AiSBbtBRk^^$(;lF&!1pc9<WAxaz+E?v&fipBUpyo%k-Zb?LU<o|F(+f9l_4o
z+Np9cUc6Y*78LLq930%9MY$5E>MyJVl8TIstYW)~B|Re}IVq_-^_}s~Et*fo#lisW
znn}E-O-oAyO1Pr19p}cV=;+~U_tn*7CHj1hc<0uh9%rCFdy`MU*prjd8-QS-w9%6w
z)-@Vt{0vBy-k3?-O~Ha!K=F7a-$;UVH7kf$PL2p5F_55<g{&o?09-n+L^Ga2R6hdv
zF0b(Hne(*b=oMnBBB1s;|D759hsHwm{MOHF`b#_Lo`xe|a4~{kz+39~hy$<=fTyMD
z(9Sg>O4ZmK0x%gpK*nhR67Z<G@dtsfH*ekmWTwL956hK4bV5psi>oVqT+?iZPsY`8
z-b4>+0Uufi95ttD6~@2I?3p@jrNG`y)SZ7X;QmYD_3zUc+R0KQqp}Rblm5IF1rx<s
zzfYAg6CwX~=IqXnMi|_p`fcAtty{^vhwKq4Y%yN$*hvz)46B=73sVFu4;|eLO7d_f
zU>z%xzk1y>NQ)0$2m6|6X`;$Wgb`M+<QnS$Cj<ACh`VdneXf=#C$mi0+8De9OrkV_
zkJs|d5uYG`kQ`AVbX#Wug+~QiO}lOqrF{7_R4$S39*mlQv5WTGoNE<>u~9{x@kP5y
z_=lSq>WN}+qpOYy$!#v#iSU@Dlw0tT$7udt=bn5rfrlA=XXQi!CTC%a@8TgJuKD;B
z&HZeux-fr?=)1NNSh;tTXqF#PZ0<)SJN?{Y+~X_9nDs|n9SWg5s`qb(stYNF5asGj
zYA5W<v{R4g8V*+;D7n_fBm}JIkm1ex<dQjpRix#Xvluf-b%`wA_I@WySgeN5jJ4I$
z5)v%tEc<G8dhG_TR*b!rLk+qy+)eF{uDn%ER8mvh`PxV;Em)uDzY)ODc&i+$(k$p6
zp*S;*e}i7JKH%xnPEK&MdWpE2im6FV@L0)b#~&BfSu-fuq$Hy22l`Wmcym4#fhWp#
zmHnflD~l#Y5tCxf6$g;+W$=l_#_qGv^SOFqm9Q%g57%<2zr2};eRG*{8!RbngFXe9
zmMM=2&VG`UWL0cGA0{E#ba_VuScv=2?Vm?(p(PZNF@z#Hl9y`u^eJR<n>yl8W*d9_
zX#M;ShN`rk37xeHZXo%&6|BgBL!*9#lu^k4)_J>BuWu6T8D}T9;~!`}2q;NQoaq3B
z+GWia*zXq}MkhdA38#^?M$@Av_4+n#OvHm>)13(P*NYhsMtLb({bx~I(=HviKM1hi
zkT?)}0b8*>)60gi9}E9#%mAxT?;5VNMD?zvJOnHDTc((xjAv(Wd@FUTIyvi7ln#LP
z9Y*KANu)URJd_&*wi}AfR~jKLj`_OpIC`UuYJk)$YGV?MmspY5-Kw~2@y6A0uyyQg
zLlgD=ewW+idpbUY!?|Av_x12b5Q&2Q#`iQPvCG|2H}YfryvJaubYxj^7(B{7L$XoL
zUmN!qNEz-^?sI}|qA3+%y<6FHxRXQuQ`|JmNtWc*aSOQ>a~<xg5o-%{5jrp>YVTQ^
z1g=o;lYa*mVa}KAYg-t%?p1HyS>)AAsBm2p2rkYauJHUJs$8_@mcr+V+(1bd==qid
zpn3Y$-!40g^`nhlj`$kBlwRQ$jZ#c<a%P#e($CfK6gvCa>?$?N4unBSjYN1#mC&k;
z0cIAqJNDtiumYu|sI2+Hj)#$~t(@8&-eeiC44076lt}u#1!uI86l_UjMK4>eeCEWn
zH@`lTULiBb&R#|)i95}^Xwv$*jxz;NsA75m>|D#-EH%k4T!XXAE)L|G-wdN~%6%0A
zdBYq2k=ZRJZ*N^@Gv^1M4Y21^Q;NI^`#xm`9gjt^XFXv`ggwK_@tX=u^B}qERi754
z1P3}-({hJR%!YOSEz@V#wdI|<f)(mbB`v|Ta?Z<q!St>BWg1M{bYblj=&YACRvf`c
z=W8-nS$LKE*-MYQxYWL8#5F!J>hA79No!J$TK#Rg6wzOKrj%*OYOlS@SIH-JOZT?S
z!jcVahkhB&vP$F`MR04SLI-<>7&oKHCB=Bed(eHW^B}j~UC;2)aETQe1A}yEj#5?@
zVs_tk_Sfi1<xC1G9L2W^fSHIKPj6$}n89@gN`n`(>FXjT_uYE+Zg+_VR)?n^%x$qd
zJB+2K9dF@g6uIZ8Y?k7hSpe%}|C+R%*<a~zutot473QLUTDg2qZU!E@j2|8P$w~$x
z68LTwP89iGYD?q~@pdrH7)v<_Z5^(>q3}7Hb4YpH_h|DAKKW)vINkCDNbAcN&8uh7
z_68xZRO0FhN0^zGRP^gau}%w@3ia(6ONSe4%qOeCE)N{bj;a@e`lKFz1<<Us1;l6)
zHj=bz?R(KZB~p`U9k9T!P@0+-L0s)3qfO9MWi8*`bW!Ab^;yPyExYDr%wATPVcwb>
zc4xhqSNqESAOZH-h0ao^fb3ick`@?xJkgds2unKM^la8!%>b1GAB%kd==4Z#NI%2@
z9+vv(D5JHKv3Hzp_xHV7`Px(D3O>Yqvt(m6%{H2F6xBj9RAwx7dUU(oMBCt||9bM5
zC^_G0>#j^!ZCgsoH`q)%?TdlZ*|8t)TANV*@Fq8I|MV%U()G}5AX1)`uvFr4+*$jT
zwULM9JBo>18nBvci{hc9ek|@hlXBu#KQ1Us0QEx7!AM<i<`aj!-pQfwY(yB$r@i?W
zIBE@JrjIYnb=&K>WX^&tL7aYjW+(BjU{#PRO}geVs=2TpU2K2)TO<{3a&(;B;Z9WR
za{D6AS$%jc)z&6A*`^X2stF4)M<nraQ-G!Z!q@NYxc29vcZB4fy}YKwrQO1DFD}Mt
zm!vblef#Eoi7vjr!4?=WoynH-wJOv<JUBbAj$iBzZZcj-{&%r8rk;T#_+WCcp4Reg
z;?<vCoV2`SR@t;INAP4$y!UI3sb=qHdayt0^Xv`I$Dz6Fl@S;@FlTLa(RBJyf3>J(
zm?dgG@}QNFT6CTJ28<Li*?lu7a(nv-a?mN7D{hjpwKNS9+`t%Sy^_^drA*_q$cOy!
zue{5^*j+s?$b46e6u4K^n`T@Nxowdu<64Z%o{iYtYxG(_q9{&U{(G_I6lhvObNJ9z
zX`3a1c4Z2b&4uI?H#Q{li+nqWyB86ch4^=u<<XEjmYqW;MpoKSAsGtdYc~%hnG`5r
zS>~8vz6VlR+j9K*06R#bwV!NP@AP%(Br+p;h|DD6rt9hHwA<{0y1lp;7dntnht%u{
z>}uB(+Dj-1=J4sdl)zrSuosE)+Y%>p&DIIWYwgC$baY|D7h(9(Z{1WU2ZHQ6LwY|`
zWojFE@Y`S+dIta|`?jPvrm9qZJ$v<Nb5L6ZVkcUOBltV`xng#WukUYU0*auTnvhZi
z99k@@&e*Mu%on4z3Et#YUQwo%+DWRBAL(1$ajQ2v4fryO;lF!ZygGaq$6h_Fmp$1r
zz2fYlpPru+!WFzwc@E-YCgJBx>(Xf5U;%x9NWg4~ez`^JgZN#gKvG3$uU&ho059e=
zv+Fw;zK%#p@b!4f4L?3>)Aq4hh!%^(mu^o+_S&AIX;bA~N7q~o)G7~8^%Dxmy1$9M
z!LDuCnuf=Vx=!)O<RKB<oj@id?hQUQ=fP_AxaEoX4b5HQPP}^Nw4L<jY$G~P7NVUl
zl6Z74RK)G<3F*r1w!T!+I@^z*QrXoWc9tgjdT0^G7?b*KB}eFwt;ckiCz0~jGX)Vf
z2KPq-4stzAVAjwZae1<>#1lj9__EXIUki<TG+KX}XsV5XW!xK=er=FFMa>L%0cjQi
zc^*eMs(h<7DIvhhspob1iDr^00g<NJrO#4FOqnk^dgD#UKV<yH$aZxaq;<M^#>dS<
zn<C?`fKA9^f%*xn`f?stSE`pDoF+k9_`FZ-59*kk(MNHZy))0`B%#0aBMJA%xM!F5
zfmWn#KD(6A#fRH$=!!+3GXLaobF*0QYYLzD-ARO&SExp8`Gd5Fi4K%Ak6{=^49qoX
zNhhPPDkWy5Dc<MJD%c-B<y|zM?il9>9YrF~n3-=BZB5ol7#po*xg<Lt{RZZ{76z-m
zn>G?(FUD3H&lk+GeLAXH3+Jwlb{;SB4=oL!=uv*_F<|HIB+XEiJ6L3P!p9y--}F}?
zuIWKuN=B`LzMY7{=kCwei4x$NEIf_<Ou6HS*DSASNSm9M5!0kumrv#`E*U3dZz~2^
zY}KKE;TkGXeY_{)+f(H)RQ)-E*<gn`Shx{buBs%*M$Ujza(&X0)jK6(DEjwwy>hKu
zQ3RHi)y4V555C%v))VdH>}5e~_3ZTM0g4njq?7k*b^Ae=<b=ydx19hqH$X#c%VCSD
zymy1O%SYk$x*6@wOn1OeaVap6{OgB??uV+OoPV=_+mv-+BiMJqj{9%7b^g05_^-v-
z|9rKWN|Lzdiee9sNhL;gNb_9ok<3E`sQ29q?6Z@F+&0vbfAaL1FuC}vZ!;&Dv0BbB
zZIY@qs^L@2*L$lO4xg(%4jraMA6>m;6pedBapfAOejO`_@;i(3{v;qxAViae(y^I~
z-?bmKWWKxHi5;4`x^bvBR?oRYx?Kdokfh+@AN<7(`kyB`bjD?#xXfP_&Qf^hv*wOJ
zPTrNG1Z`kzMW-#+(`lq)GtrVQyavBI{M1~Uz*ceC-)z1w5PLUR1zGAcS&OXo1n=)g
z@GhjjGP0nwo*KZm(OvZ`?<-zMS!((li$32X?KO{CVx|K7+vHQVq7en4cCtpA@psP<
z1Gr<ueXJ}gl(@VobcV-|`!;&ctQ%>21}OC_^YYWA$T({-y}>lR$X~SK@EBl<co`~j
z<<&Rj{%YBa*?fxL6w)gUzuru|zglX~R#*v>gm&uNUk7c}yhsMIN^AU>cns9BGX$?7
z2PQ6?Ap7mg8J@LArrcgt8uoYJ+F>N1HfM05`4c^-^31$)j<Yx>=loY06#DfKbDdL{
z=apX3m`mCB^Q@*=^U}h>QW;~>AAUk&WT5xoUuf$cr3HPAL?(tHTD7qA>jjK%pVV{i
zyU!Fj^FDO-Oepbt9<h=&l5PRPF@rxUvMLU=LDj`$Ss!`35fZ%8kNY9i;vZR`f7p_1
z^3s$~B537b*G;YugRDuz<jS&Ae@|9!SB7wgrPsiZbY=QTrAuCY*owNLsi&Vmd9CJ$
znYjB7mzhdB-+4$!I!&DK38TI|HuTx6yo~n(6A9|L^ZR$=yp}<0nZszNJt9k*MbTgF
z6Md8#d;QlZ?}1Hk;&w(QYKp-7FU3MP=McEAO2(9WH=K);NA#EqBDFwgVpfZiW9v~u
z5}kK$-B%wq1xn!c6l7&z2F&5PmwE?HH6x4n`l3<q1U1L=<K*DahCh^E@P}M#SFLoF
z|2<Wcr$ouYMc5GQnnE&XkCpjg5fCKBah%{tm1dg-Hk>7WfI%czhO|k}AjsBv*-5Lk
zatzqPFu##LW?`AJvDZ<W%q!&LQyDaQs%((QMC^uCGn9YG(8IivKdV1C`!>g1rJ=Zu
z1@Qw(3C!FUSB{KIpii%{mhbnKxuttlnho!k?Xc5R<v#X>?cB08z18-@g?09%Cr_1y
z$t&)5DII-LPot>}ay{~f<tTsHv&gIE)tVM*kFC<>>&mj<DF5y4jGld4G&3{9uIhL)
z7U*i$smHiIS*#eQmp?W&OvS+qHb-*TKgBsDM2C@gd6}{--X71xu2CURvC}n)Hl`&F
zd}Nod0NcV7zfXnwO<4$g1<VE2kVYPbo$RZz4Q4G4@ZIG@R%v(ilM<0NSXc8spAHit
z+XmLqY%H~dp0#<{V8__DVamR8#NBfiVBll7yA+7D$U}(qKiE=JI)RIO;^_0lKlt};
zp{uq(N!9Phxx~g7m&6#EobdAd)=52i&)LCvl*raFbSp}s@1?4sVTR4#$H%ig2BC7@
zjo;$u_`R2}8}QvkXj96$x8<5*B6E$jxENhuR)79n>&DfZqxvgCjZ$)ZW576-1nmkw
zMWsG!wU9ipKR90N%GqXe)vx{>-J;uBrJ}GF*;F*G;vr~R9#r}4Ah69sk{vr%^|cas
z#i^XCvS_;Z9*a(HCK?4>*^Fv}TSN@{$G}Yo+C&nsyaV3od_s)zvYxFF72)YS);Sv8
z9IQ!?Dz&?M=e!;Vyav)@`QN{;@PG4G>#01~+Etz#RAPYX0|KckYCSJfunPKr04}`Z
AJpcdz

literal 7110
zcmb_hc{tQx+n=##9U^2KN(m8SP__&rOR{AzW6c&~Xe^T~UlEBKvW1X6gk)<O#MotD
z#?siA>^tvJzn=HGe$Vq>&viZTA9J1S+~=InxzByypZnb3GhsKi)##{CQ-eStI`!-D
zTObe#00M#EQIQfQT!z-&M8W<|O??$&7YqiIkdTm)l9G{;k&}~CP*6}(Qc_V-oj7rV
znwpx1hK81wmX3~&o}Qk8fdK-6FfuYSF)^Jyd6Jo#nT3Vr)TvXftgLKoZ0zjpr%#{e
z;Nalo<mBSw;^yY&;o*Top}f4je0+Ro&YU@W_AEa?|G9JL&YwRoARutz!UaJ=K_MX_
zVPWBm7cYv4h=_`cUb=KiOiWB%TwFpz;_~Iol9G~AQc^G&Oj=r6Mn*<fR`$x3D_5^x
zm6MZ`mzP&iP*7A<R8ms9cI}$7vN9YFS5Z+>RaI3}Q@eisy1Kf$hK9zC8#gpHHMO*~
zw6(Q$baZasys4|Jd+XLMJv}{reSHH114BbYBO{~Rw{II88=IJzn3|fJnVFfJn_E~|
zSXx@%xpU|4-Mdy+RtN;b+S=O2#>Upx*3Qn(-rnB9!QtM$dybBdPEJnF&d#o`uJ`ZX
zcXM-dcX#*j@Obdxfv2aZmzNh3iS+jN_VMv~`0ybLh4S_F_4D)d_xBG72tcFJj~+b=
z3=Dky_;FBBP;hW?NJvO%Xea;xVPRoUo;(Q;508k5c>45dWMpJiRMfL)&!VHFpFe*d
z6B82~8~ftLi<d87VlbGvxVZTE_*bu9B_t#yCMG5&B_$^(r=+C3e*HQ%H8m|QEj>N`
z&6_tF85x<GnORv`Z{NPn&d$!s$;r*l&CAQn&(AL?C@3r}eE05MQBhHGadAmW$@};3
zOG``3%F4>i%RhYhP*G9w@#DwJ%F3#$s_N?Mnwpy0+S*T_K7IcD`OB9tU%!6EVzG5~
zb@lc24Gj(7zI|(KY;0<3!r^eu&CM+>Ev>DsZEbDs?d=^M9i5$>U0q$@zkl!U?(XU7
z>Fw?9>+9?9?;jW#7#tiN8X6iN9v&GP!Q=6xqoZSEW8>rF6B83Ze*7R12$PeOQ&UsZ
z)6+9EGqba^b8~a^^YaS}3yX`3OG`^XfBsxvUS3&QSzTRSTU%RSU*Fi+*xcOQ+S=OQ
z-rm{S+1=gU+uPgU-#<7wI6OSG=n8cs65z!B>&6}+5DnAO58OHIzXt-bxvImJ^t}@o
zhEI6w+0-;FlNu`77umxcO-){B#J+S9W_f~sFF;PlB0Mgo$npZ*kJZ(ORJ31t%_kWC
z!Hi3lk(rIlCX;03k=0lrhk!KcnJ;LXo9`4I9LD?>8-4IGQV!R&pTimz8kcsju3fhC
z*_}*e2%_i|pyegGP?HS0mI-2vrh+hr{r@z9v{p~kIQ3a!LNu)c7dZ@5u3AjBzP%zF
zB3vpFS%+~G?2TH(1)j3<G>i)AHg3{Js#fdGRCc#YpR%ma28lg$MB5h)XaO)G$BYOB
zBb373kfDVD3sM;=R$EUOaJ;>~7TfM3)SYj~Q*+L>LOgIk5AK9B|9X(j>gD+v<5c@x
zdM3v!yo?aEq415ei{TQNJnsE9j)oP1E{tw(-U5@<8tbaCa3~Y1m?uO9K8$Rtijh)U
zO~h)p;Lt)_P}>J{J^6H@-g#Q9rn^E#Q&R8sCb<ng>ls`!gzwJ#^NZ}d(^=V96;F=7
zz_9+fpq|5>VZf?Vk){{tPRXOGa7Q^)+@K`&DZ+<2l&zCb2!27V?TeLwUt3Nb&b7se
z!n_^X37xgPmXckww&JjQ?K5R`meQW)%umvU)luFBUe9o*Lk`nIcrj^<i9s>%X*i)M
zWwtEh;Ufc!g2bW1%eyd~(W0%5Mvm0WZO{vxtbCONCoWwYo{?Cc|3LEBR#UNI5$m_)
zzaz!D91fUvG+QL-KLj$Q&A53|h2G}M7D(dhZv$7qx!Qc&*^w~4tHFzKwTGA{>LN$v
zvvoeCapDWlUXs%AdAgt{dk+!M@GhipOMt%UA_7&Xn|WlogUCbr_`)S?+0H%wwfI+d
zocGBk;a*k_x;HZFT<F@ZYyv;>8_KMeiN_ToxA2VJk||PAmV0%Q=YX(ta;~p<U;Tny
zM#637n_}y3?j-X!1tGg@juh_yNrapwdg3jw|4QHadCk<7V(Pbn`^xIlEmA_g@60uB
zvv4?sNUQtYGPfa3+eX(zgz_TYX?5|dJ_OzRms-0T_{Ilswr@;JNVZ9jQr`o3gi^~R
zt5#Nqo#Jk-(Fy544=I;2pODwsO>Sw1q!<m*-s{$s=_IrI3bq9@-Grl#vC6+qB8Sr+
zNT)Bo*^kZ-IN_H$$xdmeL>FB>ypYq*9bhwGGSsfq*MFfgwZb*mG-{YTc|VPPyrSc7
z623`)QZm?H{u6D-<w$~VB=0P$IDq~hhu;$8s`da`%dziiKLpaIMQi8n1{DQ2ovBGy
zyH5^i?Ua_TY<Y;al<z#`{>%cEavpr@JZN^9I<cSV{Pf`Lq%0Ci{ptUQr2i=jd#>Pw
zbI$s@W}vj9^}NnDqEsB`Cl>>yhp?kiJudl_Kf9jzu|i4sh{*$@P6nW_^C^<SnZSI|
z=Ts05DE+*`t-03ImTdx5q*w}||3*{kPV&$7h`>vVPGrj$YlK1YCCYzZ1dfI)omei{
zNZxX_L|=b>C+PM`$(us1vbA&o8MP|Vz0Em<(%W)YJPY}2S7v4iKK?W65&k=GP0!;(
zZW5qIQ~-Lax>Q3bhA4YUKbcDd?${)!#2!w3`M8>33v8~~h(UMc6(6u$1$%!Apgwfe
zyoR0@$J%99<#ewClv&)D!D(|4qo2d!Al++dD&Q;$T=6X9(T0K4R-L~2GU)4tuLI+1
zZqh(boYIF>uBymDJ82*b%4W3@Mz$z@#aMReKxkY8Pb2GExtt)hiIzAv-{9&`3RqA<
zh(|oe#AZTH?l9YzlEasX-se?)v1-ExeJJ{u0l%?fbHcfkL+nW5J2D}+<zN}XOwX?l
zM&VzcfXqy>jmM3d3e5?YvqY_aak|P3?Q9dn0{5JCQ%1+)ug}MMdsK@r7oSsnlY#*H
z9ZTynIdybYEQDN=^Inw1;3<wF8B=u4#07<1Fs!y`vW=$P_%_vpz2B>eEFt_kAU3w6
zm6JQgEEzah56WMpbA=Wwhg=z?VO<?W3rIX}JCiIu(eLn$4ze%W00~V*>{be4=yv*P
zN}`#eJC>2Wf_8mqbp`Gu`7c$Iy-6mf^gixZi)6q7h8Gy~oW(PzH#;6+4x2G|CFEqp
zkR<R!90*G)JOx6aV)9{m5BtwZJck?ujX+pWTVTuJ)%Mk32w2xbwZ>7t9C>Be*Eyz}
zXvY=ziP}fRfNP>TGHfI-cmMg~5fFeDAgaY09D=nsb_Kx|V=<3HCyO5p&ON5EDkL@G
z0ChZON8d(9Ev$bgS)Q9_G^r*^#u8W>AgG^&aP;)&e-Y1r`R%V*0ChM66(_fDoY(L}
zhLS=RA8=UBcN&<m*HnH5L!ciS++x^<kw|IKpWq=mZDp&PY)F$0S<?BD9zvRiB}y;`
zzCNY;z@}U)Mh>iN^-t0EpD6Kn-u+$AE=cWkJ$|!h7c(Mv!rxF)DR29|rXw{!z@tq_
z$nyO$95ZG|&Wv@IJLi}IPQ$a@gy5E@diPE#u2ud+g;)x)O!Q_D;SH$cCR%{b>b6W_
z;qMLupXL0boy-Nr%go2D)Ro1~ZCB26OE;b;2SP~TGY|x^OD{MULjqR@r#%41bA7Ky
z>QF=I$p2#h2_UB2;{MJ}V)Xl*03^B)+`_pYi^tt~o>zyZP=)cD^{GT@=A#g>1?+mF
zqBL6=Ebs39UeU%H!XhI3sgJd&o`L+x-)QN{eL-jA^43WU16JoevSXRIi;|3K^}C)P
z?rsk~69L}Fk_4w;BU_3i<xX}7w@pzI>@}su1@5NH(sOh{(By*e10cXZ0nN4^zB6_^
zC|`~*r&Pr+aGgRHlFPSMIZ;XpuhmkVf2XtmuJ&=hTygKca4nf0j%md_&QFEwda+Rd
zXN~wie=_K4bwx9F3D1s*zFy}A-9Ox0r%<dqi|yT6g_YHEL7u)*h}v4{;@9!Li^BY7
zAIz8mz=U{nSjtlYs^S|d+<<}~DR7X2{vf6>9Gfq<H#)<ixOPZ}P4=i6&a!o6Q1ITN
zfb1)P#X5TsqJ=YEu)!(Er~XYbCq2#`&%45!R93FmmE}smoKhiK^1Vb)f*ZOMKw{-e
zc%n_L*2^|Nb-~cn!DLvus|l6I=)RL#t~rsiC;=!*S```gq;f(QW>aDMd+HGbnILM~
zSDssJ!c+6*BIi&3CLEg0ysgt%%hVJlOidTo@tc(3qPbHnRz9y<%$uAJ9@7xvn#r&&
zMsyJ$6n2yoD!&gSc8@(egO}t%IrltQZsnpTlDQDQ6@3OheRu-czM4h<%fYvGb(gm}
z{J7m`k5@?g+spUXXDfplA%t$W@l@$7dVY}GD$2wal!ot}UIMBM5*NRxl0h?nZjNx!
z0RBV+c7KPpa%q6lK1>FX!fV@VQ)<9iw7OzM)Tj!Vl>hI;D$o{eIW#=;$<j`O{gSOI
zFt+9J<isS(9*3hW*0q~qH*xDi5-l-0ZAr#E<FfFHo(spCW<lxO#+TAidKranryJLn
zUI}rtFSk=n5f70C-SAOC@0A}7?CsX+q1i5NZ-!@q#O}E7jbErdN@}>}@T_@?qijYh
znp{l0Y167`tGt^mj?BKY#Hs_f`h?}}`60XA`i5A>LkRo_aNoVCjeuKO)yG}vhU2hQ
zOKV$u8OJZr6?$tQ8&s5TTynPXg*MNTbw%l{fGvXA5$K;}bs4~9<Fm{{Tn#}Tp$?EU
zYrhZcZ7y+ix&J-M3_?%}O?aCve#?}rCg8Mk-gC_;gWQrr9qxQk#~cI(4Ey@6*#9B?
zuNvWRF#l&diw?Nv8&r$b216+{4^AxC7hWuoAqMWRoN1{n+(dY>Gw(#y&W}!azGhh@
zB}wbJ79n1xf~E?yT#(^FHpF(40#r8Vp^F9$|5+ced?NoI(4=)QqNm@AeJwZaF>n#$
zw)=I{MDnWYJBII&&CwlM#+_CSm0#8ui}-8M5NpSbiD-H<X!A=#QR7%3YO4Fu7vcBy
z$@-bnmCT1OCiEm&G;VYiINy5tqg&0bx%E2k_~>Ym*r+?L+c%lA+DiBS*D3k1(NuN_
zfxwpL9zqwlHqkM2CJzQWaz6z7Q+D0;iEa<akvnZb<KUMb=%VVzUws1GIGtPSn2OVl
z@8Q2^{HUDtK$VxY^-3Aj9eH*ScIkKNvPWZeF+i}oSQJWOazO5OYw3}R5y<K_!4h&l
zk6@HfM-=}u<nQ_a$ozjQx_|4fL-Ysni#oYPjC%**<&5CHd;FwUuZeN{zw5z&y&4XE
zIsB7*>0NKDU^Xb43Tu!ZQ|c1X<7oFnAAOyC{7#LpXXNqNzl@B_yoSSPM@IbUi1i_u
zM+;C3-!d?k_0Q@_N3%_Lgh*es%QIwxQar@+YLDU)Kw|Bre{Sz%JS<cs!5xhpLD*re
z(bs9mW7il>jJ5E=60grtEN`AE4WSUlHlEyGDcU5Ym^U##j$04w!J(xWSPrQ6AMeFz
zid#YI42($u(!wc*4z3zZz{?910G#*KRnc>sNqZ%lVt-Y0RFE*`=GqBxzQc-+6!D<a
z*hm^9hm1yr{x>A>Sx9gA@Q)OyTa1skc4)bw1CQg*wm#bo2&_T!krasr>=}~-8_<w2
ziSMQ(>gQP)fqmgZYN%AOfY;Lbr8hCuf7poq6cX2jzmVOquu|~CZiNwBIJK(LwlavG
zK8LLy4{B9#sIZ8rLA;ymZMw2eH6Dns&3SPsO04#VgFfU1rL=z$749V+&oF%#m-2cL
z+#v}V^uBT6f4MS<)7YMj*zonU=CWh}%B~1<D^iF~o~MGC2YFXL!|$cU@0HOPt=_Ad
zH~-oj-XYYv<rDCb#8O_?e6Ra$=lYx|OjLPqj_QEwc|}|bRB??S+xBr~Ro)hAIdY%J
zl73)vXC0Fu1{__~4^E6*yUDVy+Gx-p7A$Jua!CL$UfZ81UtWZ^&U|uhxUCd1-`G0`
z*<B%&Ahz7G628WK(8#&Z$hIFa(f1RuuIE_yj8)7)Bd2m+azf{|6Y9z4j@6&fVOMOf
z=EnC5PvxFBQPSM&n=}B!D<O4@JSTvts3{z;A>=F4cz%&;{9Rj}(FDdpox65>&%VYC
zDzfh4=4mH9hC}AP1#+vJgE27a5NX@3o~DmY&UF)cu)ZMg%@~?eS>u9()GL;L#Pk@_
z_sTMJ+Y2hc55aTGhl}c6EL_?2@m=qmifAus{#~cvu_zHV%GJw$TAk7~$uY6T2MZPz
zFPy+dF8ju}M~foFx5RJFbJww&*A>UGWtyXkP^`G&^liQyL)uRXIPBS)2(6QqcBb>m
z7rk^UUy8;IEE7DFuxE>fz1p6RhKq>#_RA{`skq_ODovWrQ_NJ~6U-x7kutJh-I2MK
ziNcHS*C=bMHad|NG15h0<Fw|vcmihv(pW)RJ#0J|=$N?J_RJsCg*Dec6{9{YJSn_b
z?EfOB&JPkZ^KsxLq-0yD8b**-kiMqsL0iNppZ>vG-*8Aow3&k3`5%d}dTJ^#P#Tu3
z2ET5B2CkC<0Q{euVqW?sQ)p3@h`5@N{dv##B0l13bG^Eg_T*ut){#9eJ++ffT2k0b
z1{|mJgj0OFSVbWkHK`Y$zCR@rpI(%3&pL8rQAk{^gdO^k^XNOvoE~CB0f<ln;-qj4
zDLfvWh6ATHTPs&>0?i%rD8otnJjGdi4;YSfVOmaH4WXiCVf8~h7t_euVq45>sLr+U
z#W?$GW71GR&8^UwD>4=7NebsZrA@~y4I%&bh3uOjFzCToZwNb>pJ<k<MoM^D-yF1{
zB+!+M5OPfY-2&blF?|W$8}hD>=)f}D&^7e+tFNfWcyeLy7noBvRL~X4A3UDD>%rmY
z^<f=Pd^<+4N*MOY&+OHv3~(%Ny$Fzzv5FW>UkE!`{NX%tIv*qUDlRQ=>^ryp1?S7g
zy-KuG1D=<j4MK3@Y=f^lr%R9&;$AuZjV%=s-CWP5%v5L*j*0#V?78&(l%_t^{vQA~
zQ~aP0u<=I|(h52^6*0t%i>Lj(Cdm*9gCzuI)$??_x&!yK(hnbaPt%sG=_xoZ*1qp}
z$etc{q3Lwwe(b^1Rmj5Ab91u5m{}z~g*%6pL)gw0?d;+$C%<M{!<@W=SN_%S$NF#@
zxNj3os)KPXmAOaDXY^t(57i|o?DYEpUQ0Re=RPgv-st_T=%Q>Uud!dv<X+Tam!+EF
zV&mi>g-}b7*;k%hpeyI!s2uR=UyRD!aB-h7Wl}w<7b{tTsh41d4sd7<c#eLp@ZzHV
zA%)fHHg>+@vev(y%#dQ_>iq4Nj-p4_A?$heSn>fE!k{H-KfllVJvokTce(gu$h!D3
z(9m#dYJTiZyMUqE>JDpjO0h&oqC$%Sy}rYlJLCNkFH)FO@>}zD5mV_fv$zh+f%_bM
zwO6kU&bW%cirDZUSj3;(5G%w;{PK?3*2#_$`{IJfY^|kpcc7aoYD4i!m&#riz2v~!
zMP10$E_#v&@UdFJ?zn!y41_ZD&28A{z5m*+LfJA$h{)8@?fGFAXpg+C-mAnzL>QIc
z!%_1wLiva#&41UVbv!wZe>*)n^x4v;F;N3b0jSc5`L$bK-Wl|oqNjkXfzv*R2+myD
zk&DKi;tyFLYVh(xpdLm=HVwXU9f^?}owW#P3*Q$#w{a)Mrhdrn8r~`JgDn4e86wZ$
z7mke?KXZ=?^7Pb~ntS{BDTcBk^VLyry0n!qb#^Q3`KZ{^?%sx^t;}xYqK4(47uUN&
zW)lp6-ND7Tes?aLCYf(5CoFBN;fEMsfzvwl0cX*71D1VZVd<9vqeNSL=A7v!*33E}
z4tsk~si?s*IAXtHFI`?Aps5$zxDg(naPrXYk=OyFBpINp$D1i;2$QTq1<)6rMC_i1
zW{OL#8rL&1Jdv|id6Ub<Lfp4YQ(^oKVL805%=Jpq%Yd~Ou2gpDX|3S5#k(G@ZX2gU
z<L+^!^&uMfa|Yb~N@jGhO!bkLZe#5_POd$uy>!<_suoa(JumTbGu-uRfZ$=-4JBq2
z=aZc(ugvXGo@~gW&pO7FVfgdGZSOb5g9(3De4X%mJFm?GpDK*l9%X{manFBk9V4IQ
zNpsmL@G;Q%kzZS!V>$qr({E7NnT%|wJ$j$-sNRr&E7@Jz@{RMqh=P3uEonC}?`X(8
R@&5>rx{5Zu;M$$Q{{cISw#xtj

diff --git a/vignettes/examples/timeseries/timeseries_anomaly_detection/unnamed-chunk-12-1.png b/vignettes/examples/timeseries/timeseries_anomaly_detection/unnamed-chunk-12-1.png
index 23e6cc392c1c6ba3e07b2abbc422dc6eecae759e..8be126c92787d1720d1c54e0c595c8c7a7063f12 100644
GIT binary patch
literal 33046
zcmeFYRa9JEur<1Ig1cLQ;7;RC2<`!bI|K>tPSD`)7J??Y2ZsP5xVw9BX<YxE@0@?!
z=lgaaZjZrecFXRyYL(2YniZj{EQ5|hf&u^lx}2<(Ism|a0RWh4WLWT(q64^U@adJK
ztd0u+pyK@dgPC0Me*^$3Ku$_R!!!G6#lv{|>PhTu?&Hd;M|0~@4i;4)1F0HzQRMp9
z$b!hoSvvzwp>o}nMgJ$AibY-9zC{}YE9Rf7PE>EDzm~(pzJ9fxIJEMnY479X^X+!5
z0$W;p;M`W;uBgY&oiEP5$mvyX2o-W5mLzh1&1bS>QgP(hA>&R&!~dR$g1V!@H)3A@
z|Cj$Cl*Xmin)nNU&S+KVvv>U|EVx3(7}|k13aK1u++_4(-9cVY_gAF6pVE13XRX%$
zyjl+>G}P|!6_7J6K$b*Cui;zuG#gB2?Cb06ak{`|NrzdlpHgnUKPZ!VU{eOlv?^Nm
zi?gimdmmIlB`{l_t<VcU-PTrq9Rgk<BXUb3zlMYvbI|<1xx^&xe%hzud-md6iffXZ
zdxD()GHHDsA~=71J<$<r58>g7xn`w;&>3@J=KZNTUOVx4Rp3YZcY4Q>|M{vaz)fI&
zcDB)cAW^M3FL%YMpBy9H4BUDCM0mF_&~`j!#F`oZeU6yL0($x~riKBCpPl&Lg)Q`H
zx!CXAiTRYwcJ2%e1+<>eFY%28T$XuMjcVV!g9b=#TUqyfdo13PL<LFC%*^b#-Hc*G
z;<H<X@Iu+~P(ph)COV=_rdL-xK91#1bfCJzb{U_Dxi3L)I?W84r#9?Ypf8bg!uNyr
z1YJ!hdkQbeH~yMyZ4W6KdI8r^%*cSNRfpYe{kwzv{yq1mC)p-?2`kx9$=6kLOFQeu
zN?F1bW2MU~W0Lu*40t5RtwhqAnwsCyiI9s(&dD)C(&skl(1;H<nlBD!%8dN)cDl1S
zE~iui?)OW4yb(n1Hm>w62;hjF8)X77z1m=<c-voU?tWLXF~_TMKs!1+?#Y*iKIEP#
z2aS)Tgi4C$RSk7l=rzTX^05&_p-;uS!4Y7D%X~GZ7-siBZkT(9=Dm=e6}1~WP+<|B
zElqqkn-9)dfnHQNqz61a4Ash>{O#{-d7j<twA*1E6n>c9d%nJ$#2LUcvfrO9Qb=PX
z-%N6UErl6y*5$|UZ&pxl(I-Wuj1*|B5G7{okAVP)|D`jihZq0r<g4TX%E77#(FH=X
z!_P)fUs6#qVp<bD8{=kc&a{g}>3jAIm0kF>4T^ZS`s2rsZvDl>E5zXL?rfMG>acc3
zJ;xJ+&T~g`p!@Emoi9NB;;8c4nAY*KM*;KcF-(k{A6rPP8IFCWX0qLP<oT@+Z-w7;
zq$CO;ZV&mRo2BOo0Qr7HKg9eNyXZURu;2lp`_%bei0l8ZNmFxX`ulry#NeSFl0{oD
z;oM*(OpIMLn)jTPLhhfg4rbtds`|Ggq^RLKAX>wvanhA>86wqwYrRVpPfpF5u#29*
z?Zk{G?~mzjUd{V6OZ~!*cs<2{F#Ej1e0qZNTN|d(g-xVt<Mic5?!DrfX8R>nW>&}?
zJ`gKL3V~``ci~w{%oD!5?l`p5pH$8VCnQYEh7x+S79jR<l&oL^v;*Gyx%<Bbafl01
z5U$|PY~{U^l~%9Z)2AXvU(q$=lkq*MNniB6DQ%F>m!IUT3_ASzEXKAGZe&n0+em9S
zE>rkQvEILfdC%9I!5&HsKhph=eVXU=zx$ldhIhp7v->}{<FEGbZl@1bJzFlK2DB|o
z+Vtm{JX_se8a;AAM`cvj91XtrrMH&b{90HrZ%LApdHC>_KEsiJxyW?2-a3hB_^lzQ
z!}8A>qrFeIwt6+D?=BJ?`GsMX7KF#faz%Wmij|)H3aF6%I|Denqw%>93QSJT=%Tkq
zjh~&}7V`Al<Vmidj`|KenyaesXY%x~%KQjB+r<Tz(<m{_uM@@mo3`W5!<$ywo~ulD
z{T%E%+4tQpv*|tN^WcF%?zLq%T*~us*%U6(qr!)zwn@+D>Qy1Zq7cc~oonpKuoAAm
z2=2}L1W|p)9B5vBZMJi4Ndwb2bCd$kg*7#tcYoHIcdZ~y374YwJCh3RPwT;$yQ4Wv
zRY^#&Fu8k3#KYR01sz1!kr}zJdp>&CGeT{ZG@4m_{?Fl2!?b~4Gw#Y%I(&A?t{jd0
zSY8+PzbDR5z(NUSh?K9G#KEsyJEcq{Q6Bc`TwA+of<Pdy&6iZnyKqPt%if0-BAvzU
z?FLnIW0^d*L#eD!BAxTsf%x1cFf_CMin5T5qW8PQOj9`G9}Lgy`tBF2nkdO{`B6fY
zZG5hetQ{OEx8*=bA}uZbElURrK9LrNN{zGg`Th{Jqfx3VSb&tzKAZcaa_8#gn{=9l
zKU9YkPY*XGqv(g<nwtf<xszz+>K2yQ#7@_P7xthr4U9HcpGYbYq;G%UoOE6!sQ74C
zJA>N4S9IIQ?$24RMv9!HO%d>TPU(MZ5>NjsRMN6e0Q-Y|lU~V`gnHld`1rbeE0PyJ
zF#prZEI==el$10nSO5>@DUx;}c}MmBr0X+WJHfIk2b{QsgoK};U+1V8ZFjl)cf?0`
z*8`$+1ZhE4d`vrM(djAXmYf9N)Ak#TEv45GUPrPf|EK#Qw$@Mad;2)U6%`f4#Kb^+
zI-91EQI!Tpw`|<;Qq#+IQ)dxFq=ioA$yAB_+qc1z`Le=j;XQk*)6WtRoJ;0(C=^Oc
zLZU2mT;Gwmwg=6TNc`UYI2Q2i(jA1bY3=hH+BR?cS#|8pXuxWnbggYGbq>=7Yw5Ce
z$p22~X2hZUrTu1(o#bI$xQnmDtDp6)Z@86)-BP1cg%+m&dOgu_MDeHao(b&Xm>AQy
z_HW<ztf%d{HtR!_6pGY?yT@;?HyRsxgS+E>%SbYJAEp#m-ud~73|-e&SBzMa5~2mo
z2~|hl&71^zt+cqYx1C8gIo$251T2n^$NSQ9qG0Az7<0rSz8=?ZXkTRqtXZ-+S10X0
zzrJnkxDJ1zIO)M*k^EX?^Psv$O`>Gvd)`-TF@#7t%$0U{Q6Phh?;^wDu%^7?Nz?DN
znb3^{J@vn76*_Ih1)`-3ruN|j@9*zBTm{~gXz-J1CD4j}sZk^AzasG&85!B?d?uMK
zmxFhq#ra-Yc`|h8Cz=;Q^^MVpJg9qpvB9>LbvBMd#QXB^#5N6A#?kg_DI9tPHuA2j
z?MgS?ih$>^3w7n=oBFnZHS&PFvz^v7R-M|T>k_fbI9Q;1>18b?F5tSeiL&K>)k7El
z%WS|!xKI@sl*>-<ruZ6T9nL>N@ApnR?k9&Fytng&RyIorPDR`NI&%;JEMd);+npG(
zj>`j~50#Y|)tX}~k4Lk+FUv1mFH0nchYRBan@bLz?MJo$MB&`jI=!_|G92*CeftDp
zp&=$6?td0=Kbo&JXeD4bM9WuAkfL6Nirg&MxgE|@tl}fXC(Bzn%Y(rip(~JicXn0_
zw3akut%io~`zzD;yuhc$`u&^E7e6Dm4d@*t2utCP0v8Rjs`aGB-lN2^S>K?sLyG~A
zpM&(u<~BG(=z0NnXjic`ox|;VAt%S<G=PkL-T8y$?fiAXvd5gw@uZKxdxZVU$ll6L
z3T2>!Q1!-DcmUMyWM*!3YH6+A|M^Ksp|a)l<Xz%TRr7kn$m_r}%uXH8nI*?nyC&Ae
zis!2+pFW6v*+PBE8#<u%G$V@5U%+v;;>Iwgb>|uN)X2ya`tsx&c_$Os;rxL2!DuW@
z8O5+F?OSPETU$30H!+$GKbt^$ojF%&RrsT?Iu;2jDZO$we|ltdP0e}7!->!3gbWNY
zTQ1-+mfM?N3xd&>a6VoLWP5LqIb~2vKD5Uql&VX-<c@;h@hf=GxwMAHMtR_Cv2@j3
zRK<yq#|d89qlSIc!w`F?qVT0xyL{}HI--Ag&xW3TbJ3EqyI>A0E9>Ui<}R^(5l#tL
zfg$?KI{S~y7EL|6-O+oehdLxfexLb06yFC{-~0WqqpJF7d~3^41^9IDwV^m8HN{^G
zp{#3Tw?m8cW9?|->mOIHj&e}9eV_84A6K&1jypfrAQ~(G&Sxilae;0n!3FrowD_&m
z@hN`WXmhGO?e{oHj<hql`f+%`_v~<o_G)-6>edMJen&7RB+(kpdCBPBhu2kdsor`j
z<72J>?U2s<{CF+`g%dM2Tiak#`@ja<3Je)HH#a%CfsLMUB=-wY70MbK+|D=qKIVGS
z0bQxAE!sx`Y=i(7e2qy@n9<`V;r{B<{JeI03Xj)Vt~ix3gan2P4`skJm*}tXzbFw6
z`tYsaF#||>-Ur{qDgD#G#a2m;xsXYy#phw3uX*o{=IlR@B4LnuKHu#TvS?F@MmIJj
z1jUmK2T_&I*d)@v?+0UFr4+}5uV{mA3>bT(5{Oe+bLzJbTzIXdP3L<;UUB?+^-9p7
ziv6`x65U=R0GOXlGiC<NP;3E#Og|V>U;&?>_wfN+gHdB%6hL4DpP;;*gp6J=$Nq&g
z2@uGkBEVWT)rjBSgLG^$#6$~zzdJ6^^XPV01CF0mFrSB)mw<p^(7d+$D2*!dx9Mb|
z9}5{q`1cNqTyK;{2lmT{vL8Q~6cpYt=5&+ww}av9<@u(Q(sTVY<<no8edP|@xk_|W
zUb28Tzg9sUr)GV_Dn5Gp{==*5YK#w{n}yFP=dfmSVZcjt&gK}msVy$v9gJeno%A}G
z`=ME8-+J7*{o1m{3hx5CvFx%@R#mTC(Z6MFWhLv8WLV_^$}b|kd<X&1&rA(yH#~K3
zMMvy;etMXz(DN7*xsjsAZa2(Bwyc^f(=2mC3kKykgRGSHZOn81^kej&6h*qfV=aMj
zH1KV!u_sQWr>OE!a{z#+Nx%ZcWi28%%kl#DR%>H_($X?1Z7aQ--g*Tm3nPvk-XnPt
z4#S$HmM<;~pu=F<ts3kM<y0Pfx~z@zE_ZT~aDs;5vOE0w<w*rM3jIO@B^S);w~Z}j
zam=}@wVg>wX0<d|u+S*1?T#-hN3!?=?sn4BjQqXq?Ce^MMct3Im6b=&`xB5kqK8jC
z1anNe8}!-M8twF2G-W`m9-H3BV>y(CScFTW4JE>n6S=K-eU212c5-nseVVX>=o%TV
zt*+)GTjq^@kW!0JLxS%01Y;85;c0yMFtl^wY82RJ=<xo1(uzIdP5<`dBDmM@9;iVQ
zU!?oKZy*NeS615h$C3l<u7Z2XDGk3L(m}y$bOu~(^n7NnB$_|{qD;=u<U<9&mgZ2|
zcn;`KJ(P)l7|UJxZfe0amc80?GKClcE~DCLh36?Upgz!lcR4ZU<#sQhA#nWpDvQ!B
z7wJu&|7^0sX{Qk(9G1RqK*#NKT(&T9Cd`hu{^@VXYyv!cJH*Mcrxp+&bxIyfvl@(*
zq9(Vi(6XhII9hI=B18M}zL<!IhX<5h73iwM8(p2!er&~+{Qxs#qD;_auCA^S5D@mV
zU6e6bhSAZ{btX!KEWkiIefS36q^tYr+%wtDO+HX+aCT<Ki7c(nH*&<1Ic?}QAK}Ht
z`rnDi$HxZSIV~-%NeJs`-93mYaS+%5&-T^Y+FGdzmY*xI5``Eltv_n8R|CX2saG}}
zA$)vrfkQhjo%s?;)5$q7Urlw&)77dt(vgB7c%Fu$4)@K_lHBvI8e1c8N3FeLj^jQR
zQ9iekH%XS+0nharm_q(77x%u2KYnhh+(1>vGMs0FbX@##`M2to+mFqkV*-2?@0NAm
z@3CidZ{;qQ<!P%)GDmyAEXp`M=skUV&PC4|*VWZ6*J9TApvTWfLOz{p<{S$bXeFGj
z!Ll^*-mJ#_tZisVU64#y;;%D))<|kjjvZG<qRhK<;g_SjF**u&g9F<kf^?p!1X+Ey
zG;PkY9HE<s(@qr+GZ05CHuCB~GLbDu4|l9H_wA^i1!?7et+t~`$SUdydY*k%E>XbP
z1s|}9A&<U5V0N~X;0oyP_$>yLdj+k^K2#|izFj)J%JsWm=+~=jUrpCSWyG%w+A^MD
zN*bWe?cH#k|HKP83g-N_(x7}<@@_W_reYAlcmPExZM##2OPZenohOXjBNyGCN=yA%
z#sC->i=3^#yeI9mcl}|!KeR-+h(Tuc+J$M5)fpB*J}wWRn_1Ov?fg9SorUJ$I^=@U
z*iyJ&@I%RKU^~hF!T7Y6hDRpo_hcj)Is%aPT=B!I$v}^RF@5=ALHfMvb6D!o4i`y`
z<GOh4!t(5_WrHq~FOxvOgnQ%oDh3clMP$lZtqlMy1T;gtJ3H9u=-!E!p2C~IXsByU
zPilDz$M$?~Ry<BytIS^NS6V$jXlYS&PVSC4_TeY%pE%gsT3MxYnuY0BoS2YTYO_dB
z&-EutThtjkXcQ!%TXnC<(o(0$=NC&$V&$vKR?WeQ+u0q1aVwEf*xKGc_fIWo)PSn4
z?lAP>=2Yyseu~#_k=Fi~N!`MavNHDOZJhHf!>XI|y3zZy9?ZUQSE{az5rf|`s6k)X
z`BlqM`IR4C1ag#}B9o`~vq?~h@la~qRt~QSoU?@qdLoIhP=dX_TDhI{7G6Cf`z@gP
zT|kqWY#;Y^a@O7sj_j0V!2rf>w}x-;hI{|uN!l=E`ub+0U=1!a);T}8@Vatg!GCsa
zrlRQ?gtDdC#&;{?;mHO`0B4@U%{X#09C{UtstjUclm`c{cr$O(?RfP-Gy7(zwSDf1
zqqIp8CsU;UCe9r(3%$1PRg&D&IF;5gAf9M`5t}qnGHKJW?)<J{(YVbb{pP_;w<1}n
zqy94$GFiV{YZv_(RCtU&*T&YCn*?oqS}NY$xlh>d{&G8V!M-UEADK2tU}-C1XlJLW
z$O|R(Hmk9*k&c~r^U7O6K>-&xA%$f~kHsKJfDA(c8P>kp;-aj|AL2sSOCN<4EKxG`
z{nERRm7pdMRAL)bP8~>OceHn?(!!H;DLremb0hwXv(@eNbl2k@ag8=+dmE+SWdhYD
z*I$)Ur}(6T$UdNJ!*N@oi0XISLOB}zU|rp7Aw6bfgJQ0wWB0lRmIPT-xq>m!EpJ+6
z@j2x3*lP9T#>6-dyrI&{ASPbaW~n(V0hPtQwFE4BIc@!h0je~lw>qizpf!R9&V7&9
zOAbaZBRnh5hG4jT8c^|1^aKkJTEiUiX)<aW)ai!@2gS-cnfjbf_4TEuoLK!qDXBj)
zvm83^EWIDD`$+jByFYOdB{qGz_HHMMktS<H2jo-nSqN%fxWze$XytH5g~oK>Cs_^V
zLaY-<ETIs-BtQU-=2{QLC7QI<=;1Ytq*o|gMQg}gj}H_NPMpfBlEGLJQY{7xZ?lh4
zcz8JMGS#tqM~GXmrr)bX1V?-S3{l6T;``87OI67xsYZ@pC-p5=KJVu~U1_<-g>RbB
zhF9)421h|nIgt(!F{p8V03TXfzXnqrXLTI=$<O}KoOQtg09_I<!IoF9=Poh7D?`l}
z8A8w?@)Gkfz1zQP)twuuo;`R4OJw2=gE>{`(y?KyyWIh?KHD5vB7J!}4T!>AT3Q;S
z#!lx?pdFABA&HP;Ay7z*)JdFotEtDcvCRv3_L{GGOBxkX*XAmiLxmhsC1lE>lEL-v
z$WqIhEFovb!c&)nsE`3gHgS5t&0Pq%_3Z#d1n3B@%*+bQhta4I>*fwqkb>cr%=$ou
z>5d*}n14>|EI+j;&x%bjLc-|!=>$=_QRNgOi=~l?y>yIFWhlgh|1+T0pVr5VDk;MO
zhXrqP=AB<1Pd9M#VMo8i*Kn`+NSunQvU;V)Z@Zter)lk1CXeI&J~=yA1{bk3TUrl&
zUf0u&2eD6zs)2!sTt(O~Ny)OEEp{wem~JaY`@zcZ-@QQR^be9w*Dh_Hm+LoQ6r~y3
z*x4Zjf<__PNy)@K;Q8L23QP67hCv7MPdi@eB~DaqF0QT(008Adj3JbxTmvJ<$H!OG
z($Lbpa5?=xB_+k@<^%zFGnBZnpt0AiNsiAAq$JTmcp})ddZD~oN>=s+bV^*c<?}=*
zSKdW6M5P*FWya8ig9d`~ch}O*L&4q+-bD#LIc>_hS;!*MNzzGSD50f=f(;uZ+yXFe
z__2jw$4L+@701&fZpVs;Hx9`G^|_&8|H}LJG+<+%lUw{-8`Qf!i0TXRzI*G4zZf2Y
zdwbzP@&L!(-CY+{RRaQ|`K8F_mX?-uT2P?l`#9Zb-Bbw)308xZ6%`<6o3Uw#B49nJ
z%_Rgpg(;%?YQYLNSR;tI$)1@-Oi3|sV{8QlR9IMe){O)wr-bpNk(cQ%_Z2KW%)x<G
zgcRB?H$X>-|E47)g#r_L(rVGy7mh)0s-ramb+u<(d6NzYllTn~*VUG7YJ}~Auk!ef
zDF%kU+6GNQui8iuo^ZcY8{iR<+y{&G5R?$rFDI3m^-ITEW%$Ar<2DFoJekuw87z1x
z712`5a);_}4<w1SplgDG50(h;$wv<dZ;(Dx?PhDkvGl1CHVVQ2#mDnRfLV_2cYv2L
zMVLwNhn5w4=xRGhfd8CArlMsc%g88VC@h7~r`6et6~wy!Ew%Y$Rc+aQ*La56V2R03
zsm|H+iH|cMjB<QVngL_34D}&<5KGV4uo6URgPBT$wFt$Z5&Xi*(tbFgp`pQ|sHpa|
zGrZ7&H&wRKY&(rXh}6bbf{pN_*TSd9y6fYt9%zKIR-b8Ct%dtT{@d_J|4xFy(UBq{
z6;{-yCMXv%fF&UCTb7m#V-F>$d;gQY{VZazt9>Qy_qdF24C>#bF}#J?T_<~trLFLg
z^W)1~KdXVKrh^Fg#~cvf*^R6y_)dl)gVDp`_sX(;RUAt+Z_Ki8z>IU;N-(!kdp(@;
z>(@Y2A_w&g5IW{`Lmf|=Rn<*#LZtLs<_<?jM+r#VSL~ZCAFl0#7vMPz>4LvekJ~Ut
z(DJ;0n;);w7EUAcwK|@cWZ;(aGjeEYG#YW>&$a3MtHVCylSUDJvbh}O-Ro<2U22%1
z^&MxY$?NW5BzkK|NIy+X3Dp>M*9xz;E-tUA9&a^(l1{~v7AI}Wx^BTWK~|1G8`+X)
z=-12#X3>Y?J?NdUXbKfpMuq{M>pKHBEjA?J&mfmHHAXm192FjK)`%syRDp4WJS`4q
z`V$2vYYL7oE1LHT(f=cuZ>@AAsi~>3siBkVKs*6~dJZ4Dq9Z?<ZV<=Sd@^V@5565>
z)>fqM^=Ms&f8_q>4^{o5mAaAh`}glf0Xzap>9EhBI)bhXgnszJ3#U*{UQ#?fSJ0X?
zGBTRKmLf++0zk5Sz@((^w^+Nc2Z#Az#Z9>q&iB<d_=o?x;>m^enB?w(vHkjY4Ct*R
zBqj#_q!O6f*-@v%xeFD|zy^L*t){>Pz;@E(wR^Sp<@x6wwlwB&lUP0?K-`+K?TI(r
zU(kXB*}iGS@_ndeacb(@aEyqtG97|ZG0H}V62>3x?KyOEs$~}XCG9q5Y=rF3nsv2`
z6Qw;He|Gt~=;<Ag(uszn4Nn^ufWY%R*`$ccLrxG(Dz|wJE0@$&_K(rf9+-F`LGC4A
zgRTbx#zk}D7!H47;7?T4&I%6#oVd^1U=hA1RkDucYn0bC?>;M+?A)DpzW8s3yz1=^
z?$2w~?&YbL4St>KN|80<-uQDWSdotYi&Wnc)SEpL6L;w9&Ui4{p3RzzlLC`W2(hmZ
z0Hn<zZ_>-k%9xp$jOW-?ncmRh;#0nQ7uFN4#T>hO@$dxZEN7!a<CBR34yzuF_%3Q@
z1=p3WgS(@4s9>$!!Jdwr8wQlSQ_`!r{sLR8K<Ma{!&2&PkEJ?Z`bhtnx-N>(b=Q7L
zPm5dTu4UGUbeV>^nVD&1dZdL!Rk-!ksHm^!T8AGu0}39B^`5ci(h`ExTy{AN^B5)g
zjxI!=MZfgn(FJ!XXhz<`V&EF|=SnPy(liySSV$O!PYz*NK=zmJ$3@$!D`6u12oO>%
zTE9Yuq+%s}iM4n?w`OH+&8dpF?2HOvEiC+oJTTv4QpW05)K3wkh7#2<^1S0GqGdpF
z6~zCqP|(?YSh<byp<1^uF^!j`xiI&rxR@?@jNX#Bha)0VVAD}wN3V5Y3noy$ig0)2
za5oX>l~jS_ZT0-1l7)FSWAGdZb_tcU5OrArSw%YQN5bT4mWqjZT4IcF5ek)5E!2^b
zvpq3Fk;`}?i{uP0GP+!U-3~98*-8TdsHjNvf&zhGkm^AH`cuF9Q-=Q(-8rj!>&hV1
zfZ4KwATp}kYP$AQOwa3tfj}w<mkt#c6|y8I;4PBvcU}HAeD-)*Ko4zqRB0i53Eiea
zdt71a|8fCtZy?Y|4;~&DMwH)iW6x)AStSCc;RvF@tXh?U7Ll7VMINbzf?v=r+p}BP
z?O?hS?3fn#5hA0bU`PCcX|uD_E#SQSIkuMs;>kie0MOAXhIErzf1)kdDw(o=IVj7H
zVmr9znYeaWrh|+8_R17&(tJ!!mAs%fy}MJ@c?ICq5xbu%lQk}UV(O%Hc7X}BB>LEv
zPN2`mGZcwm$3}usBdPQTS@i=wuLZYO*;(g{o0tE~^V3aQTJTTN2<d_^Dk-VXWDF^$
zeMipCmOu3!yAe3NxC%6W93Hl27t(CgsmxaCzyUwR14jp80m-6GKhg8CKh-SJq#rlE
zYCg0U6%|nv9BObLUY!kdi(`RRG0)sr@KoHmTwKMT!XOl=(DHzH@$xcZ(bCj>HwD3x
zN*fhh-P%XC%uH3E{(N8~%@kT&`wDHU=A>!Cccyty^x_YVZYn|nwyZkm_*K`$cesai
zY!dX1>fquf8{1a0R2Q;8=2UFF85!2m(n74KJC0ow1wa=M52t`YM<mMH{ndVDSIbJ9
zBn$viV%M$zVr_e<!LFszHKhUv*g?4DxNX}F*+-nHz-pN_N64maI@a-Vq~rYIl`~n%
z4!`W;_D|+cqh(x!)D#3p)WZ2=_v6)esuRT&ykV|fq)^H7^c>~A?VfOxeRbfg<=AYI
zYkgEVnvM?f1Oq@j3((Klbe|#t77!F~5p5Prz{&jQihY_I9!E}-PxC$<02FX4FwMvm
z7}s?5M&QDH;?0^noVgJGd!?$My>U)9-~9e>rG?~~_rUG4RhWss&kc$eb8<?GGa0Kr
z^i$0+GrNP8l`8mkNRV{F(K1o2DifMzh%^jwObZ^0!`=i=5LNWeo$>GJ*nX}IExIK<
zhiC*~V@l<&^MxGfvQOLDOX9Y~Iv1VR0E2VcjK``M{rvIxaqW^|4ih62GVrCTa!N;s
zGfU}F2OC2nNFs0U@X9*|9>&}6AOiq2n7#xg#amj)3C0OyA_k*`!ZzB6?oYu0^@~#S
z1N>~*!}hHPq{v^Og>M!XYLjyagn|TwmWFF5tTgB?nSi2+FRfzu4^N<irH}l+9gTBu
zHK0xqZ;F;iRZ6ph81ufhu#f|-rcNJAJRkV3A)%(PmlaPmo;$o((2L#*z&GgX>M0Nq
z1kGfgLY>ZI+Pdrm&ie>ICMSRSuUCl=@*fHUqosUxD>_|h?DX`OTvJ0A>TqQ)E-w6^
z4vpv|cURVY-$T#)0r6J-gG8L3a-GSrFxRy8cEcKHO>4;17(bk@uAt^HSY$QbAGyB1
z)(?~0ZUXB)6RZ^xeaNZkLOEa)jUL&W*)r+7nbA9dg9wuI)(Z5qN&PW!xkxqlicFM6
zWJoBXT^uHP?C#shSwt6hT>FWsM~Hz|kRMz8oauG&`9(#2SPli2u7XZU5sBgeoeE-x
zq6C&HhfrqLl-uv%iFAoX1e8#1&W{aEQQg_|w8687^dx9lRnd2a_Ljd+OSab3Jj>kV
z`B|7*5rX2a8c>|9S{BeMDhQ%>vIpfkNF>TM!8WsZ@V^v<s$imYfoxw`skx%UE@%vi
z`y*YfEH`(WV(#Oto<juY!mL(tJ`AvJSXEnETK$tBFG4&M(|dI5b?=Ev1%WU{^-?RQ
zWd$$xFqk0Sh2qlDCZ^QtYMu)ikfh=ls;}<YI5^A!;!Qg=K)NEiOA*jzJnlL=4c0;O
zinJ(=e!XY1z`^Gp3sz_RW6l{zZ&NoBu_;RPcIl25O#hV=sZxS;D_(!e%p^e#ye*s1
zWyMCj5YEHLh@$DS&LqgHeH#;_sFn7#uo*2aGPWmIWLy*2_w5y8R#uh>#n^dFg}lPp
z)V^#x1t6OkP3!iU^clc?iIp_}I7Wxu_tOdZn!`K7grVQ0E&eIeqq!PPT(SStk;uN=
zqfd=cHayWdnLa3%0|^4)`8U5+SJ%*M2mc&<Vztt(6{ZU-3R}z8($Z2yPSQic2bS#g
z5FrO+*?zwMyP2X6p<qttUf;WFv8T%^ue-Z`1;aWWLZiR(<Wc7c2w?T?+|w#DD0zfT
z$lTm@81yv-H41%g9;d4V{sA&b+yDT^t{7<;@n2A?lnXAM;Vo=33^{5kZZ22G?(8p$
z5mM^!-xut8age|SPS^~n{yDrNf$62e>-o8~1S3w2thyg8@Sd6xKdTmosHk>lp3^D6
zGL|qoZLE-C(HH=NXnW*vvA-L0U=J%wbSnl?^>!NS==9(6C%x)r{CIAxb?#dZbqG-<
z1UY0euuKGID5!U%brIjfF&`9*m8-G@l8Q^u&vSJrkL@?SD%%bEYQIil!%uwGKzX=!
z3qbG8!#mW?!I-TIpPY<6?=kr6>K%0E-7Y*vDd@`ZKD(ENE2~d6y3bTFmtI~Txe2U$
zc(lqmhm+d>orlE%&=AS{Lz66#z*aDOU599MMees!Fv&#@nr~wLvr|4M^HV>vcQJ!5
zDR=9e9IA=8aITd3r1M?3-vcck0v-Z@)1HeG4uW?HtS)m{z_1LRhX*E5l;jvfE0h76
zj@mn>6m55f$jJHMzvHJ~MO<kQ4(`g*f{+4OC!$NNmqf<J7~0g+VV0MN2(Q%~?5X&}
zq7uRa@#Zo~VN2-VB36*hLi|722L%74D$*%CJLY>5LkYW<yfsg1U|C{li+mt@%%j)*
z$IAh8B2{lDkFNUWVs-w5?d{X1w>CL6sAXvwq_mXHz<LVjg4<Nd;zcJRiA8de4&z^u
zEteRm<$A}ztiYki(y!OdL!#6c^#LR_F{Cj!zcW~#eoclE<pcRF*?s35X&f)7*q<o7
zO`GA1lQ+ePL4pKyn_925XGO#{r2f59gdDiH?f~Q~^OoGUbxvd-%@<DeE?@#ps0iR%
z4HWIE#n$l9r1kVddjp~_oNJ52#`l516j^Gymzi`&ZGtf$O&uLs>ixfwY^<T#*|CI-
zDXK=8!kDzLbSreR9D^CBx&K8MxG_>Dru5M$C22j~7GjN*ysF4KZF5>`#dOENbz^@$
z@rZXvuS$s1l@MX;=b0*}K1)q<VhN@wr<G}~xi8uMBdx6mp?=NZxK+lqtjnOvRxX)4
z<Uxi_PA=8q9bZHlQ1H19q)PWBWC!Z&>yr;%4JH%5?FWt#VuZiqOy6-HJ)M&-z+!(3
z6Ii^jyzhUcy?Kb5l13=x82BabfLep;)ev2iTT(`buR;`<5+upCP$uV&&(5CrkBu})
zQD^r^&=#kE{1~jy=FaJ;ZauyA9skA|F<zQFA~IuiC@GF~ZX_emp9E|GnpL-`e_!_k
zi}_<5Am~5<8*`|rDLH9MJK4(LTCHVid+LR9fUeqdi<9N9-!YrQ3;=`Z9@%Rz<Lk9&
z#Pg!u0oBq1Q#Pg`pHn3tW{4dp2_;Cv5b?PNapc1K4W+|5Z^kgZnvv1zxhI}5W^?lo
zZI;PN$xDui=ELT=j|U(DXV<o&U&bE>5LaZw6w1k0{X#7VP-!xCpPp(*(Zn9Ky=c}q
zXT-gvZ1?ru9UuE{_Q@Az?<LuK(+Ooc@@rKYiipzC2pqM(MkXiUYi_=S->~~sh`WLU
z5A)`YsRBTs)VF-HxVQ-RQs4rUw`6N*C;=|6^^NdW0cfEdm=b+iIfj==par>8AOCX=
zxqHMh#Y6Y8l6s^w2A)#WHfsqqo2`*!Vr$;Fjw}>0NhONsNU&I9W8Gkz^|~q`LJVxQ
zO8G9lBkaDKyYUs9Dp7fPemE^Mu)LqFS7#Cdi|qZ&uY)B<YRh{h+zl(rk?@q7`bqt7
zj+gjCw_4!2%T%Rw^>CFILez<!TG7dn<ou4a)%W96N)Ulrd-~0>AA>_Buw+%gQsjyI
zy$slxR%cm9M&9V^y?{(arCRMtUJJNDeT)bg9WSpt+I>;y5V?~iHg>|LQm+h_U3b2v
zPnBVae}#`pkzp%aemp?M?irT-Jo?u{zEzt_i~LWqAt=6O6=B0&p#?Rr)R3~Y40mrV
zvqYuJT<3Kz*zLX`>Jj(7@+=x5Soo>-n1K=g#*fWT=xCYoBroBsDnqYvF$2n_H|tpc
zk$Z7?$f2U@ci8iUy_lGo1y)vrpp5rS1<D^8h-ev60NJ9+rO9syvJ{!42d*0+_XE=_
zJUBYQFaP3V^eY3?#On%;L4LMRb?4_z%&4daGc!YJ>?PTwxj_q7!mB*FX=wwspUwIU
z2b9+0L=bU7Xvv=BBud%tc8KnCbA28M0AE$X;>%{-ztiOV-_v}9BbT-rap_sOauw`M
z_r9<Od)Z!039jeXoQuo)`@!|_aO}BJ_qrM&=X;HZ8zK9kzvzDZmqxpm_AXGrNKo70
z*s0Yz45r->6U)Zt;F_-W4H4RgXU)>|bRlx+S19W%8@J(|$e%0OQm{b1G_E5*En1Q=
zc@!T>fgC>f#O!Zcdn)F#?7gB0Wv2|qL1_6(?oSl!1}$9{koQUxA{lT085`griR*9F
ztw2<jDwy1+qU6J%#liAOz4ZQ0BO_0Z-Qf}}L#c+hb7;?OAA<A0X3BxqGh|+A){lck
z+-bCWtgly=VrvuB$jhz-7uatuNbwGsoKm)Z+9d@7;gGK3ZVS|q!(qfpqOKZDwZ8YJ
zuxuPwZDhM7=H;^zI5z*PsA!W%Kn#xE(+Pi@X^#|z7cxN<b^7>+kB3vPi4|wsrm|gR
z_uNsP5m2>p`?F&z7<W&{N+4?8{Autw72F|r61*y1D&#dXY@n@O5l+Q8!B8SG_axiz
zjCitP04J<QmUhLXh#9YKjNZ<M1Zh08)<eta4_ep(xe(%+wCh3Fzna&VD`UknH~M3<
z*XIxT5dyq&rR6O7MQ;`-agqmsuiHVWgNg12!~8e;`9DyAN4+LQRFrFOw8zr`fXYlq
zHd^5zLkrcOk4yxGm^dOsmfYFJ#Z%Prl7R?~DrFE^k^<$*>@2M>asIAm_%M`v^zQ|%
zddb$xN$Jwa&<-GO3#qww{{a;l*D8z0`&{2oqw<(yMZ!YhW$I4|0LaRF=ZU&iNUiN=
zKh!a!e{buI)*Bj0O|4WRRjQu(cXoi3QFEJ>jvR%iae42><>e>6#>*RnzsVqTef&!l
zyT6iNg_0a8&0vp-roH{a(3|Gw<SAld@8lUPAkgbV?qZl5cT6igK-9VpX6WAk_H#A}
zrX^vP8YO%?i$zydbcow7P<$`3Sqzu2zRM+9{@Fnn@`!W^x60(@R#eebaBFrc(_A)=
z+`Cinq%2lG{<k+=VT#!;>>GPi2#E`XiMGlQk!oO6Iroh9>OoUZPSO0MM+VWz;_;+Z
z51d37=)Hc0wl71Mn+KE-6rIkcVh?%KU!}++!z(7p4i(CMzB=qNXe>K``)#$oa-y6S
z_IjEa1Pe55I1vf7fcaRqm(@y3uSw+I02=}`0Q^BkNpEb7ANd_ImB+=E*u*~k=g<6`
zw4PMf#}2<l#`hRDkY9~<;}%_xWByQ5)v}Ebok8!ST10ugE@&mHj~8e#2um0S6{q|#
zEp^;^iPzj}f-Fond)e4I@Ck)5DJ|O#In&<=8c+ZeUREtE^x<-u{pFETv;%hga|UT-
z!+7to?d&eL!@`Z8akMH!n5&NVrRi0aDN^XznVDnyTF@=ki-N=h!QR6!c;}@?E033U
z@BEq6kCqgXV>WLY{$*7(Y-F+Vv_N_)>(KY_`)+6n#q}B|SJIz&sYi?}{qFAaXUY)5
zX>2EHYYS<0-%oWuyhiq1hXKh+Y^Xu7UAO1WqLlRw5jwIHu84BQD^Si*CiXbjRv57R
zUO*rrYp_Z4@kHNEA!^AIH_Au4Yy{z}(2YqM3aW)%K~XZ<oVDFV>#6c{retCvHQu+D
z@(uwz5`BL!%2Q(C%cE}?Rm(9l7$QOA0Q<`-Vg><c#)@=@57t6fkoQqh0k<3B{+-x4
zT>o5!(}ttQ(M2}I8s#6ipl&biuw088$3#zW3Zc&Y`Eh<P56QG9C@ARHw{N^<_|Lcf
z{%e|hh}#(I-xE!ufz3d6iHgz4zI#JXT)L!G3$q$+7eA31yH;<D9kk}hG9KGrV*+MN
z>pg78$VcCIX%U?w_3-2MhRYRR92XbjC5P8c$<hUtRX3T6#X5Dh`fQbPIS2zRsd9*L
zh};J-|H&Ee0NFCg_;=hI6Wgm_MJpJp`Kpd&{X#kWbXY1GycHS6m<i?}C)31aqeX%3
zd~W8-Zu!G$VNp@ZRK2heDFENTiWMVS6`YNW7Dg7jx!xU|pCeQdQ8P7w{ENfK-CY5d
zk)YeTnOJ<)>x?&lg5kbRlbSwd7Y2<t-Nj|90fj@iVrJvZ_O`Z*iyb^?ys=?DL)8x(
zu;>)r^43-TUwrI+!(<s4ZCkk^`h*93!;oR{TR(@FXwgrEK$HJ5Hrb~S+MSyv^BGuJ
z%FC)=^wcI|OY$fZv)<6Dbg0tdM!Eem{_0J!y8KRBnzsuCJjQFHwM$C$Fn{}^sy>ui
zJNaRT1_AL!7)o)NDUt7}CF98V$~A0`Yu$DM^N(*cMX0S~ev?Irt6?imHK6s}I4zuw
z($hcsql;g3=pn!Px57W4R@*X;+u)pTHvNVHDxLbj(bw1FR9O7(EQuYD7aY-pmWTeS
z+~OedW(odW$C3`Lf)h{(IB$1?eZtT=eeU?R*E8D#1f53Z5*7IPKa!jOh=9FhV&W^k
zMuvj}4Q@-<mrMjy=amzu%|N?z_V~pa^1X_(sLN_)WPm-586iJyK&}0owJb?;TJ*#|
zGGMvP4_I*$h1h#c9?kR-vYYCJ_i(2BNYQhG=sFB>4BD`l7hrxkyK0oF!5Q3V`mHqV
zZFo?B?7JxyA^S902wR%q(;x*?3KKUsE>K%n$7jEso{~~@utmFTsFaHan09hma7~Cf
zAS6h-kMMv^3hpC!ze&6*Wmmy;fg~&im1Z(zYMyj6ygptCFeX&hR{J!m;PW&4t7`B4
z>f~xLgg)V~t8k$ABrPy6;NR8`MJe`caNWNwkeRwM`1+n4RLr+qWmO$aufoY&-LIY}
zr6|D!))N&)pOh~wE33KJd4%ziNi&(sbzje_L^(%SRTTj6-X+j_X|}B3A$FO5US5v<
zJnrqcCgUHG{K3F&aiZXdc6o`EQAe9um0TkHcL+~<8C&bH|E$i&yQq%KhcTb$I|?>>
z2E2Tj%ZKX>sS;b*>g_!^K){|XD@P>0&fG#yN2@DTK#c^%HRv&#Wm+lwt|y|3nmCgA
zMDN~ET;S44H95pB9!uUY#}K0(iy%UOg`$o{rxiA>`9FaFKDeB=d+p`KQeNJ7jHMiW
zojS*nW7xkr{vC<mKS#ky^8UuB<1XbxZl>7F>cK%oRlD?+CzB-Dlq&aXb|}<6_5r|7
z-M_QIhkL?VS7Sm*vz2u~&k_E+Y7~TZpycD@V{2Q+MuPsE&-(!8@Ww^%y+hGdF=v@p
zjk6bKMt_E41&a`)A4i_stWHGU>9NVp&gTrCvaRrJvcK+*ZgB^o8=JVfg2>CO$6>>t
zZ<bY6>eSKQ%^l^c4@3@CqhV$!Uh7p{i#C%1Z0(oo`v3tEc0%EJ+6MC9<Pb2(V%>r?
zrb^wG<mcTi?a5-ieaFv0+H|P3@WZE*^ZvUw;o5M6ES6NQ)RY_k6Kh(<On7U=4>wfz
zTa>5EHD+?B?C^jr-$5kpKoon&&p8i5O)YnW6Cc?D^V(rgl>i-`g)S8-HDN?<c{8aX
z0q0LWSph!KI?8K4L0@di;vtqC?@<fB%AcM=jcce~G!Qzl1;<)i8n|g%jCq_?zBWdW
zP}fk{ON@qceC&EZjEjQO$b`qptdmDH+PWSTIXKwS;qCag$#r3YBRlv9QCM^B-i*kD
zG?knv0qLXHxSTb}(Jk}4OT6ohW6`bobQg+X$<q_QJQD6+uIJT#xYl-kOiB2cGiySw
z>-5I<mq}^)bc7zw%@^oHXyK1oS~*N5Jv&~x%&WJ6cxu*zzw#*VTbDW-R&H+Yx-p@5
z7_-sD9MuVve%Ih22gWlkEU!huWtd3%!ivM=jNb!2slQr+@AsifJ<b<!A|qhez;TAd
z9qBbh9rkmeAYRa*{1)Gu6m^2KL0q0clU2h|uFuJ7z)LF#Xo1&wID&jlv$l|uxtZq4
zlC*B)@j}YFnApS0C~uZ;3}D_V*dxl6WlZY|Av(IBQT-B1BR<8?K$b%W-h1*C)Q&ej
zWFp@6Q!YjU$F19m2MxAMn?8ijpMf=jbP+GF3su<+|C1QQs#6nLUcqzY_q_5B^URhu
zk5@M-qk0gRHT<+3O`Z6B%aH1lJJlJ&Q@_pjgZ8~JCtWy6nm?yE%#Z#ItSJNL4e01^
z<lOYC^EY0VU5Qmxe7<)7Glm2%KWMZS#|fA2w&OuLN7VbuyFEV-_sfJfU3_66YV=n~
zSZ7?uaQi{Uj|kf}-y3K}o`cJ+S)pQ{rr^%vT%*1y{YN4s6>3N#xSQT%#5$dC8oU&u
zXnVO?o(CYed#^%&-}^#t;4>VD1qyKXSv;-y(^WF#m2h5CM@>s<>EDyqQ;LoU8Jd`H
zKdoHr9bjYBXt<cNra!PX*VRHlz4{#WPRB4PX5nukIMiX4hRQ-PXyc~!!NYrWkGZtG
z-okwRoN-i=LCSq}efjYb3KLs|X_=wO%gmWUlmpLzf)tSNDs4-_{o(1deE;wR+2ciH
zIi#odLzUsv#SpuPfalqIcOtHZBa-0E{IPM3EEMXOLpb$Cni@ezN2XlsTTRWDDnD8d
zlJVZ&-UA*==5IFyW#akdm`TzKOxPc1VRM~2Hlwm)IyV0^k#+2Rw(4_gXKzojs`u8)
zSn_rJkY-=sCr!-@lhiE@Cj7Ldx2H=Uk1071>ClS{$4_(L5tuELS4$gchwA8ZIY);t
zpCv<-vEX3<bIW2nJVAdBoQQ}A&Fwt}oHfQClW$OxA&}(1Txf6YdeO75&LDEM`OXR%
zYaDynxP~ZE)Nvt|AWs7Y(6Fy$5j8;hWx{0v4gNQLD0PAtgZ^*X|3WEpsSZJJqgB}q
zR@9ruMNpXiu7ZH7bwdP7M6~q#VB#EIFm2vP51{0J==|vKx?lGoDLKfU*sSt$cH+!j
zIS`@jn6LUrK9PK@nS~&V%Cp@N%p8G}MO`NklLlS#Ew3+qfhO>=W??P`2;lslD7e8h
zByN$r&nmNeRU%Lm%bmSAMU@VU{Q7H-Thx5qcT`w^3K=*+R@$&7Tt7&3;7s?Ei5=Hs
z21+%4SZ0(DTLwS+=I*@-Uitx(bH-q!&9@{@Uylvas2P7g-cyP^SljD(3B+DY(o=S>
zIS_=z8e0+(%YjUZ^Y3R91+~lifNI%{CeEwrxwrluRLDfbCH$~iwQAlp5Ci%9g0uks
zyj~4y`XX_vco=|~c=K&$u?R)nEw^nG>D~y)iqQ2-rp*V7>G0+0*h*%g^?fWv48X>F
zYf{G}C~)`WFET_IlwnMyV*0@{{H>8L3ILuE(YuK5VY5EpL7M6~;-c_^T5OVXajLHh
zD3;1qn0!1DV%+vkHQyv37ub^A;JLMizkCbalX;*UC;CS)G6rwz20@W7xQ*w_Fa_(n
zVLg%AWal4SHBe;YU<0X;(Spf^N+AF3;a{ZG0(m~~s3VN&hAq482?OX5U>M{d1*Eao
zuOvCXC@mnDDEu2<p@C@ocN$S6W94Nk>QsdEzyK3g$+SW*Mn`*M;jO2!3ABh%#@Iq@
z{X5j6e?10c1lZ5pK<e&pBFLAY&&*6D#4#bFYxLv{=}|+olu`rdq~zg(iSuC=Iz-OR
zsIb0n9?t+&Lqkd4!oaxq8C8iEeJn^c>0HAEqfK);TJRm7%s7ONBWSW4)Isr7cBD?D
zN6yW2EN8DxVokM#;l1F(EB<#j4dCcukaCj<JrYtt&L(o^ZjGt5^}pp)B_MxVSv;?s
z${_1<)k>u!Lj*kRSd(&c_$$@FKl+OSQUz+|=_*d}Yy%Ph={c2WYN%3HZ{EPQJ2hjv
zM9KR)7NFYAZE|z7*Ca-I3wj4o5k#4A<ZU8{flK<2vx5~*hB06J?lHr;mC-5x6ShPk
z6%csz2m9)BKca^zMIOx(RD#OKQ13ucLgMwPlJyOm`fYNX=d!i6iFx>Z0z3_k2bN>X
zWJg4g;8FH=8~wZ>{2wtnr;n%_paU1;fdj*e{*fj7sy?lvByTEadnLdS0UJ9gE%L8K
zKm{*RI=kO%Z|MeE{2oV4RjTe~V>3Tm76ncdpjpyVgB%F&e}_JPBfN$^T%&C&qO2xW
zX7{j-8wbudfvo|u{x@Sj(j`=6;3_#JA2?}T6BA@<vvQP?LjiF@(Be~Z`dkH{538QM
zyateP{f3dTf41VuV=`NYAO_J6fZooA$Fp5GHFUI(^CKkl`gjXGQ%pK0FucSOUjhG4
z9At5Byu$tZL;e>uzWxH7rT}OAj#HI$vwb-@Z&6?4{4=h9YgB#r?p^T0u-$?r*dQ)p
zDC1GYx@UZ!h-umvjpOLSI1L9PWkjpxpH8i`IQB!S(+T>C+1Vr#f0!9$MZfqBGkpGP
z4GR*8|AT|zz4}IXEv>Au)ZGi?3Gmyc)ONRbx&U)79ez*~B(S7U*e23Llom>n{`aT*
z!q7>7A-<LUesqycN3%7QO_q3c|H96h_&%#+;yWU<<^TM?Y{^uGcFkr^ON;De@ep-|
zR#uV#*|y04<pLxn-4n!aRqIQG`hM`Q&-By}g_QuhV)#~xGDTOGyhM}%j;s;BHD8y5
zhu?9OonSX8;pbaOW5T~_1$iVwB8k$MpId#S1`eGg#h-9#Q{>awjdJfhLQx6*!I_aB
zUzfk|;+u2?3?iUv9dvs3p#ks`9E`!8lRDML-QZ9`#Gw;4s3DkOs(Zlq+2RfceSOjj
zQlnjdDiN!!HQyioPZ?D^ohC}-xi(wkMg{iBmUU4Q_?bwH!vDQ-&ccl5Iyi+us1JUQ
zAp#zhs30h!D^kiR^m0a003?CjFT0GPA3uG?#TYgFPYp}gK->AuYk>=&kdq@)e-o_C
zpDZQ08SS2o_)i49>7te)fQ&@}#YQ-)<P)9aav<2xhs_4p0s>XlPOwjj_zKhl%(EVZ
ze1Fp6(oZ1(cp3NK7DWG1l+h92Bp7oPxc>J6Ty272@Rfh*vj0!2X$O54-t~QTWEem~
z?G_WcDENgJ>wr&`0B-pUEIQ)5jdk{9jDM>Ln<^SNR>S+;9X&~OZM+-je|6B=)PM>A
z8#I(+=Sg=Zad{ke-3@UsCA|F0bJA9x#^ACrI~Vd@LEzW?ma@`!gOnZ<Yw>IdRm=CA
z(e+a2^Kr&fiN=fbhRO$_Gg2Bs><w4Oknyv8nIrk56P~=oQ=O+})-47gT|}wHVe<!j
zyF%4u>^wOYL<)?U$mn?qsFso+Aq+bgo*5GIMR;SN9F(PC`!~S-k8=<|V-iMgdOSVL
ze9UhCyW#NEns?WCU`gTmP)z919-wsht&XS1-zy}H0(6pCB%(croZQe{v7^VU36SvE
z($e#h{FOC7;QrEbRf~UG-trnrY-v0~mZ-Ne4*B(T9ms(!#v?|+<wl_!O{Wxd9yO!P
z9_cd`j?1(EE7e=aa?ME-5;pMXiqqq_rkIHB{XITXXB#W?$og!Y*z+yr%dJdWaI5?A
z;Mf@Db?e*GkDy{<k?^h{xRhRt1$b^2H7t>E+Z#Lyt>{r6+P9r<;w+3CrZKf`Tsfp>
za!;xV2!OL1OiWBIWWE;z^l+cM##d1!pF;7uC9Q4?5$ilw{0LXyHyt&_rM6v#q&1CJ
zAGP^Uj1}q`Jtw%!?%553@F!tr3Kb9Kotm0jTCi>fFHvu(+&DuIsA=P8a<6v(?%s%!
zzud;sSH7=!KKgDm34wkLcsbtXey-!WE`ZX)waaeLhw;5!?|u|Ja&Wz+4pB_g5xmMC
zve$Py56K_#4v|LI8b;3FNWvKok&^dX5rxqpM}qHM>cRXbxNc+AVg7td6Q#wX=Wp^^
zXXNa)2Q;7jP|NV_p0CR9p&!9}*wS^g``==!<z87@UMunSv+4S3F0rN-*x&Fuh&_}L
z<~FeV1YC;`S-9Vj2Dq&~4}=+M&>9`o9%>DDn%xiOc@eG;8vX6<+42XMVUl#eL7j6k
z{?SO-$$KeGcTnRc@8(*^;bn|aWjGo05*D5&zu^CL_LgB)H__WDDxp#W(!D`IrMpqs
zba#V*bR*q@;HIU!Q&PG?8tLwk?(RN=&-tJC!?~_=o%i*V>|f2SS+nMz75B7|&941=
zg)Hiwb5xF+41<q(1ESsJth{xE(b9SE=v@x|BuR)OkmXiNnoH~#GU9KFYjz8Jd{EdZ
zzVY>c&u?-*z&cPq;L9b6FehUzgAe^DOh(Xt$3>p^iwlH1?myz4*BD}@@urfDnJ@CI
z^QnqGPUZ)zJ)8QM!orz7RJNkbb!Y9al8Cmv2Vg@qST={af+ka0X!jY064CjYYw^fc
zDkWYQnd@8(Is$P_8d*(tf0Cmm5p5B_bz&yb`vE<E+aeXM>;52wOHF*)Pwb_^coCWW
z)jY_y>%})Huk27}r5TCfCgtQ!wP%lp0wvOKN1kuX|4pZb>o`By-S>+JP3MOaBwkv=
z^hmV9GPOrV7B?eh=@2t4TRAIjSd9=UtUoRjq(y%v@s|^P6jAmSyw7j87&+CS9(^nH
zZkwu^w$#s)8Exub6Y%<wIj7_vl442>zm0o1`rBBqKi?P_Z+F+T7FoK5MvkG<XEOBM
z=PxIkFMjs+>ni*!1O)BaO-$YB^!9a}Z_DtZ&;EFGIDC8#_krv=Jx-gK7>#o7r>4Wu
zRELWM*iM}2*TwC}W`N}vQJhR1<Z5NxRanD^aLs$v!0$9^-dIYzDZL`~zqsGwTJ&Wr
zVd!+Id-W7f>pdYVLJHZXTD8abO{G}OKEa*Y$w9Xo!v!s~QR?Eiv4cuVB{zX;N5i}p
zRv&4efMvy(>IcY0+g9nJ&6^5ks2Xr}?bm#8UJLEn^$WW{D=Oi?kGEHK_<$EuFJvOs
ziXbxd*MZfLA_{WW=12BC1u5;ADu#V;0T&_1Vxs2R;_hz-$LDVzEo6}a7~{7JKJoEQ
zmxaXWy2n$l52h!9{BGNmC5=~IvHY<(>w4b_1YdA!zeFVf`)GS^gcB#y1CvlZ`upC=
zl10XkLYl>{_kQJjq)nA{8!gTAEH8SWeZvHU>gx}T_15W9+4=^uj>$k=<K^-D7;W45
zq)!#NeCM6QnGVL!L)4%4Mu5!nmPb%#pC9*N?IpQVJ*ca_t9DZ0*Ab7sF@L#Pkb3ZC
zshQ|o-CvgoJrRJM15m+ujb?|bo2A&dNQX5Y9~I~1Se6vV{k>TZc(Ya;jXQVV>1k!Z
zkDgxo{3dSrZ#X5|z2LkA$C~fR>wXRAVLx2jBmr@qi~cUnM6gat{N`VOSX}TxgxVH)
zz<E^-$1J>lwG~JMdG{_u?Uh?l22*K=qi%fwmHY}Dfu=rdEkR~!*A!1FS)P2mU};(g
zStIUNBdzB8-LD>Dcn1dB`m%?`T*LL<-SV~p-Y-2#UqysM@ClC3{|IYgdTq3Bjphy}
zp-T<#uo;c4HiiwL+OYmsf1ZYzv-+)b#RgBl5uF-N`J(gB+VzAmk8Dc)xzDJ}W&2=L
zbAywb6AP*^0%BS_B_=y23oDvg-_Q>Z8T#r2<$5}<`(8X*9_>Ce_#v_wsLuwIkO4W9
z(6a@xz8#MXns;d4VK2b#&P1W`($jLJW5G{8f9E$cVfF79A!Qju4#5B5n8zfvV#{gL
z<L^q&^sHNkwiBLWNDS{50rjIk;tzZPpIe@@Vj_F=4Z%?w!A+-kqJ;Q^H$2B=@$OTL
zdrwG!jurgj*^fBz;uhT@XR-h8i!WtVWoIQo&MZrRBGw027Af}lg-*RG3z6X5XCL_<
z@AoHQVI>Lb6;@>b*AOFG|Bpam|MR7I@Fi)u0DJ=Pb`&z6QzoIz(jRgEUdC*|f&%)a
z7I%gG?;=YKv{hbF(f;+Jt$}UrH+j&{$Cm2(V3_c-8VdbuTc;78)Snk2-)6JZ$ouWa
z8N`~uSS*%$|H>PiQ+OgNwAh!v<88}g9I}<yWfmHSWMT%Sl3B`0)Y_*jM|zd2)oxqg
zg*Hcv$FwuL)m9|!;2hDDQFtjj$&Sinzw>Fx3lX1;qfPXC$W2`O5hqbjzQpiHZ{`R$
zeH>w;1x40}p5$m^Gq154({e=m%~GDU&qKzp$Dwd(KbcV4I#1%#ZNWB~0CK@^gd-cG
zsVy&GP3?!F@&eW?jLYGk#R+7}xoP0<-`&{RXM3&x({}@=6OPq&_N9t_c!^25L;YxJ
za!~8-)ovd<v5Ob4HxY=d8m;7G5+JY{nZRyu;-+I#iO~=2uco#NE=u^j>9ES!Q2L1L
zr1<c-TQWoH<^+g8pyB<K{tFhbk3tECpWn{@b~zQ?7`CQ0Xm2$OhoXi|AKL1dGFo5|
z_;svBKr=RJKH$IIjzK>PN}Y3?OTxYx<UK(gum&-X`}=zboP|cW$?k5k{oyBY-S`P*
z>EwYDeME`)->pJ8*i36ChD!}ZLA@L|GRS5JyHIzbXK+6R*u%c(SmedFQi%_-9*oW3
zY9j=L_<Kmm4z0y_zOsr+3{l%tI1?N&GD3_exNzQ2e{mAwT*syJ14V%Q@MMaY#Fv0&
z42i|-2^<aF<2V1G-!Xxz|MxRcw@1DI?+?O9&3(=O_^x+Cq%5rP$qjIDFTMZM?#K7L
z@B+PS1iA|M8qT*ljT~q)+{@ofM_{zJ{n}CB(%>G4g^cML_#kM=f7k!l^EPq@{)qow
zB?ON@P6Y6{ydtm*kB^g41bow5Vp<+j`=8ks4sWbz^3;MPte-J3^l20Zm}9~1`i~RX
zB<>SDfzRVJV{z$k)7}#PuRm<Y#)hI$&ji724N;?M|I1*)3+{PrOIi4HKNLbFG*%o$
zFse45Pkw|zDFs9D6W|CZo#X^=8R0$yiLs)xC;B}KZUYx?nv6;4q~7U3M?xYnBh%Z+
z=LFf?0=!*1ucE@fBM|$5uFC7-ezw}0Lx?vG<#7^I6c>wtH%dUE?c)Wii<-NjwENGW
zKgZ}uQBMaL;L7yc^#%3)sW34xjuzOr%gUI;#iF93#(2rUwl3X1B_p(TV57qOngg!u
znV6_)Xk3RfFD)$@bcISwOIudueE^gG$-h>n$Hn0>FqAk^&x880i~V^`N=!CQr8G6p
zX1Hz<Ep2I{P|#o;A3Z#fKk7uij)YE9Q(Fs)+0*5wEWUny`e@tyF@l)h5hi^5K=!M&
zw6v_u)rsSakx@=<t;=kc<#0j~<X?v{UJio}`{>2d)6>gDN1LKjulx8sH83!Ml$DmY
z{I<YuQMQMJBgGF8d$cCvztm`DOQ&pa&H-i0SP3aA@wH?8;h_zvh5MiTzdi~Pr~=ih
z{SMAWU%Yiv(q{GIHr*>i!kwH<A!HoR*jNgtg6B!PaG%>?9YDV$e@%=l8<bC@q509`
z1VB&Er4_@&J*q5s)(&KSk$6b>0^18A|BWZjXFVl$RzDS2S5_x_4A9?zSX}Mq`$Y$G
z^3@d~^NC#+4&ziVA%o7mi*3W3+uBxrKcjCzj|hQ|{Bb`*K*bk?;^5#g)H~(&#%2cz
ze@8<IBN9(!AD%A9K*`L_6@^ylcVvwzMTot|V_-#pY@=_k<i|+$+6XmT2q>hJkW<Wz
z!^3>J6(&&TxKAXn9Gy{2A1-!(GiF4QX+n{yud6GJh|>S`QM*2^Gnm+cs+YHK-;Oo%
z)Yi&--dkAib?AEC9WJ$1RIo})_CDZfHgfB4DM0dzi$lZ0RJ9t&bRUP<A0q%vdR97f
z^PSbz&(}*xbxsFML+L7LXbTGq=%oC?VPVzP)g^ByPJ$R{)e6Uha4&Wy4*)rd@~7Wq
zs*jUF4+;T&Hg(t<Sp(b=StYXW_m7ENgQ*AG+t-&zE8GF`u)<h|gYYPOvmKS4U=(3t
z<YTIf6J*7Nb6zS<vb#)f3X_LA!xLQh@}&=-5<V=DqLQIwf_MImkBeJfUzgBrc0S4g
ztX=N>&4q=u?CcXQE$NQp=NqYEfiRPsy9@q+fI!&m1m8KV^3A37VHUL9z<~A_Ze5NK
z_Oh(AeS$n2tv(An0{*IhZbqLZ5SjInp*uP{zIgHC=;#O(wT-+%Lqh|8?IJ}3Dn7oK
z=(#(SY)8L-#uQxTtiBc|nU7>WT5a3AyMng&Z87((RPRWlg^ZY0J%>wpdL8$XWNv<5
zNKo+I`}cs$f<}9!(d_~>6(0|8jLz&cbkpBrIN>PuePPTLH&0_v8!eWoXwmK5%bY8I
z(ll{kZ&q98=lKA;5+ma;U;#$nfK1^MO>M;SA))~FC>x{xvT;pc_)dAG2WL46=yGfB
z_o}<1g~mkwloTXV;yHQ*yT7};L>L(15fS>9mUP<7K;~zYx*SR*(Fr(G`y04c7umUJ
z9xPKj^^nG~F=1+cULh$T<RJgMcRDpU1q04C*zWdtrFdXLvjf}~T`QH=i%muWpwu=j
zOjc-3{-<fnY6`wOch!EI#r&cc=e686Cw&C7dlWF$t69KfQ;Bs_FFs|VCyXGAoN2E?
z){&%Jqvbe~j{gidc~IyJin|3~X`ElL#1CIkKy-gHDDd#R^{K-daLd+qf(nj2WUm;j
zn|lgA3)Uf`XQYH8l&z!LA4ot6;*Bq%K|UKRe2KM(KaPkz?maEV#g^p!qVh7(G7$20
zc3UbyDuSbV4#of<#>!S8XJD0{K+O~sv|@FmiykD=Plx@B^|YU;?wpjpCNciHe!<}L
zXQeNh9;n8~#uOkeP~_zye@z&E)YrJdl7U`;bI5RyW&5Ot_t+=|<`t)GKcx;#<kWFT
z^xSs$COQ?})=lCx3yto&pf(P6KK{$u@5G}*Zx)uLVKww+{&9FUesu+wmsP?20QCr)
zPvh}uH>n5sdJ6RiS=_b<mrDXVi8i0U<up*hnP^~w#V5@sI5adb?ZdFVw^p~uVJxLU
z?oexXxqLFZg89yNVSK7;fc*4fDFAO#Pb5`U6_m07IG|tEF<7q)CQTH$A4P~L5!l(V
zu*_&P$PX-EZYv?a-HPhx_z2c#etygsR<Te047Rq!1B0-M5xV_u2N?+XCazO|UAoMA
zty_A8CumZAHph$OvERD*<L}KNTHV5mdN|wyPmzW)b$u<Rk&uxIS8EcA>}9pXf8x2v
zQi9L_A`L&S6Gg(?+(axp`?0SXwG6j!OmC>A`I1744KGd7fjplK4*gKnDSm8h!d<Jb
zR!m|pz16-ff(dsXlP{(m3%iKE)z5^+g5ZwPvA=rVxb6*j;m-r`U(3w5e^ZoC|M%R@
zZEKZ=&2)9V09KlKK!-^WqTeVhbHsD_$B#bsm3;E~TaJa?Q+Iw{h)Bt&-bc#|R`8E6
z4zJt;{(&g^$2I300SnC&{!|9RkXpaTbsQBe%QR2uV<l0ncrD`}?`Q+dc9gGoe@>4C
zeh)JIV5&fDQ<M9Z9az5Hmwh@P9}0)Iis&Q0ksDZ&Gc{FEP}uq|{iMofsa3t)NL;r>
zvzi_P5lk6wc|Zf8Uj#ICO|E54!Pa)__UdEv-T6Z6&H3Kw@UST;im<o0myqZwHy#Ap
zXNlxum!yANN&n}O$CqFH`UNXv)~MW`_X7!_xI{$mfW$14T6TMT`;iF|kPo^n!>a@Q
z{;#iGN;b8xiRmz2IxH_QXftzgRD*imIa~(+WFDuxYX<>=W`?HAr));q|2HJ+UwO;@
z91b(DKr)o%Znia}@`IeR@*8a$oS@8%jI0t+n1|<2p3&Y8)zs9SX-p)>+ni}?Oi4)r
z<-wTcBmX3!77T=r2%)2!2q7FkKR3I+#vmsj@dL<|LL?Z}yhB+5=>TB7z?9e5*RN-{
zq=ERX_uteRbd@w+_blJKSSdt6GQa!(!wemac+sdm-oq30zbe_bhrZg{iZ|JOtUK%u
zr|?BY<@UI?|NQy0jm;qt7mS99sZ`bMoph4eN|&a9>)7OgjKSUY#Psw_R8oR}V_%9V
zJ)MRU7!q>8%{?<Ug@;a}WqU!7fsKo+y0&2rO8fw_0fa=SC+GAHPGK-iEUZ02A=4gE
zmYNxiqEYGt>d;h5G@CyHp!Tf$`2~}&U3YaUSLcd2R=KGL0Y);fiycLq$=%H&79B0^
z=x9AWJf5&H#&B;IB`SX^3JMI?(D?J$j~r<seHD#&332b=zX!~mlpe&dUNKtER6HhR
zL`T1+j03z#@bK{NpO(O=_OQjRtFmAZcRgkC`Wi7bGU9rF3yM<crl+}%ok7Vb>Oxs|
zV+siZLVrluqMZE4?`YF1N@A3lK^YL0T$z%EDjV{>MOhUU9Ef~2E!KCB;E2e`%*;$|
zV!u8mRwim{>QrtAiR+ogMX!-8aZr{oaC;~eY2kjb&`4f_q~9<J5@W)Mxa<J`D*9V3
zHMPqQT;12NU$a8$s7_Y@M6wtUu-Z}l{P`1%?2je@mVD4Xn|1{XRGX`%JUu;ibaaya
zTD<NYT!+ZweF~A;jO5VV`31<yTY7o|)o3+wV<pYZ+&GCANWFfPGC3{sgEH%}W-m8?
zm!oB2H;@_N<ax&K;bAr&czSxOUPwh0mz<n@w~(t1<V2@kclwNj6R-%SrID$(#c_Qx
z^7L#vbr2Kl{-xK3M#4+8Bl22C80(9KgoM6+#;F5PHZZ#Yht%P62lCd300Kb%b@^L;
z2dv3}+mYPE1MpBKY+u%cph|~;vrVP0z?s_xz=@?EH<77c#aAqE-)=pyvS!7;2ABX~
zDAMD%A;5co2h#CeRr&7rs(bnq32W~FNdX`XL-rGdaD&oCBpt%F_D;v?nxjVaOl=Tw
z0E#<0I%>|~Ug}^zOD+2uO$n0TDhSDNyO0VB#G{~yuZ9~M10)Fn0RjI0h#151EDg=~
zBe38sl_D6d&OZJAPFzSxR(pPMumm~rZn6Cq<o!mnF{Sc2(Z7FRu%OJ(&%ZI4T5=Vs
z7Y7QsxDAW7YV82~5xTpEp`c(ZV2sM5MuK3z^gEjn-*%UflatfZBEnc(Tk|*@SM_?h
zb8>gT{qCNmAV1!o6g5^_$IZi&2m=xXMhRvDa5^AuHS9m|hC!OHsFq8b`2if2P1UKk
zN<KXUiPjGG_9+5hBse%m#>Uh|2!Qxy{4?wY7L}5c(kppawW*k{uyHmvw{e+L{Zg9&
zqOIQLNa|j*y9pK&fdv%>OvlS#Vw4H2rX#wfII}xNfszsu+qzEoyq5DC4;LCWjOU5g
z+uGT{<jmFR^0qMo1>31}l9iOC$cW0z>+K5l=>C|vxnthWcywU#fR2g^%87tkO=Q#)
z?8lY@!@qIs-iBTr`!IR5GFHuiN+f1JxsS*Etz)CN0l_<`bK}I)!b0aRJzE$N@l-_<
zt!sq&a#x+_D<JFHE{xdReBJHbrF80#nqD~|#s;CFh|SKLySuwccmr_8<?YGf&TBvc
zH`iUPZEjWW$r0w`qn;`Hup#geeqRfARMqc!pWNRYa61sN7zI~|7B)2Sa&d8ylYfO}
zWMninHiB)PuHx5wSe}VKdyU;1v%G><5)`6WuDb{C=5{m4e`5h?)<y(;59T$MwEC`p
zRYhfrhDnU3Kq2vwNw<-WcY)aNL84wSAU_eGbyF==lON5{C&G(+g)0ihCMU1GyBV!s
zoOuC1q9hIV`HsHz`$>L21mY>^<5-lWeZ48)6x17>lyr5n0jB$L%iRs>eY7tcDL<2y
z7T^1_vSZa^sm11pJvLB)J#RJl%l%03dILZ#^YtFw=kRF4!o2MDWf*(g<mBXv($cu!
z{BtO&vBIzV0Xc00kV?ZvNQcutw8|`6eFr`~DoUvyKS?jn>)~b*usYi~I`Z56{cW7;
z5fdK%5nAu{FyAs&qoB$2o_y(<@9Urrpkg^#)SK(l*EiP37Ys{BW}Yl?FG2pMp}F}@
z3NxSUa}vJgFSS|4I>ymMa}DpPM6T<8{jveG-Bd9Lw!pP_o`%ctjzQbi?P_7Z{=2er
zt+CMrZ|U}ae<Ih?l6WPim$0zkT@Y@4W`>}U(BjRwgrp?3p`nNEn#A9=UN95K$H%Ed
zNo74*GLijSyMl0a3-UF1oQhe!M1_U9cz7fWy1|ZiZ>Ew0f)n&6*>DZ8r!3F(TL_W|
zKo3Ho<Vwf43fg-Fn3$7|ZqX&y|2Dl>1#4Vf<p+zT$IBfH$_22^fgI`E{rQjui|5(f
z+tt;$#KiN`+8&^|6V%daXe5-Y&7E(Lqq({QYE<=7_bWi1fQ5yX%IktjNGJ&f_OSQ(
zgvHl4KnxSB&B@^ndt!Dj?#17)_U7uw1D~0thKn?z?M21wV+6_Mo<|w3%k7~eA_Ofh
z=1-qqi;7Bm*_clh{;YitNFm>=7SGjwF-|568chD;lH7e)=OEeSaWfuE!eLdE0xUX7
zqjDH;rDIWcX6C{UoPy^i`Q}+;#L{Fj*oZY)OwzA8sCxKL6lDV6>fKIBb52(71{de5
zfb$2a$j_g51^F_KnfRY>99@rmU^U_4K;z;yh_4c3=+Bz%veyT}OaXAF*KCnrqpbbU
zA5*X*KhnOOpVOnFjFGa6eX>jjcxN_|frf_jmzUQ-XNdi+bAKzK90@_jVI)F}PvDo9
z2!sVj0Fn&gbCmcXD!S1apkD2CFl=U)1UB*N#)fZXq$o204YPQv=eFamvy{7g&-Iy7
z!T7JL!l>7pXrdSW_xzCjyu7>y56$_muAnxM<_fbX?;1RKnc58agA()~v}%5(<-xd}
zWbV96i^avo-5nez$_#Uz8xT&`i2+?GE=&25wdZLnpspjwDGwuc;&YaBbqom(e#}Ix
z=+ugg0|+5&a!oRZHoVGC(Usf9b>wf$Yogl)vDbocumA_i6Mz3@mlKVv8)`~Q&MHeu
zTieq?0aiy^kL#_?+Am0?+^7AAOT}R%TYFP_<gX`A^|B6TqE!|T+(4naE-tnM9i0cz
z6e+1-@>oe=Ti()a-mZj+kdSCgOXo_(+tqnJ$b^w#$d8<CR6(=#^q^Vc31-VnAEDR-
z16L&{frk@CEbHs|8-uv~9vWE%nyo8LC8CT!bF=Vj>j|x=%d|BUCkl@~#eMtsZEkK3
za5B+r)!00;v;+8be!f4HF7)~HYG(+cInp1u<8}~UEvtxv)$w^Gxc-I1h+b%ZU?kTY
z4IvCa%CG}PH~*ax0QkBR&wxL`tb8<=uF3+#j<a<{!C#MGQnJ83`G6&qB>TxTli%Ji
z1wazOZbN_KNelz^W4u5EDGLf2uJzE&)ot#%zk9Q^1pTZopWbAXZfavAB)xCQ<D^2w
zB?$HR^9y~if;Zusu-pCncaq0VR>>DTJ4b+Zt|MhP|6l<(%24`1GSByK-vIu)Wzu_!
zN|d+0tf02|#AWaDh{Swy0hFa(Rxfti@6x=Zr^?ve+{6$T{zF0ntpH|sy4+Y43I#i#
zo=+g_4Y*a#++K3=$VQ2=x}Cebp-v8jMgvb>xl#X&zQtU&Z0g7KbaXm8gNc$7CrNQo
zeCx1dq^v`quRM(M@?`?A%kdn$rO6y;+dwjD5`RCV6kt*xoYV5HvMhM|)NJPIG$Ff9
z4E=Zt$6&)(z4y?l_;`FA91*BGI%%MkwCNblyk1)jOJA{a0k56i?z{Kqbp*g(SUT-8
z^4WMJqNK4~`YWs8U}J-f@&Ov)YLlVLsF_j|Wn+VsRab|Fg)uNNsD=f*W8jNJD=Y~C
zq<xJimXdNX?e;PwVk0&wDagnOhlHdxA|mj;>Us*Fq-OO<N09yYYZsSNYUwhx7wXYb
zV?_?;+hgL2qSVVvLXdpdt1{<~Krwob>;4`S7f@+7UeLKWTMqc!T~2AXjtq)Rna;L_
z%Z*2!7Ka0{3RqZfOiZT{5ggvrkYzNarSbCd1qMPOk<0p?8w1cR{!efPIrp(CR;A9R
zoSd9@%N_Y>#C95$J<Sf1b)AJ3O7qk4*JlQP{v$)_6Hq-pJzNHzx%2`z%M*Z29tp-%
z`CaTZY!2zssHxwZgIxU)4nQg2TUOQ{O1v_VB3<u!mtDoaac&8bn5qoIXO|KcrCuCn
zIyh)H_bQ?0u;MHzDCgr1)Ts7!*-<f4Qi=|j|K2SYAqEp;;jGyU#$V^>zs;3htSner
z_8A{5qu0*Z-8~kMJl+^6Qpds}YDlYyiecEe*e}AsQdHOI(J$W$Vpu=ED(?z=D3DHc
zdxoSkx!>P2C`&`z$L~@`#u3}QQaUz9$*c3HHx{a-rL_zFd{g&F?@z>=L=H;@-KHeK
z`LZ{=b!l5?Ck)u3+m?p*W>>dNcemcU+>p@un_ip@A|`T#x!q1tGaVHN`~e)bl#lw0
zFCt=$;1d`aI5(#yARzFVbWB&|<YDe@)UtAPc|g4X$=tkqZ6YzQlG|R+>x!A4z8H9I
zH@3|*szYIjNu2Vv9&QdE(!SA?B?VyY&(3rk_|@m@?Y9=n48hvIsi7j0lCoQBV2nqg
zQ)mB{ma;v(EuGIThB>&TgaLy|zf@U6qky6LGb+hkK}E$vkhQ-aj&!Q$-`rGK9xtsF
zWbd_ZRbgt5QCh$Y;~PgI5?O8&Rhri%#C&y$vAD*+;UF2~1=QF$U~0b+hN_YBZ70Wz
zL-$rz&fMKi7I<+z?k;O;8cLeD2UfQB=OsHj>>v1D+yFYyZq$^*#wrK&idFvp+1YEt
z9(VNL(W$Lxpl;_w*3IGBYXz8=v)ieB=BKmP&~{F4DkS(<@|WIEu7YqJcwNda`@`FA
zF7~zQFo$O<&8(~poepCB{A{(gV6e_GYGdQ=8Rfy=osD+r@mXb*n4O(nKG=~Qo=-^2
z8smv1>xn5w!ZgD0GfBbX9hH>S&PcH_xbgXO)M8`7cy3ttaOvRyF$6QH;~Ii7PJmu$
zWls3zWPNWmmj?&iR<(rW=5Xm%Y0X(6_UbA$G<24k<$##)`Vh>rz~{)LW1|Ka2fEH%
z3Y|-B!dO(B>l;`YJDnkpCmUGd3%~Sye*d0XEHLW|%?Q8>$g9(?Z{=|+qt~L@oMIYZ
z;CGoUUS8HO7*Qxn;hU|kiHeRLUF(e|=8@PYgy`%ZJd6zNJBE?GDJ~Yx$Y|~9!P}ao
z?9pjX!RLyQk`8HXkYZ+Gv6`b{XUPN@NK+$){(a*4&4C}EYkOacG+Q|$BI3ft!~pHP
zKtB7tnVAGw4>7SpC^1C_Y)CGqw3I6-nuL%L2a}nceZt_%E`d9GcNf-`d9ne}sz!Yb
z4g(b`>c}|T#qmggcOK!+Kyp~Xj8uvg)XdC``(#~S?D-pN9ul5TQi1#H`Fh}QX#0-N
z!Nl|e4UL_NDSQbDDeZ#=dx2tZ_4m(M;Y~q7j-CMHB5iy&98UGZ3=)Rs?9CRSleiDI
zF1PH+tLo*->^azyLz0aLX{n_#9d}}E_mk|Em#Qp-f&#foc*SyZvT{V)zPklM!Ny3E
z(=%yn*w?}W#nNJVFy&3g7&5NiOa(b)WMYDu+4cAwHDi1{+~dY7EQFYpND8`}M<dO_
z0Re!~;ji~V64zJ(ZtoD|Tizs?r)JabLw-JrnMfxdftk8S=k=9?q=Y`m^H!FVQ&QWE
zj23n}7>Q#_Q7uN%sb{BP8MR&!S=ksg9fl<E)k~1!33%RmsQYiRQY%aKfvq|4_pI5~
zPKE?G?JX_>G6mO{4!bm#YeOfeh0oE@_#LInGg`l_i1hM%)QE@(gb`g$w+A#eH9h_M
zVP{fTvp+#NCg%DX(y@Vo3lr0|$8|&L&x5AS0@a_=gd>x(FVQzdL<?>%<5ZLyq$KpS
zF1TLBPtDln=Zn!6H@UEMf+a;X*Sn3$^R}kHJ3PV4DuY@UMnl0CeY1XuI-f4&_wnQW
z$cXco^`YqZW98I*ZX%f?!^<5NpPrbv`0)ArudXTI9vimSevyi$x1^FDaXuB>+t=>h
z8by|t9;MgDjE><x-4%z@KSO+8R{qlnS|K;}j&;m+ni-<7GoyUE8UW@|=yxRJmZ$xs
zUQ$)n8)5iDv^cT7jn<!j)Vp0=ym+BriTHVf2K6}^de{*XnniK`w(Y&U%49LV+XW`Z
z+Q#hnZxwfEs*vUQgyVt&N$bU8;Dv&j9O%@!M6x;VcBgtCTTd46o=vB>Pjb{NQh$87
zA17qbtF6VrSnMsh+*XE2G~4;vI@0e=GI-8b0XN09oAXdI5AcB%kX2M}_2HM5W&DZ!
zGZ_n;D$>|O_OnKLI6o@_ehsT;Pc~$&dPlyoCO%-YPL8Fjt&Q>4fp44$jIf)hCY;1(
z`-=#w4i3}VsZAvnZerfN#xK-cn`5*GHtSNcLAa))IhW6nU}C@AT^8Hhrzk{OdwN4=
zsyT?b6mDI=<*a3jnxT{YsN?oH$HxdyNl6IlM?noKi=>t|>LY{9A2jXY0YJe;Ez3MI
zLl<sx_&0dIeg|IhdZ(nCg+-H^QZ1l^0Z?x!C`jqVQX3cm)*u+AjEwNbK1TRSI7JS%
zj9iib`{a(GGY^mNIM~C%_zXowqcOBUnb`qL7Xfz^m#6#B`WqR(#?qCplJbhIZ}3Z(
zN5Fr<V$`>(y7NPozE5-w99#mnj~1&U#k$Q+o_DVPdj2T*Bjv`j0np>n*5#q06(b{2
zTG~L_<OuqoFjzM}dttt63DWwy8!_>UmKNR?&hm0`g_XxdU$Dz<&4RRF7%|~5i)kT>
z%%0m@V}E}qr@fbZ6GcXQ_>z)GbG28EZV%jFsM;8t_*}E>YB2*=?;G8YVq#F*zL5x!
z>@Od8;<UBO#3xcaZ2n%|;CTHI^hOfRanH4Tob4b?0M9>QXxt;e;C;B*Q{w@5GxME|
zbt*cl*DoK~IgC^Iqwn8x+3JT?R-T?kYF4MrPCX(kw+Bx^r~ZlmA*?^qyh_0B+yWOj
zwz2s(V)^gkqAcq3Pv?6EcvSs8JqIJ%2_2JzNde{6#&|i-b2a_XkS1uG&Gj+2)`4kH
z-GP6&sj{p^LkpJcW|WetcjWPUXuLc!cp)EJ!qCQuy|tBoeGNJ!%TYZ?%%gCy;E|qv
z);^bn5kxpc$d=L9hnbP7>(zG<95$F*???<N?sL4MkfYPn&&Ap&yR)%+angzG4tq0a
zjfn`CHPC_rr0MDL`gi*ajWj}SM4(ouNVB=dmO@yQC(Tk5P$_nF=piDaw5=i_-j7yU
z&R4&Hk6u{7cjs5s?q@`y2!|5l8|LKjTaHpuu_|dmlOnN3b3zuqpJ1jRj@S3}Fwa!+
zIb1TfF{1PG`FdV&1*9y@BH4Au&~;s(26&#+<|r4`rM|~(92$a|Oor;kHG|@sKBEJ`
z{gn_6ywSkHtcRDQ+w2Jm?*^_-kifvapI>LxpMXtDy4W4U3dJ_>199?+i3up+e^XRZ
z>C`V>kdK?MEmt&LlarR#CMDg%lr}bwe6KntBv`jlJNjpMxOGu?1|AAxg3zk*@zqVW
zgPgIOQ#}FqJSt36JOY*Yg!NSJ4o=Ro!bT>GDFe|iyl2mP$%SBI<z5f)e0+dr=<h0J
zULKkMThGHZM;8c0b@HQDF@Wv(1njSySJwM|%gUu!mV^CKDKdz(=j|00ae@rlK$VAn
z#(i$8M38OmoHx5A<(G9iHN5#6`AR*Zz8!OMLRD2qTPAGW*_yiT%}w`>4UUUhYoJI5
zoxv+zVPVPKYz2kES(&9Hqo(uR+>XKfi-Xt!+)i5xiZz4X;lfu}71p};X8C&!E^NJD
z`uYk}c$r*xwjI2yFNpd0!2Ot*BX~Kn_4QV<^w{j|R607)EK|L7gU|LGefW_TtSc?P
zbbP!JXWsry(XD49MlXtru(F?(BO@1t5*x^*+-`u4ESu`)BW<j-G+m{K8<wIX<oM|5
zs%oMd(FFe~D(ZW}xbdzgr-K_AnQenkC9uqK+WdVh+Lgrb0fQA46m+ZzAI!x!8cJ?W
zl{zfu<mA{c`CJWWQhE#_2PyBa$J7{n&#D+p7ezP4s0t9TU;WI+exn{n(#{BNx0pT8
zO_+%TFh|T_j<eLhp;r^B?2q@8+tpR$ddGl8tMyeI2#?bM30oixWqzKT5D)KsPqSQs
z7H4N?2P+&<KPf1HT@7y6xZ6SAh+Ic|5{RD~8VWz^z(XTnGkHP+v9wh1M!(yQYHL-c
z!^PP+D~sD1M^JWpf%)UH5QKR<jKoZ>$J6Ed4h|5^Ds$tbzh84II1*s6iQB`r=4VfH
zx;h=de4&7JaedL>=p%=Cc-*0&eb+YtS-J!c+=0%#LWG28FR&J{o9pU2MzVFx%szwf
zGw|`f7YU_=B=UO{pIm8!utb6nsETiEen|1W@_YL9K817m?5rUgY`fn;O$i79T{&!<
z_5O5!eimqbd;4l#DB72OqUflhzIf63d5&sRA%1>zmj?FAnIRBm&Y-2Wn=G8xv%LF5
zojOKaVlqju4k`8A@yP)JhZ{B8I)|Y3KCYtlVc;i%EQ><NSnK2a5-%+J6TrmyY%L4*
zjf|Wfmnc3Mave&$T<HoB15xXltSk_Y9v&Wc@I*=GcHjn<4sgH1x?5V7MkjHZ;OP7}
z>kqD*i|rbU7MI$3Lqcx)Q)Qt59PZQ_h1l1MfBJNbF0eUYJ8FA>>$;Hh6S(cF2Igw3
z>Yb<);u_=;mTek}b-m<NR2VG@=yZ|@U&%2W1=rPaV@%7WTH%uDDCrvdPX1I(EMs>3
z@F7uyx|l%bJDP)fsSj*-cUNY8L-HUxqO77~ccGC712|^EzL?+hc4~b5ESX1ScvT52
zHj>J;CxzKi!p!Vl8zVSZK@*VQ-Yov|xZ#-zH1AF1W&_>_r`KXjb_^iPn^Ed=rV^5I
zs#$GTf*k-A4I}Zm(f%?wN7?B1T}{2Q!LxCnTR?z{;ebysI+msR7UH3<rlzK%Lgk?i
zNb2jAKtj(Ca-nv8rZz?p1d@@E$X9o0kWS>Q)Brr>XGvVjmNSebBn;>zM2w$)gM)`o
zCJN&>_H}Znj*@wDon2qT`*7e`O&^oe6}fWoxE!wnH%(yROOV0is)CCxOn~3fY~>L9
zQCvJKMdzc0L`>sd)7~sHGP00vb@@1_`=w2_)%^Bd7+l_SJ!$ikM8Ni+WML`r8Ck!S
zGpU!4%b;_%V0n4j)5+$!dZR0Fb}%t9vGrQx;a!ezZ|BpE3?0U&kWgSYu3&)>mkX~B
za9#9AfKz9*8(ozN&Ox@vV-$uwyg5-Tu}s;P{ysrm7#vcdWj3x(c<M_3>w@|CJjY4K
z7B{oteIwHEJZoJh`5m1`vl4K49*~CTjkdn=Ms|4k&gf|6-26^4A^+{2o~h~ZPQXg`
z9+v{3>Zz0jZdo@sw+G;s_xJZ#S63gI@cj8hr%_{ey75NH)!8kToi(Y<b_MKAK|nMn
zKAsEM*t+KNWmeOLhJc3)F9)hkMn~J}{oPwbkHoX%<2suqpRZp(KFUSCDA#znw)~e-
zU>tgGW+azG{(joW^tH7}qk*xBiT$gIJ4)?(M~~CZaD!*WSaA8?v%gO?{Lx7l>KF9N
ztpNx94KFW$D)-r*GT_R4!Q=Q1D*Sj%gChb*>u9-j5&f^3{E(2^goM)pd`~bN!Kr#V
zI_7!1+mnM0wsv;M;N*sji-#ZG*y1uTbY&8<I^CDs1E^l)kc^K@xm@u~lo0UWF6(+R
zp->d&yXODA`l<<-dk6mf(F+VpBx`j7n@WJOb|y<o6f?C!cYaP9ED}T{C-3d=KYHOo
z%f(|s?iMk>`=t;+2S2~(_Sny};-2nqJb=r3qTfeGM(%i>RGRlynu7?KLsU#m%J9U(
z<+L#>Hi+$l<Qtj#r*Pe%>R$2oi4e0nAegU6U*R)2TAmZhf-6<C(9@IOxIJ$UHxIcQ
zr_3M?0%S)UYis=_)dnjDe-920mX{$#CC1q~3F9CS_c)v#mCb%n%K*gY9=XP)W%o`t
z1_9?TDj`XnN}?F0>%n>A95_snS~lf2l=}#zzkzcN=aa~vcxxFN8&^m0dl%Rm4W=~O
ztx42hu)Vpr2H|{gf(ylDdtqQ^8TfA$4kjqe_r?V2ws3HAkjErJI0^Uwd^ord0guUp
za5NYoofQu54M_h3<Ur7W&qsBOp1(ro<7)(v=<53V-rCyh`t&Dg&Be1gWe^ka16fuX
znQ}P5Ee~>J0z(Dgf{1LooZ1aOT)q^|YfKE;YkR$D>Q~P}h%oEqy0##nmX1!^(>OTH
zLSjH90sIPTM$~RzwB%d-_I~p9>(|Fuz&e)6h5Ug+3k0G;-+uf49h}~`3+k8T?a3sT
z0B403z0*vE>Ektba9Ro&tjo0T#~&#`_R#-#Utnf$j2=0Y3no-Ok?1cd>eXdD)o5|f
zMNo;TDs4GIzE9>RO%#GBVZTtNRqQ)Dx(;GzwqG1(Z3}rVj8ZNxwIDev<C*+RP@mkP
z5B8gHxnE@iGYrCMOxWKVeyaQvter^|5aR!D`NFpQ94WhN4mNPF6}k*JO=LBjw&zJ1
z#Ou#*(tilPz`!UC;WHW_{M<s7ahx?9`!ppg?i7zdUzO0IqjQQkADKCbtzX)V#CG7d
z=);$zRNQfuluUmIrnM#mJ;~Gcg;!nIR($yKZ<1%d4Y-_JP@JB{NbI88dVX8H+8NH~
zM!V|_-9EqOnQ*hOxE1n$8&K=G@L<|hr(6<lz(+eC0Q=MI^f%0lxgA*P5xYD0EfY7i
zrG|IZ?-UGhx(@JXSIXxH%diIfF7PVeYNc&@C#3#uNk`442<*-3DR={g`W%M%&TXVf
zbO+1{)NW9_t#ky2D%{5v8t9@-+vr|i4@M+Px}mFK{7CWe>fO{~8EE`WpfLp*pC_#S
zTL18ct9W(1BujyYJbH>ys_LMS#CLMRSzZ3sdG98D=0$tJ&##$62x##mY1TKVciZj?
zE`Ls*Db}G+o#q9?po6lO+(a=K*SGR-AgwOSp?Jz<HI3~9p7Ni#+8A5AL#&GU1#}BH
ze<UPwT1k>_U|;xI)6()kxD;6JK=SeZe?He38L;|e`Bgu|oW!SFvEYHNA{niEFGF*D
zL^vXtPbe7iE}Y5;Pkia>&-y$EG;mvYYI&@j15a<WnwZZ^`t>Az?o&6Rq%s5VKby^E
zZ&2D8J7SI1xQl8S3LiG*z6#}RW`&h@O>kng0gv})wTw~uZ2_Kmi*Dvj-?7(8?{dRC
zj(73~F<Jh6G&_6qib1fNy#_=)e<aZti#NqF!ch@FtO`>+&AJ$xiI2mgs!W1(zJ1#{
zBVFyLfb^62^1a*to>TGr<p<QNP{YX<mW_~K*9R0Y@>9R@IvkRibW-VWU69VP(XV=9
zC>{tRbkdQY?V{#O$n<eUK}i<QcLp*!R(b>SzG`Qs#jSnAsN65NS%)*g!3b9|XKBn2
zl*n0l6g|Gkry-Atxg~ZY2t?;SOXhZ<#ESq$%n{A4)w{->Nz0oDv_TvUpfR%V!s9wz
zrmql-q_XBkEFx9>ucm7;5xU0u;+9ueW64D?{MVy%Z@f5u&5lKGVPbBqzs9R%j=4vy
zp~^v|b8ohxo9_xG-IyOv%b^Z_JNT!F(aaDskKm4_Sk$oVZtH=px!~#IX8nVckE!-=
zXDU)Ec5G*|&QSeiL;-?SV|||A9%^XOVBJBNT3J9`8EoD>6ebY=x3l2laU}c=xt*_M
z6}fF-Vn1xm&IY0|QqM=p<R$b+%2{rTjbqjpZd{bA<?AzqIZJxvgNVA0aB>DnF{9n=
zkb(qE=Fe<q7l?zW+WYpHTdul12Hl$|{PWhim{6F4q2uSs7($gug<vCSYNFTM#eQ-{
zXCl~@a$ZLpAyrO1<#2Lj=*zDf-AFF>;?iX}Bl)dIb<*OHXe9g@De*hIZdlpDz83pK
z*9}_1zysIJ4`IP-L)`s`-F^K|3XK`-;*;G)tHFh0^&8@ME;`YK(RoZ?q%`oc*VZKy
zlPIsa{6um(-yL<B50}~PjJCj}q3z7uGxOZkbc(G^rj|-^k<@(sfhS*rP&9u|dFmZW
zOIIF$(H)0-e{A<YL@I>qj7}-wu65*#&4FP_F9f1ry)`kuYIrRv^5z^d(ru;wt4EB;
z>D5Zhov$J>8!Qi~#2YmHoYj_zaK2omslw)y%ef4zI$GdTayqhSTEPIGqVfJxPQJ{i
z8_HMm0e9V1+v7q!tc+S8pvzflvk3Ei+()m8l}_>->hey8ESVq>NY14T)ZY@x4Q=vI
zqvr5l8g%RV*zZ?GI{WPJ&Y{k1s$uj;U#q|1L*6yIos<e4q#zc$;=bb^G5WnqoVv@p
zC59sen65m}8>X6k*+j9=MgBx%e<^k)Lg&BRBkP^K5Eyr}&ub-DB(6FN@>riexGo}H
z`R!FHW$z_z9X6h>423G;V~2m*pOPVbY;s7BMaoU9wZ-*=`8zL>ji^X3Gmjk}LDgY$
zHkSUT$gX?0-5S~QoU^~eS0rI6HE!SezN@vx&|kZpMonjk$AqEZ?VWmff6wPbO&NDx
zu2!t_I@(&E3s0y%7r<*eqo$U*Lso|CHR&F^-!TqOz~LDdKm0p3o!>$2E~QSJ({X^>
zl3X5}U(t1GdB6V>tPQBV>+XCbbxoN<MIEHft%oOG##?m=4^{7?U@*09?VQ5<v9nR}
z1YfE!x@5K?GPb|nmKV~Kb2aTMS10*%zwnf|aPlOO&rZ;LCme{`dq({q?vLqYg(VKh
z3DF}Ea<spPZW(93(e93-g1^#s)6kzWZ_i7?q1&TWM5nqKxW`)a#YL-ho!6jz3#;?I
zG7A1>**r|4CB=E5Hg()k$q^S(ljfZm#-eUHFB-aSdcTb{+<GC>v10I0i^q_El~}ZB
zew{&XISq83?)C>pSJI;0c6=HY-LJ7^fjc8DH#B8qzB)%b&MIhg<b{c~XrpQo6_|p4
z+mArSJIq_+^bxnX)BXolK&jCGKEC-sedt54t;POk$i=q<y3W^EMVIcg2J$RWcsOPu
zeMA#31=4x9aB3H?+U-Kv&{Qir&2zYZ^Z@*{Yz%6mgq!}-9j`tT_D_dX>O1sr#CL|I
zK29}R=Y)|8+$9R^ufxn=Hltj7Y*omLNXK17>Iz0v?u`kiR&oU4L#Q^EIj~d^R^RWM
z!{xoBH1dw|e=CVLf3enc!xv+Hr}uQDFStZzoryAZn%!~Cx|Yv~E@^6g;Udvm6$^vu
z&4W9eO-}Pt+?!r1^*QywjYFvd_7*#%l3p(hXzq>odS;W|5%Jz$&%)&q6=wj^a{WcU
zJr{V`E|JZ+O1)*KzRHH;)gOcf2h^9)<K>BuNnajz0=p1M|6to=wl=D~qFLTOO3pL1
zR^sK!SLqljdGMxu8DW^hasG?4>M8CMIX(b8>0Zswwdf?mU>%Apm_n3d;(fM6YZ)J*
z&>8Kz%8gFU07d=fcPZ4;1kS<2P^pjh*)k#|2E49!5sTdJ)*dtq+LGamPd+GvtjUTG
zwCp$iR_{o3HH#{0qR1(5Fcg*5J6>T(Y733pw_i|moG&-J62_Kv;ZSSW{jkW3x0!iZ
zyBzEvnnf6zvl)Am_f+SiX+9&A)Kx(Sx(1v`A+Innx+dC^iS4<3m{^&aSNX5fQ!h$}
zmn$fGtIP}QY#k+T6Aqre#1}?FOLLVoK$JtUIQgl`ejP@#d>j^}uJ_Sg)kzx%<L8BT
z?G9FIe6jW{muU`Evh$#M?7B4~MEDDXwDn}uM+W9$fD+@xp&ylQyS~mhFF2PhPVl=6
zBcZ2hsG;~u>VvC>Uvn(}UwHigRA2vhX2v~fwl<=RLS{25_&l7rune?V@RQH~1wiQg
AVgLXD

literal 31199
zcmeFZWmJ@3*glGapmcXRbVzq94Bg$`-3UksNSE}`(hUOA0wN+cfOH9nl(e+;+4Fn<
z@A-DtI$zGW!&)pB@XRy&+53*`zVGXr7<E-SEOc^oBqSs(1$n3@64K*nB&0_lpFReE
zDcwc>2!1?smpAZ2Lc+jDd_S684!%V~qD4}GzS8!~-CurfZgk!y@b`#paVcd}oWM@#
z8U6F{cWds?)W6y@SO4HqR4R<`j=yN>%A6!wlu~+Q^rtOtg&2MLM;giF%(D2j_pk72
z-{TI%G=1JEv`}%%_gaxmEFEkqna`kO+qk-Api$VS%=sX<x@t|Y00FCE#@_c{1wWA}
z!_85_uaKv<NJwuXVgLUx|6lIGSG7wl<@k}!#Ws(1@6{(rNV${x8C|z0BSh%^F+x`g
z#BjH+^I6fO&rO#7u{+(#sb9uJ?oV@EJ7R1mm>(l?_2<7iY>&Vo8G~`JLvwHGF>E1*
zpX?{j=gcIRUwh#DZDrYYpOtbR-ws_CkRp8;JeY4-|CUF&Nr;5>67LlkG7=IY9t!XO
zyI)S9c@HsUD*rjB2|Al7l8ICOF^W`jc6q6G`NI0g)D-V>E0Q`PE}ZFy<ipkYPWKa`
zQsKNY$=fxYV(b91{1MKcK*K|(R`4uSJKn07CP>PXD5uu9`*xB}_V!%l33mDZTlaT2
zo8ci>BWBsz*~23vZasGw^7Nipo0&}bF;6gvoqx6+=X#ALn>7X9TuiUWGd+SyiHnQZ
z_uOAsgd4TGz8Jg3i0Vt0x9&LljY3XI`LoGx9PDZ?3ND;rc4Fc<!7b40?)qG{+zYH<
z@@_{h#(uRs=<=u6`1SeT>!X#s2|1d(lgyq#@Q(iZt4(%Enw6B>3P@$TLX;Obcg5JQ
z%N;Y@rz#_N8^qXAeQSTcSNorx^`g<lyL>?+bt3l_`s*jgnyy$rl^2~_Z`69I*K@5S
zVJEa6&l~(VoMyxCm5hvxoLtnh>!qI<dV~~*M1IKq)p*dMm)92jdC}_qMXPDO;H3^v
zK8m?xMBmy=TCUw+iBzIn@GOF)0kt2AKB>epP;_<f?bEnb_dcwihr1)DAcvk+8uSRp
zJ6|CmLI{krbH9<q)bCZTE{%xypN_q{rFJi-0T1w}HQsSCEnQuK=DCiOVYd3uC&QKE
z+sy1E5*PDU<B|`7{{HQz^_d|Nhlht?uLn8pU0ug6Ochc*bHQp7u(tHs8J|k|_%si!
zv$TZxL{!2a?p7ayMzi@zjNN<bS8l*QY<PjiNb>Qejg$Q%)QB$I*E98(!9{^|9JDOy
z))_BMOiawo@Q2*rb{=+ABk2`*a(@2%^?L%v>x0E+tARw$z%`7M-o{3rsJ`ct!4G$r
z4;QUHr9Co)GFPvRjEoXfk(6s2dT+rSv73HwD=GPQ;H!4f94N6Gii~z5EN#KM1=jAf
zH>+~kB~idho}hX?Spl=iMPD4ay}vnn_%|R~S6f?KQQ>zsAt&gvs6AM%M%23GB9N0H
zTRgsfb6y`p71tmAhqvGgeiZU>BOQg^zI+HAth}j7E7*gNoS!_Q)AwL^cQ+Oa85cWM
z{+|T}!E*%NB=%~WYILgy*0UC!5dQ0b0U@&en+ZI->&@%k&Xad2k3$0wxp1F;u#7xr
zR_7WyWu%~>(2AK^S#jmbLiaka^y$MA{K#&krRB9#ni;&wX137bvlqwN-R5`v`zZLb
z6S2#=!d}2LHom`A;_UJUR?%cT3Qo8aY^_7<()MUh<~se<wNo#XIz2v_4!7HKhYt1@
zOFgXBd4Z?U3pgQX@NwLZ93)x(QFfz@hH<;+=4`DI)3vNZN?5PCL~nd7+4oA^AozCv
z8&@TzxVgPe|NFzvqu}!yJwdmXz|C|$iN9ZvpR`@4@n$LSXsOrr+y`9szG(gFIr5^l
zq=zTHIY}HCQ<Atqe(L^W^B(jkZ*KP2HFqx3H2N3|(yQpD0}eJeHWrqwlh~_19LY}$
z&g&s}$#im>W>Yp-dhiB)_I3Jbjk>nqu?W+)D+s!Hr%H_;0o8!>!Zf)1{0NIIgKrL6
zJ3BgZRkvGQmby>#j&SkDPu0ZFr<Y%^n!K$>&Tag<#x>w!RsBi3pAr+byd*O9RDu*U
z@!<s(6;&}uo?ZXvkcayvw;u6JCsZ`F7sMm6L~JWTm&-BvX0C5u_G3O#Z<DZXHWarT
z%e5KFz?;aj%Wrp_sluWVAl%<YV#*YMbAgGCz2%?cE@W(In7wP8K_6pd08!L|_uMQW
z-Cxi3Z21#G78`9w4D9mw3tQvSx9s_{?Zg)s7e#zs=b5{@x`>E~*hdx*E@JNePjR+h
z(yBXaO@9zl=@YXxaB|vPY<BVrd^ILSclCFzZ}ob%RrPP}Bjuhb>VQ8=X4jMD66ckH
zs`%JZ2flAu>pq*-FF2Qdifu10E-os9JajIDH!cUU4dI+r4@M%QqEq3*l-LiVBaDfe
zHNyL=Jr8Y9S}EaBZ8Niwvp*Vf{XF-lY6@I?=hq57RH{@{#BesfI{00P#&@!z%au1r
zF`rr97STTaY6LP-)_5S}_$hdz(9qDyNll2NN~TD_*`GA_^HXr8`0d|5$^i!aXZbX9
z<%%HA=~k=Uk;XtdezgVpp9QKh!}fM}MMOjxKiMhM;&L_a&zs-h9fkA+obMh7_Tb5A
zxHcdSVm&C9g^q8hO478tt#*6wOgeK?Y;S2-HuW-nAIlTPLY2NpF5fg)<+l;d8(TA{
zrc9&o6crIUG0UGi@MR-JPgm5@)a1Fput*xv($q`|u4c?9VVqPFAD$?dBZz?>&IwT_
zPF-JD)T`COn7r>_(tZqG|BiZ!$XNvsx7%c^C!3rhU^nUR<Q)mAUt&MqVi71jN81|9
zyBRUN=UHl+&)rQszuv1idvnlyh1TR?{;Q?yjTX_VJD1z)n8};k&s#}TJ8(%DB1?i~
zml_Si!pJ71s8}v3(XPlnKv4#QCqAO+AD@7avewoxx3skMDM+-|tN>vr9pIUdb0u+~
zYI7@l!$0m878ZgK%3#&7GK2U^4~Ew2hBON9zGcyfHun{xylG<oGE@GO2E?}Nx(>&o
z6b8oB9grd&+}xOok|eJGd_Em}^D7i><2e^1nSPDHU}I&i>$+H+{+u>I6Y{yNtgNu`
z`ONE;l@*SVB8`jDO3Q|Mp>fm#QpDwU5~o?W;aQhAN>41jxgLI@5hmbI{P>|BATBPv
zyu3^k{BO-B(?s^}F0M%9D*vc`JD*C_cf(JQnQ(H)dxwPxJE|G_+nNnn%cIU^iG1@}
z9PG&)w_%k9=qC`RY3~0<t+zvgB`9bH!Y^UqSt;B;bFUu~y0*TVo}DwSB}Q&qYeUC_
zhukc!X0RLZfW)C=Mfhu%=_?40-RHG#?Z$N&9Bscq3{1s2{MkNb&EvAvGJ4=!Z_;@|
z((_N|_3PIzE(fE6tIE&^;F!Vx`e<C|KU$3Eiz8rF1;N&QE-UU+<uhk@&C{DZ%7K~L
z*)1Blv5+$<89X6aa>rZSbWBmx&Tjkm>h!b{CMqg=)OjS~k4@ioGTbN~rD-)R;QAA`
zo&Z9N56Fz{If|9?^hFFsL#mmJJs}VK3r$Hkjo|5@jBDv-_B&@7ErU$Qpm$~E-?bvR
z;zxIGMT`4`0skv(%^_KSqh-l;gcQz^SzTYR7R>IJUpTeHl}?sejT_$6eLYhzASk%{
z=BO)TKJM(Gb*FjmM5HgC%zLl4-K-JBcYpK6klTR}vHO4Po;fhh%Fhq?HwGC&AeUrt
zn6Pr1xsQuqPS4JwJ`M$2zwEo@a&k|MBk}JSh&EGdFe|Srg<LFEM*yP4w{yM9tc9I?
z?d2)rN-w#`7aAji&*-iXU$dd2UK}h50;~cuqsrmN>z^J`O9$QijU$n*q^lh>x$^s-
zoiWiCu-7a8OamyfV`F1ivg^akozjKxXB-)d+=Xh0WwM>^4_13Z#&U&?TAURm|A7GT
zkh6NrIN(u9rwQq0PM`j%-1&M$&i7z3{on`@tC{lsEYEjmPLec|qEhKCl##>f6;Q)N
zZ5A4=I)A#pW`!Y#zU?&!k$YX^gT_L&Zf*7tq`{#a!K9Pi_vQjLmki)&PLT7rot&KB
z1O%J~bcso(M0a#_Al{*SL7R(oe{avI(K;ruq_EIp`k9xrDt}B<uR?88Q&V9fz){;x
zW?{;C@)c7T4ela&hDlLdqs1pDkuQIR<46p)5g3Y5ha9$V-`{Rah8#GVjc}Ve0sA@v
zj@M9NWOP1Jv2*~iBXHi}9?1a_V_+&x>55&nC7P_1*wy$k3TR$NMoV@eV$0CA8U-N$
z`$>9z2@W?~LVrbf{+MLJN+7W7tHAEW#6)Mooy(gkz29-2#!$W%%M1f{;`%uwP0h*0
z#cM=i0IGo)*<#{21|oQHPlyJ4i}$6gHr(C#3A%0mb^$;#QpQ8J#W6J)Lc*1$C{JGf
z+1X*?<itS@ujCxDv9ZxJ_UNw<xpRUj(&LNz?pOTxC7<w^s?+~x#b{+BrLgCWj-JXj
zaF@NkJy8I8{xr`GpHMTD;Gsy6D&uUeJ>eK0sZOFac<4nFWKOqBau{6h4!Qwu?Ex@E
zYip|=1_Gi11Z5@pM$EdYvZK?4BlD+n?S#GW;m<6WnPgxyA#{_B_=(zDzCtzf^z%da
z!#9^l1r+Eei%U6%A$-JfRN?{N?(SSf@X(hZLTxJR630Z7T)W%58$1A*Cr@BpBBAYB
zU7ebmT3zKlN@Bsn)+{sqI~6E_9<fom#=yW3g+qhDmNA2yo+gGyOH)fZB1Wb32ERQJ
zJ4Ko<PPWwQ76K{&8I8dGh5=SKHUQM73c5B@luQG5N?)%A2@=PNpZ$<NN%;k0^QM-X
zxX-UIq4M-6(^_f?<fSS9>q~tKe+MMThOG|YE7LtSrp1+S>5ff)TCGcfm!7G|1+O#d
zWX&Br;cl3on(_h$G?p(eDreO3S!t%Bpw*?Ah>cLObm>4M;LJWK=+1vTFS9rYk>uGa
zXG==8;YW7#F<H)BH-Qq%h0-0L9Pq$X5u-fs(i>cJhig_>wq(@ko?PuZIf50Y133Rz
zjB~u{`eyYZpVd8P@aS?po1aaGPsx-MFM&!h*ujlc{L<JF+A+1We{jG6X<fQwi0Vt5
zO>?m!&7%d7bzO=}?Yj(#$(iKCsibdb5U8R;?)J>&HZ4_DM%_SsHgat*t*HTRi`Dej
z0A~k>tHrr)-}|fW@jTJX6eSL_<|^sL(aPcBVa5Ofp~~~~^YqNjsqUZ)0Hp6;@`t@-
z)e#eJoyU*UFd)Oz@NYLETy|{@HvdAeP+a%hU8t(6%2~kZx`J7QFyth)pd|PwKsnJW
zv@I^h>1;$|N4}k#u~+*|<13&2y;>;t4i5O+jGY&MaX!bz^*8nZ95cA^kzJ|twG#si
zzst^-P-NRCVR6`-H*faHA~%97A|q|P?)yp1jN9BT9H|hgskFZE;POVV*3hShe`#BZ
z6w@l)fbHp7=jwv9z!kk}xt8mlXS?yNehodfTHYumb`koRh?kNi%pbI6l;-Nqjy&62
zT$fps=0)z#rlz7=On~8l$|F2H9Mq}-RnphpH-`u;<he0;(7GDDF^C5d0YxJ40T#(O
z083L!1kUexOMn-9md}A2V18z%RGW3cU*Mm$d*`xlTm&{%)mod_<=e@pk7uR_T|TDJ
zr+-_oH0OxvcXV>fDwcb}F6wCRtEQ%=qcc-c`y4s+%jvje!L`Tb;c}%14iqZt`(v^+
zcY;&_wW-&9umM|7jM^?BAd)>pTan@R9@hIIe_CK@!d|;nQ{zo%iiIYLi<_HShquFZ
z7!v}&9{+Ra8qc0#ehMBcj`)S*T^V*1U65+i>XDm+gMtsyaiRKGSDG0N4uuq5T>FWl
z_VDl%1!Im(F`3~_&#bR&DSpN&^jFShP$`0>;{U7*nUF*w1(t|y`0;}>;VC7o77wMY
zOqQPy3Y93Rmo5Bo;t44_@zfGB*3*-_yCaSyBZ6Z`6@yZ7Mu6?~DzI{5S)xF#evXFX
zAn*@vzh%i3w{IsmacX#0w0=f9uLEJhxE)WHJT85$KuV2$|4Uj|p32EEgOn0yHvgC+
z*nQf-dV?W5eJtiknI3#3<AqW^?B?dCwh(t_79VdkTZatOkuQ5;W(LEG`$~|Cl$4YR
zuA-t+88dhdmHN1pmE1rc10}?i28nO6#ig=re{ymX;Hu1#QVf(TKht~9Uzayn4KNXM
z8LmE$=P;&yV5_RN=#Sl{KDA6<ZD)R87Y+jv$83!B%?@AK%bA&LXjpWvs$GJnurYUL
z#xAf>Jsl012n|BHCte{TJ<G0otm((ryzF>aclXmHpb~1a>_=bh&I`Wa?KqO?I1sq*
z=Zuba7SIyQH~cqc&Sm}cu(Gz+C0L7~QoG2MM(sXKiuOHMj!xM7fCaPrlT1Vpf-IyS
z5zkHw9s+mek_LF=wksQw*W0wYXe0tb4cfm$<Yi`7Ae$mz{2b!wIC*~dGhnaww#8-p
zaOH)aqk@9MHSt7n&#FD&aCmEse5Sno<|cg5_AS_l3>$~u$4yIlj>8fMpQb3W<vw&8
zP1{!qsUEgsqde}gPfMrgv-xjhYU=6^;k@Iyhn;Uzjhv5GepJ^digS$nmd@%LHbSM-
zidk+z3=4Mi{N^SL%#%q$TPqcP3q6vo4koNxKM{DDVPwB1)kPbRZmU5koLKyD`LI<w
zx^RATR~hRM(gOxD`-g{rIINuIBOJwZX609>F<_UxAkSkVO;%_TeIRwNPc8tfR#H@|
zY|-T*eYb&8qAt(QqTn_he0M(AWqJfUCp;IJ0HO{gQh}6Put|f4hGr*I?7ir~K2o+j
zVb5m>nk-e8x+S)<{QMKW;c8;XC|#HN+|-$N{#m283MraquOA2Z#H<r@7&E`-bs~6H
zR^>R|b__PEee-R@QXe)N9@f<zMs+oaQ0pcp6T)7@(HrIfMC{5#`AF5d5j*v$tT(^s
zcC&AQn&M5M--ezO8^7SGSV6I+uSj0|0P(v%i#&P_pXTRe@X{h7V{QU&Os(JNlkhC`
zjM0y*cFfS%%lvG$E9X~%h3fLA7E62QhBJ=8g*vv#MUKQ2v4TG>u=<%6PyBee>+ycL
zEB<a;x(~8qzOE7$R{T&JH9Tb|+myStxmo<-_H^~({&FA#_}@{o9Q75bo`ku$UP7sp
z%Ae!D4>zcT0rY2DPjfy>`*WsL`QTCMRI+h<_d~Y;_3L(HPvl_WM~*&DMz32B_RP#a
zn)iaTL>U_J%DlRI0JTX>Ev3FXH6S4R4`0F2$>oZd5asle5(Y`ew{=gew$-C;;ly=+
zMBaXC`__RrQcAh(2Nm{Kv0P<${*+@Eot~lUWk^S-C=TVIM+`vb)}*X2RCM42s3CaC
zx=L>@HLd8zPj`r#fcNXOpB2+g^pj-bmkx{eta2)(<R+Wex1RjbFl_fcV{mCfk0`3E
zON@+UCO27FEoiV@X>~ILdn}FYB%m~;q*RD&)l_;71JztkcD6F(6ho#lkE1vz<m4(4
zAgfHX;C3<J$@OFH4r7F&j7LR9ZnTc)3O60^nc<Z33zMRuwV!1rC^E<tZ+LpJrc=T(
zw5y(g5aVhqL;D3ZIQJJ9!_jaU@t=j${^jTYIk;dWU5Gp40WER1R~N~HYEvgH-UUbX
z^$O?hZrQpR_QO3Wi**N1=Do;O@&4PKh`wRvIPJ^J%goI5WO}?2O&J-n+7Vt8AxhI1
zPuE<QJdGig*iGz8#BmY!_V$B2-sBKG#RFVJQ`Myw4e7oLM7in3P8@LK(y^WInW2*K
z03HoAF@sCz_L_rlxtl-5cUwC+u=MfGPQRCWC3#~o(s^WyX<IN}zOz>xi8^LkXx%V|
ziPhlSX-K%4u_3q`czSy}&IHLB6=>N9*3Fn&SyQ9TAuCRuB!T}tM@75R62Oz@*4A-H
zM@^R%Wn|>a*Xc5lKDr)jUfUnDwc4h2>dZe~Tl?19%;)NwWVngo*M7OmWd>AKXgJiE
zsNsx|?lRfr*&|h`Mv|O*bxd?^WrHkeew8TVf}4Rn_FX-?zEXpk(31-d`BCCJX{rKz
zIu_|Yb}uFuR*AcQs>9v;?(S~X#Nupn$Zto-joFC?5R{1qDm^(QBq`yxO%WE4L*MqT
zsUM4#MrtyJXRZ628ET)Mo#k>Uc&|lh>)I+&5FcLNR997P$hy_fRWCI~eNX)Kalsin
zN{%%%m#Qj9p72_kE2_^TK~`EWQC&LxtD{ZICLIB`mv3jb{EFLdZR?6*mou3#CD~Hn
zrpI!n`2{xi3HlSf7;=@LdkjU$XxO>7TwK9DB3F6RgfdQIs+DEVkZk_A`!gmGFbj2Y
zMDjqV8b6NzN%%mRaPkKVw`u2j+^3)$`PMWAcH-${XeEl*|I_tVH3Ouos!^s`wY(`Z
zdi^T2t}aDCet0C^(I>=>6R%&Kdd$*d)OOfSlv9X^7K9&Zp_*~N(boF<dI`cJ|I=uB
z9kcOrqQD*Fpr9gItD`@I{t;`j3+xJ^|J6x9hVdY5oJ^6=i<7y@tsD||LEGLZ1$ut2
zt!at>PP>2q^2P~luz8|?SFRMd*k|Fq(OKtKGd?R+ettYWvIJ=d42Pk}*3jJ}rcM)!
zX3rsPJpp_T*eGQm6oRKgNQEu8DS7|O?4OjQx_8HiIj1d<w&Vid7*wZ6gOtB^il-40
z5|TF{ot~ay9lGb+S?!m<e_>^p#UD3YTg*H5OHs+7es0bEPh_O?@-ncVv;skTeD8~e
z>De;#h2Gv*iR<#^S_zn2Z3A?}BRASQ>EA8EMgh>IS8JI>-Td8BTcePflV<mj5n@C8
zB|W1|5Fg1}>qB#6`TLPgk-&e5Nr3|GzzH_j9RzP`P*%gh%*-reNoCGd37SpDJDrss
z7Bl$zuX#vN(86)p!mx+6Kn}gb>DDO5Hr%44GfCXUR22&g`Yj{vTi)2TMO5Fh+XVB>
z4BiOqlx)G1t`)SK>kV_)i-4?nitO|>=EskA_e-*PW1<@jt*pK$94)oh`(HR;|Dm%a
z)p20S-0#~e(uf$MPg+4o4F_2KY|+V@Z`Rw#g<OkUkVZ`3b1ecjE`sF+1EjDuGgZkV
zPj%&PX!|rPiMkYn^Ao+%&CT^StNw*Pi6|aFfAXA?UG6L0>W1RJHFxH-fR<LlbF+Th
zr`7^f3wHG@O?F2sT_eJYpB6QgKG370MTNIFHX7q%Q;B@awON=_<#k)KB|u^2+-7BK
zUS5V;#}5oS>iqF`USX^K%Cs^>0#}o~<0I}MPwI~&DR=WjlZl9dP8Tuvug8C`hH3&p
z!a|s06q8lO>XdOu6p2kktf)jMF7A;em)!D3vqPSX`B;oBLFCoHWza+aOVT4>-xIvQ
zbK>dfOb=PHB}iF+A5&3YUjFT<{5k-l$xK)oYv3-Db9b*IL4(I1DV65YB$UCRQF7ti
zev{R-L}0~w4@OTWx@KylqtKF13)(|w>Uc?~j07W%$@1T}PiM$6H+-k8l4e!|n_k0-
z^&{P3{(qjY?C?Eo2|~Y#<PA@$F+pVONm3FH7DFF(9LtU@xsAB(939mwP^76sr|0Fp
zwp040bgQ|kkGw_FS&1C@1T-)sq%iwn3^DA)B9}jHI$u9n&DEu6vD)_TF@%Rk58Ctb
zC@*yf*|8B$Pis?QI~inXv$C@B%0brL=V9{iXNIe0sy<&bhN^u|<%E%Nn%;ssRRzr7
zHX>qV-;jL|*ogL+tvOrwC`i((R#<T?YUDBEk`onOh>d-eInr(+z3DS&6Q~mCuKvS5
z+CnH}84T58BO})o;Ehi({=}KlLrRxb9fD@CqonsPN9s4GcYCcH6<u2;s59AYtst)q
zy0LKas6^=X;ISv3EbYIY6+w@CPD#=J$`b-9*2ppbqVyuRWM-JCY3aap%Xg}uL`xc3
zPs2jdwIst5Sc9#V4(rGdT)Z9W=iODY(fZ-9&{1QUai22dKg$;H)K1wjJpe)4KkOy(
zNJA4dA$`(i%L02*#nXt2qOdKvL5EvG-x?PwCNZQ@-~M7hQr~$FjmBL}4L!ngqlhDH
zZiF;p=#+)B^R<!y#b~Z@&U$2zdwROb)a1#5TgVb&QtFH%WqICyz%TdRmo&Uz_{@tZ
zf=efJ_Q#YRE^Zw9W7lX{S!){Rqc^*Ly6)~ghfJ)g&(31=3r3|7t?56j*PC{XiQc|a
z;%JAi9miLYC@KC4!-$S<H6(ISGW~ZKoNp6j1Kf0}cxcwm*Ik_;2K?>St-a%y|K$SU
z#)rM<hml1-el)ogEpK8ll;{w9baW&L3Wr7;C=|7?!JHUH7)F;IJ+he{@{h~JmkiHC
zDU*BvPDX*<Ud3v}E(_h-*rKfIrA2)VwT_k@vZd=>=1@2p{Cj^NTq^ad_`)Jp(5%&v
zO@~#9U3PQxll$7^SMEZT0Ls%VSPN)b6~Dk0YIDkjITXJhI<iw?yy7m`jgq|Auc|F2
z<F)<1-bdk|o?e^F4Vn{uxN*rdtb*p&_fwDP5g_#uK|%ml<gY<hGKCc4>%f7s=at%w
z%>jF}wF9PqcdLIt+~VMIBO@<Hq6YV)bKJQ81Um+vbI_dYr_Y?p^r|ETrUkbRGQ=v!
zmC>w>q|wu1NXh%wArHmFMhlYYSv8a5D>_rQYbZetu4ZK;qTAlb#=>%FHgt3<7WRr}
z7BAIy=oJ&<GRKbC;BdFmNe@siuQBG<vH=4dywun3jY~1nyyTlU(zdAKpWL4>9L#Qv
zCEeZK4No_k73<g9h^<=lX&R8dFy!ky$_zJnQNKFmK^`G>=2v2ElbiSqRakgA=C?r3
z-5{Qf6<y`btUkL%Gd^B=x<<$YV${Zfg{7^TvLcW49xIH@ocv?M{QDUZKF_m^fpr5?
zY?^9V45-#={C<UMPkg>HdsyrV7RO@|qwa%e*~F1FSw)JMM+f&%!UN)L7Jm0n&j}W>
zh%&r@fpTLnHMJQSgH~!bnsfAkLdtK|^0_^JCC%B`1>4l6hL~tYMP)i^dk}f!Y)2sp
zvUP()_S1~v2Cl=55fJ>R-;yX}m|knp<&27XrWvynH$8o90N|7<qe4(nP-TRcU#h~2
zwWSa&U6FrexL0g`d6Y?MX`)7#qL0s>Pg_@3MWxiVo)Ru5`EzPOPnUx~C(9Zs&SAZ0
zGTfl5#uz`&VH^G&0?KzmE7S>huFUlGyyc(Dx@=j>6|l87TsYFy+qnLB!^3TT$-(X)
zTlM-+(*!CNbWDSHzD9&xPZey?5FjJ(j;TT>pFF<K;-w%b4?7mJ*@+^)3zntMwB`XB
z$HgVB`%p*GkjriU<J#|bugzpL=r<khk}dBn{AYzFQ1^B)+iS*5wWv#D7l+mA_6;g8
zgF$SZshVAG@-3UYGtY*0*h`VT3f+Obdsts@dAD?0Mk0Cl;pvq(rnCqZ3!!7ao0C(M
zQVJy;y(tSTN=Bn5x_=`qN6qz+@L*SyvBy0g3<;qgeo;7yS}f^7U5JS@5uDF%?1?O$
zQv)vENaMsONg)b!md#-p0G<ZDTpjIRF@qg}{-w808R^MC36Fe>St!0gLx9aTzud}d
z0@;gvSsd!KgevG!&%;BTanJpE{bm8I_f_h+95ll(mxZmmqC)6f<Mc}Y6?y0vcP=CL
z>17ph@rAGj=GYU+d_9=!3|Fx#>QtAOZs{ur+eZ%o52a7CgIXrc{CHWOCvZFO$e#$G
zPLjO$dnF|HvsW1kWgq!&$qmtrfmF^E8B$$deG7dAdPaZQoLd@tb2}cDYA(#q+NRO-
zWHFP0p7JWIf;F)4YW06&)Zlo+jjwkhJ2kT>+FAm5W;#Sd-XJhj=1g<hjqxb_-Xv&r
zrgzcj7vxJLC#rX}uSlksy*K22BQVq><*0FJ%|!v_&UYxIe2I?m4)^1>fGuT4mp&^<
z?8W)B;%uXF;ix2Nc1pFoyQy0zGdeaJQ*4Pk7v@yqjN+S1+t2R^F&jNwD;AX1Bvtf`
zldp8)witaNd}<KlP>bmkW0$kqHZE;G5&3=GW`H|i#`t~jJfN#R>}uA?VKjSffPsNn
zLpiCT>&>OEw=Xu%P(}z{B1VK%3OzUU8&`^9-rpO8R-@D^SC<z2SJ4bTG`Pv*+e$Kw
zR(#Y8T}Hj9Clwr}{a<AuphkqI4BFB`8cOIVmsOy2v(IQwNj@G2hU>ERyyUlvoL;8#
zZQpqA4261avB%4fY#g6bk|-=igyBTZ4C$!)G}+4N>vnT22ML;ZjsA^8356zqwV1RA
zX<Z&)02cNW;V-Xn;Z1gIvK)DMdwZ!VR2c(eCxav@CKItQe|=nDx!TG(NtA`A**khB
zEv&!#juM_rd1}jhcbt@Fqe}!@4v>Oz#3c8HNk^K9NTLqhw^N6(Z|$SluedyZC5nUq
zP<vAL=BifD@F-u0Wv{uz_KwudNVHZexbb5!A7l+{ai7{pMG{udRcL2hOE$3o^5w0x
zFiuZDT<#<qYX7AeCxaK~-(^bF_g+&0nzEe#X9o|ZH(TBs%Qk~SK@ls_%E_sA&Aq1Q
zVLl{op^2<q>(EM-IlbXij)rXU?uMh~_ja$Y%14%BYC{jQpKP)2E?WDNuu#QYtRvR}
zM@bo3BV~VdbT~OYV;vcp3{{7pKMSf+*8sCmGim#H8uE~tGGq;=R)v?V<9^S4?V~e`
z`z;HJ*ko&Dv&O8e1V|AB6|qw|FNs+o_xYK66<PbIY1IwS_S@hd144)Im_cIHhezeg
z>#4j502su6eSN(}kuacH5FJe#VL5n2g9q>Zpg`@N_wDpH+WTlnfqseTUx!Fbmx-dP
zwIwF14kUSHc2q}O{_YMG1&W1(<Zzw>HCLSS-Q7uhzBNzo%H(ZSWF4KM@!{T|D=YWk
z2-MzF!iQqB>gzaqw(~LVx*viAw#ZU)>J2i~miWkN=}-XBcuzlM-ONLW_Cm>3Fq2rG
z1=K}<d<irl$rZ-UvCAF&*~(>Q8N|<n(5h(eq9yM2gZ<cH<3izwJ(7F1e*?2-u7615
z94s{1%~a{oKOGS0;L2iWe**^opdPn2V12<)kKfSUZCo7}FLBwRi~x0xwE{tKVZ46+
z?7v^W{B2x;)|hf8KC^Y*xYy&3l2%@M+tJX_yGNwOIlIGG1Y4P}tEjWh8w1^Pexq80
zo@`vaK6S@#KOY~gJ`@qrt7%m*qOwaA+cD6t-ZWonaiPGgL<v=C_Vc0)+7=~Kt<#*4
z#wm>)X<(0wCG(X2@0baBm8+>JH+UoDzL`Izlwj|v2A&#u==z*Yro13Hz}!AM1vv;u
z;#_;2=7Cvqiw-Zs`1FODDwvJj1ZTW3YVkj{)nU(&V9X_zt)2i(hY-KR_x+ag^uK}G
z^oU!3V<lYKa9nzGc}T@i79WjLE3oTb#%e2Do#5X^{7N3jR-+oLq_&;vXs>wVrG4N2
zLR@rW`d2XW!+ng5tbEd{J~uP*<qKkL?Jr<d7rS6zGM!nnyGtHPCgurEriMGs4mJ&<
zG>aUb`Y$b#`yMeu+MRCyg+ya;eCWAT2_E#IWVIc{0`Z=TUZGN3aVLTtZ|{@^JT2Vv
z8I^=`7%gu4ctL$CJ<AnOp6Xn&y1M6t;`x`CmZ!E?+%z#2DUSq9-MSU3bxUb}I5vB7
zc~Qy{$7$f=BO}NDbouH4O?6;y_tm$vvokQG$qNp+#z4vAIP?XLCjCS@h;BXT{NU6_
zWTkQC&^c85E2%Q&++Kg3o5|~K46HgZDc8b~g0JDGxXtnN`CT6M?3Cr`Cq;P%7JM{J
z5;xM&$d-TC(W$dD)qT{Z=ekBRv-@s!Jm^eAX^aZWBY&a)f13Ep$gc>~6>aoO(aTkA
zw(f#eSgj__j19^6_DgzuKS;fTcA+8{Tr~EccwUME;?7l|9BHJIEet|V<;GRsO>?_W
zb5^NVDp}0JBK-Na+mB7lX~#x=cBn;^bh<}jigHOxiu?(Qw~sHYj|h3YLzFewx8=i%
zaVBO$=r@Xl-q1%j*MlsO3CD!eyL5d}0d|BA53_1$&&*;!44!_bXSlw}vL0?EB04>|
z0P`vvD8Ze+?p{Mc_BKCy1PAfRKPN?Um9Iy3BBB(<8Xg;}TI;2{QDXT+pYc@5{%fHr
zq(~J;lP82bG`&wRd@rEIrDS6hbLi^Xc_*d@4UI+jS;S>MFOoq%i3KfWnz*7ki@_aw
zh>lntrJdbJQ?--E%r_T5!k+Am|4mP?S)$t8BX~i~lQWt^-(kEvJnV6w9Y2~OKQ_Z5
z9H<YdS%36M6Tbd3hwl;1I6{7(U4LAvB_&%?cBq62tmRNqYL~fwaZG5csVjVbcmDee
zQl_Fo=iTrAxF(M8S(eRZoN;B^tRrcPjO&fXhP9+mnW+*%oBYp?@R$Ry#?ryTQI}wh
z;MYpzUQBiOB4%z(yk`syp@7m5mA>{-LW}EEM`wDK=I98gat8Y8*YQA(Tif$<=bfIr
zoPWJ&P{N-Lbd&RKN|hzAR~aBakXKn3CWBv`84HpY7eSFVAzlWUY0(2w9UT=V@BB=A
z8+YAdWS74<Wel?3|J?VlTjg}K%Sz<Vde__AHqKNrgE+q9-dC>Hi6=xvnv01Oialmp
zCV7F9eKXVca;Qxko-A5TwPyjlyG0il(~o~;CG3Z8?OzyBX8Mb}*Xm-Tj+(mJxzxTH
z$@u0c0+VZ;o;jdt+97@^g2MZ<H=>U_gP~|X?8zG7flv|SKa(!Yk{@0)9NWS4@nMLU
zg-izTYB?Lm!mp;w+w!Sn2c$)PKVlFEmP`!x)})oJmuL*{luspvD1*j-6;)0&m}hM&
z+RVw(3?=l5jW78J-xfEf6A^9r5+pOnVo6IE4|_l?M{Q4T2!eY`CUBF#TC*nf%c^YY
z>gw_g;aNjV&{5AT8PucacX!W}Z47+BeYJp|?KgQ(sSISX`ON5e-<=QC&Ck!X$MX5A
z3gZ<|#q@p<CO&1?$6fiv!qeDBgHt-n7|=C2ctH1P5)@hJ9QZMf?0BOcr>gg64~ij?
zHoUi-m6i2+*w~SsopSXUvQTJgY3Y$e^VyiFrc-smL=iIM)V#_2q;)5=;&!i{H>;_F
zVU!kYU$^vYdYW~pImSGUS;p;_4ITIM^mH1AOh}wiLMI>l{M)jFyHiQz{=M2T*UcVn
zhALmX+e1MA{#ud!S{|C!-L^%broOl*oM;8W_s)nVry(>)U?M3vZBKS|>&%TJyR3|?
zLI1m1>cn2DdNh4ZcFw}(N?gKewVFt=hpzusFSls$;}Xza1do)=wnc_Pg4$<s>?a)>
z4`u`foU?NS)?|90%L|{M%=IvSUde9vs%>s=c==Y=HOV8eby=2*ima%jYM45SH$2>v
zxKXh*&*D1`{JKMAY@4pAZcMdVQO2s<SG(X%U3=^6kL7atWu;u?PSJ#PB+aukbJdf&
zmJ`vxqZzNKGf1J(Zy#$UD7R#12E+>d?P}R!+2l|9tc~R&cuC*g2Wym0zA(`A{G$$8
zvZWSIOJlFATmVFecdJj!G5^QrCaj_7V+A4FE5(q7Zt;u-*K&P!ZmuuZWHz;c752}_
z&7YwX^eq}1j@LPZV(K~+g(tZC418s@mz;f|CvQ0?>}!TK6>Y!GJKMgB9$3#`?zAGC
zyMzh4#mJs+U7yFR$_>RD2z$C`wY2C89>(;9X*wkJ+ZUaFldY^{Ol4U|vC*xc^Y%tj
zR8$-n>f1V1VT2hO61OiuwH<tf@Oc0~qS?Wy?{5T2bTlg3ox$vfS;yuu+Gq~FmS^AM
z#)Q^C&Mba!1l>U_EUY&|V^@Q8a;KQ(^yep=+fDm_{n+CA)8sd@U-Ky{D3mH97!dPO
zG>jW6wBJAG`sn(Y(a@3f8oH1r+iu*)$M+Xev_lQC!wgN7(%iW8bZl-;3c0eJk&zo>
zV`HnT=rHL@>IZmK%L`l&cCP%!^FJTrN=yLY`y^ABlCyIS_3+1Y{QiGmzNlYdD}~w!
zmuscltbdh>BNlfPco=a8>jrC-OQgK&t1~%_qRCImu_1n8DEloV10PRI6QW|@mbz@p
zpzy}`$aZOwu=0Rb?)O|x`;h&{rms~(B!gx-stL~Oqdxj5#r~+sZ|n!?Jg;h7qCrJO
z&^FY7=!%QF@y))I(l#4Ey?q`bMZVp3nGCZT*}}pqJ?eb&-@L%)y1osP+~Q}gJJf3i
zEtzqAb9ViAfCk6swVYNpN79-GbG*2Cwnj|4wu#A|FIU@~H3T9IGXuPUW~VvTt=FVZ
z6A)Fx%?1sk(b4b8o$8cmebE>s*xmOXbxl)3wsq4VXR4i+4~64M2qyl9eOSo+gTB8=
zG<>fIfoK$444D0}E7qRu@F6SEfGImFXbSM-O}w2EX*Vweyn8NukFwE6$nsI<s4(&(
zX6m4W=ELJNs>FR&?CH5q0ar{>N~sHz98^F{T6LooHWtn^t*@`Bpe+Kl3xKM$b#-@{
zGRdq)a}^j<#a>4@E%4NB<%O?p<p6$`l}D!ws6QP`dB@)0z@lCUkP^TlBaFAf=%`T=
z4_7OkK`aBC&;-UbyLR@G6DoNN0fJ=L6*0)Q`O4=NZh|l0H-CC<V`DxYA{u)nivarx
zU@kP3BHwm#D1NU*m#lB7lL-?QY_yWAspnjGW;Pf!f7(zZJOwudEs;3DLQ=@*{MI;U
zvlQfyub%-3eru0rKqi+99>N3zxHsl+iDV??Oiexd|IVsW$~m6Tn@hOggW)RiH^7Ip
z3%IkpKb7<{Q7eyAo)Jt@PP6q&8x(P)#$&M?#b@G~7d>Df{$Qw50eg-CAo(;xPEnOU
ze3t&XF*!1FlFfD|h2dv+&~a!eWfln=3z;>HvNv!-h9u8xO!#D7^ssHcD2^k0u#0Bg
zr4>u^qyG1vjzO-(J%KE-c9Q<vEy10drbLsk=A`Gh{D6S};|WP1BW~LUNzX6O7H!^u
zAZPee`TI%dLweTw8hUUTtOxX9aEYe@txWrv{=_=jpO6_zdGOb-51l<`1PQX=8+P8~
zqCazZWn=d7t&hk$2T4wso6g6D>)*$`jc8s<N=m!OU6(7OhZF@L%F3*siMYz?)@cJW
zo?~u3b$fSNyamkc?s{%-u3qhdpV#s%z#@$Q?p;N~cP9Sq0((BFcR!I{1!KdP==k+-
z9yt6_S}KL%!njZWoCnzc#(+O2Z0Z4=Z9&t(2EV@_xDwOF@q0hOh~a>Imk*Vin;RI<
zN<Da?EEA!H%a;n{7i}gd=Rvh|U5w*g;lF2GoiiQrgV`*of&R}x&}ICTYO1PrKVZ(9
zRaf6U!^0aL=6U2SMxQ~neq7nX_U|1F6wt#cgEx}pY{&C!b*lk^U$Zio-)R=G$|>Xe
z(cOLCJ_h8rESrXp<#}bwMq)e^7S>#P+S-)r%r!R`oAcZHKrwU4#N(z~fdNDaUIjNl
z{mk-$P=5O9XS`q|>-_j<8Vm*t?)L3YC2_*YJei?ftAK<h`uZ^NdZ!#Qm_G*b=u6YM
z#M0tozyQ^2Kubk>f{TfYwZ_$C9_qh8H#$I5E=QvxNsEo8;6PWXpCMi|lZ}cxZBWKS
zklgFKD%kEJEAyeK$Z}`0^hw8jED=!GFrA^VUX8sj`^8SlRLDZn$PCrG0MZn|JJqUn
zCyuk}k0k=+6enkA_AfJ!&}j+LccLOI>Qw0zChWbTbL+`;pS-rmK~UM6GZQ4mgwjV!
zO%8Sssiv~@o_6d9!2$X0`>mn^GY2r}PAP3TKR-W(fHSF&`&5}q%RY@XV7IF?mtaH;
zy6RPeIAubY?<Mj8mZkLi@Qhj*LPM&<O`uUzIW%%ATr_9dutA!F#Y7cj!~GAVkp<d+
zyOFA98gxDbSf%D(UVvBh6gAw%#f2ktuQbz?^}hv^VpbtPYFyGGAMGuc0|fi50M^B+
zQ*XVwx$W`Rqh+P?JXz@Ej~^BMbR*;xG*f#=gJf_9g1mB+eoU2x_a@K7{I5<nL7-n~
z^RNQ^@}+|oefHLs04HGEhgZ=^OvzyRl$7+Hc`PhD%bl6J^{V1Ca|`~foFQcQ3*+xD
ze6OMjv6IM!0gLE$_0rwwe(%ru`F}I@!RIK;V#>%FL_gB73<&Kz2nRf>rzEPP2lvih
z0c#{S^A!?|5%fj8I)_~KwDoWfR}1d7`N;Exy|&&E5h*va<b1WT-9Q2}dPAU>;j^ip
z*(GAr&3#J|<q#H2)|k`sMAp#AXs*tL0z0ay%Y3=Rrw0&vWvRs6f4vN#d;Svd_-fF8
zT7VD5$Y=^j&Z;4Dc-YUS7E$<t_43l$(a+7yfUdU0NK4Y+A-V?kFtWz17M!qjHoYx*
z8_564c{I?DfPb$X@2?=M`(a9BTEV!b{nit#(1N;=0GFYB(a^?h^)Ch;9yuAjxw)Bi
z$;a_kKjwl(t@GQ;Z@Su~Q<n=q?VrJLc}JO%g}+}I-$Jgg0r((@O}Af{8I%Q4l{}?<
zK_$fU0l$9@GVj`2Qp=t4=it0r8(K2ASIn-o#)YR4^Ue~ks|2E}@tP}185Jmb>u^;T
z06z@(sdb>^#s0#RwnuLXYU!zIR@y$+)X*p6qP)uY{^jD9S)&W#A;nbf>R$V!ku9p1
zLLVW;-6e=Qh1rX_Km>sy#6Wl3A@_Db+|pU}R3v_lfR++_iHwm5a9sXe-n<~jz`Mes
zvK$#k%nv9h9Cv;^2e-kOu{m_XT_eC@3wh-m@pKCucJ=7!D54>jFa8GIUHI)|(DvJ8
zS1K)qLK#yvQUWCkDB;q1c~!fPkjcICWkm)m5hyeT%rn${DCLTTmIWWlRJ{y>1Ew+1
z+G$wSJyAFBHUF4qF9eGr6gyHm<R&^gn3<cwOYSun8uswyihFAN%l{b>ueHVp^DTdB
zE2Cejtc2(b^bTz-Inmqx1>;PT8M&%c@GPh`7b&(PKIjENA=!;;{+QRW=c4TPNbl>-
zx$N;!6tN3&!E_sFqqfATM}D?LOLVKdVo)Na=wt1NQNMb#3}R1}BApPSM<_SwSFwm%
zH1(%aD0>vX5BeT&#o6^&>UzWs7&zuYGOhBZ$;)J^eoVxj^=S}8n+N8M490+fje#)Q
zKi(o&4fG0t_~X6e?~&2(U8zMi4tmv%z*^qd0~d=KU}=ByL`$Rh?9wi{viD`djdA?C
zI(NbE9|M@e0KT6PVWJ8r(nThRY*bh2m#0QshQ74i?<O?68s3tL!-ba%MQ%@|VJ-Z3
zDSCwd#r%Wf?sY)vfHV~O&E<FHvT2V9DZHcsg<99@ikDmkyF#i`T++nA@)Q(u-jQRY
z%zyeYGd))wBmwa}TIDokuUtnBe@=>tufa|%%~kMQom(0kaWUW`+JM?ofq|jY4il6q
z=I^7!&~QKlU?7nKGq|T9+OmNZH6-shY5qV;?te2H&0N)hhoGzHd!D6Rig?Li9d!oQ
zq4a^b+=6bVyKScRa{ZV*8QTr?Ox8dp3&Z42{t$Af6b#40ZXB}diB!PHtHhz2WH(3=
zvE?J*bea8qEEIM{7**zsz}#G92dYdii}aFPK@#0H&flgmrBh0djyT0n;nUn*rkrTJ
zthVM{No0+PW>};iVKFsMza_Nf<R)E5XJ`CcI1v?%paIs^$m0V90kjdlhnnxHK;t~+
z+R<sJ>oQuN0G(y(F)aGl(8YNq<54M35IGU6t^6x3fsMC_wJN0eW~L1;!@ujtD^T3Z
zKrM>7FOovM|D5ljVWA?2ky-ywWx%CCLGUbKn+n47eZ8`_)Vd)Be<wlkGu=|FRxkV0
z#{(V_pLLi8YNSHVBb@<4Pc77Dl|0-pQIAIH&6NkVyI$L=)>aI!tr8mLb8cin9i~KF
z)!XPyk6-u%!+xB)lo>SF?`!N^nS@kx1Y)u#S#6Op@5BDP%$g@l3WH=a<HbqMrnj-p
zrbT*HRg1u#x#M3Fr0mA$XaI!Kj{-*V+fhc>m0XE4!|yY5l8V~trAcH+z5kgWqXSSD
z&|C#hb_d09OV=dm?ll#v9}LEl1;O8c{VNs@j^Z&m3I(t2F&8JNxS8p9-W?ihS;~wN
zQWiArs^x~YiBJfXKEdHSQVVP3bp8Ub5rIlIy}V9A*|zTDni!v3uMD+q0xSkkXM$GB
zA<RC)Kg5Olh5zSC!6S74>*ri)^;lmkD{D~G(bG!`ilcqXoEeJnj^n8yWF)5N8fm1F
z(XQ>A89+^TmOWaW-@0@F64~DTCBL9x$nEh!M4L3TE7)a_JrM7o>$lH)&LnxIU<ns^
z#hSibmg7o((;dgrVc!>t#q@9M7ya8i*mE>K8*nLY^iZ&6qy-^*ztCzc&{hGV3G$xD
zV-ScDMrff4d9LD!lsRD^Kd`=z<N2>diiY<;3*mZBO~NGFcny>beY2WxL09%AcxM=C
zl9E-KR05dGo5h=>s#Zd|qy>O;JtJURN>hQxk#Gis<(6Lr|0)SrVW&kL(YsA8ZeC1Y
zM=TeiCmxCnabySYp8!Ms`pz$Kl86_nFp11l3UTK&dF2jER9Y0?jdwUC;D*0FfR|Hq
z{loD55{2>QU5vkS=3`n&#x~g*_kYg+Y{U-O0odr=ziW@VEK+Dcuuz-;JEZ*YS1I<$
zShADezhY%duLS<NKZ0>_Cj(K4zr``m)5h$`5DMxfCPH+g`wk?nCsU=gW|UFAX75#=
zOBizi#jHHN6f_E(%CN;bYsQg3OWA3*29Pp10ca^hG;L`j&!q(-{<#A#y&lRd+M*@b
zPM6K$bfDz`DCu;2ya$Ld5tx%DN@D@Z{rBI9&+|86hL;p_8@}xi2>^0aAbZ5{_jy|h
z`3}JIR1gW$-y!V}Zg-y5K2}y%c6Q}~C0$+1);vU8yGU9XAd3I`#_}y{1SqN=yDtJ+
z`2Tz^z|!QTCCCtfk9)cNnqXpUG{?x$aAN?8i3%M2ml!P|xCLzToM=e$AW*yA1xw-`
zi#-6gN3BU`3z8lh%(DXa2;A71rpd52pVuz&_8;Kd0znJ_fo*?^;?RAO!QKSH!4p_x
z0a+qIvp_AT`Xd5pZ~qs&{F>1-3%XdalcNG+KODeeSN-wgQ9J?gVIdD56hNeG2h%7s
zfC9I+w#xGZc0<vq5mi%}7`9%WaTZ;A3__8X17LJT`0CBd<wv1s%AzcYb3$U0k94~O
z%fA7tDIoBSwqIzp0ZPQr=d(uTy8)puUy`o|2pZP5cwnXueKYSAhq>YtlL92cDBb_r
zVX7=YpE_|p8c;0|Y6O1)dmzs)xIZf&m$-43lam9{*uT55a1D@~Zv}g<R7y3o#<L!k
zMMTGvA*sSJSFcBghXIQ^<NEjxOp$nD2>A{G`v?X;;0Yf41)7`xTCghsaJFkbQS?Ok
z#g|X7Xn2uwxyV?zfIjK>q!On|cc8zS*=i<W<pQ?n72;ha=Jb1lgTU#^7AJ>%(r0JW
zt<i6BHUK9sQT_zR0TF0G0_*RK8mZE$#-<kYJ7}I;0kj9;Q!DLW+W<9ZkMlb^Itn<?
z108soN)FIK{P#kjlLDrXq#6&ODs>iE0s>qfojJdgA*TB7>s>UOz>n?i3+7xX;k_*_
z0+}P{U~wul3`pn-_$X)G$X}kNJco@7Z|63kAu1XLGpqk6OpOthBp@VQXo>okfJvGQ
zA)n5mkxv04vWv4RwebQ;$?*|)N<@SaQ+upVPyY(=V~}!b$^HZBl>-Ppzn$dGLStj&
z!~M|%$1jsdgo3Zk|Ao0`p(B+iu7~Ws7Ri$>Za`{%?B4k0F&(fv%JL)w!hVxZ-(|21
zPd+oewPa$%N7yXN(BHL_aGG1%JGtVCHuK&vsb!omLoabbtT<s(z@H_JY~VpT1$tYZ
ze>X@%G3FhF8%Y+V8id@!Prj4m!GT0!_@V>cH}7B0BOwKlls<D~(mj0~Muzw=LbhAB
zBY>pJm_mztVHSETk4g7!{a+M{Rl?J)DK$y)^*GbO<9=tgN4c?FCg>-O4K;w~qAIv7
z_*jMs93J@40)!|9OrEa>^x0d$K#Z-ft>~j5s`}@TCnOtc)}-9j1%S|QnP$PUl_0n>
z=#e{WyLWVC272V6gbcZx3n986LyFC0F0zaiQ(%uXkBsLDZ2Q#G;=1^Wg%q>SplK_T
z!nF<P>=2=9ufC`IuOOOw?#C}rH{9Bi$jsw0X*hJN&wurX>sHUKxwkF5g7MMHis?U5
zWM)%P>bdm+pSH68=iYwQeFJoqlM@pjz|_De0z}w0kzvFm#lPLrd@smnBP~D|c7^yv
ziHk)iaOolF9sP6`0J={}a4E@8*GnwHpIj!FXzK9*VXff8-;YM)WSVpN0qKHM^P*E-
z`&PD7ok5594mdtRlh>%~pGu%%GF;NZ|6MAvwF-J3fMJ5D4aBH>cB`t%DJXh?jMHM2
z^Q{J{2FNVk`nt;WWjBl;NE4`U`#p@6lt$K*Bv-&^H+;q5bf>)BQ40x}tudJL^g`-?
z=42y%4=kW@NMy5b$GaOK`lS|^Q$I1%+b3LzbVmUoL(T*Z2X)dmNM5&lW)B@8P-J9e
z06p3<_zNF-0^&w8=%B2*2M|+Bz?$cGpAGm#1Gnz0trP|&pl`oMe1HWIJaxEYe5e4a
z=QH?rmIbHt>66&iMk?rTO!){RF+GtboEX5o`udA=&=SfOvQDuIs{@vp^%SuVk^%6d
z_>U``U%pv{;tl3}Yy4vs84q9)!S`rLwK~A!Ulg!Gc@PR<8rL&W=t}-?!m|Q!{YttY
z(*Pi4MgZ<<-M}jmu!|s4!WQ;e-yEe8k)shQiXz-RlKubw&wt4c50P|9H8#?Qce*iz
z+Ckp^A1uMlqn|=>()U*^jEo17hye!fB?Qj<KMRmeL3q_WkQ@=DrLO?AXdwZ-iD-bm
z<4$=e0iFlE9RfsI0NV<|98dkPn&AFFcY1XXUa>$Zf>34BD?I6D{xS`A1@;&;G%O>P
zftYvA{U6W?WQo6NF{~9JPf#rN1P6*EdF#$j%=Ulh+bwR8Z=Y@f$OV8}9-x4FyScfE
zQPYr;`hnyF^3hS-dg5baD)V9PlyA>r&($cSz``nUM?FCM>S%v|+Me$u>HjM3t%Irz
zzjon=E)hfp0g(_8kZuVj1nF)>x|EU@q)QqEq)S@5RZtqF1f-><L!?XMEPUVJ%s1be
zIWzD1=lmsX_r7`d9c!)oTGzVP{~s{PO`(O;wM`mD0`aC5?)IFDs%oWSOQ13qh(VZC
zj$eB4>pm@Tk0J<#IPpE<&D|Kqhp(V=@zK$&AKV%LT{3E_tD)>T{+V0Bf4kJnIWXQL
zF;RPC@a*gBV~{g|qTsQCfjQ?oP!|9K_s<_ce3Q@=G~W7qeTibbTw9E!ou-$QYr}<u
zk*;5+e^1ovI}S($d_wUqXk+hjoIy>QyqrY2iKjgyBO?G12e!;vvAKfs!DZ1oJg)7u
zRke;H$O7EPGW#WdetEpAFm}7*?mJ|2Uh7TfgG#7KfwA#%(r`(|1rHSD1}dhPC~v@3
zaOxSO>m7qvEGoy_uTFp%1Jue1bJSuYO8l^p3?R!Q!qMENk065x>_dEwb5tY&^IxzZ
z2DS_V`XgAtipGEM@K$GlrzAC8WJA0yy<z$>?|Y#ja%dWv24EiO2Wu`rP;dn{_U^?*
z-L5x%NioXoJ#t-acdghq1SAC*-e<gd^M-slM5FlcK~adh?%%YpABt6~4+=;0UKZEQ
zex>>q$Phv!Bf)Em&rl&QMP-uuCsOK;m`J{ek%Om~yqS%e0&7dkCzs8hS-+!eVp2HY
zTq(JpW73=X6RM!r>*4ZXUFQ<RW<+^rEZN)C8b&B4;~h-;pBJ4?vz82uB(bUK8k_X<
zA0#AP4j`q*l-Q{0eiwtC<F0=ckrFp|*?n?G|6o#T|5yaeEAziUk<o8KyU`^X0-w(h
z&sOn$tl~_#QW?ZzHgsFA2fI2No$bnd=l}dIYuae3^0u&tq!Qw2*SS=%;hC)c^~Ups
z4MP9xVXXb(gLmw<NS<GXNB=2FLu43A>LFmo@$Uiit+Q#a)g%n$-5s<zT42j2{P!Lv
zVl>i=m;Lv$lo-0Mq7={zeWe=W9W#>r?I+j%qDPAQ@-gsh5|o$SkG80jdo`SI<j2fz
zV`SDe5pz9%@|4vD!dFTb=dIZ$sIGMp6Oy`l`CY7O#i(c7M28gK#>4=f^N|iSH8nLG
zTj_j@C=<%%bBjySy4Kxz&eqx4N%=c#Y1`GPoS%=c)<H~6CMcLRvnCol03BxgH|>K5
z<KMmkCACIr>3DZ}zCFCVukTM)ew6*(<wH=@65T!x_}N{kS$cYQrmOH72L}fv2#m@<
z4!#%b3g{s5|B0#FlTUFC-SEYW340^YlRe7@&$E2B4^p&iRgF<Z|3+zg(UboLN*Ums
z8ygSqm@YLpH?!(}*3#8AwyjBOYz<SIkdmN*b^7f~g>5TXcTKleBQKj<Gxz=b=GInp
zM2ngDYLCWs*Vm%?*4V-5brs5WRMApj>F8oT&%RzE2+C8u`R>*E>8H9nL8MHJ8Hy6=
zw~l=O4N5fK5%yEMHk<Y&5dVeyo_9F;O;E?ln(@+iJ>lN_Gqc=FF4yzi_rnH4c+w-)
zW&~^L-1K8OCo$2-r%90MrmIkX38wIu3zm`(z89x?{@rv6DF&br5_vn({6%5Cf%0<G
zJf4{K-J8Me@JYkx21v|3G$c+YE`khWwB(C^<6L8FFuNtu(eZB%-kRE4EOBvsFRzIu
zHi;}RqJMkg@+9zSXt29+BaWc1)8CE9`V>KC?XrzKSbM*DuUI4qCAQ41tgoXX%=!;A
zGeipJ-YfY2XXd9SKcVU9AO^<9G53i?P3ioDTE0cQe5t>>v;fsBKwf$G$BLdF5_k$G
z*_>#Ssi=``u<alF`ZXy1g@Dhfa-7|PUdZdJW~q&glvcTsgprYvsHjg|TwHy^@!GI7
zH8le0b!A0GED7717M#83<d;)4TC7`JUoVuvfyQgA(9>I8S&4^@EhQ({)74c`TN_WZ
zn<6+W$)pE{5K^a5cGeJgi+xQ=hX1RYn%dCNkmI@YwdW-0u<hL)92{ij<YMK60s>fc
zs=iH42{SSlC>voS$?And4i1VG0@-@XnPF1O<||>Xw{P4~)6~R5#>dCg($d=7+kpX=
zB^l1d%DU-ae0S(;wh_nINYZLH`WS5;9j0UjMMW0}hpfR>p*w%}*GH%7Tx_-8p6-tt
ztEg)4E_uWxefaNjU;qBe6Z{wzfvUr;g9Ee5aSRL$C~$`}m0ITH|C5B8cs}21Xmev@
zqXNVRDk`kjZ8bKhzqmU(I%W}bt*nZw5KT5H#!JN#(yiS{J@Jk<?@OEd_|i`%2elEd
z0}yEaNYs6juSW<6pQ5q5`~Jg+X?@EC6as3z9W5>M;9mj*1H(htV2uN)1RjVDSqiMS
z>W{pb+WKvln3(v-_Ph8>rLW?-7)4We_xkpH2L}g-c;J=j-tDa|i^-=p$l%T+E%V#@
zm;`q~Bca=X-QM0F;zv~E<;$0_02P-054n!KuMMF*efktE#<-Ppt<98)!{pSIW|3Cr
zP)3x!qok%YWGj%N;o-%_MZiKXH!(gQ9yy=Q(|cAFnFkVY+Gv>`n%t@8f>N@q0W#ib
z9&T=WxrB_?sHh6Z>1_#F*+Tf%6{{#{m}=_kpq_<=iRr9^U{~Xb(OepUyRPo;hbseF
z)j6Fe%6b&?vW0i=igjgHUB7@3?d5#?Hq_r?c=ylvWmPC+Y1cWk<P0yrM#T^R`OC<I
zJozd<4r9QN2sSW232_6<Z=$5Y-aDa_g-hD(KS)=Sl~wXu=Q;^m-I&LV?*<JAx@aj%
zmc)bvSBOe-xn45!sjymc+*PL<87R2Vs0>C{+)>hKY4`XN!htLJxK#~!T$KW5;uk-w
zEa5gg)ZcHhsS7)-Mt&N1^VcsUE33(-ZJHsPf*ZO$QmeF%pUaBm#+5U!4v==lL&b}x
zhQ=MJ1@G((ZZfd@G#M5a2EhWGPSpwo`@J5aZARr<MUq@Nz6>oM*BEcsBt$bs{a9_@
zu{V9<Bk<Uqb_opZ((<M9MfXE8GBUSUuU@&i@mdq!%|6*8<5Iyu_oHq<)H)S+Y7cIR
zi&MYccZr98X81TbR{w4Mt>_dfx|LPw!Vj?&=xc<jFMl<FOj_{udmvt*F%s81zs<?X
z2}S*^`d=q;U?--|y`YT7rCMD#HagE+3u~PO5tPZv$%KSkyvcgcp4nJf@Oz!VhFkK;
zOnFe$S|e5%mAVKoI~6W&ECmLP?^qBfmU)GoSB}@?U_<+OU6t_UREpN5zCA29{J918
z(~^oT5ehLJKiLbY0(L%^y!M<=g%Q5>kAbjoMjoE72vjHR`gOES-ezJh6%;ah`h<jw
zg4n18z#hhr$tCi?%>Gu9x@p0gI0U&6m?D27qTR_)4uW1;*wy6Vk<`3v$~m;Dlr9(2
zK$G{MpO(VtUv(@{`>?^F6!g!h{2K#a3x53mKd+HX=fDvJ7kEs0<)3@^Yly;ybr&K0
z_elSoaWK{8Y~CD-?(Bc(Gmh#*MMo#W>j9=2yjDaoo(I$AyRJRFPhJp6hKuo`aF>(o
z)9`RHBP9_LO#`DDC@z50r68p^`3tehkfaKuv*}%RCD|O1WNXTn4qhMesM-P@v|Cps
zBku^qvW!_BbjPznW=xP;YLD;I_wzPS<U~D>s!pJI*5El3La_r%2=Ly|?sLtFT;@==
z*d9r#4`L!XqVF3}E)6<_9nsJ8b5!OAvNS<q<kKgXcxD!s>YAF|_EvN#2v(r#|M3IL
zV;2D`5xA!?q%js3u0$?PYeFVJE=q+i#=`QeHQ0iSOSt@FNr?ilDou2DPEKboY)2el
z#5*PBHU#qgj&Ia7mMqQlN=t#l6|OhVgm*cEv~j;?*|Kq41q8NsOiNNJ*b@s5H#^NF
zB_zV{2<tEtqc!#<azTiKLUr$6UVH(gntGEj#>@kLX4r|(7iH$duFzpd>Ocqv!N$ME
z`5+>bM<V#Z$;pmte^u4)xWZ})3|<gEIq|h?(Md@sU_#>A^if3=6ux`+$R#|aXpg9N
zJD+@_pH6(;SV=i^XiK^B-^5?WU}Q2Z%r<o&At<PfzPtWw9tDa(*>`sj?fCzKLQa`s
z3oxELQw5Qzm>A=jSoEeQ152E^xUKoftwF`y{Vy|8!~dc!DgGloSZ&$v`_<LV#G$0O
zEDjD>kjE)PpFS6b<(4)iD7f@=mNT)Lo0y`AXgk-?$KZfdMp}aWYIwu$SVyqssi;3s
zN;rOf4jOu<gM6Sx#9`bH?EY`_+p$DFnc_h<lOG4VnjzD=+HW9`eyaR99gpa8QTqo5
zpkTkex*Eg5fn2^f+Q;YD{`!8yF{blx+>)ZFj5nIM=ck@O&sHf=SHD@{^5;-rxBiy%
zAIujo@S@-JE#KcKar{_cUmw=DrSdt6ehGfu+OivW1TVbM8B^!DT2XS7Z?(1^ppp6c
zd0lOOZtlIK?RoGH&*kU9UT9aEspq%HWy1Y>9pJw7k8e*;Pr+Vx=YmDc7|z>@BxXCe
zNZi@kfnY+CYd-S!r`hGvTq^Q`;sWY{)$|HX|DPDC3Vq+d*OZr|AQQ@2B|EOyPtU63
z*|6m0*EXlB<dYj)TJBH@3x5d`L0DP;H00)C$<bR_4lu?#-%ikb`t)>QRB6Yx9$=1=
zxcezw<_cC;R=O<-@B`=|S_wRS2&x>>AgW@sE)%ODf1NA2_j_k&wEXhI!qL$Y$Pe-h
z3+wSJiEZw!4i>+DPLbXhKXU`T0>vwbdQ(eF%hkU0nOg#C>vQ4(7%F@=a>vK!#~ok_
zZL6wG@@_z&VZ~d=qkDX#t*s5zf>b}Qbe^9{2s*$1brMPW=-IPp!0AJ%goNtjmL=&J
z8Na(9%C!A@7s~Ri-nA!z!*;sflAp`Co4A|cqN`X%9RmjkLHKN9*Vkmi!zb!ta>!E^
zK<`4Q;_BMj-vR<kPQ^5IPEsVpRv-HcI&MHo*7aAvb!>@n7Fd?txbee(@IEwN0EsK8
z7gGIbu{l;w(w83ZMT`9&<iV5^;YJNolKF9^m*PC<+Z~jqSXc-l=ul43s;jHxKLT%N
zW@d&eqOCo?v)IiBMFw$k!%4jbH+bZGj<vPL(S5<sA^9aGX6(0+GLWhOS-s9fG=xF7
zW@GrPN3pgcknMowU^)oum{&0G($Ot;#@usucCOGc(AG{A5|>NiFR!T~MsU-FCr(a3
zz0(rPFbugF+$t(s(fOe}{*$uu4xMex+jC$-@Vluo6Dy_(9m8=hzUU+h&i?!fYb}i*
zrBH+MslD}deJwRLF3GCIlS5f~YY#;J8u~R-(obLY8+?6H5!aZv_PedYBd3E)dY|5l
z6ONCTR4ZLt32w6AU@EWQQ;1D(T3A@vE6Ha+B99-rN*QQ5Xz=HmR>QeigXayna26l}
zso#ytby&sC%G1&WN9C!!?+BBz5Ru+eGc~KR3}qzeYab=Juwmfg*>4yh@6@gBS_c2K
z449sc+6Jt^{M6(^t#)k@NxYy#v;U*>Bk_Zm5NgPi3k@hM4?fic&58RjM2Cll78Vyx
z&CSt7baZq?L`1-fFW|5Ay-A`s8+<VufwGX3C!3?fdF(+qHZh@?$hio5hjAWZk)g@W
zi!!>Sj2^#QaUKStAu0uqVupqm7v~U~Jgu=Bhcjw=_62^4J<K1Q`V<qBh=Wu|-ci4D
z8~f3P)qTRDkp?=+t(9wxoGJ%DZl$K&cJoW=^*AUhQcQo))&?*uE*THgc4y4U#ll$D
zlfE#@QUg(lIV@HtF!Mg!gY4lZ!5#MXI%3c?ZC|vX!D1xjSJxnu$~heo7S_*%gS9HR
z^y|pg`Pm&rtF_fCcK|@~tpPTB_-CMcy-qfM!NDF3mb2e=^m4$R0cMBuv)G?0u9_?N
z>Mz*Njfv{+h|Z$nVUN+Z`RK<q!EfGp-`Av~Mp02UT>bH!N&RadRrX`hAgrmWVTm93
zJ)^&O`Y9q1llxPLqKTpM;YqdCXGrW_Y);Yn_`L1!Cm!Jy;)yF|;GmqY%wr4qrL9Km
zWjs8X&oFgQ(VkrBZBb3l2qT@Tx!cKnN93cE!H-Wj8*p)#rs`h_a&q3g_k`g_+Jjeb
z&kQ9=f*&(dQ)~J+?>j65rGx?jg&2sMP2?9qE=Z7{A3=g2zZEuJgy^5LzA350kKWqt
zX6CyN5}(Eg9NWjUO>f`4`LFA3eLCe$o<Cc2Y;R)@i*&INMKSSC6cl>f8P6>}3E@Yi
zq@?HQ<#|O#Wo2a<83qP=sZ+t%uLcGh8X6V?qzjUg4<0=50x1(X)+pY6%U-J5+P6rs
zkhgn|Ae0f+Y2tC#)B1S)XGmYFy98lM!YwN73aQ7>?PUFEh4r{SUlTYu%wrLZJgZwe
zT#K^``n#H!zHl!S)yDwpOdzdeb%Pef@}?`kYk5zUlx&=IT&q7jS?~S)x$Wjn`58(V
z%V8#Q0)nX%|9E0Dw?Av@<9n65wN+tZA$2Z&;ct$lti3LTZ`@#bYQM2EU}-xYd$6K>
zzQ>5%T3#j~3U0nQtsGZBI~S~fje}I}ITq8&W;DI(Tw3h5eJt7aC{5^<eWN!DZ=FjE
z-wgkQ2M{XTg0)0JK}ngY#YaU&uc_HsS;;K~ZX<uS(|y|`J8T@E6R%k?%hsX9*NBM?
z47QBUPfeaYDNqYxG0Voib*25u^~$u`Lw>UGqI^U7wGsKZF=z;YzFhPQSH^1DFsW>p
z6)bXPS0T7{;nXhejLQHD(9Re4B;aml3`9^0(kB;Ozge9wH(lpmTR%4S=FKaiw4uRW
zdB10kP?fv1W@!;+_FX71r`mO{84S#+p&9RXi~PkW%fTDSPW6itD0KVkiH{T%6zIR+
z4k&74R_uX1MqFw<*jcraZv5L<(Xm>_%C{^%JSm>AT;KIU8x0R*z#~bTsZUHz6_#WU
zJl>^dC#0d7T^nw+ZJ@YLTqDHrjfb(*qMMj2__1eB_SVIDY3o4wYIi4;&3yzy+I`Xb
z*T*}<g_?4>7@%UMqdee#=9bmuC!Ej&@VUg%@xI%^6~tb@z6@nn_!TOWWL4AI=z}QD
zZ#?xp`)p<Xc5KYy-8);hXGL-{ABILthx(~PLQF!r$E}Rn*aksB&~Y7eqUM!s4tPP}
zrNe~+AK)K+{HZcSaS;*ktSn<UH-J^h$uJ9!e+7gqeOtEop>oU1U7{2~M}`?0#}(5{
zKDn?3DBOE!<ptE8bvzO)Fq9%Wz$C`@P3Scn-Tb-#l}RB5@$y=YqON8+xQ8Yh8yFB*
zs)d3yE;diq?a-+d46RHHw||Rv^23r%e}v^-QeB-S`!1oagKu>48C>Xpcg7$FO6rSy
zXNM>F)AWM$^tl_2DC3UO(w)Hb9too)u(Zs`M0paiw>c$9O+5zzm|+aPPkMS_;fvv{
zC-S2g27PJg5Z_>h*YGegO-<=(siGlpIJ5XuQqIp%3?Oe0@(!{J3hi)J!DtC0(5L>p
z{M@H<*R?<`&+8&r7Gl8#a7qb$E}N`A-qmKk*()peMn+24OS^d9x*u3Y1U=!qqc~~c
z+5shuxeBQQp`HF2RA~A$RfBmrzWIfP>zW#plsvfa-!FLhoMdK}y}WCDe#+W=MHj^z
zEXh5^G(^Nu<t+eV^zuI-1578nak1S2#h*2=zn-cXD=0*qmFSs0i4>dpIy=u5-BR%3
zgN(eqUP>%&PZz59``8<!gL$H~0L%1sb$uu(=v0t(pYeX+i0$h;`<-HI?^NnHW3bL0
z{hi&?q~wXcRaG&<_GI2yi;GQNadvArZa+Y6Z(op(DKa+p>grmvw!}F&NIE%TyH12;
zl;r1ga_>c#Y_3~Rkd+kW<_0>ReV*&jG%$@jTaf-9O2U2j?$)-?c;#HY{_N@TFb0NS
zb(PhhnLnD!D9B=Yd4Qy3_MRicURaV_2blU&6gAa4SG<e_9o>QKJE>puGf(U2?HrkT
zxbBzQPU+ZvN@8W5U6qaqL8YW5Ct8Kr<MO(gmE+`a=0-w7LO>AthRfWn-nBfaiQ&~D
zD>Jj)fdpubcg`;!ti`?BeA-m3d+g>`g++i+z38}l{^r^h%&dCzz8wPh{I?(l_aX0z
z^UDHuKDLlx#-49cw@4r4Wq(V0yB9=Iywt;y%<cmSV-fKUO+b<ec(`|rjl6qH68X#P
z&-KcXt%W8(uWi>dBaEpD-Sy#tt*mwgA+^5j5pdtri6>ahXLayGB2O`yn{jCRp;;@o
z^4T~a(l}9TDfH~6tE+@GPNL*p<pLu4-nC{i0RQ-frL%4J^{&dsvz|E(()X$t8mw!K
zRbVKj`LB;!azkmm&)-ACErZ<B(yG<O3bnN|cRo6y*YYMNQf_X(Up>bsKOtL8!X6>f
z1F!Csf)A~#;^yR35f>MK@+6N(kW7psL^5mOk(-})HIKNes;;t8uAsnkViI#Rvj}PV
z>g~gz(yyq<g=*eWoik2AbM7KWt<3-`Dg!3w>ORTQomAoETDv*6_0dBj+@`72oyGXU
z{r%EsUrzR#&>XC-kB*lE5D>6HN8Xnx08Y3*RFLiMjUtrR9AoQC^-eL})2am%$$z!;
zcx?e4DKly_(3_eF42%WUU<>Z?483S6M|y^DP76|98KMXx8y`_z;<>XR4ToU0()=zS
z9*Tj1WT3`N4!M-EmY)Lm@5AJb+$NH?JLZ(ivD>^Ymq?>BSJyVg!GQ-Z_vcTj=eZXi
z#;>k!>-iUdV<TgZe`0eA|9d*$bU7N59T9P?tc*mlTWPiZYV-J>QgCjEG-sBT{b@5g
z>ht2px|09uDC+Qtqiy&mV$h7fGJf6daD-g&V6<N7qhqHE4I?AtH6afj9wYzhWIl9J
z3+bGUEHq@vxP$B)>p^yQc66sG9?4rgycd)Yhe-LAHdeIg)(_uFU1>D}-ixiy;Pm-^
zsvtpgi#s;@(2OTJ2|2meGn=okA82X&g!k5LZ0xlJU5CR)_2A(%asBj-b=_C{&*!%1
zCua@ZBrpik83lgYY#Sl$w|Qig($7sF3%Y11rs5*M?g)SQ^=le@&vc<?i65H#vclCM
z<CeuWS2uyZ-}h&?XZI}2b8^%Met%0L=L`A0vH9xN@ZN8C+Pil%KPTlk<f@l@ezExV
zn4O({`_P=2m>9HSR)1jh^s+Fsl&Yy0U&Sd{AHHuo&KboQ3i9PZ{(jfe3DuR~j8H_v
zv*y=T^AM`zfx894!5E3BOZ)5oM@I#gmVW2Elt;ht?CcOSeAZ|Fk1qUg-ZwaZ-9Lhf
znblopcoJTsgfylJb0j5Aot`2huwC6d->%gQ@xSIb=(zgx^tkrVOk_w18PD?@F`cjX
zR<6Z#TJ3KJMyW(g(UQdHidUdCHV(%rD_^Vg_)AEJ?|CB4JJ{1CdY64mI-XT}U(cow
zDkg`E$&r;WuPu2@L&|LRkE3pEmgpHL^gNS|_pRiLl45tvFV(7t+mG*jM{tv0(~lox
z?eu5ABoJ4J%}GIWas;8EkiQ<&ko54AD3knc4kKB!rTFb_L_|h<U}wvf@&aT&yW_=Y
zzWSHHd}c(0Y@E&oeG<!5_Gm#{aSX{ao^A_8vv-hA=G|Ym`yeCpx~m&lv<W;fdeVeZ
z9$hfkzntG$8$K)14^quD7%LO-#lCt~A-~~s-gDT`%gf7v#HbbwnNXxo)zmP%yJG~T
zq!{#5A?#cG#o>{Wf)4H8E!*Q&oj<$r<m76?$ast=KLSk!7acW=MOW?}D+c{we_v#D
zoJ99G_Ndg>&SEVTmp{G%7=|XX$q7>K`!a(337yL%i%Va<2nmqE=`<Z(Je7j2gWrzy
zF--<)N(}AN(i+C?D|>&~&ez!xS6wBM;UTM!Mqi@uN|cwM;o@>H^#ygu&xNvicL$7D
z8jf>nRlTXQ9K_Vr{M_&;FN5mv^woPs%I?mX>NMf3wl?j)6m2t`XLj2(cKiouSg78~
zrECpVYMzES(NSBnY5&s6A|jQx`kkR9@BbclqG73siLE5_-YnM|Dp2)#{9dI%`s5_%
zo<dJ?;e8uB4(SMT3i+LbG$P^y;ot?2IkC*34&8Bv!?#3*t&c}a8&Hwnp;*=%8XfIB
z?g!||ST)}JEa~od;gc1MW|4R5PdF0m@km1S^f1H&-|*TDZEYdQNq=TaG9OCD)`^py
z)8g@w$>!Mh^J$^dxt6&mw5XYt9s?0qepmOF+L`7YcvfCsvo~#&S1uw#=P4hZ4|P0w
z*fKTc@g(w&-S3|$s)MYLF60PXA@l6(=l7_>mO9%Pl0C;#NtlksMOEu-lM1Pvr@M+&
z(F?m2cWF{=r#?)5;&Si1gG80+=7Apraka%=`HkP<zP_}Ff4XIyUu}!AaT2{_@x#DF
zkg6(+^Ye?6!ooVcCZlN~#Nc$x&*8NY5}Ph8dxux<fMxs5wKn-p=Q*1h$e!4wrCQ@C
z?H#cA_3}nPvTf`qT<XI@n#U8b!zIJ!o?=p(@cl=gsNPK#rjnW^M=$x=4aG|Q#M`PY
zKQR+qSTufo=}XS{m6F}Sq1L9w{JW5W873a_8|ZDoZrBV^5zt2_GTj8d1?bIqlUeHM
z^mTOw0Raqsucs05z$!sOL*skM&7v8v_}-U7FO38eZ-cbv>)qZ5=F6VM#6xYNDBeSP
zk;0MY+S=0V6Ni54N(}vQZnyNFez2VxVPoF-^+)l&lD+L@{_mgrUdkM|ek5{97#g;M
z!%#9b&(qenoL@3~Xh#*^i|O6m5s6`DMp-o0kw=hDiH?YhiqQC^Q!;BIcO~ofH<l!4
z8uZDC>^*vJd0G{%CUCq<OA~;q+7R3cA2ggtk2YOqzBt%L#W^`Ud*UF{(iM@diCivM
z@WT`35Oz`_EiOS_K9*i?1`8#-t+Jm#lV6Mq4CT!|X$i!r_gMe(s&a3s_d>#Xht_Hw
z6^Tl-Atf}hvH4nBRi&8zaAbEGuk<Mj0+Q&_XMa`KFG@dhCz(A>3EXgtjJvtA+PSo@
zkoMGNmFP_yRoo62=SCp@`r)7=T45mzDjG`Ts{VzMkcSMR|JJs#xR+Nboos^F`Kpe#
zW&JBIZn6{!nx>=e3-?A;BtW4r<`Dq{4eCg-AfIW7q90Zy;0<|8i%kJ|Y%7jTanOOJ
zAL4rt#|(V258o#1Khgbs?P)k#lJv!mC&_kY)5du*#I-l6LN=bC_U-om`c^j^hNL87
zKziHiT|5@D=(1VQ#olkgJ{!s=;@Pj;Fx@fj$t}8abc7(5-=ZEmIsF04aQ(XWU~Xr~
z-qX4obDK>)TU)V*U12S-8%wGU(c&cMFQc^6kCsk6dL#Q|$QLg!ng_?jp(J_*YHyP#
zutEa~TuO>=BqgJzp4%))tqZ#BJc)>9G2%>wyUFqKF)uH#qMTfId3o*N54b3Ph29=D
znx^-ZJ6coX;~`(khZvM=3EhbtnpG=jv@d-Y@%miK+Pb|pxZ=UyR}cA8Jy1EG*j|(+
z*VJ4;5!E#?HI=6s#|fBS$+uHadiZc5Pmz%*_}y2};!m#@R|l2SNl<3n^ObSV&qtp`
zY&<l<MlAlGEUm4_NM&ydyN_%iQ^jdK@{n=-bKXP-m4pRqs<N`|>=vx&&oVR7y1QAp
zxWb=ApnIzbf2qyQr6nqu*cgwnsyUBktT0kz;`z}f^VtMvV-ly%1&`OdtK?C<$^It(
z@NnZ?+2j=ImG*-dqwkn?>Gdy_>lIT4cs#petuhHs_;wz-V?K#!3@6{(vrJSPt82Q5
z>`UOt0Z4njl&!0~?qXek=a^3K^H{yB;pD=`FdNluX4;+G!nmN_r=x?0d8OaMso1nU
zK|Bc6`&|IW)EGZ2jjy&+<CQB_JNq8uG?r(__o|Fc-hCV4)f`FGP^UXObNRJ4ywdu(
z0W&!9F^#2##lrG32y3D?`r}-`Y#QO^SdH!R?0Zt8@@R?r$8XbvT+^M>yM?9Mg&O-i
zdz0&%=OZ(e$hRwy0BtX%5Dd9E$}-y+YuCRUz2bFl=?vodUgAUrKL%sIm;w@5EPj^h
zZDg|WHd*(DN$hm#%-`~=7b8}qKgj+LYa$uKp7jemrc`g*^G1=}^Ybgbg4TyNFa5r=
z96MhZ77FQzz(7<9BFo#`m+&#2rD;Kt0SzguuI}Yv|7gG1<r;7p@!^V-T+a@1Woqiw
z6Wcy4`Q9{Q;vPELq&u6Pxw(GdQwMo@;`#<4D#{p`ZMkaxU~nMo_uH86s%jLGjah@U
zjYJJU)#hQ4&B7zN*gkP0ARs_iIJ`WUHpUSH1A{2V95I;?1|IdofdCq&;WBrZ;?5Y|
zkaU^q(_?d60lkrU^@Vv=L7)PRUA;wEQu^ni*Vp=trVLS>hBHiBqLuZazQjTE6{Ydl
zfS1d7zqYn!io-?11-$CNd`V47NswET$Moa(x|1gC7Zr8(@L_Uqui>v&bM7QKBK2kO
z-i-iFueq%3%e!~@ikin)uf9^T<kQx4MgztAAch-nLj4vNfK33`R7}Lpo|FC7pF(b0
zTBqMSveDR9)+h!dqCPEKb1cJ>!lBAp(*V{YzF*wE+;+{P8oVSWC9_hS#>bBVCWYaU
zei%hMh90iN*dCI5UPnzAp=?RI0t;j17wn0ic*IEKq#xFq=aDGy3^On9dc6J>BjZ#O
z7L_d?PExWxGEOnE+XsL8xVc^op_6&+KNr)}8$Uf}f1%;29L>!jrK)-zu`=%|^Egj2
zwit74<Gyzf?~da9l6+wRUNY<cSVIG}2SE(X&0W4!4P}Z}c64>gI6rl=w}+dtyo`*!
zi;HchPHf~+k$+GSfa!m)YjRrutU5h??MwAN{ZRwt>m$A9whVn+K-WapypNSspeg`W
zoBL;*=NY!ID=`@p2E#|MhB=jTt=Et_%biIYZ)h0mURSQQkr)^VkDvM?tES;e%bXd-
z@1eu1t*zZTzil%3l(p8d0AT6HMv>shVuY;_!T|eWF*HI~XLEnc;?>=55@tfq#8yrH
z#hHdb%gcOVltXf4;}8DK<d1Ei{p}#2fF>AZ--Y%kY$u!?Z<IK%4HW=310`bT=jXV@
z4xdORB}=u7-WjWkP)YH!=gY}38I>RWnS=*82E<rW0^r;cP{R%Iy5>LRFGZ`u|A>?G
z#}}sy7ouR=2uZ=O2BL!!iKY7K<u#u5Zu<a1>VI*gOZVz%oc)}h-dvq?i!^6#^_q$$
zp6<f#F*jK;Bnuy!L&6i<MOBo6Xgb{aCX>GD^15tJ76I0*A1StW^!E0#xCD@g^pc;p
zD;XSYXq1ebv|{IbDZCRSJvsUD><iN9x@Cak-R6ygAFsj?TcDc9dq$k2S$eaepkZP8
zg_}Fyvyr4jj(!WJnGd`c{j0@=g%f45p`o%OA^?<~nVFbeLOa^F4L#S{*c<}H7s$4D
z4j-AAR?jwE=<T(U0N3<BGxO=rsJ<{rf}kY`09yILflN4~rM1_(F0{KgL<2~^UdKD8
z;^Hk|!%HSdFEF_X2$Xl@`;&4qqbnT?oJ1O(C{h_K%~sde4s7ioNvA*bfReP4N@>A6
z)cc~=OA^n4g|hQ*is_icg!5Ht`iw7q<T5()cXNvC@V9$xbU84g1Mw%hDJetEbgEuE
zI8;Pjt#Un6mX&?c9!8pk@5VY>q<uHp&ffmB>$Z{F{WQDxCKRz-f>L!=35uh9loRP{
zGH=>gm@f{x+1Sz^$`kraG1RvhRM{5;tPaJZj|~k4ACJ)lKjt)S_V4fSmt^?~C`C|E
zQ0?gy<aB@-1}VnP@yaK>dHM~WkehhX>kKrOTmb0o1^|48SL&d3(#D2Wn|?&--Cw_S
zD&r&pBFDWRqwcWGt0}@vJh-|id4`LfS;NH49P89p{R3_lkoLR5+J%SDaz0j`j{!NH
z7o#QN5w>26K{W|8j~?$iT!SKL4UHP9xq$FKQ8l&L!otF$FXymYczAfGPllK(N;08e
z&m!>Vx1bYVcj-`Wb~c>7wr}6^d6Ef3+bH$C>U4p94JSp{;yR^}s~t2BNF^l_%NM^%
zLXzy^>fSxW1xYsbLJc?v9w#4?nlIUd*wyku^kz=C<HF(R0%pk9Q~O8Z{v#G*7ZCnW
zevRtepd$v75zt`jL0Wz0?+JYER->2QYz7*XaPA;M(G?-4zI6<y*xK|-CJ`iJrD%Ux
zE&a|}lcXg=LBb_fovD~asR;Q61-aH{>OOqO#l;oae6yV>{C4-cJZvN=>BXNREcy<K
zRp4I5QUu6ENzp>vCg`f90Uw<MRvtir&_^M(0!<kQ5}mm!1zcu5O7MS%P<a7*P&UJ|
z$Nl#jORW3}#iRoMQ6>G=urJA#|NF1`nnv_7ohB7-&>2suAbq^oo+9WmjiL3oxdv(R
z8$j*JWDvwv_xUTitW-LY^YJO`w9mbu>H)%$z}upPP=*ZcSh}u0rUNn`EH^X)+7MAg
zjqCu5)e*y+h}>tF{7XWWmt%}RmoxuAnrcEP=^lWWLABMCGZDzdi@V2NSD!q1?cpK4
z@9XPp!ZloG=Dabctfj?pkza;6l&3>B9O$-5tE{YSV`Bqr<cEn55*N|EY7-RVhSvGZ
z5t-cnWUz?<{vly~nqOQDJ&z!OCXc0vw%5|pF*7#@3i_SMj^$mP{X$Z}oc<nI>*-B}
zg<ToH9l(=8Km4PL7rS~=8WwED(aq73JHGpDE{HNEB?X&jn&ZX@X68G=K%xQ(qpwcu
z??OXEtpr$Ezk>wvr)>s|^gbvAw&qrydTw^Iw+iW4H7516kQ*c3Noul#Kb87Y7a$!C
zSaQ6)rn))^U4wh7o`Nyq3*C~*Cf0+j^F(K-H1F?)g;ul5IPypp$RjEjs6jjdM#|g!
zl9Qp-kn{j11h^yea&ij}|DVBW85mYz{EF4=Y+B%48W?2M2nII8pQR8F9-`rhkB{%3
zQEY_56I2k6Rkx8!E&crY=fs2{q_Lq_UuUP~&O)a*&!8NBzIAo}vVvINV0c&<G7Ge2
zNZ&!XyZiau5t(JYLk=BEG}W%dsUV0S*%2e}@aZX2+=oEh@7sHOK!A#s_k!*%5SBwa
zQHu-rg@|>bgtZIwL)(QuPr$;{hkPkHza3TnVZ=Zc?~rqc5^=Q!2UO{Qb=#NWO^uFj
zGAbY1sy2l@dqikxjQpSlH%Pk|6a@2R!SdJW3_0Ap7W^1a3WOfOxTz3Ae_n6QP=vof
z$q_5XLbMx2Ee@J0aHeR9I)UtwC!d_V@g_DSKc9i0pa1S%ai%z%i5eCvDsP26<6ed$
zl<H#Q{YhzHNCCqG2oTU75(CvogbMk$*p~~_mAx@wo=VHXF#tWaKuiD%7~u`$a||6Y
z{O{ktA+!J=7z#O*%e~su+e<!8mHnTcO14Q0=~^pCwW1FnXoj3REJ}EX>|p+gL<v)!
zu6o09$&6GA{da~}9DM%k98q6#ow2EJ`u@NBHPh>=i_0PGF|VpypnlGom`X?>wqgtX
zs{b7Dv%&1#+^b~V77&a8y#Oc?Ku)YU2CoYq254_Uw22{!2oHAxse!Vx1a3>vvu^_g
z1neCm!W0Kn#|Oytg9a38o<Lv$d1Zi!b8~YezIihoa6X`2A$Aa<0aqTX!=T~k4rm!x
z7y(BSsLB(-xQCXGMa9MDPz&;6gc-syIBsA<fNcjXmHYcHRM`>vT~`t6p>u9-Ztyj-
zG|>RP+@+yGAR(~}d{6MXsNOJ61UcVT<dPm&s^1{$hl$b%jcZ`wS3=W}=>_~oo+cUy
ze)*I7K2dJqWcnY0W%)5YjDg?=&?Kv>sV&USQR_1$GcYhnL45i0WweK*z5TCkS2+8?
zJG~_8K?AQmtO&w?ABK(loV%mBnTCx`k@xKM*kRm1F>(B|`GD7jCjvu*B?QnFD1K~=
zjGf=UNhv76_jb7_8BU1W_@W29Ab56m781vhlLkW*5D?JGdC9G0)Bk#Uv~8TJ^zpI*
zYM~}x05p;#0mr`#Lu*53q#{dFio(!HQi|p1<fKHaym@A328xXKRt6jeC=iI*K<*o)
zQEzi`Rdz&DQa-R>fI`0*J*Tt>4!KGULe3ir6=+YSen(;ZDmh;PuS)1fp!<LCUH@nP
g;s5Bjqb{&^$l4xHD9mYI_C}VLP<UJ-X5jn30AEJCM*si-

diff --git a/vignettes/examples/timeseries/timeseries_anomaly_detection/unnamed-chunk-13-1.png b/vignettes/examples/timeseries/timeseries_anomaly_detection/unnamed-chunk-13-1.png
index d3ec5b90fe82057cb391423175ada4ce7abe973f..d40fd4d5ac288bd565fe6782d920495b8647cabe 100644
GIT binary patch
literal 36351
zcmb5W1yq!6*EWobbayw1bazRINXUo?L#H6!DN+KGN`nXr2q;L$P!iH110o&5(B1u?
zgZKTs-+JEf|JV9i%Z1Eb*PQ3I&mG4;_TCez{YV8DiwX+`1qD}4RZ$lO<@Re7lw030
zZi83y*6w@*zc8Iu4ct&ru!)fWZVgWdUZ9{bpr|Q6(DTmR$nt6+8J)f!>~m<wt#!Yp
zq(rcD@9B^&Ntucgt2$8(mH$Rm9Iqi=Oj*3`i0Qqxk7_%SRKiqmpMJ;kysL-vN}cq+
ze$P_%iwbyPkbj%YltABmX+ay$<MXROO-EY;ncJCX;>&(Ls>xP_s1FEGAG`|YI=cma
z6QG{rkys;tV-m3UVZ36%AY#x?MP9=q_nOceF605Vf$2jan(*IS{Y$P6qD1n#g(Um(
z;?}VG5}OKxYCqgIKAe!v83&Qx%m;l5Ra~4&>Sfc0kx=Lu1Y*@RCZt*IS251zh6YLO
zT6Is~G?vG38-)vfO~f%T@lAUeE*_Ff|68d4AzPbgsg*%erBjk6fg4i+-0qrqoDT<~
zh}VRINsLfWXF&*AJ<<j@Gk$I_F{4PT#7i0>L*NwEEB^9mO#Brro&bRy$|fk;DTlRk
zeL8-u#w2)Lz};K)DWLch*|%v`jBe|SzFgz~ysUS^Ah*(^3`7u*65RTR0X+Jx+hLSQ
z+1y6O5(I*Hg^Katl94y2X0Ml{5lQ>T{Vw9`lfL68<!LH#A+^MdN_z^leYzy|BiR<u
z+u6jFo<@4*0k6apIQtl?8=1cG|5zrvdmfuA__|I{Pd5>_EuJOvMW-GsEG|@U607(<
zi(21c-&9iqR&^!+J%J5JawewGDzv1iFI>G*x&GVK;X^mawu>>S(l_s1yQ86&h||V?
z2FD8?xI5WcXB4}s%s_)zamb#w*@`JlDkIniO=LOE?&90BjOD6;Qt%7f#OJc-5NJpz
zvfL-ZI?e2IF1hW7#-F0B2~P!rexuF)PV|#=WV~WBi&`QJf~aFk1mOw}IliKg*&2xX
zlcsg718bEW1VTTdOvBF%#HIRD&g}n>{2+2RHp|J}hKjOx7tn8X?G{toL*t;!!?0O=
zY7vK#&pH|@Jcri_1lbReAB;(mA?nn0eRV#+A3+~@vhp#mb@?G191j0jR8*^?;D2*k
zEONiKOiXH>=QG9K<^*Z5*tY5ju5b2*Np>`r``e_|ES+tV`f>c)KbOBeaGa>boY7SZ
zqij+!D?z_e_S@|fRrcJy5}TE9#6VVbt5LT1mFIn61nts67k}t68PoIa($QKS(h)_F
z;yHa1gfEH;(pVlmfIw1JtC2d-{j9Q!?@bss`J?wzlsqSA>!o2b4F5c5@LU+V@yV)I
z5LV*FxJ@?q({FR7L6^lP>uC>_;X<|6hijwe^hCS2km?pM7q6ZNOpLKzsNYh+E}e~0
zam)h(arvkRy-`yll8ubi%~qW31c_VZL5x64Z}*mqI8LRivdozfShVqJr>^7v{UB3=
zU-~8iO4oP!MKlFddUf<gN_zfzVAKl7X$gbmeMR!n@p3LrxO2Oa{s%sAsu&3Tpc4gl
z5-Mrm%1aFrME%{HMKFHwy=is(w%Pg+I-baa=`GW4awE9*$Lgv2rN){yj|ej@O7L%9
zL$xq?a^GxgZEbC3<pcT~hC>vjIlcDAifD}{=MBflm)en&$U3@=Bruy>-~~edP7>fY
zvYZ`+N3Wgqckcw>#hQnI2GZRd;2*nlqbdtN2R~IbHY4TvOGHol`3+Z{l1nyP_d*yj
zblH-vs%mPy+}&RX60i{L5Lp-7&u2TnKm4UUg&K!^UyMz8W5>nHkak;kx1`E`mPsJH
zcL}SPW+UIG;x3%KsHl9p*ONKEcj>pIWi|G_D?XKEhglaLo_xEt`T=L=LWXkfN5FE&
zOG?)JB604?errq|6w(A9@tGx}OWwzHfWaG1*w$EIVweo+ZY=K;i2A(vn+u^Rg7a~z
ztDaXL&%l%-99Vedmm-eIRn}yyC0FrkoISyW`}uYP&B_%-uskMetS}O#(-~0*{A209
z{(ZV@@3?H*iurkbte{Hz=10N$8T_I)C~b}L`8cAt#~qS4m91glsuKEno>~^$IXlW^
zh>Ss)GV<@M>9Y2x4MyTX2Co&vdj>MQ>PrOgl@}HDrOz8O!FbW?OH!~XL(}Sbjk^o8
zZ~S3Ke)p7!05-uJj#DgcA*v;wz9;7w*M7{NErZo122&Lt{YBict)INl&+E#{%690W
zW4LXvl|P2bZpOcV|6ZG&B(9TkOEXUw7Z(=?M2Bj}>4~YSz9%QCn?>N~xtp6{Tcb16
z12G!w+8P*4k}9U$iong%Ed6U<UfBGM`i6CdfhIgXJ)L3;0)gnf(|(;t=i%Ye@n=_+
zrPZ0&%@Vh5Wn;rq!?L$rAPpzrzL~)%xE)+5x*AQAC8IyQZN4G7mz9>5HYO&<q3Wf0
zf(LYab@e`H!nCXz7a1q1D@RfyOC)!QK-xntG7J0J>KBv0Mn>9`AAr-ad&ZGu&e<Jq
zwo*UvtBLiknvzZf83UeMU(SPb&6TmT=Syd??#pK$lX{t)$hbi8goB=6gWHcTaQJ;_
z>iWWTcASKRr665dEYLE%+7s-uHYfTDse@EIg}ShC@S12Xx%ImpUzEZBGtMwk(4aGW
z{<V~l);GUzeBHl6h$&Sa?^y0R{(lFae=a{yC6v4|hgT2kls0Y*_Fp$lh!guiZ<;Cp
zeG>yMjIs<Y@;}i{kx$J0KW`rM|FaPg?EhLnTIcuIv^R@?2QU2H<^Q~?()YjKME%br
z6jEuouHU#Yg6}V<ddj)f5?5><wBM{wgrWR~;^)*9*V%o7-G5e6?n{AsQi4stLhg@=
zhQL}a5j^J2Uql4Xv6Qz!wC%DK$hSiT*rn%N830X2{k&dpmmX$T{{UX8$I;=;9RH|b
z92l)VMYtJ-!?8ROuq+W6Fla07u*g3D(s*El3W+QR)~j<=;R*fL!kQMfDdOeYhrhSN
zBzmhkp2CHYIA*<mApdE;owWdE^|Lt}CIx=>vrM2*-Ij@08NEyDVPEXhk>=a?>obu&
zS8tD{fCHtT8nCu*qeM-(B^g5--wBqs7)`7q(~*-DIl^1Y_=L5cn<_>C)i&LlorHsG
z15fQg1_CsZDZ!M}zt0osJ0xk-`19JlbLzZt7i~3;c`TQn69UJ)8hfoP^FdT@s65x9
z>e%5^u4H{iXe*6D8=q^P5i8|8RThdZ%>p=ZA2wj!CnUg-jCjc7I>(>ia${1^%KhTU
z=vnsEVfd5Upv?J~T=sCY7Oh6V<7#X-7NYzA%O><7F`a)LDBCl;((@_9zZ+q@Lv8G5
z6T3g<vKwO))RzP3aXXkr4YpO!$=E&v?cML3lK~MB36>Q3e4?8-Lc|HIs~M?u^Zp)f
zI33^Z>xqEv9oD4hMzl}xZ(c<WYakSaK2yz_TmeBy%H$#pD5cork)(i?oCwG9^1Eax
z72HIMMI<-i0wWwtka00y0Z;t$0FKG{?V*HI`<s!!q2N_nbGvz8D!=U2ZruF&pE%jT
zlbw0w9|e!Hda9vxg~~%oF+igzcJwtwm=d~z(doo2e!LO3|HVQSd31DCU0tn0AE0;e
zmMQS%jp|XrIH^PHw6bLct@{{4ap+|-2?R`AUP^A&S}gS>)t(g7$<2ab|Jqc}aejx(
z*?`<JdTp$9w&{E;dNObAugWK|yzR!9y@)4Z>J@k>n4)W<mIz`h{s&HKAS9Ma;CAXg
zI2jdn-U$wf-Fc+v=-@zdY4jQdYam?3+#(MDcq^aw*MvKdUJuG%8*Myhy!}951gxoA
zKF;O;t@#6{D9@PSjp;L$#ie%0U=k60qX80pjT%PTeW&*p^3u<IlhAQh=^Kk<A_8*s
zF%_31KqbVCO-+#cyHaBB(%c<jb0~lBn8Fr(-)V5~=73(e^WmJlEO9{ni=I#cI7Z<@
z2hYtEe5{-4HREH^8B2P}rOI&gJ|QN?{$lsr-40TWze1!F#fI%v#d+Mm3HGXT_R|;@
zRtB|vFuK3jNBWs5uc6+!mV}plFr&nnk_}&mc19eFKN-ieC(1UeKye!mHZU8@W+NCd
z7NSIF{r7t~OWl3=V~f^N@tCdu^os9OX~A~GsjrqJGK{_Li9MHrS2h}+nVFfM*3@<<
zq^42s+uM1|OZzL=vL>LBXS~v_VGjjJcnu(;(AEd3d5ZQy#g`!}TS1756T*m&eJp(y
z6?AE5p1Ak#=|rN7it=^uSNjR3;Y`Ma#vdJdBK$bZU>@sF5PoUhORjyCEF9%Yzw?|<
zc8LLO-X={#W;CbB)W@G!5DJC%9iPQ_RVI6QdBF|UtDFG3(;-=oPXnM|hc(pQ-JNfi
zvxr5o#_7|mN)D=VMWzH*mW^3w0&C{FD#?+vkn4K0)}GFi%dv%zJQ)OKmlMVMrLBhU
zP*EIa;7d|9*xp9vw*TQ$*msx6A`d#Qtf@I(g-3kV?{(0Jj2iKbfw<<(D|CkRCNm&3
zgZK%;9LUJ*f%<*@`W5H+iMeP-KfpAe)geVif4$3R5R%EFm6#G}O`f6L1Y3aH;05??
zjmcJ46$XeR8-phb3U}HXi;BYK8LQnsNlHpmQ&Ig!I-2$R_Yc{kMFPTxgjPq3_3iD8
z?b7Rvc*vqj8^!42M1_Rj6)~^6KTBk(#_6%H_-Xs@Yry)-imjugbuR-#@s|*H)}xxr
zO1>%Jy-U^|HVr|jwi>s;Rb-U<^`KRDdRqF$54AhEkhdRyHW;kuigUk+=(h|zE}v#^
z=b2i%Zd9~oI`?-9i!_3=*0t*DL{6(8&HZv_g7;b)<j1Y~JFzxP%swrSH1c+8VK%Je
z(}+QT-*7Bl;b=7RIW-_w){867V+N@A;obKaVcOn({GBRUuAAHidUOOJw}F;-()qoi
zlYVyanBl=Kd<qc%3HxNuHv~;vP=0C!qu#yIWV1P+OjX8zBJ#gM_x~1=U7A-ZYcxjx
zabPa@`;useU+EFv+z9~&`ETHS=NEWSck-_ry#vB?*gyYerMwRe9k>z-5WcZ)AOj%|
zmY_NMUnk7|Z(#m6i+|-JYyKmk?;q&R*YoTmjBZ=v*59i?*Zvd6|8w<s!vC+2YUZH+
z&)!?v?O`NEu37(h&Z*A-1~3#fSpN&%+-&vUVpZ@%Hj8xYnUs9HXUTc%5(+QtOEU5e
zDDtU=G&S;c0kB091OV8AU>^S6lb1z@C&i+oxQN3L4U*c`cZ{WG*~SObQ>$b=zyORp
zJb?>^`=v2sL*hUXt#H5&0pY#6cpb!ZhheL9Yqqy$#doh*0QmJZOtzv7*C`5s3uz{P
z(q)vqUc;#@MxxdKTQrZran0AWY@2sxAu0>{Q^(i&XSbD?(Gln`qa0jFAxHbCsh288
z(sNr|)7k`r#BNKXNL0glurd<w0uT~Sj~79?6*grCw7g6H(0YKu8ulT!kMa|{I(xfr
zoI5nG!`z2-n6)oK7>7Bf7n1lOQt{RNjJw2(bT&7%ur@O#9SS7Zvg@7sm5lG6zyXWV
z40aV4QDn<uzFkq#v!Vd(2(Y_ay{<w%^&8`|6exT8=4(I*`ag!%4~UazAJ!!d&yydr
zkt*#zd-*paz9~QAC!XptqW`N^3W=W>6~bv;+6j4hkIsDJ8U7wBPH*|qgneoE+k4R<
zq52C9|JCV#aq54QC9nCodRwc0snY$U(OVY(B$`EcXbi_NZ(jXBn5CQb|KC1Sk)k&4
zA9eqii%}hYiT-c+AH5~fxc)bg{HwqJ{mB@2YBU)Cn$CYOqAUC#?D9#XWqjqYP5iqK
z$fjQt#L10#Y+qbv)BArsqVpOajGg|7nZ*q}1x71I1YyOv$nt#MJn_|eksY(lHC)hy
zDk<VM0Q>ZFGS;66@LC<Lh%MqI9lW`wy!t+$Bx^ojVq^k0t2VZNPx^`@dsfhf+5N(Q
zJ@Pmm@51WhycvWXxX`|18CP%l7bBid(?@WjHa>vBYW2qsTO(Yw1kx;F3F9Sp>EUJv
z2M6D#H82valABUME=s9q##EtHO$6*EHNQKvE0Luv#v7*%VCJHp;Up`7C8d!`XdB<t
zOml6s@eKfhL5@o_F4XTX(Qkwec%Ohs;F#-6;L-Clx|@bPDQ*}BV;*XWzXfh0*3;x9
z^}OE#s;ZMuBT-f-&0saQmdpvNLR3{jZC0^qmSas<o!4395^K4u#OI4vj{wr0^0q-z
zB{>Aq9&LP<26z!|ILGPv-JZ;hiVz5hqMNVe%BQgpAM21*`;jW$Sje+vR+fZY2%JQ#
zw7ThI*#u4ah)@h*1vM;KkBQ_-l*}|NHR19(1cuz7SwME@<-@BOW9(p%+}2pmCCS3<
zPMk(8PJ&eyY0r~_9VUMRV~12MfzjO((0mI+Ax+i-XW|jCv(sj>;dlBb<eVInyObGa
zrQVT3`DeB##z^H;t{={&Ab$Y<09dmSl@%Rz**F9;ili=uHJQKozU^IVo#!(|<5wX&
zn+9T6LlDUWLcqWX0*OIJ@z8l=L{}?cmUswPuQ`l7T^@mh718Ffklt;{lz`y|A3SQQ
zFDWis+naF~<LVV?q)^t&+ro`#OJqqp8|A}^Fv$wew6dQaUUf$<euuBdDLtQm>^5|4
zd_r_Z%a2=*CRDuQ<g1nIcA%5gJ}zTVJpDs)OY$Q#Y$Q!NA`0G?I=Yuqo&VjCTlm;|
zRT$7h3>fkl^UQr;VX<~!r6>TeRDWX>@6t=Ocmoca<0ichT#mU;PX@2u&U;NNxJ(p~
z8SbZtQP>q{^G*Z4d_tQSchutQdPDg+OC&vh$iKT`iFY$|TOGaxoL}C82l@=J$e5|E
zRgI<9XwqNI7|xzt+9DU$4I&}6uJdqij>+cLFbE|uPcxhia?)IHnYogo0PiQ_H^tS_
zvDcqyJeOWNb9*noKBSRIy9l?)HYjtfvA+)FDA|@r;#MBwVq*0#J+X1HgF1XI=w5f?
zVOl=)c-+g9vzHMAxzww7YC2{jlg6;)8GX^K<@QlglS_0RH2Q$?B^BJg{qCD*!n-fg
z$<=YTEo05a_k20Le?L30ikpPjlNEeGE8docGS+>W&MCj`ZksL5YiYlpmli=n$HlW+
z5-l~IU>^_S&)WOX<qoz=9JT{NmXD+YqzYDEkDu(#bids`U6AF?9C-^(Xxk2aQJWoy
z%xf49!M2_#8(+i?4hUaWKp$Z#5$niozd+KZ94N<t)H;0yb3mA>Tm%*xJCCQDH1S`r
z{gja=K}rt=`Ph45J_isd$J*Qv8p)fsqPvSoq6~F6^EP=EOD`59OXJ?5jWP6BArLQT
z)y_}-qpL|n{kWa$Q?SBwiYcBj5?%i-u}cK6R)g!l3%bME*fR4#<~u*1-L77z+5Y8w
z$m5Qe_;fOv_lx)9xFqGu3^AQ@>D(1eK7G+;0rxdF`2FYjBT|u6FLf?9b^9ALM7~~0
zP}7zfcAnFJTD%pqV8T1Q_&=8{mfn&*U;5d3vC&FSZW7QpCM&&`8qQ7IDN|N48wcnR
zAQhF}l&AhM??1J$*(OJMw^tYIry;WA63(-jZ65<=sle0k1vt~YNo!xQl0{V`==C4b
zh##fyJNeB|X0bh_n9<WHM#~O38b9A|HaU1aA_H=zi%F+u+-wu?`7n?5U#Hd*8T~T>
z$I~{ME<fx`8xa{lxecq0{5GqOoSM$oIH9~^BroZ^EnUaUt&9zGKEYRcn@*UvHQM+l
zCMNv9@q|8Pa~lv}zPQ{DGO2n_%DA{JdwmMml9jgq_2tE6HMc=|`}Oj!@8watustBI
zJ><K7#85jm>?)t-6(~pVwG(SmgwHoaHR0vsU#>m(*?Iad<4#83ZhyUS-O^O)Ae(lZ
z2;Ei5^4s>?43tf`@jqQk0gpp@c?{^{ppCLlA<g}f6;09UgCEpREz1DIgPhxbAJPB*
z87f!r+3fY@tOjqRP};*Y(}97hI=AJ)Y<$h%Sza@zTd?gj>$8t>GW82loL6UK&4E{^
z%hJ-)fL7m`$hAxFr1hF2){-*2-Yby3Tm^J`*ECto0N~7d>Q{2r-m+bSTf&1!<|F90
zvP`G^_xq6o1YG^x>{r7YV<22Xn}JLTk0KSayXAK9e6X~qr>D%(Z;PO-{p@VtwZi6W
zAhYZ9+1bGJO)Z(z#biTA@$-!eC0S;A5r^L>v@4~}S7&>@yg_rp=opL|$y`2r-S;pV
z!4{d7-|XlyOG-%i`tFgFBTpB6##b+8Y=f%jqnkCefn!iFC@2yHPH;!HX>fh=-u|u-
zzFEG$)09>!bF#SpE(m|=7P3;XR#ZF(QqHu8*6Y8%?re9`2l@Xpu#rAn`8Z>64XOu$
z`x&$ThokzT@%P?)FQi(;)`2^olo%DRpMl*w@pADPH=Hyc&-j5;DQ!HOhz7EW1N48s
zW)sh>At*7-ULA?{mcRGgs#`B@Zi2xkg*4+F4i63%t>{S$?q&P$eMhI4TK`hf@C|0w
zn>>4cbzC#;NzDkD0j_72?7Xq{C2meXbqhy$U>oH?o0wz8%vXo=GDKcHx3`B~9!;9~
zuI2&Bg8D?^$B!YUN*@XemeXwBp8~Pj^rzpQ<uYly$cdoy{SzY!NF<J=y{?Y4lRt4b
z0%^L<fji&Pdy~2MTMr&<NejQ2s61RB1E@D0fllwgGo*k|8NR#tRx3Tf&T+aPpp5Cy
z1jar~Nr2)AA=m@TBXd0ErX4;gzD(oXmI)p(+{|0T@pSw8YWsW(7Iy_RYYX2Wkko=#
zYG#N4cGky|<k_1)v63Yf6~s*j_3o<`gEAT2mNY#XPQdjSUrrm4YtDG3B&e7iUm)gX
zVaHjd8rOUKX28bve|<GPo(sWm-RUU*Qt;k)?P7m0BKrg5{>aZWU~S{&U+!X)xqEqC
zUo6YcmRm(z86;bQ+KTN|t&6eu?QF2Og6u%w)4e77pbLIbB+)E|-fqd)%y{zU=>rU;
z%GJD>OzyUHY&?7d9rN?`Z6HPf8#kY*v;$Na*JXJ12sSy7gT1{*p493So@bRS;na>O
zPU0tb_7-f@*}VE~?d{<n0fW-VMm3JpV6TP5=cD>2mt)OWcbB3{l}rKGL@nyb5;{aE
z$PNd>{a`l%c#uJ<{fE38#sP<;cqJny7YpQoa5AWNp!5RR44i*;wMZM^uP+AsN0Uwp
zqR4#wmuD@t#~m+NvQ-{@b*VSS#DBLVlwd{O)5}ZYr|<RU@oEZPnN=?ty9Rj9U9f_v
z!|$D84PJv}Verw0j&&W_(P<A6KzKpRSzgA#!+j4yzut`tVhQCktnTB(2^P06sO7VP
zC%7zM+I>$io$)_N#@6}`3p(Vz5YQCNdPMKHQD#3@a=GagRJ?u$*p@r}?Ci+}74CDN
zPkT+S$^$Qb<-2;4I58Q+&89pjUnE<R*1K>lXE-&LacRJXAP`K(d^@_pllhSh>7XnB
zEwW&w34oK8IUIT1e6gn?ud1&<7DdV_bFtF~=mEcbed+lI$x2X$#q$_)J&`R?dJLTk
z^=>>E)-d$~!l|$sy!sgzSP3dqytx;v1=&RMz>QiNgqsbueq5LM6r$3v&XM#g0rl(s
zp;UmHHQ~dnNNryVgHF_~uC4-~uHol*9*RjkR{+fG0hx13On-*xtj}^*C;@bgyZ2d{
zxssL^J}nc78vDSj{;~hbYTU<fn0z=OxxLYReKjN_m(y=u5&tv(gibgPI4SXsZ*A^S
z^pA!$j?iic1Bt9h1@KS;O}Nb23iFmSz3&R6Eh#u37wDK^S|9M+iWsNhgmV~x!oG(6
zbiLmc=Tq>Jw|1_5_fD}%-2M37`Vv_njSua}n+y#NC9=YUq&HcAzVcvW{kIq37cmy<
z(CFyll-uA^KzB4y-d3KzZ}Q7uYi3@V0E%%3#Dr*=j{NJA^;P#XV0a-+cpsMmSp^qr
zB(}B{cC4Kb2XT?6X}tOc_2p=^87Jf7<?$?f0X8aLnV{x=P4nd@C<|73?=FCwcsJx*
zslv_D^Gnv<^KPF|11SN$$Gn%>GR@O3<NXparIn4Bvg?n^$2UMckm@mk3z?PAfk5$I
z$Zi-M+R8~SZ~|voQ*a;P%;xA;wzl$3hL7rti=%&JVVlP-R&sHI+G8A5t(lN!h3FgO
zlgSeB1Zjdcue4qzuK~pf0+As8tNSXuNlEPjmb0#IiZBV(8^_tw*Q|z%29auyHSv6i
zQs87SinHXUf$_HGpG>y_gGOIr|LQTDXTKT=thJe;W1E~fT4M94so0fV*b}r{a(6$%
zMcf;ycZaFb?x(5e@ibq!I#2T&)y7|~>14k|JbTaQ<>@J}v77)+e-m!@TM!gNh57+l
z_&p}F_v**p`Uj%Z?r@+w=zxao7X?h*(Pw-6bc7r53~R-u&(o{jSBEJDtv~7Bf2uGC
zwMkb#%8lm9HSrZH<eG9xKSseO<9Z77<Q{+jbB7a$@ub{m9|0I8uC#~pvU6{>WP;1T
z+EID@@E)lOLrZ&-4WjJ8dueJ8F0T6o@9%ru>pO}JczPEt8=S@2y2<qfHhh11uz*8#
zQa~QVd;WDwt5NNN$(Y9qjdxkm0y5G7?NG-k^Gn){YoBJ4gNZpcogzSZ?>e5;jw_;!
z!Ar6%pU%b+JOh5_E*IOJ{TNSQ(B-<x>hVa@ricIht%vYWg2z5;-xGjq$@}<G0bZHJ
zsXxJ7<$vnBkoIYOS%$`Q0K*C9qeDTs!!ikK=ew<r5?LZR7P|`_NH0E|rxDq?KTdE3
zD&49MfP6RuJ9?hZ*6+Jb#hF496nR%A>P)VMuoFQSyvTWMJfS=37I1ZO>TZGnE&|b)
zrckp*sstB8D<Jk5A&_snI<r5M7S^0zRocEAz&*y!rKbZC)?zjRL2LGM%WFUxlaVjT
zpkz9RT2x}?vcYFR(<_xvvfp;_hfqH$=M&p>Wl8DKcu-UhY<YPL@4LXeA3)h*I-2kg
zq#rB3)fsKkU!SPx>yxio{gi9T(|Ho93rUi2-A4=vYlhs-V;IbPGcg9KT^@O?m8Urh
zngfJ2IVXIJim0$iopgVz@9i!1GRAtE0Mi0KS$emfcvV#Lq2)1|32?;dblxb*55)fv
zWR){z=wD<O)4f@v2~1)s{Gxc(DSM?;n?UqLV7P?N<)PWj0b&?>Iva*~O7DxQTEf^{
z-|+Ql5()FKDbuk5DPS}nN`T}73MSe0ARu4NbG-MZ=~CC{UT=9W`%S6orCbJ|pW#u)
z>;>8YE<@-SC@k5J{KP5feDXCRv-;Q0oTcB<r_^vE0I6Q6CGLp~elc#OVl%IE!mPd6
z>&@~S8m4sI`VHHU9}zOdWPI`B#pT)IAA6V9P|Or#-@<98MYhWOL&g{T+1HEq?+65(
zW*Pw8rj=Rd79R7sJRCC_TyizW5!>(^`hNMik`y5`@WFF`;<BWq1Vlz>P@~!5?M$Bp
zEgNw7W5JK}w5~|N>praJWAe&ze2mDq<_#eS_8X=-Kfhau_k>3Q8AsvdMw%W%<~$SI
z@hshZZ*-wVHXqz{wi?Tg+U=ZpZ<zofIQKmyD+EZxD#pneL`_g5`lL5reoRW^YXeGz
zEbpugz8y)%t(U;?dZvK;S|s?wqPE25YGYHdZyuw^aroOJ?s6>XGNn!MEz#W{!9Yi@
zfI}$@BF2l@UxRQYxnhuob6Jg#3dJ0GGn{`gl_QamkXX4BDxagvsFAq1V_#^vVsl>4
zQXYVbZ^W#oTWFfomy*$SbP$U9AvbS=QDSbMw|A>F@Q`!uln<xPI1IjeV3h(@rsbm{
zru?#qYob1HI>Ly(r%j@iY~{<}rBLW+Wl(LBsRH;HCvx`mxDQXm%|Ma6w&goJ0aln1
z8&wBS{LyXe>kXPLS7r@|Ky_U{A@mpF9EhQ_!lE7qIXQ5nkHvI9|77>1+(rl5g9uDn
ziG_I?^R1PYF8iwcGq`ZC-<=y<TLX`^c^dY5IFBM*L}$G395GLn0+3BQEAFv|%REcZ
ze<x1*h~l!ZMq4~onYQ{HL-xm?QZRrKXhj{3etw<z@1#U#D_Oy>UU2)6d%Fne$^E(#
zYdIc|;?em;3U~nHJ8m<ZybPg>uHxtJAzqUD`v~C;zcPkdyI<JixFe4mG2G@Un?IS3
zl6oEAh5pe$(63KREmh8kuL7HCkx7HDjf+vSmPD76M><@4{rM5|EldKxN#i+pZ?2@l
z<w1b>(;Ppk2C=o#86LyrD(n9ATxr>+NYhgCa5EA*zpV}dJA26~1;p*+7>E)zd-}JU
z8DCD8umHciyc?WOrh<3pviPp&@q<1r%OcFu5dbur&ZMKZBtgkrFMy#Z_25hT+yBAl
zZQuiVCPMuYwRTghKU`Dv`yP)?trsSg%xddi2pJo%!~1~{7++g5(cUFw-1v{S;_6S2
zZ{}&dJQkf^&!&xe-J!fjYg&fRvBD|!LOyp_e&n_U^eL(IRl0m&R^^?sFq?g86NCoi
z+kf#ER$m|WXopb#SE-4t*VdF%^Cg+c*AIN>4-5@6GRVVsxKV*`Z$?Ie1GrX7kHxdA
zlOqCf)6~Wb0@l9flMdd7YV3lY<?qnceHm<WYphNF6NeN6d3^8rSI|_DADX1L8smJO
zf|cKqEaHjIZ6rwlCUjqFIw8B$Z@aN`nyjj}?LN1d3t8AWzEKpxaW}Pm*Q>kOo)lbO
z`qde|ikO1OA>Xmz)_g_-@QqWyBvZ^~0WammrN;2o>Q;@tY*BS_aj%(aXTnQZd7VW%
zqWS8uw1L;Ib@<bcd#rdI$OJ=S+%1|~ct^S)`Obq0%<|8#KOd^H=pLvyZEEm^iIYX2
zTG!aBM3nMP#^mc30(XvP({3c_xhdnb989ai)jLscg}t#@hz<hO@_`N(bbKoM*uW-h
zDL-$)aA!(L=GzYf@}Se;7DB-?HF*Vvko_bhXP;P#4Z;Q6G4V=8rg?+_8kXDF&p_Jk
z`)|myKJVYwq4i<(+>A653X@K^Fa|)PVe|tXd%AlVv-Q?p)2*qx_f4Xo*&Rb)(m5<&
z{t*jqy`ZJTB!%?YR-PC>xY#Bok_RVTubV}=Jo3?xCV+yU{OgYYCc-MsN^)}aap)T(
zSpW}HuF$qGj*xzLe&d4*_+)*wwA|6;l8XI-q|Kj(ZCNglO-((DoXtw2-!|VCF!z-C
z6GOg5`q~MGpuzUlwa|=B9V<04c|QIew~rrqy)k*kF2~BDX>~<xw-?$%P3}bBnf0q7
z2=T%O=UEXfa%&p?pop;->@)=sZ`N7)xu*8>)t)2AcB5MHB3rDT07GYE%SJK1Dt3^F
z*UN@)yng=4^C^8}DhA06(>6ie4}4GC){<vcSdWGg_9Fq*tUV<==Gy12nf=yK5gX$a
z@1Rfvz0!DJs*Ln>te`x?W#a|elev%vrzixoiGIarmr+1YT%S7Wc{4+>;t;8ICUtGj
zQ`;6t3ojE|fkI5^2fGNZ8LEijAHQJcx!TJFiDbaZe1v1k!Lw9s6Gq*&9(Cu(pODp^
zv{I;q>$F`#)=COUcG>-RR8yxT*xK_pv;^0LO)E#<ziQ<?o$&auQue`^`YV5?N*{?1
zvwqXuYjL?+UR`5*<wC-W>o6Ij$c_5-693I%VMlixJ$90b7d6u>HYY7LoK)<*_VdxY
zObM^?V>{&=_|(Tn&wt=q%pcBwJe2!+LbkQB<Jj&GhzeJnamw@gwSg=kT))`QR>_Wt
z)SUarHti0)uZ`cNu8u!H->fkaqoEVB)n*~e**&yvr}-Sv_|w)MfQ}=ls3!Ou7_>B)
z%Pr6=o;*SwtFZ2ntxsv)V*a-GHX1%)0ixTI{5L0NrD)?iv3Wm`CT~-E;(Xw%Fj_?u
zE!hBk4FDEe2R6GYms#yoVMu&ksX!3dNgUbUeJ^_oU{ZNAGGX(h@4Zx!isAV5GX+}o
zXSgPHL`!rph3kE?aMsAnt~o86Uh?88@4HrruuNHTog_<(#Zj611u;<{JtfIX&%uA}
z%?BI&&!UP<{LXW7LCSqc;iNNbJwHnVvN;>lAc>y}9~TUt_*O!D#DzrRxip(U%*&=d
zCpSNY;N{M~I}d+mqn<Q~NzCS4sf)3((HGr&vik><G2r|tENI`o-u3np_AOh=63UH;
zII?62>^#EB^Eb375UWC(jQHLot_B7%1EYf0%(2r;kOAtgsasEB2x73z3P(AHrk%eY
zfA7At;MHQUU+rd__sG34Q@6xHEpbChLx&_btVqwYGnRH8v6~t^WNF~|{CTb#Xh3_|
zKw_;U#*&pyw19uq_1<&B&Zl{Tnsd8o6Yw{y&Us!ksespmCVomzPDXM)fofp}wUVFc
z!hF_73ITZec9{WqHE|j|zs=X2?p7gkds&0*q}UW4!}oUeQokoRGA|@>PKx(#v~Gl?
z2$MxWXieOlW&m=VvMJyH;MR9}e(Ypt7tA|<Cx(FwmsoggJq{WdTK8pBH`LzJ4x6D@
z09v3{mcqG>#vRDRQQ65vf)&;_AAAQrmQK1JbX4wGIiDFHviF7Eri|`*_Uze(NfPeb
z(jbFwvK7S=fr)ZoUg-EBq(^DY%lZ3*)SQI7yb>m6o?Qm9Pq1>Bt^H;l0V|Mfl@`l~
zC$-Z8dQL)MI@c~_)&|Lbhii8+dHl5@;zC$}Nf-xVfSQ_mv)NbFTm*-iEIOEl{3|mc
z$Z`WTad<Zj*m?ylKXdg8Ev)n$#p$+ksSPtm(A=JVLnH25kmb4eJxSWE74^LR9<8a*
z^OTADxmVa2i~(l{;@b`TENXjQz3M7`M|cG9yqlCGE&N;nw>mUWJFE2cC{0}Qr!C2I
z$#I#d%j-R!J7GHl^QI{S+AS%7@jI_|`D3d<7NeY%GFDD`m*G5C=FB#B9#)po2~MwW
z`MuvG5Qq*|e#%Hi;t#C}x-7I?#EM(bH255uNYnDi(aQz`2IU@W<<4>Qb-?yaqxiE=
zo~t%WI`W`+YaNC9VGMHKYJ{;E%GC=*5bVUJnSKBm*z<23=YVg^aU?DeSnT=9ej1oi
zKAC9cGaGQs8p>r<>kQHqdF4LiWBJ^l?3)*$kBID-D9=zAqTj6TwGUu0`uqC>V^f26
zLg;a5#J;T%y9URm0{-IEA-!41EJt!dhZptN>IQhghTC3tqH!R1KX1k6)gxggiu7ir
zzEaR2S!9Fq%VFq{Sb^i?X-G_gt{L_eGb=>kv4(C`!XNDvdkJbs<r)Fl3xr%I+In*;
z=aHNJSc@@ZtvE?_&a!1*o?7BdBxmAOF|RnOXJl5h_(VmJKWEH+BU>Ydk1-iRhsYCd
zzkolIx=1EuzO5^Pd2{x7d$t)U-W@cY*NRNW7=pBQh=69W!#7mIQ6F=3J*^A^7ZX%g
z1b^ETaQXjnB>5)E=;`eeCYH&D7K4W22D(3X9zd3cz9t==l7DcTZ30vU-(`v)wNfLm
zcBpNYCJPZzH7}-wIm1kVcCKgqwwEyXw*!x1ptY!T3te8nTFc$jldK0n3%g;fe%%yR
zjvT(n5d?CQgBIK#{!SVN4Dp56%qybCu=1Q+cjGPr35$3B+!Qe?X+|gQdyuOXz4ydQ
zA#!;VyKbiydp77g;KYC2DuovCT%OTG;dqn+q0XBGpNo(*BPB>-BHqgcob{y%Vlw7v
zuK*tPo>3T*>)%{mWem*(Neb9k!R0;G=e)5eC=9)L5-SW!I=+W%havPAFL|6l;-(c8
zmorjsE{`r=wkJS}pCp}OBJTg~H?yVJe8PJW%F8ar&CUH$Ip$Tx*5*0(Bbif6UgbWp
zAM|mV{n3dDb&S)|(m-MNm7z8~$v1vJ?*Yp`?KKPf0C1C&X(TQc9iYVXjU`m%i2IN8
z8_w2>L*uV5&)ZI{hdw;;nsPyR4@AeK!3a^6s9%6w4S=(39ixQ0^$S~`|KgHY)zIh#
zt#2!ZoEg=&i7dEOLKs(v@~PQ@XLe@>$xGf+fi1Vtk*rr#?nA&qFLot>kR}V-?Z{31
z31v8=hDSzFlW3ZPrKC?6-(rr5Tl)oZ4(=*h6tH8tH-ma?DHRak0icKR-wfBep_ofJ
z{N|JI0;#9rQ?(Yt*RA&pGH>@U#WO0HMnq0Cw+7gxU&iMXGApwXVg9DqY7R~B1(=ni
z1LUbIz%mCU*1sT48*tGxHrW9u6QsAL15r3snG#MGqIpf3o2KW5%-b>A_{8X(aa(b+
ztR==YFeqgM8(7q|zn;Xt<LZ4@ZW&orG{g|UjV%ae^=-oy#i0^<9o>Od=wddo<m$FO
zz#1wfD%zHS@og!jgf}nfG&v+THn!Cjht1Yr=K08HQ2DXxif5#&D>v(BEPL_pxt*QQ
zX4Tl-Eio)NiL55zb>`eGESM>>Bs|vM-G2-Mu8i-&iZ**CkO8s$+41&Z2a{<!TV}w6
z>amhGp#HD0d0x(lcA<IH@2Oj&o8`B4YO|wei}!>V6(I`)Wl5t=0q3n>Q2hmc_AKSQ
zewSIqNvyGN3A(*y(**V7O0T<+ZW(hlGlVTDN^Iw4#{t^0%%ZdP&QMIO?8Q8O_c8wW
zx!a12R59`vjbIb-RZD|pi!zFb?Y_&|J28OFs;sK|zT9vsT;8bt02<q2UQ{$3%^5Mk
z--hr`Wo4Yq-~vfVYxx(zKRZoU$7~IhW4m#Y-EjluwQI8o+8;l81ciC}lC2s+&KOFd
zq+@Du3+jTx+5F7VkX0?0vaQ(azCStRLl}{K{=>ArVw1}#0@qf|dj$B^yE=8WA`TQn
zDW1(k2`Y4$^kTkA-?MR;eojn0rzgvH`9TRN$U`s-Asuk$F5e|QX4K{tpsla3&qYSf
zXNodX0u%+%2Y2u~-m$DA<(PomnNS}9KLxHTM;kPU=O{$PQj6j>MLa;6=jc27j;pu~
zx<!M3t0fA#EGRl-jC%>~f+w4Ai=cNOdf+wfLC(1NMsT2ZqrA6;hZY83=*Om-1y+UZ
ze)(xDM1vP0p92aScunYWoO-OrVQYZrfL7WPGFr~H1oOEjTGf5Kf6oZ!Yqe>Cu=N6g
z0C;gp+~h#~JdMOpAcuG$`Whw~inR}DYZ?EO9aR>2KNR=0*W^%o=or4(+)H39p>1Ys
z*uiNZY=wN6lJxOtB}Uez@pyoC=4;DRkRm`?jSwkES5V$@H9<`krI5Zl`Lj9&VIiUv
zu<Qcerr6IqXsN<L9meVGU=?1V#VcXf=5j5B<M=_M!@R8ZE{Q&<=>T>FlM&gfOCJ5Q
zWF53Hdo~_VI>}OW|BSkf29=4OHhg;F=Oij-)%0-BDVNURS?`^>?*ORfaeLd_bK+D;
z5h0{RwS=e|kxqC&lB;(}f35?R#m+&5Y%|uP`1Ibn6?3Q%LbB8+I2bC1@EPNbAR7n|
zM^j4#%?@u;i6a#)EdBs8zz>mT13=LOV}C&KB8h|XCLF7alU>m5bSA!e7lx+X2U<^(
z9FHKE?%l{JJ?p#v)xM+-4NI_-r^<|CC<G^>4TQc+P*dOmRRZK$Is>0Y>prwI--gLJ
zWWJ*`b_fgtk4{RADrQ)4a}~88q7Nm-3JU~7J8_hP$|p2#+GIICxQEYMXMg>A;_UrO
zin!!4l}~U>A|;HRoQ(CmF<TRMc;r_l*h+5n!tTi<$<0c$HZ-l8TQ8Lent?p;5{bk)
zi%EI!m^$0KnSl}m@<DKHTkHeQRtppaY4IXJ<xHz<2sBbP07fn7q>Uc1MYX@Hj;$>X
z!c(*cQ~BGP2pqn@x@#Kyhei-u0H2rup7I;0)+mJRfx~E%XL()9)$1?32kgYo$%%KO
zqHeqC9Q4cQut%9ufRkJY#pZ_Jlv?CCA$*)s<&%yuK!7$I0&;ntyU!G;?kuRzG+!+7
zf)ed1&_6sXq0bc+`}br|BQM>Ts0gfk%K;foee#gJi9kM9_UZ&YR+#4%sF3{{udvw{
zenEleerTH0@>%DjCxALY55YXrI_Llkv)=;ejzq^!lb_cmDoKy_qyjHa@-$M0$GWxB
z1eO3;nzwxyBSD17&u3FhG_^apgD*9g#r-T557a5NKrjL9Xuo8KH5lyzH7`({1cj{m
zq!$1jN05g9>5QZ2Gi?DiguRl*-;5uS;1U#KjNIGQSct~InJE|=QqR^I??ftn)PdD(
zzI*&AQjtmTegr;!6a10nuKG?x2yfut+iWi528w~{DwAJRQ~r}LYA%svrTM$x%6))9
z$^b=*jqwU3%yE~{G_jc+84<D_`jEh^LKXhKOgsc2QQTb;=V>67KTbd(572235E)K;
z0_s?yr|I@3!9GzTcoG=lULk0>=^`&CuFr$6<NlmF#yf&4<QY(F|9;^y)kODEh(2i}
z5g89eINxsAhZ3eu>$8|Z4s}lm;O+qDN*SZ<Lt3Ol=d%bpnx#PA%v-le)W-LrlM*Wo
zi@F<mcwVysqU0ZUc0fBJB4DHRWd@(wt2SMK9cN<e7l#-4NeX!5;skiUSlkZ*S^=mr
zQ^-!Y&;m;=Hq|a;cz77uunx)*i*}>P3NdJ@pV-`609XRdkcGyB@PY4tXNYi@q<(O8
z&<wS|&|9j>^!uy8%d?+~fNEb4_@+^vt>*^11W1&PblzQ`W?#><_Z_@(aSpuLiIDB<
z9hU}b-WHCV$nvvERt0v{(e3jvOM+A+VLuO4e`Mq)==^uSN9A%s>xB|jTg1f^R8d7B
z__Px5Qmn!8q&j#R3}%W~>})c;134nbvF8(2W3~Kj@XTK4J(BWW-^|*?nm)e(03J;8
z48`Gx$&Zlgy}oQA=Aj5@R2v3Fv(vF3;D%yL@NU)m@l($6A6p-O@jrJup4ngN%;;#~
z{e>HqTUlB8hEg#6{OFUsZT_$lE>CGG&9k`%F}q*mpN*Vv*U5q{j}CT`(UGp+n|G%0
zR3-0-UcP1KNhzQbjCx!cAj-Xa6^33j<%-RS`p7;_%16LN)@O(Sl9;1dXGhUUplXnO
zeza-mUeB%VOvxYNhdX)xQkp#A*@h?tf=vu2UExCGM6^;KJQ_r{%J%l_AgaCXxAuVY
zN*&*Mxvg=DCFZ`hp<QY*+Y}IiN4w^rkL>hQzbrz|v<zrjp)Z^R6sq!3$NP=0lUq5m
zY?g#IjvO>IvL;Af0TTxxBVDB#H{t<3*@Y(GGkd#+6+vUGqsupJ7~gm~1AZ-N5|tAV
z_0xE<TgjyXfkY?*;KP$L_uVQ-n>|5QFR_at_MGT$^mI$6l>we7Xr|-o#?wkt@$wS&
zf8F^fD%Un0&5cZn*ORAvr?}TE{Q)LUn7kyH4zje7(9ao2%2v*z#<MA(^TN?_z`7zm
zKlrEaCI{+zd{<FXJ<Yycf8Oe!%r3mI*=7`5lgjnhkKvULBI{goVEn`?FVG{q&gWee
z*R<*Bv?xx_GIPh-R(U5+fJJolB}NC&wuUz@$|N;PFOf&>mz<pWZ*6#wAxV$yNCF<o
zzb-1mOWf8?UA1CXiEUU)Hq6|l3ZLKMY#+&kmN$;mN_ui_;wKH`#2cf-lmtuIP303m
zfnYu$UcS@l=H9b;*;S5HBlbHC@kWD5K^&mQ3*ZyZkGD(hp3Qt7z+?9+j4v#Km0_zU
zR<?D$vKFJTV%HhTY(uSfQwz7Puf&G)hsJ||i>%Sd2MmzH8Jp^i2J)=sMm7>WY<V8S
z7ZbaN?&OoVHRjdZ53o||7Jf;bs&SeH6(<7g2pJ}pj|yGjh9%R;=bthZ5s*YMzWL%~
z2ufOh;d;FSQwUmJnW48+p{X>1waT6`E1KM{`6S&K98L_#|B{mr_Z(VuNA)-$J<@mI
zvV#;ePkyG}!`{uUBBNC$BPl#QoOS}t?}XWi>wM_QDYc(s94ZHv0Dj~QxT@%=e;DU&
zfUbh&ub*8}q<#yL;U$E<yKgm8{sn9=B|54QQ#6t$x&w4E7RH;kqnw+2VUXZ^$ADQC
z2&CS=4xql3KcvH>qoi@2(1Dqs3QC?722U5C_Ix&@$jZv%Ax`t$$@vUAMY}E4PeG^t
zX9eW!_RV}=m^Lxemfo1G?y`looXzNZf+im-YU<vJB)Ro41*WzK5_xad>Sy5aKLQE+
z&rpWWb-!v$(P^>0&69tWr<=%9=YFD)4$3Y-^lfXtzL^S)i7djx!b#<@7n34l_u6Bi
zwqV5-JmR{XoM4gZFx&KG^}MNj0{$<bzDz0BJ12`&oKtF13erRlPJH7QcYp0b=TXKn
zxm+hF@oC3Q2KL!5d)ev3{t8<{W2%rV9N`+rb$qA56GKXNV_+`^3=F;0A7AdvQ=q)I
zk7xgQbtjEgZMa`__Mt;tWUlj|QLtN||D6L2W3Bs2=eV<l_dmFJj(;JTIZwSD<2a`3
zCVbX*DWtx88pgK;lW7l0?pQ|PRL>C33Z`IH#LGkO_`RDpl?`msVsKNU{RZJq>n+xK
zdq#h>(70gnZ!ds}EP{m0Yt6D9XW#g0lJqw<LsPSS6_*(z7L0gqSM1F?8npHuyFX-d
z<XaZZRWCm$6jVXQOa6_qLXko)kLKnMeK{EB1WL5u@Om|!-ubzaS?Y_ZD8>k)WN$_h
zsWL*P)vIudI6i%;)_Ki2iCx6%W^a@lEa74m!U4uNPNeQ|R8xBV=|Rk3@6HnH>oXsg
zi)gtOJe%0UBGk=wtph!ijmyJ41N_WVgiAf)h&N4BBceT2{BAW{`>!dh?IP1H$QCA8
z7nIjz9w&27o_Fq&F*ea7`!*`CJ@2pd;1PHTbQ(R;BwpxY%k(EYa&C~p<%na|NRF>}
zv3eZ(rRQk2SN)E=8)auAmm(spwp^(yMyPz$HQg`<8+3Ys_`W}$l;w4Gg}1p^G{3`P
zaNYRDyZ~w@TPA4SP!^(Ly9KsY-OM#A>Kt1~BLAe`CzwG02*mhGP>OfP8IpW%l#0m+
z7?00y2`kv?6MlD?$Uh@XX{2YI%o8&=_yxj}@9`E?->q*{0M~%g>M*dVk>!qz)7LMd
zMLycr{(<X+guX2pggKPvU_fG&lH}?*5adRe<kn)913_kFlN|??jLZmMd+M~p%ch*(
z4(%y@tKM=cyI~el&*EuO)}xn%j51{mk|=e5n$ULDEq5+!6-Q{?s)+r;F@4!Ay|)Xu
zWgV&IGfXDMiwNtBzaOQ0aJXm8{5;lRM67N3<kzU@{ii95LAbZg@Lcsv(Fm&IC<%tM
z$vmbg?o^qe0&^tbEj}Op>HjtQSrT{A5c+r$lscr&R%mViy--=J(LHACGJBCQA#%uK
zPf2=?wBW9{5m1DH?qbj2P%559+lNFBW@P3meieJ5YI?XmJF5lTdEW4p3xP0?mEc%Z
zeN+Y};GW0UMH!7l_CGj0P3|}fEWwING_9HT4?a1&NWK?!-M<3lh$p^Cz>||?m4d;c
z7lMv|G|t3z<-6NK4mBRE(T4Os8N&)zbgRG%_tthh9=D;Fjfpy3<JUefaY$jdjvW%W
z7A22PSDA-#P+K$Ag+AfAvL6L~K5UONS1NQ`f@(tJztdn;^y+WA+N=!a$ZymKpIb+L
zjFpJeu(f(%!nq~Lqtf>vwfFNbp&+9r=p+N9Y}4~r#$S3$q&^pY!&YAJOxkRkaozTS
zx^}bDDEz_h>`M07x(~`vfR?C0PD9WdQ-<1gVG*wM(1|&l#@A>N&d5-pJvT``KU`mS
zE7krYS1*;DxZRhjVV2)WAN%Xl%L?15O470`p4RyBo%CDG){LAR)&rT{oS$Dl-lBrU
zg2bvkzMo{&b%VUnN57iD#K>Tu(qH2O`Aq=;fz~&qIw9?Xj?YvsTPDC?WbHbWldzKT
z@fS|&F?64(mfbei6eFf~Zo);iD5<qPk}ZO*l8s-#e)WWKkSpncdFqQAPPJh`q+Zm(
zKt6FsKZ`(EXLH{M{VPWChGbj3d|%&Qw%@J;d5?cdwYK~KKzIzW@{l>V*4#n*-g3Y>
zgMJ@GGn5gYoooddN|1iR_|;Di0|CVaIv5ua=3BsiT5H1NXeC83dUgF<%D(_y-fx9)
z6;f}ft(})mw%P>rQi5uGe}^!E$-?<Ww|Wa<#XSrodX`X|_dY+Cny_P$HJanupyuA_
zV(EnG<ND_x6_GukTVq2{Jp7gV$wx?&t&mvyUY2{}3c8C%OkAo`KmXWhjuM~w(Y3d*
zPQur0>q-uyv~2}*gzt-Y80w1;LeL;T@eZlTt&c~v!W4@V>emR~-7O__gW$u8v9AMU
zA^HKp@O%5?ib^tb>6*BBi^uRooE1lk?zW?cS~^1B#WTEH1h+F{I1Ci!QZ7N-l)k#R
zZH0}Y1_@wjkSyG1@4LRWj@C-c*zZMfc0P*1Mny!Uo!YDd(SPJ_#q3`g)nVc>N=PWI
zOP<)}K{_r4Rto_`RM7F!yApI7grNu=OtGf{M!J7m-=JbOp$F(G@uC*;ID>CGxy{};
ztFxKJeNy7!+vtT)%<deUh8t|?upTkz^}_xGi(p7=AHBQP7@x;dG<M)5_YhAjO35?l
z$Ym8*Fum-}Ttf>N;_iL*ajjQr%mXfj-NprS$eXdo6M#0)>eyG^CCH-1FV=q8vb_Z(
zCyrg;OTc8G*}K1U2>U}q{f-!k7{)IQg)IGgNG?aM(9-hLx-F+zT-n|-hJtRV2nBnn
za`hJ^vczjn&H%tB*4P30^#*y57H7?vz*oX}za7Ra<KoPE1XHtQ&&vx8D_RyPA9wim
zUO?MMqspMp=+W4rE+gorngL&^a0I@-1q%Mg1+#<T8pfuyhxvNkU4rTuY6b8E2>Jl)
zIHQCWtE(}#sX?+CMc#zya^lT6HqR!4^pUN;1{92fG{%x&j>hbfl0Ax8v8*tF3(*pd
z&yPv%VHH>z<l7}bpWTZxnJ>koTzpvAh5W_<g<k3Z+myBgV>wr+IlnIvm_&3HO9G=}
zV3K?7)kC&Fdvo31Qaoyj(lqF>vZoJ|mvozv!(K(bZx>tVoyBPK$rYka)Nhk2MJj*H
zlMkPANqdmU!b$|770wG^rUZ6hcNsmWlmzaFqUbQdHj!W@oso|G3}8}WBJ?doj$*!(
z7#&A`cL5j;<C$fW!?3fpb$;8mBkvfI2`0g|hz)c|eEBW43&{h)EfI>K4Kd%YzA_`l
zaoj<K1~0`|SlgI(5Db%b{BaR=;(ckI4uGy!tCi*caKDfy1fo&p1kBL5b6+)2SDSMB
zcfUr>LpI{*4ltxT5Q&4oftc1D5?&q_d7f@ffo8}=9iFr9TX@7xf;GPGP1kLyt3#4<
z`&~o@$~YKG4nnQBpE99A_0sHryea4gSw;cTF!1Dg4u`rPfw};en<nATDRpqC_`JAg
zl{!KC&-8)`V$f>+A_m)SN78oGc!6@~TJ_PV`p?D8;=Q+FsUGqmniGHtHI(0Po)A8z
z^=CL?&U^W?)#&jD&Pn@Myzz%3okoNGHBKWCCUnMfz({^~GI*tp;h$eL%M*4Yx8o-(
zEC*$E9aXCS5oUk^=9X(8dvI$cy($e*e1!t`6I|l(|J3!?QB}QB*XSWcML-1Uke2Qg
zNoi>~AaN+^RJs&Y5b2U`5TvD%5=7~i1_1@>ZcyS|{NC@qW86E&J%2bH2iW_Z{lr>p
zt~uxPsh}v@yGd5(fF6^?D9M0`k@W^6S>UPYVBF^I#Lje0jOdb(rG}}i-}>(|A}60-
z+?A~svO-38f^Y!`V(>W8SHA{rTm>6sYAe@w>uQff|G$$R?B5-c2Or<4$7m!r=U^F5
ze&ss&kj}LKTK!wm9pBBI+nC1m>vexKx%cm%Fs5t!WqtbgHTCftYsH6)`A}|kB8Q4J
zK=*$wU!QwHV%Tb07FufmLC_t4bikd37G+Q@egzM5hLW#%>&j+*I+7ul-dRp8qGFGi
z%9*-9!iE-@B!L`Cy_U6eDknDn46}fey-WNJ&=myAPS}3AXEnbT_^Gv4C=*=KGi0qN
z^|cKG0SL(qixVuy)k3qwu&eAqK)QQ7A|jyh$r(`qxzWD14_(5#2}q40L)!RzxZdQ$
zAR|mWbQ_<huu1m`QIUc|Tx8-K2{jFjrl?={D<pwQe^Xjvk$$OOzgpdmv9Mjf*A021
zAoA#k?LDgzibI_k**iNJaxxmaySUQkkX_!n$L!Idjcj0y($B3|Qg9RFpg1&i_`grj
zu(bfY(2QGmfa5Wvb{jf+xi+=M!B@;Act#c7XrH9Dh)V4vMvxYSn%O5<M9*>-rxu?$
z1ivHKCJHv()J9!hzbAXDVHSD5v#3ebr&s;Y1a`=TLP7-%7=n+ia=p@HgJx;wd8Kl_
zU*$m<Y_;CHU_kqITWwimp_TG1p7BOti`AMHesGxv5z3%kdzI2fZ^Wg#<hiM(FtTHK
z`Q|$f4fplUKV0k@NUL!R^=!InXu6-M*B$<<4bQ&u-#2`-jVL}kl)`Z@@(x7pbdcSk
zu;92LK;m3AF^#0IH9HhrU4@@)tac&k`j$m3<Yq9XZ-aD8=U@M9=AA6+rNs3@nZ>UQ
z3aV`9ZIzYjCl4RZKUK?k*`1C%LCcEqC6Mdan@(>R3tds<@A&+pC4=`cH>H?0K>V)r
z&*1V&`M@HuyCyFUjz%a_Fpx%T-6;A(?ku*3x0_fQ^-~?nfs<AZNQWxJ2B~~I_@nlz
z(W!TYK8gfnVn4jS-bi6hU~m)?kGW5<$GbVrw&4jW#h54OeY?gP-fH1Rl2Bi}(#Gl6
zNAL7-&G%DEXMJr7$$J-9rAZF93JXTw=V7XP?)$H&OchK2=q`u8cxOM`7WaNsiJ)%B
zShP`z_l0x@Vbb0A%t1aDrxK4?f|x^XDSDLtv~u;}*0jcKJPLO9jLDYUymZF56pM?W
ze;$la5NXi${^DE`gy*>5n7yBZ9#Sw__2Z18C8-=V419b<&zWMo-q%l56hk|wy}ezN
z#D2P_FMO!9lyjJWQ;uqbDOOQsc#pIH+o`V7gR~m<p2fERLSj&;OI>OY4l7&qP1?=P
zO}y>)=ci|&NU1MvQ2S%Zxl9StVMp+%nRI=yxkoFYvc9_d{4-dF2>nI?OjU(q%B8=R
z`~Gn-tG@7KWD|(;`z+HQ0WRXWQCPSpex+4Z{j;^@DD+zra9yX{LEjE~G=qOmPS$;^
zU)np>D$q9lt(>bV%LP_RuW!Q@b6Y(o1&N9KHRTk`AEe6zVUZ#S-VR*VVT}gP7tXg|
zF`W<|%br0*gMi8u`y6l5f{x@xOe+_F?sU;*e&3`L9|Z?gGuqve<|T`nVBz`>tqG#p
z?Dcw6QqV7dHRMf#ZZmacL){aN_rQDrO#DPBO?A0#E`q(M$@?&yWO-<I_KHU$#2FDE
zPplWe`CE9x!PXW#P&zlPurRr25i(16>-R7!KtsU<$GaCkY)|i#hYNz*zXF(^@IeeM
z)s5;!?(@=q080(m@gS<63b4P>S&QCdzKVZftP-v8LAw+5+v9FQcu8qb*e%t}kw_xh
zn}H$YISLFvCMKSAJ9>C{uoH*M&^d1C$B)7XIu17MX09re9d3?-UV~4)UVL1iB|LV(
zKdkxnyEzZY!It3X1n4o3S_=woL5sTB1|(6us|P+gIH+Do{D+J&psJ?kM<^9RT|<Kb
z2*KstP=GebrhJhm2o4aNQ6BDHbS{6=`<wPA?U7}8jOs^5MdwMux6)aVI+H61ri4Ru
zpQ>#P-Cz07#=w2SQY}~0Yjr8A6WY8G#CZ#C^dF=-+Sp8HX-F%;$M0C+XTBova*ck0
z$o#J@P$S9$hHSa~o9*M@G`-!N4RW(jW}j}Fmp&n!-h9{C%aT0&<9`=kztDrAjakRv
zak-8G+mkQdKIiU5U1O?v92y9VP&eM9=%7(UFBsAv_MRh|(ht3!(Vdj^20C=xWsM>a
zkwfeIbn9w#(|if<6wH@>Qc_axD<_tjd_@mZxb`r5rZzmCslth|a3yvFG|TMOst?`1
zU-th!mnNV-jSbbusFutvu93W<dDx69{E3y&<ANn&wT}CRk&LhQF0d4jpaUa|JbEDI
zxwkjL$;fzgvP;ywgs&b>Q>a&f-7q@4q$5kGTdncRUGbN$1)=nWC4#Niph@o9lto_q
zZPJ4Jv;3%E^2GAJ3fy<Z752!ArE+gbS3s0axF&CGxbwABCt;?Ze2%97#5jGuu<i!r
z#b<YTmsz9gL!5T1S${<?W4&VRJ~yL#)%Mw&Ejm9H_hofoPJ3Y8+FlMpVO42`hOS7D
z|7Bp7_b7V)T96gOE()Xd)+0zk^t63?q`QBzXlzEyHeyY0nY=EwSGy^lByYN;hN^Q{
z6!@<DUQPL3LN!)?5dqR|j18^9n^VfPlnSm&C;Fi{mFV+OK9-Wnd2v}$(X3isFnc4z
zIAjON-$mFBjV-0Me8{vq>HO@wMMvEf&t4Cf!I!#~HOr6TBXQyXMg%iId@oG+N@#CE
zvvxMrjj@|tzjr?3&yRJ<fO-|o{?O$r6i(eTMouhV!z)yJ)unk88SO1F$gz{;JT0eo
zuN@Z>pb6jWxb|9SS%m{)M5CkolXQGff^E~GWWl5z-KIvAg$6%kI01i^Ja{6-(lm;i
z9v-zGx3%tiF6R%(-R!^dm*72vBJgZVOi2xu8>((u%w7VL3|MnUIDs^SqJFh)T%OHK
za^zjc*se41t=k)B4t%VcU=ZVm`ZQ=(cY|G{m5Km9$D$)54LU5M+l>n|`*-PjDojad
zW@bF!t0C35gcqQHa>dnQ6;eW%FpIbF>WX2MRyKa#Mg-cM{b4iaySys<AS5{&W;X7o
zY82n=q@NX2298eDWfNLA<Q}p~OaffJi$7*&&Y_QlBvsI3q{f~r_?UhrB>ZQSu*2rH
zXghjGo-$hXBOq-zR@c@F2kH2_%zyahT#1RGj6SQ%XYPKBOYs@AtxUUP`Hl<oUAlms
zy}<XOQATcyZZ(FO*!v*qsD-BLzk3$g*e>F1I>kG~soBf`Zi3`uz_Rk?6l1_okd!7E
zItLHPF)HfU*iV9N<wZ>cbWos;_fG;Tbz`^gIa<~g`{_z6iT#H4F^z>DM1mkH_r$0y
zxkd2-=YPY7VHLE0itRbuJ{QY?Fuw7x!J_+NJSWOFIj?2cJNy>kqw(qGZer{}aP!Nz
zNgi-#ROD5087)M<Bxjs9;Lm8tj;aJU3}}%`xS8LMveg-uO6S;{^O8yQef`jD;M7zN
zdNGMjbbzS?GnPQG{#(cl@8XAtUi`Uk!874hWhS70dTUV8Xt*o(-QvTD^N{Uf-3`gU
zQGV+gSsovbz^vcpdG{vouSz-Tiii8Xd_-@*33hB$a)+{ZJzK1ITW96hC#IC|#pznS
ze5dzL7gYCp@opfF8OV^Bvl{kqK?*bXo}L)b2Toi&JI;9I3+g*ooFIp}zyZE!|I1*U
zla2VO#Y=uBX(iqSPTjim#~LE*&D=rXOfT<`eC3exHO`d^MZ<QX`L)t*Ny_tYyPGKh
z2n?q-@9mw20)S;e4H;UJ!e7t0`_6T<s@}SR?mcGESquo`PH1>Te;2z%zq&9dCr7#=
zo7tmYx08|&SZ>wM!#L-jcP&4ekm^2v{=KgGNimxb%s(P>j&aWKx+8_y8K_B6@b9o`
zELxThP@!uudI5P;=DW-VfE?18rzQbmR}P0_R{fL`;q7KTCp6uB0Q{m!zh^U0pxv2c
z{yx4OZO2a@cw1a3r!gQ27m!tQo=3rZ#)<Ng)GFU^Ru23I4Esu)+!zOsY&V(L#3x52
zA`t_vgFd$V(0#9BotpOXU@(oYI{-bu-VDs5R`B1gdHj3moZGtL;tK$iyr-vVkDKtT
z9OQ^ezf}jd%-E$0F4%)Sv*jF&L(M#!#9zHIWA<8{ZM$_j&`d<v3zQ@Dj+}vHTJ!Jh
zh0O!czZ>cVU(L=4v7?FZ1j6`R2ld1p`hO}>bgglHMaNp4zxqmNOF6FC9@0yRJ<tqj
zxt^h>@XMmgFgYb9c)RtgodH6NfD`Ju*(M*tx#dLJn|vK2OBGQJEYS)leial4^)j1W
zzh3(^fNqNL&X6Gvg`n~~6k<6MBVgeCg$BLlTJ7=V7W99Ew>d4OJ2`59@_HXwXZT1!
zuKE<38JDOg3dhf{PG*a5{N!Zpskj7Fu_bNR#Y$AG4Sf6?z))bM!@5mk_zaD!VT-F>
zWXXi~PxV7+hRRF)y|ddy7Ff!>+eGgSjGZgC=f!p~<pWo1Ev?u~{pu_HYi|1sw9+|2
z3aa7Kd$G->ez(o94`E8|oLX^Nx*<bX4pg@j;~3MQ^ZUeO7<j@a!21Zv&g8!S8bI#j
z!lKf>=AFKDu?Cn6;zCxyUJ(%1KbLsz3bBxIo}OTjxDML8-!C^cqZQ17o+U2C2JO<&
zHZ$Nz?KX>0h&X?^6ls8a$k5$+)xn~xP)8LF5|#eTRI!~Fj5^`^?)gsoNZ<M3`)1{d
zlBRfG4m7b{cbNbciMJ?^vq6ok47(;F!K1EHZ_V)sf8W<b*R$P@_4=zeuIBjX4Js>M
zvYv{KSHjbfttAUT&psDnxw6su{R5S^?#-EyL-L3~#Gc*@mMgT~=Z1$dUR*u|r@Z=J
znk!?)upSvs-~NY&a>U(n|74f~67y{QRF&$k94nh9xN*SB&t$02Woi>{FZ9o$Ma@9R
z928N>o$7~%aym$)Pq$<m$EZFo5LG~#3#Fs|5coUwHO({<K<-((j>qZ@T<@zUyDUbK
zvPWeSwr%Y{nTsy9mlg2&vq%32PpU7SzViHCKH~V)K6&p+3_}8liR7$xsY)Z|a*meS
zeW`_bjUDN3lk-~xy^|}hGuKuJoLMJvfzGxQfvp(Bq@0Kt&?Eg)x-QyF6V$={w*|nN
zPj`+C6IUq<soqh{p!A9JY#^V@{z^=zsQHlE$=cfAPpbf|*xo;3XX<etvir}S@0r#K
zHpXNYsxz_GP8&@6Mo*1GvY8=)Hp1Z$7Zdv!1?(hz$Qfl!<s`R0Up+x`$UW~n^FzeU
z-67AP<)o(wZQDaPRW(<aKfY?Otw6=W8X|q%NQACSjB9vh>8CTn#){eMWN-iB@wFKS
zXGmBZ!DJQ64&DmDYN&6R<G*+-4cFbnre#c11+zh&n950ez?wpPuwRAf@$dZS_(x@n
z#t1s#Yx8+^%*maDvCqgR0gu89T5Y@X3R?U5rK`f8E8yt*+uUc48%x#UNy<!s|5Z8R
z+Yc#x#Sk^ehiU-COJ0Fd;1Xa;%6a9bGP(Z6l(fE7wAQJDeR?=3=E_6pIg6y%26x4r
zp4E+}5aw~zH#?+kT@QPpr=Dr*ee;5dD8?1(jla#RK{+U_fDszn(I9)jtoTd!Kyyb_
zb-0WnF8Wo?n_HN|hm|AqJs$C@&-z?ON1KoJ14F2ixvWOOjY|B&`Y9caabii+3u+|l
zgfxucPAhw%2w_ZTcA-#0C-1|?C>dg8Hs(~*<;A~qA^oP&c+-!cnK@1(-8A0532w2}
z?i7CW>)qtj!l~<8y7ZQBJ41dH!bq_{DKi0CTO6jQ?EW?2?=r*Yt&N+&(B;?AxRh9w
z`&PHwGK$yd9GTWRs+s+HBf~PYuI?0c&UL3tDd+t9qpix__xN+zrXOKad*~8NC-WJ(
z^tx5Isw$Yv(%ODE<OuGy*d>8}X7?<#R3$dCi)*Cg)k1P1_cAQ3*L~vH7tIjdqtIm7
zfE3e1aw5e#dM#`BPU2InkxeI|Spoclf&|nZt_guHwmpo|WE<IRqRcl^lZM^SgYT#B
z%sHM5Q?Dx}UYjD~WdVf^NF7<dZqiSiIKoQ1XpSKZ5CK038|}^qq-E1XLmH+u_Ab3#
zO$RYi#}0&dmmZqyU4ULy{Nm(k#->sXh^y;?;B#uP8@~wRobP@)*#TbQ1F`e^x+N+9
zw~AuW<@ToteIm&D&}-Dppa?2fPcYt13a<bfMg`;-Tt*uoHJCLZt=j@Cia#GwvKm(o
z&RorBKh$!}8@T`Zk~8B_gR6pGuyV!W%PU$4cp~#sTqDKe&_%gHHdDk})q{n-e;mb7
zP*k)IoV+J>&bb*tFHRS*i@#jn>aKVQq--z&II$o3{qSKz+33!dupDSDh2I_PS{{K;
z)(0}lms4>Tr#)QF(32-&yBW}TeILjZbHFHxo8TbP&-SI3X?<m;m8ZPx^+DY6r?9j_
zAfC}%&Lv1IugH{e#R`IkMRYsu0nzP;65%nM0=gr5D-i$e*R#F=kxL1y4@aP+pdi>@
zf@xf>p@JPc8UKhP_(*L%<>fp4(J;U;gtgp&QyNbjEeCSOXD)_(9t{`G{}f^vkjTJ|
zKd&k~*x={f&M2_4faK|{O4NPh$>rQM8hVoIN4})640oVv(*6Mw`1;1#<z||Pq?Yc#
zj*OH5{8nt!V0)Q(bo3D6JMuF=bg!$|42nu>kVRv9@($HX4u_);465gAbc#mn<2dL#
z38bz~!e8-jZ%1lvbK$<L6C$>m+do}rsylutGUwCp)i3VmU_+AI0>%ebxf&sUbx-qd
zxsgjz+~#;<V2~3km~Qkc&!gtt1PR%tii(PyoSebb;nC4sL0!^Z=Bh^m3zCU$78os%
zQcq3j-ocDcOGCk)7%#`tQ9Mx4i?deU$@8d4zU_?a2IlLT29Mh3O70NQYiq9~#I)`{
zjXw0PbG;{Zxet#%EPM^az1Q!zgNP;TZuRql;G`rn#QV6oG%=r=zOafX>?1Luk&y$v
zz2?L>-k4QseN@f#2u;$vhlA)`+ylnrx+iJG1x<sVo}Po@qP+Vwd-9D!wuUcXzHn@M
z`}pL1DHZcMs(RjUQ8^h2lyKnC6A%z221*Js?w|TvSy>_QgJr^e%t(b5)<&+2*-7Ad
z6QW5n5Uf6Z&vWV7VW_Pg5HfXk=G&)t^YHL6Ffh>d(JgH3z5RU*|HQ;Z5O6mgz0B2|
z{E3Z`A(e)#00I5D`%Zj90)`|wB#yqy(<y=%8GDB08zf{2=Npv+`CM>;{O->R4B4!m
z0DM?$-gcYPS5d*`elP|u$EBzSD4Z8o9^c)wTL9lgvSl^b#IPsq>AV&p5k$Y|=MUzX
zHvg9kU}jmv%xnPxVQC;EBRzfg?|Jee){D2f1*}WJw7W}4NJv33eMX(Dsi&?!s6<n8
zuV#X7<)!u~2YdU*XTL!j_Xyl(b@o*m2sF}_wajA}hJXARlv1mwkfl-uRt3;unx8Mc
z9s9&wqQnqG?DN*y*%>+<z}FxfX~RHA7e7DH5=?0V8g+NKQm~A)Y++_*<{<*fdgWs4
z<t0*y#atNr=g%LSj-QPwwQcrv1i`>rL7}Mla`F?WY8+C6f`Swnx;s0Wxwyt%M5-g?
z&?M=iKjh_Q-_7-UWMq`CV_v}b2>g4!Z{e_F0>9ko!4f=m$f1cyNNQ&qUH2R^)tLc(
zN)%%Q8`rIwe*i(0m2t-~oS&U-Y;HQu)Gva<)v}V1fMC!86v$xaT2@)v)7hDs?qF^0
z>2Pv+I^Y+@yS23i*AxSeE?5tg6c+<DwKPBPe7a}hY5>kA&dz&K+&z+(R)`4;4b@`4
z3oTVeMa6GHx3|YiQY;fcRr?&QXqb<D|9)kvkO3q-6B83v)qWUCLljE4L_aPj#(C?f
z$lVA)Twp2#nV!d@`@LSJWjLr(z)rSv-F~tn5L=5gp=WV#Y*X`ARccz=kPzI%SKO~J
zS%8p9S-ogfI*Xf*PMU$B#JHK7O8m3+P&Vl0M1+M?#a~Pf4ede>j)jGlo10rNuj8WN
z13tfI<(|J=1KEk~V27o|yPnpz{^yS?L<lg(c^eqmwYUfNzZbh5_ZSrWflfMJiegYq
zf9T~5nQE1cxYr(o#dv=|Cooq5Wo2dtcSBG??gRNVXG>;z75)#b)xSlbzDxnz7SiE_
z`1l&z(ZaH_GGKCQd1TYm8}KMqQ~f(w#X#tm8fLvLEGmlaGG|aUYH;5KTKnkeDEy+w
zvuC6Nwj)=Ma;8BZA6UXqU0n~3HmA`f`Fa`!D}0VNq}+#0eXpKCfi{@=f?r1)Fg52F
z7J_idoR@mZC@4}+PCVBNfCYA!hzQEPHDEjE$VW2WFq{7#oLr&65Hq$3fFL?T3LlOV
zfL%qf@!{BDTN@)Q>l$>KBZGr)-n_{Jvt2;oig*rk(cs{&|6vd(rl$J->AZ7LJCH8+
zIFKbyX-H(SL=qRPFHO_~{Ddkk`yRNisLRT>K@^1P1}lBi-d#pU=E;)@z!zFuTY35T
z9xExqZZY%o=ZP=a1N}QY-4727<Nr32`<X_O0J+!!3`WPFpJ9tkNJ!{tZ};~0&aR*n
z74-q`GZgKzIyz&)g!ic*dP>Hao0(<9?16ht*b9cXC5w7+5+y6;+S=RumJEIVER5h5
z5;{IO=<Vu~Iz5K|ehQyegi<am7gud{b({jjebrRS<AnF`Z(yQ}XUxII&GWSV3KXB4
zy9C23ZT}U+{;}`~Ql8hWd7w8VCnW{fn|}}%85tQ_;<~Tu)(mk-nbsj>O@Ieac(@7_
zQu-xVOJ7B06PSn)&|dysz3=1Wqfe*hb{qi|JeVc0Iam#(r-``aDKk-sy4!>0?Xzb)
zOMNM@IRO@|5@T*@nHLn)0jvbr8_dnkq2Q&Xr>CN%1Q<jqut1yDc-?=HoD5+77>3P_
z4e-5IXNJwIbtp%^va-^^?+Ce)n5__Vwk%?)eY`!C6BHdi1nC8g3<80wca`FDf~pP9
zil9x_5(7;IR)iN2ubG+o`1o9Pwm=}1k3Q4h&IB)bhnYDa;tMPzkVe7OGcA7t6lDy5
zSXvkeRL9xMP!5EUOo9^F$~(t^Z$A_m9esz@1CLS|bYNsIE-r{?@Q<@IjQjeJ+@30%
z;lVUls{PIB+MzenSvw$9NZ>?4B}Ruh*qjz(X0{&AQ&G??+2Yk6<1+s8v%9<d+cyk^
zwa5~zMw_=a&WlPUBqW}b6NR5XeJU=Nxfe|<FE5Wc0p0haa~<%x!HXz1HdYsGj>dT|
z0RT&Ohn@w@wF&zg1%~-&!mv@nVt)zJ02qzlzCPeQyA%$w;NJ)f3xkT_uLNj?GFyF1
zp6V26tE#HLAG)Gl)Ym^?W`-U4w#X93b-aL}AfV~)M@n2|IxaZZS<5_>knjVo11&8r
z!oTrzMD#hpb^HSzRIs<gvQTd)g9!q;kmo~cH);pS8<do~VFsU_ox!mkK7C`XsE&u=
zm+N^AwqyM{Unttv$3I<1+`b+8?%g{OooF2oy7j*AjE;`hp7921?TWdpb|u*ZS7wX9
zu(|*K{Tmowk49CreN2`T8$7mEul)n|U9n*u0SflgMj(j-!I|;K>w#rJn$TN-uF~Dp
z0|^WS+JZN+WWtv38HZ9rLP9DkD)tT!ac|r}_=6wN@}EBdhQSBXGK%_vnMT}fJIX2{
z(G1T{6K!oi5m0h}bSN9)JC9s$kAJ#1S_;qKWvN#xx+Fhe_~qH>tSnk`UJQTOnH~rW
z4-)9HO*;!y9&`h_=OzMrDtLH!BcDu7O#wZW9|ujplp7N!sv|7o=ITWG0(hO|r&jnJ
zZ$T294(sI5%DAn0lIIe%74OI3k-m+Wd{hI(c&lHD8q{L}&OSGwb5xQhseUxb-j9xs
z4x35a7HGhRDmx#)l5=>I4MMv<dbYdR1s@d0dF(jfOUHSv<t!b7@5RW^Ey!}|@NcLW
z_jGr=MsC2qBk)bCG|dd)Fa6hAmuM<+w{dVbrmEQxIMfdl<2w_$j2)@jLvGZUfwT0^
z&S02l=%8~igQBJY$9G`qBU)Np&0u%rO|l2Y2%=?q-(aJW=mzG>>pn|6a&HIo<E<Hr
zyLV+yaUj>Y%k@ffu-VT~NJvP94d5sUI>h9hQ~-l4se!yHyLF_Er<NBHur`?dL{Cp+
z95VGGWy|IMw4AG-(5v~kX4zK3%0#r7l}E|1!Z;9JWJ<lMx<DYfU3&95BQmjMG^}Cf
zjATe%l>j{nT;o4py?Q!kIBsWVcJ;0AV<LnfJaC*SLqnLEnX$98pPirIrhZ6DN}6x;
zd?;HMhtkE<)AQDL_2f=TP7Xc65yTRGkGBB<@a`bNDRQz#b{Jt|YWj+kf9@K>_h`~e
zJF#BaGEmyo&aP*9zW{+|0)lMlH-gK8)A^)sR>RY!DuU%Bh~zij&H?Bm-j;fU4|5U6
z(p)*2-v)j;`H`7^PAA|7c3OBb46lZ`-!d_4p|!wFF{rwQ$Y&{#h18II{n}qb%0W%c
zfHCIGB@?Sq1LmLgZnxrSbaZsyUk+Xyxh)l)l2UZ~ye))?L)>k5URFCO_xt!B=cl&u
z@q~e8p^RE-hx4q<cg~^KAvg$Nnxp$eu>wsScUWT9*)2tu!|G3iAZgCe%B0WB%Y%mB
z$jLo{oa)k2Jb^TUA;YcW<*7Pc3PJUz6@7kV%N!A}e;|`gR@|x{rRRWP2RZ9g&&l7B
zUyAed5r}I2aNdX|R)jX=3;N0qf#J%fVxa<UeH2PXpa!3W%jm-z{#_iwu#pWsAhnoO
zPBJLcp&2)NmU|!C9$10%9lS``pJXy#fZ<S-aFyf>x4t3Pzxj1s*6fdBtE%|>yu&y%
zD*}5KYlLlQ8%J_2$yY@%qQCwATvylD(P6jL2VP$VqqZu$mrNI4vCXdFYyoK&M5_AW
zSxA-Oe1NetEZk-FyTrsV0sR`(yd`;$Ni99BKAI(2Z7I7ZYxnqAo<!S;fG>aYSIMv$
zwW^AWtX><G(nSY0R9Wf}8bkH@>U|C%wArZS*Hl-J&M+e;rl)IH+*Ta5WQAnINqNj5
z?jSN3^&lQNySTU+5)~rP{WxI97j<2U(g`iY{k4DkNhua)!-&GO{eX{}1^l=4syu?W
z>}TqUy4Yg7lJDzkCk>}kPR%PTT3(;EM)C{Pxvs|OsM$L^SEP$>(^au@vc?sjI%E7v
z&&p-hfE^s>%FK!5bd8FO;dV9uak6HCb|qTjE!en3J+@?WL>zh{eW>~(MK@eX$G4EN
zoCv|<0R;3jAKR!$en!X&>)Zt$N{ly@s`vr9xfbik>Qo5W&m~E$Z{k0NWVRqvpO%B8
zB+j^6Kj2f_%uLgV59W|#BFbL;bJtfz-D^PjWBqoOs;rWAiY;Useo+U8Y2XFOB83mz
zcFtku(mm^PpJq2<%6`z(`?E?qU}X&WU7Xb|iRS*%WWH;iD93!aKanTr@R|OuJBOKx
ziNE^?$SZUp^%?u*x4u_0c4_Ofd>|)?B~?)FHmq+)nY>I=&yqB-Y!}|c?XR;<ZU)ia
zIy+dLn>lt!xqZo`<0m@gI@IF6p8MOzqkXLi)@wdqUQjJ+Eb<Dl+!<<Ilg|~Y4;R54
zO~K`;CI#8I3`z?ZuY~E&o9`sV#Fkc8$A!izx*o|^PFQ#`ZDuMUaY0-OJ3lXc?dS-4
zg(`6}8ogI4l3HWgM@~;q&(EbugP*Bkst<+A6McQ&qSs|lib2@t3t%Ts`sQd62{QY6
zogtP}h*U^LO-*T0QO<^KKdq!K1u1EjX4UGvvp%Mt{TJM}IaHY7Wf2swm=BtE^{<6;
zl0uzFMh0t8IUz+pf|87qGVJXr3nseMvuE6>2FP1-nZ6EQBZ_=*^uy-s!tRQDw)IyU
z#^%EEAS#!hsze)s`@devm745`w}+7tl9Am+0O!`I(UZsd)LzhlO)>E8Te^?ZGm<Lf
zw&HJ;N#HB0p*|Tk>9al6;+uht)7F3sEEkV9CW(oOp<oJ&OglO}JO{JE-$VkFOGKkf
zwLHvQzb)p++6>f%tXBszIVOCcS%a=6xk3vEbk@#i5Nza!B{L`4)v2zeiu(G=Tc=5F
zW*VOWql#WbvqUI$A!zPmzj^&Se|zY$ACUqxm<8@GbntO^FYFGU(IX%RePrzeb5JPa
z2ybBOYbGWqYbJsIdK9)?9LoJ7>gOIFR^?AVtzlKdlHBh+0x_H2k=yC%sh5`*n*Y+?
zLk|8}PAU9P$=|22A+)~S`_t8*CW=PG&aS+!i-CdBm*i+rW3Q;JEHin%@8~)L>Jdwk
zgpLIf5fPg)-XS`4W8=B8>&6hwpr^Tt*G*IJzPa}2kBk6qadGjWpg98n&K)#_z2HA2
z?ex@?zMfuIJc1*iB^I99NSJF&#3>R+5oolK9gKXJVX+6^vAtjlYb(eVJ2PbiSEj1h
zA!WV-#bF17+Ah=V8b|(MWB9;wPSpiSaAA_*P(5(noT^5c*xGg%Gi=T_iH=U&SX#bK
zCb1W!V3fag>lXj+9#l}AoSa|3euaeT&Ye52UcGX3bhH*MwoEbbP+u=IZhmRVd1a;q
zVIPt@kL}sM{{B22J@xbgFN}n(%sZ|9{rwXY6W_mM#^FI!NN%>}EC)<(5%zpIH9{bv
z0da1sfcOR0)mMZD4I(_K5p8*)ZwJRgNx?iNe|bFPp`oFH;C}Y2<)d0g_8MQM)Qi_N
zu-7P@o2sb|ETLW}kM2XpBPTBp5H_TLgM)*B7{<KP_zq9t{QTVOVS077z?!Y|vuF4>
zFtwkC_F2*on8U{FYXG%6r|}m`YHDh7^1p?_pFdeZ%MViu-1pihCK5G2!}o!TG6Td~
zLEnZBu+d*37@hCm{|LoB1sNHSXB`~)0u>@;5`scPA5*K~>6Dl6w!Fds)L@XwIytux
zD!?m~L$fyQkz%W*Ecr0f-@krIp2FD=-mt7;LHqk7h(i~bmgMR-IZ^eQdoRFvVkHHn
z+r&gfJZuF}4EFW*`uh4_dDbD|<HLek`3Orl6DN=<hlU4AKrCGH09IQ}DQ;?MY+T$X
zkljJY{3&F1PoEN!lD?b5g7JgR^HsvS(pBPIg?aa@Svk0>@`AAi@pU@m(2NXY6O&)?
zorXI8XM-d4>I0bs<T7WlD8Xw3ukm0Q+8H3?UkL<L)BMIpN{2id5-qJ^V9DFhHcrmX
z&2@I(NV3N0I2$nF1Uk2-me#`jJWM{w#c#ZlEQ2ZpcCLqji+~408qd?W1qB76p`l8-
zR~EBg4~<MsX%CCHDo0sBYzX_jnXN4x??y^R1?{E>=AHNm$?0=XAD_x6>~QXot*tl2
z!#{t_ITL(Ve=wFvy4D)N?KU<jNl8CL#tJJK;D}PgI%i<(!~eecU0wiyl24>Q()K-=
zP(GZz8v3Ph0D&D=ZjUod2#9MprfWY{RKzfJ0X`?5q1KcM;6n+t!r?4Gi$kdRegn(^
z)%C&QAru513ou-~p~VUebcz))|6%4&S6Zn-EeiN>$D%W2C8^2D<U~Y7R8%VXnApuQ
zC7{AX?974N7FheioUyU7!P&S1Mnpv^9Ljbei-=)JN=*&348g+0+?Z{m=)2bN>^Hh%
z3`BBaVNwoy0M4PQ?*;MA(z>2KA^TlK1m5l2$9sGA9@|C$5P=!v&!0bGCR71#2v{N@
zJ6}NDoRyUY9YPi^E{Kd?P!2=zN{os5@sL@ch0Fox814%u1J<CLh8+5V&yl^GTXjy(
z0+{g-5E8PnvI4#dh{JMU$~d?It*zO{FhETzkU$<&&ynzhfrX_2oM-9j7oZrmmcXqu
z`~tN!Y&L+&EC8AVMG*iYZZ0k<;AjBWGr&7o+1yl#`2+3-N=i!k&s_C5Wrs#apamB9
z{{5A1790qy0T=kcSQ3mAQeFMm!a{6x^m8Z^6B9}4>2t0i42hSi3|&`CkenQtV||&{
z7SmTwSnxHZvz|fF20Lx3>6(qXW<Nmkr)!;R;Z;_q0I9Q@8VHYkp}v55@U69#e>{)g
z)sQpy@Er^hRQhrelp<!|-$8{7#I{4I(P7fVqlMi829rV2aqbJ%rpx^NJTG81_wL>6
zlZ3C3s|Y{~>`#yf*x0N>1qQN_&+eO3Kr=Him<HT4rzGe~NwTo;6gEyLr)EIgmZBML
z^2`ggVawNlO#`($D=Vv9IQb~pv{lKgKy|05mjnpU)=UH1FDS>ZUAqQz8d}t&-IX64
zdvft*Wo1{g%gcH0-(Los5<woGuAZLZa|ydWC!>xqcU#$UKw6;D%g_aTQ#RMAt-#^m
zl#hH^5oENnU2wM%TQ__A;wS8|Gjp=B!SMrKU0o}E5L4AkbROsaSTag_g@PkLLjFPe
z{`bF^4t5??l{1ZAd~i$nSjlt4&b{DpRj*f|4a3oRdJ8V=z4s?L?;Z@T(&NX&Kaw2H
z*VorIZrE1WXJ^y#@M!Fh!%%3AYZe?pAYOSOAb@;K*v_TP8W$py4aSRcAqw{P_Mw6F
zagSL5wphxtu(Z^?r}0BJHrDvoLMLmSQs0b8mFLlh9$N$XkXvuE0t3N&WUgjGo*5Y#
z85s2DW6{JZL6l9Ac-+II#V;TLSQC`zP(J|_3jNVOV?dL=y&oKwL&4q_N|G-p{Z-OD
zNjV;%RM-Kvv>v_C)7K}zcP}F&Ba+~Wjt=qPFKruLEj+Odu%J}x48btnNJmF*@h3E~
zGEofGodW|T!7^ZMGkz9R`gH6=ef?kfW`OcCC}v}Z2?_%AkzZ5<ATMj&8geq=RcHP?
zI0dr0x_X78TPvHf0)Q}Rd~$I)@nFtSRT2{fGw|?4f4wEWLaq1OzOqk3K%F%Xo{pYz
z2q>bS3HL6UJZ2!Uq^Crv!_l`kHeJP)q2(y(xW;zntD^KXRm_t55}^eF2xYGk{sNrW
zp@%~Cc?RBo+R#}CEwZRCb4U>M#&ur2XacP1%~$O~`U>l!KjZzEPIqD$ay3~h`+c>y
z_)<=WM@Gy|Oj_)BtsNXn0Mx8BY<Xf1KPL`N!rjgMOSZ6wyUE-B3&tL@sl+%ZSmsMh
zOK{Vt(cq3CLubm}L4nX_jngbJ*ZGo#T=1*K_$wd5^OTzu2W#Kyny4iX)|;;;Q+Fo*
z=dIxgPnQTimbf3{19#zlZqi<3Ppt^Oqk~@E>-WqC!mi#5Gitoqp{&D1Is5#V4$tPB
z)T?W~`Fa*7RBfnqK0h*LY&hmS^2kT;%A{}1>-CPaE@A!s23mFN-n^wEZPxbt?r<<|
zdrCELO{mTSj}LK^p5^*IZB`;F{4nFe%ZN_f>SUUCeo>taimY+nH6(f&_C=AMO1Y4x
zS2w}r3z?*QAl4(-&Z1PJ!{>vKc(24*?MI2a_$c!dX%1&oR@cZ|e5p<>JR6$nRVv%r
zpfZ{3R#N6nnErX3^Qb#94o|b-eV-+t!8uE87parXM|I{z2acjhRG@U0Hfz%4=tt)F
zO1UK&{DatG^xy(IO`w;n$x~UR&isU(m@gVEwCW5UDME{Nipb>Q2wjl-_Hs@g1(kPR
zji?5QuW@j3lbUGS>58-q!7QPj4ltYcMK~O;H$ARr5m3n9R6FP*RhRIi4({kN#qto6
zEuvfE<wFOu&8$i9`bZ<b$yVac^tx(XG^Vq$Il0g~gH~_uz<t2XNvy|HLj>^-v?B>U
zpQrjKO7%-DH=O)evFLe|iZ^BDL>IV;mF_h#DDu|yc)qz)<%Cs9*)LpS{gkC$nnyJ}
zpR<30a{*TldOa!I&Y$SRWOZ2M{*I)Qew54A?D8bEKX4~?%KX0Q{NES2`rxB7F9b=Y
zM!z>V*qd)9>Mg0FW+$XV{~K@<#KH4kzd#E<lOxR8Ilm>WtC_@oX*-+9CQ#*G?x$sH
z`rj;T#-2UBb@ih}cTD%*|NgX!zU*l`OeLH=s$KUyJDyoi1cKCw#6R8Qh=v=5U;Gi(
z&Ng`%&}O!}|C)6Ayd${x38Uv@Izr5>$WA9l<cgkQBT@8W@F#J1GI^bW_h*qlI`^`s
zXk?oT(whHgHa4O?(-r*h18v^eNaVS?tBt%#$A1%8Srq;MnfDuzel%@f?&yxleoW}o
zyJscNTopUCzVs#fP_aRX()WaJapP1oL3~zP5Rw1b%!y?B^OK=eCr*{H^!={i`g5je
zW1mWj^V2UYc2njumQiA49k@8|>f$G+iyq~RjJq9b56jmVoE~*l@c#St`gBhGIrUi2
z@9OD|*v0YY(>-m+uT8PKxhU_3^5NUXlSIZE&f62KX5(MPrOCA%x~7WOtvD`my>B1!
z_-vlF=JkGCcW)!MKncfL5Y0Q;Ng@#Wh0Yc0eGba2ecKkC74)KB9WfY@tk}H6ytaL3
zhV0f)M(Jhs`17cz8YNa`x3)XB30{e%8UJgvR1cFqT6p;Pu=k!KWI(q|DIc6nc;(|c
zpG$auFhBf;VSE;4sNuOVvG}(1#f*xpP+?@}5@W)bS?>!Qjrjr>M$Q-3jON~S7>Fo=
z+l~@Tq8`8QaNM__$&7VCLrCr(=%jWI>9yGI=9CWJArq4%dRs_BGVb^MVKeLT{BOH%
z#}_}tE=%9zWU^G$|Jhp#|F~b|{8G$0a=7UuVH(bdPo0v2#Fj+<)lu?xc-7Q&MWOx0
zGp@Or2a4W*U8>)Fq%w{S-U`b&OHof?%1g$*H<6f1M~9|hBddXYF}mOQTB(2>p<T_T
z^!-O62#yI5oynhz=3m`fZfxLg<yqCQAxJi2NS|qXF*jp&ZKhyJC6Yktk=$oX#-$>J
zB+6LZ(7-q*!`Sc>0#Q-EAzQ7jZK$t?j>xb0G~*F*X1(+?5M%9AJF<t@QDXbzt9-P^
zEi13ZpD&6(Jk?1`h#K~ii6jXoQ$a`yzQ2ncPmt(iD~^js&=h%V&_=%}8fQjBj0Dzh
zlZG4fuQx?@>^jb{MlDO^<9S`aS>+=RBmR&VKU}Ufmdns{{kD**3M&%Bzp*~Qxb#SN
zcDMqI2oaS%PvWLh-DZKyfRJ2xQ#9~;HiP8wt6%*hN-;XS8MU)s6aJC!13z-?kvZ~3
z?)Mkp8z3dpxPxJ6qX(0ri*4Jm7zcHb(OLIcoYKB1j|e3u<5+=1NkImQl}KySwuswn
zl9H)NYC5((f_<@%12-ejOt_Oyq9Qb2$T)VNE~Fn;$@1k8_aFJXiWx|CuyaI9Auo1X
zUu|AC9cW8S_uMp)UokhJY4h?>-8eaVrW9Bo80nfx&wEzQtxeL_mZW0%K-8@NK7D6{
z;;3uy8ZED0TDqyo>afQ*s{tI=$NlHi-R<$(jO($-PG=YHH&v1j@ySL;k&Q;f+3=$4
zz<R=a+&y?fSy|7&`$}VH1r22zqxGF3KeW`h)_M4gU9hglr)l>bX%3|a73b4$mWPXo
z5Qw!9-rTgDn`rje<%g58S-J#QCj`C^L|pLp<sAf&7@c=qGswL!jzr_nrAF2Moz{B(
z{NooR5B?&>^MiT69FI%iB1vg<1Pcp?&~sN44AV91Pld&1r_KK+N#%*7<|~}z>l~5Y
z5*mw)=58BeSsowHzAKaYvmXCkLK@tjQ#|AF-b^=-E<9U$wJGX%cCOfjjgUkd$q5US
zqd(iEnm{7i=k$8o^^xX3PZvvXY`SqF5V)@tWu&w_ho?i{A}hohH3)Y4%Mc2Rr8KXK
zgJ+tTLQy1y+h<kJhd*6owO9&XS;4qZ*prMvj0of&>O=^#n@m2Ke8*}xir8y(`S{o#
zWnEMp+1XvV7m~U^XK&Q(xZZ_xhpOoB2WCatBn98gmhqHvClTFlzMNRS10pg+GVjZ#
zx#!wWXeBtMk(~i=B2_M`*A>f2V(+GJxfZX*Ipk{ocsGD<l9ZU=t;ERR`-6bYzpZw6
zjcO$z*_gwwl(4FT!~a!FSM$f(#h#IzLk$n>V4R(V)+h_1jp0eLFvQ(B3p2u3cW^8J
zzEgg3+u-j@kJaJtYzY;CDAdXU<Jx`RsOsk(ThWd>t`=kR<`v}?XQxf~YVGxFpC6Z)
z=~qWtR`GQdb3Rt)!NIO@y|u@w;yoyd9%XyXb8A_q7scEE>nP1kpy1D`inr6}1Qs;J
z2jx88UkJoko?mck8h6>Vjh|ids$96)EcE3qd=rm;-J?-Z@y$a={3uo%fo9_^iIw;l
zhgmoZbu<2xcW9~f5D4C+*WVTqh-bpANnvdy*x?Ly&9Y8k#-6WPKe$keONcTh*|1uo
zT1kzKV!V#vUJ<kyYPhp6ezWKcfw5X)?rX68!+R!xiBRt{_cya^p-oBFMo6+!*cQsL
zMkyw{)+&W)#(cjJqkMqruElU)Hwy8g)p4iM>(2H|a+?g#^Kxv&gC;&JX(2XrghHKi
z8xOEWgej4+VfNkPhoQa`cO|A+*BN<DC!g+}sfinuduv2m$#V^s6&i=c^sFg-+R0@T
z&#-yH$D~@^peH7ag{b&Di+tW`a@|`*n=4vT;H$;sp3o3Gq+t!rw<Z}Kcf2YUQ8!x7
zt@-jIaK`7I>H9%Q=Fh!3QCB&Waaa`3KaT4?Nqi$cE;J}~wC6&*B(naIn)x6rPTi8u
zl6w5faDv2p+~nDwCmxmF!C}%-qsXyNNk)ec(Rl}idZ_37)*&5LSyZcznYkIA_O+&$
zdM8~KaT#iS=jY2$$lGLHrpBeh&Al|8y_$9}I9AuNkVc1(6r1rdjhTO;`DYEHQwv2g
z1xQ-!dpgXW_A(Vk6?|P{k50s-p^yn@k|u0_ah+^8No5gty0+{tod(iO>crFJCOalO
zz&0x66y^*)hSr>lVok@YJVRc88kX9_56uX0M(45OlUT9}KcBxi5NjxSJnot!P*KvY
zra%-&;@7-bTBu``7{zgXt+iLW1&v0T{aQEr%`V=WGLuE4pNBr;tHRtr`jk-&ns12S
zwS7nm&Hi;kXq4%6dDGExEe=#tpCcnIWm#^jQ4sMM)Jn$|%cL6$WQ&sC2(ZLNOiNo2
z4rXG-xBO`2-Rdm(qyOddWM~P?EE?UPo+P+#>FoafV-ok|mE(t63_22=1a2N9)uEY+
zT+v@gSX)}da7|{r!edJt+W+Q458TB^SJzIkY%|qCM)nYq=RjLQcFVswOwHWmGC=2g
zSUMd=!mRitzxCm3@rTITr)*rU3<H99CPx~_Uk_m3yxkMlisgCXSX$ItC7aG|=Wodo
zWHF}fGLxC)U=@CxJEv)lmP{8EbX@g=|FNKD#?QYl$hk?o977x27i)~XdT~jb+js8@
zvt}3dMO5FEr8ji5|GWb4zgX{BZ%iL9Oc%SWE7tSr;HYejr@}kL-eh4WZh#(<IwSfn
zM7ExUq|e4UqV6G25}*5I_mChaGxd4diBhsTyWT%;@uIIC)ihh>N&kdvFcA~|Z<aYJ
zaS)l@&p9J{Y&UFUQyu>Z%|(^Z1bn=3`Eo(yf6II{*UgxbD10-s_pvhbJ$}-80t&;A
zWQ5lc0eUSTZet)2owgEN>3>2@&mYG=M*Vxxy}kX#Qq1R2+k;aqr|>L5o{E0#C9$oN
zKZE;vNI*vjbz%0u-GPn?gAeUQ(}DWs&S?JYBDQF(v=kHp%?u>nR3lw=V=oDDSdT8x
zq8e;fNo*X&dpOlS8g3KVeItHBH581yw#<`3rBr$oUC-Uu+2Pr7{by~RkF%8TFia+o
zBF^%K5dQTo2F`-Hr5*_f*DE+<1M_m58Xp_|k>GH$rpbsC#9UoVv`G5<Zi6O#?qV_Y
z`Fge?+m?jj*r=)h7nbhMS55{|QH)3Ky~?Z}H!E#G@<?N+sj2t!-tek3<_lT(isH&*
z?^LeUO_rCyX!_WocP_Iu*0Ci@OC?&{JpW_-E!w_~Oj2H_Um(8w=nq37YpZD?OjPt`
zNd3uPaO)uBW2`|6QZzAHCW}a}?{|gS?>Qc^f0SM+bh&Iwr*qk;zaDD4zt3t%u^wVH
z;@=;krxtyy-|@lI%hey;Ti@Q>j;o`>YZU$^$EBquZj+7;v?U@;ZmE2@jqO?5{c+L$
zG>1*-)<(Nb(}KOafx7A1_~_AqEy?uMw=eU@3ncf*0{Tcd9b3^6>SM(5yA6LQR|TVQ
zNB=b8?Hy0Xl_b@BhCSS-`a1hQZ)`<F!WNrPe<A@*R9us~@ZW}u$rn;PC@~x@)wFu8
zS+5@lMmqENSrmgkoMwQf(jFe*-639&cMmD1gztm9X*$4&z$p&1rG9TPIWAVPJbE7$
zQ38sNCa3H9?05D4d0}7T477ahp*v0>)ZH(Um&bGGUR^OgKQrq+oz@$4Y5(Ae9#JZ?
z^|HQV+(ZrSmOnRPKJUGvpEwyrg>`C9XX#z{9=e1jC_WmxrDNsSA<+<cL&{2Ma&uuh
z=EKZKwLL=p2_J80Z`4tY2BX?>_W!I|Ba!cuTW^}MSA-W-2FYIJ@-@%AV4@@zXpQ(}
zhd0hA`&`u1jlIG3EP+8|d}p<(*e$#yo$l5|CTDM_z|rkh8K>^ou<}K{r-@JzB7UMl
z!$P)=u}P)X?qafnYbMj4TuUP*%fyeY!dA?SedfeeqVLRSqi1XR^JQ4WVPb>*H4t%%
ze2uKgew{maYOUvR4nm&5kNs#g^q`?@*TV_A9t-Uy55GCiq`8(LRvCGFdPh}={O13R
z{^9*s9{niQ)uCveg?C_Tv<F_pavFWN$SAV^{qZ=LvAj8hGT&Y9M?H=^2~@jPlg&5K
zOlIC7m*fMd^LBLC2(s0kj_L}L0~5DE;dJ}0vC^o6Jrd>p@>}V@5W$+iH8L+BvtWn*
z#A;-Wo>Q%9qpQ8<q-Y*vZ_Ke|iO*j1O7Z%AiU&P&vzY7{rGsC;rk=7q?4_B&TB4%j
zaC1s*Vkwf-iE!@?S|Quhk$GuwbtPi=xrk0DmT9x->6SGE0<sLwz$@%S*(W=VBcb)G
zbY1eJ_<t3aXM9m<H@6SIx{2hcQyYIdj5nGlX~e;P^R-HT>hup6z%BQK1WTI}H7tG<
z1ZCJDOIn&A%?`~e(QJ?BEE9J3J{0XD#M_o2GHCTUuHd8dxA$~jkm;>U(Ie<lb&P+D
znQzDU^55z5zK1hI{8npYMzhE>o}BIvTF-6^IX*w*y03caE}AnnF*-B$F2B)dh3KpE
zy$sPEb*9I{Zu8!9B?WfNH5q_3okjW^^pf^UVEEmmFp_Vkc5Js0dMSLAnKj?PbsAmd
zNGSO8pIy*foU@b0k!~8BnU%n0S(>_9=gtvSq@orboA=Mmf$fmE+d{Z0_05ConGwcf
z_D5c&a~V=eBWZqqzg`;`g$JglQV}xwUdrT_`dXBA7i$%@-sfzf{ax^*CLGV;WkP?=
z7%7gN$JBVgF+4})iI2g4$oXqS6~4=}*P8@&UOm1N*?zyvS5-2cQ1~)9?BcsGUwSPj
zIlZh&dT#BRNsIYFr>(j6a%?`DtPSs+NjYWIp+7%M&)MqJKR+x+o5U~M=FRFJ^j@aJ
zPUk<;3tE07zv|@AD0qE5zJ0cnbUh_??78!=+Gh535?bG)7{Dx4ui%J_Ec_*O7h*&g
z&oBMtCz)(`{`Gl{F?SbdzA#y$A*^<SzNIR+J6<^TPpL_ZHvjq?_N2@vcDx7wv%0tC
zZvG)}dQxDc`sS2=8A6iO=dW-jnm_)DYc(e)Ge%1U*&DC@EA-*4!DQOaoo@sGXg9fw
z7*o%U!@Ie+;=;7p{(A(49s370zhttSjfs{z1VS>tgMFGgJ;Bo<EDHZ##vhLb0ovG@
z#tzvvjkG5pgoQr-uoMpO`^G~Ww)NkYADL8ARF+;wuO8R$Gb+F3N<MWv36O0Y79~lW
zNa8;@2x@FC>PblLzVOrzs<<`T>aT{G+7PZce%ZNK<v~Mz?mll$AAPj0)`k1Cv>-0l
z;8v1n<MNx^!HTntA3e-fe>Oz+v;2Z#v)U=P$63li>lo@m>8o3AQLLERE=|phN7!vp
zu57o^-+28?^qRYm7tvr}uGXc^H>$P$TjW&tU+h|ZsJ%1rbj8z;xtGhPf~tO_==cqL
zs)o9h)@;hWPQs~-{lETJx9w|!ah)^Xo;Oqs@WyI??@4Xlmr$i~VPl`w`*N6rKXso7
zP^B~okY?PZsHn8GEf+VPqjXyS0iiY%ZvL-5xd2ddmQngdZVjO&kdpS!;W_><wApfK
j{QpHeuSVwo7n$;_m%TSYZshYC0WU>aRhd$0lh^+Tyd2Ys

literal 24995
zcmZU5by!qg_x8{YhX$z`x)DV}8io=OknZjVC8YaE3Md^CB1lMwgf*f_36ewUP)ZL<
zDD@qDpXYsl*Y*9uTyyq0XYaMwy6=0fv(FwL=-<|$AiY8ggTW}YG?8~;FkBD@!)YYO
zgId^3oF-u~Jea<Yks5S}gM)*Mi;IVchmVg>KtMo9NJvCPL`+OfLPA1HN=imXMovyn
zK|w)DNl8UT1pojwH8l+l4J|D#9UUD#Jv{>h10y3N6B83NGxL=zS6EnBSXo)w*x1<F
z**Q2k;BYtsf#Bri<l^Gu=H}+%;o;@w<>TYy=jRs?5D*j;6cQ2=78Vu}5fK#?6%!M?
zdiAQfxcIee*REf`E+HXt<Hik1Nl7UwDQRhG85x<IH*d<y%F4;f$;-<tC@3f@Dk>={
zDJv_hsHmu_sv?m{H8nMLb#)C54NXl=EiJ8Ew{B@`YwPIf=<4d;zI|IyPfuT8-@w4&
z&Ye4khK5E)M#jd*CMG7Prlw|QX6EMR78VwkmX=mlR@T<mHa0f4wzhV5cJ}u64h{~E
zj*d=FPR`EGE-o&vuC8uwZtm{x9v&W^o}OM_Uf$l`K0ZFazP^5be*XUc0RaJlfq{4L
z-o1D4UQkd_aBy%)NJwaC=>7Zm!@|PC!^0yYA|fLrqoSgsqoZSDVq#-s<Kp7t<Kq(&
z5)u;=lai7^5Pb09L2`0(N=iy<YHC_qT6%hVMn=ZNhYvF|GqbX?va_=vJ$m%`@#CDF
zoZQ^pyu7^p{QQD~g2KYWqN1YW;^LB$lG4)BCr_S~m6es3mseC&R903#efqSjs_NOZ
zXVul!H8nNQpFgjyt*xu8tFNzbXlQux;zeU)V^dR8b8~Y`OG|5OYg=1edwY9FM@MI8
zXIEF(%a<>^ySsaOdU|_%`}+D`y?XWf_3Jlp-n@PL_T9U8C=}}b`}h6*{U1Jj`1tYT
zz`($#PoD+{2Zx4+hKGkoMn*<QM?Zi5JT^8qK0f~C%a@6XiOI>ysi~>y>FJr7nc3M{
zG#WiOH#a{&zp${dxVX5qwDk4s*KgmxVKA8G<>i%?mDSbNwY9bN_4V)HzyJ90V`F1u
zb8~ZRYioOZduL~7cXxMhZx4&b?(gsa{Q2|OuU`iT2Zx7;M@L7;$H%{a|Nism&&kQj
z>FMd&+1dH|`NhQrrqMbA;sA-Crg<O?Mn-e_hcmtqc?yHUVOmIKqmZKQ1)?Il8*{L|
z^&RQ^yP6s|amSTk#&W6^B>xg*Q@+1TS{@zC&#8u==|J#4Hkm-^4hlDU#>Qyn{yOE(
zzV5Z`A)TyZ<{amvbg@yr3@=I9{NmrfReaG#)gK9Q%1UtP1^xIz!DhApeS{?Bg_;j_
zKSHcM)mA7hTCJI1l@Q<kL@zF;drc^7z?nzeMa^WJN6V(F(qf?Q<8$3iBO!Br=4SIu
z;cBCSI!%Rwe8GJB**}@X$(3k2-$|QHqfHL_7l3oEkij*^<~9zV-{n~WxiqLA@&QK8
zheOFu0p<o-=9!PQwI4wT=Kb1JaZUlI<vcB%Z}PLa^XVTPm<`Z8Gzyb38Or{qlQVSd
zxzia+(?#u$w|Y@tuv&9umQ@WuNLn82MnzzteE0Ih2-q%z3xbFG06mgm>|OvVCXGN`
zHKX+rIRcZ7>OqPNaS1LMc0CFa7rSA6r{2kgJ6k&=Ybd$Ws<{&VNJ3m}#Ih~-(gK0p
zJb3<NZ3mlFClXEGTn4ThllDmNP_3*3TCRM0t+tvbhee)&*js*20@Reb%iwF|pnA1#
z3tFN!0vUE`lsp~s(x~F1*lKvdrA?&eo-$BNYZBD(GATAM4$*Hdq4*sBGRF|WqrhUq
z<qfKjskSQqUMKz;h~iKDoI6MTi9R$n!bth9$TK+M70i(uwZ#!k<=Zz>PD%oCBOt{M
z*qwyG?fBnqdh}g>t2H*n32zJz4G6~Ciej1EKqAuM@Xdi8^sPkS84Y^mbO5Qt6U3rn
z=kTUyx6?W9-i?#$*zXp%mC3K7CUz1mc3EeB*IK~B{=L=?v9miS((_gIGuH1$H2uGW
zH3GJMy)3VhZYMs@1O`+D;u3&t5CIca{bF4;e4`AWh>vNoNQ3pO{Onz&N6OQo9x8>C
zv@%iX2ImljJo<N1G#K0UU<a_`$w2!!X*oNo12KH0NE@CW-CuZ15WLS1BBG_e4=Kt0
znIzEqI@?^A9phq#7{yTt{go!p=V}wG%H%kxnlBxWU>wY4-4r;`2UKzVuf&9qDcS4)
zJ%VAs)rdv2chMoa=um)C(f*z!o<rW@1=qh*c>-wD;av~7hCKOW00c6j24GTzs0@=i
z8$nQp9HS~lj3P@R2Y=*hR&ra@aEclXX#*{yYFhi;t63OS&m%gmwt*7f+_g4zh*n7@
z+RS&dN*!X$9Z_Wy@8@}p5Ml;a?l^R&l1PHuGa5Xb!s5C^DLF&Qb!anjF=G{;#h7GA
zjpBTvs2gzJ_t{`d8Y3sKdKRC{dyUH>9C$+g{j)MYx$rbj_W!Z<F8bi%Or~ez-TJ|!
zqAOKg5+sBA%+7XtuQ8Yto#e3`q3?gp;C>r8V!hDwF4)d|Cpk>(o2vV>{prs{wpJGi
zoepl4CPpbA)hOenzp2N3Oo+{U2T$)%wi;3R4<6}}NBWm;@Fy(?E(4No3?%%AQf>1T
zr2eh3{P7Tq1;K_@2<2l?%-{$Gx46##0+j~CPhdd&mygzi3r_$IK~UvyAXSv<R4$|S
z?Xu(YOD5!WD<dckFS}QT`Bzh5?)-z=|B&PzoYLT*bB+HS)JGqTqeCJtL&;hVCxrk3
z!@w>F6kh>y8#Dgeg;G-4qWWfkNkj;Db;ig`<}u-dRIjhA(O*&&5j{gyB2ooML+~L%
zttT((1BTgJ+`Qam)GT6a?;n^C7-Y$5b;Uyw5;wS%3S2Ai*h(~$?<DW7=ek*sQnHO$
zO1an6!s;Xw2fyQO-UA7+rJ-9DS@(DcZly!P&Gl&0%o=XOdwkayNYfW_p!B5Pjaxra
ziFawlP>ge4fkZ<ZEOb>;7`AgJ_ZR<{g50Pu`$3)8ABK&^H3+Ex+8kj&r^EkLxraH>
zsL+KKiXMC<mK0`x<4LOr^(EM0O;kNsEj=mbCTkL81VQOb$81r|xpr8?f~V^7Ch0E8
z2Q?AMuBFQ31dn`2yP9;yYW|HxnBDrkKVP62@|E?mKlo20``-9GcIh$CkCYA(2M%Mv
zdQbq2iJxH$8ZADymN|RRJ;{mD2jZy2Yq+<H6@NU!!Ir80OQmSw0zc$ns44KqFH=%Z
z5(4pKAPOf${yIJL&pLS4<y~01*XKIzk#tBJQinYRJ3)xH@(N^&0*VVpM2{4uLw!{}
z1-{A?WRhLpOohvO2k82H-@WJzy8oP?$J=X)g>cIh4$8t5ZXj;kp#Fg_9x;|6dh~BM
z50y^u{L{1do?X^I*IU{v{_Rf{2OCX*d6e)!yE|T6K37|T;GlpS-fNHj2>Cu79DOHi
z0?Gt^VePj2!8^Al0LqYJk`lauz#Rq54%hdlmML-qaSi2pvb`M&9C1nc$g=HP=pqVC
zm=r$puKOZq#Qi1%!2laCoEg_p85fIc7nEwHwDQE2rd&ltjj;T+@#c6<9qON1@9XNS
zb@W!pq5SykuDF=;-<+$ib*UmZ46X?=JJ_7^=w)}aia<Fw_EwfrnE6o3V__VvHs+@G
z+-@97XKSseV$Xz~I3E9>M0p=dl;%*PEUK1Po)01QKA9*elN^@u_|il4^&N}Dm($%X
zf?BVZa`>7A$YBmw=!uSyNV??ckUijOqYwPPL>CsP624ysCuGC;e1GwO>2{fK|MHvY
zUw(fi_1_5pH}7bIamFc-QZNO8&(%<WSqikdGeDufiORLFzT`7gEQ9}lzvTUG!3E#Z
z|HJLF%fbF;&Mp=@j3x^TqSt@YqaP6bhF=Kf{x=Qc0~|*Gzdk`+cf1sdziC(`I8KGt
zGeJ5+^k15>qH^tu>eY+@|G&ay({`y)kR*M#*da`iR{lq)9{eMKw_lFIt9Y|Dbu<|`
zC2fXMXj`%>xI?o7NSYjM9Lg{2A^c7>zE+#~;&a_cMupr0jC?hc3R*)chi2ZX+p8u_
zwZ?-}qbtn+M;^l3xU-xBx0HAmSdd&sDS7WKP)R<m3}cXrL2x3;_6FAfl`>d6NPlRF
z7}38p_Kh#~uKNGv(-5k3{uUg6KQ2q5F*ZzkgXZ4}$e?Eb@fh}{a#0N&BJE{e1Qi9<
zm%3*JEL?v4(*dLrUS!z!K7Vh$Y@H6c5QC5W%l~N#jPRu%{a@XO+y#k$dTeK}?4AEc
z;GfQWA_3Rtf3$d$(7!zv{LfgI#Ya$L-u`z?Tl)_IC>Qa=`>+h?dmjrRLF5O;eknnz
z_j;4t#RrI42Fm2a3YamOKd#xdTHGi;QfgYhwfD*95B_TNF_TtForTuo4I>u5TFE!N
z9<)xApBNr_nQ%|4L$ZuM3|L-}&M=m}M|x5Xm3rn6#-MVuzQs*kPhHJKq<Ef30~~#E
zBl2eMIF#3)l2GTu*WRa?yZ`OGQd(S0#{8cW@xaWxyb@hWCoVQ+*(Pw=i*PmmY@O!w
zv`k?f^EUDJ+|l57lf2`(UpQPY`R4#7eTb(}RUqzAl6ot=oO|^6sF@bUW!XjE<~}k|
zxF&ObMk8>h>i4Wb;CiIX@MT=?u|t{kZ!s%Ar!pTz^{EDs%DAN<$l!8G0_B*_uMHLg
z-i*ml-7hn^K78udM*O(oe-BL_Y{z3f!OOEUS1@-E$F?Z34_u@a{ReM0Nl*}o^QjgU
z2Exm!80(Hrgojk1xX0mQMQ9w{uRwz`(Sh$0DK?E8Tg(a)WAiHDKHCoh_>ruPwg2@0
z8SY}jsQm@96J_UJkaUAZFsJITS+ZuotuDJ88Uvm+Ww`i)M^_M+r`DXnp8=#Os#M@V
zcRlecxIS7W_gk0cuQd?F$H`Xp!3qzC?`FRcmu*n2Nm0$Yu0_)Ihtxmn_`$MRv)}t<
zh5D-hJ)89M%*I!p@xRBVx6Ac%&PD%@=LQ<iYP86rt_(||K<D8-O6=ojzk9wah5w|D
zKfG{u`5m9v?(+2{N7xn+%Ni>fxtbLKYAKd9RXkSKneuJy-p}CKEr3_0&m0L|X>_$7
z5-3<D_wOS^Em4-eV&wc)n7@vX<@qVE0W!M_?+P~D;3&0zHd+X|ZxJCHrGMgjo|3c5
zm(ckg&l|V)qQz+d;UX`aY3Zy0Pq(BAe10Z+eUJ~8fiuNj;%cq{caf}*H>8;TvOoi_
zXOArgt-3gurIW|gVYbkWZBMVdsCjBH-cUN-d>qBuExeQ)_O`J`*RA9N7>&yMV>!C`
zec=kf_}e;LQgIVoqpsnEbr^Q~vCwS7``VX35I?^hW?q^FKbI0=7GbF*>12^C?79YW
zBzh}}wUM5j?gtBN<6=?Yq+FCsY`60Rz$H4{o;KHtVwW6z2TaiE>NK82*j7-EKun1L
zk86*}>HV77H<NG0=M`<fm}8v8mEOC(pl4PET|3+eD;q5{Ty_f7(uD7TPdzVEILwtv
zP<&*O*_$Ii#hl4hT`v@de{x`Q*lWa8^o;rhUvdob>X7<>4%puPZD9Wa<DNr@;l10D
zbLP^bMr*ccY~{&r+&4I72F2JHcr|>1V^L5mr6OXVnJ*M44NK%zx+lHw%rs}C1s7%S
zcutEfX?MK^w~>z7cvE*};rT1KH7$4~Tf6m1Hzlm9)h?d~Br0RO&*<}Z;AWD06){$r
zu3G$?OSG2bMxOjLEqKmHB1h_*<OCbRtRJuO0E6rmp05Z`m9Fyurtb>vZ;;-DF@^7n
zNQ@F--bwJN%!hv%i2ck){~Q#Ic2_0EHY9wX+p*1!3!7;%y9R3Qxi^FteopQu`%oBK
zf>`XD^=BO8Lixg{IWj+U)82>)b!i}Xpo-?W0hIoVFLlqs$AC>4tqEWN`lhUx%Pn}r
z@RAg6zsiF>#vwTp6j!6g$;l&=-O4p#!}z=|67&--eV>IK(WprW`V`<u1$of)6sJF}
z+Iyq~-@zu@fVh`PzgrD|@R5@6+4f^p%Gj_3!S85qek6FYfd9h7_M2ohCy$(p)mIqy
z_k1SI+Q3Q1h}}$^FD~Ty54N9U-IJ7zG2L@x(8y_HSFg0%VpkH$h7(w~O4$2w^J+yZ
zehN_7H9$eAJ}b%a<T=7?To58bnK&{?P2cKG(4X;u;BwqK)%ljqLQ8wefCl7HjoGEP
zsC-=bIn_BCP{eaf<5kMPbA7O+Iy7%B8L%a+mb_Is^ZiZRM@>C$ZXghkTw^X6$Eb#(
z9}0x_V%X|~`zfUGyOdwKhv^WJ_>H=?#)X2>dLY_lfD5G?a&uTrRqHNoUW!h<bg<uV
z%D~TajW?y}{9^Mi!kEbQ8FYJ`m8A#CB~fB9Qg_?U7viv%VJi|8^<x;NJ5O7w<fg3(
zZqxhkzbW<tZjImu@#+xz^E|?*87&N@0WH#s`_!!40+wfN`tUJC7J+g7$~7OE-)9&r
z>IABRvN(cN<GhmlF;{AZKI@R6tkCLo#v(9)at)KCB+MS_NAT5l?L|7!HakeBejM95
zD~nqH7$2EXUR~t;!6sgsG`TfD(o3;!;#S8~2%w>h@h7kVZ)V$Ig+_V~FHoaZ=JVRm
zQ3SgG<GiFc+~iBL@z^bo!T~M`J=~_*xE2r0dpeE;DIawZcqOD_Iuz(qg^y7letT^~
zOGv6pe2@;&NA;N|Eym$${pS%*jxj50Oi(nbJsOpGSbrW8S|AWp{hK9PaPxYe)^#;-
z@SWX4g+-=mYq1^?Xi6EweBjn_0Z3($6vxI^3+Ii}rT`i$PPMm<O<!|~0)da?b}TYC
z96g8eBr{=3L!Y@scvQh-ZUv1wkFPgcdYP3@_JySct&iPSuSO`R6NHqmJw=hdzmk@B
zhm~>e3qY7RN!qecX(>*Xm;b}y?TXFxE%Jft`ZJMWp+TLs+OwnR_L@&jH8*6lfBq78
z6-$31>7FCWX7wYwVU?rz!J19-nrE+u(D==YScSM62`TXyv1e!ZOU+`}MV9%ZecgR@
zp7&17EiASx)oHhUaqxOtvRQMc<+|y*=wu@DY9vTenfDj@$}m5kr&-VtPvqf!6|*lv
zp)dxIr*of&Cd~TG)I0mKjqXhhx5S7v<HsaR3iZD-ZNNiO9lWn#(^7d{Xx0Gp|6uw!
zm7r#Os!unN2VgOIp^Rz(qZeJ;PV02sKA7mpTsOtV^3<cgrfOO>;tbNkO{@q$3p;im
zX5Z(R8^S4Xk0MLJw(g8l49;}WBQT|RCQ}Q4G3!)yaiTn6LfCW9^JlGTy9Nb=s`Cpw
z_0cNAQaL(qILOM}L_V{2VgpNO3QJF(^~|Nm)SM`{ZSNk1=c(1#$jqE^qTS*<bbs1M
zV85m{L>|V)H@fc}-Boi`R?tjG{(9(Dz{2$YZGASa<S^bay<rI1AidPEFUGp|4S(#Z
ziCIIVyQnPD6M-u+gB}+rixM199(!~n2iYWt)vz>L{LgZoe~Nk)th;G&WC$Y?67h<j
z%$`<-`pUMPO~c2%UP%o*-ve84Ft^@Z$r14>Xyfhm5vpHdo#!0a@&8iC_CuL$`{L6(
zp<zNvyf=#)5*8@AN`HqgT-3=lGm9V-Z13hUJ)+-4oH0}&M@VVQ*Zlh05j7&6?^OoJ
zLBN(y0<7uR)o5O~Jn>#C_jZ`SVtupJYo$Y3Y6OODrtx{gL9<sT34j%!$eOMEyg&&Y
zwaR=aLJ3IlC<)6K=(v%-Ud*y9DF6^2uPQgtqKrd+@p;azYWnVO1odi_vu_fQTb-F7
z+e7Arm0puHdUDpCyG9Ba=2-HY#yXP$yB#V{8hD(@vyO}BHancd^zfYVbfGI1hz4Nk
ziLu9Um@5&OP~Y+)rw`WDa=7!BxAHR^B@o=vzRfl6<qDETs0!am*3mr4loZ6VHg6|x
zY7K%v_s;)nuK{irFa?^FY0u`ro^pGL=qs9TPQBIQ#f~}8(!0VU7@m4YGt2-V`RJQ(
zWsoDH)HD6EYBu2tHHPVq-BX)%YcYB_=DKo6xpD(w+c9A=i{qfYxwZWSng=DiJMh49
z^APsIU6KmBI!9^7yH1Ar8Szj9<WYY};yzeq?_d3{Sq<zbxt522EKy61i*i;Q%4Z?j
zT7ausIa;r*wLePgWc3kp%$d90LI76OXA7a43BZWDRkIEOP85+A&a(7_1$Ry*;j-R2
zfWj}~xS-8~2qdUmoq0WgiwuYR7544Uqj4||IVU37jsd5MX5o1;gd2=;7Q&YCW<%A@
zO<bt=Ir-g0Gm|gjs^Rl5KWaBaGr#lE7%lwab~!0PWX_jc3b{Ybe6Y!|hZ7mWWn>~2
zoCrk*Yv2<TboWEH{4Bpx_@sr%92Wz^V(y-usuBc(81vr!N_}l2P^>NF`V%09froZn
zdhrgON3$1o`8}$4QmEfKcZ`RGZYU>qj=%gkTzr$gJjl)LZ6}f-C-5^a(o#D&=`!f{
z3Huf%X~YIg#E(CxGz+ERBAZ>F9|a~HCu@Medz4Y#H|`x+=Tt#~HnC*Y2NPesu~=1m
zy%ne)EiIk<6g}q#DMk9z30=P!7b0d1*OZz~k@HW0F_SE&wI(XZg$(G1=9j{jKn!#j
z-iz(JM~$H5b57+Mooa<;Zmk-1kJ4joGJu~UabU{iRIs@ZfwX!{C@5QJNCrcx{E2Us
z&%BvXc4d$Op3`iy$%4_*uDP~G8TJq%+BQ-bdiRa0l~RGKBQ443XD=KF#<^>*QtBi?
zck?y$xPzhxHAeqV|ID3c?Ri`zXPe;UHVcLwDz->lavY`fAs;f7S4Xx6r{JL&y~&<R
z4j;0VHxY$BldM;<3nJoEYLx^jfoEu!3^r;)W(?=14xRn194d4r;{NSId2_hZme<Lh
zt&LjP7=Eqy>`aD5e06E&gNc+#tv~`WLHH?b#_kO+YGJBNNE>Qscc28ghE>|w;67AW
zk;}p61W#O)tB0CHLGyNx2-PAlp${eV#Mh9UC4zJ=&@f#qXSb;D%LJ8KN6(BMKAmYd
ziVoouGQ=5j7Etk0`@}>s;Zz``)XKJcQ;Btz0wbCMi0Iy?AOh!3V?XxQ!^WhSt8S2?
z|5T%(5t!ROESn(A;aheox&`?ms@hsd;zRoqlB#VLs=Zk2<+LE=Z+qD$AMA-Jd6g}N
zCqAnYF(?pHY1Q?SDD-46-{zmjDff0<Wx<TpCYep{dxwIS-r0^}C{~niL?%6ka|uY3
zH#_5hurS1{q=O~)udAPzDPXe#qh?&n;e+dl4cS)Xv<j`D2*c<Z^tHCeJzuwWSd;W@
z-Z$;-N!W~<EsjH#_aS@bEG}}+z$ecaq+W#Y*sT-!HLgFv2Xh7<2Ss=Cyau?;AKRIS
z;G*y|0PjK)Y3-IcGHmdiP9EQR2>4~-WHMEVa_dhOtSQ0SBlLNcE;K8GkNz&A+fO>H
z)Z$^-W(}~vU(o7Bu6(E<<huAqvI1Pxw*CYy%5QG#cYLSJYalJr>=oQ$k>j;0N@_BX
zMQ#aF+mbH3nE~`ymDAz&)cvM#c8tXS`LNor$unt4v?Om-#lJFJEZF+45?$TKQ^7g7
zIvtsMPuP@!06bh^g&dP+tPmi`Q|b%05ByH-Oe^W)UP4JB3q5~6DM>mGk4B(RTWp(P
z*x17zygDoG0uiMx*j5ixy6?j@_*F&D)lxmwT!9K}zaL2vFzERrjJ;wO7iF0y@NHPV
z4_&8d$&R5y&t)g4cDs|y?7kVPAOpl^T{0H1B1#hNm1-6k=Am8M@(Ar+77Q!=*;h^3
z!qISya<p4vkuO-NGhxxFjfhlj{Vlai^%a2$%OMPl29Gw=M4}Jl7Bw&3YdTjhA(pTQ
z9H^caIGvF*^~OS!-efU+>&i+t;jH)aCxBHoqhjPx43(QzVZ5}>NYb;54RM8~_8K16
zrRt#`Xqx)qMsSR@ilQT$RMEt6ND6uKn|cW-t>EFr!=*N@-7UsYM2AMfsHm5|O1;HH
zwWGV`4^#+L2(fjm&+bdDxcrtwX;qquG)EGqS>J+!dN1uo+AgzD2w7NSaZ-p*SlIo6
z2UMQf-VQPm0!lhAgG~>)kjvF&<4@R+r_b1j*AU|(y?yu1EIE}}DW&MguOxNv;TU&$
zCUs`!eB=PYPHPTSPIIVvP7=uRNZgmlPA`EBm~-3RArXq5d&rqn>+&Yym@I$bWUuG-
zjAqLmF*^o%r-8ETwPMKWnS^Du*@t5^{S9;i3u_fbT;Dd%mQOkbHxpWKZ`oN#=b>`A
z`YPPHA*fsl)4h@OYa%G{)^U8EkB-+PGE|v9*2Q}r6Iq>9T4rZ0x(94q-nUyP4gp8U
zY1E3~3Q65BTa}!w8~1@Wu1>$RoA`aEvpvFts^NNJ!Er?3vCfCuF3$s@TR8zB{9MPS
z**5Gn_?FgB^8j|vF2gGuStm`2I(EHrq<|;0ZFh6~P|EzLoevlU`(*Nv$kFFk_mC4J
z0*HxYa(D(~)a6an6Lw}|-X9uT^+5N2lzO)kEAC@YkHTK&ZxlF$l=~!D(wVn>DA$=6
z0zGchjaa9%gzC1<16!_jvAz0-YtH8Em^VGY-?*jj3arAglJ&1Q2{wMcP&%wu!N-#4
zkDhlVxyzD{N=lSAtRXh|<Tn>mf+rJo_bHkZn?BZ##ACj&G^O;(p<c9ck_qaRh|@k1
ziEnCA$3;oMB-QZvAYre7825H+p+0i2zugt=u0WN!6ObMlBhB=DQk3Z7q}iM=llT%J
zGCV+ld`6A46fkd@Xtx=)Y+x%7!j}?X<v0Ii*L&%^TH~;*Bdo<WKVa0Qm|5G(o!6f#
zA2=UT$KzsNk5Cu1UN>-gMkt5urKrj9RY>}l9iduB`G<ODIP5E{$J_KPA)u|h8iQ%U
zwLnTGB5;pFoN&58<K$p;0$rt93S3Wf^+Bh<)bFG-+9QDs?9)8FH|x^}^9PJx%rU<;
z$Yy{!4&8PO0Yh?I@P_Usd@SjodD*A8>@ZDka3Z67>pfpI*Ujhy4<;)As7=bsC=zea
zEF&^ATTZCf*g4vk@g_9pX;#I@N|-Cfy<8a&!i`E7e>S8#ar7LG2rrVveCxYJB*n+#
zclt+cozVs2)N&}XHcq{Xh>)r9(?*R-UvTU=v^(~FE88b^bj(lUf??GO4{j=|!{YQ?
zIZ2-s5WLJ9g^G9m5~Tr;1qeq5hTca%$_Nw?80(Rd#`tTF&vrw>j4_(I(!LkY_kvzz
zz5;05@^7747Fsd)B>CD<EF=2H4ufW5GA2B44ApJZVlL9Z1`?w=Spu$iIj~;={r~**
zuXw3`9!7Rb<9R45)<*pAUI3{#%BLENx5DD*(-b`ld_iN_ZcJN3lmi^6O<EHw3T)lp
zQ&!zBY$4_-ShHUr3k4ky@mPP{F`G*$S!l&wIUt}e_2$kXVO6i*a;sh=!(gMHjCxq@
zJezyRilB`CMv-l~syy5lv7SEybblr4^hX{iek>`S!Op)H2)4S8;*TDv8E7L@o^65k
z4%OPrp{CIp3D(rk_CC)B(8`r!VBsH03a;Gsq)EbXzxJ3uY9o+sO67-zzG$C@k9Y2Y
zSI_JcK2cM5WVtBq(z3bJPx2)|C|kZ!_N~60(-V`e->&Q!x_z$s6ZE)!+e;ONQlQYY
zc5m<`*?oAgXpI(8Dh6WF6Ccx3O5sM4R1@Jzz4l2Vb6(#yf(mX2KPMmTKTe$CSoC-l
zU~4#m3Gz0bl0x<@Ulq7m4tdpeb>)yE)9d!$;?IUweu?D*gh$%9M-#$#86U|$W7A)?
z?_JslY=3mJ9<B%wo!r#ZxEFKJ{=IT|gUi|CioqT=THRHz3sCOa7Og&qNg#Xo-H5QS
zcwr}ZSdq&%iNfK}Y{-5YaTJItO;Gx9@x3la8)O`^c=ndM1OnIy1Ob+ECEfUdJr+Wb
z<aUksUg9rnGu*ZI0`lKUU&1Ll`*PGw4%g9Q-;Ar5$;yMiV_!4yo5M!#mJq*q^|s_Q
zHC*8{wzo>)j2&6uikG>%hSx8LoX`Fcv!=K9CFi!Ox5f}3D$}BZD%ew&sk@Ew`U`r@
zdtV&(DR@v!VI)}+!5T%p9I7{1jwbSIRA|U@gG_aa9>`L5s<cnbJ}F2Zxyiu^*0-{O
z%zi=}b4&}gK?>9}G%N6p^&hPZnZLRkbzTLg$HSU>f{L!uzx=))gKmy|ECVJzff-Fv
z)s!9+lsKC*go3&^S40Bu{_@&33tuFgD|;daa=86G3{m@NYGCO&xfpL5u3Sg;If~_%
z51FEvE!Bzb0Y%diSgB`7c^d80P4#skK>;yB_`Mh%q*|r{`}3w{kA8R^em(~Xn#TE{
z%V(eU7QDY%_Z9BbLuHN-EkPQxqL|S-x|8srg#9LO`z~LwTUPPr_q$`sdp1W2-x-KN
z=PCM(Q;oS8x8h**s&ZMJnXjaT%FzqeIR?*@eXsZaz=L-_okv@)1t=Ni9um7%U)t_9
zPBC;7J3<;<Xye+nP|v6z*n@mUQMe`+=Jkc$``tB;vll`3wt#;UyP{}PA&dJDmr@l~
zTfo!3J4;NdG-xj=Xy$xMN&%r@?(j(AxzgMjbtAM6osYJ+L9-PmH;72Di-e5vp*q@Z
z!w%GsjHkms=0(&496O|qx8i2x>u^66^x~XX`tAX?8_f<O`#+40m)RE*ckV`;VJa+1
zJvLpv3O6pwr#G6qcjG9U-pIKn_h)LXWe@w<YyktIFNw#Kb_3<p6i76Qfk9kur@HU;
zzFrY3Tcq#5Q90IEE!`bqL~$T`?7FhixxQwuI4ZAnAz3)8@Vn2l#jVGSQlSk>MJZ3D
zSMo;O{nEQF^^G>4DC&@;36VO5DA63DD9!7*^%Ho8M}E6I%<{oKiZPD8=XxL!eE;AI
zhurz+haG$C>sCT2feu*s3*~T0nkD_7_fjav?uOE$RMG5n@!RVjmlcW{rIc`8ed)U9
zhkIwLgiT|P_N|skC^_Kxh_(C4-;Qj4OX6@8u76Wxd7=Ift@O5S#cx)WsQ$H~(>oW9
zVItDtht{}g!Dnp&+NPLyC6n*Iv#=uPr>F8C4g3+dHJ^RUh^+5gj9rXy_#qG<DHy!N
z81y(NDdhT4<#OFtv0rGW9%zV@SQ)Xe_t^KS{FlA!&)>0~9Es%?Tby4KJrj-rpNkuy
zE63a5x{Lgzg}G7Xd5gNRDd+Zww)+HFbMW}MEFJwvM!Cd<oEQ~&D1(JH@hG(%zxZKT
zD&WF+5CrD76_=scth}%!P5iIT1B>WOf!)|^-`L?EVXMxf1*_pID7O2j4wT2swB<p+
z&Nz#$7@#aqD(B0*_dyCpjec_<P4DzbVz+MvE}D?(rpQ9K!XIWdqz4|KXMb6XR!j(2
zW+gu|Tj^P6;S8ra;ILH0oBHx-322FA&vhBxBW!vqa}V-nYcKXC1K8`9oMgrXU&KR=
z=!=fCpdNh0L(Z3dFxLaP9EYHi+l*S$64Ct5loCS{-DtliEu?^0dm}~D;_-r+v+rUh
zl?cRp0zJqcnSU7NM2!k_=@GyjHiRlDH2(Xx%r0z}^;S<NgP!esC01Sa*<>D=PUoO-
zM-!1Q&->X=PwC@}-3JDK8Fw$#S)A0(Q&I%wXP<CM={`l*p5cBilW3WLt}OA`?M0B-
zO8PO6Kte>#w7cMxjkmx@*Zg^^_wIr`=yQ=zZ~I$~+MlfHKB<Whv24<V?PusXO->!h
zS2<*PuLiuc&~7U=b)M{>Nv_|mIMdRt6kE@DRugh^lZ7lHFsd}~mMw#Vs^OSGd%%&3
zo_3q>jq)0&u(nN@LbLvljzr_aJYM|bd1pS5cv$LxdvE#=o0*Nq_rKSY`Ni;7c)dTz
zGjkQ_C==3skLF`6<+$Dd;r*DYcTl6pcFQ>B6${>@K%c|JUUUXdmDwkG^pZxKm~slo
zvZHz1@Zq28cDj`got+Tzg5(xc7TYV5W}k-H2YFpz99l=`N99dUg_Ks6Z%?YfrOg?L
zu*x@P@fK>Y3hudAJj}klY?&JGp_X1~o^ZR#W7%u<{^b;2ENxVvCt<th&wG6zXLuaQ
zEiXwxJ52T(-F;$YN$s}1pU?6=jrdZOe6&9tU>P<}_sV^Q=#^zZW^+$a*S~Nr9Ub#h
zrYmaptaHBVH~-f%TeGeks_w~i60Zp5>*G|~!waiAbpzc`fvbh{=0SWM9qxTrOzokw
z5>cdnv)Aw~>c!lki#0Ht9d&45^J#LZ7xVDkXI<_#*D|3yf0|Xxhz-n3FrF7*_T$^R
zG+6bUIa*A(a4lvl|G-n)3`^v2Lkuufvb<pfUrSIF0c4wr{H4KdthtC}7ui-Mv|#O^
z-PNY*mK=9X8i|?3L)|$xkPHT2!Fuds+;-~x0jg!sk`lNNIX6oD7%N2UKs{y7L4;)w
zd3|RaA;}v@jIG^*VO<6%zC=ej%c7imJca`1cFF|U9ej=l$xuPZG(CLaV6W1AQj*Yh
zL_uyR0a!1Reezgd@SU<;vTXD(7#1&l_zNv+9hU^FJfD_L1TNS;(Ex3Uu5dDzz)g;D
z#Ysw!_F&>%j73MdO%#2)U)V5Gu#C`{PR9PQ_?dn(RK{M%0SqfYf6_rR7VeT6mq(W2
zj(RR{l2CODy#UCVz(A`qOj`sP!f>vqB>tZbCFqU7VeHx49iXjZt=I<c5pdq7)Dyu`
z7#(3^vF!S3X61ZP&5#oXxooH3HB?U7Yn=dtn73>q8KZln98UCBPGz1|6Xf#$P{oTX
zugd@^9bB0<s4-((u%;^KjohKg!EIn!t7CR+nDKt2-V7oVhoHGf$;)}pJTSI%<j&5(
z8wsfJVGbT$!Ew;~{*@YYTMSr=^~(S{@M$o_ZeIx;?ng5D)EvgCLodSQavb6PI6aX4
zuE!*#RDdPs_V~<-=uM^YDKgadUL!rmo5<}Jv_f;OD{!lcixP7J>+Vqs|H6eLp37Xs
zLK`v2eGYKN^X0tONHG3ns5f<S^>Ed?RRRa$*X&||;+xS=g}|=eJ3BwXawh$XXmb*g
z3}~C}v%G+~{UL)rEH)%b0~E7_i#j{H2LiTn#6qZ>Uhyh8<p-L(A<hLDxWX5z&{%1@
zE9M9WN5|GV1aGFO#XLfR!d@|n-J!+O)3U0>C;iYNiAgun#D29s_?ZAD#$%qXhX@?r
z0}@K$>}`+2JOEvP;2EZ5Sx!kf=7Sg%Zrq@-m}AnFePaJ^4q(_r22_3WKm&9Xo{S6;
z(j@`f)Dgdc-RH{$4s<^XKd@o0oq|<Fpsu$Dg~6792{lIC7jU&i6i6jP%ZvNC*xu*w
z4?|))i8$p|=Bc5*GJBISS_l8Bj3StSkrcSyf`@9j*Q?Y+yjLIvuGRsv+hqcj4vwd*
z3y4TH^}u~x{{{`vS~&Y8O`dv4)e9Z+f)SMUXB1(FC(hlLDYCQf`9tClS0LG>6sNX}
zl>$2~5tQcIhHnx=^yCL8u1D9~$)YacawB&}SWk-MI_ZL0e&9H8&3&t-MePc&PEk|}
z;9$ZtfbRQoGvTEKm}V@5k9nsgw;&lFrEr0V0MXY#+Y<O~$PW4Wd<wYac327{c!5cK
z$Ok@#;OQnCp5`)okOD;Q;bQI0R!AKDzwZ!Wx`!=ZVF)nUoktRSfGs6A>il><m!t_L
z1})YXdDR?NBtl<$WM!3D^_R#{QJP8jpv18R)S4lJ7ImT0kMJ|}_OG-@*~%edCQ-hH
zM``OD1!mj#e#|ldj@~#eY604&M#W(r5-UB)g4!_wRn0pl-Xusj$3PqyE)&=WE_jQt
z6auqg9xICHc)X1W?D%S1CW{i&(s*eD7yU`x^vpTnRf{+sJ{V=&u4JW<UL_5t_z`LH
zqVD&>6K5gMI$2K#0#b6(a>_j%r2&3Pc}jn*4sn}IC8KAY%K5-`tq%+~@Toogrjl*Z
zpq6K})EKdxfK8&$bVUBQo@U4^3lsMCDx+F3l>VZuP0FW_s_;QyA{tCY<7d*Ql~HK=
z^UgJges~ay8{+zAaWJWgHt=(`KSd6!moC6Xp-sTxJ{`yxH|-<jRZb^qQDK=H;1L8(
z%AMkOcqrz*UJdZsX|&r8wI6Wu)^!n{6F$d+iZlV=K?@E%n_W{vYfRW6Fe+;?VF4l9
zA4eS)!-S1y1Z^K?`pYY!^uqFhxjtoRB)!w<<Bw*LA-5Ab<dyNmTL^0pWhJjBmjWkW
zDMjWH{n_b+JxZ*0aUa!FiA7OInT%l6L&^YoC30Vngpgb@u#Kxem_I*4@d0qr_$ch9
zB19;@<%MuPVAP!Vh-mZkupYSlt(Crf*5Eo7wc;B_X|NZo@5hU}<0VJR?htp>s}xSU
z6wVp|rcoipfGECVi{=2adytT0j*%=VYl-)&L)YK3psqbAfn$8KAf6_3s0Sy707vdW
za4>OuEmB~J@Xl-HtpwRwEH1VyN@Oz9suc3KWiuN}$L^<F(lHD3+$0_{oT`Qq44n>*
zo1qVl?IaG3J9f$8065rGAuoRmc~k}MGtJ#0aA0-f^aX4iRanpQb)7Z9_2ikgw|vO`
zqD+#eL2Za0b;hUB&H;Zo-$4)Taz`PaZ7ku+;cnE;DiUEgzadIgW8$UjE}X@H7WkCK
zwt%Edekl;<qh<m=)%)`36=3^G8Aw5|IZRtitcsziTI7`LylZpn1ZW~vxPvEK2F?cl
zSR62NE+*##n^8K3=gQm^k$VV*SpgXqCGLg#Zq*52mqQr!(a1L(&L4Dd*f5zlp(S*c
zqE;pX8qAzXCLg%+s4$m7oDw>epL@s%{-Ml(*2+Xjy$HoO36nJ0ar%6Kh&C|4EEcSr
zj0_-yqBW4Mgh<j<S7}mmO6gYw$>de)u#fKrU}woOmLYIGS`4YA&fUZ48CBWNMkV2&
zw7Rp1{$nD4e3rfIvMLj$YObp`ec4qZV5gooEoMR1{SyJFbu24tbOZgICz#;jSNLtK
zSQVVi`DzS-Lr)ue*SB1Nl2gX^1$@J>4*@SIe&`QqeK7Y5h-$)l(IAErBE?Lqrj`j@
zz(x0nGgJ9Tcyk{P!`VA2LSr>|o=YJLxTpy6%R?U&#?3^?HmeAqaOujZoVKz<GttUW
z;YDTEx(QYTERpd&N)z2QPOCw|RRZjseJFdCt(S3CR`5Hjkt#*e8rr=+fN3-KQy?Ak
zZ0#nr=}iNC&<Ecj%XrL1b)NOn9-yt8Jl9XJqC;Mn2JOAlpc(gRfKpL$Uce7NT)eR@
zii|k3J-yA9^AG)A`?UXw#wtv8>yQ&N>Ie@qYNA6_8rp8&ix+9)qsLc(L11F!bU~-g
z%+5aGgBXOA&Q~{@Rl-x#On1mLg|%sBam&v&3uKYwh*lE+2}cQf7+5pFO+%ZN0H$^H
z_N=mP>#>G0A2RBy%_bKMXL^YoIcOBz0^1%J-XJk8)U*u>Eiv{ief+%CjPHn}v9cF^
z1ce~tc}Io^hgI}Xvo`PGF{9L8H14zd>=PM9T^@Ap!rYycY}6+a7Ij1aLl2G+s}tH%
zr0tZVrr}t-$MQE@&o+(0W)DZN?67i<hMfA+gbK5^!+YvE&-0BB`D-NyEVf4XoIKl-
zJWi^E-ovd#3YYB%!;K2PimhHsf$AY{)B|fMC-xBw?hg$9yKEAxt|?~Q)%OC{q*bj_
zs!IdxRyBRsxlg^Q$xah+XF0PP@OxXq2U8wV!h?&A9pFkPE0qZ6>^E^sRAJw<nyqQh
zE7a1U>#bRck*Lrh(5a&UsgEdFxOP)TZZr|L0`Hq8G)FWzyhEN4iE~K+2kBO3ZGVgf
zuPLSj99CL}(3Uv~C`V2D)@N%ULeMJEjE$4_k@-Xj_&PiO!KZilZ78wIf#TZ>v4IHD
zK$vHk(Wz;e4je7qHpGe5Tv~;11gAP#f=-$=Vn8+o52M7YqRU(#2n;yKUOWTza@&U!
z^uWMvVBkGJjdFrp(BMbqp$kPiT<}L(Om%S7r{|Nq+bHY>F6J&|$YaN;fv3<DbyEgh
z5q>`0Uk=S6Mngjap)k@f6r4Eo9j|~?5JkE)W~2m9w^24!PSEsOlXzB=lE$=7jD4Ig
zK@SJ-O5jtLpRX_rb<~LII#Yp8r_j!7ZmDSkO3vT+WeCdq_T~{IO{&}-epw&cfuF@1
zh?qRDdXo!-<*)E8n<Hp{_B6Ra;XZ!TzW?UR9cXnaPa_l3YQ^d(z}7dZ4k)#|^d@%v
zAi)gF4KaTsF0s+$RkfPo2r$shO??Z@WOVyPjj3gG6I23z=*Xys_fhE*;FrtrKQ9G1
zW}@qWfjBqr1~KIRLmy~F$EJs>+zU>fsuQCT);Ey*J<KE*@C|u0Xpbl-#S6eSEZ*!#
z{JjA8Y0}KC>5%BWBi|JQjOR9Fl8jF4E5NXBa%0`eUs_e7bT%MSGlz#n78y}JT$`wQ
z*r63q<FFIu{)w8?+>kp2yq;+b{K%ptN48X;H_4iou74y$wZc@ntu3u@pu{}KjczAa
zFTK1w350oHYR$%(LQ6nlSc@J$Zn?}&UpsgfWxeJKKe$qoerfTx%?)HTtr~=vK`vyo
z?K|G!9~OwLTAXPL$S?^a?73^E6j(FmaUh%T=Lae34m&(+M-L)<RAD?@`Qq1geK7RA
zOh6Gaq7tcGIj~8d87?0TIx$N70fsFEs23ieA6-WVbUi`_#aAot1wvN&R@Fo+&5x&C
zB21q5A}U%_l=5L`3_Q!5kwgPr8OI}W1tUtH321;2n;@uB(r%NFu@+A=JA65UB2s>S
zN=3OIfugTayngewU!tZD9Py19ySX8;N}f4Cr-ixTK#|)52QlR4UG?(rpP0*M9Ru_c
zl3}@3r*R+KKp{f9V?(_&#dyt;<1D{;&7Ew{H_Amns5!e7JYqDhf0Ja+FGfxR!}jIH
zh=REC0AP0xY~VviFCEYkWp3d@iRy*jOI2<vfkakLX2lW62mTL<z?MvUhb3K!$MW#3
zY}_4S;OQ?V6`dB7+M|LPuB*U+BO^C1Zl+v;Hs%5k8E~Diut?^4;&qhw(PyLd*gT9B
zYUvk5g~-6`x~ur$N}@bXI&_Z@w={Simkg`#XxV0iD6kH;wFQq9Am|ppQZ<Z;bxYM9
zRpvImx`5b_da_Ojwkc}ZSU{Q_Kn1MB^`KC>#>8pf=iB?tnWg;$GKu0l6uKAdi<}#J
zDxIp_L~79CoCqFK<}dNb3-CW4)V(T2>}KGuRFm;0r24mpRa4xOsuiu{-!Wfe^g7a3
z!~ukFWOy&x0SEToa!N%h5C_^*@QE_-X20PL<}m{TXVn+6B$;v|VGHn$-dzl|v7%NS
zM+2-pVW5FBKHrEgSec>->4(`S*$NSIT$}m<Ou2HI93od4(Iv@OwjMEpKRzmg{SUvg
zBKKRjj8)|<(yE$ykPT|wQ1*i2_0ovv8+j(H3H05K(gk1wU8Wb(fdd5@KZraU>-Q38
zFmPR&cJPlrLb?c{ppodpL+j^7t;HC18G>ILw>ck)l#79X?51}_Fs-2<X_4d4N(ss{
z7@*ba%Zh_r{+p{MsCC{3zB-A8@B*dI1;sUekl>;cf~0~a%KeZb(+ZF4n{CQR<@v?p
z|G=2&QW!jx;P<_$FL1B+v)xP7{KLfu;d1V|#JvTCO=vjQ#IpZhoF*YB;@dpupLjQ4
z>ftx!&5#q$Du}?9HBMTG!47vpIB-EIUGXfIG&2?RbhP6Ykg=s&d+s{C2OCq^PuBxK
z2+dvQh#Gc8R>__S5YY-c@rP0Y17=Q8G}DC}SfD7CFH+UWA_KbfNwBeEVC6K#$3NfM
zAOxsd;CNHpN~ck>^Zxg@8%@1+zY`Am@_U7U$?B>(Tcohn7CDO?nP?Z>(r(oCLCbCH
z-M<+z>x-w$GvoSni)w1XS%v&EPxpQqGTA=#k$7C-CM2~*A}7TBYjr`^P4GOc-U<~p
zHK3L-HnHkv*|bo+SJz0;<^$8^;hX(*5Xo*Dmfz>~n?d(|m1PQ#(#tk<pqiT?*OVl*
zj?=1WdgsSr3QUoC5<P96WZJ4&<8}7XSd(+Dxow7npam-IEEB56m%Xwr6--V7^Xp5K
z7{-mVdeOpjE6|J8GrpVQ(UlGrJ^jPGGd2uzqGE>zYIB1Ny5TxL^G0)nHnri8gbup4
zc}_XS+B-70R})7Rc#mG+nH#)4dd$!0-~VT0{VK<M$&QX*tJx14T@8A(Hj)wMHe12R
zAO7r{*>hxD9!77knn#a5UA6FS>0L2Fq|g$0i7D3ci!t2r&D&o6&aCihmh(LcMzUV*
z^gG%D_4d`+Bgd5EAhH6C1*+y}i(DJ%9NB72)~ZITlQMIlcG<j_Vs2xB%9V9Bw?rMB
zxhH#4hk1`Qw}bVRpC3huU^dFkrx?U>z7kbfoZR%h@W^RXt2>a<{XNYo%988sP0c4(
zDe%e?#r2|Wa=J4^v!9sDen~^lgur|-)89#J?*#mrd5<Gg?pXV+KgSrW@w-qCe+|<D
z%pvu_laMuqa+y4wjttE|2<Pk+`@zgP4emJJyc2LOWOZJFGnD(oyoT2SgYN7R^?|*U
zS=%edDuML~<OFMKj8JC;^4k5Rt_;m|sRJ>mIgN>=3YW=^?%&eTiE%_Vbb`>H@7smJ
z&+OXvxodJttr==-^cdM;cW7a(Q3G^B>b6)&=HU<-yx^MvhMV|hz?G)$O(HA}n;5(k
zU>rBnALY6qU(FHFrCQVs3yn3~Am1P7yq};+`2Hs>GddWm>_eb0cej@bF!U<~oS~6u
zT<YJ^n*BX@ld7keDKk@lGX{W%h$jtfoaYRtRAZ(o)w{SWcR?&p;>;n%T{gt+0R|xD
zRNW9KM)Qw17c~IN?p6u7f@{@RSdruVkjTF?aK!`f_P|`hl^~omvV~*dfb4NyQ6v!a
zP7maXYBwGvyuL@v$xOHV9MNwh1Wmhv1lx}1^@$aw<JtD7kA+eSYz_9hfS;S-o7YZ?
zrGNVh7cGu%#zCH#IYH$~fE{AP)USa9`G@UqviU<KF8Ugn>v_<?ha!-mzXhs|jy);`
z7!gp{b21ND39wH^`7M<yi)mz4xii=@iHT83EjN&7hc+g+Uoqn(l>*wQ4z)QHE?>zr
zVTd=%+!fagfZ=b}Az%~#LqJQO7d3S<`=}VcF&*vSsa$F8_lL;fRn^%A9cR})t~r&V
zX4TZ6#AB><leVDwM>6F2^?D+(k{wD0Tru%n$g@QkR%qRUHj@Gq4%KhSlP4k7GcP1o
zi49%_UP-!+yr#}{S@Rsp+bDwSOV^-;UH1xbA5BtDkiKE5I?+u+%!<OedQUdLAumE?
z#Szn-tE8&hz_TI@AOY+9elK2Aw#X1LcM7~5vs`B6A)7;06fmr8P>jfmkb+X;0gH+M
zF-ayXT|S`o>=pMGaG*g0)xorHq^;m16*c-Q!jU;uQ0vHnnatUnU{+RSfaz0`%vK{P
zk>dBrBF|>?HVMQP+yL*Tx3Wm|&u<3_ogp;MDn$%ez{4?SRf0?k7r+U;3%z6;c#4FK
zdWsAN#m`kL&%y;jVMwb3DDt3e$65?6Ib`_2z?D1YB$=tRm8MQ{)kpl9hH#S>RpB=;
z-`tBow&yB<%DCGS*?x&1|2zzas+3VqNcR_8vKpvAFc`d&CxaXoxN^f4^q&4KwnL2F
z^+S7BvLOF_?+HvXD@_~OBsSQ^&_EaMyWM=C-wyJSeF<8m#(X0r>7N?`28M|*OV0+i
zEfPH@0a=vLA;XvuYPe1@k`Yuo5e<qz=J>pMCsJj_0`;ekixeuQqD_5(AH$P~1mc#|
zM{MN^lDI-g<_jbJ#k>&KZ4(GE63|LX7VWxATj<7>lmci%O^zx~QfClM^vb)G!bd{f
z#+WgZ?INr2tY-we;CA$4<T9~1;rH~kzgCkWCCRh6kS)7>jNrlBv<(%q%t#?<(RrRD
z2|EOXv5Q~+UjVw_$t3|XObUydGO9yl{S<nlKv|JNF$?m(=n%C7XSBZ|ArbT&F6ei7
zVaTwGOusm2b3GcNo;2I1hCA<3IS(HvQY%JC+F2Vy^}02;wy1v~z$hPI2e`tvX@shQ
zZ9Kk;J4&)U!STlw(c<&P*HL9k8G*$J2K?oYU2=b_&FZu`&7J)wl_GbjGX<ZXZ^ZsY
z(f%r<6c|VKPt*>A6H<5Nka9K<>{pZ{X)*O%LG~J8RMN22h)I`<#3Gdl60O??d9NHU
z6ioh_$uKES^X7DmikEMtG$aw%2YGz12Ju=WN~WNvh%lUpym?B4-J~(nvF~J`nK7cU
zWv28v%|A2Zi1wrhm*WvmRQz;II+mu&l$%YQhf6xI)374WiR@S>J5?lHgjxZ2*<E6+
zEX<LgRto&(ujmT$)DTmaFPe7Hm-{D`u%H6eKT%-3<e}gCNJaazzo)>salKb2?}$hd
zej-o*Qn``}M_SeE9uf4&o=R}_wwR1+5y>x-a)>Att)^6z4&HNCr!c#_%HN^ir+*I1
zC_?1?zc$V?tjXwq_<+)}frL0>qZJqeA{}EOB_YxbPy~^7fCGuqGLV)~DM19JyF+EP
z(w#C|N|4Uy=J$V||Mh$EygAp|eO)`}d-@yad`K6npV;AiOo#9Yp1$_?;O$VcD?u5)
zdC0dRcq~%hgR2~(bWGD=(Zn*|+3eN{>|EWdDQt?c_+sscxXsC-tGV?faewMo!(%yo
z7cTgPP}C>o%9!6ph>hpj8(_s6q#W&&y|Z=%Pq2pxY*_-hjIbM<Wa|9-u8t+p#rKVR
z!Q@2phEK!tK%bvlNIc$85XT!6NB$E4rYS6s5)JFuaZ806kuOB#0Mll>N*phHXHdbD
z4Q5Gp7l@dKzL=(FGQ)ZFjnNZKi53vjk~oiASvy?f(>LkAp?|nhWp1t{H*=-aV071H
z7>XeHu&WBj0y(s#ktnLeTfp{z7%*Vl48hANKndyg+Bi?*rriaKM~=8Cj$hI^>P@g6
z@p&3xk+FXATyO`kk1rEE57yn@NU^=>IPV3!=5VFLuw<V+zy3h-T;opxcM!2pchM3U
zuey&t=p%T}hPF%`pBsqbaR|%vjB~<HHCMVi8ROcU#@O)77~D+e=eTm@a7XC<0^#d@
ztCSpEmK67K?WAM<%GoRC8{n96m?hC&mt*%}^@gv@Q09|XBWL0F+5D-n2Zn*5J)kif
zO1iT$$`8rj7v`qIIEe0IMx}5zH)(HJ8zs{{=w>Ald1tSX#m(9SgD%y7;gbeS%Cgh8
zfZZCa3!)lONN6OBOQ(JrlE?1qGW3{DMXWua$@+bq+naHk_BFsVdvW(B_kau%MN#Sb
z1Taw<4er-5283PJ8%)%Ge~^{y4^rg2ERv0^z3n+2AwRAYg7*Y2lV&LS6nMNDzaH6u
z1ghc=e#6Ywl<F9EWNI;mdLi5VV6L(ktVZM~gl8q_d+F)E1n;9Wf4TXO_!$Fvjj&IA
z_@6&2MINf*Zka6LGsN+p#F5U~)OeruTYyZTzt^(hX}OxdzYQM$NheCWny3qG_JnfN
zO+GVWa&#a&!Rb>&X=aECvkWVs!>_uG7XtkSC-dIE4Ol-99d)+cF6q60kaDdji~tQ#
zmwo*5&3grG5HMrncTf*;+t+St-yn~RG<FOeFR6|zMGmXgiWQ_$#QdQImL-{p9WD?3
zn0uZEO<p)~Lr$g9-$mKC$teVl5KjOfK5Ur*cDu8hys}sG&}hHwrm!5w&yuxL$l;7^
zl`XJ+2?rcJ@ybvZuNCW*F)g;BUn`CmfBhDs{f-(Rx_6)54rlHT$=yB=Q703qQSwbx
zsGX<Tak5xQ=dr_Sa1Lbhk`ReXd{D7gE(e$GR3|ev4`1Lvlvp{~@nku*&6d1#5*c%E
zx^PtnZtv!))=sJ%oLs-T-7#tR<s0kiA_oLogfWH`Pcoloj~u*aSotuL4u8U9?KIN<
z&HtcFnC;7D3J~IcCb_8_Z@N`dzr7YN<x|CO1=PV`HpXcz&H%X6{QKJc-j=yXUAfcD
z6&Q^jO%?ZNCGx9ED%xEM4!_1+C9+5OOy9B#{xX#HsH>cuxPh)pl7_!DGjlF~<+e5p
zy)~byS_$HHjr<VRI?ILCgvXe54~TxLWk%D?uD%z}5*?*wFjEoxk!b7YF|Asx>RTjl
zEZju-(UHE(3!cOrH*@`04es_qR~+_d9ANFDGb@CC>7B>y``0mF4Vpdugut@{<{f38
zL*W;PVyL(3Z{lmJag}Dps!l;M<3hp&HQK0<Ce^dAG{&33f>(b-D<~BBx0Pc%>Ex((
zb0!pmBrW!FVMkrU)2RL(5)|~$v*+B#IFE<yfQzc4G=baV8W%Cwv8f64PcR(DVSNh}
z7r7BI4Z9C`%zF?#3nOOG^5|0w;P_es&M*fHD2K}vKz#wcfrhf};}U>bc*;j}Tg9)0
zrcQaoE|u+7U|xy$B30Z0-3HlJNhsKbc8O5E66)#nmv4yapzToL%z2Tj<{8-OjRh`C
zw$_euDS`JBgHf<}ZVeu#)DC2}JUWz-3z9NR%C))DLz;z6jX=Q$?Z#)I@lRJ{1@Zp7
zHz`?gHY41J%w=(Cju}!eAB8Sbg404!4g~+8ofips*flHx8R!DmN4W;WrjS%|v;pfR
z=uZO$QNUr(P(h7fiB#PpD|?5A04vddjn{<*w<yI_R|CX5)-LOnZ2I_tQ`-I6S4=v(
zBou-_6<p)j2eu*?A@IJ>UaX*sOhSzLTazC*N$zHYCj@*jIL@(^<sb!tZUkEs$po}~
zo<kaU+&jg8*tioC1i|;Vw~4agoSeL6@doNuy3y0AF+dNC!dHO};?(JXdJLV34xgeW
z2=pI5wSfN7t%ez2>E9e&*wRj_vTV`oPxN5%&Cv4zbxuY=kGUoeG^Cq@Ty%Txpv!s7
zIiUvW`-R(Bm_7VMu=<0))k-$pqNn6>KckqeA2004(E*BnAIT2C_f>n_+7n;V5j1^Y
zcTu6s(Xt@GHzWWI|M_<)aCdvD*1<l^oef_%!N3JUw2>0@7KEC#-gbI-VC?DF#`geK
zOuz?N_E$+)07ECcwfjPMmeyOZH?w}rBys6#o{sb@tTmr%m2Ak@;ow|WNx%9k+?S{o
zQDv&QS3@pgc^R<M1W2`Ji}>glRU5zW^_9Rerv?}sA|@=jfjvuVU})v^7U|+FO+D=C
z{crEYVXM389w~a1r#Ek8w``N4NmqR~;NTczQ3g2J?xtr_Ent{t^xKJKgXaXT<<jR#
z&__IO0;^P_$z_k+1J%ikUj~nVZfSl!4~-9eCk%e}G!grwjZQQ)b)+&~t)yt~*D_hz
z(hYCjP&n9M*ht&Idz6%*xiA2c#H~lEVNuZZyq889vFuscRhzGvB=l!#5L%`jDP$*?
z#jKCpMmjg0LRfIzQIN)I=yJ1mOWg!1LCeBJl@0G#<34h5kAvi8;7#(X{ovayxb?`C
z6u?;Q@1iOjDLUx|24gP5V9;j<9awwIg-<wlN`YYaX|DGhqLE24$}aGTeQ~U#0v#wt
zzNdB91?(!8E(W@zB9;(l`zLJmAK-02G2~e{ywmS<mjI3ap?hEkXeAtP3DD)^4u!g8
zbH)zoVz3WFW4DAF?}h8nFu?@Y&U&eI&=Q<q?KLa_df#3>DS<Pz+m#1cP<_O);yf<k
z1&1|1<yxgPj!3u>3JMoZH^Bs(%+|^ZWP=s6bwAfC`CjiV-6#fY-R$G9Q<{GKxMgXC
zmcX`<WXysKCyI#$T&vinDQ^V%pZJ_ET5HFC*VnHCkBkG$y^&791CN!m>PsTU-~Kbu
z!GcaW_;GhI_E$)-NqPaMY6nb2j-~^Zk*tF9-Y^oLdiNAi<8zf-yjIA{?$AJumr&5H
z0wS%y_4ClJrr-jMc&sSI>&l5ljpqqaJ^kB(g1X%0DDwy&C)<NG0^`<auHWJ{PqO0$
zdylglWg_s;r**BCVQ`B<pGmpuX`jb4U~su!I0$hkt)-lbZ(0XcS^R*O0T_BAa^pOL
zMfog44i5yrKcuQkKrKfbZlC-$a)I9`j%<5RgJaQ)KVTRCe7XbnL2CGFb#s;_>jjM-
zWv42A=n{CaebM7xLQ1q%vaB+CyQeVwE!b1($j{3-4KeJ>w+!RE_E0c1LtGR}XmHon
zfGSY;JCD(`6P6TzI!%Mpn5@n7BJ>aVRe7y(l*#)J6*~9ZT0_jBIZL6M*^;5-QVCA)
z<=GQ=$wBnO5dr@3BA<KsfAYTK{qw_`m|&-e7LMQsM_C#CNUXg_Okf8;QkXxb9)N>u
zuB9q8D^eomwD^+KBL~IzsH2T=HOj)3>%+%|>on-0uIN0(F>-{vd2q4-VKa?bA6C0&
zFn3Hi<ux{bt#%sDZqd)+hproz^!x|*Bw=K8cU=&N<o7Bbc#Gv5d3K$*`t<c&D*zTs
zHB)|MzvMB?LDwCHyJ&m(`!n@W<2xMTKq#dM(I~l=`S-Z4*v7zn<z+{tc3A&kS}uH(
zS)H4suB3*~t+~Op&~$j!?pOM`U)!CChr}<mPjEv}Y^h+5Md%iPOvsmFi&o_DB#vVB
zU9)Fzw1{kr*)2&0+1|dnN$$w7agHC)BSShn?;vOW>TENWOL$nOtM*svi0HWPkpBI>
z&{|`WYN)_x{iMT|-T!O_RdqJfv9i!is!aRIxWw-Av~g5+buWikaZ@_FxmZ?^3l(rg
z%J}tFj+mL`oqsK2w|y#`eq|k-Pgj@)j2oJ>%0iM^cYY;qllvYG=MN^AV`T;~I~Wdv
zOF)Xlu$S`-@4l*E#=hrNCuPq|eg+ni8~=vU4Ph}$Kd3p!U*DnxLT4Gs3SaVWW%ht6
z{tK!@N-#eWV=wKyDkjU5CC-#-HdF6sM-H{LUbs2Q;8G)i;<%d_BwWHA=#WS?=0kKH
zmqElhhb|SVv@d6KxztR_uU@uGE~I3`gZ|3?%U8O|qk=HGZV>X7HIY+MCM@iT3~t<P
zrea;JFNpQY@0@-akag^)O2qMfibS*`-pJ|U!*^B-Lp6Jx`RoCREt!`tFSN?79OWHw
z{j>EeH(rcZAZO(!qo>N`X-C8CP37=(PgD$dJ<s*?q+enT>9dSBpKK;K@-=?DRjVa<
zWPNE_cdvyDpEH@F`*Vnn=3Q9mQVD6ng`fdcF@R`C&^d5V#I%LJQ0m=!>uz>pCqyA*
zI^kL9I62f<YxAG;KYJ?=1L4|QS>2~aMI}&kuR6EydIFCSTCdnZf9<ace@9LeN&Xs0
zYOfHsR8_uCN5B>>-)grOrf4F?Id;l|ek;Dh%(pxZFxKYA!tY-*AkI_MtQp`pZVaN>
zfjrb*AUV3!@U9Dc`zEgGdC#Ik>#@X61`}vCl#e~Qy=GOn-=wbw;(k%uUiMt7^fxri
zNXlJ`aq9gn<!;!TZG?ln+k`@$w)Q>x*FQ2}G&t0%9L|S+g6RAfbTq;}vzNE+nv0*j
zJX{b7-j1$RF8s3O*z#)-JWD?5usw%}03^6CTtkGl`%8p+(&Z#(K}>km7$3oW=Q>mJ
z_nx*}F3yo!+RV@c$>nI@fe7BGg0Dx)`HY1vydEF~a{8{V$^4?)ar<!AR@z76$^|1K
zM{}Dj6zj|3TV8S#dm#{C)}AAX2^{ejxU(YLq*goaLa?r><GRyBLCJ*0>Nu-woQ9MY
z1E{K%f9u<`-HFRE%)MR5v7dIr?KSHnpnCJXo8%zTh~`HjOyS=IZ<O2?RB=YS>4J7K
z=8Yet+PDM919fs7UC(MJpr7VO$#$f^YFvJ|Z}B~E7=~kSPQc3N!s+WPe9x8_wK|2*
ztL0SA?FuE>U13*(GDxy0?mo_V?pGtEO#!Yr`S`_+_}+Es120^QY3?Z4S9{Cp!(eS2
zgEg_$=F_z-9<>uAT+CdwohEMdY!MDh(uodxZ+#TjsqZEGgT(5%LiuMWsb>{*o3T^}
zu4pA(|E?+b^FR?LSl5hCzxN@5pP~@_nD6wAx!NJag<qIo$+y*O!GcHSH?+7(qa>u7
zkpxdV%8OsU$GmJ^x!L1OqsMvKgT$}A<{5~{KrOqCe^e7Xq_Bd|9$;L=GU!hhE{t%i
zX{A1GUqg#nn%h{u*j98PRan*<C4vb_2VrdA%t3~Sfs0z|P)stMiI{Nc3rR{oYXcpc
z!U(NkjYx)kb^+lqpMq*cWQDIhoWyHbP9G5C6$xIgvb`m{8q?LQPp<#Kv0#gNnWQE)
z?!;l};?ij9xXs=-VvcNw@sxL^LP+!Cme|(V+E_P1%sWjaS-re~)fF^yVU<uK!`LPn
zI|;=aDWg{By-UImY$sF}Qj<l(Mb0LmS(Ond3Qmdw;&Dk4Pz&V=DAu;o)-_rc=ckUP
zrl?75y%sbDy@=oDSU;Wl_%2bOF(ae~p<;+vH8$8D?=afzR4+@N)ySqL@GeBf#DR%J
zq;yI`oY*Rt{@zgB-zIOxGpWKn6;V$fnL5`9WK|73E-0E#nViz<3Q7fDG0#2S?X@0h
zRl&WGpIMH7M!!G~t-oNuDBo~cmZ%QM__w6rImLNv>V4WPVmbr+Plvwt!SDAE^`?H%
zjnVbrSqHIBBrm<26tbTKEthA}>oLvkzX-!w64hEvA*U^iM;TFhv7-jU@PG#9YJ{B2
z!_OG0nfJ+*?-0HXAqPgmnSZH@7$tu`t$kFAP+7Sg^q&KV$M|7Y&i}5_xtX=om0@_N
z8CHrPD+c*RrH`yojubbUIx=e}b+qJC)c@1V;c3onC8G~Tik>}|4ho@w;&1c1F$Iyl
z&Is?vcv|KODywAZXv{rgrHVaY>Bw(Qj<c8xQy%z%I-8;X-Sr3jh|Y-Vj>mP0OR5?4
zipjSc%_-OeWXyC=$MVI--i^uRURSo~G+;i@uS!lxzx8Ncc_B`-h>c)uYqA!oub@g>
zadYTa+F<CDIdUYFw;*)(b`>pGa}%q>n$Jw_ZI^u`m0z%v-kS5NoO?gb{iO{APHszI
z2)u)QY?{lE+w4O`(iMz7Q}{MP|1WoetkT*xFTs?-6XFnjWEDd~Dv=eZEmI|@^sz0d
zXJs^m#JY-Bf}Wc{B!jvnC$4aeymHe3GPo4U&J|H|O3PS8SL)VSow7%L3)Ar#{D2Hy
zgJg5BRy!Vy>e*WrrOy9$^86@S(sPt7a}u<D%JF*$)H<)SxHWqhyo;d@R@1Oz!{Ag;
zoKW%GI!9EL^h!AXW~qCnt8dErbLs9F@VaZxwuF3j+?m-^W*TGuLqo4SMyY$Ja_PlZ
zaC{xEX!_`QSo_~r{pJ+yN4d?G<{3_Y$%=7UR&SqLru%V~>Dti|_x{tN;-&r8-3u-q
zIDHj+L)mft^pFmXDdd1!(~mU$b?%v`a$QBPv7|3CdlT&~c^?M4PnFB2FAU&rk3xcL
ztNk~*?H(95JewT<VmnoHu^ltYzt)L4*#5}$@5f<NLksg`^xPOri;QHkA<aoBieed>
z<x0SgKk_tcM;2t5nYQRvVa=1q5QhtnTK932C>1~Vj_fZrP5tM6^@@!rZq3aN13_{J
zXuiGBL(Dee1~WGNzOL2d=JB$}ucHq)U!blNMC@#srTUB3wYRr_%H}VtRzC02HQeHA
zav6s?G$^fzoXO^HXZ7{?n(wI3)vI>v>Lg>QDpjAdz%i~VsJ|uDBN?HtmqPsyT0Yr?
zd0D1heCVLMhzOzB;BEGg@9#B8NQ^GgR)5Z{b`o0H_fE&EQ_7ozy-QP8;9Or|O{_j4
z_L~ttuGLPRQ<M#_vbmHvb5eNb!R&av+#ocu$WX2~s_Hg&a~7g|%9?q@*!QQU83F6e
zkO2|YZt3kn49|TxZzu#jQ|SOwb~OOG>WPsE48w8rQhu2?!%FR6(-ilnSb!2(>)du?
zXv!(O@Ww))FIc#X?w=>SABnBESf%$UE5=!g9pheFY`$#&N<BH~GqU=59mQ4mECDEJ
zdBrOs9R5jhLBwt_6lQQLQ8!x1s?}63PCYOnRf03nZ>C$;2E~6}r~eQ3c3K+*giv@G
zcq+I)uz3yz_qhRIP72kbldj%>rKPuW$o-5K!EL7W&|OLSqr#+ywZIUHN3}r8m<6EA
zM%*OjV`G(Qwq_)OBCo((|BaK?B&7Q30H8Zj#{c^i<O<WJr}WUzw(<%HCwebs`Iro!
z?ZtpD+fl~j)pzlQM*;w-6jjJ00@vAnG*NiHdeY>wAnKn<k<)xveiT5htK`u=c%+fe
zBdB0jeVxQyI1Tkt7Z;X()*bR58?K@`$O|Jdv^8EAUDr?G(bb3F*7>QW5oeIHVee*d
zSrC=HsF5GFZ{+My2?)Q$U^9{v@mdtXragr7KF}|6I#=Ii7{WKcxWWiWG5yo$H6XC(
zp>{?ADCYfN4nu5{+7<Lyc+g(%|4{1w!}O)fT3-a(2tXnK2Rry55+l=uL+r4?S8$tV
zb6z&8-e#l|$_F45#p;B*^5YU4cp8$8Nkjp9B7iT}G}2DWMtxMle|ab>VtMt+B0L%3
z4xeXtJ_lIB1RgLi=daj%9Qt?ct<(M7EMfV2XbS*Y7bgs`(*8Rk{DHwuoc&%c?X=a7
zWw{d&K(XDQ-;V97xy4(TQ75@D&@E<<aLo_D%{Cd1T^qVQ(1WiZtBt<?ikaJ6dsb7=
zqGej)RFT}pEBvu6T%<jbyC)fIn{3T7N_v~ioTc2YJP}r8gaR^yiGd6GFuvZKeL4a*
zTpl)@X8VfsSQpNvSug5MnMF&dL;AgOQ_wr)>gN%Ti1W&+shT5ePMO`3FgG)|O~tLY
zSwGsM<B8TqOjDTA#XKrn$jYonOBA#o#28mdZ7<!`9n19YDl^dyMUQ|%n?5HuC!1ex
zaocYfn!oL*&K>yP`r~|Bp*t;H^;8if!kog<601llAQ%Q0BD6FXj?YW1gM2<T6rH9M
z#v&Wd_Wo6HM6mmaVP^d~x{HpPn%c*DfYy!WO%=TldxeNl9}TTcwXyAt(74Jt*~Zp7
z;NJH6H*q-6g&wf`uGnub4f$ual82<U`zFcC@Q)zNwEi*gnA4ip{Y-4xwX!tYSO;Y-
zXYA^^oOfR)rS9zs8|^6GFfGG}SFr5zf3{sZP(0io9!u+qTXG!UH2T?Ee>Z|+UD;9x
z#aeD1{|Y9|Z_dh7%xt~eB$!OoJ&k2JFFj6+p_4xzxxeXo61NCY!{uA<w;{{&g!Xp%
zIl4EcFXkLTZIx#+2d@QcmTt;yd|}f>Eid&b_%k-Sbt3BoebT2F^?`mpu|P$<HevCm
ze`?X$6E27}dx&Cbey(f6>Wvb{gFLw^cTYDP2AZ#J|AwDOH@-2t)ZJbBvJ_3>Gc{-p
z(am(u@D`|XUTb_xKJ#{Sn9md6x~MP{RqeaHG{rHyy=Lwkb<nhsIjx&JFsb;T)(P3y
z023xuLHfVy+jWai%<LFZMjgm1XUNz1bxo_<+x#(3M^&+FG&fdGS3F+sJ^$G>-crF*
zSWkHBMzPbPe7wl~(_gFySr^IV&e6B0M<;)goGb#7aJ`7|04?boDRfKH`%0JlMwM(_
zbthC6{|Y9;UrfU}LM5E#>qhAAZ*BR9n>r4jbQDZ(;?K>eeNc6QkmHeOTJ)^jke4=e
zzMF4oY-C{bRu63WzsY&jqV4JoLSa@9J(3AMfechV+Ex7DUxzE4m)}2ev;RrXQ25%O
zmfZg*zlu34B=glIHm@m8qE2&yyka~BvRz_iAGsLE!UPbFFW!v)wFgsQx{L1C6~a#y
z9hg%*#+L2gd}<Z1l+?jAr9S#c<KzBp$6wsVpKn+^wF!yXlk~vJgY#o?IW48Xmvzz{
zqRyIs?19Sv<{Texic_cCxVmhF^;i2K<y$9Y8CCp|VdGjA7>$Z*YYFAsT>^jF6g(YW
zea&2lcIfygQXr??l@-Q$c2YiJL94DRW2MS+-det9Q=h3>Y%WUl3}`Do$Yd<`AUCYT
z-zIPSiv2$sitn*L55%SCTcU|iDhxvp8mc6*Ed}#U?<U`aeX8?ily@r|sKgoZP<;Dw
zW_s`jCzUl=`HFQeN<q)R4$=%2W>jIC;QIe17ofaAjFLQ6?lXWiPo$x$qf)H=B<TMD
D&Tp8S

diff --git a/vignettes/examples/timeseries/timeseries_anomaly_detection/unnamed-chunk-13-2.png b/vignettes/examples/timeseries/timeseries_anomaly_detection/unnamed-chunk-13-2.png
index a1a3c90e06af8e8c2af7dfea244d053e2570e08d..f5bcc02546f19dec19bf5b5c3889a0977e6e5aa7 100644
GIT binary patch
literal 13529
zcmc(`2UJsC*De}g8x{l$BK<`vBE3sj^abfi?;yRG&}&qhf`EoDRgvCX=!u9@0t6wn
z&_Zt^gceF5;cooCbH;ba`R};njB&~y5Ed&Vd#^p$GoLx<Gk4x=YbsM;WxNUkfv8np
zJl6q%{(1)jovWhy3z*4YKVJp>x#Ic4*c$}8cI)i-+|YE`2?+EzNagucy?{*owEq+R
zrG>8T6|>Ey8at|agFBLsJ=GqekjyVoeZwiD{t{w_=>{o87svE{etrk%dE}=DhMJj0
z@6OXd``Et}zkA$qQ$*mxOz+c~O2polXZDJ$f5_v3z-(_0<CKdakoF~gFG*k#C$UzS
zKu?}h*@HlLBK{8#JJDU9i_^`FDoesXy2~9Jr1Oc+C7c;G9WdK<8jXE@H5mW0x!7ln
zvb}1^$@Jl(I_4@qbYA_Xi!!Wi(Y3T(-mihO%|^G~m^99RG5F~lXt<S<(*r5JqRFC{
z9VSb4Ix~?zrcs8xrm%T6-rMD4RotBcbEygRY;ex8O$zu}H39N7Ho%Ot+4MD3(R_*;
zRHlwspd{oi#J13-*s8umINVy7COc@*R5<BQm7m*U^XmLCw7mUkpw7FO+s8#`8r_4c
zcI}rquQ0)AN83ytE*qZ8-Fx3=!K$*ecHmwlEeU5T&pcid6&;=x9<9|IhCY2aEI0LK
z4rR%9KwY%)a;CDY&#v}u9kgKXgHSkax{R`3)u+0rIpibsDvB7soGE<uXl5|d!>`B`
zWn5bAU3%*`u8sI9U6E><O)&MPiXRJC$h~=-tqo46o|u8ilx_g~Skcj69#hpg7JM8c
zm|DFsmGNy-sQZB@#V+qRTZRd?n7}xVKz!`=V(=h*mMp4IXbH~?9(ip+D3i|dLpVNd
z;WWG`mnj?cfM7tkLG0g6M&fh(+%Ct&UYB*-J8jif=I8J=#=b6~xiHLAm>hw$S+zG`
zZdVa^izqmnx^_uwfp7)xtg<wkJK6h;B(K~56uOZC3E|r9Y^O{73(xuG%DpBZ9443x
zJ<abi`uhcRA}_3qURrWXPYv3<{+qg*%Qyv{&`0I@k)vLDTB1C}ydK7=QbaDy8Sbrd
z@7mgV^h)=OiJhs6BV)g`+lwi&vahN-O(Qtwp+eK!9&sE}ew5ATm#A@}?tgA6>NFjS
zd6;9>9$Zmzb8#`Oc#10+$<$LE8qNt)KRpy)#&9geh4VW2@)q8bkSDKHPO;gY9Qc30
zB=9b*?s4ry<AU>M$D(N*bq5V<c<5_3qN;PJp@ygQTb?@`ak3`Ksbzvi>4uLE!jU*R
z?csyJwJ%%{&6Yn1m!ijnV*&!hm`uNb39#AcWtyYWr~8}Tb9o4oj7o~=C}NTCO3S7K
zcXy3zD@i|CMg{DA8rZ0rvec(`I@a?KwC+ugr3CU|bkE;VWx~Gp$B>;A{%4=PALgxb
zjmzVp(8FoHtQp6~t={oB<jeiz%r?-aK3!2Jmk5ny;TBOgj`RutlNmi@)VK_)OG%cB
z2c1P|apZhOf8w%Z6YW@`0$!%P7k7;)Sj+T|+A!Z8_sVy@XJ)|J6m`Pr0JW1^m|c{1
z{^Bl#kEpPkh8DhM(X?p$WLUZvsGmx_*xzEh6I8xCPQhY*zu)fAmdG}M9)HxH?vQMz
zY$Z)d)tcxFS3Q%vEER@s<eIT7txs=Nn-;P?DRs-Nr9<aP-7QEI??2un*3&6EI`6Gb
z4jG_ieK+c6Xek5R9Er{D2eTwU#qne6RrEts@pUD^Hlm`N|7Fv9WWi{zl&FPMzNO?>
zm);<K6O?Q*y<!R*wa(=5PNj!n#Kd&o2h}2^r0Zh&Eof<UiwV^$k)DfgI&WmU1Bk=v
z&l5vD@ZW15sX#YOLK)p@J9}3IR<DinA9WKvcUI$MP4jyjlxnA;V=h@;`=@?o-*5j~
z6il+8SgzgA!PSHj36PU&ugNdJ*(jL91xY5z<EB?>hVJ>CYwfb`$sAYMZ4*^eh(T?9
z5pn%wDFdiU9!g@&w8p;*Qq>c%hTH7#+#D5y;PEeWG%Nj_oeK~4?IsaP$B>;QwZxs%
z{Aj&LDpVOUZT65M?8+dsaK=@Bu?gqXeAB+V{;9de5vD-cIzpd3OM8k>k+sB`HDIS7
zZy=a?w>SV!(%~bn4ms7*<7)!LW@6e>t=%_14{O^r`#YYdM|(qT^V%O?`%;YkY00~m
zP-cAl3%%lMWtm0jL){@U$Dt>A`{kC(ybLHe%$+av_jl);eYtmJ?UEP#Jf?CUPFbXp
zmKIZ6jyNsjAD%RuifTOi?txVh$S9#uBa75|3SwA}_i<yD1ETJdTFi1`oERm;DtDgG
ztw9e44Cod)eyBGr5>2?5;cZ3Hm+zc&=zI994kjrnm=>(aHwlBq=BQbZkM3Sdo)mo=
zC$0$g@q<3!`}9lquHKQSYji3XjY-vM7lP}O#N<T54?J1s2UIWlGD5DgAdVx5njo_<
zaEgain%DOG?iJpMC|o&OsaVvjhGrn|w1M?fzvaUm{5Rotx9;m27G?Mrl>iRUt{!+g
zv(Po-w3tC{fV#pm4WG=0x!;@|Bed9Kz1oF5f<xPz^Yi>~QmO`OC7E{|^`EYbp!p}z
zO>9Yt`1vUqJ>~oBue|Wy$=8vxX~|i;jNb9an3_%J)jo<_NE@@bgmZKaeBioql>&;R
zw<goi_3420rXOP}Y`25DhR@--d*)0-ptz(=+-SXXEq!5m!V5?vp>E4APiB;VWl^tq
z`WxGfhs*KdqTpn1C0qXoiJ7S#_pu4PtmzVQoFw<&B;1bU%Ogo!n;kMnXM#Q~Ce;q(
zY42wfSGAoL@<Ol|+&MvshgRX5_;|Wr0XHm~h^S#prI3~x1SN_W%j*3rhZ>K+S^j{k
z*PO6517t6o`Dd?|=Bnw3fz9&O4rJxiR_8D_80sg!zQwh${s#Z58tR`|!T-o5g3{eW
zDPK{7k>-DFdW|${UdR=?hrATCNGhpjw>0;5mfuNK8CdlTY2Nxec$EIpJz2#L6!~-P
zvi2Q6H-)R@z#qFK9w)N)-*f%P^2JzpjtcZ`LUKq<uNVjV++BH>I$@Zum7GH~N0}Lm
zb{TMHATxPv^M5Tr`R^Bg?v`xw;eG19J0|RKia2hYWxHZB)1Y#^(tIIl%p@NS0;%5U
z%@OWB*bn<%p*g$y>$u|EiyHzclZENpFS{g(9mv`WRogF@%b?unpKOg`ju#8w9pRg!
z5w-p=B?rnN7i^Wn*mh3l=D5Q)>^%n8&w(Dk<O}O`ZDvZ;uKt2hhQ+v+itX1Tm{Q#_
z2q33?_=+zq&Ujkr+t~ESusrp|OLe=Y%WGwyL~nU0GoO0*OeK$S6TFi7%oRW&&iB8f
zYa(f*WFV(oJ#Q@=$ljD7*BA@C6@&JMmo2k5bvub$zEwWcAW+i7QL1Gqr?@zAwmtml
zZmCYbYPm^$*>6_Rlj0xX;tiPF<o;S=&S<IOKuC3|Vddsj6Ixp1FHjO|3XT2I%*>kL
zo@tUc;#n?4T#(*DmvnaT@KAr9UM)$oSYtuwJoRCMt?`YkR~a9B=BA5qB|4oPZX3AV
z0_E;i1qfvboE}ovn_~wSVP@+K3ddDuS3$4&j07%%UcWsXgDzH|h}*X;i^5I}wi7Q}
zemLByUeaN6XBo@8HSinAoOxaPkOsk*;}2y$e?m^m>s`hK%o>FzI}Z;JXJ==NbqnQ-
zVdp@(50T~qbhpk9sM|t`e(C)DJl{BwpU2Q?rTzur?LV*hs_700boB?A2S|>eH}&QN
zmX9o}HG70N7$5XLkY$8b5bm}RAPR0RkM!N?PKLaFGw=0W0+zxnYrc+Z9-D+SI_vg}
za>8{g_l21RN7pbvwi8sI#!@V5Jv5n=V}S(xHMC>C-L<vVmGVZQRFGn+ywWHxD9(%{
zhaL{mXGSuUy*+#Xktu+!ty!(z1=4n~QCO4SUU|z1BS}h4amR7AI@GE+C|;3@y=km$
zjg~F7^K7*>fI&-fbz2H;p8LoLpSDRQm6YZk4Yp+3f}~g%I1L?<DQWJ$C<tV1B%8*W
zAa6<8vp(^!q&`|-B+GaruORkz+)3@zhIcicDyn^mXV@@EOD^BpJlx!gC&-f!ixVqj
ztLHT1N7HF<fS#!6Ft0ttND0$3fnHB};`smf)Dpu6C0$8DyeB?~|IG!*2C!Y`Hu)uh
zzettQ_C*!_F7D_I2{tl>+v^KD#PN`4@ZG#T$>9$CN6Y=4DCiOs2=wIrt8J^4JP4D_
zvNn!wAlk2}D70}fDZ~rE2!AH;9OC1CO*8WxDB$P(SQ_<B?hag=tn?~Y@2+A|rVEjE
zWO8!~^BgloJP*=_0=66RHI<Y<o$gC6oEKqpOGJCH9)_V*b)VceRH``#>i#kR3<C|`
zdYf~<>jTa#(6JymP|G1M^V2<+EQ`(bF?kdD{7?1E+2=soB1Y^C`q^&Wm5=3x`S+a{
ziQ{+Lkf`#v30nWqSMZ+)eGDH-QOn21#=U<Zr!lJHC!HKyn0oCDlHG5X0Qf$nIvZ=#
zrl<iJ|C{eWp#0FkX?UXEg<ilc&CSNg$EQ#;i*Mu%@EIeT#`-t1<*8*$J1xf%3LD@1
z$D_(kM!}M%fiM0Xu((o#@|f7zS}Re&#0GZ}on=^=>ZgSlZ*s9O{tOkJ7;}>f82!p7
z&ot6r+5fZOUp=LQ1%ULg`F6e!&2?gJeAmz;IR*TF$6MpS9>|=Y@bZPYb;%{l%cZ5C
z5&#p%7|Exxi6~*;xeV7D`kJyw&}yEW4*$00d3GBglf~a+?gYD{JF}FjM{rY9ZIa8B
zE#C=D)@6}Weg8ksX>1g-ld2K7=@EJ21;#ial$9*t;Dw=v+rg-^1`oxKSuB&lI7Tz$
z^afy}fB0DYN*<YW_u)xf5Uw_hqR67-^%&jS3_Y*K5C+ieE~6F3+D%lkIg=zFk?A<(
zsHa%eio2|*N<HZ}>l~$D0`T{@st}>xYSYdJ*WZDCtqC)Z`t_QKeEQ9SpAyWcLF1QA
z=hIIvf}Y&HOJk3&U)miVI~Lg3!(K)yZHDmMPs|2#8*>0}ZvrdnG<_-T8$OY&dxbso
zuNe!G{?P581c6n4bHIH#p-VXe`e>}A>zZMUhr?C&h!Rq%C89D*!4o-}A+1ymU<(Bp
zIgt8rBQa=_bQ~ad`mpcsh~`e5g5Wjz1Emg1@9lR$P0#8O8ErI9g70;Fg6|T3-kTZ*
zm?GIt#uv9VHy52ST|SF?`-d+^(fMKmGjwdn5RCL2^W8_+K~J3Q+Jf!QgI<T+>^s9(
z`*nbcY5&&@-%Y3z({wx{b8aA5Gs+S-%zBSWN5_pWWqRVk%}3L{#n948`;Mo1(BP0`
zW$R02!ui-1mO##Y@KzZowiPpOToZC0Sd=heNARA9O&7B>dEw~y4Z3<7$<I*1imWOB
zZsoWHlw13IfaDU4P8e(1=WPSpQk5^0Gmwf@$=>BNc|ns|G8qp|o5O(ceJl2IZL>SB
za|Mynw4-OXoaY86^`FNM7&~2coJYQpIAh9?ho)c}i!I&7G_NdPntmTa!k>_vmxu0^
z+nrGEEflL6odPbWdWn}K4_S*}r0e1?_+=5-%9`=j5Ywc;wOzdWquve#0#{9di}!-7
zg@voDd!&0b3&Goa;6HGd`d6g;&$tuupWeMBkV@JeP#CK)&v3J`wY5bTXz>5bm%*uO
z;vO?o9UTuQJMohZlNgL~Su()GqQFPMzP{8{RNb)couytcFR#1qLLktus6Xd@;a*##
zE9g$f!qk*rH8#ImSCb1yNcc}uKl{RKbr$_DrankplIgwVf3I!FI!4&<SH##D#%FbX
zd!_r60#<~s=C3nX)%KK%sv&myvq0=O*M#Fl4MJNHgQnawLFnm~Tx^t3N7YqWWbx~>
zOWrYyw@pGePW+mRL3eFfHZ|CfNzFic){(x61{srr^w-bAaBeR6LT$k>oQG4(rKHop
zJBp7_;l)8tuPOy<Z(nkAG%i9fMf><`;uUV?m(!1beVTq*{Ij*DdxEFiSTkd64hR&$
zv0_xyy!r0;HTE(0^wyGK8PzLK3FR_#*l*BhUtB)gG^nrc>{O=vI=_ou@p)?LO0siw
zGiATtg<#z`{{9L7;E`uKzzEUcHbWS4bTnYaXw?Ifsb1w(ZcwT?r)y00Upu$x?kNCZ
z)%?tMKJ4_7W`+rWG>Wp7mCjTd2{{NUH_-L_Nf%_CxNV@PKAr2gq^C$#__`rg-P=-j
zyX6|`YaMXOLW6iKiEQ$I7BNf>m4%ffGnjt4k_LDA2j@lz^&d6rzB(*U4(&(@t3UQD
zpj9j?L6C^Y6G!I@UE7Lnt^n}EEK%7%Tg;z~%fYSM?amHWE>3*aU1)sQ{bR=Z`3~b2
zhMNc@9&9^OUZy>z@{JpVSlkWSjEf?ta?|#MK(YeLLM|ttN8}F|4qmAo5FXyg37qbu
zvtuZ}*AVmLqX+qH@l)^Mb?(nu>S+~U*iC9q2TASxjQ6c=ZWi6W2Bd7@#qtso6E@Vx
z`N>MjuIuv{=`xC>|4-3cL+XQi9uql6dO}PB8R44sz2cuqP7p{d$4oP#zTSc~=hz)R
z_iDu>zeLViVy|@G?5Ex(0Dy_&Y1B?C#iZ!5Q2e7|C$pzi?CL!4-$OeGf4!M>_zPs^
zWISZh;G~CV$BBY@?Z$A0rE0-<=U|y2AQ0AEsD-TzO<qxptpsT9?}P2Ny`pea7_%%}
z;+k1|gV*ocHzO?Ra@3&Lbc?OKscWPX_PCR`3RmSqWj5`{I|Ip#;*72HCa>7Ku7g07
z@XP@l`Msq&h;)VYu?9Vj3Kjd}p}Eau{NcJ)_lbmT(M>?g?e*#`LwLy*tE4)}vc{XM
zZI-u)iE~aZ2$+LFMi%i_FqiF+-SFLl3Pcj$HTK1%cM9@{i$1B8Oo8(ttBZgiQKzEJ
zHLgwGGdt$JEGgOYAtFL7uRy)YkW8Ufbj&&3-}}0`|LeCpfU{`kc2}mu&6K<aFk55F
zhTIE)T=*WYfO%vZM4G)PX_UF_kO}jc0+Zu6_>$$!c4`nPy9oG9%t?Xr_p(NVEXGu{
zyh&@BlL12EjNX4OnF;tV=kN-Ku<fe}uFN|EQ8V8ZwIY1nP~pEYezZ962$&(?9GEd&
zldp0`?~6k`rgGnQbd1B?B;Clvu(GtjK;M6WZ`ctsB#AH3{`4Olmxu&$`wXye_V)+r
zcF%%ToeRGCf@>De4S+z4Tn<9YSYba#al7xeFs-s1=C-SaqVnc2;)~~I%`fjlJDZ@E
z#~(kjg$R`uGtqPbI~dD6mWH3C4%*+Kua(Gb^^+leN_%ep#pT5`*{`Aca%qB5D*+tR
z+-}>a-sH{Y<L=)94I|Sl?YII9uR(fsW(FQ6^xy0fwZOGL>{V*XR#4*G{KBt(k)-=Y
zrbKf_I1h4knY&Bk#-{#>41yBbRex7>pX2($_UyMmbO6AqYrbgwmsZ7J8COS;mNRX;
z>%X6l)LsfkHkn7GAJr{L0rzE_yo<7wx+f@*Q(Z{e8)Kr&b=52>d8WWnU+$l{o+IHt
znESb~Rtfc@mBnj5=bZ={aoUz={>twP=({ej!}#Z~qgwZNW}0`eq|<D67i+2Phq<im
zDH_g=+fH>}JFj)xw$gostL6zx&N4lgUqCobmX(M$+&T}M^^(dF@ZFenlEL+ncUy*O
z!pD3k8gobTJvze-U%fypW(g>KCGOzH3!loI(YBRAe4@cZdC||;E6t-7#E>!RvC4v^
z(h=YFhX*-#S%>N2XFC!dJ~^@tuJiWs^7Zwr@-M#i+vX%e`8F65MF)Vv5I}VC`hu63
z&t$#WakDQm*w0)XxmH}PqQVn-<K>^|U?rtgG(vf>B8xpOpHn-Lyv}HSdwA%;bdp0O
zhpzl;u!82lgNPtYX=~20?Exo{3EIFiLj^W!Dip&>OZaQ1=l+ymjZ@1$`4&^n%c^lt
z%}bTtNFgMs_kQ_Irf{)Kz9NcZ{qLCN{wHRoeUi5$zUJcPYK5cL$PKa*ADGwbJ;+s8
zBh9Aux|eiby?PZ*!-j%wuMXxVKhl3>CIh6<@JyQO$8g-CsJB*vQ)c!00y%C*dV4fu
zQ~okpB7z}G|KSfC{yK{=mg7U;JBr<;Z{xwx76*828VfV?)YMdsn-3T~<QQ<m0h%pN
zevp<gQsF1uFux~!Y`9sD@r`e=jcEo?f8!qx1*-BNo<Vb4Ic@g+IP<4=A13xFALlGL
z-wR{L1L|4fv(ZiL;=4D3Ezb_DE0djZd~7VQt0}O?X6BXh&vz>k`DDJ_kaEd*U)EEd
z5w&Z<^;ucJPxDhE<;+tSP#Y5vDgC?PsgmtVkBXzCo(M5>=i%nrcRkM#%~pKYyFiiC
zyqGzZUm<gUPqDk+l&f-mCSFJJcQ8)#P_lSYd-3yi`uJ=&A?uac`YcX%+0XH5Qwted
zH>NU-GR`8oEEJb)k=$lTWSohA#Lm$$!5^E6tl6}1)hQ?mBAXO9Y|1u<r)35#6u>`}
zvJ?Xu<=wlG_=v@qg~j&d?<TAjVFZH0(Xa7JL59e=zWH-c7f|~l^@(};rDr-xk4TbR
zbEvI@4>7mqW=rC_Uzy?Zm8{+pvT~)om<g4X=o+fNyZiBF!DQMRwcZ0s$Tau&qn_b&
zpzrAWYC=pu?zRb!;W`X1+ba`~e4f;otqe&loo^P~78t_R+v0R)8eSe9qsz#o_+4Hk
z3P>1K?XK8Q?%%d|%c!>wey)yO|K~l*cp%N~Ja*0KC(sxJ9ra4{PFBqHT9)1OTpx=R
zW>NAB0t;ud-eyS79avh^gXa0isoDjIoh6)$J3hs4@*=w%y-8bx*9I9(w6F#I<lag8
zxwp03oue2s*U2qEE!^_j&bYJV<oBj%!?yYUpRPc@j%<oIhBrV^v7sy3JPQQF5uSz4
z;;I&6L7By|1=E<~rJ^{(Y<@JOo;%U{#|k~AQmz4g0rdR?!0NIG#70iLmF!j$m|BKC
z;<x#cuj9%#c2D{+jQoQ}+Jzi^{rbepk0xOdXnCm->iVvcmgP_f%Sgl4T~tn)&xKXb
z-Hf`gpXV;U*0rs1^=R=05Mtt(+spRL4Z{dp2(uT#Tc>-my{Vf5h6AbE7eU{5fLO0F
z{n9dy@ppr9a$r5q(C<a8v_E?-rcOU^T_lhZ1nTV#ywBHEmsYnn^zoW<g5`%1C%MoJ
zv2tA)m6r8ccu(!3s&~Uol(!7S+;xG5r2p=j3iGEQ$&v77gF9eF5CFVY3SyTYNce?t
zQ%Bc;B=TMICQxR9uD+FloYlL3CNBbgwX;ISpK^AF8Z(>Ga+cY$o~+&1_097x4*;%G
zN)fep8({P3<WH(S`5kbu3O$(lT-$|Tl0U~-DhVV3L)<uR*hG7NUk9XD#UU(99*p!@
z8|a@3TCQJa?IQ!R?JBsXGc#!WAVEpXD1>{f4XYU+)|n9rxPy@zY`DL%`6#iWGc0rv
zazfL^<yUbMVQ+(T-!P?g?370jm?xZ-?iEX+85fAG`_2O-yT*QQptJy>XLC68w5^Ns
zY0z(IEH-$!DnTlgZ+R|3&t07H=+2HkrW5dP-?T#9zB9rwyK?#86_MA*fYboFF1ROK
zP`c^US^5SQPpiz57BnpT+Sck!!j-ZL=7;*VD_HvQ4<FUe4G+X_#!a;nrkjnmP{A}|
z^&Lga*BZEd8MV0vEH8p`hkKwYFmuG?KuRpP`I~Rm`VIy83^j3|dr8-`v`Yh-#el^c
zhuEiTw6Wg~ne=lBgn1DB!bW7X#KB7Q_;(AFh8~)XRY7lR|K-EgCrK{)OXm=*9<<FN
zBQOYz<-NvW>BiKfLy4fko-gulkDT}|0M6x%(H~VTo;nHou#bBcR|0I}VA-v{GZ7=#
zijVDIgYQ78_h)974{ouBUD8STQzbS1sZy?fom$dqaG#d;-!y~8IiNCmbfelmfM}`)
zsCM@sJn-Kb|Kw&95YU*DgT3u?4**Uc@PRjf4va*FN&O=M0o2zxzykNG|7eZ>V+HmO
zsIY)S=PHl?S%}G}r4I!DqkYTpN0`)xv2WBmgexdy2&^2wcblE+D7DzOX9uW&##laW
zgqquJW{JEbN^o3u%@~z+35Di0aQ`7hgY*+S)AE6j4hmSECIkEgqs#DR(38lS78?NM
zkh5n2RP2(i?kXN{{OB;(`z8Xot}#zqq-|bxwT?a@%bE545y}5gweO$1<p-%3H~i3w
zw)}5UCZB{(i5rwKo++7E+g;yX&60NQYl!w;axox+N7IG~)A6lP9Y7|6dd#?P@EOC;
z0n5Gz$RkpuhH9G|^_1-p$LN@eXeNyw`Z3K-nZ>+L<rUcMPaxe=22^egt(QpaOmAK|
zMZ=Cg0@-9hc-e+WWoabw|6AM+H#TqZUD(C;n*}}+*d3IRA}_xotdxme6!Ws<0Q5*c
zKKUg6rptndJm9`YRC2U^bLxT5Xg9<A!NIkq>_B|pk}jjM%O<dooZ)oPmrEGGotTB&
zXzNNPi-doJ^-#}eZUQj9l#yGL9;t0SEp7a)2T-Wn6cE((N2g}~mud`1ZAliDS}H0%
zFkx}=5nB=Cih5ul1!FJol9zO1!KMO4%(27mrMq^!j~!-e?0V0B+&;v_T)lq1PR|t3
z*xUib_)d?;4M1eQ?P?j?$h{LYSgt~>>>)Zcx#0DZeTo_pNLG{+SpPe0!Q}vK1O5}X
z*7aTTCCsv+j2Fbpmr_{i{H(BZvD=qiMt~691E`;)gskyg^joS#TJSua0gnU9A(ug#
zoVfMbG5h&tnZC+PpjqJ*1sjR=p`pDRK6OLLXV^Ht!-hU#<7Bt8YO9{*K%KN_WlurS
zfS&MJ*MD(+^#Oi!tjM8q{lLLq=AsV`^hZZvGvFr1#t+gos=FlIrUHkx+fe<}!8V&T
zmW5f<fq&PU_|m)oODfArcGJx}Hf4uon(bo<pFX8lPBre~e`br~@EN~_4iBQP`N7c3
z2te>ry$R?vC8}dryj$c?a#Skmo!Qsn&GW+>T>8vXmS@X0fOkZz=G|1JB9P9FmeT{M
zY!6CA0J1Oqf2og@oyaa*@6QAi@Z;4Tot>TP=`#MC*Y62^cLo|Ff0qu?^}T%gk_~d=
z)dXJZOG;OOz(*#|gPus_^QN5xX|;9#IYMH^UB`c;(deuAqe@_T-*xTwH!zQqOVrav
z?2%^h5UGa`9j6+-3|#cv0S)eTp#~z1{7A3Js^GgNgls2tuWU5?G{P8j7WDpWb*{GN
z+8QcL1`9cDI}&lO59<EN{)i4j2KX<=O5l28Pf~h%IuvRRABS?PDJm-VRpy@ux)*Po
zKFZ3<Hv6nu!_^%fi+mZ+g94m|kU&4?U;5uW@o-~_Z`|Bm1-AFfjer2g86*DIn4W<&
zH|yuy%8X|=H#5&-*)T7;uy^ltVekUxYVsesJZ#5seY2q@E^hUBa6Xjxq%=^<MSbnS
zU72}uzvDv~S+v1}VQ%*hO+kiwLGZz5x;mGr-3=CLF$g6@1=>eIc6AkM<;XAhZeVJa
zE5|Yf6SK2fX=$w}Gn>F|eEIT4KtN#G-<M1xMo}|;7QD^IW?5^mVQ6S5YUd+8{r)z0
zYinyC;gOu2oRHA(kaV?FfnQx+r6bETBTi0EhLsllmhjdi^XW#fj)8%Jko56!BM1Z%
z92_ia2Te@u=;)A>mF**_R*p5eVunJ}UB)V;q@>a<L;a8U@z0`fNXf`#V3T`nk^bw}
zIyz7U0&%o10G!cnQeu}yAVRxXk44%?$Z~nOST{aCep!JExF{SBpP8AlN3Ie{mSIPn
zSXMxF?>;zB$G{MX8|B}vtmNxOp-@L~KPCy+?!G=TOD<7S(Z`P)t$)QV_nI|&whs;t
zo>WKx#OmVW0#~=Tw$|4E*wWIHE+U;JATKYErRm+AZrR0ewfGZYPUy~%-bzaZ-1^tm
zT}yh!cNvl%%g2#{_=);+=P#uR=Gj8|O!r&$A)Y#U>Eiyb3>K8_Y%%7X32UV|1<C8g
z?!0KjauZ??caH638WbwxI$qU)4Qug?OP3ENb;dFEL&i8bIPP7K3Sk>MJ1Njx12%z<
zE}7u35fs#%m6dgr5Zz<j97Ya;_g37xeS4QbF(suyr+8yyqt7rkGc%J=0fj<$s0qWv
zdi5sdBhLLv!i;gn#l=UPk2pDh_4M#yMNG;K@&?m*Y=rI&q0zf!R-PBWJv}}9`}@l_
zJb-OPMQi_VK{luW2Vi6g2?@ZqKGNas9|`pSPiGzh(?=ARkDorh5=m!fW{$Bix3>No
z8><FqyUYF@u9BXVWZO@prmCt6C(OKUg#$a}Tt5+#%XmGiu1>5p)Mzg#R7ojHT!ByV
z9e4?uj@8%G(_2C&_op^_EkvcJrXHb&aYR6q4c`7m-%ltp0vbdbWAM$PH3B<?;-O6}
z!=iW!<*#khZ|}d4%<6B$0gP`4<TZ-lSS^1e4*G2D7#ke5`_aY^95`m;A8%_tPp4o{
z;knBBg#l-=?6mJ{J6{BCa~c{#kOKclAJ=NqUO(#!p#HEr<e$|l7u!{M0Vh`!Iu9hr
z7l1V6e;lSD3?)=Qudc2FZC!(6o%}{7ofxxbpPsR?hDOtuF~8>w{1!Un#9c;*0n7kz
ziqQ6Tqa4K~+$MaEc9wokf7Ias*a^V?ujv?_h#L(x!Zw|r10UYnJOr9-7uf-t`*(Kq
z{~CVEQq9*X4zMvwHt8o(m459g|KO2)8qp<xpB~sp0PO!hn~3xjQb$Jz07Xm-9p|M#
z>?Q;Q+NJ;h9f<M1G~v9)$=CB7+))IhHf!{tqgy^uqBc>!q%c5uEgs7_mrd6dCb3Jn
z?jlDZ+3*#!kuJ0*av!}{8f-wWai09d%djaPGUXPVAF0Jq%Ob#?h25Wwp@u~MGik72
z{_*%S==9U(vI3;c=y>ecb()+KcNhjcPjhWLRzp|H*{y7{?$+CbSH)c1_|LzkSF_3`
zzn&h|DffPHAuNI$7>@RL*v%DAls(9!a&hYZqu1RiKU$wPc(_6kHnTs?RT78+%RPZv
zrq@k_BzVNZ{b{58+#9Nk`wIWq!aH>L)!orc;dn`%+X)al=HC$Gq~gYjqdg%dzV(ri
z??2FkpGuWsYho*n0u71<lcmuTuA;bymb`%<Uk0VS3#_jXuC+YO<IA`e1}RoV(+n`y
zP|W=Omg*$4%a_V3MC|?Ty~KBltsfu1(j`zz7c_te)T)bIFaojepDE;G_pWzHVj(l?
zqn)vY8L2MWB_W6Kd1r#|>W`4<!g4mfir%l2zyo|5+lW0MmxB>;n`&_)&zDIVm{Pwm
zBp+E*rf~eS7JMF})%UV901s)`XJ&EMoEjKY-F$2#&@ZCJdVLo^^`%vns;53Ql`HiI
z#aj?}B@)>LP88@+@+n1nq7&J<S*3gcZS`+M6)OFs_$-%5mdDvggaqmvzTwKzVnaZy
zOarsC^zDNW>R}8=&W%*Y_(SYewgr?{@mryTBQnXW+h&UHB0xijFH~_Te&z_;brY8G
zsRXvisdx?7cZXq%k`{SUql+$_S{J-|xC%Re76c~kww`i%?(Ko!#lkau`LxWN^eaEk
zWZ_!2-H5hno1H(M<k~Vh*dw%)iH|Z^Z)fZfaujcrcZNu{rk4X6k4x|695T9@P@C-&
z`f?FH=Qm>F?`Edi;IHW!t)$^lDu3AMVN7aq{NOd{KaC8Wf%v5&Ycr3hSI4_5Xl%2|
z;GbIc3hC%{4?0n*<spr`#tJMoeBN>WUu}(F7zdf%t>Fny=snQnW4Jl)6A-YqliyLY
zb&mh@Up|e$M<LtzVl-D>s|aMDWEujBd4bwzL)#>%ARo(n{G712>~mgG<jd$-f!|4F
zZfj-U?3913q8S^jaj^BMSLVlCz3dq~(fRoyTlNT{U%xsh1B0X-$WcR)f@|2;6&?I)
z`72We$ALBQ%XEUL)aD-Q{HNka2`1&zt&g{RtHohuC%pNGjK_La+k7itg8LOOsn{dd
z+|?CF33sHH8))wSG!!ErVfrw!`$|~3Mq3okv32@v+HK2@{B*YkiMace2`1ia?9kI0
z-yaRW*E&=ARKQN7&7<$lk98tw!5cFO0g0>EyKC+v1LjJ{>>c2{V-**}Pt1=T<<Bk4
zO!+vFlg@eX&kR^U+rt~Rh}aq3F%l~&;m?p^A>kS|sb00#ICzAhC8$}6#f!fWlbxQK
z21x@!U2439Bd#s6jo)L>tSGkS+(RX(Kepj=sMI&CX|>=u!6IbwXkdA?_U4ZO_rWS2
z$;*|QBZFTWnGT1mi$#4@qa|*WG@8Z*>=tN*Jx+!Z0G`QT0z+3yv8JQm1I>y1Gb{9v
ztxfXKx=9<BVWO`(Aj7OC9lwR`q+uMa!gtROR{9lx+&T~}&Frcm*ji5U3#^HiD`{2C
zP~uWqqDhvf)x&dm-)XZ5T9wHeh*SQ6gF@?sgV_T>G>uAp-z%9<9pIvUjdBNhQSu(o
z2!|UHMQ$-rZwnkKM=vIHH4t^Kutx|`>|1ro64c1zNLocGM^ClMlitr2nd<aei+Y4e
z%0Oc$R=si}cs;sOoaez6_U9&Mdd?WG@<+}?S!U+q>(Q=qIM|Nh0KMVMH2jvd_+N^o
z2A|eo(WfQDuQnXKNh3pd{e#uylY^hU$l-zwHfRtn)8*PAEWrSd@}Fac^qdF#R<PHu
zK#@%*waem-<C#n<DGYN!?aOly3%vese;_@nJ-}<hkV)X)HA@S4?Q*#Q%jGcdiI4fe
zn<54M*JXp2G|=JH1unfa8&hjbkt?j^uX8`&%Cm&mAWbZ&7Jji=GS#1u5KJJWl;%bz
z&=UaHa~mGDhI?VP4sIEWDXF6Fnd1#B-FW@PnLQWqX`-7f!f$YWdQKx-1ESq|r^$Y9
zV#{U7pH~whQToXQXlY%cr?D&-u~}nE@cRa?*xV-MY=Du;d0ou9OWdCC0HsUR{>jah
z8#zt9UI!oO3*5EnI%R_4*+P#SgL7_rOPZr-fzCd6j=9uA{YU!d0~TSA-Vt@1=c79b
zF9ga>WLzyyw=yvSuR6H&OGhh)I&zpEgjoi+b)6i^&5xB!5ee_>7qy#S;v^F4b}Fq>
zgCB_AA01#AL>Lkbxx#8CC(=wiz|()tBK}T0_s^<=hjp%6>3anPdbH6dB>I)<RB(LO
zIAR|DGE3P*_2WMi75+he?;J_X#~z5)D)-e4HFy0`J$RN)O56D1-o|E?6PY2LR`H&o
z+c<~C>e{iRSzmw`eqwbqyczME(Dc*SMUxRM_Ja>yz7+nJi=|ZFYNH!96hmRbVp6%l
zVtmLX@7~*MY`J?@51F(t9E)hepre>+VK=-#Zz-H<KkQut#{@~~bYu%MsQ{5*T~7|C
zllBiN(C0snDs_owLKflc4|-)V@-U_hM!A+r8gq!BNnX^@l7a+-<zdM_k&`COUKX)`
zusADt9?$Y2v?U*2zkJwN&@ZkCkOpj{U-LMNHPcAVF8eruILR-DVRUTXKAvyux6jU)
zsx9M~2(rYC*sX11-t^Z<s(-V`-oFAmz1V*JVa)bKq?(gM#U7!`?3QBp_S_mua$&Ki
zUAMD5DBtOq2V)&>$Yv{TkOJ+t`7mCfRLMH=1~;JLHoK+lXy4$mK7N7?&$)-HF!z=i
z@uH}O!hx?Ckb1AU#P_?;=sbwf4umlY;|9g;Rdg#11T3$2;yy|8k|Y}y1l$*Jme{Ie
z)yxyK{itGQZwb2Zkf%cDW9~cQH`<y*tLF1kXf%yE>ung4ZPD%%uwI)=7HkXopt8$X
zmOsaQls|!`$YpQCUB+=b{hYv;wE>8+Aq&~6Wp3{?HZdG~d=e)up-5%(ar&d6A=^=v
zoy+zXqxbDy>?LS;xVf_qGF-)~Iq6Q@=!_<G%3G|=f1kfMxJB+w{5e(X3gPoeARl)P
znW)pjUy<W%2@Ys1)k3+B9qoEz0!VqIVGH{ur4qq&^#@T6$ra{tsccCDO}8^Xw*@&;
z4#6|ciz`8+kUOxb#8QaXhR9y>4WRe{sx$Lw_DUa3Ox1GE+kt+au^X<Rs-2EDFBnsG
z)F&fYix|d-`jhfrRTqNAh*x~BYk&6ay#6p8*;a;z2MP{UZXx^;)L~vm3taW<7|Uyt
zk$gAYieSR~EbHiMj}DuD=l8vZy1|fLC*S1t&hQ-tt?NDku38dsxt3B!jf9+#ZY_RR
zjt0u%y6+X1<%Te^=?V6&$lP!5OKC!<%%~QKbT$@`mv<cdFcMqS`h<oNdzW0B)HO?g
zYSy7|c5Tj1D-{Nx>#TjuGcu`5jTNMVb@|5}zU!($?MAr)RrJnv;K7{#ugw0x&iDVP
dXLnA?*Pf}Q-&6p;xC8>J{G<82Lec8&{{yrSTRi{(

literal 5748
zcmbtYcUaR))~2ZlB0)u@2m}yRnsg~4l7y=CB0UHwf<i#RPy&dwh$1BdN)Z92*Gun+
z5mb7W-b<u}5(1%bK;66d`F5XY_kMrm$((a$&dj{?o^xh?&vdj@8R*#PC@3fx)YV{m
z6ckiw3JS^sT57W9qLFPcS+S{e%RredQc_Y<QBhG-Q`6AU96NUG`0?W>PMn~nr9FA_
zBpn?cJw5%YQ>Pdh7*3x)&B(|I005Ynn9iIzbN1}nbLY-6Gc%t*f1ZVf<-&yvtgNhT
zY;5f8>>L~%7cXAq<mBYy;^OA!1_FURJUqO-ynK9omo8o6=jRs?5V(B#@|7!Bu3o(=
zC@3f-BqS^>EFvNzDk^&I+BGpTF>!Hm2?+^FNl7UwDQRhG85x=D*RRXU%F4;f$;-<t
zC@6qHAVoz*Fc^H}#tkJUB?trpg+gI4n6k36ii(P=s;Zirn!38WhK9z?n>TOWx}~Y9
zsimc*t*x!2qob><tEZ=@udi=lU|?uyXk=t``}S>PV`CE&6H`-DGcz-Db8`y|3rkB&
zI2>+eWp(Gy9cycA8yg#2Tid&L@7}w2&(6;7{{8#*_Vx}A4i6qYc=+(4qobpflM@1g
zaCUZfadCO{=#i_dtDBpfySw}2$B&Unq=$!xr>CcvmzTG<HwuOF@$vEX_4V`fd-CLo
zzrX*}r%wX{0s;dAgMxyBgM&jtLeOaRvuDpjLqnfGfBxddi?Fb;moHz2hlfW*L_|hL
zMny$oFql`bUPVVo$Hc_M#>T#W{rb(DH*s-s2?+^_iHUFDzD-I>diU;KN=iy<YU=y<
z?>~I_kd~H~o}QkOk&&61nU$55ot>SNlarg9o0pfDpPye)P*7M{SX5M0TwGjIQc_x4
zT2@w8US3{NQBhf0Syfe4U0q#MQ&U@8TUS?CUtj<6<Hv@EhEJb9VX@fHpFcM?Ha0ai
zefjd`>({T}zI|(MZf<F5X>Dz7Yinz7Z|~^n=<Mw5>gww5?(XU7>Fw?9>+Ade{d<3Z
z|G>b&;NalU(9rPk@W{x>=;-L!*x2~^_{7A-<mBYk)YSC!^vul6?Ck8^+#C*v`|;z)
z{QUgF!ouR>;?mL*9*<vMUS3&QSzTRSTU#R#2<z+XL?UrxV`Fo3lSCqIZEbCDZ}05v
z?C$RF?d|RF?;ji-gm$#1lBs{vS<S?if`b0c;g7Ow#CMm1;w-N^Ov%6_VQIw2>fB6X
z{ptx?(I6?lGg0W1R_CsAYIrbRsWss9czU_^$@>^fStTKE7WljS%8bIsvbx$m+g>G>
z?JNaPa}oD-m3p$Gsq@|ityz0AoxSnmhT8bn)I9+S`KtJ^B20^x?^j>2A76Z;=UY}x
z{z=cMmdgyZ@8Y&;o?L?j(gxBp(K7J>cwpQx^bIt?9{{8RLTLU618@(bkkex{)vN4q
z=fnKCLQQ^krC8Da)~gpy^GQvFX|eQ}^IfvofX@Mm>Cbzc!rl;7=R>lOOo9DtvcN`#
zkuLyIJyD+~g+ScZM5rWqa5)ia{{7~6_RPyAFs_7Nc}X0yu@8~%&bk)JmW|9#=%tRu
z$pmvCQD4)&F!LsT6-@!zCjx^Va}!%Q0i235yRan%v&QM^eK)`1>6m`?oHx9@LbI95
zZH;mKYn7KYngQBelJ|A)IilfrOKVZD=O5!rI34y`R1$+&kOtunKDk$MV|}N@3^HOK
z?pY2+0q6NrfxW9quN*e_rg{uZFay+<3+YbM^kQ?e5n~%`u03B;3o&Na{DpR<s{QhB
zfP<K!?*9A$-|%BR!qA+%!K6|-&*%Fy?WOT>(Cw^-e4==-A;PKGC7;tVIKyecu;!0-
zFmbfhVT`|betDZiN=fu~dYr5B@e7?oxrvGj?RyTvebC-!G0f2KIfAr>#VbIXwv{oL
zg`ubXEG}BGPkzkPAXePbcf7%vZEt$pT~8ivz+1c@#%jHzVKc|r+}5g>g~H6|xo3K5
zAPmx9ll^0ZuNRiCdt39Deb%yWeHdPVq)RfjedHbY#Z6eZG9*Kglf74|1a4Ql77^$4
z@JCuHI8~%?dtlIym#w{epmHU-&h2DKzr=7ycO-z9+8UaWlLodUN#1*8*21^Z63G;^
zk|}!xg;|Ol<7%&&Wp6ZZ53h<K7Y=E-iCCxWRrX7HDgUlX86u=##;$#dR!p2*4$HVs
zJwIm(R&>>DnJ#^;zW8Q|X+vv5|M@K^?Xz$4E@^B6;JV4pO+NWcav+s3(`?lYjL5>d
zeYK?Q&^8C=ldF``vVYhJmZWN-+n%VQCp>HMgWFB9Sr|bZfimnZ>-NKaii1st$SZ@T
zJ$Tjyg6*}uYfJv<*R#FkpM0BGCPSoSKsEMn_ePuSu_!|n+=wndzC+r3Sx)Fk%1A(~
zry|#z#7$B8N&X{$9ISW5j*cU2k4&yQ07kd42REze4%JUmWdmYi8*X}a<={GER+?hC
zBm{jjx0YL;a6+_dIkx)J#@b9*BfdwR>F)%fnh2J<Ne{)QhQtFo6UM4)zTw>M3@Pja
zif1?{>nuwh5X8~mDS>0o(rRXS;4FtEyw^>en8}@35z=3`LHEj!En;3@NzRpVHK3up
zNn1o%gv9rxNNG@GvT8PNMq9*VGB>U#iN5PXvkbt0h8<nM#sIV0I?hvoqGmF#^`{jY
zyAc>HG$wRKjG?^wuEEUO<s_-~Q<3M{3Q>gWp_wH($D^>Rr@?Q5T@c<FggMmw(LyEq
z;1UcKt0FW8;Mwg)&#*%R{m9*+=oM{p7h0x$^^JT|)`FrkdL`T&T_uosKoY1KOsJCb
zr3N;D@jy$|XZlZ-pfbls9El;2_MklEb%}QV<-Y9&kyUo`a2Lix=YTwqn+qlBSOhnW
z)C9t+t~p6tDvsJI?H(sj7DdYQ+!LSr+`zJ+pnW*2NCOPL!uY@Cyg##KSa~}gd10B=
zIs!l*d-II5JkH5l==vZzAQj7eAnl>zU(lk7L2y(30<$AxCZ4%S)5gi2RSm6X8dZ1P
zNSBqUvelFZADI(J10p3i0jgYG@off<D)^UU^|ANJn}Hc-hQvA!$v$}ab2D^e*@L85
zYp0(QtxFMSfyn|x&7=$&-6_sV&tu6sx?XqjDMwiG5iAJF3|#Y@@GOEPdC^}3%iB>E
z;C*{CEW=Q9Z`16Q52jpsMFK}ipH3|CQ{KM{foW8Pyxt1yYu@Gn!rn)DqwT8iR9^rm
z0wOi=z}l`K4Rke#l^YNk-KA9v0!zD*gOBa%B5Cd<)BXg{t-n!E4zVpOdqO)rO2y;s
zO^>c$66XF>fz<nVEIh;&0*ygz7|R5i<HREXk;P<QR@;p5Z%W7N2BO~={>j*Xp1~NI
z!NZB?c>009Y1x(@m2#b&WRbF9Ee?~JFUlz?{9B&?W9I%wW9Tr<+Zm}xQ-9iJOkUYb
zb?%q=ZJ}}>v^*+heJcf2E`No}YR0h0<8$HCrZ2{s1+Nv_^Bnb7zMy+Shm3L#Kv2Yf
zWV7au8HnK$nVfVk!z~wmvJvbP_Ls?^8=*a=D)SW+mVQZgIg$|6vZOYY3}zw3ymPA|
z;Q_&NL-G}1f@`O0Hveth4(>7H&^p032>5uauz&Vw<URC*hi`+=@a&%5>e}5jj6Y-H
zjh81g!X}6`@MO_fc2C-iO@ka@%##C1!-cpl=&#j?N$lRHl9D|dY7=yQMS1FT*G+|=
zD_zx)@)))vwx8v%PTKvsmxH0%37L=Npeu|;)Q+(IoZASEL<%k1XCOqK5dP<iAPRRJ
zYrLI(U^Mk~t2v=wo8#MP+9%AD3&-vfPCeWBB{2q%$~8sYXUq#nQy2s9ab+QYH7yH@
z_u(}Y+h{eAAa?)ifdSOVyATyX_G=Co?Bo<0<E9YvD_Yxuc^9Ap$U~hH^h`~?lzme`
z#<fZmIe-(|HNO3b_AQIw3TR04i3OiIzQKSQcJh)6o4hP^eECEtM98j)IRLK3m6=v*
zk;|FLve0iCGBK}Sx0Mgvb%MaU7MJ773=9w2q3%u}&PZ=ZjJl;cNIXLB9#`Zg=cP>(
z-Doe5fzrgH#>S(xv!^Jif;Le}mU74YD3PJw@iFST2eK;aWB2;}yTcrF;2$bKqAQ;N
z3|HSI=3GWcl1|-$USc}=Ow}h91&KfyDTLk~LbC}#J}my3(N3)jT38-%G6svUX&&V~
zF{n@uPTyXmym_tmSqS;qZGt^i=lX#94Hj8`Q67iP>Z@aFlCe5$MIww|MspFbA96?#
z3j#ZgC65?-|32ikh1#AO{rr`m6tioc0G|hS^>+U}Z*L#lYSK-_FFK-t-d%~9aKF(b
zD;b`HQs>!*_L@VwguuO?{JL4LV3^VM+D6<|U=x4ZUE5<ke^KP{LTDtgXX4Y7$+`SB
zP(-SA*}#0va@g+JmGfI&s1=rh&E6V?!dXQ@+Q5sG*~&Jh0;2&tIouQ_Ks*P$yto!b
z&c9Sv!6dcV_%rZn;EO}sJ#?04#=aoPIf<F3{*+6Z@Pr82VP@>`Q|V?IHL*>A37H$i
zP}{BJl+KKihdcnmVh{j|p9XS)sAyYjUAvMNWoB{w3{O-aIZ=m(Yi=rxi-(f4^1r0;
z-}O+^cB>IWE+T$fbcdH|=Xi7UI44ipS7nMQQdf)Pqew6Hz%Pe?s3NNW=WY8}<#F;q
z&Qu;O4I3xNz5j#zP@24ChCda*O7;COYprcFX|nVdD!;aYw2K#spAwJy>oy{$^ZgAC
z`g~rJDe;$@bK|H8XX!DD4wbt0_5Um)%+^LFtD>KWe@aY~e_=HUOQ3_8q^^G(Rz9Mp
z`fSxKlwGlBxz1qju*{GI761Cm@C`ARc2<K?0(ne)L#|I~3e-LeMBDzCx1k0WKQqn`
zZJ#d;Gx4CrR^!-tMSSXaJRpIh{Iz&g0n5m;j{>lP<v%`gTKTcb@fU~Ke06A<Z=t`A
zB^K8Lz^&{HxX1N7&mg5eA-E;{dt$Ih48dchU~g3W+}dL3_Ei~?8DL+HeCCG<;J)+1
zC>nfP)Wv+O+-Ck^Qx9CO9jqYc@ClniNmxd5g59P-FNacnm3~Bewh_u0xaKafOLhD0
zeq>!4F3$L*`E1Z0&)YV23>+>uOv<4Apc@ju18>Vgw7g`yHYH}4=2<4&J_eF?Nas>*
z-7fNMdjcP0a@UekW_azHPuNXiPmz8_Zo9@3cI&jm2Y(D^G+~_FPJ=vkdP1_tA*xNU
zLX0=@@7ozAef*T3)-2<|*HSX72p(2iJyXM&P#&%8Up;HfpA%5J>r=Z<ZBuVM*&n5F
z&L!d$@Juy8nQfP3Lb>;MugCL)BcA4fBH(rhD4|JUR@o6PTlcnY%4v(Z*mG%n#N{c0
z^xk&#qS(QQ&bipz;|ln<Emf)4GLfoE>rG<wE~7u%rX<d%^I*M}wdhFcOh|!a<f~s5
zhHBp)E5|d=^?x6m$y-6?RY-NrkI0x1MRy+xuD)58$(Pyj`2M-{ndE#pOT7|pCq3jA
zD`fpCKtC8z!;O6fgo)CM-91vjA@|NZsfk>SDppvp+k;b_u=fSd^Y5OmUsqU{QQ~;d
z8jOxX|J#>tM*?Km`9_=6iQFbe{QVSYU6&|fk=u7Vdc30!|5mD!5?Mi*>M->)iy5?$
zN(DTi^Ye3VK@si{<CDWQU2&pvHT$hO-Eq?$3xxg3<xsirw)E2-!B;Noc^BL}H@Vhb
zV%a-{B4KJ-{ey8F+XM6^qhk(meGadIR?|Ab^s*wX<`eOcqLZB)(;m5-rDdaE7$sqI
z9ns9D`IjA{pzQg#PkHCqpRUPXHiZ-mzkZ0)H0onK2w)JK?Py+$mF`R)yNW+QEj5q;
z<*LFlR+0U{GFQR(gWaW;t)MUD*7v*gZ%cJ1Gqe-Q-oU6tek@vM=RwWj{tX)T6jvOt
zou>Sm>B!X6H7hQ;WhI$Tp-^qwAi1oLc>~TI)*xfo$s~uyu}q@(1cLb~L#OQeq<;v5
z<|7=7w%I1}g=_qDM<Y8*#SJxFw<Bd}iuBMtagprR%)mGTsqDO;&60rH&B&$Xl%2h(
zg;t~gk8%0m7!WF#ymq)8G)TRY%=_JG7=pu#SXoE$Qxj$1&z<lNZaoL|!bIYh{E9)H
z=w;*0$0{CN%<HITkE?cJH79`lK>$z%jF{_H1>ZMhuAHq*NEm7@3j;t?UyzI}Uf&J$
z=oT)@?_C9o;1gXx-|mYTj73AyqqexUoM<Tt(az^P%Og=+Z-}g7>+zYUW28W)$-<{~
zc|RWUiC{Gc{PN-+l}KUP(8hr{&0t@LKwZlPYC*4BZ-XIt9Mqfs<$Q{07+{PwhnSgf
ze4w((F?v6++&QWbUJ5pn|2^z&eE0VPLVtD2BCFwJ^ma-aSPsq()LM$bbmP-ql9=p9
z?W^%RmjCSU+d~qtB6@r6h}0yLg4EF3T6Hsbc(L<&>?M}*p4X+ZL6{N|1hV)sS_)7|
z+Zq4j(~#>}ew@UuF?s8zQ6+75!~+TJTEnRP3jka~v|!N2{vkSv>%gAalbr+XyMg6_
zAw&c-yehPfz8d+AceM;>_D+<w0*L73L0Xr|oE0qz4R@`o)4?kOwJXTh_D#Ik7UzSb
fx&Hp=rIiCYk>t{-6u2$<8ybbWvKA~4V)gVNn}DEr

diff --git a/vignettes/examples/timeseries/timeseries_anomaly_detection/unnamed-chunk-14-1.png b/vignettes/examples/timeseries/timeseries_anomaly_detection/unnamed-chunk-14-1.png
index c8572f66c853a17a005406c9229427b07e24b0cd..f68feea6841841d40212f585ff4de12803ea069a 100644
GIT binary patch
literal 32775
zcmeFZWmuGN^ffww5>gV1v>+voNT;A6T~g9DNcYg8w3Kv%gd*J~U800^3|&JG-F<HS
z{onVTFXvp>`Eb6S=TbqLdFGC__gZ`H`!Q5mQ5yFF*#ig!f-Cd(wJHRH8Ulf!lwqQR
zfBCkFRtCOc*}v6xf<Ukd?|!2U&-z?JAWtDOuV1}$OGD1O8WO*`727-2@BhY!Nk6Jg
zjM?=0sroD8p-BSlz27;LWnV>4f^ZvX51ulAzJCu}Avg4hIg1Drize{Pm${Oi-6;3T
z-Rt2xd#6gz#DT5SWk;m>3CvZLyRy<LhPC%;z(a`;-t|#-2o3n|2A>`(fl2dPl8K85
zd?k|zO@1hWIrLDr@9t$m%Ek=f>&pN_Zt&u4nz$q^@D(?!Qy!Bh-~;CW|JMJHmnr1c
zHtI`xAD^3F5#(d+_h|yYNGarvL{N*(eRz3cyzG;Z#AiLZH*MFV=Q5^lNdmU&?Q*sK
zVta;&Cj%Yb*}qqy;7|#l-CXW!HV}iKla=Z<z|L)#oumF=?lZy8x=Q0A-kdiHqteZ8
zdMu;EtjmWf;-R5+IFZ<Gk>z<v%GIs_z|y{5Z^B|P$Nr$Cy`aNx6DIB(GkR6khvZ=j
zse={t3<~Wt^_ckGW159}8ri6u0Cgc+yDd@Q<hWq09wz}Kp;unQ$HW)Mk`^1<>FkA7
z#FggH*zxAQ^I)$+1@Ax_B91_WguhlEEKo00HSoYv5{vl3)O~*V0Z!<T&i3<o{7)=n
zBIHNWYwhGIGy2W1S=JJm1Wa`I0t2;bOQwBqh!Rsin{?Z+{O%15EDa9IF5~=OF}!I1
zq~(?h9W4l7v!UoU<){gO{l_%L$?dvBMH4eKi;?gAUQc$LJ*{LP2s(?)_SobpURRlg
zowwY+iDgLQRoJc|bom!6AEAE(6fyi!{8(j53Yh!S;NOaHe6Kzhma_AZDO4cWryk`g
z<$M+Y->+Ah^CH@sFj<w?DptwQSIJXKd`sBVtCTALVF(ot=k`3!pr*DX7o!>w4I||q
zWoK8xxC=Uh_KnV-__{SJ>#2KTW_fMOj~|~n8E3OF7gYBLF=n#z!&pa*1+2@pVq>}7
zI4def*|}6O6mDn(9?~2f3>G|+3b#MnEsf2M#FPG@GBkU60*4#SyFJTp^y-9Ht_}}7
zhtZ?^u@=79%_jS|Ll4W#!Ti;}Q#|1xG~zYia2C`S9j)5PR!k4Cz^Sa)&1*Ov&A)VX
ze1U=gYb`U!*SA_ue)QE{D)F+r^}(UW)BY7ow9lv{3-e^j<0h>W7P~E3?#8CWYHRaF
z#~{~ZZ=l2Vys$8ZE>%yAvcLRqlKWiRza;UE{t(J%grCm!t`8bz&SvC{xIm5+%|rhj
z1r0H^oX!#EIG9yJD^qzzc%V3QZfFd#QwWC5uXhPbG$HyU9_3&WJo@+V9+xd%J<npv
zgwn|HD^w<))Z+@Tz>(*^du><k?KVS0vn<TAOUq+mnTY-^6E!xj*_IW%f}H13>&K0j
zTU-13zpHDBs973TS$~{!M)EIjE+1;&EM@vWIKJqo6e`d&&Sv<xR8d`>ts9-xzPk19
zPAavwc>Y<$lg2O6(coEGS$M%z_EN7`Hl5jJV*J`ln+sr}p(+0`z00QMhv+SxY!<nx
zolisWAoOp;vqpWNkWdkvojW}IP@8>p%U`!7$)vl?rLgBd!{)yf+)VN)@a7s9FL=fN
ziA_yD%#)<S`{@GY6GbFp>jUwG4obXA)(%8_%WX0EAj?ttzHv$H+rvK^7e?F?Zx@^d
z_m{%rU%iU&Phm^r-F`Yq!HK$Ns*T!?65IO-rw^UNPvj->#>Qlg!nOpF%a6n)7Fd(R
zK#Q1Gy79Lz8%!D$4sIedv%u)+lTV-iP<stheDQnw^xp5$Ej>T|gyU9uaSUs(Lj}m$
zwQDuD!f2L;G5hu9w{Bi^m#;F5AdzKdqtesE`&0Dh8%|$p<DoM9If*lTjYZ)4;k-ln
z)_N&k>V7!bEK*Q#2Z&(iP;GYI!#*`>M(-0t#wM~wESlF<cFk>*<=!W|)ba5ysAVV&
z3-1jde<_&2hgkbzPolp=4Q5n)4db<K_-_J(X7K~pCudIZ5ucm7j=#spUe+7K-<;O|
z7N4wwoFU}8bG6`)e^KEV@UVRvUU{|>n_ge9gq-3bDg)zLYo%`G_0Q_G^(9ZjcE0x>
zC9&n;0hI@)KYG62hitsXP<VXhS@{Gr)$R3M%|$`_u{Y=QyDNVxl^fdk`_G?!qkT-8
zP7-vy&I1ICbY!FBM2TXn;#W*gR3*sIf`ar`h1Pq9oQ5bRmwO!!w-<?1^b7?A8A<8S
zg-+6RtFJ0TEc|i`X38sU4*Tab_3fLqonHaoPRmzwSWB1bj*Nhm+=H;9rilx4zKt0+
ztLH02(?V;+4*wjfQD601y?vL9y!7Jfx4Npk>*eVddAsz0%s^I=JSLViI$HB3J3EIR
zhBBThq!x0DR{T*}yP4=WT3V}vA<f`XD@t{k+~{lUa3BdDPS;S{P1w>aqgbQFKuAN-
z!^H*d$(1pMA6T1KRv7qbZ!Zy!{Wxz_-LnxR87h*Lc>B!Z#M(h?l2RO<b3+0#O~u)B
z05cRA#F@nMx-Y*zNOROjikt~WL3_u}t%4@<7DTQ@k-qoxR5ir(D{&Gg<cr^f1<O`E
zFRYpS!suzcGxf)3`;#>hgigz^CE)=Nmpr+lhVU18`nwR*r#SYg>VA*PAg}!HC*j;f
zn?T8vijvZJytMQ$s~iLQ(s}xr?UQ~c32D8USPR-0$P);uJu2)g1t-d4tIAil!L=|B
z!&(!B>>XxbiQMs-NGh(8kw$Dj3Fkegdj${-!+XxHq~hqRe!8tVufKg?YQrF)DuHGF
z4?FGhSfDms8%jv9?mc~orRB#gj(XsS*GetU#QB8j{;+hoxMRiL*8Ug3r@b7%2dHaZ
z3aQ-oCHk%EIa6#<gCJjWX%}R^BL1d-rN<-Zy3-g-o?#>=!?{->A-Z32>>=JCts&y~
zl(9Nd(6vR;iqK7E+LO7NNs@j9-wCy|x$g>Rn}vLg3S_9c5JFD!vxk!y))=OSJchA!
z*n9Ww7yb&30V*Nt?*TMNT?v0ke6RI>+PxD9x;QGUIK?2*mycF{c{oAL4c>Hvb{neB
zJ4IlAmOr}hvMTh&tAyy}=e;ldBE;1eo*6ziM4@ZNN<*2eoZdBQ5mZ<Cc(_*Xkn_LO
zaP#TSW!?3OMQ>sz`YP($M`{QuCI&Mkk4{GX8(lTZ3qRb_#!s0(?Vh{;0_nfm&i*vI
za89K2MxlW8pl~ar7eNR=61;k417X~IHi-6Q?++u80Pgiuf&Y{1C0KP%G*5TXK9;xk
z`(3w-K%d$Bvo=65zMyzvzx&m>0C+l;*D^8bhd+Za0RXYa79u&h6oQX7frj3S{_=he
z3drmN_FARBbBk^Sn9Qu5({?6@(0?>`2Lghn6e~efGq!u*(FDKZ45HbfkUzY)?)PxP
zR)!hX84pAB0+a3i4?qHVcjKq+q;&1f6q`pcL)H6;D^B_pExuKzsTuQZ%J|RlhA~Q0
ze~?wFdPCi7o;yMPIz*aEV5-85Ss8}{^%sO5q#E9@%6U4t!YV5JS*#=wolG}W!Gbh$
z4#AVVd+#s16A*|&6j&1!T47HK>&OSur~?C7(G#>;^-RxI3|sRuX<6j%!f<ef0}P#(
zwWvRtJMUxNmoItDEb;n7#jSqB!-wLjJv|ftpFXj$1nar#ZF%mrO;vxP<H44HC-L8m
zgUCNiNbaJBcXgeMKmUTSg~DQQXKjryBg_3nzp!BQ0dcieFt5#*gztb{{!?ZTql5{y
z-@nHZPtZhRN#&izuk+(v*p7*LsvQ}Jc&gp}VEy)djWgA#`i?IDTd4rL_|*Pb$lKx1
zNl6L3I==oW;-vOwX5X+w90a3M<vaWy-(LA&Uv9XvlCk}~!=D5(aa^i=I;69+(Xy70
zg99H{&y0foR8F;7kx@o9gtY}_y|94O1YM!ewraf<$i-bY{qWq}PlJPbDB3ZCB#))0
zx1|z$QIWB9>Aho}2-%2;raZj{Mo8jN$o@YCfP1H~_RBw^_9gpQTP2?Cc#}TeaPH2|
z!H)QK-}@pBZEC~0f{xU*UdL#$G*59G^+3qw6Z~I<{NwVd>qC602vkmfdW4m@;#)t~
zL~<P0+rucz4@-{f$?|wl`=DX=g3FV>`9<$4TVJc+VW6`+<>f0|cH8<c<H};xmGyNP
zJw3tVBK_i`fZvEIGY%g6@D_{wcmKu37L+!3_qrR8I$g`e)F1x;;$3T^Wb)^u=g%Fw
zy2|~0mVY%a7|K}}=_h!ic*uN?HB-|{$}%5_RY<9VDB@Fve|d<UEriwo&z?o43k7G>
zYO&g2KgfVQm5J6UT%0W5iy@3{d#dLUNJC4jTZ(Tv_41D)Q)DQXfG?Ha&xJnH>Ryl>
z|H(UKtkT}7+WPeA;VXY$nHXUg%lfv3&mmfTmha4Z6TqmFlbM)We?_SFCeY+aCCkb=
ztYCVaYBbEeru=W-s%p`3bVBLRG)hG|WK^Z3ipJiTd$^1=3Ol_*3Bs2Sd+dhf3J6$0
z!t1JD$|P`^Vouel|DW&v>C(}#`%^IkxnMuT)caP6FX{=wR7;D)s{CaHpYgFJ9o@q9
zX>@!bzE~e}%H@{Qb>R==7|;{{v&XRQGJB)3!7Oe_FdYs3{rmT|Dsp@CGT&$TQu*GT
zw~X-}dxVp@H2U6FtoHvChyOvNE6ibC+05FSZ((sz`3M<0s4TpHFNLprL}@vUWACq!
zB}w!L%;I<jkS#S-KCZ8SvMO=M`0$t_JWG<xkuC`DjfY@+`$MAT8*P~wQ*DjH+}GIu
zA?J8E%J;gktIePMV=Mxp)o|il`^Cb8Bp(;o3?MlwFu-t{*Gj&R%J7}~BYB2^z<fQ1
zY`y>1alE)=DCZ5OU^t3mQa`^v#A)@%|Nh}=Ol-%=?k=wITn*c=h>p#%$?C#!{9zF0
zr~k0`zPf#@6a%qVdcTj9l=I2nVz5^AgX_yey%*VU{*^-1o@Yiy`r9wb@KN-1Y6A45
ziddigFnDb#{SO1yBMgnVqJDvXKHYf2=X5}Y6ETb)JIKV`dmQ-zj|UE9PogQQ)%e1U
zfTTA3!PE5=D`o9a{ieZUo!MLe9WFjg%cG>??v_l^MzzZIq{oQ{*Js0Oc#l4dW72%;
zOAjmKoV2dl+Y^BxVMps9F{s})XuWReWQ0y^+{YJ7^|)4&9!MABFFl>d7rn5-<H5tC
zLD^c%j85J9xmf@%VR6^>hka08hc|lEc@EaJqFh4Ue|shr8fwksUQu)J1sSR2MVbo+
zl6-n7aZkV%uXYxOjFy&4OdLShUH;YP9xWIZR6#DV9}4GRiLcI_eWIQa-RN>|PF=!l
zAXI`qyUq}BuV}nDItF#&gN6p)({Fi&dQY=mED^8GdnzU}b?UZW@0|_2Pqw`79R4P^
z_BUN;yFV^24v$jstl14{%|Brr#!H<wIyGr1?U3gj_yh#UxA<iJwkxBueL<p8Z0B3$
z1uL2*6e3RWvP|2sA_po4`nyZr)9F_@sepjMr=GW%G;`#fklK0Mo0V9DkC=wTgIRAC
zx*{o`@Rya`zLJDTr0r!{PvqY}4fplLPyI|eTEAxI2pe}7!~%;|Bpr2lI@ds!bbp~?
zcW=t-Y!Rcn_jQ$=oAEf09SO{BPFgHahbEEoe6>)aTvK=GbAkX(z`e2qM^D`u=C3T|
z_i?Z#E%K+#Xs75!>E8!69UNb)+Zn)V8Fngi>5HLk?hc0Ae92febeuSpWc%IJ=!U<)
z&_FnSMN8_XwqFozt1ePi5_7L?(BBHPiE`2e{Get3OF){8QYU(TujMxK=GtM}#xgPV
z0|_oIhysrIc%#0gidX)>P?FOH+Ys$6rFTfT_!kC=&FAR17n&P`S&OZI^qV++gcI0{
z*Orfph_k9i?r^A>@p|NYvDozy^Vf+W+`%m_<JZ@>{#1#UEC19@Gck1(M~idkuSf{(
z;h~Cd-(%Egx$OVl8WjGDk<XYn)ff8A65BR9o4t>{(t;|m{^bPvCMuihgwsu)Crsf{
z-PFEcb8|DIM)-OElX;`%&pv<FRZ%HL5x~N7TOoL9r$5k5FOxkp^LyxyqxZilrUO3%
z)bze6HYDtw0a)vk<rz`E74kXo2#SBQHIWYMa4q!(_W-J39?`~z+wqu(n4)%ldCtF5
zcf5EypQFX=p4R2AugRi>7VEJu7#g|2KQ+=W&`L_23ws_y{a;<&E!g%_UP1W31a95-
zDZq308oO?FIVw7>2xbv7u5xGk8N4q?P36l>5%YagcAVk4Hz>)(tlA*|dgNcG;$YMM
zq^Tc`iw=n&8miluvYyi&=(ea!{pW!gR7@MaxSkhNV3@Q9P)>=T{r%r9f1bSeI>7R@
zsDhU{KQC!ioh2lQ_&5G$=Lx;REux)<UFn5Qc;)_#Q%IQF|NgM3ri;z|+Z)#)y6gRR
z>?b@0Hv@OM>tF6sgz*c$RgE2e_p?o&{qsLIocxG}HS#?I?fvd_Fu^nXtGcQCv?)^m
zvDJRJQp@FzoYeBov|V@Yyz#Ga4^cT#Mf`iC>YMw0P|sKfsT`HOkt{u-$!&Z7&2#FP
zDgSHy4<1Ok3;2Z+F@gX@@#{zTe}zyb!2NFqxg%M=w+z^i1Csg~ARJ8r(a*-%iH2VO
zld~q3&_4_#RR+FYE)!YqPQtg1|Ay4@90+?8bji8S(0P6eYOlSIQ*UT|{5<|I)+=sS
z_qTsG>(FBl-Gb=<Hk+~cg-c*)@ZCoI-#E6oSvo`i=JWx(zsGjv-<Ayw6<II+1H}Kc
zuS+K3)EOchs$C-N={!)fqIO@yWeJ4cTild*g(x{5M`k)`z1vn{2jR~M-?I&i9QLgj
ziz6$t`(}5>a+QpiG3=Y~b*57Pz^>p=@8!r2-?{YbVJERw)vYhUSw{IPWekI^3QErk
zG#8`zecMt;Yp)z05DZKN*%xz&PO^AcFidf0@+aRc+#OlA%BvR(&7wH0b@*MxCOIiL
zQ7=YP{OVST$dUuNp$~`GN998g&W{Ho8%%s&U9QHT=rp`^QcN!MIBJi#i};KXQY&M&
zLFnL%9L%>*H|vQCR@LCM>iu-ynW-O}lb4YJi2`ln?vrWARG)DP;Sad4PITUvUX8r)
z9Nx?MCCXuQR_-<#o2f*0Bf%$9`c=KV*?X_X+|}UbH;E7x<yepIWpK>DtDs51<X3oF
z8ihUrECiQu=G|&VlZC<^HeV~2Vm(XUymhsG8;vG)++*O?gQ6c=@3q{ARjMJ{;4>kT
z&gNAEayUxBT_Uoa5$0t$*=oG*b^k-p4)L>AxL({+U~4e=yUlA4PdOJOla36PyY=i(
z;jtPo(r$QGoHkK&=gW&}KAYd1n!P<I$Jlu?yJ{oSgY!pEOel<|l<me<T7P-6gjUBi
zYVw&u%k_95p>xKC*)5Ew>K-Y0tLj#fwNqZ`?QHDPMfpk&91)a+Cqk$b?9kAFv%X|^
zYrVaFRU-{d|9dg-2`hUMab_mgdCii7<$P(1iMbOhi3wm0_v^_v#(yGrk_NFq`G-sS
z4-t+mE~@kL$bvVbWWBO7i<K&oGUlB>(lbKbAMeQ~D$H!DfSZNR?A#$apB!Q57>!~n
z>DHdKmPzX2)|Eh;(yW9@Lp!*v8llS;3P(8JP)olVUv;^CGF^k4Cq)4UzU0Ki;H#ae
z7~XWf#*CPc;i@=1<0CP!Ab!sZHZ|9wAUdgEMl5d1S2>3X;1kn9sW0+#jH739EiawP
zTQ|<@d&GBlv9g7a|8(?N^Ykq)*j9T*)U0^B86@aq1EBHtdR>||WNn79@;s>2@j;7f
z>D`UQD=WxF@W&4wzq5KQwp>y+b6|t}46l#L_myzBi1Jp*r9STG2-)J^PINAOVD{|J
zdQS&vD0lsfy~W|VHPA9g{YYe`xY(1=TV%UO1%GBb@w*QJ7<02G0Al3;@hPdN!kXfQ
zzLU0;hqi8>RYoS&_3_L$?B8+*ynMk!SlO&_nIyt8o>o$(>r?MN66BnxtF5oEsw(9~
z;McJ8Q_fvaS=q4q!rrf?u5QwmyfmzzVW(Dhcz9U!_R8txqOi2NXJCNj;Zdqxv-{a%
zuz}|`4@&tB=oPphu0FpN_B=t37wdQvZvj5L+A3{{+vTLE?;=q-*_rh`SovdVcLT`0
zIpkY{G`089>AdfJlP8~FzFHo;R+-_*cLe}W0hi6VvKMb-R&xFr_}*Thg1!WVRlm`#
zJCYJ|f`mKm%m^5C7W{!$f&x`qo)rRXxJJS&^YioDX&v7^<0Ahl#=kXXqIBi?3bv$o
zx;alZS?;uSBxzztKB}d8)6>_-%FJwBI?M4Fe7f&cl|_AheKV%V=4gIr=WB>!9ybfi
zRvQNO)zOHOH^JKk#M06dh}D&q6`5EDJ>To&<`C6F^+Jtrxw$K=tIfAJmte^IZ;WB}
zy5GKi+u7|)<~Hy8{HOu8E$F&40~X+~h_R8;VwPkuM5}n5;F**u6w2%?=yP?J>3j2s
zYuv#1y4r5OnTS!5LI{Se91!)qIbV|wf1;wHp`ohk2<GpElUuj(J=M158kX4GbtB-g
z)Je(>ZM@hlYz|RZRb2|}A2(?az#=6j1<qkG9wl%EdPkJRn!GPhz^_LnBrR75z0<XJ
zB)U}kf769d+p#$ylXVWuM;jwi)MEO;FBp=hHZwQR6!X2Ab)6S++xyC~l)>*{{3nLK
zN36G}2ev!c<Z*0i(CH|c{^$`X&XO%GERdAi38erpHq$l5y7j{(1m4A~D;)V7Nv=17
z<ri5s-+QU_z1=Qj-uP0@94`I{`E_AByVModa=w~cQSk!8o2rn1aBz5-$gbzH(28Pk
z&-(;!?&Rd8rltmps5KG;ul<JI=E9<)dz_}fzaWv;kskpBGrSJ|bOhsbysu)Sr+;Jd
z=?t&IF@eKC)Z`aVV(;<!S|+ui^N+H!!3}4}sA|vC9Z~O#I()Ra#Kh_ClSeEX){|vz
zB$~*DHvf6IWn9SZ)giu)wf=c*FJfsLcdF6-@L81b0V2`B7GBY1TdVo*9fhdZS?#>f
z<wjnn*S^`D>%8~cK<0ddYw2(sGb`&%gR313c9Q)jY|ed+vU!op6ImV}9{&Ee%4KW3
z%^!n9zcI<K6SU53z~>?y(*<2N^D=#UdV5*e*+s!B_aY9uPM^!WyPubvbbtQ*Ss$RQ
zuTMU)H$E}3)@GVFxT72(qbrO=h(gqT1v-!+0_^5{%F4=|=6}<kNk@8JY~)q2&bv(L
zjivO?O_v#Ufbmxtbp!#V-YqA4A`2T^aaGm31T!)3y`qG9lZ+XQ1vVwGI%`+hH|vm{
z!90VB)yF)t-w71+4gvHow`=FAe6Ji*`)hozFWTEBz%~^Z74=D*w3rWMRKu<9n$N$F
zS?ZM;q6l?>bZ|J^;Ci^NO(g{D+i;$;=G9Jsb9gJJybQg5E+pg$qM5>k3lv7vVI0l;
zW<A6WvpbIUV7oHX)2}BCe4k%)Miyyo0cGUq=osf}stpYWw{eWB1D{n;z+3}2@HxNl
z*NgbM94Qn@Eq3#Z+|C~VbU^Gj_xtzNwYA=gwA9qp%*+ARkbbC%4eEH(GU9T#g@c0w
z!?5<&t~qqfQUr$7($)s%(iFf_FV|y8aEFSz`u(SDY;25-i~<7ZO{a4vzEfcJ;g#m+
zb6d$gRvJD&%^i#O^}2Eitnhj#bM?C`z8~tjnV&a#f87kT+9ygo0Ra*~wKe=;x8;_d
zg~cRY4uV7tm_|V$X6l{z=G=0|EX}4ln3&FhHl@JB2=iI$486WQJ^XA|{l4Y;L<elF
zVH;}U*iQgaX)5>JA3vC$J@cnjFnDK-I(j9tT3+gVdZfPUE80@)nyT2b>+UnRDR~~@
zzLeODsF>a+=d;x<GyK@G`1|i)fbgR%dbb~j|LZJ(WN;*fK>JOP$P#EzXon4NI5SU#
zv-axL+E%|WE~;5VfCR;=3AYC5Qc+QHI9ea#Xt^R3cXT`evER@2*raw^DN{5(F%iQM
zXaxYY=j`mz`c9H^$c)=^L{X}~hDJhkbX&8y0p04xNS+V{o`!#Agt?j7_1SV1NTo%2
z?R;q+i$98rObNJOy!e|Y&@pFfuzGR4bv}2a+_P)Xf47(GIZ=-vKh{uJ*DfAU7jku+
z7Xqk{i+iW4T46a>z*17Tlod=R2tz)(BqBFhoF<Z0RIHiamX4xQP*5<-S=Xtt7Vy3C
zdh@0WOcL&w%;|Ml%i0Ta6S(k#!TwoK*F;oLB4I+8cvgSO_oUR-$0o~7Vq;^mA3T_R
zv+XQcYd3FOI}MWCc!^$v)dYLwqO7c}&-Jk;FGUxbDu^6L#$X;*Gwsy|*Y;709=u3u
z`(??>W^GS?ZnLr8iPuy&$ko65-uja>TCqUdIPz2EDP^*QrBdKb?5#HG#)Emp$H#9^
z*GAcOK3Dr*MsxvyYs$x}&84N3ZpRmbU@RPaudTh|SkKJN{5Mmq#dcN@g5>Gef@l<J
zMW?0~;z)oLOv}Nc^e4^>6lca@`h2s_v168O;8OtRBB*xIdLa+6vEihjur0z}xZtjy
zo1@R03DOk-XyZE;&p;xZsr)`{vU(ZMtmDK_aejUdD0x5nnKUyCiw*BI_~wjZ_2}pb
zWT<k_CNO6a5$~<>Vn89rW?#&bFX8PyEnp@H26}RBqtk-UK6&Tv{df7W4`2uX!W3ls
z0DKXTu>D3h<P{Ya<>X94-2t5z`R~~C%gV{cKl{x!GgWP+S)}z5JCWQEFrZ+%;yCnp
zWcr~?(G`MKq{-y`6r04Sc71Kz%;pk*lX$s}&<G`z((v17*Lqz>#k9!^Gi;$+;CUR4
zX{IG|sk0{mB0z}#VC}VzM4zYP?p5MVW#ix|ZES3;sv2Fi=THB8ez?Y=T@j$C(g@fV
zq+!I$%Dd_NCrSq4{N)dRBOJ|4O;-?z`uR%T3RA3ZUDbE*-hnUxe_9!r0Nbai=xVP6
z-#rlROH7QazCMVVZP?{vM{xWpC7!g5jEsfF^ZA7CUXGR`Yc>$F+mz&HMvOS~n^6ju
z{?qn_ti2#ZgV;oYH1m0b^o1D7Q>ttd^0}%zXugElfqluGD*!pd{MD<EmqHqaJOTow
zw#PX0QtGz0>_UP7F(47|Ep|-kxl&?x*WPahGXk_%@3hulJOak^_pi*YLLbjKt^>uD
z;}j8f@k|EkoGUT_D|W51&P_aZ@K;K_L)K$M1*pb$iGI@s&}c@!fY1#KHHzK3fMO`U
zINdFMCk2`O!U54vh@bkR$qyvYA;`wFExx|a&M#9XC<qB1fy^2!)FdG${#}t4A3s)M
zCdXbLghNUQLiW%*Aqg%_5IOmS65Ii<>2YynfhffQ7I9?c>LaVg_CSy}DyO%5{MpAm
z`hZaAdCN+Y2`c!8<e0QHq^NuMqP-U&Cyl~*J6i2(%lHF9KEB#FVUOzuca2CeyL*}{
zDjc_`s*%-~)lYj5fJ%I)t^KFAroX>m-Ik^SpFq9W+{Y61U;%1fe1ogoQ2@O+@X@*(
z<<UcU=a<%pa*(M<L&TLwN6v_9RlbU;7<#C|j1BQuXphT#!Fd5s?<sg(1fcAr)mIUM
zEX(c6yjyW)8#p_$p?@<(ddtyGo!sLRSRrO9RWF*p?8@Vg&j|yvz7$8S_NV%uBS?6^
zckkmV0Mr!Y`$NslP_(Ko$I$SFTUj3a0fBG|IP_PbHtNph?>S+IrF*3sks}+XXs%!}
zipI}?sD#Tyo0nse^85f5mp%d%c(8%M*|UW%Z{!M~FHL5HUn1TfcGnljJ%{sSHC|_X
zi(q%s>7n`t3=I{QY%Or)sn`!qr-?WLKyBJcp%tItz@{Gs1&-9R4yAScQqI1zZ`^Y;
z%*<M-dP;^&c#j&Lph`Tl*(zW!WObs{m3E@8%TI(#wXDy|i;9X?mxsEesLk`9c2haM
zde!DMAd(jV<(NEv*85`1z)`q(i)_!70OU!npw%~|oaQ2{fGX#~Ja+<(xQ*o*Vo%FW
zwdcx=(0U&uz3c_k1=AyOCC-fX2zbLXnqa$OW%>Ac!OwVj;!X5H6~&@iQgo7=JWR1D
z@rMF3*(i6~k`NzXuge~(i6W<0t=}~?G$hM1qhxM2L2+U}k^}KN8j`ji`~U<fd!k(4
z@(7w*JEedl9XGdHJ&yN~$qgXbndy)oQ@e3*V-X55g}#UeAuleVWQs1Ge2SQz4fewN
zw-yPlzzxD4vl&w!rBsT;({%Ra`T_$3Yv())9pQDSNwZ?V_8lGTbzO?NI+=?`N8d%_
ztFrJsw|bZ+>OJLMj}IGln=MN;OIA>D+~vZ!f8PP_rbiHeIC%Z!`ucPp*#%M?KQHN)
z*cKAj&<_RVM0%>#)zvjdHDg{(9f{xmll{jzqkg^Y8y?sGYswS#Okww!W&SkPSWpXq
zi`o47X>bJu$eI#^mgdz<_C%|TqYdj{wyij3v9}$oI4qA0`WqT4Jo^B@0-0Y4?Jqc%
zSGA-Cen4(+r*~E5I056{4%b9=h{d~i#C|{;_GOA?YVsqjOka%*56=V(PWtS>e;1F-
z%5r#*&u$LsSkIcUIdFUVp}?lvvkw{Q$<pb0oR}6t%4{ReQJ=X8$}ZRH-dFXR{jN+6
z<He*nn^(^G-I3M3h=Q83L{DRl65B|)oB&$kG5Z`iu7ThwHCY3;7a0`Lj8qO68(V}&
zA4*8N>6k~vPPIVyMhbF@iG@XtH|filuUG1%aw}tjK!dp$94ou8_vVSL{=CUW=5w?^
z*JD2=%d|6w<{*l=6Jui!K~*`Ck|f|Lwl&k89RFO-i@K?}Wl6)<*FF@{Jm}g#O+7;`
zZq7|%HTzw^)b;jB_g9jxa*g~G{)S6E^PSN1tMkJi;a7-7dMWpV*k`=&)Fup;mzTM@
zxu0`ke0`t!(`i?Yna{d0&$&Au7q&I7a9;SrAa254!={+<w7#?_l_55NY77zK(adDF
z#MfQ;7E9A+;jhTU(nX#VR{qP)N!m#E#KNgrtLE9#pUv_JY-hVg_uXs}sQ$;j+j)Z@
zE?<)E_!E11jM@WItoXN%Jx)83d6qt&qT5`%)3v)ml$w-!bHVg*JuO>GTF7vMK7W27
zM%nW{Q@do~O(CQvi9S&adCxKPd}$7$nvuHXk|97585RlChr!J#s1NcPH<)F5NkX+3
z_0`HI8r%O=AbnhCpp}#&9#vB_EbNtiGtb6vQ6fijYV79qZ=<UBwiSNbfr8D3$BwbG
zFlJd9Ry+~cW{5f3StLx6m3G_MERR1SgW!+T8*sOMNj6cevo?}X)hsEpdZGK0t3hmP
z<y(pfG^acv^Jpn4iQ|GaXXtl%(;r@SE3>t*PTfigSLy?^*%$gORCT)Sr#^E)ZUAi^
zX_-7}&FfSYWy#*7vHgp#`=;m3M6wav+sI^9_hMF@wMaY7q7q~B2&raCctKxJiMBPM
zrdVlh$KVMw_OFS^o$S#1?CeiL6=?l{2_vaQ-rt(@i?)^=CB<7k|3X*^<iFFDkjtj&
zi)5|%yiv90w|zTS#+<2qb|oi@)oTtM+QrqdYAWvT;KY;d=~Hm>QQfmrE|Gobc7>m9
z2;n&9!OT2WW7A9j&}qoniMprc_}VfQt~53yBoh-?$O?E~&!2+B>Ex5b+7VC9X4<9D
zPH=d2IKWT}m|Z;wOpByhWcAZ0)Nh5F#k}DW5rLDP6huTFEV_z{cnx~q=ZHk<H*d=K
z=})jj_Y{mzOx3)i^mrIz|1<@6YV<B;kLwjCoSiG;tq*2N%FD}FpEfvew0}BCUxf|%
zw@Vp&5500F`5_69p?hve#c%JY=P~2iS5#4eU23A~^s=kzuW2b53K{+YRdc+jyL-K3
zjk;W)br5msVV?-s#Bn1umDk2CP)(x)l*Rk|``}0!WNCRBo@%_q#Zp(g{%%X)Au+LY
z*CS1)TkGN6(9&rXU0!{wy02fK!hS!gsVY#+GhvHbSTH;jRbz%BC)?<iuNQ&N%NZ}S
zZjO>u_slCOkWD@XYT7$4iQ4DPh@HGB^O7|g*fyw`4<5wOh3a7!-Be9)&-mVGLuC70
z3aZOD{&c$q7HW*WXH8hk^qnISfzopTgJ`<RBzI!VJx~sad=6(p!0_9m*GFni`lYPt
zY%jRvVsbl=Exx}@?MFk$Q6@v7&@BAHquW0O)%Q${kN>t}2yOL4fjA3NPCsuGfpSU0
zX=rGePm=00lY0aV)YZGbAC#l@u#4=(wrsYXwy~sny0pDNUtC-SoR$8i=*F}~wTC|K
zZCujUyF+9?tfqWlYhHb4tX(GNrSrP70UeZH`R&^fPb<G=5vUm#5g66T#-~r8N}5jI
zs}mlAy{rGXn1<zAQW0ch1?nsbWBlU>)jKJkyJ@9B2+GB*WF%?t<^ruF8<lLrsC51U
z{wW%$rNUd2#&-1%sKk%zUzi*{`7f*|IZq@6@Ikc=)vY^n(_57#(I}n3*0Ruj4%5}8
zuws>xl472Ce;t7aIw87@ur6jEMK`xIo7!3Q+7x3@awQu0)R%Sk2{O49Tf&uho`XX_
zoAbclI>D@NGhA@#VxA|Q#^$Y*%76BH&LlD<WMPh|;8^DY9v-y#3#k0dua7O*?_BQ0
zh=~=)`bm`rP=Zeiv6@K%7&Al;eRxS0F>@N3p3v+w)73P&Q$9(v1t!=ZrRVrmeY^+C
zp*4HI)+$2pz86tt1*J^PPnRu?;`_)3bO)eAf4c7CEr99{z*$(B!gefWq06Z5>)kZ?
z<Ec#FzQwdMVJ<X!67dVf>ZL8y^oN%}2VCRnWb#JpZ)+Zv+V@m5%cueqB$h#8GP&P@
zNe7O%B}#^ap01eJX~x0&7<)diGF-}}-%^B$>23bgv(*?&+~;9k?FY$=sO;s+(md}g
ztJ@rQwZe))d4NnWE8`qT8YWAXPTUzOy2fqAlSuMImzM5p9Y^X|a}lVR_XCZ#lM)c=
zz`!v^d)MQ>6xNA!7KH2G8@DMQx40|QJcB*We;IO?eHIiI)z2F#UnM?HLL(CLoY^I)
zP;&<o4w>G|-p~h_Zhd{d+52*-y{p60-5|hemn%3RV8&ysn1FyFM>=X}BoDuvo_`XX
zUbFnDZ_bnh9p_1_E04uDV|5Q^khnols}SyBJ=Z7z3PoC4+L?LNP`GjV`msmzz?@rB
zl6$ln{Q5_^R&OGQ`7#Ow2fZVLZ|*l2g=ZU8eLe6HH*b>?Et?)foRO1UG&F#gu)L_z
ziNd#M*=lEo3pF`T&=|*p^6MB^1F=b#By94JEc>8NAPQIiruVvfO*i3o8<(rarGJxk
z-sHlFvkcX)U+nxP520gV0C74p!-9z*XsxTO1NP)XJL&f-oU<S_PLHDxC!gq7cV-(v
z61)Z;9q6PiipKR@I^_(lHwz)d5UKCAjCV%!iTrfzn&O*CfiXZy*^jDziQ;Zo)+8q{
z@9Y~(zY46DjC28X|Bt)*s>RrIU=hh}>$0Xs`WoBI9U{Rlo{S`;`_nKng&Zfv3r1fm
z^OAr4x%8REG7{qCb<zh=YOwJ;`dKhOHDERqV`EUxEag>JXm_MgGVqFgPQQ946>j*<
z7wNI$2Cdi~Z0atW*s^On9tWcTFg*-XD=S|*ak@3JTRou<3|~@m?R#L0)%8Tc(g>aI
z&Qp^J`p#f@!aEpa=s-OH+`;hK){ej|YXfKE$tSHzTqe$fvy%8Cxg#4A5)zOz<6ZFg
zJIAFY)3Ua~O|J@g8=#am4|Tt}Ixl`-{UTXDIy$;e*`Vw@eksQxkeRoK17axQWIVF&
zYI!5GK9}2}p`nk%x(qq1tl=@2$|s=Kga34~LdaEWu>wn_<h-i<bW)i&^~u?!NmMzP
z{`kl74QD_)(v=l(KIn^NOCOg+Js4@JJ<_$C^t2#tJ4e4XDm2t$s_J>EJ=Ky+a%HB4
z5o2hVaWq}%_HqCgQOZjT)UUuB0#ukc@@Mr;WoabtSp+$sVhr7Qt_l6V&n`&emmT<G
zOtss%=#pAmlVHTufCo2aV*k7j6@ZAGuf!se^>-K+0&a+KFExlLG9c&K<>Qwev93Yo
z4cu@Z&;AQ);2o0l+Z!4iFa1!-0ab=#9{B6mD5`p+%dt`eF_{<!#kBQ|q<S1P;Owrf
zm_m$<j69lrMt`U%q`cgiD9zL_y>ptoBgpfPlIp#Z57!26k0%URn3?4n)QpXdK_2RZ
zayN+euuowTF}$>yiqjPOT346E_I{&zDq6o3)C->S@;Ewq#(?P>2@lWqm*MRjYTxB8
zhI3^=K*T{K04**bI<5943%W3Ovll7mr3yHGSIh%=4F{_@UBg!B4(HA(`nC4xuXprk
zVDTj+b#!#}Lo3@14b?aV@WkrGR#K9Zeo(w_(ib!C3<0eIAJ8~hVzk+BC*&^V128!_
zH~`%T`$g-}$jA&%qfbL0R;nw0_9A2f4@E|5)ZiO}vm7yW<qXeC$yhl!>}G$uJPtDh
zwZqx1c@;P%RIc^O-yb0UFBSFIEM#jrXa|6thw?<wnW+$4DvGK&e7FodzG-225>&Vg
z0LbvIg@uKX5KIJ4WK<MB9v*LMKPWanM@6m8nSkzRWef@2=XiZ+Ig~L2(n95AX}QUm
z=|xCP+?lclX>V`8v*)4xDYL*R<l^E2=J3SU@mSC&j^tiM&+p$MAt8qTE`XCyXPgG&
zo~uF1e5iWq8#hC;vuRIE!epO`u(K<YelYT%w086BEuRKj4^;mF*l4^!vP|fEva!Fb
zUha;<u<xdHJ=z@Wz0?A}CP<oi4;~ovA#<{`FHW|5e^G&YPEJ-<3!8_50Xq2g6?kLu
zA5+Wenr)Ed2q`Ei9zEg#9TX7O6D?)${L8qndRR*DaoR7m(ntm|oenK8ExoU?ws=Wa
z|5E5)9tJS;d~bbf-bW9Yg@lG0<%u-#4dqDV^UjZWPqvaP=EX)Bc6WEPoF-|vd}>-9
z>x?#Y2iS#Z`RmB^S+}r0mV5j5?d{F=ZryUEM)7-xP^KsJt5X8$e~Wc$@6?g%l!r%C
zhKTocwbcaZ7&xUxxSZ`7on|3Hn=NIm1_FjjLPd2mWmWodOBPSITso41l!S!OxD%6}
zD^;EW0V7WA6t<fahJbZ|;vX~b^4Uxs{T15vIv&#m-5J5KE=S-cIh?=HW7h=s)#=Wx
zHZ2x5b_24a7eEJ1JSG>ITl1jYv$o|;v<Lm8BJ63v)}S)RZhfSw%jK~2`b2Zw1eW<}
zhrV#E=OT`k2m=G7GA{{aFJPWegANFUB9CZoZH?sd<C+gc##d+iYI%zHK`#QGS2urR
z%ky~C>_sx5DmV4Ip3?SkN}(Fy8bT}KA6^Cp1>sQ%Pa2mWZA}bV^F~pL{45DhCAS>P
zegljRh*bEKwnzf<(fpGcfzl=-_#rzJlN%ylQ_C3EvaUt05|hgXdgo+3%pm124;1pJ
zE2b+>fU$!KMNtS;0DT`r2a51FQc@M)nY*j0Ck^KS=s-YV;8FaZGj(yax8DO+91$U*
zgruZE&Ejj--Tm{PW2Kp+iT;)r-nWH(sFQ>gvQg48bd~TKu;=NDt|JP35l^^2Ei2b}
z-e6;4CBL*7292~hSAcyjl7wgUOQ4BDPEBo~tLp_Ab_bvAg*EP*CjHDWUs`ui>#Rn;
z<N5(ve9%J=T74m5VQP+xng-sFsHlLc&I|T5wI2#14tfY&(?m9%goK1uZ+=87&VvVJ
zyw*t|qupF=#lvNQM{@UVjm@;1f=yvTf$>=8&<AU<YM^yNCv!UI$x#@jl`^z-4BTf;
zAEDpmc8&YLK$A*KEM^;(9NT8(dp{U7P_Bw;{B6T@-BXWA97EyO(B+37O*hHr8FyV>
z(1J-%O|>f5stW*x)jKmYGXNbh=(ABZB|Y!sG0^G+8vKcfhqHvlrvol(JrYs+-im%G
z3kwU{*Qx^)pa?2lSnl4Cd3%M?8mlDwhlMG1vJLu2_^^)~Sj`|m(9kT_uVt9Hc2jvD
z4S}wyV`!J+Bi=+c(ytG2aU&ulZ%$ioXB*v}WMZhi_S(43|DJ($2i<U&o8LiVzPG!D
z1{9Pc`hj~Hs)SC!)dduLceVj|mAo@##KgH(@OlcDnrM)q9D&9Lj+JY57i}D)lv_0#
z4CFG{_WRgB#N_0?8A3?V8UQx{Y`lHLrN1~Sb>!h~YD!1WU1R}lIAELUwx{zuNW|oV
zA}B#(k{b><CRy11MPVMK%5uzdvdpLlf5fBXTBZ!J#y~1xRKJtaq%3v?>s_djZ|nc6
zfLjA3TEB;b&aCL!f<MT#wScrKyFs?KDya`zmV6ExE#>cvd3;%!czMBr#UTYyv`v)E
z;_}TXA0{@iM_fL51o<wXJPhh-po{UTgp<@efzM+-`;%ur?;JePSLV7q_xty6{U(nw
z4{X0>Haeq8Wf%;`3*kMCPJUt51BNzhzB^&fVvv%k?_qJgIVPCC1L_2$-F0U{UTNv%
z*t1R|CEX+Q1rgwecS<b)OSh;9&4f1<G&XkTejKP-(m|*5*$QjS$Y>>bgmWYk5Fl|w
zq;eb(J@G~L!DL&Shh(Y7k6!EQ>Xzy^%@n!>dAgpD=Bt5z!sp0H(7@+v_yAWm5N$t+
zjCe72oNU?BwSudiFpws|4a!Pj+_TsP1q3j4^y18$1Y=(pfY#2$mAtWW_L$|dC|83n
zyLJL_ZsINeAU_JedG~G<xC<ygp4};Vz%4F3E*0?r%a()DZQmo0UGqv~5h$eE9t^D=
z-5TrU(_>F<Z8+#bDKMx3U81h32UJAMPf_4y4ust>@kr}7%LC5=`s31NqSnrUSNr`?
zxrsD;f23W@bxmXAj5qF83aCc`7uL;`aj>zOsR@EU=8jKh$TGz5^XS{YSJykfEt!XV
zTSDE3+Iqv%!y_Xg%{-x|E;TiqH!^3|3D3@ma-a|p1A_!Sq^cR%D?;H1+n#bkP7Ylb
zyVHVxDIhCj<8ql^;B>Z9zDC%~Cvmn-ls`#G5+MZ*rh>8!OcHO%$KK>*&>aS6Edm1W
zjZc-BC%w8Vvfe%Yi*PJ|nO~Ilo<!fXJLu)>n}gtrW(ASV#j{4gw^MrA-v8^VC8(_I
z=iTvpB^xKxcA5!9`H$vD$AFjyEo1i<9;J3*FabeiulpnG#vXE4eLxMz0(a?G{-}jF
zM(VwN`}?~>Dmd7HydD>(Lzd)KNCf`2zdLBK@Dam{GK|t2zeT2{aK=m<CYWAcUT*FE
zDa>c`G6Xpp+DHu^*eutJgAqBY{=2IA%=?nswoX*93$8x}CIRm_a`xddkD0Z$IO~ri
zGyS8}PyswMKC5wLb`;~j4XkW>4Yj?TTA_a3waazoDnBnnbsTZR6Rr92_v>uk<yc0g
z$Ccs_gkU7wTz7(>ni1jtwjop-J3#L(6;!sF(z!SgZNewC%*EGKyF`Gm$F+PCyT&>e
z0x0#PISFdFeqtHSFF5s}J0sa~=(*M9MBs#3Nk3I(CY~9Fb(lZ45;JC7ei*iTduWl(
zqIOjBL^+YWK?H#l610shMKQP3GNRHB3vAHbn8=Msa)VEk1oT%&O;(j2Sd5q(zRs=D
z$zBSs*t&gEs$1cCYU?(bw=#`<cxCn($>kZ7bY3&lg(JpPX!3xn=dn`(W^)H^(prR_
zer)SDtigE^gO2^eXt#sLK8o_4eBaNS*j08_mb0$%{F;wbCeTN=*J>34gfmD(_9>Iq
z!0WxrB(0qR`H5Q@{p#YO52lYw*31rr&a$kH@_U^cBPs7iVpz{NHx(7h?+r!=cmLg$
z5)zk$A1~AmF1&H~dE`)Z6QK9zlV@LeSr!s>QT8P*+i5Gh<1CnT>=Azv4$m>IVQx2C
zl(=S|w1%#IatyH+AO*HFV-A>`xKm@d0)0edc|%9$7?r#%VRC0hoTeVUHZscM4X&s+
z#QuyW1Va&yV{_A}S)i({_?h|JzIMhjEOB~>Rkp91J6WDm2u94C%i#7M{gQ{yNCc{$
zVY5(uaPr38Eh<@)^+cl28$~;zqrDw)J;Ry6suZXt0~Z=h!WXivPby=MbFz@)9X<^b
z@~p7TgY1D;dxOUJ^8D&S6>sK-@kX`btY3aVwgeB01nXcUv#fXB=g3ywXfSONdly~2
zxY@vmP$J~+mM0}AXK_^F#*}9S1`#qCb?Y{10g53F1|EUklIJ-T<p6IO9#*}#u3TeS
zz)DN!#*|9vWYXzFp&-wSpBaeIsmGSn14VsXTN_zg`JY0{rxh0_EgUS?Ogb^S>+2sP
z#WdAh4h|d#m3i?q@YVLG8ZLal{UQ;T>3bPR`GNIu*zX=a(>X%nI9&0(5v){(eoUQp
zao_=YzQl&;n8};s@|*vqP||=P2zxE7@yL0@`}qqq7p5Q5W`@Yr{jakCEUuAZ8Xj&?
zM=U<WR)~@Y@#<s`!AVlK8;TI|LK9#!BQ{rcjVMA=l^8lBcv!LH)1pM8M~@IU4FSgs
zs`Zs|MQs>w8s~-bMmFsEsoWM&xQI@*K6Vv7!;K&BIa%R*=pGLW-M$VEwLRq;v6H0d
zToV=`i?v7oSQyLcJWUdj@6&B3#X*mD(x;O-FwISzC?F6_FY&#-0n()91HZl`T;I0S
zi8mFOYQAiZP^EOD<!D&0y*;!uvN`zlX`L=678VxhdCb{$J5k}8VKZU50(UMhDambU
z;B3RAmmzi|h}^bJi?yBBgFmjHC1D)u<3J~)3Qmeu^M-UICd)fn9aD~g?DIH5K4(pP
z%bwWVk=opt5{~ZQ7CRDS@*O!<%~Gmh5wVKiNygYFvI_L}?Zs9p?QVFA{C7|Yah21_
z{M`Au)!8W_slkhIL}q7a2jP;hUc>4ReRj#>lA0LZgvB|v6F}d02Gp$hVo2}K$;nBO
zpRusJX+PsI64FEM90WE&eF{#L?E46LIZ*W2PL$|@c2s8oHXa@@K1SY%{bETA4h_(3
zp`)cW1}D)n0uCFptP^|3zR4xp&i=#-Tc`2uo3CJvBIhdsM}pdEWn<`8*H(tUc=p}6
zA;m-Ca*dHchdvycV#$c5U*H>_A0XO|!Lc$QaMZKConX%j=Da&6LPOIQYwuI*R=wsn
z#!O`!yle->#KtfLB`mOob$P$Nic1Q`Y^MsqwO*Rg8MhAZxctQPVfS^=GEbXYYTkp=
zCaa<BK&X=9E;yL&LO8>*Kfez_RUBw_(;q{(kJ1+QrNGCu21_DQ-|4GTxtv`!V{(!K
zJ&l~0toWUUM7If^F!@Z|fh83R&pb0hDDqn>+ibbcKHabHlw3NjosYmc)BDI1Nm!|g
zDx{{G8RN!YRtRm8(_2!>`TR__?<PQ%M0yli$JDR*^q|}R6Bd=vhrxOUNjS-y(B#oj
z)x`08C(68mYGLzwu0EOzLD`+iN3wc1z0`>dgD+ahb;cd}cP(HCBs>NG4zUwll~X?T
z;U85hRPkjz!xO|cGa#mimfnMrB>Y0V(lqo<_7TKNyNGLS4>kB8&j~)CuFOiO?W!D`
zd=l03BSD+D?7jgMLM^n<#1$*35~XcbB^;;F{Z}b8Grs?gAz#RIu1lhVn=2B^hOl8I
zn~iTDO3F6Zh;j`hc)b*JS#SkfGo^S(Rtl}pK_axn0UWF!4GpTO4gwDcxK3GX>&In2
zo|cSDy2euXeJ`0Gmg7g=GY-vp@G`DgBrZuN9krpGVlw$f8OG*fFt|%Uz0Xg&ynpwB
zXWoOc(~x|F%aiVHL)xzP9iF$LUcLhl6;!oTr&7L*+ipSi)TmjzayTdM#}nI=0Q)g$
z$RS<jO7aa+4{c_b=kakZz}x|Y<uw3LG3=A4P$D1A_w091B|O9@1^)-gP)X~~SL_`%
zbQs1gwS}|BNhD>H*bP6I7&SWDcLo2U-}F_r-<E1A#c&y)94B4N7@o>8xj3vu6;tdl
z3eSYp*>6oOS5GJ^CY1sC16l?sv5cP7N7+9#hMsPM#XPq#8Ikl!k$e5%%(E{gwf3qF
z<HLOOWurR+4libAG2(p6g3YF*-=6T&GfH3~^Y79paQa<Wa9<DCCNd2JRvMNVM%@t0
zI8?PU&No+A-{_?p_V}$*_eZqN#Ls-b!mv+w_<eBJlOEPerTI|C3uCrJ>GRf#cJY&!
zn0~;Z>3I<@Mf~`JbB5l#u7vG_lOC~HlduI->KC~@c$@x%H7xv&P%Dm;?XbMe9U?M8
zlDER1kC(f_Lmn}F-}^aYu?M0;#A=SgbHKwNhTO!T<#p*ybp$W>ktY5zMW;<yq!0{X
zDQ&FMN@=PD2ZBYw#(xMDD+wi-tn!l6(PEVvJo*vOvX88!5fIq+tMm{FmkJ2fb9~+Z
zy`)utAOO3Xu(I-~FAFXIK5u+Gk<yt3qtYHXC)ER_{h&Qbzb8<#QM{@uA4$1k;J_eG
z(oEHJL02i;l0>lE;a2=c<qNg(NuY{~{<#;~D->}yfwtV>LUCy1`0c@)rx|-Ax`eH@
zZ_paB3*k-6h$T<PUVDDv6y!{7p*&^8;NnVrEa#P}>NJpX&i#nc=}%8$rhwxJ$7QYE
z?Cpe!x}8V^(=>ha!`z|6k)HGrMq7=OSyz+IccxH=O=oUwZ15x=j%a2k%T1ec^wrU?
zgtJcT>C3fMhGn}ga?~e6GagnG>Atr45A`t#f4rOy{APbcuBbbnEKI7RvOa22IO`*=
zO)4SD;VzeQ{o<M{&25Ru$!s;bebr{^81gmDW=u6R*&*R)U&>0JBz%f^lF$3vVWsia
z;=S$Jo!PSstE)4U(z3GK*KZb=qsqPNOHI(P%yy!xKkuz2nKjI!IDJXh(~1%JZCVSr
zyqrs5*SkBxYiajAH`jT6ux%}nju{L2K!VPfB6Ae^PRC)}51jPyJ~g^J6ZT&m2*I3z
zk&9m9fWsOV<T<I$d(&Cgi=o;NLc=)|h#xgPEJ35_dV^SuCaCsy0Zt6K#^%zvOr#VQ
zBSRbsZ;|L0un74x^J7Lg-UMi{<bBG9CF9Ots2|`J-#Qk>y@Yw~nyc7C#o#YJ3(o$J
z+P*R@s<4X|l|~vt8YGocT2e{`5fGGa>Fy3iN)Ql`l2SoHQo2D(T3UK==pK;nyZOHR
z^Tv<+-1!3xGjq;)&wh8Tz1E(2H#UFvsgVkc(0QNOe6*iVj{VK!TQey`!y`fTc9k(0
z{kjCU*G^g9Q@$;3Y&qTq<&t{-xR&qP1N-{Hu?OenQk=Z7c!>)m9`$~T{TzmVTg$8-
zIvSEy!scrZH+(N5GP4FhE3ngG93*~kcZ`+@Hw=)QZPb^RZj|(bfhoqnZqkR$E4mnF
zOR5_e+Ulo#$4h*I>hn2#=97|z@;~?R=<Y$cM2}Ogrx;f`!SuEBal*WDaY4O4#>BEX
z3g?5U9!!a}+O?p^*d69mjHfU8V1lLW=koW3fOH(@c1L%bBiZ;ST0c(VGz(7VW8;zB
zw@(v1%P4i~b&$~(tTQt#@YBW&Pg#E?$FvPRwNOQ&6!OezXYFSr(Um*K@D&O>P*%TQ
zH}Pd+_4&n78rA{g4ajv6*hhNOG85ZV-7_wC>VAML#8K>(<v@>CIhohvv&PE3^_!l<
zquDS!T0$=ot&!*vIAx|Eby95sgUMdc--wS>kNaT3J4DY)Do1T8{?Lcgm!FyC=?6Xi
z()Zo&zaN*S{3h+L6?<I_-iG!>;OR&Hk9`!N_6p0`8D`C#l*^5qHB+SOW~Hd|Kp9j-
z{3X2@&lz#9239jrQt|NEoS`w01{jbt6aKyJtL{0S_?nq4A?3}G;>MG;ws)imTiLQm
z9EI4@($f0bs8FArv2}fJC&|71iQ)!-h`x)iA`Z2y*^rzX6KfXg#HXZiuqMpdQ;=f6
zW>x@(8KBoB<byu=R0K0n{Q~N|HT<tS&TOKmb@>>vx_`o@Tj@y}v2!MPB|N;M0QkwO
zKsRn=WaRz(n><yS6Bv&*iu6{^IQvy@We&2JYFzt}mZrZ;5p)&0tDx`&)+aEy=^r3%
z1*d?R-%!jX#;7m^aMe`?c?}>X=uW_;<Rfgeq>}#ul;ypYY*CLhrrN^$jadQmqBTI8
zg^*%!VIa{w{G;pe4hQQ|f?3Y+UDkv_OZ2~H!dyJ}@9?UBQes76X;LjfGf}7Bv!pW&
zk=eFUc*QPn!L(3L6RB{Te{77TA?S#fykKWdXg12t&Fuv30;zIr7Et<m8jqy4gwa62
zYtnKJ=qVhWoO|YY>6%;K4HBZFr{59{{_*O)J^2U#Iyq_7Qxo$3-9aFiGIFL4TKb|I
z4(r9-(?eg$Ox(fhuld~88~S7EN3#(TCBY*(x!|Nnp!15&f(|y$lqfJgU%Xg&Rvoew
z+m|AQSGSIIt^pQ!!*tkkyeN8lLct4w+isEmHsx)eB`I6(2<#azT}CRg4udSh!2j6n
z0g`)jHxI6jOG6&pCI*sPR+=e+ls;C@#M_aNc~I{ST5iahVE)ls3-stH0^ufTUh9%v
zzYeed-M3mjMhMtVl05?MA*^YKcW=wmCmd-1RPQB;6JmWXEmhN!#V*M3`T|z~_1NEC
zrCu&Oy>Zb89`zq-j4E_^yqMDIJpch6xNHed+(aXkyj$Io`r!i_ihzb{h`<v1ApoQT
zdeeZzmbC0+!J7>AFhJdlV6?3zc+ID6)1S%j?I%$)#So%}(R6Dl`x*Om)Uuk8xHvlA
z_d%LP*GTOX2MOiXi@l`c|Hx%K#;m}5`=$Y_dB2>zypYvU+iA=#I|+3g-ehQpToLUZ
z1m}*e$Vo`97n8ck^johz5E6-HhP#LZOY?@Ic{!P$#k^k>mptMdh{@5~-p@O&J`i$c
zr~Vq1#;eM+5pu4sBHq5FA5ILc2Vl%PR5COUFL}icA{{*+stmXkNO2y8h{FkI4h*Oo
zf9K%mCwr5UG=aeJ@tg<H0T&mS#ASHQlE1Vuu=DsDQT*K+<|Rx@A^X%s^=ax6g8M0b
ztE;PsAkyC58-u_y&M&G}TsT8(FhKd49MlKjQy_tlhMqV;;^;ZtF~f#(>3^=Ui{lrI
zO&9ycupbS^5q;_6aJu?{oEiA;v~O=aKJe*a4Uifc5k&KDFc@FpK{GGYUbrKD2lgv~
ziFkFmUo$hITd9GXx<=<eWXVrJ$s6R|8>tMSI;akqH-3VEu?`xFEk$3UPxw$+SWBy#
z!obrQ?jE8%oBBl!K$W_j<i<Wi=DkV4DGyW72Y4`E(|{5L6-j)<mi%k`CvYCd6wrhs
zdq`tlqJ^%)e4zH0I|z)l)W;ODEv5%sH)zJb>0r_Z3b6d?C;uqG9aTX60+2jjel#@x
z{XxzWv9S<IE2^>Z=tGw&i8gZ0Rm3>09=t=iZR@pd=knIwhKR5*tt$JOpKoroD_zGL
zsPjO<XXra~@^6p(!3RLGf(RNe1)c(6``%^2n6!0(<AsKX`j8UPRo&33u-@<G>;Lhv
z%pcT@Sb->G*~gD@^QVCUKLeni@prt}$~mMpbD@dvYQEI#sTpo3PB65F;O{n4FC3FI
zE09d!TThg-q}*}rLccW4IWoi^T`GQhd{?oTCamCsK`arQpu@~0Y+3dosey1zHE`-$
zV5`^nFUT`)N_6+^k)<+=dR3Dph97)pMr`Jn)>xL3sR4z$6AXcTZ+aF*TvtAHhRKSu
zKX762WS?phS;6PBHp=1Nk-Sd()56>cDy()NRD(wse%en7qBjj5;8=nK%oKN!R3hUX
zwi=c9kY_6)elqXqy9xf{)&gn=9n*cNT1UCV<lAP42%!FYNHonuJc|$50qsxZS$>Bv
z&SBE|%D9!-_cE^@FE6jedW`Om7ZI>Fb3VPLoEGNhxOi;Uaq>cb4?{=;%nkkpGzHbs
z0b3lA%7t&|9(G2HmvGfm<KAq-I<PHxy7C$CG)9&C>Jk)&Vp*OB>Ejeg)Pj8l2ZPEe
zBT4J;AilMws1eAAAn}Mxa*So(1oRmY=av?~opy@E`rDo|Dpm@y&@SdEhy!o|0R#w8
z*=)wP3WOOH2OxtfxQ}@qR!YDSUVHQm3$lZtK-llZ)!u&VTf6njSZsX!KBx^MY{jSH
zT`eYF(Qq*$X&~D%gBW5^oFU;?n`B6g`FVKNA2c%H=;wwmPFYO5Q&Ur+kD#Yh9Rj)t
zKy!6UOaOU|W#$4-0<?ThT2@_auK=s+>Z91Ifq_?m(w9QVRNuK8(>UMR;2%POpChSn
z37Xeuz*}l8cf@ytj#Ntoy^rMz0%GevG{9q1-s79yzkP-C-@qaW!%0j}H_*|^+ex9=
zf+hQTuNYwaFdPaB3X98t%^8nJ>Y5a2+bKaoM1sa|IVWz(sXqFkC?$s>RiZ=TrfUNd
z4W0tK=g&FREZU#TVcfzpa{3|L-2CE^>30U7PNGi#S^BE%;{lz(>x1dDhK$Up1!h!(
z3(D`Xp=0r2vZ2q4Uk8}VIyyS2piEVECnqN$_^+oqg6JTZ0_7&d8i$&x8?!m&XtP<u
zLTpGCI6@m?^FS)Wz8on`^E*`UK^?XJID{uR(Nrx?PEH^aB5Z83i;C=Dy_yfzbAM65
zLu$C&^!f8=kCq+J`rh7NpnuqhuA8YQNnZE0Ni$e)nTuH7pG>2J;ROaw{xgt9h@Ok_
zCZ91P+h%GAX+_BVw3=k6OBN3f+yzt40l~K~xIe?LkNpMmkTB{ex0MUV;+|x|jmyr?
z-U=qANLGU8VS{^93$$w)f}CAkpr`N6NRzBo`QwM@*^!fxk<noS!=JplwY6{}@ADJ4
zLn@KEqI*Z!-c`D}Gf7deiSmkzQ3lnBLqpD98iJh6%(v|(SDLqqii)pa=fIn`G$=?(
zNdaD|g$LpYeqLT{E32KoJu*c3K7=`6i6#he!x6jx{soCA)z*sCTZdPqBcI-Iu9=1=
zZe?Yqvv6u<rC^Y%kB?7I<8$c2sH+bH{{ZB5(B}JtK!7!T`t<3ZhC7HR_^CD*dlJDC
zNhB!g>1Y33J${*&#l^;MF9+<*wS)M}u*tt~eg~v-7}urT-7Bjrf)u|&6H!o5ker_C
z?~W(jQ#(65XwS9s{4}(3aVcBb^n6I8Y+#VC<z>vv%ZvR5ZprJ{Xkz||fq~agm!@82
z*$Rp>G2Q6cG-wS|Vx8x)gM9;|AujiL2}0F%bh=GSwyU3*nB0-hrGDjfloAb^w)gMF
zK#b$BRy_{4VQp;<XbB#TRSgXdkWx`^qScdtP;MLMO+w?=GYC!I7uH=_`8UM1Z6QvP
zn{mm>$J<Dmk`_uTDyXJ{f`Vk_<*O<xaHvHhBO_tq;Q^%<2srll_i1TqTZrLRSwK#Q
zRg2IyJU=^YJU^I%?p9k{8!RodKQSTUAYD)cT63_r&`Pg$8he5ptfs5$14=8f*7$^k
zxmzl?^K)~b?6m_G_2T>#BB4py1ZWq*7^?Anpgrs1aR@mykl$zPz0_|<Fkqiy2nq>#
zoFHmFw5U*RpHfnI3>$q9xAc*TZ-w&Vgp|L7nuH-pM_0F}r^ibyz1riXW$d8?oQ19J
zbeV+;+-gvIhSFfIT4juXYmiiRbF0Y6m<866Hs>GUv19{z#`nmPJmMBLH8l;5nDt05
zoEVr1k52iMf_5;gwUw3Wrhq1p-2MK|4NM7P;c@8pj*T5cqnbmb;7SjZ*wf&MgywAv
zh$*2B2*R;eqx+y?GIwyeVhBYDa2J04+G2G7>hulj#}NpB==?x;)!o&VnT-v)^mLN`
zVt_J2(=IA13OKT0g}~o``|;xlIO(_W2n@V^GenDmQV@Eye0*@Jpvjzfu6c1!0Nw^z
z0OZ5JUcG~0pO>2}3<LpSIM~@0R#jCcCu>84=zsR#9g)%{jY`neIlK#wkxeE2g}VCg
z_I47aBOontI&JOivxDpdX!!h8b_<;`Im&5QG{MJ@e?k8M(EYaX;8w;Ua9kuLB%sSU
z1#JZEAjH&^?z3lhKwN`szt7Ka>);R_7M8?o7M_@>2`TRT_q`zFysDX2SD{bF_Q%f4
z)XWUFgS2!DR6CBk+pyvX2M0$@(4zb#?sE{&p?Rg@1xcNymDR+=#6wY09^J|U*w^>(
zt4>b|e&|||dW#A7ofA|=%|^XYyFgqM5fQ;=1l=a&6c28{2YdeT;RC3=_=SYVMn?1+
zeYF{gKrjrx4wfA5LiIY*QCl3A3f5X*UtdAtU3j>nj7%#u<w5odn+rPeU%-dv=H8~H
z>{@ibas4{5h=uH@K?+F+hk?Y)&~UaWR9#ICq+2rB-~J<j>}YR?vjrE*FD7;+69MfT
z)OOH2Z*X4h2BHihjj)lmwGL!@APEJ(1P&9{qA%$|<~lO>MP5_WC3H|9K;s#xn^WWC
zN5{u@GqwBBBn8kN2&O<2XNv2BNihZn20%3f02`i%H8g;O5K5jg@bEM_&HpM2mBD__
z{t+%Ck;f3|q2F?HfP;+mVg2>%7XiJ*8IY|T{Lj6iM1T~JNE#d~$Rog}lO_DTL2>u!
z(QB<jnrZ36+*}CO5UIeALDb*g-`8*S^|H0yfVvneGZ`6~Ltt)0bjZqL1}z^XJr%XJ
zCs5b@C@uZBy-l><NG3cDmIq!rTRAQ7%a<4!sz6IC4l-+~p&!f3*VWVvuDVi1f^bv_
zXd~cT;Fg1cduzNTv!o=Z-Go-m^T#@}x33SZeZtlqs7>AR9BFB3;E}F)7Zf}OFan}{
zy=PUtjjerl_7c))XraPj+z_In!9i{4ZpU@O4VT46Qb!M1^6LozMZ#la962|4=TEGk
zukZHZJ5l`8LzKLOPVmXe3AlVfmuMQR9!Oh)^_%ie^z}u`-m$W@WC*ggwXKc8j#q4-
zQPkAb#KXfI%y8>T;KIbjJZim7O${xwAk{~Y(17kTIN0oQV?)Hk!Xh=Qu&@yHuwb~-
zAPcpJVL9F1-JV@7e$%h7-EjPdgO6WPUQPo%FIdDnI;d)1B((#p?R0;g?Tk-I=n&N4
z!cJ#kc9Bp(OA!Hm2AuK?<a5En?LhvkrBn7Cm8s`7Iz8BQad*#FWe1ujG`ECl{rvop
zc-zf*w!oCrCu9>j*EKMZ5EQ)H$^HGs+1VisVUVhS&&}mDY}DgV`2PKFu(Woadm+$U
zAl?RI;MLOE10Tb)IFbnZ!k-{@OHWU)syc-K{DLwfFE9KuxGue@2xh-VXYg5ET_?<<
zlsIWL?TRf5J<ai(F1H$%z5JMvkPgxopkc##6H9}m2LZjOKJ@ILjkRu0*O2l!f+MsU
zE0V(o><vZ-bPNtYU}wj6@LL}%?hs%AvM@N7?vHnmxWMUn@~wIc2QtVm$6O6U2EAWf
zbdFN8^X^g~*u1u`F5CkMx?q`tDUqVqu&s~GpaVq4F%1R&4mF(O;^HDKjJTNCHYz?X
z&Cb;HV|@JL))w9E+gb(&QH0&9-(SH0U0ik{GYAa{aTnLfQLt<I^%bmUy9@ZUo}Qiq
zt9uY;uid~5Lwv9Xj>8oX!@C+volGG<v7qguBjAXU_&EdXp#|$}v82y|?O5yR$Oy3f
z4Gj(J*BN4-AP|T*Z{C2>p-!Yll7$=$AXliGz3{&*Z1Dxr{aNN*<iRE@-IgLII=b+~
zhwmyDj*dJr-!kg`zS`ivN|}QY(nB+vgC~Cegu&lE)CSMnxO$<y^<2RjlLJpS7#Z|9
z^J-3@x-}Ut1`0-f!hWUTNUf`@b9`vZOPR<rcu)-uJjcYGf`a$DfaOJ+@Fs(^h@jxv
z@-6`4nSQmFNE&xRGd3d=Q>L=Q_xyaQHNl<-d1hy4o7RyKY(*=qh@^p3*ZOeoV(;|b
z%cM)IlSA<G+Y%HFt4gk4!NlmpDmW&fxXYR61sei6IjR@$^7D&(w;X7z)xuvULHrd(
z(ShR5bMKPlCSq^f=xs%XfI%{`^i?|ST@5AX+%v1@K#Z2>+&~n87=eY3{#P>Ct-Y-c
zoiS*td8MzjusZlyT%NuXWZ$EjhR8u|-sJc47~4&l1T<mvVY*2`B$Jg1tE45Cb`=q4
zVZo4Gotu+pjoVQPA&N~*q~|k#uN8a@=_Q~TEy2`ORE>PkSmPA##DL2Y6?XswiH#lF
zcZ7iK@57j6nZGCAcCQy*N=rv4++G{<MaUE(kuc$sB_msE@-s5t;^z+aXmlWyQ&dn$
zPDzQV+XgR>JWtEcW=;S34zZyUJc{ArI2*-oZ_J8)HOk>Dy_TyL_z+LFC#-S7IS;)&
zJ+bo>tAhayI67xhghFHcZ*z;JoDdqxyK>kkMIGQkU`|Cf#8lu`IZiE#BOYpjCX`vj
z-J$=6C!5&evK@pDly||>(o%J`Fy`9_kkJ&O`{~Nc25C(Yqrb`5JPdo8gmJwYNF&?Z
zH$ybt{QS;g1jfqR+R*AA^hn)s8BdttoWkQL0S8q4rq)njZ);~Kmw2qwnn#8!$Day|
zeop$k%d~VQo69y6I|817kVM(wL6Df7Ov(KV5?sieyuH1@UY>*UvJIC_L|#X0UbMOH
zlZSs6q#dr}<e{xz%chA%giRTu9?At9h=(kv!1#itd=X8)bopn2hmepE9IaTR?46ix
zJv|i_6=t<52$gjntT0mfzMvqmrXgqWRKAw1l+JHOLXylLv|jG07+g~Fz<(g})H6cR
zLCzx>o6*C=Bandz?!W$cjDV1kgM$Nw5H2|p(Y8vT78Kt1;wPu4K~Ec8cK`l;BJtHg
z2JFGDji%Dd%By@{;xRW(h#bng?{h#Bf2a?aeupWq9AT}lu5ObH3W#E_Nk7KL={9&*
zR8<N2Uu+T_`mnRJSJ%~X+`Wr4dr>SaIVCjf%XB7H2hP;x=o;P6mF}Q9)f;GN^zm2U
z5ukhfCgw#;dzwol%q0i5=p0Y?z<?uA->|<xY_qhshRgy!3OVc|hch;+99+n8c76`L
zKgg3Ntq3I3qh5E!$ztgr`8KYEFcnp?S&UCjsegZQR)TT;%?jszq^}zCV{gD3(*-GR
z@Ewp$6~Es(a3q#qff)hU-#mi24;f5Xmjb_f&<F6%TF9F>Z*FXCI4Y21qcf9yAHGg3
z?KwMcRoCi;bf}Af^aSFBv?Y)UzV_>CYva&2<?sze(y}5po6aA!=_{j-9BwUvE4H$>
zK3E@9QB=e|`~@4yDW9~M9=1Jw-f9SxPU}#qtgb?eX=P(GznL|hy|TC%Of&=~7MHhP
zC9yQjsv#Ci{h_SN{;|R5@J^7asp<BE+sMu`&~m+P{;bLlu9IHWoikq1&quW+0($>R
zO4&ak-ywr?$<56TK(qQcUxem^zWPy!d2(wQ{{k-$9^C2ejQ}0vOBo0x)sLp-uI_f!
z-kqI8d2o{S8*P~QRAi*2Zvz;_F5W#$=523oz831@;Xy$clp^f(wR(bum37FnJYeQ-
zZhk&2vEI}+7LoYhF*iu5j^v%(+|nci8n?#?@$g6qHr-7i_l9!Z+&rZJ>#E4Tco9%_
zKuO6I<m%=IrIeE}+*VOhQP#MLyEgLj9pKuSf}oIx3Zm^(LKqU6tlV0EM6?7+0%WWr
z6>_lsc4{GQW!nmYadiAJSp_r~hjfjNjgSXI%m!nTb)bFwAwK?mE`rCg>=t7d4C@eM
zosokS6P6lkzF(UUnM)UWZ@<UDBmesuvOJY^u^QiF-pI#gWo7EmpEuXh4&E8N1;TbH
zCIOEDzG2p&B@pvpzI227jSP1Yk}KjyC78V7>cSNB1WFAxO-%}$e-zSBAa+CIJTs#Y
zxuU-Q47jM_VRgt>4GjE2a)r$(FE4+lcrQM@Nw>b1ikcd#J{`}6#)$1}R6zqk+<<Fz
z)KxUW+YPYM5R`r*CnpDGYD>!;aIYcJgUJez*^{y=K`k&iI{IzU5*&zLwIgHt=3R1n
zdU}#~3#+RN-_-W^_P{+rQIPfRn*wueV4yUV-H?>fjQ|5Rp|SBI{1yi+F!UHECnwtq
zS+Ku=(*w$tADrTgd!z7xo$#LrNTBqtW>buhPfU2hlt+FlE-;p8q>YVDv36N@L4j>s
z81Y;Pe0X_T8MT1*3y5t-Mt`sN@BF-sv^1Q?)qI8G+Q0Mjq*PSlnJ&VqkiUU#5`+#%
zZUEgK76H5`aIJwv|M4SPvXZ^MJq+-0a&hVD?}yR6ZEVj0<Z6eM5e^Uc`ac}rh`N%P
zt_9#%!*Kz)`K!$}|0@#k`f9J;Y_J3f;F+28U`ng5Cs3_HSpuaBa%~OJ&4K;(v6Q4F
zSE`~FT&T_A{gA-GS%P?J69}D9LcrA@CDp?AmiW^7^XJcw4hukwkBAy!I7hXQ`+Mk-
zV`0Ij11Jf8hwwjNIRf4qV5w(cRKYl>rlw$9PtVLirMiRcf>QxWRng&w9FTlDp#J*#
z(+C7`P&-4_31ABNZEI_5SoG#*Mwt7Xu`>j74ydV9z)ZJ*HVRvsIx*y`DhJs{ef=2-
zr<j<S@bU5C-H_h^4g;lG31}MNqdtA2x)=WmpbQwSQKVC$2pAFE&zbRGFbWEWc6`aq
zbV0TRPfqG+YK}n~?&EVcI6_xP=j7-pC^WRmbEjo|+!OK`sB{6g!o$ZeEG{;LI^c?P
zWaWF50f`$}H0&aoP=XSz0)RW9IJHOAQ!?(&M@w2;TAqO%+TY(_v)G`$I59C%(rs-7
zT>Yx6H+WY~^Czync9Z2`XfO)^d^uFZciGwZP_vTU+*>fisLt=SG9_gS^g7pX+<>fl
z+mig9EEUw4&{911y%1e~T#uDLw1PqbJIjwe!VRD_adB~og{*Nk-ur6+f51A!r@Okk
z0x|VVc{!P(#RatYAbSq*A?Ry@ybk0s$%1yD;2UEM3=j)lalr3nED1hta(k~m0vikx
zBrw{{%*<n7e?7fvxHu^7u>-3?4*+sA2~#BS6(ZifgQHp7*=@oQv9i+A1WxV1D(--Q
zOF({LG?K5^CM4J%r;jiad!8vP|B7RT&jJADzax~iIANg7=u|OO3CdPhtFQ@=eEo%5
zV4sTlpWT)$`u?3*nG3KQ--|OZ*nLo-z_)?K#>L?TTH4saks*WZfJj`MNGnAI`8%i~
zzzRTb51BsHym>`Mtk@9QGAk<&U_?iYCfKQksVNb<6HLH3`()3~#q~@{sV$lw)Yp(c
z0q6$Y$v}n#E?o_6Z3QB9nBB!@L5faHP7WgvMjNzXA3)Pevq%rQ<B2W>fEXJqEBM!h
z#6%$h0XbP&y;@h8q6GEX7`hbr<6Td&H;)Dyw4`kX>tORiScQpaQ0mYSuz}`q55rd?
z0OUV_b=|n}4g@<eeTV`3y5W6j@Rb6Z4S6TjCucCn3MRAE=xxE<0nvx>2tcp|^B+KD
zv^h0EapdH*1OI~G!OzIcV_l$NI8ZFKcQrNTiP`}D1YpK~wsEH65uh5_cz91$R6sJ;
z+`q8A3<*>Qpg{m10HVN%JqKfkj>hT%i3nv5UR+vA2Spl`dq4+=m51wwIX+MWC_i~Z
zR8$Vn0ICYh3N`~BgpT|_p<rv44h|imzo3~6TO5w5uceign(7P8Hqe*A^3;3o%-zTY
z#03EC#C8*6^yjLoA7C(%dD+h3EEL$FHo~Le1_w=y?zA%}XIx5;-5>gsL*{UA<-w9c
z1I|pSE+7m58Eq#~!j~%zoG$feTn*vzWmw8Qb)bEP2~RTX&E8syI*NZK^lg7=Km8gS
z+G`aVLo#zInTx`*DT`BD|DoxpRetbFGzntA=SdI0c@%wX{HPX7dYVcLPyXc_qsVzW
z7|K1(_c7ye*spl<us*zhxyf%!ws@wK9A2?P+Mu5FF6vj_FM{hamf>sG&iWBv0L@jQ
z=>MG^yMUvLzPtv<Me7<64XzP4J52rjf*_XHkf<`C)*$g$KL86sRkkF6A)N);yW_(P
z!}2|F%QjXrilevUpTOTg+3<h=_hO!8=yRkn=R2%_&ShR|vxVVFR}*TW$v=a4vNFiT
zoj4V#E9|NQw>?c6)#o^8P;?+uURC8WzePqw(Bfx~^meIZ!VYXV@hd{2Uw;Fxz4{Qb
z!T~!){HSkl;F|=?Fx+&&3w^j!qoY*-vndXyhO%e7^k>7*_V#aMCQ`!KY{^J1PV&Yc
zYm}v}Z7K|zD-H&fmT295o*tlNOAtRckSoVFs|v!YGe~{D5BVz^2|6y{|LV^R-QZ-<
z26i<hI`f1=cca1Gnq2ycVxDGy#al-r<?F36PndCohis4%C~+;G&^S}(K!%eOzP<E6
ze>5N+$Hqc3&;(t*(&W7AzehkCO~ZlEJ}{^zT`fA!_24Vp_`m)65yDvG#m$*!FPf6`
z?fnG*gY0gHNpjisR`uo%HQ`?}@5$O~E?#z&W$jw<Ae4A0EI+N>4L9Hm8`hc#TQK`h
zzvB`SaV~7K9FtOb;?Al#V<|)2hf(!ABjQ<c%%9ETyQgE<g<leWRnCZf<~PadFm9NJ
z@JYn=!E0jn?ss?|Z+(zpJyYYZHdQwDe7PvAk%9kY{odJSMS572Ccf>xw|^2hf2Vgw
z%^oo$)fPrY7hl{^x0~qa6#e>8a%&?w-y>Wrk2U0;U(F~^9tpt~e!v}3&(D>vLRExw
zZf-9_-;3!VQI~Mg>4|$hDEZLNKctzxkJK0=ogn5?aYz-Xy_K;ayHZbzF*7_S6ttC{
zFpO|z-f#bPvuOK!_U<fu96M|NmecHR<L~i@Q)#Z4N_&LUwYI4xOPWnfE<u+hFX*Tx
zjjTwP#Zg;M#l?NYCY_!5)2Up_YDu3TB4*Ww%}+k<<E5Qa{;pSLVZaVFbRJRH?Qr@t
z`BPvRGoSh-y?7zuOl3wc70ZTe!SwcdW78<(MT$11<U>m>C9N9c?{zT;<E?UHF_%se
z1l2a<4guY#9Ug~)(HXRt$EyxXN;Btz4#&+en=Y%Kj|N7cC)aIdTCiCj?fUT1hE7li
z=<%6-Ui#2?I3;mnvBm4Z!_9_4Nx4|uFW`R2VXuOJTX;!zG^J`coFH;gkZQ5Ye^K`7
zv(OA)E&BCL_tuX7B9^;975hxihCs&`&8td(;NmUx2WR_I0@=c8VK4C0TYV`deAL$|
zWRSw}cG7$cr=|D5<r2fUm8j!$MR(dn25jj8SJXmG*1trnoPavLZS%xFg&iTv6Lq?1
z2JEk!{m178g23+%G+LIk)e-*n!a&hMA%lGB&&otCLCQhzP>J{Ob>uBW{CC~l?<FgR
zJW)I2di7Qfd;(JK#vV(ZZx?+7iR!~YeGy;C5u<DFLx??bu(!pNA`1Gv)aH3BXNJY^
zW%5I-{NBFAi#3NYhcl7xP_+gcQU<V}Ne#|y59s?u$$rzE|2PrNbl6TlW-7NOp3>#N
zyR?+MN>wQ-F-L?xm|ovoepx!z5>VpeQJ2?q|D^hUz}bd@VM>k=x}-w+G-95@`OuAi
zA=qzk10A2OgQI3Epo<MVFr~ul^U}2B1PXVW+Ec}YU^uPDXZOB91|PoX4Iej@sn);Y
zJv=OTlmwAe)y=V&1`2zK-5Puq6N2>mU606lO_nX)2cyby_rw*oYR9V*_k|2%#~%}+
z3!#?3Pz`5(@{ufT@;%%5^xLm+tsGB?PWw<p*ShbtjE9;oBUI(+HeY+^)KkF&XKq5q
zAXS@|Mc2A_Mb6AH_SN)kWv%qU8((UK&CZSf_!TB0<JVQh&fmg_r3=qGIYr?>nFpUr
zg!Xhvaizoly1EjDE8?045otGO^sUs)`m7uxMptLQ2}tlFOi!jUhJOj@*Cw^fU^Dvf
zjM7TJKh(OVIdQ^kUbw!!D3lWBk@Krdx#t1pVxHXNu&f2%$!zvbDbhEcm`sVkX;?#v
zq?sX<V%EJKl}|KX6cY}R^ZvJiGFST0MDA5b3zK(mRY_}}R0nuemVe`B=^QrJFbzgy
z{5|g?>VJ?@#+seLOd~QcuF&(smC{OC4!dhS(%m}15rshyb7?tTmI)I#6IIL)7LANm
zOgsEiJvEINuqO(V%5pMG%a=BF*Vs>wO+%lM;lAsbhc`4YlahUxnA$s%drpxrZo!o*
zFmdL-P=R@!QI`K(aAAss_fm7mv*J=hCAV#fqmQ9PO-E}*wAb(w&Yri#UHEJ#Fc(@2
zo~<u7x(~JFpqm}iH@ef9^*_Q^NUrt!7MuL3fAhSJNLoCGwpoLY*I`gt23H#Uxj=bR
z>H&dN@$&w}>Ux5dP(|v222H@(1pyniSof5-Tf)S97o7D%PK}v)+%<<yJvZLOr$p$@
zQE8cr{#iPYwlh#}<~PTacB1I+3-P(0Jkb^B>L|403W(m?cCNyIu+~H8GTgrE{VJfC
zh-g!YVpX@Grk@czkXFc0#!62MBZC3kUoI&$<mr>Um-^pvsG@w>JyVbJ3ws&`HP<ie
zKAeTrz4<*!AQjzC&vePI)zzc45zw_vJlz|qQR`CNz=|Ds>X?Sn>VHbgEkzWlscF7n
zHks7SqSLQry5i^2uz!i56@ThC{#!bg@UWTc>2SQqLv*Qfg#Pl%zo+g!eRAFeeiN^@
z-fU<M&gNMrY2j0r8P&`(+i;ZyYucPzv)9S{*Nv^#Jx<Y9(INR^xj|v5^zfp*Sfsap
z+D@i7A?FA7#jYQ>H6maH7d!BBV+_e_e$7J^JCH$S3-j-r>j(Q7pL!LLs543}iky3R
zryo<x(fD>>T6yj+ByPB??KVux`*qWGD(@Z7@$p6OH#STUgkot+kY+eWQ!*+BIPiFH
zd=sOye%k0eYaTdd7}xkjsmiQ}D?(*!g6fCS(5I)x=+~z?nD5@C6@1<8m*C4SCMI@f
ztLKmBpKikcJD_z*U|uB3UPQYSRpxp@Kj_NzFT33K<;a`!4!WHeJv{oIKMQTYU$f`_
zbhW=+vg*9HT08Xiwnng-CvBLq=bnjHZmu4WW?}~T%^VoX{(Rk-oZGAzSwB=X;=4X@
z(R#P=9_`p2QbQQD{Pni~qBUnQ(Ve?~0_K)p-hVj2YH3$2+1PUHJ&;N;T_u?#*$w>;
z5eZ^0_7OTG1JPjWkHO34taA=5HxK2Wjv>+s#_Pni=ZSjZ;-zeRXN3)aG;a_t?)TMB
zGEI~inr6N1N={li<l`S^qL}R}T1#0tI;P-B$aCh4+LH9FcpQ~ZMyV?gQ@|Cvy4<i~
z>#oQ=4p1;PtN+Usl+{bKk(A5GabkOQ9-SkJsI7Y>X%|F$qiJuwnMGktbp0n$M!rxH
zJ1aw~TWjBjI;LhV{>E|lh7HY5Jr!AU_V3y3`Tmym{fjdjzrPgp|4LJw_U^V@2BH2`
zkn$r6SWQDX-v*%<zGMNJYC8*$p_5K}HnXIabogT#{d)3HTgUG3voby25aTdSrx*nt
zLG!>6NU=5S>KKcQ^*lS9FG-6IOnJ3`E|ad%e%~=29i_x~!xv6DIH@JbuDwUe=&wC;
zN51OwNikEY!{yqyG5K9KZIAvARl5RHs`d4C!@CnA_lq(vZBQ3NB=3|+wlUyK%AiV;
zlJ(gvgC59SEX6)&=_l6Rmv-XHQM0OJSzo^>lF=fFqFfEq2q#a5=~w%B<^xU_UCnhJ
zEJ9TKP32knUXI=jxjD$p6!b*#M&bUwjQD+p&9#_~$$9^Y7ZT!UwgcVY@&h78w#~YK
zeeh{WyA7kr=`y0Ne^i<aj*g89sbPt%AjIj_-`Y?pn3nm((I{vrC&>Pw>JE>9$5We0
zgtp^dYpUMtTFf&W94c!=<g=2){U`3&#9V|PUX%UdeA5ZBZcgtCwU*mndz`khRX(Nk
ziOxArf2ii}@i!<fJj((mqlYbt4jyh+CZJO9g@%O;H@WL%WML`K+$WM`zDMYga+c$8
z!D9H5goIyXZ2K0O8Oy5*>odN82NZb20cT(6VF<g(2FJ4mJ<shPq9D&2oGP2M$nqz$
zb}fBkamEP&8CrK@sqW;Z7DVvO6zN(OC|EbdeK4E5ouzaqPx<&*GR<l8_u8BHjh$<e
zQ^Y0R!^N0(l+3|kO|Ht*N1eKbuiS_)y`0r>0&*!aDzc`1b484fXCocLOUCm0U|8{>
zNN=T;kMrHx=JzeXm6Toy_wr`lBfRt%|EBz!Xs$<9fs9H<Ib@Pe(p<$<{)@H9I6tjQ
zG5O6uETqq(OS|_msrTER#t-PaqC5k!uVsW*y$I}}Ao44<m|IM38WEE)_4u66FO7|p
zeiu>skX~KfhWU=c%(G6@Xr~X4H#T&uCdivs)CZoPv6lYsq!*0z?N{qMA#is#ewS=Z
zC~tL9wlEbba~*T9eZojQj&^SP((3><$><T~ZR0n2E~fDB9fi5<gB>~*s>=Pz&emo{
z2lpGy7&fA<`;6D>sJL{t3?q72s4NGgbAG?{J)L1Ld>0`ecwWCu;gjNA#-!Zv$;0_;
zo+&8mI;U9a>Rb^8tGun^P1f2@NW`C#FxeiQ^gX+sYX5eMmUK_rhqj7RW+*K*1()u%
zWX7^4ZJp)2ajtvFYW*p$yR!3<c<<C)-<D1NNs|M^A9ORU9G5E`Q;(`<6<r#laL|zW
zE=9LhDzWTP$tiD|{Evx`ZOlh^)KJvBXDLV4=iZv<*!6{uBCXP4n<Zbwh`%-)x2}DZ
z%hwq8s9A0gS@;H%cRYxUSNA<unTI8krANS1$x=6&{8r|Pw)&>to0qAw;Ef^_Sbw`Q
z84K>Dd&TUc&W}~aI%zd)@843FB-bCkDpY&H1z1(UYW!5Xv4go_%>FOWB-@Ad`itRD
z_7d=BRg6w=&a=_i<yNk#csQ9Oo`)V6^15oTaEMy`jd1Cp%;)i9-)rl*q!9l|E5x_v
z9N#i9;zzxG8%tr*dnwz2^VA)&!4@sMH;N?KHqdkr?p8WCE-(6lKl$mR!yy^7(K{z3
zcyi`Mn~CL^va8eWYiK+!C2`;3rSQbcoRCJ7dc@0tqyRov<XugNpu`Ra0o*%lC>n7x
z!{bA2cDX6L!82sTEcr3l(Z7{NQl;QNY5R-d86i_P(d;A6I7uq=6c3J!!p((6@fQ){
zL~KIN4sS-gi@8MX)?{rIoHrs<$=$G`cMe4O2LpC8+Cx~`kAJ(ZrPNG_C(sU_@N@IB
zRz0)(aN73V{GZ>4)W3^;H!P;SSw-#Lyp?*>V%=GuYN8(8_;~vHa5Ra^{ivq!($ywe
zs2@As5bxc}?M?NI)}zvAgQR&5ru_Lgcal~LROG`9sYpNN2?S&pPES3ZqtKg91DWGO
zK*2wsk^n^U^V?$XXNe(xycvz5rSp=JxKW<H<O#&F<)cFRz9(M?j~>2hIkQ6Q84d(w
z$Y}J6;jE4ayr|#XuY5{&XRAQqlBUYKb1RO;vvJH#X`(;+P<VfFztg*WIc<h*O`X2`
zqL{JEnri)?U&yHhe0SX7=*_*wlRf8PaqQAl*XJ+y-fQt@SUG2Sxo5DhJos;co;`R~
zFaPff_5X7j|Nr=324q-BxjGx_4{`2dzgt{oj+5kp$u+cC3@GVIlvU$?NbYIGqbXQv
zO(y607T9T2YoWb1jageKL__nWcb|)%rS(x`*AkXL{aH>yh#QH4y%Xo38fU(sMowqI
zK052N*-uJEX!r2ncE#`9#i@jc`7uP+J<-#*oEbAi?2)Tg-s!Y)K}1cwt5s?8B;k{{
zusnHmMSUJsI+n0StIW%9UhGddI7fZmNp#WVp)X_$7DnNHQS)>ie2^D9IWycPFmygc
zUU1$l7dJF!NQRa{cpZ(QYrI%(Z0e?tUtMG21tCwL_8@2AO4!*aDWNnc_6OpOo~a>X
z!{)yS7ea;noz>}oFur}d9ZvB6Pnou-F!%9EOz_)^#_6<vmLry?ntke$t+UOMbw)#K
zDsBan%Lc7K4Wmo7bW)lq#8a_Fi~Ox~`Lw@=(vI{+Wgm=ht4uqwAWsI{Wk&3-^NstR
z3tZbieoe=R78}W0-1%8?hb;c~*EF@e@|S6N`sE65b0!_;z0mv4)oy4jRZIj_DgKmU
zBak?J=;(%Yv<lkDa&FY};&xd5HO6AbyTqZN8uDF)_GGy6@w>f46NR<;%C!9y;d*+E
z3Fi<8-JP$GU%!~8zHY@3+uPX}s_FGTX23CCjXHy=s)~kcLUWb~ErXK17tJ@kv-_SR
zbI=E}m*Kx&oUSw8Vhj>)$N08iWAn>&m`FM|v%X~Lo8lw4Q~$DIGQ&*&T7sfiYynlS
z1p6qtLDy#^bJ9zw-<>>O+M4Qx^$UhF3_<MPZwo#ufbTtTaes8$Y!U=Ex$X4G81wJ0
zbWC=S77uR3_Dd!xaqh%qEn0?I^X_W>S)e_(+8I;cNEL2mMPAf4iTpid^Jz@?w`*(E
zZ$eS#o}O8pqH4+H1}2BX*<49!xeCRslIglIUSBdN#pAzTV8XsOIJ)~OxAJ!-Mv2hc
zBD5AwT-t7?pk5i07H9hIchKgU>C`IeH@cTTZ!b$c6N*@r^f~FZ1bMU!etZ`XeOBUb
z9ht&5uZS4()*GU_?P&GG-n30g22r%1mv7*fkwj&E&O7qe{At~@d86*1bLm!rxFoBx
zb47Uh+L+q*knMc@yL-%e>l=3^pSk_05!Py8tdjMy`-m4g@|wbCr)MnAu<wv2#XKrc
zw0y)>vH#53eeHYkB@3>cZocWfV~5dbb_YvsKDWQa0Wk>%8l<jwi1qEmZlIlne6Hc=
zYm#$$?e4P0!G*<l>qSO}PQGBTTI73lx8uXL(NOK&SfXN9o(6>;gi<in)Phay?B4D7
z%UAk8UfA5NUn*q_9u+f_!Fwf8+i8NOy&h|mvDCy|-uFzvDpG~7$E@+k*4WNMhRhz%
zj8UB|;~k{v>Q@RHd_Gcq%G=&I{?5N?xrs(Z%H!9tNfI0;YSV|@Fhi4&xZRD37WS^d
zUefChtqT0^Ydi2?XcU`<>N@x@W^6KZ`0*$Z?)?AXKRYkc(0_{>aA->|z=6;d<dkKL
Iq>Tdq2QCAnFaQ7m

literal 30523
zcmeGEWmMH&)IJJtL_#S=0Yyp)l}5S|6eI)$NhxXR?vxe@=@bzKr9-;AJGXR;bT^#2
zy`Sg&$9Tv2@_snq4r6%Zy<zX)T64{L&1+tBE`#J{CGf73UPmAhc#<!kDIgG60uczb
zeC#Xm%J;wM`S9(U)eALS1OkT;^#^TW(&G$)ph8GKd-~cbX>HO`=l1Xk_tvJFG*|Ug
zFJHn;s%o6~US2P5HC=h2fkA{reEU^iwJp)MjnW5_S@k{!><<ICuYSV!Csn7qDk_S7
zZN+@G$}@>_cjKjAq@b&z!>-$fI#;KFk%M86ZBIPA5v^LxQ|$Y`H_ZEfY9d78Z)cuS
zao6AlywoNc_+MjeT0?mGnb;$?TkxG!G${UtD0cr1sV?6eqJjSMx^(cJ!<Ud9-tzVS
z|M%Aax6OoEj3dv_%gZ}nVy;+u8*KlVGg+3~VNEry|4&HBsjBW&h5hR9VpGfVMdD%{
zII(XRCx5f@Dje2-XDTsRnoYD4vU^>ep9=4^QJciRzajeKyx8;7W79lo<Nx0SZb%jg
zA`q$zpqyJnUb-%?sePmf@yEMK<d>_t-AIrb6ZWJ1@UtHH$8?DyUd_=JcB6*bI&=am
znyUr|FJ#4Ka7!#DG|19r>gSrVaT>Q~O=6>ec)5UIJs3GVR;?y2oes^_oVB8Qh}A0`
zrd2Kf+D<K_^AUYRm2+T7GxzilP5H0*C^vi{FL~|3gBy3s-ZuJ2jcT{zuuwf2@YtuH
zo3k2x$*@sw^!wL4KSW1|yhgTzS`eS{qr@AcSFhuXYf+S#tsEWJwx)lZ8G4+=vn{SQ
z#ee^PTW2scL(%wpi)F1}y;r@@Ff+p|tbK@yXH)^`T=zCAdsUT{hw|@-Qnm(APDLht
z9Lf2V<hEIEJH0t^cWcu#?J0U!Tq^2GC|?K^v0W0$@To_L#LyTy^DcF*$ZLFjnaJwq
z^r_!C!r(N7d#(XvVJTBn`F}Rkak@EwIL&5+&luAd71f4Aa-H@XEiH)*9ml)((y>c+
zn>8FlQ+y7LS}xa$41a2B{LkBb8@=~ew@Yi^o;t*JpW^z@_4QHm+1rSVa}}wxxjJn8
zR9rFKwJt*1EMT7;N{IXarltG)rCl4W_soGSfvY+E{=rnWeZ)|U1wngnip$oQkn{aX
z$3|%>e~lV4F#VKSC?-{BCUNW!|9?mHOa>PWu(u*{dQgmYRsWA2pPH=6nAP*~5^*(g
z5s@I-!@6=?0z<>}!_Bbg!F`pb{~m~flqts}R!jna#`&)Jqaq{P#GIdv)>n;FB5f~M
z2FoLSzkHc~|Nf(F#>)uI>a8mSgWCt=CL{?zd{7AE6wj!)B$KV~_ik{jU*cWP8_HkE
zC-xO1UQ@G=IO3-`Nf*O})D1XtEr*P?8vU2WKh{-lHs~u<PCO$=AyZT5z(8LGCn+c2
z&6}@E3_1@!QLh&Y2t3feCV%_Jm)J!Q{D8CVR_Wax2Z`tQ%uGxl!&EApp1tHa4!$8;
z#Yp3ulUL5ixxbPok#clw<TL2Z`!E9=yFhhECVd+}U}N0U$H2KgRm|DtJe13Ph?(ux
zEtJ<Zb$(5owbB_(Pjp;s+Z^k+V7-mTAJRfKqwwNI;AAcq+3c(uU$SBZ&6mqJmOnLh
zztW+6dd|HwW;Ry)WF2{{)*+`+N7aCJ%xhgl<JC>qM^g0k*JBn1s)hPiJ_^}iuXOmS
zq4SmEzjuB5^k;Hhr&QXZvfsqL&2UwxXdh!u<jbTMwko2{CT@~~s?&;Z`%julY}^m=
z?j{X&^^-34-XE#}{WqIt>rC^aF)(8|P4<RA-fi`2@w!>|pp)<hW?WJal@5gtmf8LL
zJf0Mlj#-+<U-3C_?PjYL#W1RU)kT8@5x!@4n`HU<voRan4EHl@J&`L{ymhEF^Bk|Z
zsEwf8BFOXC&kxaNmQ#kAIbH>r1RXD@cnW!(m8thbWc?UWxJ|5OGGr`sxWTT^d*cyC
zE@H(?Hm3B-6M6jGXuIAPDW0J*Os}+Z9a4k%h9tXUS+qTl7h_47)W1pD!P?h5S*%aj
z1@S2<5<XdD=3k-Hq1mDc|2lezmhE+G3@-|;^`TtLcj>Z2tzw*)L#Yk6U4@LoM>9UQ
z%RS~}`ET(k-&DKo&ou`#R5>54AptA}FOKG?7^&|2ZaEx35jxh_RvxiEJ?urx7P*;^
zJ$7}C>&|ua5>G_1H<yUwm-9qR-6fd`yU5>ubbsaVv$1HD<vZ^#rHO}hbaZ6S>fNG~
ziDP{$>c?%=oe&)#ztTN$Q#6odZ#`JsL(OwFhkJ8+ns^Lr0bP8CS42gthM?@$9jra9
zuDU;-7q&x<1X<t}q(uIOhZ~cXy3HT@`ubv+-o(Vl2Kf7LAd4j2*ueZr0Tc}>p0cW?
z7VK76v9Bs4OuaX~Y$NO_{@|}+_@lde@BDaF94tNbO*HIyHLuo?rLeG2{g?5p*RN@6
zZ~+CP>b)#$g3PDfsQd8Bu210kqDhJbGEQS<W02Lon_-~3{PXHHDGSX$785HZ716aR
zcfR{?1?8`=U8O@jxq)%Vd$+Dygkp;$Kcu_EV;&D4^*YT^H#8)KiGe|}00B?vS?t6}
zO^x*8(7Dcvh3F&1j$YKOV-caAVbS3YAFtqK)@rJ8rzeT~;p?IK;fAR1Sef<NQa3{X
z3w9hXBHT+sherp!2C;$l87;TObkuB3b!{p{$bE?E$tw&z0|45??gT$hQ!<xo@^EkY
zI<zQs;yT=`R}6JXZ(RL|*@Jh_|08ZwOD4*5|DtgDFh>=wT8RI_oRzGs=@%j$gnk{`
zm=|H)RbRA+-ozqdw9-07Z_iy&?($R+DA2Z07Ex*{apcc?tcW)B4m5t9t9EtIMKB}0
z8L1F~GLrT4VbqK=>!P>p|8b&C(>?^(c!6?`u2-u!b{%dMI$=28FHA19`y!rSuz7^_
zu|vsT0lIMh1Ey1tvzxrUX%6cH=y{r1&fTNpMK9|Rh=?eHF3h=ub#x|c)v#A6s<rv1
zoKP`+!8gDNt=WtG&ozxJ2w!vq5k{Tx#o#(|$T1pKKWw~%T>hn=L-Xm-#m?!JCkBXc
z+!u8MGbwdXbqJ`?@1bSU#Cy}$^`CAiR+a(4F+fJF>N0t*Lxp%~MSO(AiXha%)b-Bw
z!kxeruw;!IH!HnTt(+%b>41Xbe+gjDfbN$5-IbL>M)WaEj;q9oO*BK1D`a)gXQ&g`
zpB`i>e*a$PLNs3TQ`6$V3^nJzyR0|Ybhq}*22B-L1EU6Ap9=R823Z6RnnA=6*$ecN
znq#_}OA1Q&Z>SjmZv*ygfe$}hNevLrJd(e1713PB_vIEV4QoE{4B`UE4(-Hk2Qt6B
z!q@-%9XCuz&ic!3wPze=nCl$YDT;2EiNz#)k-tE%!R+{&whigo*|u0(X6R{TEkq4=
z@OO8Ylf=5nx;u3URd4?~3#umF$7rFe`*~yQd1&6w&PT1n`Fa8Wk{1PJAZ+V#^Bndk
zHCH|(>_phUV5uXNsc!V4)z`h2?7*^f;3sAjiKIh;>%V-3&Gt^=Y&feM4abT)pXd>$
z{uf-HtkGrkOSVLIyNWv^mc_aYFHoT(hGNyePJe$!T3bIBK?}e75pmu7Ijlx5V;oyR
zMg2rhtp~$F@Q}>zHSegl-5Nk<;9y2kV`D8kX<YZXx1h*XR-!(1ocBH|S4_K>@C(m4
z`j!XfK7C4cunsaY-XVdN-z5EgHrI?JVrI3qYhWOi_lM$(+^c@2FV1R6esAxm1_nOj
zVHO9#Aa)I3L?mF)(Wm~m;a2kF`}YsgTi(AX`-0tu#^{CNL`9b#O!J)-*D^MC8hha(
z%@=3eZ|RwUQvdO>CPlgDw~2@l;fr{6C;oT@=(1F|k0jo{{lX~GX7hz{`pTZxFKjfl
zd{tDE#6r~<H$@|)u3Sk+gnzT|@3&Sqwe&#~VY9MTi@z1<z$2R|<KsmafIqwS_Jlmx
zB^X6i18<a<)6yxvLG<@qBF#dirQRp?JgLKNi-<}al#IB|*tvP-WXGl};rXqiRVKbI
zR6=D?(QWXZ2oEPgs73r<Tzrl+6VDjLC5xg-9z@oy$OHu)zY%S)^%^v88Y@7Zp#ML{
zp=QBM%-popomJ^rixzHIq4?^Rv-o@=g<vk?8ecM_B&OO%T<qasx$UWUQi=J^8(+W`
zP5wD=EIFTVsu<dyoa_YwRjCPv8Z+CSbP7IwbW)z=MY>3G>|*ss@=(<S?T(HiM%GtE
zC_8KyV&a8I`&Djoy~(+U;H9Te&1V&0%w}a|yk!+9<NWmLv5Wi9b~{E!Go~9+$sufd
zGj^H(cYqmiPX%gLh9{A|k-vfdr_{U?|N4{GrauPd_rAn_Q&r>g^Czuv5FG#V^wx~%
z<pnmrxbD~Ai@*X=q)f|U@%qm0TdvZ!Zg+eyn&+yF{)!n5*KuBA`V8fz+!qj#$s;fu
zvwB}Ie}kOdZJ(acuC6)QBsTGXcp36x%FOX{f6?;SM2p|dk~?Aq(RXlrwzinW0%Dq*
z-;OeW%b?~a884BvwyqZoNcix<YVK-<qsspwW(o?dZ{M$3;UdenGifB>y<;PmUTF!z
zD=I>_v1M~&P2i|!@1^3o6U)+?o-S4R5e>cXtKa`w{73<{&{?1JfthKgdllpS*|FDs
zF;SO8uAX?kd%_g=xW6hfbftuS#68<hS>c{?avGrN`keEB4vLm$CM&N43qJ{CjP_c4
zds}yNrV_c(`OsMh!_j)_<r{a!yw<+d4A(<uiwV_r!dsmGpqK@qSW?n}jg8#<p_pi2
zF@vb+)`YF`#2uC@FIwrXdMp!?PsZ3e(J#Te%IY6}{FoTGI7JxWzU}LcZoQgwhv&v}
zFEz*Q{{GyQXd1~FgJLPEUta(D+~$<i(JWnSs5yF1Oq9;<(q*Ne%wXjUrGKSM6(i4`
zO2;=}(=M<0$HX@3iyHs<ZSdgta}<PBnROB;3B*rV>|@W!h#*2Knj8E~VuOrH{yA3-
zwS*`E<p#dM_*V?^kcO6!vD(@Wf&s{GhyREwlmYykSC<ydFs}>@=&;VDTsHaR{h55T
z9&wZDD+&ypkdbn_^Wp*haaSkqkYH6)+`Y>xUmN&3<lnN92kUXdlyxdz9-CTSpF<p{
zo6D?+-8CX;y#1&X#9nfNd()}4<kik~5Q$lD^HFV;jLR$gTTJ#9=<Q$nY_N@*{@sfM
zMmJRDyxZ2+#-`VzZ(#6K0TbZi?Y6G3e}hS}zu!4VouODPv(uS%aq#nB(tb^DXGh!I
z({8qd=`Vl&{7DzNvKXsPtc`lpLI7oW^kUUw%+Ah^nMiDuuNPu(Rb{gE9a-X!@Hj;U
zg^k+Fi-YxXiD%Dbl6cr?X`j2@MG;vB6|2qajL$A6s*>#jb-X6G$wCt@vG5E=ZRu&2
z3iMig61iDeST@6iFI0aS_csOIYL4}Wit^nEZhJdI)=*#n%E0li2}z)FvPonuJ~`js
zgl(^8xy^T#JS`;;o&CSV!6YoXTZT3^Hmf5!)!KZb*s}}8qpZ$0!O8qAS5mzou(!70
zi6Mrgwwbg`_r?6l`0W0iZq}yoSnI;O770*nvcdNn(Nq$rsWQ}9;a3Pnu?gHybmp48
z<b;l|VPmD^k_ATHUYiWAanCxX<cQMj{mB}2-Y-GvesR3Kw*QiQ(r#mdw{nuX<_=J^
z7qQGLlukO-TNFS7>&GqUD>sLjZ}l@15!#B1HoV8ciHL}}j)Sw@oj97OEsUPyC5FAN
zM@w5WHilOhVs@LA<M#9CC^w6}es{Zve^bEqV6Du09?X7nuztEfsvXRNe&08c-Ad;O
zo#2ia8agJTZs2>_I#+S$vtQE5xqp**ZI`;@)1-{y)R?$NC~khj<Pko;Dt!5Is@i(9
z`kx2?3#6&n8Xs>`f3U20cRPHlgegh3>Np*oKSBJT^RINWY;QF_N5}F<Gpe+294k<*
zuJ8{VEK=|_)C9uKoh%sakC*6vvce=%n)4=gRIe=@%q+;P)cYz76+DlHD74+)h6UZd
z@@!$@%NK-F&V)$d`+vTzxH{6-^-Ds$21BE27L$jJ$0;8^S)S{kNH+CtH0^nCgvI0u
zy7Tt-*sp7Tew=PeEd|&9MO`|Vd3eg^$B%CJ+RTQp%L)3mgqR+c(MU!muJR_FJ%@sL
zsDvqj3s1ZDg5XQDiuT(Ho?q`28(R(3()-t3LQ&QB+{*~x@6s<<R$|eT8@+dzw@2P^
zdL$LvxJ&%oceY-ut+w0t?fm0@iS1z()<n|_9+bPHsO2+X1}|z>`E`XGlp_CrIn@6B
zySRk1|LSREp+OFX(+t5(iRU?vMCcDq(y#Xa2x;O-PGczfatt&1)wPrHX;xdho!wz@
z366hd=F2SQ*=}wFNYZMrZ?}ZChLXd6{VV_5FI^%Y-QeYAM*G#Q;-$-{MTM)p|GkM?
zVlGl{Oge4&!*G`MZuj4qKx4B1EOfdVpf&xs9_tbZH-(#krS{{PD!2y6|G>%{%jDC;
zS|iDUbQSI9zEn=n7+?VZKGME$a!Ngd^xL;SJVP7J|4RH&3Uf}qmPcQZ^*(o=T+NWn
z(xm$O?!O;>c0Xtcpscarp!m#lQ`<8{t;pnv5o(A3miPTTHg-Ozu?`2RJs&wCp-xe~
zMc%AupRt4Be^!3XdV`i0K_~ZF;BCWIM85i4)*l-5U*G;`E<3)xjY%D3;fxpJmKSJH
zIi;DJbpQL@DOWkJc*C>3TjF8j^DRz`M8!C;Ie$fL5C?Gm*V-d%e~k-Y9O)GYi@G6D
zn!*37ucV@3(l=47`^vOe_WFO;Asb`x_xdgFe>v#?S$QQDE&clMEc%HPqBVQ}{bpX8
z8nVe9wb=jt@c(CNm&11{@ZTfn?28kkUxrB!+y13l3c1G(H8H1vz}ZrwrL2^c!gr_{
zMFTN^hQ|#K4sO<*ZV(6^QBhN;Y1eujPPyhCJpqyU{_k(1x>M!0Mwb`o0R)s3goHag
zoh&H=&QFAewNXVYtmMh-!+@_}zs}9gwS?TWCEzveikoc=C>jmGCzk?!BrtG>!Vf-b
z`?(U$(!|6B7KM|}`N=*xpPg!<emg^6RiWGQ&Qf>c?)AH5s4ra57d$^h=73T@>ul`r
za2CVmV_x2J`_+*uDOi!O@k^p0h!RTk(cDS+9m~q~lBsIf!`0Ev*#OFm_0nk`ai&L)
zoc8)92q=Zl8VQ8=Hm7SldjEd_m##Gec8}QT?MQF^{sKO+SNO84t*vi$AAyV6n3Y%4
z;V3e0STyM{ZZ7Du$K$g1qu+R;%mDU*jygw`=7qemu`xl{gGZM#GBOK^aih7KrlsS{
zy(yb^d`!f_&APK?*09{B45fZOt$dyOKmBQ^dn>lbJBtz&&I_<dtqQxxt>HoczP!D?
zJNS=}k7s9R$;4207s~uVvHRtzPtRvT*Ujp~o+REP*TW4ge6l-&E*!1lo3)oUXSI(Q
zP<p=1aF}#+Q<F?0*GsEcs1s3mu1mqHBeKZtxHdL>-T>S5%p6une*2vDfUP%?JFGg~
z;CSe#VrLA~>R3VAn8Hob7lO}^6JLLdg9WBlZH$+0LPpfn(<5Qlc$EJdKG3s&USWS?
zf3%~grw1Pg=a&3!)In%0QxpxZ2-yAU>xyT`AP$0c|E$e?(FKBx{kf5<s%nO8^0Vjw
z)XSsr@-UMY@F?FKmZvFA6FF~6{|Y(&>ASy4;(tT*?ACc8hV}F%&C=!BvM@c7&kfPG
z%TzVw{sg?ltXc8;IF-`#te`^pChECq^c!p`LhddXXUBeierqYpS@s~S%(sS-ykUi(
ze-p}_<6ZixnDKZ2BrYy4J3IRmwS-o=&0<=Z5TW)H)NfE^er<J|YYGy+JS!?Ib6RNo
zq?j&qmmNHpO>d)eQ)t_d8YRmo(z|r^_xBt1ehvu@t+87f7|v4h@$u<R;-#+9C$+mg
zKV-<WwY7!QKGY?;ecQ=$ez-tAr$P67dOEZEufS@p8aHQI*-)=9D4{D7viiZeB~IWz
zoV^F6@D5=X?V2jnQD&CrlTFXd(^g8)N7HX@Yzn!*($LYtxtBB~CUTkYh%*)oobL_7
zT6ewCu@C}#ePVEQ<2CN5d9@cVKmpzwgwwQ3N7i1P?e^AQoZ!@TL^FUvaAx4=6*h|>
zxOJ5bi`AbqubUMinYC+@A|f6zK9GQDG)i$Vfz7GpYLr97LwpPAHU*wk)xB_Waq;jF
za$R?^?nx2$Z25RsO-*eI8}+P%WPY;Hj~_q6V$TNUg!jd%)M9Y1Uk7FpM#kG9E>U&!
zyVN}{vlp_mJslkh!k!nfhQA?AYj%9WA3kVItCd;JLUZF_bJ}yUBieDwIl#54s;cVS
zHyQ+Sut8h+{gsuKjuhDxq4ROeTI-Md@E9@yr!DX=@THTrqFycM#b{ttw@FAyh>3-u
z0YZlUc%y2!C+OBgzGT^h-uB3cy$PJFIi*uzH-pe%)-;JQS!ro$Sy_O*-{rQ;I4o89
zuxB4E0^-}ZtF~M20nxdsL7@dU*nuX?m9Y}@@mq2N)+c)_0|NsXIK=rLqXCp2PHSUq
zmQ8}daWa*%qCAO7NGN%%bmm(^rUDnOc%-DI0YeoP6_NafjA0alE@1xE(Y!<PXO<9i
z!uvyt59L083=E9pcidP*7PW>^UIH!!vc!vL!(%&HY8jcCF?EX7qv=^5J~RS)DG^4o
zFWwic<?@(;fq|BG5w(K#;?bJprNqL*CkS**+%mfr!d4NW%SA@Lpurw}AYzD!jEtrW
zJj`cUj@9;v=QNEOT+sk|P(eXKOY3vv><q<foi{pwnH4$aqsNa6kY>PL$1Ci!xO2y#
zn^b$cQHcPmyI{qm;n=Zax3iA9K2|XQGvo8gTri8n#>BfKB<wScQn&{42msL3J?$kQ
z0wE_@Zd22<?>^D7v9WP+PAdc7i5QgKpm=g?6x{7%fA{X4(#1PK$Me%e?Q$F1=|C(3
ziV1iuK!@YT#No7ON^-KMyu1}yIAd@nWK%H?sH$Ry?Beor?QRc0c!X>QJ3l|b-=st3
z`AoeJgG%nN@%5>|(@|~D_3=^y5etir20t7*X#cEaq;wRzL13yFGrL;A{^SJrhemSV
zWXQztAdOOZlY54ShQ5A%pcdo4S6@U=xcf~qQniNd4sImf%LHCqLno)Qt@rQz8fTBd
z*G6(RPvM(-&d9Xqr63Cnf!DAJ6EpL{z}V(uXKar0D)4#B0#y)lTYB=J2NRR?Sb16Y
z?f|sn2Q-=#cNA(UqrSKdI&L3iNXIgpPuC>tF)mHHZo0!+rpu*R6d{vjmsvng1ET?Z
z5L*eb)-O;6u#b$7|5-)d`Gj8|KfuSe38Ffysw&C+0OH~5j2+*@mkD;@ppMh-v94ue
z0r=m)eOp*qXe&&Lj*gCuY*YvwHZd^2G7>kx2w}wZ_%WJJ5ps;5ICv4(*4*3-UPI||
zbiY*<vF70F?9A&0M(OJ6qLv7Kq$m$&c0JxPs5x|jg;pmekyTD^*zu|4s25vK*FZh6
zvtyGrd;(BYgwxk}!{nTzh_MqQJ}@{~k3B$jnGjKln6(jc2E6S!+MZ`3#u@j3^MUeC
zuXJ2(kKL$;Ovo)xg)z8!2dawk^#g;>7&XQibS!)U2op$w;4f>&E@_0fzI<q!C^j7h
z2dTB5ZyC&xn=Uj6mMbL|T(;vg=Yr<xI9T4Wbeu)A!fv*~@7uR;w;syo*V=?R9jvL1
ze6`~t>)@pB+{x9fWF*GP%5f@p+J4Zfq?|QuQ5qW)^I5=oCzz!MVFln1xx(gS^O#u)
z4hgfqkrDb4tc*pgN>yII&L)Uk7wwL)hrnt&Ku*T@?*%M<_kxat9)8XrU(d!gy4Ty*
zb_D317#-d0LhsgyTt9#SC1zt{S_f{j_VR+_-Rb|%0vs4zT%2yYS$aNNEFE9ZkW2A6
zS;^Sl+3D!)WFqbaJTKc!9$ZO+F2mcmv;KJZPWD&10dOD`3WN)`iE0E>;e=q7kq_n4
zzkSPpwU^BAs8#J!{Oi{qfQnh2%$=I!H!+I<TLS|Mm6H~w<BpqC-Rfpsi6QO*(aU&F
zU#qIlAO#8AEWEI=V6OZ@j^7Gt#*FKYwe>bIu*1#iUN#MQww&;JlYg437@&Zmp<$tT
zCS#}Dp==yIA$w)H4b<weR1fH~F}*2q+i8E;94kYeB%3j5QEN3T`t&K9>3aw>AY_Gu
zS!HEM5U+030O-ldeYNKY_bT7Ef2eo)>sI-cQ5cRLY?U#nrmpS+AOatU(8=R%xsq=&
zS&^O1XmDc(a36_8N~JH{!x<$k>NczdJYXW0e){yOwKbdRo)x7kW_WCDYtIc0X}#>=
z_L^~^9@`;*ta!kr>kg+--v3G9jHY0p&U=kan6&6^Z$BQby|9PO4-d{7cA0CsRugo5
zcD#FfxM_6gk}SJ?yu0kU{yRD<s;Psj^T}b6U$p1uRCUKoDXzkX7X0Qe8N24~*FUX{
zVMpB#FD1Kxy;bR=S@Dobgpx_Vd}$Em4B0Z<2W$`dE{eL1Z2}BCn)4>WH1?hsq;Ebl
z{1L<b{!oMB+Ey|mqUiAOz7q2{h(vBnVpSURw^*t_=U^2u)B%9o-jf4P4N$Wi@<W3)
zmRVC_!>6OLYsq+Povr9hzgWGfs7P8$>cfW*YYvr-v#J$#E0CPVh?;?XLRBaYXXmG>
zbGg?Ncq_{*N!CbRUERT<WL2<JDNEV9ixy(-`}gngUnROFh-zpid0y<MR5hSM@@9to
zpvuUVxDFxQaq3W6ou8lo<42%n=X?nFlzd4{Oiar`Xh>oHEq}}XR=VPPgEFO<MfcMK
zI73{-bYhJqWL7^MlCh~|A$N7>X?w5Is;UVIOC{izEZp3q`{cpRHVbVx%ZAw9?kNGe
zSo;$d9nDJ?3M3BFXM*4S_!S*8e2G`D`XP#D2i`q?{P^wiw@tbw`p<g6s*X{iVL3Gc
zvAsRtYWe%uPk8p6bo$L~Vx5<U%Ig*rW!sQd^*se)ssdCc=Vo=FVsMbr9oUF_I~Y`*
zg$nm}u4C2BTAl3;$^r8U{qR9oF%ZZdijx7{OiXB~#jrnl0#!{}6DhE1V`JmRLp5E)
z-Df<yX$}Z|HAC$c*kM#H1-}C$J^crj?7W(j%vWjRT!}B8`@V~Y@p5u1-3?lVdRgc#
zw_+$6?;M|Vt3|2j0n&&#*w69-GXJg54j%(UmSiMd*~sQw671HF>g;Tpm1|{MIE9QX
zs`+2Cvp-sz1G0Rd5OpAbkQv_VHU#Am*j9YhQI&yAL{c)ylsbaMLH&x(E|ctMf%Y0n
zmX~b&o)=X+>Ob$PnYR-bac^*4mwz6140(VYh!oHy;9i#l(k#Qv*&j{CklBh2>Fi~)
zbQNyLPg+UCyBA-n7b2qG4Rq{@*06J`y6%n@=;fkE^E<2^ZMRY)<0xABIZ!Smgq!f`
z6Piw=KVF#N!JRx+gwWYmqvzQ?Wf_-rojoNo(WobBcDTcuhs<myS2Qesm=w_NabkOy
zqgzDsHNZZHNtVjVo{jSSdz4jWp~-Hi>!nhQon`5Az2^Xu9lO&+B`8@CMG<KYwY2Ff
z3o{u#MMMy>H|R5~OgoaYKYqkLAh)S)Y|x5|jjhE0UZ$);MqMHF{i}dNZ(*Svfq_cD
zUT!b?Xf&{KdUHW#!E&L4*58{|x&^bXaH<_)5>P`VB)=zf?J6%Xha_A`m{6@V8pha}
z?b*KlE7zhf@;T5l!@CJ226ajMaiE+X&iD{i5c_urvc=1MZjCT-27<x(CV#hn)(VCv
zxUgqK?4OS@x*Ys2W@$(l8e2bzin<-@+|0?Ar(Ijq5NRl7>x)Bci_qC+*1S~L&;U}*
ztA)X6ujEW5@%;JY94pG?Xiev_l$T*wpkBPfWgHb32cK(SVPa`I@_}#ziZT>EHuw`l
zu?zWh{<rvRI@h9&TdT82#=1+ktY^d0EY%AqNt9XFHA&8NcU{Vwq-10?3Uo16zo>l1
z=vbO~udUnD*;%>#*)^)OrMPwNge^2Sa&RTvBILgETObiuZsF=bCDnXpfKKIhy6300
ztom0ocAA+)n1=DPvW9oIWCj073dn8!1J$xV3-B}^uDhl}tB_)bO*D%P@eWlnFffQH
zCX8|@LkSxA@_wE@)GC*H9_`q@Pkx}3BIp`3&y%DO`7HzW#bEUKFrW|t7<UFo9zE^I
zmn<i+eKoOQRagx9tbPfXfk=C?vI-6e5CUic)WKTOMiC0PwxznD@XJk0Rtn|ub_^@8
zcrHUtO^ucW+A0vKs)onf)Hn$V34i&9>JNfPzh;{0sKq}l(`&`g7<|W-$l%r`t!U0L
zp=ZdV@#g;Ltnvl3F8O!-E^Yg`m%F{f3>x?y=VmoxNF|GIogca)uMnZ+rjVrGzGWG0
z%GQ`9D~h$jRQ{t`b$M~xVXD>IV^ONY-32@Z53><jWc0IjEnrMiId$}pdZKMoZhLOX
z+?=j~56~xti_DL-JAWI9iHOQ`U497cEe=!I+IuP-4BeZJVvaQp>E&iC>bDt*@9gSI
z687|n4vME9E$XpPQj0HaI_w`$V6a#`z=~DP#c@}uIj=uQv*T(k=rI4g?=l)@jq9RG
zW*U7xtf+Fze(7^uo>p~c*R9tSTDnmWg?V+wIET-R-wme6$Hzy-oy$kdO_m=#j=snB
zS+Qci0~3P#$C*+*OUJ4}R=#r3>V(ST%gFZEvwwG<;b})b*-o5S`sSO(YPC!nIJ!mZ
zN}S|wZ+24xOftK{IlZA`ubsYwQBkrb;5bRao+olNeTt)m|6~^rXj*H?Jr!p>qiU_q
zggE!ti<Z4vf=f_&dF*y`uN8B*<TB{Ut*acW_|(g*tN%?CL;n3ZeDq;e8O@AMnz>-?
z+5+Ehb8+W)0q*8i0hs0*C10<a^}oj>w4f+9pcE0qO<2@WoUW7S>=$W(+}|i0HW75*
z(f2D?(J+iJ7tcOMC(DpnM;2u)QJT(<C)!Q^hS>{-X<0)R+Sb%H`n<<J^)q=vYR){O
zyqC`I@fNv}9oCO3{dXsfIV_Nc%~^_qRRJztoaGj}4(`9&#iCui#M25Y!kl}OEdoFb
z8k^Ej%q%S271zzxNUg*_F0{CtM??Tb$W+)8QVx_WIpIO*SFZ4kNJNtuWLfD&=y@(y
ztajxS)3nO`tWqZSw{0$EP}VF65vQI`HB&M)P+=}%?CfYyV=Ql6eL~VF?O;ENn(mPj
zLtZRFNe?At_AxsdrH+bAS4Z)JToBTvxR7_<e7uA+UM9Ok@$X|x8pBaewm5Qrhu2OS
z4ZCIb$_6=_sjXexv){q$kSsw_0Nu^%{9)_6AlZzf4#ijfU!UVJd&jL3C!WeXL~p9R
z=n~4@nUak7reT00;-m{s@kJzNd>N$$O3ND4)`I%k1~Ng&=_Ee;4*P<}zraM6mzL-?
ztFITom65@1qFeg&$5?Cj2|GJ-{V%F7BQ9RJO?!Y7v?Z(SV5I04rinop*<Kvbq})08
zwL>#2{ODK~-*~dZ-p<a>?~vc_Pu=^S<Rynmp9XQ=>i(yWi+N(m2%5*1P*O)}x=grs
zWfx}R=jo|9b2JzC=;u9VJH5x<OHmW;3aQYhMCnDBng8v)-7ZcxmBT8--KB14KSkxn
zz7@Oe?d|*b?<3~s=aGrJ8*B_^d4FCn^4_>}XAGJliz#@d99s_jUI1^nBzNw#;oohH
zQ665g%U-;B`}WmMBdKN(7|X9_@@l|G4ktu+A6oysz6oiEVRy=$D>0U^C~jKNCYg4y
z*i;1pQeFg2fvNX!3h?gJUSS-OpmKF3F0RSrJ9$<_nd(JI=t<@|rL|@+G=oD{I(*!7
z5zMm!hO_1;n<4$;#St))eV6J&yONov?52H_bX;ugKZ6-z1K2ZV>d&KR>BG2oe^)~K
zqs3x(KiNb3GPp7(du@X)_-+S2KE856Yu9<#>z-q%)3~C;$mJbYtW(aEcB=`rW;L>K
zMmh*^a0XZG)_nE%AD%`lFwZ{w+N6=wVI=2Q6cW<vl4p~5LpI_YFjW_7HU{taK0can
z(8)I07=q*f>;pjt^YB3VphUO}T0;3hE9!71g47EGnu;;QJR>_w8wsYHNqJC8y@*&y
zh*R5y5}QscEMD)S-0>6S`%fXHoSDaZRSimYhN3@;Td-L=em*hoqdE}fCD^Q3W!9+N
zbJARrB2ms8#Zi5y@z}x4jM0=){MoZdqw3uu=pkuM8cYt&^qjJ)s$<oMQ#i2+x={Rq
z3_ubSr-t>tLW#+=lM{p9TlIYWr{co{?BaTz8T)4XX~J<Ft|IZD)X%2$LP3(Ax|O-}
z26QXkoNy?i3eNXUnBPrSaK?tHEZOlPUnO$9BXkY;4ubHDktR0M5b?B6r&9%k1LI+p
zWyIs0#Qgf8qx?84;KNcd$T2&3Kdfn0J~F0M$IZqt{)y(NUTY|kdfk-i4Q}(_zn~^T
z81^JRmQ+wxy`z!ISs&bJ^%Q(!&5B;IZDvMiq~*18BP(eLaOXYX*8RqI$(!_&5waP!
z=jIGrvr(o>5#Q?S>M-$0^Vt*U<x2|5N8CB?;DX#0)TtHMZCE2`bT_1>W7tJU=ZnJA
zt!`Q(QU$}J_(k?cP7U4E1e0Gjratpi1;lTP47<npQ&Us3k@ZhPwDO>HdRpG}Ub+Cu
zy|nHTwpzNtN7w7pC3JE-`~A8mD38BO%sVlEMC~@+3qsCX<(Fe;kE^E+Lfx!?fPdKY
zv0g`q$llGZN;X5U`|)44(}xZRWmDAp89Pw&A^Eu{tb9~lyI#F|1sZu%ad5!!&J_0Y
zZ#{6)8sHxdjg0|#_eRnr2tuW;@=2s}bywo6+qL#q1_L%t2BexZ_g=$j8=Kpal>^y5
zZi_B9Vy~?&t5I$>U9QA#m+H@(Mcfv2hFbb+4uU3(Bf#DoTF?=9aQ&eiu*GfF%9pS}
zse%c@OR4lMR}pXu?}jsnp`m#sgiCj4ys~n$rt(dx#Tu3@J+7j)T$i@{K2h#)j8q;S
z6&2O}`){}Bn%!DUHOH+OiTR%+)iR%DP!wrHrzHBGtXbDp5fCG_oMtgLh<tajb<E7x
zV2ILc9am8u)2$Kh4wu;-*Lp%Y^<*F=p1^BCRI{S&Ff1;Iqn#u_Xx@3r#K=}^K2BUG
zcsS8Gg^Pz50@a5%{h3$Q!?7v*zeDZ&FJt1MT7KiObO#r+CPT`P*$E`y&9eE>>vHV7
z*IQe7Ma9IBaz>=M{&jUCBYMhzpKXXC&Ac+9cM~n)#Vh2xq;58vw>R>}5?FtJzwyV=
zP-qEwh4pk8;+Z<cjqN>h*de(4#5d%uVD-~{e0z=9C`A#t1e^>hSlf~LiUvJfwbpc&
zN?z%-M<tXsX8JdW^bS;p_v6Mvl&N&I)L1>Urd%^1NO=)4^y}SVU!S}KI>D%aVRip2
z9X$lK_(y<UP<w_}?CR_55mg1i?kxTe{aicwX;m3K`rLJAp*<pkFeVP_S`>Cfw%@*a
z)7x|(!}Mh9<O#G<E}>*WyTfPqV<%lUBN3<_7`@A|s9pz@rg@<#RF%MOIwEsC&}C>_
z>99^OY`(_J71d3LDFmXU*~eJ>FZO)N?fN?d*=7eJ@iOTh+LL(P6Lkz^r<2w##q>s}
zKSo#6($YY>YGX)B1k718QXi89_o9{5J>`K45&{kZO&hG%&I#~-=l-=UJ#-J42aTW!
z;#^Uc6qv4TYHEVMnQXbaCF>nH5N*(=A^j=KCCI=4u!k{Bp{}jn6>{ToU5-}KdV8D(
zsk(H^g(3H4^qkE#_rpoY%i~_pG1HMtXgi2c*SPcB|K+gzBOlrRZEywJ?yKJ=EXF)Q
z8Ig|o2D)}7oguRr5|YZfzq7IN53M{(K09L{Ogzv|zK^Z~cILt{Jr*=%)?OO;F6{~0
z639T2?I8JOq`yqWym~eNG;QrGIk$xdRBg`1=zxA>U!!xg$XE7HdXo7iBqWOL>7fNu
z&%hQ&&L6{+Bm~(7j6g#}^QfCNS+=!h85%-IePRUGYYl(xDxl&2$F4#)neVxTM5%HA
z`{Am!_*csFyVsX1UOd;=r;+Y@`}Qpw8XBgL_si-9p0k;ga8o$LC?7C<0Pfu<@F?i%
z1*QCI#*yyIp2Eh*hwc(gn>H60(bChKOq5vz>W+5&1)>6i1prCM0a{lAr*s6(i2j0-
zkAabqCD1f8F5&H_TSwD(%_C^yyO+Lv`BLwLRn&@Q{Ak=0R52rn8w4gkS=-<DaJArZ
z8poQb0ZSPN7woXaa$4ws@6n@20s<PmKJ)YQKJVXOxpD=idLA7)Z3r%b%>CH;)4B?H
z=x5gr-mO)(jBnrEz^;(uvz_s&I5_O!h#{Ym3VXN#fq{tMcX9Vio#C_PO9u7?%{;Sw
zW*(j+OV87BKsK~|4~kCOs=t9}XKn|2!BW(MTwKa`jo<o=nK^lN7L1o#PQ#prm3bW<
z4UIm?l4Z)nq2&C;b&x=zXR_qXC?2y2o4~-tEWCH{>C-0wR((AJAVp9NW&4qmle-VT
z<PIU@9ngPLY}`KwjV%1@*FSvxm?YqA1Dy&}E!9lrvAZbpjs6FEd@#fjb^52EpupN1
zWohXmF)J4<zGM(X46}$H;{}5&JNzzky#?~>+g(A;06c+NE1ofI2Y`H=u=st3F;lIV
zWNI<6YJ4xwvgoF+gHAIpEG#UPfIRR`@?|2|BH=4msV*g39x@2fE5NpX{MrYy<9c$?
za?f&Yx>mRqsS3RiNAeB|h>|RGG&g9Afzw#lo^zGuE_}It3IG-o5<>n&KQJQhC0(Qq
z{|9MZRGSM8_a4_r!J(|G>HN2i(74!yh7YNr%P*u^4Jxa-#e>s;WCbA1_{rDbf6A&6
z4-g4KB{=}eiPCxDCKq?2blJj!{)nIXNH%h!+!k$bb<`3tX)Q0ib&U^WQVc|z{l{u_
zU_xCvwiUYy=v#jQQjg$~ec039-rm`1I`FL>)DcRTKlc%kOAH#LU8-x~FLaTQ1q6~*
z{IbsIyPz8l-ScHHFGO1Z9YVr^T2Id+<OHCqmzNhDrP865t$B`W!2@b)^t$l-Vq<K&
zK72%<`In0wAGl6|EMW-)wbEU=B;=;lQV|@`ol6x9z{JF~IddioXiObkN#Hc)a@v}C
zV_97cZGZ?GPLrWW4<5YdRN~D&ufyDh9s|Yg+e4+6+6Q!5k>KCNELszw5r&7~2sP*V
zH00)Xp?!HWs_8h{Y)yfsZFUX+dn>i1f{2I+_`Dt8WU}KFuYTK&P%&igF$)imQ%BTe
z06>MQYj}7DnyR5ufh^T~3T|$0PoLHU%9IHkx8-x*h8W0F&Kc80sB|ZCWAy<i)+*Nr
z?F~Flidh46IOq(6;0)W%RyKu4iHOvpG}3IfNGREzc@`6KadF8wjKcxo*bP70D1$aq
z4Z3fo<K_})aSpUWlsusj15ycBs3K?tx9ZTdNp;_e=Z%({Ar+3-spP7_AsCdib<SUu
z;FvM~`frN=Ixen&?GmnZB$??OjZRv?B{MS7ywTdDxnO9_F2KAR2%uet?6Fo5#?W+I
z1)l&8^x5;`7^WFWh=}%A294mF0jiJ}TKc})kfpCyqMifY6hxG=kQcyUfjz_cuXnqL
zwU^{|fIQ9^2DkXf7~@H_-TGJT93b^Urvr$}lS=*rVEDc9>;ck@4w?DoNgW8K7cXAC
zemw*Y>fdG7dXf<oo@d)IVRkYnU&0rJw15z+Ift3Jn)Q<L*N?sUqngNW-ed!B5Dy`R
z1r*GV01Nou-Lv?kD5UGNAgV8x%x7=n*Zd&rp}12qIuv51&bx+?VsG7Q{LF6#<>|9$
zO^_Hgg<BZN#sjtT!U6+z^!%eyNfnwy8V$+3w&D@rzI>sI5glVkLU|O&Wloxvf#5Lz
zZP)kZX$QfeOT)foKGZK@pJlVpIE~8c;Rd{q9{p+iME$w7P(259ATxC&_@@xu!T=;i
zt#kg)?b~+a=B3tdj}}4KB@X_+6NUk!3lPlN5DnL^U2~kUmRQu@$h<d7M30b35vm3h
z0bY#4imX7(>irSJ?u2dk1Wyp&KzYA;^CpyzSv$mt);BL2@>AEK$%dxGP8q*4nDLN~
zuA*luZ$ddY5zOHZ_{W#L0qryWO_Wq8ArT*WgqrC)G@6-=N=fMzpMe%YUN-uz8#lI^
zZpoE#Wx7T21vvvVf!ZeXY@F)%{Z1R`i9;Nhd7Qh9Oo)Bo&vFiVS#bRU7gq^HeeN({
zV|At5#zZ;jSwDXKh-OeJ431&iU!*nL9DAf_pYn6-jm&45ynu}Zr8<Qa9nB4TbM&l%
z|7JmBvBxlwD0<Iy==Q+y$S$g5ZlESBC#RB5TZHxZHo!f2<CWTKz3xF@Utjvm-``=g
zniW>K8C}10?&%Y4z%Fy#ycT;nO-G!xYN1&LK)4B5#KEx_^TuKYMi=Z;)_xbS+IvPN
zP%~?%Gj>X`5A9d1B8wg-GUgZ$rsKsLw!(*KDk#iVV0swvC8PL+hQ%28e!*k%+OwVD
zRrd#>=m>g+hipXweZ09_GkB<R7sy+X5+b6j1CoE5v3(#FYO!88TWF7j?xdqV5%ZxH
zf{>693^{*1GGPt+T%tvVqzoh1JLHPp^)jCyWT7CpJM$Aa5*G36m21cnagonf5Kvk1
zvO{|VCRqC6&g01M+_!O0&$7*W4|`VZ3ORmeC#jQbxwiUqJiFKlD6Dyjmoj@)<@MsX
z=2v&%OaD8ZS$H^dHP;NfJsma=l;#pE)Jx2Gp&8p-x+Kd}IVof@!NbOVgq(uN4(P7S
z#GjUKOU}6COkO!NGXoJ*YC77pn^P`1rYUFOcm30>%fZ_5m|hqxmH))TBT`7Eyt8o4
zibq#hckJZq=z0F~O=Q6>XW=PFO1(<vb-JZ%|E>(BHY7sbHD~Yo(~!ucUQ~t$h?GU3
zOX8CfMk?ahCr|QDPn*aB3h!~6*iE)Z9&z!Ht=po_j)kRb99GWp+n|yi!>WyIVq2Do
z{)1P!+o0kFsZsfnSFRq!p?H39<j%ym$);?K24)&=cG_wDijf9y)7oca6~cv#DxdEy
zR;#XIyeX0Ng63#lHIBDlp0p}|UH8jl-CmjUKT*Z!Ss*rz8ipYy0t(Mdn(O`eiF{{<
zn*Hc!pN*7Q<+XOLjs;cawFL4yEG{laQ~K`HA4>E%50`idMBx>@8|ZfJB4e=?-e5aj
z54FDU@u+esY`cRN=9Ikdxl7e_<JW7aw^0=F2gO@<YIbe;6!N9dq?)m1e3;>9d|y6&
zw8VLB!mvnk)I(&hg7QFs?Cfyt=_LGuWALi~%i{e}3id23)6WszpJoiPY(ZXRSqn&M
zZnAMVtXPuU$utM)EX!(=CGlYsfdP)R=El7J8kLoRPVZ+yfSdOLp%VfQLfpU6{<4jF
z&5QbsE?y`#6`b61AOf|7y}`e*H$|Am<HbwWX9sal@C<NrC&~qwF}%C+4=J<KQq3HX
z)<d1!^q&Cwy->xNy)a|!)kEWP-l2Kco$}0lgUwdMI2+ndpDjDcot>S3lXWb;XL>}p
zHg>!%UoxYHeDB)T@z>))<0KG$3JZy%M{FwHK+N6^ijRdS%*2zfjfVT~$0<3Nc#n3B
zk1j3=8fq1c;jkTI=&Rx5;lb!d+RSp?>ajc*$@|Svw)X*r$X=<DkN$<{c&F>v*$T$o
zJ646z8{M}}*k)hf?I3oR9@jQ9C=z+_=7)iSfrUkPZ9t*mvWIEhydPCrbmJ<?I7c9%
zZSjO;7a6;OjLgTj=iOAS$-aJm1(esbkM-Usl9R#0xpL1xZ`cdB7;HjUaVA3y`Tl2N
zU^pC^RM-B^Q)>>kf(bX`7|TUwMti@204xlQcUCQQc@AD}f5Xmevh!2TBqLxHkmo`}
zsar})N<<7PP$n$XY`ST35lw1o^^k2)zND5dvHJmo<0oh7EqN|Pvg|{%1P*GmRP<^s
zmbHFHgjqD(Cr(uVVNwk(N*cxlP~s5Re750xOR>Q030`LPm8JN0S`BF4OsKtz(_@er
z)B7-Yye{vx{QmuW-{J`+H9;x;XOa=V1}r5V-uYUx`ug8iHwj0)q!)V>3Na~fZ@X=h
z;Rj4wc&-(+HpxurVGtLEa=cb&7*C!SSy0M^LYBK$rd|wLGp3gg(ijS^#%>Z-xkMdN
z6eR}5OY`{DQ`z(c(@1W7+}ly-6wMgihdc<$eH5?N^wD@@NLjB*al`t(1C0kTfrdg<
zRh~~{{f@gF-75$GI}1Q@H5Mlo${Ofwe70<Lx4<{Rd34P&DlTgnh}5b_i^S*~u6P+p
zhXF`4V&eYpR5PuDUtGJv)RGYZrD?@a1(;hyX$?OOeuG}x6mYD!v(Idx!k&Vzf<CB2
z8%Z%_?2FK>Z8KF=Pb)_&IFYmpD+i<%KHrOM<4^lXQZ-~4B_qVcD46JoO^OO1$Lf*X
zy~_*c?k#OhUUpT<YFgv1pGSM&UWjV60!OPne%njk($s{a=?x79fe#vko&_NlH9}pC
zjsF0x6qv@>1}wWpNSMI!?g5>Kyu3V&Y2=-<Y0XZ^&{Un!LPbO%xEIbp>t^w297xfh
zzEqIP8Xi=%lkAo<b^^Ej3wnKiAe7k0yv{>Q6h$l){PMa7H}cXfI7lbeL<b}?F3^vF
z3i&phAPX0vRC~xshE{7R0i^9svB{z=Ps*FB-|kf=MBJr~+uZEFe(0>wiG+HNIdOnW
zs_WU+w*(C*+}o>6rX`KD&h;ihv>Ue*W`1@n`i)L;z8iQ}^e|)j5nH6mwT;-L4JQOz
zhLO%hp2>MxE8}ODP^=?vsV<6Cjk(GnNNY{?4K`QVBG*<^GqnYIhn$keRkr7!X0N6-
ztx3BT*gPZ~&Met4w+KJdn4{Z19$#)jcgw9Kl?XyoUnD>NDJMCxJgrpJ`<;3F$E)U>
zw1(QEKZB`F95iALAKp0D<*C0)AUyPUB0Y6&^=p$%=JTJ@rf05?5*vOtFX!^7Fj$$t
z(mlZb&OE&sENx{R-y6bY%A2&z!G9b~s<Z*ufda8>#_(?lll@H_{*iaf4J+BdFfW*}
zNaCy#M|vKurn9zIhBEo8th<!fY!eFy-5q%T+;nk{PR*F#zU4)AUEqw+_am0Uu&`vO
zs5r^uAFbJ(>J%O6ddq1|y&X#pwfCB$;@qz(dDaAnPTIdah$&%NLUbR*y?Znc*R&-C
z7pKLL9@F?=X60Yr6h4+Nnk|DhWj+@}s)KIj%Wq!7nAa&0ybdx+jc=xKPU=wer)dHM
z)1p@xq#Yx05Bi(>#iZWBrp(GT`u<k^tbF$pA*E)7;=j*XlRo2*4PPBFOW!+n@Y1O{
zDvQ#&aL<G*o9eT{-T0&(2bvdx4CC%KWo{n~11OzR;XI7<W2t{vt`6*Iuv?O(UrFIt
zx_StUO=N+|>?7Q6{DTjci$}WV(t(M%CL~_Qv7U~77>AF~D}6SCIrtb{H>Wl(PgF9J
z$nN0Av7w7o$M$`W`7rI^#*w50q8RPPBnxT>EG0Hg0k*c(a}DPg{8JvMaDk%IqV^oL
z#>QXoK4VogXDV?7y1?FqevjEjhAk0(I=i4KT3TZAJY!QYSvxykrj2FhWMc5C_kw$_
z&BJe{6RKO+4&{ujQMXtB%DSnDB~mY6Os>u?p0sYPsd1Q<txh^_(m#$L_;#%)rL{lp
zZ|nzdZ2^b(FfKm4bJ~%Nb4j6cL(&avvRm%M>Vg_$*$N#32B02vvXXv^U2e9rKe4iZ
z)3;RdNXW%`K%VQqeFm~#>-sW2cVmz#%fVU@iPoQ+goSth*d0G<2?b3jnLE9kB@*W_
zERdyS9~Pxx+3zGGn9=;^w-46*-1Cg~+lXg8WG|!|y17T0VhlGK`?Ad(|2r5{-6L+M
z=>cNm;F}zRCg(0ydkR*DhF7fA@&ft;`}+7KhRpu%Z0^habCkujWP$qLy0Z4##;1*%
z9KF`ZPbPoAc=y<_n~&9Uib_Yl(cgF!UrmjoWl@nech<N0eXp`E%(%md;UP$NPkk_9
zT#N;7Eq}*dsBq(mJ=!*Zb-B@C;;`vdelgYlWbNQ4wK()M5N%!0?Cd6WgEkdSHjwPv
z>7u$0JY<fO_M<SD9uX7c0Avz%M+hzzZTEL12R^h=J{bDx^ymJ!499_$rG>?}b%;Av
zLc%I%KFtq~|Aa6R5$?iV+JgrVTEZwTxe`;LI)ai<!R>Bf?SL6cOvAASkF}c7Z;of;
zU(LrSxyHZE>H7K(IZG#;KXGSEa)QfA&^LjBYVYcezSKsTFKLSPE^7p8R@`1|d2%w>
z=A6@T?RO<QR=(Y1^<QHatP6kkoenk6&WcL)PR<QBf{Q;~?`6}jcH!Ln{kjoSS#Yz$
z>({r{-|fu8<Om6imd0#1ObWt8Wbh-}S3aI9Hy#|%9ep#*`2E2Yt7;d&p|QZI*ZtZx
zC^sH&Zy%jbvpJYea*1=h&lZUxVRdk)3)<NS6+eEo>9G2~=G0|JU1GXL_u9#JBwG#I
z%UUcC&}I@Rsk(DNU7OCnxZ2w?ux9n+i=N~51B<Ggv$LQzIaHGCwa%R%u9nblKTI@~
zBhs$gczve>dajF5w;@-d0-~j%IWkW3H;&|zdsx34bDlWuan1b%=9#Vb|H3@rGzT4K
zFv;jT38P7~#Jwg+Lm8sF6dcBVP;`O$m#lb*8C9)d3KT9aZL4rx^P|4Jcv5`FVnbtW
z{m<{>M!`cj*BSWNM7DN1wQyUbmuh5CCqJ^fZ#tUmzEQpmbpqT)ID@g*4=m{YRJAF-
zbG(-x1EZU!<EuCXt+|?kVFAPH#vF5-o~?q1L(Y1?&yW)JZhIdUmL(Wn_WoFJ%{)~S
zy>)o33khjw)4}W1S9yb+U$>V6F^EMlb#OFqKZVPrm9y2%9vQPh1F&UkfFr9pl%e0`
z885_LoTEJJX1#54S#txGiH_a50}o^s&nl6r$dy+$M{tv7${TJpecn3N5EVmiAv-4B
zA5`XZM6JWs-DeC}!{kopG}ukBeh~GLFgOAMj}_7&7p}PhcZY`RYAc^MIrhgp14NFk
zPP<~kJNVrSN`}oyMo0oEf1Md*YWi06Ee&yzb~wPbUInJho+#ASv*W{TvwC+a(vGEI
zcyv?ik#Xd!^#`vzm!Ht$8APs1%gDU0#MYV(c9&ub;D>Q!7`gZLyHyUNe~(GAOW7xN
zq9f27poP81`ZW<QSnMoz!WjL0yEw;qTsX~RYU;+_^|EqDm+Gk%I~R(cmkH;;O~rv*
z;0MgZZ8R9LA|$p3*&67Qt}fDyi_IJHBI32OatZvqH*aqWs$|Xf!Hn!G=vCV$&Km87
zYqPVqo|}6v=OAng1~r#?-qVFpep6z$E%259qqrWBQ+4%(fbssdE7Yh#ZPu96j1T?B
zx7D%`Fj%dysC{FEN2=>HpXY^}y83Sr-I{!h)N`QwyIU1%6+j~iVw4Ba$E!`cvmmY<
zf=uaL_(|QM12x3ErLf-c<U5TdNKnS3xycV_Co)D?^EAqzIBsaL%_7=RMPpOb6Vecb
z0B*)2hWIl>smS();~A-(F1SPoWAP9EuiCyl9_#mSTi;YhaiJ)CWF}-}kB~@4MwFG2
zvX#9XBqUoTq>>PlWECOVsf3J#L{|1*_i=rH&p*%odOfe_pZmW4s0i2RbAHbAJ&yNr
z9JqY%(H3QWchRI|dMFK!H1>3G2nZ0{o6|vcA4CuC1#b`FmsO2Mtbws{sZArT<sixM
z$VfQ(o=2snQo_P*aYsun$|e@qfDydIy0G5PHB6ram(Qx^sz%FDnPQJ+NpDDq7-J+`
zz(P!D{ijV=M?RC`0cnQ#q*X#E^YMdwIcs<D9X)!04(D7w_BC;t{Vu~NWqw|s+^E;|
zm32LprI$QNg&4etGF`2@7W%$DHeno()v6}>4A061E6;6X*Fz?riVquYWN;5mE^j$_
zOn)d8$Y5QqJc3`+AJS~Gswp(Xy4G9^`*LMr=F=4xRz6y>mml>Q<YYaI8Jl*4w*28K
zU8?q;C>_n^<$Pd7Bw%*RlaDHNTGJozIwKO-_#v}?GEj;HJx>j%^2N;-N^V;P<n~?Z
zm#v)be)aT+mHo;bCP5e+jVul710TbF-J*2nCCk7F@2=$})Qr2@sF>~BsjdLl($)P8
zm_(kOH)-APvS>2iO49NBfTgNrZ6Rg#{0GxqdV$5H39cAwk~_>z=d=SX{p!nktZ3bI
zbqRihmQiKr7HNVdbL8X<?Iq%mI(8k_z@uX2v0XlS@}$I-&+q4B4mc_3JBTK~Dk_3;
z$xfZ)LtA$AP&`G^xjD7jqf3>IJq&c8!Xc*c27C2rGBvmDvMhydk+hHy)%B#SazgSW
zJye%ez7^{49uN8P5_GELDPH_)7cUL~W0vc_xuTu2uB8UP_MqH}<-*;C158Wbtn;Ik
zm6d}t^MU!UxLDcx@lFm#5`9}lM08uUES`L6Sfk(^-J@jiY&bT_;nKpNFRa>E3k+p3
z2OR)0HZ7R0Wpn2jhzv}e%T5k&kbukMW93;DBA`KwwP4YR#UyuEmQYN)#mTfz0E=|1
z2EU<gkh5($Xz2%zrraGFPl$a2!FaE!srfL;lCrVutbQ*Evae-a6$vRR=^ay5^~0Do
z=9|e{8l3<PimB)?A7B(ZjFBH9G}VnN>j!M4iB=+nWrbRCLxhu4`MaSzU~7Pk<dn;h
zZD=Qp*2Q(Y=LswWvn=WsbevYZd+-aU%g3@cv)bQ5HUWb9em7f8^52=W(bMxxDr4f#
z8LHB1TQm^%fi<TiWCnw+LO+`{S4C5Z0fK)xvq@hY27aI4r#^QeK1se99UewYl(|vZ
z&KnI}QFkVXe*~^iaqhW!h9|Cp!YHeqd}0`b7iAkJs_UUoSpH6%d3&#bqmEJD5nEcR
z>*x}JXBB=b)NQt2#!voOU++h&-HiF&0UDaYMtXaXNi<o|wTx&;mHwc*((`%@4E2VQ
zmNi(lF_Tq>0`KH_am@vp7Fb|#rVt4@0K?&ItUMz<bFLZz)fTpSdl|(W=AEP0O`;EL
z*humT3Ek78c`jD1*v-Gipj^B<p{Ec|E4GI-SCOoZ`)2Lc<by*BG4i%;wH=8Ap4SIl
zS#zAE(lvLZK`CCHJR}^SBt1hc!o#{tdUo_skD5%J>kJ<_0|i1{Y^<u<;hqy=$vcUy
za>t1Rr?Z=~9%0)wR}>Y*Hcd=T`yf;{7`=Tjh>dK!A5GjyO^Imcifb{;<bmTyOcGob
z+-ob}0JmAcx{HO5Za#f`<1$33_6#464!PK^qt9X{`+NV+^loz2rlnu5l^xdkweq%X
zGM<b&<Bt8;JUu;AOO%qodY!czQSy&EpZI%W!QFC@)NAg~GvRvIp|%#8?PZz1#fHYl
z*<r_Qry`Z%Y2%k4Ofh>X+5Y>psBrxA35F%E3{zZ~QtVT7;^Y;y#sT6U;|7+4RIGA*
zJoMh~x9oeVjEn6%P;v<KJUpxcSPPCePhPreCJJWHTMZP;XL4(RV!6Mw5tJGXo6IUa
zol<K%!p#eU7CVe4{FA$uz5^$zE&6TeZ~#O>i&9q(Fn*m2?bnNKY;5EpFX1$y1~Rvw
zz8O+qX^hMrI$zKv@-O;sLWIOIViddsGggRg0vltwJ=Ui^8sbr8p^kfDQrh!%%pH_S
zqFLOB;JCwfNN~W!t!gl<f<eL)OhP}IJfA8g(mt>W$C>lbeb0I5(O=PH8-nQ*^eUS(
zwbUMzMke5ripi}MhW7k5LwQD&F7?^GIC`sk28>@ed0&GwrSqe0K!NQxMnaC@xx8$e
z02S4$rtag%9>?RBRZh;L$wYzT<u=x1gUGp<Hk<p7aUUB56vRGL;*3EO_){U9<{YwO
z@*9Pr7eYWGFiLfK^(|xH<>CwS=Z{rTnSF-_MJ6;i0qL|YRo3Wx@*y_1^puo|##c{@
z*F?J?Y-4mxONTXNlk;{(objOIMbk?4U1x%9qUKfIeDLO16;)ML+(2MpWl~_8cGmo1
zQ>P=#VE0rP1!_3eCqe}VnFWHvglo+=4v!t)n~PlzOzPxDxow)U?wGf3Y;2T=<Q`4c
zHad5XXL6YW$}n<v@UpnLRFnYQc0u+4wBj8!V;;TAdD=uB8VOEVzKKfKn=FcoTjrzm
z_AJl1NU-O!O{8YWt*tO1LSh1Nis6E=fIw42!}l_`EWH%a=L8^WJ;%hnx!qIIUdoeS
zykmfu{05<u@DEu74~6gG(eRrT2H})^eBqt@vN<n!MT95V|0ut*tKdLhu~6xbgu0`T
z-j%)#%Y1BQ99+&BF?n<M;b2nH>CpL!K2z0{m%-}k+gX-LEn+&NZfVJ4yjV=^yN@5$
z*x1<SGV-{&xu>S43=Iug=!FCY4;Ne<A0MB|=kE%<9mJ?BNDou0;}Q~Ks*guTY=fhf
z*#EdX#iyjKgLDtZFJ2H&+)&arHHF-W+sIQ$NGK;Khp3mq?v@tg=FY_660hQ733+)I
zmmoX4F{mnjzV&JE>^ujrsu=ZaZf?KxxZPzoF(!lfu)1V-PfrijYbhxym~=%lym#SG
zZ@VAJT%dkfKu|FLi0~ORGZv0`Xpv}XZ?3mZOiWyD^sk6|jd$SX-BUl+6GoG8y3VBs
zqqbT6NUx(jv|-^95h2PPe;G71HC4zwGK8tB=0O6)i&y*Md$Goj%D~O?^71uw9I2J?
z{{gG%Ddp%6Z{D|<@q*yuEywzo<2?<oI`01cW#r2tFXM8i5$NQgsYy8UFgdv!Sq+UI
zmg7N6N+VtB?c29;7kFV9gr$SK*_+p|qvg+lvMtA2@bsx_|5`aheVf9%pn`(J(W7}~
zWyu>W7>-3PKeRr;z)%&qxds8XnYaDnNVVL&Jb4g-aPA@^B4+QM`sh119JWwnRVXDZ
zE32qjRb0G?OPTvCutK|pvzV5ahJA)?2yX{oYZ}rUKa~{He+~vr^f=k|-I-n8KpjY&
z?G2~gqRw4dSfI+rp5zE-WMXP=Ztm^r84mc$u$4Oh@#9C_1u*U3!^3j$b(38_;oG}s
zslES)fteZ1)yC23k%iPiL;zXn&yJ4bvNDvR(sFV{EiITRbBl{+mHuU>xgA|ylE;oU
ze)=SI<cQm~Yp1oe_EA%_$$C0s*bBN8HWEgfLi7<h=~;Ps?;sVc4W-V_%@s>-!Dy_t
zyL%PVy8F2n7j$%@K;c>4z(P=x-o#u-2(pOGOl~f&UtoF>MU@c|y`TVK4i7?AVoXfo
zrDpJ}695E+hK6G3oYH91w8&Kp`s<s*LViI(s2i;cleDwy>*`k5*OB5$m${*deGC0M
z`V%DB;^N}U%F5%%kK^Ca^hl$f!_yb)N@=_(;+mp&{=CoHf*rb&sVR9>wGck~ZyiIw
z1H}$hn8t>N9_WX`RgrY=Ll}?_-0%kcY+C3l6|4(EwY@#?&C5h~O?33Y#6<I;D<<@T
zqn}t1n{Fr(7#JiGb1>n-$;{5p1-#Cp7}!@|pY`&kZhrk7Bq(=OA0tj;>h{a6dUDe7
z^5x4LKjBGo6l&w<-}%OcL$D2nIMfvuBW`Z%Xkc@4b5V0}Oj|QOHyB4o25Gg#wv~}5
z4r;g|=rog3Qn-R&zkUtAQ?-pK6s%mFoVIOgt)ue}_V$=B;<Q2=WN&XTDk^#|-<01^
z%JiI0stmNAMn*=<ligzS^6%Ex)*w<+^qKq6)Pym&KF)xQj0}z%w(H@;A9{M$;8xdP
z;*y@2cn(^5C8g_=gWnyGE8K^(o`8VBX>Dy!H#ZEx3^10+&2>0=l1R6uW|eFBeGSrL
zqEs4@7W<LNfopATg$QnV)_ZGf>s9EJ_A!vl0PvFU-AkmnVI~fk1JA_=$?)4Z1H?RV
zNfJ+Ij{N>joFwCg`;d_E0PPt>NYE4@m0QE*U7`*pJpOX!z<~pZW6&rodwMS2JtQ_G
zMu*02XncGP;!&IcqI&{pX(!31aE0qjQ#?F8*qLoDEqei(ObG5t&p?0^Id~T!P9Nx8
zh>Yy)7crtHDy;pMoskMqK~ghIMW?5yr={f=6m0*BQaop5lrHV|vA@5+x7Q?pRByl)
z`wm<6p{!SaMuw%Dnwp;8*Z(xn598y-u@1X$W#r@_nAise1+hj!uU3rF2k0)hZ|}~~
z^<J5=?5^$Z>9M_Zsrbzs{0Vo&YOGsdpFb)`w2nYI5(4DIRs3~9?;UMPZvTCM8?rz$
z4m!Fx;@G;|5yZt{Lg37uoE(At2YC3{l~=vI?mM&w1_rKdeqh*#p}2BT>H$PP_#}!P
zKTfaMdyqpMrWyhXGPppB5G(9e0^>0&RR&tnR90Snt*x~N+W(bp0tx=8py13$V-u62
zZpUC<b#<~^)gs4^RqO1GPfbZ-(yoVn7?eMtxsY-2@@4@^cx5d3*7j8-=E1cIBOveq
z2w+n1UjZ-5Cdva?i)aLfCRA9~jd?^xDE6hU%3c0&henV-g8dGHId&r})Y^D|z=HSX
znOAc|y)VDVHw=u7s6F{|n~H^zmGxeG_2S-INnxVvV<W4!?@CExVkeg4TSo^@n3=x5
z3gH7d&x;ESK+kwt93WN~XFLiH*!$}{@CJym3G#gy?qAJ~Bz*1YNL37~M5$I#FpbuQ
zNZl_eh}s6H9-kW<96W~L4$cNRcM=j3{>uTa4<ufN`WZ118>TTuY+XE1kUK6pI^I=0
z=jZ45w!pdPwZ@Vw$Wh)Qcz*l#?K?@9{roxU?%kk>d7@@p`P2K>M=XB_ktX9Rg<9;R
z@F9_lBnoG2s811s7o=jrVgak9-Wy}=Zlnn~3WndhwKJs^q3%sdiHnm{9fLhKtastd
zmm*~~W>tYKRiCS?PiSf$VP}8e+shPWj+_oTvm9$%d%GJ-b)=A$XM5pXEGb!1T&$(8
zF6BM*XtGdduRWIzJ;=RkpB9#vmzS1oMGoquD-;(M#Rak(B8IuUzn&bVKYAf}i*fKo
zjn*u0x<9A98yI85!3%Sc<|oLTB46C9CL<$zueB#cd3JH}2P~Ta!q`}7?1A<ZKW%ER
z`R8=HVG{7w%NrR(x0a^#<^1^%(b7WM#o6@$#!yLTc*@dH*2lD?IDxxEwVSKuo0l(@
z$Oh+bd+qn3oQFcywTmu4;@PujU-yLN<k(-lcro<b6>&!Ay9DL@{QPSn^9u`yyHqmu
zuAYxoV8bnPr^ta6)<(;t{jkBd2$F?=10tMz?#1L@5;<tIPmK&xR!{XNJXu4-j@DMK
zwN`2>sziBH#MMwY#^BFj8M)qD@m<>-7HFAf++u-|Sg+|qSDoq}W~x0p4IS~QUb1Qf
z4j~R{X|*m7pUi5X{c__-rfJA|zWttS3-7VdA8-G#oXH54WrA1A%%{5qqBXP)^)eFX
zFBLoV$u`3GZ{Pk1>kzd4GwC*FhvdBBijltONtLhj@1rf#W$Q1I;lX(VdzG}xqtu9G
zdt!TQdV41LZ3t1bC=<L;#-)6x`l5}k?NIXHzkdw`4g}edaUgcWGOLVY9(4~AMnghX
zzza(Vzmzk+_w;xsxm>ygTU36(QjlEVy`w#|_F_REcS(9;FN=)9S_8tao}nR;G;ew&
zW_QSHRr&cO*@fo8kIPf}>bPAK6B8dJ1%L*9{f7=68r<>lPr*|B3!#;NmZ$E_(c9i+
zN9f{uSN+uiw7&O|><Op2t44;WrnJrp`%t=}&0g1xzjyD`%wTP!VkR{SMaV0!Y4dLD
zS?{+HOCtwD<sQ@er^3DGd<o)k*(*OkW~Ho|@98D2>53Lh#{LS5($dL|HrfxR&L%!`
zLCqC+MD`{-&}POmiXv6c?sAVKf77aG@|~TXiL?<s&)v764f)0qybwwhmX+zPqgfhr
zUjvDUnS~{@Yc=25kTH^nMZ3RdV{6B#fq>!KXP&mR4V=6#mY*6Ll%`o~FIijPiYP=8
zih2W4VIc^VUEZ=9ji?wN2md{#*<<c;V1MXzTceuP-c!-gzsbtPbi~Lrm8wzU>QFG&
zu`@ShHC<w*xwp5swsz;T@L9h7a`avG^?Nirj)PR5A3zt@Xmi2DL_gaAm4e)}>!&l(
zr_pKb+h61HfLYJd$tjYze1_R&tlebn8G&PXY)pedN__Ou46ok2dGpT=rG7qRb@he{
zZEJ|k!JA5N8=du*qQq-V*o;8R5g@%@f^A4V%+oPJAeIhqXuJHTO^tM|Hahysx~TiB
z<PR2RW*H|IOqnC;P`Yfu?IIu`V0NQKxb&xriAgVPGftff^<VuN+y4fP;%K(#lUCN&
zv!s9DNB`zN!$Vg!ICoF_Cwg4ed$UXkpkeD5U_%jk_pY&EVt&5RmT$iJHH#_T?>02d
z%r7!BVv>@+Tkch4y?1>*?y0O<Pj7D~HM7q<d8yj&Zm>dswI-?`{`%y7!|c=H=oT+C
zu1d;CHGLza`l`E@mo6<X{R8=NtomV;Fe4))2!~&jIfES?9fv6{Ll-|h9IW|Dl`}(^
zkLBDQLXDGk=r5|#ke^P6zKkuPL}3qBSo^~@wz{;mG`!#^w7dNC#Ub16sn%#O`S95`
zffhueW1^yrQm&7}!x`?4eI_BHZzTSX0E)jFZ;FfSE~k2LL(8*rPKyDqAYcFWWDpb;
z{rdg;K{j3%syft#9Km-ZBWDk7LT@m$fm7VbbngY7IHMEXnaDUiJUslVZ%_Jkc6Qo|
zCW|vpea=uJS_QO^xp)aQ{Q;!{r&UlHThYj<>;3zSfcjQQRDu@iPFqwAd~z$J5c_CV
zh<BYK-_w6#ZcRl?E7FvsQWbRtyPt2&#uQDYTCJpPVH9o>Fz?FFn$7g@h*z*GiLCYU
z^^KJr1f!e7#S;w=LT*qm-@$_iKaVYdxgN*F9#E^}i3?j%nVp$|?-Tjz7^Gp9xjHqd
z-ym-J_;Ghogqo+j3@GPgAor<|B|mt8qcP-)#e{c@<M9fWP6S#O@hb%7D@f!>bBPYd
zoVUGhQ~h|mLjNo^HTBsu+fX+eq|~><cJ}s7e(Zlxx2>;>C7<c@1<VM4B%)0;YUfif
zKW@Q*c?DJgAvz{B%BWw5<?Rkh%g7LoRN!h;H|nN7-f&4Y8A-kF1lASSJ8k%Pc216r
zv^4H-(Dqb$L!+{mmJ_M~N_xT1U%s3*G`v*QjUNRUp+{e?z_)>ep}ni?zCs`0{wA!~
zzP)>g88<?RrGbRF_=nMK)Hk>{w1Vipq0EO4L5VC9bZ=n|1bx2<FCzzs36NZ4wi9S7
zPuwE(!2IIbGaiNedObB8ZHx&hGGICL?Hkcokmv(Z=|A@Oud|hvC{_@CJ|YxS5Xv1Q
zKNG#-Tz+?3Ta<Ff%E}5+mmwyG3PMXu3)Kfj$m6^`)cYtpT(df{OHeAJKB$5*(MAWQ
zj{~{#6;SaNA@4?a3Tz2|72+cjWPZLg3U&l9v`i@}CdS4|IXUB~MqJq{3kzpqL31A7
zCkU24wfmTkw=_1|p-Dt+huXBEtxX4WgFBe}em1TOl;7&ug`O6b@ii%C2M323bxd|_
zFJ5G)4J(iSfq5<yBja;}92D#*Uc<vF5>Ee~U&X4yEC8U2JxU4(2Vx&v8Bh@yrw@@;
zT%)6t(<MwRSRD-nI+~kvaXsj}ybYS;`SlTr0yft?;M;`O1$70hj5{J8Gb}#D#OBdH
zv9hv)hth*ojY?D2-KfC|&lr_cXlg1;R6UVv4r-L1zCPO3S@}gA`vKwo=Z%e5R#%}l
z-Se}?>+02*l1ug99X*kj&?+N(z_SCTmcso9KpCpGH(eMbA(P}Ikl`NW<mGE?YiFk#
z@k8N))Lk%{dDOIY25ka8Jw0>==tAM?0?`z-5nwH8XCXe7zkEpu`gI8d3vk+{@VwEC
z91##GpK1RLmS!`0G3+pbBS#=<K+=ZGMri0MU*9#<4$ySrfE%Ee&1#1!+BR~QOJC_(
z6B9r1W6h*K8X6dM7Ff90+dqho)-^Gi7#uunWHdTCIXONq<M>?!4+(}^P?cb7q6$zu
zc`}ksaSO`?(}X@`KFq25I}|sBFyH(3?R8n%DmDQPMF=`Uh%W()oFofCjsV1fgAB8F
z4=XE9b8{4!c!&^O?cMzgHR`Lo(SI)XzxBZd;rL+WgkBiY80Z4}#4Y&yIQPFrTfKrJ
zn8=aK$)TVYc&f{X*qRIx4NB!KYp(vC)OLIST{3!#5Y*n{c(SvzmPOszLa4CMoIVYa
zlc}a=Gej9gd}X=?qZCg}_gtbfR!tm_quuGoN7y1q1O<r(5u2Nv32#KUi~b(dl!YV>
z9T5hI@KwUwUArcFzZoZqn`{U$u_9)CUguy;Hs5?-7!+iLJTQXe>VM^(FWSv!hF}L>
z2z*xR6${>>P!jt2oDflmL*XrA<L6hQ#Lg!ytccvM%l9HD$JcKakr4&$;zU<TNr@MS
zskym1{8iHOv$Huu6UD{F<KyGu(<uncD6AHK3h4=PNl6PZ>_LD;cRF#wlk@}{{>!$u
zsM|+|hw;DDKPpPf%O9}cdHC=lmKD8pwC3&5(9z*x0x1m(OXU5rn|R09uf^q3)6=Vf
zb%8AaM2v=XwvNF9$C8$mgFa&Rei8Z2J9qD*4L6H0G&9?r8B`Iv-`m-#Mo`77AesS&
z_3>H7WcAn3kdXOXpbYCHtxPxBQ3m2&57TONS;e3*F}ZMosO(KhFf=haLK}wfeadjf
z#bp~M?FXx?k~=jtH8gxyW;jj|R`%?~jt1$^7uf+R3A(HN(o$nVdc=E0zoi$5MnIco
z3u)=-LX<6mvmo?A-GzUnf7aBb+?oc34VcKp_&C{3u<dd6IN`Qk|0E<P;w$=*PY|TN
zwE^9Dcr4<m<eR@E!WOiov-opJC$9OoColgb&Lw0Fs8{iI$OCXp^M{uq#My|qjj6{G
zQSf^6W(gY$5p!pMaru+X%#-Yd0h3<<%HTU$hVBw^3(g_{S75LW;LivX=2kv`g&<^c
z<&0?dJ5b=@VZDCyW~ibHCkqf4Y{U9sxORf@0Z3}2O$B*W=zCd}u7z=V+Yp7|{B?As
zMuvtb`^9$Rml5rE!K{ENXso1Z|Fv~~IcWzz$=!87WEJfzf=NIr@e|iyjiuD|b#^8K
z&IM-X6EwFUpZ)W5&DLNEU969sxmQ}(ZhUG%GeAEHv-(?GYq1j*uS-g*Hh++Z$T!dv
zznrdV9fGs_>r3-0I1$8|8*eZ18u+f?q;Rvb8OF5kIsrkoD=R%vpji_Ld&8;YF_2*^
zJ_<D>e&L*+t|cTkzU%c9#H;$e!$0Z}{?_Ge@{20dM_&!)P`UT4RhM~Mdb%)uM8ynQ
zXg<!6rO3hXy}Z+H9iE}Ur)%2z*_neUE)VYa>7ya#*iW(UAcd+K`4T$OwbP*I3RjA~
zThLE$va7zi3ymuU<>=_Bk%dLUOIIAK;kJK{T(798sIfXBMV9q&Fj$!*uqnq1Ck%#w
zDwcx)sfK;~yd{bD0$5uL%B+@0N0|P-Ox^#i?mr>_x$6J+;r5G~GnsnX26dS)23*A%
zwJa@KVTSmeZ=u4IKvlzUDMb;Y{N&4i`uw1hixgB@EqtUW78Z~5Y3u99B_`4b-*UxQ
z5ijZW*7^Uu!L`pWcv})a8on0%b)EJ`&EEh2hcVQ}B1Q%Vx_t2%VQRFQ7TX^PQRY8-
zG{M0TTkW`{B%&-xH+k9FF<V_XcJ9GJ^SkI)QH_(@(C{1^ni9XBYV_WdCtRVm<P;S4
zc6QZb+ZfnMNlE?r^QW}DTmg;$Damur@@XtQ1Tuh1epi?(^6QIoU%Yq-m;vZCqU_ss
zpSJPYZ>6UBP;}wA78C$;aEfw%(N`YF(wQ;l-By|7EuWUEC&9m|YN1Fdk1&7heIlz}
z>&zKk$P9CssugUKb8;@|>CJ6*Vd1W=;WQ$dV7d;w3osJ^7S_zyS<S-fR9r#=NAtTT
zf5Q-_u4wU*O)<%VtxhV!PA?7ua4U8KnSVe))%KR(O?J%sku-7Wksx1#CJU;;V&}8w
zNA*QqM7!m^yPt$g8foM5R5=SvN}|HT?3|p4+LNlPAQz{#pJTJ%P=RIV=cgqltpI{N
zUu26x{u1#g$ad4DO0Z1-G&=C2zDPfZC-djZb;_b=&qh#-H2gp`Gr_nSRbgF-G$$uU
z<jTlP-Q9<#RRFi3`vzy_s=NE)uG-(%Kt{Npg{1!D$B$Weps1*M!^2Piq#r?oz{my&
z5wZXG)^>MafN>TJ3kwP}Gcz+b#lSOfxrz~t(I@^ILmu7-<uCgkU_{yv#L3$b!6ZVI
zftV&J>&Jfm)op1i)KyoP*+%IWbs0u!YC>G}5zS3a`yWKbv+7I%zC*n99s%P309EE^
z|1Bf0S6ng<GW|sb0_IVdvkbIOSIt^7jG*-ZApK9<`4Z<J7`k1xgr^ivErt}z8Ln_a
z1(?f4#wjUjq@uzlD0t>Dt%tFRii!$?ApjwwC!%_rq^Rh4ZH-`z`WY=P*l8hrgZk4y
z31MRJXS}~E=NHg>W-Z(8?4cq|7rRWA=6vOfE--AeAxO^_7Z<&4!SX@Kdt9`p>0)z=
zplWEC0JA}$m{_&`hQFvSQGf%&1){OAxQJ?LDx$@gyBL8(MWq&-8`w8;0ai9!Sd{CI
z^W{^q&NBVQxKS6IEIF;yL36lKpsxD*?Mq9ZH`(DeirT={(%H!gr-_5?ab;y7pgl~6
zxrw)WAEvM&vg#TdS6y8JnnFexllHSDL>la;#zt}-3}Uu`!;+p*J#{Khtj_9KL>)X%
zeX_DqSYAfl8IC2(Kxsc$g$LBo@CUeJc*9e%cQM$LvvYHBaJq(8<=-1~Mgc&U0i6&g
zIyepY4vDGVAm20+o=PMFw`A};qzav%`>k2`d%mhXnMI>~%zHcLx%#^pmh85a_^n|T
zmW>ejf62;Lz!nhcAMnJXg9isz4Fn_v1@Wnt=4MS_ks2sSpBvaC%%cQ^spG@w=<B0?
z-rh6bZ_%5!rplI3fAtQ9OD^=Ta6}w*#Y_(Mz>Gf%E4hrvk8z$M>%AeP8Chrb&S!qN
zaQC-wPm#s5v%?5O^Q)-9pr^yHWf;XNUFQ@F_T{Yrp|Tqu!o9r}I-;?iOx<ID{_Mp;
z2RMYH181oMZwz(i+9`YNx;0F^ux(cSw=i(R*zftu+n3I^U1KsoJq{C}U}*d^jWMIO
zwU$Q_GY7;@5U`0XitPbW{(%9WDuJ}pY70hyr|urS(N=*3sA7Jsx4YZJxDDkt%ua7+
zVQ6C~{=is}XS+Als0CxC+QH~A^YimKcBnf5a~2herFD09J_F8=f<Rlp0)Yh()vWuu
zbDdZ+@|%P(Rn+zHvz&Y#dfL*LzXE=vSFnRHZb?299F30l+_}WZ*VYQmn}npLyQ`~r
zv|)=q)yodAni<dh1_g8Fl`oufMTLbir!$utPr5A^acM}hq&|D${VqN=R-AvcB{ZP!
zF5v;{CM&BU@JZ34pZEkUx~rq3qqB2leiiNefiTr+-?GfiOArO(IbtdZJQ;18@h-)0
zn~;O)0cnYh+_%-<-=9`z#m&ph6$}Ow9UUDvtp>V26htWCvK(B!ybAL2v@|sxTwI_h
zG=6z;hcc#8fHoke(&3CjL4%ioAeY^{cryb7bRonSjQnm@<DMCz^{k2IT-5VK1BRL(
zn0tcyVYELeZS^<e9Q^o;;R$UD15tzwcESecrYNrAU1+HhQ=V>?C8KqI)|X$F+fHvS
znYvO0qXm?gK$<XhyJidDu!PgF=X?|u)dr(fC_Pb>ejjVz+}NNOGDQpJj-U(Z8L5Qd
zzIIy;JzJ=D1^^dO95{o~qhVx+hBO*}UJMMUGIV9kb^;8s{*?FP1&Rbn|7nzogHUPd
zz5M<P7CXk=KWldLhnY9sgNa93A85FUCP4IIhrq{vpURhCLPw_!i-Jwn{^7$gNI#}R
zMmx&h=B;;ooYq+CFmKY<cn}|WTbZD0;j%MY1w%8~0?$2ES`cjbyR#!oUnQDFx0HG-
zoLF@H(TN`l6o^jE&7A<~yXJeum(phW#&~@;M*|ffv->4tm35l5{MElz;Qyme`EO;*
z|Lljlo)~SBWuS|Mo^Wx+eg{d@|G6`gfdH^uj0K$?9sO5kGSIxY<-f`5+)eU_-?eV<
zbv+<9$kr+E-UR@BgT92I45d4IUi|8We^OdKuU;*~%pBecSRc>n{z!}1n3%zV0g;RE
zuLDWMgJBN_{D(pOCkWS>KcPE)YJ@HWbDxWLcG;M)W44R!zv*-bT`e*;r1N^GPSv-z
zP64wDSo%xyS(t`|<oC13kCh1@FfLGo#Up`uLI}!W;Or+Owf=bbn1n=LK|xzv8~B0%
zPEf96Uewys;^^Q&Nlu;&p2rCSbgdX~yIi^Ay|qz><^Yz&4IPQ`@fZ}saUest6b9(1
z%y1q5cV<HdU!VK2<I&HvW54GAA-TQ*-(`eC{{7*oRh8I31OI7n2gr)3kJ<*H1>*ln
zvR!-jq$DQ;*cqFcsKU%1p&2+PwxzEjCd1%#N{EO6G6B+v?+ar3hXWm+xZB8?;Lq5H
zXiY8j^sFo`GyZt(xE@G>ZL=~198Hf8zo#-%?U$>RH&3LR3K{0V^WJi3fAJC)E<m+}
z0B#GM2+XMSudl>ZR&Jr{Kgmf`CLoZ&NKX$dN{nDbgyXl5lPWj#^)c%}BAb+)48UQ*
z_Y=v6wUXuTM;6D##UG-H$A|}QZdAQ2EFzI&PzC~OCt92$_O-N}5xTD_lpY-&4G+p+
zv*@nr4j$yQjt3`2?n{5a`Ps9UHa6eplSyWDc7+_ePOaqNkb76t*Ut|J5lYoxOz$m%
zHYJq7+xc76rhp0+`^4bvSsF6ZVS5luuqFuKtNvRkSu%y6{p14Q;@GjXoH0<!9b#fK
zIe*^cs4&U<-bW%kA5B8}{vt2$l3+5(g?KimZ@mjMG6sp!E=?|s02X;N7$!hrQRkkD
zWc)ont&{Hw0Ny~rW8nP_6dqs*H#9Y&y&3wcHoLGO=Q){#-{ORTYii2;cVRcl0@(_y
z7>Sa?F>omw8_{wGyyGBQ(81aGx9DFf-;e)9DGhLBd?JqozjZO+5&xqUret&+e?L;i
kRsMVWDa8JlKipzVde!J3`DyGr-iSowq_%3l^2M9~1?GINTL1t6

diff --git a/vignettes/examples/timeseries/timeseries_anomaly_detection/unnamed-chunk-4-1.png b/vignettes/examples/timeseries/timeseries_anomaly_detection/unnamed-chunk-4-1.png
index f788185b05a7a9953a0bd45f401c82f0e08167c8..ded93aff97dbad28f68050f443ac41a89ef41929 100644
GIT binary patch
literal 42503
zcmZ@<byQSexK&W3K~h>mTDnsyLAo2HrAxX*N=h1r7LZ1|L5388p-Z}lZikL{_^tQ;
ze5~af9PZq?_na^GxA(r`DoXFL&`HrBJ$i&CCo84)=n>M_M~@y?q9TEJ@^_Fc!3&y`
ztgh>$M;Ley{~nJm_}@Hw^zxCM)LRYDjJ-^c48lo7`(bOfUonPk0==|~RN7|(!l!0-
zjHthf<P`|8a;a^iP<nJ|ObKY|nN{=tKI-`LIYJ|;U~HFd;JxwRHb>*slk2tqTP$uz
zORJ?8i^cTQ+o0jK{Z_&AdG`as2D!8XYH$^g9?b_Te|osAM~^N7+BuNH>*vTvU6>DV
zupwb`KfE@gDup&#uHE0=#f11I8sC*t-t25J+&6hbuLkaua)+&MZU+zRSyL}h#*lZI
zpcCJI-V9@86Ec0p+R3T7=e3`Wyw4^$D+)^)^mbO|6b}Tae1qQ6$6*y3kapW(MRa3o
zT$h)9Z`N;yBZhEBt5@2|kshZ?tt&6F*$<;C7cHUrkzD3^Uh{L3R6<s?eszgkj78xu
z$F33KVTJUe=vn&&RMBfaKv*Z$zXld$e7hSWecjmia4TwPN}+;(tuA6S?3Wyi-l9@#
z{JPqWE_8FK+1@3zAeBH33D9W~G#fd3yD468vypd;gxcfpecM);LY`L@H*lu;ex=uy
zAJKMuCMLW&ylIvC7@V{n&&U-f7$k$K!tZ;oj!8|RdyY8YFTdZ#%AE46qYuW7Z>U(%
znYOxEo`1NQjr?cpx5luW53XCA?8aHgmvT_G(%IvgfP(wAY)AjPhO-oY863x(z5Uej
zLg>}#Y|!&yT(xs{|Lvp-Y*PjU7DL2rl=3-2s&T&LjBvm~vKamy&t;ksxS%C}wW+G4
zieJrTKQ4B)vofwG#Fws`N7QKiU51mrh7~ifU}ee;+SXN)OKsPy;#v3Sh59T?*iR*{
zwc5@rHnURiPUhr2w02gf^$?*A+5WDE8|=d&{=VniGfPRT{SM$}Uuv#cK=QwwhMfzv
zIT%)bS${pCbscKN<^J!fS1DAFN?LC_)a_arTNL_QsPf-Ns=R#ShrBQJGi^~fUy`G_
z=z?R3uI>(}m#*)!BJyl4ql6Eer{y&W?yRr27K!gKgR&#s*}PY)7iqI+aV6d#=LWR-
zj&YRcv)rHFF!=a|YG;0__F0^__v`&8n0K9|jhN>RxZ@+a|MxdSJ3E8xuDMrByIaXu
z*lf>h&1jCwIM>^*r!;2Kf5W~eGLKjM?i!B4_6uHu7GUgJ`{}kUJxtN_r<8P&!@|V#
zTgNoS?Qrk=slhhak&$=PGIHO{pGAs6xcoTlmy;q3#g80JWN6t!9=5zO<q`YaJMK^{
z`^!C#-BQA(2*UiJ!^7;k3QL`fxQd3pO$xo1rqNh&1p0SHRO(<~`yr_Rr)ef}M5@zT
zWxZHFxpf_{AEoZ;>eD7!%9QED+zoM=iW2{uV&t{W+oKcdOVr6lmjEbBp}*7h%@BP~
zVO#&vJB!U$FX8KZcZFx!#w|M~2g*a4(EYf@%_q1Z4U7bjGLzViVj~mLqOWoE>p5NA
z7{7HtMUEvE+OPKzPaPYiYEjsC?(0A0_q~^`c(eEZ)6}GGL!U>zupar&b$4qD{n@Y-
zBaf*>%cgcx(XD?R)s`DV)80t;$ZhhXDw;Xj|H2}Dx9y<orvO-iM=70-EgDDjD?!L)
z(EH<~2n*qBBOV^*ezE)2Q*7T;#g(%<GFs$NoIkZU`HAA5>)#E6aZh-yr=&Cqysl@d
zT-hnPd8nv^RSv$F%J44MI@p$(Y^&A}r#JAvTR6#vie7C^oV#BK3B!k%nD1#g9Y;cR
zawlrN)D`QK8F_cS+~0{y{F(k?z4rUp@IL7q)Jsx++cM^eP+y7)pOf%YdGR#f+hg&2
z|IQ_%Faqh@`<a7wU9)nb{q(0x8nzpe;s#UO&AZ-LWCahkF4o^^zBfbIQ);Q&Z3%Yi
zV%C5WMl369!?n8VTJI+q&^No9)fL9WDV>I<rsZ%}ygh*{(Y*PpudpQ4P^+rwS59)n
zP<lhnS}e6aH&Wo6mw5wEyXN+@!~+J{lLcv626Ck2&VLSD{TVvCvhKQ8fYdR~=bpc6
zxby3`h*s9*a+-pa>aksW&fG~;m2ZiyG&_7qd5E%aRqDkXmt8GlNBxmKgAQr)GcmX)
zbxls@DbTim)%Uea3j53NWZka{-c$to*<|0W&!xo?uTvA~-EV7KY|aO0jX?is8yZp5
z61>?NF4ZRPga74wyU69SxY*_~<g*jVUCEimZ!Q09FdceX<vQ%l_T0}gUCCmY_|zl%
z;+lWhFg-d}P2|)m(|u<E*(#=~`f{T}+|y})2LEp6IkwM#iI8=(amZ&UTQf#GHWKOm
z%>QyV7X$r$%=@6wbL^~>ABjm6e`}n6!}5Ea&pG5?|GwML8HXDPukVN}H1DQSYY#aH
zZ5*Th46-tEYUA5nJU$n{+v+wpzU#~)&qkE4E4pqHBPfImxJSiDY+;jD1Vh<S!Goih
zZBn)j_rxK>I8>u9+Y{;cBKyZo{LyXOxA?l!*|*miw<kw$4mq{?3~xryZ_A+tqJJ$O
zVwU=TqFDVB&32~QLaQZ`A!>J;9ywsNWY9YG=ewvD_Yc>0Xw-5iygR;&UWTMkg)-)#
zFV&bD4!A4j&9W1xbMBWN6#Y=C<>ISnGlgrdcKJ4!oW}oPlROuMp8pNY3NV$3v7_7g
z*SOcb@|w06n?ijk_1`d)q<-zbXroW0QnPku?GHUoIk_L_kaV_#4CScO3hA(M*B>x<
zzpL7TvkkW`{ze~e*3p!i`r#2H``O=dU+GTPnX+H?rTBP5Pkl0`E?3b~YSB_cU0RIA
zyk^b#ajZ$P_(gYH9R^7~NI8mZ@zJP_G46+EFw&I0FTaU3A!fY5ZMP%KI#JcAtPjTE
zkT)TQTf{O(F!#6~79>UI>koHCWP>s9O3x0Ohat8W$;0uVd;hoJi&#D!<@Edm<iAEI
z{n-TN;SJ))p%2GBGLR}Eg_}>6g<t)|%_%Oj>vU0P%X_0cB_VxFlxXo<W5V&0k!6fy
zpqe~dW;cM1Qd3hiPL{4}DO4h-+n;g3MOs?A6RIfCF*!LYk@%XOU3sq|B9|^>IW3L!
zuAe!A=KYM5K*nFk1l9U7){b<jrY1PHuCDGTNmG;8#DZ5*US_7bxw+dvy|Oy3AXC7Z
zW@U*!;e%>Fy&Y|WoKvTDd1<LDWLX38-_^f-{rG=j42@tHt2PzeqA{tE*IF)~P<_CC
zmCAf1!^15-D$1rS_lw%QTHf}E$r=@`z5Tf4QxIiic!h`}<|oL=*=O57?`2A<MKLCS
zIGj{r70!LNMw~v}dYeTS;B$Xbuv9Tm#iCoq<pcYn@7Qv<Q?@a19yI;milW6f_L5^v
zK6P=W3%>CZ)C~J36+;;}i#04IYdbdtJ!p&&x^%;NNAvvDh6h21CabN<OBDI4o=!P=
zqnk*-q$KS~Lb1%+6zwdrIwJn!yHa8`TA{8gf#NtFQ=G6uFk`8;UZ+b{b@(lxNlqgk
zx+>oa;T{@B&yluPvV~l{R1)hZoOXcgJx@ZFit86-P^p|EUMtOWY-UWdn_c-trf2Kd
z`EHo}G^E3O+wn+379FM5N#`x2r_C1_ijJ@x(wL@Zl7%3Na7sIlvrWRKeJ9%QOg&Sm
zB5OzEHN2v%z#46?$^1PLb>fjDKA8bc@DUC^+xJ9uzLmk(5{c4snD`-ytn^@y#<gYR
z>E6$DW4dzn=hD3=nsY;rr&HyIsST*&C^#T)>m&Gxu@;)ERF(2%b^@6R3BT{y{8W9{
z<>Tt=I_TIwB}7Y0`*=GkDM?pXH#IdC?WVrIp3lL1M>=_9XJ_Z;=BC@fl<~bOT6sxH
zKMb~BVFv=w^dX=IT)>;<<=7}e;FD-Kd?prz(#2(E3OX}7tjRiGN=r+@qPV-eBS95q
z9^Qgfmz2oR#C78tsH?A?UkkF7me_cCd71W`-iw)7_=y=SDJg+tb#xWSY2su%`==%+
zUuAp%E6tP+F~JUzKx25JtHZ+}QBzZsNhfeDXo)zApC+EnbnW;IGYols<vzSO+cHy*
zcs2U*_-f&`w}+ys>VKK#nbM6_bXaSWH+OSbw+XfNo^or*NMVr+@)w35`LIKygrcwa
zze~x`JjCE5x|Nytf`$QzYw`P?`}?K!4BG>tFp0E>KegAiJD5HE^8Y9ClNXjRv6x=J
zR)jy{Y`<mIt1YUmRQbgGC^RuK(Qc{D@Dp?QqtH~Uxdz9jMrYc9CKj!-7o4U~0-BIW
z#QksLU&+%eKq{?<DMzhMWa;p9q5h&0TOH4+%8H7Xvp@6NPTd@J;oNLYO-#Vjcx^9F
zHptCyFV<qne>4UUrt_yHC({Je#Qn*O5o6}#bcMsbw)0aO!&%tb+1c3O^h1;!up5Z5
z$HAN<bkO*qNHwn;TZH(vUa-Ef<#>VO=*Pt3?vFe|&a2%P1IZMk-UP3lWpf5}8ypt)
z8<xJ|dhh;hz@`#|i9e?DzgX!E!RQNE@;hH}gx)3x&5Pe1$7BcG35uQZrkdqvI^A7v
zuIWiNo-g{au&~(L+S10Ms*m2J^4ZU}dY$4*Feqhs?i3eV4rPSn&NjQ1w%v|x6c=Xc
zRh!Gri@$P3_d{yEK4|IerFk+;0`Kqde>nZc>A^xvp+aisy92Wr(X-q0B?ezDCnOCG
z4H3kIIx4laDkeghG&(sY#Z|s<Tv_~f!DA`z{u{ZV%l>TLM83$$A6m!O<KJbvHOZ{H
zF8{VhS65f@8I+bWslW!Dv0fe>2HgAfL=cAIP+5vns%g4SmmAht4Nvjd-`!mPFm>&6
z`g8_1XzHhauhRUVI9lZlfk0gJu&zdPVpg4_^rf57kh0whsQ+lL^p^RlZkw;*9lOL<
z!2R`7fnqwP?@@=gkyl8_NA8EcueQK=_BJ9{Lt8s9;Z<aqZjAWd#h0fTFXZN{%-*T0
z#@(eeWc%XbV2|_#6<^|pOU-lNZwPkzg7yA&e|LRxdUtyby@ZpAxqxeI@w;{pGO^qd
zNKQ-yQGt=!%8wFo3F2fJgvd~aAh<_!hS|?65|N-45Ji9>Ihd`pUF{A}pjVu2@n9cY
zNK;XN_>nZsB?R$!v0a$G7;t~PndS4XuIwX^o`%K`>)9O1Fl-8;A0<016|E18x6&0>
zgWv41&`crh{(b$R_9LS{2$V)=GYgAXV~A;9Ufvv)uOW>mCnw0yu%+@;4MN($1+V^H
zQBMQ0u-lGf0xr$5?UMf4+me#in(tV~T(2!8D;j;y58~tTx=Y<DKt3$`97?0cRl6f+
zal^pr*4r*oe$?4cmm4(2hZrc1gT2kYw**_3V&u()p({l*r4(SS4r&Vc;UQfB>)dkK
zHX!>?5=ZpY(em5i5fDJm$|b{_bcYF5Ukt3lsn~8)7>(X?KV2G4g;~7E9)(gZmFel~
z3W6j6pCuzxY6c-g7D@Da_Lt9jx?`JQz@6_+e%r43k3;TZks8OPw#8pQ33FyuDg~)!
z*0absBA@e~do20!r!xI*cC%$Anya<OZ@G-i&CRtQFCf9|%0a5ot1GFjtOS*Fn-tr0
zuRiT(MM=pLK>4bwqaZo{TK_Wd!!s!8>FH_Y>NVxe!fqA?x4$)U(4}(pkYd};I;fM8
zr9Xf2;wX`zO@@7RqFt&q>53xZe$jXPZ(R8pa-I`L*$m<&2CR=Z`I4x#_9{qMaZo!N
zwIgW`LIqba<T6jYaKsanl2$H6rm%e+b=dNRZa~7^Zi?SGt%Q>OGZWL))c&a#vX&hM
z;kFtAVfHVO%3wsjP|1S$c#H>Yn8r5ktdAl4=H_%*XM=>>a$<5)F_p^#>_(K7<3ck6
zWZ<hl&!diKgwkkXT`eAmmUPcg&*mJ#-zMipx$MQ@DWxUNX1(9`=Nj%o4clYgkZ1o@
z>y5Ap9x2f*{j>00zus;JtR8JRR;O#It})I(!razYUl6zoyV;Zc4f8G-822*lasNK{
z#!$udm|_sKX%OU>l|?nH(2zDTYHC^yXBq0~d^~~_lfmQT<4a1IUFq?sBZ~7yy#E!6
zm4qL1Px9<XvbSwb6yrzsbj65WZKW9dU+&cm$2&T`{D%1ak~Eq*LIQ>&@e$-SNTcqA
zxG20gM31f0&Q6EL&-Z5V<pn=&;Z~)v2Is1W)rVHOIXiRGj)K@<9~}MokLv#7oxS~j
zwZ(u;G%2su5IGTEwii6r=kVA0NRA|97DmUc86TOY-?L6VHfKH4s=wbTgxfAwF=X(_
zf;Ta_BZz&|MSac+v;D6|-%(ZV&aP^mgHlLr<Y@v5T2%4`Df1Et=*_`&Z7r>{9%ADU
zVV^@(exxUI#RMLmPhiyu`d}?_*Xe+0Y*>`*$+1%$2Op1!j3kL}mL+JYuI3_*CdK5&
z@T>AuBiLB(2%<9s8SD)b#hqdcdWWdU6!us}&nvN!mVTP`zBj<4L9bWERdzYeTKWo9
z7f|=l_?LXgRlb3?o;P}XeZJ(>g~ckdr7UvXZCml{VTso6jofD}%#+VEJr+$Cd|^3u
zG;wZrUUT*K0#97l|2!9|ckjDm2PM8Sb9%KWa=OCUak6PV-qy3Jz&a!(<a{YW9DBSf
zxrh<nC|t3PiG_t$F|PSz>%nY4hx(zx9u~Lhe5Zx(Z*JFhXjw}VxFfX$W1syxu+w6c
zHLL^OUaySfUr(_9pc3`gP)!8w^LqkqkduG{m-}EEuL2qCWxI(t5_PS|;iAW)*CuF-
zynnnmGd=#8KaRAB5}<#JG7e&}qjbsE#>Sy}n_gyM7uDMWhf1u~evVIZ{D_Emit@B0
z=oxs&Kvk8~>^J5-Q<HhxOiSoxI73!ZdHEEM?DK&YCxi62%#T}!LRxi`H^8+K3~SWH
zMz-M8nEYN4a*LUvf3@|)tif$pi-^X3?{AgQxx=kWP;m8(W??iLKWT;_wojdNG;f*q
zo94;Gy_d6n)lpsB>re$OZazHkw+g(k{J!L@XlQDJf=M3GWjaV--!Gu;{P~^u4ULKK
zPGWL05i#+u{R${oK^_&pCw&ZGwH;`KvwRK&MK+#L)^IqPK>v-&dv0(DSwrd)D)DWO
zAaZpdcofEY_G^a4y0ZJ?StLnc^zlwO25r9obvEY%ODnkDRt{+1s~*@wmqs1T77zFx
zMFQ#-N*An_jd}s^(<hseq`JU~YfWc$X~SH>qr56{jbPcEt-P3{fXf$_pLXWm4RkPV
zjz`cN`ayg6Sg3SdO(p{cXd#{7af&4}g*`S6Hetrb%sd9A;y2-dnZA@}8#n<yQG%50
z7^3^`_o{za`8!%sd4t1@xN38+6q9A>$59%QbS1-b&BEeCxDL<B{q13ZalkFFglzJL
z4r{3G^Mh`Xek6XgTc&(n<d`+?F5gH}xyS=PI6Gj5CQ_7ZUs?MUU?oR2obAu`N<`=u
zO|D`}c5dZTu^rCTgpu<)ekD?Qmc3cx2nppiW}V#w1@70UZMEQ$v3z-icbtM#*W19y
zqiDE~4xr|`VxC7Iw6utXCe19VkJ$q*2Lh3w#Uk|Hxo2X;sp;$MAAZQlHRw`U7d@X>
z*U`D?#9-$Tb?vfjA)pNh*r<JRZf~XrAn2}@qXcEKa1UHgp;qs{_*Wpy-W9y1Gji%}
zcCyeFbKldM7x%lrJ-<FbEMyzyDC$I@S5Qrfzlj3}Qi=KgJ<@ZDAP<8FU65KzF0M}E
zkxR@xoZi;pA3KA~@pBhYC0GYQ&;K~~1C7hxfZptjNmuA#hM;Tv=f|I<`BB-uVHd`&
zQPShJPcew>=IWy}dmkd*aZ#kTe>4LRGcqveZ5)6CYjOnGlr=LPRG?{tzA5Z~BakN>
zmqb%0JPwu~>$J+OS0ZQRWMjar3X;pX=O6~*vnMqoH$Q`j-`}|2_DxQ7^pxP(?d<o0
zb^5C)Ln>lf(0U0^2gNQYN!pwufy1}bE0pMmT-r~9W5^-+3j4YG(WGn8PuTVAF?)8=
zcon<W^mZuL0Ez<e?#rB#n_FF*Fb$=vt}e;A-=)$~hRN^5l#~QE_*#rO8!IcFzQNYi
zby`{)z;g*qoGf{NI?O;TZ@5#SDsWjl(3l$5=0vwhPj{xnBjL3nh*asAZ*n~rMQ?O+
z)~e90fnOZ2m6VVwdF7!N=cC-e2)MhuJ|yYu*^0SnfBO<?yiZWA+Ow{bl8EvL>wq&q
zKB;*X_ZZ)Rgsu<lYS`09=*;BkxjzB5zJwE!Ow<c!Pz$>{kSDOXvDMd79A~O~eBXc1
zWz12Ly-=tLlb8{|{f9pj&;)?qDkrz$GH$d?*9ymWXKjM|&;0y+fr#?^EH=0^N|?$S
zxP_Ka+g}Vz0YXgw#UjH?Y0U6J%N{fg(Ag{ige#IasIgQuGRpGjGYn9ob8~*8R^S}{
zq=_#`HJYZU!F_+$D%+1wSeWeoybVC`?Sk|od8q>G6hXZq)7V1mw96+&hu{wxu0tLA
zWG|?@&A!<}j#J>3e9~VrzAZfz#@l(myo)DZukXH}N69e?6QH9l@oZ3c-Arm5$6;|d
za`z;Cz8UB=<BPcl^Cg%7X6o(dKo^Gi_I*0eS*f}ENVbEx0}0VB&rBfA#Ktx>Jk0;c
zG{>5isEc?4GI_Zig!2MTn!&!ZgVMHua2xbqPgfW!bwP?Swc(PXmR6`TZ^!*>R#tP)
zj#h8P4ru8l+wWt&-|(_IUk8E?2;gPGnw}pLCe`2P9{EXa*a_o?Tm8IYBm9nwQI2DD
zdNsC97eW4!e8~s^kw9TZJBNr8^wX#7#5!1rY{QQ_p8A~anWUj8jz>g95V0F6d|P8!
z8YirE-5dh_T0=`~$1GhZLJUdHu$*<^T@2-<5~XcffcSx$?t~}T9&c%oqnH(cNX6mX
z3s0ET7Jz*J3e%jnI(QXnOUq$_ihiaaoiC0|_HyEFYHTK%T}!sm>rN&G%6`%wvNJQ!
zLIVQ4yqYe3?XoCq?tI)`<U*CRL`aA|0G(js#rGQXAq^ep8{_uS2)>4iShk`l3pZsv
zP8ESW3A8$`bS~ZBofCxUDL#+Gu=q(SA3ts6gJm3)CytvgX%8xX7?po}7~zX~Y?1u#
zPL71_H!M#{5QVeFk&9J>%gdbL#Lpl4rBQ(<Bbzy<`J!5<sbf`wJvS?+tAVqK@UASK
z%gx1czI;mBA#F~%<9v^?WlU<FiA?jX^xiAw7~R7fG=5mtDdgD7g|);|=yj$vqyD`y
z`EjHa4V*WYfcjbyp|ek)i<*k9;>Iw_o%EG_osu?2x8j2U`C_FBvcwi>N!%HQGQVVg
zfo}2V2LV+~sxO~aF=LkkMQO%*M%Qd#CT&@QYDF)MCJ~@N*|UW>m4qhrsxt^;O<vPO
zukuKx9xRI=BU#vjlW4m|`$v+WO@u2<i?VcIa0)8DS>;wk7czVk=(;&}pNWPBk0yP^
zr4h77xVW!5xS}k&4vUr%2lP-()Ie5wBk8jTg{o04B2OUu_+8Hw$F<50nEE-=#mAGb
zcS<Yb)lJLb*tkKwP3AZum$wR{`Jt=SF(i&(JF~-0=zk+uP)gM9X0Nor97`IgR=|el
zvk?gL!6rB_`hCZaHd2gdp&1M0+}Uwqk8h<g12fnMp0UKFs<wwKe_u2r%07y@+XzKt
zmyy_wo~uE%n8P{2%(Yh4(D(!Z42^ya3GKYuS**#F4-76CfV6OR4j@~vU(*UB3el+y
z+`|RKqBqlQsxYv8-n)?6E_A)`120J*0r-NLt;TH5s39jl-%4dd!YAG^lu>7G>SInA
z*vRrLcr$@Edvl>#s01}@K0xD#i@&xe-tRh<wLfztv5a4!bNb0?RANRL&P6UR|BJGy
z#AeZ{&->pkNPfU6u0ro`**68@-nY!|t}qc%L7M44o)^2psWk+c+z9j7{JYS@ZS<09
z4m18x!)P_l$(;}6GQIK-GlswiG|C#!e`51Hlcjy<BE+MyPb$wKnwmY&y2;vpu=G4K
z(r5z6iK`dJkodZrgZ3>Yw`qOFL{<m}@vL%Z)z*M)V6(|F@h(r9VM~2&twuJcSg7Cu
z2Es+xAiOj{pOuJ=%UpI|onxESk|wc?D@f4tc&pG*yG$og93pC25SmIJ;&=qn4Xz$E
zuI5YdzDJ=VXtzSIH|l~p@Vw+m9Guam;&;fnyF#z|L=>v~wb#JW^u=*EzA~#(tIj=E
zwu<{aaoTGd;V!#66&?q!BGjML96`OXn8vDiQ%W&!=ZTGdJ6<R5ugeofV_BU{8wXIJ
zhx}_P(v{_D9cLyZdD<-%QoY7@xB7D0Mi?tJ14-t&a!57S9p^!s067O9@u}*{-b8~J
z_ky5hyxTg&;r7~)w|6gRU+u41xyO&;+QkG^zA%;gOL>RWZzTtdZVYJh@!q1Aw|e3c
z5)|Kdiemhw1b5G>+8P_gM&h?ZRO)j)#&}+nM!PiD(jH&{_zUTG`}mXutGJIRJE<2I
z&~D$ZPw8rq5_d|@P1GxtxIe<dZDW^#9D^XWl=+%w_WS4@lKH~YJ%N@FEst>&9zu?!
zboDgd!omV>K{FAe_iIEXna~t6^ZIG;XA`0raLQRcvNmdG*@8HHheezD^Xa*<zq}{w
zRJO4y)H!xHZ_7;hScVu`1I6uTs`D4EiZFQFb1s+VzjES;9DSa#mu@Y|NN`2A&Jnli
zd$jwllR5GAJckcsqP<1ZV9k(1LZGcLD4I)$R*=UHXaDB~V9@y5%W6~ntx&`=&=BQb
zb4KUAj7+Ft?f`Zvxn<^Q=(#?HbHCNgTxby_#Nc)gcDnTh)Z8{C41So=Ib?Td2lITr
zk}}WlY{m*TXKYN(g@!_63!uV$&mX3&M8duom6twZk<e9iQ@YV610sn9JBaS5r5o+u
z4@_6oWh4%i+`G6TZ_Dq91k!tp272rkTT_b-!VZ{hk&BIAm_X~}xGA|Fur?Tuy<NZJ
z3V9yA%qBBJyZT6I0vIXVsj{3`rVGt(@VR5$r5<CgVOXYcO0$7#*IXkfgeMd#AIK6<
zCr57n++dYcpO<E*&l8YxgT^NYN%qLLF{!M><>zuko{Wh<OG+kRYQ!AT$WfN&i1NqG
zptydG$~73GJOx-xXYYB5Hk(zr6kAgi$#K6;S=q4F=0j7I&Kr%2!fckv3>em9+cg)+
zpqEW|qrTao|Ew(nciWx%bvh)hltrLPD1h2StS{)r#VOSj@Rf*liVC!+MUgDm>=%K&
z@5RMCxfAUjAszsrxNN+~4mXL$)qgcm?cds?ST4Dx#hu<~hx`E|Xd)&s8&v6j7(5OC
zso&s|kR*}kNYu!jsOa&AzQmk@l)700bZ>=kZ)<y<RE^?MNkRm&*$5Ubf0;rSmZH@X
zcTciHnhiiCsW#G~bXlQH$T5;p*E+As>_}YFN|m4Div&mc>kM-;+|$vvnnE%Ye&_6&
zXRY7YKUtsUU*l2NVl6*`7<;ZfjgT-js@~?jUB?|JwN#3P0;YQYtw$%Bl7?n<kLgg&
zuor>0DnThlut0}aK;2b%7Xv`N(eo%wk?gTf5<H~&MP;r{eEsy5Od&Vjm8XNBqlM5I
zcSSPK`fbvtzH9alA!kV>CRxlF&7k!4Y>H2$i}O|V{497YktrN^)(<Fnngsh2{D=RG
z+vSz94#d-q_OG|LinvrqC{o79yDj==q?9T4el;F@X@P)7r!T*GiHn+H8YxHu|G@FF
zo~7{sC8*<x(F^RbTYu>3(yU|11g%{Bq%EOz)unCx-y-~QsTWMPr?XEIfAX(yNjdP4
z+Q(K8<>*%rwIDb53#G6~>1GdGA>cZ!mZ#ceO-eg8leWhKGGD9ruFy-TSH+T5)9X};
z$Sn0z%kC2jtP#v}-%nibeneLn1*R&f5WI5uGKjWj4q1bv4(V=cNyX5~iDeI8I%X1w
zE5_r=It4LDPo|TVl)2^r4MSr}kRZt(_pHJ#pwA}Fl8SIdQg!C!#3f95?6q!2pOT0r
zSq|f`l4F2*Luz%?+a%)Y(xG~{-8;%xZL<aD;nl45O>-=f@f`Zt_?WJ+N4&v~Derx9
zfav8s{K+hOP<3*zsHn&j&@n2=?<V_bZTv?~PIG)#;xXpR79G6Yzs<0O&ZzI^h~$^O
zX}USqzy>q|L3aOUd;&DyD}Hru%<MQNng|87%#q-1^I&bv8rQM!ho^h)Oj)RX=I7c<
zME%|!W?GtcWo59wgl!Xu{RlC?C(Qa=x=$?(rf?^ryUYvDYl-c81_@D7Q&ZE?*;?=W
z#{t8ZoLT1d!7iN|g31cH?MR)69GpY8Y3NY4T=+$?juLzK>iBGtJz;+aLQ`FB8v>gp
zms_<)f2nn*-;pb~rgj-|z=q^-x}xg-{Q2{7p}B5Fd6|>;@P(zTjDDT_o#xwR+!a6<
z{kY(xjus(}CL$y}PTAFi+Wd*7#?Jo1k|HEkl(4VHx_%;+I4HaKJ!+jgJ(HBz#{2>w
ze5G*c3?^0A-jhH|SJJ397G~x$HcdtGj6~ud3|un06pCfe-JEn0qUlPZXNPBe0ditK
z+)}~q9QwD~AKu6JZjf<qOL*uPx*F_ilMJ1$(Wg+9^TMo~5^@1fCRlQOIz>}%!e^KA
zK1f07QN;_mx8w}qK<d5Y%MG{%URkT$2iv7|_A>cAVYp0AO6sA0Mean|Z%LQ3Tw<|O
zCd}w+m->`1bS8GrdV~`cBFq{R$&i=t%8uFbjvxsgi)Ie<jhvAqY%W#CJ)Inp`=h}=
zFmk+Qk)^GAuG2YBIG@;n`n@8c#jpNtcgbcHfootVwKO1HC55vYHjgB2{9s1SFe`|R
zowYmU<hLH96BbU_H`Q6A3+o}ZB$5d6f#t%R^lCqj)>^A~rIh6vh-mku#r!w`Y~OH}
zC=)stM*j7G^#wtanVlU2FmpPe{aCvhQ3!^pW%ls9X1AEKg~7hQ=QJr4ouZap`~41S
zFCDP0%*`Q=oTi4+m+O3-Fk-_`ad$qaFEJ%%j8<UW!%_uTVKGbbbWB*7HAYXf7DK_g
zewipCJN4B&E-k`SA0wl1#1D{cK8vb!r*UlAuOC+u#;PLu_ZPQeC*lM`qN<|uA$F*|
z<BN$q#Z}bZc=nyS(WlSp8L`_&1vz%5#h-{8d(G%|`^z&iYBh-soB{{A8^4B}vl4W?
zf~2JMzRPGL1+?MS$pHI1hSOx}=m>oyIFpW)&t_Fol0&x{oQVRnrnEy#u-rsRJZ9#|
zMTWn@Ohc2mtaj&b6|J!I3u#rDF(7*J0zP~)EM;`+o7H~dU>U><(ki`yl)qWnWnDHN
zPbPuZauu6p+o(fMhWdwExZw@2-CvFqMexmA5y)M8?l=C>oV%4*626$OFL6h--xbW|
z?}=LOm~eK{&mOP!s{2BdR8CXOZ{m_EYy77Ba?J?mlvnV~tHg!b;H1)1j4J_68yv$b
z?r+~dUgYZc`Zn`@4RDrnV)s|mP<U*7<7X3iu*vM?GH15ykl^3R(@S6SWYlPKLAAps
zF1Kw=b}Nd0zJFQdTAsgoVN!#&x;kOh&q!uPXs669IorBJ7)Lk~4my`PYv>&q)7j_}
zfI9W>8I><?I3MiFaY*=3gh*mnT6TWojh>tS5T_^1D$|X~`i5L7o=fB1RL-85-&L$6
zvJRPL*R|j$6K5>$gKp=5e#R%vBU)aAW^g;mno`RukJBk6o!2&$6_cAbCbc`+GfUHM
zN&IUw&h+7VkO|S}88dn;c3jl%%o+tq<5Rn@71c&b3k8Z~)hQ|Iy-|eb%-Hx1txjBk
z%3C|Vyh7}TK^4;X)owN+As_zsmhr<vS;7TaHNZeY9*XLjPxdTk#ttILeph`Ak){a$
zvn890vO?HpSE&hhZUSW|bmmd<)z!`)6HUJ79kt4%WR?zZ#pi2YJXtZH_dOjH%dQBx
zzf-{fAcpjXBCTL0{RA^z0f<GiA70uM1w3~$WM*XCns4I13vFi4XE-3QS)@zIp_IlH
zgbm8<Y+``DC30CKwVaaJwE$w&-q<p3#S2ShK1%L!l`{&D`5fYRc#_%?CIq<^yWYf8
zzW6rrTaq(-U;Hg|EZ<5Ah4mN%Mt&9k!zL{`J$)`$8l%hE(y;lH;BHyXk(FG05e{_z
z0|cS{#O0t(E~rOd@}VHiTGL|-DaVjt^lY+ozQPN?0a}am{g1;6h&QViCY4x5mr=a;
zajwa4%9=$fx%f{&E~N?n((y7<g|CTvErO-|O~IK=LmZHF!YEIZw2kCMn<JN&!1xb{
z<JeT6f})byV)a6~1o~|#olRQGlKQ&{4_t9t4b2|Wl(w@gC5gNA)YLw1hv7e-KsVf7
zs&M3$ngM)`YBpOHnuS@H?|$91k?(QOr;y(WNKnUjF~{q(eR*xhh)quM%k_javW<iv
z<>d}bZGOV;dv6R(5#AX*5k3b^&)|?tlRqCpTjzj5uO=&SCH$e;xxHnK@8#ss#;_#T
zLaTx5%JpeWMNiWN2H}h(W~aaB#YFly&M=Yw<QM`d^2;fGM_%i(;GiBgf_%+Vtq0of
zZlDbGJs1o)2ySkCWvoYrKXWZmIN(O~KE9gux!Nw=#o~_4a@ZWo)C$;{YhV|9A0u==
z?}~HxJ$s2)zA)<&5nA0>8Do0>3(yTgQo$oYp+FeN%Zj}Q{Omaxk5`9(pg%==ENM$B
z=(66lnI$j31+>W%Fz2ofpSi-wMc|D|{aB?u0M8*lgh?I>$Irf*Hv1-%x3RGyt33u}
zCm==D<p>cG5U6AnmK7IMm!kj*JX~NZG?g_>1{IuJY%U+Bmd0nVr>4eb+WnQi?cz(<
za(h1?DDhv)U)n*~8<)S#IkaG9iZp}a^8lqw;g|X5aJD1jVH@KZg>tp4L0@~#!U9-y
z2?<wT%jbK%MW0Kn7R~qssx>@sj(ZXJ(PaGvMhQSn(cf#zO$Y@f-`Q*ZdY+H`-v2&E
zh<#!u0@TO_cX6q?U9W4r>B#rDg_-Vpphx{PZZ;3fXTj@uEct1A^OOXW`|gd$x75F$
z27gINNHzgM_d_lL#yww4M-$AWIF56Pm19?MT^QbLP+?NQGBGcF?JMScLEOT#C2CpP
z_tkPRZ4|54v}^vSKEDWq%fRb~?+L%s#S#Z6wBpn!9U50qZUUvPgn+345cbQ%mN8x@
z>wTV*&m$8SFezPozfttfKXwqi`1~?r?x^kNWFR+}=9VA6C2H#QcjrM^KOq^|8Ox8<
zWzRs-b3srLjDEAzox)+!J*PDL9o08HJYA9J3kE6nrggjm35r$1t}=!fm?%hrjg&P@
zhlN|5WII%pX(qZC3ck0-);ZQ`CIp?Y6MtHh9tb<M9?Jx)aJ*QzgkTx(XeQ8Ym(>iB
z@_fY2d-6hJ#(PvTji>4EYIpnK{ACTbtW}jxST%w<wp6t@n(SS-DshQpsQEFmb<O4F
zrGVb&QvM!t|Fvj-bPSB=08g$+zq#7$d77D?9$AzYJz1))t)a06rCzhHF)2VjHW)2t
zd1c=@pwUiY_7eBy8gqIqSIfSd=VxA5xWb7d$h1Br$tuqYj{D=7fC=N?pXU%JQh46G
zn4P5nkgT~0r<$km!Ms;JpG!)vsG&Le7Ut%ayB%OQxpwsxu>_rwr4=)B5y)k8Egp^$
zuWz<S^T_xe_C|7$xk%tZd%s=533z#LvBF8KUPVe2z=?qLY2}0Txb|(+`l<IkATt-=
z`mgS1+>=XRZ)ikDoSf<mVt4w%wei=>e!6@C7zDlwm_nXmz8C@9-dvCVZW|P=aGB+q
z$FAgG7sNPkGl+LRxemS$3GJ2vQV(zY0vR;Pf#laM?)yRy9XOiIGKDjX0|tsh+irQw
zp8#$yE_&rG-yTBkbvS(W_<0_w7tklxa2oH0KujJ5<f~5~tDQqq*o~F$Y_|$9(9!>q
z8IfcrDX#j@i@<@)U}t!P?}eq5Fl#*!`a3%%@Q_By;H|B#!Uv5(s?V&>Ag(Shm4S4$
zj|wPVjf{*cjD8($f~5dSjYTein+Y@W9Y-dFjdP+<^x2?9Tv3Yq1B{E817PIAL~+8R
z+0dUq9OOBYPxmlu?Pgpiic~Ms!CK{bQl0|egOWUB1b>Is2In|o1P1>L%=pN1#B(H?
zZ`&7m)Suv~R{lUfyK<Cee_UGa<+L@DL;LPq?ER|4#NjzH0m1Tjmb{WT<><+D*0Z(N
z(A#~-xZ)Gl6yPJ%)FixQcd*M8a_dw>!J&*ocK$keXWfc}70EF_8Mr-JO1vO%R%rs5
zZiY2Vb1oQ|>VB<Kp0jo<ZE1_k#ZOciLz`STX_7;mz?y)w*i00S0t7E&Gr@Se$}TgP
z|2vv2Z0!jub@s3axGH+Zv|8I~?8ERvojjQS9j5bht|#W}8?y0<2?;fTp}%PYGJfEj
z&vsNPN()uVbbrmJl#5oF7-~!`H^;U`EkzE;i&Sa5&<XWGr{+h^0qPCS0xo_P+Syj|
zc>yl=W8^STKn|;NplEx~4>^C7J>=vT3Nd!PmgD#CrSMo^UmwYQ&&g`o@#X05*?dZs
zFb%p73<x$o+*_b{bVt4x21AX6(`&uj?~MqgOiYWlVUTL&<G8W0baDZ<iYHs_39)qB
z@N`G9QyBO0)-%a(8b25bwlQ5fUd#h*9YZO4^aah&DHjO0IHLb>-I^|t`DB4X<9GHk
z>K<qX00t4|inh^ZhhdU%Im`=4($a+LLcfI5grYnoQ&d#c-k#mft?~Vpc56pGx_Jka
z|0P&LX9b{o<VLb}2Jo_yDpIn<G#Nwv{h<MXoB-XTka6fGful4)GOM*wq`W1dvMY#M
zjvenf2g&{R9OBqqUtdP0<AYsn;mA<dtW+8!p;<K8&BR}Q!9M+F&$S0Mws1N*3ZYLg
zK>uwIL}p@cabEiktSjN<9ACVa1D_q?^**R?X}pmT&gMKS-Q+42xp=*LJ3+szPXxne
z1wIjx9UQd5AZZIK<fB3tOSLQL5C2{IuJy(QY$R#>BT9{bA*S@DHsFKl35c;;j(;Bw
zTim4!ZNhNoKdf_Z!r^BJ3v<ittH%~EaUmjJT(5JySB2EN&BP8{PZ<wFx_~_Y+))P*
ztkV`LhV*iHMpc2h!1HG}Swfh{Sh~o;?=bHP2w%D(*vIs?w&)D!-z3upu2{$9zGA@5
zwsy2C9Pu$t29A^kRA%lW0puQ*kNnGmU9G^+CPLa1+5}eY>C>m(pW=FmHAcuRf9+2T
zZ%yR!^R1)-+2I0^yxD*ven;gu0roh;<zcEg?xiw|5p;-g2ViP7P#jUBjLAJlp>LlB
zrmzkUv;3phmjzy@a8Gd@LvYiD+6=U`dBD8Z@u3ogn?m6!g7@+_nWHaaQadhWJGZg;
z_mPXAJlYc^TA8RfwK#)(7}CuqpR&dLz|hu_d&uVr(-{Y(my!2&{#E%}dCS4oaKN3U
zYI&QEUm>OE=jdEIIbf9QDx|4q)Rlnx--7Q9!yf_mE)Y=+S)RGzO9?h;7Dj_nbCzxo
zJUjzc-QCarR+*7J-N*lp4?>XTbCj2}G&xX7fhS3-t(#LwFP1qeE-r`N#ZNW@5%@(%
zkt>AT{15K!1h#ZCyRrXr`{N+95-I|Z-Kp37{3%yk`iGo2q&$E2N~-nqqXrJDdMRxe
zTfIOz{x=tPb+#|;b7r@*^4OQoYF>PWh7xqag_%#mBf^KR_~;==R^<I`jEuknSiZ=p
z7K8hur0Pfdqh_GzRrW}rT3A|+4VwItx-r?AD1HzDB|ai{ChccyO{^f{hCHmX4pU_$
zL%<R0+ngsF?+8lYgb(bAT)a#SCFqX>61KOg&!n6|1zK>M9?TL=Zx#rSEwH{C#8|%u
z)|#(nNS4(6k3~!%2&>T8GKO9Y<usmc;Q2Fs?$ez)00O{skas#3O;)lrukCvmeWJaC
z8(7O=(<+Ep$8<0n{t-X_!l!W{AuoBZVKqyHoBXc%c_Hu0n4a71BoL(>*^Sr)nhT|f
z=h5{cls(g1#}atswvE(Xm)iq@&+PXi2fIOI?A0xLz};zfWMpLA!5@!LE{EA~h4NJR
z%G-bBgE=AyhNzwslREL5Z*-P&u6BxNvy2gBU<fOFx;<OBJ7bmoS`JUxVp#Ot0J<dZ
zdvWxP((}#zukK}XPZz&-AaO+FHlOsf$dRuCmC6=SvoCQ!F1899=2Yp}fsWA4k)Ct~
zJP_VzdvfrJkcxizr}IXgRhc5NZ`UBu@Tr4w$VPeg!93V)>%P@*4_s6rt;~VJMGkle
zH@?3O@)a}Yc+q+Q?Ed_YN**C0GaPy9+B&g^RZ|4gED@&l=s@>>0Qh4R2a?35ki+KP
zazL21-go$xkx4_D3D7|)!$?)q92giN<F<qWG0@{?DGR8Uz{?c&xY!B6zOAqu6co%C
zl^;5gh$Q_p9DygeTO$4V(ihZMzIm^tRu!ar+=fN(T_9Nm5<4zB_fmifz{JDNVLk&k
zFkr@M@dgf&6Ta=Dy89f4SOCCfv9*a#4gg2%(cB&|i(8{jZXQV!8xo+0=ruSfD#diP
zf<-g;AUCY8sX6Xe2N@QIMHae=J7buos-*=iqqTrb9`5Hb2E&c#Q!$sYr9H#0jD!S9
zcdXUJ2bV{{^?pNb^PyropC9-Qeh4hCK&;;^DJfCSlgR}9yukb4G$?<8tBlOCwMsgY
z2n<HM;E1)~(Fl`M;9UbInW0qfun3h5Yq+3g@CZoR<G_`um!X0{2+v0I?t7K;U>{u%
zd>~Ww26oP`pJ9^&czWE{2tdy_)(<8fphw2KX?k6(gn=q4>tP;V)hsmK4obtn!lx?F
zBjV_azkmM@Mk5dv={(kBpC2RRN`S?I0+%J`GYWOO^6C@=mvH}mQA8#I|Dxv_@PFJg
zyhZf`gA!_vzC)7*=<6Eh&uwgN`!)d`P-`<;QeHlEgc+G0F6TzGR4HsP$Ie3(`P1lE
zGw{9r;eBR0w`|x2O9Q$@7vF19Fz#LfW81$a6=VI;Hoxob@xtk;snv@)(1;ND8sJKB
z!n!o*@Q#LFNn!&G5xD1@KLo51O9A(Ox7ctre0tFG-m_O}1H9#p;F@1dz@!U94*X1A
z-*=T}WIk88JUIS!upEFD5X|zo*p4iKafMzEIy!0X(&7RQUQj|dK<cq{nTVx}=}SaM
zEu9?jO;L*ZHhL}eb%*1JtLVLIG3i1%NtOJ?+djjy<5A9MGx4Xu${h`$OYe&?CINXm
z%}Ow2T^(Y*WhRPEfbQk|ZEC&RDczl}IGnDax<oNX$Zl+CZKc3K(+@6Y<wjbw16W&2
zQxlpPPa%Oa?(K>GDlZmHrNHI7<)`bu<qEmNoABO<*dg>2d$3SymFaY-dm}Y2<Q0C2
zm0`2Fd=%#c@Qf1?1%KioL0MVRs0RqnnT2SRS1GPBT84(vv@A;NYUD100XVcdH&`|8
z)F}zaJPX)^CT_WGU?{6|T=KuYLJ%zVA#uQ9AubM%aF)r&=N`wz>cM}RB@>glGu}}H
zG(Dbu)Q`0?U>f2H%Pne{Q5(hz%&`l`6%_8N^Bmx~-**hCfj)ZoCpMdH=-HDA{?CB7
zlMd+-5RMxPQIndu7z@g=3l!YTRH;wcFOG8m&5x@5!t#qh37>5@DSRkXxJKW_q6e_#
zmlZV5s6F4_B>u!>OoSTz!k0UDa$^9-(ci+9r(m`qp=c%{rVdSA4*jX3B-=_BeRBk#
zIIr(RrOrQ*T^_T_ZmT_6OLeUWj8u)k>u{YFyjeDYTC|uEiq3xmGW(PVCU9KM3L1Aa
zAa^o0gjlt9s&KRXQQl4{;rKqG=a;3AA>j@^@bXmx(JX0VV)83y)V?Oy?spK5*q^0$
z1cd#T-jc-9c>J(bzkusc#=)`fcH6MaOIl?X<Krb!KZg5#Q@7*VRAZcUZwr?5kL*(%
zt}=U{YvoheVnj|*#W(_yJGDJe5+k4I@6}q5UknTm4`=h*YURm9U$AP>FDHoYXPY>m
zz4=vH`CY;$vB<_3;3l=D=R;r2eTZJy9<TkhiXxfNSuiHpoy`lU<K3Mx6!wf&@7=>(
z!AB8!URoAwtv%GZWTupU09@B+SEX+o<G_;>fNgj(J@iM#AW6|^tWY^Skpm_-gj({h
z?G6Pi`|o>+mD$+D%-7Vx)irf#dY2R0DJd!H=eed;I>4ipz;GPSCnzY`@c}p6+DdE<
zMO{@jhcV)|$pc%tr?jk$j_Qwc86J4%2Z*qCi{O=lIZ`>Nq_mXQu|^fdHgi9C7$6o0
z!Auq|wU^X`BM=v<i4oR?VAeM>Dz_#jw0N+oiw8?kpWv513F}&!oz?TPplA-dkh1W*
ztyQZBe%WN;y032|)qf>Gj=g(I$A+EMQzEFEY}$usas0Ts@%v9a18wci)62aj1-ARj
zXO37NMXV*tnW63~;8Bou1}!x3Xi$PhYDBewI6~SopY`qzBs5=GaKi78%D{bfr($Th
zFE!&WEK|xeEpL!>l{d=SM`D@pc^RV^W{?(yi@v`8G2aA>T0=X`WtMc2?0j8BHyJ#i
zv{yR^N;ptT!U~($uH8eiDP|iSje}Jf399wd7k<d$1Io%RxS4eT7;dqZAJc`=@&SS$
zz`@31a4+Do7JmNe_YQ7uZa~$Qw5WO~y=a+d2WtL)($Dg8dSa$oN$mFi;o*n=l*?Gn
z3O>WlJuWi?JYjEME+0kEvC=Rw6dG9oje}0E8n1-$J+RnFT7+Z$eq~!>2h4qPM)TU2
zxBg-86wIE9zQCL+wG$w(*RHRw9$OKWUaw6CV;GO)qJie*M;;qZp_!Q(@SKd63wRj9
zPb^!~!ntAj{CZcDDCXj4a7l@ts_Kf0?wIBEzir^St+5=WiTfH|&G!me1zF_cZz*w+
zjJV|394l3cgyXa%coo4@6|SDhPui5xGRuCa6@+w|l#ccF^;I?L(-5G`H4l{RQxjWH
zm$KHIJj@gNPjqGg!6WG07M}1qNg5SABc}UGKv-B`m4M75oCqr)v=S2zdZw<jJ>!-<
zX{c=wD)su|U%|EAg-Jbqx8b%jHd-vswnn))rFh7vO(^d$r~YlMfoE*H@z0m%!ZQp4
zsn8wg^0ML0ah~Xi>UGX(X89^}#5<SxA1*3Hc~I8KplWYjktANKkwNttP#&Twgs1xK
z7zr%~t#Ih_luRiS6PDEC(%<x{*@<2C7>jW0-b*q-6@PQsZV0<uxc|aruT4pdc`=8W
zOug5M`La;aQrhIB<xl-~ygau+)ub(g7c0#yS<RQKhOsF_+lOSiwQM-ZUTdPnPBnF2
zC7u(e)=^P-cmGOEO{u~5;NJFx$uzX6zc^?9|GWSPb7&JSj;cVnV~}8iSN_$U$*q@n
zW}5+Ku{#MgZ0^*1cc$%iZ^j70wOPIar1K?~5V>d#`tWOF=E7zb)%y1rOcEj+rXm~K
z;higE9(t?YlpL$Y*vy(GoG;+&A6B%z&^Pn$CVJFENxliLeJ+%k!HRI9o}J5YH*=?O
zz_w6N**%d_^7yyL6f>iK4ar<)ve0dB8Z<Q!>t&t)c@k_w)N(OUhiy9h?;|BA?10Lp
z<9$OS{*DS;Pi<4cw;3$a`Y=rHs4wF>Y|4M4z7v02CtyoqU?M(JKv7myYg|&~ioHYN
z1opz70}snuoaPKt$Vm!|`RwG8l7r`w@AVVeXpI6I(+2ssF~9W~@pu+?j~)r^v|o%x
z9i_@Lqoy7Z^7gWpT=Ecr#zNg<C>m09^!YF1OagRs8G5UB4u&I4w;Wa{d)8oS&(GwP
zdoCj85NR|w`(MgrO%7*};<ck9cfH1dVky!x6N1RXTmx*Z8mDtiOjkWL>D(^ljJy`g
z3^Pi82l9ZNKRs2rM4l$Ta$A3^@)CO0yvwl|+5ofzlOsZGFVrNJ0h1eL>ak0^dY&Wo
zsku_<jrE#TGkBJHc9{-E(yt{kmiz$VH`5f7lU{tpusS98Z}l9zA<vz4_Nb*p53YvH
z9eExR-rMy%`(uLVrFKZ4O_W%2I*%oMPM5{tz^Pvc$X%#`aw-;siZM=)c-T{!CN)Of
zcrG+WlqwJ0jySo!qh<0MT(zQid0&WubwuKF;AsrW5f76F0$*lN%#vDguUK7?yk8N~
zDEfPlk{faK20fZxoj#`zKn#@i)guB7k?Ju;S=Qj->R%kadv`BhmVJ^J%=v>j??R_?
zX=?bT`lGQC0v$n_VUFO>Zn=;0j1bUHHpS-u!=cWuOU)O5k1HkH$Hzz%zhggjdo9Rh
zDK9zq?dVT717=OE0g(?~A!AN1{YeO_n`t>?KjB|p&u%k%L0A(h#=j(r2qG?8NzTND
zmt=>ba_Hyv{5y?a$HZ>)B7!M963%^#{@44!VF72?{`0R;vMYlUf9GbSv&2rU(aYER
zr3GFZW~E^SS$U>e0V=iyc)%vBPzCAV$30wO$k5OboJlD$kiauM<mgeItH|i$JWeBN
zIJNn|AS_pOVCFd<`ppp)+eyLk;ZZ<N)()fEN8ePis+2LT8bpj5i3W!}Ct#q?cAvGr
z*?@0s91}+v7)OV|h(~$`fpNnXRvSzo1ftx^X%u)Ez(xVpI+zEdLfmNh{*OewJX@2L
zEb1*7p4QnXsF9{QPKe)Aq~Q2%t@;)}*-f0lcwwovJhWOr=iut;g<&{9zZhKBU+}eE
zg>O@sYP63We#q%#Gfnjy>#QB6*`5dv**}goAu{D|<r7Q+QnPW9hr|q!h*sDMyeddr
zro0I2nA1%|AFOGgjJKXZ%-m^_v@LvgSi8P_K^jVyTrRPrPH+VeO>M9e;G;gt5Rnm8
zmcl~)ADX^9p349I|JZvZBqU^SC7Yzk-kXH%J<CY;k&r#gUX`t5MA^s4F|v2rloHB{
z@Vn0Y^ZWYa^Z2~wocq3C_v>|y=k>gvmxau=J6FyWf!ZDQiH=iD6!KwF5#@)2+fNT=
zhD5cgD4Vhsf@de1!iss`LvIYWL8|ioO=96pigzb5z*MlzrTMa8Pf@I)s81U$XU6E7
zk<1ZuJ{X$&JCk5}_~C1HdwXo&8=$tgs71;amIoAG55PG0<w?p>I7a>)^wHFtla+I;
zeJN68-Nd2;ykf^%xh!&lJC|njziYFtn0-)esa1p6>2)@z^Q-m4lPfF}8%WXj(bAVB
z^6XSDu+5bu4&(K;T*RYQgjo2q<O=zjH^F*M;+CFPw#ue|m*zcUzwe`V>jN)=3xveJ
zuC6M~qdz1TrB7p`+JguFjfhnOs`l^B4DGhN8bUkj4s-J6gk-()sLuKHat{43?{g$O
z3%MQHZG-p^2eU2_9jqZuEHgq%r*2vZ^iO3mO(sdt{KW9xXT5QaS345IbRIEY<BNv=
zq9M~l1&48&f}i7^7m=4^99=WBvzghv|Bf~JIGiaG<o@Hd-4E?w*(m*GY%S}35IQ|q
zV)0_<ZeX_vS&Zg=uWmYMH)%`Z9<%zGs(?M3j~M7fkmQ#;O>Hn+k2F~%f(Cq74u)>}
z-P!n_#>YU6e^iYb!eD5wvpNDMo5=a?U+LSo3hWl~kX>1ZobJm;1tvmFEE5oSTvzUc
z7TBb{^xWT;Hre}14!0UX%uc3~pv16PB|w1x)<_HL>QdWuG{+#V|COV0z&J!jP{_0Q
zcO!2j#KQ!@9eM$ic2gejX!T8RC}&*>9QYo#u)VQ;_c`V(cvBLo3{tD-?TnmVF_1QL
z1CtGM=F4J^iRfrv=^UsA)mf%h`4`?B6RU)%m4#%MHBuTnO&Z+HM@nlV-HA^DGX1#I
zNhU?+r&ad(Qt`Ze3I{e>hl(5gDuioH(-N5DJgtI)yOGw)#P|Is3>CG6mlIw)eU|+*
z@X{fdSDGzd=O<E3Q)a7f?DwBPQe<w!1s@exaY^!>%phaTj{B3CzntHgI;%e+xi<U4
zI9uhc*uoHjI9xniUs>ND5K#Hvu8mnDLH|5c&fOH_iT=*g_x*QeODSdD6-qI&l)qE9
zZ}GC>>V$~r&lsZ+3XF(5a@?WT@&w<Mai)@K<QT=ShSJlzd^~wbLntYD2Jta^gY&zt
zq2X8^v>vyU(J95;Wk(6f1elD~L7O{7!T%Oa1eE2o#!8)DO+0`=7RaX6goPt-1^QR5
zD=G}b?Mz?XWjFG$FUCf7x<uU51t`gYRcp;L1H*b<amIMF7SnV)0QQx<*o_ShqcOmU
zjZB<)a8uJI5%s+AoBAYIhgRxMiTK80y@mqLa$}=0F(us!8i9V4W|^&N!{17&q$SU3
z>nw=ICMSEoJPHe%?n+hSCgb^qZd^eH`Uueb$YUN?37oYyG)%sh@O)d*R2|!GV?^?z
z+eW;!6C<5k&hs3?$gHx;%J`49*{Ky5V_kmz*gH)N3rtgbabpHirL~Sv6mc6(e;tH+
zeo+ufrp0VIrPxW`!2XQ&p_4i692@`@UOu0koJ{__?yHt;yKf9nW!i9D9tM!60;>W`
zPfN04xL1cs%jNULHq<+lfdUjDLV?ih-9bQmxTL^pRa8+{)?p>W&)*+ok@tB+$MlI~
zlP2Z8BHazl@#)hR>}U8oAU`3-Xa0J=WHF<|Q|!b@21^$Y>q`Rd;~iGRn}ZO;KW$rR
zLP3nImD<h4$IZ>{ei(l-I}_9hAi8C;OY(~!6&u|6BPGiRf0j8vKi?QnMuoY;mz9>b
z_UG>&wqG2AS_X2y=QjKHu2-(`{Q&F57fay-edY5+p5LJ!B`rEYXm>%eK<AUPX_j-G
zCJL<~PE8u+S#7?ZLXEMHYENmKzq*5uxU_s0$}Q%Wut<U?Z{WKhmaJ(!{{Uq<*VGll
z9hHIIh89X?)in{h1o^QiE~)(~8Bbv^Mj5=nT&NP0OQ0r>1AmUt5$1j0eF0>iM7k*A
znGt#ygcFE6M1n5WTrO2oldP^`;WJc;*^IQPDKR0o$}~;#?zo6hp{cHnUe;F2)ME_*
z-=J^jszRV6E*K?F%&@V0+koqh@6w01zl}H$hixb%5F*^$Pwxm-n_l|H5P*x-kx=0H
z^_cyBZgZe&Q;d?x0D?fT>MI_lSKA<Blup=KR%GC7w?5&hJ_c7|c5%?23^_v=I-_FD
zlkSSWq5J>c(5h*>NG-0=yT#Cx4B`ONX`${{VsHJ2pJc;o{A}h6-5@Q-(TV?ym@7dj
zLm3DUk!aR<i(XMQfOqNc>pe$SPHz(`2CUXa`-^duGtlWnFEZF~vpe5UX4;ek;j3Q#
zA{^j!;vj0ePDK?mxT#cVJSdkt;F{Sgl*Fc!8&AzSsZFT5Q^hi;k?Sk?{OAfpx2(ei
zqF={z;mg70^iP6VJ8$aN?Lu___OPl`9}hQ!Ti8uyU;g>6Umh}q%S`ru3ocU$C&sF8
zJ@?bS*QQ5Lgj&xkY0`W!a98u>3@V`)yeMkTe)ka-$XMo`Wql1uZX2r?4mpa(Jf>$^
z%n?H%XmD|FF{?zp^^Q>4XIx{J5X(Da@nKy|h`h#~@_O7LI`PL33Rvyy#gSpOxVB^t
zxVx=;^-TvuRYmSiZb3#~Y3HRuPoFr<y!<TS(ktVHczBnB{EG;yCgHbMoKULLuf`l2
z7U;MqZ{kn4f8U@V`0I-P>=!1KvVQAtxuvG;$i)EXQc5MOD2ETrXUvZRet%0Q#C&K|
zZVGEenwEd94cz}#HlVeUyFkJg`Fu)%_lLNz$<&ipry}G@dAI`+az2z<%R|<%(IC|?
zq*^5jR3logtLD!pnh7{b6^Z*Mx?S34GyTxnoTV<Aa!MG(j=dfiQ3mbHDhR8)|D@I*
z8Oxm=8?LtbF|k~Jy1nl08{p#Jhl^lc^E_`16Ut@u4j_`aJC&XNFKwd9T!w>xU<`zR
zLaEgV%-4Re>fURt{7vg6XTRUx+56^<2BWMZzdMF1RF{{Ha6lm_8?CJ}pN9^@2YBHn
zj5ls{PZ5?qmHMGh`t|HvDBBlQ0GB$gW!42)*6UYUeHcks8Hg5XaJU&LeERO`pZ&9b
z$LlH=wB&d0?>_%Mn}(p6ak!|2|Cg|726uu`u3kF(uj(GcIAoLQxmTNqk(>0Txr~LT
z%Wr5wj-h0ECX5ZC)1f~|)SzA>r;N^?a$1bap;qKEvur-xz+@t+N1cs^{NL4#eSEyW
zt9^A?C%>he04e_@RD)3Oio_n>uJ)PUqkleoCDoCykOu}0Qt#c?UKc*13LZ^{&BaQw
zGamV&Kjq$)?NovG%dvAM^dan;md9ID+V`(R^Ln}XP>cL~ix`j2F>;BxG(-m87apSd
zygngqmize+SL4>??ejQP<3B!@+C`_Ed<}^BJteTvd&3k#G@AdesqKUip^%VhIKw-o
z+XedsZMbqn`Toh>^O8%xmb*t0A=|qv+E=*pdSsk<wJ?SR?!_IU?Q8yJ%K@8~ws6ey
zmHySDT<|WnH2ZprRjVF4Sl-&j^m0CKJ^jef_(zLTZhqk_B2+rJ>>!tMGeFJMgJwbG
z>iO@Eq=c=XPycF4EO?_QsI&%m)qc<`mkD3^ol3DotAj$c=Qa8av}(Qoyzp^lEIL}%
zCAaBd{%gj(=vK%AOWckYBchR}#t<C##5AeIK)tEZ*>8!>>&rdS_RR;!ds*g{IcFbL
zJudy$NavZ?;Xsg$=G*EQa%+BXPvB87ZCCiGsk4N)Ab7M`lyK?Ehhv?7>xWfA|N0$5
zZsu3+vQ>J|C3k1_8%916Jh|_s`<{9K=pb!jJ7tJLtm2-}tv{)FeYg$J@thL-hP(ny
zb$azVvqs?{6^T$!iSVCX8a!LI`t0>t@MJq+OOQJ=?bKy5`<wTE{_AAnYA1!{$1NwH
zqy8Gl8*FHX{MhXE_>tBjGEu4kYl}c+Un}+ts#I6q$GP3vb#vhC*W*_4H?0qx{Yuv2
zgZ6$6%NOb1ULjYV`THez$XN5ruzvsS-t-6d5@V`E$w8}YIIQ<~t;P;s!%>OJ+bGGE
zeB?W|E_HRAgpTFs-dvl7PvO8qVS6++AJ2|)DnIde3LhR?J;Rj=hNCP;u~(tKO1hm(
zo)97&J#lk$)6}Fm6X?JG*Oir))ajM|zVUxS3ixmpj>1dDWo56Oj7ZY$q-bJHurW*4
z!mEpyNu#oP09$e?643Ir!7RtAnV1?H+L=CLe}5ap9_gwku4-3F!yMZU0Z}}JGd$0~
zOH0Jwl{OHkC1s0rc6NT!DjBQ7h>eZJd0bK!9mO{P{H^#3AE;^7S_L=@0K>Z=I21<>
zRf-oU_O57VK67JZ2ggNm5@29KNT6)0nmmwWX_3tzdr{xq{KuiWwEU2|kuGfn%7~=F
znBMqlG>FVB>zKy1{tyRoiWOU1`uDaM_30nG#Kgx}MB3ZJPd3d%P~170vpC4us1|fg
zEud{X%xGc?xoL7JF~~Ew@#keNRWz@wb0$SH$KNON8*Te1r>oWwhCO{IbG^I`da+fL
z?-ZEW{T6XA5%+fcI?rn7l`gG2xqUcC26TSPC`P?!{CR^}ZmSs7K%wpKp*W|@ALARm
z1r@+jFRm!CeW&}rv`2iC=lOE;Uxrm_-kC<KMauDjtchC{POsvn58Oa2XaiHIPu8z&
z8kdGzt)}GM)OW0|2)^m=*e6}llNqQ!8DlN}_~${3C~8PqR$OhswY1FHj+A17!#%U8
zSN+9NAg!2UtFoqdTIP83%MWiflmvB6UrzqixKmb(LJfGdqlwTe9i@AYVgPtTShw^C
zA}GKbk7~=xR%>zowM;yKNLIPXDC8zJIpOKkTLBDIEtNDs4^Mz1IFSY22|jk5s@`t}
zga8V>mZb7OX&;m!C>n*vqfi1w5gm<mJ#ieW5cT!(L&)i42NWN2OplFdIsR0+D=W1p
z&r}mik~A$R=gRw&_L0|JsSrzZ72$@3e%`3<vpp^U=~X50YaEJuEApW^BVDNzVuSzv
zv*rx%bDI=8dphO@Q&+k@svnRVg7(lm@p7kEn#u$bx0&qT^(wa5ii$&UT>y*?ZY0F6
zV>-;QUUOCmpnj?T>`(+HAd-Y8_a7fU+!=k)O$BQU>$wtA=W*g+7S4b6yP-ZDZR8r~
zF6ENgv&MV12C`BZB^7t+kxUB_G$y8&Yo(uL{^(v<`w;E^U9X{B<+AtZmP=wOxL>&D
zeVo=?Po>(<R*tH28=qrX4vnM4Y79fmJp-cNym<Ty#3-^=d%};I@<fkDZ}^@Ie+yVz
zlS|Q}K@?MA{Slxfc)4!!F+l^oktaFE5Ky2aH0QD!F!~TO$@JFZiRBgta-k(ZM-}_V
z<61;9N%M#dRIyHYWZ+s57=(6(HlT4n*+Pax$irIp?)J6!VHh1M?@T;TLw*-MEmLe9
zM{$7(@k|}FUyM+=ATm%9n%iwy+`+@a>1<Hu;2p78UGbKbwd`;4?OIK8ZHD#KZzG`w
zIAtkxH4)FKzSbG$LXtcqm3t*0CvGQyh8kDQoHYleNp}d9Zy^><=i5dHDX!2m*c(Hb
zM}vlVu_9~}zY!m<OQ4MRowOSjn!FZzlA{uO10q+PeHQB29Z!qv3V&kAR%7n%L|Pw=
zuZ1F97RBh8n#Z2tAJ8c_ICVf@1U6m_bugFIw)#km(7}P&xQ2ibpE*=Ri*Ei_!7t=(
zS%!6A^eK+EwxG06z;94mVzbJ;AI8+<h(Hi`oUW+<cfF;v9>OA=e9QR<@w~VgTp=9t
zKfIxpYUJ8Cy$|_+{9cvnIDJX(*<O1E1z4e&cY6f!_mSqY_&{f(BnOIO9*C=PSSdWu
zk7ydyarSY`Q8@_YWfkkiZwv^$D6-;<R+(e*sXOJe6Y+s+rU*nb5%~@N`Kz+KufmFz
z?x?3b*@TL<pGG)gZ$%r&l@8%g50R5LZ<aSa-5okxiq)EOl?a!s1<IyNyc=QZC|=_n
z>?PYH{EJWMCzis)G{_I<R3uUbKr1TBE9BZ66*Zn8Ia$;=p;sU*)?P&Fdd{w#dEp_7
z@!<+=6d$(XZuA&GStAgcxX71uHTN-Y-bzcnc47FT1uRwwY=(g^ZPA`n+!YvPl9n@6
zY*C5`Rac4mB~ub5PFqco-5di8(axytxA3c1KLB^GL=u7VCWOxw6H;7>^52}jw<O1l
z(7}gwY_zVxZ)Elys|<C;LGXoI+ee)geavNyf{4w6jC>yfk^B9yDOV1BS)BH)c9c9f
zJoFp5-go>s$j$v>iQmZ5%R$Xn0YcZI?zKK2L5R&*?|Y#g5!kOUP|>^aO?iy&FvFE`
z-s}5Qh<nH(TX1CfwcUC@SC-y?<F0u-8C+LM&39qAwwSsm;W9#2F*G6b!v|?YLy8Ab
z5^^#VpjTgvdR};!{eo%w75S^AQBW7(-S{t|zd9y1xCx3nWo6F$#&{!+2I*kljVAdz
zPy&#XmP<&efWR1Zo-^~sQloSyx%kN$z#?fH8kXY&HJi!K&J?O@y0qNh7l9T6EzyG3
zc3zb=YT#`DoSYtDUS%nR&jMc-O^r>wPu2<31giXxlMQLRv6q-DR>OK&M$ak%7Sc$?
z0_c$|Z<96#zjlgl3|*Fo0-l>5U2JzY)sz2paUj?I7IfnV2|rV8cLj~>+dqUgD9W|v
zuL1#fs(sRzD{b&>c})46_W5F&9Z)zHIRn^)89Cbz0>2IpS#4dVadX*NsWu>5L!peD
z$^}bkho9JU@b2pJOrQGtk~sP{t5JYExWak>7zD&W*|;Fia=->4ZFk(a|92-Ee7?80
z_y3ke98~gNR}}IzhE}jNK#6g}N+c6f?7LMQs^C}Ph6i~iBqZE#7}0Vv#&cD;3eeZ8
z)dcr}8Uiu?as%a!p|t(KPlyY!3A%@S;{VWs|DWtbb?o))p6x?@WBSL#$I{h+sx+a1
zZ7y5kWCkb;e9g^CY>vOgWh4xJVFjp9sW%7<pn^w&U2%srt2;Rop_N5GcHsKpW-HX`
zGhP_RiYKO}l^D~r0ilG(h~D&Z)&VR$)?JVwb;J`pi|_hgLWFkhj0ZOcwu|J2Gd;)m
zloDZFREVIgd6<(~*L50e)r{SPRw`wO!KI_YK^p@UY@xlKm$CVC{yn^AppMY8uVF28
zMX0&};sf>@oegk{vu9KY;oOVBsb!VBfO1CG-bJXW!&(2beK!yjK<ABQpC)IO+cA5v
zbj5BTYx@v{>BKk)s{|(H|1x0DD<a3%ni1a96yO-Z)*~8<eG}4Y2x$fgt+dq)9?F4n
zT{<gMcqil2hAIqoRb^|BSw-y80TP<+uFyghI*vOlVAT$7YHQ`F7DE3`l0+Ky;ex+4
zgvUI>iDyVDPQG_<7|qMFMhz}gw9DJGPK3ezv*zGr;UE;s!0PQ#cM^4m-qGU}Nc!OK
zTw;zf<c5b-X(~e?c-6|XD70~|6kTiw%;MY7QR3)J@-J`chMyoF`FD#1IT$u1=8m5l
zJH^nD8VN?J#5iXY2Gi;YfA`2Npuzio8vupjdBpSfGFJr-_qU|7yQ!#uUTvp@h}`0?
zd?W2+CVLAON<kd)#JnfJRX5EjH^eO$uVfTPzTW%xv7lP*i}bt>RMQ|2F)Ow>9$OGu
z`R(aWPOH=ejE{Z{#)RuR8?S`;YTrg)$k2mtAi@KcQ}lyhSe66XDbNQ0k|_}bZp7w|
z^V%wcum5^Tfh^}M)e!W&4lv4IJ3TMlq1{B5wO@pDUsiV1pL5-gG^7AfKp^(0oXJsK
z@s0;n{P5S8zJ9+K9I8MV!fKjm7i0q53f0ryjE^AAa3(0>_XUaX52T~>o#<N}f0Oa&
z(*SNP9=1iGnXf{VEH|555*A8T0xW+`*i7XebZ|Cp_XH0ET?iDGLUvPQsii#^+kt~2
z`QHsMIf`LSzX4N8sA)AB7&6}J=4NIXK(+zV1O|X4+5PwKK<eBA<q64t&QR+Y0JXf1
zgy(#BPZw!w;73D)ETHk%u+nC907=CKAb)_`+Ic7TQ@_tF)P7Rj3E$j-MM}6LV`OgL
z|3Hk8+~-NMhpWhxO9T%$4Y|(;Ql)t^W2+w^@fRI(v_zCkLoieigz?9@uh!96z;$%`
z=&%^z_NPD%4p8fguPiJKf1^V#h@KxFp-L;kqoBit(6W=}eU9!@$~p=fP+VR9(0j{W
z`1y9*xn-(j*r50+h0h&hyN5@Qg_y)$vF(>{o6BxvEIm623?jveb+tJB2S}Pb0{lqv
zaSbB5kUR-5eYhTW3283J-Rr3j99<CMCOpqI(52;5WHO5Z?IjH*$B4jmsMNB>$A`RA
zORjsgWu_}f=FkL|rA-Id0go5+u*dW3Z4OeiKLjyXipW@Me=OcQHlkHhk)x);^aPC_
z_0ugk85tV~*$T3W4-}^E4>8}MSrYY_YY_B{w%{dVSU)1(&5RmvgSQjVsy5DML0kZD
z0Qv#`WT^AgkLo;J<b}J;+JDvCj3xa1nBqRqr8T32E1Bmv9`T3ysX~W@x}KxW&H~pn
z{J(}k2n5=XYHQaip`Jl+^hPEK4J7^3`FuK8(nToy_~&%eKXpbk?St)wyyjv2o%}bH
zNGbV$gUX<S@xhN@JEqu_RO-6wr!rR#_tQ%u1uBWX&_`sit9}~r{2EIwHFPB>dpPQR
zX|Ei(c5%)f0q+=I+VuT{&H)!oOBNc?+Ld>uTS;QOB|h+P%|?<a`)<Gh?&DQ5Cz)q@
zl-$(HxCt-WC}xEWb#y?BjXxD5q=l=+v4Q6<{rFd0*IG^D5U1az5YC$Wq$DWA>b5g!
z?E>13OnuHB^uoi}jeJ>c!?r4L)L22k`RP-6H9PcAX1(*|h0eTo`HQ+my1GSA*>0+(
z_6}TTBg%TV)hcI`Dat}3`5g!C+9ea8?OT-XMwiaN$!vCZh+Lwwoc^7Zk`i%?SSqp~
zO5^$xsat3DDV=NVA_MlGLF%ofTn+r84u-7I3wy2C8;)$&KiqVDYCqBs)J@qz>7dw=
z@kTN{(a*#t+CmZc7Qb;76bQ3It;NHcn%-XA_80da&G8y8ZjJvxE`a;04fh6X>L8`=
z8w1jp%ojfLnXCAPZ|Th=(60FJ-^20SrJXIhNA<5mhULcZ<nl25N|wdAx8JI$(b=2j
zsEgCQLOYCI9I6HSnj&BgnF3=7+*h-OZ*iyi$X<}?jgz&{lv@+^QGWJkvjcbr<hA#Z
z7T9eqLCYqwO6;^MpQrsPSB33qT05DXsAVIgsX+qQ<U{)KL}!h)pI^IvrBEwk7}r()
zeh!MIG|scQQyuwYYav8{qvu*dRN|kfO5KR4d1(ZT4`~w37ss-o0g7d8n`n4LJhQa#
zH=ua|PocS{s6%0y&Wy}0FnBXgdouXstG@PtU>w`lx2YSs|F%)p4%%5~YfGs=BScmc
z@pGS<sf4?6Y;b-9{^$>o%TqIjO3nU3VdQg^Vcif$E2bX@nn*$$nbsi+iPRCG2cx>o
zYfKERrVtg?1XJFBOI`U33;S72+?FG&y;l=3ocu;0DB}CW?eKn1{gc0E<NL3i0tR^B
zaqB(n#Wn(kfO6n753ZBbGX>URl%4vSc!W*<#2T52EFFgaW+v8n#y7q<)s-2KBHn_c
zm$sLe>bwkjcJPk~hv)>?`%N~6hKK!Rbw&4zAE9qt7@ckSQ!P_fsC9fNbor~(Tb}1-
zWidHiMNY2@d9<F!<|Wy^lywzOl}uW9sG+C42qg{6MEDmh1oFv0e$m2I<4zhN0$xyA
z`Ft$O1CJLR?)9F=xazeNMHwAs8UPFTUNFCu4-Vv@5xo^LmQsTf_ck<yUze3}RaX=+
zPX9~1OR$`-cqeUQDj>)%#v-`j+XGJ3WFI?<m#%;Ac=J%-x7$*W&Q#{N%js8olb)Kw
zXKk31O?Gq7{s04D$0C)1T0;S*XUi`JrKMcnMfz>bb<8;N0p|l0a-m1F=JUplxIfUj
z?(D5Ov&%<NFk8bMlW&D!;~>B@ldNDZmJkNF<-2j<sVmh)Pa<qG{J^N%>D3?~p~crE
zm12<2`}JVusRtKL3}|(t(NLF(Mjzd=vnvT8tM{=J1;=>o7#ELf6o=4?BaI>&J(Q-z
zo$=VLMV3BR%D`x>ZgbQ5H%kx{S4JGg{w7zJmqV@UIQph$t(qqXO~*?MQ&T7R#70!o
z==<I#RB`1?vZ#NPO(<lB<170#E0ptlz+!%#02*-z3PWx>ee7zOSLzV7ubwnwlHI#v
zEDR07{Xxv<^2+e0^5h^!frmEPZYCcJbus34hA1FQnN%6}E$>NQm}Rx=J3MA|)JneC
zw7$?o_bMmHZ8|fh#(Hw|th(h-T<r2D<57;=UW4y<r4Hot=bSyy-f<bxk|~QRU(_6#
zE*xRlkLd|#bQ$MRz7zMn#cG>^w=h4PJ>NGSmHNf*<Y)6$_w2{Dbvc*k37XES&U&mJ
zhO*~|nl>8dR|UovhfdK?f^YUe9DR0s@)d3pp(Ns6X3fE5Ky_uM+m`X$3Fmm_JI6Mu
zAPy*g^Jdhws}y+piJv#ncfk5U35=eee*g6AzyLW~1Te3lAQ1P52EeVKKllDBLeZ|t
zLaX&|rO}F{<pR;%^mIF!eBOhTxx4I1F)=anm}FtLHSr2pe}8{P#qgLI)f~+8yX??t
zn{XGTp0{WtSiJDAlJ`gC<lNutqK)mg5SkhvH@MowfP}FAd#S)bt#DRn{@4|NU?t{a
z%s@Hw4a18SzO=Gds)mLJN^)`yO-(4Y`SF3GAtW1nNwL<Mm6ernU3teGDfk?0lC2ZJ
z!lDZ0B3&NZ{a}%q_~A1x*ioJKwfhF?-8OI${~N-^T6Q-XHDPO=VSf9|{_Y*-n>y?O
zA?ck$c{@{q^xS^d{m-U=bhF5XM?^GPxVshL-zJVyluN6xs~guf;!2q~><|(X+Ue}+
zd3|sKyF!F~Oy5*ZiY&)$Ynvks1I=#1fdfWHM$)`oX=*=Cte}+WubYQLgd*o&9w(}m
z47Z9MtIwi27>pPXKhHSO;9hDSuy6YCs*YmWfn$v~O-fbxO2jOqHLl^s4*OacqLc;k
z9<z&op{*bqiDZ1*S@Uk_UAxFYJ1<r+70aBid(k1aGK3MYK^0qz*H{P(S3@H&<JnuI
zhdDa9<hh-2FPb-V62<p6gfkB^rC)|POdV^U)9xM14n(_s4LUaYRgxo9or%=mc8lD)
z{)!aN-jXMIHm5AhXP*8#1VVtme+M~Zvev3feuHyha8OWCkV5#9jg@d0ECKXDX|iX~
zD2oor%gYZ(cu;gtcXV`&jg7&--K?<i_VH0yRUNspvy*N$T-*KgL)^)S*hz{xA#0xx
z<0pU86;)JJtgQ>eHikjocMpi0(2gnk+*<tKq>_lQhksMg^+dw-SZju8wCF#!U@k5$
zXrKbdbzaySqg{AieZ3#Bcoh^B$SP}7cXoICK7HEx{o9#B=n{gJhv(DGOq0?<=7J$~
zvYdg6HG>IHG#aQS@nKQP46XkHf`XVhuTKV&|A7vKP*HDj4vwzQ&d!~|L+E-S-`<dt
zmhN1cqaa6GBNM#6nk>-9c_3&3{BzL|s`wV7xCtYLMC&Rhw1aAf(ET{}sb_1|hP?gE
zYVrNRU!fN?+cRo=BTs%S@f?&$iHL}9q^eLArM@*<et?SrIbI+}&w2k4aQhmJ*3gmd
zDkNFcqnS03c~PJG%{|46*3QX)0SWKHKQOR?#tz-t3A}L~(Wm8#S<mTp3*&}9dSp6<
zN+%gO@g@~cj!h^q5E{gpM-HctDkRm>qctYjf`cB6Rz%_&^qIeyPp2VsYQi+64rS{a
z=^-A@l^J!z=f9_M6it&q>~m`xd>l*de7C7G+9<77&}_zN?ZN2!7l{Ee$L|HM2EP`{
z;A^X$4W(QJ;`9n%75C0B7k(rpyu5CIFGcL5*!8pkU6JQHfpr$>dA(J3y9Gm?q1HdF
zXbTDCb`xVP{rx^&6LM<{v$RU*rtZe>C?Z(22qVK^NvQ3zd83;-^@eM%<*)utc(>qi
z!Ka8Sox^?gySH@ze3@6kw3gTL33r*~v(&QXJByncHhIWl3h4(C`IjGU7Z1Nqjz#9}
zEy1k{I?AW&IvTFED&2dm8+*_Y)sUL)W?1lZvJbmZ!=ra*-u`trQjF)hw5v5**g{=a
zHzXZP?(;=Vwdu_<v6+w(-FS?vyJ9@&dqdvXzyVm+Yu7kFiA;s*23GYH8rC}szPBsD
z!OT216w*7ml1_FT*+gk(s0%lU_P(&qA^1VEK6Qxy`T#Q2x&6;puqxU+snYpO(R^aj
zqFk?`<P-K@=(nPVvCF!5y?)=oyRLEVO8T7l?OLK&SvuD+H%Z_2&i_<c+rFPRSD2>M
zu<7<t;GAa7)#swKFLuFR&yvR?E$S?u(+HL*-5?c|b}*0BdDy4)MqRh0n-}{i)c6H@
zrJ2oPNI&Q?9vK;d*4~Hgmvr_>M9B4dXvsN8^&5(di-B)lV4;TTq<4c(I>IV%2$ev-
z=jP3{KJ~YJbA@hcx~V0*H1x4027D!YrC`uwtY5#WSn6#MC0AEh2PPT$14Tnq+sv#h
z3`ja!+Q!C45djLZ#z$L>cg&UO$kC88TxDQ*wj#NeF-%4UYFW?@Cs-FXiU?!_F?0IZ
zZo+h?D_2fI*^oi+?K?-qT6zQml)LZOSmB^EbSVDW+S#qGu8PpA(LL-|#Ps*4q@=uT
zi{9K$vUaz$Vn~1W3Yf|rT^C+El!KDMV3)!uB7F)#F60H=vi2&%I|X8jGa>;Ld7s+O
zj`{Bt2=cfKXY^BSAHSX0goZJZ01^t1&_Vm&o*vAaH`1H;G`HJ^?$RabrAjQ&VQg>)
zS$Uuf*|2A@lp9sr{PvHHZNCN^85|S>f%IZxT(c}15xGFxf3>CFnvhb@t7#rOLN0$*
zD4TG$@<fx1dj7EL(m8-!C|%Ip-l~9;lM_g4H;Q!-p;1u;BK}{j-<co~>(_J)Iz)Z~
zISYEe3F%1@JbZlDi*<Ogs?6l~g=Vpbpt$n-6LnH@%f5KWOeOZDrL>zy36~jUVPFyT
z;)>VYRk5)t)Zn^Nqr==wk%u+#?JCHb9bVT88&}!_2IMK2oSw4V*bV%S{6_IUgS{0w
zUf;aPSS$|i!&O(Xnb34d&P&^@JIN#Ea<o1NE}x-K1XN-^Ha1QtdS?|gj4>ox3#E!f
zmzK!$`Ns<)W6ypiGjuC6mfRpBuM$Y;#p}_bN221BYmtY4`1twrWCLuh3Uo{}hmf?;
zZ(K6sNY4BxQhbyLdAZjs=B|xOld@AjtMYeIJOq^Q$OSI_{>5a^$WY!pu;i)0|3j3i
zRViX@(Cp3%IBOh)T1j%By5r{;;xzzDt*))%AYcHGyGp^ZGUL<zE6pOFqT<gl*`&Hl
z$LKPWhLZc_h`~19{kgM*zTGc2AchG}iA0rR>}qGgEh&@q5)cyNBF3Hl)XD!1llkNJ
zydXBrKYVCadz;Z7Z=<5Et!*hvF>d0axDoM{%j-yyoeI~B7cZ_WmfBRX)K*psiXi(v
zQcY)+OIL@1S3=%QO=)w)a-*x=CoP@<Lr5sx)5G}mdb1AP$BizN0@wAl>^^T^ylvyf
zZ{LXHtTPMJY#QlPzOD)*yJH@;VcK(49>$BJ$+$41#QzwkM3^Ko6C4lS1nquj=n~Sa
z!ePgI=*J!}^Vg*#oB#qV!W1$z9#p@`-`^<4&l*HNOyt(iEk320J*Y1(j%?6jOdNa<
zKQOLHKo{Hk@@VT-SQIV-1YteyOtNNvGOn_K845Yr-!ZuVR-?N%%a^7G)iO7gk=L(Z
zBjQE%33|Oy!Q<TTzs3J!D(TIB^(qk3EC4XksU+9!el4`M*$*<}sYJGmt{)v8fm#TI
z>$(Vp*C>lJz+Q4KGCz4vF{(gw!jOlWnwqcLb@X8Er}dAZz#uPgc}@g67Jrh3>0?3d
zqI-u^>{)z#NsgviXpBIb8pMSZ0<?}>5w@AZlb-_ENTUje1WzG4OR26x`rRuvc(ncX
z6a>hfXEP)aVgn%G2RjrG!QdaXp6;9p4<=>PMT`j5EA{pD%@(m%1nvL7Q=mnB=Ybvv
z%?NSn1f?C^XS{N)ZC`R-otU?sxtbbjQ+2&s$p`;k8?ZFcX8H)aO#!S28&c+N9SmKn
z7cN|Y^O3@>EhH>VgeWd4fw+=4O|8iyH$5HllTY_JX!>Sq9g2#IAoRI%;lgc0_M_^4
z+>Lr?ICSU4QJt5BEuY;j1UdLn39c<*6oWDxj8*6bUgl)gef2v+6|0C)=$(e7s;jp*
zWN+ji%pf~7wXm=_KRbmWyKf;nI=Z*J8#?t|;5YEZbM-D6ZEd$E1}0wMz)@Gedsn4o
zl<f|({pbMXC9!H3$GL_)wD8hE&j~9x#JhQXz9g9jgcwgMs)W82Zal&Fpm&}G?P$=f
z>jPv;tz22&jQ*UQ978@j0W$+mTtPb=L_tBpY=aRib^E(%&`be+l3T&Yv|ksXg%VJW
zmgZ&}8W?DZe|!89>z)BU!t;2~>GwJaFF<iI#uhqOe@s?J$Q}Jpg7)cd_azu)G3hAA
zp2Wk-8k%dKe?QLd`b$tbRlWC}v%Wv1C^6;<bYXq_=7@5wFn|ARqzKmGoeB9vMaKAb
zkhEU?%>??xhDL4$kT@bj;ntJQb35L~DHr{PaUifOt4|F;?!KZEjOfzRl2~A#ZGtH)
zH3*nTG>2>qU1?N<ZfG~?*(7K#9dT@KX~`PSH8eJ!I=Ho%1I~f0K@tD<Hv!LMAi2gq
zf9}||RbZX2!U3JANoplHI5?`Rs*qRl3J8cPWR#XL+Q;1ez4v_#PTms$Ch6$t;QvK(
zUKbK_pC~gt2NE{4ZG+wwC}K}%W-^osPf^-|B10q+3BT9{O~x<a2YauEWCLkwweKQx
zFN`+u`!z3h)DM}=&dv_JHd;Em9NB>1|Ni~~h3tI3j0z9f?ldv0&`SUnbtfk$Xv&d>
z=??}5<1np*ftFSbgav*l!CV(uxzCxRE6`91Hw~{6AuA`>=aPf`2|yqO0ds&}fUmzK
zWN`$~dJg(DFirv&!7LZ(3yR**swx$@_YE|(of`W**-31{1u7QQR8$B^NF2h9lgE!A
z=fMVsGy^bKtuw6dw63l$b{0p#Z(BXRQ8)=O5XWf%gS;g90vgL6G<x0^pnuR3NXOU@
z!8GVabkO^NT;y!kM1O{GM0mLEbyHtoF;`FC+fP^FH3OFXDl?NONo9F?876dO!M2`n
z4L0XXgH*Jqrw6Rq_UF7?51RdV0qukhq*pqC!Mp;r+%RUb4QX05;wU4EM@TN|w`l<P
z^+qd)l0}w*iHQvHX<%R;25i8-NmVPk%YGT;RW&pQKwRKE#oGu8^Z<w;gswF}AYN9s
z33PObQs$s%_x-}tlR=3E?1TmgxC;wyc9*-s1^WN}T>=2OfdMT=u&V9#8Bi#<AIRJV
zoF6R%g94!Q3Et68we%UtLyUps2RVYZy!I?H29jqgNNcdl9+@Ixtj{1^g8nc!I4rPj
z+W!3}L$J!o1Rd=x>D=}7^t7^COy$-8Mj@K02u)$C)Lno7Z2bBC`?Xe1;P=sz#l=NX
zfdIMjpIEi(8xAiNSXH#O7v4RLwh(Fq=~=kB+tjVCtuT#3n1=^;kl29DJBC)Guka_m
zuo5tW0%+U{FEnmn9q;c~uQY;4LLP?WfPq0En0xEh?TZ{Q-EzFZNkj588zytWxe~N!
z^n@`fO-)S@Phz{YCo8iKR-kWoZ+BN!UA?HRB08F=c^bYP#93%bl`JggfdGDuo!wV9
z_dU*r8=1d4u!X?7PESw6r$Qx0=n(CIPG6*<f$|&-^C8bUg%;_N^yKX9<|ltaf6JRb
zmY~I{F&2!~HMp^IqjFVs2>SR^`otJDK|-~@{-#`p+DBqVKTqZbkfJs-n}r<+!OScO
z$iyqBq@~@29ep+A-$-XC?(rMR7Yi*xm$cp#K97&52&gnFPgCPe=*@zGB5)X+L4UJA
zrxBWBn0vvRr>U)hiH0*ngcx=dgQG+f6Zv`LO-TvFqXXGeDajTn7^U{<^XE&k-I6at
z{;p-z91>EcgvE$}6cK>Xj0n&5fy*8q+kni6h9Vf{r1B>U(^69_EL+TqM%uHb!2*LX
zD>L8&BaerRTUb;CZyE0Xb7*KN^fiN{*oHn-KxSTV@uo<9gxM}oCWZ+Dbcn=^j6=}m
z#X&rR#&Vc5FulRL@w(%6F*Gs&5&`y6tRZ>W>5To_H6`|uXBV!)s|W}P09OGZ2JC*b
z${w%g;$qwB8tc_;$(I`G+1Uf0&$A8q49h1#w+2x15<T#6*Kw5Cb=}>o%qoTW`StnI
zz(-X<1E>rJ_I6!{It~Kn-grZ&eqv(c-kz^x1nlfO7!^Yw>jxVcbb;#J<|ZNIhVxWj
z-HMB_XbHS2BXa^~NlZFvHbx+$zu;Dj_ZkMa0~jr+0X<}(XJTTydUc`JK^uV(bK>RY
z#lH<0vEv#hjV7k}!-rR2>sm)gv7XN2<D<cu87ZI`iHRAM8mzCbwyR`W2qnbDJvlx0
zg9n8InHRIRsw&%<Ehu_jeX@hYX;)wk=|)W+A2o3+45I<l3*zfodo+G+j5^)KUuW<+
z9lwT#2G(rbeoBP488xIS)6s9}&!3r|u7I@#)iqci5Y)=(vxO7@&K4Z>_4RdtD6rGA
zU>pQ4qTS1))L_))%hytU?%&dKa`#IOKm$UjK!+zy4R2uDu>)yZYEWfX|GZ*zcX#*c
z)2A?LX$2Y^U?a|Wj=goY<(z2?xh1g77>{Hwf#`!Y$_Bn2u7*ZZ;2~f&c&=ZkAuW6R
zcIV&#IVEQApUj=xDbMp_Wmp2zjMv~#lm5V?!4xEB`6rpZD@iBJsvI;yK|zu`cT|Nc
zToWO?gFOlT_E%;7vHj~i>Q<y^6c07=i>tS^hB^U{61l7TFob!F<%a*tZsI;s!-$R=
zbeGNtCbAMY`fY(@1Ls+)hHoY~yVei0EWt4Ad;}$drp88i#dHWD{XvVW7~>$!_yJsc
z{>zt#4<A}TyqC^dCM5$t%D<gcgD1z-3hbnMx<J?zObpmrhOh_`ij8B7ixG&rqGN-!
zDA*|X_|ldoElo^_5hm|!lyr43oFT{FXuD)tJL=aHb?-+D=z>v#!z=5(N*uSlDEai~
zwCE9cfsX&lfg2*!LdcjeP5NHWXeQO$oNG`>O;q6kz5#9+C?8ls9tJ2P{%!aW3_GBV
z_JObuRs$A#e|zDUs3=rMH(_Q0lst#}`ao@ipR;K;Mx#Lr)D%LSUXM76kVk0V3k;M-
zSV4Ua9;KM<Iyhq>L6jXQW1v#Vz+Y8bs>_!KkrEyPtY=5jC0bGffl)*#STzuY0LkK(
zCx2<a7KqYNxlI%*#fFE6BM{eLy295X5Zc<GkB<B$3$BZb>atQp!gKp0F+$S!TcTN|
zmbP|A|5}Z~*e_8usR>NFTN$YFLOun#APK7!_B6H6;ETY^r>4?Qmm&}_v<(`3)iBNy
zI@k>OHZ{~I#ctey(@dBdF&R#A<H+ONJ}7jFu#(*8_$?oANoh)3x1h~(12-ODV3LG{
z<m=b3v%$PkinkBYZ#!+?`OC7<;x9bV*T>VmAoh9ojzEo$GYkwN{%Dgr{HqDG_8Z*x
z>Z)zb-O=_TG|bP5<{sdoC7rE}#JWcqXp@w6S_P62Wo2sIJ<nE-AX;?JqXc_Ga&PLb
zA`-<mi}MwX0ld<77n4JL+}qaHY(K}13=Bjro0xs@{SH<F0M>lToKNQf`t=n^q8!M5
zvIoMKmu+VIOITqmIoBV9CW|M56v7IY-qZRJ_Nd<@R9s9<K?|!SJA1fn0OM;G-+MqW
zfm{8Q5SF3{NlM*@hJwn<Z7Ru0#AyO(fkE9q>nG!~q!1V>obu9A5=u%gvUZ3a9ECH&
z6Mc%U2bh~h2X+{b<@8GTVCd26H7_qOF`HZU6j^uKJ4DEblq5kNn>^QP?gYYSv<j*r
z`ktYwv9U2CRE1;Iv=t{yV49Yep1#Vo)*f_TY}wl{GHoxk38(Z~ktz3y$Vf_>n3%M)
z(=ssuxD5spPIP<LLE^yeE&B%?G;W~G{qve?i5?tlH4KO-Wi@kU@xIhJ_`RlXoB5;U
z(I}|SEJE^|pKfiB5<v=>?|UtE5|E><5=n7a{Pw(|W#?sPrsGvHIJ>XUn;z~Ny>LFv
zWUJ|RUHr!F{gf!BGEh!V4q*OIG!HiDd+~3RGl;Yg4QV+XWgjeM)+l-U`bORtgUSIV
z16p1zlKeyGTVX5N+p`cAbah3>@qc4B)!&e1%}H50++NV$$Cu7oPY__4O-Fp-@|0D{
z0<|yQ<dQ*f>42{vyi3s8t*H?iy~d_K1W~cG!y4?0w?lJfpL;a&<gYY~jv^5CW}55a
zS_;Xzv8He0*%n~tOM$-kxO4p@<Fq~D(hnBU><Nw!wg7QBYcQ`#RQ-yHCu7(F0mY}=
zb~(jkyGsB=c{n)Sd(Azt2lhKFGqZi^ks1;!k_dsGdFvA?1S>bU17s|&;xhdF>O8ca
zdUvJk>#q$!5u}D43KNcp3VQ{Q(8O%Sod{%8iPl-!{n$^ZqQ2xeES}%-c&=0Q1qI%@
zxR_&08J{rqTL2yd($vh%c|o+@DZjp-zb+g^X`Jf*iO%0?Yb~s%AouG$9$*H4wB-5e
zfS?zq8UMRz>B6KAIhvm^KYxUM++WLs+4!WPp9|9Qi9>^MrpgAxat9xzP-Q~rF;<3U
z_OaC%c<DZ8_4I7%G@96M1NAeIarrb|D(*QRx>w`H>@gcW>}w-JJ^Dq~YdNw3O@sjJ
zK8t**L9J#V@p*%YkZ`g}-nQ^Iw}uQwU#aDuq9!#FQMB~-56P=~rL@Yw{m#mCc&z%~
zNcU9y31aSrAYN%mit|hOEa&B+JOlz%YBu9?bm_|e;G|kAML-75<XF7CvI6nB_F0B<
z3ghL=rDLdZqPTkJIc|t3Xi5E?i*jpB1g;M_ifq=tfUL<Q=zDHcQ$SH$*q;%`BOJqw
zSoBgNU-^V{W9VI?>t=LUu25v@uFGF<%8z0Zd;TtWaA9s5t=DVOk(NIz_Se}MN-4pw
zuP;e{tus%ac=>5HQ)TNc!ho;RM>kzwi!TPWRH1noruNC@oJfDw$Pq<wML%i3A>^X}
z@sZ<UzsuAEMx<%pJ?D=WkXwm}8|mpO*7sVx=_e3vKcnByx3;(3{kgSuNziQd=g*3b
zZAhn0NWZT!XAL&g))wuuUi|s<Cny0r&#9E?nFcX@bSnLfGxgf(qa=9WxM$BC>z>1?
zD~W(QK3Uo9zLnKtkyF6?T*ZTS7CR`;D@J^1oLq;--o7k&Usr9t=kyV@2NyXixM-yv
zhgsLR3!KH<v#xR0jl3Ceijn<4F2IbnUY3afeYRhBBB85gjunGZo_4;b8&y6EmE5d)
zGYaO+;Ns!^a4q_pBWk}D|Dn!fQ=oeMs+#J`%E}#<E;8rJ?-eN^(kdk>`K>9s@IWQu
zQECJ$u^x5y$!yI$L<JBnUmnu14mYcud|(Q@0kR%Q7p`A^GI|03P20^^pCGr6C!Lrq
z8M!J}R?*UO3MLu)uD|ZQHh_4IV*5Nlr`EImqpn><L<B$t48N$9R87szvF{=IPvvIO
zh})E}x6xI*?{0NdHS@|*gaicO!*9x_G=ui}((a|L?5Ls`ivCj&2f_!SXz*UCxQ-^S
zA)pbU)@Fx7&?(+@fUgE2z0q&YMBeyecV&9qAxIT`tJ?(CAZ-Z!_rB=@XzP1ctXJxo
zC0K6;1c`A<5{5Amw3?Q%rS^ATh%&DPm4041mJq7n7#gR?Z%IV11qKFA<`$^ox!u;R
za>Yvw7h{Z<_Xeb8>YI!oSN}K?iFzX6wsckOgNBNE;#qfQfJa0!is3(byixmXzd}J`
z@r8Ik^`L?>@!^chQ;Pi&NFS}yNI+rA!xqFoh^XGvDZ(5dAB*?xT!a!*b*-4UQPAhB
zTsnFU5L>b*z12lqd@LAte+hQi(IiZIj&#<ZjI=VxtA3jSfLo~@O6}(*(p*#pwEXM4
z{yU%W^{Ng+e%v|lDUt7ox}fT9Pe#Qe@BPb<+-L!<bZa{1QNNb_I-&tp^Ksq)lN>-H
zVKnx){L%VmseljOWm?+e$_llT(R)?$Z6nA*`yb;Wj8f{=PrpAQ<R$)K5s$Isp+$>?
zvpTN27Iim{C<s|x#;Go-bT+60UVYOk!c9O>u$0fm#-?-4+o>7{=X}sqE^UIlc%BD9
zM1!=FDG_HX1iW|`ve7p;2|cvlR8mZQs1P)`ny`6=`C%1>T>02rYkQW_()fgggaacB
z!idecRQ|PZN30Z~hzpWYgF5UKO@7zy2N)PZwZ*i9OfI+;gr=}sN=R0>P3B)P-0(8A
z?xQ6n5J>BqQP33ufFXQyrB3b*3@+TPaLFf)cYj-2>I4Zu-??~V6NNxYiP&GV<&6z2
zkdbm?FCsi~oG%#F?5?mTO~DW!N*ce1bAI#UUWGyqCC;m~`@;1UJ~`hh)e$Y14gHLC
zQ38QML6AYc>TzM}86Qs-AWeNP+*He-(O;tbj47Rf{~Tty{+CiWCbqCcy&Sq(;f{6a
zW<B-cqHj(gup~^uS(&|6G?GloEIIt;OU|bL?63>IYKFS@ci6qqCSAa$!eqkWYkwW&
zMt@<U&g4*^;P&GqEc600SevzVE6NZaxr$BBas;;$ew}GQ0r16aA<wakfsm595$08*
zcqX0pUuImrQa6|W>>2UVOz|}W{%d)kxwpJ>p3=mJzj<SNMatVYxKd7T%-!Bz93=X3
z=5PPV{6XMeP_?xm0_3{-5W@E_BF^<nh6)wdbpS<N{B=W<($v{STN+5;Sy@w;@3Tor
zy%e#=ji<baPAB}O-v~1)A!AMXB9Hn7uRS_4QUef<vf<-0(w`5VJmzo_i{4F|6WlfL
z-%s_)WUk8v?jbyy6OjdnoNR1tJK+PY{`p41-^e?Liy$XsKz?tAV&@(Nd8$2{n|Nxr
zwKMTm?(HHPGHPA+wJ-7G83!3*jPTa^Q^>mk2`dz&ER@@Cd9?KdRNR~23JYPDJRLLK
zP)%=ZCnbL3ViP`t?xY2$hYwe_x2d*-Og}6#+XJyFklU|E{i|g{6G%3J8Sy)du}rv(
zt^s)K2dxUkQZQDe83sG;EXsa@JlgrB%#)uaP%Z+7Z;l^$_OXAmdB?c*Y%D9M!{!~R
zb%}yT@}*0c$k8fyh_iIXvnusF7pq2U%s|4Y%H&-*QU_#myrWFE#KZg7UcG#IJ6LrJ
zKi3{-dpK4OHRCFtb${+WH0TJb6YyWxDYAw|BF;DA5>^kdT)y15vJ3dxW)P;ko4|7T
zPphg&Xx%dkDxz-;_75IBV5I*6X!G`$j}m`bco?7FvLFVFE3viC#3k=zV`G!wBK7!}
z1)6U7>sOXvp9wBYd2>vBxmeF;mqbTo<K{*|A);~Te}j?9(NL)_zCZXq2E@=LpZMgo
z(l)IsZ=|SXlK0ilxIzef@nQ<(fO^3`TLi|(&+rf+?dlGJ6VPT!G~7^2QOj07IMqMm
zyhj`p_kG!W<1Xtz_)=`_z-HUF0AX5RAIatQKp^EXQ#t^s1Nwl`k3oP?3dWJ<<aN;8
zslE*e;{%j!j1~`4;F|#b516Y(MogT)yN+D)ZUWk>9Ei;o6|F$B3(sX$|GvIHieKOQ
z-ND{oWbG^Kt?<{oK#X{EsEQTcgOS$09uV(USNd65<)@_?!ckcyY_<@353|w0pKvn4
zDb){Lx(z245UUq3RRY8*X1Vyp#KxfqeQN3ugcGmvb0qYR4c!By3RQc^%#Im9BC#|0
zo|vnsd;+HIqB_)AurpDiC^PZC0H>6zg@%rvUfgq00d>z2Ca$io-fdU`xNQ6dqGQtN
zqjipy4_6D$IIJQeAz`Nr!(3!Mt?bDB#DpKrDqBtPqjyJ{2n3#CXP{11niKVtzqFWT
zLA>mgWubZzvG;Rk4CE5Ky6~T}$TdN60@}&J<Q6ZGm;CXUWgrVX5A^rPCm;Z+z>}jR
z$Blg$Wa>mrK}?LD|NH0>L@qZfVn8^AEmr8eYRb%;oCL^&HI~_3#pmYdpTM}>w(~O?
zUuJ0I>npw{SaNV}RkeCAuqr7j34CjgeDH_rYT+k%e@}Io6MAcFYC<_HVIk|~JQ%&L
zDS)+%fug1AdE<c#RsjQo+e4Sbsh0yDoJ=o)RN)Q;R!PasgYvy%0;Y9frZ#=}@Cd9L
zddbAh44yfsrVpO^<sLL+L+t@DFyQw9EeQ5KSayN%nuaOLphrI@<ClI`LXPn0`zV1L
zOqB$*2@Q(y;nvfT1Y_Y82v|Dx?S=1uSpD^j@^j1Lu0Lo5E_+JCV(mqGOEQzAO#~((
zxrf4x#tRb=B-;hZ9BMqXT3*n^=kG6B!2(mV&p{&d&UTlokIx^N)C#FcFr?M6D{5L`
z5I`+J+b>_50?Y(iCYkwrn8)FGm>IOc=?7rd|5za-F5^I2Ci>RV{!D1#9%XbbFik+W
z(8Syvc+AmIH$<xxz`Lsn{d;sYOol5XBU26EBO)@~*=Yml1Xx8-Ru0cCX=!Nzuu?XP
zgbS-<79g7&5pf5uyVN>XXPyps<~m)2V+j=9kuP7aa50XD_pWpdM5Pftx@>(|4w;q)
z;R*z>_~y=;NYEO)Ml}NB_S*o6;#4<T2#Fa$X#<i#c=)|+!B0@p*oJ3=Dm_GX=;2*J
z$D#P}Y0cYIzz9Mr5D*)GqkZ+;(+;XJlDuJR%*G#>_Uqu`5dtBXNdVjwsL(w+GXBjW
z77`MYnLGQsdto83z`Ei6`^P>$E1r_@a*E!(*?_VbGeq{icJRNV<XO^wWkXZo?kHa)
zC}KdV<kENH&f*Bc<dPE3Iuqa=&CfH_b3==Vw@dJ|0}SqI*vK=0WvU*uwYHXK?37C~
z4*`0#bY+W#NF#6{_;@euY!Xx?f&QrK;hIV#xKADX_AQS|BRm6?Ljf_3v1X8pM6Y?@
zWE=#HX{jJmLopu=J`}h1pfr|~ojp1>2H@D@>Z*a68EL;*-skP<cPp^rpf<xnLz2*I
z4n+#6%YbN(gxl<RyF~&atMQ@MD=`vFz+?8mk2zofwtuY=_V@*hoU^@s12)S9P@&1Y
zdv^8%SRPO=wQuyA2G9qzYTX<h-UA8>0{sx>L`6nszI^E=$?S$E=U0Gw74Xeq;b9if
z;%>%XXztCdK^9S)uKyL6+@T2M<fMAw0cv^{Jb+BDZEru&(WwE31pwDjECEsW4Npnf
zwNTK4A4m~lv#xkI0qO{8EdZDBa(Q@o7{na0Gh<Kx{p(4%8UjWc8-3--90F=vc)#zC
zq9UAJP_hFxx>`t6#|8#|!oyn#U3w(ndUosr#{-HcFci8Jo(wEtZf-6d+1_3v5f@X?
z$IHvigl?Ui5XP10ksM0B1Vv?l@Idd@ngGHFMVN20_YQb<z$N*dBQwP@XZCK!6|y>(
zgpH*oYq0P@4sdZH`%#^jw*aLCYjukA)#YWB@ApyQO%JC;S(_**Aee_=io3n4tlWb5
z3wDEW+dmpcMv#NAfpr4cn=9#6h(ggED*)sF4CJ=8^5nm?1_uQSagj(lTe-Va>ibPr
zm_pGg)KUe+YN5I%>oi&nB{<-o5a8p3NdP3f0hDU$IAQk$fSi%1fIZ;e=x8eJw)Sqi
z$fZ$JSH}#?2w6^k5u{pwj`v+Cqd_=15vnv4OyVMWr$h@oQ1^p(49LqKOpJiD3}zb~
zT77LT2yYW+4ZZ-0lNT?L=EsuK(q|CwiQ&3ZQc?n9i3o*p=Rj|1fqU4Q_TReNesJ~J
z76z_eAdP}lv)N~Z5)q1C64`wUclFTO`2*bU@UYITZ~>vUtN|aLtVALq6s^6RvRR)2
z90rDw_G@aNEmUB9y}iN8js4m9OcD--MWWd5S*R@^)IzBvcNlg&0M%Fy3~U>CW7mia
zD=IWf^u9pG2lQBJ>ErQI{r_P@{rzOQWA_U6N*9U7@&T`czdIQM9q^^+dSMFG5E3yk
zWCB%ZqjoPf80Iu!E2kQnH9i|a@r+c2S|im`o@KMYE*!UNH&<6Em|2{70p?>k1A7H3
zh-|X57Eookz3#t=bDJ9SMJQF}idc8TaNWfms6C{qDN)}AX=?}<>mMkp&H!41Ix3xT
z2A~_t{Nb#y1_Ku1tPr6=Qx-`$)ZAVx7BoVE3Gk*xcd1zn#`d}z#<K0H96x_7D{+#F
zfxdn%97+&o)p#Wq^Y6C3Jy$aj6B$IT+M%3o{X@arG=~cK0MS!G$bfa62X6<q1TuKy
z@WlQ#3n2+<X=%v%S)al72aWl}&&!fp_VH%|7L7ZgVk;^t3LFI}O+&*8$VcLaSwUG5
z#vojI;s!?@wmck*7b|d>pym$~(8$x|V1UE_tL&?zqWYq?FANAuh@><IF?0(ufFhy9
z&@s|6gmeo?2_j+9B`G2x(vCDp2n+~Fh~&rsBAtT341CAmTHjjlpYMDBymRKB+9&q2
z_kOm81xIQ53uYiOEx=g;_R>|FjSY(gcXM$5fudkkz{Qz~k_<o@KtXV*HUk;31ylcl
z*x-Q90pbjV6GRZTxJ^pt+!!82Jy(A-tM_ASs`cT{RasdKz{Ej=fOZdr?8&<VI<iZg
zoR$U#g+w<H&TkwCHuu`<>cqR(w<pKO#;U5S0GZX_4j>{OTX6{qd)wRMfZft1{`@&C
zGb7{v77hoF?AcjGad9uuHPB6-Gr{zJpxKF81LxLSfG%F-LX;$@8l~#oyN9axvIXlx
z^=bj}AMcL%4m<g^xzn(jKc5|<5`iQ4hVkkYFrVz^0K^E`qJeZzDFel>pc+FT2H+%*
zI@7G=<OMvwP>Ne~h(>)?^*X%_$5|=x;S~!TRGkF(h=!)78*~_*yLTzipD$HpE?GWz
z&HpNMLW4yGaupD!uB>0w<F8BY_D1mYYdA%woelx3T<|Ea%dR>=x=_i+s(r_FmyZDr
z%o%JMODVazRya0(Fn)Rhzjffis^BII5_V2cO&NUnQ<n<pHNhwMi~TN<T}kgR6*T|n
zYVJ{CPzoZk0%qX?0ws!~uK+>0W3B(@s7V?RC2(sPqzAM=MR>4<gD%pnf7GCh*5-R7
zp$>bg)gxX$?*t;ufTKGwtRqkQ^iW{4j8ip0p-M-4L2?<N2CvSusYtp~BTdg+7r6NE
zY-IyaJs=TE$^7>p2sDw@ts(M;JpInP`JQqZNM81I(WM8#PE5Z?4prmaIcnImW1GJV
zclLq4^}9CI#qhTMQuS%`TeX<46lSj;cuJToJGTr@eJEC&=JfXR8eRqp`SwFZDD4-3
z0zYE^c$qeQtdvcf?40w>xpN}^yDY$b$V0aUoF%W(QIe5<dv+@Zgn07M#X6En4uBVt
zx1_JVvQ-CZ0LcnjBdMkSEB=+zA>-1&ADz{CIswlU9i^%iDXZhZ(mYw0{*Rx!#(h#U
z)R*E9%oQ03t*zU9Lw1+H*Q9idKl|!98p15zi;6j~_V2YM!Dh<md1~z!O@ZKxl8-Rx
zC1EJvcu6?>tK-PvaL6lL?-4^$bviP4iT~Qolf}%4{a-ZvaSMpWvk$++NtbGRdbl#O
zm%o)y`~5l6F8Q5W<&=LL4{@g2@9CW7>xRpaYWJII)?Zel#bb_hrFmyZ<LTy7Eir$*
zmt~LRcqeWUAFZP|Ti%#DaGo4zHY4^skEx!Zb&(v}b=2p*rN*rVVQ9{N(GTXnh$ce!
zWZRZ^@cyp!n|v`C`hz%t42yzR8rl<e;fbsQYuB4vK|sSKZ)I+g<6X|_5HtNmA+5jY
zUxeS-!Oli`+0|!w`F9g;jVa0gOU@i(a$QlT58;BV+9qN6>d^)(*v%dI-El?^YstNE
zZ&~iJpA*jC>Bi%l!jGkRK{}^QyRAdg#kd~itTs|dM+b2dqK#Vq%;~nzqGgQ=)WNon
zha_w6d2loetc(re48jYjL7GiNKifNFnWp6Ddyq^)Z4;5YXh#Vqfb>PYD#jtdD9(yh
zpRLaIcRqV;B#7;xh~LDC`t1h%96PcOEvOW}f}Cr5bV5ujq|mUFv{YWltKJ5qVT{r+
z)vy#52=yZjOFeD1%xPd06p4CGJXH4&2reysL&H>{oifh&+N|Ng$HV(qGQT|AA!O>C
z#n5)!cSC}~xbGhJcBcpzKeH<YY1&5Smq{wGrQUiBgDCZF!%~}T^mwiSnx*64*tJdT
zffuj$f+ouQ`y?aMi(XHETAHA$d{8oY!_=v}&G|%sh!E}M13_2df^;V7%~<SvBRU9D
z3kq}5d7fXT0@#|enx_K?z6_rDefeas?Kz7zZf#D+OSylpgZs9w3e^+4ZecZ_=;lto
zv=$HDbsBA?O0p`WDH5iIAXWbnZr6Yr(Gk5%SOF)QC3OB;Mk&zH5*d!`3QT;KbQcmr
z!WWZsUp<FG(&OY$QpeRMC!dT9g4QrCQlSew@DBh(()RKMSFjGzajLulR#w?{K6Pji
z=gM%jvG20S96ItXI>|`KyTg8M9`^QL-&Imb#pJ%8m1S;nsyW{%_%oJpwYt2y3gje&
zV&m(D@R7bw$UKGaYdcfEI@wAL!1v(VyXc#!gN)t8r;#bfXkG^j)q&Rt|3RLsSGx7E
zSZj-xWjU9*y55F7azx9i#UD%xxvJW6YLs~ZM~Icq*a-0hoGJpLJioXp(uNZ;y5vuo
z5XTDU%?_?eEN7P{`YO{r@}AJc;*a2!u;Q=TY!V@lJ%4oqcYAUedfqAB3b!+Q+I8VU
z+`x4wnGT;(rR4vb^kUOd7bh6y_#h*0bul{2<4%k}tnc1iG9JOrvz>)i%j)xir}t|Q
z+(avo2>@CO)nFh4lQ?X|XsYl~t{Uo}9KB~`Qfiy2Kz=}T;8w?CMuiYkYSF<}wT4Ze
zcewy<r@#A7q+<%@pC^b39bS*yPRc&PtRuYOwQLj+)T|oO5Rw74`}#Vir^C*6a5^4C
z<hS_fO3LAW9+%<fXl1t@7Db0jdQt0#gIkL_-6xKP<>#P6$<1NKj)_RSBioePvfZTS
z9trkv5O^?s<R;JCu*<)56iMufmP6co_yooi{>W=Q{zDHQ|H9qOU)-`Hi0Ncatm&*F
z)%*^asw_Oc*2O^C_Hx9`c8yPBzG_y8OaHOV`*1BIbN$3JIP#$bv!u_hp7<gM2@+}Y
z^iN{eCnq1Hg_=y=TlLE>@t(Gz*YcclDFInxKAfq<Qc+l<J>amaOd$0VSQxSRq1H`=
zWbcd~)^3yYbxozEZZmgw$%XA388wouwvL0O5onh<s(9DAq#6h^3~A$1B%D3}OE5kB
zR4fd+^~>f2Q#7>3!G04ac(9}%=_|PByEY^$!UDGXF8P<?{Rf<F4)NjWFT@{fN1L3p
zCDJU~Ddf;kwe*Y1pLXBU-*G4tz9J&j*AQs4FlFo-fD<el(<xK)l^%!|kk_Ai`>sjL
z%hcGw01og4A5Q6zgOnAj;znMJp>`4|?2coMzkfRfMH@u=)&n{)Z@GGl+cqZdQyBk`
z#pSHIOlJc!*Lxd7t0ofu`tbrPkS1itjZp0RQ9(!l#j^9y2QU+LF0e=1fYZ%X0B~tR
z?e!cgq7xZY@@8c|o|{(eQO{|%+{D5D7F}3Yu0rCK$#~;cO?~nHug{AC$uXFM%dz&w
z!p@`NQy_|0^km*QudA&xP0uTOX^7qjrZP1du;O?kH!njA@s#5d2*Imk4Ze*^ZApbt
z^oW1ElVNbDN4-h`ep?PO-Oc~Jkg56h)h_Pj!OHFvBAG-rW8F9<eUeUN;HSflxmLNI
z#AqmD`0vXs1z&8rntlDJEyf3wA6<t3vUbA$9UTRFKgFNRTf9&y0Nv5~{IRK-gjjMk
zr=4O<$M&ozEtQec1dot?vfHJN@}IU@9<yHtCish%LA=RAb7?h@RN#T@vh88z=&AKA
zE~^Ie+CWSl%bPd%spp)rsiV^5^E^%Vb_rr@6(2L@M!c}3kaSI1jUaDvSt6?@Dt}Nh
zuqc0MvGstlh1(HZ6Z$k%DeyFDsh<yJdA!`3k@fq`ZZW5ARW*g<Otb7J2NU^X{8hBU
zk%l5Pk>>fFq#d*1%S%sVGdVS#(oGB%E*)v^|B)zbQGJ&j(ty+Nxs|XlhJUVI;7-uI
z5a`yNU@EwW&l6IB4NMKD!5~K$&-a5Lu9z_g1eB*&H;JMdxM1&&@4H@5W%~(r*>y~1
z-!W1qToX*&+cA9)_FMWbhobVg;u{4&YD2bT;W@7{AD)T*?d>sdd)#{$Rk><TQIUO+
zWpsK|&EoWCQ+cR@#vK9IuXqJNnd%P6Yj@9p*W^J&ZSF_p>H6)gX4M+>H7#EHRQ^RI
z7nK<%03!@Vh~Nwa_34D!TXnuJEVRwSyQabrvtLy1i||FnsSDeprp@k8)}I__0P385
z8{V~2I20jpPg}4Q^VfcX;W^n5eeqcH*5k}QcO4Rl?~%f*D_4Xi%v+7;KH`Ro`Z$Ai
zsjF?{y2zkt)$1P@&F!F{@_~6b_qp!bHj)~mMdCfVQ^V`=?+>+iDxRkGJl-KWr~e&W
zJY<TS{PK=+K!t+LqUM{j@NNCcN$w+8Cf0#f<;JzVF2>XxL_<34O_ige02O2v9MU6p
zXu&UfS}l~C=<bNccbCFf+c#UjsGhrDPF&LOZhyH>lV4uU6A&K1r`Oiee4(00h@$?4
zO%>Jd{{YuF!w75d$Eg_!kVY6pY$!`>bP@y#B>m=M;9+k0ee#6a)1vszTRVi?2I#vl
z%j$cC+gds3D4_m^tufDIQYeD&Q|Q;e#j;k;JcNNC_3~&r>B9iRsH~w$*kL5oxd)pX
zq1<@|t8pxdN3W!?UpKui3f3rS%hMsqsy(*S#-2BITbF+cJ}dJt6tAN_A;WOhyWh$G
z4Wpk!dAXp=#Jt7f+fKkEEU<L)V{I)(+%TZL)LkvML{i~>xxszoSJou^vIEajJ~CLL
zTCy|#bY2b%@ToAgsh{J@xGHjLoSw=4%Wr|CkkC=~CyxA$<rwPx4lg$?srPyMynD=M
zqELqaryc7+7zC+Y?wU?Ex>+9XZBVVNe32>)s#dnzp1*~*%R)_ig}!~96n%a0H*dej
zP1{?4{I)v!OJ&}~VN)npl4zA0yjj1W+W(M&Ak0KQ%9jFMw6G)Z#YHWr`rs>ZsNr-z
z$oWE2?sOohuDf<jW=-ya@xc=7q)aHRE3?7El-Mn4m}VM!c6fR6H;QP?BAelNVz713
z%qVsIb5Rs#(^Ge{II^Io=BY-7d>F>AQ2Aw+SqwIwi(6UjfOw!7@JlPsRk?+RNeV5I
z8(3K(&pSD(Ha>!Fz6$Nj3=E`<FXmc=bhv-U>se{m3@^h%&AV^9yr`|=R?<IYWwhxh
zg=#U~qmo|M$rC|rbMM*HM+PW#*qNN+@Ls2n8~B|qH=6;>hf4fGqG4PO*QJFhw}OT1
zqP^XO#-J5tkqUxKvz(W+?hUSU(8T9*ynbAixv}F%7k_;B2~#A))#HYU8bqSK9&~0M
z)$V_>E+pvYFHOusZ)w!b_QXr>>$7d406hgc`xtlbH2t0~kudl^IM~AyPUpL?nbhvS
zKlVBIxU63r709lU*A5<tKXth7@b?@hIQy=nyzAQ9kdC4+-oUQZ^ap=|$1(C;4;2@`
zZzIjxdha<%s&-*f@lwbXvxA^8d7v1==<wJ<<%ana9_yuV-OIRbM#s(3!Rtpqsug(Y
z^^isoL>_ui^|q0FkH)_DBI=03i;GLqN2W8lx9*Y$^YMcd#Fgo5`=Opx)Q~LB;6Pwv
zOoLb!{ozsYkz>VLN$(Y_8SIR~`+TJ5G109jz~gQDi!P<5m@vr^ToU<Wr6Cn(u>U*h
z@!WzMBs15BC6y?blJ5!mJL^R=UDr^~z!+>+gPhhDr&54WkH>F*?xU6Lu9Q^3KzQqJ
zf-lqNEWy3%Y}KRkaK^F>Z0nGo>96wx3gitwYlJOxQ)VLvK7PKL7XSYS6aQX$9XH2=
ztyw1`$~-ufycN|r<h*x7kbK8)CGMb6%%biv@JvsBr{q-TSb2L`LUIRvbUGJL@eXMj
zV!<FCWDpB`+xN4P?kl&>PG(+!q7m!;a>1jCJS2-NwJ;Oh6?))NVI6+Q&bf3oJ^YL@
zr$DyOtZ3bw0)BGfVvMC%^<9^kAhq}X%7L&D$&};S@0zOHsaj?=5jOY^Rn;R}o0@2d
zQ4fDdRR6HhV$tO<V^GU{oDw?CgQmIl-heqr^vQp!s&ApZXu8UO8twSw|EbZ&GDqRN
z9NOe3QlW5v2bVDsb%o2#K}<l+4XBIlvj$ov1RYc!S+CdnXtm|uHd#u=SX$K3=8BP`
zoG#|Zv6xZ(g)h80lM86;olzbhr-nse-%}r$H@q?Gx=wtwp(CDO`Y7NOFwV>b>DzA&
z4IMj;w*GKhN(;N97#j5CBC>}=Is7DZxT6O*NiEdW)RsH3D~a6D(igzL?6wFuaakJu
zzMX7G7yBJrmAFPH5+lm_T6>!8&;CccnI9p)6N~l-H^ttada?WB6)oiJ4%<D4dO7Oj
z-m4EPUL%z`OuLv{FBDbPF%r(YH(%3a=Z~GTWjO1Vsvbv}NvzjOHRp!|u0Q`@36a)0
zO!v3t#V}8vZ1i@?p#QX*<b<t1?ORBjDUs+lyz!FSGB_QJ`A$iuqMNla2*#xl*;p4S
zEkeDzX`gzqMT0pU^Jj0P?l@lNjDIDTa9pM!TP6N0VLhpaA^X((*x_v3hJkKf|J3FS
z_b28sLisR}SamNiAV=QLxp72jd%mr*EK!vZwoklB_?i;5ySMeO^-E+(Xr;*bUwX|p
zv4ag_1!s1~7$4%{+6wQMZh;RiA8F(;3}6zmX;hB(uAQV2KfQtu(gL>QVG+fJMbGHa
zv#orNB#>NGE7va<`QX_~F_Xcwe0QbFP7Ui1mh69j(IE65+0DnE!L>hZqy(|$#aK~6
z?ZTutA6An<8SYP!FP=1zLjQ1(tjqtZ38clinp-oZ4||4Ff9j0gb!{UQR8=AyD*rl9
zKj+?Xs65SLZd@F02mEhPY*MGz_gE)>3-<1wE{!h~v(AfORdr0%BORl;6X@C=>Hm_R
zS=kVZx>a}cUj_GR-7wO`<MF!uFBrQ|97>*3F@3gs7p_e)cdcu41QT3Gn}xJkI73c1
zC|N!?EE7ET+~ot`u1r)GQ`9%u>kjpBq}r7_ooOa*-&7QJ-Oi^KwjX+3S-rE7@WNAz
z{cw7VR;Ppmuq)=HCMi`DA{Q?G7VBsF;YWLbsX18rHks-)&ot9L3!F&?W~A!Zvfq1H
zwh)hX**2|ma)ka|ZaZEs-T10KO~W}&zKCY&%Xun_I#UZ&5c^xU$>6BH{Xw&zQQ=}{
z(*yOY=DRD?wVb%J&53|C3qtaZvs;&UYQ{WzM@l?{rwjkpJ{pBtROO)yE|g>aHjjNb
zbVyzNFaWT=<dfe2w^a9|m%ar$=(3DC^U!&+jM@KxEa$cL{{?8;3Ab|W*FlGp)PlKy
N?x`VFD^zSA|37*xrvCr{

literal 33191
zcmZs?cRZEvA3ttnuUljt;~W{uO4cd!93!%&jEszo%wzBExU)i$5JFT$h^%<WZB+J_
zl|x46Asi*Xcc0(y```EB;g56f`&`#+Jg?{Lb-k|F_0ZVx1`|CGJsBApldg`YDH$0B
zNJd8fj+PSshVPEkI2jownX$f^7W|Q%oScG!f|8Q*+_`g9R8-W|)HF0Sw6wHzbaeFe
z^b8CPjEsy-Oiawo%q%P{006MEvYtPGo{f!-ot^!{g$o=U9Gsk-TwGk-+}u1oJO~7W
zmzS51kB^_9UqC<riA17MC_zENix)2n2?+@c3yX+|h>D6{x^ziQOziUI%i`kV5)u-U
zl9EzVQqt1WSFT)<k&(H2^{T9_tel*jyu7@Ef`X!=A{vdpcJ12r>(`Z(l$4c~Ra8_|
zRaMp0)YR40F&GRMi`CH3(A3n_($dn_*1mD$hK`PouCDIQn>Y3J^z`-h4Gatn4GoQq
zjEs$qO-xKoO-*mzx@Bf&cKi12J9qAwo10r$SXf$GT3K1$y?fW%+S<m(#@5!>&d%=M
zy?gfd_6`mXj*gB_PEPml-*<L)c5!iWb#--fb9?aMfxElAhlhu!r>B>fm$$dKkB<)y
zhx7II_4D)d_xBG72nY-e3<?Sg4h{|p2?-4i4GRkk4-b!sh=`1gjEahij*gCriHVJk
zjf;zmkB?7CNWkOqiHV6I2tIuHFexc1IXO8cB_%aAH7zae(W6J{>FF668JU@xj~_pN
z^5jWYR#tX)_S2_Nb8>Q?J$sg$o12%Hm!F^i{Q2{Of`Y=r!lI(07cX8E7Z;b5l$4g1
zzI^$ztgNiOy!_RxSFc~ce)Hx{MMXtrW#!wqZ>y@Rs;jHty?a+vQ&U@8TUS^2{{8#<
z`uc{3h7TV;G&VLiH8nLiH@CF3w6?Z>{P?l0t*yPi{lEYI`}FBkM@L6zXD0+fU0q$>
z-Q7JsJ-xlXpFe;8^5x6duV4H6`uh9(2L=WP2M33ShK7fSM@B|QM@PrT#>U6TCnhE)
zCnu+-rlzN-zkT~QGcz+gJ3BWwH$OkWu(0s``}f7gMFN4aw6wImy!_+GkCl~`)z#It
zwYBy2^`Ad~ZftC9Zf<UEZEbIF@9gaC?(Pza#J#<}{r&x4zkVGY92_1V9vvN#NTlQA
z<KMr3|M~Oh<mBY<-@m7)r)Otpgm*R(@EV}=(XsR+BV#!K@1J~RF7kwo3`wS|i7^Y#
z-<+e#VV6gcecxI8cr+ij^YNk1j!p*i%^T{W#Ps>(OP2+IaAt^THPeDcQ(7iTu|=_|
z1AKf^XF{@{g=Cd<Ee^eQ?(Hi})cHO0b_rT|-sfaWmSB3REB4%{`50q*;e0HhA$F$}
zfQ*wah@F#p;-EDWJynh;UJ*IH5YOB(FXS|?)hIpZ#NPUn_-z&29&22z+u89`D2Vk9
z`x#~Y%5rN1?+bdhV0F3EdoNo<3P*ys;ISOwcS1<?HU?8~rLmq$ORLw~h|#VmpY-j7
zzHy%k1bqMBJ3ZK#1XsTVzGK!e$A{BGuL4_t<w<vR7mHn=*GjA?z0}71T*uz&O*{SB
zr`db1$**{mQz?}D3HoB!hpjnpJ-V6Yx!#eVu%PW~-sCi0ZV&!toaa1WkU9{=brUDJ
zuD&@=&cqvVaroK3&q%Z@KWpK~-;N_poVN!5c;DKqmIOCJuB-@o9<!ckuIm0Q0n@4)
zUz_+gMQ8j$QvUO0jONJX>D;sL=9GlCK9(UweepgM-gH*yI<U~6niLuTQ(`)YTLvr#
z9-OQwWu>;Oo)xnZVoJ@oqj-0uWzCZ(E3G*fd4D4eOBjHVqjRCVHha>H;E^mJwmw^P
zVTR#lW2Qgyl-y3V?O|C1;<%D&xG~c0ctxC)kGmsIK@>wcME&QMk+G4Dvpvqu+7O)8
zCN10sQ*v*Wn$3)H3I3Gbw9foGd~}Evs}g#<k<coc8hcpbFK=6in(qOg-jO+-3i|%A
z)P(ke3$Wv2<)(aCftlh4-dgiQp3kBbI{+|jug{GyVn^x;L4PE6O2y+UmSrS?Pad^h
z0!WwI`iHnDNuR}K1<l9aCb6Kmbi>DkXAyw|v93P|$f_@qrz_S|$*B3S>28kxZSBTM
z7j+p3^rYEfkEc>e>b%*$B~+|@Rr#$XtCHD%pxw0d-ylKN*(73Q_c4d2I42Epz)x&{
zkUMwV{DkmvbT+Lmi1Q{+gzDpE@3s9;^Kr&WOk%x&H?Kj^mBujOX3a$^Zo>R~jo22-
z7wmebL&+=uPwdiT7++ds@H07#U93!w@loAV>ClVs&~iaXfJfVnlQsu9bD@!*GSyu9
z+5MVz`F@|Lvm4ugrhfgFeCu-*kpO>pkQcfr)jX-PTjLgQ+eo9@p><3jq#<rSMkTNj
z?)e5Wka)!|@AuRp{+%1G1N>Jkn}Ga1C3kK(171i#M0~O#$hvqjyXa4ys1(}-9*IZ(
zrsaGNnD1!I4)9w&a{l|dnhC>*`Nb-A9W|hops;O;<K3DvDwuE68c7B3R{2l&G*@ij
zR+KunSalo~&s~&yfN$B3LJQ>CHkEO1$A--r*o75;2vUsLd_L$C8W!|*a{l<6ne=Qw
zgW{Zcgy&Dk8TW~@BV|-1X=0&M;p3swxAyU|@aLuO=G84DUE<pTPv<JCY<|@Y+-THO
zoj9fqI*7Gx{QBr!kgCGRQtp9&qm7+?Y}t0dB{1{bB4Dm({+O}+L(rtO^(e!t<K5h?
z22tpbJEz2S-7H#gdubA%v|>EGdC}URqrWGu<hWhn8@PfWU`_Brfw1vJA%tw%+Z#um
zeO6+X|G(m9k1mw~5TLp0{y)QglsYGKZn(ycw)l~O1|Z)7+mn_vb}igMAP<jho=iZh
zt^;R&j8}O!kX~wC-aHJb)6+UWscnsuZN%jDLO>12R7e7W4KMGr<ag^O>aBk-@Pug1
zjxi8kcqarD02gzmekdN$5tSyJZ^V*1qt*_CaknfR*CRH43IOTCMa9%Epocl|?LM81
z3H4-|E%4#YMVZ(B&pLTcXh90iu#MN~jTn-f?vvx3#@C$CCusyrg=)!$GU|{vOzBHx
z$O7;<Gvft~ZK_k@)J2TPAL{D@JGcFRsfFDVSnS@e<tRIB=nTPHk3Cy>oRb(_FNzgS
zR8lw~$LW8>R0Vi}=4IO5M#-GpJ}70>#FT?_>DiO2eRAApM^_8IdaIGic=e`%3aah+
zvz^+^MmY9FXjti48SM_lMnH!5tzRvy+Gnbn*BWD;pP%|NKgL(p*^u{eU}quC_$MXh
zN)4=H)M>`u9|(sE%Qaqqb<Zu-tq6=qJ(rr{e0162dswpk(TsnRO{s8W1+s!Xeo{Jo
zl%OO=VT~0b0bk8-y@x|<U0<6%{GhA3&flp-?D2oj0uCOrh%=JW80-E~ZB~TR7*+y4
zOdjKDm#!Xq?|(mtdQfW+fj!yd;z4tWq4*6Vc<K{BoE@cne62Jq>uqZge0sk`UH8Cb
zhg%|F6vA~j-flxR(24?g+ugi>G7vUMHD8tG5+W|!0@93{qt|!C&hG<-21!HPzQOCs
zlpePPfDH;fN!KRwIf9U}nPHFLRK2zEruNG`n}9FAP+PS6&VrMK@T~6Sar9J+0C)c(
zo0)_3-BqTVt&?kWn*pk*ZAIkag9vV`IyPDX;Hj6&h&MeC`kTe3L>&6Bt3~FmS9}td
z!_jZ|$!josA#_&CYFvB<*IV+~36CD{E=|3dy<7+cWpO<`jXw-7!W?#o$GbPts;N$H
z`cHcWq6GlO(}cDGylqJK-3yGME+crKI1<k}U8ahhh#~J@I=p#uT>z-l!a7hAo+z?C
zsXKxa^E>{SR1_Q&Nh!pJ5f~JCw+>LiQ3+wQQ7mT4w}681A*wI7hP(NF+c{+1RD~t6
z?eTDQMxpIf4T&c5^Z8g$<XuHEsN6pE<Q+cFm?c`-6s!>EG1E@_88l-1gUvyd%U(U4
zqdqw9?LJjaR|ToYG3M=17xOBm!}9=F+dFb?Yph|n0c3;h?B-RZ?Y5dNeT_m)<J<E;
z9~5r1pd!81Qyv-T?P}YHY^o4Z<wLe(6yGTw=@f=xQL8%`&Jk?QJXr3ovk@#$yz&N~
zJr^5{<QR}FCn~Kofm?TN%+$lnbOI?I9%i--`5o#$SF$G715r`K6J!Sk&Ku$C;q+x)
ziRCoK+ucgf20D(@>q;rGAMEcf&hFIQRz(^L7}MZO)EL43Wgt%)dbG9g9kaUW16$KJ
z+1W*_WHW++8H%TduC6`L;_j*W>R<=BW1^ObogMr1GYO?EpyfrfAziL>>BIl<LBEOV
zCs*LI<E<>;m<2dH_RkS|f^*O1hGRWW+K8mSO-s6aG7PKPqib9KSCF?WvN_2{>I~+5
zLnowZ9~(#Me9)jL){}NHQ)Oe4<PY>v6LGJJMFZ+%2Fy6m!3W(<za`fSIiqi%Cq?JT
znM88UuL1AFK8S@?x1lmiLR!M}VxGSXQ$G=N&uvfaVE>=Z4WfqzuD+W4k8x?Lt5h14
ze^E$m6ehoed5j<gSyofdo>5R!Kxc6Y`zX;f_A$nHpTgedR{3dIlYOLvR9>w~HH?Tt
z)QziCx9Xg`9%Cz8g(QVyMI=vesaHN`v6Vq~&%|8UMH)t*c6*D}5aVtd7ry;PUL&Q?
z6vsnp29%4$`f@qEVI<pR*d>?2owV+nrb!Ob{uDzJBF-u-=0Bv2>T(RwKvwR7lqp=a
zk2h~)qNhYuL+{(TjjbdwI=%$l0WfdL{({_bbj}Z+|F$08izYs+dv+R*=~Lv9r?N;)
zE<PI$m<QCbS}VLS#Un7vxio_hEW^6{u82YBykEv#aV?6SD!JoDS&V!B-iCY_ZCg#}
z?yy;;lJ$ZG3@Bn#&pLfo!Ap2@?q0s&-1`-tnoJguuEQ6z#EW==8jE}mYeQ@|^-Jmb
zlwMBvZUa~V78cxEo#<#km{U<YC_jmebQ^mU_FIPX-vYdo*w(!y;`k`2E<hFO9Le&%
zn>qXyS>W~tgTp*47s2Y}lk`7WtZGQ_ecaSeK#AI{Qxa%YzRR~h8g+?gcQ{eJg%~dI
zI<2N~ILN;;96x13fm1$x)=Fda8aaWFC%f?bg%mwVEMiRPmC6WLPobH9VRpzgMD|7x
zB^5v2?P%LO$Mx;oVz_$cwSX34W<pGhIomn#ZKv_i=9k7_4cBk5+{^nNZIP-z>$+KE
zqtlFv%$r6#CQXqyPW%Y_=Z9I144jZ<TtgxC1m6r3j{ld?!dF8x+S<`9Rb)dS6~p#C
zCae4<`8~Td7LYivb6>M!VN>85w0*mlJ2A11Y{pnsOw)1a4p!l!lS;z~>{pUxsW~>E
z(fAvw`7^?~`X{BJI)_fKtE4hhmYt(`g!(46Y--PJdY#O>*SK)?{Dv0iEzfQ}j(C~F
zt%`f<@voK>ZoXLY98Jx{@IGePW%|mbNo4?=_Ub^+=`?l3ZNQI7J2h2r@-=t%S5%<r
z!dGT9WBM9rkxTL+yl<vxoq!;lln-d(JEP0qR1iI#1emy_hfM1X`;f9(Wb^T7Z1ACv
zLFP;8pOx_zUy$lo2VaWM0Rmi;G1_V4#sZh$ou8$Fk`KrxFZ_t+<t}>3H$X?6mrZuR
z^Jj=QM?m|w^M~qKvYHFZI{EF(Wheckshfie0hm@(T3YBJ`m1>EEYSW5U|^!L{@dgA
z>rZTsA6dtP+zwN2sEGsGdIkGC%PjrA{z1!+S{>A#vn!hi>=sO|Vu@34CAfgpqjv!x
z4LUeAmnyv<(i0#dd7W(4wQJWPT{2-Y=$ija1i{8}A(gix<yP!fVpOsF9I*OKI-99h
zG<qcvEeaj%hNfncQpBYhrF$=H#!?U+VrPY0uqE#={;n~24qVJ!rU?FIVL{W_TKt&?
zO6h)9h<hDdab~e}!49Z9LIDVb$oz|cSp8o8O$SjpKr_8|lM_0axhxvnSMt63Dl);s
zlR@I9*Yc~l5It1(xtv{5e4{q=FuL17RLLUotv@FG8pe{%fgO)dO@O09V@-imfq{85
zYM%Ovp*k#dY4-t@d*&h@&3^Cln}Z;ZUJilWZ;SbWd8fPlf9QI=@)bFyii^oOaKYQ1
zt`M7c{`sOCH>7vZT2k#%L$!tDrc4VPfs!!Ey({meX8Kn&hU9M6F#kgj5yJCn<0^Pu
z#Bgk@jq##`P)p+7arzg(hl^c<-sCObW~@;#qtNN;XhH=_QV$_Ba5j&liWvzO;qtRj
zH}82ij8NgQC&5B`uQQu|0Cwb;v{3`aFf39lxWglBW=ZDk%)|9wAQM?x(JwXU_)22Y
z!+*F?u`uK7=KkdSDcNm$Jt{>bf9|dDW?ZH4b?isE&Of!V$)yP_tH*gvVIRFu+7sR$
zQq{<)^2PFYUvzI@!{;nnOz^MobA|A92-0bmu>aI=d#ap%iyCJgveddRj5fDsuKA*E
zk<-_U8JI2w7!c-7%L!pRNPc}PzRG*1AhAOydv;mv8dThETY==cuXrjiDLlzp`6Cli
zjFVBLduew+|2-6^9f($h%q!w>zvS9#bBWWI=CGji^;Da9{jM2}Eg?yEAnfeib92O*
zw{wLJ@Tl4_hxsXXx$`79!4K0{tETt#7l|x`jY9`%OA*Z6_r!*R*oFI%QXF{c#L|qM
zNck#3e#{uNnR!0yGddt$UC!{<H_I@ZK*xKW5OIZkIJklyWFtRc`0<44v%=#le+{`(
z5%T?6Ie~Y9m%YFMa$C%L$_CHUhaMi9CR9~y!gw!dc7Ja#eG6A2`n8|m6!UuWhpq)?
z_nwW|;23pfdd~&fQ<o$B9FQoF6j<873_e?R{`o|T^Rv7;1(78>Z9?(60X41?=bKYX
zablvwT(f6kI!-a31$@SEm1my^yWMUg=riOm`)Tm@MBaP=?)GQmg%IwDMSL(7G9zP&
z$uu_-JK2nypw0nDgi(uAd^&ajr`bhObxYI>o2f1|^t8~OIS*-2Ghy7{$|Iu~fIVSp
z&$FE=nr^w*V0Ps(10mTc<H&OYTh~mc2<iJ>4b;htTh6^06Fh;>(W1cZa{2nwCgbX<
z-!;o~LN`Om?d~jnxSTlkEc`QC{)xKW&BQxAKu>L9CWQm~lqeOM>*UBe=>=-bunL^b
z>*!3}9Hk;YF;Q+!8Og2g|2Z=UEF>J?CW=uzDV)}Dr#tczrVRV<$qR{1bfEk6kf(aV
zOwrVHG4IXZWik+oLYE`*>WA!VeWXBcxa9Y=`^4E<7uiNRH2<4N_A@d~N_Q{kevu-S
zbS>q|uPLE8aFC|2P}h?Vx=ki|3F7g1%Rt~dyfqR{x}6y6pr8mbI5(O<<ZEf`;`VAJ
z9?@M&eb&_QZ)JGqm|yr7YkBQ++c(&c#g4L9+O~`PceiWld~ASrWb1DY73*uAR$m)Y
z%&+w3gIF#=jIyI2bRTCe=ZeL+o){O_w{<g~2jjFkX_R!4&M0{wXH$$^lt|VVj|!(|
ziN9+VIZp16*rHrk!ag1GYj!}FhZlI3yU0OSW#q(UYz^35JylJ^R(d7+sVnnQ`&Fb1
zu%Eh4)cVLU;an5Tx&y@7yRxMIYo2OEW*U{>h<cT}|JD&k|DOreP75OBBo@ZLzS>34
zuVX7HhwJZDVS5dJdrG;E61UmFqu;#ZfaupzZ#f{tXrpI7m!)GG&rkNmE#(Dj@ODMs
zqz9!P!Ttx2Ju0=b5rhTj#-`{Ewyh*Jmqz>Ug`VOuXXlChX)5BS|459{?RUbnj#U{r
zM*$j@8Fq8Q3vQPZb+)q)*J#FZKl*qYzEKeqeE;XPL`!sezKSv0VBZXuDfe9Qi+67i
z<jb<udC|2X2ekJZdEskzcEUcBjO^#zRus5vRgXTH$~R0<;e@iPDT&D<&H+PC%_B_2
zR&Fnj@5vk!WL=`raH1qwS>GxRcH5Vb=hQD(hTTZf*j%W3h~qz})jRGb-SFB7ocn{#
z(momNW*YW*fB*Jrh{i0f{p-BaKG#Uu^PuO}#>m)J@FM*>MxR>DAL-S-y1g==m9B1$
zuf9J9+_115xrB_2K_u~dX-chcmBo00t9?GQy(Ko3Ta1Yn)OPfBZ_;i~FY*FWLgkAj
zi2Y34GP#v2aq5Gl*b|1w6*d#=LyC}dL=na5vZclyMa})?p*JUg@)P{NOq1D*ZIsIm
zPKxLudFb!nwR5~o2a#W-cg09NO*f+wM~OfS$(ETMS?bmg_;q#Oe3=Nv>c5PhLQCVP
ziY|XXq=$V;D(Cj15BVQt61k%IutC*1d;{b}8sDyjcDoaV%|J(&9Hi>0vu^sTidRrf
z`g|+d{Q~lB(iYyefY&NOYE$(U10i(DGqFq28?B$I?vX0K*Y`x98W$WX3EuBSK4gZ@
zk(B!-bocd;Qsl3UN^aY`+W?7oEln>W6`%_zg_DX=1#Ih9M&L?UV}v*a7X$o{=)b|{
zc*OOifv7~WAIQ|=m4X8YKX_>(b46o3ZWRKZ1&Ns?U{~tyR;=qhx5&8ZRnT2Gp#7Q1
zO=JV%vW^h&r~S#(xZYgtCS<luMeI00oRQZXkvI=}=W6BB6ZAIAg%U}ORZS<A+0U-!
zR|ROCboR-B%C*#&q(Et&C_bGa>1w_KreU26sZqMAQJWPY4*YocJV^D~MtXv2Mp1K(
zN(&btu9V@U@!qTVSyt30zLkN%^Nz(}@jNKmM1P(evTq7Nbx+1{2^5w;6L!1V!k5B8
zaF|W&xtTR>#R=Kp;n?S^`)bb#QRkfA*$_Is-#o!Y40_kbn7<k3JK#fdohsb6^-{{G
zBpfa8G)5RZEG^7PSnF$SPRHhZb5=?vx;DJppic0yd>{KZ5F^KuaOZilA~Zhx!Y%gi
zwheH17pNPmmjeIde~iRh<bD;7<Ep-#BWZob+8eBnvypTQIts+_KDe)28m*HQ5;`s)
zQ*{tO^<4Se(Q-IOLO)3Q=|2!C?H!$;IIThHJMcO@S~=%H?mJn5m>E-q2GUv0rzBlq
zzIgl7rEcH{l412p-52y?Y2k1%ou-DpEFY-h2cz&ML(NsPmP#VN5fon&2G%cVV5u~H
zKk7u8RQ>@G1WVb(5R*=xqDCa4SSe-OmieE)A7FidGjMWv*T7s=b5}O}n0)&4+lN&i
zw?sAT3_rao^Nz@_3dFYF9sPpgw`3FMgo4($&=*r5tIM$~3`b7|=_p8o!XcoW1)39j
zck^Ds)z2s?r;wD#5-=4IwZ!$5si4q6k#HmErWfea$GgIkAg_+YIISiw|Ecdhj?WqW
z(sO;Y2MtwSYcw^W$`Datj$tF9#5KgJpe*Zm9p^y<c~o1D{5#-lT*UbeCR|%YmvEep
zv`r^Fu(21Wm<_{;J-&wek}tS<0<#xlG4H+#(vjD}B~o#UR_qk3eDD4h>3}Vn=lQ;O
zXct>i3ZD!b-S+TP<p!Zr`HIl_lZu<j%$czFQwMnTqY8fMK2;A+2-A^wfVW5-RKUZ-
zZ}zU%-4#Q3GJUVB862P@y4&wqur=r4NiW|aJW91Wp$Ax1Be2kdyH7f!KFbedEfT+v
zpW;%C!bRkGfmpOR7(Pl%2VL3=3jL=5b5Q3Hf53m<VEwU2$@BmkHP5#9Klz_+`Y?k#
zbYVR|?>X`i`0W)bsE76Efd$q*UDUUp>)xO=?V|1dMSeQLw`)W^n!$8taq33}L$R^;
zH^mGaz<qC8yXHBZSKguRX=ZOk9<J8a;6Q62!Q=1|x(pR5v`LMSy-A)iXNAMgN;}OA
zc3k_>D!J?!KrIEPM!QSxuK~%#=~k2~GOFA>f3qSGlmB@u--fsvdx<w!2yowppCu)5
zwNNg_PCXjzZkA|9(ob&=N;LvX|FZoG`eS9(Hd}4YM5h#}7au+kASi-{KI>05BUN|)
zboyhK5y@hxk!2=aa{wZ}^ofVN3!0d-#AJeI5zvKS;^*ja!#V}rmI5OZd>AYn)zwSc
z5^prGMg|)4wi!&N9{Sh26L!B~+kW`h`w@1ZV;gKF^8KW?w6JZHmgv=!Z$|smS3~d{
zV)1!jJFbaBQhhFZ{T<iJSJK!!3`#0&*b<Eh0|+FQ%|<TTTG=+S&S&6bMTZOG%N?a7
zLy3UyFuY^?0Q*T^fC2CPU0TJwgnf5{&y7%@!cwIV+bttmc?3TbLz?E_1NVCZ1);Oy
zadW;!h;oykV>|QW>;|??as8*eBJc4z@R>^I!TAo!dy9g6$D}*S>A49lYrTQ+ka1OQ
z+l-3S5BxnN=lr*9VD(LZUf_`G$S{i6nVfI=-W#Y+dvw)?Z*f*>MPNbj=KaK$)r{2A
zCRoj#6m~aHo|g>6H?M2$d)X^axz1M=TawaSlbwm*66GC(F7yj(?i5?X12rc(6V10Q
z<!%t`Q{lP&vR`t>`^;<YvsVx$bU&|KnGZ`~e@W#g3^sDlxVgX|7|Sbcv}D69!+kpq
zd9AhTXfv38v5l^RnAl4k{qw-a{y)Fp&@=*Bu?pTG;aQa^B;FPCc#Fy8<jqIa9YP&W
zg}^rr6~oLPh;fq6!!IatY7$R!57?94z?S&AdrL{p$Rw>gQOz5}eg80bd3hgCVlUjv
zK-qHE!qHuPc|SflHt?O`XGAgSc3s(|6nNp-gK6E0Qxmt-vHpNl^RSkbbom*~ktPxb
z7I`1OW1W_gE+cEH9~y!L^6Y?39L5^C>=^XT2H5zA;3Q7*%`by=#N70L1fd73s<0Bv
zM!<dh@k9ug-+YUXD78?QDUkb#74|zkzBkzTI^;Y|hQ*|d&RJzLOeU=sM0_tN;^9YB
zQq@P)0Owwa2D?|g>Q$)Xe<m=OjLR7vIP3)mX{V?BxurnDBbEP8`DCR9p@z<D$fUyO
zHfsFr;QQ?cz8N#>8c<PlSboEPB^GVo3m0*j3sdLHd@-A-cvy3-UU;NO<UU(Cc)A{5
zBe0!rx={wpy9f=U^Q#W=VD+o>{{nQ^VA{v&@9L#Nb-$1t2lz&fVDn9<ekxpA`XHj%
z=k5V)P7c_d?1TRIfu@l$L~(H6aD1<`m=9)FYMB9cG7yFl+_$0+5-_?4r>UD#@s1kX
zkeuc%L#(^vRQiGxc=FI6X9FJpOZ+VVQCg7i-al&!l!-!m)@2XUli(zXl|ju2R9hNu
zv~YMz32QzGSMl9Y2t|w3i?cM?HlOBiR5hYP^5VP13%8YFn>r|>coDv$Xn^f*%P`*a
za-2m6>}B`<9vBIIz+MRU#QdNB(!S4r>L8O`ny-S^(sy7t&BJ66*V%$JWZO#^9alw)
z$a=dGG9o)ReM~`*-sj)dg_u&8*0B;lU~f6D!C1Vaf-3k2Cw4V`BQRVmAc9VlD4|cC
zu`#*BWHPHXVLwere3G7;`Wa4Q4qeI=AO1rY#GgLv;i;K);XyA|O~r0)<s*B24LKJ3
z((w?hccK^{Oyz%D9^5U6LN>ykMVlJc6e;jle7HxFfhbgmZMcCa2|RthUZQdGa3fzR
z&NXuUSFACcAmZ^174GV>l8PJ4KdZNB(AQ~QcZ)rXPJpd0*ec;J$;}Ivc3M%BDg3Et
zcev&a3efL-K@fycg`%2&T6CZ5P@A0GmcF;h>r2XPdwZ@pxXj&1^e4VoewB#--y}n9
zgzAopT2@yZzz!8v{-GJxMF6;!IpE-ytB0aD__f8Bcu$>CHqYul{$w8b>Zf#f9zozR
z0ND<Z{4SQh<;-MIdz1ciVW3Oa#t6JY-bkG`5A;{(+5&Ho&T=#8+88w(p!4&B8>Ptv
z;|Ns^qExVs`w72ff=(hnaMMO#0}>5*-FNtWx=19CLz}*aP@)lT$wDwuGUKVrs$nR$
zXbr}mB!B%C3kzRh_tEbE_5zfN$Mpup{!a_xx7!x*$&SpDlj7Z9?}!{WK>uJ`L-7HA
z>VoNG8Idmlcw~vDfebtH>hlz-$sO+3S=!jX8jt-4p=kuc`TOsYo5-}uI1=C2Of^q3
zWWnloe5+*4rGoWP><O#ghngo(U_O#lQtBUcP$;>L>9fZ!Yft#c%c`|lT?an1U-{9k
zgQB;oY0Ol`qX>iH5nS9cZ_#5!yhYl~^yPVgcusrS=~5dn!D4YP414nBqy^J*?>e_8
zQMreWaP3`}RNY-|6$)r1P)*evTp@|OhiYh_7%HO*<d&A@^0Kr3xxT+<(d~A-A|#cL
zSC^1ZoWcax1vVnpXMw`jqa@>}22X?&Ecy|G1y?ik8HF=6cnzA5f0$t2TXi8?%!MMz
zkf(HqKMs6hAZLDMBd`Sba!trhAd0X3?rJ_tJz{+%exi(AQRo>q&@^vBrlI6Le@V46
zI#`+7C^<<~Iq$_!@sl9!>vZ&*`*(~bU-eiyhGO*Z4Vczu!U-Kze_9%tZ$|sWelg!y
zJF{TjU*8T+VxC0}+FM_6O2@DpHL9Ra$qx<Evd;e_eMX0CEvOttO}~?@gN>9{Iw8O)
z2e+34H05=s`E-|Oi^&gB`_2PO8qLqpJu;L8V;j(;^l=Uzj5-kz`V_G9azi6R!7-1a
z_*x#Xa@->)`5(aNPgk;d2|jy4*Jo7`s~vR2qT{34?%Nn7@{~DHBj-#P>BZ48U5CEj
z!S|`|Rt@fID=K%XI{e)y5=Pieqc86^Ij)pgI~v_!bA%`HmWl?MK0@3-=J5sxeZ4Pq
zOzDVXVsdy3JJlx1cxoqdYykwJ)5g;q3_d^Sbl=qSx%x*G^Tnugm!tyV?H%GV%J#MO
zRgjP+kmol;N0gXKeQ6F0FoL)E+%~hxG|U@ye@T0ZCd+~%W9;_JxbPL3Z1<PnX%%OK
zr`LB=4r&7}f8ujuFtf`{HGF5bZ0|A*W65!jH)nHS`)Rm-`NscmYZUZeMs;@l01t`|
z8_c8Y<hQF8Arv=JxcA%z&5Hg=d)HgtBn8=_ahO)K3PmB=dq>a1cZN}skHmpp165x}
zM*|}mQ$%wGZ+t5$MT=R6b9t~T8r|1uo&Sk?=MX<dcXgoqnjzQI>AT<g8CIjGvZJa$
z$5OKvGJD>PyDJ;uc208odEf%jSwz<+ZZil2PgC*!lElQa!aP9Z(#1LPUSXJx(Bo`l
zjGHp`dP5;NIX=fm@P25EQcGw5y!(D;9#DLsQAt0(XUS;;^F_=-LpUuJ?q#wl<cK!+
zx>OX49)o51Tqe7VpJKh5F%c;G1DM&}slV-mD(k#x9OOyiIUOOq02C&ckpxTI6ruX=
z!dvp69U4fzKbdg%n3YurG%rAp`W_USz8go2XR2?im!Ce-P<|dL^IAv3hVJ{{bMrtK
z{?1i!{Dm`j#qDAkk5btzA22dp`Re2{((GSeM3ahfP3Sy>-XC<tKobUnwDza{7mCot
z?A^EWa<2eI7wCuH2d1Hn@#1D=cJ~KzwfL3BZ<Q}Zp_X8URXKy&T!W{7mYo-^OR866
zer@RuDGd|=K|h2$PqX9RX1QEKuGI7h9mbQ~n-zCIiQ7(JCUbvWeh*$k3h#{VuY!dH
z`uK16iPt2S!5yYy<zUYl@|TP18s2ckv_38hHrz{tL8tIvUyR*GFKS-y=I=f~d1USM
zxzzVwXhgh38H&IF^*<wi{IjiCB#WT^Mn{aavQDhMVD){4HO&28O)?*l=dXC0U*&ek
zFKxQ5$r3#YR3E=s+v0i8lqL#o-PW1&t%nN73WeibdHMIgjZ(opEd2K_X{xY1Tz*OR
zsk)PA+?NZFF%k(Ctv_>tJW-g73kEzFsD3Qg;SWDDh-#X;JrM|X1U)~ZPMyOo16Dq9
zj~p((uC)bL-^&d=T3J+XZ%b*v2G<WNnN~<e8`|eC5VAcQn~Wdj4V6&LjFA&X+{Oci
z-1V%t_B3Rr!FnRy4ZQR8t{b0v`>aI1J8x5m@kt=xx62u+%<J)u^*k|C%Xa?WY4i>^
z#VxeGK~dTKe0tsQ&ywWHCdzm5y$^QG3C&Dl95-ZU)vozaVgLG;G8uj?od-TsdBr{#
zz+51`$xzhXdqcHIWsUk(sR-TfF5Hmp4T)nHt;2_6^5=Wein))1ww~|`Ha&ThnaW7r
ziV6mZRD=oMM&GhkgWxtsSZEmv&W49WDHFj_QNQ9X7z9>j<-exQNKpJ5?C)sOHn>bJ
z>^)h&^<D+#^&0a{;Ctv{$W;*kH2l|tmp<V}py<_Nffp=tl}^^O;prm?!Bh0rMa$$n
z4Nujf58r-bDLV%<{f9)JDQcSjFuylx`-*Ns-jHLU<DsvrAUi$x58Ce`4+x_NhTyV9
zggsxK@Z9E$K&JOxUb&ytX2O*Tf7X4Ld@7^SbyvLHRrjrYP}^VPX6%ldsUq}h4fufS
zMZHdaS7*fR71AqFD3^pTd>Im0H@6b~&)O;JnOF7$ODr#gc@J1Y(!kQm`P-$y{!*AH
z=}DEZi-HH9jX>02yYwrr`}WW8*Tua-lDXM9`hLt=J<C2GMAR!&rS2veDY!n(518`D
zTIaHK?5V$abf}?%o<?h-8AQLxQ~*ip?aI-J*CAMFDN(_~Y|=wO$t6b=Do0@ksDj%Z
z7;!YS_u>3i*S)Jr%g6D%%B4o&XrMEqCTt-##-mt56tderuvU|+m>V<&$B8XpzRg~$
zd%z{iN9kARQ<c_xe{pKd+0P7692535!)?4z7;OEoo8*T*5pBP{VbWMlUXus&&$D5p
zApsuq3YSZIKbTktW9=>eD$H*)6H(7NHN)d-Rpq$Py%Q1j*IA?@R%mv+^@m(~zZ#7*
zjLt1(bXnBA=7_Qv3HO~6(Adnips!hUKj>q#^agJg#$N5={C-!r{WB`_%gdm=Mb;rz
zW`5{-W7`h$_mz;#!KVt0aGV`OFFQrD^@nmyHlHrj)$ph$^v!ofPmMIT@<Sp=KA25g
zR>OHO6(cI#+^tF6A{cf>BDAbp6r$`taK)Vb%{midd=sp3axEF>cKwQFA&<6wlNQBr
zBfYK+8B%aOi<#~Ptc7xyEv2gh)JQSN@Z2;M1F5~7%?L@}RH5yEQ!Mhj!xC*WaH)5G
z`jXZ?vMR=(CPQq3ReY|kBr&-x-p|S?!`BQKhAD!{I|lU6O|3I!m`O#%U4bH8&mm;4
z;Zd2W_;&RfzoSG>ug-f$U^m^37~v1Yn};O)no<8pWagr>UqdCGy{s%{>l$1o$|9|L
zskx^e8Mmg3Lh?qBz3wzV-PmAC493o~{3t@6s*4w~6unV|3Tum7LV6U=in57;s`~xc
z1zV6jlU^z*qj*GyYp|5-A{DDq%_LyOEQ{obmwN34>}$?R)b3cvx>OXGxpn?iNiV&G
z0TQ66BpP(iCzJwtZ6*R+31-B51&W%p)6S@02_%;vKwb{uO~$_%<n#{N7IQ>UdN00X
zyyiz^EAd4kK;7eBzIJ;AqJN#l@XPSfZ~uq?M@m6%lxWb`r>I5!r63gu;qa8FX?vld
zj$uzkf&SBB0kGf>mL&)aUwk3q<=@9fP<=(4Fma*ZU452mycNaY?_BxEE>F^k+x&@3
zd6<TTy>)ZHBvtUn*3bxppx3GLYa92a_5iIHe`vX(RujeK$k@z)3Vhod`@c1B8_#RA
z*z3>rWDt!DegTSVS`Uw$ye%IdC-wO0Aq~47?|yuc4+@d~W>ISehG;<b`JdZ_0W~^)
zC?QxnQX_27D84P|9{jr!3GozE;B_j;TcH|4HedG|=o3@we!hPz2e*wWGAK>TAd6{+
z0xZK{&BXM-{vO|0E0$Yu9B|cYJ})(j7~jbGMtw10{3}<&@D1ZR1OYLf%0e1w4;tjY
z5d*5VqBs<hZNt>N1G6y`wgArN&+O-|6w0OMy`ewS+)y|*M0qJ>;Q{`G4~O9sS->|i
zq1~B%;n~8aOVr~lp-{3?BB~+0g&gZ*>}4SoM^dmlw{InS91++_Ckj{KzEEA+;i)Q$
zqx=*H?j#jH()>eT^?7JVN`F7zVxh6%ZYZKZVp}dnFQSjlL5q!`nR7H9`m67XG${Vl
z|4rdcG#AwzBjKPG)i!$GKzi>w_qVWza0`qvUs7fDX$m=U#{WEc;Kw6y#nx3M(Sota
z9HRPfXG`jRO*z|z&PB1q6$3Bt{L5av)Rf97LV?9`i5G_w_{Qj!RLm5oZ(v-M^PxHw
z2N|!@I^A$%8jJhKCyX4~F22iGTIRC3yfW#CWH*2Jg!P-8GiHzinNn}X&@P=d5WkOp
z$3RbQpI#gmO&H3z53+|zshsTQM@RKdU+V6c`*@wuhGlP+hjMftYfP*U<RO67{|05I
z+7c7-ZG)@!QI}I~+)IO2)OL*3yHLA7FN3kI{qoIqpG2Mm8$W814bhj6V1?bSkAAg1
z>@~Z7RP+{Rgu3ZgF}@q0((|N0y=%984DSIMH0P{?#?gY4>FS$IP0gKAb)m(vXL^iL
zw!86Q_4tg`MW(~O$OPL=_`Jkc(L-1*Xa_!*mZeImvu{oBZhXMRqq(HaQKeZNr#1X>
z=ICCFDAfGUi_kXX8aY{aMRf@W2mbyTg~z%AM|kNFjmmFwEcmGxG`|TOe)E9cPV~y5
zt-c6kH8I9Uc-vvj={1ktfhrzYa~r`qlZ%BHgM&DUGeo10KE3%v{W@jv1cP9~w8!b#
zK*^%PQuF;3c85=wYS+)4;7xNCzYnoK=GFH_diyh7RGMp4Qc6tj4gxNytN;AEHVAh$
zM$Try)6D5I(PNZr*9#n6hIJGwv>7#^2&Zc8Vni@m(<@HqU~Hx1iPiR;b~(#g#%|AX
zMggF^Ki%P3mLB5B^XSXxj%%np4|I_O>kkV<HLRz&=akQ=pACkGjM|M3k28Z$eY*DE
z-gWiNV<Wv5g+d$In2y8C6W~sO9IEq15d~gWZ}MP2ycj!ivoXrH#TzQKNhMhH#y8ev
zpL5{sonw<)Xiv^PH6=zY)3F+9zu%|y8@2i11k`bDSQWmv{mMX4qjP7EIp;BI@+lUq
z*srd6dQwR{V-_DQltDL~J_!Z8*5sdlBCnBCUwP{M%knvo9Zey6jS6?o%r%cwL-~(N
zB%C;V1ud~z?)?&TB$j8v{h$<bulyIL##&S1e#dkCDLvtJo?bVR&Um3wF3qt?*y6?y
zK4r4HPMgFx@IS#ycwkem%x@=GdbU#%{L;e&tS=^4|6ONdeMEU0#6AOCSW}tBsq|wO
z4b)CVl0S@PVf~QNR)ue*xultCB~hZs48HWaUBJ4{K79fzLbY4_b1r$ISY@l455$Mo
zA4Gl}#0QfX;pYB$+@gahW`p)cH&}P#bLcX+t*Y?gD$(-44qX(-Lx((OV(b*_-t3#M
z4x45C&iUHK7qil?n7%4XqIC56*O?GA$Z7?!tgA~rbTM&U6aw_rdN7;4&Vv5xykrqg
zETe3MC^A15cZ|d<238eg4M<i=axBimjM5=&w#bHw^{Ak+l3a#&@In<nc)pR|GgFWR
z5MPmJJOBmUuo5xus{zRD&1^eLhp099_|h&s<`sDjntdcDSmlDy67sO_A9);HK%RCs
z&s{?joDp4e$OcE3cW7Od?JDw{!JFrRpYNhzOyoB&@cwUp7FEsJC+m0~f>-pQKa!)f
ze*Zd<d@vF-#X`NNesUM#z{&^aVOSglBET+rWW#V`@F*TVi*WGb16Rqu9B6$?L~}d}
z?6BS2@9?<AN{})DwQyU?3`ZKw90PPUI)xNGz@H9vKCnAvz?a#9MZp|oa4>;~+zp=M
z0dSLJ<_E~^q3EeEZBMZ0!QkvHB_6<z<7TK47)4>W>7v2gzfznM0v5Wd&%UOsp<Pz0
zppy9%otPN>g`wD_wQ?xVLZ8ino)2uSJH<$XZ*M+O_jtiXe0GTmWQ(V)WOR5*hP74;
z4_>%`<YHtkQaK)trvC271S<I`ARN91tz}X=pt$1XkW!R6BN$2t0WL_x>DI^V$bwE;
zzzrkug02QWr?kbtbGHH#pAe9>4S$)z{@tk~E8qwv2_6T)mkNyQqtqXP`)AkR^Y{OL
zZ1)mAfB(;(37nBfam3f;0t+`_g`z)Q2U6`0!a@PPpX#&V%tbf#O**kNj1_Prs{zG)
z5Id!r?)NaF-wXT`fK223`9>La%6#STJtV3ioboL?EpT&n(oLQBf(1zR9#yr747AdG
zdJ;#Xp)04Xk>lLvMO=iKh+%x7N_?g*wM=UYHKZNO%MBU*Lhc3JR*a5RW(Hlz1B=zM
z`2pPj>r&5BZVJaJGU1&TrUnDqcY7PYP1$U_EhX?D`*s>{Pse!hPuo&id{Oa~ox5nq
z85|RK7{K$$V*z2_KRcRM9lR{gDBS3xCPM6O=pemmDp%<aUEUx|XmaUc3wDxKZNUj|
zl}Fg^^_sKn`3^>sKl11vT-MlLk4Ety<0;PWpOUYJyZm9k$9MFr3&Xz{@85|a9DME;
zj>c)uy1Y>wotrVK@;}Jl^%s%@lafjYRi!nbZo@U>2v=U#lI>c9yHYRpCLx~?du7gT
z9cK7kBYlf6y58hys_eNf{KT97IJ6Uxnyq$QA5oE+<&I`)^igrH^HGcs?<$hu+#W2I
znAJkf(2(6I-mX(wS9>U>1uiyYWu#`+=+;IeWb%pAZw2<F6A;&95|Y{X<fCJ)8<W+a
z@>>J75sW^_pX4)-Ssm9*OB;ihifB2d&aW{mH||Cc9wPqpTKbl#%BK?RiQ4AVRa`5Z
z2ack^p%<yH+nog$)BbzRQX99si1oNt<9Z4li4I-L-TzKJuxJ#hgZB@`As-EgXBqHo
z?ET$|O<LuyQusvs@yoeU-;-8O56^vdXp}O}XE>7H05>n?ejxbpX!%9FkT{UYRqGDA
zyH)&bV{3Uc`L<cP>uGA=n|=j8+1P~rYS)6liNp}8oO)A>*W|cHen>JG@LOcP@*e3m
z6o)pWA-F|*@<Ih3Z_GI%`dlDtEN03IL4eoiLHr-pt6(Brz@Ln94G#>Xq#Je_i6x~l
zL;Zr#9liC;1wd^mW{M4uTQn?x0aI%^S>PI`vXqYaiW#EEdZsfH29-4@8WzZFnAul4
zY*xyW&eukyz?Wu_X{?8a)VL^*Xi}yEBz=HvtE3!8-<U0>tf6Twr6igwqpCTeFR(E5
z63`MaWN!<&yJ?(QvpDGTf#the!4lDE{b-UFrr|X$?$&=OuN9=~5eP(mTD@ikRKT1?
zCeZO7nLQRCjHPj4S~P5AB-rzTB`?@Hw?&5deAMM4p~7xvkSPG^H5Q-4#v)Tng=1jJ
z=-=3kp5nuAs4G8A=9C1-8+}%%@5y_CJ^sjSPULCN@2F-}!B~ui4T8X_tl9GTA)5m?
zAK3ogU4WV3!y)C2Vjhok4Og^NQP!MIouXfjCB0yVysltr2y`bE6VVpzgT*COL^KU9
zTBZ6en}e7E*uI95ctc0jV<zZGgC#|kBJeQ^=om+Zqm>(pX~ZB1ng-xqN?IaUTl!e6
zg$5hpAsO}r_-;mq+WushzybMQ2aE`}1}w;kS}NcMc^k6?<ar%v>CoYWi#Veql3=|Y
zEODHL6@2c&2ad-p;msQ@c@sRscFkYF1xaJKpLE1?QtM01P=Fvb;XMkUn48O?ye4C&
zsI=GBPw1ZBQGjggEa3LZa(UG0c?+<e2w2WUwxc2|C~@b^#$tkb7c3r80W=zpUP0{=
z|0$&20rhmd3V9q{$O+!B*E1;zCdyzs2Q&ful1T!R<%OQY>H1_m={dh<+sH4Dq4s8Y
zTm+JNB(9MoYTODCb;HiSrop{uh6IQ~B~&%#%n;Kz;CREd<*Xb*nEnrYVoO8tc`-7^
zR*Z{`C`croXW*;@Xnta>OXy^1^jA9mKFNanlV<xdo2}_m*O02B@9PfYNtA7Q6;!x#
z5$M)+;42Iz8x{vT1JFDe3>0M(g)UzQqVOI;XmBZ(<{#T15g+QRqatBAJ|1d?tBkDS
zMuwCQtkywaJEluSnotE3aTa&j2!$~TAQ~)bMYai<<=#Z{OgoT6<e`AMyGDL79nOdh
zE{3`n)g$@G;<(P#m<j4(vG?1NUgYR<nwnB(=n)bQTLG}6GPu9QvDCElLq`<ML`Oc5
z&pw|S^2T))azGlPz^I@K_yhLhWQ@h_=-{lpE|cCvR%kR9t^Z06&Q{nT6i<=W2*~{i
zJgPYP7kw{>U%RK_Q5P@OgftwFZ#1WIpm`BeA_{r$j%y)NWh_^q_Fu^CkvJ|pHU|TE
zc0%*GC^w-E#uZQD*fbC}=mhtis*T3wn6Np77=TfJz?PxNDOSIT3irx5EDEt!%;3Ps
zjzHdl9<RSbwN)_Ta+#qIy5dyw&>?oeTlO^^Pzg^PV>KBFtJi@^4QDmvx4&w0z}YYi
zLly_D0T=+^!y^k|Wx;X%bDiu@%!xo%>j?TrUz%vuXl$@L4WSNOadq$_!9LvPjfos*
zcTk-_(s?x2;*RU?R*2>-<8Gnhu(j4JroH&ayKDqNLf=MCAg0{d11;zb!^kv#Eku(Y
zb()_5<KW~;+|+$?8((z~QT1QSf{|&19q-dXm;g!SCV`r*MM0Wlu_RU@0?PNV8oYT~
z=VfhA{0+vTRV{8n`oW<xhnaS(j(JIdIT`P892#DL7!`2wI-nM3t3%^pPs(I)d313#
zHYYg|-CHm@L03~O0wrGux-Jsrz4nw*GdI#L;Fd>{F&v}GlbHVLYR&HhCi){AF7WIJ
zW0k==H_XXR7KZ|LtX$HHUMNrj=LV@RmnNgp7Osnn@Fl8QO|=26#~%(GUrhdMIwIUX
z<M|88|DS;*1)WS(G|F}&4sF5WU}^v|`6DBf^X$6lK}mn4RA-Q1J#t_qp47w-nZr+F
zy(NCC>z^g^LnFC>xEr>$L&NF0zBQ1zhS4vh!le%()|sE(>|zvk{Y|Z?@Eyr~f6+l0
z3PLWz1UG#S?73R!Q=BhcDqO+i;Qm5paQcb=XL6iCTcM*ycqH||>OJ{7CVY+s47}l_
z2vzVxtn)zPPmF#QUCj%As0=nJDqF*vJ|dkTn$88<zas<dU4C<*8b;zsGF93z-Egj}
zThoZwBePj~tUe=cC*#n40$|_=C7JxGsxNGWNvAu9FwBKNv@9&&huyKz9cM|yPGD65
z9|s_}KV@lO2XF!ud|(9yqYRcsa3SdMx?)H{x~F34cT^gjs$*TXzS9xK!-bNl0NFK+
zb!?*7cr5w`ivymP50p$$V2A7GOYQ+oU;^_+70^8xjDwp6M&r3&Q`XcD1>K>LdC38V
zcJ$w+(^M7@^N60hNkh0xN|H(kQoS`O9T-%=mfdIT8XE6u<oq$#(&~)mdWK*~<Zyhj
z1`7e6p#_=64+XabAR~31??2K268C_akGfB-fz<~X#=LT>nx-Lq2OEJo91G}xd2qmm
z^-(0B9g9PovN$}D1OpO@VL_8<qnC^44D8eqkOpr@N0%wsDUTW$jRo7WlHe%ZT-bR+
zNk@d^yI?rF<zg*@AjvPd;Y5gSgz>=x8!Cet>LjK<mO>D|ev0c2^7x~`GBZYpE9HPH
zDWIii7HAsHlM8U}EHc}C7~goI{Q()O>?2H7mqy})^WxAW2!faj=yrfKgj3efjdyX>
zggjEIAp;?t5A=QrG(Bt60u89Z_VOY)6S%uu@eQ!cfiD+P;UvANi^yyIW#GKPOAgCx
z2xqOX%z6BY$TWsan&A?|v1kn%f{3z;KY--85elT%U6BOKn|w}|L*TBN^t43u1RKGu
zlyW%VLxKd1nm;I^A>KCt8Q}P?VFl#TeyD;A-SOse6gB5Y;?Q>(9E5Gaby!Lb9K8dl
z0VqpBUn2?Mx3yp)PzM_@IDC=@I0yKLjF~g?)@R6Tgy7%U!pBC}fgmh3Y-l*67G&V;
z1j2q|JOZC4k~R_kpQ&*6_fP|HJZ7G!rH52?Xb9TYx#!M1Xsdvu#~)i<40N~g;9#Z)
zxQK}(Vei6$@N56SQXZT+tpsN=1PNYbSQUUCbns~yf$m>znCPL1|9Oakph+(1KC=@`
zYT<|aV&a1@HK_9zQ4v#JGB_ZKE=u~EB7V4FSz^J2yUK!lA__H`SOI>#@QSDG5(A|h
z7%oAk$^OVp%1UMkyYDCinf}S(hhB#QL3@j&&-?KnO5!N;;<*`krSd1~0i*rgJjP(Z
z;&1Z*sgC;^AT=xSc|DWYq0mht-wx1iB*XSK8ZkI@+4Qo2OYn`>XaYNAXM#jItKum_
zG$nS0N3IFBJ^FJ7r{G;aGE_v$**KdeUX%F$)7X2*Q~iej<BpLW>t%1wLH0;wk8?P(
zLiUc3k&$DUb?jFvIx2f)M6&nRaLkaAO>&ZwkO-OIm-pxQef++U?_b|PJRXU2-}iOh
z_jO&*Yx-d$xAp2#L(=LT7xc(tTP)Q#QT7YG@d#yP+A;Y*n8^|NifU{fWbZ?isKv*S
z;JKZyCaoSjB6ACs-a`9kBwF867#Gdr0n{F@0)&jj7S4U}lcG*|A!2GSjwnUg3aB;Y
z(*d0oeL(4nmx?soU2Fv7ZQ1}jq7$nW0NlNn8hBz`EbhQK{Zi?Us(ryn^^bb#RgTPc
zk{$waUY$582iES}hiUX7f}K_^Q@vW`^9?(JtedZ&NN>T^7aptxifYm{cc@2FHF;J)
zuw5PzbP5YW_s={nc>KR>0h)-S$Bf<+Fqa?WEHd`n4&g0tzO-$1+Bq<3)EOqyx`c&j
z_gnc=^KCk|VAy&(k6xGx=HFd$O6e>=RnOnxJl5Xd)FjW}C@Osa`SC{I`v+k<;}%{M
zDs4a|vyTp8@8MeNE*N@+jGB!)1>BP(`6ECCIt2>VtPOX08Jexokvt65P<?hNror?J
zqANL!+zT}n$PF<V;rsS$7Z$M5aTO(S8M71QN=+mZGPb;!A0pEfamP4CTQ>d(4LG>R
zj&yFx$~L)J_&gA~m!Wm;I^NXUWk*SzG5jGT9`rF1@j2cT9`U+HN_?Ee{DB{)yPATO
zGr|<aXvUSv9}^E^ty=($%0J55Ept{Wa{T556YX~Tu7rIBP9y;e2SiIjv4w|7aQ-#_
z8^q09%fu^$X=Q$1=fvr?aV8S%D=Ccf9a9wIbNG8i$tf)bqjN8l$P-{EGByxV_1?u$
zT@Dkr`!^%LUzuMVF)n!=XYE3r)DA@Tzmum$7aOY~S<jdu6X8g|4fK(-pEKrSVXh{C
zUjjaCpd(Q^zd=CbodbgpSWV-)oJS6>%=IPA1jsuiu>b?2-(4=q{Ctpb;H7;2T0UNS
zX@CUj>G6>TDJ+EK>amwI7i*W8q1dp%IU#xHgrp!ryY1saV+n$GE`(lbgdr(d{l~h7
z3TLB##ODJfu4xt*2~m&}D|O<nPqdiR5hcp8EsG~ID)YPBi5?$y80rD2XwM(qtww_O
za5P~6Ry%_(Cd2aN1&t*rPEkA0rr<Sz4U?zGF)eSHBWcirqT~$1XNWFtYw+oX8ov|<
zf|)0jQlM+yt3~?n%q1XfMc)F@^>Ii^p#|BN>1k#7)3l2DY!(s@eg=SvGoB0%NjJgP
zH($GRjWyG`fwO6HZDCAU$z3JPp(iCd(MKBOv($mSBOh%U3asZ=fAwN!YcWjbn@wa$
ziknOJA*IZTw8#N)Wb6iIn20b-HlhJKMwhLKjo3AQu#HC;g_it&oF&~JFb|Ebkk@qW
z%;jYR;{|>e*(oiyZZOA)#Cm)j{VGf-Onkxz^xZ>C)U*8~8{^GF0)X(o`*H(vwbh=_
zLGs6lPbol{A;Hr$v-<$uvGiD@p5p-hj-Sx~J3-0wg!reLD{hB-)WpFO5poll%%e^7
zFUZ19rjn?yBSS9}6AsKQ&%M4|?##MivAiysbJdm!fJmKtk(j{FVGV}xvN%urh$J9<
z1qO2GjW_|b<K$!90k;a!CrEh1IPr3Z3g4Aaf^L99WOY7e(WF$^yBx;{x<Gr#*5@ld
z6PpXcN%ufvOFlYA(TNet88R5$_W@njOfMF?$L~em!wbeEE|3y`lW0L4fa=`z;4$P_
zeGS&l0))EAm;Z?Kh__ymU7;jw!xB7F7#JnMYhw$Y&I%#EDimChQ)JWP5#LhC0ZH9A
z0K*J`2d3X-L)@g)<OuooHHO+Cb=wP%xPHSL3C*E<=XZEmdrca1MLv8;tD-KKK9M*k
z5}hzJrGI|VMnJF8wIW{kXCEZ9kKbx`aA^S=QU<|!>JgzJpWXj3;OSZPF>dA_(gBVx
z{`l+N9a8eKJR|e|rdFXO3#h9ujM$H_zS9C->nbvdZ1e&<ap}Pi(;-c6t@Jz+(Cnzr
zZw?4!)4!=O>;zGuRsc|Ty@xVG1TQ`ZP=J&M|2NnlCDF*&V+h)8K0Y`-D9RH^0#a_f
zW%phJKV%m<p(Ku#AX?+zYOY);r(z{6>|08pU!Ff1Z_wSe*6$QOv`q#_?=m@m;|W(}
zKGR?|8f_@$p1pVYi4Tm;&Jk5FEKq6AC8fkhS}{KcT9rOH1H|VLt!or`nt_v1V5^4*
z>`Q(@_H+ayLXXV3PGjsTHQtDY1p!GPEXKo_<&x{e2!Pu1{z<BebbcYNb2<ftUQM`h
zQ%wA=CQM$)RFwoIc(|YShRdPhFY7s#qCk2h%9xH|*my%uGiNF>J`B7@Pp@8?I|{ui
zdfM$!p(am>xC8pI`bbS?>GzGu<Pr$=#M2d_G?|6#?z@7|gA3VXS}H^K5<W;sv$fIj
zMV_Aidhj7`BZhX{Tc34A1DuV*M54Im@bT2-1UxLEzip9%oFA%8bObcSQuz3&oMT_^
zO=G<uZ~J+uW&IV-m-Pd_2WcU}=;(twUcwP9A>)f9tJEm)uOaGl;fY5Nw!uZ?jU}1$
zcz_7G3<9`I!C9!LmGVR~SnK%DSL&}(;Ts_hQRS^@a=qy)?UwPE<?knog&!x1@6I-J
ztd0MDsM)mkaoNv%B3V}L{Y&-1#3LmNhc=;v>`B-g$XtHMv<2GGq;F8GFRaD!hV5Tt
zLfS9YN-VG-9g{!)^ZOnhsX8rvarW~aaJn{JE8$+8cufj@2<-H4XUK=2I=}y2qyK%;
zDqJ{!LoCM~AMd?o#rFR3x>kA>xIB1wrCwX`;BkEWi2J;$<wb>B;raFt3{{wVz3*P1
zn<HDHfS^v=-9<ssE5s6Uhb}2q!~27wE&BNp`FJVqn~ZhCzaCB38s=wzW00DJN|qix
z`!BNh%Sf~3X7%Vxapx9q$DQR7q1}u7*JP3QZ?F;zSD`~o|MNQ|ISJ*54xJ_^&+A(Q
zS@BW>s~$fJjpc5K3vFf!gibpX$BPQVi23$%%`Rg5KR<@w^bXAXzCN_bg@Z2MFsoaT
zy?2XQTKnwK^3HwUmKA~m!hc<t37rCKS&eH@tzvm3<^w$sG@wKMmr|j%Tz-GQ{Bzb}
zhwl@YbjBo-<9^x5Jyl5v9vd1a{;(Q5HZoYz8S^&I`!15D#|76&a<lDHN^-kmmvC)N
zVuWwSM(%FKV+dp=wLjt0-q~fdR(zAR`bPMXBM=nzuHyOoIYoMzc@y=8H&j1T^MRf@
zgb!GB9^8H>ePbc9rJXdAh8?pc3caFWok-NN8U;05A~G3%88fvZg#cPjo@xRc%;E?-
zI`?H~B*9M&q=$T|Wev!{$7QUjLW5o|66T{5{*$IKeuTelM4uqW_qE6ygp^ObGa06h
zJsmj`niv0&8fYIr&$AG9`)j#~jqEYHy$LDzK$hSl03&`+Q?R+}D#GoA3Q}y!?wMr#
zo))Y-K+nHe2yWh>q+ldgB^_+bBp&Gavu<MJtiKpA$GmzD$-WUib2xa`Q)&Ro+j<`)
zZ^Z`D(U8pOA+LFALK&D;&Fi}g)ND{&(<yaN1bDak+So;G11~`ZI`YR+0oVWU$hSc0
zgnB<2I6j$yG3Lj0ECs)UmNA5hcsZQ^r97b{<Uao`Fdyi4msLs%^-Db+iXSCu`D<Ju
zNN5zb(BmJ`@<AVxq1yiFDv=|S*idj$p0Ius&{LVuNECr?hzhIauuBegx-zLH0BaJ-
zk(xtiEJ^C%52{$*y_8Ntg?Ik;F)hWk<1r$iUqLV+Mv8*GaB;&3HkMbQ)4z)Ae=Hyz
zubZss1|hAq<JeFnuS1dv=(3CWOpD&VZY+k8H**`(ERLOo(o5g=5ImrY*jJsVwThb(
z_Bjb6$#E6b`0N6mxQu`7kivZeV0PnBzhn}n;t}KSDKsWIO_#{W=FH=C_qGa<c?N^;
zdoF>+xjtVIsd{jT!x3t`2U;tujZg%ySjHGanUk#lU@}A#<mwS8q+yGL9&$)oY6@1y
z=9z=LUrkWlBTW~{G41oY+z<~_o(#kATsl(ZV@9=5E%&-D8_`+~+;`?6)bR=Mfb->F
zbecq4*)Mm;Bj!XWu<8pvm)n?CkmKjCF8AE<X!);et_WSDx=5gsu)oM}m_mgyUG~-b
zrIZc9u{pC59chBXJd&3<+Cl{p@@*iYf7pvFMmJww?A$bdK~#mo)sle$U1k|Ne1jOZ
zJOPTQ7E?n>kdCEuYdQ=)fgL7(j`40N!ZBybqLF||WtIQ~DM?AUyM!Pr98d^<0k5Rx
z!C%^uD8Z#i9OtBP7{7q@`TPwJt?&m@qKNT#cc9a*$;5^Z<oT-$A(N@m#iG3k|9Ztk
zN)v7_^%6wawgem<eg%<wGvPdL`XiOOP;H_aVvvkJ3X$x-Ok~7hPhS8SuL(sO?2HR7
z)+eHv!Xzhg;n;1+FQ%k&^0&)xWa>R9O{fp1SUT+jL3N3r2Xt+El$iJ*3DjsB1jf|v
zB}}GUk2wOkOU_CpGlaq<?Kl)vE=)s?@h^Wz!xrOFN}|QgwKF3d0m*Zb0UQkcR19F;
z?|Xm?63JL9B5_ZI6PZj8o&EqAn4iRnB-7sg=*0zq>3mQzdxInroGLK|JKe`$LhG&;
zJ<%%8*Sc89>0EXK`M!yh+vkVWkLjZ&DzZw^X+hV<P*-`6^B)<0Vdn~T!*#{f5SD~|
zNa-Gcu&`9nN5{YB=$Sw*gz+7zgFFn*@IG6xIOM06)}uc$x<aC$Jh;%Gg#?4)bcAUY
zga+a>jU46>KY)bdhFU;tF%|yPk_O}-D38s+OuRpP7%Nrd6rVxQT3v#WjTdxy!@8!Q
zP6*nLRN$=|b?-ABeYwsS@68lOwcKW0VvyeBrx*FH`y_2tVBwkXf4<+A^?$zKB7JyE
z0B}nZ%KWT0_ej5!=TRLOu<zYE@MHTzy4X&w*`zJ`+n~l!hrZaris09({<T~Ct<qB1
zzNO1?zkEci!hE~9X4mo2KR-OxNDU85U1bH^!1wG=-7Z^6Ta527X`uge`h0vWZf&`#
z{%@zxY>lw2p4!WR-+dJNLx8v=`y=%$C^C{(5QbSPfjxm7J@W6MA~GQvBp#!W2_l;E
zivrFA#!p5Vum-fp?8k*l!CKe%#TgOmHEEgvA1D7B);HlPTCgZPdpU**pIt8@g$}P2
zU_>O}UM^&T(D6S%2n}~Fkw9vo6w=7Qb>egn*N<TM&1Yo%nZ<}Xguhlt0%A^&iT~ej
z-VHGL?4vad8(?_M?hdv88j}?&>P4$Pboqr7X{zshNsYaD?q?9aInFgokOKJ_98$XD
zDUNj85Sq;@WPlc#G4MCj)<bq(&V}?FzXC1pqRfUUR7)%jhXf(%$*?i#nh_xjNx8wP
zrmNY=!)4YW&2wDS@`9TB{PE7v%6w;FCI`#&(W)3<a2j$ud+0zH$lf7RqXl?*`MJ8p
zjnE6`)Ed1UVDX>z0_0pkni3+_cE%%Ckv|wMSbuv4*f`A}qc+L;wW}8^Vq^-$fKR0`
z*rrFqbo>|kZl-1z2wy|a#ol}fOypdPjuY8aZK<T*)^1}i0$LxK#CfL(6GYAsdzshL
zNlCH&0-E7}l~&j=Z6b;6r`D+JL6DPzwx7w2r$#qCy|`_Akspr6)WZ0){naf|=V+Fc
zu+PcQ%daq1;|M5B^`>Ai@ULw%#fZfBCu}A006b4~)yII1vO;H8q6h1UH0Ual>k;d*
ze?|kXH%plcSqiO`D4=KdiYhWy8VWG}TzW!>`Li|)==4fK>VSs!`~r@{nKhfB^8Q$1
zcp;Ro;UYS<F{dL_Q^sk8V|C$dvi9~+!=so3oCXs%y3QUKDoj46%lGMB3OezJqK<05
z>%+gMkTLzpu<=i1<vdsmd0lCm+*gx)km6ZeC~{gh0QCMOsC1s^f;8BV#{X12@vn?B
zf1a`KI)c<6L!(GAoPt{MzmEq}D-pIl6_p5CfqF1}@XVv<5M#S^x(5)c+2(F7FEq((
zdfA?6{ZfdRqPRd*RQ7yI>J0PJgVo-<-xlBDF%mcq7minyp?v-Vmq9qJhZ9Surps^#
zMU4OgT<-Kxv>EGo#UvRNLp#<s$y)>@K988QHAM}%ONt`o4RxBIpgX9YSlZwqYW)%m
zu!fj}gW8OCS}bCS%|=#Ia;$Y2rxCmsRWf1Og3xL(!n?3Ko7f(TLiN#U%mCK}Rf9P|
zX|v1>HDp7;Wj1Q1yMR1U@v;)ZhQxB8fr79$RSU(?ECh-v#$-+cvI_14unGZe4bt|F
zB_-yHu+}bG!oJ%!Rr3+f{+u0J%0oqPB1$}ml<3z+1_!5Z4j@OS;x9U8Q2U>2I)Bh2
z@8vQP>S~!uEXK_QwKfXIhd!X?c!jgBX=_};>E5}e1kP4oqQr~`KZu>dCFd{#;8R?(
zL1~OfJYFk@+NrTWRqOhN{48o%%*8e!s^)!<2qqwnlb6WhWu`=MrmeUnjJd^MZh(?@
zWC3=p8X+p{#~0JIDvam%Un6V-wiFSmSPdU=0=HLVg>LwHcLn%f`w`NvIsJn;k6ghE
zaAy@xDPL5Bkp&AtJ>8tf4KOJp+zP|^2ayV=N+4CaHDpyX*L)3-$wqp2Gs_XSvCV_X
z5%M{V9mUvdtRG5H_Vp~lMq+JoUkq#?j(>}~z7B)X(YfBF{Q|N`^#f6j^3OGYIerm^
zo&lbYqa1hfdF3Sv!lS_EZ$R`&5+?u#{OD<9EVTHFNNl;mMU)Rf!)p0^3Swr`5*~5^
zm9MW-5IX)=T%y3#?4jtkFTHYw9)pmo$3_z=|KBZ~2p#t=R70;;s#cD%bk%d9l9N)2
zpk49F1ij!bC4zXLK9x#V=rZLH0a{*-x`8~U^;>|j(E(h&@;hQx?3QF1!q(Xw5I?80
zW@gSuhMOY~m#9k_m8#iMN`+;3R;R8*N$}Qzlq=GDUvmX`Tz&YA1wwK^*;nFb*dGAX
zHU?s#kB1@!VXMLqop4a!SjO!v)W&2ArbkCAh=Je!yWg*gi3eGY^Kh^MT?H?k?6Ic9
z%Cs*Q$9OD$eX3;-&8XzGS%7UjFc4L(eI(V~fE3&8K_n@Uwl&4rq3;UNU62F49d@dk
z(K#Oq%-l~Mim^Kql8FbbxU0OWaH7I$3^$SMm#RITjFR=^ujar~W12Z&H{r*R@lWm1
zI-Y{F6#-lCu11Dz-7aC<(g&wPbs^yX=uY}Z>P<NI(z9CV*S}{%gzO6Zq37qInoT&M
zYu^%0%m8_vz5%q;Z>-N_MHtHq7%w3O7Q(kFD448F53Xy2)qQl}D-tye3(W@QPf$P&
zb|?yI=vxZj;8xw?H5S6`YAR#lUs<SsDC@!?vT8IgBQ61=xDcZFy$;2;kCz@rEemZa
z&Vy$<mbw-=QR_$2jmC!ZIF+#&!Le7pEQ?}19)A!?R7}`_pj#DLO<PJoQi!ZPIvp}<
zCc_CW&nB<GLfCFV6Tg-kpy$g|zgZHSCI^Y+v?ZjDBuE`?q5z~3;VL^RraTCmG)IfO
zWT_}hXhaD=g(RX)1d<RX4sy^k3`U8;k~5JNUecl4wlV}ocM+VIPN*Q^E}Wdwq`)^J
z!YdH0u^%fmpzRzRnTe7(g$k(cZ#{bC+A2;rLFpUPNIv};Q07H}@$yD15zH#|Ag639
z|G-a#g;qGR#h#03uo?sB?=-HCnTbg-Lp9K8zl*mfO|yx5qG?aK^ZF&QMcUyDg~b*w
z1}b_%%{SMysPJuR=#FZtZ3RoJONkj5DoRTILGc-u9Bya${fG0g?I=Tl9;@Cyx<gM?
z_D_-c0c^bwx)owD=90`!RJjlEk#yWSUyw9VRS;UxIGU0FoYw<*EmbIsp36%(gK^iN
z2YLR*(twr<mMnyrsZ_Xg04G$mAv3CS6*?cCkWIEeQp!C(A3_Lyg78?1bM{rDRG{1D
zEX|q9G1Uhlb{9^P8PRP^cojpBCK|foa4W9DAa@fAnPy1Q{m5{+(kYKA=Wz8yTE<yb
z7`_Co>fRgE$-pWPo@y<?8nklZg+a)&xX?xTcQy#<cR+Ix48hD@L|5M?q*N`7AAeeb
zv`lU!ilUZ-78?60ZJf~{`18L=OCvpf?Q32_$W^3J2}=|m=z3k4i<ojRlbZswt~n0F
z%f&MyY;$YPfl9YMe=w@<NXsA0^n{XNSBkP|u~*+3pVM=?Yt#aXP=)Ok1h8~L(K~*J
znwX{Kv09xtRb&(O$BU2GnfAWY7o6_l-W;&xV#VeU4!q5pv7uv#=ve`QfoE9~3mfE_
z3E2kzpv%`W-|I%;`y0w_z{gpiqnAK^%%t26I_zU1!p3e_3O7;TQnD;&<{5<Mj(;&A
zZ*~Rz{q{jTxrpazVB%<Bw=H5PPu!vPoklxV`s2R!U-x`6oP6%<&D?p9>PYsvgX(aT
za$tGC_;9}&kh*kzo5R<X@9T1M=Ng;en0<>-g28Dvuu1$;u=<`yX5m+rX3H4Watw=<
zX@B}EoBDzeuFA3niUC=vd<pV4EAmj^khM?ai|_of$*?&*)!XgRREEW<I>$c*{(L*N
zz1YliKZ`o#Ja(ecyFOF0#dWE(HPC^^>h}`>3S<^6Z~mFJ)QIPsEpa5_Zg}tT=DR+_
zR0+K*vT*+E`@3(>XQVz?ZTBCOb!d63aYP;2w?4CDdpUKLsfqJn%<oSvcTT&lnjUT&
z!PiZ@t@D?A_W*@?>e7f!?>at$VdCF?h&G*Di=za;!9k%_gA>X84M!ZEdiSumWQEFw
z-<=;jKkR1Kcm}8V?POD}r@_L125k9nvdJ%!eEdhVf1m7D?($&`^h<9HX5@HOH)I#a
z=iaxGk*I`k=GLsyLLCLmdnaTc9t#lFBP-iH<P#Tb=-+>gvoTa((cZZb>>wtc6_#FH
zn)9ddpPh0~MT5sRj}D7QA?d7cgL*$_8?j{neUca1EJn6+XD%UE7`ku&vzcjMkjv@t
z_&jhp0EKk=j?wNPY%X;cdE9*P(7N5@bDz|E9_@OU<N=z|xCExb>Biy-%K2TzQ9I64
zmf^dnU|Q6D4}q+sC*MIM@G+AodZAw`fCuO)JNad^#|*qFrS*VN&+#KVu=s8@?e;Zt
zYDk!e_b(+LR0|P0<Z`%)3{z@)bok+GLWu_!;SS(qPel$wz%9RiNUR8Nbc{2C&Q4=&
z#N*Fp=H8X^LC6f*h2PIC$LaR{GexOthB=EN#$l)5poB%71yG6H<pBl`q~!_o7S_T{
zqJ-n~EgIt*tUywJo-dFq(A-g0IrjpG!e|sn34&b{-C8-OpYM==j{_-YfGwYr^j0|7
z?Ez&sI22=ILoZa&&%<P9^PLryFQ>!#d5}hM8L?<y%<gR{u}$$8QFu0PYq-G-I5zeU
zJ=4O3)Z4NxL$dmUfmS%)gU%ubLWOAp@gCF7kOpqnWCl#a)j*~-b|mEY#wm(o8kj4p
zZDC^zbY%!uUl*vYS<vhV3m9T`sql0|FL567q?%%2xA;*)O9G7w!N+#t=QN);Xbx#g
z9$T|gs6uVB#f%C+h}Q~_ryKslcWw;=@3sgL8XBHHV3*pFK`v05T}7@o-H#VRq?%b2
zCuHO%Gw?$v`yA&ijtMxpc@=YomYs4t8Zso;u7*Wk(?F)mkhc0g{V_`8ta#<l&u%T(
zB2#5V>Liqtir-->beUt_NmKAg+edFb@jwM$-(Z_Y)M;hF3$MO@e9qt{7#>GN3e_AA
zh$4(i)QBvAO^Y>^Ov)tuCJZaX)n(AX8j}$Vogq<I2(Ae^42SN#{F#%?J*osQ;wIg6
zEc;6+?SD0?1~NKW$d!%{##o|UAxrM8;CUo*okE6ZBUm`$L*PQlxi<89$OkOBy2M?a
zwFd&m3pk*}o}CofAn{xa(=Z5mil1~LTgacAWCp{J1XaZJpJUPcrlm0%(?9c|2{OsW
zy0rM4M~VpL5}#T^8I%wvz}f7<3JD8zgNPy$?$RwFH+M&>C`Q-4*#uQwFAu?le?N|K
zy5C+QN=g?uNy%jLC&dY{F&UD|*>IXxMApuefb4Oz5fZdS<I?$30ufp{U*dGjA8CUo
zMxaZ@>J>3CP>PZQqkJ_E&L1F|s4Gor4#<xBhWLYW_Be#p?Guf7Z-|6qE0q0?94(sy
zV*IHxB_{moG(G52Nr!?yelda)LwriDqqAa#cCB)`L^!sr1}F$2*3Bx&Y-)bVwUbzn
zJZ1mvw5BX^at?l#GBZ^1z7#XiXA@ib8X;7_i5EwV6nEw_1DRe*nEMVtak|$uSCMYT
zi<>Kery>i1^uP7KL3Qj;hAHs8PAtImROvF2oG_LJcpB!C5<`~&y&gGwnbye*wlKVr
z5jO&KedNr1`9Bh>oD5%|ldwy|g#d3?MBjl%uoz-PW+&rG`76uJ&<%4wWr&iRITVyj
z{nup<%);a^1F~bMkcbf{=s}9gn~-pbLp>Ph!~|{Nez2#`-$|-Wu-<bKw|@ksR^s}r
zq)5r|P4lMc>z<GAR^fEN`$M4qk~}@0$4Og>_?98qFc<FZHn(9!hfjb+O8p)fU(tI&
z2PR`RhVyozJ8z5ZKqOC|-%BnKuLu7XYzOg`(F(d2rv^B`%G=)CV7aw81u|RdM7CKJ
zf*&niybvC;_DwnW4%Tj0UBq@ULDVoI)%`S!H=4|WwP*2=hT}}2);WC4e-#sqd#+V%
z(JKbpc$^t4VK|K_@PQFJ`ruwAa84HOcRtJxYv8Zg|8(W{9Sy|z`4Jilo|E$XG(hY$
zs6d-r0G<yuD9%4y3r?~-!p5qiS?gL6<9={(b+qOV_)VkkJp~(odt2esivbc6UeWWv
z5pV+qZ3{v6#WXm-kL3%TE|+9EqApckN~=NqG|syX;2|4@I**-(Uo%VqcOSH?pdmW0
zSfc)ih(&t}LVN?PfIazBzz$%eC!MH!w@?YexT%Vn+ck&b6U@*OcdqtYp&F{C;-EV|
z2T@A(#@_Lv-4ytntB@kjQiia;?*|zE*-RwLb)$(HT7|C?=vm9(2=j+1l<6WZj4%~;
z4M4m49FdXL5~alL{g8DZ7T*5P5j^WjDKJy^*I@YPyM^44I<rfL2)X~0598mh4LL;%
z##It55EtswD*VA!nq}+-=a#$R*Brp}c`Z}nWxeP{UjqvK=iwlfU^=EN>jU`({?6?3
z7~SkV=x}~wkm@8Yl+@{WGurrG4l8g3hV}V<C-hY+AvZ{Ih8u1fgt;y=S&?DF%}{D{
zb#x>EBTTrSpVdsmqC@D}SQ6xmhdUoi5A_x%72eCd8vOG7Y_An{ikHo`0|rutw;W95
z?luY4TJkX11aK%33^`7l-VdaJ&qJ)2W2mvt>w&IO&w=$UN@2)xzmf5%yh@qh&DT?2
z!weN(^l~=r28E1P)XttY4pHYD@<*Uho>1aeuP==0gW_vbvMW(6m??cU9k%{zb3Rky
z4rqP5odud2<T!A5gt=5K%dA)SZiN_Pq^7sp3_UbAQ=x0cJbYJP)-ItO;ntwMEtcdq
z*QHIhU89tPBPIS3+Io6sp&h{C<;|!)*1<CM-*=gHt1mMdohig2M35{ARNlkq0JTiK
z&SGB4Xa-`$cO~L=FXme;1pBg~@*oOAGiRpS_9QSY<psUDh=*|JO?|hZSER%w_^RtO
zF>tan?*I<Y-(vaaNKq@1@%V@ZXhzA4L$1K&_aXW#{eZB|SM}G3k|v=mrl_3Tw5dfg
z8B1A7mN2XyS%CZ%7Q%I>&(Ck5C2nmrK^8{23GO^g4sn5y;54Nvir8!@`D22Xpmdvs
z(m*|MNIr)-4TPaZ?Vv@GS?M4gtgn1$X8}C4{<<rM=sJ0(G!aL}zlMIRHALg~RRx-h
z<yoH5UWv=-<BTzbVQ-5D(4klemQHz&;)u<JgO$DKaOZoM=RZIOLP!+FXItM=j?CiI
zAJTjz7+d!c(EF;wLXex%ZGnRUZ-0fnj<Fuh>NA32J$(WoUx*;_-`U9nUWc&|I;Lg>
z;o#Ya+o#vffqC5QGdmbI#X_+RVOdksW~DiV8-nUFCJTx=pzAEP?%pUvj4wmz>AS70
zd^sYm*-gL<Wp7EV8*t$&rv3_K*jO@dmqGa{TcoC3SkWs)%sp>Yv^V8yJTx8r=6c49
z7;CDbEDZ`m<rBJ0s4UN|TnFYXBMS+dQDbjM2#R|>Zl<SA$^w-`xjGpSt;5r+RpFc)
z51BX1lw)K?;XT9Zm<*OX*~=@yn_9ONbMz3Un}HT3#$N@b-j^aNi#0?4a&q{ggiuac
zxpq`ptw~YKlri0^4|+b~Fruryz%0{E!NcEL`!i0F`vDLZJ8KKErR(Y%bb(sy64znd
zi*;_EFJm%jhX$RXAVlSw^Cw_9JLC@3c5@~M=2lRvQ4q+cM(f1D#bvEOZ%B#qx)07d
zbhY5ge;l&ex=FE3ahXf>0gx;Ex`1K|b4?6%sNkffAnfatv@-pS(=CZwl^>?NI^<W#
z{k<h$!<wn?6=Jp72xI}QU==-QDo`GzM3n%Az95#NFYD|EqU78czsYaVW)GY}mO>9c
z8myaqzGqNnnR6@+w(akx1#pC$mB+xaNK}x72}P}fIWj+<l$dp}+FDFTl!Hbn+`LRd
zC?Z+=ty3IhO=6nr_5luV!Q^`O@#a9g=9McSUXv2tNnA;sSO{G;_~Yuhj31xKiW6oo
z{>fS{<RD)9dUXG7o_4CzUktbsnFe>((0%j$Q^X~L6|8>@Ls853cB18WG$~Qp>rfR&
zyre+j#)+F&0pQt_5QF4!BT$1YEYXg}4>(KyLp7EYW=PBMTQY!!u<2w}P>2g<|8jfx
zOB*8*VI9R4m3IMC@Mi*sr9|CPeE^LN5ZA>rpYWUjl<M)D#a@Y-JjyXu4ndkZFsF{_
zb@~LJU$=6!IQZ5VdD6ptJH~sduq3En(+#N?AJ19?)`csioj(lr@HhZ$980u$xu>6B
zTCk-!&L?jcMI7C9Q0}3?*5L0hE?@ic^YO<!#Ome^_oH;42+61nGU5dsZ@S9o(8H;h
z8z(>7=GWTO0yKxH{PrdE$Cle_eR@g)k|O=Y*B0AV?12=nYR-=b-dksZTqoNPu=;=8
zS}}<x<}FOlXdF>7j1W~>x`@7OA9?F<tXKSLvcIOiPu$VT++sT&Sd;^Vkrvyxem)4|
z!%Wkvsyqflsf7vBu$!2zW=#`~0?NDi-s&hHou~!}Vw@uLR35GD>C(1%mV3?@)qWp&
z-`_C&^$xC}Wlmce3T?cGp_=h3JrNlkch(#t^jnSlgzsz>Xca^QR=R|UX%pK?xl0#)
z7Lr<QLqzEdf(Q|%BO+LgK11=Dx}>D7mBDL^zSt(H=ff9iVa51nW&h;k&}s5w=7W;~
z@s!oS_xFzOe{0Up{TS))_Pw)SwqB+E_o++1W7qW)E(&p7;3!>vlGJ{Lm|~Xr_$Fsl
z5+zpsZEbl0%TbJRr<oF-s(9+R|6Q&|hEPB!)tjQ*_|2qI#?_I`WL`ef^k5_@=*Ymq
zLwngrTxl%C>7QN^*5hf_a%0~P;u@1fPc@fgm0gF7mzh={nrA9rT#e^1lORdN$D_|N
z0sEF#oezuFCZx+IRiDGJ(K;jEvcC}kW{#e3@ukQ91Uh(lb48_a8T!sI`SM1#joUV5
z{pj+lscTAu7~5GYF4S_Og6vt%RtE3k;*0F}gKgDIsNaTjfZy+HH;#0MZOq&Yf3IHo
zd7pfWi00M}^yz5N)N;(yaAVC&6!iYFByLBtErL=DFfV$Q>uaD9*QOBVnAUpr(=WJL
zDA0!<uio{a^Zh108sFtPIxM&UUccgPr@c4ZgCIHe<N5I3s7HM-{yw(;`m$s`Q5F39
zVDl_Vipryb=|7IP4F?x&yyfs&+99*%hhd*<C?=AAqEekz&*OG*_gIMNB;l`oKd3cm
z4bp6qKG!MHD5LnoBZxat!&`jlEl}Z%rEWo^tj0f-{SylvE9IR;7`@OHy!wd5k5nOC
zm?9@8yicA5C1~Hg=I~4TdV3wWexz9x)6JiNUCFD9&X@c-Vj@4ePv@^ukp&c>?Y=|j
zV|!H?KuGGjyb}|4me~vfP3f=r$!ZHd2N5l(*P~@T$g&7gD}Nfw+P*=H&D8$Q4;|g4
zbz4$?F~eGHzV=41Tdid8{6^3TwN1k!y-2#3?oQoOu8>!O74z;NJFIUiqEO-@Uvo*$
zev(<(CBDrYj(mf+l%yw;TSu^~s^2fQNjLV`I+c3MG_Czy8n$oF0TtA8U-`W7EGP3;
zz&qJLvRU=b_W%W8@p@@*&wAv87#2a$<VGAc1QxK}M}E*&1*u80@Ee;Xw+%X0xDvDb
z8R$2&`n!5xUBEWTDhdAXM*NO;#%wtzr+L0yUMM}e9DgXL@-A%H;KfW&^o;>393@ta
z*3Lo$l(L7$6*;7N3*uU0efJNa7Hcqmsc(bjhf@UEgDQKqW-Oo!$mU_tGpzR@j4X~!
zZwu=0H!wFDsW#-n-O@h1{;>4ScN+$Sg39POg`Hm(XmAWY$l9wP^zTB=X0BfiR(3n&
z9_3i&wm0L5rPKFO0wKJFf`(4bj7w|0*NmS!WgUUAx6z?7J;V)plEOAkgqn$7bDTsd
zpb%|L{hCHl82c|<M_0vJl7_~b+Mc34TlbpNzR9Z7E7D(I4jv7ck>0AHtoGLlFkyXL
zbjT)RjgFalc)R*L`Nef~TCFNrOY-=wn~93$Xjb)(EsP+uDHrZ8ej{7_H-h)pd^fTy
zTD$HwEjjKY@p&CHm9c^tR@pA)=E~by3le-&z}v6%rZKLWH@Jl4DS|KjMtt5j72|Qj
zNBjG~Q#fv4o{)I3S5*7g`{skw`Lb6Ri4v;<jKNQd>vWg5Ztq1gae$wK$6g8OZEUXU
z*LQD9l%u(;`unwS{|3@}FFSdDL8+GY=jgKPM{Cqve@e~D6=xpXl8?c}zg+uxw1nTO
zBmK1NFtqukOlNCArc6~iJg$}b$)g0BB`kXM(c{VT1&5}U_iF*n6$o^AVvlL&Gc#e^
zTe_&-#-|pA^c#C=N%oRB_F?nQw_KD&?8|;9v|2mCj+DPVu)?yW_f17PH`uO+N12S@
zF0lAM$az^lW3rDFv?j$_@K+o*OSLHnAa0tJc4S91^4&qJUQP7%p~x$KLghG;^g2U@
zl4$957#&zY|He0<FIG`=_Ik_DxrC~NiTY1eq5ekcWgi23)&_{MXVRgx7@%^chH*RR
zdZd18Ec3;Mov*t$DR2gU#`K8qoaI`#ZX!P2p1_%JR@pr5q&-&_Vb)(g$MTLt$C{Kw
zjE!zfU1c5TPV*MWDORcKJ|^34xE+Avsk}~NLLqi=Vct>+coT5xIo>Odv8jS{SF+#+
z68A&ezz*<3+_Xr%HfMEIz>4Y6YZkNdD4UBX+YWvvR<R-8WWh@?X}Xv@ljQ#5(_-WF
zb1$abqlM)6GeQAAn?jS36gsCeMD)7;7+$IU>g_l6gIXNH1_s|Y6D(5W&PB0_Y01LI
z5Am7jg=ELtFI`0{M5HxWac^{JZ{6%;0WFf+OtI1O7@xn%Q;X{RRvUE;$R-OIp|-B!
z`6{}jiq6N&aU+o%4Ec4-(7fGmWwhFNz9PtMUW}gj2IcFSqcSDXAk+V7>X$0mdatFu
zzcqWS0`A=(sl(tp+iq;J>1%f>+oUFjyp}rMo(^-|`A8^~|DV7K3yknUa;5b<s>5c$
zOpFQj>jgZpSvZ<(dk}Y{z2adfkr{Ptlc!8a$d;<yxb<2&(~^-BkX_jf1(?ilUG4NK
zR|Q?{9#p-;Zh2bzQi8p)pflM68_w^((dlA0%5)A>HRN~f9o|jVm~(QO)O^3N&A~^v
zRcz?<L~rMwr}NzFs?OUSKBiHM3$Duff*&I;Y*odY%a0fT^J^JsxXJRvcD(6+Y)JE{
zRJrrrA-d&7wMxOq@`meDvQ?$#y<J*Wr@9A@e}+o@%T#Q8cuVehPHa8VuxM4Ocih}m
z&bLtlbFcrTPh^^x3JG}-&}`+fdx_cV#j^^%onh?DM5+fu@3;^ATBhpiGt}-qyEHRD
z_euW6{HY~%LuD&t+Ba*)KL-W+!PoPS+B@2ti|g>=KO;>O=cKPqhlV^b4T&vN3Hn+0
z+8`)BG^F!oow}glPPG1BvueG2Xvo%lchZbi*eIg0ovTp%jzVY%*3?EPOSy5g&0NTB
zj~cDn$LzLK_Mz<sxsWT>9>UCKRS{m;<C^Y9_oyT=d-eVeV_m7n?N^<0^R`hrYAcID
zX@ft^MTcMhV-^-EmDv2fxx%vTzallPDP=!X?YGT^I3yi-IV)kNL{)j2TQaE6F-vB`
z=5^THb~^`*kSm{xg=5x^hob%W#SK=$PZ8X-;|qU{tVo$|)WmoTG))VAvsigL|CaaF
zYc;6bDFW?A3ND6H@W?Quj}68DWEMzJ0L^+XFL9pe@Jac>LhFZq<wjq9pFl6+HC7l0
zYCAn)^IJv6wDe}PnG}%;asv3|4>PImUZCpo;zl;nT`k^R2&cR5#rC|raF8gp<y*Zw
z(=&mf1yi2mCjZmE$=c4Xeu`w-V3l6gcz&T$dBvS@YJ{^*)hn17<D1M5m3cpggSRha
zq{(;;gQT_c=O=KXQ$j8q5>X$}2HzJKD@hoO`s}Hl<8E`0-MPD0`i2zj#sxjH<sEm`
zCLd!uy%Cb0`J5SzIr>s?<>&kB<WM=^%-hY|4g(AexAZ<JHf}J&7y(hmh=sRLEyph<
z&GmN;w;!Q%51W(hET^up-O0Tt!gcmQ=N+7p(DbfAFjY^VpOfThH&3Y5(5G+vInQ3;
zLqu9>_wtbH$+JtX@O{>WSvuK<ICH_1o;xcub>FXFMcB?=Dh{0i!)D0_q0{wmvM6i0
zyl-6e=qaPXay}G!U*gfXdFvvg%HO#qp`b*J2`xcqW!@<zeE+H+RT_@njcxLUMz`m4
z@fdb8OMBhvQU4<COpWen>%n04S)JH41f?_U8mFN&nzG9P4d(mMFVCF38u{w#<%m<;
zc@=?pe2uz>^juSRA64+<k5_J?6C;G}pDvVpH{^fzzk_!+Ryd)9Iv0{;@VPlj3^WDo
z1|hfo@3iMLEzmxYa%pYrSi>e-ns1SsJ2gL-L!2%bE+{;V_uO5@etFk{DrR{a07wL^
z-!<g)#7>mXH6W^9xN~&48$=26KxT@QZl*(Q7_Hb8ozPH~n??yPwcgYUd%0$uSgazC
z$LSi$1()}^106Iy;W?)VImcS=NuFp`nl4{`7M<6l^pZE#;$0HiZ#2ESl)txJSB)nd
zBcEiH{x_jqulh-1y(DMwZ4@hh_)*Qgj2DwP6D<)%Ln?Cfz7pK>38IP9uL!!4N4FmQ
zpf$rAOKrM0lC|<TV8WKn$Yl=5>3BQqd7cIAiIYW7zF<LnE5kRhuR4Jxyk3FcaqKoB
zTxf~7&S+~g;#a-;rIN}%mq@M)Paj%dR114}Na?n{=~qH6$74-T7tFotjTF$$RH>xc
zw#@jj{zW$0{#8~Sqv7o4F1lXH*-aQfVs{ba#Eluv>-GORn%SkKPtD)!_B5E^Cwz2I
zw^eeq*feF;n?HSOv2A=ay20kF2J8C*%ZDVzY8HpDpw8_ugH21Cgu}euF65B;=KVt!
zHnYTU<wI8}aQRG~N<&#@(SziYwBU?Fc9$6f<V`w6K3{hEmUH?rB!uC9bL~PHrMvg9
z3ORacFgqP)e3IH&`O&#mx<#e<X)@b^`$f>B7`pVMe0}EaKm2{iwB~c#v=O%vC9BTk
z%`Ib|U+u^~$IIRtVel=te?INwzp2T1cbfdiwyPHPc+i8Moy$d+!p0t7z!28iUQ=o=
z?$+rH#1}*k=fiHE-ny>vpxyTJYAmbY0TK3Q>I>%Q+;|WIR6i`GU`Y9te9$skGIgWQ
zf_vFL@KZ+?6Et)=dG}O6Jn%jF4@FoE@&R*bE{Y1&?R(r#!0!zzj}M)27T)opO~-kt
zD-^Z!4gE9;Uid&Y%+)CIk+;8?mQ;3Yv()Iax8z$Ct{A@iSx%1bNwHLoCx6vN`)wPS
zAIrN^k`FITs{W|pk=3E|sY@Z*N**n~$u3w~G+3y)B)mv_Ci{uBP(|LZVxiiVe%5a@
zp{2XyY>#QKulnn6A)Oy_U-_j}oNuxI8PE>D@3))tO6=vbp|{z557$DlWn#+OCv{0M
zDdI2u<gbx|!_>PoKk9E|t|JcN7ks0gsj&fGgwF<g^t<yT-WPP~wyLK5K9XfxnAIi)
zv>z&HEkqHpTLuf0T+dbRl5qE!edEm$ZT`iTa`NGe%c|7Vfa!lvH73GXJQ<UzWEUCL
z=j`v&N~E=PJIZUcM#1?WZ_03}C(Tih1&zP%NAHR)?tCchVo)j2|HT`+^^$MT098D-
zA3MJ<v^AfYt^8(;MmDWWCMPY1gujc-h3VF)5Ki1K*H$y=$E;3j+g|!Wm8Y>xdw<py
zz9;aHLC0wcySLZ*!Eeje^raD$)x$Pa^ZxzXFFyH>IdF>3zZAFm%1_?BBudq-`{CKJ
z_P52yF#^&~e?Aqb0I~{crw0<fN>Thm>9bXyE@pHtZEK?^v}e6HzO4mtfcvv?NoN_`
zQ_Gxb(>D4GxtmK7N}p8{LX!i<!zHEfF`<tA-S;Uoi{Dn21o3IT7Gt`p|FlKL=*n}m
ze<5;s1Dzo?)h8z#nA`qW=!snq558=FVn38al$E8#MON*SUCMkJU>P~v@-x;e!@^Zd
zVc}485XF6TG9{e5>b}IOz#EW~*OnGmP;mKo->>rS!G9S?#S>EV*BA(q8&dYU5#kXq
zb_^#{<_{y5uK#mx7Yy<qNm^>Ht`7dWH;9$>ebIS~fk3{bld^68{e#En=gEoU_ajr&
z+r1<^#Xhks<X7DO7rxHeg1<*x%;nBLeebqCRulGpM<1GDYo2A=dj+s~TDoax<+^FQ
zVRJH@$_2wT80NJV(WT-8OzIv}GPK%+mwc+|SeO;gJoa^9Ko>tP^`+J^CnjgtLzYZN
zz#vog_PxKG-^%Z`x*$x`*UI*MNWqNCyq=Kn-q${FUYU@TP8yM4stq?a3EYn<eQHa5
z_A3iHyXOt$uWjbFY;Yicwtm33PLOSLYznUQqXYYMT@dj4gFpU#!$h;Df#9G|xi$+P
zxc*F7p{o@cCNdjVsI^H0+>=Ggp`NwKqCRE83Ss<wrRP7yLO+D@H<$m<uTm~>AoQWH
zKUn?Xt0?AX4g@E{znv3t{$1ePTv#C?_y6^St89c)ViP5!;w8#~K*xnNf!V(WE}Q<`
zLBi2ytDDkTyQ*K${DvH=iZ%r+J}l@NIoG>R_a(Q=s{AyntB&1tZ4++W<fa_E+<p4c
z4d-E9aG4C#4UdNoWGOO4GFGE`3#wgrD>!_UGvm45Pc%&?tLqCWkz`eV!su{q=+WeH
z3r^1Ro0seTw%x9=vFX3JSFhHmQx!2VI`{|fjs_jp^*wZ}3L-3amwk;3x$SxT*ALr(
ziqZ*bc!}Q&kqo0ZA{7ijStuq3l45MsQQ?<;TL#+l9%iIIqOQs^OWbxLxiXvX_;po{
z?nhJcmt*R~D})n!%WrM6cHD!6W!LMiYq;bJsjdR)`4b8$eeH<F)w|fw-`odh#u~3x
z{-3)@F#Ketn9xhotPRJt2;cjMwb!^!{59n8fmI;#K5GbX=%0U1>ibt_TWJ<fJIUG}
z{W>zcvU@w!>-Cec%+0rjEo5~4Q<tQ_J-o&%SyB~<3J%xnEuN?Q!RkPe-cvQ}O|H6h
z$00Y;<Z?Wj_PfAC?E=gzN9OCoqVS$SFk*w34$<$#?fU0dFQtDW|5#LO$go%P!&-m+
zAdmDgd;K9x?y7-fn#xw;`Ck2>cLz%U-p20u8f&cGkYfG)wdKleECZjRjM9twuAi$J
zQ5ygMcGk>o+66xXAI1!Di@-~sGRHfAq$d_t?Ud+ztI~MzrgE_{>SJm?(?DjM<4~u4
zlbcM`na*ni1($L9!@5dg-<{Tv0zoYrHjZh2-1LGf#Laitb(|dnm^GHZ;-uWyQzygn
zL|t^IvzXK%(nR1Vg-*)Q*A>d7_gR;#L?rf(5u5X%;@0$EyUX|bs$y5po=N*HTgdF2
z+9b^^r2QlsyAaV4HpEAICn3sTd<C%Sh0-c56ijqVsk3DBJc9&=>z}{jHivBN!@p%-
z2@aI(v#~05WKL4~pN(!9F`fUT_P>oru4}0_+&S5OU?!V+mUD@Eswg9@%=_Dq2Tb%z
zFWUc^48?Fqa1{D+4ledwzv(=cu9mBUl%Ix$upgBQ)Rq`$sVdCF7QfyLNX?goC`xtt
zKQHUngB#5v%psa4))H^Ng6q;hytvajHczJ)zi|=N#^1d-9;695Ti<-7BA&oP2g!ED
zjSN`ft@!sez>ZIf8*%fMDk)9aO<B}X^I~Omith@8)67m5%(<(w{rm4u6VZa84>~Co
zoO$uov9FEay*Zi;_u-?uobsCca=$nAqb*);Ma-4&eV4O25UcaLJvyk}R%~g%Sit<o
zQMRk9cYUz7qw3{k@|;JP)hdHWSlJzks}q+8tlrYpXF_`z!nP#|aQe%K6B%NawiZ!a
z)9|sd;ob3^rpS(cJv*mxdIF*9EHH+;@qBq^K!tIt?xI%!8^3yF+!~j1vjB|Jt6H7m
z{x~c95);;?FG~4c{M<QVA?2^ps<z)FSG{yf2`e`x({8f3KkBZJd^3ZVxYx(rwUTe-
zS0GPD2)8pTIu!%3p1M8k9pfGE*X?vsw%LacI!!gP?6ayr90ml#y1m_S2mBI|e>8Gq
zH1A#jr6JY@4MUT5W*k$k_&7>r{%@6Y0Zd?l|HtW>Gct~5pDLs;>kH_w&6DV98)2$6
HonrnM+j++J

diff --git a/vignettes/examples/timeseries/timeseries_anomaly_detection/unnamed-chunk-5-1.png b/vignettes/examples/timeseries/timeseries_anomaly_detection/unnamed-chunk-5-1.png
index 9700adcd2a67b6e5906a728b383671cb635f0dbf..89fe25b04d5daa66e7dcdb099b77208f70297cd2 100644
GIT binary patch
literal 35821
zcmb@tWmJ`I+bxWRAR&l!NlJI8fPjFMlr%`U(v8AWknZm8?po5_xu_-GUDDr)_x(KY
z-tX8yKgZDJxblo6<}v4S{Zx>XM0-Z?3;_WFP3qm-4+sbk{SXiylp;L@N3ymam4YAS
z&+pXj5D-va-2Z(rI`4Unfba@I>aB>fbK*{tQ{qeUnXZDI*E^QRF^OTR-i1gn#D}qz
z<(^<-*77S4Li}A3huiqu6W^@9T^wgmdg}LU^BWQpO*o|w61mjJ)^JA{Pg&cQq!e^F
zb9Xaua}RdmaW_4*^WI5#LPdV6WGintk)iJ~7ze`SFJs61aUmSsy@`DQeo-FiBZ41>
z#}It*Q;M;MWOo|vC4AS5d)2Oc7ZuEyayv&TgDwNTWhS{ks#+J;g4%5+jKvT~dD>;W
zZ#6jS=-Mk#R*@0w9!^#a$z9zAP?P2vt_GR~JjXY|KtOmi-$tbGcu@_#8<$DB%P%?V
zVb;YL-aN#;)qKw>%h&AO8^O+{s%jm}@6;ys6C=-g!VBl-BNditrc%M}#Z3C6s%I7U
zpU*NSUkh{I_ITYXH~6uO@>O3KP2t{siWLN7cEGTMz(%rkQ~I^xHpSEsj1N`D9nqcZ
zobzR$RN~$vku?_)dVW6Jn%XXxDu7<Oar2Y4?C#&Jc4Xx&eVn2q`Lo|hEbcTnVl?Fa
zq?48%JTa7pI*8TeYzwYE)9C3Iim?Q*UAkJ!xIOFQB8(|H#Sn$e6IMGnoNOmKSYBv?
zwZKOAI$g^Vb`EwIys2UyI+a-MO~GkmUyb4x(rSiMkFIvypselgupO28+H6&tw0O_2
z5AvPPH?XVWkZK&B*+u9&FK1vs9ZVvmft7D8bLYH|yMX;LxV;m$g%4ho$R%Ho27Yg#
zU;TaPa2I#hJMP7FU3RT?_3OCLsRHhLnngP%tW%vMiVck};PSd&@+GAA6dbJB6=rUR
zHV0bWf_=^{h^6e;y6L;r{%pM5DMpUzSgd8R&3~|FQQWOl;jZ+xJBQmwV6J}OXrOz<
zXbi5U%D`v4z20Con0Se1mqbMN_RRh6I``??4!v;QS=FMTjY|oSrGnZVzt=)>G52#T
zr`z9T#9lh3dNFU1$lmf*nXlO`OU|*G$-0!<&s%Vx6W#9V|NK)eUzctA3nE`bI#=;0
z4vwpR+G_kfOs**Y;+!Ax?&3!5+m_W`<Eg)e4FTKHMb&G0oA@u=s}yj?Xrx2y#JRiC
z5%y+b+kEHcE2s0Vt8=sms|$0%J}JQ|<q$gOj&RzL+x4&PQ@M#xz$VU2%a-@mG&I_@
z#Z1}czpyG$P1?KZ&QhPxXFeCK`xG;L{O1R$?Ule}kDMchC{h_`ljS()ZkNqI+mU<>
zcg0VWpIS=ut36d?#ogqs2Rqi#D@V<P!Dyz_st&>9_5He&v$P`1Mgha_i=hF{((Rt~
zfL#a%c~aiSVUN@OJxkq<GR(Sl9bOsho4CQju~3K0eOsiwy4^5qpVwZlUF-8dSIn;0
z@2+j!p9=kP+w}eh*RH;}dd&~>@#zugww|V%o4;Nmlb<qKyIa1M(S|r*hFA(YIFya=
zs$HGQaig1Eo~_f<NzNr|He2pZ$=axobQHP+XAb=oTdmdmMGCrb#t65eSFf^_pMswb
ztj)!sZMd8y*r4w$ly<<Ibe+`3+3OM))J}WJAsCvTBF&UP(8&TC)`#v`m1ZsmmAvk*
zR$y)}UGgY6*EH9z){Z6gn{`@Et*+&-o;7e=_7SjD+Mts`LJM`oiU()Q4!d5FX<2s-
z7P*M}F+z28bVjIl!HQ_=h$K_`vz1jfuhy9`@!!A#Y31c>O;(g-1k!14?l5X^($mBU
zypAupRgPd(c52mitI@#%4S|mk5EOD)ZHVnu_iwfPxIEdnw;73xchpiSv^?(~;O<|#
zwv3|r@||r*42cVd^>y2Jy{kT0zP@nNINu!WH>;W59jsP6v1oQ`@lJx$Rj^Ym*R1zK
zFrpZPE2P5T3g6t-aB2F9z_U%R9PoJ!d`m<|ez6XP+<7e6L|8ygH$s2V-!&W^EY%eZ
zlDfz|-CzT8At)C2(<*_^WXVm2jXb|H%jM#+v#4LuW|Ugh&8%?tanK!09k#q*>UMeS
zpv&>OQ<K{*N=S9aC%T=`nA3RZZ(pexEuzKGZaE~{=E+y^`mKWEZtC_5v&EzG0Mfc{
zk>(Jyw{zzc-YHkYcLN1&a}Yn@1%|_3Xj5&|@o>xcDd_p1GPESg<`II1S}>ep|DW+K
zPPL8lEt^<ca5y&1>t+e;sjx@*_C!hdVFE=RYK;dRx1_0VN1=#51wwp`L$u?1JbWBU
zSm8wAYS8cqY%GT2Mjaxz@afxS^|s47p0hMJn=SS~H?GSmBrr->r<<wFl&KQ0Hlqle
zq=TKYWl3FjS2?rJ7V*1l=_YTO4u}2cCaM$<xYf8UV`?856i1tUvH*-@7A=9J-&9C%
z<-l4t=_EixmarlIgjA?d5`|k5sS9R1v<<7UDsGyfikUoDLz1mOF-<U8F4xqaX%G`6
zB!ev12Cl`N?0&pUJnM<-=(tT!%Xpd?li^oTBvKF^mssEu!!$5oy`wtsnxgw=sH!gB
zVRJUmtfKFh=Q3Ya4jQUte~hJ+g75Mz;WI=j_<F!3xWcGv#}W1<Xo!%-R!<3TCO>&G
zC49Y;sd+Vy@@GqXyh2PwCt<H?X59`lO>p40aMyT{w&~>9lP|jt!sl1J&(5TFwiuqy
z^<65$a(`U*x*z^EiGf%B@sG_X-!~X^cblnD>^G6*qk`1VG;HusR(0(MYx4E`R~BH-
z2sjU0!AXRVNJ#ftMvrXEkL5MnciHngV|-4bmTWxP11+V%Y*nU~7zPUt%-=Fnoblat
zGxC+8EKeVP(__iqpDQmf*R81y#*zwm7fKPzh>D8(_9pdgnSAlk`6TJ1lG5_2D;n3<
z`Bg!FJ}MU-lBkPIU9;VdC-l|E>{wJ}bo7?n!CS?XtDDO}Rb}N3N6&d)Ep6?h*$W54
z(+l^Bi3vnPG(;b4Z0yUy)2o|L(+Refulj7w%<_aXykBjcWgQ*a&JUI}Sm)*Ek0$BO
zcWj-yT9})oauFasrlY4nP1-)b;Mj`%{ufFhU>Ji*{u2Fv!;Jr5Ptw1*-P0+Vx?1!l
zZN531&CYy^hj>2*hOV>S*{ic%8R4s&TRoO{NW9JmzqOs_Pk%jn9vjaQYX1uCjvjA*
zK><BI{pPw<1hs4uUlay;Upb9`IYYwz6+6bS!7Sd;n4g|H3!wY@cp|IFJ^j16eQA?=
zGm6wu6$=I5Z(w`BVuu-B6gTYgeOZDq^m)I+h+^>_LV4tXe81Tr(fM*4?Kelm!^1!8
zBRUZNJr+DV`hj2oM|@=Y5Xrmr^J42$Quol?{C;rLi@z86QOQKVk7N6eVSBXF<L~dU
zUSaHUG9u~yiip>C<zTVxB(_MiCO%#wRO>}j6l02@vx4<l^;e2^uY&?7V;kCn!!=z8
zcx^%arC0pjS|MpuyT=G=Ct3qbfI+wxfCRHUaHv;TFmX9UilAFG#?qlP5}Eg_KR2F6
zrueM8^|IuG#$}i7#U1<aua3UHK?;-iyh+)`4fHt5GwIU#U{|S3r(LC|JLQ|=x|Nnb
zZ>O5*jr9QGD-NYpxRLsZ-la%5Wy+A`y0H(R{6j(Gx}$^VZgQ?(4lmRG;OgoemzXrW
zRHtsAiV;XRpd-#cA6(}dJ#=?DWDVo^py@w7m!Z35jl89~XyUI}wHe70@63d6sVw9h
zY=^~~f2H+Mx4c@BVjD+2Iv^kbnBWFfIo78ob!c7bJk~%9zuC|z19ND#7XCO23&Xfr
zC#S%7P1W))5ku5(BTuLnYCaCP8Peq5-Mt99oosEZn<pY`-;FM?fWS5An3+u$PBQdK
zJ!ppBG!A<loh+WZ1_^UxW@Ke$(He4K_V)JLr)UDN*>8WFeM}kN#__?tR1pfdX;f5l
zm+~EnkB>jyo}#}SeR$;B1RT`fnu{zmJ)OKWik^;cNq*C&fAbXv(nj$+ELzo+A(KYM
z`R13b0c`B|e<t88vljtFz1Fi8UR|!db4K&*Id@Cjp|mMyJrInN7(*@;R;3c1rUUxN
z4<7_CZy<6Xe$dd+DA5s2VBG6&q?Gc6!gn3FSu_+C7abdt<Z^=Tz~X?QVW0R>TidDv
zCMwMelkgDIeTC6IFu*yQ`wB@#MFmqkOZD1Qc<9)%f#h7sZtcCkzW%#Mswn9+h(0ZW
zfvCh1eSLiu<>e>5&^@B!{QUCS-OEJX8pV<+dL&Ufx!^DfpB6n?Sy@ld=3qMyp_I18
z-!4K6SVrnhjEqIP(B?BGbRwS6s3>gPkjful_>u&M=#z>X8yij4gxt<7!2^WW&z%J>
zx1xJ{rE|vehAtd%3%ZuWC~5imXWwOfd-H4ibZvEYlV@uCv|fvek*`B8HlgaXhnf~+
zjV-a@lG(6eP0(+xx*BO~>fI}<j9ry4{G12~nfOZC@69IiRs0%94yqXXTJ$ik{4<gJ
zA+Sm3U!`j{vM$!xk_6}^nkzJxUUpY;HqGTfc?O}Y3!?o!hzMV&{nFo=c;-Qs5jh`y
z`^xq(mm>Yq2fxSPVUwT!R6v`*1@`SZ_bm@y-^s@jTf45<{M05!Kp6Eu=n+v4tg)V-
zDbW>Xh>^xBZ9xLw|L~NK_Q|EJi;Ihtl+-H>77FcOrUx^Upx%16D7Y3yul7PzT1EzX
zeIyg{;z=N)Ac&l$k1?m<DMB9ci72K%91s3ERL(%kYumnRYsNeJ^#q~fKC67C3N^a$
zc`o#I;~}K{0m3i3H&GA%h9mLeW-p{k)+wv|AQ$>6)j+%E{`h~gP&C4g-kZM{Qh~=j
zyAPKiP~SPvBi^T|Xarjnn)?u4`WVCbe_aJ?#DC7rINd({Cz63w_RmZ7IO?q*-Pe;c
z2z5_h{*7S|k=Rr3Q^2nm;NK_zNB;Zx@Z{eE(Fm=mf8X|Rw)pizZ=Lz?Ti^J9OLh8R
zL;CMUsa^lv_5T<n!vhGu!aqZL)4I*|&v-*M|Cz~m8ag^Qe&k<<f1^LLX~;#GBnc}&
z*?$h?vK$N8IEsGl`8(|U{Pj1NJlJKOwd4w37WM1}u0tXIi)bhNBgm4A$IqXL{N|B+
zDLy}idcLL+W~7esIAtiOEuB*8+nYD=gc*B6Y;nbKN{{s|E&^0UaYc^lTuP~OhHWl7
z|MXW#e1lYwKkh=r%^+@C??)N&g5ba}h^b2s!P6?NGzDqgKiyjQ8KO@(CCu{6xxOhj
zZj7P^c{#&#0uIx|r4FJl+EYVSRaGOSEL+1@7=DgMTQZl4!eK<{h(3IB@7sZo|5V1T
zmw%s$NDYBZF*-KPTi*rYBvACD`%)~mB7hl<<`6f`7cVb{fgw=(wxzYD53N@KCm+Oq
zECT=8q8GZ%;}b+3E>j=mpPqXqXQaO7qT}>#c=7WFhOtJ$mQ3l?Hkua?XF9(xaFqdL
z_U4a9>Ri(_U0=ZbYmR4k9f_OY7Vr>D>^i#X>ywcI1Ezh55sr|&o}nE#nGRyfOL337
zmGdhDN?NY^fT(8`EfQF~YA-wd_^@w-cw31SMa;wt8`xG|bbMLth)gPAC-|}lb1&JX
z4J;4|Wev##2aLy(WX5yR^{FB6!XrgH%%hp9!euVyA}MLZ9&cN$Y<||;a`3<6O_CFj
zHCv62i`zVOj#TcDvui}AKA=LQOjqIvsN8tj`PWW|zp!R4Io4_Q%Ad&Oj7i;1OnrWw
z_wcU`Jns4;9TJBzTJK;%@PM37zmr7&-(3EXuEw8kePsnh6cL>$w_O<fzghmFcr0(Q
z8%W!)<&)nJ%C6b+5P>)lM99s>`|}sZKQ=htniv)q78fUlwCLgqGBWpq^KuQ`lV9%_
z!b}i=3wFjvz;lsAh<mfW_#L)R#+Bqb&Br&s<LbPEyu`+K*ce_+cH3*gvzz}ncG#<O
zY5R*^4ZvWqAOg-0=%p}AgU-Os$s$5yqkkfTo{&ySO>J##;w;yjxCG*0S4kFy*CI58
z2nhY_09yE>6O%&@1iV(G)VM6C72v9=sptszo_(xnbHr)8@~)o#HV~h^UlyC?2}c*e
zdG4(>9l*43b=}KcXdh<(WtEN(uWn8rEp(y&ou!aJi$N#n-<Wmr)~$`=o!CEzk&GTa
zxepbwA>{oZ$)EfNa9dAkgZ1Bo@~<#bNzHdz4c1Hep5Jdmzpr(qEI?p?o%*F*&e#qm
zLd4%db}_thQm9@rQ*CA8sPV7+y63WRTg{B`KoE}qxkf>MvV0ep+H1n)pSLSpR55&t
zHo~8EoctRp^{`WGzy9@n`?GyG)cNpJ&{K*??e)KR_Y`W@Si{PV9$$TVD2jAHn!-n-
zuzxR)9OiKC|M&k(TQmlNb%TFnVT2+CC(ZKB-`5eB2IFJ=jf;?KtJ!KxSovfdMsM-)
z-xt2na#@W+Bj8}+$NuI1_j0xw?q0-Agn#1CzadQz`Bnedhxc3YFKNf{t&@ZuZ0kQ0
zLgskJ4x+wG?6ZmuUXMR|hw4!GJ-CwZ^OPe$23Y`QiJySje)f3iImz{z+wD=G<{{<5
zabeDaplu5k@0C+Gw}6YT6TEjS+o~8Yx8PPk>$CinOJs@&0Riu;f@9L_2ThR$XI|%<
zmE1xLw$kJfKR4*?!dkVY_0@9I`8jXd)>HYz{QWsGQi9`k=v?gKZjJAV1xTo-dc^Ds
z5${4^Veq4+WL)b72_&-i$-On?B3mMevs@z5T5?y?Tqo!LI{91c$waub1i6*-L|ls=
z%e8HKY^XbmcWPUakvfhz>nGQ6*J4Vkdih8rm|jLY4RyHQyo3XsR#45Ugd7ib>wros
z{KX-a2!?&)yBEh)%f!<!bzeLt!_-~kkqggsme1VF8KG%vw)H8qqjek*^N>n8&PimB
z`+!fqB!lVDYg}>Is6mE>Y0Q-2xs6`5ZDM(M3UxizUTA|iUJbI%4j(32HAKJgO~Pp9
z-6<b>{bETW2^2zY#DZa584P_SA(C*9cj1{OAR~;dA8*RAoEmsY@CJ`L6<cRiaw;e(
z8DuKv7hVQZPu~r%%W$;R7a%LS|D1d&LBQD;Ln$?C#Q`$F8C4L#X{2OtUy8d5IQ)L|
z3gb{u<M?euIYXaLiCPWwW-%((O9zkFqv4#RrrrQSy0NpdO(WYO*><d*>u4K7_ilTd
z66*az6X=PHBHwv@P@!6{MrVykNFelluyq4M|8<TAEhFQ=z`(>|@E57@YftFx?xcQY
zdV1@wBY%<{@%rTWxS^5JCQn&?{fu#ik+Cs>+4%bLlZW5@vZ4h*J*cdrvbw&$%`>%s
z4iaX~_LqHjlfe}L)N-dxL&zv7#C%7Nj*bco89W9{_<cb|y5ra&v_54uzJ2=4J2WD~
zAag{I1=Oq=LouM3ZCgaj<F5Lr+LXvR`3VB>iL(gCTSfaiEzIZDv%Bi*>LbPOXFIUO
zL?VyDeNgaDFamfjm!p(&yv?K<(nTTl!O^igOakPfsqKNjzOgUGQ`-)8S{nBLutWy^
zA@93qZPbCA^W)-3WM^mMPlRpg&>y2)&CI6e3i97PMfh5e3=#m9NXLp6+6+fu878?~
zf#K~&<9bS>BvP5~=VM*T^Lh`{o}JAxsf5Lv^U<-#0De`1nx5w8GrKnWTX!r#h9#1C
zd1iOnc=px)$#!Fw+c^T@3$dM({$XF%mcU1y|Ed2!-s82ufat&Y=f6nie?VHrzt>P>
zP1({mhWr~+WWJL8gR1_+-^+YO;QU`O>%UL^4`4D}(qMYBC2$`%S@7w4(0#?0TRt+W
z{6&RdFHrGVewTHTu!GohkJJ!;85*h&0_zwY<O0y>@1g&^?0>|9f1cII@+|j0JpO0Y
z{}LRTFQ)Y}{|SNr_Z|O>6o_cyGfw>rJO6tl7}no_`@f#}FV6gzPx;!TSNBrkZ;1WB
zpZ|`kH2B|z`kxbyN`<aXI#S`H%`e5nmTTX<O5KkKuG&>8p&l)+h$<h51Bj6e_ZB^c
zuDTYz((G&r_|N)B>h(d-ykGgJ?>*k;2M|s@`ZP(#qsMp*ws+zzz&DuQOzEE<Y)2(^
zbm4xFxjscvf?)F6blh2Y)JhrSF$h8Oag(Ow`{y8f5xq~+#en!d4M^WOzsk^7k2Wfq
zDj2$$F!Mu+r}P+1d1sW{;6H2?DYX1kIdidj<ugyBT+SsS5I1%meZBCJ9`gj-@euJb
zZ5&^KYSG~jz?Ys-;ZLsS;uw&?z`#({iFcp~Q&o))2BlXu%rY)6PJ$pSK5buXUc&5q
z?!FcjK6KnTU)<Hyq+?@joE_}{@nwe|OKezJS{sd^BJiR?eW~#Oanda&)p=0arPBD<
zC&3p#9|(}{|5Sd(N#MYj1Z;;lcJen<4E<7zz)*u$v8fvR<A>HYnkmO_0wf*|0W|*q
zr(cG*G`~c8T#0Le1XvnSy^PS_mn#2nbNYX7-Uvj!c#iWfN^;+0>sSn0c<Vl|2u=I>
z5aFx+Lz|d~=Hw_9Cw&bWKcFf*yrstL>WAayP*l8A#j<2=P<|mGVBlhYSEsv$=OpvP
z6OMe-ErV^?EidQFuv6T>jBjM#k7ZMGg`{GB@XF)XZtS*wi|)BdED#Y8-U(sWSXFQm
zJ33tz*jyd&1&T#oqbyf@lpnRmsB&@hnSpEGTw=Nit7tAK2>+>`U$4Ni5?ZRN@i1+%
z^Wj^~=mH+U=Ht-sknGpWpth@7VQiCCFhCVk1!9GmMNG8AhMUV3k-^w~KF=f$`(F3y
z{OBSDHvmWp%EU9J?lPUECXC;KcfDczWlNh%eS~LsOePQIetR+1aM9!W{Q7gW7QDUG
z@b4UcMkhhp2DgV3kz5zPqJbxVt@WEWhz|BVJXz5dgF%q8X4OcDcxsZudhSKQ(FniR
z{#6GemVTX!Z5`dj<zT4O{#c8ZVfp>7CzQ&u0m6jB`g>`mrKeZAb4R@eZJGp;G~lG-
zFWWpm`CJxUExzky@BC7aGArJ7Y`jQvaPCzQGWOaImte7_7!ujH3g0u*JS0H8*IhPD
zWfPkor!-Uq$&-TXLRH@`T*zE+23Zt)%qk64l%#J`4bV9fS>Cqa5fCH-!u=f<fJ?bQ
zj~w%;?v_b(s;xUv9$j4<wQMmSOBf|#Q^@du#~|Rna3HP-TCQz8o{hGi$M4^R(VOhn
zK=I-j?gu0Iv+-E1fQ%wh?YZ09sF7M?&DLgD3RG|QO!YZ8*MIk~WQX8tKbv`*xO-6e
zF6&$xert0$PYQPsOCFKBk8GnMerc8${Zk>j4j0|f>zh@W@Fd~UpOvn=;i5Nv3rVyA
z@O_*4`YIa#Et~i6<&#z-<b^IaGD0@iH^*}J=jz5-#_ZMyi8;-c*9Vi(DmG4CuTHmf
zbSy7Vr*uJ;&*y>z1<a*(gX2H~xBSK0pTp%I_H)2z{&1)0I~vWBi7*+_+G{)>oYM8W
zA-((alxzi@OcwHR1vg)N-f7x28M%K7Zwhv~{&T2P5`%G+BYa~9Z}xzM0x~Lv&w<Vb
zrw#M5ek%l7LPL4F(kz13apLW`re*>^&d#0LqAv6LWHZN3c=|Udp{sp|uB~!~i<{dv
z-{(&sKDgZX$EJ5Z-)k}+h+k2EDSkL(QF60WF|+}-m@LE>a4vmNynpJtK9u6s$5`_|
zo`V)vEjr!kay*#Cf4&bFc01c~IbORtXnX!6&U3%%e62t3csWG4w_E_6+1lDN)ZAHU
zy4sp3n2>E+Pj<83ovE@c0#tReR$Vy82Q97Z1-E@rE$zvBU5-_njUkcoJMPpj`O{Cw
zEB*QH^Sl$B11!>N!tKqK@KHB~&(r5};o=n-;)h^?n3$R24Nm*zMt!QqTD<Om4jw&4
zcfY$i2h~kHztdjrm*v4^K?_GV{DXEZ`N)_UEYZFC?P9RTO_h}=!)ao0_v3-X<t}kW
z_sh+k7zXulfo;ud%QQxW?|@TN)AdlnfUMwCLHyElm9HS7plY(2E=@BTO75xFq*vr`
zb+?|q^>SURa)Kw=T4bOp?YV9&Ka!yC!?SKWoh)g-^v58VVq8qr0cQHQ7vKfi3Rs!!
zRmBe<%0YF|MdHTbl5H{&Uj>-s3X>t%riBJDzUo<P&DziUW@ek<g(m&6VW<IKf7(dt
z=;#1NEOfb%@dO2jh)2)#+h9DW#Z0AHI3=TQ^L5*E;lkWpg#aAQ`%6Lv7?_x{Nl9E*
z!X!yOfF{_iZ#}+zIOn_^1Z*(tZgjA$UXw!V#j(zy6Wrq*@a67jFq8Ru2Q2Zj+S=*;
zbFbTr^$f|7g({1f`cD1swSk1YbLbtWsE+4_(WNX4-t}I?{$z<RDbi!dy*ZaZ2l{Ig
zH=cL7iW%uQ8cIs+UzXYr<lL9Qk`E>F9s?hUF;aio0TvaQii(=rP_m$Aw9!m!MwMj#
zam!QEm$<lq_%hs^t4pWxPyYj+O(N_mfQQ<AGAbhz!z7!pDw`}2clN<u2;SoJRINm3
z9_r;K+MO;QfKJR8i5j5md9m6>>Lp;e){jXcK|}5NXVG`LE2OX7^0L!B>^scK{w6Um
zB(S#e#>HdrYN!Voo5$~l*L)xfJ>mglMPY6%{j%Jd;(7TKgUxYgIz0S^=sR)oCuEd)
zH&fX3su-fT7eijj;8MWEr3P86M}skVfQ6&VkiI$UWA5wi1>4ix*GDBqM@!4=dO~AI
zPj-DWCePe>&<aMJ;&G-f>Nm9R_WRpY0g|CSrJQiJBr90?DzITyr+}!g9Z-^2{xGtx
zY;1HNNf&=Dm-q_8pi#NeAIA>C2<cMN(75blh9Y@~2IhE76>9^V_XHaRk~gIkpl^Vg
zjZPM6hL8xxXZEhH#^~NI1u!F#Nuw({&6nywA~W0?&%@QW$2_KSb#ORM<h2`jS^)$}
z=jup?q^|p6M^3V<f03eScTdFYJO`<8E~}YJ*Hd%Rcc3({lPy3QqyIwPUET-;46wM$
zs;VzUDW%d!EJZL(MP5tAf^EM%-4-Ny?m>sktm8f;c&@ChJe<7?hJFVYjk;|wn3$hh
zF41|VJ8Ve!t`WEoy=oy3*s5las|~C8g=v49{-wo^{k#2-0gENYb2VPOiL}q>$1~y@
zn2J4OD9wOb7O0h|l5E;QZ=x|owJGzAlvGs2kzjzP0ooOfzjCaMwDexxI!QPLb5GFi
z3{);X<wU56Kp11_yZ7;el4fbIByO7p;0QgF<VHX<y5AXxDeW4QAreu)scr1X`XDfM
zg^<eYoPF>}=Mz!Pm&*_Cmz4?UaM)6|9W((JhI@G*UqgU@`C(G17HLEUUS-MBE56*6
z9hpS9;l@DfF4y|-q19babMum(vuZ{g;4lXV&Z0RO0oD3UFUSw`b$083K{*6*ZB$pS
zNTXviv9+jXL3+@LHs~bWZZQ-nMl*99$%cWfedX6R&>Jg27M<wctAV6skZM{fTZqg4
zLeu@8O_v!w5lIyXvvk^-{<7AOAJQd;1PVSu+Z80u3|)`Y3EjI(HPP-egD%rCu>VR=
zpOa)_Q#1|m+O3sx5-?+aj}yMzb24c<I~ZuvFq{+@R8=6KwoeVlC<i+<Iyx#IfD>l;
z>==aIsCXIeJWz@$IvFzcudc2tM|iiq@>oD33&3T*-S@haK~jf$-n9S3f@x$9Q=>Q0
ziXtTuy-hhbL^%edG<$rt1FHb@$Pn&X*Rv?g&R(mU(owoqvEBvSl8!IuR%sa&ZU>Vo
zkjn}X`yO$&rQ@=Sy+%RN8>Fy}^;*vTsONQ*?o^4ce7HjGi8r9l8eSoRwcG_+1%ixa
zDh5d$3p6~@GE;g*DFpD1&f1Un@C$m%^T}fEf|J-ydFXYkCw+j~Shj3DM_zq>JmL^C
zSvn2Kt@W9i5LKlFZX08r!j6s(xAVO$67l^oj1Q`+6412fOpez6bUFFzVl<5xAfbHt
zlu*vFKU1~6T>{Nc0U~%@V9w3ub_u-P2<4_QSwsx`JEchMCfewwM%^E<YGyUz!O^sn
z>X@jAW*Z?~G83eclvIx{O^Nv(3zc%@!YOG7h{G#MDr}igdV5a=NkS;4q*o$}6g!xb
zex<+9j-dQWkR@_OX_on^skYV`WYyl1O0TxL4UiC#$$&pBG`dvWR|t0-BN>3T%7Rie
z$G$IVG@r=FYq{XV>Q%kF8j}|u@du2_XiBWSHYK1=PmYI#)zs8bybnM$5di6RtD4%*
z?xo_(sX}!z3qA2zAobDmowk4doQLzO|E;#csC1BLw2{lNuMc`+`IwQs+5E3>ulDKT
z^do9(t%XQGULkS#%VCOt*&NNPjLEo7Mt+J;<h-8b)NNz;F)cg$M-lH6b0$1gRW&t`
zIYpA<sHmu<#oI6(+JbYF%}b_aIBfyj21YL%t}TKAaL&E8F-IwI9UuGxQ7&>48|_BC
zur+qd0%$#Hq^9(~o;c!R#k$fGRJWakHqHXc^<?!LNOfaxUMXm3NGlV>1s0}?ZysMf
z7U2PC07N=KoAQQN9$Na-_~*zarLz@6o|`Gjv!dy9MI6lOmVN5roYKm;)=*XLbR?Ms
z@o)3(i&es`9tmYKI3@BiSZgD7Re5tP@k5Z6Gv+a?wd<=arkHe_1v(#8O3NZ{5h5e{
zY-LYm%O;pAr=-<!!`&mQE#JkmT)7M7hXE@!G7UQ_@w(Xt>rJU&_LQJkr`%AA&HD^k
zfXY!tVc|}97zJhlOSL1FRQeb@GTD~M1NCU=<rX~^-LJ_FyCFf`-MuY55G?Di_8Qv0
zgY-G@ma4y}`<ce^F#033Ph<Mtna?Wt42zH|j?jt(!*PekkRTV_i9#M4ic1l%<wx_5
zQ5oeDc@FQRL8S%V!ml~z%_!iFgtmuSGBKb_VAy@S$^us`064SfMw;&@qTlCN5oHMW
zJ>gW_lSN9mK(T0yYOD5Fud`h}hkKPYm)m*}d{xZeFaJL1`w9{n6_p3g5nKbglNKMP
zQnZ{&BRA0seJF`vEv^;k#flb~Z<Fh3IO}y><|Y!(3Wx=ZMJiNmit6fdjU?>accSaf
zx!Y8yCTos0R<k%Gp)b71H&_#7ocEL@74Lc{8FTN%=^a+x)|SIY81@#LG|kvYqsUMh
zJ015xaHupL@lF%(-K~Kazdt31SkBe53gB1L|9U1B9+^kJ=n(*tH>KI5XQ#@?p(n#O
z0wnSvARiAFT*6rmIw2e=<RsCrwj2}|$2Riw^F`0H^}HSR<s1&PL(FSg#%zL#c*6zs
zRqM$?9UjlZF4+`nd%ibMCc)Zi$(By@L%s>+?uQlf$kSKx0rX?^u_zVM1H01_HkB(0
zFX~O+byEl$^+q0CeQJo6j5f;pE$Z422?ZS%SsXA`tC3<=V6X+rTU+L9G81Bf#Tr%q
z#of=p?J8lhafvLan1`1|5fTf@JB&<-wp06!BDh|x#%%Jmf^Z%yFQ$0CveFX)BD!+3
zd@>8bA@f@#e)V*z);~xpdH`*|ww~fue9H1eehq;18XGuC27R8TLWgRs+sQ^+KdS$?
z!`Am*y7&ZWHH4^$6B&PAR!bPEFEo2J9L3>HeD^nA1@%cU$HQvHP&^4SJ;DeDVRWEw
zD62}*QX14nwxsRUM$43GcM+A<q>^N&qHO@YROQGfl4l26EhduESH73yGHKxz&E&Z8
z4!gDERhTh+EPd0bk2?S9;%K$OaaW_s)d3izcZ))-xZ)hBR-&_u!H|&Z&KKP938?f}
z9RLq_Fonv3yB0ds$x&1FQJ>%DO@^W<(yT|+#45JpQG{a9C`m=<8I|3(#0$mPiQ`={
z9AXwqjTTzFn^>`&Sn=zh=vyP-SWaSP#g^OptBs_Q)GJLDi?Z>bRh)e;qiror_6wtI
zSllk3mUC5e^P}+xr9&W&A3JrVc2^+CAOW>@wru1no9q`d(X0E=I~_scxNAoXr&B|T
z>g)_{7W@X7lDPwRy`+B93uAa$QD5_)rEM#!6V?|)JGaIoG4MTFPF@;}CqaDerI1J&
zT5PnTW6Ti4k0YkKOqQ7}{6xWqcvNAOWcR++RoJYg9v#E1o19rA3DNU6#?#dYY2v*g
z75h>~7^;3W<xgwl*6UtXvG7b_^bmoch$pe6Ue{mNG;1cMi?Gpyauam+`T>ceA#8U^
zpqVgvl<(D8Q6_=oc{@a?8_8|1z~|K~*$|kUvm_2yqp|C~a%*B-9R5{z`^W;nA%7|H
z)dL0VxvyN#e>}>Hy4nQWV;|5dL14}qWKRghV&l3HL2W8GR%6WZ)$bx3XN~eScD+9?
zBMZ5&`lE#h*Y-@I<>q)xgzVJ`ef=MQAz8U=TZw6?C)`C0(P^L6@WADYo1{`BNBzoO
zrdoveQUtRP5L$+W6HG5am1bxZIlcyy&i(b3#Lih|aMH9^d@Vwa;*<M4nSXkx2Y$ri
zxB_``p!5ftgp06A%5(}^H_3=nViR@DFKOE4c@rud3kccrx8)V}81Q^u_*PsAv_bZ^
zW9<BP?I?|{S;T{oEYs7Y`MU6(AD^jA4IOHoH;q40*CzTSR{@gUF@madbO6*UO-pI~
zPi`I)FlPlW(VSb?NDg|1bYTscvfu?;$Ync*9bCAN1de*REX}-ScjTpOWk3?G>>KS=
zYUFn^-(;N{A8&8A)Y`6XB|Qb<UKcYYFsv<{^1{8jHc@xE<=dNA5@qIz@VAGqO$6|D
zyf)2T#oZ19Tf_F4ve81nMaC1Z<s+)ykdJ{mQKlp^&mKSAGWo{m#et0}KC*u<`Hw6x
zWjr>s`3zq>duO(WS)dhnS<WncX_P%^AP-OF@w^Uq@u!K{Ry>#tye7>?QePwN$981f
z;%C03Et}ZW*p6_k+4f+yv;C<%?_beHU0xBZ<*S`$mHY~9qOfPZ3Km6@TM}+2-WUG6
zcu?xRu12T*h0plt@;w0D<2SSLTwhAflYFxUIE2$;l5$oWDZBl=Qs8UPG)ng$&r<ci
zEOH&`-L_t$%@s!ZkZ&z%mj0V*ma3pH$o2fW?fI8^{*?$s!Ddg35Sx$RDofszqe-d9
zX?@o3CVE{nOYHgc&8uHuv^a|VgND}o`uoQh-dzKX%t5kg&j%V)$Zsi;L?t97-pRa1
zT)bH}<tA*)W$5EdV4V4F-!5Y&Flt;O&MwVwH{u!OS3a-e>FN3BP10qjfnmujq%ZiO
zR)lsw9a{Vh)D<8cfT!!on4j~-^kLs`Zhv?8@ZPoX^}(YgAq{%gJ>_u9hIx`h{XoaY
zP70Ugkg%pN*abLhIo}7rXa(+Ii?cH6jut<)uHj#bG8<CKQ1ornubGYVF`MdLwG|hL
zUBAG)=R)VKohnVWcrZgy<3`^J8yXtEWd}0<*k&LX7w+LD0jY{ZHRKI^A-n10rgfD@
zaG)YS;=>2PAiV!kG|W-K$}!~_Na_!)Z=YXR-Uc`viQ}%+?AzU@b?EM2e%TDCAZE6}
zbvY_?G4(Q0{-Bxd*dUmWfCo1K8m%Sa&))LrA83Shy66NPexC0>gs5LJQZ}?<qR-as
z#<M+?uHcT>pzO(wmdVSpA~dV9R<H<B|E7-}7Fd%y>3^0+KIYcnFAA&15qmucE3+LM
zQc2n)5YGHQu%N$EYoQG#8f@Ha!Q71tGOTEAca=r!`huzUp4A>DsCOz~g}%BuH&6yB
zNA$X&kc{+H*qiD8;}Y(-K}OW|-Ttj}vV1->K7uUt3=dT-qfDwZKj8#(&9C|8m-r`b
z0*WYi6Kc8xTjgcQurw}-p@~RGAA%p-v2}&2g2hdTofh>q-783BkP`S|?{-PUlJ~;`
zhnSy65SVercO*M~{j|14meeky4^8D!XzI~X2M(ZEMke#PI-|-bLlv(K7ZjO6ASVE#
z(UT`nwrnaT8yyH`xhhN38JlxT(NHmtjU^KnOxObboCv;1d<=K(2TDF`u;Ir-GQqV2
z>g=Va-T<uua#^R_az7T|^H@nxaqIm;c;wm-boebVpsR^xfac2H0Q60?Gr$W1kk=cb
z=NiHbOHzyJn*el$=glQ8ycB!Q)Twt;8tG&glZ+a4d8FRP1hWwcfB2PX>e!}9lDAM3
zoa9;M%IdBWb^b7PBS0w@%Qk0`VyQ>76Q3Dlt(!(kE?lDnf>JSHTmiELxj-ymEqOhn
ztP7v_eEd+rYcnf`jyoZf#4b>Cm^)7a1yc|*_-to9FXAZZLD~)gwdY$CGBQUeB5u57
zB(u#PZldoBREsvCGdXvr?dGkPvHo~JmSn&Zk?lGh&F(R@l`B9o-V4q|DUZv``XHi7
zgpd&1e*GX5yD;s|a@T7Vqn&;01j_ppvG5iD7bLxFJcGs+wDSXocv+BXqgex06J}1s
znprEZgb#W6aDz!UBPr4mNL+PUq|GPEuW@)_Kct8evL!Zy5&`dYBBSwz9#-HH?4%_D
zBZ=$I3ZH6aM_Dxn^%-$-m^oDTgVA)F*i*(H8KToHTU*<WtKXY$_NK4e`HGm3j!RGp
zx!&I}J4k}}eWpp6TmxLnyJW?V&>0liS*#c&Zrvq#d7RtAJ%smdY=rhtCH-u9{wR~`
zR-!{W>;Nh61==R!EDG@g+pIUo$H$-<BXSGC9Y~ziGb1J4t9QaDd-MErG81shJhs3;
zrnOqdnS*M-U6OIAl2Al(Lq6tAJkh_!)}WN8e9g=}v}q3-+o;@*)coj&Nh%(k#t#@<
z0D8yMQ|R;-G{{jRv+vR?4JahPsiw=qSjG`0C-32k4+|3s2v4-YDzyrfOn7Gfm1V44
z$@>BH?q(mXK+#;%?C$Nh?p=H%2ZpGvon3^h#|x!ziE^*4#&z6wt9s;qu+pHr&sx<4
z5%Zsg%P=Y4<p*DuCS8r@s}uw<)%(58L&IUzcuArr9~w%=rE&gY`-E8(MvWyKD;Y!}
zez)p!n-L;p4w&Wgu2@_~jZBhg1D%>qBqKncxKEerPiff=EOi8cYojRjvlm_SKY#|e
z%)v|*lO4dPMhzh{&}Iv+eS7qDZGC<0p!|VBKk)hh9J`FlY}HR|jGEPrE8#Mc4~<Ar
zm2;r730#m+KzRlf{hXE%+W-aq^~JGEGIS$7Ae?VvkR>u<U}(tepyes3(Ti_Yv{Nm`
zV8AK?BTTI?6af_Gn%SqQc%eHab~l%&*xQvfn7!3RTa!ShU9JtIkTf<lq;v@6WoNGd
z73Nk~f+JIevQLdA?5GumSs7LF&=lT$lPcmfJ;;Ny&*}>)-+a3EoTTad+?*Dl%11#P
z@p(EEQzt+^*<&&meR?@(%{_DssAG=+97fX+*jHBL;AP|f*p^}Gw;3lA)Uxs8#oGLr
zC9WUaH|?oy+TQVXFwEFLFo^!ipEL;CL+j6Rh%?_iL@rWZYUSq80v<gN^dt8|E$u$y
zyCA!8Y)zHOcS&xP6b7oQsC;i9k_^Nv13;PJA~lNTet(L(=0HUZ7B<{YC@iy#=zBDy
z<_l2*PV-FnjCNu5G&vyeXOn<^y8<x=tWfYW58s`DOY0oym;q7+WjYIc<Cq@HA3*Mp
zXm<DYWoQ=X7vO=mus{t~%S(Z?X#@M$iwjL|1yJbOj0GT6jm%@tj`5z^0TR%O&-G-3
zNxPl~?z?Z^iW6@({#^s~wlujpi6Wu-(l|!&(I>2t*<mfVPwz!?fe=45IGD{o|Nglf
zwO=_dh6mqxf-K!U;!LW^tjwZIa`66WE}NAE=x?K8W|kKu3%~|3MWf0*>ZqhVFoJ6-
z@6l`8@Eo5?thIZLd$KXY0XGYhLaJL1<~RLFAuR0G?uSW<+<F}Yy<P<r2iiq=^N$Un
zZD-us62+vXq%=M$k;bY5&D5ZKngr+<1I8IqD4P0Uay3V=#m}h3e3hxGsimc%>lZ>E
zmz9>V56Glty$fSG^1_GhSU`SAjmf%-%bkS^3~>Vq1Vnkr<|811Z2E@6y1#$_4oo7$
zu`7sBGFGAoY=HerH>T)2(6t*08+kjCZ^sAva*xkuZGc?FNuRNzWpxU~q8Fmh$@>7d
z4ISV2Ysp9H30(;qQZxQ!J4~>8LvFbI`c%s(VPFz$R?ULzHfUIM0emJN0cQY)zUOif
zceZ>=vAdX{zoXi8@O_&iU$T`j(1-vR<HSQWo;3Ze0=^HgvC*vn8V;w0gfhCAO(>tr
zY_-+Z$!6PduJx!~EMTna067F^i6t7jNk}0XoN1lLLvE)pWV>?D{7%4!_1F|I;DFG(
z0-6Z7>F_J5d3z%SO~nnsV2|A#Og}pYb+()^4-tBN%K5w#9FR#VQBe*f;<yt9Y9ZCZ
z_{F<_n<L}fz^5S<Fhn$h@oFngSA_ZK#2MKiEWVFh#xpjh4HBiwo_Q<i=y-y<HO%f2
z>03^YH+MJ7q<0&id4vEb1?b}PAUfbRC``4#9D~7QfGk9v+11ZA$N`#{SC83;1eQPZ
zY|IKpa&=*=X^B)eZHw~vR=i=a27caqx@oxpcx}+)QY|6dvVTs0r9ZNsuG8e&n{S`h
znaVi<W}^;M@hvAte;On<JVewr693JLMl*W#I|{C`k<kcK`RJV@iXC8DbX_;8sMlUl
z?n2@)z#?H1jBU)-vETfVQ0EP%XJE)=%H80guThXxq>)w;GVBgbu&)1!0hAnhTkA_9
z`bfHLqrT`Z*F?ViC}WBmoyL*^6a+wk*cX<G_#kc}`~r(82*`Hev~sa5-!Ry|EWQ09
zpTy7y7+9<6QZkR<^Va7#=kU8h{{7)cz~x-#9i~RwOSkf_Zai1DiGeMiz(DH^Xt)lu
zy0rP@?Jz%V9*Gm3-aaIB#c(sV>*%yKP7C_Z@iYFEGpw(z0cqnfD@G?Ds24z{+<a^q
zyc`)^-dwZS{<S&{nGCcL0-7{vY}3AB9@sFD$}-?%)+c}rz^_K+vRQa9Jba6VrZr|W
ze%r6W3))ROo<ZTdP*0E5s9ouY<D?)Rft+bp)cyHO26_K|ssJ5~xs8KeoAz`laj_<T
zuaLMBc7ZlQ4WU8f7muB-G-Ln#`_0U>D9x<tIB)Sdb>#v`BW<B%V#UpuTL7qx9Z-nX
zJwpyY-JMl)NwEeTE?^I4iZr?T%3E<(J^Af7KJQjd$;GjWH?^yN)5lr?=&`7{IGa&7
zrY{SJGI<?zO#$DL^DShm{r!*{r-wJ~$XTFwR=B}oTY{S;=P8H9q$SYMKw<%%A;^z(
z0J9EQJ1Pi=*=YOXXPO$&ODd&H-lG~zEA9JZftpzI?y!qA>ekowQq=qY^NQX`+VMHC
zK)a4t00y0bqUD9CC6}h+{U6LcE4TqM&8K0I7`7BEm2^ISoCcQAu-}}(WfFCs!Lk^o
zrjtg>#>NK91u(TJ0F#X|JD(Kf0NoJ~L-oipm$G5f%RbAsKsdRMGf7TkT{nDxYYy@-
zz-z`pMFUm|GS7XB=+1N*h-i@Avv{8ShTdA?n@xF0=<y(bh^LfIH8B4v01tp}5zlQS
zXTY-dDGIijoeKG2#ieO&&7^Jq;p0c=ooQC8)PYi&9QQSzVnF}L=(y$`0op=bT--j9
zUy6MYbYAki9Ib$sFFpts+e*)H?vbrl@bZHDu`HiB&12uTu!Mwy_n7iTbpF5-t;Vv^
z7#jcqH@$lXRNHFPk#s<~Po~KW-`E4|0~};uyJ<Iz2te%v#to~{n&Z#UOj10eBO=6n
zn%>)fWbCi2s{?ZRP;YO#dD{C=3NsH-eZ_^Yb}9gjc&mH~7`$oA>c5)MC{TbYc}v;c
ziwiNK;7MT1Z-Iz7v*aRH=a>P6TnE700T6MoPhYw>@9gY6C2{^LN1Pf<)nAxe14Q*N
zqTToSg{up=M|?tpCeW9TFTOn`QTSo`k<6cb=-9d@LBgrH>1+l#>wT00I0QV5L~sq<
zP=DA-2qQqlFQkNwm*_%)k_d=W;iTA6poik>IP3SvvZC&8y(;?7OHUslv_9Y9G}Pa(
z)!@i@BZS|FWu`KbBcGDQ?*xSKCb9wxrp5!&HP$zx-JtTm2d~2PI5v%kod7M2NW5WF
zNG~Mn0!4;+BXiD}3n<3{_om9pV|6MM9tk>qtnDhT=Ux9CESd}^D!SSNKKi)ne4&r-
zI@cYv3<<iPID<}QQnww}Q~F#p3!BkQsYL_)O9d1=@J)k6Fp(7Q!99L9+gMp!8~w;9
z76e3mhM%Kk*_z(+i9BBbSr2*!bZe}Mu>xL-$42FW4<2o$jE&cT!Yhx`B%dO5uT(#K
z_AIsr5r^U9L5SCNn{kE1?#w-(x9H1hHcFWl!!|~A1)QkIh7cb#!~Qu=xpO~SkuoX;
zZBX|*ZsE>g9Q%5jFUi(i-Lc^L92U}JQ?~Mo3Ngo*_-uyVfcM-C9BdCWs>{yqG_E+^
zfsI+$&@ab2Hr)Imv^Jf{kIFJe%}in`NLlFb?+4a#c6~gQBmSb<FG#;hg7nne+Z$w*
z#_OX#PEO7(U0+oE)^=$m-{+)2PfbB414txZz=e(T5{&S5TlB*LR=R%He8y@S#Hdg8
zTlobWmHN`lok2iL=p9yg3$lCDSwH*GTdn?MAde?a1gwnPP^FQjAY)O3-b!tU36<V+
zz``3Z#ASTLxNJDaHvLa8K(D{w+~+nl?zzv8$RKGf-~+bua<<F31x#&@GOs(&tQY&^
zd1LpWzsc>qoUPt&U5@SoA0HnqzZzm)+vZ%I9cb~|Tn?*8HdD{{a=kg75<?fdyFON*
z-vqrD@=1K+F#r*sh4w1Qf=s54%ue<;oHBSBNbo?g&(rZK?thzBigvL<u?tZTYFpfZ
zRphX}k5x968oRmvMB3BH`GzBqlZ{(CBTs6^l^wC|2bu-wePA^rD1SMC5MT}fp~zH=
zZ$w}_4aLhUlAhrhP*vQk#;fJM=l$Fk%Ev(3j*f|O+8F+}egR_k_Z^5z3`^g)Jm$C;
zmZ>`SEl+UUP0>DnJHcK9{_-74!R#`epAVvrRFM)nK@->?pkb=C1=6)%1LNX0AMaj1
ze9O;<&#1&m{6*C<DuBa$Ja_Vwk+m{xU2FLeIbAyzKfQg|4%CA?7LQVSqZ_Bf)Sq^6
z6L4E=S@f`_{74s%jT<_WiZXk!N#73~p}W67P;fKeMfeD`-DF^?8xR_OQe=$>I%gy1
z>gPx&jFBGDvP;#u=UA{qtgWTBW6y`?Sy$ib8WhE=1c@VX;2{DFOUUy%H7rOV$f>uq
z$*9qohL+p8G!duZaFOGav&9>3v)BF7!Gza%?Gri%v+p>jaid7sBMLeK!W!mBi7bGG
z_W|xrgythwctX6GjbsN(iw!OwiEujYg0~9wRAA9F+}!Ye0F$;I(W>+etNo=u*EH@#
z$s;z@U0Dn)G1z;{k#-Ar6f(^WNT}=S7HeIq#KsPt51~u8{TQukvh{%b$4tO}*#Cq+
zkU)m0KbmxCHZ&-;HJ$2K-fzr)8xkjlKE^*m5@s;_Y~*7}8Jo0jx{-8<`f^5Zuc<kj
z1hr^4!Iw6cFM+%&qA(0&AC{~xcaSB5B$PbDztbA@XGi9Fw;1%>yHE872`J47s^+#k
z7);iZxb|0tJGNR*DjE7~jU**b?02#v>_+9O14ghr{~U`@o-a@#7Oz(>ia-D}@)KL_
zuIExa@~2Us2)d4V$JRnULu@RyZa<eMobtYp<*eK<Ojm64YJVKWg`mW<g$<XdS^kJ6
zL}CpGWjUEW3V~93FOrcaF-3}F1qs5gPii)*B|9lDW#?I}*&A|MQJ7S{p5=Sk#M`>m
z&bALub;{k2Ic2Qk6!_s9N>4`>W9{thqRvSOeAHXr0mfSn>YP~AyUN9V`4SJmM%Zo*
zRVqP&>q**9Jw*L9{3k0ih1z#2(HZ0lhCW=a&);Tuxu_zoJ~v6|T3VOd#`4f7aMhU$
z23pO+Bc%skHUM5I`#qfV3B7N*EdD2hU|NZ)M)S#MR&3jRx*wz>NGi@?QrZ2zTt<wK
z#g^i)jZ_Px(nbJ`?75#~rfNSs=2Cpgueoosc`+P9MR*2WPr=tc{w%uoBjP{sRalJl
zW+Pn($`h{RjAE>4d6MLOEI#jARQ3}*sA_#nHlL|aGuyoy`7_k9_5Evg)0f6;>}Hz@
z7D24}p+aevN_1l<$VU7`=sp2a%@2_b`lZOx#K11gEZf!vx*znrBD{y~bhZ;a783Ou
z2OplsN5@2$cSLH-(rmu!U;9xmn@1j);*%aSZYsp`nu4M&lB9yxT3<Q@)@>xyKK?sa
zomGB$y(t8bRkW`G7k?pdXq{rlmDhMc?cL_;2Q%#6Yzd{Tu0(=7!{APJSfN68L}ptz
zeUrM9POag!07Fcl1Nb(DtxmZSukk}721ry9pv8r~QLpH+kBDgZ>lF7-nTpaN1|CYF
z{iseuK|uj%Qfa%TUo%>x^EB2#)tz2vlqphHGaC`5g4V|61Iy|FzQo?IEIoI^MgWIE
zdl2_D&5fS>`_@0kMz1hlK}^HY*6^-XWtCE!$FpQ`m+g3%nM(jpM<k^BA*pL00T$5(
za1j7Gs&4B8z}R`d{+{TuY=lC!V}NDK82}Za?5L<G(8ODy-DM?&FWB-W-8a01Ni2I?
zH5p33s{<wJCQETn0tcE=2mxpBJ3H#&ub^5CVn#Loe$QtQ=;>$UrL-B#V)hOWP96yv
z(^uYLi8QcOG>t)TPfGyg&<K9NNyKGT;k0Vr$&CbnCTL6u<j7Cpz&7^zdEU{43IjcF
z9k{w$+xq=#7-ske#yWW*{i*=%jh_g>ruWx;hXkuUgR8c2pQiLOI3lf9-@KC+OL?d?
zGaMp(>kPULQMt@(W(m10KW^)*np$Qhv2>|GOBvI^l7SC&qm3#Sin0%lVWXh<_hF1@
z6VC;71L&D!T_~2A-isPIc8~AQq~)-j{@fXeFQJjy`~~DVQ`ls(l_rbI(ln)2S1j5O
z3B_n;U28|t^wu+RdC+#V4XUaF6mt%MFJ}Q&LJ`#YWL824;KEJ{S(~fZ-VA*u(A)Y$
zP22bhM{Yd6_~I||3F;u?#(X+R)ggZq+fu^fzLo5}gv^-n$CC6EH2Ktlq8+T&X)gy8
zjHQD~3>3xLtEG`lN}qJjvavyxJZc+@?M}N3ffatg&K7Xkw5#}nZxkiEt(J|tjk}F&
zAx+5n1j~t&wWGLm0K4!v!df7!Ta%7ajPw)5m%PClV}ka-51lJlvWM2O#KZPlWRO7R
z`?BMpUE#6F!PQM~%OdjGg}eQdX^|PlI@fL1W1kk&81U(Rnb*WZY6NA^`Y~;c{}=U?
zq1RF$Ds`WTD41wizExB<(Ll{uguMuNR*&W+0GD$m$nwTseh4g+0#X!EWc90Fgqw}W
zq&NM4mAz$DR&Vqz`U*%&NC^mnlynP7x1==EDP2-ZNC`>_C>_$>(j}$RAl)V1rGSLM
zS^WO@j63d^JI>*Q3ch>u>?hZnYtFd~IB4}G(&=*A&!B81j%Rw!B&Wu+9%-eU#om5E
zY@^Z0ZtZFe*L>I8u($b}18jJ57WB7N=tEl|sfKFT2WDFy5cQ*otKj7RWJj0y5ny(K
zuM`#M?@=~!KgxQGClD`+^=@?b>f@9n5}>iOFh-~rA0=@{WjBQyYmyFI>Sqn%SxJ-a
zDS*-hbRH2LjdObk?8yFzPN(x+bcnJ0`V?84H=OHtUU_M05nn!!qHS5A<P=`f_hCc}
zqPi*PypPi&DN!7oC@3fhd8Grw;AbZ2u64T0pJ^ImL!ApqWFU8g8D3>nU?Fw!7$vzs
z4&vK(hLU@xH+3DdZ~<~0xAV&zUGoqx0eW}bfd*(;Y^+!6_1RL_W4D7dN$#M+EvO&#
z{rTFv8fp=fn%{o5InPh#-M#pdpFF~<MLr5p>8vNd{6MAG4ISHk)p<Y%58I>-NVN=F
zt~4=?m$d5Wp2Ak%{Q93KaK)eZrl<Hsvb!>4bkep5RxbNy7iv^^FB|7{Uk#)CZV5k7
zu4-uCbnQ5xQzcA!X34YuTvG2onK$FgU97g{L)oYwUkx#Re~f~-`~R)=xH9cZOywHC
zo0PRH<|vLrXH0Q7W-Pw0%Ec`(WZy*nX7^;0EnYs2sbh)Tdb>4Si}@l-XDV5wr>*9w
zM98fb(;U+z;vJEE7m1=ZcDCn<sg0V{V1*)4pLY&XusEDk$IK%d@yzOC5x2CoTONPh
z8oIZeZEv3N>Kpkz>uutK;PzyZ0*BEh=XWZvD#RIPWU2bxNY&qPRzL8Q`r>ocPB@$g
z7g5F7>rnN{_^I$Bs8AV)P{f;FkU(rR*d-vR-yUc>CVn1>rY1oX-Wj9vst^~kNxvKT
zq~wbaLvl#_3w<T5amST5D}(CZK=))S#l6wu9r=<I;=n2NVxxd$S=~9irEINBq^&y#
z0yO>{oz!zYTyEl=V{&*s&s>IUP@f5!JS)n>$=I=aT7lWq7XMA0K~VEU5`Isv_E}rG
zge?`UVnV#02=S!ZV#ZkOhBJKG2gTWhW_`MT9GytVx@sn;pVJezXbUtGky=Bm2rRZE
zb#u-QDJ|MN2O|}kHnTpBmr?g|*}P#8%z285|8N)Ncv=a=^A*ivxt_gTY{e=e_dx|H
zgJcpReJ<p7htu}=J#;jjJEk9Xbz`-}62hrUB&H5c|CQeU$V`!~OqVM1nv5OY{JzXI
z26t*hL+EkP#&BSjXNdGU!_&f`3iU7mHtW7fj4-Z>F$hv?v&Cu*uRkL-U0PvZF885P
z#{REg5~*XbO`~;>$&8;B?%S17FpuKmtu-j>YUV<y6R3e}o)@XAE_PoR3r1laLOB6+
z^GC5q1#QcqLvsD7m9aBQebX()3uLm#DzEgjm2vW_><|i{pqLS(otcTac@~Xa`k#Mz
zE=f61nT2*cK8n1S@AG;IgxXQSR|3MhzN&}En_BTy&kj3)rI;}SwUjz)>vfWD6FE&H
zr`h}7RW3my&-C`d2QGGlW;aui;V#F6M2%r4FZCirg}@kwaY#0Pu$x|Tu4(+Gn}eJz
zZ!D(wsD3IYUojD3gcL)lSzPDYcZY(i`oEp{Jb|k?YRjz8DM!>>=8)9%@F;Jw*e+l<
zDbt4dZ^O!`SKmHsVPC`bbss9^GmgW#ew?>sW-*$?L`kGPSC<tui3rnhEYlJ4G&9Zl
zNH1)mesV5*oJX%ID-^9nYR1di`cVSsj!6yqrg{K#=Y7W*3}VN>h*rG7ig1)~b}0F6
zx0r`};)@sN<<mGVYO*77w~@b*he4REyTBlq{FtQpLx*No&QS_{%PD6JX@z26!oS$z
z`yZ@2G3Sn8TDkHY^NfX)*{MGCen7sPphkXq@z>Isf`EdgH^&sB2Uu@2&@Xl0{Agq2
z6?A;ViS-71%YzL7Yg7B$Y4=q%tc(Kt7BnTdxQH{%-CMxz-rbx5XU6Tn&T37zMDeQ0
z{W5(CvHgK{Z$e7u2n(M->?@Z&mdR`nR~ufjFsa%jHMim`-98Xf9yBW*f$r98nTy$K
zCYkUQk6MaW5a%cMZQep?B#1|Sk-+{8;W_5`LP<Afa{~G}u4t|+PUn4}AEoq%`0?zC
z!)Cng)RM&ubn3`3Q3^$O)Av4!l-#Ne<_n5WN(hGI{UeD;N8d|yql>ZOasTK?eWsf=
z9wU&xm)xG|&CVt*2+M9@*A<OP*?Mw^<=x>;Kot)JwgKa;R#_6PO8x5N%XrZlcKoHO
z)BW4?{<uO1jXlcO)w#n0)#=A(scx+f`Xr*ypNu_{HKQ+a`;fG%0P$L9@s9hhZVQ)&
zF;k!O*?VP+66OL-eCnNxjN)bUT`q1x{qb_sfrp=RNYjbb-)}5=BL>Z$5f$$%bVO@4
z$I!pFdB;^<*%<T2;@FqBrg^b~01e}zgn14t--HZ$cmFCAbNhmBVQDFGyGl7QZkGAI
z2QpZC!9PPcqPU>ou5@{3ro<D~xYZl2CI~iv{W*&Xki@%jcX4l;*9`}WKEA~0#9e9C
z8xfdS)4VYU$#;f_v?O=NZDKz660fRpegGb%LKI;?*aL9-U@K>IE)SW;_!#NvFv{JP
z5qzOD?f^=%8))~}B)A2rf!gZ>^T_W)BG5jfO9ZonVj~z0>+P>&PcuSWC8coUbzJrW
zk|#TV&XD}YctEj<{uZzQOfrHt5J=igL|S*U9!6@TAD%}wWSf$d6c$FAKbbZ+*NOO)
zjrX=K{k$OlCJt#o$N`ZC$@6h5p6u*=b|9=+b#}TNTSrx$s=3RBWRjFPU;rECDGP|*
zq*Onw{j*sryd(Lb9M8%;pEF2OzK6sJQ#oK^AD5jhzBd+5=&SdM914%(uGfQ}`F_po
zUc!c6{yHM}7U`iB4(5-=JDUjr?H|yGP@w|{M-KxU-Mc!x6owN7ND*T_2t+wZTm@48
zlY|Nl1ieUn+qvw-Pwsjzb+S&MO(OVQz^5h6hcsejQzMT9Mfa(cARmfn_D#58y(i5L
z{tLIogHLSvL7o<-dTyMLt3?wEZU-Y)G3y>s>>>xgG(es_AQfgyQuQo-Wu(>^1Z;K&
z0rH4oG$^dNVM-pj3*B7(2b%dYD)iI$hw~4Bma*y~lJuu2gDz(%-c;}<6&{|ns#$r*
zYU@gEWCB?sDOvdY7!3}R6_3=t&AXiMhCuMk3P00$oXH>$BOc#XFBtBAn3O~o-b}dE
zN^5}w*bE8I&ao$WCVW6){-H<Oksl8lU6$VIZ94kcj2q&LCO}{0I8WW@@Y$$~19e!g
z99HLP_DI=1BOWkZY&N~ZH5duBdqPYbz8ZzEH?sFo1izdu-GH1``u?*U+BuGmaFmv&
z{3qEc+Fs`>{LC@$Dx?40TL#t2-jUF)>*D{{$q&(rGNQ6n6E;g2(K<Qg_o=7Ls{RxX
z<@Z~F8d3MdaD!xu#N)SAKiu8Fu@kAO9VJ^aeo!rjWkL4tNp`lCqhuPpN0TAAN<b`n
zOb4D1mk?N*(lXK>uGG@y%*0!{N_q9FWw8(Kn+ZJz!-zEG-U`LrioI<%FLX*vOaGg9
zR%kpO^_*OZyuYhlN%Cd$+3~+K*r!NcjP3umJ)$eDCZph=of#Wk3wvb*NfiWC?-^c_
z-&>xY1Rb_&o;C*e!g<`ig@$){PWzn$rMo`stKs%}=NI!nUOx>zln;6?bd5EM(L0yE
z2UVM1-c_l|9_!a7r(aIrTx(@S`Oaw}f&Q?Dc;W+Tj2TU!t^2AZfdGZD5^Akgf^|kk
z?G={_v!bvpiaNcoPD`lVrs-p+#__ti`5auv-DkPRmrN4awkB({;kxy{mwlGJc4MK3
z$La*z?D6tN>q0q>#H_s}93F$kEi%1?+C<Fc+E~Gyoz(Vo-$+;t3C2Q?nbmKaWAmdp
zj|e~pB?ubca86Oh82tU>gUdfBzi^5we(p+AEC!(*BZOg$Y65tsm-NB)kihRC?IsZQ
zJS*wJgOe(zw5=qv{dO)+>v;Td_Sa%j^q!v5<NF@gSb?N?Q5Jg671Z8akGF#=Mz8}b
zz%yk{JH4|Qd3SStcym{cBgrm{*8WX5VWah`LheK>-kL5yKzymfUie>wj|0pKu9o~`
zHgv)5=L%qQi#+O0=zpg_F39enw!*G~YF|eLh3)tySzidjBm9=&V`fQaKpn^(v9AEt
z2G^~}9-A0J>mI6vH^fjgtcwOIC#Ui@;(nX+zGS}ro4VqGz_{bv1EK4is^Sxu4%sXo
zWhekyW~)jBe_x!eu}_YR=ljwg!?|{S^|u98l>bKGjOCS32_6dos3)5yWoK}7-tKhD
zWVpImxaJ+spK~ZHC?Msw=!1eo%&(j!&jY9|^7>xASzb1S67I+;@1yrDIu*cMp}{be
z3X+lW%20~>#qZD_N(AK!xIw=~H0@@qKXS<Pg2T0!r-Iv2_Aw4}_HQBqt>BF&AIFh2
ziLlkGnb@WfaNYR!4k}#Zw*u>5Z3rsj*GuR}3Jm;7=Cx;HWqthHdY%90{@)uV3RFE5
z4%D3HSW14}$aT_(+`$v#d8)@6X+Svs_jgcdgnBm=BIze{qL)4Dg}oOfLGS@YjF@0m
z3fZU^FL6fxMv$;P&lOV;J=_T|@_|eSssZNa<``t*_s;ip39>dIBs`3GIfyG`lMVGl
zccCBI1z$$1Oa^f}q0CG0H6p_F@#!;0s3^I-eLi@vhxfO|X0OW&9o}Xl>fYbe)wbbP
zvK0JICK0vT;kf1_Cvk_nsQ^1K09p+Wax&973H)vlbL&vT{eQo$|1l@=l+NLxmA7P~
z#&N*`Kp84L*?-iBHK>5s#Mh}(xHlPow`0Vy%XVf6BHd}N>imq8Vdt5bi$V(@uv}@4
zb|&%c)9UO}srgN<Ni!!QX9x%1+xTIp%8u71=noZT#HuOndORO!ITJl+(DRrp{=`@G
zmGH^oNgNK^Jm|pQ%9P+Tn?&%%D2XO;5fFc~Q&q$;=iYg_L3OBYt3T-3|4T>WNEQOZ
z-9f&Y<m3GNo2{W7_ep<ucXibre8UP^Y`>x0pffTuat<?a(!?k|{tX2^=uWKjQ9N{)
zl&1@LDWB-Ffk_OvP3$l|e88=V13=q!8h4Ij;^bI_4y(eJ0_r_mP%mJF#hHI4q4dl6
zHVf7G?=1KdvwKEH)Ij)69;U>C)Ix_6b)d(>l!uMv+*H4tS2_U-rXh=St}b4EA%Yy`
zuNaa)_nHJqPE|FTkL%8$3=NrqA(#BiDrUAUXtaKD4<iQI7<S+j;@m!@^fpdYHLv~{
z!-%$HU63?|EKb>mawIRc<|<xM2Yv+_M<8B6-oBd2jf(G+ObDs4g=gN6Nph{gClpMQ
zkMS&&NH&Qf^*st;bd2QWAD4F=|3yj{DSSf0`A;=Z^8>)OG({wpo_!#xs+zKcX$E@p
zUDkk!i@V%`9tXeLfMqVidD7<8I9C<Q;jWH$4$&)l?x?L3X+PBLC>0N-16H?&71-(C
z|4jglsQZQQa$V>#NHVMk)BW2b&;4%Q6t(3@h-~%mfB3>hc28*F{#T(C%u2>0JAy~^
z`OY9fko(q7wvQ?WtO(mBnPY!Wi!f99kpcjnfat(IX$78SjLBHd$K!hOMn3h27yqzH
zxqkc-WMyHoTWD@-`!)-Nd078I`0(o{sgxl(Nm#a*0XvJAUSX*21V%4N*mH}Ux2LO_
zIlqRc$7XzL*99pP>^tl0>sg(4st$FF??JHjiDWwrsDOa?FMFhh=vf`L!_-}H>ZJH_
z(8KfS(fU?V8LvpILM@Wgpw(C4y0^M#f#$)vaZW~y^U0GZfC9r;$Am4ENI3X|^s21D
zy)E1**fEp?h^OBjYG;0t@4+Sp#itKETdo37T@tXH<Up^Ku7{G-1K)p#*S7mY$eDEw
zB0d&a3Od@v+IZQRNKDcycNI(}r5_BDDDZ@`c93|pmeZ5zg?8kVsPo8#m;cDmFUQ}Z
zC<x`beYapZuA!>rk58Rb#lYJy+}o96Ut5&FbGck|zyA5zrMI3teIdgRfqv)C(ed$^
z%iANUnG%tc7hiV`(R2Q;4jyhQDWkg%Ytz3I);^pUb{8E5mWyN%&RC1leNmsUSKm8;
zT_d>BSg|=BB}>c2!7<qM-Z}*5K5t)dufjm>F_0K0yg$p`o3Cq~y^3aGFc5xiSW;YU
zXJ>arSvtLf{L21#^{Ftk+J;L#A%b5}uqEyxB`Wg6^LvjuIIN;yn{VfL-I>#?y5pdd
z%#$^9qCV)TZ*1YW_c~adJ+Fj1=t_Fw^8@Z_owR!Hm{O5EX~Qh5po9!}R)U#6tLUM7
zm-k=l)uWRq%si&xb9O~n<xMWR{?s}_-P774rkCy6TskAqce_}rRO7k^ivCCJ=elGb
zcLpzOxU9;nLs;*k_*YgOKYhCSwfTvE{c`AVb!bfzv4_F;$X#FBgK!({FWM}ZzUu$N
zw>XzuXG^V%PIS?ZZ7EDo%au9`rVBei<Vjn#8HdMRA#+D|LVZfW_v&kEOWKf=8EvZB
zRqYVHOP@Uv%>tg~H2tzhSx)Dy>iakI`A^n(yL;CDoB=upvdS0wNPAyv5IB9iDG}Qf
zB*D!O>$&X{{*BQqQ3_j{e@|X%o8jO<_DeMqKVQ3JzT8^;d^-F4kXGg(@y{wHfinBy
zYMbcR+4uWB<Jb5pIORvBN_H5A&5h6FYyDrvkc0$PaQ3&FmfP6a9I+>*riy^%Q8=vm
zAq5U`n5NOqS6T1zu(#_FxeMqm*&JWKh#S?_(dpLylzC}sUOBb%@83UAZYU3A!fpaY
z!O<K%)Y`GpQ9=Y68JX))j)LCpwXQDN%dk`dHyb$4JjXBfnv^sIiBZJ+KnXm+-uVc9
z>1cBVmr`W3as#klrdau7#%ZlcPhe5Rvl&D@5bmkNvk+bbPVM&ZD(?H<q{5!A5)w#e
zs;GD<cS%X>!N$ouv_iWWs=FYN;Nj)nQi|!f;N#)hnXXnCuwWs?yd^deu1g$N4b&R;
zmVZ%B7<~<!@b9@wOWgnNqN=T(kd#!YtcppVpch0HoRC0bpnnHhZ^gMD0|Ud;2aP?f
zy=P!xAL^uR?>I;>ic3o7AShwWbQMpTfEv8Fj}IawH!rVNjy<oiFcj_|r?9%V2FoJ8
zoIx|<M6gnu<tT^Co0o0qH3khewd>{>=gYgWgl^9vaaWSmgK&`|2yNr7Rw#6kURq|p
z1;~ky8RZW<Zd+K?pnG0dB+Jqw&53CAwI-60kbI=o0nM6r%q<pT=0}e#A!dPaVmO)M
zc*yqBSFaLMRr=^nrI_KdKOb?!+C2}K-bGL{S#VfIZfV80Do#kJ)3PA)^Ye{adi;P#
zp_nS9GaziCprC*lT=zryTD<u0-y82x>e}mz{g!PH@#%%F^Z1>Y>3JFhKSyV0$z8Gf
z?r?B(BO^kjs6WXb(TC=G-4bIV=04$aO|b^b_3qA&SPKP?jb)-@2fwZV6u^eC&*AnF
zCPESk5@g%NnekNKP0p;RL<sN@IwGZqBOeMYaVaU6P`%h-Rbu4fA-wr5cSvqqPzZA@
z=V=b`erWZtcWZdPkApBRKZUq#qdX4_Sx?G9;GKli;<wwVp?zN}Dv%I~isQhisH@|*
zc*o%`Drs%ah)_)6FoDy#Pm0v9Ir{<xlTef2QX+i(-JI)2x_T5R=I77}B(~+{<;3Hy
z1Z?wwByMxwkIyFJ0wo@jc?rEa{W5QwElaE0{N@Rw*r4$QO6=!%rN2z=#Kgpa0Ry+K
z9QU)^1Kr)EoTgGxx9}lD=^|CD$xi!(!J8b;3#uqvTifHQ=l;G*#`JQ(p-#kSkbKwY
ze{%sG=Rpqh=IWn^vGK1P($mNY^u-DAG%~zRO-;eU_ok{ny#&AX)~}P@8L$d?CM(<3
z)%8)97L<W<va;GWb~?_^T-BCJO8pS3UHpY~ynh3{27dPTHo9)&&oN8XFnp4gkd*wp
zI|F*KW=fyajO=V$PEP2!cfbr}WMh-r9fxI{AsOTYA^i=}%6U~@TU%RSe-;>PUdLOv
zc~Uk((+%i40`v^&-GJ!B#soePYtV5DIm1Ba1==YNB2~!aH%R&?XJ)QJ01tkcU?b2B
z1_w~FbH2XvZg_5EZ0vW=2MJ_X(fgpF#`<~#a6144T!f06+V3gLG&$NRcs6*9Xj;<y
z_rolRq1oU9yU9eYVhr)#H5eLHz39eAoPzikxNSLVn^4{1v0#Js5`=yK6^t`H-6Hf8
zY$qTBeQ2XRh%^@gQvz0JSo>10V0LaUP=mpnpbS__pj?HxT*=$}5_$&M#0Uhiti7-P
zJuy%nB}4e7r&GZM|J8;ZBBiFGQ3cvh#3c0p4H^|eaRC94TF~f!K3Y(F$j!|)Y;d&(
zI$2+-MurNgPuD<q4a6QCOiVHk<5u8oM1rC)EzL+<yAC24)hwAN0L5T?VTvsTxec^e
z@!2yU@O%W;69qXr@Mx?;1Ox;!hvGqVGYg?E)XMlAX07b(`ntQPKuZohO}HaI35h!p
zsx*tV=IWe~AEQ8zBywB+yS({g*ROkdabn-Vwlw3|LLuzQgZM1qw)LwmU~zE~F*rF{
z2h@z<V06UC-c=YXAORKT=XZ5?i)Y=~;evC-5R~N3Pj)Zh>LE2Luo<k1xCw)RrvFHO
zlpn}@GBT)$&w{TVfQ$yziO~Z^4UN$+Cf}j@M?ggMjwO`s{rmR_6;)NI)t*==%1up9
zg2&z-aFViy>8WXGpg^Ohu8vC~aMb-MO~r*28yg#123v?3N29;gs`D9y;t&qmSXw^U
z&~Syx3ii98)Po^xXlMYYG-Is%ikQPfvoKJxVM4!?;DodlEG{UXI0x>LXc`++O+$Ch
zR+xvSq!_$>c>uwjs)51WPV`06!x?O~0uB0TPC;HFq1hilUX13e-pm|0e=rgHY_GdM
zK13MX+c&}U!t#V6L2d<lI*?Wkw#0JObc~D^oENuDKn`4rIM~cf`=;RAX;C@!9jxTG
zAbvYLKZnDQ{?Q{nZEY!OX_(1b8%s+|nT8hj_V&G*uteAYoOOw(%kKaBbpZU%^pS&;
z6XFml=q}jZ;1?l`O-!IGINzMUL`;Ild~6mh)`ksZXytHmab17GxWN1X#|&hIS!V<t
zIyD>H&z_#p?~5O48S&5u9Z<y6Zyo&l1tX1yhJx}%yBO$iHtdBDa1p;issHomPZV*8
zTVPv?qJ;xT0WVONwxYUvceH?@#{(v=A_Ugu<%h6#VLfw!juob{nmI2oZ_n!QI5%%^
zZ${O(R6y}EwzX}9V-#3^h}Pe$y|AA;IXMB(5p<^kZ{H$X2M0fPE<<1F!q29rNXf{4
zCU6EYsbr||@bFZz*x1@)u?LJpFb#7S<S{R}bin3Rk^Y9{344v`#SXVyG!_od$;rtc
zv`j7vAaz`9?BhRw)Y&V?$WGwzvPIa?z+}V)@(~7%G%9pTN=mkPTPv$mpn`s)^|<<X
zc0*<ddL#k8q6)66e2$ve5U$9nU!fj$+VaW@Eg|O6M8RDkfXRHa{?+Oa6dbUQVRQS5
z`#@;WP{TXn5%{ivU8;65aQb|~eaDX13r^dvM}dKXBqStwD4fQveoITH>gwu+XM0Oa
zNC?QpvU=>StW+2XS+_v~2u!jpe0QOF;C7c9wM3*Y92~riqJ?$bwdxFW0S%iB+5q2b
zKYu;}++lyQUu$dcarCRK?rJ9KvpGVHtQq4ujEsydDIxJTAmEW=g0Z9H{@NNGmQApW
z{6D7Q_xT*8%%PX5SYy69Lo-1sP<Zd=mKB+v$fm#(L>OCGbSxi&?Bv_GZ=XMZMt}Ru
zxgKbwaQ>Q>YjbjP!j25%>wSJC{`6g7U_xSILHyK~Js>5_bfL~e6ciLGDJiO6p3q16
z({NnF1_gM^2@u_p#Xxod7UQ`N?!W*XGvuM6LH0O>?GzaP=&>BG8-q8@gA{@4RWl=L
zZ||ma$x)>1s8%7%pX3@&z{K)fgO)sd_6%_VoUvi-zVh;NvYZTU1#!F|0K|;?LbW!X
zmjnr6(BP`duAWR1LXCo`Yi!KIW@l#)=N69u6SDv`^%+;>N>6!Su)(m`L`jOd?FTEf
z@H&D7uo-YqdiQk~)BsiY#=6X<3DD392%@(q%J>=+E?_Y{jUd0}2UKkcX4v8t_&8v_
z@UXrDt|P-Mq*j<&gXT4ZzDZA9=rz-M{Yh`Thq`T#-x7m(3oMM@THt{TK~@$`tg;lu
zHlPCdXmECMQ4~LAgMD#vfrf^rE)6u+#H6H5Ypj)NhObRcL(C6=P6qNk*a=FmjAG?;
znyj#jNMM|Kd=3u|(i@IJ_E7GLb_w=B(_y*U*(zvYR*}@;1WORr5Ad~2(ND~1ejr^)
zCpF#5r6EDFINq+Vt$kXesSyo62|6`)q~`=@fnYm*bCi?Cwo-AZ+meC@aoRv_ZEaAz
z%Z=0_?IunyP=`4M5eJ{g?-!$8T9&XAI5r)}dHR>2uEDPI*h{%!CJ6Q=nGS-xcQY25
z&;RZ_!U8q4D5|YqJwgxfjILlKARq{zk!R+JrmZ(_LqfcCbJG+(gH{&uby{b58?RNp
zY@<N8v0H?*s(+6vz8C0LL#T7#kdcsJDvsW(vR>KzM4OnJT6%>gKRxUehZSCpBxYej
zKW&v15Anz3k(dE7cfP1J87U8h8)El?$!nVwd&3fr%2uR@!pO3R73mh&-O+X=F#`jN
zElNH*At50Vkp~EXqau7Vrm=OG78i$HA6$aD#AyFqts~_$mC=G!k|*C?#NdQKSe<}V
zNbmX|N>Lw{SouG{e<wWvpv}~jdivz>FoPL(aWXi`uKId<kPtkUxr@Km)YPtkwTfu{
z1EyG_zNNiI%{OMRzSmb46I?v~xKAa^f??w<%<CBuK!Y9jlWbrd5%?si^5M*>+u$8T
z7Gp)j%#1Y_u$Nf7^;*jt<?8Iux<m%v<A7f|%5}j0gmaij-dJTuV}x>y+u6ZN;2Ka@
z1c%8tlzROKq@>-;hu~{f<a7!lzy_0AFDcKU_MBMtN8lwRhbxxH#1I~<<TY&+8M>z*
zx&z~_wDehL`^T%RD@6MI?lXD$(8$O%sNk}-+epszadO)_REepn;WO2bjj;otNP>MT
zqaB1!C5BzFWOzcQ?ceb9W%g^@s$+l`Q_gu$aPEG)bP2Zm)FVIoQX>P|`>HXsLtFq4
zDIvSOdc~V`#lyt~Qh*V|n1S_VP??jG-a<rnj&R)-rl*(5np@u3nA{pG_yd?gmbJTE
zW>!|U>hqvDP-A%~V;8|j1nPt_XSq}r1`XKcUHIL$2*-3r(^qR~&KB4r;~1a}T9#`g
zFDZ#UmLX+^*wT3Zyy($%dI15ev^4b)96Uqtk|C`XVg`2cvpsP0e?BJGE1_h4KC4Pg
zM|TNq;H?&j#Ma?tYzVXY@jJJE1C;kNavW*}RhwHLJXpDg#zpHK5IMp!Jk)$X9h1x5
zh#&G}csMF<1K7vw{mDV^h1~q(804bF#l_(y-J%<vnCM+~c1wO^VD;H%<|rXY&3Z&#
zTp7L6&;q6_KzxrQZh(v=lQOk}7<sy}VDK`{P^n6~ye<qbI;3ubc%k~EURh~ocz#Ak
zhAgexzBun~;ZPVckYr4Oh0_ftQbajCIVoFz$6o6D_wV4dGU<#-z52`0ZMv#-bY@17
zo7)A5$>0oi1zR$BnGKKfR{8f^s7qkJlPJ)rAz$_oG6Z^8zJnDOV8-EK^j)e}z@Yj<
zm-Sw?X?v)jpC81du<>(vQQnU1yw78WS3)M^FT}!f<4tvkka6^~nYSh;Az@>4b87T4
zGa;s>CHP^#0CgSkcOT!+WRQ*0XelWP1=9>J9-bqJV73UN(RW9D2W7dcOVkQ@-rFP(
zOZYy|QG2}0!GoQ-^*jj2`dC~{kf&NX6bfU<MQhIWWRmZ<#;~Tx4j}BiQ6KhVArzIA
ztV1#jJY#IKu(%RG#Ic4*WkOwIZDqyk-%r!>1XdEsS^!DlnPS{72H5Y&Po(5>#~}P<
z#2^1uRscG~%Hm?iVH`0u)M8|P(vEj4?UtW>PLzDkXj<7@-9js4<H4+_xHcKRppF>*
z<C;H<i7Ap6*&q&w;T97_=9;Q*23EgM0mp+v=qNca4^Ot|qw(?aE&r(-P{UM@==kuE
zl*ekYXx!@1T?jj9xav5oG(RW*)2AUg^Iz@H-)MXwI_Ko@l$4Z&=uKXhDYjy2C%1_c
zn@45{lI0(!P9q)q5JE$UH;8@^!>A5$prxtlTY9Aqu%rqBJfvQKzSzFV41j(Mi}!}3
zV)x!s$$TK;e6^!}Oj%u(0~0#4wQtq=lVo0_d}^CL2q$qdF(smLpu`5)qwI8lKkJhx
ziJi+$Hdcg&I0oP_po?nv;zb}APx33gYG39?fB#!_fuEy1n`J9`C~STo9?B^zJMuE|
zDagvA(}n8Bk)2Etilng<VnXnzm(qkEA|)as;^5#w#$H1%@}|x+Hj1`D3H!$P#q5rl
z0*o3Ren@@&6J$#(Q`5|4ujBCWaO)}j{3c-;3by#7{__5B%ZIx5D{U<n^@KcpfwpiM
zitOhrxqZdt<Ew>$0pSO)ZK<#$M1M?O9v!U?Y5K3#3nzph8H+SC!^YcTth*3-YhK4g
zv;fENG%-2~xWjdIS%K({fiqmb)Plq57}SA~aNOWJXuLP&@pMAQ{3AO|3;Hm!7i60j
z{SvWddzjF$WbZ&-neQ{Gw{L^1RAT{FO?5Rk7MA1;1V1jWuGunUNjUFo<>7D#Te@Oz
z<k6m<LeO?w&CJg~SRLU?hrEV?dwh0w_S-k497<#{F|k%ifehUU&5<F%GNrPyusG@G
zTTq#a*#PVTA}Q6Vmh&q(j|)kUepGWv{bMq8L&u!#pT-FKu3Dyzs;p#XmG9}08yOi1
zeGu7cyJ)_%CHK;fw{f%Br6};-J7Mp0LM0F0{tgnZ7wDQzdwY8~o_KXmG}F0EIy5cH
z3W|z~JG;AMjo64^+uLxqjae>y98p=w1F@DBA8ICy96VZ<#kRej-AAu==lZVA$Gi7b
zPtVS<!B4Nvn1Af$XJIeiW3wVwpSiQYC>Ahp@87@A<P0%9O!2A>;<IXijow`B*0ydG
zKn|z-)tyt98n#u-w^S-(!&dt)_4+&;-rx(Ur>E!Z>&rt4Gxr@n{U$ycnE<%USM5R(
z&c+73xTUG@Cilb2hrw%aUxh;8jT{krNmxh-6t28*<RP>hTnixfgcpn;x2$b!K-9gl
zx90$f2RPC}lsN-P!|8G)pcpsCdp4i&Q2bNH1O)}Feg_JC)>^wK4G7{O5(nYyu?enQ
ze&+-?QYlD`K*ytog{{LvT<K3tLqh|VX(bwfFZ)x4??15uHUhXM&doub>mf7`i4Bx3
zZ_>BLMJI6iB#ea1knc(21*W7>xP60=**Zx#vKVV@3zUi?mq+Rl<3QvMX%ve=LrhyN
zlAj8Le!1y8n8t9j!`Q$(1U7bd?2L`^@bM))Ea4R>i!VQfhY$5vS65%cxj3Zsv<RD&
z!~MWmB#n}U1Xd6i)GfCP0|Ej-ciOr5a&+bul0U*@2IBvW6jG7{kd;BkO@oIH$$g+e
z3Xo9Uzh3`a7wzcifD-}i<L~5$&t#Avr}<tb>Y}yzJrMQ;Lufx6k90y$GqYvbPvJE)
z0{r|cq{_E$-LhgO?Wsh*0S19cqSk4H8`)V|Rkbsm(+ezYmBi6+58hJ2Hvg7f<0js1
z%Bb0a+z57tPj1!IjE-^8C~zct4GAZ#qno{v!x$ls2Rats$B*Z#tO*e5G95j!%#W0k
zq$tb*^4i>d0;xdwDCw>=*j@TpOikInlKJl=7;iavc~kiuQ|s$JARyyPg=FUz0^T}w
zMTxiq_y=|X&W&V@ci-RveBf7N5wlRCBN^auJ4!}Qh=lZ$7{n=rv@iYm@uRA$3P4)u
zO+rFf$P`;!TY;Pg8gU5WhX)26Ve5q>zPH!*cx!Cbf0`wJSd2azs@*<N9X2#HynFZV
zg9i`bXqL!;tPZZZ$*@pSQ6aRMpPv_n5r=fy-@gsaTXR%{dArYpL3=bfs6t7Z2IvqV
z6*qo$bBz=I{Wpew@G6KKuy(Sgy!;bYFnqygzfk8CkZ1+i2}s8BXBPMOeZkVNstL3k
z3qBXSP!l(T6&BxX03Jp#f8cYFw}4f<7**XHaEn%zGB!4LcRvMVNst3pIWCAidXxtp
z9v2t){@uHqm<Q6UQGtAT-xR-q0N_+tps67%8miJ^la6`;i`&i3O-pMW&TO#t2~eXX
zhS|Ql`WPgw5Xi%ON%E8_ZQOo<9t}kIa#3Kj`9YSp!SlG#(;B)e^Au(*m>I$wEAruK
zA*zR818BY_fD-^<K;aKkDFxUQ!RGcxH68UrQLzuOOh_Eyy+g^w(narAo$IBjpMbOw
z3e;d{$O$_H@P$v8!XN<c=<MuW{e6CZzOu3cku42P7T`Qlw6XH$kPfqDLRpS0k%gUo
zA6nPe)&}ZrfU<zTeuL6MTIdVHsMqrzPpR-uPfwvT_X{Q-ROQLw#FEcB2TXcwpN3U6
zR$ddZ$CYk^QJp*3*uu~dC{03O3a^%^g0LP26A&P%Btmfc8k7jMazNyAsh@>!A)-oR
zdfL0HMk~ie7z$}MWL<FLdjnDj7^>Ir6%4tkp`oGPcy>{cnrCI1!Da+5X36pKo$G(#
z;)XY8zabweDwCA7s@IoKfz>)aGXoxdf84)9nwinQ2UF*^AGkrW=+)fW<%K?mgdGr#
z_K^{I!c-6ls<3B|t*UK}j*o+S`C;V(v=i*;z;RO>mNaapfDC*fd%Oko{!^F;kaq#h
zi;DZ8xcPx%7Wu9;fc|`TlYXgxCnqO?c;Ew0xBz_|93O-0IaDtIOM@E80N{THDk?7z
zm!vH%EvsTaC+2G~<*=x*5QC~15~M2tIi}OT2sq9PSR5spSA`?l8yKXV7hxFQ4ro<W
z)zqYmm2Wq70@9j9hS(z`L&FGO!>{iCskN{!b32EzpA+JTWRC@{!%BcH@bK`EVR(Zr
zDJqJZIP9@ccC)48CHxD5Bc88!%v9_3VM#_1H&JOY5qV-!xgo{0SxK?5BD+*Vk_E;y
z3bgn~mXeYI=ejSCM@CNHX&H~F!|$0&FT(y{TL-iM_OdVN5w9hxnwy*9+s%F8HJqKD
zZM^DzQIVOMIl;gb-3{++l8dq<LmK+7a5;5kXiJ$eR!~;P&d8`*YgnKGGBN3QdOfX4
z@5GI@3N#?Yr<H@3U47tsR-i#fNx86d1@RFJv26lpNo(;98%sX#xhQ26EgYhH-V7wO
zS{miB<YF>C-Nwep0hexhr78u<L?w>+{VLqPs+OOZEt2soW#AT<q0(iXEz!GDW>HXD
z%FD%N$}x(HfNQ5&6m8_~?ODUxZ62Dh`97SUoMedZ&ZlELzpKe`)0?YAR%QQPcQ>pZ
z_JKQVf>j_Ljw>u=f`w<%F;MrO4j<wT6N`Tzq*XtAjmZ~Yf6CWqD=R7*Ngx=D>gbTM
zk7`Mhh<f4S@#OI%c`VlGmZY4D3N;#WDX*>V?ZmS<2-*Raf?J)|WJcVoB(?3x<Vq}t
zUDDbi9HL~NuTI@Yz`+k_`-j!)PYUl4<4Aagg46fkkE&>18g~Fg@=h*=KQcFb#+YLl
zDzuz8|7rd7SL+A(0g0j@5}L)$4UDJn@^hHMr%^{D{x5G;mDQ?<hZ5kp`;g?#fvQ1#
zdH$7hostyB+ribqWA?%@R?gGbv<9|H<Z{+(J=G3P6oJX8h@T%1%PD&OuG?!yBbqz^
zYYQ~=_<x$$_W##9;~R(Ql7S32GSk<6XDcc>ufL?_kHrOYt@k~j$#kw@OAF>bz<tR<
zjp#Z*(hHG)>LpTAEMD>Z_q|GyCiKgIhU98bW3%ZDY|%MpSy9SQ4n!Na2Hv(a>le}O
z9upCti5hW_cqW=JvzYknZGPSxdTSD(AFiuTSG4@f_smw!XKAxi`X_QjqKIg@({|3=
zVYEk)xw^;u*5AjjoQWcR4LGhF^PaU`?$np;D#Yo})G$(W|J@7e#TGtqOu|PXUU(aN
zvIVo;vDPTZ$?Q<Zd2`}@6!!K}H)Ffdb3SI|T}`(ybOGP;OY{aTRy(CXce|1OGSBet
zV0F>^y_o%3(2Z^nYFoH$6sM2BTW1sJUE)*Nb$>lA>rWmKVpq8Oqg^O1Do~RhkpD}1
zC9cG>63dk4>dm51x?uQuj=2Jq!v>m34VHhs_-QkrS{)F<dW*+NYD3x5tNA2UY$OhW
z=n6%v<@%5!K#HGzN-x<0H`;19N^TK2Q~b9wyV&@7-kU<NI@CAaOku<C-d_hIT|qr3
z!z(Igc`qg+9nz7*e=xYeIn7$K#e+6D`Eq{{-hN+~-6^7u<X6~`{-xr)z0&gCm@|TE
zJM2&IeH+ZMLOBFtb-4)RnF*HVn>wqP$EzF8QN^+?Z)pDo1$U0|R1S0({&3eBWQY!O
zl4e4DjJ-$rTyxFR{N4o9Ex%&MzA~F)7M4=FTZk=6eV_U5ERPY<)EVw=HB(8=Z@cG<
zX?@88E}OJPr(PcNyP!BP$L*KFE{Rkwet4_v{V<YKHUA^#+G6qBh@PH*_o7gpOmt{4
z5HF;U%)1ArT`bs=w(wpbm-C2j)^~nrH`yOZ9$mGThzRYdW~wkvb;aI{rc9njN4(e#
zd;5DUYNz<CJR=4|&B@2wz`)7m9TGw;_VsqnP+fba1-nP*!CJpQ9Ri_4(XwAe>Clv(
zRS=!8uW!Qn;v7!$JmveF))G%O(e$En48Pj)d3X1bhQ%vyw?&V9c|>7XFf{CIHwmBn
zF7f$;@VwVVw9J>`wzwBWxMsnNfAW+Q1#@h*^m)TC3InW`qnTezN8cMrXx>1MjIoVe
zoemqWprfMGsg2zKx2IxSfSHs)jIF9$gF+UuktVvyfTW~{WG**+cDdL{UAZlbh$=T4
zmaY<dGUohkI8j|BVW@~-L|uGYZF3S|%*<zJb8gXVpYZC>3tkbChO^`@JAovDDSVE<
z3Q4Ykt+NT_(fem7g15A~%ctc1I)a|^7GJVa3kaC@zNgs?FWkXue6@3RY2iQt-E5$W
zvCx)^9atb~X`ZNk`+Zz#lUL)AH5$AvKqrMnZ=H{N5*L42L5_-fZ&$DOFb}J|7T4$a
zxL!YpdiNA+gUUaM`H&F#OreD(<i$Fz6;4&9j72=89KJ(%4{sHQnsN@z#>n{AGo1LV
zif#OQKy(;_kFRk%zS-!{@qAHv{A<TapC%1Wg4g{28jS|%<9lx*G;-Q8_0U9a*?FOj
zgsLU>25PF$F>&u1)Wtb2WO0WJ7WIY7)ox!q_)b`sTdN_=*V2+j7d$s!?X**D`$m+h
z#?4MrUM)MSDi_5f5Is?*-{$?%j0dMuS&54fT|FCFue;f_j_l;_sY-iT>->yOYd*8N
zTw7TB;i{|OFsY{GB<Jm*amr_8|BvpOiG0>#Su|@#`ecG9TbqNcuM3$)aTu#y|EzFM
z`TBl`K~s|1uH5%H*Sq#sR#f31bPy~eMmqW$d6zjb+=$ZpET9U;Z4~k-DtO%tcKT-U
z(PBJ8yJI}{@{&SzAd<s;Bi<FC=yus@{%grsJ2_pQ#D+gV6Tjd|hke4XIpukJB<q+n
z-Sz{DM(t%QWf!`7wwi3yo%p1|TSzWuHg=^?{Tl1b($2n5uFE5+`cW0HTfTgeru_Tn
z%@aAy?7mkAXzIwj+VD=qg3o5Hw-CQx#;7sMk<8RJ2%=Ztk3L1KpyzzHG}GA}?W^qZ
z<_#l;SdV=GnBWVzSZl9O+LFp6UQ%$b&!uXiAU<*xM18-FpyDSt&x*0Pp<B_SfBJ?X
zb%oC@U-rVJb!k${u*@*dz|)O6)JsJH=_6AuXV{wJ<;yd{S-n~;&FpW3ma)tff8cE?
zj&f4s8sBzQQ*+_2Y6Vd+<Zr*P%I2sc5So&jI<X%<PCT~qE$Jyv(3n9`RytfgQ#Kto
zM<8ZXqlpJT3+zWfT#es*9$kp~rx_BySBtk03^{6Ih59&ClldC4^ZZd8A|BZk;ZL7E
zSB-UwEdN8yeqq;N`04l&^#kiNI0+wjtqyyJ-NF(vsI)UaJj5EP+H>5nP@qM8Tz$>1
zfE2K^DOA!h74>$7EJ{=QHr+7`j@82Qpq$3g*ut9XkF$*ptf%ajd_3~iS~->qVkEjx
z8U^((g|{<JbalgqtLO$_N_ieCAOZ^vCQ558eDC7~B^nJY5hDCzv|=@7#ivxtyxCJy
zS#Q03hXe02RlI6us1R7(c>L#ZzwY<OV{g|Ay0-;JHkiVW)AZT%%C;;8*w=A%Wd(OT
z*PfB0vAASB!9w)0o*q}LMbUl18x<Bh>)n;e-1fD=RNN|}qdm}w(v2hCMcq`e!8fbe
zUiGz`!wet3SK;m?>-Z~4FX1l2F5`A>jn-`z7X@*hqs8D;d9wwl45ZR*8JyNnW3qQp
zELe2*+IPJkcqtiGPOGOk@zE(PEh!W#op>h~7iwC>J{oHp3*o5Y;6p*AzQg0PkrOd-
zFiCptwm(%RQ`Gtw38k0}14C?leDn|Nhfqp7CLyn5Z_9?f?g_yM8h5bEN98roSk_FG
z<Q8!k=2f+New#k%{p!y_kmQ0uZ_cTwu8nw+i2km)4@*>9n1<s0dpAyXqS5?TOj`Wm
zzJl{L$B8=^&xl2pqk_{!nYY*2&KJT2D#Hc5UN5`&|8jMDd>tmk{`8(6{(;i%69fGZ
zepD0`+95<th+VQ}foXh%&MVQlxeJ0_`oC8)e0obcQgr-EXP%i@jp9|uMS|?8%`<1d
z>^4jrwZ=IfifZLUAKM<NBB|T6$L6y3ldaolYDJS*UURXQFxP!bW$njuGP{zf_f5A{
z*#AlPZz24Ng^%1LRkhD?+PjmzzDFAD+bP>-m=gM*s|(SSYIw}$jf#Btm~_(IS6mGu
zyQ>GX=wf^NsLLMIbm$rsI@CyU9V;kpBK)7w&%FK_7kPQdkMOC(DSfH2-6Lr&JH_rV
zRKXD?Zi#jhD@b@2J_J_-xClR%znrH}nt4ulGf%1G{c#j2-=mesN)kIJZuOvQ{-k8a
z{qg(=X><AFF!okcQ6^=D{)Gz_nV<QHsGW%~tT%_+hLI4>OD|f{k0%KX#SPVbL$*Wf
z87UjsvN1n8Tslg132c=7gv<vj6e71eqGIL#j@ULfuE<y>)Hi)EMMl)94+e$3MIct*
z<bB&X@?_sWeL<6UhWdE@nPH5p4wv60Pv!Wl^co+fGZy-GhpqG207Z=a#@ge`ggYLT
z=HlNjKXazM!_C67N7E_iy4Bihc7%iruXHY}<|jydRQbH!GW1OcY0`6)Q4Nt^rlR69
zrmix?E>fjK=6;kl8t**r=^Otv9!{M_5xEPp@$NW97NIU=&i3!`dzOYjG~f)M4-Iw4
zv7l2i<5O0=i6gQ0O-wIhC7mL4F13tf{H#=&rWMkVxGg=Jem+0HjBO_RM(;uWyLCRJ
zne7~Vg=_4*qR95N>x8(tOe&=;ONBCwq?)6JkFm88T!#t`A@bZmuRqydZ!%P_E6cO$
zvsB0C+QLO9iz5T||2(whJiNn&YKskfTZ^+;yGL+1|J)3&&eUP`KV!$c_x(8VoqtEs
zMXcRc&iS___z5Gld4&EfQ6cQlDo!(7F8fmg{aKHYyLW398lJhgqy<CRsACkvT>aOD
zz-nPtR#xi>>o<qxEj)|0o~k?M_vZfy#fqOC=kqM5lbQ*gRo^}<E^WqP@8sxfeD!Hw
z;KWj2a{uq6(}cL3PTC){&467UZ}UYH?Rn2tZSM$0c&~A+uv!uz`PsKQx_`9Y$n0V>
z=~ZK!>#C;U_6{Vd<*FSW?d)CKUqjVfnXl*BntVsN_t84T!9uGPM|0wu-;Gsn>1;mc
z*0^rOVz9a?Z7R)$D=!tIPD_ReSH4)_xo`<@am2;8;k)(hzi6-CR(vJ<6dBaM5J@;O
zub$wscDibrIrxRj%fAcq{gILHx`qxU{Jbi?Ldq1<N_KOC@rI&qdlqC~#*_T)>LtlW
z0#i}`1<7cS7Jrm8lUmSNe0a5PyKM?vrwecXZ#8bEH^>Io#|dtgbn0QH1`n@ZNNMON
z{%IU}|7x6T%8*aBS@lu!d84n?Y~S68&%{eMndX<nH4qklj9b;vId`ixwT3Z&UdZB}
z&~=z)0%XA-cU66DUS}Yx>@glPcG2lye*U1@-uEx+Stbj2JX_C2r?P$o`<@W5U0qfU
z#;~hB1_D7C>(Sm}C{m+eI@taF=AlyQv^-h_Csjpjewb{<C;9KZ9GyE|OWv;bB-SKK
zx%{1nQlF|xlo-?^6q7MnSt#T*5pNl);poRuljEE9-ya+I%@t^;Zh7`889QOK?Cjny
zGwYG@Lg$_6>S>FcSF=iy*0E!2iOzZ)e&^Q&b?+MH4mk_<kynZR!%+K8BL?_mf$^Mv
z3jeBs@P9JV5S%ainExfLd}#llq|}usroiZg;>xTUlWgxjyXa@R<=C<^Hu!ASeF4+o
z_fY@H{Mq?8P8lJ+&NF%Xb(dv8z5cYf?D{;t^klDm`T58Sj^Vejm0q|qxpKP=^$9Kr
zLb?#k$On-hm{$Z`cl#`ndDiBhIPov3Nj&Tvsy;lAh?n-zo5(ea;P+XR*N>R(=%4yr
z^yc^a)mj*l#{3gJop1e(?koC=IM&;H3>%1j$yQ?loH|HZT{&20y#$gm8c&WZRC7L`
zaxy8OpROc*iTvA)ryCln<XiODYbG^8J2>)Z0%2;d#wtaU&{A}7tV~Djy=%vfm4GKb
z711s-r46hE351k5SXbvB@$u4kt@R{29F!lvvv_>(9x7|S25#tcVrKTHYTGtJlg5fE
z)TK&F+$2}O@mfkq-<f~Kv7wH%;Wb<cBxAn2C0smZ?R))$^sDn<O9G|)C=)({LZ0he
z1#98NeAoQ(%u02a_9wrv!_cNH(l3HGEB}%Y95~2YUN|IhJRUU}4lb9rco2Hwdr$M=
zB+0Du>2bw-o4geoBT+_mY#;M-dBm^q)1?i*j9d=ibS~pHc8=|B-eSBvYeuMmIVc7N
zad6!`K=`#UCTZ)EAtTxrEBrO$h~m5NKQL%XLPZR^bzEFoG&lJe)*TjM2;ZA{wRL^9
z#X5h&-ur8mZWn#Vu!onEo1uiRu||Uh@g=3fiv$_*Qn;y+IBn>H$c*FK;s(&Ts(z_I
zYw&?Edc640nxD||L40}K+LY7t1{M*D+GxV3Ww;m@HP>=S&Ms${CD*UJ*JsCKAE1V!
z*&@f)YE^Kt(-hsBJxlNM{A^z^_J|Fv$K!6m23P(`yG*K+H+_=@CA$;$Cu)DiM{>V?
zHJNQ}^x^yW`+dgJ`PCDnM?PGQ_ddBlUHao6)`s=p=Qkck)U;1TP`-9zYjFB^6!y<>
zXzKVqE0MwV#JSG4a}cS_{uAS-(=FStya2OQR&XXnp!R<2|6aFJKV0DW+BQ`w62W&%
zZ?vhb@uM`N4d3JB_>1vfl(veao$bvPVGC2rXKj^ghAr(2Y?LfZ!APChwFP}W1sn&d
zhpX#mDfTL=Y+1Pj<Xyebm{;7D4Jx-piRp-tMALqD^xwO0{OJ{k^+9A>W0QO72B+7b
zjzhLO|H!tlG6&3000$C|u{OfiX5n}8$fz;e1vtpHmDS)5swA$&pxG&&8G&%i1Z9AS
qS*!5>Mu^-Q{%6VlA7V`ZL|u{C>V(^S;6-!@gq-v<sZt5!xBm-0tT#UZ

literal 24266
zcmZs@c|26n`v-2xzFqsy%vd9&gzRR9M3KtAWwNgg*=1iYk}O$^h*b7{Cz+IN*>^HT
z_K<!1UG@2Vf1lUy_4~uT#&PdC&v~Br^M2mv+;c_U(Y`@L$w5g#KtOX-1AUi(fCwZY
zAS@*(hFZAwt%e8)hzai8x~~TPN=QgZL_|bPOnl+O1ribxQc_YfGBR><ataCxN=iy9
zDk^GfY8o0ET3T8<IywLV=;`Sh7#J8C8JU=vn3<VbSXfwDS=rdw*xA`RI5;>tIk~vF
zxVgD`czAeud0{XZ91cey5PW=m{QUd^0s?}9f<i(<!otEAFJ2T85fK#?y>#i4n3$Nj
zxVVIb#O2GEB_$=VT)A@f>QyNzscYA+NlQ!1$jHdb%3i;IT~1C;US3{7K|xVbQAtTj
zSy@>{MMYIr6^TTmP$)DSt)`}?uC9LL#tjV(jhi=bYHDg~X=&ZMb?f%++uGXNIyyRc
z?%dJU)xCT7?!9~W?%%(!r>Cc{uWw*rU}$J)WMpJ)Y;0m;Vrpt?W@ct?Zf;>=@!-J&
zOG`@(2J`UYLn|vQYiny88yj0&TRS^DdwY8a2M0$-M<*vIXJ=;@7Z+Do*GG>YVX;^@
zH#c{8_s5SPdw6(wdU|?!d3k$#`}p|y`uh6$`T6_%KY8*bARr(xFfb@6C^$GcBqSs>
zG&C$MEId3sA|fI(G7<#AsHmvu=;)Z3n5R#l#>U1zd-g0YF7EmB=kf9J2?+^_iHS)`
zNiSZ!NKQ^pNl8gfO-)NnOHWVF$jEs4^5v^nuQD?;v$C>YzkZ#aot=}D^XAQ)+}zx}
zyu7z>-{$A%zkB!Y{rmSHK71%BC@3r}EGjB0E-o%9DJd;2Eh{T4FE6jCsHm*0tg5Q2
zuCA`Bsj024t*fi6udi=tXlQI~Y-(z1Zf<UAY5Dl^V{2>cr%#{S+S+h9+~?1q+uPf}
zeEIVA>(`Esj&I+-b#``kb#--jclY%4^!E0C|NgzNudlzqe_&u>aBy&FXlQtNcw}T`
zbaZrVY;1gdd}3l^a&mHNYHE6VdS+(k$B!Sgv$J?Ser|4Vet!Pv&z}nm3yX`3zkdB%
zT3T9OUS3&QSzTRSTU%RSU*Fi+`2G9$=H}+s*4Fm+_Mbn0c6N4lcX#*p_V)Mp4-O6v
z4-b!yj*gFyPfkuwPfyRz&hVvXPoNy2c%)(EPC!7#aQ;KsKNWCHKma4SiB`Su{c>%J
z;-&V`5y5X;o2MSDV|xNxRu`eq`<mfjn<G}oyfnXQi>Sc`;OJ~8u~XxuumtI-@>t6*
zho{%i{Cb)eC;fV+o=KMYeA!UQio-56_D=2`O`l{EYlKI`$Tc=7?qm~1hP*=vBUSIy
zGiAd*kViic-@k=g`w~{#N&bYyuRP&{hDJ62(@+_HQg`)MgITun%fmLek?%Z5Bgwl(
z>6;j`jg={Y&Ln%dBpZotxck>n`>f3<gHL|6=yxcYW53M}A8^KMn%=V~@G@l%<>K=Q
z-2ln)yN-U>!!$3Q$%{HLqz7v8JAL5l6S#TiJ^%3HXNv*P5g&6=w#~;3hxIP#7dygN
z_G(Sl3@(Z5Vy`3RH#*H%)u^%0BMvb8Gr{R_X7;PO@ZGhi5l>?}{kJS+vsDLcz*ogw
z9o6yv>35yPIL7aXvK`L<%yKJ^>1z7%X@QH3&+@@!C;Z97MVlu2MCYXg?zDgr*t*zb
z;-+cY2Ul|u4<n>-*YfWC?7cG}Bg2fO#`BUZRzCGVRIBdnXh59}tnio|>T?$rpZ(U5
z4ej}q^$UR$Jc)MU`nDNMDp~g<U~s=%r8_Q~nxh{$3l25H_Z2>SD4zyYO}Y7T;~#w7
zjMh(i&8;T>DmadBQGtgdiu|mBsO;jlh~CO`P87QLsa)bUkvrRO5TOyr9jLV3@#4J(
zf0~qZAn_t@KBS;3E0lx5lLN=>QDs3!Tc&bc`~Bit968P`Z&8u{b)9a!C5i`A@_doL
zK3-z}9WAzm8DD(Kpfbx3*~11%ECdKRnBZzyzzvoonE<8LR-j}?@6o;2HaFDNX?(zf
zA3?hn51W~Yu=jgzvw%+~=CV3SSU7;b5f8Wf2Lwhi^Of|9$;SADfG=|^DMFI4?HmO0
z4PVm{&M{?f^$&>Jxn0y#`Sgu~38lNttvofn_{>H=mm^ud3s2sexQK)xn9$6s^prW#
zugOVt)$W9yV-G<Cd;4{$<88;ud#m(^udIwO_qI8w4oxV~$xv(>Cp3v~MHDToH!7}u
zyXI$B*d!vTEg}Xt7Fnxr8(@waS@U3BqY}B^n&~-axy#+|I|P09<4ZL21k9>O1o$UD
z+!Y@mD3+i66z!J!B+{B^?eY;VBP%Q43(vLaCh<)7H?V{|z~?&vGaAl>j=Td{w|{^|
zXTwB5j4()<15oA&FrgVJbJ!?zLWILYenAaghRkS5CNzmIHHoe&6H>J@EJOpT%7jwA
z4}{+VxRI(!>;)mh5h3sX^@9`!lkWgGC(D2x55|@)J^g&BbL1o`I5Rk~AVb03$@Fdc
zyf<)y0PDpZ*50Fj1BEnv1>H`G<%F&G!)VjOo(0uqGXJ|~v%(MX^;9^#sVaq1Wt#I;
zDV*F`{n7X9GJ(m>a~pBaL8J9<QK(S$*SyT=t4wG~7h;_8i)h}}>sF3@1k%8{(TE>_
zS6A^1nMDn@eTSPtij`MTs>fH^^Dbpo@eBB@|EW136ag6?(y)LfkDGaau>Y!gE&I{M
zI4v0V_9a_V5UrESQSbUp(89|F(swY;r`HT&;sGB7RRVu)W8$AuNCwOWUHG#b@wwsq
zhdH|=rMoFPH}F+VCTPKl4+(Q?_67`V9%Z{9*dpa3KkVDRugn$pGeagct>bA-hITIa
zjax^Kd9kew$-b|wEx2nxB0jOgP(1O$#k5jS?1KT}Yl+8M#^euJi=LhbbSc);FkHJp
zP+df9^E{sQ*l_Us{DnW?v=(i-%zY`>Ce$5r#6Ns}BuTaRX^=$j=~uc^$tI!pYxqWZ
z^A7QNjb&c7+GM9QW~kT*Q`lPOcsjR=pe_?qI5Aanj?pN%`n@<!<=BnecctJY<Ga7d
zQ#<vD5ZOn&Na`~A$Lz^M#Hd^8ik9oK^J?z;+r;+)fp!iq;HcL)2<{o$L+ozTmB<`-
z4k`|Qz#_GTmrws!3f~(SdNUM+630g>t1^iy{*;tTT*a{TY*;kC9e8!Y5+s39pXI~)
zAzYGQ#x)NthOkp2Xh?LQ5};NGyh=&%ww1=`V1xcR4Sx-xiNQWNg8V%Ai7XF5#p~cQ
zTYN-_aJzH(yJ}DhLI4w5OBm!|M}vyzKob%B2sDU7;J!cZ9KL_CNdFxBMAT&&5acxT
zNZ2q25JZwRsV?L4&oL)*`!B-r+F_$c(5QL+|LM&il((-sM=AXSQ0L-#cjuG73w8Fs
z5;F81WY=B$f9d0ez|o$Y_X09cN96p{^bk7=l6c%J@=fgD!5WGHukQ|jq4^P*_-ow%
zdx_+K?YZ&S9?1NO|Bk1W^LLmUX1u3y(Rp57V8gfl8|o1_bUs@IbUAN`-g#R*G`zEA
z?D;fvNx1OF=iO>R)08gNc2$!fd38ZjG8lc!DMSo0=d8DBux76-xmzzvtJine-r!Ac
zKVE!n1%qFO_+}eel%k`p*tk60NgKE<MVFG-kKwl2`S64XFG0y%^y4Iw;NGPWxJ&j>
z3YX1{W!jC7ZdB!fNTtDN+3wJ?u96uh9A^hV14E)SeVSE)W^`4)nRe;O=h1><xxwrA
zKbysCHn3Qi#I)FyG<Z*)=3{p&h0jyMv9fJdKKHY_R-S-d7wY8`o?;6o5#!c~_2+t5
z4Y^R0>IHAkhx09w?P?zQP$%*t`O%nVZUkq^X}TF|Egb?OSHl9C<vqG`6LE?$LZ$2>
zcK?xEWPdYQ7FHc5ruaIDv`lKg&ixtv5&<(hs*IEh7w3zV6y`Q&q5L40!w&HvVZ<m=
z^=_9b#bIR1mm5+41*Q}2gzs%bGU9T*9y6`K)Wr7@$Y}l2Wa|{J+H!jz>#t1`A08sC
z9t4*T*YWM0H2(Qgzj4!Z|I7bjxxKVvZ+l|-SIeND37P{MDma|Yf<tL#`_mUvoUcQS
z2E=!Kr9d3<s?qk+Gv7m3N(2>@6d#et^bclsaQkil)57tib0zAJ%)cQYzyoN12mV8V
zA(%zPb<XxtmC%jkKTQ6C;1X_!VCEclXyXU)VnW&6e_0h?OY}qNzgLV9JWB^wT;B*l
z4NQnqAcg4o9DNM-gpvD@h&l%q3tqI8<s6&m-G~SGp5s+etQ25vV`Twps{0fE=8GX1
z^Fy>s?LJ`LPeg-lf!e<1k}%;n=-hyz;<R%oslba#WzRGF3Y6LV=UMjm3Xve!e?5LT
zL5t!dc>DhK_e%`#bw0a97CUgJz^vjQ5BwB=C-xsoIC=ix9yEWIK)v3zCybC)`J2o{
z)SK+*6V$v1h-JePY8m0Ql}M+`@Z`TGi2%iwTm7HZrfnvMou)m-o4;(nEcrh*0tvFa
zjl6Wj+#wp@k~D5`JR_A_l#)>)(S_)gVAI)Kx()SoMixeR!Vz8_(7Y@=-W)e5utd+B
zWfk+)vHF9f<ZdODMZQDk%xiZR?=brriHOZYMPIzt+^bKCE~*Ryf1OV)9xB`Rk_J!~
zzFGAsi-PWBt+O!r2y@<@`&B}toSUZw<zr<%gn#=Xv{v3{`NCqhhiO6$NVMIRQ#r1e
zK+7Md@%)xm;7J<#{-CvqX{CW#vsFp-<NbU2Qjv1`A2i-WNha0d15z&d)XvM(X;}t&
z=&v+=5fryuXx&$CCBtupW9|Mw#eye&)~ovhKEOY@JdiQH<QaYU*E>Mu6gyzSD~ZJ$
zbi^zaF8nvYXj_P3^$;yz9exnB&+@bs>C~9^*URK2eGHU{D0zCk#@|RljHLUI!XALV
ze~Iq@s1RGpj7hv*d5%ux1tvTt+nfJJ=NXhU|I5{~+W!eJwh@Z`Ij!R_{BNAMVd9Zd
z|3>*ChL;k>z2*Hsbc#LvS0+ju7KF}`qnUT{uY?i)KTU)Z2CqD%Aq8EPqW)g`uSEP;
z1ZgmW-O~Sh%Z3>vRr#UHcCM%(B}AJI`}S1tTr3@Dh3~7AozLX|s)Pb=fz$5hd1brv
z7mWXZy6m(UX*x{*KVAv@C$%W!1p$-7{J**Rf0B&P5Z?YJ0c_@QjQPrDqSwXjH=bwb
zH)ZXomxOwS8hW7x5j%JP%Z1sTrf*mtuG2yLe;+xy=Z(=%(x|pe_p!(MPH8e`C@t;n
zAD5dFwjUX}@Z=~i%#Pny%`G2dIfr?vntu~xJzov@IEXR__g{HP_m_Pl{#80|y0V{S
z{{r_<4?@n&<-{?cVCcpZo&O03YQVye(pm$&XKe7)QEDhtjP{Od65F|!RKP_`is>RL
z{HkUDCKg1FTY%P$Dzw3RwQbDbyCKyIr^8MY|6I^B!}PNJMIGemIYDjx-CJoVO^RdE
ze{WSKFB_6GrgQ^&6HEX1)-r<<^EE!>v5L?Cq#r3*;;GKgfOtk_y*h^fTMZyhJjsRQ
z{JHi&zYiAIjSwsTZ_8xIx3jVTOzGSxs<Ke}T_@~c&x8W_(2aq+cTkhgVTvYPuS0|-
z^&uU7V747=LC1Sqoc%wIKH!PSyj`+w>VMjxw5hG6`t}^4vf$mG!$L}of<)H_-_M14
zzfv%b9{L{}zM{Bxe|LyH93TD9Z1jN7&m;FcE&qJ(O8l=akcgj$Pn#VLY3-UG5m>_o
zlRw3X^xgF`ffXBte;2*cC1L>bJ}!1Bat~&kAs-3!3?9CYn)ITUQQ9vt`gs9+@gS{y
z;I4-ja3pT{RK4-81omRlR>Y{UnAmj`7k6C!X$0*{TRk42B(Cf@j}^CJICOi`QAw(K
zi;;J%BK&y-8_oE)QL}B)iXIBg<CPuqCqwIvdEe+U$L-JV?TfI2ZCOW8oD;peAv1~y
z?hdC@^^sthKi0*{49LmjzABdG*GqSolXF)qOs_*U`;mbo?yJ*#Ek)eQ69i|KPa&$I
zSZ$h6F7kb(?7Ew2OX+}6OhD$O@3n%*Vf!bp`z_lKXg{%Iu9%N-H&a;p$ZUUai<QU-
zn}93}9X*-b?<`%FGbZ-v(i8$q+Bi2)zI~DiY-#_%1@k<p@oy;2KajHpH(I}r@D^H%
z-xdN3Sn!|k(LX@P7Zvu?O0LI$BWJ##&_W~ArzwxSy;s;+;BI6$zSeZtCXXB6y8HSS
zg`g_8?3PSHCF<7dEqJ@O;Z0F)nEI#Oh$0TvitnFFRC9Gxd_j`H!)vU^iu9gM1g>S8
zH^8_qBt>9=5ruSCuAVlaY}i!UlzhjMvGlVC<6im^v*UJk+X9OGI#{$E7(noYx02}+
zUGjC+Rf4?}=493YbK+GL8Cds>#g3_eRmxzIU^68uP(MU9jf2I^e)zfy7p3=SeFE>s
z%M<<ARgPJ=a-6dt32c?D&=_uPc!IsCgUMH#Dsz73;`aLEa8L4JxAEu;L7DVUye^S!
z6y8nkC&AxKK1M*@+OnA|)2R@DRVI1YGKjkL8|e|$qW<aG=5Nx1$3dyJ1qZkOSav!e
zWbb$zFJ8c6%CL7H?R|c}^HE)bZnKs7N+g&7n0(wGFW%s;WYbK0p$(W2a}DbEtgn6?
zj^f5|KIijsAf`syg&y3ECB{|Vp?J+2X4pKe4UiEjX*Y1begY@$qTOs4{u-N1K>h3B
z<GL>okQIETPR=Tn^hK7$Ca-)_CCZ=ZcTR0?x1*4(w&)rl@iFUHD4!!Q@)wN`xYzdF
zcf7RHl~Dufw9O+n2A>Wm+x`L@m8Lbfe&GoUDG85w%NZLyu$F!|^a65)3$HT0FF~UZ
z>F_5J<1}qCMXAt+HUdHQYNAf6541qwNb*8}GrUlSHJ=z)GMmpeMDWcO`37N_M~u66
zhv79SUHjAK6)~>)&LLCuH`?EZs>8ZD+<3k3FmWQ`%|A`@%ib!hWCCsu-<I{D4ILcy
z?iXorjWGHR1#wlBD53HI1TpvTzYY_PqDRz!>ACk^SGh!ZJ35439I$Bpa>WhUo@*^!
zC&~Eo$xh!sQnnV8O?t2LO%^MPU^F;^8(%Kej=B+)avJRQfWGLK5O^DAno&hzVW)M*
zWzt3qh}w{P>v#Io{h%(YKuyv`wpF3z*do=_$+$hio5WuYs|;=Q<y2NFX6a>4zxVTq
zeKh`Bj%uWS&L~Fkb*2t4wCAI;N~2bhME35UrD<RGfE|<oZ^jKScnk?`2p3hNCMkdK
z%n_tj(xeOF@DE6*X!Ch6N(!l)W)Du--hB~S1IUoi<fbB;iLbUY;tNPjTKNc!^xHHI
zRYV6UFnr}<?Q6us`a34ZO9W*vlV99ElA*3NOQ81zEk0!f$0xi^72>6HWb(K<r=czx
zD8-bDthx4$jy6?zcBAvSG5jhQ(#f7T-AkMOSpCuyRD2ENu`X)3!NUHD<6Eh!!Y+YI
zbCUSs{0Z7|4d5j9-(*wyiun^wh6~4j?q|Z7E2QeueH!n(lcF~crwuAS;$(hDaIXmx
zzPG;L)k-7jwB$MWD9u-f4|&cT(M;B?kDP$s&q<QW*2+N^c17Iy#nM|DGC}d-$XTwU
zO4Z3TN|S4OUB+bKxC?Q!p&G+jTaGlSRZTdr;!c2+XA#8{V$}j*^3_M*48)HM<#E%p
zi-Y$L=egfjNrN(!c3(QzDwCu;VyevQ=%vAfK(T}fU}htWjLt?w$$RnoyY3HbfTN1W
z&Y|%F+7%q%q?Q^y?G76GVGW@cSXtt!49xy50NJrx@vgG$Qq>Ml9614Bmn>iJpe&8l
zABSmETNk`7ecwxqC7Vw7KPA0sWt#v&QOJJ6;ZH*Hc5d}&_efde3IAERo&*^QSp#Y{
z%Kdctoz67e*fDg!d3UMj#v|<>dLCe#TjFH|V6l?QNoS+NV<<<-^I*sm9JiU{S<9tw
zcQ~v6t|^aJDDkKYyk|+#_4dWmg)&+r<a`XN#dl3k-1@dAN9Wy)UEXsKig(_%ee48$
zF)4Q|HI~OAWi5MF_+6O_nBhtvxa)1yB=B|Qdm;NTCENWvk!U*4?kUnT`5yazWUuua
z+b9fQ@xz%eJ~vVkws4a7+}dVoDrgXCdT1rCkZN9MIjQmPGvu?XD!p~3#=%Co()%Yb
zW1M7JKV33TQvwo2jgtqIps9^|xKQxjkRLyb7#@cYS5iCEKu+Xx?4Y!)@&v7Pj2~U>
zSoz|h2Cg{}SgA{}wLP4Q9D_Z(awXbfqI+Q-XfP6_9gUUXMEk!hk21X#EC@M>OEeX7
zkPkwL=a#{mEvN7&!46toc~taA&J@c`;7CU5ts`|;_@ce@p~|X@*{Yv?5&5RZi<BkF
z8qdoev;xim?^k<vj)u$Dp|az0ccn$-(Qa0mUg9@bS|K~U&)A-2?uE<tU+EWthBp>^
zHl<dsiW6!(tmfcQk%VTHdSpi!2x|S#Q)0xJc+T7ltKHuuS4i!-PQYP8`(O=N3SN5f
z^0ufmKs8|BNg4V;uIM%|c)c*@757jS!1|F-ty^kAFD%uGJJqLXoD`dMt%$Cu3N=W_
zrkR-KgdK+|QST$ZQCom*MGrk9M?a_U3KvzW;yCM@eeaslfpr1jT-mu41GU6(;cFd-
zz9O8%bjtk>^5}YYmj#NxEp8vCQC;^#l`aRf)z$Dz^mqk=&g`E2O5-r^7oq!tMb)ZY
zmbVLTLOI((NdGHl1*<QL#F06`#q!xgAC7Q31Cd5w<kD_eke5|lE8;F<hQIw$K{}^W
zo=k>&8M&PY$B~4${eB?l;8>LT({PG*Q@m4S3cheH@msnU{}Wh*3G0z<rGA)f<jbBg
zcKfXF4)k~<TKr1oIB6e$c5E}p`Aoh39C~hALgPB2lz11Zu<y8V+IIn+Ih~AOH?Jwc
z77XKsXy82qWCN=5N*|oBCzh*nWf)On#IL@G-O8=1A&_)w-4?x@u&5oFQSc&!4P&<I
zbWaHBA7$G0mMmz(u8Mrql@`%3<`lYrsk5(IbkztYk3OkYJsV3OiUv#)dzY~J@;JB~
zAg`&a5UxB_(u&j*)+}%aD%@k{_2J8roxiOF8`-6MbDaP!;~R%ekozmSQht){I^o1r
zChI7~WXl4H)7unx5X76LH7`dKUn6j1`yzbbQIxyrWb#cN7rPVtDjGa2`OR95$bhjt
z`iYngT-l`Dm>wTGzisOer%et?H62XfB`;!upT;U*dxjJ7QR!+IOQ=wHc7Qh*(!EWq
zzT#k9Y(5H8|Ek)Bx&2V*`!o@bmqqQiT+8_?S?5gjP0G!v>6;Sno%UCYSp0-usVlwB
z$E?^@jmj@sy4Xz_jU;2dQl1z+vpI4xcH9`dryMj?5-{?x3bT>h-LaVV6yDsnp!)J8
zU^t4ib3ygj*fv6j&Us9MT+#2;>81#kn(1P&JZYrzt3}J{du7AZH+lU#oq3*Kz2e^<
z5a-$dOPcQ3Fe|sJdiv&7qk)LNag$%K{wzP=yjLQj7{2S=U<Afbx{c#lf(Ue#8DDcA
zS2+=q11*aY!!NqSWo?^`v@M;l^&3xUDlpk_Pr5?r98KaZ5XH5K-c(hPAlpn8crV4|
z`HIZt%^YQ!dm%Gsh1q5M@++}x{)u)q6ACyts&xt|6~A()&}2i1z#gbpVn;NcV`@i?
zMJpTwKFeMFEm_3k)qN;a_A<WR#T;x{;G6tjKj1J?T&Ji=9L&IhHg=r2R<cOym7pKf
zu&`jt;U|lnyvF;ZU$D;sJJOYP(6F9<R?hz-2vIp+?by>rv#$I`fApHfdPHHvLgZ)>
z0d~a6v`U7SVt3&DdfrDKIYap3w3hL9W2f41v+Z)%z6kCknd}3ZCr?L1Ggv%X5XhI>
zO2u~mp`(_m<eSeu9*m6=r%8j?{w$&=5*)+w?22gcaxVDOoFN3xX1EYA0MEYA#J=K-
z21<fF_}Y7h)WG4kSyLXMu7IJ2y06y3)s!9+M+yRrKVj|e_9(j@F_lJH`>j>Szgr6q
z)-X32{jLgt9)EueD=?QDqDan%RIuYV6!3KHAT7I6cNkPc=E!NZOa6ieZ+soRO%r0Z
ze5EXk5gs5wwt2;HgSHNN9Lkw4k6ypLnhfXo0=C0peo`u7poP2lcV4{f{=MDfj;ob&
zX2pEaNWo3)!pN7ml$s-*M7JU<^4Za#1!o)l=#9wg;R|v}HTe~GavDppI-X?v=>X{_
zv=iMv)L3X0`C^P2pW+7awKo1{#nWzjx!8q6$%7G2@hSlFv8ag~1!~|v(_kazP`rK&
z5SR82b!o(XfCS<7kh%JtUeH@uX1D9sVLl$JVYM@Cv|}FkZD-K*2y^Y;PopAlzSi-Y
zL`j2=wNVe_=Awa}8-DP;lh{o+kMR9E$(MIr!HJ#sLMb20TTu`Q_cQp>zDlppbY?zl
zis9ZnJ>75US^pXh1okUQZQQR%Dnu&}wRnz4G*A|4vw}JqRQK++PbV?LwPEaf9omkA
zZ5xn92(yuS(t#(1X~Oh~01129TfeTWA8|Moywsn76~9enS*Pgp4$Hlx=<jod%uxU`
zdgSuGtYevAXm-l|G{axw>=ryCy^#zE8LAp;-HEF|2TJJiQC417kRKk6HtMfX())?M
zo^k#5)|t`Hqh4jOMP{B>m6u?*Ig(trB~U-s-!IegOXg2zILC@{i-g$`g|QeCH}X`y
zHEEjhtM%A1M8;SFvn2XVAohqekK+fYbe$=jqjxpgVVMR?#netl?Yy~kU)Kw4aRaFO
z9OyC8KOHkk0U`SvMThd2_bF)`_@J5**+w;nE29Y^r?=P@lfS{o84?#Wz?u^f8Ay2A
z$6H>uas`2t<S&x*d2ZG9kk7t~99tu~qV^l^+q4UL8Z2RtqsSP)aTK8ZdX3W#VoQ_9
zx(S=7t*53z?_BY(F>m1GE;1#e=()71E8UH7m*M@Pc0cL0ce9xy&QNE4{wwW_h=%9L
z%tD)zR}=v2tf-kF+C7C2IoMn;k%T~f()n2?Uyi^@RMMsVn(JnOpmk$yyMgr+zjIr>
zL=$|u#HKN%=K7nftTu6XqJW*q49A50=`Tu3Ay6Eaw#Pnyo2LktUZoeI$CFUQ{e5Tb
z=dXKsft^dn{)3DdGsXA5bC%;gZ-Yc}y?-jpyPM!HAq?Ihh6e|pajXZ9kRhHpd}Ky6
z+&6$p-ODhEe7BaS-784PJGYnwJ<6cRPv6y>)qB#|;8u@1Ry}Imx|{mp$quEFOVwK{
z^oeew_4oqN-ZTT6!I<F7binkUqS9xMy)aRnPHdiex$cbggh542kQJHMf6oH6BGCTZ
zn7hO5C$<SPUTy;poan7`N*;mf8<DlhK?~0yx!W~AJ7~hkE;f;4^E35}6%XR3@T_o`
zL0Z~PDqm>dg>;E_leG~`a=05Tid$jAqcSR<(l%Lx4+$~&-<&2vynwBaY82o#;zVIJ
z`?D*|tqz5g`ShVUH0=(T>w{22yvb#tgrmZQgg)n#z!J5^^eK$C?L6va>W1vE9~YdW
z`ZzrO1&|8)dAB>2+&^AAhUtOwxDmJkp<Cm@vt5Zn8vL1_Ala?&l_d2wFedl#S`@49
zX`u{`<_?QzO5_Qu^2Ghko}p+GG}V&2i?^gwnR6kgbY#LDDOc!-cyt2S_RFyQYKian
zy~C!+Ar|ut6g5<>oep<vORO{w^4^he<W)V}<4(fKu=GPfty<zv9nqXHN?Y54;!j4N
zB`D!=YjvIy+pvrM?EA<oE{^mmYC)3;j-hmQ@jXpTAxKB1a^i9iN{~@;9ZGgZp1Fxj
zUWdU1Wjr6}yqC=nd4>v!1*Tx;M;GcxzW1*;|DwSvPLj2F1eZiQ6x6?vNBh$hllTNq
zG2(TY@QFh1J;fLP^cXn6X$irYHm^1%2l5x)M|23>tE#LJz1g~K=h%zFaiTcV{p|81
zK<@pMoo~eVb9R;pHtlL?y*X#R<O>Z68O37H$bra}I9nmu+DASsZChw#YA*EI;rGWP
z9J1U^XD~=&;I~I!{2GdKUrF%?KDqZf5aL^AiC+&gJHrY6?^M%<HK-0gj*q$1N&lgW
z^jp1ZR|U2C=NmX?an*&cCd~%qF@aK;vIWhNqc>yFjY&A(@zGkVx6<V<W`qy*1q-KN
zsegK^Vj-%M4~dS8F{Ui7lBB*WaTE<GzdvGV%n(!dKmK?tX@nV5MEE&+*6Xg0DzlxL
zvKf3`yzhdA>-Y|P<Bs&L@<v5L75|V+>vzJ;+hm#|$|vR!xFX|WMVi+KQk5qJhe3*}
z!pK2sx7mG{c}L<8USV7=KyR1=sWl@qON{KdWgQv3f)f`#&ukD>Q#FvhmrayG#)J1*
zdwRGd4v;cSpf<REWn=#k)8p&Alp{0!n=W(o5xs6R-jywe7)(Tt;Z)PxzA>_AiE>l^
z*wim(_3&Z;gex;9NBRu~){gUIwczDpCw_ETZP97xZ&ftme&w!~FmGw69TDy*=Kj0j
z34{Vo)$5IvB`RH$^z@=BKV)@bxP3C&6%z39NZF{eXpr$T<827lYVzp9adP8$)=$z>
z`9nmw3i>ouG+xWPgQAxw%;>s(6TzmGl(?ai<QIwY-R&Qx($9Bf4hiWbC8<8Xb9#Pp
zRTfDz5~=SQ%$DE%vAp;<;=X3#mXr9OQc>LA2GYHF)#9nTQZAua<U4NDy=NU|h)B|M
z?k+3VpK9QEzw-q;mca~@QNlAeXyovHX0<>tihi60E=jxLt>ltoE7Fz63j)dxviUpl
z3M8}JzS~tEi<D(5d71m~#{#o;(>!KSYQ|Z#fLA;t^XQR}8k5m}jx{9hI`pC=$g29O
z_ALGc)v~|-`OCGUTJW~Bvy*xO3HSzf!;bwS9S2p$BrG$;I8T&cTIo!nSxug~yiVn5
zpP=)Ov~awhsPaD6P>&Vduf}suRX(xTTLWe2mgh(}_l~H3-ELKqY387Ik9_w^vMaiK
z&jmL~L~I_!Tntpv*c3Q5X)g8R){?^Z$^E>;BiX3vkk#|D{=%mnu>u*Ev?nTIwRM}5
z>wWSO%yola%pc}yC3Sw9?Oz!-@_r1+jYu^?kM;JCpL^y$YO5!>0PYW)ll!FFc)EtY
zrOVqv;OGc5qtveR!edGON0aoafJhX2^h-Nu^N8CCo)jw{mRykJ&Hcm-F<|6;ls-d(
zky09y0}i<Y!o=hR?$K21EyD3y>Kk#rWdt%(G78EZGRXlj6@Pk*=vE-^<Sy$%+Lt;6
zdhOWG(`h#LdcJbZG_~0ZWf{MM?E5{hR|S{aX7D<$^mx5@ADoTmzK5H_CpJ;0#ej^o
z$%mpyHn&%nG4i;Vd#kN`lO?RD_f6u+`Z@6nU3K{9nM;(CM9Yg@TF*49FKq^StH_&X
zu*B4#X%9z;;<BG#9Z;?uo5{Ub$gM_*mkJUYx(3?b*!ifxE~l#e18&~8FgrCbNxGUQ
zW(vdCtOua(c`3g?*-X0e-CG8H_?pO>%WviRy>)<hKzrcf$cfRfqu#zU7x|5I{fY;U
zFc#wlALGB9gn}*ZX4n!UzWN`YoSb#4{w)unRcB%F#?q>qDIxwzw%)_5vs3~nfoe<C
z^fOMQK0()bN1j!6!&JIx<C3mrh<k<aE2)@V?PkUUUzr0U^6HQ;DKD=!jx~jIq8odk
z^sY|l6j9+4jmKx&aZ?)DNLfndgU^CJ%oq$`P)e25lcCH<q2By-{AU3{9<KD*Z^F5>
zh|?YqnTDEQU)B$(05X(lr$XVQaPqyJxl3y+l$A-^z+UP}VJ)(YR3}Mh(TUVA*&t}2
zM@8bQ^F2mm1{iI`q*1S?>M_v)3KzH4C_z%@8GRjF3`EY=5F3uy>sRutOumctpw)5@
zm!)*IVC<*FBuV#YG5}ev8`IxZ3T<v`xq>J>Kg=e4R*@gpOcg7xslr9>)2wM}e^$Ei
zaTTC5gt6?uNcINf3Bs$>fNmiJZ!nNh!Om!PZmctr2nXNSvk3$p3Ei64>9D!geK+O7
zaj)V>5{#I$6uX;2qdF6kY1_Y9{Pxj7hKydBfD|xO_D5momnLMrO~mKqo0&_bo7{Gx
z{4gcrgG=_sK>iPzuP1Qxpzn?vlsO)F!CF<Wu#ASy$Xe1*%GX+MwUpf$a{>yC;0w7^
z*5m-|AM_QVYQwZ!i2?9)H}RkYuZh{uls4P;6XNr0*JLTc+1epeMtlY23Da2KnTPfn
zoC662Wy%2;bSd$13_$Ok;7J~AH!*mWhbJ+(ma@fPn7GWB1KCLM`%?|n=OYV2I&hDK
z9_M8ZvkA$&0vNc>pbfy2AL3f$ORJSuk=|WIWv{7l?)%?fQ*2tYf`JKdsgL2)p(cJ9
z`-Leuq(*W@Sq?MEaOv-Z?068HKyPA-_1G2OkK^9VFwgK^H(-Mf0d5$<l-kF8=wT6f
zPogn?W#!c53!p3Zi$O7Pq(XQT6e<LaQiK$6XTR0I5XGA%)u?jWhJQVb0AMU%Ql^gK
z%@jYy>{}XCxq@Zn>`6B-*+1^(#n`&=Ob0xLI=c2d)+Tw^*~eIu7j(3SMf@-qK8E+q
zdpeqM^g+&$2{8Q3i?=%V3{u7gZmQp)#9!z%6oRU6WETDJ$Xe?4YGl_9Eorc-5O9Ab
zMfpB-pVxlkBQNI0F?kL(ZoHKUo6nE_IOTkRS|wJrhVACcNNBw5P}lMV&oopNLPx5Y
za5t|2+is%RCsa__n;p#e)H4BI4HS4r<W#QWzn})SN`LZtZVelzfP7&BtG0D++yY#m
z+ZK-Uw_FCHDJa%8Wr^Z=_VZ*eg0^A+RhjlRZ}9BLMioV$!uLh^sVWD-Rk+d(Eh3x(
zbj|su*j1uaC=Wbd*#N!ISp!bs5wkWGW4I!#5~oqvaD`{0$+P5sGQ1oG*pbVQ2?VD$
zUuh-p>Y^Z1JlkmaN-G1WVA@1rQqaD@`o20P{yyZAI@BsZF%}-+5S@1tpOWkZtQf-=
z-o{=a2Nw5`SAaA4ZCI(b((@hS2ce^Sq!^EPyi$OB_!Ai6rd+@?JtEA#znci6E)Ryn
zQeVWXhowpw5#a^xU=-MTF7%2^w|~;@AXz$PBMIU%a}K0}nz7JUBwKXWPJFPem?3Qa
z9e3F@(dH&|B@Wf7?o0k4QNa-=MgeHKxi3(bQLWP;Nip_T!G@_jy~0Q*e&S89ptLK1
z`;Eu(!TaTxaHJ6a%+=@Mnfx1C9V)9A$u^&8jko>@PxTAEfJ6Hes}o@;T+CdZRUjYs
z)J_Q`?^>`x&d`R|YbWpk^4mo~cTp4#ep??bJcB{Y{d{X!GiX7)8!1c8@fMD2A*eQi
z(fS5MUO6%pbYXT3FbXoHky)`pT6Gh@rNSL^3cv)P7k{*ZHN%L@7@Jw4Ei0GnV7CCg
zr~Ii$xhO7RcusVC)|?P;HF=*}>nO1)!#Q+xlv0w;xd`C_obktM9igaaT@XGhlT4R@
znvsitvPtlah1*s<7=@)5k+-*E3JI@r;>NB3@;?OWtzr6=$V+ZdlhL4<HSCmh5?6=p
zy7n%E8W+DG@QfNKXivWB)&C8o#u2&!HKcU7gbaRIJCSKhup!spqmKUSp?m>CeBzz<
z1W)O!Vq-A<<ltI*<!yeFP3wzbo8MOze^O<AKH@Ta4&>m1WeN%Y!LWTA4PK=amqx=q
z3Jbv8JZh#E%cH_kW{3e<R%U;G0}l!AUP@H(BREnClhq-`&rN=QU;r}yp4K77n24g?
z1;$}LRY<F~JrJPds>)5H+oJ^h+;obMKTmhFz-T{zp%KJ2R>l($;v1(mNbwZ?46l^Y
zL9JJSjnA=*7pQQ{Cj2mS0Y#Z_Dy9&HDb|>52N`x#mI<%iG7n0%e#Vbpsj-R{#U(=b
z4(v~{vV(ECaNOMUGYKl3t!;k-4Zgq)@FJta=`0VQXb|C<$x&`VQJh$SX&jV(ac(Aw
z?z5Fvp@t|@Jn5u#wldn;vD}ap;UnJ4iy;G3LS--0z)y+x59=@BCRcvK0$gq$WlNnq
zpvv|VAm6B}rGV=Usxfv-u)ruDkOMJTofsdLLNQ{jLV%kZ5yItgG|X;Vf)%cgo>D;n
z#lgep!o*mO1uY1-J;ZdIw#VYp$~dOlMN}uQsV&{2TJ>1O8|I*V3|A*?{&k5^a$Eii
zFcB{OJk(H&60uGOxz%z-;g+>ndLAtv_1x;Fe1~xX0M|z8g9mnw%<n_Jll~BYNj`Ya
zUdfLvEH@&>41M2br^fIv*_adIEgd=7Whv+}Z*7t*x6|x8NU=f8m@+;n^-SZMj!?(f
za+wl>HOxgfBS#q$@?aQIh;_-AcGZaR@{4&ex1%~#p4~-mSVC+vMcLdHV+J_>A)|DM
zL`Ba$32603fKbvugbVlcF0fg7Q{@s61)bG6cvM1AW@M--r*a9L=-CYSCKfkYw2T$S
zISGfVHnK$rPG7>Uxd6QH8}&)?wQe(SiE<8pM3Bd4M7}Rns~A*us=v^o!XJqhL=PLQ
zEyGUlbDddr`U|hPPjez1^X~>3<^hK@L57h~Midm;vwpI>83tzLOTt;sqHb$Dz<4CS
zcDG5%fY(n|?*ht|Z%?|@Z{Os`)jhf6t1|iU_E8zXSROIj?R_tAj>~6WMBTzFUAzK%
zjShRs<<s}T7>&$@7JIPR1h<+W4`o-zb856(_w{ZVkKn7v2gq9yq7dv9EwTw*>GS-A
zrtaK0qHn9O_V<@uVNP6m^=IR9e~rM&PIzXa!L4Sme&meFfVASr`s(?H2B(V2^tW_e
zk>4v8zb@AL2=CH}Z7TW|yYzIU8b(GH9#x#7R9$!kX>GPY?yaVsgk4)ZA$8)}&s5|r
zu6ldIFnfjWBp;Wq=-1x9aMa<$W3vNR?P?!>UFq5W?SZ`_@Mk2L@zo-nqsLgtMa|Ke
z!B?oLvNPTD^@XFO!9DF|u2U2AmT5?x+Ug%UF&`KD#XorhmN6PU&o^ff2?lXS>CH1=
zW}DO=Ph3khh&!nPTbS?*tsWy?uJqj_++kvvxo%;a8qg~0(3O6tNyucO?QI`Ji&`UX
z->-!laH-xgIqY4Q&g3N=ca{A`58%rrA<#j5Wr@0Pm3>w_RO6YQ2DF|TIJPIF_X@`#
zFJV(p!dcPVX_~HJVVN!=ruyoy1$a8mO&^pM&CNUt-0T1pRBz@EmBPlEbkEcshKy4n
z%QrT;p9tSW2B|R(j#Z3(JiIbV4$S>28we`1E-a`r<Q&w85|s6^JU)UiFETa*iC1{g
z<u6a*%U-ji{D}qT{6v)4KimZIjrwi_KzAFJD|pYogE439Dl#H?^gz#;k`k;HGGWAY
z@SZnAFcv~roLe^>*w>Ac;!N0z_{x!>$q4e9!rNi&;EU7V^tTqdxcE(YLOYw`cz)Hn
zWxjn6>{CWf^Jo$Fu2a1DGQ^~U22#qx3xG)<lTQVJhwc>3K*<zzI^EV_j6acV;-)a@
zJ1K&8i}|qt*UrwPri*{clnyjp%FBb!t#Q$n4Z@8ng>je8;Ox+eMg%*UtM^<*l?z&i
z49j6{Sh+KSIHrezS+nqjpI!Y*4OtM!<>p^YX8B2pWq=a9`L`f?`=YxJXtUj!2|%$_
zfTxS4|MN3#&Z$%j1@?;ACcr2F*H;L-!iGMH)sgPV?SL`Z6UWy><GxXY4g)g6p4C5u
zE1gh^kr3l_TdB)VrTK~Dr7Tl>fNpWQ5LSdwE&m!Uq0{P;^T-oOpz=3umREKb1DkHr
z@Z{5ANOaHpI*+S%`46YtG_U`l=yNjeg>qTr8)uG>SW@J>7GhWAOSBdOC;NCC*dYny
z^M)_4@t55cd?oo@gA;t5(0>7nEis+$*{6!K^tB*cQ?R#A<<|2s!&yhBT%bWbJ8yz~
zGnx|@57{>TMME`-CJ>yTf(3$yZZ?Uxd|3tID*SXQrIl0H>>(-dz>SCX@9-mW212($
z>*oBtASAVNWIupd3jV~4@Zs*}&6y6%robY^qJX&~MqAImSqj~j!XT9DEoSsq?JJ5t
z|8FlN-&vG28G#E_4~TH}S%w2$(mYM@a+dD?3wS$La5DkeI3*>yh;xatzxWz333zZj
z2HM`X3H`%Po-;jtcy<|mvV0+9{`1s1W~P&^?EM_RC(xnmOG!XJZ_t{q|E>z$MlSB$
zs2a%!fYA2@a>>^J1binS`6!nd-CwEo9f{lfa{^D;`$A08=Rc(yWN0g|NrE2ENs~cu
zOXmsUB*lTbD+;*v7`t;T3Txh)-8KOi?ky;wPq6aR`W32&l|gpW(j0d{UkPhL+~#>k
z{cKTVr_Kq`$-2FjP+U1rfpzsT-U`+bx)&t-GSYBe{v3{_RKjmmxK=17>7gQbVn)fu
z&Wo^L6<LF6qpkoq3mk((?3%s!R0^AjA%!MS2Jtz~Ngg}SkSAx%;MQ*>zu0y<;CCq(
z^1M@XYdVX&A8c8IhlD(5Z1kNBWWLcdZ25^_2Joe?|BMoS@`75rqeS7hRwl5g_eOvh
zd~7-mbYS>qp+Ic3|4rcbaM>Pa{51%b{#8cl(@dn#{zQ(t6N>pDO~=EN72p8@643X-
zmKND_C?)wAt4Pf>+++j`$Cz9}9qk;i8;bB}yjohBCg_S1wqEEU?=RB$Mr9WyKP#t3
zE(yBAHf&JU|3Ut>ZYL)xXV}6U)|7iBk`XhAdgOc!=<9AT+s}(BIrvLRdl3^f@sIfu
zJOPx1@qkj1n%4W!QCo3M8uX(G@7IV^dn6YMn##Y-IVj&p4Vn!8j@0no--;Y{;)8OQ
zqe1g3cvz5n3ZQCt70lqnhfE|4h^DtGBH*D1`I#7^?n*{Hym>q=S)m5BxSuDW10Mb~
z;F^XQzL1<k2V7X8eIF`Zr#25ykB#+3K}u|2HP^N@6A=2U`gjUKN{@C!{-o1&1%`ML
zEozU$zsk5RQR=$IXvod{nL#YOe#{$V**xuiPK}%1hRWZSYp1|j3B&0{Nd12sJ$HP(
z9O!R*Jpq^#fY&?IlYlNJXhJBkPYLM7jRv~*TwEm3zNDR~NelKfDfqLu0z8;Xe)N_C
zqn13DRm_?VEKyX09MDpT)FMJRSLvN(3vdyk`}OOTaW%ZCvp0ayH&rQlv2VYDol~bP
z(&!VKTB%U)8F47OT&t;o5$bLLu;g;(1(2xQ9-%>7bkiFNod<>Dd%VnJycPDY#WJ{!
z6XCP?G)xG+O>)^X<||6-D0(vtg5(^;J!F%dn48zZULv(w3T(f7<$Rsm-#_o01Q=_Q
zf%C0TD?`8Tjn7lYyUzDBBkbd6c=0A{kQmsohO1|PR85*h6nh1~lhCyQ4ft!od5_B1
zVH)ei&wq#27V?SZ(P9`OVwIDn-4H|AzYh;vWPwmB_yVXCJFibrNf|Q95<@sc<cAl0
zEez4|6Ky{v)aPktTyh9&&a{MQP&F8pV1kwbp^p>2TNdEO1QT3(LwpxY{G`-Qq{g~`
z22eww)`}X8tA7Gtc8xwlK_G0s5U<dF^AlsvJd3V!PU|djpLL9|3jbP0!tbzO)oM(z
z^e3Oakx=}%0J9p92tDi+y~l#(oWG_CVUb5)5i;cS1qt+4$jEJ!{~KaxB8Zvs57X(q
z&tPG-LS`@*9Xc@ZJC;-u{e<HLzI8IXv`6>|+V@#{gX6_Ve~D+#IUq9p0Hd6WOxu_u
z2OJ+kU%DmoX>&>)PnovZ1fWv?uyatA{~lu!&|y-`zpeltb}l&;+E_>uhHLO#gGxnm
z=sw@~XJ385!zPjHu6*b@+=RiybN<Dez|7S;-1A$y^++i389M!UmtkAor_0Vi=6*+L
z)aS$FgW>(VevR;)R(4kW^nu;^q4>(G9*C5(zVPM*aIMhAL(l6uaynV!|LEdd)J#t=
zBm}=}VO*us{dK308MD)*0WATzy8}mVO}{Qvv6@=nf}UTj_`{cDRgX~oi@)uuHzz9(
z?1r$FPqM1Es!{U|bOXG`pWrI<CMF(J2ufh88nma0Us3AvC%hG8=*$bzaE%y71YYhk
z38|K>Ked1KVbk7AaR+uw7b{&k53VgxM!oAaaKgqPIjd2vefcST=-9hd`8eumfOjii
z?r;Tv!)RQ<ar}i-i{XG$Lq>8AqOP)jrO-b9QM<6&!ocCJ*@f(W?L{Bhy@x%h#gx&Y
zI-k*d$r+JpQ%1%~*TfjKO$Ms<JklaXcufYhiNq+jRo=G0j7+eQ?QNTHo>VZs&M_-X
zmo?ee-L@axv;A6!M$5z`E;(#S?^x7ve0EZNTP36i6(WP9qkWYXrw_fv*d(`vW+=l(
zRfD%mPpj)+?DEZuoK$UWS28g>-j+|MJA3-JrFl9txfpFSz)*2!fBPthNbIxeLjST$
z`P<K*NZv4cMQN=l6Q5noi2UYfzXTnQC{Zjv0pqe;loaE-iDix~Ud%06dg>V0`wykR
zg$z2X<PbCaxtP5y?l>N(qJ`Bp<w#3U^*(fbUlNP_q$cisNB|uc=V1tpyG)r<>$AUV
z4hd*F%WdG{?ScG=Jm*3ktv-X=GM2NRbWa63vCmuED)<j&U5U`%^vmn-M|Pn%%#Je@
z)cM!^Ga79A3;8JmMhcZvTcz<<w(neh56rI4JV`zMKu2pi(p6LaVK48d=5cY_A6?9P
zdV&8<E$BO!aB721MLYp?U(3s$c86V|uN&#%nTp*880fr0Q<lj<m$+P*+12ZJE_-zs
zs$I=8ySj2WcIMj3y|(1Ku~^O)$6-3AQCwP+Ek%Qk+hRuEiJH!Hmo+Q3351|TdPRCc
zCmWFHN{^^ZO^N{O0#^Ge5Gys|k?(lRNM+FZP|+}+mw_kruxWThh3b|Dl#Niz24qH~
zf;8SPyoKohLM|mHfgkt5!f+s`pAE5+2YU^4JgkRaEIO47RYo@-U4Ak&2faUFHI%5q
zj6QK}J)grbsID^hks)636C&znf6>KDC3r!CdAh<IJfyiHjkVb@$^^#Jd;t}Hc@#q?
z&@p8Oy_B=rE{5cfdyor97xOkqMSM*k-VXjcKED88Hn_Hd26N$mCTQbbo%LE$f+6#s
z8M?~g920<!U?Q!Cm+gR2TNXq^&6raI*m8HyK5E|2xnBZ)Zrc2xAc)5+#+Z0=0(?$v
z;OVZ2LzUc^YizJ(og3v*SN(b!U<uG(dHU&=98~o+)~z8X*2s(!jAsGv^^@d`mF1C3
zvgE<hTdBnq@z;tLiQ}tkzM=-Rp4|tQN)mvbFVv67frDZF9U%fN<Ut>Eg>IOjCX!!<
zi<mUNN*p=?wwf~sEn+gg;IkC*js4otF3y2i&S^3^tt&XcE)G>(8%Jiy`qK*#iI(7e
zk>};)zYge!>W(z@=6By}<yYwGI><u}WMwzmLiEy(hV=ii4HUaUUbWA&bx<YN;AyS(
z0lBp~@+yP&nf<;^k|EZj`#CY;>FszOPH?rIh)y@KGyV3<F|`|zih7b{2#e@jOjd*!
z``8so&Ow^JS$CeM?9d!4gQJ{wF_$56mR*?<?hQ+V>K_};V~b^)GD_!LAIT7YmRXSS
z{$`D&&ZdHj#+XJB9rP&N{6Rl2zCwjAK@r|O!{4%L`idZa<Id)T<k<F*tq3*-2me)K
z?5eVQRHVlEt5RgHtD85{)bORtup&Hwc!mu{4@=1B69bkU{h@-s9!60n_e8e~8cg&%
zBc7`8ApnUJXe$|EuaTB_!k9eeO?jCO*1uw@fNr9qmF`%n2*pNzWW?_gv|nX|q`{a4
z6KT5)gv9UNRdi_Iv=L6kcN=ye5ZRka2DeHy-wI~K+{%~18UJbaO;>sOIlTlAUzYcj
z#vW<)J>+!-3EJO~E)hVCk3n(~{dqmXh!-T7^G)T&G5U6JVpuMF))S$(L{ccqNWV+Z
zKY{g|TSBErK3pGcA60<6xe{d%5O5lv|6hPpf>Lvd1C<MNFQ+*%W^#845!cVERIFS{
zpfJ9=WU?SpW;+3SfNCdJK^bhfo_`%=*HtToiz9eip-hOF$<86|vpfFAiJ`q*T)VHj
z`XK27SR1dvh~a%rHWdJz4MA#%j@bWo$^}SG1SWejKgc(SUTq*p?1a#vcX2gBxPw*7
zLLnTAKz*YTa_kC~mORY^(<UlPussV>NzX*P2QCx5>AR%^jt}GPiq(U@)+ooih4y_6
z<3;-(Bt-6u(3MGizwu51E>5(6Pht~*I4X{y@T$`9kkscS)yo1<NTdhDMxTn;N(I+m
z_;m-$bUh7e><z-FC|_tsLk)i-Ws|sNeUOtB52WZ9$*_5lfvH*R4A5$mkckm<+s-jb
z0ZvPFI3LZ6o7qVa!aXE#jHUAib8iZ}^EiA|Z~8$~CgwCur38#^$131-9A~qf(pt3l
zWTye0QkoE6T$=CsE>kgF7HsE&hM{>-iQ-e4m%l#<<CtoS2I_Y6b>XSmtc~z#y2?2Y
z9gr_*I{znpS!~k~&Orug+c@GzTmK&vWh^_(G@AmxxFzCv#qK!@EXnr_cqUL4syqjq
zWW=)vE%{y)hH+T);QMm+^L-QvK6+OTFy?qVVN4g8VOuoWZ!nkDB)m~hgq5F(co0i-
zAH!sm%|f<9_Y2JNa7xAhr;h6mXR{0YM(wJ^-bAgo6xG@UF>6*^t*!Q`^{QBjQ7ej?
zZ_QF$Q7vNCstTnxwUr9BBO><vp1$w*$2b3Zt|uqYb?$SYbDw*i`}~64EK>dVQ)Dqg
zpYBUggLI)!g)vcSen7m^+V;t@jdN|BAJ9{!Sq5{OaFSa<B&=-#F&xudXYgSke$xa5
zBJ{RO;y5Y-eJij}nc3fR1;Mzf_jK)zP^r^(jz*|B()ebbT+k)S(?UQRiq6e_d9Qsj
z0cMYIPg3FSVR+BBRj|4xmA9cppn5#TlxKeq1LpldETwMH5JX6prd0ufb5ErlS$p^3
zA2GPu>C(!tDmW0G1@sz3oM*y{z4I3W7TC%jC<S^n%7nEc!LI@CndlF>!Tsn)viQmO
z6sTo+PC5-Zeul{op?PEk5Nk>1QIT;Pq{FtW9_j;|YVO_&KWAKaOh6I)h3ilKtua7`
zL$64F&f1Bd;)!SLYJ~{X7ai&cqKe|~Gub(i21!jZVTXhkGprdQf1OXgCHp+IAI`4Q
zn{nR=g%2A!3jn6j>QPo-%e|ZtIzVl?YWZR*M<4!?KBXckW>NLyS+IJGNj&d)kh)SP
zO}v(LE5zEvLFvgV{m9mhA4~$}MiP|4>KZny(m}MpA+$U<bDf+o{%n>w3cCrKlS|;$
z^&<^|JU~O@NHxA>(zg|2NU~G~sE^fWan`{3(@9qr{DU{|QX}LXvYl%nksJi{7K0n1
z6VRxd6yCea?!iI}I(7;j_?8K*Od>f@R+IxiR#VbWPXS_9o=g$UEihQC(^Lh7m-$?|
zA05DN!PWYUJlU^go1UOv-@XqxQ$H!<XVY>)Bf-c7iH9e!yL=|UA%7B_Xfe1t%{2e+
z&zSr;k9hxhKpEY3NAlM=MFJDZ<qIrDcQ5P*pHE$UU8E0qkV!z{c+DcH127M4Kqe)q
z;qQyk4Svs$5uxU9<kvUorukEzd=z7v>~HK@rw1oU8FG%13It-t$&!`1*2t5$ER*R4
z0N>^qjV1w1^7z(k{V*L~A3&oGPzS?W#OOv`7V~06_qJk(m%8hW0fCyVo?eY50TRF%
zHA}A-!$4etUozwG)=?KcXS@!^a)9mVmNtq>s(<;YcvMzDW}rktx?7ETv5SmHrH(1R
z6zt1%$tvPF@?@FHZNUEacfl;@=zT!b;VOVZ;~TnTI`io>wN$K8VQViP==iqNK=e2;
z1y#Zb2xlK)r=VCjvz6n}vCCB;+|jfUuKL?2dlX>)Ql6bZWsDx^T&X?`0T~UkK=6;6
zvLX0u4z(l23rLc26F`G77EzDliFeOe0l6~ysDPk%vQWfuc0IiJv8tG@W!mAiK)sCT
z>5*y?@)Sr7?{cQG2s8;MARs9XhDj0{Gr|oF8Ycp&3>p(w(TA+b9tJ?4PZR@DPrAef
z@*@GV=|6$NvSlLpg`+uF=~X~<9FywEdXz^(GfV{}!eOy*8>W*hs?i48<anY3e5&>K
z3JOu%PA$Hbe~Bk2T&|s7p9h&WG^>Etx%^Z>iX3defv;pYuAm%(-NMWGejlrAIF@gu
zmm~RErK+M%R1s5+q7tftvkx~HqX^G~B~|<FEX#qz!>5dwX#Ky)muTMuX1k6iKU~W~
zK?Ou3<Pf1_=qCm;qeBC9L{fq66&~34^dDlqLaq)#xa;@)Zs8wR^o&D5c#8AkKUuj%
zXS0<(#xw>DyjwGn4~0&DX&aU?9$@~1O2CzZ>sY-AuY0O$>82C}tn0mJUtyL%czi%+
z3n40?TkMIHyP&9RECcg~04LI+e5k$t1qM6iVZ~Sb;W|TSfvBKD_?y_^{zldf@?_rC
z5+I!wfMT(eS;_@75R7k}jos-@h}YX!(#WsnF6`q4hA1293D>x-D)?dQDX_dnfhdP>
zp5Z+~HGs)qodZm^^xpy)2Ovd&|2DV<2Z{>>LoRyAl5>G-1YL4pQ<Qjub`7BPz$K2E
zW8I6up!AH%FwKTLuIS8WfUQosll}0iWliMu6>^_4LRdC~#!PV*0|p<5S!cqxYU;mi
z-+OlY9DIu-vFPYI*wqNw=Bwy&MbW+RBh7l`a%j6acIlnX7%^NYyNMWnR74;{GhjHn
z_P!mwB(>A9LdqcdWj++=$sMV3o0{0v1EMHFp#ox1gPbRe;l;6jE@iURE47-H+~s*P
zr>a{MS__6psyY52LtZbF`#@g^-ul8|$F~$K$cOrUvU$qQTX0d{vfmHYhwdMS%Q5Yl
z(Nm)26Z~ZYQP8w*=!YjdYEWxpxSW6n-c2*C$DclY_|?S8z7`)UhIH@uIz54Md2A>d
z!EZ`>t|G)hc-82>ZEe<GE2@4LZe(Gs>$NgMg@ure5+W9=j;)q(&v^^xzvMV(F3_L6
z<>W(IsX_&I!jF_*2wj!}3wKb#V&^%WIEmm_0aU&{X~D2NQ;4}a-hYY;JIfNz{o+*h
zj9wKQu^*kG&KGzX{*+};`Zga***8^0MFUk#fr|T1)^r6hmWPgFpt?YQT+XRl%Z%BA
z+_7q<yHepgd3E~;u8IL;_71wheicLv3z1tS_mNDx6&C`s3F~**KUTGTOa@=_69UX!
zjRk)}LASl5(jS4M5FpFY>HY@$G0TdM-?w(ry*Lb?>I>lLqr#>zo?JIm0ZEW3C593s
zP)<sB$cKvm(D14PgtOWDO)Qsdiybg1S?UkY7w=^KKvzMVu%!lq4d6Y^FMv7eHG6ow
zsjzp#ULCmsUb0s*U;aHaeulRGMXKtnd$VDehvDqBh4>EVR25JKDI!Kf_8+&4KzEvn
zHs{i4<5P?@bqY6on!%vq(VNVsgMhRdzf$7-4*QYq*dRvKO!_#U$!vp4-PvSH2mZV2
z&A+nY@iKrg+-7u9E`7dwpB!}ZdCQ^YyjJ~;9UXhT@3!z0wgWR6iP^|^clq_+OaSv9
zMB0}wA%7T{x8%$;3>{$gKd+o^Xm#H?pK^V5XOXSPB-#Ke<6|29qQ$7X7^>3Ziy;ML
zsb+L+G>Cm3A)vQhq5G-+CT+25i8*z!$}5oL=t4HboJ>_#7Y8D_UYpzt7laW8b~WM~
z36!TuP3gpaIW4(%sjVz=-V&l-Nt$$PFQ($MmAQQ@U7J82Jkp*g_nG(%Y{wTiz*XB&
zwt%Y(9n|+A9(8aX@|~<NAiFxRpln=-JLAi(H0=g94JCu=_J>ufvDxkub8xmcZ@Z|B
zUFtoXdwg9CmWe4LXJ@0mm&3=TzV7KYYB_JIy1laER`_6i;&G!o+b;8-+DI-9T=um*
z=8VyGOTaaqEUeJZ8N>81oCCL)0;vPB>tdH4ip|R|(|x_kYjJR>a2?XmppoQ3c1&yP
zb@Tql9jY45ESP0dp~qzdq(y}}2WcmBl@$_FJ}~`Ll2;w_+y;yJI$q1$+aMk641%<=
zfXVjrHG;!}=^PT~D35Q}zfA80YWr`J^(zutlvsChR+;D}BkEoXkF1GnEc~7mTC|&b
zF_Ekqr!ZRP8*|+CIfEBtAFaG6L8hqjI&_kAT#FQ?mdM-9wnia`0TX4LkxFGSLp62e
zmpS*&l-nY>^?Rg4#@WGig(o;STB0#i;(lM10c`%-d)C~23N&uun?j_^V=_$JjGT{f
z+V^a;(<UdDyWBGGN-hRJ3*i)TwQs@sMo<o!5)P#_`J#@doFS4lQm_E$m89asc~laG
zi&j9gaG*10yvK_73nfs6>1R>5ZrpNaGZqAuCbt@*GA2<0L8?Z(XXyTNd_U9rNav`r
zJb%IR#vaKRSW~3g+lu1fQ7OgRs<57vTFmEe<5eRiy>_Cji2Ab-rmz(Wf6aOwU(`8C
zc^L0v@w;|vg2k?`^V#)f&_yQAU7IidpFec5htZNe)`d$Wy$w39J(%54eN#D;SU~cy
zCI!8caHo?EF$1>@O+1kJHeYq}X`-EOh=rOBGBE)~R8gz8S5h*x2!j}^g{pFBAH#)|
zlZqX#=+ZPnQ$y}La%>_!&<x*KTj;cY#Ro&f?}*c4O=$CEa$c8!j?JW)ZH;9=>|D=-
zKdP81GBA-^qn_D_rrK2n=Xgb+VxB0aJE8l0--*vw>-E@rfL2L0G?-(QhV5-9V|ejr
zEWPH2xj(B_EuHl(L;-}6PvK%qC;l|GC$w*=kom5@DTOF<jXkln&d~X4kv}r_+)KZ%
z`kE#RHRYzRf|%^DRm;&Y_f5@X=~n{VSrM<h<m&Q0;-Ch@X@b3x!g@D<be)j$fy^QB
zNoXYl0tMnLAa~_}l=v8dTHO$!?vKKxSU-iN95Hd(+q^4dgKQ)L<U3zI^e0|e+n33p
z?AMI9xn70JaH@uM$WF?k=zKg_cg<o<lZ3c#)G<LkSy(l1Bky+=CN)%U&M^vL8gx7k
zc0zhs-$!Kpi!7D4QX;sg`{Abi;O?>ZL?Uri-R$f^rI7gd0tveLYb5ht9W0O!QI59v
zp5ADdM0@!<O23ikTX4%063h%t$NO96w4xKMzGb^gm&=zVFn6EeDZbrpg!IF}g(?qk
zZ_oavuVWF{@Kx?FADBfxk~#Eeifn8MBd6iTm@}gBK-c7@kVmpS$};L{HTa^Z<}>3P
zg+X<?XQoS(Z+!Ys4KiZd^)~m(A3g`CKWFx&wNqD$&BzF_P$~+2D9x2RYF+Uvt{!nw
zOdy?>dsOn!k~Q>2mqHHZ1jlEu&(&5+B1PwEOxyx_a%kTOTV0egJ{VV6QT?VRe8Nb2
zWPK|i$>{zzhG!=UX=_%GvQOCDZiHjdmt~c1Nfo7;@JBk$Kd#ltU&Z<13SEGr4LJ9D
zZ2)hT`6%mdGxhMe42I9ldB1Y0q^i}!Z`}KSguR`4aBGNni+I7(-esyMZ2|aoy4P;%
zgP)H2v~5k@xU57M9l|q&``P;{gIU?+Qk6<BpFK%-=KHR*c_+kPAMlQHFg74^u=b*B
z@|>-f9;eutHeR1mgY~uFtTKaI-*Q95y19NkvbyiKz=E)v`RuMm<vSA_wMLKm%YOV?
z%U5YfMrjY%kr_fL2mKk7u`S4eR_F<R6Z%27_OrR+or))ORe>_^tKP)2Me^NUfe-W%
zW$F-?@4BRyLq~N${=M<1o*T(yaORiB5t6~V^Sqqt69ymoUO!Jhm;~=D4=8K)j-p}l
zp&U0uO9%AWa`s;+_ZYoqZ?7Mi?hB~?mGeETuHMTbyMft@>EkQcQrh1bcLl*^Ph@yX
z-fs`sDcFvf(WgCIvpjrp-Je_z<zRIfpl4HtLyrtSFMI3riz-?Ub>s42@y(%C6e)nU
zbKToMA&7*Q`G;_&jN;4T62BaAdcMvHF}fXi@Au_T!2M!{fhb?QnE-qCzrV<p_XCGB
z(6BBwrRc=370S60xz0(c?<CJ;c&VhM{t~(?k|#SRTu!P$#qu@bjkFwUh((01DoBH(
zCd22rV*~b_C|OoszkcHp^H)W66$Nb8fEUVjhV+*8iy??bkuv5^)O|OLzq5wapSeI|
z(8yjjjo6>|7xXu-AqhnX2<m#aCcNmJ-;9iNyGwTefP5!y<l{}t5<+pPyNjNi*+?^E
z*WJCTbD`)#iHnWc2zf{m*`uL&w#V0JhYRQg*+>z7l=odP>+9baj!+BH`}=}Ke&42-
zyuuQ#Z+3He3nYu~r*dJ8#7YZZ69_5f<Q+Cz@pV_kgX4D7vVE>jcfXoQce5<<U+*8Q
z^Rn93pT4D<%rLQ{UtRhWRW@s^7^kdYi`78{nN~9Yh-31qxQ<>Lne%rHPZAq)v>}SR
z%pT@MLAjSmz{-W0+dr$nd_2~3DJ`oz#ooIVZF)MWrm+@hh+p(j?P+iNbx|Tp31__u
zfh|Ei=Z=Ds-RRX%Sf|T)bP|H^XSeisTASHr$M~;QMSZ#zr~X%H?yAFsHF|L37~HoJ
z?Q@Uy95fnRuj!D0jydWaTD@*wRTW&lKb7T$dDMkJAcE?2vL-%nM8CrAuJD{XtS}w6
z1t}1(oeCB%%sf_PZSV^`j+)&+j9`Mv%+(I%*4KWJ0Uvs(xm=X@c6P=fYisSd=klh<
z(5X`td~ovzZpTsEefM1tw9=I7BXY`pK}-dg>~nZdPV+i5?LAea$Pykk3))|7wi+mW
zf_v~*9+=_k)1BSEI&J0GZJ9?Agse)*iA^ak+1t+(oCcKxlrMU!pT~?O;NOm3=$<%x
zWk*exwpX9v4w@>In>V{mZ=8Re9Wxm`H-F~eoIroA<!Hi7gqLfWMl*)3^U3!>RBvhb
zymayG$F(nO210$?mm=815#>E;5>f&6(>QJCO!J4c`r<!i9(YK!n!Tq4swqO*Il3;A
zb3W~Ky=F7VyPsYx&)mx%ZS#$aZI)~mJAtnq!Nx}}f1YbCN#~_m|4M=zWLNY&G+WTF
zr|uDN3t;x|F>#DQ?TEMaRP^eunJ}J6P@Hc`_69xmBH$900|ta}Wfv)K2qkN5@)K?u
z1;Hwtud;0>J233gyNszteoh|(8orldI~OwwWdrc7ynC%zB~6K}3whSAd#pBl>v5v%
zmG0)F6h*b4U+u4t^}qzl6%$wYX81UjiB;M3cHeL8mC6K9x=@8kB|A6)HD(0`1X~m1
z0!OJS;cbDYE^WAZdq5H7#md2gyYS-G5$jd*k?RJHlUr|u=YR%ksV>vsR|qwoTfypk
z`b^vRnu}=;vvj8*ul_vw1qjx)yoMki0JaSq89d<Wy1p|^P6L33{dYvA{~uQJe_%2J
zKsSHD&HjO>{1;aW=<GjnDqjGQn@*Jgbdc*csorO|3%L0e`nw*-{Jr++E0%kMyPT1S
zH4J86(@ECoWD_g;owT6yempz_De?k9)=GNwD{{=*@8_ao`Wndy@Z8$@-TnuY%Ra#;
z16lfb$#x1N-V3yQ-vT`@qt+Ki{V%LfjdF-#rq{W6;P#8Nms@%D(syFejclb7%tpQ3
zqXi}Xh=)q)e%v=|-X_ZzS#3!L?X@!o-sp1{PiEhY9x0@1!{xCy{5ob~gvDjbtQg>3
zugAR^{AiRl=H~sKXu}{l5UM;3KB=TiS!v|;E@H)wOGQi<U&Y-;{5g7k@jX@z>ql3L
z7UwIhYd#yBz-dcdZs$c0TR(&IjD?tq$TCpLygPSx?0&_1v6@%!%Vh5T73zU2T`O)^
z9_uM6wm;DIeM<j2@M1F@xEw}h9~lC?$bX<>d>_^5-Z{!Qyz!pNS!nBN4JiNdw~+%R
zfr|>I_vHp|QWU-Kdx<V{e^R1#T$=AYhN1YCR`1nu3(yRUIQlq$T|jiOezE9(?tSC)
z>VGkyUDKU8&CuN$=cC-F5%#&4LHw>4n`(Udc;LlwGhD+BVhAPTZn^B6qo19=E}r?Q
zXOxee1!aldCS65GFkyv88AS~A_Kcbdz}X3z_x)4&VOOuV;I4Jx`dNJE9ipUmOUPCl
z+^^^8{p9&reVY&bhttsA8kYpA(_;Bb%5<gn{dc9uRlrQP^Xg=c1@R}}M^*D!XN};x
zH#*m*UhxF`i=Q|ZA2|MWw;>yJ?PGgoU@2E@rLTq$UCcY^$R`qoqD#i{KlzgnCm387
zC}TUUkk(GIVo;RK_O6lMh3!c8ysN{=C<j5;H*76uwHcuDFO%Vh>5aU5Izkv_yujY*
z$oS@Pw1;BbkC26SngOVX4cTkgQxEOA*RFl9-~DZKQu^V~p(*?l?O&IVOK0Cq-yF8E
z(e2-8`(Q%c5uf?g3B{!^xh&rJhqtJ$^c9kkK<~DH7x3Hy`-g02(D3!YL3z;F;U`{|
zxl%snWE{G?1@aZiNE*44Ruwxe=HuDe;Oo3QBGXWM2qAk-34J4#typJSXdqYeFglbo
z>ol>SvFV>1*%z)p60MCL!7dIXoW#vP)TD+&m6PsX?%T!>dnvEMYClpbd#)>BHx|x)
zUn+^0iEM|un0nD;Z-2X`&T;cUe~-UqVd?m<0J#Bl;8UfU>4D$mEV690SaA&pA?DM(
zKB}9Zx0x{(D8p$g{C0i&dTW;!L+fdW#@%-PuJ@v!oquZuqZv%1({Il~qmYf252sUx
z<XY+qzfw~WSk-5DTtU9){*beTG#QIS6%fl<P?x*%aGTT&RSe(f+((N~4i;+=&Z$}L
zZ|0e}G5P%S>T94`$fL8HoULy6o8E}uI#4!8>}@l*xT`b;J;yu)J}(8CPeWQo<WMlR
zmZ)J!{bZ$(+l14Qw&-nEE5vGS{S1|jX&}*ftaGc=(FzpQfDAJ72F1Jt*b4;EJ3Gps
zFmm$bDE#hT&tb!BNq>8{6xb#WS!wq^@6cRV?oi|Y!}{)ds~BhmyDL=k?&1C`S+i24
zca7ivH5gYGh_B*FNn)qs=FNCxke88IcFz1m&@wRaTw@SYU9CmgXF5MtkOctZeSAgp
f|4>etOLE_W^XFtu^2GokH;JyMp+=Ru?aTiFBmh3;

diff --git a/vignettes/examples/timeseries/timeseries_classification_from_scratch.Rmd b/vignettes/examples/timeseries/timeseries_classification_from_scratch.Rmd
index d9ccf151a..153ccad4d 100644
--- a/vignettes/examples/timeseries/timeseries_classification_from_scratch.Rmd
+++ b/vignettes/examples/timeseries/timeseries_classification_from_scratch.Rmd
@@ -69,52 +69,8 @@ get_data <- function(path) {
 
 root_url <- "https://raw.githubusercontent.com/hfawaz/cd-diagram/master/FordA/"
 c(x_train, y_train) %<-% get_data(paste0(root_url, "FordA_TRAIN.tsv"))
-```
-
-```
-## Downloading data from https://raw.githubusercontent.com/hfawaz/cd-diagram/master/FordA/FordA_TRAIN.tsv
-##
-       0/20094049 ━━━━━━━━━━━━━━━━━━━━ 0s 0s/step
-  114688/20094049 ━━━━━━━━━━━━━━━━━━━━ 11s 1us/step
-  458752/20094049 ━━━━━━━━━━━━━━━━━━━━ 4s 0us/step 
-  925696/20094049 ━━━━━━━━━━━━━━━━━━━━ 3s 0us/step
- 1458176/20094049 ━━━━━━━━━━━━━━━━━━━━ 2s 0us/step
- 1867776/20094049 ━━━━━━━━━━━━━━━━━━━━ 2s 0us/step
- 2375680/20094049 ━━━━━━━━━━━━━━━━━━━━ 2s 0us/step
- 2834432/20094049 ━━━━━━━━━━━━━━━━━━━━ 2s 0us/step
- 3407872/20094049 ━━━━━━━━━━━━━━━━━━━━ 2s 0us/step
- 4014080/20094049 ━━━━━━━━━━━━━━━━━━━━ 1s 0us/step
- 4751360/20094049 ━━━━━━━━━━━━━━━━━━━━ 1s 0us/step
- 5939200/20094049 ━━━━━━━━━━━━━━━━━━━━ 1s 0us/step
- 7675904/20094049 ━━━━━━━━━━━━━━━━━━━━ 1s 0us/step
- 9625600/20094049 ━━━━━━━━━━━━━━━━━━━━ 0s 0us/step
-11698176/20094049 ━━━━━━━━━━━━━━━━━━━━ 0s 0us/step
-13975552/20094049 ━━━━━━━━━━━━━━━━━━━━ 0s 0us/step
-16367616/20094049 ━━━━━━━━━━━━━━━━━━━━ 0s 0us/step
-18743296/20094049 ━━━━━━━━━━━━━━━━━━━━ 0s 0us/step
-20094049/20094049 ━━━━━━━━━━━━━━━━━━━━ 1s 0us/step
-```
-
-``` r
 c(x_test, y_test) %<-% get_data(paste0(root_url, "FordA_TEST.tsv"))
-```
 
-```
-## Downloading data from https://raw.githubusercontent.com/hfawaz/cd-diagram/master/FordA/FordA_TEST.tsv
-##
-      0/7364408 ━━━━━━━━━━━━━━━━━━━━ 0s 0s/step
- 122880/7364408 ━━━━━━━━━━━━━━━━━━━━ 3s 0us/step
- 589824/7364408 ━━━━━━━━━━━━━━━━━━━━ 1s 0us/step
-1277952/7364408 ━━━━━━━━━━━━━━━━━━━━ 0s 0us/step
-1720320/7364408 ━━━━━━━━━━━━━━━━━━━━ 0s 0us/step
-3342336/7364408 ━━━━━━━━━━━━━━━━━━━━ 0s 0us/step
-4112384/7364408 ━━━━━━━━━━━━━━━━━━━━ 0s 0us/step
-6004736/7364408 ━━━━━━━━━━━━━━━━━━━━ 0s 0us/step
-6823936/7364408 ━━━━━━━━━━━━━━━━━━━━ 0s 0us/step
-7364408/7364408 ━━━━━━━━━━━━━━━━━━━━ 1s 0us/step
-```
-
-``` r
 str(keras3:::named_list(
   x_train, y_train,
   x_test, y_test
@@ -336,384 +292,568 @@ history <- model |> fit(
 
 ```
 ## Epoch 1/500
-## 90/90 - 4s - 41ms/step - loss: 0.5310 - sparse_categorical_accuracy: 0.7198 - val_loss: 0.7948 - val_sparse_categorical_accuracy: 0.4896 - learning_rate: 1.0000e-03
+## 90/90 - 3s - 32ms/step - loss: 0.5310 - sparse_categorical_accuracy: 0.7191 - val_loss: 0.7907 - val_sparse_categorical_accuracy: 0.4896 - learning_rate: 1.0000e-03
 ## Epoch 2/500
-## 90/90 - 1s - 13ms/step - loss: 0.4771 - sparse_categorical_accuracy: 0.7597 - val_loss: 0.9159 - val_sparse_categorical_accuracy: 0.4896 - learning_rate: 1.0000e-03
+## 90/90 - 1s - 11ms/step - loss: 0.4799 - sparse_categorical_accuracy: 0.7590 - val_loss: 0.9127 - val_sparse_categorical_accuracy: 0.4896 - learning_rate: 1.0000e-03
 ## Epoch 3/500
-## 90/90 - 0s - 3ms/step - loss: 0.4717 - sparse_categorical_accuracy: 0.7573 - val_loss: 0.9025 - val_sparse_categorical_accuracy: 0.4896 - learning_rate: 1.0000e-03
+## 90/90 - 0s - 2ms/step - loss: 0.4590 - sparse_categorical_accuracy: 0.7757 - val_loss: 0.8196 - val_sparse_categorical_accuracy: 0.4896 - learning_rate: 1.0000e-03
 ## Epoch 4/500
-## 90/90 - 0s - 2ms/step - loss: 0.4113 - sparse_categorical_accuracy: 0.8000 - val_loss: 0.8173 - val_sparse_categorical_accuracy: 0.4910 - learning_rate: 1.0000e-03
+## 90/90 - 0s - 2ms/step - loss: 0.4088 - sparse_categorical_accuracy: 0.7983 - val_loss: 0.6425 - val_sparse_categorical_accuracy: 0.5534 - learning_rate: 1.0000e-03
 ## Epoch 5/500
-## 90/90 - 0s - 3ms/step - loss: 0.4164 - sparse_categorical_accuracy: 0.7920 - val_loss: 0.6884 - val_sparse_categorical_accuracy: 0.5146 - learning_rate: 1.0000e-03
+## 90/90 - 0s - 2ms/step - loss: 0.4141 - sparse_categorical_accuracy: 0.7889 - val_loss: 0.5519 - val_sparse_categorical_accuracy: 0.6588 - learning_rate: 1.0000e-03
 ## Epoch 6/500
-## 90/90 - 0s - 4ms/step - loss: 0.3984 - sparse_categorical_accuracy: 0.8101 - val_loss: 0.4227 - val_sparse_categorical_accuracy: 0.8294 - learning_rate: 1.0000e-03
+## 90/90 - 0s - 2ms/step - loss: 0.3929 - sparse_categorical_accuracy: 0.8163 - val_loss: 0.4225 - val_sparse_categorical_accuracy: 0.8294 - learning_rate: 1.0000e-03
 ## Epoch 7/500
-## 90/90 - 0s - 2ms/step - loss: 0.3878 - sparse_categorical_accuracy: 0.8132 - val_loss: 0.4749 - val_sparse_categorical_accuracy: 0.7018 - learning_rate: 1.0000e-03
+## 90/90 - 0s - 1ms/step - loss: 0.3870 - sparse_categorical_accuracy: 0.8146 - val_loss: 0.6112 - val_sparse_categorical_accuracy: 0.6949 - learning_rate: 1.0000e-03
 ## Epoch 8/500
-## 90/90 - 0s - 3ms/step - loss: 0.3739 - sparse_categorical_accuracy: 0.8163 - val_loss: 0.3880 - val_sparse_categorical_accuracy: 0.8252 - learning_rate: 1.0000e-03
+## 90/90 - 0s - 1ms/step - loss: 0.3717 - sparse_categorical_accuracy: 0.8247 - val_loss: 0.4257 - val_sparse_categorical_accuracy: 0.7559 - learning_rate: 1.0000e-03
 ## Epoch 9/500
-## 90/90 - 0s - 2ms/step - loss: 0.3684 - sparse_categorical_accuracy: 0.8292 - val_loss: 0.4363 - val_sparse_categorical_accuracy: 0.7767 - learning_rate: 1.0000e-03
+## 90/90 - 0s - 2ms/step - loss: 0.3698 - sparse_categorical_accuracy: 0.8306 - val_loss: 0.3653 - val_sparse_categorical_accuracy: 0.8350 - learning_rate: 1.0000e-03
 ## Epoch 10/500
-## 90/90 - 0s - 2ms/step - loss: 0.3590 - sparse_categorical_accuracy: 0.8306 - val_loss: 0.5077 - val_sparse_categorical_accuracy: 0.7282 - learning_rate: 1.0000e-03
+## 90/90 - 0s - 1ms/step - loss: 0.3634 - sparse_categorical_accuracy: 0.8260 - val_loss: 0.4390 - val_sparse_categorical_accuracy: 0.7836 - learning_rate: 1.0000e-03
 ## Epoch 11/500
-## 90/90 - 0s - 3ms/step - loss: 0.3638 - sparse_categorical_accuracy: 0.8247 - val_loss: 0.3574 - val_sparse_categorical_accuracy: 0.8405 - learning_rate: 1.0000e-03
+## 90/90 - 0s - 1ms/step - loss: 0.3676 - sparse_categorical_accuracy: 0.8281 - val_loss: 0.4205 - val_sparse_categorical_accuracy: 0.7822 - learning_rate: 1.0000e-03
 ## Epoch 12/500
-## 90/90 - 0s - 2ms/step - loss: 0.3459 - sparse_categorical_accuracy: 0.8396 - val_loss: 0.8054 - val_sparse_categorical_accuracy: 0.6352 - learning_rate: 1.0000e-03
+## 90/90 - 0s - 1ms/step - loss: 0.3467 - sparse_categorical_accuracy: 0.8406 - val_loss: 0.3697 - val_sparse_categorical_accuracy: 0.8460 - learning_rate: 1.0000e-03
 ## Epoch 13/500
-## 90/90 - 0s - 2ms/step - loss: 0.3350 - sparse_categorical_accuracy: 0.8552 - val_loss: 0.3838 - val_sparse_categorical_accuracy: 0.8155 - learning_rate: 1.0000e-03
+## 90/90 - 0s - 2ms/step - loss: 0.3310 - sparse_categorical_accuracy: 0.8594 - val_loss: 0.3341 - val_sparse_categorical_accuracy: 0.8530 - learning_rate: 1.0000e-03
 ## Epoch 14/500
-## 90/90 - 0s - 2ms/step - loss: 0.3303 - sparse_categorical_accuracy: 0.8545 - val_loss: 0.4148 - val_sparse_categorical_accuracy: 0.7920 - learning_rate: 1.0000e-03
+## 90/90 - 0s - 1ms/step - loss: 0.3350 - sparse_categorical_accuracy: 0.8514 - val_loss: 0.5723 - val_sparse_categorical_accuracy: 0.7198 - learning_rate: 1.0000e-03
 ## Epoch 15/500
-## 90/90 - 0s - 2ms/step - loss: 0.3228 - sparse_categorical_accuracy: 0.8507 - val_loss: 1.6007 - val_sparse_categorical_accuracy: 0.5132 - learning_rate: 1.0000e-03
+## 90/90 - 0s - 1ms/step - loss: 0.3295 - sparse_categorical_accuracy: 0.8497 - val_loss: 1.0273 - val_sparse_categorical_accuracy: 0.5465 - learning_rate: 1.0000e-03
 ## Epoch 16/500
-## 90/90 - 0s - 2ms/step - loss: 0.3232 - sparse_categorical_accuracy: 0.8587 - val_loss: 0.5349 - val_sparse_categorical_accuracy: 0.7184 - learning_rate: 1.0000e-03
+## 90/90 - 0s - 2ms/step - loss: 0.3244 - sparse_categorical_accuracy: 0.8587 - val_loss: 0.3219 - val_sparse_categorical_accuracy: 0.8544 - learning_rate: 1.0000e-03
 ## Epoch 17/500
-## 90/90 - 0s - 2ms/step - loss: 0.3119 - sparse_categorical_accuracy: 0.8656 - val_loss: 0.3336 - val_sparse_categorical_accuracy: 0.8669 - learning_rate: 1.0000e-03
+## 90/90 - 0s - 1ms/step - loss: 0.3082 - sparse_categorical_accuracy: 0.8646 - val_loss: 0.4187 - val_sparse_categorical_accuracy: 0.7906 - learning_rate: 1.0000e-03
 ## Epoch 18/500
-## 90/90 - 0s - 2ms/step - loss: 0.3057 - sparse_categorical_accuracy: 0.8646 - val_loss: 0.5241 - val_sparse_categorical_accuracy: 0.7351 - learning_rate: 1.0000e-03
+## 90/90 - 0s - 1ms/step - loss: 0.3022 - sparse_categorical_accuracy: 0.8715 - val_loss: 0.6302 - val_sparse_categorical_accuracy: 0.6824 - learning_rate: 1.0000e-03
 ## Epoch 19/500
-## 90/90 - 0s - 2ms/step - loss: 0.3042 - sparse_categorical_accuracy: 0.8691 - val_loss: 0.4023 - val_sparse_categorical_accuracy: 0.8128 - learning_rate: 1.0000e-03
+## 90/90 - 0s - 1ms/step - loss: 0.2996 - sparse_categorical_accuracy: 0.8764 - val_loss: 0.3537 - val_sparse_categorical_accuracy: 0.8447 - learning_rate: 1.0000e-03
 ## Epoch 20/500
-## 90/90 - 0s - 2ms/step - loss: 0.3037 - sparse_categorical_accuracy: 0.8677 - val_loss: 0.4818 - val_sparse_categorical_accuracy: 0.7670 - learning_rate: 1.0000e-03
+## 90/90 - 0s - 1ms/step - loss: 0.3053 - sparse_categorical_accuracy: 0.8715 - val_loss: 0.4925 - val_sparse_categorical_accuracy: 0.7309 - learning_rate: 1.0000e-03
 ## Epoch 21/500
-## 90/90 - 0s - 2ms/step - loss: 0.2834 - sparse_categorical_accuracy: 0.8795 - val_loss: 0.3017 - val_sparse_categorical_accuracy: 0.8696 - learning_rate: 1.0000e-03
+## 90/90 - 0s - 1ms/step - loss: 0.2827 - sparse_categorical_accuracy: 0.8826 - val_loss: 0.3893 - val_sparse_categorical_accuracy: 0.7809 - learning_rate: 1.0000e-03
 ## Epoch 22/500
-## 90/90 - 0s - 1ms/step - loss: 0.2879 - sparse_categorical_accuracy: 0.8819 - val_loss: 0.4130 - val_sparse_categorical_accuracy: 0.8003 - learning_rate: 1.0000e-03
+## 90/90 - 0s - 2ms/step - loss: 0.2892 - sparse_categorical_accuracy: 0.8795 - val_loss: 0.3128 - val_sparse_categorical_accuracy: 0.8863 - learning_rate: 1.0000e-03
 ## Epoch 23/500
-## 90/90 - 0s - 1ms/step - loss: 0.2778 - sparse_categorical_accuracy: 0.8844 - val_loss: 0.3149 - val_sparse_categorical_accuracy: 0.8599 - learning_rate: 1.0000e-03
+## 90/90 - 0s - 2ms/step - loss: 0.2718 - sparse_categorical_accuracy: 0.8889 - val_loss: 0.3083 - val_sparse_categorical_accuracy: 0.8571 - learning_rate: 1.0000e-03
 ## Epoch 24/500
-## 90/90 - 0s - 2ms/step - loss: 0.2743 - sparse_categorical_accuracy: 0.8861 - val_loss: 0.3331 - val_sparse_categorical_accuracy: 0.8460 - learning_rate: 1.0000e-03
+## 90/90 - 0s - 1ms/step - loss: 0.2691 - sparse_categorical_accuracy: 0.8934 - val_loss: 0.5609 - val_sparse_categorical_accuracy: 0.7420 - learning_rate: 1.0000e-03
 ## Epoch 25/500
-## 90/90 - 0s - 1ms/step - loss: 0.2654 - sparse_categorical_accuracy: 0.8892 - val_loss: 0.4245 - val_sparse_categorical_accuracy: 0.7892 - learning_rate: 1.0000e-03
+## 90/90 - 0s - 1ms/step - loss: 0.2635 - sparse_categorical_accuracy: 0.8948 - val_loss: 0.5264 - val_sparse_categorical_accuracy: 0.7573 - learning_rate: 1.0000e-03
 ## Epoch 26/500
-## 90/90 - 0s - 1ms/step - loss: 0.3058 - sparse_categorical_accuracy: 0.8660 - val_loss: 0.3187 - val_sparse_categorical_accuracy: 0.8669 - learning_rate: 1.0000e-03
+## 90/90 - 0s - 1ms/step - loss: 0.3109 - sparse_categorical_accuracy: 0.8667 - val_loss: 0.3934 - val_sparse_categorical_accuracy: 0.7878 - learning_rate: 1.0000e-03
 ## Epoch 27/500
-## 90/90 - 0s - 1ms/step - loss: 0.2619 - sparse_categorical_accuracy: 0.8965 - val_loss: 0.4151 - val_sparse_categorical_accuracy: 0.7850 - learning_rate: 1.0000e-03
+## 90/90 - 0s - 2ms/step - loss: 0.2599 - sparse_categorical_accuracy: 0.8969 - val_loss: 0.2974 - val_sparse_categorical_accuracy: 0.8558 - learning_rate: 1.0000e-03
 ## Epoch 28/500
-## 90/90 - 0s - 1ms/step - loss: 0.2785 - sparse_categorical_accuracy: 0.8792 - val_loss: 0.3807 - val_sparse_categorical_accuracy: 0.8308 - learning_rate: 1.0000e-03
+## 90/90 - 0s - 1ms/step - loss: 0.2805 - sparse_categorical_accuracy: 0.8823 - val_loss: 0.4026 - val_sparse_categorical_accuracy: 0.8183 - learning_rate: 1.0000e-03
 ## Epoch 29/500
-## 90/90 - 0s - 2ms/step - loss: 0.2594 - sparse_categorical_accuracy: 0.8878 - val_loss: 0.4138 - val_sparse_categorical_accuracy: 0.8128 - learning_rate: 1.0000e-03
+## 90/90 - 0s - 2ms/step - loss: 0.2517 - sparse_categorical_accuracy: 0.8948 - val_loss: 0.2699 - val_sparse_categorical_accuracy: 0.8849 - learning_rate: 1.0000e-03
 ## Epoch 30/500
-## 90/90 - 0s - 2ms/step - loss: 0.2644 - sparse_categorical_accuracy: 0.8878 - val_loss: 0.5111 - val_sparse_categorical_accuracy: 0.7323 - learning_rate: 1.0000e-03
+## 90/90 - 0s - 1ms/step - loss: 0.2581 - sparse_categorical_accuracy: 0.8920 - val_loss: 0.2743 - val_sparse_categorical_accuracy: 0.8752 - learning_rate: 1.0000e-03
 ## Epoch 31/500
-## 90/90 - 0s - 2ms/step - loss: 0.2653 - sparse_categorical_accuracy: 0.8924 - val_loss: 0.2788 - val_sparse_categorical_accuracy: 0.8863 - learning_rate: 1.0000e-03
+## 90/90 - 0s - 1ms/step - loss: 0.2550 - sparse_categorical_accuracy: 0.8986 - val_loss: 0.2936 - val_sparse_categorical_accuracy: 0.8655 - learning_rate: 1.0000e-03
 ## Epoch 32/500
-## 90/90 - 0s - 2ms/step - loss: 0.2546 - sparse_categorical_accuracy: 0.8903 - val_loss: 0.2598 - val_sparse_categorical_accuracy: 0.8821 - learning_rate: 1.0000e-03
+## 90/90 - 0s - 2ms/step - loss: 0.2528 - sparse_categorical_accuracy: 0.8934 - val_loss: 0.2528 - val_sparse_categorical_accuracy: 0.8904 - learning_rate: 1.0000e-03
 ## Epoch 33/500
-## 90/90 - 0s - 2ms/step - loss: 0.2408 - sparse_categorical_accuracy: 0.9062 - val_loss: 0.2632 - val_sparse_categorical_accuracy: 0.8793 - learning_rate: 1.0000e-03
+## 90/90 - 0s - 1ms/step - loss: 0.2390 - sparse_categorical_accuracy: 0.9062 - val_loss: 0.2961 - val_sparse_categorical_accuracy: 0.8571 - learning_rate: 1.0000e-03
 ## Epoch 34/500
-## 90/90 - 0s - 2ms/step - loss: 0.2373 - sparse_categorical_accuracy: 0.9052 - val_loss: 0.2966 - val_sparse_categorical_accuracy: 0.8710 - learning_rate: 1.0000e-03
+## 90/90 - 0s - 1ms/step - loss: 0.2327 - sparse_categorical_accuracy: 0.9049 - val_loss: 0.3039 - val_sparse_categorical_accuracy: 0.8558 - learning_rate: 1.0000e-03
 ## Epoch 35/500
-## 90/90 - 0s - 1ms/step - loss: 0.2560 - sparse_categorical_accuracy: 0.8903 - val_loss: 0.6435 - val_sparse_categorical_accuracy: 0.7309 - learning_rate: 1.0000e-03
+## 90/90 - 0s - 1ms/step - loss: 0.2447 - sparse_categorical_accuracy: 0.9021 - val_loss: 0.3123 - val_sparse_categorical_accuracy: 0.8710 - learning_rate: 1.0000e-03
 ## Epoch 36/500
-## 90/90 - 0s - 1ms/step - loss: 0.2391 - sparse_categorical_accuracy: 0.9035 - val_loss: 0.2705 - val_sparse_categorical_accuracy: 0.8807 - learning_rate: 1.0000e-03
+## 90/90 - 0s - 1ms/step - loss: 0.2340 - sparse_categorical_accuracy: 0.9069 - val_loss: 0.2912 - val_sparse_categorical_accuracy: 0.8821 - learning_rate: 1.0000e-03
 ## Epoch 37/500
-## 90/90 - 0s - 2ms/step - loss: 0.2384 - sparse_categorical_accuracy: 0.8993 - val_loss: 0.2720 - val_sparse_categorical_accuracy: 0.8738 - learning_rate: 1.0000e-03
+## 90/90 - 0s - 1ms/step - loss: 0.2352 - sparse_categorical_accuracy: 0.8997 - val_loss: 0.3983 - val_sparse_categorical_accuracy: 0.8017 - learning_rate: 1.0000e-03
 ## Epoch 38/500
-## 90/90 - 0s - 2ms/step - loss: 0.2339 - sparse_categorical_accuracy: 0.9080 - val_loss: 0.9019 - val_sparse_categorical_accuracy: 0.7309 - learning_rate: 1.0000e-03
+## 90/90 - 0s - 1ms/step - loss: 0.2294 - sparse_categorical_accuracy: 0.9139 - val_loss: 0.3342 - val_sparse_categorical_accuracy: 0.8447 - learning_rate: 1.0000e-03
 ## Epoch 39/500
-## 90/90 - 0s - 2ms/step - loss: 0.2404 - sparse_categorical_accuracy: 0.9038 - val_loss: 0.4007 - val_sparse_categorical_accuracy: 0.8294 - learning_rate: 1.0000e-03
+## 90/90 - 0s - 1ms/step - loss: 0.2268 - sparse_categorical_accuracy: 0.9135 - val_loss: 0.4730 - val_sparse_categorical_accuracy: 0.7975 - learning_rate: 1.0000e-03
 ## Epoch 40/500
-## 90/90 - 0s - 2ms/step - loss: 0.2265 - sparse_categorical_accuracy: 0.9104 - val_loss: 0.6253 - val_sparse_categorical_accuracy: 0.7559 - learning_rate: 1.0000e-03
+## 90/90 - 0s - 1ms/step - loss: 0.2202 - sparse_categorical_accuracy: 0.9111 - val_loss: 0.3007 - val_sparse_categorical_accuracy: 0.8738 - learning_rate: 1.0000e-03
 ## Epoch 41/500
-## 90/90 - 0s - 2ms/step - loss: 0.2159 - sparse_categorical_accuracy: 0.9174 - val_loss: 0.2307 - val_sparse_categorical_accuracy: 0.8974 - learning_rate: 1.0000e-03
+## 90/90 - 0s - 1ms/step - loss: 0.2064 - sparse_categorical_accuracy: 0.9233 - val_loss: 0.2924 - val_sparse_categorical_accuracy: 0.8849 - learning_rate: 1.0000e-03
 ## Epoch 42/500
-## 90/90 - 0s - 2ms/step - loss: 0.2081 - sparse_categorical_accuracy: 0.9271 - val_loss: 0.5934 - val_sparse_categorical_accuracy: 0.7434 - learning_rate: 1.0000e-03
+## 90/90 - 0s - 1ms/step - loss: 0.1966 - sparse_categorical_accuracy: 0.9316 - val_loss: 0.8303 - val_sparse_categorical_accuracy: 0.6935 - learning_rate: 1.0000e-03
 ## Epoch 43/500
-## 90/90 - 0s - 2ms/step - loss: 0.2135 - sparse_categorical_accuracy: 0.9153 - val_loss: 0.2809 - val_sparse_categorical_accuracy: 0.8613 - learning_rate: 1.0000e-03
+## 90/90 - 0s - 2ms/step - loss: 0.2074 - sparse_categorical_accuracy: 0.9212 - val_loss: 0.2334 - val_sparse_categorical_accuracy: 0.9043 - learning_rate: 1.0000e-03
 ## Epoch 44/500
-## 90/90 - 0s - 2ms/step - loss: 0.2152 - sparse_categorical_accuracy: 0.9149 - val_loss: 0.3832 - val_sparse_categorical_accuracy: 0.8211 - learning_rate: 1.0000e-03
+## 90/90 - 0s - 1ms/step - loss: 0.1961 - sparse_categorical_accuracy: 0.9313 - val_loss: 0.2343 - val_sparse_categorical_accuracy: 0.9182 - learning_rate: 1.0000e-03
 ## Epoch 45/500
-## 90/90 - 0s - 2ms/step - loss: 0.2060 - sparse_categorical_accuracy: 0.9233 - val_loss: 0.3182 - val_sparse_categorical_accuracy: 0.8682 - learning_rate: 1.0000e-03
+## 90/90 - 0s - 1ms/step - loss: 0.1923 - sparse_categorical_accuracy: 0.9302 - val_loss: 1.1167 - val_sparse_categorical_accuracy: 0.6755 - learning_rate: 1.0000e-03
 ## Epoch 46/500
-## 90/90 - 0s - 2ms/step - loss: 0.2023 - sparse_categorical_accuracy: 0.9229 - val_loss: 0.2104 - val_sparse_categorical_accuracy: 0.9140 - learning_rate: 1.0000e-03
+## 90/90 - 0s - 1ms/step - loss: 0.1823 - sparse_categorical_accuracy: 0.9347 - val_loss: 0.8034 - val_sparse_categorical_accuracy: 0.6935 - learning_rate: 1.0000e-03
 ## Epoch 47/500
-## 90/90 - 0s - 1ms/step - loss: 0.1855 - sparse_categorical_accuracy: 0.9309 - val_loss: 0.2882 - val_sparse_categorical_accuracy: 0.8877 - learning_rate: 1.0000e-03
+## 90/90 - 0s - 1ms/step - loss: 0.1635 - sparse_categorical_accuracy: 0.9444 - val_loss: 0.7104 - val_sparse_categorical_accuracy: 0.7115 - learning_rate: 1.0000e-03
 ## Epoch 48/500
-## 90/90 - 0s - 1ms/step - loss: 0.1789 - sparse_categorical_accuracy: 0.9306 - val_loss: 0.4009 - val_sparse_categorical_accuracy: 0.8197 - learning_rate: 1.0000e-03
+## 90/90 - 0s - 2ms/step - loss: 0.1587 - sparse_categorical_accuracy: 0.9434 - val_loss: 0.2001 - val_sparse_categorical_accuracy: 0.9196 - learning_rate: 1.0000e-03
 ## Epoch 49/500
-## 90/90 - 0s - 1ms/step - loss: 0.1643 - sparse_categorical_accuracy: 0.9417 - val_loss: 0.2160 - val_sparse_categorical_accuracy: 0.9071 - learning_rate: 1.0000e-03
+## 90/90 - 0s - 1ms/step - loss: 0.1492 - sparse_categorical_accuracy: 0.9483 - val_loss: 0.3399 - val_sparse_categorical_accuracy: 0.8294 - learning_rate: 1.0000e-03
 ## Epoch 50/500
-## 90/90 - 0s - 1ms/step - loss: 0.1599 - sparse_categorical_accuracy: 0.9462 - val_loss: 0.6680 - val_sparse_categorical_accuracy: 0.6768 - learning_rate: 1.0000e-03
+## 90/90 - 0s - 1ms/step - loss: 0.1738 - sparse_categorical_accuracy: 0.9382 - val_loss: 0.2680 - val_sparse_categorical_accuracy: 0.8890 - learning_rate: 1.0000e-03
 ## Epoch 51/500
-## 90/90 - 0s - 1ms/step - loss: 0.1623 - sparse_categorical_accuracy: 0.9420 - val_loss: 1.2733 - val_sparse_categorical_accuracy: 0.7254 - learning_rate: 1.0000e-03
+## 90/90 - 0s - 1ms/step - loss: 0.1530 - sparse_categorical_accuracy: 0.9486 - val_loss: 0.2891 - val_sparse_categorical_accuracy: 0.8544 - learning_rate: 1.0000e-03
 ## Epoch 52/500
-## 90/90 - 0s - 2ms/step - loss: 0.1465 - sparse_categorical_accuracy: 0.9507 - val_loss: 1.1664 - val_sparse_categorical_accuracy: 0.6976 - learning_rate: 1.0000e-03
+## 90/90 - 0s - 1ms/step - loss: 0.1320 - sparse_categorical_accuracy: 0.9538 - val_loss: 1.7182 - val_sparse_categorical_accuracy: 0.7032 - learning_rate: 1.0000e-03
 ## Epoch 53/500
-## 90/90 - 0s - 1ms/step - loss: 0.1453 - sparse_categorical_accuracy: 0.9465 - val_loss: 1.3677 - val_sparse_categorical_accuracy: 0.7074 - learning_rate: 1.0000e-03
+## 90/90 - 0s - 1ms/step - loss: 0.1277 - sparse_categorical_accuracy: 0.9559 - val_loss: 1.4269 - val_sparse_categorical_accuracy: 0.7254 - learning_rate: 1.0000e-03
 ## Epoch 54/500
-## 90/90 - 0s - 1ms/step - loss: 0.1370 - sparse_categorical_accuracy: 0.9563 - val_loss: 0.2190 - val_sparse_categorical_accuracy: 0.9098 - learning_rate: 1.0000e-03
+## 90/90 - 0s - 1ms/step - loss: 0.1249 - sparse_categorical_accuracy: 0.9611 - val_loss: 0.7962 - val_sparse_categorical_accuracy: 0.7365 - learning_rate: 1.0000e-03
 ## Epoch 55/500
-## 90/90 - 0s - 1ms/step - loss: 0.1319 - sparse_categorical_accuracy: 0.9563 - val_loss: 0.2376 - val_sparse_categorical_accuracy: 0.8932 - learning_rate: 1.0000e-03
+## 90/90 - 0s - 1ms/step - loss: 0.1214 - sparse_categorical_accuracy: 0.9611 - val_loss: 0.2158 - val_sparse_categorical_accuracy: 0.9071 - learning_rate: 1.0000e-03
 ## Epoch 56/500
-## 90/90 - 0s - 2ms/step - loss: 0.1354 - sparse_categorical_accuracy: 0.9549 - val_loss: 0.1614 - val_sparse_categorical_accuracy: 0.9334 - learning_rate: 1.0000e-03
+## 90/90 - 0s - 1ms/step - loss: 0.1258 - sparse_categorical_accuracy: 0.9604 - val_loss: 0.6349 - val_sparse_categorical_accuracy: 0.7642 - learning_rate: 1.0000e-03
 ## Epoch 57/500
-## 90/90 - 0s - 2ms/step - loss: 0.1298 - sparse_categorical_accuracy: 0.9590 - val_loss: 0.1651 - val_sparse_categorical_accuracy: 0.9376 - learning_rate: 1.0000e-03
+## 90/90 - 0s - 1ms/step - loss: 0.1268 - sparse_categorical_accuracy: 0.9604 - val_loss: 0.2129 - val_sparse_categorical_accuracy: 0.9071 - learning_rate: 1.0000e-03
 ## Epoch 58/500
-## 90/90 - 0s - 2ms/step - loss: 0.1188 - sparse_categorical_accuracy: 0.9569 - val_loss: 0.2961 - val_sparse_categorical_accuracy: 0.8766 - learning_rate: 1.0000e-03
+## 90/90 - 0s - 1ms/step - loss: 0.1088 - sparse_categorical_accuracy: 0.9639 - val_loss: 0.7215 - val_sparse_categorical_accuracy: 0.6657 - learning_rate: 1.0000e-03
 ## Epoch 59/500
-## 90/90 - 0s - 2ms/step - loss: 0.1295 - sparse_categorical_accuracy: 0.9545 - val_loss: 0.1905 - val_sparse_categorical_accuracy: 0.9209 - learning_rate: 1.0000e-03
+## 90/90 - 0s - 2ms/step - loss: 0.1219 - sparse_categorical_accuracy: 0.9569 - val_loss: 0.1600 - val_sparse_categorical_accuracy: 0.9320 - learning_rate: 1.0000e-03
 ## Epoch 60/500
-## 90/90 - 0s - 1ms/step - loss: 0.1281 - sparse_categorical_accuracy: 0.9552 - val_loss: 0.1835 - val_sparse_categorical_accuracy: 0.9140 - learning_rate: 1.0000e-03
+## 90/90 - 0s - 1ms/step - loss: 0.1190 - sparse_categorical_accuracy: 0.9642 - val_loss: 0.3411 - val_sparse_categorical_accuracy: 0.8558 - learning_rate: 1.0000e-03
 ## Epoch 61/500
-## 90/90 - 0s - 1ms/step - loss: 0.1176 - sparse_categorical_accuracy: 0.9646 - val_loss: 1.6040 - val_sparse_categorical_accuracy: 0.6741 - learning_rate: 1.0000e-03
+## 90/90 - 0s - 1ms/step - loss: 0.1177 - sparse_categorical_accuracy: 0.9653 - val_loss: 0.3984 - val_sparse_categorical_accuracy: 0.8502 - learning_rate: 1.0000e-03
 ## Epoch 62/500
-## 90/90 - 0s - 1ms/step - loss: 0.1363 - sparse_categorical_accuracy: 0.9500 - val_loss: 2.0555 - val_sparse_categorical_accuracy: 0.6963 - learning_rate: 1.0000e-03
+## 90/90 - 0s - 1ms/step - loss: 0.1309 - sparse_categorical_accuracy: 0.9542 - val_loss: 0.3462 - val_sparse_categorical_accuracy: 0.8211 - learning_rate: 1.0000e-03
 ## Epoch 63/500
-## 90/90 - 0s - 1ms/step - loss: 0.1358 - sparse_categorical_accuracy: 0.9483 - val_loss: 3.2565 - val_sparse_categorical_accuracy: 0.6352 - learning_rate: 1.0000e-03
+## 90/90 - 0s - 1ms/step - loss: 0.1291 - sparse_categorical_accuracy: 0.9514 - val_loss: 0.2303 - val_sparse_categorical_accuracy: 0.8960 - learning_rate: 1.0000e-03
 ## Epoch 64/500
-## 90/90 - 0s - 2ms/step - loss: 0.1276 - sparse_categorical_accuracy: 0.9590 - val_loss: 0.3056 - val_sparse_categorical_accuracy: 0.8669 - learning_rate: 1.0000e-03
+## 90/90 - 0s - 1ms/step - loss: 0.1200 - sparse_categorical_accuracy: 0.9628 - val_loss: 0.1735 - val_sparse_categorical_accuracy: 0.9223 - learning_rate: 1.0000e-03
 ## Epoch 65/500
-## 90/90 - 0s - 2ms/step - loss: 0.1189 - sparse_categorical_accuracy: 0.9590 - val_loss: 0.3571 - val_sparse_categorical_accuracy: 0.8530 - learning_rate: 1.0000e-03
+## 90/90 - 0s - 1ms/step - loss: 0.1138 - sparse_categorical_accuracy: 0.9625 - val_loss: 0.5628 - val_sparse_categorical_accuracy: 0.7920 - learning_rate: 1.0000e-03
 ## Epoch 66/500
-## 90/90 - 0s - 2ms/step - loss: 0.1111 - sparse_categorical_accuracy: 0.9653 - val_loss: 0.1403 - val_sparse_categorical_accuracy: 0.9473 - learning_rate: 1.0000e-03
+## 90/90 - 0s - 2ms/step - loss: 0.1065 - sparse_categorical_accuracy: 0.9667 - val_loss: 0.1540 - val_sparse_categorical_accuracy: 0.9417 - learning_rate: 1.0000e-03
 ## Epoch 67/500
-## 90/90 - 0s - 2ms/step - loss: 0.1120 - sparse_categorical_accuracy: 0.9635 - val_loss: 1.6044 - val_sparse_categorical_accuracy: 0.6477 - learning_rate: 1.0000e-03
+## 90/90 - 0s - 1ms/step - loss: 0.1156 - sparse_categorical_accuracy: 0.9618 - val_loss: 0.2319 - val_sparse_categorical_accuracy: 0.9043 - learning_rate: 1.0000e-03
 ## Epoch 68/500
-## 90/90 - 0s - 2ms/step - loss: 0.1123 - sparse_categorical_accuracy: 0.9632 - val_loss: 0.6339 - val_sparse_categorical_accuracy: 0.7989 - learning_rate: 1.0000e-03
+## 90/90 - 0s - 1ms/step - loss: 0.1100 - sparse_categorical_accuracy: 0.9597 - val_loss: 0.8668 - val_sparse_categorical_accuracy: 0.7406 - learning_rate: 1.0000e-03
 ## Epoch 69/500
-## 90/90 - 0s - 2ms/step - loss: 0.1150 - sparse_categorical_accuracy: 0.9611 - val_loss: 0.3409 - val_sparse_categorical_accuracy: 0.8585 - learning_rate: 1.0000e-03
+## 90/90 - 0s - 2ms/step - loss: 0.1165 - sparse_categorical_accuracy: 0.9618 - val_loss: 0.1460 - val_sparse_categorical_accuracy: 0.9515 - learning_rate: 1.0000e-03
 ## Epoch 70/500
-## 90/90 - 0s - 1ms/step - loss: 0.1120 - sparse_categorical_accuracy: 0.9597 - val_loss: 0.1705 - val_sparse_categorical_accuracy: 0.9209 - learning_rate: 1.0000e-03
+## 90/90 - 0s - 1ms/step - loss: 0.1178 - sparse_categorical_accuracy: 0.9608 - val_loss: 0.8988 - val_sparse_categorical_accuracy: 0.7046 - learning_rate: 1.0000e-03
 ## Epoch 71/500
-## 90/90 - 0s - 2ms/step - loss: 0.1070 - sparse_categorical_accuracy: 0.9646 - val_loss: 0.3976 - val_sparse_categorical_accuracy: 0.8488 - learning_rate: 1.0000e-03
+## 90/90 - 0s - 1ms/step - loss: 0.1068 - sparse_categorical_accuracy: 0.9642 - val_loss: 0.3645 - val_sparse_categorical_accuracy: 0.8585 - learning_rate: 1.0000e-03
 ## Epoch 72/500
-## 90/90 - 0s - 2ms/step - loss: 0.1076 - sparse_categorical_accuracy: 0.9663 - val_loss: 0.1426 - val_sparse_categorical_accuracy: 0.9473 - learning_rate: 1.0000e-03
+## 90/90 - 0s - 1ms/step - loss: 0.1106 - sparse_categorical_accuracy: 0.9653 - val_loss: 0.2813 - val_sparse_categorical_accuracy: 0.8932 - learning_rate: 1.0000e-03
 ## Epoch 73/500
-## 90/90 - 0s - 2ms/step - loss: 0.1025 - sparse_categorical_accuracy: 0.9677 - val_loss: 0.3274 - val_sparse_categorical_accuracy: 0.8613 - learning_rate: 1.0000e-03
+## 90/90 - 0s - 1ms/step - loss: 0.1117 - sparse_categorical_accuracy: 0.9632 - val_loss: 1.8641 - val_sparse_categorical_accuracy: 0.6741 - learning_rate: 1.0000e-03
 ## Epoch 74/500
-## 90/90 - 0s - 2ms/step - loss: 0.1056 - sparse_categorical_accuracy: 0.9628 - val_loss: 0.2165 - val_sparse_categorical_accuracy: 0.9182 - learning_rate: 1.0000e-03
+## 90/90 - 0s - 1ms/step - loss: 0.1058 - sparse_categorical_accuracy: 0.9625 - val_loss: 1.4979 - val_sparse_categorical_accuracy: 0.6893 - learning_rate: 1.0000e-03
 ## Epoch 75/500
-## 90/90 - 0s - 1ms/step - loss: 0.1303 - sparse_categorical_accuracy: 0.9524 - val_loss: 0.6197 - val_sparse_categorical_accuracy: 0.7282 - learning_rate: 1.0000e-03
+## 90/90 - 0s - 1ms/step - loss: 0.1223 - sparse_categorical_accuracy: 0.9594 - val_loss: 0.1508 - val_sparse_categorical_accuracy: 0.9473 - learning_rate: 1.0000e-03
 ## Epoch 76/500
-## 90/90 - 0s - 1ms/step - loss: 0.1205 - sparse_categorical_accuracy: 0.9590 - val_loss: 0.2972 - val_sparse_categorical_accuracy: 0.8793 - learning_rate: 1.0000e-03
+## 90/90 - 0s - 2ms/step - loss: 0.1140 - sparse_categorical_accuracy: 0.9601 - val_loss: 0.1400 - val_sparse_categorical_accuracy: 0.9320 - learning_rate: 1.0000e-03
 ## Epoch 77/500
-## 90/90 - 0s - 1ms/step - loss: 0.1061 - sparse_categorical_accuracy: 0.9632 - val_loss: 0.2158 - val_sparse_categorical_accuracy: 0.9182 - learning_rate: 1.0000e-03
+## 90/90 - 0s - 1ms/step - loss: 0.1050 - sparse_categorical_accuracy: 0.9639 - val_loss: 0.2169 - val_sparse_categorical_accuracy: 0.9126 - learning_rate: 1.0000e-03
 ## Epoch 78/500
-## 90/90 - 0s - 1ms/step - loss: 0.0959 - sparse_categorical_accuracy: 0.9705 - val_loss: 0.1449 - val_sparse_categorical_accuracy: 0.9362 - learning_rate: 1.0000e-03
+## 90/90 - 0s - 1ms/step - loss: 0.0986 - sparse_categorical_accuracy: 0.9712 - val_loss: 0.3240 - val_sparse_categorical_accuracy: 0.8724 - learning_rate: 1.0000e-03
 ## Epoch 79/500
-## 90/90 - 0s - 2ms/step - loss: 0.0995 - sparse_categorical_accuracy: 0.9674 - val_loss: 0.1231 - val_sparse_categorical_accuracy: 0.9598 - learning_rate: 1.0000e-03
+## 90/90 - 0s - 1ms/step - loss: 0.1081 - sparse_categorical_accuracy: 0.9632 - val_loss: 0.1525 - val_sparse_categorical_accuracy: 0.9362 - learning_rate: 1.0000e-03
 ## Epoch 80/500
-## 90/90 - 0s - 1ms/step - loss: 0.1027 - sparse_categorical_accuracy: 0.9677 - val_loss: 0.5311 - val_sparse_categorical_accuracy: 0.7712 - learning_rate: 1.0000e-03
+## 90/90 - 0s - 1ms/step - loss: 0.0995 - sparse_categorical_accuracy: 0.9667 - val_loss: 0.3123 - val_sparse_categorical_accuracy: 0.8641 - learning_rate: 1.0000e-03
 ## Epoch 81/500
-## 90/90 - 0s - 2ms/step - loss: 0.1012 - sparse_categorical_accuracy: 0.9677 - val_loss: 0.1622 - val_sparse_categorical_accuracy: 0.9348 - learning_rate: 1.0000e-03
+## 90/90 - 0s - 1ms/step - loss: 0.1058 - sparse_categorical_accuracy: 0.9632 - val_loss: 0.2851 - val_sparse_categorical_accuracy: 0.8779 - learning_rate: 1.0000e-03
 ## Epoch 82/500
-## 90/90 - 0s - 1ms/step - loss: 0.0929 - sparse_categorical_accuracy: 0.9663 - val_loss: 0.1457 - val_sparse_categorical_accuracy: 0.9417 - learning_rate: 1.0000e-03
+## 90/90 - 0s - 1ms/step - loss: 0.0958 - sparse_categorical_accuracy: 0.9681 - val_loss: 0.1417 - val_sparse_categorical_accuracy: 0.9362 - learning_rate: 1.0000e-03
 ## Epoch 83/500
-## 90/90 - 0s - 1ms/step - loss: 0.1008 - sparse_categorical_accuracy: 0.9663 - val_loss: 1.6042 - val_sparse_categorical_accuracy: 0.6879 - learning_rate: 1.0000e-03
+## 90/90 - 0s - 1ms/step - loss: 0.1005 - sparse_categorical_accuracy: 0.9660 - val_loss: 0.7325 - val_sparse_categorical_accuracy: 0.7157 - learning_rate: 1.0000e-03
 ## Epoch 84/500
-## 90/90 - 0s - 1ms/step - loss: 0.0981 - sparse_categorical_accuracy: 0.9653 - val_loss: 0.1732 - val_sparse_categorical_accuracy: 0.9307 - learning_rate: 1.0000e-03
+## 90/90 - 0s - 2ms/step - loss: 0.0966 - sparse_categorical_accuracy: 0.9684 - val_loss: 0.1322 - val_sparse_categorical_accuracy: 0.9404 - learning_rate: 1.0000e-03
 ## Epoch 85/500
-## 90/90 - 0s - 1ms/step - loss: 0.1006 - sparse_categorical_accuracy: 0.9674 - val_loss: 1.8495 - val_sparse_categorical_accuracy: 0.6976 - learning_rate: 1.0000e-03
+## 90/90 - 0s - 2ms/step - loss: 0.1018 - sparse_categorical_accuracy: 0.9667 - val_loss: 0.1202 - val_sparse_categorical_accuracy: 0.9473 - learning_rate: 1.0000e-03
 ## Epoch 86/500
-## 90/90 - 0s - 1ms/step - loss: 0.1005 - sparse_categorical_accuracy: 0.9660 - val_loss: 0.1263 - val_sparse_categorical_accuracy: 0.9459 - learning_rate: 1.0000e-03
+## 90/90 - 0s - 1ms/step - loss: 0.0995 - sparse_categorical_accuracy: 0.9649 - val_loss: 0.2956 - val_sparse_categorical_accuracy: 0.8766 - learning_rate: 1.0000e-03
 ## Epoch 87/500
-## 90/90 - 0s - 1ms/step - loss: 0.1054 - sparse_categorical_accuracy: 0.9632 - val_loss: 0.2135 - val_sparse_categorical_accuracy: 0.9140 - learning_rate: 1.0000e-03
+## 90/90 - 0s - 1ms/step - loss: 0.1068 - sparse_categorical_accuracy: 0.9632 - val_loss: 0.3050 - val_sparse_categorical_accuracy: 0.8710 - learning_rate: 1.0000e-03
 ## Epoch 88/500
-## 90/90 - 0s - 2ms/step - loss: 0.1051 - sparse_categorical_accuracy: 0.9656 - val_loss: 1.3499 - val_sparse_categorical_accuracy: 0.6630 - learning_rate: 1.0000e-03
+## 90/90 - 0s - 1ms/step - loss: 0.1002 - sparse_categorical_accuracy: 0.9674 - val_loss: 0.1675 - val_sparse_categorical_accuracy: 0.9196 - learning_rate: 1.0000e-03
 ## Epoch 89/500
-## 90/90 - 0s - 2ms/step - loss: 0.1037 - sparse_categorical_accuracy: 0.9615 - val_loss: 0.2271 - val_sparse_categorical_accuracy: 0.9029 - learning_rate: 1.0000e-03
+## 90/90 - 0s - 1ms/step - loss: 0.1042 - sparse_categorical_accuracy: 0.9608 - val_loss: 0.7157 - val_sparse_categorical_accuracy: 0.7115 - learning_rate: 1.0000e-03
 ## Epoch 90/500
-## 90/90 - 0s - 2ms/step - loss: 0.0976 - sparse_categorical_accuracy: 0.9670 - val_loss: 0.7824 - val_sparse_categorical_accuracy: 0.7129 - learning_rate: 1.0000e-03
+## 90/90 - 0s - 1ms/step - loss: 0.0998 - sparse_categorical_accuracy: 0.9694 - val_loss: 0.1576 - val_sparse_categorical_accuracy: 0.9251 - learning_rate: 1.0000e-03
 ## Epoch 91/500
-## 90/90 - 0s - 1ms/step - loss: 0.0995 - sparse_categorical_accuracy: 0.9677 - val_loss: 0.2369 - val_sparse_categorical_accuracy: 0.8918 - learning_rate: 1.0000e-03
+## 90/90 - 0s - 1ms/step - loss: 0.1187 - sparse_categorical_accuracy: 0.9583 - val_loss: 0.4125 - val_sparse_categorical_accuracy: 0.8031 - learning_rate: 1.0000e-03
 ## Epoch 92/500
-## 90/90 - 0s - 2ms/step - loss: 0.0986 - sparse_categorical_accuracy: 0.9670 - val_loss: 0.1185 - val_sparse_categorical_accuracy: 0.9584 - learning_rate: 1.0000e-03
+## 90/90 - 0s - 1ms/step - loss: 0.0973 - sparse_categorical_accuracy: 0.9688 - val_loss: 0.3084 - val_sparse_categorical_accuracy: 0.8696 - learning_rate: 1.0000e-03
 ## Epoch 93/500
-## 90/90 - 0s - 2ms/step - loss: 0.0957 - sparse_categorical_accuracy: 0.9670 - val_loss: 1.8619 - val_sparse_categorical_accuracy: 0.5270 - learning_rate: 1.0000e-03
+## 90/90 - 0s - 1ms/step - loss: 0.0972 - sparse_categorical_accuracy: 0.9677 - val_loss: 0.1446 - val_sparse_categorical_accuracy: 0.9334 - learning_rate: 1.0000e-03
 ## Epoch 94/500
-## 90/90 - 0s - 2ms/step - loss: 0.0999 - sparse_categorical_accuracy: 0.9642 - val_loss: 0.4507 - val_sparse_categorical_accuracy: 0.8350 - learning_rate: 1.0000e-03
+## 90/90 - 0s - 1ms/step - loss: 0.0965 - sparse_categorical_accuracy: 0.9632 - val_loss: 0.1265 - val_sparse_categorical_accuracy: 0.9584 - learning_rate: 1.0000e-03
 ## Epoch 95/500
-## 90/90 - 0s - 2ms/step - loss: 0.0939 - sparse_categorical_accuracy: 0.9653 - val_loss: 1.0529 - val_sparse_categorical_accuracy: 0.6893 - learning_rate: 1.0000e-03
+## 90/90 - 0s - 2ms/step - loss: 0.0927 - sparse_categorical_accuracy: 0.9667 - val_loss: 0.1507 - val_sparse_categorical_accuracy: 0.9376 - learning_rate: 1.0000e-03
 ## Epoch 96/500
-## 90/90 - 0s - 2ms/step - loss: 0.1066 - sparse_categorical_accuracy: 0.9608 - val_loss: 0.1416 - val_sparse_categorical_accuracy: 0.9417 - learning_rate: 1.0000e-03
+## 90/90 - 0s - 1ms/step - loss: 0.1050 - sparse_categorical_accuracy: 0.9618 - val_loss: 0.5005 - val_sparse_categorical_accuracy: 0.8086 - learning_rate: 1.0000e-03
 ## Epoch 97/500
-## 90/90 - 0s - 1ms/step - loss: 0.1068 - sparse_categorical_accuracy: 0.9618 - val_loss: 0.5324 - val_sparse_categorical_accuracy: 0.7961 - learning_rate: 1.0000e-03
+## 90/90 - 0s - 1ms/step - loss: 0.1134 - sparse_categorical_accuracy: 0.9597 - val_loss: 0.1732 - val_sparse_categorical_accuracy: 0.9390 - learning_rate: 1.0000e-03
 ## Epoch 98/500
-## 90/90 - 0s - 1ms/step - loss: 0.0916 - sparse_categorical_accuracy: 0.9694 - val_loss: 0.8154 - val_sparse_categorical_accuracy: 0.7656 - learning_rate: 1.0000e-03
+## 90/90 - 0s - 1ms/step - loss: 0.0946 - sparse_categorical_accuracy: 0.9698 - val_loss: 1.8677 - val_sparse_categorical_accuracy: 0.6949 - learning_rate: 1.0000e-03
 ## Epoch 99/500
-## 90/90 - 0s - 2ms/step - loss: 0.0942 - sparse_categorical_accuracy: 0.9639 - val_loss: 0.1314 - val_sparse_categorical_accuracy: 0.9542 - learning_rate: 1.0000e-03
+## 90/90 - 0s - 1ms/step - loss: 0.1018 - sparse_categorical_accuracy: 0.9632 - val_loss: 1.1308 - val_sparse_categorical_accuracy: 0.6782 - learning_rate: 1.0000e-03
 ## Epoch 100/500
-## 90/90 - 0s - 1ms/step - loss: 0.0938 - sparse_categorical_accuracy: 0.9698 - val_loss: 2.2256 - val_sparse_categorical_accuracy: 0.6366 - learning_rate: 1.0000e-03
+## 90/90 - 0s - 1ms/step - loss: 0.0990 - sparse_categorical_accuracy: 0.9674 - val_loss: 0.3846 - val_sparse_categorical_accuracy: 0.8558 - learning_rate: 1.0000e-03
 ## Epoch 101/500
-## 90/90 - 0s - 2ms/step - loss: 0.0853 - sparse_categorical_accuracy: 0.9750 - val_loss: 0.1660 - val_sparse_categorical_accuracy: 0.9376 - learning_rate: 1.0000e-03
+## 90/90 - 0s - 1ms/step - loss: 0.0891 - sparse_categorical_accuracy: 0.9705 - val_loss: 0.1480 - val_sparse_categorical_accuracy: 0.9362 - learning_rate: 1.0000e-03
 ## Epoch 102/500
-## 90/90 - 0s - 2ms/step - loss: 0.0957 - sparse_categorical_accuracy: 0.9691 - val_loss: 0.2444 - val_sparse_categorical_accuracy: 0.8932 - learning_rate: 1.0000e-03
+## 90/90 - 0s - 1ms/step - loss: 0.1003 - sparse_categorical_accuracy: 0.9677 - val_loss: 0.1713 - val_sparse_categorical_accuracy: 0.9390 - learning_rate: 1.0000e-03
 ## Epoch 103/500
-## 90/90 - 0s - 1ms/step - loss: 0.0949 - sparse_categorical_accuracy: 0.9660 - val_loss: 1.3661 - val_sparse_categorical_accuracy: 0.6519 - learning_rate: 1.0000e-03
+## 90/90 - 0s - 1ms/step - loss: 0.0910 - sparse_categorical_accuracy: 0.9722 - val_loss: 0.1415 - val_sparse_categorical_accuracy: 0.9542 - learning_rate: 1.0000e-03
 ## Epoch 104/500
-## 90/90 - 0s - 1ms/step - loss: 0.1033 - sparse_categorical_accuracy: 0.9622 - val_loss: 0.2086 - val_sparse_categorical_accuracy: 0.9126 - learning_rate: 1.0000e-03
+## 90/90 - 0s - 1ms/step - loss: 0.0998 - sparse_categorical_accuracy: 0.9622 - val_loss: 0.5677 - val_sparse_categorical_accuracy: 0.7753 - learning_rate: 1.0000e-03
 ## Epoch 105/500
-## 90/90 - 0s - 1ms/step - loss: 0.0981 - sparse_categorical_accuracy: 0.9663 - val_loss: 0.2216 - val_sparse_categorical_accuracy: 0.8988 - learning_rate: 1.0000e-03
+## 90/90 - 0s - 1ms/step - loss: 0.1037 - sparse_categorical_accuracy: 0.9649 - val_loss: 0.9369 - val_sparse_categorical_accuracy: 0.6241 - learning_rate: 1.0000e-03
 ## Epoch 106/500
-## 90/90 - 0s - 1ms/step - loss: 0.0895 - sparse_categorical_accuracy: 0.9681 - val_loss: 2.2126 - val_sparse_categorical_accuracy: 0.6269 - learning_rate: 1.0000e-03
+## 90/90 - 1s - 10ms/step - loss: 0.0850 - sparse_categorical_accuracy: 0.9708 - val_loss: 0.1388 - val_sparse_categorical_accuracy: 0.9487 - learning_rate: 5.0000e-04
 ## Epoch 107/500
-## 90/90 - 0s - 1ms/step - loss: 0.0939 - sparse_categorical_accuracy: 0.9670 - val_loss: 0.4274 - val_sparse_categorical_accuracy: 0.8128 - learning_rate: 1.0000e-03
+## 90/90 - 0s - 1ms/step - loss: 0.0833 - sparse_categorical_accuracy: 0.9743 - val_loss: 0.1413 - val_sparse_categorical_accuracy: 0.9376 - learning_rate: 5.0000e-04
 ## Epoch 108/500
-## 90/90 - 0s - 1ms/step - loss: 0.0894 - sparse_categorical_accuracy: 0.9719 - val_loss: 0.2745 - val_sparse_categorical_accuracy: 0.8946 - learning_rate: 1.0000e-03
+## 90/90 - 0s - 1ms/step - loss: 0.0775 - sparse_categorical_accuracy: 0.9767 - val_loss: 0.1223 - val_sparse_categorical_accuracy: 0.9445 - learning_rate: 5.0000e-04
 ## Epoch 109/500
-## 90/90 - 0s - 2ms/step - loss: 0.0937 - sparse_categorical_accuracy: 0.9722 - val_loss: 0.1248 - val_sparse_categorical_accuracy: 0.9487 - learning_rate: 1.0000e-03
+## 90/90 - 0s - 1ms/step - loss: 0.0837 - sparse_categorical_accuracy: 0.9753 - val_loss: 0.1275 - val_sparse_categorical_accuracy: 0.9528 - learning_rate: 5.0000e-04
 ## Epoch 110/500
-## 90/90 - 0s - 1ms/step - loss: 0.0965 - sparse_categorical_accuracy: 0.9656 - val_loss: 0.5962 - val_sparse_categorical_accuracy: 0.7989 - learning_rate: 1.0000e-03
+## 90/90 - 0s - 1ms/step - loss: 0.0859 - sparse_categorical_accuracy: 0.9705 - val_loss: 0.1209 - val_sparse_categorical_accuracy: 0.9487 - learning_rate: 5.0000e-04
 ## Epoch 111/500
-## 90/90 - 0s - 1ms/step - loss: 0.0825 - sparse_categorical_accuracy: 0.9726 - val_loss: 0.1671 - val_sparse_categorical_accuracy: 0.9307 - learning_rate: 1.0000e-03
+## 90/90 - 0s - 2ms/step - loss: 0.0780 - sparse_categorical_accuracy: 0.9757 - val_loss: 0.1142 - val_sparse_categorical_accuracy: 0.9584 - learning_rate: 5.0000e-04
 ## Epoch 112/500
-## 90/90 - 0s - 1ms/step - loss: 0.0879 - sparse_categorical_accuracy: 0.9694 - val_loss: 0.1298 - val_sparse_categorical_accuracy: 0.9390 - learning_rate: 1.0000e-03
+## 90/90 - 0s - 1ms/step - loss: 0.0801 - sparse_categorical_accuracy: 0.9736 - val_loss: 0.1316 - val_sparse_categorical_accuracy: 0.9459 - learning_rate: 5.0000e-04
 ## Epoch 113/500
-## 90/90 - 1s - 13ms/step - loss: 0.0785 - sparse_categorical_accuracy: 0.9757 - val_loss: 0.1539 - val_sparse_categorical_accuracy: 0.9348 - learning_rate: 5.0000e-04
+## 90/90 - 0s - 1ms/step - loss: 0.0740 - sparse_categorical_accuracy: 0.9753 - val_loss: 0.3385 - val_sparse_categorical_accuracy: 0.8544 - learning_rate: 5.0000e-04
 ## Epoch 114/500
-## 90/90 - 0s - 1ms/step - loss: 0.0914 - sparse_categorical_accuracy: 0.9722 - val_loss: 0.1216 - val_sparse_categorical_accuracy: 0.9542 - learning_rate: 5.0000e-04
+## 90/90 - 0s - 1ms/step - loss: 0.0907 - sparse_categorical_accuracy: 0.9722 - val_loss: 0.1222 - val_sparse_categorical_accuracy: 0.9612 - learning_rate: 5.0000e-04
 ## Epoch 115/500
-## 90/90 - 0s - 2ms/step - loss: 0.0754 - sparse_categorical_accuracy: 0.9740 - val_loss: 0.1167 - val_sparse_categorical_accuracy: 0.9431 - learning_rate: 5.0000e-04
+## 90/90 - 0s - 2ms/step - loss: 0.0767 - sparse_categorical_accuracy: 0.9733 - val_loss: 0.1115 - val_sparse_categorical_accuracy: 0.9584 - learning_rate: 5.0000e-04
 ## Epoch 116/500
-## 90/90 - 0s - 1ms/step - loss: 0.0786 - sparse_categorical_accuracy: 0.9753 - val_loss: 0.1977 - val_sparse_categorical_accuracy: 0.9279 - learning_rate: 5.0000e-04
+## 90/90 - 0s - 1ms/step - loss: 0.0786 - sparse_categorical_accuracy: 0.9750 - val_loss: 0.1667 - val_sparse_categorical_accuracy: 0.9431 - learning_rate: 5.0000e-04
 ## Epoch 117/500
-## 90/90 - 0s - 1ms/step - loss: 0.0758 - sparse_categorical_accuracy: 0.9736 - val_loss: 0.1724 - val_sparse_categorical_accuracy: 0.9431 - learning_rate: 5.0000e-04
+## 90/90 - 0s - 1ms/step - loss: 0.0760 - sparse_categorical_accuracy: 0.9750 - val_loss: 0.1519 - val_sparse_categorical_accuracy: 0.9376 - learning_rate: 5.0000e-04
 ## Epoch 118/500
-## 90/90 - 0s - 2ms/step - loss: 0.0793 - sparse_categorical_accuracy: 0.9712 - val_loss: 0.1316 - val_sparse_categorical_accuracy: 0.9501 - learning_rate: 5.0000e-04
+## 90/90 - 0s - 1ms/step - loss: 0.0797 - sparse_categorical_accuracy: 0.9722 - val_loss: 0.1555 - val_sparse_categorical_accuracy: 0.9459 - learning_rate: 5.0000e-04
 ## Epoch 119/500
-## 90/90 - 0s - 2ms/step - loss: 0.0781 - sparse_categorical_accuracy: 0.9750 - val_loss: 0.1054 - val_sparse_categorical_accuracy: 0.9598 - learning_rate: 5.0000e-04
+## 90/90 - 0s - 1ms/step - loss: 0.0793 - sparse_categorical_accuracy: 0.9729 - val_loss: 0.1252 - val_sparse_categorical_accuracy: 0.9598 - learning_rate: 5.0000e-04
 ## Epoch 120/500
-## 90/90 - 0s - 2ms/step - loss: 0.0808 - sparse_categorical_accuracy: 0.9740 - val_loss: 0.1425 - val_sparse_categorical_accuracy: 0.9515 - learning_rate: 5.0000e-04
+## 90/90 - 0s - 1ms/step - loss: 0.0803 - sparse_categorical_accuracy: 0.9747 - val_loss: 0.2298 - val_sparse_categorical_accuracy: 0.9015 - learning_rate: 5.0000e-04
 ## Epoch 121/500
-## 90/90 - 0s - 2ms/step - loss: 0.0761 - sparse_categorical_accuracy: 0.9753 - val_loss: 0.1359 - val_sparse_categorical_accuracy: 0.9431 - learning_rate: 5.0000e-04
+## 90/90 - 0s - 1ms/step - loss: 0.0793 - sparse_categorical_accuracy: 0.9736 - val_loss: 0.1140 - val_sparse_categorical_accuracy: 0.9528 - learning_rate: 5.0000e-04
 ## Epoch 122/500
-## 90/90 - 0s - 1ms/step - loss: 0.0813 - sparse_categorical_accuracy: 0.9719 - val_loss: 0.1067 - val_sparse_categorical_accuracy: 0.9556 - learning_rate: 5.0000e-04
+## 90/90 - 0s - 1ms/step - loss: 0.0795 - sparse_categorical_accuracy: 0.9740 - val_loss: 0.1154 - val_sparse_categorical_accuracy: 0.9542 - learning_rate: 5.0000e-04
 ## Epoch 123/500
-## 90/90 - 0s - 2ms/step - loss: 0.0778 - sparse_categorical_accuracy: 0.9767 - val_loss: 0.1328 - val_sparse_categorical_accuracy: 0.9528 - learning_rate: 5.0000e-04
+## 90/90 - 0s - 1ms/step - loss: 0.0768 - sparse_categorical_accuracy: 0.9740 - val_loss: 0.3421 - val_sparse_categorical_accuracy: 0.8724 - learning_rate: 5.0000e-04
 ## Epoch 124/500
-## 90/90 - 0s - 2ms/step - loss: 0.0822 - sparse_categorical_accuracy: 0.9705 - val_loss: 0.1931 - val_sparse_categorical_accuracy: 0.9265 - learning_rate: 5.0000e-04
+## 90/90 - 0s - 1ms/step - loss: 0.0848 - sparse_categorical_accuracy: 0.9694 - val_loss: 0.1126 - val_sparse_categorical_accuracy: 0.9598 - learning_rate: 5.0000e-04
 ## Epoch 125/500
-## 90/90 - 0s - 2ms/step - loss: 0.0797 - sparse_categorical_accuracy: 0.9719 - val_loss: 0.1279 - val_sparse_categorical_accuracy: 0.9459 - learning_rate: 5.0000e-04
+## 90/90 - 0s - 1ms/step - loss: 0.0800 - sparse_categorical_accuracy: 0.9722 - val_loss: 0.1821 - val_sparse_categorical_accuracy: 0.9293 - learning_rate: 5.0000e-04
 ## Epoch 126/500
-## 90/90 - 0s - 1ms/step - loss: 0.0761 - sparse_categorical_accuracy: 0.9743 - val_loss: 0.2298 - val_sparse_categorical_accuracy: 0.9154 - learning_rate: 5.0000e-04
+## 90/90 - 0s - 2ms/step - loss: 0.0787 - sparse_categorical_accuracy: 0.9753 - val_loss: 0.1105 - val_sparse_categorical_accuracy: 0.9570 - learning_rate: 5.0000e-04
 ## Epoch 127/500
-## 90/90 - 0s - 1ms/step - loss: 0.0740 - sparse_categorical_accuracy: 0.9802 - val_loss: 0.2257 - val_sparse_categorical_accuracy: 0.9223 - learning_rate: 5.0000e-04
+## 90/90 - 0s - 2ms/step - loss: 0.0799 - sparse_categorical_accuracy: 0.9760 - val_loss: 0.1091 - val_sparse_categorical_accuracy: 0.9584 - learning_rate: 5.0000e-04
 ## Epoch 128/500
-## 90/90 - 0s - 2ms/step - loss: 0.0772 - sparse_categorical_accuracy: 0.9760 - val_loss: 0.1223 - val_sparse_categorical_accuracy: 0.9515 - learning_rate: 5.0000e-04
+## 90/90 - 0s - 1ms/step - loss: 0.0811 - sparse_categorical_accuracy: 0.9736 - val_loss: 0.1287 - val_sparse_categorical_accuracy: 0.9515 - learning_rate: 5.0000e-04
 ## Epoch 129/500
-## 90/90 - 0s - 1ms/step - loss: 0.0760 - sparse_categorical_accuracy: 0.9764 - val_loss: 0.1443 - val_sparse_categorical_accuracy: 0.9501 - learning_rate: 5.0000e-04
+## 90/90 - 0s - 1ms/step - loss: 0.0797 - sparse_categorical_accuracy: 0.9736 - val_loss: 0.1143 - val_sparse_categorical_accuracy: 0.9584 - learning_rate: 5.0000e-04
 ## Epoch 130/500
-## 90/90 - 0s - 1ms/step - loss: 0.0841 - sparse_categorical_accuracy: 0.9743 - val_loss: 0.1104 - val_sparse_categorical_accuracy: 0.9612 - learning_rate: 5.0000e-04
+## 90/90 - 0s - 1ms/step - loss: 0.0850 - sparse_categorical_accuracy: 0.9750 - val_loss: 0.1581 - val_sparse_categorical_accuracy: 0.9376 - learning_rate: 5.0000e-04
 ## Epoch 131/500
-## 90/90 - 0s - 1ms/step - loss: 0.0697 - sparse_categorical_accuracy: 0.9785 - val_loss: 0.1062 - val_sparse_categorical_accuracy: 0.9584 - learning_rate: 5.0000e-04
+## 90/90 - 0s - 1ms/step - loss: 0.0707 - sparse_categorical_accuracy: 0.9792 - val_loss: 0.1129 - val_sparse_categorical_accuracy: 0.9639 - learning_rate: 5.0000e-04
 ## Epoch 132/500
-## 90/90 - 0s - 1ms/step - loss: 0.0739 - sparse_categorical_accuracy: 0.9743 - val_loss: 0.1153 - val_sparse_categorical_accuracy: 0.9653 - learning_rate: 5.0000e-04
+## 90/90 - 0s - 1ms/step - loss: 0.0725 - sparse_categorical_accuracy: 0.9740 - val_loss: 0.1425 - val_sparse_categorical_accuracy: 0.9501 - learning_rate: 5.0000e-04
 ## Epoch 133/500
-## 90/90 - 0s - 1ms/step - loss: 0.0704 - sparse_categorical_accuracy: 0.9781 - val_loss: 0.1186 - val_sparse_categorical_accuracy: 0.9473 - learning_rate: 5.0000e-04
+## 90/90 - 0s - 2ms/step - loss: 0.0705 - sparse_categorical_accuracy: 0.9764 - val_loss: 0.1098 - val_sparse_categorical_accuracy: 0.9612 - learning_rate: 5.0000e-04
 ## Epoch 134/500
-## 90/90 - 0s - 1ms/step - loss: 0.0753 - sparse_categorical_accuracy: 0.9774 - val_loss: 0.2290 - val_sparse_categorical_accuracy: 0.9196 - learning_rate: 5.0000e-04
+## 90/90 - 0s - 1ms/step - loss: 0.0767 - sparse_categorical_accuracy: 0.9736 - val_loss: 0.1884 - val_sparse_categorical_accuracy: 0.9376 - learning_rate: 5.0000e-04
 ## Epoch 135/500
-## 90/90 - 0s - 1ms/step - loss: 0.0734 - sparse_categorical_accuracy: 0.9743 - val_loss: 0.1498 - val_sparse_categorical_accuracy: 0.9501 - learning_rate: 5.0000e-04
+## 90/90 - 0s - 1ms/step - loss: 0.0775 - sparse_categorical_accuracy: 0.9743 - val_loss: 0.2899 - val_sparse_categorical_accuracy: 0.8988 - learning_rate: 5.0000e-04
 ## Epoch 136/500
-## 90/90 - 0s - 1ms/step - loss: 0.0773 - sparse_categorical_accuracy: 0.9712 - val_loss: 0.1266 - val_sparse_categorical_accuracy: 0.9570 - learning_rate: 5.0000e-04
+## 90/90 - 0s - 1ms/step - loss: 0.0807 - sparse_categorical_accuracy: 0.9701 - val_loss: 0.1152 - val_sparse_categorical_accuracy: 0.9584 - learning_rate: 5.0000e-04
 ## Epoch 137/500
-## 90/90 - 0s - 1ms/step - loss: 0.0807 - sparse_categorical_accuracy: 0.9740 - val_loss: 0.1143 - val_sparse_categorical_accuracy: 0.9639 - learning_rate: 5.0000e-04
+## 90/90 - 0s - 1ms/step - loss: 0.0839 - sparse_categorical_accuracy: 0.9750 - val_loss: 0.3530 - val_sparse_categorical_accuracy: 0.8793 - learning_rate: 5.0000e-04
 ## Epoch 138/500
-## 90/90 - 0s - 1ms/step - loss: 0.0755 - sparse_categorical_accuracy: 0.9753 - val_loss: 0.2602 - val_sparse_categorical_accuracy: 0.9015 - learning_rate: 5.0000e-04
+## 90/90 - 0s - 1ms/step - loss: 0.0770 - sparse_categorical_accuracy: 0.9743 - val_loss: 0.1267 - val_sparse_categorical_accuracy: 0.9473 - learning_rate: 5.0000e-04
 ## Epoch 139/500
-## 90/90 - 0s - 2ms/step - loss: 0.0807 - sparse_categorical_accuracy: 0.9750 - val_loss: 0.0994 - val_sparse_categorical_accuracy: 0.9626 - learning_rate: 5.0000e-04
+## 90/90 - 0s - 2ms/step - loss: 0.0845 - sparse_categorical_accuracy: 0.9719 - val_loss: 0.1043 - val_sparse_categorical_accuracy: 0.9626 - learning_rate: 5.0000e-04
 ## Epoch 140/500
-## 90/90 - 0s - 1ms/step - loss: 0.0697 - sparse_categorical_accuracy: 0.9774 - val_loss: 0.2056 - val_sparse_categorical_accuracy: 0.9182 - learning_rate: 5.0000e-04
+## 90/90 - 0s - 1ms/step - loss: 0.0701 - sparse_categorical_accuracy: 0.9799 - val_loss: 0.1351 - val_sparse_categorical_accuracy: 0.9556 - learning_rate: 5.0000e-04
 ## Epoch 141/500
-## 90/90 - 0s - 2ms/step - loss: 0.0701 - sparse_categorical_accuracy: 0.9753 - val_loss: 0.1829 - val_sparse_categorical_accuracy: 0.9362 - learning_rate: 5.0000e-04
+## 90/90 - 0s - 1ms/step - loss: 0.0712 - sparse_categorical_accuracy: 0.9764 - val_loss: 0.1121 - val_sparse_categorical_accuracy: 0.9612 - learning_rate: 5.0000e-04
 ## Epoch 142/500
-## 90/90 - 0s - 1ms/step - loss: 0.0669 - sparse_categorical_accuracy: 0.9792 - val_loss: 0.1132 - val_sparse_categorical_accuracy: 0.9598 - learning_rate: 5.0000e-04
+## 90/90 - 0s - 1ms/step - loss: 0.0692 - sparse_categorical_accuracy: 0.9778 - val_loss: 0.1139 - val_sparse_categorical_accuracy: 0.9639 - learning_rate: 5.0000e-04
 ## Epoch 143/500
-## 90/90 - 0s - 1ms/step - loss: 0.0751 - sparse_categorical_accuracy: 0.9767 - val_loss: 0.1953 - val_sparse_categorical_accuracy: 0.9223 - learning_rate: 5.0000e-04
+## 90/90 - 0s - 1ms/step - loss: 0.0781 - sparse_categorical_accuracy: 0.9736 - val_loss: 0.1202 - val_sparse_categorical_accuracy: 0.9515 - learning_rate: 5.0000e-04
 ## Epoch 144/500
-## 90/90 - 0s - 1ms/step - loss: 0.0660 - sparse_categorical_accuracy: 0.9792 - val_loss: 0.1440 - val_sparse_categorical_accuracy: 0.9404 - learning_rate: 5.0000e-04
+## 90/90 - 0s - 1ms/step - loss: 0.0699 - sparse_categorical_accuracy: 0.9781 - val_loss: 0.1131 - val_sparse_categorical_accuracy: 0.9501 - learning_rate: 5.0000e-04
 ## Epoch 145/500
-## 90/90 - 0s - 2ms/step - loss: 0.0673 - sparse_categorical_accuracy: 0.9764 - val_loss: 0.6545 - val_sparse_categorical_accuracy: 0.8100 - learning_rate: 5.0000e-04
+## 90/90 - 0s - 1ms/step - loss: 0.0712 - sparse_categorical_accuracy: 0.9771 - val_loss: 0.1097 - val_sparse_categorical_accuracy: 0.9542 - learning_rate: 5.0000e-04
 ## Epoch 146/500
-## 90/90 - 0s - 2ms/step - loss: 0.0708 - sparse_categorical_accuracy: 0.9757 - val_loss: 0.1321 - val_sparse_categorical_accuracy: 0.9542 - learning_rate: 5.0000e-04
+## 90/90 - 0s - 1ms/step - loss: 0.0730 - sparse_categorical_accuracy: 0.9736 - val_loss: 0.1421 - val_sparse_categorical_accuracy: 0.9445 - learning_rate: 5.0000e-04
 ## Epoch 147/500
-## 90/90 - 0s - 2ms/step - loss: 0.0733 - sparse_categorical_accuracy: 0.9764 - val_loss: 0.1479 - val_sparse_categorical_accuracy: 0.9487 - learning_rate: 5.0000e-04
+## 90/90 - 0s - 1ms/step - loss: 0.0757 - sparse_categorical_accuracy: 0.9740 - val_loss: 0.3533 - val_sparse_categorical_accuracy: 0.8752 - learning_rate: 5.0000e-04
 ## Epoch 148/500
-## 90/90 - 0s - 2ms/step - loss: 0.0797 - sparse_categorical_accuracy: 0.9715 - val_loss: 0.3700 - val_sparse_categorical_accuracy: 0.8752 - learning_rate: 5.0000e-04
+## 90/90 - 0s - 1ms/step - loss: 0.0818 - sparse_categorical_accuracy: 0.9701 - val_loss: 0.7787 - val_sparse_categorical_accuracy: 0.7836 - learning_rate: 5.0000e-04
 ## Epoch 149/500
-## 90/90 - 0s - 2ms/step - loss: 0.0709 - sparse_categorical_accuracy: 0.9760 - val_loss: 0.1170 - val_sparse_categorical_accuracy: 0.9584 - learning_rate: 5.0000e-04
+## 90/90 - 0s - 1ms/step - loss: 0.0717 - sparse_categorical_accuracy: 0.9771 - val_loss: 0.1214 - val_sparse_categorical_accuracy: 0.9515 - learning_rate: 5.0000e-04
 ## Epoch 150/500
-## 90/90 - 0s - 2ms/step - loss: 0.0761 - sparse_categorical_accuracy: 0.9726 - val_loss: 0.1173 - val_sparse_categorical_accuracy: 0.9653 - learning_rate: 5.0000e-04
+## 90/90 - 0s - 1ms/step - loss: 0.0736 - sparse_categorical_accuracy: 0.9757 - val_loss: 0.1445 - val_sparse_categorical_accuracy: 0.9362 - learning_rate: 5.0000e-04
 ## Epoch 151/500
-## 90/90 - 0s - 3ms/step - loss: 0.0728 - sparse_categorical_accuracy: 0.9719 - val_loss: 0.1596 - val_sparse_categorical_accuracy: 0.9334 - learning_rate: 5.0000e-04
+## 90/90 - 0s - 1ms/step - loss: 0.0725 - sparse_categorical_accuracy: 0.9729 - val_loss: 0.1175 - val_sparse_categorical_accuracy: 0.9528 - learning_rate: 5.0000e-04
 ## Epoch 152/500
-## 90/90 - 0s - 2ms/step - loss: 0.0727 - sparse_categorical_accuracy: 0.9722 - val_loss: 0.1086 - val_sparse_categorical_accuracy: 0.9528 - learning_rate: 5.0000e-04
+## 90/90 - 0s - 1ms/step - loss: 0.0726 - sparse_categorical_accuracy: 0.9750 - val_loss: 0.1329 - val_sparse_categorical_accuracy: 0.9515 - learning_rate: 5.0000e-04
 ## Epoch 153/500
-## 90/90 - 0s - 2ms/step - loss: 0.0704 - sparse_categorical_accuracy: 0.9771 - val_loss: 0.1305 - val_sparse_categorical_accuracy: 0.9487 - learning_rate: 5.0000e-04
+## 90/90 - 0s - 1ms/step - loss: 0.0727 - sparse_categorical_accuracy: 0.9757 - val_loss: 0.1806 - val_sparse_categorical_accuracy: 0.9293 - learning_rate: 5.0000e-04
 ## Epoch 154/500
-## 90/90 - 0s - 2ms/step - loss: 0.0667 - sparse_categorical_accuracy: 0.9781 - val_loss: 0.1788 - val_sparse_categorical_accuracy: 0.9279 - learning_rate: 5.0000e-04
+## 90/90 - 0s - 1ms/step - loss: 0.0684 - sparse_categorical_accuracy: 0.9778 - val_loss: 0.3892 - val_sparse_categorical_accuracy: 0.8682 - learning_rate: 5.0000e-04
 ## Epoch 155/500
-## 90/90 - 0s - 2ms/step - loss: 0.0702 - sparse_categorical_accuracy: 0.9764 - val_loss: 0.1109 - val_sparse_categorical_accuracy: 0.9501 - learning_rate: 5.0000e-04
+## 90/90 - 0s - 1ms/step - loss: 0.0734 - sparse_categorical_accuracy: 0.9753 - val_loss: 0.1122 - val_sparse_categorical_accuracy: 0.9542 - learning_rate: 5.0000e-04
 ## Epoch 156/500
-## 90/90 - 0s - 1ms/step - loss: 0.0741 - sparse_categorical_accuracy: 0.9764 - val_loss: 0.1325 - val_sparse_categorical_accuracy: 0.9431 - learning_rate: 5.0000e-04
+## 90/90 - 0s - 1ms/step - loss: 0.0758 - sparse_categorical_accuracy: 0.9774 - val_loss: 0.1077 - val_sparse_categorical_accuracy: 0.9653 - learning_rate: 5.0000e-04
 ## Epoch 157/500
-## 90/90 - 0s - 1ms/step - loss: 0.0750 - sparse_categorical_accuracy: 0.9722 - val_loss: 0.1820 - val_sparse_categorical_accuracy: 0.9348 - learning_rate: 5.0000e-04
+## 90/90 - 0s - 1ms/step - loss: 0.0734 - sparse_categorical_accuracy: 0.9747 - val_loss: 0.1594 - val_sparse_categorical_accuracy: 0.9376 - learning_rate: 5.0000e-04
 ## Epoch 158/500
-## 90/90 - 0s - 1ms/step - loss: 0.0714 - sparse_categorical_accuracy: 0.9771 - val_loss: 0.1982 - val_sparse_categorical_accuracy: 0.9182 - learning_rate: 5.0000e-04
+## 90/90 - 0s - 1ms/step - loss: 0.0753 - sparse_categorical_accuracy: 0.9750 - val_loss: 0.1992 - val_sparse_categorical_accuracy: 0.9168 - learning_rate: 5.0000e-04
 ## Epoch 159/500
-## 90/90 - 0s - 2ms/step - loss: 0.0732 - sparse_categorical_accuracy: 0.9750 - val_loss: 0.2841 - val_sparse_categorical_accuracy: 0.8988 - learning_rate: 5.0000e-04
+## 90/90 - 0s - 1ms/step - loss: 0.0740 - sparse_categorical_accuracy: 0.9740 - val_loss: 0.1483 - val_sparse_categorical_accuracy: 0.9362 - learning_rate: 5.0000e-04
 ## Epoch 160/500
-## 90/90 - 0s - 1ms/step - loss: 0.0679 - sparse_categorical_accuracy: 0.9760 - val_loss: 0.1040 - val_sparse_categorical_accuracy: 0.9639 - learning_rate: 2.5000e-04
+## 90/90 - 0s - 2ms/step - loss: 0.0677 - sparse_categorical_accuracy: 0.9778 - val_loss: 0.1041 - val_sparse_categorical_accuracy: 0.9653 - learning_rate: 2.5000e-04
 ## Epoch 161/500
-## 90/90 - 0s - 2ms/step - loss: 0.0649 - sparse_categorical_accuracy: 0.9785 - val_loss: 0.1281 - val_sparse_categorical_accuracy: 0.9459 - learning_rate: 2.5000e-04
+## 90/90 - 0s - 1ms/step - loss: 0.0686 - sparse_categorical_accuracy: 0.9778 - val_loss: 0.1327 - val_sparse_categorical_accuracy: 0.9515 - learning_rate: 2.5000e-04
 ## Epoch 162/500
-## 90/90 - 0s - 2ms/step - loss: 0.0623 - sparse_categorical_accuracy: 0.9781 - val_loss: 0.1535 - val_sparse_categorical_accuracy: 0.9431 - learning_rate: 2.5000e-04
+## 90/90 - 0s - 1ms/step - loss: 0.0634 - sparse_categorical_accuracy: 0.9792 - val_loss: 0.1059 - val_sparse_categorical_accuracy: 0.9667 - learning_rate: 2.5000e-04
 ## Epoch 163/500
-## 90/90 - 0s - 2ms/step - loss: 0.0588 - sparse_categorical_accuracy: 0.9813 - val_loss: 0.1049 - val_sparse_categorical_accuracy: 0.9667 - learning_rate: 2.5000e-04
+## 90/90 - 0s - 1ms/step - loss: 0.0595 - sparse_categorical_accuracy: 0.9792 - val_loss: 0.1236 - val_sparse_categorical_accuracy: 0.9612 - learning_rate: 2.5000e-04
 ## Epoch 164/500
-## 90/90 - 0s - 2ms/step - loss: 0.0615 - sparse_categorical_accuracy: 0.9816 - val_loss: 0.1094 - val_sparse_categorical_accuracy: 0.9515 - learning_rate: 2.5000e-04
+## 90/90 - 0s - 1ms/step - loss: 0.0637 - sparse_categorical_accuracy: 0.9806 - val_loss: 0.1185 - val_sparse_categorical_accuracy: 0.9626 - learning_rate: 2.5000e-04
 ## Epoch 165/500
-## 90/90 - 0s - 2ms/step - loss: 0.0594 - sparse_categorical_accuracy: 0.9809 - val_loss: 0.1116 - val_sparse_categorical_accuracy: 0.9515 - learning_rate: 2.5000e-04
+## 90/90 - 0s - 1ms/step - loss: 0.0607 - sparse_categorical_accuracy: 0.9816 - val_loss: 0.1480 - val_sparse_categorical_accuracy: 0.9473 - learning_rate: 2.5000e-04
 ## Epoch 166/500
-## 90/90 - 0s - 1ms/step - loss: 0.0586 - sparse_categorical_accuracy: 0.9806 - val_loss: 0.1275 - val_sparse_categorical_accuracy: 0.9501 - learning_rate: 2.5000e-04
+## 90/90 - 0s - 1ms/step - loss: 0.0594 - sparse_categorical_accuracy: 0.9799 - val_loss: 0.1044 - val_sparse_categorical_accuracy: 0.9653 - learning_rate: 2.5000e-04
 ## Epoch 167/500
-## 90/90 - 0s - 1ms/step - loss: 0.0605 - sparse_categorical_accuracy: 0.9795 - val_loss: 0.1516 - val_sparse_categorical_accuracy: 0.9417 - learning_rate: 2.5000e-04
+## 90/90 - 0s - 1ms/step - loss: 0.0629 - sparse_categorical_accuracy: 0.9788 - val_loss: 0.1343 - val_sparse_categorical_accuracy: 0.9445 - learning_rate: 2.5000e-04
 ## Epoch 168/500
-## 90/90 - 0s - 1ms/step - loss: 0.0696 - sparse_categorical_accuracy: 0.9785 - val_loss: 0.1038 - val_sparse_categorical_accuracy: 0.9612 - learning_rate: 2.5000e-04
+## 90/90 - 0s - 1ms/step - loss: 0.0740 - sparse_categorical_accuracy: 0.9750 - val_loss: 0.1081 - val_sparse_categorical_accuracy: 0.9667 - learning_rate: 2.5000e-04
 ## Epoch 169/500
-## 90/90 - 0s - 1ms/step - loss: 0.0612 - sparse_categorical_accuracy: 0.9799 - val_loss: 0.1354 - val_sparse_categorical_accuracy: 0.9598 - learning_rate: 2.5000e-04
+## 90/90 - 0s - 1ms/step - loss: 0.0649 - sparse_categorical_accuracy: 0.9799 - val_loss: 0.1136 - val_sparse_categorical_accuracy: 0.9681 - learning_rate: 2.5000e-04
 ## Epoch 170/500
-## 90/90 - 0s - 1ms/step - loss: 0.0626 - sparse_categorical_accuracy: 0.9802 - val_loss: 0.1219 - val_sparse_categorical_accuracy: 0.9598 - learning_rate: 2.5000e-04
+## 90/90 - 0s - 2ms/step - loss: 0.0649 - sparse_categorical_accuracy: 0.9823 - val_loss: 0.1037 - val_sparse_categorical_accuracy: 0.9695 - learning_rate: 2.5000e-04
 ## Epoch 171/500
-## 90/90 - 0s - 1ms/step - loss: 0.0582 - sparse_categorical_accuracy: 0.9816 - val_loss: 0.1627 - val_sparse_categorical_accuracy: 0.9417 - learning_rate: 2.5000e-04
+## 90/90 - 0s - 1ms/step - loss: 0.0609 - sparse_categorical_accuracy: 0.9799 - val_loss: 0.1239 - val_sparse_categorical_accuracy: 0.9653 - learning_rate: 2.5000e-04
 ## Epoch 172/500
-## 90/90 - 0s - 2ms/step - loss: 0.0601 - sparse_categorical_accuracy: 0.9795 - val_loss: 0.1023 - val_sparse_categorical_accuracy: 0.9723 - learning_rate: 2.5000e-04
+## 90/90 - 0s - 1ms/step - loss: 0.0636 - sparse_categorical_accuracy: 0.9795 - val_loss: 0.1037 - val_sparse_categorical_accuracy: 0.9528 - learning_rate: 2.5000e-04
 ## Epoch 173/500
-## 90/90 - 0s - 2ms/step - loss: 0.0587 - sparse_categorical_accuracy: 0.9813 - val_loss: 0.1001 - val_sparse_categorical_accuracy: 0.9639 - learning_rate: 2.5000e-04
+## 90/90 - 0s - 1ms/step - loss: 0.0608 - sparse_categorical_accuracy: 0.9816 - val_loss: 0.1296 - val_sparse_categorical_accuracy: 0.9626 - learning_rate: 2.5000e-04
 ## Epoch 174/500
-## 90/90 - 0s - 2ms/step - loss: 0.0611 - sparse_categorical_accuracy: 0.9806 - val_loss: 0.1214 - val_sparse_categorical_accuracy: 0.9487 - learning_rate: 2.5000e-04
+## 90/90 - 0s - 2ms/step - loss: 0.0624 - sparse_categorical_accuracy: 0.9792 - val_loss: 0.0981 - val_sparse_categorical_accuracy: 0.9639 - learning_rate: 2.5000e-04
 ## Epoch 175/500
-## 90/90 - 0s - 1ms/step - loss: 0.0630 - sparse_categorical_accuracy: 0.9788 - val_loss: 0.1159 - val_sparse_categorical_accuracy: 0.9556 - learning_rate: 2.5000e-04
+## 90/90 - 0s - 1ms/step - loss: 0.0666 - sparse_categorical_accuracy: 0.9785 - val_loss: 0.1135 - val_sparse_categorical_accuracy: 0.9487 - learning_rate: 2.5000e-04
 ## Epoch 176/500
-## 90/90 - 0s - 1ms/step - loss: 0.0593 - sparse_categorical_accuracy: 0.9792 - val_loss: 0.1535 - val_sparse_categorical_accuracy: 0.9473 - learning_rate: 2.5000e-04
+## 90/90 - 0s - 1ms/step - loss: 0.0625 - sparse_categorical_accuracy: 0.9771 - val_loss: 0.1024 - val_sparse_categorical_accuracy: 0.9584 - learning_rate: 2.5000e-04
 ## Epoch 177/500
-## 90/90 - 0s - 2ms/step - loss: 0.0579 - sparse_categorical_accuracy: 0.9844 - val_loss: 0.1511 - val_sparse_categorical_accuracy: 0.9473 - learning_rate: 2.5000e-04
+## 90/90 - 0s - 1ms/step - loss: 0.0616 - sparse_categorical_accuracy: 0.9806 - val_loss: 0.1036 - val_sparse_categorical_accuracy: 0.9570 - learning_rate: 2.5000e-04
 ## Epoch 178/500
-## 90/90 - 0s - 2ms/step - loss: 0.0552 - sparse_categorical_accuracy: 0.9826 - val_loss: 0.2204 - val_sparse_categorical_accuracy: 0.9140 - learning_rate: 2.5000e-04
+## 90/90 - 0s - 1ms/step - loss: 0.0590 - sparse_categorical_accuracy: 0.9823 - val_loss: 0.2086 - val_sparse_categorical_accuracy: 0.9154 - learning_rate: 2.5000e-04
 ## Epoch 179/500
-## 90/90 - 0s - 2ms/step - loss: 0.0572 - sparse_categorical_accuracy: 0.9799 - val_loss: 0.1048 - val_sparse_categorical_accuracy: 0.9681 - learning_rate: 2.5000e-04
+## 90/90 - 0s - 1ms/step - loss: 0.0602 - sparse_categorical_accuracy: 0.9788 - val_loss: 0.1181 - val_sparse_categorical_accuracy: 0.9653 - learning_rate: 2.5000e-04
 ## Epoch 180/500
-## 90/90 - 0s - 2ms/step - loss: 0.0554 - sparse_categorical_accuracy: 0.9823 - val_loss: 0.1197 - val_sparse_categorical_accuracy: 0.9695 - learning_rate: 1.2500e-04
+## 90/90 - 0s - 1ms/step - loss: 0.0615 - sparse_categorical_accuracy: 0.9816 - val_loss: 0.1587 - val_sparse_categorical_accuracy: 0.9459 - learning_rate: 2.5000e-04
 ## Epoch 181/500
-## 90/90 - 0s - 1ms/step - loss: 0.0575 - sparse_categorical_accuracy: 0.9837 - val_loss: 0.1206 - val_sparse_categorical_accuracy: 0.9653 - learning_rate: 1.2500e-04
+## 90/90 - 0s - 1ms/step - loss: 0.0647 - sparse_categorical_accuracy: 0.9813 - val_loss: 0.1836 - val_sparse_categorical_accuracy: 0.9362 - learning_rate: 2.5000e-04
 ## Epoch 182/500
-## 90/90 - 0s - 1ms/step - loss: 0.0537 - sparse_categorical_accuracy: 0.9833 - val_loss: 0.1074 - val_sparse_categorical_accuracy: 0.9667 - learning_rate: 1.2500e-04
+## 90/90 - 0s - 1ms/step - loss: 0.0614 - sparse_categorical_accuracy: 0.9813 - val_loss: 0.2156 - val_sparse_categorical_accuracy: 0.9223 - learning_rate: 2.5000e-04
 ## Epoch 183/500
-## 90/90 - 0s - 2ms/step - loss: 0.0571 - sparse_categorical_accuracy: 0.9809 - val_loss: 0.1052 - val_sparse_categorical_accuracy: 0.9695 - learning_rate: 1.2500e-04
+## 90/90 - 0s - 1ms/step - loss: 0.0637 - sparse_categorical_accuracy: 0.9778 - val_loss: 0.1028 - val_sparse_categorical_accuracy: 0.9584 - learning_rate: 2.5000e-04
 ## Epoch 184/500
-## 90/90 - 0s - 2ms/step - loss: 0.0528 - sparse_categorical_accuracy: 0.9830 - val_loss: 0.1022 - val_sparse_categorical_accuracy: 0.9709 - learning_rate: 1.2500e-04
+## 90/90 - 0s - 1ms/step - loss: 0.0572 - sparse_categorical_accuracy: 0.9823 - val_loss: 0.1347 - val_sparse_categorical_accuracy: 0.9431 - learning_rate: 2.5000e-04
 ## Epoch 185/500
-## 90/90 - 0s - 1ms/step - loss: 0.0550 - sparse_categorical_accuracy: 0.9830 - val_loss: 0.1042 - val_sparse_categorical_accuracy: 0.9653 - learning_rate: 1.2500e-04
+## 90/90 - 0s - 1ms/step - loss: 0.0593 - sparse_categorical_accuracy: 0.9826 - val_loss: 0.1402 - val_sparse_categorical_accuracy: 0.9459 - learning_rate: 2.5000e-04
 ## Epoch 186/500
-## 90/90 - 0s - 2ms/step - loss: 0.0605 - sparse_categorical_accuracy: 0.9792 - val_loss: 0.1048 - val_sparse_categorical_accuracy: 0.9695 - learning_rate: 1.2500e-04
+## 90/90 - 0s - 1ms/step - loss: 0.0659 - sparse_categorical_accuracy: 0.9774 - val_loss: 0.1873 - val_sparse_categorical_accuracy: 0.9293 - learning_rate: 2.5000e-04
 ## Epoch 187/500
-## 90/90 - 0s - 1ms/step - loss: 0.0539 - sparse_categorical_accuracy: 0.9847 - val_loss: 0.1090 - val_sparse_categorical_accuracy: 0.9639 - learning_rate: 1.2500e-04
+## 90/90 - 0s - 1ms/step - loss: 0.0606 - sparse_categorical_accuracy: 0.9799 - val_loss: 0.1210 - val_sparse_categorical_accuracy: 0.9570 - learning_rate: 2.5000e-04
 ## Epoch 188/500
-## 90/90 - 0s - 2ms/step - loss: 0.0540 - sparse_categorical_accuracy: 0.9840 - val_loss: 0.1189 - val_sparse_categorical_accuracy: 0.9653 - learning_rate: 1.2500e-04
+## 90/90 - 0s - 1ms/step - loss: 0.0610 - sparse_categorical_accuracy: 0.9813 - val_loss: 0.1250 - val_sparse_categorical_accuracy: 0.9473 - learning_rate: 2.5000e-04
 ## Epoch 189/500
-## 90/90 - 0s - 2ms/step - loss: 0.0514 - sparse_categorical_accuracy: 0.9854 - val_loss: 0.1599 - val_sparse_categorical_accuracy: 0.9431 - learning_rate: 1.2500e-04
-## Epoch 189: early stopping
+## 90/90 - 0s - 1ms/step - loss: 0.0570 - sparse_categorical_accuracy: 0.9830 - val_loss: 0.1227 - val_sparse_categorical_accuracy: 0.9667 - learning_rate: 2.5000e-04
+## Epoch 190/500
+## 90/90 - 0s - 1ms/step - loss: 0.0678 - sparse_categorical_accuracy: 0.9771 - val_loss: 0.1367 - val_sparse_categorical_accuracy: 0.9584 - learning_rate: 2.5000e-04
+## Epoch 191/500
+## 90/90 - 0s - 1ms/step - loss: 0.0580 - sparse_categorical_accuracy: 0.9802 - val_loss: 0.1005 - val_sparse_categorical_accuracy: 0.9653 - learning_rate: 2.5000e-04
+## Epoch 192/500
+## 90/90 - 0s - 1ms/step - loss: 0.0651 - sparse_categorical_accuracy: 0.9778 - val_loss: 0.1170 - val_sparse_categorical_accuracy: 0.9639 - learning_rate: 2.5000e-04
+## Epoch 193/500
+## 90/90 - 0s - 1ms/step - loss: 0.0593 - sparse_categorical_accuracy: 0.9823 - val_loss: 0.1235 - val_sparse_categorical_accuracy: 0.9473 - learning_rate: 2.5000e-04
+## Epoch 194/500
+## 90/90 - 0s - 1ms/step - loss: 0.0588 - sparse_categorical_accuracy: 0.9792 - val_loss: 0.1089 - val_sparse_categorical_accuracy: 0.9570 - learning_rate: 2.5000e-04
+## Epoch 195/500
+## 90/90 - 0s - 2ms/step - loss: 0.0564 - sparse_categorical_accuracy: 0.9833 - val_loss: 0.0979 - val_sparse_categorical_accuracy: 0.9681 - learning_rate: 1.2500e-04
+## Epoch 196/500
+## 90/90 - 0s - 1ms/step - loss: 0.0584 - sparse_categorical_accuracy: 0.9788 - val_loss: 0.1169 - val_sparse_categorical_accuracy: 0.9584 - learning_rate: 1.2500e-04
+## Epoch 197/500
+## 90/90 - 0s - 1ms/step - loss: 0.0553 - sparse_categorical_accuracy: 0.9802 - val_loss: 0.1380 - val_sparse_categorical_accuracy: 0.9556 - learning_rate: 1.2500e-04
+## Epoch 198/500
+## 90/90 - 0s - 1ms/step - loss: 0.0541 - sparse_categorical_accuracy: 0.9868 - val_loss: 0.1261 - val_sparse_categorical_accuracy: 0.9556 - learning_rate: 1.2500e-04
+## Epoch 199/500
+## 90/90 - 0s - 1ms/step - loss: 0.0575 - sparse_categorical_accuracy: 0.9806 - val_loss: 0.1010 - val_sparse_categorical_accuracy: 0.9681 - learning_rate: 1.2500e-04
+## Epoch 200/500
+## 90/90 - 0s - 1ms/step - loss: 0.0568 - sparse_categorical_accuracy: 0.9816 - val_loss: 0.0996 - val_sparse_categorical_accuracy: 0.9612 - learning_rate: 1.2500e-04
+## Epoch 201/500
+## 90/90 - 0s - 1ms/step - loss: 0.0547 - sparse_categorical_accuracy: 0.9819 - val_loss: 0.1096 - val_sparse_categorical_accuracy: 0.9653 - learning_rate: 1.2500e-04
+## Epoch 202/500
+## 90/90 - 0s - 1ms/step - loss: 0.0547 - sparse_categorical_accuracy: 0.9847 - val_loss: 0.1019 - val_sparse_categorical_accuracy: 0.9556 - learning_rate: 1.2500e-04
+## Epoch 203/500
+## 90/90 - 0s - 1ms/step - loss: 0.0562 - sparse_categorical_accuracy: 0.9816 - val_loss: 0.0985 - val_sparse_categorical_accuracy: 0.9639 - learning_rate: 1.2500e-04
+## Epoch 204/500
+## 90/90 - 0s - 2ms/step - loss: 0.0519 - sparse_categorical_accuracy: 0.9830 - val_loss: 0.0978 - val_sparse_categorical_accuracy: 0.9681 - learning_rate: 1.2500e-04
+## Epoch 205/500
+## 90/90 - 0s - 1ms/step - loss: 0.0558 - sparse_categorical_accuracy: 0.9802 - val_loss: 0.1035 - val_sparse_categorical_accuracy: 0.9667 - learning_rate: 1.2500e-04
+## Epoch 206/500
+## 90/90 - 0s - 1ms/step - loss: 0.0551 - sparse_categorical_accuracy: 0.9816 - val_loss: 0.1006 - val_sparse_categorical_accuracy: 0.9709 - learning_rate: 1.2500e-04
+## Epoch 207/500
+## 90/90 - 0s - 1ms/step - loss: 0.0593 - sparse_categorical_accuracy: 0.9816 - val_loss: 0.1023 - val_sparse_categorical_accuracy: 0.9556 - learning_rate: 1.2500e-04
+## Epoch 208/500
+## 90/90 - 0s - 2ms/step - loss: 0.0556 - sparse_categorical_accuracy: 0.9816 - val_loss: 0.1314 - val_sparse_categorical_accuracy: 0.9626 - learning_rate: 1.2500e-04
+## Epoch 209/500
+## 90/90 - 0s - 1ms/step - loss: 0.0539 - sparse_categorical_accuracy: 0.9809 - val_loss: 0.1529 - val_sparse_categorical_accuracy: 0.9487 - learning_rate: 1.2500e-04
+## Epoch 210/500
+## 90/90 - 0s - 1ms/step - loss: 0.0566 - sparse_categorical_accuracy: 0.9816 - val_loss: 0.0993 - val_sparse_categorical_accuracy: 0.9653 - learning_rate: 1.2500e-04
+## Epoch 211/500
+## 90/90 - 0s - 1ms/step - loss: 0.0552 - sparse_categorical_accuracy: 0.9837 - val_loss: 0.0984 - val_sparse_categorical_accuracy: 0.9639 - learning_rate: 1.2500e-04
+## Epoch 212/500
+## 90/90 - 0s - 1ms/step - loss: 0.0548 - sparse_categorical_accuracy: 0.9826 - val_loss: 0.0999 - val_sparse_categorical_accuracy: 0.9584 - learning_rate: 1.2500e-04
+## Epoch 213/500
+## 90/90 - 0s - 2ms/step - loss: 0.0573 - sparse_categorical_accuracy: 0.9840 - val_loss: 0.0968 - val_sparse_categorical_accuracy: 0.9667 - learning_rate: 1.2500e-04
+## Epoch 214/500
+## 90/90 - 0s - 1ms/step - loss: 0.0568 - sparse_categorical_accuracy: 0.9830 - val_loss: 0.1000 - val_sparse_categorical_accuracy: 0.9639 - learning_rate: 1.2500e-04
+## Epoch 215/500
+## 90/90 - 0s - 1ms/step - loss: 0.0534 - sparse_categorical_accuracy: 0.9816 - val_loss: 0.1031 - val_sparse_categorical_accuracy: 0.9667 - learning_rate: 1.2500e-04
+## Epoch 216/500
+## 90/90 - 0s - 1ms/step - loss: 0.0531 - sparse_categorical_accuracy: 0.9847 - val_loss: 0.1042 - val_sparse_categorical_accuracy: 0.9681 - learning_rate: 1.2500e-04
+## Epoch 217/500
+## 90/90 - 0s - 1ms/step - loss: 0.0532 - sparse_categorical_accuracy: 0.9826 - val_loss: 0.1096 - val_sparse_categorical_accuracy: 0.9695 - learning_rate: 1.2500e-04
+## Epoch 218/500
+## 90/90 - 0s - 1ms/step - loss: 0.0577 - sparse_categorical_accuracy: 0.9809 - val_loss: 0.1009 - val_sparse_categorical_accuracy: 0.9612 - learning_rate: 1.2500e-04
+## Epoch 219/500
+## 90/90 - 0s - 1ms/step - loss: 0.0539 - sparse_categorical_accuracy: 0.9823 - val_loss: 0.1072 - val_sparse_categorical_accuracy: 0.9695 - learning_rate: 1.2500e-04
+## Epoch 220/500
+## 90/90 - 0s - 1ms/step - loss: 0.0529 - sparse_categorical_accuracy: 0.9851 - val_loss: 0.1005 - val_sparse_categorical_accuracy: 0.9695 - learning_rate: 1.2500e-04
+## Epoch 221/500
+## 90/90 - 0s - 1ms/step - loss: 0.0554 - sparse_categorical_accuracy: 0.9799 - val_loss: 0.1036 - val_sparse_categorical_accuracy: 0.9695 - learning_rate: 1.2500e-04
+## Epoch 222/500
+## 90/90 - 0s - 1ms/step - loss: 0.0562 - sparse_categorical_accuracy: 0.9809 - val_loss: 0.1037 - val_sparse_categorical_accuracy: 0.9736 - learning_rate: 1.2500e-04
+## Epoch 223/500
+## 90/90 - 0s - 1ms/step - loss: 0.0544 - sparse_categorical_accuracy: 0.9823 - val_loss: 0.1005 - val_sparse_categorical_accuracy: 0.9736 - learning_rate: 1.2500e-04
+## Epoch 224/500
+## 90/90 - 0s - 1ms/step - loss: 0.0509 - sparse_categorical_accuracy: 0.9847 - val_loss: 0.0969 - val_sparse_categorical_accuracy: 0.9626 - learning_rate: 1.2500e-04
+## Epoch 225/500
+## 90/90 - 0s - 1ms/step - loss: 0.0569 - sparse_categorical_accuracy: 0.9823 - val_loss: 0.1261 - val_sparse_categorical_accuracy: 0.9612 - learning_rate: 1.2500e-04
+## Epoch 226/500
+## 90/90 - 0s - 1ms/step - loss: 0.0519 - sparse_categorical_accuracy: 0.9826 - val_loss: 0.1019 - val_sparse_categorical_accuracy: 0.9570 - learning_rate: 1.2500e-04
+## Epoch 227/500
+## 90/90 - 0s - 1ms/step - loss: 0.0563 - sparse_categorical_accuracy: 0.9819 - val_loss: 0.1037 - val_sparse_categorical_accuracy: 0.9584 - learning_rate: 1.2500e-04
+## Epoch 228/500
+## 90/90 - 0s - 1ms/step - loss: 0.0518 - sparse_categorical_accuracy: 0.9844 - val_loss: 0.1127 - val_sparse_categorical_accuracy: 0.9515 - learning_rate: 1.2500e-04
+## Epoch 229/500
+## 90/90 - 0s - 1ms/step - loss: 0.0521 - sparse_categorical_accuracy: 0.9865 - val_loss: 0.0973 - val_sparse_categorical_accuracy: 0.9639 - learning_rate: 1.2500e-04
+## Epoch 230/500
+## 90/90 - 0s - 1ms/step - loss: 0.0520 - sparse_categorical_accuracy: 0.9847 - val_loss: 0.0999 - val_sparse_categorical_accuracy: 0.9626 - learning_rate: 1.2500e-04
+## Epoch 231/500
+## 90/90 - 0s - 2ms/step - loss: 0.0569 - sparse_categorical_accuracy: 0.9813 - val_loss: 0.0958 - val_sparse_categorical_accuracy: 0.9626 - learning_rate: 1.2500e-04
+## Epoch 232/500
+## 90/90 - 0s - 1ms/step - loss: 0.0530 - sparse_categorical_accuracy: 0.9851 - val_loss: 0.1024 - val_sparse_categorical_accuracy: 0.9681 - learning_rate: 1.2500e-04
+## Epoch 233/500
+## 90/90 - 0s - 1ms/step - loss: 0.0523 - sparse_categorical_accuracy: 0.9844 - val_loss: 0.1019 - val_sparse_categorical_accuracy: 0.9709 - learning_rate: 1.2500e-04
+## Epoch 234/500
+## 90/90 - 0s - 1ms/step - loss: 0.0539 - sparse_categorical_accuracy: 0.9837 - val_loss: 0.1595 - val_sparse_categorical_accuracy: 0.9459 - learning_rate: 1.2500e-04
+## Epoch 235/500
+## 90/90 - 0s - 1ms/step - loss: 0.0543 - sparse_categorical_accuracy: 0.9802 - val_loss: 0.1028 - val_sparse_categorical_accuracy: 0.9653 - learning_rate: 1.2500e-04
+## Epoch 236/500
+## 90/90 - 0s - 1ms/step - loss: 0.0540 - sparse_categorical_accuracy: 0.9826 - val_loss: 0.1068 - val_sparse_categorical_accuracy: 0.9653 - learning_rate: 1.2500e-04
+## Epoch 237/500
+## 90/90 - 0s - 1ms/step - loss: 0.0543 - sparse_categorical_accuracy: 0.9809 - val_loss: 0.1528 - val_sparse_categorical_accuracy: 0.9459 - learning_rate: 1.2500e-04
+## Epoch 238/500
+## 90/90 - 0s - 1ms/step - loss: 0.0527 - sparse_categorical_accuracy: 0.9861 - val_loss: 0.0977 - val_sparse_categorical_accuracy: 0.9695 - learning_rate: 1.2500e-04
+## Epoch 239/500
+## 90/90 - 0s - 1ms/step - loss: 0.0537 - sparse_categorical_accuracy: 0.9813 - val_loss: 0.0980 - val_sparse_categorical_accuracy: 0.9598 - learning_rate: 1.2500e-04
+## Epoch 240/500
+## 90/90 - 0s - 1ms/step - loss: 0.0512 - sparse_categorical_accuracy: 0.9854 - val_loss: 0.1054 - val_sparse_categorical_accuracy: 0.9709 - learning_rate: 1.2500e-04
+## Epoch 241/500
+## 90/90 - 0s - 1ms/step - loss: 0.0541 - sparse_categorical_accuracy: 0.9837 - val_loss: 0.1325 - val_sparse_categorical_accuracy: 0.9487 - learning_rate: 1.2500e-04
+## Epoch 242/500
+## 90/90 - 0s - 1ms/step - loss: 0.0535 - sparse_categorical_accuracy: 0.9826 - val_loss: 0.1028 - val_sparse_categorical_accuracy: 0.9681 - learning_rate: 1.2500e-04
+## Epoch 243/500
+## 90/90 - 0s - 1ms/step - loss: 0.0555 - sparse_categorical_accuracy: 0.9823 - val_loss: 0.0991 - val_sparse_categorical_accuracy: 0.9626 - learning_rate: 1.2500e-04
+## Epoch 244/500
+## 90/90 - 0s - 1ms/step - loss: 0.0572 - sparse_categorical_accuracy: 0.9795 - val_loss: 0.0986 - val_sparse_categorical_accuracy: 0.9639 - learning_rate: 1.2500e-04
+## Epoch 245/500
+## 90/90 - 0s - 1ms/step - loss: 0.0520 - sparse_categorical_accuracy: 0.9833 - val_loss: 0.0980 - val_sparse_categorical_accuracy: 0.9626 - learning_rate: 1.2500e-04
+## Epoch 246/500
+## 90/90 - 0s - 1ms/step - loss: 0.0554 - sparse_categorical_accuracy: 0.9809 - val_loss: 0.0978 - val_sparse_categorical_accuracy: 0.9639 - learning_rate: 1.2500e-04
+## Epoch 247/500
+## 90/90 - 0s - 1ms/step - loss: 0.0498 - sparse_categorical_accuracy: 0.9837 - val_loss: 0.1015 - val_sparse_categorical_accuracy: 0.9681 - learning_rate: 1.2500e-04
+## Epoch 248/500
+## 90/90 - 0s - 1ms/step - loss: 0.0493 - sparse_categorical_accuracy: 0.9840 - val_loss: 0.0994 - val_sparse_categorical_accuracy: 0.9612 - learning_rate: 1.2500e-04
+## Epoch 249/500
+## 90/90 - 0s - 1ms/step - loss: 0.0543 - sparse_categorical_accuracy: 0.9847 - val_loss: 0.1054 - val_sparse_categorical_accuracy: 0.9695 - learning_rate: 1.2500e-04
+## Epoch 250/500
+## 90/90 - 0s - 1ms/step - loss: 0.0550 - sparse_categorical_accuracy: 0.9802 - val_loss: 0.0989 - val_sparse_categorical_accuracy: 0.9626 - learning_rate: 1.2500e-04
+## Epoch 251/500
+## 90/90 - 0s - 1ms/step - loss: 0.0539 - sparse_categorical_accuracy: 0.9819 - val_loss: 0.1018 - val_sparse_categorical_accuracy: 0.9667 - learning_rate: 1.2500e-04
+## Epoch 252/500
+## 90/90 - 0s - 1ms/step - loss: 0.0542 - sparse_categorical_accuracy: 0.9830 - val_loss: 0.0975 - val_sparse_categorical_accuracy: 0.9667 - learning_rate: 1.0000e-04
+## Epoch 253/500
+## 90/90 - 0s - 1ms/step - loss: 0.0538 - sparse_categorical_accuracy: 0.9830 - val_loss: 0.1065 - val_sparse_categorical_accuracy: 0.9723 - learning_rate: 1.0000e-04
+## Epoch 254/500
+## 90/90 - 0s - 1ms/step - loss: 0.0542 - sparse_categorical_accuracy: 0.9823 - val_loss: 0.1145 - val_sparse_categorical_accuracy: 0.9528 - learning_rate: 1.0000e-04
+## Epoch 255/500
+## 90/90 - 0s - 1ms/step - loss: 0.0506 - sparse_categorical_accuracy: 0.9819 - val_loss: 0.0976 - val_sparse_categorical_accuracy: 0.9681 - learning_rate: 1.0000e-04
+## Epoch 256/500
+## 90/90 - 0s - 1ms/step - loss: 0.0521 - sparse_categorical_accuracy: 0.9847 - val_loss: 0.1117 - val_sparse_categorical_accuracy: 0.9626 - learning_rate: 1.0000e-04
+## Epoch 257/500
+## 90/90 - 0s - 1ms/step - loss: 0.0504 - sparse_categorical_accuracy: 0.9854 - val_loss: 0.1201 - val_sparse_categorical_accuracy: 0.9626 - learning_rate: 1.0000e-04
+## Epoch 258/500
+## 90/90 - 0s - 1ms/step - loss: 0.0498 - sparse_categorical_accuracy: 0.9851 - val_loss: 0.1039 - val_sparse_categorical_accuracy: 0.9584 - learning_rate: 1.0000e-04
+## Epoch 259/500
+## 90/90 - 0s - 1ms/step - loss: 0.0479 - sparse_categorical_accuracy: 0.9868 - val_loss: 0.1055 - val_sparse_categorical_accuracy: 0.9695 - learning_rate: 1.0000e-04
+## Epoch 260/500
+## 90/90 - 0s - 1ms/step - loss: 0.0499 - sparse_categorical_accuracy: 0.9840 - val_loss: 0.1046 - val_sparse_categorical_accuracy: 0.9556 - learning_rate: 1.0000e-04
+## Epoch 261/500
+## 90/90 - 0s - 1ms/step - loss: 0.0529 - sparse_categorical_accuracy: 0.9823 - val_loss: 0.1022 - val_sparse_categorical_accuracy: 0.9667 - learning_rate: 1.0000e-04
+## Epoch 262/500
+## 90/90 - 0s - 1ms/step - loss: 0.0492 - sparse_categorical_accuracy: 0.9840 - val_loss: 0.1063 - val_sparse_categorical_accuracy: 0.9681 - learning_rate: 1.0000e-04
+## Epoch 263/500
+## 90/90 - 0s - 1ms/step - loss: 0.0500 - sparse_categorical_accuracy: 0.9833 - val_loss: 0.1006 - val_sparse_categorical_accuracy: 0.9626 - learning_rate: 1.0000e-04
+## Epoch 264/500
+## 90/90 - 0s - 1ms/step - loss: 0.0533 - sparse_categorical_accuracy: 0.9823 - val_loss: 0.1088 - val_sparse_categorical_accuracy: 0.9709 - learning_rate: 1.0000e-04
+## Epoch 265/500
+## 90/90 - 0s - 1ms/step - loss: 0.0549 - sparse_categorical_accuracy: 0.9802 - val_loss: 0.1010 - val_sparse_categorical_accuracy: 0.9709 - learning_rate: 1.0000e-04
+## Epoch 266/500
+## 90/90 - 0s - 1ms/step - loss: 0.0512 - sparse_categorical_accuracy: 0.9830 - val_loss: 0.1016 - val_sparse_categorical_accuracy: 0.9570 - learning_rate: 1.0000e-04
+## Epoch 267/500
+## 90/90 - 0s - 1ms/step - loss: 0.0496 - sparse_categorical_accuracy: 0.9847 - val_loss: 0.1012 - val_sparse_categorical_accuracy: 0.9570 - learning_rate: 1.0000e-04
+## Epoch 268/500
+## 90/90 - 0s - 1ms/step - loss: 0.0477 - sparse_categorical_accuracy: 0.9872 - val_loss: 0.1016 - val_sparse_categorical_accuracy: 0.9570 - learning_rate: 1.0000e-04
+## Epoch 269/500
+## 90/90 - 0s - 1ms/step - loss: 0.0523 - sparse_categorical_accuracy: 0.9861 - val_loss: 0.1012 - val_sparse_categorical_accuracy: 0.9723 - learning_rate: 1.0000e-04
+## Epoch 270/500
+## 90/90 - 0s - 1ms/step - loss: 0.0474 - sparse_categorical_accuracy: 0.9889 - val_loss: 0.0975 - val_sparse_categorical_accuracy: 0.9653 - learning_rate: 1.0000e-04
+## Epoch 271/500
+## 90/90 - 0s - 1ms/step - loss: 0.0489 - sparse_categorical_accuracy: 0.9840 - val_loss: 0.1055 - val_sparse_categorical_accuracy: 0.9695 - learning_rate: 1.0000e-04
+## Epoch 272/500
+## 90/90 - 0s - 1ms/step - loss: 0.0503 - sparse_categorical_accuracy: 0.9826 - val_loss: 0.1050 - val_sparse_categorical_accuracy: 0.9709 - learning_rate: 1.0000e-04
+## Epoch 273/500
+## 90/90 - 0s - 1ms/step - loss: 0.0514 - sparse_categorical_accuracy: 0.9844 - val_loss: 0.0980 - val_sparse_categorical_accuracy: 0.9695 - learning_rate: 1.0000e-04
+## Epoch 274/500
+## 90/90 - 0s - 1ms/step - loss: 0.0479 - sparse_categorical_accuracy: 0.9854 - val_loss: 0.1049 - val_sparse_categorical_accuracy: 0.9639 - learning_rate: 1.0000e-04
+## Epoch 275/500
+## 90/90 - 0s - 1ms/step - loss: 0.0467 - sparse_categorical_accuracy: 0.9840 - val_loss: 0.1473 - val_sparse_categorical_accuracy: 0.9556 - learning_rate: 1.0000e-04
+## Epoch 276/500
+## 90/90 - 0s - 1ms/step - loss: 0.0504 - sparse_categorical_accuracy: 0.9819 - val_loss: 0.1065 - val_sparse_categorical_accuracy: 0.9695 - learning_rate: 1.0000e-04
+## Epoch 277/500
+## 90/90 - 0s - 1ms/step - loss: 0.0505 - sparse_categorical_accuracy: 0.9847 - val_loss: 0.1187 - val_sparse_categorical_accuracy: 0.9612 - learning_rate: 1.0000e-04
+## Epoch 278/500
+## 90/90 - 0s - 1ms/step - loss: 0.0498 - sparse_categorical_accuracy: 0.9840 - val_loss: 0.1267 - val_sparse_categorical_accuracy: 0.9598 - learning_rate: 1.0000e-04
+## Epoch 279/500
+## 90/90 - 0s - 1ms/step - loss: 0.0517 - sparse_categorical_accuracy: 0.9830 - val_loss: 0.1037 - val_sparse_categorical_accuracy: 0.9653 - learning_rate: 1.0000e-04
+## Epoch 280/500
+## 90/90 - 0s - 1ms/step - loss: 0.0501 - sparse_categorical_accuracy: 0.9830 - val_loss: 0.0985 - val_sparse_categorical_accuracy: 0.9667 - learning_rate: 1.0000e-04
+## Epoch 281/500
+## 90/90 - 0s - 1ms/step - loss: 0.0468 - sparse_categorical_accuracy: 0.9865 - val_loss: 0.1119 - val_sparse_categorical_accuracy: 0.9653 - learning_rate: 1.0000e-04
+## Epoch 281: early stopping
 ```
 
 
@@ -727,7 +867,7 @@ results <- model |> evaluate(x_test, y_test)
 ```
 
 ```
-## 42/42 - 1s - 16ms/step - loss: 0.0998 - sparse_categorical_accuracy: 0.9652
+## 42/42 - 1s - 16ms/step - loss: 0.0938 - sparse_categorical_accuracy: 0.9697
 ```
 
 ``` r
@@ -736,8 +876,8 @@ str(results)
 
 ```
 ## List of 2
-##  $ loss                       : num 0.0998
-##  $ sparse_categorical_accuracy: num 0.965
+##  $ loss                       : num 0.0938
+##  $ sparse_categorical_accuracy: num 0.97
 ```
 
 ``` r
@@ -749,8 +889,8 @@ cat(
 ```
 
 ```
-## Test accuracy: 0.9651515
-## Test loss: 0.0998451
+## Test accuracy: 0.969697
+## Test loss: 0.09383929
 ```
 
 
diff --git a/vignettes/examples/timeseries/timeseries_classification_from_scratch/unnamed-chunk-12-1.png b/vignettes/examples/timeseries/timeseries_classification_from_scratch/unnamed-chunk-12-1.png
index 3ff438a1e2dfe04320c909cdbb15fdf10b4e91b7..31fccc9f45a861eac243a107c0344926a8c4deb9 100644
GIT binary patch
literal 65538
zcmag_1z40{_%#X-A|S1_Ff>w9BArTicb9;4cS$4CDJdl&Al*5HfHX*VcXxlA-~WBD
zbFTA#-*>ziMrX!n_OtiB?|ZFvuN|f!CytInhysB?&?O~Alpv62!w?AE)r)7~l|2es
zbnt|1C!y&GfuLeNJ>W*?eeNI-GKi$eM-{iUy?HYY95Z~xB{NeO*P4+=KME8VhACX_
zy=LkQEOzShOG5UGDQ$+ZcQsSSQO%clp5h`naNB(s5BF_D1fs@vUq;ey|3>v3w4An~
z?M>IB8`sW$d_f-gGB$3i3F8C!TeLhWE;R6^(2HJs6+7@mPSvdld-_N;l;-a)1mnlJ
z1@UO421AVoLwlNNpMBc6#lS9(rh#VSC(ZD#*vXF-Z574yHuxLJdEtyiT#?2EV#)YK
z-S!-*vdJ_m6fwyIriP~2CtZkbiqjg23B4v5pLcRD=y0GO%?srF5_w9^dbS*?#G@(3
zbT}yD-l>sYr+h=F{sX4v8iCiFGSqROtF^!d_PVI!dPOK=3@Tqoh@}oBg6}0u>KnJH
zW~oHJFpBG2N9*9C2VbjV=NGK<-wmP`ERlixl8DUl4|uvS?nso#$FKkX{i9R<d3Lz?
zI8|;;x-jzv9%-i9a&x?(cQj90u!$XfFtbVCaFo`a6dn<cYd9Cd!mGx<H~W5YaCv!o
z+D0m8#7VF0#h8zu1e?Rc9wEC04wKH*7e!?9KrfGikX1pL0Sk1PNRb}x;J|!*qcaSf
zbW}@62kIX{js^FjunY;4gp2yk)3oE`LLZTp3UMshowOizK8$Q^XTn-<HOdS`bQnm)
zC<I7CFx5)+=HJmhUAn<sV!d@`33JSBAjO%M-&moVUN(j8I{Ig)(Q$haY7(RN&&gfS
zp$y!YFQa2(mJ8CroeM~+*xjkeGitls-<(HUFp`lO{EerBxMjGiq&~fE$d(K47Ob;4
z0uqP_AqVLHuRyq@D^^mZ#liR-@=3E$g+4|C3YN<rTI;+!5suHSR;*Djoy5fOHeM=b
zl`C=;>Wl&nU*#&-u&SLs<&nT62#)DBY44}T3lDM^5~X<6SK&YK$3Bfkp0<O<p<X9>
z`nhL$SeFsk!9bA74Gs!MiP~<o3l&xvn90-ujL!q*iOl)}2cMyYoR8;6Syxxz?{6=M
zc14(vK%6w?$bB(B=Cdu=C|0x7L+X7tbn4NpmpqMw62u~3AVZBECIUi<jnJAy8@%hU
zLO;Ik1covdF3Xhk9&NS23fkRk!WzPMu892?>C=f#zI>+GZRF8eJ3jGXI&adX{nG^b
zhM+`71(cRDlZe87IHM#b&1{Mk25;Q)X0}KF^@V{&R=&AhOphP!?=#-(|L#s40s`*Z
zwP5Lr%gxiXRHvS`RQV|cr$vP6!9@Rm*=ql<OYnbufV5XGXQC?EbLgLoUyKJn{Tkz^
z6P}k(>-d0RcwslSG6-!ob6a0L7WdV3+&W2ktNwe=hq_byH8vQ?nr`4^hoPH?;o)z_
zMkS)4afX}63^Z-2iYlqw-;DQxnxx(gOu7whjvM`}9>%rVS?^8zUxo#`UOdHOcHm`V
z2i<8xG1O(d?X7;rZfu62K(ob$bxv2aTGT@IohZa^t>;yORM)B3KngptgGsO<6&7;0
z{nJ{4Ev!O+-zG*Fb&H|w>!;G-G4xw%w$^ofw2K_3k$8V(#r=^A%N}Z0Ieqe9>@2&3
z<%BwvuLt&H(@$AJ_FO!{N(yasQTJO3wM-EN+pBN{XMKkFy+p+**EWzKC+%94!^EFE
zDuOdyzMd{{UiIb07HIm34X;yYN6x{aR;V&js=r7sZb~4=fQNeODA;Y!H{>gUs#b)n
zM91pu`xVEVkofuwx~Y*S$LY{?2<N&TLm$d1G4W%mP0%&T(K5a^#;dEYKNao!UlJs!
z+=<C8{hOy)1>Je*Q(1h+95Y`o>MT-SFsUgqee!!2<o!P5)6{Y&#Q4v*UN1Io?rCyy
zdq>AHi)z!#Y4Q^8M4_x<3qhv~skfwu`)2Yv1}q$Rc<;Olr1EZ#3_yVMWyZuDFUxX6
zhfeXryDX|VoTC})8XJqoEeuT^d%6;_O<2E6P$?CSyAxBPvVQM3XG5uMJ3l&7C>sBN
z#5r=!MxSguW3N6-VMhKoiES#+qek|4qR;+NyuWDEb5gAz8w07F3kAE*XG_0f@NZQL
zqW=s3yO0-^A;N2VWP&tC5RFwiTdvGj+f=>vTB|hP?hV|BPa#;3BRnHEwYG~b_aKSp
zNhdwroeouw_!cS9-drALVoyJ9B3En^7>xF<({}uItByk4$Bz(&+FQQ{cLGU>&;9vi
zOzpuBMddWE#ON}+;U6E}(?h{Nao1<#IUeEiIHPU249p~*ac#0&7SikfpCG5_5+J-j
zSr`5%3~|HT-PlO{Uz_z~v@ku@TT1ar8A1*#c_jx**V^*2{7q-*Q{3zS@L$aViTrrC
zv~|N71-wMXYc3J=v^hb2zyMEyYh3hP;OS-T>i=0l{$D<Dpgi?ih00;d6=t6{Z{w<B
ziP6dO4Zc+tX@yI!XBxtS76-}}HC9;QciA$U_)nVTW*|Rtq%xKUn2?`_P_J3Jr&FLt
zHcx3J3e3W1ios&8TTYe<f(<n%6hWj!2fRb0oU&?XAU%EzXDn4ViD0U+GSK9BuLm|n
zia{mvwNZfZq7HO)u=_xnmg}2Xj_^!O-#UmhD-PaV_HL{&k;oek;>b^$Fw;s62DKt~
zaE*^y3du6m{}tH8>2R_sE9yAO<mMMrc}is<FAKfDwINQBO%_!>z}_-rc$ywBssBII
zB5~2x)g>w>hT`((X$qIGaMTLDJkC74Jv<sRGxhb5+>^e)Hydp5AZB7>WcUd}<!jeU
zlbpy4a*?`%5Meu)y{VX1*~T(OV;<+7+_CKw9=VCpu`x5<y1??)5!<D~XD}oTXwR+)
zNOR94P%vQI%_gr+``eEck&4DOT7A6rx_eVb%mamvS4dIvU<N&7T;P*|U}X)<wMwE-
zdgcN~JCyU3)}6D9udlAW4szRUEz8XCFCG<JeeVBfAj1VqFYzK*JrZam7bjNsUKiih
zs}`xbr*D<uG3zzueQX{#e4`Wvb#ZYS+<m(G^9bq)l)<XcV>?g*C=rpi?>h6t!-G5V
zW(h%Xp=uEx*uz6+CMIkD1)%@M0z$|EEMvSjz>^c#mW0oIeT+hXFm0kdl#M3t-s~Nk
za;aW(liU*k;~MqEDAZt*eel}j&e3S}o$2h^379F^_%;=Nl)2!&yf<ArTWc#XE#0^K
z1ax}Dm+nz88h0j$^D&HP5$1V(@~on4Q+-AnWaiSv6QP3_m(MZf59htFJpAeKF>lmz
z#G`DEZ!bzox&mu%`N+x1OIM%vfzYK70@5l}NyE|Ye&fyQCE`R8tZQ%4QFHq+$?<ZQ
zeeTM7pkCeOM;ZTYy@BpGZGqDm_rAfR(>k**bGF}guB7}PSsy8#b}Ed!Bbx@E>g0<>
za3i5b5qeW+*EPxu9Y38N1gp(ke$7@LeB`cZ@A}=+vL5oz(S?6PQy&kn<_dLEhp?s)
zYvG4%{U-ilisMS%Vbm%Y!8hiItnt4l6?_p-7xj9+;5tgX`ud%@W+xfs!_=N}gagCN
zRh5SM(Bnbnq?QwQPm8bfl~n`={7vz{*wTry5T61>O6sc&_3ppIIAM%|ieeRdkij3X
z`rXrc&@xys1X|8j3+H`qWse<Di=mWw;f~9+({1+9ZhQZM|KLjzey|}(xc^~JuysLQ
z-~gDjoi7Wmqi5aWe|C|3DWak2t(!MPk;=o3YWb{Hz3Wu%)>yv$?nH^Ql9G_Hux8=~
z>wh^6=7|SQl#UZ5y|wx`5qX<0`>j}@cB})LlH=d>NJmodv=cMS0g08=SA?MCe>myd
z(X6Kh#Jn#3UR+OgBqT<H3NZj?<x@&dRIkI7)OYUDER1m<1+b4=(LlQK{oC)ib}u+m
zyNp2WKa~f*q1|(j{{u@y1TdN<=@WLGc>d?Dq`ucrH8pb8{D`<+KhH?M(RIfoTRzWW
zhtl)+B7gm++K1}cN4~X>CxKl?(;)88w#S%t8{#K5pO!kDOudk0w4I&O-yIRc(12F*
z{n&P*G<~QtB++8=f>!)fwu1NNyz4y$=>Vfv#c?q=KwST8ABor%sX!p3La)YAr67LC
z^pm+3WJrJV?9TrT>?B_FP9H?1^SWMajToP8jSOW94voNj$EwXWMI@HI!und;ve4}5
z`d@9JAbT`Q_}f+=w4>AVu(c$)Xj@atxR&DBXt$OSM<eetTZI6RxV$}<3=2T;w7l|N
zpRP2MPGlG{X9L&&t&k-YNm3Lk<f2j<A`_fUASl=h<;MdBEUNDo5Bif=dY0omRJTs+
zGqdB_@WKSg+l&5!&F{R9Vvc=b*fhgc+-`IzAyC%jZ(@`~r|!f%0U9L}5BIm3I}_N0
zgS#~H^;LEs6%-T(iETPVUx|xH`3o)l8_&C3>Ed=J1G|qj{gQ#p3CWc4rMK_4hq^DG
z-@(aLN`RW!#Oqf#lFp8SK|zL-q_k>KdR19GQqO9*C4fHjo4gbXSRicXP6u=KP)4ry
zj*d8t1J3U<hm}bHu|+?|U%utqYskg<)XS2IU&*o7^*6|3=*{fWXIv<wR@K$POm(F)
zwM5=%OdQRoa~dYQ)l}K|XZlWoplP&ng2~b#ukp9|k<#Oc;98u``Y>3X=s)`u4kL%@
z!t_25f}E(qRFAyTGo=;iU?Ng51~HF8n_rXMG9-$GUn@Q13#`NL8mndRnSN_wiPjUC
zFLitc@NtyK5IHVW?&Htu(Zb4L*W`lOE4)(nmCJJUGzX4p--uq5nHuX<sXU8?rt6TQ
zI=0k-p1ses`fvkUHigQSY9(5vB(YCmI>VQNtE$%JH~*{DdQzh;vwa~!ct|dE(F1@H
z^hbF+^+S+}tfwow`ud6hMp;-`sCStI(Q?SBikOkH+Th$;Z(3bidKglLM@Tp`YBg7G
zitE;L&~dBU9pR!P2Ns2j=%H|_Y}To82pel*N+sTz&H!tr#7|&&yhvS8{{kBkP0VZV
z9uA(vW-ghxTUJVH#?oxK>>Xqa9nihA(^JorHOct~XH*>8=m^Bu!I~{z?w5{$+bW+Q
z8^5;|0p%}nEy0(biy;3iKebAK0<HT27$3mgA1d+e?7xKfy<&O&`t{)8V7;9=4Gm51
z-D4yR@-b2l3G_uzR6F_A&1{^b8U9-}hSOIiYP_cX$>)kuTI;8&eC`!dQA4~lCYm)`
z)fU|8X-}AnG;=9g;$p7eF@G(IA9Ia4w8;0FpIpd46`FjVlmdzbsZG~Czh&X0`zwsk
zy-1_)hwF*x)w?DaZDwos*T(=xtZlJ!?@w-S7EL{DeH3u5@OH8R$nXAg@$qUk3XIb|
z9pI@R)I)F28~OR)x{br9Zv*jQGvfr@RT<%@!`ov7)R#3JAO|Zqy6mGN22@s_1|y;y
zR{O6;$8W8yoSbmS#Kf4-U_nrx=jG+CYt=+asHmuv#N-^0PfB?Cw5XTL8#iiz0D7)~
z*^Np<LJ}GU?}w3YP+qLxD!|9bm-JqODz^Vs8hZeu(V9Ykn%-p(%x1o!*-J=7<T;0}
zzg`q&2*|>Y+zaB%Km;(K$Abka)bIEZDJdzP@Vo+<+0TPZZ)Y^Guj}F7HH|0i-@o&Y
zkPYi(u?Rx`+r3I4AxuvE^15D6%4WA5d}zv5DOB?Q6X|?qKQ=a|tE)@AU}b~y;ze6w
zWq}5=+OFljFsRNaFO6$`&%$Nd2qXSOWs;JTN`>*O&uasIiBP=zE>r3~V!)!)>cdY)
z*0#9#2tOMY9)5eXTN)KcB_YStY;L$cnzwiwL;-R>)@!t2Otd9hmqzb@1&Y#4kI3Pl
zX6wq)3E9e~t=x&}QBi_~N+(g@YbWBTz6|X$id+qJEN_-#z|&^HGkoAyrBBc*QTvZ|
zOHyG48$RJgEuh+DDRAW$M$P0a*l}5KfJLd4%dgZ}(11@9A57yuy*ge6TE}xl#J?u2
z&3cbJIUoZwPh?U44W_&qX~G-OBX;i%HtYy4`Bo6|!(X9@h*`VJW}XMqdUw*7B^-i-
z(zP{`gNo0TwvPG+08t8im4QtStgudtYORuqCX84Fe0=-}G0J``TRp31u+t_9s!@yT
z|0!CjKtWQAwz09HPmua=*nqJI8#?i%gJEkXdT1YiMl0kxDy<nX#qRNJ>jwanKSKC;
zWo!EOYVSMlTQx7nsnXVmuO!Vc)d*c~mzD!~T@UWaL!>1ozm$?fxVXQ|P#2AR>`j#~
z<3Dy597d<a@_1cMlxSCaWeu2C&h9_E&T6JUwL*DM1c3n|3b}fAUue*@S}W-KC)3@3
zkewqn!jRy<qK|^mP4ju8S2_6|5eQ%Fxw>+ru1QYeC-4~x=V#dFvqhjVTVw4ym>%?V
z!2ZArM5N#DWND6=EDIVRpFB&kTt3jUNZoKi)_K%av3rp)juMVQ@oID|UTQrx%5~@8
z=fNa?#YzqmF*&QzE%>O3GQ*C0Jw3C*G`jHy(fyffxOirLlA?(p8unBIr2~Gy(4UHm
zKMKYdEnd41N$)6MCAX?sC>^f%EqVzV7%X*^Xfx^5Dg4mgr4aq<b@i&FqeEX8S*O|K
zylW_9KZAW>V8FUFjhl2q;Hi#O0A<YF!Qn(uaGSrbzWy^CAs`p5-vQr~S681*^8C7q
zT%wj5ACE&o&|{w_l=X5?hdX04s6L<OiN1;^=ad~B9Q@ywF-9U+t>{T30-$Qd9A|?M
z851+As*3IZfkfu!=Afhh#3R7A%pWmta6R;)s3M5Px<yz^Wif1T%=A6o7)VY|{+saD
z)5C*@BTOjEZC2_RyX6Fw--C^c%B0N?uD<4x=ZVx=SB-7$UhGXfW_!pw?S1<qjHlmv
zeGsKA_;6*M?vCm)=dDm|oZr*aGbdU1dF*G_r0xAXyD6{7J7Y(KTV^nqoy!GFJ%9VO
z3#dAgd>^_tGkse?QDMmBI*#%tV1-7c{8(($X|(R%9?Kv3#olQYUs|_3vO3323%>n?
z9LRXivZ);NGtIYQxgvaCp+Z6LjLzPr?oUlAV$!P*8oeB}2^c_341$pX^hskFhV{no
zm|}<X!(8Q0J(7Ay1u%j&VFlPRPo7e}&;1RM-xgZ-1?BSKI{a2&65aqT#)%sFsoBGa
z--iJO_xA(sNnf5cxX@4rf2Lg8)!5dnk~^<I@)>EoNN+$|5_paa)r^*)0)-3{8Ts|)
zC;7mn5WOhvT0JIye!#{y1^b`(ofoMX-d!Do`SiNj(*zMsg(bOx4TKmarP%y@n!b&m
z&;$*J3_#ib)$&=?&e|?C(Zxu7#bZpt7WXAem%BV2672r-hr?>>sB%c4Z$KDGP_G2&
zxhfJ<QarYXv%(^XL=P62;(BlHs=jTyL)41SPd|^nvt$xk>DpCO7jP^47=T#%TRto@
zl7z=;i!Sdy)`~n+a%LZ@Wr><4;7b>@x(2MyiV~ZH>8X|>Bm!Pn8)X&db6(?rakWLW
zQ`jvxX0iYddO<2nI3Q$UVG+YTgDE#DBhv@gIa$El1K_a@VfqA=g@u=4a6in;K`c$`
z@p@kDDc>U+cA9_7WV4(oo<omx+#K9L9taUw{PEaU{W>KClN4N-vzt$}`h}Z&j`3zk
z2<GOD{gXz91EgSt#(4oSCCd47z3V3xpZc;tqyGDaZg5%r8W|b6rKRP&yrZjYsaY*?
zgBP-Zz*J53McQ;}b&gum@<C)Or`^4Um6a8fFFu$J(Ra^3Q1lM)E>0-+ME)ALb7)VW
zg~4F;E;>(}8(|q2>T=%P+`8ZL>8Nz1MHhFmpHseZXO7wNLPQ&GqH$vKe)pQU#n=?g
z4_Ae+{QAi;p>_EFtm6G6<+JVW(*2fdlipaM5dIWsnsgbxe4>KYV?Sk6y*onMCNX7G
zbjP<788lD!YZrO=_Gdx4%R8v}6vFv{m{Xq6wHD1dv^eIv=-n;z-9e=2P>m}0)UV}#
zj}2yRZPU{1Fd9~7sP94iy+&%()k~Qa(eE5K^BVsx#yv%{u}0*o#)d7keb2;(@=uxV
zw3wdf_#}!|{Uv5YR4DlxC3rm65p8WL%j;_lMJAeWy1C20p>}gS7QgQUzXA;xtbPH0
zp5N__nSrF>y3>hH#Np-Ebh!wbGAd8wb`%4G<8Phsz4Qx6QgU*7wKl|l7yGlfH#egb
z8`Hwtsb=uj)*1X&)$Sa_6C2zg&H14d`J6`2emOhGIH-1%wDPmDpD)8{#Ky@cZsN<m
zX}~X17(EoW?aIibSN(Oz<LGfeyduu;y|jILvP<(NZfH?!xJTnp!YYoMrG<=)Mewhe
zeH_2Ex~7`m{IfXXV<U8POWAj^$7nikji9BWu_8nZE^WElQR(AY5&?K9eBr`#Lr3dr
zm(K^P*-eeV<}jl{G)U9M*>S2S=7aYPp827k7yfAl`!rwlJaw2g8lJ^^rfO<RBTwo2
z;!XCW5MY_-9^Q+h;o<X96-QnXL$i&W%s(o%&N)*58jVj`Srn-nSDR`S(-*PN8};tt
zkIwFSq#{6`ox+Dc!^{8lLueg8f<7Un^&si+0^T-kN}B3#x3pK+iWgQ`sfdo><~@~>
zZ`jnd(R9V9@M$Xm3QLNLc4t@Yp)^?Q_q`3Cmj{Kz*|h2;kYA!t7!&MVw2r#Ex;5u2
zhBAh?{qoKbjYMY$2P~~eBkC_YR<=f(YCZU~eL=@U9W37r4(q6<!Xs9>Ad!Em`xnpV
zXsX^1gm$HMJ!`Xhr`;LbPY3;`9xwtb@rT!A1D}(yQ;}}UKcw?G16T!M6ag9rH%DIu
zC12VMPne6ifX-;vd|^+4E1G0M2}i<stWEh(%@Wx^@*g&BAO@_sUT0<Id4K8fq_cbD
zO{zazb3nqV9kYAAwDX5ne&B^O>iZCUi<|g;xbWj+sFI>z(sN0wg_YKH!i>!bfSFnb
zFqaPwB9l-ZNyI(f-M<f*#z?FcC@*PH#Ppa}B3Ev_1igp-FZexO7t~=Q?G?o=%p74N
zMjWX@*DVxy;ouVg$q0RWjm|*9HhiufwLzVQ3oVuxki2`J&aXm3YIej}Y65Ayx!Q^h
zj4gTV*<n>9u|C>Q|HEM&pBOo^f&YHBJ4%}8?d;*yWTjIthp2GN^+qZJDqhomFIF4s
z(Tawhl`WniDUaU?ZrSIKP%ZWu#G7WXlm%1y3ayM(MoB#BwCPxr@;~1Pknt>=y09!+
zEv-3HC7#E;a1ixxy#1DoLzQRJq5!M-h-LpVj4OFF;LXa);NXKxk*l6B&e8Glcdf>a
zDZ_xavQ@$fW{t8R(^jUMYRSH=kQXi(AwpR!-(}kS6E)G+oxeUU{@Q=yB5p2ddRcM6
zG87_LpVa%B{s%0II>{WD@x06pVs$)qzHcad@3J*N+>NNJpH>gL8r@)M*Rzat%<AN1
z@&Dz%FIj~tB8bb#$V7>u=l?67A*+M<VenoI3OcDUCBaX%`aS}<2g=ZFwdEwBACr3K
ztlxbAT5ohcoB;m@$>Pwrvr#@{BAJ)8x#gLkr8;H1)F;|_`Th+y>4`((L8!I_1zo?i
zkXs82GjdF5z#A;&R8MRX)=v-Xf9J##s3S}<K_Nj*VN2m>`^-jUK58!aQU?atW+UOA
zV&H;#rs`0(i$(IY2Nj7&dmLFluK3`~1C#9>#egD0NNVbAPYk8>LgRFn$SVl@gf()B
zGAB9t$2=w7l}-X)5^>DcDBW9pCNxaUA+KIgmNWT1)v7I2U0pBth@Aiue=nB~*X^y!
zcD8O3?^JYN-kWy4NSX8QuSq0FYFxjj)h}|R<Y!p+iW!*8Z$#ZBU(u2NmA=wR%aKA5
zzfge1psU!KVB$ijKOl(9y)=Ry+j5sJNyy0^Oq8TjQ12Y&(7gTx@VeRFRB%n^>WS!F
zOT-^pvN+>a+Zb(*#)gK$4F21Y9`P5*urJ<8J`daLHo2u0(^kjF#~B$JhZ+M}Bj&|w
zyN&tnbU0t@zI~(3;T|>jGH2yyOQDe6Fpujs@w`8?M~l;Epr=j<K3Q2&`us{y>8iuy
zBaj7r&bAa2KLLIB%7I|_4S)xr?Pm4EpVAWmE<}1R2mj#>yR~<PDm<I10?oG<CB!&o
zl{Vf4XfNJIED32;yvS=;D8hi`Qe_HwH%P`&3qpA#NjD%L6o&sq5QU(<j!jG3TV0Lg
z)^R;tXbwQa1Qh#L>w&D3f0o5b0<d#^{)re$akbw%z4kib6O<Gb5ARY@yA$FZ*Vjgq
z^vkFA-%+3P=h&ZKqpFXMI_oVU+LGqXlUTEgr}le3KC->*z*%{kamm7}z<6f5_5034
z&jrAe1P^wcy*QD@I+hhXRxkH)H{wrx<6~kBE2kIAF+i6?BJzfIk<S5x=(ONy+L(8i
zy)iM`>*gRmo@xH*L2|zFe|rIL@%|Nr4b84zjo{0F4L|**nQlm+QBpPwV;SBBy5e0b
zhxmsNA|5;b2&maPIkWxxkg(_c9+bsw3>)jpt-jSuq2b|{t#<FIW&G!BY}@N@-$<Dl
zD~NTn1i4AJuS<K@+K9G(=}|D;`XSxuI{%*F@O8xHx1YC&tMEC!Cbf%CFMQ)wWfmZ8
zaMbx+UsV;){5CEzruU?|`JoO>X}D-v6As*$u&^-Dl<~aY0y7M>g8Xp{?OL08fPQ>a
z1t<HH-x&{fNgLFcehuU@dBqMWg>N`YWUJxmC<DVL!?NYtWmJt8r;WBR8Rfb}J~OKf
zwJT=S8tgtMo`51ry1^N-tiZGO0sBw(2N<Dzc4nwZGAp`6S0T+iR6K^^Z6^W_tRi*T
zJUm2XW%Z`DWWCY%QJ}pere1-*xF3vX0b1fF&uKIF;rvAzgU<ZY@^bb#o0FsEi%%Ol
zc#ZrsCIWcaj@tS1&d#mYYy#}sN+lD<7W?nc!xw}E%V~9}OURN`qx+=$uGDMgOO*0Z
z{(KWIRFaS^jEOneyqtUnZJgID(f9q94#7|Ib~MGKRrp~wT`_%&=kO2Y>CL(N5Xmvf
zU6qr4qva%M#K%o$fhsQug1ogn5xEuzgP+4y?X#|!k6He2pA21XwuYGGCXs`RE^}wC
zMy+Ny&D=`I`(y0KJA1|3i+w>LL~IUa(!6;SK`KbX$_;UIGchr7b8{nmgMdzGG{RS9
z6)?*gBC6th{^5;Od|R5k`4oW~(QiFE9wYOb_ep>xhWm(Uyr!ZVXM3$jM2GXP|EK1A
zd=b%(`J9Wf?tJwU*3&!2fa8<}J6YShU!0xE{l}(#B$cM}!HAF{mQN<5QYh}C^;NMt
zYPk|I)&Ypc+STm?10{?82&FbMvGMWGTfL5sAbwiTzjf+kc@esSwhBZT)hwYP5TGsB
z=2B9BfiBS0+zgU*`gOwDnUkA)v)fc3FC!x(z7Q4=s?c$vrIn|nS~;zinkCAJ=cRC#
z=WSkw3mJWW-<d`u_#2APjvF@ifOa&L#vLc<>%+`k3KHhy5fz^3S3{VNO5xXLU#G20
z>lg}AlCwkK(Jpkt{o3*<LCg6HAmxC*$Ly@<<kU!C5(5y(4&x;cleO*6cgAlo4?*3S
z?e3=OqjB@_s8Q`1blEkpoGuPR)o*Yr6beFNCRHsONBynSky{9fB%ZqLCiSfXw~->G
zCrdu$CG>0jCV$XQGrzTXQh($%8^<g{`h&k5rXRxBqgaceaJKL4jbC*R-mK{HGOese
zXCTs;rSa*x7(BY(sf>+|+AK7^hC-n)Umk7bl$VzSJQpWHrQ6_i?4b!Y`2?7*x9grz
z9z6k{)9%ujQtK5Ge0=o=ZC5tiuwBl^hvV**u5h?`TE*;(v1DT;%nuA->+GvfdJGTE
zGeNGw$Wo>Eed<4(vzb<2ud4%!S8n|0-%-Q4hg;+IYyCa!8?#`S^V3{!@yI@}`zJgB
zBjPmtV+G-IZyH^guQVN?S1TrXXcYjHr_JDFOn&%S0(SUO!y=^L;!*VSV1Ci7{VVQU
zu}uCbhjn{{#SD960F=*Wu(PwX=`VMq{;Iwr4e^1`nl$bY4ZOeAJ9wX?RETnUgMv*B
z*A^oj?!^b1TAA0;S|yPY5sWGY!+nfL#V)tRIotknzKTV>`C~M9-sCnfm<zL`)f4)D
z(XF)@z48OKeK3`C$bbbfwY`CTNj(5WZ6dw|C{`2N%-<C%hfLPcOUlb*Uf-df-JzH|
zcGe6Kaag%^f6a&_<|QFKC)NNg<8)o{;iQkbRG`7@`otdVHNDC+zhmP}+j$1O@DjBm
zz0$>U<B)L+>yhlB4;OAfRq3Zc6sc5%s1=&Pmniij|LZtrV$`lVJ388$uH3BhKqu;b
z<Bk3+^zO^@;8#1dpcaKmVq`i?OyZSNy>}@x(Gp9`npNf_1qFZy<oo$S;(7rxN=r*y
zVBqREJ-t5NWG5V0()DDqBq1=p`X^a!G4A1LxWC*H^3H{V05MOxRJ-<Iz<)JL@Zm#}
zO=zUpH420*j+~5zj;?odlJgXqiHWHh1*ue{fq*1~-Dc6e!a{XpGK!e@;<|sDR)>Jo
zw*PRxNUzxg2;$I=_}G)YTrXx<bx4%##wpLeX2eC494RPngt9SWV=4cfATTUR)u7^E
znUDW6eVNwQ&JI46191;EILbehRI7HL%X!5Tp?keANihdcD*n0M&doSj$fURG2iCIX
z45&G;SYNYgeV4hI$&t##k}L$hyvgRf-aeBG&@O1ABnPD+O$?!1Ff=sO?Pv)GpqpH#
zz`WTwebJZgnOwt3vyqQH)*)Vc#3$}ucyF&}nZtc5%Z4!a^oi9x&I`wmj_L9g-zvow
zA+7OVGCUL&juCO%+*QB;BgZ7+<I2*M=JN@RR$xKvH)ZJu@Q}x(CwkIl)qntz8rzM8
zK0&F_>wQL25`IP(c`@6>-Pc${3CZ6vF%L6KeLzxat#bicsbr>Zgqim}Kq`@i@9oxE
zGx%QP+&|a9m}7U5NuXC{CDjGe$nk1-b91wJBy?J!%@$;NRGUT7FF!&}2e6vFDI$J~
zQ;$)`eb!AuDL+QY^Lj(4FI!l^kc1$)Fq3}D0oulHRmWX|B;u_+%;&Pe<|KGDZyp%9
z@^B9qj^A=&p2}f;alYfcG$Bj%W^jKB7-cjWH1`t4K!@UTB>J!Z0#wOe*2&2!fbi94
z!G1&CISVVz$r8PJJmhkR+3?z`o_}ZSJYyR-ve>9NontiUG?5e9#WeJ)-K$&Y0Z4lF
z4xcW1#}=YU-!yn;`@jbV8=X=YjBlNyCo%|av(`DiI<^o0DdWz)2Dh}<8)yA<=mTsC
z%hg7;Kp_i+(fRJAFMDjS$<Acy;>8&npSRM;d<oDMp4?gv8J(6hwYI({dm`V6g2I%4
zOfEk(2a98t=r*bqjn`PuUO1R5f{pnuipX|;3bX|AQhmisOATs`*UGxJIX=5Td{gt4
zE8;C0Tjadk*Mn~{QpBw<2DL}EkoK=MYku-Vqbsvf8<|Q;a4M(I!RYn?Le-Pl+;j-B
zn&3|7aoL|??`s2ILf}UN%>^KJFag(^ot<4Gr*OsWzI+Csy|c5ky?s_hYlV1O^Bh}l
zEI)Xlg9kQnarFt}K_QVO{5&3KTb-REMX(>F9HVqBzD2^GH}&-GP5Ls<RjmDrn2Q4~
zU(`d;8rZ!#P(s>_)M_k#UT4FjJYT+gcXj=d0oL3|z>ilh|IP<$F(GtD=_QEE`r1@Z
z@STm!@k*CIiDLd37@uXiSV(BfRI95`H9cUq9*hZMIlw@7SSL6P8jLe?nQ)Q2+F5sj
z$k+tInqb<?D7#;tzw~bD3SLzBi3SQ6|8gU^cNuPbEdExZFJuq`{*`W~^c!lbEID0>
z2G`-q)Cd9!j@z^YVxKNsN+i{r%(3l5!f&MylBY{`FY=m#l@x`^T}M8b>bL6E*%|u!
zK3?(?l@jrLoT12*#xsUQ?5cBEjuvZH`g$P)CP6uM3HiUkI?ZtO-&P=brGp_r0RIE^
zRNCk63N(JaT=8?p@m_7|Q*ss5g)29&U3v5#|FO}B%6z-lHCb#lmkh2%{hJ{EPdG4`
zf(ZljAWPDbtMNx2sz6cr6wM2!!2aHE<G&W`>{rMADY?1bfj~9YL=3s{YV^H01EME`
zxV9dkBG&mbUB&e;!+dpGe~&)gM-Z~_04?9m?Fy(q_;U9omYg4ce<-TyOJq1YJr&k?
zoyPO!*~ic*%cT{S95HmKL%2ssflBq5^6BQ#qs6M8k)mZ@_U~l)Xq#%i;q4*I^G)ZQ
zT@T@};|;=j3t39=FA^nog7GT3>G=0A_U3m!bq&lWj}Su|yxk2tQZDJ9oJ22>z${V4
zNML|O1oXf-ZBCkPH2l>S?%?U`rK}L030kfKAJ{WJ2hI1YQtws2X34F06r0cXQr@op
z1+YSa>;(n#Z{3fpKg%>;p1K*VOQFhT2ze0VNTu;9Zrmc5l9-R?N=lMvNyr4&a%Kn|
zUtB!YDvn#Q^YM{m!G9RZjv^Htyz}UdeFK`@?mz<U_^wO&<gZKNv~4jO<g063?fV_X
z?Q*j*07Oar#Jf7f4n#AvO(wzn&ZcG^yY)Imxxa{VQV^ch{U$kIr5xVEkp&I{((f{Y
zdC5Lfl5v$*)_K`2ao}fE`Hb0Sdah^QS>J+wZr=XVC6V{ZQ~ZjRQ0cnw7Ol{Pj-1>O
z=8upq$ZXT#*;>(>8EWWjIbwI5IcfR)X46q=Lz$j6(-cvTr~c3buhsQFSjgu0Bo<Z)
zmL%vEibaulKR(=(`reILRIB|`ku06Oy}DxiJWAy#{uPuF=)=idhGf*pQy&6%;1Q^`
z?tvrY=hvl<`^5Ct`|1dEb5POI5rzwu^8?O_I)*R>nRw1S&ay?rTfBNf-ku#u50O%}
zJHFfa2oKPl_xj%iKpODqRSJO5^^o9a@0kFnt#pk>Q~`Sk`wX_0ZpwNW*)!C@zE9^S
zqkov>MAu47Io-!V$?1+F^_3{G#@`b3ys)a*d5B`pB2U<RWyQ>iUTA^|9`NXeYk~4v
zOT)Y87@x)I6H;V;T0LqV%+?-WF8a1QUy;F^`ZTAsE{5+ZHgx*IclV2i*SI*_Y;O{u
z{q>%tP#%lw<dYoTz7+I==BS<89DqEbRH)Jy1r;sf0iDN4>q`!so=m|<kR(jMTNh}d
z&(<)NydD8ajUmdWa{A_CzmaEy?(Bv;v$om&bmJErE|;Qs;T33R0DC0x^2*A|fu64S
zMvVb*iBVEgp1LjplLVknFHlkY{yayG1U}#Qz+P}_QZ(<QmouOG^M+FQd?}%M!CUr^
z+((3zhI$nRT7jjO8ch?>A!=`G&r>P{4IWdDa*ecz3SRtB5sJi5i>F3?pMY9sQH^pM
z2>OZvFtUEDCod)bI4{d=YTwmkfIyHx<&Y<r9xf44qs90ye$R{N$q_D7y6y|wOpA|O
z?+E_n=mnUb3tEy~QpuFMU*9<e9GfaR{9171ez<0cK|^D4=Yi<lI~s{;f%{>^>$7dW
zcL5+3w2Er2rdj!caNG99b8Hm#jlkvE8G}yU5ol9-dwF4y3d}7}NjhykhXmZ_*uA*H
zfcW7G?jMf6<Kw#<j|R-kmc!>+Q<FuiY+oW{-yiQ)y}uXx+v=p;JPZQJpZ>OC+_m-`
z;f3w+(h+jeS<>THU^4MLWQVK*SO>7Gi-QBWn{klr#cZchg1=AvbOw^cqbl#eju>?C
z;QwFG*yb1^0SM2Ze*<V{uFf8UhS&jkl*{1)@ty-E9Ua~BSz(^%CDhK@emQTx<13#1
zCGE!c2K_qavwJu;RCJr)6C{$7ys@!vu=%n0MAM3toAceIloYLOkuWe2A~Zw@NYit(
zWpo)`VtIL@&Rb5)_Ei@D3iN&N*%=wp7amD1>|MT`J&sq&+-%Uaga2bZBO{U5Z8NXa
z;Is{>gX3=ZDZZi*AL}7WHw5W&(rRX}X*$-tyjWVfL|XY64Ld7-Qki~-i>|V~wwqT(
zRCFMNUpr^yi7kP+d)%o8eegHDt}uM!qqXV8NMd=3+~i6~Tk>`s;62jsr2#`G2J+8G
zrNSzmdWX`3)OtW7fq~ykxyN;_{Jo8mK!?dgwubrsGq}9wW<HZq6uGb9M-e}2+v=ik
zXeF}y5C6{as>R4kh|3{EnUN^Se{h{$U0ENoq1<QL?<otGGIDVsAT58q#fa|xA!SRp
zTK_M%nP@V=c(+Rg(By||193JUi9E>CZmPqvmCZfh!v^SL2~XN6&_*Xt9PO-E3nNk&
zj#oxzHm&cS^vu$9S@0H5kf>R{$|PWjiLXx1uCZ!xxwSF#T_p9nBJU&STKEGx6MXJi
zA3$;9iX;N<s=|kW<&75K3rwVE&lbvywo@aa-2Qka3ib4U*ddtN?T`4_xG3BJeUuw_
z1KY^zpNPu2ybNIVUabR|V8b~WGjD7gIvOcMeH+aUJ)~q{D3;F@kb$)md!Aioo0Sd`
za@MCM^PNZC*J5M+rlA&;kgc5Fw^7n#T0yDfnnXK$WaYR2aCZMfJ;_WQL$}qZS)@xU
zO{BtcU+d~-SrMmQYE@C?bSfE-QLAVDgd~FSSIwghuk&bESDt>W4<r2D>>4JN&+RDH
zvuHM>GmY031EO}%TCZN(*4D<E^YiCV(38_BQQPcGI;;+PGL{4036fDmU0q;cV1ddn
zFfoQ;AF+Pl8<?Ng&lxcXI-_N|#>wW;V!4a;VUfC)dmWgpzowPvdlS|nand?U4%Z2S
z%zJYq@_%>tFQ~%bj9WBYiVGdb7AWqDna4O@9`rRe#WU>Q$zyAYQULq}P{1r&g6Zxm
zO|6fvW{Fm{B8^;@k8l}7?x>M`W)sKMnT@@@{Vag%y-Es2_G>IFD@&w)xgvWr?!XLC
z@NSjVSM6~)aiIF%>(QP=#uT%O1&Iw<84N<k5qLu=Pk-|R8z4B&Q0B-L=vDsi>1+W*
z-zXGg7#gY0ZGs_w;dQ%TONT!+iON(axKpmgz_hn<V_jdCPv~6V_M!g1^jQxI%k{JD
zl8g89M$DaWMDlZgjay`F-rwH?Hu0H-;8!G3)X3_OF{7;?=W#8i%K3w-eKLTVYL&DA
zx(X`Q-z4UtQ!jJ#Ez59$3;}PggT4&&;TK8F3375)z)}tRq^#X!aj~&qUH5~-;>fX#
zcP=@T?zn(mXg-tu!w0fNrWrx6Dc*&4{kef^t>;X-R!bYBE#9|4EBQp-<ghLkfSx+M
zFAbTh_-vV;k|&j*1#&<fn)3sP_2y!DCz0+e0Sd+BZ_G>yiP0ybd0ymU8Yh=;8u^AY
zf#IUmRJ~M3t$t;)&$s<b($3|Nf4%S7_7h{e6qMyiwFaDTcNOc17KKrZJR9+1uRVh3
zw9%E=X^kKl4au+4%=C0_cH^SutcDxv(s&OPU92#wQ#?OXbD)I*LxrP@3kipnhL~8e
z4iji(Uw5k`yJ)0u0Ihpqxd$<&BL>@CLi8nQVpf2;x))@WW+xyiVs*mnt3`@xkk0#6
znWPzc%UWF58ke1$L^EU`8qW9}XXX2Eg!{9sYfKaK(s+zktFGLU9HQFK0aB_EgFh<&
zN*0f=3Pbaxo12>YO&x*Pje>##0_Aqdy~li?rII+g`_zqh8a2Ur`-Nyy9g{FRF%b_B
z58xI$NbBPRug6(nC8}I{lYqnCdEix?Zet>?;>uHgmc;|KF$fDnuRh=8etK|*7S9x5
zslH5l1!Shc;NT1{d!eFnV1GNWZixe=D@R$qSQkB#SK+i0Uj28qpq;@;wkQNxCZdzC
zR@}pe=I7^c&v#u5G~Zs`7<C>QaqP}tq$@O_G3kFTP<pd4!2j3W!=@anq>WLie<x}H
zCzz<3n~%3t_bbRh=B4lI(no4;WM)qN$=z1H2W%b}Sm&YLCCA@*e)|;>Q&>!NGz)<D
z`^ZC$LBTTLSw-m+DBd#kF=>=7rFvdc+*|-SUu6?c(hTfmpxPzsHs9=2YSh)Q4nIi6
z(ZH2yFi?yv`m}<>J4Qy?wY8VCctqL^iBIM{dxtM0|3t|_xjSfWiPf_Q{-;uxz4l>6
zw+y=~`_;7@^sMoVlM~RszX@TxeK=Tb6*T^X764NK64evIcg2u;@rq$*`AVAP!mcUa
zohK0-gmM{|g79_MAH_D(6?(Rs*Wdh^M=Jbmb7135$bQ7}TIAPRYx7Gto0_!2VZFC6
zz0Ni7&L=Wy0$>Hp1h>Q{4l=1;?y#n32q<ps=Al~7cKh@~;Rg3SRttz3-4D%mb{Nh(
zhGRDsSoVOHQDODs<e_E=h$?)HFZGKzxnm(N(OPYH_c}+X{?--P`!f99xo5$+fd&p*
zuukL;|EV%OHKMP0!q+HMm6keV)F3?yfa5p7%_Nn`P^eUBDk>_RFMN+ktBesgRBrb$
zCQHP+xCDGZ+`D`3WqX`&!jtG#cV<gwkLE@CGvR%GpFcP1U7^nF0LR-Vb0y<CJ3HTV
z+7$En9y2hbqhTZ_&|XiOzMo!jc%xmyvX!NFyHka%MJyF$IIG6?8L<)>bu+1IwxQO>
zdt(^3&p6vY55MS(6uGNcyahmIyi}fEX;fq+C|!2IJOQYfW+FnF*-!?FE9r-vKS8Hu
zv$Hs8JiZTim8|PWETeDPKY`7F)^VnDN4fTOLO?GZj}bI}sj&FL1+I>k^AiR`ea^b)
zxSyK$l_AD}1`8jGW*eJ9k_9*0lNEv4;n8iNV0;HyE`kFCn`)N#0G*-cs?WCK8qDk(
z%QwG7fZVZITb42dSO<Enp65FW^<p>dkvGHBKwiLx4{0Aqf1AksONkB)3Y<Goqs5uF
ziS__ZCh+*;!R7D_7wY>;2zEUB)T7v+BH1d|nQSu5QzdGpB}uViS&)>AP`2FvHXpIc
zk6WSj<;!bw^5vj_jxnWXhEhzc?r<tj_L<$T4jbOaxfb`vh)~PLLgf(~euHYG4JZp5
zqF4BJIIM!;Jr*|g^3g{sngBz)3l;M4uI}#5;z04N_N_Ve5KNd=drNbH8nM1pTl#D=
zdvl=BJD^i#>~dzyY8PupY`=&5VM|JzK*0Je?oY=!t_Oc?oMJvx3L7k5a!EuYRD2(R
z|IX<#yUPy_4wLOL0RaI~Fwtto$z4Xrr>8`$COro~gt(lajhPya2I4ac)(M`$(h<=K
z_vY$PyZh7)hFAOaX!WpYqL4~-*H<?$(+o*39|%7pz%30)(<-4kyq>z8?f+*A-$%iZ
z&WHIO+AH~`Ts})pP*xpg@p0NA)I)>vqPI*yKY@<G%V}}pV3*TiO+%xEfFG*%*6W_d
z0gZyxV2;+!L-?=p4FhQ2oHXB_7Zr9%u{o##jG$*eNzz&(tnqoS+1qXApTqj!W-U8_
z)e;`=Q2!N%56!~2Wh0mcWBi?~-6sYh;-A~y^LW~zSx#3U&cWVZIu@4U@p0TXuw%4-
zYlgJI+id~7TxE^`d7r=P{!A<)D6spCj#K?^ZFl!pE0uW>-pI%Zi0w}rEFgIE&7QRe
zFn@4v$ZM!I6)}LkOc~!x5P=X~3^ODj|0@nG=yWrY=~Z~wQ+vzNVni_e|LRQs7dt92
z(9#3#LvG*|l3bS#=W@C0bsWFD$1y5zl)?L+xIq`^t;}TAUg$_&v~T0^wbyi4wZUZF
z>-rmqG@M!fGbTAbTdU5`<<l!{GnPYf2jO{`DK+1ec=#85$pq*}^|G+k?gH}!4X3ei
z7!e!M_h(jN-iL=`o}QjXD!=l5u78ZQjiYB})%g&Y%(k1Jr?WR(_8N&Z4x6)G27ha-
zAbsa4K!BJN42QPnP%!#R<)uK&R23XijlnElUBmL$D~UV>GqPiD9_KDI=XV}d88kn{
z<@e?y{b8X#&#MXFNC`0n%Y_IsQ_{ZR_b@z{Dzh^1nj0Ja^y(UNif?S7@5T8_ndW1@
z@W4!+J^Hcz)lWf-@jnAZQw_kjT<3{YGV??7Jw&6C(8i`rgFDaP-mwzrEnG9k0ZGrF
z2weue<N<}eaFt*FdU`Q*bi)>k4QLmSaN)Qf@K$P7hoJ+zuU-co@RvDbXLNnc6Edf5
z@I_{+b6U9HTrWgG!F#_2Tww|T1%VSSdf>?|?-!RnG_O56P2~-_4QM)|1o;~#ThG=!
z6_X0$Ve^y1_s^a^18O@sH6fJ;1me2)j!dy}<vxtDu?qBPwYIW0)7eUSQcu~{=gv**
zzdzvQ$nV)C3Hc%dT=#MfKn(+p#qBQ6wsyurhNVf60xm#eTwLeXKj;x+%4G&HKR-Y3
z+Y4fLZG~(qwUW+BZ$+{ft-tNU+(B!HEJk9_rBN)$eQQtYuNA#ww#WwhSMJEu<6~~u
zgSq|LT7dd&=NoVZEf`E?0(>Dp$wA8bb~jWIH+Ruaf$!ip1m)iOfCkyLFoHM-XzM?z
zl;J;6B4tb~-N9*Jz;Hc}mcCr|8rMrut!Sw0wX+u23c0Y8)Ae*|cCrefT<@=`G!lOq
zlzP`k?0J%#7KPcJ9iNo{mYH#DL#eYvHFn(0JAB}RO4zz`mU(NS+{4>{I(E2t-4U~t
zD#KU6*<w5}|5AC(Y%+`s@6>6ar1AB<C-ApHxdY>9@I2_v(wbHNboXJTdEzk}O|Ol9
ziWyoodn6r_PWUt5NUMorioNCl9CJ#zP~K-WH1VaOM<)_;?UT(cCeIm6b(;PCVc>*=
zI7XJPCqpuP2?iE&E?>(n31vvOs<zc5aokTKYT-kYtIu!#M2Jm0)VYxmqTijsY(hei
zM2wz4Z`xw7cgAjRX%Aa{f$=Aumq75<@v%lF3LRZC4?JH~FZPMxSC4k-<#mKqVh3)q
zNZ2STdD6$iz(8@0*CGqz^&sbDDUh3(wDl1|7FcUn^~ZMG!hTG`tb$=LrVOiv@&kmz
z+8}dv$jx{>#JQ|gZ@w5BRDWRbk0J#sU9VD|FD4X;hQ|D#8#ExX#BS|{3?eX^@_QPv
zQt%V=VMYqOA*Aub+cG;HMB?_n!bjrdJmTAP34J3WiwH?2Hbq5Q6ZZnAJ7D8)kB&vp
zD4#$7;Y?I~gWD2c03ZFoy#PMo7{ui75A^T^8Zvg5eo{egcM|de0~j$cps3%We3*JD
zskj+>JqqOd0_If;v}Aazhhv&AFN?<(bf0uhbwQ5*yJDNG_z^q2>-#P-pq3zVjDPE|
z5~y5UHfAI{k!js~izHNY9(weyw@3xG#yRW{@OJUdqXO3W0xeWVy6<qU>DJJx0Y<R4
zVpXFxZ^G)E*YqPboQQc~Gxtc5b}qLm^izPYE(FKty78qud<?{I=TpfAt4AgGxB<h;
z@<gG@h&XaoCh<&Og@!1KVhW5aSLw>JM6JJK!J$6%Q+5V_`JX@EEN2XkV&bBz2!%gD
z<TCskd0TGqihMpgu^EWxP(zlsgPwhys(2;(^$RvRKRPw#8cJ?HY#g`cdTGVK(bb?<
zm`ZHiVyYSY``f?qQpw-U#|g)%jr&~AZV~w~eSPn)X}r7?N(EBbgdMP`<vU^U=HrMo
zQ8|W`--rS{T@~`^g@p8~l&f#B!x6l1jE<kJ4Mn~61vT*4KA8`lzmX#M#czGldocnI
z0%bD;1z(V#!=6KdmbkPWdAd1L|JI75SUZ41RH2INU@=IJIZon?<e+Tk6~T)L@crL$
z4BAc&rd!k1ze6DlHK`KYqqfat^eWH&=*RUQJ$lRVAICORCrN*)-gZ5j9Pr&n?XX|X
z3cgQTfV^Nl(PQ@Q9m8(ZCmL@4GcKDmVsihId<|O^p}QLfoZBE35Jqe)!2*DK%F1e_
zufHK<rE^)f5+|o$hm)x=HBg<KWES&B7Jw$Dt^u%wP0I)OE~3IlDzD{P|4D^$=@s!p
zi7Pix@p2+FxgQx?LL0TY^$%%Y?+|`@(5kGDZY(MiS;>nI*O<_ycTPBgcG#ii#7o3K
z`sy?6#<}NcVW?Ile}3`7w%mMRBi%Z6cJD%bo8CIXfiQ0SdN3lG0%3e`R4C<J-z;aA
z6}PNc8d86TgXW){Moi@~=LG?U{M843+E1ZTK^oua_s4&!?8aU%`Qb{#;E*n}Kfu-T
z0JFwm&*sp(GBNS*9v2hV(}D3+3G=4m9--AmBlpsLU+Hi53Nz`??~Y&wKYVYVt7#*b
zZ!N&!CqPDuMXP_7FDFzK6osjQXgvd)CDOk{>?31GI45j3Z(A1LnnGA_kGX=rVh3ai
zZ%{Z$D%Aa(RD%6As~B3)yxN3CeaiG(g>u$JCXA@?h_Enlh{K_tZ>9s4-*Y*CJv1yh
zm6uw!>m+aKO)L_o3@$ymK`5{P53(K;JC`78cvb{H{+D`i_V#XPFO87*TqLLrGzlW2
zPnJ6U&{?^yI9Sk3D)6=2Z;{9taT!pNF`W_J=m<7MNfBwD!9Ur4$g#*BTS5mA^-yx)
z(BOF?ypZI9vSb)bnHh%#QlGzJbQ+qzV1Y_ckDm@kDlW%c0t^dw9o?ff<heP6;SSS~
z@n_AzHj*LmW#4x%-PHhRed13h-K=Zn^!%9~8Jv(L*@uEG*J$b=KgeX6vfET&zWf-E
zA(ilx_J>mZFBQ0L#^*1{Kky)LdsjBgTt8^6IUwAVA!;_?-oE`l?Y`OH&-;ET?Xqdp
z?&Vj7lCIFnqwys3MY!(Bl?%qJTgu<;N;X-lmQa4yyZzb+f1>UnRQ$mJONn(suI1>z
z!Wmm`d=6_pumB`sAt7p+<R1#OP*m>7S#Z`6SVm~So}8SZAR~XCD4qt~-_g-gGPX*C
z!I`TuTQj^biTP6D?P0un!JC!%im$)3KH{H*6MBBONl3KJ#mZA?dleW;)ofDX^2O)7
z>Nv`m4@_7dl6lUm2kWMBGGxONxqNP%F(0HRO?Qq7jg7hAcSJvjIY2w?p+njPiAY0s
zX@jQN3MxU<*_AW2{s!EU@cK*%+xrZfdKRY_35f~OWYV(K%~8zhnL`f6{~xN}GAfJj
zefz#>1f&I|MWjo*OS(lGq`O19L0Y=IOS+{&x?8%tOS+!z_xJo?+$@(1Q3qyb&z>`m
z<8y`^3_beq4Wm$Sb`N(==6pbX?nnMUb0Vd7e8M#kWo=)RJ~C}y_Zolym2i@+nXX9s
z)}(!%hzCPOmQ8vkz4mi(Y0#y~^3<2Q)`x6QUexD2e(wTiPFQ9xKR?@Brr%v8oL{S}
zvnEK%Fq_ObHsJV~5EZGv%^)H|OXjqoD89Wo5x|pwg%l_h+46AqSGZ3Ou_7#e@$+ZY
z0BHQzgHjmU2>JO`RD!AXoZhI~EB)<}Yv_34rZ{01AtFvV6@LWtp(TBAI#x;c_rVWx
zIIB=WY=4ritcm?*%@}ioC&vzp@KG`?!~Aoxu1N>e+2toU{HB|3M#{DA*MYtYh=IN{
zei^NH7r=)8?a;SWX72dsL!)rE0y*ZgiEHf-{|dASqOaFGxA<EIj<1WSQD&p0UfaGd
z=0(%;+Z1>_hV9EF2MHEna0m$r&CJXM1OyBW3>ep_=PSCpxO@-#_mL_?QK#`$Lu0Vo
zQdn3Rn77o`ljki_k&q5%N@l|eIBGTMhpKayEHQ1dLSV7Yq9vX>!>DLR$24u~%*yKB
z@asaV8vfq2swiRZ6;xK0HcOZqxcWiRw8fBJ&R8WE)FL|kD=dhk5KW;Zp9^d-VTCZi
zfA?w%p_KCp-?T)LMz<ZFV`FiO*roPcbIDkDsY%b4m^5v`ASK*Bw503v$BoJh&?})=
z3EWj9hM2}+E9Ui#LUdgtwq2p4kYIr)TuHe(1eQGzS&^m|hqlY^gBTW?+S>%OSI*HA
zB2$E8(Xe`YbBeK_ZR(^=eTB)%A+gKWdg_6jaO9omx~?iB6A(HSm<*FngyzOK_$Y*E
zDWv(r3x*~V(AtV-Rh_VB<dQVs{me^7hM9QLUdc|8f3{>*gr&ws>Si^noL7&yi3)`y
zH_|`Cuj)2!06vO*wQmd+YnwLp0v7fxiEBk~89G11fScjEhDrFc?k#IOjr3ou29&V)
zhkc%h7~3sZTDX7@8BOb><P__OiprmvU(T;yX{08RV<G)jaj$INkW)8Gl1esW!hr3p
zwrt6CHWyW96ABCqxT)o)75$9e1HKq(;=3}(MSeE?<j0#XM}r>skxq6>8+)RA=!;Da
z+Z8a^P7VY>12b|c>-<}R#vkjP1bxvESw&2!?RU|^RG|g^G&Q317?-?w8DGMB6jKT$
z{74a@Bwta-kPRs0YbK~1GGNy5?lNFj5z0q@Pb%Qb$VH4HCB;BI3r2K_E_^<f$?jyT
z`CSVX0wxQCA2`herERAniEA#zoH{5$^f&#xsf>p%LbfjrV(ecq+S|ti9M*PruJ_tO
z%k5yg7z>hML{cyfo?{N16<&?P2;uCgq3sVJJ^-InBo1z+Mj5QoV*w4Yuq`$^>;m1v
z<d5D0*_4rkiCka<ieLvFDNw+_K3(;uW0c?7r+@u==J1L;oq>*yj)sAOhUNeWEdWvh
zI$Obnd3n~{!QP%K^k6(YHcyTS9FrxP{bW8iBZHwq1K-`fKvgNkbWoguZ2?cdc<by5
zn)+6v?q|2@MU3R2ERJ&(RdL?%7n)a&&L!1q)P4#fqGs_6O!=DTM#`lX@^L{2x?)b1
zlOv`k6s%T{$`{8BoxX*qxUh}XmV^X*1^rX|FmI3q8tDo(c8ij7(O(V8?r_uGw3xSw
zP*ql!M&*HJRnGzo^-*>TtXuVuvKXX@(8QQ#`8{;mW+6VtTr0iR`YEX2VpnJPVYV*F
zaz9gJLMeob#N&j?kY-6t_Er;)!GPz8Od^Gx9C~5Y-epY!8Vc;b=e$sU6q3;BirAu<
zYeML{oZ2Fv>BcZ%p@^|1qhR%n)S)C}VD${R0u9V!$PmY2&5ZHhiQ4myUGVl@k_(}e
z{?chTkr3}jL<e_-lY3X`8ra=jyf{;X{5iDLs#<Qh(GY%?11(6!>f^d&(W?3_i577L
zZZNP}hKO*dpNJA%z>1JzbTLKF(z;?N7y$%?nwHkl>1k$ivRTbMKIqGps<VXVeB=IV
zZ(p67s_E#+COZYnSf!<<!~Olg9CjnXo`i^KI$!;vizz-nK9S+8l(e+PN{g$E%<wYk
z1uiwdM@MAEe=UUW>+1`Iw6fCDCFSLR4zJ$4d4u}(12&Q)H&GW8=)(%rj52A0KB`7p
z{k)}HYsRjtAE`?QZioo9Z-*`upm_APqXtFsjG4a*&7lhG!KA@3!k5uwmaLyRl8YvX
z@@{;LSlN?|#!MOgsEO!j`zwsFZ&9$ge^O;pVsM=)7+taWjWR!TvKScyGrj`~1=xi?
zV|OtD;7+rcpgLXxX)st#!-fTtIZK9iSg+#AK8q>Gzxkdhfk?VAbZKrUnKU^vdZ`jP
zv1rNcn-wwml%diGLzXf;aln_qL4_teJHEDD{5E#B+`T}U-PV=UJaniF>@dU;yS_E9
zoWF7`zuK}0gpwZ4B&H2D<V5^+#%Z$sjI4mYUBnEN1qw8g<@OE^7Qlp@D~*jL3bZ@m
zN_tK8=Mwbx_C6aQ`#`}Eub!U1u_=K94FdHVCvMo0Txm;o&Cj1PgwUa&ppXJKCU#4E
z^g#(nEVn8rhfF{~CtoojC@3*54sv#V?ZlZ+LqntZ<qPG|wo`L!ULHB9=QA-?78UJ1
z-d%bwEp2RUfRe$y<@MRwFV{<BA)(Id0$Gv@;iKOHVczVc6r3*pzHg)<=$Ik>hD;Yx
znDW}epj~HdV$$5)d^%rKsB$ff-fbg>kQTaHLUsN79injVM2jiyvoSjCx`f~f@9q6P
z|Kr2?Xu*W3pPwHnP7%LU860n_&w*k9Nuo~VP@T!fxgE!@(7l?Q4y7+CqRDfM(<o)?
zr|8)PtGWxbbF_|mScG@6pK*{OEBf_PhKd``xcckA1Du-E$tBBF9uVmH`GAcUfYjC{
zo!G$wpSSdyP~nB@+w*atu){<O01UN4N<jF+M}gn<K;g3bho@nc<taqaCdUL28A^g|
zFCVG=$^CDmuJ1XmOn@#GRwzlPK*KDYTryu#omTd#%#OWN=st+{a#E`Ce((nh!_kL<
zQ^xO!iSN;r%X(}fYn-E8t5CdSJcW9MeOxtGa_&M<m2`(`^<2kr+pa6gRt|p~aIN~W
zY|~gx>W@V=&tW+u9J-}jO+P(<=1g59H5@@`So-BUpcwTGLjfCYx!_zQH~nO|<W2v0
z%+6NSenn2GLs>VYaeQ%`)f~GPZ1qiU5dy9+_y%$*EQ{AUWiHI4mZ`Bi3jaHN^K^pM
zsK*cQ52NhTJs=7Nn}h8+T6cV7|Ns4A$?g^Ee_vkz-Q&>rffM`#a7mGI$A}Ww{`VP!
z7O!)fRl?xuZ%vwuP^+I1ogM><Srtq~6S+mfsagjL#B8cia>?Cc1o_W#27gY%*tcGB
zIxN7env_75kE*bw)z?h0zRs$*d@ng^cKQ1$p0jXm4T&*rhV4+{be7!4!}0fhejlu(
z5m)28T$9%*FQ>etIAQrJ*9Q_pn`x0ihoPbwq~6SFUd&l$^<qOGj~k#wQ#WrpYL+_X
z1T06wF?mx9JGFI6z+|gk;jFQkcIn4$UJTA0TeJ}tA<a&rC&t<6^nR?3Q&rcysPAqt
zOyH2GmBneArway4Zpl%M3oP;fuFWk>+E|y+z!QEiTbtWaw5XQZ$IV4CIUf{6C8dNE
zS1QCcptUltNiNtwJ3G^uXAwe2<w(I2j@hcg*}J#MGD{&v+^r4Iq4Fb@n7mWEsz@K<
zvT|4$-T9RzXOSvMYAeU+ms`WN8x_jp%nqqDN_AM-_vRbD@p@APeHiFIL_9Lot65g+
z11~rZy<lDojRfIwI_e)9`a*?WqFPZN&xZy34ks-Ay8hb|WHfLy99T-VYv0LtL@$qV
zF;WSE9d7!+?<WSX2#rJxdHS^joSc^0Xfh9hS3Hq7@vc&iH=n$7(3z;_ip!NryQ)|?
z;Uw4O6XPk-_tco_eODijJsJDDN4<0QzJrCI`8kTkfL}1xYs+ah0;qTZSoGhwdu71M
z%*+g|MmS&#Cn2er+BaXW|M~MLFna%sTfcwjZDVI)0Y5Z69Nd!~|30M?d&Vv@64Gd<
z00}WMcr#qCmX?*4F15Pf0L2Bpv!Vp_@hTn<Xt?oG+c|1$YlC_uus^+~L6IiMOewf>
zd#bU%#a$czNQfWaYZlhPbsRf_(e+ug{xkM=t#ZVRREC@_dRqnEol@iFVt7T59A+&Z
z?*$_^as2^iE>)JDZD$WpL7=v3?1ym5(C5&nrrXs(y`;xS4;>wy=)Hy~dazSc#PVsJ
zHdj|Ov#|kyz(3Fh`2Y}c#K*;ftW2y}DP?45a&q$b@83Z5JH_hhH5HisnyhW**U8HC
z4cJpb+ZmXKJv=-B$pJLzh}qRtRO&%_^ytXazf-qnp6va508en_{t8-2?m*^XIDl!V
zZ7tY03V<yc95&j@%JEY9faf&_5<q*oOzba8_GVm^vF_)VP+cG)Kq2I;1imhy89}C{
zZBo&q$tlwJ|4f9z1#||$$4Ws-3Gf1{s;ZfQh*4QN4nQEP722R0EGHoW)OEbP9zYz_
z7eSOevO^{B_@iOA_lWRM7cDREl6hS{sOx&aJpU8mG&x$Z)J@7#X-$<3?ao?TKYzOe
zO+}C61jXEBM^(U+ic3uVg1x=H4PGRB=!8tvmn-fB?0jzhU{|^AyDY6!N~)GVTz$)T
z`S5V?>$xH_mG`a=C!?-{$E9n=22H^~h$G%-gsXPPempN*Z(h8Cq|M^Wn#WKAxnlKb
zVtd`im)8bO9Nc^$ngKq5510csDj&$nwc0!yf!_&uu-Z))Uk=y%M@QSh&IH~)F)=ZI
z5_n59fPWG{v<+H8=I=1H8R!Bh`C^2V6@kB7EnJMn^ZswSdBdeEoN(x1*`IUaXe|cS
zIXWVLm^Z@fzM%(T-D2noIVugttHuha_FN>JQSe}G7w;2d-d5zc=gH$;eH~nuuh5#E
zx17@GT`4Z=pMARkdut`xM*jv;!Y>LLE;(KUO0@UhZI!;la%(#qI|S)<U_F$V^B`?S
zVqD76@YO_@{7Q5CkrWx)u=$L~<;bFb3Jo*fVbRPn^iy8#n!R3hyRfmIn*Ye353Ht7
zbzA89nk+gtAfU~yma?b{rTV#GJ|)rPJVe_Z6hJu8v`d*d0Oxo9F?@9Q4FyhOe*pA5
zNe#7fL7Sq}Y?Tm^H^vSQ0-1~3Z<KLL#K@}V-X`oWf6rOm-hL<5pqroi$NVZPxTeJM
zgzU4MXndj!O+m3j{)Ncs6JnhdO;%WVFH+6+=5-mB9~DP(A-2VAh0cpidgu0f(OdlR
zf}cMJ|7|donq}^s#7`o3ncHS?2!&((FH>l%-Y;F*sUDjvp8`?$aa5eva^IS`6Y-qO
z%^?#jI=+g)?Z<*c=wbp|22iWh08Le(bgZea22T_U{3laW00g9=qXTsgpwR@FKVUG0
z^%HMB0o?*F*NZ%<{DY~&F(4kOql}zUZnSB1EW8|7g-0cv%ag=P9Z^wKv<E_5WDC_l
zvqh@pOQuK1c71~`O!d$SHXxt^j*-~dSYXpKuTl$qO@M=o+W<%&V8fp~c4}TG#>AYg
zF_8x{V35A7G$qV+eEOot>^_*cEbKo?TO#$tI6xsv<MWmEP(kxafoclfrU5?D@UXFK
zMu9}5;2aC@mc3ARg^aN=UJR+`!@~*#Bd&r{Cigl)1tmC$d5Eoo;qnNFP2b&}7CjAE
zRxa<_D|MDDS#B<={Yp=>so9c0u=e=h#7MnnZw9d`NAj|WFfl14BPglhJ9dvfp=`8p
z=#EPUIbz1{2k20;w2vOy@NsyuQ-gmqy`S$`thO-;*}6bt=Ob;)iw1VWGtpF?{hGzH
zm+|aR56;22j&wavu2(_xtvGEOyKq5|zhK2j);P<Xn(pOS;L+gvz*%2hUT$t~zJ~Am
z_g~E8z{>U5zQFXK7`3At>vM5GmXxm8$RZ0t6@pSl3ahPU3Kr4GPspZ<mcYixAG~b2
z*1ir7yh1fzk)|%$BgFhw`lUJ1>4lZQX(A-;VfPX#<<9L~u&UFra$e<I$O|*Ik<^f}
zK!J&<Yta#*j(Gl3Y>F8I#$o=ti-@UG!{DGK`nq|(FljB5OrBW0#^?N)U`cfQce3c;
zlXA>rvLqFSzrZ5aiC9*$kcGx^+CkQ)I$*?7(@6y}VaNQ*Bx#2bC94PZ-@WpCR9**3
zF(z!^CEb{&dcvkFB*KwytcJG&9f>QWfw?re;IB$*Qh`|v>zoz2kOHg_NFwl#oZiaf
zzLk}V9yc`E67~%@lJwhh89j&L>BE3XqCX?K`mQ8tmT7wP-E~zou$xcsC8Pd7Sv7B6
zQ0~^Q>n1;Y%hga}din4HXS2uBz4Cj4$3J~fezvb7@NGfWI$*Q+mUc^2yY1gQMk?RT
z%nYP&fGvc9fuX0T2LWL{ppaznxr2bb>)iw(f7NNcCS_r90IS@%hKry78OS(AIm%ln
zv$W%a?hUPpKOy5&zadt%D_qNxh7-jwccvY>A21SY{wjm~9h=&%x5EaxtCDx@JPO$i
zx}D(Q7PrGTZD16UTFRsjB$Ma7k}+bb%cE7Kx>Yr>P${H!>;z*V*Ex@t{4z+(4UgdE
z&9SWWF7ooSRa4u{s1dqVTQCL!?w%>r?U2!|v{aq1knUX8pglCAqccx>D#><nqqdg$
z?rlQ~+luY`MnM};dSn$^n%Z~Cs4&G2_j;Nr`WQiSd1wA5t-WGkREwIf9Qs@%J&M!2
z{gt>pp1!_b!^~Lg?D!X!YCC^#qNg5liq-=HNQ_e9Qs~8W3%;_=buuz-Q#WI?$8nps
zP4)aC(3)VLKbe1WvP7;m3(Jh`**0D-`SVdfq`4Wm*BcQ+Y~vqheq;Ix=1Yb92+qBY
ztmUCVFR;VXW&W&VZAic_Yy_vSBr^4_rG4!9B2>|zy0=gIef0E=Li*r{&h_#2-rk-G
z-wP-jv1s|_!v6xRYK<B$=xTbg3K^zt<;uD=C9rI5ZOiL02aV8!rGuKV<z>jP%ak=#
zBnM~!fjR|91%cRXk`*Ym*MayYhMWs1(?FXaLPbvx@?=KQ&mbPUzPyBC^0+;-A^2WR
zC~(prf)uMRGlkMOK`x0YBBkCqJ+wGBCLfh~NE<t1`sSf$1Up=f5R~^E(gbNPRf`ST
z9cmK`ll3~fWa-?ZX&?|?11{PjdtPyfvPwz%HsK{uP7c|5WOuGnVB+BaU{9!F`jM35
zdIkzs7^2HlDwk_-BX(UYc~woB;F<<vseRk*Saa}GytuMuWDsp*6W`K*6^7x<pL{>h
zg?owI90<EiT~Vsrh(mm^oxYGhL_3$5F(W|s7&QIEBb|R$_s8|cFHP+``_?9M$vhP$
z?Q1l#u3m|zmAtiQ*>=Hh)=wMCnUfoUMgpP_?4jlHDlBaxsE982FP>q@fP4aSU?7^r
z!o&o9J8SE5_4U9T6a7=`B>F6gC~OFwHax`1{b>LT_~6B348dFFpWq)vGa#ls>}w10
zditys>@A<55|^4HVlZb*$oB(nM|VW3l8#)l{S$(|XRADJ;=aBnDI@NCvEQ?}4gw;g
zToGx4V%%+WJ`^7%Br}%F4j=OjTlkSIR3;J`n9hmbVG5OXCfTjI%B1+r;o$MY+O)nR
z(K+MQCV`8`kEPVVr~4kDw|O5D?FiIzDfHMJs{K2o>5IS}Y|=XT|1O=vo`V{PyyM=!
zA~Oj3z4Hk%+wP+$&geVN=Rzu4TEJ!jnfxnMRNz_4FZi{VxYR<So$*Dpd+RHgz}aEN
zh_OTE=k_}HzEkT$ufKn~yE()t$;r2tWZd$Fw$9xfQxg~{W{5upp?+YQ8gYc-@oaE+
zY(TXk;$dra{ibLpn#ui+rfP1m5^NYm&dzp+c%h^ay3^_BkDuZo-(QYPis~1>vD0?h
zJ4MJd(9@0+=#=X?FW043wXHl%RqH>s42|pz5B`)2D>JYKwMy_(;3d~b?mDol-|<>8
z=FzmOQ%Ryn#W<plkAl6+6(mzG-r~T_<z;1$IK;qdp=DbhtX%FAF2_k&qM@f{!mB;z
z$toRYeBb)XR?y98M*MiAx^Bi{OxvwdTSP4WPj5AdT>O)7;>gyYB)_cf1g0<YSkTH{
z6MV-KFgT$sD(xtgZv8fs|L-U$@O<6+B5c3cKjASIi4=~>x0#twov66Fp0z$UVO*)#
z%}!)j2v(fN*qS{*-euY>bl<{75^*_!U?h~P>gd_idVTV6u|1%tnqB}cBO~UA`&<h;
z8OHrJZ6+(!lnL1`vEnIC`~6CEKJxjbAo;3@8Xg4vt_Rgs6(62UJ5oRpGCPs;h9N_I
zlOjq2eF{au15KQQET7;91^HL;3)!@L(X|3R0k0O}>?DA8I_nAhtl%@ctiS;s{|3la
zK7TIghaWF`m-wC=K0e;|OO80B<x^9oXCH}l=tR~>iuN$JlJOR}1UivLr%E;5nvK7V
zNi1l)W4?VGKp3|#WtjJgN_U6xclhDjF9c&O*vH50lHTEcQ_R=BsD2`nAL}i}TJKHY
z1|88UsaEi~VcV5;GO21d9M{-i)z6CmJ(|CCU1j`NBOuP0?CLCFIOSZ^e_=xo>~XPo
zHBoKnYqnhIskm=!KV#Hf^B(AmPD<jmIU-;c!g(ueL~*Iw+KhAXI;7`=|KZFr4t|=p
zI@{&n0L)xF+ty%%`9|4RzV<f9!n*c~a3fG%OP;68-j5E)FKzZ>$uHG6KE#2Wxi4tb
zo><AqFj@M1^?zCb#bT+)=Gbuch}rp4m4u<PKgpKY(EdX2qqmd}Y3NlNecCEmm<SiS
zJ5K$laPeJC56qOZPX-lp&1~EX=s5p6)(J~|A<1mlFW5V%<Sx?0eWgnNp+_n|X6jIi
zfE|N^LgEb)1Viz&pANd9$kI>|IaRiT+SjjEKZ3h2;&1hEFD?}d8RXIl<*T_%t|V0C
zuC|;p@V>*57@Y7<TR>{&ZH@r%LL9&rn&Xq6TigpVbBmvR4We0<4O#zK%z6R)MaAh0
zhK)RwPi~OiGa<M+Kg>XXUf}5WR`J{u3=U&g>CL<7jxMamo3m$xU!KSh*SZ~~wPE6J
z_&XlsJoWS0ViS8SSdc%Z+|1U0f%pqA@GZZGi=dz&2pXnuUxEAYzz-Q)--JUC`Dn+h
z;u(Rw8=bagcOr8)j8^3WrBc?DA0{E~nMKa<RU|S4xEzlca!E<pd2({2Q(;<(&0o6^
zudN@$<YJIvP~k&CjjQ->lE(Vx2qJenxLCrU>cDm>uoFmjSetYJafq*xZVg!}s`aX$
zjPIcR+)<^5UPHDt2c%N%zFhNaMpcJz$_tb{Q`Rs3bs2&-UC-j8?UY)Jo7u-5a9jU;
ztV{cAJ$YG1MEk%Uv~aun4umv-@khial`o+LEcpvo+ROu`Q@Hoy-*@-_D&rjP&c<{t
z#e+QQy1=Lx;f@kJ0i|5EVNO>`y1NHHwplGNK?3P2Fu5!RgsF`J1x%%MsT$hq*7bPD
z*Mb51mm_gd(T-kiiS8w(guJ0YdYgT-@mNsVQ+NFL!+7dtOG=x{r*6uznjSCeNb^%8
zwGt|3D~=|gWweD7>YIRL0wmlIv+d7yI#j(2;cE4Gd~Rgx%sXC|ZSIB28}T}RpU&?I
z*qfKHPn~yA@i|YI;D$58!Vf2FYM3@*ET8a!7IODJsic;4_oO_qYFT}epwg)AemKo`
zP;KD?y2Oq?TnKrLunmcg%_%5Anb$1{2|Q*g(E!C39y&V2hcq}Q2Giqw+y4_iz3;&7
znw94|FEsQ-%LlJ}3~r(W4xgSMe_6MdKG|W!#Gn&Gljv;i=w9NUUGe12WdGU|;`ty7
z(&%R1Z=A4x+qr!$vkj*gd4`pA^;V)!@KFP#n7KuWjITTLkr9!Ebp>7fMxzEQ_i6j{
z_9l>F5ClwV+5l*x4<gv5^NAY9|GDUsi|cG2!8xy`)0~y9NY4etB?_AphYX`s1s&Ff
z7#fNS$%!X}LK8!_hIItmh4&}zPYeieMnGxa90tl3s3E@)Ww4<I!BAN34#wepK?vXy
zm0Ea#L;IJ2tPr1@Cp9-GD*e7_Q4+jCLQ8f(x!BtIxSGGjw(z!5VZ3{arL0=&15PV8
z#<P`_d7g$NC}96KG+Qfnr6y}hZcq(wUhIN*&0u6svLtk;y3{Ue){IJBEI~j;pk8L?
z%xP)ya1lxJA$R{Cu(kg;Ls4?INpFZzCv~~@jW#|6qGQ89MqrZJ`m6F&s*e3dn`C1N
z;%j-!Y1r)bxdszD8?O9@QeXiF(#S$<ip_`CEUHq?u%A*t@L#_`Ab4!hu`5EB$ffEb
zSbY->BujCoB#~E{U5mg{y8e)!bPkokAiRvI#T%KQ=}XFFu;5=5pq$xhC~QW3f890D
zf{5BVuz7af>(kjkoJ55M6iN@xr)|>m^76H6pB0xV$o$9$9BR?UV7g;4grEQ^;qBGD
z)ZWka^;eB3xqY41eGAHGXDCC1pvHd<JYIBkGBWbwnQi_3-`fRRJkYU!{3)|LAV<X1
z|Lmfxn;?j2Ljuawj{|XwC@_J(5-3(5XlUNW3hx6Z+uCth4m39ICmPYQ6-ZP%U$^<u
zeAz=oDtpH8035wfkKZe-s$GO78DU*gunhBQ8NQl&?I;yhoZjq?z9+-%5#3uL^4Es<
zvC@qC2BY{&hN4KxlrY@Eq|y>@XQVSYn$wj~faE-n3c!9)axq>LWS63l4EKx>P%(fb
zI-QEeQn(n<VPe@{QhXQgH^lqPESE0g#BLa4-P!qu;<B!V7lk;qrjRrqD3=WLz&7=T
z^nID<;0OTsasCd;AWjU*HU7m*$kqW$NyNNmYfM_&e=quHUKy!GVRz0SRih4#9TD8*
zu-J0mA|Qh?9L+B08^goHd3kx9;SViNWh-1>3%%9$*@6t{G34i7T9&_>FS=@Qb6Z;@
z47XdbF|O5rhZDATuFg99M2atSJKn=RxL2Yb99{@Yx%M1ahnCXNcU1`#tRkDe-KBaf
zh%RDj`42w<;5Bj?ysp3i7aEG>E@o~3?rieb)33SI4@UHLDl)FP5dQnK;K&&A;EzEO
z9Ea1c!0-B1AT3N(8fxt~S5Hj9ePH-+sn8mMM??qX&0KkO%+Mj+^E-T*q+q9Wreuv_
zko}@(8&GO-2JIE>Ex|ePcldPv@_HlMQ>g)g9#VyLwD$^;`t-uAgVl_nW92Uue!Lo7
zZ&g*k_`R+~XzJ{5_mzkTD@pVfM0ap8YGAw>tz1h;kH4TL?4j|O9SjyFSY*$wtM%6N
zAVv9T(EB*ohJ}P|dqPJu_%43t<{JMIU^HY?<C!g~yV5{;H$e$C!rv$G_i+gB-YX^)
z6gR46DAL^w($lt9PXw*D;6WN&LbofiIvjx`zk%<nVm`akNj`vZ!fwCSenpWabL&Nk
zba4m^ZwwEse?nO~fA;JiySdm3U>$&-l;rwI4(-2SWCHs0KrW%IqSAKtN0YMr>!$})
zg(m-9^kr-dl@9A5Z$Z`kMZjX&vgd-tkkiuAYRbD2*vNNYS<Kxl$Brum>d8>V_UHpx
z2o_R6!!$8?0?!SCgVW|p%_JoUGK9LaR1EGgwc5UOk*psm|J@JE(uLq%X%wC*l&QG$
zk)uFZiCr9FA(qejt69%Ld<z=edbwf-D;|XW!8Y~!zeQMse<!B45!`72&f&-s!T%fo
z!v&e02JavQWqeiS+oxX~v}w+CcbEF3xwvX`L@4@Pt6^1Eo2x)IY|PTdR5!VhnfLRj
zIoN(yy<eVAPKfH4njGXy!q6^M!+K!eAC|Yc9S07#f*lGJ<Ub@bMnsy`K3E6JQ0=`X
zW~Qn}^`V99smkt*8#yo?6q%%S5zasVLG-+!IaOdY>q3@8?BvnB6R-DP?1#&@LZAuW
zZ?}H~kJdAiN6mZ=D8|5P*XsS^bvZ(SlW=ZI+`zNT&*|h;kpbYvC|y6^vQThyYmlNN
zZffYwBSn!@JRy?|`$`3<29dVGX!Ppc-1H9%raJXEugXyUf@dDo&oR3PB8F+6Jq`#o
zenq>lpU1G90OI3Q7P={`=?b*FmyZH$LTB}xBC5W|n$3|w&}fdGzd>Ilq`=BC7%w1N
zNey+of)LYhWKFIAWt3<Oc}wE6X?F<^?SD22iHqMD3QxLw2RT}Ou!}r7aXf6RF{0&m
zhJ%C!Y;azi{nShzSyRJ92~Bf%BOX7*uaU9Wdv{{=Y16m1Syvqe52XZdcT|`vbW(Ly
z#VkvKw3?XRBjS(M1T*Tg+uB9Xj=T3SjrMn9;bNn$Z4X{9FHbj%qv=8G?<);>j6EO`
zJ+~hfiUk!N+;7sDjY17x;NajyMMZ&D$E%j13sBNm=WgMw3w3v`IV1FYYk%ejS+Oqy
z@-SKw{kXa2)U;?ctcDexD7hl^unufY_iy=f25q%MIU;c$a1?MXU=dWLiH}G#>s!e@
z#?#xcw#|^C;IXBA15|bfbSJyl&w70gt6mnY_(87;ocI~hU_+q%JM6xmE-kcqHs}7B
z2TDo#h-8S34W0J5v|#bf0dOMSfKpN~M}P!*4i)ey&-@D54U!-S(1_7HMQ8A_ph2Tw
z1yGZqlbhkAMn{48+md}&%7R?7sgdtWd`KX%Q07tjc}Dt4Iw382l!5qH;|Ln9MrQi-
z=5$deZa256ecr<T1~4ds@h+0)p5XwQo^013XtN~n%jRWSpmnx-e`6>_KU9i}ZlA?+
z4%l7@+4w`l!T_^k|M=L`1vFaBMVozIZi^i0+>5lDD>O4VSDts`Gcz(jw%Tp>T7PX8
zz>Dv1<lv${kCo?l{Z{yIt`(OaCND3PZZGTdxk1D>-OzJ^xjrZ%EL#I#VT2%QmYEws
z34d`u@v9&S4Xt!}t5V^!97w=%b}Uz_7UAoUK?oiD(QE%V%MADJ{X1^>Ww)#xXk`9C
z3B-7O<I5kvaKN|&0l;xNq*c)yut$Ww^<cAiAc3_dr}R|QW(L5@L>;fYxrdrg=DCZy
z-7#^{X=m{19{L0n_y;q6c>}lGu*@tUHyka2{4(0bjI`rDEiylU)X>Xb-sK|*4=$?3
zRHQxSBTwbi#Wqa|=fcXD(Y0~i_zGmxZw_c3SZY4H5kiJDXWuSKrV^1be((9wgM2<g
z39;S0_lUK=mJb4_#5@@tGs#0sPe!)U;R{392P6gn5^n?;E8Jd>zW_HK%+~-;aIgFQ
zLZhLC503j0E1pnmhm2zpL{D)8G@!!=+*uAqUlD3819EbFo1Fpd%Lv$uH>YaGn;FYL
zY&ao-x(DP`R;-Hxj}0G!`uyuXdjSx5w42ESa7)=AMu5**eHF4d`MNLC^H231cojlf
zF8w+O3mzbOYD~tiudd*9nUhfoIiq4?gkuar<IL^q;7S2yduInY?6Rt>t3hX~J-VLl
z6BTw0`F`3dWwwH8!(w{PV+DXO;-P>JaBfZOzlFI8rKwqu0sHP{GidAuUS7RLcIGD$
z+qV-{zF9H%uXB&bu3y|RHyNB0vL+S|ut1+Q`l(4p;~^HWobsU4U4r~!sz#BP1{tN7
z(aq_OZryE58pdQe$yX;*Lw2eH{vPRs{KlMce{S5;<eEUHx%n@D_IbUWXQA>vUrqW~
zUKd?M!@L=DID?>NR9^_(9nd5LV;k_V0~@YBo-qi*DKC)~6d-8wNh4petHtF@yUrap
zx<ddNm4=UxkK-ps_l6K`JK!;cU!L)fXud*aNEd#>eopC}5x<4$v;*%h&o;f3EY|m5
zK@`=swTV|!eQoOb)p8~TNnI+e2>eL@^);}uyQ?h$2-!PI;t&1(t&*85OokC2h37t*
z%fnj73QIUXK!I1O{TM_t<sEOfG5K@7Uzpi;Im+YSHXnNbn#mB)6o1)`v>wyiI*I>M
z_43u@J&}B6sal2VLg|T07&Yatm$H#_-A`_whl`GPYPYX&7@)v8ox+iLwi)zE^g4kt
z0Jjfb^$!Rwd~aQE+`k8cX%0dqK?A{4g$mkElYO}M91!%-tP(Hu;m=VONXRDzN{^PT
zS=O*e*Gwz?z*nSRW6U)S1ZH5E)LS<5xJ{jsnFUZQ1$I~td>&N0GQB#`fFB<pFFaZV
zx=E0orGjWA!qff1#oXBYv1-P=tf<JR5Ep{?7fp`S$A@&ODI`Z2F@V##V1l4Rpba^t
z01RKVsA%2U-gX2Vz?_$7ljGse?R?SaT;jU75Yf*tAi&_X%`$j5c8Nyt_xv3EKqI*V
z&#|0R9Xz`aA>@*&BRgr@u#H+Lwf`g?`Bb5#!3))8^@B?<%dQ7mpBJBE7e@zlh;eL<
zciQ=-?N3=udYY}+M&EHgfBZ-Y-p+2V{7g(|V0B=Lhg+^?&rcS*CmuIPq8zqg%H4x@
ziKXBf*ptmt;QJ=j_x-jGzsa!^j$wrA3;#~Oe=Ywe5tY*c%0CFwac$#$Fw4WhaI($-
zyyQHl&(Mh9>0kY8CtQ9xAuVTXn@7^oQHldNgHm;<XuO$)_sF>2y}kTPOdmhmfd+&?
zg6hUd$gLjbN;)cJvHsie0Czrz?mQBN)7gpG=H>a})TU(G<%&&2ME@?vCvNMJ+i{`>
zgoOvz(<GP3J}O~pgeb&*zx`BchQJiRr|z2d`7(qZ57>Mj9L6GN0ZPX|gs|{>rzdZj
z&wIjf$0mQ{;5(i=uMWizmZuI=8xLWL{Z_+)_qWk|rafK^LV|<BX44(UYvCcXP$473
zY-<CepsNubczF1Nyu5FZH||ulo$f5-<oK!;GQ$I)(7V!=0DwgRm9DL$v+BI!Do&0W
za7_Y9NA1!PH-bAc%5B0gTNKRhNEoqI{gjQju?g^EKjY%!%BNO?CR>qZy`JxOBjFi<
z<q~}KEt`S%_bp(f0V5%pHv%B>nPo6f9N?8ZVe_MR?Op$f(C;^YNf2X;H7DzU(9wFX
z63mzatFjz4rXy|kYP$}~@j%rhbiD`)e6rgKG!t9oOYaE!;6pN-h1k|I-*JM~?chG^
zls;0>FrQbeE1G!Q;LR@@YiJzL?)jSvf)hv-?d3f*_~!T4p`WJY_ldQ+Gm0P^DVj$=
z;a=HfB*DU|05oEdfpu2R)qZq-(cbqKr~J1iFH~(7sgf+V^*(<hMf5o1-|_OZX&~c<
zmo@mTHs}8E2O!&cm&gSK1a@?6vNh3^;&NS)3AuwqVj;S_PI)T<%YksxkTtgrakpb}
z@rZSQv?@D3DXHHyLtrOKO-^<N=q9kah>3{-3F*}{f<w0A`fpc4o)qtm-Jq*g5LdT9
zobW_!YP)W~2@%0$x8^nWhJ|p%R&F*Uycdd(ADQ~n+U(eD$f{iVas_#V@m;LqlS+Yx
z=8R14yoP$fH4#L2Kd0gFd$H2b+uPgGlGzO%8@&lX1Q?Q8Oo5Mwlrl+crrCLUh9LOW
zJSm8<nzjXWdjL|nVfX`3apKyN;I#_1R>pyW{i9m(Wg`awnO_6A3Fwm0J&%>?v*MH3
z9NfT)A1i8Ufxe0!U?>7>IT5dm0>wXbx%}<R$T!CYXV5&IMggi|C<Y+NbQvT?VKCa<
z`jRW>wJ=Kkrf-!bcf!T$fukAWdcVpcGgUz(S|m^G>XH`5CN54^qN(}dAUtFJU7^FJ
zfYp@7^R`Biv_Ec01<Rv|b7m_221NJ1*<dC42vN4_dYeMr{4!d{liHIG$9|!lr1I|<
zblB0%fC0p0Lh^v0QYZvqK<GJ`#%Vu)bazQ4@W;Fkyhf!pH1N5dgk?~07zJ-KsQSA`
zkI}<;@|L-*m`Gbt*~`Rr(Y^S&Z5RK5VA+P5<?Q6!(IVfUSxs_G5IBMlE_|(O!vR`>
zqt5OIjH%ki<+vd8F7rCT2E5i<ea<Snd#~Nx;4Cj5!KbVSVPw?eRAppfP*#1-!BI4$
zx~cKW1R{tf;u&3;tTd@a6M_?V5c-`}h?D`R1F(kbS7hKJVQ@X77Tk_d<Hn<eNyFpj
z*|Wa8Lg-YCU+$kMQv-nU9rQpF_<F8Dr}>lXyuY`w@J%F1GoYtQ+sJ8^DONnscIQ>F
zKRi4LJf4#vChmrV_yokIMBb5vTo<Cr|0&X3?z?rOTQ$p47MwjQtY5e0V?%^j_VQ;J
z9(a0c=e!}no~;}zq7vwc8>SX<87I{)*S6d}<Zzx}BCT3ZVe^5SR;sr|nQ0%YvhBHk
zPpw|tKlkv@s)yQi+%D8VEmEpYTeMU@(2_WI|9!$}y<h86M?|!=X4{<lpz8Uk%anG1
zGzFdn?mvm~@oX;J%76C)j<nS1LCvG$M@Rp9t42>9>R%vpI^ca;22i;6x?=(#(}axs
zk?T`lKE7Q5i&bmmz1+(NwD2IC=Adu;lSZ<xciG!v;BO1*b)IeSz^iU_CPd^i0b+Zq
z{d=aHBY}aI&w(K`=Ns0l+|!i__E_IwUYSvh3bQ(LvD!6HG2_POrp|nWP2X>M1*Kqi
zD~v_Lj)396V+0($b?zqcLrZDr#H_@gRQltK1ewpweB|8RZfT0=J;VF`V-`EdKzz$s
z%s0+J47o=Ga&M)wZiyzKtmzR5DI`>!ptRO>+*^5nqVv1#?l9IpfHPq2wi&G$%zMJH
z+Gqh-5&x=wQ)ZTHRf6d+p(2#_Cuo*9eB|T`l3yS$p~i>e^ZF?@{AdNhcbPcr?Z}zZ
zATrt-xmXWK2Hu+!2Ir&Lc(0G^y=yFH8GP4XVI?{r&7F^~JOMKXZ6YSS6?TXS@OSf$
z0rjejtLsENO%fP&a<tOQ<GE#8YmE%S>aD%J0GdvKo;{r|p6dGZpNbi*b)X;f-XY)G
z-e&XkrZDzIgfzs=gyXC7SJ#(I%n30ZlP1<hOM1&2@<wP^w1d1h=)(}PNF>A8Z`-+C
zp!W`>u|Izzb!1klSIh>zc6)2n`r5DKARZl&_%^(4vA#yS5STVz&Ud^=qNGM_H5eN#
zB=U`a1uB{Q0DliCcN3$E{vG$8H6f0sfTc#~30Sz*?7H;O*WUm)H;0$?a-k02b8i>I
zad%zr*>-uAobt`c#uDq5C=?Y;G`(XOK3_!Q<Lqs4P>}daBm#gUtsfo|U}Azm*z4Qz
zPJfW+r~nf!?LVw#<mumdRc$jMr~zjjn><BwbuEx`eO{R9rJ<%3=*zgd-nIO-4vci5
zM127mN`YebZ*<NxvTxa*M;6%$z6-OOYrP6a%qf#uP=0rKK!N5KV}I{`es*T-?3_8(
zGT%4#{90G8f*4#9V8``vA5x)O3UqM=Wzc)|+`fM`0jouJP{?DqTF4TW{yq!k^T(<n
z6pnT!9Xi_7vI{;EiTQuB(lzRmo`fU9<>#(2c8IR*c8?Kz<n^UD_s6p2FI(%kslV-(
zKt_O!#Slat|H1f6gi4?7lIUvIYdxq6oGnBcPpG`XWt9MtlPKyy!nvE`tyjs#-THPV
zDrJP;L4koAYkJ=fLC?7LNT8p<s!n6JOw+kkLx7GP2-1xr#W=YoBqfEkXkp<DR)Qqt
z#YOKP+dQ9Fb-PURWAq~TUJEE6jwQ>mM+61oLfAwdzw|fJZlz;0?`O*3ph8>#(S}kk
z4bdluDs|)<*zeTpp<&OT+W3~yi2hL^r#Rv>GM+C|#y)z&*Dg#eX9}O};0*}79RO&G
z@o1V^!=l9#UE_Zblz-uH*E-)ub8V3SWPJ-v?0rR^(db9&Ya6(+%lY|vFpe@?!Q;FK
zH=|J<pD}XDe0k`K?2nc-KDxSzs!|@Q06~YN!h_<8*Frw-F76MThI)(H6TSE9>gs^t
z1dwJLW$a+Qa`qY^<qfnHOgI$E!kt}dV!e=7Ed!{EhvkxpDJ#!dCh}_Kx*pk35SUza
zUt`nOrFp75o#eD?eYX+D$cis(mHu5ZkSmXcaC)B+e`L7gy&IIUkbFRCLn@gZO?I%x
zi3Zv-o>(ph4IAO{!#)3@u9mPcbkjF~8Pw#*pWbUr6yZCvFC0Gc?*kFw;Gk@aU;kSG
zw|wRsIq2}$gbNp4>0ERNd_jg3E&Nw!&I0*rkBeyTjqzhJ^4ywUo~xsq>Xy0>*`GA-
z9%Y|s05b;^Pp_vJsg2&a5n~bjG2lcMOci|jD4WyE>dF+kX<z3hq+a)3FVU-hu*J!$
zx2g-2#$7=Vis)@RnAbOLbB@Ux(m4bB*ZtL^vz&mp*FvpXgcZ1-d$%cLqoWyI4itTM
zz)>IZ<Huw82XUG@Tg2d^q9YO~UTbp?kH_B7P|r;ISBQv7O?scoio$tQpDI^uD>X88
znFR=CP~RZk(=6Zr8p!>y(Jt8WQ|V`${<^$4YUdEa>gmz*RZYA&z)SM~u@d~(sEta!
z+-a(drI2-#EBN0A`=byApRMqzW!HYfAHq@3eb=nq0H4WyUy-$dwpqFQe6Yz9zDjWH
zYQK)NuIX^^f(6lWvIMEC9*Yhpp-xk{hl7eu#fuxq(^o$+uyKBFxm-*$PfykZ%EWLh
zOtx>U+1id>V(jLm1yqoMd?ON8OQ$6PSPPX5_1b^KX3!cu@if@XCiA{_X?Hm<j#s?+
zUu?6i(DK|)<?=A&p=A5RZ>ATMN|xT!dG=kecGJpc@;_O;=!*2|K5*#V94`ZeDLzTu
z|CvZ`yrPV&h$jfp4*qhCybMTum8Ux>i&~U>^w=#Vk&FQ>Fnb{6cv<zFKfWFul(IQ3
zq#rs#G<}lQEda)$W@j5^Wo0m*lsLWRJ`RMm+u+TQEh?MA+X(O;V@-FTw1jfmI^M8q
zu>Aj@+=CDe57m9Ub&)z5&7KQ0ZgEXkC&Crg97gvt#sOxbr5yW9S!&g)(Zan;7=(gy
zc`i#QSWToh#HhW+;^g=W+p%7om2L=-hBg4g2oy$$SWU-4&n4A++nz<I0<n&!E=SR#
z$}2Xe$wH9iqbyYa#6kT^Q;kG<Z@&tJX>S|Mc;%(xBqmE7$}aO$LMp=n@~0%!7z7MF
zK%EAM1VDxYGgga;HFSvSG6gktF^DQ6*)IktGrwD|q*c8Hje5&jJ)TrudgRR8E17VX
zvKTH}q9t-f2fH^@gH?4H`W?*BX&uYxLLlih=i_(KSYQ7npZ4W43cw4mQXeopJb04A
zLH>K`kyP^V76U1uEAVy6eEsd|(#pT~{rbkfE4V@pDmDz}S81|PkTW-;-2zg`Z&~Uo
ziz{9cSc&))+J24;AtgB{U-tfvPg4ItdBbC4i9_1}HD46(8B0zbJMe56@;+h{<adDg
zV{Y?jWqQ#;v4&z(#q0X`!sts%->*9Jwp)8)B=u*#)C9+wpO6NGI*;poBY62tz7o4l
zFs`f1+*jhq))obf4k%orrRGx&e$;aWnb5pJUq%cylRJFVH}RFHff{3~c&T(b4&aLY
zBAb%a<fiNRVUVW7judgJG2@5V&i`ovmJKIGAfSRpLrV+r)@ndZ4K)mgw@K~@umFJ+
z>^f6js9uDHV-QV=*B=$tIg<V_p`;qk13HQ&_ZHJaIWONix3;#j)aekR^*=)U1GrdF
zP#?*WDVUC+R*Ow`zCYaz>UP^eFPhoQzWewSMldE%INNsAO#_W5wdUN^#DQHR=Gy`y
zu^%ZI_!d^bF#ud54zOM`z+~ZC_;SU2JijI7mBDdh(52b#v9Lc_8c{RUSEj>O?VERf
z{0<g6R1b3YBOL>gD6>V$<@BBQVWOwK#z8Nj5}~#E^Ee(DGU0s}*0WNMj2kdyPd%1*
zYZQ1q>!KC#XfM=!1d?p8xq|KT=6!<{RxNw3=<P;0+;Y~l=^y6PS)Ne>5acfIC4$`x
zq%PBx-Mic{TzZ|h+wXD$H4cR!(C6nZ2Wuz4prGD#f{cb{aB9i{BzQNct2Y-H;BsDG
zT-akbKArcG$k&+$5Vt~(zBcVmJwL5@zI^pj6WDnT(rlnjCk{FN(JQMe_jEHeQ$ieh
zQ1p>pcC}UwEj}T0e7_=j_^9G0{yRB$Beh#Itvl5l;)v9QUE)?nFnGJqeQKiU6%8%-
zUJ2lSlj!v2D$jsvR2@?#M;r5njeD%J4urdDDP5my{*xF(XOx>)v3(*%|A*D99Wmq4
zEK?10)Zn{Uy@>Pcax;Ez#$EJiEvZ5O)*R_1o4Rz`YL$-G_EzCAmnN?vv$Ck8TU;Ux
ziJreIBDA(;p45^}Zw2v9;@xtK&Lc@&v8P&WX}vz@vtgZD>Ige9I%Zy<l$AJDl&Azy
z#REAb@@ui0RyRSBxee(Mn-*TvYK?bWTsRV2{jRuJ2Wqk?edn^8ptkjaf&xtUvuV5Q
z%`PT!`U}o-?AJHAG7W>B(sZnqZ-#O9uq2!mZQNdxG33Pj!So3L=GF3ie+VoK($bN0
zc(p%Zd^$gYItD%_rs8JJ7U`_9fE7~)JKh+<$bvYzMwZXXLBU=@6bM**A_hq5EIeZZ
z0pfL;XFJ`dZ3SpUDLk)}kE5d#DVa+g`T}$;1!cYv&T+UbxXa&>y6^^eNfy0vd`Poh
z_}|s+Ql_tm|MjznA1#A9@Ka{`dn*opF<BU&@pgxsBCg%jnj+6mp(0I3<?^}8R+$9#
zxs7l)?8KL$1cK)Zk(Gt|^Tc2Y3bu;vTRpY`sH+w*^zTz3O5<GHdsIR>{#%37ukHy>
z<{am9jK(~vZ?So^j21IBd*LOz*BTwn$BJ>wtwFy;g{-kt*JVFH5lmZT1=9SA*Wl4l
zS1DTng}B_1k8)|8zUmG_=%&+#eFbCQi~&YqRDt`+f}>&AZc%v#{_^d_?>wXcpqA3b
z8)u!Ig4TR`Sx=VFi_sfPl`%mnV1YL%KNhH#5soR#i;+W8r)(@O-+Qa%0n~xS=_e0;
zGU?oGJAe0UYin;sku#IVhyU~aaI(n1?aDUCV#H^1Rz8I5(&ue#sVNbVwPY^qba?CX
zK5nFM3~`9IzMZvST+UaO!xWeP4T6~0nYDJC96Oc5lj?AWhZFzZ7S>lWd>lGBa8FOX
zo<zqyg8<Ge<LPoD|H=J^`4D1Pm7a_-Wsdv7I<C3-c=2?tE;@?V#^dh<Fm$&wP<0QZ
zK|R`^^?{7$=6LZ&Z-`a4%hFFNa4=LrqU;?u5)iPbq@>U!4#B{{kbD)5*}7pdK}kL*
zM!Xm?{s?yH`;mH}AO_xviHU!}r>7Abf-xQTlu`VV!q>f;#zS55<e0&jVcb-PUzg&E
zSj|&A6{c8jP`-f)b%0e0;K>91{iX}G`*GA?o;gRS|9553<TYE$R;Ira>@|jIA0}zY
zxpC~@{fUV8T+W<L%|y5G6ZU%2JouuD1QEXJV7l3@?Gg}mbGsKO@f0yKno~oaS4Twd
z8zB!+xIOkIxpX?!;?8R5FEx0j?-{?o2@|AIV2C<ns?m1dJN*C`YzDt}biA0Hocv-I
zu^aLt=gtNm&0H4acTAOEm3yX@OP)J0{YXLR{nD)r=0wvyoQBY#ae<e{p^j&G{qgzJ
z5q+*UFq~e3Q@B<ubtGU{bUw8U!%QMR?}kU(b7`d0X0V`4G^av}FsmZJdqD}BCI+<S
zbJZ&Xz5X*SaWv%T$kqo`3$jIMh}VgXo9V08VxKJ1(BM#{&`_f05)1XJ($>=Dy0F#<
zI^bdbT6y@>;xx3ihO8(!^e^hWJd>R#3{e%K_vFk6PoAg82m@QAohZp&Rwm0$kqvas
zAQuDuTPq_YHq-Ged+T2jdO*t+wbF{6=Q%znD@5Ye^0EisE1+8=&u%fjX%`pV_uzWx
z=V`nV!IhljBj!l5e2r@lS^Jz}XlQ=@wDOcC<rv${@$U(_I#B-Qb!mkq83JIMzQ1-W
ztzRdla(KU4wSP?7C@ccN^1%>quwgVda_I$^wK;<B4sc%B+uMTzg5DInZq2)*`KC_~
z&*WSUUPSO7UTG(sq0E#6y%c`%mMSbPSdo#T%LLs5FOOKn#BwoPppTduAA`4V>4}N#
zi2x0$HFxly|BaU62|57%8y}cFu*oR}z!MH+mA%0@T}&vnFrGHLO84{gL@<<TpMAOQ
zHFU?Lh(6lq_FjeGg3-EPrw5g9h_3-AiIMK;Yi<)+(aK-3tJ#^qq+ep|U>o1lKP1A4
zpE@7GJ6(@gPOR>~1<w5CospnYx+G={ittpec54?W2FqnQs|6va>ntJvZ#B7L1~v0?
za&qv)AngPp2f~8WFD8Rgi)cJnp!mUYz1ih&NTY$uM8k#)Pz?53S|1)LoY5-#^Jvo>
zi~+UW1eC<fJ|FkU$;nZcstz}%)psf6j;r*I<m5m`jas&YIuQv4Dc3GmF8PBnrrqL#
z(R2lIM<~)ftrb?THIoP04-m$J4dD4~{gZ!RB)Hf6pT^vitm6lN?spREa-c49glMtY
znP6h=@_NgH^Wkc+=1fiQ-rfo?K6LliJ@sa(&T+W#uq;!9ycio=d7qu0r#ARCIr-1w
z-8v9G@<U~afVGaLQRh4Fp!cj_q?Hc_W@ZgBE&UYUrqQvYxXb=QXKM$SrFTcOc8J6I
z9?M6J2JW#A19R199Ja;c*<oLCE1q`o75ng4$dHO0_u~yE`y2XHP)75$A=0h#>GH*A
z1GW$WyED+2f#>}-Wj?WlkzCyVbiIf$2GF*GmM9P(#aDN((7q4|5)x9%52A(!=}7<9
zPnsi1;2}1fyZwmL#pJn{XuGFUTua*~^@0B6HJ0jMPdGrL1(C<BDJvGG+^{lL0m3ci
z8|*wO<N(WS*ca{<edLs`kp4vVvtR9h%Bz~cECM$ho2M;qcS5Bh+!5C=-50NZ2?U64
z6)9~Aptmkd!16S7Wy6d-d*WeZ8|GCy(lXVtdVbd6B>xz|=8g)YM^eOb)3wegp|K3!
z3V?u)koQJ?MNJKX;J+?X2Kl#wlFOf7b^(FaN?%}e>lh(WAoV@61iZfmR%{{%hf4rd
z6cZ5vU7Ov_@~o(+C=BwBxDY0T!747jrg<h%3<+3p?C6CT(Kp4u-J5vd)%)M=_O~lh
z!UF_8*Xj~nD>%Oa#Q=kBL#q<wPjVSI)zvaoQ46VvufAhSjvmaKIcM~-m2ZAfE+)w{
z0l;9;@zQZBMS-BdKqKXqt1`BGTp`5`HXbjisjFMfRc4=t6CefToKFuJ%@K;_D}u7u
z{M;Ov1KHRO^RK9TNOIUQF;M83in7XG+{7m+nV|5lkU=}D!2*pA=XP37qb+#3j1~ey
zn0%DC_SYAp{liOqX2O8^kY&Qi#Nz+>sB$3;rnYQ^_x1F=!NF+)+B{cjeof8A*cub-
zh1!rF{cw${rS9%L%VP*c*c5Q&9G)MtN=j(EdH+AQ-a4$RFX|dSgmejr(kUR_-AYJF
zOLup7N=gbyONVqwNjFG$w{&;cUHE&y=RWVfACG@nXYaH3UVF_s#~gDEs0cg^s3;=$
z`>~>d1A{j~dPd!y$~56|>FMdVw;wd7X}FPI<TTYY(*LmEnUh;iwklFt#h{S+H8g5y
zQjolR%ld<NUjJ%{lfhc9mPp7LD_~^=Y9wE8S-ZT}N39aa&<OQ1p9jclLWB^2a=7Q;
zXORxjFPweYbt3+p0LcHZTi$MMY@iZwG5sop&&e4m*11}fhf7kp*_+%e*GDA@freCf
z2axb2Cf(Dq*id7jJXv(ChC7y51m<*R?NWfwGHBQH^3D$oNY@^+@pw*yKEq&f*J6$3
zyXFw%wl6P=v>Rb7j4C^)!7S+O?<~1Wxe9~e23QP?m!dg?zP2;<E~DGe=b~wMa{Nr|
zS|L6@zQG5-+UUd+vOW9rvzZxgklq_(C3lsu2j9)vCFRkJ9w!H*&O<Yu5_L;3!LHcC
z^Deu`ZwB*HEb&un^X*_*S^**?qq;&d!13^Ae#T+k7zQ#@Ww4yqiySYsK_etTxH(M3
zZax9n6NkS)!)Wksk03&Hy=TUqe6|&Dya15$Qn+p95DNV;BE(1g-sT#ls-JQ(Oy6-i
zAf*X5c$|J1WyngxI~ZDMv{K{VnR7W4p|}w=F=0S8qjlwcVgph&&%*sApQ~72f<>nA
zBvPDF^bvm!eUkwe!sX$f8tVln9whmmr|`}`qOn<_ef*k=b_n!+D<4hXMPU|ovA~w%
zXu$nex13!uuQ2j0>caDHUR)W|2&=!+aG+8|g#Ah#r#|C&6L%sO5$?wV#)qN(GmI=o
zOz(%&=T5FwUa673uoWY#G=IHfZmefuaOO-K`Pz<>lJeO4I~3^z7z0es^HHzHBUncL
z_>DqT9+c0J;~l{UFm?NEn3mz3`r1Sm$49bgQp6mtlRr2~?^_bHUV8wTkywoFtzBOy
zjK$8U#)GIz22%3EW82Rh_Z+2gAcdAV7iiX<f|hKxx)rYTQZ6-8sJ7@YffKvS^Bu))
zP6SA6?~SQ?_h(5?Ca~90j<T+POMM_;@OC>Nzn3RhE^5~hk0ioBI<gEuJvRbL9O@XF
zFa{q{S(O?#fH=|x2$42I@_MMc11B9#`jHgt8UYpcw<?MMeajbx7#v3JV%6q*IZ1OD
zsvf<)Y4}e+P-MQWoU{P15)7n3(7{fzKU(^Wx+hHLci*U!nhH!`D&Eq=VC>qw1x~Pw
z@EYE)z}PIe&m$e184}ea0DL~4q}PJT#)`-O<s{a({{_`nQL~dqlk0=riSyGFp68M+
zZwdy{R_jozo12@U&FuY&c9sz(Hj?<%v3l(zdigK&%kz&{{2r1X<e&@deS8eHJqv%h
zd^b7Y4~F?oqA7rgQIc9E5^3kwSrRUO%HX5<(dcWv3f20Up;v?wh^llrA7chQ-@g=r
z_VZ1uV7}BJkDz2_t+7e)eox~#H6@!mo}igjthL}I)#~@eUe8lFnEPn*dcvr)akPvz
z=M>C<x`)N9*L_}$iJO=kgNEo=+IRnGI7O2CJEhgCFT8LTEq~R~-X5wQ6C3LQ@(*C1
z3t|;Ef4)KiK-2Y%jl0)QzD$++N4G#hEEjAKryqfSFvH0NO`JKvsKC7ez1>xiKnQF{
zaQQrXmaS2(PLm3a3t>Q7Me<7ujz3Gu`;JiGIlg+0k53ydVLn~@aFQHN#wPQecb;za
zr=_K3i$%~_6$ZDhlE(eg1Ev^63h(gwauncPCA!|6YvqoWGvA!;11W=iD+e#BYxdbQ
zaXjQ$Vk2O)DGPe}9$=6>`^$oD=~Ua82<q`xllW`xmV&uolat$2Ho>w1B6atI^O;|}
z{({+T?-9J<4zszccaVh^AL4+cJfCC)L3ziNj#cR&Tos*Cg7PJGn%OwhbHZOSNDQGz
zh9&QdX4;d^;zGl>6(T;_4EbSzY#K=1KQzlqfUWus%ua)3m%~8m(?PhB{d~aSZx+vP
zny?{cDrIuvw{c`xu}AgZ@tO?8T(&C_m#N|@qb8n(6|YWV1MSyJ?haw#4sLe<3u*DU
zU3Z_6TtEEKkKG2#f=7B_V^XDFce=j{;~TQ_`NZ4XTL*gvE6`rXc!>F<Ur?SOMUm~y
zdwS8P51S(X))Jv0S<-`(uTiG4dG|>+oeCjkx#|lm8aZkKokor2PGFdvoG#A9{-;mV
zrDk6t$8=PT)k5{ho2fD;jM(%{UqeFye+(G`p}e;CDT}pmau;A*BW#2O1e$*rqJgyq
zu+ux!&nzy=?f_j$$4Pw~oW9Q{#ya0|c9-hhp&t=w?KBig)%vh8vj5uGr5~*3Q@cv!
zTE3a@j|f)kgF`R)4n?HE3@0YtuA2e$Z5X^B;Nr^)AOFriZu1hEeN_+G%-iZC78H1(
zSsnHd4rMgC0Po<*$-54<8B1pXUJ2N>0$M6VBpx!^$!ah<O^Mb#klKpFovk&phX_GR
zq-DG-aqf_>jkwss6AAK?;aa6u)7K|cayt4R-*}^3^)4Z{VCp==ZB&{e6>H(QN)>Pg
zEHj@fzIoNom6VE1#tad1DkF@HEYhZef*@xyZ02Q`(DT<F@rp9&G!wjh2{a_cBk(r%
z_wQfngWSA>>0$NcF^1iI#`bTe&lO*zapC9(6JwjEyO6v4n}Ay<$cmVKVM<DhKm^E4
z5V0s#YmY|2ivkkGIMQ(Wi@o-elnfP^^aKloe$o*}va^&!hP~ANFcW-W@3%NzBFJJo
zGetQF^auI}E^NS)RmC+;w|s-)oEd<0P_@33`{&x`rz7v-Dm9b>4Xvnl;*km>pi(xS
zSJvK8JsMT^$)=(akz*ijbC!c-@3yM)S?M-TQ@8cocpdk|LLHcy)xqwbh2Du<%FK|1
zhNR@(seKh!`${nTb|A6N?Tiw%3q`Onk~P^YB|d$st*pGua(G@~10ObemcQ0e>$ShW
z*lhh?m*sCF#%!!n{HX)6Cpe^$;Bombq*$6><|x8Vj7)#K7rE^8R0}$V&1}9xJc>v~
zMa7*1s9&I$yoUfk&dM7drtt&yOiMlYDJ>v$$%POYN}s1>&+@rQl6R@UAGziL_9-Ys
zPRL$CUKyL*?Q|VcR36aaau1ZAA95Ue>XU6xmU2(p`wLm9`nn>=p+&3yN4^^_LW2AE
zK2J?rL%S#Q=}T01L&eVe%{tO2KY8lC>~KfQR?iT{&qoEzD7?O_lg1{e-om-U!dJeL
zm+g`Z>4ZIOTa;i>teh#bS@`w$<ZhyHX}sS+cD)3rBWSWvc`r;I84Yczh6?%?)pvV)
z$(kYE2dWh&NncoOQ@H>I!UcOPYREh#X*q0+2sBd0up+_G3Z3Wu_1|`GgiJf+%>=!4
zd;U>25RSIJupz>ZDt&-QW152nuIS;<c2$U<7N?T>FKg@EME3XM6u;S_CgWyXEZ0kC
z8t2~ITedS|UBMF(wfuT&v&+&tQ~y;K{<u=0(}{BLaD0XsOh?5}FLSn<9pGm4N=C(M
zX<+4L0j6xpyuFl2C9Mn5?C_@m2*h+}t;yn2a2lH4n%PqITk2>G0n08_Oh2^+zFW`X
zG*PGAu)fA`B^spyo4lOJVwD3a4U5J43unO4^LrKxdXK~SxS|#zh6Rulj1RiHBp+i=
zBqzU|53FInLlq*MSZue9i7?8Ifs>PDdp7R#hxaiHxP6WcuCBl9>TZBtmm$ln08(3s
zJjB*;KA!aI2p$d&Z!HHbFhM4sv+KjVHoaB;-P1OklN?~31XGSb&|-%(T8C(dD5;*3
zqIK;3OB?Z%R|=x|161Xn&voLzgitIgQ9|m7s9wwLxVSWH{z=1@rP(>XkOGqG@^G}(
zyk=a@e&e$CW8BW~p$MOLO|Et);}rr+p<6n)4=EcPNJwP_oCn(5^>_R_(r4=Luz-Ld
z_>ts_R$m0$y@NH8ki6f#%gnvN1u=a|o*C@O(t=}hL>zCzOMZQ3X67aitDs&*Qcfzn
z`)go~1ar5@-eJt_fRBjMQEo$^rAoqDTLDr~5-fOl2j~o@-Qnhr$IS`!!aXx%{nK<c
z{B+mQkBKVu@seKe3ih{z$g=9G1X(^JD!+iVGP5dnQ%D-B^v=?;uFf~olas#(oIILR
z+}(57;8^wS%M6@BpNkW3`nX4nEz2DB-De4i&jm>L3&_@szo!nUM&uwPe?DLs+i2A0
zc(G%MbmLRs1OQuh;V_lfHNU3!Ie{~CbM@Iu_Lf^5Bx+WQG4~G-h+p#aX9U7X5WckC
zm9m@>qtYDe-_ZqN#2So0Feo{ayd8PSX6CL5jm&}#dN8~@MJH>|tf}?f4CHds`c%Wh
zajHT&sp@3XBc^|z+5)eh*^W#;ozCcdPL-yQou?d@o2x({g$@ZB-s;N<WJACpX%^>C
zftgpKTU$5Zji~VUSOUDxCeIt7AM+PTEWmMdiu~#9TnE_#`GY5q$8RX1r!^DmP|UEd
z+$IWOYst-B2dW1hb)^vz3TP@5K&KJTu0>B3Z20XICKMdAR}1BvrbUMiXinvNFu@(T
ztj5we79AX>BanDHKz|Zu@bBaVqyPXufaWI9?gL!O`em_zZ-IfL+@H>~g%OgHlT{qC
zb7?eumOlGp=<o3lUB{i=3@m>6RQr*cmHAA6e8VL775bBc(KgB^oJ{i86a&SI3QW(w
zQp8LWyya26jLzV-(Xm%g<GRSZE4PVF+9vo!M168&>yz}WQN@-4E92hZ*orLy{_z_h
zKO8F4cdFZL#H+rUsc5iD@sh_f$8A=p!$kTR2n{g})a-qY1cu!<Py7rhTNns6{8OFG
zmxy1pm;|em`(`!6cUW7YLAdR^(x`?n1-%|g=+}TO>cyT9TOKR~;`4AZUDf}&b&#<8
z)3wW1B>E+L3((yG5fRvy0O9hhwq+c7`!2Q^WSJBg2=#jF8VqE8GcPyS$iyVUm4TlV
zGW6%qF+K-0r8IP|z1Q%PP2{825V|lrdekf7h^XQ8!L^fa2DKjv3GfiGT|+YN@NBHK
zIb1EySXwnw5>S`s8c}GxEgZ5=hk1#rTu;~#e-!cdr8RFNSrykV6c7`pymLIWXYY;j
z`cBH?V&!?P-|>27bNggfF)zHI!`E$6A5on1tkVSd`%gDbf{nS4E|niZEkutB5yEdG
zuLqYRcr>>Wli*%%;qg`B;_R68&UzWf7qoKB*pmN>e##>hfbLg;hPb+me)#a5^yNnb
zZu-^YE2z|l_Be}J1AR)F<Qxw^sqEDBv^pxzbm$e5b1b>C&vZ@lgIF{B#ucLxc@Nq{
zE5v-R5ScHnQ)rNk_+fAeFE6kDS!2_GKiv2?$-wVpqdx)Igr)$wqR|Rn5VLTne-%c(
zh4{|&U_l@Ok?eas&=B8h6MhQ_L@MewK9{eHZ=(pu!c{``i0UaqBcAR>dqsb_WWc3b
zXVuxEL%mCVwb^JpRzULJyh(r^)v`U;58!?^tHUi$M)B9KshAF^Dj5$ZJ4u%K?Gvb0
zUS6$ib~zw}Q&p)S01}RYzCKX72h#Yq>wjJX<4I}h*73czFHDe;k@L%i`>q|e8+Cd}
z6*Fs+g~?jGo`v&4qwVS!^%84tJI$L0oJAlM{Fy7j5fb`k>l5wgQq3W_7H)tAO?q5@
zi7z2tyY<Z$|6HspKl<`HS%B!2pS=F5aGW6kfe<DT?U^>ZLb!%X#D`6_&aJ8iqza)R
zC~2c3i6&lAd1e47=@{0ovMY%xlIt|ykf4ZB<Re}G*&eFwI}%e}q|WEA1R4@Isk}p~
zjgZ3nj*CB<Ej;~`*j1*L8-G@-0|PlYIKF*HnB>nCYS*u@?q82p95cX{gxGG`<IK&o
zvAgkZ`FH780~SZ@OD`xu!kHvJ=$0m09+=2^D$h($^wLLRlj+~E_Y*p1|8(mIKsKT2
z!;#;J5X%%++Tx51{k1lb_Bo!*aOIy@I=~IbMK=D<x;Zc#G&V$CDz8s}?eJ3~E>0v9
z0TtD$UsCE%0vnIlENB?6rg>3(6wyXgDuu2ucymZ(v(#9i)o=;6G>t0L=<nZ^T*j^e
zyP)ly3E~UoKXiZ60+vk68~)oL3X8ryqKZFVTyKKlFfcn|E&4O#$<XvV1(Kg9P!1R8
zT@MmAfoYxKpbi8F=K%H|C`>9%Sf%xEh!_ve+h8Mg%QAH`q1=sbC)rY$4W!$W>&`dp
zqLt$XAEdD5t#zih^#t=UyA0~h0-LV#*ZA($2++QlcTqs3OvR<=c5>3FKRSCPgY@`?
z?W1LfH?LDqLtWhtQXOKT2tX3?HC!*NB;#lRPV*9E^nC-`-{MlTC!vL2JB)KXb6oBp
zIWyj*<Io4?()tYwxUJ0r%N4Om0yR=S$K8pY?N+}n?e6y~Q;pwh_=uvCuvh`?>CNU|
zs_@SBMt#IJq%0A{Jv$U@ErY3QPW9_U%*gh35tGMxEq-Zk%WCx!DK1dxF96^Oum;a=
z?gjRyy)=sr!3e>r+4#_=6wmy8gjO@3Y(et380ShkG)O7I5N7a|0rfIH4!+|eVx(-6
zi;apBHXjiqA0HnB!%%zsv$+G_pm@r<K3KZ1gqM|RpoA9~DnE0okBm<zOE%UUck!ox
z)LvdC-UFB)-hEFpLEvAf=FVb0-4i>%XGfh<tzAl7Fqu2`%3oRjJh}MC)7vJMQ8xHY
z=BdBTQ}LSq)^+4@k{}QE;^X~^8^voeMYl*ciF8#|J00(TmswA>k9dU5(Ev?OijXyz
zaP%GkqTK{W^M}Gub6KxMz!#eAIk<~u_mE#U{#1#ekzPZ7RlKX))@jtd@jWqd8)xCv
zRaiJ7Gm~7S`gSZ~IkvYaFGo-@=Zn?h|KS2qLuD`oH`T{v!w(;O%oXdjJi5v;O_aD4
zWcXIwt{yHnpn8Y0sQ$Lid>YLLXi&*(#?$lj;%(dKg2(vAv;E|JVrZIyG)^bgO>cME
z>vzkc@Wq2-t))9#)l+GZ5^!|;?c6!SErZuBAA0<7=M*Lmj#+EBCKf{hdMwX_0b|;<
zKeu1(B&9f<R<T)yep?*vxB`0;%NS#crrt48$oKEqrXz14trO=zKA6vR_lJ{&0KFF_
zM{Mv=-n{n`SpD<KJBdssbPQ9wg@J=WHgTRNT%oB1Iu-C`zO<zO%$*rmXuSLLF-|fL
zP$)Sp(WFxm)eka1#;#o*(b@&CSOos6=ru#6VTV+)vc9rZ_hBMI;BoeY?geEn5Qfv6
z;Q|09afKPo`95kC5n7qzE;agdvBBK4g@yJctoKOb5g`J1RjARct2~yN4_*N9zMQ(0
zE1Fi-A-xUmaJvV-zMQZQ4q!mKWoAaz$LHxU<GY;vx*fIoUk+vvTAaF>rJC5fo!0Nz
z<rPp6CU(UFCAg?cz}g9}TL2*nDA(PekAH%*A$~41QDld?36##enG<exR7VrP;s$3H
zk^nNd!*v`jAv(eM*u-`99)<`J5mEW$CjelY;$;yxe8SVY*W^Sp%8pfVBE(;$8^D>J
z4^^Xp0RPjq^bIM}9GHdQQ;lv8?*l(Krh-rh+)_J92IFHHXu5H-CJMHNB)gAiXm|n?
zm^nCtS~8_0OS0Zm`{@bt(;S-Qyu*w8b%K@Z=Gd)LRK5%|lkPF1iy^be%0d>!%jeM9
zPMzRf(rER90{D)D!ivMru#*7o{p9rO*$qwHuD=yN=-e0LW8nfON|DRB#0EAy@4JiP
zb7`d3ph0!z*@EnCt2(8xA6P)KcN{F=OfUs(s(75)25O24(6*#_fw<wv4<7=6w%KG+
zLPTj#7XkpsfW|X5wwQSfANakf{6MkVTMPtjq!+9Pv>jcSPUW+fUr4@78i~RyKy##&
zjGCU`aXrihh-6Ar8W>xKH_bS*SE9Va7ulA&5}~t;la!R~15<}Dph4e*Uy5M)NjENZ
zcOm~U%A1$3H@j1WVt0SqY`<^Fc)JoEFEovxAlb7(R5w@%qGlbB0)Z6dm>YB|;@H<M
zkQ^Ml@&RdHm8UIu3t;Sd$M>EA-Z$93>rc)ES1uB;&P`#kD}NPhD31d8qN~7Rpsa(a
zn_mz}Is#A)6%$_=qh67Mk`mZh5BK-)jnfKFnmYknQE8PW48hWQ>Ed5>LN*02bB2U4
z43-HET~+B)oc^t|SN|@&e&$U`mp*uY-6~;5_Rjq?VKUQAdigUFF?92r{rg)L#|(<V
zunT@`@L+Q_Z^Yb1x}FtC|9x@2;}jg+FfaV=b${B?Q2L?E?9p)c<3Q?IdUAN`s+ZO1
zsll)j4r<VF%E<TMm($iovmF#t34(e?3k{5R`z#)MjQyBI`&xv5(TCF;uHve7l;hQ=
zZ#pdE>=aS>b&S<H9fQ*8Hcp!I6Y6q6S>|qy(_J##CxMr&os{&@GC+VT6Q;JV;4v?|
znMX`;c6(v5J^LEcx=8NT;`h3E>9<|>_}{l9Y;P`!HJ7V_pW~tw?<uIYKL02Z5;B1J
ze*dmfYh8VNs@x2A3)L5Y33*k&+a5#m;3uZNwBR9;#0)&)zkZFPP?N51&vBX#XOE@^
zt|e<1d)sMzIzwhXoInUu)pgH5^5sX2@R5x4M*7aPw|N-+P$nK8iaTxn5K$T{<!-Av
zkh3V;PflqvWH!ag?y8;SL*R9WtvnEvF0$EFh__-;;?1sCUNgVB7N()jY_~nAs$DSA
z5vy?2zYZB`SNh#IIuw6#sly0#JPsG?03Y=kiUM2lYzw+b1DfqlJ5~H>{2<8J^>(G1
z*aLCs(HWEPz6NxRjbSRpEPKcPFvJ?`E{_J;q+WMkbv2_T7ZNfuvVtOkj0|MO$hz5J
zvas0Yilk5Eyg<eDm-AlDkYxspJLmH0binGn;q237ny{c5F7E^N`RFequcL1tY>iIf
z<Iru2k{R;b*`2JH7jAU9y$SfS&}Ng2iqsqq#R7RSAtVi){HFU;zc*v0h~C_y1#WE)
zOAJ!yP(uk1R8}6+J`R_fJTv?zO38oMrwV$w;|SAL)izb0paT+=u)i2C2TF3DFax}z
zqFhlZ&$y(ajg4T7+c<7)H%u6Njqt;#{pe=|lR38dC?l)vW$UinSsEc<;gkY!c;1;V
zOMk?Yo+04yz1?h#_PMjv=A65knn@?Z_K8;(p_JPf9-B}~N`3EF$wQ@~kX9h%KBW+C
zWSrtU{maWE49j)ugt{?Fv<>~6?Ye2TLF~Bf#~g^9c<cO%;ZfhJr|m#y4z{fm3<Loo
ze5u|H7^wo|Ob%BH&<x<ge8R*ThCfO^KHoVuQxx#J2YDM9cas=r+sgI@DNDJ{5a@7D
zD&}Fp3dguFdcI6Zed%^BMu<cQ@x_=&Uee@m+)(ZtZ1`G;f8u^r&}zn4elu5)y)|XP
zA4;~^U*RpftWFaFc7h*#!!B)%&GCJoUq98Z_&sGMFKetGfw}=u|Cig^R>5%W>WM1^
z<6T^YgshCBg(^Zs1?H!!3PaL!K)nMgqZCC9C<Fk91r>6u<P_oRO6=*mdM${G$Ht`K
z3M#(Wbm~ZSIE|jTKVq=oyt$rDSoymTR?u=R<E@eje2^%@T6NxmMW`#}!-XBZi+GP=
zvPU>U98nyDJ(Ct!V1KpTaQeqn2tp4Lau4!vj%oq#Z*RlS-taZNH>zlT7(8-|G&(XJ
zsf`c4+xBX3MUI!*eHA}kac0hQGjcvfESqp&1N7Q##uMdPVet8yPdxAa;fnz8{73YI
zxG~3zl>t2)+h13&kD+|?c6wu}#2-0`KKyl{O;9dWDgb3Ym<Fv>t@zuO6bnSV!6Y85
zLI473hUv;_>tuwo0=lZ(2%8t6<Nz^UV14$?aF>43!L*+up4f0Z`R<Z`d1-!_2;>yQ
z&=9&JD9A)UXm9iN;2lCp3Qtb_@hpDh)U4Y;+j&tjyc3Ue^*YkPwvsn#9pi_AKp><s
z7flyq>D&&+lG*OsD*BAHCpEOSb1j}{LZuL2uustvSPm{<=z#fKF(m2MK0HW<TNpe#
z(L~Yb;fV=0I=X@K6Y%YTF949V1C^O_nKmIL2%Xs5v0FM{gP3TEn9nRr@6+p-FPRS?
z$+z#Z7V49Jxt@Sk1+@Y?jzH?Rs-gneDrdJT5`ZjhKc&@vr^Zmc_Ov!zr%=R%2yfGh
zYW+-x{=iaH1XMF)Q$y8lt@NYPnnGHXfXOk6I}0hFo?D$Oeu2J@L@HeDTs6PHud8iY
zB~5`TEXK3G2@!3JlL88&Vxd&PB6ztvrBa#>@6+Q02!fE6t*t9MvU}5cm5TXVFxiKD
z=?M7gzReG<V@M)AK%fuO3ihu1Ae7WzPC~ppPvEN-#7_wFrOWJ5##JGGn_(J(HCalD
zoI1C$;`|8ylt?aDqd<e}3Bs>%%hEwZK6VG|f?L^Rp~fo;W-gCQ4dN!t6?eGZGfZV*
zDEe~=VWZGYi?u<7keq)H|CUN!?Vo%94h)fK3l*2SirmSV8w-u<e4Uo&m-<i;Ri7u)
z^T!vSd#xeMz^WSsf)q$bOWQv@eB|G>E<n%7XagK10Z}(3D(b0I4<JSuXRr_w_Q{C}
zYzUyxLA9L2wA7Cj^gmXi7$3fU-i*9x&-YjVG`YPkQ(MLv;)9UTnP2Yc+OJDw*@U$0
zZs5b%#7Vp*)*?(Jv6s`iUQVw?ax8d(Ij39#rsb;5i8$QkD=@Rq4=QK(!8{=y0XH<%
zc{!8dhWc!(#%Z}cX#0YZOwd{w1xl(0A5P;{kn6Ra#uWfuK+NyHzEI)Nx?3$v$op`2
z)Z$(vJbwO-^!;X%dU4_5qV9S=7|R@sB;d5xoc<%l<YX_=>AdI~nw$GJCl?wbhmDQB
zlFRTM1KqY8By)vHT)KX+qZ+%Eyi;kZYqs0)S8q8PItXF2VwwU30h)_T@-q%ayjwd0
zZB1UhB5S#H7+-1H$@3D}s{PTLB2*tUGL9v01FL~S2IUe3{~jPObnfIj_q|k+@=LXo
zR5ITARywGc1Ip`kjnQK6f#|Z;V!d*@7$z1LV^UZ~njl2zE)vjD=;=QKsc6vXTI<%2
za@?CbXb*WQezfSfurNp-?BhO<4%Roy9h=wZ0#_NXDz~$&jqC>geo9NPxriqZ+5Ul$
zfTX8AE0dW9!PHn8jhk{P1SyNydre%_A1+iG3PY*ma1jjZF_oY;1`^T&>B~Pq5&Qp0
zAweLD`rI;;ehT0N9@`a&Fb05>fstOd@Q?yton|vwXg_}Z0Cfs456|Y-*6{9#05x?!
z5W9tBjG7A7zG@K?lK(D8J@Io8B>TelEF)|ruqP+V_~#R6fe^rHzS^ISO;2aUKoHQb
zvIhcrf{)k!c{`wJLBVDD@MPRV8AAbd@>wlZNCU1t1y~IL!wU3r)#z}9F~TE&NWO7%
zW__RDKzKyO>y*RyQArhE4d;jHBv$u47x;8YRS1>`Vy2?bBJn*KFlFCe_a1CsXf@7#
zquD%Plcb+N3%4JN;q$t6+|KM^a>9|Ds~rECbA6|ws`}>0F*7rh%5(3_jJ<K00Dn>4
zQFR5p3wk2++3l$jc;CUlKG7w@3(KtbZ8UcTZO`V(Q7<r?e`rB8@a;&N;bNku-UfFC
zSnPlb0T?9QR_R(luF~N^zDed1ZVmYdO8Is1)OvnxX=(ZSbL;heK3H!vQR2%1?@w&a
z<_}kCpb9I0iIuPua@3dbSO1ErUhU!dHWTw_Xv!DAuQ&c{VIp>_@Ws0NFOPZ2E@y%r
zP$(XTVfJ6gLj`xpQ)7!x>St$V0ktm#Xawj&<*0~=66K<JVcGKvAU}kKEo3VNl5NMM
z-ZdsiY)ETHCh&YdJ3dB0M65Hmc{*P0;6dI5r4%*}&hD%|c=9as)#G0LfIR#3HrF?d
zmQL$ZcMr$JWb2nG=+`;e79rJJqc8IS7HAxeGL43!Sl7CF1rWVvNML2<$dt=}dU45q
za^nd~jhUG@8ual!YrTdn#@*c@uO`7D2>|uO#~4q6WfB>$@pFR2gmeuoZKuIFnWn;T
zIrA(2Pv@?`P4AEmJXJ6+ADxgG2%<up-1+9-wWhq^5{L+w%oe&kmCt$~VQ<NGx^aJh
zzp`uer`PYhVMrwKm7qQ;2fbJz+;an>{04VIQWCDAaWnl`Yz&I|tL@XYbwhl7{6*&H
z))YvE&GbA$4fNDthb$HzR71*Sd+Y}T-$zgan!9{HfMN120oWAm!FfGb^!r%nnz>}d
zLmRx@h)b88wgg&)$gGtSF^pJ|gkny$GTxbpZ<sUCh=s^wu?x-QDi_r`obC#9HTDk<
zhKEw-4Rl<$ldx(zipQ;-AN>Fw_JYsJnMVKSiGz}crf^^jG=lfw&i%g{=uK70skQmN
zYw0+`BC#~S<K+~2|Gxj^ePb{Nbi}QdEO~-Mol*Lm{#DyU>)XnwKvw$cGPSAdFn@o4
zVKRVVbqM(7?d0d@gCWq<+V3?Lirb{Q=gU4rO`gjx-x{(#(!GA6qfIV6EPr488uLV?
z#sOfY3R4{YN$*Q@T8)p9gQ;ui(?eg;)Iax2H0Y3@75W_%Hs5vE%Z9K+d^KY8y15sB
z0;7#mjoL#j_~JS*cpOG}h^;N~uFr^y`jj+02_~LEZe_-X7cc^%B2_hlHL%kLAdtY5
z<O4|nj`JQ5y(J~rXCz6<?MgQXH}xWgTqI$Ptq<{Af3?$g7TiER2fg@11>EuhgbDKt
zz?6_p4ilopH?*oV9nBhUUhk_RukqsjY+{y1xEQD2lJbxZLzSI772$o$Ep+vnOokTO
zNE1>|ve|j)@O4T+5@{dGbf<FsZi#628xL{w(VxnF9{;A_0BQI1ayzOJN1Bdr5;0Hg
zC4TVIChMznDH|!VT8?BN5a8?=8D>VxI?V`4P5o8Yq9QoY;&E35R}{%;{^f`fk}>&3
z3>ES%PVMh~AEQVNMM6Iz=uQ5v8Zat7Ea}+%ATEwiA_0NS0KBitpvQ06;(WIYz<Pb$
zP#m<h7JDfaZv>FsFSvDnE`gPZ`icr90(l1b4+FQ0PXuXAeB5SuD7a__RF;S1%k?vg
z5ju?wCmV$xMYYS*zPE={)$%d@+3<v`w=e3uf*xwfm20{Dhgur8{wON9Z#Mf!{c2kH
z<!sfQ1^tZx_vZ1BV~V?9THJ0;LE7)5KZ;I6715a~$VHPZ4NUdJ^{??}{jJDN(phb`
zsh0U1PxuE-d<gja$EB=~_-OBqdJiA#j|!%#YEpxstbX{MH{o8&{upuSqbPSVRZ(R!
z9nU)jQ{3iPRJpCk%Ydv055}>^=I0yk2Nkf{-RGTEn~^Z;?N06i%dKC{Fp>lzV>zNg
z9$Tb)9v>v0D70a%k!k(1qO>7cqs@YYceAy!)%cvm{L%x1EQ5?3<X?uGTtg3^wtV9`
zl;XTSjG+>p)LVYqk4Dq6B9ksJ^!<Q=6l(Nf&CAZu%!OGdI_C`EoTtxzSR6S?#Shiy
zz+#*>i0=Y;)L_RT7KfIX>nYCD?c3?vSn7!~hUKgINpva#;EvrJfp21DglkR&^D%}b
z!CwrOj-&A-;CC<{NQ7p^Gicqew85~lR)RKLU46ZG)@<#(>-HFQv}(-fB`1FU3RhF*
z+6aqBc$_SiUK{l&T;(CS=Yj`);l%Z+tyH0)JBrw+s1P2zZ|<EDU}@94hidhugBd`=
zg()JzoUa8B<KYPw^-q`e{WIwr!4pkM$LZM^V-Lg_!5v~5UiEn{@5GyV@gOf@W#rjN
z{CuNAi5`9oQxOEn3>vpp-OGMaJHtU;25u?Ehgd1MtEZU|2mxA8c_k&fbjq05DJ%wp
z&#&4Cw~1W^bY5HCKuyHTO<`lP)$<*X%l#Y2Zx2QFR#7uA;%GXS;dyP|O~az40-EF1
z*9hQL0VJVfdIf%%1b`ebPSHrVqArHua@EO4v%xuHc<b)&Zf~-T&UCEJd-qRDDq3K4
zT`<KIdlTWS+RuxU<r1;Y%g0D+Y!*`Isf4Rb{yZ2jOHw5^^6sxBU}JJ#@5DiT*Ot4z
zPhv_Q{XbA~N~xFDQjxXn_Xyuw35xgHp8~VVMIO|_%BG^My{Ns-Tjhj^@7;2Dwjm$L
zfSkj8FsR;@k%v;@L+5zyL~L!#h3B5gXyOW8Z(^c87Y!$~TfFxd%2&dL=<cqE8;A>d
zJ&dw>fvn^F{2Xwb=H}+`Xk;0{M4J6u%&(bvY$TLN9e}!WcJ@pxGpH3CIp~lFJ_4hD
z-+g=UY*p3OH#av~SUI}2E#a1f-&0_EjY{=7>)idW`pIX0q|^IIK3|7%Gb-LbOz{;)
zgbypK=@>}&T^FwW#2NE{Jy`E@N*&JKe7BFk*X{iSyRznhFdawGUe#i-qJ>P2L4Ry-
zOLh{h$wO|uTt}<fBLCXD`_<<0AJ$qH-|A{F7t42v6fRib`K2MkOW4rwF4+vGO25UR
z$Wk)b8{hc+NVMM<Li);|iU3sNyXCadK~c~VaNg<P9QRG#aH8=#ds^bbOy!}+J{Jlu
zoShw^>_fJR(<}cDMDf{`fNT`oc6km!1M)wjT-C+q4;D2vc|xewA$nl*nUI()Jf4cq
z?&-esX~E(G{?Lf*ji7>{vn~A|9WB2p!yWMxSg~20odRhqg*y8UyZ(4(M=p@yyt#Y_
zf=l%Oy+J$qAn?-;78bTOf4OxL#@_zRsXfRnX?U_?E`V-S)o$7I0lKMg<A2)BB;LDS
zPSyBK)YLV<y-(i1A&&?|R8$ecGmml)z)0vfs!#=}P?TnkA)o*Od%ccYYoqpf6@Yan
zX`JkZ8nSS3fLvrC%`gBXs^>o#bH<Y6Iio}+^%Bo-VOi|z9+*k=tnd@IVSqE+()f90
z6S6xPm>NyF=)t3v#N)X^9(Gyz$Voa9gZmT&T6l0~J)pw^7=3K%%T4QhSO~BKnuzTi
z_S^($tih8>e9x=nBL(KaoF}4ZtwNO!j8H$Xqn@<_B^qPnZ~7HnjEo?C1@uJLYl29|
zMT-;hv<|4&0&Ko~bcj$-ZVwM0W7;n!%gv#I78u_Gznv_7OG_R=aO{0Uq#+tCM*~c|
zjyb>gRwz3nyk**_(!ib5bKRBWxRt%n85<rte8aUGjx-TH4(%Vb9}<nv&hT6=bTpH@
zQxf#io>46zsIv1X!t8hJ-BCtcP##L_>%%{nqCt8nIVNtmUlPh*M9klWs)2o^XJv|g
z?C%nj-8F8JDiDt4ay`}s7Am6fU!EbTFmFV7%onhBR|DE!0vvC7X?I3fXK4>_*iLQ*
zraSeGYCsh~h0lTZNw#R#M+lhihz9#Ve=a2LBaE8YLbIS+<?9QpOSZ7`3o~B(;X*VV
z1SN%cEX;a7j~6phx=TrwfRPXJ6|W8n33-8xqvZJEd0i-OFG>l|!AC4&vt0^Cy&Gm!
zrvnr|Y!iYv5QrR`X|+&{oLp<bpwx5Xw%OW2me^)V%%1gh{%+;(7#FwQND%5PLbT`g
zMZ&e(ib4rSFq-Yw*4GVz932j_jygESjK@F&ZT|me1Le#QD5ba?on02~DB#vUmKpFU
zZz$V_tD1o%8~o7lAc50`+a;=;#KkZ1J2s51s8xE~q&XZLHd=4cYuWPujhSQ|Hy%tM
zL5X2Kw%2@qBzN6R^s5;PI+8RVJl`Z#7h}`E=S=V?p3*F9I=Jl3UHn8cpKzk2<r4ev
z`Hj3rDaaoq(rU!Bc{wu|F%r5zpa!9fTG5Tyvw5mgD%?1Mf}GrFCRKfdgMwD4lnJNZ
zXEM=k9dA+D#~vm;?1%9M01@y*{Q4magdsqQ0b5#-dEZK+nxgzx!@}kA%m*_0=jBDI
zW2wnd3tAaTa@57!7(f{}jP|+7aih<wWH~@Letof%VoOcLU^UOL`BunDWh%cquS2Z1
z`lm`wMd@5)lHZ!J2x);F1XA*i_C^E>8X^RZ^^?cjdrD-i(XV6Ynyzp9D@3))9dsut
z5y&_>r%ASerBiIL;Xv^-Fu!3>8DF+)%SC_|ob-Z^n;=pwThBOJ8hF(fDD@3Qk%B!K
zSkHIq;}-h<)U&J8jFapd;4vhNrWO=TPijE`jS5gciu{p_vqb!&d#Y2Z6Oo1(e7}%<
zTxd}m9<lrMdWYkBMX;_U$l7PrKBl2foD<=Hvm_F<Fr9~t%@Aj+YK_$OYyJ$0a7Isb
zcZ*#m|B5@Ie+fD#jqoFEg9EM9q5Afg;a=~dgg`n4f&(Z}V}6E5$hb5q|1a7PwkT;p
zBJO;x)H7$rWa~8b2P`!b1i}?r8X4&F>m8G4yI;VC_v!L)qh19Pf(DeL52op&(!56k
zHgZusd(*>V&~p@~y05Vew=17}u7KTt`<CEOmRgAN-E2{b<y&qJmq<ZPUMe0_E&4S}
zran7ses{6GESsVWH3d{i#^UdCXEAQ#f3LXM&n=o)|M$GMwzT{CoskxFFH;c4X?{r9
zKeXwTd;TkwJMEn&^x*xQ6I7ob(+Rr<@xZEWHt#FeK<1A2=l`(pk!8}JS;t$}bDkpU
zyt)&pXXj_ouu+!SNEPuFx%J;M3`ia@(Rc)Tx~rzo50$eR1&;}tvL1%DEPi%28vgZu
z^?YGL%-_JC!s1Xy!{6fZbgN92s=+~uB}?35^4Bz~_D#>f@5ktlR!|O;S8A5-%{6BH
zw_V`xpX#F1JNkW{Kr<&*6k{;B%@1-L;OPUp|1>-}|1N-NyG~Lx{y1jYY0|HP58|ud
z&sfcFEd6Q^48x2Qi?xMwlf<T`-g?aKC{ldAvwuEOvF7=HZblYoq?*@NuxcUzX<(f8
z0Qe2NvZDI8c?Tr?<<|2<^<F(dH=U)@9s1w*e{OUcR(}gB>T!BARvK4Z;tLBgRH#&3
zri=`Kqs*)nEe3oi`Lu0qnWWR8mzKW8O=^-l*x3nh3wjicvn`rAzoFAc(>zSSyJQn^
zc^;~ULtMcE5sTGAK86BJ`WpO$MFtRFig*g7|BE386{@hpOdK3B6iGKhWC5_yPyjmF
z@9H+L^qYO<+ad~f{r`XGt^49FiZXvp$0{{8((Bg=59C3CHtIHDc+kBoG^*$Y(CREN
zhiFAuekz-pnE@yrE`{$WFcc6@AJe;G519AQujI{Jh6LJ#{^xv&kwYMTc*eZKQ$VLc
ziKS8qFRJJHb^Zs{2$2XO;L?DvhpN2un38AGt_PZ1VvASH=4l67uVk2+bvQympa43{
z#Z@Mto2VghG||MnvW)_N1wV{>7n=DTnPxaYE;GN}DG{|Qn$-yHidUIZ(eJ;FC39Hm
zp`WV<;$a2cNiEu3UiQ03fJpV<1|P(n+FYbG^!U=ch40qvF+aLcd9{RGJSyVK#kdnT
z5=qb6TEP4MIy4gcUwFiIKV+lnj}2KUQV+4X_>J}T|8N03Xo-8i81x+0m|Tjkriwj}
zouQ0}rSV3D;&D)34J_Tx<kbvb(63GprTgSCv)|YSEB#yWcVzBpVc*gWoFB^KgnF&8
zI65Bxge$*GeT`NcRQun1BhBbOv<g{KG9Mc~n^?Vw=2mu>CZ4Gfbz)s;^LxJTL~nqs
zX1t9a$DiD3S_SL?#z#lhl)1&4ZdN;ZsON9ekti`tgx1^3epgi1R9zYN7TS0AeLTlr
z_1f}CU9-E(H%~<M=+wkKOzznf9QLw78P{Y@AsCW0&`Uu@zbbuhj?L75XbJa=-;#2>
zyv(yXI?Ji{r)ON^D$?J%=f#gt(`!K1p;E%-Dk;Lphxkf1{WjXdpT>)aWhpb;GVWmQ
zEBJRuSTVutO@X$n>b8?1K)L2|r&gq`z3_WeW-$r(GlM1x9Y2gYuPw@J^i+p6OX0I+
zN_b?0+959D`Oy+!xtYIDr`?g3nR?)}P$Kwz_x3gNOwh&k^ds}VWn}j29x5P^8mC2;
zPgaX4t)<+5rZW)X<D&B_(3ymVxVTd`4gZ)ii3;_zjO*ykH=lo1Ah+hxwN(_yfwY22
zW1tIGr)e!G_f2>x0^_eO*8rW^FeuyjD4qAWcg;ZW7rZ)ZBG9Cr8}?u^$1;LH<AO_N
zsuW5-)CW1Ebn^^`NKoj3KT?Ju9lr<H2l}5!=4R_4!1=1sR9oUeNN`ee>6D-op&>hS
z#(!+8?`dA1tX7hh-pL@bh(Z8XR!~qNl35O@3vk@Enha6l;nFfPZ$FMrPfr6iK$1Z`
zSkS`W4U~Q#LlJ9`O&yw)UI}T{uSf)6W?Al;NNTW&j|2&)PZ?MbLtUB-??MqhQcM4}
z)#JivN4SvI<g+kS@^uo9L2ITV`Q@pCp|Pw4k|{);CORo(*_xrd=Hwjb7(-~I&^g8s
z5j77%7jz%$rwxsa;+{Hpc6OHdJ)nezRB);+(BaW!PK!AJrQo#I5+l5%AioGl<Ns7a
zllnl<T2GJk-|muEgwsC|7STJY58|^DjnO}T0E#w!d=V|vqrtd?Lcr!>Bhk1G&1~S$
z*Fjw`A`jwQ^iR{+4A%Q|97_6nUGxp-&KsUiE!^ycf!Z#uhnH(xp%kt<&C%$CWrF!s
zcVcga$uLp=E}FHn-VDO__-jgOZRdk}g#xzrf`98HINO9i?tc5txb}F%=daE8tuC!0
zmx3-OshDSD)YVEe+JaAi9X-Z({bN*M{JSJl?5&2A5*QZYCS?{H?7J?h(qKF$nO43>
zUoBm6l4Wfp8#{|U$JQB6aX=(7K2*+q*>H7cAj?ymCPeu|z<W4n4rNWc^n*xLjX(cf
z^IeVPpBZ3*M1T8aoS1}{yA52^+qIjZNL?TB6k<r<y8ZCsHy$#64vytEk!cQGGf;}t
zRp}L!{}Zs1{tH+e4eHtEfGg@ke+sm)fu<|P$Elskj`8uzEdPI>J%V-k2~?y=V#T0@
zKv|Ki04Bx0O`(KUC>A-!ioM1B_)OLa`h+()JZ$ps864+~{;H^|8W<e>pJyUqrbK~>
zn^Ryj=Q1hvZ%=u{x*7oUymK&Rvr1l6mkr2y_$y!sV(4$mX6ybY9xr!HAp|eQ7hO|H
z8F_321DHH*OGxWX7%Bwei|%GnOnq2HQU#PhBL|bFrSzlGQ*v`KHvzbHAbJzXHas5}
zg_Y-r?eos^#@G=g+0TD9ux;D^-;Orwg<hltrp1NlK#qqO5tKB)R<24H(M{cD(f9Fl
zT3S*=E^{3pe?q@<fzm)Ys*a_RFJsLwJrL3KD<G|M6+nm=0|8}hg7d&bf(sid5FF?1
z#=h=H60wO$=7^kNp5Bf#o{^<x_@bE-2}V9Bxt`HyfLCC_wgF7KVLV@snvR-U*U}Pn
z7o2z_@SbCnT|1Pk#4{q*QC<e2`tNFR31zwdzgM%|kt5%oNiCx8Ywv*YQGn(<nHni+
zg1>@a72b%c5a+wN4yc$MMCfag(BYpip!+(*e_@Q0G1FtHoTK|b@8@vFIRXBloA=;e
z++LxLMa(Pv4n5vg+Kk@wb|sWDA=<3K&=@Z|P0ZK`qmElHauS{`K}t<s(HTu$gST%=
z+rkJ&x}^RYj?r^xk!%-!ZmMa|{PMqUe|r2r{ARHg8)yl7_4f2oKSs>x%)G726iW9;
z{Zyjz1bSnDKv8ew)ISv_N#}h`FJAI1em#`phU{Znfe0LNiQQ*gE*1uz-p%|O0`}l+
zGBx!NatQJiz@0uU3qvNBZEz}|%bZ=j1T#Mly6#7bJb7-6^hYwqn3?)~)nZM>Ft!Ti
zpb@roM%aK*@hsY@!n?A=uq;R@W5YDR_B|vy5An;jtC{yGbC<^Bbz+t=XzK)b0O4-Y
zVIVu<?&fC4HPDaz44Z`-ga`ObHa+}l9Q+>SO%VKgI<mVEe`3q?JY$S|N3iGGqnk*x
zW$hX!Y$TEm^04ENKf!rFNC)PL0i~skn4jVs8ip^z;%fv_2sA0Q9I8c;nCvB#?ByhV
z*N2J*k9f^TKHeiy>!q-T1=x?mlERq^MWP5X$Ve|`U_@Xc?$a!Y2$?PsFd_&TBCsSv
zx^1g~e&ibRVUkXiNIyDd=1cbI%)0UHx^W$hVXqNq?3gvhPXP(3sNq<WWHE5g00b>^
z3h49n^03su2R_2nI$t26Mpg)<>Q@jyk{v;2EZCFg7to}CwW>s|FpuzH1NH-;ZD{*(
z*>0wFd}in8@$apkU;JkIZV9gKAC_}{!Kgy2>M846eWoiJ=)`mbk4Mb9yW?J`bd@jg
zCCyhm{LNNKe1QTKXse&PHUctCg5WlwdV1ijyVuoiA9i5Nz{psbC1(jsjSm^;dDl!`
z5FQj?-)AM{U`r627>Sx}0ur%dpCPOWg$RXkIbwV7M1Twig|MrJ+Din4a{=y5vt;*q
zI#Kzmpu7p}l2LPI?A%OKdeK~L>^y92?T^wT2qzy^*rnuUEQ*pj;=xUNr8CmDSnc0X
zB$Dk0XbVjIYihD-@p-;*kPP8zmXcgEdeMS+>hI<8LKPIhCudb*=k6s^_t_u>I5Jlv
ze38KfEHuOqWbN~Fp+X_PwQdNg8S_H&QsdY%X4K$Q=8{EI3g)3>sZ!G{ej|?zq!c19
zisL}0j2V~<oVGA-wCMxL1&8&D7kn{SKJfDES6p|MI&1=?+~+*G2D1SsNyw#wnUw?7
zxe(p_84MyEc*xq|s_WC_UDO0R5#34Rm^$_93P1|Ml(wIcP>kUXIncoq_4y40`HBHi
zwdcCnx7y>LQA1wRb~X<U?d@w$PqyP^=dW8ih~KSmxd7yh<xMR0WJ*6up=m2$1la8Z
z_F#-Gr&?S32-7dd@~~kU6R>2}>ecaM7!p#W6Ae*tuW}=X2E1e9yjI?^=v8_hSl;hj
zn$N4J*m7MESb}r)O(esc;esoTLCQh+R%#TQ_wGYEg5-NFbNa5VDmr=vELi1Yqs6F?
zFfuT5Lg0aCn0<ixYPuM^ZB>g1fgpzD`Ms!$7o+<!r2rd7rYg~&XUd*CiEW{Fb!6P9
zM;j;U-^jQjjA2um7mAwA25uB9M}S;li_6=whp_)LBsWEP2mqdDIF+s%dIx5jwZIDf
z-)+haOd_`G?>?SjJNxeWJkI1K4xmqJAb(kpvV8j*`U43GhWYyXNTK9LknCkWZ@zj_
zQ*4a4!0KRiCw7whdoq$GH2u~FkE{80Xx}`jQ8t;D17{(&5jjgue-gR+DmfQUpXotJ
z{Se$=t?q36DX5I)rDH!hp*o}u{EuzXTPFy9WhFAUueQ2}Wb+7|Gb-PMU!K;F%`+xF
zYuY&GRh>RP{-KXo&aN*68WG3O=b4r-{2(JXydoG7s1NRXnbvF}bW^+v6I@@H(96Pk
zJPI=Xy{=<_I&`?;7Xq1NW;Y9m#m?8p-C$%H5XoqHLNSk@Tb;_Q)5>2H8>_h!`swND
z4FBUwYNX&l2NpNJG~wjiJY4cslKqE0?0r5WBXc+Yxpx`-!9!l|2N9wjPketWU9N|j
z(=AuW;qikiE6+ynq9Td8bmOiI1xj#Q8c2VRY(sr}A}d|!jdON9VdQ|q3LPxI&#3-1
zTpi#NA%`uB0V^smddB7y<+(r^+%bC5hAo5EglG$&!86jFVFzE5^2bv>`xXmt(&s~L
ztzVpQi(vHObylU&j?cLd#wEPYo_*xs(J{6`{ff|Cvz^FjF0T6u%eaV$!L6s!=RKJ?
zC>oA%xkSHYxw9w!y@>!9uE}CvbZpg(KQRHI{)Art@d)tAK;G)Wk>t7N_dmCV_)1VH
z{J0H%<@OR{$|0_p9Y5Fe$-=<mTX+A?xDPyu85A#1=-Cm>_jGuO@3)%JqF>hUtzLaL
zfX|~{qKuXZ_sU|?wMP$Dr5pM$uX)H;oc@3Dczy!FrLU{2`>(!i0|_}<^z%(-c8ni?
z&8*u_DkHTKV559DA@4Drmox#BH(mBtwBm2N1JTmk<;E%c@yWTlqCHLp4_9ctVtOOH
z|KM+!AGY67Wx4u5`0w_0!FyoD8gJJTS8~KK62-u~+6BKpsK9L<$~arHC8bF|I)sp8
z7U8j}cucZs&;846(u0nKmPz~i#71jk9gl4aZzD|6g8vMzrt;7%+{53Ukn67NaD&J_
z;{w`F%k*;-O1KkQwtwCjbx^V;h_7#*$P#1iFMMeBRvXQHj_`s6KWsg>Oa*j+d7tec
zn?5mNwiFp-V+6E-GxVdk_La0eY|n)f{Fg1;dkj;ePp>d?E(<y$8aur)yFz_ro4pIH
zCqAf^k*=g#G3`GcEV}4o8$Ajh51$PTOj&ZI%3&gq3c-pf=13^Q;)@Cu>WU8YcG&~X
z(IMxwjq~HkGw<LZui_-4OGA(lB(V_K{YF0epBvr{9S-m<3v!V7y`?KpN%}Utw|6qg
zx-Z~na$0^jYtrYB5z6;t)l^LNdzZ%@_0>79{T?=%RK3nrO7nFHEq|^vLjZ=zv;ZAf
zG9mZ7S^gXEqw@JA*&gNZNt-@d^NmoSC`A!SvE@~5*`xxlJ(gD7SC|XD;HX2r`-eBn
z1hV~7f=IyuUi!WeKkLdWNwbS>B0tpsBJ}B1%I4sP!&M<65^2ClH6n84=1}@%zt=Lk
zy*({iX!nuIo2g>5Upday&!@+u<HDe>I5~WWetmlmFB0(35TWD!CHON2kn?|yLx2FF
zIava}0oGfhJWkt<+`M7I^(aCWUMvxWObd7%f^AL(F_WwF;Ehh4&OK%Zjr=<&EVhU0
z2k>#Tw(4^A;+{1`DM2{@wyiTW2N_XejC*e%axd1F^kQ|9%L6#bR0$P6I~we&e$)n6
z<QiA{@U+A%1ALENXqKr%06q$wJa%<bi9G!L#)?~-Lg8Rr_gN!d|J;2bUW8cLbkU$?
zG9d_&5TQ4fxDnm9PLqyW$S!!O(@5<+sY81g<__namt;faV!Qf(&s&PA?p$bJ8KpH)
z{i2o{frpF#IlT!zuCAS?T{i8(M`UF@z8|{kx3R;XB2(Ev4$fUWU-It+dOB%D$Y`za
zX}0w}Inj<6_@j3&G<V*eB@jEjGBT$UK_Hz7&};K!kd!>yPCL!ek03>BV7^m)yq2%V
z0#C@iuzW=bRUQ*_@k$`4%KW{jz}l}D?T3W_eUjq&lYU79hhN7Z2w8>YQ7l@<ZLVK3
zY6&BdI@W<tCUm<l@tQZ5ee!{f9L$(sJL8V7&qgAU2CyL`qhaHTbhdRgl8Zs-wzp}n
z{k`ejQ~vjCC}EJ%9C7c}`1#3uSktybW${{W!@2(FoNT(_oKX`Ml_q~3nUa6^**jeR
z%qbK(_yCTY458y1F^biwwv%a27uEP$68~h$K+BcsOa@kqdhiSQdQ3^rFBX0NlKPk@
zt}AEgCpd)QV|YQ}UqlA<?LO5<<<`KY5C8TgO|Sj*?~qmCi#l#3wr!`MLL7Uc)i}Is
zMQXh&=e|BK5^S)Bc8~dzY|C0S&~>F->{7kgcpB;7fc1qeYF@3nSwcf|(N$w+w#rz4
z@>gb)j$28NoEZYi3j5SAfr)%3MS$Y_CT46r7KK8jVl?jmsq3qwqWYq?hX#QW5T#?3
zPLXZ|0YMt+2I=l@kP=iv1SFO2?k<rM=`LZY0qKVK!0-G1_||&AS+iKcy>sW>bMD#u
zInUnb*%@Pk#x$Ign||{@W?PaxbQhN0k-D{?fQeu)&dR5GemcB({V(h?H2=+$+SF^C
z_U79ZZbk|Ja*WxUVNbX>tkb6=V?LNYMxH3h#%9khgTF_!jvXFO1nCMwNWVz;uiQKN
zD9UcLNSj7rUHkWlMncoijZh>cJlM{RF3zv>W5<nf*Y<G?d4JvOgvz__{Dvp`{(`6V
z8txm{VsI+Xo3?lI_Z$73zdz6AQMcKU*6|lAa^7K|vzsJU)tXXa2DI4?`?CD1EncGw
zpl_NBm5*%j+oE7pQnTFPwY>jV^e)NAQ8LoC#+Fie3`d*e)n^|t@UOabK7%1Dcr;_K
zWG3n|#<j8R()j>F-4HL>d*b8ViG`*8pq;WfK2eqx;}w2}tqRNd8YTJ@ryD5^Es4Q4
z)vvy@zaGugMZe&XmYnXBPj8N+mvxk&{L1lz`yr$ZWHqM@=1gvb998l!zArIHHeJYj
zL$hoDt#~dN{$A;V4j;z<S*$p2cv+)aM{D3HiI!7Zncz)jd=Vud!Y#!Uvq~tb%q)h2
z3G-^!PnoU~n&OKYwE=oz$wGmQ&!?)A<oko8@d!$6LYCm~0+g_!1f$RTPsg&p!H^-Z
z$N51a2>Fz&(<tG@h%A<W5HXKHBE6>pYeCTq=qXPF(BfCBTQ}YdJyr~GMA&=3GgIGt
z>$TIL2)Ez8&TW5!{3#{H`{Ky5a$3k{s;q1WTm1DBf|I2=lqmZxUeFhJrd7V(E1OBz
zU<PcTa}HrOIii8!O-`{~Rpyi|h8dq!G)Nk}`o9Kdt=RzEgB(DOIWv&6){YLD8JKVh
z^@va1JH84VO~0D&qV+D6`IR-6LYXRTZoJYh%R^wB?S_G3sL&|L#1@G<ZBaf$W+8<S
zIseOk`RD?y+8~%RNqX|=$iHhODw?~7+QcDgJ_mhmr-emDO<UIx2p$>Pwp-)1pTDLT
zHao})b*djzzz+c?Yx@`~eJ!bGyt+Bnm(zVHvn+Fd_)%z;j#EGMkpIZ5nMS69j;JO&
zD$2QD7qoV|-HeOWGfmXL%rbby0*pqAsyZ-|n=V$IDP|Zceea%zgfNr!I(i+idsRQ{
zQam)?$G01ko{hA!Nys+MJz@2JTK%`~+W7cIV(fU$yGx{6-5Qg0j%@+qGaFOx8=hmH
zWEE1YEu{Dr3XD~67Xk<*J}BG%YCG3SD}O#tHDQ8LIZ72TPbAwl(tPJt&!O%zR%fJS
zMxVPnuq)B;y@8%InO}%Xlk4UdH<j@1$q<9Ht=|nuoIf~lWA|OXwQb$aT+VM-bYCJ6
ztYI^#TuQm_`0Xw27H%xLEwaF{$me4C3v8dg_{d)>)9!6-FVhj-tQw6P(Bo=BZ>Lvd
zgRs+>_$%W}5OU$nY*#DDa@k?Qb>Smo>j(Y4=`!=pYuPgm?x{@3e#LAStrn~EAPo9@
zi^-}m;e~+2D&^0=p8AzxZ;`3O!Glm4ERTmOn?DR!6?}~sME$d(XwV?9@7;k{^Zi31
zg^8uXBbz^Bf%8T~ekS{&fULKO@6D=#n{Q>e$7S)RD;oOp>V@zXyic#H9HkRz)qpW@
zKC*UO;c$4}ZC7isgVeDc{64_|#l*PvWlPF?E@Q4@DwXnSoBL6GANICH&d0|`UOt+q
z($cLg^!k%5sC%c`VD*R1_g(7Vep}HgJq1avj%c30)qFdSdcC(qc~~I+aFb->(%x5@
z_j3ien;6#<=~{9{t=8`M9!lk9c<uqotU2Gp_52S+!c!B-67z!V*2$WnSJ!fC>hJ9H
zlD#*!)g2{>_f##)Su9SCC;vIGDe>n0$@jczH!#nYu_IaUTIsZ}zZpvw2T112XT&lO
zpcHJ?!w@Z=l8fYtxA_zpun;8^?BXTEnn#jYwex~jM_n-W*W_I1_8(J-hD3jKcMiY)
z&Tt$<N>5)+tzYVR{xqxa@CJAZNd*fujB+=*+ETq()wr1mJIy%uh8?$*&g^dIw9jT7
zFX8px1f0nhrlHHTrgeEN;+WjbFiMpXTp!1Y43sE;qi5-q4PKfmxrsif^4-9ZzbLmT
zG#}&3h!^#G-HITpUZn<pS_t+8S1y9$YK5p)3F2YfpPy)2(1#}d>7*~;63E|Qr2GB0
zU0|T6eX~h1t^l6eJdO|#sPr1qDUK&+?d}iUBA(yl|MV6$8rZOjIg9iaY^$w$ffpYm
z78eS8wp)_>9^P{7b#g&*d2s)Nl8YdEO#W>=@RXjTrLO@uY}9T1^v(Te2pwT;iUyUQ
z<2e<oHxCUKHEToZVv-WB-#rwHM$yCSIb7g4{VH-B&Ma7L(AXom!gfE7kwtikxr)TX
z+hHR_HYe9t?DHA99TZsH*s^lpPGd%T)}lgu8~;u)dSIx8;5F3fjdv6dXyKQT=aIw;
zxEvg};6si?OL88s{nvz-b$<j^;`lb6)Vn_TeASa+@JMt47Y7znqBYqeL%ghgl8Bcz
zx$UVRbr}Npv@Nk4I27u2)QF3g|Av0+ovyBt?5A7&A%%RwCh+lA=ahLJo;VJA@clE3
z2B4P>k2qWNSq&3%u>8G0pf0j!Guq6}A6-e=UOVs*GTQ7ld>)>aH@LRRn7w?#(fOuy
z@{xE66IrjbdkhC_n&RQf`OJ@tO0Q6k6tl!!Ueaj&QjL^pF2N~C0NisBfVqPsy0|_7
z0`Ke_8SJl>!SqnO<F~$d55T-6rBr#j6{kAdC31}<*6N0<p<m!e4u*8uT@4K*S`k|V
zIyP{Qj`LY{24Cw%(N%JI){NsFBGw8&UCz@LaNcPY945D>4q25iN0d;Vj-~H?)O<14
zq_joCKXEm!tF2mm>zCj@=249T0N{gMyge>b@6`Z8Mz0O3o0GkVH<t@hsxNRS0!Q~^
z^llgyLQq0+Ii)?NVo+p5B``wtX2)ZK@x&RviGN!zE7tQk8ggHspH0OG*cEQPti_#A
zXjoR1kdow~urm8Q=GNjD?2^Ju%yY$1Ymr~2pe%?Fu+u+G$M>D_Gd<0Y<0YhM5HXgE
zn?~O+G|u}i&n7!0*<P!5`JrT2xqLDm<$cq>sxAn<|IGdm<<3u&BE+ss<wX;g6oNl~
zK$C6#Hd`MZ-ZbaH4?oL?CBwv<++$3?SQyumXp}<a7la@HTdh5CjST9-^$-w*Gp0BH
zQZOON@_+=KYJ7+?d@*Lj@qL8T>6_&;{A&N%MX_kqHAZ17fJVQ^9~@U`o~AwArJAd>
z{z!%aZOYU!Gc!wsmEIO~lfBEd4xST`BzIzlY}8y0P>E*UkB|m)7YVet*1pv|y~$(#
zhg(k@g*MiBN+{;K_+VP8235%pS9;^*KZMh1RanAT&HH)L;FLM&@Z7u_NHY!U#hDbj
zX0f8Pb&M!g!kdp3_!*Ntgx{}WM}%-%-68o<j|AD3e~kIeH)q->7l=l%VGyLQF8E@+
z(e>b=iev$EZ)7#8y-h8CyHzKG{?pSCrP0S*zI;xnYAH>o=`GdOcG&X7$3sMlF1`K6
z%I_FC92W%lc-V9p{=&=8UX|Zc|JNph#1i72T*bP9Vhqvi^zAvgPYX>`wVK4R=|Vx4
zih)hamnSz_WMo#qxY5pOq3B{bGMk$+=mb>~W8|+9oLw(#H+Zk|W-F8<q7Tj8)b0np
z-)+-<ui9Ru$#KIU<MwJw)Om<MT;Wmgjq}K8bC782^#6)+^ODqSZdQk}XWb)myJlGV
zg_Uvs!2YmsRE<w!!gdCf33{gL2iq2JCXCeC)wS$?&KDX~s&Gl4>r~-n`}cfh>5rw+
zJM%>`*f_pWy0}}l9JA>;eDL+3*Mtx^`q||9R&23~^4X=G&#|rtr(c;O+Lbr8uF%M&
zW7sV-__tu{_sI6^7eYvYny<T2*y|{_y@l(JBg<ZE+rU#7ul2!HWC)%(7E;FFf%H~i
zBvedVv5Z}7@qn>wq`=+{zTN9(i3WOFFOeipr~O}}&`EpYN2DIa(tx!_8bTzYZ?i5~
zRBfO-?}^-AOtyMZhJm%Kq;4L2^{z10)lFZ$AXAg<(lZ8Ij<q-{g9eU`x4|m<Ox5Ig
zk^wa`<!=+ci8X(olj)h_a?i6d2-a^_x*xn^wEKy|i9QqTf_Ztmcj_S|G@P588)j;N
zC(d=bAi?+;l~rNqpO7LZe~K*-#Eqac!LI!LJw7*{J|q|Xd=@hCm>7?NkQfDB&PdKo
zE;h;FvuO{F))&Hc4Z?`>HX96B8@_melK4%z?WCw(J>#^EK8M}dc8dRF(}-vB-$T$N
zu^`lK#1N#MZ<|`mSCLN{Ba5Dks)8#JKksQt;d@VTu$(prw$#XfeV)00s$*EDOu7~A
zKpF=OVXq?`$|Jwbr*KD<jX?=s8J2O#r#Tnjq1okMIfMS2(d-1L%Z6DD*g_CI{Uymn
zOv8G^@3rkYKTQ`-*w!nlAP_1;nO860{v0S;$UmA&WFKCfyPlw?=+H3mReSsPUAgAX
z^`5^0!VdFFE`%5deek9d=L+RFsW@i#7$|AV{gE6Q5{jfssHIFJ>nS<1|1HKT1TO{x
z`zyB(#`c@4Cj-vAoSzR4nIjFpTt}{GfD_|6Tk*n%Pk#-!wg|h3F-M(Gw=^%#HO-eY
zK;j)jo{3>CZ<)zsE(e0G@jXj6wG~;f9~?4PbP&^`LIircck=W3$yz?20+8F$?xw|%
z9k`$4_KBJyJNdc}8A2B)gTQRJ^ek&ns#x8!tR%4Yf$3*EorsWU?)d(%=zTNxMPo7!
zYKCptJMV~2f*%qdhl9K*i=`D62a$d!huu9r0;vkE22rQp&r=5365k*QUv3<Ob&q}R
zEXtt5Gii8<rtA>8%s%r2lQthe^bZ_4&y@nuZuWL`Sdpkh#=M$aDUPYd0-fR=%&o%P
z(&IlF=dP|r>nt{Tig|rF?j!SRxjKB3Uc~T$Bd<^!dA>rd77~J&GU>gZ@jyu&`PLM^
zR$D36+Q<rz&b`$I5-DewtU#>7uC~8o?riZQ7P!a{QqP7Dp<xmT>X*dPQh;MQZy}(f
z0`0qpw}bCq;EQ)8M6dFtDiD1sNv`_+02<5ZBfHh??^8TD>TcSz%?QEM6Ze8om=WxZ
z=*p_k#<F6}>BYJ2aGG7~KNQhPdb60bAf+m6gCd`4xM}@HTl6qDvq6|&%`!M{Hui$z
zX|Wfm!K2}ZxESPJ9GjR(6UI84t9*$<r7&zKdf8QHNH%N?!J+==<%zj;Y|>roZj+c0
z|8Iv?VnIMf>GV~=;V&auMiZ7V9DMK}V`8@kb=eAx5vBgsWkANqXaNm4mmnBSeygHD
zZCLU0bl165#qysB`GBj3{K;0Zaa0PO+(sNKeXFm(&erOqSt?QUXBxL0aPIu=jfU<*
zcMXP!W@$y@WV4`)hb6c34(}x-;e!rbq&E&Nj^mRV3@Tv?3(BinWl4(7`s^~XJm&Z-
zD=RiUDfifB8K*ZP;^W8C{QMC!o<V3$*|0-b&&mo8><M|>yH+_={it4r@+GANZ8QwX
zzg{SrItyI|OZA)61>w(n4yY8^G0fkm+ux)0CbS2wJ}_*@$AXn)tDJWWODqhGuS7kw
zs`+sg6IhRX^Os{a)+Vb@79U`0v2g4BuKDA!e{;3^!}_fz20)WaUw<L6Gk0EFsGX9V
z`L4y~am2&Uq4e@mtu*={@3|K=q7^kJ?i3AMCc(Aaf5*|TIu<KmwWbwMe(#zZ<|Tca
z$@cD%=U~$ldlUHqTl93!nVmD=A&D6vh4*((vSUL_^utZ@Ad|a2gTnYq_b7mos<WPS
z^1QyOg9(XAEf)j6!_6!+1@}bbafpB<EYat7*E-#<l#m1kVz6JLTFbsT@(gOu_lfRj
zU-8_dTrwhmnZm*W?DhJ+q=K%V!g~hOUc70yt9Cg$*Z*}IA9wSg1;l51I+w_8UT5?Y
z`Hxkv6y%%~Thg^m!iy*3uSxEZ+#OZbinY)5cs3#{WOkQ=+iL}nnFgACRogp=4X5ge
zQvX!LQ0`cg|L_-<Oc&yXnZJ|*x$3<O6##SJ81dE4pphn*uKLk@8J&U4G=e7#8{`2d
zcMd40$(;7}6^Ad<vs?rTdcD5qw;<voKeEWqR*GI?Jh|<V5)pPeF8iUte!iO$8TR%Q
z-5vaKZ1^?<z4E*MgC@SKfA9Yvoc~acb?G-~A>LtNK5Z^r8|!~D{`bNU6YY8G5T{?H
z<*`$;e_NRut)rl?FM8fxPb60>*{cuzYW24h)3)l7FTIN8v_2~9B_hXvmW-WQ*vY3)
zDyX&PDEvHUu}-1Y4%fno)dMVR=){|4O%G@N7kD=>UD;h$&+Qb4JUgyW%_|jG`+5KH
z$okes6JFw?LgvBx&OR}iZj$XKX68C|SK-^8u$W}@`B*!U{&mkIvGfVTB)Le$jg}iN
z^W3KpV)+gZt()66PrdFJM1#uV?xfM1pVz#tl!x$;AV0Rg)jQYb$lo-oE0A)NMjLH{
z$lv7lIrbBZapW|3=dsCqqTxgGUP~Ub;ls!*vD=qUKFGB$Fb$86t8pREU)P4w%DiN-
zF2m;*&Pa-bP>`z*jP{_^>+mRh-kLVUQ8OOahdh|uDQ`3{EVsn3i8{!=MZ6B&{q-tp
zlfM?{^9lnxz@KD6<r#g9x|&*OcKX+^#JQu4GTPtjIg;e;YIXiI&8Z5xgO59hb<)zl
z5GH+LgM3Vp)DbqJXO!t<GK-3iuAW8A=%(pc1`irO|0lHPA}|gm2x&J2@vfO7>eZDM
zD{Jd}yUDuKKib=scLkHiX1lg*$yr)r%tfX?x?SzHPCN1R`8Lvq<QU%=r<2U3Tpe2L
zG`S9k{I)6+V;VHW#^hR4x7AgpeC{WlK4TR<o$e(a6E|a?N2%Bi!ToLX{I>&7^sRXH
zH+cl+%aGFq`4mx;)kc^0^8IxGuL-N7{><{^m&1~Zr+-#PFhba*BBI<CQ;c+L0+<<J
z0N@Fw*xeqNMkoPz%c9$0Ong{+Y(fgd6ps!f0r|kAM}>*lVbG1q;!N)Gh0M;t_g=ny
zV3IQjmsaLY8wu@E@DEL%(n0XtY<Yc!(>;W&!oX2$@+ahZT2Eh*dsLp5lz)KN2R+JR
z7tKmovogOsN&=3aADFwjDmHyg`fgzx&3LIwso^DonHk<5gu-nNBG)CBoPYnSzaot8
zC))Jaf0*3lg$JH-0Tw<b=U$d4!||8b^Ii~>1_R!tmNZ2n?IsQgSk_x=I#8p2SZ%J2
zK8UG{aum{V=n&fT*LBRC$;1=iNDe~75O=-OY8F!6HJ>0}DNA|cSF(7(NvRK>Xb14c
znTx>C=#HAbCP{q><a>F)KJk^|@|<)7)5HmI?Rk`DTwXnL|DcVj;L<LCg!WKiqKfw&
z_ZRT=N_aPEydF8iKWJzOKK7PbMHr0}v~kMdi*GQ!Dp)dHDCmgGrBIO|BeP>#*veW(
zOQ1xDP)`#61^EczMZtQy!o-)y@iT&TZq)p&WDh$8^W0N8-LH*mNJEei+?*P|mT&a`
z2c%Dz9HmU&)-z|j_s7nUqojG;<=J2NrW4id5hzAf>o|GE!U*#51*ld%95);ozsDIJ
zKd>s~F43fM-S|u4CRq8ZkOCKU@7LFa<V9Zn4`WiyMckv#m+1YqydB<pJ@Iev`Mmmu
z)MCMPww-lkRo}}Zpnc?Eud+r1S2myPD22;c?u`&~NQW^fgIFqj@bB}MHnJEd3>A{G
zyIl!zZ-5dqYPcl5kh4i?zMjCL^uy1=^?Q3$l^aX2P$XJlkFj8=eAMyCNa9%hJ_^I+
zu;$pfvw;$z#Z{wSY<0VV3HuQPnkMU+KjtJxh%`gjt)jPUWC(-XyCiUr5yshjufx9M
zw`nMEh0Ai+G5=DqyVyw;QjR1$X!`&w_5}N<dOt>G{dCk-QBkSOawhNn`sCrXlB}o5
z(@|xOh*duEaI;E|o~?q2i<|wh-YnOKg+3W_+IYMeycmKQr8JeaKVqCp0zkQ;jd)hq
zrC@3|d7@aAENU(EhF5O)&t_b+r|07Y0+ZORpE!g$5V4HfZ5=wamW+@VQJ>e=Z^pi?
zsDfL$**3WK;z1ccd|8>VP#GMpQZU{AWXdv<uJk)}hs)N?nl3nc#cK?Lmq`Lujz^<}
zV6y+qdV^Y~TQ)Lls5++g=&0q70o=E&YU>L<ee!wGEM-#CxL^Y1<-yu{iubK$Stw*6
zP5AO}Uz~d3@u;FG?i;(9RXMsxPskH#pW_>rYu@4cCB#*SLEbRDa?3R~i^2Rx$|&Qf
z+|vrhFN`pOm{CV^MJ(DtQ7maJG;PH3l{tJNFu&N|HgRT%^m-=4Pf#(DP0*)l2JxA2
zUG<&J{o3NeWwH6h3;`NMI4<tSx#I6mOjdldT#nQ<ZXzzm?Bk61tz$3S7B5J^w2iu;
zxC4mD1roq?r>nFO+%2!Ka%8F!W+-q1_z}B#bqoaWzIR*MXON)n24gR#8hDOMl*%b`
z;;+B?wPW=HFQA4Z7D4NV?usV8bg8DU{+`y|#-?a=^XT&0Xp^;Rz^t-=O+`==4`Aui
z$ytP3buen?<TkBLtSSG`MN>D|-JxZM%n%i1E(c9iOvrPH@|Wh=E(kG7C<hv`Dlv*u
z00(-bY(E^nE;>^=yEMkbTP3#}i+5{kcheCZb{*M)X2xwXKZh=jN&Bz8#J4uLHl@;f
zaqv|5t*Yvyvl~!L!>v)+gQps>t|ZY?X>f|<CeFzJxb!_{AUw^J0;nv2uEB`FXmqFf
zVu0_!1!R*p%>yVKZJ}-77nI!+8Nm~hO21N!^x}bRdsX}IjI=d@IZ6lf9Ubk%B7!vU
zArJ<%H;bBoE|NrX#}R$`vSi=F<5V}827zEXQ2qg>=&+kIba#ICch_hj3sv*pzK^Yd
zH@IyXMtJ=&uN_IExQ+9-aox96zm51%ew$iwh8~jNn(jHN=Pr2!>^6+`8GaJVyOK`p
z3JoekojEQe@pKx#bHwGW3RU{l%^5#g{l&-p#Si@k|D;a*z*_J3*lc>*NTt{H%6KDl
zeAmKvdK`JqaewSK66j;@agrDZWkRD{%IkI9t~FE$e|f0=k`iO{Ds4Iu5s`W2w8Q(=
zilzS-5%YXo4iEWBNFEKMf{(dRgFw<m_%QudgG}wA#?x~oF}u{K&;~8L(VD<q!*)ls
zvZcBgnhv^1EPPLqJ}OYVRDV@~lX+BH_pIJfPNMWW(n{u(t|vY5Yky3XcW2?M#G`UC
zo{bTfGen(E2>S3IjoL0y|Ajk-(#wU?|BZJuZVhyr<*FWqg?#!2EIz}x_^+CKsqU8-
zxMTM7U2wRQB)z;O{n}sk8<Fd~fNUJ^f*59|Q~+=+%eh^&9=qfcmk-D6XS=8d)?^hO
znOi5id(OQU8C((Vs4o!#z{O78T~;N2%3~Puy)OXPkU9%kS4;V7BZRxnD7wG8BF{Tw
zC58J_J?>7e(5zBx@(d|mq&BjTu^Wc2IBe9|e|EDQ&2>L=-J`93)4R$7@E8Oe8bXS~
zN$Trse=z@Bk<Ulz=XrvlUI#@VFJ+%b5ywrD{!PieR{WE_sc1lC-g(04@YSm%z3U`B
zlX-<bc2P$Rr+kq9*zW~KOGO*Qzm`(v5@+e361MmTX$Ux-iP~1h*ZqSs`XUqg)bQSH
zw_5yuFZ8e%8a(oMZ>~FWTFo*w%S{Rpn)D!PKn0hR*r(a<-*H1Avmoov>>wPD|50bE
zSpvMvdtRqD%BGga#l?=tfH1GoUH5Ip=~Ttr+sfEcwo?eQfN!7@6}PPh{{1dk+tnjO
zQF`dUkym9CotyYHB$*9-MPwb=3JJC}f>*j;!_PZD_$M)T5wrI~MHO9uKAj#_TuWsP
z+^PU2AxvKV#KZTp`6f7vCe8C*MyJ@rKvuY)FHBOIAzR5`c(#qm+~l87dVD+=(9n*d
zU~hgv0r*@@(5X+Y!B*J*g-Z-bnV4^z7*Z(Gb9D19<ogW9Pgb=UnAxZ`NCPk$DSP6X
znn7STCTnes);0_)jmoB}Io!&!m~yLZ6t&itleM*!ksrVBAV%pS#&Q{8+epHDfXV;?
z-me2+-2d{S#ZoC&pX`UdS+--uWvk!D{3$vPamIw71g4%zN$9w7s)D)Z1;C|Q3|0W{
zrYgV>&q1sYb@)(d*t=OD7`yhvrYwd6_E>(uD-TeZLXlB1yv-3=8cmj8oDVoyawKs~
zB;>$NbhLLp#Lu;iUZ`FH`sAy@7f;zrf>rNon-NY%jtL9C*?Q^NhiYn61tt9Y*36E4
zS1(o>ml@VDVR&a$kY|v#0LBk}9~kBK`y&UwDyqMe+-H_kc^BF~ePP~xrVLTqJQ31B
zqzSs)5xb-17sI+kS+ZFxdvx1J(Oxd`DWu#*$7Euw&YegfKVEaH2cg}{QX7HP7`&&E
zE7??}$l~GCHqg7!SoQNx{1UV^5|g{lJcM5sI-=m`4%c6oB{2azRXP1mOT9UW5EbdQ
zUUj=F0RxbG3c(OmJ?88oM1P=46WaIT+*H!6!a&^z&|h*!HnuNE8iI^t#I!#gUwmN&
ze4rapYB=!u_#{GDn+7=mc_)5e&B8sNFSXkZ&X}tj|F_L8lp{2-h=D-$ssa#bG|YbL
zlbk40flA_#fbU&vrzvR5RbWf)D}y!rFH=^Qo{X%_AP#CNgYVb~Y2)32ZeEa;X7I`~
z!d|5SPtPP1yETsJw%|!IGnn>Bd$?+dP2T%zvI9rj)YMe-A|g84RUmbS{MxK?dd(?6
z^(p3^3)@DxHO{;Y`n`P|OOvnfuruFyIy{<#4mLO-pb+Z7+adBD>JU_@^Y6ANvJ7oW
z=U7fBpB>yx=7~b!7KtHsX{@mgbUo%hy`b6!+|K}AX~luEFke{Rgt<yEl48g>x3Xi?
z7^wK-Ww_o_{hVgU#FNT85@=*tpCy<2{&}?y&){j6t9SuzK{0KCt;<`f58TgsW?*zP
zD*ofd=nhpAC*yqtnTzrtg-?jxlibNUNTY>@1<7Ojz0_BqlSYH5^iW<{M5HArM+&&T
z%U@<Mk}zS*zj_>X|4%%!B$NaiQdI<`b(o#qKzZ?Rmy!0>+7!^REC$?ftDYxZf6x~W
zhDDn3GRBN}$KR4z7s%C+O4#DUPpivcfj6##t@FwL2NfCne<=fg-}~&J{}gEcG1@QC
z%(6ikql)@+y-so6cXzrqSF!M&?9f0Dgu{t^*ik?>8rgp7oFUZ^-$lI{&zh@?<ikFl
z#M`pRo5XFFVcT_sZWR-5O%rZKJv=SWJq8WtL-S3`{^X}qCLi_sppux}@?eoc=omy@
z1&Tk<u>!x(lfo)q$^IZ9g=4F1wBUiZ*2_XB3wiG9alG6(hR7vHY)W?&MgLq=mMU^D
zw8cq!W;II1`dQ~+%S>*34{T15Q|32M({<RR2P_^!9-eLpIL`6_^Jb+E%>Rb=t8fBa
zwXEV#6U()>6|bzUBSv46yA5qZYNNb_mK-0IhK1`tm03E(PseOuUzm@j>5F7|Ln-5S
zeyxzcxctbvsYScAwgXXo;`JtVi0LE4r2Oi0<dA2tj6Qr1A?SD{%eag=t){)e%x`&I
z{s>fVwya7G#QTGM_x)_dy_ILE;fjy|bcN^>4<SY2<FOj57J5{R`Gl$v0*2Fmq6Kc6
z4c}L6K5f&jzqC$lC%eujB<wtG*9S1K!ZBfpjvy4+TScIF?@<cW;9!t8YgZ5^gAqD{
z9t5DDVFlbFVB?g!^swO$y1R&Aww$6Ox!lpv!rHP?v*%z6xCr2)cV<(~fE7S;bGrek
zs<>36ZH0lQkkSrzwt`$QavO(rsXVJ(%&fXPu=&sW(EWinfS8y4n=TDyM>*3(SyFDS
zbCz7XbaG$+f%@7f8zX?%l<(tR^r`f_zBpa*Rs??zgyTO0LcgOezTVZrT(n^CdKi?#
z3sF5e2L01@w72NL_A3F1*IHp%%B~NkZadYdd9IUG2kWYq?ST8azk#r~9i~%dMgZtp
z<ysi4sHy_>phCCE9C(c2Zg@<Lx}fBvQBb)xFOQ*Gl}i}Nzf9~nG7SD27{HlSDc@ib
z`Vj=WCNsQKg30{n?m<wHQ|&M2C?4(s4IP)PsdemySK!}~L93zMPX$Mdc&8k8(7J^Z
z(4-aP8;9pC^qjs4XuU<4wx=-gfs`mA&lwn+{Zk0UaQ9EAC&>S-ulJPXK@SLK6|cLa
z;So5bZNDZ4z1&@sY)>ny)7Y*@S+#}6b(uAfOQOT%Y~yfIn>-DgZX8Pr&w8^l2rk_0
zVN`jOgbwWJXHe5tA?xp|OlK#B6w1DK7N$D>yM|Frz2!uxg<zf^OG?Tj*3lO08*w)b
zSu}6fFH@|U{j?g?CJPlKOZ-N|y^ARqH=CXr8wX+8%)*<hmqJ#5^A;-9l6@lCE}Hx>
zx-t$c__ud+R;*^al+UMELZdgy)_{kh^KJ<TWWcJ_?VGR8D~kMEQZ>OL;~`w4AB_6k
ze^PXD)_hwcvTF=kS~%{ESonBjcVY7Ji0{HT2Huo2qs1*Da&?$ib(vOgzh&aTVbI{1
z-gskpxH^UX%iFNlX*TYmV$0r0=AW2r=;qQ)s%@^Y<Kl-T2^&$He1XykDxvLt$^pM^
z#CR{Q;kb^U!<^9J@};?g{B+AnMyuZ=bN(U2ng<$Tg1hOUvg}d5#2*h544of_?MM*n
z#DrKLMlXooHjl2U>8I)!_+ab3stWw3Sp^OJR-iZw1UMOfKgpLli7fc#HAp%a<!TYO
zKP*2)2ocK8zOi7=7oED|I|+IL^;k*D)ZH&RW_Gqn#HqY=(RwRO1`VgpuGUpv&YSFN
zc7v|ZG%o3QTaiH_@dqmTrm}=cCGwHk4RdPhrJL=wLjtm<TAnwN{!Pb|8jE`I)*8Be
zQ%}*bfpDP~;makY08F7nuYJ|^ZoD^zA3l5;2~_U%$r`xRd54>pdgwng7~H~>un~`M
zUCQORXQ6Rl9F3l!vJoNoCW6;$sMSASca6{=8Q%7QK++Ps4^#LiK&PZmZyWbS%S=9k
zfQS(WIuTgVZh+M2H_wNt_j%_IFLTqqm$T{Z`G986svWt_X8=n<{D`7>3h^Ld#i4A&
zqDh^PhHHk2Z^sYvZ<<1&a9q2TBwAt{$87d|ge*O6XFp7Xs+l9^8}Aej=p>_>XM%^;
z#tm}<%Ol>}d-Fa6@3<Rnq39WW-?#(>8uT9xd*^q~@f)zYy2?1GF}RKqQi>qTHS_1&
zAzw##V=A<Zj+2$^X1AWdnN6cu;iH~7I`L7fz8lM>Mq~&i{qtjY3+c>Zcc<Rvt|A<t
zk%w3j+sT_VNvh0~J9D&deUTqaCqH!hC%XQ)%I^Q+I4u0TG=Xv{sNJlsoQRP5YQLXK
z!6RvJEc2;q;gg*03Kws;`2Lb6irmo(l_BeuzK^c2b=6zl(jG};7tEcgqCGagi=7}a
z=A5Que0z$v<&IOlsz<HRob9gntf$W@>7F2j-tESl^|U`>^q`7?s5iP#Yiuis&lIcn
zxKyrqo1*+hF&Hi+J!oq~MP6c_`_#<RYu4<y6W`9*aF~J#3E;2szu1j52{fVgf$y`+
zZ7&z!8y4CGJDoa1iA+%@+uvu7ymZ)3=cQ3kY}rN)+~H&RBy>*ts)(wYjsl*+*{{!S
zZOmipp{0HLzDhQZqaj$nu|xI+sj8fR4^sQTT=~pdZalZ?*hs@7ht^;MiVY+@h~tNa
zz5U<)0ASJZ8gOKnlW$x9LKd<K{>yR2)I+Z~9K4J=ymJ?NY)v~<nd{>Deb;v94L;_^
z=_BqB<+uf$D-!D)YeX9S5}%d8DOT+^4bdU>Eaq#G1`l4(27Z@1wrfWUxIH>qW3Gu$
zYQ!PqE^=O~(4$LTIOsl*SHtq{l_!o5I&~Jq3`I5v&hLrlT0yU!r^U;1JX$+X%dXTt
z1W~FW{MvD@i`ShQ{j`V3zWI9o+E2Y<qow^J8(;S!EjA^w%i@Yk4xumq@fUja$&Gm5
zy%^H{+Idi`;C8}2-9XnQnkk7FQPPU&=!1qRy>@tylDRI%FS3`K-XL<YJecaaadH_%
zub)<Xn1(}l>S=eggQ1OQ`eXA;`>hsl)4aELug}**D=LbCjkTTg-eKqu{zYMaTJLRa
zJ-)SFg<EaoN=`@yYfsC;$0R*IgRPd*TlX%E=6|NBFc-@+*+3;{YK?fcimhPLVpclH
zr{^c_lo3_&bmnd`&mdwtZp|@%?kkwA){DFn7?6zqpK{FeS}s{9E|>e8<dENAo~3z`
zL#`mY+ESz#kPsr&$Nn@Rw;>1&r4{t7hQzX=N-*7jEkQ%ybF+oTI9&cmMm&{>N<T!5
zaaeRm(Jn6S&&sPCWl>1Lp2bbyk9hUsSsFr0=6Ixlgc7gw?(A+L^N@Yk0mfgKuM;#?
zOk2Tced5&Le;C`JIQ8r0=B=!GQ3M@EMAPv+MtbFZg3r?7EuVkwJCD;ZyCo7-zHJ8^
zVy3I#)P(iV@~sVqIrrU#&iMTO)yFHQw=dbxutOhN-7;Wi=3q^#yv(atDmX3GuVGtE
z9Xs81<gFe_7;0V}>M$4Gao&L=!<@U$mQz_!#h7O6jnCObf4ePxRi3O}LTb}!GTr=D
zoa>rYD_<wfnamJsy}t9>*JaLChyiW!+Fg2UugAT8m=!bgO(&p}*CXyPyTa@5+e|-!
zeMc0qb7OL=zJ~@&gRtK;F=l2@ASXFM5O|S%R~{g#=MDk<-$?=%#o)+OC|HDo1CoGI
ezdy0=7UiYjj8+krn>x^vg~&)Myebhl4Ei6}qfuG_

literal 53506
zcmb@uby$^Ow>7-!O?Phv1p$$gl8{td8fhu%?(Pl&1p(=n7Le`^=?-b>?yh&??>Xl^
z&+~ofy1qX?F7+n%-t2p=x#k>W%rOm-lM%y2BSV8gAea*3ZxtXA_#p@c?g9!AzO(aG
z8WUVlt;N;sAP{uChre(mbDlR42qi?~t&oy)($1WNJO24S%E7sW?eKRKg>1=Ey^cqL
z;pra{I(1k`u3Nl!c{PLDtgBEzeq_m-!U&wPt`bHMK(ok`3%{P*k32wXluOe`t@(L=
z9LXn{FxL>}o=|Ul#O=GxB?x`uk8?ylG6RPr41IF4*4FCt@JZx<UNajWHDh8R`)R~d
z;XrpBc?BXfvgvx3tets9P{pIDB}Oc(+<3*orz>0$smyd(oWybjA~=WwDC8pY482Fk
zNQ7tsoQdJ7(-*I&Ak819#iIzqIxN+c;?HOZu%dsh#I6RveH#5szifK;C;NGnWcHrx
zx4jM)j5e0~e#5Gn%S%TZqA*Nk1c9<c$C8i^y@c;BiT!$|B@ElUl+QkkQ}ZN8q)t2Y
zPEq5b3gPVah{_WXVKslCf4DnW%o1x#e;ncN<^OxF=T<IKUtLI798BShgmM4bWdM&H
zx_8FMA`^jFFSQ1czQWU4ipq4H89w~Tn!t`5@^&680IUq3zXsPl&aYp;<T7a>5JvCr
z+@vI!6)jjY|Ho~uhk?ek)h~Bu&}+D@Y4b$x{0<JR|Lcbz=&gdm$1U6U+oN7)et!Np
z%0*8l3ZuX?oe|Ny5;J9zc^tOJ4js|Jd(B0>;3EO=UHEZO0K)(4RfQ9ajFFyRz1BWY
zkP?c9M#g1LDTFgG1D*p-SCQkaw1#n6FD5(9$Fmwp%FD~!+WxFseprT$@vq^Gu-)Gk
z(URF@;UIo;mFU;Qu^+!KnKq}EsI|eeilTlVYCAbH8%zoHM5dC68U&k0fhv&7zIdoZ
z1@#@APVz#Z$5Mb_kM3MIpO=w`KW{8MybRxPpeGM_lc%K45F3C_Ji^ZSMLg-1=jpi0
z^XJbS8yk6@_UwDrp}wDg>07FOrDefb-i1D+g0@ahPR=~<A@QYLv8}06d`+B+k|;=d
z;20VOcG*hInTdje;^E?n#c3pfj}~C5me96j-D}&!1ESZ;$hqd`#yv(-1m~@$rlz>K
zco-35%2koZM1~UmscS>Op<VCDtQ36=83wXd4jo(@2q@zUbt{zT;8VxC<W#RMPft&>
z@W8*Z)2{bYdXy)5U`9xik|>Ok2)*~`!)?Y$zE@IeA2hFjWR4f&PYHE0$fT3ZQ>rzP
zH&CCR8Yr_{A0U5a3GtpbUt13pdRYGk!Re{&kqWcP)!sOCMEDxbyTdLDH(soVDAAIh
zrh^Fy34v#w{@rS5O|JQlf)MzTTzugFvf2Mz5d8o3eK9v)Z*p(|s~CbPq3Lqs@?SM#
z*R%*kYPNN2iKR!WC5|}%o?};V#G$@ABw)S3Ilf53gm6V1JATIFLLmnS^AhP9v^AW9
z$x=;5?HnBI#QwF|Rcn}0)y%GA-NOsAOV>TTpb!p$$RTGU*hVhc-#F4)HsYMe;*6q3
z^<A{!teYDcL`H?w_jk^;tTDjAVUDV1gInyUgIT&y8mh&nrozFx4%tY0f4UEJ&a}e)
zeI)VMWiEarJ1*GGjoi8y=fh_A{v5Dkn8!L2ZTbCha{r$>?LTwee}3(TzqTYeAC;PV
zzZ@xG%AR03S36#yvOm1*+Mq=f?edZYyl7(!BdpsY)$8suqkDrwy=t4q>@1nr39|U7
zBkXQ>KIE!NsKM==iGkt7Q4oZNyH5Z@4IMM{ey%wDxT&BVi@nQvdk8+H#Q9L~!u9dH
zc+Xt3bE~G?i5XH{Vdx<e5!TAnYz0*A>Bahq=5X@i;bDlixDg}-5ki?Y>xY6}Plf}s
zL!{1<AXA2k>-vgB(R#lx>c`F?<fhXJ4G2rV$;S%=zS-cQ7^ym^{k`2!Aj~e{vB^Y!
zm<?xi%vd$`vwTu6aF}P+ZS|jOaEp$LQkPC4ASd6RbAEUv0y>H<tuG~y<IYs+M=>iD
z5T)(R)l#dNx}1sF6#VX+BhUVR(_hjXq(wyaBF<VdS>q7`G<5#CN{hRj%YX9GZRAgR
zu1})NN2|R>$_2vdBs}(XZ7lOL?aXdxJ3cZW(xf#Y7ba`vAsrncQT!{>DswvK!CpI;
z81^TOewRkaz_1@U_fB~`%05+yeXm<3Pp~@91>WP(Xiq0m1pWJbL_|c0_d6PVDB=LQ
zqJjbg5xR){fOZX%h&-&T#>~X>Z4|Y%z<I~k?~3h+NCAHLdwv3<PbO1>c}k6yIs_uP
zh=;!BS}X1xsJ;h|pQDD^`v=XLADe4|r&*yS9!q>wn*e!X(S771epj1yO8lp}d(}C+
zZ)$4svs&lFKgCUVJO(`krK%W^wjgX-DJexdg6)YSyWXScp?V-@WuX#L%$`jk;#GcV
z%G1XTpjK;<`%*%E7uVh5kC&4`%@mQx3Tm5qE_pW6vc~BC7$jp`SdicDdr<JWU-;Jj
zEM>UfC^%|Ti+>I?oBT1peVXcV7T?bOWflME_;~IoTDC&o<hI2?9-E8k|44YU18prY
zFxsYe9X%VzuP!f5*vUK%mv?pW!#dcdx9xdAF7#1ls8nURx!F<5QxXvox#<H3kLHUw
z)Yn>tGc-eMJ<w?W&vIp<5vWh08M(MrqRza~hg7C1&xt1H?tZg7Q$;JCfCzcSV{bfB
z7+;0|P_Q-#zlOKI)bGZz61%9Xwps2xT<&_IlzVVCV<q4szI=0eOnmDBzjGFo>N!%N
z!puzb7WzbqsK$1~T#b7Np&@@XN46?~$09*<dVT1N%kdgt?JYP`VS5Ixz8xjb({UPq
z(9qEvNbo?~R8#uhYO~V)1^f~l*dz$l0O3-L8P3Q~VXacTb(srW#IYX;x5y0K4r{}^
zYHDh7jjtgPAwg$qEvLsIfC}g>mcDUXsCPv?0x`|-MH~+gukrRw<Kg*B>+2}o+z^PN
z9r^TfnNI?{2|UVqTM=Tl`(e>e9~vROGviZGZ-3^McUcn<5R{uw+xC5X`Jv7c6BSiL
zM&`_zVTJHb5LN@He)sU4m!Q^sw;CJ8+z;gfX7~HKwf@9scoIz<W*6M@*kD2VU+mI~
z-CmtupYMBt+X=y^l}Y92V_@jZEd8u<ad~MmQ(06}GBi1Po|kv{l!~h9YU_m+(i^aD
z6ETi)J)wjz$T&>Po%XdJ>S22<qsD78beAuwsT4BpVF3O#NY(|kxx_244}!H_U+18F
zNQ&*U&ar3@oB0}R@NW_N+!a%4%{8bnU1dd)lI;r~ldkgqck9w?DFYDCk9Zy1%ajxp
z6aq0w!L}7kCvv**qC=YNU5;;f9q$g>umPaju77yw-6NhWr;U?d=BBTPs65G*8$;=L
zA8N5kSSZ=pMx%8>db=f$IZGQRe4%VUU+)4k40tDyPvo!m_6LLf)IFTXj}8tFz#6%|
z>}&BuAtokh@$C>}Sfj%4<mW08;N`s*m~m?LeFkAM9Vi4z0z__u^ON3K#+l(Pq_H1m
zyxWaBYkeGM6ZNq>;QsCy!R^0EmwPKBVli871O9}QaiC#ipBppqvyp{+;S+%TAtAh7
z8OQYD$TDSRML$;yAL5Pc^IJP}SYKaXutQQ1M-WHu84V-jZZ^7vgoKajI2=T{XVac%
zkVX_>=Y7TBtuH@zega|m5`!xGS7#)JfZbYO+`WJmUC7%JJXPldh$>M8EMKGQhtqMn
z@i6?|&7-5o&EXf7%~w}f%Eelbn6&-B>Q~1h=g3a{_;8(#ZZcQPVg4H;5>R5~^zL1s
zS#UTwSlR(}tJ&J%ELq=PIfT50hlpmDI`00ivRXK^t{mwZ9CYfn{VJKQRa`Y~&fH*R
zYP#i%gdQ3iI`45l7s-Fs&1mNJGEd2rg(T&ChRtZe98|T11+VVUPv~RNAR{9q;8*wl
z+_}G9xEB!<YqGzDby+Ml-hwJsT3UL0eNGHT%*N<^K<4U2m0!=jl1TzfXLcV_7f6rm
z>p8>5AVYTe^@sh1Q#Z5lwEy3*)bAGr5f>NE4;hpuIuF3}`?c1?tHZNFJoF&566<Qn
ze_v?7g0N}yxaKq&S@FZ>U*xh}A4mpO_RP3Y<xio|JxmA(qwV0hNB>`VLwfW9Z;(rm
z-ngeeg9=mQ9kP*woGwU7DG^mQ4b3QDSkPiye*hu;*qqkeL17xPiW3d3+THUFZjq1q
z&>)N&l^{GbX%!n1e0|}o7`i4WRf;D8s;Oe=Z=l8t88$v~48>!?pdvuKyv&Xs)ZhZ^
zuUxSD7u9?be>+;PReTO$5<pl5<2{4Oo9pW;nwouoA*JO)>(ci2cC1K@3G+SYeG3Z3
zYX>Q?XSEL7l~xOT@#4jO4gJZf3kcrDElGLKiT%I8I(HtM&qo|+0*G`>{B>};{!u#O
zo`=!$`bYO$EiEo^Hjl_`|6+a2v!mAG@6w5rpAj0qj+=g{w$^obch8oL0l~=}BiWF=
zQF?Q8lPwiD-{jQ<s>*)TedAx6@F#g>pv|Y=@>kTeUiYs-w1XF%`K>$CWqh?viT!7*
zaoSf-n_KrX*U_}HjVE`RVv$;kb$8lrqF!f{Z3rF@S%X%0iH_~h@^ZR?8N~YvQ&CZo
zul7fgMpb8kSj#LDi`DQoTa)KuvRdq%VxbTHOG*OaxF0#Psb|KVpn4eK5=F3Px}I)X
z|FMF2OA%91QGsU{7BbLLU_vNyyj5@husMlGnRh_bOK>2Iv(u*I`7Z3_52-jf(oQBV
zN0lMgpf~nIlV?N7Fdg5e)7!W0SrXB&Ydg<&XPA?C9`Qed<~`>V_SNEf^=newi_d9K
zGy0cOu~x&64|T5pybu&!jFiP_Aj#^+Pxp_~RGC|pIE!98RJOPjWE+(4(wE!CO~S&$
zR}30wBiAbVO5m_Mq{slMnY+{4DvwP4>>1chIEej@yY8J-M6Pw*?+UZHwR|g(jO0uf
zU&0Io#l@l2fJJK7%hlK#Sam&c04-8*<2pwxzitPUeGfAQ>fP@+c4QxZGK&S?Pnr^a
z5GFY?=ZtbDcWoC+mF3)#<I{&~JO1#3ZAsjAIz9GsS3>ejpjdicETR0vKDFR1m*~g=
zIJ%&NK->vvqJPcU{LSos8eBbrDSU2wYU~v1m4TT-bq?EWyMJ+2K})MV=gH1A8#VRH
zNVe2z=coMl1+i6n)-b2#PRuB38y<P^i;EG_#p0JvLtd)Jg7Uvn43Dh(gSO>|dPOKD
zlz9!jZ>7Mm*&n$;#eiW2RVm)MeW<ZrRSAIwy2GX38<9&aRNJ)8gn;)FLu!y|Z4Hb2
z`W0lxr+_-_GTdu1^ukN?-JpnkynLu~t_C0DLZ{6mA|t)F@R^=H>(DEeBOt`VS^PmL
zPtakLR2RD@N-fcNf@Yec+~j#HQ=oERVZ&E4TUYxBFze6M67mE>I9h|sPuB~6&D`>{
zOCPh5Qy^=E0MTMs$By5Gq4$B`B*~Hql-)`plS{;g@1LcJ9vmN6yjA@EE}q>aN}SsE
z&hFx1$;8A2Eui&79A+k6sDYX%{}aHFmTr9wrkA@Sqp7D0)!5ckafEOXg$|y~lqFHD
z-?$4s5B;-C^E3sc(&*5o<O}sL8yHrlN(W##?p}5(oUsN-@ID$u@4p|Gs>hB1oSZDM
zvc6tt;3JvM%Ex!x{e`Nfr3I6W-MW``PUK(c4gv`d4@ZK8i!R^L{)^Za2hx-NLu{=o
z0I^jsiUmZCZ7Ss{QJ6`>#HRfN+bUHxFIBy`oX%6oV<*G>tEWg%<M~p~#)N0kWh-nx
zWFmuV>*^1NhK85tbwc9Y)sER2O0~Lefe(1cK74&J^&X%&xW%$TezV!?QnASgdgX$<
zr|QbU#DtZ=iIR@v8B)h))B;EYUu|t`YpV>K;ojq;)VVi74ZLnLT3RV$DnSp=kyoRK
zl^OQC+Cze9L2?vtL}%1MknsI|Pfg`nzTAut7#C8-wyzR09_pA?uu(h+byvTb?A9d{
z0Fw%$<Iu&h5qevbvkZx%h0_maw;pJKO0jnb1`)buqiTie`@y1Kd=Pl!rvGql0|cUo
zE`8Q%Y=IjVR-T?F6{ib_E(w(Du}kKfnwm1B!4xi=KmCcX?k~5-Oiu@FKK^Fb81tNN
zwV(<_K}NniHfZGBo-D@Rn#$8W$EJ}^Soy0P^?3iZSKm3VUS%LGQs%Ya7(!sBsP=rs
z<K**W9|7`-BDvHq$W9=lZE0h=Z2d5W!bceHMcjwWySkyFA+N@Hg`<hmSx~`dDlHP#
ziyi~YlbV`(+zFPgCl$-TcpV;6_`%LtP+&dpg?rm^nbJSIcGeE}TB?JCJ|;gmH~6o7
zalk+*+sp{V2Jq%ku%Pl_6}oZ6lsk@X_5Ok^Ty#Bhb91w6+()j6$3IZ$IwD6V`AB_Y
z^4ZImrN=i&<U9@^@NjQ0*HZ;*8%j!WgoPj9QN4amOWVbJ`EWQNsd7CY**;y|Mu!ZQ
zbMH(c(?+ye@z(vyj%=I&D<gcxXYKh+`W>BiWBqw+BTZK%MUc}%khfqisoS><baeFO
zDlrj}iXJZrt!(r7ixMkYfYTbt08x-5^$u80-<zvzOW0q!czJPL1y^ClNPs>u{-Xw@
ze=#SZ?KNTC<I#NbvB{)W`Jvk6>B0QngRdu!8T-{T9E5?$`Jj1Hts6oYdIdxlqv)Ni
zo22&s#GBjOhWNkaz`M)(wIORfKuClH1Y;Q@VeReh)zw@S6sd3s$gh8QFNw_m9_4o0
zn|+oeQxz2~_K9Q|R0^YgZ?^DS<j5s!xW{cvWUm9;keo4sXmysH-{mQ7ow_FW$A0-D
zZ*(9>a>Q*Ms<C_J0U)ndeRu3Hef36n7d<__<<4*vOtQD~W4-sOlfeQ3e#_N6JLUoc
zO=8g%I!o4=RsskB)gYq>p_f=Y!|y3PZ}}BXn{G!8DqO|(<I}_bzCQc?BWCT!*X|bw
zfD3%j$!R(n0Dw)iP-&l(!&B1o4ipq^6OoXRkjY|gF9e=dsZIM@fLXd^67VL9y1=3~
zJVJS3$Or25&!Fh&es)d5VX@6WKgTc4e<WaKV_S&lc_bQ%<hOJdjB8_K1482J?s}g)
zVjFN|F$#~9vD?Yc1%Lij2*loF|70Wp2o6lEp`sD6nlDsB>BjXR*K;yOxtTqUUnNHF
z&S`v2na{IX-m_3gcXn~HCpModS+Y)aO0&V&nTGo)w&t+6vkOR`{~ZcZ1RlHDg!xzq
zzEo!fV(bS2_it|NicQu_@Z~1xcW{f?v^@bGU0vszCD64#l)BySDfw4c^FR!+c6PqV
zygm;N4X$%Tp;z)RGb(sraOV4L{rB#qbSz>ScZ+#L`Z$fp&2wk~#2R_=-5S2pj|mH?
z1Nj~Ti6h(dx0Jy+Xc+;%^iRLDQWZYIyBxNhF=+d_1Gv&as(>~q8esX{?2SGv7X<X^
zLLk(r!hr^AMlBCS315{<lGEl7z`?YxdcOc`@fUD<Gvl~e(l%E&0Kla8W^}qnB!_1{
z2ZBId?OvDdY7h6mfW>ci%d3+jvq_9Xd5fuX6Imwug}rJ+PVJWo#R+|reQU79TdxUf
z0{)v8V}?QKDk~{OQ%cIOnR2}!*6fkE+t}}T6*Bi#7rr)DRY~^g4_e$wyfFWJ9sKTB
z3+`-GwAi&C)tB8ZH!&|j;4U3yXaRO%DHRy$y~nq-m)ugyQDOmr-~x!-xqOx4#x#*K
zn`P08s9xkq()XF7nAHyGc$VI+#OhT@5S9%<e`kxe1=eptz68bIGfHxlJ;AwvuR=9i
zzw}uG<E#GIYt9)(bZmUI{E%-XqN)M9e_|Q<P1hmSRijZr12amOgF+}gIrh2_h8Huv
zJ=_}6UZ~_K=CBv+{^)R>F_mUfv)w(zNaSn#zK}hR0C|H`(Vo$%V1u7m?(>$~qeJ!(
zP}8>dJVBfTp<=tTocm>?JrR19AMTey96?x*(5xoEJ8Vxp{y+>igm_VnKGNg<_8f&-
zn<*l1qTyp#aGKzAi5`j2C#a+nk`hrg%|Th+_nu1vKvs8SGlG+T^}Sm(oUGw^fET=q
z$0j{HR#3q5JBAz^E(``E9W{hrn)S6k0h>QRMNzD(5PZGe&bN>+dT(wq8O)ZG&h^oE
zBJi+ZViOHjbo^0$xqMT+fM3VR6#<0pq;XS=kt|ec0rOZ?R8&w$_|t~53gtZ}3e}MP
zbl3Xg&-WK1Lqii-4MLSE>%s1GCNRhM#eJf=#Sf|&MZ|qU#1SKzAl}hgX2f}SfY4<F
zpyc*K)BW*)0x6&CN#I&hny`qt#Gc(s^G}0b$$Rx`q?GJGWb-Lp-bQ)V$b$+taNuiH
zkfg7GRLMF~Y%!Z>cOA&6?JkQXf(z*gKuVMIRz^oe6^0`qQifsxdJ?&3xX!hXQW?J*
zih=w>1nyS0iuSn)y#i%!GxQ_L_xx{pPxRpLrdfe1a&0#kDA$0T#LLF;;$zR(=aHVP
zi`9vSz1*3X<t^3|#ywR6hr``J7iY(h=C;C?JI8Y6GuvC`<>Xi_XR8H{j}}f<1`}WH
zpAKRN2DEY1-(obkE|D?7n7{=Uh>7VJ7Z;6$8F*o68Bx(cxtOwqpDELg@xYK1=dPOS
z@Yn~iQ3IGE-h*Dn+7vKbV#vF-sJH<Z*}26xU+|*7MrD1XCM0-H^&C|~94d^QhJ*GN
z@d=g0O9@u$C$!?EA~Det9y^@zb=cC~eSQ9S2yhTGcF82S#b4U;i1LaI`Ok?k6)xeM
zEy`BEwmA3bTxn#EzYcwQBEW}Mp>4SCh==-q5Qd+xkmq@GD0Yqvc)pTo18Tq<2BQ1p
z0CuZ+_v!LqYOa38F3ZjIF$x@HsS~?QA8JpW%Bm<i2=!Sv=1?Gu2Q8NlrW_`}?R$2(
zyL!ahLY7&a>2_uiUXG`^vnTO$Q@$Q^!GJ@e%%Ou!{q($Dvt7a=!h*24iB6nZ@WZzh
zbb_{ra=~e;Kwpkhk^1RS$nsD<($36iiEbP5>*$L1KlFO?8`-yfs+9||7A($C@hrh8
z6ZoZ#v6B$$;SQCTEP`|In3e}aD%qt+fvm~wh$sZWmZxE+$!|t4MN3*7C_um#oW6^j
z$RGck29E}zL!524#I;=T(SoLLl6KLGGXSGXr0*E3^(xjCA99PyfJ?h%PGmaSL9m!<
z=|IP+LglfChr<lQa=a-KV-c~0y)tYI#EkQv<MzaWl5QM5h6wKRB&1H*Z;Z)j4pY<6
zC>N^AXDbX>=)GcAMNGx=9s9b6Mv>m8>v@W~30hIBy8H$}r}52}vg%8G54a9<<HaAO
z0+z&nkLxo&<pQQDbH<Z|;72xKaT<hDp9P3oKBucSVBKe6m{(JUQR(q3UcYZh{Sx(U
z3PfZ{2}_j`6CeM+f{mrmx)CQfCTCTeNkY|hq4Em>@~pQ{mv;9a|Emk}SNH_P{sEDL
z-ERBP53)<dx-xSyK8f|M(47|+tU)8oC*$dmv5p(;<Qg%S$)Eg?T&1wPw>ObR*Y}sc
z>w2<N<Mm$MK7{{jBh&AEgpv}cF@;q7^DEP_T+CON!bQJJoMeldAwg~O2+gc!rT_p2
zF{>S!{{}k!Z}ZA;-@X}9t*)+S=j5mqslR{v<t94Fq(WVq1^`{>EyqAzns?t0Hd2{Y
zJ2ppiBHtImeON&8UPeOl!iGKIUdTo}3x4Yx?3I-}P1Sm-OeuB^QA0z&U&x;@IwS7Q
zRt}Ic$*N)36o2k|HQ=;9tpf>Rsm3`me4UvtwakTA7&R#m@PX9G!&5h~{dFhcmn`m6
zEupz^bz`4xIeJsw_q>?mwUe~G6_qO^8Y@KNt%MWYme15(GynW6vQS+fu)Ol}az7fj
zzFRW(4i?*;WMs^JcWwg1zQ8t=^n33DLpEdfp?&=BKJnT5rRgYM&|*jpW2Yo&G%9lk
zNu8nh4nPqX7Z;#7)&n5eoa!0a^M0`>+uzGtej=RwRm5yL(#jDw!dPz{vy%lp6A;2&
zZ`-^!9Aqs_%&Bo-MFlrup<UFJIGac&yr=QYrTfNy8m57|ZK|Fhmcs9k1aMIk=0N1h
zH#xc+_mP!tf{!J?NfNf6QZ}N?=Tum@hL*o;bSUz)<a9YX@VYyufMUScVB;XNTiOC(
zhnra0+{b+|rw+yNMaG1Kz{C4|r@UlBL~2_LAIj%|2pUh#xBkhQa6aFg19!!u$L@Z&
zx-WS;U1k&niY)tcm;AAu`NbzdaS2efOIx5lSPiyny!`$(vemR=Z>DO0V>ol5|J1O|
zWo3<PcjSikzq|md)rj8*rnV@`trqwvQ<A_gp8QPE`6c}Am7B3@O>S<k(SGLlyq9qK
zhA-lLZ<h%4P2T5M)+U~Lns!=-4;<F=+(kW;y5Y6onyWhpd#XK126Wg40?VZf3Ulk4
z!odE<WY%AB=BwYn$*jT`KsRb_CFaI7q7|(O?adwJ%(%YBxY6-((B{HmvpG9gOb|#H
zoRD`ag=P~PhnbKLr8^(KyT?+{ewvspmpu;uCBewPgMxrZyc^JrE)a<VUy_B-{It{E
zV1YnF`Rgv@f{aqpNN3pW++(`+O5JY`yTl5*e8)#z+5$~RzXR0`<L0hT>)spS=E)RN
zUrOa4n%ryid<owv`hC?i1_j9N_5X0?%HeWVPJ1=JwlI<<`$UV`DVLrWS)5Tdk~3RC
z!KUFV7Wp_i(UXFRiCh0AMFI=w{gBXL5jtQVn};VQTPlCfdPlvkcdNV#wK{Oq%2anT
z$s8_9jDkLWU2_fJ2tf$9GIbE!%YtE{@#c+ha;OJ~_>AVbD*doO-5Rgkf&>>KdEL;A
zn_!~H3hKdYa-sbCbto2zh=jyfBN}S}2*8PvAPzLZ!jksx=6FzdzY!%mr)8oa$EzV^
z<h;BO#dOS{Dv0^+zyk<WQWUMjnPP5_d9f%?;329F@KMoAy-NqTuZA+7=^M56lP_*x
zrwBK?o|-OYh78`v<1xz?Pj0iZvSwsv1_<l^?5M3wGcFJ<uq%NG;=V<XU-1Z#6{}HN
z@+l~|0m{5K>RY*DIH1)}OT36XK0a|u3Ti{*!VWUCevC!D_!^c1m{_S;K-=YDMn;B2
zG|dNI_YI$6x-2@6u-JYS4Cm{knkt8mnC!?%osusrA_~?^eba2`c4xd?nf=HMw@{Qb
zpJAxQ@8<Typ%UDz??1?2=Yl3Tq{L}|9<&K0vPVGY2ym|w2DhUXiJ-P&ppy6q9$fD?
zVUw{Dt@Q<U(_Of_E^Hiug*97f^Z-!5?2Q*VY}aSI=$M#~*bG0Z7a6Kmf|nC}nvu2u
zfq0WDQqmzJV<XjJDl_=?VW^g4h~s(A&2O}F1HB!zBp&JoMN08|y(Q<^6ayzBvsZ*B
zKW?1z@t#PqM+MLH=v%CeNPcqrjST_CsPfxu!oYC<`dLoiO3Q`DhIZ_SMhr0_;rl#=
zfV-};(wcv`+J93vASI;G!(;*i0ze8{sIj9p88Uihf~8=xk$$`XR!mv3KxtK2Js<#C
zl_K?(fx)zt=y3@B_#lkiac#7v!<M3CMtAH$70;Yg!}lE)(j^Wq5<X2!xbrveB473Q
zK%xb`1TJQTI#77h7w#c=sC_9Y<P$zEkeBau&r`ja)ieQJ$qBpz>fP1;95w>V8%f)B
z0k03BPJ&ij-F*e|-W);zvnnIsvkZ&8bDe?LX~wEt*e9;vxXB?pD{(C5)Nf;g`pH!$
z3MW7($LlC7MwKfxEDVSs$uTkSfy1b{@oMW=(en_Z?E$^r;*Nw*v`ietGz8_b5*oeK
zR=={cYxKfebi`$4BR_p&IApj~%#rEn=*S!2T51a_P|7zF7*i?M;=65^fTEei#2k*w
z3jEUlsp8hR=N-_7q{^k~_xJJ|4!!H+GolDFcC!8&yY;n!WV3GZ9Pyt?nz3dq^FP!x
zO*=MZP$v`p1KD#mhX>D`Et!ZY<$rL!VS81zU#+NQUp;y~Tf_D_5&ixZ6db4cQ9(-o
zkm%+mZ<jK!t3St;7BhiO=ll%n$jhQ_D<&*PG=16_5Nl?O?&)Z$SOypN6QH1h7G{<C
z^v@p#8qi1x<Qo^0)g%Qd#E-3yiR`!FLG`vgp56o!w&G-=nw?UP_J!_X7%{UYCaT{7
zG<XK%m<S%SvW78rKM7hbS~ulDBR${y03?Ry=4K<WUKR`?9D&;tMHv|xQ089%kqVM2
z8h({6grBR05b^l2)yoJ{wzW+I`1ZuUpwmu2o;ic))#1aef&Rxu1|t)!CVh4T7B%*p
z3fILnLNO%b?_SsH%Y$Q*8du-$!UY1T+ihRK-7@mg2Wn!#H10bFy@66;4M&7E`OihR
z*-unTDqXG|6dcf;1g$ba*4BYQ5N0)p*5nva?{Un~R@mU0$;FkguCH|bp@}0r)LDvN
zL8^V+KEV_+REY#3D2)g^{m4Ah+WV?l>rZvz1{w|?yrclco9C1U``)%>8jW!t-q$-_
zyIEyf@F`pn0v(!yBIry5^@lVN&Km&*>qM=b#B{<pnpTj`_W;`6Z?aB>;9WKEimqIU
z_yvj)h=r)q=zh7eu>rP>=lFbg=DcXZGb}Pv;N!~=(h|s}h@NWN6{i}{_*H}2zI^>^
ze|546UUe39f;`)3(w--b<?4pSeWPGEmeYLe0>w<_bzfOI+XcP@W(lXI(H01#f1$Qf
z|E}yF9d0q{l}A>S_Ln)ES)JEeqZAY5M>`SAuFEzJHashNYLyL=(~?XUG+#)5mKhEt
z<&ACvI_wCx6r7zbRyrmv@IHZtcpzC!)%)|nI4n%{B>%X|(8%Z>Xi&h~HCN|kloDV1
z&|N-W1bITMJ|OcgV1gj60rj-W`7ncO+w;2&@!oPKLD+=7aoN?#snZLQ)s1lJH+~4&
z3R_;-G2Io@#U$@kw#OplE;Ti-KZedX4}5@@843FZ;t(J%{6MaIG<GM5M~U~QKf!uy
zWkD)#Wo5;kOOKjC6Mj{=#gI`zKkfJgXe_jyot^OBGC9|vUy-F?0@5kcU=qep{a;Zv
zukc$IC@L*|Trtzs9k@J>PPX0yVZ;5%5+pU?E|^{r^i8EuDuzQwu-`<3u(;@Lbz%<g
zc^~XgJTe^c(Fi#hciuQhRVu(@``#KnELY6GGtDPLO~PqDmB~v>O}(?wB*4J1({MiT
zc65OnYj94%HDtAJC0EK?f6EDdydUrcZiRcAEFPxMI(4nm_hSKH#U0&|w@?e!rKX`*
z3=_un(EkOd7g=dm<v&EgCwMX}Mv;sW?rpmYSXDb?;<%~9bL~A8?8i#m*Du}2h|$94
zYur6S(a64`NDa{7(geN~P0-|cxc*HvOau|M&#0+2Ywb5DiZlQ=z%2bHtUtog@I;*E
z*9hr5M>lUJg*?-DeDPUY);iRuC%R~^u~G+$UA-%>n6&%3?XI#$1gzFHOQsi(2N!cN
z1*1MEjyj$1^Ca(KXOe<i6ho$3dAM>2@3ZJE^rTyV{G=9`z<ABNqJdCgxAv(;2UsN3
zoFlOx?@5yIK$X3)sZmKc5h5w#fsBwDWDWEse}Dge2XcTW+9v}0Fmy<a$B`jA1BDtd
z;y_ZRjbkQ2BV^QQY0YNQWsn<dOXe`i8z&ybMneP2&)#B-9~&E6TU#3t@`3;Cw@f_R
zBi^>sObhS1p9W#apWg?y9d72N?%}dD+fd@Q4BbN@?b=pkL<x^^zUkhks=SR&&X$!8
zI&lLYi6kzp=f6OBtut9uc`9V1$y)lkC~@K8<2PNdk%RgTq*S#sLq3wLXB!R&UB$E$
zMFqf_x~cuAy1JVAI4s@Z1;o1p{YQTaAs(K>)%~pmNckZ1bNq9`w~h;)0d#hpd*&5n
z%DfWPS}E!%V5I9_J({VqilzpKu2HjIRoZ%ACEvQPBv-yBP13VR)$1OoMoAP8GZdYS
zfLPF**?t!a3Glj)p*ydc%hFQV+MVGK<uyD$6u2K6g(u-gUUSD3rCQNXMnwy}!c5aM
z*qFKhF0|mdb5W0gy?*+0W*1RNj(`Jbm0am0?jDJS>PV6EHc5<S)zCd^lZxMw0(Z{Z
zTAqY{cglh|_AbOfwEU`m3&!MI??ByWDZb0XMWdpjyXJizsMKYRSB;I4%!ZU$&hfCs
z>+Fc@ylv)b)`u@8v9+sn+~v`04+ik(XimVJ{{7-?yc4}Ldt(1k%?qHW&$+>5q~lq1
zTOX7YJsSmuk;e{{Ae=iR;S%7>Nd1MUYCINcZ+SRPzL}O`f4E+bnc|_0hG!f?2@<r1
zvYR79<O(=fR=gEZKSAl05*p5~W%-};JMO6J=;(mM@KD--AttEJ=3Rl`p*Eb5WhytJ
z>A1@j$<1vfAXvbf@#)hippaLXj3PtM_7}WBak)ALtqyIksS@3~V>ES%w`_)e!66~s
zUUxSjJ*R;TMf$7P%`x8@QVrZWvo&_-X7-s0><KKo@LJ`lPyCatL0P}IMi(<FVIZfV
z@B$5+?r!nW3{Zb!Q&KF4w{3rD7trbHA>0<^-DVK*(@UVzYICV_PU_SvX4Nb&?x6;p
z=a0W@ic3jJjGOshl<6uB@ivL}DAqnbJd`gOk9!VhJ-bnrz}y%QU2gK7^J?9UP8$`{
z!SNTP^PKyEijGnPYkjJ-ZwX@HYzH|RYmE_JTf5pBj?>CNB3b8tKG*jqfy0c`c!-jn
zz5IQFvR<h(@cp98q`*hP<VxX1<jsn;`~swWmKS@|hjL$PHex79d<^mcjSEszQqb_`
z7qG-@w}6lG|1qaCt@pV%UBiimWNFcp%K;uJ)1OXsO1bJBG%iYBwOjbidF3Ml@uh}C
zuVtahLTeAGd-gmZKs8`iiSV)2{Tw&5G56!S@S4wCw84}Sy_bH|@RyhG(T*(noFR|u
zn)9NqnF+7X{9eQ@ZRl2~qN2yaO9;=dNs3`VJKaQiueZ3KTIqH53uxC~tO@CAcB++c
z-z%vej1%N4VzVbWE->uRH(W1Lcp=sykRbbGkYFhCR$2k4mU@xr7QDmVE^l*rJmtAh
zKe?B=(*2y%0;3J}A`##O06+ZeD%;$8IAul$W3~oG7o6`ufAl~nqxCv|?WCN?)L}Qx
z8yfKXEe&Rr1YO3;vB)m$T##B4MKb!EvF%itbgbf5Gs=ErYFjWxp;r99js*a^?WHBw
z&mqkdjNy*A^u$o6wH2c9R75c>AyMSFqAj^EOO}j`Cm*X4O%w%G?k?_U!bwz-bDFCb
zsaF8YK}uGD4r>Di-bvy#+A<A%uW<2W77Lo4DYQ{TI7rfK`Sq$@YKcl<#%d471xJsp
z)`M2xNT_R|<l~EX@o+C<eMJ#2U53^_ic9!(iw|uJW%@Yo3(gYZpiypeVo+cbAcd_7
zbx0nhE*&hBRoly(mwwaB_Z5UXYzd3<WM%s*O?ca>N*tLMa5lqFSq%N;2!4A|FTnxB
z(s}eb9#tK)o6vfj=G!}dr+w2^G!j;Vh(|sPj%d-A8snx@)r4z<KA|zX$IJ)It}JAW
zQ@FsO*}Zz8{33S-LdM}+c+KL4l{+Sv%}QXlidV{73%|ZyK#X?!sr$Qv-?9GOOjYM>
z1Bd6$1s=ZczeBv%!*5RLU5bUu+dVNa08Ukur~V{&G0VhYN`SY+j2m>_-2tyoLPEln
z*5;v~uTURC@cR+A`I)b6`lYb0?w3GbFD60yny=A``Dd>@r&PT53inn}yQWY@7*G}p
z$5vMD@M+^)srsmq@H@w9;-d0K`8MXTuiP;pEIa7AD%e+;-aF<~l}Le+zDRaYUc(Cu
zFIFW#Iu%hyNmS2z&Yl?VSZA<+TWw3a)*pzKyr@Nwi2P>vw_MI+j4MQB4BznWCzYIE
znX$d@kdy0#J0z_S;eoDJPNrIqLx)yQh68QBwC&SCjHZS3g8u0(il8o>`4!R@hc_d+
z?v!|S40QK*+hLOrb-Jm;!TV-EO}T3XDtLis-5!$HS!Gxyyv>FPq3}@<KR6O(E=U5h
zueVA<;<#Sms?<G<=fR%%?C$a3R^0!jch2v*9#4ZZ8i+{_d{JYD(Jhwfu@Wz?pAXIs
z4zMNY!wY>a&InbXe$Bw1mo^lH8+h4$+o2aNPLQU6#0_J;$gz4UnwY%G)gWh#Ck3NU
z?`O^`Pw@S0%V3>Q2+LI)eLEc68>cUVWc%b}`LsWQLFG&~d<$O@XU=s{;b$MJpfp=T
zE(bjLn{Wp*RVDgrH<Ulc-1+ZEllc)KauxYZe53Q}2(Z`2yEBu<>}{1?X(r*nx_>{?
zsp5qdS{|`iyCAF$u!@N!t}*)NMeDRue-_u7uqSy*hK8<lsSq3rUjxI#_g<!CMWcTo
z=nG$>zT`=BvW4(!St!TvF`rKcI14No!R={hDDcezq!YW2C92JO$kC*qL|MoLa}fL<
zhoYq`s$5u*Yt~H^D0IZUP~o?N`}Z1OV*R>vuyu(#MiQtL14-OvRtt@QB}#-}DK(V3
zSW`mU1-gs3icN_L8~0p=^JSQG^2qM;-xa}mzgVY2|0;e<{WL&M5H}0E&Ay1h<Jr2;
zTRLczBqD*3coyCl@g-}hPqfQXf@w-xrb*o#97ESj`0#!qlTZm?mLd2=AuB>wK^((?
z*1~b$Srjz%eg-qN=P2j<ep<98Gso0;Ksef)FC{b!y+JT%w_qZ+>>Z%zBul^4u%->+
zAPc2a7sSCKLOwj?e9|^iBu2E>dw0>VMWvh=+i6;nri>u*+eFfe-KP!Ck@teW9~bw#
zcUsbdSJ=`N@HhcGUo6SL$CEIy{bL1v+#Jb9_IvXBwR^KSob2tHwgOkmqpQ5UJkZ?=
z1vNN0c&g6HN>ES`2!|EwyTAy7{fI{ic*Sgz1Sv~qc5&HV02IoXe1<!s;6ph$cn@;^
z=}MG)X3dnIA-6FM&%P|pJDm<y8l6tTtYStjcDdU0E?=)K06|6+9Tonm`^sue<MR%(
z<Z-gy0&D9e?6!-L_@~j++VsQU`XP(VF~<GM+PH5Z8-<lRHA+UMNWEDmWZ~~9A-*`n
zUxUA&5RpHRM(9FgBod|1M%CGIEcGT*U`K&<`6vlS8=MhDx<_vcEANG?V*(GF!k<U~
zVcQEI)dTiWa#GT1P(Q(p2ymDL1_rwF;xYqQCUCcOUzXX)*Z?du4oo4SI8T(FPi~(A
zO#zr+2t?qTfqXic%$uW-_s|kN{3AloG2Ych%gP$1OA;eVh>xGCkO%A&)_b$uM%lf+
zy`ykR9PvFIbd#A5+h9`wX`9x}eZj<}Ok!^?Ctjdj)Y{gz908M%mS#)rClCP&6>#?{
zXuy`1-{9&va@n4C>eQ4s8*@&XE2O6l>HkPOksdXsa$iw%Md-EBgL}jfL-eh0Sh8cO
z{8KD$dc69FUBUaWdKTVF=0ADO843tqN(6Mo+_(o&zn9;1V`MH)kX&8lD(Vnb6twmi
z7mVM<O7x3C4X{o3rbI;(d^_qGTfqgB{1UvX`b7GRAXGwf0Qv|fh3oH~0iP8GYmlcQ
zD1oi_Lb<V27hw#vl&B$spg}EE0;9!+az)tPy^GHJoWTWw_eRj6QxKOoEfuH70WP>n
zb^;vX;PVHQJOFgB0MM0;p?5pqBMj>R0xP07uv|v}dOK=wZ(pu4@ieX%bQtC=y)Hj!
z^L`Q6=S<Z0Y?_pzBN$<~U?<ade+6t0NlArrbU!p}T7NPK;V9%OrON@t4xD-M?|^6A
zUpQSMuig`Yz3s^W;dJ42xjZFctpMGC&<?#Xy{l*ePwD9BczAfYxVV^^nW?CjK|dXY
zKY@rmfdLCipJ7>Qsy5KR^gezJ3k@w-XKZ5uJpka%0Uxj~z)EA7wdXCXh>3^{4GaLw
zmdzYiI{q@xIKssh0nwrB(S9oPG9iZg+oQtISulndxI^p`-#0PuK7C&1(*Dw|Q+CPH
z`ZSumPb?*GG^ob_KD1|S>+91_eW&tpj4=6Y3yJF6+q*X)+Me-00W{vVAthCt$Ax!_
zB=CvEh(4<eu!jM+>$ExWbAfp_;P716-~&!lacX#olL3{40#jVw->D!v7*N#qfNKtj
zn57@PPFx#~dG@tEPTz})25K^Cm+A#*ObE*p^!~hY7Y4nzbvIHW9AF`{oAkCDDH^Fy
z<4o-5#;+eY^*H)nNfRx>KqPQgbR7HX6UKV)6C5z?g{B(Ik=Q>FHphlxcf8hLU;i3t
zO@Pk|>_f01X(qpytQU3Mo>5ESA+`**bk5LXLW~pE*3an{506EosJWOMR<Ia)(X=<&
zt6nLAQKhFblF19{VCU_8r6Awfj3SPMymnv0+~xkPZn|W&o8{GyDQcLd%rWBO;2dnM
z8=d!ZTKqm49t0<IP$LYul^#~Yh}H^){J|>&IMBH1Wqf#*U;l2Gwaya6J8C`ZPa?3V
z)6mes3+n0pQB)xVem;Mk`<U?7VL*9KPEJ5J$R9U-tw#X9O>fuJ$wU~M5r|E3SHDoZ
zeWg0^XzQl_H&K_(4TdNVqE#6jL<r}mZw(A%B=v)d2{4BC+Dqv2;q#Q<57QSnB-%MY
zZHDL`eC<8uKTD-Rw+DDi$%wGP1?mF(&#Ua90`PN(1P7<xxy>CAYu8H`Pu3Y^3#SiR
zRvp{9!sZ7E#pMYkqHdBH!v__+STR0Us9(loq9meQm(&DS`GtQsfq<4_6!o)zT84)|
zfRoV)NPVDT%Bc2pMSjWouw!PHW3EA!fiwQyjXN)xy}POZ<8Aj{e%{bd_xteVo2Oq?
zi<JLsGUM;Sm6g%MQi8$8s_BbMUmhbehlg>D5^D%HdeFaIkl`0V!^0%!MD*SoFIZ@B
zn^fbP^*rSCrA#V{4HA!Xx97&zN;gRWb9{!GbUiM4#1GRRb|wG#pO2ANEjWDyn+LHS
zK-YJj2n)EMA8y0*F1yLI{@NX62ynuhK=Up76>vNm@qvlWNXIn5B|933t*HJQsm}(F
zhR+^r*?49M)HY)lf4%!-QA{C(1LoxZTGEn;t(Mo_UP(mN-`2uI>Tg~9*ISV9iyqy>
z1QIv4yB3;0(=K()IJ<>pzS;-RMA546M{+*ooLuaf(ex&^I`$tSiU7gw<>eJA>H@dk
zF_W*97xnGiRGE>iRfpicEifm9g@sW;{{SnZVxCeHnA$KkH3f_p_=WND@sZ$Lz~lfh
z^{R^qDj<>`DHi~$dU!koe8LT$*^Ds2atnZHLYwM7lKv)WvZj~u`SZ<Y9|SP5@u9+W
z9I!zI1cbvs7FL|eGF2|ehS9&qGR3gh%`8e)NdjHwZ^K1D0t2wMwKWiSjg5_UbQWI}
zG4b(vY>njsnJ;d-P(;u_Fc8?(vqZv(fY!tZ7<u)a?M$a6HD0U%ur7%%$gpt3miUhv
z_Zw71vd7dcc6-F$T{{@%YR+@g@93Q&1ZL37wMSSccd*~Kjgx~e=>_QI+;6gh!3-<^
z5iO~XC2JTMy8*Wh+*_{u4d++q`xIb)YtuU0zfP$>SeGO;<LKx}%bg@hG_wYHe=d(!
z5#Zs09~T}To&YVNMMsB(hK)_JV0>@7OpK1;Y;$xSjBLQRUeqWm-SlUhh5v4dc3NV9
zdz&#$&zh)8(E-*+;Qq!gtRpHZsU9@qrQ(=S%UEwZMqWp!*EU`4gIrcxVK$bF`{)rP
z0h*tmUm~YP!nbe0SoxNok3+=@N9=6{h@srvZew;<kiME5$uE)mh2Owa<Kap_7s+^?
z`CKt4Qm*^CuQ4&94Lw^yzIq3o>du*abd88J_e<%8$<3Q~J*2i^I!UXt%TSU8N`3ez
z=t-)A(TGpdO(50ZlCQ<&+5x1ObQcERcHoxWQ*e1!{qJ7M*+wame?~Ze7ksFfE*Q52
zBPcf)htv|W%-WtEp@g7gb*fpd(Av5r0tn=9mXG=B41rsvF97hcpG{IOXs})aU48}n
z7;wsY8#d#JbXPqin$&cR7SOP&6B4_At>_{;?FxAo=f~`s*t2P{S57_h8_;x})Q-1e
z{m|RLulZz4Q+c3uqMz5|x<SHN{GRM((`<Eim3r~UCqfb-J<lco&eI94I_^A)+oBMO
z)bae9SI^!HMzN-+Nd+8rvF*QtqV<iC+Y5_YR$PR<6@-X<j=B9B@9N<nSvX|=K7<5L
zgq461Z>YSkc-2`uhr(^8edztd1kkbDdH=mPKxeTu1)60*)nU?Z1VYZbR)Z_RUOT{D
z2)sw)V2a7UJq|e9C1G7^3`AeV5h0hCmp}~yg*;cY<_LI`fp=Mz!E^rJxKtGpz9rfS
z#h5(-EFiEi%uStMYx6D4*u8S-D&noU!X)9cPWz?*U~+I$q>YR21={QD%*_<#S7jp0
zFN0+LFT2%6{GSmL%uQ~SBcnVcHd7A>O6l#{#Cb&eH10=Tirg>R=UH^w!C#<AGzHSn
ziR2NVWEtlH6sn)AnPSG?gIKOfM=c>N?501Nl5D_;HPKSa6Lt!R1JbS)i#U4%R9hsa
z<(sA^jP45!At7;MzX(m40(B233F|nVdR=DV_X9`WFU)QFq!>H_w!bIK$T>_;>L2`?
z|D%}#77W~m4rrwSq0*{qCR{XAG8^6-jAtD#;m9z+7<y@FX|FFYxwVN4#$6Yi;i9M)
zJg+U*;a&lA7*H>0_Ot0p(pf48KuHAD^6aH*TuPJssZh%2Ea8=rI^^i4rlx*Fblu$=
z`0wCFb#-;Y^ML4~G*iUMmpu7jUVu&Y<YLXo8Z`<9$e8&eB3c*Ms$u~<n#zvc0E!XN
zbvz~C9{N@UV9cjE!}4iD4E(&$>m;9S(*><6evu+Wppci+fB+>yAV3)Gu?7EZ>GV6+
zjq9#MpK4f9`mpfs1Q@ctQBW_4c^VA!3rU9-b<h(*6+*)_`~!Y8rF_mG-`vGQ7NpNI
zLqYCbUo~MMY8Y2O*i{b>MvU+~{HVTS=vb`zvNX-w<If|-nEat)ft`8j$9(M*R#5H1
z0A}Pj_p&X6*s)9rN4kTchXZC48z@<T>;x7Y^Y7*0A^n3m3J1+GaLPc`frPD5?`(5_
zcgy$a$Je{nfHreZudB@*LiEQgBp43Ym=`P2vR7xj<Z1_t-3_12qq*rHKDM?(mSf&~
zZr9b9ZR~U?#!ttnSB_}CrdXmPM&A$&sm@3pF)q85dKN%q_ep(KN9~D7lpt6?-rk-7
zVK@=A*KC(QF$UZbqZ$^D5f!{suQN-dBweb+qw~fIzcGaOAl-iXssK}bpQEI<TpU0|
zC?4Q@fr2El+Dcu8_lYy;6Rx{lO>1BD`p27FCL)5#qnI25v$#qRv!(bjV&7h&y0f&=
z@q3HJ&rg{R1fb`qA*U+Xd5UHGsn>dAo?>6&npsvZ%XSh{DQn@|EwqRf^FUV}NVBr<
z-W3AJs;%u&g1Uolmm>>li|HeUrW8%4^V8GhlXKKub<BuIpi7kT!uuUi^3xRFEYp<O
z<J!Am3k&O<r%rlJ<G^4y^(Ybl){RXs56IN(2nCfD<T@Ax^pw~wXSp~y#@nVPw-?m7
z9D$bTb$<s&^Sr$wt6y$C@>SrxV-p>5mcI0JZHL#xC;AnEnbu;}ZHP<@i*OV}u@XkE
zy!12NY?Ja;4?-q;C{eS1#oQ!$Ay&|Y>9Cqb(bHDP+1j?BNx>P9Nu)U^bRD=70s>)@
z{o-_VNei{pD^0SpmcP9bR9z19gEIyVj5JJiKfU|IjBC;8qVche4TG)Ja%X|`{CQOF
z_aj~h0p9TeCe7*t&<SbMU-y>5e(23o)A;u%KYK-r-}nJt49l+#Zg<}2&2=xuRO%m!
zAu0xYP_VIrmiyw_G}im`vJ~zP9E~=HP$4>-8E_^}@7|j+sFvse;d5;D_zv$77LBmR
zjO+uU`Q%Z3<9ic%KQ;X@w}jsWpNY%ffkS}dg1uOl4m=y*-W3rIOoO}%G;e-BN0@?Q
ziwbIE<7ZFoqvI#6c(y-VEKw*ZzHtMK;Gau1!?w>)-lUwr)BE>avujcc>q0_A<nmH3
z80QKKQ6UNM1U<c!*#@bnIP931OcR?9!~;eBJ$LDEG$7T*qhWMC-(nMfoP-=4;*L=F
z3N?IA>wF?4-xw+$a5LQasmBvL$dg~hNGfaEkj?Ey2m+HFea*LIsn0lcUko=-boq~Z
zxv}?5%Hv|_J+5Us)RX_zy+8Fspd#eI!;?r5#m_uSw7Dq^HyFoz`9_{yLg`%P*5OrA
zzWR8PWWXu!eHK*Dv%rwzwp8UA5g{^SS7)cp_E;X69RZykn9CIdgDMiTpT}6qobtqv
zqqyGHoc=YeR7_HKW1KFVTV*c<Px=xy9|Iq1p@Pk!^tOcRqO`N1f;inXcS`7D@Yi{d
zbc?!2o98=<IlyH(HaJ*(e-fYS$w&UUU*!9^V?6%@z=N=W9z|AGR%@&9w<buEk<-lX
zXLsU^7tQH@!1ntA6gQ<@#14_6a_stXw8j*;)l4`DQxdQ1?iKL;BV)m!FI3>Ve3}HP
z=K^a_7dLcSNYeUe?g!^xkl)%=SX`v+)}3ID5F!?)WlVK2`Krpb1mmKUNtNirmA-g7
zWwxUi(xBJz<7Eb&FVb_OEU6%axT51*)wVA5w<?Hp)ONS(prLFxxKnc6$1N)~i%&r4
zK~|onB&~9MJx2;ghzaN;f3m><1qHPE@aRadRFzgH$=cCTtISaPO_*ZTeK$rIIQ?-9
zY8i(#)9PZ|-C>hdu!IED+<jT+?5<@c@$5!M*Z_47=kSQ-M?JbI$$oZrb+Rq;yjxlR
z<=G=!tYn)@3-w09MN@-$?n93bU{|AzIeI<EjykouY5G%ba{jxt+051NI_xJOcvwuz
zJDez}FJwjb8zCjSyMtEIE7UckMOC|J9;ACILn9M>B}SVG2W4iH{^vWDC-ZF#`~o6$
zrqe$SfTH!7-`!a4Yw8c9js1NA*Igd_ZJ#*iiJDWu{IQpE+onj?2Pml!EPnqPegm25
zyu_s7uP-b#2W^pFt66xVw4fK3nn8Ic%hj|*L_}5nj4f-6(rE6YZI#c7`b6ZQ?CcO7
zN63n;{chm(`QlQ@^?BO&z_0qb1qf3dzaEAe!P=GfE$#1neb1(1C@T4wnCPRWuAcCg
z$!cfk=6gAVhzbP{&&ED7GC}}S_b;C>xEJoKs=^E-4D>(C%NfA`S^`Q(bg#Tpd-m`N
z_s)qe+|RPKqHtgMq0QapWtqLHJ5m8n)4)tDIX0MsU&2x@!Wf3wn*&1KQjiWJA|hfX
zv-#Z4PR`COp_5IuhuSP$z^4_-fOaRRKB4q~*m~=zs@AA)bR!6;G}0*|AT15j2uOFQ
zbhmT~DgqLMw3Hwnf^>=q2uOEGcZVR&ot*c3?>Fukmp{%y&r$Z?Ypwar`KzfF(rv75
zYuj$Is%v8{yX=x~;a=q}U`r*uZe=srGCW=*D~y46qO6<KVNOqUpRMlvtMjZsm=Ude
z%^lZ<{HBIp50$DN1`}qgS$-0#oK_A1%b-cw33!)QR#srtXoRK}WJ#Cy#gt}dvs;`9
zkqD%~xl>IY5*A`E(pQI?Rp@t-0=tPkCbh|kiS5u6OIlyZr3lzfRYlNtLdpd_J^j@9
zI8f2d26b*)36x*IqR7xVX-X3=Dhc3=BYm0-LJZEBUWrH365ei}t~G_~$m?bEE<89B
zmu6Ig`umK5kX)F9uvgF!5pW?3CP$Q4I{m{X8vG)ee$t*%;?uD_%~ak3HW&woPG|SM
zkJ`~41tZ!<x}6Tchh}-6p|GZYw}=`XnE{P+h)YXVcY<0rM=>EwI^YcK+Wmv#m#6uk
zzH)44Ue*4j{n(;{#t@w4po}U_k%1BUR&?SKfn<)Pq=9Cc_A~&0fOc_be=W_R{3cy!
z{3w%%&G7Ym$~awhUz@!1wR>7iw5Ltk!_{y6eso;WPf!KCbmtmS9$R+Otan$bpno-X
z`wg!f#!e@C`{tO?^t9q9zWuAqbL_@*u`xu1JQ0s$!(K0PsJ?@-Id=}juEt3LR3`ak
zu298X*IIotvJ_A<ruP1JUv>t^zBSh*J902!W74dcuat>-+<t-36Clx))KtH#mR2P^
za}K)L39vHn^x*-Nwooa|z+e?p$s%>(ws@@26G29Tfts!QuFW&nMosl^?ms2Z(_8kF
zEx$2RMziXf&~L5wiBntcaNb=Z{MFr>nZ@UV{)$sY|Gj%J?~t(ZIQrI~2rSmTyz#o1
zHv&7Em}&S)#qX5S(9pPft8+VLe)Y6eQD~Rl?<Rc6i7e#3@K55zT;OxBi4Ee!T>9?a
zEJNBjJQQ+W|3SyS=1v(V#+u!k{_d`|Td(Z1q~g8na<4A%Np(GA?+RcRQ?#sl5NLs$
zA4@b}o|V4SB&jHfR?hrDHK|CMt{WHqba$pRh4<W7(NLp)i17isICmv3N@myO{WiOR
zifCPBKYI~fNfrji+oK6QBSm~Cmn!aWO+g$1>OU|>-P8zu{prLfraTVQ<NXUCv&;Rr
zcz-PJKco2|+vv8ky!>)^=2E=b!J(t~B^NiA$F?`!vVxiT=hUVhdCS}tRPUnZokf?j
zY-Q~iIbZe}m~x+LVP{Dpf4^$1Za)KXk~r#bOUyE2Z#B4fHHvM0qkL(mRddL~)7$$N
z4mt`{=lb;KJ$oBl2ZyGx!dI*Y4;Xg8Jy-ylbl8=gtn9cQbknu)1VWMvYG9j*<N$I1
zC}ng4Am91VcU3|Y62?YOA0R|yE`Mk;>y&3Nz791mGhPjJ%`FqWulhI!@586}cORhN
z%IcWA;D62hD(}|46)`QMdql|JQ>SX&kY3n63Bb~nre0AJR3g^a_x!syP|*4%JNpyV
zo9bErZp^!4OwtH#au?s$I^3roS+16eT(=tE+8gL&gc9dJey;ddKWti1KIv`h)Q5D-
z%P0S!yT--EEj4OZR8ms<#7R}}p7DW%oMNEzDWXU_rYRu3s>;2m)DD`$R|zHGryDv#
z{(boHA>f~GR!$Dx{ks~6Ut-%#LfaOoia!%3%G0GPmgeQ142avj>NJ!6qB_ESU070L
zXlPiHJgI(cVG2_s7!W>tuSZa$k1*$7roy#-2(?y7NC<9saF38<fp{oe%-oOQ=m`1e
zvI`h}Wm)onr#>AE`Y0*rOFEh5Iu}u2r@u4IQ=D6MdYygnhPJCXSzVj^)a$}CL5gnE
z+vB0w=PkB?&T>_ORMyx|>iZX!U$%rJxcP)@|MlZyVPQcPaC32)1SZ}=5AA~M*m!~I
zTw(L86$xT*__N$;+?b|H5ADN;rrIaE%zAf%zL&RITJ%KAe&KbP-uw5D>WRrSXoQ*G
zB&DP*$}mq>wJwK)508R~Roq{gkfp7AHSq~q2-bI3bU4CRaWIa4)K@2ppzWj~zOQ`m
zM?_IAo~RI$E^Jw|6GwLs<WPVpL5(asN6~=%GmfURG7b?DeQc+c<4o8KLnpgnBDVg|
z+}S4WL36l|mWkI7WpHFZ=b?v96MHybTbK%2cWXYUMc2EBEK#KUb~@_0Z|L_^tAT?*
zL*pMFN?%Q65fL)~SjVsN!%wZBSGdbaf7pGSR%Wzef62vbu51zA2z&0h6%7rG{G?u#
zfS|&GNo69WZh^wUyfS|)CM-EQxywdvB1PdztBTR&nwx;Rlc-m)#D0VIOv5SIouNBP
zjENbHE?^LD2_W5cFm2qN2>;{ISg)5@zMj6qdBWk9E&A6d%D;>xsny{RBuV5xtlsTS
z;dPwb|7K=7HN!P|#GOgqdB-AxoSYm{M7l*%;rZb2(Rp0Lwy^vA*7;8YFAKJ}_^p=T
zX=I-1B=NbEBFK~XJq4kd4Vdn2ENi3H+js+d6HLe=(&oWei5uPqQmTge#e4X%oSQVC
zkB*o)y`aRIE_Y#%v6@?2iWuPp9a;kC-}Cm_U5SUxx?v#!;*4)TGp4*W<*SsOR;zGF
zNsWy1di(_Y`eZSxyu7@rBMOjENAZ3^ZEe0Lg?*c$y4{tC(a})^h;K(N#c%{(wp12s
zH{qS3I_>PqFod(*)<$`uuALFP|H|`Jb+PsQ+>NfTeqnqUQKF|8S-~!Hd@N+eH9k&b
zHBfOv55S=J@aIQ}3hRHuG#IQDNXBbU?<xMYPBKCd@^s~as+zKTjhkq*E`_*E%ro;V
zH6CeZWl2Y5-~qy@g39F>k&uuaXBwdA3a_uPPf3|NT<rtj=rq*DCBXQT)6$x(20jB;
zF-s}@t~Fz7%{RGG>3-PYdwJ=dX(ruogi`GYb`f^A4d0$tYp!G($FLwUAA<<IJ%WtM
z5Wlgp(Gi6OHR5etcSh!`c7^^~%uFWc*=+FX{}>-1hho^#k*j;*3J7{cHFl7=XqM?n
zd16nB01C@R#WC(bHyP{s;e55iDx&Z<C{Xe?ySL6QzJC2)=TzY#_B(xT1PyAnjRhWv
zgmUBXW#t+rgMbZ?l9K8&<1jLsE$AXxgr4RGMcs(m7|)ms4{rRv_=;AgR+uHBT1#uQ
zpxu?$ylJsxk=_XX%^x;t3L)<sX(;_{gM>CpqP*RdOP9|HTqxE7;;$axBi0frrXcdK
zTtxz{M9BH*AaJ#hH6^*#k}FO4bX{i)0Ga#i`rTb!%q%PfBnGyRxVYR5bOnlvoF582
zIQTxheD(R0)Fxt?OP7}+cR)atKY|CPtJev5wFeI#v~+dZfQ=hk_lYJypWv=D>=?*1
zaNpJyP8V@m0;O~xyt68XiyF%d1{NN&z>l=1U6MMZJboSTBNDmgh8+=<B0=ZWz8B+0
z(uEd|@U};));2aYl_2<s``*D5Hgx*PmH=`=!6EVZ>_*R>!#RiIB(utj3WwLlGj%VS
zWMX3DSt_flq+TF(ohqwIZ)I5oqiNVSLl2T2U+jEI`qoE60=Xspi35h6y`FyK((dl=
z;^ITRP@oZVb8;+9Ov?Sv-GitBH!aX13v;fl5Kgs-eCA!3{nt%K=OXV(x;s9~wrPpr
zL*#wubD?Rpstv50q_eeAiTOU~)xR?MydJ-@zdBr4d*^6uf&mEhxi{_|h>c;z?^xg1
zfP*ICoAuB)Zl<Y9+kf391|Z#RXtel}n)>JcayL@H6_=jg6q|4OchFIz3i`yu@0)L!
zVM;cz-iEI?F`VRhp=J;V21M``R?x=xX5+UKtO?j24zB%K&8*;RcvC3Fq8GKm@kv40
zTo+&VOLy37G20&m0dL>Dkt9x`p+jA#dOy?1@4WCD{%XsH3<%AVr&D7N)4(GjSgy9k
z)~)*a=8%{Jqw(Bi#;#}zo%?0Ru}-Y6*U^TCw)Xg!pwE(*XY%R7jgvF!{%7lx4R78}
zqm>aq^!@7k^wVU9l8Cyx@%%*tQ#xIztejkmXT<l9*3>d~Q{@i7t&-$HY?e_#7^AId
zxST-sn6asi=+|>R1ZH$@qqGWciT(li@a|~w6aef@$P#ISUV{OB{%}Y)M1iU%JmD4X
z6B7o_*W!9)DWf8jUk%H+4ka=y_%;ySttz~EB#{E5Jx*LGvTYmeNxyGC)qv}!S8W#_
zojpC3BLh!V7oS*H;2tYHEQ&}~z>tYy#tlV~`|nRSm=|`nH(wltPxV-tS-QXI-`d)$
zd&A7LAwn%6P&b+exsX;1J1<tYIUyk>hBh`fc6Yi~&CqZLe5kYj$3laRwq2=H^O`FU
zqbUD=wGEr9Mo_OGJ!iW2DDEjge=5H#9h4OXTVzpS!kn*J+au9}t?~J?<aF*Nk0THv
z$aIB+;}KC&A#&2=<3HL7C2xOoIk<J3rvAeWkc)`HM}v}79QKQN?S8)+3^iFFcCL^g
zo0tY~q6+vdFU(z#e~c%$o9El#xdfGtCKhDQO^GtO{!WRIZ29Vdfk1iHA~BuSizQgY
zHr(2{B6#=aCX#H;gk$aPTPb)@Q&Ur!EJKw(@eLRkT)FBudLP+4%Y;7vX~B*<#f?NA
z)1@i+F4Q-TVcE&g&o4pgllD4WaelrRK)PkxzMz`)@_nH1^MKbweJ!j(*stKFo~9<x
z=84QRJH*%Bex&{1Ng?5~#(#*vS+uqj_M|LcHHuW(QQlbH%1pngS_3(Js3E5yFHgv<
z6O)vbpOw|RxW9+_`xC!fjg5xQ8i%XvVMd`Agz`;~J~Q6p%g*uC(^02G{rdH*s0-e6
zyiOUb&5L~0Fy)#{=PjnEA+}$9`4K#5)cZ6q%Bn^yt}ZD>{HI;_RcmVImYj>GP)zYF
zcE7q`eLB^U!UjYcqX_5joA}_6ANiQX7=m6V!|l#WD(h_(@4JPCK_osje6fMD#Lv&K
z&syyg8WIFw;v^rr`Jgty7pZzzKs<<|-Y<qIVIR(MtRaLyezbodF(td0@j>#CcUQV6
zad~&wi1QjFybUs`LXrpwdqYYnJsTcMN4_WJF0e@f4I%Mm+(=jUz<i<BR|^)z^gF3N
za8x?&d<B_Mf4?#U1J$%8=r%%DR+dijlc$^-nU2nRj=Z~vht>+Su>S=yDQR*-!s^C`
zfd7SOn-ot<qucsm7f)|*uZpXML4)T2oMOB*pIz#|d|9r0g>2E2^y;G-U*%=R$d^R{
z7W<C}NOP@d+S&+|_wqB6sgl?J&^J8$g=b}zk%92?V&{3Kz13An$^XNc-BWPp@Yf}@
zysG(b<0}ycgeayHvIq}HQqZkv)SIMBGsZ!B>rsXB>+-=be%?w+9FnJ|kAWb$T4b<w
z8rE`Q2`l&*7iS>oH_0&Hxj|`w9)#0rmK&owNlN1Ih(wzLi*(yOm-TjS5?mLUSW%Us
z(XbsY3O~Kw$UQ6}At59v_#=QaJiR?b%Q4#Q`zMl_nHh~#{@=fUdnY6qG`~bn)~)f(
zlszbHCik$>8x?Yz`(AvtV>OH9QfoB#TW}d8@`l$T7q~SVQM&WT-s$QhLu`cNIRS)~
z0P~5(U44FbRt2X-$NcR4Fkz6(-f+g39=dlAbV(PF+(wIayu7whh)EwDm(lbmP@_k?
zAczI!cN6a1-ZqGi<rJc=85N6&hyW_=UPeR1#dTlm&#@K5_mPpp_9m#MxClbDKER~E
zOpD)qBp;vY1TQXi#)e_T6AlXF-SM(TD0yV^BhC0Rw18_J?EMXWKX^dM$_nje6n{-X
znFKz!sUL#G{Ni>s)qQjn;hwD)Grxk(4<DL#RDRc8&<J9`pD6Q#IhqK~`$y))QA1S2
za`!)oE;1Lrw!hma8-$3WkCQD#EyqJCg8>r$+JtyQtf}T*NFjWX$z*IE5q6i^;I%~$
z?&0cBiV@_<uC6ZQ0b&F=R7p6J<dZI%hz{?G5NK|XT@PGN1z0*=Q1mf9BjdLr=D!<O
z*45qNqH~zczs9+SfFKDeqdG3e=Kf}l@G{PT+w5KkOPpdYR=Ytpxi@SPsBTM2c2?Cw
zk(JfdQU<De-X7tyb8US{-YRpP)e42CL-Gq%VU@d=7r(iaGJG+FmZNqwF^cqn4yLjx
zd-NVhk{!rvfLjAlEH0eSS6GCGMw}XFd~y<?iLzI39jw50FXr<%r1tH~oB3em&yHkL
zsk%d5l%o0CKXaaxH2I}#s;bi9VTw>JYqtkf1+XPgK5(H@OuHQ%`u;B%p;lK{Te}Gn
z7#O#2AqC#Sr?VpyL0VpZe!b@!+deXAftYitjYBE{9&2Z|Il$fBJw>tw$|_7lw%i`>
zJ<=-x+3@^MywjIC-i(%$(~gX~gPnLKd0yTeIU4sHBQbftm!+dhQPvQjKh)OzV)ky>
zPQpDzprW)U;Z+UU_hsf!BBkX?;6PiZVEFJHDmoz*RkM@mHAqgzmC{X&jWd><5JADg
zA@uY_L_~1N_zj5WOQj3@u@91Q3512CKCuGBi_uFycu{QuB<7fr1#9c;-1@btV7=AV
zfm_+@O-P8PnOR~|5>k!Ry!cJ*2~b9agz@q57S`4iqFG~XaBJ@G?}N=))2t2JHhLLP
z-s*0~MaSJ)wcs|ISJFBYQ)xe|9nJ6(QWEHKDER`!q@{s9Z-S^-h-;D$Ck$X16|2$z
z@O&7C>O6zy7t!>T7xPO?DBq{2FT9ps0Xj+Iu{)UicF0Ph=)W%n7Z}4;vZN}j%Brga
z>TFj*{}zcwh6N4PvgzvYleLxO)7$T$Dl%*D!PC>w&J(q)3JqnSNbhZz(_P6&Pw$1z
zhP5HAMA<MaJG})`DeRU?41`_nz*s6Ea1Y;~-|uwcN3IgE*=KFvN~El<*FmIJnh<~6
zeK$wjMg`Bn;7_)xG@C(Pw!QgifdAZBq#VQ)?;EW!eUf4lY4$xW2i^%Y3k}p-Ve7Gk
zEV|h_IgA*QpFVwpNHt`u$qH*(*{~mTd;wR!Dk`{OYacA!563}ivG_*vzK-|>w~VZ8
zGWXA+Kka8$nu^U6u71A0lU4RUFAm?n;u&SZJVLr3{;auKUiLBb%=&2VX3OH8Aq#Pd
zZ+hR~^U^D6>*?{@j$$AZx-U(SWGn1L)@)<P-d5XBX&M<_gG$nl$LCkHO4!yU3}vx&
z3I#+xaa{J36~G4c_e0_XCTa5BC)|1*#25mpSCC={ncYe9@zs7rfB*iIGVpwAb054B
z?(Q%oxL3;QGR8V~Ue*ZlmOf`h&XCGIU$ar-=BZfzYI|%Ho4{ryu0rVW>+3i-ldXnE
zengA$u)GMnu^5i#+gBa;qNZ1UH{(-MQaBbF*}1uk6Zk=l!$%%zlOqK|?di1zA~>Mu
z0@0c#nHWP#&FRyunhABa6vkP*lN72(fsEuY-pwLi!upL34U8htmz0587Cv)dVI$nI
z&Se)SAq{&wJ9xe`b8_5)1?}RUoSvQrg$!ksmeCLkirBCg?)<{S0_cwH*Eekn3x(on
zg(4{--R_g5`&o`~3D#L1BC=VL1@ErE^^nbvwzYLF$471($N3o6FIwQ)XOf_Yp>K6z
z;YMX+K8ON38U*BIhWVzxHuUpgF=*-;%9<As0pIz34u2Z-ThdtKbe!aK?j~0a$yF2)
zQV`RtpMHo-7|*aQp+to+q@8v@40H`)QdynQt5VN-Ul`q8z2d8%5|_o!O<I<y_8zl?
zjKnIShZhnW9%j=5v~Spy7r)hz!!vxM#m2-G9Lairz9~`>0@o=iDZ8Q!qZkS#E~7I=
z`U0{|JTJww2G-o_Y=noA@``xkF2xD1Qws98)5D=;-gME7j39t~POIvP&4mNHps*`&
zNt=5K@$y!~#|ple-!n6Y1_lOYWe1Yqnm8h$)MqN6Y$r*v>DM?pt(#=3%VMoPqVJr0
zij8MT>#(yqS*=G;8T%l<b4<j0aP+5D-?vpwU0sPSyZGiFQTOBA;2{yKN{5!!L-1f}
zLAHge>H^TGKF4;Ln3!))D`B+yjTw|idG=SR(xr26zO`9pd-g{ECc^7DU12@6d)&kB
zPQG+7?`mP;zM<Zu@LDRXILhmF+zW)$p#Y_@=wUnq>y^56aq|?_<_|jaZI^=o)dDCa
zKfARm*A>6wE1&QjDUp-J72_zx)|TVWE%A9RNztBkk#vYx3Ke?VSX^vTqP?GV@|Y*3
z_*c2^%ZFk(e#8}PT5JP(xV>r4cMSI=XuiV7CQ6;b5#dh|CZAP6@+7w_Nov9l4y9U~
z8!V`JhBLy2@00iX2315k+h3@wRaI4l=qMo4g!2H9Br9q$_#FVm5MUQ$lGCE4r?0jd
zVWg&p_MidU3{0-S)u3A>)6yDy#CZ#==HDOwjIqDA)gAgZx-@v(<o!NQLv(_4i)G4!
zXFmuyF(YO)2}971qTYGAbQJ9a{mFB=6g+{X%|4diwl>rFG%zD0BR0)&KrDQ0?+vgY
z#pFS)^S%Y#Y;in)&&whr6rk#5)XZ^_$}8Lhqg-X?fE7--$mKv^abJ&B=`SGz9gGHF
zB*ed6P<=7#&{7_-`iXmM);%kb&J9T-7Ot=tl$8$Gp}(xFtAnYbJG|~IRF*pPStxo#
z6+c&JI<V3cIdpIx<)LDV4fHdB<qqtX<Kx;dZ@!Lhutw<lM;8odBg4aC64C?Zam4%g
z(7wgQ#+Dk?1BzV()J;0SYdaLMAL0{5Q?vNB`c05(8SC?Je8cJ~7e~&}yPP_K5TSEy
z!GLYgsNVQwQd%)P#cti_i(%_quhE`}C;hb>OaZz=gxfZ!xEMGpPM4)$1!Bbscf*V=
zP!R_8ZPM2PFJ|)9a(Divzg}Vaa4{L6XizW(`dujPjeh4<O6kI2usQQx1U$fWgDa5>
zj3G)$%@*5g0;x>{9&X=ba1yy(dV0?6`zSFnBcRfNUDPa1$lBh~Nccs}X39Sk{?5<1
z_%B~tkgQn`Zd(2+=v~{~WHW3y>9B!olyOp;S(Fe97sSNm^N%Pg#yB3w8NtVo^Iugy
z4rjdY2wvg(8pqP4oI2K!#i40`xh+l)vd!wYNB5q3^qYO_`V!-uB)R&V87L*+4w`lD
znH&>C_9Yi){pkC(&E6DR#=sC*p4USe!P>ekc9jPBsH;=x!}E{8k2*N`A8$>pty$sY
z$A14Vx-209+B6T2ajQ}C4*#fRqLViCRq(4rS&-M$^MT97y@?G{U!IiI<YZlC<$k!-
z1#)TCr^~N&YIoc45C+4ZFCyM2mlsNJ^PrfUn*-=GFf_CY2@XAv2nb#=&S;|LE!e*d
z{y*PfzA8tRc5>Q-aCmqIw3$cYF8-973O=17!0Db5U|_+bgVFloDmp2e{4Evt$l}Pz
zH}UO7$M$O|Dl~XXwJwYD2V}M9)HQC;I(qqhNag8h@ur*a1tpF&rqJYiY5*E||5XbB
zY~;KS`aJ>N;7^Zi#ES@x-P<WXpNKvG*!#kKP8hwg6^WY3^+&Olq~W1J7CRrn6xY%X
z@`N#$`_G&oBZLI}l0sXY?2C67DRs29t#}8?8qXI@cRcA}1kvZF+W7w=+v33bktXa1
zs%~AzxU=(fA0HpyyrKXckZVJ}yUg?FUrjkZ+J(mEzn<bvG`vFAAK^zF+!ptv^yvHE
zM1`5YO81D4*XA`-LdKT}Y-@I&mdOa}&X?VU5uLw_m<?rPmVhw(R$iW?lzx>Imh<Gv
z6XH2kQcMNA&j<IIs^w_$;7m_)BGaN=jnBJfhsDA1o{_0|7@yXuD9_AU(J)h93h}U#
zJAssJgQF=u*F#<X{#)<I;{M0Xm^$j3#}P-XFqAMbY`iJ`#Xk=?qvxKn<i{|Zw}*a~
zH%U&1XPC-bj%ORU4Vas%-GsJM?vxtUKV)_??tf0r{E&C+@2Q~$ow-Tp`y(<y6W~Ky
zsQ>pf`URf{s_J6|4vL)}XCX29JwoPjiT)RVi2WC-9A+m$XP_(XI8s2cGuz4_w`BHj
z=<Fkl8$|8unWl>~f^g&NDSZl<9cFdeiAa+jHX82eRgW&-mLUj5{^ml!FG2QA8#m&8
zT|vNZh)_z8l>b1k$g|I-J&$DnD?g$za*_-CdV=2u@MBO-fv;1q-rXLVKjJF*#A}hR
zNC%CM^YoZSzZT`(B7*Vt>Np459R=~2E2))cY~nk+Wo1{3@Mz!fEQQolj&(l)#NUnJ
zwic@qQq{+4g4VTG4-f!<fLc<h6tBz%wB^?6LOxX-p_1GHo`Sja%5y{&s>7T{Ap#7q
zb}Wo4%n85m=*-Q}tBi01LkEaHBP2`y`6IMDsgbFy9d+v7wR~{o^=9huUK=gc549sV
z3z7|2|JNED>+4>leY<Dp9w#Tx*4O^zSkl}HEH~Hd$~I5`W0%OAiVCi;<H`-9dfFA<
z2{fTzPEE{X`ZTN`JN*S*_eXPP5uy@mn!2h<Hs4_K+atv)BYxC|tnfai!BZ-0pnYd5
z%kx+%P4HAad^8V7j1?{7?T?_o*M4{?PawLxK+nb1^~IGIy3)fTz&xPU0<HCP+8B`Z
z-=q`K7uy?KG`V<x4^|*!l!$D`<qwshIt(5fbfr~AP>j^f(@DV_!>~u^Bk!GW4FR!M
zadEM|y*-PD{6k1bQ&zsP0zBn;3>O!dy6@=NSg8)}mdFnP?_uBvf>Tdj-HZm$+z6Iy
z=#htQnN3>h^pYtMGDY-|47YHHW=8aoJuxGuY$N6^%NNvFQ{&v;N&=OQm9_OtxZ)dK
z>d@lfeud+A9^Co!LX?=~2=|Wk_Dpd@72MXw-I~NC4=*J;{&GRD(%m9~+kmdW^M*hU
z^uQY!h`q~GFfRbDFze4<61T7AC-3vP2x0u&jG2RDZ*cTtJ{;^I3Ll8m{zHXcDv!<)
z^Sfjh7TTJB49m^64DM1@fRxpkvL=k6s?iB&rHgKJ0}$B!k-4>~%bPx0#uTFH^AP8F
zWBx<ya;B2Vx7H6uA8BoEHz{(Jw6<S&ERH10rF1N2=AqNx88iIVXJl%ARcYM~N9p46
z4hcw458HVHY}QdwQGfJOdV3@0Ch-&4{FC`Qf3tHHrYSBKMt@j3RlQni2wEkeLD;y}
zHJ4}xzDm;-dh|B;zf<U+m}I{1ZD?2Glao1O1f%yD8!O&-m^JKCxh}QY?h+omiCS3f
z2hJ_o&7iMh)%ZF?gdZfkt2DWy2A6FnbmCj~1=iJ7=XWMbw!BnMgj>ELg0npm*~wRd
zL3l-+;V0EUr&Ah~TfdW#FneC0dK+@)jAgnac7MjqhKEHFCRoU2Bec0p-lsXdi{n>f
zDAEhLQU2V&#)bO<SUG5o2ypRD%4#6X5sYBA&tDiJr5K1%BBO{gbg5h!Aj`>hR}6CS
zKjvN7kHm7<bGgn;E$N+(?Bx=5uw^6L5Q^!%kNdOY60N`Fa%+*^&dyHXz#xu@9ViHp
z-jr$U52eN`<}$e^b0lE|j3mE`A3YY^vE-G~?_v86wHN)upP^hX7o_t{Y%(oqWlQ=h
z`}q)O6{Pg`4QlF2ZYgh@@~mn^hcWwWMv{fn6CMrDOE6?G6UuQ{KKA<N7eI`G8X={o
zV^1qnVE2lPF#HSfzNZ!vjl%jHU(y2vx=6`kMyvDr=KuyUKpLUgG;RE8%*-!kM~R_b
zb~&9ricd@=#>M@Ro1>zuD@&xA2<j%Xaji1B$)dQQe&|+QzsvwR>R4D-*6Ntce99x`
zwmG_OYyMTcD4A6FVG$!ME2|K%W0P0poiZz9oz8c-5514Kta(id?^oA<(eU?gib;@T
zrl-1t!Yg+=)^&214HJRVZy2#!n9HfHr&bmSJ}8+O^`7TA0;wfSFVC-huPn~X^2*tR
zsMR%7f~Z)s-PJXx0+uWaO9hs$-Qc1DSvEK{6dV);`H~>X2Uh^8`lyyl)CY8qq@vLo
zAh!h+QX}VZocOdW#aY&u(@uzDKk4EI)=1i_h+s!&)0DoIr$dfFAcCU1myf<g(8`$C
znXnTB;t1&%c05zuwA!o*n;RR)v2vv|wYUBJJ_)ba`HWef<x_WhzGzEd4if2pgMC}P
z#i&^sr9ZiQzhB*hmF=6^0GV|G5F}m82w-l5Ta<f@Z{AXQe$UOo^vJPlH_D_-*1S>%
zNP6EqrY2>L(c>(Q;h-OLd2IITNj|sXQiUabfe;Q6fLNiqf&&FJUv}B^{Fo(~J2*`k
z;kH{gENc3$k>$DS!noqF528Rm;q?66NV^W5LSUhD(HW`v7O>`VFUn?-cv7}^b_`#3
z0|^8KoNCMm9=CHIPD!zv=I9u+u;=7P54`D$-r?U(zb6^Pcv!N*rK|PQknLq<6X`}f
z#u;o_sr$qh_>5y>(gjQmv+KS$#{k0>0mC^p7>FPpb`Kis=|6v9`HPJv=H)+JXsAGN
zk(0BszD`0(85yIht{&L==~&q7YbqwZ+$H^iLXj&E=iauq<&i5tKwf(;J8Z42VAF;q
zL7XUcO@qyi_TMJ*0Q5f4wt<pCL_`D)Eu?R)6GzT4>{%$mLX_$qBRaN(SW-9A6se)C
zq4y*hrgW`Aom`erY~P0Mc{Tuk#vn{2YaU87P<VI0Ff1&A;IkG1&A@fZmQBC)cFiY6
z8=KVO8v7dZb|Xt7e0<lBZEaiD2=^ib&YeF`p(K_zeEy%>U;wun1gk_xM;n`)SAs4a
z&}z`ZP<#h2jGTgkOmzkuru8h3J#m!ZQVzYs|6WxI2w2v6B(Dp(C8qX-QKsGz-5OU^
zZ=F<zREU1qxq<*g{kx4p3JL*n1`KTM8+S@&JeZy)0yuKe0y`&pZQU1Ip8NS-+Sm&}
zr_BxP?roCsIZAlgwmM+muRx4}PbZP0IF>UcBPt$qtM(#T`#_PoH&!#f)p7tGji;VF
zVe!{;#iDacdE(zmMOkW`i{_^Vb#J5Z%4Pmngh}BIp$A#Pii<$V`PL8+2f5(Mtc?@&
z3>h9qR<;iXD@=6OT=U&r#={i+^o3G*P?ZM;We%E|9#_KH6uc6N_`W>DT)KP*@KD<d
z2h7cOM1Qd~R69UM01pIiU<^=iW;$C5!<O|J4icGL)CgC)Z~O`N?PhrcR-|0I`Je8`
z9th?$%hB|Fc+N?S_e6$J*8Gj?MEBi++5Ew*p%m8%EIapmthu*C>^m%;#O-D50FxbM
zKjn~aDZxystZL=xc-<!P-;Kz?c<SGD`<fYc!~}I~Wfli38uXXp2kuNX1#@X542cMv
zK9`*{>_;E(?TvqP=JfUdQJ8?Q%Tqc5kRW+yXLPq1j^}<u=ved(c+=Y%Y@oKZg0E;V
z6NJb}NWmX(ACQpf+56T+O&iVh!QwgXgsM*DGhOCbg3fG=AY(i(9;ADO-Jd$Gx-7_s
z^WB+JdW7B9l-=C8Vmr4gosgE@6mELMeGSmaZSx5Tl-SL(dd^agPazXWCsazO${^YZ
zK0jG4<w&E8-#-OLpxBNrw$tY90ZcWgc;PQ$)5s>>C`fp?r3c2wW}m*8uJQVYWb%i>
z5ssps&iusd$IC=_AoeL%Cpv~k(ck=}n=O4z9?>rU_Q|)*rlI7Wy!6b`d*c7Sh|1S*
z-Z=ApN<L!f>h6X*4#>n0s?P;j%Dedi-C*3QTeGQd3g~!bcbb_vB_HJZLCe^mjc304
zQfaIx-EBqR*H`eywhBktVx}R)-1`bQ>(ayA>J4Yf-`S|7n%a(^KZnWNT3a9P9(-x&
zCJMRwtL7>8#NxL9q>643#gs5Xi0$YF4BVLB<wwuKkO){zM2Pq4*%@H!{?*QsG%&?Q
zK|pr%)VrI9Pb3=Fr@&gLYbydHBy*?XNNr}gitJ1me1#Gzdd0~F1M5_>-449=kDqXc
zRPn62xouMt5uq7_UBTlPk@50I-j0yR*0@S>IuPOMFYUtP3UNsZadA~PwP~HoYbs2;
z>M?Ii8kg-oAsQ?Bt4(!<AN)PSYH4+T?g}!>xlZ`U@WD^oM8S^6q-Eqxb%UVna;`zh
zdXFF~rl425^YJH8^7*HW-S8`<==tgb5IfDpRER|(Kub%Dpq6oIe@_Cy0{AbluC8ch
z6q7iX;q3qjR$RL&-Car%r>FX%(+p(A@R?SB{hC)~NQZ;&52=U;2QDx!Lp*|C2v-?o
zaJzlftSDpOG&L7Pn49dM(6eE)XbY^K+__IEu7>R4P?aiz{r#xiB`+U2@a1CB>l&L&
zGfT_X+J}g*{MBzN`CRBU-mRD|3*SdaBIYD%kwQ(a9J%MlouZTuC+_yYZMj;foH3j}
zv0&7=gaqC_nK5j@E;5QS5u>9zye04kAw2W_6MrYSTkLF3D{!kc_0`;5Aj}r<!5cMr
ze)#lBUtN6|xM{$oZ!~28eMQqwKvoer8K7uF_7g0;;T9Hdnq+<Yw4TZjwQ|oq*c5g;
zR60QN^G`OT1I9dL_%Rjb<=~rx1wnnkeks!y)OqZt#l?+)M1Qo@C>;b(00($eRWl|f
zB>^P7V`bLgH>sM?-MNy0UGFS%Fl<d3<>5Hg4ubRjQ(rLW!qIdqackJSBT3C_``s+|
zW~EgX1iBFsC4gD_aUE=LoEsgnWw&6CF^nqvRxA*fNIT}oJg%?I5Fg7Nlvq^yv|Cgw
zLMpf_r#317a}O{8&CAbVI)MdJm1ahmJzK1)qm)8H#}rOyO9-W?k00;fyBGeJHGw*|
zQ+ZuQuK#&Vb{Nsk$l+QGuL*>c0P#V9J&YMh6%K#AA-rBXlHBX`z+6*PQ?l)vUiRcS
zYurLNYKLW-7tkOjT2Bzs$;p$M5@-k(-HPW53L1jUm>3v<6KScdt1ByK6npwOu!C0W
z7HVzZPJhma{IdMF-XFf27<Pr+<8>_4W=0$zAH$JVH0ck6E{kHCydR~OnN4|S5xGH~
zJwH)pS%xtXAPB%jN>O^6!7vUV_q(KKO-cf`qQtbMSnUV5`gJ0zt3IgU^A!3knStCX
z1&|!G$iachuP-EogaaSp-6)rDR%-cQ+UVi|re*JaY`1X8cbF!0QGJa{M%EzHh})C{
zVLwNYGU&h=^}_VBzw+r!qXAb2mAU=R$CCxnU|dU}uC6Wu-rGO0Lc(4-Pahu<@4tCA
zi+x+FkI^3y`9P0;Lv}deg||UcflY=@8FcDLdwaml*T7z)^ggZElEC)@1C?J`_)WVg
zIvPhZXAgd^xf$rw3rqED%MIUJuoDB92c&5`FKaz_86EmZIoi-v@2Y=AhK5I=H-pI?
ze5D$$@jl+(hG}FN2-W<KoJwG;3SQnPktvbwyW8_CCMFvz7}vpm-Zt;Lt>a;;;&nkC
z+3sw!BPlz+D-q&*)1T@KfAHUbI|vCtO_IM6#tyW@k=8e!2yH__e*Q>7Pe;e<u$73F
zZB~WZud6@YhiM~hU%q^S9Kq_UsuKC}g7PmW`&#RmBl<R%lPTNvV!&Qhd8UNT!Zyq$
z>B2mUxW<Y5V(v%UB<lNJ1%l8q33&fq14!t^*B_uxaL~%Z8PU}Rv1f<CI7L!CHLQ0p
zhLUoFv?(e=DuY+00dV$5pBwyDYHi8@sK50J5va_d4snZif*_XY6S<xS3k{U3KTUq%
z9GpQwH{ZE>sKc!3r$7aK6l@ArP*8AHzzhm}Pm3+6Bu_A)d$eg(sRO3mwQnfI#D9OE
zIKd+>`$3Qr+l%Uj@jGm51iYRhC=@o!>zqFq?(7ay?R6vl9T*<BvmSldcK>l^dXGi;
zvw)QDLOQsN{ex$JLvl1>kE)-+7eEOO)W?3#Wg;Bzz5zX`lZQuX<zZOZ-Mm61gi<lg
z0-2Vl+RU0jNQ*?=Ljdu^s7D*yVho|%Bg4m)yi*>*Z6*U&TqB=YEy}d<LZ5N7Jg19&
zj28+_P}huISnbr=pcKx}F;a3k++Z&QJUFPRs1K#yBjz?YpL!jb0oSOfrw8CB<gdX;
zqNk=d)Ys>@ztA?~&fWk1JMU+g5fi`}zyZCzy$$~)&?14&h-w#HS}ZJU3k#o5q9J90
zP2$i~F!b3%Nnt1VFq$a+yT|srr)kYGsT3K7g@snI%SvK*x~+>H!Rz=eozMA4M~8Ho
zEv$5GJ6xT`(d4_^#YRkQpgh)|cG*O)1nP%Mh>S%z9CXuLni0vElLu*An5^X%Vyc!b
zVHSG|KMr7p1L!jRUg;<P+dTE>hnBsEE5l;}SmABr*6X<<va&V$CHjWvIv7;K7Ykj>
zPTQeRt3#qFk0ksx)vIU<f7EFx*c<tL^Wz0?DWffHFD$SYD}WdJAMXP@*u-#-z7%80
zgc4)417}=d-{7(=Tgd_|%L1z9ZmaMxSJ_l*`}mx{EhOArSU3PU23GTesE3pD8ypNb
zQNv_BTQ_om2M=xt&gya0p!nq9jUM=-7adAc7^zZ)XBN~6-7UucXThCCx((x9T^pd?
zueDSJvd`2VGQ^_y=QGkWyMIWAU{M*SYpop$XFX^9w)(Y{yX^0fy|&VQfY9OIy|mZe
zwV7N3dm1@bZj(0;FE6&0O)!C68J*il^<hAknr-<(z2VcH;4nH;SM6375+dz)=cZxK
zIHCVsge@&9`W?ix_eV_rOi!nY1o*=&^p-G}_J|t<|HUjb<?->UIs?Nmeq}EDQFjud
zhwJfy_D#dS3Yx6hFgdvQMRvZBi)apoz63b@L*v=}Ul|OlJ8GQK*<z|DX?~mtiApZs
zL{e94qwT@GoSYoh7iCl;{vV&y!uNoN_gExo_Il;r#MYWy^0~cy33_+9Zwvm|#W~gL
zqXw&JY94338NQX!tuet3yNen>e}=n0r<|88s^cO0=cbB*bvFu%)c@g*P|;*!{zvfw
z%=r$?SiXLK#YIJ=+9~D3DJ_{<YH8`x)Af&bHW*_%sRTVUfYn%As{!&nBQ{qwi{&4A
z_PdGD@NgywnH<i?Pv|b5Py$E+KA}{P-7>pJ%*FS~NKndtjvL1qQta}_T~=pfMAnIs
zkiQJ%;M+ijP7bMS&I%po5Cj}zz<PanPK$#M!{q-A@2F_$eWl<$jgKE?WgxyMj-a+1
zX?+xaAB)6j3D8-;A7eDebe=Z)s!g+#pL6SUQ!BZro0nlp3T!<CEEp7yS|P2E|F1FS
ze}6LREptCUEGl6;fC6#Q;Sk^6+QK5|>(3}x4}l;R?aMfapLNb~5Eu8rS~e;Q3WG{F
z^cR+;)lN@5ZRp%4M{*RJ+9s{|z)&Lek3u`UM96q?<wg-{#edTb-M`j_cNEX8gGWXN
zx_bw6^COUGK9Y;)%4Q_^45z>61^S(R&$PghfE&q=MGEL)p7y1RFw%`W3@ZUwgV@W^
zXJ7L3ekE_w_rj>M%sp>6+wkZn;AwADp0BcH>pN_rVF;Lm!xpGN2X^B*=(fldDWOi@
zZfiz0Z#9wme=*JL$>@cEyO4$nMVyO+qd@*y-{O9n^p1$<9(CXV+OdYkpS&($q`o@c
zvEp*uXt3;~EG1x){$hT2>3~PDcbCwj4r8Qvuy@``S@o8^{nDzfjSZvM{q#PnMrTgV
zH?LX}tA#(q;mgCrgIN#@aO8rB<RAUpe~*qf4P=Bq7qvVA&`V1T7L2kHhQD0+c?TpD
zsBuRUCnx(^)#0wLevNw8BxhJX$2)%twO?}b@L*wJT>NOct3{m8H9IW-!rpDpIZ8Jv
zf9PYf1k4x3$_#d+MH3#qofO2xo)3&sF(T^jbgvfa<u0kfW>Kr@6SeJ__+l7O9|r>(
z1TW_h(D)*-$%8ippd8iw;n~^Qsi}G~^Gh7K1q6Th9<QALJ#0m5WMNE#EoGqC|Hyx{
z;&bJ*F_+`3hA59)7I&5pj*U$1QP9wKfA`*wS~WDRf=8fWeUOY4(1<9lQdkuRCN746
zkCl}#$BN$j<UdwcYdx|xXs2-Cd-FfH1Xk0<Gx$(?dV7zLj)YuSFyPkC%d3GFwDk+m
z^(!0{+=LX~FZN9L6e1|&E$T?DWl@c4>g$`QkEY^Vf-5Y0aBuYr?mYQY47BC@)b*r9
zf_I&2S-<~F7PP0|AHhM{*BC{G&5gByUK*tdpPg>B`QXma){?UPcUy>jU~85Kp}Ve{
znk#T?m~1O(oX{p1zQ|OlsI1dV)Oc<_vAj%5N-9gqG(4(PqNB+i$Fur1pqfi#a1soP
zvMM@GaXp$7q;eymt1zgtBe{|lyAi0sQ0vuX=k#}TOo`@$AaLza)n>(%De^`>e8Y0$
z`*hx!4>M01eP8#IS%0t4#>Pe$xi)p2cW%zI>=y;Tt?-^Lj3lEny1^oyZy)Sd%6?tq
zgui!uM&)}ML;dc7#9MXSb~BF6O|DHMfmFrZAq`D3*b~^3tbrm1<?eq9TX0LpfE8n5
zgko;t3Xv=hN>qoMf<2?pw<k&R!?Uf+(kj~8+M1dmEz;SXf%N61<m9^AT0dt~zf8M@
zQn1Dr8`NvQEDzU&Jm>(w4|ik8&vsf(#ojy!cCHP5b368Q+4PDihW5sb0P9U3dOoS(
z!fiThZ>WWlw_sArd}vWQ^zZO5!vEC*G;g5BofKjzhyI^$F7|#JRSJ(mZyre{(!03d
z(Aa1j!Bj-}<z|xs2>9WGAw_qZz07-{)y#0tT9TPLP-U-YX}Q{^@fKFlawQf7V7SEm
z^*-2o64LSLU2A;K%db2@ZNZu>{KRCj56vT$S^&Cm|DPMa8I_YCH$dU;-swebcLJvE
zWVSfnQ=Ahk7(pwVM|Wg+^lUWb?ry=rnv<OkcP=EXL-H2Xk|+|+x82L6BqZAqLIDE_
zRB~v6p+7Fn%gY#Ahi`5wYEQ(YM~|Qvf>(?pRs^2_wD~WeQkY|60XPSllywpNZ0(mz
zNgY@{Na+7|+m*8B*0gQE!D<8`R#K9aVWqAC)C0!2_MboZJO!ntJ7A86oqah9WM&o_
z*(ju>H6CNu?WME-1pk9|)UkR`W8+ZK(9n0L4fks?1M`OBdX~=Ksy6FlL_!za*4}Po
zYb)Y+<_1VToMm9T+y&<{APPf+gN!LUy1Ec8!q3341|VXk1LSP10FWu*xi<$blVA5b
zR0lU`3Lvo&>>2Uh%Oxcx$fy`6@haEO_X!9n_+6ec#(DerfOG6IPa4Z(AG_HtweU1Q
z#WcZn?UWYh)n1BbHOswjt_Gh&ptqr=zJN6YyhSO2ebt^$*Ah{N8Mr%7-KXE71N+qY
zj0TL>(Xp{=$f@tt8HJ#e)YY8~L_*G!xex5|T@5J<GgWJ+j|mA0=9Wn|dB~o6>Lxa(
zYDp4eC6{lGl7Tluz{)z<FG3L2A*ZMa2G=+rP&1>8Dk+r~NQ;d`K#CGPY#>7_UebI7
z*fC_o2*L6tn3~tz&fvX+Ah`joD2^m2RIjy}(;)EHfESxRiOIL4f!|9UC@e$mCqfU&
zP>@uyf<X5kQ2or~`pOT3@-8KIkDe3-zuS#yq(|??FLqjbzl@=IUP{s7t46hFpCN~4
zWHMNk8XXf;;rVkXH#gXXxl*0&mCEaAnB9bHmm<Y%=)%vpMz%|N^G?IwO3XdYJm5sK
z({B=7U*Zpy-{BF06x9Girh(ue3Tpequ&Qh5<6||jhiS5~W@^qx7LWmX2TTPsGc$0b
zO^@y+fXE0a5_~*7PeIB+_;Gyv{Fmn#So!K_8zW4hrn9MZGHUdGIaUB`aZm!Wsles4
zw6r0>BUBhfn<DeGG-##udq(baz89;#y7iUS)g7_gTxDX;WlSlOxkF-INP*PK>+O<V
zBC)1!%Uy9#$);#L*N)Z-D<a66KOO(w4V71XlP-dgPlWqKl7#%fga1Ad544A?Lsndm
z{)m825)Pm`_w9|XEevB2`a+NhWbrKQd%}#y!ON@VG7Vh}KpnoIX#ryO1?q4_Df(QI
zFOC1|->KOUXxtnf9i{zMWM4R!S|dSXeZjQ3v2oz<ze)kV&pFAr=F`*S9>hC!%N}{`
zqLcJH!p`crO=fNq*aHes5-)S&Q&DN#mccIy$z1C`bKfhGeOgv|{YlT`sYP*uySr^z
z?&Z#b=4_7y8PRy&{)Y8D=kTK}HDPeuj}R_X(?!RPu>Vfb^B)(<{U`v|z)qYl5qRwo
zOf%MOTlC9!0I<VkG7yVX;NbK_ga}9~{N*0p1IZ`gcODzV)-dzY%A^Ur>;i{UuS5YI
zc+!(I3ukieIq{tIh*U=AZqAaCLQCa2`F<-`A2*RKM>pQ)Xu7z&%LnzGt22M9r&c;+
z`<}7jqu_!&P2XJ3%xno(l#L7z!!M!DlZk=gyo>iFmM>nsI<W*~M`WkkYG6T6@lgMl
zc;BgOM$Srp?o;>fc(y8{&}Xm^dUR;W)ye53-uVWqW4OwM=n3l^iVwnMs_c>uPT6fg
z+fia608?suxk(rZ_Y9Y^W^|njn-?Dq4b3#ith<#N#`nXe(_bP~16RCydGd4*WjJ(|
z_YXtl@xHIC5er}-Vqz`mo0CNXuBvTMo`Sxk^~VqBBVp6(;@q4uR>)xl1%&aLlZ%0_
z9$4)iZrjjLTL`9buqKV#Ni5Ijk*QOHo8*7L$w>C%YkC)^6|TX9{13{sHJ2_VYJsT-
z?ozmSv_Tq^(bdEWB^kZ;d~Pp%?9kzlqp&A&v6O?0Kx&%8!q{(eyb~GeLdQY5h5)qu
z`w47yH8os$JPHauK&Nj_RVztIn8KbX$Qyw`s@cKJ7>C@Ol>D)8Yz@*c+P5N<()<@Q
z{-=#G|7f5(0gvWQwt0oMMAiTV1p*?FmXt&SA>%LCGY#Qf_`f*j?+a(xHPR*y{UY2Y
zeiCLVWnaJY5#sy6&~!ydr(y6IFop@cDqw<N)6##jB<wcBB6rUVRdIfH^974aVfuou
zWy3&80rbcBd*`UFoc{)kFv5Cjz%3-Ap5vhGjiDqm=1AuJH=%W86t#ftFD54D^K<*O
z_;^Ab9Po0EL38fS%Ln|1mAN@6w+SdI(>{F2%zL&0zgS+rZ47I&02wZw#btWJ&Q1(@
zUgp&0`mEQmW?3=#ZCIGhN=}7!voN7LuogZ&WX{$oV?WK{F~H2u)`VijE`ov(c&TVW
z^2Z5TLtULg(;oDhZtZ=$4(>cI<(5^B9m_6c4$A)(z5re*%V6q=)A$AT0l=5ed8ejF
z4}Q*m`2V1L#kyyL{QOC+^YF|KpJ!Aaz`Qyn;7xqV891v*PJa-ND6%%QYAw0`c0~iq
zZ=cIE?yC#r)H>>`BrsZ4q6WY&A9x1AnU<rNs=F!&E8{ufASk@`;N*fSvk-*8=dkz%
z(m_DrF)=aWcPInk8^CWKDx-6I$HzMZSs^rfzks~~MM)JTuYoJfo9O52A;F6#eb$7&
zRjofS0oA&Le<gCS>-8P{nI^x&;^Ld!HE)t+>URapPTI_bjL4CHre%y@XI`->kdAe0
z8+E@Y;S?^yXq_11^J=i>53rYiA0Mx|JOd9KmSH_5r{zPD>z*PVfOm@Z)(&86cFD0v
zNESXyPz2c+fXldnAt4>BQ8ehEbyx7h*u-&OS5+NHY3^71)|<K@i{U(zXjmP-Hye|l
zA@%@$!B&W{nZCg5+hc$6O}P}Cx{!bXn95+|nKKk~xI??m%0Y#OjU5H|R|J(v)68wa
z9qypqLWpO6wXvQnFX#C1T>WU)Vx&L=()6UV2;esA?guyStydWw^Mzo<vVHcGMF1JX
zyR$zKq!2jd85x=SvA}8%E3+_br0DMAu_9WT@`4y(4u05EySZg%`K0zJ$jTN-MBeo9
z@YI-`$s^|WS&0b81SjBC06GpyRtkXRz;_7v00CpMAT2-#ULUs1=D0S&?F`6fPx=Ga
zgu-O39{>il80T3iBd7G-{I5E<45g)WD=O5vuW`ZUz35U4Y6VKES*!@aPQp&omznSa
z%q=Y9KYqONN<MjlD++WCnRqlvVn7Vo*ysm53=s8<)goR8*)XMQ=)p9=C_Jp?<&AKu
zZ8#wLZpHy_G-C3@lfV+BSZen~(nHa!Jo?vQc>S`i*Ix*-w_UJUaFf6+$mepOHI=$R
zR5K?svUh9)HmGKHglT-OY9H+(euCGRrr<bRqDv3$<s}hY%a2FY%a%ka^i=%WSy>N(
z`acLUi0^}M3jH$+fS14N=Fi{6(B?xt!M~{}Le+a=;W=HC{uN=mz}BOP<>gcbuy(y|
zW7Vx#gE&(t=MT|M0A9S=5JN2nS_CMQ=-Aj}V`IFWoX|IcckW&wG~T*$S=<a0DBIAW
z1Wvs5w)YYlG<ShQ*zp$oo^Oha@9QzZ=?k`@9X>Qcdy3M#K~Z*gG59Es!0)3aP@L$x
z?_|{2?M|*MJ$9W{MozE4&!&TD>37n+FuEZw8^dF9FEIs$2Z!SEybTVugDdeCL`U_b
zeIrFI9!#V6SaluhbW?Oq^3AOi69!fBhkEWk_7qbQ!$nyUcI{pL*P+5eOYZdc5NW2*
zPjE76ah|i3{}8gfS3jNlPUbXKmP?O4wfw{9f5GP#g(f1hhYchy2&r_=SA$^#`iFhD
zg;Ln$0@enYXm`fTS{<rA*7|P(_M5O)A2>!h!88j?zQY{zYaaV;O}8-@vM3Ha+LArU
z<IVJeMg@>h=t}%NXpY|y8ET8rko@@3GVf_NuLpxcy=yw06+eZaVjB7&n~@l%Bs6+!
zSXnVBUT8e!+t*^5gLx+yKValm3Vwi!Y@B3||L!f>KKd)&G80KagaJ_vCuCu?IJlS#
zhIL4na4tH`euGeZ(9v^p9v%zXb_|KUp}>-Eqc3ZnH$LQ`+C9@!k%=F%n5!XRNy*o?
z2ipCz(?d&i`qIhix_8*4Kwk}j(L;dcu<qT13taN&(M#wOj6^QI*6{8%n~9tsY%%_`
z2H@aqHT_&^WA@Gir9R)#{m0wzQ9W?3)mMV;Kw7U_j4Orz{QMggaIrWu^@+p0%iteB
zpck5&qd-IW?qaW6w><GX`wLAc7+-EQ>X4BPu*yZF2v2u8MQ?y1)2%?QR*9+neWCHY
zRQ@Lq8m>^SP>A`gAkPqx_`(VaAaJ8BbAQCfW@Z?#4(CGzazR@bc);d=#P@WmGgSv@
z<i@-L&n`LNSHkUwuW<_U_ZD;gIoXdA_`9x4QMA1-)Z1398XX+Ze?1KleDGi%b~<6+
zxdYY?-LJ+i@G6<ZOu*IH>{x8y1R0M&dVv+^YF#PdE;{<x2|FdBBXu#c7LY$O9$)kY
zTtN|2gDs)P-viO8e4T#0zT;ls2(<@{Z#vLmz@Of7f7*AbS7lebnO>h}*!l$i3+xNB
z1tAKQpcW!>2U(+E%?lr`jk4w+<KHu)e~3WnkE}KWbNgZ%<y}1$D(305N%7GHrpkR&
zQ`5qH9*BP(c?hk_I83<K){hZ;g4gY2j_^_;J%g5wZTI*OA+t6+s8)d-X1ezc!lI6j
zh0?t@^5HOJ3jdk&<))dnzG2OfkN|bM9SEVjmKGMmd*6|vn$_D<DoOLD`RvpJhyRF<
zj?n{B`x+q1npAMt`%!V;NnVe`QcXp?g2vN0mu`F|!Lr38uRT$%@$1&~>-~REJ*;4f
zEP3BsN<=wE+oPFej5)Bj+(=L$-T*rLZHpsO^apNi{#$#Szm*_g22^QilOBk0?#>sN
zQ?pl^`%29og8{&zLBD|-2>ZiM&d;wecLR8dF(8ttB=J)0M#v`YcZLW&nBDmINzkjj
z+|t3p0rVhnh$BIP)zEiSiKr(chh>B-&}PF;_SF_b6T)d_Fi)7;n#U=^OAOHh$C7zM
zaZoB3MPX6Nz|C{hu(mz~FK}mizN~|itl4$7qx6d4_exx!LUEgx`i>i5GHHE^!7815
zN4!eA<MVN`Zy=i2CcE;us*k#OZ=(D!3~-^J$065M=?cqkq+kRjzd$zwh~G74=mSn3
zYiyyv(HyT<w<oI|o1vEj*CHkKNzmZKj9FY^**515vkexd;5f_ya0dhL4g1kjSH1eu
zd;c_lfhedg_j70*A>Pxl><J`SN?|Xm8D7N8>ePDuTIUqewfB4fR>5Ak?9k)Vpme!v
zYCDM9b5l|WLJmZ9pNX!*szX-`qtmUS*+SDjJ;uu5S=Id8@W;33s?}{?^*H8FkED(r
zfQSAjfE;iYfV2I8KR@zrEGM^yS0*kY5hp1J-74f|Fob=ETogq`00vICeE<c4xS)KX
z1^_((RODtxX6^=TNB4h1(4j>KOHMVw_t1;Mv#&+fH&oizRsJb9*0g$Re)M}U^k25l
ze7-H=w>zG$9;_@G;qo8YpYzyPY+yU8=Pgz*L#rjyf!V+NAlm$-nMA93=GbS%O|r!q
za$}&7{@wYLl1frvHkU_X0XxazU(9?qnJxFp<@eoAE|V<NPGOBH%>MTwpaDj_r>klI
zPhZyoj&;BHZzQrSQMRVtG(w3=sO;=0MaU*2L}=%3kt91=AuBR7BD+FmW!y%Roy>SY
zf1UH5^PcN{U(a<t=REG)J^thO`+h&`YeVA6Tud*!NE&xG5YlN;9x*|6L^8tM(K}HQ
z2zpIRh*}yL7Zl3N>}^%(y7Y52!u^}-0T;fu*Vm5Tx%YCZapd?Yk;vTbS(<^J9{XfN
z*xn1ct1lN_!KHpMD4v10YGm~56~msE3t_XCH3vNVL-h4d=Tq~{98viIGT-KMOTtGh
z6U@4cKOPD*po7`R7iyVe30HDCzs>I;9Xc|M)&fTAtk2aLa4D%0$-yHV!9tRCA0=$t
z!a6TN`poV*Ro^ka{^Mbus-yD@MRw<%s*fqM$@@M<L7Jkb*3-YAH2(3nd({3*YKB&+
zFXRm$Mh<$CDbc-ZmNz=B_FW#{@aC9&BcuP@fy+jLY+CHCb$W3PdZ7rMt#Ycfq<BWw
za|W|P#P_wX()PmoRaAvDJ^t+YAW&Wq@iCV#&%-I{?CcDq-$0XrWN&Jdtx@V{`&_Td
z&d7ZIxV;gO#?UD&hr3f#V=SM~<;Tq<Zp$uZv&&v~eA(om?h@{GH6eGO1zXH<u|%j}
z$ZEwK{j|Q+&&6VywcCsNhG}}PQL3%wxz|~qf4Eq*bX_DRls$|b5jfNbq=S6-t_h}?
zAufoRQaitoBOF8oAgemdDToU&jVayjSq<*nsZu(AEF*A+{COZcCnx2@z!Q3Z>iEoP
zNa)$A$M+7-TUF-YEA8^VD?Dq}){8zCWqy*IOA4`jyq)cj%Ar`7D$);q&4?4LMV#z+
zaM9K0C#}U+zcESs))W~KI_Ouj{5h9{$Er0G;Hb@w`BUoZ4NrW<@yH>=q4}Xkv#*(+
zXpe(GDf)&D#WB3?ajY{a;M3-$I#hXKlj<=tn7cYoR$NYrRO7BTQ)mrAYhXqLv^*e>
zCjwXxj^pj%2VdLvI$-?8nO#b~Z=rBXiuf)KOjKZk2>7$!Yt@t%9o>!bV!TqePsS>_
z#W)xPJM#0T3ay=-GMuNX_<f11ZEkAN=T{!qt{+=87|B?g#6?AG8yk1-*)usiK}j$A
z{dA{JMk3Ol9e;~`wH2rxr<6kNn1Z(gw!cb&Z<XQdaOQ>xyN|Kg()o99-P-3=#Pa$I
z3*8kzjlF2vR*|qzd%3ne<vcEa;6=&(kdXdiCgR-~H8hem)zHWXx5vc91dYoVsVB70
zR}LQ!eV%WU2HWVRp-)Fgj+Cs0phRl+LA^u67n_`&5*~5(rqP(bHa3x7yeGV#W_jVP
zhldQQ5bam}Se5dzbmQDKtW=lSXmC3&$cxKhr3r<?5E=h1HW=A~1oqJ;1KS_1S%cUO
zTGS^BV8q7_UXky8M3JqM%<Hw#Ja9|7Tq2Zl6%~S!rlwMAcj5CI3|N<DoGJBhJh(uF
zM%m{sV7H&b5u)k_X*$#8TLS6U<$=}<QJs<(q=<_-=gf~x3!5rMdclg@Trb6B|C(Vt
zi!IRR8aPI`L$}j6{nHT9NNd;jgj~8v+|^985Bv)2&g{yD4RnMZZ~L{(q!dB@V7tUQ
zj-!8vMkDiqc0OEGXo5ttj9OeR*oXx9q*|<?pTn?C98p$oh*Bs?OcVliY<><{27p6A
zW<~}EiS|mwb6eP5w(tNSfU#zkA6DRw4ik%*DYGkc@4C8tG0h`F_L5Coc2edburyv4
zy2hyuU0;ga>m*7loOr@RkBrM}(4Up!jt_2~*cGH)ZYgCc$BpH|O{zaS<BjPp=MP`&
zKC8%bSM8q3EfRU+6<=?eReFX*(qhN|$^2-ZflLa?ZQJ@WLhs_I=c>ppCOwajk5KSE
zxNDaq<YlIwR*{2Yr*$5*=5E6&>_yn&%<180lTHmk8IymxtUh}og_cZlmE{eOVBTmt
z9o*XRBAEVLO{Hay{)~;|K8%*W)cy7<xIxIYMAt*3EZZ`C^{#SUiw!L)yk2i&`|gcQ
ztOsY$Ap7Jn>w#d@Yg&#%%SO7oeIYjxq6Ipb`Ap8&{O3(W_E9$igvHW1xPimy=~WV6
zG(6MD7M8&_#pJh>HW}L=C1qy^*W2Q|r{ifnJl->ZxMVe|JWYM3ThNeV>szOUY)niT
z^CpeI3tbBxMP^nOm4KX_oQO!W`8joU77`N9UiStw5W)VqCm6|C<!w9iGN7zk9U!=O
zZ<Fzb-J{Jo*^HbLuqR`<2!=or)Z5U|5K1{L{~b8LSWWr8#})Yx>?2;syP`=)gylwD
zuSr08&xhgZgA5f@i+6Sw8#v`+mEe_iCHd)_dcHe$?<TGWL9;2>KqBlAEbl?Q8u}U3
zeG9<c7xr-wug0|w6(B0U>9;hjFd74SNi{9v(x*C#$L<fx^Y(nj`UJYv%=gdqSn-0y
z?%hrRe+~v+!;8mUuW0f=hfV@8I+RlTHZq}n^q%Q-_u<04Vm2k_TM4vPD2(WsHoii=
z1Pud9q(!9G{@b5Jx5$V7fnQ;+?-9grSfNsJaQ^KH!~_}~3Lr4y)#o<{B0~6ELI&~8
zn(ur365kM7n}eDQx)Nb`e8fv;%EXt_JWTmWe?x#C6EDk&6_983%Ex(>Fy)A?!JQC~
z-Q(MET+;IV`B|}3^&S@mkB})CdV#s@Y8St4qo)oZJ}@cYI3=Iex=Qhk3zde_O`+Q>
zJCpT3E;8jwedUyV_)a0gz}8Ez!bWWU1xV#7KR@eWuY}$lKL>-fFMc!nppnwjHsWR?
z{#rS>#XPLO@~2IPY(#@)Rz_G5uIrZ^aDiLu-)R}1)=LTb=XEA}J68W`@)!9ct86eJ
z2$N<eU@!(a!ND6%TZ<95xa8JIvZMS|-QF&r8KRy5t_ywo9_vqAqbmP?{oWz^DpdL~
zWbZ~TisgNztAwzf<G32Qpa#W$N|*~lQR&Dr;zZ6@HTfS-XrelV-(Y|bI|L7p9#Lv_
zZ;AtQI6P3Y<TM%BvgTfe!rqo4*W?J|5Feln35Vd6-rKl)Ge!tt>>f}$G<Z|iVimzq
z1*mv$Jh6ROn1Jj&PB^ZKsJ0k<^p+j?=-AQEBPRuB0gz8KGcyk3cXGAVG_W!aecF~2
zunV;^i{mePO$K|;z5(12+9OBsjz2cxO6{<pA?BD4ggfMSjuKi^hi@GT8O{B!vI!@`
zCPQ|U)Q&H$<CH@~q$UU3){oTOZ_YA2F!rKj%@z?H)`lF7gXP3CNiRKfzplN@(plyb
z95|-$598#a1y(?hf<8x(zU&776%Sa(u=aN)RO<$Zh0UG1v2bQqg`^UpFw-b)kCEA(
z#pSp=i@k=H1%y1RylUG<zD3->AFcf<e=p9(qRl!SS)-}!Zj|_5_V#Ie=GxG@gKX%o
z)fI7(;Su1n|F-s>+7}g6QDZX9t8S1;zkI&?E-#p!6OtirFRvc!Qpo#o`(r>RRWxo7
z$R7?eZ2h+#<6Q@GLyw~HW?*Eb@q^W9c0RCS%RPDO>T_?0{2o4>D7{)k(c86=X-(j0
zvG<y?GC>!wMg{3b9T6PS4+kc*yJ1K6!a>1ieuTcdt`6|_7>F-Y_d8BACT8Yo^XR|3
z4ie(gEDQ_|ihkJvJsbv4SOfVn^^UfI-muT^8!@T(?c29#X1={C$12n1IT(Ss9gNUM
zxn5weP27OfoD+^mqu{{R?RXKqPARw@eMgF`Px*m)02`3VZQ|ucJ1wQfjs=56!Z`$*
za8rOD2Fe-^($_aG7p$$5)DwogN=o$k+Rp5f;bD;q&HXUs%5vlI2>U5yad8Dz+UKV5
zK4;(x!Z35o(EI4&S8iM6A>xjWILWpux4nLV+8SsL$DL~{_gt#pary~8DJA+e8oJ1P
zjtI!H16neIIuo;7G1E5}%L`Oa^PBI(Tf32oByc;n)mBf6i|INLF@P}e^I^<J{Bcea
zu%%gOPB6b0UjCxV4^rWF=u-95Ob`aJAKWFnD$~nD+qxp;lh=|d!HiZ0<x}#tt>+Cb
zr)I>E-NGf6m9gl*PRXF5a)6P6R3{N$=X@c+_+sT|hl%X&C*Vi89SSpC2Gs*f9g&_(
zOaVDYd*h>3woxkSOjfIiS{X#<{mOtPp-&0+r{E0gN^xx1(78ZCX<2zUio`=ls~=7A
z8SiDvyYlCIUo{3p&U~9KI_Gxss720KXs2@mbg%Zem>*VBLOB`U`eekm#wNqoP3ke0
zDVrS2tgwGBQz><yJKuZCPJH=amv;tGd?67L%p`OskRaAnVh=%#6cY&wW$%5}M36^}
z4}11S)&rNz;_mT#ZXNrl7T~HdyR2)+7A`P0KknQhhVF1}z_tlmX>kHL6VmcQRhW8>
z<bcqa6Vk-fkev>pFQC^ph$C?aCnO{g%RasuxK~keai8&0I-iNW60-F5YvmsSxUk5X
zys_OA<~`v^EMsxKL#kD{P|?=mtL)sq-4gqm?|f4f*eh?|hQp@&<F!#`S(F_SShbJ(
zdv^EZHj@$ruT;Jd)qKM1p{t$*-E0yrxw}!?!s6l<d%3eBm^G4`$<^nPhIp3v&-&p6
zU9@~fpLB~VM^C9oPN|hrk^BV^T};eNlW%k%X$wi3u72WUY+kA*XW-aO&B2osKzmI5
zOlc2er0Kr4jd1ud^U5~&J16n3!i{8OC0}%%sDPCcKfme*#1eSFQ8G|Rq=rd+M*Xpc
z(#>3>DOsri;YYl?!-FUr#gi-yEv2|$wY8G`S{qiAc&bjFD6M0U$3Dcoan!cE$X2YR
zby)jwo)t07Na=2Q=F0AU@CGd%v730aFs_Utn&aW%!GE$%zPUaw9&-EC&#?!3DVGX5
zZ5yi4+IHCIwh1Rki)l0t*jkJ2)JSg{dZrFj@T;sAYPZ*IXJ2Pw>m-fqNs%c`s{@EF
zj!UWZY;JqeC$i9OsqmnBvkUOI8f!Ea1Si!Kdu8*$>VVQA*Gyus+<tpA(W0JnvvG<0
zXkcpEU@IkMrfFg0;$T7X<YQ!cyRw|ICDRx<3J{C=;6IPb+0v^-sStIMo-;^Fy`o2;
zKi<SpSa|1_0p|vDs_+mO*|+v|YFF7l3|kbb&`|f^*W52=!f=t{rNj$A9Np4#|HD2N
zXLt1XS6KGj>b+{c*+i9jf~Vw->{EJwQ(@tkdghXkGPQeMzvc9Iy_DO^Ws=&iqV;j~
zH*ccy_@oPQW|FVb9^;J-78UJROLKYm?)meZ1x^8?1&V6NZw-&e(~vq0w`iIkrlC%X
zQkZ7FAEa!vOEZH0@&S^cSU`D0tL^jbz!K9VX&h|fr)K*D$7|P!SjDq^Y{;|MzN2S)
z#UF3*VQgIWKbi=&5bl(YK~!@{<{I$)_9&vYDkQ}D3&We(i{2k4wOvC*B{r8x2QfA3
z1SoPaM@ws_KAx2S{>!m_4I4(wZ|3=awDnB1vciwSoeOlDaZxu^Z?@N;u4{UBK>S(F
zc8;upipHnw6>4KnK08h6jG!d09#+ReFYbS&@$^6T3e54HYzRVtQEnGH$fM|!A8tF2
zB%X>X?`b(v=F6RchnhezJ39vp&xxF@J4PQ9EV7*XHMT?ZHMeJRkk(y>@jH`yUr+7L
zp4hw5v-s2D_rV-(yqtEw?{KtR6i^<v=zVuIb&<Kl+wg<`8b!u}uUw7QXsB|;+M!F^
zDRjqkFO?5*_2u)@Yu4UQ8`@h^m~qLa@|`HRgvsj~=XdvXEAob&x@E`|ncVi%!Y61g
z?o^-W+HGEemTx~xx{fk^Oy&C>w8sEALK#gZ05=Lm2uZ9Wc6GJ21%J}2&&$Gr_O9Yi
z%h=C5{uC%^92BPZ#h-~jbX7Z=w91c~jeIDqT-AQVMtMQ`DyP6zR@cZc?vr11|1MP=
z37`p|n>&@i{Sh_$Q}Mytnklu~hAyEZO`Bp(Q%m;(>r>+E1>^;Jj(I4_D0wJ(kR1a%
z7WI|rX)k9Piqr;EzYo@CU{Bl=x93#a2HNCBm7}#CsjK(nd5g}e<GKw)lCqs2-0b-5
zy0)vNhkueV_fB~ct{ySBdFEy3$cx|__Va4OLiOKl4L#ig@fnYj>>Uh`(^omkQ$;!k
z3TX^}sk8H^Ug}7${>^+*r!mrJu0OPBsSP*3d6WhNOIGX46BLo#=I@ktgh=V-i{9?V
z5hGaV>V}$wD4^b8UVKT(G91}s2d1!*wtr+V>smC}=9to<d9cAktus@sI{TKEEjRmD
z387bZj%OGZO4NMojZM-DKV>pn^oQ2e>(_n0sFz~=nD|PRHEU^$<AeiAq*N=8^J?2f
z3M^W;)k=irZ><)uU6)!oN=2plgO#_DQGTc6v+I;u<CZlf$3xwp*#ek!U+6xxxaDvB
z`1He{FCHF$c)CVl4J~O`LC2<&KHKX3k3Qs|a(2~SA0U%d>r)=vW<S4eM&)dbww9CH
z(4jTg!Vh!piC-7LhxS&DM7)KeTXp8b>&FhOw;v*TL{iomeki%8_@Uzh!zRz4N3cZ{
zl^F0PI|p80yE6V_SNKUbgTj|jC^AYar`-e_op=(OhMLv2B$ZSS9rD9cS?PR%hU8#z
z&c$>tfAEoixz#?SR6P<37Uj5%nxIpoQ(oe_Z>GQ%gSN!448&$4_d_<EytTkjn_-Wp
znqq9`Rvqbh@x`Ezf`SoGpN#IMUAUN&yQ~}5^gzd(QWuozvdzNN!s4A&_;t!Yo`g61
z50<W_UKgLa^^M<q<R)M8Ta{`2uru@L;}BNoe%ICGfp6_-_~ZrA->j)}sbMZ#8y%c)
zAws3KTA(jnTt_;jb5F?m>TO!Lv_r*Ie+}G7+rt&p(Jd1zxQbL<;4$<eq^#nzIR8Bx
z&e$)1hL#%MY!m++u^9iswn);OTKc#Iani(3b==>U7uRomb-$859|QTBkjRgZFRib(
zZl?FwHx*-D<1bLVCww>r6(wfNN%=AlO<(?7;qg?V_(E}_n9%gbL_*5!1E}bfmKw`T
zwYC?UhNruWuI0;+o0fiO#W_u?mabi?K6uf@Am9R!nuxSbPMh4!ToFfd#~Z-!H9=_A
zy&B~aVuMgr>qc-%Z~PqNCSq<{`WU_b-tyA=Rbu+rlMnlU4sWZvlNMFz_UHV^X;Dkj
z#s|-O*Y&4cQy;QlVwE9v%UDqAKT?zWy=jiwv#7A;y*a%<-gzp@o<*E3a`AP4_?AtF
zM2xlv>Nl9^${#rzF3nxI=@wg=OO1k){MBc@+|o07Et6yy&*j;r4~!G6@*CQdd>lX2
z9`F)d81fzCKOj-9d+Yhn-{Fr#s%Xoc)cnkh(>89`wjY)E%`k3y#obfSDPdv|dXXZ`
z8T!i5QBy6f{zZ#Tt4Wz=+($MgG^w_wF|JZlS-acO#g3a_pp4~3raNse7V_Grs-~gg
zE*sI)(*t9>EVO`@mIE+~I&>LW^Y_yY)`!o0DOoy|{zZTOi->z#cZfd?kzgmzIr%<g
zi}$GYk`0wE&!))Gr#X(tj;N&?)z%wnORgpbDWB`S%k22>==_M;c&G%mUq@lQ-B+Rg
zukzR0qP_%pDf?be=O37hkx`B1YuRC2$!)h|mE(kW?$no4*<fHQ{`vf#kl^4~|7eki
zQ*GrN>FDUZ;*&1dV=d;LGA}*bUZOGEtJ9yxv+jp|yXkG)9XWS3<KO3<jwluv-axh-
zD*kgl&qO!DMgC!5etPZL$Q{~aBWFsr7*SfWptSO%y|J@<6;p{|&hXnD>p>gtZ8>8?
zbgPnik8KOg<NGd{W1rGBzsbzwXvwAZ?sE^%z1`v$N9?pVj606biU*WFU8K|hGFxeE
z*RlD=lb608HLa3-q-=#hR$s@>%(Iet8BkXGT%)C%_mLm-@VF^%y{g4(c9HS=y`}OH
z_LK<Bn%67}q@~MebRRTmuVA<NMX_h+{+!k;aavanFk#F*9V3gYySB&PTK>p~hr`*z
z>+fegE_ssRwi(;(G!qBU`z`x+y_Xd{YV+}S-m~7yOWxo1=d8!#oUDPmCo=YOkzF1C
z^T%r|M%F19H&Whgz@PDz`_r_LHe|o<Q&36We5+mDem+pNDCX}=(s1x3S|$p7RVF`D
zD!*ATVsCQnr<jB-N~fEWz%sewWFvf<Jw&+jdN*HbTAjH3s0E*7Qc_Gt)9(FM#n%rg
zeuHoJCiE;gw2;B2-@JssF?wi}?fmN`iIh(fefA8ss^&C4Th8FmkH+?2sLmTMI<I$<
zX<m8yX*}KQ&`mFoJ+pHz@zSmZdt20rmpRRR*7I_By_{JvKrZ!!NA*Xa(GnVFn}kH_
z#;J}sCf3so(ZL5N*zLNsfF$>Nnm)FW^2#$!cbDO@C0K+Ntmc_|0`NTOowur*Ptk5T
z`q4f&Yo(;5rzUyq4i~!p%_e!thQ7ARSd1e*8#v;Kn_PHN7eSX;Uk-(UI>&nK_O10~
z-fQ)nJ_lb&iVT?v0h3wK33&yo{(s>)+lkBgtW-F`z&CCvGIhu?Y9)x2y&IVV+5T0h
zz8B?L6JT{nV0Y;7@1L3k_#()GDz(t>$FJXt<)lM0OA01!%qZ(3aJ;_t*f%v?GUJTa
zU7pt4{O0fB6-SYq?j$cCczd(H^J$s!qBcX+xEZvJEPax??7V;YsYoNGJ|+{FawsT<
z&FOKygq!Sk(ypc7tU4dj-rRsuj5h^|7uj5~q_S>Y-goIG3d@`(;@Px#c5UB>T^GK1
zkUoseKb86U7vo+fCUbvg|H7{Xs;d>|pL>-uaX*kqcuPjz53=i4Q~CXVFSyipYUM4p
z`SBt<On66Ka;;f2U$#$d+x6xq4KD0_T=kP1s@@6)f5Gx#qrUjkjlf<+QWQ%wfO&)E
zg2@xmQ5UTe^K~Bf-n(PWq<M!DLASA&-|wItj|5!dG2NmMpda>!<Md$=u|mcGv{8~N
z&Gb*7-y!py@4rX&R*tHqXTJO&sIjo>D3f_Gw*Vx&oa^`T`Qh~CZw(K(iklXGHO~O(
z4p>F^O~!lyqG%AwAV7LjKQKExTgB%E<AfoU9i&}IQbD3b*VOuBm#*6EBP*6y1YNwg
z%<Ja%V{@)m{K`G_W85{cyvGrqu-omIi;AB{u=2>TZQ2BEcoBOIC{Yn<Jzr@cWRk>B
z&bcb`;p<G8xM2kCT_$Zx(EPj37ZGwITtY2lv+K?qj>zEnHUs|ohkhbT&fOkb6o9Bc
zOLK!Mt^1{=rMGWiO)@t(2Md<cVesd4|1l`VQ@;#{OJ3A7Fz|q25WXOg0L_!PHcx+D
zPp6|;YgUkaz*%KPY~{6q-T^f|03iT?7DzfERTu(fC@BZWR6r?-6f5+!et`JY`g0Fp
zH<SPKtBG(1f)R$O7!5OIMxhd?NoSyfz}G^kK7`eYosUX%5O|vVbW&D}y@fnp02(=E
z`GU3P!-d-Vd>^kxir1s=`}sBRvZ3tE*|I3{ST0;WK}Sxh1sHzY+AI1h^aT#G_+@UU
z>chSew;yt7{s$wy>|ve-h*#5D{b)zJVb;Kmn-o`&Udjd%TwU+NUwx|lF-Cm|(0zH2
z<`v<psw&Z-t=qRlN2?|*qnBnPth!FHZ~oGl*a<FncFc(Gw9nNynv#h7i208w$!Ot|
zXtBZk?Xk=*pVJjD1H5}Sv|qnxTkV_p$O=OK4twcvhw1KaLgTiwn*u3Tj!T*i2o#PL
zlI39utA2DHHar+B&#bijXLhkE6@~;2&P7KHr0+YG-dWV`F;y=OO3(}E5S*c~F@k2U
zY!yJt3b+Eqkjcpr`E94{n{9GhI%xesxK_P;C%hsZ2FL^|yRW|=FrP5Ds9G2<C!+&1
z$y(blB*g~r;bxNGZxtwWoke<a#%IY^!ngN=dxFHJM?tNW)wU%sY**{N3S*Wt9(l#X
zRQRR!t2g=5*ytC>I_R{Z2?3Db?p{j_2jY>@tiSi`N6RpfJIMM#*%7B4b;0G!m*az4
zx@i3%i~=ifW<B0>gTLYZ`zO$;Bd!Tt<@b;a<R4F(x7M{5agx5f+hO+AvI^LPKDmnW
zn&XfA5nWo)$wK<8{s4D3?Ifk)c`<N{j0~pE=zDLvAza5I|HBn#6@5Luf$D~e5h?Fi
z{-VXB_Fqb+Y`l5TpZguGrlO((^~(w%6)R-3?<uC)=+M+A=;9(&Rb8|4dsBt@vPntf
zOO;IyNdz{2PhrUep&FL>u1K+m9T$|b&HY5>bNo^a9;Pk>+2d0hD{GssJ9xy`@RWnk
z2fhloxD~`|&{+Pq7Hig2?WyOP@F5+_m3vg673;=>_39E>kHd2!M*o48Cknd!9^G+G
zdW(9((o|4jwF?98v%&^2BmmqY;rWjAVb~YYmg5!$fu1Nx@m!$9C8<|&ntmK<(3N8*
zmdI`pk%(fXW#j<WN6a_v>RRq}?*j7Q{Hj_}kX(9?%FnUyKCxgE%lqX1dkie#K&(D)
zw$QtKgYU8x^audfi|em($?8|Br_o&|`Hc?m%rhT5(Rm#UzaifS9?VO?P+mxuXPh=R
z=gF*Ft$H$ieS*%Sd_bPzqSPL2Ne&aPHZH!t%g8yI{{FGNtIMqQ;e!V~S-Oc@(ee)5
z2P<cLO5R-5ztmpxik18C3Rv80UHSmv+{1_3))Yd0w4LT?B8SzX=mXY)v#nK1znz3P
zf7)<osz4x6v%Jr`^ka$dvYa%|3kRy5dmVOh&kihDvHolt>5!1VC@8r`pQwcd$EJQ6
zi_p)J;~Ggxm$-Y~hNUmC-8j9nJQlJtvQ&`BS=)l5S8s8TGnkpyTiL@V4~ogKtxt?9
z8Tt6&&`rl5&mH$2|N93__yY#v%+$D<!=&i;X-SlLe;=rBx}$7pxsR*p6HArT<swr|
zL5-Ij!!U9JkMe+u;@id6)ht&<L*KtKDgS)(SJ+^Zy6-(y)zQJ1cZBP&RXU<|p1Ac-
z#(i|;sj`)d9BlPgTqC$T;OYuqs>t@%i(f)TQU2#Of<_y0hsQiFHWo^ENCt<)&q!(#
z{G=NyY>=q6?|PYZ0fm`oBu2!dzCZ~Q5)$H>qG-OGTVYl@<H>1WxmJFS;woGFTw`-*
zpFGgtpFU}?EKEnVnyvOhplw5!<S%p^u=ryOG)<-ds;FV%rxygBI_2%?_-ch`;_WfW
zceRFf3Cq)dGHmZZz!ZT~!iHC+RC<+boz1VT15!1wdbad5s}8*vw3v4xs?dmpFOI>F
z9}o2PA;-uO_X`+jfa=I(P}R_wv6g9;QBB><TQpDYXYan<v<{^tb3g*^F}U9#7zk-6
zRu`w<Rav2O(W~FKE(i*5eyH+rX-eZ(dCbZ!Hd6zR6E;tNXzehsm+B1u3QXY=x{?i;
zdr@nfxRxT*{z|Kob*a8n=(EqCcL!G9MCB6brNX&^a3UWRhUzO<mdutWL!rS-xzHGC
zZEby!`3BeCesh*?8GN0d_kt@oQT`5)Kbc||_dNy9j8(AKCg;jz^$rAT1A<KG(cH2l
z=eDcJUYDN@eT)(*ymY8$M?5d9Szzl!WF)Kn_@Q@?;jEh5zD}vc`PXQbS^nE7g_LFQ
zxJ#d+l9nPZ2LJ><=fF;eg`LRxp*UJ<>*msD9FYAqTUzqw6Zd?&RJ$^nMya_0$2PH2
zUR>a*w8NF2GTJC*#YDB{Guh1Cv^)v>U5vbDquZPa8tr<*7M`nDuY#T!>nw1$vP#Uf
zz?6}L$FF^GP)RkCn`ink(iy*CpV0jBT61hj+b!a6?t+aKO)s3R5lx@n$Dq{W`1@x$
zr^)BO#m)BBuNEg7o4@~UF(e&l{3!&zFbElx*I1ES`@*jwO1KLBscUQt-BnE2^z$;c
zCsOJ^e_T(Mj+yl_^ORBADzIj2+1L;tf4=L6Vxp`r`$&1!&0#&5*lbF=C_h6JSL&q5
z{DrQY;G(g8Lp>J!DYm^?Q6^C2!BZ)BdHGc&SGtlfeUaL7EV%VG=Fao6s6OP_X=<Bc
z05l#Wz>`tvbYcTU;erEC^%nYg1k;Y$vPr6^<=HRxA1|ATn@@jj5an&e!lJd(OchL3
z%kw}Wdh_mGgg2<e!_nBNujiAJy?f`*^rr>5SwLQIkgNyK=X)U_u~<k{6mrNx@=w&4
zj;mw@R?!Q-eGtCf@5e?RZM#Z|zK5P|1(Bl%xTijWC7Vrv6sw%IZEoVp_7{up-!{IP
zU`%Oej?ShDWTOr--d?g4ikDGRI#NU)3_lP&{=40Q@^cTppe`MDonV+`;E5*6xm;YI
zBsOH%gomdz(}hSO`Qj7;F|Y$Xi;U!3zkV-7ry|@)mv~Zd3A+?jCMk34)2RiI#7+G2
zYu-(!tWwmJJGud4F2U#Xw%}D+OI!t$PG3}_*CbK1lZ4&}*XX*{HjC?oid6&G`SAka
z&sLI}@YQfhI#NZhRn7V`{G(p}V0qEIH|R={-1Sw<GZTU-WIj4lyq(%r|4)E_0K3_T
zsp$d;+p44bc$=E=hn<y7oH7!B#+Qrde~4PBmdU;V;&71{qLIca_Yqs09#2!YRfdmS
zShQ*5v$WKR|9vdSP4W=s4j=mk+{;s@4!V)L+d{W3v)<I8(qKjm3`R5Dxq}dny<sO#
zNr9=Fhu{t|t%!QhcGD<cr*zXa{lag%q(Ul#!}L8e4OaW~vNRMGsWc4*c<2HL=^NLQ
zJ9zpug{H4E={`;juUQwbuIi@x@c0XR!-oz<0>S>_+P73BHc13O7O$3_T)sT>{qw$`
z=Iu8f8qE3}9&Zn%1Us?C&ql96e<9m8b!2LLzNfCDI%NCkLlgsJ3hNl8VgaRo5vj^D
zWS#kNw3lz4c(Q!cNTjS`(~OSRZoO8gv9B+W3`c4V^z8aqo$4^9#+_2a&Cj2{sQZ_s
zOST;!^Ia}*E;rVH1;-|lRZ!a>^0}}UV4xUBjt+r$4Y}r{g|Yd)35h#Z3p3U1LLn9m
z?0(dzlBv{${hXy}Nzolco7sc21m%x;lsV<9zZ8<Dr;1)hud=5;UwgJ^;%D_~S>}5-
z?%A)tSJl$0dB)sz5jrfaHl#J`T3YY*b}f0w_|b+|DaEVDS^oT5jts|279_i0w`A>#
zFlM`|cBjIx)4-7ab8`u(Z`=TD@jtaQrDp^JaatRvz-Eh{Y`KU!0||g&LP1VXzWNA1
zD<<_oVUK@<PSty(mC~Z;0_rOVkrx=*SzTLu=k{%CKRNCr)9;`&WLjN*jNd0XD#~}T
z>sRon+Zp8B?d$x8bB%+!1MjyRdyLgg51dS6MluCX;A~TAoF~Wdh`4^6;fr9l*zCo7
zg`&K!OIb@rBC@h=oEn?3WN)h|SL{S#0*&y8@{2NC{~ru2O3fXy+mPO>3nK~Lr{@~y
zdBfsAW}*jp$c~BcAFXAH+H9YjcZJ!^(U3*TFlx4Gq_!F*EWs@cDg9fmDpRH<2N}`x
z=YfBD(|d%AU<JxTRR{{#%zNS}aP!4Ke?r7TSQazsfx${tnFidZVwf~hB6_Q(*S_f+
z%HVfQ&k6Jj+#(~nt<*n`<5l-Ktwwr}imGbqnq6$+sI%-U49a)#+cfn5_&w_Ui~sL>
z8{&Dn$Q|)QPo?b8`{3vNu8l~D7;ZA4Z@Gz`!RNy+(ubwl0LPz+D;vwH0VqfIHp)+-
z=EPYMK6r2m`>(sH%GLY%01?m`^(63Lk}G}~9*Ckbg;0WgNGDW6FrLEA2oKV0_+{Y@
zGb{FPg-$1#)~if_*xl`&rgl`+^@d#meVL7=4-zkuE3(=!&IK_8=pc(vT+>W-d^R$%
z+gX!dcqgaA!p0|sOPnBpcVOr2SrMAVVvsjBaj3=ZF_%Q+oS?3pF1y{9O6}CgI}2my
zoU>NGLno@s|D%Quv^i<N&eGR)D_a?_f(TDX!b-J4isa-8ypIkC2n2#T?Vz~jYam^}
zFVZn0C%eEzBAsoNe!4ed1%7!?@u#%to7KO6|Fyt<r{H<$)3L?PK$tW7AWq-xjf<FO
z^0M<@eb<vjY1#bM(})<gb{6*(6w*-fAb$&TFSHhcqZ<T2or~+U6u((6VL!jo_H)59
zJ+{3laQ%)lJR4lD$nmiJu@lu3sMwyomM3nzrh@ON@~yMA%4eMUoNyGk;wU0w`XH>?
zsCLAme22^&_KHFrm%a^%ZIE*20q}Tda9S=mi6~mQc4}PUijchcnTA#V6>=-kHbD4=
z5`PX-uy30s?M64snv_iM?eSHpezCp5{y+=C>ji!6WF$kHuiNH4JAU?zb&i1&eO9XH
z+VIj~-8xI1ys<k@9A;mxlpM>waD&JEVViOiFLSq^JQlB@)-!J%r7F!dA7>@;#xQIs
zIZ)X0RKG20Y}C|Ie&J`<rTY|9k6{QNZlcDulINXHI?hE7GRPy5*lbmgDC!1XuOCKH
z&fGX9cA?;nh<mTVn?70h$tO#9H2Q;+_KBHv+NxbR=km@G$~^&ko+^;tY)ZnX03!!C
z{MGYKDt5nP<xJ1k>FWGbK8?;bEr4gH%reU}p0~)<cOfU^1hx;N#>dqNF&9x$=U++{
z+O+t@z&m1R-igntuMf=aliYZ!4(MK|edv=yatOhoa|K;*wPm>xP|xt`%&xB^ZY#xi
za1S#9^L@kgB?7EQOR+Y24E-jzWaR2q&KB|4>$z|C=@CVV<Yqvg%7I*eY`0VamN}iD
zxM0J$YybPZsFK$Ki*<TXRyJ}kh?<NcC^4P6#w%Bn!5LKJyzT`o<UmDIwA?EVnJc<0
z4+s$tYyn_*;R_!rx^eNLg0J4n1LMFIxrC3EoxvXosP+zWqhnjShMB?SCua1@ES!sr
zAGfSHEA9O{@=0wA!SXY2!b20W{%5!kTwT5TL8dkYOTCVX#C^^s2EZ4l>MN-Gl7uF|
zFuXe%#kT!IFABibKmwb;R=?0}znKASkF#+&*Z1lp7lFY0JMrPzRK$(t_;{%Tt<;xZ
zF8t;uYHm#$V1Y^`QW`npC(R9NzuzCOjKusQ`tg|StD!frckb-!Y)#oo`d<7;*=6zx
zabqVcmip7gnC+k6o@YKvr}l0n>S9BG^2(J>F4(DI#`wR0dU`36;^Or~uE?VzBA*Hi
zz5geux3F2tMi=@zplI7vHwS)bmtfy!erQHOXEtcK>;gYOnOxArNg@TZ^_z+QhGN4G
z7Nj2sp%!Y<IVdax-w{B*FpQLe&oL^h3knoD>6OZD25c}kHptDtL+k>FT{}zwKpKfQ
zVY9eR;L=povaLbktJwPjBzC_7LtQrB#+{dEV<gVTWE^89cP*YmaY>EJQ^xODd+uC^
zJ*%3>d9?O62bqEMD|l-=_6-a`!MFY&5MYRZ+^0(TT5Lv@j~x@GY(g*QFTl>h0mJNm
zjH!jJ<*gr8O`JE-L&Q<QwtzLs^wvpwgd1@MX^^WAe6rdXE0m0m6W8{jrzHWN8$K4b
z3$@rk)%ojC1es0X5Iq%{jrvIW+DOX2KzjL-$?p>-zec}(>T>uq9&?A4S2$yT@TWlS
zHYuU;cYLg*w|%K*p+&Q>S9yqMu2q6>Euz)nEnzIUN|CY5I>jY&?#3s=0`QjyP1(5F
zuFre2$3`lgc{L5`W~`M|IDu<+B7cqg5G-YYslZDil``W$byq@SVufxaPfO2qdwvtm
zSR=?&_5G+gF!uu|$9iSR{kEDd#u3`d6wg~r^W0BJRHQI%r6eYS>gJ!$@fN&compP%
z17=OW<oR0gO3IxMB(e1-YzU5;Y+|s_h5PXIe%G7@!zlAPT8sTCZfvh4(vY%RpZF+-
z%xn*D{joEOod?ajO%AnRz2#o6pi}lQmn2BY&-r94)fgeC&;AaYtoR3rh(#KlFl`kg
z8iIoW<(b|Rr}wY7#Kdh=>-qwKpYP1IP`(o|SVD@BXjm)(JLc-P^u=hEU+3@0yV^0I
zqdk$KAxujhOilY<{LMre*`bMU+QFP{d1JfvfV5XlghV8zc)xF{c+<LqKgtW=feW-9
za8K2{m}8Vu_d*tnq|j}g+<K^KVnI^4yjM$TG8E%jS-Y54(+E}=w|Yc=8J;$^PBxwT
zKS)%Y(JRY;X5<@7etkM#(+Kbe*45WwR3Y(z&a3$SU9J##pe>m<W%@2(Dd<eQ8z>rb
zV^Y6+#Q5L~Z3dY(TG|&o?wK6Ukx&wfr{y|{`yyd_IAOBs5SqyH9XojcKVlL^Sr@ud
z;yCUwyw1Yg2CZQ~=B%qBecsHY=b9SL$8TIc^eZB76kc0u=#Lp1hM_vwtfxN~<QO`|
z%zbdPL;`zjgOrW@bfwV6zl9pvvH9knGA+}F-uz;*dyY>|ym<I<-O52JJXnShg`XiW
zen={(J14IdlhJM7$iPDCZj9QXu<%61-d#6)Wa50@;6)#pkZ55-qGKc*z4b5W+*qW%
zPVj}w1G|xX_f|)KAv){@9!`(L^f6D%{AmR5<+ZzKv-+4AcAqSv9DlhVOPey-p7%Ow
z3;AE^4lo4LFE6O(%IOfRxNa0yP)+7M^Mb!L%edmmx6>=@<YgMHlQxyDsR>zTn@QhU
z0G)NE5cNzlKVB@clBZ~1_VNr3X{5Nb?eN0N!aq%dgsbIxmV<xRF$eHFtuY6@HHmkr
zkOgLzysgUbn8|4h6~Im%nWyo@YngKqe-U}W>6K`jC@LGv@bRg&E1*dM-sC$Uo4IeF
zmVVuzLrDi+QXCf(+X&K73*?fk1APXoXI<xh4xtu@WhIy13ps>h?T&Y*k(d&5cy0eY
zI$=~Vr!%f;%ASZD>w7(%Yd7corQy@(5zOHEv$&|s1k~aLvE)B>lHt)fIM){vwS0^u
zLuMmIvXSralo@TgQp9)bu1aR;*!=vILh;g%%$>Q?c5S*Nh4@B+*EJoHA)cOnn11c5
zs|ADC4*6Ui^duGqZSuz2*1}?y0rxJSFB6s1i8nd+5PhDmn`)-<W0QxAV_T|R<(D;6
zIB2$P*%#Oj7vAxmMY<Mf`%?942dE?A2!J&L$^)x?!P-)sU3#+I;jIQx(zX08HY5Z6
z%Z)F3>!?9@N0cW46Y?6u7E<9m$=rfRVlz)99Z51qGm19aI99^YO=rH5<_5n_3pI)7
zVTDu2*a2C$*GBg(!r0(gzlQ6#()P77R?0=W*Q3NMkGOfI2&+CU{Q8trORbfua;k8J
zcWfW|J)fQhF>kuf$N9Re`~WD}$U=VL-Pi;k<zDGLDC3b(=QBQ&(ZaLI?EDl1)`LCg
z-`PqLP0Nk^N&n1eCn2KMWMEk-Vnj!p-I-%>8A)KkGD0+#;TV7v9F171wvx)b#GLB$
zPXR&brJUNi{V(|#Xmgis9elo=oC$6)tt3m2S3v*#Z}LHt!!q%`sl!fCD(4u9XX($p
z+Eizsi=Ma0G~MmK2owuNGK<g*#R_H4J*7^Gs5j40d01fyl?<8kb@BH(q~L$FIzJ^z
zt0!lSX@ew>;XPuOzH3`{mkoqwo!SkZ2k5{K-{n6*KQ(~o=R=VU<vUo~$W0*j3dV}D
zeH-TlH1B}^Uk<LX;q;O;xBHg|U>GAph_H+Ro``yBkWpryLCiEA5nyI`d>UVYkUTo7
zy)#~oK~a0cz!)`xHUv(1e1v^k1F+PJpyx*zToRPYeL1_lXB{mY5APl>UKoYS+a|%v
zYW%MQ#G6GU379?fiTDM{Z>uVBjn@WZm}C#nfKAoePFQka(ygEJj;CC|k)ifKi<eVz
zAG;N&j;N|u!!Jauhyiv(a5Ib7?%|VeMEdZ;!U9FE`SAZ#g9qNYL2;U*<>?`j03RGe
z7!3GQ^x2bBQ!V6j3ALf2riA6WM&EoBi9AYP&)Jusu-$^<12@3W$A|bwQ+X@Cnz}k7
ztRAXx1hByG5>Qp|f;0wVC!p$z;e+27ET%SMQ1JuJ?L^)jIA;<Gh%l7f|HQQ1c-&e?
zeuMWsCAMF0h<Gb)SJz+QkLXD&wF<u4hiJtB8167Z!6w7t#P;=k(ySX0Zw*&!jU7IQ
z2)23eIcL}qG4_shFTC*#WVC1gtz00!krq3g9jQ$6WxQ3)tmi@?(A&tv^DHDJ2c08!
zV!R9k3(JdOL_@Aq!luV3t?@pSFgye8)P8e^uKqA=IZ{$?5dOo^0<>)-pEQ5zMHv$5
zzdc6iNLt7`4!W}T9vgidk%Lul2{D4pQLT!Affz&(J*xf%cfu9qiD0O41!dqC(ui5O
z_Pqy(vhoH%m1~0j)&o6GQ)`J+bMyA?nudneet<?`?8>1ubBB<*`6@+x{<YWgt9pEp
z|232l1-C5^2{v=VRrkVoywt;`Yz!LqGRh(t(R*+Z;d{HsE4La0VLjo-D5nf;kzu7H
zE;d+-)D!l~%X`C@MO<t+pG)&!N`6ewtF0U>0_r6;*Op53<RI#T=<3%a3J?9ZzW`h_
zqI*x^Qu=_kKw<uS9ASUO)rQ9bH6{eZxKun|#+WHaTy5A&aO3y6c&O3h-ckBLb;SFL
zCq~X=Fiq9nO>u4ve`PWMy=HLT!QG2!4=GvM%g|(#UYWnUjq({%@-&FG;naJXDbR&`
z7+ER{K$lAXj5yS2)T8DKTd)GGL?T?xm}Wo(7Qv~IDV8$p3gUuM-Hxmo_X^(+1R;pI
za(@8~drLRUVIaXm9*+DtQ~-$4LkYKM&mL!h>If{@i-AH&U22TvxMKk+C@j%=xad*6
z5QhTz`(R%m*t*|CO&7qTKm)705C*>r9loCpyW7~e(<Uaklnh9D9uvJNW4(TTpLoCJ
z9_Br=lOnw*t9O<xX2R%0%xl?x6bWj5FiZgbMA$%%ksXpBV7Vpp?Ei#Y4v{`!BJDfV
z%u3QmAE61^f}OTmXd}~g5SPL(fySwNYD0j{Dx9P>k{SHF-3qAu*kWw~!alahZBK2n
z2_f=UzzYd$qULyk=|X{A*N~r(W?Gb=o!xAo%eraL5yQJ8NR_~41Q%pEs{#)ZcR)ph
ziEmbT3B6%BE)gMy9~_<zdv`E8I1V9_d)F!@aySOk#3;w$*y_AEG4!pd0C7C>p3Zn2
zp<Nxwivx<o$<7X|JJJ5jMvTIwdiTy98b5`_9#(jM;)B7zWCP4#xj~6qUQDPyuf5<Y
z5cosXq>wpA*#@Z)*n%7iC$6_C0+*nU)Q0;K()S*EI;Egy+-CSVU_$;r_z$i6e))LF
kU4W`U|Mb6Jj2+8#0veg8tX@8EAmK;#sQQsKr8EBj1)y^9EC2ui

diff --git a/vignettes/examples/timeseries/timeseries_classification_from_scratch/unnamed-chunk-13-1.png b/vignettes/examples/timeseries/timeseries_classification_from_scratch/unnamed-chunk-13-1.png
index a251d01ddeb83772f99e15966ceeff664f2a171e..497a999fbcb29aa62c72df9d8e49b6348793f034 100644
GIT binary patch
literal 58558
zcmZs?Wl)u0xIX;QNQ-oLcXy|BcPOEBNjC^c2-4Eh-2zG@4bt5p-5nyxyZoJVX5J6~
zHRCt~o4udC;=Zq1k?N}Q=qN-e5CowsD#&O;5bQVv!Q3Ikf}b2wsi1@Zo;fQRxIqvq
z_P>8HlPdua5JUkf%1CN^WgcaD8c*xb^{a6u=iaUgmwb3xYu*(#*;B>fVN4tqHIIvn
zQ=zY0%dD%7gIfzleaxe3Q#$qkY?^Qz=i=?f!qQUS@<(jv4<<PmKR-Qx;voBOwl7A=
zrU2Iejt7PDz(ol5zY+4}kK83G@!vS>ri4d=`EPV#MqrV^|IZL7xe#~gq@TCH@jUt{
zmHR=4ii+%}{F4Grt^R|W(K&mVGMxoM_^Wwwhk=ggII=Dz^p54#QdIF!(6{<_ln@H!
zkcXd5xKVkUnwn2vai2t<@VdITT{|qCzz;|;<Y}<|`^BE#m~#-Ud47FL{gjyaBDz<E
z>@<gOYYjR1WiLm}x5{>u&@Ks$q<`-hm8Z>g{ieNAfd-pWAn3CuCB=W^j+o;A{p?cC
z|M#t(|KDG9*9n(Ip!$1WMPsPGj{N^jMB@48g|j(tb;9Sz2%~ZF$GE2AzeM*RUNf2K
zI&uW&r*dvbkfsIwH`dMi;?AAVn6w<9eAcI<ZC}07lxp3+-o)5VM%}H-EX=s)i92)s
zpJ#ZGc_6?<Qj{xCwaED@Z^HKq-JNQriwpUSP;(tCx-p66^PsIQOT1{=LQMw;hmOtj
z=kU4e3<;Lq@HvuH@o#w+#T|M(aByS?Ex9<w%w!X8v;Iv@q>PkC(%p_9F1%#G#S)6-
zr@+Pcfe7Mq&gwYsDj#PGTQE3BccU3CA&CJTuAC91PY6Np-@h*_V<IF>3x-1qFnjTq
zhe+dUqx=g)RGv)D*_@<C(x<XDsf&EYWnD7_Sq*r}y?6Jg{Ev*18dqTEKeWjGSkqNP
zh)<MHBNc!~U%Of+S<<|=-0b_95SNjj-oDwDl$4~;g7@z|IS9}&A};@J6IoSNRSK+!
zpR(l5!6k*0=m~~QEO>|it+Iwvm<*Lh>2%;?RA6?;wu;`k=^E4Ye!f{&`tb9)Vyl<e
zwfElscK^E_DaBQtKP6vX;|wB0AM8v`{k7j4s$9-0Mhf^P3R?eFpnWYB{Xa`X5Body
zC#(N0BQ9N{;$1)o3Mcs7o}`!jSXxXg63V6T;5ZAxM#aI!eX?Qx>-}L}@_!EnIR>Qw
z^8btgfxsmu-2XW%<f>X)Q|Fug#KgqKs#ynHTUTls(r4tznE#!ct|wo87F|inV08xL
zcE8_Av&~qvecR|T1_lOCUrpejg3tcjiVv^q>g%PYq!ehwQpik$gM**G3U**74vP5R
z*jrd^_C&rwLL(UOZhQIi<<Ri(pW9P*cJ?Rfk;9j!rl!u$9GXc^LS6n-MJl3-+Mhpv
z_Tn;Eanebafw5-XTD)z74eD)dOH4@cxjlJ|h+UnYez)_JkS{~jq7O|;X;|DF1Kuq3
z{~p6*G|O#OO57(QW3punyj2xBFE6iuf0#K{d`V-YE(@OL3Ih&`#p~BqCM|uIT;g_S
zMn=q7$n8gOeqU_gAQ%jDs@`3gt+sl@LCSQOiw$<1oSag*XXO!Mo+ry22x2qEu_K;j
zDM?9NySw~OtF1jfJ=w%#o973Ihk;Fm;jgT$ta5X6S65f<=BiA-Id9m!$l6nMY+l$0
zFMe2Ak)_h`^15SX=T{-LAy!uJeqA3X1$k*0I9OZHF0k<N&Gz@dS!s5knVCsoeO-H3
zU)#%JKbSykQ7+E*i`(<wOHW7V`oZ#90RE5efA#IG9}6XHQacJe3j?CQa?iiKq+RNL
zeSWb$%;;@!EYdzDwHAhoe^b+)nMr~jY5D3^d}=EGix>GtMSN~M%0wSiQd66HeT!=I
z{NghwSKEA5R8(M~{&NIWyl)@AUA$X-sA=2HrX3s_f`t+XjlaH)ceF6kk?(Icp|t3a
zO@3!;VDQ~xiEn@MSMA9rMqyz-vMa6gF~abwiVfrCy?%d65+WiZ1_4JW_&$S!idtIr
zQ)T`AajO{z!4TTytYOFEX^+4gT;5|s7(2nGP&(z$XO(U?Ha3h*OiT<63JMDAyOuVC
zFZtgdnt`b_ij5i?dMPg6(HPPS^~1OtTT839Me2X~@#!^c*tPv=bTmFFFxy*ORi<rG
zd3zNG^-0OeEhlV*d3CdVixf-rg}ArHJ~VoBs^45vRjDZZit>IcqoPJnPfv5d8U5@c
zMAaYjftB(4>dLsy$9=8cziBkvs(>L!(?r<3wLBuCe>_*3nTe@Fcro+cA1Fx8sXUe4
zOf^TWEdoD&-SfXL)7619nBjxzCS^fDhlSi+ZT-W#9u$w&jzA|13kyrj>F;etMLT!r
zn=E*CvlRv&9v&v18?N8q3AHse@KR7ne9DrnJ8JN~b!25_jgRzxsge~|HyL@GrX}4<
z`*HKvFA^i-6g2`k=!IGi_--mMUvhDB8nk-Ve*JpY^wYG?GSXCsP>;qZLQ}^mgCDc8
z<K}3;lUm+-Bt1Dk-eti*6GfV-w$=6eVCMPr=Z1!cw6wJ0*Syj4=m=1(`rX5CH$6Q)
z=);E(`jsrgK0ysw-NyWg>1k>HcW1dyYwbo&j-n;i?KGiA!@pWf%i6ZLQ6N+@ku+PI
zGFIkFD%xw0vZ<dfn09HblJXdWY*(aWe%~xV{N)y7X$-5&%ME!*_~yk;+v_Qh`FMX=
z{rB(R*w|PeA{7Q4a1>L1^U2!V^94NI&37ZE!FFjYX?aZa&zEL1ge<e)6S$xMxZ0n(
zyu4&LYhPYmjH39E-FOo-nAp(Jz#c{}=HNiGJj>PlWBT^hgJNoYdA8Wt+qbPr$>Lo-
z=0#txsykTvo_}8Jzu)Mu>&86!6%=Q*Cu})@1usb^>ho@Yvd}Jj(@baPeX)Oyh87GY
z?=WIvA$K&)_#kT}TUPb3o@uZ%6cvy@Wc{)a_xR^S(d?i6h?G_@pI;Y+<z#HNv{s`&
z&9j#0e=9TcW_Y)@w>djGJ2fO$=XeE(iEg%9<j18y^q7g0%KWN!q`(ez?~YS=083Dg
zNiH_Ou&}VS6nLZY_U+r*N(PcM817cRN<#!FA=O$`R1_aSb<8yNbfv}f)gN71p^vYp
zoNiB6JWp2<g8n}KDK9Il(`Nx|2<bP!W9;Bii&VYk;EiQvbYH45xJU<ea_jimzaq61
zMSAFTu--whikOH97Rnhj121P3FAnQ3HZpY5O>COdX1BFLwcRBZ4`?kcq!kfqZEW=T
z)e{MN8~C{4!9nl855Mua;n>|06J|YbGdS7g?t|fuHEj<ZCoPMW(-P9s7;EwRUAWRj
ztbY=qqknw-b35FwVruh7_;?gfkZERYwDDtqpr&rE_ov61<&Z2V?gPKw%#i`b%1Yv*
zayQ;4^E;MPd_C$0OUnRKer$+J%+kpF&;0!S5h`lAqS_iyk7=-6w8Gf0NLg2HTQm2D
zo+u`FL<^fIu|kJqwPB{|!on4mj}S$Bg;cs!a$y4Q-W462-p8)-r1(A<>>(yXoVwsv
zKkif|U3IL6CdP?<Qa_j|3d6cdR^#m03VYFvkuRSNU@4!;+R$L9N#(+-D>8&jN2pUg
ziEHp!u0G29IKBYY{EvwbHFlsP1$@665~V<MFc!(mnLk25__AY}f>Y7mnU;>Og7j+}
zskn*sY9|jo;^I%gqRe7r2w|Xh%+1tnv3^S~OcsMJ{>~eJf9bTe>ywohT2^n7RZya{
zWiq`4MqbE%da<9@p*!u+z#;UPty^1h!^I}7xYUI-W2K;~K}+pXV(8DMqea1hWw5>o
zZ3_(-*Ts)8R2TVfl_VUDk3@)&CyJ77cq|teSKib<At7Np`->vO^k<wu*XPrswow%U
zSg$(03<(V#9vZs8KBUKsuFe#6yR@FRRZ84CSo{cg9J*G3F*m1J?eOK(I4+!ha^aqG
zFvrn|5wSZiyhOaJ)5bX(?v?^_!qjv~Zb@Ys7on^bBMx>H!2XinRPoI`((hl{2Y*tf
z%V%Kf<NW><S?q<`1*1>JY5tif?8ozGg#(uD3d&L%@DLKz31&MSE^#cic<TM@7<z0C
z_4l+mF9(^`OSNnDmFW_V4i0hN@>D6%sN{aVxMM<Nq{~_B!!O|~=pjIaXy7+K$VhP(
zQFg!Ga-Jcaw2xO$mx6(whm#i$vFRT_wT*tjBEd%PI-in6j&;0yq7_I_PbU)(sHmtA
z@wwW|mbrlIZ1?}OYSO{fcOGdsJMx)x6O;6$8a4L*{{HUH3sO*2)L<S`s&<OW5aTSa
z_4faT$`twjvysDU>%#Q(6+i~Xo`@<<Tt*ISPIi+SnAab9S#;r~e75)&H4RF?e0R70
z*<FL0O5;Y(tXn}aGx1RK?W!&U6(UjgAH$DEdRtaJ$7?&lp)O&4T?awKAEK+R)fWD8
z(GJ*rg3TJjLutZK(ioswNGWoUg^TSNGhQ4EkD+oQA5I_Cc4H#!Iv3&;Q;^k&SKW+~
zS@<|2RL2D-s7if6UL7&ylSY!<{ch+9C&_MP4@1Z&_4grpRA}WV<~GBFx?j1kw=7YO
z7Av%q-V^9fFB%U0O3%nRJUGBhNvRteMzpnq?VOrKNzc_VG0FN?D611?x;oMIB5LQR
zwmII6XLMXbIy`=5p5(UHzEV06Qm8DAu5y~ic{Wo}xK+v{%h;UX*f&)d1y_2Sb?xp#
z;>kCDh=uHkk3-Bj5cu|OQla{R74Cp#ou*rUZeC__+F<V2D}!0QII)!BsnXD|yzJJy
zNElzI;*$sM-(GybWjOjFi-d$k7B@FO{-v_=@ob}elNc^y&o%;$oFx6>#>M74PuT0q
zyrb9lU#OWkiSgB_<8?7=NF7S>P1Lb94I2WV{(@!pi2XTJuE!pJ<S{veK`18BUg0`3
z_2&K&(SE*$QbgpzLF(bp@A@s9>86Hjw-eC>ex5Qm7B7CMB2VRIR(3r?CabXwp6;-Y
z7;^Eyz9w{1&I<om1vZ}I*GnoZO->Fjl*BV88*h}+SE%J<A5Pfz_xCx8XOD>magaU}
zJyP7=tN?Zb`Z_CX<=x%g^t3iHafV}nWe1ahkdV;D<>kkZA8DuX%tAK3939z0D3X(t
zi}?_n35a`Ab>1Pr$1L_{i}Lp|-x@EB!&YsM@zc@q<r3AB*KJf*)G2EI6|yux*mN4l
zHNF|$vl)Z#xPza;Fkn#1q^MO=OIz4|{VqF2uY{ApwrD40=%N&_C~`ab{=)Ux4<<I0
zB{^^C6Ktmje4&<tcG8u6|4nt!X6M$WA_J95u6F%=rEzv)lxBci`S&_0Ta~uZMr$d9
zZ7kA}q;w%)m5)79FZEtfqxDzs;3T~^m!;-+`G$75YTUlQUH?WnlLQ@SOJ6E?EkeXr
zi8_{=zEAK^pBSlCNjWQZ)A?`a7qkd({7Jc@Zsyd@e$#L|e=)HkzBDE_>S*gVFuS#N
zl;n=^wa={2D!LD=ST;l|wi-@>=>+V-4BlZuWF7|kk+8N<fleaeOab6&tJnE+3kNzg
zth*!HESYxh1$0OV{)hK1`O1|mLeQISo$3+a+vBC%&S1E{W7ciu9ewG?Ne}Bi<YG{C
zg%j3zS<NJ*K7l3NXC#b|j|Wu6n2mpYW_w<L*T9q9^!(lTY$LDYns3O`UwfqpnR?|^
zbiybbhDQ0CO09gg`qQ}Ih!HE->Wermex;3?<idHWKh!c45eI0~_hW}L(1gcEXO$IS
z%_Q!klVJ3g-2&M2wV^?LiL{#|Ffed`Ss7Z_Um_9lQm=cQueGpsO4wP_@2IR4r;>;z
z6HU*`YI`%5-B=Mrgxwnwb9IsV^;I__{BK8Dxo2IJpcjdgzF*%9&ql?l)z3^y>e;=_
z%F6O3n7Q*8^uPbQ-7>csg9dd9)3F_GSinoR<EAle00{AL81ME+AEtAte{2%#0v@s9
z%Xb8yTb)1p0^N&#psHDW7yiE79sA6ge>vCq^(zk*l@tRGD3Aa<dbCvn$``sbN?2B$
zWo&X-=jv{6&+@%MeRPzQg+&>=HPY?-;oRNfjBZ<HIluO&jHzL+`dpiWU{gbfwvXvi
zXvyj7ri^O&v7FO^S|*1jm66a_XK&~WjcmWmU4Z!AQ`je&U#Rcx$@EMNn-dl!rJmzR
zGA#;U-OuCT;4(+v;SEJ`hLI-|YVwjEy~gnLW>=zRB!XCc?ylAcUbUDN!()=iK%h;v
zvTr{!rp9-)k13ax*lwL2l<+V!m;Y<IqQ1I278;@%-BXj_SI5dChcDv-Rb>?))^yGB
zc$xr738*gNX()<Sr7J}z{|@-vj~`N?rw)eJS5-LzwkV`>twUN-xy!b<zt3v;^=*Ao
zLUPnaObccMqA)r!U&Zwzts3`c!!vf(SmHF(wQp3c7Bgi!3Fh$y1vJ#u)VR3GB_$<&
zeSM~FK9xpIu~cd7SdlwF)40j-KVfaf612VBPuN0j4S2ktDH+#R%*|rob2}alKw3KQ
z3-4dG>WPeh^{rD|<J{hQG~O4JW(OA+mzSUaA6U$(RW)<hf%ZIaE|$e+Q5}3%SXsQC
zEZf}=bs=-Ew8Or*wqF|;Ma>@9Q@fv=t<BBX_a^dbMMb>MD43a}N36@s%bk|Lf%^YA
z9Nqi6w#uwybo0DOIW4cK2oVB!6F~ReJ|Q@s85d}d*t4pz?R3XA@Zs*cBvv=uMh?()
z0C!4DO9KrzgU{hUMfYzkR){n#r9glXa@Pd9C=LlhIIDh*IaYbHB%RqCR19)4`?qiP
zb#?LZ@r$dfPIh*}U+LS}lvY)-g?8OV(={|Tb@<+%^nD~IB_$;zBhxBYO}6g{cobs2
z89g#hfjDk#c?PGION;6vhgaU+8|(GE>VXaF*B`UIvGwgx$5sdFuyY_6@M9Q>-M#&~
z!pkPQwpwisA}4#SbIObjAr}B|q0Vhr|8NX35!=raF$itS*?XeC%O5s08Bjz_rLB7N
zeQ&Mz?YJ2%+w~vvnjJqkB8JHT)+tfQ$zZu<w?Oe1g#bf`R5|h`hlcimljAG(5^d~@
z*4A`osh*c*nhvavMRL$tMQg%LF}aDHZV<(oX=|pG{4viAOeZxH9oO|SfN|YCxZMeB
zpcFREP3pgRYOS725_H?S^KFh0#C)dU#VB3q^_fU%=e18(^Sf2|pOA`swt~IihR;k$
znqJ){c~TPi(?D&*o$t0IkKITb@cW`JlVH+X43}nS>8Pm_GcsE2{w{Nd`>->Wi_SS+
zot?e4vr9}&WM*S)cixb)vpY;;)FL4vv9ttRWf15`w7;q=D~|`_sX;HQ0t<6;V$bJ$
zQbTEkBy7pY6V|FoyvaK_Hl}7~)<I9rh=rUj;vKSi_nFf|Xxr7r#bx~~qs7N7t+K<u
z-_@06WkW+lL|m31;^M{*=c)nT$jr(T_vP#F>CrY3Xj^rWfXm>fFSb0IuLY{g{!&wQ
zb+yx*CP_)jR{Y3xE-rQk&E2t_r$bM8NLxqexA(VLUY<MkHI4z#(x#?cv(w86DP)Xi
z@7M44Dd6S#uhgBFI|9WY=Qr9$Rh7v4@{y@}I5gNZ!uhLbg>C@b@l{T?wz8TpE4Sw3
z`5i`1%JWt|ARxfm+1Y3(!VqO-e%|YBy$b+0D5$ly6&(G+fq|bn;!pKQu-JL|`T4hq
zZ}|D>N7E}PN6vgNmfiF|<mYepM&kor$<@tGv|+|h-@Y+6;8Sc_TZ){p8HaH7yRmFh
zSy@^CKQ~}17F#^oOAFK;1~@!DZ-H7ltD|>{xhk4DhQ0YMtEjmliVK2oQ#P2b9*Zs`
zQscIOI$@#&MJyZ~WIFr3j3)z&Gw<DyIXDqy5A}_M^|kw74^Qo@7+0y16Fd0Bai(RO
z=wC7M&SXl#lEe+MCr!;F5Uq?)TV_TSV@c*Yt_7qf4)0!CSz7w}`Nfh5{$ORhs;_TL
z&^pUlGOH=65Tl~{(cj;XzR5gpLqEx_Ca#k7&=V0J-aM1CU}pmb8BMPYGdk@GEUu*;
z(e>0!k=Mng=#mP(8<?2j1mK33hQav@Y4~O~zlHT}WBTh?C{Uu4_*;#snvGf%Bg2n!
zanh@FY8-qq3S)IyoOT;O!m`*39fzix8XC5Dc0kQ2D=YhZ9e?x^XgQXaG$9oAZ^r(X
z`L4%JPE5onCxgc>$jzNdyb*{c2W_91H@V9kfVf<{l*%fwWnBbF?Ck8gez(xRUUwD%
z)w;R48NjQx@-cG2*=VprHj6_6ZySHnRqK6eeR7CT1E03k&(!$sUH^D6#Ib$z$SO_p
zr=+yCs?1;0uzhG%H(Tz-D}B5SyMjU;69XGM`l*2pTLo?`_>IV(cdnkn2tclvqK#Kz
zq0>s2+K7^w#HmwfaojnV3HrkD@&^a64Pxx3EUSYhVrch4jkYcorGjXL(b((Z$r3K@
zm%?hDUq10#6&S$60IfKmq3}&`;fhxG>-?a1$ncAGlO(}=4kpr~5u)9b4*`Snq=E~n
z$-EegMEsXgArY7X`kEfRO%(aLHl|MH(F@+g_e(r=g0hA_pjcXQrGa_;XHE!Okpy0B
z4GyBiG_@LAl}?Y0nF#~PB<6RQiuge$Z*p=HbfI#+YOzZKI|}yOnu4Q-cv%UFAPI_Z
zo~L9PBOY#UhY$YM6&1$Y`+N(_JoUATQx3m|l9@w8LMozM`v?gi?!~m;?2ss)*-J}z
zBdRr(XKy}-PCtLcBO4p(DGBejJza4E^ciS?wzjtKghW9H;pTQ5{~|d%J8M#TAm{w>
z78@Iz&7^sHeqJG0{hhn}T%}QQdHJ8)j0tpxD9zI8C;VIuapeC$>gMi>2jXsK=5mYY
z>9ZGf6v(gyeWj(%HLl*BoiNp9whRnq?CiyCkEF&uf#Oson~^=dBxG>1U#$uc7nWHl
z`|zSsQKDOasyO1}RJ!b`Nlpld(VprwXmWESt>+y}BeBK!r?1Svo3zrlciHC<&C2yz
zBsW?jtzIj=&Z5S#h?ttxvOA28b6-urvYxAv944bR&JmD|dBuGyM$$SCWHdcV$&LE2
zh9*GGT-exl@%Zr;Qedd|n=5}xpjB!X3<K4W)u{2qMO`u}(&<8dak<~`n46m$pc3yL
z9xA=GKo0wXiqE#SwIv`Z=<V(8&KDgS`Mu6f81kxmc6N5=b$yVWm-qVV@1rkQ_EXY2
z0~;IJw*WXVFR2GD{We4*qVSx1UM8jkC%$}Nu@@u=2nZS)8Yn0z{|-|-tsy+DB&hOM
zR)x{g!>MT|e<hIth8*B|dB&bsQ*#ESHslZ%{hD>7lZ6f#wp3;WuZPMERu{|^!u>g#
z9zHy4!lA*upB@Ia0_N)wL{O1`y<mDw<wBZR%1*yTiDNy@AH@+9m4ISolsbR&OihX0
zp~RC>IM9*&@T_TH+TH!7;Zh&bK7Eqj(A3*Uo@ugF-JdmF;`b-r81(XrbU~$_uUAxJ
z$_z*+amwm&9_36oyey=VSg-R^8Y!C>>8H5$7OpK&5h$zI5Ct&C;}aPjZBPTUd}iI1
z^)HM9{{+zBIBcKKwt6=$|G0W(vrr^?BWzam1v!estF-7O${hv>XgV{x{y>A1boYkG
zLbvHl$hc)IEXeY#F>-MF(_SU{3u7WBSyfLnw)x5YaJF(Z)k&p+%jm6zKbNT6B^J_Y
ziwj*8$_!}*fRSLQaCFU|`J2GWGbH3{GNqA%_R5-+j0JH;0rsX)TCOC8&?2d*SgNbS
zou8{=W_n^QH#$ihu+lU)Gb0-K*H4YDq_y?oo6DxqQ=vM8nTil-v|z0MJ}uG6_nGv@
z92^-L`T3KC!#pT5a?tbW1}IJ(_5wI~ou6GKih@#ZxUtg)Er$jNuWKT}kpkj{mDTn}
zcZ6p>8(en{;F>^ZoyeD0;wpU-WMXEHsJMveRr*r6-TSL|uG;J=U-awOuRZ<!TBX`-
zsdo1E%Q`;*Gr1`Xes<+4K-Bou)yxYvCT#FJ@Q9#sUc<2UoDg86KH)7cE%A?Kl8N~U
zKoyySkJ^5}a(w@ExwL-$stq>(ZRi+h%+mbbuT>M9`0<I!5K8P11{LW?%DC_aj5umg
zO)VniU_gk4n%jT?aRkZnwIuiNB8Q!ZecL@+I8}OC-7B`0g41x05cPN*o~p;%c&s9;
z$i0b8oN`C<QYmL?X|bhg`~H0@yZ-XlGBy&Ej)`GeS$=*_xS_q`!pOtX%c8$;rQ0eM
ztN+-QIe6XJ+H{Cvz8r$5s`X??g}jUn+v_WC<_`Jc=v3a#&V5hO+q1Uavi~x7dbBOc
zw%K2$-yp_gIb|M!d5wPBbJm(SBxPW;n59qbK2uL9Xt;xCr^G}|l#Y5Ts+nm0UQigO
zGZudkxF-qASpcNEIA#0Z*0I>y<lh{?K{QH-2N-x+%I#@pGPLo!qTWSoEpcICvg+#N
zv|3st1;x_u8jQRjIQ1Ok&eFeH4r2TJ`(sBY9d5*@rG0;uZvs57uZ2o71Z3Lki<6VT
zDyo@PXwAl$MFgGA)v8Rq{u@d<mseR`Z9dmbY;1BGU+pk1o{E>7-R%ZPmN>kS;o(h9
znzz&x(!llrB7~NfmQa={aQ#+S*=T8dcQ1iwvhIi$LxfswDnU;C)+sb)HoN>|U!Pn#
zA@{3R=2oQ~@7+<bqr~mTmzRx=jO?tf8(;TEEtL^{`2v|HM&&s&Ffd#dJOO67vttb`
zmC=YtKpVSOUMP=16dN<q=}7k++5eXXNGD>g9BUAUc);M}1O9@d?(92FEiDV~exxJP
z^+#c6m#_l8!yNI`pGZCpUrF%T;h@jL;(UC%1_ldbW9(MzSdg=T3l-3*FI#%HU8AF;
zX-%8RVsX37{`vb)cy)DkpuA)-F+!dC)Q>-ZD$~;!vuEi1M=+M==lyil&VE0Dqf5jy
z%~K=TxGE+mCl?tRnNn0#PZv|{vVdGq`l>PU^Ra?2@nV;Z(C0b=KE=8>2h4+J{s+b6
z0#)ko^N>Hv{6d!HHxi~}y+Oa*PPMR*FHutJPt_6Anq96B#0-<#;1I++x`D>dp>;gm
zM1IHC{t<#wu|_x~jEWQU%DSekcYm_O>NLM69JH19muMdd>EZjbMu<AC+71s9Fb^;r
zUgu$ukzCyK%9}pZnTqGEn_pO5)HKwX7`5NHNGjNK)dssgzrDTP>~2CwK5kZNP(aMD
zsI-*O@90}}G%5rfg-riDb*IlGD~~6jOFF*vl{~fhQ_w7A9p%fell*DUCf#mUy>$9P
ztDkmPw)m6q8)AG=^y`c04y<Ehj)Ss<Jz+aXvxNEK5o*0!6D#ct9yW<{_M_gVpSFvF
zm3DM=M8!CkfBx3o?GG9;84=N<;F0cs12VU__v1azvb1n>y9B5hNZfyyf<=wJOLB9c
zlaQcw=t#DV;={&{r6dYEuS=+TJWU|Ohp?|DxRBtZgsmC+Q7&xHRvOj3LjA&kM)L0W
zUamNkz-tQ&|JWSJ&23=Fc%O$xgdX44wzCuK!nLc2orm9!bg{UkWI+G48hiv+K*_Es
zAQqj|`y6kPA>(I@jph0I<+oHqLO$oeqruaeeM&8>K$&_23#l{2k&9=8-4rQvuiGp*
zXpkf9$+q4lJw8FIt3yUe$Y{V=Sy_o*>$1*+!+~a72B=U6$;@yndv!A|y3f5d^RQ}`
z5bf^^5Fmk<JXvi+t>2Y0L}9swgOer@v^~A&eNvQQ!-kDRdMAW~|HjHnu<8QoP;C7G
zL%6YMk5I@S)9}^nukMGllarc%dfUfaFmdoqO&Tu#G}j^ja~QyGMnglRvOC})K?<5q
zZ?5gA%y4pYGHb9+7GTdPek1U@=fiD0e&)LdHJnsOA>Vej9I*sGKQVAx@VYrD<C-L*
zw)kubpk@EtzPgG|6OHU;?MUO8y}x_FIq`$rY?AGTWBgDX^01xR+S#KwX?xH%fNV5J
zK|>>BZ>haF8w*DP4QXlul_J5j8WaSJUNv^b`-?iT?-nH`A$48dMh*&z<~lTP9>NZf
z<DGy<8p`|bK!X$aJBla?8ca7&Pw~@}QFD%mhritZ9Dv#llWmp>>rb8ZFpb+tS=667
zFhk}53x*=1-%q;w`u-W+KNv;2v`1=LEyI8~h8IPyj+fTg*Z21J+6J8lk?##(I-Wb3
zzC{Vc#=(J?l+0?|KRC!1az}y6%FE+Mu2W=6l==AmZ%<4BHc}H+{N{UGm6!J`(g`2+
zSjND>8#Minr6VOa<8L)LS2i`rm;ubt`s6t^s;jFQ>f}R4X=&;C?QPsHYIvfe@68b>
zA76@n2Uwqk#Nr@Tqve*lT<NFlf;YAL!zrxMRvu7>=qFV;GJJx`&exuxwQ~bp2Vxc2
z;(kHIO~J~+alISQ>0hn*e7M6EX_GJ|^Gk)57&*^RO})Q4m^sAE(gsZ+em;WZ_mq^~
zdX2d68GEb%g6}O{R8e~%s;6V%;^Hb<>UqFIuM!2`X_|;lR|tZMC$PsCYhHD`aq>l}
z`YtXa{=tChyxf`tnq`7b#dB&o7T>=Q`s;>R$blxU<8yPjlEN?MkLFiegy`jUO`_h{
zga%P4I!#egQ|nb3d#`r}e=2?!wCR|BFg}K{r`c%<3z-v`>+eojgnGoRksw<}wKUtJ
zsi0V^MfAITN`Y<j?M~W<oN2s@o%|moU(k03lI|S@D|q>gu}L{UIADc@uK)Q2)Jmr{
zF|nM}U%$BCjAn{YgvQ2>mgv|m-yp(uCUvUB0q6DpiZM%7PhI^iu~10G_0RmG*xh>9
z5CXTI{?RV34nV5eo5yV)e?>~FKSm4e(@5*(O%;@C*X`3+mqw(y%>7-EdWHVwT0{~z
zLt0CVoZH=&HKHl7_rlC16J}k)*5_~c-FYA4UtFEX=@~3|F}AFyvbr5KIWf<NL?`nf
zLpt1QWc*s1%0-Gf*@9YiE<GQf5C6Dfn%e(p>=>4=`=(hM4lb0K-_h^)4Ic=xfJ3MF
zJ)$}`RDmX;P}8@(Oyu(cCve~{CJV(mKU<cY{vKh)8+yrq5oKRIDx+})bi8WQwjUh!
zf7iQ24YEMx)0m%}{kaVgFYxI6QmSAfUfu@p%bmw(zBD94@AgJA_yISfpja=5r2^J3
z4GoP3Lr5X>hV*@LrKYBz*pnLhS9%(HdU7#8f_@i#e0*~99CqUOVPU;vX_!9uW9b=E
zQc@s|fDZYI1rF|W*F0fSDD)%`03pW2-xc`%wcYYkPQ(JPdn&Sm_6x{#Me5pHbCpPW
zZXX^V9suI=I9U}Hz=J`O3YI~HzGw!&L<q(zAfUjJc$0ZVPM!x+64WZJ&LXfL_F*A@
zZ*3NiPvr6^{+2(F3h9Z+EjXfBFNSx36BH~?De@&+ta&2}!8@+~+S#ES?5Cxqq1pO9
z#QUY->dN9-rqJ&U@0iS{K@j(`d;7N2#aLQ;bMZh2xF$e>To271H9v)nPul${6TC1;
z`h%7KCS9q#EIV_^&B!pSD-(Kuf)wQ2(uf&wzn2_>ZNCJc-!CQm`@5BraGc|tH**BH
zSP($D7L7>3k&Mf6!NhZ&zxfp@X|EZXjJe(wJT*TDvNf=^nQ>5%(22gfZV!p`WW$U8
ztF?hafH{l$Tq$Tewv=NM_Mbb5t<H(`0UhLTDmzU^nxK$SKqGtXCP2-Bzb^_K8m>bS
zP}_+*dA%<z6%ncn3JNl(=uZTNdELG}j+s0&TF)}nYTC4_Ls=IHF&<|A;7nL3QGDJ*
zdR5;K(`ocaT-;j<%>q^CBZ(?eq80P-hau)=uAb*GpNiDwvOBWZIXymvU!32bj@li7
z-6TJ}GIh9U#TDhTx5v&qp7qvRWOUwqE7|CDcckFUp0j_JH2<k$^T^<!1~^jD0&r0I
z&k^CAg@c2Gh?CV8_rsKskRSVNK&HQ#nUcXO$dLd0)n;nv6#a&SyIo74%dLCI$u!3C
z8OFci$TUVQU|&F^M-Y`~mX&vpM}uwC->b0Z{|6p6c+Rz>s&}XZCAt>{6O)~h(Z$0f
zo7YZv_a-JL=EDbMuC!$!C#>i+)X*Y-KCb<1M?~JMNK~_MlrJBLhl7I#Ht<NH!r3DN
z_KVI*lq#7YBgE<^g=rT%T;{-X0fC{FCD2wpuJ$zWqPI6Uxoh5NP(u`0$jgn6q3h~j
zCVpX_P=olVS$o0MJ}AP6XlM40j)<%d5M+hRm%+L)KRb(Pdyqr+(dQnto9_S;C%=5$
zPBr^EG6Gb`&LVLcQ~2WCbphlH*Z#E*w}SxXxoQ~~m-AYSelL+#u%d`)1X%d^O)YhZ
zjqs94+_P`kSd5<0S$_c*PWX>i^SwFpY>}Js+15YnWH>lMbzCA!@QjF%SXJ(gC;RxD
z>AAVkDl+xb8fN&L2r>PR(b3V9U%i06N3!5Ws}?UUEjf|$^X*9x=--COyCB7egGf=G
zeg!y#_=F2NJ`)oT_V!0i0^f|;6pF^Ku40G}m14;eq@}&?C=$AY%u>3;+-kgyV)aQn
ztSo{n9ZXC}uNo1cPYEy0-DpKAkn0>vwB{$e&0nL`-d0cTad@L>Yx@ZWwQuw(W`pSk
z5&+QXfXZejS2H^^0}>y`{EsHv+tKE4EzOKR($f}2_I4h8mZ1b)6=1*Hp}r9gAGB8A
zLvWD4Cv9wO$eXr@khTQ%My)jpQc{AP>Uz1DY=}5Ih!(ah0_pADJDxD9^NWjSx7|@-
z9U<k-w@cm(ekX^8gX`|@7B&Jd(%$eOqv8Zi=Sa_<wDr!L5Nax_e|i$xMa<))Jy~(<
zynz;VB*EVM{>_zk|BUVpFCU+>k`ka+d`~B!f`HAfB^&n0csgB3IFY#0OCTrt35tM8
zXxpRV3g5*28LLaSrfY%=-4wUin|^Ps$_3Qh=6%*mI<MX8E)LqnjK}PpN_pieeK(Nm
zA;?o}Gb=3(s<EA6s;aKmXNirAb801f2Bi$WB$|woFWeSus}yT&NUZdM>D-?xMy-~^
zn4LfyVx0Hjyx{)+B|m@4U3H87*TDhMw$ih++kN~yPErp*RCKw`S5HO7VJL|)r!52g
z@~@+N^78WX{Jg%2i3vM<^_bKpjlPi)fSXkj*1&E#FFS?^4vF0MVL!?xBKHHtr9sTs
zb^nn9uj>G}#)TOQ$`lzGxbPAg-1?EiYQ$!69d5;TJDP*Q%ga$8#9d6ui%_H*D*Z=N
zuj8g?KU#UGf2KS?FK@uo6)d}^W<qlE49K0eRF#3mphvWSmpHU#>2&-|QP}Uy6Cvt7
zkV(h+p+V({@4w5+=O~@2lal3qR8S1Q|5#eeP%T#b%(1ai>r8NpGM0l*Dx_VP-`srv
zmhwq)l3QIfoLH~Bpuqf*4(p7)`pUP?6Em(q^hJMe1Mbc!5>)dws?E8od~yBU3oex{
z!abN`eeKGm&f@^5Rv#^4p?1;GoU3aUhKQKr7ra2+`=6L%yvcT>*|XEr#-B5n1zZJ^
zU*Za()5R-jGr7;FsB|1QU?al=HS7BN8mJ_6babIenBmm&l19Y(M;k}6z}3u&YbN|;
z2EW++B@b4W`j_U~&rKKADpMU3ttsC=o~k@5{2)f!kBXcC<})xN!gOFJXNGUkkD94#
zXt=n#4h;<3P4frI*N&HC7PPvoG=lX4@jM;ty-yD8W^fq@`Tc$u=>q8f#=4H86R2#a
zu&K|~)j9mfJLTd(hd@vTDl)<KUw?P^hM{$nlQ~BTC*p4bxG^6kX*`&Qnrds~{g!mJ
zwcCA)nDwfZGq>Tg6JHG%k!GKuI<ED(h)_}n4_fN#>B-B<sitw!ULNg(q`>951uF{+
z^e+>>=z=`33rQ0-I)qmWn}TT8`x8aMn$tiE>)&f|@e!NF`4Qd%^Y)Fv6nUxR-L6lf
ztGg*9dH4F<ryr)-wGLYMzFaL#;>lGfAS5)<*MDnc^Do?$5qfVkpuql7=}eHo(avsZ
zVdi+LiJ3in#KBgKHvq?s#x(UTFKswq|MuqMf=I-xCNGaF{8c>&7Ai{P>|IIQoXWFa
zwSglKDSvY+m?|$b<fcV%Sa^E*V7$=v`jf(Kl+x}^?aQ)~#j~Rs`F7`uBBrP>F~J}(
zEY-|sU8}#;=uoculBG{W(rxNv<coW|5`+%bsq}4bt05f44B)?jOe`}$|D1|eWqfyS
z26uI1<Bf%JMbgJ(z!Tlyy|Z4x{E+qw1EQ@wM&cL(&hFp)u)Amwkz9DJ!?7L<3-_6e
zx$P)N>v?@1jnD2(;_2#2l{g48%%d56-4zS!hC=m?2gBRX%qIOC^mX(+=U*T~<BMF>
z2W2vGV*dB<<Xnlod|;rXHPqK=54zG;+k3=%H^)n<+1ZY-UTy3(-A~1mDqqW99^9TF
z*$~bX)R>nf1_yWHQ6f^y!9pacaYh2-Y-}g$j$&Rul7_ZHWs}YBheGHKpm{EPLMfH-
zkV1@jS<%1fz#6$cRI6XH!38s00f&H~zl-TeHeN0MKD;y2dwhB8f^Qpi6X+9AzY`Jw
zC#H^^N$%Wmt}^~Uxw=|aSlCckfxRpOK^Vso&4y<;!AblFYGgBNX&7YLXZOiYD%$|*
zK@`-~AOb9>p)ql6i1YI*d?c8|0`Ef)-yHj!xBHzX6VCKRJiL;Mgmfz5sJr%Vb)a~I
zG=X@1CS#uv2PaauAi*pho*QRtZfKgxMSM|3MM2T%4-PiEoy|_1=rDl->o@#E10hK`
zJbX-cCHXYi#8_zEGM06FNVt{1?LJ}2AWAM8uzZ<-`g0p~RdnUD<t)@0(xXtc;gjie
zad46)HH1ROu|Cnq7pTMUo~M#yL>Sb=K(4N{O}xKZO=ToM{UghT!{#@}CDm@j_s%y*
z3swF1SI6kasM)5UQV^NJwUcaGG7XK-VC_5w@ToA7AS^Nr*k`c85}FOUZEYsMD1cU%
zr6B!ZYLH6Q?*6Q6aHfI1+}2t@ZLVJMTd5#qztmW1W)<mYXCyaN{)~%DnmQlN=emfY
zj6^)Z54cu95PkT4nV@aBR5`+KwcS=$_TA|2JT<^S%4q2ncc(!n+o;ulI5s-9u%O_l
z5%HT?5)=qwWQ1u@v05G-Z&Tozq}*QMkdg6p*Jx>FVy<S@4j!6G?LD9YGYGO%l|Q9d
zTm6OHu(Om+asKqUU*5!1%LjPaf@I&n_)GIUo3AWD@w7efzO$z+0ty8Vu~PBW%JOnl
zRFvtX4-fDgGy<9DN?Q8Kdi;kDTP={Fe}Ul~G12;V$@i};ajc#p@bt|MoBBP=8LKWU
zo1B?>eEtjq0XeiOYbkPK=v&T1Gf3Bhx~HXeS^m2koE!j+$lmV5rit87mjZRq?7J(%
z4lBGCcg=~vl|`o{+Tp9<KTaEEWquuvcs#t6tSWV{5g=t1m5nV%=UaGXw2_ew?jmt9
z9|aSW729Qj4B6;c^T^)_7R+cstQ|o}=u?W2MPJO#?m0+8aeLa6gKVyx93lxy#KtHE
zmW1iaw@&oVUFkiTPw_BH<+m+Op1cx9M#N$%P!6UxQ>(Aten(kEN}@u+*AFiuFv+dC
z(x_AK?pNXvf+0FJxQuq_jDxrE`*(2Yy#d#wmW#QwH$x1ij12oZIvU=>$<c+*!P)AI
zO3W9U%zJfUVs`J+67~AxZyO&U-!Q)uVo*&@&G@f4^SAJEn-VWkxVX)0Ud#(Z5^^r_
z9l@&~m-x#c9wZHLC)c_Pqv0YjxNpr~gi#a^l^Y`b@W}lnPvvgz^`}X5l(`fZLw)V;
zw@2IBP{rOpuczSe6D0X;nFA)dEp=@wRSvXr`?rvUgf1QyME8ONElG#wg|emtFpbL+
zBnnefQZNRMSR<s1HG^b+ynRRXQSRTBq|jF(POVo_F0G{fxo!u#dQC82U?7M~Xnuu|
zRf8BZcWA>mcT%N9mmGi<J6eR}AqpB1HEAofN@jYO&VxaaSXy22cB$IT$Cnu1F{DP6
z0t?DP(k5zEsy{QDc)Y`UphU*(SL;tDiv<*i`fq>$h9!=M6$h8|^uoe-Fr#e+=BFPh
zp><c$MO_nzjKocLmuDl!Y_^UKAaU~IMdAX62S94ej>~Q0GjI^d<4S3*%6^MU$W^yq
z{;2b1LM@h@Fgu0p?@hOld`scupPwC1Tkh^E+ZSOc8l2!S6_7PoN0)oMc1AKhU4JK_
zR$0JEe0Y^b_|3@M4?AQ)qco_q$LKc1Ne<|1K;KbH_$~+_WpA2RU)04Pc?^UgBZS`b
zp>z<C!9lTFsIvkD$7IZ<F*K{-BwT`p@B2IMzTO36gNY}bgx%5g#zwxIWYkNC3K)t#
ztBY@<e`tR*(6WGhPfpRVmgalZwC{YjW!2?&nElI7byzp4es^cDuI?o`ctYaH&EAtI
zcsX`ZgS#RINm)qQra`Z2To<rpf2SURw3COOw0ZGZM`A?<%kI>hR}<J!Q1mXzrE7wq
z@!3_#b81PN_t$p9)J)B}zL@gB!-*voKAGibJ8)<Nw=G5m(}4yCS|KGe7vMF3t?=>l
z|GB>xV`mNR?w62+;jkZQ{0O$mM8h?zufg$r?IyR2$N8^bfH>hLK`PzpV-}=F7J8{r
zQ}Z5U=iBB+>R;i!o9*G${#ep(7r!6xp8?4wT)g-Ca84~-ByF1qtw<S(B*MMC+~UN8
zoe+E_!t6vOm5+uW(ZMzZ`1JXZEr&J9hu#`T_svatSTv{Jo=Pg0XoaI%H~OF1sL=TI
z+wtY;@C^qPlrO)Njd4WZhfqv1Fhx^krleS-AgE*GY{6wAva-IoUnDr(8_HuKJR_Bl
z{qgPmjD-wwURn<E+c${fZ2fQbX!nu5JrqQT7k0AX#we`Ukyl}?W!K!k3MQz%A^SJT
z(lcim=H|l{6jIU?w-$9?Gjn(|?GZ~TMN99U;BaEx--p_kn<;UcUwS}6cg6D4$sO+q
z?(Xkbvs<vFJS5dQJ`Hwm3d(1rqNBgq5J2sQf*AWaI$tCt#?A3ctObB@^LhCS-3P)+
zDH00{0sL++g1#PS47#1yUfgkWKR1vEaPI1*a#+@vvcv=jyMQLr6A1gGoATe?s7~O#
zohrNDw&J=uMfGnj-SX?FkvHQuSk-hVW6XZt)2E~(+d8IZW)>C{Go4+wePg~Jh$5O{
zz~k~Q>rvaI=7?to<aNiaw|HE?MLYZ{n3;By{S~gG6k^)S+2(BV^E0RE=mmbAYp+1;
zrhpcM$PHWQ;EOK=w>{s~vY(+u@i_}{Y1?vV2s7UQr68?b=|(J^;~n?8otc=}H!QIM
z+XDo9E_X(l-#Y5N8dnP((6vgEQ6m_^k1S^Uvwd`lZ(`R#97E{I-2p@q-!HB9X6`fn
zh1=W5KR3(u(p8+<S(wtyUk^M-G1Hs7Ps{KmCE?4cq%JQ9S=^FB!At4C*K>n8@)v=j
z(h(S4u`3kRa3doT7^Ffjzhvgr0P*8nX-0=$aTkqc2{#?egLVVlR^RDaF;xvp7M3F;
ztFM4ES<}!Mue8DlWKI|osTKK##&e$i?(Yuiz#*~ZA{5f&#KEP*Bt!%%&;Xl43=$r*
z?Php(rpE0a(1lZ2R_~nXR5Jx!e)Wt70&V<p*^a5FHzWuXqQF*ERFsvG@w1z`02z3i
z=xixz>EN-5wAC6<esvqwZ(jbXGp*R+XoOsr1D`s)sDNCR!DTfBg5T^8DDZ5zTM4}@
zCFn4*Z<3NDnmjvq93b_}otY0Xa580O&84Nuy<rlmn&`sjpN)DB>fT)@E<zCQ3|De>
zl3PASkVnlO2}X^<O>=X28Pf$rnFLi>b__HY8APG}spCtVRb*2W%d=!zjRU6_RIgq&
zNxP#_Y%qHH1h>Ng>&l<fsLC|i&vxtWZoMGZ^mJ!4e^LOv1O(aIp=oS2EpsQq348dZ
zooxZ{#k3_9hIXW(ak>6YTo3p8KJhje%*|1Fh@Z1NQjRwk5v*p+E9B8$7vu$0%b~-o
zTP**}rmkOsyxJe6azSH7MxlBR4#)0r?LyYRv(Wy!OK$5!;<{Jb*B;;dQ##Ksn8n^-
z7@U1SIEoZ-JdW-zG4Q?=Fy_nIW@M~=jAL)8K~#o!ckeVGx3ug*vx!W{)37Q2fmt5d
zq3H&X{Q-s+GW2Hr-jU1`HrcaL#!>Jy8CRZo?>hnUHv(c3|BPILVflp9-}MN$P>szL
zU5>+&t&sjDnaD|D9=4GtlPujX4==rrZVs9F{d9yqoVOtV3uzqlmZ{EPvp1r?H^AHi
zmzf&F8@+FH^vf6=2!ApBSz<?3qZv(1;m#0YEw8&@{(T7`)f+MqWi_>&r}g0W_JH}Q
z>N~!MuU~<$<9;>}ADgcGXP;Z8epN#~NlF&v_JMvSKu<4kU?9lBp{}H42a=$maGEA2
zI9glF{kpUh(nr!<>sR(DPEg3m?jX7zzHqm(0oQv69wX2Y^X-xufyptB|BS4tZ<ouh
z$Y<<W&LLCCor#k(zRMgQ5i$MqXXG#`bB-?!2vh<7=NpgUmIx3W*&~mb85tLnuCQ;&
z5rf=?o8m|W>+BjTfi)t(>Lw}AUzKOfDIg#sA|fsxc)HpK8*F~HHvw!qpuWc&$AZWR
zhgi<b!u#L9gIi&tPNy9mJw5QUmbSJwvyK3e=ma&rp72U30uzdy&KdSD{sL(;*`tPh
zsqEF@Z+8-nK6zEE58U1<>(1;I^C|obSNzKJ)4^E4`Oq)3(f#jX!(5&7BAS{!f#+AI
zQ%*!o3{ngH_VZ^m5rojX0LX1-CnQkO(>K{n6*;10w0WKX+8Rt0hy^GAUn0B3tJMR?
zq4!J%uUL`B7B8AvzvBHB<}pM=Yn9$CSFiswg(32L#|jY~k{MzUAKhnv*SEZ|@93hw
zP=BCGgIBh8FcF2S7Unr777UDZ<sxMg;6y+mZ?3j>^7X`Gq0%3rn^Uy1V;Y`qUgaxT
zNJ3*_ktmd%Y*0*0EO)g&co06v8-ldUBq;WTWYF2zmYSTz$^20-Tqz$YzU*mZPga|C
zoXD>-zomaq2BB;}WG!J#Oe8p)^Yb??UG9ihmcgS~O?ahh$zw>G8GTV#{h29dGfAVP
z$4{XJLDMr12BiDGp^}O=?=0F5hpfzQI_wwf4g<<1rAZs@;Vmq@?!S#r^9TqCP*8la
zqd5Wz%0810GCe({pB^1zZ9WMvF<{gw;&H@Oh$kGY%=&jPOSQ{rW(A%$9yOWm<{MD|
zA<wj&wAws@wZO|A-S}_l-P!?8b&>08;6CufJ3qlfY$BqUKEGM=b6q|)TEF(5Nqd$*
z*GF%1;|vqjaIWa@mkRFqNt&7loGgD|YaRr*gKX{WKmZRU1w4-zd6g7ByaO}b(gZ=S
z`XleOIFFCMtZ)CtuOZ-%K+unaF9pT%Q&S3lVhTIzSU6Z->c)mg>pX#fAH+RPx3{+?
zxr?&1IKruMz{N|?nD`fJESUe71y~);4UIBfge@ZiiD+>1&E0!ky{0a_=MG^8x3I8K
zjul<LKL!f=oEN8~N=-|4$6nt1ngEZK^bw>Hx<ZjWEBnt0L0QjoxfNqEg=HP-jw1sM
z2$1X9grx~}Bc$?&t^tCR{v+K4vsN!oNCd=fTwhNXzgN<uI{mJ;*Ie=jB(Bx?>=EE$
zNXTp5fK6<nuxv43hmM2u%8cPFUw2~`1PM1<=0ttMp;ZH%vBr%*CI>oQMHHR~neT!A
zetM5SlJN;(GyGs3p+V3<MMVXXY48j)KMmG*V9>9pmW%>DIhE*F)Jm^DWoBk>*v$oJ
zAX!@fv>tD=Uzl{$<IWiy8yT6IolSX92qIxeeI+hmD&c%RmpnKgA0In>f<fREpT4-u
zp1h=z)a~yyKy0*2f+VUkGBTzPmB)LW;TRaQ4#&cdvfapUB%YU)YfVf#cDWo4pR^+?
zQeZ{Dr@&$2<-=$?E|)4~idT$cD3hUMPE?4pOrUzELls{*u2%H{q$Gg=2y*eNYl^zM
zDQ<gW`uNBs81^grm+OC1?`=SQFCxRRYZ~YENvoJo54iuNqy!J!1)$<%zT*#l{BQnB
z-<#_M1#o`;+;CaeMQMKBivlqK7p7Jp2L+-3d@XB#^YXV4x3%QBBM>r=KJwyb4KDY*
zc2QN0&CAOJN$<G+VK51Eb9%bEh(VxQWqBh2Gkm$@mdCQGu8u3Zmy4VGk8UPb#>nW{
z*uPuQx3_^}Z^Q@)yW3j2n4+TbSPvG_x@*8aXF)+h5TJoNT|~ewVn)Wu;cZO@oM4Gg
za}JO=Zo%6+3<nH)tcBni$!PIsE4I1Kt%+*E{dTyXq$qe`V7_~YO9K3JP)M1Xl~hzf
z6!^1T{=XZrH`mc}e0nD7l5!et@%hZZa=;CP<$;a&d|yEUZOgb&#a>)o^ph7nCFj+C
z{|e?AJAIYG?b(L3tZZ+C-5k-lP13YpQiw5DG|U$rhCFJll!p|0Y~q386&41eYnt2$
z!J3XCK&!`c#MKHl_un;(qx+OH!3E7#`tF(*m$di>+#5WPG{!h;bKdwdU8?1|S5n3a
zA{mJxRkDgH>Nf=<k9rFoPfzp0hd~UDd@<tny=*M#R45;UK73pWj{3mLgg9IL+8pUM
zy+^L8hDP<r`I67bA66p6Lf-QX*Y(_-_yw}04j01H$n4(Eos5FpEa6r+X>O~x-c1fX
zmC@lJA@l9J`NaMO`x`KSE#54nf{x*xzqckpqlE?r25LwS(B>G`nVk2`zdnVv7Sa8=
zIW{6YM;ASM?lpXKTs{Yv<@}Uw+efxLDpC6W68ZmQ>Mf(HYQuKXMTdl9prrbuBHbx1
z-HoJlcej!vDIHSME!`l}-JMD!AuWAwzHjWax5J+fS<Lm!JFa>V9Pa0uw;mWL<(*?d
z8%4v(rl+QI$<%K8pNNaaKX&+!z33$^ZU5q8Dup;(5@><P8r@GLJ3q$;FCqSd1P-`5
z05DgA%U{!SuGe_cZo8#5;a>CF7|q4|*d`aK6wQog3(0h?oR#0nmJW^S*3T-d-rs+h
zsQzzwv*1f61vAPP(cR4dn2?KReNdY`*cxq%yw_zk{q<~|xyut-_vUId<S3z%#DaBt
zwdbLw<qjydA87_-VPl^ktV+TwaoB-{^!NBUIXNYWZXWTzda}Ho%zy0&+}=0vh~PJ4
zWn-f=>H}UrB_%~ws(=5t>%b&G4cT|+I5Vrl3aPIg`*wd$7#LdwQtuA3zBpp`1SAW3
zJ-6eQfti{5!NJjumo=J??yGQU<-UXKU|y;oI^=OZSoum=@-(jX2jZU2@VAHR6>SNl
z?p)3`bW5M2B%v(4f1d_xxvy_)j%?~gu@<q(=v~X1U%y>b7a9-fZesq6UXt5awB39#
zPAV%aJ3F{1!(M1kOqiua^-3!KJ8YQDgW3yjhbLpB>!RemR%5xt6tf+nxW_wN@#+VP
zzn7Ph42+HWysy6wt`pAW0I$#0%)G#IEMzvLF`z@jX}1EqsZ)uj{NAzRamGag`GP>Y
z+!zA^Ctk>74v)5ru63;LdmY(z`me~EPyZT^p?`owK*sCg0)h<?8WQq)Kl^A6Fq1G3
zQRqo?0GrD-4#G;)e7=rV;4b7U5aNO}YGiGPTP|7`CA_oVeqBaS?>9a^AjIaV3w{bg
zHGA_<SjG)pGDJcHl9GO1aU*JMc_u$-N=p7&O-Hf+$=7#ta=bI#N>vH^I#*zV!P+jJ
zz|@{2RNs9ni08EVPd(JvYW|s+s2=i)ou79{uyNObpwlDmH@)tEN;Il>L6vEj(~0r;
z$!|#xF#`4*9xKp~UH$s+O=gd<&o-d?;ObsgB;McOhn-nmQj)q~8RG-wL~uLr{&i^}
zXGfEgk`f6cIg2tw`4Aj~w$K15*NpRr6f%(zpywi~y6Fiv`}=72##m1OWj>QH_3{n)
zIYwig`mD9;<)O?W8h*L|yBRJ{GhYfv)Xn56Gnai5h1Z!1|57;eTRru(*z0r~pJAK|
z79)I-JW(uWeb!}dyD-AZ+B!O2MMW%M2AxFZz^B0>VlkQet_wb$P|@&rQX}(C?YTzf
z2W22hxTjh|zi#66uIvbwS66mc>jjc@=ch3~ovqbf4W$yUn`3#SfS`R=B}UND(6s3>
z2H*yML|SM-|L{k(*P{jGwT%JM((dPf*FGD*aF(^6uR8`4N$|TLR1em<vR%m05$wgD
zX9tF6E^T@~!wz#U<H@hc!RcZ1#J5X~0~duJibi??MjT^HON;O^`J+b@RukK+7NLx%
zWON!;g`gGc3@2YSHDo-~?&^N#OD$$8w<6eYb7`Av#4$2DIzK;8K3DgITb}@dCeGD(
zWd~Mvk5C~YAsvg^3Xo8viJ$P<Z~klT1W^QaN640i`h}^-@3v44Y+kuF-`HpGd1_iO
z|H@QG(w3ONww#sMHPdE%bxIHQtRBQfaNYrc2l%+WtgOmnnt+UK0gA-Ca-)Z<k<rcp
z%RqXD^{Rf;a&PdkwL7oSZghKC7tH@OZFEzUZW7dskn*6a>I?vK?dkN<(UEi#o3N;;
z^Ttq1e}7T#rXL8vk2IIU@RI+&@K7Awt^Fg>8i<dd!b3hityZ0VuM2(<7^~de-BIG%
zEY*u6DaFjpw!r0cpMYR{d)rR36q=Feb`rx8F#>UMGMDr#)>-5Vtls*k<ot(^TrU4k
zoIAQED=ERFij0Z^P5xm3A#5FC=e@+V>c6{}%?ni5XbH8$GPASIOifplnW40+)mzrz
z^r~)Bt9;A|i#G7I+;L#7`ZnZFByRUskl4~v64z1jWovXhSCi)~btlr$$lIlU+{CS@
zm(p$kFaR49EwKu-zCG<r7$z%Y(*x+{e@HrVe4m6dbMP?k@7#x&-tSibAsX;CNC+D(
zrkRLB*Or$L_ujUz;p%Fz<xMmwU0(c4Nn~0(@b`GQX$9#lzuT}NX8`O`mmV^sfWwD7
zR{54PI$4!m{kj606P6zFl5aJB-;6l0ucgmCt^qm(wWLOcN$KWU@7JKlPJH&^E}!5c
zfEy0;wJ=($8ONT8;_IJFjWPUO4kX)b@b0a>ii5trYc@gfC>HU(i|56X-C44stsFB6
zRc_iaUo^~WeXC5Bw11i;BuuRgTrM=EnmDx|LfLQhzVTXQUKc00In*4~D}|zd=&W<0
zu34;B{Mc!UpV7znu2FdkpAoOOAms|;L8bJI<n0GWYY&|jTobq)x1hP(FWS}4fnzx?
zAA_8noJI$W))MEjgTtQyLA1^?!`>6`OPs8qjh^)&VtpiaHaa}qZTvPCKTnCvuqXP%
zDfxFBMhhauUz^l8cWbR$yBG7L%nbk8>(y65AEHWDA~=-=Z4-D7*DmB_W%0$uIjtk3
z4(E|T2^pqr{^R|(z(#OsK^)BCPO%(iwoI~HKwLoKcE1q%`Xgni@0cn@Xi-7hc(6RJ
z(s5)eKs`__`IHqQwti);xr0aA`eMW;k42Lvl$ed6Rb;J#xv&L=KT^tvL7$OUP3q30
zUZ3{DUcH3U_-vY&f0_$E7HNsNy16Y#4L&y3=do%<N8CQzbquukK%U>bLwK%Tzm=O)
zZq`o(%3}P1foBpg-ae;{NgIxcj(30I=;*kA27Iw;gGcc!PtBG90pr+|n1oV0*GfCo
zkCvA7H*}8?H1EpZF8#87&1v`^w>7ldn$WQ4o${yl->((3HKV8_Vx4TZqM^O#^MWQ{
z`KnngYeX6aW3@s?V#HkO6LJ!>vCQR9f8)rg&^TI&TvLlT#v+2rb^rDCXRqh!f5iue
znBrB-PXas6m9hgq#?N}`sGZ~zQ099(VpZy9?)dKI1htDK*B8-!`@TMsC0XiVx~q{O
z)nUZo@_g7Q%^fGm01gR!IcHXdes6!eoqurg(0O>$F}aj~{3Mg%W}d|7cE(Jl|6wMJ
z&|tHEOh#Jt*^v4nIO6b4=5V;q5m*R{ghOP+(05%S9UYzVKd_jE?9X6vc{(ogf>780
zMCFa;Y*q5gp0~{^Q&%Rn`!w;BCqeCc&d$!w*27_LT&{vaH_SpydwY9z)_j;Ve%Yae
zV+<Y9Po^#=e;yKo1$6h<uVFZPkt8#{z2;g=J62m?^AvL)u^3ZFzfE|f`5Y@wtfcYA
z#(Iuo;}6Od-c#P!*38meMJYrDyupV3D&_Ku<x2VqOiw}x)Qo4QrXGA9q5hPZI5Ti_
z{(T^kmD*q!8DW!p2gSEZs-w%>F2((Li}m*mYjb5mL`?0tt6EdyvHe;fmU_co5!M}w
zZ>BfWQFP8n$H(^N(R?2N7$8mON8k27+XQ$E>8X-)nD!43>nF-^>rVX?I89r=BB7wK
z+`)r&q`o68VU-tuMjdP@=dKT!^dYUORP^CPWxbAY@*9($XpLtrV&eOUHaGit4_2`&
zGY>k`)fG4$HsqXsv<FEOYn>B38Le3TCLAnLb1|AD>pWo78{1uVvN1B(_18*j4lSAJ
z=8=dzWxTa^E>DBxCqKl3*;ZkdnTz9}s`y|+Y{ws$<6ZTc2QWbfX7vtblWI4=VYL@&
zA({X+%|QC8eWWxVIW`G%&%{Jkh3L`AiFWsRK+tP8mv5e;e%=cs^{TwacnBb|90PGX
zJ808JOU*m`^P=oqSpijjKtcje+8UGLZ{MW24<f&OX)$0@_QnIXiK7MGOl9kJ<IG2P
zw9v#KRh%hr>KaWm@*k^#G{rZlSgoE6^<#pS4H;hEgD1!wxW00qR)ZC-7BKPf$aX95
z_iOy#TqaI2zatR=t@o7Ai(u4<pDwWTaxad4wU_SY$2}(3nzskyG;j*|Br|m=B5{Ou
zfrODa*CHjxn@ucINo~*1>0o*tL8^|`wR&)L<N)(DDD#z<zQE!cDguZD@{jdH=tp<@
z(^M!Qc<?qyt`plmTC1qAJv@J@RmAklCnmv1l1f`!`%j2jhPWa^t%+l|LMPqml|mjk
zy8pZqs>lAhyboR$Dn%E0Nh+$`Ct-YO{(|=&J<86>Q3by+2uicEvk$gdH6!V)77F%!
zhBG7=WWl0brCrAa!8OXI6$3_<)lJuhGjGE`8wAdM%X*2@mCgKUghQ0;MBy&#)|j-E
z)a4{yVd4;c|AVh&L47$bm73qv`T*NsG>jxDIGC8>JwJ#UZvVn)>5d-o>B5T0?jrWf
z$ApA;rq*H5iZwV*nZ5>QSYGU=@&OFg-q%<7zPzpN?O0Kg{@eLnTQE3E*gz^AjoT?%
zaO{6vRF=O4mB+(>508xWc#__`_wg4Uhb;*kF6yZ9v%f}*?DW+aDSbS_K=X{14`3B<
zk3AtA6p|%x4M6!&$2B&drC<s(ZYvS%IpjMnPcvilTy=Gcv6$f?aM1g<C|<l64h`iK
z&m?~vX%n<k{+5XDixKsmFN7jEBA#po*-CC}W3O2(K6P2Q>Nj5<PwG&Jr@-hx5lm=W
zq(1IscdompRNMNVLFrri&>`(wj=xM!ffHb<x9?=?AUT_b8cYFHR=EhDa^ay7^baD^
z&vzd_k$Xn~YXl&e5PR4e@<cJ<tU(i0ci*iNiNA_IeQ6}6b@;7t9CGJs3vdx=ljW3D
zfjWi(LKFm7HS$5a<<kmzO7Es<w+G)+J_sXX&36@17j&xnW_stW*O*yjnm0u9lljQE
zf0(}C)JpY)Tf?o<C#$&*PK2D;FI)$NOsk8Zzr;p#8)cMhRg~uSw{^$!5c)XOGv|Pa
zLcyfL09?Qp78amT{*2vA|9NO=C^wgqo4a1nJEa#VxE;WR&(HIfLWtRSEuz+P6*0WY
zakR?w0dWGiDRe~YV$F6XcFYlX?Rq<imKwRf_Pn_UUGI;sE|VW0kc9n@yH^jO)db);
zBgF-4asQMhQPnC1q<+t0fyl8B>HYrC?X(|0qH(>euwwT1^=-pO$QxdA>ojX}!5WsI
zYS2<fYt>zz0VT^C+g<Q2@VM)qOMO?Hs$3pNN(GgrTjQP9vCU8CY#S}}a}QJXTnO(?
zJI%X=JVtww7ZoF(I7zF9;fL7D%^P3OWnHn^Djg@vx8Jg#nz_*o^@uZp9~;c;EgYVL
zi9Kha&r{BzSVJHfxdQV-WMm}x!w=A1@;?O!*Lz-6fgcS&?&D(|bb-4}@5Eb?(Q%0!
zdYjmfaKC%+5vO|kCz?eQ0>Si(m?c13=+blPV5}Q$eM>@YAWrui^<$jvMPl#c&tYLQ
z;!YHpJ~BD$Np@jib96g4gE{fy_|IvSqenj4xZdaLZ)eJ=7?qg{>0f!=W1raQN-d)!
zw)Xf4elt}(iVODs+!UKGl&4e&tDI_y_OZpR(Km7FwH$^BiVcT6OKS0pfu&p&7vQ--
zuMQRBrEH4kUlly%HV&N+ZpXHpKd+Ltk8l3G*2W=bvzSVW$4LsP@nCeAotRV5<&$OP
zCBgRpmOVQ3wat(I*CAiCYLU>{`T4*jgLHdU#NU|{lVQP+{O`@o@)IAtjs~RobNg?R
zP@))#L=bU<{SLO}a8hnIc5WDkgQLiK+NbtEh1+wSIHwDMbyccH&kBiDOSATY*P27q
z`X20$7K=?h7BzRc96SD~-d|{(sW4SsKrbyVh0}FA7LJe4%&61FfhAkJ&LK~wNFCy)
z2Do)1B_lP-UdA@u4_XYO*lYVunOC@^)e+r2^y_fZ-(Nr$|5WgF_S9x;G{-OQc?o;s
zRRAlR@F0|Pqz6P8$zQ(g!{*|4yp`ug0}E6*gBk;??~Q+4Al{nvG6;JuxW{pQlkYxZ
zjOjmjp8w8pK$mwJ0h(}h`}J0REW{hvhL6#!Mg+>JRKevN8z!Qno!=shG8IhK)Z$A^
zZLJ0@A&~)~iPivIdr4GCYWa4$Y5$m;3~@>S)oilPZzgMnm!z3`Eb=q9nfPirKR6Mu
zqcHPZ47}F%&24L+DCEuk%(ox~q3rZ%4s$A>5l@|{hZNG&a;v91Gm3eKdwj~EMpLpy
z2$%lpR<xMIMs#$n4<u_OzGA6#sp4~qFVD!^1WXH_mU2}&<wFN9QoVah@6c2E#`6?K
z!QvVt0a*8&dcOg!bfw=K1{bDQCJV?qKTqpI6cYH?_2H|-P;^<cIXA+_BYUH!<YOfe
z5*iBriG;mYz@a<a+j%sb)nT!d*Nsxf!opIG{xHgbkfQ>V^g(Dogb)F;0i#Ly)c20Q
zK8x-$Lxi88uGNnWu$ydEtHeCIlV(2YjoYbYKp9;k{4MTzH}#i)Sz^(Yy(Pzv8%H7*
zmFV!-tPtIzDwIevO2*#S)^)uUhlOZMUGrduX=+%W=26^=z|fB$h*Cy|0^g~4kNvw8
z;xJhC%4Q=3aQ^E8lx{`YkzTuDqtW3>;>_>)S!(LIYLnsLU6YuF^9V$udK{frBBQ>D
z)iP9?_ag=@7Vs=iX|-$4bM9<iyR$ktL_}lyBRg>Y<eS_)^PM|LeXFJAIq>zj*?M+%
zHo(UV)ny`M)s|Nmp>WtaI9&cY(Zl~S*}JzpmmX<VX*T||8Z&|d6CSE`U`*Fgl+SZQ
z37g9S5^ksMo?=o59}^~%u7XM3H|dmg)CJPbNcEACCt3|JBP3!;x>8Zq1UKZrY%dYB
zXj37wQa)w|`p(UjjWu?yat8TtUN-noc!dalu`gdcteuN|6nx`~JW;q?ZNZ2+GlhOU
zT3Oq1h4G>L>vaAq35pi|X{{X9MM47l5h-R8x*S!GYacx&M-wb<40K3$2vKU>&d28K
zoxwAA9hGtWNETANJ#vB57K%6w7+jxuz~ne+R6cpBh`G}p_3A2?Hn2*@I&22^T<qi@
z2!7;Wdfocz8wp~U3Y6zmU)GXZ8FU$wzK;`?-~LVfKAr9F?*zep2F3!4cM=kyYp}X;
z@+%fJhs6SW*2MJuOs&iSYjOSxoo`-V*16l;{^??~_ns(0JR|XsX_O01h+{FQEHfS?
zE7ZIrE&b{3LVYzkryZ)G?f|_BS8Bmo>K;K{uP_%mB3eT1#VW_n;Clv-dBPBN4t$pL
zqpBCluW~tV8SzEa$BEUb;seg9@C{6Xh*odMjUmBidwFN4=%AUMjqQYp#b)4=GwoiA
zFGh@qM5K=(+6xXZ4d~qDq*KoN|CTtbh4f1>a+#lmADM?Gu@?h$D3~rs74t~Pww}_&
zBSn2kx}&-TA%-V9Qkqowu9y5=h_L6^lhK>%Dbh;m?UB)I=OzAb=QnKTYg6raboE9;
z2a~>gdq2U`Vw<V5OtQNox`oYNG8^mn;vwo!cWU>WZSyx1VyIQul-ja~`7hn2LEu*b
z2|0f}T58UZuW1vCvQ-XV+%RPn6*c}Wo=uGDP0QQN?5L=&rf<4}ne4mmov|tQ6u!g#
z{ciI#idWrn(?0_f-i7;rdnG8Jc*$q`7#+v$4qgst;Jje6Wv|g6Mz3u~z`q(kobBeC
zf3*H`iuJzNi6LH9P-upfTV*JlnV!{rqE9uRmoPyp_7r=)xIiKc7Q8rZ*S~kI%8@>_
zz~f*cvYh;9TORB3qboO#(*2+JACnL=9xj95ZA-?oU+(TPJNDqq<|s|g&gNWhSj(GU
zx({Q*dnO)9A-?vOlKGwMWVV*(?Vo&;_y^-;@X}O4V_{(!a(@t)*01M!yp8?d<44g<
zWn6AmtbpZRY_!9D&xxoDKS3lS>4^&e8bx;H8uSkkrwEqL1D6K*43Vq9s?x$R5<?^q
z@V-WRdJ6fSqENGQ>9oLz3sY`>+Zzb;z^yc2Z(wAUmXT3uBw;oRsh)Lk*bBnnW9Pu_
zv1|qhaKyXA^bGr>xr7OF-Z$I_>?|zXyaA5(^z!LrsRqET!7u{;2Rd;Bt50$`tZaz*
z_)HHEjHM_wf<#)J{vNLgT7osKHlw?56~oju9AzZ&a?o=Aa)$gi9my2e{IOb7Qv-5~
zNirsyoTl6$(}Q}|tR1R2&|#cN&$=G3*bbS51?|_45(6wOx}~PqZCCj2V_5S3#;=;q
zQA7w3<S-8)fBezbMxi5y8`sMN*vjeN;_=ZDmHqRggC~r7ZEINfbth5o>BP;PQegg^
zo<2L7Q8t;zb`}EiMtK}1^PhAYYNxd%V968rXe#&iD{m4g*@joJQGI@&QenIGv7Z!~
zv+&|m4GyJvK487RkHH1dQ~mv%AKyiORP_sphycDvD_*ThSFKcS7fAuGl}lhQZ>0sm
zv?}Gug@FaYIXJk{evfX-46|uoSAAf)2BiAN&59txDt-Lssq@bSitf9msQE6}u|+e(
zU8^T2dZjhXBf^GB22?0ecVC3Ovp1ifyEuA1HS>F?e&ZgUG@mHN3Rc(B(5F}u8E#w|
zZbunv$FL6B5~lBJG|Ej!@T!EJ>wHk=`TB<Uy_W5-gm$5vQG8Oaqoo#<ND5#dq6<X9
ztifo6%)+W`7O)zubDrE2ia$OwLS?r4NhBl2nrN&fC@*hw(%u+Ung(pS%~YuF5SUX{
zRh7!?X|~^7>M!D<C>M=HCAsSS^7ZRYAoxq@Q&MNX<;hVU#!$kq28}PF<c;V4-RECs
z8lSGssy-Eu?2mQYpt2lBzwvyuvhTuHzOFZ{fN;5X0J!op+9hGS!S(Qw7y3*^dsc8@
zAQzC2C4E|7C|qx@JwNgf0XZ~>=$@mAfsz43jfT4Vqt!1FfbgFmnFhW6y}b>avD#%R
zWDMK1FWie&0bpa=Rv(~Sl{H;e%6kNxl_64#8&jC9Y-fY1@d9`IIk$^7k5jm2XgMst
zefs(<^=-NwM9hI9v<d)jf1|id3P=oMPHk)xBW&>CrMuJ@SUgS5RC-6{O{5fy9h3FV
zhk&G+4^jZ3vBFD5%wjSWO3V)CP_v0bPB<mxv!sy}%U2pF$r^rQsQfC;>6tnI<!OsK
z<iJ8}w-H)NoR*HXH-DWW>Ik7o&j$=wy1rD$fp74O$<lQ*-nH{^BL&!|{rpgYx2?5b
zKhHB4R3&VwZcgFJV9J%l3L~V~c1UVQG?MJ^_Sc<Q-uavZwv><%96-?s)+!IqE}_Z6
zWYP$^zMwmY4B<=Hl6-h$Y&pJqOnlQT<)EU<zcn|@o~uT>JR?}_>Z?onawhz||Ehs5
zcsjITIHq9Xosdwlj8IVT=;7HVq`bo?$)|d#LJ3>q8`lFc5P^6W;HNXGPH(7N4^Uok
zP`0GQbMd|lunYSeRm+>VqyGO#3$VFx=80{sWoEQr99`kSd(mO&X2Al@zG<q-df#zU
z8xSxEo6b`*1OKyaHu7?3cr%wNk%PnE&E+w6`p?xOlw$n+KW<KTL`e^w80Et&D@t8B
z5vXXpjK#mUw?n*9JLhWc@tytz&Adb`G4idN@h`|Y!J=`pfU>d0_1PZ-pN!_L>=QUA
zdLp=oMxDC`-T>^AJ~8=1K>2OI4i=`reG-d$8O1BQS2Q$zRtv~#8q1RZib9F8W8M2Z
zUzM)diL9A=iS+T67O<pY1<p$<whBbehD7nnW>WYp14deYIgqAOWR=$qKdEn>`Gi;q
zD@-p972^emk{lyE6MV8a@neOIfh{J67H_?er0C)Zt&jyG3*%(pIs0POhGdQ!?r@3E
zIz9J3ssMe4hSG0-(|6TY$&<dK5xu8UiA&qtG<Pw<l;IefYSpPi3_|tT$ymACk&gwu
z&XdI>Czs7U3sW@@9nO~t2W>aW`J~es*s&B}MoA}g9UItS&Nv_t{y$}w`NDq+rrYf7
zJS|+rS#tZa`r!lORSrSzJ+QvUOcobx3t27HtnKZ=j{kQgOh?$nB)3_f7Piaxf$LjA
z;^O(rPLS?isD?{QYGAlCS9=KLkN++iZA>zu!B^Qq%o}4~U+KGAwM1d+ni$f>0{5n0
zzyzY^{f903vn5?dM;NKb*pv||=jS5rr=PgP$p;{KOQFx1;W62K{X=EChTl_=l~PrI
z`YdiQg`w&FLmh1F&pm<gvb}=hGgGP`*fTsgJS=~#3(E#dIA19%lhnP>Wzzk3l!^c6
zSUq&Jcz_Vb{V!-a&7&p#VLz--4FrncU5Ul)&I#}_FaqJd(6Qg}&8~WejO2c!1A7Fu
zy$roc#aDx7B35<ma_vH$J6c+uQUZHtoK<2_Z1~>I;%Iy8X1pR|6=;jIRlF`wdHzK#
zqCPh>@132bSDsR0(y(Ga>7C~fu)h$S_O7n+a`YH5<*Hy$ZTlSFot9Wd6IgrQv{WmX
znK*86_pK`K@iQ)U&x)G%K<@y5|03_3dm2?J@y&Mc(&?69s(YbkR=dv=H}Y?)>Hd%G
zd2aV6FRg$cC&RVra^qROWu?V$36eYzmaipv-wK}1XJ2?W&#?XV@_Ik%Uy(R)vc9tZ
z%fimCD*Wp<I`dB)z3)JBI=hKdmmaD((_)jtom%lO1B&ElJ5$Otb)M5WY+eBnDx;=`
zKnVO8V}bZ4US6<oqkO13Ij(SsX39}bm%E+f;ip3b=yjWR-<|m=5^8(31r{0QabBp?
zsl@X42njLnNRH)xl%8GU-OEU|ZW1E|Bf!0YbFWw70<z@QRvalO)V~;5SXo7r#<*-R
zDRFRZ58av-`RD``9ymEUCEr|MkbT(RM#_B79;!E%D?d3=BoJriVq@Jl<$dSCGg>S>
zGb>e^(10o$ftgmx)hVuF|2-_$tNSO79bGWC{UJ04K{Sj|k)-Lm)m@Yexg0vW6ro-f
zPGTCG-@kuTDcZnPwSp7>^u1Z^*RN7oc-Wt*0zd!CJ)w;0iH?3CB-E-&eT%ZCiZ3fK
z5XUvY_1c@e)!)|cD6eiFItnfya!g*D?%M*PBh?m$663~+_HcQ58CrfJr0A`zD6M_>
z)r;shAf(e211=gKN?x9+ot<3<qy8^0w1kBDVy#=y+IG3)U->%FQG==dMt8yBO09eY
zK?Lt&qOm)hFmV$^dq_!0n!K(+)D|a5T`T(bZAH>v7*^C@TDy&$TTuk<Q~*Wy(2yK&
z>LQF^-$q{F@u%p9Wd<fc_|`5BY<%S?AyZ;(Xy|)+c`tF&88lJ|vIp^sx`u|N*a#ob
zG9}+t0)sJjTYUH9q5>X4)!D*m|JpN;rsKc&;yIouh!fr%!c`GeBDY!u3TTKR0)#5>
zjKTG>AG8dEUQG|>Ps%B8n_f*wN9OCUfi0)wveZ-^VbSjjj3n*+?KV*Fs3=?H8-qeZ
zG{3+9D9lAK5y5EQIRJ=yPEHQH`Na0-reY2{r&+1ms1<7X;+d+pYKV`I@7bEadKazW
z%m1E8LAtnUZ&hviZ_x3Ec80X)4vR_tx!Q7d6*gUI2asejpEo4b_TXPI3a^oo5f}y`
zvjU8-@Jv*$;udP{U$L-M0F^8hfb)8XJ`jSiXq8*=($?#gZ-2Gy(%9Gu_@{K*4Qd4{
zv=GGp`L0?Ff0?G&hfY2Sq3+{1<!3MM;NaQOPSg~6SFR81<L>%;r(P+{=@M%5+_bcy
zz}MF+ooa7?VJ%<L^NfpY2GmVLLbpdI#-=0(Rg_WA2j0y8c?`Dp_8KK6y7^fNR1Ye-
z2yF^<1|3S<j?XQ5*to?*^}!3Uvcdu~<c9i{ha`RsNuJBt|2-%9P!=1MyZ1C}z?)16
zBu~aU#@{5_5mR-W^Bu?jGC0u*5O!U6|D|rb8B9@qijh2q5}fX0^zX`6_(va^%%Og{
z_n-CRl(I#xDdV^vH93?iu#{O?Xxb4^gzI%Lj8z=^NWB(EW_UlO<>1({VK(`?5!XcE
zz%UQq2pBuyjgTm(rxsP3tb(4i%6cg9|F*O7B(ORB$1H?!BdIFlnpKybY}xzm%h*~=
zvlf;^kDHu~fk&b?)^ucj?<;X!P@&Y2$;iac3)1@ZAN{I2I^^W7_6wQW(Pn7r6yK*U
zFh}h3jbF48tHm`T&xM0L>@LkayR|m?oS90MLJig3E7iMDg?F3v9!g2qXXfmHuoaM=
z_G<k8o)ekXKm)@9vZq9)<Jp^k@FphohP@{KX2^lr8}cp%D;@5VJ9p@(Cu_A&aMS9*
z7C&LUX&0XW+sJ&az47W`lbqaTf}Br=8#7!Pz+PfW7rK>tbIB^MaVoC)jZP!9R}XpP
zy}}=nw{O$k@7ZxuAF=%zXO75JkkptP6qm36>s|~R+g=D;SEUtoik&oxv#XZWV|aX#
zBb8G29<RbNVCxTF;fI;$@B+enKCQTWrNU6}wDa>^`hEGHh{z>??oce~D-t5#@*Ay%
z8ED6=A%8$@*KAp#yUo{R<uz%zL7gS!azG_w|0q@E+?u%cg2QQd4r>rsPPLhkxE$V2
zJ7?0Yv^2^G5;jWK=y)V1dU^*?t_Ad8(ta-kcyYA#s4rqe&75bxIi=Z&^pCm?;q%u5
zSAF8@>Qsn#qid17cogMbuWf;s2RUa#4A3YYG=`bFSYVeWf><&&n(M|t`Z13pDbnpQ
zhDQgxx+tgk1tJMi6SQByiAhEeei~q6F2g)z{-ZdgCJJbu+~=I}eRp-I6+H@2{RNR?
z?iZwJr%ysc!MEpWdK_NtpyKfK@PJf65KLmnvg<dww%|3P)rPFtwBcQ?9$VxaL_F+)
z5FQBCEP!$W>*7dp(lr+&3DiHW{WVhc>Xw+^>)pdc_4Zz|0}HnRRqVYaCG-zM|8;C`
zNwE(VbQeyo2R1O0Ky`LdE5-IjU)~|cXO8MQJwAT|0AJ0&ZsrQ}aC6ncJ{#(|%_md=
z8IN-x<52|&0R)oIP7svom+-((Dc_0u!x<yaho5ayW7e2?dGA^>3c&c$%3VkAH!DT)
z>bS;jdaxvu_X(93X$#U4sA)hvzA;x@4n~Kq`%fT92aQ{CR4<DOp-E)v)3nRtIsa{2
zpf}D#AOo~1Fd{>ok``dRei?D}urmRpqy@#v!lFmiUePDxF=dMY7T&|?rQBjL1pv5N
z_Fpr0W@HuyKH*;eTUi+c9bVDpoR3et*htH;gYjvJA%g(QpmZY3@KEkl+D8IASlQYg
z9;sKOeZWmgb>5?ixoV|<!Ex~XRkyuQAFX&^e%-4Xhq{8M0RtYbd5JGXR1#_pJ8^<9
zqZ*<A{P>Y(HDCA8?dks=?4es0ySTg4kF7gdOo%@H5(Y{Qpu!+F0DkD}e-}NN?yYzY
zB3*_QqFQs#3<dJ{qqE<5GXA@`s3+q=PYO^gJ?L&~d_)yEM3n`b2PTflpQ?!7h{#Lg
z2l$GSe+Rp!zV-!2h#_qZZI0qNXdOGSZpK(5YwUFO^<6IBTobBK5_EU(h7!?0x=+67
zTY-I1n6S5Q-S3r6bO=k<M|(R-hC=9$Y(+by_bOM0enH*Jo4L!BFcaF_C%cD4Db+80
zl`Mv9#p(7E>_K|RTtg-sbE~ku0xv+>+|>K2vGHpAm8_FnU!a1kQ}=J3d_gzUXCDiE
z`ubltgzI4;=t9|qb(|AF7CSle=Jl)q6qK%2-&&0EfmV{oD5$+w`5cncsL0_xR7J-p
z+io(OK8me!{??phJ0^s|#bjFX3%786Z*gyD2lqp4N=K7RKuV$%@}G<*K95IiGP4ZT
z5gN4-8a}8v0fOn2ohg#R&CNFeLSp$mh6tImk;K%SNIxS(_m^2skAjeLM@{Y@I~r~L
z*;M|G+++PgFrDAZUyw^HPf2CdhLZBy0Xnc&X?gjwT_atfAItYYT6<%!5vI@gd)&jx
zGY{Z?lI@wIJKRgi&mMKZD{M*4K+fs+Rd`b2An#J|w7QYS2yrb-j^;y|uXh7vrdE};
z=VN1m$eh&gEbY0^|7m>-5Vr{s3FM<69ySmJO}@wqFfv4u`|p#j8<#!KG4AU;c{iv=
zH9uTA2=0IR%98PJ)Xw=!IuxqgL)}pU#|vmR#ABR}+{ywyW+t7IzZ6@u6sdgOYo4#f
z6m->=1H$Cf{IGlw$#~{}0N(<zC~guV67qN$elid-Rf)Wn*h1exK&?B<kn`PgxdXJA
z<4QIjF=Ff#S=!xQ6i7rvvXb^$rMjkfimtE?jEukm3%W)CLvCS3WmX(g(&Tu_Yaj#j
z0XBonzzWPBKX~u}n)W_7WKuzZ^6Z!+cD2_PighZ^gc6E)j|Gvq8-DKK@~<DyD$H=d
zVCw3wdYyu7&`P-T<|ZNlmsqb|!eJAg3pbpME$Np^<??G3ET*%y6{Y4Q98Pfpem=|o
zc2=Ci@qC?7f`^LPQdrv)_i>Zl`jc-w_EIE<yeY&%U&_bHSqr7kwUR!tbrXa-!sN+N
zJ1L+`aXq_>3AdDq8b<!p0fs`$oLCya(dfT>-5-tB&qFWpk>r3UNyGgE!?ZPAp*3!-
z!%FL|{ssmk{F|Ab<>d@efm$xABp42Pr{MV8Ih)ORX(x(_BX4A;H}!R;Ww|}`{=Fp#
zqJgK$L3pG{KTO3S!O9QoKGk^|SUdv7kvl3G`p)KqmRB|OBH%M=rlE=+f1nV@$H#AJ
zb=MjBr$@>IH;qB=F>r+|Y-ybyhh`?iai5r04$4_h<R(pPlM>l1w^o%6ksiGWq!9NP
zY<VBO%KZ+)lt2{}eGx(R;I;U5a%|Io{y9|77yXK==DLPt4&%NvpFM#z{V!ipvj|v-
zp8qFZw_06K>kPK#UnTd{%E=H84p8d>lvw%9ONoXiY?=<AYYDKpO5SGj`LgMR@L$|+
zPSW>-v_!(WE-tgEWc|)g9$uWV!fj!#L=`RneXVrNcWcP%n4EMz`J(@x(YVUA5fIRP
zGMRQkWK|}n$(Px0#SIZ=%n64$$C?h6n{WM<oQe$MrPhY0%R6Y}QDSFSjHI`BcBs!k
z;h>{@^OnkDV&-&P)f|og{~y!fdXkxGlav&h++d{0-RocX?p;C_Yfgz;(;N+x=8i5U
z!c|Ms_9oqcRg*;=3`;xCM6=GwQ=T^7C7t23mFBABzWJ2928Jc#Qm(IJ5kvYz>qJ|!
zL!-J^j}-ok1FSsrXj3SXLjLu7`+Ej9JV*5kz1L=A-v`CW-8ST2kVL}=skohVh5$~&
zqK-SRV&J>lx6f80N}$6`<+{GZ$74Ad<~E~=C^MpM<GENdFjHc8c|#^8Wxp`SC@UtG
zJMv5E!CqS=YtU=wvN9walePz*jS1*(>dhs(Cabx)5AVN{A$jA%5Z+l{QIV-Uc!P}n
zc@cmiC4Bhm@KMn(E-ps-Jk=i%*fqIh5DAP;j-nxmr03=ZCK?=-!%Mg0lagk$UIxB#
z!<3fRY@V1teXAUv`!y)0d-7G5rr4kyvV+g#^o&*c)*m%K!S&as5X`S!;-ZWFnGZML
zWwt~H6j{(wA!^Ppm5^Smj)K$OPOrQJQT9M$rpxv-SC*I*xS|Lj$xE(+(i&JiNVNv~
z-vz|~QZN6E<=rn=@OTOVNDxgfMnS}85f2b(xO48~o<<jqM9tFw{7;?L>!K-;I85x~
zo;;cQUa(k3m~XsaCE-i8pn{{c&zA3!gbnb$Ajr!=)#(lYuVfq@EFehVop#g(x-32M
zE}{hbAtDlRTEtPV*!lE*GZ4;w;j{tG@yjeO33(rVupw&vm+@W&<Cy>+YpB%(6I#;z
z)nOo@VFh6qujImfgmRU3*RyV7dV1w3kt!6{3kyxBin-S775-;pkjIH$N<$6AHd8ny
z)C2IVNUtfZSi^WBtOzdc1FBy9BYW71Gh*rLKw3^NA|vBJ=F1HAAHv9TV})6a4$)jW
z9&?EgvD+uyPfH<_=KC~zvscU1KEh`(#q00V$K+(KT9Q?K0g=eg(UIz`%on|@2%2%V
z-VmG%Gh6S18aWU+e^w9_R=pB@J733GXX{HUtHV~L%MkW*mcf-^J=M?)_oX6LOquo@
zU6@`dSFr6RhyRwY@R|KbMP&v6Xu#DuMA3X4T#x%T#6pByJ3a{{R~BbUR;0oc{jc24
z14HT6>H4W=zb@r>BYv!g^vURadqWTz1>yc&p0!_Ev?LF`qG!_5(*7-{8*z9&QwvS%
zfJ>*=_OAj@Fyx{N7`GP69r07Q%BKmlR?vzQ)&EK(4n+Fj;$21Lh@3%xv6gl6)eWWI
zYn1}Xs@zE06kqyoTfj4!;K@S2cgX`^bl{B1xLW!q&LB#^gX#6fw%_2oEVmPhMkD{-
zKTbOvC+C$#MV6+b3Z7U4J=$^6=PDHFY0^1^3`5D^N;=c!C4e&aLC$4p5x{QZb@?l=
zS??HU#A+XRvzkY~%NG`rRJA)-XjF-lSYQGx4y*ArQ=Z^6Wy%5XYq#gy`qYlcPhWWE
z_x_W_X1rKcM2tMLq?jlDyXYsBE{EAlgpeELIuYTY4g-5yCe#CpmoM3@=HlD+`Y#WJ
zsbWC<@O`B_6y`*Rk|bufKj(1yq4%dH7n%n}l2WcT50Kv9Z=>*>e~0c?_ab-i97MJ7
zLPLa1ZC#y$)g1K2X7-2AIihXjoKm`~_DTpV^ixfy`<xO{`Hh-6zSB-QsL8G+Yw`Ym
z!91g2--pJ)$_i;~CUSCe-r;DdHlF|9Bmechsa>ufw6JBv6r9U&PTwr)Ph_q4+{NK?
z@VjY;P8a8j04f;_hOt|Y5O*t4Ump{r99J_{Z>{5-5jI;r6V9!&_C<+MJE)B~QbDlH
z4-t>bC@5+!fa|ONqSrYzG<2ud;fc|pFaKW%AcTAWhBlRO-QZL%#(w_vOYY?x*P?lM
zv5QzdA$1C#D@qguB`@FAg25;K;I&4Pie)~$sOz+FKCD{JYbX(OcXw|g6;hJ?PNRYc
z@A#+>XMS7c3D)knMD;uG63+P6*IyK$t3(MRhac_m09@Kb<l+S#0e%m{Hj#dh)2;oK
zW*l7H0yB!3AEKoO9#RcLZym&9BhYb1hu!2#c2B^2m&=BD<MHn<UKm3$S`xZ5ZYYd1
zz2+8RFSt#y(_q1eaBpEW`FqSH21Ma3CLO3I=K-2XI$}P;2&p<vN(JP*hId>VqzKGV
zZ6NCrkSs^XV;IU}moYJYAS1t7tiyW0t)l~&B(?#GoA|`FU-sXjEFmC)tllswV~-cF
z*g7vO`(-Tk6wdiKJj-mIMiOoZlZ8af+M2q$*uT)tGJY36uGQRy*MmCuP~rn<!1D4Q
zlE1p3)2Lyj8|DiwLB)KZmNI}HrH(6^x;Il}tG~T~Qco7qdyf?E%gid)#^V$P2#T!f
zf}W+)f4%ONd*kiRA%~v%a-Npa5{9mh|F~P;Gr;Gyt5xQS(7EUH)PzNSdW%WnQbtJ$
zr?a`l&Po<1DP@QL++yCkMdCUledEh0D6T$%7jN1^*y{6J&U7Uz?@I1vaCV+<PYc~`
zgT|V~(#<$H1*0OKWK;mUv_I2X*Ul^enYrh4?uRe;ujkbXzC_FLonDV_h_E3Fj~#&e
zhNu;9qsMSz<6Oy!HA+;s5><?_Rw}#TaQ44U>PrW3uxV?UyY3{?vWy``|B9?K8#gPm
z99<<lOjk-dR?E%KR-yBbYreTArk?a^;XfsVO@9XbLW8GyK2JaC{9W84{3w!f4^`mG
z_#w8yUGIPGc~VI?IticelK7?YhIU4^Mv5(CIriK8u$ol5)2c17DvHHDwrGBzmUV6e
z^0Iu-p2>f~t-k=O*Xc>8q@}ey%l(CMencTI1)_F(1~pMp42bq;U|B+5Vt8U#t4TqC
z^ppv8%Q&q3#q-G4dne>>N7t7`i{Ca;chujG{p|=<w(P+XxXU+<@&Up>go7WUWoC1Z
z=hp&Ex!5rp0J-A^Wj=Vqd+m`s-3`=F5~%Wk7AnmR@S=9a`;b~250|y{C(mA-@qPZ>
zQ0?e1D6c*5!*oscXk5sHQ+sC-tLGnt;>Gn|8an1YN5$x`u?+6hN3v!)CihEl*zSCO
zmb%+%!rsr@c&0r1E7_o>#LUtDQn5@4#`FgLPQ3H#60V||lYEVwM{#%3a%!hRVTDGY
zLPfg;IUwer2N!ImiP44IPaMe5KD-oL|H><JKQoF*^UI-0oj@o^9>5<RNl|Nkj1Fdk
zeBM`xzeh(nkGYT?`{+F`)2dw$VO-C5d;73G!(w>gfwr6q8v@y8Xl0-zeer8+@s{mF
zP<;08uu=~da+2}HHyGM0;|D_ZejbSZP9c4KmroyL73A%t^VgFhC`^*^Uy52R`3$L+
z2YUDB7Rp~uDd~H!M@D}A+5iu8W9a?OGn6q6R`tgoH}WDKN3s24OL{JT8G@xQQ~c>t
zye5Oi{l)`}aC1oTAr%Kld~D<!Maj^YeJW<A!2#jQTN}tjqoYZnR#8qJ@)nbp7ACbv
z<XYcf_Ad5wp9kg?Y7WqkWKiPiQ@*5A*fKMl!*I`>j)_3xFf*mgO<!CABH-5O%ucKB
z7Z}K%yeaaKIRiSSV{Y1GJi*y~Me+jgg~wGJ=9q|VOOk>@D%Gsz-Q6aB)%n0)Wn(79
zQ7H?bU@or9*_WrE>XPf7o@OGnU%Yr_e8+meE+#T^gguEbvo=)eSvuCLgqCOep^1WI
zeZK5Bu|#%juZt03ht1D_CY3%G^R2*LIEq&2dJ*?ic!PlJ0Tty#p02WI|A{vrg)}RS
z@&>;^JVs>{)3A?C_d=;OQ<QsY?ZymMu$><HA)BUrL@J-d;}P$}jP>x}1br1Or_FBM
z>pvFZN+#61j~-RI=Tqy@aNcb8<_UOe^VC-FjbFqJpsYx|2$l(dBzYtn`fA)eC@bn)
zJiB#DOR0Xzz`rW%MK6d9dNI8U6)v6$g#=Hk|K~(e{t{BGzm)%hjmJpwzqUPv*9g2&
zo8fU=<{{>+tu94qYnLoz$raz^)YL`qFyWnSa&8?s`sdYlPLNVnBBjBXuFr0%1zNbg
zFiVT0W1@{eq+S=#h?ho8Qp<&B?~higWX@2=)v%%aGV1>HQ`pfnG72csfo<C*&*G%y
zA<Dg1QKU;*<m)*FRP|d-8}}U?HE|t|y>}eRZ>KCXROOVW%b>6@ZP37~OeLB4=)zSL
za2FoXG=x&YTrPj#?*ux3C04s*cwkXG7twlMn7%rc8Co>B0JEfO+Yj({K#jaobh&Gk
zs=YD{{Q74O33*_g!F7Y0`Pa>69IojwHZ^^hAPVH`^L=fyo5-h%B9Vx`F*Pez91#;U
z6Qy`%@nioUh1cl$0Q1Pbqf&BpdpL$8dR7iCJlTj<ikKVC<LYFtZXXgjur1j@hg4gl
z?Z1G1Qt?NxOE<8<o^)pB1wqB<-l-Fa`?H?I;n{u`hp)pypJj&@2pMp&zfm84YEPrE
z<)$_#5BH{*%stQK4BRnFWADj~E&E=f29E;!bE+Q!$!ZW}Fr1utJtFWZc07DiWWk*<
zad2^7ncD$(@$SuxZ+Ufo+AgGF%Te8kdi(6}-|MMOtA{KNKx(T^U?9*@NF1T(W4PS}
zMXzi+w0M!-6uia$fPOUcyKeJO<>V6L{aw`6I$3bB$=|!Qg*m7PLK+)55^G!>_xt}e
zxC>Bp?JmE3IKu?D4elG>VEKi}QAMAlKW5zB;$MKKl!QlaW@M!Dv-Q^A3B);&GUT8o
z7oY37^_{irGm3-@k5zwhl)7Q5{6Y~&{(C?C0L|164Z$XWGdZ4;Ut+H8IU)<-5?p+u
zgd^(8WVg~aE2)Zv5T798@o4}2@RsM8qJg`84Ab4^I#pO79_$h14a^B=isVT$prWCL
zayUE|3ivhNJ8iKBJ`ahqNgy^+P*6fcLI8w~mohTNSC>9}8oRx^PmjrzDM7%Gi?08z
zX~2btv=Nyx9ps$~dC}?$uOIv`Rtm(n9tS|C54|Utt^Dmk?E+|thf8=s=-)L*aC!n6
z3m3aMmz*Al6`4~R=&;uCaCq(p{U0sBLW5f-1#9pw8z<*G$_H>k<@&|k;v%^fb6Q65
zIU6@&ZJd}D0v9Zd(`9gH2F>O>Wn-}=8KaX0mh0p`i<m;IE|nyswDruQ?r##r@FvPB
z(ZMAt<Q|@$%d(_=lUe;0#|31G<D{0Rze<gyTa1A3p2!T+K9&l7SjJK1wiv4wb!q)*
zxwIK>9E2Nle0oX6BH)YYt;bY=p}*~|kSO81ca0Un@-*Jh6<)YL`)`!n|5W}t8b5-k
zh6u39dOL4Z=li#3Z}G8&O*DQrq+w5_vj(<cIz8dqn2RBeqklNF?e#vbI<%HIFzR*8
zvFq8J^PZgydm}=_gXGxQs<TaI(DyBkyld7K5P-z0S`BdXUYoA<5gW@or{5!^R#j;e
zR~ALZZvRqogwk+CQA>gVoPE<odVI3-Y?ZeAHJk-6Yk>H8)zsA91O?~i<Se`Gutmv$
zH)swHr`cHUmulQ+QI5gEW9v2uld|~!<>pjA`-QHq++1FN!Dlr$Qd%aa>lzrt!^50j
zmo-XMhNZdjq6o3*?$g76qGy{4*e?(~=$;M518+T(iM-V;EVOyOuc4cXl&V=68?(bn
z&L_izx=h}m>P}CmV>Fnuv%7o!uWdT%KXi@Ftp?duOR%6q{!}e4X6;w|>j&)Iin&T;
z?)OjV5^^SsL>a*$3lejFF5Arz@`_5JgTw8O?^K=bjCb6sf3zFTtgLt;B2v2lr)pS}
zM5cFEcJkG$mlb%yo7pAY3aWuZ7o#YwvA`8M4<K)BESyXmKY1_)v|c{hJ@{CucR<hh
z8AvkFV~oLd;oz-=@0gg?EIOtb?*sa|TetPk8_OKI?^lCE`Wa&AQ6yaXK^bX0apt<Z
z`>od?CEFu;0I_=Ge}1~sRgWdmj3-A#M3fp2=>DYexH>Rw@wvJ<U7x3l>vf)Qa11KA
z5?|!ULYVZaB$?*2^GikvwEH1H|FS_R5LgWO)YcIg<|S-y&+0QctK8k)-J>M#xOqBz
zulcU*0_lM6*c(c3?N~}0z%7&w@GSeo$B{!SURH9(+~kRtOh>j|$Wk<Vl$uxzo?b{1
zU}t8g?pcTI)>FvGQW#iMy)7$`w{Wz0Dy-ap{XFxeNHA|bm!eE`S=Sa!{`zMkQNW%(
z&$5%2>GA~Mul1Kh&i$4}efiPH?8ViN$!huYe4gjg@$pk2zti(cyggkKc*+dlwl!;P
zbOmN(6%*N~=Nj$XDyiYU1}|$zSLtjF&EZY0%-48(Z`C_Z-IqsXEQvL#QFE6%y7!2Q
z&4019oh|q}QLA43`xkw(0b8fSruFV;fckhKhYx`);Otl>Vfp6K=Kx{Pz57q1-mARI
znO`6Dv|7v@6A^&1ea-H_AGQ}SQ<1I!{Fcin<+h$-+8$`+LcxH{MWs1R_>9*iU;`=x
zgG!s7|5l|`dgibR$T8v2fi|SqGUMbQ?;qHb3Aiy59DN(=Q{T7}MCGIl2q7fr*JIvf
z(^FITj*lO+DdN8F7B!OEgF1cob+Dup1TB0tI$l5Yn|zB0MwAP7jqg;yJR~!?gEWL#
zO;R;6yW_*ZqKY!AuFNx~bMOBB`{2MD*@Z&I7BJV+QG|G}eF}-*z;)B#v_M+-r7nPO
zArTa&=H~T1i#Lu62dYg!dP+<9pF(>7;B3JDHSHfSOoVA?@yMg1!(;xtsL|NUI{dFM
zwZT{5G&?{>$|aab()`}4$C~0++RJ#FNslNQrh`8YT%Nb?-;e8p!;zU|7mIpGLs)v>
ztJXAyuEOu7z64lOYhEBxGiRDP-)(!9UqxjC%(V1vuTCb0)$5%s1*)rigw6&D74iye
z%zR+M_kJ)IDe*2sA`77K`dYmp!`P*aFWv0LRTBlqyH^)*7M@k({`dT={W6o7!bD+K
zrqHBMatJB+BMeWQEsBYv1ui|8rXsi>685_pc#Kz<mj{-R=gYu?iXg%@Gn2?F`zl-V
z`BP9(g8M2Jn|0^fjy)X)tf$ETC}TrHgX}{>YEx+q?zUlS?O>>^xl4x>jjI7*7J^@6
zXE<PyrEu1KbF~KtUfC0!uNMwHy~DN-Zj$w-^+Dp#kYy?iuH>dNS!)Z;z^Kk&&UP@>
zDYFvXR%tA=fD9TiLZ3up<xD#wzjxnQ@#-YAG#^)rG`PV{fveCFCi{!issloK(Hi4~
zrw%j~_x1Ic<9p*kmu9K=8xZkC>w@}v_uN^U$B!2lDqGvyGGY0tO#+^$!E^Tw&Xa&o
zB~eLibmG%hW&%=DIE8<Nj^9{n-MC5>5+*xF(#u6qAWfOu#SuMWxh(M}6py@8Eo|$l
z9Fl9Uw4K+w6O4{~T@Fcm9ifmT0IE@-B+e&gpM`z?ELhTnkjv<ZSve@AQ&KnTiyyAu
ztA=ImQGx2^`D7@7y6S(EE>)D7z%~^Xqj(k~bC*c!C4xv`2(AOCWqo(`^R?hh#rVWv
z3><-BpAW6SDnJC7Hq$jQfS1K6*NVwP<TeGKuQz)taAPsy#i=Dlbklr2ICf(XYn4d>
zpJ^TK{1n1}#6HHBP@LgSL#rha>6A$HG1gILB5(ft2MrTVO%fWK^demJ`^ohnUE$iB
z`0(Ka<2oJ5*6^%vTvHiWd?{QewLGQ!SHYI{y<5Q}W$t*OeyYHJ9cZ=;<0zQxTp=)E
zHRLn=Hs^v^VP!**xs{O+>08n|!RKK#QP=OYRR8KH?*MkgsP6X7r#_rx`zAlJ%v@ax
zqHU=W`!yisO}nYZ%}h)jlzYBJ;$#s#6S%b*PyPTu1_T6vKcji{ye>O?E#vBf3;Um0
zN9Ax{dB|H3&U$N&GCN8o+D`fJD!@DOR^+J=qL$=z7YO^cdAGwgX$S;@SilpWf4iPp
zU)y9sZ0L~|I3F*$&c0y^FmpwI;GsT5=++miZPF=oZ7<lv&->rTNRJDiCJ0Mq1$Otz
zg9io$MLZdbIbf@S08>05jow|Ie^m?&>ch#Z3Ny*QlNWo)Tnmy-cW-hzV+IFq$9aHK
z1Y#oKQkua@^V@(rcI$zzt_^UE-HO&W+w!B+3b$984kQm%5HxCqF|4@fCRb^cq!8V|
zP~`~CP9e`>WkddKLJ`vO_s{OnRAKP-%gg2<2WorW21LNcEXq8eBki`vUc1nsC54`l
z78mOPX1DRb4YULrs6?nHo%i1*q-2@B(VTE9ub)Fhw>Xm=)%3dTuwtY*o+!(h|5(?(
zyKocUU^r<L3y2H{F=WZ!xJ0l&{g0J(yuY8;FU@zbxzSMZ*20qt%H1{QiObNS2r*TP
z?75yk<TQbd+9w+WjAQh=XlE~_6XEj0njvZUbuS<Z^DLOq&CHBOqp|Lds1y<c@OpkO
za@NpieGmYv%tHe2whj?YV0Z~qY+jfD=pt!q6<jwWTGE9A?ey8E1=t_U-%A@6MhT!g
zzU~iSJv@7v$oV@Xtm+X263zaCWD>!zqAl4-W#y3daFEk4X-plH6sd@RFk(m{{tA;&
zYX`_Wu;wFf9iNX$^h%0u-2wp2#-!T=&IZ?g?ojVBDz3l>ebfB<#(1x(HIb_x@gdc{
z{3-tj24ugir*v(!&oOr!9|{+JF5eX%=CajK&b6E;J}qo7a@;Gzr0iKpLSQAFUZ9v?
zJ<xCSqp<NBJmm?eaQ>R<=YFtonBx3~WAa0r-xF$~N>V&Qu^io>`<?)g%t=uEYcJqL
z^flX_!mtFhG6FGiH5;-X{_*MCWcMCYbB{X7P$E{nz9Kx~Mn?V86H$>p%)_grX~<Ao
zIz&Q*gsj@Rt~L^o+SKM6IJp^>y~Rc>h-@2HP3+#p`fJzP85*jiUsj!zi8;6q6c$Pn
z7ARCo!~jbgdSq`rIy&n5hD^pt0P}UAaK>VwOpfHW^qn?4&SV)3gvkHL+FM3d74>_g
zTa=WN?k?%>RvPK<?vj*7DFNv&X^`&jmhLW*MmnYM#OIuI$9v!N@j8ZsPi*$uYwbDz
z^OwVas$leuGU}>si82m>wGc$yz2<Wcvz)bB$}8U9-p&wQYWMY>uRtc#?>tH$N%dVN
zC@C6@8R(-bK$4P`N|1DSzD0&@p5h=_7D^5{19Owb^eafmI4V6oG}Px=BQfcTJ!e>0
z*kMjo(X{|_g0%t=I>%>2V5Z;95Ko8jrwN=~6BA;3c-`=5q!%+QbMxdIJ#lex5CQGG
zT(DSg@7p<i@SH5pR|{f*s;P|U;4wNTKmtJWf;%hEi(mF;P1Aj$G6mjs$&ryL^x7{N
z8KV+ri7<c>N-G45G$5)872}fdLoEl$2~4+*XwWGWCIQMnA;+kIfB=x$K)d?>)azcY
zM5EKFk|wmt?fAO}{kCHxJytPLJ(!3P@p)mBq7A0$zJGwFvuw=b19|QWG$os{z*_>e
z=kZ#a!5Hz&bZ=hx;fFOQ*iXgS{;kyo7JZ+Cc@|I>ClyPjsjaOpmqvgrK^7gCkf1iY
zckvoPl{6Sae=CcM=508oVv6Lw%he=fOjwg=Zj8lpCIs=9%DywN|1Ik1=s=2?y`t))
zUTX=vvfu=X62PT6JtE0z#>xa`432Iv2+*ESu2mJncD?9G$hjI%a|zOniBlFCz8}@a
z0^L*;x2Opg78ZmeiG7Qexhy82aAc`OWb3Pv%z9Nn5=onO1-yfo<&Nw4_k($zCI@!y
zXRR2~Lz9hbNcl_9T*WCUxXLT&+E%}Fu^sLib+*K1hx1)f6bnnCk1G4oUuc9A?74@>
z$M#e99VhGEuMrSbgp8e?ofjgWr@Emb<tLfRm<s=3JLs_&_QVFvq@?Rj#ZDEqMb&z5
zMB!ACk{rA+%WKyP>yOXh=UEdA+Rl{`f~M&PoK_7+`?LgM8L4awXT#&qaqv4+gdjJj
zd@K!mNd5cq<U^#;BQSJ=I|yX_;2-ELRXM_aZc*}HWMX#T4=gzVH?S1gh~2QvmcWn+
zlp|p4x&cJV@#5*XOp6SAZRW@Nz`2Aq>q_UE4hMkXgVhuc4(`*{L2{k32wYYXOqh7P
z6@a{eVnE>L$~JgH2*f^NLqe_RtgHS&Ui37$`p7Nd3ZwomYsGNAA9HQdAn~ih=xG#R
zOiwjM0akaEr79#Wh8KRtq_4^FIf0QP<M0zB7xD10feX!WO6<>SP0#xuQ_<uA{#3h@
z;RCD$u-NG6>Qc+4S(unyj2u2^vC_ZA#9|Ema{n7ds}S<#4{ADYo#{`qk|Tuqbq0Jx
zl>F)sd%fd!aMHbQ-2<ph^OaT<C#)|On+GKexY(oWfBO<dv*y8}7obM_bLCIxV?xy$
z8B8zyQAKA^BY^$_kd2^a92EM1@!^FqJr=PiHg@eJ%H!aSaoP=p#Bz(S^v~l~#FSQ@
zpK&nhk$}x(4lwqbqJY?8(e5B&b4dCNB>djKKA%S8D0w<=FVszhUkeb(r&{1(jbGHC
z15&Gk<yK$Ei?QMcH6VafHBAJYqJJVb#4q{ArBReD+RG_1OsrS1^7Kuh82gtbE7Q{w
zk@QlXDH-&BkO^=IyD=99eDWtY@ULEo_Zb|6Jyv!79S|1s{T*qJ&FN=?)>75dMIjCF
zc|}E_Ukj7~(QZc!^|i+hUcEq&118+p0D@*=TOqzqfr&CPqPjX`aO(%Wx}BYZ?`^Sp
zfe+E$$9ym_^6`5B^hF^GC6Z_sv~dm*nvbN^)=Isrzt{Qdfy9dYKFxBb==qYsB5v40
zVu1g-V6OA}wCLdX))urmxiD>D!*x0$QXH2q>f0ZhmdV{{%1$PZ=2RQ9NWswjGM+>L
z(e(kyui8|Yf)L32YVd-WmLX-RbunxU5C0H|0;T|492|%ZXU(6At%cVdFd?X-x3BFB
zy;SF0VJjRtAb#O1w;3<pkBD@V4>n^{^H~SjqTjF9mfolj5=q5?R?-f06kto18A=4f
zJjo<q;j@N@N;RmYs=(g`z7k+i(Ctc&1)BQfiJagjeJo&vfc-A;`HGN~uDY30Jzf{o
zP}l8jrs=c~_VrPb&`46C;o>C-rC;X#aQ2e+DA9m|2qmX7R@ZV)1{Zcg6o75b-&mY*
z<(4nFJGsf{A(EZ5RJ9n3-xGm>pOmF@Z9*?xge)NHvCFbjp93hA=-ABtvq)P!cs?E{
z#C2usd$nk3Xm3j8zPV!bmUvb4c79O2U<#0~uB{PAhNhJk6Ck9N8q*C7L=QJR8TLuq
z!qS6TnYe_!_xc<#&=mq)z8}Tw0<vW-YgAolA7pv-`Qk>EhH5mj6u3d&*URJ8>&jY9
zFfAnZ@_HZ?^wNHwGK%l8tmI~>Lj+ICB}gYKbK3dLQ!mrSmnG}b;Ko&!^;v^4^77B-
z=2mH$0-b)~okKuSP??i!APf;nY@Kf9=I%reaR%7`@^ek@m35#XZG8u8H+6N0tpxnt
zf&|hAU0L8E1V28*OW^7@(*9so7j=0hf`f<m_$y6GNJwZseTmI%*Z_N{QCrgi2?_G+
zgWcmu{V|p6Zf?G0tzV~s%7KrFrRCcsrx%djZ`Ps{{ZA&oJJ{bMWrh<6G$|VASM5Se
zYad<bGbb!$sg!fV=Pt0r#dBz+-SVyeVDP5Pp$2$7MX2Jw5*fE(zt<qumEs1aGt-~g
zvBD_PLTK+*ft3Zcb^ihma}aU^$VKlrAM;J8gZmJJb5=iH6J)Mh&}jxraD$WT_0F>U
zcr#Or2S;<bzLZo*Qnn~-mD~3AOJTy7vsj)>w=ncsg6%wj`<LKugOT8lqrqPv|I?0`
z;!jS%j(2G7Vm}jF1`5?ofp)C;J=*S!{ZnIgXxEp@Dzq;Y83@u?`b~?D8kyzCzOl+c
zpbl^nd!XS0(ix@hndqCF8(_==0TX~(o|Lu@2K0~M?Z!2js+(T@_4M@Y52f6QnIe<v
zg)ol~JBJgFMNO||AqAj{rUY^jY~k>KI*E4=gLDW}d0)(RYT3LDYz#apn27t_S&&Kz
z9Gk>kL@<Z}HPfUwJU~zQ_w;9bKIjy1@e{h(gB3OGH;)9SJG!*BAEpfP<TruUlhpwT
zf$-rv*5*@l)0<0oyDv-D6`;T+ci<sYIt0A5<{sSbkL#80+j1o3froG|x2uKO3bQ0<
z3qEF;;zLm1ym`_6GAo`75$6*#j~GC?8FK-E@h({I*Bn-pYr9P!2&QP=9?jE<)?p!~
z^meogQpmFC;%wOdtz%mC@2dRkRR&o=Wa|2IBZub28%E@R+YV;J7JVcLJBHx_D=r2!
zB_$%vJ10_*O6J3j4cXjGN=dOg(vfaKjgOF%lZlg*@+(wZUYJdA*6~LIqY(nw@g*!R
z3+a#&=4xR5_#^WX>zlUhfS<_MeYwF&p6H$3(RQUVyH@x5;i16&bfNbUmPp|zA8jG)
zKaOZ5AC}@%+eyf^MHTsZALKpp^^NI&O2Cok(w3~#^ryjZaLraw!e>j8!4G1V>DJ2)
z^ns`x0_6T*Vfc(GQHhED2F$XuKD$`r&{o``n4y`|!?A%-5DU|u)op{J9Q6?uOzZ!A
zz7;|5;7is!@FG}M&#O+!;TpYCkP<22AO=&onhOhq_2t8zFpTPXa+~{?YV^42X;{j%
zRFHfP&EeX*%ynA~@Jy`1qeqNYmZg>Vaz_2!0y&!uV|p$q1UZLaYxX343>`f3B}G13
z){G$69iDI^h{9zdbOyYH2-C**_4H_HYA)=UvgipyfBCezd+X+<^ye235FE_8Ixmvu
z|M)fi?Hx3D+XI}-+Xe*%D}!LH*@J2c5Ec>X1mVGU+`L>r2ilf8Fd-BpS{^lfyIF@F
zH8fJRgvoJNN-A2e_@)uH(4*oaaHJRH%<F!jd2e}ld8I`2oZO9xC7{=l&-3FS*F|LM
z)H19@SxKp}#DQ=r$Jka|l!(eYa#8SUUo)>mLf(rerPzdfDJW~CQAV>iUzgT}Au2*r
ztDF+ew`m|&l}VliH+J;-b^hQtR4@$+6nEF!pUU0Ju0|!}6{q11#)pgcHuK0DjZBml
zCx;23g@u$yF|nzihjzF^85e?e05NiV{jE@;CFFJ>0hLk>Zm)F+Ui@;@>3SJ-D!+&<
z{|+d`!N6r|XGesC^LMLy!f4>6C+<~9oCs=|ZArgYlB$Ov9BDvNl@(VPX|$wxBvGrM
z2_p-O8!=}KH^a#KYIs!{zOJqnTn4`vNO%Bs0zl4@dnh^c7U(N)#x~SX4!(p1=u;8A
z`SQE$7#*ELi74HtDCngbK6usXvPDV)Adulow-Tnax0p>knoIb23x`kh8SGfBfn;q8
z6{cqaP4Y+IzjJv4q52}Snr)e0E5ZZp7YTBUrN@%A+xa<F5fKZs(I2_fXo1;0OD9Fe
zvD8Zn%~@mts6{bFD-4))baav*81AmSV)$k9S;TV6fBuy9d1R;MVhSTGbFs7^HgPW7
zTR#8Y>v$}<OHLFC>qZRu^ioX36)U*=g`ZKNk)w2<rpOz}-_(z@r?!X=M|9w3jQntg
z4+fs3vn$&j=&@as0i-L3mwxB$3DxSwDXW54cW&md#gZvSjq?kI=_hvg%i~MH2t{6|
zx58)7`BmCYK7H7Pb%SzpU8;~J&xluKxUvq+Pg#dJ&(pQ(jPog{n8*uiFIZAEJ$czI
zzd@Qej9zcNTP0u%MOhxZX0*1=DsP~j>W{6K8!GW~CV9b=R4@gUIpkGU{gegRgIiea
zi*gct_T+VS@1`&tKBc1lzyI+WSJm<tXB;398jQXjd614znyyG6U+3b8qf=g&uHzxV
z(z_5>U1}Zdwyg$xEKp*!00Uu8mW`qC9s7wFlcYtH*o<Gn*vxHM_Ze062bxN=HZ7#1
z$vZtE=KaKOq}6mt>m4K$2I7!3tUY^g-2aq6=k#<~zg9H)5Eq$?acS)_+tvaF*|oK*
z3bD7ARLiy4018du+;nhr13Y(alkt1-v~v_DI4JCSiX%6gaB(~IbX?VrTr?q33p^5@
zuM<L0-4143+WOb=xnMH{)oBtg;1K{EQc;&PmFbo2<t)8eYGk-@Msr@mKo@7z0XEIx
z;}w(@&yFnep`4ulaQuQJ$>&d6q-fURD_h2QJ^~`2U&D0bVK2z1Gd`)vA;BtAFfTU8
zqlV<V>?d9w=wD?@QUJ$^v&R<F6(|A$2i=N=csC|;Ju<?SBs97NW_Tl~4+VNbB?!b;
zAZgZ|N*4O%OUQ18*91UGrm^#{43UD-G`h21%{tJ0<{;4MRurGXL&hqXTC@Bk$N#Od
zx{uDsyWQg+Z}r{Iv3nB3oQ3m#o&t_|8pV6Ez`(o)nes>!d^|igtvV6KX}~F4M`rA@
zfBKo>`>4r7-KwZFx#w+EqS2Kz^;EWwV{rPe{Tg5ZQStB)1(hBXqhlwqn9?d<iWbv^
zHa}cf0{SN1Tg2U2zrxp`AZ5Xq0gqSG;dF31<yO&su*EI*`9|)2VQiIuxJ*=`JzS_R
zsQj4680Z0z)0Z$d8VxuQQLNF|aBx6NgeS1^z2ytfo+fLkBbPxS-|3BzPAV%XA^KaW
zY>$U0qPERsJ@<oB61)M`??oI0&HcB;I~Ta@TfeF}?VeU;3fpN^F$%^*G;7!1%Saim
zv?D>t+L7Ox!yuVJ1&b5=9B6=YcGJ~Gu#iFM{-6kHXd%3U8leQBRDy}wQ*<mWWQCh5
z_0V=c7f<P(V1_Ti1p)ZNm?%@U@o(LFIm3t(aQG7`i-DD~Eruuct=-7wDSkbd^sC_n
zWncW?MX;2)?+bsPEL$qjnAo}KReei|!}w;Qz@$!LL_zuO?3@RP6!XM>@9e;0k9L;g
zJ3QK$8&<}~#s<l$NGlr0C=XSI4<szFAiG)(a0*t{)b!ZA1Y7jJN{iW7d-~T+e+W6t
z1<-Lyu?gYxZSPI>Lq)5z3L`rM+e5mb1_Q|Vr~SLvO)y^PIG*X5ZKdH)QyUp%y#6x>
z0*}<?D8N0mk08*q{dqsNU<fj$0?f2Rc7`?|PX~Nohg!a>oSfzD=_@iqR|3mDsf!}R
zdS>WvwE5Czj_t`8kfi&o^grf3RC^MgG#d-#9yW-m0Y9{fnN9e(xlfv$Gn`MY*tCft
zs1Tyrgtu?s#$%x*4j~~TUR+<#DfF6tdB4aYxYwr8Lp9(X;hN)H@*aaU09ENVrhQ?O
z=UpXopiKi>dO~XK*}cin#eO93KYJom%&gR5E)T2_Nv)ef!t@~`5Hxiaq--wN>W-b9
ze#b3v?kcSX#=;uhyjws?HMPilQ74@2Ua)Zx@FuHRNzsFgxZt3rLaO0qi?GQG4nVfX
zk9~KB=bG+JQ&NyzhBAfjm@LvG1<W%`e4*Np`trkg99AXVjy5~1t+G^3-O1JTYQ1+#
zyAhyRl`1bc>}?TanEUv^>UT}()I6{sOe$k)w|uzTcF-^h!+o$WDm79;4{zVykz>>2
z&qG9%35v{<9hv^*MOY(HZ{XvR50x2SIUxtjhkRIUtZu`pO@pCkX4A9{LPbHsR0{<S
z^4V1ixWs~Z_W8du@HY2(uQ(r-pzwVSH_BOy<k)H2h!<<?N}&PQX=yr8Yw^a54#n9c
zz@G+fJcs;6W{uNWQM#=HM25|2JKvlZ0+W(V<oug%fD;u`b4U4iauXM|$;#I7<NNmk
zb=?5guoCH|i#?e<UazwzGv;))Yo}#BoEh~?DCH+~zAX|@b_gKv3t4iGYn+*IrN^Tv
zfPbLQ`)@74kK{sJ;9`4WP!hO6)rRzrjZMD@G2}gzxVV84h`g+^l$HJVJeEBF5lz}I
zsL$>U>TGu8Zl}c-zpl@AbotLj=OolH&2*%km~17oL_`CPG?c79`H;S-h%&G3N(i3P
zZx0zTab+=jH*HPxHfX8<$OLK$pO@IzMWafmRv-8<s{X{gkpJQ1>$cZ#Jkv2V`i9MA
zJD4~COF^N}NNX6=g9W?vRxQOgy3B3U(H%&DeS97h<pTo)KZJTL_+DSV?Dhqjhb8y(
zV9UW%YP3*^aduf_Fn)sE;X6^4c-qYxuiugozrM^Ph_BWv(aWLPH+DC!QrXUky4Ry;
z$%^(QIPeM&R}qqPp}`v&8zTlHMI>u&$qzT)k{zHsU1j-An0S0gwjyPPg$g`9UT^jU
zD=ig4S3t00t^}LJ%b5si%-%G<6E~B$Yg`>`L)+a4vhG(K8^L9LQl%-^NEZ6B4^IOK
z;0y^`?`&5Gl!c?hJe*V6Ehn`rdxS`NKG0U@PoeNFaWVgG<7u5CU7rV@K{R7!iz9dp
z=yut$wof9xge>52f{skd{O^Y>_uvnWzlC3yd2PGNReg1H75kAXIPtkk%R6sMvDt(u
z*eS2C(ox-#IjsXWHUxGq3K*JZfwO$1*#~_#pmX0PjEMjC&emqEA(y>yIAnlZd+OA)
zI^0T>0_}4pJTRg|{0kL+9pj6_Stb{Sw_CYgFEZ!sG}xstyMyFg)IKB+l=H2`)bgLP
z`c#XaC|n$wwN|u3Cvubi{ZN@8^ibBx32P+Dn)buni7vcqvgjY+AlBEHe<uGx4!nn*
z12^3$4E)p~ret57jaedj_~uzIG8c-~5P@I*vLPeO+U={SP>!mMjyW=EK#g24T|!C4
z{4l`Y{FsF#;2^@1p?UgCNwjD*+*Ge1oh%%*v^3d%Vi3dal@*WK<zIXyP+A4OcEr3v
z&+#y%q@B~#&HxG&dNdLwlZxqH_wuOU1>~sb_SkwV&JOo1zr!!T=m|Fr0gJ@)awdAX
z&$mYgW-zOzVy`UqmH2tyR;!U;hnDt<#+c0PmJ|gvHo7k^d<9?g>D|SFp+9_Pw9yn&
z1ye{!NF3OykF<28MGD0oMUf<MUr2BeZH~1gs@Br>?<>1RW?74{!UuaqD-U8B;A0F8
ze$qdA=I~q1O;QWK4r+}4GDOGOx!Fkg0Y0jq5P`>x<1+C|oW)a3$NA1Bs2=_WXicWZ
z3bSMf_*Jczo;)scu(Fa8S)#9oUPSf}pKyUuq=R>8Jt17jlU|GAxNPF#1yeOZeX)>O
zmR2|i{GumEgurL7WtCAsfjKk@JnBW~BPv@Ud@fh1@IBdtD@oJWP6H6X7)eZb@fJvv
zzL2~pi>`Y)A!h3C<?^Tz?+rWaim_O*2nQSQ{wePGOEVe-(#T9ru5Gl}d~i(H`C(`!
zeRF|jN;xnNpHHB4I$v|)LQ~8xV!3nbi}<)zQ_7ZkA|weS3^2)|D@FSW8VHCgjhwM%
zO&<Q*NdgZAIrtW?x3lv{eSJwPlzU@U;e~<_)3u8pW6eoGP`I3scNH&n|L{R`>JOup
z3^WX}^}BNp3aQ_z=Nt%uUnr!!oCd{WOgMa<EZA!qUS`j!$Z*50%EPYSUmVNFY_?Nf
z!{^|8+bYp_UhZ2o`3!a1<hWTfh|gxdReD@F|0YzC?%A`-5R8vQ$)8i7(N{c7MdZwz
zzO&?d`9ST$(AX~I-e$E}BF%pjxtNsR%+3Z2fg0vM{Jr{3Mg|&^N%M1TDhL1F^J%Ct
zlwp;sk+*VTT`#iU?Zyh}g=$J_8tE4AJkFFKnpw+`X)~|GrWmeNG8&sFY+2{dO;CP8
zuM&m}XgUJaBsg%L-gt3*m&s8x-zpm%eBoD#Wj18%j?m}@ti%$m)xsq6{I0`(FJ96*
z4DT~D)O9&%m7sAmsKKXU6~B@T>$R*jRgvX>Z_k+Ckny;_TH125^%<!gMyHv%Q_;5f
zX+C&EhK|i?)@%o=!$ijJYxqomzl^jrFtw%Yb-tywl79{^_q{u;2YKfp^A?B;)WRjo
zypq6d3LlL2?*}XufvHN#denArdxY2ZItr4=2I733am{TN(gPZ(_V&EC1ENaDO-aBJ
zeMtKPr4jYLmV=8JlCPyn|4Y;7wxn!DDj95!TOBM0*4JZYgs$deuE(poOK-p!G08(a
z>)_DhgItO?lxb>ahJ}qiSE3pFW4x#cXi&eSQ+4GQk|f&W32>fXfZ~eMRk*i)jeDAW
zj^{6ased<(iVyUNEZ0K_{O$M=e*W|YWU<!w_JW0q0Nk3k+<lrrvU4R#ri{!mw5ZjT
zKO0<j_Hlx8PYADHYyV!eU+c*L&pjQd#2=vI{B`(}LpL)4pWX88_R<>D?OQ~I&+Ww?
z7l=!HYo#T98?o#KWsqE28)1$Q5sopTMmLs}dJ+1CE)f{)iVnYmv1^YE6CL?9+*HJ^
zwR^kP94hBBo4m6oy!~LR{%W(gx0flBXmPIkya~B&#UvYalc@h9iHL|ep@M<<Ws`8u
zuCD6Ew?2!y@arm_{&qf#F{PYMFgRE!Hxxemz0$DlzgrL$I=U&<s-^vPuy-|tU3%eg
zJX@OjJQ9%RhJjl|6Dng+Pc)Y#0}C-YTXHLFKXHvAIJhkQ_KqwXhfV0>1v`fUrpgbo
zNFcP{k0AWwn~V%;k3<A^v4?F&kRjDP$!0i90BQL2dmiB9AiC;e%ZxQNB`L<IsN$<Q
zs-bHO#Aym4pa9cL&=(NKixE&@zLit14C-E9ik#sxc~`vkBD@r1fD1GdEHgKJ|IV=4
zx#d_06YksmG~f5t`<Kgb1P4ByYZ{=%Iy!`B>eXp5PAdPrPS>-`5$6&6AT>fW#L0Q)
z{&M7kYDpN}LY1uj-?lDh1Ph`l@PxCRRKc4Z2fTX<8uHbH2P=6*KzsMBj{`W(st_Lz
zKECfQhL|D{bhNl0rc_ruDJY;O4yn#bfrt;7R2;&^O3Zoj!_{`%>!ELBoKecto73Mm
zprM92qvSy9)dNr?i;*b`3kw6X{~!0;u3B;m+Im`1`T_qzE#Lvw4YmIFO?`19Ks6WY
z?e#gzSa1KH8fPi&Zf#vIoXN#!*cxmwQC6VzOMn$>E|2b>7e#i&Ky^1f3K~$uU4jJq
z{@wr&6C{`q&XK)L$;<8mtWl;m5Zl04RjrQdazE~cfzRD{?hQa<A_#cUK&kXmHCR8-
zFE1Zhmy#~*>ncO+{f+S@-Es>G{Qb!_&i#77ToDpK!z|zlye-8WQBoZ3?tXc+9JIu>
zz7z6m6VX3d(C1#`as|{#EP3tM3?%c|s+Yo}ps1TBR@rRyv@Zm;<Cal=b%q0wQ0{vg
zMwmdTqZP~X%+Jq5;2zfuT11s5?TTP}Ud)?Rl{~e=XU)4T9ydkLJ1+u=u-uiK%LSl#
ziHDO>4ta%7%7N2<ZCOuFaYK0$9EXGsS(b!(0bi}J*IIO!@Zm6lX>q>r9Y`^*0TZEQ
zyK0Um?O-&;bAT5RaDzhK2hprdfhC94P|Tv^QM2l-A8@kVByW7H&)$iEZGCv^nO&N<
zv0nlZZoiR)BsrM+Ye06Rt;A>4BC?`6`gr@Qii&wacSF0TfqKEmjQ)2FTwuDeeB`Ip
zR)SPdvPEKHph2t;J|)?b09Xi_mQzOTe~tl@ZLOtN>|8QiAV3)4>KxKixpUWGI*VOQ
z6uglni|*;|<>BR}<$;D~rWUX^w-bC(F&KgUlVeJOS^5_}CS_Y%9!M(Uv|qjd-TNA-
z)Bt@yVPSQbKF8PaylBga!KdYYsa6^M!7G#Z*T-URu?|b8EHE#gaxPo!q?CU=@=XMY
zV<fUw*)e(go|7~UR1oZLWK^-51#*O2@FSDM2URxY)3aK13oy5OAkbjQ@9H|&(n*=E
z&TggX&;eCcQok_C0gmZ`$gY`+)y#P3s+ep_?-0>qVQ@-B=r)Dz_gZpQ66BYiX>v?V
zcCTcDxtaao=u5vF{-FtvDmjGEkPyHmeKv(Wz$~|VsxPgR1VptdeM|_g5H`>on)$Z$
zQ}>=0?{Dk9J9cEtcx1HdohxCzP^4FSEZ<Y#3PAfT=)S3H@)$0-LzpAYIaGJ%{KSX>
zWZZe@27$<DG&D9^@)3YFU%iNuAfQp*dNC_Xx4Z;jb8P|F=WZ~5hZWuUp&_rzWB^U<
zTF~<wJxp+!)l`1Cn$ceYYI_$oM#w8|ZP|Jy?I~v!cq~SpUEfps;pEGyFlrzaUqpr_
z4Y-IO6w8UpExxnEy@Q=!OJXxzajGVIJqLTEAsH41Z*1IGfBi^RTF*I`*6I%3X_<`C
zp+Xl1h&|h2f3ciF18b-Mn)y|s!o({rYKBot^T-6b?B9A$RB>AU<Sw%6dP7P>6q`m>
z@Iw)6yL~~+nvVbi==cANg(nWo@Yw&7`ANqyB`y1Yggc9qH~RUB6V(FyYW^}UR)myX
ztda_L#Hn~2QXw%W6Ikt{!~ys8Mm``LK!HY~XB=k74i!1*w{dpw#l%*?{UD3B;6nv3
zLE_u$zL{OTRWdFIZ&bG8Q7$A?Z`_sT0|47!xzk30*|<v%8RQvx>I;3q{PDnPv+sS?
zHJ<ej?E5u9)B>mK7TR@<LS8+n$*3Cbd@x73<8)!RbwyhXNoD++D0$aUy*0FX1XI#2
z=apm6u$iy?3aWGa2aZM!2Bb{C3Fpll7K?EbTiC#7`dRg_jlIQua=k)9z1xDD5bg2A
z(Tw(J>)<QCvGr<mx1l9wC;4IY#R}-!F-B{=%RS{*og}S@Hk~y&ZnCFENAj((=}>07
zanGcB?Ypas{pOmUL{?MQr+`hH<mLEY{_O*UqPahIsKUE5^N8S`<Er$1R(#;(d%WsL
z>;DlhBrofH3Ox`W1=KIINAtg^xX~}Bm?XUAMs=`1c?I1(<`Lp@<$GOC%j*!{pFbQ1
zyh}<<jy~;u&pkOUB|4$AR68T7;vq$%{)!@(LOQn&@mA3sRz^U!mz!X;=ho$Exq*nj
zbg`$;xj%0JtXHpp9DMbcUt~fco*za{DMJ(lflZK{t*YikI9eeVPDR;)bEmM!KbDH4
z%D2f5H8{AFzt8(k?2&`T)LMKcC<jJI^*}t+$ql-l%rPjQ1Emp`M>=^J9PBKVRVp^p
zIfRZ?ZU(!5T8qH?1N&ZLXJ?1YzAdU#K|-Dj-1fC#+P7#aTG}iS2%Iq=pE6qJW^aPw
zhCGY*UykA-7S6}^=1b}Cla#9Nwe#Qw#?~w3E{}QMYgO5;@Q$QQ$|<5NO&-4%3!?Zq
z9-_bIcX|3YQ1saj1QZ85TU(Zlkjo+<lK@p;%+Gr-1L+}NW0;*@cs!IJ_wDS14YVZD
z(TcQO2zTve{f)7<)=$BQ!Sc<w>Dj-rPkq(H7}EHq(6Cl-X5;pbU3_cjmzQ<i3Ys9l
zc7Cp5+|(`GVj%PutCs^Fn2?Z=&J_#6z)Z{Rz2O^w5?8)6?Fu9zZ&xsF1#_Dj7{7iA
z+lsJeZ)~hwS^d053u^du%IG$^T(_;oWTY}~@ZvzaTa8oy-Ux+zcQHvayftg}uH5L0
ziY9Qq_zcY_kmz$dEibdEwY!|@X9eHiKV!6Y^gDfmxhPSnd?*@QglNS1woe4OwNWHm
z4^@k|XU&<k-43p{U_eMCGxG_;7}nXD+4RR$J~un)(JC79JA-F%V>DuJ_3zWsB1nBC
zkRJ2<7z=?Q0&Ol?-#1+4ti&$>ir^(phdK0KT-TLF;GSN8ZH%TNKC8w@T=(_cD5qng
z`cwn+t$RYF<;ZOR)2Rn6qgVY{2J1wl%}bRJ+;C3ivo#{5%}tMbby_7Shfc}Kqvcp&
zl350-%2#clCW>UnSU_ccV<*r&l(AbOtaw6o+y<Ugy?#Z2ywfaK$-;L%mc|d!FzkQL
z%=K`f3FH5w_o=RmhK6~2;{*nvX_%GtI`~=;hy(sCGz3EK<8S5#fm9SNfo$)NuF}9>
z_3x7cw8)C%`*KU|K5Gmt#Y5RQTvWp-Y=5fB8tTf%x(ry8fdE&v>az+h^lbYj7X9pX
znY0T<yMv|~GUSz}=C}2fc>E6gM^Frr@iHMnmom?r+Wx6*gZvrp?L#TWDi+t$WKbhP
zZ9v3<EA=Mja(k5yEZ^b#1r4h7h?3ILG(R$iKg`h)HUWH>hMpcMkfx*Su%0*0|MNe6
z8J4#AOqaaac3za8pa?c4Nh`-(zy0zm&4cCIqbdP1_08~}KSOE;A?t2)RXBF`q@h$H
z8uZLWk3}t^{7oN-F9Z5x-kATa##u&Pq-~Y-yJc?kEmyJzHG0kzNp+`*-fumo+I8VU
zu@9iJ>MTyhilfVAVx)s9@)Kw<!jX_J><l@<g-Oz>q~$4LBzD*dl(8ftLpnC=k9B<4
z4=H%&i*3#5)62Hunb=2aFvsEtpwN`7O{!Apx1&?!2n1|S#A0$6*A8j@N#MME+k(V%
z)`eH*q9~+LUp;Qeg;7Z1g{!5n5h#^o%}f%gniqb$tGo@ev8d(Gtc)kS+3u^2#D|$v
z-m#DLIQOp_3c^Ft5YJUW43Upuz|@}Hqqd%@G2Gic<6cv`TAy>p;4M1ff?dmSjMr7J
zB@B${S|gx587b_5EiiHk2Z_a)^Tm_TBgyDDa_wExtnS6W&8U_jU1~yvKbHBULIOsW
z8sQRbtgJE8dHbbyvC7jwHZOt!p%ugbyJzQR-f9vZQn>ik)K9t>VZAW}CVwB}!m?@T
z@gDH-!o?MDZ@EOK2HPH0ceo08skJk1UAx{m`RI#=AIY6!uO)xFl|P~PRJh%|_x|%)
zuWrU7iH4iuJT0>%{YQFm*5`j`n}Cafdi02{=?3<)d46&h()Cz-&fK@<D%9IkY7s~a
zjd$m-;r)=jQaStZbY1}n39x-_=?<SFG&kY9US;3Sca9JgZ(-j7C|J#88`44WJ-$7;
zs8kXY0Tfoh2hs{qp$&?$*>Oq0_ayg^N(*)NKcp>Zf4*+3O@aL_rS1#no@M7CO!Z4b
za5&(*9XLltp7S&t2MZcIb`~-yu>e9sZ;;p!Q4{|TAg3WpUYGXF@egw?Ye$l*pA>9W
z81PD0R;;_fecxzuaonl2gb3qs?FtMh>i1f6qftzD^r*0i-QZNJFZtwj&Y8Q4i+2vr
zKp+I8fDHptx0%tXE|IcS_(0vp^PP|eqZMeu>TL1l;K3gwPE1H}z1UUcOeI7`%^V!m
zj~P<F&yW|uJy6pdp*7rQbN#uKPB`cd-UNTy{})%|h<~{k%HQavASzv^qu1|O+dmN@
z<+D>lQ25G(HFJuTe5I>Pl#OLSn_B*R+QvfHi7@1^_0o1>bTo*kSwuzv6Bo*cE&Dy6
z90=<3^4g1ev}4S3yjlGf4D1+Be{~~_DosWi^^bfUpTe;#s~DW|KNotNeLKClIjVm*
zm*!5#Zp~w8sepfWbp@*8HlT}VTk_2Xe8_%XBWbeCXQ#miPk+9qPGsOAu%DTZAKk0B
z`62<$hKA|JX36R0qincX+|Ts;%#!nnA*DG<^-xnV593nV$SUo<QFkV><7TSQQ*J(a
zwK-A>2Rc1{^Y}nVsW!)(Tstx;w9O~iEL9qZPk^DJ?fW%+8VMi#qJ#e!<wIpEM7U=l
zGW9KesVVu86!$ICIp2ksG;t2D?PKBj*b!u^_?&oF-`SvXAjFUyg|dz)3J>=sg5_wd
z{}|LUK@tH+&QgZk-3aY*C$3^A?BN_iQhaN}Va3Ij4Bfq>%7`+BVk)e&KHl;E%D6Nh
zF}vETuSCyaM1pD!JV^L~f4dhMKnA{9n<n;SCMv%`A4BcX$vGV(BPYj$*Mf&Gpqr|_
zZk{WD!1)67ZQUD$0?dADRlEMN*84RL#>=<vz@jpPNezZ2I>+ya{Q)_^yg(xmZf}GL
z`FIoPr9KxqqK#=LtsazP5zg!8VgT6^$Xyvd6hWU{ZKJUuf7Dp+v{@{zMhe$e>eji{
zFOHKZ#t`x(v8-NBHvK)Gliqzw{E6oo4-YHm8pjs$U`dmQ@;f=@Z{KS+S{k9f?u&ZV
zPD>_K)Ob(*H9f)=%#rrTCz$MbAFivBAW0x2V(+N6vvzhSC%6$78B~NENJNOoY=Y1|
z>h@0}ITJ<lTR^^1nC>v26!}C2;iT8S-Tteu)R?^#XGt#pE$W9C0Zi-*lGb|X$-*GM
z$E4p5zPEk_8G_n+b2sutxyeaioz5|Sk&Fys-v4|L!JJciW443W?0F*Sb#6Lm`&-SC
zN#cbn%ldqV{$F!BB1lP1t)04nmR*|M*sD3PKOXW4^0-{Uq9a2{;IhH_&FU*6X~1|p
z7QF_^!=?JnT@O0F_-tvcTPjTlB1E!`fp2enApO+xnut>7sEs&t-!i#WqIQYhd@yGF
zVu#71mX`JR)8``V!WujCDx=xSz9C;L8`6j;^M|v7?^#GbziOKH?LtoGE@UA|76XKx
z98<2`@ww&cBZb$;9le;3jmw~}1RPGlc;SWE=ceP1D;S21)0&uiVPRkg5uQ!{cyC~$
zQdhW?JCMx8I&vthc0Xv8ES1QRi3^Fk|4e6Ar8!HaFDQxK*A$Mlx7TAjC!M0(QT|u>
zWtLUJEBD#f!y$Fzw)3_qOXc#BpYG>S`-SIAzP(sQH+Z1N0~M~#Z6a8sGFMwi>9fvi
zXAjMCwEW%oE!WVnFyLgpFtViMOl8185c1wMMv?&Osvu8KD}9KchaloT*;|Iv2nO>u
zzy<x*qSfy<tCotPXJnjx#+@moz?n8(NR1tDWLd{Dxfk2n>XcLSKEypt&OG*2@oDZ_
z{wGITk1S=5b%mc4z=foZ_9}cdxA?Ls_cDN7o(fbl;4nou;a8AZ@!}PFO)${X;tBcO
zQzk>5oy#%k{My)1&WfTNON58Lh5H{81#l_j*8%TzRhmnb7O4B0mvC_LQZ(jIhQkpN
zd9e6XQy3%;zhPK%w+)giD&)jW{iX!^6r!UZb%7;_AISRG+aJfpqW=<a4FHhLMiM}1
zkv`-t#>f9^nsc$^-q6X;<ZjR8-szq7<3x{K(-=a@sOHi<JzR0Yhh`>cIF}0<_anLD
zIlEbN3PyxfRabW=pg=lI^srUeE)$<$j8cVQ{b##dPUc|+GBONAIE4deaq92I-e%v5
z0Odz1#}Swx?$*_g0KbTVXcO9`R<wJ%HvtEF*Y?`_)@yDr2d|Bwl&`4In5r9XghNvO
zg(0DC%jxf-fSS1ZI|-~3&;4~v2lr!!*jvUFt0q03QMgbw(=AAcD1NaWK`a767I18G
zi6%1r9GG9Jx&he_GlZFapiUF;WAvn?B1@m2hw(ub-Ih|l@eo2j(K*?@C#pn%$Z?)e
z|BhIIMZ+#3Hp+oLXKmXCVpB(HUTEg&W#^bJw~|+P$v{aA$Wr<6&#EcL{ShDd-SM<M
zIxmv@&1KKe&!=30%p;(PD!!lEV)lb)B#!lw@%Ge0U|i}F&csUJL-xQ=b_N+L{O@s8
zN($d?x>a4*>d|cCKLR67QZ+;pt)=DK_JI%`8BAn&*;ZYOw>GNkPY&>@7FA_!{e9oH
zbve`Z=X#~>CAfMpIKYFcO6y`Ue$9QBd^;}m4jp(HzxEnfz;&U$^gk}s2ViP|O78D_
z`H8LY^)h^W8~g*=oW^egkJlaMS)QwV){QjK%mN%6l?o=ZF|cWb^n0omBF&+o1qt!o
zH_$fKu(t{X!dN{69iplUq17ty8*hTHt`1N(AhJFme572<%H+R3Xk{cLEx7l;enrt?
z=DE~FpMT~L#3T%QEuZfHo&pdJ@2lbvzkmR`-1i2a^_{r(ry||vx_{Hqg(`+v>!(vP
zP$|+M8CO|*eS-e`ccQO;#GU;CiEsJSIC(~WV4Q%~)1OY0&L?|K7TU5rquIJ<)5s{K
z1b&h4m)&2@Q?Vp&AJ(&)tqZ~C@C}b6G&U{Dx?3N(Pyo6tDCi=mfRU1n1UYwC)8+14
z<5vMT+V=%^<Fz{>Bfe?B8~~%sfm6?kgNr*;s;$1H2p$Y2aoYF@T(<4BLLRt4;}Uw&
zckkL8hr(A1+`Dh5vxZjhYlT|He3y3mN5!zTYw?(WE+GTak+cq($qNNBMROEAxjHKv
zH9rQ3>iu?CG>W#z+p+*+o(HP>fq4&lHwY@Vj*!>3yX8oPfAieV*Y6okWXxlp_YFOH
ze=77Y+ZWDsdbMeDr;=XYdF-r8MYS#9R#CT^YP|o!q+P39Yd8=?`Nw*mXn)Gq*Iz5+
zQQcjc_Ip}N76iw_jJJ1xX-WwL%z}1A>?e@AHy4M#z=iffJRCs!55d?sefw)q3#}o1
z*!<XiAN=-2_<~@xyw6nrxWR87F}k-Hj-lg95+5#Jh&p%=CQX&(3o=fG<sW@u(*6-Z
z>*c)b{Egyce?_G)%YNhL!P>`nP$b*}d+iKtrKxw*E;h4ya%a&~ER!v4u|dZ)g@>Ju
zyr#Rv$0l6`&D4Q9oR?O#i>K#<{cCy`!o6F2&?r~Bthamh{+~^S=+H)w+92f3+`GH9
z`k<wKw9E_g3%61%8|84$Y3r<0s!+^P%&D%^qfL5BNixR9x}!-=11>**^m_oue`^6w
zHb2_B5x|dAg}jgX{@u>nddS3QLmUSzMj3i8>KT#rlw;zYL>0~o8&tK&tDcYaJVO1s
zC$XV$v`-o5D!&q<2}?0Y$&nde{knKOkDBUd*dt&gf2S;?Oa+lVlAv%)Z?1I`NKx<G
z%Ti~BW%+5=ZGLsi8fiVGn<E<ETaS*sP)#VZWGBeywstaFn?gKDmv=gs!M-9SYK|TO
zIwps8?v}p7xIB5_&*huzL!y+XcXMk@LgK~@4qR%~y|o6x8kIbhs83IyvY5R9C^B|j
z4kaoO4ehpw_(SpK?~U?OZIG!__Kb~7RueVN-2b%?dCv-(FYn6i6vz&!uoyvl^M>-b
za-9i8?Q7=+jE|yNGZr294dIW<=(lpj(~X|ho*_^5m?Cix#BfxxB7Xd48&TypH^;M(
zS5is{(9o|^Z6&?DF0{dUBWf0Dv991f3Odgr?<a4XC6e98_8QBgW&!bFMBJ`*hsQRj
z0;#z+;4yyXcZ0j&CLKWkW&Y5@{8C^_X@9m9pNOceQZl*-7^Prei*}^F4h#f!x<E^v
z_)jD{;%MZfL^oiPcE52AbjLsd=HcOiR;07B&?aEJBil50*F&cD2>)62rJK*5f5lyh
z8E)?r1G;=t?V&XCYaRvb%-{2$C2yEQ?`xjcVjv=e&9V{@0}t0q-J2!GF_!HeS2*5B
zOPST%w!|FKuFmTNwoLZ|*O!uIH|^ng&k&YwNcl{A{*fFFhz_v@kzot%C_C6kgCnqh
zGhV5ELG!kvOrQakndJq9nKMC>;vXDKi|^lQu~c_>ePB54dRES+B!EAOyV)Yd+Qqaw
ze4<xjKAP-SIBe#BSB0A&;eY3lB-eUn7sCB%uXg5Vxb_scI0K(=lPSWC`$0fry^<B;
zzH+dkqUVra^*djIK}$l{iUNFH6}0h;fmb9Adj6goK77O^G@-ZK;t&Ndp=sZqzTgET
z63YNlG7hWi?Y_cFju}@OO5h^X4TggJ8*fZlaUuL@p|?HMGBj49m~NzF(&6S4-XH@V
zXhML;3}|`pxmANl%w7j($a_tCgN2#A8%4DF%3))%?!^P}Bi?Osf<On59%@>>0Hkgp
z|FyV7llyu8xmgU@w~81rp~VQTip0D{OdSC(j)n$}rO&%$F%-$NW`JY|{&^>-^W()l
zqq+CMYld*3YN^t~K-O~M*>IThGpTDJmcnp7-Tp)J<EpayYi_~#x3S^Ua<K|3_{KTz
zv%C-bqKBaQY>H*WU=ky9s$M?%cKg>0#`@oXdM$K;rMYbG$0?|argHLOsP`i$C+~LK
zW@5I5feQp}n}VQIay9{2JB}18PZR%Ffd7(-*C-V4dt3@>{EP{!8~6NVk>=AG=x?-`
z#?jH~9KiMR0v!>6{Pa-)XAWSpeXMny06#?UkPVI@cD43A+@QR<Ev6d~<+%YxLsOo$
z`Gg1ymxE2wm+9&2_2oM5jBG5kmw((GtooD<3t6O&Tx!_z_Usx%g1EW4JDTe=6(C|F
z&T-Y%)pg{PC4lS^p0)<98c?h4^|iG#|E7k=oc*&4mx4q6_Tz-48Pkb-eDv+$<V(#r
zI@$$H>pZLr0$E>Q&Hpyg?)S?(bg1#`->}8yHMu_^foPyJ{{;QSOK&WnLo5^AIWvg)
z(ft+$EVDu6!@n=x1+i+FIL8Ncv7Cjx-`13rz@K8p?}(XMAeIyV^8e5Xl?qQ-#OGSA
z7yRFRvt-W4LVuq0poO4WLeL@d_hcR9&`Qb4&AQTqNux%*H{hL@vF5n$afSO~a)(9h
zt|YVV2GgCAvv;jV?(%cK5WR-O{ojv)qRLNi0gK1Cz8rWi(oS@=z(qqN`Of^A{Jr~c
z7;2^Te_$vfa?U*o^&&AX3~*Nev_6SdK8yanb-BXOKWOyqWc}UlJgx<M#fE;)*O(YI
zr1WS0Awv11yN8p#klI=lQfUG#dz})++H%9y$L?Pq`fH%J<}ot;8XBX!a4h%DhY@&4
zRJ%Z%O~1bpf*<WDn)~Pq&W|Or0@W2zrC_jalR4EnvA2FEhQ})de1z+Nr5x-bH#a_U
z%fJWr4xCO&b2n7d^OVf{q8Tot#tm(a9VPX;%VLoOFWJe3m2_lr@Yw``=*P$T_velI
zz=vVt5`CW|@0_eMc1X&SUz94y66}1iZGPR{WLTG_3#u|uiD7G(N$93>OH3PJYHXKW
zO?8%M#wmk>f~wU^#>X|E8G6BZY=fMI+Yf8oTn|q~==$)w19~+jBj&ilWCC)(rl%vs
zqg?j!R;Azg)7@=2P1c*+C10iu+>LjW)6vP$#>Y@t8f0!$%1OrLs<yA<7Xi(Wc+SXo
z0Y31}cXEd&{O}E{Oe?SP8c4ALof$qg-Z>LN0(U=hgBL^b0<xCo_~w=3)xab;0>KHX
znPw2?hG6O!{x)j-ObS5_3Zggs2oHULas#dEup=rrway{<xH^)~$HmAv3VKU>LeQ<g
z`H{elcw3i40X5I?@Guz}Sq<UWQPStKPp~0a@uBAW!hPwmyOmbiiHz5Ypp1m1NO8lF
zlWSSNg(&2aRc5`pGnP=Iy&Kgnx^X^#(uzBpugp0hPSI{b@`kwsl8^t;)6xdS$biO%
zy#j91kO2S9<|L~20i|h<@Nm`Xg9n!4JVzn{Q}*uzYIXKo1H;20Pmvi|;Njpp%Hn3!
zfL0IaoHreBfRYYWbv!faaH^Uq4VCY4IEa5Gx186j{wdSEI6ROXTq8(&=zj<Q)hL^S
zv5c3cNKHdySJ37!E87q>`z&y20k=~fc~JuxQNMou3Y?<y#JM#40Dl8jOc)`~?e9`&
zK`<YSabEy4Qldk{r7xbEipoDC{Kw#A4;(CHD(gJ%CSB-J!*kcZ_60->mTBhaUfUYr
zJ37l5*ll4TT*l97*1$dV<1mx<H}Fe6{^1g~)}ODoAN%<Rc<h)wpG7#qNkIm7s~&%5
zf&J1*Or|kAJKK)8S$Ied0+9-Nj`d!uFMqaWWEJi4iE~rnjxiAkd=x6>=F6!#U}Rh=
z_4YJxrRU#1{S*~<u^t?U@jxi^96U#_@k+O;^F9h_01^a}($l?ZxQL-7M2_3)y%u#^
zJ}EYf?9)FR79Q%?Jse|zpkM7-`dK79^Cn`_ytwH$SZ4lxtjnbU=hgCl)!uy*qgvVF
zq7^%D`02?BT+w=it1BuE1|7hH*7An;oO!_!#{i<V*RN>~@{)UgqiDy4YLx6ZXq3_Z
z^?E#WefBOuVi2#+OZD5^{^>tyWqN0f#C#(!_&0owZ^Jqc86tz-+x2Jm%#H4ZI%}sn
zTe9LcX-ye~FFaiTf;gQe*s(7#Nb4>pJ=uZ4mu4CXDpkwELSL|*GeS24D&yX{?mOeK
z(lZHGWg-O{PC#uND7EW0@LJ_BV;$IX1aXE~FGrOWq09?uK3A*6b!gIZ%jA8tA}5HW
z;aVGC{r;f^0)bGqQ?+jzmO)J?I#>j{mOEL2Jp4Dlzdza7loS{LDG0;^<K4|$9!(O|
z>0))D%A=GR4~2e>mA1QxuA{@G&u!l5eFwfvwK#|Xv;LEd4j{}gq1J_9PO@7dH}R*G
zk1~H?{=Jaj8|~x!Y4>5K$%+}YZY?*QCPo6aw2BWiEpK!T3}YDtQtGXLu$cc$yTe(g
zg^K0;Z^w`Fh>Iy<4~wrdNS_`NX%I%myjq^UU_TImXy)BU#gs+fL?9mWXW<HvoevU_
z>OsI|R{(rd^*H1|?f_|ne`W#T6EIad>L&HVue^2{sT9DGFqd76*YTV#!hB~{?B2?u
zV(Y-2Mk$kE!G{1|K3<~NKA}oN26nWW+AK~-qmpLU+~cXMwi?-Vw`croU5S4CYLPZd
zaoGy+T<0EYfGZ%?Ezs%`!4wxyM*O~WOyK@txe<--iMT&U!$4bh8&Lwna}9c%xixCB
z)XH$1?7UrwXrIsQDp}yS9IP6#^m#sAj8!-x6M9>QV3QHypW7B6%Xa<!`w8gGA%5a7
zB?>@Mg{=g5b8vg?KsnCbvSV`)?S7VU$wU|AXm5)^M^H&NuBndjK$}rt&!xlty727@
z?<5nGL?U%2nJNZ0HYj@lXvmaD>o;QX8myS+g>dLZ;?YDS1RQEqMQ+t9dTm~~L_~I0
zR=hx`|DY8rhHX~3i;<=9cS>Gp#cBnn=XaT6IncOK#7ox_#xZtj(*6@|I3ys9D!tA5
zO^+$IIcqrwRbFNrj;<^SYEd+5t>Wu)Xfi~zP{Vq`KC+1XUT7IOZ`E@0$1B+&Y4RUH
zxbMf0A17QDfJ<fON0P(%|1O<XUu&l2{R8^a$h$cG;0rP+-GF*3qb&Q&psA4&6!Z<T
z&(E>Dns@x*Hn=FdXxSK6Cd<nc3w}t-P^K!!kw|wz5mN-e+L6Mn(Q;r?$_ujO=3L)r
z0CRpS_s4A2$nRjH0P#%YuG|pFNGcbxoC0bSA{JB9QcYz}PR@V6&62%ukYmldcew73
zB^FY<4WmP?R)XJgN#!Xqa|TuQrF-3#46E2mTr`4=Ho`yS(92fDK7^Jvj$4EAbU?n1
z`l=FSt7;}EBf@kqEa3hT2Xr{PN-MRJmr#wht@dlixvt5<80!hLziLFg%Xqn82TYKw
zssq2Wot*I;EKRj~x;?)(fUO-s3AsYK4``97>%KvW?B8%~{HG%;q6g}^^}Mxi+qw6B
zHV4Z-swRuY=`R?x-FXdO1T$&@qcSl^`nI|G@{_kQKq+bnM|beJ>=56W3sZcVqSmgl
z$oltHR7IPyD^&O|-Fn%wZVMZp_~qG-3(pzZ&z$5znaa$;E(2yzl<?z_rp*elw`u8o
z8Fo+A?&sTJ+Mjmi22V9@h!yTj!=srLP8t*SxwZcYRJ8wIH4YEs*cWCV_ZcwvMG}`E
z(txK~l@rrG>>q8~hORxlUuDsDxOZb{>8<vyRM%TO^_l+MZT{gUAF#K(RD5(3>$^JW
zM8s{cQjY8bMqI@iz(F0XwP%Ru09Sb#5HSBcI7JaDe+=+PTTYoP8i*QkQ+@?~%tXIG
z`QxeV+HpiYsD|$CTg%Pw-cQ4tpLllFpE#@mh}PbU0-m1>Oo08tjoR{SnNK}<do#d6
z;Bf1U4u7>W1v6QG6U$Lm17S(*cAyb;d{1HQ0Ys_+ec9cW9k^-*cNP5g%=dx4Fm-iQ
z#7EiVB?q5i*wZkz^V~mu>S%3Npzv$q|8&y=B&&-N&*LI^lL-iO<%v^8Ebqe2ku9w#
zL+;G~^yQkpuCmi<mDZZRv@;1Vc2N-lAr$7=qkyuaOP9-NF{=gb+yCb4+<ujVZpD~1
z^VRIn^tWdd77&OYTjIfavnzxd@0;Z}cnQ^+Y_Zn|BS2~*$)IxBJ@Ik%F#-=c4^5}Z
z*>bn(QaFlD%M&Cl{F5pHIZEJQAWm2Pn^bwRf!s=1YPaKY&+SgGyc9$kbPBYAagIN~
z>T4@JR&LW;K;WG<A25Q5{%BHgalifF*UAN}%ErdVj-8j`Fc5_TJ0qj)e{VaW=rve<
zF#gxIQZ9deetLLzw$4`S4ZkOpREWbi(}4+i|8S~!Gib<Pl!t2n4tD1F2Bkag?#jrt
zpq|br*%$kkj?~*Mu+ovc+_!4asg*f6I!^w35*1vsjCL4<uG7K=w(VO?ABIA+mbG7Z
zAxM<KInTC>6X-B<adT_@L*PVbkVO=E6+@$AV{`g)J~rxn7AA?^TlkPM*lAWBD4JE5
zGp_hw87BL>QqHD;&_@V_5>neTs$QUw0=ipkK2+WJF%+=_C8NZ2c;RMtc4db&HS$m5
zZ=d&vTsP+RXGvyM-`?Q9HCPR;`0+$g>9N8(4St*^slxboDjYZK#skYbjK2tKz_XB?
z9WOka=3XNnWD=vl`kYbnZ~T1(8m7|1(eel7rWl7G*H_<xj!lC;VQ9;W-z;7v#>Ay9
zxro5O6NI7x+WGmc3~bt8zz}WXjU*L=AqHsVVDTIfLk+wq-ryB1vRzdr_6=Q-#1#LI
z(gYvNH~#Ro<U?Hvw3ll<0gQ9|mAoMmEO`F`?8yHBVmM2Jf9hb=Np4y0e;4*K$>CX;
zCAsn67iic8%#3iP;oSj%9R!eaF#P{dFA!072tl2Ei55b^&(9B<-T)E~_&q@RfREP~
zc+j)#wHqAP50<|W`Vhh#zHSG_?*Aa)2Y>Jm*6P?;5t>|KJw0)Fwt=qeKY}+Da53K6
z)XM0n7*LQF?~TpP+<!vACj)TR|4(&S9?jO;?l+3lo>Xxt)#!kl=W5MkHPn<+>L{g#
zQc8$6YSziDbRbBns-kI|Ac&e<p;|+o8k$ofD%2D$5yb3!_51Ex_gi<}weCN6t$Y8-
zTFFjU_WSPrJiq6Ce*1ZT@5o8eZtidS;Y8t!X2b2tCFZZy`1D~g@ij7#1OpX&jytw<
z{}9Dh5nQ7Ga!sb<ehf!lcl_S?o;6a>d6MrFDlIlHVYUmVZMsY03r}Y1(-$IXnblSM
z-#NUp7?-yiIM+t(`>w0Pn>rZz?oONgb9a;e+dpnerI>QD5VcA&e4$5o%lskx(QVl%
zt~GMM);r?&20x^4Q$h3JE}jz~Uvq}pUF$#duUL$NmENE|2xm9LL;z9zR5G3ig8nCm
z>-Hql9GqjTGzE3puRiw;SY-6?x@e2s13`>OSYzbe$=PtmWZd#zOXCY)ZqPTW%k1Q&
zH@EgZxwstp#)I=m6f&dPZ}e}s;xRlIWCHKC*BM&tZQdnIuLyToYjQHCR|Q8ldt#HK
z7z8RZ&ErkD*?DG#;rDbUGce*KE>rSwJ6osc?UxT&rO|nFmrisOeH77ljwkqLw%7gU
z_{?&%>5F#?gGTB=xB)%{iiucb?d|Hf96HEBoxP;kzOOFWEF5kIlcIB39Bhxy);pa?
zQtm{<OPuPJI~}YvDhG=*trk3sL=Kmq)*^&dkr6T2B>7Ii>L)1Ck0NLap>*h0>d#u)
zW=6(7BM&`aR*h!X2e1vBqLLC&kphqL;WR<)<yuv-)GCUWE&ZInm$(~Z_$Wo)rtdnu
zoq2@yK^;~sa{SdrD+S;E=E%{rE?*JeZ$;6MCGH%kJz!;Se%Xhks5ldl^0jEb;e869
zGx&R8jI=FeP*4!;kMW5KfCQE@3j-s#=#AJob@4?ogN}}lJ@IBLDk|>$$rj`w{d7uI
zK=#O-HmnGb|M<^y(#Ve@F65Sgx>N>c&)oEteBXEFE#CUed_@EL#&!o!>YO&)J0M4C
z<ijXM-3UGs8Nf-OX=<Cazrh7r9=bs1#N6?VtKmc;%fj*Kj)pU}ijmPYcAU}pLDulW
z$V{WnRHvwv;HKCYgs}*4W1}6=p@pZqY8)jF9&}j`5_46Ye2eYSEiVog2*l^{xGI+T
zPmGU;=eI=71_Dd`ArFTj6a78>5cF4i3l{`c6@ixN|I>#n8wBRJ#{v0=G+G@8HU=uJ
zb<qs-vM~s{@AmL={MaY`IXCySRlZ-ADhphyf%^NuzIqk{)Jl!>12R()mMCIo*C6R~
zDitURTdJP1*NHYNsM6}3mKA!CVbt*L_U{dLyC9R$Gbt@cr5?vFdt{gWDVr*$Qr@3-
z;CMqikY5P3Co&`RP0gRz*6Jo?U~xE!J@N4N<WX-~SV51NKw_hvP-!479Z~f8O8<@>
zXxUORx#xe7l&uxox&pEtI<&eiWf&Miwc_I92+4O*tkCQZP!i+4as{-hc6WDcz(;`E
ztQ$EHq<iw<kb*+;041_zj>CV)wCr4~LZ8;D!n4-asZ>+OgCJV^n9M2-VX+b0>?vMh
zm~jk7Q4gfqP558N$;O{zl!mk(r()FVV{UD%3s3ickHe~|C@aH(el4{Sf~IU9ewRou
zD<Tr@>Nxis3^Z(W)zP5K%w3Kfditm^B)Ynx!2tg<b1gC%@x5CGR4UBt2|-Xk%vI6D
zR-yVwg$3BSS^~!B*Oi0*cq}_S|2OEST|x^Bsi3MFJCW<xU<b_7uMG=;>i>(0{bwJr
z?xED}O&p2v^RO<ne!q7P700T^Qe54KT|`ASNQpMJp?{|Lviku>{=DKc8j?=%VAYfI
zo{V;97_6e+2CaKlg;HG${9y8`s!dPjjZe4taYHwCioN40_~<#xtkTcM>(7VT^W>%V
z7H<SiiqSu?%I7v%b3{p6O)e`Rc<009^p?>t)Y?@_j1<D1nwId$)H(1N%G-9Jkf25+
zV5JK#XAf~U?Om+Qfl{2ki%ag4Cr#ljhPdomtJTd|D#m*yXE%kU+^4l;lggc%Fv=u5
zLQ>|?frO({>0{yGU2X|Jv!knPQP0u<05Rdn8gX=`IW7hr>G0KC%t%wC%=%?NYUSe?
zZ3bc2=`3|BQ9uO{v}exjPNZl=N7BgB4}SA|AEr?3wH_cR+^1FaS-uPjKs^+m=s%#k
zgSpaH3Z0~ft8MmH#m_^I-NLpMhO*P@>XH`EEn(i}H39T()y2&Id99#9^_eEA^7+Z$
zWddlW2w1oHg}p1ATW94(lFhw#W}=>k#qQs60UDVyst!&RNDKRQB8qyBi=+mOf*JOS
zMcee*Adwusc+PbZfueS8RRIIenniBJ-0XDB*8y@==r(y_Hdd4wS6w|CH5OBUlGh@3
z+u)}?7jz`zN^NA{6sG$}Wu2rxymppcD^{9pe*0dj^l*tvU62p6lwX3s$Mrf!jVQqj
z9Z7znQ%lk$yDEb9G@!CxZ<0Iz!%6X=Y4NOlX4lo`Ul_cy8B0+I8>lr12xq`3<w5$n
zrCI(s@PO03bpDXi-W?pay@3z!!VCkUHBh`ruBuvzZUXEF7W*%Cy>Q9B&`qDroOo}L
zyVjs+sw&;RgnGEaks{Rxg<?kR$CFPo5FirHV)TIB*m<GwV51(>3A(bVsODAjE^f^w
ztLXt(6s`ue7L^roLHE;yD?@-<?}yUFV}=)BDxOyG4-^#A4S5Q>T$vzcktu3zux>-=
z0cnZm3tJIXZ&A5A&eR2`2*$=%<=P^kB_e>HEv(?hafWKvwQv9S68-7m|0stVGCny8
z1h{d>r6C9{rBUMurs-DrzZhe7!UJ!g764MfB_IJp{s53%EK+7;F~el4bTXFwvpvOy
znRwjJt#G4O2ho0o=`<PRA64Cw@>ERYC&4H{Vni)|wpi!P&0z$uYu*@%;Q-#d1X=KN
z=OZX2e~HJ}a_CWw0^F@?0b&5zRNcvrUd$9F;)jlo1wCH90>Ivi?B5@m&z+A1L<-P#
zX|&^Nt!^E106`XgOsK~3**hGGguB$wgii&2lF)_8T$dMqp||k0B4Iu|u{|T!eS|@d
z`_06>HF~NrvEh^8In>g`WovCm?dWpUIrjbF<xWR;elxe<<CcB>y-hvE#%3QpvZt7s
ztq;9jM;gwLTNF+hGTrt1$4H|QqGbjff9sdpH8h6R$FQcj>u1v_S_uMw$$mbslea(!
zF79^Ixj}F4wv1tSYZ~P}JB9i-?;7XE>YSvPp~nv+>~e11O~1PwXs9>yg;zUvF1?nA
z(;V{@Gg5LfURyQi_;z~miH+ha)zFEfui$S-j$If{I`539^D0P%mC=2J2xFK-tr{NL
zMwDg^`LZB^yc3}<V7{%|8zpUhzr)ZxP6LQA-*S>((bF0Q#mcA7i>=9T4<y1Qzgh2g
z*16!?=9iqb;DWOL{jkmFqgl8S&h4gEe2@~|?_QoD&u&SL4bXm5x^HXx!S+!9wdE5r
z(gclC;v@v^f}b%rb@c2co0}^u0U2H}Jf(jQubf26dylYa-Mwq*_3PFKrkDg6lfwba
zDX?x}rX+d<;=+6GXuIWj`uLXeALJ$`{c%N;L?V6C8D|{+H7Gt<r(t=efrwrm`x4cY
zC%oXT@$~Blk2*}uEO0jnU;CFPqI^9vgOih<!w&hzy{VLqTnpA$N=*=5=`paOdra=l
zQ(?vKON^Q+!t?|iDW1;uFdl0OG;;suHiXkI82%FTshX!dP8^s;RS$u_z--r+46fIq
z7>9mpq%OaTnu`~s<LwSa59~e1zv5Iik#;I$j2kV@+F8gY=_iwmxj|YW`0uGj-aCc3
zF=w^8d*PBiCbxkP>YS>Yz1Qx(aWJ08B#^bvd$V;kZ>W-$)xG|>E`4jct0JZsN8hH@
zef!Hjal20czP{ImkI%QJrGyEw+XG9%+4z<UAmuyjn_~GUv|Wj5{i-j~=7~|nC7R@0
z)N|5eO<CZp&co*x^3l8s^esn7m|F>May#UzM?I#u`>JB}0YdRUo5p}uIf8Bdc{Rn{
z&3S&ifDq<pwny#aqjJ}!mOK{w3%usF_tlB?$84`GX7pD5L0Of*LwduE-1TB+kjB3y
zC*7@4+x(m=S9v_0(8QW4_~~20V*)F0;gaE-tQ;^N(Ol$Lk`t&JZ=rSUH+U0Cea(fs
zzQ`CU%c0roQdi{e#YM9-n>vkg6$_h*2%6?LqeLh*wzB4h|MX|(cKK%0(URkpE7@w=
zT6I!h*XzSRs{IuAN?42D=<Q213)PU@19^x|3LVifM><mLme$FEFV9LnjQi1$S|{Oz
z;F$7U%SsYAOb|GE@{q2NV}Y^nO%*Axed3Oe)l9Z@sbWtvu|Pozk%7N4tSaRxRC&eZ
zfr3=W!m7c>14u@g?0B>FlBj?@DV4bP`5|Dy4b=T|th%QIlFnV5|4YHyJP)_YPZ^Hl
zC{2a=Y>$>FSx(Jw6-|mR&JO4|Z}cs^NP6Q`W*ELbxW~f0RbA%<l%#TA=Vt5E$gU>l
zLObd)W~4c`p83SDfQZ{hU<XT|(Xv0e4uxPijXj3+MM`!lY)%NSH;S#ztx3a>jaY-H
zIpFzn$ttnIpN*<PNt~}iwRtqxPP8Y$S!)(t_mfs!6-^i+Gsa)bQS7a65#GMo4eQ|g
z8N9!9)<|RMVrg*JDxYm5$_iZ~oa{JvVY`F&-9abp{i6M_ve0g)OkxZss@QNUWzJ1k
z-dgv2ut7y-kIJz-D(H@rTakVRhuyXBB4wc_)jj<8Uktr0ExG8#nvdGnxZMQ3y_T1B
z^KRB)+(v&MocJCR-1jHQS)oDJDDZ~=F16+~SL5vV`b@4UgP%fYPTQFi%`V;k2Qu!4
A`~Uy|

literal 53428
zcmZs@1yEIC*EYNf>5`HTLApCc8tIVkQo0+FR6s$xyFt37yQRCkBt*LNzkQzf{r>r8
zzCELdIm3y)_qx}*>gJoW;#+hSA`}RM&}F10-a!!T7zDvwBf)}a4qhstg9l^>X>Df+
zLdE{~4`yQ7=N^J6AQ=f!HILMTG*=Pay}R~|`FSJvdHqJ+MqI3(_FfV8ii_^{ZoBi5
zH%+KYqWqPc=OY4=xG~BSFnD|MBl<kFKE5SR_Y%+XLXSQ@1|)f|dafp=CGB`)Q_5h|
z{m;=#^YcguLix`T5|8=HLHWPOCs#WE&jJ5AHj2fONpSyr7?R#zk+l<BSzD)6@*;fp
zH`YQu3zk&D?K?RV3i%B$sRCJe##kA1ieY`}F#5H9dL$J28(tI(UKA<ddIio>ao0sa
z6pPvB9rLDt_PB)&&z)Dq#@$&C5()OVTtY9?>Uon&g02d)gR`ctZKsqg25)yh5{v+f
zf3G;=p<j)go}OOyq8LUF$wosEXNdDT^5>%c^oG2PElK{e1#u$v=XXRDNfp^OXdqWi
z^gqW5dL}%^|D0br#{c{L&hXNtI5{~L6cpm3kRZR!>yb1;Yinx-d3hSe)a#28x&L{q
zoD+2}5^9<J{CvB`T7@j-DckSgzOi$1>eg7PYin!!_&m<?u|E6H(G$_HIqo2xH<6p0
z>mE=rIXSslXO9VmhllU&?MX>VL0C9ARM_7V<?};B<v1}!lK3+f8zwkJRAPHfypL44
zXz4;;B2ApqOm~{)%4?5z(n#iv>#U@Ki-v}<=@pTkWJy?uEE;(v^o2a2{G~(1{6Us@
z@Efnol1Hg1>F#xhSTD(!QBIqG7wgW>&eYY_1zmQ>noW%=wU}8?G*R!SCoH(O`wqM?
zbNMYj>ty&z@$llny)u)ccmDM6(+5@!LQ|)%$a)ecI^RfT4A;rs_iQ`$n6N|A)YvQu
z#Y^>{Zk)EsKYaKw?7@PLjvggFb&BoFZU1KGD4YX)K#JO6lL8l!g1&y5?ES@N*Us4o
z4(&m?`)OD*I$X&l`QPBJuXYwC3$?VhU(wRSi8wh`#l=nY3jXiqbKlWLA}K5^yu18+
zc6|J6czC(dwKzULe&YXBr7!#zVkI+olT2QIeh^3D{hBgOTn22OOkQ*I{lD|dl4lS`
z1$8E-qNcC^@UPCqUp;1L{^z*fSB6FWpBtqajLrAIi$Nf^{lB+b^kh2g|1X;naQc$}
zttPma_@Qy%Dqz6dP|(r+sI9$BF!Zrpto?gMs)hRbf8NOlmV<+%)^3>r7scD#I~arX
zcEvRJou!>!Sz)0i4r4`bem(*M!hSq{+#VqY0b3?`m(%S*(Qu9Z*E37noHObC_EuI4
zB{A0-LI39+@mUS-3v%+x8dgd&g-j~vW@c(_7i%vVOQ~sS%2K#mYr@^+MJcP}@Qz&5
zwaJg-7&Wu$k`oeM;Nu5n|7dEuTW)f9b#dv9ARpa4;jvwC>73UIUNqZ?bLfc+?bTEC
zdbeJZ^9(X2{hOVWL+_tMnDyzuSLIDoRTW171vWOeV#b@aq@<g|VVUUf-@m{1y0np$
zlw9uJ-?wVedtBLMWMGhIkVW?_;)3rlDPrMw-|sjozJGW?4v3V^<1`y0SS<VdLnTKM
zW52G!%bYC6bE^NeQ<?nkd~35WLQ_+dm4L&+;>NllHn+kMj&`tU@pS|y5)uRzsugu@
zM?!J$=I`CgT6y9>m=vbA!>sujz^Di2cxG|^YfVpo*7(wslZ&#lvWkoETKZ0Ll(n?@
zz&G4Hjec7mDnLl+k(9!;TuaEyuBoE3H&uue+7m6C*A<NUG}cp6R5UO;D(tlNRnjdv
zFxhC}H@X+AAy}~~!oIDqA1jqv>MjU})*Xh1h7iz+)MjP%^;<-!84B@skOTjspKEx1
zgxVVsE0xk;e~b8b6Bv6%`o`gRnDm>_XVBmC3p6!#nD#E|C#mj><2l4XO>Rz;lj|G=
zLPPQ{d^U*83X4TVab2Q)%OqAG-r<uzcqL*&KgRandVhrTTiqru^(8#|{reKVZ089>
z&FU;8C_ZYV?#;iS0`M~y^r*2FlDYa7^kP)?C(V|=6si{Z+?`NUQ^&@~yB*A_>^0KG
z^pm~zpuj?U({Fi_lE-6MYctQ)JR_rP-9GQj5<&0<Iu~+g?_QKk7<5E?Q=9kJTphuA
zyM?aFpoN>8n^q~kq@aKn7X=Ouu6WagvBqV3FeL3~%>+Zy2-*pPUz;lD6Sl0&v-?`7
zembu|-v^2q_q5+=;7evNIHB7f83-X4{%ma4d#7F)kS%%RiAJJXK-s8=0exlvF8+iv
zXa--H#*u>cgIm*A&C2;zZ2QUN*lFu~@TfYr@hn{U>3;IK!`6^k4kB>4%rG6VJV#iq
zh>eZaZ*r5#n?S~)arwJ|_iuK+eEAX$4ejH_zWQp|8mCsZ89MWC|3befB^{~b;YBnA
z)6Or*aJ%mdwe(Cj>%pp_xty$T_cxwvdwSGWSD$Y4^+iVyQz9}}Q&W?cMvCbt5&G!j
z=5`r~`hulV5Ls)eaMteda5>FC+nT()<l-RGkSc_piNwrQojJyE@YjK<aS+aXdCxq7
zh_`*XZbf?jD8(Xl99xCa5TDFM+d(7u7YZ46QP`MoRa)EG3q;@G<78~PreM}YeMBpz
zODOtvSTM`!isSp)0~{u&W$2i#xBpJo`TD*K42#NDX27%f1BLJ$wbS{FBfoOE=la>q
zN(dfte?tB_+1W3JgqC}Ir6{oI>ZV@sVUWF^6IicIXh4S0+B_~qwp!d_Nw%DB5;6XY
zBYK=94)tqN`3O0NQ1|Ci1&uga$Edw?E6bKLG&I!J^_nhH*U`}-C;cw*QoG5`cI9tM
zGFh5NsSarT1@33Bj|3s8EZ=;Y(+)hrj6)XTSn<pcS+jFiuX#%1W~~J_SS*Xr`M12C
z-62@e!PR>UB$xiz-CdgHmd7_pL6jWyMg+gMxf;GJ(?zGhF7(hhGJ8Ka&<!8e52^Zz
zQI;3>%g9!Yk11n)QKo#srcH+~nqsKoXO;9la6<MKHcU)je&<IWJ*z^NHTI;02@dZP
zCM>^qmDB;bn1tLR65~?`se<F<hj@~t=;)BqL7s2JAt^~(YHv8{;CqhMa~UkErR8vp
z-O>(?B)5l|*cQap)YRCfcozM}ui@H1kfr?|9a%?G`6rtvJrQNsl*MlTA)$K_iMMYP
z5)$0^CIUxy+Nz6vs=o9_$>yOUzyYY^aeci0c;gJZ%fT;3t^2nQ;>GuC9iQPaNIjUb
z^E5T4iBg!gDnyi&SYN#gj8E@$vZ{9dGZP(+{X*Y1Psy*m^n;$B9smtRBV?p#7dUbQ
zKGhr-HI?#o^!hCyFIv6tRs_kC<U#v>=U8RhhjBDua%2p*i4r<nrr%^XO!VEOdVdk;
zYKWfpNZU26GLIkn<@h>3(;?lRyg{4M*Vh*Sft@pPF)?^gZ6dqBGmn?Vq0~;C9A$>B
zB_R_Z-Tn@6@`y>k%$Eo-LXlvBYk=**yBW>E?AvB=3Hliph93+Efp+WgdaGsA1{0za
zF($+PJ#G6QnzOcqg=EFmcMsebDUhVhBVB&JqbrJHL?TkqQ`KjVE9M}uyA`-PL$TqS
zcI8XG^L!@OkR4jtvHs-+N`TAM#I+!^>C`En3(HkR4{GP9+X<}qZXUVc>LQ|QTqPvn
z8HKhDR{kt$KSEgdOoeTE{2>%QH5*#^zJ8@}ke1dwgZxmfB8mO)9$)4X=2^~L)4p)`
zAgP>jo>YNqhdzufHsAB?b^%K`NEtV7odaq|-ayoLePR`{<ntlYS(!+h0BeVin*<pd
zIfd80?8lGY<)*9alg+k?23R<_8bg|7<ZD{PRsj#HY(--V>~Cg6i5>m2tO?RD4whFd
zP5T;Y04f1k?!`9=Fh0m%d$5IXp8Xz<dF`o3l@Rgh+7n4B-Wx{j-Z}Dp&`3=>Z?ZZR
zll%?Z1s92@rzaZ1=V7zeCikP8vz_xFUe3#lXiW|m(!Q+D+@da`#so3EKKIw$n|*1-
z6B)yqN)7nz7Td|Wxzu;LwBT0cJhoo$_`P9p!G%7J=<9gj=QcF(fA07}OePaOn83=~
z=qwjt{3)UKWz_EWV1lJ-&=?kV<<EPgPXT2Nb&&E4o|rxpr`>Z(o_GS$R2jvZ$*CVv
zgP;>#sPo6YOCq7Ymd_8MzL7`Z;tkB!Aw>;6FR>wJA%-NJ4rJ}fLRskwe-OOSH4HVb
zt3JC=`&({ic9vLTM-xQdem%&0xmu4buOa>iF31>OlmuKAhJ(@Ug$3hU4S&+5GpjFf
zN4)|90ss>=iU(sPB_#o1Iy*ay=y$d=nz7mp1AQD#<hGok{y=L|dA-pSO4#>w?p$^E
zuuc^~t$*S2()rZQp!vOt$vgM5#KgoHS|uwBi=jmJ%@`7&>gwtj7^F*GQ+#C=bM7}N
z<m4Hw%|T(}mVB|wuzgD6gLly?;L-ox+U{h*s3NBjDtAY@d(Og${v5awX2MYtFOw#V
z@Af)YAO~(C#!`KAU*D&-wKd&3yX~zl8++5$zD~98#_7C=uKQE{!^5_}A}JxZ=F&eI
za38H5=N?Ayna4Q+#KCc$hSSrHfmX~P-w>WzfRSTPs#~e7Mwg%@&x(YO{tU`%HNHLH
z8`>G&;>((;=}YFe2E|o5D<ZBR$bEhP;lq&F+}T!}>ICx?yb5(ue!f(6t`?J9UoJM9
z5>;k!@_zB@0Od?6f{wP{;Lu1(Nv+z8Sh-y9AaqSr_R!N)XO0PN>MW(VVb`xK#YCgq
zZ8aQW#8n#928`LDwVEICwk8Z?leAbwG6gXV3?g9F-ge{|tA@F%E4$2Y7WCAZWXZ2|
z{4y%}Jlsk<np#uSlt=*k4}zqW8Dt3iJ}Qc%!>LcN6tEs%>`UcP2B*{5{#{3gLhY7A
zV@*DO{0QJeIESOZ@>Y|}eDFK_>0s)906F`*yMGiFu^6^4b#`vV(0G8AVm_MA&&T)E
zc9Dmcwr_kqYS=E*<VLUC<RtV3D}Y!(Yid%8M=?c_No46_P@X;OdE4%oniOZZXMzgl
z{Iykl<8xmK8!Yisg$3&D>H^Hh>ta7Jn=X8_qoaeJo!xffCo&R}h>F^&Ta@yvn7rIv
z8$i`?9UrVlS)3VVbqZ0THR;Y!N~@))FIxwdRKB<`Fi8cb^uo%^2ZoEETUDrsyW6@S
zTfSrD=7vNdmxhDp=4Qvu-Z1@^?1;em>GP63K{v`oSX`7qt^0%i7{zVLM{h6K_VL`U
zWL;hwo#eQ9yu#v&csB>R@)C}XLIF$bFy{(MEQu23u|y_>fK2$u7353%509(D<>_#K
z?0%KEz8aJmo1e206r(*+mgYKN(!YwM$1~+i7dJiMpJi>T*wwzic!QtsMJB8}Hxk!p
z^7wch6`CGiTy_~gcYS&G0$j}!V($Cv_Ew1PaKg1_h9xK{2nGhmhDVnI3(!wM5Kg^3
z&{wXT7(R=-+qAG-m;%sKo+I^he0pNj=}K6J4eywvs3`R3MVUd1fXDG#e_wc|b9Fsh
zV_1HpmVXN!bHXMcj^s9crS)tX;0}<8v2m6Ng+;b4s%WI=*>7&2@A5QFPTL`3S;r?Q
z>rM?9?x5G9lfBlW@A$L3%eDPoA-HS3zNv}VJ_GmP!u-D)af{d0%Id(FM7yCk>w-_O
zjlKO*{#6C|q=Et^3Z$2@gMESv337Ruo<x+ClwMbd>Jg2C<MZoBQ{QQS&oFA1p`oCB
zbiDmS89^K$)if|o-<au;SgP(Tca|8?NgA>7qSigd=RxB>w-O0b>>Tr|^loyOLvE|=
zt@c)|sgPjWvozCt6}LPzaKXic#el2bsbk8L5<cc~;X+8=FM*{-8T{L8a~+K)tb>9O
zEjlkcKJLTd2RJ|6u1>ju`F*1A6B~9kQeq7o5!c;5r9;`MD0^HB66SZ!MC+$Z8*r4A
z#I?3N!r}XNR?V!3^Iu(w8(sfMVM)l#qlIsh3c9HYWG;bcZZG~Y;WJ*`a$~}^>;{Pa
zX<cz@SnBQlz1|h<)Szc*DExG2o26$wu=DrN<mLzMo-tp0PvYI)7kGk}*66jBuQO5w
zc4y<jm6Tdtcmuenn7BA0T2M;L;=|n)9*SgtmzR&?<(1P}Sa5jiYtO2Tj8zs+A!qK6
z+~Rc^Byml(Wow=Gt^SylKaJCfT@{8tH{0>Lf<|nwk0jIi95+Zwr8wUxk7fwJRhb$Y
zDbxOVM@@$wj4iu%ZE4|YSoK8ePsPGw2bj9FR8bejDwsXK#<<(F2wfGoj#>Qt{5Ll@
zK<OFF5(hm3(0H$y!T<`a1~dnf4mJv82|^Oe1h{Z2YX9cp7uxD-E<#&l@aA{suku7h
zh=pt+q|pJ#KS$0xBcz6;=-V~*1{(7@rf=az0iRZ|{C4l^-;gX4U<?M9>YEITZ$xN2
z6<YGq7G>8+*uPM|#eU0vhsH|PF|gvVrUHWiw~H3t1B(!oJINZn`elgATm5kRXG#Ab
zAI;fuZLInd@)!-~m_W`3KqFo2PIt$1M!UiRab#)meyUUaI|P3}%<kxW{<lAC@_{9|
z?u69J$VgdL6^EF(I6of{D-Kgr(`f2)gO=pz=!q6DcTLUN-nY8Cx+zIXmd3^%+fFjk
zBr`MH-R(X^HgOHSN7r_9rw8+#1qGD`zi9xqaN*^?<l-T5d7Ln8h1=yyM(caqGK7MP
zdZ#>!iqE7~D$Bc72<zv*6v}BnvbDW!yHsBZsO#jUbMJS;VW%2liFfbbMRfUgZWTh1
zZ<^p@p6b=%@*ncC^Q)EMPa%#Z34`0<I=1{Wf(ef>$@IEBP+6(ZZ=&?2jBU^Q!cnNK
z{K=I^A>}mT2vXN<VrH+@EH@x!Gitvg8*ifELoh+6#3q}oOp8kEkgyc;MyTQ77?6=E
z^E_GdAH^Bi_WfDAo=wM68f|aTmBoEgU5?1sF;509?B1KxAKBWWu;K*o-@o`fuPI0R
zE(XRArX3FAhJh;fTu{M7rk_<d9fXXo7>e`$m0x44XesO<6I~?sm%y~N<T8WnG{H0#
zWmRLC_9IMk=D8;O`J!oSCNye-R-xbhwdwj;B4y^kUr%u#PUOiQt+c#4;`BUsJi-E+
zR%!z8MV+gx+*ABo<ngfsI$m7lQ-+V}{q@N=78V&989P>08UaB_BuR^dfv00)g5!A1
zOUx94L?dC~5V%hI+P9tl1TYN@4i3YAvc({V5OnK4-kuk>w0L*-ID@+uSYv|oM#*#U
zi&J1DO0g6TP!}w7%`t*iSqj;=3I~pPXI`3`ZUV6=|NH*n`q_>03C~7j^Y^|@Zz)=_
zEH(;?wMH$J&Sb_vGqv`st#-3XoaSbv7|%#P^RTcvazLpV3xZVq+J=U{zlV}|w+W`y
z3|H&@2k35X|B0F*qq$X8rvSThB-w2Jit6p{B^Qu?8D&wau9k!b9dq%#CdB(-Wi_zv
zG@2&Zw{!OQ?_csapDavFIGnbmdl853Sq+*Y2oW)u2BQSvV9)@#h-?q$Di;<O{5HSJ
zhkWe;|4Da$)sip9duhV{68mM%%iLQ0LcEtKfmMIsk;3fO!$Qp?{<HiVkcJe3{Vg6c
zj2}XKCN~xj)|dC72-%VdL}S%8Lbg!w@bG{VW>g&$9gU_;q!@i7`3m8+`m=D(@88o>
zQ*luwhc@f?@0(4~GK{Q&SdbJqZ3!Dz*5dW#d=tX=H;$ysvb4k}X-jFsqTL8eQ{+hW
z%Z88Pcbv#&qqWBPOqP^#RT1|9+eVKhQe4*+1@oigM)mh7vV*Hn<dx1s2M#aKp!bc~
z@Q7I0Gu8oy&gkeG{(~GDR!rN~;=;oDUp*z{&*2~v4D;i)jw7Lo`G3dFMz}43D4jYg
z4vjp4pqmZOk<R4!)YMSvb9^L80lkr=gu=FK#QMg@H{`p5*XvyjJof9QUYyH102Ags
zHT*6qjnmSpy`1Zp`iP2zl*nxzKX|s4@BweKj*Gir%h}k<N?>*7<4!W3i1~F2k*^j1
z^=Cv(APk$Co14pLAlokb%Qt$T-Bz2AGF-{#O=$ZDb%ZZRUZY{-to$@)V?#qlL22YS
zV&khrsPtH@WiC)mvs&>~X?`(~dpaqSq~=3HChVi9tc*oO)HtM^()Yf^2g{l5KAuBx
z(#w95we7x2O|>SjQKt@Pr$sl3-r;3Sm6gJah&#`6@-`NonabtKjc8x(W}k=K8oOm3
zCVcbv?>#&`cK7z8sQImDm|BL^hyr30YZuP%1n}UqObH4jsZgFNuu!o-%WlU?g2&Y|
zkou@IC9Y1R8W_G{@q-fST~0ffmb_bxv+y8{G>Lf@9VGDrxAA<>>q29T!j@YgOygAy
zX>t5H8Q<=Rf6Js^D$LqDD>H@BH*g5(QmaJEXQ!td8yi5Fuj|)Z6qR%S`+06s@&gZ&
z@mOg=PEL+k76He5X;;^pqN07l)oaJU%W*1{fH!`Kiv}<hPDDlJxhNJ8U5~<n!(w7;
zT8AJ%`@UG3spRr#^=T!}(aGsFP0+ohv=p>OWN5NLDMO@7&!xt<>BaD$g|4Zosl~;n
zc`97HrTXX1ua)%!0|jd4tId`4v5S7cqM|ZgthEKhjY{hK4@SOh`TMxYNGw7^%MTy6
z_x1{zrXI)HjR@k?(rk^4#w6}wMX|tu!6bWKQCFA2Z9S9lHe5|Z0~`Ltpuaro%0NwR
zD$|tK=r_q+C7x8oGjBM^ue~QvlSN%q6ZE#b>0(V@LMAN|BtDGl{Lki^?Q9uU8<CNb
z&rtDG2{>S5&C&GqK5kFEP|R|qiwXJ9kj{h;FNYDu0_~icj_&yM)M~nDx{nwt(X9(7
zI}=d|1Tt3X(E>6A*2P{|AD#Hks#gjF0|N#LUu1H!l}>D1wdU;RXv{oXuQKuAY7SXg
zWnHE6CM|w~5Cel<d3lPOjvd1q@tNl~_P6jdvKR_ATqm+(&o?^oqQ3|6f;2SvG`Wj&
zf7aK|OpR;zEv2e$KW89WtB9GLn2jK>t8%{kIZzZzC&<Ug$H8&xu(p8_bnEVPHv1)w
zjgvEFF7Irf@<Dy~hwiS7oGUkdkf%b%-q8^`DXF*7W9oXXhI!0n?1Xs6<O^1ei*>eL
zuZ{G=h|A(~0wiSQ2JgGt+FCawqs-RU$NlNzVZ&Fh>^w_1d_KnmH9{ZHEjGTt=|^D2
zMR~g3j5s(rczn3KyS|?KK=eQz5)v|!E;MLX{WhFro{c*RJDM687@I-g&1e}J|E&5x
zsjI4<U0ggY6kwpE>qo~UcS*J9XgwXAE2Qvvw>Wd3t_Xg*cA&@{YlWkJ)!D$w=EU_{
zX+I<3@7fv|OBY7F(iPRgGCpjUC>9nXa~U0+<`&l004wXYerf;}EaHTMMvknsxl>r>
z`*5L&*2SV%9|_=GxnWzrHBJiBD+jtP`0xw_*)<D+%F4<nx4*D{=|WyShx%<Ud+$<8
z)XYgs(<0IzZ_N{1S{{PKnOIm1YOLa}=W(EVzNrGemr0Hfm*Bd>7+G$8_W8rZFx-xA
zkMlBWtcWoxd4J;k=hB~90b=5*Ut4tO>ppynRgT?>0mGWbSl@P08vJ?~;-o{gNUAmB
zm5x^SOiZ*G;+IDHi7m~&xrHVp3l^j_&F9mCQJ3^<8f*fng~HP>2@U`Tn#YL&jAP~E
zdT%0f?3R!qCMM?TW`|r~UmrmJt*!TRa&jyQx_Wvx@89PZ6`??Y%>hiw$jp3UGBM+_
z2SHS<28*9=_JLJ$MQjO3mYH7LUu)C(4r$+k>68KlFE|l-QRDi;i~F0ixPpHMmDRDo
zdO8{XHr}9t=9j?0bv~c<&Oor1$$-AQJGOqtGDxc;T;n-eSW@znj31@<)?aOcd7~G|
z=tLU&g`cVWfBc|p&|7J6-oYdsOb%^semzUC+F=hH8}HdCEC33q+C?t<y^W2Ht!-&R
z0XY3{ZEa6^e*BqS+1wc)MHu4VcpUqU?$iM?V`oZr)Uvg;#looB(V3^Lvr+3?myn<d
z)IY~0*v6|bIaw8!jF5<;V!>MG;d0a6?JW^8vH6D&HK>Wmn=~ShQ9BQUX#%cQm6eVz
zF7k?stuA{Lo>Zz8KLcX=fW+$J=x7-)Lk~QwSFxCgl5*l%$~MS}ZjL`NgFCbw4yUNh
zcW#f$ZzHL-J+9owl*u8FcWPAw_DrN%doBkhdg%!8@LIaMv!u;HL|>g!vkAUE7e3>K
zEUM<EJ{h!e`f^%urNF=esRi65K)9cE5^?b&meLi(C5Ty8W>b{g!PvnXnb<(k{YN6y
zXQE%XPQ{|;WX0Pvvt~J}?sO6vkT#}5HE~i&k+6SDE;&w5eq&@Jdp|9!ePr66B=`Dl
z0g&}IsXjp>DZ}XmPp4B8c*LUy5|&?E8QjIEa!h6V#TxVy2InWbRO33d$R8{mz3N`o
zk)zeskjYEqo9B;`|AvDWNEGK*eP>_rfu#%%gVs;{#lEXB@%Xf7D7UQIJeCu=jXDfu
z|2Z_v^rA9^%9c#ZdRbXn%{h`@Ht6XG*3`-cvuN6D&q|-|mZWGm@FeepelnbTZdf6%
zr?;f$Wewl1p$d2Snpy^1avYE|-*xXN1PMt=(ejnG8GPvQwQqWS{ACt_Tr_ooLW-M;
z0^HmIneY8`eEeRMN~8b}Fijg9Zw2RRT2q1e*m)eawuAtc355U|@_Opj=I8@ZpbL^1
z+Be;W94MHWN)(gcn=^btRQlg6fVunLiRtbEGXH$ncl5S<2?z+3z|!*_oqtwT)C*K~
zJvOfJw6x_LZO#btM{!kE=gajjpauYqG_W1dk3u9zR0er>SEN%FNdCgY!XIJ)C??>>
zPt}F{*-DP%?Xo`w7<L1z7W`i!iYM-&R78l)a~}}86wS3vWib!~U=9QNwKckH-eYD#
zk_?!d!SEP>j1iPkAkhac4>H&Dm^ftpaQhwh%jn3!B+z6bG0ONVXa8U_`<k!G_zkpX
z`horS$U2_Liq&(U()>~EeU-i(e#2DksW#KOHX3@~wB`xpi&_qna&_Y~g&`z$14Y*w
zqz_uPu80^fRJNlS7?8p>X=(4|31&M)o~)!aj;pC@ZA<GrH%>+{xZp{+xF-%4e_qur
zpWNx-hdjo_uN(iU>+I+h7NmO3SWwrn*G4!`Lqh||>?)wk<2i4Oi;J%^_;<FpzEDVm
zdmMkZQzPK&%EPHkOngUTLHl|M5((c#2{0P7;GgCu%AXksb-O&!#t-E-=|=Jw!NSIN
z`eDe5^Ts}?_$+g>;19Oq>^M+~3yU=fPCU4sZl)PjCI;f@LKMS4&W|e#m;La#yVOrl
zPX}lL$Qds}Luu_6_YMy9W@mbOw##!U+uUptRJ0Xz<{#HYJ1@q^$Nm5CuHK(DW<Gz4
z0Iq8&66rk+Z2RA~5h0J`_T^>o)2;rfPuz5$yge4C{NWIJ&e}K9#qKafqLmFEJuAtM
zj)Al=$^BC6H7ayi^kb>jyNM;!Y^Fr}6{BX5LDTD@F{6w9X+s|t2niFjZFU^z_8iU+
zhuHV2=I2lPLvG--spB81cW$0!{={{0G3utv33NUUlQ*kzEj_9$&dGseRc$Y2#)4PV
zaN<h^vTWx7w@88b7PB&)Ux(j)v(iAc;<yV)l4R!@8yQ6o%u2JAk8L_d2rC&hJ7`lO
zD$Un>imYM33AHTEk`j;X3+(w^s%z;yM_rMJveag6Uhz}A*`qQvG}KY4JhYI|n_1o(
z4<%wf2Z3*{tbOY|v><XKx)dY+;*MI~^GiN{R-rc^_4~$RPDq;llZMkwho?W3sre)g
zK_V(s<Krj0Ey09@gsQ{CKob!a^*1)z@Jn0dnz!lf@I1<vbP-sdp1$Y|L>)a1d2*>T
znFNw>i(W+PPs=0!O+G<Nj@SHqZw3S$*8I{}e9ax)-OC5__&!}9?@bn%YtAMjU&EE^
zqVwHFP%2&2)TVpO^angcbvlw_xpaszsecM1mW}YvtGp1W(FA7W!)#fGpgRjAqxDy_
zej?s~a*<lYlJ)igy_wknCRxU)VTF-sS1q6f7#|-#wm;k@Cnpp0+TG7R+^2W<^w_R6
z--N5ZS?RT!%vU%(JghF$U#$CtOoy8igl?P5!}PfWx>+nLb>dG`6m~n%dVl1zHvSx`
z(|7!S!VHhExuvB~N+R{zmQcDoIA6&c9~C1_F}SW@Emrl%VqSK3)rY}&W`$k&u8KUl
zgrK}!*PS0s+Euwx6u1bVPgeJEQO>JTx)Bi&;+b_^USzq#Io%0AO_euhzh7(jud^Q&
zY%w963K8EK5>ttl%7F=oNyC5(Zl4Iu95_?ijCj&e>3FhMenTRR9*g<b8vw0(y~AJ)
zC!iRDX<uAYa(;I9XQ{#YL7y)igclelsS{#j>4Zl?@&}-D!=hp!(LALkJLa~{&Tmc4
zW-mPZt5^KKO3~O{r=iZhw#vn*nyZL$nR}Z1>(kX_G#wkyXADR;d$SiY*oljq9Sy`3
zbZV@oj|4thqhFkrPZ;pm0pS17vb;K4jWA9G6I>8Eoakl3V2?><a<a}t7=VE5Z&wm`
z4<a?GQe#v}@bP&DmM>0^MY-TT&l`8MYsUd15uP$OFjxWl!}xcusS#tKbAvUX!eg8E
z?J#u0={rS!BOgqA@*R7N&uX_3n?^YCWLw-#*ZnLUoNpSNacA(rr2SI8qsNUk@CG3Q
zw{_X$haXZ)M6C$+9GRgKT|wy9vaXn~6wcSj4i_<JY~gu`%M4aN<d*9(4sJUUg<XzY
z0w?oa+VfS#{F|CiU|*C1|DvuI(VF-ledtMe%pbQHpaLPMI-B`-`uai^W0@*BQrwOx
z=*ApL)G`b!QhdJr#Tun31H*`9<@O(Lz;qL)+V~CBg)`KU%r;D<KsMKRN{C({3P92a
zR^(!hHLv^Ol0b(TOM+4&+j@RDnVRNkC7W)gNubl6dGi%Dw#=qKCr8rv4kNnqCs4^j
zHgoWbBgagSGc(K@8XCh|WqB9_!Q#ky>zUX6%5=D(lqZM!>G0$l<&##F7cn6qXn-kc
zSKm@!&qqPguAz1F^T(IL_j49M+SJqm<4Oq(?CJQ77(ph)fc>o`ZSnT?r^K{0(uEE@
z=x_(`%gwp#^dt-f&!)Ts;|~L6<%JQa7$Zm}rayi+)I6SymUIf-^Ala5p#d^TB8Ym3
z+q!C7-jK*yyRE==9k$F2C90^ZZnik#l^7tp^oF?>XXt2y_QB1eN)t1GjCZuNQ^zCd
z^~*4U5M$)e3@|5gv*@cG9Fx$J<TakwuCK4#9&hD<ufEa!4c5PXZti-k|LGkEz<_*T
zix!A~=1*Ff8|2gRnrWz8TUrtl5Rf!_3bM104^C7?;Zi|shlf5^cfV%@bO__7oluMI
znQSAe=RGe;u5G!I;GZwnC8nhKa(ytLq(p*ElvnogK|@nY6NpMq764HjV&as1uGMDG
zYG11&Jr;Kd2NA69*;)6y1AaCazTCVK2U}HITc|zymE7S-Dr&OMb1%2##Kh*9n;gf}
z)6>WMGbMOYfGdvH){aLo*HE>q&B6?&f~2XPoSce;*YW{!hYPobpvP7yhGzA2A1?HT
z-_!ewnzoBk$(1N7)qHssvi|oR5#++@iHUSrNJGQJ#DC6BKdG630Gg{WC>#EZ?#lCX
zm&Mh1?oj%W?B`%aY|w5F#4&om&+Z3)ib%$Tm*?yBurLMh7yeR6C@2f>^N)&VKkf!H
z2H$xVR%$hHcgBHbyWFHHF21Qi@Jw{*f>aOtK^gyr9PG8$qvmRr8k36kEDR@ipTe+A
znbq<o93ln~O)KYZ^1{7$M$&>0n3<;Fr#Wt}V^b><`v$twg?%v;M>1`8$0_OQ3myG8
zTgLXzYpiEG7iwH}1;R7<?vBnt+$X%vqvmldsqrt+dH1W$Naa&tdL$#@U=>sOP82nP
z$61IMM<80iw>Q+}*>_bw=#;v<FD#jKD+W?St*-at=YWV-R5%=8`~fXpVX}&sHZLzP
z(;kvV?6(uz*xXD}S<4)^XubXTBAd+b8{vJcfZKGV78$F>%6&IgaGdM(DymF$o&72c
zJ^dblFWv>FfH)Qei`%WX+8!*|wK~aBV=qe*W`cSg%h>@VIhn-)DNs!FiJr|C0Xn$M
zc;|K|RbB94VN!A(48(wi8dF;NbOaSlqg8SUqC=^@grJt@^e+gr$Rv<R&Uoy_dMo?>
z{=OPsojn|gPu^$7CM4L;Rb*9EjP>^BS5$zNivb+8+1a}M{NC6>7{8soZ7z6~hL0yx
zS)IYHRLSqovw)WwF$x3ybC+o#i^}=zdk!d5T{CN*@cclYixxFXW*A09l@zjKE0JYl
zUq3U5)3n1HxVrLc+iPhh5~0T4+im_PRcJqJJza!H{mQ={Us<4AN-+H4)3&F#*+^>A
z?p{({9Ny1TQ*YrWP!{cD#zrP6H2LeiWguCG3LkKbK5uiGdy#_hKqM_5xe`Nz2oFzz
zbV~U9>7Tbp%v(licDe^BxcIvDj}oXAI+SW0N>cYNS7T66ROI2|vFo1Cb_9{Cb%*QY
z!M4$d>1tk}yg|E@tPBiA&leSpjMkpRKf_Ta*tw#nu5TswkBU;!ua_V#Dk{=<>{+gQ
zH#YX=*J2sYicscrY#f}d7R0$ZSm@`~*NBL#<Kyet8*kTz8f$mQnIYh4z2DsfoRj$D
zM;QOEVz8_}>pmw}pku-Q1}KfCg@xADWE@uRcj|o%<-GA-;fP80XaBh`OuDr;$QjeU
zJu~adTHOSOw$3fz1nVTFq)t7W<&@HmsR_@ijeRo?XNdMqNo?-EkW|&wY@c}CT^-Rz
zXQW)57&JJ25O1pnMKzkt>Fh*I`%H1D`8j`g{`A?z>F~<RQl$6Kfsn>7X#qH1j*ej_
z2UX`l5$~Pbi=-Xa!n!{dX2Z$(-#-CYURCqqGBjto=6i|HCe4iP3nX~W*^&hinUIOb
zF(0*BQah2Mi?ITq)ZGoAZPH%DY!rg2UI=m+?o@%lfz0B)ea4H>=NL^*TsKaE<m5Yg
z2r^sWdQ1!rS8vHH{;_bG$FbgL?^H4Gl<u2<V6h|!nvalVDPPa-vWs8=jx)eGIYK^!
zo2^V2-Ygq<kpe+eQ!EC}e4z0}q<;Xj8C)rl;y5`u0a3uoU<_*yL4d6(g#Hc-TNZ+J
zuR_WUZyvBp05FIj+y=kw*}CWMA1e1Zv^)Bm&s0!Qs?Ensj?oG!o2akYQA{2iJD}th
z(lZ&2^?EHD);MV>Ph8{8SZJ~nOWwUG+*%eFv<AY?L}B7NCzeQvlJiKNZ&k?9Rm4@e
zaq-41tj>9TEyjfV-o|aHW~EB`5<SkQO2zG!4R%L=a3@3Xy22jV?qW62|1%FGO2Kxz
z?w3>zVkZ@Ps?!QeJ7yY+%AZV>IAey#BTQ+(b118$`&{a+n2U5xQ$x8Rytf=?Q{FpU
zShS`YpXJ)-3*S-W>UgG+sLj|pmUz!Z>*?W@5*-r*6x6h|G~pT^2~M%-EzpObzJ?-F
zj56y_034BDSNh{eqL7zcy~IB97N}F8H{DVejF)!<#Z1$%0fCFU;_Vek%lSxQ{(}%s
zwuq%nFnvwshWeU@;{63QkBu&S(NB*Qnd9+Smqmj{Gg#5ouD6$;ZJQv`HM=buydZ$O
z%+Vf6!SKOn(Ifb_zPz^^&#VjL0Q+taH#-@jN@)mwwx&{}=|Z|RG2G?EeFZOR`^BDV
zd^NstlFWkjDpkfc@gNL}+w`Ch(rhcmyvpei!9vHK&dq6bQ43X^l^EAvAZi@miIm_E
zWB8XD6{ce<m>HEDB>l<lZ2jd`%|^mDUdyn;BH=EKUqoDBO_;*d<AXWo@uATV^5pB6
zp8FoaW{8Tj&IH0#%h2~aTH2?!LkE)K!7TH<x)EvTLZ;vlv2C;H;*b<=x*?r)GkoeD
zxk+}AZGI~uvAe&2dwc6c_iBn4PBhY0wr)ZRq#%UWKL-W0ba`hvfQnRb{B0rS<m{~Z
zR<dh`6soY?;Q^Z1Xnmqn14-<ypN`7=chR+@nT{HzGiCb4LkYKqXe4|U^!_DC0S75u
zjx$xSGRu$D8A(X7PYHLde*GR!p(+@Hyg&bnjt1=wWI!wIX1{>Qq<rF7g$%I&L+}_C
z&37J6EGymqE|jlhX!~y_ad_71MKNo>XT(k7&<-bIwCv<iij$fYQE?R*cxwrZyFF<E
z8{1ouRYVu|nSz8UOh()T$#$o<bR5pxzXulW5B6zaIH@4yry!nqVMAvUpf;t;8Hxl;
zCG(wH5*`K;3&6#x|3Xphs3#wZ9gX{30xw7tM*=TgbKQemQEi<MqLZ#7=n7#!*j_Bl
zw@^71J^Q!sKp2Z4Xu5EW*JiE)m=V#}@Bh4cFx$q{)X*S`Z&UgFDmE5)rZcm%!i1zL
z$;m`@J}pjy2{LUdDMw}HFK{z<`Musi6p66I^5(gDnMsAx%TD;B4nlBCoha26-L4nJ
z8n2&TgVEpm;k5e|nXO^>lHv7nRu$~LlB|pj@;N&Uw02Uaso}d>{o=)o&*QqYh^8^X
zhsZ}F0jQDBWfv=e0^n94jDeUvY7#H`$B!L*F%@=ja|`655vt2oA8dtu7u_jj{@HYZ
zVANGsf_Tbx^YO1Js<$7#yfoF-|3OMk#p<FWhOucdvuE&80u?N)z`0JQFB>O!otP`}
zr|^{e-hKbX(?bydGA?fbuROO3w~$GarM!EkybOd?c3r#?^fN?xb0;qdT}TD=LBVI|
zEc`2cbMC)YnL=J@`qi+no5WnEVhdvwl8#PT-dSVFNk0op*!18JPcZhV6jKS`+~&(V
zv)wFxe4AND--GE7Vo^X*etfBttIUKUFjuB8NJ^UK-Z~HH(B6xmn9%J|V&N16XOjHe
zw|#zoe(a(6=WLv1vA+V?U3{fFGEW}i6wnBLGbF1gqUz*WqVochw2kmJ>I1hE7&(cY
zwg=+y(iWQb_gvVfo~}CaNcx0XJ;x*3Getj-{9QyB59*BR3vF_F^!RMn1Kc?0&EC7p
zalZ-5egC_StM{0y8q0%2LkIKK!22-;G4wjgP-91m$gOWTs1u{9wLqtQ<M^6~=Q1<O
z_$LQbXAkqk?X=X07l>ay8;v%Sg}VEMPOj`hkP=j^aBO}pC9mC*w4&l6AfEq-g=Wqe
zklKmwuEdVsV0jh;=S0CWIwJ#flN5q<>iQcmpRCeASU6SC9Yb>bIhE9>ACDwb(Uwk<
z(Gt<rQLkh^uzskz3V9*lpvMvelK?kc7EnxVVbQ)GS=hJ`)F=Sx++C+^vwowCr<HM3
zbhHV#lU5h5!SGiE(&@|`4Z|aNTzeqphL8u0jg7H$^-!$Bm^|ZgS*&$uI+UeRNIqXF
zAqE;!t!5eMFXR=K<E0D*ZPK~s<{}0~PUR$hs9&;k^3<KL5XRf`QFDO>BDH|3DMb2A
zbjruBsGI!D8}C~^71bu^opA92i~)xF;Q&3Bh>(zt6OXW9+6t`_U?%~tjRViVzxC}(
zgABlb1uTrUwd=b9b<KDdkbfN;fpc+zffj1L9`8MY_4xDW&)ch`r}MqZ-LDS2Go+6#
zGxg^VM=L@E!h~}+9{LWday$9y&ks-qIGXF}HI2Z9|8s`Y-Xe4`=lbcs9(<oqKV>Ef
z1>H*g{X25rMl2y<a&dBoe*J0zc2)=yuCH{AkF&qLM>yYw;Vnx^NdY|-#EDShVWXtq
zb6RnICwNBS<2{iJ&@Ka&rDL>wT>s{)36F1FDlZk{2i~JE-JKJ-5aSO9_53tuw^h^9
zV9FoPqy&Py5ka>^m!H4i=JDx`XrD^^MhOWg&)=o_Uc?*9qT*t5VILb46BAooTT|1?
z&QG(C2{$UKQj#xKI4^s*onB!hi(-)rpYQnhzaRMX?x_|jQJyA%LYBM^=qCzs8rs%B
z;M&*D5=LK9dyBnI42Vw;YL?E-YYC{c#Fyc|&{G$+=^lPwt3fStg|h?CgfE4K`WXWx
z>1OPi<79zM5!+9j{8i#*VNuaXTEfq23~4;JL^rRq#28ZQyzFr_vxVOv1voS;vHFVj
zn4EyHioNc9K!q7>Oso6R3hQ3jZ1Gi!Q)XbG>d{gvuYLN%IiyvCFO5A@B8`k()|3S@
z1HeuM6zy=Xa&st&Gmb})C@d&@S5St(4#2ITUO$#)qeq*$2F-a7qp`yvdihvfeEg!q
zLV9LaP`^#jf%h+>C&<Lb$HjFnXLdyyPQuS|HNair?&sycff=l&{m}NIAtVGP$XM3g
zGfDnmzTs^I86}bv9&N`0!r`)v0DgQ_6g%Hb>>vz8h|g>PDvFB1<zmyS=cLAD{K?mt
z!}6<8%HJUk%VA?d=Q%eUJW)8ph1Jz3n<r5K?n8ff6xApjVDLT$TEak(CToU^?E~S2
zm7-$!ChKOV9Iv$8GB6K;2Db$wR4lB_%-n2jSfM>NH8lhB9m3BCf_MMB_soJXnVpms
zilqK#Oz@cVQNowZMWDsAe8Rn;zCMv!25bt0jxR3_gnUOPFl<5kT17Az$f6DP^`E%l
zu$hGu(g`B+si~@X9|@@=g69P`E)~DD8h2EipY(-CRMa%a#<RCin5cO1(g$ItzR7!-
z>=xL(1-o<ZFAtuP>azwiQB&vACm)0G<L#3p5?Dvk9@5g%`p62nk|4Sri~Snxe*0}U
zZt>GhCei!vLd{~?;n*r7^tklX$I|C9A@<jP$A{H;8UH-ka#SvE?u+yDaQS1wRv+#F
z3ZIw<aJGLC<a5UNO;rETWDlj>jO6AkWu!NK^4x8kMUxFN)YSYe0!aoE5fP#9u;7P+
z9Uq@aMYBqn-Y?(+i-0H$L5%X)Ls>a9FYhAm@NsFA*7%G^1k1q@9RtIskMnbn$vU3_
zZwJ5gZ^%#Ts8#k#*TF>qIG>9h;+cwn*i6ptsYO)2oeCoa@KtKtktkZ%hT!P-Xax5B
zII<DN3PF&o2Y$9`+k@M>YgB}PtIN?y%6444?;U&#%h>Mp1814Ig$12FDx@n2Y1Mjk
zG<%+by&gd&ccf@&wiEB(O>v1R*hZq9fz<tQ`u4$}6HiYaT{7X2_VI&G;CM63uTjWU
zy&AUY3BlvOI%2cnYOeVMAj;D6vSrg%TB~DDs*#mYnDwY>TAKsUu>AzcMNCvMzbyOk
zTFHo7C`Y|HEZ@u-cg77x@a@vCtuP$Aa6x|lG&jeqrcYE2B__H`N|16^&PP2vS6A0G
zqxCv#UYliPu7%;<AE0rAERT=N*5h4((V}AJc-d;$Yd|1D$mY+{!^iPKplq$UJ;AJ9
z7xdQqP|7|p^7Qd?r9|U<4D=q!5PrG@fUayj*k2I?rPFb#udl#PXxU>FyoRLO8*yxJ
z(ChLC`KD~aWE^)tfbYO0p1qUTaECRl@N(~32?vd<#hVs1VkDE(>>Hqb=rwzsw4uK6
z50Y|+&8YT$n*FFIgDs7X1<!^i8tFa)F#p|(mwn3a{<D<C!<Q<%d+K$gUpg!Xjhmd*
z^yV2xZ185D64>?yP%tnIeAexu)8p@Z7o5p}F_V{&bHX|7j%67pFu~I}2UX_)pW0jx
z5ndGPTam|?j;6jkn1>^-gGHCou9Ya9x!oDaYXxRKG0%ajV8M0kJsm7!&l4^{K^{Y&
zsUBJ&qwSVmP=+8NJTv_705LbmojaF;NsF{FbOimT%O`k@fK2l|&gk@U&dFFqCJ_}$
zM(d$MbTccMqq8KfHXm;Q2&}C3AD=l7!{2`|+9y&C6|nqFEn~zc+HgHKQKa}-Pm0*Z
zW>nm4o2{=Jd6qYuZ7CroWyi=8sIUmBq8V0IR5<V0>(y)x?kzXf&GyIeG7e-lehgAC
z9vyis?C$Ey+3Izz)VeQBwn~!cF<W$-YbM=2NB2EA!orzk6D$A@9AHBdFUNyTEUc}p
zL{<sF>VV@R?qZ{*Etoha4&$=K51>&%(rSUVv7tO1-=Yd8=I4+r^H64NWi0~1zL#nt
z&p-JvLm?DyOlE?38iKX5<}l^-g)~3`Xt+1#vRQ_KUw_gylHD008qhmEGTjGT+GrK2
zkv6JTdqLdN_KX<QcA1jOv@dZS!lI9dVvuWIuC*z|6_@I(t6z9NsD66uQ{|2?ISBS1
zYKTip-qa!J=C#WD^=;yXScTm*b5!Pz=)hpJbF!!393a0tqmhUin^EDbRpI+dBaGkl
zBP(lteck_XrS9h5)0Q)(<pLKQDL~$O^V9L13h<;+cY+QVZpNb~_s+lZ%w!cP`V|u1
zySg3w+pF~oFfyVZ035Z8HpWZG%KC)u)$I-@D4C!e6D6gnx_XUIR8+5wqT;{JtjWm>
z05?xhO-V=J?_~G_;Jvg2umRb<fh85tzslPlCnhF{gnd4V6KeywPLSevDsFug33e%F
zxvYhK#rTYXzQ9HW1%>-99em@_aJ^Y;EBXlofsGqS0O=_zD*lOU<N3Z&s%*r|P-Lf}
zzFLa6Gq}3Iv5SV=9@6e*Y2~tUw%|DT={)SCFtF=(JD-8Hg0OI#lvH|L+%q>Im2nXv
zP5Ivat(z&;jUeVdZ0kFFIwPDv`IA##p5cD;PE0HS<TECo(^h!ZDFkDP{NagAgkFsJ
zhm+mA*80wkv+~PP0U!xBoT<^_h7)lY?Fp-e<R9nog}qCX5|$0^=K_#*GCDeX44C%t
zrrEnur$lexzIAoIpjXWYs}CdunDy$%a<jQK#k~L#aoe~0eyw#oNd3WuvERhF7Q^Z7
zi_Z26Z3OY?Ex_af!E9-6PENE=rS3mIJ_g&n0rVmuY8CuG+N$thMFjkokNwEE^Ye2I
z)E6Bkp2|8C4@YN5VH-Cfup9C9>yOgXi__EF$S0IADOF9)r`{x~IG}0&L{CIkH5ZZe
zjQFEDn#zwuLehM%PIP`6u=w1Xd58#qOh!rRh|L~GJ`Lklk2mlQz(yBAL@aoAlkTrg
zU2i`(3LMN|P+P3-1nT|X*$EB~CMF~VmJqjQYuG46)lZ9UnOkwGSbCi#5z^+JF$s~V
zX{D+2r|a1vvgp?MV|D<H{yG7nbIVd$Yxm*!a6E_n-X|eGe&H@Gc>aun8Siij;i)h(
zL2$P}CsJxM`I|&GaK2d*7QySTq->(7Jwa3!)Y88YO*V(h8)bl@6y)Uo0I@?>R#sa(
zsjluy^OX1%k_iY}x_?4|i@*%VpeDh%p>{vT6#YP?4!m(;bY}52t(K464=S-qe!i5i
z{wAzyfDihb*|O;6Mqh|mKENvEewcc?tgoR{q^aHe-S_+N4~)dvM9EGs+=nXBZ%n;H
zyjN^dFfqx!*a$!<1n~+ZjO5MxUy(t5uRTw<&i6!Q{ImZz3or?0wrYW*`o{$yegJ$Q
z``QWMVe5UEAMP96i<;<f!eQTPY0Zg!BN6a~g=|)%>({R&L-7L<%OH4|=<|j8OE8L#
zxR60ZS2Xwc@`C)(|Gc|8vc1k`rKbm*qn>}0$dL-OGD%yvN0W)Br>B2SR-PqMoQvB<
zB~im`{l(q7hx^*MHI+tQidWsvUES_2M$^eH1yB<Whi9P|nvkDsaWw`aZ}yf4aF~Ip
z2%?`ReG2f1Zy~?9K~%+z+sMUyWjf&Yd5w+ZNMyq_Xai<xNf*PcT}6$n%*N)&n1J0G
z?z7wU7`EyKQslP0HUaZz5cJai@8mcMztdI%$npPYlSogUI8Ds<QoY-XI|L=7TqlW9
z26rW`+1O{%R4C)jHOte}(Xma;no|q&hIxw=(;-44SVJXoouxu%X5)JFqP`-EX#&p%
zHKEXF>G~t~+D``l;(MDD-)X%cZZCuhV5p>EAc~(~zkZbp$@GES-$_z8v9Up$Efc2&
z8x%fph2-bcY;6a8qsPT<*4zi_`vAo8wG}cpv;o=GlsVbAZ_NSx$WW8p)&-{G!^1;4
z>Qq*hiA>52z<e5Zw9+0lQ7gFLUVf=T#<pCRjgrf8-Ee9wcKUSgqF^D7V_gW`oUR!2
zwGEU+SO&&7@5xn6tgV}XYz?BJz4>|mpQ(OrQ($vrQ#*kYMkEywZ+nCsKFnT%PU(&^
z*#Ew2>>N$22?i2Tc*yX5w3MycvG<FQ$KTp+p`S`fH=CH*_KNr+3HICIu7V^7+s3fh
zyP!=pgf6gQ=l7p%QH@eu_$XC<(|56>4Q?=nWw20mhZ<jvMtv>SREjzG6IGW<Wo9kD
z7E`2j>lpls%bX3*)OLSa;cut5lm-EIcE{ml?kk6?HX>i3c)7k417Fe{3|0GnvG&Re
zPiOF_0DybSh|k=p!hArr#=KaSXRPUccbRs1yx#E&6{b&-RsWPT->?Y?BCcMv$As{Z
zuEyiOraGyTxIL02@0aYVk9%)Qwd7gTtytCVyf9K^#38>?4a<+)wIISRA0CdO-H7FI
ze^Aw)-)#$wiu3V_ytfG~@FY)O(*siBTMm53Gs`GnI)f(A(i9QDZ4C=^PwUnEX1zG|
z1mvn~vf!3A8I!ItQ*H-?^o<r_;s$3X_`?F|%w4%vASzQ5x&tsqu>1%J0o;6g1nt7`
z(X}igDa!D!mcfZ1ij>dq?9u<d0yB4H=t>HT7Q`_?&iL{E2K3p-TFz^v4EGd+mXF)~
z=44trqEb57`C%+9s}}k-Kz>FaugokG_=@0rf70hgWf)5nH#awjPR#SQ#{?5I;@`%b
zBQ_BIcXl2TQ&tZwP{POG0tr_h`zQ39DXYrWXiy*{OiZ3Q&B^gaCbm@GckyPF(zGmn
zp?wz{g*+w9ojYf3UNu^;_`*TQK4VRS?cV6IW08wzvh)}N3-*QnJ)+D1N7P$JRrP&g
zqX%iEL_kvb(cPs;hk$f<h;(<Cl#<d7f`EV^(jkp>iiC7XrwEE5aVP)xzV|Zx;21g%
z`|Q2fT5~=%-CLyj!RC0ax{;ZByz~wn{!Af1-sf99@kYD>{hGS^+WtNpJ_bI8&Xbm*
zn5K9+I)u1u<FEPyB8P<myuL}9r%&;)MX+kNBplewb+!F}T;Qc>ZnZX)0*nv&35A8g
zOE$obJaS-48v+;&GB-4rdXN0c2L}fsWn}rQ_smmf=ToprxNbffpA7pXhUaz2^$D{`
zBP;GtaEQ}HgtlhP`QE8uiQgIRZH&k|^a5~_0hs_=_eNLqI764VLicYqC|vxNHUHUu
z{krsIs_n^?W0|4AL1;t|)zcN##H=h(dXkrxs6&W{o*v{hEWMg>81g<+yAw$OK_bT$
z+&}>d2%sfO6XZ?Mbn9uQAjE(D{pCqXG|EADMT-{xfYUhm=~HohRV~zj(CX@HDzUNS
z>_={0=(r({t;$EsZc^0F9v&6*DVdqRd(-UT69LKN5kcsCXsf;^lL*B<9bA8Vnz8*I
z_St+N)VPeBzNG+kK9_nVM9?9JpZIrjBvfhq1%qrYyL%05^^Oh>bsP%@dtcOaa;BQq
z)D{dVN_1r?_L-|Od4;{suI>$4QNDWfCiv)xQ;EY1@CTn?OaDYwY8d8~bHZd}SHIu;
zcHsyh{w_$EQ6Jz;8AcPcKVC7!lfCr*wY`dst20&o`+O}?#}yS3oOBoQJ!!X1J9U{H
zGt2c+o(ga6N>B6yI3GRS+`c)ATyt{VfKpS!q<fh#*_ql`ypXQy$gt0?8WFtdnldt9
zl37>yn)8x>g70d5;J#=c-qvVnJl^`oxq?F3_k?P){uk9B6o|Y9P=HJm1IZiP>Z_=z
zT^=VSY~0|dD!q&C?Y_Tl1JEvfP$W*d2#l2#gCBdWNLpv*Bh4+|fLKRFB`?m8^gLej
z4|*25{q@VLA9;t1MxWi{kiS-O*-{`n8eg+IQ@H*@Ads!Qe*7W)brloNk)r=bOH=dq
zt+KM(a+R%E@`7zCFsojC`PzABci(DkaI=Kw)pULsa;dhoHrl^GzrCW3^w^rUgyowa
zs1B+U@<aLJM#GD+Rx`A2vv~LV^?;LvgfCl&s)QlOgQzHR7Nmz%G<nUfGY2AcbSMnU
zXjTub7u;1=*pQV35eQQNQ0mDAUZ7Z=9xncK(+OvZrG5qKEyy={98ZcKm|1oI0!*+_
z(N7+RU`rP(P^xP_znmZTI$?34c(>yz8K{!~_I<*5s@{40#+NT&dd+(!=l^_in<+L9
zyu}vRq+W)GuFfoXOcIju8<Pv&rKfmx_%;mrD|?-OPD+Y}jm?1ehwo;B>704UO2}~^
z$OL*n|LF*mb?@OA7Y(|;%vMDai$e|W`{!x$P(b>#owUd9;L(SfN|T_9Pu>wupP@=h
zU39dy#nRJT*}r2;RD#nyn^jb_0ghI8UkuzxWSQ_IvY;0h8B=c4JF8`RVFjZ^X}N4{
z%_b~Gh>i)YcQ%h7m3nN>i#^3|*bK<Gbd&)k7rj^++3LYtbj2)@cQlxh85yK6EMAr}
z_rEX-FL}pYAjPyQ?#{%_Yy&$UPEijB&DZ$Y)^wm127MjN;+EAIEH^l>ZV(Y>h#O_+
z;?AGw0!uT6@=5W}G?6;u#jZ(2&AO%x1gSaCzAW6Zjeyz>4-6DqEIG6JSFlpcSSM_?
zbahEcNW57@-lY|t14~~~Uhb((B4#t~&PDuNko!<XYRJy)lu4aIOjFgp`BP&#YRiL2
z5%Cw3GvE8`%>vjTVsvwjmm`<_v<=P+yx7@TA81Io?;kmL*uY{(3#FzK|Db6blPP52
z2hCc0)^Yi=owp0`Zzbu_UJy(mAs)s@n(3mdL=}|CCo$xJwf2)};u;`Lq?2pOJSpB8
zkaGbf<x(Yygj%xJ*~6dzd`)Z2ZkEhNK0`Wz%zT1&@*zv?D8LF0D))cdW0W$7a)w_Y
z_fa9<s+Sh3?=afs+N%97S7&A@UFj;2LY3j%IMR;$&O;h1BYPDayYMJw-$5IfmiAqG
zI!T_|yL=CWSAnf{LCe9xQKR+uqI+l(`Vq?-{Ksxb=GG}uXk_p-P+8eY8GX|n`X_BU
zLXi-$<!pc2{q<hAt_rn*;&~t;0Wl7NFPZ#ZYHjgiucM-)JG%(@Gz?k-AWY^o-FaJC
zU!M#m27xH~p_w|)l`*VhNe*2lV8d>%u2>itOz(S(LcIUe*&n?hNto8pM1~+Cnbzv3
zGJC!t)4}L|{(f(NKimCc5-j973^8>L4M8^_3knLrUk+{~95TLsQ1gYNr+ynjk*Mui
z`K9;eC&U#grE{5Se%J&D-Q}s1bf%?9?8?#n(&#G{<Q5Q`3%mT$-X%{gK9g><hi+;k
zF4|1@fBFDN!>sC96@u5&Moj!I{z#7=Isb{NtgLLKoxlW`Ovsl6>+Xu7+&St3C{niC
zeTJqP*ezddjbsT{PCSiy^$G(QcXVK&^z`B2`{JD%8TvqZo1FXVam2(!OmhhX*;W?E
z!y_AS-Wdtzi@%?Zh)`v+pZU0mK8uD>gv({tX)F!#$}$NIIN6N-*e(7hKK~gDf5a*_
zyK*L<O~R`^M@v`H|IXEn;_(JC2p-Muv*yE|g?O`>>f)HIWBy}tVd2)cLPd2OUjTn>
zc_J#C3)MBO_bedyX1U4D%F)qLT)boJ;=K;uqLF$&(=MRN5L#eAS@IPgps3!KrukBf
z&u3vzRHf*Z4!x$WF|9Rcc*I6&iD-_8&c|Kva^}hTtl#!LV>zeuku$6{F?^G+j*P$^
z*J2{EV%eo$*{cof{Wr*bnUavO${Y0Ga5@gcS&&VxpHkG{n?GEOKjNfGhiE%73az>D
zr9b~bP0#}R>78H>WnA=vcSDYSlPsI>(D$F71=HD}NzE2iRBH+Qy(IZoFM_DGn>noU
zlNNn$X(<qN?H}+o-#{o1k6fUfhLssa;MB~R_6OoD2g8-?ws9U_U8W;w2n6F6Pz>Nn
zZ*l#1E?#a}McYo@_(@md8^Tn2MM|dEd_*x@lttv@)1%Y^Yu`kA`tQH@^N(J=ak{4<
z^rq5sxoIz>@yp?;NB#Bv*?b+bk=U3*@6q4SGTZE46EZ#i<DHW8gM@tktpsu_TJ$6u
zNUsB~hd8Y`B0OKzY!2rb479#k|Ln9dF)`+G_?HEbZo<I_bw)vf|J@fy`)MKGBncYY
zwpFFp7d5kfj4EWpZe*XnYh(&c`_dDklw!Ec7xi*w=gTx1wgE^Fywv%{1;7SaX+z&<
zIJwP@k;L;2^Qi~=GG-0(p6neRUa)X37rejO@(s9sXI>t1l=`!07JuLg*aBGLKHep&
zN47uj1RQ>L4h}}1@)qznz!(8BciS9x@$%vf5~EWxVPd|b)wxvsbGS;0a+x=*B#h;)
z)#B|6`UtU`%cSBR5E6PEEnV4B87COCX!+xQFgfs+g5O8Rv|7TAeG=Oj6EM=|c`r#G
zsrxx<fzCaZi>b=bYu7(MD&tn{+ao$|)|=X?Wc+{eDjQ%%>Q{hiVg*XqOhG46`|n&8
zUUXtEYhR$AgEx!|5LC{O@m#}cr7YNtt0m1{U0uV&ih_cI#L-&yY=40E4poWl_`<BJ
zyYkuQ3XB}0XU?cfRLo3ssf@~6Wr*Mqv)0&fv5YK%$zC#@D&2iiS3));OcdN9VzI@R
z9Xr;~<oYK96(y8TOv%pHBn#tI>Kdl%m|OeuqfMrY3`H?A6B6CLSq)8T*Te`6Y>hW`
zh;f8%)WmPjGFa?m@?%c~yuZhMS%-KFJFhKhXwR81kE=Sni3c=6Q~T|3G{KgE;PsOy
zPekS9A-(6VgI@<17dK8fY2KwTYf*{VuP;jGNtRYG-m_@kp~!n-NwB)x@@*+1E_xA^
zXmAfckQh9FPUFoF`cD!?>$xWk48>8x()d!L1h|Df=~b2I|LS>v6d-kLiYgU~DbrBq
zfCSa{aOU&OK#icrx8kQg9~*DHzcdYryuU8nGtGVCbLR>zxE}=#8y636=5W+$sRdIi
zGPED1*<+cA^!I>+Tq5jV-shzkGyu@T%K^01grnvw;yU|Tf{(^%b!1D$UR~CrH)0#D
zJ+`h&%En2Jg++Qzv$l1QLerfr?(G~_FZ532_JEPlH;w26@u%~vxQDk}4r)9Tar6;H
zuXOGjdU<(yKg8^jp=J(`TwPjP(iiT=8X6r19xGRu?c=o+?eZNy8^Uv}G!CPv<{+n+
zDX-rdVj+?<n#%n+2M=$M+W`gvZww29DP^#vq(s8g6N>fbIOrg-5(zqsIKK1*8V0_N
zf+HQBv*4}Fv;N_I2Z!@xG=(D9)9tS_`;_^$?u79*uu`i(gX>H9uu^Je3O}`?@w0`-
z)x{j!^H}R=An^3Eu#B)bGza^vTX#hDe9hFZgU?#%8V#llQZQxGc~aAk^L_I}zdb{h
zY#}6GD}abbGMDw5UQ86qn><+nLiO{Q34&d}tNHiS_t%Eg_m=wmx94`Xpu0l{Vbs=G
zU3K-upVxt%px3w~;1trUWV%r{#F{$u)>uYPDP+@mlEk+?m6U>F7+ZvYvPLw{j4-&_
zLfFD!ne$IFXJq}RGsqiJ(a>^7Mh~p)AVA3WU|KO*p$Jy;qtxtOD>HDYSI2XHFaep@
z-*=4TP6$ND+>}dJ=pe^5<VOA7yrd8Q@h0QZFW2pev$kQ|H*<WE?-b9AMtC)ME1J$D
zv7-1`jrOYa+I7=^3lda|vXpXLjb;ZN2!|x`$|?9=R_%bp-{i^Zm)G}I;wkR-_P$m3
zFlfK{x!SEIMj6X8x=JEo|9vXMEo2~Ut)LD4&n4)3=CI0HPOqEZ6ItU%-j)<BC`m&`
zd15wO8~Nu)G3o0THN^Dv^cBXf`ZNi(Hs2yfELjPopitJn;pBC5I}ri1=GT!?pC?NI
zzcvI0B4-xo+mq781w9V!>k#MhTMxKhfxpL>bHZMN-s0l?9I1mm#rK)xb;3&;dL`sw
z#>lIzqKrJ%zfEgKpw_kPnk+T*|L=vxb^Yd96h(5e&G`PP>4WIW5(_ptoR<W+>qD=S
z=?+uIJ!OwFKU=0KEUO|AAt51tyTW?E4q60IEA|$0$6j1dx*g?`Hvs+*r!7P=P*PF?
zL@apsZuIVJ4K`vBPyR<wzwz&=y|P|13SQh@*Od+&dFJHQI4&Xx^GLp;2FGw?lRdY$
zUw5nD4?%RBy;tbC6pfw-nbFZhPVH9_fdX3k{{EMhTbx~Uu?<YS(*-^{Sq(yOi<OCq
zKUk9?$Xp(^>!ttZB1pbjgdoe?iZeT#f|s`jiz7Z05*;#7qu+Yi{oTs@DyH(ax74~1
z4*W_EJq)ERN$<a58_#}ID*MpT(gG2T1`1y#$+SRQ{M(~vd}y3TwoLEr=+i?nL}sz~
zC4bcb9+e24ks%nWY@D3Ln#}**4+w(HgrI$o{U}@qgz55$&Sn0)bj3~8k;mQZg}MKA
z-jY>}ULSuyrhw8)c>W6A*)1~Q^O@()`q5Ekg-Du8g|?XQsT?&vNT~gP^0fM#F#Pvl
zJ{&T0gV>NkC~~}nOu~=$-1zLXj|0>o`m+v?+PvlISy*<VXM&Y46@krtS#L{*A_2^i
z`AX>172r#$xkE|};!&U}eRAkTY-tpJ>FT9Q4^)Yoni@F5rl+U%7aXS2dMqt1d%t|~
z_Vc?9ZQ%Y?_V=$_<=eL%4%^P@?HJD3+HBE}PSaMU|2?`bR^W{ljqV@W49o30IYEdo
zb{H%b^^H-CjvYBrySOVeG4XA<CSomDbiLwa+IO_@SnWdH-a*GkYiu@RV=81S^PMx!
z8yQ53hnghtdR4Ge+n}q{lxwcu;orb#3JQw0`b^1l2)XGTqr$lJggjT&U3F`xHw~z~
zf}t)ME?+h}MzLH`qm$1yvet98%=?RY%O2<)4|b<%k@SxndD%uDW4Wi>TBk!RWm%yD
znsD3n4I>%w6phz1zmi5E*bRUvBi&4T?T^hKEh8FUtfxZT82M-I+^1th#M6_??RYK<
zhg?*bvjM<2BKop{*G|U~v4KMOCeAdOX3LDALIyTrr858#ypa9T-?*p-%wdV})S}`w
ze{*q13@ajn(a*>Q>~)($$r^9O8c3ajxMqT$9Y0Is=zR5cWfyI+IZtk^gt<GaRAIxl
z^kq6PUZWx6=YK!j_r8ZW3u#N$hiHEN9B$4+RHB7{ej1%(cCR}`R_G0$84u~!gr4kk
z3dEZ@Wjy=rKb)EE`5K=avjt(B4Z68_1py7S{t9*aN@~fL(ybsp0AM9Pm6HG~{6za>
zxgiO>uB2@~tU`|(#(Y1+#ve5-V^n8LWHnn2-#|8&3>5`M^RN-k@yn}wJ0En~b*lh*
z^InZ2y1+#TyC*e2>aJg~0xFN*BT5q-=>o|nZXN%!lJr6wgmw4UNHw2b-`0*WM_~aQ
zJA1tHmCw;!3ZK%AkWE2FMMWE}Bn#FE6h>uMxabL~U7|rzZ{(PQ_b+fz#ZHYI%tC`K
z6?n;65hU)LU2*eXbsjlwmyalb$ZF<y;^rd3Ke<TTPJ>{=$w3K^3eWO2Ka+DeAx)Mk
z*dCP&t%4POZ9DxzC~J#bxh5*dKX(_EcUWm|x<C^-ML<v&w%LgM8w!b90N%o=F&-ZA
z=6xdU!D=50^nSpkkk7E3oP61Ae_p&<0X$dt!(}JPI4n_3X429c769@Md}Fj}L!UkN
z1b~fjt#6@E3EAALPN>pksTh>Fw+K!W5KqmQYm@)_-iux;Yaq%wK3DMW^|qUb5I)LR
z)b6Vo@mf2u;OYrhmNJ(~w1>b34w{94AWsr)^<f+lnU(-##F(3F02}*m#w%HG{zk7O
zW6)vhwE6D@@t_zkKG+vl?cv%S%_09Ygw4g>)Ew|3GV<x8npR?uqk8*22J3B{2cvjK
z2KiX%ui8lL>l5CpqFo72r90S6M~r>*jTA$XXzLdjW57S~{j*QHmG`vo;*?E|rM1DS
zyUUT`0g)dL=|@s?YIOo=aZGubM@)(9_G-7c^2Sb&!p(JGcq;3lqa$=bnIf_2s{ZCu
zc=oKDgl)Xs?Bw7tl+&&RaZqs92mYdi?dmF5>|lSt3<LmK1n*THhez<3S<b1a#=Q(N
zejP|+_6gqV=R4&v5Pimd4%=(<W33Mnr*rLA3*W2C`|yNT%}B&WSB8Vt!eS3PeD3;j
zm`IwYPxfB_p3KuE+&CMmh2u<H1Up20JVAYVbP(n4vqsON4NkGN_JA7DVe{K*|Nb%U
z3pJaKjg6?Ns3C{Wa@Dp;tzi|n;EkagCMp7P@V7;eX-XU)IXJZ^@V<Vn^}T|^u`l!y
z?IY{F8JXnl-cKa@T_n6oaK(`gy^me@222sRNqGZ|W0nUCi<0bOg5x(6E23t?4R&_g
zM*wT1=RBGXSS+Iy^S^1DBkI?A<a=5&R{7002gl>|C?_}f<B{{}PMsk)!+}FzSn`_!
z0<5>8!Bo3z-o3LDN3#)diKUtsu}Bi9%8axSvW$j`Do;lM+Cr+JOEouh8=D>Y9YyM;
zA3p2>CG6p1xbF7X{~83yppw}Yw5c;2`#7xrI$N{yrj3=AKckTOPB(XOe(rB{Y3XDU
zI#$-9?tP;SbU|J<5C}Cj`ApHmvr@(>!_lI>bp456zKB7BHlj}<<ZK!uz!5kl^c8Dk
zZhzr!Pf@0fR}M;xUMlg34~0l1s`$54V2;mvdqWJXJja8p$D~ls(RN;h{Q$od@X0{w
zj1CR4{u>}7Bbx&qIaKXYJ!_Mbnnn~4m?NvIj-j7sP|0n%(bnxgYF4qI>nH59Hvf8b
zp7?5*4bhmMuHN9d7#4<-NQd+mvLc_F^ew53HD)LON@sTlxb=*bIxl4PsM&#=By;94
zi=aX^PTp^xTlC!Dg&AjDtG|7J&AV%$6Tcb<(FmZrabm>A2eZZA+5fh%?%w@PdGF<R
ztg@^Sz%Tk>6+jpp8w1hz=g%L+8^~TRkhzP1-w_=jU#amvjsrI~8IN{K`^;(9q8$lQ
z7yY5dD({}_IXK*OQ1UdUT~=j{gnn&b-wm;bwg5EqSz|e3ftNy&-Lt&zC7m2n;(<rY
zg3iCON{*1s8lE0Z8$5G~D)2m73b?rr;N+}?yRY4CpQzjP{F$-DELJk^oomBqHN{W5
zGkX%#($MQ~lzr;!IoQ|;ZERi1KV4p4LWT|)hJbqqR%ct7qVUhS{Awliz(6hzfwl9`
z!94hAGTX)=S_9$*udlD6Vz*mv3&;}<K$t>gABg?I_Y3vHH(!!|^z8zgseF;sEX@wC
z0;vbG@y17I`G@=K-e#&!6x|=`ihAp<1?{HG<IkMIln73`^S^&j59S|8iIn5W!T|uH
zKxnK^|60`Bd@J6;92kGk85Vhym{#X=V!24U7b|%$J=vE{ScjMW)K4R+M;VtrHmq6r
zdC|t^roQJ9l2oLWy!;qrJP3W9{yJsDOPqH&hrBx-nlsN4NA>UOY<W+(fGnO)zLtR8
zov*8N>x4fg^92uS-pxlr382xy=;Z+sMzA120yz>3RWKyFcYkvARnva?`n9}^iwirs
zabWw}67-~?z6t&D7nTepfU71%{@AfusA1%mXC8UZ?E6RUvrmU#tQu3)^ZO`qW(d>4
z4d@{PE~B!=f&x#Elb~=KHNlE`mHx=hl4$JJ(8ijO@!dg;HWkkBd;insKos=k;XgU#
zK9>Z!Qwj2%c)%-hipTDt-JkpCh?!j{hn|24&Mt1)Tz486N>k%QxZlIFw3Fr69ISlo
z^^!VW(UPC=y36HqBR>>cID)7!WUqsZwQssalKHBon%Y$pOiJJ0^HwkMOPn6L9k0oZ
z^w+jcW-7dk5UbY0aZhxH_#N4&PXU+~d;HjU{H;VL`H!_Vu(zV)kimIaT#bojvGdCF
zS(Y7%+?1z7zKy<{ED$_FBUx?>?dVF~pJsWVNOJ1y>L%alZl1i)CQl2dW4^DUmnV_j
zM1OfU(8)F2s&f=>cJ^ds<m_py{(~ncKdd+b&<Cg*>MXw}dQH--Do(TVQ)G3&V&2|4
zVgOz)I-W`*%kxb-{)*vN=)rCJI*XQ8bJaJ5RkT8?<yJAP7GnHOx3P`+ROVQPoPFN+
z`F0jcl^(b*r3{{;{l>>4soKhAck}#c#qfrQSHDWBP;p*ei_F9cYO)KVR*cJceJ#`a
zAa?yU;$-pvY61S5L#`TlDq$_d!v|Q15(2lOCjwsY2Y9!(w@adb1kyt6jqu|+H&V!k
z^7s@T8_UkP+98)1@cYThh8se4#v|!wJkm1AmuC?~lTdcv`UlM&#u>e=Ei6!-COSqa
zqoQi3lu=9hVPn(&<bB*W?iNaM@PO0k;@SvoOACa*33SU>H;cijT>gbW{DEsvOU!<x
zIw>NZ=M^cjrMZ$!Sf@!=NxpWuZl_VxFMr3S;XenU8?HQZelv&6@AQFtFoM|4&F%fa
zfS~@{=kv-ft--j>*7oAHzy{QCBaqjXM9|kShmNgL>sV?34XNxWX-;0A7zA_LxrTG1
zv1*rts-@&|0MU__*^tuY@<X<x`FHSru{!H|@8jpl?I^~=vcbz|5-0Tl04siPy6fy8
zc;*Oo?7y|!ulw!AwPd=;187<fPeRdcEt7RrG36_U2j=rk#F0KZ+qaYsK<^+ct(7*T
zHBX<i8<yEX$~M`egv>5d;={(q7PxfT7ccDfoJ7E3J`T`qNPhQqJ3V^jkjeesKB`mo
za$M&?d+M?Sl<T1X#|nQrVyWYzL-aK9MaPkysS!RG$viHVz1x9490Fwr+x(LNBk^na
zgW`L#Sc#ndu&JcPx@+m`69<N~|HkwPWv+-Fv^eK@%<Sx_KY}WaqZL7{!V=iXN?%mN
zTg|I!i`Zr$WNWB$mXNZ;P&dyHdnXIoKU(0?8Jwj)`La5dGj}IAk#4K!{VdkuIeC`8
zo?dIa5(E%IGy-TOKSP$xv6PM#u#nIwnA-gQ=un;|T(<H{26kBFP=U^ZBJr<jscf%A
z+<zu9+ibkBvI<SQzaoSP{_F?(y9f7i1qB-%*A_nHTZN4L3AnC3C|_oAefxyJNr?4K
z+Uy=emxrW%UjH`_f#xHBp!YCS1S*k|6*iVy{Qz0GwS!05kNID_%LBvFUvi>D&P-=K
z_E33eRs=rRJ_~$)_E|ht9;YWI?9gO<e;fU&u4{7gR5&w?@!p$MtoJG@-}Z67zIy#S
zq<AD`1c~DT8UO3p*r~NqBSXW-FD%3--Q*O)>~vUo{3Y>ktxRXW)60jZexPNHn%ol7
zF`N((J%`Ig<C1&fUzvM<PyoM>Se%xEzXJkh=C@jY*VGu-xZjgu-&trn>6<SsDq3t0
zTCQ_6icj|5o0RT24ytx_ENbOm{S@ivgAlKh^nrdiI2bYV2ck-|+V`wGe_TAC`~AoT
z+j4kZ?I!>oz@x&UUvYZ04-fumx!ok2MA)>a8ZudI1Y73%y1S8I3P`Aky_I|AzbpD<
z=o=eLT&Jl$9CGl-waNc`>Ig81L_UPkJ-^P*D*ID+6sAsN_UDTT5G6yFsI&TZ)A`gG
z>UZCUy+iBA`Yd8E&WX^S{0JyhnwYW>^!qc)B*OV>Qs3T7+C96A5W1h0K0P<nqV)dX
zUf_&<lD_6z`OaND1Oj?wW|=19CyuYx<)=NS)YicYv;D-4n9ueO=ew6yV`ZiTZ}478
z7l0<hPNaTraBlA6v^*oo(K!Cc-(;D)b>1Q}-4M0M#ijcB-=vVXDxVnZ`}D$el0QjJ
z1B%bw>djQ}W%1WPybbGie%-Uy(IGy2VwIrYq{|_=;hbx^CFCo3<2?jTo5SAp?AO0H
zDa5F3o7v`l64sM%825c>zJ!fz3OMl}wl<LgK7EqRtYM!gTa=cJb~sn@PUCyM138=F
z9h&>9OxFvI#jcp>zv{r72zSsg;OQ~ZugUe8nEmJ;dc?|rnNv-}@w(=o=6@&>Wk?D_
z*lLh>0!J!1^{c-0*6suwcXA((yyWCt=Zej&|1RXX&^P}X!+RQ=0HKh=LT3G6^c3q3
zd-*JaZ8{qLjChb9Xa@lv9^ae)Ha9mJm9s&OPLqJq=fR&6c`HsuLSS@>6PwaTZ2G=_
zF*jdsUq^;cV#lEW++jcxnH)o%JU^^^1;S=3NChoj?;nY%&}7C?At8`=KR<j^HkK`X
zHt>5}T7QraDO#1e$Mw8&#&Ju*0*8G#q!V@Ij-+2?OXQhozrdh%K+6>}+&vt<DwH{|
z;J)~XYaxaAeUi4>1c~~IIew~!7>GE<O{zZv*1c#iC!d%6OhbYV24I*?kPD>gVeDmP
zIC`HZhnROi*`tb&8*-X;?0ktz#Yi#yeP;jee<Y<0>}=+lb-Zq#mRuyzf84qShrSao
zEUf;_FC76(Ljk}EHgkWpYrx;v-`@{I75o>d8efLD+osN`z26R?BD|hR{YS7f=2SD0
z!h&rYlQHm|^)82T05W;WPaP#S;U!hLC_rDh2*bm}%|E+SOmwu;Jr=Fe`gUPiEXIFg
z<FT@>@$7T);#DB&qhCKFKVAELY4aem&U7zhJ)v6p>lBU+*_;r)R{PkEfu8;xSILo|
z8$H4LJiUq&kMi;@VWh0Bt*yJeJ1F4FCU)1{pyY#}+BSp2O?CCtxrIvrL<xi0ld5J%
zo8JXj9=(P3_EEC)HkHyh7z5D+-%ppSvh$&1z``#7tk>Uv&C1N-J<&Jcm!Ka|5sh&B
z!E>UAf+%Up>Tln}yvd1_!2c4AgbdQL>6W+<Wg`S4DhkRxVds?&Jc0w@H*Xq`5PVKZ
z1~)?O)~h-xIkKWpE3`rTsi&`BP|Qmur^>`wWF?w5fYdRj+t}GVm>e?aN!u@oL{H%>
z`|TEy-Loos|LU&cF6_;ik&6qxwhM8KaB7~P&JfAFazt<ea};6prp%MWuspZj`f`_l
zIpAy&ojgjKrhg#tEHCNl*8JR9yJvttZ5~9Ir#ZN-_utd4B_zS#{_b<MB&OD-<^c!`
zfG_P<mPk$$9F_XBwKf_7PhT*_Js=zz7*0@4M)alW8^rL*fOTY@9I#XoKS)FRHb2=z
zInN?j-4pW8W$h(J>7-kSsm_ebOmD{|i*mobtm~bXzxVAT(|N5V(f?>`OyY&q2zv!e
z;@_NC_5V1o$E6bR=&gLq(D|NQF&!I`HGL(@M&C0qu(OEEjm?CG5U;pcS(V#BK}B86
zdO`$Lpn5aZg+LMS2-92*F%H-`)o?sWyCcr~&?~Yv_I`Bf?V~K!W?CKR#r6|)N>joQ
zeX?L58T)Xi{$Pti#MJfC*{Ryv0Xa3sIbJ|_*P)+5CNcuK?;z|YFwgNiIVdP7;LH38
zmh$e0Pkg<SO7)-~n?-|kk@jZmhu-{QRw_YrpNt=DmC*7jIf1zejvat3!2rrDC^+LK
zeFT#`KuL%|#GCxE<k;(~AQ+fz(3`d8NA51xOCSBjW&Sq25)cFl%*b3HZ|}=hO{dNr
zbVZ6F8^<e+<+eOZOD6W!hyMVSEY48$$&;{s-<(6bPw#hKOZ;<*Fl}>w=6r!n=Os1l
zct@Y@bkgNeShUiHPhklrFU|v6t?P6I-Dmu{Te##Z<C$qXm?i4@c+odGL&E;TP7HX>
zws3)01k^Q-BP`q3B4Nun*Z+=wQ^tuh+xYruAAY>NL)MWnu&`q~pWXM>OSMkI%xo0^
zdx-b$2tkHZy5z_izff|#S~nB|66F(L0J{8X?DAK4G}d|NQ}OJlZjBRxNXa>&n}48+
zMMXykvsXODqv}g{37UkkhzOnxMQ8!XJp6I)-W>w|<a^aSj?@qS<xxY=t%lQG#w+o3
zb?+chMh8WLl(RauQZ0#0U>SRRH#RhQ0Wfk(sivjnt7o4r`DjO}_ioc2#AR$4`9XOM
z&Ss{Ok_oXeXl?fv{a$7~x@<aGEsllL7`2vdIym2=ip)B!o0`vdvLxH@^#aA=YbolK
zaY@>pj@9JZn@UP=@NN0}YOmw_i?)}$%`<|C^@YJ}l<d8({;zmm_TT)&NUNcYqiFHC
zkA^Dln%JHa-qw8%B9z-7AgxvU?nYT!SPn7i%dObim!n0A8u{XuqEzYn=a@>?5F-!)
z|CXCQhJG$rVBBF^<?;uDU+^y%LsCdnc;PbsobKKG|56k-(DzK3qUQdrGheph0=dH?
zB;*Z&Vp?KvlyOLT=)8tfBUy<dk9f?WReyPv`e?m$z+}nNm^A?*{_KD6KXN8jnHFS_
zzDdHd<p-?sfyVo1i~&!>sEJ^`hH9res9J7v`O?H{)$P32ugC@McvY<+brR);gD-@z
z@HPAp!eyaXC0r3AB*H{_$<p`HO(;$!t9WE-Pj1SHgCi$5A>};-DT%<)#J-N3#+?zo
zDFL)l7yHru;Es&oRj~%!;~jMojOX5}UHRzu?%jiy9UA;8&3|pMp@Ogi$rRw&z`#I@
ztb5SKYCPpCO02C-#UbmwkC8t(!vjE2-0b)6FL@yavcuV_ixmcXTlGfw85Oey$@eK9
z6RpQ`rDLYcDXA%_-Q{(==3u&mctC9Txz$7Cp>0e7E4VQS0xHjWUru5GZ|w*p9>zrd
z=4QS+UT?;Br+vi7K#>w~3b$ZI>UQKX`8C<s*V&1LcvVYyxMQBKAg@0^h=%>HY3_V~
zba3!a-(v(q2)DP_wJE#rht1*cOy#oIazI*M-qH59?#^`;d^vQyhlCg?Al&wJcb`-k
zezodHlX8`un^LHGr-M4-Sl{4T+o(Q1Erz;Oewfr201h8b+7DM&#y{wAQuI+>)*bC5
zf)L)RIc|Ydu4r*mb|$7gNC+~)*U`~#&d#QEJ#Lc9&kHkay^?>EOWu4gkuKHDee6*N
zRL1ghyKBz(X0M|*_c;hlT6OSsGJjWXFD1o9sBK|k4<lg4Hdsx2DVUh!as&?G?7)ub
zyb2KkSs3IYK-Ls=n*SCusSDR8`C7RJgUmky?Cpkj?}goBcL|zdycB-nMZU>@&A)++
z`NM0)$zMnE;9QayM`J@%M<P)nF=lx5>VwW%tdX$zI&c42WS1mH`_aL{NKEu*M0ST9
z|Bj}PfHTR}62UYeYB#?G?5a8%kJGCF%1n9H2a2AhUYr^Yg6S+0cr}#lwlHC<q!NqZ
zSnL?a&dxs4pD!*fwD!8ug6ODekMqHe&FOBZ1^v&S6>Z(6$HqfMiv$`o8|z7YSD^_%
zb3gDxO^<R;5&92Fe`tTc0k2mjtoe7x=dU46HQ__4{IJ@WOOrk_(BO6=5at4Qq1;E~
zQ9Yxhqe-cpHAGvkqj_ji>d-fzJruux6BP36<iv?vZ~Pr2BG|Ah=;r#&pg=PlMsJnt
z)zrEN-mGqJqCMb{*y4DrEu~%naSI%A>O>GU8&UcVF1{sT8hw=KZ@5szGnsE$u02`5
zex(qy)BQ{l(3TqS(m|0J3@5-{1jspbSh2(aZ<0loMfTaRv0i$9Jd$asUtw!z!}eRD
z;SnXzb3Ppeai*K){+`!tgh4=tNzh-X6uJyWwCCL8e5NEXEZ2zwJ^o`bIssqf-n=6~
z;YKmxNq4rK>-dhOqam&5KUU=7VUW$RNV9)sWmaIwyAar`Gekmw{)5_pwe}~G#2Ze@
z6j={eCZ_G)kbw9nE#hpV8Z(D}4;T?Y{m24Nw_!TY<zE*N7C{e;atDc)Bh&Ai?@i%c
zQEtna>BYOu7(<Q*8_LM<k2$FMT>b7ftyYFs=+~89^G+17Xp>dI&Ufp0$Pz;Sb;41h
zUmDHxw(cF*+P|erzOsTHN}o0UrXw_gR9(FoE~2R~uetbJC$<^$72cuU)gqgM9<7-l
zBu{<uRCq&*?K!E3<0Ybq!nEX6ttRWQ1>u`iCc^><MBgzbqe_KIyD?3|sa|70eswIE
zUZ5xho9p$@ZoG{&Zhm|KHsMn^IXJ+Tua(Dr0a|rPc!iN<ppW+2)%+LDDI;g`jknr-
z0}HXuvhMV$uHK>m59S@RV8uFXMuIj4+UwzrZpA1ap(&W^<Duu}jrc#S)0D2%^CRPt
za5}>)*b5AsT*2q=1HGpOy#x-|HyC}^$nE>&`&<`QmI^Ko9CSZ)L|QJM3f0OnBt80b
zcsDu?DH4Y@O$2kUq_pgLFsBC5s*o}kO1n%SB{Ne2LK>xNVUDKbi|(?E7ADzh3A1e<
zIb4$UTPmjJ#N6D66clVQ*6Yq4zJpCrdMT<+qSJGb$YWByAO~gdL4yN2c&fivzPp1A
z#`u!=ynpW$sryLfqm1k>I4;uC(6kAxaybKg3jg}4t`2C}#lABJM#j)cbP{r8MD0mz
zOw1%5MIE+Gx5+u`hk_qHEc}SSq?myXb=5pAQ==F!NknVev>72!s`&TeM%Lfv(e6*&
z=Do={v;pZ3bYEPh;>_tvKkhN?{X0|`gcQ1!zADu|OyAQdglTUhw^<{4o_jp!p%z59
zChIy(UJm-$3*Cv*Rq%gets}dBn+xC*6<vA@q7Zy?WOe3}sh?a6Lqi`fLMQ8(GA<+6
zYcrf`zOdBsM*GypT?IZBPIiE{_|bNa6eH+~djEQRI@h8Hf7<f`Ly+k8`7=63*HKht
z`7$HJjaKy33V9c2=io0WpK8!c3wf7Uh<#g*&X$8rT7B!JX>RjmV5fh7<Xp=PIH7E8
zh-@s9>DgrmJyqT3A1CJz(m3av`Q<9K@hBg)q}J^XGY+KaHAUWyVnT4ljkvj!bZ>o9
z&EbQ76G2a*C8;(EI63V7;43)0u9bN9Y{+U&a->}&Dhh1LV&2pVxr0-N@-AhpN4(ot
zdIjdaF|RWP*#EADD?d?QS!kMaCQjL0TPxVPqw(*j_Y)>&G{jpQ)qXKO9ssI0;jJq_
z>~}9uFZ5iG_Sz&$RwqQ9@)3NjuKQJAJz1^}V>?UvlASdNKse)&D@MAp-)Oy*ZDh-T
zpOK>3zu|dL6oBp~U`SjpRn5rUKbarK+YP_}NN<6+kGn3aw(f>{9)S?ZYK2~ta>|dU
zA03Q){vP|^v&LxuV`a7e`$<7CtyJO#VJl<%O1yuSuF{=g)~vdP-s<S!8}@ABm()|Y
z_kh5SWBm-N74cTH6pM5`ddT$hc<clM3ZNu_4MU;SP!b>Ic-jAD|IljAU-EU=!D(dy
zXQ2^kqNG#tBJ~5HtU#>}|C3B$|NeQHIJFSO_C9}J02EL~p7F<mop-1*d#)!nfR%kR
z=LxFeqO^TJEIy+O4c$Zd?wH2y61kcTrd5hU_O}|>5r=Y)n|Z1A&=1#$;0Ldo7_ZSX
zl50P~A*&Eni=-|B;VhqD4}_S#Q_jXqIhf@&>k5l+=yX|_nyR}n|Da4y=TZL2+bUN`
zspt;k8Jc->(X3no_uZ$ZXfLVtBha3GcmX;~T4Q=#VXqyts&xf*TUWEM)qZ$DVmt;B
z99WR>1bU;Qy%}$M>57tHXV|YeJ9uP2Wm$!Y*70w6F){y{TI=ejw{?FFeNki^Xc?zs
zxYjbf&zGVgjeoa%n06t%FC$8|K7{%e$J={xETHfMZ5>Q)C3yV;!T_w6+URY3-`C*8
z50LyjNv`{Ei;qYMfna6cr2Kc6m8DRcVWN}PM%1>fR0SZQH<@=lvR5pUVSXtpmCWs^
zqI#e-ZR{V*Acv(WTn}@$7^^EFA%aDs&)>a1X3|JFe!n|UgYnWjI-ib0tCH79<liOT
zOKQ+GgCiXbF(>3Ml@u{yi1GQ64FDP-xf_AtVdnhHWvnSjPCA4Zk3#!yP)6IqO4ET0
z``eH_yK#%IsMq0a`M!h)h1J!ypUiXbKl`&@HCw?c9_Se_07yGBo>k!<PGi<l0`d7{
zL)F4kC0s<s-xf?u29a2ljj^fm>^QSr%_6g#2`_4WZ37jB<^MFbZc%ow<gn20RcjX~
zk_bqQiz=vX9E*a)Pm`n!TAd`tJMB4{qBDIk)W(#B-er?yedA8NN-RmA0B_~OC=6cl
z%%g8L=OQ7A*x4wgprnKi_EHNE8shtGCaC_MImzqnkSR4+yOWefIyY2Uu)`YO6SNgw
zN%%kQ&UYcv4o$w<iI#hLIte1yZQa<rg=M(iTra6@XXR~n>t?Et1f=|=w3ONX)L2GT
z32-wMi+fsKnhe7w)vj7KzuW(5<lnnEnnOh(2VISfjw);G4~n>9Pb`nKrP_C>CiX>Q
z<&~<JE&xFS$U<vtD<l<L+S-;aTFD>rEx+L3C?v!}wdM%Z)87Cth{K+u4`)5bcz}oz
zD;gX|8)fn%?7tuF=hOTMp`mp{Lm~?NwkkBb;fk|fSX-q1M;$}}3Pshazty?31vRzs
zQ96HUF|}7NV$-jd@PpLp^ajh6Z={K&OHLW(<(qRnx(fR?Tb*0@wx6hDK}N@ESrepO
zx8k+PXiZ%x9zHrPV^!#PpO1Im@7)p39*fdol8TCwic(l3QW#<HMfp0WOG3@#*!AJK
zo$ww?X%qAgQu|pt>D-lwzKy^uxvff$fUC8hyXg%ih_~ijI;?5dB2U&oD8k&#lfy-5
zGVAoX#^&b@mRJ3mWJ<A^SsrO<US2FSlc7YMV|j<h#s+c`yN<u&cu9Q`L4a#yw%9u0
zdqKjbc)f_nA{ei*pAYV>nOte{n$4g4rC?y1rj>5f2Iw04hsq<S>v*(Ri36NOX~xXV
z0q0J+A;`7sZvRLyuVR;tU=lJ4Hwr)vn42{if4cn;Z+{+?|8g{J%7HkJr%nFnODjLj
zpO6n5eG>nGM8lt+S$vZ)Fh)#d!j{^iRXXVCwe{&L76pl+H?G8={Uvn-Iz5GnM3MRk
zj776n=PA+b+Zq<&=l5`R1=%{4eQ%6UPm>ENB6-As)u6WGbQRe1qxPF?zZ6MX+jzR7
zU)rzNwM$QPrCH9sP>JvM&HeHbLAgAXy(64#MUC;iL$&Le2r>>`COa`1E>RRVKnH%;
z`^&}fzKh+o7Yq%tC#PW#58y_zhS%oZ_PAD1CrzEiJ^7_1Ij0l_rxdw~CGuhXd(t}p
zwMVLs>vcRJC^3dtnjfJc)F$cc`*`vL9-`wpQX2X9xu;5Bk#8<**j77ETpO)ap8uaZ
zY3c5`qR9FlQPQMT>1ms^rwL5}v(6j7iY7|*pR2BqG5_X-U0NFA?v_VT%Z<Dwh^rh2
zwDtG1)UNd7H`$YdoY5hIy<C~sXa8V7Tg`bKHDFPg65DbHw#e25toupp=CAkjHE%!l
z*f<dhA)qMBh}51Gd^EI`Q6T>o<K?6-)9PO=kW%OnbHFe+tq)U}3LE)pUuTlOp<P~(
zBTTU#olTmRH@@N_9=O2E&nUrRQ-4TCEs-DdX7Hgxp~{P2)tRJZ+Snk9sbtTaj~M&V
zH@&&64h6EmzdzVa;XJQvG(fqnt1<g#AgUZpAWDdpgF<-~7`walRs!0t5V|HS6@sQm
zd}*xYT-f}UQlh2eN}@?aHwlAzgD+GpT#wsP#rgNshqe73?H)E%)F;L80Fy*k`yU3G
zV8{k52fGO<czBH{P$lr!lT^FZ<URlQ11|QFCg%t_rD9TSz4rf-OEcP}KI9}58cAxX
zClU}eBKqvUV55=Bv?sIDSp2f$nH5v_N&F?&O@fVW*J0MM;;R4Chuz1}pm=mS22=;(
zMXbRi5xHXfAd~<UvsskpCAAzW9HvD7rnM~h?K)hVZV(gG#>=ad-UA9#ZW&M18RAn}
zztHJPs`b$6DXK@E74yRw^geGAjxXE~CDK<Kk)v&fUg+~;jELE1VDkp>8R$&7yO;7A
z8sal2fpzX!A?2do%z&3vsd0h9&{N}K19w_?<@IaPOxou{SIZqxkGtd0QxGL4B)I%q
zmwy=l03V;45NpWMN+j0w>I4I^Zsb?C*Q^y}n#O4&3c8i1rgPXHL4sE%7F?#Ge(La=
zPLgHDU;aVjdKwk-+d^ylH{o=_3OES7BC2{Xf|W+z=$+x`k81Qr*@`!prxhbS(Pdu{
zUa51$Cl0(@k7CMrqMfgKbrx+Xx3WZ`Y|inXIetEDs5HVX#x>@l$i-k{_;@Yp6Rf;O
z?}Uf6@Nme(2yhX>;H8cX4+og<@4tUPj;^)~Z7v}0wyCKph|`4Zd$V4tI{%-!I(W?q
zzy=lbMd(ijE8?yFl=8;6!i(;>m>3F|pHAuAcU|2y-mBo@;!Yo4-?ALgmf=S7!u*5y
zy(}%G7{zNH<LiWl%a$`s-W9~}@8z<Fj#3=+QJwf}jrT(Cx){FtA&b)4U8Nqy@*gv)
z(hyVG8#)U@c7_4>l-&x2oKkJ8lF1!~l&}H?=G}0^W6A5h(F`_hH&bp(J;kRC-$DjT
zx}VQ-lNO_xN#YafArgz!8CkUk6fCOMQwmjknNyCj=+Bq2rNr@b?X_t+-rr06(Y-Ng
zQRN)k^!FQ}<xDDLK=1!k%&>;(*?=9QoY#Yq4a^%))_$sO3nFr*Z>8l>HCcGw(&vGj
zS+#44X1TwGr(8gM*l@%Mah#t&LsgnR4ncM-XJ}ELIil@;l)k6i_uTi8>~X&f11|xc
z`^a#uK&UhuvEYRZcFaR$p4kzXAAo-Lnp$sF79|&q5~ombFh%|K(v_6P&p~-Il2$92
zM$2?28TyJ^+N9Y~T`sXyK?}ZAxh6d`A+v&3qI$|ZZC0DdYWa$nM~V12ce)iZ43qqw
zohcB7gr%2)X{r<MW$N@-4Q*Th!cY!-%>FqGEg8lMMjn{k(xfS{{^Z+^jD??(84q?@
zs7A4pojYI<h3ViJHQybbhPNGh7CW-~`kC^Q!!I4FxLe2nuNEMuYuVCRPHuW3T~qai
zqhnl3s`D5{pQggZoc7n6hBhpJA79_D)pzj1C8tU%*LyyA91$5Qm3Gwe`5t27weLeJ
zs-so43}W}AA+02G*OPc6LK_x0a|C5fIFe*YUe%k5sV!A$d>W3`?kg2RPiC1S7A+f=
zDe{!ZTKWF()ZeqD=-GN8$tUnBO}OV~4r>IhzKlZ+bvCQY)>C{SFdyD?y7tNFd-<6Z
zqivVHHf(sikp#G(s;WXJHGkS-^vH-P#=yGdB5~!%L{uEur!EA1XB5(uPh7sajBZMJ
z<M-m`=9wec>2yZ{XiYwtKmW*K@@2!-s@x>f=J$uuxj8ZpLm(^!oq&!ns}^LIaTsI$
zORG{Y&u^a*gGe|gA!mKS@suV(Q)Dh`WwjAK@}Sc!$Y19_`6ziF6D8s|g}O>CHu^OI
zvm5wm!N}1Y!_QO2E?!b&y!6Mlq)wWGDMYo6f*Km%*4Cca=^G`)!Wbr)mX|g}pNNa?
zKU-}9zBBLB?S$pTqTtov&!Xl&Vcy3T?p?5iiojk#n4bAHmG5AR=7fE{dw+a#coA_o
z2UXr!W&M$5EYR|(i$MswCbR7~x9UDeMmjLAuo)-iq*Z7m2ct)d%uKXoy-_hGiSO&?
zD$i6=m?^ulnBb@A1J5+Pz`rS93EV^Je8IEf+SG>pm4TL)HXLL$Ol>VTBe#?piUdqf
z-4@2WlEUm0RwzpB@zrq{C@{d&ZAJS|(dZrc@cz+U-3OQ$A7KVO^sI#4<?bU5fcV<o
z*D^iyW6q;bfAwm7m=cQnk#eOM^j@=?iS_Ij6Pi}+WxHEr#?Wi=fD+CWT40xp#W5p0
zB{3tx&Gk!JuU;aN&iPTtCIZpeJYa4NrJkyrj`22WyYDC>*!+9<Ct>p`TZfT%e7Y&3
zoz_jy1GPsN)tM2XaVQJldBLmkah2`6IBel%I_x)X<tR{y&Fki2ea3<A<7fWi$sPU%
z-o*QUrnYa95R-&jW$IHG_V3m}MgXe?><@W)R6Sdj1Ea#Ks)fZi3=y$hL?hK)yibk#
z;3L}{wpmZ^LktQT)sw#rmmhETGcre+8-?QCt_fwo;YHJbK3p1OONlY6(qA?s!%C!s
z1haq&HH{VL22_VpD#AA^>WFdU4g(wWjbkT+H()qRP9}tbBBFVfGv4=5B=AMVG5Z)H
zN8t5+#?-3^Ndt}V<IP+j+1MQLAM|agGv|#x?IyPXJ>JKRbRhN-uaFUm)#f}~`k6R^
zP6K1`4SElS5f2}Ja{k&)IS(DphX&`{j}0i<e$lb8V3^Zu)+m?g$3eo0|II(7{KzzO
zfKNa`cDDDQ=AHj)&jQY<>U_i0ufc?RWe;{dm_z(br{hxsSHLDjL6k-^-F{YUnRVA!
zfGDA%hP+$lBJ~|UYUE=OjDbW3k_JXcSf%xOl(-55(7wv{ngp&lgcYm?H-2)eB5Y|1
zDPEl#3b9glwBEnkVKL4mCsnL|zNij&_RU;jLHi_RegJHJ4e*ExWLQEa1BB|Y<KwS3
zFMMmC_h8Xneth3F8-71X#bynegyHCG^o1s)ZmDK71n6yVTZ4=lFd^%P)6@_@dY$3H
z{^?{!FH7zR<ahOFJ^9ENZj#8)4(87`%Phvn(YqYuG3Rt!6QIMSzBu#n5CXp`Fh@L)
z3{24nf)8-NOw>CVRD8)ldrV=$6h$`GY|$TVj1d3E7hx7PeUOs3lga~<KV(Oow|Z_u
zKK?ldivaTJ(C_~u9rbl5m8l#XS<YqD{>F#iU96w3T*GA^z&s0jo2dABuub{sVR~vt
zzLORA5VFP6jO;eee@sC})LQ5Yc`?9~c>DN(geE*90>W*MAhmL0iV-Ktki!JHaj;$x
z(UlH#abZW(>kigaik-ay>-u;z4p+u_!wo2LSiIjhHa3E?)YeO15rzx^o&f;@b&X>x
z&Bin~M0vIJf+kAea!@Zne;lo0A8Ai_Km<cm4zkR;`JbrV&QOEFa8?Byo6<I934G}r
zmP}^=$lRc2fBpJ3bY4GCPrtadK%mG6otd(teDJB+rT*OYMz(rNYbGntjG$An=u-PJ
zv@<e2Hj1XMfj~SI1E93LU+s3<yywm@HML5z7!q(yvsOqyq@+CHw*OmF0z#+T$mcxm
zKj&3<_(3mI{$J@1DN7xHQq|1YKaEMP_n$dcyue3p@IU|M`1uU?hH;|FsU1uz0g<89
zuX45cR<{94gTz{x4e04v4=B1x&=o2Ibnp`v_EQR_nr%%k6%BSq#1i}4Uwf39RC)D?
zHLS3P9RA);M!L?oGCka{a#Al=%ur82HfnT6y0aCUCpMh!w$g!AffDp{MIv4F{^h0F
zpM^)A@9V0^@C;cK+~pJ>8LNgNVH1!>PI>$iB-mXOmpcIabiMSG5E%T_j)zU|vOCFq
z%ZR}xx=(CFfIfKWXn-5K(+Lx}K`1I>JIRzItL*@8?PBGeL4~Lwh6I_7bsvMwA0%%O
zgV#vuHa_6%g<(s#2~=A%4-*SuEKjK^PaW@*ju$)<ZmX%q!kFIrvsM|BiH5m3Ubk`#
zTt(S8IUeXpWeef{NO3>jUVA-Cl(Y-lT_!A2e*~bP`9NuocQ#Cb1K-M?Fbzs+ViW@f
zpOS%o+g^;VqoeV(Pa*Ml7&B|=X2zOFwT6WyAltM5iLVb)NZI@{G>HRO5K@@3qBleQ
z0=)mYujMZ4&u+m6;1xr9H#SS$e*Y8qySi$XSTHWT@l>D46QGlrTUK`rX2)OD&=SFD
z0oS1RaMsk|oPfy5b&1EHCv>7Ig`8K??E=#Oo>WYz+iwg40sjA87>t5Agl7wa2VJ)5
zfh0y~87jVgD$~%`PR)wNL^Nik+uP4mAglUah(`-New>|`XX5@r8!|m`QaLtVoAhVj
zhlX}%jk`ere;Fv1p&r)$qbI2zlJN*23PeYE&z!x}V$}N_>jbAS&_;oJDkUpxZ)fMV
ztPsn8{QN}z!pH~=6hV5P+gD9X=619R&AbPA=;;VVK*0>t{5pb>K*})VID!dC|No1v
zNT_Mb2!NnP7?;wMn6wI`>s3#I{eeXQS_nr_---7V1B3a>D-6pj-3?76z<vO}dXnq;
z@%@v;LA1ga%(OwS;MbN6|MtG`Ja=VE8u-Fho|=-#dSg-cnhX1fYZH&2McYq+!e9U)
zX4i`T8DRPN(i_^NI^;l*K3^6s(ZS1o@(^qN7HGcyV=s_JH=wVrTj|-7?@Tc>JvKyE
zQ9w&UC<CRPt-dK!iUMq%Fx-5X=hN;t7({EH|0dw57*qDEf5>DIwS#wF{FJQHH*&WV
z!ObZ7%z9INgMIsEtgkURC|^S%A}70W2jrYW!ou-}HL^ey%W1)KOc!!_4tY`lYZ%m8
zA46V)zIMpFQ1z}ZCg1{F(I>i`ypCnc)cGHYfWZYFA!fdcM$W1+sr^vm^!+UYEbKYy
zMTk*AN!04>3}_z!FHo;rfFceo|3L0N#`*9V{L57|T}a}`Mk<4&63eUGBg?2V2lo?c
z3tf@ooT?gTk4&Shap1_3Heh+&)oEFS)bZ^bFAdFe&_ct{hP`6q3su!O(ljt44I*=>
z_yu8bCRB;wBOr$iB^%+FTN|G^Bku&4sqt!c_g>&W6)Pm<XZ^8wIS#HcPd{?a(Yt`j
zeEk>|=M2FdpW(cseC+-m$OQPh4)*qgvoPoL|5y1xQ&E{&@IE^<f>1wLdxQ@jq~%p^
z>$|ud=SxLG1mj00j?LV=y_yPx)5c$$K0lP206WVPq;A0om7&~PTm1K`o+=8$YPw-a
zm$qh}qq9gB<ZyVsn$~g;GY3s7IMcvU05`lHvVh<Q4$caJu>_7r#LC77Oqeuva>7Fp
z6BEOS!d%Dxv9SlQrM2){UFir0Ze0eMSjo0Ln?BOQ_h~cx(k}4E|4D!Sf69B$u&BCi
zOSnv^2r8%~K@?P!ETEDULn$Z-3P{dKR>?W3NRsp_f<!Tq1SIF6h=5Q8$w`tZNRXUk
zkMi_Aef#$9pI`s^>N(FrU#qCS_gZU)F~=MeT}&kh&z2^O{1d~vF`5b<VhjI{{TeFb
zg6!&-Ze874($U&kzt|Q1QU3VZvkSOXXcb|o&u4RoOF^=b@<H!tc|$`wzL6r&3n+qS
z0^l~-NAm3B3ebw@$S)rEmBvQ-+GQC`^J~qmYH6`W3lG!+4ANB$Sc=Y=_e6z<R}IZT
z0bO$`cyWi9uK1ye-~bC~+J6i3*poc*eu*zYSLM@=hbhr?L0TeDpFaII;JlaV;kOet
z?eFsPRDu&ibom+b=IKT`Pnc?{E$7rQ6O7sVSS_tB^)j=wXDHcrY%b&Hd8ycP0Hb;s
zM5(-}wtKFQEKcDY4fr`Do|_pj7zk`-@X}?7EB`(UmSTzZX5FtA4ol0hMzG;6tuMJ`
zXS|h3?>&sy=S>xEUrh)wPcT>YJTC*eDwr>DmLblr`1H?`OG(@ZFPY;5Grc_T#i0rF
zSMS2i4*ix`H~T-G2M8D+4;-QWZjPO@TBZtjw%S^iC^JU?zWupT8(4lrov}mY(Wc}X
zJ8Nr?=)%`G#Qq^2Sqn%^9Nu2PGjO2ODYWtVdtKj?PiapQWO4+wb&W`$=Wgz~JafbF
zrr#J`$d>5BOMamQ3-;rpm_(sl=B1szou@>mg<C=*rc?*gVeS@UUb>wro)**Yf4UuY
zsO&ekPu7o(-Onhz_A@Z*($-^vY|j$dFY;CHv+ey-e>=9W(bSjgi6V<_v!<FY(d7M=
zTW_-V^T<bdf1$C8p%t0$>1FMTbNr!}9^4UKr0p(VZ)hhlbEv7x&2G^7OL;@sRP*5n
z5xMQFrO5|OZEJIX3j~(w#u|-W^Jn){S?pu0CpqyYR0CB{oHDne``T=74$k_$&MfHA
zeOP?6kDh+vT?ozx0!L_DJbV6J@l<J~Y`mZnfx_eUk7vJ;p-f#t-^$$XzI3P*sHM1#
zdA8wzvh%1*C?@{7_wL%4<;V64oFRtaZL^D4r%BSCCE`Z&{>r1mMTHinZ?`O_ahruE
z3|%L^9giIvwV6*g&dxRB+b&kDR!)+5D0Nhiq2ki|S^$>p1pC?qtAHSeCdZt8nJ=qA
z9f{iTyz|1flyF8`ar9hNoBI+mHt>Fhhn1aOoIjQ6-V18fCe54!LKjZ^8L0Xa^#Y7W
zPWsy(;oErmJ-u<)d&*D)t`g@q!y5e<G4T;}jNJ~k8UMa?jce4V08q#K7E`njKuw%n
zzHCp|N34TMRR_xWDLwU|0(pCUB|Nv!!p@GDm)GrXY3Xm9jMUsrRYo;Qj$b~O6r+iv
zm}-VLsmwO#_tD$-oc~&VYm_?R?mO+)nfsqcyOr}b>iBE?Sf4xcpn>fV>x*k|HJLli
zU^?qpMh!)7oE`&Vquc&Z+^GRypU{fcDNFy<iH~b#SJ<8g2P?R_mE_wgYiXTF6>38I
zdSb$f`*quk`~{DEeN<Lj3e<ULa$j1-u8~{ZBvBEOspv=UaXtMB#!qx{&+&-Crt}f{
zGo3ai5)sF`eX!6-6R~_;Tg@*O`@X~@C85*7Qq7}(AZqn={t~gdTsFO`Eps>1=s8&%
zo7wHAr_1;^JAc~PUtV6Wc<NL}l9{FDsrbU+syjGid|X`k16|{}Lj=<YiQ6i{rwic2
zxC0AJT}n$!J9N&;BJ>H@k5S)eUnG2V70s`XL~9}-c7|0$Kokhc&TV3KAl*R)o1Ke`
zOF$q4_5DH0yS$!i-C<e&G1ApO#>svIZwkh_*S|@Ge~Rcnr=X&uQfWI{*KBz-`@OoI
zA%I#_o%q3sh3$l7&%_`_PEu_x^}nXiGFmr(Vn1-}V#zA0)?Y1U_ihExk4})jujeQ(
zPL{5G{%CuWA9;+C%hPj#zyWVcIZsc;@*I@oQDfM@e^yU&lA)y}+d(?^kR9K-E3u&h
zHI#fo7iDj&Z<{U`U`;GK%vHQ)6%F>_GhT~rL?Pm~Nw?v=BxfigNHN(nMC<15zNb$b
zW^9)Ptt>6m3qH*FUuE~}txRZf;$yvC$xD3x&p(s0k)h}3?eM?IA3>PxqwIV}BILU5
z?~A938=v|ZfBS~WEwZ=|-sPk1XlQEc#Eje0c};mHTZw_!vf9dQJ^iEIYWK~0bPiMD
zAfxrVUu+v;M|k2k$?_e)h&DjBm9-o_BgHk+&#il$;A=Y^95Od{nOmB;Y15{my5m_3
zQ=Bcv)1(kS0sFLo0O}C&quCMf<gb?uuhbuA9i(Mtm4Li1rEz<IUCP{On}lmvTlJ4a
z`{ioC%8yd>k?z_rJGO-=#+Pg*5F9My+&1GqsXyj$c(66<Zyb`9`k+CpEqB!G*8G@$
z!i>Fo0G6J&z$JfS{j%u>MW#gW?4v?qVe@fyYeO2+)zfC3;N#E-u1Zw6l=z0aUSezL
z*f@T#yI}5Z(0PGhxo3XS*VWcE?7Py=m9ym3p*zV;AoK;{Ucr6?)S$?DJ5jOXVsY}e
z$jE!vBMdZ&pG8g9ws3|DQVE*(h)0~Q=XNHGB#bw(8+J77`!EQE?h*(;DHm-^QRc0h
zuRgQO*EsEMS$vOexiK3hflwqcr#^v#GU#L42#%pDoM$<JK+1xNq`?r|INzsrHK}E3
zDXC0!nk7DhZX=#PMMYNP(CZd1DE7$`G2R|gYz$IpC`!CfPCmm>5pTeh6ky-2(YomH
zrjtM*^n`axALHleKXRnBt1Hq6%5@+>U<iVK1#{G_e1q|w-f~~_^|n`OTr47WuhIku
zcum`3sT!M_;3oeJ4k{rcH2!)s;=CV~)3VSRgNc_<`V&}x+mtCZ)xQYIq{{r><oa09
zJosn$=3>R<%}$9byJqFnRes0n-5Fs&;8BVpXZKzHjEtBb*mA(<(Ef~t@n=$shdO%S
zCsDH@gQDEI^OTkYveahY=l(Y&yK;~&N2F7mhhFXCJN?H?K6bBf&a$Izc9~QGMAc5p
z@CZ}vW@ezFX+T3*q9PL}b}%wlnQ&)*6BBt6z`$|+WBkm#>wIVH+lw9)S#~VLEBk2H
zY>P|9_W7K+kuq$(u+EZQ;c2k2xRUkk^$Hh(pt4tCS0GR`2xv1)xM?J{z+T)(n^{za
z0KF@Vq$`V`Tk=inl9G~`JxqB<4dPz9Uj7V&!^4i+Z(T>n*A%lWZq&Fb^iCW*$d$K&
zEt*|z?JuO3{{qkiDbu~*vAO-{z1God^#TftBTzdrGS2Fut$Mh8ezmT0UgMir8Rsi%
z@9YZ)cl&vsVZOR~B9_hJ?DO|M_TzYC3GGzwoGUCRf$%`ByjhYHwOq`8eg;|@_6gPq
z3wbq=xynR9`sU4>kh87>ndQi3ujW+km<i$A$3nSQdT(%j8)oX1`Z%>P_xS>JWpxr|
zipz8ZiWRl~NzcCSpV{@kqPFvfgV@=#FD8f^0}C0`x#K;2--fj^vX?%Tu)e1z5Yp~k
z8jqcRgyCbZon^slxyRU?)@rrgmSPbix_B>l#awUs)I6&fUE*1D=EKXN%=YTRMhrCX
zyxNCvd0bd1*8NeEV#_J2*mn;0S<i>{@1MB6Hu0UWaX6x@lX-F>QU9Lc0_F_x+%YYE
z-8KpjOt*LM1)^zo0h)KwBh)Xzuz>=58}-5EI{(C@bRX$^P6`W$O%RDh`{oF@M3W^^
z&C3UewcNE8590m~V25C=-NsD4M0|Wa#xjqC&xN7weL|t04E#Ds`?f5~PNmue#7#sy
zt1{hlP5k-lW9q?|?^IGiuVY4l+S)qbZQpyZDUQ0=5)@A()6yw)YxaJ0cL&iH)-OFf
z8z35Qe0txle5QLFg(eH9)y#c=MlQSAHe7EuB{RUL``^y?yU2&KuJ?J&PV8y%QfepM
zB;g>IS5$cUpmp1B>PI=a!bdqctivfW6=t|csvtaBzR<U0ZT);n!kOd_Sh<rz{zSHC
zTnTO-{{B=Y`!*3Q>Y5+1!+qhw;|kyl=78ym2MBtX`k@c50$4+{>0FaCSL^UfcYciG
z2OHfEkM$DjQ&#yQX1JYb(QBN?ehq$$?m~^zg<~_JFDg;{xpInH2tebBiH|@BVN4)h
z^3dmlMt<Y_+g!W@J3GkTz~5z#@TCT(r^Di=hUt}1W|#6$Yc0oI4`C;br%rUWx5jx{
zKk5J<&!PUYrFM$uvEZt>RBaAA_13lbpdg*tPw7kiY`>SV_xVtm^sieJ#S2<@?<T8S
zHRt3jDsqO@KX|x5)8FTJ&9F+iO>@Gnpbv(JBP=9lF5F7zDO;VC*V#p*P$nYcnXkwl
z08TPyX}%UXcJ9RY>G`b#7Gh!2<!U;-2mJ8Q{@kNUx9`n;$uPgDH2(079!&#<b}GNK
z*d`4JohpbM?B16BuC3>Av0F`z37#IUuZ-&mC1KKL(d_dcXlD$+K9V)Pud}dlV2a(m
zc&_#BsLQQM9ZjuGG(6Kb6T>>n1B|a*AvAnAsDMahqv_}{5n>{*3z_PH9_1kN1PaX}
z2g}-_#54Pui<bI4_S6-f9q1eFMpN`3(1L$;v$BR3#e`ob26snQ=J#naH$R5AYOMQX
zGtB1dtn*K9e7ZRkLFIU?e7;#5Z;67qx7sa>-*ZkR7lsDl97x>7myOnf1b(%zFp7G-
zwr5!h`rhqOMUj0IUU;KD!!iB6jrBKabpQPj+kN3rY>O$!jR<J2u+q{#3k#DXQh4Yh
zZ^7PLn4M)ldq1jF2RZ)_0FXV7L(!7GKg8Atc7CLmymaZakzicw?GuNVx4Wz_4T0l4
zL9DlsOksQ$q4eQwmh#uOB?ro+=o2cUKq00EDrcfMzseyCL%=HC?<rD+&p?2Su~Wy;
zkj+7U9i;L?dp8=Z02ZHzWH4&ozz0&Ml>iScoqts6U{~7k(S4(mH^|S=8Di_T(Uq%b
zYNFO%H>6II++wVZP+gfQn%Uj!`!eoA{HXnqTZ!z#vy@s{3BHpaR$B<KT1C+8;zPs!
zPCn7wyV5MR_$m?WqJiRgo3=Yy<r&aRwQKUr>Ch{VcBQAl+Tti~fr4A6SO4SZ2C)cN
zs<!Qsz1L-Re`x;dxplHPFR?!7RiVRR%eVIO>cB~wi9~M`olOMOlTW1-8zgIluQy@*
zLyz0aALz7GF$w*KQfY{?TeUjfua0}d1-N-U4Cn%~va(`8Azs_Jd!w<&I;K|KrNBnr
zdbnE7ej-|?*xu%w`qbLGzWs2KfvoMyMZ>!MMk{r%rO$0$q-PX-y$A^YIdT~e-hO5N
zUP&f_utOW=;uqR(i4DeR`5zb<@I3+001z86n2@SF&^<Y|`&LTR@XOiIZM5rm(^2Xn
zxB6fdA4OnpKt0E!@0v|bO^$(cJ~x&;#V8N)6O&v=azeYNS%$PF))xmP-Pc~650iAv
z-_zsn4y#W#6?^lbx&`?2tV!b+w1o?sEFmW4z@#y>ntuB93783xlyitNN=uJ$a&n4_
z7F4T;F2$-XW|lu8zCpLA<khRYnH~NtMmzHgblp=<buF+s*_*#EdFh&ojS??St&RP#
zLgm8=;f(=M5u4KmPf_LP$21zZn4XS1vyIUAT!mW;Qy)S+5r?H@{De`Gb0TeS3GL8h
z^aAUu$@4hRyofVzPjNko<Ofey5Gx;49FusO!AZEi-p2ZunCxv_n$4<Y<LF7Al*kmQ
zLO_&~LSe#!!N?Ez(LE##7<rmvnWF+d3Nn#@V*l-P&BKyC^$xF^wsmwSFC>~hjMf1!
zw)Kv|TAFwTn7pCkadlw7byhT2bx~1V28k2Ex;xGb%qR$9-+Oy|{Y6Z>e%`w(>Y>|S
zJiBj?boZCBbc#8bcXE+23Jk#vx4o*mk~beW&!h`0s@2n-KEsYH0`(1|M#WmJ%*@}x
z<v@rdtKUM~pqkHdDD*m`p?6GuQxdM8kYs3DnhwwfyAWq==-038dtG+WA!nBi6w=@A
zc}ce-<rEcraMi3+F=MWY-OBW?GlTTB4V}ww{rBs6`;OB3I&OqGaM1ZNdeO9B+&LkH
zcNaB%;in>HB&T^5=&H6>R9xlGd39)xPu786@t)xbdkEGV%|qtq=4@o?wOzZ&1E!Dx
z<w$Z?an70L%1h5KZ&+KZjSSzfPTeuks%PNvPVc~^RJj_ml=9E)2iO(NbjdN{`ohqg
zq$I8a&LSV$Vz=hvS=!=~0F9h9-a$je4jZ?M`T9S6plhp77eIK1=Ax@a;_PUf9A`*H
zc{#>iS}{p~q9RVj8T2S7YL<_k(Kd_biG|#>pkPy5n=;$QrhP~j&ivmZJKD5*tC>9q
zUikR5Iy<X1FaC-{e)ip+j{E4wGK+PFQi|HIp`l89HpD(Q^@kSC7AO1B4ULHs5)P9(
z*RJ`>CyKi)rX8TZcmGyhXsBycupE)qDer2#x|wuB{rMGtI=<Lz3U&UGd4{!2B?E}w
z4mix89RqL!<u8Ax(@$3TH@|q>BAZ9vpFI|7k#{fU?)Nj`XZVWXAVHR8Q`XRC!ULws
zu}1@QZQ5<dLO9Xkr}VE<Z>8l7DCJ+)x=Q(m2tBF8gloi$iaAPM7dSPC{+NVNpIx6?
z9!-r=S=yPsE0$e?D)Myj1jwC}4BNJC!+uOwPVLd}?MBN#I;BuLipl?`A!B;^@KY%_
z$aR&tu0+_|G^iQqDvg^P{IYub+s~<IcXFaayGk5{xu6C(jaeL&YqP=VZzkVWGbJ!x
zSmj{SFTfa977p`5G2qsaYcs%7eWl`?ZZkR<q68JRUae0b;g=Xd2cV@36iH_0yVyY*
zwDk0qs+_My=F`3iSF~|x9pdZQ@mg|<gb6*<BtEzl!iBF#&j$aFNzXtHZk|uAJ@?iF
z`S`A>9#2rFpAg;M?N7J28?Uj>bfeAgUPZs?u3fvb)&9UZNcuAz39f<$;~C&i_q#<|
zSz`dwEiO95%2#FFg{&G>Bw0086cvSZJ`tj2EBR|s=FxKe6*9sSh+GfjI`EpCzeO@I
z%?fVHpdxy(usBJjWT+!$YZmXCm{;D<oI_+SX{=i;I8Ec~{z~t3aZ=%#kyG2qGHZOA
z1wViPUWBX-<oM9>ZQ1bV3Py{d`&{nEjT9AV-8sd$&UZkQ7*d2Q?VjX`Obgxiu;ATT
z1$q%?1&Iw&AO=t_#9@OWK%dlLq<P6#Eweqv4|}fFbAKCNCBB*6W7wkbUTioU@;R2k
zUKRtfDd=tt^{d{p0NioDre-oHm;Ftb=anAngGyBTsrD`v{BeN053@NtuN;#O{rYPf
z4q54tuNXJCzmJB7h7eP&K79|24<rYu$}pa1{>tvk=12PI{q1#`sZ}>GF3%Xl;*#eg
zq997&VaY=i8EGH)uEtCEQAvlh!*8kSskQtxo#Dn<tjK^E6B4YfjDCGiEkR-uJx@Wp
zgx1?8`xtQpluS+c<>*8ee-Rd5L_mC8HmV(qn%m=)G%Xqwt(Y+V8PX?Uo_EyOPauNl
z6QFzn=m%ae9OI^6kbI2?Yr?Tn1SkWIM3vP_1acqv&^Z^xT)!5lRqX2%6-6SEL^|K9
zrSzofmSGeZeDPVkF_nim2DTDzM%G4)DbK90K=1gA<91gU7yPa!bVl~|>9CD7#n)#T
zkp3lu1)bEf(NVJ>ZfHCtM<)oJx<Ef6Js3G<JDuZ2eGs-j2IJ3ts!1gEY*SmrGNu!A
zw?0P}@o2~UN##c-Hq6_kb&U_X0*jjRc6)W!`CE9?soF?n>*J1RjjNAdg>z22Gr#}P
zMM<~Co4R6%Vd9frkR8B<z}LQmyOW8fX_dzQAL~H*tQBc5q;-aBO+;mS3=i--MrMEd
zd5<oMDL7ZPS%VW=Kj_$~m`F!U>H$~DmoF061ih&bwqjdhy&)rBG+oUFbo)cky6Mg>
z4b8L3G6AjXY)j2ayfkyeO+#I$-Wv557RN8RzrS@a3{b4uZ0G3FkVbbW?L3SAlV<C)
zO>*7J$r^)gheuvmL*p$)<wM&jE^_UmTJG0|*k1`J9=g^*b#lfOfrFjhJZ6w$+N;VR
zQx6&UcLHcI1Fau<9Zds}ND-z3U!MK?`BUmB?Y^YuImd01e$rp4GCSHx*H#ybQvoW)
zb{6}}K(6M>#*H_}`5Icd<XDdkbeVsIdHve(G~&$4?ApTlMWZFCrpk#F%Ew!<y;)tw
zlUo>W@@}fR^rmgJsj(4V-C93CuoE6!!Bj)&t-DP92~)FJT^tBvdzt|ES;1N#Bf-kR
z@cV~e`G9q#Y-$@>d&j<Xi#%uVo?VWEK1s@?5NWGvALGD6v$XFyR|VGpl<kpFl$4z4
z$agjNSk9#5ajK3$+uU=v7-p8lte@GU8%oJ&A$uOaw6Moy;aq2Szfjsg&!0c{XA~Ud
z@H}T=aQ~24965bPWR-R}E~g(xgDl$&Q4k(lUtZ0EX5?}xz~ldSGYWj+7U9mPu7sOk
zW)_uUze+>H^V0p^94C-ujued#XKjjY&#J}-rYVn2&CrCKQ(4cPCvl%R5hi5PXqg@Q
z#82)EuNi1d<?8TqQB!mGGtwD*#f`DY20I7dZf*4%K&~9a_3Y-#?ssiXH75C$fz#1n
z#eq$z0h=h#m_Nfjs0X`<?&qIB?+kn!M41+9<{&RyL9MmswuMRtm{L}D>nqzihJZ?)
zkL?SKiH=N`7t0vC&bLQU>7e1z*w`vu*^3u14j$aBd<K<hl+Q+NMD^u3KfoT5<)!`w
zfb$j4%2SU$Wd{u0eE1(9siSNMBPF-fQV<eqk=Fye0+ZH16(<&1#C<a}b@as-jbbiV
ziq>TTk&uv*Tk@;Vb7<lZh2GyV+F@<(`;J0Tsp)Nt7Y!-n^o%Y{hp^LvR>a(q5b@Rd
z?vtvLK`Im8Ht4+2>kTKmE;pn+pW=yBpLj?hSd_x-0R}?lvZCP<X12r8t#la0Kz!9I
zyjH%wtE;z@bphMp>#y_OMzTrYJ+d-3N8h({3Vmq&&)ds)qDOKEQf|9#tl5w3m(@EZ
zB6{oG2D133LNE0Q*bS|YoH^Lhaq|)gQ`Xr7V<hRB9^B0)3eF;UL^h(Dc(kZ$Lr&83
z)vLqS*%i4z&oCcf{^-oudq83~Sc;ZuxyzcLHJSgVfzt68q<#DLmA#?=<~*TrZ6{CE
z3&+Ln7}_S|5Wn-DaPqcJ3l;9HetxGwiSgm#VQ^HC9>Srry+Mj%8CtzO#>u_>Vez%k
zZ^4)X4aI|(T44Wb*H*^W4<UFoOja}b(*nmL?wjb|3%PT9ToUTe0^sN_&!SreWl@WW
zv2C$&adly*$_B@IG|MU=(&KPlP)=}6I%VmEptnz4o=C2;4#8`vsnFhh_V6~sUU7ww
zeiMKbP&Nd%W|hDCFFbchsMQa(=LW9}y;~sGG4`hbWyf*5cW1x-oUK8K=viYI*Yz<s
z6IXs3SPpg(=K8)YEZzeA2nE9rTE!7jQL8^9JCQV<?RCjS3lzdOW$y1dTu4Hal2uVW
zj)+h*s2w>qx7+)A?EA1hrC9ln&eq=!N-Z}vG{)%rt`Gw>qpUV$`tM67O;6W{gmwwk
zP_W7kOR(tTF~=VgyU5COID5H0s>8(l2TI0yB*fyM;~I>~b6WBO_l1N6SXwW^ylA^b
zA|rDh6#ZXOk+zF{933;N2~LQ(z$=7I(o{LUw#-ZMrSgT^MCA8hm?gf)Tb4z;A1^-d
zAS~&eAE)U03;||?Z8uZLYS*<z1YvLb?2ghtP2t(L6@oP9TwejEdcZ0Eq5GTi`ckn-
z#YXR?tY4iUXX4MKXNxp_+i1|nOs_=*5K+<5Sh+*{_6a)A-$j|Tzdzli>G%L8NM|!s
zQzi{>$~V?}HxQ(vNC3DSQ{2qZjeWrEepe0;rZb^-A}(Hp!98#wN}M5}K`#FA<HOKT
z@JUk@IaNagf`Z%!r5NMR%OhN^Kp5st#rUSdIQdOqn7E@=@h$zcUCk!v96RBX^V&Vt
z%)_3CCZ1yW{Ad5h(6j{4KOP<)!#{U&o_t+HRbZQhAV~vwD#76y7LHZK>Rm#DkB>^I
zW|r|S7y(^fUC4L*YC+BgEyTG$xxGvrh9v0k7Y)Le;;?q5>0V}21IPnN08!RxU*9B9
zynVOuUr=YCM}Qdvyf54laOhN>*hI2=?gzND3RcMgra7#Lc%@~Vu$L|bbMVUVDv!@!
zY760XR?2E5y{-PnjLOFLxEtLiCeh%j6vm`V4(&={gt`yN2%ifpHa+m=6*9d!rIw<E
zta#0(?5d(#tLE(-i~i3~a*xU}u)N2AhExID8<euTy1MvOGl;w|$XZ<0VKe|N1Z25Q
zS9{}AML;wfh=Aw@qJo{fcCoX8e(>Zfz1-B;p3h*2fy+6HXM^Zbh__+b25)j4<fH65
z9cy}n3|YM&9cOo~dG!X>1O(yw0Yi^E&BzT?mF|-ByIseA;?I}Ecj^2YxxH@j^!n~(
zb<{gm!)54bX{X;EIl(nEH&<I<zcJrso{KXG&kehTqIZ?>10w?rC?Z!!5t2H>jX>W8
zWm84fOueEH2-?UWA-@4q{lIA5+}5qB6IaJ1IYD<2!@fhJUB5m6VpMj~5cYrr2ake2
zzfFz;mF4?C+4L%pFdu$lkDIh~Xw!C9mdSZBC&j}Gm76YX^Ci-vrJjk&^m?}L>_}27
z6O|Xa*J!xOXleQv;2oqPWOS(rAgwh*j`mn-`T@-+oMy|5i)2){{j?#FFO)Q>_w3<A
zOLI$7+sxou+jqB$wPK>8?lOnfJLsc}SFpY?Gd}IZ)2lD;41K8KNi(`4v3*fuex8E<
z@u?qL2Ad;!T76QRa6T3qn}=uDy@^C4+0?8Tkx9d}SDCWu7)+Y=bl=@#S1>OCJ7gX6
z3}xlyFoU`q&3nj6N=>+>=i}wG&!qn2LnqO%OL?Kz+SYb0DBx$|qQ}oM$e(;%pW`J%
zimQ7RoD*F())r6~iUC9yALHhPq77V#tiCFDcF$zp8kK(i^HPVf&xj@O!<SSFKRO?F
z9}%efBdsSP*PhwAqj*yR*YaRP*R)iZIrlwgJY$H_-Ntvms#irbucqfaUgON(!lm|0
zGuwzVtQ<^8F&80ws~!*8U33+dl`|}r5kBQVu)0Y9R2T2vCL~15xW1!hWbA+y0k(yR
zR8T~uqp2ylNd-Do`O_2LCtn`wp%s_u{{m`_NuBsJsRaFaSCYQ)pLQw&;dBLjDzwQ^
z>>+D6wxvqIgTNUI2N*mQq9`jg30L!Y9a*fG_I69J<yL9V5P6fGaGZ`yf|V&-2z!rS
z?1>l3?$cq21J><Sd|p^M4|jx0c~nKl1+y8-fISN5=iZa->5f3Ru`^d8UP0$@>`^(@
zoaf{uSmzMJEEKksu~kUb)ft(Ycx*-bmAhs2r!J)^rDA8GoUH9ScRQr#bFSjZ0TD-{
zP+M88!lS9@!p%UKvMx^<^+wxBtakw)m}1#ZAk>sK*Vh5j(}t=9qP@Dham33oOu%1U
z8mmR4@1qI~!AhuEfaX)-iB^vd*eM_A^1cJ&bb!QDiKMetM7UwH!E$PQb5|~GZfBl`
z`K!cXRZfp1D!d2#e||Op+WD!Xf_%#^@vw&00&rWUI_Oh8TViKx>$+SvhpN!Z$_ge%
z0s+F;G1AYEw(&u5=FFKKe9F9Ns>1K_;8*OD-`Qus_PnH^7I9s+DW7KX+*9`=M4Hhs
z_J#KjO;A1wFJltz|7&$^VYXxo3=pzc;{N^npFewsP)K2_6$)9Hn9U`czlwwa5e2>s
z6^ReyN5m=0_~jF49fzJ-SX#bzPW-7H^g-($wp>aI4~>C|(5Vhb;%td>cW9u??xG`f
zegRLnUgo9~q*7RVg~63jQ9suDeT$D@lcK!59zTQsK9N>b+}=kW-?Z14Sbuu^<J<U2
z>-%N)Vrt4G{`K*lWA79r6IrW}MWKEt?6#Vxl*EFDb&Qzz`CT2t735m2B`=See7<O@
z4*}K0jqCh*x0<k12PD>Erl4Gf?x@t_d9}GcKP-V9`@YRlRzbV7-SVt?W2595lhuam
zkc!umhTlt+-!*NkGRPcV5;wV85>c9weCnAFMt7sXXFKLh3ArpzzNPT8vMN#EdoO(9
zlZ@_s(|#6~wW;vTFZ_a+J0TYH#b5o*(<$@VPwQRmY)_H&ggg1Qai(#T`29aop446K
zcJA53{@N+fnSf&f#**BxW2l8X-7N*dA7GnLLQ#>%>zj()?u(PJ5x9uITZN_b=c_sY
z;`+5~RWo(5xJ|McTD`9iuouk7aaCTOEPQl{reV6^zCn4pp^OUk=P-p=8P(Q_$!ggn
z#xIml3O~+@*eKf2x~~$ZYDdNr6Se=ay}2<^ab8rCjlRY?0)vMoO6KW#mcx9`t7c}j
z<y(V0ak{PI#+@kXSU|#F3#5}Zv3fj@r|g3aRBfg?nz}5+UqH`oAg>w4X$a>ZAi2Yg
zh4QFdrAzYeADv}29NgO=Odqe4?>`~DSy@FzYWxt9sJAoRL~9NKq1@U!KhMVV&=iK;
z8PeF;-ORf=$Im%TK3F#qyE1+VT3|&Ln+OT#WhnzOei?`fDryA<1u*htzB3?qBH*WE
zTqD1Q&vJ1{znu%Y-m|c`kH8%o2l=-wT^`MX3)+9e+T*fSb1)86)3no{!HdXdIxUax
zSk*J?!4S>u5czdHTKwWp4zXjJ<g!_>FM;shH~7dgmC$GbgX`r%WRgpB0u4=Vlt{2&
zw`KoblaBGvY;0=%x0R11Z!Zo_#0<_M<y4G6b-2*xbHBSF+#nL&>2KQQPUNweWY1Pk
z)Si#f53OxI3K~2jM1)gtW<A<2M;^37q|;27Dyj13L8?rF3~J&U3k%!WPnXI>p3Q$>
zHF*~?&MY%Y2{BoKK}ULwdBXRouTM_we*$=M3XoafmDha^{`J$8&u*J?oGMKFu#(wE
zv8xsFFj$4;HxORNQ&&Pa6nsr}H8m$Zt$}B`y>5|_iY^67t;y9@9B=Xu-T$C>(UyM;
ze}P5+qpnf&ytyZ63w@)OGAOnixF14Ts9E{>`H5K%QkC&zs|E!I9)AOcDU0v`R%?OG
z5Q@uY`G0O|c`rTcJc%e0t&S+1LNCczHh#dE@7$Xqg&F}6<~4#1k(2&3<|7b7!qqGw
z&+>mJ1a#UZJWdX)x`|=BN)9dMCkAiMoL`+RD>k1OJ<s73d+|&EeqX(vj9oDlLdjt+
zh=dhzd7L5F-!>=~(cLjy@)lhyN@iw@==-t%j@g`k*Td7xykaj#miE!uj=9W@+zy-<
zc=u3Z&EfHQ&+9+x&a=Nu8?*=nPQmOxasoY?`q2$Ta1L-Op_@*3`e#*C#gFo5W8eS6
zCgH>5pY7=qVkw$uVn^Mdm0s}P1Ny&J)(dbjf7RI@->~eT99zJgHj&O<gbN1#!}>)r
z;*J6sfz12D9^n_{Z3b&~=U<Z_WsI5eOwzCFHuK&lMk=wjqF$RcL45{mnu@(2LvA1H
z)Zo{e6av?=0=^(F&Tjm>BVxKx{-Iz?<C)b}mlBV#eO_mB={_$OR%YD&C34P`N1?DB
zy!B~B!zfCO0$+f$A(J=(c*+L194&rmXf*hn%Ikw2B`NvLt56Fos~-klkWBdyzASYN
z^&6Zm34pyCydaT&L_^OflVqlxl*!TwgI@wPncurSCb;1<Hlu3)6#X}c4!HrX{Q6Z2
z;1pb`B~#LI1G8*aqaoS`ZhK`3-^XMh{P*27%*^8Gm_W^utP~BJF!H~siuc!cWmx&y
zE+5Z1GoU%?kMhQYeON~{YwV<^K4<mQ$Iow989(BVK_R7p{F_Q88y8Q}xQ-3;raz!?
z%Jx=DI)@+^Rj5;q#|Rhh>Bzsm6>v2ZJDyR{kilI8YVI$dng))?T<h$4uity|;k8cF
zK-u_%=bUC~HVCwSg~{3*zg}oxkCiXEyoo^YmMJJZ@-I{z-)O}g7k7{AM8OBeDq@ZJ
zp0g#H&Ac#el2IDQpAL^SP2AB5?sZ!Um>>f7B0u%cuC839Rl>sUJv}{nM|pVzGZjxC
zkxNn!4VENM<mWF-H7I3Wy%sCa$+fcLJuPtXpkUDGGN1&1wKc}{v7c%GAZrb`2f}`n
z8jk?6171;A6LoiGT0OBTppYMSP|iA#VgfVt#Oi(KT9e8%DGHw}gDhk6DM~?VGmNeo
z&xmj8OWIqfniElf6ui$00|q?fEo@CyRaM19j1HZHhk<f%nF0TOO<v`3*PAj#>7wtk
z#$46j2vy}sHO<2|U<nL%Te-QNcXSMvA`<sA_*7L92;2{W!=iJ5T;~Erg-l$t#KjSL
zB2LEho(mI3hNHetqV2ZX3X?4H^O*70r)O*}Cs)sFaEyab^Du|elZ`VjZ(mU)ZXu+}
z$JzO<L0m2sHDRz0<*?z97@(|x-_>0AdAs}bllK@6UvpF(I1q;b#%=Midt->SM5@n$
z#;{WaLMy3RjJd+l6=L{>4pTpn<e-F?FH`>dQp)(im^fR+;P?-&Q*)E01;V1DP>Kne
z*pF%mpMr!Xk~-|~Vq84R?<k%*J2_b~`H7tpaoi@V2}%fyvJtt`N@%9T3XtIv8JXFI
z1#M6y;FwcvrgO}$w<vpFA?_a-*7e8%uZD2Cy!p}Pq?UCQU_gItHO&Tc1QIK5<g0hM
zh^R;ZZRhf&d1EdV5N=un%L4nu1||Sj``XcGWHtwCkT9xI#{~-~p0KW~sU6o1^R(?=
zs@tqX0ePa0fglMC4MBZfnS5d59#{kf><aSoM`?G`P%{gsAxPJV6Oafg<(B~BQjsTo
zCaW}yMQ0BA`T5Zwtj)1pe<ae_&UL?ce!m4L#57kWT}9*OO&u3aa2<2+Yu2$@>fO^S
zbq#3sG%(7`pN!u2i3(x>7tKFjzM=IdmtzfxhD+rpf&;l|iqZgC79EijQyD)z$>_J&
zWmq|1<Lc)pKYaM0aa_nlCcy5-{Y^BrQx+*@8TcUomhIcOS63YmDtI0ZF2q-IW@5yj
z-S-<f0l+T*RAAp<R-4K%bLTvI%}r-86goUtd-tA~XiLA?t_O}Tlcn321dU!cj<9_h
zTj5fgL<HW)D8oa)dCL~W&A^E=Km?6C=WGXaw}pbLPbfTJgOjUw<5ifCo_M#ajpw%4
zlQT2$FfrF-1y(c)2?P!PY-&kJ2LgR`UmoVj*H{NrK$RrSOz|ct!-|?%%si%b>+4eB
z79GQcGlWlSK#hO?om<X&&Phl>ppitv0;+>3_M+JH%EuQWUHmmJa`+w)1_Mw;vapzh
z0yKVk7i=BU1oiUADzQ$kGO~wgZrIpP2TLsmZoy`7hT{W`=eUs2EZ&ZEhN^6=r#P}n
z#n4hZ*3f6<`WJ<72k0JbG0j$sZz4PJ<SpnuS**K2L>6jPks=}{h6=G+cBeq`dZ=2{
zccmTm`F~l$CwouUT;qKqNX4G!LEB>jg$T47pgM3n6uCm`#+txz7HRuz(32-Wj;Uv-
zb8|hFdVc*_@5P#`zhvJ;MNT-CV1C{cp$5@WQH!#r5?FVYllf9d{Y|%gdb5h~d6ad3
zQ=`fgO_9;U%Lm?W-VBWJ)2`iwzDu&QyO&m*{vnaNitJBpy6_LTK|61_j3K!mjkY#+
z8l}s((pB#ExF75BE!XqX<^K5!qG!OvHRa{upB~VMQ>i+{`|5^C$|(R2`JI{lbNr%#
zK{w?z+Bj8`n95%0GK3m1>QdK~Wh5(d9`M~dcUqo-{`seM>7+B#CS_B<_x+&n2YX*b
zgF9{Tu2p_fa<c!#{=uWcRYbO@$HZTqW}rgQb$;r)QATdOrr1W|%Bbm{cD|w})WpFx
zV|6uthJiZt^c=xe5Hz}LWK=1gW4iCo!QIZz&e>`_X=j)}=e<%1EnnSoN5oz@FTt7n
zU8n&2#Y0bzF)Kn9d*_NM3OYnZ6%;$ZLATA+yp}fbtG~bZ%<qdWDeCqJBSPBJ*}U8*
zS-#XfcDm^z>2Pba@Yc+|&9!GAjG(jS-B|KRcPTmOyA7O%NB}{#nq@cPP)nkM(Usow
zf`+ok)H}jnFhP>zDcS2>@_4l6fsZjvFoG-qL1EFDfwb#{Fk$^TonN`01`UA_1%@Gv
zR39_tzkfJTWUJNMZzZHT;hlZH&{AuNmt*bvkO~k|h)f3*Cg_ykgut2nUmT*_X&!8S
zTEnMll;|fUGwHBfQ#A`K?CH!h5OX&v7&ZO~3YudrLenS8!w(KzF1^J9|J^at`%kD_
z#zM>4q57D5du9hO)PHJMMk^8>{dUWR?LXy8(QBEEn)*>m4mP%rP8}T`WHZC%_2{iA
ztSKMm&SPwp#(4|l=Sl`X32MLuPs|uhT&Sx<jlhK?g9)i=mbF;y(WcH8w2}P#qu@sV
zE;PTE8n0a+XwrjJW0DGEBWz~*6Q4hKI(sP<3W|MVPg}wS465m)#oV4djYi&LcUl;)
zg<VS13l&gLc2vKF{;0oaOHJ^8ra-hT+<F{{kH-KXfgu<|`o;VwK>$39o%@m^o@($K
zF~k|}9sw@~uEo~gJ_cHhxw&WU#r}Qx2BYi4?H@()wEl}ue>X=(SL5ct+4LublfC$P
zi>ggC`Vvlk7Q2DJZ=(b+^ikQ4AVn<R1M4u6&4il7ygUa~b3j-?u`gW6gmLJ(8iBC)
z5Sa@=OUfXZl>8ouF%xuTK?v<`PA;DkY4Gzq1o~z)c}Y*+v>-eXadjQtoo#%DID0kd
zZ|w5#f1pknlCC=I>*FSC-17ba2#}xTUr}!-gf+dn;&*fx{ot2B_aSPOcQTl4yh1l`
z!#@RQX3l^Gv?#lYK)3yWUMaQ|`GWty|Dnfw=6%z!(gzL5B##UY(H}Tqpka*Gy!+h|
z!NHioEn(7l)Jr$79bzK{mBX^*78rZc)7y(#5~Aqg=HW@vd;Rw901|1G-oUD@Eq6ow
z_esFg%9|Z<?$M)<ePD9r<>di6`~3d6v6fm~yoPdh#F=~efaX9102k5dKwSpj`wlzC
zj#El*TwkT_xwv!MN9vVw@Tc8Rwl2rS5CWwL-a2$q`<d}8L8bG#BcKWviKQ;!2bg4N
z@dGCu9~^AWFpy?r5;T;tvdRb!ZUU{YHnJ3>yowHHZAGu4izL59Kg;vUKp%3celbwT
z4bufCb{HNHxsEYJi?X_goAX}Th_dw3aBt?==6&h)2`(=7H-J;0NUQO_fEnuP?k>jJ
zBZq*G^x>B(>>+hf!|<u2BOfhSNi7kUkwpq8_T!dZ;`!kY>i|sbzU~l@-a4TFpv26u
ze2G#C4UswuX@QBugFpNF&^QbG3#W07fqp+J19#&R-^a>vPZ(Yq{XISy=6sT$KTS{9
z%1VF+<c7i@PZ0nSnkn!wa2#<Fm{DPCYdbeJm5-<w_}S&lm(lWszVs+Oc)<A9>Du@}
znv8#`PM_EonnQ_U7Z`HsxXRcMBt4MiY~38B$W3=BkwLc)IpqIl-~NC1x2UiQg1f$I
zYtg~|zC^>`;~yEkXnJV+g;dW!l_D~0yk12F@MzUs?w{?)9<U2&5p_;(^Cq&L8feA)
z{K_|rRe}ytv~j&iUo}*h;UPAf>^>-xiQnKRQI5UxkEz`(b^_*#s%(;_luhM&>1##K
z+Pu^G8XuIBvei-;bF0Kc4Qf|Xzlc4R`to<<{1cDpV;4TkT7!?0^)!wD^Ijou-u<il
zn$Ka~u)hBfOhpO8ts%0K$K%J3!D9m_B7>U*bDt1k;l1TTALe<-_l%xRHeH_<+qj{H
zoDt^lRp_PCO@}QYQufAjU*bi=+;DDx5Es1612xKR+aG*#baZU!79$dm9ZMB$C!@lq
zC$sv&mIR`*l3Eq)E}Zr;^YPkY(FoYyo6@iQ2z!OWz2PfaycFN`I{yS=l5E>ew}HC;
ziEls3z=TgC|J{##HBAd{?UL^)MyaTIqBIurH~YS%<Xl*O=)8+C3JRfaR>@gE+nqD5
zgIG?J3~~ApoW{#1f;dVK!d2)!KtC#lozC6^;661MP}4m99LUTJ;tI-?7@8d&C)?F!
z-<Nw!zmQKO>gm&|A6=-v?S0UxZUN-u;g3CyW!|%Gb8UBz3WSf9=Y~f`$y}_hs*PqI
zx#=B<0;k&)`3rVj_~ZHx8Vp_M^si3z=C_oU={JtF7<L$<?DqTPwRcMUNR7XGBWImS
z+>wZr)9kWy`wqWB{|hp)7XO0?#{Hem2Zu=dKE`0opvq?W9iHUD-20Yo+vzc{2CpUf
z8G??gsrlcCCgIe7yK$m~m|BKy(BX{fd~-81eovkV@Yaw>k*D;%(2zg_46_zIQ9n)W
zhzHw#yz3KoUoZOj(H+4B2Izn@LN@m&Fo6V~9E~AA^@6Z6I#t4s?=U)$cK?1N@#4jc
zFl`l{h*prU0yw=PCDo@#M(B0Yh4ag9{b$gbfxi0JG#!sx*4+*((8Lkq13u??140hU
zxA+C+U%nulE#n6iP2o}M-}RG)l%bdYs^SKXFOjNj$JndK_7Q(OS&_oJY(FLVRO*ht
zewjU`HLYJcH>tC;a~gK}vE{{`V`T}Q!Xi&8J>MYSC3BJvP<mSG;@SH171#`04OCF;
zmo^~V!ad+P;`}mvf_5O(K<1nT?+4oh#ODloLXML|uKx!ml#ygVqxHIIpD&S{WD2%B
zQUaOdO{UrZVGN`InJ7#K1wUd(i_yuGiI6k{HiG<%pZ7a*RX;i&%y@no5y8a3V64G_
z>Yd9|a@Ini^t|-8!8jX%Rj|EYx(b0_%K&yJwQuj=M1{iRqR#g6jAW4P+b?#2<0r%f
zgY-8SSqjvl5jH79A<zkOmhJ5-AXbCjM=t$o7Up0fz<fgB!^6o52GTkD73?ws>5IzO
z%wsb>{r~E$q-$pZJNA~ES~Z^GGt#gBc*e*L{<cA$CnrG;-j{XiebFO|{shAKcE#iV
zyPg=Rg2;~B-Lu6Vdyp_>s$UANh(zS@D1ks*-1qOH%zo5jf7Xe%SnDI2k&~2D5sQdQ
zNJyMMoelg2<-Gcun$_uf&4`0HnEotS5B2oqw)OP9oJhlO+|1Gbw?gJ;HJv~*6I$59
zbmp0;ZMT3eZEdvN_0yis#}2Az9HBIQ85Q-tz}Bp!jC|yxbXXG+)u6C*OpecL=G7$c
zQ96bAy5<oB)82=?Q5+#pu1kZ-g+kkOH1LlLi(8$&%c`n;1sbbJky&b-T*r(!t{2KV
zopWqHmM-itz5>CzV7fYca)68}1rH{7=JbV~HP@P|376c(YLgbi6w1ANC290T=DC?3
zAzI%aza24X+UebdXJSfIO2-V?iyS5Q;=_aoQlIXTW5G{Wuln$VPzMwE|K^A9((cK#
za7lUDD4HVq?7645`5{XK=pfV-kbI=ZBFpH+SLyKo>kv-i3}L&W4OL75Niy~%ch2*V
zBZF)YkQDY8o};1pn|15BforCnX?UZ|i+XAIX1PqH1RzE&^f(7i9G(<{Pz92ZkL$}^
zy%+l!F+&SBI2X9YB+J&mshf6xcg<tFncCk0XL~Hj4|tn+nKU2zq_pbnDIF*vFoA+T
zn6EE+xVdZU>w^GT$sMhc<isP>xiu+8-ps-rS(2RprMT!)`S3jgJ;R!?{_$~HA}Z}O
zlars8GA<zM`=Fe9h?H@g_k@NvUF;8`{*28@hqJ_(k&G=YEC9F1EJJB&=>f6bijdGC
zLlpq8ajXy>NAXGHp3%;+RK*d4Bk1(=(dVU%y|-oe@p7e6Ug8o)T#O*i=LBBIj(JW9
z-@FYIcJ*on!nl33v^$1K)0ozcl9WPx6Qg^w)gy}%ZemE&V-jWZqi>XfutND7vP6OK
zWDF|BheI@Y%|Ssy>5lA)@v(hFq{gpzGT^6~%<G!i>G#*Y@Jz6Zuuy$?d#~C-@)1^2
zx=MtCDLItDm;Yx&7sa<r`PbHd7uDk$N;(G~+j;-XKKc5<fFm{qsB;^|qS1PZjiYZc
zx=~I}?!SCO$pWJ&dJ64PNo{RuX{oIACVvHH3u5{lhLfrzYKPF?|NbHVNqQ`hg{-mt
z-z~rS|F+j3xUYeK0y^*zxsi77|H{ex7g0_<C}jK%y*>~}Vk6U-FjE5>-e`(AtNVHD
z08x7v9k+3Nny!vcANoxKKF&2rvNCR;!y~7sL7hxzvab~04caWPv{HwrEC?5*9-qk)
zxsU>ONFoR?t3X?6ecu7e1WJGG4d{@_sJqL{e_=zmXBhZ>{I~)ALGYY>PY4MK9X<Lk
zD@*9ek$%uRu(bjDt$}$#MpC1q#wI2<s<~35#2qv6t6W$a2|G67P96%|jwoKdSc*P7
zpWmRjpm?%t_inej7G;2C(nK7;jec)d^0Wjr)D6^Vk^8L3Z0;vd@Esf>*CAF55EnFZ
zW;!}0bbPRxVmAYW{<wV`;nBlKTP_^_X|;rj1?YN}B_eL2@aRuUEur;Yce~5^R4Pl2
zhXxpv#^;P=t<*NtZ283IyrX=4a<TG^yI}ifE+J@m_`5Zw3z-Fi0w{<bG%wPPDj5ZK
zj2l!-vBUKOxy1|RdS%WKd;|Uf{uKG;=LgS+A%1ucr|SeL2|9LpWt$G~#b3d(xeLPc
zZO#^`Ly(6%WAeBC?{7yi?gy+X$d`P6yc6z~o>#*djBpxi7SY+-v%NwLROFf3t;vwn
zUH%y?M^5X`?<VX$F4IGCSOxq98EI*-;b?j<0=E8p`18Dx**}`=E8UW!G&(X;jh;Hf
mzv9Jn;__tc^55InHHl9u*8_6e@*dzVglku1i61WMJo+CJNe$-!

diff --git a/vignettes/examples/timeseries/timeseries_classification_from_scratch/unnamed-chunk-3-1.png b/vignettes/examples/timeseries/timeseries_classification_from_scratch/unnamed-chunk-3-1.png
index a2016e7967bd9864adfa6c3b86abc013556d48cc..bc06ec3e6a13a02e586f2540803795e6b667bfaf 100644
GIT binary patch
literal 54970
zcmd43by!v1_6E99kQ78x*>raaZ0SZpdIQqZ(i`bk!cBK0Al=<9AR!>#-AK3e<}Uos
z`JL~?bN{-3TppKf=CjrsV~#oEeaD;-sH7nC7~?qx2n2fk`jw;#2=rhG1VXw%djO2=
zQprCCzM;cjY1)H8m{0F7q>*{A8xV*B^jcC3>hfhb#rYk{`t98!z3YLHkw9Q2Dtd&U
z{97bh3WM0Hn3^h&J_c43&vZQ5zVLZ2?XR~Lvgx}*y%^L*exh`V1BU_U5yq~s;v8B<
z6jr<Jb4OjQrmvb1fyJ+KPhLBpdkG&5K9N*L7JG94Pm#r_`Qe|>Z#-VX{tUH|J_Y|7
zfx(ec|J*f4+4k4n&i}u^B}B1l)>iim9Y<N*-7cxd!D_A<T3kDM^ZxUlTjcL9d;LA#
z8e`eWh&OVwZZ?{?2Aeu}{WVCd@viW_W})U9F-5Bbg(s3$Ss`cGPmpm5#HBAoP%wqa
zL>*puHkM8%>j`_Rn-IQZEc#LXR#%mXMJGWoJ|1L=(WZqzn1ZVkoI8;J!@tq_Zr^i9
zuVEE6ZT~~5_mK>b8_D^8!*v!)>pbM_YOrRZ#;?Q$nbUM=OXkTcUSCX_@Y%Q`(zn|U
zMW#WQInGP^JH+-xXPfF(OEq(s&1TG6X~5<HTIZ&4-F~`>`N!J5y+Q?{%}+l|7hN`!
zPWH8vaBFKR(0m@Em)%gJQ7!c7VjYCd#$}=BkFwV-$IHazWE;l~Xu6)j8-<%!ck8pK
zUuJVFHf(2mdQ}=M&E}Q}TYH%(nA&s}t$Mo+(x(W6K9=5Q3V6Bn?{53_hum!Yw%B(X
zk&tE}PCS3&fYaFC((C^2_qKg$`{NW-|3f}Y?5jm7a?AC6zT5o}@!;CaP>sp(cUecF
zBm<f{8Zn7+ajLv0b-`qBXX)K)7u_y=7E8!KnQiXDH@y#}h!PZ6k13xZr%TeH8G5X`
zt+rDkF;;pt+t3_Flx~)<Z#}{Ax|k{ZHSM*t0tJb?b>3tWDh`x`#I>~C7GgSO23X|@
z-Z2VpZn35Pvi96)k`0b{xqd;e>o$D+@{*9!e)*sUjNkn8&RMI4B5M-txZp1$=qXmN
zRtCMj7}rbNKS|yfVS5}mH`8E<=~tx2j@9Q(H{z^(lfEDBrD^G(y%OcD`w8Ck*qJo)
zQC8aVV9lO0X(BCJA`d*}XudK$Io0x$<1qm_&UEeV`Tl`?b-Pym%7&S+?Qb`yvQ%Ed
zAaWux8fN@uihwN9o|o&0GZGQq>veraSt3c>pR}bmn&;D#?0UkT+dGR0>*&eGYkhCe
z`gyK{B%%e^m9{%~&oiOibu8P4yAloS)Z(<#9irmV!JD6Z=OTz18TJLo%T6<e4`(62
zg!J767C*Wp4n!==&jnpg8?4x=Jc;zDL<+y=DS;%-yfu<@y*kg?>(4`ZN`=l>m}^<s
z%Uot_Y-`sCgtuaRgIJy%o3aaR2E2|F+Zbi&5_@uUAM~$3Wg9NsaAErStYqL*m`t>x
znjkSsbzTyRo?dbs?{!&|4OgyjNQ@$jl|gUTLcNs1bM>Ag*!)LzT!qd%x4DX}`3cSg
z9rBLd`lp4JxtFJ@?9<%hZH)0cPA-l%(gkWQM3F7Qv6r<yPpq4Hai4tn&==nLyQDB~
z(z@}a$OtEuYyHN|-Sea37+Ro?y&^Rl)xOdQgBT5}4+WZAE7B7dLZnx^r@`Jj?hdAh
zZF8DD`D&PVm5vAj_NK#L%hJ}fVVS|X=Bvb3kq<ww^}*RE9<Qm<5;Zk@`2x0eTk5$_
zUl3*(bzG&8y|pwOtZMP@H0UE2RC4YbpS(eAOXw)3U1!ShYqGl>1W|X0gi-`tNZ?}o
zP=K=fH96?jo}gBmmFiSmj294a&o6r2p5-XmN{=QxFStxrUYqJ6!ueXw_opt0dJWsE
z&rcjvDyIg`euVs5%pM<7J#zFq%+=Q0In$(P6)|e7;ai*&iK;kUX+$_|y_5@{adu+T
zXL1wL!#%X(@_H`zWbP{(kc&Y@qh5!7+=#2pc!b+;ba}IWvWw$tKE89s#@$$bvyJ`P
ziyl%%XQ9*HS>~`L8eXqs<+KFe>pCO4!8}7@k%!XOb%0l>+dnSf%+?2AI+D$KhBp2_
z!@BmXAn<P5uiIU=El~B(ajZ>gTq*pao7Q}Nh8D5nR4<A4vi{`*4|&_U-;EMBh?9)h
z{EEKmDpY7xrVQ;sX#l+~81z2fgR{S+GdpKi6l}F|89=FRJwi0+5^4qwAZC+sZdca$
zyEg}gu6io<PZc%2=0{8C${_Namj*Voku8`;cq^|aOG+n8?@TR6->L6WGM9QD`{uG0
zRlha6xj8u3E6eL`yi2x4#kupoYw$9_ke(1ZnPV%>zS?DGe_6z2y(n;ZRi-)Q`8?iN
z_K75#Se64L;3Ga1kUuL1e5jM~@x32F+mBfK^W8^)26F$i&k{B|u&DpPz2&|ZNlH6A
zpTo($AQSF?Dly)Z;LZxe&Y-a|C>aU4nUB%g_IPQ_)%HlXT(_GHFqMEnQ6nQG35ky0
zSY}}4sSl6k1dC2xSxZZ5cUO1f5(qY?ETijj)T!J$w|Vjy3u|p-qsuNnfI>KJaLZw1
zeLeIn?9CfV<=umW+>SnUWqeX>880uy+k+mzPMcc#rDfelr@cP*Ru9DGX82hc4f4;Q
zKW}(b9bSU<N}XXaSkbv*J5>-Bsh(Fr284ryW5S&?(vCmHs2LOQ8GqAjg%R|yv9Yn^
z<Kyy$VD#W|i{iXI;+^J%vbQX8y}iAE1}V_S$H$GDiT?QjcG}?N;-U!fCWE6(4{W&6
z9~qrD-EQo3EEx<ld%uCl0<lOJXy~_)Q8k%#DCPW_jq?vb8zOYlJCfIEQkE$GjKKU^
z2udQd+e!XX1iJ!SCmx5P(WA1Ap~=+rU$-tO^`|jAxI}-+!UG%U|MOvj3aE<&obShL
z{x-4yd-z@?pM#5|U8}p@p84G*BiG$%?Op1ssqXFctD;$SikXGh)i4~ZF2xnG5%j+a
ze%yIeFAnEJI-GK9`q1H6+p@?Zcy*BC;N$z3s(`e69c_OdCViW{`h5^LSj@u0z?5d!
z&Qtor{T9c$xW!<4EW_tGkzV(#U5^O#ZT{k_R54j!RR`L9xT`I^d+NVf=Y0KeQ3%l_
z($wILHTEb<2?iM9A8)>&{bSOFLzk9@^CUB7o9Q8wow?)qt4Z6NZc`kP3yHpm>)O@e
ztUyb>GeRaFv72`7acI@N)6~bu>TW0}GaJvRakJ^8FT6-FK~p|}=<Flm_4Yh>=DNJ2
z#p+r-tRaqMA$Y#fh5kb&earAm1PcLw(+!M21s$Sm)y|=SbCASyF`8yLBQF1#%yVo`
zRh96@?W`f>n5s|7ISn(;t5HG)7ns}SEFSX+@|R<{WmxL(U&*L!npY#i&1UnSmy{@s
z43NF6G<lKT;_7`h`tGW<rvF}>)zd>Vd#nX_+p2{(4hODVGlk?AdE;7~nR>)s=_4I+
z?Mbg+J^-QBN9W?tx?KtG6<hC>>U#XzhTbt3#$V0N?78yJyUrAN@19`1WG2^fpA0n5
zJT)u37qq%nFaEE~H9BE+3lHO9pjFo^cyA@VRu`JKY}s`FJ^J^n@arorY={&-`ZrN~
zRsF5qikOU*sC*I6o1J8w*GL==?@pY~W59cl02>@M3Lu=JcaJ=y*x<eefDfoK>7+9e
zEGRug(fD=vS<2yY<46!w@aXc^#TxmbH5h#KDz10P{e0o*b*)*}t?%7={2;c_bxg&W
ztC>gl$=J_uD<_v`AxnhE?=5F9>l_=WH~(!U2bYd0uDcbH;<>ZFdQ~&_%jFXukqf=H
zlX$(mbdpzz^)~^x8mugZkhLJ|T;ZdsyD`lOuZvLjIi0gzXEQY@+f8T9dia99QzXYk
zBl#mSe3{lunNz{@Is5tfd6B!z!OVpR{}-P3@w7mblU?DoHxsY#by+iOL$X8q8qKEl
z_P70@Q0<?D70uTEuOMqmPY?knPxtaQJ>;*L{l7@ZwvCDgV&=KGo&UeCqtC!F_amQ<
z$}I%>;sxSur~i*Xq5daY@lQ4}fRq9ECdbylrVkgslv2#=Yji#+wDr=m*O-``cBI$8
zYW98R-y!%{KKW+?0J2Ae%f#I}Dto`%#l$*Fh%8?Dt$$(3&P|$(w=nxIskXPk{`^7l
zZ?Na~Zn?PyTboEb<=c68z4{%TTA^F14#e&C`N87i)Pe}t)O?k<w~<V%MysUhWQp#e
z<=bYrYKEW;1~)-6cn5ALj#K0OGEDOJ!xiQ3a`o(<Q40y)87hBM6He(*Q<9VUgl{4=
z8_V(mdFiFjn?%X8#3|xPqF|fpx{@}VoD8U~Kx&Sem*?<&n%|<dg@r|3uH+yCTG&Xm
z&Fx|Gq?hJ+72oEX<=xeiFOE)ql|!4qx9efswYQ^Hi`R5n-}?6G5b!lVP8&o8DxD<h
z#G$69R_PaMh#$r25^)hE<8RXOj$jM>ODV<;k&Q-9W$#!09Xlt9#FhL(b-$Y(yVncj
z(i(2&g6~d`OVYn}q{h|Sw!qf67Ms^L;+;{+91gs9!t}qieptWMSU6WMlreK(m=8z5
zhGSlexYnz#wVoI}k-=2#P7)XtUONmX(>y;mIb*&XCmNS7X*%vzD^%}v9lS4l#_K>)
zYC>G?bi2^&9NZLb2;A)+2BxMslWBVOpAC8*uhX+5_Un2x4`%joo3HnU>FAxeoe_t-
zs0Sjq$*as3-dECTil@~=*SCz7Ezakb<jr>ncS1SaU-i9(8D_(<21WH1&p&k!GQIt+
zl&olr`e4p!*81_|$6pY=!^zo(tO8dvMCxf5gZj6-6+{Ar<eU}<g+BjqMyaeWeL>v(
zBO<e>l}8#bytZAVirrQID~30v^`Gd(2b3oTJvYqh-}HT?cb?uGd_i_-z(f&*db{#T
zB$aR6XEmp)RP%JGur{sXw7%8*!&vcbi6Au&q@Zf~mOJ%kC5Jl9)M_M2O(`ZeF<JPx
z=PMa!m*fcV!_IJTw~d4IdfD#nsIBayQ$bvFK_NxS=Zb@gcKjlqO&oeS2Wvkr_MMGp
z#MR@b&9s<I$C{iE2o`l49i3G*%#s0;+(CPSlVpqplKV-CN`z=E|JiKj?t4<Htxo36
z4szj7YRIX@?yicydFE?g{>)_5G#t_R=qnRJiBc?vMxLxcGc>zB`N+-X<QVoVJ@Z7p
zWulv1=ki<D?7sWb9Vx%bMWm|26s-^&Hsq>#BwWwz?%nS(9XU5?_1nvHVfS&Y!qd8l
zxaeP)<`y?n-|;MNIte15dml*x_WY9GK|-j1G9V<Cdvkh|nH!&TUZg$yqz9ScdeM7(
zotJz`=&;@BAP(%ln{|5G&9dcjzSmW?va&)oh}x?2VF7Q?D2>3PpI8#oMffiu@Y5yq
zG8_-cu(T$K>BT;w>zimN_Wu1U0u7B>{l@j>v?ee9B7$fwgvaTM%(mdHGlAY!nkBZv
zr?TS>_XVMye&2;ef%(S9^urM|wxUYW+%jIVSN{tsyw>&DJE`3yz<4FJ`&qF@r_6Ej
z-ByTZbOq{fuS&TuNm`mti-nsCX^9hc)tX;kxSLAFp@52_2VLLaG8PLs*J#uIOw~+f
z?5WM9+nXX;NYGOftO?8nGUGptPEY`(I<@>m(<b%&kO$(+AsraEBDsX}LUpgMIK8(s
zIzbWMLk~a+)p>8u_$?L2zep5L%__Hu{n(|JTWK7OjJ`*7N!jgqk*4{Fl3AKeU(yZ8
z|KH)%UlL*df0e8MFYwfVKrfP@OMCx6sowKH3ILtKeL|cC()Zf~De-TPum49ipyMR*
z9-o*9%5b^fs|BFz_n_H51v$B3J8BX>o13F>Z*ctWa1>BQOaZL-*LC>$vnPy}6UWxX
zWW57NL~z0#7gbA3i%vcXe^nZ>>5ZUg6X>AX`tTpxe8y(iFw`LV14I6*+y4v)03+5%
zUBLMOoT>&$`KP70FiM7&*JW2qv!fOM3!CWykTvRmjM5;M<^S$k|0s_iXpbgG{zrvm
z2aw+D|Dov9yF<Fa1LS|8UO-LDt^Wah;{jw%^Ivc{o;lFl<==F-W!M-5pq|d_n`(!D
zL%e@M?LGj0KK&aG{>NzScaf%l)1B{6nbs?zL1^*+9o+vx4}i^p=g|KJ`Tsd8yF$M3
zZ@T~GIqVuwy;=Vkw*KJH66^P0h#RmZj_tqcE`Q4Do&318nQ+oG@gsntW54Um(f{3e
zCEtzni?rd_2jM&WZ$$qBDSy|GmQK!Ome*?f>EwLkM;fnwZ=LAByy5>~(IMSc!d;Cu
z1CzL!nU1qrhFDzCm%38Y|5E&EbasV1-vB?Rc5RnqINfvOeB@`u-;`xKU6cQPfnpO_
z2-0xiV4t7Ks|_-Eq2I$V!=KwW`%g?jY`Onyf-#(T6Y5ZPLxhLyD#vG*SPJ%LzZ;JL
zL&4g_|MZ90U#zivvz+d{+dJjXwoT3!;()Gxv!N5}{VBm0Y~=spZo}G4dwY2WKI6|f
z33mLbkBktR61MD2%#q0?n@z+2weHGYH>B8W>z(fUbQH$<uduKv1V}C~14_kT{ij38
zp2o>4!QQeQUamB1;GtJQX&#~0qXEA4cmMk9CVc#Frojh#@-$CHHbnSgL9qXdp}=t!
zMmA|hMGt3lDquRunoK|b23cvb<7Aa_-P~9IDd<r|tW=!rQ_=q;f?!;!@>UUbIe#kf
z=lk__ppc{0{1^KS6TE9Wyk3-fc5-k)-)8U#P}awXx}^^eI8rV^_d|I9MKmM5??{xp
zZ-mI+7K4<aemOaH`xAaXEChjVV)#VS*-^;<B5~~e{Ge~k0Xz)yxY!bZ=JpCVtVyr}
zO2_}N_XXE3wD-ma2yH0(bUXnBLuf<k{v)amGqO<}9U0h426_U_1)$l6Xqrq<+t6e@
zR%8F&se%U-Vqy(?@y+m8^<<<YlD6;a>k^RRNP*-h<fK^&nd;0<3obUi|LI;*X7s&%
zoikeJCAzq+6U(ArEq9w$FY0uT=Bon%Q$JPlqu&3|NI%Rd#O&;BAIq0gMvi#@A=$h~
zC31-dNMwIcW<+}IDk|T9M9&L2OW$b0AJ*4{f15vuy5PJUreuIfa;7HB-*hDZomF3`
zXz$FnimYhlQ-NP_sFMQ*nX~bpQ8cCd3}3C8{?EW>a}j(c_mE<3!_-9(w{^}&11)S$
zXUJL&wE_zLt^s7{zh`J7egS1I3{hl3Wi)Kp5r!vsP6~Jd_i=~qvvfyEO`YmP{hyZc
zf#P9;4yEZKYq()`&Ke1#5!o43LMbe<=#T#|HNes{C7yzPD6(>MWo|yiE(Kh_DnSR#
z#<W7R47eF^-M2mNb0C|8<7*!Yd?`izlmsYGm~sBEXuuT8pD7^=uJqf*wEs~#bkHOr
zAeWJmCtE}=@6i$y)AfI-48vIndSszr3BbvGtEZ#5Qs*y0fDZt%_)i|>FNzJPKzn9l
zg4<318BX9sy-!`=6~<$^(j-;^ee%naEu9eTybw5R+bx2?osawRXTdnz_riviY5vl*
z%~ZmG5~Pej|Lhs?4!ZY#M(;pGdc>R7bU@_bM`*(^+J+aye>>?o5(wXcsi~E%UjEZ=
zA$50NbDqvDB>$Lnn~jXvr&A*(SkomP;s^)paiYuR(33T8Q~IE#sR|TpYr6$HO(wMd
zoOmU)Fx^HDTQX#%lMRE#7KWNHKw^t2U-&fqnT3FQ!2|E*%W|uae|29_lI1P3DAB|f
z)M9Cf^xV+dh&+UK+|29Z6^GEebL3%-tAY%`Qvim1{+z>;A2c?ZnU2J5D%oIC0ex`s
zne?`oGAK12(H2?98^u>k{e?Al)+U&L)vRdim#7|@o15X*f#P4Oqy4Z9KjVp%9fU_M
z3X_%fVmQ2?^$7l1oLT)1xUJldnvh*uM&>>V1IzdhrUWp`z>q7q-D$keIJ~>VFrOvZ
zf!9Vw2h(wIklapK-Jz|%KvrYo^{xJZ){(b$W(lNDi|O{Z?o8d|nl<;Q{YLd9>}o54
z2UEGf6yqP+>!*@d5)F|K+t@@><2*Kw#6q)C22JSxT%b{SXeH1=L!tD<d`}G?%|ZHB
zW&ZwsfMH*tZ9a?$Y`E!N((Ja1KDbK`+jZObK?n@i?{@wk4!87Mvd9;_Vu)jtQpE#=
znx2@lX!=lxf8R`#X?_e~?x%^_TwFejcd48EZ(bmaF%%U>Bf}}aHi7r{==A7<%6izc
z@Tl6C&&2VSbxF<c+ijqDW*W%I=rYD<L~@dn>O)C#!0S;`g|tZsZ`OKOnMb*qup4gz
z=?AAzlSvlJ6C=RvEf*_XkJ+0~M-t6ax6a()a5zA|$>#xPEfg0|a&y>4vZAI@Uu$Le
z-Lo(A(06KLgNenexP-73!N0w?J6k(wT1nfU#;xql;q78QsSEEWceY*-+_?U-B!bnb
z(`H*bYH@aYYBe<keam2Kj8$8U1NXixJANs=SZ1W7hqNZLyLiCFK+M2;R$UeH-LB8m
z(-Zv|HEbD2LPA*VkX)^Be12wT^N#8<w%e!W<>gx;p^Oerb@=&|QTRFon&f~B3F5=)
z7<tRsa;(?cv~Zo_z}SJvsNltKDK&b;?ML>CVqMqu*h2N<AQSI78)tWScNT4qcUvmu
zg>r(MV1?w(e!K2Kz0#?9)>KFJvDrfxL2Doy<`~7pCU<~OJZI2kd7TER$NtNs0fCnX
zW03AWLPh@9SHqhpjx9%@UFIE0cmrCmVi|%~%Da=M+DEdN$#JEW@y`#48XGg`2ZsSb
zb2K8?i6eq<TSu9iY;A5HxVh0qe|H&vvE++L6S+~kStSP^=xh<T>kJx-4inKkb)lO7
zU~g}~+vq$jMa1*s#rGCcx;biaT)xQF{PAOv6z8FezO;oM{|N7P);omBqGNEp&K)A5
z&b6r!Dl0oQ+F^Slcy#%${A(UsbF;GTJ#>?Nmv6NYk{RBWpW8LU_>!?~*0)j*MU0`i
z6h6N&w;OJqOa-~(28RoMu3ZT7!0CRF3q(2tJG~q&qE09MOPUh4jKst39SOyPd@+0b
zENul3Ie%}>BBc-8p85wDTifBkW7jl66xU14`irm;?=<&y`^BbRSZSDmaGLGAcNBbl
z@r&BumDVCbXyPGiObWT&>|y>6ept%**Votbqzm%NW$%za*c((|9grd8B`!UY8fHYR
z$CgpWv#>-0o=!zwNEuf9=UIT#d~|h758lpXgcqAnhfVD!YM9f=$_Gl9yVp1yu&V0x
z0;bcf2L2bkhR16GKnjC9JM<ck_7bHwY!^w<eu@>99c~(BzrAG@5usb$ouROCsaMMn
z3AqqhGc!%vA&OD@o<FDe1*XPCO?|OS-t9*JAYh{q%4_qz-5;c+FJd#AAcK<3I(3FA
z+Dh@YwKW@ML~Gb;ArGUFpokL2$~x65GII;oee@_I+P7W&^sLjq5}N(s+q$?`(%t6J
z&x~q<<0TW#XZ!-x^IU2GU$|k3UC_fPKoYgygEY0*)>6Os|AWPBxbX&6>TjI{&!9JO
zSxFh>I9U#H0dp1n@@X1z^L(g$TG*#gU$dXCKNVreDU!QeTB^Hb$f{GYT2^>uoG&`4
z5~=?QDJCwXka>iaz9svfz3H!(9<f6B`Jap#5D+M{?tDa_yy&y(c0X7S<%2DH8Z|S6
z`uPyYZU*HSmOPnUzR_r!R*LvijU7m7>@K{1L_nQ+HB3CpkGHL=vgFil5HEw2izjHc
z`TgJx4@Mb^j0{{HCP}dT>MuNzmi-hT98H&=?cXhMm{rV*KNKtmd&41TvZ(;%x<u|}
zhps4D625*drnWeLiZ<LZQGI*#+);Efw3*Ch6nH^tJu-9q@FW~4_La2DXr?7u*=?(V
ztEN9xIEj`Sn1iW+{-<nv+n~zU))8jdilKw^i3@6K9<of8YBF6yc7<iUcU_yZjKO?E
z`2xvFTyLsSdWiE=aM#q{NZwnJ4@I6KLKy>qD%J<=jJVYW7{$fqdxv=xw4o1rdd0yS
zPXHs99%&pRD8=G>zkf{2rYe31!Y88tZUNz1i<32>y~ls%vT*V+_EzvtGpB9-3%=_@
z77$cp^&3)&J5NjbtnbDrh9IX#-PD>VExn#D38c$I9&c2rw%2~FQro^DUhWx?24_Cq
zPH64(-<w^nIkWQ-G=~n)=;-N_j#tFq^RBe)k@?USLl6TS8N3*>h9m?OPvqi2TDdZ>
z+qVayu5z?*C(zYSjKzIBC`(m<&7=(C2J*EvzsC6j(^0uH7J@DA{6Z2wABiY~qfl5L
zYpmlAY*@-}_6j0Kwmc-?7uxGi4Zx!%v-~M5#E_L$ab;J)z(5^Mr~-#vcJ;VE3A+hW
zm@Ay1GEP<%Z+m+Mu}_9G0HNO3Cbcwv369F<qRNV03x5^9w!zgWZ=6sA)zVVZ4NXcq
zid<TExRS}+T%!nEeV`WCd)P&~5UM>iMO@2AjMGC6_A<^^DHjFSj}t?U$&Ou*{PEge
zC=*kL%{g})3xP70|4Xy}Z~OB`4SXA{u{FF>5466}9LBK%J>Xq6jW5c)h>F=gHaSz*
z)YeD^mo|eLmB@;$u0G@1r85d<=2RaEf|CQi%G=?PwBSKuqCYaDrEl}Nq-j-DsD4h;
z0&^zzXNo<F3MdI~*9Zv-G|qQ$Os{%zy3^!(@q4y>mISoHN{}z-83AD}oGkID2nk1V
zLuMd~SI8YT`N1-H<_+(-R(8Y14poTMGJjnxhnMy>J1T`#u`Rs<`h8D;>{V+Q;-=Bg
z-S_CDW*crjpW>gXhPFl}0-}24gh2gC{-gq3HnyOIDySftx4c<IMMa%M9^zQ;i|fV|
z+;t^nmJ|yn7uOpRp%!J`pjM`b4W|p^<oLL21*|V-BF!V4+B><Jl%M6YoksS0ZW5Z4
zpX^4y&?xlSLvA=wWY%KK?|I1$3&Cd_QkeuYQImd20FV=XO5`rL-<;Ze42q5mieaIk
zS1|8)|Ggxzm<aQc_-vey6_nVi8HpBF#B%0agx0-8!No0aoK8m!W{v&ysU`A@THy-g
z`!19aFt#g(Zp6sQD|hcm0O222Lf4E<!wM@I5-1738wL&#CcTQ21vpKwc&qGaOc$YQ
zTu_swR7M_yQyXO{I8oS>3KJUySaa(qOK-@__H<lu)K4BgfK>y6NCkdp^|LVf%<_}v
zDoPFZ1W3b9PcP)a>yisrG529{?&20`HxZOs%06}7qgb^HrHjhW=v{v0HwFgeIMuib
zwuG^=<`v5a^lN$ndbIKta?Dq~CnGb(#7?O;b#oiLR5p_EK!MMx;FA2VGL?Tgl%^(J
z$6g{~epY{NBX~MAmDXdC;N@{a$6;|#r!F*^KCF~yfu*YS&oeT9F<IXW41!T?E9)I~
z#AJV1`e-=7vtpEg`a7cz5?Akq9>~yJX-O;BxjfLLh{;+vI>Nz~(Tb9j9Tm%C5Vw2u
zb2*~>#@*Q|8yffEFfPW~Ce+2|l|$1p#zawDZd%wtV<MFh=|Y%IOibP#_3;Opd&aI1
z7j)dtgP;Zv$Ng5Ik;^WWV}*1Hh63WLY@ajZhF46z+gyj3T+=*y&ON;4Ff*s=non8(
zgzjgwW;PWX&;qDS&I}RUgS5N^e~{~}sjYoyfeVt8(p(H1#yge)<95D%5oT;Gy1Q>6
z38895186i%(k$OE)JOK2fmoNOpD!aMqd`iZLr$uCo7le1pjTxWBkV#7^5lT{ns{`P
z983`WG277IWaF)P#>+PoDBN&r-je!9tvBm$2V!IS)|bul>jbVUEQ3JC3T)2%q>s@l
zs>cH@Sf0ypaV?RI5SB<%|DmoKWl)<sq-c7hKElk+S3I`&vNLNnSRom@c#1S~QtIqm
zX8o#k@|sMfq}x4j1Z|!JG+gYHo>x2M`*MV$K|D(=lD=s{w}YSW^o5*n0cQ{=D=YD{
zwv)kRPp`RUfJbtAr<d{-XP}ANcqTTwzMulDSeUMLl6(B<8A69gW^*A0w5wcR7~QN8
z86<h)U^5|iLxN$*jTNZq@f|<}66uLb_+A|Wf&Bj2<2Jj4=0}-sfHpRK;a9uw{HR5|
zA7tB|p7)`kf^tV+S}BL%ppROO>G&DJIAXSNYd#q(;wV52Sm+f{Wjd)A{w#*YfqW^;
z10_hIjurXiHcB3M`R{0`s<5?98K%8^a+pyGZ6{=SjK!eR4H$#^7#x<H*0oc*%Gata
zB5c4yun`q=-QD0RuwVCZP%!iS&eSX)CQ7!PpyVTAlwVgA6f?+?!hK2w?j_4UMteV{
z4is_*W|F=mv}2%wVGmi03v&DxsT+y;)ym6N_Tc?@Fijh-ykEy=$0DMOk)2+^`n%<8
zlN)&F(=CB0z%3Y9V%Pi1-{UU0b}wCy)SF4e@>pVz@Rx@$!eDW|Ir4QI&%3Cs+TSwG
zNKoNkN>MzuX~Jc14dGiDy1qM%3!20%j6HIx3RUw#rUv85!hkXrFEva^<#F#~F+Zi(
zCsDbJAze;nqrcuV>9HSUB7@Awvt3Ju175x=zR9SEw4RfZc@b;qb{dspcna-*xj4<E
z12Q_$ktuBndTD>SU3hW5_&Ak3r2QFY*pYR<B-!#*3u&bp_lAo+v7D541{J6I)1>73
zW=k2=I2JJj%A}-(JMsk;WgZ-KmAAy1Z+`8ZDB3A9$gO`uHFR(0YDfB7Y5y%LI==$i
zE(uA93sQmgl;RjPCSyLpz~};gkn>bkeO%~6EoqbQBKXNDEz0n$^FZh{l5J!7#hB|Z
zXVV3h(_WBJ`_@VRP7_9xPzhLmvNZ3$F#+i8ycoKOARHg>dYGSuv>vI&ba}aIM3+Tw
zN{)T2Du=&(M#U4$xh`*+kA=1L?UT#??ysTgA1!wFZ&qQD8=@tXzB1FtCydLiv0>O<
zN{_y$b4_fXz!jd-n>!4W-RwJ;qkM_M#TI)~QCX;RG9DK+Y@%bE@rjGWa6zSfHZs0@
z1a8-~2ltg&4^4~6rV7%teQ%*>vHiYi2=D<;PtRXlZKqoh#Kn|=%X{vkW5MPIA8PS>
z8K9gk*Fv3kMXigyHND$VTz+ddZ`Rf%1p%%=AVy5C+QeturKVA?Zhlq#6lr(bH(wR$
zYmIvI+@u!ML2J0SP2eI~I3Z9#!+eZ375Wp9Kxb!Dwc(dBhjFs*0!9oESIm!T!?KUp
z;}2a7uOU8EI3S83LwkE~H!+&Hpwy+Iw@hs)fp$GvjU=?RMSQyD56C@@4|Bic+Q;$@
zKMA@#KR>7DbNTs3jtKGy`yp~JwG{7#XU8M3BR&Z1&LVtC)8#qjrGG5#_PN3VS;mi^
z1ZbAS7gbylt{&%&2spL~Sl_$&{HBH%s>W$X&7h>ol-Y*3<s3-a-Rq={f48SKai_`h
zD9C)2j^!-~#KW_FM5QH{T|d3_i;m50;(N%msp{ZC?fn`5L~f*az4#B?aJy87d6y#+
z0(^}2v~HGA%RL)kr~w(Q<C#B(9POL1+!_}xaS?b2#Fz>l`z&ZZR%2*BG+c!DN_z#h
z6-}l`WoxfG7a#Q}ilvRZ<wIv?-YEe{Mb2`RH`daSgi&jpIf_h4qZx;)4D6To?3*^p
zzZPi09fv+`*Zchljqlf1{AEzPBnuQ5`w;uXoD21U01l1_xR195@f;{Qr>6nQGB)!;
zmnTSHOB7P9zE(q%l81*ll8%-xzQ(`ccjv9K-8_jCaK1v3u`UL{2&0@oBg7CW`)VnH
z)DU3b)}N@$2~$ZLf-Lq*j`=4;ZxBU0B~SE7Mv?go+kA)0B%Rd*?{4POrC|2cbL!N?
zV3;4kd=?O65uui>y$4oHHsYw<1=P&Y>GFq_rbvg4J>~Ob)jrhS<J=LriA%7?TTUW^
zdr0FYw;rdBPwp=Ld12RgLjA1BmQJfJU1#|_(91t)hT{l-1G>X~rdgEwQc}crGC3iC
z8c^RFxGh*?yQIP1f;HB`CW_sljTf{aP0hIt5GpHy=PGYgb_PAfkl8(A5rz5y`<s6{
zWmeA0kVJ>W8m2E-J7tie<<2_=b5mrKayLn4$W%H{Jm&Z~tWid$zVM2!J5s5?wRSS{
zDSz9D+=<({-JLp0tbw^XO)<o2?>H?1M9B6%N4|X5R~=@Ia}F7$#se1@Uu-jXrQKs-
zAd<p_wT^zx)nEM9-Z=3kI{X68Jm$E3s^nv2vO7}^+2vT<pyK9!Bf=#?!^%_a_Wsq-
zo3sZJaXFK^m*0?D(MB<_i?c_j0}nh<cP3?;lZ|}m$$H<bD(m#;GDZt#|Fke!&}!Fa
znxD5~W~iz26I_U<zzu75=k#>z@6aH{9`Bx;-y2|3Srdl`0(;T;;17M{F1|uSUe-jt
zJ=hDGE4U+kzH-y=wWn2R=zQXWrpD*f2pjrL)z+qyq}hK8SF{}*_innLlsQ_m*cKCj
z6D5M)g+K6YQ&#$jBLWps0H)Yxb5s59Q>x9GOGMNIzw0myYA_h3+p=^ygd1^6Ki7C3
z(uKMh<i^d)n(zXBRCe!HoNTrAOrt`tw3Sg^KD41l#xnoH<8716O}5=HO{Rbj9vM_u
z<Cc;zO;3^qm0_p+_<-UX$!&DjbXNK2cs`oVrZZZ<s1EYcfddd|D3>Tf!<Nc6BBpY*
zG`r79lr>O6!Gz1nwp5F;0m|#J@8twCLn?O8Gjr4F>=sTY3rYfu<Mlecl1elsFXlPC
zy{ssCJ}>nU*PLl%P7O}rf>~MLR%4KReD(Trk4A;$r1W9P=*$oxw8DxZEf<EK6Xhg)
zD{G7F(<dsYM?$Y*Z`i8NFW*|aro<LCF)CRU@S|k_UG;i(4L$^uku|UPEIO+WGJLxs
zCz|g$P=~DPUs=s_W331oNe#ILSXCEL)un<!0U1-%TPnBNiHXv5Nj5U5p-TaIvM9&y
zKNp%21!X^?$6wEQDBaJd(UoK4z-H%&z%TABuKlJTw*?pb(i!p~nI+*HC8s5!({6wX
zUzHEFXU|5NBpn1B{Eab}!{_)uxOO#}Mso}f^!B2Cn^B*r`T5F`>;jRHJ@2H^I}aR&
zs9=1ycb&Nq^&+W}Zg7F8{i<OHElkK=Nx!sFJ~o?b{6g|s%tYHxXt`MqqqMX*UV~bH
za_(kpr)Zg>qG2!o2GPhE3xO%Nk(G2nj>yCm9(h}OdTw`gS}Ejn*Yx6St<fOagg1kP
zY@~YrM@Go@wzLc^j|yzJYeQBAotwu_S8|&E(bsqM67u^{t--7dQRgG%&S9<fOT#mB
znntV5;4Bo&-LZt{%ExGgrXKoR;@<UYpbq$@g4z8Lx?<Mgb(4Inxd4Am0e&S7_*m3U
z5w81KHOGilHP`5bFmYRw%Yh#CJ&8*KdZ>7VJ}HHzU4ir~zcP;+7q_x6KfvWoiezD9
zW0L0izi_n+aw|mC#lK0wfCIBf#+#e)!q?HuAGW?r$L59u+?-g;2rcaRgF{GH1h<+W
zby(ntm|W+mks@Q1uG8@eGuIv`O_smRb{-wBn4iS<4TH?)+>1bUV88B)k&8dKm<Cc+
z&vU;hxS_eVvl4cWtB!#aD$BHxmaBqlH#-h<eroDo9z_mZA;l`^1_u^`31^|HGdXcJ
zWkCNdabigr65q-v)QF^@aN%|${dWt3MOp-KYiThHCSVIn5vq)io$}&}BQB`M`a?vI
z7HPh6b>S<&7h&(1ryv;E%J_|qGE`jQI&4-8{Z}~ay1*|BaJG1(_E{15NQ_)$jw>pW
zEC)1ZQ_Ya(cXG8w#%rYPyz35H@za4Bn<W#1XZEvBHoFb;rRg9JZ0#x%xYlD)K(t6W
z3kvmVR!{fOc+KD{G<|?fl<IR={Yo&pY1NbscJ?CsIs5AQvyV$<0;(CcQ0Snc7L#4u
zW{0Zc@`UzntJmE%@$(h6^j>0uos(rhyu<y*5nI(<F4AyfCEZB}>G$w*=-PvqC3##D
zPt%oXv2n2{z6vzXn=J@Fe<r4tlusk>>POAo#{eOMhETTQg2ynw>%<vhSo=}y4S!dP
z4VsVCD+qwanDIY<zH-9$CZm6ONE(98|4POvNt_iTsa%F84ZJduPWNloGR{3In?E|`
z;hC|KD>%q&%hsDkz%{VDTRd&8(}3PrK6g%)SJ%F*c}NT=2FU4JR2KPlW@u=F;LFp(
zUSp^x)A2*%)tTMOLD*U(1}X4g6F(&<7yCgY&e@sE3i?*9KW4d|k<OwD!Sc}7C_$G>
zBvJHt0>F`Qwd|1ZCVrv7H;f38DS3EwpMVO--=Gy1Pe&&v8|FxBks<L+V6~0XIGFw@
zWtlO509KNMgp8U}Q;l6%zN09Fnu>3($7apcV_>Uc!!_H|t*!9%+m>Mg^{x#*zOh_S
zn<n3v4&8fRLmWn5V(`fChu`wGeMbZ5bX*Cg5I#R{wSUW$XUD~fN^m!NLWI(WPsM;Z
z<qykC&8~m~yS6?otfZ-)gqtb3F2-9bdL6XHv`hd2fRl-9C$AOS5rO$f*E-%Z38KJn
zCcuj<OTRcYOKtpe(BTY0WItSbp{IzEXmaaoUB>Ax)0LCs*^t*p-@PN$3=))okf=Q-
z4mR?mCO$cLlQidTPHSp<(3d9->DLR3aeM<Rd?rehhXVik*~cef=8J>#_27r%eV~!O
zJX8&JK!yv#sj1r9IyV=p%w0AfTsCvqw#D@SeoqBX|NPmhv6>-`lC|H_hNA}RL+=`5
zX{h|2Z}Huar=7ks;7@@5UdfO0l3}lCnbnyG1o3Q}kW@{^a5J8Q`TYQ|D|cp9uIWd9
z@j5e*0^%b%!J`u9A%y#lc}sA8x-kCx(D0h<&(~=J8o7}+va(7%(hxz>-#G<cOqptV
zRP<rEVPCUX&X`6cPENRWoY<a~`}&T^Fy2*FpWQ@8#t}L)S1lW4Fa6rV3&NnDd3&gm
zuKV<!i4kiuu>e#JKa8A&_r)5S$FKBsRFxcgHqt5b&Nv?~&8J{2o^KU6*1V3|`DJlI
z4pH3<ke16waBxP8g_)_LG0nn{WKfllQ8)HoOf;=r)_|PdDZZ9gnGI>78Y}ZpZh%&O
zPEprn;zSQ~!J3VoER@mD>wyjUg@ibWTD$)m_jh_xKk{Lux6e_W<ui4BqKja<p_O+d
z>$p{{GA_6P6r${xS%^ubAJTf$xfbUOaKFNv7C`WO2E?;+0Vlw=f)^CTM=W^nO|LHl
z4Kkt%f|;#&<&?!F^_P|=)!oJabRmrM6BM7Dtc0iVzWC1S>Co|+{WBq(7L9vWL&F$>
zya)S~*fS@03v;B^_MSh9Tif!K1t`jPHd9uS3#AJ#5BuV|$u=D>)wG$C`L4THd)>S`
z<vzBYJ{@=ockDBkufn?T+&o7QBUTYovs?d#(%pj=CcyUlSHwbAa?g+?<m$RJ1Pcdb
zY}^xH=9W_0pr!Nda;sqxc{+Y!tr%(%n&~n+6o^4F;nMQDs#lM%{b;{2uN1(#33DHM
zRF2S@K0Ou-u$3bKU7VyHO}5JB>kYps9gWL-#q#!v<8%0o%GOS09I-UyMQ5oPfutJK
zDo9t(WQ1lfk%g%9kwURYGi~lyPz^Rw%bn*JCoPX-lJi|x+r=0DOFe^wN75&>n5<uR
ze^*5bcBKS3lEz+H^e9U~pmDN?(ga$&(^imGEFGAb<m)VQPruc0GeI)@MG@ZZ`%PNN
zFA7n+H!I_ztA(Yo?&5{!&?NRJeN1S!`G>e*rVJu5d(+1&x28+e7~7kz+>zF{q_^ny
zjH%CGSn4G|*3`85Xpqkl?xvV5P&gE`#9^MF7o<sy8`djZRg*NMe#5~OJYR+0O=`9v
z@>`qf;lqwNmNO5C@)urKwj=gu^74>N<NQIp7r0?dMKA(PqR)Grt)2n`hdI^Ib~Nmt
z<|O!FXf_u%5Vvn@zUbhSyx<!8+(FyhyX3ZO4)CYCO)|l2t5X5iWhZLd?Bb4$!z+|F
z3*5|$B>(+<VSTT3|MU>edTW3g@RL?mLTSV|yUr$8yGZP&3|uY@q8l4*$%BGp?Rr3&
zG>{sb&YMd_nw)Hw@zQ;J)C{~}dFSPE;5VSIn0O`f*2k!bg@9lfGpw7eqxCLyyfD*s
zZ=(2Z%^TTS<xps71+uPDJ_IaoVnX<}273BAa7sWo5hVv1jv;hQNd!hTsl_fSeb$5E
ze#)*e!QA-SQKU(y_^35|1qNYto%vKH7r&C?Of|3R>62?`&6<?-_L;!)5c3Wnr}Da@
zx1r@z>yf<bsUekNk6<h+;CLAgSeXyO^-UCJEPRg4)_?{l{p`3yH7!p#TPsjtX4$Ho
zpFd3V^M|&4LR#hU95^xiDUHK~)qGI2LOX%RQg;05g2|bMj-3<(?D(+bVtR0DFl16!
z-ydk9*IHxfSgssTczQB0d>ooLQDj#o!ibHtSncBkcF0`)B=9-8d}g$ajU}m{c%6_4
zo?z`aU_MWy>{V)_>#?Z}eD;h@)Kh;_dqrD_jgtDgT&IOe?PfeCb}))?nlvl~1NSY{
z2w5^2`&#&=0Qgos^(x7{{EJJIqCf<T8^A(UT4r%&0D1^@c54n7{iNqd&H7k4&5pjg
zOPV9tX6z-{(^EWhAm$Z%+|p3rpJN?>b9~)OGN1W9>U0~G@MU7VpFzxqjfJp35pSob
zlvTUtVk6Zd>dL^<O;%CnRGx+sZG%400tH?U*hGXJ8(#F<UIDvEeWUTpqi&zS|4U%9
z?W2#&1|Bz@ajZrMgH`6cmHg89xY*<V77I;ZmjdQhG+W7ni5||+=f~1oTt4;BHvBU4
zWbcO8nxu*WMdt9>^I{oSQD|?^pW@xRY)Hu0YBd5MCf3jbWL>=Fvf0Q~VXQn>f@U<C
zf<gaNhj|05iiai5{@g~79zHx?t_HpJkvQm-1>krqg5bAi&QK@}Ioa@MiskXjMlBSK
z4UZ^oG2K|pYWPxV);+Q&{zF6G;K&fvZ#SO9WF#QqXxVK#IFT=ZJUaHBtRk}pij^ZN
z`(8oY7Y#TChz>8we$Yj$H(U1Qb(wr!Q2ieHZpw#@yZC6uFs>-D@y>+v?1m)@y!Pu<
zIeW#KVV-;<Lp#N3VR>ZLYv?0j{|5G|?;53hdmOTid?!Jl;&;9{*<vQ~5a6}z@kTr9
zb50}*0*@zkbGNRaQEVv7THzCYQVM<n`ZiauVw_pPgiQ_Ov-(B}2kGk8tICdX&M>eL
z)RLH+H{6lz`MK@sa>%yf#v}ts&TRR4xd|pA8}i4_;Bv>pZ6;lnr)nlF0K~4YzWm6q
zs>p8sLf(OrmDS2mBDpw=h<j+@V@zx`N?rk1bl1_<4se3CY|!xQ?b9u32(QhlH$%?m
zvny7D#`#ZzYH-`N^|s}@wWl{Xn-_?4YV|CviBfpXfx4>iLsV$eAk^7D){*FI=0noz
z;Bv3#g`GK@wY6W;5a>(rcDo2cQlWjBv~HopeMdHgCH8%EmLoOTU3f}`@Cn#{<2$hr
zQLWYO$jj0ag%cOMAEHLaJB>qUZtp<T(+xfnC**piR|}=R>2caj#}^Uaav$34w(~Cz
zysHoD4IW|(6N3Q2bh5FW4u?%>N_)#07~ITSF;i(5^nUra<g#@|U__~}ikI^SsiCoN
zBPK6CfFC!EV|D9sz=X=Mf=#d*>`f}Td}*p_I;~dVzE=~j?FaOJv$7vQ$cd6}*30U+
zcQ{75Po+7(2}_8$BQ)|ah5^*{dLC#$RvdjT_{o43CGEV|zB?_7tY7>!o9o?ARr-px
zjrJ{_CgL=&AmE@<yx#K@E#SzRi3vza`O|Nw@qCMd-lD?bEi<7bWIZb#ztWZwcU^n#
zFo=(?LKbWAW4z^|B0L|U`KYuhdtb-_w9<fb=Jg^jI4zyhiGrrpXwu2{UB6cUv#E#0
zzqf*kAB8Ka^N)e2vuPc0E2ou|zPi^dUD3LALO!g;VMxGU@KNU_IG{dlH2vpU0Q{-X
zj}dCBVn_-iq-~szZf}xvxXA?{eNmWBYAMJ45KG^9wT&eh_;JJZO%k$$vm$?84DK@8
z9A~Q6%w1;w(W#qc@{0m%>4<ou<R!F#8Z<OQjB;cdpkFHgJbgZ_s)`aQaqml?-QM=}
z!Q^~=J$EqrZN~<7fY#QsLqWKgBafDLsQYeM$k#&VUXmJ%2n;Ni*d;6JgyIf#EvAR+
z@(*%%Ap>6m378Kjv^UPRe0A2dvjV8e)LNBPeI&*W_`WqB7)G0wh%LNV#iLU+)lkT1
zCFrdF?B9~zKBEut3oKxFR>x1`K%h4awmZkzpY&`-FC;}~?{`g`hbendRl!q5_bQ*v
zQXGH(24|{uN#I1-85D{2gPasC<dJ)<2A9~ohT*nUBD$kN<;vM|C)W50tC<DrPq9+F
zqwTlwyh3JY*}yAmw4QG@ygE+LCW?>@pDE&a_wWZznqM0w_E8Nz!3Pr%;F8zf)9mL<
zinGeBn3-0*y#*?X{ME!tsQU@HVF<6a?8XpenWa+$4D4nwcKy=3#qWg$u%&MWYSJ&m
z^}PbA!-j74jjC|5=a<>Nnz6p^)-u1%=1MT6fq-rf;pOXJcNappMtxgdo3{_j^1;~Z
zAbZ#oCjqTl)&wpo+hS@~m4rz(M?Y#7FFt<1Pr!jIpmA=L-&5kNSK1KRvJmT^(s6{i
zHeY<Hj5{8ZtuZ!HyaV6lbXPW>hzZce#l&u9x2jD4DtfM^{FZ?TTwwf&j!BNzC2Dgc
zqs$1^^rwnoGS4H%q7@BJM<%V7YNK^optLCn5jZcBwgvwj8bgi^S5xz^;q?m-NX)r!
zI1>gRUlH~w)^E}2xEG6fSOTXApSg>@&|r(^V22I6yiP~&#0}Gow<(TFjE{R%D<B*f
zWU5s(k6l<z&Jp`!8%9At4uqq;Sq`WI4L*^@K@H*J)+}7x-d=7lPDznI2|C~JEYeg~
zsGO9BIMh-r+HvHz+;Vz1&YO=TTgFa6bx9i<jeb!#c{VQYBY{GUH)3;`TW*x~Lqrh7
zW?zut4<TpJW+uJYu(3&Q<Un=L&tiDqar%^Mr0?%HGecMiO#CFYYe_ZzR()}b=+p9g
z>O;~sVpa=NQaURWM<8Y&0<-BW&QgFKK5I@DC`~PN4GH@47_%fb+ps#L=1#jmrO;))
z4Ylw?2XsBmG=ZN$F17x7>*eG97Lqg%pHPy-e9Ga`v8)MABpFdgC2G3U>a8uY*MSRM
zp0m6<p4E4dXPQg~as{U|8vYXWCH*<MG9sQzsYv+4XL`aVPefn8E@;-aF?{5f$xwG?
zF*e3=66IJv9TCL^N~i$p;9AbJ1)&H0v=(x-?DpT8v+~lQYCLt<qNX{@k$$JT0SX2{
zcd?<z%vb~ZwOMGF*(XkR3Jvw+>*#(f71Zn{Vzz<;Lj1&<<<^h<<pF0q4~H{N`c9GI
z_yj3}&#uo4@8krq*szKu8@?(G{8aM*j;+&xfx`gSG}EH6JbOr!o`YjPGr!^c`Lcr)
zT~!5xTyEpZ*sODznikCY$x!>O*f?1^Ol???D%u$f0Vo=^rSr?nAPuq)XuMVtB3~zE
zr#>>8qmH<1+4VAl-jy%4T*F4?cXdeYUq#P|=SiDU(LfBVpyIgHwIuoZj*jmwR77Mj
zQeX24D!-_LwwtC&)h?OPhHft(y&Gq`IY}$JQ{^c%&<g=}PX-S<U-hTz3go>%@A{Hm
zZAZP|#9a|d<RoA^Py_;w*-SICWmbFz0bQ~u%MtK%S=B4)4?K<oCMKy919%E(?2rum
zBp^n43r@bz{diIZ4Y7yN7(Ul2lfYFZ?I|rdWJH#^U?NBf83s-`$9Gux*VbVv(5sHV
zwyZofF)$s28*IliB?kdLo{k<B<y3B5U>}3=>T+|$N>ys!6wAlxT}Mf#cP}AW-28Q_
zKu(tO{3atSLA?>P3>)OM07`+5t>!0zcN{1V%Mx-qsEpP4>@_&p{mg8sdXj232&<Da
zvUJ(!O(K<l<tiNn1gfr1>MNw!ouibEm(O1fh|GTavj|O<*rkgp)#K^O=FXT^f$&Vg
zvGn;t_u7wieR@SUXY;JYN8zaB<B^`CzP^VpyC0{~WSoNXFY5Reha9kF%EvbbM!uvH
z_8opbI2{c|hC{vablHl5o66APe)VW@{1c)*Yz;S;Hf7b0h$U0yRNdQpwe<xwYA`=f
z`A7$%b1R6z<<@b}J}X4c)TUbe1a&|&bF2FE4^F1a@PF7vPopYw4ebMj%oAS(!3hO^
zA=vZ#k4w<U-%e}pZb3OaFCtl3Sq}mJB`eoS(T*#~;ueST5%{AE4Xm-94+_=DLId8y
zT$<ljni@_22nql;T~>~|si%V;blPyU^K;43#(~2kq-fC&&-K-XsrhORBwF!_g)iP&
z%^1JW@1cL(MbM?BM!_12D>c(d#H_JBw+1D*mgK`zOIK1;3zX9(Dv0tQ;hXzK*R!oj
z74#{gHPE~qfd(&MDto(hk_a^9Ov<Ii7h!bNpqCs`PB+8UkGk23mjkG(I{`fA_!7ud
z*&RE3_5t-dRA6Nc>{22hiaUj=%EBPsBV4eGOI^iNYwK_In=AfAr@UW}1&EQ^S{uwo
z(MF%D(f--|2XYg`X)bS+6>qyiWqP@C!AfCX#4#okXncJ5EXZwx(e`yBf<87fawE71
zK^ybpc^P!786OBav)tAqU4Ox8Ih%Zj8E~DB4W{LM7Wi~-p=5SC{AV}0(bh=P|3lMR
zheg>%YoBh0?g2?@X&8`B>Fx$WB!?1ELh0_1?(Pz#86*UeF3F+0JHO}s&N=^c@yxYj
z?X}naTlY4Sa<#1Zug${K_)Q3!^|c^K<uIBD1}6i5GocGzV_~sddY|N2_RpxP2}o5`
zja4q;al^X6p=Qe-3n+p52aNaS;{S$+F&=$vf~{LyjYw$1Y^ht_nQi~_6N5$W|Gc`G
zQ;D*99b;?o_$l*PHNOtu&yN5=M|w3`=0hhbaz8fKes)ue`hhDGZ@@v)NC47RN(0uF
zhqvEfDTkVpgs<D@!w?3Qh7WP{d+580wQOK5>{xCJ)#YWCUf`Z9AV3SLsDwjz1^hqc
zFksn$4VH>ZGpjB2nxAz2`2I^vVSls|{HP1Y;H@`ikDJVsagkHw7SEaDZ^*>`D(W{i
zRQtHL%oLKF@G}UATH70pg+JayxeqQcIY{~uml{hsD|1}^Oj_}&gwO(_Ah5ICm$fge
z*8DZVP5XkOQ5PTpzm4$>U|v1vxzH~Cf>vQ;8MQ1FVT4T|DUJd3*;$&zeCSl49OTl-
zmJ+5yY2$AheOYi`=;Wab`SfE?RN#-pLqi3-pS?hN10G|q<#b+^I*|)93LSp3o)Mc0
z`Y*U-Nk12p$BCJo-0o3~$VYzjCfpJp6a@!+)6ZY5Kb9ZMQtru<W*5Vvq1;CJ<>1+O
z6*|6VJPg|ItPzE4@$$pJca^G#-Q{!odI`aV`Q@F1GzYbYbNF2nW|CKwa*(`M_VUPE
zodDGTE<_2%Nd5N9fb16z^}QnLy2m0jp?xZueM-Qa^xq_S+&wj&U69AE1AsAyJ>p>%
zj3v{BfDFflGFCbqI*D3<nJBQC_?u0A$48DXF5g@4_m=cc6~M<fo}X=v6y|)Z!3Tfd
zpl84$S@Iw{{;AN8ay?jtjLm8T!9>t7*xlPtdh>Z^#?EP`Sx2u{-Y3-0Po@<^!OV=f
zv-^(%q{tD^W?LPhnB3>^DDv-U2?Kc8#xzbtYggA8mF+>25G-%E{nuNx&R{+j>tgEo
zL?INP5RBH@-LwVzEDWy5^vf9&jIVfB=`{1OnaQu8$itA!V8b88KxbeqjD&u7U}vT_
z%Z=OmXD8yRAhBW~zCwX8&XoEdmvuUQ#TUHbK(1KW%l;vcKZmD)?iQ>q47TJX*>cdq
z)CCSh5lfDg#Kg6V-i~6zVfgJKpNS9C^W9k9YhCQvcy7V|HCiA5ua>XcEV)i4TGe04
z>(<kR@72HXFNDlWVhwQ-1KBCvfY;Ifn~dJpA+{Zr{hxx4l$1p);yAsUCOEZ1ZE7p_
zy}gf($Tuygu)c%-w<i|MB0p;qPhY)ya?R`N2=twJVSb1GPE2_#+Y362d5zbNNd^zE
zxFKs)>bJNeZ6h7gr1^c*_2=p`y&Yno$m#frPC(9YTvuo=Z*BbPD}SJj8$x<27Eu`T
z-l=~1?kAx+K=w3b+rM69@Oxt22gi1a<D^a(l(SZSj<yB?V`g2#8TVho#=6-ymdAqS
z!-1Y|aAClB20aULH*0Tx1Qdk}<gbj{lgMPyuJ@d^cAu2qmybc1`5XRKI=3RV_kTxi
zeSJ;1!gPKQwdg5z_=nT&sLSd7!8vy2zlvCHCGVT%)<fQ+V||a&srBE(?msd*6IizM
zQD$Q;)jxH<25Xo(ZKXz9T`t#{`W06JLBj^m#140z1qVRJG!^KQwevSOlL%(29Gd^z
z9x54(AG1zn7y<3ug+@(GR~G>+{Ucrrw|ZB`(y3;3GO+>Wk|QOCil=?@M|&mhpI=91
zB7EGHI)Exj9%_L{mLvxO60ObP&U|zr1M$1KzmHF>m?fxqEX%VbwCgI)Z_(Gqe^V{^
z-!OKsag)LSEB_7A)C+U$V}vb(^yR22sc=!kJoNc~==k4jf5K{C05aVqUV`f+Q0?4E
zB(x4>m8|`Rf5j}I`yNJU2CZ|01jdEPLyAiZ@+4~x1F@@5v6NE#EImYqq{Y~-)OFb+
z)Btve0zUr9W0nY<PC@^si(Fm2Tm+Luu`pZ(>U+l|eun~V;e&&!AfXnEx|u`!92>JR
z3871%RmY{AMM>%A<P}RMO#D=nxiH|GB32l!r$~{b*L5@Dc0G#ahdmq2Qrty{63Okf
zm#vqet!jiGT;Zbd0F_d$S+}+^V$04ht3W!uer!XMx06!SNyXAeWz%!X6oeN<wPl7C
zIq(2E-y{(*8})<I#2n&r5p^ckg$1ukcMpO}HI7JCGKdMmZs_ttX_`dv_(zYDVpvi4
zszeSATB&+40f@xVITP2u*RrwKST|^UyKa91!sWIo_{ox!50?M7TQwF__`pMihksLX
z^gRYhPblTo=bDofz-~g!fA9D0dhntseD4mQDP$qP((ql6BK<vf3@r8%LS@WSI;1&D
zTZVM}24nyyg)jiycD&+WNcM`M5PxtMt^@QIzE3l6WSukN_^tPp`?!;A&LpTbC$Jxu
zwO1thhl|`<h#m5s&UcyJcFjdF{J))oC7N@2NPJD<KQ1-jgb9C9!hP$dq()?nVc!;+
z{_9Qc^>|g#6AC>TUN}4HNN&RedhgV=l^a%}*R)`0kWQfVZ*)W^K%!6U?8dBR;oRXO
z=1Ig{ZDn7-fLgGUkr80$0Z5sJh1`GL@Z%gTR{doi8|;oFde^T?<&j9UWoNrE?|7P8
z$y}nRA0}HWI$z5U;cqiGa>{7Y$x<tdbmk)fUHCzC-s3FZ%gP5cRB}_lNz3`nzcAu<
zV)kj(+BR6mFES>YdqswnX11`cjq>TK=2CRKNtx5~+^og9|L2a?hJ2Q%9!vSvl)CFJ
zyn^}qrZ2G?vdhW}Cp^+cr9*fRf7hRFCS1^K0SPktoDmHj68~roG~p=02mlx%4!T>d
zjE0_Ywyu5Ki`Vp09Lqa9^KlZFxZ{e{|Is=Nip0xT|5-pusVSSwkeIaX?LADCS#lSR
z*|;r~TC#ZV3j01)DZERZn0WV1A}L3_+T0MYd~yG`-ss^=k6T~OCa~_DGVdxqK3rq=
ztS)}LqO_=IC#lv;T1f5*W3a0$FflnkwyFa54H-wdzZNz7?|K_}<E>v5!%6WVb%MSU
z4~Cr2svp_v)0^khu&lIaCPh{GoJ8JtlM~;Dl?*D`Dd#A+V(wqYw{26_c8@~1%Z;vR
zW*X2bDWwsmmhqEc!z^1ExYRGviv>_^%FFk$_%v~rC8O1H$vIM@^SjJ6tG%8A=j}QF
zP*|c7|MB>mWWu?9Yu||xwmvziQj!`Hc2&1{1ciiZyQ?<eW3ca@v|;0kha0=}|7PCE
z`SOU@bpcWnH@uQ9C7PWqsB0%nwFvIBO0anu%XKk>a(|f0Bm@6rO9AN1)^;za`6au!
zK~y$PK(pB0f$lEza?-Ta`B8zyA&-k;jXKZQLMC50pC8Okd|lV*n_XU&sPL>*6qof(
zQA-8)tp>|{vf*Wb&^PNyJFdnqG{&F+stCp<j^Id3WLNQ_<B;SbBR$|T)Hqru3nb%a
zP!zm|hUyAwC`#4(Kcf)v@{F#**YQENW$otFNQvK4#FZ7O*1cy=%KnCJe#;_#)Pf2u
z!w>tF2K7j&sGSP}jltO#Kjt-FBi~%*=OGwsvx^8)V8%9%4eZZEyO^otX)#yI9Gxrw
zFo6Pmft`iEW+(}miD~y(P`G1sM?36EyW`D6U0eP47i<Af`@1#q?WcYf^w!QV^^O#j
zl;kfqY_N$*BvxWZ1tIg-!Bcy|eLTq-AGDz4ascBY`Jwcs;>&`K_l4@q@5WEnf-nGO
zO+L5MyGjO$h#|}uHw?G=BKPLUdj77^iN1#paggYedm|aJb(N5DNw{r~nYsITi9!EW
z$&I9O<#ey-XA{Z01H$KfO{r><vX(dHR7=j>TBe#yHs4VMAQ)-b7dz}Be~-B-3K6uF
zu{*>PjM*9-QGp<mH&9>7Osmb${h^%2d3N!K<zKMrs>pcVly^-OW$~H?aRmb85vq&y
zlR|n9`{_QxBJneMFWn1hG@Zf4@f9Js;US1E)5c3*DwVzIs)DVl851Dyiu+$7VI0`_
zPQ|x8k}JqI9$l_GJi9c{MJ=}%vl_OK#8#X<VOhk)YrH;NaAe_-J-&?0_iGEZevd*;
zeLG@KUf;Z8_FlExLKk+MfE{bq*^a~hv8hg@478$<au;_<^b0VJu*?-NF5carl!x&0
zlJ5jw2U^#2zk&Go^0?c6Jo#G&)PirU{(`S7{^<k<85UScLv{4^m6H}#jM(H;I%_)u
zFUD?O9Gll;-$F+P=G)0^Y>e6!h|@iH?%%Cy;vU*s1R_)lkRxiVx~G)7a&x3~h{`7N
zi$Kco!4>@3u6E_oRbSTd%)!q+s?5e+{+A~Y-8V{o&&_i<FGmSi6L@_}9Gj`bWBw)O
zy8O{t`xNb0$eFq}t8RaUdY(^ZxY5Q>R-CeUuxP70H0tD>f#TWEHvK;hynE3?21g_A
z5CH2cpfHkz?uMo|H1crc7p_@s#?W}gcWlox=6CDV%A)3DD+2T26_K35;?bW<ZKxZ$
z^p!~;8oF7YQ%uOQfx!;B{<OpSef{=QyKtw>LJ%%_O4(|OOv;%^b0>JTHhhh%F^ZR{
zfdyjyP21iDn=EJlPl_(h3M`<<xx}f|TM`o+s{Vkpk8AYBtvA2&R9D|Qr5LF>Wk6gC
zNeXkhfV7gttoA=TCi*kzC>W(OCE;q1tc>pismCr>?3%)WF9QNF0sA_ihV*Kc<m#Cx
zut4q0vCq$lm1@BZOmZo<AK#bzo+7G=@?0JyGO@71P|4O>x@{!7SqPQoKGf$g;;4{z
zCu4(V=H#glunb~=nh#KrxcV6_quf6?CBW~p2dH4}D-S(B>yZqNRVmsIS=q~)o<8p+
zfpO0e0B&4c0bvhXykje@e>t@n++Tk){)F%(C3eykrL8_AO1Ud&DQRo-%|eI@ju8`?
zlC2{jSlH<P{+As~iD<uwT{+8oB)~siJ2I?(J~nY&rc7U(ua^<`7qM_<Nf*y{($18v
zJxB|tl_c)N0Nust?WG0R%r{VFcI0lHa!>LRQ%Ks_?Cx3^0C@@vI|&`eKwx1p>d&Ne
zZ8Yiva;Yia&V@_=-m|?@t$h5iqw9|y1DuD|bxRFPdn6QsroHvHxUu4t%-O%<vHSYo
z)!G!<qG0Q8PCza^Q?x9AW-`re+KO%30o^8}Uc}iH*|#45>}~L3M#j)>{|pJ&CHVv-
zC_pZCo>w{n-Mi^y^hFQ_T!XA5*e%BfXbM8(5D^AbZYa%ZS;lTWFRq^Ml|n8xuD>`0
znxbJ^d=8Ykp%id2tJ_xa-_!PcYTD}~+-3krnjvS5*?^7M%F(HXAEv(e-^@|f__ba%
zxyN1}-kluX_-P1ug+?^(xJr#{EwtbPKwl5yR)}&Ym2Ds^^1e8qo04)00qBX9@U2(H
zU^&va{%|@&UDVm<Ax@HX6CqMCI*5GPXA6Zh<=G#=OGOg0orvMz#i$lr6kbD;4|*%u
z^n!!xXJdnB^+GkNi;7Uh#4PWjy9$uU10B`hAGt@oZ=7c!7<-@p*6GL+z}65N8e<$@
zT_4MB{Slo%A{foWCi?Gk1omsXeAmv&Fe>zCmY16PU#N5H@HtYNv*KB4E`XKs)V9?w
zj@+x&!8Ww=BNwY71wh&CA~mwYpywU5sPqe($!ptTdwMW3a_W*JKn3hiUm}0$1Ip}Z
zD_U0b$*j@Qyg)kBNevvY>1-rnh34b+I{ayL%NT?AS2&iI&GrXM(so^Z^0`JVQ>xz{
z8%i`VudXIhKC%D}b_->KnaqzlJuG1|#X1A|#a88aza?H=L%(QGH95RI?buK}tUlii
z2n$JgFW_7peJ9M!D?VTN0NG~&k|U4heB$>$F=*bEwI}z*TZgrU5%v>V8L&$!pjWJH
zBKc1|KOvZ!dK(oWD>1V@zfm}!5rZ?L6YV``q!Eb7q=X|=D!&yYz3h;GE7lr-B92cL
zDw*Qw={hb~=;bCZmp14;3X{H;AM)~`?S731Qjqmi2Wqt4(K4hk*Ppa1>3?eOk9RTT
zA<~4ERaH<~{o8p*7S`TBRoi|vdNmCN?|R~h;kVmVa@T^!&^jFHOwQ=y0u!YQ)Uac$
z<_^*2Sxca4fdMR-sbP?3`g}`-tF2vK5auIYj)LAu5|+%Bm^&6f;w)rMQWFdH@ZhdO
zO-haIvn2E6AW`*lw{5^i@V(84sUwQK^$e50JunuSaM;GTkhs4y?mY=v&k*)x(}tF1
z#IRYJHgH)^2O=w3ih?*vXp8TQDo+B|e?QbhN_5_(-$Vs;`<{fL?=ddN$^ZeF5n?-P
zMOKIhYREWgz;FKoE*IHLMriU;x6J+|{O^w^DJv7Ip>rca?0EB}M{u8SFI%71!VlGR
z@#qZd(f|6%h^bqm0epHAe+Hoq-SaP47a7Nz_`#y8v7*v-P4YWx(~?%gGJ{MD%BZK)
zwWH$#5{kRdd(WB4Q5l=^+d^(cQw!+U{;uR%kqQp396v?Q!kkIL=Q61rQ$~y)ScsNk
z{JrxVB5*!<;qp>vV$?B)E%G_VUd`M+Di0rwCbvFS3iRx}_)rH^SJ`gqR&b_32|zzV
zd~KG=X&*$$17)bj^c5f;f5XNRIoO8X`F=I|DmA*Ea9#8%O%s4@mC~F8oiR$MBA)%*
z?F3=@QRmkFhdhAPu(&f8kTiRkBh@%jyK;O@y3ZKR=k@7)(;`38P+9J^5fBfbz#x*1
zaV<pdriYYF|0c(?!}QYScF!rI?#`{5(4o*=W`3#v>}CFBhNbgu6(b`&t}+m@%S$k1
z{t(#OyJrO`zlv$GD8~6!M%%me)T}HOmrZMo%S`()z-`7n_e{%N?SGc>S}8y@{r|F_
zEQCLzsZGdVb&%6~+HVvW1gL=^FvirVVevH5GPQKd-;Amo8NB{EbL5_uRjU=z0Xa=z
zl)GsWdo8&unOXe*2;^QAlED$6;Z6dvVuH7oETM_E({~~b4F-3mSsC=aCmM82OG`^>
z+6;5$`SyA-U&Yoje4mBI@88iu=Iic%{#<^io@U54OdGkcS2k%m5XexOO9z1gX>b-E
z-EYrnO>|*$%${?Hv@@0Ia1@1Ihcclv76B3l2EgJb&3tzGKl2NDh(Tab!bkW;FCYn^
zz49ZzSt>L}I_W0;$U!33OBZ>2KAJOV7AFfG>O!>TBry!@JtY8TvyuEa+TS*oE7jZ2
zFaEt+xqV_41DW^#ImeyjdYK70Drx*f2=+;d?1!~Ul93g*8=r_&K3&HhuB_@?Of$=5
z^H&f(eNe_G!URt`V{%_%vbv$YCjjY%|LeB7^|fRG_j)q+X&JVFlv8OPy*4i|=`xj1
zR!RnB>8o>v-RE($JT*)<ah-o~Pk9<LK#;6Qi34RQ;22n-?ue?7+yGkrm}_a#Z)#Y+
zXfH-?qv83$J+MX&{&ypu%9-SUt&RplNu6=$waI^fN1&$I=;6uT^$8t>@;79A`wxY{
zIr;k`l_>)qQtVW&!N1wB+kSE{J|GvcH0mVR6U=144Ft!YpF4gCn`N8%8zTsO*7dL+
zeL;LeT$DSnEOC5AI|z<x%XJ=4Xz$S1VAJ~i*}F6H=#V#q$>-&%^+<4en)SEV-o-9s
zOhf1z4}VgFwCed+khsB$ukyE-zR6AFJ<M2SbnfnkhYc5Hox6&PP4i|EEDmNsz-jn=
z%N;0lt*gTnQ$Ze<8u<{lvzp^RH5KqY(I6~GYM=yqnPGUz?K}a+S8~h+*xG)uT}(b|
zZ*2dadKt`^j3RyVPCJq|@V*ftl57hLgN;n{=YA)UeqjyCl}}a{5Ed~fe-DnA9=3J(
zF}JX=kiN{ib2BLHK78X>sxtD|cI;v<neW5jEa9DpoD;v58=nBTTPiMJ@3Q3n^Tf#A
z=aE(K$QgMb;n1*|0YedKDi^0$7lL-a4C!{&gz$JYLtz|3P+xri4@@E#&Rs$fPft(0
z&cA=Rxp^Bkt<Zd~R&|CgWyDmWh>l+}umU+nk_tw;19CdB>^vKttsOpSrA*1_5UATO
z|BZ=ZRV%0q`W0!c-qgeXe=Gn)sjM1gattoXp+~TXnKs7?Dy&by<KTOEe;!X1`@Pho
zDurM@e#lE-e_{U~xE8c9!g|l~pN93JGrDO0DeHn7l8HLWNnbxrmgya&?Az<t#l=F$
z7cf0QVSk&we-T8z{xP1A7=gK0U@Z^%^+lT<Mi8?eW&TS&0IQS;Poio*P6Un-rfCq)
z<xdds%=e|%H)YsPMJz0c9*n~t>$RGqP=Cv2S6oYTR?PComZM_&UpXLqQ$CEtJhy%g
zNd;%UgC(@qMk}#3yv#|~N+SL{;{1?bWaSER+OS64W7~c`@aS4dBy}8=rn4y}<+I2v
z`OZ(P60PWiYE8A6#%$hM%=_}-$Q!!G@jnoFoe{2S1J-`tdWizPL_paWr8Ma24GnAg
zkkuRjjxZ^2s~+nG7-LpYF;4$=NEE9rmXd5?kPC$FCO7>~V9`my80<mF_SM%54R{2q
zMhLWI`CD70?WSjaOfdK-nItHN4>m|4-Qiy=Vj>%SzM;5r{hezYA&BAZ75jkU7I`@>
zcuSY4I=VwM!uRHK8oamIZT?w#xWR@1l#93eRP)We-}a`?;C*5wF9Clx801x!Tf<hW
zodMv3iG<Zu{9Ii2To~7hf7JfErAK+6N!SpA;8f<`hyZxh>~Pk30`Gc=JkpND2vhPs
z>w)&_OS}@fF`4ne>Pj1i{u<?&)nEX>|93MG;|8n$B9zKOkb%zK*JK~`(<pc&NnOMK
z;vuz_=wlud2HHFMI6}DC-Q8TGjuzl{w7D_Hk*sq#Xy&848G9iSu=r~Am~!!Eo2dK#
zUQS}~WhpIIx5bkC>-T_FD)q<b0RP)gs;-M{s_vH!bMf2rZL=V2kCI+xwsRk^33?k!
z1w)F^(CNc(dxlg0mI32Oo+n%B;`3B7wAV0X`A-M0Ch=shE*Yn4gB9fDq%6LQNix!c
zh}{(Xw~z##U>O3Bf+E3yZgrQW$Y?ySF-lc@Z)GQ$0h%G}MIw|dt8f?XB+ti0XaCe>
ziwksq&Y59<jV4J>E9^Aw%?(Z$=PB7YuKpr=`VjK~EhCI2Gn-%Z6;y=KzC(?SkfG>L
zFoqaau*pz6!sw#2ChSaQayr(0-pjJXdQl3EzFXbExz8_H73IWWQaeyPo$_RP@<Bx|
zQ*yxikc;3gCcEkQr@367r^~2cIi5@;Je4}^u(P!w>AJEyd@{4|FZ;@qmP7PlD#5iO
zh7ep~(8J*HUlF%=4A(n14+Igit)kbw*%exdQ;{+YlHZxZ8ak2j-D&CZrSgzZC-Rnb
zl$jjARy^pz5w&TnH^K6<N^I%0;B#{xY$g(j{_11gRm&^k^S|Q8J`dRHCsohe*at2W
z1K%&iu$Sf+B}6~|KARIiRa*=$9zSv$2=#it3g9yBJdX@Bwv{Y}Gm+e3`J7qVaMW~w
z`}*uLqc5v|tE#?@^xP$<_+xzr{R9fL48&)>RPi`4cS=l4NyveUR#fm0llg?-kg%3_
zyxAhLZ`q9HF!)C<2ppeqTs!>OyNihxXS=a;sU9e6-9A`-XO^jfxhRi43cBRWx8f&M
zAF3k{QBwYS?A-t7HRCg7ybg+YCpi`0d^IsWe7=?7Lr%m97a2}}pJm6H{LMLE|C#Er
zL`}7vSBsTpX_51}@2IG`%_f31FX@obV6PRO%0V*|$!qcv+sNf88}<Pv60dI_=;bAn
zh-GD`vgFN-q4fVv(>3$3!w;~l&9lr6!78WFq=gb&0ZIA3O`R8)$(y}5?T5UgcSW*V
zOb3!n0TMJmC<z$PPcDx0Ujy43L|^JcpJFm<GZ6HG3!8L&6h}xH5CW@d^PD+25^T2;
z%2WD}m!krGV@tTum`H>%V5P6i?G)Wa=L3227{QGiT8I{`W}kaiy(I2#gjRhXn#rC|
zn9W5hIB=zyx?WzKeLn9Yzx~mDo%Y~=b+H)mn3Vgx-hwB7bIg3Q%>CS$+g-u4>*?jy
zEuY`m*!bf|>P9PIs*8iB?aeJFW@>1-Fx8ims&-bZqaXHfk<Lk5i*WCNiUQtY7W>?~
zOp~6K+({UrGRSdvCt|+;X;5CiNWXP`{r9uKRp#Hn-^UdE!C*4<=*OCO`3?nNRBjkO
zyxeDIOt{H6Lql5|c<+o+Q8Cacax*xt^`OZAzI+lX9US}#d><56Sy8cmc02U0sEj+d
zWfer@59wBgfTQK}f8-`0nd>$!4sqe0%zs10-faOag|Dwa4wRB|tX&@+9Ov-6B}B*c
z9>t;uN!7GxUfeYa$@}|@l5?;{M#nb%*xQRpPGL1GuZ(JLmMa9)ni!E&V2Wh_$|)@g
zBvoD~NsFSQB3eE7ra$v8_uY7YHZY_ZynCW+&)$H<F^hBK=i_;LBwu*9T5x033W@V-
zX%ijo^LmWnfZRfsXq`KFo_7oLfmWt|GWVDJxy#ysQBya?`o+oBSKC6zd+Bf3H`dR*
zi)vyQySrPzQ-O((YoQ;Q^)=~^p&rR2V|ZsS3%*Py5*w5I$J?Sgv_jlHDQU(Wy^@k`
zj=+e)VEBWsXOWTbyjl`orndrDYach-+BXCQF7MK_vM%p@Cv&N=A{Gs*lX~+!$Tv2!
zf~WwylI3Zzbt?v1P}Yq7mer-Pk57Vy#@X=vyob~`$7FJ9T;c4H5Db(er~7+pnm7;g
zYu>rF^NpW7xoIcNlre}VqBMB1F`V<a<AGi@j9tqoJLear{V`iEA$_iyoT27VZ2m{h
zlcmEWJR(NKRCs7XOS>%kI>`{sv)S4C$>OC^p~3E_qj{$m%if7%__1^B27<FypsS#-
z_$!2U<zn=Zkc0$+^=cT5^=^$~wN0SG_or{>f6jiW!m*h`aSk6XqY?ZicH9)nx(~U&
zXMDE+%kzV`98=pwY8tiAa$}=0gf!PT;F6NpXZ!nhAD9PS<74La9PF%+8m2}NVVrzo
zkfEQ;kH%`uWLV`U5|68ZS6)m^|Hhd?w5X@r2c1Wf=H_O3XZs^`qt1azG&!b%x4DnM
zs~P>zH~;FixAcdh%RA^-GYUspufW4daPUH=W!g<mO*2_>aBz~-(%ivhG4e9`8d!L*
znVFHlNqTv86fyqk*3u-{-;Mtfk&^uD{y{+j?#3g243!O%S{N`jMby`|q-*MKDq@^m
z<?4j#>Z>=m5G_VW|LQdBXNM;c)a3Q7okWuUl=kyqj`{LM(~tr;M0%-5vZG4nSx=Kd
zNO0hGI==b6zf|{iPWInDNrcBe>Q}Gg<O}@+s;Zo{eXsV%FPmC<@;(>H%O`wj3BIK=
zdq_K885Izka?#b3fBUxZvb4@Z%J*a?NzHA=4V#`I`+|=b_H(&jgC!Ul-D+k=<U?1R
z$n1il4!@Yl@U7|Dw{PKdmA5*Yn!fMf-#9zNrPW#q*UvN{Rc9oC{hYP+I(vUV92p&U
z?ldsCAtwjkxe22N#cnJe&yQ<Fuc({9(l`7aB2D-AC^jxuPfzU?B+q@~9B*I{et+Te
z)1fBfb7wCJ`#M+WFR@Dv0CRoV*Eg-X9oGn6ZLQ@p(i@N~oG1rSksqX_1M>30E%fN1
zyb7fy_z;G^2JPdfXzrp2zhuBi$e;cVx|&8lGh<~FK^YknP0jBctp<jM9W8T1G3F+J
zfJrBIQB{*a`gF%`ZJjkjk_&8^2O^<_J_2mCekaEt8nm<$oLN(ZYhch`$#&i+axi{I
z7URl#+M9Fz*RQSa?i$o%qE$qI-C{yZOa0i>lf203q-{XH`e{Ijv-{D;!%b`HHBB6_
zB5=`;^i4EHSG1SUogWhvHdn<(C^s;}nq1ebC25UFPPUDV3KU*d-;$7-dBb}6@-;9H
z<_6WpQUVT&)7EqEb$}nfy}irXqVjkXFj$$IB76TJ9LvrQ?Df*!M)=Lc26)ibbvq)G
zgmhqFAiM<&3wHhx^>EYOH<HS##x)qy2g4Yj>2JLq9Msey#txCr8YH5kAzd9F$;3kY
z-EQHQSlNUtFGiCY6D3dbvNunD+e>$br>3ehByOfjaB{pIr*PDO8M6KrYGl>dmo?be
z-A@FnmZ}8;7Ilnl!DYESgr10Ka!&waZYO$5shX*ds$-PJ#MNEaYLG`(|NEP|S-^hr
z%7*`q4f5}hLoNf)Gt@q^fMbbuA^+#1K1w#A@9OZxY2kc7=;h-5hS%zO#mV(|>saB?
zecq+HxfZ8I!OEF^C}enK1d;Kf=in+fIvNM3HI3uUpc)DRE2pJi-zYB{R8}7KuFV7f
zQ<G`|9ZkgWI>(LUV~rVD0e#eU+U%UQH~B~!c5Plw4I^>jx&G$XHwl`vlm<zJm>9Hx
zw9YDie(sufSMyifqm5rgw{CB(b##BPy8XnDfY>|yR!3dN-alGwx-rq-`2KxuOnq-{
zEpKomLSFNFjJZnR#P)Rc<Hv8ERhARKP}ta;g+2k65lvIOE!$%)u`vq~?gnONv2sz;
zFH1nP;g9Q2hYK49KeZ0HAN_T7=0`?GCT3H;mRv>9&5a#ifvYNwMFa=t2Sh|NgaX9R
zxcZ!5b{P&d74kbevTj-?@37(qNpT`;`33)(e{q7rS>peOBI~i^<D~w{#j9*?n54ww
zQG?Z(mX%BS_y{X0oxgrf`iA51SkPm{fP>}piJ0%p!#cb7t)=*>xlyF|Qh&hQW>H}S
zAt<b`-i^uZOZ9svvw#4emsD#jB;~ZoS2qt6(>kzj-GI$N(+w_OwQQ2Dtu5~xQ+|FR
z+>wzn5E88HpA0tGFzK;ZiN8WZBqzsOat`hr{<0vwQfJ&!&_oRUy_lGoS#3*42N)Ig
z+S_EQq^3s2e?wg*ud8Bw-D6li`JZL<;>ZXm>AJ?qF!-EW1PqFmajn0^Li*NFQc^%a
za@pi@s#n>!k?_N{oxp3XFRg2{btmUM(}gq?>1~xB#JQWA%#;|i?GO+VcKrF4-^4Yf
zZ);tG1joq1AtXRy9#Y@uy>!+byT}>P8u3VCehV{fG1D=7)&*<;e`GqZul{Qs{Jfp+
z^EfyrHiw8nol;ZJV^Dpu!$8haT-%ou3#y$;7aDC-M|AjDE=EYFrXv%yvFv423RJe{
zs#_u#Pchjajd5{j69@)!kOA|z?K84&(@#P((~#>!I-%o3*+Pe3hQ`>H&2{+w{afpx
z!GXH>iaD*WPSw^M@(Y8VX0nBKMqB!@4xi$rq)EX93^w)<A9QVP7S65Jb%-o|#avt;
zTvC?=>6^hH_*wjvtdFoX9XtgY9fYCpa?V$HYk{|?lA@?V09srhTY|bHN!*@sz4E!J
zUOZU!x2bG1%?6#%J@1|b+*$Q5dYQUEw{4B$#zg7mp5L8FOt!6ezntxo9f17Lu+595
zA(CuvyTik`01lonp3*n5BEJ7|+?}SBu$pLe)N_30%~;H2X!9>33cYgKWU0xo<y~R+
zSK#GT5%a8L?Sijmosqh1J`Bir?LUTM@|$=>A|pOqzcyt{d0WBU{L{x*c?6Rg;EHBo
z!UTymZ2{f>Z8c+IIW+P@pJ;f^1u*;j-%5PPZ<EwOJHlZZv0`WA(AJK86oOBxZbCJF
zP$LzIV#T3bTYGv|&GK=y(lZQ=eVBb7(FQRgH<r9^`oZP~N%j-kG{pFkp)^LEPQzGC
z>*0Ck1l`KME4m~zl!0dFN;0C=nFQZ4y2_<s>H0CgVnRYBUM$_jUE&{BM9aEry1>lt
zlX@L9U{$ve4hGSO>Axdr-Dm5#uUKF4IpAz+hC7(Ev%&0I0bkf~eRG@RU`SMIn%SR5
zSIC9RxsZ3K-yN>^mG0p}5=)-dGCii(#peMoGz4=^g11z;=<H!ZD#}{h6C+7WI{Pp0
zb=j2ILAHj6Yqa~=!TcZ6{a7S*%egPC!&O|Zx-=@aCeHAJp^Ln#kj;&2jw-Eui%PR`
z9fYC@Wv1HuTL=HEKX2c@4MWO@LLjs4E`S%*ZEq<VWg}>e8%Qf@@#7z{5b;A8K<Fi<
zhI+vWVk%$=w~;}mLtxE;^PAQuB0C2@HHc$z&TgImWEpHmA)WX)?=s?7<TCcmX`V>J
z#*vK(wiNY~=a8!}LKJN6s0h(9bW3*8(6VX%3frEaCP~Od%vly?URBBR&c71+c4XA!
zDt1sBQ?5FEf->LYy~X)zR!d5wgr=!CaCNQp(|U5^wJ+Ut0~wR_+7;|X=pCe_)K$fz
zbN5(IGG~-|YwFIoW24<;S_ziFtG97QsxSoBk0@or2>W+vgTKi!(B|(=`Kz2NfZ+Rg
zNlTi_Jq&CWn3R6`_|cvelOSQR%=ZgN!ACe=>$y@Fq{yuTd-X;9xE;?*9u9D?Df-{v
zYBDl)-`k(PJZ;|*omf3bl4ZDri{0Wow<-FDOsEP^)XE-lVS^zEaSEO~7Bev{80Z04
zL*iT#Xw7c?<x0gx!L(qH$=x+t`$HtSouiKT`95O|5)>l+{P8Z=uAb#s0d>yi?w}HO
z5VAVvQJA_@yzQTY#Z)fqC4Se){3+a;QfTI-myb(u<<z`9t$dRdfVyMde{+5AiUr=l
z*(2%tNr1}>o0#HD45-nMgC?iR95a2*pzNrw!+3B0<>;tGp_Bl^F1F;KP0C44{auLb
ztf*4ik>jrnCv1Xh(q&p&r@VFJjA`SQ(cku|wca3et2ZWr)LeTm!?qucFi@&1QwO7>
zs@26m9wPtIBjEQH?|7(rASZ!IEu<o^?e8nE?0JZ!A)-A_OA-R+2{q=nqpT^|b`OY%
zGBOcL^0jFIUXzh}mMh}Co(nm6z1{AXX0|6$Z}hvtAVv~&z^Gh$)PooZjTijrJAruQ
znjb6$>rF#dRZ;Q2HP%dm{@SRMzv|8ywD4CV03l2lkz=sDNnc<8WnO1T;Q#W)#7(hd
z-tQm>f2AYMTobdAU{#aFRd*z_`qZ*w(t@n(f_ihOW$6wn$L7X^yv9uY)0p*nKq7o2
zX;+`ug+|cH>X31I(fIJXY^95%tEv^5FeY7=obHIQ_x$tL1fgxp@ER>;HOy1qn}44u
zHX~y%b_E8z)A$z1LtaZ1OboXm1f6jsefv5HiAcvBA21he=QEf7=GBYZ?S1vzffA2o
zhs#VWy|#lf`U_X$L1Io$ZwDm<0Vt$MS`IR=+xlA;l11XY+=cP-EEFrkcE(umOPoI&
z``+%|OBXFu3OdcL@o37WtsvF+(KZAWq&oh}>4}m2g^qAx1r>GcqP*FEi_3YF!^z*C
z{cX26F>~?6#O&G7_+a%Cye^&}*R_;$pv1%m!fHmkl`Tz-bgK6>c#r-;s{Em$+!7t=
zQDl6pB(+wFh@>Pxouq<LKkH>ORjE*?`Rka!szr_n4w@$aEhy(0FMd#k<lzsRP1Sj)
zqDnyr28Z=Neiq4-Mc*VQi2Wrjz*v3?*sD+p-M{j^Rw(TVNnyX6W0aU;i0zrJevGm*
z<Uoi+hdX_N?iB-g)#+}bRAwYPOEK+`VHAp(V)5i&G;x_fDiIdp`PraBv(ob$S3DjU
zc~m$zKAdGESUG-IT8aaEA-J$pVPlSB_pusj6l0JhQKt+e%;~qB&?+yHf?Isg+>7XA
zU0{FG6CXBKRxf@}D-;VmMP~5yw654B4Vp}}B9I({kO{3z9kORBv%Qkju)46#^82)o
zDL|dL;fB=LH^lB{&&9MgU{3Vy>%+na^IG47hf66^TH-{Lh)oO>jYEO8aV8&Io}Q>y
zl+44!U++DMjEtSGuIxR+!ieX%6MK3C%mcdF;Cb(bWBX2CL{7p68K*CEXuz()fAv|I
z!Lb`w^{p&WUQG1dSb3C(V$SkdLN^tap1tJT;sBTKKI3B2Ljt@Qe(*=f%jh5INF#`T
zeX|;xce`Er87Y&+EQ)LqiY@twcw#6BM@PdFY0TiyK^(t}ybgQrc8C_k6NhhaTaAK)
z$o>^4<sWW1w=}vW?0t+TrApKxE-{NxY^c+4^(3It^PrTXHDk<8D%$)CU`*__r*l+v
z<vLF%SKg;m;%>Azr?ip=N_8#?zkZfbE@ON@Jlet%`}Zc$X#NLo?n}pGNDhr$t+@DW
zDT5isxjX>HkHJFbs^?~eh2yjn<Z^w=)3+8GKMpQHjYIdBXmroag5EwogX5qq1A$@A
z_kR^s+zB4`xn2%msweqB=Wwnuk)%!fIIoC0F3^85)7z$@LL*0}U@eE5Kqsw&p)B;^
z<oWNi*mxwwccMQqv*iwEtAQm%sLnFevt`9>3c{sx$-!G*rsV7FNK<-0^&0Fd`<^){
z`_%Hi01WQN)L_bQ2Y4>Mm5{^-mXDYvlJ}(I<JZDZ6~$;2Xw+L5=lC%dgdnARpEX3C
zYKL>s4~#JiN(gp?2XE}gR|5{)4Rmo^Beq&?qUKA$&Yx#{(k*y34j4Aeu(N#$1cTK@
z?zT&VWn{)6o0yot$FskGZ^2xqqt}C#?UBBG*$TF{!?@UHbrYkcMA`Wf3$h`?vtBWO
zpE#Pz3CK!TRIZ&h@jA(%)5lJ(nA_p0!OWP+U~79UDSZJyhl#|<^;i`SsmMs;h<GC)
z@KHMKYOD2{dcnJ&v4_`w?IQ<JR~zbOjI+Wle<b7Be|C0~bLab_TsFvCiiW1KrpgA(
z|JvY0Rdx^+r^up9Mv)vGlzt@u318=4g(M{FTt4|Y*so_c(}UZd2z4wuU_2v?DUo-$
zj(=)dl$tpN=V*VFrUhXjsI?FKf<lF9!z}|hEb8mrRup$d=6LDBip9i;a^i6eMRE{`
z+P2T1M7$ZslnBiEOmZH}?GH|Z#tRQ8>bd?PcY?x+Zq@IHWL%-2G*e%bSddmUHPZ#8
z(*h4UmH|c@PuRaN|7gE;K@oK9xft8HHy?J)H8vK2fQj@fXLj7fH$9>Fy1JX(1N>Ae
zZi$3p$~X9r*DbC<&+uaL^?JxFC!AO1GY8opI3rA=P%<)#HvMe|ShD?zSZgs`SfBam
z$Qd5;`jIkm6EK*{b#y5bbcP1rXlq{<4Oim)ON<=i2uZ0)G%s7;n3}5F;VUuEHVefx
zb7c9bgZ8ubCsIhr(AW-Z^3sXhXxNRT*w$xP_81KGGgGvnuP91Yvvt~huwJE&cq>dD
zD^AO|Rg)Bah)i&c*-#u1eSma0bSS9nv!NDj&iH-^1AXz@Yz01w0ZL&bL0uNvk*KkD
z;uG-gG5;=~qNGM=DP9&1{gYd(aU&zQCBDObu@Q+eIF3I&9E}e`yVc}&CC7m!Y~p6?
zZ?i_*5ybA~^qmbd=K$|CHN95rtU#Oa7NUt8oU6SFIQwji48rwthrb6bo0r>d(3SE#
zshM-Du1iE?G-Zvc^lCrC8l4a*shW9Rivuhd1${TDDh6r87M%dI1gq)VnEvqQKb<%0
z$L*Fa%<|#uRx>ns`)iPz|B6f}Yr^hcz%do+XlkT|RK4ODp_$6OKo5_QpO_7uQj>A#
zbN-1+W`=>VGRyhx<Vk14+dFh_rZ`yun3pU&CU~9;X-%7Ln~W4BxK33eOjM`YgZ0yi
zts>i7FMBWzLUgpw)74#MtZt7cLY)IyDraE}FMavx9j1Ulgt(NFqFf9VqbZn6BE^*O
zvNI0@@){$|8H6qcVcBq@0Nh<(9!sSY=exx8v{pVONTW~^fao(OS7P1AY@7!9&{8_a
zp#cG{A432}=tXfD9<(#AYw~J&CEP${W;QaNje{jgKK=B;a)<m%I`q8Q-oDCD40Two
zO;nTNAl=}bUbT*=H9_oBnvOwQ|NDbQ^{2h(2g2LY7_uzrlmp+i1Lp=1%GTw0NtO)F
zY^)U^#K2ikbL-1G>~<CT9We;XxcI#*uZKe4GZGKYbtn6F<%V#qz3u%K8<GdmtOb*G
z_J(XjZy7?+z8pnr;Dd1E-D8o&UopTYR$~7Nz+ehr6tiS^1n=~D-?qLpRYiSAx2s!O
z=S0PmZ!aZ?;R@clQj6dX?4e?X<>wU9Sx(URr?1o@4$@#FXxTe)Lr6y0FFxw}pLgdZ
z13p<r_twB<%&q=t%kUOrhvkb(RO(qyy~e?1{Xox<I4sna@QSh%dVqI!*obi2iw?QB
z`^=tz!}7Vj4g>v{ELI-@s5EI}JZu7ZjryNn1QZ|JS_J!>c;DofJ3Peu{$u|aujQF9
zd!BBczrBNHnnf82<SYj83K$rYKE;v3Ev{UM0zbX=vg*gWQb&^X6!3Y=Mo)x_+NL@g
zZcRGlk^)lOoLdY<58wN0`lJASO1l{m7ol|-5G^`ovMe=%gQ&dMN0|Nhynq|(`p#<i
z&WW*L_tm_qx?$Jn(TI<pQKNL+Oe9GDA8b-W#JE`L!JVC_cR2Epp@(<rrNe9OzX?HK
zkuzyd{)O=U?Qv@Pjdt)?tw0u)Em(#;9G6ONLHv2%XUX*t=hG-g82-gviq7V!JS#T?
z@Nde>N!FYG@g>J^l~=0~0gC0YO0v2*!E)l)KD@zH&f!El)uEEtl{0M?-;^IQ1HV?2
z!SmeTA#xapnzk2}0z9evUjiEJr=-<@fG~&b>x$Soi=@Oz(aPd#RW+x*s}^hj7$-?7
z2ud6{5(?NeByib%g?3y5RDzSx>TMs%7<?>X0N9%90Wd}8399qPh95=r%wT7)L;|2j
zgu$cUo|Z2fCZon0K95LU^C8Pp!7B6qvx*v4H)78><lUpKSWzYMHxy%V4F?xiBUkRh
zq;&(1@1?BecHRL;^HPk^aUyO%ITja?A3hJ-gvuWv`(kTQQ^RMDU#LqH_Y-Mo9}P7)
z#4ppz$542GagR6mG&T|^;qejx*84K+bR+g$+ikk@xSNy*_n#F8t=vZWo94_3J63W;
zQP6_L#7^%@VisQFL)eCuOg;to-8PjnN6H%h9}5tX`-F3|8mI23bWNyBgq?yYZBQm8
z09euc#-TbtTx+;Db{wmLsFx|<yHCIGjfP@;Q%b`oB0Ai>tLO0#CqA?hawaByK<iFq
zrDSK1HVdFG&B^eFk+C?h;PgUu@ERmSHy~*aT924W)H<={#l?(p+8FLEwctKkj=iUw
z;#Yp>zX8sA?Wjcpnp5T!GQ6{Ub(7@eIx()nHv>XgY#Cw3H2V3*WX&Cil{~YK@#rv_
zFNK7C&0~GPGG(>EnDk+{5CBKose28?b3S9We}dn1f7l*9Juq_G;7WjE@Q~X~mKpV_
zTlyL_g+sHaev%Q+277$5-%g0i+hIi2yj|L9?G~leNPT<U<JBdzboEb{?O(MSDq|Jo
z`<6yRetw=Q-^b)F2^KjBKcBpmF^3jCF(UmYj~K3g4h$sM(Vd=h-ZW#tAzvfWKNK&`
zlb`xsUSygNGwe}%*l&Z&NUPp};|yYv38<4dgKtYM&6}Km=LiXnNiyG6ePu0|OvS{r
z7}wg%w#9oN*<MqME}rYZ&x`Ocl3}7e3mu4F*0A0_+pS&m?VG~Vut?eCRdu{}`{2S3
zlbGVhI%|cxdd;KPbA~f#<8@WDG@=Fs+N{sh)4@*<6p2dwkz@X(WF}urb4i~=VyNn!
zvsW}<uU+J~q)oj2Y$2%zS@F~VQRD0^uYe9HnqjY;ciE>g!w{I@?{IDjLqn5+uK=p=
zxxv#AhA0hdFHjymjSHad0vhk{4w{no{ltCGF=3WU4d#v8b*Ck#b{K-0Cf23M&&LD2
zgAnB|9k+FuH1H<^WcFeCrKJ;d%tu9e2;EPd@O4R<L>=H$r>m4nI{yWT`w$ekMXIy?
zSMk2cFE@$G>v$4I`#4r3q(JHB4`0OHvPE!^<sB2fYsU9r2A~vQS3(J&Z&8PSQ2rV#
zGCsre&l*j$m!Cg)fL<=4#jby1@Ext?_rk%AO-$laoRE4dl5zoQNbn{d)8^#Nf6#)%
zn8e46n>V);2Sj1H^OteC&vQX>%bsHr*)bQOlp;+5MY$ZIf7F&u-fQ_~oq|?N$zPp=
zgOYWKb?-tgnin4#-t#^ZsjoGRl&RhT4_q^MBp@JutVyJAbUYQXO-)Tb_dJ+(c6-7=
zFT*htj@}Vt9A8Mkm38O4KDqHOlR8c7{slB6o4>$Jy;QH!hp}Yl+jRBWQh}*y&$P5G
zxGdQ{NMhPE;<%mbKE3ItQq@kw?bWOF?2mNIfxdHKL<VlLyour45!$d&Tr9qze?|r+
zuugO<=G;OC(%HF<ezHA!ZHv*2tzNuRLFd=+8AR61^(<rL8UMafMAM03UG!7y!@oXi
zN#DoB)pX-!DaUxJ=24WJx9>%wmuE3_L(FrMC(>lK?Z(3$|6m|x7#W3Z-g%7v$~?5P
z2Da=Kl(8y64?h1A)={weF6(v4&#Wn{^k@+eH^}#3>)=A7`j{dk#TXrL76;3SjO$*h
zKU}n!=-u4!mDoHUSslH8R!2(4zq=(~@ShHM35TLUoP6w371>-~oZL@z_C7qH4&-(>
zEM@Co-K{S9xxHM(xm2|8#l8GIkLQMeZ7ZVB-M+@<H4^2H$AQx_fSa4oH988)@xI#e
zW6mw2(4!tl&}hX00Yx!W|8Uhntd+lkUx@u^L9x{Y<8~u_a4@L1i5^VJsgC2($bi(X
zC9L+}tS}QTKtO<lmy0hl2)(BLc)B~F<?imS>sAUH)*7J(9~Cuud|XKMqx(jdJOzAc
z?9~ot*yU6i#$a8;uSfzoIi)o0l(OgT3BKS>tMll<pwXNk99tx8;6q)OC+QW?vZDO>
z3VPC^{3~+zs%0=}ua)>aX`sr0iYTHd6LJrGZP~PDJU;ulqN3c-4`}c_Ux*!sL`a5u
z__;dojc9^LRvI(v7|wiaBfw)&^LX6bxYRtW107Ohzas8T8DN@*teojhIQ=8T7$WQn
z{DhhZlys6IP}#R_AH#%RW77WR7baz`9QX*+bkN6--5LpzW+^F>CW@8x4|!W<P!-9L
zxL&I$1-w|7LxXEi6l?gpyyJ0sipqgB`~C|3I+Zg+-=cm@;zR+j`>~}oz;BBnN2Wit
zFTQAU#qzforG`>CDKz6<krq==IBX<u_p=qTYuezdq8W7g($BsFMMhj-lgAS24Oz*j
z7T1`Vq@*N!8hU#IQ2)CMDF}`%BzZ;xI1*Zb#tX1dktElfP7g0Q!LeWF*~@e8wLS}A
zpOa6A29MtoU^I49msW(VpV6xSo5lz`!x%i@0`A5L4BxJ9_vSjuvNMWGm}R7Te}v?E
zK0OCu!lY`uIY}>tnHso?6v)*rtV)hf=NtMhjA}#&7s88pA>qTrjFmch6Q=PmW?OPF
zd;QVPGnB9rR%*Io6_q@PiI3~@!KNkh!G#6Vi44uDMiIp%KvB$!CwrVT3YwfsUu!Lq
zUkVH~zE9!%BVjqn=Bgkh+`cjp8eT}ipcx?tdHk!SZ;&8IPbiKGe<%2nSX!`E5dPgJ
zahSUOUS#!tbZaz-Yt>E9eRPtq>+>S=%aX8f*?X5$Zu)W*q=D2{G-F>xjUThb7R?o(
z=Z=;un{@b5qUQJ}s%gOq80_)+R&qVT{QAHVKN06;s8>~1Q=<m3z~6bu-sq@m2uMQ0
zkyeh6e`!_KCwLWV1xx1mY7u5}q<&DU9l+*d+(G@mt2;moju`nPX(CUgq*z>1GGH&q
zSfB*i^-<-7xgP%nGD;7Ryw>j~B8%el-dDM}AO1F+0&l6UT#t!c#oM~o9{WaMV`+z<
z9hh@|^l~Ho)ZW{z^~ZHq;CY=aUQv-vD<WRwkwHq_JN`rH7GHlyb@&%01ugg&jIt^7
z7~5CpR0>g0#+o;wzL=0~H6U1~2%2dH2GWFb5xDQk;)C4I3K_w_T5v%Hq9qojf>aRD
z8KixNa{1YbzqWwoojgQVk*OL(!7yEn8z$;Wzn5i`h_zN_;&YwSvi#B#*VZny{L;Mj
z?W!f<v+I-P2lI>5H>0nK8!_w?BL@o^QQAlr!RS#T8_|MqY;Glnk8u)P0096AUYC*>
zY`rQz>nwgJJNl4TqCUU??(bi_ZgEpp-PqXpy$xltRZ%)h&HD)Jl@S0q#C(#CgbqQO
zVqcwCq*06g2|-1Us7XL2^{sZb!ME(Ux|ry=&A`v3*gtT++iDHj&EXdZg)uA;(<8~r
zVaWe({C(Mb=*1a+OaNrPm8<P(W`7%`F6RjIasF*?m4Ctwj)vOlmxp_W;V%?1l2jLI
z4i#Kr4y953@s&jSvlywNVmBbk=<6`Jh(nq{Ne8e@DSbQwB+U!_fh-7%@+D2g$GuMm
zbzi-+Dd1%_dibD3rvpVcCV~P!{ESRpbVd;v50EHk4aF8|>1}7w*w#ShCWS{BNhlnT
zOFTbhd;MMbkloghcQxAWH}6-54-!yBjrf1Gy=7R`-PS)21}%y-j5LZP-HkK@NDM<s
zNh{?L(jgtv(kUe=4N{5@E#07mBaL)7zm4ZU&$-We&hzSj{ki7F?cCq_?!DJud#$xs
zeM(?xMYXaz;@0s&^|-j@DN*hLUpuasamM#*UCN%mQMn}u71cNo^N5G)W51IXTyCr2
z)n;b6VTG)J_WSulN(xA{c)O2QBXFczaFJrw!*>dh{x;?$$SKA~TC}C|0Os6a`L4Qp
zh+p|Gq3+(JV|Pa8IMBtEpj6i%z+b(xM*Z;bdBZ4~(DqI73qVnQL#D2!^Lu=*dU#0C
z<Z<4iz%aIBQK?Y10mYz)yP8H$dz`M?uB6+hSba*4fj-pehP48e!2|>gyOf~aiJ+fn
zjp+c6Xvic@&t>u=@eR{=%<*e?BU{7!q7gyCjCLR8QbNw=9P0F04tW?L$}6=+ii2K3
zPB^1X5NT<a3?|%;BF&@Y{>g<Wd$0J+KCmWn(iN%&m#TssHZk?u%bFw7DrGn(Mr*s?
zjOV=Ntmx%d!^x}5?dQdtVW$Fz)fXG#LeI|HQdMw`Jzk;cr@ma>cDOdI`|6@nQje1`
zP0>R`Qp0Wr^V%Ny*rWuBw!I%cK2oJ-wImOnyw+`Hkk+N#5kAl(XT3W9W5NEAMG7wp
zOp28ZnWu3Ni~3Vi28I}P?VAp$R3r>sf={-Np*dny8}#ezUA4;Uez6Epgk4-_Vj{(d
zk|e<{f<ZANh4Fk&i1#A^WUH!=a0DY3VpS-ai10Vcf(u2zSH-j84)WKUF2y5-<UHHW
zBp778OQI1vVfk$!=8I5-T1^%M2Y?WX=wqRh5O`z^PH<^%_7lPCs;)MJ7t9e)Kb*C7
z&;iQb+}}STA`AmHYAoseCQUchV(}&&^f-cZI2B5*v_7CDE_^f30=5l!(G)U)rag}X
z{2hD*NYEkW9%La#S1z-C7w_MGIjhCmUfUTTZsuQZm7TgDG#%vA`mBdDGY2b%;*Btq
zVm#7%))9FF9i6Z6ke>D7UH+V+bo{tyB^6{GWVqN}!wc*21UcpH#MJb-*p%p>qYCjN
z87dQS)}oMSd3Q-kAJg68fKt)YYEzGlkfqFrPcrR2)%c<lR+{i&HHGi!=XG%F*Y7Ak
z^cnE)*XGY-f_&>{05vDu+uNWcpSKu$JyQkd=-WVw>aOBA8^aPA)}XO5l4~|MlJ?Vj
zaO7bcFdGFyedGJ4c#FyfFetFKU$i_6kftvz|G961<dmd~!%OoH@J~@(s8tWslcS8i
zJKUV~$t45_+lz8IJfKb>g2Zg+4!&k*5+&xFn?02iSy@Q?3h;}S9#!U^Lh}Kl9Fj?O
zU7lc5vl|B#LI*TnykljSACumf6?>NsiN6m)mOBcvfV^>_jtAS_udJ|_NrsNob3Z&p
zW_(T(#|>N&Po7fQMM_D-iEulb+@2Z;+z_4|4=ySq5&5{S)0A}R&NPQ9H?L2o|H>?7
zY~T*FWI?dNyB|jg$51VGvdl0YkOG#LhM>kLfI*}C!mrud%3Zs;L;qU2No0}2bIbP*
zI;3sj$D@x4ij^W|1T{6x^vr0r6l8e#^T@u2cTA+kq?zZe;-{0XSR*&36J&@6N~7Hm
z4cu(#%mKZ4iT95=Nr_On=jR2Lw_enZzkSw_>hg-{6YeYAVAty^oewObD0QtMF^{{%
z6MNNZq$^9v*Kzm@-PUc%>JHX}#fIfG+J1HvN-y8tLvv1SP*iFrsWdc=hPkFaEDXTM
ze-$g29FrXr^EzGMrogS3I2Ml*F&u{|5U)WI6^@qk;K(<j_)C%mWmMkYM~5^hed0PW
zU^CPL$G!gi_|)$BS6A;wwdOh`C@2@T^b0pMG$r<1tISX=-gVqmZ*I7isHsY*A61zb
zJPPf7L_b!d?xA71tXr$>j~qdY`8#L!S`=$Vc~ZqA?pDy=jD5l0(od2r^s~}@qDHCU
z`;T(pf{78`>KD&0JP&p9{g5n5%LML&Mc_zdh;F^M>B>tx{=!ZD<XV)d)IKZhW;=6y
z3f10H{}R4tw)(uSI|pvNw4TzniR#d0oD$p2wgki%g=}_?j>X>4%LGus``(kK7S3ND
z!%;?>t3G+G8jfT+-NhMr1<A+7u1JH0Aqfa8#RNz}A=xsE{}|zI4{}z{jKllJmlv4Y
zJoZ9`nuo-Z^7GUAFu`X{co-PJ8&WExTP5e;+(kc&r(3xjiEF9*aKfcVLD(`mmz(?@
zePLXJ$*j2N9+6gwnyMV}=GHtec8*u})n+h3R$=MFTdrZ*;KzKAvG{GdRe)2;Z}~7!
zM)qAz%NR*iN5|+e<wzQi0!otf&NbK_!U~QI22+yPqhTmim><A8{<)AC%6n6KJwN&R
za~3KaZG8>JagLU}VQ4g4fE!Hqka&l{d~Wpy#tnZ^ZY&RpK^foWWCt(`6C1*0)ciR8
z@y8iKkEgUripL_6Apz3b@cLEg*{U`-Dh3gc_gb@fYrEr@-p-^-3~y0*Xedtt6?))6
zeEb8PXw%(5GbRzSXPp4kTeuMv<OdRMWATV>ZRZxA$XEqG72vZ-e2Tj&1gW5Cve&u;
zyVsCzN7sEEkArRfd<<DWM}&-vZ+*o80bG@gKM;+$>H_Gx!c0OEt@6%6K?nl7OW3Xm
zfVQ2SqQ9Pf(2H=Mr27ft@OgCi8>!yLjAgk8n2iNknA@n&UT7)tbl7`e=BChQe}AWZ
zH?mzWJ5eC|2KtWtel34dm~(H+H!td-Ac}k6;SV6MUiE#w4&=i7jSPsH62cHZcpfli
zgQ$O+u>Q85-TOc{8S$1ZDHd<|(eM%giwTaP4s^`x3Z;CvEBDckS@5mhP=O>Yub$H9
z%oK$sJ_#jdqPqEj?g7M%P@Wq=EX2X6aaK-{A(R*mgs+-%O+t{IZv%pY{cqj`S|NqI
zx+bug8VB{Y`eL5`en&K6A04CA4&87k4iz4Dcq)Sw0O(Vru|;K3qqjN?WbF^&ZncU~
zM^Fc{B5_#X%9oX)^Wdz}-@l+lISk3UQ<lj3HABoS!1UA^;+!;?{tdld_shGY$zKwr
zGZ<IB$ju@S@K5u<M|f*}c4%ho?W|eV<gab38Up8<Hf)&Symuc1!4KUM{f89Ky#a%k
zX?=FtA9Zs%JDcwdX7(Sv0m$X%<`xngz-S+l_gRmqatKEqJ{TklpgcU(^}6|&Jr^lG
zHyjITiwN2qBNvuTjb`9$-$#GEX;@`ph?#d%WS}97(ujcVF+vFDILb{mGEYcU#3?ay
z6lJ@=r*#bF?aVWM0J%K@{>R#&7%(|pQs4z5m(_h?)Gs$%Rj^Z>5SQS#HoL($I?Nl-
z;|o5CJsp2s5J2@!<w*qc70-tF0tM5!7Z&#5&oxbsX5-gh2I?!7o}&gS0gQLl?$ujb
z$A8m*9h95>+8v@_=bNKGzrZXjNTCJN^h}miVv);cnuRSZtkq9KU4%{jEE2mE=~##u
zKlpZb#^!wZz|l-k6SX->jIdZ+@mD`rB^Gb_UPghQsEsJfC@bp-B+ua5qS$D9dxQ|~
z`Lf#iCdmAr-EGL7VRfFzf@aT@pd&R<(6D^zeNSq~a`e)nsTH)4Gy^MkRhTT8l&1en
z#-J(VOJ?#JjQH&qI8zOf>*T)`p?GTREBlUuJ3C+OO@>0X2(gP5nW@ZgaL!jz56iZ=
z4i(t(Myl5F)1c{yLDqHqJ;DzvRiexBy7Jgm6hwqIN}5iWN~;^x5SdR;a%K`So#(@s
z%t!EnbRnHux$AW6(z?REGY;WE2m)2tWIXb2guUS+x>H$AnPIT{Ohyn`NU8h}apWy6
z@eB|$6YJ5$gmATqCSARZ2PfWKcG{m~gF`p$eC*dpNXBPqU<sByS{B9Hb2(AD`CTl;
zCg^J-GM>|~W(|&2g)1}3cnXS((_$whx}Kt=`7u^}=?j%rhV~GKCbB8WEj9i}WLrDk
zf|qB?_lj!6B<Un*V^q_P_I^(1+ty#7J<)QQm5~t?KYGp0J!~wY*rD1>pez*k@HQxF
zzyP&8{*WRe$hW;4?NYK>^;r+zA_u17A-`n-SOHOFgYeNg$eH|fvUdYbCr$eHt*?%<
zL|$m=%j7gCZSo-^VCl;`T{2pm@No%-thx>b_}7)Pl9{K9V1(k`07ozNj2h=lf~Bzv
zEc@c6B2)<57AA`sFBSjG1bN*rrDze|8$dx750FEyW0QzO9zH-~yu$%D)<hGfJWb)Y
zy~%S^*2rT;n)|=Dhwy$Z7qC(6-KR!B<Md~Yj4qy@s9+2|D0*y6@4OlG-MIWaU#>g=
z$1ZAFAJRzPmbC!q1Qy8G)kz#Z5&P}zm;kjs<yg0cQod&eCyE4BPm9^brg{?2_r{`m
z(T3FugX>WM)Pd>HHvut1u&l#Lf`0uAl|W1=3(eye^EE1Hom;zE+GHL|l&RK`J?a|3
z$Z@-0BC3OS{20$ZR8=(_<d>hw>a~6g2Z7ju;+hc<f0%mJXrOSkydn3hMC2Zde&W44
zjfxyja1tyh=rINtxh?ta&dMZZ81vFI1l-qZ=D9tW%_r<5ni5#D0O!d^%z~16EG*`+
zov+wfdivbPAC(dtP-bsCHl}d}Y(^PJLXicNJi#qU;YZMcS=yJj>3zWdrGwD=q#+Eu
z$D1SZIS}S{h)!Y!2QHcHtSqSVLlO*bS;4y%FMz3$Vm2R$E5{tKhLe+11YvU<AR@fA
z8?@WkVsvRyU$Mw_csU)vCqOiu5|(SiG$o}71(gcC5W-QMU_T_pi+9E=(T4?uDwJCk
zn^nEP_N!d*(G@OB0Swghx>2il7D9ul#0;5RZD+U+09L&SD%_J}Y)vmGII%6ea8X<w
z7!Xy;9v3tmhQt|^v+_hs)1Z+IV|jPc0~v|eFk3&__}`0}lDKR_zxzdC+(#72#1#Mi
z>H8Pv>O^VR4Vn}(L1Itx!8Pl#BSKFWqDYq+`l1K1e#7f)6CyWX&lT+TDBK`><9w-%
zT8JWw(EeRAHt2i2drbYIiYH^Ox@|c{E{^%-Y-IeSh|MiOd{C_jy1O1U_N3L0L^NaY
z@{TCi;6KpURf;ByB8s7?r;H7oVf~tbcrHD*|C`5jF9X9iIxA)i)QB0n-Bi*R(daY0
zYjGNiOF}l(mj0cH(0p8Lk|3D?t@is|nQRf_OaXAdXt^0}d&D~C*6xzB)c6=Y3BZjZ
zV<IgZmCE*dZE`n4=GjMTcu9zF<%e6hZmoUQxM>pL?{)a_P*eLs$)m;wH4=;V>zb0?
znt_2fu)I83XFJ__^^bbb>E4-^^Y8|oZualvij7Bc%qd90)uWIk1o;%`XZmdNw9~vt
zzMOY9XQSM9V=JD}p-XlSHV*t><bTQ>w2Q%rsX>Lb@7fBi#MFGa6%j#7N<mYLb?ul+
z7vBU!gLY?Ihl5{w?&P<ow5%0BW+Al&tKk&yekf-i?#$6OZep|otJEz;XmoG94J^>v
z+<jZtbCKr-Z2?(QsgY=jR+=jVz_lGUDAS>DZ5b=;c$h!1X?=&#jga7B;j%j;1qSRr
z8yTz47wgBlze#S#bAjk_XEFOoXZ}H8IwGLujt##zM7=d5C|F%f%i4@hY93krF()gl
zM7cs?pq7A0z|PjZCQ*y9mEn*E63|j=n^kn3J})X30fm-(e`57D$W!vKogl&a0@o}Y
z(?_keCa?+abjBT>NVcXiLO`BDm^=?TTJT_kvE7iICSPose9TquNBfsJy4XWv>Sttm
zJiE7k{BT1J&i2bPs7F*h?zM^Y&51-|(_&zBRNR8VZ}OmG5@?@4c_Le_T~|^mBiP>Y
zH4cGD$TJCe{_HWcwPqi?Dc9OeIwC<CZ=lxsV!zD(>&?QAlk7KJJv<m&KoTu8oOs<r
z(M$NOC_t^~-pXO7*<*q3=~U=ti+rL2a69IuxbEY|qsjJBFKJc-Kl(X{Jq87-8Hx+9
zYCo!pehg$CHEy!h5AvkT2n`LW7jZm3Ay;&7+pr1sorip&dJrAV9Vn1zf$fW7^NP>P
zIZm9`&;#&S(C~W;9LGo`6MbR99_AycpND>dhcf?*I7z{s-9u8^$itbI)*Bh#RF;{x
zPA;rEN}|Ua-khIAV;?}2jTA}@*(UTmK|(=$JF&_a4OQPcom;qf)vWVTS6xUK5;33c
zF6Qo$7z_YgHK}$iH|n|fw#@8}$M=78bH9K;HX*XohZ9X!7Nu&SbZIeW+^{=nGY`~P
zUwa8L|8ywSI$xQaFx{T<NJ^2C$F5>~Er<2U9zUP8u)tR`zF(BMtmzuxoBj+M;kWJU
z?bx(xUNJ0yQe67|^PVP}+E~h~GFMfA8uLgemz-MX=PU&rDMZ&7K6?Sz7S3lYv+y0z
zZC~`FPEj#-=VFODUg3s0M8j*u7ak?8PQ>W<R5$cFC@X{u5Bg%P?RwdqLf+Mq{T|Qx
zJW~Si;$nIL0_^hkTvJ~S#Us*Qtmk8uC9+9hd^oW_+K$?eFZ=mEb?DQaSO(tdMC*+=
zND2++z2V;>ews9FY|pKVN?H#;TfVRx!;@tqLhDpvrSfBb8CLCnnqz@cotvEyV+rx+
zdt|Cti)j`hJ^Shl&RuI`U~Hk>+)!Dgww@QBU48qLmiF6iFKF`!d70v!jB8BRpQbtB
z8EbP$%@rkNN(AM`@}r+QEfo~`wfNTZ$9O*KBY4qy8$;q!XycPMABz4xh(no~WMIGb
z3fp;lpAjNX+t5fV1TiouC_8lbIuXAn<MwVR=vS6Hara!B<5mtWfP>Q6e8`w~)=-&`
ztt8fE9=@YChRCT)vY)KeFDQbaFg=i936IKoR~Ym}P|(iUezWx6HnD0?c&N$&=Ms+m
z-Po`uk7@#b>Joi!8Ai<a>JK`Mo2~gun`yKSlYO&g%0AXl%?|k)e&^~>Y6(tC$W8gx
z5sqZK_IW9ty!q7bgN&dsDogP@D_P0Yz$d9&eIigs`7q#j+Y5|By9MjjdfmWibfJF9
zZAk(GjOB_qaZz6t7KkdGKMZ_kq5)S_*biGu46XtzCFsg8^5!qI7eUl7%|0h*=Vp&Z
zvw7*iBBgP~2whP$MtzM%fX)OOs_N{{xFJK~PXk)k$)LK`RlMB58~Q<3HA)J;vI=te
z&(AS|16oq>E?WM0#)%PPJ<dP^$Nlk;9CG&Ym9vR45o2u}2`|e&vFfs8xXda(wW%qR
zO>1RqKXo>3cY>5-!AMxlmda;=A~qWF&g7CTqMa30$#vdV{5TBowq<(!#%GMU1pd09
zS+`)tgQa3E0$wQ+;vzyEw+fQwqu0vOKYt8cuVorqTfIUaEedS6(EbAM%#d21$IP$H
zUn7-Sb=J@38N(w{R-`^rDK1h&3M6XWOr#(k^Rw4#M6jwj<;%~w7-h|o#44RKkRTcF
z&wlz>)Ip&K21+T?m54iQb5ZTOmH@rJb)oqe*)!ni12w^XGaUEXUfa(6GZ#R&Xd8F?
zy_WS#Uz8)I5>;TdbS@nPG;=JYrAirzLUwY<PR?pD>dhP;Nd5RVYt?rBc0t_GBXUTS
z&hTiUoL>I7==__<dCUfA1TpqkLfWTYCRVp1T|pkPYm%<Sm#Ti6y2CPAJ3Spc=IFOc
zjsc9gu$zw^d7kQZk^lH{n&!0X7}lxtkSc0t=ldC2WV5y3r_|FRUOf6vublb86iJ8$
zX637DXAk#smNYW7PW@e<OA))~>CPBx>}lDT&mJ4mjmwq1FjoITD0;%uC#LPm^#QIY
z#LGk)FIcA_h%hKnQoKW8;!bs2EP%sDh*%!x$JyDLog0e<k3%rTR9AYJXn_uR-p5K8
z$-l_dU{s^8T}6txu-pX5-aB>XgKqIiAo|#~<LRv}8Vv+@+1N~M(?wSbs@`gPXSe1{
ze5gcaCn%HQ0}!PeB8#v>!g=y~r$B4Ob`idIgGCPLA>+ralyFZ|$FXByQ&lwPZ0R<1
z5rPJq9dlr+r#$`ni<O7^ccLCFS@-!nD``)JzmKY{49sg0aaIe;2JO1GO&CQ6qL3IM
zQ8%WYzMF8+PW3h^DyG7iU9<eReXMV{>G{tdJ^Q(c9bv|ro8~(n<0Ww8^v~ZEon9IL
z*6M{Xzxut^IGfcdGHGMtPv@Cg`KT~AjHDQ$aD8yz9yGg~yEQ{(#D#t97PH47=s7Xy
zp!Px3?>3FHYJRFPn{OXFDI=(bi<g9#*CI@X>hvH_Sr8$)4{MyiTcRI|Rmye~MQ`*;
z0d(g~w0c}~pDZOnnmafD+&04vE752J`R;fp=RswaLnqzUl2za@?AkBR=RwKa=_4|N
z+xeB=c_%m~HmofHE!^Ba6ix(5$SttzjSVBX9fR_7pAJEPr-Kk1Ic+$M9c5o0?H$mv
zfQocm`pMxF^Wd8q=ql+TNT5}y!wSkifEHMs^=>1jpzRPlH2Zajk`8NNO%YniAPWmb
zYhpJ}*AWXptNAtFe6exyGT?sw#zw&7Xt=VnlYr$t5;{HEf+x=6pgC3Va<lY2l4;dG
zI9NIeCs@a0c8DR&kB{#zx^u1Et(Wg;?Blb0BT59ul6^&aIP!`C(k;{aRPUC>UVdjv
zX^t$!%5%JVrHdaFaz^m!J$?T>6&qWF60$56A1*uXc4vKDrn>qaK*1#hxBd>AHxbl%
zS4$)!WQ@g=R3M>r#Kw2neTi{?zH*n)_b$s5I^gpP*Dl1U`-P|PoNRH&t_|;OSzZ_K
z<!cA)zH=aw+l>f3BIY&32_>37tuA+dx*Q*ke^=-}9htOMDPp5=hncjdtMok<NQN3M
zetT}wTLz(uwHT0zMHm~`({tz!x?jFkZEXu*mdwE9<VoOdIvk~M?i`Ln7Tla=Mycnp
zB0^>EgoZryu;7SKlXi$zG-dq$y^~z;Y2D(@>(>jBH=ll|ruH9Col>cOoc~~JqIs4M
zeTNYO@}e!7O%61a-smL8q~tnm;bO-Kq<9BN&yY*Yv|s4!4zakow;0ba@kp3`oI?lm
zY_AuZj4dE@9~S0Do1V0PPhF3?c;Pe~UCujQ{V-5V(Fx>YuoOO@N0u06UK<^hF*LOI
zM$}<6+{l-ahUG-_k;m;fkuk>;%zqkOl`V485&uO6fR+<~f}l_AqF&D4#rN7LhD~O*
z`Tas*YrweQR@ytn_U*(`F>9|>c^^rFaGw>YeozH6Yv8wDzMS0(9OSTI7~>|Dn`W}b
z;u&KVd|u_mvoH#`nMX#)DiSj0Gm+M$NVCCYb66Ad40F|FMP6n(p0-H{o1fnf!0Bkd
zOqcTF<|uqVCGP-}WeYYnVHgg`sh^ywHyTJ17I8yy@$;t#<i;M6+>Sv~JlYz+v3{Pf
zDOq~6o{+%+t*QAmKss+qAr*=9h#|Xc#ADNVT6jBoGDCbRIxY9ww~I$){=MyGQRf%A
z<7K*Fx4t3^qB$ydPxlvt<|11}HZi_<Y<4yFn%5~ZThlQ>zJH($e+^O?C_s{rECGx;
zSW;flA{tpB)Mu^<AG|fhp%?TKVmTiyH|=p5@$3#Z;WEY3iZ4n4eYt43TyU*`==Z7>
z0+AcdRTv7;w5<W;^pXmz9@dwYJ*p>Ewl6z=yqs$#ZW|8^3(yN+xRJoL{N1}2zd?Ln
zI3x=v*xj$;sVkL3FKGB-EZZq9be0mU*hen}yUE?GMB{OoU7u`3YbgWo0_f%{rem+V
z)UzvyqC~STUwO^AxuS1k0KfpZ#*#^f!si4yh4hU?kutgx((o?h`MBKd#dGH7vslgx
z?)!zywHQx7LqH~2#-b5gZx$s>j!w|4_PI$M5%K)RAav>uA32|ZMB_c53rD=!5?n4E
zyl1->5c?aiV`9qb0a{J=XBvs~!Zak^S0V1lg3mXbYY~cwXsP%x_s>aS6>06h;Ri{7
z&79<_CB^TtP@>4RZuB$nV05@TjrI61?PicE_`(ih80mKnh~GImp|-yAZS66-+l4Am
z3e6>%FadU3JC6Ld_qnm$BRJSZQ;PYDfir2O*y`^ph(K?{$`Trm&mkZm1pq_V)ZiT{
z+ef@es?0L^ndS&TvUt*&-yT3*F-yx8$BTCbi!tNosV~#|z2<1E1w>kI&_^?8Gv7Rx
zWna(;EkyE;?6wgsAUg%DY?&2e9~RC-ViORNg4Y~QYa^)br|gp>kY-89I}vFyGVLr;
zHP&35&mJp&{)Lz3LX*H6^CRW`{X&Q*E#WbqOiiZM_fA1H9W$xs<#8H@`XV`A@BCM*
zy3{wFb-CHU$Bb+I=0Fo|HQi(tOaE+!weRr<4H8uaW%AnO7zDRM1(JFfx!rFGLh9xS
zrDC3bBm@}Lx2T8gY(!sXAZ-}TYR+N4AN&V`lo&?z+N=`W6iwrnSx@^!WN!TN#=(@J
zZEZ<uaVbRrBwhB+5wonG-q0s+0INfw1YKBQst3->62l1VQm%np7sHwDw3iBXf;|i{
zSzvk=3Nrf+4iZ$s;P5V2>FsbooeJ~!Q3gevQi8x_;|6%&<u=I;d1c0|Mv={z2EoYO
z%z`)wD_Nbs=5mxb-SR_>GjCH^4iu#1PI+*QmB2h&pFfWlCf`wrZ?rDI!9ViydPL=f
znu2pZn^9!C2<9Cjc24Ga$W^rGHBW!*8;)yVu9C{1Rp@-cxiK&rKO#y9t{jUzE6*Ph
zCX1C5Z7Jb<O|0uQCb+T%LgOj7L5nfOa9QYrBzEcW$DFvIP9|1iaR}~7yGOeUgGE9j
zVBU|Fpuj)hA_Qq5{}M(@8KPXG<J$)y5QqJ)?d?J0&YXnAFOO9tL}}dC*(%0dJVjFo
z(JXATFCLbD;$OdBTt@!Q&Mnt?we;KFv_3PNp8*D3n~4UEadA#<cxzLk&q*4ciPG_}
zTv*Z7xz~Oc6>oJ&T>D}gh1SUYWrQCHf3f_*xFVBlARy;Kd~J83q&7T@Jc8#gnuR6$
zB+MMR-9^sL3E0W6VFHRD<H&CiyjBKXprAcXFc|YXj=W%rs$gj&VWNI^RRlYs2*LjT
zQ_S!ocy)EI-f227d(&1)DK<#62A+m(c-A23S!Xkq6l&2Uv$TQOCwU&1Fz|vQS}%_0
zot&Wesh;QQkIE(BVg)I6pp;?CG4xsJUu=$n(99|&{5TV4nekOqd%NtO8xwo8=*mkM
z!OBr@__Zg3geI(-+HlNKc6+Q3y~m6(ZuogO)t~l)is<ktkRwcq<|Md}rqik=m@C_J
z5DcGZTv~d=&R=Lc$?dkve_a&#f&^08NXS0B&ENVU;jUwovUVsx)HByLp+jG`csl<!
z?%WyJ9jv|yA{B22^r!Az@9gH7i_<{1Q!a0kf9l2r5$bmhR01<0Wj^T>{#Ubx#~?63
zlNl=N?bO*2{On*F#gAETwQ8Dy7v2CGi2LNuyGMlsYStJrwjqSVBF?$#{#dW6IN;oh
zTJj(Y$P_;ni8K?Kf6J5<;D1oUYjtL>4Cewy3l1LsD}8e%rS18**kUNmHQ~kL9Ienr
zW#v{!=I-Rp{2u8sZ*JX?3wE=u>{>1KK$|p=Q8*{>6?o)632^i5_G_fhxIOSb6cHX5
zJ7kVm=EF|oQ;-GSYifQK8})W#ZYfGgkP;D1_dH(nXl+6Ms11Lt4ez?gY8Vbcuk})W
z<)ETCs2I%a7D7{Y;7a?0h|sObn<>`!m8UQZt66|v2#&miCUoR!DkVCJi2KQ0iZv32
z>p;HdA0R_chsmGHv&(2`^>bJ5_~d)Ki2eC~;ksPU6oajWr+DZ<$bhv{uOngVt$X)a
zpzyMcsvq_j$9vsCx;la>cCpM4!?m(+2P`Sa-iN$Eh2+LE{=}8XvgQlDkxnq)3ep~g
zAn0hui0Y=EFA4#S^-$7;T&b<M;W|xbH%3PV&-ZNNeXL(2e@28Yc=Hfes4>%wms1N6
z4q3>tp|a}hh+0acBqBMyxmHZUp-+<zi&b2t`<V7kk&%=kGSlpdO^SjfJozE03lZ}u
zx7KcN$h%7vxp!K{^3%Cx^8s_be{o`-e@qO4UMv$nqY#9SiS&E5b^sTQ*@XeZbCUr@
zq#=@tP0N7pjLuB~2^AC78`CZeSbcIar?KCF&+8#*`Q$H(5o2%4g6Z7IyR$y?Bu>Sg
zRE~+X{-LE{NlZ<;0qfHM(E8d-giwT)M`YgPd1xnM`wd1Jxv6PxE_X3b$Jag`gp^?-
zlV1q1X~M7&MTih7g?+8hjyD4qe7wE;w#q=Feb6n8t;JNq<9VlpR*vn?d6(>NaF^<*
z{j8rjcLyi8l8g^5z9k&A^um#^W%#wf4{xbII2dW5dymBkvh5Q&10x%grPPV!<bk`M
zk`W^F=-Es)v+17@^25ULh+gE><J%^6Y;!cvyRr*dj-jM#%$f`c9Bj72BoH#ZNgX;L
zo(qRx%O@5SbN$VhhLjTasC-KFMm-i!BWl>uJX3=9-aQb$#V3F7z;w0sN%k<=4g2H|
zH&>h7mX|SJ>kw%D{<`PF%TU;f+tI^5C~@iLx$vmkLpO|$DPHC@gy8G%wh0PgQ8e98
zgv~GE+XoH~QnJKEOFw}R7QubpQ_Wk<EAFE^)nPZkw2{vRQFDr=zkfnxbRj1AoG!aC
zziNl_a<jRt#Gj8q1KuUnQ<uwd=--vVm{@RwX6gnZ#al!OP6t6>((Hw^@cAPJGf*8F
z_sq#@3<i2l$=fZ0dMk050#rzgEVT->;d6xq2G%<Xi|<k@tIkz)3{Bm7*)}+RYv;T1
zH`kP(@*B6Pk$Z!MkUx^nLLUyg3O6)x56JmqR&YC?9`xuaQ^-{UE2TdPM)y0w)FvFL
zkCpTc(vSdCEg8X~L<F4kLb#g#0c38jvp@eSw4biY5P;X5q_8&!t!=#5ijtXON!I%I
zSe;1xf+9;Ttflr!IQ?UwN)<up4bjd3MqQXW2*xA_c3eL`&oD-zw>g<er{B-<+ydC)
z*tpp1rY3+l1FuRESy+3RQ8Dzq{Mj|6Vq&1}cR~32@_@?EpWELaxz5Lse8_rorE2hF
z!zRQ&m#G%d!dIU^|59r3cu|@6R?E9w2Ey~nqTyuR`31=h?r1b`7wGw&lbaj6%kqf$
zmm4UG=hn|N(Saw=0ZtLxK2_Iv%P5|cEOT{B(ZLTtai>p=4CX?`^u4wPv8rIoM8ukN
zp)<SQ6P2Bv@~b8qx2|6otlQPVi~UAVjD8ohZDRoRmK#DFU;GrjoRzHW01*Ic!>6w$
z1slW6Bjp9>DQM-ry=(Pgy%-?LYOLMLyxf(`c{3ry*~#qvi?_m8nv-)bHVkJ5#jjVl
z2lqtd1J6#?!A%>A7{rv_!*PI}8qExuEVu<%fTpBBa0HS@3cFaJ(>VVU^mo$zrU3N&
zI+<=Po&nB_`w`#Jhb?Heim=RHQ0~-z-sgcIw!GHEkG>K(S7#2RsAyn3`15BQIa(EA
z+y|NAZ#eqClb0LIz#noXexIaK6oN$O*~>8KF2|bIJjH~;lW*aEWI+F{^B=3wA$2P7
z!{GuBB|aC&;TODaN9Rcbn|B(15QFyb+T_KF4+gWuPl*WsV(M!EPR(;KrW9M*y?k=3
zm3WdM>(4lk-r{s*#kF82&wU?IBRkQmn@*eM0N5E)g7Y8VID(L!><6D8d%V-(eLd)B
zQi!}W?}>FGB7;6befILDlP1xwd(Dj^Kdybe;xh+Hxb7z^0K<G8yP#xo@-}0vqFa3&
zzlcXe6s45yhsT0g*t(B0c-7&kh;W72Z#Hz+o}4eqeB%($@6!8)W$i}3bpSG>6B>Ut
zNY^pEhm4o65E&YTRb7MmWsi4fr}ExH>UK>vJ8(yI-h=^|Uz;BmoR)40UjeJ#jih|Z
zkfJ*>iW_)HSsJ>4-hbNBK?=@Nx4Nfor;YCN1K?oaklY?&f<$Y!M@HoozL29{H3N|r
zXt6xkOQ#?FCHl$hEq`<Yy8^VP=DBqXuuJOZEDj!XgQbKubx|_bc<|?xs)Qi-_Fo9R
zP{d_sp6cwZn;N5AVkV{GG0a9WN*K{c4Mbe0jK$kdW-_Mok(3hzi?JbgPwiI(I6{5G
zZ}!`V^8BQ~0-WWX$ZrS6f9@pYOr#hzw=Z@j98F02Si=>dTGkuXT6s7z$h9Adfi#~C
zIojJvt_TL<r^do|lsW?~qK&LM8D->{N#h?SxlWeAF@HMaCyxYkqMtukTN!Z$^*QT&
z7!%bGO`m`s6LZ!t40?o6G5El+&CftFC-&?~nqdN616=>~$wDDQkVY5N?O)29pK@Ed
z`R~J^RZJ*#b>DHL4!@f`UJw%WyGOYbv3LNf<!&OQVWeD@S_$tDHQk3JR!8SLO@h>N
z!9oNdsoUC?m%n6)VP}1a|BD+i9gI9xiNH8a`bk(=1bf<Pnk=j&Ypfm~RngHK`wlrd
zcrbH*eYw%mx>GxS=SgMSU`*84QaJ2>u>Y4VCi3^;DO6v7dYD|R&?)UW<B95KZ&}Ho
z%YH+M8U?-nFgPoIa|rkxv@(vTZRc%Xz8DU<;}~vpw4FTb>b~_Zc+61-E`$Jw$#)=>
zwL9=D54JcL^KOgMi-$qM%rNFh@tn!>ulN$!L^^{$+H4t!$#Qmw0Xw$~LMR1(m)DYe
zRd%E83L&6Li47y`lb{8Fc6N1WI!{gZlDwK2FBI4IYQ)ma-r&y5i+<Izr-RNN^-Gs$
zIh+_s-_mq8Y3ybIKCA!@h7u@3n@Br+uS45@S5eB!f~rdqvE!m*w(Hkpb7SLK6c;O;
z*-9Jj-rv54@zX()#w$mu5=1#uh70iG>Cy&zaj-EmRjrIIsx^DF0{pJ&dmn#2hHAhi
zoeons`KFMwwu$NXi+KWKC*zj~7wGSe5}~U;?0TI<um!b1NmyW6c&}E~6lL3zc<}Gi
zZmDb7H-4zB><xXM5(*twX`!?Pynw+O&HJqsZhV)J6e`KWX%TS!@skJ=bmw`}Wr^HI
z=FFPB;96wUmt~60;J4#@HM3{$mNQ<S?c@w7Plp1Xtw|AwIKt_mhY7MbL;U?G=^?VR
zNl9we`2sNnPL6_KKtmN~(tUk3DPJx|h(elkhDS{PxD2p_tXE2iBbycy4b_i=-}GH9
z`bg|u&|&}ZE8e_J-t>Lbci!LOY_((}b~xo|sph&-zr<em(@%$$j4=4ej~^PY1Jlz6
zhK3nC9lhA!Z#OhFK%r140d}^C!vi_zO~MmT9-p3MuREaun{_!$^KYwmNYTi;%1Uks
z;m41)4qDccXY-Rg@9G3WZ4#-`${NJ6ac;*}SXj8IxOkzA0PEVeTF4Z7-fYUw(p%PB
zeCbfN>#_`Tyv}p_daf=jt9(=#mS(ssX+b+OGLqYzNvtZ$^(NqTxA?EQ%%;8Hsb$9(
zZ$GGC1S%ZrH8-Oj^Y6Kg5}(_Ts>(9N5aEsMJN71i-edOmf6kkN@?Tk50qLi45_aMP
z-k<^G`r+v-0q+&&jYyF?V%6}DIk1qbWy!PLw`YvN?<m{YSg*v9zCD+0aNE!$;<QUO
zR~uXpKij67q$)6R^Yh>LVLn)^nlf;S5rNPwvNwbrlNw!ckB{B|<S+!={<DOiBO3K%
zS{pmvbFC|Uv}OmAaw?mA<NCuxLnw1>>>P!uUr!w_UR^qhUox$3-m6BWyY{9(Fw8kG
zY-(z{`;D!0#R+r=+lsOc?K6XkVQ>lwtvJ-fPwS2!pZbnkrKhDiHBNu0<}e8J*bHnl
zL7VmoIc(KmVvv$%m9o3QLT!<EMn`$PMXP@M{)l+|*q!-K`4%i^$M*jH`=X#3q8K^Z
z4vaGFo{DM<aL)kROVO{8DoKGXl3DSp+eO6{rr3&#tba|M?vBFco{Fj}qqX3KIAr^}
z&k0JK_IGBzKD0`axwxFN7_M860qW$SQ?M4ehDO^P^rug{TPhN*F=5+@Z{u6CW*G9`
zJ^r3Hs^%P-{T|C@CbZwvLPe_(B9o*E5@fDk!l{#Y?dO=D{qGNx;%U;i^fgD#tRG?Q
z?SC!LDaNI_fqvI*BIK_LT@CP$LWbxigEFckPYRm$mHd7`?l9emDTz<msYo3>swiC%
zAFP)xNxFHlPmt|J2n@k=!Tk{%+CKW=)zy_g6^!Lg0bKb?I#0o0_95_i9O-r=4GpAV
zk`evvlua(fudNm^g>s!9X#5((;rI~tkL6kTesbo;5uH!|l~d~K94hBxDLsieoQuMI
zS9aSnLufddr&~>*!0ic_ZLD#1BHpxRLQOp-`VK&6MO_EB0rXG4iy~Av&>q$%gZHzj
z=zU`Y#=DQ>!2g0@OuAjS7pFpRq5!>lGggj)frML1um=Y_Vk+vm-)jD1CEYQgB@R)E
z2TZ0=k}snptsq1{9vl{yES@%r%~h1Oh^g3qhL88ysJ}f{4hXVaS0g(3u!$|<1V1b^
zBfXlmj@Re2EczSQkixy~ODmvF9vZ(t8Zl2GCBu>{$2^Zow67dhkN<B|z4|Gn;|Ygh
zxW~rV?T-GviXJ;oz4iATg`LIETa0;7k_?BY7gbdCKJH79D{HF+bv4V#Ilm6tD5En}
z$rkJu*_-HgfDrUeu^r2-XVy+WVD-fZ_Jlh`ggn9ipvKdTw7xkZ<FDmi4PK&U-PK0!
zKUeO&u1woq5};EsG~3_lYpeiOE>MD6#E0<%N)}(455OVtkCh=I0(LdOK73TZ-8}--
zvlftz;W*gRaAPS0)|h<H0(=ff@_O3c%+b#CVUfR41PLK0z^ZUqP^Rs)34G3Pg@9ug
zzmu(E)rv#-Ny-_nLjKV1f6wp9lU`-oHdeV;d?%-v@0v3Q-?SIxX49afMY2?tfoCD$
zuZcPSjUqk6sf1S6_CGN&2naZFu=U_sw@|#e!2pNu@kYm)C7;9PUY`&s7zGS+^=rRe
z69&zZQCNWgi!=+C#WFiv_I@-Dmb!l0FMFDztM^`UCJ3PbyGHA;U*(222s^o4f6RnJ
zE|@4m0~m!t=VJ$yv78{zUrz~P|FykW^y3A2Nl9aiAdO{X)}rSHu-AAUBo-x#V@~s3
z`u^CnS3LXYNp-m2pt`fEJ56|!ge|{SiCzgz@cGp{uT+bWyx?EYUF|-%AKExBp3Bfu
z391at7<9Q#?nLNcPlI{>+YI(%5<vpo#C?bjjF}QeAFNBJ#Q5L%Ndpd}1z8-fm~Y6d
zX3bmA7V=6(-{<q4d<i2W9DWfO^gmp5uN2&qu4a5SlOcJ7do=DiA}-uB+|WXb>m6LI
zqKp5W>c5x=EaOC{fkfg6Q*TWz9uWd1eUii^ef_`j%hjVtD01ZfXwfDWNCO`mp>lUD
zy*ij-Q$|Ytf7`yR^JWH|_1)6E-!|DC<(&(@TpbKReWAa$;vd^ASB&5-4CuJ-O>0r$
zFZbUV0ldjJm!t41zeAcgtVt*_rtJ3BCjQ6a{_|D@d@%g#)@;+`bZ+uTUi_~Ci#{Rd
z0~qi6PM1~?M`7pWSQP8<{~x<<BW(m9tfJ&&gi!r=elsI2iVd3zmX1Ph2|@0opZz)e
z|7*mcBE%k~`;(z0oD27V3$&}H0PgwG8JBC~Y-f7)ivLDSiXu1{;8<XcUeRqH@;8R7
z#LI~dg9*ZA|1auM#3Fbd1r-&0AcUG%WV+(n%h{RqAbDM*TNDy+lcR6OmNOa0$fEAv
zVNsZ9DaF*4kyU$Sr5kr`<JUFP!LYo_qoNv8H41q#Squy^CYX#QB1~sr^YIrgU-E~~
z9!thlvfLSCQSyz(WlCm^Hd}O)xbg5i>{?e^@&mVx*?eR6&NIiJK+6mdi}ERyBwStg
z&qvb)Uhf$!?qet=rFhJMM#m<0pwRbVvZSy!Ck1HHKR!(lNE;2NHZIR9OI!*`epJan
z{dll~F1oR6v|BdPBVpLh24N!A{PVFjH`?X+Et}IucrSLb*UK@tC?Gfc+gX1e+ec(H
z0ul4)<E9F2{^<adky&ZZD_5i9-I^>c0{JS?*szuVe8$XdrJ0#Q%6)(5>Y4vGAhY3a
z^PcLE(O1&Q?=`xDO^k&F+?S3o<lo)?`SPE$(qz6LzdFjyvH9zVO6fB+s68&!)j)g3
z)BeXo6_uexL=Gz)5s!2PC-_0mzV$_X%5jHveV~7CcG#a!?<wD9H+kgR8aFv%+<#eH
zd=jE`2M=2jBm>IAx$ga06<Ao&BSolZ(DR!~mnlAZ78l!3?a*|E1q@f8=Z{1RLqd8N
zKkg#9z=Ril*lqtd-fCDZ9`yCq&?dtFvmEU>1iKd<c4@|O*|E0OB5m0M8R0)i;@=zx
zc4IvXP5i%KR<xG;%lS=<18u^x^q-dvgIn^SNfbau3hVUYIYF7Kn(_a2l5+-7?JbUv
zStS4UwWxe=P2fHFH?}EdDJ(}KUjH9+Y-!0m%1`l8$OGm^QO~9RcL#jzs~22-G(%x+
z7L7;S)cWWGivM5?Kwx%b@b&+}3rl#x9Ds?EFYBxSMwFtvZ^J@HN**rtRUi>}{$rks
zN>F?I*c7N`pMq1h&uz`T|GNI|6rXKW#mmfUtA$x&&4>Sm3e~yrCgl7AY=8Aa3=#jI
z48R@x?c=PXa=4EwV)XwSVEBEdU3&G4x!E}c-+vw8?p=nMn4#JaS1<W<RFs4N$(Tjq
zw=b;5!UJ>sdVG4D;XIeFe>-y3@y@-M3yFEJtR*t?@W77$?<11OOiHve)*EG6nDXb;
z8@fi8dO1!WlAm3EX^T>U56b*QU>r2(`Gpxb6@hbq{s<zsvAubi(Idb<t-e<l$dgzd
z`ZA52a<p2yc;eYU@tLwcJoM*o=qr*3@tvKWk7;wQpeoRCZB-Q&6$AqDPF3BU%TSX&
zv3q`QzXcUjYw;}_X5HVxZ%tuNP88Erz2~TRKf>_tciWQ&+2+WaLRz%LAGW=cHRepD
z2#PsPSy1*CMt#|J(5$**c=`Q@McE9%bKaT$)kwYXbDSC%H@$h}@#7AF=J=lj(q8=S
zY;M2XTCj+ZawnLtI5Xnp_xkyRGh5YK^@GeiM2X$;Cq|W(l~yAK^bA49BqStpbM42w
zOS?Ne-_ASg)lVYu$=Nv#N4FOj7qQMLG>_y|R3a%O)iqy&FR9rRi+>sAA6=F!o*iVX
zdrXV>m*`)*pT06Zk+^VMBwd5()!HQQ8^Pgl_rr~K=SF-yyfhU{Av-e*3psiD#M3Cp
zoIUELzVtVtp}0p}MXM22mi?du_0>T8a^UZi`~EfWaHyyMcF|0MQL}ejvx=u@(+tfs
zp72MFKY#MnpKANWv5IvNM3y=4`?#(Q<*DR54&}lYyj`O@-lnE@zmbb~70pOXgBXq$
zv~m(8cDK34Qv}7Nr`zr%PHS4}>gtATN2#`ZT%I4QrVH2TR#}J(4RIWu+E&P&BxbIC
zI*629nB@0F^<w+H;6D1v@dS(ZEH7O3J-p3TGx>5M)Z0kVL1WBTFllp1{u99_T)nLk
zf!Jmz)ofCdrJmCS|56TiDJ=M904*Yz$;<BgX#o93Q3l<O8lT0NSMzUNYUbi@-`ouP
z1f=UfWNw7@MbXt05{U<=&3SOPT3JUQiN%Hg(N7A{@8Ey`U3<hJssUsDhhQu{E;a#t
z`|qL#NE<Mzzl&<{BK|)tQu8l8IoQtUqZzIpf1IhL_b<tPCA>|WfRcPK{ZD-=3m<F;
zWbkQ9&=dUU{Jz&MmW6E>T^+f<X-dGuw&B|UCNTf84nT;%i9+h}GtH}|{JT~Fbcxt5
zVkk@^{HMtL$2!daU9-%kG()eJ^6%>RAH-oI{g?3n$2$H)lk4wl|Jl)h6Qutj&c7+6
zKkJD9cP;s^9jpBp5rIOyB2En0|9|f6Ki0wh@3P9i*NAjZ)BIv-=x;;>RQ~^ElV9C9
z!O(sHg<384BtOzFWj)jec)o<erl6UkxF{(*?(N&E3JMcVZjSG6!}$UO0-S!2u?YzY
z4PY0h`1$*L9BzE_K0mn<Y?YHX*Rcq68l7#yljgVX3GdB!y(K+k=!l}djgP<E<#wUR
ztS5*?2oXg7{1WCFC>>1Lo6Pfok+DTv9<`v7Cdl<_713VYy5a=>J7yc&)!X}Nw@-w!
z>@GOAadEA#`+L8~U(U5h_=?NEU`<3|V7$JpKw@BEykdo5U_8EukBxz$)`7yrz!=fB
z1L$GkuX4&|T>$VHICQQxLXh}1ubB8ZV54{l7zZr31QNSJESP9n0|5^FpJ)Cs8f_UV
zIUxqdPfLA?8yFa?I)Y#zVKVT|yU4)n7#IbnXfWpAzwWz0;d9(LA-FI#TfbbYjMB1I
z#eXqTV)!cI10Gb+W2Zz0ar<Ir+_0m&r#tKCNG<86Pn}h_Z%N4eUK43s29&GUgUeu!
z1g|>J>4UipcQz_{gEORD`)|{Ysqmicf%0eG?p{!z@ykBLauZXBtG^xXd}|k5y>WU5
z*_B3Z4&8&}!(^f-zf9OKBGBi3uPX<9Oc^~|h{aBx95)lc__0Xvo8RL2h4^-fRJaCh
zIL*;eEh^J{#kQ%Z2_jnP^Xb#@*7)bHPKg@3r<ajA65<QSO`kQkrb}`SZ~J(%JXrcN
zE9XMF_f#>9({Q$XmaEF=jYnb{j|@fG?RpIfkNpo35A=oFcJOZzZKqwhDko-DeHMRZ
zGF>IGm)1nz*eY_puN<$xnVg`+ajt33u4OAtkE=*BKk<O<=81CrXy7NM3~C>ZZXEf!
z;AR$<&)vUWlih?!Hd1CjZ%=4W1{qUiRJRF?t9UMiq;8)meqHQ%>hxK#V2bZhz)HYY
zo?hb9=fx8%tHc!JEPpZ~F)SaUqoHF>)6)8dPd%B_+SP}7D*GExoySZmHi+@D<ZkP1
zy^J~pa}w4T#B;Wto@OZYIj1Ai>-~IPnC5!}+c3!^BT%_ISTOgxf;gDB!l>cTgxqcz
zc+4>#I+f8gn7k>`<~WnOaNB%q*0k^}ihBP$^<ho<n-KNr%$G8w;RL_($MeHWZVhKl
z2%T-`M!#|4G|b#igNadTJ#v}*QTAEbXC+~dTt+6m<Ylw?i2dW{6#kfZCjv}0Q=iP`
zeaA(vd$|K(B*sH!cf<4suC`g^e$oWe@@de6FTq76kcfs^_tULh{u*{o$+EduJi89a
zYHA)AhvujJME#R&RTt$&==G1)1SU<N?~c41{#hDfBXTJUEp$A^_HljIcT`6Q*r^j)
z>b;sgV(*hoayk5_k)mv~F6#XSa&I`0mexM~$N5-6-?@e7+XorpJblqg6IA>}G4GZq
zmd|)rgzR3rV0mX%^`L-hpnp+B*}{3WeZX8cXTG=)=kaWE7&acIlgjzB>#glQz}ek2
z0!I~=`4T*WK4`~H8Yld@ebTwTosN7<(X%Cc`-9BsURSRcA_=?ggnOwoK5i#(cC|Qs
z=niY@Gc7`mL)I(RsZ3l&XEq50yICQ%s}&V!$`x{#Wl>9a@AQf1-JXZ5O_Ae$J>6|y
zHf;ek0%wClCS9dZ$_TC3o;7_oHKUzKON)J|Zl8Win7VgT6)KWgDrCoLNc-^YB;Ha$
zrj^FcvVGH%U9}#0wCQMQSTWn;>eWs*{N5$M-I##^{^7}z(am&GbX=1A;8J>@sI*M$
zoU#Np6*0B(p|j=B*?g1qOVnto>Kf8@?!A+=jJ}cG#ALSo_*36ZpLq1EtMe7Eu_?AU
z;j3QUtj}Aax7PO2hTAZ+>)52dvoO<w<WkJ@z$rz9yQtw1#ywMP*}VAv!Ts4O(Mq`p
zDn6+-liME*#ah3rmi5=&Zsw|M@h`5DmC0QmaCtcTw6@4HYHccKq3>3zqfD#9Z1wiD
z;0)YRo0$!Q<w;qYRtpN>P2me;vzKXUf+t2^4{qH$H9@boJC64^*RxCfTv(bpg?}`*
zT-><S$b1%ax4pZkEG1@pqj&sbMr^T<aP-Y)xd7y7IrsA*d;4$V_GH!e1J~tenf4#6
zS*L2|KgC^a+NBLQzuy}EZSAHc$Ls3S8KY^sI&swf)l#%WqON#6ZdgE!vsztl=d{?5
z<{WH~*4Orxg=Uy%yO4=gW}52k%#8JR1Em<Ft>GrY-Z5J>G9+Ya?enWlTS2vny8d$d
zb<;KTOU$?)h1sU9mMP5J(<Afw-DX!}v@8;R=72uie=wc6Q7%wtj5-&~u?-*Bp6Z*<
z{BX8(v9|i8fuNyA`|ZFl56ZHar0WI;-!{mirlzxwP9y*ay0mBPTMMkiZrcvJO#D0;
zB+u7Ms&(uw>h18;n-*j()QtNOCvyGU(<_yNgY&y72*~{#1b<U7|A*fB|M>M1&K@Sl
z4a33G@y(1gSi{oJsh<_$@$^Kvz#-x3bH&4Fz1aPRJg<5-?$Wk(PMlm=W~6Q?g;%_w
zx%QeLC@7Pc?z_W0!UmkJHH#}mzumo$zLkFmc)8Qo>w1kOV`HBQR|`B05)+uk!yeeS
z7|pO)`KX$N8yrJ05MbhW^bq>_H1-S*8%W#JkG+jd_s42N6*iXMe7t9~L1gc<{VVB%
zy;qdi)>ldSsw|BQ<nPJj{^ot#K0a7fziYL2_!B0V*`fL}pR=;+fiY=XcX$0nVg5tk
zKw`3+w?mIt&*!V7Ed)e47ruVpUt{=XZ|CK4;Mae9zJn>;b+0StLyIiIlCM$}x3hc)
zk>#wLsgk$Tsa|H<#p9(d4#Ttc>^T37ZBweIMr3E9Th;7$BnwB;I$M+qWxCoMC7Q?P
zC~HBCxLnTlQ?^qcQHG4YO$j%9Ifr0N!{o80y%I@V`4Sr^(5Q_Waxs@X?m`xNy48gC
zw@6o}IGoQlOnx~dd^Q@#z@=Yk^oeVwkga#q<JZ@Vz|Co)`Mi5K8J0diQKkBPBsGg0
zHA~$5^7}0pcHzb2$(*(0y~8Nh@>vnbKGS;#Vwb79Ki){)rOoux-mMc^Y&!Y<?&D_T
zOY@_x2l-D{<AhD;cT=y4K`^v~A3-om19Cb{t2z#CpEX{TmhyxJo!Q%LHVmDVM*1AD
z#^o!J^`F`wrCw6Z^2p@AF?C7pqFwiz7b4P|h!`dp<zsEXFJ_(Z47nJ4|InxM({6^x
zR9JluvEM@b0eP*#;rB{fIn~nLOl&UE@fm^N^?E)F(Y2<;OQoWV34KKB=&zC41u68F
zkq~@qzWSFYhs&+U)(!7R38q*u;;`|tUq2#36gxjj4H5LaH@e!Y-$cpE07ts|6gl?u
z2&@;~XNdJVTkHCqFM5~lpnd%Zsb$npeq7t##e0371Q&y(vt_T!Lqgap*UP9f1s$pt
z*!2Wk2U*4Y=-Z4=m&T3!8<o>CTsuD5M^HKcmUuq@^8Za<a`As%<`1gCG%T9T&&m7R
z;Ng1jcYA->3C-4B;%55O&t_fDbVVg4x4LbsHB@v=mPE~%`1q8m${n?&Cp$hl9dfGw
z=+v`d!6d8vg6?Xuw=$O5_pg0jEq;FA+h<$8yuB>4Uz1_qF<|-#@hNax-|~00UHP>Y
z!OQ(Zu1yM*RbO|?(B(++qs#nez$T5J%%9uU@!MZrHPu?=y}s<W{ht>e3p*z6{d%}Q
zrR3esXI%FRvnAvXpNK5XpM3E1`71luy^rnpmz4beUjE8{eFnRuqChVQU-@AE|MR0`
zw|{SszO^p!+dtutm;cp8=gwZzwb${l^5v>;Glb{<{5!GDUw6JlKL4iP`nNlKm4C<V
zf3&M(@$Yl$e>0zM(3p58X#Xx<Bdb@(clz7i{q?l^y@l<n=;KeXA5VK@c`4x(mwjzH
zLk+N$cUy3t;ebxprEfLGYP-C5R~=MRdS_g{eeUY}y8E_Se9yHMzWjIB426$N-v7QO
zd3$f^7GI#+yz^`Bo9^4@>%z6U@11>#-k!|YX{AmsOEP}H7n19X+4OL`Vtb$1mEQb7
z^Eb;+uRPYe+5XM{k2NLT)2_Q%cvXoQ&e4p%oV;s$-?mP5?b74>-c-G{Hd!;RROjh~
zmy4o&qs{v{c`vjuOc4f_nXKaCefrC{Jo>zgp#yj}$O?|Gw^pCF1=!8zKH$Ft+*}9t
kjd8cGaWuM#5dJ5>qyF2ZcJ(ctzy>>mr>mdKI;Vst08b5`&j0`b

literal 53086
zcmdS>by!qu`vwe;2r_^oAw6^t9m3Eer64^tf;31C-JpathzNob(jg%&-3`($F?4t5
zyKwL4xA(L6^Bv#6?>mPh9>Zd-`@Z75&g;A{3|3N*d4NfZ2?Bv0Jbxys0s^7*gFtsq
z!6?8dn>6wdfG-$`XIc&*5EkC;`_ACB_az8K1$r(i4tGu1Omy=kJBq&%`@l9EyCYP-
z09&NOphH#JM$kXOkf|z7fv^P($`c(|eE%L$-p@Au65P)w^HBC8wO>V38`Ga<>3)De
zgzkNG3x>k{X&qbVKDEx^<u$PdZ@Y_h+Ja^G=)pB(2y0{;ZEw%Yg=bPT4HyONO9lLK
zHB8(jzx@Z;@La;z@Atp9-_QO2>MKMC`+e0MHU{?ZtAG5z{u0C7V&hIctWX`BN)vPQ
z4#Og2Q&WXwQIsbTm(0}Er{_El4i1U}OCN8pcW#c|YuoCV5U}&}^QO8mDsVx%!Z9`u
zDlWal%1;7HUVi?Qlao)evFhsTDA@%C<=Xk7L>vloa$knY@8g^#<`fq8_4aP<?97%~
zjupJD#)iOlo3J>DYpyR3`etiAPJSi|ePVw}TIfXurcq(_rFwBJ3fvrj_UxxilGpKs
z$9CmPBRXZmrWaC9N=nMiY~^C7-Uk^#gyOoA=)zuqX<s;2?QyaltswY(ry`bB>nY$K
z2J&J!*pQUQg)0IgU|6KNzP@f#e|3C$c^S`RCc4*xuTx?3vdYoI#%3FhNEdN=YPVPM
z`Sa(3s`cCtA3hWp7n9l646HeZ^w~}no1E=0Iak+~mL7Jo6kg7Ua@^OO9T_nx(v+2z
zEw`C=U(d;^yWH=F%Fz$HaLobVuiNipK@)e~YbCl<b28&$mSFd8+J*xP)h;q@-JY(-
zg_D6=5(hj;#BqM%f_>xWf4MG%gMeL9tdUeqJ)okcmB8_DYm?qH?(6E({{HIiDU6NH
zdiz%;@B-ECh1-IC=j<!Hy1T9Urq)Nk^@(0sfj;TE=k%>?tgRJIZK;2AI$9rDN%j$a
z)78}#6@^<n=Hl#(($U$OQ&4bkx1#<^zK$oVE_9uPje`Sy_xfT-u1|QsgBCQJr!_P@
z{4qY>qGW=-YBhC8QH=b_%a@5%VB7jDXKQP=H#cX=Hx?Eazt>(v-6)QuO2OmXY*+*l
zafjoe?(z9t9gSi6?#_7Z+==72O`-zVpSE&<E;s70zx@c{=@Z6p0PMOK>2rDaiI>+^
zkx@IU_-WzIc_9!PeEj?ywR(>r^Vg1!j$x6J-7XA+F3MLK+KqdMLyC7{D+vyK){~CC
zkH>_M$Blfc4o^;|toT58@80#LV%4i5ZUDjvO`Oveu&OS#TmaPvU%=;8Re~TM(;g6&
zG;Rx(zh+&L%Z?tX{&I<<mGnKEPC4il@K!9Fj=GA9{bEN{RaF(fqHs}R;Z8yIc5Xqz
zs7LLo^V-@P;6qzCH@DZhG&~}ME*KXs5ANRwW#{DyUu>4_{8)%EYL7?+qO`=cmz0tc
zeeU24UdVsO?MpR}t(Jo$sq(4}s!qhr%$!RU)HOG%V=Dq|C7p5QXX2Om_=*1h-9b5y
z+UxTTGJYG+&7$@s&;8CsA(!=9J<s=FzI;hc%%2W_Hfw@x3TX2giy#*Q4Udd??KYyb
zXcXk-=AJh`^xi#Ph%D4BN-Zo@yl{`^pas)xEBONm1;QB;DlRD*%hS4DS=Lac!-_&^
z>;CbStr+0#va&MU+9P#a+no>s7Ga9Mc1oYTlk?NM(7jPUn_s8v-wL%^V=W6Rm&5sI
zb}z2@XIvMoE2bI1zUtCJZxg-)kMSjn-dY1PmBrK^X8JRM`Fi!UWx<`@&5}M$-~Rr7
z(5vIk312E$;?EBs@V<?>Pdkmhv}+{*t#52h+nl_6Kn7Y|T+Gv}J>Ss3sfx3!w4D{L
z9SaqJw6?bPBnBV8g!|;gHFKH$Oi)!6aRNefUTX8p>}4huI8yYS2~<7py_WIG)Wjrs
zFe^V_9><@hum-O|qyBsZKoruNTa^042;Zl+k^z&p66x0%G{4)Ls_@y+nLP(gR#H;J
zQF}PJlH_4CQ{x6$|MFt?rp|mI^=M;^L-&W$&e5J1(Xuo)rDzR+a|Hzj(TbvHK(gEn
zC10by+L|mI)APuu0?W(G*I#Yfww6QYBe#KI$dLQweYw|eZEbz;)<!JDr=fooi{Hxu
zs>A<ht42<A?cJz>si_NOAf>{w9uBHFP780#>oPsSt9)O=+uO3@=Ei^6Em#)TmV2J<
zch2D6R!czuT=&}A+5rF97rW3=kC$2u&9{bbY;3q$d4^zMgUk}0Mo;QaYO{T&cTOHY
zc(Am*+-CO)j8Q+k<6OPvd||xPaVUx7-`FTFSd)N*+PGxDw6r9D9PvUYLBR}#Vg_(&
z{rUPgz#Ch}JI|i>4-PijE<FF#j667d7Z}Ls!I8=4OVy8D3Zw!{NJvE25)A<;D>6~d
znV8c2Wsxp%?)!w5iHXVWH)212{J6gltUS~rbmVn@c6QumZCtq~Yiw?AF7V;neexMP
z#x>Ny*&Rv<B;odZr&r#ex*tz?a9QITUEkh<P#h;gVa0f-)$tv+?T2VbUq4oYg7+Cs
z1jTSy2`-<4ex$tp2L;3*pZs_li>c9}f_(7DebKd`x6s96oP!8pC7AiwrM67VeFTCv
z&lLtSu#4gU`C&{gXw9EHq58iczI`N_V2%6cM2T7F2YPoguj{iF+aJhx;}x_(^;0aZ
zsF3;YV@I=w`qx8u-x*%(rZfw!xz^4+6g{157rWj|XRV3;_iFWcgn^qlIXO93Ujg7O
z=D7O)G<|BT4sb(gNXSpvLKsKg=~K`Bv`A6y8dp0hsn*YrEzd75daW)qzt~h=9Z$)~
z$hd9QQa@*e0ow<5@aMjF0i$U~P>Ok<OuN+YwULR4h{(mVx*ZP4w62U5>icY!jX66z
z>$$IGSdQl5qNV^)o%^lee5BBOWO7nlRn=i3od5I0b@KILme<8G0TGeICD~!Td-If`
zWGUR-m5fnRE+8}1?rAA}b$7pc%b9V`vB;wNkx0!KKBSpgs0G%@$Z0d>i`wGU)cMoB
zg^Y}htE(%%FZo$nGM?T56VShy90Ogh%80oGNDVNH{k?X|Wami>vt%#6H&eFtwV9a=
zdt-zD+Lmbu#f-zxMvoo&eh;nu(LhS?(?XvsOQ5JIC`33`<mTrSP^M>Qjt&nyIyse<
zmTpd#Sq`L1%^xr(xzg}REw5!I>s8p0zd@&Pode+V-2W1{3W(gVdxyFyf4ys-_v{^F
zv;yVi5J=JO$+B<R*{hW+NmJhq{7yX`97;|bRt^B%?~UV1%geI@NJ9O^^bQDcx1!ip
zX=-X}b+t}e;}_eyv*j<gNgZO1Z_>MwU!epAN->xk*|GuraOp-M{tVVOHoE)zDyFs+
zvJPB;Ta1s#on9VqP0hM4JgA?&*=>F}2E0Z`^_(Y63Tuxi%dMvX0NAd-x%Swpb+JGB
zb^l)my7jxw59GaXBib)t*1jIe-SN4(&?&caI~virqYK<e?B=XbPEKytZ6%RN&-Km|
zC#NGI@VtiP`D0>Ytf$Ji;?0ucH@zM{)PhP1|1F8$XQZ}wGA7qpj<JIPbMJKGNf*=}
zYwtAz)hO#KU+tTEzIItiCApEHprGO8wE6QY@V_$RH%j~~w|=ML>&sSxzrBH!d;W6~
z9Feb6q0n{V%S!O`AF0nEgUHjZv{#uS0+RgQ-=HB-BN8KyP)G~kaOs8r+z%i{cw&A`
z+w8q<v?UkyJf(jC6sSi4;-VIgZP-_)y3RQYzGT-Ju2*>2M4^{8B(PS^^^y`mIWJdo
z?Dq7XHeT=dae*$AG7oB_UIdj*SX~|))1bEBeJH$}Z|h~bHkgs@eQu%p`=zcPd)hEp
z0dRmf?*Km~B~ggufKEN@m?!`m7Rq(A>YYeIEVk=j>*!i3io+2RT}L)_^p^M_`EWX}
z;33|@?)jBB9iCk>p-?5hApsb_^|k;YkWKuvo~k(h2?+_SeD)Vx6`_9nw@ng_GY}5W
zKz=K3i*C-;dhE@&`cm;eer#fFyg5~23y?pcwBB6y-LQh2l5Z}}Kpgd##p7_tMqf1V
z`ud6relanhp-g$b!V`d`RRO$vX=*@4Pj9BbAIDj7G+$?Ie4KbpjJ>7h@paXF2n)dO
zjuUr0R{$0h_nm*_X?XpQ+c%BU0!I=UsLed8WqV^YwXVp3xN;@9|IO(`9*%U2BD=Jj
z8qel;59-f)ZEu@`U{8GkQBuV|{I_jG2a-VECB&y53}s=e9_{{22qs?1Gc}3#H29U(
z%J2w(dh(ZH(J%Ln{E8X(4b=KVR5bw4a{;_ogX*S031(i5G+R~}tqlgdefyypD$kU?
zM}(^g{H1{kL=yg6QXjk9Y%tf}ENzkruXp`xslk*VlLq^d54sPN2rY3@4T?{wQEn**
z@FLRpv@3S;2be}2w$*Ii)rf}v%L<$6+k&{;FY)lD*U`w)(UCA;k|XZ_ztS<I)RWBO
z=o+1njt_wYg)0{T=s$MgDiO%v`CPD3{Ew>jCv5ZodF>xn>v!P^Idcg3XZ5=+SGKFW
z3<Ccv_y2zsN(T4Xu5^k2ej0|yZG;b`$pDod52ZlvGq6nFq$HNVz4?O$lIk)urE&ZN
zDcsEg(xDE7_2A&3;ja#${kmmq?Iu2@|BatD@wtFM!NG2>EKq0?(9;3+9@alODQMId
zM$ErIz2nW#F4Ro7X8*5(#i3!SK~{5XmOuRae`pD$%ZJlIexStu6#z9U+?W6dSdyJD
z%=ll06)@k2e~|;d0=jFY{NL?flM*MUvASzd|G$dOJ8TYsR6PG53&p$HWk{U+3(5X;
zHLy&}HNcF2dGI$|3H$^w5xW2JX|Rv$f35vbUZVT_vdr;+P?K)pZg~HSy6%`3v6}yi
z1^!0|E12ZBo7EQlJ1_Z@3@riihVd6R|L0TLdCHmpT07bA=;P49^I=Q;JN^1I0PO=<
zWBmUjW&in<e(#~}f302qkjp3VZecyem{<He0L5ck^yL`;ZoGmP;~dX(X797^AKO29
z{eNTmfBDX^7>*5TsPnR$PPBM7-+r0J`!B<#MxIf^-ZVXRbZK{T$s{w*=?K1<r2dyv
zxwN4A-oYc^ihO)@Pt2>tV?$c{aefN2{G6hbqeuSb&)X%PUcr_gZBCPoiF~%9^X~C!
z8vlN0Ys+rN^q+S6!en&4wUwRSkSg`D&EUQk6tA$xw?-KJRJbDkpYag0$`<2?&gCPz
zzF~K0UgPpw8(2ffIPv6E(W^W2)t*%Uy$ng)97pHWVmxMMJG~+WAw02VF|Natv{aPJ
zO1qzA@|VqD{_Yr8e3PEpCEsa^o!^$G;pX+<v|3!EuEYGdB?fe$ALhPmBTw303N!JI
zbAs+abif5d^zUH`ESBIoVsWJz8MLD(&-nZ4h!DiTawvF2g&F!(U*FTkLJ-&vSPR;T
z@_$SeL603PYe0Q|u~4lsgi#EaeE05sj8v>F-jn}>BM1P&2u^jm02;`<TT2@PR?zy(
z84)#yAC_i3(!$OjVt<(9ysHJJ`7jr>vwyp(+g1wy!Cwzs?DLTx=JF{7WCD<qX3+BT
zkL}S?fIdVa4IU9hWqd&Wt6$)#ssc4Eg*;}G$HkWTy|*A-?CM0;m^JMGyWg&1^F3WR
zdc(-G=s8XxGI152rv=x?D*P9Om|II(vxS7@a&sF_0zw7x)7X%pN5~-Cnh*b#Ag%+`
z;e#r}Bb8~yCxW8bU)XkkWCx)J-U+5WNG5JkNMmN+J)5tb_-8T`O;0ni;J<#|hD{l9
z>hh9(aS4qgE2D8Cc&9*~u|owHmlk5j_|HHm5)om77WaHsBbEXMGgQO_1PNCG=%D_d
zu&&4h>c?tv0uGz~DM3DCOHXas1HC@~;ehw3fmAIHCI3t_{nxKA<>hg2Vxdjqca4sy
z<)n<QSl=H<#|IA#{IKTx=PI!DI0gqnn)VJnH8Vjz(`#^W#4p``Yc=#@IJ`vzK<B?_
zIZ;)$vI@!*P>nJe+ij2u=f+uKD5xd!Oz}RU+?=?yv@n{?KNAnFh=C9^7pj2h0Bo>R
zz!^>j=$h_=TF@cVzq-1o8lRY8tb}sz?(Qx11$YoU%Ff5XFZ87SN4N(hQ79t-NiaMW
zKRUN9ij~0YHYyr9QQL)fBx}q5=@_g=FgnOqo0R4XB{GbPhv!hg7=B1ChgOV7V2;*4
zQ>$5qsqC11_ercQ%AeioZJh@%BY-wIY;Z7W0Y5Dbvk~_sN;kSaQ{HZdN(%};eoUl{
z6(K{>DDe>KQXgv5{YOF)bMX<Tr9Jy2VSJn4`l5h*slrT*B_YXinYa-t)W2Mf5&*E_
zy1{s2)5wh4K1U@eWY<nUUW~S|pll`ZA4m>6UE<hVO)Ii`^9Nx@%fdf9o0oWws5v6^
zJ`r-*?&<(9@|b3Y&W%?*Bqz6<VF1?jGm}Ue9MK$DgrH_6WYrBaYEK=v2db_rqX#+z
zVCXQ{4YtL_mf|~q*qjK0T@>}M)vz-s@6l2dOJT#p+rGdZAF*GBibLS`BoSh=(1WkU
z#qd`+NOP+e3`WosFA-8QE;EI69R@ji|7N6;2oLif<xFf>m6NjB<qN)-FTYO>j1kHC
z`26_*X_;1>04i`A-xnq#ZJoV*+mw=_mcqy|fAzdLQD37In*?Xls;c?Q-rMK_H8-E*
zVk;XDk1cdWF_kBr2D$+Bz0T``kX2JjQ&Ur4|7Oy%Fj&!DIO*~gu>%eAb<Tb4`9^d=
zb57vviyzZ(&sB@#f2iD3i20`%d!K24W+D=%7#@zRo{orkI#X*@MHfyX=<m=MUIK6e
zG-@v_b6QA=+1-ZC33ID3!|q_)9H14^tajcq`Sg;HlXDkn^$TBC9|JmA{2Trezi%1)
zfnu-D?0NYeCWXz8w_A%%U<WU(Ch(5<fTmv7IJWB)wq5&aW%~s`5Xj=z`L_Z6F(y6~
z0Y=Fy##$>4LY#Wypu*K!wSc5TX;{F(h&Xg~6>bEM6w<0lgG8GEq6ylZt-oOuJ9`dL
ze4zI`8CF%Cw(n+JU{FEmd!H>4km|WD#e6HM>RLGnkBH!V{8;LwVa>5^f8*Z68BbM!
z92a?A=-SQhqYC2)3JRm4ZK<g>nI%irCIX_Z03<^4)FE@@?m^(#o*(7hgqeuPiNd)k
z<Efu=q|Etom8wdOmshp(W1>f35|XsU#5AOj7yOL0-e>qReIaNcU0y!v@-pQEo{!~*
zE2Ki0lk+{5&m5g7G0C0|>)W)?AGk+nL~%}O9z48wGcYiaStr`K^f|Bj_CUiPeiy|@
zL2xbIW!6jMd<T$6Y-*22fs2@CfY#4ezr<9YRfva=Z**dUq?_=*W}WWp8t(O4GR!)O
zXC;yG)bpur;dghZ;VGakBL_nM7xUAIz7<PbTmB=JKndXT(+&Lq846^Ls%k<J7cw=p
zB@pPIE_(R`C@)`)u_I@%<#AAvzZ~SP49@m<`)(UEdHJWdwxt&WMGzQO-b)tG)0Exl
zWB^U`^Yiz9{Tg1pxmI-D?`&9jDE7$vTovd<4PHI@F(1+&`cq7Q*PV9S9%#=ut6dIy
z`T6<y+Um#|@NtovKo0^OQ2=Vt{Pndr=$3lBY>wNFYMIG5=vX3OW~k_f)roq4SqYwf
z_>CYSV09<b9biESdK5X69lZoEKK4qwScT5#Z-<zB-ODst>5q;Of;!qLMb@g%(!`qg
zdU(daQ}2r;-2Y{>xcD_Mj5X$TVb9;*`>6mVJUqMbVQqF5DH~gqjldpTOfXX@Aot3V
zk&^P+&ah%jclGtn*0>$qHmiWXi{a`*;9c)yZqCHd|7ljIMQJe#q{+&qo&z-D9uTZG
zFo`eKeQfN{f;Aq;bBobYUqU~AY?$_YMS~CVrTW1ZtNZC&CCTH*p!Z+M`uxUzg?;>Z
zY8Aro=ckJrIJDM-MX{CMy71QV=J_Xkn4qAN7KyvEHCqH_drv&BHGp$Fs4=ZS<G2<8
z0lgDjopQ>m{99q2`*f8P{=v4r=dWCNTl-z|eI6bz?xqEDSG+8H4}(dyoUq7+EV0DR
zL@-TUp~^%pStt>3ta9~H-Y5wNg5A-zbdQbjK6ZKeQ(czNy<}=pA5cbdevGQ~K9;IZ
z#_#UOvA0)Jdh_<Jt4(Y!9ONqeOb$yJdzy7}iu$2mY;0<^&>)*?jBHb5pUT>fi~%)h
zc=$yli>DbQ9SfZp<>yRhhIsGTE&%Y)Ru;tIG=enW;pyP5FcRe!(`ax;hKgGg1~Dl>
zC73>)Ityujx<(?I50r!K633eo?{}!YP0FCDN4d+yEUu`SuA|^7M-MRm`Zj%49fYHk
z_@Kekk}})I8;h~i?zI+=9b_**pf^U>vVeN{ulLVhmEC*&IE~vc8mwidh=ZEdjaIKO
zNY;cEv1gb5I?N&w<0vNw!0TpqMUV}@>-f*$+4$EeudpHt*aPN*piNPKG-YCPQg{#^
z)30SQ?LxchX|#{}L&Xs<c;uG)sabX2RXQM@+ds&X@%Ly$Rl1+Pp99N$FRToj+IrVY
zGm(Zw3(PI71hX`8aRTgJq=X*%o?ByZaPFwhuEaPeE)?-GRvH%|=Jd!pvz)^tG*lU7
zTz3z*n!Ac{U)sqW`Jo;a0Z$m`uerRE-L<UfTK;9W7+RU9Bs+AlFgw9k<suL(VJLK&
zGGnfzcjWjQ4_}#uWnhM43R@9uILP!SkJSNxvj;%OLsFmTCl7Ip;hE@{tORywcfV#c
zQL8scQCGmBP=IAcWilsv3zl1d*{tQV%_%KK&1D6_)l%}qH}vGK6BIgy0X==qv1^F1
zNux3{qL>!)Ir8}n$wUE%*Nfq@xDlIHEnLu)h%*mim@B7^PBKkLv@}6}@6yqTGMi2(
z&|;oyL2GN7peS!rQ<=v@Bq;i6<>-Tq<pKM3GR><SjjBX-38uVGdSQ;JTvUVFROsM5
zJ@HFOKvd|ASh?fIX3%&ul$EJUBZ9bD>u|BlGWOT9ew;1;u=Wa}ibSktE=e!d0_+#D
zTLOnfJTd1hyZ!<D&n!Uss8+P3i;M0D>8$vS!5NC|aH0|rV{LPYmRv?9Nm-elk4quZ
zCw+>Od<DK@2=;&qNkuxr3Ysa$;xc!)ym4GPTw6z3Ens~;Cu(lV8C^PiWsxdk0Yxpg
zYjykH?6w!oTiC{yli<A!45dRh!|*LYLXA*z;q!VmqPQJXWHFy)rnk*LYe9W+Fdcrz
z`*GAK4By(=(UA$Sf1d8`$yRe)cywXad|2_`r=hh&O`h(}x6*<@;8@5>^?djYjh;^5
z5LZ_hMhk>4&sML^QF32Pop$Rl9N$yD<2GmxQfte8_WE`DS=B9<!dJ@RR>K=`Qa%ld
z;sUsS62-@l{@2wA>bg_I{o{LY+~lM-RXA|7OpcDXzH4T-p@^V-43z^E+9K1~rF(VF
z$SbB=aG-QnP=0C4Xhv|PagMQByDB1^l>qswxo~jDvF_03*UT&ED|LkAc5gIb@)u7Q
z3XM$xQo6755OlDdl45ZCoab__8xMvZWk1HYWk?*ATmM+k?R<F95?{o{+`y&{rH##-
zw;Ia#J+g=K+J&W18=KKGGU|5FaS@YYc?K7k2&m_{l5dxZF>hIXq3L}-?=Kuk3sOZ$
z>XFFB{ze*u)lp{uwL9>(7mRXJPPlPGM1k1yGczDgi=3cN6#My)ZJvu@_FBhD3~Z9r
z3~p9qLK7((SaI>NVlfS*xw)f<H!9%sD?};$-Mf2j4x@R&SzqJvIPGDJ%#L{U^ijvx
zG=HRIM`urhidoc0Y6gQcX)`Q&M!Ae?vW$$Bn~<T83`DB2AvKax>EAh>8>s=2AL+lw
z3IhRkpn{`>lV=&QBO-`z3|<7MgK0w_JdjaYAtW?pHGwK5ha~OUh0m>SV10yj1=mb4
z-3k^vdxtctc?9hZZ-y_Jo|G=YGov$7yO7WPhatdooXg#X>mu^=vvZyY^=Nqq#ddw=
z_(nBdYED6Jf`y$}X6o)fe5M>N;fWUbO9huSvec~i9%q~ypC@a9gvF8Dq4MQ%Mex|G
zq_y@yTa>hSx1CaQ1`Q*HTo;Dqg%byBSi#_}6LZ1nsT4~DVlT9y*47~cV?mJ_gR_yT
zshZ2@^mM*deFX$;1Y?@gdfMiViLvkY?5@~2*<$kYEJ}<cZLtG|muG#l8Ou^p-xU7H
z^YCeLkVrMUa<`iF&(~pGo_-OPgsrc^5fVBg7F6>q>UIgh4r>tdS74dajM|}2m9;gX
zJ9w_Rz-f}x@ue-KC5=^my#|tHmE7{gh@ooQrd>03n(zab!Ju%w8k7+0@*1|ij7qC0
z6ahT@mx>)^t1np%+@P%|*w1PA^g4_ul{mC5D(pVE)+-Dwvq)qB$V(BctTEq}mAGhM
z7=}o};(Tdyb%uFM;P7t|&xs(JtOWUULHho8_acfEx&8bGvo%rK2m%>-+DY+x4T{<0
z1RzCrD~YFLK8Zcw>uZ=FipiPeSyEM3;XGOH;XWA%eh)I#Nz86bwWFiGdyQ+2l;!>1
zl+t3#Ewzia^~8@?^sPQN?ETX6C=_vc@(oR<)N+0e5Odb)@4j&FN@teK_$EglAKWf4
z5d!EE8%vBY#C6%v9ytS`v~M<9lSvEdY11c#hbxC!Sht?S^xQi)Hx{tP8mS}CbncxG
zbz$9E9Q_ng_CnHM;ze&*&8Z&;35g(*6M+3Y%9_wG{0rq%O}rNr6zbpzW<uoqs$x0+
zZX_{rgB%AJsR0n&Qr(i^h;~YT;t|U(6%w)q<zzM~*#3T|GHLtVS6u`MQ9Yfe-g^3;
zvsb!T3id8gBA&v4?Ah@V(ie-tVZM|XO1SqE`CbRg{PX%S<S7X4pv(VfFb>49WK1#)
z!HFiLr<XpM1UQ~nP>$Z}?O^|V8Ye5j4K#ho)2Z3r<u9jeiX(lBWO4zObW7Jo6v9D)
zlx?9L>sP-vSqZR#E_g^5Va`gR7(%ne^J3*bfK<gYIl?HU<pk}gFj+Z92Vq~~Cfr#F
zc4fu~?!_+lJ~XAOY}PU#8bj>Vxgdrn<jKNV{qo08(TRN9C6mNr-?eGIWNtObfDHRi
zde{-X<{+YRqt&4=Y)(lFs><AyLI6Rs#)x<zRuaRc#!HLfrGb)k`=L+hE?0c!M>O`9
zzqpG%w9Ps7eD>Txn9|ya0E4-SD<2iTTV+p^z?!;A28b|;2`jj_7w1rjk^Ksnd|1Z(
z0X4@g3RQk?%XWNFoZomC$LG)kfV$*lA==4>k8>Ml`D5{uDMoJLoLX+FZ0x%^uAT>Q
zz!2WJk-?NuG#57+E4zpr@mppyiWs|H6+M&&ASf9DQ`OE+1{2(m9(}EhKReis4as#2
zk`xyY<2HCj#n~#7!WyH01udV|kcNRAT^?pE8t1acaN=TTW~4rX>~BbnZ!Z&K?H@3z
zB0A)}tofRA=fF`qP(9n-@vO|u(}fd?^B|%FbJBI;gZp~kB!LgxDUgR3BTjl}b)t$9
znHw7LTgk(pHAaTAHPYKZ7rO8kKL2L_8Cy(EO%4}6H>l)c;B^i?c=zm830G!~1zSuC
zb!l0vgBjzyZJXopFGi$+-2^{Cg_>{Wzr=B105#Hy>m4^0c1)4C50&Ew*z+`*2;i_e
zf^9B6>{eyi+?3TxnpEMc3fqw9Rk|nLqXSi7+;ol=4O_4lgFv9RD1f9TzI*mm^}wn;
zSTiNqm)7v!z5cO4e1(Ry6{CY?;uG;N1|}!R8<-K`B?<t9Wg1j0%%8~1ztRE*lz<|8
zs5d)f18qtvhO@@zaJ72u=>4)ow~Wc0h2l0u(Z)T-0=<4c-0hX0v44HxI`{T;YfmzT
z2sy_D!zOP^5CAFZM;f*Nu@h%*U?)io`tzF5sq)5+@U4aSk`$mPIR3G*jfa=dSP4WL
zQ;Oi}Xu)<J=@mq+t$F;l^O%<ll=pmOAzT`6_zzOzZsz8w>@zG)wW|V*<~W0R$(f5A
zcdDI{c_pB^Uk-bF?OfY_#}|f_J7k@MxDebVtJ?^=Be#_CY4@lQJxTD&(epTdW<v<*
zDmoLwzdkX^d4Vb~&-F8LbEjSGhNs-t?5gMWW2G0r`<6RjtF%S*BrO3hNDP^7k?Q*#
z$AcdD@W8FRIgrbR9T+Ym0Y`AghLmFjb~rwQOjRsYZ5UoSTjav4s-$@^AH~YXK@_d`
zN`t(y1MNB_A3@d$y3dbRw_0T=Nv_!T{Pg{4jdKo9gj7T69H{B}ztu)n6R;d%N90-j
zSf#wHS;c#78Ly65j5vB=whqSjvi!mvcY9QU65HjmW3bfKpvq%PjC$1yM#o_=ObJ1}
zae2Q`cd;98JC~DXIMHU7Y%l^%&yx8iY=7g;s*MUND_b|do1ZZ_=-bv79ERSGUYkB}
zc3`(lm0Th0$V3Hu{rb=z1{A#)jksSjXbRU~KpKl&IBaUT%5~mwHyIRZ4i3t4W0uea
zA%c3-OPMT5dv&0-c?@Hxy|&-Iul$`Bmx+*oE_`+6SN=Zyi%Q(sbjP%REO#y~3*5rI
zp~ek_!ftQH@035+v3K~uM3Mp4)`r4GMDVAF-l9LE^pJixv*~y2e3WO)pxv0Fi)<mN
zr=;|J^H8*5%C^0j#I8fbfV!=nMpZgI9EL=F<776h4Q-EHxID0%d&24=h{2h~e5a#p
zo)29U(CNabfaFh_3(DUV_a+Lf5dslOOTQ{HSR3RdKWq9L2FQT@GolHqA4tTt@+bx{
zv7OC(7sS;F+S(YTE5DsHv_~oR)Ys3(kP0>q%57e?|M2-nXQHWs_nZX~%I-2UylD+B
zGE-!|_(}koNb}l0>_C<A5OUspb&2ukjy6jO`q(ua%ni|=IN#~(<bpno=C~sTQ*X1G
zjVSa)tE}wkxjz)ygQG0hTR=)uWS*&b=b7BlM!~O9LL38N`(^N|YP+?J@k*Zc=>$l^
zPbKLuA*nQ?^6`tz*38U3I&!XZ85N(b(OL;Vy>IuxqieD|WKw`pFC)LPe_A)%-XiaY
z!WW+g1|wEe9zpE8M_m29yzW>Nb#=sUEAN5VzBHy;mF8P9<KchEpgBCm&J5Tt)>?9w
zIcUk4kl{OhsdrX`pg!J-T*@!=rFD?R-BkX{0oy;=Lk6y|+_p30;q?llT8b6!wa?_y
zn!IGnaIqtN7mmj4&X!JQLfa#By*J0mQfFRRvY7TXY~rnaZT)##bS*o&g2mgKTcAF6
zV$KG5H!G3vXb^VDO3;G*oaQIYt*_IaSNFg+$Gmztv_)Nq=GdQbAeKQMQ$@WnzMQbC
z!{apyVJM(H_H}E7>4n-~aOps)4O}i^uPyJ)4Xf^*twim}wq)DG=7A2Y7~WPzZZwg?
z@!C1SD?NH=U6-AVh6zvOvyEmcQ~7&R350Q&;62s-iNaA_NPfO%$vsNXuim${*-u1H
zN+0pep8x`o!FF|sPQ`YQ;7P+BUC-HKqm8u4onT)h)3*){ecIg{dB#ifZ>{*8a3oD~
zg6#9lo!8ac1R-zB@wI<|u}7Tb8EtuMGEyfaB(0`-?$3)v2N;V;mTRj~=QDx;h1b!=
z1MXvCni&<GIW-_o3!}+Zd3&knaA^J&YPI~r=>s1nTmWLXX~W)n*)Di~d6r^QTSmp<
zkYMXCaZMi4zz~fGlU(ivX0=$WfkM4Ti~Z>_cjLh5yHE~O)TQ5UDuJ8!wGE_A63Nw~
z<EZ)j$&qbF;ihNgU%0KOMH(K5RE7P_%&wq|SS~0R@@Ui2cOh@ITLiCfI922n!&^v`
znfwL?Pd&Mgdw5$<UCp0Xgalw6T~-jkhaO9=ugB>>Q!RnV34|4qP=CJGj8^nS4SZK8
zYO9wtPfTj&Ln$0YE2Al<^zI`~zRDd{#PFyBKr5sG?@a3g-f@2!i!x!-Zqv6{4Ljad
zt1@f5{^3h2!|CC?T7NEG^1RfSwpC1YeW-GD6)S>-XHi7}blW9NsSGIB?0@85h$}5;
zjs28S?7>gSvAzD(n$4=!O>5z2wxvN!=xeFWGzXaP<Gq12X<rGGoLiYxFbPcvn9=^R
z`xH>iyf2MvdG)suIbmuT5$o$rYy={Pa!2svc3H417(29e{>XG?Dd5}lXyo2%i*!s@
zB+xfv=5wLSXpvMg7<R-%n{eADEjm^9js;Y)Og<&n)a<`7yN_Xk#1+a7mc&8g2#IwK
z2=L3N5<t|yd547kv|z!+wzdHdvM^*M7uFJgD(OzUzDzBHyJ7&7s*~RoqCaHvaHr}X
zJPTfRZk*T}D1vLwI^}G93d&7=3WQbwXV4sqO-rFS+Ls46r|9K2txW-^ZT$1!dzrpZ
z%pJ6Pq=v{MR?`Op0WM@6O;2O#lrB9|K!ZC7OIQDH_Xtql4v!;bj49lN$E!ab-CcUA
zT}Q=L=ObrQj@d~4`Z9Q(F*_X5GOdD}z^)VZ0x`a&R>|(1$E47KCgWUg(U`KOP-U*n
z8vEMVS7L>RmG9AjG}~=_Ikpd#B&Y>8=WS(?BU@MkjaGij(`F?(D9rhcen^$jNp&Mr
zroH?9N5b|naMU>REiP78clToN^1yYsvB9VQOg=1PKUc{lEad3It`W`k<mH}Ei76L(
zB(c)VF-K{0U^j~>0tMOJ5_l;}^w_xAqhE!xhHPhqNhxKtKj+Xrb@QiX>1Kj4GPk{-
zdw=f;h3nH&+*2}SkOnUoJY(K5;}a7N#3a6;t?jf1E1JmxR*D-DtuQzcm#Kt>E&g4T
zO&YjmJcHiXQsl#17CgOoI5IM6BbU3A-I$>_bseaI-CkrwhCTY+x0x@i;0oVaL)NGz
zOc{g_3{;_gMk?YB;W2x6{f(PMEjHv5)b@k#Vi7;VE%PaQf*V7j!n3fRZpI|w_~ZIi
zX3r;-$mnGIM_mTMriQXu!ZK($7=l@H!7?<Z^Y5=mSJ48Ojer44KAoyol=SpC3<IZ9
zR~`q|-r;B6v|rmkiwnv>HH*FMN=;;FxTEIZ4728=0mgIQy(^YAW+<K3vvRNp@Ct%m
z;wv#V&QRyuyo6`JRpsj!!%Z`7S*l@vyS<lk0ZTnn78d|m<~MFh+Y|_g((Ek8`(1oa
z9apim{GJj&B=9+@-TeG3EeuXB120t30?v?wsO01V>o29YQdsqWTK+&ylUxFb|I}=1
zn#AGA51{q7*f(G$Nb;5Nwv4bPL^#IfY4ATC@|OSxu*`)`pzc0voskg?im#ok;!kTX
z`HTjMF(Afj#HLU8rPc1X%@ZTv8(xbRQvJ+)_^M(>hmp6*;MhJr&j>}c7eph@27_QP
zuE-28r&sVJVlAD-fCLIG$VjDyEeb$zF|gT1Pm)B^928?`v9=h;Oq`|hk>JO?j)JO~
z2gHL-x-0jL<fH^Ujdby`P(*Ja-LcF`!VCAUrU<Y29w_l6apZt`cEpI9>w$b6G%W%r
zX0D$aa7<*aWpJ?x!`J7e@~Dl}xLRY+1on?4agwtSKAIKk;Ynzew8uu*y?U!aJ3S14
z{#Y6&{Ny+@w+)J~mQBMLfr|~m!r4cQ4mURv3dT=i<4A+Ow2wZMlkV3Y&*Va@2qssv
zRS@^@Yhq&2OvX9p^s{5!rcxLyK`hXsVYWxUtR{8%#QKudoyO_7=R-;+Q#oCyu`+^l
z7}PE52phJ}XwrtxKc`5WJH2#Bun#%UO^grT+jC{KMd`<kc=xX1W3)+k%TGDG!$-j3
z*;-bfdRZQCUmR-;v9q>GP757up36!@uaLqkg{E1c#Mn6f?&~RuLiW;c6Vb8J>gt*J
zhFp-!o|cUV1V-qS);Xr%UIfi4p}h*Dfsm8`SR%aKx%vF77?_^$?Aq}#WssJhPsW|<
ze`C_UyBPhdrR2re_$?x;cfqo`>;(u|Ucecy*V~N3{9}LM><<~#p~98S+eL>XWNjX#
zbG};eQb&d`**$kNJ~tOnk#|od$B#QfLFc*XLQ_XE{PP!ORii^FP>t<uV34Sp!WmEe
zPcPeClZKWwtaE_D4dO_ajWYs{JLg?2r&&IR)7)D4Y|WYg2bK1veFs-m(-m-U<PwcI
z_!%l9n9je(z|D&r(I`$$+l&PwA=#(T(_8tLq^>BfSGCPAew@MY#zLU8z%;|_n2Hkg
zILQ_-zj!PhVf8C$qNK>Xwv#Q8m7rdV{<}d@OGO4I=yJ>lNmP%cTAOO2m&K-xD{!`$
zJp>z$?P_=;TV&ubp?V3;p~_W>M_JW|@|c?U8CNp#Azm+SunC-<AspQh%@0o3HTF6g
z2X;zZqG)%>zvSaU+}s4}%<GI(0-}fF)ve5a1#6Be7ZtvPsQCjW>=}P<zhNJ;Vd>Mt
z`39{l%~iV?;`%IX#c2DYU78#@UV_e;h{_c4yr#--bNp=}<srF9!5K<Xp6_lfCS_v^
zjeFx>MSV&gg>5at5*cu;bImf+5)vo3H~dxWKoHW=MS7Pw7V{wc#3{^#APx`0z(%0P
zg+<$>to-)aS#>IeQre^l=!<F&dM?`U$N|3yHT*dv=^-Py#H^kEnm(<PgheC7D<Ulz
z@vNjtja3g)-hSEWx7qMzR0V?;#+#K-2g{Z<!Tf5O^W91SFon7C*9U>2A7h&ida!<M
z3kn|->$x+ZClZp7NxXh7^GOYPq{GUrP90LHwVJekvO=Iq(iImol?g!HkhqYX6rH2Z
zFE9S<1F^lBLZIP!vAteo!~|7oqxO68er~*;GL#UFf`}-zv%z6yzCQ(4qoImlQ2Dib
zS||7VjvV4eY)0y>J$iUf50Z=QjCkv-eZ<uBaE6@Wv}$Rr;$xan_D{eiSV}rj_h!k2
z>7*V{RbPpKr0e%s5l?1vWf4pg90@@UDiakCv4{+3grgZH8dU5mf2mhHQ*G@p0l#ha
z%tf7Yjkh32+<}M1h!Ns$a>c-qd%0<5C7&u!D$mSa3g?}r@w*+M#gG{eufNZU1Bv&!
zqP3|b(Lci!z2f~aZEUyu_1o*ZV=|Kxsc2qC+s6jpKc(g5jAKJ2SMw(7DT!YC-<;g+
zcConc0KK<nHQytc+dA_vd#sC?y{e6Gz|!~2<r^*Kb!XgPfzHmpA0ESo<VFIX($i0#
z+P_ePMpGbhP{ShAGVhU)41oP9uW>l)f|73_q{-A>N;M1}+X&z!m*R<+{+P~uma9l?
zm^o|A$4AZ{-j%<niY!sv{c+xA&}<j2?LB=PSKs~Eyy#nDy!uAK%0=1T&eWBXKC0#F
zM7w+CHfqyrN1`Zb?zj=9yr)-fiH8xC=q0PtiOpP`ub7A9@tB+NAnpRcXEmx~F?nC|
z$WM7s7U<LP3{6gxw086cjD+AThB+;c0f+jRSnO5Lt{i3EU+Zfe%F`JX#U<0_5<uDj
ztLRbavTH>zCqUBEIcG<9+C@%Y4bWVk_}pz+WfDCHv#5VRF*~+8)*`130SB%>EkmTx
z!3-<l&f6^(S$Aq{%R>yyJcFB5xD>2|oV~x4KhXZ@D^cZ{?Csv7s(5i>P&C)Y0)WY}
zvl=)zoh}#8QQp6KP160G>1#RskOsn6kG{56)jV#DuIqti)^`8hyTxY9JKs?QXE<0a
zESr<BA04Y@_J<>$$rll_a=O`J-Nsn=hpal&<;8K~?Xv?T3;Pt~c5Gm@0kr(%wQ5Y!
z^mJ93(ZR~bFaw&ojp@S)WgX}}RT?QEm&AdoycmCpF_k+Dc!k!QKP6w|7Z{4Kpv!Eq
z-@i0l>SChp#Jf1DpDKFB0`okRDlm%5Cz7!4LcMb@ML#?|T|SljJ6u{uLR*vy#%I%l
zMueJFwMtWSQ+F&vMpop?Wm8LVabh$r=y~+*$AyjjY#$X@q>5qM53s*>`W+A<D-JDL
zOpG1z+iFW91p&0xwa$DN^>*{L@9?&^_pzB5L*|UbmOH-6nAp#4etvFE2(WCa^U|y#
z1}O+VxH$P%Fxc9vtk%Yj8kmvE@KDKYeT_am6N4f>BOn}MUJ`R?cyE&AeX)HpRZNWO
zAZ*6<uFs9fxCU3hcd#QZq}l~t^wH44{kyRtLu<Y-N04OBLb?SyHjVX{SaoN;K?KZx
zi9*!N)oL1P`SKZOFP!)_v|Plis`}L&NS(KbFKUap78h5gVL)58-Xuow`CAD*tw8Pf
z!I5;VZ0YmCD%=QK%1xFBm=Qq8aiMZ-PSy9;elJ%)@KoOG-|HB4N!d8+$0;sx0_YmL
zX!7|?VOL5lkZPw&ujK3-?K<gQFRZsbwkHGV<v??-M8HUm_W_x%Q_R4uiW1HZ-&9CP
zY%Vt=45nA(-KyCEoPJjceyylDY%x5-Q)X<!H5IPCGoTHf?6fYzjELu<C4_jK+}n3v
z>1%KFvkL{J-<Z-;g9i9uc3+^Q>-(evArt1)k>Q!9Bg1?9&=}ApI8Y$xV!K+Ij{;ss
z5+FXfifoP(qYRq+My~~xV0Lz*k+Wg{wGHSGfD)c1);d|nRY>pduB43#T3TA&geT@^
z=JvR|$9a!Z-UNM$Ob#IzI`l5gKr>5dWM>?VZG9JvFdxNG@ozR%UXi^$`y$qYR-ql;
z;f<Ah6NT$cMOV17E3S{|(e5D`m)QD=uV6wf96R_SDz<~2EhRLr`R%3Wba%iY#{f%P
z6m12Z*f%W$?StW<@~diRG<A72G+?6YU^&38eWJu^*{}W06r0Wvr5JW)SM*CSr+Z0~
zPvo$G-%xb-Fk@vlkc9cVgc0e1g?S$*=|!4HOPBS6H;>0rWt2*#-oTk!@F2?4FkTCP
z1;)WaY0$fj)LPT~*iA`X44Bx*`7E${WuZbAUvS`5zUaA{nz=3ic9D>1L{buy?Rc&@
zloL3ip9tgB38cL>+h<gT3A+R#A`RWQ56F6U%I;Nm$;lJ6)#iGqWTFP<Hru>+nQf)m
zLGh~@ST^lR=qyN3(Se3oKEU#l3n;4x^dMp&wtj#EXIws#BVmpm+_pC69$d~8PJ%-3
z4&rK=2BD#~i36AsZ|^*Mv^o>~5nn`5XhapQw1bTxW{!AUL$8ySFq4e!sD|_TtBB9d
zZ?~o@OHGH4DKH9H4U-8eV3Hytl}@eekyK1~g#>r6c8rD?sABgH!Txz^ry|xqtPWbJ
zq;!qSdK7zah3rv*v8A`A<RitLU8ZSy&={?f5RlVc1PPKD;x967L<jTDdZ>z6T|#uA
zp<giQX$6;;Kei0Br>L-QCNW#R9wxT-mvA|id%|&-^J42mF8rNpI4b%ZGMgYa^rdbN
zyC#gm8&A<AmvJ~YI?EdY;rm(mTZ!t!BWw9iEmD{VRk;xefBHG;K%m~;3S~g!FyYj{
zqRKxoR$ctv0XE|!wl#=FpxmqJF10mV@8C#XLL_DCRMd}OGzUIy!baaea4IA~L$hy8
zku`mlS`1H*7Um~Hd--7970vR#ie_t};V5gVH?l=d1KCP+n~sMAVDE|hy}e3jnlUk}
zt4~gY;w;=fly$AjpoO-Lmme>7+F!b$HC0FhYGDy$N~L|gzuVi;HaSdrrU*n&m8&H|
zZ%rw0Sa@r@%D{_IC&8!>+~7D_Qm4xIMh8?PFNflpgIJqV9y9lVBYIwpLMwA$ADVl(
zR1DP2r^#^H>U|vA1f;R)tqxU4PtMMhp4m0WEOXy7S<BpV!rEG{LtIf;&Q~WOP#ez{
z{aOfz&oI$Kq-fL9r{*OT_DadZ5RFHWqoe!cscOjv{VmB$8qd_wfr&gkY+5-f5@i3V
zwG1Khbk6T%3=!0x-Ds6yY0@sc+FVqXbft&2{8*~IP8CSd4?4mI|Ct3~lg7kaiVYEd
z1C{Y^94UdTs3`HAq7kwxT`qs1&;7-zR+&a(+S^o2bVDw7NloGPh|(u)x;j3GdSK{R
z-l&$i%oAoMT86{2wG9coq8aY1hC7!R_2?1NQ5ksU>9U1hYB0z7F01qN!y)t~k|($?
zca|0~?>r)!N~pER=NTuHYP7B}XbbCb4-<eqlP5<31ZQ?4`=R?g6yk|^Wf_KjkP|JH
z5sutE$aOIr&rtTWr<d%KS)70gUoP?#cXtm=kR0;^liXJRcN@Y;O!;#BR1jf=>grmL
z;>(fsZv=~i5{3K*(g;{~x2t>f3DQz|%u;inU(s%H{^0zMw?>zk@a5ZS&~3W~9F<Hy
zD}Fntv;vIQt*$9+IJWdxA;*ORH`tS!FO{9G?ap5rZ&detO~YigEpkfBi_rpw1C;ty
z?r=J+VkNky&hsk@pjA6}Cih#?TRnNHt>dmQ#9l`X%2OBKL*1@VPP8<_wo}mokF37%
zN0kYSm1Q#I<-7MKznrkQcYQ3Vk!SB6Blumc?D%HMDq&4ty0S=(UY?fU>`l@Uk6H=o
za;qkAbb_K6)^U0ExH-hxwns}F9eZv?j34u^xt$=Fi;cjmQI$@GBjOk|G(f-h0|yG%
zXnYXlO!}7Nz|MA5?xK$ncWuN?97+#Id{<6nTvoW(slxRWi4$yr5ci`=68xy4wXFe;
z*j0lHlBC_vAWO(%dP-#=s^dQ^=za;{z~V_0y{}pbpMc+HDv%%h3Bk`0*@^>fI=eB2
zPt+<%{N{)!olZBujQcjLs8~^kGF!k;oMh#sw#U1ou&`^P1VO$M1jv4W&r}?Ug`(*7
zr;sU$o4)R5poKXaqn!%5-%1@-h@@LzZ@JL?g^CXPnkntzR;G#l7(8#Hx(PEh&XMF{
zX?k=osVHP6sX0|j@bQ=2NWRpXflP*;1d{Weo_G_AaNeQ0qtskz@k4W0V!c9qPBLVN
z!eatTjTMZ~Kaotbc3~zK3X0#H9LMWl$3)HHY}lrDXFMiDzW<B}0;Z3GxO^Jd_R|B2
zAR{C9F>DBzCLYJ>Xf20wHl?WIvxt&wCg^z!%(zNika8Kt#OvhuUp6|7n!D#4*oSp(
z0wX-PN6R0_2(0M@d?68Hmv?5*3E_M`US2JoXP`Gf6&BqtFctePXtY#5>@3m3KE{TK
z`%4J2y2oO^(cO7d#TGLX^6`-yFO6X%^1xsyb2@{C05_b>wxy&`x|jehWZG2E(KuF?
zRRBUA+GQyI^`>~Apt3T&`d(|p?g#qulJ<U;C`HmRmavrEC)%X6A&!Sk1vM%w3|Z&q
zv{muhm;q3hP?7a2r5M&n^+;)carHhtad6wE{EN^BlCUtez!A*)SwNnKhO(934>RRP
zc`vbGv63uegSbow%RI_qGo^*(cE1%kVx!QTeIwWLm;f210;J1#&Ba*^GiQ@1Mp0jh
z`wunXU(@dXIDdV)W$R9+bVh#P_mloKFO|PVa4z0&l$Vahd~~i#_MDFliGU$4e`vO9
z=&>&op@4F8Q#S~X<`3%TRjw4Fw6wjHT32#R7QE&;-#wV2Y%$y0PqeBr-HJc`Qc*4y
z*mZ5o%X5mh?{_fEvN03mqxAA$>i%ZQ8iJ*2La(BdYSW$3|Eg!EtuZ)}OB#maOB=`1
zf&)1_Yd?WWlI3RJ%g+bu@PR14*J+x)TAZ)nL`YiNT{ShqVDH?>@>CiWA3dN7vgN7-
zw5MpH1LMAx4iC@*BSf)UE*}pW#y?LZ8GSRI>7mNR)X?xmGB!jP@kkRwyB5}0+N%GU
zr%Y|?oj47lT&tD8M9iQGdlNN<UcCp5USsIl69+ew1Y8CId&#K7yN4MXgkjleGI3#<
zGkTeG#AJTGgpVL<YKY21(t4gg+yHKDcALSW-cinPlh?U*@xQC2X--ES_P`24_!?Gs
zK)#nU&&wNklNs6W*V{+Qp_glMn(=ER=mX(X9z$Oj4QCccc}i^4#H6H*RM;Dt2uI@)
zo`T3*_0eX{2Ui9L{5K_t-p65WG5Qyu&!Q%vz%=083h{#f`ICNrVjDx)=-Js8I>gXl
zV)HoN3>fM>oxOmhts!Y3?rb~{3>d<>>p7*}mCpt!PCiy+JFV96_UvFAtEruiyjC%x
z4DxNx7N2;-Fmq3nwef4|po)PY6+01k{vGBNj}s`Xq>s<=^E5YKBUm^!Y|^@HEwNor
z){${Ah`)H1mB7^BZ@~B2(syZ<jQJA`lK9VZ&mm}mFck&q$o#?W2;o3KdYI%>RgXRq
zex;HEYEPe<uj`LP5KrthR0#HRt}*{8jA$VvRe)NLiySYyE95+PzR`M<<K(mGSDNfi
zIxBdE=<;+%2?PeljM);O#$z;Avs@@fvynZf_orT5iMb6lRzb&X7F+CvhIZf>o1$gD
zb!5ccLRr*j!bG|6i3S+E)-DP+0CT&Zvi>wc9q*puU}L~X?gUg26VvYT{UD;+Txe;<
zhBlOpKb5t)vV=?5zu6`-_2knTI52K2rRNeO^$~5~Wo?Z|;Tky5UTe&ZnSD1WA0@_%
z3KN@_*TJktPcz?2F5t}KAvR4j9v<XgZGGtyHqapFeyN?_Aui4lQ%!y-qE!a-{`D0s
zl>o(Q2cy%{rG+5kELj+Vpi5qwG#7M|5u4Ezogo4B(>^i_XBv*5@q|q5aDI@^U@`m7
zo5YnW?jD{4iTDnnx(ggb4Tv&91J;;%LyM7DGb$KM2+w^Cn|kv4o7(i}BfhkBVe(wU
zVE_ha90hW<KSOA$=0{v-Lr49%M2?dgs{2Ww-8l<*YLmm%a*AxO4R9bP(ppM*E;9B`
z-Y9K+Dv%l$5;AunHP)%a28;7?-n}N&b3OHRT_NB6l`~~0=>y50O-g1&y+*aWlVg@s
zbAYX`KX<u%fWVE&cK6}>2EVUvn_^i6S&H(s)hTdk_|>W$ZG}W=Tji|#6;xHq`bwPK
z9Ry<!+%eXMzR+b)r&Y>GotoBVY)Uc9$)2)}e0V=I*8z6#9;dFZ-}3VJbuM6>J?wp0
ze7P_wN^D5-$Y+7JHz^5M7ixeyi2Mi-Z93$phw<WKug&9phl79#&8?}W!l=>Y9tgJ7
z`|%gp=#S3fW%mI+a{$9A=Mlm4m278&ETZ?=noyuX2mU`aopn@|&)3BPX_Rg*Ez(_=
zmhQfQG}3~=r5mIhq(!B>1wp#IK|rLtr9o17=lfgh{ntf#=Xqw%oU`|5OY~AW4RTi4
zl?!(>t>-6f+^<GdDo*jM#|VxbI|};JYmQ1|R8jV1{qa$S=Cm#hvlI30we&=wDZRnN
zJ{zWLoh0woBn67G@8eQT#Zp7VsgZWS>KC-$%0aD1arxD5Xz}CX-ZbK-8|ASeVAC2Z
zl!og*&&hcuX!b5j9|^cXkC#80uHV!$N{sd%l?d~4mBZ?Sr7@+yMZ^rzK|rzP(!DYn
z!Y5eX?)~hULVc-wD2s#=k4o@Z10T?1VRrY;zM9s0i;oN0q5H1~NHo8RovuT<4i{c$
zuI@YAL=Bd1EhjT%5nS$7dn-t>KIx<Nsykp~{puF{`M)G%UF;J27bzex(>L8%*t*ye
zmqxas$z$OUeoeK>#0Kj`DQpovdu|`Yd|(IJWib3qAfVvs&OO?zV1)AUBKw{AdL0YR
zDwZga%UT9oEDHENa94Gh{F>88Gj)7~x6IFqj<jM2I_{;Gyk6DPOi2;of06pD2FGt^
zXfN9@*mWF{%K}vy4u&s39OKN(i|8Ob7jk{HG|Fiu$ntx&95L{ccFN5B+u5e*gw%JX
zZl3wI_;BG1y9=^h6d_F@ejqA&({?l;^2O=cfef#@?|m1ovik2jjNL4(F7qp2)zIyD
zYQ33Wq-W1giA$D)$#|uLrH?!Jc6jXQAbHv2b+Av*u~GY{d)6+aKmkE#Kf3<G01kzm
z>geVdGF15RjmL<>KN!CmiG}ZtJ0)94jRQ$vY^;|<#`CVOgCczKiS~LT&ej}-d8@0p
zZY^ZN^rNP|2bwQs#R#)+N{fdl92McV+EIO>)3{ysFTN+Lpw9;}C|gD2fmhZ}=crrl
zq=JG>a=c0v`p?ZB=Y1vjgzb+`GdIEniFlEd-iwJIJF9UwEt;+E1dq=|kWt+BV<O%g
zK?<4wCh#aGPCgzi_s=EvU;dd7ISnifmyVDrj$%jVH||r}NYV4P{ndpSj2C~ZY40o$
zTT6;oE3kS#A+5u@u)$F>h|X}da6}5Re8Up7a)>Q&qMB7>x;s(hWJG=RO`m~w9~c$C
zI7$UP?omI{9UJ^U+c#|3`0Bbg#%i}xn0(13&pXn;{-Ih$jJv*!<)!d;SAcO>?$q*z
z-E?c=hDXn<*o-Lnr<0ysI*S=@8ZJ7BKFa>7*X2z9kV%?+LhU^jFFoWy`@P@SGpgFt
zZ&W%uI;4=xg2jw!j{49c-{lT!J<~Y<dtd@#lDE6C{u<DH5Ec^h&6*Y4$Ih-%##cuY
z9YG(e<y@>6Z_7nGo|BgKiXKv{zVPOM2DZ=!*d;y0=P+sgt!3+jXjO~aJ-vHM%@5sV
z#~ibNK7RL$jcZ7j*5;dXge3MNrnO8=q=)3_7oTZzDZaknC-*=|HJRhw140E<)Q{QK
zEegW7mS=8pY!)yh2<d2vih9H<wd{zdd)vzk_Fz10D@q(J=PDmNmWO~_A4WIrnFKh5
zVvtkq#s1*1Qe8baTj~d5x0V<?*VvQGN{)NhiDtU$6m#6@^Y(j=_OlmDo~Glhf%t{5
z=+za;UhW@A$)aeL4JSAW4pY2%$Yp0|>6%)_8Hr_?^@R;(*}i{{g6uA>@fkg08l%M+
z9l4Mn#Y*5ybGHhvlpt>YG$X1qgN0>|sm9^*2O2d+=0P$WRKiiW4+O^SJ2!Sfd${p)
zK&2=xECM)9GT}{f4oyV{{-H#d5>G-|e)u@zk*Ta_zb%3oSG@s(smx^0gTAnU>(6nR
zY&85$W$m9Q*_cr*?49>M7s1kog*=pOsXY-m&qygKd;zZL@n%56XVWE0>3O#y%bgI!
zr$MFMbJXP&A&m{==`WBZu0BzDE{DCKIBh#Y;nW(4&#fBwD*x5(=j4Qp7yp_7dp>n2
zb}e#73j;`;b|wVzA-;cUe)IljGiAC<tNl(JQj;7B<VvFg`E*_A@?KrEl$KV8I$>+*
z27Ubem_j~{#Wd8p2|BICcI>IIlR{`<y85vw10<E+h5O^`68uYVT|@I)tJ}g^t6m2L
z1bSSFj?i)$zGq4fLIbOuP@%tV>>G<bFxMU)M?tpQ+AjXvx=i}$T=|C)T;^S=t(l%?
zI1#`%?H%z8wT_+WMf}tDr`j!Pu6eVD#c!`Wf6Be{uTKT7F<-vM9EF^MOoE4K-PR$D
z6f-IR3nxMMkw?!qdCj~kCU!o$ydgz%%QL;@lR#=P*@1fhaGH~Y0tc>vZf{xEPsw65
zgck6(r#sO=XKFa~Q=>Jhu`>$uFrTTTDQ~ymAB!B-O(QG;yBNS-bXEskl?UE81b#WC
z+>|9%mL(>dF$e-r{=#|)D*Y@`pK-XNMo-VF*Eie5+Mwz5w>)x)zm?AY{eeWX_C3ln
ztM#}3nrcoaZ9XoWQYOb=;4}p28~;PI9<|LT4!e}#5X};#waaN8j4f+)W6sM%`GS{M
zT){xfB77#}ZE+7q_{-F=c{bC(TQ=TOhSe1^@)5E?fA<4b3gK;A>;%P8dgf-d?T<_o
zKlangzqhB`=^H5(oXQhrC##6x<P?~aZy?x&gf!Qo=ZO~0!4`fHW8<6EC#&9HclU|>
z<zIC$2wrnQqc4wnDhj58k(5*2sC9NEYSnSo0kOsZE+r+WSRULS2hF}>EnvmZ4mYkL
zBM&S!G5m|ji-$wRz!1W+x=a;lrWsgXA<OINq(Ym<3JT3(^DO~lRc!*NPzDqniCD1H
zPUSPP&n}1scUFkZ$v?!~nF&7o{^ny6oopv;+GzV7<hgR6qx?(7{GJ|6(8rq2NJ<+9
zPQfOd2`a#Z0jR$Y09!k0y%=A;Y-vyv+DaQ{Zy5A_jet%-G{ft081c7kxPAMR(}#cS
z5oXVGp3apwfg<6Z6$-KZR$BC*&*i>-Z0tl6ULmvXq7gs_ML0GY94swZmy#Kmn@fs<
zM(=c4ki5m;@CxH!p$`KKy0JnxI(CzC!Hi5U>yJN!Wm;R|6-~sR3RwsGELvppZ@Qf8
z<^o(Y?JunJ=Ss{3)<3wtSgOY=&SDY2kX=ZM9yit)!@&pLe(oWfmiDCu&|C$Rb`sOr
zZmH*2ab>zo{{DlNHqQ@IE5w@CCUtJgSR015f)<o4vT3?<LLD`gLDxaN63mpyt{@8-
z7|hJE6%!v`XNm}RDC@x$sL3wk5Wr`z|NS?uA?PUdWy!EYzp62!qD5%A(LqW}%lMS(
zmA6B1;Ok+zPaKBf-BLz@VgZ-*e-E+fYhZ4~6wPCumS@UMYa0FzGc{E?v$iW*nzB$>
z_6V#Vc7Aao=(J*_p~2;OyfihXnJ|y}gZaaf)_fgCf<)DWb0a6i#h*&o69S41UW}F6
zkUg9O?`vZtm#Ce2vj&W3Jgn#JYTxnVCrjSUD_XsTHn$86#D$v=YgaIU`-h&W7w*L_
z#qBRFVzZ}5AP|%e@q2Ka+|Y38#(4i_;iZpphYTArx)f;_ZO0VLzobbbLl2AOYgY)R
ztWP-IvywVOW8>%lb6YSVu@PHzO)W$(k$IQ)a7aexyvta5fA?YScdI@dA<{R!H<l9Z
z32j7ahH1W1-r>@E@c-(^iGEoU?yc@Url0ntrDR&U(W2xw?NwUOpYH*e1A)veq_;6z
zmlfLCoMaoJR$TBcqJMiE(RF95j?#R^bPEq#EH&_w>EMvtS77tOOb$AO!kPQkhOCG^
zm9|_BJEFhwXFcrGW+PxTy<dqfRGvSn#PA?{`C%Ued)a=bJ4|aV;o)cT-QBzC2-#72
zC}A5aU-n2!9jpqVudp~8*A`bF|Ncir8Xq734|E8g_WyRBFw0ViNVvATtl^jQ?!vJb
zypqcwn<rwwCgEF!N(V;w=1Lzpm%WB_n=9yA1q_>k=Ut_W0xSN9K)y&dWJrq)l7fNV
zKrufrhZp~!xYZ2Em71t3DlBE8r1nP&woW#rv-n`!J;a)hke2AidsA$uq*lnb$=Lym
z3&E}4eXcBvYeq&T<?+M$sAnp_<dc2GG}+W>6Bk~*x_H;TP0WOwFwf72{P*nBvxN?O
z)|E^n*L!J%PAjqC)9x+{GNC>c>B3Bi31hLah^8k`T;p;oEr~{nMV9wz_8hb@FRFOe
z&CO>yHukF?W<93lBtotW-{U_eCg?dhASJUJMUU_i@xjC9)nVb7%aH{gXQOa1^b+yp
z36)if{!#TZbVtPEvNVimnQ)NP9y3+LlvC|vxE8$RbDZ6G@9q<fSQjWv$x6w9rQ`9c
zI`A@|TI^z!_j<{7;^yXNIF<G3VOK)^A<J2+7AAg+Y?B!_Tb_T2T}U%x!`Er(??1=M
zZWBnU3X9o+krd#3sfS6$6<LNCB(Cip@Nhyw*VXM8icMpmHr$p+scc3EOH#!Md7DYM
z3N;NADcb@$O^LY&<+W)xi@Ro7+3yb1N^#j1b*oZj7dDmHk0%FGCiRpYdNTg%m&o!t
zvC_odXI^HdUxupgjt`RPAhc0B1_oSJtbSkzC$VtU9pUC=)zjB}iTgSh?)TV4=lg{n
z`UW+(2d`gClY?P4SlTnpKsW7pkB1&QgNGf_qxG@c_z@ZZg~JcRt`X5(*niy}=~0rQ
z5sSg?8t#4;Bg*x?1M@dn;7XG|!o2r}<@52*s3i%)&8c))Hj<m?A$A=4OoLtC+$e1)
z;^jp{!AaQ0XCZrH;Qaa=$6{_=%lURD_?gmMQ3m$Zl$4bJJg0yRm6@42J3H&S?5-DY
z^7DvL2EZ@6xSgn4Nd3J&&HAN>O)l=%?iXkyP40+=xi4(n^3LfO5bxT=WyHZlt5SiP
zWd}=_ENwXGS>E;AnLoBGqI`Tz!j5FqlL*d{cDTzSJ{wVGsHFW&f4kn;=m0Y!`zVru
zW9i^=Vy<EEPP0Iu_}`;v1rQ>A<9e?P;MWLZQqdqGHEL>Z*5IQjpr^-E4YvWnfl~5!
zDK;efY@Hi@=K3G+961Oexsdrww=Z1%@8ej|gHr9rUn`f((n6m{_wJ5%wfhHM7z9&R
zSa0i;v8TfEu$!=Xo8P9SS~ZgfOH)kf$wS@7BPJ2q*y<9vd+Yp^zB`|AUi2t|Ww2IC
zcMNt!Z&-}$k}Nh8=+2Efw|MNK23%Ih?2b-VgYUkmV{)DkZF71#$z0=0^BKH_Qbgau
zJ6soz2LVMMOHlzZTdfB65$SF1?p6c(*GL*WV9s_q6SS2TVI~#2tRWC{^e#1>|3pLx
zI!TNoviNW&hW7cWIRFzV`;^=#=<xmuM(GC>F)%VtliZRb{VQ6a#l!+SgVvdJFw4vR
z&O!RX7Y!#QG&7?C)4x}Vy4U^H`;Aty&DBj^UDah$hK8RU$5{%A{~Y6%x&5+i<1~(*
z5P$Z8v%RxBEH)-@SPR2XGB~)$Og8~=O!%|0a!A$@;L#TJ<6?Su)ij|=)-$)lwY6_<
zi{C2~eex{%i0}B0Y_K%#a7Kf_T&^<<I85JnNj&D_Lf*gmw*=<pSsdSCDUyV<l^TUT
zu_*GtXr$qx6YG;g)KW=@72nldxbZ)1&v9&Kl;B~+PtIvLZ6uC`pPhlS_n^Fa5s#f{
z868a-&c(z25YT9JKDov_u~swSro8rSjxMe+j}eNDxVUoQyA!8?9Px(&%nK6TNuN`%
zPnK$Fa)SpMuw(%WwxO^PrL)#*WL)%O^gqUNH<+uQvk;Z1x(9Kv0Q3BcyYBrF^;7LT
zn$;5ZaN&0_#=D6$;#c_4-sRd(7*!6P&1>Ryv#^q9hT*H!0^PKg^x%;$&NiRt2|bGc
znkBw*y0^SI^GTuL1$-mw59+uu6URBKpy8+e<_bJuGOz|UP%K;_nBI)eo;<dMEw$5z
z)W}&6w)evzJ@0v$BXvJ1B&6`!iA5jgnX?%Woa<XfRx&dd!$<<$?sI){%vrs?JysSL
zJ9~y}**5p!$Bd$yk_ufT4>D{@o~g00`4wthWKi+O#+4i|;cdaebKKHWr(5dbST}?&
zAL%+FHgs0=cR!a_HZ}Bb7E2i8r?x?9rIP$-Bqot9C*c#tzsKL^>nC+QF#M{VR?Wm`
zPerU&`^jXFP9@FD%sL~3V>^U@<kgE1grl77RLBTjSol_wcb(4}t#eyewq;r`jLE?W
zjM23|H=xq}s{><*EBgfMJ;0D{bzE*99#%pqEiDC%w^gtbjD7=Nw7^7tZF<(LT_F%1
zGc{N2nK>(G`%nH0<d*lvie{#5B@j3oTW)_MgcK$A{m@ZujYDUJ79c$9(Gv1dc_8R^
z(-U0}9nR%^q18_K%_n(KSyXB?gFrJ^3He7nSywKmA<EeEXI-=cWti+9slh$g1Q6^#
z$>;BB8aD(6nG><%f4f50E>dF~5{M@>{>T|;lD@GXA0h<f98;6rO(yzYJp5I5cG;M6
z3UVn4Obl#YoT(E(v?5~p=;_yzgIby^$g#<CadOou>ATqOcJz>iKgbT)@}EvNn2Sf9
z1FPav8kci9DS5kTGrs`sZ8eGD(g!qZ86V-tOU6ErOO$~<rrKJdjr01e|5vw<B^5Cn
zxM$?xTn5hXSJ)EFqjV;R+kMIeW!11PGz$gQ3Ud*Oeyk9TGwB)&!Dt!l*U+={tl5d?
zckRt}%@8)v(zJhSF|)ERvbgq$-~Zv=k@>?MZCg4R5`12x{+i}Wv6@`NcaAWg{{`_F
zpO?IQX{g$Xv~_NVch}qc&v(Nt1tDsJV*F%eZ4O!Af~7^TXVRkCJSf2t?qo6s4|z;T
zwJnmAI%$-HUI@G4p^GHaF+sai>V`0g%fQ6Xvq9jtss!+$<Ge2rN=S11`iQ@%o<&3;
zM=WDYci{~hDk!rKQ}V7NDC&U|&f683KRt+k$F12^G>)@WCt`MK8tsS=4^q5{I{EuI
zx-eBfJ~V)ExYi0Iepu;YAn?;o`R@ZN;+Rn_%~!$wNj-0=_kZf9+G~F%#fuL{$|vCC
zpS2%>9B8nLG!0vEY!ptUFhh|}8Uw`AI0tI28h5EguVC?G`H}FwUJ}E59;dsVL96ur
zBJIfUu)F@FQS%qTj-__LQY*^=B`4={GBNW);00H3?ztE^L^QIS&}LiIm+J-J2PMR^
zeEF`^tcM#4eHKj&Z1(fR2~g%y1wNbsVx+O(SsyqgSMOJ@+yB;Ai(O7ipZRnEKy8;F
z0Rna;5KXs>*u?Y1<(HthwfP(c)qASi8;;mUsZBe*5>9RSV{gVKYG(a?eC5<tdSQ`1
zLyt~ND@o}0So>?_L=Hk1Wd=bK(B&`AeAWIccI}fyhX*v+5{tdrJO6VNDQc)=#9ho9
zLKhuu*|*h$8%XgheowtCUQd*VFFGL1)C6ys^YJrZYOkqLw-$vv)n}!n6;D4fm#{*A
zRZEE8m)ms(ebC8ibY{wjo)gEX)&5M}eBXuF=;G7HYh5&qm34m~M+Z?Mo$V%fmk^GQ
zeGYb9u($5ONnxdg@w|VvTio7m0TXMvIb;4XhwY15)_xNo)%tDFb5GH+BU}ggRHUxF
z3bqQ9%(LUe+~tq5PyvVTB!(aoOr&^dG>6H8J?_1$a3(1=HHp^zh?{CE=BG+%%Ck_}
zE=1YF$h#A(S6llS@!_bJR=JL8vi0?#N?ltVnea`OFZco5poMM)_JNfc7ddGKP(ObA
zUrQc~4r%qb_5ZI0DE{g)49o`r>q#`?8yg!7IsnfA-(zfW5Rjk&yWDkW^mcwCK!8{|
zA=DO}HMf}F$3|9cRgI+n{yrG73oS)!WXdB`C($UcDMl^!;U3r#?}E6pQg<D(o*b~Q
zYI%4G)Cwv_bNn=AgsXuvwm%NgR{Y}56pqj4FAn9#!0zGo3^h3a2eGPZ(GRu`^iLcp
zzuE@uT@Qoj3378-mEl0+Dex5dcSS@ksd@Y<(EVCP{rGsu+p~MeIL+#<r=*JkA)yg2
zhR;{Dyfs=<sK#-`x}RUF$m>>VWI`b#Ne0ZXOl)-TJcCfutXw~TbyZiNxO6L9rPeWC
z*UrYIv@5dSxp*iqZYgdlD(XJG2x#kBjN&ig>DJWhcK2jGelN9AjiD^tx!d#m)6ly@
zT$F*oN55Oi^i0|L2ywzr69V0;g-u)A6>kUX>-Uaz+5V*bRL!c+(9y=*&s&N`IU93^
zLnDi>?$R)chHLc;0riLIA}yLxT{whKI|m^p-IqSa0ACvqJ39gWo?x4wy6Va0*opd8
zyWxsLh+S!9r2;=rNKb?&>}95vo=KF%KSlgft;5%%=jc>{kVN#AK<dRGwAEF7BN7$U
zk8<VJpAp!>U!EsgupN(wkXuQUv;5`_{@5`4G)s(_AE|fKSA+IXb+9emVD%BLsqb4}
zd*R$Ugb>%CT8J9Dx*mWR9Po6%+OZx^aZqpVk+Ea?bY%)|#kYI;34j0om7_}x3JL;u
zwubxblc$Bb_M!{J<>lq-vc0{%lFG_$C<j#nT1Xcm72rEXM@MH-RTWr%4lF9OE316*
z(>FSX-rt{|{_Dg*X`q;h2#*l6?<;(a0Z3C5s_DU5ubf<gO=EZW&||=c7$s%RXY%v8
zE@$V?(N&@$&d+;Pyh|9%%SJjxB+pTsa%rpTV2@fRRKb$NPO8Lr!%c|))W?S~5)xd`
zyhI5xb#!(6H_s9i6(j>7PVHny4e^2MOoXv=xmQB;-u$Ukj=X{IY)?$s^z#Brn?Hlf
zK|8>H2>*C6@%UJg7j5V_sc4oL7wJ}WOVelae}eRaf>h6m31#Hu-<|C5N2R2(WR;f3
z)c=snhtNO_NXamEws5mW#VFmN_!ERg&K4FM$|g=PsW=lmnmQzO4ThZqo?spV1I{K+
zer7azD>GA?<A>Y9T{M2}H-if9?sea%XP9}RgWIkljaR>J07L5v$eC+L8atlE9=>Ng
zr<<DWIbU8!+z4J@hAJyJd5Kb$mT$&Wva)TVqJ>?s&P;WsrpHi2g@?Cy)Xq4Y?4p|j
zhI40^$GqDl9nC(lvH}ZK&<D5Q&tEv--fBYz`bI|RmzS*{i)Cd`?^B-|kB`0F>wTV7
z)YJe@S>b7^l=(~y5V^US9QtpdyF2}B9?FdT@=Zp_A(G?!ZZ6zc%KG|=W*TRszu!PP
z)jd*4DKm$Mh9=}`XxfUaorVun1?XGqY~fe$z>RZAf{{E9(MXgECq9mS{&BI;NnFaW
zWp!eEC%jK^#1u1ZgG2AR#J_g{b6v2?!rHTEEBRiO^tnrZzLBJ)Ib*M(fjO(5OT#Kv
z$L-GV>y8A5FG^C&%a8%YrCww!r_<7)-gG6L^9y=<`d;PR+v)Qd8^5`#!GqcJYd5!?
zI=(^Yr99cCsr(ey533VIM2ct~e@wZ=xFzRfZJCCfAnxu)@(~+>!aQ_eWui=RP>^a(
ztIMK2+)mGY=il=3b0f|lZzHUyVBlc8P~kjjsbLH97dk(jJ{1KcIlgD3N*FNTsu~%J
ziVl=@aHz~kHh2nJsi>*TMUS6;S8zf3AH)g7BRGS&3t|!y8!Ib-*UoksW=@cJyaC)@
z_ki;6-!GmKfj|a8Q-=)5=1+kdkdO+j11eGwSQKaH$iSjjzp^%CAN7hx=&x09yyVdh
zSGO!rnQ8pp86SVGX%!%T*-_UB{kh61)}y1N+E9;oLPi#K?-y+&MNe;Jk~P%9YM{Y4
zyDRAD>I4Ec@&Lbhsq=N~e=m<DqM(4>Z6&SE*+y%?|K1)nq&3fzn;S1a@~eA`!b*|N
z!(XalJsV#T$0+%EqAT7H5C(Q1J{6+jxB%@^I{0anl$uvgZ-<h=xEH{rFHWA~6?)B{
zVq@V{>hJCBFpZ6C%2-0gOm&D3_n4T_6XLZVt8O}9_wH&x=t@fZS#GUKI__(!LGd@v
zN;MRAS=Y{b2l{($FtMbKjcIRv`v*2rP$0Y45lfEoo1$62CpA?53HkdQ$)X7I^Y?j+
zo=r_fN5l2fyCg>>NU7zgoE{@u&iYFxNQ}O}dhEtpata_g+*_+^`%%Iq&PTYux->_^
z&c2;Dwddx*ZeVbJVL?ydt(;iJ0n?49hVpc8Z({_y=*A5Q@#Q@J)QF53C~l08_i|u&
z{_n*>L7|P`G%qYXTQT8f40uZ`C_t>;$|poNHl8NIE6`6w)W^uk`0ue-am!;MSn^Da
zV-7|`v&wE1Pv6<J{8MpE(w-M;X6KUR6=m8yn4R${a(ibuz~0i5hmlbt8w<H0j+*q-
zU$27et=$37>?i(#wre>?#=h&`P3QQ{ye_l9B<&i|6>abLz7I76TdU&2<eMlFmpi@6
zL9gq0R>mmB$d~IB&-C8UyWDk7gEju$%pWi=^q|byiu<(1C=D6l7nFDE@ZX4#h}ElN
ziN6HxMLmOIcTe^5>+sc+1Xr*rl!~yVWm8j7(3=d8U=2-h9{;<RF4txD;2a2(>L$wh
zp6qg(kdU841Bh{b-fcC&%~V)Xv2XTK6zdgZ6Xu-R%%OWzqb=-Af|BA{ptHgQS~24*
z%MKG0s*~fL2KmEp8K0w{bvs7Az;QL<e8T61UUK+^g)QybzQor#5pEFS;if$@NxsI)
zL=*r1?xOUyVoVDNnz%uO02YMT>6ARRasG0{;&Z@c26*dp8;N`U+ho99x6b&*q=wK5
zfaQL*cEfN}(^UX0&QvR2JlY8qn+MWtCre99v(E44VRQSTi$^9+wFMofJ_+B;SVTx6
z6s|CsWqDv*0dnrgKV!?w+NsJ|n2<YD1Ji`{w-qMrq9Fh;Sjx=w6kWk~x=0W7l;}A%
zQ(LR4#s)n6T#h$8?c!_(DMwdn>zfIh`pki)CFqHfEkDv8DsyZU{r@`tQ?IJ>P**v<
zFSdlkCl9aW@|pZ^6UI4>-ko>2UNY4xQ>3Y?<m5>`jU5Lee)X-oE9{HOw6K`(6=!2*
z?WwiW=5!J{lhHGzz@~JmHZE@<3y>Hd`93l{Jh|}M{h$<;!qo7K81#)^gUkEabE@4v
z+g}@fy@6!&^VI%ZqWdpXQd7_VRX%U;K6FP3+2D<fo3CycUia8l^zC_3On{CGLWjbq
z^)=VOe$M+7`tM%TIIgyqhmgcZRfWy?^Xjt0L5ed}<C_Iq`@QiQ8L;M1-*7TZikkEW
z&TSR6<RLjWEVg;DoA}pzS?C4^(mo~mnj-=CPsj0ferf5az|zNFhWml+;y=QSAXMk{
z?;;?@v9hwV$-Ja%omoxLDUZ%ngWB8SqIdQdc5X^C6^N!&Zeaj3fszs^M1D<AZ(iJj
zZ~T~^PMHvMQd-WMntek+P*OeWxwSRunexjz)+Rkw2+~`Wme#yYz30&)gdG<4lNTLX
z>C0zKOvw7Tx-;+?9Hy>pp3xlZhuT*;OPs!*KWe~A4}K5Lgg@=cH(vhi9Gd$_EAGB}
zGTukhzQ)0_5y(pF_Ujks6Xp6?6i)2G@?8DX;MtnPmg!j^rwM$w(8_9V!zs{p*@NQg
zXRh@4*q84&rX5@Of(ULpgmH0Q`$ny#U1m{t&_C~-uD7==!`|&Oudy)?#mhue&Q7ic
z+&j!W&bPJfjf?bMpJK2+D8G0yo`^Thh=#<(JhXxH_<nvKW_%>>7U}0Ik47FN=VE8S
z?zXd=|LvQ~LuU<l#hXNi_NE`d+WcGo%=wGU<imxx<-7A_8IqHK`qJUSVCdJBD(d+K
zp@C7GxKXxL5{?bPqk)cW&fPGlyz**zn2Vo5bb5A{0n#?KO4f%KF*pv(2*qFjodO7b
z>W@Fu+y6^!5`Dbc-1i#@!4dZYY{LHlmf#;-LEug`;JOp2LDvCv?-nFTq>c|1p_&#-
zxqX0bO{S)u3qk6n{B^M94Hruqx=Jn@S*C=!mzF;K5b$<P5g#{qvV4<|uNl6DSiVDL
zH|y#qa$vV{AjE(vbD~qBHAGG}xcrpc%AF#fpZ}>uaIQc(-$mZFSU@l#(ChMY+Ig7y
zbC@e#Qnk;YPtMJbsxN$en5KVwTl&ww#KgL86QNZJ(%`o(&S8sT=k!!jXFXpHl@?6w
z$+cUgrP<c>Y9pm6FVBz5%GD$!lXT>$|7)7%!o<&IZzrW3Ll+?}<SCsMwpLbks+PPj
zsf8(J!zh8OC!V3M6qxt|S@4D+vKPUC<gK`(k!E3DnRt>Z0}>*kwfYEJ<9<r=^R1b3
zk?ayz=wQLro97%Pi9M2Ar<lx0bEXM;CYoPGoLYl^=^A+&OCriidjn8wb*t1rTysug
z6EE?RzPA_F=T4`$cP<pb!3A|=(tIr|B8$H~cE}vajW9z<IQjj|+ZK4D43c;fOtQ9$
zG@lsDwKRWh<fQdsW9NyhE4ZfB<DkHFSY{qb>9oH#n%V*`@g+_ud_n6Ae6V`awN@$2
z4D3A?Kd1>ZRlf+-|M&rtJ|x7zco8Is`HTzgE`fvz5WRCaSFOHlp-m6UcO=A8Q(A;P
zc<+%MRK;Evy@Zp%*f{^TK3sM3Mavhg<wpzy!!ayuy(mH9_h0$S_73t@q6_$!5kGLu
zq^gngMdN=T*^FXIjlp?LzYuB2U}-@?h>fG2u~mcH&hXh;cM=h)Vygi4Md^Hv7I{-f
zJcOTrZ-45KkR+v5o9)yZ8sz{bTBofpR)UsfN}5Yso7VBCy85VgT!Z<TB%0IPwd*g0
zkaaFK?Ds@1zIcP*ZwvgIKa;m_w$#lkb>)82-Mk=@8-(a77Be76{8}D)xn&&d@o>r$
zIqmo$%f%h^z_u`nM-J=Ct<us=D2`yMNyLS8i%o2Ihv00CnC`?#vlNM(EuU<xN6=am
znz-Sa_(a{&Mi*@9nJ_<G1xk4K`_FW2gcUjU23#?<|8qDa8|K^YZN;w))Xyo$GO6Ct
z3DOk=I_HNe`g{Ld0o&AkMIDyDzR)!d#yGHI!iKnjvl*S$imI58uyo8E_~udF4NAD1
zFLn6y?$CbYqU)sCBR_4oaG^sf%)E4vS777p21LL);wxQh4Q#t)daiiC3s8_-D~Ca=
z6apqM+>K&LlqXJm;$}YL0pCjEQjDKmaDK(}a}$s(*Z(D&IOx)%x-#x$>RtF(Ois^H
zGZ|%H5n^_whsk*iE4z5z(0tEfFAXfs!i}f_z1WW3n@h1*f>aOquKSpO{JU-{-Q0`h
zY4$_|+*`j*-WE$_+@l<np!4(d1Ao)R5P7il><<Wno}TH0uDsefOQ(zk|3~oKV-M5o
zcT$E?F7FW$rKO{k<gybI;G_cVn}LTjQR&WI4l{b&#y*8lQPYZPeFw-cwj%lq4#viz
zYcz`mXR>GqR3u6&nAq)Xw&v#DBiYf-!SqxSymk+>q9IMpvQUH)$LOFB8y}h3R35O>
z*n2nwf+hmRO*GtgbD3MuxH|IikL4a@a(gw|seg1wlqv;h-s}h_rh8syNo<ncsYG-g
z)LY>E(|pV>xI9xFW!0}k7xB~*w=8RDnB1Hd^?&E1?~M8ql{e}W-ceHL`R&#ezFq3@
z8Vz}7x3k=(EgvzEh_C-$d3flXe4R@=I)@<@4c9=4Xe0}I-Dz*qJBmu!<veHXITU#5
zB|3n?T6>G&RLZ7Hh3_XhZ9cXsL&vE;pU_ck1J`$ti$ag+{%;Dsr_(qj19ce|JEaIk
z0~dP6%#<728b87*a(Ny;OFxs3X~SaI4~2Il_Bu|kO?)S@K?o4mFQ-VA@h0)dr*w~0
zl58S9Z)3M+g?(En$PdQGaraM(byt#RW`LAoq%TJf8ZS$nsqwiaQ`rQ*wJjv1c6U)H
zZ;+%Q=wD3|&@n_yE-n2AA`E2IrPa~$+@%Oun4(o?G-j53NC;m!<ZYuz#*!V4ep1iQ
z4cx!Jx(XI&6q5^3Nz^2r-_-0Ud55z1AogZei#8)mfF26P4r#jDyZh)R&Y1wIEQ^G{
z{9C%9mG~q`sFOZGo#3$eVBAH$q;;yY5zKzYYurKA;_>#M&)~2*qE{RGk$C9S13r}(
z!r>g6=Ol@Nf{eCI7p)#UK2U7Ss;NQM^T$e1G4Vi%5cJXOw!Ez^Dop?ZzZeTI5$JLC
zx65=xH(xrUTg4&R+1EIHxx}wwD*#jBtX5IiV0vah@wz%&3~WuJF-7HV7{i641I^pg
zhhxuZ#-ZB`;JJ48`GUmL%`GbWhl#fiHXOOaUq+1N)+K33XP>Ai`z0o4D+Gte66b@9
z$je>Rm!wp3x{~qMR$>TwXX&d6Hd)@48%o1+b`62($QpaRQ@P3q8L~lk8y;?EwPhIW
z?5#^vlF{~`6xY@|syj@<Jn_9yd=KZN(^S|1LshT!=L#k#D5%0M1E1-@Pxoz+e~Rc3
zFl?!)FqQrL7ja+gpK8!W<vcwYP9GPyG9|G7hR5!=u#Sd+kIYmNnB(FL8c!)X*=tu}
z4mL%)DKNBoa@Ht?3n_>3Eu^|GegD2+tN@Ljk^=n1yi&c5SEWB-TVfS6?~sGMf*!J7
zI$RWsGc>b=7HZvQuXg7roEylj^n`q{WtNnX#lJV;-;*<|Z4gX4nLVK^88}`B?<pN*
zrU~yScggU24korKtnm~dOWM3f=AEFHHo@V+Y!l_4y3|z9Vc4pA=F+P<vA+#4!r)D0
z2W$I#o_J`M3HE$#VciS?8R`oyINS*S^Q{~l5IR`Cb4TG(krU2%855w&FIE|Ud`xDU
zd(GILSRX0=B>S%g+>wDz_wEW0&FZXp`d(Q<D(AVHM&9CK=E8k>rK%lLiEl&di$Sh=
zO>tJNsgfc9uaig+4IbA1>?bWBmd#8BeKBtwzX^Sv1evLW6{@q-$U^a3Ui&v)-q(2Z
zZv8EF9r!Von$5NJn&-{0-=C(J>At>_)BLL8?7sGmZcgvxJWq8&uH@cQ(I2jNKdwnK
zEHBn0G^`S$4&6(6Q_IK*ypLIzA(hto)6?k0I;9sGv%A3?u_|nmDhzSMlSa}Y^2#L8
z_yO0su)XT=Nv>Qk3X_fSXAFF>qr4DhZ0ypL4sQg;E5?O8e({QET_Xpwh&p>#$W6u@
zbpB9W6^M$aNxFf?m~`IPDB<Mp-dZldTh045JtxZXKj@?4{<IBtpXn1tWunLhy0d88
z*E=6kPJ3Jk2TP?h4r0n_$_7st=tSBIoxBXsv}eYPFVKqq5KLa}tE3Kxf34W%9usF{
z$s=)AQK9km7i(^IlfrUQF1*b2?A|s0<Qsb~ka_FSJr;250ue4CnVU*QnvqHcY~%_%
zsS#ZS<@Alyw`0Ae`yG=%k*3>h>O#r8emDRTujrE>`?#u6kAZ=k2QyLew7&Kx6*<0=
z8u-r`tD5xRycF3WO?OGc|1x5#vb`F3F?KVRaJu9QzVrP&62dLD1GUICPv+6OjxyaG
zouqbh%2M?xsZQ~#%y5}>W!s0Xju$V$_R+^F41qV{M&(qh>Pu+qXSjvZ6Zc%!-4zYD
zJO_Q(YBnrPd{*Pa!tU(N1L6f>)^RmP|JYy}q8}c`5LsHFNp{}8gL!@d7}be*DJg^Z
z#*SVmL7wgBB_$A`F6#B96p>rHQ`P+A_Sbq}4(f#_>z$&|BwW^OkazU0-*OaLsY|l*
z1sO!fQ6myQ7}74Muhl&#@>nZkI?tS}muKly<JbL#Y{x2f_>8<xLu6(`b)yzQPGd+|
zYZHqxhwWTcWPnyr!@y1suA#9~#1!0P36qf2Eftgiv)e3x;^NG-o1iqoV$V&!GCN#~
z6VTe+`&IdPWuuMtC6*q-s=JnRjNb4=AoEL|Xq;w2JyH8rH~Y65&FYb+OVi(Ni+^Tk
zHl0xhm}8pW#J(93y~G8kh57kf*3e(EXb#YLRu1f#a-a3p5{R*Z(et+X_fh3+>$URv
zwhnzwyBog@L7ZrJa(fz%F~z0QkXP8;eCcTZN5L>%Ozf+nZXKP7MIt&)^u@Y{k)z~P
zR-m&S)8dYW13ysvje@tY`O4Mfzc=+t|2TNll;|XAxfvCJ@{yedSwD|uLnj+6`h$fO
zoKy~q-j!?jz;1D2-TF|wks;=5-e={H%`B5cuuhZ5{F(pW$6*EU-~?wIxVCMZ4yAIH
z=!)gBYutF^;iPVl5vvJ*&<jllVeK&Y`tJ#`aE)@CQf`Mf^ynt^0myfl7dEBvCAhI%
zl_94adt?OQoHwD&YSC(Uxel`!r<|eD{9+Ij14P-WwDRio)o$is`uVQQ1#d=#V96y{
zoc4>%-igf2S|1|BXI*crxDqL-;+vKq721dNTj&(gv5&S^K9ehj=2yO`rw9#8cuayu
z4LO`w7nrOUlolarp@Zzq8ec(Y=~-zH`VHv#?iT?<i13fZ!CI?mu{H+^@iPnrTU&+5
z6EnDQnvOxf-tExS6;MFTyqu+GMTzkFcT`p$@0pwyZ2*S^ih>RB^cxTLWNe|!W?WNl
z4SP#_giRLW2jD#}Qa$jDlBDZnBYM$Ai!N&rDh={4V9)%ST#L7_DdN!4YPfW5XiKiY
zx!t^<4=&EqYV}z4Z_DnGJr>MJfZL`;yvzduUy^R1D`lt$pWJD{2)0-iC#}z^D5F!C
zm$jthoPaBk*No-eS8{hnJ^0L>A~=iW=slGp%f|<N{yM-b!A2Cl*0}T8-iSj_Utd2k
z@F|SM1j@w)KWV!je@N3u2A}!qE(|8Tcg-Lhn%_Xe&krQlTiyku+J$?|mMCbQL0L-;
zffC~q)F4>&kNNTx;%p?OYz>#Aw}?WvN&FGIJEOf%_NN&Sn|5cykVSYfBCZTnH^9hi
z%v#ZWd7>eeR@^S4n!^B15o$uij$BUq01%fE)GW&3Zi&mELC$0ngX~5&4e4(%bykug
z0Y>xtYh{V{OMH%24KKD)^?=>XHre96>Vq3}fp?{Y-wg~5w}0xkJ$i6fdAwjkj+msE
zXMnap`5P&|q@%!$Coy^LP<i;*u%lr||6xbaYmXLSaiXV@w@*E-{FTAXR5a@)J-`j&
zg%>`uNqb{Q7-``;h6p={)<*Qb+_>{mQB$htHa6Zn+Ex=)3^3-&?#bm}(R$w}*_K7v
zZ!QUbMDaDqT9aAQ^na#HjW?X!LrN!YFe)v`0L8@onoQ7(U4Z>Z=S2M3;_|i>n>7=3
z=Tl+#_rRw|VE#Wrlem%OB7G<hyekG_XJ`K<Lf^7~M=vWxvNGBIR-VUaEd{^4%qe>H
zzUm*2en}o!1bOJ_>R-3pvxKt|(~acp6tTjxsIYz%sW&Bp{Ofy~=ltYLSz_8_W9Q8x
zLa6QhgX<+jQgLwfCNEjpPoHuI*3eLIp4ui&JnlW-);OKBmKN32cJI~C!N-N7sBe&D
zWrsd;)CN8u0Qr{5U=;=>xB2S}LJ1+3^s!$dp;9tz`<b2v5_bUces|ajQWd|rh6k|1
z#DkYt)#LkQrKPXjC=mY4(|0JAWF9^+@E_BLnG>WHP*S=S)>^-CaRGKzrjEXEO54{R
z0}bt|nQ>-pB<kW(z(~y*AX&e4<6fz#EF0Oy1Rw#Z0hn_URwI#BuBf|GdHlIPqWtb7
zvNP8&!!#N<FpqHp315yZdv1PNeL3#)kA>);W;IfjE@TGzXmC=4YthE66cm#Av6_=f
z<Fh)oE|GwF9-GG|D?)hoP^3FtZWz~RzICS^`<yu`SVvifm*AijV6kXG-B}1SvG=C(
zK};qH>g?(1K|w*0e@?+SPFa$MX_;|~fJ#Z!&h<~K2zm%`4B8f7o1ed4j;7}>{Te|B
zO?ubewdvw=&@p-<ub#ky@BQb~U2Qo7ba=Q9CLGu18xj|1z}VDP7B63HRsYx*Ld>1I
zp{}Y5^@ueumB|OjOz{P_2a+F71I)#w4pzaR$uIwES<B-*A4C~gbQ<|JJ2qx(-(!)`
zWqFqVV+~gWGrkwa)ZA8q8T<k+QMQEHgxJo*R2Q{zF_OT0ij74NzZB2t04;yj7ub;3
zOa%wMKi>8@=Vs<kF*h2yBp6ekUh}0j54`(Zs70{5yV;32+omD$oDw^~TK(>0t0*{x
zG-0K3pr1ku4SlK<Qtq=14XH3yK*zS*FTy?$tAKiPc8ET<tfp@v1KK1wigw0&|1E<r
z`gyMWvgS)kFg<93!;8q*p)N9e<7uxc-8J+E_Dp#rF^h6(`G}dKTl)%d4UjMjg?vQS
zj$fap<=Xx!ik|t52a!7eh~MPS&&ZzY<bP<$nCbSjv$K;(&#1=xVn;pTJo%!OC<1`?
zfzCW$Lr(ORDH>yp&r!*Ybe)EsNLHtv`RhE#6M9BFa3+)?O(DcnYDcJ-#32{>64keb
zJa6wG1$kfd%*D2uTm5+0B|Ki+pC)`2Xs%rC9s6r6OsWs)yiVFc8z}He<haGDRzl;m
zL|B42PhB*Fo!#WJ>btjje^jUD7?ppzLr)j%K`6;sQRxUrms+vw*k>LSxPtvl0zB+6
z-lWK#sD@C7MfgGK&%^q9xJraii@8m}a3|Z=NK5;**en)>Km~TYVB5cE-2z%@56UB%
zQ;`Mp3?r{+bz@fi)rx8Vd~woRCg9K8Hcb9o{{LD4%b!|!j}1G3a&N4hhRZEg%0PnP
z{$)xA7A<@tjdn$)tB}^+-C5tTbas~7_~^3~C;_|n%tViPSfWi}5{g=uao|YbgsaFc
zqY6P_UEJNJ1T|v?28tmtuax?4GOG7$G{n1!!6BxkTI@vb56mgXTh1&<l6h;|miMJq
zH$}+uAEp~3z$SJc=i~?SW+*{&>l_XQNRe)jn-8lDo!Z=i6i_|p-JvTryRvM+7ZDx>
zva{E;WL=VS&>P1C3_@vt1ettAw>DMWp$nDEj{&`h)ygZa8y91<A<BAbJJwdd9!Yt0
z$PdoFo10x`26)(+iHxs`=JA3|OCiNgWCQVuQSTuU5>I4;O9%}V4NETLrcaf(O}_AD
z_g9!}__q8Ml%yX`v6w4=j@gr2h0+UpekFnY%r1(C<Fd1XqB@uMCArBnkd#!5d6V{H
zk=jXUJI0jnH=lTj?E0^>3d4r(Cywi1i!v1oZI6Ee3Q0NW9qpiHRTN{+ktKrHK7)~4
zV#Wk6>m=BzZifg#wSg0{sM?f`+Xt~c_NLp1vMOU{W!P@2HNduPdMCa0+ME8qYCS`(
z?v1){Q47t_lQQ7h=O^T3XV|LvYWp&epI_>=G+bt}WtQEll<vdt>8*MNAuKg^sNHv>
zV2dleKZ|I#Z&>Vr4vGCblrTo;EV|%kuE4C6kT_gE<Sy7EUY~`4fB+mH&eGC7APe=I
z@AYv-MFoftIsm!Ewcfse3%D#w>*_^ZTveb9z35P5{8AVOlcpD0bSa~C6p8mg$}PmX
zN^$kEbhE(|ow83{lA3r$I6qwC;*yh-4a8*X=pfOorSWTAc<UOvPalIr{H_54yq5sQ
z`|Ngtdernp=p6C!Hv6hWO43^>Je(<5I&JRb`ysI?EqL#yq)FX74D1WMco6jGD~O<k
zbog$4&M!SHxuBSlp7Uh0FrDb%h6F}eu1TN>WN##~xBVm4Tk)W%&p&#=&F$5+%eK(C
zHCt+zBn<XazYu9zyyYMyRG{i}*(Sz|$~3o7W?t?fkck}WM0eM*F9RJOVGz-BDouT3
zpot>SM$C)8{1aJT5u<h?0WScP4;d{{Q0Z4zIda6qBDmdb!_cwM>|_*Sc^KG%YO<ON
ziL!K<;$dqF0`CZ=1RC4dK3Sv<(<g}joSVCDS^dX>ALV!Knr&}qcX#aB(PBSWHNOC5
z1)nW4(oS0AD|tUejrd=LaAAJdY8+~{(FG->hzXG?0E<gRXaATx75yRFt(u1xnn!)&
zS)s2DOew&B?mS@B4C9rY1^0DKua*`9<(huEcA^M6<nv+)#qi{wGq)kV(_IP1IV>*j
z@>n`3endu8V=RY~a$$4x!oJMQJZ9+3HGd`gfGi1svQMtW=h3<B|MBs>*9))^c1`OS
z`QkYL&xP7KvB<9a*hmozvjO*FphNfP@z%@Q@2ZcieGr?Te$xje9%f}(7Cijuz?WMj
zI4%>DcPg6dtQ%heed>k`Y2jSY=6A$H?-T(fDnG{D6iG-(2yC1>3trCL=RQpSQ6<WZ
zQt*#4AlZec-2skk#T`})Z|Zc89N5qQHM6-Y_k9>E*fLbQ4)!w|gveNfT#^<w!c5mu
zEyOjKrvJpO+5i6793360i(RFRY&UNuDW@?6UVLXaNQ<5;Cqxb>T7;lQglvvtbLdd*
zA`sf3mKtJHDoMf{l@$JC;2v3ELo_Cma-8IN@Z!6r-~$GYMMY<4XO^x|LQj>{MfTjq
z&=7gv?nG!V(Hy<EL%7^+2{(2`aU?xwQZ11RDhk?w^XBHJ{V%gsan-I|Io57XgXlXS
zBY*zmM(JCu%}sbqvMMbd6Vq!hrDIh)>e0tMhAHXbHC*$n2%Pft?{0O%V$2}|`uaW=
zZZchOp~_+Mk?+gch}sG?2lFm4wnkpqiPIonf%k}@-kb)%{c6+URgA(II(D{tY7e&<
zrF)ZGFcU&soQfibXvn}fON=v7{%%d3LR?6}f<P@Sn!eYPhELGm<l_up{98TQ$58$>
zrFZW%y*@Q}6Upv^yVDj3acTz@CBRyfl96SoE9M)dVdCp2cPHi;J2QT&&d+D!5P*7H
zmONWFUtubA|GnO0KrieLPxmdpC)zD5gHwLUTGQZ+FZ9)+lgg3R#4|c#H`1Qo+On+m
zhZ3d>WuA-Brd37=chg$gH!q_{04;d?v|>{7F>^`^tT9}%i#8^J!PKchh8Y^%wHE&3
zcN0O22g75Wr3nN>pO}>uv19)gn_-b%SiL*zxR_uZJ1U=bp!U6)+8n?75nk0^x2Q<7
zR#EOOn$-Yy)0ccxBNkKS!#6B+HKbq1w*YygyvDQom%tt}RquP8gD3q;M)W>_a;qSM
zM2QIZX^~OIdx;9aqdPZE*M^1iNHBCu);qGY5)nX#uycXA%oaT?xnK=+BOkUHFVA0H
zIrDE-41VEyTUsKp{05$q;+22-IJcoU%B_sjIXE~7GRzOhz6gS_Z;AUq-{*{6Re>0d
z*Dtx<J4kJlvwc4W_s1MDq#I*tG3LPT!hei;SJRc3nm32dD4t4Z#9P{>+;3enth=yb
zN5o4*d90)-<LBLLdvfdGpIVmn)UF)*TK~QJ)Y#`d@)<qoouij8Yh^R??kY+#<P(%C
zp(g%ByO}mOLH#o`Fd1LB=7>EH3&Egat@&?=Aj&i<D%7x`SsAx42(0jjWVP<6$La$L
zQ!hjk)35!+&&aR7&Y(xQzWYwPj=M9TapF^K^`A)}@V_6T$^v9${42h}Y~zc2Xh$w#
z1}p~i<-&N_LYR~}trd~F@fqk5xrKRP>i>PHFX5*$xx4UfNCF+`dYm5|#?eDqy4K-2
zFW(oP5$Dvx1``+Mgdn$)E;_CcXaD74P(MlDa@4Pp$~5_of<zB~&VE=Zy;Qb7O+X)b
zKsj1ZyAKWy<f;8Hlax93_x8MQx6_|)v!CYkTf9EI+Um0){Ho|$X~XEERZrBSV`Ij5
zkmF}0ValzjN#i6zK`Wi!-b__VHa&`q1LsdH{M50apR8K#w-#Gjq*-CoxeE<$Bncr%
z_O>^ksU!ZpMX4<wE8S}pU1g#^0~GkjEn1<{P4RjYM7+wXs{3_Xa3-9XQ8{S;z(6|Z
zz^}^2hx1+v7_WY0jDUAaQcp~3uet3zMcPie*Oj{ka?t26(H7F(JD8q4WpCAFN5BJ@
zwkn_^2PI6ye36Y45`ux4>$V1(JY|NZtqv(0DVpI0y_cK^e<<m?bH!t4qeG#BV}zFH
zkzp*Bqp_Xq2=#0pwoI@$wvhE4j~zXR(RWspczVn(d9{w%1B2+b7F+e6)34qyD`oFg
zidWq=MgQTJ4Gqca$?c9^n~a7>I#ld_Ds<V&u%Qz7^IH7j@N`%GWbl@-B`Y78W%AbE
z)G0U}gwrtI-)ZP)Y-|Z+ZI9fVBv^`e#%3z#m&B_~lap)d<1VbfAjKA+HO@;-?=PlT
zi{E_3$XH*60#aoZeB~tV%QIdFJ>b87wUX(_v*NE83GX%EJ4$UxQ*sfmj$K&TU@lzK
z5LhQAn`#rL0Bo<XYT#tNS=z;S(xBwBR3r=v2%A~7WVg2iD{IGKpdGRwiL17EB422)
zQ0yy)QQzWi!^r(*QUe~zxEA`bQ}N%i0O+5+FW4Yy<#d}6?2jI~e%_V8d6gX4|CW&$
zoUrCjplY{EN_4+<S>m=V=L&K2E+#mG&H@0&F2&0@EqYmc`k?nNHMz5fncrZ8IiITv
zr80%nlk@d^)bJo@E_2^D`f2Sx^P$MkXm7+!P1v1%`wR~EqlVRX=up3sD>kKa*hC)z
zK7l*P)Qc)$C&G*p_-9!#<JbQ+snyQ(Z$V<W|B2Z9-#J%#U4=HYER<cEV^`0u=NgTw
zSdIpZ^_3a$?lnJ-!pR=hBn8ln{_9y_>G&@Gz-!P6k^O=lL*m;nOc*S!FJ)6pQu&Gn
z>@@TVgK0*?BGk5Ha>>96Vq3Y9)ZbdDWgPdo)aRG0#1~zWDYE4liFoiU2>~>e`r~^0
z_t3GM3z3xY-Sx@DZ)nasbaYBJw8aTew6MP=a#j=NX$sw{$HJL*FVIOnAbMEX=0eFS
zYow5Z>m`_mQS6jvbOj~VUv_<S@e|dJ-!(yb)*JAjWyg%dREaj(Rcjw^Sr8EFHtiN9
zG0`+Rm4ZWbgDsh$uUwRrUrUtBP6-XLrG|ZoFVxZk<ad1kSM^C#y}Y8{U6N$kdv0`8
zj3^NvTE&PaPLeQuNFCx}=XmB03YxbG?nQe_3Yx)!vFe$ZpOHwWWcmM9_SFGVE?@r`
zNGK(s0xO+Mhq5BwolBR}B_JV)gwiaHGzbzaERBFjOQRs&D6n*jgmnC7z46}f``+&#
zKmWM*y7zfz<~eiDnKLuzd_JU?*kEw%K=GUYQt9D&U3ai&-RQ-dgq_Tsi9NSv=vA7B
z4{ZSh6?VWAgt_9xuWq;1%4_d~KzZJs^I&1AiTa0^u$pNJX*#?bV)vK<Ioa^=BeemR
zvL|UjEBI!H$l9JGHAjvC&8YJo11PxQ&pd}S!{Q1wFF6B8b!fiLbo5;{R-hmvEX;$0
zvpH+#<Xf4djPN>vvCOn4Z;3wh6qMx-CH~S9r5+TJsdaXzh7J=J+u9AYRo$F?04cxY
z{nUV1fvQzOFlO--2<e&a=<1_RQs<;wodom{5Yp)50xQG<+q?Romv?r$RNuPfLZTS4
zaV(htUgzZ0=SSxy##VU_#E0HN8f>8or5(As?lz8eDfhArDGd8PirNQyo~F-;p+<-_
z*0YuIqdH{G5(Q%dXNgPKii-=NlLS=IC<iLaY-JlIW9Q+K#T(CTWnlS@FI>73^Xrjj
zhg3UOIbTNJZ}6km=d3eg_3znARKfEb`0VY<2=V~y$z#)1J2U*X`T^rQ+^-hN0(OI%
zwvHVjQ^e*<rb7b$%>aK=fG3#J8mk;A<=YJxB<9;_rL)4i^DKX!lU&O`AeB)-7}C?t
zALX3SBxP@(U+{_;GCZ0oZ02%fqM5=%TAtd1l0ri2{N>TR(Eghc$~ZYw%-QkjcD6sv
z2PLe_3AE_k^O(Iv9zr@9=HBAx;Y<v&KJf?@&wnAvVQ!%j&H&Y7crB)3I{Tw3$JSe1
zaD2>_q)I772kUxKHdQ=-_HfeRg-aSZz!Ww?CN!5+9nxZl)Mr+Nu0FeEKZZ{rAa;zq
zKFVf#@<sRkkMqy^cbi%jSV+c}hOy;zbZ%wm&dNi~QGP|CMFNtQ@V!_$*orx~o&73-
z5I`N$HHrl{5<Ft=9D=vy-s_lMoKKq!Ys-*?F)nM}+%EGpx|iGrYmu-$#rASo15Bd&
z)|cMGQ7cvVm-TBjGtzUDS|!4r)h`luz9|=&@hDT0GlgqBuX=>KbDhv7O3ELNMkKJa
z(Z3YmyjfKx1HdPe)H^3oCAXp+zf9!SU;FHn*(@QjuxFPvqyCmgx4=!mag9r*nFQCy
zq-Gi1a7&7JYHGZ_i*s9b)`m=9A0ssOs08+MCmJ@`jrRqWpFe;0PzkpdTD+#h{W^h`
zk%X!ERaaNs(Bx#&6mfJX(OrBf|9(Uk{Gt#$({@W2NP4=sAfvRjGm&$dNI0r)i`kd`
zuH~$owT#qzylP`(Ga<WcS3_>pBC~~b78P>#zQxOgvy$$d`YNljRJ9Dzp-|>eVd!6k
z5Xv*@1p@lBmOJ$GAxrXW*9o^d?pMOe@;ip8mYppuU(L7Wk~TmH4H6Uu6!E|=({Y%K
za_O}!LC#a}mZ~FMoRmlW0s=ptt*fLiYLxa&P7R0y_|m%tW(Q5Vy^YNh!*y<JK5YFA
z*7aOkB6N6!2F-7>#K%_sICY44mD-;H$PWC_ZACm#K1f3RzLHI(t};s+rYeN@&;xo^
zPEHPA3oZ?SByK};vwk+?9PbPB9)!Ep?j#5U;+C_Zs-YnxLqkUWH!Aum#fBl#qH(^;
z9Vx`uZ?DkJ89E+SS8*3IoY=B~JVS)7ejV=wE95<axR*Ld!e@ml;e;f;SQVIOd@BOm
z#G54ZNcr7;?U{DkuCFo+TDNZtdtKDfOh=|UD*i|d#qF%yDxgfE$fvRFRQrBI`N5ZX
z98gk##jX=7LHHFyXz6v?GK*eny8ks4Haw&mIxj;82_}iE6^e)ixr>o&L!|xzH?);i
zubJ`A$1}V8bT8!QK~3+X`vLpM4?hc)BCdddd;|+8EOh=0T0N@fN=&|Z|7}R{=cZR!
zM*4(noC<pvDM+)2dz|FhltIv-pxJ9z1+@!91j%^?pc=X%C0sHzQSqc8cjGD!1%Rr&
z0+17XOG%UNf0;`zUC6o_ofKauD`(vNmU{vhM~0%vl-r}O+{5YSyf(m!g$Uyoe|;Ni
zFR$h-AOH`V%%!o6<mRJ7ztw|QrbP%}3Ufqpm?)D#`b5Jy4f(223ZR0<v29(cG<x&+
z<I2b2T<imI)ne->(EOOWhR9Ayrts2T0w5h$LI;2ckgEwDB0UUJ6jn8z{UVo`yh>V7
zVsnoM-Swp<^CE&JgXa+cX;^7v*y!je<?SaV?EFyytyw0;NNTkDQi-+obv&By-}4df
zc?__?54Hu=s#aG+k`;1fXkY5Z)*1`bcAJ?cA~t#KQR4hgm1V>w(_FsS>@~kqvF=V`
z!QswV0^;ldmcVNUE<iA;6jGc0nYphyK(AcV?>cb#2Ae5TbPEgdNjXho(}OpA#RCI~
zAvAvJaFcG`CIqLUKtdXH0gtif4!x~S{tXlaA1mc1By3!uT{G*u#X=CC6=wp9Eh5U!
zu5v*G)a#}SdLIuRHZ{Q<i+#WCuAju6GtYf^_ZXFoR~-YG6+K+<IaXT*O3`5okfxk=
zi54F~5L0FEXIPqM^H?rY%iuSFL%RV8-UT!&((j-P8^cl&N2GJvO*9D!Ydq1ahSzyp
zB%-;YVoe2#@`o!M(h97_xFN((;KrthF><W&q3*T-WN1t271o@nSHNA3Z(;}Ot8eEx
z%4e_2-An}GSe31%#3h$&LeF1|vNcQCPhNKzS0DP7!3dl73eV4wM?ln6PtdklkUYh{
z2<chpmA*d4+Eb0}ZG+b%lT%;`^via4YkS@BMC08%63fpo_8)B;?9v=g^vxZwKka?6
zr;B><1SNQWxxZgs<U8|*vn8j5fsicY)5UjsP+1Q=;EEi(YTfOvC>vPR1-tqcJSWR8
zaW@1eth;FTn%w-cPfFU=r+eFDO%4{XWoYTqWdSnGHmu+CEPtFYvnINTp|s8MbrDwO
z)_65|BDg1=ANrma<ylM}wtWeaX2a;e;h+^o{xr+siHy{hvpLkOJLtEWK41C@3Q0#3
z-n?PO)tG;kn!j0a>RpR;*~@pB8EnfzSN4m4dfVLFPwac7R5#PV+2}*T0*a9yw!{ji
zn*|2AR60x!3mld>(}_O25q^tl=>99lyuq;-{X}*TH_hXphF_H)#b=`Ii~bHb)i*R8
z7}qn2LO-^qjVXLxMSG1WGS)OY(v07(@fYV;^;n&>m%~(69H!F(KH?vdcn#awe57c3
z4oMXQXv}x^)kH)s`_OLRe4NtYw7N*Xag%4dYWmY_y<%b`(V{<KbkrZf23*wetTt*k
zX899PsK-uMNn^|=(Vya&>T#r>y_p;M_zv8gJRb>4JRBA)dkvuKS{CO@84%+`KY0|D
z?7tStghBv_gMa|2u(^eI%qi6-SY6rQp~TMJmc36634CbS6jX3YQZe(3CL3mJ_5SL!
zNH&6FD=k_qIm@6?=dg5$_aU9CFfUVsXaEPSIhI5T=`7$1P#e6>&c^Lq<}106k4B~k
zMtRjaLl;iM2-L#|QAfc~p488-Y%SBr=J!^K7f!GSfvs=d)@7{D0V-H7{=Jr19kKqk
zwH=#MXFYvJu&)C#k3b^QEYRkWID@@*370u9L^v6qsEof-{p@_A@8?9{yW;`F53{Rz
zmjFXLVDDNRbH5tmeStJb*+?))UWQ$GJAB9IY#vZiV*pW%iyPS$KTj=V&S<%hO?kIy
z2PEh)*#J8Ch1SNPRWa&vFiI-CG|47zo(n-$HTxbFI2Wu<{Qhk<r6=O?!<D$5FSb5I
zrsIR}UcC%BOzf+2icrmkrP*1{Zj-}*niir1Qs#fzvm>as)8e%(xZV~#kVoTzl1Wlh
z&owq9&#Z~)95l!NlIn@X=3;F9?A!gIwpw?|wCfLwi)|c)&)kHM7-3QOV`yKE!pQLf
z<-l#ztALR0Em4RnxjUg8$Gfe26r^7{(^t~4x-5(p0aX{{t*j+Zat%)VXReBFqc|_C
zr(10EkPQjf`ib(c%E+W@`*nEFk%AEn4cR#5MPc`Ue#E4Tj^`xjWYs+Kzy5Pjso!pU
zVkoCKciR(eu?uSCp_Nxd?v&)&iIx|zUr||;_7;tZWOD&vD;kCm1;_;Me+|997|Wpd
zJ`lHY%_3qa-B&2XQ<8y6;raP)q9UWeX4IM}<=ee^b#?Kuw>ofI$^^Wq_-AgkFE`rn
z?d?Ahf5Q2}QN~B4P_vxkvp!UySwjz3tt?56A@9TfP|I@8RE!_0mJnbepD&t%ve9eB
z=Qp;!x+^-FnQi=A)7c-Vo5+zB?Er+DckdAR-pdc8maH1orEii%5m1%6vS5KK+jo<u
zJNcaC#SwrsQl~W1=OYy@1Ej+krzb}5x)!*y|Jn_(a&6~9I_IWQ=JRVw#)<nnlyA*Z
zx#2@$1XRxZ6Nh_wsM2-F$j+HuIehRIBz_e_PVQVcFbPFCwP(;#?Oge!=Ib9qJs9B}
zy2@uD$j93kP#`Mvg7Y;=R4S#L2Rk5Af2d))wL5a>fQ0tTl>`>DXjK)CzM$K4IAWhx
z^M*{cWMN0{mTE*gUa!0FENYCxbIM{|#>vRD-fK><JU9w(IfU(`7OeU+P#WuuSNJ)d
zM)30uNp}j}N}rW)0T8s&iD;MR&rA#gH`S@!T_929#jRfj=O^v?)#F!o?<E~YQ_Gar
znrGAy(wG^<Ss`eEY2hCF9uIAwY>g=eYRSO_WYSt+%}00kz5zEVk?eAae9nZu$Y`_-
z3+$QSRV*B$R2n|(#K2h|eWZWj?AzR(P={}<5!_*&;CB17JE<VfQ;m^o_#lR3*7V$e
z$}y)gIUAXMFNV{nq=_12g0*cb4g>~Rp^8i*Su%%j?1qg-=mfUTlR)+NDg^i_(D}tO
zYSAs4s~c_EGG}L(=5cZ8U^mfnM|l*on2=&PUUpoy24wIR=Uh>`&RUMl?T;PyJIyKY
zkk4)G2b?7XE|GYq2QwF_&Za}jAIq(eM-RKL7$2j=bIuF}C|z;8JN9mu`m;9vl=NQH
zTJ^p4GA=GS8(lrE$#k--&CYy{4uPj_ENN1N@9r&bCI54N)m_xIVRSfc7SzHYP9He`
zsIirvV#|-K<)ei$Cl+--_(>uwhjv5wMeA$@<|>e>YjOZ6h|XAs$$*ePq3fkhgjfKa
zY)Gw78WPMXs7uH7i!;HEjkSNfg!Q!&sooP*f6eS=K(0BrWBL@opMxAL^ElqB-c<Sh
z@F3}&Lh-%9H%+~k9N@x&FjmqT6x22P+Sm-jTIj^cEH5(bCUzRDZy$4>zfl$536*7Q
zaA|H%yZ7dejhnlT(av5Bjlm#E3FRDZWowAZ`{dD=3>lsfLhgI(IG+0xB_ZdfpXTvD
z8{&<}VY9&E0Kg{TW<v&fzb`K4mV0xYPEWxuGtb6e`1Nk%GBr3`{jz*-@k-dZ1AH{Y
z5Sy1n{6o|zjT%vE;1DTfeH2~l2Y6O3th&GEWMm8SJaULxC$%?7ov9@P6_kJ|#G9{J
z)9`^%B0?5oL<UJU&JPbKxj_%<aPkZ1_F*G6RFD+tD(lx&2q5E3kY_9|qRAr_{2usX
zYtc_L{W7IOj!<Y`PNG?int7H%h&SSC2X=^1NKStKAl!aw!{KosjR<sf^z-W|At9SZ
z7DcW`J@X)>0p9}%xFJ@j%6gYfNh{WcWH8D@=uUm)bGWa#+)7k#)?V;>r*P|BXOZS2
z`E;v~^}4DvP!ud1$9I))U|y9hehR`FMy9UR?$vHGX#;y~oS^2Y-lMc23vfB;HGb>C
zxg!U|g80?bM@l8$%@)g`ru1hpPpW}ujm<Mk#f}|KFI^B-PeV)j2MyGjFIH={p_Sit
zHD2|qZcHK#@qNCq9Uk=!e1GF`zg!`-QEk)!ii$H!z@HEocwSO}6Hv>|keEItqNG&Z
zf6Ge-G3%Bn-oX|1Bqq#D6XMM3MdOB`pAivYs~GdQt5V8E<#NCTG`=d-ow^hmhqrH!
zl7Z#!7rhKjpcZ~5^<H^DIYd*+%iS+6FSBDrx9+=5qt{U2J@$|r??K?_lBnp)A9a@@
zQ}|hrX`_WFEa6w7T@-Cc2OkFv!q6HY9u>w65jLwB=v95vTy9JHY$L{3*i9bJ4?xph
zNTO2c0c*Af$v#&tvEN~Uum-a^xFT$Q{9dsg{pr`(d>D({#ir@u<&Ms$7iTHs()exI
z!v;!B+M1Pc)`qp1H%}T#g=XF6Ks{NtxuCL&%lGei0kp{E<niz6A2eP%kY~A96r!5Z
ztrTZl#19~0Uf<AYHoyT?)F9>s7r-toWNMC`=w6BxoLe^yvF~E#4GcEYRw9XtzZ#<I
z;ktSmdA{)GVf#|4d8&kMUD@xi#jh$~8puH(FEuyAxd_)J6Cz~3KM&our<aS)u9;ZC
z%qz$UbC}^e^IVC!g9cZC#QxH;yn<R?X?|B$w$yZ;apS}%uF}u@w|XbwzvTA5q-%nr
z3N);^CU$s9r?j_?S(CQ*-jrBNtn@s$;)}=yEm2iiR<VoLLYAq=GiHl|fl3Cb=y4zc
z-hk#n?S!^2xF#)gG=4u*bdn$?GdxB4S1y3m;9Y@O#)M2$tAeDWaj%O{m!9%6?9=vP
z4M#A&nLe59BH96nm<G)r3eIr$>)P}r()4CusMEkMFs=elNi{I3DEfSlb*$dv5iNP~
z+YkF49kaVk2XtQfL!_Qjclp_1ZMY#Bv$NV^$!BhR`FcsQsYt$E|9~zk6|ok&%y7E6
za=#J|+xTw{S;Q``s3w402d+s;XELC!Tp<i9N=!kJAwP$Dk&}aqhh7%)dZ|CYrL?dz
zxqpinuY|j~z&<#tNUW{IC1@v8krTdD&C<#SiJo1-*S}bqgeNLq+nFBzAyeR@=i3h8
z%=@{T#{fHhU$w!+iT#1&^X;2dYEGEXaB*c@beo@__9CJ|%kO_=2~MD$Cn4liPI}B?
z4DC6-pC4;ja}$RUi#YLR*s~4Sc1^m}OLnC~J(6$2AIlKp*rLYo80td}?uB?IB2@J`
zzBb5bA)XBkri&z<&2AKX;F3fg%&29Frc}e-dKqev+2G2KH`ca!;9L6lr^-K*g@int
zdGFs+ZP8RXy%o(0>mff}&;D_irz=qs@(m|8>gN|lfX|`fn9B_dBq!muumi+FE{H?Y
zKS%%8?t(nLd2w#6)7NisUAG&|Rx%;^4%0K=rhx`@<<QSjL##6M5bsH&;6gY@0004r
zw&J-0R%PDi*pFly9OP6vfczqRwto#~jY68;$u+IQcly}Z_lEKp#FwU-wzl>LrNO`~
zXTtuouTy2769YF#wY6Ef)#%JMle7!V{F;7x&WIKnVAHa6THm*=EGBSj4CC)5&SYmL
zR^>HdDPz<a+aIcXOT};?p)3QdBpn9lJ~1yEyA$!)C8-w|lk}k<7#=8LzxPcN4k5~^
zNin6dZqxki9Vqpt!(p6BPQjdk`bAIv$S-g|-RF1h<(SFq=<ZbZL_{w(-MyknXT~qb
zY>9}-@&sh|$cd8gA=AKNcRJDxcG(6bKFmbpr1~LB{q822Ap9;o&YJuc%^9MoN7%m+
zi`*`D^qrUhl*XEeR>p0#tQhX$<6l^^vXUS}MU`-Eef#1()5w`N>j7AYhZ`;owAeHx
z@U@|nM`i$P*o6XzZ?ctxd@42_c{yQryzy>cI=w8+0UY0*5e`x02vZ5p0n(im>$OHy
z(T(TNS2o)AH(kQspgT+x>4;Du#J7}Syi`>M@3<;=*$R~#kvL21Z@3Ca<<F>w2e8^W
zv}K4Sa};-itynNsr9s7dt2JW%Td?+%eAZmSwy@yS(c~5BjRQH5b$;3N@G;o(dmBkx
z%*)mvm;`f>OA>e89Eg+jjZr#+n?chlH@$ha9zb>kt15}<!!QE3Rk8&27*o|xW}>vz
zzB&XlwE73!S%_fdlYI_CpieLH*pIH8e##SzBxhfEq5_Ctts03!ys@(%sNH+Q#|4vD
zlBu|;=cN5|`tF^TtGRn^TDZkGP2P=Bw447tU~pI0i0!oq@V~q<leu3jRxv!5Q$dOq
z9Nb3Qvk2vTapU86z%|tpE3`-k*z2B52^fT?S`W}ao|3kxI#ZB1kQgRNlCG3%UrTCN
z9Aho7*f4Y=A)&QGjYs8jCbUR`d`c;8Kso;2@-!~H>fWlp*f6eKFlWMhK20MQHrQx!
zYouUm@-@Hh_QR1<0CTdlycB~3a}(#59{0#ipw%v}63!>8E}TQm{-LL*T#20Dp8-#a
zNeY~#73$bl@7?>lHUd(z2(HQ<hbewWJAax-N&D>?H>+I&RcT6;BaeodL1No|2l<Ea
z(SBMLt_20{y7m04LX&x~RP`NL86SSrDeY+V!3u@k=22=7W_*Ie0!Si#VnkR{QmmXo
zt!OuTVsUG0&4Z^enPGgH3VDw0boN9J*L-asD?ht`-;psxDNE5(=h(SRs$g}jZ72WA
z*lKl7eAnmZJpo&kh08acC53YBihIc<5J3C}vYUn;K!R~TrQV|mRR*LYQG^ZNL-oRY
zdp+{Lwdw4orwd)AvAQ$F_CB4RIjgu)aW~`RXWyj_PL~~fWo;y~M85lx3Z-LVNxARb
zBB#-pbas+(7a!cEHu{9hKApwlt$F!}!eC=oTHU70FFR>&%N41pAo=<Y^V)qo)UQWn
zz&p6|g2C0Vi9oMnAOp7=uupKb^JXuJ7l9zg1*Zg?43PaAEKO`VfdZQ)aMWN6P$I8u
z5jfUhCz(3_q?U?$NciiQ*QJQKn-Gyy$4!qxP~<ML%rM?SwVavh37v5u47oN{ZEbdk
zpSR|rwivb$l4(r<;3~yyfxY#`uGq&TtBF1kBX0S1y7r-4u7^FstAIPm4T3A~FN(t>
z=-!7Qzx{A0>1bJ1m_*m~&0clgcb&-0crbX12qdKeR6SBDy?u1)G_lL;Ix@(n*ucLt
zVEZYcGyvjN$Hpmtj}$mp<PK?77Ikqf8E<EH*C~$6bJUu}4jHZ2FYM~`J8N^QKArWL
z>$hkUi?gCu>nI`Lv1`17qe;!eXaKdqtE?xj3;i({{PCsmg>bJZ_^#+5j^!Q?upWJ-
zglnTg0{%$p&CE|2kAgs6o=s|!HfK0LX==wFaPmw0A?{k3UnK3<v^Rg_-Z3#xC4AV1
z;AI(Qc|h-UgI)#m6&iUXZcV%RA{9wzRu1wPa<C$TBSt)J=lmBssZUNVdo_7U(OtMl
zN7sxkl@x#GiGB49yfW^$Q<bBtu%xWqwi&nB)2Fo*8=0wmtGwwtMSXs;w%f|wz}MRh
zl$64Uv;g86fOxmx)^=(Df%@t4nvuteAvA@O5N0jk4svkZ=D_~%{iXV$z8OSlsG}9}
zGWB*c^;9Cs4NmftGVce3cO{=Zxpocplh~i(3PJKBKyOiMpC%!EUf|0Z_sZQ{Xkl5a
zg;P3^p$1AmVvt6OPA2CqUh1HFdc3;BHO?<>h9C2?JK^;ayNTV;=Q-X!;DqdOv|97z
z4zFvt1HYCQ10E$wPWf>Cl5-8nVDi7bxmL5`;Jw7ZmWMg7)+~qfAx=M7DD1sjdGOkT
zG)jQi95sH`#Mz=u=e2zQXJ?fnaMCl68<H+CxWgBaf-C2u4HZxd>Sv*_-Hsyx*)7x0
zbrSs6#ocBj5sSVRQF{3|qt`2`1c@fBK&nqx-A9tv5;fshAO}0Q`W41nWR#d(y9`zJ
z`Q(E_=UMbg)0o-xpcU<y+bP_WgS?IDIAW5hz2yqlpPsV^uM@xY99ypP@Wr7M89$pb
zhg=B+XOxrdX1$u_YJuh-Kphru@!5mjP=5xgL_1k}$XtJI^-Q9m^ZoEM&IGWr0%0y+
zh%i>{c8B`qiHX64tyzg|w=bGKdeEJB=~*$VqaUgOu|aDHAhDlR3B37}zfzh{`;>lm
z_pw7nC44D<HhDEdhH0mkk%YcQ!T>5{14j`D#Z2aS-^Q2jzW8P|wNm<Pz5EFkR{gn9
zuMqniwYelui=BtOmXV$L%S5In-A{3^YU+pcfs2%$TEM95tVd2#TAFmNx2g(xADj0c
zW*|l_WXwOT5dN1VqUI*7`I%v%8lSv}5+gJ)*UE#M6jW)8A77D}edL@M>5oIcr9Fa@
z=PMRS3tSBWbO<DbN0+F=#ybGjZp`a|3~G=IsVXM&M$JWtBpk&-UO;0#oZz!<R#`p&
z6NGH3UlM`e{XLYer?w-P-3RC^tFMT2T+4H1K6#i5Xr}$mpCzs68VC6uC6O(?8lvM!
zXgdtO0a$q)Ko+&d^y_ZMni@%?$8K)76d4yl0)m$|w4b`dp_U^Nk1)*uip5oTW9IyR
z3tC``Y<qDgAb^=ZRaTTebNizA;TIvjp}%{_G1r#k8o*1RxJsB@0SEUt9Mil2sL1ex
zg>tk$I}8!dD<rhcL0(QCBH{`Qg=R8xkf$ZIPu{AT4egB0bjte<NX>9Vwh!lReQ(Sl
za_XoS%B8Bd-O#1qrak}$A^3+tpls%&oaD~~b1L|+n3J)FCII4Ai+AmTwsBQ(2ft3`
z%iY+W<9EQa@!q4BDb(LwRT?%}4cD|ez1IW)0JonmEEUK?`{+e!GM+R0s3n<7J@Y8k
ziI$#bP0>-zdW!`Z!9IT`<DTl#EWcX;&`22|iPxiiPOS8t8SPO$j)4pno((aXTNJk-
zCfx0sybvMLDew#^eOVJ_?$P|r;nIvgU=W*81*$7>KO2cHB2s{hqd_q>U6T2!d3R9h
z@N7%Vf#56fdXI6GWYVc;#_|N6*xV_p%KgM*YM4{s@!Fz7lY9%;+^hnMIJ)X?Sin^H
zK-nKF2)Z){HR&=0*Zmn<wV?p17@Ha$vqjoOGct02l|NPC_uj<t(eT*y8aLskt8zlO
z9WW7uel`4k1#<F1VQ~6KBfYv?rHzPYiJpx!Hf$x!X=&K7g|J~#CEP(xJE7m&ZxlPE
zm99#JoEK9@j<v-51jX?acnOTu3`7*|ZrRvL+Ot22Z{Asid8!|mH;C{{rByx#t6j3Y
zl1!k}V|((<3*p8?nQxvsmZsvJ5y-i1`5MCc`CT1nJq@bdGPoOy+GJn>^s#@!-PE$e
z3x?*m@8dO|pll)ybM>KXlJbcH(9SKuTQ~sq<!UvA#_zt_;$$5hcw9C6=#O5o#VwhR
zkNd~0E%(qp{7*;qKW^}KH@=LbP(l_dQ(8aFbb-+-&4)^JB*)FW30?ZEnQ|SkI(v5Z
zBSo5CQJ4^1BJd7~o^?E2M{Nb3pZanb2t@bpUp+cn*~tC6!^k{rNQ3^Xr%&6R^EuHs
zUVO?8=?)S*ANWFye-~7H*`gR>OIn0D7I(*jgE^87(c8|xO_eW}LuX^6xjsA~Q<Aaw
zHG~yW%XYfJ@<U7FK@pdOp>z_zss_o*!Eq4>>Y=Yk^~g0ruv~`#S^)QT00XnKrfGSv
z&j<JoOH0kkjc+NUR)|Rh3FjMxgF3>tnnpd55%P&&jEL>Md7KD=$Fd`#%SL4F-ep-V
z!^N>kQ1lghU-5cfiC5RgZm|C*0fEgE?3aMPQeQ{z$jXQFNtZt<<XIUU1d2btGrJ}(
z-vL0u8qPh(!~;E=x-4YV)*qP@JC|LfqS=97PHA{Q=`zw^VR?+Q@%5jRV}{WdM%<;5
z(;6xw94?ROkc}M4NfxYqAvju6<;9;2mYa=7sNvD<D)K&&kRjEy?*yNPtvJY!V)zy9
z4vgNFsD?W4{aPimRScUmbF{E9njZbMpNQXAl$8H-VF|iH@@rNcfs{N|jVtRgjlm;q
zqTw0~9GF^WG(G3z`uK9cr9fLU#Q$KXQ-MXlfdXTe;7hPP8xA`JH)u%IDER?ds_jLv
zHXA*D6oVW`^1wRqYRZpMJ=o1?m&eZiw?e!$NEYk~>;k)rkl9bKlB!ie$+4)TV-!T9
zu>op%M{WmyUt3}6WsrU|{&8UKYx1wEN;+_EHQuaw39x>HM7N1)-0R@LB2;fA*kYg9
z6+@e}vzT(eH)qy4o@{~YiN%$B{=nn~XZvbWN)Rz*4WF$R2?~g8+7`hHo5poS!V)Dp
zxnyzhO!Q)Rv5lLJxiB@^t?Y)oA%tH)>9Jd(<Rd;WD`O=i52NpLkbgQ~dVxqThsUh6
z2=F{IYn)4`7c;f7o|v+ikgV<NjEsVEB-A+H(bh(vpFC|RuV}c-d=;lbktqd2=znB!
z`<=3TRsSr&tSL|aZhoG!{DuAL^m70SLjj{qMqb8xvA?0;;N`BMpy2AdyW1z$C-y)@
zWHT-DSBeTp!V!W(3A=@nFZ_CdHIXPU({55y+|3(AN{I-j1fG<<0{lXqug<N8LZ@&w
z5<7k1s2EdPCG?=XBSOjuOx+EBbU^cwEox;ur4PdtRNW+U#ndvSs#27nWUCH9?{U?x
zWN%JYZ%tOM7ZjF2+1Vfd8rHI&eSsQp(@ZG^@O_6mfYgApecd$$?4%~lzbZEtKs!IZ
zv_R?Il=%!cn7!g3(%BFQ+UQ1<(tC^<=V=hIH6NrCS22e>W96e_Ws_6xYV&7yb+?Y_
z_Vwt3^i-_<AKy<JwtV{5e`UCj>Jy2b{`@s&N)y&I*(06?#!mh^5#N-;ex&T1{vv|i
zxtsGAqqDfp6x*hWyQ!n0%a0n^#VMG7Eao`bRc$B$*q~#=#R&#*4RBp062g<t(`AAE
znFfA3fFnxzY-7Ax44|WhkChmoe0B15z1En7UjgT6&qcP#!IqG-pjzn$p#cM^a^vC6
zx}(-7f2|cls9p#M(5uh5Ska$=G2g6vGiI{z!_tE6q!ei_>HN)o%-HY)xY75&E-NdG
zB@8e|AA0-{aMRB4*`L_e76$pYm`rf<ay?cuJ9~R2w1kI;2dGl~($q`6BDS)z!NJW9
zP|_3>>Z=7^em*LDXSHAbpn`}#w*R}wLArY)lPL)x(t|=Jv(ycht}nYyE~sU&6X9Q<
z&&ko)PKk=*jb`EUQ+x5^1<1fXO=%XV6}>0pJG)>Z{4nhpu=RnoE6xkQi3O+3?*P+^
zfFLNnSHbMyJx+>`y0P(_^X2rX9Tc|;E;-7Ak({>Dob&MVHecK!49WCbLrpm~887S7
z$99DxFd+E=eg{2694V^>&SOJ5J3F<`-S@>WBw&#Jle1taUo`fQA9XeSy`i1+6-{4V
zS0qk*_|F&^K=FKCT}_{#?`XBPqlo7(?H^+2QeI2%(&p+s*r!RuqyakULSI|P^WeHI
z(VZ(HAvH?eNZ<CP#<r>tihb$Nt-dk)ootB0mLVm`pt|EVSow>IfF9AVt{)BC9aPcO
zB46}dZzQyjj*J*TUj$^7*jO20=o=dweLg>20XBM4Z>a@<z$@(f{5eI{Gq<G8;(abP
z)~xwC>P+ax+lx=DZmAOug*s@Oi<hnFS@Yie6P_nx;?IxZ)+|_Kv$O>TRz5G5Mz2KU
zyn4B(OglUKT6S>@ZGJwuou1jpV#zAX0|}l|IYWFiQdUvugWb&dQgpFdaq_kR|HVOe
zZkxo0$v$lbk*5;V*FL<E*0Kowl{&Z1yB<BKNr*Bixzf{=LQJX}9t%&%LtB9Gkx)-D
z4IIhV-g8xGWRnZP;F)C5+}X*k9=i60OTD1*X`jf)!#QkyBwqk{eX6+tD6a)Ljs>(I
zZ#)bd9_$j70P<=iHrDa+quLzyfDqvA-s|;RBp^P_&N4ve>c(lwGg2vU5Y5-`E{RjL
zr+1v!dEC0%=>dw3Q!X@P?|_TQo4rP$#dJuzYt%>@|5knhzK0u@xStrvJ(t`=hj!8#
zXw$!=^ixr5-F)X8bh_yfgXqFzneBaub_QdzFTXJGx<z?r3S`fJfusiu0vRx2hI`=a
z8l%Ooy=)Y9m=>N&<L_kU2&{jn_nonIgM<86y8h&pexTICMnw<~<`obb5n7@25YLr@
zVlNw6T7Cq_HC`|D0~z$wRomtVqu0pj5y-eCmgkK0T3S|DaR9egxnGZ;s`<tDy4kBZ
z7^AL7w#Z*9sTJzI9@QuspVYU%)FaZCAFM;9&bO4Ag!60Lfy*{t1~v?WW6h1GL=$KK
z0DWj;3kR5$(cuL`0y<)GRYEH748qs1;~Rg1u%U!j`rW(A+(2R4>^H0B&LFHpVghT-
z`(Tx2gLDRg%<C&evz0?uFASIN4{Ow|0&Z^Tq<W|fOn!rAH*;v*?|=)|+zh(FyzC6Q
zExajkI_6i7pE>Ld3lsnx%y78+w0Os9MMYMI0P0C1F$rN}s|47CD`CXijE9lQZ;!R=
zcFXh`1&k2*-2D$d3T=`J=e~-BH$VnX_7fDC1|Q6bFpKdY!qmeD<shmo)1Cpp%9IJB
z<!st{Ve(<l>Kx?INvkm7MX4!I{g2~fOJaHD7e9>uM-pfl*PR<?o0AjMF*S&Bhv6dR
zWdn(DNBtrj)YX9a`eAHBjHJ))SHY(5&mPGD+JdpM$dn;(2&zERPYi4CK+9?%xe8v%
z0NGJpAthCIj`>q$5C3{4YJU8OO30*voBO3tpRi<{`$3sr0FQBVGd)D;Uq(g3K`vV`
zJfh$qFbVq#?=h{bsu2!hxFh}eF=Fb0DMqy8ew+<|s6o*0IZY5i6-s#sVoB4-ml(Mz
z1+9c)C~J+JfkVFpz5qP*;jhVmoT!)@N<wI7?+OxgGG?qwhFD1<cK<ScJW5O|m<^L6
zjdGt<*im6xWNO;2?yDh`7`;6k8w4TBU#bB%9=NTotei+;wQ6|2zZ63Y+F8Dwl=%<k
z15)@~65!xfP3Kl^?TQL)B~g*jUy6`DT-*F2`Qo>w{*MBH5wB`#1%Qe{cAb7xWZUT?
zvP2kN)(g?&ZD|&YEcjap%triw>SvFpLT0Vd>3;Qib@Wc0zEV2`B4>*(|Mb8081oze
zROr0xRv|T{Nml@OdrU8d^VdrHA77=A$?Z#GdW2Xid{+R^zgP9n#ApTc&Hrgmm<}w!
z3HldFfiHdrOCKB>F$xWS-2{<5#)N*Gk5wVu4H-Cg{hVJfUn=s<{M?oDWNZ2#dIiL)
za-4|h3n}N7&gn96ra0>O_r8Fx0p61K_B&MO#!bJ+^PdvdLx4(Cu%nfZc~iVgXTmKj
zH<7O=v>$!cXS>n#;&wV_EA!W|?sX(0G}1I~q5HuxN}}0C)gB1d|FS8dIWj|<u?J|L
zn-Fh898w7UPlf*5>M+&;2een6E}3d&@Z?$bf1gU7*KA5bDvmiA=kuS2j!|E;0!8pI
z+kX`rCpj}rv;Ca+;z8t@$JT$0t*9d&nVr_EVbtRS6Zl&ne)~pDtC+q903tJ`5SgTE
zRBy<_vbjqf9-+`byf4PnU>=pMh8uCUnGHtEYD0xz9GESBiE#W@|LMQ<3R`lD7x?13
zfSbtixrg~H3q~gUrb^uZ<M=Sswvvc1LHjbB{;XP>J|FdYT<4Dm|LL+Yy*r|auU-ah
zdo!BwK#)*?dF)?JvE5F<{CrN{i<AT$Co{wT(!$=<O@d_GKW7CjDtd@_NKN&ZpB1t;
z+5%?}+qq|V^~-!*Y5s7l4w!Z0AF=_Ftcid!F5$cTm1Y0AxcvE0VOHnqBB1*ErtZvB
ziP4(>?RS2QI4<JP{|mnR@6`e$X01g0b-L7R|5fwwu|mu{K3Hd97L0#Ui!YjwbN{dR
zLLiet_G~aDV&Q)+i2oR~2=y&+_=*_<2x9+J;n_L><8yEk_813mz@Mi159g$ZC`aDJ
zeOu-nTx}=%_+Q<Uj=4J0%*=&K3g_SaU^$t&Qjq(ZBGJD{oYxY?(;WNX#Icv`=PP=c
zlS-hefO+}<c-V3>6{U7G@uX83uAJAujr<*wz6DSr{mXg+jAO*6;}k>}q?i}~ulKS6
zM{FSRD~hi<791k|9i9ATAKBI@fqPj<>?>DT+0!N>Q+5A)AJ3b?Ig(CM@&sq9?!Wi7
zq7SW~O-`4VXzp<wuJ`}LQm6ik88GkM;7EA$P{TyD!cbu5zv$&a#F5~Avvr*!6OoEE
z{+a)xn&6_@Wp?|~*VL5n*+<v@(EWd`g9zjufpa@YD}M%+|6+4{X6JwgF@fFM*>H*O
zPj#xWdU3@Hvj+Y{I#{ry1&VKpN8kzsV_$AyBgV@<a~(W%<4=n-69$1EwZnk{BFYx$
zBNi56mx8ne9JsU%+$d1F*!SfCNa7Z_-o&gMBW1}=$g=PD7`M^~%(z3aF*eV_%*@}9
z=gN%gpMLVU2fqO#g~{JfR>Q66L*_fh&bNn$hZ}uPsJ(wSBM|slDk>_RoSghWN9^Xd
z0ronF8EJA69d)#tva+Xe1_*nO@H(Hq5`VdA5<YvEy>#s16cwM-_Q@F-X$Tk{Ku`iK
zZZy7!=qbm#&9%Z(-;1*Y05|(a72s|oeRJuEaVztGo}2{*1!2LWu7|X;3F8Jyf*N$T
z3(W50k(XEMriqe=6VILWUtg0uiI|z0H_9k>e_fhtNEF4rdU@54&@ZGaG^HC2!v64T
z$M46xnHQ(QRE|8m=U}IPk%zc~Lx|bKmFsap2#E*?kUt+eIDk@ljzv>zqCJ|3sHnuG
zq`qUTo-rH%R!&Y%UeO4eZ~pnySzCLg$f-lv_tUddBvC<O;m}3;Vt0}u%M}QuQ~zMq
z4i<IN&h5Jm2poj>$IaPJj{!;G0|9}(s!|`$MS{PUNEnzyyiN}C@ZZknb?o<(MQ*<%
zP+9Ki>gp<cKkU)FLO6s_PDVz?&JKX3>m?u(oQo9O6=6C^Gv1<z7R~V=Rb)CyNp#f1
zAIV4D<d3z%$^>>bLE1opz-EN<130!h*M?p%sd`|4YROi#orH7-{{XFQC$;~c*_nn4
zT0l(xW(%{a`qS_+p)cjF-D<MmuiLNc6GH$v;`bgu6z)G<{BKnPY6Kqs=ZXz9RHqe4
z<v*i!9;L+#r>%4If`SWJR3^iL9rKMEmo1sE-Qp+ix6^{kR?G}YZRu?WgM01pzfKfb
zx>jwBEgc(=zOT`%8r9(mVY?QI<mb<0GMI8Vh6bW}d^dDv$+?XgOVYP!tW7%*PQTiB
zmomDai=NkJtW5X$L{^Q7aLe0%+r2GQpWSnzKj{(_pddr`8o|m;uAvMa4Alkyy~=|M
zHPa+%&!tBw$WU%0`26Y<*&)i%>}0q!FG_wsqltBn-159{@E^Z|7h+}W!4t|M|Jh#r
z(R`H=`qAIb9h}OLl(O#nC4TtFub@+bIr8>gGoinF`Ln$qn$ksloWF<nahV(8DU)(h
zQTdNwL8rN(NoMk)$A9(mM|+m%o}9X{--E_9*CFDk(Z>GeG!yxcUol@9=_$jDN+$fP
zmp|Lv?X<6HY5pw(Omjbu!s$&kmM%D+{Nq>9scjoOLZRL5uU`IaFGb9xo1E~sOfk(p
z?K33<a=fs%{l~8uSu&UaslRLetCv68`*ON%P~!aeBt*VA&kog%KN3?i{Kv0<OpTc)
z@7=$8`Ln&H=&^rH>i_>#f6$|uy?C9Tembbcy*bn5_pL7`CgxGaglanM@hpmG4)`NL
za47n0L$8R4?*YK!-5fP^67kxz!Tfb%a~~Y=oc%<%mWy^R<QA;$EOrx73*T7}P&Q!K
zE!p_-qp^hrlr<-d`qa`b9~HvbA;61)Ke{_Nii?X&o6fzQoJuw1T5JHt4k+1WFrfqB
zLPVG<PW+>@gBgGq9;({~9F2s;rR^^@^HC3@AP;V?Yn|ThRsmI+kLEOBSXeJ%_&8Wt
z#ByjXEUY<J2o{!j`XYX3Je$@#<??TJ89ZRdJb@Alq;dB8a`QDNi<xK-RqACv&IU;f
zxw*M|nP!7vs{ee773e{G%Oxx<<~s*Om;R+~NYE85tk4{1(AiTmLXdFQYY;R)jMAEz
zm{85D$j&x<^oVw}GwcNr%QFv;gD#rRh@_rQG&qJe{pkGI#l^+f*H=sHwQsCvq^7=p
zN=8OTT%4wn(P=k#)5`6zF}+xM3xE^D;Kyi^!=e9P`mA(r&g{(WTU(*NtN`U3K_?S{
zc+7ZHzshbToV)SPI%#`(R#q0EYoO=lEw8E3Z{oA6w-6<~d-pECRS(Sb=OY-z_jJ3n
z!K75SzQfke&ZX`<pl~o}B{Qh7Qjw7flC8gd|H-7=h_>!(bl8|UrmWx4(9jR_=i<|+
zm$!7c4f35KLT&*6T*=S-^l*E<q>k&ua+p1a@JmJTyG6x>EPiM5cH9wcW4htR(az%W
z{Po*h`302tL7`);?(Xh})s7=HTN4%5fNhX+ra`u_uyFUtn;|H`Gm?*oCnh=hFfaTm
zS!ob+0$}cW-e7W8KDI96@x#cWk917*cqK=iHYPPy$i=e9^RqVu5gzWbHC-$QjteO$
zDEO^=ftUc9;l_A50~x;0uMypZgaiOUqJ8+|9-QZn_wi~TSV-kch)Xplr>36VpURdA
z*VNR+$H(u@5GicauD=C^G&D3Y_?4?@Y7WlM0%=8U#xlR=ooCv8n#^k++=^BwLpC=0
zey?>0i2d&*>t%OO&Z0&3$5Z&N!CaEf-HUo>OC{*&y_leD#5L9>Gk5Pxd-!dqy3G%%
zsi~%ZXCv#s*4!UN&D}ehv9f;)hI)SPd-fCt`2e5_!@|l%1}%P1n5d7BPnw{UsL$~p
zkfM)hCg`bdxA)<hub+y_*U2h-LjwaKPC%)v>gDC7p)t^<HV^eapkxsgtj@}k@T+%U
zAKlyA+g<MK)TqBz+Zvma@_ea>;_NAA1(gjhDEPZ4LMbUJV1R@n^Cw$PV^hJB^&QXW
zBL$~lY>AvbH8Gh66O}W-x=o2JFE6jH6=Gv!<LBQUL@I&X1QykFT9X`s`}e8Ee2x4Z
zs;5YA-h8K$c1vx$<2Xc~SrzCfIJQnQ^`(fE1(y54X@(l;XG3LW<v<g1XACocqDKp_
z!MUnZj!m0=y3fh}^M+}0+J1l|)5Q?YVN$QhU~;MXk@KDtinx?+e`{vTrN$S`VRsU@
z*mfH!<|OtHg|tcC2kQ6Td-p4KbY6YEx3e=3#^!RahwO{ZG3{QAfjHwRPYwGvmr;M{
z7_1a9ww@mC-X-irkh23t_!Zt}Uo;RT#b7eO@vX+S*&mxLsj#wA;LeS;+w)8q_mKQe
zANYXL$61*euTn1gKfdMv*nbqpTeSVVBl`b+_~m<t3BJ1TfSy2GyK2JWcz$wnGCdu#
znk4#sXJMvEQ9~nkj{-{>>$d~D^im%XM8LJRkGw@?G`LLa$G?3`PfKGzMqC_1pVj#Q
zV8*>0Sj<=+Fsz8;s;XaEA=H{=iy<_=9~S}5E~nvr`%#X}K980Dtor);8rS8<)1B^P
zU1||8H*`x-g>@enxV!N?ug_)$1fduPR$x1EuV3HhHu*L=S>U$_^c3U2!AYHg!Q{};
zZl3PvzCP0?--fBp-tO*)E(@XAtV~Qy12zKe?DmfyJ=$IRs+u>n>Lv=>;>ElADdDX%
z)}gg;(*-(&=Eugyus|^wMphR3)2l1o^*<#UZ?0}`8aH_T0ubUt4<Cw9_XQJB?fF^G
zH2Q#rVtsX$jg?ie+VM>`D{zQu_n#CL6aXX*MHLmFy#cw4Jt(j}cSlD@o;yb6W<k@?
zwe|HU_YG-@ZES2{?rU%*rS9q(-#XxN5)%`{2(UgSyl+1S%ZE7lmICcUxVCYvn=Ru_
zV5o#4=A%XWO=tU4=++Qd7Z(t!lgmuJw+!!`2Z!;9?1iNz&;%5*Gpwc7@N3i%!1ZJT
zolU)Gzn3}gd49AkLaheE<Q3f%A$L3Aa0?0mf*!b`qd1G!)Y6j5QU{Ac-}3tUdfG=1
z!cVkdNx5)X>PaamfFj~zVrJcheud-V;UTi!+}sY&&OuOqmOZw<xwWNB25>)(o6IUE
zgsIhZb#-~}0MPDZnWS!HK|y0v1ORN2S5}5RUm8f{(BJuVg~(yNjE)TdJAljwXgJA2
z?gjm6*tD+{us@MMtpm2?<3URd_6lG+z+c{oy!epKDt2*NagLqDZTbVqRSENFePyMe
z8&GKx5)tL)=0^BJ8-^lK6%}CB%Pub71gM7(K$rlogvO~Z)B3kunnjS2lLO_scmMwR
zb|=kLwUZw}=h$0kQOhVQEv2ER_63w(;9Fm(r04^$Y-sr6cyHB4;CQ2=uf(Lm>r1@f
zugtg2&CS4*0$<fsS2wl!Y|7z-JR|*42G*TuOb0z_f?I&>ft4E+-F}aZj0C%=EZ>H*
zvY#M$@Tf7$^x4-}QHd-l7&CeC1TYop>rbRVa0OUAK&j5Z?qc{^f$;<Xoen%NaDtw#
z1cD0-3q__)COaaRc&IOE2(dZ}OG-qYXFmbHOUQZv7SHt`zt{fLM0~9I#pUI`Ofj+1
zJgi6b${-*ShrGuB&rsz5*8}`#Z1IP0Im5Aa;I?mBiB<u=2undmRk}p-k^lb#%@!G>

diff --git a/vignettes/examples/timeseries/timeseries_classification_from_scratch/unnamed-chunk-9-1.png b/vignettes/examples/timeseries/timeseries_classification_from_scratch/unnamed-chunk-9-1.png
index 45d0436e03763e43c399a94f7c74d867ce8874e1..801cd19b824f12dcc04fd2ac85a6dcb16de37026 100644
GIT binary patch
literal 321147
zcmeFabySvXyDg5(T4G>ff>@-20Z2%wfG9|ZG$J4k(j5k(7^rkhcZh_vqO^#_OSg1O
zNpt3dYoD{v7<>HAH^x4HoUwoV{nq#v@$$USbH{bfIj?!$>zSmu@J{k0<YZ)IJ8#|)
zlqMtF;z~wFA^rDX_{oS9w><u}<*ul(AlVx6zjygDo@8Xl$!-c>m9+^PZMRnO99UiZ
zl`3~-(|+0e%EI1E<5Jt1bVsfvSXR3$jU?ul>Z>N@>YE-*2rsVr87sdjsfW*<Q_}R^
z3vRMKf>*Cxk#}<LwVo>D_#OOuQ{JORyH(0$mEYbHTheXqckatpVcCe!$Lshu8O?)#
zz9e(FQugzoFNa>2Zus+M?Fo;6-Wn-zZ0nz&-EiaC_2;YCyKJfce07|9iu2D`SGapk
z|9s_Q@xLzUcQlub%;lJTidw$itiI-LTwdg$vT|!#*$v{GAC$OIo@i}d-M+nhU#M;-
zBV$lCG%|Xr`WM+1ip1Oao_PJ|ZsGs>GyNC8V1Ee<#yroS+hXv}y?b<qh2{g|A|mcH
zvks|kZEaN4)UIQ{e*L;|;X-XAu|!nYha;c*vembi7Nw@uK40}?d=3hWb{(odK0aeD
z8T&H%9T(Y{nBvPSeOXjn^6jlPmQo5=7u&5ny{|ap3O3u<*UO8N&!4-$`Lh<yIj=2s
z{w#UvS#S)$z2UO*uP)DNEXpg4cTG$thKH4sRk8=FLraT``}+F4ySn-+eN&V&^j*m*
z-@bht!socyUlq8QYMJ-m-Mi9Z*IeBx&$Qt76&3vh19k4*i;jsYbXd5^&Q59~>HVDK
zcXf4TWMnKaFIQAlq+MmOva-rBXk@vatYT(nW@u=rp`r1Ne-oahnive1tYq<EZc@D5
zzDze)S8W}gO0A2xZ#Q1hEZK4RVkCZ5Vb99SDiOky?XWPFntGo3(PeJ#ob2pvtu#C~
zul@WOJj;WEf*c$j-F6(Vjg>mpm(Rt;Rr-p0X)LSr74_w4?MdA2CF3@QreoC9!LMH*
zx4ZfL05fi*{)oAm8H;*B;_`dPrMUwK4qWvA_WAQwi`VgT33hgN<gKARwo}#oYr2dd
z1?!WPvouRz&925+jkRE+!g)ja9PLILl04}p#yj#ab8#s)h3S*xXNO7qGKB>N^BfkW
z?GH06XQpRn${RSWEY6Y=ZfB@y8yH-+82o{UvK;*Jxw%<!mL*%o372{O`n8LT%hRWS
zk9nx|4pbBs6*V?mk2bv@A0Mx<Z%kBVHELlSJB0;Q)=TDMUAnw5{c@#q=M_57@{6pj
zLj2+R_P-B|73!0iYE~^r8rUyf@aMMFmzVd6_HfkD=&tZSld-=pLYSgvyRgSGX3z2`
zI}X3~@rmwh9Pax3i1LhlJ)6x$N4`)>0H<kqw5V}M9^>fRV9hBKAEr@Br1oxens)q8
zDqcI;^wyRPgN(epLQ1K39u&A@u^6?amp3%X8^tcpjvT}wE(kuRqM|Z0H+Ma_I@f9v
z!1Dh6d&Nvc|G6yXu+-H1YpW|*3k}jPbJcP$kFY1J7qaAK`K9pOIgc=Rl7E5BrS=iA
zf?aC-{Qbw;vduA~+Lb<}`qHmowMt$buJk>}ac7y!ynnPMqw@QAaZ${}Vi2FBlIZEJ
zs(yii9!EJ?PM*AGbh7u+RtoVD9>!6Av)&uUJFbc?&J3Mu`P|v5o?aLj*ovFDR~?eH
zGA4ii!i9z5gp_2p{NcVnIW3(N2M_vA4^(p)G~BY#+PDSll;#K#wePSU<uInYWa!7j
ztMkpTCdbkse+vzt!#q|<qNoH12eCf<14rI_xT&x-CiqbZRt2zo9OI48EUT}smT0PL
zZf-6u6|R(AS^9nO(sdyr)f|hPO&xi*(?5Rvz}St*B|Uod<=4+r`?=A$M!9?U?rq&g
z6|0t?qef(b*pozpxn371Lr-lV5D@S^gjYG`GuF~O`+)<@N)Br)zjJ=C{h0a~85wE+
zdmI~3HP7b4$&)9UnZ;zPCHNM!mgj%PGk^a48Ih<c1h-0JyHV^;dD&`|!SG-()xhxZ
zkwb?V+VsRv(9+VLJ$sgihlieC*e1o_U-8Nnm%&}u0k{0lhlvGV3K-XRLg4tZnaFA9
z3yO-a3AIfRlg@}a+RY4x4b40{DI_fH_T<TRVPS(IR_k#UdU|^5t;@ToJ`N5J$|cBK
z4%e60)Vw$AtHAW@*F{&wggUiE-rQ$zzf|(_s7Xhj;*6yqpJUFjmO!>jfyS3D8XSmF
z0qh3ud={Mrj;?i-+qZAu_}5>LgBz1om7_!`HI^{bdHMOlK|w9)dUY63>~|KGkCHa?
zalygu4N1yEmAj+t$^;O0WMyS_bq6DaJ*K9oEom^d4qcq;lw@RwnhPA4qQwHEIg`_G
z`Cl|2_`cHZM)SBG1A!=Lxjb*0zlGmneiGM5w2qyBf3GUQpfP!jl;E2h<Lymqdavoc
zI7F}dgqeZiN5T&FgXh#_4bAWA={08>DQmcW{rWXFmO&$TlUqWCR+@5_@hmnMC#Qnw
zsLk*3c9X)DxrOP0>HaF2+{)(WY{d`v5GtiI=a-hwspTn*H7V}f=UEdWTpfJ*Hs`#{
zhK-M3(u)V}4$&JKAMc2fpwSp8D=R}-iu?Z49)afJ=IuZa8Nd%2->JDRCTC`N?B^2i
zQ;8w?`mm_x=+#9d(w|XEOiaY6SvKbbrMOJXZRk=hT8L+$6&ZGr!@%oQAM4&*DB!m1
znbiT+wD=*aTiGhB1EJ0V0r4qY&a<=gn)N=_UE4}=EY`BYq9DHKwU|^+P)fXPwwZR#
zFAkHAL_G&NlOYGW_Uw-zpB^~xxM-0Pa`cknZY%r6nYjDMf;>HCoSX_$KMcMybLX<V
zcI?=(@Xdh%0VQ9*{=I~_E}0bLQ!OVh{%X)n>C;nUjyjmv0nT;LkPtq-+SA#0_BLS7
z1eZn%LS*SY`vwLi0@!YT<a?i*Dw(cI=ZSEeA{`}?vxCKM!TuX20GaKl2fuN1TAxm+
zWB&sG&LeEQ$JDcoRWp2MM;f>1%Ed`XAvo&4RCk>9_V@Q!($v=<R(D>Zr3w@EWhp9&
zjg7^!cy>TuLE#V;)mg)druPx&b4a~dzTD>hFJ1*>mp)cV0hE2w|LMU-%-rrwgO)+D
zlwud1>JUm66ImN<!Q$r!g(w_!k?kZUCEq;S&ieJYhpVd)GCg)r4IU`Ra>R#OS^l@d
z*w`3J>Ex+X0@tqP<>l44l)gA@K0m2JI#5#_j~(tf-TzWa6T#7d6i+*fD{D$fNQh-1
z;XXUi7J1VXu}U{bgz@ra`}Q15Au9xr9}fvUdgx^|buJrqr(*Q2!|huO?PlYB>{C>8
z=>kgSl9c3pKX)$i7*BS6=3iU>5IudsT{i66W{rr@(9rW9OUu*M7}V9?@_oe~a`CbY
zGj$TKjqV|d!#zDG7Y94mUyE{QJxPg*GQK?Hqs<?b#HleVKQ@oZA<fr3J*Ifvv-~k@
z*pA83W>)6kbhNb3GPl!PIl43?Dt@s4ZT#p9r~fUwHcD|i&&``RZ>x^LI-8xHm0N8}
zQRnZ@skw}3l4~W#c?W@OXY!A(8YC*khZGoczFP#E{-n!R#mdw_6h)q$ovk!Z+P^Tw
zxmm(yi|gB{s3?|GUP{>btK+#79VKN5ki=@t^f<yUlC2Rm7`79Ua5uGN%EDWr+6(St
z;^Hiyw*pVo&^080i$s)HS06Mw_oi-dh3`3&Px*0iaXCATH8jG8ShjB6s(PIv7?A?=
z7CqGQ@Zm!@3K}|k`hmVaMx|S~Zt*RRvbL8ZbS;GoJx@$Zk~^xZs#-YT^Bi}RlarHa
z&?r9kc6oVOHP`CdtGTH@(RP}g8{8oSuMv9x-NSXQ$QC*|ZXB1Z{K+=94*`O!W)6Xl
z`|{-x`8#^`+2%UJ!onK%YHA%tuG=|`S|SfTP*hSH&iKg5%*=fJ_?@<orut)u`@D8D
zx`j?FkM~hLdi02$nMM5O%{@lWSU|WPixiQo*>2Pr$=2%xjp>k3Q&Xd(qYHl!fSUoP
z5=zOj|D8PSfn^rn&gN<imQwrSTWyr>bbt8R_5)`X(zKqXwwIUR>~vGeG7diP0caEW
z{Q2|2^7wFiE(Sqi;eq~s@`HbeUbZ&=`gG6Yfh{0chv$ggHW4#SN_gIX`J%t!k9SwG
zzV>9|Ho1|@Glc4PSp$C^xCFq7q$-+nlHJHmM^iJRk&{+P=KCQeu9glAfFn&$N4|ZG
zM5w$08!M~w$NL%y@=5hAGSbqR6t<n--rlpFYpb`LR*Mg+4`w_MefI3x9H|PRnf5uc
zHS)565}U$>wr7+erOq!bJe#A~woU4?$hEpOVF`)qWYygGOfN<my5q;+A;Q|p`Mr5_
zqNM}ij5>?;jADRzZLCs8d}>Wgi{1Ait}Q(7gCi3oBWF|+a{zw()GWhaQE{C#Oq7kQ
zEnJ;Fr^<v(Jk(jZHrASzYcUiq>WnbF_p_ZG8QEk<T2>at&YkT5DbM!NOG`^L$tMnu
zjqw^boi{iCwYItCqiKYaDB`(;Lta8cf>Q3YXE*QM=_*`XnE)O=uz&x@?Ch7vcyoX=
z7v=qYeSLj=<gt-6=AP_4I*yM7=(~CIX2pxMmoHyFd-j&E23905FRx`{cTe&2if?C5
zoVc1^{o*ideUZrJ%eer@LxKEEo3?BL+DVm9QWBc0Pg9WV&gyhpT3wnrb2~Y)fzNr>
z0a?RhNLtKP4J^iO7d5xT{QZp`?0BlX>)>UA)!cmg%_(7wLRwi!Bet~dRAeeIC#MZ?
zZjsgY?a%S7;z6A6{V0KoWY{i`)L|<59A%a7YHDH~B~<$1PB2<2uQtoZ%igp-|L{I-
z+(TdAMy!oY!{)t5j~d*&*8uPi>ToF4Hs~6VFCYcBN?g9Ojm^iFDUC2BJ)~L?rZss3
zjtdtQKi=;~?xr}#V}s1u+pCqWGTznIH9Fdch;+%Q#bKsK_z5Ki8JVO55n#GjPUqAW
z`h1VL{&FwxvPPv3bGU5~9WcxLw6xm%wQ^5-%h1ZXOjkEIb!1?`)}mZsoSBi7!g%)K
z*^>z^nMSSHo$S;y38|W;yHuaFq-l#H?+bOF0_en_AC&}$`=ngFy8dD5x;5d5CBwBc
zEWSi`tCLL}X-qL`%@ix$8jKtZBx%Vk1rDHv;8j*$UNiDGOMj}jJV4xE@|16-D}B^3
z13SAVaFP;tZ-4)I^9SM5ni3Bhn%%qaD9~cv*Pi;b^CBZpQ&4yyyy{NKOG=)1TAl}U
z(H&w1R@;7%MVE&D+&Pol$ea8tGoA&Oz}ptX^>K@8vvVp14z^9H8W@0@4raUu6Ip!o
zGZ0ryUk4^S&*s;)(lByz@+TBDd_du11Dd+No}Qio0s5a@@$paARg;ONj~_qcZ&FmU
z!DVF9AKf+T2iwf>@O<;820K9AX+cOv#Bz0c0TYs_Uf7Y!jc4b%^U3AoN6s%S=S;CE
z3_1%|Kv~%>Kid4di|@0H+m(WZvyrg<S#{Xj6Mo*6lf&aBs^u?&GFt{6To|m00O49)
zS-PF5@T)#fW|0qY*5z{(KLWwv7$ym`N11uXW3%MGqetz>+qg9Bt*kP2*V@`tYF6L8
zd9&!nh;c_8`~WD(@9b0^!V6AuJ40U%I|d-Usj2CU8%1Z1rD&;X+J_Hd3<TsIZ_mv#
zYHa`&C@Sh1s*915k^;(q`tqgRT%OgKl9EzDOCPRO$G3`fvFjL56CMF;rXfYW^ZR#D
zbTb6wc*n<Xp&bQ|ttBPbOJm+W8{8I&&4g89glLyzF+_0Sty|qjM@NA>fW>K`OoTeG
zN?CYT_JvP)y*f7vjKNLdeu-PRc$}6$@=5fpiHeFUXMn%TC17gn5a|Ghxw*MpT3cJ2
znzVFuK=QH0c`jYLggfBjS;oC9J+b@M<2ooKdg~V3yYrY%IlAzr0~^UiVm2casC@j<
zBu5q{3I_Do#*MCHnk6q7`!})V8G+bfH6+R>1t={ce}RM<dA)V9_DV`Jva`#1pMc?a
zpN+@m&hOf{Z=ZpIfky6z=fDqS9VgZmv-7V&xZ1$elP<8a)w{Ts<E<0B*JZUs4>B@l
z@!PY8ov9j+W{UBf8_x)pgY3Ag-Rv*{aOga}v9vY5Ji-0w{do64DU@C!WVExe^AHIi
z@sDID?T3}d(diJ8UroWk{5ml3j2JI2Qc^?&XEHDHt5>h)T2EZE0&yTSp<dVaUQ8g;
zV8VSX`$|~AE4q?2^O>M<WaZoZ>H&)_n)~`yaOnAi4cHCKD=P?WAt52gqe{3!<dyYQ
zYZMk1hM$!r|0VR?-Cde4#w?o$;hBqT0pt+j)ZN2_!`KM#mtK#iF8lwzNO;F#l;%Gn
zJN_4!=>8+6_5X!OHYY@7h6@D+1<A?Dn0TOb7G7RQKpf(mNvk)d8$K<4X>3Hu$o%Km
zEG(}8RXXhdTkOS7_-DWF?}jEtw??odP2cIyT-M`3<!c20)jYU4&*b_ms<=hvdqzgo
zG&Clo;RqJY>*H>drLCp)`0?XSTeo)R*|I`2<l<_st*tFDj}!4`0&yM$mIUmm`u6RG
zD&1r<69a?aMQwL|(Y^|A5O!rv(j)7l;LFy4GiJ!S^YinF`VSxerK2;zBo}XEXSX;%
zpRP};z`jq}02l*#<<S29`;Q(C^z!n0^F}I8nig**fad(Xjjb&*Nq*G8Nq->eG3bf>
z{OZ3n<>d)tG&kC;*@l2Pai$+?=?*V%?+he4-~}XN!}gq6JUqlvh^;pzBs`uy%f8=T
z^p;k%CCdas@C$fcT~qO26HLy|YtS9yGH-?Q<r_Aqg$EZd4xia~0V|6%wVTj{L<24v
zDZQ20xYh{KRIrnAY#2GYw+z4HwV$&vH(!Dv0@dj#v<Rd>K<UZuFCfOuifNHhg+iSc
z^#HM6ML`M%Xv@F;@(3j9?K^hlSx<C4KX?v$AW9ZRU%q4+HrIk_c<?RiLJBOlzqtcf
zGX?m<r+f?LckcT10GcQAOi^Q1Wu>TZ4e}AI&Ey@D4<ybfuCDpPEgQ&axDoGNPCz_V
zL#2eqdHDi(vVAoKCZtU$A?&=ossSxpOqH)xR8?v1f9wUfKGnxj4xo4>m+Fw1vGF*7
zuADMNU`UD)6P&*1)aeea?WW>$_-OjG1Tp}rW2`nx6sZEVN%PG{h*=!QZ8o6W6^feS
zJv}{(zbE)1<YK;8m{>x<{4#ZG?v}ki0cCly+IiI+pbkm-pw6c_nV9&DZ-E>soE!fl
zd$$eA;))4~)b!-!jT<*4mvxPe`E-B0+qiWXB#J!n9WY+3yTD6Ei=!W^y?lcp$U=mv
z5G5PDiwuv6mNGibWfmD69K1oG%AZxZCvnqW$bga<Of2IF(E&)u+xF1`C#i{s4i4%9
z*8wI`Z*Riz0ro>Xvm8|?XE$o8|MEp3qp!Ez6QsfUr5puW7cas2en<!k3Szl>neSH0
zG(3Fl7?hw8jA{cO1o(rJNmjQhwHu@K9cv$xg}uUfZsymo@~SF!wXsjRk<edkj31?>
zq-aF&V%~7m#Vq7eve&K^mAyGNJvB9xqmL=YXztEWc=zrO#tbyfZs$jn&VrpsFAZAW
z)H}6I2`u6*#k7Geual^C*p*dO9O1lw2XatPDTqZ7mw8E5m6*&L?kj}L{QBW?Aa1}W
zW%p|jtkv(_xpV8*3w_bLXtC!Xk_f#0AuY{a^`J-u#MSM;H^oXvRba{d!U~v~nfY0Y
zwA!YSB;e$<3QQ<OFTS|2pqg(dmsC?!ba$vW%GA^paIL$$n~>PmICZ{Lwa3(ZUjr4%
z$jFd0+O}iIGGwtA!~i-95d-+;T;cNG!-wNOssd1b`g8>lX3UpOzn)m{6(5!xl2z>q
zgsMzCskYmgdY*J!ol@r$HVr^mCg-WHF~vNa_jwHZ3yHPuxz-+3T;UB0U^jrWI}e@z
z1Od090g^#7Bs}O_?n+sclaqLc1or-cffR-J*XFORuC8`EujMBs82De*ZfkE}gKl%K
zVE$T?1PV1!a*kE!9fFF1)iQnc(H3Tf)H@hDf{vN@-?~tG7IX>dSyq{R`}UQUmFHY%
z$;kAHNyw=vFPCaJ)Wufi9@p&a?!KU3&(P-)<vdmmPDU>lP@k?R0|~;wV1(G*VD~^)
zY*wnp-hqKj7@=J>d~H=#Rd@ibP=Hx*Q;gy4^tAoza<*Co?dj8J6;j^;b&HFOYgc&1
zfoGy5@q4s6;^(;3(9qC^4I4ly4BI}whi(Bq2Rs5YK2X{|(0|{`S*GhOzM-6^cR#so
zz&c7t)gdcut=CiH-uEEe-{Y$JTev`LAdF!|5@h4<YHQ2AzbgdQ9*e^5@#8@`iC?_V
zt15nPpNb3DRsxu+<=IGmyDCepVI*Y7#Tf!F;zRM#!NICYN*Ped`@i`ks)bJ|#}RtZ
z?@s66b92d35l^5reE2X8WiUsp{B3$V8?~^_{A4#^H^Ro+uTmP|I%Op#`B=&B2oJvD
zj{GIa+e8&GXi8*TjdvL#Qw>IV$SRMHjS1;6U}vZ%(Q<OmH-FI4e`#!NEbwH9lo0~6
zVva@J{ZB3~WMc#TUS8ngggRzgF#PW7BV6#Frl!oYM&a_5Xoe3~(!puZ9b_&!C)ZBw
zB9pX3s8z2Fok0N4Ckc1H*L0hD5BVjgF#}XsQq4d^<4sV|+0lqwzoT#YhdZVtqA=Kj
z55#9~zkOS8@4<ugdA&cv1Z-_>5#{SFygfW5q@~|R$0J`wjQT?(j&dUP8bEoBx3-+<
zQ1xAObac!p6ufqjMMZMSEV%?*3yKqpOGO4GHChM<cZ<nEL268)wnz2Jy{+tnF21xl
zl9H5|n7+RbvU?zhiT)7{<lgvmvmbP-`Bp|mxh4M13os*7WBRSC3Xv)}G*rEJ#JvC8
zVQXt^W?EYt8)!}=lOn!5x_dd1Gf$=rLqPC}d3WA&ST-XeJR-uiJ%CLweC54y+sB#@
zIud>Lx5bXI=}s<96dD{BX8X~cYn`IPLhKFfW<j?MEGqme<2E7OrxYw!eEW7w)V9CB
zA4>5o^or*WMxs9V3%psTf3n3v4VsJ#Mu2LP8ME2F843s9O)`G8W#?Y2Kz`?f!3d&+
z(X{aO*}g6O!9BZIQmzbq51u`8nwC~*{&N8mnR2}1jTs}Rw@FD!T9kj0xqB=?7*Gx7
zG`(i4Y*`S<f;npp6WCfj28kFE!dpw5RtGx|gVkU$0j-7GayW8=C#|*%Q4soU`1|M5
z5C}n4<Ya`zKwQ}k8qSIdn<KCnBiUA-@BrD4Dl9CVlQ|Aj-;7JEzGA;6Awh`U(IE>9
z3n;(QpaeD-Aq0uBx34csPt)ENpA4Qp4j>Z#8#%e7*EwA<CkL8@pVIS|_Z~T{>?{0h
zg;~g~r5-d#6&ltOm9vsjrp)=KmFK*+3Lvp{$D!A6-h?#@jJD@CwzewQiEBY(1ti$9
zb0^7S0jiI3x~^&0rwwB!jwl-x7CIxro<r&Owk_>A!Zu_|78VwSjrW<ER+g4Vq<9jx
zItX;^iq{hWEQHs1_5eRWHWc-mva!ezWb|KR`6ARw8)cjIxlgtLiakAWyd`<5@RirK
zSFc{7JSA<ge^ul!GD#|eL8z*Jynod;S0+Y$QEhx^C@wEa^!rJokPo3-n9kD@8o~MV
z=Yb-zyP25up;6Dx&AIad|4;^VNe56KIIy_5$RyD{SQk@`sR4+9fQMJl&!3MTKD_tM
zuX}Vfu~K(!q1t(WU7dr2!>ePw0jO*vd_r>P=jVsgEEL+BWrF8cS>nX_Ap;xVzkgpv
zr46#shYugt76w8gYJj{laBFF4?Ks9Wk1|M0sXvdcnU+>$T-*?}E&%}nh*RDGT|hH8
zJ?R+fua6--l)t47xIVw~yAxBCa>EF6_VpKsOtDFUYT;u5W<{1cbNd-o00Pos>k1X#
zDz=&%Rm|`KP&|669@8l2?0|Xk;9u2kk|aMPpXkS9o2I5iJAo~CJ4G!gbk!C`*zPZG
z2I|u&)m}ydshO>T4nH|NTlM{W!>{hW`}SSNlI4Bv*orKH6frw9R0p+3OIsTWLM7kM
z66ps88aXW8mE~nNp-=*4UE=2A@;;}omR>l>qe!|65HIL)E*5J{V`IVd$A!k2{YxUW
zWG=sb-qR0mAe*`KbvB;RKewi(r{jiS7MzMT=IBUPy{zHZlX6?`$c4L)!v)@mE<g6<
z?EfBoX7m6Jje3kYwRR{H-S#D~GCR4=B;F*ar7?~%iLe{-SBD|&vsqSFRkZ_gV^2M{
zj>{K;&Vfu^WDFcGA2K0@HI|phQ?nH~rt7_CDVw*Hf&wRGtUk?WEN^5~Q{v~ZY(dQ<
zHyEW);PVQ;-xHk&=WHjtuHrBEJI2e#nL)v*iO)1{k5?(Y1T{^=4NHJ=WT8LMgpKzy
zb#X~a#U}AkKGj~=@DVxWNCbi8mb?1;NA~Oq%M(eNjmONMGkY0N=Se5FcNasBnw)h}
zbuEOPj1(zSBmg_sRsF!ri$NO(HbO<BMdg^YK25vwad5_2j`6w}kJPpKUq8L6*~BR8
zXLQ77W}aZ8KhCGT-5{csAFsl4FvLN+z+nONmL>IJH%dNT0-u%v)5YI<dU|K6s;IDy
zQAZ0Qae8Hmj=hzjp$SDXY<nwyvAQ^N&fMI**hAk|lJOvJ_ajQRg#SaVeg9Ia<L`EB
zp<Mu8LD)6qttf2qAa)mzA=>LHiGIXlgyf@A*9Bn;=o=Byx2F`ehSK)RSKNL5cLCW!
zM6W4POJn2Y!J2T8k?$21Z_BmdRR9<XkndesToezyBxZ`LT05*dEuu_x8O4vtQ5EDc
zWP=8kG31VN5rL^T2w)d~_q!m#;uou}t<8=Yi3%uy6dm>hssa$1iYfso4^dJ^Eqy`q
z3F5WC1oH|ltu_{Hxc#p9;4@6Vl~|7~DiQi7CMdrRJFhO#b8!5IDxGUJW@c$AH!_W3
z#=ad<3;7B;ye7*;y_-s)CW4tsHujw!RZXt!)*1`>C(#=n+^Hs2>r*uZGJIxcENb)~
zK|go`b>N*R=nRv|h)rVr5z>Sqt$mT^w%Wbv?E3YCP@?7AUcGpcXEF3vR*3R85Md8U
zmFt|$ehUo~lN8ztDb;VvK_q9Wn$m4gJv=;Kyr{4JfZN2PskN9usP(?k02PyTDG9pY
zGAgS3vp{EBf&G{%*Z_`;6Hsoj1{J3cdEEWj%_$%-5V)kdYXjt0z?r0$y;M|&pDw)u
z$I3S!c$Ml1RE%khznv1&6Vw(a;HCyp15%;i4yDYZIt_zzWbfXd?rseuqp`ukdDx&<
z0BDPh^+!%(k-O6H0C2*tArZ`-0mz6+g`I`!qE;E?VAsoWNLh@|>LOu?+S_!#M8(8x
zx7S|~JD4wOsqt~p1xB8LexU<gUAbk&#X?Q<PC+Hx$rxw}xXUfo{=B`Uq{L;9Y|7r1
z0v!wZT;gz(KD*+V?g#Tbua4c8z38QL?w+<brL(9R4Gm4-UlL(kb{=8yeR^OZ*8s`^
z^VzfevocHs*#RN@@`O&v$kgHan2dF_v|?%5Pn<YWL0y4@OmzNfME1T8t9VftN;>hN
zSQW>2eK&SfBjCkm5QUC0g)NTEi22MC{=D{?1&=l%P8=l`E8t4T^@9fw*2G0~JN*T4
z4RRem_k9-?rz8*u!z18G@};v%>9NSXMdYq^LUna@w0uU<_?CZl{wI*{&!0cxfx39{
zqH$+IJ}{U@KwK(JL?mbGH|1I>>V*fa>ZU})maC+r0<QU#xe!139^v5jw4~F-{_a__
zu1D7h#`8=+fNs$!=|+6#Z#E+~V-<wS#*Q#}Jt|B>CVmV(K%SEK0g&zjNXp=pxy_a|
zD2vi=lEVA&9xKyh#Xv>2++T6G=`&}iR6%Znr!re|qhFAK>Dt@dpC^iB_jA+HMVEaq
zKEdNIE4n5A6wqA;&zZAlN5N_BcA5e2Y5c^OLYEOX)jxVDU%U2@B_#w)@qTf_pondT
zhrg~XQpY~W{iNzAb9+Qoz>N^{DEI6k$_K974+3nFii?XYDk@4!hyh~N`JRVTr>-ri
z$AbqC;+exkcQOm>JU#X`Z1G@$;1>A0F4=^BuDvoN#872UzkzImiqIs)si>*Fvv#%~
zx-rY7T`wK48%jm*S$>w8*`Pi)>|?N>-~8Mero*UiA}6QySV1MS?sOdJ?d=^KTbZur
z2Z6cPneFb)Bv;#Ub%9rYX~>5tEFi)Vn4V3qr?*!&(@@cKP1N@B(>axtGHWr*R`v|*
zG0dg=Q7m=<WY#sB4{H}zXoPM8nn=z#w{A}+HIOQ_E^f3k+`l*!6Ix%g04*myR3rYn
zg_9(9y5Qy3mKOJrytvBK5vVq}qbNTR?|K!?w<rM{)A(e6odiFnmCrqo)CU{48n<R*
zsfo#Kg<}WK8+Y;RyHF)vk+Rshp<DRIl4|^;r@LY~LyA*3bP9z(zi-L^{veslNobj5
z7U2O!r%#`LD*p6`xTTNG8~;~QgHxNOv$w0kYN5;$a~d*=VTBiC#Sjw;_}Cx-mZ&8r
zJbEO6?L^e^ex5j*GjSzmsMY@7y?$sHgdYJmzg2LmJ$v?4hw@jW3dZ*8ewttW$xe3C
z%k88Nc9KHU6`}4m2H$t_FZG)COSNxP*V^dx^Ra$o@=6s)XKW_VADMz%EhIRYIY3x<
z!BSq_M>eXGUd*QH;lnwVi(qS}+`^_nLi^1hzm-NphD5Wyl|*;Q{aDPqJ_@fEv)h|V
zou|ZfuIe87P)0nr^Bb?{zc!FD@U1V3px|Kd&euqB-ojSH;+Apm=GZ+-!Yha4tu2I4
zqbdsYjFP80R4RSXecQ>&QR%s8(2xk6iV;aH`*Q2~5E25a0~m(7v}LiEb8p=p8q@Q%
za~j_*`O#kEJQE+e_Y<VU=9ZTDOw?W^OLvG?P!vYbW*2QvjTec$Hnx!R^>J%Q`KAZa
zgP(JW5))grS$|wS0+ud~WTl_bu6dTdD^ceSCuGUU1lK2v1oYoG22L|#zX0axoyEU2
z`k_Zw8pf5NO7<FU>SVLA$o%~L&lEF0!{=)0Lmj4CO$tGxm%%Rp>JDAhdOCbS6)|JA
zv2c|gwZtr?FVHi+23R?{xT=D<reSM!fAIpYPHmLdhP~MOF6?iiTu6>%v6i!M^h!v@
z?6NeLwuW4z<!QY@OB*E^UW?2SM|~Hr&yOgVsU|buP7WGeKF&PdE~w5>bJLtQoS-tD
zx+${S<2n1`1vA(9iR8e(qAw^Uq(>8Z&6ONu-SF$0|E;;M&l{-|?A)F{@d$T=+<xZ}
zF;&U1M0saH_?XSP#LcagZC!xu=;r2@>S0E!^yY()psz4xXtYV72o`7sb?M`niMZs*
zy4C$1(ktBXzamKtA$wbowansQ!0JTV*48E*tklM4UfFICFqkpxj=8mUA&3uD2Ecr<
z%{wM0>hR-`%8j{`|7Mv|mD#zyzJtAQj}PETvHF0*L$9sE46ryE`2ULZg?|eaBO!CN
zEnBA4K*;w^e6Ui2m%smEYU-q)$7zK)ZQ$0Vv=1E`L>1K}g~zcy+g!LQc5V&<S#0LK
ze9fb0nwpxf&d$L2(0t=tYGGxh-s)AZ1va|x60|1i^y%Dd7QO_O6;b<}tSf(A*CN5y
z<b71A9e0k2t6gdFK69Ii-fqVcwi^~-82iE!gp7SmKIto}sz5ohcf;f{J-XDLB}Nj^
z%;0y;k}cFY!eKl*Fz_27gh?4P-i*(&5s0i6jz4_PDu!#n(tvFDN;&4>LbB_VCrl;@
z`34n?p%{MXNcDZ)k-y66L;pTH(OIZ|@)3%X&{QQdgns*5V_OuJfGR$<T7~j@Du|r$
zBf;)>6Kn@A2`sHwG(fo$kkZn64fZxcyMz+s&U5qBg$q(M)rt3c^?*TK+XX4ADUC*J
zBHS`EEZ=Ng1(zx+fu{0z^^f~zW_t_vOdRM=34Hld4Bp)r(mOCJf<*3<!(bDXY1y>s
z@iy*bRkqzlu8^6ToZ(5$(dFoqK0anjf^_J<gU(BaG^2Fb4$Hk}h)1MwZ8di)^S*oN
zur%G6#MK#tBqmavRqa`0YS0fe9Tt)&hltPcjue1w1<`84gTdL4u#=G|XqN4OnR0K<
z%*@oMe}Kd#$92ktkmdNoH$k*Nv*V0VjL6}8D8>_3;I1OTwe3eKsJZtaIuzS5KQkng
zQ7_T`{9M8LE>p`!n1VHa!qg(zD9cpVH$~hP>IBHx4QyP-<89eRKNTscIPWOD0XqnS
z9~#>{4GwoPddtY@Xz+3ll<aMMC%M3F$#WN$mJ)bJz!kC7w%Naew`zX3{)b7ei_I8;
z`e<pOGZf^7IlI7<u=?tcz})sYr5pN(tFP}Cf+-p3H6;4SF{%f{l=EaQmW*uXJj5K}
zPex<<)2C;E?#twyP%*&c%80)TL_ub-jmE#iNdP|s6Ckm5!CRAZazu7qArvxHW3n>D
zq6UlGpuc9`#$TT7M3FwG;d6Vt>fv#eebwtYkyiEuhlh(;mad%C+3xc<X~Ms_^MU9-
zQhTY_8E4lJ9@~=vN1ob&2?MLKwZjc#rEPavt890MxhTAMvoP>88ACyg?N7Q+AM^~5
zC@=j;SwUeL_P8x7P55}=jBTw;vm+b`jedT9f7mqKUKvKMm#_qr0{#7);Y!#Ur=SBO
z4x`00CC$5c?;<gy2yX@71Kgy&0|OunOx)ayklkdWZ^bERjYCv*!X#%Y`AKSN^+H|8
zmg9rSiDrQjkZkv94G5^YDC#2~Cy0;9f?QDE3xIxD^qPg9zO=4x8f)u@m>4QtumRYh
z#{4*N1tCn9^Btnw;AM7b63}CTLxE2oySe=yCXK);q3;M28}_71Z{4{*!am<Hm0Uf)
zk<5$RZcJ?l+4*Ep8G{8+Cs4?3`q*qw^kZR`9wFQ%!?4h5kZvA8g$LBBre1N@UKQ-Y
zl^10RQhx&%H@Ew)s3Sk{Sq8S;@i7HVkc6)^F~g2z`KBF<O-NWK*_jP~ZAfsg)bQOW
zWEPTeZ9z9_mZ8~SziOllS#Z<809T?SBXx(S!B3b(Tk>o_wB*O9)B=5{^9Pw-XorWk
zeq8#cBJSzzq6T5V2;?3Z8A)i+Ap}Nhn1o)@@Uu`2L)ob?tf{R9K$dPAz&?V)BsI5(
z&sq*)DMmbaxFur*6I?&8ge{X6Vp6bhz4ACWOg+zA=<7(NjIxBqxv?|*IH80Mn%6je
zGX;whr0IbQlM^I(R=?-#v!V^Kelm?_TTiIf2p=?#HPP0NY8b#Mw|HofuGM{51yXw)
z3{aagR`tUWHH#<Sz>_Emp5pF@)oTamO5L|_u)-UJZ$tqgp^BbVgKvbk?pNK*r%|-&
zi<-dZ8j8GEFt^9+HNoH1?Z(5#md+V9+y}SO7M@ee%F5+To5=RlOoQ_*>LGmr;dxyU
zR#5QE&0TtZLI80((y#LuHr~?j36tiuC|C%wnbG1RJ%g$sDtxl4&*Gx-)0am%<lFeL
zTj8%gKt&Y<a^>rLkcOrZL><g*WvcQVIN3o}wipr1sTMFvppd|j5SDYpk=H<<{BLF6
z`TP+6WE5m$A|nY>O-I-7tL|iKZZ5;3JqBA!qq;LQ_d_%Jrz$KH5I$ja4i67UWf8Q7
zFb~1f0S)hKX(=>p1$)|fi4*6~-$w-p78Ir_zwKfWmmsey1HPfuf`8@+zZ|9yS3q5A
z@7}%GHz^^!_AE;2H)^tB9^SZR$NA9*?lVA_xw*M#j6=CC<Kg!~B?8q2Ea01>qT6@w
zgur_b^NQ8cSm0=1!#wPYqz*V&hcp0d_ay|nQ>RX`vQERtft~wnYD)ThXAq=2=)i)a
zu!IHnSfL<^of5!dQh@A^Rhfsq4#f|}z|W7if}kwHwW{Bc*bI}SP_5G56t_!oUP>QJ
z3)_v6JA7DC`e&e@-&ys7Hka+F_?p*jO1goxh<I?R>zX6sxJL)V!>w;Spr&ozdE~Sm
zd`cwPP)K~NLI9cG(T^k%1gs`Hm{l_3Rbf>{IVZA_HcIlwji)UWHh44MrJZPpA~gNa
zk_TjD`Cd;JOv$0raF@`M6O9a_qIKE|%nyQZ(S1lw{hORT;Jn7e9$~TRpS@<;g9|Pq
z;1ce}`mBZSw<_R>^9l+|>|>l6Z&zOW8v^PrR6<u{^_+s{NajV1no8*V5wyQ4g4n}r
z+VlAS<F^4c^uCo4F1@E~*oeXqY>#GiUULhknA*KNTV&zoW<bSUv<i%eA&p566-E71
zUh~VGXyN-7><e_UvbTMplo7I)P@Q=_o6N1x7?7ou_yDJsMaGM!VS7VqV!9?KQYIYU
z8@J~O2?{=ab#G?D6uLgp<2kcl@SNfK0>6L&6bC7}x?sF>P06N1)wbTuYwR)`KU8lg
z{a_@|rJnk@bEwmQa&sZ1dI2~9m>>p!ez;l4>E{ZHIw0qhC=|6eHxFGjl$3l84P!9>
z`}gmVqEIfWiz*^?CS;798QBcbA0njG<-@bf>A09a*qtf@-cRY|D|3X~LKlC{qV9+J
zo}MUND=0IjjRgZhg=KPX&MCp87RdwEM|59^QNs-2jBwn3iVWs7FxzyKPPG{|EPJfx
z&6_sqs(<3V`}HZbeS>W=`)`SfL3FW0a6;SAh=Q8wi(|Y>NyS|oH?b)eIAm))m%!?<
zSeWX=ia%&;i%d*3Dw%ZhkyQ84o;T|)gF?51nmfHro#V`z1cVc{kM~b7FqDBa{&qWW
zBR7iZhQH{BgkHHohE0z3L$r2bJG+mzK^p^XKDZ#923G+VSk5i21TY9B#u%a!)gPRY
zFaj6^^&??o6YnGSFIfW+>h#yBpmZ|X-G!gHIIPq8;s!FwwD<4*q0^vt1Ql%$6_o?X
z6DS-Cr>MO_F(wTED6M1E4O%mk)H*++h61N4A>W~aiIURW$EP238x;aP1Hz!}?``KG
znG*ou%9Shk?@t0G!J3<vmewo8%*w{WK~z*9Z`%g|ZHA-`QzOKC80t{1H6HvCCe+V<
z(LWK*CD_iFQBFktobrqSwjJJrn(*afr5hk6i?ar*Aw@vNsM}TI10c@vI?SIxb?Qqh
zZ35V_Kc{IA0xgPFrU0!#T9EppqodIg<O6^P!JJtseIAh(Zr+_82XCE2VFVsS6yd=t
zb8zQy7{X`=pC7Kp1v~M^iddEGDd-=B@fND-)2B~iN5RSfY_J@v)vk|~3Ll{On_aKA
zmsn)B)A!&r1{)+w4FoLd?e)Uv;>}`EClEM*IHGR$>_-Gdksr><1jB7`AUu8g!{Vg-
z2Cxlom>q!C58Y2fdO=SMT0QQA_QGW<c>Q_-ymH{V*!PzA@1ym}ju<M-VU(F@s|G}3
zu`>~RVRGb#`mL+03r2<ASyx*Nr4X_UuzDK=UkUz|bRgJ()ty&v+O$OX?Em`<`G?HR
zV;8W(_neZJks&<ANTf(-s8$ngi>$15FkfH~*3|s+l;G2A&&lY~e7Dqv$;PO;KYhxH
z&%+B3h55dE6;)svj36%$4}OA{qll%zyE-~q0OBy49^*1Q&B@7$WsOP@E9-~5#qPLr
zRUpT2aJjiT7PBXDlkDl~=~S1k62~RPj==sYdAT7I3;Id0FbcE>=!BeDH;{Q9Bi!^K
z?}tU(pHt1LM;8&}h%E)`yGa-w_f}vvCbJAikw*a9j1<n;vPqI*4^2mG?O|b$RS8Zy
zZ#&fswT+}O>)?l8Cu(YHBuQ^(+!nI&3alkB6iAn2@TrJ9@yrllk$Xbk57&JC^5rL*
z3*(uQDLv=&Up;@GfEKiZ+k~*-B(i}lii+svAQOqh+}#GB%qZ^c3{7d<n~M@!iviH5
zO%Z24CM?cX1^d_A=4Js4nl$3%V*mvv*vYz#juVb!hA1dvuwV$2d|teGLAY`fLMUXA
zY=%9cQp4`x|K8nJ7%VWJ<u7(@tOCbe0Iofvdue(8YI6C)@o(*CSB7S18`U}VY6Ua2
zF(dG4l*A8gjPxm8LFeIONE>-Z<9%Qyz|9h->z=UWB5=fKg0aTp=1StXl1)0TJD?fz
zzkT~wf9VAz?Y4A1&-cR%Kj6h#oa(Erph4xjbMElh30e}*IN+pw;H;v^S+zV)<eLRe
zU<vs|1&yynX5{!zG+;B3;0xe%sZ_xJj^Sx+DlP^=aJpd3I}r}I2|<dD(dGoXwH;-u
zkF?KPPz-|MifHEU61WORNk;Y*f$FG+?ggTx<*x@<;O6xn6vFVf-e>OcLn&SN#9AoA
z;L%*FL82?N!a51kMhOe?=RYXc9jW;0|I}^%|AGe+qRaa0KkkSB7w<6r-8hE+&$$5q
zdb9n%c=3OK!G9c^h^I@SATNmy{PQY=ZCMoUBbKNVPp&or&=KF_wg17okr`k8`yT)Q
z!^f_7;?+W>!s<JkOHNke<>@*7g@Qje6X+!h;yBjr6}>wsKaK8%HR}8uIF6ICCCPF`
zBqXvve2_6BvL2G@3t3p%$!1E{P0RS|KQ+Steh~hzOxb^55L5DBdQz78=f+uC+!Gw<
z=g$fW>iy-u`Tbp!rrhhF)lvoDH+~h2U<}YreBYsOEXr1+@U%ujljFeG1KSl&SzQ$p
zlocFec$z*?yc(L%!TO={uS&A1B`1gXF#Gk7rrrOpVve3aGP1tiL^2^>UIIh|7x9wZ
z_3Pgv3fEU~+V$6e_{LXEpk*l|c!SH@VK&_#kTWf)#Uv!0r+S}Ri9Fj&E8@iv7V`<v
zAKu4O1rk2$$}aytOYC{F$#cM2XwHb~ULd5JyLT<9k$|tESn*Qo9DaXjE4DEi*_1D!
zBe2<#g!B^vPhw@6G`#^(U4l*#asG87WN_jb0EH;<mofz*{J|%Nr>K*V^<FL^4+j$R
z@Ft#M>SzzB1lfW(@*t6^r_F#=%^=L8)%3Xy+8eGEXc`!_B8^fV2$zoga|MC*ho6cF
zl29~)L>VRG{X+UMac6so@o`y>MFAE9A{XO>V4^`{ivG`MNY#W30Xu$h%t6B+5avq-
z8sd+(um6z)6aB_EbdC%{NcdWTpMO&L^I<Yv&*Cm9q8;LR=gIX)lRyKl<MOZ4X3h(4
z=Bjw2*9m{#sos3F-)B1iIR5yx{Xaj#P{zu}Ny}H?O+(AX6{qXOTpnMOj*|uM{#MHf
zGceS$s;EW$(eZ&le`KIYJ&LXveo@g%mG~X_h1sh=BSCaU!1Z(}B(xl!X|e^XKacx;
za<?+@KHuVCgi6x-`=74A--!cCp0|kh4e@j5&H=8qC8ODk%<SQxU*5)17ZA~pzV-Zb
zF`D&@{q^bf6=j@^0eL+7iZ<T<df)nBlS5Y80**LbtvR1goLtj;Vtr6Iy8ti(vCRx<
ztUus`E5rl7yYqv_7<fPpdh-4R;+E20u7B$1AM4L0sw#Y#3d=-Na($IE9D-Rn$7+np
z{zjBEj7y9v#KdS({JA56#^WD}w&6qtspcH~{CME{Fzr7C=9rq1;T_FN%+5Qq3#=lH
zU<_!MjFEB&vT!*D!3+JFb>grSW<xH9#1ed44!>bIE#k|<3!}xU{azY7P|i7jjc9(s
z%Vjwp&avyyA}<h~^Tg|4-*f-+=KSwB{(r|VMYY=)_~XK|Omfhk2(y2zR0R4CI8m#_
zOF<#eap@8tA0Jw^2$9~n4Hm1N<m4426_{HxV!2*<AP1%jW9N$_?D3$bwGr57c!8~e
z%SpxjQx!!S^4W_QvYF8|f5X23VFIoeyolDs8*fh5@?d8|c)gJN3C&Ls;8o`&ade8v
z`U~DRtXXtGL!E|Jd5qtQYWME@5Dw6FF<Q6}D=E)=eQA}A{6>f%-vfgrTE|Y)(cO#a
zeCy=}8sXhI@$NW$l*7;zGjVyy=ePJfAQB!@n}Wps;n%)zI2uVzj2CnV>`qJbA)YQf
z66>>hB>(C9#@S35Hz5@VdL~ywWuIOY+kZ&t`GKlhorGIF;y+OtLRFwLKMVW*U6sNw
z&N*{jV)IR?@Aw#5u+6;Wzo%RQ#vp$7L43y$*{TwUK?&?3vd`x02>Kq>m*v9+pAql$
zg@CxYwB*aId@1D&#`?}~A_ly>lbV*+0xN0A9Tn8N!41&Y34%L|ETf>H;Ds_T_y(9j
zJ4%)o7U^#@T#Hbj9vc}k(9!X2q{Q%?cagf6pGF+LQD%h3)4IB&zh-AGEiFmrc8J-+
zFzbOW?`ZI-Bp;jo32p(&cQJHrIR+Pn6YurP)kE4VACcqYSuu!S3{v4}M}(@PjxOmt
zciydv6T+|AfF>Sn_onqRa1}v}>nL<iNjQ^c`tv$$Oac`b*x6Tlo}WVxsmYCdIIsmb
zodo>=R(kMAgb>)EP>6y43b|?qJQ5rk9uD)l@2jP_rOHYJq!hR*pq)PR@SxbW3vG-`
zh?=)<O&N+ma&;|+2gGS*wh?Wa1{C6`BfcN4F8cV3bC0(OvSA9eq2VFlG+P#o0q^=6
zJ`t{SAKi5WAiyu}LB=XDkY};%>(qXr;#UZ79BxfVsUL7UGKSxN?v|qk`c+{S$VuM2
zd$(l7^__bD_)MRwPV{cBc$->TiHnMQhRnl?addu9-%TEI7}mS-2|0WTuV3$ua3+Fo
zJrP@7ii*B|?S+FaE%-HL7?g64{5C_C?&ZsuRnqE%cml|-HR_4UHbL-RLi%OR|B6S5
zJduq%!ZAxXSqWRWvv#z2Fq6GE%J7Hg8T&MN#CLCkt)i>zj%p~Wl*ZZ62F>_nJ9H9l
zsFt{h&<l6@brd1q@TOxopa^s5mlEr}Z%tTPkac{%$4f<&V2<H&|6s@EEs_C25NE%f
zt_z|S^=U&HM7OsQ>KfFy6IG)_LwZI=NihU6xv>vf(+*bdTLy1wV3;PV){Qmj4q#LS
zW^hSdPwG>-P}c{!(=A8jPpRqRj0h+cWgXd&`OysN6D$`m44k1YpqU1lrv)iYDtwe}
zGx9U0c$g^$D+TGrJNjVaVG6W>*HH3WSX)<rq5O^KyC@_9P}OUVS@IF9E-Rb-{hOmf
z4(cOe84mKg191h?6|!_9(@nIsD|r#8GHu$l>4y2PqnB>oJ^>Zan~u<?AX;!Gh7_UI
z;B26=k&)_$dt$C<uu1Q55}WXNp}DE)8!=SjY_Y}OCtrlA8>aJS0VH8qmP~*Tpw+`=
zH`%NcioVTjmtq0~14AC@*`e^6hJA!{tg!pBvTNA4Wg+m@25V`5c}n9rL%2I5haj$`
z)uH>W%^#;Z_1O)?p|&v0EsG;+XblhvUbA0UbWOffWsE1@{2V50LyBI2i;|qjm^w4F
zv{AXa@%A%2A9yp$^g~@qTY$k;>)eUlP#AyS$-lo_3?FRSJs)v)^qpX#C9w^l0sl<W
z6O<+l-$T!+s-^eITH=Z&#Yzb@3UfmuJeyCf9)Z+Q3;_ck?W|VaOU{rH^c*lrM?Pia
z<>t17atot%_{`+webqgoh@0Aam!IPHiG~rcdH9Z1JujfwNQe;vJ~7SxUcqThmlbDW
z`9<OMLTt{$O<7s5WC3i1+i7(8G{tKKxaqY<o!(b0D=qC}11-^O99V$y^Yc5@EP%2Z
z^t!jvP(h%KexDRwgIo#O&@&qFR{2t)VXA~x+BK-M@HD3chZ}^t{l(pQ;<oQF9^#)}
z?*z{qeW0<4iAka;VUWADf*or2of~Hm_=tE~&J0R5Tc|!N)Dm`rBMLE2B7o({&PJ40
z(TXD#f{k<7I#8(?D%~LW%-oprpV^^Z4=7Ftj~rnN93LH32^ZWJp;A;+BXvo&v4DKh
zW0*;diI%nx^_Vm_oQbAYXStKyvz?@YjX+1YHZoBzF5iyB+K}=3EvuX@jAl;!{V;ha
zq$;whjel03C-nmD-w#Ei^if5VVhJ-S=2-cXXhF^HumY|4Biclo1rR^C90<?Ut~@aG
zwF6?NesmAK5h%ESuidwOP)qLW?wN1fckS9G|G$kjgxGq^8&;ogUm6V9o8YWWrYdad
zP3Mgw%LuY@&I8w2At9kLRJ#L`7oLJUL>?`M*LWzvGO#*MHXaG9;uyM%DNxG$$cOuV
zvQAH?a-lWk;~29!qPL7Nb5DyxqpgS@Ij7Pnr5e?DsrijD)*&-~^X`<D;{C0q5KI#w
zL8rOl+%12SE|!DK+x1wvOtiH^=1Oye&UVS9!%$y8;EO0=Kv^_;Tji@2i9^Js^|HrM
zpQ^PDke8RoCNqVPAD+v>A5dxuW25ZY3mhC}3c#B43dRJPa)N<qQ4#=Hk_=ucCU$nY
zIY~$}I1fVS&W-Q&^%k&Xz_orb3WYn7OKY&up($RA!V-%fj)K}5$gtR=N2Cu&R3hli
z&G|&bB_V5dzYVESt<WiWY4+kpVc}A+kijEChp;d%n;`KYUpr249C8<g-(^@)b6&kV
z%-WB7CbU?R%tK@`padLH<+Qq>V{dO?Q-gxb4Gw|bn?84RbfC8(XH*G74sMd(J`5$7
zqXkYuHF(gQM?PW~9%8-sg3bIvV50uryFEigM0*YDT|*Vb`EROW${10Hb#B6N2#xbA
zlidQU!cfR9Fhhu76o)T<C6P!C4Gr)ZppO-TIa;QO4n_rqCbYas;k2XXx}QBggmt}H
z0F#IIt7&+LEi6#Ju(q-~2B3%_472&kKxY88W*h60DL@}w<>_8AF|kVTGxQ=mwr#6n
zXF9(o87GT?2&*DICq$Rs#*G_Mr#|O+?g%{5S_IQ!AiMY?>=UmFg8ODEfE`p<%sVXo
zxy+-}Cy}dZQQuAN#WqinWooWaNAFU=8cGW}9f|O5j?$LAqE1+oL3dQ{2|2<Q|A5B+
z&Ye52_`F`cfU+%@*$c!ukg8>bfLp_!eTX2>D3R)UIc@*>lYqCTkWn!L@WuCE?<TB)
zZ1qv<hK7V=l=b4p7jjXfP~>a&^~)d!ETQp%GVawi;KWv<(Q%Oo(uWSNudUZOhYRcP
zBAPom$O$QW^aqy!T6740n=SkPEK&Pp|1JL${SP?wNiv#`oD=<+fN1=Y);I}(&u~4k
zY%pJ2bws-v5;1(GEJtGs@&7mitg!~gm$`{f{)D3!VeCai&M171&9M^3Sdiip5{;hS
zj}#4i|G?A{1T)ke)EY6OpCs3>`*|8=I24Vub8=V|(|S>1-x7k;#Uf~&=06t~Yh%;d
zrJ=tg3;Pm#*k-z4ELxk-evYtNqQM0vq09#)$xi@F(77$C9p^^<VfubMdm#sE^*^R5
zRCiENsTvt3_S?&Um`5PO2J(D5ZVV^oB_bQ5>khjx9(^c;f$F3X%I&ZhfBEud0?`?t
ztFGRO401$m3O#$oxp1fp;R0;vL`Cm%M+Z^-Att1z230?FP9T?GMuG;DNG_-vZHCPq
zP6;gvd?X_}6S^0*<gQ=8{=XBAZ&*t>vk{eifIAQk?JZvc)c*R$kbBMHybN(VCh<}n
zB@XE#Ui*27AzXj`hYHbulhFTP4$nPul?d2K0rkjL{sz!z<l|d~rzy=1;U(_AHe$;v
znpE~sQ!|U*_D}dG!mx({*}&oi?L}JO3i+N!LqtwkM<lzhVTp;OzheN-No@)Oh;$=o
zxhzk>0jZX47R8o-!~bL-osUcb-xWTuBY}$+-E}HVK@0JJ)})Ksx`;!3Q*mOwGK8-f
z+z63~c)g=qSC{bBf7v%)s2m~hdW4LWOk^c>oVM9iQGw1DnI?Gm*3e&DD-KJ7EeQN{
zJl`ibdUo~<=e1Rsa{~4w)S?ljSHqi(w&uqG^AWpupTY(v!&;Vq-80{3KIV0B+qr0}
z;J3GD%R|)I?QKcZdufvkJ}&muJ#Rjf8ZSVy9iJ9_7V}WTR(<-r+5?(nJ*H*M2u!G+
z$`y*kGuK+N{}OB_5MS&NGwGthZ{V!6RiN#4Mr#?)v6`HoevWGT)2FGasSK()>Cc~|
z*;X~p`jKx)2)uDoIO_`MGM-&OqO%Ig8WIdaLePhioo(AM)N6tUGF*{=dGZlmAB<?o
zpAauSrjn%>ZoqtivLQN#=YhD;<PTg6p9fGe1qB5M2Zy(acO#{sIG+2&5j!n`Fr&dv
zAX8u_l-Y4G?G(5)?h)!QdM1$sh|5CkOHvRU<Z_4&;O1byt0Z{2&vfQzUi__0?+nJ)
zL<d%>p1Hu*5327h_Za9s*!;`0oR{VXk1@_U3pmydrwODdudLY(Qw=zJB7JG_;qh;1
z$8T&m0k8+&h#UfrIm)K{I$0E_W%PpB?x#i0TZ0L;%+68cHQ2+_{Jhp|B668C&h1^C
z2GYc}&5=cMr?nB_xc5y>iM9nC=b?9J2hJ!Z3cjK~XW>7|m>EO^9)LGw2{h_{1HnEj
z{mwKhD#{hddu-l}&XeN|3=F{g=&SX@NlX_A#V0)$?Yl&mW=r8!biMY$RdMSU`GnbS
zG<?7up_rzHBh}s+K0$8v#_5cJlAR9IVxVFBlaEbmybD`9Ui0IqDrN$61BWbe5OLt=
zuU{jP32`}X5bCtWcV(reI`A5#xuL|1(?A*RV_*NKTCH2)ryONEl#ssjHb;=HLy1Dv
zhKi*tJ*HqLq_01ri3DbOm!D}7D?&2{Sg>!_>{T_0a5!x%+X@DFIUXF<*7qCh%iaBL
zG>#=r#UJL3`ykb_7g`)4iU}ChSeSo?bVA~YBb}=6!+?VBIh?LJSUwGq4Ss2YV;lg5
zfJf@py8xx^XNRNC9Ed=#V<9pz`hzqvcU1w;LIe9kaIgjlAG(y_{Z?}pNAryUd3r_$
z(fbOL!p4YY3rTqZmot~G`3ty986%vFlPp&a3EMD04NQqZyoB3WdzkKT;?Oog-_2KX
zc!A#q%_nyo*$*7Ci>>+L9IHF{l7X4o2z{c2YQ<+(=Y#EnK6`-Z7x+9}_lfTsRz05P
zx}e}-PEPW^?Y7ln#zQ&mq77{FdRr%`+2p0JgfQ&an7`HoIHa@2->jp~a2iLS;t=bj
zV_xWa)&A+<iatR3WEH(UyA%A7Pt}2-XMfkQv#8}Y!NsSlDv|N--8)F9BNGt@JkG0a
z7~Iaz&UZOSzJQK^+=l9%XAz|n3((Tnj}N{GIKxZ*i=qkGkuZc%zkt-!frYU)|I-5u
zWCA`>m<>%(1Yo0S#3y40(Jy|xSl<S=EXy;tz=p*v`hsjTk`ZBU-Q>ywDL!6a^=O@l
z67?OXDA8k`JXH%VKYf)#gu-sux&lSxg>yY>s*E81K}!tDuf+p0EeTT7Jb|l&SOb&E
z@rz#2q0nwtc1%n}WVj}x6d%3g-1!qH!Vt%>LN6|REteD*M}c3qFhaF!Ds*;MBhjY-
z@uG!+osUm#r4wTf!whl$H{OL~adbi8x9m7X7);PqhjTh12*-ZFW^+WMgEQ${uKUK#
zJ9hH!xXpQV5$+DiKgW3do}Y@r9`gl}&|0$ipHF~2M|b+PZVXP#X=!fOPDuw3hBkXR
zLaodihuj0ao}p&!9~|UF@WK)$Oh<&)6Ji6CJzQUvJ_{&}^VrSY@Y20^Zw{&rkgkc%
z;Zq=K>;H{_-(X@m2E=px(;;f=6&Nw@{dVl#PMqEswLkL&JrYGrmIDwauIp@(B(4gn
z1}GJpxUMv&)?%WAmzIfqzXRLMc}WCOXu5Q{JJhkCG2jKD7BE^@VuPdQ8Noae%A9;9
zHi%R7_Bn0$kS|;BeI4KP$a-hPDo3BfhE^3y+n$Yz{JUIQ2*Wa?np%%)!iQdXLl&Ta
zpb6GY`fsyxVzc6;Lw*Ugr)kxJPC5Syx0yD>ac<Zy5L^^jI^o$S#;C<~Uj{Y>JofQ7
z-Z&ebLMPh^rRp0-FxA53B43EZ{ss_d+i*oh2fYDUM+eYq7Sr_I1%5H4Mx4N=L=bkl
z1Q?exaFio6{pL9k2W<Y91Yh_Z(Kdpq=H^N^{7g(X#MgNB##pJJc#l6z7t+x%k{i5S
ze?No($_958)_lL6(1V8_PYNW%!m8s@VJv5%z@gf+mFEhZT3Ze1<G1<%_w->aq&(;&
zh6>^dm->ZwKjx?k_*We-m@lxOQNC@+ThMGHbyweRc(R?lbnS_t+o2C;T=)B$z4Ma)
zB9AOO<{>IB-jQVzT3EOSiUMO3GHW|xB{WR<l<sB3!s&xp(UxP04#972ZA8w4mu_xy
za(9N-<{FF-PK!sTr9OPv-_K7M=VaiN3nVCe8yhXl3qwK9UGSN~Cs+vi6b}W{7Yu<a
zDk?+}gHunn_0Z5SwD;g6u=W^uc-n9dc|MgHut1vZ4LGMgJgTuei7xxsuW_0r3d5U0
zc5s7Cc{NMOb-*#P!P1Ee?=if6k>6mVgAdcFBacf*Zpjx`G|ctV!UB>Pq*&ZT2?#y-
z5e6J3USsIj-}$rf#MGSkkE5H8g$2H~nyRWKOtU|WO(G8NaKOC68Haccha1`^nwp!v
z&nniUYwydK2pH0!k-o<Xp&7>Q-1Y-@_V)jSy|)aja_ypp*|sQ(3I>R%2uLG>7=#K4
zlG1{NBHbxSnkW`1B}hqkHv*!lgdp7rg47}f79HPs^qlkk{eFEv&N*wp*S_|PxYo0t
z=f3AX=NMy-DYI|rdr;I!RMJaEQFp9n_0*Ll-#v#8-NmVa0Aj`p3D-HGD=Ifp{tCF+
z=b#e;dO^iPqwu`Ed0??2&5o~C0ZS38OqkNl=^B1^`Bg{*Vi(tmaMUc1XOWa@(JI1=
z5*aa3v?)!My&1PW<{?wG3VSmwb8Hdvk;+`6hBEsO#Dh372Q9VjJUxF>^UoGM^VE6h
zDWd#1>-{=97^B5DfbjSO3C`&G7#z^Zv5m~llaCER+qa5$4PJM4a_TB}5tP~I)J$pd
z<JVkuXW=gYi~ujrIHPFYT}2_qX6@J8X5;Il3=@K5QlyQx9ctCpS|?{ZRcfBzm$k)!
zu8C8%vAL|d<ZW@zV+T-LpOzmjIR4okb!1D)u9cxh(S(t{G&gRWT!#tmEXSC={)eAm
z_{#cC)a&l_e1B<LX})Defto}zoxhn%u|BzqnZ0v=4wp`}ldb7=r(d3;m*V}jjg#=~
zSkKv`R+}<$z4!*%a$vkFFr9Oc|3TWrv|}?LSx<beP^>>5D?##4Z7nbGdLH^*Td}B5
z;@LOWZ<4H6dR!bPFAqO9viZTs6WM!G{7*jaTZ(p(+UgRJ2ta7yY~iEsjLw%kxd^o9
z&z{AK)5G%FvfseO<Q#0~5e1$e@rNm9A;e=?+K2;5sgZ<X&aY3KFHt9|r6hX%7Yc}s
zXxdqIMkFN~QxBu-!3=IM&OgrP5Zu9XGP1Ke&i308a2&w0_ygW2Q&LV=_B?d#ko04*
z&?#x=idtWyhTRG_5CK<!;TWmMR)D`@&D&G`NCfM0R2LzPPIH%vUyFw_gXlH{Lglr;
z;(E5RA`G$4n73U+)RFC}$6k(A2wwwNeRVhCKZdIZ7vU4S%|`~Yj9HoxqMn)15r<d8
zd5|Ttj6diIvrTL(`19=N5>G+CZ32BWPG%m($VaI{J#O=%@%<;aeZpadCUxSR0rNmK
z<aRv4wmkzX2l@6II(pFF<7|5H$dNU)m|!=`E952PBC%)88e?CE*-jT=goCcGu41H@
zS0(Vq=pYQGIKG+<=$?}lYj&J3(k{k}IKu8U5~;rJfHWir7XIMK%v$(6Qgw*4frX?B
z#AN2-q-s)%PO;``<s9rF*e38zm7^!Y4X0>uEFtM=aOwheMWAHv#W%|~tj@>aV58i=
zeb1)%SP3|+o1@J%E#$=R;+rCoc`lCwqzm6hJZ>BREb))VC9!Y?np#irWjb_@9NLAm
zP&DufxB6V|LEvvVatvu-K0;vf*LeX-yB}rO&v<-j@A|D<w)(Gl(z7%bDIov8W%>D8
zFm3Z;TL)`ouis_ApH~|h)~L=>eXRE7`7nJwW9Y%+^`C3uJZp!;4~M@DHjbN3Y_CoV
zb2>XidY;51QapV9yN&#}bJ~dxrxaJsgfq6+utHO3QkB8>Yx}Rv)|`y=5$z<$JW@BE
zzlFZ39{W4VanvR$g1MGdZNUMrLqo;b{o}LK8$THid~sn`FgZP7TvxvshlVM$M*_!X
zXn6!O>nkpiw_?W+SJ|B#o2D$KRS$PN>dQ{4D8vo_<E)2HAZ5`6mpvcUvY(b#0BuT8
zibF}v-x!JLe+6CzuqSU1fkh{XiDfL0nEsE=D5#>i*&B`ckf*h1skoSD9o>dg<3Nm5
zr^+R2{yhFI=kHGIBh8w9@eN3x=$)}Pv}uFYfz~#)G&{t#7l9ikSmgZ4y?geYf~mcz
zXeknJvAOr7C<O%=9bn>uWE$=;`p*e+0|yKaE^=uC*dFX?`5$`=a79G*J3mht4jb=d
z&VfS8*gY0|cB1b6!?5Z@eK@WoX-24w&P2nR@7)<L*6NnTRdlq3d#n|#g-0FT1}Fo`
zHKfJk2)XtSkFB~3V-|<`!TgpYChTFla)OmrMgb@%4j<Gz>A(layv3*=;g$gSX5r?R
z_n<=K!z7lL`%H}>B7U=~A0IAs9qlxrlG@s?17QWLiP9wpvH=K~myCA6=HY7L%RsnS
zM@2>P+d<b9?G;(v0Eif62S-f{jFC<BaI;|mC7x#rYMD4&;J*-NI%I@6wJ+uoPKHts
z(ZEk0i_hkQ+tGb|HK3LO@?oHjNC`GyI*XeRWmbmXdhpk2Wd1cgMJKD69N?ltp1(FL
z*J7E|jD&$1?W(x2%(%L`#=6m}i()*N#jNbHIc)pNx0@U#nBW6BobCG<7=-o8w~a59
z<mKh1R(|q4`CWMHZ4**g_S2D4&v?;S?po#96<<r}Z>Um>GBmjd*qMH3GrD0U#(hio
zySr@vDPgm@!yeYIAD`djY!>&6OP13x5i(n@8Bcwc`fl0kPNwY}u7U!YneK;<^PMd`
zeJ`CvdlGl4?%s7Fp24muD${MCoxyNEwcTZGEu-G6-ZLb!K3<DyvgBvW{+3M1BGt*3
zwt_!C+G5dG=VS_1Ml<tg^t9FMjnzup<<mlcwtff<(SB3hrdOi$w9lpk?FLB`d}oa7
zZo<G1ckCqe#3*DT@OMYn?+}K<I1zMRN3op-mSa?*w+>_ww?2EK;;t?)KY^|)tV)D5
zl_9hv5VK*HXGeVBO;%Rm=^@@ljzkp|6p#^d!6$+uzyk#ml8;dXunB{1ACy#>WQ6It
zm5)787ff_&Xli0>WFXRjLE%!26t}gVhF%0miOuhdJ>>B&92x9r^d6}yM-iYtUk^_R
zv{S017Do`(DY|2U_zQ!+2Shpqk?a6!75EhTaNUCGN=Kd!VkcMx7a$#I>;v8c7t109
zb#fde`<ieMK?y9iqX1?Pv1vgw7IBxw#3bylH_=w>n4lp-7;d72rATFw=_Y6Q+{_0O
zh#b3fY6X<SI*=3S%lkp4_p+fVSc9;Bc5*)ygq5f~t5ZJ}!ODkCIWv8V1QrF&u3&hR
z%%VLv4E_k`d1rb1HcrmB=n+#>7gXd=zL<iPJTu6k5jz4fc4Sb!9Z2&L9dq16)PVtE
zNlFQ(u#s5oZq0XJbuN4P4J|BRDl6HJ_vvagqj1?>qK=gI5ZZj*Ei5g=n(x}-D&V{n
z<~*SwEnS8G<1oJA^%LV^li|&D&Kv9MhPKcxasHW^@nxAKHOo*x!VX_tSb$If1W5yo
zd@!d4TLeWP?*|b$e2(C&3#>B;T80$<Hs{C|e$tiu@%DZGgRFYG-*r5;yjJ=mV=<PM
z@grA6;dc07gehNDyWhL!TLHIE1zjw-a%hzA`FSq7EUOnMyBoS1KHdzuvdTvmoNebd
z9*USZ=|kUzN_rA+&+U>mUX6slm|Uu^mfLpT@#W^0ZNj;K+9o?aten<Pz5M2v$0fSf
zyRh?;`|05akv?Dh0S{nUM<FOBqI46?z_}id{xj!i(n?AV08qg1r<9%~-yB1E&W(=J
z(-cK`pn!8~Ad_y|r-1fO5p+=@ckpVV76`!7M>&fdVZ8JJcf_B2;Z2O_1h9@e`@^7G
z3HKpIPuLS7Iv*Bwcg3d%1q96O{s7Dv&y%n{c=^(4iiECauz-Z)ABJ;WqDFas9uY*?
z_Zmnh7w{oaJ7ImHP*Cvwi4z1#5f!7z3M$qIM=`9#ZMu@R;r-f!qgu#$2-}`g$dP8D
z@e`kU=!tG5P&gP#0K#L(gDK?G#3sYLEr~=O_I>3KYOX!ZB@?9_R+Ne32Z&6Z0XdHO
z=OQiOB-b}XjDt5J4GpQ6y1R)@E4<~s<iV&tPBcWS>$)-1-LOGrW^V3Jz^OP<&}+|=
z#5$M@CpF}ZSknf`1+gLg4D$i<@GIuZaBaJcx;RFmp!u<J;Yj9Y(p_Ut7q0fpls+h1
z2*0`u1JeWMsZL7wv=I1bo<yR2mUwQMqI+`NWnhB>W`BS=3|$ejL`1^TwtrI@{OH_<
z`X?s1wWn^hMvKxp<DVtj`Z|smo;m@>vf)R!p^gP}gm2v}z)_0OLFiHJLlO|6=einD
z*;ZRxyjFdjHRO||HKxVhsa`9j+tRMl;nq3W>E$8EZ<(Q+x75+aKSSp4RJYX4(wB=$
z{;xGc_ThgHTg{6znm?}iYRdDSt`ss=Npwm89?Ww1O*ltHxQpt%$#|RrIaxk6w8WI$
zjBKJkUvEuOs^13mJql}#_TUj8VfQx$RSj=vQjQD}-p~lR`lB5EM_77Xp(HCkJwBx1
zP_H?Zo<4rO4?sYWfLVZI&$o5O-DE<8%o2mD&@;`LL#8o|z&bl5)s+)e&@4BC2D(_6
zodZ)F9j&clEcrfcKoNv5T9!dtucCeguZ?9j8qCME?SV6M2Z%Z2`TAxI3M8yR0nQ~c
z4&#yZAy01zq9MlJILBOvFEQYpN)Z)wk**+3>F(+R!x!-Sbq4%|yC!0{Zh9(tjqt6!
zVZ_&`egG&l4^J8@U}Gh;fWr|^XXHsh0>W{XrEQu{IWdC@^&mnd;Vl_287bn3`2h`w
zI5L|={UFbMK`&U_+)PFRK+F^{oPG8cXFJTxfH@e+5xW-Ub&P@=CYPYj!&n^f=*!r8
z=8#M|cfx8n60O{7NoqAdbnKk$>{V!G2gTA5=B(oP{WAyZUTq2Vl1mqjbUfCNy4X%R
zTBUom308}<3;_&bQ}^qSa7LK@e(a8NK7gy{Da=rH!6G1K`*j&G`e&oXdgKDg1|-G8
z)s2TGve%_dvoTPhkm|7~j68r!8Px=nfjkXY%^<tNX(>I^b6nyXuCDu}-OU|FKaQ!J
z9G5(G^_0uRSiv55CN}2EFSRk3g0?t>l)JPVZaXCS+3F{j&BA%!LGADFehV7L8qNjG
zH8UhRru@wLU~yM*jk~$fFNbaUxJY%U<M8`kU%#CAV!pl2c6q>J$LZ%HIZOHK4uOMX
zicwEeow}B@=L0`+cYEc}oF&<~+PW%9zj++=*w<gVn2%hjI?BS(aVGZ9!O#6RB=1>`
zt}F||yv@bC_yh&4v{L{0Y!QoI{(UE*g}6MyPgWJQQOb&0kuzOG37c%b^=gxf+#Kh%
zDT`<VY1FF{RnmfsJ5*={Tfzm`S3VYRywY>+@>P}~qebfG1ZBzGo}=3?gZL#Jx7`_F
zfu?gvb2mLby|Bv+7c6NS=`7mPF9|nCUaZRw47xLq3UQX_=CR)k8?l&v*r)UXsxob<
zX-D3<_8qlVOHN`oy1wS3mLVMT;=?=^lZjiz%(kJmM#yqzlG|Ea1(Cgbq}bvqNJi3-
zSB-LEXL~W*q!+VGX6=pbv4LLNzj+lD)J`~0;cXykLoH{5u10K3!f7!&4x?x$055OR
zkr7rQTrq}GdqGQ!XnW#<^+o>jUX+>smM<`u&=If-mJB+0>pp+Jjc!8tegujfQ$S{n
zVR-kOm35HWK)Nv7Dt%|<urSj*{noE7+~vcY_lr05<U=h&6Ct|X9_40dP@&hR)M1n`
zuUTV79L5gXk3Dr>dW!6>uQT)IZp3L4Frt1Qk{MJO9Yqu-GIDYZI92YXH%xI^@~NJ6
z*+rQ@Jti$V`Re%(?zy|Yc4vN%G5+TvD{_9Zf1J^b@5Bk#<Z0bzL+y9&ol7;s!j7%u
z+9kW)R>POVcbs$V{du2xD8OlD@Yl>oNtI-;s_4hhzw13eM?**7F+OjZw4xpgPGac(
zkNl1OcYb!T)u@%e_<pu8JwJWqTb$rxOl7=qR6wY1E^0dEvbYtz>0MXDp_I@@Zw1Hs
zJG);}(S{D_4e9kx-!)wqT8|r@)Na+)@W3TjO0pgZ*B)DZVkE2df`5ot(2ZWOl1|Gb
z-G2PT!f&S^DPdXi-_P%L`VCkEtL|mjX{GZtwJHYZ-%H%N@_T^sG5--k3@;Ln`kCFz
z-^}?usu}=ivXR|M<ZzH>CFx>(WOu36qhqmEbzMi%*DU2o^@E+sz1NxO`(8bM;)M9R
zTsQjcXAc8xD=_5RK61MO)q8?24V#iax=@#r0J)3q^Ic!&@_pq?ER3Lao*^?KQ8#v-
z+6Wwb=h)JbWJK<U7kou$=Y^z!r@60`-XRA@ld!lKfi!EkYRm+Us%Pp~qK{kK#a$Zk
zJdljjCi3-zZkg5DR$cjuE^V(m;Ddya9o^iR)!}FfZ7`0jR*+aoK?X)gi*V{f2#ubU
zi)ZZg^djwdDZ<Z;KW;dcvr+8NpY=|os{gU5fDoHmJzc!iP_mejlT&kLdxz1N@`~B<
z>BCgT$BdEXEqO$}|FNBJ$y`cS)~|c`^F@MBoO|}HW2EC(H_`}w6WYtGl!#b)#qP__
zU28TrljUW)<5Wr)tVk=q>k5TuxM(*WW9BlbF$l7_JL-RCZNpRLSMqx-<O@f_<r>Vn
z?{G)@Nh*4G6?Z#!|0E9<*9zz`?PJR8Zex)XTzMHC8|Q8}`>xzie7wAOxc7^Z!Q2K*
z-GXBK=y&ejiO+9PUZMu@yZTw`k8^^k?#klNB$vqY+SE|0Ao+Jzv<*%nYdKchT5I##
zKkNnh9UX(}qdTA`jK8aWPi<43t2hYR;LVP<I>E`DZoOLqqDqUb78J&D_K})R`T3RE
z2`qhRmt4)ypE_Btp7xb5fL!IRucO|JDUFxC%6QJuW!LkscoyQ1zlUDHJRn7N_jb*z
z>NB+VJ#3Y+NWvR@%w8Tmb(wCL&z&4tCaFQ8(tshnhY#Cx@o*IvIW5mw<R0j%+zWv+
z>R(X&gry^RFg*QL2%FIh3aST`;~r}2bmZ*q7QM&?jxE)^8D}27dymsas)#!3z;$Z4
z|6kp=V=1Ck*2Mc>#m|>Ma~hfbY<P-p5Hw1NDTt+shW-r^th3$^;j0WA;=Hu9NkT^s
zd}(>5cX{$lbtSaims~P7;H-oU`k;-6_osHd3JX+!Ibv3sd*7_{lxSs$e_3-_0!CsW
zF;6=x)M~NlR6TdUvw3~6pqlfriIvH3RXwZH4{obp92EQdhrcz1ip4a_`zpOrYE<5Q
ziB~Zw-`|G5<L2;pZ^KklhN)OM)s3y@%^&upIqNix(yd+L7AlzHoCxAyP5W3Xe(C%r
z>ak}Xsx}q+qOul%LrMO7B)8o5)l(3n;jxUkGnnDT>u_pD&(h$_ZbeF;4Tq(_Zy(uP
z(UHB*XIS|Ddg%E+uVH7t3T{&$QRFb@XsRduTm&EuC#MFa@+b>&&vQ{^YIHAkUuOi1
zR0v7{`93DjL`6nI29(V>VwPqZ6~}C6ZC#fWlHxfWr&ch%9<aIjh1)j5T6kW7<Bc(o
zT%>!-$dQ4$H-{R2+MU_2c7P31D#);M4m54sy;~QQ1vpbG&R@uj8e#%5na?5Rb2oud
zL=Lm%ac5S4td@^J#uK_m7aJ2}(sR=cO7qxDh48l=+vw2`AN`=Um_E_P-@fRrAGzI$
z{0zC43D5Qu+T7X;>!Z^ZZ(e2@_s|?F{`f%Z4%;Dez-va{@k#wT%VXJ-$H<%#eyG`~
zE{XBf)tP}<IYPG-k`DX9Q>R{sj-9ja{Ya!?ea4U}nnN#X|Ew7H$Y3E1cbUK$81+O>
zsSxUf*7ZE@q`y0TQN*u3MzZ{Sd9p|7+U<8QQdA<90x0viOn*;K#qb7Q_i(<l)H!cu
z!MR43F}lS*^SvQ8zMEA}<j{4io7<~p!;W|zT)Pte>vGQTrr*nH6N2omQhpP9L(?av
zKNLRTMns5ZxAFK4b-ed~9u<Q;x4E*O$O?QjRS(w|n0mG-ckJc25#22k%kL#)_br_6
zzACpoaonS}{Ivnpfgl`^btWFz1V(ff4IY>j#JTM<3Vqze5m?UO{76v~E3-7PvQa@^
zUI&dGpr3cT-b;&nXIU}mSm<sFX0~<K6Yw$oKn<h}_)^{|8sAxMtoXW@R|%K`z=D*y
zZzUL0wmNe!TLJ>IePnwB{oQa2D*yV`u#n$>Y_C-2XN&B~Sme@NHPo0X*4xownrYQI
zymOHx50e9|!UmpBH1y{bjbt+DwI`f(-eqz?Z+63u<L&LTkyL9ULLDAlpFca)F4u`I
zC$Ajxj(RT`%}?jNs>(-6R5ABGb)e8Nu(04-)d%$fuLp^T9q|3k2O2z5=mMYHLl{G`
zM9lRN6I;)lY5_x@nbGOx{<oCbtGqY->$AAI=WQAtDf3!9R3{F78dhcsFBgKcJ)C9N
z9ht3$E$ph*Md6MKeeH)_Ef(m#oT9Q6_seCeRx4E~-FX}K^08djbj?nl6G<5rt}#Bf
z>H?9n-?Yt)Dk>wlRGC52hcJe0g|qOIGzLV0`a`_|%X~iE{Pwt~UukJ+o3ZsKOO|V=
z+WDQG<2`r!x000`8^<eDEO~0#iw{)UUl0<zj!BfcJ6=40jyZhq@5h;mMzqw;F!2fs
z28M3FS>f_JG}m{8dUrdWq3T)zkpWsYNyWMxE2Mv2T-K7Ka}`DvV-enzwCwt67qy>c
z@8Eo@Um6pX`arkN<Ohv!XjTB*c=2!jq0oVI#68`J%n5MD5jtmK{jFb#HaM5WqF_j^
z;U&tSHFu^b(yA6IWk>aEss)^}s=k+g=mq}0pL%cf_%{8KuR6|Al=fYRms{|I8;|H=
zJ}zycMD_CA`0?1acbNxXRKM}#4^1+Xe(3q&m5poR^|yy^z26(PTT?~p7z3k``1bAF
zGnen08S`ib7Z(pTMZ08ke1w?<QBLDfhtUoYBNx=iz(!5NF2IEr)jBY<MjR53v5j|a
z^^kl7)7}LQfKeWGa4f<1_Ji&lg-)<Y4^S5f2ZBvZJneHFR+~`Asj~Yo8P($p0Xu4$
za#Vy?<@0HyhrYgzz>Ra1UywQH!B1#IVm6m+33cYFsp4vpk>=A6jsmTnOR}tpKh(bF
z7Hj5u&_GDOq+`3xor#C>pF!?2jXdiv9Bha4f^<&=F&^<#L+1PzqC7%$8?$F`H>=#Q
zbgyErA~dA@!WFlhgm)pDNY;u7uLNMd(*BveWG3h{U9{s4%WtiY@qFVM=K1p?|18}<
z5)XxAUPQ!E#mLkcdV@x?Y<#J+z>a-=7}pzdH^oTJ>etx0--*AQWW%22B(%lKNd(n?
zeO$MLl`(YZ^BpqMi>10TZ@$Q^HTARhtN-cYdwGO3ebrat@N=&VtZXK1Pd4t>-C49i
zs^O4JPu7mP(;p;Aj~@AdT7Y1>^S?Eu3oX0XchSB+=Qq;l)}b$GQp3Ja^}i$fSA@6;
z+>hMjVr=EfJWg#KHgfeqE_LPS7~$tMY?1CCn6_O833&11M<kjMNymrj(|{+y!@~3h
zCT`0_K+<M-NJ^!a*gtsK(Kj9@Ap<?AbQ3NmP>hg8g|J7^Lj`%PvM3Ygi>i6BEIyib
zF;3n?1)CvY7gN(qJxUQI4GGT>j?S2E+qYkpmWDyO-;qK`g%2jLj&1`CSdd4NZcYWB
zZ4Iqm-=;2ygl451I9dA2mr$Q@^vY!D173yYWp;+L$!E0&-9VmmGo8pSZ(uVzXPB6n
zKuBSN5gNl?7-x%F>Lx*(g^$FoUeiX3Br=qgGGqr(1?x%#j%7=wgx?G#2EXDU&N803
z0wj$!^hK=jG;<oAG=avrKJXNAF&-8WS@0|YUxs{{hM7c{%lR&KedKvPZEcv?Pm)r~
zbAFqco~q^p>d*I@`6a)XdYO53=?dSY7kgZ;XzkiJ$>IN5P^tNn;d=_<Cy{%z()~>J
zBfl%h+?%?`_+!=B=@`#G%Bg_z&1e1ID(El8i@U77vft=X!`LM|zuw|*9P|YJg>aPO
zM@t#jRgG8{ZjPzKS-w<AP2zRqRT6PH4wIsGJUH6HC>MxME&Ms@N7P~hrc@A3EX&ux
zsx<Kc#0W_nWF|(}`XO)7ORnD!2`9Bf?#`}^66A{w-@I}60Nwc4tP{NBKFlMybzLAU
z3iUuk*b<t=!JV;O;J6fxPG+>gWpZ>JYy}fd74+2sN@fV3r5hzqi{na9XqT}bJNAnJ
zU+kJkuX@4LP-eLly+eq)frgDMYlfOPfoI#f7+<w?S-AFr#|8NHbTkX)GKQJ(r(yBE
zjoEH5iut^QWGEonYa-lkM6Um+#rLw!7mMfSpWQ&^wsui0To{{<(N(M`=DBSA0MSs?
z*eP%!&)B!tyhinL?vT$DeL?Fs8@ng`r(|T_vV9jf=+k3eOBH$L&Pnxk`-8d<t}`K8
zqfu3i37dlAZ2ZSLWJZ0$Ub_ok9^CE6eJkTugO3(yZ%faf4~OrZYxDr64G80X#{102
z2ey^>+nR!}SvtJ}s*H31YYL1NEkiTE3<6<eX;3W^DYCAV3-$dC4@sD9fdu*eVw(An
zk0<%PTSG!abMi!e-;7gNG^{IQR!&MRF<bsQ%VzYQ#_Pgl2_S&*79GX^8lL!qL=~87
zVO9p(h6x7k7Zhi_Oo&NU(W3=I=pg~B_S$d)=hqo-i=+P1%v_@Ad4jlLNnz)%U1&&k
z1-P0WcfuGoTx9l*@bN!9Sj14qbU;H$_L~Bc098-r5mHw{)|>|#AOxyu&6fX2{`nkV
zS9h(3@stG^*jj!G>PG}bF)<8u591_nN!@>N)CsZ*(a)SIA5%`$3-Mi=$eVuHgbt*d
zP!hPgY(di|LK+F?mDGxAAAoh>VEvKAqoI$Z$B}mirCOw!oXl#HZm3Iwit`-Z(%Ad=
zt#h{oXMsOhw#Xh916fi7vRd2=1RyeoJ<M<!sG9io&IM|RW4kMBOE_;c{mf~stQwgy
z4Chda8arvIBU{L20$g2IuH1`DThlqeYQ^hJNLI>B#L~m)5n!R3!3P)4Pt~}%6~#Z-
z-*#vy{3+w!Hrkwb9cO$?e;};`D8#^UH#@_wY|a1Jkw9g@>dlS!F%V(})lP~#(fu@%
z$%pDoy;Mp;VV>MvAUHz^4ZHH+m$E>R5ZroB<7Jz*K~=?=pt*hU)>2~d(~%8ZH*dD+
zE^PkDhfxk`L)(AjBoWAyZE!G!a3FRh|3|O@$L_cTUi_HSROtR>&y&-Ly2o6)2QV#G
zdLNi_`0NRemmLK>h>pTQcLxh|bNvf`SNOrFn)D&|Au;D-IDk1lnSj^<>*PZ-2${r>
zP1%^0k+nCI-sI3pD7+yiK?h72@2yX8M$D14fW7IF^Y(;wdgcz2pm|$eGic2Ph)<+T
zdC|Z~sI~-p)*%u^v^UMAYZ``lBKsnwkt1y{A!`J0Vq&O*@teERqIlji24SINCYt~9
z<;&6G@5l5w?s9|E<mJ8Q_x5vh9VI2k42*X!fT3Z8s;4oF7}C2qv_3jQ3?{e};N$PF
z3h6~tnn0Fb41{r1+nJ5&syK7r;81<}_jql;B9`{2OmKD|i>wa|kl;1255emqPE!$(
z#59DffY8rw6(hvAhc;H6(Lo?<*00O4u)H;K?)UrQ6G7twgGE1DkC2d#4y0ZxlHUKL
zbmq*@4<1b2hGiX}oE76Fg6~BId9`D$w2(DeFBEdg4BCgB&Mv!F6Oepn|Ct>ZxW?3q
z-|-JkTnG~S<k#Wqmc?)5^!eB5QgV`dVc5yy4;G6kJgrEjBeQLdVstxSZ(-hXxW)VQ
z0|+Z1HoN*n5H8Co?38A7+iRuK$_#k0NudCII-y-2nLdP4;P7cf+1Tc~BYy4(E{D16
z(xEPa0%chDl%&gjPqe8F{I+#Ao+yVaB#5?7h#5`7uq;p)(*6AJ+4JZ080rVzXmt1-
z#BgY4jdr&H1qV__aSJPJ99kZUd0!8XqAv|W*$g6*r9evfss6zeE?J(qptuA@GXHK0
zXeYbjjlFS{r@Emb697_h8B1{9X$+(dp-Hkp7=rq{g4FaDg(-QWdl_gz<SZC=tUH)f
zbfcdb6k^APDl@{pHr4$_23EEcrdBmP`HTz#m*|7}QiLr7EysQa1_tQ*ivA0F14Bu)
zq~nljA^OFUkzqk$|3I|qb0<~+yby~V<UM`BGP1Xp#^cB^1snx#7YM~T><VK<EkVMM
zpYx={NrPJ27;=?iKZGJRG<$(0LJER=>Kpof0f{49uBJ4ilZ0byrdbyP8E_D#3tt)<
zloW(2dU|wcSu4=S*WE2_<j=(mgVPIWHAVc}{=g?V_&n|c{(FS39PN*&LqSOJJ8Gdy
z_jjii0CWRuJ<e^PkYNxOrk#|onC$^od{uTBzAzye_94I5Ax<!1s$xt2*{>2};rk$$
zS%t?tzxUN9JXnwiDwr~YAn~od93!cvdvb6a26-N8LdtF6=uexG?$dshFc0idxTa4#
z9VVtz@Tc#}C}oO-mxD~J@VP;KV_D;|qBh$z10l(j9|P!sdh^>@9+w`|jq{i%b&K7`
zky(;Ny4IX1#iDz8V^GlB8i();1vS6F)q6jGyWhsaT0l@B(D3=yXYG!XLi+`j^K^fT
z?6r?}ark5JG|xUUq1x5&+y7bb^BW~!pdaV<+hu2cUrW_kL_3siA|!t}z07`oSVR!w
z_rQsv3=j|2fp#?~Ho$|VT{I023=r_BmhRfN4Z!x3xrgOgk$CXPaaIr^0EIN^`a&pV
zdOj4ArohzQ!OT#iLP=qUTM7TSlcAKbI~20)IxD9<P9nkn_SIm_Xnm48PtAq~rYkRq
zA?<U%%^>p#R2;lMI`TW-T^zyuA#*b`7m(NJ%R=+LqHmawPdZ>vT&ZcSIw$N9avu^i
zYicgDBBx1$1u@#*#XAe3FotO>I2wqCbFlI-9`jvx2y!0NuuYIHR{2zo65Y3oRgeVO
z=_Fr)@u_~eg&_(3%f2E8sxaA`o}X`RmYi&esYX?lW%DIjv(&vfHDx1=%0Z2~L+*<x
zK5&;&`v$ZDICD`_5ywiLw+P_;x$`Yd>o7>w%-cUNwdoR^`k*x+L7t*-RS8$bRI|$(
z&@V*iMLjr*p#Bqg4%(O3<55uU^SbmTje!8)C;TKZbexQSXT>w^0*?IArDmw4@-i|`
z88*-40Z{^Kr?`3^T?eStut(1#u+BZodGTT=Bs-vj33_Nokl3IEO&sRM#&>7zdJW;i
zjJsmi^AXMJvt#XcAWmV#^M*U%L}bub8k|UmHFO)}TpQ|k5PGa2U^`U&v+WT|3B*%e
zc464Re{gb=hyDT0rd!zmTh<<*+V8yUu3k^KTfzum(v7!?vK*i;ujjsJIq0|?<8;QR
zBy)6*yWL*Ku9}Y7jQ?cg#eFYMx6o%~q?=rey_BMqqyNwbtr_P8SY>=w1q>Hzov0Tt
zwn~^?XtvZ!wNZVN$f(0BfAmP89U=q}rd_*t&reU=+S{9hFab-8c8hjM<!f5B3FgeL
zDmvG7(Y9ty8S&2A!GTaO;65X}>hJBvG5QGdF|78=GwvO5`$M@K<?e>81*L5hm>z7<
z;GBcm-%RL_b8(G>?jr;|IOAZtgf`V~q~;KqXcce%iPW#T9lh)5S<Tbxu#=;Hgb-jH
zrp<~;*hBo7>6`=V7uR#g+Ij)>Ce!ib>}ts;i;jEw!CLkF5S|^dBMk_Ak<7J%WyC~2
z$bIYW9k=X0!daA`e{>r(eW=h{ljjdZ*N2G^5Q4=*Qiv;5RZkZ8+Kcx`5s?h>N~{a%
zP&Hs(@s0oCgyzv^JHxu4TkGvhhB~MQ-CSJ>xFIHs&^?;9atk|-T+s(L(cjOcy}+i7
zya;FeO-q%^lPn5h7>%x{EeZEeXs!Zcg7(0?a)l6&=wHX5wCYxKvc4xiXlMAnfbgqE
z#_&1dR4{#`{%k)NTJ}$kCwL&~`!hSs3Hc)SJ=$cNJr1JwTg3zhlZnT$qaFvhkMAwa
zOt>Rvk*<MLh3z!X6DTPO_JPl!ir>6#3Q|#oe)Js$*3#0#1<NNk2iCopVTC}5)J&KO
zc-dt?!m%)wTdeW$3?)^u3ZBy1+Ok!<n}fq5d=|+&8*7<o4^N95n-Bk3_b$HR?LDHi
zA}ZoFrvLiUSj&~oY|KxKX}(CXks|Y@6Le(iHE+C*eV~+L-RY<GGDh`Tq*6(MEk_2k
z-!>(e6R-6%tJ$l+PkOt0*bQB<GvqrR6e8;+%(%uIZdW_xYB(rc8#M1)J7jmMe^K+7
zd+jPM>lcv|v$hK@zh=CDs26Ff=rv842`o{B-L;NY(pfz>(POjzO1jD6g*xr#ACzku
z>x`|9i9t`Vc8`DWl-#%3&^tl0qTe(TCloQ%x7nMQm|lzUfps%IJx#-bBL-h~Qq$57
zM`jHg!~MSPEkQ8C8IF=ugM48HzB9z2x#}-fSgN!`VHhEU`w8VWhx^-<%5poHH31(&
zwuk8K3|>F-187Nt4O|(hd5Vd+V7T8qT~|}P1IQCl4@Sl(na)li6Ij9Lk{Vr1_)How
zRW&^^kY8AM6>uR+z-vdM;dXrp#Y)q3_vf0Lafsd^?a0YqnVj5+E~r;8U#6J8ou;xu
zbAtw|_*#WlYaA=2(h$OJ;YybTRt!jNgO8rrKZQ6>k=bl=AtzzdfhL52$jQoAUe{Gu
zbHIxSC-eD<S&S*|#x6tan3;px*~jQWDE{-6esHjGt=@ekCXld?z4dv^coI^MVUM*T
z4+LJF@nwvb7$r)iaM98ZZA9!-&=tXZf7uk`0K}LD=n%O$NvjpxafY)uW}iEI7R+sI
zYKO8qLIyEKgUCiKMg62sK#_w4rO5i%gDOVMJPHKsPQqCMtw?I_An1;hKSt;A|DpUR
zIwi>Jwg?Ak_@F7C$wnuh@|qdSN;fpye9@S0V7{+Wc_L)^T)kanD;4V!U7xs>I697z
zDypof$D<*B1%ESfLgBzZ^<56eGmmR}wAbXh>z!`9yE?C2aOtV!5%4b)+g$lOv;B!-
zl7eT%*u_uxAD419P0Kpi47a@Bz01w##+xLQ$I==GZHwu3S?`0@>Ljk*`kF7?e@3@0
z`S$Lj8x1#}yWilq9jXuwU(xVgZ8<l6mT4Fe0OsW45v#gC8Nnk(!3%gJC(s)?kerm%
zTlkMxSuLZkZO7q?OqTA9dAy7aI>O;UEdppKc=t!dFbvathEsjJngLwlaD2i;dN8eY
zZSSo%MT8^ZBJhjH;cdwt$c7JG$NV_VBFy%_7e&l_Ku-+9qvEWFkJJeSMvNLg2Lu>A
zOX8}IK89%T-%qSl$kA~Blh~=uFcdXByhjvVIzf3sXH5Fs*tD0N6wh#B)7Oe*FNkQH
z()V11@dA7wR*=u2Ap-40@wx{(Iy+0Tgb<I2;R@&*G)?32WWi^`N(e$4fKPvMaEf9S
z$b{+^T03|eVQV%!{=6$$?U1#UqK`g}Lh;=Xpk*T71LzK+VT95I%YM&?LNb{giZ}<b
z0c9f7*<YdOVOfL~jLd!+3rlsUP}<Ggw`cN7SGl0T!TC^i|LGktGsW0SKrD1TRviY?
zGWU9wcXbJ&`=6hmFl~lXwPEd&@{8RmL~DnOV9g=fRb`mM*fXqKih7_0(E}Mb%oc1Q
zyC&`N4-M@`j74ne#2MqU@axI>JD)+-3d1KGH;Wi{xDr^_-hLMj;fNB2j3GFph}k1}
zlO$Vx({ElSp38~*S)IFWBGb0Z&+^oIm0<6oP*{TM3;9NjNHF|}8!=GhI1>{ZOU^<e
z7(6Mjt97m^g=cL+?~}ql_tN#{tY=151Fa3IDQ7eE#=LLGNL)F1D<YRRfVRY?x7*u&
z`J{-j4$Jx^^<mn~XUg;Pa-E_U7(MeqaELV8w5q(yv}9F~KA<xvJTToi-52F0+0#WH
zjGcLBS`sL&G0=KScYfz&0ckd|XL3C?j6$t~JvC0dXMd4JdF>viJFm{#kNQk!rN(`q
zWynyyCN5s?GFcpFGD2uF&(<XqZ3FC{k1;+pskaBT+wa8x-ZSsLiUK5O<(e|$;@Zmk
zoi%e5EK6?uM9@J0NYYyPyLU<XxdGJ!5IkbFVA%jn>3Fh6tD>Q?h>|NgnG;=8c%~fQ
z8-s&_5)OK3HrlDM_)8<<M&D$j`}F{+6YlQr&g3pNG#q05WG%#GKYk$R13+qnr<_>)
zOrUQ^>KKw%f`(#ew{>tJ!@S>}0j>iy#q>?#V|cRQ8BgrJ3QdszP&x+>j{-m<1qJA`
zj4=u~=XRQ@5|}tx@nf=7U^Tkq#`^oCx@a%^Oxtv#lc?Kcu9X`gKuq8_NTHgIJk+0h
zc_}g;D*~*g$$gQ)EHX2js*8(@b)83;89Fu8)X)wY20K9&mEXO+b#V4T$1j7GCfu}s
zR4RrtB%n4W^^E&_Q*Op^d37Dd&@{<w@JYb9QyrZ!#86ZTsI$36mt{`8!6}yTBKo$8
z$=!h}S`QBo?K7?LiIIS98}7@kTkqjB<Vy@$zEN|54rsT5zCNr6(P|6`AG@Pv%4i(+
zJ_#zp7-p}b(T)j#U~vW}zji=58LyRqkD+g|5xX>XuT=sG|0ZQG9E`;C@ZiWwPtS+k
zmiysFdT6_r`MVVj{Pq>!`fYm`*NL`g%=yMwhW8S#cA{8x6SOy&2;mnH;Ec1>*_pj{
zKRx{o0h60RZWD{GQa^lH0V<C_!uM0<c?~?VPl1`$S$>vuwqE9JJ;yKcw40`^&X-^h
zt&{+xlRN}R$nM$Wyo2*`p4j$uvJ;+8w(;*O{2z)R*8r>lh&M&Yvip07nWcbO<_4ap
z&ptf+mc*DuL-fVrfFAT0WxX3x!LJ?r+0Ud<)?*EJ-sWxT>!;UeF?Tl1`&c3AH9#P^
znQ#OW)-)$i#^X?9gyRX`fGuzqz0WwT`sU;6ED=qSS}Su0p;t#-{82>>Z;epuTA#Qs
zBX)e!+=sVk>sz(l&ehY0dTE`Ls_3U+Xfi$WZ8h=Zj3L-k_Kl%qn6B#CICV*vJu0zu
zx{9OkJL9~N;$4C<Ke%Vltp}~cgv4)J3}*ZY#@VTK8+#=_di|uaj!Y|9vW*lxpL&==
z<(T)zFfsOr%NW^#6Qd-RZe{~UWFpB_6v|FdUqB{8rc0*W>qT&ZRl6~Y0TbF))nC^C
zMmyrQuE**Is87I&;({rhF5#VTgbcro)WN@Aacrfa{B85mE*?u04@`-B5v7{nP_<YJ
z1iR9f6CO{Fu3|?cmC<@Kp5V>S!LdFQzJBru6v<|20e5ISBBx?0hZ!CHtd<nVd&PD0
z{^t#0P~SyCQy%6@cke9D79q?F9oT!D>c(?Zw)LN*Lqp$;`L^VvKyhDNmp^TF`!>3X
z?(7q~$HBpYh6=wPUC3vHy|bw(3h|y#i1#c$TTf5Crys_j{fGm2Upm)e=Ft7@KogEF
zTBc>9$om%55s^$-u-~{(Gz4PN=Z)`ainz2<#DIcC$+~pD+^b}e^u_DvaK0@?#0z*~
zy+}rGiT|aHsx?``i5xG7ZWxek!d*HiKTC1(h|yD=P+K=`Qdunp+U(GIaM&0?4{XI=
zKg!!lag9NKmb`}*4&&FneZC#~^o@AMrQY~&6^hwb2iPc*H?l4*J>4fvK@l$*wyp^G
z_`k2le{SOV@25`)PcGs`{r`G{ANgN4&O*d=W8!ts3Cru>>(BpksqVwl`1iANvHxj8
z@xS{UW>DW^#G!%;l;)NsoW0+o35cJoj|MKSO#{DwZwsid9xym$a9nT;ubjUEhgrs*
zf>Y*r#QWb7ro4DH31E6l)CrTW|7kVz|NV|}zEF|@bcY1b(vL6#iI9NRx0Y6jtrw()
zd3lc7IWV8<{ZbG4NmSEqmcZ(lN_+5Y;!_yxg`#8N4IpV$tj&sqk7b151Yt@~y#CL5
zdjAJ5*#9lR@_vVIV&SeZqC}8}$yW9(&Qo?~P#Nr8T+K?qkVUp)cuPM&5!?g}u@y!O
zsqmi*#LPS=w8ec*Bxr46;cg^%*A?2;TK}gOfEIR)5uW*kfAPm`*|)r9Pjial75a#X
zgDj7UVEHlI9>oyCHr^wDgs_e>ef9U}zYlVV+c`Ke04h;T#M|7<_!}wThQL+=rq|6M
z!hmXaf)^Mh<1c-D`ZuD=%A;Yb2&$tQ=X}D5`Qs*H!Y0Mr$Y4nmq^;<r;FrFQU%x5-
zHx$Pj!)*i>4)Is1U@X8dIHU)2+K<1jy|&<2o<ULw(EWxl-M_cu|Myn$!JABRx#O65
zxQCty*th?#30_5Xv*bW5X5NVlJd}3|f5N-}Z|L9G3@NPOPp8l677#L9b$unS_xiu<
z4ZM|?jFyt_*4EbHm$=e<ivPZ4DvHiCrcF{kDQ_$B<B!BqiSNOUu|jb5QcjbDgax_q
z4ViKL_+rh!KjnA3P(%jCU0t0NK!;1a7yNhV2CoTN^E)uGb#t>+Ly+RN`TONrYCts1
zFDsz&4}F)~Nq^so>fbw^?=Jd)yykm*dx0A9CGXK4Ck`iG2jp)iko0n!*=Dih$0Pqf
zzyY}=a;u~}0+2)ki+x^4JjkX^#C(j4!Vxmx@^I0OXx@KBeD?^Jzu%p7A4OiCS*!f=
zdE9)OIO2}QZzt~9y`@_yGBpDV3k!)6%h`W#*v-EeP5z9u52Oe`mm6S8v@LuwCw#<r
z{yj0Ti6V0-g!U&mfRDJmi*A2U=4)2Ql-;P^;UC0idmF#~cLFKJJ?Gp1p8ho+OaqQA
zIM?ccKO;YU9AJPH7abP+Q`m@0Qp0^lZ)u`Z8&YT(`!rp3!H*;URWS<vdIg5~g?$;J
z_#s{&XKtdSN0U!*!tCD<{_lD4?>_whlo9=lW!Khs9udkXYP11Jm(QI$*O#}I+>P!7
zSnjJ*!e*rqM$7(}k*R~{NovJ#HxE3u4%6_2KHqIvvcw!#r~9(N+6aED6{9~;@?dH;
zsy4W|0w0r+mQIY1r;d9}{(>6#04Po!fR&a}(G)<%;KsUF%N_&<jAOVuJKEX^MyFT^
zUP{N0M-EzuS-iz0JAjlbl&Fz6&*Vc-L#A^FWO3lMcT%&$aV$D!toC~cDnKr;&*hbl
z6scalS_z2|csYU~QBhH$r)vBHEKy5aTa_{^Gm{u5%Q~K*icg2%(EHv<v=f61sTi=8
zBQ9Xupa4Gaxc9^DbRw=!7cbs(UIgh+xnsu{D{LR?hgw;PUd;}e(ZCxE7}50HA5|H4
z|Jt{wqEhL?-@c9i8I5kI^CsW%mp17}2I@Fho+KaGzu)vO6L6R`gB$4PDNBK+kFj6w
z`!i^&ODHu!T1PS?LfF(Q1Pgf!HD{i|`?k#J_r!$Gcy7+>RaiD|p%-#W57?nOh!pCg
zI^59Pk-)<uMyHkPZar){u3xYFMi@bpp#;a&L=EUu(d$ntq~+m97bft6RtFAZb#469
z$%hdwFf#Z7M^ug06sPxa%eDn~hI{VNP5l%xjuRe`#X{)7U-{ZpfPk~@)v0g#IDp=s
zB|^>#+@AC9EBSrz!3l$2^4y7r?%|X91ObjU17h~L+Q#?jf3idUg)=5;)bpY_@<s44
ziKA+7pWzMBMH`7~0h3^^VD;Sl4P^_dX9^l-gc8WSu9L}NA)sDUw>UC1IM{*%6=l4-
zx;jBznVF$U5q+TWe!(a-bk}y3dg!wF9vam+7y*_SM40uRJ79k^Fcw!w=O<(=F#Cc&
z;iZxXE*70?edFU6Ffb1aA}$V#fPR#X2hwVYN2{x=PfJ^n;M%oe!v;h_z_TDHFp2>4
z=%bZP@N*3At#@n%RRkj|^qRpU8;$6Um=@T>EsydYbLTLA)gLbv_Lt9}1GNs0^+(Oy
zhVKR0PkU`GP9-Ah6LY3aOcHcl_*a=e1i%GsVQC5WAfV$h1Wx0|4QS*1{rqTXXfR`S
z9pj|HJ|QG)XbUQ{{$pYRCMINM*}+`_leBPoC<f5zY!<a{f(VOtevFJZ;M2fBa-gC}
z8g9=PBbF}_hY7XXwQEzTozXp9h}I*L47%zz<<XtB?uEt^#Ut{3`mI~Hju)L|XFm@J
z5k~nC%;45Et2^Nw!K_+!vIm|8nHC#r^D-r5%HrbU#vPNOh6r^*hgljiuhsf7xLm@a
zpfi^X<9L^q>wc%?{R|fMa7ReNspMCUt%zwEM2~5394#<&9W<yA_QKb_Dbe@`!3zwB
zh}lALnjvvu3@```#`E@sXEYO2R0kytUAwZ36v)#A4N;x~Mi!#d5pdJf4Y*$Xrqmy6
z)as<IgpnxdiIcCT(Dne)3@=D%e-mz4fuhH54K_e04-^&@z%**LpK%?7DO9<6*w_|v
zhftCO>dF>k=s)sn3;JO+;dSPG*VUCkcKl_)-Zdnuay)dTIdo`oxSr|DX9ylruV+M(
zyv&vsI1qQRxd{#g)cS+hHh6-F2x{bYh%?KiG4q3|%taB3xtV7G{?^r9>`fYQ@6p$a
z$%ci{Z?H2quw&B@7T0&iF_FB9IX)C7wQ+HA((n7Ikv)YG?qi2eV^(KUW-W+eh`iq0
z`fZ??T505)CxI;?72=RSNZ6`vhGQUD^;o%RrG#D<$Z@E(JkY(VS(_4!$)v#OHHveg
z$DqIVPKz5xYmql+oaZOp(C5v-!N$(M)cbJvrcs!kKI~=TDV}`;WEU$qr}Z#qdka^w
zVPtZ~<0D6ouy<nVZcj}f5l{r1mkkGB?K{91q3|GX{@eDE1<k7`-k?>tLZ6RJ*n%jf
zo<MVpTPW7KmpC5;RNueLa*d<4IOn+aA+VCb6$AVM_7To?TngXvCG#(%<cwvFj|3v|
zpr{RzvMLyfH(x{oWm@rEu&o90@BjSu4GdOSCq2|X_M=JoRQ@e^W<GPyX4HOwPJ6<y
zMM){Le**2*_CX>N)fb3N1-yevj|@Q3o18e&Rt21<ZgLHY45CWC$Yk{QRVIPpcDnBs
zIDVJ~5td|3EeW5#e>cZ%#L?+C;ID4qg>ilI>7y_ABRCU|>ll8Vlhy+098YsZHB0<B
z8#i}qa<YoaFNjvjDUv=ZDdR+12fF}^R@Tj+LC{}?vjdWfuJllx5Py9hUL3Rk&Vq(*
z4+zZ4N_;61uy(_$T<Qc^kVQyPfcRu56HG5SZqo}TBQlSey7&w7^I`O61oeqSx6gVr
z<cXNC0EnLLiE{@wC;#!|dI<RJ4D1d~IKtp?1Y}@5sRkZ45F2f2+WkL{^?T`<t>Fg$
zj0b{AhWh%1aPn(+XiUriC`SaCmp4*AfnLNm*%bb24`JvFN~kb-@hiQRG3j>vRYV`C
zG*l=NsUdJMD<&x#85$DZ-q_pp=@vLfa00ahFLo0gL#r8dL_6D}9z;d;!+A;1z(938
z7>po2<Y9P3px%<S+dDhq2-qxO@;0JF<dd9+lvD-K?$JWpy#SUGFH8%C)r#=ulWiGd
zJ>2H6UcJJcNx-*-$%koZngC90?AU@#w|oILl@$7M{3UNe`aCl?cLbe`u|CZF{3@s1
zefMMA;qetf(+PF+rAwD2Bpw$R6EXp~NpHSzO90X++$Lgi&E7TUN5eGk?4e>!2UJak
zX+mINQ9BK*dR)6!hsm3MR(o=?vk6ab^2@oVB-)UCF-i1!f_MS*9-+nE>XwA+;q5!K
z6$6#P-Y6{X^ApuoRU~||!d%(oLD);s6+w=Yoo0AmGQZ-t?IIo-<iKGNKwZ0bO;ghp
zD`#Y+9f2Od8=nS`&i!<BbZB*L-J_?4Y6BlNPbLfGdqOsdx!JDpq${+FTt3GBS_(t6
z0jHYG?m6}!1+N@YHAof0vJ2uiV2(iXckF1)yo_{ex9AGt)`&p5<JChLfQsOJC3-E;
z@@h~I&=@XsY8Ja9VfJ}od^H`7{4iBgR#HNN44?kh)m5U*98Wjyz!(i(tOo9rCx<aQ
z8TrYkhjAZ2egp!~%FH~XGmgWWV5RVJ*9xpr*WiF|<O__y^YiCVO-nztoMNVmKDsa-
zJ`_Wl0nNpU{0<yDSfw$oZ)cyqljUG%$47&sfU@o#<Vje6>N<BN(q2iybsqvlx0MAO
zWC5V5mB&+_f>#6u&KCrf{1&Ny6!Zf^^sB??rk^mnQ+IVWy_(WgQc;OkIv!X}IDldw
zBdIkBzAbq4=uI56&rGh<ARr*Ll0nclOW$zdMef2izXM<>P_Ln(A&f!TO~)^#Ecg}h
zEGGD>T*%2jU!Zk_fx!**Mv&~8n;$9U!lO#jlkqo+{aXbRGxOGwo`HtXNw!;<`UNK0
zIQ+d5i!ZiVYgLsgv2_V+QRmrVg3wER0$xr=N-BP79a(OZ<XTm2EqwZ)FLC>A{iIuH
zpWw`?#0>jNWMeop{VMgV24Fh^Ma3r=q~{3jkkjXD$1#P$B=i!M^j04TgOjpOb{sp2
z>PJ{#OF<yG3V57-x3yaHqSUDyF(rh6iOm&R;{f)o-|}T}x$l4t?jIxXJAByu<Kfqc
z%E36w&@1J?10LS0KlgBfZWl$PY=Vj=*&1638p_tD8F?08xo{C-KE6!SFOaQB*&%>Q
zE-{25*XEhCICgW0FAGPaj@Ct-Lg;jg)G2{j2KGGz1XYw{)w2f1vV#LHJ@jy%1$Ep!
zNyW%}^WJgqBEam{*4EAIXxRE0BXfY9Sh6xLQ?{t;FcOj^Z++a*OC0-Ws`kPp6sY2g
zdbe(6;BQFG__d=50nv*H;5>f*^l3YIv*IF@EoGbEda}cnrat=WzK3xRTPPw#>iJqX
z&LnN12s<4-sNP(U(1S_V`pI97gSV(Owi=clIs*-DQnw=fID20m5q6FiIhVkPx$<&i
zawfseirj%XOs>1Ff(X}4W92woO2XDo3`o@~P;w*u!jb|=0;P#SaAGU3LWs=MmW%LU
z;yghrj?6Aq!%c25xG3o5>8Q&ch%`Cz+f{7l`s{DWf)kRvAXbvIXpFg~IqdML_3KJv
zGcf}Grm8^r^<HGwq-F?6)vO1II>Ru(s7SqtiD5{cIL9`r`&JJGZ^><(4XUZApdS0c
zwy>*eU>ZajzQFLp-pc0KoXr#wYYif$?5oHp>zZM6fQhmBXb|>T>1QO(XKuGe297)u
z+ZXvW@?V$%Ck^LECv5cvUR<xq`PzQ0t-i5Q*}5LOCWMP@^Banadk!3kYn{Lt3kc2P
zb=+aS2R%ztAU!YD1M%ZHCi)3NA0!|9fmDsGSO~qB)X1cyG)j4SO)aftlaOSbKH49g
z2m!NOJ+6uHjOtHC;);lak;ADjX3dYEJV|n70mt*~X1t=BS}weH{69gS0d$9aL(9R{
zjI^>%5L(tYd=xTEjHwY{RaHU;Ff{u2g+c<PHi87Z<I{a1fd#w57w|!wtAF6+#DtGB
z3Dh{)5&=v1>%qc!c&wwhlEHD6=zBsNZ1D<0B~)7~Jc46Zfa%D){+4QopyPV}yc)G}
zPM1Ins=9yVu9mem_xPsLN_tt|GdS+?eOJM*zsQeb$HJ#|)j+2JXMRa6y-5xXB?&_o
z+#0_fIE`>%N)2>Q*rOLqu#7M*GCkB!3>bL8lN{TzX%z-oOFxsa0|K_lof?Jc7&MkD
z<$&!XeiJ&$7^T$JlvANhUAzQC*DDtY6<Cu<UNX)}WSg*2Xop`eSEG$_hgzu`op&tL
z4Bk5UOl5U7F|S~=)yBz|ly?(%Lt@>Ze_eftTD6$Ko9_LFfCPx1x)3qc!QOrhEudK7
zT%4O~{gHX<esHoeHjZC94lM*wgkp@A$5t4Tx~rcZtUMwfYd!e=5YvRjrKzxMUgD8#
zoqJvHI_!a{1X^oIrcWM#218TXh*6HeSOoi|HPb(q_p<c~J~>l&v3R7kHvOK?cir7f
zdY53z7<aRvH)Ln<PdeobpQxq?QH|_&CBnG**s)`pXF^b*z)!5~z9NH&%|mayG9mf2
zey!}%?ZD`GLZ&b=hY+D!%O>G-SKz((Ksk*VaR4(Mlz(d6*;!eTPx$rTyjs|3rdMNX
zvNRYH9Bf)%eA1XSrK+Z8*u{Y25VZy{QL%z_tXOuYMs=AfLsc(vB-T*|9CBHQj1%R2
z^kR2C6n0xunJO`nOUv!GHs9;EmwKPjCvpxtw#&Svq1oSL#Cxp`H3PCc)6CKHu6+&;
zAqQw^;vFlK^=AKF@RCE;R*K%o)8GmLl2Xds4EOd%{Io;S9POzn;TD(h<M~9^W0ED3
zThePaeD7ZI)TKAo1DXMJ%-8Gvl!xUTk9r2{SjbJjnh@?LE$=BoqoTLSl2kEa=Si%L
z%wLe2nI8Na6x$(Eo6|L`P45BrFJ5w(5fA;0Qo<~jJhVQsON6-&G4AD0U1jA}Yw<*O
z!h;@yHk2fKEAANXO^9-asnp(!w0Y3V`3g1p#HP*X{OP{I;+t|k<FWxo4(|8|c)As;
zM<#6bK~Wx%!cVm+HsQ>r2aaAO375B#sh*P$Opg32+(AX94LPy%BH|nxoYBN!diOX4
zXiy_*NlQPQreDHwwzRNtuuCN>DXCwp`j|9<VU-bh?zNZD&z9NxOPL`jfXL^;J8$^O
zv9Mf;>1t_Fdz5{d-Jdc5;^@*`N3I8uEn&Tspa7_POh|Bwx5%Y=_kQ<o=5F?L7h#m0
z+19Fv92G@vQ+4S=-;Od{0|X*MN6nL#oE#$TatGwPGl?IgIee@7vkwnxdcJWz<qd?V
zylfRbn~IpSI-kE{o(?LYvSy+=;E1Z!i8n}>Mw@rKT@X{P=YQ*rG#a7u_Ql7shR90U
znVnw-V@!%c*))oKB+O<VB-GsvNx4H}7L31gSOM*%Z|wpY!Pa(LUIHw}rs`7MRlxMy
z-FcBR#k$PwTs9t+XndV8A_X}L(q}ZC)g_$)j(dTCf3j^H9b23HWn+3mVo%iy)Z9Fa
zW4%~&daV64`IG5ZHK^;Ik)riRBGSRo)ubJRUp^q|FU(7q4*{bND&n-$3@YE#FzzWP
zY-$_t>7g8?+_{rROJ#;lSx$qv&=V(5N=>D#Ai!_>_C`5wa^ECMd*o9STr?}jX6<SJ
z@iPPgVPEX*S!2mJ#`Z;-M;gF<PHQz?T<QVhQu#5}XyBhD?PUHeiq@*#0#vCz5V?>)
zK^V-1!#C|KTC!A%>ws4(pNS@1t0)67SY;X|8qVS~g6hRbQ28${_pBeme5bSfViGif
z0+b2AKpu{=WYYn+*?LLHguSGBQ13}XFeGEW3QY~XWvkN-n|9>5J!<tN&Jk<UZ$!2c
zck@wlIn*G679G)c8=3b~F<KB2jqUVBSJw}*4+*dUYWQp184x%iydX{D$zQ_<i4WIp
zrTA#){Rbzm5A>hx9wvaq5J_j}=O-by2uNUs>c@`}_Jg|CIS@taK(h-9|JfUy+C-qj
zb95F^D%fH`pl`?=@$GP=(>%tts<>8S(9JlgD^{li7>eEse1mAO5IDMuW(<H}gcUuu
zuWF1B`S{I3_thoj4rmTHZW~0W7wUjisB)Gf(P|43X{gX$K1i3!%HE@pz!=d66@_cp
zd~Q+2^O!Dg6D$OV{N&N2_&}MMh7TWv%0lEdfG^=%P6UT9AXGs9ZVlQ*!V4b0sEs#o
zeb;!oQSt&me>U(PXwiTxKl=oV=Igg_qpBYM;FPyt`7t*)hvugOtUC-1LYRN|t`ldH
z0F}GCre-0bW}gFgsX~xNM+towf*rZz#QF0}kZx8n6Mi|>11OW&%>WEf<7cQDvB9nC
zA}$fA8<}Kqk%=fa(7Hvo1!n_0w|E-Zyq}!T2cUrOgBgqx$~BNcp?VRFxEkA!<DT%N
z>owolEPeH=2HN{$`D;+R0$giWn?_F;z^RIR#jQ+O9Q5ipZ}y@gZ1VvEGz*-9B0o#@
zZWmEg;5%TECM095HW~^N)>no^2_%J`9g6O5{Nd(@+T%F**sl`>8Q>GxPq82PS>*U5
zU5F6y5V@fJJTc)a=f^E21E@C?!h}xUoWzm5naDewlj7qSkW{Ns`o4LCB%1Z0IUGL{
zykm%t1G&=@+s?yr6sb;ABenL*sP6hSncQr@vBM6ZycOaRWe;V0n!|^?@TurEIG9V!
zH2e0Fe!i=xT)k5=-v*l7il;e%<<=norqv>R<FRnq0)dPD9vnOa0r<D7H0>f-%ByrX
z>H^*aVudwZ4|Xq|&K=&#Wow?Gh$vnV2Z=it9J0c*WqWGCANU0aC(1CRpj&!%2TA8@
z@a#uTh^BBC4fuk}1m5OjDxzG!tEvolAilCj<45Zbw4EE^fr8!e94~ZFXlCm2Lpe5l
zR^y2lcR0#*<Zt$AS}eX6UHSEOu_MP=S&i`P#5B2^gn<f-8^{H_L;s_oruuv9V5T>Y
zu|Ba{oB{&mmx4d*_&fYQ;qh<Xc`#yxMSu%9s6VKHNi|i}R;=N!uqn6)MIk94gCi<2
zTWM!!=mH>VGJ~Y927W?_`%xA9OFMM^25f`GOBE{)nefh^R;x<5o0T9?Foo+o+6%MO
z)RdI?upnUS>o$rWof(*qP+5c?!LI`%h7+$YjxBr=&=>4zK?r5>Cz=-Y6cju(L{Ll&
z^!3g9XQB!-c|yCb*+TTM;v7M_{4P8^t%RSA%>?H=<j0;8v?xDtVi6NIcTT({qs2Q8
zSqsE-9RR@$4P#@Q39KFJULqmFGT}3qbcke0ll2EA39X7NB!{T6t7dMWL|Fl>YZnz2
z#shpo>52QU<}2KCA;<aAyG-QhP?yglg#^2Ff2m<*Wd$~L5CDF5)<E@vLj##@JA0H;
zizhHyq#*I=HbusUk7#NdNP*lA@5tynbqUq|PN{<!v2=%6IZpRipyL9FT9^!-q8b(-
zP`)>BY*B#%cI=*i6CPfQW+LPvgjpd9e4y*#K`?Fu(Mz>NS(e3j^VY3NR@@5~y6@A|
zt<gwF1bYlh#0%IFB!LA5q}}GvDi3c$uuCKjUI5#IZ*ms4bmz{Yl*GVTydmNi(E{)B
z3egt`whB)x%@f-F`zb{<<d_<tczSATXaqRYVwm(lUw{Of!GQr2D^+**HLMN(8q0+K
zch?cjP}!kP4I~x#Dt@yTaL(<p!vuB4$|}k3{}pZ_1_mQas<dvp9G6hSVp^RlThlBz
z@Q58H9Vn{zX^FyozP()|=2~@<HofEm-mZD;yM3+?fO!=jmO<wpwU(Tts_(gzzLolL
zpeA}=fV*Q7X)A^gV3%Te0tRjSQ#b;1AkF8Qmfq0f4uq@{ujQqsI}=^)aCC>Yk=wJq
z{&6N#_00a#*hN4Jq)|ixjRDer?!pD2@KdOAupOaq1q$Qr@U(H2A^DAu&(Vb~!X8(-
zDJVYu4n_izoO%PwBI4C_8ezZXO@U1**lOZ+^#6**h}-ag^>M^qprGi>;B~5oU>T&*
zm#1=|7IQ3s+UUCedF)((gaE7UFMrEx16hKQP2VHu0t@;Zfv>s2cJp2`P^gL)l_qzT
zPYKmxY60TF7Axco<v>+fu3O%^Rc69~q6U6V)SQBOr}AvG_7<GFXY$6sYPS2)vMN4|
zc>SL!PrTmbu<_ataG*VCt0k=?xynuNrQYi=%YsZlM2lP957-jMY3A;Fx9MULs4j2~
z1AC!z1ytRn>O%1c)^|*(1CGDuhvoCpqY<g^W<FKc)D*j~7TR7#ngLto(L`a2#4leq
zP8~lgd*2;R)XV75QC}7Pf7pA=uq@ZEYt;3qfC_>lf*@d$qS6wAh(&joK}btThhl+(
zh%^YI(y4TcNVg)5gp|@HAic-sdfxZ{{<pv5+xL$Zz+KmMp7We@j4{U)6zm}8N%kzf
z?LvZ#vo1l`cHq-L^XF7kii>qQg27m*s;eX5-~mcCQuR^JC6++ww*SpNk;)~<KmASu
zHW^J5NOU=*;eH3y51oQXY-MD_Wm$m3zm$$cwT{{z$@|FqyeJx%v@l%u(y_754?~EG
z<Ap+JUI9eeoGqt(`h=jG`qls8f*2nJ6CMEA@eLx;_YXO4VDkHrpwLi|AkuLyFJJBl
zMvQT>a|&I$qN1Vti%xg9VmCM*xu(yo3=gy3ZDD#)(!a6czvh;zd~WO4IvO10Vn45b
zj$Oz{DuIMmeS`2q2YHa%v$X9lr?|K$_@26xIELskm1m%?j+vP#H$gT;I4CJA+oI7Z
z_+lY%Yce<{uygKt^Ru%~$QM%d(qdu=JR3>05G;J1oeSRLN}#*s&TvftU_96_B~7pK
zyFAJ{!O3Zka}s|`j6t-q`7nPBLqyOzQw-D|Xs_TANzk(ZW2tpp6BL-``Z>KGJb*2B
zf>T&Pz&>;iAV+s!-dwM3YTATr0q|b<Uh9C6E3Rp5w5)k?6txc>zgdu}$BpgdkI;BV
z%qpro38oX19}nmXp$=cMF1m=b2bc7%DJ|?zkSK>rzd^YmbEYKL)AdcDc8QY{usD<z
zsK++f7ZgTpfhqx_#8{f~D964JnVBu_rB^cEhhV5Xe&3951AgMAEQ?A|WO(?>>MDd(
z1o?m{z#lV;{vA{dNufPGJs`&}e^Wp6r&7}K_4Jk0DVOfWmZ!?c)ye1{N&U0|Q3Mne
zAUvpSG|VgJO(jtyk3bWd+7_-^5e6#v=91pM4XHk=)B-~k>$r(toSdF+L4?6|fn9Y~
z)s;Ibt*x!d=u!@}M7>c@L*|8#w`#?YQWpu|)WjqT7??KUi;q7;ZV7}Gezv8^H?f9+
zxc_?TNLTW?8yx36n&XfwOe64lamL2QAss^lhdYQ`bj*eWwa{OLi-pWcWZFf1a@5~D
zXcq|<IYS@m+S8XH)$wwmB6TxL#1%l~pkU3$+pMp<mC;~fLt2A`{W;(uyonV^ikrDp
zP;sbjd16JK7ZttKlk?%u+_pzhJ!(3K0rK8c>9=>h5BAo{BG1$?HERjlmG*2k>?P@+
z?%;pn*g?SHTT2&NbL{6pYk|WeO3fDK`@?TzV~5fGK(I$hVi*n{6fmrLVYYnx_il43
z3`sz(4s*5OTjzEb=z1XNBlO^X1hEm&b4pZ7cQc9;6Mx~^rEtGneG9*h+p7Tz^Pdew
z*U#zyv8_M5{TRuKk&}l$T>Qiq)?C)gDDF|bwG|o>q219a5mnpJK#VE3CHh%ld-gkx
z6%vg&CehG?@ZAO<RW>QBg+vXU5#0TU>*}NfI8!N5c<dr4*ZpwwQT$nXn8|BjyXLLG
zIMbDNRfSF?4MZ=S_$f}#H7q}hlyRXK0Rdk@pn$^Tcw&P#MD)6x2dV;3Mca-_TwmBd
z-UNN)We{j42}8lb2A#X1MkVEy+mq4bF$vqAx|7fpg2-&oiw2yEg+!mBcNS#TW#sA<
z7{Afb6hv5|0Mpu9?I^vJ{yqd_eAd?1uq5<Cvl<Y{Fqb4*V*{2?ROCbU-_=)s?D3e}
zChA!;bZ_+sG!od>s;9%xyXM|lFh!>>%b7FH*QUJKwebb@^LPoe871mXbxj5o&>aY-
z5PxVVi=&P})=pwAhA$I5^9deAQc`CI6>^Yg0+Oso+MUdVMkvHqLMVYajr)b666-ie
zfJy@lh;h^T0=Ns4pr(Xto$s%50=|kfM38>0mIEG$UXV0C<?UNyfK1)HU{lu6dJGN9
zf{*@Jj#R>bU;-2)UdTLY6W5__Sjq-psUKg-B_=WP+DKkL^g9@FSoyvif7Fqxb<5YG
z>;NbP8(0R6kDz@E6%ba_!D729++B!xUZ10Be+;<Kk5NSZ8%;8(YR-(}7JOZx_VfZ5
z{bX4pF_k@Dg!AqVE35SR@U+B4r+-9JLV|-46s7?{sZnBYU;soehSR})z;0nnKd`gA
z`{QD}4qg^P)S6~ljA+RQpf#9xC@}%FSq2Xabo`dCzkl7(v_w@e1RpH!?wu;XQR;2t
zC>$RF0D`7Nd}x(T>a<xn4RLHncQ`!C`GH^owN-1YdozYCLF@@r1(1oYd}6*1L<GsR
zop|)@$xm4ct+Kdb8ICQc(Wj0L<*i}AL69d;Vx0BJlOpW9UT`P_yHRNo+;&9kz3y*G
zDy%rXrY3IF`)i#J#jcRtc>D}e3p3?X#u=L%8>7TMHwo^)JseL1WukiJ0x!-FBIp62
z*EoZjPDH^5FZu%=8_)|P=3xD|^BjCx_~pwNpoMMk1Hh?LP*A9yURee|iaBa1cG}*L
zx)85{D3!KainI<ozijf;Rv5Gfy^hGabHxiyB5K>O!z+$Z!#cZ}Z;OEcq*PqRfWdq+
z!Rq0RdAyT4(UQNw>6)~(@{BZlz_o*?G^zIQSLToFBQOkz^mvX++StrtMfx+TMQ!`~
z3pX993ySaL@9xI}u7P?LTc0h*dlOPZrLe$2JNQ$eT7mPvZuKgt6f`73-DK9;M7uQj
zLpgK&fZgu2SLhB)6fT$bF!i`;{5D=r){y6kYu7T%bd+XC+Pz)Q(F^%#NX6UN8}c|h
zE=LUSwJksv!@|m{qpgjBSg-}4`w(y!0x}Vd6pe}bgi;lQwb6IOqSiLM698X`blk|u
z2yzu*QT*Xb9hx)RIRF24+7Z2LPo7-WD_KKFY>9NOj&+Gz;BF{0y}Z1Vqo_V@!xM2>
zp&*01HFOhrZS(|riQlJ>M|BB(qPTc}TU!oT?8kESFs(xAYiDOSH8}}Nc5r&yN;yjq
z^SBC7G(Wa<MI=Cml6(N2ag3ObiZNm1HxQl>qrrx%k!-K4sp-Vs2Ha<!^SOH#lL!|3
zpQ3EUAD{(2RT(oV7Qj*w`<KYypXa@R4<?6aMxr!bTgE_28;CC8Rh*^K&>o}dG~(Rc
z<pLgj7iEtGOwzCqAF~E4junMr3VxT41=Jc6V-;VMbniIdPfLk{6EGL*afDeMhUw|)
zkK2w1)&h-z)X37(5{?P<f>x>46tI*;i)BMoQ!eWMzg?CvzT->fR?CTlcUW8;t`&}L
zIyc6b0RcyYrJ>M(6bFOsD^|2FUbeFOyW4l0JI&pJC=i3Me<1Zpk~-;yg#;-f&{%v5
zTogS$JqoeJ9c}A_Z;%a*&&T7&BCCZiKK^WYerANOs0DymqsQhrBGI0UFv7vKy1I%>
zlYv75scBsbqC5dn8J*U9r=iI}O!E-qxQVVd;Jjb4QG;z!c>@Us0kVluN_$%OOWY=L
zapCFp^a(CO+uqNdzaEu1eR-*q=6tKMXVW_6fWAVhnB^AV14k1UZo3b5HhGSrGQ*U1
zOb0`0&9%Q(HdS^tb@kQhvxK5scQka=<e(9KvB85b?QRz#r!JkYpFzX1!wIT&iws}N
zWp9pJ%}n1O>im-Y10BDYjV^apb}7786W`37PNmm$?sPrZBWyxr5=R>!s<B(P&+-7Q
zo6;%t7%dN{(QA$b4;#^#(1>y-dhmWzoLR_6FD?Wv8XKipYlzUek(jaqq#|nmQ#gDB
zTwvaVub61`lzTB5H?{opXT;j*fPgoeTt|)!lq~l}Vzy_9H=v_8<JB*q9DyU^ZBqit
z8f->7trwiJjtoKrNN5<d+_3;mhG55p);JO<qyV^c6;atE!$1%{3@IY=fP(Kpc5sR!
zSYfT_Kd0p+DIHLQT9^`O7UuLIO#U1mW~;-XSt=^3vv6(#!e#KvQLYPgD^#qw>Co~7
z)kCt`=Q5E-D2ZmH%C19yq*SU)CuD0#kpf0aH~V$~T>dF0lCao500qL^>3ww5it8zb
z6b78Z6O;kabxrk44Gs+j``_mPrkPq;Hj7ghK~D(v?#*a`L<vZVNp|jpM{RQVL9|sO
zANdKP5d;<--&!$@tr*RP$m*gX&z0I~M*?kJk;@Mp?lfImG@$5!TyBalP4k|9nqoOH
z`z7W2<;(kAC2t)#b)>WC!#zD}{@g^Lh>F{MRZK-JOBIP<J6&5fZ)r^Kq@Z`*!}jz)
z|9NZQN4ZwK(7?;9bFX*joae^+kt|*l%A)$|l4YTvVn}&jL=jMt)@s8qiipp!DGf5J
zWRVSvk*$)VVOJ(3y5D-7Kz#xog-KE6acpcRK`Y&Z^Kg)Xq(CX8Ul;SBqA8OuBQK_9
zu7ygsJWn<X05!Cke9Wm>8OYIMv$J1aadP^DEQ4>W!|m{Gb1vNC@9F6<SZVnu76%ZF
zC~+M7T(#=|n3n3Pij06e@Q?mZN+^hkiJidrPH<ivhv*7QVC1M6G*JC19(NNb$$OVY
zGaEq#B)rzvgdGBKMZK93q=Q`mS&0^WaC2iF9#DZp9sr7NCke@TSXG?!+nu$N?f_j0
z4-XIu7L8R^@qjo9ESP{a*W-wh7VwM*3MRe$1)LP{_zk!(LVy>VlboTIg;fz7dtxjA
z#*l|z{?lQJ3I^3fWA|szID8}ME+M);28Elzz9RU?AYVXcvzMH_m9aRS&&m7wbA32i
z0nEqZ4+RW2aRcuvw(<L$qkEN=i_0n@^r^<z=MqtVd&viI{ArzBS%-DtvVrmgJT4Dd
z-oCoPLG~G!IjHs=J3OQCDDm3@o{RNF<7t2AjwUMSkax?=qE2U@3ZI;)5j(J7g*xN1
zF7=K#iFRN6_+3fW_So93D!klKYIqmV87Zxht{ZFnF4J$^wLL>QR{9^(m5vLIt;Led
zd#4HuI9L7Fr>@m?wRfc?t}jxr2rK6r=GX?*r7z8orE5g%e~rzK2r{3SYE<}-G^r-p
z){5WY%faX0MKT8(yxw=7l@Q^SN$EamF39mMQNC7~X1B(j>;Q!izTW=%o9nC5I<9+r
zMS_ANV!OSMj=Qu@siY`m**2!0+FGa_m#d3BzU+TWq;zuhS)`Vgdb~whX0O;U^*qDg
z<jX1ATI<b%`9)`DqFvm~ENi+w4K~bEi+F>{r|fBc!UV0OpPTz#EOa!4+F@L%W&aS4
zDb!`@if_Df#R!Q966x+~`F11_3HH9mfK*{+sH>v_Rr{A$>hG>l^p?J}FAJJ<<h!Ga
z+AF7U=AtPfRN9^d_5&&krI~j@QKX{CyHTHEDRuM)?msD=xO6E;yvw<r4*?OXp=1$1
z&6o@Kk+9r0|AF;Wz4^MYh_?c1y&+?wc206%3VD}H4l?G<6pRH*5tUo6*UZYiP<ZIA
z0ABJnzqvYUklc7!7$p^R;KW6vKy?OC6O4S;@Aa?3CjsWMxbz;|3CDBGS|@#|ZT|y=
ztjg-@e&~2`@B<TeMv;z*jYs8>{v{cUX6P2Y6<vW5MzVoA5&^l67<L$AaV!iWBtm=X
zr_?<I#wB|MStWp-MEQT7`}=qFbYq7ZYGA-Bz3rQQ==E0U!pIcZHQi`ggXuwVdYmvS
zkNDHX)VE4=Vw$w*5BQVrSAeC#*RQS5Xtx-v^$}Zt$0LDH7CXm-;)(W*vx3)?ii%5;
zAJ^{w`XaGNdw^N9F$20)c`bPZVt_2tp>CJ`2M;bG`JRxoSXo<}a9RjhG=U?SF#Tb^
z+lr8TO;@uD2;>4eD5SN;Naq!tyxrPoU1q;P2&+X3Z;<3=4oMbUVtiv}it0~#e;E2#
zGoJTY*X)v1gS$;nmWluKc-aG@Cr+O9V>;jc;^ckEik-oY^i{X#`_hT(LQu3KQ&3o#
zOkz<d6$*1tXXo&22^A_%)@zW(^|ASty07Iz(9;qEW+$2c7J4TbTCQ7>F~JHdb+Ysb
znLP8k;3$>$hAn}u60NRsYD$G)`6t-bHh<m`ANS}IGFr2mVhE*UEgG|FxHb2A*AtfS
zE;jX_uCjBVv}yOzq8U~H((WfeA<Ul9edq3j$!t8IO(CTy+bU<JZK=O6W2&z|zIZCH
z<n{Ol>;Z3$6~qN(Q25opt`uj;zZLVm+HR?~t1jj7mQeBM;@uM$+fB^{__^+Gj;(~7
zDJ~AX$V@O6+nqfxIT0#7H!YysJHX#xZ;>Ipd`9)7HHPh6yvn-J>aiCcOVBd56nBA#
z&A?%fAjqXWTxwnqHvp)0DZ1)b!C(No`05)~n4jO8s`ag>M<+sv7*z#iE4bXIpda!K
zLcj=GqtPE65iUM7+k~kLX7HSm-@BWFf|xoy9iO>!52v||;3|UIm!L!GYAGRnhu!9f
zr2w8(ME$X^*~|)g`LYFMHnOc}<wNLOP)k;YksBcj%ijSJNjwUhI{sLC)MOUxTMz^$
z=k>U-?@=>!KG6;V!s-wfH|o2<N~>L7N61hlC#obg;S@A{Fq=$#AjrK@rVF1wO>OrE
zbZRet1<1x+<Olu<w4=~j#r^$Bq@{rhVq-mdGmz^p_H`gPgfBvzu)di*{?qYqCYy|B
zBUzFTJv}uzs0?Bj;ik@bsTL*ARUy;*?r!b7M>^mws!ezKgrA~cUkv*?B6*$VS3+`Y
z+=iqI7>fPo`U+AphWU%Z-5983|2PIDkS(3K3dSrNJsX3Z$#4Ejz#4j!ycuo=aj3!h
z`AT@9?&`<_TD7iBhV2{qG4KYd6ZC;p!~yZ8_^LAuYL_?BViMYS{yDyXCq^$d?#{vY
zY{w@A8{Wb<NO&M@ZMs^9l9G~UVA|=c+k`m=GMcswgRKXYVlSS@fwBX(!7y)=1+gBg
zZgcumuwN%UGSpI77!6l}1Yu`J&243+1VxKF56;Y>3x#FwFb?HLCLSyRY6Zgg5gtN_
zM`{e*R<kd_=C1Q{tIwXrT9S2t&WS<C1DLkNcgN8i*MMlv6mF}b>LO=C769%NZonI(
ze(n6`Z?POetBO?jhh`oYO|0q!FRs{l24I7V^DLHT3uzrv9k21Kj6v<m?CfmRML1lS
z0BJ>-iH4iGp);<9MgWn%zG-vv^BSXQn5v$CVJXRGM6<k@*a`PEkDl2c=|>b%S{5dj
zf`w(jw9<y{*nK~Lt+J5|<xk_tZE^4_$=PUGlBuUv)|${OyfAk3?i03W)21q|1-~-x
zmpUZ%om4rWO?7HCC%j{#T)$iXeP%ZO55d8rE32UizZ5&(O-GBlYOCHpDo@#J<9Oce
zcW28jnGbH2nRfmx!3py=Kqh^>z1c9K3#c&CS;7wqM-p&Jpq@IX=I#Ii!nPc<G{YD*
zj4A<0-2+4bsT;uyqL_!pl?7lZUlB+I2$!MA>5g4wWVg|pXKy4ypxGRTW+k<#C@ZGM
zAty%;_Xmo2;s)YEAvr}M;OFCmgHz%9^_Zh6OTP2>LQiSg;s8Y8gATk1X%uR8<WK8(
z9{MLpNl46^q^@0qwN+l2WGl;Bb2sv202Qe#o&>n??+kK6pN5kP3KyiuNGCgFX(7yJ
zMyG(;hSQKVwCG?sCS%}!(5}4(zbeD46&;cX0juSiwMpm3YGe?d#7O=M3Kb*4PmHI1
z3kVA%?7x!x`0fl`klY-;3=D>7Zizq5U5LWB=2$5PyaMY+^Y<G#siHazOpgju_Pa-1
zD@!p87~t6p-uz7>C&8ca9hVS@E~<>y-rf@YH6Xlpav?O3p^xx3mM}Z}s<|e#P4>_+
z4EjS7fa0c2Bu+yR+h*k0+~mxGIr$fx5nu<xxpo<(Y4h7$QA19Q<?7ohsDwlm+fozB
z^xQSvu;ACl-3K>D+6I20FBF=wmi1EHI|66LAHqZsu7+py3J;(j+qZ8HdJidPK}+cH
zF{pD0a5i9L$X|cp4iTzvw`r(lHGI$GL}X3>k2)kgX{alJR0(o$aPaZ*p%)IiJ&?k<
z0VW9Ud_Re(B%G!Uthf~-mz@%`OG_x;pL=_sgP;$yI|21$8C34*hg!w3=+!gy!U5K}
zC@IM=PJ-gDeE}0Trf@4E{6Shy{dx|~Fo?8u$wuLyac%^Uk>A=$3HMq;kay6Pg$yf<
zZ1-;E=kI<2m*qyEhb-6i2KiYkTG|dk53=q~sJL;t<#imIMSanh>Ee<ct^4-Ut`WzF
zKnP6Vb`d)|ci;yW6`=K<yALFyE9~`a%q7oDNFcHVbj-w74%j1vd;(@($qZNBTzIxy
zEG@b3Ws<G#YGT-5A^bmD0Ozu+dn{5j^=(xnFJyd*=Jzd)aFU-8;_p?P?@)GVcjy;v
z7mf+m`WB&(qM85p$4}4k4b!5!so&A<{Ry5u(j;706_a&i=jT6-JUYy)6quK+(-thj
z$!PKWz}Gr<XAWlC{rk}o>5h+uYRQJY3^gh022d>2{S!HtRh5C)!9ZFZ2PQfOE3Bh&
ztl?V`!Zf|Y?9m!Rb-eJz&j>mkKn%>h#=cI5Ks>5x;MHj56bVjgUji#qOJiWPnKL0U
z$u)-~073=II0%{2igGN(2GD1N9aai44{j5nDkGW82<SN|9iX<l?XeE700~-rk$9I|
z`(9=-QO`}c#x3-f;F#`)`RAB-3FOHTqQcrUZbd#UD2T8I9Gg~N0=+1yMU@8?mzH@N
z66VaVQ8>9F$Tn%4BiGZ2ay>(K5euAXQpmz6ib?R8XbvIid!{)G8Zr&Bth-wj;wo%r
zLsD=Xv~9atypG^Bt7~p<#yH3vc->>9_9B&?KQ=GO7DyWubDX-$pK$cV(Q?M$y>xAn
z#WbRvSP~ZI0)C)Eq3zkg7Yb>UmzS0nMh6EvSYQvD*kCUMJ1bm(tEm<z$v&gBG7mZ?
z0Zq&ON$<=HhS4V)dnZmmkFSGRUD)qf*71KxB2d^O*Qop+c!Uf|h_c5#I4<eM`C@ue
z??7Ml=K=HSdm~AzrTh_7)`?-MU@D=}ND}#$&_#9={?0^0$m2u6j|nNOh-3O_b!)4t
zW#AX|@e*?xiDolcg{#l?R#lC@^A`g)U;L*BS4CT$Rf7el56E%RRRr-4%>F()0ShsL
zD=R`X4Qm~;R_(hthoE1nXmbXycg|zo2{mQxQ8|cc5c5xY94&?6mc3<;4>5kzp#sR<
z3tQ7<Y+{z4nF*xnFGIBWz9a4vom%((j*x*}jR-yn?nJd@2`V~4(SHJ>25um}CC{8i
zfCHPb1lQcHkT_r<*ka^|Cb`9rlThMUEpcbxo+71u%y)Zt-9TA;`x1~DYYfFeJ$_Us
z0;bLDgy|!CJ6d9Wrq3TDbZA&sHp!`v9#x{mSM@AwaIt<s`pw49mip(jxcG!VXXl^V
z6**yfE{}IdmwI?iRHn%&oYwQpU!6(u%$a(7HoH&poylsk%({u}$#FT=YnN)iom-3(
z-;CQhue&(;B`hT1;6Vn5p%p(o?UC3X_s!LUaZ~jWb%WNm&$CSJY@4s=-v5dHW4^Cr
z(`TKPU(cQHx@}gid~SFjcWqO*l~%P;GA&CS%Les|0$w{MFC!>V{j{?4#lT(RB_azo
zDl6yyow${Fu$!1K|BVX70e9@)Wt@O&;#R27km7P}A}W7QT@<+L{jL$9FABm|8)HQD
z%8LTsh713kQH=S7O=Yi<Nw6Ap87b|-p|wp}m^`oA=((HrG+`)f=~_h?2;N+b!)8qy
z{g3Z2_!uiOH8e%5ApH!D6Dx&DdAUi74?$!UzSYWlxB~1JAI2H@1h$JBCwWsiZ}lvV
z4k82c0!cXW;h+FNPz>7vqc^W1IKnGS`2#(<Qpg&gh%Jud@$2s?6t2K$E!w2T=GKt^
z4o)1U6$U9yKzJ#vs<c#NotT}%!fj;z6bGitmcB8xQ-!4m05T;4c0mL7lU!<?L5_cB
zZ_Hg~W@c^+RXNDE<slcOU1u3vD<?TF#6fdLUpFU5_oOu`C7e%Qq|2*~V}N<zzIO;t
zTw0lz?w&G4R2wuO0-UN`^1Y)&V_E(PVi|GNk|&_9-2f|5xQ71oO!$ztaU=r1Js2ug
zRawa%`P9#k=5-ySCs6`+J23r1b*p*L8r-f+!n#Qyj}se!J7d<kr|9i8vc3oAC7jPx
z`CtSVX@J>^H;mzr@$e{TwGjVwBZ&hgOWyVHYEaGq(OrtCD&@9gTY|>L28sqlk2Dxp
zJ>c|}c49nkuF|LP_XZ@S4~as%F~PVdj>a_<1QjI@g3xPvrnlAvly2P-Uh_UkX<J)b
zzIl*uel_t?m}pCT&`!E?YR)wCZEck~L)^~_)=F!N-qo=s1#_<Sk>zIwJ^einJ)LRa
zR^|OP_|gj!M;%E^Ri;O^L=9}X-Hk8ps@-KUy^`72yeOP8b9mPLllS@KEiEGkgQs@v
z=~w=k94hYPI$ytHfZwi{c~R1HS+{yvu30e@-PP^*Izua7YOwAhv?cTSNJhh#;)`}7
z<{N6CFDH#91wNOZ*BYeG7EA1naJU=N-nBn@{<-$2Hy0AjR?Va8-PV6ppVS+XxTfvW
z-9XJ-66%zv^DTX0wJ9gGohipN$kz3{e4B7EdyKLMYxPKyn=U}ryW|@_BHfs%{EN#9
zAqDj!gJqssyUXoOfY16}r*9R6|2gyu>dFnxFW!{-dv9_tf(jg$XE6GyRfMpJb};@e
z0(!{tYc%^k)uYjHak?i@!65}rQ29r$>E=Zm)%}n_KGxg&Ee&7^boMWp3ec?5m%%BG
zegb7CV(JHIwYYr`P>bLW2H!2hVj}m~f)Kb9eK>P6@<?jb>hIKq2w1@RSIETxNj18$
zu9<Y<jKcY<S67k^<!o4v<yclN+BGrv8)dTa5=|FQTjW_rq-qZ|7}>4cB4hpNN+eWj
zTHmrJLGyrvFDbqi0$oCOJ9ib(F%ae&={&=5+cVyYiELfD?#?gy9>iOIk`x?ME6CcR
z6HUqFJAaW8(v#f0eK54MNq7X@@sux(=>aZn)rKw<WfNk~h}OdK8y=wNZ{ED7Rd*dl
z+0F4AV2lc>^#=#La4xX1=7tG7DfT|9x`WapKy5Ff-95<iQl?|h2WzwR_Sa)Jm5+mi
z+lWD%+UJg)JJ&gD<g2xZ3E*o?DorOSkFF|3<h8(_l&6<sEXxn89CxY-f`a5!#|1aI
zC?J7V)rca_H>8*ee{Lb-1-oV%b-6N#hIMPWmTKEt`S872`qMUKM7ybT>d#L*alt=1
zWnm9)$%}~|ie&TDVmZNTvA0)rZuXK<eU8j8EwhlF$CrHnEVREKJG8Z;*k@<nyqYo*
zoS<eCm|^u?#+SkMj>XW3-_KVO)cg0n42xhB`LQzb(@wroX|%~A&BFXZ$_*z<R_5?x
z0mHO&T&D%N9VW{@kj^lvMt8+&k3Ri|(KiVpZp;i4R37J+%IBZe_s6fBbl=5Hyypxe
zRfp-0i!|1K9#@%B;C1ykeu1K_uv#YW#>2jhE~Ik&vqW0jruWd6)rRd1#HefaCw>3i
zn^44H`$kX53hXd2lF`KYr&eIdrM4p8BtJhtWUHfFgI;7`K@wH&lmPVfwKqt=&iCa}
z*VkhLHiMVC-V;?kHTB3$y)br2;=!+Be5Nmvx=mh3nZnnG&aFoG7pCp0;kMP1ZKUb~
ztx4ZWC8-{wjOQ@1@9ysA;H-fvCEQ9P{3SXL@HPMd_oD-lUCRmgBR@YM@HCPl{rnrS
z&dr=5PnL494CLP-p~ywe-aGe+$fu@X6kfRhAN+L+`T@=Iw{Wmvk8o0=C{$T3f-c6O
z`btZ<TK=6IBV~uBHY37+2Vbvw-XPtQN87e&C3L!(loDJ{y>aC+lm)$!olnaoLCrfT
zhuoZvag;gK+Z+K(Fzi+dTs%_QBI!gSWm{WKZoc}$LF#Z6kLPU6YSXQA@HfzU<Abbe
zEdS9_NTJb|c3R}NrR90hF(U*iteST1_V|EA*GDj`<5v`n>VlL$I)N-rT4$7dGH}$K
zPb6^sTv>T}d2@3HihyLPC8VbuGHgW4JoHzC{>aX{S^A6{iRIa9;JDT8>Y6ca7ZPni
zANspr@mo`Oy8ZnYwcccx(Ao3}C5;7z#f#1B<NVZFlfQ?)RWlt9WE5S<kUm-AwRb*=
z!QbF~-~PhOC=*}neYD+Kug0mJ=1|)`cxHcS;(pc8x{Qx~2JE5&x&k?hxBA$F??0`(
z%puZa7A26P`cZLZ|LOVE-{oOX%rB}MJ58392#Tt2svlMq{xn;}%W%G^vQ)?ZOu!3|
zT^%AP-sagy{C<d@5a`0`zLdS{K&Kie+ndGH5aJ(5R$jX=X@RA*KSSSEh3Hv@bO9YK
zh_i9b!!7Sekp!SLY;$mSi01$BGAJi!5fwtEUt)!Vd6#DevmE1HcG8db<gWB6u5O}=
zj^kUWz*tP2j=4{-Nuz;7W4aBZ<~Yln*HFs@AAjhwhs!+CLq(bKXJ|=-D(52HDw3uX
z<AgssjkmJfByfE*MQnmM3Bmpr*LHUo0sx0F#S$r@El(xtVw8+5QC$hIp0M5C^^9Nj
zf!>&y%N6S{2ZGqIQ^?5F17$#nl%@WKbdns!^M=a`+s!+7W|4AZg|adMVziExDM4jw
zs3M&uwW#pNoo+BBgtM``V#k=h&!jB1Kok;pVl1$V0sqO9aR9y{vmXa&*DgSM2TfNS
z-)p@{OC%Km8=zy$+QMQIBQ)bi#w#h45I3=#Onc&_RkCE4^)Of573ogI8o<k77N3fN
zG8mk{@=owQT(R}DK2(7>ZB&n(3E}@IMma=T=c6hi+`0%F>!7Om@<eOWyR^~kxNg`)
z*4K_pBf4U?K!&VQ2)oKC6h6s2?bLqCy@uc&aQI_;9uYF8Oi7(#jM{#4@exi8w)Pmb
ze91%26O_e6drA|$r_<aXRk--#SdH6$8OLNgK+%0`>wP4YEO8N{%9$TBP1WA01=`BF
zN&h+E$e!+~qd=jtmU?REmmB?-E6JxMs8nT2P3!Gw?y!AHmoBO@V$iTlQUCJw%Q=qH
zk~ccxjIMWN3kKDw9<F^|-f~~eIF(^ydSgtjb;2!dT~0qzdpakGbf&oY9TzkGhJbc%
zevay<vX#75z^=ljNP$w(V1E@ZLzb8EuYT_>xZSh;+KIG1k>)9#Y!IuT(R@DuMD^=>
zKHytKB`h3>1PN;aP=7>^#F^tfb=x<QkHT;OsM8Q?dY95SW6?#?Q6_SS4FDBc!jXL1
zAy&4;SXyhljMm19J+P50vFPF%3!6n(Ey>=)w+BXATIk-w8T|I1WwoqCRJPLc^6im(
zLejZt*z4=aypHH}#Wv9G_KL~lcpGHq$MB9?!_a(aa<bU0jiW7b8gjgr7ut*d8r24`
z$78RV9w(M&cFq=rGg1lX53(s<N1z1jU6)A=GNN9vLFtFOE6GX4Xa7kw^jm%3o3gp<
zXVd<Hs*{bM4K0H|*ypNFH&g@kDMMsJX&%A23!XUuM^H6fgkv#`Y?zr}V_p^35yQFn
z2_m*s<d9S#b4R(2RST*|(pnLg5D8NrcP;3+T9#5#-+I4(3|MYASxv>KH4%Ez251lq
zCmtOkgTyN>bu(gba@m9DHtKDWstihX+Osj;e^7a=e$(%!5*^+gD1$Bq4TCu{&@u&E
zIjW%Vgg7P=6(yDgX^_KXkWW+oD;z}G>zA?q!HOGy7c0#xdoUJj5^Lactbg@Nl2idp
zV1&GKQ$q{qo;8|Z4#(#o&&OmcM#LBij}H_+5YJdHc<1L9T%!Lx_p!e18EJFVN>`=j
zpoK}E967TaSA*H#a&cx`XuXO1;g>HI=+Ar7^1*AZ7~%KFPcmjZKau`K<~56A63rio
z29rnuBaWkQ4ITvTs;3Db@4jL{oSy0L{_~t|jq(p_E^t9+*KeOB^RkHYmUW2EId!!v
zTzfQpXI;yQD+O0ERREc!)bbNRoZO-(6ISnxdB1}364$9yK{DF`x5n=5R!g#yj>~%)
zY7t`8o@jO8#c`-mP_@RN`p&sYW1c8~!g+4+QLeF?^-F=j=wYr@cA^5av)?WlhY<Qq
zt-ZRdpX!YDe@}EM<gCY;QF$UFa-YPsj)W<hkK-3*Mhsl~j&qgRqig0=_L0UoW6P^#
zD)r@M<*0kWf<q`OJNMum505QYRz#WxG%YB<(JIc)$sHfGQFHY8@mml`0omkp6Y4=d
zl07o?Tg=F7$%G3feLc-|-`Bru3;`EF-F|<_sN4W2k?2Qv4M;3b^?thK(C$V0&aigD
z4QK8iV^*gxJu&RV&94nsc2;zC387Qe;mbFS=ozZ2d(;5Digv-Kh5ry<8wsRRdew{b
zlk@<d2g6b03L8g$B-X>F5+7wbW2w7|EjO$p3qlSSysgss%@CZYXvq*(xBg}Q2@`+j
zIqQy3EBCyPur25OnUSmI5G?34x=C;0K`NOrHWs5>8#y$uD%Rxv+aoYE`g%pnw$Xib
zbe-tYK{aX(j0^cui}P-fg}5sCi2;C4_+44I>>|S@nC{9-u#aTL9YE_8v<(&LnOWz+
zU+R3Qc|NU83AVPvnhYP<h4!#ixaYLL+0L$0bkG&!42?bc8oLnhWG<Pm7WFe@Xih;&
zO38hiTubcT>T;Z8;5kOledM7jyH$fv>uDcSuD__+c5+zQ>=~QJ2v%nqDRWr5gi4}n
z(#!EV6pNgc>Pg4W+~NGS{_-qJ`*O$b;Jvp9ORp5Fj8#^PamSb6KQ%Wir0OK=UX0VQ
zut?31i-MTB{h&;QHq%}vIP(DbBjOC91V%Z6AQ>264mUCY={X1mpnV!&3GvpiInfhB
zO~{V}m~8IcVFSv70RzyGRj{Afw{PE|)plVoQpqb;+2_q$zP~Lha*a=x9hcKJ4jl7(
za){R$5bxCpZ;V@O!2YqS$qA=mQdLPOIbD3x>C4%(PA^6^`Lo^It6`}3{CT|>>5#zf
z-53g^JURp#FGH9j4Z4Urw^z9XM{%?Qp(1e@LNFhktjEt$gvj)S&sff07v5FtqP}*}
zhJz<=<v^REZbMHP)B-k?|A?3uQ9i5Bd(2xn%opq^)7$@?pFYGik?YX%iD{|fiEAH5
zj>%uHJJ?#N^G;ttUAtP;B2D#5QeX>@5(|$Ct9f8j$<4+GZm$Z_|ImADrsb&ENKDF)
zfW#!h;*Kf^<o-)I*3{(xOI85oxQku}8=rm%*MWZl$g_HzC@di0fIiO%!&mTN0_=#s
zZ<Q=#Iiq!Z(4K7>U!;jub1bQQ{MpC>xc77gpiq+xaURFa4Y(4h+RSAhA%h+rIz|i1
z*jQ~9^}F8{s^zdxx{U>lUL);?x4p&vEVL*0x%4R4rH$CVF!Q16%1<<Vd<?QeQPE;U
zsmkdAQ`6)duQhk{9zYK6@9#fu({|?8V`e_a5oHzz28I&q%M07a%s%3<EOA@mQQ0|)
zWb6z{6x@;OW`+xX1PUVWM$78IdTvj`{%ik!TYMElqITXzb>Xfs;stVe04bRI!L3A!
zi*b@=2<kc@v($8S$oDRfbiYE*clj=_+#yNhYIE0i;99Jwom0Rdg>cPVj@iBd7nwRA
z@!_D7I_eRyj1iUf^qyNAd6|3(9BtyxjEpKJQg;U<KO_oFkB%B&Kaw0E==Fp_P*uHZ
z*OtgTT~!Dqa2OLCWp-{3<3?g9V{)DD4IEY2@5tU3?;gBVchO?`-HgA`+0nXhTV3pl
z$S0~v=n8H7*InE@Wt?YcH_F8iZyY@(AnWqpd@-F@yGb$HBIa$=c;$ti5_jyBy99L$
z4AS-{f7{MnCF<)I<+GT;ZJnijrZTpsIti&o;WYg#kWYhKIP18I)DG)^D-yntQxPnr
zrKzckodpaEfWb8x8G<g&zxNv@6vD^15=!k5(1kMOcaRgN9>3krefKhpNOQ!WA+Si%
z;-2Yg+3#0;in@J(7=fz{s?LhR;zMK)JeZ{Rm?gO`gv;E>=%h{pHB5?rHZits6szW`
zpRLP?k$kk923`ZNlc|t^qDB#TfAJ0s*^&g0QSCS?Cf4_&O&Jg_JoVI>LQ52pt<Q{%
z#F0BA<Lilg=B6+Qy9l=GS$?j+!AB=AecbF9T0BkLSm$CgcM1vyQ<d4{0~ub0OW$1!
z(NGh>2HCjv5i5j!EFiVu)8QR5+BJloS{-#PwT`{#pNQ$e<&9lT(9I&nCiW2IAZQIg
z1N;lFJrR_TP(=_82`mu?rAU(+Jt5tN=UCD$PLk~pp8TuwGL39Zelw(n2|KL~p97-<
z8Q8>cK(}OYi(d?KejMW0`e!M~nN$;~qD34}A3rY1e<+P!G>orDEnfDJk&{ZYDtq}=
zvwnD+qIDVAeBpJBUPW?s>B<!japqv1?R&fie*QFR0SbBdo_dcCXIc4L((@;E^FNf<
zuIfF@XA#k?Yx}}@&F-OX^@Gle&L5t86{Lfk&n?uPc}@0`Jz3{8nb&m_ON~l7Bh{gP
zjqpF$T@EZ<!}=;I=T&In8@An3_PMM|VrI9|!jQFQRD4*(UV?Pl%J}|V@0Dn3p^J>P
zu1}Jm6s>y=#-BqmDn%-^D7IJ2VWiIU3Jvy6LXmN}tvQ!v=0d8^&LcY{I$^ALfB7~^
zvfZ<wonu*5pFiKM@=f)<4{kA8UV0S_rfM_cNf}HEyNHlSMFOv~4`*C~<wLe=sQ_{v
z_zS|1A$U9u*=99OrmjLB!Yw94eJ%36jewtA6vg%=CXSkk$k2FauY*-G<mC75i-F>S
zSxi;z@ne%dBlI7s8pxnoR*^|{L*8iokQ+~3!8YUR3wu`yDbH3jff@hD4AlpL=Z<b6
zD?~jRK7V*DlaVncd8dlU{Kwa2l903!(hhX0n0d&8Y=Je4&$J9V6Vz<+t_G?qJ18$d
zJ;1KB<au!3yR01Nm+EqV1AARBIlQ;Bax*?CBR76b$sGM?4+^@(^-E4jV}Nfbi3(w4
zFEsgeLo5_b$i{+q7DEtm!YVeU9NGkEcbXyKWw(aVS<z$M=tkGan*1`!uWv60yPn5f
zLbinS$xH`!5C#b>mWt*06<#X%*?C02*altvLE@DIJ1Cu_$Ai*M<~W$YKwW0MI&B#6
z!0&-=h(SXSoe@tHs{UKS&$|lFY_)7=a>?<FC9enAGW%fttSK#1Sg*Tvt_1O|(58bj
zbaPlGWi>14=<YLjb#GpEY3383w;3n+v?15es2~Nc)|;nmAOx<Hn!K5hF($6SgDj`O
zWKYO#Tq9j?%o2-(j^bvC&NshL&jN>l{HNY(&H$c<oW1drK-NMkKl+4Tm54*N+N%0M
z`c}k!%t^6*?!T}42c###NqXg%aGohM!8+$dX8HoyK-H^F!8CS$GC5D6h0%pWLaq$%
zBp<qp5Oigugw=XRkHTlkJT=DkrvW1%R&dB_P}_s;GOs^^p`sVp-*S%p`UU?Rm98?7
z$DCU+L<je+>b+wD2t5S2S0?~vhV?kfP6<s6ouB7crEs8#bvYxxKKo#j;s^TPaZeSa
zSH50l_SpiLX)n1w@ig5shCaG`J)WP@xVkYaN(goav!l>?HAa#z46jTX>N~_b8&`HT
zeS68V&<r(I;S6{E>*5)^*&~eq;g+0@alO)mHQ)-n#_c_=B~Ru-Zdb_Z1+kNiMfYkk
z^z-a9%GLOidG@S1PP4#A>@UE3sE9rGI^qVsL5o=~=0V2Bx~+~S)Ro4&t;~nJoXx<8
zsC$49s1$WO-UiYNp!uq)n#guBwfqJ>f^nTf3EXq|>ZABvkRWqonX7&<OQ-_KlA#ya
zOwOU2_`-I<<((iKorn`G(SCiE)K@~w)GdfhG7l{SX?FQD`)0BMnW)sa2x!&c^wmf#
zX7!;@e0kT~Wg({;cGE)~B1_q6Ck%uI@}~y`f`3~-41;l(n#Z2z|3tSv-E6k^a5Z1s
z$;@@fv9{e<t39M$Ja<JY#D!BOrXi2wqgu2Z08MfU&pDB=kwJp2%(bOT2OUlorw;M7
zrD^PxfaJtX>&-~SY4hBr@7nM3)47WeSGcI%g$e+`i+WV+ak9@C$5{C_4CxR@!fu&<
zX2OsUta4H=FYK<`GWROCSYs^hK4FJPf|q1veRH|0F2TrS(D1b^&2sI54-8O>C4cMS
zhtGo59|1QcO1lr9!o<u0P=HEQ5C2LT+B66gA6I|Gu-x3-%&9`ac}M${(2A2K9$A9K
ztmPt?S4hEzmH?b)k>}GZQ>dCDAA>?5fI=o7w+&t}(4-LEI>14oEXpJ}tev_NRPYnd
z3?b!v1qpx!!#N2mh@%h`31hIka6%)@b=&w6E_rz72|cDiGbIH@<K)ow15hO>Q4DbX
zj}~AZy*<VaVM;jZKwaUe?Q3kDg6xr?19B~O&|@~PMzl<1K2!&hgdu+1XxizT15~k>
z?raRYH<0Ur|3)UD-ZVDR68FsZbJSM`zDGD_-NE9u1b28Hf!ZPgyA8W<VSXMM1f&-F
zzzKM`zv3OnnkaGnqMXU-b3LV%8Jo#52!0|eYUKC}9)vTsc#a@Dp#{v#NdXpuS&oXl
zqjR8YiwX)F(db!y5)vO6$>jW@Ne9%~Z5G&;s3ZyH5po%Da!F152oEu%qF)?`#M<OX
zA%=*h43c_4+W3}>u{HqeDBOMe=tpR}8wN`Zw~M4=b|E2Wf`jv~^C9H_^JsyyCn!wB
za`p5P4P|#xKL*ht1ZV7}%<t7;9Q|_0&!WW)q#`4~nX2&_$f5_K&oLwg1Z&=!$PM#X
zX#K&SqMCuuUzUG;4dV&r^gel@cL-ZCc;g3h%MJk9z$bC-&uFB-diU-YoOlp-;OFNA
zi^=2ZJDU!GdWKNq=J|h=FfR}4Jsc#{LqmZWUkhlgjXMpdA8edShd{az)*b{GW1WD?
z&;w@-$nWtVKlV{mXTd<ZO&)!@%B%Z$p?WBB#8o~&@^T7Q%b`sA)$^20wsoZ$n`v6Q
zVjWJh@3_^zC%<xX8f@#)IH@P&klOvhp;mTs$8C>V)7oYO@lZ8~EUHuO4thn?<Aq;L
z-uNW>s6AC3Q+oI0iFxk96f>vL*}e|PesCAorNx%!rV}bH2^~*Lf&zzjNN6&=z5Do~
zz=A@vwQ{<TMQKMN(_McGQHO!1`}GQ^uU1@*jZ2`hKc#uffO}?tUye+xLr(LQYxE0$
z6%$KKR08wIE<Z)H-Q_}0$cI;<BdT2f4q;!i6*~<T)xh+0dQy=P?hD}Jnk#P+@ri>Q
zNnKB8=P&&+7DjQR!`$`~rwT6PCTc;PNIB@fT6}l<=+PTcG7S31*N4GH8}O7O)kCO`
z0M5bkvI}lwE*CMNmrzv8MaC~5-EXv@1%J62;sD&0j2=!x6$Vkk?LFSG_1eN}CEUQF
zGR#z3MX7`q0KEmGr<eX?<x2rVzA-uk8MU7p?_20efIbm8eSA8gO?c3}vAxZG{=7Tt
z$z&PY%c$NXO!`}~Q)v$$K1@S%QiN&${<mE6jS`62AZG>6c4=eMHO2$#FUDgv5F!MK
z9$)Z(ge3_!tEoA$P`jfKyc4`_57`Z@QFU;R8uIEOQ{Y3`W26-p7AENPijp<vx`1^^
zy~A8+3v?@+o2@~q4W{}5n!(_>TL=cpaYg=&B50hYvWX9S@Sq%@$|$(zGKwrhTi)7=
zhP`$8t`bOXmk@I4O=9w^M`~wJPbj;}72czRFghg4u{$d!`V>qsN5`DUbGN@7kt?_|
zvI~jZIxZ>SR$gxI((-b1Q}5Wo#6(JX+7h~52=g3V)zHPC5L+R7Y@oG^oSX@*@v!vA
z5j}%r{KL>z>NlTf&n6IA!W1#5F5F&W|5tPfGVq(lc3mk(HwG3yqmXT^nRjqNfDP12
zxIj1)rC!;w@<3J%Zni6O7I0hE3yE)|&(3_*#NEh@54M6+&LKGGXBJ*YtuozOe1r@-
zkxgLnJRS87aCE@nX(?#p4TtQ||Ax?}!a{BLO38~9h(3BIoLpV&H_ZUE+0XnQ0UoBk
zy_;a35)ys_whelg0pS}ScXg3r7@5ZRk&`jmR(xH%Dxw;o(3YL~e)l}aNUy$D69Ylr
z9#SBfVf%YiX=F3`w8Lr1iM_#$e066Ae~fd-|Dfyrc{ty0z^p>8_a_@f9nCWu9y2;A
z8I$?%{NBw=Icp5}b;vcX;$s^_APiUwyq^1fEh|D$4hGlZs^UDVV5fru59Oj-RJyg1
z5tOEEa&hBmm_k>^62xdwVItl2@$(~gBhDGfc(E=C^ITVkOyxLfHQ=Z!L57g^5@8y&
z8xFCG3#AA(kFCWZg|mSs08RdAM8tIXPBb3kWw3EDi_P2Dm*`Ewrm1W!#o@@VS7oG&
zb;YKYX$V*QBV;)2xbzBtL+J!;5>`R0sMl}`S)n(Rde{$Y4~!RioE_e1lBW3PB5F%4
zBF~N0oFil)V7eg!#*IO*C<I5ZUcCZEga!dn(++Tj9GP_aW{ML9et_BkL1!j08Bwj2
z&ae@56V!+P4e2dYeI;vvqbb}RE<xUngm-IJGSGVWX|jvxHA2$Z3kKBknZn(gMx7cQ
zR796~x0#TLurPtqA&+;FQS^Bg6@~SeX1Nhc9G55`K(l8;M4c)8T^Lmr0*#PL(Wueu
z`64tR737V2``q5W^{ju_GFX3r)VMHEh1TJOX=_`ANeNzy(3Jm<%Ei|uNa*PdW(c^t
z(Qh@%PmHOC9E)%Q!!-q70~)0p+GS(^P{>%HA9}~CY=I48d}s*$;)G#8&N}RT@2oet
zqWVe7Cr8wr`HAk<zZij9iZmu<qw50yXBwS^zz9tUdNC}~R`~tFO`K`*J^$F@U(kAW
zoZ9sY?~YH4>x~+^3pe;Y;sT%r+z07v*DzUuN!UKIZ~gfZGPIVWMHKs&lAPQEwGCX}
zuq|+a5sgYj6O^PRrUY2lYooO-1RV%T4<jQYgppEV$-ux4+ULC6U*94jK{B(OlCl&D
z0m3cceZTQ%)xWTsS?a67o}Pk5d|`-?2vjhT$bQGdY|zig^H5+XHbr&3xfXB?B4F-v
zbtvYEvB<dP;oWzx{GKME#NmOt33_;Wa(@zl55?#Uw=Ab2c#)5P2A^Rem`ApzCIe2w
zx8+?JP2NE%baQKJ<8{Yc@y#2-Y2${YuU_%y2Gq^Z>JEnPDc#|`vrB8j`D@k5%Cv&a
z(LMcVGTp99B}uu=py020ShMR1@A8Hc-&AMz;L6jy)Gs_)w8mtIG6D@ursU<G0{ep#
ze?OYOksTsm`*JYn434U+|D<A1q2qf1J_X4A<Koy&$kES&ol*N@@c^>l_6p}@YGu$(
zgzX8#yo?lD9tnUaH%K8Fq7fj=9ZFqXk=V%|fGu!Y&6(~aGi-X5lqpyT19Eu%-MC)G
zy4I`>F)biZS~|MKmA!#sVfjY&2OBN+sG)WiPQG;L89dIx%owT+Lm_Bd;ie0?hXj6R
zmVzXVm|QXgoFA?hCC?K0zW4SL2`2!^hGp(&P`kscz7z86a+O4gUFqQ3<a|Xe>-{ly
z*kmIW%ZwQ0K_i3^P8M8feuk|vb$AovBpuVQFoc?!sDp7-LNxkKXHc8!A<B?6t*ojB
z^YeWl-BZ?gCu`>jH^Zh-7l$Hvl+F1O83`)1|3Uj4`)(i<t_+A}YI)WnQ37yd;Qr<~
zSg~v5l^3*;7-?x>EoN2Wy$><zYpTb22yT%=#^x=}^#3(WNtt(>rGk$ve1I9%&WU3v
zgJkA)E%aIW;W+kIm_vG3x}mNg42aePUi5V|BgB~Us{VdLYd|=Un7xGX-3^;SWBLgv
z{n_i#BkaT=tdv8|&oHv#ynE-R$mY5~D=Rp2rZ+dX)a&K1T|0-%2%f|DE-m8t3T`T|
zPDoRCw+nzW!sh^$oZ-Mp=vT0nmS)d>*5=sm4hRdnfkW`M9z9=21z0;mz%s&YO8XtJ
z((D%u<iNm?wmNrd1bEED4xP^ah$tltV;h{2Bh9ZpQ8OmzY#BlhRDwQ-wq3pPlCv*V
zT6kI>Pts@RRJVIQ*Y3~eedv3r+T?Jt>si5{L!WPzm-a0$es<WCb}BVHD)7at*)HbS
zMs}WUrqu~K87+MY59{JR-(uD4N4j1v3CPgrQG^<Wwy2a&*|CPKbf0>NwkKowKy_J`
zLes~&^(*but+AcW5>XITBD^5?PDy<ax)|6jVdzsyTllgXI!#;_dK}jFl}{5<65y}Q
z`#+W(jlviIgjrU;FX;)CF1O|C@ULHg5T?=a_O+E+0O}ON6CoEjVR{;%5}{W7L1Db5
zMZKe}zJ3A}2ebtp=Zg*Eq=P{uJbd&hoHdyV?=xA0?Ly(hKrToqT3+7RzKG9)kB399
zLr<t4mNa!i?BJ6GO?c!9%^z47M2kn$jemAX02RhvF#z2uigVVImn>*cjD@9TmZ)qY
zQd1y6dk^tksun!2gv=kC*A-Pk?_Bm3d;|#7G3Gq(-3t}KQ{aIKlvUE!<cfvj-F@Lo
zqX#hLZ3X+3*bj5QV7WX|rW45Z%AO2j0z+m)PDVyXU#O9O8=?c2o+nzK3;Qifii<Jb
z0)oz+CAH`LfDRA}#l;a-u(e2J=rrvy>&zUB7~7g>*iZ>c93eW+ivnAXgx3(PO;v<P
zTq8~j!rr|w;mHxQzP>)BMrJr(e^4Am+XZrHR5Zz1w!|0=-Fza&D+)mEs8{OYj+s!x
zeQqFw(CPdh`>E~xb6JT)qHtfR^`3%*fjoTP_t3(+jxgjyI!A%lQ$4qLIF-PQ2?Jjj
z3bdg|%J2v54y@*4UDNY|gVnGgkY(Q;s-m#0J@?LYJp8m1kt`zZ24|i)MA=ndehUB!
zA#Oe4Xk0-eM@ulu6ZlTZ*xR}JAu6&(Q7`Pinthb=oy|Vc;h3x3{=%sfcUC{Fy3Bh2
z@jfc@B(*qOd9hu0<Wto4*A^?ob1Buyd5<qWwD0?}axp(Tt!MG6obQHVJ*zm$)*qJ3
z?nW(}B5NY4W3xwA?~kXENH<?Iu(gtk?C-Hybyc}(_Vzii0c&aB@8$UC6T(Yf-K>r*
zLl1Su#G9ZNKOEc=!j2idM03mNR^fxo*tm1Dj*c5TPc(IO6reuAy#_4x2l+CR1rRak
zMMVkAdCabmG``%`)wKa&3(3KE^dBImpl5dsbJEd>3?V75+U+KG<yh}dPR6bmI7_ht
ziQ5WX+#OwN7cY99TNY6nx&|F%Bp@Ob@sbxWq8eXB{qyVBudpyWXqmt_;~<8yK{kr3
zs|vnr_tsWckRcH+--OAk=avVA`$)fKg4p13{tab>`A0ttF!G-IwPBRCmAcDoU~*Cv
zKpvoY&<&e_ZBWyAKYc1KD{BuUMn}gCrFW0vp7^|yrKQ7iNzm>Kw$j}4a_GS<vORl5
z-2S{D7c;7}^0qgmL|;U_R!(wp67rD3qMu`91u*=^4u(?w^tp3?aGs(y2WSnrDWb~)
zT`(BeML;Y=d*Ud?{#?+VJE82}9-FayY@}N|Hzx=F)6g+&04F3GhlrHGX>Jf{0?IM$
z3Ty{W9IVJQ(Lnm9=m|(8wWh4dvp2kQ+n66G{!7vAkjTezg%l3I-9S1F^LPY^o~|y$
zS%a)6p&6LJp)w5k2z!$WPV68cP#hg*VnSBE02v-(+UvLs^n%b;s!)siwZ&vccz}NK
z3ufe0Pwhmof?j48t}B@KLx`Y&0A=Uqp4NPS9mgM99*|A_-YfHKWMYD|r{NOSj<y|7
zTS4|4Ud_y9;%L5OWAoLGgXp$_8*#*xyMC#MUZUbrs5S7yL>~FVI+6!GJCPFu6@oh#
zTsUDGIRX=fpFca0Y0IU<Y~zaC_OEk)N75r%1fm6(>rJF71c{!?$H@GTGmjJD`V6%k
z+HZh5jq-A%5)}|=2bGB<@ZYvabBJD^S4blR;J2&(?GXbrXrI!ET=BMlwv%9)695c~
zV89(xuuDUdjBdo<54Wy<pu322wbJ4D-tYtL%7#c8h(<t!CFJn<u<^T|3Oh_D7Ahri
zz(x*Wyz^*7IEVr=xF*3l7!IE>Ji#f9SJOkg0j~78Q>WZMQF54i0a=N-st=FnO&GYg
z1$<6K(}HdF9mZx0#2GZZt7zR<q9|#;i;#^4j8E}jXE?b1!ZQuHETVBKA}lH^GBUEW
zv9b3+O%t#NFeH@O5!xq#J1YPQLj4S;2|&3n*dYT00{}`T>_<luq6rOPAyqlT7@Klz
z4dfH(sBoLR#mB?b($tiWycX6i0G)m>Ej6{iRKCT3yJY|_VA@zqe%{{5v6}lmw-_;9
zo|<}Scvv-2@eOXh*<h!|K`V9ll|z{>Q3DcV62BiT-1uokYF+xJ@JzuGGTP5)IyE<&
z1p^Z16etQZ5^NM)t*7+TZs)L!M5vz9<%STjGw(yQzW5%G!-v<|&YeH5NA)na^@Ne9
z)knpyAQmwpedp0~x)qX-c?REevtD}I#>FPZ#!{<LQ@F7;J&Cf;&D46pLxDE^<vWY-
z^Kv~`Mn?iWSs`CtXPswm6gsnCpG{|v{~nFpCBEsT`G!P6C7@9<O<Y>knJh8(7p7k3
zJQJq9n{DIk!QehKQnKN=#ObM0LSZJdrQcd&<LMXaV4>|d$PWsGV@$iHt*@=CeFa_k
zzw*9PzGK$+uoe34oH+g<Bt`RAr|c{eJYf>LpN)~rNDw09HH2;<`?(q3gYaf4nNg$G
z_u9k~0U*=X15gTvXX(f<{@nPD{gz{~A`X3Gd#W>z&~IszO|Hb|d4FAEj4cn{Md)%v
zQ;gkv4PzMA3?W8J)%T!a{(%7KlI3(|8LKMU%1#MwU~+N_hIMNT+(DrP9*oMDX2%;8
z*_mnK3t@y70^M&ehA5a^I(C1`|A_X8u7>ZB^WVgegxX#^?kjuXbRxm)2{6k<kC{8G
zpOSvD8tWO64{Q7QQ@=hP^3z!-`}Uh8$+u4zlRp)`K^`4(U-2sv27p3AlD*Cfwc|SQ
zon42Q)QKUU5vWiyPqJbs)Gf?qA0dOxFKCLN8)nuE)!(-DasHKEMN~jl;;S?BAD3FM
zCm=>l>}<cgdJjlkn%y>ha(L*TV-+HOtT;?>8efZxCkQWiz#IDg?`ovjVJ7c7i!UwF
z5vd1?NTIevSpXP<Eo+sriJ%nxsp`3{m=!zWlunZ^t~)j{e$*!k-vhL91TbMv;pKLY
zrH{>IOnhWa_e3AN9iDAM8;%ruIe!z*`S2I8O%|Uaz;5Yh)o}draa}LF)AFLbtn3Dq
zGvOSP3(qlC?qli+LSCAwg?Y5(b7+(fIi;oSVu&3tJl^&xAHCKbQ7?#wI~;za*FF1`
zc2LsqpuE6T&civN$Aj()tPfMV5*!b}4M(nr_AwJADzh_?Y+x)#X1qI~oIwaj)%Z-D
zFne^_<MHz!92__ng4Jy>k%XA)YWH4dTOTuu8C+)F_t)?LzJBf6HNeug$ucuErpJ<C
zS%!soc1PI*lCteDmS6a9?>mXa9@(p={PN&=nnQ=!uUtl9jL_`RwDeOCUNeky+yL5#
zVb~v@wZTmZVSQWFvwuE`LKI@OFmXcw-1s1AKh%GEy1KTx;xos*NVhrDteo3+wgxSt
z2!A2pN}TdwasySBXo4$@k;7}oJ&*1oW7;;XHaxUhUtTVG*!4zvh2zd<`fW!@=eU2L
zMzi@UzUWQ~Cua2$Jmb61r@2y;<V4N>YI-Nn6BCMUFCP4%bD?bv<365H1oP8A$H``l
zG$9^glz8{Qk56tj#3)|kF)AqrLsjB&mU$j?rijPA6<mOS|FhwV8!X9*$91t`Sls`8
zEF6V{^WVq+y)Ws1HzWgnsY_33IDI=YrTMxUQl8}M*W#Y~Ic93SjwIW*4SV45POk3n
z^I#BQZD@%4$2P$?$hSfN<)YhnP!jml=#lz=wq2MGh=_1Orhz-SZJU!xENfiC)%c(G
zXYM)DO#F|B?f;8Q{>+y=%{7ySh2^8u+~-$VCfn{=0iH>RfynO~4fxie1#<+uAHx`i
z2L=?O!ne%Rp2%-OQGwKUhaOhT1;W0m_Bq7<R9Sd_v83kzjHds;_lE!fo9F-Ns4?+s
znuu7N)yNkxS0TGM0%l4g`}gf5e4iL8LO3-}15m9RBcfeq<1ql)32XSzdt2kk$dOC<
zwM9VN2Pqt6lBTI7*cC72QZa~fTa6wj^I282qx6=l!jFeJ{>Dk$Py}UKHpNexX^X$U
zmTMBu*Pqz_hTY-QAa_u&UoowNmlSA$E2g{gR#g{?NPMJ&yFE4+#K1ug)f%?H&(i$&
zS;Bw`Z5<t(Jw&RwE)((h-4P-TznB2deRryxKhp1<urOqZWo#RGuY3Cfh!|eumsx5^
zSOx)1yfQ|PXZd;)5&eRdDi8$JR)z^X9f*PIHBtY*i*i@eraClsM0dysUi|%8n0Vn|
z1*(9-vfPN_kWh8`b{@jd==Kl?%01m=WL;1v+-OfX`ap~<Prv;4n=IF9Yd7}xo<DNr
z(=Uu@Am+mV{YWJQ@@X1Ab}tP%7?A(?U&Q$HO#i>1YZ?`g5{yq8^eb<w;W^)Z|Nb~V
zq-$6kfC5|qWm}2&He~+yz0+|0fK@+!-Jhay<{O5;ZQIs$_3ztja6%A|uu36L*fGaX
z5MTbkZ(?Qg=VcoZ4z)Dxwpik$T@d*9YE9wvj(YWqTqrP#SO~Td<ltWgQ5daymDxRZ
znGfs}lBdWw;`njR-oL+{*(S;274U)Tf3yG};X4GMix%>~FYNI5J3EmQQ{7P{7dDT9
zyl@uAt+_Y(_x{c9#b{dsbg}WmU}V4ojqSv^Xv0H)zn|lS>60laJWfF>()IUQr;ZbM
zeb(wW3LB`PvU}^AFp+56%<g|b3<*9Vk)iehii1s@EAKL~RJ8t{@&i$CEcff8{SK^2
z<M+Q;k>cNXxq_S<YLpn!roR{6(fcytPPzQg>J%znAV@;Jd$7_L0BI0?Kv#h#0Rk|-
zaU(#5I4%k<A(aAQ6ZNg+o=3Y-?>}_I0BB*YRy(Fo+(NeV=QlwA{P#sbdxU?}yC?B~
z;+gV?aM?TZG-Zn6PfL;XRVs$dr!<dwJ@r18xbG;3(uYVo*Hb<pDiSqs7EFCueth2H
zz2K?DBOEO|BrZ$D9k(27Si8~h8eL#w{3I<aW~)D8WHQ{m`Jh9Sz8w(&h==F@W?tL2
z4N(5QrT_k7UE+W8G+a^f|4rYp5=e<#_}|Co-zD|$|9oQp--K=3wm%PuZS>#A|Cin*
zBH4=gRQJYFJl{n{*lq^z)QfN4!`&3cIVb0{Cy%8*17B-x4bnml9{uv=F-tz6b8uTV
zb_Q&vs(PUMK5U$UFsC}g|AU5>R;Fh^@roG-|5+@9xIlG73OoEy1ZMAEb9DeqQ4fbl
zOh5Sse4d}5pGH88k1r_a5m5b*quo#)mC<tv3rDP7;N|7D$xs6rV&r0s*ZY6id(*g{
z+wT3>5R##&jG-thbA$#&g(50NQs&Y?RFWt%hG<ezNJVo(gHkd_ktP*Vh9oIM2?_1@
zC-?n)KiIGTYw!L0d2(HstG>hMJkNEmbsWb!mh3PfBHOH~;r&l_Btx>^<nWl-Q+c(c
zenaAfQZL@2pYUx;#s4i(ZUWH>fz!?3P2TZCJJRr$PX88FEIcXc2Hnco@tIB+5q_y&
z{oi^?Sp%)aTQ2NK%j;)-?0C0Z3W&Vp<Mm}FV&mcl4;canm`u%DtxP`)(?DDa&Tp^m
z6<)@$@ZTP|Wd)}K2O`&QvUDl%ydWPsaYEAGgLWL-W=CKlT+Yc7Iq3Wp#_z&_g}6BX
zgsP5LS^h7_m+iJs7V$59li;*;@#5peHS5=pS{o}Um$10Oy%@?eG_?K6*hhth*hVx0
zep1unZ=2+~U8o_n&hBmLs2eW8CuD(_TzP3_XJ_8<29w9#L>6vTF!UdUtfBzNE_-i1
zd$xmel_#)^&<(dBRkTj{Jh4;$9Spx5m?pDF2j|K?Fv5sj_*L*yESTl;&I9@3!rlfc
ziV<sas;~(<)ep}XF6mBw2lwvnY~KoleQ#J~bo2lj8O?R~L~TddL!EPdz?m}suVgq;
zu2UzMKKDDSqXh<*mQ`hCK`lv;nuj&)xPGbw9N97;n}WWz{qIcsgZz`F&^;6_tge>j
zzhmZEO3(8ed;3S@mgwvEKk4hdefwX;6{p^b%weSM?UIsY&;fQVQPS1f8H#*;`56@}
zeai56o79Al5Gzy5^wLh9wlX(;K6d_EvW82SPMI|6a(cR8bmZmLBW3g)2^x3gMfa%|
zfIVG|j=n#RzQ1}Na-0Hw-jpWpe}AKmdu$s-(@*2e?6EH-IR~9Q)w}!)80d540!&7U
zt^;y~He8dP4mUyDTt0)_Cz-1(mgoF24;U;p)oIC+C79JR-BmBNW3ReM{`-8nV^vkl
zo<2Rf-Xk5UR&uc5qh2Y8IzI6o<wVIe;bQMpFLkV=ob<DtH}5`J0RFcq0B;*T#@aW{
z!oc7nrFyjwlaz|e%cG^g*O%FYgqNgCb^NZ_zeBNMqm@+}2p^;-N6OJ8<;s<{<1t4#
zmFk;oVe;>&4E)bhdt?8NxU;tKT%pkeW|td@$J{bZw6n7_d$eNudB4kZU0l*EHVcpU
zXXcuJVeur`e|)_1`0?WjE%ZvbxCB26;WwZs^)Qrjsh6^spnFO;?E%lgD8fuyO6qbj
z!<`*R*sNJ2yLl0><+(-j-?r(Pkq)^H8!Lz;B6WL}e=^jCEwN%=$<F1gR)s}Y+zpj5
z3CFU(+=*80#;$7p4_2=eK9KwANgZKc%Xq{r_fu>M6j@Z5r+)m$JPJ|nhL_A<zdPg$
zSb53Sy}RCuwjQyMkBf_A98|`xZe8{YXB;%f&B(}q-4NmK|6;23?={r{)&VfBt5jOh
zu?ZIZU<tPG2%@yJuDjUYSfvq_YP^g(mOY+${tK-iqa5ZYY02X2WIcY*_m3fW{_}#j
z+)2$kb^0J7GTFX-T)Brh_D~46^2agJv%7sPx!|05FNe}tf0Gs~Lo>5Ml9G3KK1G;w
zZ<v*>t-w~7XAPg_9(yWS*pTky|8uk}P&yQzQR7E3H-R>{IY@U6amjkwDj*w7=;eKx
zSDOVs4Y$7Vg~?zzq=RWZW<l4LAK^@X{0mtVw!CJ(nLAtJ_q`ClzG30o8ziNjJjX$w
zF^<iUFK2$u=5Qs@Ox!_JQ&ZOX`Q^1;FD6GtSutXE&6+ivHjQgI&s&KPnzV*?1#BBi
z%geE~JbC>fl;(%Bgp&t?D3yVHbkmukGHqi4mOjI*)3DWAkKzrPsIRy8SS6*&s;Vx|
z&O_nd<KmR;pV8L9$;x{nB%)soWW#5@HcJp<Q_cJ8>XD7(@+a$l<0M<nybvFMR842`
z^5sHktFC7R;ASpHq2>5VOd;ex#=DmxJ(0#rNlP2q<61KLbiW}(l$T7`zL%eWYzVdi
zpPC-Xi{5$lYPH-J^3qNV?mT{MAh&U=;k25!^*uUHRG|J<_cJLe&ISg38jOgk*5z9`
zbqwj#wCwV-$#Lh-od6Ukr_@sVyxU{XYiK#;J}4V#Hd{DrTJsP9gJ5jTMwn?tAAIVa
z0|!Ki7v%?Q>O+|ujA#@;gFKlxe4(>wW4fFc!p4`f|Lnx=tn?VseJhe8BCcQ?VciO`
z%bKDra(O<y5mGpLnBsou85q|Pm+%qkOcsM(qIo+pGt+!<AK_k!{rsO@Q^jkc?#$i!
z6eU%Nc8y4vIb3P5NicLsva}r{ZbhV~enSmQ<>t73J5C>$u8$9%$9OmJVvTKem$$r{
z6=7&*|A*bi$k+=YhA2VaM!4}Vom!;6cibNmJZd0P-<Bl8MB{qqs#|Z`l)rO1?~`K{
z&1U8v(Z;N5`SPbE`e<wuxt$|t@DN}YILiwdFt=!t``7Ajt`*cP<wZq)S^jC&I}T%c
z1|zum^t6i<X`qWP+qO-eHcdD?Nl6YyM#aFZ%s}9V9^9{=|Cux6HVQRaryJeu5~qv{
zPZAVo1ZdN$|K^XeS#)S1aE`2BruAvkYJjPKfByfJGXDQ3`Td{#P0q*FclW%G>w-y@
z5F7jo(06zrJ$mY$pkpt;IJEr}mxE_!A}0R?rg(>Liv61$TVBF~nEvM%$T-ZJ9g$-Q
z;`G)nLatHD{bHN@z{Bc)xBkC#|NlVU=eaYlsK^l4X6zD372_QqrRO4<IDUtTI*L;9
znMR#%5$1FkF7*C&PJ?v7bh?=5Y!bX0uXc38@eT#k$1s$Ji|Lni&F9aT>gRRV8c*$T
z;rmU&daQol?WfvH%lZ(}5sRof|7f(d%Y)BpX=yPn8y?$+PbiAg0eJCCN}eFcVYUcn
zWC$Yggyq74!$quI*4-okXoAK7OV}+$0L*aUPe(cNjJ^ucY+L~$R4mWNEZon1Mc13!
zjxVw+k_8L16Qk2MzX(3pz0`{Ok*tU@^!nSjrPiG-DKBTn1VGCb8Vk_&&^&eM(4qL#
zJq>-)@W@Y}e$qU7+XrLmv~KlYXhbm7wLm#UQqSk)rD|(ykxrq3AgLxBrs9h46DyCF
zB$H$iod~@q3?o6H&UD;G1_rc?GS+T@goI#4{c;Gc2d7}T6vm7>p+F-`>)*D|E-o%^
z_Fdf{i^hs0?uokN>i64nqn#b{)xGk)d-vAa@=)mDstVq+katI26WhL@8gFidUz(H9
z9o}9E{q3arFxKe&DKC@*tJDg6Vqbx1cJy5}>^)CuY{G<?$}S^iJrxwSmFtQXA@ZM0
zc4#RmDJzSKiH*D&ch9!uLLOWl(trzMVW&^quU~(6$W~TPrUw&Io|Nptz47GdBM%o>
zE~WjW>$%5@N=hr3%*>dv9~$pozs7ZT#jK&-#eUEt-P9+i=fRDn%8^#7+?_e5NcP7z
z+>G$|H(IuAz=QLbcPG@me?Rt3&(3os?oqmB+*HfaTe$G8kA<P(7nC^mc20fHUP()1
z9((Em15-XG$+}>46&gC^N3Xi-YA5z9{~ouG_4CfUOh5A-$JYoAZCdd89ra2wsF=(r
zWA)$wR#x9!wB!zX?U7?6X0Mv6bB%2g@6ao@gv7zN>Jm5th6;K^H_|96GDk7V(r{o<
zU?9Vk=$tq>ZS=l*ic%BIQ**Iqp6Y(A&{H2^`3=VFl!nZcYYaugPLsgUl$b}@*61Ot
z4vbjKHe^yrX#n8lK`*aI67>yV7}Y(0{sV=D!DJhbPZ!bSBPxpj-0HP!<z{b|beG=v
zl61^F><*cZY1IJ3D0IflmM>qr5P_+y?S6hqn8@yUe4o0qn%Xsz3B;L+@$uEMY(se9
z%hZxjW|?b>m;_KArrMR&^EzqY{c!i84GLgSL41{&nPT(Op~(58q7>3LkXNULP46tf
zmnzBg0yeNXrt6I_E-4XYj=dvLIk|m&Ho03`ulkV@hChkIuW*FD^9v4{>YmKKc6@li
zfkTIwH-o{5qS9ZUB|yd4<~RAO(<w@8?2SBJTIx-yE>%(}Vbs>$J4{WBo&%#ZR(_1U
zJj0F<SA^BbQK3?!`Hdee_T}+WU*Fjtzow?)G4Y&8bjXmsvAga2#eEgsT#THYk}-3o
z>-9j3&{x517IwhPF3Ng_hDAUFv~&eY?V+?$@!eQjTpXfeC6>lX{()e{NGI#{N`de6
zyg)B<NLbk7@n2h7(vp({##bOY8LO-;RXib<b7lVR_3f^4eQ_!dcljAO#BeZ{x~P)^
z)3Z^`96fXT*|++#M+F5w=7R4A#saMsrtRoddJk4PAam;(i$?0q9?%3B$24X2>YL<>
zFt}l~SPIYWjM{n!hsDF|K7aQ8JfTzGk<k%L7n1YKu8d_y4d+@Z(#UY2>GI{i%*}&G
zG#=){*P&EsmNe8u9&^RL)#+W%Z2Bu$?{`PO(92eAUF`7Us)5^c?VsBPbVxRiAoWBa
z_3%XJRFvoFxpT{8W0}|$39xi$93PHG5_enM4`k(nbf#8J31oVG)eplrorC5mO8E?$
zye}lqX!-IH>?@3vNckCB&U6*ZS*C+cqF0MrI`8drQ&Xz4_>2s*g6%#-r~52xx@h@s
zfEZxRUOb>8XOp9YOf-D_I4)_omewsY?7TpM8|48N&e!f-f0IexAVf<S(qm?}{76s{
zjU<Rm<y^DECsZU`^>R5ozu)-r7m<qL>KGIp%uoh*+@kWkLi6XUcUjD*1?6~yuxL&2
z(7}U+IbQU_<~@7}+Q4K(vgZ$^9f*vSO@C8w<nGy=O#06vL<euMwkD4@F1Xz3Sz*NP
zTf6%z&;Pp{*ultUrD&Hnwb;2JkDqQ9lF0AMbNBfpqm!SXPl!C#W8Lf?e~Z66<FV&A
z##~FQQ}eLj=uvITfE@28)>c>F(blqJ=EmUWVjdBtfnuW{zsf^I`Dz&O6ZP^yRh5Qm
zTPxet4x`sUx;-!~Y}MMe5jBM-awZw)y1ES7hv^n|fxMbBGaW;};6P)vU8A8QWO?_R
zfscspCRGDIz7HQlbszJ`m+=D5BSt#R?h)YYE8ej8S>fJ|`K`WubAL5+`~(^Tdlm1r
zyAT>0O1o3oicdT$!|hrM5$~?)P)X2Gl{2jW(4mw`MFczZ{d5lO+PTwsRLb_@89$sQ
zo?mkw^C=b#Y{3tccZdGS+?X|JM30APz$J$aQKzK|Uw*_n+}!x-poE#Os%lzcI5*-8
zcG1{Q;3H5}R>lE?)}kX|SbM`rrAh&Z9lx^n)2FHndXkI^VDuA|2Def{sOnh0ub*{x
zJ9j-1>>F5@?UQKkRY)YDVPp^|wvN*?H5Ht%NhY2H@M0jx3#?zizDEx)gmRxL{!&v-
zDq?~e*MkKLFKfPXWkY>Eoum0rpZ=hhGJfpzi^6br`|l}hZgtYo$YK^<T|>ikN;>*V
zKy)7F<@sOLa}w*@H#$fVvKJhG<Lu;wlz8kC5$juk1-~&4<v0$^ZtC8>JBmXx07}}~
zNNImd_LS`3pH3lRFe>md2xi6e=O|(uk*)V1zRJ{8n&*ajHc~_){dAimE$)MgieC)a
zX+t(eH}r*g(*x3qBtP=<2bRS~M9f*~ZYV23*PfS*vkvhg(#lDCxZz-8^{KZr)6&wI
zw}7c^&M?5IE2w|<?uSsT`h{-ZR2o~1dUS(*JYITWMrjP+5isIp-2_*TGEI2%{$2y6
z<WuT9h0{P%IJ<q(-@JU4QdKdvK4k_-8){n3Szp2zS0$?nqe)MU@j>C~k)<=ba~-=j
zA;BJ#6Pu!NfjZ6DX#yIK_ZmM!y0dZJx=0Lb)~<bnGL;sA;?FB>&zaLJ1XP5U&;|f*
z%^K!sPNF%5ZJ{qM&sr-g#)2Ya09W?pi9K>%x+8#W=$^(?F4Z7FGNt*Jx{RJOuJM7t
zQ^y|RGT~?Y3|7I$Sg*7vx8PbLI93)!h#k1I41TGvyib;-Pa2P)c+;`uA?w$_WjWf3
z704!nCozWeHkw&fyvVG_01omaUAyjQeRl2Gfdoi?<jA>UwPpQ~O|6}fUH|dpqO5rd
z(a~3FwW)nW@|pT+h!iU#T$Uua<`HMlRuQjqEYtdnijW6tU<J>vvO6ve3k5*;-;<e{
z`S7>+Po@|YS<4+DHTZIO!0SD+(}s!mk8gYXmMIRpE*K7JN!;V12s*X$?(E>kiZt)H
zZ)_qEzHb6~9`f+sy&51cnTXt1gVOqrJUz=py}@~d&({Gl<!2w+K790OWBw^Qho_tS
zVIfL>Uw+s2tuW?d-isez8?3Ba82;uHykX-;B<KRDbY6ioMx#%S=v@0kpvRoeYH0Lc
z0BM=8KF4uH6I)kbf7IR>m}<u$-MKlxW$4oKYLn0lkR_Ky>k>Uo@$SzIA{|GzEb~c1
zT%4_;A(nYc^A)WFu8`@9%n1w*7O-K98c)wojhEqpgsX?*@7hY6EnD``J<>RE*<h&|
zvjj7v{P)bo%QFuNeJwklcAdD+L3>@Sg6YFjmftaFPUla3kgnJ8K|_2M^M20QCrdUx
zM?+{c5wnf2F;Sh>NC>J~k6nbm%O+#v{%(oE!F<C3DL(N9*N>|AG1}OrT>Wq&2Ou((
zkxZpch^+KVcEdU~{#-<j$$qa~%K+U6>YEOz&2nHJvCd4KgXILYk(xOJ(}jE2adlul
zYd|rPA)5M~p~p`7O3gohym#=;<zj_Z$S=M?u)_x5VBd1{$@t@++Pve>FV!DLX+FM^
zq9VxYlj7ncI-VRYWb~^{eySes=;WmCR3ke73_#-)-%w3Al{pf0$PRQhowszUGhOkO
zm*%Zn^>WM2KMdr#S9OP{Fn``WCokg}IX=EaC$AG}cEfGt$})%N$qFe=L?0oA*^PyI
z?OU8w7R%kHMA5e#aJQ&v67W@;@A!@fc&e8M0dSR^L6F4#2Ob{a)KZ^`a$MQyCe#|T
zxzA}sjTj##S3tYl_3Z4Ni>v)ko}6x18v84w%fLAjJVpw%cf-X3Pn{A3f997Agpr;s
z3FT7a4?ZhayeEH+I%<vJ5>W5`a0jCFWN4b&dfP-Y`6)DmIBM6EsM1oAacRTuzrUY9
zdE$(guBK_;B5*d!q>xE#yU5jsUgRY!_T=3xTW?_@)9m~VZNT8UaS82(11}jW3hsJ|
zpMJH&YY}BKC;`pqt5B>lIV&HWw8^_<B=r-NP~u<-f1zOn_sGg@wlK(KAr+S0Cq1GH
zJ^brXG5+>l7Y_VB7?9WIo)?LBNVJFS!2fCiLPF^11+wLKIDO_!zQ?8w8{|`pa&iu)
zso1NH9})4z^C4phU92M_x1Tr#dqdYq-s}e~iTls3Ze2>OFNuom6D$x$^_qff-~Q{5
zL$lw#rYGRd8@Wu&6)W_HZ8MC?-7_TqbaF}x()kq)Yp!tXFA1F*S8fCKLwpF!DBcGh
zMExn#E`?>^EXv*=dkol;US1D?UF|m?(~=!TcFozpSG9{-)G(5bNBTi<Ik~vdWT-j+
zEQ7E(!x&ATYk3JToBUpDsP5R)6q6<TGz$-tk=Y=s+n}kr>-({{cG(A*gj3IW_Pho6
z$$4>jVJ-kueiy`h9rpG8%*C){$Bc$-aCfAb49nP`ard02TS-R~__VTLf{6c9@%bL}
zCmtMuf|3u+E{);~*j&{oO}o!XXv%@>*EJj07rRL?GCE*S?>H@4ip;^Pc2a3muDLWU
zt2nf{WZm}Rr`q3kGKMvmN@Nf%oFZF-MW)&^Sg=4sjO}s||EnL(&7%?ovE=O8=QK>9
zO}I<+^o}t07v%a6&9G#)w6^kFeFjbPO#BXuSN{B*NVN=&-pp*<CfG^yKRv!y>vI3`
zt=Kk;7UlCWIRW<B7D^JL82L^?@e&^&&#gkf_mij2R8MLbaum5sF9u3W8=O@S((pNQ
zBo9-5D*K=_XQ*%ujf}oB){B!sW@=H|d!H;#l;Y(}HB60-OStgida|?sP|fm^hRc?H
zhlHj9#qaz>avquyK#f<75<NH<D@dN$k-vZpl>iH*^<XnZuyyOIY0t~M?a6TLpBOGL
zZAg%xU2C?p*nQRDxiIk|?)Eyg<4xSW{a8Jj6ENEeyXHjHxKwL-%j+*DB~hHN$^K;^
zjY}A@+1ib2mRg=uCBNkjvvROe8Rh12aFi%c&@+gavuED~JBLJ9P<Tf<MV7w>H=>(2
zXE((5=-2dQq1Bcxh2|kGOLe@%ITp8mU&9^k1>#nA1c_oSS7W4_Arn3s^fQ<#wNwHM
zi>`6oo!QwlhJLz$-oxLpzVpWR6Fb;4dCsV=^%JsLjFRDtv-`@e4Li88>D;0EM0#BP
z-a2w=W&yV|%-&p4im(q#!b99jM?`9~_Y8zl{rglCr!o7G9)D73ut0$WP=C0d7L4+p
zhtks1@v|+4>SlKUgy`&gB+{eDiC#%8<#8Fc<nVA$jRHpY>9p69XTug#aKJ5k@C?Cy
zuyfOOA&9Mb&~{i}US0@#*E8zgy>nqWDzz9~2cYQT!-rAcc+YF5&O=BOa&E~)5}V&m
z{zasyw&(i|9B(dPcrXDJiE?pi{D&8GZ2ZN9M1D^Zyv>h>rxTG*WL4nhmLp>`T0#`3
z`3eTV*XIm!V16q0zSX()O|RLBLK;L`0traW<2kTD$j>O0Nt@Z9N)xX9#2tVLpz_FS
z*3+??;VM7_liMO$P|`D^ga8FFbrwR1R2LH6+xrm>(h8p!F+u)xMV8Q(ta<wKkxA`)
zguO4FxVKaMg_2@7*|?okSDRjd%w=9{zFn#1C(?=0dWwvM7v3zvXt_FM_odj_HRb+$
z%6q506I)@L3CI>{gW&>Nk?MGb(W8k<%S=rte;zre_AtF3{wpr>=+zS$#O2dsL~nB@
zSaMU7ljCF2K7ACjJgcK~i!0vJr7k6A>q3%?WG_#h@Nd;eb?qII<N`)~tQZ;`5<*q8
z(P)@~!DB{`Sr2P+9xEYwN^)*gPJW*jzFPTtv-b%IKe)X7m(epf+gLo%{YE{Paa^H^
zGNtr}Kil$B6o)r_*)e!`v-hUT45RAw1*=j&_@${#ovJBt$%^G}<<+Ba<lMV=`F>1+
z2msr`^}|w2zyL^&itleTRNY+hjU!&^54ahZaN5?+E<p0jyO2xz5+bdm&fRNHlnlyW
z9_4d^XHomJcHn!$>V^Z9LsQ1a*5@b0KJlAdn-CzCaZa23s#S(Og}JFpNww5Py3HcC
zHrkd1FSB%GIsz0%<bvZ?c4gcp*7BwnLs?a*+(oknG)Sq{7M%ufn4Az56SLGSl|~)U
z6*1SZpILO3A7Wt-r<oF&YX(07>u~FSAuLU|mul{jYb7}W3JdLK9&HJ@3y_N+&~$UT
zixbDUx|b0t^_+@NmXY6|>TI9A?!jiduTnjQ`+rxzQij<#aH>nLFHBx<Fv^KdwZ5wN
zd;TV|CC6a8b(K>#nf$|EvAr=_p4dER%%W4bs<wROL|q&zR(wBi8od(}Cob7`QnT6g
z{Q2`C5A>7!`bMs~QUA0@YsRmL-S6I#eBOI^s;Rk|@zuIz7#`U5iAqXhw9QgSasBK?
z4>L1YoPGSjzgJSq_L&NE{=9kf;lm0`%g5%>5i2*OaSR?+zUCwNJAy@DdqGh-$|2?s
z&T>v3*g5y9y{|{TFnUimlUa_P*8}onHibml8I*M?`_?r(k}SpiX)ri5<4XDYd+*+g
z?gM@6aZz*R$bFiIP3~th{Cye5!nx-zQyMpp+H7?58zBN09#fhyBIeSi_H*sO=TCiB
zgOaX#T}klIe$!oYDOtHx$mYntXh_?TZSy>QPS@o23!6%6-->kE8y!8`dDuwDlyp#*
zN53+vGndtL(H0Yr?cMZopkP7dxE5#0yyyNS=|3IyIcUm=rx#aFued^wT=0*Beuwm%
z97|o}*#b`UD6rNni$(D1wpiO`+K7Q8Ml6}ODD}pTRGZdI0Pfxi*C29}D?op(9GdYf
zkre~zl(e3IuzdKQJ$oD-sS&2Cs_K>@J9wNZP_8Nkd~&r=ovNW;ZU(wnT4$8E9;xg#
z!u#-}2k&ZXing!5a{02Nvhqmrz)2%;uB~ELo3+8PVP|=>F$xMnCr@s)w(j1eN6LiO
z_=pJqz(5`EJbzz-%ZE<o?qlq<(EToahGDfY)CHM>R~S_axheY$$<3Ck%(CoVuGXqA
zw`T{+jTj>#N-i_o{ijwhK}=izFyFL`F8)4uhOBS{@~SOvgc6r|+Kk7S&L3&STZ=!b
z<el(cZ@FOOs~a9oH>LW|oz>W|zcoI52We=(^mHwn)pLiX#aXU$L{yqH-dR>c^l$UI
zcE4*w9T)B`HK+-T*|6PsIG?!R^e#D%9(lV@ly|$G6T)_RZer)-Yl?3}h81H0U6v=B
zk?mPpdN2{i{2)o4_xEj2N=hO}cMhSMZS<0`ahZPrZ~8ZV`fWP4_Pbu<F#WvAV>%;D
zTU{YbbWJ?>TS#W8xf!ZW$b=pz9Rvsaims!4LzR(t(4j;3+1@l~ZM3&fKt2O>&SeNR
zHpRZgditavouoxYXrt-Xr%y{$Q~ytwFvMshH~R}Fk!nXK)l|61B#hg=R&$)=ljM6=
zYB@nC9EGaa&T^0EU5;qZCl$dDyS_%BnN_u3I@b9kl9K-<$FGj<*`;T?X_RegY(KZG
zf`T7`@28q<G;&>d@@6wSjO%M|cX9k-e986U6F1rEa~3aIvZtk9+SSU(U=rnL8zt5e
zhnwIyAf2gVpxshcIAPST3SUoEfeB5&Dwb=i>xr6szk)_Cd4C%b;nhpq1a6_)z8YX!
zkWM^8LiUO;kFUtl?|HBX-`M@9#{8<Ux9jOTuAis~k`HF$IMVk7yg~c_&71A0i%ykc
zg%o@43#n+iZ45YX8~xFg6N(kHZg5QjbL8i^bK~u3|Lf6nW$sP>Zt|r$*ubdR+u7TD
z551qeGbtvlPQYq5G-^`$G9l_{g|=)zk;ycm)_=f{+AaCw&-@GL&)==={CQ=are_n-
z?NMAJavdjL`)o(Hb@5dbx7WLjr|?u{Ns%uAWh!@#DplfIh#bnz?&m!AE2x5hLh2;o
zbsWu%wyg|HVUB#?=X>6*S~z#*%G0~fw#aE8@|!JxDqEq+ZWX0p?s7dDr@^<$dYQFa
zkzG3YNwR^PELp?kELErbDC{)Uwra_b8@H%JcP;%{c^0WKf#^|8Q|AK3QnIROtR;lW
zG=v)&oSL+^{Pg6hQ!hN8FtGL82T3=LDN|0g?wve+`o>QiQEi-3JLsn#ytfCm8o9_T
z%$Tv{B5NZTVV5$CH=|EmQ&W@8Y>u<OnSH>14VzevJx2aF7YUmSvTbDH8v!tNK6xxR
zH<v8Xf6p|VqSr*$%3a0_54KJ`6Bd>f``TrI%ZHKAozzRuo9+1=cjin?gu3_6#hT8w
z$IcX!1i&@sR5jGp=rk@K%Px8R`1tR6)s33lK2?+z7yF;@lBT_27Y?E&kdQj*^N$ZP
zZJ4?!nQUzGD}qw;`&%EEFK7VX%Ac#IAx^y*-Qoi(ZrIRe^Hfk;>j8a-Og)@gltOE3
zpOBl;0oSvV0piT(3HE{c3i({6u_+&f2PreO@%{S;i`J8zUs+ZB=2~+R&9y{()<hj|
zWU!XR#E;b5a8{%Dv=jl>d2M$)-C{AZQD>rrbJ9d5CAmvWgbXeBiJRiYiS2k2gd_~c
zu;fVQ2SHfQhndiPev-%Ovpo!l*Syub4e3aQSYX6xxqa6kpY<i#SYI$EQqC)`*TK2T
z)0h2d0p8s9mKzvUGjL-+N1w0zN7nS3@LqL-N?+#-gMw00rl=hc3___w&a9Je>uf57
z#|blmqs;A3&>uhNW$>gp`Jv@Gu4=pU7T5?0cFi3mfI4@(D%`VBG?$eSNRgJ2G1=KO
zLrkIqwpv+~dkl)(()q|QJxK$JaMLe8er&XqFql8z%VL1vx8JSKR;$CLZJ#Dp0ct>=
z!qcdcx)t}og9Mo3o8!qB)!6=Vki80i1nWy@=XgxIH{Nr+!{9Y4t2^on|It&XO?$~s
zBiUt}>FYn_^PIXFoDd&x^Hx#!MWoTh<&Y@*+iFIRukDpk8fZ0a&YwYt7q&Q^zoB3G
z>U;diLfhC0_9Yj3s?FR^Mo$N4^{ipeZyPmZ6B16W*Q#mlj=D$I)$6sx#jGq1S4ALm
zmDO6ip2pg$#;J=#OP4ebP?omZHvC&l3nZA0@10PUx3?<X2B|s8+rL|T_t4~<Z>Vn+
z^=>90qXIR_chXZAPaMD2LkiD+>j_6x_ub(%w|si>gZh*-1GUmnvTR0hBE0nRSxn)N
zY8kqtm1;G4_*lL7d)GJgE_cpG0_%1Nn$+rbo#qS;jn%ks(ne16yR!1p%x`0qKzqKc
zP_>V<$ng99<kqd%Tp86#OCQ$0`1mjSh|NnnHo9-i%EZ*v!1K}AKwR|rLl?AvdQ`<~
z<23Mw`957UcmLVRcq4jYX=c5^sRpTD;G&@4H1JGY8PpF&h2YOY)67c@(1548;Z(((
z`e`>Nsr_i#+^@+}^7!nLkK$I^{2?%hwnu$kUEi)?*T3b#HRBkAD-&KvUA#Dc<1<pS
zCrRJ<utn+vD$OqC9?2RswsC^qga%uC`;_62)JGlr6_)JqCKF!-)ybz)BJAz#<X^2c
z7|tU?jonL4{(vn*T=LUWyLFjkXIIZT$eo?@xU*Dx?=u-rDn56*DtDd#h3bS5#_^a7
z1yNAoAne}}Cq_akl};`Yn99G;;7a8i_DyVB6Mga8P5GOBOBGx;3_Yba^XQ_qQl+Ld
zhw3u`dr#avp1$dM31$)4IH1X8Kx(jy;(#8yF2iwqMh&0}^NA(``C;9M55+gi^5!gY
z+CDrrEiEtSr_G<4^=fx;^(rsu+LFI=goZdqzTw|@ck6PSjDLk|iPZe~g8lI84F#tI
zwh2?D^MAQpHE5D#A4+>qrc$tIieAgUT#zekxMYyRn;9<aDF6hjFLZRY%MLRJyi}Q(
zjY`>_89PS^toL%#FJH#=p$*Xia>{KkE&>Qe`*pd&@YpBk9zA@>5CJ`pk>J(=HOFLw
zwxl$7dWiH{@sx?i6sIv^@@hQ{2PXS(>^v$=sAR|^O1ByFlsUANNv~+`=6Qmglw~eD
zeYj@gj0XwXAg!nhcm~^$)Cs!J+}ziL!V5emuLyHJUNIgRH16-E(DsoV{U@z|IIT!~
zZsNq`xx4RBm>&!7WR8DrhP&8O2=f%bx621OKkl_x>sIEJ(Z`{&8_IIb_bgw~|N64T
zB})~e7lqt0U;J3+LtWivpTCkny5(fPPWNoObOFpkc0xu%!dT~5zLefmV)AR=y)#;U
zpU~%*j-m2ZkgRTL-`XcFx2ny(zgso-mAJs;zu9$6{b5}B86SQ3oU=W^IKUPBHI_?=
ze5F(bH2y<{NCttg{wcuR1J2!QK5KH0(YdBj>mi<>3Gu<Dw|)DxhIQ{piha~9dZ}Xf
zJ|eW*{*xvq4n@nN2>XHVn;%nXlz-c+qAp%N;L+QN<}c(d>d8LGLqjW&$)5rZ|4{mO
zc+0{g`ti;)TiO>0M6JbdK%pl|Kj;~F3`4Zyg!%(T>W$3L!fedZi_gS{2md|y-^DAn
z4Cw8*?7Q;chopZ2*ihky30hMz<#9p5ET<aKyCsNY$a(^+0?4roPEFQsT6nKgQu&?R
z(;4%MAIU70GaK_u>4M~z_3NiFdBQFaG&9J5r{9T;R#h>JeDD3n?<Mtes1;RJ=e51l
z8hTAOz~BG&kQ15f`sO`*c1k^M^_$MpNj#Or_O?XlYJ3cH)hqU;z3Z~~HDv`n@;aX*
zYd&q<^6c{7MA->@lt)s=E9jpsSvlX#+&ti>9|S$d7Z;9eomDn#*w<6}tBdP!07IdB
zj}pc|>s|l=JCh!`wFTHtX;<hTFgqYC5KHCOBSX*j=y!12@I$=^&t0h@Sn3D@*c!80
zvwtgp)RYi0a<%T4+-N=a=c|IQ4M$eVzPt7kMoFj*D$Bnoju}{f=i|psO<4!OKfkf{
zO_v5w{qhh+a<mNx={9jnZT)WubvY@&)HZhJ;#&PL0Za2#r&7^iX$AYmyubIrVwlbW
z*UgN3+8bE38u!i8pkhdoD^yI$GfBZUu=fDUn>X)}p<-fE(&$Ud-Yd-x%;S)#``6>K
zVEjpXyn|ccgD`HLogEHlRPIfs7eTA74r4syJEkR3*&80K7adbQ9I3eB=c4Skclj3G
zmdHw6-~A^POuFFgfH}e<aPj1~T<cc-bm02&Cy_AQPTq53lB>lZWw#wW{L(*HUK{o4
zO{w%4AZMi|3j?>+{;Agx=laDcWe2wpattab6@WTQyc4bC_Q~>}C0x0AUgnS1kl5Dz
z?M)_3{*#T*3pg?<9z|K4c8Oh8sGd%|>+$cwMjGNcxdg>G%T6$glt^5ybhKWxXuB<{
z;q`Ydl%e3t-)PopD*DQg7fOdIc@zBJT3iy&aARa&`!Rdd45xNqcKdPlx5wLuYw7v<
z?$5+m`FwezNK4J_%Fl=LH<#`FWK!9C;%;EQpKWcI%+rM6gvOkkbJWvK^ukyT3%jVu
zwJ@LxQ=qjNjR5H37Eu<VjRoaGSnYbJLH!Hr`niV2jC`dE7-09;KeOvV@vdE??pQ2X
zWt63a8%Z6k>4o5$+S=qskNfx3C2Oc&tLsx%X%3uzR!7ckKQ>6l@vjy_ni;R^6~)=Q
zqdxReNlC!3nA4|+rubxJcUy7tM9in_0#R`5mqhiAyDJ_AsqgbW(mFBm*`Ja1<7nn-
z`hI(Vk0|wNYHA@C8mkGcv{kH_|8C<dD)oKsXZ4mp&=99+WCCzL%Yl-(m7pNV&qCs`
z`G40ZZGxgA)@Oco+XdiLQj*KN;q&LuU|)#F5)u{;`xqTtpslOh?b>3Qi#L~V)DTb1
z8lV{8T!gcn(|I$0o{@e=K(fhMuj=Y`wSS*h?^6`BN2{}yY^QqkrlyIYacO6^m!GJ&
z7$Gg4GcGWuSyoT$pnYY=Wc!C<X(w)(<5zR%if!T@(8I|ar94KFpZ2eL8{Rllw3?4v
z>5?l()yi&)s9LycL7a<9bZ_VFf%l4vwsM0DJoK6q082`?5rz?lgK%odoQZV<H_28L
z9~>{_5o(Wf9gvooc<f_y@#1D|lQ)cM1Nsiol-a_N;w57dW}BFNL16x@qQZEg{Ppsu
z&_&j7P-K{S>B96}wzf0dFMKCXvVN0DFEKG9iYu3P&U9}^Ff9p~jvd6=#CNGH<fw_Q
zC#3(VD;`xlF$(0dys`njHVA%=uySYIhEJ~sN=Z%HQ$>=}TzTH=ufzMleSh2Gj~K9i
z36GwNNU`+_I)+g_cET_1bLs8<u<$xrgh;!F$o@7ZG%0uj<y9=|#?$*s0RV`hTPw-j
z?wAV?;28Q{6Rp5_W=8ixm|ib<wzsFa`1_<U;9Dv)^+`arXEm6-zkToC0ky63_MY`Z
zrCxNkpI=P#R?6G*hphfOpRQ>%)>v-WGTQm0@$tC8t#=cUVo@M|xjbbTX8=sn{_zY+
zkt^6ATf0G|+YIb#Hr>za&wQU_$2K^)QTEQ+RVAbl@}D&?4>)={I6*$K{pPC$?>h&I
zw9bfqg0YcWqrs9TW74Mp1Z-S2TrO))<=MXn4jeGgnqQ-I;#{5*7EFTSu5LQ<oO0sA
zvH1~;L)TbOi9=>0|3<nBkpprHXOkIlQ`aJUIB(LWU07#28Z7%yXPb|-o;Ii~TRa#x
z8a{8)p{|tXj}|?l7b@ghk~4-%1on_Aqp__Fs)#~YFxfd-8^r<7QZ0H50}~T1u9e_m
zg7Pi#Z85Td9197ZFttatC~d}A3VvqB4Kbe)vlsh}iE`S$o#IL<?OtW2J9tB?9i1hR
zS{I}-9Xa$oh$uPb;+B{%o+H5&*4#Lz@;U6y=g);QuLqEs(=A{wSk#0pT}lwmdoMPQ
zsxQCQQS;cy#9lb~lMBuJi@JYX&oQ~PcYN`*c%>yPad>Ff#il8j&cw($J3BY+3TjhY
z5b&j%IuUm!r<0573x*!mJ)4wM<@S2_s93~Hchp6_x7FnKYnt}S5{b&llvd;jn)2)9
zV@eD5W1T~pY&D~Z_N^5=z+wITB+TDHBM6N=iNDFcAAQLPX#V<jZDZ?Y$QSq;XBlav
zKnTl#>ah?DOY;@FH)>aWW?LkyLR>hpbva-38A-X;b327iUf<n*<PkS>d^BY@XJl*1
zV!JZK{pX~#11%pXN{B)cX*WnJYldsd#>U65yK$%~zwIx@!Ks7PxY5UFjTT)8yPF@H
zY83hKzRcP1sHlVPGP%0(jyJ=V*{;TKG!`u0Gf7oddbyWonrmIrUX2Pelid42LPtjh
zYRRI(;+tK|Ff3h_OBEX;jg8~<<hd9QXUss9yFt~l@GBid15JyHizAZaqcuy_j!|t#
zRxNb6gd%8M;|z}cfHK5q7pioJS)`z>bL@w-23Bfbc`EpHx!j9U5~A1(0w}J~CVu(9
z{Yz$zY|2&|!G~Ufvn}YLkx{=NDQ_Rc<b||*fDzZuI_hE=Aw+m<il2+wgN@n)=2aeT
zfxcs^vXt0C>QYjzb&ZZn$=v(ai)w?n(qX;MV;}2ab>AX;r)pXOIqHfL6=h{D7{KwG
zR8~MVWTZgIf+-iNK-{0%<r_VkYYnV#fg(8H943>S7FAYS3h)g+f!H*KPJU<*U~xWL
z%Bq{vIGOdDvPMToWb?Cqxs=avCFJ_JeQJOemn-Cxm^{+f!iv7CDum)c#cKCpR4;kG
z3E+It8@v#Q6XlIyhnRSL!=JwI@BjLGn*Ja82P#iQ(6!8sF#I)UyoTf00)#koo{Qs+
z^RcdO!K`&;VLH3)oY@}Xf0N$MX!%EDNX(I-F}=U(<*QedV1kaHICpM|!Mz?LBA1tq
zN?9*xFg`7pYH~+oxgg71T_?09A?T=#iwaa;g=MKvC0S#RE&$w2m%2&}S7%hPR^7#i
zkDFjJcWxKARW>y0q})M&kvn7dpKp>smcD(0>Px$=2WHtxEwQjMVuOSx!o*f|tLEts
z9~TzJ3^TQ~G)VHE8+HDC3Z_ALIFq~Z7+=3$Q&Bd$U!Ok5FI<o}UsOJLJUhqQLnCfY
z?0>ZY`v8R3E!+M3=j5~kfdFxV>|@5<PN?JJ+6zeNDRH2_RwZxfxp{xD+LY0v`&<TA
z0NkdRyB?5Mb6YJdL2=jzM{i-pS7j?JDuVJXp*@l0Jfr1lX;~T4$lJ%RQGi=p{YMoU
z6LS((B+PtBE_xF5Xq12N<zySi{pAcl{66HU;c_EcW+|@D-icrJ<;28s%F6X#BX+MT
zrwtG_b<p=WX0vw-ouV=evqKedG*oJl45q3daq9TSoL|me>t$&v^V%{^0MJpA-~VNr
z<xoXkf-K9;O)aGW)+0|HL8Wy$UoXP)&A=Zre96V-Zx-~^uO=;FR1$?oUzgkKL%)67
z<}_0ZxtO)V!J4<=V!2<(LCz-5UPu!F`r6ouMi*OW?A0o|xu>0C$owpip48`r3rYT?
ziq}XD0$kXcK;tvQEEg{>pm9=&W!JqkZxOuCsqI^Exb<y_YKX3a+bkd1!xZ<E$1c6{
z*B!yysFo0YVNg0JIRDhq#}E36$Q8Qx2xDBG-IM4(4fYEpL^i*-=n0{-z+x1b?VHw&
zqesPR^Bwds&vSGiadgy@Sq_LvQ@g^ojrlVr*judGQDhD_hWmb<V~*8*e~m1v0@cBc
zCm2)ioOBqKk(g#5qaGrNm;6?|;ryjuQgv+2Aju3UG_tT;GM7XYzf`An>qgEWR{U&j
z@hBFrq|wx5)`VHJ!j5qRjwlzuOYUW({od~JgPtP$&=sp_d;I?CukLf&B`YR5S;g)w
zL2t?{k9dR_#6^|WBK0d)RvuZRoY+tub>YI+5yFEAzi=^U?c+r>PM8MF+`QrsV|?;G
z#R?x-SzA9$iB&9YoncG;dPLP4abn25lMlJHT6@MGbFYM_wO#$Xe<&K0-r{4H%<>7z
zwcfrnNUSyE62=8Bchf_Bxpewl-lAZ8Lsmk5L*mTMwWTs+PIka~y08Zfxc|$HTeEcX
zd<Kt!PZluI60OL>g+O?xY*wB_d+8^x+^A-J^4cGK>|wE18XC&p6<@GVHh1t~2xmYT
z7*!J#5&U+1LIT<cggQ3_sv%M{(fid{`4RJff7a^m?A(mRj6M*8Bo+Y=9z5Vv7;UpH
z#Xb;dKGLqTHG@;`9X$vW6+wu*t#dQa7^PuU*j-dXd=s;w7Q78tZF2P9y|+mtNZy+W
ztfYB}-}x8HU0~d6s6-=s)g}&`b@2V$Qzz}1b$Q?8ut_T#v{>)gla92tf^c}$ruvuz
z1w!fURSmz>>`3LYc=6(rtA1}+JoR@!6BH!<U{D_UboqgSULb79=f4bK*YP6E?~5{!
zVEV%0HbtisV@Av$Hf(e<MgoH?WW~#Bp%q1Y_AGw-bj*n53m4jxAd{LhH|V0FqA*T@
zS__{JgOzp<pFa<2sqUj^x^Y#vutN~@;bi*ds9Lt^=;$DTF;sN_M*!)5w&WG^!~-(i
zD@jUnX9$;+k56rZq@(QYFGOea%xxb~LcF}WXEl@HOik%dmK-pE36I;HBtifuNQrIw
zNd@_iQq&d&YXhwGl~Eg6(Kvec>~?6IJ-@$t+5EZ&QtIaDuYtMS$!}l3UgM(0BZmjX
zfCy4`gGGyS?Vts!hy1QD8(Uwohn9U9%U<H*g?V{#1TEMI0|UW5A0j;Oo3Pn!=17J4
zjuPGH7PO9&ROz)K$XD<Ues)<dW!4eZDY_E(h*mRv`p1yjC5BkBR-*fG%MToHFe>&%
z>kOv!`N-uFW!EiMc}Gvqt;ze$u5=E9qZO?tZAv`RX@)R@V7tF|t-Hm}o$6D(EgLKb
z=X7v@?iPQ(Y_5HaL{qq9%?s}d?W8+07*^D1axgf!9N;GJN`IL`f0vUZ`~R3ULAg=d
z*oFz}h3dcN8MS6qR$7$*xfu)i$8ORK%g$(a1Y=%jc{P}^RfnH9L~>zosJgmDW_Ocg
z{@7M{5}Exvq92+a5o1*E2$h+Aj$^X0YE9BIIxzC{M$316x1YaTQ*dCd880O&LKOj5
zw!!0I!`xW~%;}Vq+pDpOe({<YR;hGCjdG1X=J_y}AlQ$SQu)dJ3}9<Yn5ev*^DH0r
z%=tQWCKIrdRkf{^3x%KR*H5l&`Fs)`1-A$jr&`mGwufD=EKp{KCgZnNZjYHT;$%=z
z&Xq|c8(K|d%EhnG04G>8&v4DGM#{U=K!vf0J_!Cz&IM<EHBb{`k$SBw_tP{qEax9{
zJfk-I+&g~GUB+dW*8<+=nz+C6i1`xtgi<FgI>TLNjs&Lo{`EUwBI+aptk6~(#s6m~
z>#$1{wnS*j+A#N1=nET>KB49<IRN^qTf@}wu&?a7PifL5!^#;RRw5?G!o5H@?xd3m
zk`L&fuAF~mnacaOvD#O4>Df|VIdsDz_sXX|+SIhpey?(Uh5hc*%HaC@_wN%YtQKoT
zqL>p#E4EZC2VoY)sgSo2`^L9<9IATaLiAi}5`i)0Hr21X{{nq3^>}q~x`e0*kqil*
zpYjP6Fmt5sha2LMg49A~q*~3gVS7`FFtXDQacSd&<kVCoj33~1$0OjTqhRi^3~pai
zFR82`B$ib~P5HZf>4ZD#<3m%)(Xg=Lt_Dk&KB2LQM>6R_4<+uoGE|1dg7JBa&B!eU
zHvldZAwSOBqnxYx-b7XcK_BxBR`g$`bu~HJ96v}rTuJb+*&|yQhO4lxt8eVnp;{q_
z8XzrgYTs3AxSw7{uhK=JxyQY|PeWY&(7=lJ=YtV@baZv|G`AaD=a*&SK=g3tts47x
z_4U6v-CSJdnIVTzN+RD)T77duUau(1oZm4KAl5j}&#P>Gnv;V8ibcR7`u|3#IUP{%
zYxJCX;O@VFeia&GT8vnpA#`_5K3h~(^@p$$5YtJd%Waffd1DVR0>WkLb>D{I8}i`q
zz-t{M3`s9aAY#7SfbOQ@IA#_lIaM)Xue{1DN`-Gu9x}M^8UCI`4#~Cs*({;72Hic2
zlnUS^dfG$Uy;RlJe}VBNy40l6LV--4ECdiT_R-RE{n`R;gvxX;PNO^AJZsfR&Lf7n
zGh7>GTx7@Ku5F(4;6Y)is(0HMEfiV&A4eo$=6jSBHD!-$4x40*77?gV*zgI!=-YTP
zM8Gwz=M~p9XbN(t_tS1YoqyzJb7!eYmYoiBY9e*q{Anp9DgY{xn6ekXH5bly{*kYP
zBJtIAw-nMG!C@IYO%5?vFnwsW;|P)u_xZDFH^4>(%{@PIFDC~j#^C|&!QTfzfAJzF
zI{F519jA!Y5VTn^nQ(I>ZzVC|WspEOppV7dvS!5$Z6_pq!cab{6voVO=LVfUi@S_1
z^XqDAqy`NVCe-w78*IX1`_|kn2(#G*+#%=Cb^PlGpg=%%uOG0sxU>|@&fC<O;2f@8
zMlij6G2AndsL@)&a}*qKI1bsg`OQ`D+&`2P1a%dsfWn$rU;0`*aaFduyu8k#vp+dY
zj8VA8ECZAvh+$vz3DHQjwzk&S*N+}Onr}iXLN^KZq5EHz3-q#_o@!gdmBLgO0&5#D
z40ZMH{riCf25=BULqZUdXEGJn&CRiW@qo*pwm?wLM2i&iqMU4w6n6_C0o#Pbj6{cN
zhd`Y`8wdhbCrz@!KY{4ZaV~*5#(;Uj?w-rFwbzqNvQ8XfJ4zzx0YWzSAH^^|Ll~4{
zvR+4jT&v6y-ENpm2%TusV5|WePsYe`&^dmBEfH(Q?Ph?}0s2EFC0n{Bu3dBE34&gl
z26f?fyqb1KOZL>vUDMjf$BpXUrw@GOZuAh`4tQ(4<Dx>Eg-{i3JI{rH@%_seJTbHg
z78iaco3@55|NEo}jh9#N!9pi38KKaI6%>TcZL0m#Midd>TUvsCmkrX~v32g;J7`s5
ztj3s!?Q?liXa0va8G00@U51m==(hi4Gohbl;}Q&^fp{NJ+{k-x+&q|@7KjJGoj)Xb
zV0wR=-gzTDOG|UXjSdXq33#3D$uUn#O9lINRD3@^K6fU`Dz<mnyqPoogMje#*H}vd
zsG)=f^ye95?fQ{aH)WLiqf@6(rz4}J-Qd-$b1u`nDqKlTl|gpS;^hWV9`nWWuF5Np
zojEg`J^FBa|BvK;Ff)(|Ha2ynH|XYqPM;<*rA>yGmJ&uE7Q1m^R@S!E)&MEdX&KL{
z5s+5`5+59h;bCFnK(T||97t9vkv{VY`FhM;&d{E$A<mH$+GsDXWluseBk5kv)7i0O
zhnblfF7LcEbrCXNR9~F(W5<tUKT^e)b9aYUo#6I`rX2@hcJ?_Bh<4_qbyLkPEH<%;
zY;6T`BZ@3!!RB6Z%9^qb`QmQVGGeH69`i<2kd7ZpqJ)*?%UD?<Kc(XZQykZZ^2hW}
z0(ZuZlD<b}*r!*o^T9m3r|R(o!?!CBKi4C`=##r?nBJ@w>H5BT?+eTAQ8}k3{DGNI
zE0`>DaJ^6Bn`g3v%usfetP-wi$49mv27E^CUDYyUjg?F-PhI+@+B$Es_G-zvT7Df#
zY3hb$;CH4~v0Ho;=gjjMNXg6&imBgTHU-Ex>TC+<<b3_SuWR!H6kH<SA8CzWJwk8+
z@(a9rb=u~y{wyCcBR+oSt7!g}1byj3LGq8P$TTe`Y+cgrbdR>}AwTLb&S%Ku2=hEr
zWYd!Voqo<^`}$PcE|(uY`p~rU?t$Y~RFW@W_A?ha4W!MyV2jknX5zDKSxImUx^cD5
zD~yRnTv&1OGry<`W^Gzx0#EzN^9nrL^U)*rZbGX}hP>!6g|Hhlr1Qr3?aKvZh~3j&
zL_`?A5O3Fe3roua$0`WRzx}Csr%tv4!#VAN8`Mv{U4HtJ?d2%OLN;mS?|wuN4+5hF
z<9Snqi*0v2y3u5ZdR1?H$n-6%E;e-PvY^B1NeL&LC*Sn}a?oL&jIU_!6X_C1J^YGN
z)KoP(j6f9SP-_ncU#a>rAr&u28fKfo6@>keaCv5@FUPk!IfbH2OOu~E^&2}m-mY-^
z;c~ZUS+19Adac;EfQ>2WOk=#fdi+S*x!Q|e3+cUdAp(%Sw;jVfTrU@lk(CW6aFzxD
zJ0nXsO?$sm`;#$Q&LKeD(s~Mcwo45f$7bK&8du6eoUW_w{HB<;fq`l66B_z<dX-E(
zTKZH(R&*qT<qv*pjRu;AI6Ls$nITnxLCTw!%+&ql$<@R>fz_s2?q8Hpo+D~7PbPZg
zxt4T#edrH{x)<(S^dJ7y$jofPgw819cC!|G-}OX>*qy9+`qUK%eC18Ow{XKSEm1I5
z`8jX!(4p^ZYGN~`UkNI7|5xplAHs~%&AaNA?B1`%7Yxu-_$3-@QEu!oP49BUT=4eL
z$}UEZdBLP69mP-W09!m6R!0%!B}5|XmzD0yKl{A1*5ZG5avgRJw?-WE-LF1|#)2|~
zPH{#*vT(+DJ$SKpi;uT=uS4THce<6{`)!At8bAJb#I7I1o|d-!+@m?JpH)xAr0N0b
z#w#!Gl3kWF=d7ag(W~aJD_8a?P_K`TyjY(1;GpN#0(0F`^Bz7&b<fODRGtucTxG+b
zzqy}JFPU|1x}0B4mtQ+){q9<rt@cBHZ%xhFDV?;`YyN8py)E~@@A|>-zs}JmaviQ-
ze3A>=!v&`${w(m)?C|2k&!ex~e+&m@hc*4jpzwd>3r6Qub*i(&HXui^Eis(I>5qJe
z3SzmdD9+01b;Gak%6+S!jihjZt0aymyU```XMpQip642~f1YOi>;L_S{PB{xI_(MT
zkmThE#z2O@Ri;dVF%4SC?eO<Q%((%}cj4ibXT;2#YKJbV>jKZ6TV1)C_wc+*y0dxH
zroe^Zo;R^8U1ujOr`Yk<c>cj3&;Q*B9nax^_kPD?`2WChf1rpaC*y^%ciV>m$xtR-
zY_y?05)3<NA)dDk`s>z5T4+<Djro_|aDH;z2J=b`X*%6%x1sxV>(<NP-y=EaPn?E1
zAU<-YhVz#E|03}Y7sSAE+J}1Mc~QK4GdVg5?5}v?w=+ix@lkOF-f(oT#b2rb?2o@B
z9Hj7p9f4KB=f>Pr<hfUz1Vd)Mjv#7SO-uPrY(p}izWwkaZkK}aHWP&a>e>GRZqelP
z>dV$`*q|dsT*EH@LNE^0$y~C6lz%gI{I`-p{6<|rA^2+Pl`_?i^v%+4%@}^%GvJ62
zox|moS6)r@@|y3z#lg-_ed}wjnOBAI{cCQ=YtH?ka_ce4aB;Db?{UF`=|)Gy4>Zi&
zV7EwcidMYRaT(|<QfQzPk$!VXy}*UP_nh<Z(^%*Zt&MJ5CHT%wlIi&50SR<XFiTkZ
z<Hb{nLO=@m&rkBKcyvmo21xIrL2bjo^GjN;|DKNd+8$Xn-QA+0OsnIAH2!;E;gPA@
zmguxKhjhHyoqsR3tz!Nr2C&sLJbBJg!69zgfByd8npB)2FQAXcuCC)f{<8v>sETH_
z)<|wH*pRF&#Kid>&p^yZGL;F)!eGw+Qw2-uOxcdL>;`hQiKg{dnNKlk_%((P`u84P
zm@GjlQ|8|HgkadO->u_om&|dl1V2fkb9_O$uq;x8JC@}GQ`0(;|GLZ{EF-L<`M(F;
z7u7a}tJF+Nh8lPlM}l-G!>Jt)BkEA*oUJS|Xh?MGX5K=p-@kVrLUnWX%9Vb>Q&!;h
z7A`fpY%2^7$-F^;s^BfyEuQiiQxc~mSFU^|pS$_1KB3x9utqhRYE!fZ4{1B`_X)1J
zxb%F{lfUK9^<G>PNpjBXJ8TpAi~r#l`2XP~X!7_*VaL_o%gwE$xs3YA84$C9nlpQ+
z3zm~G88ECDqNDLT`AsN7u0vY}+rYtb`T$|Igq<3-jLToZnh=F3o5qX!;&Q<sqQ@Uz
zGx%gBfdM!PCG0|X_v@+SJ?5i>VzOAtW8u$5jw=$Heqm3*B#;G+AFBz)n3a`P7Xf!n
zX3QUqSs*`A$6yNr8;DT!CO9zl`C`zGUo+XFimO`d>4r<pSji@ry+e5lc*G$bvQwp=
zExu#t&H+DWDjL+6U(|pUrBjapEB*5e197$6Lg5IE4{V%-!xyqRx)Etui42it-(P8X
znVog0pto?66s_u?UfKgB8ToF86J0k-jT%OFffP)Jz89yuow<W1pHb^;GqSLTGn^XQ
zJmb4F4=75H-@r)&<mG&|gQ)%i@mW9D00{6cKprMEUgc+OyoE>^I0VRJ7BgTNLUuGu
zPH_d*rZXefzySxiYV%{M9U)U+m`Tj%w-tR0@`l2N2{j)+ezf3XVX^q(X<J^pBPoZ|
z0?G+YPV+OPnnU>ve&wAd$Lp9EO*$DQ*d%M^T_H^N`4fzCfwmUg3^yqG?CfmFD*wV?
zjt3*I@3wGo5KNYL|9sN}WjPP1c+A|14G#pIboah1R~-4Q%#-}o*mwheIyP3Z&&{rY
zTOeUm^Y8LjhBH3C=O6mnLsadqd$cXHjIrWz&{kVfx_;d{0e<c&IoifK`S|h23<V~9
zIkReXoU_ubI*UyO^apw$?GIg6;L%S$h5b263lP{R_zn(2OlLYv1qmbp%!d)s`nT&&
zKUw48@E6*e=h;_ghN0WmY9<Kjoa^<c)=?SB1Cu$y)z<f!X*R8psHM=kV|81m_V&@(
z^*xcco81$abGgCbMMX#3)hXr8WXB+GXQHk9UYA2&UcI*bt;v&YE$%mJ9;jtpmOeN)
z$jAYgUq62y4toU%`S^GoIE{lt(w9Sg-zQCS%E6k83tqkyMwZtPPoS?kyL}c9MruIE
zdshLfFw|^xg(c3|=y{@dPkajPj3};;hkOKTXfQz?=d}8!8C3mXd;@nJVegBLx(0#}
zbK!#MlD;_lSDr#tc6e135ViR%{k(vj$^X>?<eYktw?PGS8Qh^<e9KVdrf%Wbddk@p
zynR;4;R)QvHcm8-I$9UO&YDSwx9`)YmX@mYE$5gL5aA&F%FXt_S>V10v(M13PVGO7
z+PO<%qS=*d{=;P6?rd5Jsg!pl1B(L69jf5)9yo|#;~LfI-tkI3VPs<#g+w;iL7PNv
z!j@KRt#GS&1|$&x>iaUAsP5}rFvpdELd6dDYuA3|nIA1sE(-_IiYz`A%iY2atrfME
z2o-t!1IqHuo$$dgLBQ5tvoTdD$4)(B!px8GN0XB>gu705C)aHF1>|d2Y-kw&OWE_y
zC5V?3#vY#>f|3TyA0+4B%H-wHHU<U2STf~Jv=Bt@K&C!3<|mAE$1#DV%;)h5VV0rJ
zql<}&XoVFW*Ry=79T&L1ItDq#O^BI5bn6?Gi4AlUO3d8$?qLv!|MGZ~yvoY!s5yxc
zg0FF12p&B{x<}hGiB6AGMW}?lDxB4}s7k&POueUTfb{WxINBIrG*?}F4PdHnTLB<B
z!l1e7S!PTGoL3creZzw1+^n8=iZ5T4X~W0cGvGB2h>;#a&;=frckbUGJ$CHEZ^|HA
zlJnLui_h)L>%DcYXyzhMN>mljSv)AVEOCV}&@6IHub5Cqa?%7kZ%1wVtKDNa%>#4Q
zN!UzSj~ITXnx7Uly7j|L5Hd^GH}cKv26nt<$z%A#R14p?*T9sGqZbWc>qe(PS{@i|
zlZSBQNfW*@m)j-c>i)w=j)Y9L4bTUa6fZNl4@RD;kAc}krr#-RF5UJpzHUPq(4){F
zXRewq56Eoj8Vkp2e3e2r+)&8IwDjmdU0Lh|vCkL8{Vr7#Rf>|5-nSwg9e~Tnj+;Jt
z@`uv66W>n~7d))3qc01WR^)r{LmrVUCr<@WqaWg-Lj89)MN17e#X;NRgj0C816xqA
zBy+L3Mq|s32}TLa#uqnnAK7J;q0AIBJsk^WaWpt#=x+t}CXAszLZ;nu3M$4=qlL~O
zW3@3;rYybrqEw$5!K<nw*p>fQXYRk3uHv_K1U`HfVLIF5&h6WCQZ1@(h=uybCm1Md
zXtW^*j`X5CL+4RhA2G?QvbCsy@856#^!do*Wx^<OOp|&!*yo*hI4>TuDB9sJDZgic
zXUDa%EppV&4Jskn7?PB?IQ4WHd&73PbyQ>Q_l5`PE|3Es9saPOpgg(%HPA`1G2b~y
zrjw=9Atlc3{uTG-P}H{M_(h|iOqz%o!Te#c5SGL`!qNwNT4Nji>O3!vG@P!mtodrE
zdLQBQvahPRn>ELe^<Rj>ZtH8$%NAcH+;;91n|_qkASrW2x&l$c2;^BuoESP!S68o6
zD0M#lJMa@UHGc!_MS4Ljm;onNP3$*f#6<I?Yfdin%gU^6B1ct0P_It1PzwqDnEb7;
zPXh=soRyCFS!ViPwjIrv5B#?wPM$o8##Mf2ZfWT@n0F4K`l^wmN6$;W4v;>?0sXYs
zTwW2Kd(74_CfB{&c+O)PL_5zQPG%=(1D+(x?(8hJ88gV8=T#LWtx^&adP6fI6APF6
z34*~rt0PW7qa%r~`8+4y1S07S6G8P`*>^gHBPiz>(po3SVcxiL<61W^O~}Pm+hYQA
z)7IA3>k-%e1oat>|Fl#p?pfdRZ{L2yf7N#%Gi&F!1>0Uu5U%4nLOr0rmmf}AnZCHH
z7(Nh@gfPryh~1>=(}f04WU=T0xg1S=a9J?V7`$)MXtZ(oM+}^hwG0}CV0vU#9T!8z
zr4cc%YzfKkC*hdzAng}sQ9n9i$k3rUNm*oVzeo*HUl5|^9GzK5Di?m>UoLl-Fo5~W
zrA8Z4HdLjOg9c$2VR~uROlmuBU^qLJ8T60ZakCMds5b2?Qse3#FEut6I*tzyuV=oh
znb|4sf&KgUhlH#^+JVW;z~RGJ#Vj6H+{_`cJg_T9u+bL$3{1q(TadOy8)Mvl-oa!|
z-dgfCd;678mfWF&^(i7T1crCoBDdBR$&F@Ifv2(T>TDKR+O>gQ^)~@b8vB+ZJ13fl
zBLBPGM1;pgpuwyeh245ga4LQx5P#E(wS|%EM(8NriId5*9{UE6Wll4oKk90*_%0KZ
zz{TSFyZ;g}bnS%}6d%>GS#HhInRz+9N&k*aaCa{O5Y>tke1Q3I*rg7Q6axQ%K`yqz
zXU=FsBf=DVsK(iVt53b>?u62Yy?gS4T7SVSh9}zRTNW2?`09?`HTBxoH*;AcJTZG4
zVFnXkSERO?Z{K^!)LgATV*?4UF3wv7uB{0}BHM~ldkLO+z<;;QoDWu8HLqj8gpeRG
zJZ;?S8=6z5q=-%z+;4q+mbq?&rhtqd_wi6~pm46d76e+}MZ>jeb1Z!H=*iZ(aWC@b
zoCO{99W*%-ArW9<PL+_=Z|*p!CB!tr<cRfK37Z6inR>kp28s@k*U+j`QWxH&k}gmk
z-zzNq%tg|zyp`k`Z4&KgO*}^PY1&=DsQdQmW9s^i_eP3fe0Bb3gmizIUyKc98U6nN
z^%<vy12;p!E;-pFcf#btzrvhzb2M-vdi;FjMRCk_$fOGp6HOGabdPeq2qSYLXvtqm
zO>9>;Gr%*ON6MgE6!|b=4<V1qsJKS>Ze(`sR~nUPQ2p~MIW~QEPWQGZoj@}!F%vb)
zRi5~f7fa2^VId;<gD8ld1=&r^*hxn`<R~W)YGAO0a8xJv*){X>O_B`ZkD=MW=_&kc
zpw$22I_`hbabX4iUyBf~H?LWVn;`a!AFe0?^%ao})Ax4Z9i<S*>%@(jLIB=HK8Tg<
z<;rK3&+gsV`Srb%g5mD%yB8ZTj&Lykrimo+MjL`jZ+8ky*=$I!J<X4&-*F-$6Jr*>
z;)XT6gJ`sl6izaD<&2dxe6|MOn5oq>uvM}7dFIB0nLV2Yz)1M>Kbu$)P>=9Hv>1hO
z*EO@z_}pVk&W;z&H;568x6UgpTpoCgE;9OH2b~ttN;z}p4CN-4(_{4^RT&F&^opM6
zJ)l7s^wc4XmXR*QtqyQ?eyM@Ml^K_-Qy#43?B8GRR@;etb}4q-F{Ui-QkzHmXH&5y
zy62k2e@qPiF16y-A&;4--QuPGx_^D|?K?5pPi$eYrWsO;=@B6S*35<ol-jcXrS>cL
z=Et}zMg81Dg-*@leMwER8*voAeH$tstP%zJOl_^;P8w$%MXwvF-n1!G#(o$3zu0@r
zurAl_TiCT-CLjigfTE}%9V$qpC<ZOv2-1jjmjNiXKvF;i6qH7hPDK$x0qIf@q`R9R
zXFRU6&vibY_uKp7&$ah;?T!56d7k^8_nc#nF(w5j>ulk>;BQc6dSLz_Jd07F!L%kI
zKpGhkJ)ik$4QVhML61jhPhf#>i415!C_qkX&`aguYC+F|v5XW+COJ77R0E=->Fg{*
zfJS$fV``@aC8RBk(;_Th`QC&Y0c9RS_#w#~h|YLZ+$)w<<hB5a4uIGJSvd`j$=$oU
zMYcw8gM`k8inFHG&1Yw5bhHbM_rRX60P}~4v9z@4eS)cq>U08JOVm?L7JK3zIfbgw
zrneo>Rb}0{FxzWj7RZ&RsYc!N<o%ZyUL1I*WJ$mJ-{>i2KlTK#PZfzoza^zYsT}fA
zbYBNUL7{>@NZN}t>|3iW7C0A?Ew$?O<46E{%jQFF!jl4X2N<kVEvaH|5-}Wk$PS)&
zUTdST@wK4$ka7lvCK(Van(j}}naH6jn^%<5!+t&loB_O+s4sKlYym)!`UAL8D%+8f
z1xTnc96XqFr}J#%Wu&S~YTX#8jGK!lJQWk9xn+)%F>hzx%FM55YIcE54jcz0T4{%2
zf)IpOF-+xvU@PWRIGf<8*A@&6t1-k4R$AWGnyn{3sK(}l>Wt!;kn&F^ShocOwMnA_
z{9z4e93Lz0T!X^;%kJ(QFueyO3Dp9l$k_NeY$$Tkc*9X_wwjZ%jKv}(<O(ha02~>4
zc^c6V9UX=EYWNRmBUbvcX;TDM78Q3`u~yk|+@tr5Yu^mY*M6rLw%fR7?EfBcFs>Sh
zw)6oOWRb6&>6Ns@#n&f+p<(=CRW}3GOYAxJt@e(NmGta#MW>v(_~0DuLYUp8ts97(
zI(6BZ-S^xPF#p0w&LwN$o%rId!Bm0&3u*{6fospTK>V`6;>^@a_hd2}*-K%FoC;}q
zKyYxodl&ZLE4jEeFwl^K>aEuTun~kKWTC(;2|!j>F+Hl>K%j)EYyCq(JA#iI?2}~H
z(<^RcUYIbn(G)kH{Q$rM&R{WC2<$IFhS7`3hjO&Cd45YiW(7X2aJWq_%3^4Axu=aA
zimeqP4c5T#zP`k%4X7G0n?`>Xj0(aB3nfgW7Hkc*SFX0bkPN6Spy@iRFx&@~9y&YV
zW;I#Y0~ElRNvPv(0gT|WOKB%EqGp4iS$9JNx3I9*M>DW-kYhrfwt&hSxz!7gP5gh5
zoxqVf#NfKQmgh@fJ{-5J&y-asoR2fTx@((<3wPzBB0)kmAq9`LOF7G<kXXPTs%1jz
z%3hh!h?NJtXe3*Q#SfKpH+VXAozazGp@Z7PrCoT!$Y&?U4K$%#r#bSO)s*B7>BaAt
zT!HN0AxlFV6fS+3OFM<J(yahC@Rdl&4txdy8>$;(@gczy1ZW27GYI_c0BEXKA)Q2i
zrQ$+y{u!oZRS8dHpC=ckc;Ytxz-23o&B#ZeP2T;`0*se@<2t&{JTj(6z$b{ODTBhl
zaV1D|Ggd`!PE*rKV6+$TLe&iQM8QBMRR~7Gl?U@x@}q#|cb9EFmIF2yM^8LExm-n~
zN0pgIRBH%d!-o^SL24fPS+a~Zms^3eRugx%B}lmm(Ps`0xGQUGbXg>1vRi~Cu)GL^
zI5E2_O&cT2MI&fOrv0#~2qpSkZduTuaoccW*I1RW8e_B^)(Yyg6+kf1M|tA+KvZkw
zQ6G<hs))A?$nLwy8WtLwWApp!U?>1dWc4|wdANBJ_9I7oD4<gz+JZ~b;Tz1z+Le0@
zza`*GA?n5kpgbmISSTGxe?iWu(mC<ZB{I39V^Rc2fpAeRO2P(1l_!~H4Tc)w4J`V~
zsR>pi5EmpU!rB--Bz4H^Q6E4vn1SNb=+hM-HNg2n!aJSveLuiW5y-9xxM>#GOl&2*
zrG};^s$I};$H8+0bmTKO3)B7A@A;heo(I?uRD&&waC(3sQAky*Si9U^Ly&)SEInJ9
zxtY}t{A5rHu|Wq<{YLzU4Kn6fwaUh9Pq>3@9umY_CQLB|3@(|@$;4!Yh+3;OJX4Q(
zADCh_fq<h@3%gHPPW`~`0`V8c57byG&#a*z#3l`<zfk|3n(hmV$170|?0V%JSQ7{j
zYQ#wFbMH_RK^Kf?Zz_26qCTpu3esBoKS4?JGh9XS)?OV9t)rj0^`zH!@Jmt62_#S{
zta@rOhY_M1qZQXgZS+4dArB41#!MZa$3RMMxV(Cvg7(Ja$m*&pzgP4av7iN_;DGq@
zXtN1y;1QX=e*2~@c1A#81(}+XhL!@cRDlIxz?iqN7%(9baKG)@TT$QN1-=7vU{}i)
zfF%8ozN2EJ*#yz^Ym}PqK0e?}*f$WuG`<~bk|4>j`bv|-mGknEGa$*q0do%Ci$o9P
zZ1e;UdTb<`tfQ0{uNTUd8t8v|p@DCO2Mgk+ed^ykUq}!4p?U_FSD7`ppr8U`2Yl1P
ztz+j+72v2r_%hgb;Z(R9%%B4R@1d&cZb}A`{l`?NM@I>97fgr;C05@z0smbGvAQLP
z1biMQ0wX>3%lY#szGYICudD>r=Lx4%>3Z3CbkfcqD=<8(v8VG}-(ojr^>NcUu`4_%
zjdjshRc*t5=GHDG47q_0*G90T6$r0^a-8x$nV2FWhzWooAfieGdGNUV%m=JQ!rk>l
ztal=zx4;%h+Xtag7HuEoBo7~6MrQy}XbTphgi7-Ib_R@CMiUD)Apmk?6O-=lZp^I2
z9{{M4^ZB`{@anY**4iNI)Va-i%ntyZo2H!F1e^#|KbqB&!51~(zEOD=ywc}2=3HG{
zgC$-s;#IXP6%rdTYL6m=0SFHr|JtGj5E0xS%qqk~^g@wZtT&Dji^DU}&kr>tY&fiK
zY({Z}+z&X!ju8_eM9-tchLs=#qOFt^8+K}yKQ&;_ULt+u)C>>#>mk5l9JbtDmbf{L
zfyqF^vcJ?1h#{3k2~BpH!pmz5#V~EI#AWn560Eg+(eZsI>{Jh9#RX<0WY~nOxE;Bv
zz?wVJuT$D+ud5?W1$|=Pv#a5tLB@s6C11pI;>5+DzM0d{aagfGXGIx7tnyeb@M-UK
z<=)^u>5Rq^JdO@l?4xND{^3#x_98f_NNV6G+T7d>$SYkrb*}N+aiAR3j_r@3t40*W
zMZ)u_4UKeE&S>vJt)S`I9Ud3wf$swu+)oHR85!Y@)&hMn(n$OY7td8WzKSJAPep}s
z7Y{)g?C$EST^3!zVM3;I1I}~TuYUoWGmlga7D8@r>j0pjh9g|Yf~^X>3Y}+P)RNr>
z{zbHtrQ=YyBMJWb_V%N@vg^l@(&HT~aGYVU;GrXd&;k(a5}XT$J$UlPdJ@(3m$5#H
z_#P)r1DgzVC?g|{0YgEFfP58*0u(%BJrZT#ljna;Pjho%q)qur;jma9{Yo@INThYE
z!Adl&5Tnw`+i{@yzCqukZ?~A06=7VRUUCCzD3Xlyx~l{O9Dk<vT!_cKg@uC8Vy4HD
z;rxIqL(mnmT}&p-G=gI}`~_1#6P!LuQLt;N)<+F)%}>M!MXBt%{FE*>Kc(q~ezNc4
z#CRuEv1bcF%#?PEjf{zQ74P$GNl!_a&mI~`U5cjol<Xg%C0a1X-<tU?GxTie-RhrW
zv%fyQ9Jto?qb(=U=&Lzr9J^3^mO}Td4B6GQgYRr}dJHRWT$g|BO6@RTRVSQ1HPRzt
zvTInkzD(@RX2JtWAMXY45$}-@H%8tYyv6Mq-@1Mj8|6hFS6zRTV`LPY!*3*0d)kD#
z!gYQu{`D5JBVNIGR1-JU$L0&cH6TQ|q<J(^FwhxU3?X}nSS&^9WR80@JbVG(+nnhj
zzCMM<FFroL_UU7cNhw0#1nkEV%&{&`0~lq8F{A&d;4(MeLYah2pL@#%0lxO@Jh<5B
zPMspM1^<Jvmev;-93R)}a0vi>vxfQ^lV`k-zZ(Qqlou%qdKzX*T2GE!N6SRQfd*k1
z8H(h!Yh=NMIX9pqWZlS~$AsEk-P@=D0Og^{iRBMhgw&(Zu*3mRzR>GVfMhvFOkJ<w
zJ$1$9wOzD~v8AQX09TZRz5N2>xR0LnVPf(ufX_*cq|%RK!B)dwmVe5VDB-hrtbP^W
zOnD6${v90kP@>@vk3pb=0zZNh$ur??46qY$Z6-FF?_qeCZwzA$@cjItP1&$?tpcKJ
zUshUbPwT8c0edGVjs@5z9)=UP_K|ovt)ij$xn-fkio!Mf`+j(Xrz?We1LV%pLI2qU
z{^)vy#_(!uAm9kbFW($*q(<hF+>{KKspoJ>%ih}H0RRf=eXSS&jH|LAwOv~<i<VhJ
zpvEEW1u+EM@CMRi{~5KO;|gk8=%Sodc#fb9sQs0|c@T(!9I^#3fa9NReKi7*s{v_l
zz1ao&w`YL!Yy0=?oh$0Y-;icLQj%q`0dzzom|(oR#N(PR5rmK%kq~Cg3W>ZLZqo=P
zh`=na5~zRRSDDD@7AoGDv*nfo2|3_Qv`#%E%q9?6Cch_sgj^{1h7TC(f_UV;URyfO
zrYP57m!b~AEHZJjjB9${zrB!MYI{!$sQmr+-!k!HW$$g&y9#mSAT`~`Ad;e=x(?Kv
zs3380)Ad{<Z+9VF43KSHu<S}=^cww9G!JcmZ{p=$6pP`A`ZlosVk0Q`h%nU8Ck6+Z
z+;b7o6OYD^mPuT`E-Op@>NcVc;XkmqG?PE>5j-_@T_|4wxBM2vmu~9vY@%#hWTfh!
zOTym*w}z#kn3y%ZIv#h|*kDO^-Rf##phC~RyyqUGk_;y1cg8fUTVFo8Rzf#f8^xO4
zchGOtZ`a-9>P|K~=BjDMc4J-bo8{fhY(B4F$ge)ylJ^#U*Zj>mLZ!M|M&LZ7;JG5d
z+hI&Uqt4rG#5X?Kq+43gYIbtVp57!iO!p3@z4~U`Hq^FBw>UodBi*PqlKFFt*u+Nc
ziq-qPd-opS`?2w2;;3S}L?+&0cR_VrkN#R?$u*kIqLD#*(Qgfw%BhXR$GxVeCb(Sn
z+%?w*hK3nR9cG3D#H(L@8uJwV$hzpF4eJ+!f@y#(D8bXUI$(LG4=pLm|1s4MYFTW;
z!rj%?*M53$;V1!GhKk)`EGH0Bm)dl;ne{Mi7gnw`k$oD>VL(0PlCP{dGKJMZrk%ce
zkPT`NJXwzTm6!EUYk*NziUg={7Xx*%$S;6B02kPnZ%S%w|3FoPBa7HRYgIW6@q9%{
z9#L|!c!l1a#ZpJ>7(7{w-o%8rOU{*$p-{`B`K*frmQP!bp3UOw7sDK+0pm>P5M7=?
z{OBv6+0}v1jMDQ=U3CesF~p)kq`>so%P?ClzDpDZ)Rb|R$a4|3eoszv23|8dtATtG
zD-Hf4mn9@xM46E*92Ie%i4)%8-a7y5!*+y96tFO6Y)7MUzGf#zxU3-)tp4-~mQGsg
z>K&b(sBQ@g9AX)#p01Mk3BjB^tJDAG0tl`m6R4a6F0z*le`9WDmU=LenTZLfvKFs|
zDv_~*qr-Dz<X<L`L$(O0hESKRo+^eurg&AXYXtpOsT!pI2s_OXV4ZB&(8Ntuob5{@
z#*^tNdm@g^iO4aBL05BqWoZFW&>-oaZyFF3YLhhg0oY;r)~0_)r;t!rC_bF~60@8%
zk7jrc>U{I3d2vxuL|%xU{K&`0W3J{5U7D1<JVYyxinXWl?GY42F@vpGybo+4_`1}^
zQUe$Pljk-T2e5!cqL*f`pd}C4H4!-QAumi&Bu$~&gjPhW`)fk%3ENW&>2`7WTYY+b
z06l>fuuLR-Gf$6*U<|BAwGf^JXq#w5ICTEF${c27Vgl`)bZqhnpsv2j$uby+p(PJ8
z&&7qK+Z`D2<S!V1I}*Wt#fJol78!vj+O|-hvh+;P`y+R<9}Tsxs7GY=Te1c4bBaT`
z^QI2&b|uuq-n}Qi;8N(S5x;=)5n)SZ1178dNY4Zg%ezh64PVmJRIyU|JmBS<vW1OD
z8+KM}-O5t!Wn67d?e<G@L4qnjUlsF5y<s=GF!YkwgKjC>+En7xyWNd;8MW$r9;SI6
zQW1YDtY&}sxv{aqP|y3#>gVEVKMn{UIu({!_eiKyD2Qa4%X2N=LT~x|<IA<tFL&lm
znm3FKW$S(PyiFM>D?_u@BOMpns=2njTyU0V>uPC~U3PGq*Faviv~yIRiGktS<jL?o
z@iIL8O}Q6R{hrj^W8VIQzOX?^Y<wo}qh$HB#E3N!xlRIYwVH%j@5ouh>tM{(Msm_0
zJ647!68dwif_QeQ88ek$|1*?^*>9}V@|fpAJRk5T&+_yC0~3VZy9wEJ{5gzU!769-
z%GPCryT#Q5pTX9#1VV=7nVMY{n++)+v`>IZBE_{+Q_Wv`2p&5=gXK^weh#vx!s232
zEG(j_1V%0rKjdTb&je8Qdwmi9fpa2NrzC%lK9=QYd>-NedQj*me%;c7<Pk|jQDGsN
zyo|6C5D{?))H-*=9Oc!*%nZR2^49A>y)50x-SU+b3gAJ3)vR4^(eko1wk<TS8T}QA
z1IPsx$#Ym9fUD!j^&-0-unXYUxo)cpGx6RflsQ=Z?WOP5ktHCyaTfm|VKxJd&ayUZ
z^$5>4d~a~dPIp;^e+H}sQc7lKbs+8hWI=D**N23-sE=U>>S`P?ho55#NDFi!Rq<)I
zO#Uk}SaT0dv&&F*p?XBTg(SWjWhY_6Y-fjZ8w|N_WCU2y7J`P;0o>)zYYVCU6*z^%
z+O-W)N!i%j2L`OiOUg)GzTEqWE6<kH1bOc2`aess!vL9PpdtiV52Spd6pC#<C0#Dp
z<>cUtuUHyVTv+J7)M*0CCCJU|(4`CrD8ckgWcG+AK(o)2heyH!=*iYujI>KfIgD!N
zTTKlJL|q7p173Fl8GxFH67WUuNAzC5uI~f~7hl@~ArFjSR0;6Sy^HLA=GUdwbOkVe
z)bu=B5pW|i_xQj>l9lo|pbY+7@9!a*F9%@>&ks;6VZB%I8ixy34Dy2gnSF@aH|y1j
z8JSa4J-B(G+yPF2{R0e%wMZin_J4p2G17=isfBJA=Y^3c)`u&~g(&eJqVW0UIq_W)
zCkyZcpWm;~7h;n9wc{T4Xca_s{zEs?;i8T&3qU2$st?x*A{=hES|FJoRndIHSx*Hy
z`%x8#K^#t)#Gx$n6M(kw{=<ivIl?QdfGmk(iUYtDF|i&@O#ASGP%R@Mw&t7*i-{Q<
z98BIQDbWA86XCC~O})!4UYK%cUWesdo<JpiwR*YvJ#1_d*}04FGUF#n6E<_VkHs**
zu$asWDCR!P+Z4WDklt87d@7z(Ek7k(>#2q*?Pu2W*-h?WO|*P>QmS0|C70rB86<f(
zpvA|=HlkLoOVrS5`dHbarhpv%g11j**F496j+qO;ugEad5O)%PWV>&xXf6Jvw%mWF
zl^HHzS{^EAJX5HNZ>~;SLq`R4CS3kZ&6h8rfz%Po<|4M$jCA0V9J;0Ivu`u82GI6c
zgtT;URT7UA&!rPWvWP{o#X(@9NQco=_V?7fc=4XKO{7HG{%ewwtd++oDT5OdPDC`E
zn=>;d09-K2?(6LRd3*+#E^2yAh}6*1N<SI{#}jmOfsPE0_94-OWSLL?NaL)Aw45A(
zvo|7v;5%rcrUY6L6topvLX2n!I*!drpkOl0pjE>PpcClHd2uNr1ou|N5tY3G&+f{^
z2`xfoRy(c>CgIpu?urcMRK)}4Wc&7UDIYl#6&ouIl{Ua)lm`}eB<OwlmR}EgVk_cY
zNp*J)KpSdihGpXH2C5LW+{G}A=tgWIMBih>!v&~pakwvzHeYkDL=xfSL$}k;@3VPr
z#0XN${dK>g0RW2=Z6e9)8`v4hz%@`mv~I8sqx=vN#URU~_jS8+8#6S{MTBh54K5oP
z=?&(U1@QE>fBp;rU1+Z19O453-aYxag97BwM+g*X%1H2Ln}d~#o(S9P)OB`eHDOA~
zTm8LF(UpSXpMs!12H2x_jvSnOMsya-8aoQ|Qb-}EQCqwy1#|-(24m{Of`>{FeSt3T
zB0q9Mvj+R)JT!QO+VSu4c#oy%>})Hr?t+7ZHzK^i8^>*+)BOemI%-;4v%7c8K|mh-
z3O3*^l!WjI5piF6Bf~6x<Hjn81(K(d01mk?wyMp=<9|SmFBjqW9#1A7g(c<?0Y*3#
zR)Yfu7n0@POL0knrRmF;lrLd27_^T1NO&;Y1OvPhMKHbz;G=d5kK7a@B=Qv?E4KH^
z4tyDHa)F1T!U6skF=)w=3nF)SVAg!gt~q>ZfkKBlEP`84SKzR&uHQ+qW^CUR6C%J9
zFzAecD}IjrL_p)zEiHU+^9mb4d8!!v3yEqE<$<2=`iFf2okFeo&HVk`R9a-+R>JPW
ztIIWOnU4AeJ~1B!iW<a8?%l1Ht{OF2k`ad=#_ro!_^R0ak&sfC94{Z2fL@w%dDHUG
z@_6S$mH_6uF%ju?9yxvE;ppZ5={PU@^4)6ZbDh|NVK2NiKM$#Ss1#=9HX<4#m=8){
z!)ymaC><m23oSm-jNPvXfpIm&Uf>oJ(`Dd+qfI745E4Jvvg2Fv?sD;SWUe@i^7K4Z
z!~084V3r-q<1~8h?*O21zd*P$%aX0$^#jaTK!0?RXL0tSF+kXM0ziU057OHV7xRbz
zG!Pvx0|U|)T^K=G(qa5sS!sagB<glR*H2}b=jZ3wfh(VH_|Ny7SG>QNnmF|G9!^0`
z2}DP&M^cQF(-MQcb}+X{U>QXVIi$VOE*B$M<d|B7(-kPU_wRM=-E&5@RDBZ%Z_f2?
z0$-(t^c>5aXgL6_sk(m12e2W|&v>@qC(4n?dH?-nS{Hob@xH#B_&Z2IvG=hxbeQr4
z(T>3a$FciTmW(MfB;;3LpAF6({cO#Jdh@4l>_R&giVl{kneMK9ugM&?y0VgTXDN+?
znRy6js(|glb;PXVU`JrB3eQJXVHEuS{lzlav4g8xd)#M4;ek`|-1Uzy${zFG{h`iV
zezAyOG<4VD+6rahRb35pZrFSNlv&7(4*J><1`KHLuvm+rJC_d;J$xa&qcbqL6CD@;
zWv7F0Vk(5>ynvI_GVB#iN2);98yXy}>HTyZCOiuxHyg<%<VgJ$Mqdy2<>43wP}N=T
z;kJ6Ogy@~1s4DEKM7MHaS_wT*BtZu^f0BMZq=Yr$RDVAb_b_ym;P-JvDEN>gLz-H4
zmQg6v0ETdkj?#iA!~%z%Ipp>&J6_S@iAfe3m4dXElS7BWuXx(v-_TQ2=ef)q_pu$q
zpomqZVaUi+>U@wm1Ic{~rk(YjLa@FI3J6CA=+^PT4j(zfB0~+uxey{L3N(vm3BS>T
z`4fQ+_7fteS>N|x5vBz2*8b+lC6Lk5)5lTJ(4f6N^$ZlhL8fa^bYmiuYaHF0*Pm_2
z68bVJFsb)0$#HTuFK~kVjMnSJDw-)7^~_;4b4IIA8ct=4S1>HHv9X>Tb!f9tI;12_
zCJb=wdA+Fko11a6y)Qk3)7n(xlMlO2b5nei`DS>wvuwrf(vwS+VGOD{8RydmQ+%ma
zDs<j)oAkD&PIf*rsTvC@tMhPUU23W;jNM}1h>eMec@n38*L=!!v?D|6o$jLBo=5pX
ztEPcVF3&a?JwoRUod%{#vZrFApL!@yp2|L~JNQj`y+L$r+4N>|Ro-y>r?i{tH<Pcv
z)cWUeWQF^@?et>V1>GOcvFrJfI-K}(hY6C2?Q`Vk{Zj!tn>u#${y8U~Uowla(;6cv
z=I2j@)EGjGf$3)98^dU1N>04^2V*^be`Z8+XuMeID8d4%3wRpN;=jxf7A^*I)}5j0
zuxg(D3#}z~Yn?u%S8Q#Cf}n&Bg*mvifrXo?9)iTR=Y>A#ahBm;2O&z}c%ve8aNpev
zXV0eT>ZGfXUkO4AZ}hYeN#6PFGc02`<qycE@bd8yp6sw<gOsfc(FCpqk9>Uzl&CTm
zY<bnURLb`UKJEqly#SKtKMbO40N-Ee&83s?!eAQ$9G}U(ckdnmm>>qxF4!@lONHYa
z6S5B6buBckIanvHK>ZRtR?Hadi;oyNdx36f!7uA}rnVnC`QEVB(@*r}k-H?ec6C9c
zzNTgtNi4}ZVUTUHYd_ch1wcoN3u20fwa2sk^M%vqGFvzY^h$eH>o4CgdJbvC{ob1%
zvsE->JFmk9TJMP`_zsc_N7>oa|5H6pbe(xaO5YcZB7~%OyVZhswz{5%sy+1?a=ll(
z*)bjEHKVDcwY5SZ3nqgF1@TSHHC5I>EUs$L&2^qNh`1o7r`LP3RN8+hEr7z{z(DP{
z20qZ)Ap^yI#0kUKNYx4hh4tm|7Y+=lcUgXjb+@%Gq3Z%{4F%ZaSXfaMEW;GZ=W-vC
zXp}G+M)OJ6hA%~1N7o~{{Et`i{M;PT<>jJ6Gk2B%GFGyYm-=RJ?j0|i6gkq>D^GG<
zbkcD&*7#Bua*o%;QF3B!Xyn4f|7?|{>uaiZ$<ug_y-EFe@8|Zzq$2BvAkoC-%iJR6
z*8D>gAG&u%#p_Hh9WMZ!%+^XrPg`JfP=o&FwKELQM&7+^sDB)1a{T$`<dD^NN6JEa
z?wl2~cWft>QVo<0OpeEI&aMSt+v0a^Em8r}B|z;lR{5Lk^_J){cTwl1+0LIEC3~Dq
z9daC+Y*$5!fQT>s=&_scjcxe!OmOwrK+d_ZB91PnjUF$wUMn>1TD;lP;;0m4n0D+Q
zJ@?(h)sj`8)wn=0m!-bWV_Or%m(klICEIrG;y=D%^`qjATKIe6h-yT1(~%df+%$C5
z)X7C@1en~hT!*p%2;p2jESkx*&y)`+xIcw`MQd57tNVLCczHeAcm(xF-SPMKE^UAl
zGo5e4sG)R-85|Bh-gZS7N$Mg%P0vZ<#o5&xD{se|wj+P4oVo6XJ}eviEO3bRwY9_f
z3=x*k2Z+un%t#_C0fPJA{q+X0rzMy2TZ1aZJ@y7BR%kj#?2>Yv(ndnZK%YM7i!pjx
zV5}&xqCa&&=kK4^K;m)8qV#M6J`X84gMjrdC~#1r?-vym71a^YP`8&!_w7C0c)6DB
z+StG8WiBX3dm3T>_+UqEgD0T~%L@rHMggwE>H_aKUs1yz+S=ouI7a*x9?oUjyYx^D
z$4!esBr}q8y6D+ZVzqxgTyQfC1lvz4*fZ=sD)B<ZSbHQ)UhL)-x5`-s{|h;oO^)ot
zwER0dx>NzTcAxn~TReZleULRAh-<ot0s_k7^0Jy!c?z0AeIe)w<Rg|3iar2r1AZs~
z9%lDM!q^XEkonu2tcLZGr64s?WyN8Z1^ZN~KI9{OcgFR1+cWr|z?Is+q~zpOe{->8
z3)EPQE%lxg{)rPq(kY!K27S(33fVoSbWXRlrCyn@q?cYAd)+Oj<EB43pF2FpR#RU`
z8qE2UUs_3_=n($&`O4&5tB5l%j0X*O=#o)A*Z8RFkk`Cx4`uPHQ@7pihVcUPcIP1Z
z;9d{Qf;Zah!l7YN^NWqy&F{8`%Y=n^T{~%?l-(0<pW+~IKhI*Knx!P$>L7j2?rep~
zLUL_fZEejO*MM+ej#C-^LL138)aCV@K%I1rfS@3&Czthb4%p&mVjrN6n6+b#Da+p2
zPjOUWrYo;9=s>GNtU@A#HRpKav+>cn4AzFY7URQy{leCLbhNlRQyX6X`CPGQ``RBC
zdo{EHMQ_X8GWm+LcImNPVC#3bzMRdoi}B`HzUECRV$e64+d$>rw=y&8er?%Pb_v!V
zEYUKKV=b(fnklDXe@=A80EMAvDuQ|5NC*MGPQZW$;K2Ae{ec5-ao{&r1!Mf6GAZYZ
zTP|Js1hr_=pV<31#bm0+yCEPvpivJ5{N;C-sT4p#M8Xzw?b>+?WB?#BrB*W)OUCMt
z+~3A@e{T2V7asc+%Oggd{SB7uod}m7!gr{Rikk`tRb9ZNv9fuETT`i<@0MWd7tuF>
zH3+<k@5)=k`_G>#SL*h#J`!8D2+SjSqYyocJ>{*Tg7V32`S5C>p_P?5Or-RSE(ueD
zN?e)U4^QV-O9`fpeRJr7Na|{2D~yPYNX!cqBBF!xkCO`X!eG<n=eP|#!DPi8bi4*b
zvQW6K>D*o;qGz~QVTJ7iGaKjl*6>$8_Lid^hQAi}HE;gd<afE#C1`5$E9By8#dL`u
z$5Atr#9<YK<)2-zO#>@8|8C86L$}Df%j~M7ugI5k{AV3g`_GnE%?RoYZM3Yv=8U)8
zbL413;i>0!SK5SHNdG+ZawgrVJ*2-neX}Cn+x~3yxBLvtW9$D}>QHVQ+j{FXJj?rY
z{de^lH*b#6>t77yGZOP3cjy?r)|9r?$td7-bu&A_q|BMux}?IU(o)9Rsz9y$pzWa6
zZnKG&!#N@QHLm2ZZt^ZWO%>VI%NU17hk3Wp_XW@54vK{do*O<lJ-=GoTYf)=x<s>q
zg`Q3J>y7nutMZwj+j=sMq?;o%qu9Kjo!5E$bVVzpqhC)p)|{KJ{oVaL|9LQ%3Y9!g
z{<Z<J5|m!(^%G;Du`t1Vgnl%qqxGex%;GL?OqB8S18wmePC^*6!hoVGs<$fIbRphB
zYCdbvZ$imx(<#h&|8ZLon(Y=Oi>ZafO;56JNbTj)$ho3TE1Ri`()8!dMlH~WsPnQp
z7*v<_fF2F%rmLCkO~n|_q|(Jw(jjZ%_hN>@&+ps=zZ!-Y$RvCh-h@YT>t%^&5>{Qv
zN|@u~Rn*m;AylwgV=&X9YTar`JSEM~e3JCn?YUzy53@G0IM7DY6j0tteuT&HO9(nX
zri<c08qS_oRL>-a5Th$R+qYSFp?72}>s;iWgF>=zN26j_1h_lpch6c~7fpR!`~9ns
z^AV){*bjXAg#duybL8voO$@0=flc70X$!KzBdUBtlgN*YKs<XCb_Bpt&{srCjVX)=
zw?4OpUsXv}M!gOItevmY#?G!e!dN$9G0RWdeec6ZNZa<>h>m`_RLFVsBel(p0-y{a
zgZ?zn&@af;F2a1tmiMC9rOVTl(~xhB3Y~@;6%3y|ItmU&jhqh;CuUx~d82Uj292~#
z`08MENXUs2j`z7oM8bVRJ_VI@!k_^?*G(jJ666FS2F7qr8|XLiVpJ|oTL_wG{^E4N
z(NH6tv($K_<~Vapb(a;@R(Ew*ldE;}hu>X+=}LKWZ|MUByJJ_5h&kvnw{3B#F+Vpo
zVX~`hs_HRvkDdJ(Ly@kT`ZWETCb<{m@bTwSmR>h!`lZGqRC_XWyDW-ZN~4WUf9iKu
zczrad*b<d7_DzsVZjxert(ZEIIT_@7)u>c?R$!A=`V-9sf1NvecUsTQa^IgTGQS!H
z+QJe*nCoAOj%Xc|D{?z<IPAyUEWrocsanpyd7RKW5wfpx_oDkW-HTh^>~$r~G9iuR
zO5~$~j5=(hnf*3t{aK^5DR^wCC_5p!{C*G)cE^rc-eOG$kRnaL_=&@C8Wvt&_>{M1
zNruhTHF7}*w&uW6j`FY|=Qx?jxPDf)K7=j+BeA1FOIV=Bw9uO(Ns-4&$5_9ebc=F6
z!`z!FIMTI6v}7IZ^URu!fvTf{*49OEY9vIa09*r8sVF*!wwcxg@4>YiE~BQjsb$53
zUfapu9(?)uxQf_s8}9L6A8KmMKwLr}wmpdg{KIY{xi525VT%A62YeT#pGeouztARi
zOxXI8Gl1O+V@Y5=s@p+cGBh0Vucn=&$Qajwz1JZjf(fLiKf?o989G2s5o4_!LqF1v
z3WFRI6;U!XqiK-Lo-j735R0iEyQcP6Vn+#E+b;}Bs)S|4S`>tq>Uc?;cOjs1V!lO;
z(=@IK)5!COmT<MqV&IqDit1q(cIJtHw(-nOJ}+v?!X|ZKu5P^d`t}>XJV{L{H>boJ
z<)&UKrY5DI#f8|@n%)ln+*`hyd2KLMrksDnC&_ShLoz-lGtN=l$nZ;3>!l&-hU(7h
zlm*AnpHEWHC|3QJN$C)5>ZgzjyLZtA@ED$3FIr>3RX)+M0VPB>h#%BY2ag;nG^pCU
z`SqqWH%zkuC0ym~XcUj(s4G#O2*088E$+I?TC=3rsL)3sL@R(M`$-L89BPZoQg#}y
zx{^rXN`ZpZWNw5_%gxVlrsUJB4(We<ljgZqS6B1y+`iq6V-sMPMV`f`ZOA5X#+ic0
z6FPb3n->PXYn)8?6%XY<Ho3PKdd-tT)qd;_=EnrlE6K^@GxB=3IQ3IY;OwcBC+#=Y
zLpOKF=E>LzGU~E2m;BNZ&(q|re|w^2D&t$(r-zT;zkiQD6}}<?<@6WnfYJhp4nM%i
z!BxM2uKdjy|I5CQ0|Lf>mK5ByFkU2QAlEqbdR&J+pS+BL+I~g>hCqa2W6b#*rUU}~
z=GOiD)L}u#M85bokh?RIGbj(yiclRsTw6cIG5hO7$g!Bq2x!T(g-8JdqP|(6jYG8R
z=z^R1oZq~Ci|G?la+Ber_Srfm%;e|H+cF5nmCi9wqg|9k^8>4g@}$-a{30Vf)FL^K
znC$CxIor%g)>=hF?vU?9x;x`EwQtIJq-E<1cT2RW68~Km;b3{1m;7y2o%N;f{+7Du
z{Hzo-u*BUl95$hip{;(+Ay-+bZdq6$!vy?s1`zWs)*WIOe)_{|arR*}7U*6>;8sUx
z(-ZQ$2R3+i=v|YMlV>1XMh^fvgwJdz@LlqWt7biXeHGxJZ>>cE$<+Dgp)aMmvt7Ae
z2L&)VFic{z@xo4bk*)iRsPcbL7&iAp#GgpK`-PmQGxt~(^Smvltmox%8uiTDs^m#I
z#yx2Jm7&5Oec&?Wiy%KwNHKK|+V;mib$>K;LiTkgW9+>f`8VYSbk%&WX<)t5+#MHO
zGQ90lb-2`X&tn?vJxr_=Q3%=YkC5s`8IRpxtuN(X2G8}U2hd)yNKr$qqNR$VcrHl#
z`eG|mQbxenBj6Z<%P>*=^B@{&oyg$5Z#{$508vBFZIaK*$_g?`(B|}Y)eQ}ko@*BT
z=dyu$aStDqlYNRx*yMyo%16EHxIg5*g@1!J3fh;K;_y@R8E~z<(bMv$34%Vd{(bx1
ziC>ra?<d^1yDiP{%v1BJ+h4#rBy#9`UINkeZK5%SQC!vueVn@lPcTcTnM)>>-QnfU
zhmWAr`*=nfM4mnL<F#uqW^4B`d>fX$#H&^%)Hc=eHve1K{BNh=YrTQx@itp7Brh?k
zRyH-QR;!57l9dX>mydh{cYQ9P<)-*NNHa8DR=hm08J4P%D)A=6F0TI2c8a(;4yHP-
z<=&DzMpJD3F7X}B!%Xt5>^%Bnrb>(qXQuJ()~eU~F(E0ez#2fLw}f=(tG+wHC$XLQ
zX!Fp+ItKa!1A`~gG?GF_A%IH<m>0uu6MBN5-w`)6rW@xa+S=ifvu!-3^%t>#^bJv_
z|3=~+!~grT*>$~}paerubi+3gC@CT7STR4!G+FpxF2EM>5(<}uiBW=tT5P*O(098=
z2p@36G7xh${d7QgJ+1pfi=(rp$Vk#vUzKT9vK?Lk__Zm;e#Esy2M>Y=mATjfZ8|`H
z>1Eb^@zKNroAmn(Mo-eE%bm10*0!@SpDABS?8BVKHRbU^Jpg|2YvAh(CfpqW_T|_e
zaX6qS65^4t;-<A#)9!B4gF^wED?L>(K`|9p!+;V!(IH2dxX!#EpM^~YEIQik%Q~{8
z-Fg$)UK*fBW%;D++{b{yMrbB3#0T1!Spcp?OE{)R0cz8bIB6}_r0T>wuihO*z*vPc
zqsMi;La>6b=o-082G)|Jcfv5&yPYQtofsf0*7DP$v6DZ*=g;_XS7)<5{4U{IeLgRK
z;!HZymVQ4?;!uS+cMV<R7V{Q9%jgdu(^I3oxXpgo9%65Qrv0+D<BjHJNmpfwCYl$j
zslA=6^`q-uUt6SBCB{l>PVfD0r>6POfmG$=!(X&j)|RWhu#(zm;<UXu!gNdTlqoN&
z$l_Qz(4WWO)Ja<zb^C2zF~v8j^n2h|$;6yz<>CSt!6qgW)JFP3*SCM4Javli0vu-v
zvksjP0ufTlDqm%Sn@TMtYQeS@G1_~Z7A7QDo33tWXQWC;QF7dKC9$yT{n#AhYM3q?
zUfM!Exdyd)au9RIU)<7~S0oN`O&9<}+twKpRs(d2K4*-Lk+Bc!rpYn0^(cmL<@gx*
z1jjyQUJi>y%WanD=Ya6}N9byO<<f}Ytme=%yICwym61eSN1;8Y(ulTO#d42ZrB{1K
zBSZbzljvvyQS7V8Sxx&SF1_Ft5k38FJ&l3^xP(@o`67ULy%I;+r{8P=N}%f%Jxc4t
z9S@8V7k*Rz7~A$;l*nFVzJ|E&o{oVu6p;D~m;Q`3dQ`S~+cbonL0QdDeWijFTX}rL
z^w+OTOQl{z-)MD)FAeUZME@zmcK{a9Js1&q8PzER1JtUKvN|T3ua1d8Ju2n8D0=SP
zJQ|Hn)wR#Z(W$!ChES5Ndp0qWU>|W=nas?XhjN~n`6wB!S8eIh9aAC9WN4<Z(JZ;H
zzRAAHZN?*)thi%!XT(rZaoeOhW1Y@x#a4$*_Q}gniboXvztMzq<y4(?f95XND{lFv
zLW)1acFyj(2siiY-+zDlJwWtg)-K@l9}W59@*ogyE32Gtz@*$wpFi%|oARg2e`UY_
z+ZlUtmhWQcyTTj0=S{IvKPA^D8G5aLWveS`dwGqrFaLa*GA-B>A@#G3n%;_ZXt%Om
z`m9a7OizXf)Hk!K2x2$(Zf(Zf{|2w-c6>Vv3jc6^&xOWoYx7>>5XV4IjCNxbGwJ(N
z5clkN>-HD1iIY}dVn#74i^T~>7y}%w+8<W_8N&7J7rNi*KWqz5MuDGQz9v<iywHaY
zBbB@cu3)m)<(D}|0mO2cuNQ7|v{|`2GJw)9NvoFO+t8`a4YvKOoM*&HE4TZSmLTV`
z@b{oU38PB&G}P>|=uqA$IB?A~IX&bP6ix|Sqf+ie9ysS=jD-t3NJ@YIi!d^A9m|DL
zIIXAss8>74M?N;!5nc0CxAychfFF#|d47hL`oyLxOjr0f{`1-sd6k~$s?i<>Vvx7l
z(^m-z7u{!e6Sn1U9xWLUcj0c0lF+vTt0ZgdK8UMp=z$=|W0uoC!NS22{(R9su9HH)
z>%&9n>lC6c$`l*-Bhpt!J$>3%)H5L5!ZZ0R5f!avw|TV!1{j?@c`_-EFgsFxi5J|V
z(3^LO*hOh1>9}Y_@tRJP^SO#huhn|_Nl~^vyXbeZlTG^QF+_-FaOBZHwAi^k!`>Rw
zQQc|cJm1%}6x{8_VOhF1Vy^o9z@8sguYS6(OeCt_ek~Md_?p*0Z|-iM(U(Z8-j#gU
z2!q<xs6G8Qb5Izio4j!Y(~2NP+La&|ZKC1<n9=?NlhB2n6-<r|;wJjNQ~G)^`zgEZ
z(5;W?w?{irC;D1H;XiZcTEmWRG*N0uo{=gsPMO$El@n`tInq+SObr)A-lo0>dtNhu
zYjaq*Hk=O&$m}-o9Gvi<4qZUn&{^i@G@1wr+G%0cbBPG4H?nQ<k3uXgra0a-N6SFw
zKwY`m<2LQ$=7#!Qhu<XVDyVwL59UZ_lP|za2Wi&Y%8EN0)JQ%1X%C}+iJSwDBYAh`
z)<OJSTU|Bd!voY2;yejU>i75lsx(N3k_4>;>DS)Q&d!gcPDGnMVv88V3D{JTc%NzB
zFzC3rZcL30oM^z(IoP%duGs<%zaz3dZJmf(r(~wrsn=uq@n42&sRWO{<Dwexqi4^K
zj(!rgG02M|6l&rtzqa>PpywW;I}_D~W*kD+`2v_}JM8$6_qlA;mxlsw$jOnw`&|&f
z?blG6J@+f|D#c_x%g_n(BMa5+php|5T$zKk=xT%VOE8*tH!#&16Vn(OB`)9!b7Un>
zwt;msS|xI59|PW9HK*|D#IHUOh~i+uZ#~4x!SN&CGWO2YFT}p5!kHbQHzoyv2|z1g
zonZf;2_VE3>r*A*G$Vj|fT~uk8k62?KC01>vXP{S@F(+>=aTFDD_5@)`@eG53s(ii
zRGQlytwF8($I`y>IDgkX^<wtJuH*39-?N2Y+8Q#0q>rTt9<zR)&}vrwbJWnQaRv4l
z^dqc8GyaqewNh*I_GZ5~rsk`Z${twkv^rj1rfD>vwSzS%qm;ueddD^4Y9B^2pL;L!
zk{4%&K3*X2Iw4$A$D;2=M!(p_l6X2jEUx)M$PN#>g(r7cl_Yqhc#`u!*SKF*UDgwQ
zIZvL+mL7RHGDwsz*h2JfZ^^nNd2eH)>aN@I(!2w~&1kLKpqm*YCn+Z-<qF=i&sOa|
z4irKLAVC(LgiLP(O-iX#+Q3MT2s_?+`*XkE^-}oSu%6?Eb4)4_{x!1Kgg;Q?UM~y<
z5{V<&sLelsGq2;Mg6xCi=MWeTsAP+8KKTa+FS4%b=GVbe$(Wur(p#~GxBpZzl?3_@
z(Od@ePc?H*+$RiP@7@6pfUic|GGqHTqgvsYmYME_Ru1Q&SQTkD*egS)WU24+B-y^T
zJj-rJ^s(Tj7NEKgE<GBrbmvWf4Gk?pHc{ea#AcVSoCXrf=w9`wQg{e>?@!_+KEiO}
z{`Os6-QD2B^kNDJ5^(U>SXonL=3bH^7J*lZfF^`r>*?u%KLau@kh1aUx&i=iW<YBO
zVdwzjY09E4A@D;Aa{lg*4mgTKkddQO3_len*((M_rj0&1$OPzLow*Bhz_n^|5-(cv
zm<%ug67=cQU)t26nEl7OQW{Iji0q@;lgzbtVvd2`KFn7(^>~d#Az=FhZo>W?N)_cp
zWwKIIW6;on9>h^LcA^90#&#WdKk;oQ2G%86tb`XciA2I>sii7Isym5njKhN+?PVl)
znFj4`1Y0jlTVwuZV@Ikk_u#KS^xuyN+9r`&ocm#Qe@w)b*<}-F683gP=6h#MC$#zN
zy8)nzKsr7)U~+5iaA+9;F_W(VlBy&n8{o=kWMgvCK#Y)bKtTqJWumtsKeGn=iAfN_
zo8`q@%`Z&;x)k>8k_8)@KG&cp)r|S+?|~c^M(UrR3qg%t3VO|G$Zgn5q9139wm+B(
zK;aF^?rN&5bEv;b7!c)`LE8eaWsA$rgW4MBF<59VnOf~1J}}~2FfmEy$4i}U{gsF?
z2}xt5oHD^?Ky+#c_yCU>uwSnCPDw?u!Rtl!{;`;k$p82)yH0B}H)7F}?b$PnVf2fO
z9w=dfoF>O1v!*JiJx{*p(AObaYkT_%gNtJ~c2XMkSAHaG5|F#NcyHuwqf`73kC^qA
z?wam*Qae~uZ5T%-=;Qe{gTO_Z>Q0#c`P#{j%yzt_HQT@<CZn@0vTeS&A-2nAOmEby
zQ?qleXth(lhDzLhZsW$42TR&529`nHLd8|4R<F0BO)QP{`O527W-ULzHcW<+geHwl
zhrP&EdDi;PlT>4+eZ1mwx&*`b&>;{y@~wKg+&8JIs3yAd*jZSvKb_fX)3U1Zq{fi|
zAXLjvn2QV9oo9dcFU*R=`GnH}tU<(9MK5_m3G!DzNGC}}B0;t*Gku7YfRt@PA_4c{
zyVR4A=g$eckA7QdP*BjskHcZah=7oZ)i|e~)w!?SRWvg|d(m}T!!d&}+B!p)nC#w;
zL6)d!wxVU$(H=%ZgH%N&hUX+7p99Fw1S*xCP0TM$UNr!OeOfzDT2_|t&Acl)Z0gbn
zv6XxWD0m5A1o)QJ%zN~MYjoc$rz)e#21-~Xb>KLj6Q%*|dETgwBD*|G!x?8XaA?Fe
zf@TQZ=PUn_GfPOvkf35eX%|{wYH)T%LR`Gl8bLpb)TETEj2l5nqUz(qg5to&lxZLS
zleT~VuE6*ttOLpvd~F8^Z8~phgMicHrPs)`6Qad7b4WA;U_}+`!J>t(2{I_EN)<3z
zcJJHA049x$OyJ6GnDmz*lf>l_M~Hw*+3HZN=1nzD5NFzde5tP|b@!5(Nf9av3U9FE
z5)u-K46@@~Hs~F1>g2^Y?O}qkUGFFGZS@MzGI&Mf8&rP$m~6oS_7@*`aD8WT@mmFj
z)PvtgX3=1>xNB%);xyVch>cIxa9v79<{9s8!oUo;2l-JE^brMwg~3z+r~Kx|RTE=l
zNL@?u(GmxJL0S2tdvy)5(I?Ot;g(|e?iG-XLc_wUx>H)m5D`7rCkxPkSXy50h3MU0
zd@ngeQm=UOY?noL#4qU{%<jaa$?@Al!3oi{HGmMEOIhIRe$7(3P(&QxYC*q{E@FNq
z36(j>MO^c@o~e)SqJ-4W?9slnO4re>Aa*u<89JXdVLjR>$MLT2ldZ3;Jc$`JOLiX6
z1fgTx`ItzZ@lESQ<(0@DK015AhV@0lb>{el2p<)l_F~R5&AlHA#lvJHrM?`^-zw6{
zFAJ4T_HCsrSN+Y$c*G=)rr&UQK{KeG&h308#phJzv^SUauX3ukI^Ucz=+E-&beTIe
zAe!#XpWRrRtoF4_#X>Fo)-!+SXI#+9V=fFPi6ER7fZ~h7tQ8ptsFY>1pXnu{GeN)@
zDZ2L#u%>?Q0cvVkUV&!_w1H5}6c-f{?maN#1W_}m!nO*~0jyw8)J*~&XaiCS&jJvm
zspgm;4=dBNpjJPBJ{PQLWb#;;<oosk2_Y5%J`?Z8{H%%~25lKQo59OXxv>6OSlGGS
zpUFXM0s{!_?s8<;z&x-uJfW^94j8528n(@ne4OHp?F1wN&3I#>&27Abfa)Q+2jT+G
z@65u&QG72*c(T9kJrN;AkcKnK518j^rpQ?wL?&n1@qNTq2>o;D(M_9@{B8!1b-p8X
zpVaBM=ouKeVzp(&acl;Qt$jJV$?Jzc3Cz>f8+PI&4?p<odj!&Mk`GY;)nL8}%nWEF
zhD(q=V+-ZxtJ*(bL7kAN2@P(h;jY?;xR`hd)>n9yl$94i>PDo)khwGv*7oe%2jX5j
zIHgFoJ}EsU4kC<--_$#XU3)lEZ41>qx<QL`b9-^`I&+fK(=D)oKo*C3UunWsB|?f1
zI=i|cII2Rw04NoAO;=!kj7@U$ASJ_IGBPi$cyK~6l{gz8`PW~6LB+9q@7}qkCB#w^
z<ZdrJztz{@MHDt>%SuyBMZ=&Q4C%T_Tna?WKdP#dk$&LTLztF@Xb09vbbNd-EGo8N
zp2V|>(f^FB9i1WzOwy*Jf^zp<=7B~?q~WA-qGT6bD)IgIv~LHBLt1U{YQvF*<N{Dv
z%uQKzW091Iqxy$Jo=QukT^YTGG``={Ahq~r?7#2>ig&4yzSBHBJSR^|SXkaCN;(iz
zLuAjH5IRphyAWe%4po$a7?fxH4_CA|>BB?N<dE71Y5oUH-`a|;&1wJ1??*w*UWTrQ
zUL|*#ASO&GQ1Ljk9CC{9(Q(UJ>5M68va0h}Tnpi;<&N_5E22@FFn*$Tsz8UOb!794
z-nG50Rq?;egEsl+(h~Rd=V`mx8)<(kyw-V@TRFI#`dIu?y__?A>?p!uC0U4}t93iV
zaK6VON`Q0FAFc>iEfH_uo&M0<yM@6vhq)gX!#6-i<{Pr2^X=b&wH-fjK>xb30U~mx
z%ZoQ}W^fA#6qoBor15iQh=`y(LXAax=#U*|vg%(_KrXm7y+yL>$~9x^dB;dhrmyhw
z>Z^dc1Y#xnry%L&81rAjg9Y!LBlCBv^k^?1T3R@SK>E}zm4z({s40Yt3*7q3`zOc7
zFp<c7N-~PBDX<haaA*5SP*Y3g?@R&(!KD5fVfQ#OC(4OUja`6oOBJCs3t)v1#wGxc
zYq=~!v8ssBh}=^u%(T-QJ+Mg#J)dU`XcAt#=%T?(g}P)AVG%Fl&~jFW9Ka0l^A>5j
zB7i~~P$Tp3#Kci*YD}!kHSjtvBp~1p`X-7&z{IMrtmx2sYmKo*jkz!sYsJ?2$o2)W
zbU^fEhdK6p(ZgzuP)h^A2M?7>%@-BIERnqf8#54R#R&)cvV_>*5mFK1;W(FS3JV45
zsHtRMF>rHd8@LHTLo_S%3=J?Gw7X4obrw(RKKbww@BrU@0LoQPU1x&U4p-bxTHaR%
zCME?S=*+ryVC$iGf)kb~I%6V169+l^<IHp`()ldiGR^rjvT|~eo-M?CkyxK^h+(<o
z2951G3!4-M)Ft>L{Z}H9PY`ys#>%Ja*%5EWVcN5OU6l?Is)^MW1soLb*YZ1T-h!2m
z5tL`NekrEmu@D{zg<Vho1FAdpF1#l7N4G<ctC_n3I{`G5tfsnV>5S<_Kvo}w=o41Y
zoQ?{(hh;be54#OE5#S!+Ks=`#;v9<TR^YKWWyq@?#DSDMG!R2!!k?ld6xI0YOm%RV
z7~83W($ka?DoI?fUd!v%Q(veA4sSf0nHOK)RBB-7`!w~nM7(43{&TulK9WA^w^g|}
zJQA}g$ZeQsW#TzKJj>S78Q)e&IP^WK_=kLTImEAo`ITYx9hc}m2ZEbZLF+@_gU$WD
z*tiM)5J0Bl1#Dh)2;4-C4qUp&W%yX;vpO6uKX4S`geBi#9T^{Y1Coe-O_Z3%9P))y
z)5aH&{sShsaV-*4ju7rph801ec78k(9!H&6j>8B8l<?Q3Q-ndBBl<V6C-^M)oqe+e
z9YfeBI3te&z?Ft<p~)~8)d-M-ZhRzKq|%N?g?}eqYs4dl^M}j|l~w8N)f&|!f`8m5
ziGh2FMB`w_+kY7x8R<X@kG>Ej^j65%hGH)yRwIqaL!&-$0288icRnOYW;l!R7t{E!
z`1-^-QH}RWU(12PTediOzC|OJn0?L5I|~B}dek_O?A<x?74CdEL<p)nZW0Ce&b4~s
zX>ItUBAj}yclYMj9#hxU+__^1C#)H8^x*08Sr@-Zb!4G>uphLdXv6zjh#sKYx<c^+
zuRY+wnQpL?hH{Y@+ktGa4sJCwdS_pq<Xk13Mtp8R7`%oo7IokjGBo6@P&g{B8G%MU
zi%M$?uqN_P7}1=wE<o$o55xPAORMg@fawX?@qqS0hwBa5RM*mipaT|=Nk|i}dRUZ&
zKYWBsfD$hq`zxme&JBvV`|X)pPT)K4m<zDz>E|wAk+8TK3rA@*Pw}xtHCDs*sO2`c
z4Pq>j&{a|9=I2l0TPZ8IpsEen48?S9EmTg}W+48f1R%@Ira|aKtRS)%gyM40kM<Z_
zMg^WPX52i?mW;+_6o<Si7!fdt$QHa33lwlu0x~CJ>I0F*;^MFdVqU!a{5C)TIfSBc
zTfu2d4D99njb1QpH3bl}fDjNihHwWjmSb;v_dLzs$Dr<la|$deaF)S5nNZ1_(HX(;
z0B;BHX}x0mn)er~yoJyA{we#g6o~Q^3X)`|dobI=%6W&B29JhJdMM)4uf#i?g2jd_
z!5=Amg*ION@(QKWxn9gO{ww3d7Q?d`%LnA!haw}Q_=bjgo0nt*<p1GT+$tDzy>#=^
zLfy!+WBKyveR(6AVLJ9TRr~gtEDQOpvgjr8J%{Y(?3Q$Y3eT+HPd8AFAZaw&HEq3J
zKCZ*>s9|SY*ZJvP(T*Zbdc$F(;}vf1qi>UsU9dk&!#lO{aZWMqX03#LTM|z2-7Ydo
zAkUB*yhzO=WoyBRihCsOa6p|}FC!+ZC3>r(_+9w+1&%NQeya^vWWBt+Mh9oHL<m`E
zOg9WY4}AV$o+kt|cgR#RIIg{2%_~M%1H}Z0Jng7CP_iS9VIrk&jWnKORLjk3XNio7
zSp!%yF>!}6UJgD?fiXs$@DaWa8nC>)*W7>~s3~D#N(j(N6_0bD&QNlpjrk6D6dg(k
z3eXM>*xHB3CJqtmQ2s(y%(TuST}cof;3tGeE9#@DD7CkT`ae|@+I5Qf-q*d*DCVZ^
zB{D*JGi93#f1~JcZr%cp->Ll$*37)6Nv;T{C|D5U!QjWI6S|$26#ZL;XAn_DbxJ~S
zLJo6&gjI~~Z-jWf{@j2c3}DxQE}*_ZnuAOmCapi;#~;F3)*=;ca{IR3)DJ<kk=LL0
z?%fMV*LWOe5fLDe4*wzMYpl`w;OqyPjg^%Z_Y2(*%pv?t(&g*hhw(jmS6h*4ad)03
z+XOm<iU_l&;Fpa<e{6En^2-&u%MkA?#t0LW_+0lbqF#kvkl*hjz`lUbc#_<`&0uY0
z0OOS=N3?u9j|bHPB6o+>08=qkRU<(@23gtf<Btc(WuR8m($H94TT@M6>g_TCB#C<V
z`wwKgVWhPt!WRy^9PM{TGJQ~+yj8-oyM8gk&IpJqV31R%KKAqw`7u%+t|e&_fTEVu
z-E2$Vp)oOREtlYpP9;wrE+r@D33&$mlmObnE#cFd0D|F7cN<4ouGl*EYd~NiN^Kz6
zG&Jd$zK-<c60Gv@EO562GQYnF82WmKlM_@ixWl(%p>2`fjJBNF4}pV2F~&lNKi|tn
zz>95-xahFwWJhp35&fk6^_7()M@Xuv%4j;^USrXMx#I*SjFS_Mm+_*~otY3MyO;sF
zFscN#4N4P8HjX&Ufzd#dqh;O#im#YhdB@92Yn;reb>aAqfei4Hgowk{**V;H&dtm*
zrQ;VPZU-M9Ve*RP#}UH-=H|}U>=5bv*Z``GS6STLIcLzmzu?qKo{Wf9SvsoompR?G
zy<f^=*R3-hKfDMD2?2g4+kUy$K}1f!*tdnHK{w8667Usb^a1}*02c9#aR5rcy>dq(
zRT=A}z4pd=C^kY>J~|r_GrhmeBsbzI;B1bl`E-K`V3-|5x&4vA_Yk|-d&A~vgVPIj
z<C?#Jy`m;EHTVu0k?f@N2T&)d>!<w+dR#5mm$rtK3f{aCfTW92g^%&p%n6qC58G!2
z64aR<SHIPzPC0_I5fdY*Y`)?UJasaViz=w;d+og*R1!vetIB0^NxrxN`zR^BMys0v
z?ZayB;1&ecj@ZUVt+d;+ALE>~Fgo9j>DsL0la(x}^3nx{dbdbU4{Sag%p`pH^_O$a
z{^0FwfsnmfaNg2@p#M2*7EdPT4M1$Rdn+$LpTO>On^F?hiN?bgYJUaxJuT9s*p8KJ
z9zpO&I--92?`=|-*#o>hJ<GHH^MU6PI1`WOAEDq;V&B|N{^8f9haa}jgfcShQa*y(
zL{mip8eADIb_!Q;l&5OKwhoi84TQ;S=HJoq-AsLq(rjzxH)UT1P@MSlW{R{+*LXDp
zVzB;;$|)|=f~ttGMpZzVtY{yhF;1qw-i121hV52q?GX`E9_MYkV#!i9doJpuS58=m
zH29_Okb1ZJUoHS3EZU{vh?5M8&|E*xwPCV`vMw-iE8H7a1_bsoCn*=9rmlVjN>s61
z=qF3X8`EziC&uI8Z1&UA^_lLml@M`uo-SLKQc_)vsQUN%=7{S%jYcu78MwxEOJ5M<
zf(`pp;air<=1{JPy>cpJ6^g+5<z?<}63fThT7TtB+iDdM09OyQu_Hl>1<pH^Pm+rH
z?(<>$y(sEgfo(k3kz@fjRc|Xx>44s-K2UDk`}9*REGZc!3VpoJ+h%f3GA!%ik6;#&
zH{=MJaBcoMp8eBw+Y-oa=+Yje_y>Y;p4y5QgFS>l{U&{FCw>Yt-2VDKll9lO+WoH+
z{aN5moKXU|R9y#dFD{us#;My9Hck+(%ft`MBj^7;buQs(|9^+q**^a7$hulw`v3eP
zaR>hI+mj)^@dY|Mob}OfFm&$y4;-9^-DG?B3S`0sIu$8eO+L-w;lbnq)*rM9?rhh&
zdDeK%e<IDY@|D9BB;tqr9hyv{ycgQw1^(yf|NU0KZL^C983~k9>TwK*+U5^#%@KNf
z_PevtS<KF|rjzi?i)r9VxK8CMq__GL!=|<+quF>4$G?p<95xWtzNQG{w>)&?|Mk-y
zM7Y3j@DqOdze5QX_F<*|`MD_g3AW*Xe*WhV|L6AnyAu9?vMAJ}9Ed12BMy`eFh|mO
z4}@T*c~`Do8y*@O>+Az;ji!T1Og&*+-@?&i?x3+um`HCe)|`pwfKQgPn;qa`IXP(6
zcv|=2pA@7rFcfziiyr2u+^sBw{3PTSF(9r;f_U@^?gW5?$Sj+8q50}E-bumr&+e!_
zrnI9G__Eu=h<>OJ-sipYpHLbujDs!r%^Q|YPl8!|`?k{BCSLX4`+xr{(mBHaVRZSU
zfeP&7Z66as`J#*RnnKLKBJvwT#MV4?l8D9%3>3g9YirZayB&eZCz<IXoVV}UK4PT%
z^BPnGXnjN&Q*!*7Js1D)M<^xlN#Vo6in@w<0)uO|$@u*F3_(5&90Fh&D(VQkqF0mu
z_dgMS6aXC}C*I;x&smz_Rkzswz0T}VPGW#wCOpn+go&>f_UA-V@PwW{0nNJ4V*XON
zQQHP-iTH7E?kl<sX!%|hg-Nw7FfrpXOqkjx75wKr6d6w#7JwM&>gE=kdJey55&d%%
z4QgJ$PRO78y5Ff2v)Z1W`g8eNF!wuBPHI6m1j|a}$v^Mp_wOAuSqU#dMFGA>Ox~Y&
zV*2-;bYPKzI}8HJzjtNp-@D?3qJx;H=SYl%Kp%|t`*D1H_|89Hdhh!XKP!}Y_7E^8
z{CP8qKW|2Tkx=_yxSQb>i|I)$%{YGl`x1X*=85Z+!@nd8gE*{!i+%@)u$poN-W_K8
z`ibK`0Bhjk%hLJhqsae!lz$>}7Zl=64%`t$IA+`X?<e_YCB*)_04o2y7m?B7y?BUK
zX=KD?dHm1w|1E@D?EgWt!2^gw8HQII5KHgkTX;Kpq63!snQ&JG+ev1Pcv7r1G*JKr
zMoYCayMY-~Pn^Lo@iTLF_ntkChwN_a5PyZA|NF%IbGwLN{=c76|C2fVe^0E&bBMu2
zDbAo6p}UYgi3Wz8yu4buc*+P`3%_8E(z=2=9%LF*J|G2f^O9QLkEu_<(X&VqYVHUB
z+Ln?)JS$zsQ&$5qBcr_{NHen@6YAh&SznLh@dIi@fQ!WJbs0)_f7DB8^GJ1LAct`%
z%SBUDVhGHH4&X3!j9cf`f!V;x<a#%{)lkJq_g;AV5soJ)q3}iE<kjh{4Pm-H<<?qz
zg>IemGYxS+L^YZJ^9@G23$ot^yTWA-xG=b#=v#-z#aaII53K=A2_Z7j3_^Jc1lBTo
zp*jOl8@l}*ZffZE0Ex_0O?MD&>*>+u*R_Leq~C_+5s<xhP=?Ib0hXeq51%IH)IHEF
z08Sa@FpX>`86fYN;4C#ey49e{7`3kA7D|=aBfy>=g~Q=j!>ucLj|(*<kSi8S4<N8L
z&#S{6v-29TlG@WA4+TGc{w&p-16*S?60pV-4=>!CSMCM>i@o;_$GZRjhr3GAKugP>
zEtDCBs3b|q-X&zDVQ*K9gskipviII2Bb)4qY|1Djd)<%s_5J?-xc|M6<NojG_#DTl
zx{C9BpRe(JKGsuO2kmucT0F4mYLTwKh#UCHIv@mX_hffqmO(jD?0<)@<G8lHTo<iP
zXH&%0f!<(2q7;jdCmzWCP{iCUqPa@8o4buD2q;k=t1c78-x<bKj$J$MXQSyCI(IIS
z`0pX`AVAInTurNs_C*`0S@aO6O^_v`T*{+I%;3os96vLTT`nj=+(@11@+fHp7&aiQ
zh(tdFAP}BF|Nc@O(m-X16rRLaUlEtU2W$L=vW?8-D?4K_KD;UH3i<^(EgIyr;hXC9
z@5dR>OB_D&xI|bt-~9RcGap<S&|j%MbhOZTJkOXDkdbM@b3pH2SX6Z7znKtFydc*X
z>JD~vbCB2K#fyuJ$J$b{$^v>PSkyv7VAF{d4ZKtU|8AJdcK0rS-4+C0J&B6vv5HAa
z3G25e*bK-EVr}cdLJ$rV{GW8@rL$Oy_@;0U<2ym)gRugi;^N|BV{t?Ab8s|`$|!>a
zf_@iB4PZ|qu4SNX;M>`|;Oo%0L2RFzl41hj26G8O0hya1#bFHZk<Z|3Iix5Y95Y}t
z@Z+&p_$E&_Y#(~U2HO^(7BH*NkkI!0c?O29<m5QuL`AYM(z^+$27%4cDP;bR71Pkv
zL?#*%)Cq|=aAOSVgGjHcstSmB@WU-bG=Z2>Kroo#L$M0*uAuAaA8pyWLv;1O&(H>W
zQ%>9K1El1q5VOKs(v3j!B{|>&&{QvPY!Y8U0{>nWx3pZs9RfCqkO061@M{mi3KOVr
z#&dC7n(xA^?J2khlRI8aVR#<45)v>le83IkjHdShwFVA2fTfDWs=s9<nm?>ZJiz(J
z0|mo%IJ^W34qU4_)^m)fl0u4_1@KrbNA@wt>~G({!yL5+Bp&3c2gZz<@5)hg0I-f%
ziH{i=X3RHQc=$+2Nu|Dnb4hW{xA@&>12h_B>XCsf246qoZ|>g?1FiOVhc&<q&|(Vb
z`~m{Hu{WI$oQzT&9|moS;7LGQhc5vfa(?rs1S&stzlZZcvu%JCfQHo$zVh&ZhrP_j
zBtY~q88T)D?+d9~!~WoWia<E9I!58BL&lNB%2RA@?1>ngp+b=@NO@m=`}PemIg!GU
z$v6TV5Gn?PCs;jdmZw1-`f9qZ5uxixj~YBVZfE{$cxZ?VQy#%v5JPmQt|xk9U>7y@
zI)WYW5%hS8&frJ{nvi|!&Gin@$~f78nHqmfkL=<{IC2o5v45A`z8&QaGBPur|F4s(
zK71Y#J_%EdKrk~YPm0**PcK0t-f8xvlNE3oFygin;7?g3cJfJ`h%E7$(E^wZ2-aZm
zF2uluM6+iT_3ND_MjUEbqR?h(|NXOydUx%*AfUQEOX;}k)jTbpKlkoe=^nT-U9$WG
z$x!^G37tk3U|8|^bst*p*-BA3he>)_uqk|6E`_ddI1DN*#X@O}?!6r|*<Z9wtd+lJ
zu*+X7kHPJLQ@bl`4sV8<*WlyCSl|-bah#izwlCDcXE3cFMQ_^P%YFHBA&DjxocZF#
zc@WclxPKb|&JhxfPie&Y&-yzM8Bgi-n-T%+>gwO|cA;PFE004464xjDci~p)kU~Ij
z0z)KDaewmtJ9o%F$4HBxg3EH+C<w!_=kTTXrx1f0_C^5e7zASo>YZ;$BnQ^2#;_~U
zyf{7WN)CEKi6CpQFxzMe)U+^{AU|!#W=PJ{0Y?;uH)m91?jbA)MlGYoXU(C4Uw%s{
zVGHMC8R6{Hi4)liAsUq2BiKhBK<;#9FhwK65y=>t56E^sJc1wDUt*Ux-6tsqVH0jd
zJhRTEZm=DQp!oZEdSv7XEv@Es)uo4LpS=!{!HGmnRl)PFFJ#`9gG3kK7in9HG#%|N
zm6gm$bc1G2?Z`Ru4phY;003v*+SZ3UA{?rqy-SokM0bl5r@jp?C{<+mPV006jsUaU
zE)oXP2=@qn<ip<neg2%9rvP@Xo8B>i$PX+ypDy8rpA3E&-+)iFbbHG#<qLq#t|aA_
z_|03YDk_hGjpBG>wSJ731*{O$kd3By5iA*K#=}C<<$!D<7eX*!^%KIa4B;Tu*OkcJ
z`sRTd5E#B98~nbGd-;%7IBvi)9C5f){Gi~#JdD(_j5WlV*rI&?a`%CmrR95fck}1Z
zxpYU6A_gk?!^fwXvV3C{HLfTmaV*Uh<Ir~?ryJ<8rkFoY?R~2ue?g=ap;U*60NQFO
zj{v(MEDXgx@lUSoO#+AxMl}%dqNKLVUC3y*uqdQU)z3l{T6rlPu?&<~H|}?AK~+i&
z$JAm%JGOxReuBE!_qS5+GnW_8qr#d`NV{{?Oje>Hj?*_xSg4O@eBlY{&xsWcyn8~t
z8YwKF0AY6t=tKbfn<0J?%o-1ja@K%O12@N=3&~z1(!&)Je88sQdEyX{`Hp~7*z+%P
zbGOAaS7pPnf}izkVnTUX_`peS7F?b13S8P?bf%9w=ZBexPY^N3)<U)yx5N$NGgrjc
z!vBm1zW}?|MxJvxIUzqtbJ{e<t%~Epr|A_!1_j~D2D%8@m~lU!<@7h`RgO`Ua=im*
zv1OxzV0WDIGSFf)HCb3#Kr?25<37W%^)HUyUR*ACt}wdg;KQoj6OI!T4;AV^%}Ogf
z6pxS>tkG|;xa2=aw*$99Ot@Oq8lpsgs-A(l?p1tzJXi^g#leMIw$y8!83Ntn3`9Mb
z&Ydnvug(7WaS#U-o)8R|NYZ<Zk^&L~_u#i5yuCAUqv3cVrh38^#cp|an6OtOYY#MM
zoD998qXV2|xVW+GHm0<<eH^T9u2eL!V8lq0c#6^=tW)e~kVJ(>)7ZVlO1pOQn3$+&
z`M}mv9!AGz{q5WVuVPo0mH;HTco}f0fkY*OSyK(&amyW`q`Z)uo&7;AMFe><n45_s
z5lXSxDYL&2W)iJjG1Gt5Ts-{f*k~B7x@enAoV!FSb=fsNIk{H6Z!pWaPZBKxN=h8?
zNQwC6)jv3Q4j(?un+L%YOVfs*zdsS!^-KQkks-n!d+Jy#;`D>L^@vftaB1b2ppS+a
zavM3>2@{Gx?<pwo!l;R7xt9ZVR$ta5&*{v1NMTP<c|s2fHOL7beRqM+LHhIFvMFsU
zxI+tHdSZ(b(7!e3D5zvqvTRqT9>Jo=8h_{{Ev+yNd3a=Q)X1aG{&7fHM@I+H_J`o$
z#MM9Wgy6U4O~vQWojV25H0n>j?JyAsTL@0jmYX-=C)a)%xnuV_q!V=+4)#>15K57i
zk7y3sU@4D-v{b3j6lD*(Pa-jEn*x+ExMM3!`1#d4i?em_91%y^7;;I*tOy4M`d^>l
zL?C76q9S=v=y_K4w>X#e^!15BjnEffbIZxGK!@^Gm<zQa7Bu0%h>vr`aTW5k)CLpT
zpFiRiB+nrZO}q$J&Zz@CM8%`xx^^w1r@nAf6W{hI6;&s^MDi9ll~2lQUJcPZ{}`Qj
z&bQOHm@GSeq#xyiTBt%}Z?Bqi;#o~a+D@ZI9Q5b<2`|9S5|y5{L{LSoeODkl`Bj3l
zqOH$Z!9fuPiDtDV#0BRN)}`&F*=-!q{H|>~@7??N{49|nw`+>hl#(KZY?#PT(+Zm2
z=;k4efNIpFd-hCDhA`5$SH?5<(JU&u<zPMZX!@0=m*cGaR76Ba>0LtpRLSE@APBUk
zkDOG|&OsNE8G`Fu({;w&K_3$<bx%L7FMVZ{Q;1U(H_@Ig*AAad%;Z+kSer`XzKS#B
zIxz+W+X|eFJv;+3#yz?Pg@tSD>rlwG^z;y#F`~X=<NOu;c@)%WCS4N2Aui9)`;<(f
z*n`)gZQ<L0g_BBj_;2X^kP7UTi<zN6qdi4@asxgdczHWU4XHgY^o~D=h8Uk{WE;rg
z#_{M^glSdR&3Als^k=<q4qgM48IbNu{+awyvW`STa{Eafx444FCnl(Ws3)L~kve^Z
z$iU{0N1c_2n5dL5D4`L$ayuaZ)JoTf)k9i3C<ln@EoO#@S^S&EGKPBsl1+S#j}rq#
zAtZA<T~*hm$60wC@7$_q3B~P9f06Q1Dz-G_Uvjqh;haHEdR0vg^-2fyyCdKnV`gxE
z8vQ6P$S%<x*o!A9K%iY0n{lXNq7a($zsaXi*A2nVMu;FnLZz<uXwS2<QS;wRP#{zg
zwtwUIM%;Rccb@6wk2hQQB+ZyH#SxVlwkHx({qng41gv4<8P>CMCOdH8gNf(6cWu5S
z?2!5`Q6?Eb%|zs~^26@Q+jUBYU#awkadP9@&pw59h@${iK~iVxcFNhS64rG@IAo?9
z=R{*rbXjgP!r@TSBaY)M!XJgl1k8oI^0k8Br!w>4TX$av%}$Jd?9m&93emY{#JowJ
zcg70MC)IxdzXJGoXlwqR)ke?^*md-4NWrP=RU6mnL7PK}K`_O=tq9j2JcWsbH7-ly
z<gL(?DSxxsy!T=ZkWR&;L$#@Uk;PG{qn_!cbv+qMODN>97jyHUqi!r@TmKm~^#{Yz
z#f3fFsNB7pZp2FcI{rGba}r^e^1b00=T+pOp%MDH=&|%3&8N~h*ro2{G*Dv|<0WK9
zPFwcElp9!ZIEj>p|CKvgI(wLYB6Ch4`n_3o=@E3SDgs0Q?D$9&5|8gDL8t<*(K9nf
z8Sf3!mwMQno<}3#0T%PenJkqj6;-lS3LOQnTF;ihZejY$H+25Og_dlY4=C=B9m|EK
z^Z7$G13BF7$E7SY5^u*Kzq@<kjArKn^BHM>*@WmS6jE$OPt>}#)%j^d>I2Dc2?|ws
z<Upm{bBz_6n1Gb~4EV@9;aa-A(4n_d(J(21v8EsH)^%K5{qNDcJ#U}>XNPqR$e2f1
z6Pc_OggJvp0>dD*>t)5o<BXf#hB~j2<Jjgmyk(4IJM)c(M~qU^VxO^3`MXqsH$&#E
zTpf$uU5md@h_t6f(GNkS7|w4Job7-oj-$VIp&Srj0QU)Hk)U;f?yR5RY5rT+42?vM
z6@mFB=o|rV%i1d6zdwkF{AL>g$F=&O+7Nnx(?Y|@DAB8I0GKUWHMT$L#|$%_$Th|3
zZcBEz*qLoCah^Med%|C7XAbeWH~%<YvOis@zXN*yUp607Z`?0jVw1A;i<kRLD*x&B
zD(8n5=Dug3kV8cr3bP%cVk5TiR?VahBcTL@=nCz|3OOyVF@T<~MMkE27NJvuH(Uuo
z>p8I%&)mdhhQ~!pw+QVNk_a*>iSd!B4<b9Y67I3?j-nzusTraLSe8G3AD0GnlAx7`
zzX0VrRZ6PqMBh{IS^XpF{MxGNnFxve7kJi$?9g!ChgC}Vdo?tLM6$Wb1?W65#v7Pd
zj9eT}kborX(C`rrnR9i@3j(A&L{Ad*z{eAXu|9@+{YGCuzL?g>!RF_DPT`y)?lJdC
z=M3LSUUOFYr>Lt?l1Qaj&g)|ah8o9By(auX_s4c<q*1q82v2%k`f<_d=><|V`k`?i
z3|Q(p`K%$}LO>W-T=l?!cJ3xRUI0tc)1;`9f}O3>bm>i^e_rOcpshoJ2&ppq`<pWu
zrhVc)fq}0nU^a2L0(XjyS)pm@RN{RBNY$fl=ZCBJ8$Qf%?SzRu@WTfdyKa=V$R$`5
z>cZGav|i53>33)VNkR0^>S&D#kucq7_hV3i^0f@yX~GZ}Ev2>v+~H_xba)*ii6&=S
zeYI0ryKdaamwV|=9)#e4gq)XUm!eZ9Y@|fHy@>=*v_SPg_^^<mI5M=uYI>}w4TR|^
zfiV44Kue2ChHy45{UGX)?WFSsj_93ZJW%Bj#$k`iMx6bhJkT)}>Rf+Gpd*<xP;v<3
z7KV77og=4a7mD$xC5uW3WBX;eH{$<j0Uoo~&$B^?U&Z6zVYG<KOZQJODi$I<hHt+q
zb{h<%CDThzP}<@alIrr<_qk6ZUy#W5&W!PeI!13L!=z(>9z=CGwm;J<B$(T0BUmN%
zmg-*I8B>=9h+#?CTyE;gDJfulH!<9!?g<Q&6ua5}{<CAYjoAsGMj~M-_q6<<Qx&RH
z08YQo3Zv*eyY&YUgzhDEgl=GM;z~W8oo(Fx8-h-vB}()i7Jf@BWc>^mH{NR<f@V%3
zkA!ifmPqM9-KwYj9CV4qjT=JtE3&>`Tgd2vrV5k4aI19OU_5{0-(LWxh&B_l^xI3b
zC@9{a<oVLveAl)cflCnioS>otSB5IK1NDBlmYsCBBjMQTdhxCTPonOPvvYIARqftz
zD9s@1=@AI~hJzQww4>WLE3l&1(WamPxMcv`rKiv$>2ao@u(?Avyba%Q#bS&S&f7>X
z*n^Ss9lvReUR;^~?=;XvJS&`1NDRcR#LD(|9t@mOo-ab949kSyV9!50{yp{|{z$}R
z!PFj2XNo`w&@FG(37@o55^SN_xykZz%`ln)Qih{O4+F(cocM;dcG!F#Zw_4Bhg`st
z#PpaXE;vx$)(AUp2=MYE0xPF*34=@sChV|z!bTvLpBLkG-6Zm(=ef9^<3uWqGK5$S
zfSXpv6+y&+;DeExou+zQ;Q6c_8Z<Q|yi99CXXZ>I<Fvhr+_8w~(z+(}|GNATlPi2n
zzWZ~|j(b&F&rR;vyevtHHx|?Au0PkMn31e`boc(k`<1yb?~zo$OD5wd`LwLH;8?rR
z%+u2N>7dWQD`k%M48B3WjC<>Zmjx$89`xDw0mFno4<8R{F9q5Vn3RyxE{Cm&40-h6
zkpF)P{x<dpAARz7JOf-F_==~#YQMsuCc+5=0Wa@|?&)>Bg$3=(*A3sluVa%j34txz
z@s)v|U=p9gKPlzU9%%yZ6R<0{v5piv4^=5*Sbee5yZcl$G{uiv=pd;-0RbsLAKw_?
zKPe|ykoykfXW)f*UPumy^OT4~Zl#bx_ag9ICo7`r^e?M?>x85vw~QRXeT1JQ*=iAR
ztP^^1$h-I;j>DAKYy4Y4Nk&I?B`-LA9(o3X;~)~FjEv5y?FVf3fc7aFUMQ33CP5&s
zB_t<@3Or{r7GHq87~;mg?YA(?&mpPb2}mF#b2@Oe0<0H6!`|u!!$9M#J68|CtsFv_
z_bcTG0$b>jMPQXdhD0mf+B#FKm9@MjTw09Brv{%1U_IWwbLVL39H3>Q80nBta>2L2
zz7TQR(s=z4cf_CZ@vdp#X1t<%uj~nOR+t!sfHzb;j|X6g;%q|E<lSGCufPfCf<Jw!
zf!h)i?n%3&%9@(}fFVQxey*%oGg6+taDh;08c?2gz(KH>5-p>sr6q#f8vBMwbFsOl
zr#FVhjBV!|8p=LD;;7BCN%1rj*!34rdMC_r+$sWMT2xRF22!tdhF-`TpS5g9R+0-I
zCO~wQ`br84WHz?|6k|I)Y7CUq(&B721FyFd`QHc6prWOP`BVeIq0y-#y?<r;Bt898
zJ-r;%$rTmX#!J8#VfUfn3^7P}2bn1~bqH=%SD{l=(9%H)Mbr%k4!}h-o}&&sJBS%f
z8n8l%*-95L$2%G{bQ#K0SJyf`1q3N3Og)r+E)a_5-UMDKd?WBYun;`1Wp=b;S)+Jb
zkL8WE3q=!_*6HyQXgM*o4N%k(lMk#em2hFS;WRT;Jj)uHm}mqt2T7u!HE{G4I~OY|
zEB|k@0~P=kV<oX{HX<GIA#I2PaY5JlrY9>gF%iyC;1XDV*x-D8iYpRl+1WMFr9(cA
zhd>-FIXOu5LWUirjPX3xkcnb>R|thJ8bv7FyO#-Mgt#<t-QtC{LnQk3tCwGlRl2vE
zU^?2zf2`)iP<>!+=H^<W>-pg42WSeaQ@j9t&y{Tt6glV5o_+9S!jFRZ|ILRsim|wE
zi=v{KOdg7f{j;5g{1OE5v!!)%ys<xo#v9?(Kp~!Ea~m5^;XE`lHa0diM9>1L8~h<b
zw;;qIFgAa3T_XeCm@GapLu#9ndH~Z89C85<RTq|lG|J88H(`!=q<;9%EZ{^Qx2BA(
z7mA9C6~zI7zN-pDX7k+yWji}N;WU`W<pLT2#TB+F4aI4M`mwUU@h?Y&0I!+j`R$%z
z8j1<58&D;9lrx=40g&7xDALUJaBsxf`69#=8siGbfenH_HzT7JVG~_lO7ik}JH(hb
zNl6SyRxR06wRd};DaNMwFYY#S4AQ`7W=Zn{E!L^^Y5e+szDjw(&yDz+RoX)p@xNcc
z#6-~1{?C`Ftrx+E|Ie2f-6$%=|NZ(zy86GmCH%qv{U`tTiT%&I`2TUicpqQA1@k0e
zsdsY<_-UeTXs1vWbB<dI3zwkIlvtY})~a+LzzJAiYWh;!RAV{=mD;Y-|H~FXsjlen
zq*Ow2NSGe1=A#D>80UxZw;3EjfhB*(cAv{3KEyl0!NIuf{q(37kH*+@?~s<)6xVd1
z{QQ462p-OVU;ke=yZ6)SBme$^J-8mIctIIT$NGj54i-f~<DgJ!C|vrY(2(lucfE}x
z*|E{t(E)^3Ht)o#Q-JOZGMY(6*I-9L{P^CNvdYSNxw%iK+gFD$D-}a*gA1#{4QOkZ
z;jir=xi-6FmRXd0_v8fPr$BrUZ4tL{l(^d$;RyOBfTjSEWcp1PClyCWL_gvJ!X*g~
z0dOQ(b$o$eW$XO-x}&pmY<wKG0w@m<PS0c%>n9j4Tp)Z$>FF0u-NM6R)rpFYjs4K_
z=`3Pk8*rP~*Plb!R%fDLYAR=`cCarxEL>x53~x~$Vj6s6hRlS^$nW2dLlwK{V>2_k
zd-C>`OTb(Ku3KL}7B?a;WH_2bLp6MjWUnPAC4mixh5|==PIflociBvPOhVaoPv%wl
zi}sy{)Vs7uHypQ@Zwk_Tb7X2}$No9h`(^A>z`&jSo*Sn&w0M8nW45Nfr6vEe6$}G>
zd^ul)>Hqx#C2PLprtPFZm<MndS-<aWE=MA3gd3&j;z4W!^z+#sCz=`Lz#kCRlW=``
z`9}zGcUqn2<P<@y3@R`%J9vF{Gf;@Y{|K#k5S&@)j%Gd2FYRFkGeJCkNH>6x;}Y#C
zI&k0s=3`s|#|z1<tJih*vu9mBJoHMEU|B9sJ&z)w4K~wx;yHk*#`Bt5|Jei<?>8!M
zVw}sdV@g_DcT1#XWS%0Dl(-oYB+y=bf7c2pa%EQ6*X8q&g;9=j!D6De_SWR1D|M)`
zBqUtAuS3FssyIA11Kb?o<izdOm$EFUgfMHOy1Sb&gd=+T(-UPmxsz}`0N{&7Y!l3w
z8zb*Z#P&+wx|I)47sl{nmbs=TZGk3K2^}5guqyyG{N0@vLB;`rgyYKWivrE-*9lpA
zw;b@U`^JyYf9cYSD@%<}IQa0@4xQFd3(qk~uh4e!-10JLM2w2Ug1iYH#oNoPe6t_-
zx43xB^A^%mtvJ;1smBx00pdketYWv;WlvI5pC*#s&u7tr)Zl+MDB64Tz$34CR1=|i
z@}&|`nJg0nHkJjQwu)hyv9-OtkcwIxOcCA;#sPAu=dx=SDj?VK8}*0RP`e<si6FkA
zoUw3$i~@LwcBy?&dHH>GaQID{@;YFoPntbG@7&x_x2XPs*H#<BOK1q|I8fhVMeuQP
zC3Ht$I*lM&c&2BUmN?njb^93+h6V){noxYFDZCpA<We@7xxIfszV#zU>NqA2B?xEd
zGg$9v1F{vb$BD$3I#-H2sEfy%*<*tc-w0!dH7&LeNAIA+O9q&~O3?<aMO+{gzxMgM
zxOg?_9~8Qpa-V$)&R|o*T7YT@K<?1U2oEffFFel>s}RFp@E@?LI(TVZD~DK6(Xz48
z!S9IW1LTc{|8d}~M}WUSs3QR`E^nIvKR>FpyCTCblyn=Xvs~1pPMVv_nu=wfpe8xE
zKY8M72e^wzgWEtFjHlaQ-1Nn)Vwis>k71q-VTI2N`|#nY+Lf>DHrCcePsC?+lh|uI
z)$gG1LZx=QKumIFX(l8rtRJh+hIJ$!v!Y=!n(TBG72T1h8!8IY@BE^^qGBh#(*~oO
ze3K})12T}~;!g9Oj7vzsM58B9pANu}GWoR{t+5qo8PIPJOYZo*c6UE89};k-VZP|q
z;6M2&@<1%dtHXEt4M6i+E_{Sgweie_9p@EC`3%_&YWw5@o<x|0pc})$N|k!@+qZD&
zqV&eV0glGJMlb*<j<wR3;E_}A(hc>hx%YGniZsy$v}fhZ4}hD%P}jkM*6x^zFge!P
z%a=R3h@Sx3G_Bj*83s_euMGLm?4=j6N9Zp?U_D;<%HU@gTW8d|AVjD=^#x31zP?UX
zsl9+3d%k}M&l+kcfAgk3&`?|qXk#%@{3)nOG{^G`3sAt^4m)NJAr*8Be7D1o53c6`
zNA5iaK~8J18dPqtWJrTkafRbu-AnSDs~LI*hTayV2om_tFcOhXm(2}ewRzo3xo@9Q
zE&ZFpe?Hn=7oixuBCDo$dy0rHA&evjDafY)pyoDG{P8R}rXLv(NzICii|z4v2zlCN
zvl+Bq*fRciU-`(YstP`+AkjLM=<nYJAb%92WAyde)ZPSR1LCuv4OTsS00c$v_G4-3
zCQ1R79CTR2Rj!bpjsM`icoDM-h0x03lTMy?dPu}NLq7_p3CqDPblo1k7?>id?a+Tp
z<q->%9h{D15;laWoMUj-A3n_Ve&J=T0X_{7g!)gqzkW@JZ53HJi_F2<NZXiS&}bvS
zU4iEZEr$*H0MEsX={UBtvWy2xh11q9ZeV61Pn>D@{+BNLvH6JtY^=^p3xe9Ve{2kl
zg3SR93?ibICS~La9y$gHF<l~LY`6ybKJE4j#T|n5)VTo{Wnp~^7Xn%xf9-|@DOjbS
zYtMmp0`j-My4t#=iX$HOEkmlWRX;wywzjn`#l>AfL<*r#h0m64WD5-{BF6#iFC@IF
z!r);{dz{@9{wXBnJGOg*I&i%lB$#1$?8)7ro%{ROer|a8vK1;n1k#99SL3WibNeYF
zLC|5%{K-GbFhEhB<bJI-aB7D&Ce7?HajO-!(Lnb`jBPM*-RqU@OhSoUyD{<uiP#4G
z$Jh~ltc$pYF)-tfx>qFXg8u%^ap|?4Z*`u^9tJjm2WXfjL-tZuNl6P#9IUs!c~51Y
zG(s+c^9e^DusdembsCDArX~)s^Dib=EB8=xo&C!?ev3By%Fp;#<-oJaXYSPA9kjT0
zM743~`ft*1sZ<Y#V_#qMPS|VUzQY+iH97frago}`3a|-Q4ZMIr#b5$9L}hV8CbJcF
z2_|M?jU*)|f`Nz7IiFFQZ+5SYq$Em^#MD%x1c{7%a9yc`syj&uQ)Hof3kZnAAKkzI
zRV8f;1+)PW1&vc}_ib!eaXyf6k}<qQDFi_sFc}}8!>|aVU&q$$;*@PgmyJ9mg0VqY
zj*lgZ<CGWrW5^sLBGS<GW8J2u4SZ613W4KC{KlY0#Cn!NL?k@7wYK(*fPl*N>#yvx
zuwrl$qol~z5hY#^r)DrDgEw#&rD>OOufIc?fS(5-MeI9pi~rF|lK$z<k*Yn*0+JzG
zOZgd(dn1|FM_ee-cgRMxfGh>{tK3b`$_>3$Z0S9ScaG}ETvb-?#ySE*d5}a-S$Se;
z=(EP2!o5a6K3_-J3r;>&)gaVuFJ6dYEr=V(p;!Zaf{-lSD|nQ6or@-sv9aS=h#*;T
zQ36EfN5Ge??7<n`{1lv?*f4nptrg$Csf-G+pm~QD5`kL)&9FV%e))fdZ<CUOBG%;b
z{aL~&k86||70u|8ehwna`g+uJ_~eXWZy;WPqkM*u^{lMCd?oo)+_dW129eRx?MIIt
zi8H!2`wYlL@4jvlF*t+AA%$Rhij(a}{`1PdzT*7+z{2l%Z6fw7qUuu0)7aw>>;x(v
zvM@JC0o4r-zSj~~T#&Btk_tkLfnQZ*69^$tPv1+xmPn2b=`CV~+RrQfs4O{c5pv`E
z%H@&8>GB??tDy%N+i&Oil{$;K4B9uZYba@cF>X;xOvzTv;XBGVDPX~>$ohurO-~H|
zZfdQ<Q3ZEi58k7E^Ysh&_RAX!E}s9%c)j+g@lWHk#tCN=6}uGtmHfv2$BrrPOD$s;
z5*HNTMZU{~Y9=%@)ZNE@Qq8}*sQQ|!RzZ%MF^3+9xM20m`Iq=_!}U6!UHHfc97H%-
zS;qjZ^bH^r7)5GBeLY4w*+P+EG}MStF34FQfs?lPKk<;?&)XY?00hmy&=;6q0<MFo
zB}}LDlzBIVQ=zZ<Ai55~JR%bjVPVpa#n?fz%w;%j&;*z|@V;b#D&wBKydeids-}MZ
znzJEOg3=SbHTL0cNy*431mlFp1AscDf2_LcO!G&Y74O|UeDtU}UIV5|0RuHH0@{Ym
z7!+m4In-FflszVa_@ZtiZjkjQ4hw*b1#WH)(7N5o*FUJLqr=3b0eQ)tT_MO?LlV1R
zL4nP#??d&CjY0sZpr`@Ce4BbHEg|98;NV5m(OR503uSC@n~*OGmxG=!54%sqtOGU#
zBJ8PgIu&R?D)XLfLwUS*lJEz3*P^1LxQK7uATbG2!(9Z858}5l1tv$5b^HbbaSg}6
zpyKu|;NetN?LifWL)R~N0_=&wFP@A`XZ9k!P}iRdTp2UZNr{-~=mkVc?tNUGoo$Zu
z31HPIfm#wdFt6=t#15kRiXOWJXm?=xx=|aLoVrJCJw2)`tnv^gqNr=!)Iom9FMO;C
zF=P37<iOR-H^bH~%9C;?Cyh)CF3Zcy8<F0_S%uQ)=FL3?W6(o3u1W0KvqzV0&Sf7R
z3KszOtys|Q@4SKo1KnZQfw~az?6J%Q2=?GDSy@~2NwTR&biAr6ctfN54`+8$%1B8O
z<BR|{Zf>laP3o0lUqwfc;6x7%3tK%fGF<wVWGsR3fa6u;mEOrk%<Dy<+uBZ>7AXF&
zZOP$nUB#z>XPsbdP+M1r;sqL`hxqAHiVjpwcz7t+6V!<W{X2U?%ZLCqWY}lBZr_2Y
z8(Wevxoc?9m@wped&>c%JPr53^mOA@YhQy|FU+<^fq=J#L^4#UZOzZ{Ls3$1{g9aM
z>g<Gg<9$JYW#y~fY8-W)or{o5WY2(wkEIwqZ1?JL1&!$KgH^bDgX#o`!2<Yt;ZcT3
zI_O77rNK&})^?>S*@o&u<?oaBkfj(|Ir+KEn|xviZ6|Gc?3&Fp%v%-LWqdZJwgNuK
z?UA?QZ|9=^`aqQ4S7iM1IEUiTfERs{Hp^H0C>@W7D&O1|h`Oh&koPsZ%qjDBrh&Xt
zlX+wIhhLSN+V*yzr47j^1q^PsoOJlU=*PIVaGXl<*yh4e#4b6~d;8?cevImlJ?{N>
zz~JgftB@bDDbk8mKd-a5yR{GKmzD%QTi9}0_v7<R_8_b5`qq^a@ZIbE^-<kzvj>Vw
znoBJ;PVcK}>;g9Lj_Mm;Hn0lZXxIJDpR2i8SYNl%LZxx<?mcVM`~j)5wB6d(i_d&2
zhCH0S-(q#l3o5b48%untuFm*<or4Py!S>ds9afvxL?6ynRMT3N62|9*ghFMW=0aeG
zaxN_v8CDJbDbSp3Y~P%GeYSVtjBZ+TL<Ie|%YEc3dX|e4;V!tVj;yfrjqXiLP6qU?
z37C2%!qwaR&{R-*xXg7eK6GW5FI>2=WmrW-0v1_%DDOuTOup%~lm6}rv?ZUyM8Q1m
zrekevOTxz#LkkMAs?THld*GF<e+y4AT|f>G_`gT!M$=*;uWS}NhZ+dCK}$q*RFo;2
zrmYs|w6rv|-sL>Je0(KN+ok0bO2vUwbgx}p;KRVcbBUWbOReXI(dOYIui#ZwS8oWO
zBkC!$3DodiZk@@=>~(x+_F_jjZsY^&z@4em8-aMNoG}#M&jo0@7|W5rvQvE>u~P;T
zXV0D6XZG;~<Uav6^CV&?PM(zf``p;bDBdiAXf4tz!8S?KU4fN)oC0kpvsG<N%S5|^
z<&x8S1~|7{B~$n|`VOWBP@zcC&A%d{L=jw}vF<^8#eb2KB|_zQ#Id7CZ#4Xpmqry>
zSQy+LgvJH8s^)#2hSpZ^78lAHKpplgy!l9)2@XEB)gtrY!4PyID>9?JJUsZ!Xw!~W
zq0hqrio5rbP`%Uhus*{CC?ErgjQ4yPydoGuisC>&oG}SZHgq7qGV+J_(P0Y<3kkgr
zIt+^cBRZ75_D`y%-2Y2&3Ud13?|;m={0oh_<iXdVQ&7v0WSZ{m#34}CW^H3r5EH|i
zGln`O{mVCm237EWJNnXP?-90(EfEZ4IT0O~r++hyMMyp6<Z791osh7nNJx&{zGC0F
z`R##~i>k6dmEzIX-?y7(*eLs~ogQAZeCgG4Pu!xqS>lTHa^U=BkGY8&9?yBFF`4f(
z_~(uNk3GMs{mwgJLtin~+K|*Rp%GTOmf!DCNbf4*qsa95PPkks*LJVHNXtZOO;Ml(
zM^Q$jdD^elfC(Eub$RuytENsRr*!*!e!uPSuPU;j2|w;PTDe*_pZD!4V~$nq&^`aD
zrU#YIbAomHIpHX!>g&$zX8k`}0E}<Qr?*5qg5xkRC#SP@^-teCU=Em}ANXh9y>$yw
zsw{bb6y)UzTmVm_qp=lr3DVRXr{5c|;gRQ}tx8Nh_gh*}umHbBZUw%G)E^p}qgb>@
zk5-i>>=79PT-NbML{UK@U(eZYaZ;R|6qB5w$r4Sis*$MF#<ekt+X`cXUQE*k1O%Y}
zv_c|Eb~Z~~InY9xz>6;gT|O8?b+l`H0F2`9@LbWxYgA0jRVeT8J;f|8Bni1%tC5Hu
ztio}MoKB*}S350~glZEqt;>RfkT*VLH%F}v-wblN>Wy7cBclH=v#a^hb5Bu`4|gsE
zgea7y2OpwA_Bg@yGN()-7(;iWVq)|jJZQx?CKQoKUc=+SqC(@5?d`ye8lzE69p5aO
zwg4G19TjI-j9#AukTHbzuTifR%%qHdC)9`-S_VE5;@LBe)PBgkHSRd{5>t>GpKRD>
z<mJUl$D+JE+(0VczC1rBWh&+lR*M3jQ7sqw3nKKwwk%(c9EsMa5o!$PYZw@aC{Boq
zs@QZ#OBj{9d7OrZ{h7xI;)3qtQC+z%EhCc#qy=85sMtdoG2#f+LCXc2h|sG};WP#b
z@cS4V_TUX6sM57J;Mk!<LqC6>Y3%ku*gGUI`Q(#RQ{Pcmsj1z&5(G2>9PFLuZO>Gl
z+fq{S14C%RRN#sIYDFG|58bB-$-e>u0?535mvd-t4{-`KIvshwn-WB6QunezJ1#!N
z7%`4c9t3SiBW%lS5$}ER)CnP+2KaYKcuh)np9tk^!$)iK`0=StGs)4x7kpoS|0ua)
z>{x)M*XV&3q~r8$l*>D(D$+(ZreX5Kzs7}(7xPlxBTmc1&Kjxa-`weh3mK=fj$E~z
zWsuALl3NqzX?&)2-6j4{XU=A~r6I?3tq!XVtEVH~6^XAk_ujrqd^uU*I&ssS?z1te
zp3NBZ!}sP!)!f}hnhln2CdYZ)e$Pe4rCRl^u2elaGG*b+{krjow_V&~;1^?3LDhmW
ziUucY25Z}W^!Qnq`kzqn(B7n=99?)OUnEl^w3u1y?0B6{E-F}21YSer*_1#Gg?pl*
zsmTFYHljE6^~V9s)*YrKs(pZf<*|gwO+zCY>56~;pwit0zYALqd=@}<e4bCB9Ccb9
zJ6^XbNI-!DWmhfq#qd$D^K=-ON~)@!EYK(ce&Oi}z@{1KLAcO=Gi>mz(AFb176!5X
zJjOEYGS<K~Ay0;1M;*Q^v>rel0kXqACgil$*V7YJBx7j!3k7-|4bug3r%oXuAvjg4
zHVKj%`G056oLPZ|1U(pnO|beA;m|_MsQ{E8!ql#{xnV_ld3ZadzZ<}@rDlxR9&5rl
zK*<D?0y=pTliO*cSfR)>-)CmYcmV>g!zyvO2m!lMC0oHDp&jUCa}RwGd^nR9Z@-~^
zYh@}!7lu~}&=7mW68iFv0$+vXL+LMG{F6494vu?l;4NTMlCDrw)CEvglp)3$@Gkx~
znuTXYG$6Wh=pe{7VbbU`A3N@m-20z*&c}jGfM2_uv1B75KK|lq*Gu*>Z~yFEY{|~c
zOM5D54d~GHhN6+tB#@&j>nB74B19VlU0wctZMf3MFI`BJbZX+WehWoBw#M=@HU$#b
zz(Eu{tn)+Z1I`#nim>n|Nc@foj{vmmpftKWI$&%V#jk+Z1HSM3vq`9KGHM{Jt5Ypc
z#?d-9GUAJRI3$D`hv}h1bLdMUMIu65@$8CZ(0y2_uqL6SLD=m<k_X^D4ju9i52r)7
z1u)9bk8t8~e@23!jQUupf~Mxo@bF=j5+#^Krv2Og2Ve`-6oA1Q!DQQ4!+Qw@{*5Gm
z*1af`AI0)iQ4z!Cus@}$c$nYfynqJs;=0@lAPyoeyqn|;z!-GkxGj^r1C#-Tfd979
zTmE#t`131Q*S%BC=&c)#U*jk-ic%#uZyE+^SGsU=ast?%29Eve6){f1)m0KrH3H7!
zs|Q!|Vf2E)q&8%2Z4Ej&qrsl7+Rv{><IgA1pRf=WyZhHCx>!J6pvdz0_T%CHyG3@V
zGDp)AK8*S~(${8PPb$qN@9!$Gcr>&eVZvvZ-DB{gs3|9W-RyW|Ec3;nUx8uIL|Sj>
z(DB(6O<D?%ncfx`6Frpjr7d~>IU0uS%uK{7CuC*u^tdoEGxOlo!hZ9OSS)5bmk4hl
zu1XP%I=})N^^3!&h_Deb4Dd1JtHXR&OLLW-6Z8r6H=zDU{Q)7o<~o4!BrE*es}uC5
z))f%4AOyj%+aKQudYT1%J}6=gA3m%s%fU4Xqyx>vHiQPi&Jee0ATN$ZU;#iLYu@xJ
zVjRn$Qr>8Zdy&?Q){@wBoBge=b1-UP2SqJ?1Eqy`fvp;9CymBw6_elQuf{mLd3ruN
zRy9LtR#K1=@bzX0?gE4aYG!w#bZfZ06wdG#0L}OB-?Q`|k-dc79f2vw=H+U+WpI{&
zDk|<R1^|HtU_cFZX1&_y*9>p5tB}YEH(9z;jB5J*e{$$_+T~PL3veu<`^q|YVSOV=
zufW{F0S1L+E4g+W0**6~%6h(g&mK$oUSO7q>bwio2rX4BW+MTwY-}`_N0o}}KQog`
z>>!krR8rm_iAyT?jVeQ?YSj4e)#DNmSuu))_NLrG_zp(Qj&OPN#=GG3ONOpw)yj==
z%`AQLsUd7e2GWxWiHR^t64j{x2ju_2cVUSW7l_dI9eOLqx9GuPP764tf9Cx8G_;uL
z0s2BvI~VVX3<(T``DO$F1gzNE@BRQnKzamnR2jD~!!4y2rV$K@Ft#E<Q!H_~I$Ziv
z6A}ubLq#*;y=x2~feI-<4hp}}Fof*S0~f|@cG>HM)DpuQ-xf1K6c*K-;_n4Ks`nUv
z>Vx{249uX2OIe2bPQ)lkd<Yu@N&Y;$s9!R`a+%M=0U3^}C*gf~BO}VoVvV6{|4f~B
z$KeHJ?$xVZt7Uf-H8iG&hnem?^8D+{WevP9=I$FtR~jL*_Kz$VUy%YUBW9og2}TEf
ziNla+pm2y|VD)1JMr?E=Kdd>N&-j?20zffDMDZj$-FMhb$L{-?_{(Rc+Gx_~u-Rel
zxU%L8$G<9isrm$Yg&hnzwtjB2v-@RQv67sMPTq~_Ad&OY+kX>A{l`l0IIX6<b(r*g
zctzqN$3N$JyG)Jb_sI|Qj67`D8Q~u0mf}{m^7ZvoH_B63;hBCi8?lu>TtD3XD{Z@J
zEid+c3HAL3FAk%3i&|Wgl3KsM<v&a&Rq9gM3R(Qq@Hb%h*zQI<w5<FTy($APzVG$@
zJmr;?{!{BkIzVm1O^(27);qMnVIoX-?8lbFiQ;n!X1p6$zTXsE;zz;tp8C@JmT>gf
z%=>$0+saQD14u-FHKASi1RRZSakH9ocr0UDFtqQ>0k|$nNl6#fcuGM<=`O|q!S2%d
z_rxZ)WuphVs+kck<m`?KB*C+781UV@bLYU6yZ-N^<Cu7f`V?p^4%<wn6Mb+*&hq5)
z^72B21rZSZFFw$O9z0mi{<m`I8O;6(>cmZFtm8u@&O@L4y8wEiC7@x<H@}iPqCocj
z<HCkHE=KT*HufAzKJs=YIr6jZ@u19r8?SuIQ=DYX#_FlIj*EjE{MDT{0KlZlm{7j~
z;{CVS(8Q!^&GPU+3){-q5@KSVgZqxghBN!(7Bx1OzP&&}hg(J>EUcavy7L#pH=$`m
z9icz7ACY#a2dJUddMMoY<45@RZ3diI@PYUJ&PFwc4q^T(y|gqD<JfgB9uN*ooFpdn
zaD_m-#iCsgqdZ(_p567kAn0S<Ka1q)TZ3NPdvx|t;+OG%J}UTtT%XyM4)q)GFeY9L
zl<-~ihd@{YeI0Jj##mD(ZLxC{F3gQgLM*K>T@Yo1fZw2=;E)h?G&ZPyL2`nLLSnFZ
zn0uoi7}})c>1Rb}MXL|c0r4hT+Qi^qObB+<xboHz0)-9$j~wb>4seSv08aH$QdeGo
zR29+Xk`iVgroox7LZ3?qDzSWO$D~NLU$ydZYGdi;qV7m73JF?Ak@bbqf9pmrxYJvk
z-fwv&>{PT)ZN?`9c$C|L?PiQ(%=ss&>FW(oD&FjM@Lbk(G_x_G-DJA!?3MA<xW(35
zO|G{!F!~7Rot0nZGP~{`I5Bcb-?V)6`9zH%jm0y8$!EOdXIwri>z^>XtRSiN%<tJf
zY0ai_Cw_JMT+4z%?zO(3KO}6m-#=WvVUo~hS{yT|Ho@C&cjsxeIOR2?DI+VP-Uk-n
zjQ$q1;zy(l>SVF$lAGD|&H0af&(lpvYX3<Tq@A69GTrnAQRFU_hgABidV6gVbW{9$
zta)*8%aLGl&pnhrxk?NAq<>aL)?G$`UYBbT75A(RHQ%Tr?i46Yxt5m!naxjzjDZwB
z7W9TFcxJQ!T$x@wDEL%axAC_ax`C~@ng{@(IOOBihet+sYo%U<gxjm26+TrKme;4?
z{Sg3n@ULnfs?b|2fqV=M_lY}Jyct-4O2f0SvHTi^=B_<<U{PxTw#Z8KAg-5RKp?Qd
zUkP);3>w-ld7%x>(|v&Dxte?)Zo=V-35oBur;6t#ZL)rdo7_P4ajk?+SPr5{nGk*t
zzL+g6bClk`mmpigMl_63zk!)qOqVVN<UZEUy3zD;ZgO_e(^jFK{d3p!E7FW&IbaIb
z4(<RAz0Y%0Rad)yRjvZ5ATd6T5^v>CF-l31N=Mkyq}iMUV-7$hE)z?KB4Oopacx}|
z`r^hH96l6uD$0?E!cozsnEpi1d*j5dyJ0mFU{!u&ht!><MouUa5FT;Rz_ZV|!Fv-6
z9ee4b@ow~C;kg$-V2}vV)Sx8cS+$E2Z3l%qq%<|dYWWUANr@nxy311=M94K9Pc~V&
ziW>QPZY&Xtdub{~n^J;4maW5QzmKQz=F^P9vZ(DztKx_W&6b)7`r}Gp2mG;Y1BN+G
zY|~wPGCjQiP0(jjaWq@?cuj2)x@~_J3Kg57&fVir#tZ-UG>KD*PyJ4%w$>T^J==ew
zU0bO`;j3jWdtD!GX-Lo&`495bsRNoTJbm*m$I1d6N7M5w-c?vwb@xjbHuRZA>21Di
zKSVOr<yQKhT<oCJ>#gU*J+AYU+XsqLM>;g!%EoPKd~-uW-TNLV^(;EvIBTG^DzLQ~
z({b_Ipx)9aN#ECfk-5447*`eRI$fUx@@&O?FPF%W{a@_vbt$0WsH0Y=d%ToA)D9~S
z)_=kUPC-G@H&D0t&?l6hUPoTS&skM=_C?@H)(2f8BTs=XWIvgJ*hEp`>(zi&RJh6F
zyhxJ3?Rw|?H^}XuSkRr)dPr}8EQSh=Ij=?67z+#T?#~Ql709m@i~`5&ttu9qo2z7c
zq&^VlWiY3(mbgK}`lPM`ivrsK_CAZ>javB+aK2?5@9AQ>lJ@x&iCDV7>!(y}20#Wc
zoNo1J>PaDW3KptN=XHWHfl}a!xrIf{)(qa#M3$GiL(Y@_;DQV}pR#*59Zr*FG4UK(
z9C>3=svd2;^cf~D^IhkUVdUzpc9s7Dx^wWT`ed@aB7GrMKkI~*-cd_41)|?jyr*{X
zGyK^<!X1QmE`F%}aTaqck|TsFd{Hz|1;-~OL@ue5G5@Cs_RED#+7ZosV_1y`b&Z<0
zY}@BuS%gNr>rFVg)E4`NV!P${tCwz5yd-UXrpz{$^$I!KsO^J-P8$0NLp7v&jSF=#
z2$)f6P=;E&p8j{_9=5pf>qp9R-pU^{mhoG1V@}tlS#4r0IlfzT&KT+G>KQ+7*nA-T
z_0w*<rW@YkY!c3n`(jMo*Xr->l~aEsV$NK^uxi?;H#x0F&g${LmzHCZ-K6)Mm(wk|
z-BPmsxx3fxonkvG6nej2OD!2<r=`-rtUviq$fb$f!l<+EmT$P<R_OL>pOpV0y1X9)
zw>dsJ9bGo*R$g)OJ@rpgwUhRWjQ2;a*ZZdrA5yvOQCNwx0LB8gs=fTB*r_0*%v`9F
z)$=&n*z{ome6@#s_^RQD<5+|mpasz2BN{)}(ObSU82l{Ee)E@$8PiUwO5TQ~orlL)
z<L$eR+}O=)KR%duR8OxAx6hnqad6l~fRSwF06r#^_w$KwtNzZ7NQ|p#;!r5w;|w~{
zx##+s^sn5<r~D2ZU%fhG@<lB{Sz6k|LY$+E;tM8dn#OR{;l_^u{SI4mkQMwfo=_X1
z;=>Enp@?0<7#e*13C!z(BM0}=J>WWSoGzs)Aw%?LJi8lFS>qbjf-K|7M|8x02-(}t
zlH-J11z|SL&F}j%G1zEQ;R4#xUaz)2I-AG04i)AnjNKNe4iu&3Z1s9j;nS}1$^|Y6
z)2SpSWo4qagyjx)=A$OhcOZ{sD~*=J`rx<SK(ApH?M4j^=Gv&hc3$cn);%_L_c_JB
zV7(3c3e2B^BDAjt{Ax6t>wnADi{@>OJ5B8XA1`lyrBP|k-~;py5Z`soIioQGC<iP%
z&x<47@el5S5ap)NFk~WWhOC{;*Why{G>kzBjcBb-v$AF!Yt*B?L<=g)Vp8G3-Au-V
zLcAntwJ#XnB8+D@35gfq8ALnbwriyEvq3-`h#;@yvXF*DF4yrs3LuE8n%cX$P^WK$
zVo@r-5_k5Q?qj|qabQxQwzsxdv|V&CGh9QhKI+cy!I34qwPf~}3@TD(bM<C93WkEc
z_J7MMc5zs6<S4%Jh*mqZzvzSN&p6p}cjcE1BI6DlYO{X+-M)U_IbXBhN=7=>F0?hI
zA2)lbbC)U2-ayNS&cxLC`!N%XPV=nWt0|W_9GcG2*QnLH-;fsEJ8e3Yv*)0z+VR{v
zrHlvn-~d1}JQhy=B@9)<7nT+V69Z)<7(9S}8@x;zLPJ0&5W-AE9H5LN(Tbd-FZzMA
zb6PtQ*LHUr{kCO1i}0g&s=M4&H*b)@<pZBn^wM5PZ2%N#*H$LPaNPn&J3;k3OJ5G+
zexlkMf0<w=CE1wCO(OPHpP!4znM8~%`Q5E*wUL|CwI_Ru<CBsi1kIfJ5;2G%#<?ST
zyV!nfU4>Nq9`zwR6tPL0aAw_e5hg^2Q2c=vYs|cj7MyM2Ie;T!(@`>;0B`SBEVL@S
zisbDI=`50bYq_k8QBB;j@BNK}SM$eD38Ailzbc@JNlaY)4dnj#!d-#=i7+n|>&6cs
zGV9jRZP&gyc-i{a-MdSRix_ohYj0m%{t;@zx7{Q2suzQ+rW8k>35@1Fz<7&+(#4<F
z+jn2N0MfjtT*5%6gXTPY&DcI2eEgcK>K`4G*OJ+TFLLIv7&Br^ti@)({(VJ?=Zmi{
z`OA3O76l>bwMlpH4hjx#+ls^L#!T+oLjeHiV*SgWVXGrL`j6l?(ROt_pVVppDK&7B
z;Nlhrn;N}Uce<u){o^#JI@jpYc+HT+b12^J@$pTaWw*qg4*n&giSH`t<{1MX8CFvk
z&<Bo8Pv+b)GFJ3wjH;A&p}a)PY0q3ZSz9A{IIUUcP`c%`tlSlo(&)n%NyG?R2HbLg
zzc2@fM^Q|6_FtTGef49wIYAC@BFAAxp`4L#V{aK&2|S^%52TR2{Md9P(y%s(%~V$-
z%`lwlzCYWc24-OLm?jfmDDoJJt=Y4JU2$$Qqn|*lz}6Oe$}1#9isNhZI}5pUFZ_xP
zqS?hO3uQ&4(I5lw*FE-|b=$2#h13U8*b7Gfi=0Nx|G@@sH+=j0e)6$FbO;opnVs~m
zgN`Bd1@x$uYFj>hZz(A$Ps_LAiZAAzS3Rx9z1<!{Z}ck$E5-HePQHM1z3&eCbh*EB
z0cnli9>FZHHkg|0>j`0})h|T|X8@N$3etMpor_P%6t}-R)*VK(&bIfU6|){2At}`*
z`J<eYo-RAO&uYMM7QXt6kS;<qW{TKY6$C$cxpZH$zQ1XGoZFj2ZI;2G-6o5YPG!3s
z5ZO-XWuS7pctI2bWx!kp)Y5nE?4h8k4Xxc+no;Mbmk9HXMEb}0_?J)*F}vOZEQ&f|
z95$=&{F@I@7S?ez=t=9#dGOH}JsPD`4PuiUIiKb}#{8jsXDqMNfz5|TR)=b4_6gE?
zF_4WH9@{QSdN;-?9b<dol}RMi^hIHjKpq=v&8+$wSMT!kHj@-O?Booz$fbs^2Bir!
zCBR?~S%hA*uvo-42C*ruQ}q_`)h+WKgggZ;Kth8M?vg2d;m|mL-Zvl<#fh#r?N$!m
zsrf{2Wio8}ou_H!B%P)yjd!aaVpq*1f;BpPT2MmZ$5hZXTyBnNS<u$hY+CEx%j4j@
zxGgov^xR&bi7ZW$D&D9&m6qpI?3?st`P$WvV-K3^8Oer`FyW)|%I7QOr(H9U&oz3C
zkN@47M>60>0xot)xi`l0A%wyUi3eNOF0lWS@rgxZ*^6ZL`p|2vvTzC%d<ui0(!iIM
zHRU8JCAbnIRT`tzfa^fM1e8=-+65xza6z*#-Q8A5f(NR3(1up#L4HvY#E*f6gu1A_
zTvT0MT~gAO<{EO`@22vXpSC_2JK@}&bhC*Aph8zt{ipIOy<L)=*3i<H+AOf~9l}Vg
z)rJdYWM+>YUXgZvNb)-J^pMcAy(fia&5|{WOQ+o4y;HHPG5q|>8Y=^^`o)Fk5IO76
zI_p1q(%e`xw~r1w?YKKZOrwU-zxqU9fb78s{j6~dJ-@YAXxq7evK%Zr%yTLF>x11J
zF0{y;y~EgjP}WCz@`0%Ksk_sOMyetltJ<_?i%$n4l({)m&Tu5LxII>tkdAxB-_m(%
z_dTzXo{7jaHEmLws_LHH-iy6K0k$br=V@zeGGO%@8CbqMGyKr9a38(vZQIXwT2Pb4
zp*RK=OyRQdKV$)vH)vYC<Tubsk6=y@;*=ue5w-CQJZ+zDXmav2bQyrAZI`C&+#}GD
zsa%K*BG!udJ)6$s#_Zs}vN@opskpOn^cKQlP)3Iq-#G|jcy+8ad&4fSjIIxQGs`1p
z0U)>2yrs$GFi$QfMm%IN8QSRdAN*8ihO0NL)R<m(U8oO^1RQc(TH2)DIpeam35W4F
ziDO~}7;+zAM`75a&YcwxpUTksPtcw{`)I%;8rF22mNpFe|7klH0NLt(B~X!Kwx|0B
zikn2NzhayIPg(|WB!DO=bx$5UwloXr-x#k;C88e!e0<t~2eF#(gx*z`8=N4JoOVQp
z0KIN35ClM^r+LTsW?<`Fu@)aWIk{=QI#k%4+OZd;y5G8cco_P1SMj;F_4U~Uw}g#Z
zgD|G~kLy;iU1*u8Qu(7pyTZnq>;YBbzLRfF`8)?1MZ)#;Y39n9>m|{^;36s+J@tVE
zDk)So?fmoaNKwE@jh>4scITq6WUpG{`V&1t+j4wh>ru?dT#jDn<|CHlrGKTTxI7Bx
zKG%>s32iXPTzpgb{9c|URAfO{_Djlym7F*ucCgTheae}qJ8A5dh4tXXMvE%zVxUv7
zHs|e<qpsIP-uL5tSnVEk$F=436q|NNE7MMTaC%4AVdXX!X6AN4+NgG$LoTJkTZaxk
zRCEn^HvZ9qHmS5OA_sK;wfp&%QACU<BJw;hxlLzn2wGzxn8mmEOVJl3(n9kMw(4S@
z+kJ?WAOVOumZ0a%mJAag0}!;3IA7pjfq}J!;m4=i_UWu1S-Hr^=i#v#k$5mCCkKqi
z@yu(_(2RXL&cr|#m7lK?NYJ~A1flUHu0#$>J;3vy57+>8<9^QMAUy{d>Eilq0mrRW
zG0F6t9KA5|Fh4)>l(idAbaOu@H5_KTg_M_lCK%DUH{#7iFgZv(5Hw~rSCnx+udd3)
zHSWK`F?vbA@x(Yq#|w^)GJDqLru7jLN}ywa&hdqSD_BfKET;pZ315~XrV0`<F<)2K
zP!>OPl9QFS29gpHbK>;@I+$h9$y<Bqj}0#*1_cH_lNgd)Hry9Dkd7cAD&-M;^mEJ0
zagKxgPYR_|FuFsYk5;wLpbKSaPbz9#p2`p)^ZN$m*M;{`LT`*yuVe3d5;6O=zh}>%
zcU#@8)i@Q%Kn5Vwm4C)FDE0qn0nY4Arp+!EQ9C5G5ok-Db%$Q-Nanc3nsM)!T(yc+
z(cUwG=jo0r9px;_NX~Q?8E(wq?Z9xWB%QNB^x$isVxIy|skATMFKz9)9O||!`c?)j
zU(tG6l9}ZnV2qCX$n?xP^Y4&TAZ=dYDSGf8l&e?F1T9TxFHTK!oHJ!(yKp${uNX7@
zC&R<nW=4;$2)&)IcH91Y4j!J_G}SK}YHk6}#YVdSkiBF;X~8+leEXlNZ%q1Pw$Q59
zTHCp`k%FYJzI=ul(U80jk~y^CIiy!iLfNzNW7Y+5hbF|V*_`Fz=m)9~-@nQq2h3zV
z@v3&w9H;E?uU}`MooWS^ucCbAfu)!{sgMuQn+u~vI7XYk(c2O+)a6cFjrJ68T8i6$
zPEKZ)660&D`f6>ZS$`d}=WWc`aZlvM&1{%X=9eL+*sm=*1D*>>XQNwa7)qZFT^kqa
zFuY=~uXw~~V|<hA6)8En2WXo#^}P3X@QrZ8jDO8jd=Di;mxszY6Iou8;&`H<-Fn1C
zc~Xdlg(W?;tVbm<>Ex%_*mt?9T)rrM!PMUnNCz*|`1CcHO|XI&t(qGe_8&N)`y{|%
z7H4HO`ydrHwG`*4h8k=PBm!c3Ok=TI<<P~l#T$XSGKZheevFH&(U=OtAGru;K4Br#
zK}mJk#7nC9b0w6kjiJ-4V-3gk$Ac?}(r+FmqYD=9V$?iY&942`*D08Nwf&IyyO;aS
zK4?4Ii13_@V@cy}w%52$*N_p$Y$7S2DyJVEICTrF#<6kpjQt4LS+35Uzxz2K?SDkh
z`Z?@K$pJgfFv}cIN%|Ldx9R7fPw(8dYgOYk2u~7{jYk%`5KN3SSaNHeCo)I2gwrg+
zGn2$R#?x-K1w|0fqs$*Eqqc)}pPxhgHJ&w2_{qVhzImL*nAZC>@_JqCI($zF35ODS
zV~^rf<E~kGn?dggE7GDGpF1zS`Qu+u+j31^h63C-XChRUOBrJwp8*R3tX)ijjGYEV
zoB`qZGd5OjJa-V&TV5R4P~IAg&dfAL!vx?OL2tX#4pcy-Ten~H9AxgOn%nsyC-hyV
zv;ako(@-uCE$9Zk{*C<jH|Kxt+51w!V)E#Mpf0)Po2s=lP8)NiGrJi1`MaR&D44)Z
z#M%LP#=Eu2l8<}=le_w-yn17MOunoj2riqQf{#xN3x^f%;G~C?tAnHujt14ZLir=C
z8|{kFYH!fXjXwQBFKGSI^2IGi>%ps0GidlhK^)JN-ZDOe+=LGr9dCkkKBo3?sBQbd
zzL-@);_3P#3ex<e56C{ju)HA1LN<)aj*&`PtYlGzg}A_b8n@YSiAh?GOG98%@Ej_~
z<XuhqQPI(ee<F+q2M>x;AN@io(l)w6U~Q^BeRgT^#iDTsZuR8ex*V+1?zCUbuRya^
zYZPlulOG<Ypi@v#h`k%5+%f!204-(qp=Ib)H#SyzOk|)bH&V;OCAyQDwCj&8B(Pgo
zyc|adBQrWoRg|Vz3-&uqeq;JB&Rj&MMV67#(Uq}5#l|7czPR;Bpe(a^{)HUr0Y2(%
z760PojR@xLPcc7w^B$Mw4ycVcM@C!|d_6d9^(k{Qvqt-;@`n^fb$V(mcZ*+&FBu?e
zbPl1odTFkYs{4FD`O4xDA5){U9!8@eyzBGjco&AZ1Q1Zd1uo_h1y!u02;{xC_*9c;
zlGSqF<s|Q-T1;fd$J3HHTS2Jha;i?81)@wdEuS_lvUn{VF+KaHW-U@tW$!OLpbcRm
zG!9VwZeqqbi5O}<M2x24F&tfa3&{h#71?8jK+Nk%3BgdML>9P`4{Gsm2Dx&zby0nR
z5u}Peb*2;jnB|`z(o_D)si|~)MjsG*h27f+l`;+n7<%+~YrJHDmG8V7ovbx<T-kbq
zWG~|r6ES6b94ZL-uwbl>jEPZ1wvC_PBCZj%V^GJhiBRfz`}!^*e8t+%?t4!UIR+_1
z!U9$pFdk9{fVzNi))gj(2R9y8gTfau*+7xi1(S84w8%xdWWVxk_wF}PR}(v}=C5?9
zfVbe_UEoa>C<`zSKrGBg`rr*{w?V)5_xB@h0G$NzCPZgd+_|%Xp|~A>^x#zyh{*oJ
z6TT>FeZ?wF2(hrRrDTzYe49t{GfhK}20S6?$y2e=`b0c|Vn9*`JB?clqz$a-KSo(Y
z0Y;|v>f&@h?Pe1uXr>QXD*0hmbGWdSx_E%(N~9(4Rnu8fhbY(#pPe?DlY9^N7p?*A
z74}iIjR+aW0UBh5TB4w+$oEpvE*=T@6Odd~3!elfj44;+8jfsT6Y|BMuLDRy*w&2(
zhZ{F<zB4{`A8Jz6ig|b!Yji|tG^(L?ez+O%nK>W=7;P(>Oemaysyrm$#uQt|Uptpz
z(E5?}h;oMhDm<WQWRY7=xS;TN@DhmePdq$ew2?@__NAbFk?G6e=j!S=$)`o@TUv;4
zp4`VlNCiX!2l#H|h%Ws6AioZ{cv#@#fzPK^<Sw#|p@rnf3_4ihpbo+@krh6QH50GR
z1+4k_$}{vp_wV-nMnLk#tVoQ^ip;R*Sm|y5SETEk*}riGsjJZ~p8M+e_ZsA!?6wG%
z4zW!pIpd(1qZq^*lwJRX`?dQ3OW8tNbB$h=VuwQDRG`d^NT*7Yn6F|%vU~EAmO_W2
z&Ns_~f2j1RQogImpfnW}c+Wsq79b>XYc}($>Hovtn}+qgzVE-*gq1l-M46@0P%@NO
zh71|fEDa(wP&6P^)}qLip`xgi<~bFLYMDZ5p2)OFvy=uB>i>FM-{1e(`^o><``PdK
z?cI~Lj>@O^aNqZJUFUV4=hb)R^kM7ftm?7Py$XkAo!?jQTHVq4wzE3ZWT(MSw;^tF
zljS5QODfbxX4m~r^Z*+W46m+UO+|3j)U$z8gL-a)hErE(r||_}i2Sf?C*Jr_2R#8K
zSF71O7P2ptK4`a;7<~cu<-HxOtZhDqS$GG_16RC!>BjW%cncRcBcT0chgFJOHY1s<
zsaZ07DQF3v6jX1-WUZI@Zy*t|110}CWRmVL7U_)JQkXgOjHEtMLWb0aYWVigZ&7=J
z?NdqIWA~()_fS+ZTg&h30oqt?+VUUUaSxi)v72ju8#~4lgAAM~(OQ)^%fJ!93Vd7D
zW}+Mg?1m`CbNZ~sr0j4&PgQn|(7E~!PImikxt9LMjgojNSzSwq7M0W2yRg}eZjjKa
zCIBq8`v&uzg;Y}^EX{pcGGHt}9$&~JwueLYqe?claG(|a$z}u*_HTRkeb}~j+2~%w
zjn3NXj>G=u_U%XOCHdKDA0G|phT(RNA802CGK9$wIM{hCOXHTVW&RHynesOx-LX_f
zADLo#uYdI-!W7_wSa*6@d^`g@4}3UujKd6vU4uuBA#gVYDctLvKI+GtwPOPo9>gXe
z^Xi&jMI|L{*Pos_TEkGd;DshFoG;KYh5Ktc2yn&*!_ZNE;^6@(q|gRXjB)*^B|mF)
zICGZr<7LytXy7b1JAVYj@i}-Mf0+H21P$9)lJw0(Qs#{-r27VBE>7NbI%Bf8`aXdJ
z+p>hi>ft1-;(Z%3onw4_hNVW1@}Qq?o@PBFHFMee^&|Sq)ij3#QIOIlzTQw7a{PE6
z383I5+Z+Nb=N5g@IRQ@_?u)9?$A*+wistx}#raE@@;xGyNX9G277IWZY$BZ20?sFa
z3G&rF&U)K6t%7pxpZ+?k^|7tyUP@hOXrSY4M@4Z(Me)8l`B9OHTISaq>kOpa4?Z%r
zFwk4+<lQ1#t(Ed^Vy;PU@OSkutxnTF7RGfNefw<LquTsBRjIJBN#(?KGXqV(FJT9j
z0zZ6>+?@cimF2*(gaS&iQl>&dn8#FKe8pWI2YQ`6`2=OA@zG`k^9dVLy*9>VqJ?5>
z5DQ!!9ZO3}u#6F2lYrUkV#um5uyH)bT%#1>Jvg2?_x^QGPI$N#G!xVulYVh?a&mez
zS{k5DFmJ+Q6zLWl1Xy;VOH+74g2Lp<G@?jeCCMxldtc&p+oY`ck+rk~6avl3?|<E7
zmvhh(V}#fh6=kHO(}sYVW7F~QZK5>xx$7^^mF$kiYOsl3FKq%s!LOR8xA)DGk+DYW
z3#US5?p(ovMNttjaR2un7kzy(C+Dnp5YBC27Rw#N`5J!DHFJq$+t}HHde?1x<Jy4P
zQ)kRr`N-%flKH(ff27o>WXsIc98Ra@HV(|Bo~~;a)n1*UG%+bI4r5?-^YqIc+LZY7
z@tpf}YV-=C6B&8EN}smSmnQVB!*Cz(RFm=ci=f=Gd-RhGI>w%MRZw+xEK8U_Y2u5!
zeeYBRArw1B4NRX2D~z_n^oF62^fTx!uGg@^lFbDd+1?A%cqn8Ds&(~bWY(B$JVY%-
z(}FG}aRy%ho5YVycO1HI-A`xVV)S*hOoD8CIYAeC8>CK0<us&r$o(!ZSE>3Z^xCy{
zL8?`FncyXiUWLQ`z%4=`U+e1Ly>~CkG-Gk1W(v&D#|eirkSM^o*%kFGvyErb$!95W
z1ade$dh+BTmSvm~Fe(wJUy$E0X=cunA2fVqj$o*%sU7gM5e~5r2#I)8eQ0jhR<w?$
zma#L5JAC*a%kZY(E<_5<xc2zx@1LNmgU;aqL6Y%rdxLsjZGY!6$*0NLuJ`Scs-_(@
zrH)*->^B!hrzK#Q4rusD(7V)yt{9-q=J5-Qj-EbU$YBzZtpBs-aDY*s*!C>(%M;yl
z?kqw2MH-+1_Vwnf=A1dH0D|T(M+v(XjWS8qM)Sjg8ycINX)A2ZI}tZ&pS9S;)+O#K
zUn89C{3DvI_h`JzvhSDQ**5KML)le!?yB~mU!^VcTUKv=d82KP{<{IO@iAiEGQVnC
z`EA>IP$#E(dFl1yo&Cr52eYsG{KwEcGh$>%v}jHh`7&=0i84`n;NZb0eSGRir89oA
zZN{PW9nY?@EFczLm~afTW!<rfplIv_kvSo}p2L#$DMME<7w{7Hk~2NT4A3Jp&uOQu
zv>)vqs?aa`d+a8lFCuM%D!)7A&&!uDk6&n9sr8I<8Iy3&C*8{544Q#G@Cbz*`FU()
zSlC;tqZcLO@j%*N^u0w#<IP1grqt^ZOVN9l^mbd@1qh4H<7Tx9hyKWs=i#UhysY+M
ztF^Vn@ZrS(ustDHj}b|ThjmR&vIj-bg!{t-J-@anYT%r6JXz3Y?YJ3hx7gcDVY~#w
z^5L$>w=ohDzxMCn9&&#Q9dk>{RgQQlMI(&g22L1A3l#uvrE4M%9rHUxec2DGH`tPB
z4TX?CeL5LD2NLN`$Bgc+h-Gp0sI0wYb8F|bYo1R!)mb?6BmXWdW4da_&mFr-W;Lj8
z_I_$6x8z$(1kb??x`I|x74HupEic#C(9l?9Fi-GGSS`L<NG@Mzy2yPI%{lPdzQts}
z!C0OOA2MgVZ-?WD7xNjWL?uRk?4z<WJT!EKsOVkOVHcD+7$71QGKutg2&~-)9^J`D
z#6u!*=5KhKLhjj@;BRJPvL8}2_EIOBCxWW>k{I{+M*xN)5UZm&_)J^#-9!K%Htp=W
z;`dBtCo-coOyG^x71X$E{eBfD6=&F%1T(okXk;*-4v&t0WRl_YxUy0J>O4#=-VZq*
z#8FRfV3KB?g1=p5wTCj~Www;n_+o|m1Et1lD@5(t-S$>aFUdNt{abO}@Num(TW_u0
zd&zN;V){&%+7`b>SKZ!Oz1!EKXJU}q*6KOL)WT)s)9bV9w#1(@7XLnZtKYUYx0Drb
zoLqEZ<qDU}nGsbN6tA1zeqot$S-wc)=(I`mH&@k|eGK>*rXIc2QhmT4JWiP0&AQlP
zXbhr9KEsAI$@H~oH#Rin!uOFumyeMP2X$oZ4_=O17taR(bPyOGoEtl5)~q)wWk<tf
zV>7Y3MM|);Cx07J2%XjK#H7JPh5!s=4<r6Yaz}TZmT(ja`{ugC>%yD^uX)bTpD#pY
zykLD7UNqJl@zR^L`o)CXWNjspg9q8L`bgv8Xw+ED*ETZR%T>}m@!K~0bR300EkJeG
z_9=Z2x(EJ6O2a5;Ye&a+Dw%z!CTv=Jh%hOZajE?E>psu}{%D<3#Bv1wu(07z)kv>%
za%n7`iZu=8<w248HrCb{GgBdooL@L{%i;L*kM7To+=K_dx3@RT|4qm7G}O)a?$vON
zl`7qKBn+6;RWl^$pXb*y0x3txMLfQ?sP;-<r)KEw?d9r5T|8N1f+UtqWy9;WR49;}
z!_O@HOcBNa7s^LD%RSSC^PfHS^Yd9VXEL{S7s~XTUKSrQWbP4&m5Gg-!;kp+8SLA4
z6BmZIA;`@asHwf67ID!(jX0BZs->mni|TCmZ{2j36iw2O6&>|~<j3Eo=rFFDckb71
zZM$$ko^@Rb`hnZmJtKzd>*#co2!qkSMO>NF?G+JGOZ>qbQ@+4zTou_X9<;By(+=ro
zB>SrHv}wp;B53}2UW!LEtr!WMW|+#2%U_>Q*4M$u-kN+{l<oZM(h*C=hL2LwdT@O6
zWSQ#{qc-ZluN%8jbmQuBJN<I8Q@O@lM6YUX*}P+xxTv=Ll+g!<?0F!%?6>e0%YXgk
zno{(#?^Vz<OUcv|?ixv(K0SGMPjTjU52xezrLQ5J+56gUFT!uPYjfrp8t)G}N&)_j
zph@UEBM~@SLIOvcJd*Clj@f~ug9Z%Byx^Wx<&$;#`{YH^-*#8le7u#LwD8v2&h>HE
z`s#WupL>1p=#(y5uB4XTR#&IGZn|%>%K`7%?Q^_#N`Bh<{z694gYDB=Ze=7@4a@ov
z{&Iwy{<;TwZ<E%#cje})UcbNdvC@+ZG6801vZmT=w@M%IdAjKCf_w8k{d|1vFy-w@
z^i2%^`}J!LdHK~{>CW4kbGzQ`8aDGchN4C4d!Jf_zrb}47BB#Y*3jqo@aQ8c!6(!p
zWc_;TX%w|xL<Z_oH_|>VRKUitaaDb_k0~9d29exsli|wniG&&su-Cv@>?@(CcK-ZT
zsjoYW-NLcl=)?y*oBV{VQRo<B;sDt+IM|-9#)l755)u;Q#=S_lGxGMo`1v;&S_OOx
zR;RnjJ4sbhsvy^6t~71f6c%Xgy2;46^{VvV4zugI6{#a5oRT;%lUWH@iok-idnH91
z^O^VVO@iiu&>5U<`mAS5J3d!-7L3zpv>4xr%DDId`VAptO6IzN<T19QjiC?+Y%62i
zLOE!)Rh39;zy@(HpGUUD*d!_eMs84mK<<YmXzp6;N4moSWViF+M8>~Q?m9VLrDX4#
zjvI?RoXu3Q(Gr~d`2i={0>CuNIhJ}m>JaE+%qf?a4xwFq#LEl9@~Rs%m6X=aE3_wN
z$=v4AaFsSSDYUMP7$*-ISq9IKutU>;vjeA7S;P3~>U(k6@OWvxBj*Loyu|J&G;>3~
zj5YK1B|XQ?-i%KH<>9*$%72%cjuJ5{0|rgaD4oTh7hHTSvcHia2_+2K*vgeGj3oUS
zUH%mhp^8rxb<p;W<`+#a1;~hrl=MTn_Sz%K<^IhtI!c3i#1#gO+z}wS;m%fEcUowQ
z@>lVKffnD`VWXG7u=V!HF=I05@gQ(wuGqqb3t46~8E=27yA&ILACKkl!<N2HS~O|W
zTVPcdDSD7hA}b@t_v;7VpZbsId$n<tXBrxP7+2x2t7A42VIdI%J-F~9+`1}+k7L$`
z=$%y@T(<9xy+?~N2&>B{vUy!s%Z=Ak%GSluF{t>2nj8TX85J+)Dhs>m!ElDW#3n}F
z-#87DYcCWCf`13+k`Uu(&01eEqY1iEgH}Y?&p-bx3<qKzW1AcT_YIW<-AHVa$=7((
z*0i!PhQ6O~J~rqmF&tZibzOOJf`18}MbQyj1ZK~LUfQ=)!3GX82PjQk!eGY7EIdIf
zxH=YxsHEfpP*>gtE_fOU#ZXF7`l1UNI%G)2_LYB@Hml(mJzrL~iOB=H2?ltjRRI3M
zmO!P!Mqz$@7B34&k{Po^Ma{b)vn;LmY)I-{FpQfsM?XS8q5MJ#PN{gH;AhVa0KpcR
zL!kef87fAe5!J5NOe7-#=NQJITTmF0Ck6+9pqOcF+>qn$W}s1d-}BNkkMErUS!?Sx
z8`f3_=?z<JGye9+rS;kGL_FU*-?bU=yzk4DcA4h0i;B8rI!fAF?nxTjNJvSTSH;=8
zMzxP|9pnAXJEiY_?yD0DhT?K!$G3YH9cjt<wA4y3F#a!j%g47Gt0lW%?Q^{Mc<tA`
zuG=f7oGVm%Qq>VLuj?QED3>7fxGyHp4_)sa^)09+NY7DEYMYwcuI9fh-o?evUgsD*
z&`$f&>fx22WIE0!TiM#%YOcRyG(D&1&D~+4DvLLEr$@gveDL7TR0So4MN{2h7r5OG
zNxLuN;^(vORA0kC%R{8Tr|quDEr?a=i2KLrPR3uCu5SF$@gnVZ+AW*CMe1sMXEdH`
zII+!O$CSAvU!Xs`KO~o|_cq=g5oNOZ-EU7QmFK^RjEZW`>=KD<Z@+Dq^ZnDMN=HIz
z7X!<WMpSn=F2IJYDzPthlO{Wt7;UzH&MgZyTBHo3D7~HEQdhT!vj<Ff(DhaQX{D{(
zwCPG+`yMp684Stro)P@ajTstY&rNYqh^_2AqG=wSUwlF^&l3LD;!p<FrBqfvQM$;m
zpSBZ9*~MztIIBhLMgG3Mt?+V;%Iwcf!rGYO7!gseCtl*JBb}mO9}21(7P8P(GQ)7p
zv}XA&f7{mfo$s^u+f(?QZC9~iO4!9D5#J#kLq;zL8VJ1$PDJ_g%NKkrdkjH0K^CW<
zobxEeqc=vS_8l9@<IUGH5l=zvXL4YT;DFuRt9<A}9>+zx;d$figBh+>xBl=>TQH2h
zGjsQ@7uv@P504ER@XxZvQUiR)AP9=7y8$<Y#4A;Qi}_lNGD^Sg!9~7(hZq*d`-fn3
zhwj!PCil^_$&;7ZY5PMRz+g^0eZb$^@O8EA7HA(l_>ICtz9RU7jsS>u)T{W!o879s
zIqHDLHfJgQ2DN@l67BP@JQ+A_SZi67Pu*E$9iIS!{B4W!r;!pmeY^>UIQnz<;a=ri
z<wTy1xAr7x<`?^}=c>v~&e`ymrX?=@rC-lQsxHWsb!XX1E8~rR>3{`xtnSc#&Vjw}
zpO*a9GUIB4T6D2%Kl7QZQ-96_%lC)WVz=iiDg9^07v=x8X20t=1o_|dK;YfTeERS7
z>gT!mtgo*#UCgc<AqU%6dJl5X=?BJz3@tk2xD#t<p>xBgJJWvJ=SmEZH%{5i^#l%3
zGQC+GzhREHN@GOWm?^ZcZ4~MUxD~tVgm(11z3)-Yp=;rwWGt3+XE}3SF!d&LAxKdU
zSU&sYyNtA2KgWya7%K4VI}W_F1bfJIn$_>oRaUA=P4mP{0I6M`t*`rfZ!mFh>vLfV
z^=k{8gd1%Zo+W^pzx&%SdnQ<5jU2h9(U9)*$b_z(O}{P}*3Vr3-R{1JIQzR%HH^#`
zd8N|t&~LlP$ey6CC$^0ed5}M|89RUD=?hNxZHJQyFFws#Zj;3S{zXQKY-*f_EB1x)
zUsL~Md)YqmznpO2E%~pJ?T?NM1EYkmGs9f5Ll(X!NVosT8267w@BVAv`@u*5<(&KF
z%zsUO{~!8+$>!oo|HTFPKi+<dYR?loRtapDlWQ2K8v%Q6P3Jz4AWpHZ_RL96b_Zh$
zd)Q(BPTlz1A9cC?Dp!6nwP~1<TfZBil`pLj9)J9DydF3z<X<wXU1yt^hMeMB#p}*;
zhprv}68rBvIrRVd`9IW(DbJX0=kDIi%6zwjRnjjUU<{1SqT>-F5@h_FV?OcSKcX4A
zMBvPnO204<zRt7PuXDr#f8{5W8K`7C9+6X2QGYJTYo-O0^n|bfe~tWj4FBKU7XQbs
z@Zalbd9UzNcv0Vc1k~!lr`1(gnImT+WNQb3p1ehH@Pu|`9h}Ri{o88ps)*9ko&5ga
zecmTezF<&AhAmTEzhmh_S^mUeB0pMv|A&2Yv*?&HHo4WL{mUg7j<Ct#$Eg^vO<+Lh
z`o!Ql6GxsH2)pIUU&6uqcP98bMA9mGgZcC3bpm+FGiR*d-=_cmyVDu{)6q0nz?b{K
zeSa;C>^OtCZTr>h%#I7^^EOT4z~1D7_W`o~O`8Ej6!El}t$lzWZqoX3N>@JWoDs)2
z1>ThQmjBE5qX831ie^Rh-y@8BPhR`u*A4T0D_(Qa_&aUR&E?ns>!Qjjbxtk1L83!(
zyKRRs?fr@8kDFN~&~&iR$qmZ465Q6~mi@S@tdl2@v&0d}Y(M?@Ao+jqu8N1TW9N_$
z$;-5~utV|F`f;%fegv21T4wK}C0bIn`I;<YN@cPS&ezGdUS4)WX*+bNV#JS+a{l*P
z6Sx|_*>gd9w$+M86;7pok^lJ@oM0f5l{w$Aa9Nc!i=WIKFGR9K>17^ywj}dHdm;7|
zPyLYhzt?33(hZNdYsTZX!n3XP`EhgAj63@TY3lxFrp3_Ro1MlJ?-w~ixGtGrI>&S~
zyl){G=a{j4qo3Z7An5xOdf7N;H#!GUmF1U7@RN%_Ui_88yPTchlPxX%iD?U}uMg^7
z7tY+jt$*Cl*2A5?6`*t&cjr|K|48DLa1B>>G!%y(rU#q2{MiN;(Utgr|GFgeYr$!i
zdd{G-U%K?jlNRA_cK^7WyuoIN{}KEV>guLB{dnu+ga>?R_SJvi`Ut7*!UO(e%;u~5
zvI;Nz`^qm^updRYzrC;&ZsQz{RZ_uZ?iqG60H?pLDBl-_W$+wJsECb!o_n@2^QU5M
z91UP`DF_j2(<w;V((3!IgruckJ}6zP^T;cs>*AOdCVc%5O$+{;o_`0lRe9!rx97n&
zK~^eyg2tBM{JniU<uP20{NkBzzy9_cgSbU}rcaq746`%2l#}B@vDCJY=^wc8O<n7}
zYnS!&>j@1@fQ1VR3+JX`jMPs_ern!dK%pi*s=xI+G<pL*M4?9Yl+iS89qILfe_W%$
z-&BKV=CX}9cRZUt+4k!{hlee-yS49QSf{s;TbRdzu1j>CXTjcM;EzOg#GlK4q!WKj
zjFQ#IULk<#5+E}8^o`>DXcXrcuk84->}Lvd0b#?ua{szhTUE$Qw)Xq6vBx~Eu6Ace
zA#`JZdts7C{J$IinK16o-%R`8P5oUsXu}y-6^~LOR0mw;W=4i^PEl)RiZInM36moL
z<?|_j@)0sye(a`jL{v2p=b9YQ1tG{1pbR#bx4(b~g}Pn_*JVsAbANf-ebi|5KcGB_
zj~;!g+@4P{miuwI{UYO+ck$vy5Wt#$yRCPBrq93MQ26Tt<HE}eD5m^XugCtaAVHKE
zf|CK<w}(9afKVZHC6A34hVN`jhE+h+GVPqmICar=t;M1DJbF7qcwRy!jgeZhNR60?
zCw|$q+-OxFAaEe`x)}Q4$A|nlY;td6pw{s5#S2sKRN}{?FYnH?lNmd(He~z9uCDK&
zo;%#w@eFtR&8y{jWSiWmVENBrBO{}d8o-|lm*Wf!2IZp2vubc+6kKstm8zN=fO5T8
zFvH;c+1=|F3Wmeas;kk8=yrzoeXG~VEDY4}$WEM@c=KjAGkkz+`Ut2?2Qk6Qo@!T1
zS%{LAUO&#>O-$o$`}}#d6i(P)iE4^6GQ!YEQ7I!?KSmG5=titUv&XQfjaW~B+Y!|V
zY2#Or2$5>qb)jkzfA}Lj&j6*#FDx{a5#uC-%s?pl06xk<(ByF47={Cd>Bfj9)Sq9s
zh;=(c3U=%_*c5TD?|79-S)Z!8BfIkGP=k43M}&gh?-Tg!WYHgwQTNk_mvQT6Md;EO
zeg`Oed=U)3ietwHzWI3ZqC~>XwYb`-V?+RyYVwe(n^H|6#H>5dM{ElE_nF25@IlKk
z+w69OpB<C3NZocSQ4j9mGM~N7QY^0H`Trd|Q<z@z_ddEAJeF=yG-PB`j$_c~Ep;O$
z#Rk56Vk%lBfT#4Kv}fEQoO#_4<Ihz7n{({rB&S2YdfmF$9N|l6!O1GA0b)KMS|nqt
zFYJ!3gF1xWv09hBg;w|)u$8crkjXQG3r`w&!_XF2SIiva8T=b%^lDq?CjNcG^P_P7
z>Eh>O2nCQ91D;IH&31Njjtk3ja|PWeI>YBgL+hzDjq7ZE7$lz&D~grr#QnD^G)kw#
zlpAX@ozKem{`a$JRg?(ELpz@vg2+*9hdmFCJ;<=WSFc`$4TEQdv^(t2D`O&}qE1Kw
zxSs3u7S#+*T)>r^uRSTm<>D#?;0SRqoH^6>@uBn}Y5IJERAvGFr8i439u6BNppahI
zHTTyHENirl<w)c<6PJ$?Sw^O`)Q99yUQRAw?GzujZ{m-bE>TyuPNStf=6r*H3`658
zQ7+SzjB)u!lNT&UT)~4I)wPL80bejSHI-i=l0+{fbJ{yuvE!U#-cqzlFFza0?rmGV
zDO<>GEPYqiLIGR~10gk4poF${lPxTZV^>_zTxXWDdD-DSNlfSGC@QY)48>rtVL=>o
zeYv&KZZdFT)f9xQTmSFXy=P1mX5L2|marva^`cH!-r+O=GGA)E57-9)v$B2+3_Zk5
z5VkCrT(Uia=Y?74K5$i<m-6El7~%X3c>)I~^Uiphq5&Ch!jc&GV@aHx1<!>u>iN^B
z*5i$aoRwU3@Mhf|PC@`(#SPb%#_a-p<50$50@B=+9(izehua@>;VSSt@lM-^-NyHg
z4fXYmHb0;~NE+T5yDBz`ad|4}O9@DpDU1OAhlz$y30Jg)og5vR6tHCRV!R+fz&|%>
zAtd1s*Xb~a#iFh0eeI9U3F2&`OeAJzMwZ?)!r0TJ)gw69=PfOK_UsF|aAqE}!}{(G
z(|8Z`27!l-rcPU-XwldAxOQ*Yupw|NL@xOqTD0wFyc|3V(Nw{LoSr7RvSxi*KMV%8
zxgO?KEy8&IiQ{j}Z6^E3NeU4`#vr$Lx2C2h1mm#nJK@n7<NMvisU7M3yH$P-s@%Qs
zf<`u*LNaW0px7>!b~W8_r$^W<TBO#cSP;ne6vHb))|KGW#ybbs7BgE>SPe<9>%&Wd
zajI37%zJ=gADFU|f2TC~M--_MyuV`fMSOs|NXNAIsm+~xo#_)yDwz495z@WY3vVsX
zc6Ykk%LeqU%r7bm0ahXnX4kH|YAL+be*@#kUi^}HE+-Z#Gymz$@cAHS7cX`Lxf6;$
zJWW}W<@whUSZw&U+tFL$cIj`;N`K&m!NZvOj#eCL#1{OrH*Wqy6QeL-&UE@dM6=n-
z!wA<WH+u-xz%oO_N^s2?J?PCsJ2SYdj<&YpZVk}AtNd0!K;)F0i)&UI3=PnC8rdkk
z^(%;|=K}&<Y3O%%cY}^v+}x$P;JP64;_o?6)}HcF47*tdX1Lb}?H!B%Juws}`u=_U
z^y7_7o;{l{C)Z5CW~vsHq|ZD<S}I#NvvT^G_n#X!M}E^z&KF_cg!YYp(^KOlA_&lk
zN#-b-F+bKkzuWJBJgD{GG<*HA(s};>%^($@tO&_|df@6CGMVxG_e{YnidI|OJ}4z7
zH|!NWC}j_+<Ue_W8B0vW#-E%f`N(x(AZ90EuKn%-pBpOsG&4N}nyJ#izkGo4Z>}!}
zdHugf^!}$pv|EKOeoFW%z+_zErT>NXP$y3Hr|C+vs;w$=$QK?1jL?Fd7oZLsGS?qE
z2uD4xx0WXQg@{WXr7N|yp?{Zt8!a|sgk_`FPcnDBhDh!}Uy;*m+o<*Cfu%!Ze@fp}
z0yHOxtrW#Z=7|$0BC3CBz7Q0Iv2ZUDJ@>%_*pLyCk%rMap~PLfh0R1DqC5$m!bZy(
zj|&SI;hasXqG)>a42{0am*aJ0fKD*x8A)Frt{DDT8#e~n4*?@MfDVR;o7MA3$a+tX
zDZ8Tii|g1;1*p$xFn6*b7|*)CYwdjM@B;^gu*e8n!WGg>1akiN({#KFIAEDR*zt&|
z&kE(@zC-AtR#OSkciBUNk0uZ=7DA5^^Ni8;Pnr}4`IImJgk)a`f7TqPzy&$5LTSV^
z4VfdBd11`SBgj&ad~S+M%B#GH7p@zlY)Juir2`hY*z!O0inrSRVs5p{_cu7w>HjQ@
z?lr9^B;PC3lqQ{yRo#lsmcM;|egs`461G5(!xVWaSHnEX2Ev<__ELt37BPot_?sC*
zG->>{T89b5^ynrhhC)|8mv}uoy5W*T_Sv-$;Wi4j$hU6?-KuRqiD<7xjiIma+oR#9
zNLxiPyL$Y1R&+beG9iH#q2VpN`z_^Y+S?q44Zlp2F5j2weuNg$qh}|1C1e>wO4AcW
zQdFpv^KH`hdz%D%&h<H=raD$A^;Hs<-r_lDions-?JP>hi;k|2Xwjq|3sUziG4}~F
zeyeH;lM#JRP3_EyX$*u)@XDjl?ZAPU&^hpgNTJG7W40wF`@Vo14pu{zQL40Q*RD4H
z9wzTQbZSXG%3QYCU_&Z89U_N$+&40(HD5JPIrsItH5zFtP=QZUI1)-d(6AMgso05m
z+^oO-AbA_0B$Jm(^wloN85I`q-}dBUsJ?7Lm7?)g?oMdTORBE5TYbbIG6$6_L`Pkf
zgR2Wgv>y}nSc!AUQt?h}JV(a?kblG{=ZI4bU4YP}r5w-glXjG=U1Q0!8OE%yIPr$X
zMH~*AvwbZmoAYB{I3^@XkhKB=0@~Bod|YhmDilLKBfJ=R-+2A+bDVD&q=vh`qgqu>
zJI5OI(dd18@YG*^1Y|>&SI)Qge5O`;Z-&V%F>}-3uz%Ct)2*VDed9Nz*7+Hke<4ZN
zem@bh#&tBDaBoX;+8J?h)gz6BnAN<j;}Pk$_HFy0djtr;4d2j%J&jR;%i%`8a;+af
zIkt!W<I&SNeNOtIN*Ptv8(cM1-ICOUtR{M&AQO`|dPYoPE+zYO)~poYCJWXzuV-^-
z<6RY1d{tN5t>1vJU^Y|GT&~$TIA~yu4O<yp1R4$9Mh0SbVHWA$l3rCFHIBKd%F5+8
zXNv8!K?c2*6%e?b$_{qfspuwV2V^<bpAM-0&CVIV8x#`sN(~!&sX)dumwET@lpUk|
z&*9SwjgYxx@xDzSz!s;ZiW=U%Q?`FGuV6B13&pJJ@>^}!1US<T?&Z{p6E!<laQfiR
zcQS4#iZfWGb=ITfOL&OX{(Tb8dEl_=3l2t@SlZl#X$*aAmBWGaoRuS0h4Ro%=!2i7
z#^<0ZN!#P6P69=KX)tWlj*_qk>=SyIFCd-5-i5e;t!#ErQye*-Fs_+4?0W9A%bS$=
zTkFrAQGrJiU9F&j<Qf?0=*&cU<fJT8$t_tUDn-FDw;>)HBIFXIc75Yj)N$(652bM)
zJPwWC?yECYZg*8~G)cAQoHN+pIMG~+TKY;7(4@_HGRyE1Vfe2DC+;+rft5{g)`8`R
zOG(dne8=gA#!+CIw0iVyytz1!9kA@Py7B?fh>3V&!&|%?o?BVDhcVsoQ4ivf$3FVk
zLYO4-b<{d(0>}4ehP00ZGTyW2x_8Ss=H^tMz0jh#mthQ6R#td&G(Cr%L}<XHwo&#q
z&xHSvot%NDP2d}x_WF6N1w2LQODaAbw8s)f<D!t9o8H`Y=iQz=W2rV5I|VHt1;YE=
zAaTh14A^T077@zC;CVPjmyFoOX`zg=rt9#qMZ={iLB7#~gIkWxp(!k!z`*?wCt04?
z;4<~EV0XHewXW$8-=v|KRiLX|3LAEHy!a-Kqu2iYK2Cn^_b&zbcb4v4FZ-<Mj;Ig~
zyQ=DeezUBy@4F9`<uryxb9dykB29Obuk)WGC)eG``Z&`14~sRyU<Y;j-OWhPXX_h}
z<+i3D`~<({er|5l0(kt5)KA|i6}n>jfL}`EH#keqYy3V=AEY|c#Sj+&gjQXBD+})a
z^>x1W&?V7^3Ds~KyXMPHo2E6V?bL}AaxHsIp8Hh|Z1Rvty_@2^@oCYW+1cH12PNcu
zq&8j|5@x3{Ae}dPBKVQ}(AgskJs_-IElFq5m^$Zq+N!IhgF}(Q|9--)?aqJ*Gs_tz
zc^Uc?Wtb*72RVXt5ZWxDWZcigID)cN*-pGgZAEMSm3GPuyC6XIh4<X`&!>dkJ8|^m
zZW#sYHpo(}egcO*cfj42Q;Q3X4>;UjhD-svMIhvos^T{YP0)uAqgc>zQ2jm3rli8b
z{6v)sYYg-Cde}V;4aT5<(9YHbD$n4yZ659K<6ogMXymkGA;UYv$Db<`S2B;UH$KPz
z(VSm>d0q-{7+`~X86p(zm@)Hq{X-v#{AKX>648+(N6Q-*Q`^{9#Li#1Z~!X+J|Ua3
ze}UtESbl&HulLoQV#`7&W7k43<gEZADNmjHpm<t<H8cm4<>n0c7^&*TAf*jy)@Ns^
z43xv#_WMSMtO}TS6>_cIKx?b?xN+Y;+-kLdcI`o@_YrV{ASQAHtjX#}W4j_QAu$if
z6Hbq2*ZuedI6HS9m3GJW?KwO>%s>{-pD)nvh(HWLe<kon(vK&Pi3ks;@HT3`5)>3v
z`Svh`79vv9tmU>KniZ=Y3(PHTWMpK9pZ<&~8$x}1u;#>6<ULH52``#ghVyGd8nDE1
z?qW`24380bHn{FAH7UKFu|^vM&WD8L*jC7P-QaJ>>V*EaXgYqj-<vJ^bUox&kBL9E
zL2*fCBV^d7=5(?ha=efb3&35@d!n$tjsg1f%M16~gDCS?xkwNaa)B%+il`g;Frwx~
znoSu%f#8ZlQpL~`q;2FlDbY(}0q`dDTe(zIf6K$zMX$&9T%|a+fxW>BKq-tNT>=RQ
zO8VaJnqh}!(R^%%RfrrXmPZRqT4a=LIjeH^Y)5VminVIm%4wo|q?R6?Jmwwov$*eC
z9Fg6g*@kIEC2n2M<!3qlXdOCN1QdblyTJL;u^YPQbz{`1-mR?O$tB%pm%M%RrUZ5#
z4gD9lUG<8_LiKa+esdJ`R~c}bLxQD`(uDy(5WJy9&v$9^ol%e~n$y>NYgc0h?jH#U
z(nw(7ZVnI5wkeyI&@%Q3@2Q&zOTHQF!ocud1e44UCNsp2lUEBb4eVoTWzMcQcQqJ+
z{(UF>o>RTQya&gsls_Wm6(8@2cT4-9`XzLFv$QnSr@aON;2Uy&c_22bc9pBs?f?$H
zKG<>UewmGd3q$R&NG*wFQ+5Knt<;a`J-Pnvglq7JjXJlBo&H!EaX%<Oohj-&c;1<N
zUK+;mOhIrO8ck|y@O(~6Z}kcE@hPH#FvSFY%Xvh+5d2oA5zV)AaiV_K>Gii-v&+(J
zt%1upE;qhyhJdLK5`|pGYz+oe7c9lRr@d)?0~tMhUCH!c36`$Ri?8(7I_aamYSn&V
zhgB~cY0=KXK@k3Fn5^Za?mW39Fn*RN%%$eo43e4B3LIV`ir%JOP^v%y_zBjt{P(;5
z{pXza`3lDFseyL-qP1ngp(Q&x^M&9&+$>-#OvY_7YEZM9hg3s5<kqJj)N;#xIKwXw
zB+xqODPyowl?^=n$4cKoJksS<k+8Aqij%{Oo1u%U?%bI;fBpvJTwNn`Co?bQ?LYg^
zI~K2bxPR<92q<N0j7Zt+RxK?dA!C+P3}lTwUEXXbP|mWWrv^TJNc-RsZ0EFd#29#7
zE7RVw?|?l0JJ-a}wLu-W-e`zPi)-(<jQf|bfpj;1iY95K0klYT4M$ZaXK~lQhytt;
z&7XQ<+O;ag;BEIj&&7g=HSL6ik*6f21yLBMgh<tbVqH*PY4>i;uwW3&fki0T;&o%b
z(G+lg>GvP5b-2YTn3}9<UsCx@PZ;C#7%b>)ux<ORkGpw;qP?cjsKY$%Dq^R~6ZG%s
zCL}B0riF30p4paB8)P+{>Q2^OlOVvu2@Z4H@#9IZdUJQE#lerB8v<je=RVBZCm}tc
z<CatXS_%^6OiDe>Q|v|FljdJqTuZu+|4?bVyQYarHLzXYTM^S$5!ZObXit5mM4WlD
zU~j(r9BhROaewEERjWkZlu0$ISOTTP(wItN2M@gTUIB;tfFWb2n5wTS=X+W{QVDh|
zUXs_4JVvxyG5%tny!;L<f|_bi9U-z(;4H8$mx(xi;5uP^Zr|1(Lmw&{@m?uF{#VCh
zckJw|mv*2J%R4x=v#M|O$dP4)GLFg%^UlpuQE_4D3q@qlYhOdg=!py)HhRq;`tC3w
zfBL0=!+z`&9}2@(m2Gsr)xE*mSSI*^8}_ar97sDCSTAwx$2Z{mwY3IZ3BggxB&`9h
zbZI@pev^GOF*r|4ZrGimcK7`#e=lO%z+r)Qs@{BTA9-Vw0}}9Az-)T5<$+mYIx^%q
z?dRm52#Hx&Raj%f5o!cPDY4n^ET_6Fzr=0s+4^V!qL6p#RpxTLlb^z21HabWypv;+
z7ZaFR@)s-Z13>(J>1A~6v-^4vH0A^nnfVx?ME|D}o{MJ@svNSvXjNUs-s1OR!!86%
z>_%6|f)?BZM7Q#+%E-p3bG#Srhau=vq-iIXup3o>luJ1M8-W%PCYw-r4}I*E50%X}
zuos1Ac+yHxFG7O`Y`y5`x3yT07DfbIFY3FIZfRIMv4v?(mALevSmAO7s8=-PGfg8}
zQ0E&W9@X&Zm<<*~u!L@CkZLe??hZ5~ZHV_VfLj@&U9b6oK}Mak!bLnMO`pD<bUHBT
zF{{INXM>fDWxZy%>QIRzYD2d%|F9Uf2t_{?zwl1jq*UA#X~I}j^pX`1rmensCOBBp
ztO6tgU>-nSXiq+#9%6cHj{->d)Tvv61c2$iGM^r3&Dx$W{kyYJiqW?<upl#cK{o_t
z90M5{R1n(QgwL;ammJ7?^xw9u@?5+d(1AGwh-KzN;3g?Sx#dA&0*Z>_AT-+{#LrI!
z50S?Goa7WXH%Yc$K@6Wf`6^a!tPuF3f(fTV0bqdi!L~meBFxI*Yct(7T1N(89IL?|
z5DQW)2B@`>Z|L~Q7Cm`FJ%DyD)_Mr{OCGR@#!3?{re>mQI}Za?pJ_T|`+WkN`0!7{
z|IzKa&@CIVcAmevN=VhQ*g#c>M0Y!VUdDFo1(7pA&?#{H?+j-8Bm*HM`&MNE^0B2y
zW_?W%s&?X9Y8$=)Ot@?L(Ss1;0>LlR!!=b;)=zlTINR`ZQX@p4P-n3KW1aRA5g0yu
z?DT$}^?jdgW6f?|x3(;KJt^TwOdan2qW$w{52_^o#6prq`oGb2f6$W`w41;HcodY(
zaeR5(HMLIRGV8K&&c3B4s6Z>vQte-pD3;K4y6_59(<SSwZVwCHUN~t^nbHzNJ2p(y
z$9V!*gZ>K|Af&|1WDQ1xA}M(b6Xt~G6N==$21DMSz;@#FX+wnz?yLj>c#qXzX;6n$
zV5+$#Peh$-0<`*084gn_Hft*v7;e5xT~5x9Lza@Fv}?!HWX-Paw4NTap{AkYudU^J
zome{Lv$3-&=0}Wgy()0+$x5Rg6upifg_B*xAxUGR=+PsLUdoBk23Nn(o?sH0T#-Xk
za>s;k^;e_V`nIREF00u2&rnT7?xDQ<?b5@M4h|$7#QPo%VZa^<Rp8~z@@DUa#_&*O
zJsB~u9WuVFbp;YORLBf1I=jMU>n?7OjwaI#R2MG1MUq2R!&2eB3E>89(tYodF29{1
zgb9^<+s%@y{i#OAV^LmDcz=RX?a#1UMd03gyJIco89hw)2oq$0c)0ei*41l+)i$K=
zeTf`gZ?2`@h~Dp<MA`l9Cj5O4Ud^x`*(DtY22h00o;x>5J~AWjC1J8NCJ$j76~^7o
zfpX)?1-Yz0jB`6T%(?4-X{$jaE!9|CLCL~A_QcIBV%Eq3S}1K3yY7$kBf%JS0tLGD
z&^Y}3uQjB?5&s;Ql#LZlK;pKg%a_O9I+Pjt$R<4s;Sc-7WGhpKk<6EGrJbf|LgIh5
zeCsQ>&t@fFpqwfWuS^YcU%l#OL)@L+^m?P`;)$1s+satZ?X0#-^4sU%aIk{*2t~Qw
zc%c)kvaZ}CyNO~z#Y6Y7vQ6pI&Kvo@%RVkW`cU#?fJH6Fe&t^kX8Tn-KN96{@5Hts
zO}50fGILrs*{649()^Ld??(j&HC*xI59V|fE;QM)KpFjo8|P@L3hGa!eRI%Mb^Ur!
zuP#p3Z&laPgDL^rgN&`RW9wGfl81Zueaqw=SZ=n2cX;GCd6!z1s;!UYI}9qG+zeS_
zZXVpTovAF%RO~K(h`|rp244|R2vr114ciz4I=}M!Qrh8}!7?P^LN;5iV`?fSlwD5V
zYtN1xniasROEzA-_^-M%{@`DO^7-NF@z<-LKX+ikrx~!f4J~Ft81yvyqag0OZD{|I
zzZ+#<1U2*C?)}OUMtctxecDjlu86LXq)ggzZuULL17}+Pm3&vU_36q>K+cAqFnz-g
zjPQMQ=gu3}k=JeQ+>xh7L@@s2xV@KC`@5AT?U|91C+suKDc4Hll9e&FGwTkgC_DE-
z^*SZt{UM%dxsNd`p(`FEHG6Az6B;tl`li&OJ{dDDx*jZ%GMej~poZs_wo$GxLX^6?
zNLg_yDf7(+FuT|(C0*fxH&!k)QcbN^TbuPyrgfE`bfB~v)hWAi=?N086=pj~TeMX7
zQDism7rez@N&&o@EgV%>TN_gqMell~{blElx;0%<pGhfJD9@NU@y^SaW(JN#+`B&`
zt))T??}`#6h@wTMenJWh?r1Y!gUX|4>GzI#W|O5!h@2@6K{R2Wfe<a}c9sPz8d(`5
z=i{_{_ZLnKgTrfRLSBtqn=sYgdCmAHCOQFP0>!ehA-y=1T=HiWNmH+$%^YPP&H;qJ
zcU_J-7rdhDAKlQ=)MOGfT4FXh(H-CKlgK3^<Eqh;b-iEVhw30kT4k1x-OJ4M$_&UO
zDqWs4k1T7SSFkr0G<#a5PSo1*mOt4&s#%>ec*9Pqz4WqO?&&u-ORkdZ3k0-GJV#Zg
zJNhg5;7NNY(pi1K?k`_{UnpzR;(hE`u|W~-Y$Tor50G0KSpEt9B)0hrX#14F@C?<{
zZNbM~JLZ*q@h%-C9oLXo15Z3{j|KB(a9q$0iQX<{#F+48%O#d8ksT5GUXzpZAJz^_
zU(~F<qOkA-)s%_Ls&~6-pAxE^2tf?DA_O5IJ+k$PL$km$2w_^dw-C><ul&)v&=I!I
z1FPp<rN`9PhU)0Y8MctRn;0-N%OWj#m){?LLRy|yC#*aP_3A8g9nsGnWVzvAX8c`Q
zX_%UCwMuGw=9xn9iZvcHI+sd)eruU6`^14anfXaP8gcdh<vR-$lN8}iX8O<9krAU2
zoU!0mNxIlE7u{CZ*tOHwq~QE!Hd%+{x1!Z^k<v{?4s|0{gE=)LA*@B%Ju}<By`Fm5
zxXT3xOl~t>1eyBk^r%air*9i5^RBrWDuHoD6_un>;Pnn8?<3Erd+(9yntu6LWd>$1
zk>fdy_+B7%fdzaM^LStqU=xTnja2p|^`ib_FV}1PGQOTh3IR@;;*%#$&MTm!0F=hM
zc6$!*8m9AaUfIylK*X)+jXgD1r3X|@qo-kV!e;WW$fDKq>d<lSx*BIjZf#sO-(|c}
z)fPG!OT3$htuW^FPc-w=`1*cM!DBzLq*1-+B-V7_skph$aO;U}e_W}*9%~{<6ML&|
z>kkAvj=5$Mcq`)4D_2}ZnY`$gxLl=mZ(ymbj>BJ<&z#vs-=guZ=%r1F^ZrS=@k-at
zjRccNXE&<5JzvSd0&3q%rDCFl4}mi}`_=WGmz>Y9p3<ZAz_;))beGEPQlZOnBB@by
z{CL6Y@6AWZAA-~ym@U@L5zu@;THQCE+02_L8m|qTEF|PL1vQDJoI&h6#$=5ck;3pS
z%M?%w>t3&@hzRes3yHBUJlsplHes09Q2HY(z5C!G&KZ3@%}VMUWpbdk{obmH4~&R5
zXnfPyT$%j|Rkw>TW!){^<Xk^dNy$M-dDVYafl;;j{=kC#G9Z|f8$WJ<bg=T~FR${8
zUH@pD=yUiB?BP{vv9I>3(4@KF*G@3Hun(zt4*NGY3AN(1EY-gZVhj@kn@(_50XO~j
zVIZ{MPkf@J<R!}3=<K<*wr_y;Kjf@Rnkr29%PQPYB&d*{awKbN*UQ^N{I!^BR-QjF
zGCW)u1=YO1U`$!=8dn`=G>h)BTOFTDLDKZ+FS6XDm!im5EWieiPH6l1@eAtUw4sO!
z4}R~ey1UjGx9s~1tOJ2uA}TKJc5*5UG)ru6D4E<<#oD}Uagjvi#3yp}V1>(zqv!i;
z_NPpxkRnUpN&A8sgD@}P)&OuM4*80VHf@nV4k*X`^@n{Z4$J7cFs*C)NFiIHAhXSD
zly>VpyX(S95LiJ&`90y*t*`7KnXKvQnnyhD(gmuEePL^(Xl7*^dDYdwp<eyOoR71u
z(v_3E*+?@}u2ayH1e^I(#u=^4T9_~+=qQd)T^>^!3zv|MG*u?Zt^Xr+Npx4DI<7Pn
z9kEGA2anNxtGwEU;>0o=)U+x1aI%GEG_|A1u)wFT7u8&^8Maze(N0w?j;w`QH&O~z
z@J8L@B3by4P2Wt;_rCO_h_acwpJa8kqNO3z{pMfl<+6vL9@6?OHEL$A<E~x9v#w5d
zst+aXwPt_)0P=|K8^(|v|71qDcGqZiMM)_LU7yS>1=f^Hs8PesN}4y(t?PcQu2uDK
zI5zuws&bCRjUgG;cPhIC*3FuK-LQSc{`m`s$*8~EJ#!9}N+Q+@F}#WmJrsf-G7oHE
z{Yabj!}N*@3JgBy#=qvAPwaV1#HP-lxpl;D?~jLNn?}F5f55b9|B=ZLl*v0|qRt+7
z+_T5NIxVKjE+In}=uB5h{W(yM=q3HsZEL!!_Is_$9btCR=<WHF!N{_%j8epKXNiiM
z&bqsvVGz++i_07w=^SqyA8T9cwAHl4O+w}M&ngA4l`DSzDdA^2F=2zkNXug`{HDLq
zqxYcA>tF+$r^#=?ijbWObDCfU#rO~3vujs0R7o;!<lSEY)V#ek1=?TTjZF7h&I*ST
z?8a<zG2g^Nb$IFOx;Jl9h|ZLiw6lN%#?09^P(y*5$oAA#{hcQwU2knBl`kDVddcq=
zPjyepCmg9u`ZF=wD6lSW%J<i=bcF~x2ReLL=ToKZ?#=XtOD}A(DGeHc2Thm6aJ}AN
zpe@DhJ$lBTBXDO#9OL367AO_(uTtKka(KZo&Lpo7;<jf>8`Kcx;dE@He!|*nnr2$Z
zOjMsGcYo<WczEwokM6e6Mc)+>)Keh-?$cc6I((;hL+Y;s1`JA_QZn{QVQy|DunV=C
z_KFq8>9z)B|D<X-F_K_F|C&WL$^OXs{OUkheN&n1nWqbubeaHz$7bIR<RF?*;yO@7
zq=$Yc>^qQnp`Vh`8D+vouFRJhJzA9>PS*`O-+)_{E+M7!l(yIZ6agP&Tcr7zHKPmq
zHdBC#PNh?~>V0!gd!AfV(vWJ4(rZH2AvM4mVIFCHMf?UQ-Z-Zx*kG*Y-mp%OI}sKw
z6n2715|uGR<QS9mSjZ5VKAq~{HOZ^g%&GC~bJ?c7YgBqyxS2W)N?5n(%l+P*_r3X_
za`e}fe%oapTNb%=bep5IvvNwck%gN<^I8W})60er7Y8<spKqqR#YH{5Xs0>}bcxfl
z?MG~~qrx`cOx$>*P;0=O9SwkU-#;79Sq8W-w1uQZCn@vP-<VV*Os4n%ZZPzSvm1IP
zaqB<M|A}pfm(~KYjKNXy#*G_?4=>rY=^sqV(iW|b6@Y)IlN}URDjM&HNU0e+-x;kp
zaEo1|$2q0`1yfy4N?yF;>c7A?MK;pZEB*!lsyyVBQM~bbP8nh^K1$z-WMNBErd*8c
zI0G-OK=t+DvC7e?`RSJc$dvUYB}#Q#ZLYrqTA3vgV(781V{8eCjp)K*fkv;cJ$04f
zvbMDvh3Xvq+Hxaop-t0fo%&Gv`(X<mc<a`$Zf7(U*_qZ`Ld^(M7gng{6RYAuoAHA@
z#QA$0tSGbcR0>vsDVsg5SI1Yp8Ig4|c8}hOgbkfHv{OZeGuOGIwWVb|kQ>+rfe(Q`
zS;j%I)UZox)@wZqi;ASx_9fR1yXM@e>y|jMzw7@#W)AP!<~uaFGo!z?qScnXpS5Pk
znP>G-XsgtO(}HlLBh1i9Vd@;rN{kKHzn#5PwkPKw(WFllk2DeR9_tet_49U5LTSJG
zP-bU=mEJ3r0(+3^-6bBQ`Y!Aj_fB40(|>=%%FbnT^OxTw&$apO4+2f4U(`TaGXsaM
zI+!P#=o+dcvvhB6IcLPEl3ORHx+fM#wTpW9UKF22GrFWE@G%CL=NlR{%=)e@9sSMF
z`rC{EHp!K^cJMTj&)<dfeOKOUj4|u-Z!#RBtBNd?M$KG3?LMDf2HU)qKY50J=@Z6U
zYJHO~8&|8pIdaPgrCs%*ggiPcI<2NK8j6AzG8pwOl9;lhqNKG!eGgv+4ajv?@9hiO
zF=fT1iNAZNSDu@YrdzI<^3XiRu%gtkr1q4!LqtM^ZJe%weoI<<(o~tiGsmXg7oGBJ
z%9Lx<udS;4`tX~3M{v=Lh35kvKM7i?Chf8NL)P}fCsQ)En>+byL8C6-o%l!N!eO-P
zfEVLv*Z?x!<nasGhtPaXar$lY9;xzIvnvXhoIT!N?V6mDvVNmq7#!2j=_=qV%W8hn
zbCyz8@Sm<(KT%!%<>Jbelxx=lm2J)?tI@Gf&xz_;zS7V9^Ztt0ldl?Civ%`FtJ+_#
z{;UqGMM#F*R#(i|0yK*K`f4!79c1WQwvBza1A_)!9O^nw{(|B!;cK#v=QP5ll_bxg
zCPG)EcOX)TTCl1SY>;s=uVZJEKh4D5LgwwUEZ%MLR-0EU3RfG3B$ml6o!&DhVk8E=
zD&)2m=ink-oU~tN>Y5I_lQ%D35T6#bF8l%<{$pBOY(vJCYbnYIrEc2VicwC!q#ycG
zpCq#uU**}HZfMPJxGapmE-4WYQ@HTh3O+|PC)BRhN31mt0B7cnRIZH@Atc#OofcGT
zHxaIdZ^#*F>O!a30b25J+NLw_$mi{?x1_<#G6kD3_#hf5z*5LW=&cdEs>;j0QPqK*
zwqt(M<X#imab=Tl;XUMM6_rCV{4AnX6_=l`+uL?2rDMW{e1+}ep$+!euk}559iCN^
z-+8-Al}?xQsKCD>6IzxzD293Ru#9?luK)UM5lBG8p;aXTGBBZxf|GH|P}u3NvT)eU
z@$K`B70n_`qd(2_l)KYhDE~UtAl~>R?fIMRT$=stOo7J&7$4)lT_}cOZC!NUnnrFZ
z@PD5LbLo4DIP@U^UvU9scD$~8T)d6t2$cxeD*fq*f18u`w~R`#aUFV<Ats`Q3B|tS
z8=7$#(D_<oc5_$g>A3LbuYTzbf6QUZ#sisq+w)#b-~L8+-|T?*z(irwPi*^Ot0KY6
z0Lc|iEK(@Doqf^Tb%yGVk_&BfymvjUaEu}2&hubsjMnRX1Eb6DadojV+01d5(W-_w
zk^tTC)CGer;L_#NZslRi25k+VdLbV&%-G?>uYA?9`{l{y`Fa@|he;b5iZc`hj5Wth
z9H@5{6K$s27ldD_x)u4!KW2}V0C5`CYwGE277O#xJz3$Z^E|>#@q*p6nMMY5!<QcI
zD0ZDW-poXo2yFRCQSL$3s|Mk%D7h?M_<NY=@ma}Q_Wp)RJL-R$hHF877kG}S-tQHc
zj_%5l+<M*M?cS)sZTnW<tEf<+<p`ii<5qj+N@3`TV?iBYX33jpx!36aKG3)v6oj<G
zol~Ba2MSc_y?giYMsb-xc>MNG>&${U-E2Fgk+EN{HVriu9Xz=F#IA|1TUsLZk!lDW
zt@J7Q6*#xfnOG7qfm$`NaP;?f(=un#tHTox(x}?+W<GY_t}Bfu(HCe*mpn5Tpv}In
zj*gMRS8W_LOKRU$?0GeFzh~rxp5cMLUMKu7IsY1Rrp@T-&9BRh<tZm?bBGG@Rf0*a
zL1K^tvTLRyznQ-+r^CN)?v$@Na~kDucjy2d(54l7grwDwXv`sz&JLhsqxw>(nHgF5
zrLOttU}5nZ5fL*-uo4#ZLLh2*Sh*x$N?9<B#52$3ZF$cDarVN?^uLXRj<R59s2(=}
zX7xQSH>0`2l?kTeA3PFbM7NAcz&CwqQsLpq-!GNka{r!bol$MJF6n?9TEgUH<)8s(
zvdWp@+{d*(g$4auvKZx$!lm4?W}s`kv`XYkT@QJmd3eY1acw)@7h8?xu8Vo={C?yf
z{V&T_d7qy(<YMZUNd~d{0$Ci?+A;Yt1?1{99cxI5P&(hf2K3jYADey|(Mp1L5R{B+
zz4F_))wYJLYnl5Ra>CZ>Um<Ol*1nli#!^i`hookU!6G1IAcr{rhHf(pn72nf*>@wy
zi;s|$lnX3BX1G~C%#V^_<Av!L&z_w{kDI=tM9zkv+4EJkL(Fd22XUkEXXteOe&IQ@
zl}5SiVhKFDz?_Q3zEot5CD``e#U;INT3TLhWoDitqo#Z)5z)9|*C%OgPhmadVK9E>
zaKLmv+i@0&56b=DTz~_$fB4`#dF-_1tgk^gly6LY);IpI(C;&FC8L!(rh4_{I4Eyy
zER2-=T;IYSUEOI+BR+3gF?D+Q)<cX?ia$O6A|+SLbY^#M7~>P0Wf!|-kwR?8KUe^c
zn};9sJq$3)uaA&fQ*X(Z2%AoYw0z!BjKsB_pZ&C@kzw)4IuC~=ZQinlvtIsA;d|l>
z;Z1JVZvXsRxO}7iL6a?7^81wxv0CPFmc8P+(Hss(T$a67Qn8qGtJ*Pf+dzN2NntrT
zNh|){ohJmkyu$L4qN2Xf1NOo~>h3yrO)@;I%mYHd=&j{>8zYs^>#+|34r06aYHHr0
z3$*&9Ozin-M{FWvA;ON&DN$e~I!Sf<Wjy)aYmZCaO&JCKQ2*OxS=re8uQGg(ggjGr
zdv4S3z=5w2rzCb+QnAmBbfsOk%d8$u5_*o#ye(mrnA2KxA?E1Kdo~feEbR`HNCyGE
zWv~7Rm09-i>Ti?-qko-IaJd#-<c4+0TMqN$_NYxfv%HVtvCwgdmRC*Q*45E%(MT_&
z-AO`{V0e?6=$1y=6!*4n3$GeQx6!2^WTAC<ZxN=FVncTg8a7Ojtd)z<BAa?SusTL>
z`V6B@uAL2Mce!8w_3A){KXKvYa1H;~3IyzYd4dNPinTrcX9uh<d%83AwX3$PBw}p6
zUbBi`G3VcJPx+vhW?Qi;Lp!pmP%<}SftmJ!dd;trp-r@5!VV|<;ujbD?MC&j$}}6j
zt${hbhu{B_co~q^@ec;Q(0YGSe2;N+ve4pS0UOF4gz@0f_toOfV>&W(|Jm)rbaSN_
z9hS7SXr@&;8vKX5LpPosapKb~2%K`an+1b)FO_S_$!!1`JIbR=XnCu=TmYXc5{m8E
zDZe{3^9g#A_Dbg;+mkwD30NUoaw})`G|4&JPMU-iH;l@&x<+SO#NuA*ldr)7)V+(%
zZF|+H){*2O-*bEHDNz-bHx0j%CAZJ>J#Jp(ZmCH1Fso)^^yOY1D9>x^WgJ>y(v_P(
zCuQ(G1y|YMKh|n?09<e(>c_}dm_H)ycup7ipLrvJqZ4C-VL#Q{1P0e^4{?i}7Ybb{
zHZC^cbkn$8WSG(84eXQV0tB{nt)@q_JCfc_^~8!hD%i87M~+1=$4L^Umdasp*U7i#
z>|-m?S;#2j2?D8O{)eL6+-c--Ek{UbW$QC=5S+wp?XRilpU`0U%yFZnE^+s3xU!g<
z_e}sz!>OOPiG9~r2{ep{u4qNj<*XyuLnOP)N86L`rbx4x{qI%bD@T@0zX@~0GKj;|
zP<=ZGR@ktY>@pyb*BMR-qL}6*G)dXKupb*3S!mn0u5&}k0+&_BZ-m}L&Uh1<D*x-(
zu7Sxs=lQ~4Zii*aKH+o%_&e_I5tqy`LQ7AfDS1y^jPBY2y0p+!$Z&Dy)^q+{;3Tfn
z2{lzjLVCFE9qwL|-)2K+*D(&oAs@Jv5r5=w(2BCZwQHAX9q%1dwYKwNE3N&iS`*9F
zn;aZM@T==PKlgIvsO|nr7WPd?d3yAEQ`$$M)(Ly6xA6p3a;$q|nPc;;s-P<48xu23
z@vM!Tcc-i@m6lgpa9Wyj#RPX?zp$toWwQvEEiJ0zwci=hnrnMh|LBSny*36B<6LHA
zt(d+fcDz8tw2OU1LKeHEUf^LX2(JwmNU44W56|zjCp>VoeRm0>eqFbz@vB1uec?b2
zNVnJ}$Yqmy=y7lT8OUBV_zPL3=#9KJ+S+=zVnI4H6pzQkd}TOKE6WV`A0f{MJFUL4
zMg&rnPyxs3Icw#E{VWUlhbG#Q=|5=ummyn=6p{JU&VS04jk<Immd&I^i+Yieazg4o
z|7US`e63?fgJK$r-sRq{$7E{Xxdp7QV@eumgz(aTbjByMjutIqxNt&XrA6xX>!(ki
z6syc2Iezn)-W5w_?TQtT3JS6uYmd)-bpEjw%m8=Hp?@89yznhd?+nYN(7&2wMOM+=
zLRw3+{0Z|Vjva$`eW>j+cb5C`Ve{8rVH)ln6_vFn6%MIl(0}4Tk5u#&@H$DQRh>2c
z<%hePK5NUX+dVrV9ymD6!<Nl=pmsi<(N74J!A?D$HnsD~d`AWA0o8pPk`q~f6K}2B
z`6iPVn=$yetFf#eXUEt<QgJI+tk6k3Dh!ZOE0b|Gh5YQOrp+O}RKuSMcEiTZUfMNP
zc9svf0vk3DfNkM)4Gj;<-aIFv8(jSlon_e~Sd{lz(rIe9Kt5r1)aCSjPii~U^s_+7
zt<d#8eMCBV**$rK%=+Dq8Q-8>-aT*N?S2plJEdt(g*K3a$e=-<E@M@F&ga)g1hh`r
zEftkSZ^t{^%Q3@rXT9lo^vgPWlTVq4mUP_2qUp^Odoy}CHeU`zue52u3H2!2>?t1G
z-$QoCT;y`|YsnL1a~fe6n_No;Er|Bo4VW4B+4MK3m#(MwN$QRtt}H~P-w_4cdQ>Xc
z?uH4kw`lpAlHk*@BWA&HA(>IAm&qL}CxzYiik9e<_M4zVGR=*^nGM61fZJ3TXNY#p
z3s^lDc+b*+l+bYeLxs$jdSqZt^9iux*<WKph9OMI92}wFS{-PAaQ_Ii(MouBfBo?+
z{GOVOEU{{$b)EU%My`p1O5D)8Q%w<GlJbe;t1drmv;(3GbD@W<JLYZkR>XUy`qszQ
zJ^=UFdi?F84AS+BRHCx$<(ZGf_v~MDTrKh2L^P1*1!{>PG{=?NAWg#76vMq3pz7Q?
zN4l034vvTb5EnX>JVZ2XINk}C>|Rv8aJWDf=>W7KE=)=A0aj?&R8e1EvDF?rf*aJ4
zlP3rNvgnVzTF2UxCpnmzL%ow~ify-c-8v;%Ss0n$ki-l0&`cShhPR)!9@o6<+dJVf
z(_FEiB^(6%`S}F}2@F}-3^Lhc857XL<9w^!;%Z8MEv7dZu^^Fe+k?>;*9EIR%XYif
zr2d3iv=HXf?kJhxwyucvYkvYH?*VI)cJu*mh?h(|jpq$547zD7#5LLMnWJ0U#;E&@
zbT{}~xUkhaXK9k!TEz=-ihoTGZ>%YUPVZT^V3j_-TK3iP9}gWh5znmWx=w!T^)qUT
z^aD(CS769+qQAKbP4su5)8`C$JA4cPU9;ry1Fo(%J?qPb$KTX2)06EdQS4PZ(?M$b
zg?yN+rchFAYp_r`g~cOqvB0UINwRJoVY3oJPlbPA`^S$XzPD>VHr;DJ`IVQ{qJ!2B
z4oPZGybEFicS}<xgJ0%g5NkE#Vf$*V2yDN}p(5KXB<<*KK3!fhsCL-xq$GjCE|@@a
zVqzCSoQLc|Q|&dPlJ!F;0mNtSdh?gbklehQcd|<cZk@e>V~?^j_UVUd_+Ke2cWU%~
z@!8ztN`mNXp=euF<6AQox-GzX51&j>Ldm4K@%odB3IkWPbT5@vx*-R|+Vz2@Di|kE
zV8qk>`5DMFz+^dBp;QqEjB-bUf9S^DZ*aYAsyNzR`y=?ajqolMAwWNSwqalPIqTUJ
zxHzeY)yl2`RhSegFR3nUV0<FH9w^VwFRWRx=Y90n741NHN6xQKmz-^N%zwoF+=OB=
z;X`?_w;HDOQ=olq`SilL{`I@*q1+X_bw6Wv?lLK(-ZKyVh9qs1ssh><sDS&X;}m(y
z|0oW{Wl{z(?Z>O{*Pm~=INwR^WyFe=UuP;BR2!{WUlwPiR8peK)1uwG^mCuCLFi&n
z<XSUY&8s_ZqFjZc6PE|D`LtQr@8U(7%$*_AO5g3i>(Gv`Wcpyyv&|t7J-xhaT?0)P
zEecXUVCZr<L-b2`_rU3wWLTBl9tUGf$<?DD6nr0Bk*ipMVK&I^_}%p~#8|>jy3r-(
zSbyC!hX1cR7Zeb-Snr*1cCv2nzD!hcrFFH;;n&QW^L+Hizq~5Md2DBm=M3KVNm!pp
zJv|${q9#w6@W+MGdGdSBYhHOR9eqN-wy|-Hv~<$oZjL*8TPEUKVp$=!lX`whdJM2P
zIsVQu*B3Ap3?^%@C(R<1>GWQ<D-tNX8jkN)w2We$xEgZ<v@fBEh`&Dn+#YT@wqw2+
z?xATGVMfBb7xsRwjV-Ip@e2{Yt7j2}3`^RtVFm${g!^V;<Jl1S<G#5!jeTub8fr`y
zOQ|C0;dp6Ld3gW6b*AP(YGs?TQ4YsN*8DH_-a4wvcIzJ2$3hfAz#uJ3N<u_XKtx38
zkS<9H5$O`K6$Aw77L-m22?@oZr8f!)N{57`^l#oi?|Z&~zdye5jqw}ryT=*hoFlOJ
zz3=<FuC>;jbIlc8g%TFgY-d<{D+?c=a_ci=@7-VeuROlaMYO3rDtLc&z;2LoMdV!q
zCW|ivREd0Z#l_B!I6Q842Bslt11WxGzD`{ozyQvSR=eX(SWH;Hnqh)|l|4PTp<KW*
zaBB2a4_E_eM@adl59Qg?QR+h{I4`RTLUInO8H=%0$f18BbXwKjN`ccFrcD@>y4rzq
zDUb+#{J07`1)@3K?@`xsU?U>s^dnZD4RoZyq{a>KzYG8=HXm??DK?BMQ0?5jd9&%Z
zIWP{;T%d5+7!R4M!Y>k@yaRa5PTzK>qkj_mE>AtZZZs#Qt;lS<KOBuL9t)Y6c3X8K
z0IgoX0lY)QM!?~njTu0Rp?X4<e{yZ7Dn39f=c*)r>P-Rs<NWe%NuD;Gt|&5<A_SyY
z&fvF02-gpgRokD%oD9D?KZ)wi^3YXP1I%4n0IY>ge8Nu94~HR9fIzz##(N_8+f->{
z1qJ*kJi!`9PF6spco6^sF9i@+!c1c9^aVAdkjCGmnzL?^af2i5o<rgpW}-$oU(QQO
zIX&BCD)0>eb#U)tatL^)a8DC-mpRLTqe7Dwf!+ph;XP<6MVV)*eX#8bipZA>JRLZL
z<+?q!PQWqJ8#MXm1Hez_QHVhbz+*RQ6~v}MpiH=D-vL&<I7OcV8Visg(Fp8)EITtJ
z1KCt5P&fd{sk}<sNAI7Z1g|au3uCB8<$k#z2d4&Mk~L11v@IgiLOAx3WrK|BSl{~s
ztPoB!{8ZCBj@H*8KC;E}T`(dhMzj(;AWj1cTznTW3d((lq$9M3jer+Z7j7PWwo_tw
z%f)8YOprkU4!DN90;~xUH39=Xks4!^?rA77C&UCXio+cl+8UJgSgC~E+1<V5?OP9I
zCzyXtsF7FvZ=*~>LkdI#@La7TqAFTjwN`a|uR(|M0p9?UP`uFY1@9hUgLuJ10_M8k
zFrTWci)g1Llya8)zclHiW<(EYKCT9(J{HPZuk~&ciG<uAu@J;IQm3}paq(98Ca6X?
z(5{0CUULXCXz)Z&3!acmkJ~k9Axl`ep;>}n`x68KQ0lrsY71=yN)|{D)Kiq-qpJbS
z8T$^IJK^=<*Rr#znLX~709cIn=Qs)o-17+~hH&&jEWnUM69<eFI4H8oD5&5KF!mN<
z9}FZ$Lg5OrQK2V7=!e8~&+gs9GePm3TwKJC!Gx~gQxzL%-NOvV12C9~gDH1Jgf`YX
zP9%5{6hVW4w$0Xc5pA|N73u(xu)j@jcImviUX1hs;q{`lGzM%9;n$Jd+@2dzL5T>~
zAFLLk_(Sx=ahZp30I6{E8kZg<3&=oy5q*)#;UFnTvW>hCBG|mVJh=P--C4oHsPOXo
zKI}@LUWs~*D7yARZLR9jlKc`MO0?!+&Ik~pvsei5HbxT}$_XGgphaQ6+zLd}LHa6@
z4O@#u$S*)9e5sdumkPdAHW~r*8pY4x*dUgJL0Kf&IGjXiVF79c9END5#+6!jWW^YG
zU2nKO%(G?#CuD8h)%DTjIK?>eHH2&6dJr80<VR5p<5tZ@O%tT~07i+)f1rPowPMdZ
zc8yS@f5?MqH%OSJ0xJ0Dlahi01S*LyUz%WRKoR>5=O;@3Et8$w@vc>PJb*zRk<vh!
zgT@=APN?#5{|I&!Jvq%TZUVZxT_9WBYJ-mx7(Q~dec8omeFkiZWK_Q|*>M|sMzMzn
zp`gSt!(Qy5eG-B`Z+(%|78bgIsRn&4?IOsxmy(je_f`Q3)e=2iplT1kVM*aYs~_H#
z!I@@meqPLepc;Gr8Im&KQAlh-wV|yLje<obC8Sg}89T#C1m^eS4I$5A|Bt+NeWiB;
z+j0)q2(h{sGERs-6czm;xQ4j~nv>vlHGv64LKaQuJk%?{NMW;Lz>JyJhm6cjtU_FF
zBP<D_&_tcKg2SFjm}O-#ABK~g+Y!y)c6OGi$YHSyz6f91DZBgIH#g+^V8BkJr4Ub*
zwy+sl9oD0Vt1A`*8y{a2v@{=$(k^BzI}XWMo+{D=VV#HOZ7wM8P+;L9Abz9E0qrk%
zf3!SZ2kVKIySzAM+Y(3~iC!BbJxpW8x@Pqs+a_y1^xLh;&uuteR(>%xJBP8PF^;Dd
z>W%jD35;XovhzRUt9Uy1V%_%lRQs5jvli%aELrO)rk}+|@$IDP$UPczS-bL9`^><;
zVu}!U_XhkqJd)ZgGXR~2zg5X*ls=cV%&VoXjV1A=dxFyh(`VA08=(|@Tg<@v5cH?^
z`Np87v_VSvdz{mR(U`LCtyd83W3LWbHKwNm`@*tLV}*m+GPL9M?Xd!~McaFEAXD_D
z7w%sA`S?(qEWE$0^YXX8<ErfH(k*VyszfH!_PO(28zgK<P1WVZmq%VAYQUw)ELdka
z9^TRJJT}@F2Zx8#dF0S)Ckiymp<u`0PPmqh{KPHN?AmOCOL(Lpe$NQ*LE$kmnXB2E
znG-<1(^j^sazNLRzU~5igpej`ny}C?FnGYjr~NsW#OPaVo=qdxnAB(G5DoU+DC=`+
zWZR@Wl6s-*kH2YZqlX2Gz?V9?rZvy07awx%K=BEemlPcV0OQmho3}a=9+K|xIJnno
zV2Pv+znxdyIZLiOuG$zWRELf+J-_2G0a&UF&!U>zH%YI-%cEC>;fGpFhS*rpnMAx6
z+hnj8k#ueP1EsngP94GzGi#yrUYjT-7~Uq&BXS%F6$3*<oEr!!G81pdLMQ>25zrS;
zs$qrr{r;=mL<?}-5vfPS2!#S%Eh~sb;;~#Tkoyit=AS;y9L=yB!iovmSi_^({nfjm
z-uvm(E~H^rH6U&|26HDj8JH2s-dN-&<g(b!0wx{?$5H5j#d|(efm6G@qN0MXOYLH+
z^E-5p<K!^S(ieY~k1!5ur~Ns;&1iYh_7_Ym>Hpm6c_3FQMUULFmh~Mp7(`=*{=(%4
z^zFP(h~GFmY(9U!lmzGys=uKd+wN!v!-&z|KDSu+w-54w=$WaCu&@q5a}8ZsNd>dh
zIOv3}+Nyp10_!2-MBu+Rsqvzl@O}=a15S8>GqDhC<Inw;n>Gc@;Y1Cr8yvbJ9l_om
z!F298|9|Xu_w@H)wZmODo@6^`o3^6=&!)rtH>AwIFxZy94`AK&F5UO9lNKyT?l=-|
zMSowT|M6#T`Hw5>tN$22U;2OCM8G@$`xl2nivIh*&WZl@I>Z0@ANhvY8@_J}CgEsk
z_zu6p7jf_=Xr94QDA#%fc3!m!9?C=&=!TX_J{KC?BfDrAZvi62n!qiSfBGM9m;d)q
zV@<v85i$XE4t95E;!TF&PbQ4<XYJ1+PKg!sh)frQ0gZhdx|6#V?kHvq*cTUf5f4ir
zCe5N}&ZLJDu7^mjh1DGJuP=oEW5fBsPoKC=oBnsf|Gf?WH|%k)V!a!ar*OGHydZ?y
zi`-z;0x|}u@vLzC6R%Kmq_rax6E+!yRpN_aV#DIKdI%X~7ldn*!>AGPvJ@q9Ya~bv
zFZJ<rzCC|;g;pq}NpQA1m`?bpp?!=lo%pn1=D)V0X1}MW(eN1c+KK^25wFu=7k1}g
z(>*OllZO`Qmc2Si9FuZ?&(f=Q+bNl^+u9l~yQ6TEHH1ynCf}!j-7Nb~pE<J#WG6TT
zDM}z=maldA?*S`!W%+nIGDi4Jl&k!;5k5wQgiSdYP&_#}JM$_~{2IXfT#fj9m~#h~
zm93#|j-nielr!vRE?$S-+9uzB&6v-JnBUO_5{Th0zF!aG-w*Ek`!ybNDkOdWYtQ`j
zUwh_wxCdj%tNFW!iP7V5_S_Wo?}zF=qHy3^GoZr15ULpbR~_L=48UOAV_y*A56MmU
z_Z<geT|^HbK!Y&C%o;9_7YP61bEE%$wUeGZ-cryBU_{OP3xvz^fB$<vk~LQkkMY~$
z|GpRF-}g#v6*G2lahU>leVGAY{f_Ft7vl*06d%(vSdX492=vqn3HRm1e{HWvL42){
zbZtlN+5Ua6L&STXqkAFOq)~W7A56+l;%?q)`7g6(!YT&yGAD*d#4`z(QS%FgPqp%P
zkaX~AQCJa?Cw#CuX#U>i+U*l|MCuJ+09I?l?0WabzrTHI_unri*{ZUgc!a_G{zgW#
z`>3-);klcMlm6QmK$;Ru2`JO$Zf5-X!M~QKFY234l4Su*0HPE0h6n%7NA>rjBbvcM
zLt14d^7pIMQt9$kp~3?*eL`CezovZj#EERcSMAgo{RD8q6a428;mN$I4SoHNh&J?;
zbwX!uuNFO+K#U?@wfFDb*~AmUFiSi$;`P57z~9^O|4-86pIOX+k5lGEWe>v296A-E
z%7*tj`Z!_og&EewoKGZ|0BZe!w)psnB5A-e(C}Rge|U#DGlI2aV`BlZyTMe$ij;Bu
z>=@{;fdO>tWTL=TQ?UomgA#NVU=DDLmN)s)&?f>{@k5gPuQU+u9f!{*1ASpvjBa{e
z2}$W3O7A&koSy(n@dChswpG-dC==1P3O*Pe9uQMu*X{NqT<eBdJAMIm7`nlPi+Td4
zfB?Xdn(Md=L?WSwMNY&DT`()iiB*_9!wokoG*qLnt_uM7cWZGBa@jofLXX!AEhfMX
zVY<K-xf4fdxIVNUp<i|ri=V>;oXjnZ<NILG_z?Qt$d${%bZFmlFM9+0R~AeMBNPUs
zvEsM@qx@mJU&FYpT51W23W&VXI9=aOaaTG8SSkQlfJeANrju{*D<eLC?;(M3!_l`0
zaiW;vpSl+4W+6wx5HSbBJ^CCwv4LM)heW;>q*uWMJ7CY46S@f35VWxXNl}FpD-Z=>
ziHexd0!0JdnN;8Gr?&1XT>^TGXJIw@i1@g3b$S{k;@}Q7f!;x9nFsm<2q^+X>(Knw
z3q|k9h=s^_)u@!yS`i>UI6M)B3BWSIt+bxUd~<ITE|#?HTDrP#i;9SuTsSBo3^bCL
z_YIBPT&#B+<6K)yiNk2wNP;-JhT!^82o7B>*$mYL;IWc6wA##q-}ahomMMuwNq^R~
zIsnb)`k3hp1so3#?5OxeL<nzGwEqG5vRz$UoA~uhcCmiF3ceAT`@6Be2Bj$^u~?f}
zCWMU=%q}uA8%`w?;tE(KqePqq1fafyo$!90E91MtoctMJ10Nqi&c{0LIoApQXYS2^
zAMR}aZvQhF#|1w>yF=(iw8p)`JL|a1lH<v)qF06!fUI#0AmYk8=!<A&;EorB3UyQk
zD-oO&XNChh2@{49)O0;e0A;CPxx!L<6Yb5S<rmCr`e|{~oYz;~6tdfaAo{J(;{Loi
zK}0U)H`=<o67ayyf2*pcr3DwNe(`QVi<qlWFc<g$)b`-SL}r<ef?G5MCWH{MPYO!F
zoyUp+1YR>WB~$8wd&I(LcsIz+sbJqG_NSYEKC8$UgrNZN2efq0BznL-h>b2U>A_re
zn{a4uDZmaSY>OXbGcX)FWc#ry628k=T;NLQOVaRDP)~q+nihY8Ck3D-0C<?X(Gl2o
zuG!Ph>TH0iz=-8sXAtEHx~C2O^&Cb&(M1U38rkdKtd|rNV&H|4^b#ZQ19JI^1@Sjt
zh9XcqNA9udE&u+6@+5F5bhMFjteF@Y2^X4X?L-|{cXywQCJ8XjJ_9$tf=X=LC#TFI
zc#9mfL@NfS_~q#3b3SzfB&jfJ1ahvvs1M=mocBm7K1HsBXzd0@$P82V44L0EGo3|t
zCYDxE0gqut0si)FXNd>CmmY#<K~@0)Ro5XjubosEfKJKC2)6RTX63$}^=1;WzmVVE
z-259Aaaa-Y1xo(TZ3xW*a*{9zrdv@#L5TDcUCnSak(_J7cL#L&YjV;6en0?iZ^H))
z{Ow+`){YL%x_2jeAgBX@XaK1xlrt#FEh8Uv5h6>)+ys7OAX8<fr4I%H>n5~{;53W@
z_vP}oVGN!bk$VVOtu_?bFd|deY5(_h;g*CsxP{>$0I|>RWN1Nz9bvP~x3@0YeE)!W
zFo#NSWC@NC36w`l;_lz$Q-(aq0B&L<HI=DJ5Wrw#Q^ksyeBNyYh1SRs9irJ~-u+Yk
z+qnDZ0oDqYy#kv|8Xv`n9z3ujf`ctL6MHLR@FHVlI{+m>t7LU*YHIBA;nOpNas@Mh
z+pJ`70py$O+|!3HO=42KjuBGeq?&LLJgQtZ)zyDM??1Taqo}<vCc_<>FWMAiBCRij
zU_8l`rdOOB%!5Rr>2Eri<A}s0VG!C(ev%X{rS$WOo&^Bk36Fu+;@Gh<D)#eJKzMQT
z8o~9T*LP<f;pg9YfMVrBp)tDLh~;ACo#e?}OH$k;%ch@(B=p_i0{2d0U3E1TkBgI&
z5J{>HQp5aJ)Xq9t1VL<2;0?Al+@d(?y<x(!3RrkSOLm-FlYv6`>bp(AZVMdX@8<hf
zA=4DwI%%{C5Yema$fEBU5NX++MEIrNwQvJM2W`6V3>BEo13Vr1+4mM+Mq1GT?kq+B
z{zyF3%FXr)_bV^BI30waT*3UtvPVLY`UY;%+}zyrE-SZ_HgiXUYEIe|;htRG`R{wv
zFfFkL)nHEo1GWu*+?3xJwP2(X=|5I}TMt9dRa62;;$dXS#HI2TB-BmgY%oOjoLRtV
zAyRmFAhfwA4I;m~(jP@dr;qC*%Dp&1979(}f$F7<28`S}vBe^iI^9!ulY)#;J6mvE
zT=pm;@oWDO4m4b00KB1A)-7+;4G$PT^<Y+ZYliNwYi4Hc!1bG@aAx9IrX~{Lb1p>8
zde<TiIt5z&xh5RO<G2&ck1r`I?&s++C{DuN5EN<-ya5P43l{0}QqJ1WJF3uG4s+3`
z%;bCakc1l${Q{_oGdjl+B4RL4)MBRA)^8!~1K*%3{t>6CbR98melQKqkYnmmso*Rw
zo>t0(8!9*@Vmv37b^9LJF<wWDe;SES4QVY@k%^-t*yoAFjF>y+#H~F?2O-)RGy^`&
z&|J0><fHEZI0y`#nrW6flp2CHr!%(-Wc>>IZReD6)FbvD<%Z;gz|na9_(hBL2s0|9
zB|!etJF1xAk_Ny%$u=wywgE&n6uF&U+;#e#V-GJrbntJ+Z*~p|4Z&105hLNCT_B;0
znnwr*(RN@9px=SJg`Ja=`WcueX#X@P6NrHLzzN&4(yRkt?lpQ@ns-1q)ato*`eheP
zIQWX@VAp~7ddMDDT$6^=1niZskIxvyb4d<_p`AiAa{4<D|NZz6U_v3JNguU1ckUem
zHGCvQ@V=lrVdaK4I5$^UaB05K{2+MasKucig#&nG5XKrnm(oY#p(K0!-~k##h}#GK
z7j}9ihHQXF!~MU|NO1a_8^34)4|2wVU2q=N4Jej^O6WYT>GxTPU@2ojrw-f$;77o%
zSH#E0W{&V)Ih^1K10FU^X1OD9F6$c(?1t~6xbO{2$n3!fhW0%5Csd&s-F0<bI8TFx
zHemmSOg6#q*^?(!v2g(bFGg#S`gx0ge6W)yzd{yAMtfo^(OdM}?3uUs@%E=j!w8>`
z1R=;EUj+-Lmq1^hR71?4px#AB*3@Eabc;y+1QD@FINU(os9jk{o=N?ZccP`W6@8q?
z!|MFnSw5hv<-A;dvhtNHL{r-37DjK3se$s~-319E?m~ls9Z~xnhK(Qv!R?19%mDxQ
z*%Q+$4O1lWmGW(TWwiVDSw?!KJAy|<Qg*48c_*W&V+t;MaFlNIraLo#=j;UY100ya
z=nL53;SkzgMDSHG-)|*gXS91<>7}4UtIfpl@Bv1~>*qY=2~{<Aj*cRS!x<5a86-c@
z@9#dycUVxcwcCa`Z1??5fjsbTSRGIC<cYp692Tv=(=d4iE4U+7?HW#A9mOuhmcqhC
zBp#sj<T3=09Qg^s3?YC5$KiDS(N>6eT)_+}5E^X7=+j5cJbZ~68sN3DdaTj+iLc-A
z9rtw8+SkapM9H^Z_R=L{qNaIcEUp<X2g68^g<;r&C4qti{0HPeHTCtkmgh#n9zq`F
zjye*{al~L2TG}^oit+N=z}5#3OoB=r1S`l4oI^-}v0TY1C<s?ikWbLMC@i0($OE;H
z_wZ)!M3#WV^5aBvm>w^zHt~XP5E-|UoSX+THEc_qEyw=W83(C|$H&LxA(m5tsR+dD
z#fb$?0yxEsdmC{BiP!%PK-FKg1lRgM94;qL{J*dN?*WIe^tTeB;7(6Z-y4PaD$gS%
zB!uoqEeyo&Lc7Rl7pgCAJ!BQ|F3#yLNht8k^P8L4%pFjwye=@2Xb^TVwhGGUS<u;n
zf9$cU${-WGg<&9-_F#6wnalPf+r9b|Y5fA!DCBiak*8gKe+QI{-Pu;_5wK|5KvTvG
zkBfi5-~aG6`DW<GrrL1GIx=(%6@O0wTfp?|Po)!X?U?NAg_;e8g0=woM#PsWjIhWj
zdIXWGp!^D9g;*A4?@5!OP2R*fUX(o0si7Ld;r_BXYPR+p%5?O5ywJ%-UW{`q3WSI3
zh2eNTE3M3*9b)AgTxXeF+|@p^<>gmhy_>ybdjXj$*Xr-SX9o@$<Oh(~k<XkS8HWD_
zQdqUs6_gJ{HuYR1(5(}ZEM%jgun;Qy<fX{_0xxiiLg52#4bIC;^77oAoHNtYDai7H
z*+4L83|MY~T;T$57#wC{to{e@3Z4|Y-hWM7@l6N^cIe4jti3@K{wtLXVZ|r!+Pxcv
zT)L@Ejw$F!5P^tv8l`H$1yRKU@3e*Z7N{X|w`-HzE6H%u?~B>d-WVEi5qeXgMX<LG
zW<@^)R0TpNL06Z+!eUtaK5A+-Lg9s(wVz;B06|(Z%8y5^#c1ETfLH^W>IS|jazkzm
zyAc-d`1Y+Bbrl-Q(HvwA@imL}&X~vcHLvfzG_;W@sPsr(tZxZ#nRc9^6%UNjq}b51
zn!_D@`lTPr4=%I0JW%Jz*`f-??zll~=!_I1jTKE*NTq4TwC6iyLf{R8{>S9eBbZ8c
z7R748`A34`ZRk@%q>7(XN1%;?hslBivMF5Y+IN>;qP^DD5$OhaPgqtoqY}e<$f&zd
zUv{1bK}ipUH3(AauGA)^QU9uLno$a7*DAp@7cDx?QV_UzT<L<shnUX{Y}Bj<VtZ7`
z8JUKo>c;?Jf&X8IstzT+rD#=Gmj_gMQ2rF==04B0IamA}SCRS%8u#mPNQ}H4j|qw|
zmcwK%5d%?HH_cIj;9Y~kOFKCvWkk=wc(1|8N5`1Q0GJR<fgT763v0SM;$6{@79rt9
zm_<6S=3Sp7S771gMP1_tn?Quvy<+b>J4?{(3Yixrnhy(8BYU!;2+DvFe5-o4E84hj
znwaxrboNR9GtE!-{d3W%*{vHItl+zqD1Yj^4mzl;{VFu6+Jk&#_RS7+LRj9up`)t{
z%k$(&MUAz26(^W90U}p0^o8v+dA9NC`<Fj`{P?6?tyAdwSLKzRsCg-N@AmcfZnpP7
zT8K=u3wXrZYFou~0jGy-xe)NcN&o^|?G;N1m<anSxNo>FPU?T>Qj(PgKBs4#;Odoy
z7Kzq}sBdZ5<h3w_Iv0-W?nu_L_1o5(>ms0w^F)HL^ZfKf<P*;h@)I+aVJl(*^~p&a
zC{?tTn>K_kjUl$fk_Z{H^Lz4T3+?luASf5&by@(95}t3O643ghAyex%-Wrxtgel!z
z8nzC(ttkq>mWr~b5N}aRJQd=)(03l{faD1Jwm&#O(3aQK(4f#qt`HA5a%f|)s*r!Q
z+3P^0LI`DI#m->JGJbQ!U<3ajeG2Gd;64lgpUlBZ&C=P!j6CY}3vU)KW>D~jY97@)
z(MqJ(lN;>I1=I$63C$d&1XeYer?dOuNzSmkYYmZRrOxlEze9!qJhC%SFcAxLWrdjM
zCaWFq%?;l-lpCy%>|{G7j*$7GuT$-yW&%(k7>lHYa%<(BlY5|9&I8Y8bOF$P#<VnN
zNcX<wn&Eye!D<B0iqOR#I&^4o2_P8|V$Br>3sK0TYtLZtwo;;-a(}|?riW+=P4oN_
z2V7|Led|~><)byWzhGCW1+)rKj<o;&0_EE$9BCF6?AT;#Xd_|5NCq5z1&{60=Ely#
zfReEOJn7rIEoR4yTL5HgxBsLq#@?gTXvY}?WI%sx|NHY8Be(!XC9uyD!{;F(M0u)k
zEChN*Z*GWHu^p1qy#PO8`w0D+fm`I<uWxDv1Hgx@1e*-^VvVdpNff-&GKYy8NLs-V
zLHM+EW~!8iii!_#33~qshf3(BK*#yC0>1#@16`!htPmrI31MiwzI3gf8eM`AF41dr
z;TuRl{t|zy2*(r*mmM2l(T43ty6zOjAEaSHZeL6|!U8=y`CW?W0kf$rGQjZi!6@lQ
z&22AkLUV^1Rw<z{2wqxQVbJ@Di{mvcuXmRpYyoTmJhh$N#>NIn3}&05b@u>0eTjJk
zeZgKDq`|oT#L$tg`y$V4X??{(^%rA8Nc@E$0rrj<0tz#CWawCVj~+Z2Y43+PU&Gsa
zGI#_!H*Ae?s~P|un5CdoYgHEtaJKy`VS20{C7(=;g#$FYlZ;H+*SD^rfr6ae2X<m7
zE%)3>!|7mdW|kUC)P{gWAGL=D2EImfT6?MkhcG%Bak<LM$}cpKkI&=Ly+Q2^Z$<=P
z2t+l{{3LGpV3;8r0O~Pv1yk(Ogt`*-EIl`<j9~gUJvWz+Dg?OmlP7;WhiSG2V-2nX
zR}Rh%bY!j#dVAbOW>O{YMszUG=z*^}90PXX-4@o~pd0Yi_3+MVC&Ddk5WO0QP_~SA
zZVS>Xb1xW^7gGUVPs^huFHfr=ypH-1+nZw@zKI*C9h^p*;QypF9}WFxBihq!Y$OTG
z1n7j|;<$J656rvjh|=yxBK@I5gijUj^N6Y!dQD;EB_}Zo{`5Aql^XhcgjiSzg>z3&
zPhv)KhH|M|?N4g750R)pA3rTB8jFE{&`05LLDvzXSc4VY1FotqRMc5C*?I??`M|Hm
zh<m!j8%$Phr~IpvF^^HM5Pe>mIRJq{nh*l>z6z-D@SES!i`S+Y1(+Njs+#fVw-IX&
zAbG-vF$ikm-!l$h8J(1aaFb*%2iD6p{tU9dQ#LxIT}!pj%nyn~T3!jWJGo^()%jU)
zts^v=rq@o6El9iCe9yOS-;ADB)_*703Ydwgf`DwQtMCLmDFCvg>_p%nLFmU+2ULvZ
z>KKOy`V_;3tWld`8a%}1&N%GQV!>K;Ia~*F&%k3cK-4kaFghee9n%vD^(od40uxB=
zru4LyiXQ+);1E}Yu0iZNd{F=n1Ym=3hwVH5b|t?4(v%LMFtjroSIk19D<e|{goS8m
zLPh>#DwHB9ER22ihC-cjy(oct18pbR1?^|5i1WbU(Y=Oy31k!33^-h0Umrlv3^8jG
z&vWjX6O0vrsUUxpfJ-bu4<uLc{Nb>(GiC+<fgOdt3NtP-m{x+zfV>>)LhJ<yDrG>#
zaGSn+f%FSFfz4M!D!TWxDgZDI^K0aQUt&d_()ilGBCWy|%j?7+hu_AzBA)xGx>4UV
zGBHV?fRA_&@|Z<zjWN{?=qiy-1aHhQE)oMM4QO2V{hFA7@DhFe2JMwNh^;$w3Rb@r
z+Q!YqAY~+$IIkwK9V|oxk)wZ5K*S1O^Z)4ecwXKA3?-DVkOi5->K;@Z)SY0A!C}C@
zpTf(_%hSd0(qC_bDnj5b<~}gJ@%3$EQOfIV${%!T986pHQH2~$Na5g=dwsyiNnKe|
z`%~`zHhm|X^em?d&&(HE8lQTkCG{Dc_ioF0@<et@y7S$v@w(EfC{qC=`jm>Pssko$
z4(x(-h_xP#?JZVFvr+HEpu@Gt2A!@II6A?3{p+X&Ik&H$ABv$K+%@Elm+j7aVY6V>
zA(EiSD<M1kMC9TNBBod&We~y4Ps9S`!@!iWHXw$^p#x|=k5Q@U7O1o^=c}j?ZXnS$
zv&_rScZ66GBGe$8*7aksS3(59o5R=v^OShGxCHt5sH9&(c@5SsY>mfsH?ypq9C3)C
zB@z-sFk0x&^Yp|@7(XK_N;oATKIPiR>s3>q#n1T;@ap70sgLoNhX)HT7rKxayy$`?
za^cLkq1_Bu9zdta?=ih?>(&R49{G4&MnuPkslAqv=i@_;Co$G^H}3Yb_e--qy*AVD
z0t@La8ts{ab)R!NrO9ck6=v)A?q}nnw-8P_nihNOc5c6Tnc#y+ADym8HoD>YIsCc&
z8P+K;Q<3A3s>II!w-!KKStG#k`3Bo~L!EEfGYPS*lvh>t<b7Y8UWX~1a1e7~QizE@
zA@bYu_g>p$&+UW$lxpZoy)DyT8Jdq07PLsxl*x|0Sk~gxk}#8y(U%cMcfv%?#L%=f
zz;jm=gYcr&q*=VJ(F?Nx_2`g*5UO?2v;6ETu;_3Ua_WjIu)Rjgy6e29Ow^q|jlP(*
zm`<8bNnEUW<>!qXHJW_vKQ-f?0l*<tlq0`^NEH+iHWgx6tU6*-7_EqXh-AqZvq`S0
zfhNZO>&`LSlBx6&geW%;52VJ#bjGYyBCg|EZj}Umxnsu;hwB^Imk@ifxR+wPSh?K>
zhYD)-RZ`M(Iq(?aHL3V8O_y%tix)o=^%DTeA+tA^fG7{+(PN~6c;@A`hcsIQV`9pX
zEMj%Cv9R#PVcbq)YU(H=bFzHIQQn-;euHE1!R0nAzlA3WjZc0u(J-HxDTbf{<uF?O
zUs<|Co0}>aL$eiK&0J2^(kS%47FZvcee;S|{MMfwoGUO~@Yha|S5Uy2O^EfcUOjvx
z28J7c`-ApwwFB-ZCG`Pmol-d8PgJvQ$>?q*`YZq{I|BrajErQ|{evPC5efL(-T5#!
zV;h??0B8`QXp1tFig2j$9e`0Q|AFO|6~WePNM1pmy^QRvNZwv7oI*z6d^~%}w!#ad
z@3|d|xJV2bCDoOibPNnQ^s`ICMp%&i>s;~#f@v%MicO`W1_i1eBZURP>u=bn$hT>>
zs%U8yAtVvk{pJg<vt}ThgPE#B$H0I&LOB*<S7g}L$v}3IS?F6SH6j@E$y{;s@CfP;
ztBCt?gpG}jgX4+D*}E+C42V^^NV+axPlJ3NhUrKgl4GtB);b8oAkfjF?jMG8!yI7{
zkpg^i1{cm&b5H<`#q7=gZ0j58U(m0Kt{xC|ml-~n8-F5)WgCL~Yz;Z)J`zpVsQT)Y
z0Y{oyS`tmp*nZv<)w}XGY=Ma;7FAt#K<0*77qY(|f2u;<K<!BM8ThL%V(L4h3r6>B
zt#Z3y_g)HWcetwoVXq1FLXkU;(zE;fcahH2K^P{&82r3$N<sqBK|Vwb)~dJlz}e6?
z7eSl{p!Uq2iSY<ndO?45#NQMC8YvD`#c$uDD?zj57US>lJ1Fg6+m47kg=}=wF6m9X
z4Sk9jOB)mHHurU8yc~0<{d4uB`l{5_l=+-7{-)b6(i5%4{M;FB1NEalIoI1rZ*^zo
zW<O<`*qD~R8m8{7Oy8*~#k|J;P%ODFnnW?OX2KQ|fg`4?HotwQr%pFwJ=ta8&i*a8
z$7i~lugQez>8i55x)vdg{k-EV<+Rmgd%22-`o65EF)rVx_K1tQ@-4r}k&TRyiljfO
z-{T?T;NBfEDR`p1q-d_QW$vhmZ)1V*>q8wz(Jl_Qq(*_HwSz_>(~jMvJT8SS!>o1d
z&OWp4!~F;iHQ&m)I62L~L?6$}A?9|X!5y8!N3+Pq3sQkA0NsAVFdY{hJTN#|fHc;w
z*cYN{6*cOqI5ao}*vd{kmZ$oN#b_ZHp3HUcFdDe)*7lzS$V~K#Q>^YChWrhk>4?<E
z2GdQ|GY<i+5as3qoh2Sql46WI@(gU1oE|OgYZ`cJsDQgj(G)ldg*8gyauoX5b&wK7
zaG|bx()|jN8`V-w|HPjc?Y@V}DpC6Jio7J;akg%~saG&#2eUnx&e%`M+&heqxDE0K
zaU1bOOYiIQuWy1g>BrD9jCV&o;A?%kRg&ysC2~w0H4h#?e!8Lv6BXF_8I%UxOdrM{
z!r2U=>kJ%)0|WOi){;A+dO&W(%EmS{G-PFINk>nQT}F(?!I-a9&1YPDP-S5Pk_j*R
z1&w$yPrZpmN=j6QgA-zH4Rfx&t@2(Bd5G9~4@3wCIv6Z`MrO3-pMO%qM!+xkeO{Sv
zPLw;)`5XixFf#a0oF6+?u^+*9;FI^GM;D!**5VA(2!3{qtPfN`eQPUDsnp+_WYh)+
z2L~ea<;u|@)#`tq$c@*Qg)+BUHAYUcxIXdq1>Go7l`?@Q5mc@Ec+AM~P}wsozylvm
zAW!PXkz-;iH4$n1)*k@>FDol6X6FKi07ax0-VSJsw&r0MN2Flp=1FM5#*wAfQHq6u
z(+zvBaxETwKdSAm{w?!B%zzSrEW<dHUE|+mcfOax9yhxlbodD_0S9%pKrw|R6|Nfh
zb5PA^643#78)@}fXo&E9#zo1=nE@y5J$Ufa+atAe0Jt3CXGwTHl({=K^fU&Ex)xz9
zx<k1lXfP7TA|Aq3-d7DNyo=lS9XWdRp4$O8G@03_D$(!TSAer<reb+{8D*jA!l(Wz
zk3Gx-i7(w`DB3q1RlLI16s6RU{i(XDDW}<4+va~*b4zyQc*LyRt`8-T?ke70)0vBM
zG*_J)wjen<o^%ZNPILJ-w`&&BT2NC>Q+@HxtigD3=z`>hE8EgObnUjy=yjd@-CTb*
zvLJ9iBJf&@h|Hdb-BCQh%5R=~y4gg@uYuRRVv1GKBX8k|#G?_5+}^#2>4T<|0|$nx
zR;yQ2$*&}r$j1zvy)6AsNH{g^oYT$p_dU)%4ksGLR{Ap9Y?CbvJ;nX4urw3YJaCM8
z!VVbSJQ$LJxJFOE471s0QPiqa=sOAt+2^7L=5z0-R%op7O!5m0HC0sHPy>Q)#BhTP
z<M22_aq-06a4nq*cblyr3Dhl5Pe@S^Wt6Bcz`a*qz8;CHw4vth@~>^>PkaqswY4Kg
zA|>VI(oy$g9f2H$K^5XL`s2VEqPgxK$B#X`cHxuh=0ARxT_{@sl?yb2_P~Ls>(*vw
zLJ0Zb|F>+pkL&>NLAX$#yfgtK83zI}4kldqdVgPEb&)dftlY&57n(wgN?Mc{&N^Wc
zK+!wk)!NbmoLI&=H=q)TfcDdor?WV2P*yOZ;@NQRf*77)B)bLQ%3QoiG&L4<uVYIi
z6gE`MbU@?)r~*<5F`RlLKn`n5^fWbv;XH;$w!Rrj#4w{TDyS}cF}QQ6_Mf}&zI^dt
zRZBc|KB;k_=f%;SixUV?>|}ko%|r<Wha(_U${Z&#TjI@|$(BU4axpaWdccclV{VjY
zsTX5>p6G-|;`B*PwbbeK@(Cv@W)HH|)TJ;a-QHi$tRgV)+WJp3<qWDD6g;xyuj+dX
z>CZxShfE@HhboFMgslRc35JFhuU@h6@Jt{{hqwrv4e!gylqe0P?btFuu#I_qRb<LA
z)rQ`X7>YB}m<hlTFHzCRd2ziPH?lZR@$vJILPvl^s-eCf=q~JWa7^J3ctpw7)zt|@
zPLzr-6cnmY0_0T@$O4>(0_z4~Y`Aly!bSoZh8YK)%wSVsZy_fqN7@hk>SDudjCqIj
z0IavAwKc@p-+`5)8)v_#2l9%J=4R~zt9^NUm$p;l((6@LXULhMnQ(<B(EQ$=WO&iT
zDNA5Q$?l1d4~nM}k40Xa{9`-l?mtEy4?ZVoV?Q3KI6wby!fsUS&-t+*cS&B#4+`ee
ziUyIHU=6ZAo~$jClrC}@-s2EILQ52PgtK;dIGQ;W6ckKMh{-I3GUyr~hKgb4h|Jmq
zbW~+%)2UB={bvQ=^PvvOnof{T#zd<lz+)SIkUQJ^`}cONiH%52?w~v6C%tI?J#4`s
zjq^wC%3GZhe>?g-<k+3|JWf2ZHK$dh9@#ikc8aZ5-?U?z4CkKb30u(N<fUUbR&5-x
z?X6gu9FJ1xZg}vW^PiB4?W^r|k+wTFM6H}ox!Uq&S(gs~W<K7dxU)R<RyP_3uc)gd
zY@(0bbZqn{Za!fJLcxw5TnEU!@cOsP@4*&r?f8meJh~SR-r(6U66Yj_k#l&dsH+P>
zZ-$-MH0D`bhfw+8ypgNxX}UtDv#tT!(TO@c`k>T!UK0Z>7PxA-v)D2EM?{k$dZRAy
zK5n{EiN!J0NZh03^mI87PK<B^3{3d<{Q8M{48f~thi`RJ2iOY4&I>iCot);8Y2s8#
zO<9d{LHqJFyrqCZ4P(d9Y_&to2W&%R)ZM+$eSz=6+EhuG!^znMidhUlOC*&MvVA;S
z1$p^b785{w+GRg`lRZ4|<+XtTfvodU<1+5Skbkl?0GBgpu<QJ!Yi!&HT}Zm_BMA&7
zsH$SeJlr=FdJT<@y$}UJLWQcc3!>{Ga-bw8yaOZkpCUD0o;r$XaaZUN2F$<^@g@){
z(rr~#jK~Qv5#6NtnLVO%+s`9Jb$WVf(;*}~Nvv338@!w=s@5KI{692N=hW89I*O@j
zXrNu!BsFLkBY|`wjZH%B)1QX)YII^kL06&M@I7`jBW`g$RVFq+?s5LUf3|E%PfG&;
z$^AR}ATVDV8WU4fLf?(>98`xcUu59<1k3I|i}*~iF-Va!{2;+cQ(jYRYfz}xA(+CW
zK0*VkfO-AF=0Gn@NyCeA$1FURu`3l62s5pyWoK+%02E)4<-y@tj_QJeK@Ya2mf9sk
zm<bHz!ZU`G7KB$@=eeu6)r1Oy@$@um{jHJ!Nk4aWMJ<2ZclX<I)x%Vm#Ht;U3!!Pc
ztxFJCXOjP?^iWL7vtsSP067ODlLv-adOsw)V7ikT5b?234?4lP0UK|A&N>%pA*5LH
z<L>+eMXC$H4#GWuO8W(j3KE4;{SQfX^#VA@AT8)~pTaQ$lMs1hiB?>On$#jNt=iMi
ze5zd-&T=T4tbK0Sy_EDPUD{C2D|98}g@fdbjk8_jr|z25fp$!vCp=3vc|(46p1MBX
z|HEPWc$wAX(`x!jX{ndKoQP->)NXO88OUp)Rfz0z;@vQ1v{dX6$rd#1RV?B8bcVB2
zIOG}Ei4(-)0pe#-x9y+z*laM}I?Bo9gi(xG5e|lrpiPCl-y;!jLuqL<wEsdak+2nY
zb0^W2h<dR2w*SUnztNY>O%PUntXSsenlT9Q@u`K%;-eF<#rKkV9HVgDD$))5y-Qd?
zz<qXzLgqLPPI}0N3qcy7?UI9bva{f+hsAwg4;xfl-rgJO{KcrTN#xon&{07`KZKO-
zD48)zGc-K!TIAS7CoPjV=e*zLnMwk{bOIw0eR>PXDIz7Sc}fIMSd>qn^?1`L`_M*X
zYKq)08C;p*$|53$LX2pLNXdhc6b~<??ied|6(VEgXV@2kVhWd#j3SkLIe-5T)SP*_
zT9ajS7zBmH8#ysp3LGYTC@2!mY;Z(FH%fo-VD+_KA>5cDgkJ%xV}7wc<4~x{FCObd
z=d6wq332#NYZMQ@bW%~jLg21XjP61fOE`HH$-M>Opej>hX<-2c&n4auT4<`jyz=tu
zgzWT0{KCoL$-^Yfc*ynITp%Pu&B~o3k%r`8_R!#B=61?~hv{WyVuEz+N2$mFeX%Ua
z@SZpUSIY{=QEEYUhzdL)u*TRK6?%fE&R9Bai$)$pLw}ANuoyu=O$c%Yj0)Ahwgc7C
zg!>!FO(Ld%yTaQ@MOau^<?&mLQv%Y(Kh8tgXE?^Tw6|k9bB^0~RaW-lbOvDz_%$Of
zO$X#z!mM5WkN5B-zsB~orn%V;pMlnXwlIL7#5kfp>Lf-SztzzKv4_c+m|_6gfBW<)
z5vc(oPzQRESNjnq5zf1}nCB;M%i=UW4Fm!)o`2<$wWTEr8g$aOH#JQo@j*gvY`zaH
z+fQ&9*4A}+0P^MUw+MQ&nM}X?j_hSSC9-(N!9B!eG{kXGFsooZ`$!nMOI9#o0dgDm
zA15MC&24xVzQ5lW5b+XtRh7vz2?e+1#8o-$JR62LiR#ymktO(~_+0jO3!V98_PH+l
zd30sVc_DQnDV`5QpBu!YJak5yWm-MHx}R*ZnG>n#bnvoM)>KE@<TNz?f&X#uaH(+f
zsio94C7!X?sVus7A$!YT_OojWZ%#cCF#1+D;PKk>^;B-pVAzn?@r{JkB+E3L&JrW*
z6S^nNJKmg2KDWMfx><9Xf2zNi->ah5H8Cq~(sXE`gOr^mDt3DzXWpVKXQDMY`Fg_>
zGh1Vck~NE=nYo0ymyQ17N}dL6+}2iK+p;s`LVkK}A4b+4Hg7;W_UdttYNU@V`6r>F
zh4j`mbuQA@&63Ef=?N445jzcy{SnT4Wi&OzOiqpC0R)A`K!`lVYpj|cFnkKk5y0-r
z5Z2E&0=Ivn`<r^49DPhdLO1XrKb7Tjdq9vMz+L^l(AJP-8vTT$i=%`Z7OkO0>H@~j
zHx7_(9U3Z4JD5Yhd-p3RLBJ*KE@F}zNbybfq^PK=Au4qfB0<VSJ@`-@9grvtqB1fq
zU&XRFJ<j2M;X650mMz$4_$ekUJ)38f$UJin9+24K5$TBe)2JKq$S2#=JJ4c|YMO}|
zC-mG+q!Zn&DOSp6W`j8M!o4&y%x;|iJolRPI_YneJi^ZYxH#Y#oecGSxzYDJoD)D+
z+UmB%s*f#Q4L#=$m4K!rb0GbEpKQrDmDFr%&FTts&)N6)KbFSgy!q&H1-bBn4#^f0
zDYX7JaJT>`R2WR&Ym1l}DCx`+<{Ihj;6Nk17@U;EHh;Sv0nm~Qci^$S2J-8GW6cnd
z$}ezKVDOrKu5q25;QbwTpE_J|9?(r(5P@Ub@h{Iaymvb|%p%J}R!hlvy6M9Q6NufA
z(3&+QU{l}S5RoqVVT63Qel%}>ejd0$k>`ptEOrrLv4P;g8u4wPA_pyE4v6*Rhwkm*
zG>^cn9FE`Y`r6vWP*=OlKM^hAy&brkotDN~Wm7S>4Arf)$MkQ>X93Jywi&y<m~J2Y
z{rT?57D3)G2i-?{wJCPV?m8u7<J>zc{%7j_(`=qS9?Z^z4IwF%D#A3jvHlWeV?${M
z9_|k-_9P3XA5~t8=a;j&Kqp=5H&R&9yFRie+}nGW+2hu9?|nnFTv`d6?*V2U>h=aR
zrFFBStL;XpHOFscp4sO$W>wjaH|^-vmzeogji0(dnlt^<)Tvd$-#FuT6hQRC?)mQd
zs~vtH83JvOY4&J|Ouu^<eS-V-@+Fxp?L*oM4l!(pM>fBkZ5u8-?`~6+cSPXG%+N~I
zO6;$3T#CfhdVHpN{I%aEb%)Yd{AH&V2K#Ytf7qL#Quh#$!t=MO7!YV*zmoLwr68uc
zH20%7$J?pTnsUaNO+vzOp&_*Yz?b(QK1j_S=I0Mw`A0sq%z16uY33smCJ6`I=b4#2
zBBbX!D$=H)t~)m3e$9tV+<Fnl%qJjV%=YY$n!0+TdLN=P#@pR?>WWvGWuc~}1v(|I
zdm88<X(uBFv(*gyjdf%nqNkT-l2gSbCz9uds52o2uZRI%;WZ=}Z-6WCeuLU4uz;+j
z&%)Z8m}B_KXCF{a(65BJ1BfRi6Llppy5M)n;j@m@@HgBwuGUr=NQ5h`S-Z24V&@D!
zg}$6}tOrDgM*;axJ9sDUwYiLT(ke!Q-rZxLo}rnGmPgr;weGfNw5_-+ChvtNvJ{iK
zbeIUZLb=UWS1fyXH#3X#k@y9=YnUFwrsh$<t~}r^FC}FHGm~yxVjM#**>;C>(Fyl=
z3<W0ZfXTYJl7PhuP0-OBo*H0a>Un3kN)nbTZ6fz}-h(Xm>syx-v*%Qlm8t0HYKF@O
z7|y~_0!6~}H#fn=ASC5KgeG<%oFuiE>g=U>q=-%ll(9s^i216awl*TbrLl+`_|=R&
z^-IHVDVS+FUvijKrtWlOof|mebgxa^W97O}XRvC((X2pwqop@1+^sVL?3Y4Rob47e
z<nmIssK{~2Qi_Rv)N56ZGnkmOA$7$b5&IDPcweY@=I==D=%vNhiio-;uK=G~AHx&t
zI(qAI4ld`ND{C3J9X7kzkU}E#`js`yZiKPTl_aw6W2I@7f9071I;6MK%Otc6{g7cB
zUB?@cp!M^zwxT0t=Jm&jIsoG}W#_@|R96``k9dx0o<u8DvQm17Oz=ygBrDIKXS;rW
z8L!FhNHJkw@;!QF>aL)*X+BWz@r|G?UWR?M3om{xe!Go0s2H;@Xr`p4b#VNhmVc&k
z=mh&m!}c8BVLDX|<mNkOZ>}ZJSI$LR_@uu>U|b4(!TgxIxGu-ByXTs~oI+cJ908{6
z<_NApKamm8L%xm-9Ss#O6P=%QR!1Nm&adDmJqhhMKY?|1kw6q~URQ$G5USDx%*?!;
z+dqni5Vl(u-s`BC)v6c}fF=Qv__LO13rjUKZMNECMM*Zw`yeITctS~~zkKNjh5)-E
zsmPD)Yfs3`Y|uY@Ws*32**D4{7LXiQRNM}Gu|15bT6?9Wq;emSIu4$N8s!eDy6@A|
z^bkRmzS{*giAjOwp4wS~Fd&CYp;j-e>BkWXi8wc|FV)qTx>>H&djx=?SJfn?@83$9
z`(ttk1W5a-se#mn09=BH=Iia7kh>@HXhS_Gba}Y1?`SW5E4Z~6%154^Pe~Ql*p|3a
zcS4xoLf}<<0B~zzSKF^5H;32bS16a5sn>K{f)a1OiVuxW{}}t^jO2~oqYC>y$C&RO
z4*#H75$v^>b>_-ePJarfwXWrjo{7Z<PvI%%{FN7tk~8;rTQ{hV2W;#pURz(NDt%Tz
zIBXF6l692qUQ_l<TQe3Te}y#v{Y=uwoGquSBRLy5zqrRPN55&RsrPoW>WrLMXn$rB
z%9WJB{E0<zy*+K8XVd8_@{Da{R>`2g&@|~|6^B`^U?H7>v;kiQ$c-t=F&~>>o}5{(
z<l^E2d(4dZio_2>e9)F-6E0S)^pko1+#Y_u9RO5PTYCbxU9h0Ud_SJ$?1Va9zGM2u
z^2Z7OeXQOBzko_=$k1>JoJqYv4F=ryv2_DjHsXEtJ#c?AGR^vj8TY?iNYU1KUXact
zyZC9D2H5yu-a`Qt%uFuYi$E_R8xUN%w)0^nK`7NN@(oKLS+97shl><VHviAXNea$;
zMGMGM0FfxIUx9g#1aLScvz-eEXleC|Zr)|g3=Io|g|yO|8}tQFSF|+sGo(i@1SxoR
zrBpbK@*YX)2w+TnV%3CN7A&DX82p5j-O!vMNH@K8u|+wI$PNGP1Rb?rRHk=CRj6q;
ze&C$>1F!%sEwHVvL7M?L_bUUFr%4xC$r4ISS5fWs9c6?>+|F$1a(>yxn7*Lh%+H@c
zuP$X4v_E_(sN!cfziWzN#X733lr4&Cs;Z{|h=8<=<>5=d#Y)x(^p|Qe2sdlzo;^uC
z10t6b9p7)Uc0EQ*o*Lnp_Uw>$)>%k<v>$J)K5%b`HmsxIW|p&Of$(p-^@lIw{_Olb
zTuP7Si7R1jyS=@D$fG$VP;C+v`ug8dH`lu8<V>A%RZiiuWiebyNll@!6f(X3__YE>
zq^w?qY_@;2*N4UD+V*QptlUc%6x2gKLYJN9s+<lEcXf80ldJrGjDG#lK+Q0FMjZ}E
z!&em+wxrlMrLS-I7i!-cW-?z}pBU?6nlT$aa*1kd<^x7aDyO!wE8|Qd{ZX}5c46G{
zJ>3DPzZ&#qTx;XH+bLpoF&cmd{6YiGv~en&D3L}oGBz^u6|^ZVb*7`26c<jEf_sN4
zCy4ZZx4f7YYQ@m9i_QUiA0EGT>QN(GjsGb(#v`M#%u2vI7G`Ej;9p$Ka`fm?0fFbk
zS2R$_)<0wv*+ojOt*e6#A8Tl!F^}fQPN7HcgYcvMQey3@c=Sl+GXJ1|BjZzu#UPzM
zK1%};qOz0~=|_l(@qSWrieEJ_II50yK!1uWP7wSUeSESP6>Gi_2nxlaelQ@h!E?3~
zzA*SYP9ExD!c==9>ibmG1+jmEil1A>m!~-o^_RE*kkr&%1R@1)f+mE5hl>jhrZ`WB
zciv0UEzE)m53~*NcmESD$)XX~PeUl%)=<u%EoS(e2+{>5J#qiWy;|J{*rdmV(XKJ;
zLB?!eAO0H2pH=-EY`@9P6<0E@sr%o}&rkxrJneR3^xh8AXMFL!XD{6v8z*Pp`p;c|
z|2?x&AZ@X;lANBPY6qozEN`+0$heFjJ*)dFq{eDWll4cwat?PI|F;$(cLn+3r{laQ
z)&t}Ybt=8gPOPy0!E1ihn^IiRPeV?1PgxOzBvrji#8yeZ$9FrUzCPZ)hd-5FVc2^8
z5KE@=uYI2Gd&_>ER%r-MG|~^fA6p#jR#4leE_@($Yws0tMSqrsz@)6ha-07d6g+hb
zxGCYXh;V<tOyxK&Iiw{JBfdmB?DFfY9qwQE2NkohkPtM&>St~~)7|`v+e5)P?6MuL
zJ_cisR*E?>c|k$Bo02m5#zwS=LrPw<-Bq5@#@}lwmjO->T{i>s9vJ@2gvt2W@6cm$
z_UqK`vFT=j{D1`;zIFC{6YV_s&YizrMfbt{ak)A#ImLh1K;gsAEx~Q@Iy4ve7gb}Q
z2%7~RM=BQ5&i-~*OsvLm2msF~eeT!mJGmSqhji|Vs`@&;J5z`p+sbbn$?lraB*PwN
z^b%qoS0RE87-_5CEVOI?wI-}a+$Cn0;0xx~pZ+L=*?I7LEBIV*zm@XV=a1f9*3`-I
z!_y(#DLXqm;oB1~U`AdHhTRrL&)IA8SvRKNMT++?Nm8Aq^4ZD!;QSHmDaF+LJH#nC
z&!h7ye5nW;0p-W_Psz641CYu$Yu9&g2fi}o66PI>q_|0}h-$u<IDN6?4ivs8qTOi`
zL;Ezc<$|3J7pMGMSo>+?Q+yN8yvmWCc@XdJG`Fzlo3&nt)@FwJLxbmEw+m{|KFV18
z^D&iF+c#p#`6lcRZNM`VO=abCiuXSI=$Kw9aqGJfPDjp)1P`;y@bdBYE=I>1POB?d
zXy;$eKzW5rgT@7d1Fjo_1f0Uw-`3uaIuX;FW#r`QK<uIY=&^4j)V?1b1WmtL4|p*M
zS<Nl>a))q#QknBRK6J8Zretai<G67Ci~U9yJ!aGUHd<@bP%yfzU^1GP*VrxhqMXo)
z0IhmCus28~K9}?tSdlnMcI5d#GCVy(pBxbqDtPtkfzvDM@tY$51@H{!OKa`od(As=
zsp@Ie&%kz&6E#(g6=Ed4*pEB2D>k=n)*A>8UWH2FdDTt0s%e)nb4=E+*zBOhU}Y#A
z^wE=#si=hV3&}P}o0)2)s&1GngdeV<cA&1BkmZ3xfY-*lsemHr5!8HW!WF`MojeJF
z6U)j&bk^TB$<n2ci;gTROG-|b>9GRGjTCGD2Bm7nZO?Tsx%X{tb4#-u!CwM3r}>~S
z#^<T-@Vn9V#&7sqHr9yFxqxfES~wYs9ZT2fk&?GD_HA!tXJzG=cmXw$)ApSNgozY8
z+n|{}iB_}Zwa$AfH#d=uJp|=EP}`SQ{^S;crvz(UD&iXI`S`tTXWiG?uFHkTdM~mK
zvjXGjjZ<B7pBJA0&a$DfdV=)pNSy0cy~lR-rh3IWtE+XhE<saC2a<wna(NZr%8yhR
z)(O(&-S}RiWSMi8^|k9E^~nRHa<$I~v?z59uz<pXgFiBjtFdPp$ClQt*w!DVJt?I7
zh!|uERC%v@o2<NiDF(8_*HvDAw<Oj1_?y7q;7u1~`;(;wAOUb<F86N8Zc(}mCjZ|5
zQ=;(>k<9qwzQX%+Q1`dw{(7Y=s;bZhvxWuOAsIjw)%JD`wBOI!^Q2ZUOtUVc7yJGD
zcl|@qujD)=Yh&Z{T)z?$6Q!9Fxb|9)c1Kp0^2_mMjSw^1Gcy%-Q9V3QOhpL_qM0Z_
z=LFHRb}IG1z+s-f$9Z_n*eJ9;SnSx92mIY<r9R>$SWq1=z}Yhat=)yuXQ<C1ce`Ie
z`SNsI)Q*f3ub2yqmCAk2%-*Jae;-Me#h1<Z4#V3M={RJkG>nYg^a_wj4o*!Sf&t}=
z02i#cfxDilz>sn66Q&BeHXZQ%xvGZ7XHNm@!Vat^WDU|W@erzMue6Y8ho%AwS|Vt%
z*Fe8u9zcaqg>-?;pVCkv*Q<V1r(;zu{<&9R2k8STqUe)~<AG+YpEPH4Vw2)v<Bade
zJyEqBW_InG%tQ(+$cXnfHN>FW{D*xdJjY!Bh`W7t=R0ttG|#T?J5T;Y=lO}?TxnI~
z+X0*joIhOOt(h+K3DHSBDcCC)zsWy2*C$;?RwyrgFsmg#Tx!@bcDT2CTt0NBPiQ<U
z(oh|-!uy4PfziU3!<7Ei!(&6*AsRa)PuYl?)dnLa!gKZs#k{<S!z0qikqb)BojXhg
zzCue+!2d8=r=B?@&J5F<(%f8uL3!U#YOu=*-{tn@!C@5$o>5cE>sLBByCKF7Wu6CS
z!rKL|R%GbB$d}EHJ|f-^IP|}UKq<C^6dcrzVJ)+CPW!;zA2?8@r^wI{pO8=~s=~cA
ze33P6?Lz;rU-A^KgQ$uKvd5{KbKJ2$Ly$*cM=}R(UV=ztyn*2v95cw!f;PVAK4ku4
z4^rxDq3xk1Lz26CTOIYJR7BsG(q7%^7YPY12#Tz7p%}a9^Q)ekhGrHl&9<^*Zqt^e
zjBmhnMTX-0=*|X*g+09?Wq)`vGA%9b{{1cU#|2yoRyULWPW^6X{Mlf8W=mDoW9Qog
zba;A;ptTN9o=sW(5f5QjeQ-CZasuLKcLlJ3eil<6eRjhc3TCvHH<qp#C@bF_C-Dqf
zYd6j#anrQ8@AKp5`{!0}uo)hocHT<a%TC(4bx&1}9+OM%$g4jMfzN%;CB7D4D_rJX
zVFCw9^&szI9`{sxrnd0rJZC1!lY-YjcR#94t1UVix+VB@Mv!#)cx10qdE;<gXo-fM
zo86bNPF1Bl4{kBOMLlYleG`y%^A>gWb`TFOv8>q=INPhmem4NXEbe{c{=sl3(~oWc
zY@TO6Jg%+BTYdW^u>7{v+HkGxRc(~a(059Un}>#_Nuq+R7Z=Z$<j4(|@G7U>=Ro`w
z!!aV=s_*0ip!NqX4B=cL+uoK;>GD9nWyBkO%7)%$Z7MoC8hHrNP{CbZPrwfHOWjOe
zNka+o5x!O^9~pUQ_wQHDJ9wqO@id66AuFJruk^V=u@D-&Vz${qd)t7+Pats^a?#4P
z`gR3luOS(!A8ebWC=Mi)9a%NYCK`-Gz1h?P6$Kw*@Nqu2jp&_?{1x>Kiq}9XCsEj+
z!X{!!w5|Sx11+^Wh|tBZ%br7P$k710Bc}}7P(MG7*-N$jN>C}VZWta|G=NCzXnz3C
z%%+PQfillv1dH=<_LdCmRc3WSn}2$L46sp=LD$CmX%}S3HS(<p%F146UqKv@*K<hS
zV)`%@S@;Bw0C0~{D-XfWq13lRM^VAe&6hc71sdJ^zWlmXE$phSuFg8l`oh>h0#>U>
zDUGk>JCt=Z&<5I0HvyDVj2RIL#;1e(6cC<4JkNiK#hT&phx5xK<JmS^QgaxE6GKaA
zxK)(muahdrby@sM#+=l*&D}L*o0q?$KC(-n{_R|D;Bl5y9fF5?%uhQkoye--J5XNl
zs;#Br_s^tl9=yvp<aNImTV5mG@Vzq8&^g6Wa$=w=V0$NZl3MoO>Z;f9B0VCL<AJgQ
zl&;L&SI~mm|A0+8m9PVxiDuT57?~v?3T<O;87aqxOj8-_;~+CXFSFMtK-dzdJR&!~
zmSZc<FGqxMdo<M4c>mG|ShqZ|bLDzNZE{l5=iLVezRthHv2Gk(1G4!^G2f|EB#brb
zwCl&Jg5=V4-4V{D+9pCG=-xK#87%kOD9_8Ag;LCNunc`*fZBy%Kq4rZhkwe0^u^oX
zzx@3>(b0WM81Vvd0*zd8=&Lt1Whr<JFyj+`Ehaiz>uGNxh5>#4{OC;%9#s8}z38UH
zZe`!Wh2t6Y>ho``7`5|2bn9hF3hJ$JRjpp#<E9u<Wrc84FY8Z$^ZAxEB9kYCg%{zk
zZtQ<s;MA$fZ+YxXy^t7>ic)60yaJiSrL4m@1Ri7fs(F=|l~IAMF4h9_%M0sJp9Hvl
zh#N)14#b~EOhZLwZy4F(#?q~lHwTPW)zq-7>el>K<J>BSf5aO}EwlO@4cNTJrJR@s
zv3lk?QLn#DZ^pW)Sv1zjgXs!OM7%Bx+Z*;DiHDBDxTikIWV)%$!)0{FIODw(#QA6_
zs1eLo<^XJQ*m~kBfB@;ReNm!LoDM$$dZ0z_-LFki;o%Tk8BkJE!Y2Zm+fxQh19kOJ
zXUEmk#g7Nwy8jr;Q$a5%vGu~S(U(@d0Ea;?7A}lY-C`k4hvV=3xO?j21ycwY(9bmB
z1?WV~qUo@Ccn9EuG4I<59@NKnkhth^G}v$RuVE~o{SlvJ>Q!UTtJmppydXVH&Lk}7
zr_ucQacAx)+~`z{XyqQU7|R$byC;7-4Yjzf;H}{oF&SlDX=W<Vq}OfMYL06=2|0x=
z8ceo#KP?`5t+lg|FJeI^Z`JSmWTBg4m`J|1hWb{{1-1o~n9l~QqkYXkWxjKNPg^mN
zy}xW{jJ7DEimK15#tH!KLxZsLE&-_~%#Rb!v_L>Nzp!9#UWArEnu7;9>;zMUsi9Re
z&G?v)g##wWK`9!&^_K<IjjXM?>bmT5a&xK0s%vXSp~L{`5<Y$lZSkN(+iM-Gb333u
z8Uuz+h55(QxN*c1DJwBS-3^iU3j7v!?_L+Qn}H6bI?4%p>|Z@)XEr9TVfe!3w<|8Q
zLke_VuOgAt13HxOO6tA&1E$l-XtY83_UH{U-~p42z*d7@1{b2HN?ljqy;YJJ`>0oW
z-T;^ku$rS|pDL=s8}8VBP#KW9wY9aFSQ$L4pa}{N4yI3irU<K*ZYX>3pOZpDJlx#!
z^73dq!lP_xj<Q9@SpN|B2DOH>i%YU+G)x6xQ`gqhW1z1uCN3W96q<ztf`3;aQ48W`
zAlK``8IigR$pWY<%uedVT|-@p3Ly8(>m1~yR*%E5DRLBd%X|<WhjshgkCFG`yq|9Z
z@io2Bkv|Q(ASv;+q`!6H5+VD-WZZjh9e48~@B%3sC{{{1T4c%HD=sdExiZ4@&U<7#
zcM@$-;cqea>?WX6>MGLc_;?2<pF1{dea|xJHJ#3ZEEyv1$re{6$K({u7h*~w&WJ`H
zExM&2i$9g-jx16I0AyYm^EOoI34{eb<xiN=_9O7N4;QBxAz%iYeyb!48yhBFY`_?{
zdz(Qob{%r-zfx_z^7r>sPH)NN8o~8LGGl5PO4wXsbF_;&!DYjyr!<1j9>}yAq%sig
zfssJ;q8NV=_0Qd_s90NU`6bX^18lioQI`lQXvr=wzm59=hr}qf*@T9EE{UmS$^H?K
zJArcds`Hg@G^?R>f=mzr3sS51-QOV|K!U4&9=XWWRK>MM=9yF8mtgt<J_8RJFB};L
z<+p0OFM;cusXkoEu28RY?VOBEGS{<~ATnmFkFAYL5pIl+x9<tO8+cjdGMWv{OU#ox
z4c`aTE2+gg^G%)8J^iYkYHFzGa`9yV`J{`aPWqYD$jqdKH$}sJbd!QTW)-c<?P#M=
z(KL%#<r|j!QlsT=c*T1`X6n%B=@$7#iM0&%ZE5|Yz6>2kdOAirR+4EsvKA?376~@0
znlb-}z4wmix_{rs)zzeg%t(=~WJJgeB_m{Imc27lA)_U;WD{ALS;^iN*(HU{WQCN>
zLgGGNb$uS+-|xTsaX;=q?ms^MxULHC_xtsFJ)h6>JdfizkJCaVPJhSm(H^322_YFe
zM3f(6!}%#GBRJXcS93q2)2x7*^!4l4IE@rSkn}^GqAa-uaHoBL$?M);F(i_Jat91X
zK7Eqe=8uJ9)jgjRh24&o1DW6q_*GbDaOcXOgrODG6ivp2t#^)>!B`4WZWc_yq8FBH
z8jhbP(6ZH2@EETK&VJ12kXA$W=c%cc)}tHKD!-6hATYg-5xA|#&nhWN$Y}|;%coCc
zPF&+FG*&S*WODz&Cn@=8qTw>;cab=p@${5PtYDFVO%qgjih*<pC4NiWIAkdRci|~g
zMF%fH(stZ2f%^M-BlKree;~_;EKo#*IJ{5*|6W_U{F-Al6NL%_1S1X2688UnH==`J
zWV0xQ5T7HNWhwA*i6ykQEMPa&26wH2p-Ms~PvlF!@tIfy2)aMVn)iaP#zd0(0Hyt=
zPH4{T3xg5?_krg!qis}r`e&m3z_kmT8AeSIg5%v3ehZ-=T@30h6jVerMZy!9rK+@6
zyn_eM*~KIvu+V*WX6N8qaQ0wI;82;f+(@kML5c^{Oz4%w8%szJ0%t~@Uk4zduI`Yi
zXpGor0Tvdt^zIyEF9P~qSg5kqS`>Ye>VIAlj}2Zj$v+fVVJ?FcumX|G(6CL&jURpS
z@81Un2fN~gnwf;_610Tw9v)!|^g8**uRR!TU_oxKP>4P#aZyoZoWL<6E6xIxhrI-b
zk-+;2#|Umz%r`C;w~Nd^<{c#t2HfQHcp1qnU#RVdYms{3`oYYr8HU-R6wWd08oz{V
zANjoF&Fj}*5Ov`I866$9lVCW_etmWpeQV|A1+FaITwHE2c*o<mODzJX<7{CCLTF&;
zmRkUrx@)J=fS#Gj)g8<tOQ9TfSi@CXb>Egf!XxGb&4Ef$9=D!s?y|i5x=+~lG{<N{
zMoVtQ1J*l|d(4dT((){Hhw>ciN&tx%ACN~k6aI{tm(whI>maR>kyh@DC#-2SX?xGA
za`BjbKj;W?MY>7c)q8!y0@m}zY@<8IM(cawr+fcCF<t<w0Q8B0jrtBtO2jF6l7+@w
zW<w^(bleHfrD~@bafBh*KmZG27+SZX?IZk);ADg|MK9ka3%V|ho@Vo4g3Jxe8p9!G
zX3lo2CjH3z%neN;;wiuraWE`!!R6z_i{RCOT7gXCaHoa+(bk`@o%Aq0Gq9>1%S-~C
zFF<`nD_z|WNLThTFkn`P02&2x?B%32o^P)C05z{%cOTc0BSy8}5VTjLQxiu3EW7}L
zV0fRkt*tl)1wnGmWwnJumI7~FL`sQ=KK8Z91=4eCE8TA^uDf>ch8`Z7y<V;^X5Dcj
z{MiWgY}>xQkd|)L($rKCleQJ?Lks~Km=T2xZ;oMkm<r#*@aSlrB){Rz?v_W;?!(5Z
z_>1SY>F#e>nzpt|-#ER|+|m~YO&ibUe{BDO(<@?ED4h_}Xs8Zx^-xs94Z;<cVFd*O
z{QUICtSO4^X2E0MM+ckNB!ECxKi%5vKl5;o!{z~WH8GPTIoSkH2hb?yoUz~6W#3^6
z&^={45GdSJ)rZI6fyLflHIkmPe)snFWOX-06d;#?S{Uf*MbU*_UXGr|xY$^jDI%j(
zE_#UFjdf3btei)d;^xhp2wHTK-jI8M#e|5$pxAb)Zjbaa7<0aSX^J=oO+QfR@mO1A
z98|sxXvqUDW7G3Axv=muss<a%4s>ldU|M0{Exa;vm8M^tkY~XH%MJoI3=hNB=L)pl
zwQCo~T|leB2tL$#-q0N&Qu5OB^|_I)iIgL2uu3-!B;`ay7P>6i9PBq8zbV5{hyVmH
z2PS1?fJA+hGs~rc#|$YuO~DzL7yBhLv$8-_Wa-Ntgq(I6upw{mpLApua`Pv`Hs6Up
z^&gsSEN+>U2v`Q+m1c5dA3WlhKG`cBU@NxQhvSP}M12=uRh83hmP*I3rE4F2wZqz-
z!E5qWjTszF_^i+nRKVG2Nv(Z7$;15QSYfBtyXvfyEC(79ie`}Tq%d-z0)e*N9mj2P
zY6PklL>SS#Y(0H_Xj8r=zUvF_F47zrai(1V3^*H8SlAMC>H$+WE!<J(@`Nq|he&f=
z=Cj3c9&T=TWEx-#!92o1PMLhi6LhpfdMJ!Jepv>eQTp%4YaH4zfIklEW*qeg4jlcm
zy8|ugI4A`s=BQ8qs7`<Kqykw4?xgyM=9U%@WQm}J@WfA5k2+Xe6D*s;o(gIcoiUtq
z5lsEM+S-wqb}1s=0OHPX(L&uleEk@}J@5uq@JbMLoXpD3h8qI2n9H3j=l6!k#`Z#X
z2pcPN?xMCK5O{clI*rKqhQ7X4_+R0JM3xbLBLI{2IGNKRqC)`v1|bPsdKz&KLT?DU
z1(C3hESVtH#B4Xro6v*3=eT`b6PQCAqf4j<0RZH*Ib^b9SJR{8!*j*uOUFSp(P4NT
z0tuljezVKW(a{I2JT{f$nsP^HC&16mdV6jj^hXMb<n`3)(?K&y{ya;NyDN^+0xOB?
z+slcy$-@&Uu@D${bmYWNX2_fseg_f!>3+8!=x6=?cXTrg%%@~!xnmO77q=c1=n&pw
zE2B*)Fbe+BV<_m>Lb4&3LU5t=LO){`AvsxJ6DwZ$1(h-W(jID#yt*2_G|+7nKNyF?
z3t~X*g3K%9Fz!Yk*m}9&+{D_t=_@Pe(5GZ5LvzCSD~}BiQ|{mYTKomOCLRZ({^>^!
zDm&4>fr~6e5}iO9gpo54c4Nh*OqP!-P*Y1l@B#|Q9cL}oZE;LELg-bly39>Lk1t*f
zBS&nLrcH-+8PN`ZOv)&hLZUMXonv1;pB4sEoCyA`_eE%ff1@g_@0>)0wuSb1{rkO+
z)mE{eVq95&)aTRrj4!8p#^0BgoLXeNFQ(!f`SP_!obFoT$pKPNiRI<zc8MR#qbHWf
z*6(P)oc^A-NAj2}v?VsaE{w(;r<)oB;5Gq2Lb%m3^+|@KArY+XC@zd@K^}4%oP$7o
ziXOg3!!R;EtlDvRZa9rnVs0%pHQV)qF=DC@61K&;IsB~2sVLwhfUHn8c_M=@{1P=_
zV{410{e!StJ#jVFml~G|txrF|?kD}SYCoz^Q4kh)kfhflam2zA3hH<m7x!y*^{%fk
zJ13{|HLLTaZ5KMAU`NaR*fOp3`toDfL&3<r;iMRaM+q;`p25LK3@%fcPc-%33jLyh
z#o(J%GP0`bG`qB<s;EeeWWynN-bV}D9Ud{5lI01hDQMC+8e4<KzGALpmj2I`EHGmT
zJcOGu(ioH_sF1slMZvrlDh>3w66KeV;9KOMnDFh`SCoZ(s%VNF;8XmgM^JyxVOKE-
zS<ZgV<I3OG9B@z(a%vZbPsn$wA<)Ofp7yxo?Jr*3{E-7#90P!%T#!b8EOg83x`OjB
z<L8Aj9<N#KMW39qGVG7t5yjz)cp2`EG20C=0u+87xQQTUt%w7^*B+OO;7WJ+yiE2d
zOEa@VJif_EtT#KUZ#a6fiF1sr`#yfG(LbH*DCmZjPYk%LuBtj8hh9C*zi!2YfVe5`
z&gw{tQ57n&z9bRQY(y=KM~0uzsbdiXatO-a32u~~bWgD9+}sdYQ4<@C$O0m{jCu_4
z8iw6CySk?CH%9A(L6J2veii3TwcFguK;!;*CYHxzPO~sz%%n_(crqH;A)luDEsrSr
zTH+uvv8Azb0`CTn8r4@+V}nk}!-su1zKI8cbPrXzV8)z=t}c@OQK%__B_X%li*8ou
z44Th+ArBl<&cE8a2lW*WCm)>Hx3-eSSVT;BW{rOP26J8}VBVKAnhIp`Lx{)7kQcEG
zVBzsRe)D;V5uuJacI+7RX-tCoDyXx%zr<zx9boW2GOTa&$|sOoQq^62qh*c7?`|qx
zbMng@DKt?u_r(M((WTmcE>T<O_lao_1q+3-{-%z`(;KwwjMJY6z67fHf2AF8m>$e^
zEL_**FDR>82A%W|`K})7;xj+4HS@hm@b>ojFlbadFgQOuS{D{b)klg~DGk4rUtZkc
zmGP@5+$_;E?}(<b;7-ve9KL>j;a5~*06P2x;>%;1z0S$UcR&7E27Cf?8XrOU=R$e}
zc+!1l1`7)d4ULWIOtb>ER!L%*Ssx(t0OpCYu?Nmh$h9DsM|sfI*@>F7tab4j6{#;)
zT5DSyC>68<q#A#LIrD?!lviG-o&b?RWrIK#y5x#(0*zeb?o`wd@Cm?Id^{7r1H}g5
zNWpx6>#=2DP3U1bIXGx(XwX`Xq(-CGkI+-s`dnCySuvq0{m`Tq3Y`fYGR~b#$<8KR
z<pTDPWY!v7zkVG}BW15+h5SUaKOs(urJqW>v@HwIpQ(RiZuE4wej&)@wG6GJm=KO0
z3)mU(@E~|$&sd{+Xm@Rgn&(;J)SHpJa9%w~az<WOONgPotc*WAq4UDl=C^O3;<Y?K
za^WUH_yTV!E^1`3a3{VoF<y@}+lGAsHvj_iHkIFsxhW~E@Ls@FfTvH@Au+}>fgQ&W
zW?$S~1V|*)CzIa-o$Tx^s`!xYV*mE#%UiVCSfFwhXK4rE!-d}gX@{GPWEKrQ%dTT%
z1f1H|7mBl#N%~5%y0wkXnb#B8Qq}GYciR^!=Zn!oisOSn8uzWh_!y6$I#(M+U>MDN
z`0z30_E@3fuT_-*MX0LcF&004nt5SksJtD>6*|2SdDxqo)m5#g5akV;L1bk&>G`6D
zxI32F{QZa<7dJPOjTR(Et*zOb8TW7Alm!mT#ykdXCm&A<``n$dgoLwiZi9V1ZL2Gr
z4%!V5J`!Vbkgwz5=?qs;QX=iWHwWcU((i9*^Zkh~qb;%oVI;2S0CykMVmJ_x832ts
z?0lEjS$=yW;gI@-2h-hPfiSb@B-dd#r>j@fAMt%3tZWpx-GaO)X2BY?RqW`tQ_ve*
zT5?<l3Lw_LcwCkO2{djq9s&|vq&p}`J8ShWDUmVb@PYvgv?5(yU8qRmh|{zXPsPNf
zqEkwZ8R4k6W9u($^+7)?Rw{lN4KnaCL>C(Ro6Nti@hRWNi=cSKzoStYh#ZzXj<-YQ
z1JBKVuON&#wU=sMzvYLN2!3nv<yFSh9_fc_Z;9<DiAZA08e7m=y`uo-PmSlwv$E*U
zFP`yX0c2UpBEP(U5aum7m7C9u?7a1~s7M7PSoK*%4(>2TW?1x49<#~)TnkM+OM&y8
z(OIUN!f1JLG<K<@se)*u^u)B0PIjdH$P8MB!un!EetCFFXCylY!04lN0$E*3`6w|i
zSt;-PPUP-nWWP!iDr62~IQYIqJ>i>oKHb#PeDTk`GYQq1(2zPe?Hl=W$VW3^y=N2_
z5#c&-Ud$P|P46vF68gRL#rD4PR<<58srFccyLau1+rxwYu%b}k$WBf!xcK7M!#;YF
z<+E@!f5-EXn6lfjay3~W@!dMyjA*!0Y|w_Zi*m~zk|FxBQIeW{*na7|FRmcX7po|{
z<*BWs1Kr)DwS1iA$Z~8fI%QI@n!orUP|EUdUQ(@aU*Hh)zF-b@pTq?HCXxqRa|3<c
z-OI~%-jbzEkn&Q>J@5`<Q~ednPyf8Qz2nyQ8;-khX%}__Sn4~Ri2<SWk9c>BV&0-C
z=fu#?+wg=ZCaLTjNXQRW=c<Cda^(UFRZMp5{Pcrsw}X0NRu=b+<3BfCkS;<vsS*8k
zpgde$fRBs%NGM>;OH-E5mlIgrC}d|zX1|4Iv2l`c^`gSWy3F(%*>}uZ{dWI?QS#kR
zphGgfqTB?AfAqVL=1an47LZO$;+WRTos+`=jx(iWkkK_8TQZW|$Uyds(UMUWW7kE0
zbW?SHGJ!(qnM(HR;K|v8w?1w9)PLKP>G=nM(dZTA6RxHi%0ky^sJ0vS&X};~#5H27
z^;paWoLLwL_i&JT9Hf;()<Kd?S=6hdq+Jo{m?4}L7Gth+s8Vcy@ehe-_JfdrNIt#7
zW`FZ0FTHeRx*J`!i<4VOZhZH{&QW!Isk`UjpZgV^pP!FO;gba=C0Acx+BJKpdK(Fe
z@Q#e+{0{WQpmSVhb0sko!Ipg9c?0knzA~qBZ+v-S{13^EeRtl!{m}H{MRNLa!V<UU
z+K~4VmA~$t@H_7NX9fP;|9*|jO2NMW@0W}_@d5w+QfJ2U`OjCx_W%CMf9}Qq-8&|n
zbs;t?ik6O!vr(0JRto4IHZwBT(75>?-qa-g2;7By`G>O(EAwj3H;{CmrRk;V@>Y#>
zL$Y@6`6<}J{{8Z7g(czZ(91Yp{_{^H|0kbRW7Lfh4nfMO)K5*k(gQ^<YAKIJ|B8AP
zR09K5=_Rl(#wZh{9WUl`lAb(KvLQ@qNk{}++S`FGKIYRydm~UW4Y{~Ki>d4G+!Nya
zf4*9<tN!!n8}kPb{CO?nACe@Rh{-~Kzqt0`METFxe}Cn_ckJ(4`0s-G-`X#ab>0&m
z@gyW0#8jk+2x{kc6nF2U<l#~l5#e-96q<1`jw|+m)>V@V0+;st{=dHggsmH9dgEOW
zfp$Y#0-$NU>mc#HrX8byMrzqc!)>9ppcv)oK?Y*`s%QV*#_E+t&<(tKbJjSw`FQ1@
zjr~mSPjE@<4%MZ4-P2PBCnFadQsULdw*CnyH}o4}k$#Y(>EnIO06<;-^&9cs8%!*J
zu2g**L*t4Gf0(QM=ZWEENwg>bzM@E&WTd5u$uRmVqKQFkBqU=0`3^Ci4f)XFBf<0n
zCl$87fVDM=Ozv%p-Nf&U*n;o`j$~o8>Gt2t;rtusY`OKQY=L`25g0n!K)m3Ms(-Ib
z857;)Q8-k5fB{AGH8bLyGSC$gr(0$k09qh<hX(CnKQdlP40Rznc=YdC)@l51x867S
zksuu&IF9&@>)!nLy8E(0nxSy&@}jYXBR5{?^}jDz4GiVP#pgF{tC4$Hz+Zs0kGKMo
z2mAk?iN*R?PE?{~LiM5|OH7R;A^Fd3)|noxECoisu_W;KZbkij|8+EVbtk0Pwa`(8
z;nSppe_xRD-xu7!!RKgfYz**&bddhfTm9!dk3cx5rLiXp|NDAB{(ZfYy*sv`z>O>;
zRtZS8$fuagiR&i&doOD?dy3cWrsq!rAFb5$_a(RfeX|QUQTO)t_g~H@I^)KPVHsFj
z|K61R+xhuN8w*iV0O!!^`}-6+|K68napZ<Ow`_K9UtM0#grV8x>%_0hi2uFSp8JuP
zL1lJyMz1Z0SYflpGJjUer)(C!slO{FW$f=tnfm)&XS?XveZCG_8EQ^!NU!soA7v1H
zrEPdazt^8E^ocfgY4^Q8+d+M;BG$B9*OQ9#^y_6`X=r8Y6!i&id37_1wTyyd$Mx@^
zHxq|e5;hoEcAk;(k~}b=n0NWnw4s<7o9@SneL|L~?>gRN=2TXu|5c0<BYEB^tJhS5
zDhIYkO}RY8CA>{Rg)NN;8)og`;DwT3B#W4ANb(j1Eoc_MD?*8bV1LHXs}cW7eBDsH
z2_Ei23}b`v-+#b=oA!+SPsp6V?e;%0W9=X@E8_3hL23X0iL(FwmH$WYScr>(F7dn|
z8A(V00k*h^A?IOXR4X60OI=0xyNHm`V>TTvt*?WX(P40S8-Z`j(P4Bp!*n1@`*$Dj
z8YeO`&r}K=Q8(I~fX%_0ms?^_VtYRN7l6Cx<q=FQ1WM33XljOg+AbHs{FbmEmi&#M
zJ{PbVn=LCdlm8M*bs!tG&xmpn;{hh8Q21l0p0(GY3@z<ZV|xg8&wyaF2yTRp@Ox!Z
zywL<hNcfZwD?WeI4q{EtGQ^@>ffLJ-8Dqj`ab^HQG^s0Jw60t!KFq@r2)q~uV&|xW
z0s}E_q?2PQlp7c?V1Wiz!a=ySbYfw}{ucllfd)HZR1_h%uUdl&2PV0=2+{{UOmnRi
z-tFKX&#MFgJl%(Qps1;(74hI`4Xtze0319@w8!DUKc%gtrsf#l0+=*zj@oPP72P@R
zS|qSdi~^u8S$=()ni80ajEUijthjlYjg#~J&@*6Z9%n(uHT?XE8#A9@q6MYqojID=
zn3$rn{G5%wcrO`KJf@TTiq5waf6)Tg032@v4uJg?tcjKA>9Md_oFlg2jS!kYn{=cX
z@C?+={S~x{!idBN4kVW3{h9r7<BpRf175@)aw|=Uiwr)D6qeuyp}Iq`D9PK=WB`6h
zMutT4Hy-zVzDisr%JosE%jeth-bs(-h}nNP!T~D4L}(-nM|xV?I8bsZ1sN7jh>69<
z#(FK!${N#b;ejp?iXGg@$GN<l=ez&na3y&P*PVeKgP|$8LJaDZvHx>mI6?V4RXdBv
z2i8su4m=*-TetcG0WP5rLB9=*X#GGmprr|<H)8LzXD-MopyPtC5nxS-N!%q8ii~}?
zpbf{w%<M7zq=08Zae{*oTmeiKp&3z5^F-|f{R2R$-MjCBKZ5<m5FkqZqwau90P3UV
zDO}0M6=?|`K0vB&jA`%c$_0MF%{}Z=4vLkFk1uL<7;1-^uV3w9?v$DehjTCVRy$XU
zg4cyVHX)3HM$3@zIOYMO*~UH=Ae7%OhIg3Uh<62F46BWGSU3YQGu>bC`@*mw1{9b9
zU&PyDJQ@yOTwqGdIgk(_i;a!fK^NeI;-i#a5mtapEbK$yW@&OF9RLN7-U_T)Xnw)V
zHKlCWfG_^FIAvyH@?vxjaAbH`7!br%Z&X@0e(aee%5qEa-Z0~RR9ae!E$@|$cs^>(
z%*+geitye9qH)F5_0EBmii%bA8sV0MCq;g-xC<p4IMMu1cnmC(xA7md3(0t6C}N)O
z|1dcC8#;^qBKGG2JJAbC0Bi^8fOQHGlX{)12d(_h;Ox=k2kPx$1|1Czyv(OT59NmF
zxc=>#0)A+3Ax4ecKzs^V{NBEO8z`dWKZ)%?_Z8rUh(Qvvva&&2(Itvh?%jL2lP9-O
zmIAMXMgX%NAC$ri3oc|-YJ@>kEeRe2obp<Solsap8X*Y|wMCE)Q4dfWM75k^9heq0
zG~@%y1qz$7MeKkfpN+wxp;{R7H?@Y&(C}>n(2c{wV-*r(0NN^N;Wu;^`cQT$b!+QM
zaGCxY$YA@6Y{EbJ7XJRV7s50cK4Q=(Xd?!>2FNT3lWmET(o(C#=J?5AGY!I_3DyfQ
zMrrf;L_mwdSs@nBqPSRiJW(qNw-xIKgcfdm$`)HhaiF88&am;pV5$on6?Q~V(P|VS
zLGWLY!H)A_UoS82)Vsa}7Rtfs1fH-B4pq~&mVVeh_wyO@MLS0m@bc#&OjmpGT}K4?
zE(U!atS-VyucW>&u^qnQ<~a!vYbYrN`ul3>>K-+!h$tZZ!i8q$U(sEdoSK3^tb$uJ
z=<=6xV%T}C-ETWPBcA0aBruy<mRvq50+tnqpBm(^Cl;X4BknqV;odnzXVG{F66f3`
zR=Z)$_JHh+np(yLT1|$XM8O2|VQHLehuV>H6CIq*0KK!s!ln^c(ClO5ia$v?aYPnY
z%a9VsqZ}c#3|jyXLZubSZ{Nf3Vgn>JJm~@ClQ^l~Uw8%KHd|f{RErw-XS0Js;r;{M
z9i*6ogrThLO|I6A9q!vPM})w=*SH}32sV)=PRwW`)A=rylaHkCxe|P7E?h`-eveMG
zR5y%9?g!S7azwd6g%sxzY^eZgmVystok>Yfe$lgZY6pyb;noQPe7F~y8)`~Q7M*4$
zatN%`%gbZW)d2<cHy}R*dJRV`1mz_?O98+%Dj@Y=US38knc`J-kPLv_Vg+`wnA)4d
z^9VZbU@y*{Sf%Uh>rCWU0I7X`FGT|&e$YlsOA8R!wqP`_Y|P2GR1H`loarFHu$>NE
zQv+rLXE|u0l*zyB1U{L8&3VVyS3_JYu<6w+c4&R@@Lv`})ZvM{M-_oQa5nVhpEDgG
zQ8%uXVS6?$*lX{#^A&#6Z<Ff)PJ%8<H8^a7SH;H(K6*M3gHRmJ%nm8N$*)-r%nf0Q
z!*a*|(4>OMl9<b=Bj$?EIS|L#$59tAoH$Vls}QjKft90h$_41mF3L)@nu0*q($KIG
zn1JIG=t(ri5`6=vmSki>SQN$SrorZ3zNJ>UZYV4qruCiB5CkX1p1X~t$B_r%JPPRd
zRfV0PxVS#1U`4dy^I?yLCtboOIB=+qDDj><!|<W~WjfS4CME)~*DE=w4DMgRC!AvU
z?%l9-!g<Gw`;wDmZ)*$B6v9%luu#a1GY21-FerjNG^~bP2Z}Jg%B#7ZL21+A{3m09
zW@BVJpUw*2Q^>OI;2z3vxLdfQ!<Vc>j1P~#+|P>GgKpmxKfmpX?Q~xG*iO7W&ptBH
zu+Ju=<{2pfe+)ZMR@>L2@enQkT3JE1MB3i%o#!Mry0C&g9ntF7?64d{gOSlA9C!$>
z@OH-so?5|hKuGHlI9^qpOjVYZ9R_WRT#G6iLXil_zaWG^vuGq=D=)@$A1%xHZkp5~
zHiX+>cm8DVjo5_E&1dS{pyC6MOBsg13cX?>mL9^BW&Cy{h+|qJxDgb=wxrq5&&~A$
zln-*ID4?gO-|^iK3=th2T{PU*b#*iH^WlDEXJz$aq{#YZHCmexI1puEF;5S1yxFH_
zOv(Ysjc}Gi_X`0E(wQqzFpbp%K7jB*u1P=TrQZ4TI~cg4Vdn_OJ32Z8)+S(}00Qt^
zVx&(Zzsarhm*4HZtrGX}Az~=l9mEM3WkY33*v%D)v*Ht-R#2$++4%VGoeM<pN5~a0
zp+XjNzmP^oA<JD}-d^N>k705>K7IuwQyK>Q`)T+N;p!r&r4l!gK7u%xKYh9|x4;ha
zl`oy#rVO;Vphbj$0A(h7ctzkv(9b6u$fUhik731BRSneFo5MKTevlqICpWmD(nd$6
z<I&-mM04EOoEs=V`;F7s8mEaZCbjoZs^iZiExm!h1#uX#K9C@+P+TV;%+m0<OMJvK
zKpPHr6IUE~I7sh^DV}coC@DjuqBz_Lrhc{U0{rRVJC$<amIUQGkujT4-mI#sLaIgB
z>NJjqz6X~}G+kqw221{}-64$zAgr*n10t|pzkv1%uHykaULV+q(;52=^!yEp7qLkT
zlm+@}WjMxrZiPUDhX#L+pT11s;t}%bum9e%WsBm|Q)4&+k(RIh_-5de0?*U&{gkZm
z0Yr03>o{owAyi^vxtTk$KcsPCZf@u9-82#q1bY4H6Xul6K);un#MuhnG8K?~xQ?Ce
zg|1WQz;Jl1B6y+IOXe(;R*}WUO9)MO=wjmIB`^a4IK*m?)3s~YSedUOy}5CN<i#j7
z4Sl`69YKACR(nQQpQO0HNi#F_kF6Wl($RVIP<R@bDIbUpn3#A#LsAflAbLz8FgMJR
zqobug`|udIh)9$n!;AR4)DS@lqZi*z&x(i!0S}6qpAv(}JWYGLb!jwU$cj8VO`D;_
zLZ;VSL@4iRY(m1vk(*E_G3;4{4e#nmsyX;odrj4X+%HWN3&_f*RtR}YcaVkQg(YnI
zTe-NK$&MHb<9{*x{B$;YZJr)ETVeNsuc1OUhX3Zb;`i4Tc1r8wSf{c$d7QkpvVCF!
zTYqXe6|F8$_dA0_jA%n9VSi<Bn(0aZE55C!UR|`9C*og_*?sizEHn`w1;`03+C0aa
zC>v~X*cQ1Rr2L@!4gNf4B2tO%A_e>VH|)U-n@y-TTzk+y0WKW*lD}Kva*+iwQBlhH
z<p*7b4I(|%fYJkBhE;CP$cq*W?1ORY;<W93e0mS&N}y~);+ir_Ww3_>RBTpPOx@H_
zZEb8U!)Q9HFqj1G>@7vYjG0F{bvU`cp@dBE#x8}D=iPhu^V@FL4Wq35i5~v%omqRu
z1VlyM02z(YE}>&UkngO%re>XpnhugZw6x+N_6mm@8mXy*ge&s;kPtQecd|I`1Ps)w
zg*xD+$8dWe4s!7L8ElwTnv|D!U#gx>rsmgwv;f-mrRlHpodSIC-0_4>^&+7GdEZV2
zrRZH8QTRf5L9j+fRxZahTG8gP{8Cavg8n@euaNuEA9sLxATryu3-rj<o)s4(vCZ=L
zTgOmfq!*{vYT<2wML9JU=EZ@#=jlq+#3U5$Xh*<gC<u#ugZB}rSx87bq%I7c6IA!-
zqO0oZ>rOF|6NA+7krJu*usNYorQ!Yj`tR!vYxe)x_`<A`kV`>Q))asO9q8GZ$+Z+G
zS66ct%mJkoZ0to#A%ogs1DjxQpByvijl>(Y_w~HYL6rn|;0SpfY1s4lh??(wYlXi}
zkVdzmnxfxsbgB25!gAj;FzZljEzZwF%}V)<X?uUoNY^=!p1d|_X$rrg3-<?N89(_y
z*IZqxC7Bz$9W^QvF)x_p!2bGC{7)2x7Y@CH3COXLJNNbUK!7K&7du?(f&>QRMUba2
zs^UAC)ph>q#Z6v6xF|!dPT5H+aS9Sn;HhrZn@3~R$^GhR`%NXPQ$#UU8+#UlM5VK6
zfMGPSIv`>{ftGKDCS&^-I~rn}dWyVdxR9%=s-|T`M@I{N#wa`iGx5=fyX+lRuUEcg
zM>837OW489kQd?3Q$HJpm;|P&Y~=LC!(`FR;#R?p0aqTieiOoa|1^5*s01$Ktf?vM
zXuJ;2fZuqa^wGvRt$rgw+1c7Q1?=0m4|~!y{_q)B(<3>(tR)xTK%{U^g7ju?ovxhR
zZPz`WZ{MP)gNPL_DfjN}wy<2K6(H9ccM^rI@W~&h2Cr}A9y?X9w4W+FGxH+T@z+;N
z7&6R8{r7IN@zX5>d*An}O#KXIB8-x5y8~@>&iR5AVZTdtbRTs-xT7huS~>kjr}SQY
zPAR^TD}~5rmUF7E|Adkw$T>g`rfweXq0gT`*SX6hb=BU>SpLEhEmuH+$$mwfH*cn=
zbD^O_;_*zI=O#^wsPAMChk*Hse*5qDCnX1^e(-u?PeLaAIKvU$MyM-JUS-1AFW%eE
z-@<e9FSFdPv6zd+JI_+e{y7#*$_!^v{h^>!8ycIOL?@BLNEXun(O|93m@Dp$ZpG+U
z_RrsSwX_I}!!%<P(pB?~&GtbOMO6Lm*Y(zZe4cYdnwp}L)D11}_nKd47f<$hkf3PZ
z(HrEz%3t!1Uf@0zP0;K`QX}3W-yiP2L3~_X)+o#u7xBihv(a0wIceMk>j!L22soM?
zCn5O4AmhW%+G%;Yd6!wDx&FXGv422{B$#$>4W5d4&<!phdFAu6UvTtAcFaA6W$EIt
z<5X_gu!ud<y870|3HshUcebEc!({~Qgu>CA-`4o^)_O*CN+u^VlD&+FSz#53{=-9(
zUI6!lr_9l~3&G>!BL0u2XB4-$n6K=U^l+^VY}{G?##e=X#oc%U_iz5@A9C`@#c8Ee
zu~$%57zg*lTvB<Dwv%W)(tm&L-OeZEc0N^2zXa2M7?JHHwRc2!yUZC%yY4kytWY{y
zA^`@(Ipo7}KTx|bL$+JD>?^`C|BoB}2!CcFVu6ra3Z`$s>KD()YON6oAa%7z*l)Wa
z-)}AEm0?H%fgk~EKc9xhe`E6R8P$L$%UoWDv>~Q`rU`pqdKAw=+Tp9K9v-Gad$TiA
zQ*oFJT!MXf1P+x4UZ{kznYlc6_siSce?<z%J~x9W53zfChMl$bQ&lbzc+#W!wm~)b
zf?%;^_!819qGE6yCr+M;u*Z9U`XcHeJoy{THTXENG)9v2M~b=4e#)<1f9U)kwH<U*
zlT$>+s(k_Xj)LqDnw5iQpsT<V3B)cJ0%LY*i#Wqt{9&;k=NlxWf!c6dCG|+Td2^G2
zy}uK>;Tsl_vO&K1w!hy%BeaiMT|*-lqo-gvG^!Nw?OtMHV!dSWvZuGa{Fc#la2-S|
z*gLX!t*z6jyn|Q3#Smqpp0E_@vznI{CIGC0l{pV!u+gHOK@MHslP4aqG!5l#8h*NF
zcUhFFL@{;?lq)S)o$vbjg)kn!AD@zwvxXA|`F)sxkG%XHCh<kwdCz%%9|AXT?Cp(R
zdyEtt;>Hi?$_5E@GAbeX{(TN<U`fJeC0OuK)1l!uK@rL13SJ3<y!t0><y)QsIXi_k
z7u^WS$u|`T;6gWd$On5FE&--YuWT+|@<yWp($5_nsZe+THTvF_9SynYLDChCEin+&
z0-3-e!QDH9BN|O>Am`E?vAEYTEegdI@Fdtf3=9kqO{qHfYin!GX_n77t&&EI!}kMN
zqlb8OD{wNi;P;`vckk9*|9QV|{Ur)NtYudWxMzL!@+B+WOq`#IvGlALW9|#=%^pTY
z0S1<1gZya)#ky(aQE2jX8r~M7=ztCiHd9FHA!h^31d&fKM=jxkLrxy`X%aU}=H~ym
zV#zVWk7Be#GdIG|r-c<1ni26R`XFP({ezOO7;qTG*eEpFGRFWi)U14F^r*laK@rat
z0BHH(h%<mg6xW1%D{pUGn?3xnskE|wFXP<8^qZGueKJAl$AW3n8(P&uG>%F8cnA3s
z)`_T&sF;2qVx8tStcZW~sMMjqp$tX#*_176DA#!X8;RNn`317R40EixaxAS{ctSsX
zs1LOyyhu@{HayRBf_t8+3QkI<zqUcm0Z<Bu9cI1;4J(oI-K&A@y>?buWeaL$D1{Hc
z=peT3=0EUV0pLVTHH-+0LHbQe`GUUZoQ6hnd)m@)-S(15R(c>c767Ogt^K*6(WV3X
zTe8VI@di3Uk;})gr~7kZq*z0PB`Tz_ZwaxnVkj8ZyeSfM@|~toRfU<_APBF*=#b&~
zZseGdrj5ZVH2vK>B2T9z^FhBYx*T$%YjN{|XPi}0DZ=E>oIA)S_<tOS1e0Xv#dRW)
zoa_AXK`mC$*ZCCDe1UT^@SGwc^C}!fjI9use2Si-LJVPK4&ij(`OOevuwUZBk&%TJ
zpazz4^k>T;&cDwj_AE73{+XPQp}7a306abl=K^x6#9H&_Fw7?HAe)$T1K=82WNQj0
zi~+8SND0|{2rL~9=)cEqzriPQ`NLM-peM=`*szt9oyQ1s-3wl?MEa?R3CD7jX}3G?
zEMo@%#D6uYe4}v`CqgeUO#q^-GdO(-Gn{5}P9L<DjJ$cQrek10NatZB&KJ!)!iV-8
z>_E7BtK-(xH~uceesL@X7Kqkbc!Q}9_uAUoA#PkBso&|`3diM7&~CBkKB7m3VrXOp
z&n;BZoFjawim;6z!hPJqAy;J@gzl`KEKxcpSs<f;(31K1rFfLwFoJmy4;%ns8G4uD
zPbTJNhq*FH7}3PQPNofm2Q=3~=B7FF@jSvpKBVj@V*nXqR$NmQ#yekuT9L?lkaV-1
zP>-{I9ZLzZygBfrn%_VAmzUk<SLw|?r8e+6aZKPRjS4MQQQO$vU+xH<qWPFWIC3*f
zu2z1pxemp@mmM6b0bT=NGp|0p;<=8tQ1p+SjEy`;i!=%+`v(9?$W;LB7F=<+v$x0e
zs*eyh0@`Eu=LeZ|!zo6ZsF4rhHYtM;Vz&Yhfl32lBZl^3Nnl6HODjQd`qztp1j=XJ
zr2PB~2#q+;V`I-Sb2p66A@iB66k6}&gd#l`Kpl>5PzLxy!rRx^7Z}*dXkTwi9!!+<
z^qy9zP>|>2pacq-o(=>h3KaoB?Zm{RYr?313ZFa)On5^}j(ptIv%p^H{dU4`KU1Bm
zwWa0GhbL84o^V~a&o&5L=IHL~B2@aw^8fM&kVN1;0o!21;^`Uk{e-13PM#EM_JcF!
zrr235xI`t^{>nA<i#8PgMjh9%24E7u6(P@JWk9RDj52pJ{Ix*M^SgJqY2>A06Z2c{
zeRNS^yC8E2?6Et5i=ln&gs3Pz1H;_P8XzIi5=tcKBMo1i%YAt|Xkldyc9DenM@x$>
zI^jWm<1Q?+D{&pw9LWx99Y;Appvrhpq%-&~{(yChSdWj~+uIAugCbile)4;gjK2Q9
zQO3n71s9w}F>}4hl&f>?kdyxXnhl-9*kY>xpFC{k&)WU_eHrgRNyp#6JUH;Di1_=L
zvF(4(wZDJ)@2_AJ{Qdgx9s3`x=>OdXqY>q%2Fx9Bg<=7rkT_@uY?yGt$$(K3)FuY&
zOGLcAdI^IhvE@=TvVUqFpA~%I<3-Y!e0R_8fy;&kS3@P3tsoJ0a5yqIL%fC$Hc#*r
zprB=aVpE8tJZo@O5LTcuqY=~HmgF_kHCI2l^{>R~&jbGNx+k)PzhD2G+x~QUE-%S}
zShS#khbLpMEHCHg9i<B6We%ZZWNe*S2%rXbOfWPikt8x@$Xd$EL|p>WVvAVW|3r#p
zJ{p-A+SG@gP$6OHXn%!x;v*zIkg&NZA%jDsKp%eWE$cC65~=Mw(r>sCGvCp5yMsA8
zDhlj<v4fY?aaeE}7Fkn@jU#^fH_8U>V`MBhyo4}$`EnsDC44=3Vq3|he#%Y47|zyz
z(&hKv-LzWTZ>}IgAm6bAU5k97ruO#np`po-A0L`{SY#!^oS&KL-#U&Egi#KzUpjXV
z9Zd`aYKdma%cAkXB&6NvQ**Nydi9V*BP|KYK$xko)+S*=ZUK`_d?DoF3g9S=Hp@Qk
zn}Y8^Od;Nny8xU=D23a^*hb{NhC0qE<@ZOAdS4C=qfel!5wpA;*Lm7>NX+YJ-KPUw
z0lbfOPNkhL4&N+I`%Rpg@B@AM^5hG%?c2Ab`=%s!k4!%%n+_2UJQmMyZ2I`|V{mNq
zh7?f{<FniQAqP-ZQc7V<IdNMB2F>8zgDRsjiwP;NejAL~+@bqN8B9E-f`U**WWlcR
zdw@eAvTRVt^3r*oWD%iY^#J-Ie*Z>OA_C}`DxM}WSOw)MA}qT<iw-QDI5~wb?L$)6
z|2+`RdEkMm)Nm&->jnPL>@ywGlE9eJdjX$HLdb!;&>V->dYFg`Um7R_Az9RZz@Fp)
zkik1h7DF`&y@{n|LyQ=7F>AeNpO%z}z!3ceF~9|=7<T};w+o(gzROEnLz{}`W>I{?
zCKA^o<-HaY50q|Zcv!&X<sEcgAw}@_Hq4^qIs~?vm|F$Afph1mLK+Xf19Jl<8RldV
zUQXss6g!pPG&ckFrjaxP-H7Q|4*Bg>bDuhU&jhBWjiq%8cbzGaA<44-FemEcvjG-m
zCo?h?0B%JNUVZW?W59)fDu9w7lc4&$xA(mTZ&hZ<CuRYH4{PG@!vW$3UuiTnnGNtB
zKYk4mU4*==)q^b5nThB>9jJz_E8*ZJLCn)ZBR4!KC*!bu;mLq6_!i2@JV%}&A8<{v
zT1F3#7{$yG-U#R)y%1px59&l-9^@;Ak$8jVH;vr*RyuUfyO#<EfZ!wOk3-^U4xAUw
z<?<L4+JqOYaGhf9?vQ53$^@>kgPa^wL#)~`UfT&P1Vt6bD_sNhjT9%+Y7p{}tsx+W
zA(G%Jl~X3$w?}4q*zZ8odqXZ>-o5|qpG2ekQF=WUq5~0?$kT5$`}eC(RkaU0>7WMp
z+3=p5CSwiN7){R0BTyx9nOCmt3JpRognmV&w+_hPk$S-?#CFdA+)?#Y0QlkVN(Fm+
z@;XW&1rXg}qAK7bc=tRtYa~Y{*Flz_m&Yt%3fb#6ukc5Y_V=qwy|Ef{6Fq*fyqulk
z=iV7T3%zrPDP(w_CdggC2CdQXh8pI980c>AQ-QP;>0mQAo=n0qo3OHRb1N&&mYZr}
z(05?VF_>AIqyGtWVTft~2TMFzgfFUk(Gw@ofEsXw{xs~P>0pYavjY0p{zVyWu5tCl
zwW8$YVKmuC<z%YKfb_=jKrlp2FGS(F2InI5zaVG4J#d8>NQ;kS0H=r1b~e4H_IA!1
z_pOf)ON(1e+;|QxAiBt#&_JV7@boD~X*|dnK<hH-7y6NjRfg99W*%H1^bEQR46j_n
z{4y6l^;0O!5e}i>;ch8yLurWh!J)H){b9MQA)xH&%9PbdH%Jwr7MykNDYR14)}|4&
zfE!3yZU!Y2lcA0dN1BK1btsoO3hoMFP9)LXlCZUfaM(tXgl<@#$Mc2Ho`HyEotweA
z6EXB0A_+ac#L?V7glSxzgy*vQtd^{-32cjT3<C(qd>b`IMK_>FHa1fzL7M7r!N3;7
z*E>Z57B+iUp<LkN#^#RB$;4%sR*1j##%Xg8vxX!LVx2ste2^XkQ;RmZDd+;h3Opo`
z*0*G?!49{(ii-#n4j}bBw>-(t=KeW?bN6#w>>)iHyfb>a-euAioel^HAjpXf6yp7T
zedJ7{51hyPaDE?RWgXK@I!qMF2@<inzIX+K$OIsSqfuEz4#ZuY<aLq$CM69n9-edS
zRzMM?>fSs7*?|Y+{_98XCkX9qJUxji(Srj6)HECUswsJSQt-ZlE(LRC^6Psha9)7G
z&|bf?ZCR0>jZOXY9S}05u!l2G!?d~yux@`oB>0^bbVW~2VM9Rrgf?S#ANjp-+EegM
zXswXZ!wu|0N!?rVJr+7WnE9MI0fUWu4&L;`0E{N(%3s|=SV@1DEXaK#VE$)`miGsV
zqT}B@00dC%AR~J`P{t@=dI#uv{Z<s=PM89b%e~3m3h@K=(aVLWuN&JY7I01~9&i%y
zT%HAEO7?LMff=L`aJ7eFR`?AFX_DJ+a4c=MFvrmbMb_dX=;K^W%=-8-!TKkJTCk0W
zg~_QVLq%6aV=RE&oH0x`0S48?U5m0-P5O6ga&kvYm^3Sz2i+DYLqFk%=CB3tsPN?_
zCCv_A8$ugk+e2Yx{!Nf6Jpgs@?rleM@iNdprNH`Y^IHfzbchI=H}5@Ymun+oZeh_;
zdCo<Ya9rHJZ5wG+CvYgjhtWdzF3KoyG^Z6b=wu>1Za#xx<`|l)S|;~`g-%Z|ffW^s
zYvL{{!oCNqgqVndcrSlAv3&wdzZAYbkxOg#jC;=XcCvroF}PI8Xmd_ChC82E?Wl@X
zf7+wNpJ$&eceO1DLFV`NzGur^8FnD3#|1YDq(R;F`<FeXqFd20JM%XAHlPJ;T58_j
z&+roiafjt7eC!ySR0%u1hbj8*F99YXjX^bxpF_%Ir_&kPU5Sv68=;P--P~LUTiZUJ
z?|)3j&%}y4o7$1WI0RGN9i3%Kc48po=ZOgz;UWnk0QTu=_}TG)+#E@a0mO>{`GEuj
zd<vVvu^X-9=tn|E1-%;zS3>HT-NlGfbMo_Nj@}yl5PCTo*gFOW*1+LlN#4ppWsT3O
z3i*)AHZ6n*&}9H$VrjX^0F`A7@j|SM?_+M82s<cB2RN}DIu+c5d5r#Yidb<n>qad8
z{O=5Ak9_GZ%z)Ht#S?|T!l_el;#bOx+5p2eeeVRkF0VJ`h#ib51@P&!lc)r9`-czF
zHM?9-EZQ7rM)xT+q)5Afkm9ni6WMwF8b{GV1=JgmCGKQ7!qrF=p7kPh3Gf~SzJMYh
z!*ZZJRn%dSwZR?$R007TR$Wxt{&VW;&&OTxFhE&B<~n}a;8sde(IPH8qFzUY8E`GU
zflt8Zx10JU&fbTu<LIzJ*E%F~9k6a32@2GhX552YeCxBSg+&@r{?Jg3>*rPCbA<2t
z;j?Jyl;JW!9|9y}Pte!XL#)ds8hj%o$L!HciL5LoCyf`IY8Z08MLkNG4nuSxcJVoh
zQbkS8I!<HcDTFcvFA#8U9eF4~Vt+#3bmZ*I*2zK#i5&q3TYYFGx~}rD3jms5f3-q&
z;Ox!E7n4jbrZ~Kum>r6$%C*TcX|+i*Ny<;o&rNwfy6|M*la}cgvTQPTsl58m6Pa<-
zF?QM3(VbCa5fjH2Ud6qY!waN1q-3XOXQk*3X<ZmpH!{>T)EJsyYfaJAI4960(xv<1
z=o9Xi{1!)h%kxzRJe?6;yM%XN>2lSM&}y-54YUp9ndUY4nCrDXIqD1~11<x!5{RUj
zADEYw1)#7A>}&I0D+mL<FolH}-TJJazB&^E2#BE(z9565=tYr0hrUmEe6X<GEh;+p
z;xz>zD>SXMEJ4`FCS?pIMbnX5=m<Z5wlwuRM(zh~Y(#jt9CtGAGD^ae<S)1m9U8^p
zGl0kV7-Oy&#fsShZFv5`idaQ-*rbLx1LeZEaRB-U1@Urov%HIph`?KC;MU_zYMwAU
zL<oT(f5gfI4kK2L7l*V9g6iBYwG3>;;1NeYH+T2qXFW(s0k{4D%pN(ugPNL}aQGtJ
z1HKPsxc<UOP}mafhhPa$+joP*9foTw`zV~QT%p;;k14_g-V6L1vxLAmKvlB#`!{0A
zN7LOXjNy=xZ&*%6M|WB;UFY;^D`=x|ml805rINJm%qIeJss|NDQ6RJlUY}_5flbrx
zu!Yhg@5guAQ;m}jGhq;>RBjI(q$Xd83Ia_Ap+HOIQ0t%L9C2FuN1){c?J;L=xiR_G
z#qyhMX0LI_Zd9u|quK*4qcaDLr4%j*3d22#mG$K8D`FNF(9IBu;&>T=gvb-NglxUA
z-9AlNM1-of3v#1S$Aem|G<W}_1xQOz$H9Y=WG^?woT=?T=J1D-r;^TYsFU7Uy9_N4
z<~X*nrGm;}z^Go(i%k3XD;(xIapD)0sz^5KZ4;1Fql*VwBQh$Fph}EoNs#tg2U(7|
z#j;{mi=BbA3xX?rSYOvss;UmAcUj(04q@vc=KH_HJ?6*iL&PBFK`*DkfCjc?!vs%I
zb@wjVi=d!Z>^InZ9PG`>4=cyo^YW4uO9OS7BLCd?7r|1TeZ<rI+@}H}A*=#o9s*K<
zi2GQBpNSet!u=TprLzrjDvWyo;~{^6X5fs9N={y0dM7cg3+d5;>mf!cn`voJsr_79
zN=uE2i-Sw|0xWbQ7V%iaPB@*ZmSIS~b`Wef?8X)^GQ~oiU0i%6$QSx2sK4*kqvwAK
z1dGyu$3`3Do^!+2!-^~CGVPuyH!DNm*`VKWxh7ty!_t)V^~ToTp~xD>x&9q4pFf_e
z_PcUGB(pV1>XeuG6HT9Oq!piERRxyMk0_nly)^2tqO8!`7vu5UDZ4C!CA<2b$Mh48
z%1Wu#Sw)I>F5icxezo@At|Zs_wDz4wyHj_-`^qnwmp8x8+moMQS76^x=4Uc^S4>}2
zpF!WoHR)<WTXvf=uVRo=NT557=8#tVd4H~W@e{B2i+*_Q=NC4z;eEdASoeUp*DvdL
z9dy1&+4!PTD!qT&{7x6s6I~LX$xfZ!__Z}!FM=VC=KAW~)K81=X;(z`h70#fov8=w
zdgm)I%~<gv5QYp34>IVF9y#K*GM@%!#^4dAzbq^)^pc<E<`EImX&SUg$iuXnk`$s;
zY3i=^@fpq!mADu<zfYlx_c|=fhW(QIVP6@vn;V57<lIGWQTmr(PESvR>4Gervx)=H
zH(1ytN6qe|@FqMwr2i7ZsD_ClmW7cIqN1)&Y}m{>rQn~or)@?O=iq!)Bg%Ncw6}ys
zK2%g5BE!)I7tJ@h3&@Kg0^wlzH0Nfo;I-7HyI74p4z-$Z6%(E+++>EW4#CmCtu5!(
zK12=78!2+TlonJoBMZz6qw4_3gGxUD{t8tvC}WL4TZ}6Yb|V^j6UuXPg6A4}qFW=4
z{^ULKAt5+F;OMTHwuTXi50p{G^CY%gecf!Rr`Kz0gIY5EYZ~OgE%(pC*_GL29E3Fj
zrAkPDXQw=~@)CBJfAKH04*Gl55_H~OTpJ|={1$zwwa+j6{47B3p!jhSq?J$t6Dbk{
z)9CipIdyO<EsnK>ZElUsNIj{sKAecz)73<fNj5Z`DtU;KtNo#H_N(VnadB}E_8d_1
zLa}>?x(wn5$bp6@G63(R_3q_%7YLevAzQO&SFE1nhv%`<R+!25jToo6jt#&L-Qs9N
zO#^(ns9lnBat?d!W>~^#a2_5dMGZk)31s*LM=i+q7JL#8*CPoAQ(>l~!#EEX>C&>Z
zTYD}(zE&~jB#PPIf=AS8{bTnw;pk00=LUWQs*-Yo(h@2ufTyTD_fyGZ%p9<Q&zJVA
z>-d_{>!o{5Y~zuSM}l#k&(J!rdGspBm)g4Nkq?8{t|VSFF;+0zYN4rPLt8K!6?0_D
zcZxM3t)ih~|HjPIK8II2_s&>xbo><mImX1^U+cJ$W_4Not9jwj=hfC}`_uUjIYv%4
zn?{{~`7KJvyTur=+r~-M3i@=6E1Y~Y@n+-dyPx{b4!&K?*+L=hPigu)2a2S<YtL^L
zw7e?yb?sWikc8B9h|-ADL_&PwN5><jyzlK9>b@g|Q)7u-OYOdlag59&_A#I8&Kek0
z`fRMj-{jP(jpHQ}!oppk?IY51K*lv<(019iCyyU5fT=>^fdUNVCW?tIzponpA}A10
ztVm>cB*3K&cfAfFwzqdB=xEJB@$ES7QM8jDeG<u8g;5z*Z!=JBTzohTRaIErZ7eX<
z5r=d}_NFX-*T<x45X2Zuc6#JT8QM!wZldUPu($VT;YT%xBQZUMg=Y5W&jahkxO3?v
zI+Y5G0X%Fqd$S)umh<)i=m7!A&g@q>`r=WIii(RP?|k60H$eky9Ca3yv_={%r!GvQ
zr2)g6P?!~107!Xs%;sGXUE6U3&}5c?DWP*gnP23VgW4MQD*_*nMrI(ELsYMCV6X>n
z6d-Ys!*z9ZFo13ZX_{tr0D0`T>i1F4o(YcL!-qDSujkt0LZ<x%wKFQIsn!&@9|ib7
zL6V9_7-*`hy?^URkVSq)W62ovEhHl8Lk)DO3CS$@ji77NlH%&CMbOwwW_n&<pPW%d
zC1<c%S_(?F3*0&?fl|W_b#S>QT+k%6b#>`HPvMS-eps)>#K<5!Cj8hCkv#>gqi^S`
zt1A;)4a9=!nVD<c7c5H#;ol6#OwP3k+X4U}aS;Idntr}R?E=+-{7Quw(&#`M-Yr`n
z#>8O$a&J%114lvR|MvL3wT%yzfqbwSef|1di1hlwc^FHQc0ZWvc)UIIh@|9TSJx7%
z;b33IQeyX?-Z&nrGim~Tlc`t8d(1{RL&3?KM^5;4!HO}oa$i77<UK!7W~jPx8@fB6
z-q$-sJY8^M4(sJ?&Z9RA?hWFkZdjg=zTkaEO+`(OGm#6p9$7G1EtOdT&JDQzz}4m_
zHv9{F2~(Px>1kzV6CvP>-8t=VR;n2H2#$C5PR0BbaM1qpn`OSZ!h3Sn_-@LN!#dhJ
zOZ!JE&8Uly`f6FbzTFW=CGuo~t7MmLi>+TZxsS<n{_aY<#ysWtsJEPhW@b4So^8K+
zsQbB1yVP$!erna0;JD4RE=u-{hE3YEQmx66J!+IMvkEmk2e{OD)l#k}J*ay<DSgQF
z)SZBmGj{j6G!wNXE}=p}`2iI36>3vFa(HTm#l)OPMFoh&G|&%O{v)x=>90~~kLYXZ
z>Dic?y5Lu!NXGI7+=_g`xWYBS`dWHgnirn*v+H6=IGDv<x!lcfLUp0Oub_dDE>pC2
zVOB1HRcaa<_{@9(q(PXfx3xjBeJ?O@{>Kkmi_QeCvt^J0gU<ze3GaW{C5JwEFblg&
zpuecUphqH`M>h&ENk4y<XEwhpKTl6zN9Ffo6p-CI)J~|b0Pvu(8>+w~+}yozlS`_k
zfa3X?`OeX`xp$BgMa0A?o+&CU49_XTNuhDhe_D1kav}hIDZ8c|I?(SUiF9Ul@}vlu
zVms%A`#2)tGAwWj_7kC88}oM%bMg0p;-F4g=AF~`2G<Y4JS4pZMMZ=Uz57=-1~alA
zyi=}p4o-^MlSHn|-w*+L$trdjK&OW&r}g#0$nUr$Hu>cXcGOp7#UR1m$%C-j4;xp-
zxIM|tbVID6tz0C{$jdu>fbHXIB&!tZSo9P+&}SlM;m|4{@&wE@((y9|m$o290NyPw
zCPvAPUglJvg=f20e`zu9!Dl1fhCVq%W5Rmq5NK+^DA-%*N<x7TTo<D|&qz7iD=8@v
z#yOpx@HTWt9*wdGp2jbJ5+;rTC5=dJ{575jZzE@-r=kkO=QlKrM{5mI7>sB!HueS<
zhWJG`51sCVN^L<NK0d4qZoj$$l*q_mTU(XUT&Su#?otkXfo-!VSpCO2E@!R5c>p=E
zwSDqw3X_z!j=B@tpV83^)gF<L9;sH1;syfxO>2u{C&Qgd^%F|%as&L;hf4mA@Te1p
zut__Q$gg@)vn+eAqr`r-?|IL|200WQQ2Eub&f|kXBf(_gswS1%?>6}^jm*pvXveA}
zo};Bpk{-^)%?6{xM7s?>2jFEOUw|uY(YuhQOeZH-2M{2P^Y-2mtV%pU)&v|vgfc2O
z=i9`9+k^c_v~&M4+vBPp=Gmel^0E7)&@W}5V{>^Ol{LrA{4a!O@nm0Oe&ZphY}kM5
z(zH#F=HlAIx6dKdl6kM}rG7|V_(^3Wy4lF_@yEvM5Jqc%0ZzxEvtB-N%W)olv>Y9_
zyx5{M$X8HcpkDj=;|IqK3ns3>$9V=4IaaCZUVSZdm}6O5T4FoJ37Bo0Yt;m$g4P4*
zaS;}1QVH?p;(FaX7YHX&x)IZO6~%CoMh_7>1C2ie@ei2U3BD6KW5!xZXLmP&6rvh}
zYdXR445t;(-+|fzcX|VeC~!1<n22On$WyBzZbI?4MGe|GZ|FaVmu$u>uR=o#3C$le
zP7(?dXeP!^{D3DP(jjmyG;16=&}dY96wmS$e?-<expxGBj*`wp^a7-nx*Hk}pb%o`
zecjmTSMXiZuWkzX3?h-4>)pCxLrgFb6+K`7nBu(zz%g`@#OAg^K8WzDP6buuOVQgq
zvE(p+51qaxC6@7#mQYZQtx)ubK*o@il!P4uOXeRxj?}qKad6~ZyF-6Gw2lezJ}sAW
z;Z8{0JG#4npz|ZG?UCVy3mZT{LDdjLDHauvGbFYr!ZBQ+kjAf04_XH;j-AB3TXtV5
z*)h%lrb;>~v3-Xi=Wsbr>im6+KbNTJGPaKV*MPd=wb{Boi;f=;crV{{{;5_xC*vyk
z?(JK`suYgI+_LUK=K#P%;74R9A&Wa{Z=El58pctiddU(7j=6Y{c)YhV0PEO4dRyr_
zML^Fu5YfO8FF1M~>k`i2aS;*nRTaR1S|rM;#s0XZAy33vcmVdrp?S!dz+0l9W$*rR
z5IPJeMAjGItJP0d0aC?j{GqPy_CPA&eNop*in-eoLPFsZCHkjuN^o<d6i-o!Zm@=q
zFk-0&o9uLgrG_hDabWI?cX#6%0Z+V`>RVGIHCjVhtxA$~JZz{#$;%*Zf&(Ad`R&UW
zrM08sXlMefq!AG(InVj3;WjdScdlKsBMrB?unnu{Ff?c$*+K}tzz8z@)Ge>k^?>9G
z8KN4xJ%;TBp<{XzcRXa0CmY}_s$Asp&R!+=w)ak@_6AHI?XI_L;c8KAR_4xi7}k-M
z)t2=z@o>2`A6~`zz4c>ZZpzb6_pHuCUROVFRA!tr)Law&KIcdJIa&IW(V+Nr$<vc&
zLbYO*62J6b6pjsyb*^>7<GaCLDNw03K03g-ZasaY@_>%sh1js@`S@Q|0gaxF2cI93
z(Ny?K6Rs7h9dcH*rOeDr^Y`VMg+}*~R@>G|gX@e-A4h5w2I`s{IcY*r6$bhHUyQdN
z0<E`0j*X4YUvSS)>~DtJOK{gXDJK_@0n#a~t3;LJS{$-i4waaLZ16unA`rSePA6d&
z3<}t*bIU4MVaF<e{gJb=dti5!9RS)}+tI&(-XRkMA#Wr#^<R~dHJJo&RqOz5Iu($s
zcbORJLbbZMu&}qfbRP>4W@q`ri#ryu?LqDU@WiRhU`3CK_tBGwN{Wh$0XaifkLXej
zdUAHwX;oet`Y&X8V-Yy`=O4ywIeoejS-{ROU_gFgRcJiU++Pb$HDEe<KPbHz)O22p
zTSiG(S?sL&FR`Bd@r+BIcFV$$CNbh_IV1%y4H9q+Pujed*kjP?Q>uOY_Ko2v=gzkq
zJ{HvawvJ6NR5d<dP8AXswm6(?v{J;lNAB(}mXNb)jBv>dx)zM$lMppFTxY>`aOT~^
z>JyxqL4@>Vr$B!!!5du4AjZys)O17Jkt;rbqbS)I<P-rWVNf+Kra)#IQ0fAJ7>u8p
zfjbQEw$^X)#GbM(X!YUkVB7<yn66ZlM^;t!-u$NXnwo`>s_qOFhwIzuA|<3!qb}vQ
zLTMpVcH7>(>{-JVG)uu|Cn4dn$i>Qf%eW7O0|M=ZDmTgl2R3k<P!-HyG{3pH=vq>F
z<J+cvd(0Ob013+l{alB8qma`l8PjhE#Zj})D@pO{<Rl#zWMAkNVK&c={ZMuMQvUCw
zKOW1SR+Bi<u}*g#l;SY=@R&s8QY78Eqiwavd_E=L%sXvZE48##G;0(7=IlOi)8**x
zwTr*{LuwvoWR-AJN`<F<x%)+1BRk&bN`mLr`A=rIbx-Bmf7bA+=XRntJZwaNtv$C%
zGR)O{zPI@|`6A<0&zKK3`8LDvIpSkKdDM3;`A>(Y9`Yo=VtF@I=a6=!K>~aH+j_e`
zkL%g^aC_U&GSP{=nR{v`&r`=~dwS`9xV?VDbi8Tz4L#b5&Qpp)`R>1J`@eY?Fw%)!
z!dsC0?e{KQ|HsRH;EYIk#TD}y-Hg{>=jO^lJd>6$ozAhYn^-W2AE^KM(Hj7wy}?#9
z$t}b<3GVFRVypUVP7wc3_uo1vF?^CDJZlI;0Ni>qg}<W@tQ`rV^ovmdk532VXb>cr
zDSg-AnaX7VRbs#<+4?f0fDXq!;MvIX6P!RvfG1|lkibxnSBbo?$YN7D!Kx$XJQ^?s
zAPSRcTCa~Y5SDlJ=YV4+nmoP7R1&2gIqxHHd3zW8?EUwT$om0>F^(CFU{_#2Cg-Qo
zpe?lb*<voJCYG3IPf^*P?54pC$yJnb+yTM&bXiy&8&t>sq>28*lL2VinTvC*G1-*`
z$vw2l0d9odE=24tokyr1=Wo?8G*r<m`5rERKdE4Fmg0Vu?yvY;-@+Xw6vs~@8MXLD
zuW{}itB$^+s%qpX<Io%E=cgo+u)PWHh=wOCe#5r@{*GN88G@phKkuUS*n#F4=2_Nl
z?T|CPJ5kb9G55TT^xh=KxG2YpEg~@uPdX&#&UD}f+<B3s)%)*eXHSEsLEtPCC0yiI
z`A;?P+XKcvshvjD<{ewhOLepWVBqj#7jqA7Td43c=K7bTb8HF!2uMr&;1LEteuM7i
z4?s*Q)qsgj**!b+6|_N9-<Qd*TDJ9*<e3(c7W=C=e8RVGJY1fBU?wNBRy|x>x)J$7
z<7g|^?n2hYZ^xDQ%2)XvO)_J*=$y)}it@hKxG@y%qUF)q%V{;$yXu;9|59FNcF*$9
zhT%rP?}39$qTij3H21U1$!fpI>oHpFi46=>@(Ev0`N03_>hr0_jDSnd<5$KRH$E<3
z`#iK5uW4T&7k|HQxNX0pQ6qAXHP4BO(LMJ+h^*)L3x~8<u6>=q%gD&p8$liMx&JlY
zPk)!$Gp|PP3>{@Zen@{U7MAT=ltRaj9~a7F?UbS{orfsMK-%zRAg7@};097=6h+9N
z&MCgZ`8~$1VPF_r1hgBXkz0-RH+Gza<A%V&rollYH;p4Hrz7O`eyKq2;nU(^1<^-^
z_?L-i6rNIv&Q$l%e~IOxK?N3KYvF?W(<QH@+1c5oe77byQ|?mnsp*Tr9|)O#$XOhB
ztGfg-k>weq`Y9_1hy4^3v%q(;qF+S#f^;TCD#5{Dd&AvF?&4+7c?<Dx>+Kb{a}s50
zuwiZHWhS&96dzM(VBvw%0R?b#&p>`|E~bQTm2?I$KlT9~3f?SaXM$SazI~9;!|~#R
z*r5wMQ*xSAPOrO}Fn#a-EVa@gXKHH7Yjqe*7e$T#kq$R_E7_?wg;`4!m8AyHUQ>lX
zjDJCTYN|Si#i4zBklOO`wf&}S(q^4G+gv;Me7X6MIMS>Z`MpSo#(vSJxS@IX>{)Vl
z;41DY?HAkja;RL44^{8Jx7~wi2M8vUM5TAnd#=iD?$-J~i|zNeH}BBlpb1Nc$OktZ
zrKshuXI7S$|BJo%j_104-^aV}=B=cJLZM+qC?jQ$R49ATgff!sog&FDB`cxGCPK+7
zdu5Lh%HFcF`5jO9`~7(Qe*gXc`+gsf&!<1$qj-($d0ppqp2vBd$I%ook9r=4mh3ai
z2==M{sfNN`<sR3%YxAksxilSKmq>ZBjcukUCMKnHW*f8@My_;*bzQWzBVRi=^<6U1
zvF?(0+3m7kM{BR3(BH51nJ&57nzs9=WTrf6*)uz$X|6I!w<2`M&;00_qfZViDpWKi
z<g^&jP0NKSy_Oi&WO>j0^4Dca6a$*9Dc!e53(qbUYkScI-`83>KXFciEN~>G@#0`h
zVBc4r-M;rP<eKPbF9rJkK2APB|BJnc+{AK>$LcX)8fRFh%YV(`gWMGe+F~Ew(L`oq
zWOQAf{0L%u@oxXjXMO)1Mu$NHqa>IrTlRhRBc-aoe!A-;XV+A!WBZe^*cZXUAWl8;
zOn72pR=vG1F-1#fo>ou%uCq@iW}Jyp?I8<vSio56xOM4`GiFp=Q#nvZ=L8%_Z*gAW
z5xZ>GnJ-H+-ZR1?gIor$64mfjEhX8@Y5mY)FO}BH<Tk8)O1KBp3%@Fq!jnU}fetOE
zdgJZiIV?`;Kx8=gHhls839mSfGhe=cXH0*~jIQKEy1%L9m6d-WyrY)hNxF_|7RW`E
zkW7IuEQ&)yLr36Y8JmI-C)i?!I_L#QTx#hO=i0>|eN1v{tQr+Ntn9tpY!1DQL?TYO
zD12Q@!k7sfrC_xXA|r;r+e5>`{&AaT@%N&#5JU1%1fM$quwZ@*>Ils+s;H}f|M8<X
zM>_aM9oH`}kfD&QyZ=;2%VmJ!V57rXZtj!(eT`cCOAQd-#6E@xX3U%gO#m~tR&FJD
zm!2ZG-o{3SQt~53i(rLz3?0STCF*bTxC8tMrD#mP^WuWy+RL!8stfuY21TQ2Cql8~
ziDFfoXnzdvcg_7r_g8%R!XPO4D%1mwGXo5)%<G0B1`{JI-t_Y5M04M+JJriV)XDgE
z6ys||3gL6k;ZnO;VhKyK!j#aD$#0X_ZLeu*w<(FLT@Q-cCcd^1#Lwzm^6RYZdO6kb
zSVE7bc9dmF9@)oH!?)?jO$DUouD8V1PUYuH(9cx=x~ChGe_!v<Yw~s_W%;c0IkMI7
z?pJ=GHa`FKb*RFNuT*60iR>}^?q;`qm_Tx_#eS#2WO+5V&(iNu*ul!<@l+PM?^FM}
z#8AJt>+bE0nC7M?{(C>apwG2a4xK>!L4BbEZ>FP1p`tMW9>B`k6N#^WLkb7F8A$dp
zG!b2<XI#9R_Hjt(<P;}<oKVekxM-M3$1NGP-B)<kdj|>~z#h_t%y^l6*sUG3XO{C9
z8)-|W)v}H{=c9=k3A9IVo(UDqc_A95iZCyK`F!Ka$I$C2(>PdlWcbrF{+pN1)240)
z7OV!CjXAA}-0zT(Kd^)Hz+n#4RG&9{BvXApEv97$sG4V|s+MUkrnP>CPEYn!>u1)M
z!0nXdZEXs>@7{}faP8p3wT;T@XBOekhs3Ym-KJMxG(XVuTtfU4^R)Bh=gsJ4oN~4W
zGhbm^QRcpbt;@=(-I7UaH|I&Ni)j&-KN&0UuK<-qD;KbsLVijU;1(6f9{?_LfwQ(s
z;MhUJ>A5BCndO5V;W7_sL!3wqzMnBzXjGKs4rEJ=^g{iMJ~NB`DH!5uf@}x2Dh8X;
z;Q;3e|4GAZ8awPS)Gf4IY`U%lB|O1{Lt2h5>QGlAl-67S<ny2RQ*Hc{27~j~*562M
zk00OpX-4;&02z`WwcZ8s1F&#lGEOOa=V3T3R^h5z<#mefAa}|fgT^x8)KdnW<|Bd>
z>vbORO2Z$+=iuTc7It3;{}QvWpiRWjgrSM!>+P-J{|T&_@Rixa$h3gXaovsR{a)uW
zI&gZnPN?O_%H<TJ?3c|w5lnPp!JOOwur*AY7OPmsIK=RX)!b|O^{_<5WY+tw|InFL
zmy4Za&ef@xQf1hBr1Kf+ZmUI`D|R1aeEj6`vZv>==ct~y*CEGOr%m#fO}16h3XHW_
zbW*%!ys<+JW?-U;ig6HdULZs_q(JvWybERv=>pgCZuiP_IpM~PzXPOy>HUVioH?4s
zPUyaiHH0$1i8(tR{#B^|@9T3s4;)O$-wGOFwVAJn#}<m*v9ikCc=l?bChqO9^1Ex)
z(4LxUc~Dw<X<=t4qtspL;Av@Hnzv_2Pv$B-&{$esEm~;P>hp9~iMq*H_3#uL0?n$a
z0}^AwM+(H18or;>9P}XLX%7W8mG10~2M!nSEgV)g7$`%Y#u!rcVxkn2xZ0Fb7*a7=
z4dT4msJJG@>DUsY``K<X{4d^w+lh}S_vo84ewzD@kwRc3sIKK~kIP=YN>F92##_0J
zs%<b031R|`jGs4U;c&D(%R6ul;7zk0|H0_iEz_F*S=X8!lhc~e3i`_SgR;3!0{kGr
zdnoTxSs}a@;OEyeH}7socWN_#oZ67K&GD?Xv~HgOS&+yLeSWfKRbvw5i+!97==Wf5
zubBn%0_KfxJRMHV0h&;De6{jwA1nmOgqK7GX(jrN)UZR8j)9hyV30LR+vRvRvb6S6
z)ma&x;m*h+^TpIS$!F^QY^Rm_<Pvg7ukqHJ3ME?zZLgRdyGVzFJ5xxtxhS7gDiJqH
za(18C)khQm<pQ*oiS_C`@jtme8?NJB-S5ZJYcV^v`-r8<Hb)!yt_cbfm?c`I7o6r3
zRxWsga1N#slKFK=W}ubuR%$oHo$k-s;Jc1>31SEb(s+g1xmIA$C8cNLe=D<RzI9IZ
zD$X}cvLL+5mt`*~GH-sTuc9<0CHnuMN+fVD!^x)!#kOl*@X_J){slM8FAg}z*UwW|
z)vFaol#YgaM;D#Q7F9uesZ7VzpCe^}!s?ir?~~hud)$8mmyAdgS4z<M@XpW?K47Ju
z*=A-O_t-|(g3f|!wr=04Qsh4aT+f`XRObGp>+2!0EHJZDOx0q2EgqZ>)_dzqyPTYy
z$Jbx-WRw4JgmV2Wcx+VAyXC__0I>2al+i-_Wu&B#{(r+<bOlliYz2?SnU)HOzNx0D
z$LPY&Pd-N66NFy0=Mo=3C8UG0yB-A8c?jFz%_=Y(>$nN>-rlTg!@{pd$HUU*&H|}G
z|2A>=Gw=$cTxPw6hpDpW7^ri{F3!BH^9Xt!xqHD+MXD=Xn3q>&cIwmR;jf=4!^JU+
zQRb)(9wMVx4D{Z$UR9Wk3+7{z5Zlw~d_N@Rk{Ux${1JbZb8;(s7#a?9cYV}OwBpvR
zf4Z+fNRO!^t~Bw_f%hMBoa5bx+U{KriDa8;S6uvP5ZAyO(^>q<E0OF_(;;Qv5H^)G
zQ;pO--l3Xu>2V=4EjI7GmUOGV@2?p?+~4+2)VZwMLr_)|b9htE(_<f(O8ImaroF?R
zN}SC>+riW}wu+mlxFv8q_-%~?_QR}i(&_v7Pk?*=0RiU~()k$~jSxht&(<|H&Ats3
zUis-k285AwN(7@-I<st9>iRvrttS#Y+YWOd20q{)gs9>A&T*&)6T`^9ld5Xm89+e1
z9?6;^^fYuJJTqu<-A2$53_f#1rC;cq!NKgZf%^wA_X@|&lMh+2TU-v3d1U<KaPT!2
zST?{S4}6+DkKrL{yH#c1=!iGsN_+O~d3q*fjq9CT%;=MNF1*NqfPB^t`KWL9U}Zyw
z?y&L0Wytxf#B(z+h^|HEvuHcYxD0;<sf2K4EB6DFy%n8uxmOq9m;@<6{lRt<Pw|;=
ze7)a0I^fD!a*6Z;>|I#>*g&!YXB@YVSxdt4*AWC<dG+qQM6syGna&<wG%H(Qd_;@F
zyz5;qdnEPp(vopX{;d?L7q}AXc#NM~&1T|83n`VcmJV9095)UfkO<n|>nE9MMMg#@
zB=AO>@$KBbg{0Ej3%`m^R;!>dnyj37^E1iy;|vTx!H}Tw>|EJyLnnQzl_RdxCaGI0
z|3KGdY}Y}BgNNzXZ^uVVa|8$MYqcLLTw5t>*3)+Q{pYKO*X=TFpu>NXy3Fo=*Pi+6
zG;0&2yHEMRjzC57)yKYz$*p2m4nMmG^8Oj!8R{0k*7S5i@tTjIY$qa{P5M{I<yoWF
zA^_tlGYMmVGz&j$KNa#R=((q-B&0>Kyxd@4p$7^8Dw*ji9Xg(fHlG*8xW5BfRyp<q
zdGGiL&-jtgTldXa5iSSNMiO)4JO|@DTV*9<<V(xSylY4s5T+nWnZ&8cHWhv~Ni#bM
zY`=2B$k^Ce;APTN!;HDv?h5yJ97E5iE^@U#=>S^cXc078d;WykiIT-7hCSk%n&HA9
z;=p)BgN>7;eM!>R(sE{T@vyxkDo0@SL$CWlS3A&KqSkhUSX4L|S)I2L7k6Jtl1mb|
zB&DMx*V)nR?d#jjGPAJInlYp8{cCiT(2yE*7{&k(dJD;NVf7&^;`j08$?2no-%Bo*
zwcDum@Ws#FKsyFv4Va#VgTi@scJqI2TdRGG`FdUEAbc5N?6Qs&>c!^!jEswR6QKdA
zN9HF4;RGw}ID4tfSe5bEbU21OtteivoRfUGM6&wx>vjVbP~hh=bdk9Ewf(~!!^KCx
zF=s&HDbD5hazZeaXEI4K&gYY}QlcW``g99bMV$nwk0`TSnN#+--@|19ByVOxA)og>
z!ydFnm{J=kPFXaVt~N^#|N7SQ{mRXi-=8WiHw8DYynU}o?P|}!yLTpF_i2sOZx;jZ
z2fw}8S6d-z&2aVM)w{V2d)y<lb<D}U^JvI)DSj5)a4x_8^!sP3$6P>N^RUg{uf<=>
zkM~kn&-Qblxt<#uz8~<U;<{G;u{_h??*}H>ZIi-6J40=nymqm%C1a8`v3H<$(e~_j
z&%K+QY3_g@r+8~hwaiSELhY|-Z+pbBhj_Yk)J6;!bFSty6zD_IZK#fSfbJ?hbWiPO
zTt~kmsA6-eKd`SDoNwyiEh|cT0cxk=mE)Rt-y8yx@2-V4!G^Qy3Py;Ju;2Xc^chR(
z{p3ldJxyZ}qs%FLfrGv7s`S4TGWAP#3GD^6ntV>FWQ3B${T^z&8M6ep*)+rmiCKHX
zb-PczYFhan-Pd?AK`POqU7?ML9Lk%z?{7TZ*qd++-u8fNtKFa`>WXewpFvCf<VdIr
zd+2QS>Mc9ESCNrLm^eV6j&*Ytk}>e3jTp&Qj`YGOAMH8ZN)hR0HCU~Zx)S9rm{eDS
zea5)h%X4_<?k=!;A5WT{85$fUY|FRHf-p>}cM$YD!u4zBD>`Vn?hk}SWX;JVQJeW&
z7tdVDJT|!mEp9Ld_#ojrBZ2+m(#n0Kv(U#<fDBb+U%_ii5>JQkZw+G@_Ml0|W8x`Y
zJ}H<5O>w~v!V0)=Kr=@E>K~R7!50oe5}#kkN@+(s7j<{de>?Zo!{5W<h2Ct9q2-a=
zx6Rw*>_UnK)$R@6@Mpgs%-=G`&~xPkBio=|pZPkQ$T9mB+q9v>U`y&mPR<rxY>d9?
z5xH)Ee$M>m^t0KMfzNX=m2#~25ZrC}EH*atPOdI3J=vX+{Migtc`(QtROP@XE^G{3
zA$4;cihuO@aZ+0zZ}%~W)}I&9Z8_$AAM+;b80Bzk(+IUs1%#5+xJo3tvH`j>2Gw8i
z^aRT*o-F<X=Asp?Q=$wJR1OYkj^#S&45OVT({u4~dZ?Lp5IeGlF|1d1q!pS8cuuN`
ziblAvI?t1VRH?l8V7I{L*l#>J{E}L#rY)$Eo#c<u1xzj+NwT)FahWby<RL+E%d#Ol
z+3{f*h@=wdMeH|drgElPLo4H5n*4!BsDw1uXr%4z7EUdnL3+}(<N@W6T%2$3zzH;{
z#(EKS_qTcsJeq;MYxiyzZ;>MYHVY1Y%&)}I<jf&0S|Gh=dllIw*-zn(C4;KOsWCC{
zsZXZ3v_D5R{$!hu_~9sDwDW|-u-M|0900^?({7v1A7s0oS%@)%>_p!SH|2XjbiH5I
zVmcw~yp#RVcG5!!59)t8a22=Lj2X+GV&*89H-+J;40}-IZ{U=bMKR%@lzMB&A!xhR
zX$;!_?FIab7rySMhs>h^NC-3@#V;^@RHjxyCbHJ>^D<4r9D^e@!OAdgm5syQx;Ym(
zRLKB%VP<IIa=QFkyJ5_l#8~Kkqs*otDMPNod-(^aRu19oOLOHVr=G4*o6B&d{p{q$
ztGeTnuU^^EN6#|Z)73Jn9nUb3TMZ9as(8G6zl9q2rvHI&jC^u^@xN^sTx;e;N|q;6
zCqr!16*FvN&evq-Wa+lteRg?XwJf{(kv;PhGGVLXC*qZEGXlZ2?O_#DCimak*xTr}
z8U+uuCA%}IYiOe0TJM}|pIpyrPmUg^drf^#PX&cTH9jNb?u9+>XVEZj?G8n<>Fq_5
zyy?0IlGjju6sHEEmBkohD<8j++K+x2zny3M6%<lA?B&Xs1XJ;FCCU0E_b?%x0HiOv
z=$L>Sv|>@BhJ6b=Z?r-6K;B^3gIsm3&zBQD>zC8qw+s!xcXpcIX}1N_rS(uNM$CS{
z?LZSMwNTzS+UQQvSG+U*ks}H9rqN@U1CHjxp3lE42oDb96B6JkLHvDsXy^}O6G(i%
zYF{XK96xoc^!-tJH@GLYeBUK|;1TXMfW7w@yX}$O;I`z%qz-z#;^M{Fix6uhN`!{?
zuPk5;3xjYDJnXzlQ!I`ecK_n<-<7qs`5?=n>j!Uo7RNS-1hVRiCyRTMM7jRyfbAA<
zHAdigktMu;e-<?oTH(kEd@JS`ARr<5vu*@wG8)Idj>E?m!rI*#en7tfeFxOjvx|%M
z=s|jWYvfwRT?jSqEi726Bl7^%0{&HD{mA4byDsH9F0Q63sk!~Jk1;oDseGKSMne%C
z>UHn~Kz*)urbDYAPS{IeYumhnYZx?x(^g2-l2zHw&7R{M$YqDN<5Cu0897;X?k0Y?
z`Madrwapkinx26=kCnYfO?4htO)mrPox%tr*yAhlPeH(dcCHdux0KB2_U%9#Ug-tB
znJ=|avVaDXd`Bpfy~kc&zQeP=QiZP(R|&;TED=pd1p<p7FKI(rnGsMw^kCRVyulm+
zS*z$WES_D!I(iIqPYMb{h&h4_0S=Fp-KH>D-b$P|L}1lxQM9hw?}5C!7v`dPzZ~47
zn%SnP)Nm1?ra<{~2fBp^hyOnO<e&f<N>!y)=4dD_|3GGf(ShANc7Xgc0q5S?*KME<
zma-*Mqeg%7<cT`i0_9Q7fYgbr3~4P~GNoi?!F~udJc)0deBWCG>UxE882(|Oc`=DK
z8{?*3A)OWst?@WaHm25a)a+=Z;G3K)lW*+<3muk=^Ze{X_;qL;58q%GT{arNC34*@
zDI+t&^F~HtJHoler?mkqRaw<ZtL)Xb@o}Yh^3PR*X`Fa}h!}5{7xDEt+%IWsI{!r=
zdQ)J%%%gvYn7W(7_?~~vsiM?>WNOOzuj_7i{X1}<0J$yq*jmY4U^CtD(BMnkGg>{l
zE)waF{qt#pGdRbbUq7vof@06cXnYjrQ}_(Z_bse{0BsnhSNuxE_~_{A+&E>-D4^U6
z9>0q%{0YIB#{LbgCJAvrz)NkQ-7a3^TUZ^&v_qeO=1wjIW6Tiv(eSdu?0Hr(<k_=x
zP-tm&yB-qCW8O)CR_+Y&<=zJ_B{o!?bg-fbgMffDt`i`_b&HIOrw8)>w5sV5Dxw{U
znuRX7RO+ET@+MV+5IDopN>$J3!(~HbKML0TXX|0v!6L+kluas`PU4b&Re@`d+U(mC
z0{4NS27YNLKPm-95Hu-!HEU~Xy58%1475m6Ub%8*FwepiT|>VuHhl|A5!79C?k7=$
z7zS5i@%s1KigD5q#>yU$K(}xS)CP2T-zqG_1{ldnT&g$}Q_U@W8rgXGVL}ZH)i|&a
zX-;Z^Kgee{p?-`RmfRfb@4u9$u9D3}x<FG{Y5Ig(ESxyM!IhWS)BwF|OabWN&K+p#
z26#w~p*qvpP!Eq1OnC6Bo=^k*ijb$Zy1tH9IJnJbxLj!}QgrhE0Pm?=%)Dt~p%*7j
zIr!iSrpo{^k6c{;5Nk(AAc7JSA~?8%R{nk4r+Nxjv8JatfoEgBMLRUUnQ+3bZZw5H
zO!&WA`>BnuQ4NBSI)cdvCp$Wff<cq*@jmLlk&*9lHEu(rvG3mPr{TFu$pRyBX;(S&
zU{q`8Eqb(b{T*=LZyI1a=@?r_nIa1X<ft^VO<NHTl?|qp{dycRpBq_D{|i4s{WRKS
z4jT^YQp)-7g6{%=AQCnQTVGTv-HrJSd+>`Z$XfQcRl8f?yQiLUQxYMNcd8zbb+Wfe
z?<iAvqyj$~y0ZW_!G?+{k7qh}PGM124>^30nJ6vJEX-m7SEf14+Jdx(0s4SLX01ta
zuUl#!)OlzeEp+{4ZT>ALT$M_-do9m)ZnOSO)T{iXW;JgjlJSq~`9-ECc}*v`UooYi
z8QeW{y8qBYnwZtlga`#UkqxcIb2m4c*7Bx`KZ}v>GI=4;dBLamMc<o|jh@hUbkr__
z?~4V6VZbW3OF9f1!HC3{MD&ZMaR*H+188s6uNAbb5#yB7l!2ojqoQg9WgcxmB)W02
zv9CbK!at(B1KKM7M{@zG22Kj9yBEGeAQxT`5-Nbk3nq_~0tn*XLa&Qxj69}segihd
zL%G}dMp9H%6oot34UobB36sdAaLXm1-^4(7A>?lR$FQwFezbRS!IFbucm;qIijuME
z>FLqYWl(=`90_H~j;aVCNB|z@6F)qihNEv6id3jrfy%v8>^ndrL;<9S6y)UKpC~0M
z6IS&Yzb7=&!EhrFOyi5h_d;QxoW2dCFjf#us^Z-BI1JGk>@3{|jjh7N-@kRPTLGSr
zorv86)1fX(_rwn1l;r@w#z`=Xa7mI-=n?NJDe1@V>}<0F4;egtJ$%189e3%nswxu{
z&z(bVM8r-&Y6emYmg9!M?I+v!c-7CgW)JPYu?f!D=k^r%PKBRjdZWFhw-9*`1v%z|
zjIm(If+2bah2%N(Fwn07tdqDIXKQWUm1mzml}?Z6&;?9&*8T>5hFB>8h*&esFB2;Y
z(a=FUP9%H2Pb%Z%zbC1%4f2$uJ&Efs4L~Zu{iMbQN>*h8+FMEq>U1OqaB-X=nFVj#
zZdeWC>SpOp1RoS*&nSVQhNqLgy9{HG*FJ6^!hhXg8D$=P@dZQNcR<O*<UL}Lt)1O{
za7-~Ul5?jc=-IPh0|PujB3<T=i5<YC7uYA5#_YgnIFHZ^1EBVHK5p(gOp4)CshAt9
zqbf3Q*2Fk1f#Ylc!^g~g8!BQ@#vTPWc;g1_;L&{ghK?7WEhkt%^CJpI6_;m2Dn?)(
zm-?*Y7{s7q;_Av_t^d8N3wNw5b|6~Gn6U;X`0S`ax&)dV;LI?WsQ%;502tO5E4{Px
z-sQL@J37Mp8JQybAmhEXDJscr$dj_V)6P11zWa>`p$mojuIO7*$cO;G71c&0H#wg5
zt)zVx#j_myIImRImWJx6*4d2Q()r&f`n<SBZ04qK-Hxp1G#^WD(c@fOF=>_~z5hn-
zV*0ErmDmAtF7kIr&wO9J65Mg+mfF3v14D;Z8xB#rpAZu0e$UK%diH_|mc0lU3_;S;
zEl5gjFNBuyvAs9^>4f+pii#~ovEw*OXu+G60PF~jrP}x-BKPw%qnIx;7$GHrJa)sd
zmm)MtKEs95Khb3jN7ioYN1}nzM{z5}vX6Il92`E)QaWH>3pV1;Z?A=M#o<XiF$sIR
zwc}DFFgYA3)}j*NfJkAkU+feIh3|QQ${W8j7(nAV&`1oR*J4yESXW<f0Wr5uJA(#&
z+zOCrg)mgD{p<@x+}A{=Je4<1mG8d`9dv%AJ_ynCf$|t8faDrWeWcv>Tc<$h#!aGX
zaNUv=djyHn6r885`bE02PYFqc?&qO?x-qBs9lg+iF+)7sWe!XBJ>sAdg8!iwHta+Y
zAl4|Aty;&(EwFL0BjGPPj|+RJSY=h!a_{CwyusIWD0v_NO5g2GvHT7-Mir@EWLsqh
zB#?*4^%7)3x7cwzibzVUJJNxt;?G#%eyB^}g!Fxob{Re{ER3XgfnW6(Q3eX14*iOi
zBC?fMEf++;e-81e#btc_*qdV>;cz+kJn&hwd-s0hEis$VKuwO(c34RGES9KH^V@){
zFb*Zb7mIV`<m5JF^yeFgKF~mA3$ey-_0MMN%gATXW{Nh~({WXmJNck#af8^gIxp)w
zJDs2;1Zu(()^%QxF!MyMu`qM4SesSI72!@LNtt+qS&J9Krfn&p=-#QyHuJB$6zPS0
zsRI<*v>(h=M~;9sdbc~HkjNPzi;pFPv#|?W`$1-#CdMrZjVNwT?Y4>7AGzDQN@XD6
zyU<PN-btfl*&B_jdwg;mtvCJFKd=m2&YYex_!FqnX_*-88JuIlmB}r<vEDB}mCi5V
zWEbBV`FH-Q^Xp1dl~+fKvfXaJ8=7Dg8azEK+&wq?<jcXRovycx5*I2O4_J!~MHuvI
zyE?TJXW)<ZQy5=mIL-|r8TV0BKd>VP2@r0FiukB8l0aZ*h-FaWythEj#tMKFL$_Cv
zgw31OqN2iV>#Y~Ywe9U{UDxQ5TvY%)hW&V3%i#!oC%TJPm%0{{F$Y3u@FQrgsHrd6
zUq7$|bQok{&?309IJu6n1KJ`V2Z!uyT{P`Vc%>oQm?l>w=7v(BHIyqQ9_1L8(-w0=
zpVjJSfBz;hYSdx_`}$wt?7+S}428>hkDuIuN7#2@>=I*H#2tFxT~zR|YK&z4?23hK
zEkATc5LInlnw3zoRRP`rCCzp*J{Zpqu<jtz&T;ihFq8>m15Q&5vjF!^&!Fd(eq8UO
z!J5Qh;lh0%U=N7a2tR^ED#-0uFt{kDr`nMwe_L$Nj2-LxI1PSw(aIQGx7(fd^>si_
zHu#7xB^Tm=h9Z*{pKY(|?uM~as}sj(+`UDplK=#`?pHmHjxSoe16p#PB$Q~(WI$is
zfV8NI(0L>p#RISD+!=W8eE*Hw8$H$)wUMc*`}iQJdZYrXA@(keju8M)TeT*9C}7n4
z=!LZ_=;>u-YV3Pm?Qx)joR5aNf0}!5R6}*L%>oY15J226o0XA4*l}X&2C|V!&NKcW
zdz{SBt3h)2oZC+cX~SI6=7s>Uvu=Ze)RYu{y9t5_jbzMpZ@=IDR&Jf)7^K?7`d}C=
zfGiV2+0bd22P=E@c_ej)c(W}X3UfD%pkpLu<VZLmqikd|kx00sLK;7f2!=Km3^+&f
z*$7+eISDWIXH}fE^Iuy2Q6CgYby|V%ZBOr?jrGHejPwOdOE&|r3)Phlf3S5fRLoQ)
z{^d=mWLSFKZ6?Vczlk-oVF4-oQ}nG$uTNN%B~nhdWp`Pb$f(M`3^>{Rq4}eBXOodc
z2w9GsJWt7mm?wdObBl{=-N8*SDVCO&fTyu;rsjPxHZx1+&%9Z-_HjR)bwMzx!-5mf
z5v2@d6~UK<858bTxN?kgv%;5q3cO}|0FOXxaA*F;ZHw#&1A{rZbn!i_4Nz;vn+6B)
z5QfQwr~zVcuiNH2wmSqcTfa~L-F+n-=@E{E1q6yg`GXshh}(uBU_VSx!803g4g7pB
zT0499HaC*@7yd-*i@wV$2A#+t@rjZLuLKMbuIGzA&gr<*sx3|;mrQb+UDw~cnf8zS
zPh<+v6+z|zISy1QD`<^EvI6J;^2anJ-6-*p<=I|4udrFbzIJQBGziN;rN(?Uu;Gsh
zI>=i^aanRJ+r)a|epai~q!dxcvF!dfx3Hj-X;=krfm|d%<)K5?fDF;6fQA)TA@B2N
zBs|=wPQ~Khr=it*QUDQ;Bx*U<py}`x3(u9&PSc$1YzXTE6O=$ffzE{EuVPQ!dIb(M
zC$^@<QZcgtx&g^Q|IEzJHgy-HOY2k7aXpMnvTK6O6|;9ln*m4(B!-d40VXBlEx0)l
zRYIl>7Nr$gIs_1?cuv5e=NQZo089~YgWC-c9g3V74{~yFP)<=l$HS8)=}Wbf=BG&t
zNm=qJR4h8xN7aej578doUJSw!LdQ<Hd~xN0lZuo;3)RcdpFcs>aIugXNX8{W6A`w@
zoq;+!LF>aU|I6O`b1~AyQm~~<qe>#2BCMRFz$M0^hhq-#3QsSu@i}v)63BbP_lZzx
z04ju%8orijnW1z<Q7aJ^j0`nl<H5*M3sQ#D92|FXH{ruWz4tB}J*LXx0WzZA6M{#D
zU0~cf4yF}+hm>^>J;$(*q9Qt)$OEt)k5N#}<2D=Hql#QC`KZ9l>78YOg%{qMtR_I;
z0_AieA+v(Z05#eYVkBM&oLIv8BdP2FvV`a)=3Xr(q`^pyhf#8&96U-(y9(h;qxb~;
zR1j^j^z6dII+(n}-y(?VPJ9_+@{Eazm4XCx7@P&<(IPxN$`v(ted<c!a}Rn#n#*AD
zeN3aFp?&P&WnjUj0*}WXZsSl98d0Wa&L+K>cJ9WL2AmB<`8LR^%yNhL&)@qm7vQza
z+LEP})htd9I7mObyT@CT50jI>sSo=5-MHWjHnqDrduT@~rm4)L3<0|i8>|@W4&B`a
zxbYCbA+=S_M^3J{fY!>XVb76zo2@p)I(jO|hgEehgLDsc5PksQ=)oR}21Z%Yr{l_#
z8e${8kjrD`5hp5e6fy3h2zn0WPC-UNHxTdqcA|pH>esC*JrUPi>CKJ@@}Aoa+x#%`
zFQ%=}nv&tzQyjL$eg5U<GSe)}<TK&=1(t=zFO%DQXI+xPn&JIncA@KHeEv1Zpp8>Q
z_pj9!gbGT&1|n^x>)mBN8CLp^$E3rg@8mR-+1;Pc{T-wNY~R@0$F;q$^0P6^)Q9xh
z{1{ag`RMeqI`5lH#PW-ojE^~Pub0>Kt4l--HrM!m=?$K%=X&jJ<}2raxcv{y!n38y
z(%RCA9}|@Es%uU&?N6(Z`{&a-@;mZ(USKM?_IY7Vb>Q388M$x^&(46NU<b3Nhf*H`
zYtz(XbUl1X^J^SJGC2DU$6qY-bB`J|b;srn)4wiSo6j8ZRgG02`F`yz2CFMb%hBNs
zpyn>;N#zP8puFr_a{?gNN@r2*d1X%whc>9!rQ<kY7o1gd?zx7B!k)dW_3k^-5fZT;
z@*BNn7a|CCHpEz!qMFHp0h_^UN|--!=Sc5D|2`ecTPw46&{spGtl=x;{*1#db=UgW
z;9#6~^P!!yL)wp}gRtv~fh%{07QmkdW<%t+5Q4GGQk|+$>%ymgTmQ;fayrlDA=G2>
zIPlzR&ORsk$&$#eZRsF#b9%YpZA(|2U~h=+%X9LEM&0#JS=!vg=FRW?t&WjP)vE3E
z@j^YsX7MNqNaHx#Nw-YvXr&EY*#Q3nK=M4BCK^B|lyq*;5H$j0p$Pk)VnhqxKjERE
zQ?0F=Y_oW=MbTo)Fsvf)PqTmMVq!zbibGlX)zX52pN!Rp?m^Ha!Tl`fF?^i5(Tfd-
z4nsupISWE<FKLqxLL0CBZ_1GSx1WXT48|sgO#SX4DDdPMCL*Cl4OiPrx6NV?t`Gmw
zGepn>%$eIcw+pV0h2Tj#RU!K()yi1BetB5TyKcLD@zaXG$}VFmh<rge6EaBPutp>c
zS%d>mO6H^+o}Nzc>?d1MnpT9aZHjKeXVIDl3MHJPHuWGA$c1@od+*vd523vs78Vu@
zLBw#hdit(%TW}2){_@61w=)_@wclUg^r`b8kK7&oNI#rX{?!@nPiNstQb2o*ni`Vx
zH@MG_ezSxGZ!4RHkx@2y27jAB-X?Yc*BeH;DO=rce+v(ucJE~rg;X@I+tflIYve#$
zS;qS1On*68W4xOXJjcfw^CUdl8loTOhR<JCAPdSbDByiL=OH{I$kOPzO`2}@Li7ck
z<Z6uH)pq;vDWgT6q((tU*OQ=daxN$S!e__1cVoQsfBtM>@c3I7og*T*)RgwVZL=et
z*~#TgPL)go0$R4eA~3Ir9O7Ho>%J#-+aCn^Ko@$R%M?wcx0l26D9g0A)M2u6wf@@n
z;2KZcru$zbg4<XUEBiMv6BanuvzL~M>Ttf<wr$*fT*^mVQ?>Yngmex(;?r-V-L{P-
zeB{DL6t)+zv#uYhqF;Zf{k5(3NsYWEkNuRkH6P?q^GCl&ZCOadDVWUhKkly#^t1mF
z@e~yv_>bA?1D8s|A%ys|jA#At|C&AiUoqDKiT}FAR(cWA*2JIxi7yzK{q^?!xo>>P
zgwNkpg4}E(_|#G_A+C~zejpd&<_rhXnVFzLPK$g$tpm)~1`pmy&1fn#DDA87$38_-
zq~sw7ty#${pk^a0tnG${o&24Fg6bNXN^X0$2Yo)vN&hbJ|Nm!FAs*B4nVC=%2ckD{
zhw!wvk&%-lkV&Vi!^GhDVZG2~!j1U=;8yUr+y5kF2XJGK3vk%AhEw>5BP<|py^TfM
zIp2Z*>vM1@83NzLpa0zjTg&i2d*lD#A3OrHa}g7_J=pvNfF0*EODG&dy$Z9)(2%*0
zC%i}A4GwF$2@wTJzKO@khvpFnWeJCOz5_0s8Kxhyvhp!T02!tXR(K+a9zUlCd--T%
z+*W}H&ERdp<OiE57R9Yl>GKj4P<YQ6v;x=HGf*MI>{|vRV6tc;X+&*u>qQbqACWMa
zydX~WjWb)pRx%)CS2l$GK;NQ{W5wSe*-Hdw(Pz!j=Kyp<RY=#*AxWrF|L4`LFxrXk
zsY<0ZbgeL$f9J&?{NtA=|9*-$#c~>xfcXI+D$wA~x1Bt(^%;C<`0h>&537aJeAn3e
z694}78<Gzz5t!uxI^wtG!Tz6jKiG;4-J>3fc05X`gIf;r23z5u<qdB8g*t)hBX%6m
zRd6%-oFT4{2XAWLtEYk`idopp7KCBDpZk{S!3Y46^f;r30(Y>kK>N8Qo$-(I4*!1n
z5Xe3t*C%nJ`uDZCwqEN*YV<En;3IH;y4G?8KTbjS@0uW<$w7r^J&^F$7$v^ngW{80
z->(S?KfwdooIw@KrDO2_dXLBrB~`-KAzt>_^Ye(v+dMB3x0ig2Es#Vs30IC*$r5*%
zB>mQ#D5#;Mft+x_IW#WL2)U7=G2ygM;<5F0jx^l<sdew(du)n*gi*Yx{?_$Ku^o5}
zvd~u7YXDsbOo#>e?{B=%&c+)h2W%&vKAbhRhy}QDcI%t=i7mcp1rY$HV8_<u@_0tv
zwiIO0tNV>&_YaEz5qBvLZ!K4u2R;#+9*p!>HZ~KeW@;z#!w>ZTee|T}9zWrvG%+x6
z%9{B8H;!z*#m%>9P$C8HvtrtMsgqPBp2a`RMG55vKHz%RAMpmEdQe0}73t)ScR;xd
znd#bA+ynBLc#Iov;4O#l?wq}CKDA4f69$%0b>OjQ!<Sch5fVbEru_RFQCO7Rh;0RD
ziDN?i`Jb2){GW@A>JfR^4cmsm%>03x+J&qJV}no(!z4av`d6<$jnQ98(E6ALy_-*h
z!onM9vh7TTW<xw8Vc#sY0@1q~z0g=*{tis>ATpHp9B4X5v*SfsGALN~gIr-c)e949
z=}f?^?z)kfkG4c*3+R~Fq^T78GJgKOA1$H`_Y%pps5#LaH#7&z^>OM@cXxMof$DVr
zT!5Getjk$KVI$&hXL5rI)zP+XXD@-n2*q#aYoYgmtf1}f_2^M(o+Q2v3*FY{CP_wy
z45>?uTk_gV;P2)4ZJoO#598tn0X4Q{8p&918P`Z2BTn&yMxdvl?N{N$M~@h|so+<1
zAY<_C`SX)|+!1bJzydrCKta|qZ8Y{#y`uxGTfPftAv$y`&^5&2%|MluUvGyS5)aQj
zFurt^<d*|8Acb%k(}4pvJT;}8b4LUJ1K$`!$Y%gb0DqI8M3fO#klczg+ekc5@biC0
zYZ0Ia@Fz<8pQt`>t3b{PG8wm#@S`M2P2Fr}gJX&?I(fjZ`2g{bwX&VeB5p9rUE^CJ
z;X5eBn=9j-RbHs?W`zn+%A)9kp!@*vaAFYXU^y^8*!bb&0S^Z_v<gBxm$)yMK;H1*
z)#}4AOwb@sFn=A?1BhNE)HQ(lEjUu)bE7d~-jP{`Wqztktiy;tVWRoqxO6ZpqsD!3
zK+WSJMW2G!CqSZ|u4yPWU|W-Yr(+i7_)ek2#D^vA@5g_U)B>_Z*a;#7+@{^70VYKQ
zRXM>;#3B-d#28g#B!gTy1ZNWex{fU%s<2g3$lt&`fv~V}-z<#8jcfXWQ3+r^*)mr=
z2J#v)XigRtx#7Y<^?UTl5eHOxGtWR3vp~b<xE2w_KNxI%W)9$%zy&~HQ&)e%s*;2Y
z6Xc)0Fv+D)lxMjdNS7+la%wz`*X$+s!Y3D$#MlIioao}$5^gvEAc0bu9VHY?&|y3B
z#{(}N{bcJpZ!iM9xpMTKJPU|Ob6|^J71qS;1oD0F8uSx*i!a21>;h&QMz&?&3Ib6C
zJP*jSEnW&Q!V(@zm{iNFqewFYgDYU68-su+H3%<CQvoA9wV*+O<E+MW_+k0jv%>gN
zU+w{|1X1^UzN59BT~0s=cpi*32+`qu7HN|ykiNqBntw4UJZ@mE?X)<JU-cm&g8#OX
zU{1Y_<n%%m@&t7$pae6Kr)mI^syE!G;AF)$_@K{otHAJ71jP)y6mF(s73VN?TJFSu
zQkC&EC#PJdwiP#t5LxB7Z|uqucu{Gpa$-l0pt0}$>J`90S}?Fcbdizq?h3;|I$Gj{
z;3(n?BO97@e5%eM?)$3*#{%1{vVuNMH}gw8lW1FPviz_kgu`>R4BG4XIz6zH!gf<h
zRFt$5P>w?{XNz$h16d=3>?>Rcc;3J=sA?9t%$Tsk75RizS$qS&LtC3}egRa7UnO1@
z6%|myR9Z{;SA)}wl$PoK2zom~lypK)S<&VV1(uh7nV2OZ4#G_@=gLm81RQHaLgXBS
zsea8wHbee!>$blEK&2av{VH}P5eULS_YibErMHBlF@S?-f`rf#lSy0*p<|VwFIc7E
z0NCPn&Of>C3<O~&a}Aw<%VtD6=l#V6G3bH20t*FrfMElLyTJw_B-TO*N4m@;VDX1X
z5)qF`Z1VrYTnF6SPc1h97s!en1mzxiPn8#Du>G33@G3$7rdD5oNk>6c?W%AKQcgJB
z`xoOCg`NLgbGe7947B0r9WSF(fnFKZy%M6Y^L{~aVXTSZHOm1pa;D)ThHufU5)crO
zc}+=8Z8Go&kCnpu-<VtqJ{*y(>g&fSq#ZrM2a$_~VsI4@?%`iA-+u!Fcq1btsA0h$
z)|l0RY<yR?*&<fLex%_mG<kp-t$c)eUmkwLiOvO^1rd;KldxAq4+4Wm-ejvF{ot2A
z0y9Uq3RJvLxw$)cL9psV!?$nOGdpo3NnW`!-g}F9%I<%cJq*a!#A5#1xnqZM9?5v<
ziO(QmR4D;xEtMDm#7$7OCYL~qLhP{}Z5k|C>=|#0t3hW09XM-H(zCL77EVF|_r%?D
z#7C0mZ}L|5_G;<YXrFtgE5*pev!jZ|;=s0T4OhU0P)PfEmUs43-_~Y$&UGd*87Zc&
z?&|>!P|^<3>!A;dqh4(Dk6}$erd6RkBwxvot3Z3{BW{3^K*6+6H`a+`xE*w_Wl*0(
zToN#ShUnbRvWEW)Sd%69skXu#Fm@nOfE`PI(xCe(3?kXd^Kfd7F5L<_XMf;XP>^_j
z`cYjy1~m+-kc!gMe4zB`O2q5S5gAyCEYBB!eSoA{E;h)0M2&z86Dc&DBHAfPw<GXO
zZC#}M2*Y+Tm9%DKV9Wi)tMVUrcn?1Zck=P0M;IBD$o6#K4~i?{*$f5{;%^sPC&f9R
z7%)MATT&Tf)A;<DydQF2L*#|mUf=k5p3AZobvTh!U&|;W?hHc56wHw-FFYEXC;_Om
zp){)Uf*o!$dzcTpPTt;@Ha2(gjnS?InvRYCVgb(|d^5Nd{6y%$ULmM?Q{ShsILKc<
z(}r;#ld}YRhCY%%EalJ=#vPid@b)oOyV2w+C~(!(>;%LJa`7e(2C|5X@^Y$7q#N5T
z{@z<fO~qo0ui2JoZ?ToG;;$)mpu&TP$Ysy=ROEjF4y?$w|3&tQRrvo@E@swz+dO%p
z4YCEHQq+RDg?plO5GvOuomq)%L=7gXVQM;t9{cSn)i0ry{XPYVXkmH83BIA4uFD=)
zfiwe#6qh$pt3P;ihJzzOFzwM+dPTh3|9AveGy`$e5KKvZ{TX61ERWsd4H~PO38(c6
zU2ffamrJ`-Gz761ql^h|RwAnt+}3;~PwNS?Y0(;b5M1`wJ%pPz(v~qdcW@I8r!N0h
zSpjMZ`^Go|EuZ~cd8@7ZNWK<9t{;uoF-Guu;Kj|O<}5!CkE1J+l5o{nMB@dVHaIYy
zQF`YWAAcLU7}yTqVbF*YeNNT*1hPs@k)XE?Ws;ld@u3yKum_4HsIbv4$8EgtpMNs4
zUW~TgK=dVg{7AABq(Qb@ZoERjyt-P`jOWkJK8aifBc_|khl8hIppoq5bp@mkyc|Xp
zFu`C9S<I<hf0q$K7y+$S22uFdyPRkTI9@a&@g@enU`Yb7iie#&6*J2iW5WAm3=i-O
zWEdkI*?69z>k8Es{3$ChXF7FC&~4*KU0vO`Z<DDXd*Bv}Yrq-EgppY-K7|Nc(-;|0
zsX}ucD)GRAjm)-7bM5r;Ks^yYb4=Ma9U6?!e~m^5qFZ{p_eH(%YV8km#}5!-RzteB
z&J8jOtVmBg{sucB+tmtk5VnFcAX}EjuOTUTw;YB!bP7<`c4<o>xToiJqK-j=CuYw-
zG%|7lv`EM<yLOsDJ>e9l1yNWg6)AT*I6D(bEj2au&LqKe=NchOnOThVVeG;=2y&m?
zFawA4*M0dClhuP%9NH|mxp7V1stx@$F_E$wIXgE;@HbzPdNjgSLUrM>mDn?a8p%UG
z3Gl;AzyYMiAbJVZZx{<aR)1nyjS*~%9U=#C*Y+W``@Csodg$=s_`D(b0f2O*t~E}>
zkBM8t@l+_n6rXeeO9%=e<SiQ3|K7D@2kbhzQAvpgM}gYX<sXM;%-Dq)aD*E#+y{lz
zb5)FV9%GE%>3-74FElitxZf;fjSdm>*mc^A1&##R`Df#asi?QBywHN#z1z#Z4dX~L
z@+^Ms92AmR20RUdJY62ThZJ=ji0ZarIt7<h45Tz%l8q9c93JKa&;~9@_m%RB3O7Lg
zeXrB*(!0(#;<hF=R>!VH+L543PDyDCV(#`NR%-a8=^^<C%``SHZd91-{P`U8kZ=~E
zjxw8do#mif4-#HBw@1hrLWvKE;0C&UQV6n=chSN4B~aU&9sI=)Qi2H3rey)}iiMf^
zN`*e?%z)!)pmI>2{&F8>cqLl8>L}kZY%|a$A&j;Z${=JENp6F9uU(sjQ~2EcyfUZ8
z`x|0~?bk)=fqO-#13yIuoE!oIgx}GNO%IZY#1~{gnnWEIANUAkzmn1S-YPorUoL=f
z$}qvsGHrZQh7O2fa>FG&?{-uTAXS7WY+!*MF&DTiibL1h8XL{iOu-WQFEtHKlJt3;
zMWTZbG9Utu5!A++nTb<M9A%V-$^3(0BLn1o>>2Zgb{sh_HlPwG(MCl(lr#aSPX#>K
zbBcNL6gYJd_$x0f8+BZRzS!!*IK=swwLF=)P*!3kmXJs>C2c~J;+%uv+sMdn(1BWI
zaN05ZaMRE3*Jg%&)I$Yj^O6d_*r&UWKcx6O=Bn4D6IWT94v9+dIB|gAhv!*es{78^
zga>=vpCrupTzb*tdzG}{(4MRRFdiURyvo?ISW>tsXgyskn^U)IX;awzR=u`!j$wU=
znXF^$GXSRsu-;(;_0A<0)BK1ZLz1YwwNYg9j(NHpBUGZq4U~!}GW_(C+szMh?t#OF
z(eW;G3vyaovtLZ-!7xvkk8xkB#7O{7Xr@u1D~KlWbW@z~*#|TJd_5Pi9YM8(jxC#V
z!cQP3KxVK?gp44;2FIKYhJ5ecgo}5t&Vw17=M807U|W4dHvvx_2VL{6BFhtu(qU){
z_1*U8b(-i;BZp0q6{QUW{0&CPox^SjV_-hNKUGXTB0gU`&}?#Q==xg4P(S-20-q^3
z3CMv@fB;7zl4hphDK8%>=xBv+)6|cDDFR^wS|Ru{BA6~3B{>amp{L!84P$?q9|s7A
zK-zTeI;woY_&LZwQQku13VK)CAV2|o09FQt)$_s|Xf5D_2ETlHOJ4}jJKlakC9gUX
zV^)-u!C_&Es~^BrLfD7G?>S{*E-ovCM{(&XFk~@xukiNG7vE{{WC2Bx(_7lvz0_fB
zyNb97RKl-pL0pUn{L&fzKz_t<%%tUl_k)j&W`uz4*eRhI$fZ9$_SfwNJR<<Oa>C1(
zobs<eJ3*X$lsYJVUqnX2p7uFmHi3aAOpTpAy@3yYrs%u|sN^8#JqawWs7QwwQ}O$4
zs~7W+e6RmH^{CeZV+fh{I$4^3tlxW~z8qU$g!K(8XIT3CC=Xe0az>Dq<Crs}JXYra
zSmQ#?It7XT*!g2FAoDEvT#w{e(F6g&iFHSoCrCxo^xeI^o1l(9mh}z{+=QGd0Ty0c
z<1AWJJgQ+wdtq?>s14{wZ|tGxuLXUQnfeyiiBC};4oAdq-1v#4-v8n_PGH}oX_WNl
zI`!^tv*BV(H%DQRUNByKfB&CzC^iZ<X8@`XD(#ybX4peU79#X){irfi<R(GV|L}Hj
zV4zYqw;?s2sj{ZVc|7%FY?<)#z#*;OSqBRQLu8GOjUkjAql9vfaA5(ZX^>WskMAQ;
z(Z0&e$lb&zBm^uAKJ>WEU2lG!RQA38*I>!gh40e3x;;o+pwpuW18*b@D87+{xPdl;
zP0Pir?Buk9No$z5LkwwpY>ei>fd&je5XWHM_F2jlf&`w<zP)>aE@XLOWX{(Ya2F>0
zFv{o%`>NE_<LmK=wm5P?|KRz38U%ezeITM1Zf-cAtO&vtPe`7CVrUFD11y#C5}5eF
zBgCX{A#^fvF(^fK{#&>;ffWsvIZ(0?8BuoLM{&jU%?dx+gXr`fq;wKWRtk2_7BHkR
zl8Jmb3`Q|XcoJ1I8&H)k1Z1I}D)#>cUKEj@NbNBBg3nd3I+ceb*HSQx;X<$!apT}L
zeFxfjChDpK_9kqdT2eHcAu!$C`~icJ2>Qgp#j|)Y9$gScqwcSm;8LprZUI&*Mke9P
zRR{$Rg4v7rnaK8ePYEA1%tUBRM+0)}>GPoR9|C|zRetmzNMk34RAoDfYQdY}Wx%41
zB{7;HgV!UCgjx|X0B%qo20Y~IEAum|Dc>>5{w<k^+BkKr!WgK?!1Vjvn}U0|SBBvO
zw?<H80d5({-f2te04;@i?esS;3LaIH$0YBtj(aC<fffSa#y&YlM+Y0pB>+j`6ogzl
zGE9ele?CkE006_t1rk%hLaZm%0;M6xQt0A{<L}|&fu_5*wl;nW+21ZQx|CH{voWQ-
zd%uctAAqr^1DG8=LBL{wSuJp3GVKJ@d;*P3=qP-H+7Cbz=)tRizR1dIiuHw%@H^Ei
zaZGAX+&lFgSzg!Dt@b0pra@$#K!S@eYa1zwGzHI)&#Y|%kpotKfz%dpp{mK<fSQz*
zm4U&bqgGH*fcQgXu0RIARqm;(s>1UjnrTQh@qq@BaR2(Hfb$BO2Nzh70FnL}>$I~J
zg?6k=Iv|A+owv64?uBw0`OMEFFB~^-zd=MOj3y!Kjvq^rk1590OsbRJWXuE17Pv|>
zf<@sBs`3>i?t9(Z()D^L+`eFdT+4#081UY=CH=i?WUKGC>&Lo_2&duR`pg9eW|L_)
zU`+>L%SLPyV`CUCC}I_SrJDeAn_WA1Rz04bnIYz8#aVvwT`u&fgry?W>d`oZe9|iW
zVJjh!VDT*Zvc_{{u{tn~JmLJgXXz8t`H{)hlVIHv3PR%7=|CREP+c&Kf_MxQu1hSA
z!k#y3Q3yq}omN>GKSY92ySN}FBeRMG9gIs-l|)P$vHG1j`27BU@FC+}e9lQ7u(w}C
z5it^JT)y-P6jY2HF!V<10=po#500p(eA6idJc2AubA)(=J5otVGAZu|WVwoc{t}un
zxsNH?RB`=e&+awbt9q1vvCE=JDoPEZz;Lhcp`-4cg-gZn{xTP{GkS^hrq1X9W?c_$
z7D!U~url<Gp);16sdj+GoS5x^*RiVbzAh-QVAsgJ<dA}FFT&<7hUo64K{FC77;7}q
zI;)vXO<upF*8YG7$hbC)$nVQ4Knr&^7}JXp-#NIXh`2#YMy9qOh!Jgv{>i2p0^G5-
zw0uUYh^8or)1Ki-B$i=@rBcYEg0c;{8I4c0S`<KJ&x=Mt{;L-nbKx+72u7JS-)WYP
ziO`ZKYMuPx1;^Qbg8tR4#&SSlvoR0o&EF6{1RDKjEK#_OLfz6Pb-NG_u};pNF!5J}
zrBOXxW#6kN7r3tG#3XY8Lc-;xdY>2h6^10=o{;omX@8G!g?2}EOiJD7{(ho4>#B>1
zpb+*oyt&lg>CWsLtl=+A^@RveVaueDua$Zq>N7eA4k$EijaI*x`vo?eW;b~r1BLZ&
za1xF8j3p<1dBxWi7_fB!zQu)A(-RREb`dVI1#vw4aF<_GR}YO@*o)6W<TNMJzQnbf
zVli<`Fb~FvVm5RV*JEM(JJ=L{M|%eA*5OPq?D8E0Z8L!^LP$*EHxe{^?*r7t|9}|$
z*(p2)5GyG_(}t&?5r)u_QCyZLT%-_lVU)8^-|=K3Di{QH2k1;|Zg0OM^Md8qZ@rfm
z3$V$N936g--@o;URe8j%&;e6^=)>K(Td+}t=0;1Jjx!GHa)6-=I&=8)P(Ai9Vo7}-
z9^Q$t+Oh9R9fa;0!9<f-1acRRCqRrOePljJOWTE~fQD1Dh>zGAL>n}XJ|Cn<^1>+n
zA_0AO9C1W*yY&^@L2E?uxZNP^GHa>-`J^jcjJ!)9yP(OVt@HX#Ogv#!l4^r6j7Iol
z|0jXZS@Lb;1STygH9{Q`A&bpBCQ{w8ZTU=^F4+s9_7EJ27R|_~W(~!Wg~LO$%WhxL
zBCIh(w|tov14{_|r2bTX0Rj8>?p0cfAHj$N1|Wks&0y8ng`5(-zd<c(|I5vQAyH(t
zHWYv&VcL-yH(UAyU$7-zZxczVGK)2u9SC4x{mcL9!2u=_+FhWKHPlh)n%DCTO-=;K
zu!GX+bMV7G?U}rA8L2=N+Bg2xRm3zPF++&y8kpt8dVK=diq<0E5){Y|XQ=aUfTRUE
zrjmwHV%%IQz7E3ybFM##WOaEcf~0`L7Ba^LK*-Yz{%AfsA!+kL1)V&UE&>=wh04^R
z^{C*;jvYJHg&!-wZ6qX;v9<c@2GA{y{TTB+F*b7d=;nnR<g=FVGm&Zoe@6C)nFZO1
z^OLw3fOBB@08t7>z+3q{2o%6qs$<YLgs}(aK~-a~KC>uX9>Suda)M(8s!C$ySAGJ+
z@@L+Q{IgFZ<7OGMA)AjFF1(23fl#+(X2uso9Y2G{2r;yU_FQ2I8!yc_RCAawT*qHy
z&*HPpqalKzj0PuBA>8YRj+ia>LA9leiwh>mGBFPaqBopy(1wMq<PBUL=#RFmNF=x3
zpRHofyq>3i?OH9SeRycD1KfxIpAGgC{CH5=vTNmBJEHPBD5web1p;>E<*RaX5S$XR
z9eXiVk1s@|r6?bz;gwxk`3b#4B<P-BANW$SuYfu~>l%E9`It<4M7`PN1QY}$7XEce
zQm%gs@WvoU0*K^H=C5r*<wdNeE>g6>%q|k*iB^jYB5{L*wiCuWh^{%<%x4Q9V*`zY
zP&6}`Dx2I2cYrnA&{%3nHo46XNZq+JfvgY?N=nG#o;=ypp+Lr89IuuFrcI`&k%*+z
z!4$d%s#uYa{|>Yf1GboD%mMiS`Jy~nz<58xVXr+!!@#iG7I#~p%mr+i@N*tlUU)u!
zJa$Y<PR{M%A)Z?YrklEOpq1nf>v^5!s(-^(6A|<G_=kGoL*@PXNWSn8bL}QIVL66K
z3d34NCI||>X)ip2XQA26PzcckEbPcjnIh~-)a%G<Y>bdYKegBkkK!^c_1&(VNuVmI
z5}PO4KcHox|8)V9%GgXy%*<>R8kvUPQ@#0!AW*ovkM04kKIGqs5kr%cschDxO^h8g
z!u)8ZqM@37&oTpi8>Cu=kakBg2IfDw_gDOB;%L8xL_`Q+PuL0<AP1VSsklmP4O;&T
zXsgI@XyXA9-l9y`dv%eO<JvS|^B_VO{vN`kiX4BSb0EDt+=M3_$vDx*b2)ce1NP>~
zmMJnnvwp(YLqu4>O2a@0f*0sF>V38`LE?poyPt%F1_S*urG1F}IEbf(?%HWt5GfD(
z=O%hQf`UK3e*Fdfk>2IE*ww2~K$-FNl|c><oCN8Ek;`o>uz~w}Y!TJiJ04xXva;!n
zeIIgzbjSDSI#-6js&h8Ds6;{!;^gt;Vw%_Tle=GCwD$W4j^s9TL)T5t%}DP?Vo;0R
z_hp7ln_t$Ds)h<-hV$7Ny$*u7qdWo70$o41jkVl)4*i<}b;vK`%236Xl_Y`hz_Q9h
zOgsW%!JP9x6$2!E;)fd=C@vPPz|R>S`0?PX+FHLSPYzViPj;!DZ+Y<7_D5WxY(Gy`
z$Kgbq-PQ<gLQF_=u%C3f%S+>LBc6~Ln;SzsNavRmF0GJn#c>}yMhM!JGSi6YEg)C!
zyKNeTTB+v#=QA1<a&`{+)J4m-NceZZ?L8qgdflrwBtCu^KVT$LRrldM&&(u3MdnNy
z_g|@W4P%D8Q;D%S4>I8f+UDb`*qBI_*{)z~ymYYTMwHP1yO42CC0jG~_es6z*_1<h
zwQ7`sv38c03QI$LhDcMyA7S;S_I^(u%vR+#3|>z9n{8_N>9c3Y1uHP<1$6odE6`@y
zZb302CkY)e?~Fbw!qw6Jw<yxU$8@l_F-7QoJoYps1Zn7nUm1q+l~P_kkUmq@1l%tx
z7;6Napot4HO+2#|v+H1;XJ-2Z1W;+#qRP6eC2E<Rs+I6nyeXKY5aR|e^+5##XMtow
zQ}F6^w60#~(sB|p1~K#|Jq1~?%=RJYGT9DLKj!}i&Q~)f`IHklCX$1&S&lHd^Zp4b
zsH`$0e!)8QT26k<XnCeq8ZLltJLEG&CsKF>&+WykKX8Cbdx_%(mYJ9Wlf8DWrQ4{=
zr4^Z(bUp8_V{~Tvg<~~dE<b#Kl~q?D?=6-Oz_n1-+^nRQg7dfh7!yGasO$F$B`k3z
zPK@2!d7hs6%qNbn1)j%CURRKg0=qHt1(EOy3v_Z$eP9zr26Vg*&{yy{v;$5{2v9=R
zXV172*!a?XvV_R#@9_e_X(TEzIcC$$N=O?(J`}ezB_c2xlk`1BJ=m{vLN)Tjh?ls4
zBAf^s496>D7vZ{w$_fA-N0FriFU02rPV-Xu6%n1|&XErGqoOHVof7E2L4Y7fd=fvF
z2F4^x;4Hv`zyh60G&S*%4LLOl<E#;pj_}0g3;vh;L8#O5s%)4bJaTG;(_+$s-+y`9
zy1mn#aaGcfPN-8J($d=ccCmFxk35~(Yb$smdo%q5n_6o{JhYf&`NE8V3aY)mCzX?w
zr8J`%2KvvSdb}e%g-|bEW&@jLY0gbvY#RwiLQjJN!7IAM7IRJ?ZN;QRIM^o>Mp~Iq
zoG6wjT-;$7HR^){eO;5Vy%2vD1++Luy05@8wp|%(^gJ>uA^&$qK#YqvD)pWzHZ&%H
ztrmM<9_q)VK)D&mh~Zm7Ovzy`3z$3V4-{rK9Lr-PBS#J$dV5&uIK3JGAv__1T!e)L
zBzOjZM3&#YIi~+(CnN~77i1&2_u*rNQq-g+;aOIwX`XKQGa@-hpaBHhxHlW~<Q>IL
zDu3G|4Pc_}z}EXZ(Kv<skfePM-b9=ml=2Dk$1Ovuf?Qrri1Y8II`e)l?kxH8rx3C^
z6@6DiK{=*iCrGeN6AA27voyf0jf}&#`(K_z4e?Jdt%S@&ifhHM<HhZo$T2L;%ky^}
z_GGcMe(m17mDn_%Nj}ytnXu}#Fz(O+%z?>0O?v?d7Xj#S+KKBquB-W5M~B9%9Rx?H
z1uYrdkAQjYQ;h`N24z0G2Ho*$a%hqg1vqz(_bQfif_YP9bGaU10i+<Nfn_2I0``v@
z6#YFZnd+$o#u|*xBN;*Ji>rVY_zVI&uoc}*Ln#xff)3IqAyHnM>!^pY#7($xyEl>m
zmcYh{=n^EVond4wn+@~Mfno#lw=)vWnfr_`k{&zUr!ExDK9INZz#ZT_b*GU7vhlvE
zV8qdv_pjD6t^YFZ;Y>d4dqM2RKu(eqv1C6u-p>iKW1pRwxjXvOo(?lDMB_|i(I?$r
zv&r$md9eUrU-<p`Sm}FSKZ_ZZ8DcE(t{!M3xSpJL60x`2|LM~MgFE?Qcl*1X)?(yt
zIyg<a(o{j~(iyRH@&G0ZT;pIM8vA}J{5b5?a?!D$U&75OjG;^<QsaNBh^89GW@1TQ
z@7_&&Ch*djCmg#Tzr(}IT8*o-yIbhQi9S4;AS!4`A9O1|bO?ilk&w{&2@P9l73ADq
zU4rt?G0<c_hfX8y*iR>DY)IN_zq#Q$%3BFek%<&Ec_d1X7<D;WN)UJL4ZGpvCr`d5
z@?AQ2(d5=Y<Lz=k1fl8BnNJxNM{=0sO!r^YDkFkl>?%>IA)lp~qWq(C9inE7S_HR^
zWD$@-wsE7lJsg`$s+~_#uwYMrrp2uv;;H2w!mg#N9w%81!8W9!pSxaWQDKmcY9+ne
zObSXKEa@M5;6wJy>rvLnQ`+AkLlgX)mHNb)lcCUL0dc#)xs3bWw}@E=!tBG&?2I4k
z#FP3zAJk#a{hAI1VH?#X$p$$ec!Z|dH^2!KJL8#MF^XWyHnBsr=%(melp_LL*yYwC
zBzbmW!la4J4>0;25ob^;nod9J%h<>Idw!mfq=5l48az=r?7)BVb&#YT>3q)#{haCB
zRYZ40hQvc){)znt`Jww^#4ZzbU1<qwhCZXrC<isFW5h~|jWe2ABiL0h7GuwQ&+A>_
z;HcC5y7$aXljP8#P3i5bjh+tzdyz28{9wYx-&1UiVjkGg<0<q{limxYGDdpA<?0u$
zRPG!~Fb`}EHi`rGdlq`K5u$a`U9iu?(^{4-q2MQLqMrj2BwjDpnSQnv4Yb(Z>=S}e
zmy%Am2XyD>a@e?;u=5%HN>J_&JGGrwNqX_Lh>E)~j|<>lfuAl?kT{u8unoO$H-oFE
z#eY5@_hJ*zONeStP#y~wAR{Ju0ZAfx$D+dF3I2~jt9x++k`y`3qp*~lp3S+VWswlm
ze#Ce@3JI!qR1P4;_px(NgM$}(oTsJKxRH0gb7+phi^zuaR83rY=EN&N_Wl@APob0y
zVN;Whc>o0Vp4LVB;h#SlBIRPQ9C?Q}EysL0Ng)8TvAejNQ?Q?3o;i_r_P+cThp+7U
z<&34z0w@j<OvQK4s;z8oQ!%tNmD8ggjJCmEnR>usBE76&_33^xH&m<=%=DdRT?;zb
zgf{Vf&|ZRCcGn^LS?pd(GoEmiQI)y_Xt%t1s!YjkWu)EVdZf(aq3D`a^YsG_YzGb=
zJf6p(3G~v_=&zlaB+gkF)QH3+GXNqHh-*RsfKWGKKaT2f;{Rdq&BJ<bySDFeI+2i6
zD)W>kLx>_{Bnr_e6(Uil5J_YvMT#b*NHixzN+K0PAx&tOG-!?@dOu&T>)oDhd;fo*
z?RmDlKd$>c&%5FGJ-)}Wj<xoEU;C=B7X+p7c9G1onkst-N2A8T<FjOrOFx==N9^X=
z2{&0DeVucA_wActdUH4uPE=)6r+%Y=IQQG^0J&z7`r^X78_Qlukv|)4HP;)nf9;kF
zw^W9w4(p~BW@%k!vZ(z-k)?1k3ZB5dv)A#7(8KCfd^YvpQF#S_>J%Aq-G_&mzHqE?
z#Cf|9f(@-cdEUUL?oRa=)T1`6s4Rs&xn3u2E3s<VP%e8CgP89Mw-hILzZGzTG9yE}
zE;s68%b?#}&FL+Pr=-Wp$;y(U>s8LZD}T#lMU3UlP2(3$J)EPVu8wNciHr^g6HRCp
zi6`tls1H}CIdCUyp}laS3$4kwD&?~G>gP|_x*LW2pSYQ6M2|b&k7Y9F{Tc6=CQAru
zFlj~09|q!s21R_IBup9yaTWjF($HWUXX|~X=KSZq|J^y0)>==|N-ehHY84$r`$N;$
z8i8Do-}3gi*EhlObnW2RRK@}V3z7zHJdRSCYF*yKb<*$dCI{j`Z$L>Hs9ryfkiPyo
z7Jrxd<U7RsORE~zT~J<KoZG)ht*~uWX}n|aC<^%Ws^r0)5wBgU!bktM-<rM2;8t%(
z>`6>z?jj3!GI62~jmeAEm*2M_^L3i3)*jiNE3?wl7JM@=kX>8$-EepBJYwH;nc)F_
z0(^aKI*M$41cxXH(+Og|Jgey+*)?v<Y3-d{iKG!dUF0rw+qh9yUF+#Fb>n^;dy766
z0QohKL{Kl1zLE)JkF%iu>9tJMtZC~mEp(R|@jEsTOuC@^YLyX7Rw=tIdN?>$x7u&o
z#JSEsXKfYrg&h+Qcil7jG^PQ+0Pe_KwgK?qVRE8YU!{9)H9qid4VY1UWqS~iK-!TW
zv^>l?Hg}oy?xA4sN7LF`e(WooI#N?pvpDlee8qm;0@?iH-w%LI(3&itc<%ZLd558=
zBo#*{HeXi_IKi>cn6<ijdOzq*&#xMiN`^WYHcYDOuDa^46H);Vt6DYbWVIkltEMiA
z%q#WWqV13v_DtL7*Dlv1LZp>BRbN@Thhf0@ZA8G;AtM3U`~`Hqe%fB~WiFX7Us6cl
zXmfoDzwRgfcy!E$$}V@cX7x4R@HW#u!*%Ss_acq+ruwL?xpoU4fbL@5OH27+xzt94
z3rPJwE)4hi^`jzD@8X%p3GpcrGeJv*0Rh-Daq!_avG=}`BH~ct_T1-zUD;N9Y^&;A
zDt`G+bek`i{ov40ZLj^FQB4E3BHwiC6@;!*Hsgj4?2Ar6rH%}*=)G6oY+S!5{atx|
z{e*oEs>Q`p>HlQ;)@I*vV{XN>p#yEs2-oe5A6c&lp07PzEcYhaOURtBlyGo;>wc=*
zzEh|=?+;wn_#&QRAEmolJ%Ad!kPE+uRfWG4Ey(vh-uAPU1G~Z5!*!uX#l`GcH{bZ?
z*Wd|6@t!kGHpaat_Y(%$(q6EX%ud-T{F9GN^nbMgEg)`k8YYjLGPTeLf(SvI2`1fS
zT}b|a0~*|N3#c)TZ4UbG<2s|?{gD&wDjzfIWt-6{aOv{A+?UUuon104|7!D^jj@_i
z)B6#Y#tpB@S65dT!y3;?lj`nc@c<V^SO0gpZk-+i9wU|}G$<9d4n@EL>S}7XLi6bt
z2%EC~+u(qvAf;8#Y3Y1$A)rTo-*$CL^K`MZcUzTy-!%ZP0n+%5EEcy;r>qt}9rbtb
ze*L!LpaUQuWjASi$s_8&K`ySQd*_9WslSk-;B<SP*|FOBvm<tu>3S;(x{*dtv_T(6
zwyqhW)od3T@8{d~V&>RMiwFek4c0j&4{p&O|AjN5KlQp}+TcTA5RSIVIo-@GXv0|F
z`Q5rQ6+Y?tnY(AB?rR6%2crbPxwdZP`kw^|`Nc(LG6JiwfddSQtvX-SHKn+Cb=$E^
z0Rj7&>T59S<PFi!iE9p9Jop%Jf+$%;lm<ae86SHX^#){hj~+d~*W_j8W_Qq{2GH|6
zd)aK+;_cacU&k)~<09i^*CPh6Lm;>GOA((>6rTy`5oNq;`-c7*bsHkuIh~ueVpL<J
zVn{5$c4a3;eYGAmu;2(ap!CC4EwerdNm}i}#RdKfus4#T2+FmuWd8TyboW&?*pxbR
zY1QI}^xH4g?ob)_Lx_Op2?Ub>)j|I@@Q~8O4UT+jkhc%JtnUy{xYOo@E+cElmB${@
z0MPq&@QBZ!MNPc>j8nP#)N9WYO;7YD3*^5_ZP~cDZ-fGs>#nWGu6#TcHjA`7$)ipf
z7yL=7>bm7EpCKFN2c?=tPICTy>}Q3FMfYXFZ_JV2-@Umx(_+G5;Z5+HLTYx}iN#Ei
z-}f=))fT-mzm|+Fv~LZ0cWu#t(r>J+-b{-JWdd|DF3k0`S&{F<=92pQ&epuuf)^@k
z6^M|69>bjVZ*N#KA`ps!tTh)3q5Oo|#kEAol;2`hhET0lJSB=~CP@1McMSFNsn6A$
zE}vxQw9P-hc{`ZWx#|6mON=od@zdKXD)UIu_^)PB#%rILtKt?=P^!6W=;o(>7bKO9
z^)9T>Gku-AN{DL-p`$)5Iy*CTP&c~@TSX^p@}mAt_6qSRm;AOF&->`8^F6Wuj#cfJ
zGt{O$(H%mzK)wh)!?l&nH0xVr7i(XA7NFU_S$BOP-mA7c?%hi-i8HVX1r-IDPjYfH
zbeN-t*eIJ88QafkNB6v9;`}8S$@)hjDUary(;4V>FL&!=z~7&|)U<^dzKl%#?ti{k
z?<`2PjM9XCjUQj{l889l=l9op#TQ!k(Xz4hZp4*Eq=eQs>yYZXb>c4W?W20<UQq3h
zCZ`$OgqZzo*3m2V+Y8F44tYFw%&&oW>kjnqZj8g*_I<ea$Or6$&RF}^nY)bga?y+Y
zq>8v}vTr-;^%A-k?Nz|Dfri2Z$d^ch0428}A)U8DY5Mf{yS1V?4}$4wq6NeVK>!?n
z+{lqaw`EaL5jR8G<E`9esO@^!q*k?l7u{SrcXwV_JBOhc*sq%<Be0FxI?80l)Uam6
z*Ybs(@Abv)Jo_AgP7q$8{O0g_veH{mTg_&kd;f#AUfho=aH4!rj<?Fn6;rJz2rukQ
zd3o^7O_4baiCJtTnKY94)KiF0W>FAL!5P*^yOPv%9P$!P;?H{3uh*OWxi<eKGD$F9
zVF<R^)PmS`Cr(R?E?KtB<II`d!l9Ga>u}>rimr~|{@Blt<VipCDrl|f{#i*I9wH>1
zH+CQ4K3(&YEhb!Hdp?0p=8EE@(hw9hys`hNm{v91MxC|;FW!1GQg}0y)o#s=Trjl_
zn!tVUVF+m|?IQkYYs8p0vFjGDD|$`e*@AEJMUiJolyrSR%-HeiWQ*?#V`HDqb#b8o
zQ~PY}Ei`WoJ^!^BWKh?okaQ>XW3cz{ttP+wI&Yf4U_q~S%WwlsurJpx&k}rWjvWG3
zYB@XPiPd+XJg*Hp&B<C5&j-!}1?P{yjJod45;hUX;LbhHE~EMG$myXq6wxUoHwv7i
z%o-WAQtNws7K+L4bfJy7wy?EGt?_JD*POi#jg6jtFU49_u0@FE^r5He+r2_oxVaSQ
zd-2mlG|DXAUC_mD6Vr8Cf;<OVA}ZTm^>)FrfC3`kEIg2%Cuj0{gprR!(d|DdN)svQ
zv);PXQ=rsVS+=#eS0!x~QmynoL)Z6{yWnFt>Q-lRj80{jsq<#_?f#zRQL64VlP2#9
z)>E-9O($RXSy9)5?pC9?&hz#9@W~o!FJ9CH>?wLv2zSS@z=z|8F3{XG@4_odVQD>0
zc2QR4MXkD_Ujid0Y20%5Ow>laPwLWvj+(y)L@fGo@zv{`rIX_m5_tAE?gIe##x2Rl
zt8-35^iFIK1bA<<ypnO`#_aJ99Pp5HL*NYrKS(9LVn>)Kf&9mQL(s|W)_hMQf35dn
z(@8da$<MjP2TxjmGR5REG2v+AH7lff2xIyWV_*e5>JiIWI-cOd=hc@mC?dq7o%E=>
z^=sr%{Yu7}oPX%Pq5d+lXT*>aX-1vkC6aUm%+hP=5(ulX3(7Y?ChaqsK6~l0<wv_8
z!U&asI=n+;bN<9pW1n?my_NFgU%Auiuk*Jg-g_WDRSR`%%J=XlgNgE`A=E6O11aPE
z%L1lF>S|i(-fs@-&_2nfq_<n#XAOBy%9f-(QJn92Q(Yhrltx>*+c1fVjzEgmSL(5#
zN^g(Q;ZaE{+}f7`bS^?|pRz5f)koID?g8!QRhRn-UQHl?VcT;_!;kkJVup|t37f%P
z4^`NXO%eAF{Bmkm@;`U3nf-}fVHJFb+PhsiUNObbe5&W@Ex9vxeF>WO(UL0&jFSAR
zY}Bo&WbI(?Y7fgm6mVp~$l5|;u8e7Z$i1?^>M?b4_32r;wsE8bXMNrc8e?ns_DwH#
z$EWf&{m-TUO0mpkBFxi0`hLF!j7ZD+==CHZoQyYu7T)kT)xYSKDRQTSgUk6NeU3?d
zpNB=*sBsPGOnP<4b9$q~()sg`>c4!wvm?vpcd)X4{Pwkt;kI{2U4G&S>XWqZ*w-8+
zQDK8^u<%Oo`u+R2(;n9zd3u!L`htHuOugtB)P}>yje55*_zebHAt52pk7vAmDVhFD
zWBPPKk}0LWr|#0ktPsL9@MuQXQZ?exx!>P~7Pxkmg1N^?z}|Y%mEZnt!>(VtGmjfb
z*0Z4*M+^mq*k9O%`^>9VNDSl*62_RL)>t1{LgF2wN>9wl@y(!D&t~mH+l0VP@T{c4
zmXAgN;u`BXG3)@W&XM@eu|Fe`O^oZK&Yr5cr*-J&-8)B)Xp!^Tyz~>AYbjwXKYfCk
z-By@j^zBEKz^sLI;pj(C+dTIMSQR`a&xKdOXCkz<@5D~@A%;Ceb8H3Fb{yt_>u#@3
z7tEF9UO#<m17ZwVR8UX=Nzul&XsxXHb4aw_xAN~eAPw3XJ!J_Uq!%t<-umO*)W9oJ
zHV_bNYfdcH)r~N}f2-jn1ZLjXu&5G(Mp?0g=n{`4GfO))EwiY*U@-|@0m=B)x9l<f
zO?7QUCR$)MZ?}r3ruqHfxg+o7q36m?)YY7C_ASClPsRa{!aYfesYC7yeM_gd|NOS4
z@%4Qq#tOna+g10mKtB|jT<@>*Q1@KxPAA^D)V)Z-y#p^?IDmO^l1l^px^tGms85jq
z81nRaK&tw1RRz{wRRP`?FDgg6jWlM?V+@1}MrQw^lQG^d%8?&}S*4$!Op@QN$!2J-
zRDS-07-P&hXJ|)ze}P|lL4Ge(NeIUhUM&Iw+D!=y8g&4blhYt`oxnKb2=BsU-{4ZR
zz7O#nqgRx-aSG5jw>o{%jXW!i8Vs7iRc<brq*W7x74R2O^S6F5gTB(}MK4dXUQF=`
z%}MHU%~cbsp|sSVb&%?sky#c)XVc!8D`y;~x<eE`Lu6n8m@s)Vy`b$34by!gIo`aD
zMyczwF4aEW28*12N%`T0kZjEM&IR7=j%Z_4)B}}*z(EQJpFK^Iu`F=Dq#dld>TYDM
z$yYw~g!5ro{&6ebc}a0~Y!aI7vp&z!)Fh^C`=TNE7|oMCCV52_p5P5iK!TC$K+W6%
z%xZ`{At589Fm98#vpKzR6EBZph}1WmMP_D#?z3%7#<XEK-fX1P>mq&|Ch`ooA=jDb
zPsW;I9M6WG=TE{}Ojs=~pp0}zu*DLeD;i^2(vDa}bguj8UyUjkUrJ1#HS9)?E%3nT
zaVqjZo2&Sg|KSKZ5IXD3_%k`S<NWJ`cFf8&Bq#CR-v@rLDD%EH$L8FX#bz^ua+>Yg
zk8k}8(|h*mW7IZlN~)K$&>G*ebTb!JRaXAU-7lYRihEB!c(OBFOG`^?hmBA^sLPsQ
z7Ev(O<f-Mj7^#aR8V?2uSBO~&;mT>X-SH-SqJ-4t;4AMgd(zo~6;%ldo6`9=r-$Uu
zv@$e2UzC)a@suQE(aCHSrv?3<fYY}e&#bs8&?B&`XOAPYyzDsmUhVE>sQAk!aRhuG
z{<H!E1?LWGSs;Gz{c*Wj=blBo4FoiD_GxTak5bsO<>clK0?n9SyGgkQnZ@~k?>NmG
zKWkQz*MbE`nNR&~!B%gnA{-b$x^c-QRL>4w-;W%&ha5g>!uC-D{_TF^_UK__m#o*F
zku-I|thx%VS#@a*wx7$xjibU757skP2g-+Mfs(@ld_5#|nX(2hl1LTK?vH%%K(o;=
z82T0!HHQrDoA<Z}E5L*;`vKFGG()dcDp-F0CKMrDS18V8XAkG4OdK^zL181DJ8H($
zuC9B3v~1#}0JpQAB>5_q=NIm7+!yhw{7&bLvyb!g_CRhBHT4)4tgDNc-dl#K1SL+X
zE-x3B(?Yev*&NlqgMk=r;6fWtVfUB79Whv|emvg(;|HxS&y3p$)}&C<i*a@oLzlr1
z;0TwMk+FE$vSgRe-|RK)!E%tQ6?WfQ&^M61@>Vxam=G~5?V9r?;|Qq<heug|sm$gZ
zL$aqgNr{UyKWa$9LF25aGxs)K<KDiWt6#cu;8|z$IdcxC$ULXFwBV@kVVESLw2l+L
zk-Y_A=yzfGcrfZK9#<#nHrbcGSv#HkbXss<pPSnR4pUHy2}7gJDH^y{@X9Ua#z%hx
z(2#_A4Z-{Ry>|>{I9Z!E=nxTCd8!utn!SUA)Z%Y-g+IX+NWbxSO6Hq|m13G*Zj`d~
zS+R=p4Vjk|Q)wTrWHr2-=ZhZg=ZtbwUeE6)PSaonsEKDH$Qd!B6jGHWd%(Y*V4ocR
z*$7L#J+e<77q^{q0e1lvG&A=Km#kRv7BWXrOBc2@bB=ot9O&I_pW^r&!*@0X@<nyr
z#tF8>xbKaP{ZA45-s&F{I&Lk0#vuBFOkj1kqjW_^!Vchv@RgX35!{HGu}-bFNyGmv
zn^~37DX3VbK&utA>hPrnm3rIMwczD>;|M-~c*@G-Q>Y83-rn3QyeeT9--IQ>iuvvF
z=uj2J7CnU4+T|5?i!10@|J2+*8{LF&W9Id-+R-B2>1_FI@oD0WVbDfnUs+YoFiB4!
zL^}Toldqi3DOkA9ytnCUBK<vfV#a}v_#NRX*aC9{q?q<PR;iGv>p5qyjIUplsG<{*
zPI<<UbhGw&p<b=)9P_!X>=w1ymDUxRPia(LvAe)57PX<yJ_{@E-di8O$zEMLYpK(i
zWDq~mBlGWEuPi?^SmSEL6sZHVglPv49-BX8eXgBb^3UHbga`|#S=rTEr?m~P&UY1h
zroKsy@7W%I_Uu`1NaYip6SV`0P$aerpE5W1e=SY4Fs2?|P_;GhxFJ;W-4PRy7eGUQ
z89D?S6kz>=NuS#VGld9BDzWvg$wD6O^xiAAZ=1~S$$D9+tWcRav4P8F-r+3$S#`js
zqP<6D+`V!9*@2Qi?PzO5`bYYC%s>;XZmp=Spr8N-)^<s`vRt?~*uzRaB9{lJQyfWF
z)A!vg>6PdCuyY;*vMOJ1;yP?nl&>hiWA>G11gXo{uL!S(_N=>$@|p$q-^**{pp3sy
zK=zVnx$jk*+}}O8`D6U@Z}z?C&5Gtmq4LY{1TM1;J#_ZesZ*1tEG?GqvgN#>#w33E
zj5o2-A&0!;VfFB3=2O=bn`<z#dGuhq-)ZNMFHZa#tRIwws0s3E>Yb2;d-uM>ZC%Sg
zlM1bi9lUX4nvp8VDQ86XzT_A1><oq~=q-xa?%*Iy*TSb@bM}@8D4aNdb{qHhKCL}K
zQt}RN&F3a5BG<>r;PREr=*DusfBkUd-pdn1XU|T?-dC`GpYJM>;ZadvKYMM;Q2~B$
ztM-pVqg8fjpWJDAgn$XF?js+#2^O*8Qa))N3?0VkskFo2Uk}27ue2~NP4{V!0>e3T
z9&s}x6PUJccM5~=GPu@VxJ;3ZJZlRS-H@e`vvCSwf*hp1=fJbj34)4Vs7Too_d)Q4
za6@DKx-ZH)PhY*-0>sM}etEB}aX$s?>h88-Y$%9nd!WaRdaC?NCSmPBlXm;*(hExx
zoc3@#c=zyo5Xqtm4LEpIeoFM$MyCczg$QivI!gu0BA;nm2qu{)7JlX1J)RWkZ0dC@
zDkYNK0Ula+;^>HohHC(Co4WrzE#`u*f9>?O!h;OfuyPii6Z{7Gfnoe%g&f#l5|QUy
ztCT}cVv5L`unO?o_N2dr#I5b6^Nt;x8x=~>bTYaBt-d(0zV3;<FszJ5JK3>RJA&;m
zT92$fb+#nyJ}N7P+>C?iXS8rKc2}Kb)>Kk^AC8J)UJCHSUBW4W+RgXIOW%Tk6X=7G
zBL|bfx9?73Na8$-nIw7z8s{usNnv%LYl@QJ+okLI@fO0XXBTFE_XW(Wj=k1R{^Y4s
zf+Ii2%-hXd(cm*BEd*9@#;^sa%`4YJ0Vw5lnJ{e7@5~;NS7|UptS>14$$Y`d5&xRf
z2mv{QJwX+o1_pKhRYZhBuvOALZZ@>k?}@m-Iy$JLWu&BXfR@Vlw2-tnQ5G^j>>*nj
z?FcvH7wOKat+F%=G;+Kl0|d(xE{F26G6czdk?DTM%c`<+;NZa@si|3`mAAGZZ_MzW
z;KD1SYi%9xntg)ZZY(Z3Yu2nmRTzBrs?c@2ezw+To`o9MoDYoP3y<Oe#=G;|BqZJv
z>)CK}=hkA=`=j=DmbSYL%@)UbqnJEU6saRDN=r+_kqFf+ihv)hwk72NcSh{$>OKWF
z3-by$4`!jn!{!=0pZq%(qrg3&4MK;89ACkhnO#H6S{7Ravw!Lk7VQ%qn|uH{x%Xhf
zNGR1@WKVI)0p4%qvy$25lMfK$1O^z}Y6Jy1TS$yqER3mw?|?5i{2dVK`Jp%?EzOz?
zYeQZX#Vja+Fhz%@<BW41-v>V>oLs+to!rUM(Gf&6lBSt$4;r<>Xfe64=@?^9(ZK_^
z@&jaXfZwxdc;U;+>MEUqwED1;RO4_P3(G4V&6L+zS{9+rY3&Q*Y@m^Ca}5M9%YQPZ
zp(@qgN1Bq6hZ1FlV+`G0Qv<S6$mc3h?tVSTw)L(L*(P|vie&ljM>)PE3Z2b?ozO%=
z*DU3(n}>%BZ;>9Mz`)f2FBkFBCZbS2Uudzm1M6zqu#vw-BI{459SJzV1X}i$7^FcH
z3pV4tyYyr^y2xN7n`n&32xVJgljN-r+LbkEbhglh6x~cGOZURr2du2hgySTsC?P4y
zr4+@UT`+&X-y7SDBUDs`4@IgZczCm8amF-5T0?z@&@q@JP2oAsVeh1|V{Zr@uLLCO
z9YG~kr`AtQY%_{UK3VfG@96TH8bSUHZ6XjR2wxMj+@rn0g8KCBi!k8;S3>R9f2Wo~
zXR~esfy79<9pvzvioAku!3vTZt+5<sO4Q`Nr>3Uz3M+^Th+;AboRoBhA3yTO;E;mt
zeDQ#OX}K+P-hxw-{-n#`?&LB~6!!R{kdTfYGls@;x_VJn190FITD+NshlM$>wP;2^
zWSabI)Lto&3Vlciu;6pv-dNz4qPqu+(U6z_%(eubprMCGh1PH1*pWMEPh)%WTHX8A
zk+0K&f}(quu*dFh$yN(Sk%hj#;rn-_I$Lpp;d>uGF^n3jzx(rvF@rw38O2!c=hbua
z3y{V0PC5BPbKu2`HDA9Dk&$89o*>tv=l#v~QBq>}VH44%q%L&czrT4%PVY}yndQxE
z2LH@7G4gLTSiO9^%xsr$$9%lKL-xcK7Zhj|YaG8hz3Q3Vu|9`KC{LemR<G95+&uZt
zrQ8?m5fC^nGhe$v>7r84W~0^eF*=i5CaG@vUEBLBG>p^pNMi2$#1%1vEe`8-jHmqW
z;6Q$h&g;E_m2uj_r4?0cp{ORb8Fjx>iAYgUiFJ8oii@L@n+Fn2CW=n@7M>NMICg9$
zTCTED<o~~kGqpy61JCd3^N{ygP@CkmdN@HWxX6=M%z_Yk5zGFlO;VO7PN!=%Z;)NL
zRc>d_EM2sy|FCvuIfJd+#{75?)H1yv&1~}HwzMD4$jErR%I6(m3WJ}&h4-_R|5PsW
z=pR8U@u5pOQXSQ8P5oc)gkqr}C|$;u<=x57J|s!}ukGInv)_(;Vt&M=aM#nP3z_cM
zro)?CQWz&KqNiqGHa|%t6R)8oNlMMOZ|z!z#G|P1?U-9A^^l6VdgujAS1aCMYS|UT
z9Cr_ofx~2HI_-K`Bhoc}I7UkH+wJY&d{7pMZsKP|QgyTF+nN;~^l6jPehYuuTWQoN
ztq1-wxC!sdyXpIuYv>|9$sJY92B2?pMsANs$^zxUpOP8|MEQbf!}6J&wX9r=qCI`W
z`d$GvO4*>;bV1HJf3HcBuNLmg$WKxiaz$j7sGIg~R~MPM1Tz=(@~I2?!vdbfBHE@}
zRC6urnzYP$C^T@>?q8ZEG>^Ckn66eQo`FXkYZG*PIp(=T9z|@%*S}1y3yC6u>#J#?
zNd~zP;&%BCjWv%B0)NiPt?WLg?jRT)hD5I|6P-f^5q4~BY}qI-!)XV`_L(F5GotIk
z=ht*ow1yxDO_3<!@R>1JQuXKPc^>h)g#tsFwC_Xgr0_5VCQFx|NwVuwB31GAYrxsF
zYwFD&TonRZ>Ca>;7qHHo+xA=Pmt{VoKZGi8{nL!l>r|3Ru<q>{*YI{E>+iUhe#!Ia
zPgQXD$rq07BXZDp1()+3v#2Dy#*}k)FZ({4_s3pIu>E*P&#0t2vboqjQfnqP9`F?r
z33cD_<rpmR%;PFuL@ce{Y0>`k^$ybi??37HV|o94&6&x@`}@^;_J6+d?@IiCm*?e=
z<Jjm87g;oat|8y^zbsYx`g+kvD}ZBe?!mC~;3mX6>pifL%6@GjB2wbL;CqRHj~tuE
zq1<PJ$&2YzE=O)H>v~M6L>}G!<K!m%+0YZtf{QF)K|`DV>!9uQOfX3kzN$wE7H7g&
z=!suOtHlUQmsQstXRC>b)J~vYela0gau}771Ghk}(J*Y3wx3s7`L@XD;PS!3)~OMG
zb{$jVAVFO{+=tCRkaD!w?r>p8=%ow3V#3$|Y={3j1pkM|h`we~xkV1_;YXcS++27F
zT-n#$)I{kqW>y(aG_?PSZzr?oqRtH;%J->!;^^{!OI&FC*p5&!-FF!SUluQ(v}+r;
zmfgG1wWpmpU5}>!bz8i-tG=iZDXZBVDzgekNk?tur*94xLg~T7X~o6K=*DwT<pwY8
zXLk%&!h=SC_L9Q(NbmVH9mp&Hap0WtC*F!)LRSfwX*P5$3g)Ka!Y>;C=hSyU<Lhfe
zfKU-DD=C>H1neLY;ZhLM7hgHcI*y(!lKFd!t@(*H3;zDDs~8P)s8>oH3J@ZQsQlkw
z<cTFSDd@{th0mW?LA}OW?H7J%+}|Hk(ET@_$}pq2xcHPePkAuz@23q?c^;<v$KCnS
z-#<Ht#1WI@9QOg^F4YL@b@A_i-|j!{<*TNo;ll5n@%MK|lk`(37QMUe7b*Bto)z3d
zL=@t&wF9c&`c{!7W<OR5n+*w1BBp1AfH&vRhUzu<z}Rp)7-WRworuq<KaU%f0|}w7
zp5NA_XN<R}Ldl;{D7h>BMn0t)_wU;548Hr(yuW`t0aSp3OvYdm3kt52+bTTNnE(6`
zZ$>C1RLA)L`CtuyKA2+{nR6SJxjHv)3>E#e27kZ(fd~Zzc0!@%>1~^$_!~Dz{aMio
zBEUJMwMxEgZz=JYdmj=W;E~&0jY4B8Id(*jl5o}D{`=ci=WA;xV91@|`&bC)6GZ?1
zlH*i34_k9r-P*&)MJEtuE-Z%MpXmRn_mkJd;{gd^ewG#{@ZTTl{dqh~8PI3EoV|wH
z{Qc9Z&x@oU>6Sq<74QEZAlw3vUI@@APhK{P{A4<vSqq2lUmv5?MWo>rJ`QtHgnzZ@
z{Xp#RWpL0}AYk*e1-EjM1_@zO2G&O#>&;56|Cme)|M`C|ga0j;L2hYXT@gtFr3K|&
z6>Ko@>QdkY;JYPDn3Oq}rc{_bm<KfmUV*R4$lfN`Hq$`c@Nh|>+dE`$)^;lxfa`5%
zrHONM!E|Vilp>>UL0N4RWPitX5pnFLUXa*T<lsO{HZF|+56D1<Y%Q*Dg<)=K5$oAA
zlq!*u!l4Jikuax?-BO37Jt8N3)oQ=tbgF?)9j~KxcqwJAV~@4hC{$>FUAGwPP!&Q3
z?YRIAKs+BwxIy7JTg;T)j#vy_T(oy@#EAVwMZKie4|F(!b-KF?IGE5+l`yR`8Dl&^
z7rtokaiK;B6alPBOHK7_=e)H|5YAC^_Z~d}i4%z^*DuRpfg%XK_y6cX`$d@#aCl~1
zc#jw0Udu|0kg5VCSy4KM=k$od%7rfvU;*HV&`<?*#bB0<K&DM3x5(&7I0gKLDVroH
z)ug4V7snwsbQqck2S}4o@EZaxz~AEfA^>%MVmF3d2(p&nm_?Q@l#(JFeqh?b3^Qx(
zNlPOEG|;8xhD$_7AfpBn%dkE_VE4he2aV~2D7vYh&(U#_RMWLpNXRIu1D%@|0gV~y
zrdJ3*L4}?`cRdwPw{G3OS_~oRxDs?8t%ptn2jP(u0ABd{LDhwc_fYHX@F$e=${nh^
z?W)bU=g=ek`WD?YND+L%(C{6v5qtN|^~4rSO(C?%Uz}`n=En@BIs1WOUfpxYLJkI&
zaJY*KE-lhXa<CwNLk~byvaUbHk4=muK2Lllgp9{MhABwuEL|$}z*68-9=3wVQ$>S_
z(h`PdLKp!sWy%y4l}Lz+a3h(|g2@%u_o`!XHoAw$&8oZ};<SVc5)|O5J9`$?j>3zs
z<mJbs+QIgiegKcLff)R^?aybH5vm!#AeM%W%SBDuC<uGNG|;iLx0K}N<x%gT$pZbH
zI(0Xd6#>!$fdiKY%m71nHhim4dFH-aCK|Z|_d?tz>xLaQOglJol4s{kVO-EqTQFJc
zMn~oB&gHj{EA#{f1m~Pa6lF~cS`N_#<uGgkZa!=x6yYd-t^l;Y%FHxYuc-`l5$5_X
zQT|uN(O+<0UUL{sTSo3ZgQveyv)5woRy-WOmd<mqr2Vk+Kt43pMOiF4%_HyFfrf1*
zBgXIrdWC-Nk1KKvXA&Gqu;N>|P>>9P0xVj*_*TXfWM=|pproXvt{zYP2gl%@C$eX_
z!vN}I@NTcjhGbB315$pUmzPJoR8CGDa2lfgW%P(IS&{@iBlM+SQWqw8a*iY3nbYR7
zlSbxiPr<|nNlISZ^O{B_ju>t+r@vOOli&n#6H>ij1B;}4m0b>i1u0tvvA}~pM35pN
zxXA|!n|kc9w-?-pspjF`z<`(!jtWoEmSm2+pP0z3ghh@vIQ3KlK3<*Q3r#4b8medd
zx^bKU>lfM`P(4uD2$c7t9OiU?rR3(k0DFmxi=Vx8$(c6|sDPrk8U)>EHj8P1$de{a
z#>d}*8sHQ1g(ex1E)*I=)A>kQ{FdflX9rD~IrA6z5z;~q?`&akLq)}yaoKop&?7%{
z{|`eAh7_1!G{B3&c>#<Z{Ekiz<d(qQpkGh}&gh5$-Prwp%a?g4uit!wJduweggd(U
zEFU;Rd`7M6^ho;Y?pF5W=f4l8Jf630<f-nxwN-jrJsF=r>fhM43%beKU3Hb)r{bC}
zeOgXRw>D`&e{sY43&d4U4!O3dKB=zRDsbPd=0J_fGgg`EC+9^}ur4+>K8nd26~W2L
zoe-;KWo1Wwv!g9$@Q&a*rrPDceEoXj%$YMm#<Q~OQLv#a_l?;KO3sTzZo~ibX1yN>
zrtO|FQ&!?UhASCgWGBDaY`Af=y<3}_MsCP``*u3aE5}JmS=q>d=|61`T)UhffjRwM
zmlq=X<<RZAix<adb|YdN{g7sr7k!lO{Ng}|F!aN6riQ=UX|i^0DY8mIA{{4BI2hye
z8cngM&J=`*cqm3kq1>EXtrD^#A|tw;CwTel)!q#7+5Yy0ToF3I9wKB-bx2A|TAk#S
zwzN`oTUbQI)3mhM<`eW!Mm>RRFHkl%G~{h<kFv70O`vxHnCN$siG{^}`W1M)4JN}!
zjy#;Sb;Sy=^_#eP*o_6UVq#)nj$!X{B6*USk&zK0t(u7&?|zMvaAhMR9ilhosgW7b
zDfi-nk%+6^PRgPnd!Fpdhb(%yf0b<<a}sfeQ+7D)o02lGteOvM9-*@O;U&9ZVP_Qh
zgoISk(1WF9i194=@*9B%Z1Zo9aH*&=bn&`!u4?tgPL=nB1N>L)jUL{)vx`f1-fkUk
z0pTczhbviW<mBYAsGbjAaP5k!dgx`RfD?=N-rMrUvtb^IcKk#2@?^iYTxs6k-sUxM
zGX{=BL`rsks;YYK@UO_(US;+{gHJx~;@k()?umnha04&dy!?#grKqTJs%uh5;niO4
z{*R_d2Gm+aboA%q`3?@P&{}&le8W=CV}8NI?(h!N(G)zDYo##YI(_;ynv*3ecRBN^
zSD`f)yeww>T^3~oDl(g!c?r(#lk$`ZO@f>Lp3CtM9ys&4G#yG@p9;-jckY}wRasZP
zY-PDU&AC#r(Umjijr2@N_(6IbBJJ(#n};>IMwva9LDTKaj_T#$w_^A9TCZgR6(^r!
zgXn6vGgbbQgb?WI57qN_!9DL$mjOmDUTZ%Y8f)UAzZFD#g>o{H^7n;7va$=bwHJ>Y
zD=&YY9BcI)&84Fy0RPU~6=Dk^8q-b0`HZ+nH|x2Yeb0AmR6iJVC?;k?qZlKw$2Dje
z8{bMVN9%>;N~L&jEz!RstXE!e6ZzUdKC3M(UQM<Lb9Rt+OT#er+qZA2lPb<Ul^rw)
zX6ClZ1N-vhI-_l~ulI?~XFG`NG8<_0-7zbZ56oOPgskJ_Dp|&Es&<m`-GBeNfn7Gp
zWJr9O{nV*byLazy-d$<T(6Nt)JwY56^JoXtxgq@LnWsv5OjcFJ3fnU{lXRfW^-e~A
zkJ~qImb{nFOPvINbT;_2RV*E)=!+g}{xmlnG}d44)KfUjWjf_^U8yJSjaDsOScyKO
z>k9@LTkGqK+rO!l@)+PcpS<(4SrROAN<<ftz#*ySb5^o26>~KEme@VY=`(TCq+7@D
zDi~-A{#m!~o?PoUacx(VCGkVOt-S&Rv()MjjZCA#ZNS)7=VuI^21`|TiloX*EOn02
z)~)07f|h#SpCQB1+xz+C=&7%Du?=|rdW{V{!X4+w>b;^H#y8ptQ|c+^HA6tW-}GKS
zdZ>x9ae7XcJ!H*}fw~LbpKh|VJ8}GYhQ$UwJ+Ul>bE!s-W%e*hpWlne?*9nH`YHOp
zsR}$lfNp8HUeu-awhBYXo<p=3ZM%&z>5WZkfonC-L>MHSs?c5S+F&d4$fif5y1w*F
zVZy>;V#vg<ea@5rhQ&|X%y=8pO!+fE$A4sJ?oF3BxD*{-2bq+1Zp4fkKk3{T+Wb(&
zkejr$wyLSA(PQ_TIKg)dy*!bT?n%bXC~!=X-7!2f(#+i4nO5HCm8@{=<S#WfJB`Yr
z9BNa!S$8<tt4*0Q`tA{^Jn?<s`c0mEo7fi`F<|Q^W~8uLIEZA0g6SS9owTIn&D+0V
zI=Jda<Q{#Ml5&-g+;y@6JcGb%f|PSj3*|oUyt=x&3m4vilterOm5k=cCI&($31-7T
zAUZ1Q7YgiWW<5#;>i~-C*lYs2U?T+`iYBL~s)|^ypswy05XEcnNZ#q+&=QKoYTY$c
zGqWFUZ3?$Ex^+9kBlRDZ7MQl^`f*_zESXEK;Knn-?2b-rYglhknKZS+@WD5Fa)Zd_
z?4Be3P<-RC9t~+VMM*RAWP?5*kC85#wH57#c@85@n4Y|pLI7_k8!M|xbTD2%pbbwd
zpL;ZU|3}UP;~6$k#79LH5?!gCceL6>hKJv}dv}BHFgdyD{D4*Y1OXu+CdRsPpHG=_
z+xcrb9qDt-Yq&rShq5SE`|iM|O1xkflG1Ldr1+Svn#|}4`#e;iv6JibV~?1YB15?O
z-3D6S?bNA!e^R83@&5U<&UpFy-h(vd5o@ZetE;LUke+5|3;0ETzMmx-$wE@t32|RN
z{QZT#pZry1UTbYGPu(JC62+tjdLIfQmPr`Yw!DA$;lqqEV~$m>x*2T2-JAN`4|bw<
zrTn{`9M^R(jR--j99-9L8)jo?_abL#)YYqp3r}vMx-l;nTD|`G45&q!#9ZcK`Sgv`
z6z2}EL0mgY?ftCrvuAgbp>D;4>h9g$+`y2+PrQZ61+^*nf6R4DE2Cl)>?qhLY!Y5D
zS=uML9m~sZ+YK8lfwfNZGV`R*@7=r2u{ik6K1k){(?dIk59P&S(N;RW8)m-8?V<?k
zJvC0~1<neHPiSy7H#L<}t>jb`oIA@265fJeGBLl+^w|IpbW!Sw_C80BbTj#~7XER?
z_o*HfHn(3~lmB(?+=QX`7&2aF2F-fiX4LK{@K>K~f8M62V)t(*mTcLwW%q_MgF2Va
zUkf|Vq-waJBz&rtH`ac8?5R_tyi)C@mmNQ{QBFRdyWnNO4Bz;1yVZ|JOx}d>f+H&9
z&tQ5a1J7C)<Tt8EcNS0tibq66L|hCAc$Jwfj8NEcs4<>R|Ka`n=#{sR+gy4Gg*{P}
zEPAYVM4DA2-bfsB*>=XLAgy2alFgP8?Yt#mz3up+zUR(m!@+AGh(0S!QkB(kaB#T4
z;p0Px8tI?Iwqv1|dPs51tf!72$=^;!40!TIq#=#SdcONFipV+QRv+gd&hMe`jY&AS
z`bjr7S6baW<CQD9L;XoaGGo~7=p#!n0u6LB-8eiVLbi%7gk*-S+e`Up*bo7R%y>gB
z9WgCdj~Xm=ZrK;=4B0KkCoPN#E;s||sT4eYx==-WvP7RgeQ1+?th4#e)$XykN5^Ih
z{-~};a!Io<Mkp6>I~2HHxNre^erm^NWCFYqra_7OtsY+HDdjrfJ+_okSMueH&Pxq+
z476fuTkY5}Depm#`Kgyv2;Sy96$VO5(w}xZU=}L%(m~c@15NR(1S5dSXiD7(SNvE<
zcH$Zl)t)T>Op0SvmfB93x3Y(08>nd}{|Lz~TXg+1I-V95)O~*R=Y6$T#Dr3)qh47_
z=>h}Ee-4|duI@-(&}4jfJi%||b;cm+r>QXtHLHIaxWVwACcCsoOFVxUeMMeOX5vA0
z=X~;w`{{RIWMsgts@T8bxsPr8K`b{f)uH&>eTnP<{L_1E{3aWldld9g_{UG2fS1Pt
zcB0vNA-#*gx}waUwPBXT+8ZdhJsZ0r8K1XcLH8MB?Q1nWbtDH4M6I!9_YXKu8AX1>
z29F}rY{9qgTQva4$gSfiPCV;3UcwdiqfUqFPN}rrh@>u;e!b5W*jN6OsoE;3RVZ92
z9o_FPAUhoy{Y?E9e9yXRtrD*DscO7>Z7)CvDvKcTnin&vkzB03QA|QRB<m`wcAo&G
zU^}hyHiz^S6T7B8`P%j&kD)lp(8LD{945F)?XEt3PLnRbF|chMZJ>4Q_HEQtDN6xi
z`ZNwZI&-@vkc;aWjH+BC4&@^lSs%F}Abs*EK*WQF<dh-@ho>GNDzVu0AH}Ow)~UOG
zyUNG=?C{p!MZ-n(0&*~_q%}9}?CfsOaIoDd$U;Vu++bnM$)~c~0Wd)MIdS6aw{Im@
z2rhx8IiFLcmaOww<{J?3wZ1+pv*3zWkLV5aQ~%v6FvuMTPWEXL(jfO9_6wY`fOwug
z<6!=&aN<_#)%=set&LxZ;xxn1WyR_VdW9RyhAAs6KaV$?XKXXLfMm7*?PEue-dr_&
z@yWC@j7mm)z@Tr8YAPnvnYU*{DC?*!X&djoXxFY5803Pgrhp#t-^7CNm}x$LcNXJY
zD;*0J=>boCYp1R@pS|Gneu-hjVp@)e#GO<%9UT%Hs`J0@JwAqJo`BYe-jd%8P?6d~
zhSe0A7I!{k%E-%wDYtuR|AVqPiz<ySuTz>K&3En(0L}4G)wwj^)>g@0Zi$tzQU5*q
z^>n`o`yiDI16BB$AK9=P7=?RsmWNH>b<*^@UJtjxhO3wacmHf?U?AW(`1%E0y7Vb(
zVDqyb$C5Fg+$&nTdhznI--b?`BoHGG#ts0_?uJ*26`J~PB#9}Lo3<@l)N6+4kkF_%
zfn_#_?8Ki_Cx*q#gBWYf6J{Iz1A0(V-5jD5Iae`%NZ9DML5kbQRVXbvvxE&lAe)F+
z<wEU#fYdW(nwb_RIs)Xe`v;Y`>yT~a9E&>+lh%x$I!AfJgxR}mWo85vS$nH@m~Y?S
zfcC}J!KZ3XW4Ud0iavIi^kUq6$+p6+!0^!G{@We;`PS!U2BrQ|lz5=q9r<QcLxapL
zXGbTe>8e_%FZcMfg-@z3>1CCuaaZn_V(xzKr91cTjU7Mot~*4-8lo+a(0}JL3#>IB
zUGAc9g$mtviut-g#gh>e#&rh0H`m?k^f@EJ6+xkRc9m8pYn_R2L7TK)Uz5bU-~|%<
z6uF@#l7nJnRrmS)+sn!~hx={O^oN@^Zq#E4R*9L!q#-H?y8i6mJ22@QJ*a}&YmoLW
zqk}TKAw37Lv2^`uFxt^qLW%(a*RBEaT;hU3{t<iob`5UiCg=2A<!1JKN7y@l-rL)L
z+{pCe0^61wrr&uK_nLvf>CBNb;aO%a{B!*D&4bezWP<^lX`1(yb>9!4$2YcnmYKQk
z{l%kfiyqt`tXnnC05!$pm**~COvFYQgAI@YUX;@H=W52jha;n+xHqCA-^S0|2aF?s
zCVuQD>B_P)tF2ozP5B58R0z?1p6|MOe%_kUd7~vx51V~C-xVh72kB!~UQ8gB$o1>;
z4(q!wZOF*c^Ny&wbo5d5>+s@0p}S`J&((ZzL6boi%x(~J5l~x0+YxF`LqpTf=>-@C
z&rVD5sBD$U`=|m1vFq<+fpk{du3+N&7-y%&A>|il+#;vcw6h;4r9L3!pJl5p@LsXH
z{g}a^KRz9C-!wQ(liZd+<Vcgi+__!$U-~UqU6n31_gog|3J~4Z?d0jg(T(a>HaXxF
zjZ;3APB%q#XFEwkJ@N4N)KSV3u5|nf<DP)ONIIyzl`Kj;E2>5FXyXp6+4f&8z=X?j
zR4AjaHE@t6#vT(KkRyK@ELkFqofp5g+}g8V^HH3QOh^uIFw?92h7f^&?~w^~Lp4TY
zkaxTS`YbLprD(>3{wX^}M6O`=aR$WS{2Bv6cz|eEH9NEI+b726tYFpRFQ1^^b~TuC
zYW`K=!P1N08X8LE=5GQttgb#<vzb`~QftsT_P1{h3JRj2m2h*|wM)`+<{?0gFGu1P
z>BF)E9Y|fkqS|;Q=OQuzI_+<BjwaI50{t~>hN~t?2S1VgrjjtyGxX-o23o>WJGy#o
zc63y?cZ@;0U>~=AFokNKp`RE*h$~^;jl5&O6mzI4U8^|G1n}YG5k~4mW9C-@rGN{B
z+z>W#rc+zP2vvNpEeLAy8xcIJtsEtV>}}Gr$i29hg-0^;)cGMrrih5FVB2)_=GpjX
z@7}d={Ug^NbEaZ-FqhJ(kt0vK0l?*33@82SH)Q(!&1VcL3?2Jc+HBnl^k}J;`|=3Q
zQOK8+?Y%it`hoY~uis>AE0N9awiwVD&n4HmC-K}9s1nyI=53SquD`+U<TvxCAqbC9
zO&a5dnepO9K^16B#rqRt<gU=iGw{(r`*PBVq2{DaRvQ>Q1=>C$&FB1i#I*efqP;K1
zLDzHRrcL+x93#f)etfX^T)@V4>&icWmcMzFG!8SXJ=Cu$)50Z*+Nr~iczV78Q^`+0
zK`#VlVxTF|mCU-6_k=5EB^}YDGgEEXub;1}+*fAe8ek`-Fta7GckZ+_HfoSP=gn>s
zGMKJ)92c=a$3+Z<Ax+>Zad&vJ<i_6#r3Coeva+{m!NKvlh`}W8+2sj_9BR|-AHycO
zR#7|BxAPpRCn_>>w3;9Z+qh97R?$ai^mJ5`4PAbY!4pwp@L<hdU-T*QxO?I@Wq$?&
zp`&fQeSQ;#4>Z(rmkkVap-8z6YILqJA_AK^s%D?Ctz8xG-3~t*5MaTZ<^)<8yRKW>
z`0~^6<@);gQFJNG^%ycEeRlMNurPu2g3MrRkt@7N9-+Z_kk|nr=2^eLs%?B-0!qjA
z*;QBmfVHR2I~EV$P|8y?Q&ZM`rJ0%2-o3|8oOtn0Ydhzcw$0iWK3FSqZkq*aYH{Q`
zcnGv1xuRSjaVEedxT&cL09Vhv1-U7PrU?lN7ySJARsSi*aLs%F{(@hi<E$`ZR3WeD
znS(aBO;3C}=^F+@y?giW>C>;EzsbMCo=Faio40CJA|*C^5Rh4zQDScXkwWwSBojkJ
zZYYUmRlKf$Qf0G6)=Lc;V(}r_T0~zE#VLOuoRZ^Xiz+a+gA&E+6Sw!iA1z^rb7+r{
z|E4PGIv+I4=;n4;q<a$=_wCbX%x7m1|D(MIFc3&6khg64g%pSu|K{^Ahdz%6;7PqS
zcIl-c8#@T_^(UuKWz^z`<P&rfh?^8AI4I{SK}$XOxkyFYjoCc+<wHXzPn$MDMP;;)
z_^@HgNLQ)VWtk<kaiffkj46*5AI~U;9}ps7b%smJj8uS9Ej9q9p4~rE!WE=IC(`J^
zf!_>cu-dYPK`2{rL@X_BWQ^S9UpFyj(0=~@{d>ya6?=ljsixQGU8>2m1ezrp2;*_6
zaK3y2Ib3CGiW)vGJG+^T+$*+Hfk<ub`_kbWHu&A^j1Iu3fMAQ(orCb&{bBs_uL0}x
zpXiHr<oSpnkPdlp;MawKz_u3x+-+p!{c+!sGZE#TtdiseV7T;JFXZ`>6Z|t@yl}Jg
z?fS?!6)Uv+HvQUcdsSy39GLSRFs0_pm+gJTjg?ca8Yu?|4x!I>^`On~TGpMQU3m3D
z7pxzBdAh?X^$ww{$1YTD@R5rb2Q@Ad+Qzs%rB|fz5BNLv*qw0V#8V8Zstly{ah_9f
zxbv7AKziUn-9jh)WYEw896TKwMQ2&H;Hzmiua5?nKme=-o<l4w3Q5foB(JRVqDldX
z-cQ+(ru1wRN^kBt!imkbh_Eo26u$Ca@0RktbV-xOJ*yKN4^i-qU2<Wx{1V&7m6z5V
z*0wiSVgMRi*dABYHP%$<dy76E!B(9AMZmizMTCd5uT4QD?GNtM5Dz?HEi@rulOyqT
zvIJC!l$u<pgj*W1__T+I&9!0l-}BiRjQ9Mx2IkN^l+$xo8nqn<Cg`ln9=lsm(w`8M
zoo8M_oCOi>NNdXaG-nUYFp|yXipe9#2<D};yug1y%e`>_LI6b5KETi4MxxhM2w*lZ
z1j^zzC5NXR@BDG1-IWv-?f2}d?4_00cup_MlnEKIm5*N1oFIlc`^}F~OAugQ!gWD%
zv2uhkJ-12REsZ{kI%*xdDo0k(I`;a>6Rnk5p&Z;ZXU=q8<**z?o?A)i6MZV}LSJxt
zeEoc{H6MBKd^hd@9D%*@HHm{~x<s_Wzrga0(+=+30{l-;x*dI5PW6p__Z;BZT3VS(
z0_TkwIdb0hKw8)nIr`ORFM?s%cD$Oq2oeJOq`}eqpIJj4&%3(0y}g}D8;-v_H1e*y
zR#7VRkpiopwE(;b)^2BA^b&aaXpaKy(k)L_FPr3GzG+h!mijzT$Y*zEAcDNe%~-Z_
z+aAL(@-S9f80mmY$S!d_nR+`oHY54ILr$S0|3g~NDP3k@6{`-3Pu)d`VvT9MkhC|R
z-#&lg!Ve!l2;S8clq)6=l#+stevzANoU7lng4aX6$*nH8>@v6(=1dKDHsH7&z}Q3`
zou11^aM3fU!I5?V=hjcq?vlHwEu;7a^&K=w`N81ml0uS+K4M}g-P$T`&rUzrZ54er
zQBhw7+l7S{EHv{iz{zHF!K|D%cK)J8&+++VkmsZ3;{F2%PQ(7*U+(E5LlcvIPENm>
zKf~IY?}d-!Fq0u*E4=XK*>mTf#BGIP0*WGS|5Q_>{NU@_dGk`J|Dfz&4pmz2Ew}^N
z*)g4|Z*+T575j`n9;<EJI?K{}k2n_5Hrn7jU8&!zvTX>g=E*x_O`VCy;L^@NT3p)<
zu%RFLRwJoSXVIc|T($raCqjcJe3N6x%^D~(o7I*aDecDaRpQqe1A}bV2?sAh3=bml
z?%f@74-k2hH~h*_nR9;Gz5ojqH1OKCbh!>z&kYG|CFmc(^jYC8*4DyCkBM2$R2c@d
zUA|1}Ury`Z!Gj`~9goi8n!}KO7rY$Nv4@xppmT3ZJT5ybwj|hKE>vU+p6?%zQ&o*6
zc@`SX)a|YKIr=rO1_z(>^z`)d+Q9>ZzQLm@nl8I4^uJ_Vf0YvaD=AG#T~y!3#l;Ey
zI1uqolU}`h(;wK~zHjc;t0UVl=yC504HE#hT&?x!5!PQYL$*K}zbKYB#fBogaly=h
zdyZ*at~*_omM>c5O5>cMW`Pkxkqz`+T~h-laqaqb?F9=0X1f$&4OQoO$5f>V|9)aF
z%@u;BG4~H$Bo&pFE<ain8LUm5$85pt5~dWX64$1_6oViU6%~eev4u!ga&XgRMYOb@
z2=XfWbEKu|v4LqU<-_vu{BZ5ig19k?idX?0U_Mj;3^<EFc*qbYpr3YkXDX^6PK+$R
zpI;j=J&7b56t44UhViF3m8;ov*kpdIE37FkRb>YQGms_yW$wANnH1$^o3GoUD7tBQ
zq!rjKU)-SFBdBH2mJmWY$%J{Ryf?)AYO@}D0mo}YOU=42ZfvU1UA5}N`)HJKBI23{
zI^y|_?y1xzB%op4+Pf&|LK2e3<ZszQLkXh^mMa@|6*=oo#|{C@Hc`4^%$z?1%nwfR
z=N%>`&7AQIu()UXAteDo$e%k)K|!b0o2xI>LP#P>!=E3(#nwvv6Du7aBjzl4Phea%
z*ZgQ8Bhs9Y{YlbA+ByUg3}K$@Saa_bHMKpp`RJE2Ee%(!C@U{FSQmFB<{58|TdVNq
zTCK~P<ZWz3t}dy4Z%B>WY;N<%+D5LOC8(PZW_h@}J}0}wXX+aPYxU~lXV0ENE&{ck
z_wiw8Fr->={vt+^k;VTYyprr1ncK9DWUM1IY_=3X{KW_UOKDRm@$RJD(wLw|hNg8~
zjA7l<d4?~9uEAL|ry;K!2*da14||jJ!RGU&XEzg0lx?bJIHqkn0+wPp*PI-q>v4eP
zG=Z6Dv9cneFJ7-k@GVlJ+{6vZbd)vnJwhGW#lqCNEqQ!L&~AWnzM~;deS68l;NW#w
z7Hu*4=b<fW3ogZeXpEe~yNWzA=`A3>A|emBgR8PO1*FsLNuqay(TmJX=J@6l+}OK+
z;<-p%um61FEU;5*pLbkR2yYl{@EENJ#mMSEUn<g-{5qz@L*vu1bm9aA!3_Lt??#BN
zzD_LEJ#Yej#vO2rDO1jO^aTNc$we<q!hoCs;D~JmWCVzAmzcq)i|hBFKf4SvgxNlP
z2@kUyK?N=G3$u0k9{)8R|7k1WUvmE3JNyG*a~%HemOuZZ50UuiAB6ugglzkN|K)#v
zguCPK*Z;*U_Ue7Kh_e*D2D%QrL}{LP_z_4-X1A7I<syiPXF9ovnouPZA>%B1x0!K{
z!Y9J>vti%u9>@Q8&c8tXgJvJlv|o;&;VEU{0{v{-tTI|hiq5I}docg3<Nw9W`2Wu#
z5?+>wb1b(~`PZ-G+8@xXOf%a1sH{9DK4wxhsYn|rLiVB?jZ9POa?p^n^npv~tX3P0
zAjss3xqZzdB@QaloWGk8<dp@az7hE2Yi_08PI4X~8n?^00PG3nxyJiT+`dDWLy5gL
z;F4DblXxHA@1T;9$g70}1(gs9y|O@^;#T>`ye0tBX`}TkEI3%0!1+F+uk*ho;`o9~
zq6tGf=Hp{*WMqv^$lA4U1iy}7KkYVcdcOQK*DEr#+0Jda+!^cZuL0UYNLW%b5o(Hz
zCB&tBJ8S@$&_DnF%Z2sVr4e}J0`6V(Cu(Tyrw=nc(w!`voVZGn--iS;K&eLG7b{f3
zd4t3XVm*rq8iN%{1Q;*_(n>n^K6<!<lB%j_!roAD#lj5dEet3asA9^eogX6j6=h~d
zOGo^pHd<NvMRD<LbbY|HwA-mj50aC+yD2;U@nfOUJKVX~dWvguq-1L>Gut7Ux$KOL
zdmMa!*m1D)h%eikk#noFgzuyJ&G{<Yw=W6iq?r}SW49rkQ56Q{gq|3dz;cpt_}dL2
z`yfkqPv!3X+1B>Va<zCWECeX#S%-1Dx-Y0!n<~RXL-Qf)csd5;=Ic@|(f<D(3mNR3
zZ>@cDyz4!T3tK_xk?M;3+Jb;3BnUlvREPsj*DQIKl_mH}f!m#2ZGII%20xBstiXhy
zQPR}V*h?LpdN2hMa}id(WoG>Mn&RZyzn*_0+q|S^Cur=~zyC4tIz<j$_^dgJw3_M@
zBoK1bwrWB<Vno^NRj)3eZX@z^nh0-)DZ&NimorI+aGlvLSTs<`-=A7qQ}Yw#d6EQa
zHl4nj1vR~!go!M)4{Obv#}0CpCEH_K+ZSNkXL6^a!ovwJa^+)3s~+fRLy@{YiaBEM
z*-i#7pFe$qxc|(j_SsY;5;#Wn;3R6O=8l)b_=jw%{G@%;;_eg6@>)JuUuR1m=&WSR
z#90in2Tyg-XI78-hn)5Va=s(KXNN{dQz^ahE+FG<3yipaU0PQ5c^uQ`Y4v1t+~4p}
z{3Nt9YJ&BN>2C8!dOjlyO3o)ShOqCcnDO>)6Ka_iHC)gw_)&@~-eierPS<~1=UKpj
z4Lv=*f9leC$1e;WQFQ^V++Un{;Mer51jMK|p9IHesh`n11R^xgMBiEsiZ3_JVC<oV
z;RvNV5>2YW>?hoB!d+&r_Xi^%UcriK{eW_K9Mzeo<W2@W%PDl86P<dHshJoV*;iEB
zKE!VH93?_@9N5P*OKUWEYRv&9E_|Qh$X;^aaI{1KqmHjPF=6P5(L0WHdmeRUP-WiR
zw;Ydioe@fp-o9N?ec4m2XSX2rmz5$76$<0VCH-`v+s-y|y1t#@`H9Blje)ku&_V_x
ztbEbmEe+h($TBHtyc6C<oc;Un2dgqWigU=Ur-Ngck+0LHOxTgX_Q=aig7J($dP{i=
z4b5nnkL?M`dj8xqtOLIRVf3Hn1W%pO4CR@RDyvEn1s;93-Z6Kw3IV-;wp0m|7L-DW
zg8&V}_&t6K$L7Spea|qRIzAxz&vSBGVM&&+>_dbaeZT1;(1!IOqpR=U8CHM3g5L6O
zM7OvN<fhaFE$6%UmJW@0(Q&5@e<Ku4N->r}gN0GZ`@U~Tt}h*teN3403deCa$+EMU
zgR0Yc`|4FI4$ThVe;e)#+*XP1le)KGAWc>;Cm&{$7{_btFQoQi{&5rNL5hi@f`Sa@
z7BFTe$APMtV6yDGEpBwfCr;GAZY0^ls?s@d^>YSARu!CMg{8%kC0n2xTFz6-^61Lu
zF<DuZtwQ5ZVBm!I+Y|##>jb!^J}G1!EzHf??)Rub5r8D`ZB9o-@5OntrQG{6qu+4c
zzlM%F6%`yj*`eC7eX%e-3Olh$)!j2ZQObZD^RaXF+tI%fo}dzT_?CSBOyM?vnSgy?
z82D(Qke42uAT-va>iD8a9rA8n_+TFts@Ipi6<zH`!~(Cm>={Z6JX!$^EG!duZ=u1f
z_IV#nCz1T>sJza~@=!M}q-z)+=KuvPk9Ke4o5KN*KYRf1G`(CuJdqI{*m2y>rW8PK
zgwE(L`wN5DK*pK=1>$Zk?XQo!8ejlv^a3Rf@Q)?0I%PkB=yg(u7+rgrwUldy5QJJ`
zw9B`uY(aHL&-d^RR43oQe-E6oXT0~VHIKVSm-g;*@G%}1K|u#EmsM9QeUF?$qK#f>
zg?7>$jlaU<f+~hhS=?Onm^uzYG8_3Dfs)ql9O{7FCSgwGno(QP$fVyqLZQjCFTbXD
zFri<w1~>Zh{G>aW+lO5?-yuu^0i5D!$CxGsX@gEL^`CTXcE3E8e;^7{<xN;38AN%X
znvp~EGXjHRX0|vB@Z2Ap)MGQ>HiPx1jBmiK@Y>WVSWhrrq_wW_;u|f<eL;cmUZ<pX
zg?EnOL9K=~{RhqrvMVzTXRSXns6I=_R*ctT{Giy1LXy6Kh1&vv^7&kgoE}uU37ZF5
z@N=6%gDw)GmR0W>NkD|gV!G*JozWnI{sia!qYXq#ZedieogtvSird+P0JL{*-CA)?
zT>liR^xKJHoWw{NXk04XpJyi}C2=$3SK7!n@2%2F*~xui@Zd7dyleg`;hrK;-gve|
zWzLWjj)M<=0iwll1_cIK24o@XPC<t(J-CBQn#GsKD#XtZ4cvR89VGB+*~!{<_uL#7
zGOw2sn*kwIstCa$#@1L_0S5aLn|lpgckyaT-({d@qN1lxo@{HWGFZFz3cw`eSmDP}
znxZ;Xva_Zv1IhUXjDxm@1_{6zhF_-^6x`=erS!&gueX>OiPr&A8CZNi*|B2_fG+?a
z4V_!4;>g9abapv77~RarGr}1UME4m}7Z@5D3D#YsM~^1WAu<9F@~O17wT+F9i8OT>
z$MdTWUu*+yVy}|GUA%C?cFmgCu!iJm->Y+UOR9kqr%#{G@8Ew~QSw80Csv3fGDOo+
z5R0$hGTBJW?C$18fnIv@yu%4IR9c~0mOoZ^W#xp)A~S~+fm6ur45apM6VyGM;nm&u
zIZ?6lYfj6k9<~S)EwiUQUb}38m7U!n35k@}2s_{%(h*qB@z**4Oc}c2@gp1cBaxW+
z;}smvr*|(Z!vM3>smz}Bm=1LE={8Cx508F{Ve&kUU<C`)2d%hZ<nsCRQ#txQ`p6mH
zb}({ZF<#5TfC2+uJ^K^eM0#yVbX8PTkYrs72sq~Mw-lxeE8P&FMHT>LlJr(EC@enk
zGm72vNP_oHT;gx|T)u3%eEHhVn>SlqkMU2Nmm7ADKZy4V@m;|Ayu7T|tXZ>V%h&2^
z28>Vh7hR;MhwX)%yE{ew1O<hOQ>Q{M-3OK8EqFUVmJ*u(+y|@!$>SwUD&CHa+)Nx|
zbz`iIehvd<z@iLPi#r2!i7VyD58=fE=hA*5DBH_^3JwpfsH>YBwLdCiR0jfz2$#`4
z+S1rqnvg3Y(Zrv@bM)}xu4KlHIboNO&1wuCfBx|P?gL%t^W>LQ*Zr%S%o|605D}ql
z|J(dHw;30TfeXphK1WA=n{ECr)j2O1`8I2~HYSg_P7!P?6#A4j>%VAT^vJjDw%_Vj
z+O^(^3Ff<2v27$*sk!2OD>DPbko~<MoEQrCf(qoe8y1C+3BZo0=yKFFGKAsJ+CwoJ
zvcnQ((!-6!b|H5}Ng5~ULBFlzKlJY0x}Jnw*a5*GB!zw;j`Kou-<iqbA7_}WNF&hh
zBn-fceL8gna2POu;<RbL5l8u(ulG1uTUTHVwD>`9vgLM{&R=7<cd9V0bJg0N3G4r>
z1z2^_Gz4rhiuqz-E;`E>+oP4x9s+FEwkOhW-vOfpq<I%(eVm&k(CZ4WE$BfsKfHbO
zrm?AsF(KoYX$UNIXA0YAoU|KD4JF~KrQc6x;_wk84kaJwmGY<W#l@*D>%;XruGfaf
zfg!tU+l@%xChXb79a0;!8^gBUnaD|Fcgo4h(FO7PHtDxVz!r*x+OC{V*8vx`=g)uk
z%i*8u>>K^o1rska-ftzw{E9lrVIXl~d_{h-;j@2&1hH03Xj%JbTl)h!r1i@`2qs{9
zHR$QjnC$`StrerpS!bTQ7pbR(*G@pzmR;2s=2>(_oSa`7Le3a#vb<x`9ji3aMH(#R
zr<hJMnb6ad<li$pdm~QxWZVn4<k5EWORD47ul8w=vFtba)+x8(?X(W!O{zmdbHq^f
z?FSvx+5RA5GA>IP9?I06M>`kOFkBpGXX4<XX41=Rm5Qwi4Mre<{kJV3S;0QgZp6<@
zB8g_U`P6MiD2dDGWh|HeCjJlTvu}cQUPpLKH{$8ebhD;cF>+9y2=;ylJ;1A%WQJ(4
zH=~`}r69S1-jP_-!TtNM-L@@jyfFYWIRWz3<l}56vxHrWovZTMtU~Toea2{fnZhHG
z*cFRQNmkr_4<o=IKGZ6#Ib!k6iyg}@PF<pCB-OV$;;>NBW;Cu$Ete_1!Zp=^S8}md
z^FWf7@xO0>6RMP<g+Qrut<Qbjv*Y=*XC*(*)cVNJUs3;FmTolKnZ{iE#EaD})IC+Z
zJ=T;a%Vi_fm|(na#R`eoe}0TRz$K8FIKyRmZBNOcy&+{yZM~ho)%tsS&Q&Sp@_F~Q
zgrm_mZEPypNmhQs*;Te98r30ywI7(CaWuz^{?Y35I-}p_=2DFHGg+RxRgz>3YU*>c
zCk<+{c@yTVNGCI?i^{#L9RTMel8&%M6W9F4c>3LhTNafMh$}4(@sOYY4}0$!$aVj|
z59?}|qJb!@(lA3Pvs5y(M`T1OlE}!eR8&f2@9aH7HYr(=>^&lTZ<+V;cKyD;`+q<D
z-w*Bw|KIJwHHwe-c)ecdc^>C+949gpd*uB{b=SjTmgHLV0ahwwRY;~vV#2UoYtt=<
z>s~yTbEN=<>w>^6NcBvMKMy=ZD;G4)OE>$;KzP!cL7sbt@(nZ<ufF#p1aYZ=-J9~s
zh~+9<Y2VPsD<H+*_d1a*Fdt8eic(fmDk&$YptwfCZlhTefr`^J2w1?Jh3<z#ydW0z
zGVjrP8EciO(EaS!Q5oFKvqEvB4ei}G^{yaa7F7>|u86=03iwaZAqml4H-{r!jq(R9
zh^O6VSW$piOGINe5^<DZ?fv@ZdmkWj3CA^S_~kQJXKxB@Mlt%;sPxU5?chqhOtWM(
zeHc~RcQrDQ9C{X{8;2eryLQ|oDvC+A%DDVJ<Crok8l=YG<{v2Q5`qnm-*7Q!$d4N!
zChk@)+LWm3qsgW$`P(g!6GmX5IbZ8F8?$*#Xr#XO^K5~1foV?lMM?wm`weK>K#PN$
zC1Gj}ao3kG;LFS-kH_)n`R5`KHQ)x+dT%}n!A70yyeEhkXrmyxN(Kko&+jqla74o8
zIf$Mj!0XV?u}Px+y~8jwBLnjh^U({5oWG9tK7f2+#h~R}J$``aFD=z~>+Hx`kAMJm
zoDZ->ZUhe+!>Lhmq5y{TD?a2f1z5x3gg>j_x!*t`N(SV;2+Q#q*BLlm!Q-$$(6mrq
zRh4KLp>U0jzJ$IMpo<C&t8`@oqCGd->`wEv3seG}lAzu~;)Zq_rodqjc-~{d*)=pW
zVhgiXEADTQ<l&|yh1*+NrY5g2Vk+8^v&e3IQer^`sWYsQmXbnQ*A7hNPck}^J(%zE
zoXlKXMTHlj0veq-6!+X>5*Bs@`35PmpbcE^3Ls)~aA<vl9^GVbFW%bmhmTPCpwf4m
z8#<8o`=t?F@m~5RfBeV^6b9W`YtX&?7`eC<ssf_Fefjd_h^~sFVtQJd$7(wFo{zSP
zzTVyi_J2JK*wZ8uPX_Z}soQ(fK*a~qHT4k0u^JxYUO_=)qoYSDDN*7OuojX?Xgraw
z0gswKPafTlTf&hLCn8`ua?*Z^mL2~OrYz3Y!7Bd<ois~r1A|jMU&HX@iE9J6S(NrM
z*wi4BVU?t&o=d*TVC2#3`!V+{A7AsgZ_fN4ZLO_HW23RhckI{!9u00DWD%&ikp*Mx
zBB@x7LcRngjS0A7sE3edXQicKsr^OI5ZOf>cOB#}=)+#UzeQC`i#>-gR=a4O=*R)^
zfanjWDb8~gT|2gITUuULQB{3&H5{wyGbZMa9NvaB&S7@oEh=wJVGj({fEg4n_4x7Q
z>uZbJ6<6P*EX1yMSQxvBn(Y<SSZAJZ*JwUA2s-MiX=xZKDUW;{??e{Evs&%IEkOAO
zyDaX!_sAK6T?nRY?mbLTU*lLDofznX#XgC(h$Z(S&z6`B{qxH=CIC|)>!SuD7gj4n
zB9Bg1wFuu`EE9m$aMm|m2uFsxIrDxWid8x=y%B>ON$n3n3?Po~iiNp3($M{z{YZ>a
zCPo<I3z804*Q1Zvn)1x`_=yw4F415+AOSf1`1_x)>%Z}|k(uWMm3n=6ublP25}8oV
zl?xSt704K=C@Hx<q3iQH_Y$^%B%7r9ZzOsi=dELA=jREli<3xgY0?zk6xzy&4FAFf
zlos_$F9B$6OW!pn#(0_Dy`!<wJE6WGj~i7T^1uqM$P!w1$~WU<V<?{y5@m0}b?b(c
z&g}H`?ZJ;l=Dx==QMPdLcObUa3h?_Q7b8p9lvmvYXL`QTYNF$&vhw$G#ap)oQO?Jo
z(Jc~Po@{Dw4;;<gdo}+oRUN5Ii|_>S9T4R|#$ba+M@7jSS2PGuAoVjRi?W8ffz{Iz
zahIvzKCthrB$8g0@&w1vaCDCyyT+I2a7+aAw80@I*JiD+5%j_<cw}OW?)=!Q=YsaZ
zxgFOn=DJB~Ijf8e4P%zHdefjVUl4;Dp9j~IeBBl8Wuz0AMt=MS#|T5<jM!*tgYlmg
z$FH@OnRH}(q=jQqhudAI%xn~XCp`BULZALKw}E&l?FQDO!>TN~l1&t&ncg%>!~s3V
zG6^X}tg_c*Y6bk&zfIx#4GxH@i`Vxkw71uxk>Sfs@Ie^{;*ygKZDymvmbi3@#O|yE
z2lf+)!Wgg^I4z>4sVPGyU4vvT`y<-Z09~<v;mCM^8$KcmNuQzEaLtlX!UABdZ)z%h
z`1_v|vRFzdKXLpX8<Xznz}AZ!RYC#%iuUQDg9l?5&mgrcv67xk;8i_(hY;MBEluLc
z`B78jn|pjZ{4e^eVt#SB;<?N_M)cQ<(!C{s*|$>Ja$aA#@RW}XC;ph27uJObM<Vtk
zbFquKBOcCJ1vywS#K$#iU%qlh4-`Tiwmlp<v9YeTDsxEdJ+Hmaa8F1~ypxsOe{ZYC
zp<lxLmC&@5dso9w2MrMz3OSMyWLjV7{yx+6g6cF+z^w#{THJkzBJUW=NebRL&d&a>
z@)GIo^5;gJ*emSwT0xJyqc<04&V%T);A<)=()P|CjW9h>6J>$aR_QYeuD8KYg+Hwz
zT4Yz^Nk=cIfJ~F7=TZ5#>_C~DukCfF0oO8Y{trU!G;99XqaPnjdlW8ya-=<L68RfO
zBUqxpi_XJAz2hjA#(wKOR;TmFbX(vmfUxFlM5(cQO3%C>y|xU?0XxKaC#MZWrJ4Yq
zH)>aOSF=u))3|?7`*de_UMw_dpB`bzu9=A~lF_kM9Ar^kXeJ!Y`bm)}XghaCARW!E
zUb|O2%>s06kG=hYPy8`!fpVwPuk~eDwVcP}4tz$-C}2a0-7BJgZLiB}2VaF?tkE&1
zUnxnzUT$M83ceG~d=A1^4~OI{U&g4CzLP_zeQ^9u`<`k9_{MaN=u@m-1gopX_Zp`$
z2fSyUvu$tQq;5R*6S}k`=K*?hu(J=Olp;RHs>X<{%^echswYmKd|*^kN(!`^*!v~y
z9~y#0F|Pt7l|S(cVq!ZQFTr8;MeyaE=?e>Yc;J|D=+kjdI#mUQuM<O;+;#T2mr+ns
z2Kf7@aZ4GHUO<+C_z8L?ipZ*;5K}KNjQ?Q&aK!yXYG!6<fn8Q;Xg5mC9Cbb5Dz#E9
z?Cb{lIbKAV028$)=pu%~A)?Cb?$w4i1x%>&$BB2ug2pt;mC@-AbqXWFE!eK339j(w
z@T;;{5_;OO=s24}b?5oxW3dOdo!Fmn`1$xy>WP`~FY~-Spme+~=h10P?y@NYfM{Na
zYme@(hzEc(J?+t#I$##Rhd(a;Hgn-ORQgMNA0r~tF`B+&5Ih9z`;aO;2*CmdLKiZB
z3GBD9NbbE+U?~TPm=T;!7n9v+@}S6()CvL}CjGRATr%3@=`0MX^;>E|0Hx*-!T-xz
z-lk+16b$O*mPeC9Ev*+(2i*`Tq_KZ-+V0-9ONXonn!?lhDb^^*;l^JG#0M+`6gZkW
z<}ZYzL&Q(ot}bz|)xg3;x2F#k1CrlhIKrVPg%pU4P7sE<WsV+&5X&uGjj}8QnAKWM
z@A@Hnx+CJy&Zdu4qO5X?iXYYtv~-^Gz%aul7}yoF<#$N^5u40t`;oJVIIRaQKZ56e
zWx#gt3$we;LR>gVEw?~!G{ch=mBK_l5tc0fhfB#<zU;<`mlrx@3Ol!_XJ+cl@-WE5
zfN(_UGGK%??~~VnK#ho7VIf4TD@qnf3tYUIRu<^{tEnj(Qk0@!d-m;%%B=$66_CJD
zBz}^Mlk;QlmgSWdXtZaVB`=Q!?p(j<8`+k6duyL)4$8@k$(}g4^!Zks8yZ0MT*jt9
zbt7n~x0g$=l+?+|sh5RcVpjHPJ7%APx)X$D^Pcev@~~;}&(QPvl+O8p4y`Hrit>{d
zNNteRAU8OQlon494FFJH=P~1k^AiG{37Q=`^M6h)Dr2FU!oB#GX&XQ*D=?NvYEXmJ
zZ77ki3@-BAn(s+?f!<pb%11H#2&P(qIPnyuJkR%085@rR5u3g<wXoo04>lrZ&2Tkk
zG|7>1W675^8*|C`?jQW!+bg>7=;EJ>Lm@#En0ov2<t-_&G6?qO-ylD~@}*_4#qnl5
zYv#^&y@Fx9RDph!d+c$fmq+5S{6c+DqV1(LMji>N^Avl~eU$%h#TLLOLE2$7zBs26
zei$aQh_l=W!OoIB7W9JVj$Awdg=P>vLOx+`q7`px0d2wBE`pdirZb8xjTql7rhDtw
zu|B$jXhf<xm<|I>0iAeib~X>s9qSfq)}M{W25991WS;W^PGrR!qP^GbfJ2YCKE)Y>
zw?;8nOHLnpimcn!-5!P2+~YJr``GgRU3<1pXH;x_`u*clupY{wDfznLA>*^=9memD
zBMo)SwSnXHL1CRURG_tl{CqO^D&%mE%@KZ!*gscrQBHF{xe+HuY9n?wD)~4RUDqrJ
z6K4=U2ii=)21XoVRQ){70_S=|^*6M-{=V^i?I8I(=k}p+P}1FK=Ybp|d#q63KRrPB
zl8#X>qW5y*=OD|}KygC@64Zz}30}<Vapij~`hX62aI9FS7k(6QT3SL`_IZ}z<Az1Z
zM}fcB*OVTF$1YJFtpfxkq)auf(kf%2-OUdhol_`#&8ED2>0?whGy;tnYYO*slF|wK
zi4a#6msf$*yVx>Oe@^o}rY%)`X^}kpRwkPEX9QZR0PH+#R%JO9+-ZbLD=!!|FjOMZ
z20)}>e63(>Ojei&1ao5>PQ?6x%rlN-pRXyjvq=)zswD`cA?C=8lRds>lXs_p%(Yk&
z!`fSa19xO(6yLB9_G5AdPtO3OA3LS)=5Q^Man0((y9>e7$mpX~WY2;FSAfF+m>Ev#
zMZg3YO?!mvPC05u^cLXNf{wIg(0fK+-X^q9LPSYjrZcn|*u$y96aO)`us04_CR~?L
z8lYal*|Y{U<<)Ej(GYesoJaE;S!}(YE4p?}R;iw3X*4{YuT$B>_(EA(SUyB-TUKAK
zKh;Ee4G=7~Wsa=0bP&u1Hns_Tl{$fnAzYZneD)HDi%9b6)mShi&jox?<Rb$=#ET{$
zwgllf2F3y*mf@G<BYPeC(5QFdHlR^AQJjI-Wu7vns*FiIzuLhlugkAG%l;DWinPA<
zF~5Kzr$sQ=g;$h=K#OXgq8dA=D=&{K#{GGN+yg;XAsp=hr9#mZg==0}nafg<50t_~
zBFLFhM}_$NOGLpC{%X#lfm9BA?QhCth3bCV6om>3E<3{5cPO0R_>`I1UlW#+(w%Wn
zQNr38b<dt8U6avYAHSg!d9r?cX-P>-Yb#!z11cASL`fKs>}MdA?D&nd5#Bdi=-|Sk
z5QXNwduP{f`x&KvbDH+CZ5X0Ha{cY=If>}h$htVAbHb2&s?3}&*r_6;(>=V0`uNp&
zI8pyH6g#xp74?pbI-M5C+pMSp{x8HHEt8_aIWD=KuRSoi3w0fi9cp(Yx1Tf(oZ%;U
z6AziKUj1A4ONkO=A5b#RRM8TiUg&fpU;5CQSKZz1gv)IRE!p2A`1YnN=Yi#|q-mEz
z_%;;wChoq~pJmd{%gX9z)I5upbS^+Cn1fk~U%pJ(Y7}G<e(FQP-eCR8?xT)7T;f4~
za`u{IYb+njahuSRp^cdtnMQ^B!#Sjl`ijxg@X9DSNooyX6(lz*Dqn077@=pOK)prV
zk~7sU_$##iuPGft0zq&F=126k_4O$ZAO5wGH(YNtyV~XU_HC5bOFS;&JHnQ;&^X9`
zpKr0=-(Wa`OsPxIFCYN)U_Y8`S5F=UcMf(<vrxEDsEOXc&uaHo5imySIhw-@?H4?H
z9N(1$?$njFS+3ri_7aieR<HrM#4WE-{+9SSRDOM=yaq^lczliUT7S@S!78u>2gEN!
zN_e>x4Mk9Z#K=<6%(n9Ne+eFu`)`V7*1qSzyLT<^=^Gs7u)N^2tA2P9EI^dhaa%fJ
z|8O#82-N`cDNHW7fWJVv+I?2?tiUROJ?vPA>E0Lb*=9t(N(dRv4f2)BN6?g`(NodD
z93S01nHEDMR~cr2M@9>z&4DEQ$&)AWu>xIT*UbWW33MfRe66P?oKeOeA9D~s#D41=
zMVdz0uTXo$5?7a(UE~VP%Xz&4C+;w?0f7R!S^W601-fk!8xr}~PcK~Tzj>A8xmuVS
zD?3CayfdEaIx;=XPQa4f9NLip5UB9!;9dLoNB5e+J|Ub;paq>DEYCCPxyQE1*v7<j
zP^^&#VS>l?-E)~E27)&{e#mKpf&&1Bu)mZLTW99rD1wLs#rW0Z^5}nn-rGMmCWuG;
zRF!-^aHs7f_2fVo91{rBsLf?`q=Z6PGQX?{jc~8UN*MUOx5rTARhkEMlu$}S=D+-)
z8yx~C7j6DtW$)9%Qn!@k<vk!bov)L~w_*645Y;V6ZYqTq+-O7f)@J1zkS;ivpb{cP
zFs;+sK!}hU#_t;j6D9T<+%Ya-v`-y}#Ms{dI{lXu{y8IAzI028?(cGE1LdJZAF#EF
zy^RL$++2KoJmgDw00c>VvP<av`HW~|1;_+2>H%@=Y4HV*j68|B>u*Dznl(gz?&(vg
z-)f#7sxdUVN`ZlEPUqV0NDQ2&_{83vS}f@D<nbZaKY0%=(|((JFy7J8iyLXE=W>u{
z+S6C@bm?4G{aL?k*8mmIgi9e@QIu+jvfDoY(s-h<yf|5NL;A*zXkl*0i-0u|7sqQ3
zl9PX7H$-4pdI_?6@}5IZ3kuq%_FIO8zPma!MTV0xN{(5pj1qg(49GC*gIoEJIZqsf
zugS3^N3v0RBk{p|IuN+`EZSr$!22;_f$Ddsfz1QDJGtG5)+XiqY}y$c4Q*$Cp|_0C
zTB`^S^AKCo(gBWYL6=KXQg1GFJni5kQ)oALq~s&Jr7G0>kOfo57_=&fsf3qPa&Z83
zb#;v;&_%pSC_%?5juddzU>x_#dxw3hLQ6&;9gaN-f8L=dKkob|@bryvF_(*j7|~t;
z9Sl068~6!iRY;cMj<aR$E72uDRzr&>7@Fcds&A3W*Tpa+0}mv)EuT%$#s@A0>jjz4
z=g`~iwP{t8le;9NdHt$s-FJlNPyibj@%E=o<?AOGTk3n5M+=Mr^8(kgA8X-c?nnFp
z+D-9^fPheyte6;r0Fl<}<*Ijtfe|J|FX{T|J*8Du3FnnP?*z1B3`9;^ncwN?X=`s!
zp9Lu%=-oj(<^y>@YWLp5bV+}IYCHCPLVX%YC(f37D%GXlhhhJhtI~gVmlfG9-dG}S
zo7?|$ELe>3bJ2VL-}6*qK<R)|d;Ty6Jb=a_K{rX!QdYi3-z-r)p~GV_t^9}U<OlT}
zn{%9;tpFI2;^Of<gfak`p0MqltI?n`)LKZGpf82AQhYosVHff<qah*mrJruyi7qf2
zE**gSuU~iBGXcuY-W(T?VMp1yqU$m^wxPc@XMIUVe<LWz^$n7ysVZ}{+6?o@l$xrY
z;q?S)VR$wRnWbuv_36)If9^5+`TAa8jr-v+%V2p<G3tDN(r};)c0UwisT$Oe_6giI
zFeokLf26qt<?M659f*+OqEpCg1TVHy6{h@-->J<8w&Y<70DYmCL0D8}Soj{ezW!Ed
z|8~vscLpd3K_?ZE5^Dh22=LiFrm4<7A*4_<Gc@OPo-|zU{i7%^e|g`H+q$}#;Xo0x
zC1kspGU)ApLnnsTn5eJ6&E+P>tYg?^V#M!;V+(xDY;4t)l?s(?eP~4DY?N~1khI^B
zxMN@-=Kfm}VhVDnbBGuvwH-!t56~@ht5t}=!kc5yRWSQbX?zVkNWf$ID$u7qv=liX
ze8Sa0rGbL{$f$w7zJzTD_qnfzpdgGnenNuA^zO-xAd1N7KC`iuQ}R&4z7svDadLX6
zoY%WG_Y=JjUN14#{`&RM+&tXz23Dg+y20D2MsP-APxGTNv%0Fb<gKctN2QHy>v)@L
zY|`>xqI#2fbi_kkNmG-Sg2EC$MKHsVuwY<eF~-)AmzU=`|3(V!a&)Xth5|`L!}J#n
zdv$L-{csp@6LxvCS)P6TMrgI8!164WgMy`GAS?65nt+l@{WUBVpm+fPpGtaI&{KY5
zpAN?)Eno<u$G%=Tb?Rj>0M@W_*4b58o-ri&W~Uyo<FHiLdRsm=KF&m2I-DKQ(Nz7v
z9<Z-zwBG)}=8SXEvKl3|Jz>*(PUG<M-qN;49Lw-Eg!~Lpg_I|tdUIehG;JN=EyXUZ
z0>V8ldybu(`-2m`LL?oJE!kN_%$HSgBuE*4m3F5N6H=_gzZAPgA)o|+p(AvyCz;aB
zA~cfj!x8kvTQw=r;x)=BqqaMCw}MtAk7|bop7ASA)+nqc3a)Cid-sOH9BLhQOAbSY
zT=Gg6nvK?1V;VNluEH!@%xx02CWTrLSPHG&I<RxQARv&=%_5IQ`6}4jm7DqC;N9q&
zZ?XT3?Xbq<blbA6PzjVGp?Si*C3?RFYB;|6Rwx;RFE4p#rKZXiQ;X)DF{eIQTllO&
z(Tq07Mh*u`R@MgcFO+umP{K&`B7++edgJ|u|1XG9!NLv!EJuLN9Xa&T95hcI(vk`>
zS{?u{D3<G}<oU?JC9sU&%C$1Uh$&q+tC@j>mt_m@$+t;`$x_yD#IRmlWx~$H>iG1+
zm78c#e&1hF;IQ%;Wjcx{EOF2$a!L!7!x8L19eSl&0DG6-;b((RF{h9$6Sk7fEF@u)
z(4L%Oh82p_oC`Lkd}yonj=7y}sNbi&AiXHlKD4~oiDE7U&GCq$t<X&A?=x}jI_Gks
zXZy9s&e1s3XAQ&N1)zA2h&bVJK<LgFx`_fx%dr-Z&eH2ku67ClTSN<gFF=&|^#*jz
z<SRZ2$ay$B+)AM2YuS2(<v`j?Lf2Wte;o!s>F9bR#)A<{&V3BN8V-Be>&0s<u==s}
z0N9ian0i6U=+Tn#bd7e~a2D)o|D6#4(r#-j(Jg$e7N3|19pO%a({@B7>r)^bT9)oz
zkK}X_RmT1)TA2{AipfWONL~C*IEiU#?gGM1lqB`2*tEqYK|R3FwOlmj*De3Dz5(%$
za%FynWjD&xxYLc1ktdh0zFkGiDz3>+2&9js-+I>7;k;3K6!SGL@7s;^vebL4di;3r
zYBdXzaf!%Z1_!RI$0@o60lz1!r9E3#+|Eaay%kmSs8Y<{<Wh-K18UmXaL}}>j_^O>
zw}VnjEBoQ)9!ljWYr<F@;yP{$%qWG>Yx<FOeA*s*K=p+$CXjc5Vs_im3T+6z`hRe+
zr~Lr`8I+8(8?S*7FTtY-Dnr5wM)xJcZ#sv)IiJP+)Rc}k5BN7N)m=5fl!g6K0L6WB
zxMgL<>(FGrn<9T^TPz!~D&q6=gA8vJ1O*)%xmFMtcXq{tpzJmnTB&kKVw5g8u}j9H
zlbkW-<;7p2Zg0MIon+4%6I+eqb?iR`v28xobaX#wUC}-_M_aIB5NO&t-EA<^RZ7W4
zjfYZuPsd%E`i2JJ@d790U&)X`qY&Ape{fA?VM0JSo_Ts@Nt#AN{4}HH^Y=Oj)nh;I
zt-Gq`x<{P+um#)R<gia+$xq%K)}$P!-R5zMO#f`M!gkT!Qi`@Ij0IE*gThW#>jslQ
zgLC|)PgyGnPf|IZ91O9W@RukL{`>Q6O2I<F-1iPNZ9yh|^r!rQW(8;pX#U-QSXGXB
z(K@A;GJG)J9FFNg>kS14@Y<%pZO{Q0;vqmb4Q%;Fh+u1vk+nJ^wE*+&O0l%kyXOOt
zQ$fUj9xfBe6dO$D69|wS-4wVM-Lee_W7lGeXwpV~JC2^_#*~Z5(7^)SHj!t$Jclp7
zoZxneADx{JKtt+;i)JcWVr~T2Icg{<5N%pwqFPFTdJ^xFBEp}SuuC*Y{i*pfk#+f8
z&pr&pqhwVJ3tU8FbABMuRMJqy#?J1|U>C+4ExYa}z0n-ZW9*YGie$`fUS)YB%TwxO
zBLkv#Z`L%rd50zqjH?h_y1)?c%tGnMxpbrM`*%!xy}FWF1<ih~-$ACK2Za>5vnm6;
zkp~MN|GIk~X*FNdPe5Tx6WQwh+JZ|fD=ZHwI^VbemGZCE0&08iNa4{8h|v9~J@HHG
z3nV?*#rcdH&S;7*ey7q_vp#Sz4a{zdS(?3tkB1r)d~I_4iUYp_3WcbWfvyx_b{rv|
za<aq-r^u|6KbA2Ii<J$7$~=mxxbfsyCh{hcOv4B)7Xy3!{RhxEeBZT1`XI|>5BykR
zt5)z82CXn4oUYFXUC&uIyQ`qfE3XS<@s67qf>e{chu&e%Bdtr)X&PLy$PY)Jv|S-#
zz05x=ejfunlslio@_LvZ?9}7VmL*xG$WSZ!S`c|nssX9};EPg-Lxd#ylJ_#fYM!<u
z2C8@TWTI@$pumv5l;9I#r}_9)O;<#m-Q^={hpZTAxDDJ^x3V^)C5%h^GwLeRL%f<U
zc`2day=xjc>J6vn@bfsiUx$Wgb=gX#yz8<x1PFn|`8_WzpASv%944l_HXR@<*}H33
zHh3Yc5zwj!K=dId_EPIF!ieo^sMBg6{mffGT@V~>H&)Q}{2?^zt*Rw2u{x@~;n~D`
z=FF~LyWXobK^^E_EOq3VP80=ZNda<z#9vQp7l;H>ZDGapH<4c6pBkxqK}-UQQN12O
zK8(@{f0&QlmgqI$Ll){=A}HOr)V5-Ta0l79Y*@Pu@#yR6%~o;OfT8*k<?jr=Hi2#F
z1fp@eBh*SjAP@Z=&;XixxP+MdWG!*wY2ND~{<nd?pqD^hh<Cf===to6acM<cPjP)d
z=pPe9jPeF6*Uh0S74#u)c0Thz;`bDB4xU{sfk*H?3vdC2xtDc#{Y^ByhQH)Z7AE9{
z7Zeb@Cw=BnaNY&!a}XiX>@8(}GnhuRqWu>o-5dHDmXv(m)V4Pqg|SJ$Frn|Si(Ya2
zFPu^q?ONBbT?5i8F$!Z5Mbl>s=*UJkS_sg^D#Zib{UP$)yxNbm59#xaB|+)U>qW^i
z|L%2Y6bL&0RwW4y9GJp78T*(;R!@ZB)fPnc!y_$B^}4B%6HbDC!tH?M8l%*~l!4V;
z9lz7xAKE2Fq3Bd{x!)jvj*Ooew_G=a*D|6dh^Q!DctBA+=6t}6VN*}{EPyacW1l?9
z9!w{{J9Q8_7O#~glp6>qtm!O^-Rlc{_nddQd;_vCX-u=)69cA@Z6h#qgT$@i8$$3i
zjE4ioV)0;QO9XHBIVsdB7e&bWkhhT*ipwQ50<6cP=l$%ai4>9dXYzZTaT0-q>$=QE
z)#a~oc(oKxCEmJ~<}HK?aW?YV@;)7E%Zu*y5B3p0^XT!t*1iBr=4XH<xkf12;ET-n
zm~C3B23(nA7tOC)$3{lpLE7Htoeoio;)GtqzF-W+@+d4#A|JOakByB5bx$+<AxQgI
z?avDd5#ANKvbLBp5|vWKHvin$$LDpo{OYv36DYZb6L-zbrg3Uy8jo(<xpS2Fk7rF^
zh6oS3hgfeg$}(mFEe4H+EBV3=#wi5`iQizN)1VB1tqUU@-~p^$zz=;6p8PQ$BQL9u
z^uv3Frzexr1j9umS+wFBN00>kO_#3e+jg1U)PhBy)xYlUdCV)1uX<IAUwG91<NayD
z8|2&4MZ~jI=nTW5WItl>9I0RGciq{Ef(QtTwH4oy9An3FwWurqF+;AL`yxoX8UBI=
zBF?9{1l&LD4?!=5r!PrfUropzjwHi-Ir13_AK)PYmtk7>Zn}FO`v?!5qUbmL^}%R<
zfA6N?+qQ~|d)T+Rc2600q2i{;U~0)IcpiV8AahL<ydn65AY-F6C}lTlZGV5^G)Rdx
z*uoYrAc;YhA6~%tT3hf)RUg<B7StA)-bN?l$U2+v%>t9Srl!wWJovL**!_3aqVRw7
z0qi0<*r=<PDcE#iXIK;+1UCrXQary}S~#?8eZKF%FuS$W){41RkR6)myt}|UXiGTK
z@C!>)T$Pv`;kvQUDjR8Z{)1Ow{^r}v&P|xnyd5C4iLjA?k1IS)u%aW!T_@X0AJ|iT
z!b#t_f#aQHk+pvI!ew&Iq-0D=)MNFBnH!(!_bRlPkGl1-g_Pk&{R^(A>e>V`?84<X
zGcBEh=iLSSrBv$Xcxk%Rnk|C~DTQ1SWZswQb~1UC6?2W4(`6wUBxLaqG$B(*=(lyM
zMDW2GsyC*Ne!OjZDJVYR{D4vwrIC6O_avxa)uPv6cdZei3JTj%;j3#1z<=m7{krVL
zf!U#zNCYxu4#@i&^WrwvCY^kpe3<_}A7!gI(9^>l+QYQ8ofwnddG#oK*zs0}Lu~#E
zPJkVjJJ?iRorIqQu+|tzewdp2>};zgQY;X&lr|F3=fdlo01ydaA?imd%loq3&loiI
z^??7)#3-F{XJKRWZjJ3y<6_z+diioOtZMF$U3CUU8cL}9shRQReJH~B!Rsak&`xRT
zn39WNG7yvAtgs!gzV(}<XiNl`5l~FsQ_WKFz>e5+L@HAd4-~p&3iZJ}d^4mH1DZd^
zu)HP}5@xoBvQ&-`-vIt>U7Qn;DqQ_IZUu*oh2{^G*pg)@q=u884h;-6fe?na!O*%b
zz*#`Za#{XvX@V1oGreb0ac_hIwbKS#Y)c+CyQ8@<bne>Kgf)k!d|gazWPK|Zi8by#
zch3%DApqq_PS68eosO;xRMpNam=fxgBOCp$t6Kw`WPGU?fpKJCN;0-Ox=2xMYRvA3
zw*5wGby&N$xR~+e$w<5;n0n^A?+bBQxH)~!-d`{o=UuOV61pn58u)cygMCMdDxpVu
z>BbK+R$a~35SsyiT#2zj^x{gQG@Wlq2s<s6J!DXS)0w@>(9TgccyUOt!i!-S2ySsh
zFK>YI1gSZ1OTc&HvWY#W`8$o6Va>HrY=Z;g83X1f*UriQ0QIBFF-O;m&ezW`c5z)a
zN#%`*v+53;V_sM3cK(R{Vbg&}5h`eNTx-V-S@79&t~6riXM)BLF=ou?L@m{;ALAxe
zip<2HmMrpsH>vX^V-ztk`A)?_XEwE|-BjCdm%34Ng^t<4;sdi7pADU24lUC+Fu+CK
z;`6(~40!3{#hea9VoYD@K=&)WY=?zb!#ZkqWsfom-}8>s->Ld#<K*a0@A`^o@K#8{
zU+{)8EN5&3&1qnZw1}-hV<*I>BtsXI33pllzZ|;ITVOI@d37~Ulb8N%HlbntI}n(A
zakwAkiY3fO(eI*sjdlIAr!)dicn|PWy!}be5Eh-&vB#W|r**-5+&MkhBK{0sAR`CI
z<Of;dv)0?CF^;`xZQPXWL`bE+7U5ofD)^`)kihi3+aV+G%kDvlbZvk=t{%-+jA(D_
zu@H?zuIijeYK&Gds6(pc&&`QfBD^xjtvCMs`SYRpo2$IAXAv1Tp<&>~ND3$$rzy9U
z&SV3gqY1p1%yn44;#_}HqGXiqfWM*Jny|(>cNQoX&fk46&4;1Vj*#z|%^VNKz9e#%
z>HSA+yI}Qfz9A09h2vE5&S?9)2Pi3#rSsq0OGyNkUm9~)q7<lum|0Xl5^=;oN=JVN
z*96lXKo%*!b~JzBcmY!z|D)QQYX#wqkRjom=%9;NibuG_wk_{a*?c&^QMl%EZsS6f
zw+5D$o`S`6nl|R%!|$H)V*fey9^>taLZFA44D&A&MUb^Y+ZiVv&EWIl!z>bYSHn2u
zJJ@)D?u<_726*CbFy3SiI|pZixbH};9Gx`+8}MBahgr{`S84;($OQaBm>(`9Vbx<(
z<0ZBlcoS$K)M;2|qb~uLPNPi4c_HF2kTs#`sI@y(huRVOf8hE8)G>=dYQ|9(SX}%A
z^b-j;XhV%HO8q#t02a=PskOAVRj&Fye?5pT10AT8th_&Q0?NtBrNTZ5=qUIjBf$?$
zP3dWA*IErjsc$JtB#sbI6MP#Gz7HHdijmHG!^%j;7n(Fou>`U@_%!Qkt+4FSF7XtR
z0`5SJ2SRd3PeGwS<&X^m4gLC*3kaP`7P@*jZv2D|iPq!a+Y78;p`1d*K(J!A?oAH?
zfJLOoz}=Q9nc$fcfg#_(gepx+dW3mV#>dOa-oZfNmGl~6C~IDHF+t`66B{nXPjjlr
z%)T}@K^9lW@j4|JN=O?|<1u}NR}FfI=u_1l{(|B5kS62w<K~@i^}{bi`G#erZXXC=
z3AoMlqNL5Vl&-*%YWZxUfods=@eQJ$S}(5!{s&WfuT0iBLbrn-iXZh$?ocCE0fuG}
z1@Oa|7km26kXir5WM&j&I;oh@8iQ1ZC_+zVrs-feBVrJTw?kwz=s<}6SkQYm0`t8B
z@dh3OutqEQ&Cb&xwZaOnyGPQg6jo3VAxZ<DV@q5APWBiIM7-0tYx|Y1(Vgxb<%WmT
zy4ULgGjP>&IX~L|nZ1w%>=tUbPtae~>HNeX6=WbV^EhKI_?V*KOdV8QD1|^V46{h%
z1>FLdV?R@I8krBrq6hYHx@`<>_kNa;eXCSfS4aFPRn@Y&wwz8l$z3%6gV7Z%HoMbj
zFAN+1a9W=Y^&WGhF7FFFnW<x6#Q4R6;NsmBx{WfBD<MbYs;KC*iBf7B8lx$Wdtd@S
zXbu6R;RlD4BZ%iXM7$*%(N)H*>VgQw#6}su7Hk&KU@Tk!#nB7d5-x+7odTGs$T8_M
z))pGeXi#IwF{L}9s;iZWJGXG;JKA1^DPdq>LOX38ym+YWl*LgD04~9ofk8BM=V0cd
z8y7YGA!DnL?bq)J;9w{OdwchfjL?vihb@{<A=)CRWWGWMif9g;e(M~A{r$l`JibfR
z+<XSqMQ~2RAeW(@G)UBgg(xZyq&8Uc;EQl=rOzlS%EEu;1`9FY!~3$e`{$=W_jLpR
zRGie}d}}bNi;~`envk{2M+%cbzR#HqZ@(;#``MJ>Tj5vWYj&1paZz#<FBlBDMiw_$
z*Sp9=Gd@jZiI$a?iW!`YQ$v#q)dv2FTmr`eTq)cNg5N_u-(;c`pK1Tse0q7#`{b^#
z5SyaafX{3&%FN5N!?%(OZ>3OxHxnAbK|!}vRGMIRgI<ST9>1S>L@oL@&;((o3=j~^
zaSpiPdMQew05d(HSv%~TztCn3DRr{66DcgP;I@yW0LoNsnmweX=k-}%p2c=Q)%iXR
zmH&*B5s_Vn7{n^XK?cJR{P||1BMi%?rNuZ|3ZER_#)u9zOfgTK_=ZM4jLIQa370xC
zEFmk4Dcxp(aS@a+U3!w8>9n!Z-6q40(Lj~Kd;s1F?<Xi5OnyZu#39OC2xzNzn!u!B
z45cLAH{J!tZyYrB2Lf%2b9=lQeRmFPHxpbgvNvQSa3%YEQc&=g!Ca6V&;o$s0FvO+
ze*OA&yavNi2Ni>d{+N#FjwKMWmW$epi+c=f1>EeoHGWoBX~(>e_{AuP;pzmdvYk72
zybg6iJvqXg`ChmMXD2a$EJ`60-0=LHCdhQ~dto~WPX_K9!4r-@S+1^>T^CSj+4S94
zGpLI2SxPn;D>Zxxg75SI9kcfcDA08d?Uo?tGIq|T<2P|_P@`=b<l}RS4fDv1&`^7|
z#tMX6QnC_+znU8BM~`N4Ba4e1#L@u=rnx=fR+IwV{crNCsIO{#?CbB>uUz8tmh{G@
zBR}`{);BesL?JjiY1%8)j|9Y4YM`~f9n#(#%Q^B)V11%Hg0DY=gL!#*K$g`;sLo<o
z8v0qbFVKse(F=lgU~-nN?A^OVDCk-KW_6MRY5}~8f*!VD2GunB%g*XQjU*V1s1n@W
z71dsdIerW{w{~JNfSrly+9a(i1W2s&70l8nJ2i-{f9faQi+5eD>S*A+scMZI00KJq
z@+S7z&Hkc?nKbmOdmDn=<xGy#(yMB`T71_`sJt=-ln2EJ@6)t17Wh!}JqYUBn)0js
z)?QuvV_uOKU8E$j#5^V9i2BY15D0`rm&|2FzF{_dqRWeJM<A!pXZAd_fboK|?;!h{
zIbmY#0`~dPTYi5`EqfIqLh?nBqi`v_ON<YrwDoefzR<a8%RvW~=-G5Z{Ukies4Pnj
zOdh{~KhZmP>lqg)+KoOUERR+T_@Q?H^3V1?gfHn*J8E&Bdvzu1!EqnwLcaO=`5_M6
zOyGW9v57v%Wk&)LAOP0P<*i{8egf9XOYH<Q7Is642gBb0wsvO01P{|=f`?bUn{VMb
z|HZyHY<96|L-VhH{F9j_ww+JZ%K?JoT<!HlcL1c2diAw9V~|GQ(Fw(`5K3@D`Jn4$
zbjFMh47503g+>*E8jLaD4hbp=00GO(B}w)Sb>yUXQJ#k=4C3Zjcw)`)dq(R?ad8}?
z;mXdqe9H2lKjD~Nm>mhb60Sh)t(mWCDHOJl@U-C>U%7JSyI${{Pp@brOemxw@merO
zIfpPi!)baA!hRSAN-vH>Bs^NqF(t)OK(v=SvrqwJdSt4u6G_8aWI(o<>+zkbqv-I7
z^~219g7!c-RL0zNv5)73&-rw>?2DBt?}K~Qr?j*KMF}pRFxN;7G_MQ0yj5yENFxR&
z(BrM?*H<!;Pa#$(w$p$2E4;!<q80?NF-(gx9oK4*$MCgWrd8}u*34>ap26|)>Xc(6
ztTV8BVy&vFNqI*)5YLZ7EdSC@k_Orup~a3zHA05tS|oD&ZI0=@Lw!2Dqv$-E$hpB!
zUlB#r(Z@h0bvu7z4l>PyD?GKJ28;atbBIEOi{wd^N7<HvhnAN5Jz{SUA!9jq>^?q;
zWaJFKQ6Bf3*twnL(s;>C(5530$rE$j-M?S`@BcGhxss8v945ZJu56li;lDOl+~%JV
zSGD;cBw@#hdmz3pUHn5>OycW*&<9#IO^}Q{$FYvMtCmLvkF$SJmfasC&$7fLR#aCv
z8fOpN&*<)au#nKwz{#bp08hxL7Pi;M)-X8>+o4y;y!q7rHy-6}CuKFznhAA&Qs-V0
z5~CvImgrx;mO}d#j**Ftn4miH4(9#%9huM7$+sCV5{7Egm;(#5P%2+(pcO!@R7MDU
zEoQgB@&8`I|DYH5KX9%7pLok<Kig=rcE4<>!DLWZH#{u@S+y+_`h@T#@VYnU?zE~W
zF7B$&58uHjl8Squ>5L2G`+JIjYoCV+wN&N+Osu@T{va)j+krhICsCyeOUYO*B$c;|
zn_xE~eFQ%%m$Z2@Dsd2=;1fB6tc?Ns^%(X?qC`iWgqJe);>fF!S)zYJd2bK?+((o+
zHRZ<PETfR5qFNY?5R9xxwUD@6HS&Ld2&$wy^j9jgzTLsN`6ouMgfVRxf4sdH<*{RM
z-!Ob*LCk-q1d@c$qJ`<hS3ta6Y$c_oZ7ox}aL5m9`^^NK)k~ItCMPlLXR$Kl%jQRo
zyxRPzfOzS!1%wi0h?UQRaeGgq{=E-Hpy5V<$*Kgbq0;afhs!((>x+MXW@c_~408)?
z1|d=D`1HU2l=gtkO(<scFj?xm01+PEFc489wh+a(NOpF%(j2RmIsQ2I=VnBS=K$D)
z3Ql$C@KIXcGco_+0vrug)pp2yy^J4yQh8$Y6PlR$Tau8O0e4bvpvFi1=eagv(4DEN
zY5geo<_PY9|N3@x^khPvMtXY44;@mBTEm@P0y>O;DOpF8*&GE&3DYUPt%#@R{qNK3
zg!NDoy13B|cZh+}m*%f;KGaTh3(&^%(WnN?7sdLXIXr3-0&T_<-|ELTWdcnb@`XK{
zZ{k0H4n@Q%l){BKkcx2;Pw0~V=5uWzCA9PxN5+9tSet(5C!SII=5;lY!r5CXp|SKG
z(1xn6F0G1n{M{G9%}vR31ru3-;J&5RHzXE9z_ZOaQF@Lt7mlhxByavkp#SghE`=Ny
zqn@Jrfim#&<|h-cfO;&kbIa=Wn>Pu;Rw5A;<T~-zN09p1zCap|F7+o#CtRq^0ptUR
z58oNrLd6x=@Qgt%k4!F3DIOaD0IcyMd^+*fa04JlAkxxIyDi?Y|2sZ-ZoXII8~*=H
zfp`l4A0<ccf-?p1%j7aK8xSs}q;!}y$^r-mUzqKVFLW~D{)6thx%sEB!Q1efJ2zLb
zQ4XlMzFM|%qA@71Fo$UxA~-N5gfTRdVV80P3|~naE^nTAzf@7DuxsW*89xltCl;Lz
z<{4wHBzB%9zU$!TN;OCBn-6pWm=0mz+~z>6eUnR@yQOeyY6?kz5k|Q;O<;^2iIL!D
zZup_4y&X+PTc9-m_2;|`kMP%)N*F?fUVNfL9PnZ&Zov_P%PN2*6T{Ps|J(yJ3(7D!
zTKFn-61QZsnTU?Wf)j#cRx{-n;59e|fNo@lbcQq}N+kj0p*)OP1|+pd6LK<qb->64
z5CRYfSPGmzP1v1qbJypuAN_;g1`sifyoMMXG(?Om+rZHV0X2mYT!jAS42EDd4F-$(
z!s7;+J<5I5L~xh4vRc4h60MM1x9*EALcEEIggt|Ugu5(a3F=wk0XS<2{iV!(OfLKP
zRZMB)qj8;WXrH09G58sD?_pvPj37X!hacZvMH(~%+VSLJ56};1Nul`KyYK-5<*7Rb
zZh~MLMF+!i3w()te*r*%tDB+_2QMArwYhrqVJ`CXzh(h32~Kps(Im&bKA*|>x43}&
zW9WIo=;EHKDTksWCQFmiYhhBV_QrBEWXJu1%5h5QGUHWuL8z*cX{2v#%#s8b$OTj`
zP7q7N8G$*A2Zs`k?4|klsOHDJ5g(0W5Nv4j|KbX@*GoVgI+ACG9xq|5u|5ape2}n1
zTZ!sFz|dY}5a&G;oUA)!?NMhn8qH&x&rQJo^^Ur_y0Gw;UO_0%0YTzbLF3=~X6^pa
zZ7_C5i?4rRAfPTlEP|K`?&9KNA4r^lIF=HI3h0!duMq_ZX*-@K^z@{3!0EiZjlpY6
z104u$p!3t}zzEP%+(65u(a7=zKL%qG11Qj?Z)+(pNS>XY6+p2Ji}d0ML>=(W2(zFU
z?3(N%B1Pzc%76(0;LK~;DAN{tKYS|s!Rhy9@mu35HrI7T;qnNJ-kTJ{cB<v3DJGZ$
zq6X-}Y{8bPE`)v}QOt`PIzZq9_5Hc1Xu#62Kp>@4Pt}3c!}LY620H{KP_-<g0}y`t
z<%<^(AhclE)h8mL#R{tt&goxb(LBHn<pUh6yMoRa&~C_hZTsP-elyd^ySNFc8;EsI
zqfKnU(uj83Xp@Y(A<}!q(=~W^OOM^g7_nD8%FK=UJ-E@tEXkK(KNjET1bz?k5=8oF
z{656Oo|?Lg;ukiA0Pv)9nt+~aWZb<y^$UdyWc=VQfqrs9=L-pdlY%|KA2%j9){0;p
ziOIYT8P^eQ<s#JVzOt@6&-dX5jvJ)V$KW(DrOtCWdTbSa^_$M#U@j)|X=dQsfT7g|
z;GFT8IC1;_+J#_@;{m*b{<$5nQ#7$k;|nkp-j=Osb>>H%5w80@Zw^!*bJ!6`b*#<I
z($ePph|N1Sf{%h!`fxK`R@T8U-lswb2G5u*E=Y-Vv`m@)k6Gh(BIQz!^t8E!nFnd@
zi`W%&%w%|VI@Qn>9JzHQ86fD6(IT|DY<c=;v(Y>N>vWVE2smNAw8L03>k2LOkQo8R
zj8APb3<;`T<Ei(>NsIs|F`6aH+vNUzG%rsjIbp7yb6^?*>Fj&_Zeln*3Bze2q0Sb@
z;TvH{1H4oYZ6h%2XAI(G@B<^T;5zf+GrZJL8`q<s!lpHd7NH$n*UqWsfT(NBw1e*%
zC^}binybsp(J?K`5ai;j>uDkfS}N$UoRM)WqSk)CQMGOc2KazRecFiubTkA&L~;p7
z*1NPTz!rK<afMKZFeOpa(>sCjr}y?4KyP*MiB8vNIs08O3qL<5fYy6)W!=qK#FjcE
zdiyD26i)Z!zPC`-s;a6UvJ2P0|EmMJfm^USx^~X}9QjkCFWB+><W8i95*ghwOe@!O
z!Hn<k-B{6hj}i%hzf7?nOKI-G<EdDYXmmkXV4i0U{0KO&<PR8dp@NM(u@p&G)<YuX
zEjf=}OT72C6tqswp>qK^3XaPJ7BEoR55o#5efRmm_&fzIs?whT=hL#Xnpwu`qTW=!
zz!o(+v$;htLJCp3f0%6pLfI?MjW{mx#cZnyL9Ajtb<mc0Z-`o|`yvId5qbw4|47?>
zb0jYjUr!!nw1Lq{FwUW`b>7=UV@C%`?U_frou%y$62F6TAOSDAob!UC%qY^_UQ>Nf
z0dBvR04@3wptv&mO<?@MA>UM2-&+y_a9((!<ARR{1sD|%7JrgOSiQuv8KxNqp#kJm
zW8d`^5h;Em_f&G5Phg<&*n6=E&S`ac!i<4qA)R6PJ|$%mms53Q*)aDtyC#vMp4iMg
zj{qR1I&?^H>K7h0Mt^;0!j7QK&j1rKDxlazBNJR0C1z!)8}8is6S_VZgMMvAMqC-V
zC<OKF1*JXRf9|P?$j8qC_`PQv7&`)BPcfyWM=T0hF*<{&s}hHzV293ZE6If43FZ`%
z0ev8!5S5}sb`9txPG~I4`wv6+Bw-Q5>&%tPlY2q1D(E7nJ7<V$#@}@y7Uxr9aehff
zMvl;9w}T9EGf6W~C1naNao8D$goN1kgEGstlAv-0|Hh0KydIb25_muYrbZ3`c3X~?
zJ}eON0~-*K<$6L(MoKSyt8EN56)~$fei+8tmt?-$Bc~>Dz&Vai2?1)R?Li!db}&Ti
zHu7b(Kcc(|`B7h=-g2X}vojoTrBhO#;5Y)H53jK=J@N&bR5)TyQ`@lqWQxMolYF{o
zC{TS394DUE5>e+Mj1y*rKw&<^k4)|Kzp{WGWJEt(9TfI#EF!(6S3+oc0UQMp+{*ib
zXIL2;W=LjudHai4k-{OZ``{yj`ERS^=&%BaXHDXT{U3hZJVn>g5VW*Z{4nPIWpq^H
zo_G$-5~ERRF)%49SS(&80nR`FCQ%XL;bUWev>0|jJBJ4cheYS=IOQFGf8BmB0CKBY
zBDjNpOxQz(P>_@V0>X|rA+)hNTREwNNsllHT?gw8l$LQgvS70jCK)Q}Sy_PXlZ%z6
z@z89UwC$%~SU3oQ4}-&38deP0rvTCG2!ldIvL%!$a1BOZCizy5?s^3FN^B`nKwP4Q
z{}Db@KxHlw4OekwoV%W#!-<HnD{-X&t413c3VI77sL(O*lwlWQ9=rold+)|fn}?HZ
zY;bQOdbL9l?#O<LsUYBMfD3>z(CG640ft6|xS2LT=^9jIQ4znvVMT*y9lhKUZy1<D
zm3jvrCeUqXTaK!rSD27sjF<;}mY}yGTaCAUj0fDfn}`}uBmf0t(}=)5S-lYC$1;!`
zIuRg_2)02=0V<Np<w@4krY7E#CrjqGj}^k5=rVa^KW^vkn>R$`c<wz~8);EP_{foe
zHh0*WZtPUp;i^115|fFGgWW|Q8tCzM`j&{{8eq~DUqSB_w<TaZ7yDI{czElZ)z=?l
zelKDm;bi6R&bO9`3&}8Uo<QnIQ4k~+0rSE<*gap9RZ;^937v0Pj)m9Wb!&uu;ur>G
zW0*$(Mvm1kC@LsGU8k8NjL%k5{I4PalgH4TKsyf+NC1ZBJ@!@R5DgMD5b*?8>O<EX
z(13#a&A$>@H-^ubfTpuJ-i`@N_aBs1dq5%p5dz`^6bCzY?)1&89l@E3&{}}mFG%gZ
z+q1<YP@?`s@=1gh)F%W2!g|m*G!$6{BrmYMDF)rI{3MPE5}tsz{AS<O?EkwueWFeX
zv{_lJdKlk|;TDK5M06Y<I5m)Y!(BCAyafDtob+%O?Lp(Tw|548E`)4sX~Yl!D+J$n
zeeVC6PMc17;a1RQ8>N7SbyXl=O^Y&UPR&Txal5MEAq1lRiTn;{g?64*bxYT$=vZtZ
zoFz0oMj(9rMw>$+E`trq;VA@7NYG9>ZzTyE0oW>&mm`??PB1xl6EDGiDTBndth4q%
z89M(z5suu<xbS*6+cw0%5hMOzG)VLe$M=6Zi6&hzpJ@mX7?Rj4g~&^hg=?vI;lu%5
z?`$_9ehZLVUSLeby_~XHI`;8fUXx^U`}%{ce=PI-Xe)Rk|Hq66^Vc*`Wc~Xg*aD<J
zP@9`c{X?RL0m8qnKYdRAyuWb&9_t;kA@?;+k~+7S-~Qk4NBkeuD-tlaCV3(csu{Fq
z^^Pz{dIa=wbRlMcE~sJXFHUG6x}bf|vamB#14$Mb4%sIe7#MIMsjqGEvWD~R+?<8C
zWItw0CTrr;Vt!)mWfxfAB}^b5Aq{UDA&U=0ngA!O;X+F*tBGG9oheqflJL;j9FueC
z%r+xB)NLs^r!Hc=7k+hej^mmwD7mOJESZpzZ?U;$0#`&&@D4mZak42|TQA~-f-?HW
ziVo;}{l40$6;pB#VptHeYv%3`Bhs7r2m0(lJqU^%z-ScxO%sP%2SA+b%rXrH2e~o6
z5vEf~ox!;n(;SPe1eZ|jopHa_GgOrA#4f}}Q8-{BRw742I4deD0y6R{W)ej<lKWg!
zl=f&w+uT!%C}^H8kmx)$OyQWl_^qIMQ9Vk;U}d6lGFPNsGv%D^O$#GOBc6NY?~l8G
zbAm~5N3NyRiY4YLg1&CKvQ>?k-B@xv2ySo~4o=#1cw_NE0$4?ei~8(lpy|Rnzu+tK
z;I`K3vYU)BH0VYMDO#UTSz3AwsxymfLL+Q`0Z?G5<>SSP<&;us4_{xf-)=g)K)-7M
zr^~5dKp|Q;o53JLgX*}n(!KFOY^RYw$PFN&!#8;n868Fr#w{W1Hbqo{2l9Em5ItGe
zC+4UNWRH~sHHBqILZi{WYD5NtKC<H|-v66|f`T+DOm=6xc=2Y+dukG8Ha50#coh&P
z`R?uz>WjhaVJTKUY5(56qqPxbAPQH?G!bLGoYq&tI2q4+#z3gU5NLUq>=afkNGI;b
zMZf;>nnN&0C$5R4K`y`ln-gKg;7lg`%dr3>jS;ytvP4wmVG@o3x#P{iDrOfKvmQ9<
zTtm3pz>+9oGA;aTfU>5zvUfvR7cmrd1<0~5(h^8%`Oh;lUQRg&$>+529$?Xa(52x?
zm2d{VW7vg?9Vp8H7|e*@P&#NM+-Hf76IdH!AJ^kEFh^|Bc@9tttG^#rLary~TVY^t
zAyP_;Tw4<bHMMM=;%A^9zB%pw1}Ge|O=x@P0cId+CWbC}k7>vh8Q;Q`#j|H6q7;xC
z5p`YD#Qwxn&=P>Q*y}kp^-yo?EQYAX94$&QM2-TGfZNI&sRaC`<u@O*As}>Mb{8#%
z-Y?Bw6MDwbk-ce(fZi<H<Aa!~hb8Q$Y-MdtkE$NtQ<hA`0Ce0aAu{-#tSS}w@@QdF
zAmc@joY|S5K1Ta4?fdSy|BDms)#g`4+WvCIrh_fXysna@LS)YrCRZbBj~rFlIf`<R
zaNb6ly6>f#EwqQ(P(|;_guTCk-tAjCMA^9hIXioa;IgO0sWX7Di-Hz($^$ND02T=W
z2fyW>-MiydQ$mtgk#!&!C4^M>E8QE98x?Enp-O98CuUo<IXvZY04#?Ss@D^~vVNRH
z_dpNso&T<*iG2J!s^;T|N{2z669)!KxI8Kv;xRzf*XxPJ3JAx)TZ$w$0jcDrZh)`H
z=;=Aw@fS4&-#|=}Kp;R0WNYA1D+LgPef`1T|K_!E42bAK&dIkj2Q~=~8?y>LwK)~F
zloQAk+W{bSe&$7i1zQh5qcv4k`M?=)7!a&CfBzg5JJUhEAZvm3>c;2tWVo)r9;=U)
z9nENZC^)EPuqZbA=r>a3igdOEG9Zw%Dw(qG?hUA+4*22*^l&%I-8>|*CYx7IXb>bp
z{xP;PMy{HC%UVYQ_07AtZ`oK`&p&SH?k++Ng7X%R7`&!D$KHx@L(-<l%_4JKPmddD
ztC<9HV*rO$4^m$;v$Ml|c?jo4MmHR`AVShbs*9vIGwttW0u@M=I4l0*G|+@p;=yz_
zR{tB-V>=m4G5yol_%+d$!x@s_v<`JkWb~&l8y<q@ci=!Rphm+Gn^HAQnLk$G(PoSH
z6rnl;eyMu2lU&D#pwrXtz4Pl)RFN8a+LwaAESr%`i6jFkhmn@qE5vPmpU*fix>0Wc
z47oL&{P7b)0I2p+kK!<dkC;06Q(*EuAXimC4-Ydytbor@^wdOa>!P;7i%7X6Lgd(>
z1HF>MpUh?ztyRMy-;=GG0c-Cf+%y-A5hMUfoi}c(Qq#{=U+c1)#L^;|c+VI>D8`Rs
zaDhht#T_Kh_~jU)9IhI>7<cIf0?ALDTi_#mk5aM1ZGSW=s|&M##Vsr{CQiIXl`0c-
zVH%iK-SWojpa>Fml!A9tGpV#5Gh&uTJ_(%k4Y531a~`p2le4YM=cipsTGMDwhkkA8
z%t|eu&W=$-!RpjIHO#jcGduHA#y?weJmG&WxZekd`67-}JXL!7P|(EQ?Br0wv}=M%
z2iR%rKmQoD@}>Z7NB01FTa@GTuoDL(zPd{+eMa^%3{bIVfSRH`0LR*+7(bvTo54Cm
zj1|5IOd4KLh;Z)iVC#n|e;y?!A8J)V3|(TZ<o_E?;G?i!{ci0DW+4hRkP9)G2=5<V
z2Jm=1H7U@Tu>r9V0oujoI*4agd!2!8DJUuemuWQ0ba!{h&_n_-v{jQQ3ZjYKLw$Ye
zs7--mjC$MI*z6;;!BfA%ZG|DR*V%Ezl?{UUZ`p0cXL3&OEZp@{!NtR=HYABpM)62-
z^r%K{v{5YJ8CREWSVHQj-0Go!5JGu|V_TWoTPy;iwe@hm>3mx_G05PEhnGI)HKUx?
zE-K8lyTb{^9r3)Xs#wpSot~S+@L1l@A0%?l)5w5tj3<l22YghZ3)Mqa*eck;V5sI7
z6u<*t|MM0$RY+IR7Qq}OJW^0ls*K|^GL-C^p>BbD1ZQ5bWE&bRA5L!^qTaS`n^#^<
zbUvmoAmn`n&D?u5U<-`j067BL2Ivm(r>m>Wr!x?&%|0M&Pzj;@2N%#FV+#4scoM8N
zW6|`$AHo=pa_siCze`IX*c0%7UEpI-oM5Zs3sEm{Gwn&BXcu9}4WN79=X==x2;v1t
ztN~bM3cJ7QG@4_#|1*A~=J62O#g3U3pZ@6|)rCPxZ<A8qn0{N{H!C$WY0$w|e?-gW
za+1PI&T@`GZnJf)wvxJ+c2KC+`w&Ghy#>qaf!{+1dMI^_^#XlDef$E|YU{^M#@960
zG)HszTQ5?09sWbvA3YF5=|dCpONWkXr!Pqs$#sD;uT`%<9d|A>sim56X4yW_`l&VX
zZsHlk9l>hxYFRdBw)k-4k76g>uWZM58)=%ZHMv2YYak7}s+MB`+xB~_M$3jixX&%=
zECUH7<S9ltlmHwxDH}J$*5Z}{n~z~WOky041B^+~Sg{14sedpTe80}*7f6$OTZXVi
zP$jns4P+pcPQ!|6B(n!>%}P5_A<%v?Zh#1pFyAama6|Es2>~ajn4l-ZMJDHNBb)7m
z=Zo-w<%<9nnZ+fprq+U(PSDZ(*S4@NDIOx?2u!xY<|ooO&j5ap$p3HwJP&H-pf7^$
zwLarp<To7@C=*dD?)w}4wTDcB1&H5hs&v^521|a9P(~hs%oVGz^!s=HJF7K=35aWW
zAP^ul3ryyp;HJ11r67dOgIvV+1=5Q)vTyK$ScJG9vIJ#5=ZfOukASs8oz`<+ZG8uR
z!zzY#m3;q>#UOR-zcdJSpa}wztSq{Q!Ie)lh#@XsFU@q(L&sM?Z%AmU2!iYMr&jbp
z*KyibYqZ&;4cRMrBT8WnBX})Oy9a`=IN^^;#GDR48gR}LzT>DBfT1Gk0PuB)gDpls
zDevXN;r}iayd^D>)8Imh??vw2*aSd(5^)*%6tEYz*>gZ$Giqd^n*hM7Z`6e=7I^u+
ztE&)4ZLWD|_oL1$XPBAc`3o{Gkt3gV0=5YiX;X?eW)<C=?s$l<8jcObZ>Zz_3iYS3
zsi8+t1&TAC9s??XktOmcf_bX=*+N=eyag{E(j|Bh#Mp%>Gb<xugCb%XAqVZ%a{Glx
zION9FpXbIbhvJ0C8X`M(A&mAFAP}^2^>lTGH&!404Smr9D1=Cs0EWHFl7nUx)*sEl
zsvzsv(-okc#tRtoCZ#za>lx}oK}EGOpW?)0r*Sk6w*|Ou8GtJ~kag{;Oy|y};<zMu
zD;PksTJ{0Yv0k7V;S8VLwx4J)L;pm4FRcldKH58=_}r-+^#>T8sGd234+81*mqrnI
z>YtF8AbtGMnw9a|4Sx?UKbJMebG767Ux!+TtRh8&xw;Qk#b>C>_d1?Y%gecteRX-H
zC_}tLD#zCbq-&l<6JdU%xhDo&gT^+vP3(V6Y6y34<RnKYy){`0Td3i4EW9G;A~z{#
zXt$Qi!^t;oKGTm<h%VdVlJM?55BKh}YIYAXjp=R*{M5!6>)aLk;%nX4?t@N>aY_rH
z7rtqbx|`6pSFIGKv>Y-&Xr<YiIa5`cubZQKnSDRIX6SfZgqi!9jo4FKGWSi^2O~zg
zK8`sKtj+EowASjh8K|XB@8dq2S^eIXF)c|iVl~3zMTJep%7ZZt$|BZ?FS>dZ0z&oo
z#7o{=Jsl63w^zBP6@Rv1>d#k?yz9JcpU-BUX)bS#{jg1m4Evu;B$}Y>#0xOcy-NL8
zWZx$V)dI|}SFl&8N7KRDMy~f(^F~S>j+^p)`u(J&u)Fq3i!NbNL2JEIP;A=hl&!n`
zz*E-(^aJ~aJeD!|?C$-HqId4@HDn^IauexF{-<w|KL<(2dqt%feF;=BrRu3YvNDD^
zux9#(VeYoR++1XIp2j&w1ekdFMh*(89A6gS;@b0wzcYsZpFXRiZ{z4#*!5}$eco!y
zv?r4kOe!l2f`WrP(LI2K;B6*1OvHw7qAsS~wyCST@|Q^IH}0cj%?~m~%P>DGV=$F{
zO)EcHe4zin?qs(}P4&HXZJ+&6!5JCyhP&IY9gdQef3_=)rG)N2qfg6QCM&%+rCQ->
z)=Kx)$kU#@4tpZ5%*lMbE>fJzt5=xgRL$yOh<~t2>X9`EkwP`hig5|<EU-c|C<rl(
zr(uEm+75K(*~!gaF$U}EgSSFY#bJdMw39asXU8KtOqUrHbLh&=<zH^qgGBQx*#?TR
zxUzMe6k`+tMh#(fnE%5<z6(fy@u59I=kLKQEjX1nLMs&M+jJ2P4vZ7c=g*xB<1jq^
zdd~-M@9!IjNm11OCVL3MQhYo=CjR~8uqsWkfKNM|i*C15V+I<gDW(clR94O`E>iis
zAkJ#Z$uZnNp_JPTgF=Wxo=5FHunhqu4JM2pDK+TGRNQ5ChbY0v*LQEbsFKoQo+*db
zCHX(xGYLsa`a1(r!+hH~WBW_;$oB0optCRCdQtle+=fF2!Cyhaol+iiVpllwZp^z8
z4Agab4NJ^z{}kh0pl|uygxHPfi2P~2|KI_;lag;_VxpkU%$)^}r!`v!<#?Rd?L%B(
z>c^RzcyJrS#=e~K^h@Or&}lvQD{Q><v-!I3P0xz1Z~HusK2O^C_KLn`G*p_u)k?2j
z#v^b<q>+1KkL2#-^mp^VSy@+2Jd^y`txjsn(sd*MoQGiz-_-{?B1Qw={%(!JO`Jt0
zb=^@}oh_f6pK5I!-m|uSJ$kO%dZnn(zN>4eU8t0hs<I-}0KKPQ;LxLTmqD_eKr`R9
zGlQK!-{l?aOYeJO@;Ng8PTdQMgC_4E=@%Vxl${B6U1g~h_Db5L%~WF@U3}@?SvgSR
z$!GOlN?$teXNZEMJ(h}Qfvq>gE+j($Fwl!#E_sgh4~2B?qn*#;?+r1CDTbX~CWkFy
zQ&SU2Yja8|NZSb5aP@;r&2^}L5N9DAk7WXogT63IPcRE2PwfYB1Oo)RR|7rQwrgiS
zp!Gf&9vRsH{0dSPtXEq~WjJ7BL{W%PxZ>#1Ke2cf)v!3#a*6~LiMoHXDe0)CEn%tz
zruvg7BtO|8v$tD!8aqY$f3^4J?^wQH*9{`1R3t-@%tVEdDVb8{AyEiPWJsb&6eWs;
zB=eX#W0ZuHsf-yyLWaz9=Hd3PtMB)DkM}QletPfY^TX#5?%R2t=RWsdd+oKfFY8l{
zwdW*M`|xpBGU6>K@gNg>ggsL-Gf7auL92jEJ!%{ZY>4AvND|$zc*`){Q2;vcxawpg
zb|cyUJlO@&3Ghjv%AzQvh3X9@{rdVkMB6xi%BMOEk|1poECxetg5`vs6}mwHE{@*h
zQ2=sBD0QT4z(Ng%2>RYpoM?pZ7jDBQhI7J|jzJGaGc)#9T0(D+=npFF0zSavw3;_z
z=LTsnKny_=cF@=ZL!uxjCys&|SDF}<X&=PqBsGZy!E+pHZ#S>8jf{<fdr{ZbRgB9^
z*YqI(IxLUx_=CH+JMjsE7_V?~as8tZInOPB1SJn{IjPx?p<e;FN6)iwZP3#E&-57X
zvEe8xv@md&*3>9AOh9$R@bYCgmm(}Z791CtH2otybadU0Z&1jB4$#^vDkBqZXa+77
z25!d3f5GwIx)*61oFxv~gc3|v8aF>WE8bu5)>>B>&(L-_%AUOC!KnpjB}_r#+!Z@d
zgp2co5rdoj2Aa8hq!kSf!=X<qT^G{}Oa@F9h9BF8kzC=MIpv?MX+YWc9gaUt($Jvs
zUC`Vqp8WN1165mNBbyHONV$~ur+$VZlfi{o=_#M2RP@=<o3U!ns1vF$-EZYvB69Bd
z_Y)mMMfMklOlb;3H+tTY_e3)~RW>zhw^QwN-AA#$GnapCH(76M{TM|AYYN$^zz^Zh
z#nb(do+%yR{3$Q(R2+O*{pY*43<aJg-lWO0qEh>fVuc8c{gz^a*L0ufA5)Iz^~wyN
z%r&klnCpp;Zy{@?{4ER_^fOk-_y8FEfcXUs7NKd2iV$*g&G7Hl)<$vt)d|%u0a4u4
z@xU+{=m}|QqX3XFGY3Kf_8^3U_A)Rix5n`Z2?4^OOIX(5I*MTy5S8-sg!9Xrnwq>6
zn|y;2y^`HO5dtJA&QULdvZ<w|g#-_n;5~G@z$7gXqxjAy^d5$X(CHz(#&D{U!Xw9p
zZ7i-&$lD#ckbVCw`x3%FgwN|sWpi2cvBm1bgdh94UbD8w8G!<*(BZ>gJA6JOTSRY7
z_cWm)Z{cU><7?#G(``IVTW=3^2K#T06TS_xgBMlKve6>>nZ!i#9_dlEn+cf^ZUJ>d
zixuU+P+)pplw|1NY+xhT03OV#RIzU(C)Zw<^q&A9X8aMZNo3d<b0c_BZ{Na6FaO;;
zq}lnn6VS^`$a{PgNW$FLYV;{ld|>h})WRP8os-U2NW(9Rb;>=qtT)_(hRCh%NPmAC
zdIk*oAecFAon%&gsnw574hWEQ`{(X%83Z)^dHC!Q*@}(53{wW(lwmL#fe2ssi^52(
zvhw3!@}t+EL<ZM$eIL)m8F`(Y9Q+PIEV5>Rb0At{*P=%W^3NOs2U?~+!Up;8-?sxP
zg772|dT4dvP~r4J*$GeRJK|CbOE-XCorY@6P1zycgf=SRWFv$Pc)dd%0|_>cmyNCM
zS0_v;scWPU&>cL8&bhlM8(de{0pkHoMolgSSQk%!?pHfRced9YAi@F0PE(d?Q88;k
z+zg=<FJE7P2u}k77H4PKr_|J;UgeTot*_!dZch&zs><J-Tu=`s!hI-&ZvLX&(w@4A
z_CrjJiHeHv(yrax$-pkLocq!%S1%Q7BKBtv@ZODuU1};Sj2OSIe%wsJn3!Chac?{u
zmy`r@m)JPLBJ1^tjfy&8{#F0@8PVv>t41D9LnOf<2dZVmseCCd;gM^p)(55~FURu6
zZ*l4zwDri{@bx$_6XOswC8NiA)r>2E=H3an>uhGO&x7KdlH4QOgPkixGX7@&DSpY%
ze^!&i=0;T+|J^LQ;)8D6;Cj!?$;EZz`VWEQ$8RDdnDbQxucyRmY+uMth!S(mhX2K)
zpqmQ_CiKWfrP6&v!^3w#RvLY0<m!5IXh#uw4%B_o#zn#wZg?C^D=C0O#^4bF1Vi_D
z4Js_unPuWz4^dkYU^P+(7U8QqHhx_3C_{FNkwgc0s$e@+KGMW2qbvY3Jg6%&>M&!%
zGhbR%u1tVSoQV6{o!J6kUlnjvaS;u7YzEN)U7Tl6pC;6Yf?Xo!Fw1a7fxbEcHpB)W
z)W9wT8OE;!S&0qM!zNo>Y#20h{(OYTUu{LjYY-kHOIv}Y%fG(ws}KTb;wC7a(4O!}
zMpp(O+V2rqE<ojaFO?!|!{v?AHZqC@dd8nJja&A&8{CDYATl`&!Lko|HBW)^1Tw_J
z%nTu=XnfE_&Fr}D*?_ksXfx0>{RqA%IA@}|_vIF8EyaOanG=|bSn;!pe#>A`ecT$F
z55R#TC!wUQ+<g7R$ml3RG{q|DZnr7(+?LCI$^jvWeK8whYG4l#t_4z0p`&sn<?IWr
zbg&pwadck>hli_|jPZ?zxmUiT76$jT+GG?Amh_aAVS*C~8)c^sU$*U7Sj<^#dqFre
zD%-CM8<(JW0A)to<?@T`np#?nM=zg(nzUZ(1P}|998#<DC4rD4hG^~?`|@9iyI6Vc
z&9Og!8Xj-|b^`kT$KPUN6rwD*U^-HHI|BkzKzdWkX3*zHl`d7(!Z!TXl_OOdO#l?Y
z4dNLXQhXs;mzS3I4RHvzX}DUEG>_q$M2oM!3dCpK+E1EDh+ya85>e|ySD?0_SO$Gp
zLYMxS^YfP0R>Hg(yu<pd1^}wSV4$;H95^g4{wiucIXxXpI6@?fT#8h%H3L|+-Wt0e
z@IR>?<`5zpg7hwtK1r{2R?&WJ+q{8=0lrW^RGwU-R+m*fbV*G`%gS|egr?{n*H13p
z9sKD6{KtfH=C<Bmf2~!JpkOy4-}Trp$nV8F-WiD*4eIEZ_jvj*)v31Gq||B7+-~|Y
z*YnqB?b}-UFNItw)`+AhgVs~6efBTSEUx8O-lZ8bvS9rdG#jMdlXUaP6!UkNa@w&&
z+ERzzOmP%Puv4Et32DR*&<@^){e@Lgup^v^{jR5R(9?(DDC(46U6}0Rf0t^Ti?}5!
z%J{gHst9W=9$IGMRt6)jFFjthRaKU_A_gW9u_+_M<$7XUlx@*K)qQqKQ<GM5mrn~y
z1{4UvF0{yA5fKM^I2l2g&dK`n)FdXt7RGpfk3kWMC_QyN88Vplr;H(D^!dvd@B%^Z
z*}zjpm(Ouu2j#9^mJEmcb6|iAh6jo!jFID7xcS4<)cyu^lg4Y`gM54m>>1D%A@PSR
z5QL=2USC{q&7Eu_dvYsfVc1-U{kX&mqKz%QJFPRtIuQ#;DA{o0=JuuMUM^i^;(a?;
z7)K$dSdiMQFm>**leTpN5-e$#Sq{rlsGM=nhk%mgj+<h6GF&x9Hy>~G_eC63vcu4^
zude@%_8P#FzDu^i{9EpNbuk4E^!0u2p;S;)n*jJRuq|z6#WheUA4BcoqQ8cW09j(}
zhQl7IU{1mYGLs~wA$(@+zeq<(T%eG7i0iwDs&%l3vm@s^Q}yvtj9iB~Yy+&J45G8P
zjWyQWFEc9<y&ML_QBF=I4>%y2%For=Xnllbz09nvpaOYrPzsG_B6BEKhx^&#y^i2h
zj?Z1hb2-i+bLdbn;8sBAXjPg{^+G~YL0#cFI96xt8B6d0lQqT$%iYt#qd6Kb{($aD
z+LNn~YYR4N+#6@)fTs=T7S~Mcd@;{88y*{YF*NSmB`k8|TIhsPN#{maY`>RMB-62h
za@scO%*?x&7Rx71-^ondP1Q-*>nbsFGP*x-@0r-JyY{<r(&M)5+%WCksFxp(bktd@
z@Y?JA=r&$m@0}{LnQB_SY~QoD<gmK`{yR3YT_(<W<M$!;ftMw-`%n2U94$TW@vFlD
zZ}2#p<tM&(O1|5-ki|mJ_3en<l=7jZLg~R=;eM(f2GY$DhQ~%bWywfK@uBd4tE`ij
zvd#saHcA^*OH_vrQfRLrn!2onqx0+YK`ra>7Zp7{C$v01Pug~WT%H@*FmB)#5cuf6
z{^8->pI3(GlQBdf)c=ENdfSj(h!qT<26~<eD66s2YI2SOh*%Q#MW5+Z#jDfu@{EgM
zP6mbRT3GR3KPWTw@dU|v?;6sI{oMADZMecw=i856KOXJ?*Aft|;!R5I(4PlARQ#`R
z`G~k2bZffNZcOd-MlO!hh~lnjWhX*)KX6t*CqygUd#d(#lHa^3HvQsFxKig_g??Y6
zjsxfXNHcfpUw)v$omnQb^&CC}0|T7rNLf8$FJJxuki+QIlUV6K1(;QRK`6L^Roauw
zXs_`~0uyhksuH9s;w{O9L5~VJ`5;td;hKCLc?r2of*>+fqodJ?$_h^IXVlc_4W6j)
z-@A7&-&O9}*<PZQOY%DcVSx{IGS;XUx$K8WMu@C-$N-xu#b6D8&J~lP8;*_%<T9pN
z0k6omY+l_HD3EYajuy?}t7PG*i1@S)Y~5oy^AP&{$QiU-HzC$S=}SWPFcLO15=&EC
z8%CIt&&H*urNJ3E%rTA2+$++nA2$y4Ml}56aF77|CFnD7LI|Sff_8x%77BiT-rr~w
z;q}w>_=Kr=51;m>TVx-e1kfMW@c%eFezq&wBTI*Bds#xHa+!wb*NvARJSMm91rE74
z_%EK|8sv0OF*lZJ$g1tRW8WX9J19l2$Enx3kELziLW`91Yr{{r!y<&u=jX~#S-Xvy
zjA_QZYR<O#Ly8R^QhyVsD0}1sb{4xW<{E!mW76PPjc|1sP4Cn9`%`elrb4LX{9@{4
z@?Tj+3sO(pD!$%d?_a~)V45Xh6(UV7O`YXz?DvRuOh%n5`c{91zlNs=MS0rKwE5d@
z6kOYKS-X?(vHC{uE|K!7dhi0Y_^Q9@El}28$uE~}eahwoo+Wj}fm87&O77FYA8-_P
zya36FnQuo9G@w*!z6gu_rmiPELR<c5)?a-=XYlMNu7gxFe%ZBYuj=CXEzZbnFMsqy
z6Q(7RFR;}^p)Vt^GDerD3FSH?{V+z6D@Ih6s~Ei@KY0!HsyWvG+h{Ax@&e-9KJLwG
z8X8j|d0Gk#MCN?17ByciW7}?q4$hpxRl0hrp3lcGF;R7dy|COyO-+qC2`I<haBoBk
zo@4rjyz(=IqY?U=n`PmqQ7OZt-S0N_b$0VcR;<SDpp-j|k7R_y9ZN0-?|pet5vf*u
zZU8Da*>r84HJlpdZgG$*6%_N^5h0f%+MP#Rd_RKIN<re9UrS-Su$B-FIAyWtFa25)
zL%+G7>y`)iN4cAc(-OAorxqv%JB^N1XKcxMav9Qt3yDol+6&{S8I_CAzcD<P<_T|x
zbB2^NI7{P66v!JRQ~O~28mE6V(MKYKU>5%R6<MMH6392_#DraLXbp^i3dVer)3G2s
zbpF6|4YXTi(f|Snxjk@l7;9j?h3;D_7&BSU&#iSYCcLAH&{6p&P$w1a>V2@dNvqka
z?PHX`uYW>eLf#|c<*c=>RibB1Rjjoor>8%4KG|5F5nv<p)0NQ{rn!3ec@g7n^X|2V
za3w>f37XE%@SoZqkGu-yjko<h@H#2u$LWj1+m3kiTj{sYlr4%*EtcOD$#N7KT5A24
zFeJwE#@;iAJo%%VsZbAdmDcvoV*etAhV3Q?DMxp1Y4jBkYJ&Fa>|#!JBwf9*^9Rs;
ztpQh^KfmV^eXPtyf|!B~cC5^#*tNB+j4%>JZuc_t+uf}tzY`%A1CAPwDK5UoemVds
z10(z%dp8+8GgPhJ@GpTBz3$m9`OJRIEi6rFS7PMo+ntp%FX1@(0XG_VKwE^)D>8{x
zPQr`${f$E%eEO=9rq^7xIifJ~0#52-axG(H5(Ka6m+Nl<(XiA=|Bl+UrR9nO*?xQB
z`ty{7gM*Pr)F4p>bMnM<BF?+7Men`5b*mSW8!d}ORC+RI<=dUjDu4nJ6-t8=?)-!I
z(m-G(2o=BXuhuSiuQJW{51RxvS%_*+7*Kj%`)I!-HLx{nsm<ZmEsk?aKib-olag+F
zGt0=Mm^~~j2{aZepJPaql%BEhsxOkCK)0A=t}J9^Xlz_>kn%LMq=bZm4c9bkqq(uY
zy1&o`&Mql2ZWrl9J)~t6AkNFfgGsw!mtTHbQ^*~3d6d!87)Vjp@If_k^>&U8kN>gJ
z-DE`age7m1^+t|`alS~{WWT!jn`ItQ;X~2z{UX;iG1V&0Y}&u$nd4RW@1@Vgm8F@P
znN1U-0|R#-rjSJsSV$;9rqDPVV~7%W_gPt6L(}@}Ps3XgXZrXLCNgQL-K$%!8mchJ
z3scH_aDc+KVE*l~0;^2k$)m|xf;!2(1?d|%*1|0mhR^&=`^CE9T-Ql6d^C1QO!^4T
z5m7w_<p(ORqpk)PZuGLey(7Q2U1T>_%;0*(kS)pNq^qr)w!vZdIX{Xq?=__ytqZ+N
z{k~uF&RxkTD+kW29KJvOM^0*IVtEf)6x;UE481F+`hle_s4U?J44)z;x6F|H521m3
z5oym?D?2DFA*ey@2maQMnD2uw>GU+M$N4i}AAe&9eZX1){sYH_KX-_0Jyaw@K-u4Y
zLia+~byL6!%v0mxj;CohC|`+V+oBmhQtApx66`;I_pcuHB703`Lm@&UaO4Pr4`a>W
zzlG=4n!cz1ZpMi}*5}~YVH(Mt+|a$pH&0!2D;NTJ^{?F`+%Ksrcaxb~S-GGmh;1g^
zy$ju4TynYvzX2qqQeGpIBY1h^#qIaSd+uthuC6r1c~o**zc~|UdG4O~j5vN%VDpYt
zLFsZ2>)#Qg*UzJGc=jw6*2~2Or4TY?Q-CK>%>>pZEhUxl=1s-;9v)JP^7KwO@Tigg
z6@tbMP)vcu#w#Smu<)@@fj%{10eLRqPQEWbGNTjdt;bJ2QHr|JbtVCW+o&9=hxZ<&
zrlKN*hcWGN?A<4cq-io=@u@wQEpX@@%m}S25m8*5`czdogAZ{=_wFGehbnyzNH_tB
zn2TjM!Mx6GId@4`z2v7{+Kd#x+j#>W-kbTn^*TvTF<kC*PvTSgX5{baww||mc|%nr
zGxN`xbCwKeo3&dFNAhoG&tFb^^l0g=u7pfnr18go!D~weeYSeG^!=jPH%&Ti{F!?e
zFScnuJ*;@^8ME@{v!<$VUt3+JG1*hzf9kB1Q#j{qsN~-~U}R)8>n-4@Jpf}HL>CH7
zV#R)7$Dt*t-B^MvrgRv3KpFx9mQQ3b=PES`$r`Ssye}f3ryHIJ{Ym+H)A>+t%|iik
zOG}YzVZ!5!$l`_$5%u|;l-xl)&&X0cmr-kn!OKq^&kAJ(##eh@%dMiunL>Mur)xmr
z81Ur0hS<0H`7ox~PmSj11gzYmRp*vw%=r#hRexN<a9{O)vaLhQ8%Wkt-flrtz%-7r
z-uGo+?V**dgvA?Y&YcT3qn`VyICAo|2%m3>tnWt4r{veKolq=b!ZHdF%1Y&3Y%<jE
z&sZc*(<%Ud1gdUX7tWb$n?9ZpeTwY@ighFp74}OP^z=9d4sus$CCI!n=}i$sQG(?2
zyU<&kpKpDQq;&V!>t`HOlEw+B-;~3mB<-1A^PGiXf<gGiQjC{K&nPM5ZoBOpNk5Rt
z)DY1m#@;qCD&+k?TmXaqVx>su*O7GFa|9pzq`psk)g|i5wmoh3z%D0?`R2nKc~8HI
zzJC(HQsI$W)P77&T9mTSp{_;olGq*<9Yq~ImHZ@<Y;z_p6HA{%<6OB_*x5JDDam90
zafsZ2_rS6P0JH-oq%JdfRd~VVs0J-IMRk*)3LqPpo*4<_!X;8!S&6C=nKz)mmZ|S3
zSf6K|@W?r0@Qi6)pr*A|FTQUvGBkskYRLXxJs&A(6qCC^F5s%Gc!Z$cco=8!Qhw$#
zoMqWkAFSb!#74R5=38sI+hC^(3q;8LS2n-P%{8m)Ck938`Pa~cxl+~NZY0WsRCh4r
zSCwz~D{&zqjSEH|^^4KB)@eu=F=dSXbGAE4@TyN%Muy+XR<yb-=x(gBoS9o!S!*LF
z+%TgsDhK{4fG1>3=TubU0=RINK{P~tPzB6Z*>C^%l)Es5^dlZyC<wRk_QGMkq^9P-
z@*DHZT2X>7<TCgx@3rI=6zpSsni!l@v_;l+Q(zNYLNp*lNZ1b-_XFdp*a#Y6(J86c
z(pZ2Q^|50)z)r{G8etNM^eTu%yKYU9T6y@mM14Q=F4MC2k+YGW-?s4OnkQPc6u5Mf
zcb9H(3Op(IAeRfDI1wSdUw<Q2CV?YiT%M7><+^2Ky{67e{L8enr}Mr`Os!lHnw)1=
z(ch<JWN={SeHP=thg)c;`sPc?jU$7#45vS`Hv5<VoAX9ii>4tc4&Vm?4Bfz((?Bs`
zTlh?>tElu^z2i;*DNnmh932uo_#)VPlCD~V1S_s1bw$E@Zp$aIWf}5ReOWHRN{0cN
zfVSQjiyW*iwnuZ%LvUesCs%nQL4gNv(Gy~DDBJf=q@yh7$*nKwhAP4I<4BX|Smoe#
z2bchq#P88(^-A&38VP~VGS2{GtLfRCiAF<4uIrhvFNXJnQa-I=wlO7mGX(|3LPYjg
zWtxAdbnug<t!*(uk{xS9l7n7=qOx)ie7aAZcsc$R-aE(eOMqH5CqEI(0OGGzCT?WZ
z1==l)dJPQ?weP*;!Rg%1ZGWyQL8+lVzVtyd`fOHl-jxY)ai-U<RaZAzr;K8PcQYpV
z;@n4R$%r;?{m$Puk*qG!2Ub;;q~U|(TY_oNs6(mX-a@+1nHCN{5Uus=ayZRV9~vdo
zc94Vfc+k9g8*2x({p%jXy*b`m8n$<addEsk+MY|c-FQJ4A}$#=`J^L%qt{(YsdLah
z;_-!Dfh&c6`Ti38b*a4RBWaG8p7XqU$#y78I9O0ZD6DJl<zw-Yhtu2SrFMpQ<8EHo
zM~9{1+sb0Zi*?5b?!G}DL67{3yO|yW*m?A*0)md)J?|hYBLD;I-|opHAV9$5panAb
zt00N`@@3=A)su+ZV6B>7C)|T;dH3hCh^~0xj+BkL!_u)Q^!w-eI%afH%fF*HiL8AJ
zl8lC@WzY4tmB8vIDioWdv%NiAzy9mj-d~n!Q>Ny>ZTLPw{gAZa^8)uF+7jfE+ZqTH
z;9|}8wk<<9&CUDCySa^9xP4$rf8AN6w+4)uzJUP+cd;KWEiAI`ZuKmbX*rEg9UTg!
zt!DqVJ8*>llx~ur`dlkK?Hg&>KCm?gwbj2s)qtu#FE<zZ-#5X!@nn<1&@WPx%BNMB
zvwQV1E<Sz_BjeZLr7yvYXSr{-9KNPp2M<%yqssmB<t`@+t=cjV96s2lnGhWm)C9c*
zxv6;Lix>H?25hT81~guYjrIn4f<zJjjeXwNc_QzMUu$GL=21;a=7|*hY0vM|&On%e
zUt3TNqjW#GXAgVs0V?hVHbq*6&I134;Q0`9F)E%_%@cpW)do1#DjcD{_w0IY3?;uj
zzq6$;YK?aawy0UnH7-aB7b#2B#x#=2-8wOnk@#t?V>xrwNbe09O`U{f@1$Gy--Yzf
zKB-IciU~=X-ral*C8teD4X#UTT5;^r^eJPC;|Cq*ID|-CV*?!rcFhRyWO|s8067|-
z{0)Rh>qPE&G1mwP$TfsR9kB0;&F<U-qUg<zE+8MSa5=e*0EHRUD*}lO@j>4a{L52M
zl`Gjdob7mL&pPDPQhZlpB^6Iv#CU;Ijy96J^-7?E`<4g9U@%<9BAboD9h%Jcaha|C
z0oiJ2^<l)A4(=Fy2f9n+N7_3aM4GIn#{YUqOr6@hLXb2%nW1M{dfyNlyb20rE=b5D
zWQ?B8O87CcmrkcnT_C{q<s<Xl(^IlDVKi>gXGM=2YI5AmJ<zeJ<o6GK{`^$;a95Go
zhJci_iMl#;ABOy7G{JLCy})eVA}lI8(HOVYMH0@g4z0;nIh{))zf&Ma(FUZN_uz1E
zZ(02-Y+KynRCh6x(ucv{oKVM|Q8X1_zIYj}`~?`fR-AnX4;opk<WT7SQ*4(5T(wm@
zE!#0G2^Lj**?SNvA3lD3j=}HndvadvY^5mE^^j)G@0V}$9!V86rRZ;Y(dvEcg-mir
z4&!ylOI@|;zxUS*&>SfWcMkbe=>2nt#k(Ms_6iKoJSiy)Dfie)ofHl)zG3q&hn+R$
z6wi|LEX}U%45i1mxuwzucqJ`DZ@%;0vf$yy-~&L-_AfSR=q_nURhu<Q4Xf;(4eK|D
zU}w~2G2?w(?~&7ZXezBG+2^w41NhY(;zP>?xB|mpl|P@pe)Y{s-0a8*bv8sw?uxA=
z%iCUHc`@GMmop7&TO`9!#}R6#^N<|q2)+%4&Z`a4M>2w72i}3XgjU=X;F0_p7`Tz?
zh?bGYzWOs^;vym<O8#c&jnHj_EaGK_6Bv>S3oQFM*nxJtD`zeQKJNHD$kWG!jN}Ft
z$HLu!b<&zP&4H6Iz75yb)I>n1Q9J+w(6JQ1xiq3{4gRb<?iXDuwL$DkSzsI23W!~&
zgw}-A-v|(M1&?1=UZ<R8oCi!%Jxaro`elq=Yb)zU%Hm)hv+LjM2eI?Ow3Om!K@|89
zI}Zwn@FWxkj=CI{lw1XKuum$%ExS|?0LcCWZ!Kofl{q`Z1BOriZ4w&K#-;9RCfXM-
zN}7HqbH1#vtemXX?*OXFcmkS=VEkV;Rg?wHf@<M6^g9O!xyB=sOG=`pN7Y^(m3wJq
z1UFmsQaeoOaE+i>{^IremZGQM+l0@E`mSiOu-5*lGwgh|)?h=4>}_IeLBxo=bFm*!
zWE+QT!ANTQg6!Ti7e@<zN#A)#*QGyFBD&-pS9QAb^Hirtg9OD1cIngc{#6!LImsCd
zYe_?B;XiDB)2!9Z*lBCi{xDnZo>*KVAML9wsu*9~<j||Vxvz?tF~x0K-3etYGYwkU
zsq5+J9Bx<!njmG>#8IEVj>k_gaqje5BBjHNmoG<0MtoJJW1~}MzWFp>-&IjnMa@6q
zfWio<mv5;75HB#us$H(YVhUtQoQHyHlZPk^(=sxy(8b_bp+4(T8bphJeEft8q-<Fw
zo&8EUPMjz};Eqeu#vZ$kI<rrAIKlDOEl`|<=7Y3<WE$PYSp&i4Ek>>&Rt_0>p!nlE
zlu=k8`@*QLqpuGR%dxxKy(s-~-eqmtk76wK@ZkiIWd5Tgi!h8F^-0fMcAUVa_k^u0
z`nt(NJE*9j@E3N})br6JMHQ8=h<c#de|5*GkmCsJ5Qc;>lxjR693Byom*5Xx-ChZ`
z5#;{k`|tIuO|Jgp;pIgxT8qfwCd(@A+S1;B2k?Z{u2hKL;={Lo1&L)RP=mfpf3a&}
zSA*X!foX~bBM9uGn-@!}3F{joZn;AsEJOVPR<xnT>Uw(g(si2q`K}m%ccY{8WiQ*l
z5quOZ<Q@ajAkt~k(ND@rc{$vxI-u`hB%8;b?It>%$WwbGw}pj-U_RP#wg4HI<eV%9
zcCe^H{x#@4H4`cz<QbqcJI{Z@OkUc^2@S8S$8t*bfX|`jkJ{FvL*O++0&t#E*ZH)>
zU6*pq6meacn}W*o=FJ4=%aB<kB?mN6KqNqdwV2Ga8?W2~cKr@YO4IT})c%kX=pljG
zi|!yYLg0BwJUTi$FkXU(oA9nRC;)`wWpIwh|6t>RFIS3NvcVi+Rgv?bb82Q9dr7Y3
zsa#q*Cc!3Q4cq&8XsN0mv&shEVxPAief(h;I)%@24_L%4E~}sC+|HgS5?tJ<+BV+z
z>M8dgjgNdTuQO7{uT6^Q)|_#ciKTPT-AFz6RFL7INi}c7_vXgG)<dEx{cO(Z9ZkDl
zhc<IW<p0f??4NA!Iwu`^{AIv9=APtxF^xlU0X;H;nS$rFb*(eMf6RBk&P|=1nPlW*
zsQO8N`;7QCSNvV*Yl6=KH5|X5C{TjH1-9?i23fDiV<T$)2C^K~!Z8w#-fTV?+Xjas
zJJ=gl<8iR?P|l8W3vi>=izzej+NYG%)X)+N5!ma3Gw=>G^5|(3!#{bGC8#JNzm8?D
z-e|IPIFW&?lb?s@xhx|*wA-M@PW*zLdYowtav=I(vb!QIm#<%oS^UOt(AO}+wcXR_
z=~Ktsw+S9LFj(=Jbc|LZ5}`s~M+gLXp;F3^#_UN4R7A5FV^WxuC`HP2^lWuVXec))
zr>O2zCHO58#{;w3O%wpo<_S)V?@^@e;|I6xz^#2H=DX;7TmjK1ou1sQla-W=7!Bzd
zz%WQLz8jd(@t5e;;WZ<bjYBa<>&ueeya~xP@3`=+BAj7B)|<zGQS}DFsYgzUK{tlg
zF%&5O)g;nYQ-Yfyk(6qYxL8H2vpT?=Lu>MoJZWgpJr|c{9O~2tEkufut|L4U5JDZ!
z>2ep;ZlbJk7D2X1Xe^*9r%0m2spYV5X>04jpwy@Po7oplUSW<9iaq!P2%UusW#8~8
zX@+!xtin8%P9&<T-O!6Rz=s5LoAGf{O3F1zvS2KZc6m=A)*SYlxO=hK`z<s~O57wM
z`ja+B=;*h<`J|?*dLF|mU|f2#0R=<ZjB_wHHdeEI3B4<{!&dN~KOwyYbpJp1k={s1
zvRvLPU~o`C{(R=v6F)zjlshbPv`C;z!8%LKyM@vzNj2>A!5=8(mgWB<LmY0D`(Sg+
z7mO%d=)_>zECo>dN}=!p(!zX~qTB&=)%rhksq3F1JAx3W++K{X*A)}nzK<t1(eSWK
zdv=-V>8ssN6V5xQ3V_g}qG3)|Hc#ZQ-VDF<YTJ*MH=%a~O$BV8%a+Hz_L4uhyY5PX
zvVCaEuB{^qO}-BfkpB!np3eTZ(P}1mT_ebjrsQ<YNHSdsAJqRLdQyX%2JO6~Fj?m2
z=00#B3Y8~FS%EW5O@55R;u`0+d%luIT=+Zmb%Y*b9%xM=rV{2@#CQ+3D~>>ARn=ay
z94y!)>M+Kd9YArol6T`8S{!S$$AASg6||HOO(DAG=jQI&&_GQVaopk?Y8@;(Rt#gB
zS1_fNnQ3fjxPlFX>`v>o7%aG$guK3@vDVhM2znQp92A|foMkX+N38qx>m#_t7)ixH
z4V)9PYDxs`Kouc%hI>_uddChBx$eyl09pk$3TPO!IQUovxw)A5Nk<G86m$nC5~3QA
z7%9)N=9UE9L$GP)Uc-FC`$^SqvSXM!!a3G9kTL^11U6TfFg?o|rLzEH8HC2Q;vb(;
z*3%P(Rz_W&|F;WtF)v@9Hy6`?tr{b<jw=e+@7|3=pj&RIp$Tzx)N?kg_SwynuU=vU
z%rGM>%jCmc4}SwlW=P5L;V`0`lbzkz+zd+(Hy~NK&H%QiDl_iif2r{1BU$H;w+Tw1
zT7rc`kXEow`}_NmA7i=!ga_mU2PAGjv^GdVQ{Tpra3%)Y<2KU7>B)xls0%~^%qJCo
z_+V{i1-+Z#ueWj00=&5iI4Q<yY-B{m=;3~x8a#2ZVzBTj!y%Z8Ua}{FN&z}4CUVlS
zDR6TDauGdt40nbLRs-7sI(~=@a&KyPdYy<xf7SwD9_WFj{w&Zce+<up#0F=VfpfPo
zy^$MBA`Am6Z*TWpf*qRuulEQ7rVxZd1jKwgQUfs96j5Q2lb_eu4^9V<b^}0T<i%HF
z_Qt?{JL==HMo<>}`O_8gf%q&(kEXzG9wbCqhT@pxXBoS$w>%?zLAXHnIbeO)TCZ|K
zaIk-Bs-aq%p$dT<JsU-j2A?n0)jX&-0@ZrX5KZim5Jgf+kPiqMZVTBn*%Z7WRn(tg
zXuQ8rJ~1;kkenpYCt0{=zMS*WyxAzJ#xg50dxBJOV5_swr6afi$C}!<O73{E;w$sx
z9^LZI-DJC?81G5Vyr=8TdXu#-u{4yIP+ws6L#e6ieCv7Sd~63P)EcmjKYunuSO9eV
zc|{vDm}WKmo<&P?0G9<PPkYD)3u2M{64}vFHnvj(VXbJbASx-<o1=8XE1`!OX5J<;
z-a4@JH3-yir$XY3<RUP>`Qs9%8<r7dpsMJ}1_%(H$MWcKh<X+~4yhmG4ykr;!6CsY
z^3c?j#(HAWz10&MYA10Q)N5<f*r9{zcjp(jA@lS{Ek#8ObiDi&WF6*ULV5C}E0Buu
zrmK-Sn*iPf6a}MOP@bSJ!?4ON45%MtjnBwnp{1P$WLop4F4EKUGupJsB*biga;TTU
z#I+v;4++OyCJmU6t<v2Zzwn*jvs4Pe6tuM_z$5#*?>V^IXr~YGF8zT%D%>n?a$OD%
zV8YhMow<P;0Q;_t(XR3P_d@3hriS;^2-7g13+g(96rjE53vwLDn$U+uS(}<3vu;WG
z^y$>UIG_4f^6dHZFAEDLHk8_-y3(HidKm6;bq0)iad3G73H8aza-x@cGz?DgxFX97
z$rNcHqH%>72v>zWxiRQ7AY8a)3c^#a91jpI+$)tUxWkqsxNmT^iHUo7Y=Gm?G-L!T
zjhpHrAwP*(Xira$H(Ie+mpyxCA&=%8b7i8hZx0<^`56XVDPVC#g4Z7)i#v&z^hh4i
z9ze5=w$;Dc!c!&na%byvYjQnk)}5+89T@M9eSwB1H-!c%ay1=u0`nbuz$9jz`(1vx
z=&f0sO_y(zW~iQ32o_TBZoKW4vrn0V^4pj3$2x-QyJ=59&6(xT<Xe#LE&5?;E1T|a
zCX-0gSr2Ow?wrE45ykHo*x0aHh8mf$IL2RmDV6ok;YJ4s9K_jKhG~K*hZrP&{n36V
zCRn0e#-HCeu7cnM#?@h`bl#H~?hOoafNvS>c!QolZ%o!;?T86{j1lZxw?v8-j;iSB
zID(i>$Tp?y@E=!xgR}~BqLe#!yv@n!HK+q$7=$BG8a2#I&n6;UKfd!4Oi}10Or#(h
zKlz)D-8MQH3Rjo}06?~^5rA`qg!@%c5PQRA78Mhd4eTILt)S^3v{6oUvtbX+juCv6
z$e#x<I;gW>e;#xSm)a060F)~(&bf#sAY=MHdv^8qI~yB(7*o1JM?}Y|-OJmUF-GF*
zx`OsXhqU}K;dP6v<NbT_V`FDdOr!vYyff7;Eh{Tqe>)bYA>gt;eM;T1PgP9~<FSyz
ztZ*T87XX|+RAaJO!iiu`{PPsLLgEZB5xt$^-o>Z=;|@TvTySnp8E!%p7>APk``k%n
zK5)pae_{wQAH^x6Ze^vL<vr{bpyB?{o(*~!<H2Bkx)I<5WJ$Chi-|u#Gjv}k4T->t
z{#l-Rs({9!t~-hG@wn+<;!apvS>fZGjVeIqOH|v^w~KOs--0(NA|c_@SMIx3u@~o#
zTP=JJ9U>%K+o>x>yHnb-F{Zgc|G0>#=<a~vfb1bt^LoEtfifjp1vBHm#$fsWT*X^I
zNGuu8E)@6ucrqLCXS`^1+k9w~AdR%d+v(yAX^&`n8RlCyjcBM!e6h`+&o44__Lme>
zuY7-DxoVbC|Eiy?7|E{Ei1gmr&Smo1s$_}ebXQZvIg5?4-_-Zl#Fz9TP=yv>N@wI4
zbi_2reEjf<B?5=q{f_g0_G6Klg-YFfV427fFhdPa9d1e#MX=<q_;?yFdM#jq5uU|-
ze506rhY}V*Dp018^l}dU1kUcW#!Tr+bdBK>gSldiAK>gNDsB^%*Np#~1|b7L#<gIO
zsjA|^g6b@U6hcrCf?LhszvJdckv;*Nc#vPSkBE#@6Y@6rBBnQxlz`obOMY~C_*GaK
z1Ws3$m$f=n@WWTWNIH-%bOeOQ$M+G*pg+?*k{@J<=$yUBNBa2j(tm~>Q6ytDtfRxA
zJ?qBG>iLDMJ2_CI@aiPtHt>dCZSNPUA;r0_*l98cKrFy+-lA_=e|Oo^4sQyghCL1@
z(qyE!$h-iv8yOgcd3!6tC>PZY^hqHeo|cgCC2zMU8!>v@ZE0#5)KcX6ne_vPp7BCU
zp769q*^3MSzX)SpsC^LeVItKxIcaBWdnC7MUr~6;yLY$cL!K7o<vsQBsjjJ^rliDo
z<EK%Pe{X6!j@ie8g30d*kBf_?H%4N_+l3Cu(Xw}XX>Kdv88>D$p$GU+NF*8;pTzQD
zt{p5~sEHwI`}FBkd;5DlEU^1S*el*!DbpZZKsN*RFlq@Mg44db@Fp|U4tzx@g`IYa
z9kjHw1G(QFXAdbIm{Vv#FflWGpbUmmz&PN>LUkl*Uu7Q*y)af*9)ju}h{7~9G@PXm
zQ8uKEYt3+{BQZFRH-Zr-ib}jtKe5b%9+&m{wFR~a(uzNSDm*qP5yo*LArJTKh=O{b
zns7MdJYml(g=+`d9{j?9CbU5F91<r8qXdcOvo&@_i?aeLhZx#bRfV4@356Vxe&k*F
z9SBBJ08D1`t{Gjr1fm^={tmOV<A%k(4<{kC&G1<5z@xxa9njI{ewf_(r*ZO`3sZc9
zI_v(9861JZRHxY2uPYA9EqP2GG05eRkO&j@RUHtDM6qfDc4>C@Q5*+U!)T|2RS$d!
zeHTO~<jc_VjpaKuI}8jrVbCjkr~cgSN7W&)mMbXeCMHhaTttw^bn_{9+G<u|{0w$D
zHV4{8>;VjSTXZxvFF*!mBjHYC&yk=$tHB#k*;Lv~m?750X-^CZ@Cxx}d3LmNzmw8$
zU!fPJ&oRjHd&J0Trgq<_?#aZx?As0Yt{?K9^BVIs_0J2X@qOd1DJ1o+G<$mJ*irW|
z<i3kBimvN6v~v->TX$P;YX7;u+<sj;J<`wv`d=PKWvg2j$R$jk>{a0g$!GQLqQkY3
z%kj4Hr+5ZR8$9MGTSF}MXZbg5@4YmRtnW6gsG?f<LE>FgR`XpxV0_W!;!4~txpvPt
z-!Dpcy8e8deXK=9_q?v4o5`yep|{p6oSeOM1DWXxNEn2jEH1H(wYqG1)$+Rab-Y7*
zW_nzP)<Rsy`wvS7y?INKOQDKFuhq-Fgo3sdt%{2H=H?!cF1&-COz%Nz^EmD1CZ)$G
zjw)ZPkx53wq(KWH!nMwVBhqpxeCCh=&^Qfl?SowthN>i_4NhIWxQaDIU1ImSILHez
z4!Q~7HJnhUqW<`yspeiGhV&0fcTOns*D8!SyI7WN+Dg8IP5vMrD~_XDu_W6*TU*<k
z0<T1`?;EP>ecC%PfM_qj5(FY5(qpk!l3NsL>$TOj=&Y~cD-qTn;*?}t&@f&3xEa3n
z$W(m8o_F3CmwfeaF!5B@kY%Xu_C}6x>^0p+rb_zy85hbrZtNP^MF=~-^7Hpk*4T()
ze~)d37K~cHtVuGUzvEXY*FY#RaXb2q%?G9b(b=8u8|$~1Df&14PS3S&<*N)5k|Zs;
zn9)_)G$;8Be0A`L%F;uva&w5z&hDz*yvc1d*x#FeXJ0PcqV$bV!=CEe4uYPr(e~v_
zKg3R~(xL(*P`eQ{p-^JJ;mIV_#`Yld)IcIILb!|&>`~p-zw{VGP&soZRtO3U_ff~`
zOavUo`AXW*mzPh_y34tBlUv!pw`v5<UMZ#w{7@)vC;g1cqLn_F92W=GlFfnhgC(0V
zdT#E=Wn{2v6WJvd1v%Z8m5!2oMdKFZn>H!@EU;x>?;HeVXyz>XfE-gta#JhNwonif
z%HsOPsy=Slf`At4xQk}feIfD{iA|esM4WW8#)YP(wL5<RC@SDswfE;t0wB`_2vldo
zDDSF&b>F1PeH>eaDA(OH0Cg}BaiUjn^QX21Y!Vc0uaz%?>}9N{SJ8ZF3lr@+-;2V+
zWsHb~->`fLSzfMt&w;BGc7fqHAgoVn{`Yq#Wn{QPVKAxtX%_LJQcc!yOAwgi&-K7v
z2o(dDx18E*#^LQH)!NdcxwQun5o(DDXAkjX9~!o9x-UTdmirCrm~_OiX>Nh|8^nYC
z`r<Zinyw&vqTbJ*?R26+V9vI>O<w*mfqWBNK6V;Gup^rhRV^Wz5%|c558<D*2sK`B
z?>j6aG}wR~rE5dKQ#NlpL-TrO<0;b6A1y_Yr`u%0_0|U8_@CEv)^g}`{`WONG4o%2
zLVRHvtg8R{2ZSzs$p60n|Bw8ytk|EV?#;9{2y1WNyoqoXf`tixP)tlip%f?QHhHKK
ziySfa(jAFvnr*)+F}35<(1^}_a#HuU-(GiqdB~$Q1t~yh3c~H1w2X}W4^eK~B%?7;
zKac;^is4!m|9?gQuw~XzVgJw5Ws~*!A4e!C9q;P<ufr4mdDiO=p}_v{b$^X#JDF!v
zHwD{vPulHdz{dam)_?ytZ2{n@xwX}Hw0M)}ssCpaMISYRgzkS{|64nX#G2*D7C)gB
RQAND3ilVwg=IP6x{|5sL%)I~r

literal 225954
zcmeFaXH-?`+9kTxQcDmqf`WpIh=70yNRVJDK?OmwWCcm0AUTIpF%U!~2@(V)C&?KU
zC1*i`2uRL3=bp<tXY{z^j@$R^zJI>2Z|_s1v<Q2z^}bJ-b3XH3=fzEF2?{bAG7^bI
zaqX(89Er5Ul|-VH`*SP4GU_Ctgg<s@OG=26Hi`eeD~R?Xk&ctDiT<Tv6Fk;oAM9_w
zAu<&hRC{C4$L&<~%RT%G_ht9J@H|5n{`lj*JT=)ptY&9Y(k0(E=`oWfo^80}YPOJg
z@W-B1F^L;NFB`dMS9R#RR;`@}3kO5(rCX}|me)on%pH^J=0jIQU)av`pVHIQJB^DV
zk^Vj-|8qO>mru#h1@{ntdBotg4S%@&D(@x!^ElrOAMxj(FaKlT_EEt5z1n%;wA7n(
zcP^VMZz1`ft*EHza$0?W|G1RjXG)5!nCQrJUYjoKN!m&}qw?Cv$35>4(#$72#s8TP
z`u}#B|FwU6=6+qMP+-H{SX)}Q<f`TLUnG*RIQGd!d-kV!oft(E4Y$F~pOITgHx;nn
zc<nuEdzAR|4+*-R#GjwN%eLT;OAr6`?cfhR?AYrQCr*Tmxl^sL%#Z%a7;TWC*T}OP
z_x1IqpuLd4*l8~xExo?6I!VrFv$`<8IzQSHe&yLy3fkrvS)I}5MEvuXOTU&iIX~5M
zwE1%E=2f|PrNjeYxQ%~i8+YVp-1{DPD@EIt?9jKyM(fGWkf5O4J71pmR|Pn4EDxFw
z)^?=n%a|&!Z}J$_zcXq{Dq5~F`H}fREm}H6HRGOQ%*{JxFORinnP_F+Z+o<ze4yrS
zmQLBrM<Ow};B7ni2u|kD2Jsjv<(Tz9XwMW{8G2tQI4kv8Vh8Ev!$6WR-C~!sv+1Vw
zw*7c5jI|w@S4-Eg3zc169BN>=xB2(Bou^No;x?%NHQ%CDH{8DQ>F;fe9oAYN+S-&(
zGZjo6DygTUKjW%}EQadO>zb)^PBqEPh!vzL#w*FLw#3NdJq#P8Tvyg6tc$8*L)8@%
zZlxTj7Y&|K{T?c0A0-tiSLC!F^WJjeC;x~>R+p3g*!ju;jw0M=IqhI=$lFFXHnq&)
z3lCKB8hf-j(Qy&0CSW=8)y8gYGM?-Zi{ScP<1<S7TC4fk=uo#_il@Z^!NEt%C|OC2
zuJduEOQr&|)m#A_T29OT0fSB9mtD!6R~o}YomPe&c6UZf2Gogc=3jZXhu`w5v^2gL
z9W4_^$MBj#Susu_XE^GuQlhG5gBZ1#rbfC!L!}=pL;Fq&nuGSO?(DfScJm|N<Jf!8
z`%+R;YhUO$KHhIj-B@1@U{z*cDEDJk-XOk)?KQ9EXR@k`VNr;g8hC3We#7j*fdjLg
z$4OI0z0N{gT(lcAAGEuYQ#Z#ebF_rwVhZNINzQUFwdscmSx@|oQ;Z+@k)~DT<ijLW
z+xROmFwk~n)95B5utn285u55Aa9TR}a9b;W@UX1m<;#~RI|~?St;brYo7GHojn2wP
zReK!aD)pod@b!IGb8F;#e3n5&1mf}hhFa=cKZkS9R5AIi9X06^QR>cIxbk%OHEHRW
zX|p<BLjKX8ckVfQD^@O&?_!rl;|<P%A+gtGVu}c-qRsV5Z}YF34$~z_h4Kx~sB-M?
zbZQm5`+aAucH_5isXQ-fd6Q(qgl&^~!yT4qoHtjy+S3gWY_fIaS<|1peMwrm2357^
z1pamX84<eJ_@Pqgg*N@P@mi`c$tS$pbIb=tHm1GQjB_Fp*An!O7V{(DeO|w2QBHCg
zODVIO>~#G0;o440I*<H#-70_Yyq59ytVvgD&F5o3^Q<RhSmh4v-_KsS)Dxrfpild|
zJ4fMCQr?7{KsnY#z-sJPf!+LdKb8=?k=Qz+MfJX&%D6R^G`$j$6n2>$Hqs>p)Q@*>
zBBP@A&<n3+_r8q3mD2NqDRil&XoDv@R3q1-V)Pg7h5OmlrL?p>_Y@Fe;+`}eHoczs
z;`1MAPU|Zz>gG3szv$h)+n4cL?(XonNFn=$1m)yT%NC7AlqJWd?q_c=Jor&yzu5id
z+2Ul^<`+tl{Mp)z*oWbtIp%!|8BdNj2X0`aUN|%ow>_TGBA$3+#32j!i}W~~&&bI+
z9sZ2YO`!PQUyoN%gSwy5PjnWnB3rAFP5u5@y?+1lP^M5DZE4-FQ<K<;O4j82tsn28
z?z|LQXl2*P6`cNh;G*?J^(Y^jp6oOo)0Ru2d|H0{BR<wD*Hl+aUAt!3k!zW`;k;a>
z6=CIsi<seFYSD7mKsCF+j*TlosK=}auxkV*noAGH)1j5g7Jaj+Ri+5w(P5Q)yn`ZQ
zp)Trqt+HoVf=b%SutHRZn2^ooI%jdKTK{C-SoIvUCJwfltIbIodE$;cH#QcF1{G%4
zmS-7Ooma=R+Hy;l3M;Ft^H(SGW|x%yYA`EgxiqZpbd+>6Slr|AG)k1dk;6cseuNc@
zglXswBEtF{BEy_FY^Q#GmR$|A9SB5PzSQqyuy;~lYfex((AF9w`@Z}zH!A$diJa(Q
zvq}~H5WmFf*OKgr`RkdbL^`g`RJG9{mgq7*Vg*k|2Wi<2N0yiR$OtdD>Uh-*`<L&d
zv+eWdySR-c`Wg+SL=UCQyZvv*Q>PL%3j{_kmj^G<xR1mq7p{z@)kSEkrrxbUvYu>5
z7}pBUR@>~TTRSvHAMonc!yksx9gd3~94G;UVa~(~A-B@g)2Ez&)EEZcY7#W=E_Us5
zn7wh={;nX6WyvT41exxg$T_lq0D+%;E9K<9JG*x6B9=8Gvn+>3fQ!p~Zm3~;?hjJ=
zi{JT8PIZ>gV(8{wd!U`BUOt9Fsgf!5w32L^9Bv-T#;6;;N7l+~dn^6f?0Y?VleCI9
zaOE@?t?qXfI@a;^dLX{PW7V)*(;hd|^lD<p4=_o4+|OSbiT~nd93vN5$rzx%zm4sl
zMLYJnyfoKxwDcAo?)i0029HUn;63{b^v-Lh(rSCRlD=L;<30EmyE0&5xVf?J?fK%x
zi)qRYq`=qgAo&mbEwx*cucJqzI4l;Eb9_B=LtGI#7Bk)$Ez^}@q?}pGq8K+f-p)H$
zCdpBdjy-EzK^fNV%(wNHsgM_?RSOZY>=|BI9t@*yKCs0_iwzzB=6&s>;zuky)kDWR
z@_0+XJFYH>J)jJW@uMU%s(hUBGcG*WYFt(H@g#m=gga}BF<w?iX}wHj!#gxIG%*Eh
z;Cf4K{6|LbaHeT*>1BsAq^Z$UC}VClpC9iu&sCV(c0ju>xN`XC)25~-n;kOOue<%w
zOjOOVG3H3ITbyw0cBQ5oW*`dpayQY9P{(gvy^1DOI27T2a?;K2(t|#QyhqQ?nT><H
zPLN2PfSoQ3@;;|;%$V|&`W8I&CTB?tw)l2!`lqJ0hV%NoY`gXN56j7d#ZapshTr29
zS>>aodN%D)TxM6E1Ctc3FSL7)BMGlxy}C4R;;fdUU2;D_cy%l-X3*rQpv}33v*i)L
z3^YjZegXV^@*YNM@-)pr{L8i%ZEmbii41T*=7@eU&?ERh?dwQWto+?x+$xIZ&u4UY
z23Ydg*w}7;mA>Jx?XRN4oCH!+Ou7ouMqk*a-fOBbRY`T!R6#rSdt)s-*uRBza62~Q
z;avgrZy=KDm;eDl2xLY1!ra{4{k-|jnJNxh2?LazyE>jUJS&TnVbLck>4lGG6rhvX
z(nQ}>z;dGGc#peNvAGEdv7I>4osd>rS@w9KE^Ko=t4q&FN?Mxk{N1Zr#vQ2{^n4F*
zmwFuGr^vYd`B8Z(;)(Zu>p6PYEu=rLAO?@75MTt+eBYDDs5zM1P_Ai%C>)vQbO2WS
zGX_yvQ;(ep^7ZA`t@K4m55yLuj*C7J*;wkG<(3YAu6`znF}h}^mVYpKt=|+W_&X~c
zumKE@%lXX*zP2(=gjjdJxKA|X4(qOC`ixkeBf<_#NO)>}XOhb!9~2LXgyuvwqJOTh
zE{crAya(v8ZjzNCwphzyS_~_Rw<e%K^UQZ;EwS;os2gvsyBxC+!-rUv5Vv|>bV9<w
zd8|p<CzB5iq-%7bk6SgX==WO;iSBF5G1u=q5{36bweb3BJc*X(&iQ~%8TsTDZ2yL>
zr+WXqVDqcx*+HU~q^@VS8>5rm8l#aG82<2#-f3xdvd9^Y7ce3~t%5{yJ-t$7wUu=A
zNYI4`G&+kZ57n}aJ$JuRN>a~3zXntjU6^c&ZTuD~Nq$RHdh`-eL1aYMxz&1!{^#Yq
z#t?d^3+-}={Le_ABS(&eIWDO5)!G3-YJ9x?xnQxAXlBM;h3iA_y(YfLx0HJ`FbvX_
zngHpeROR))yr^vrOd;QNDE0)f(R*>vpo`YVnG0XPeq{=CNCo%+#9PD1^BVuO`Ss~<
zfZ^s?`Di3MviBm;ajLY%cG91rD2)Wp?*?WPdid*7W+}pOxG}m5S1C4Ek5Ye1NJuDp
zd-!L2?9Ka73sVp^>bn=z6|l|u$Sq(xow-0^N4uoKz{2@vHQeJIf{@NOmLLm1GFWf=
zkw{_$nfp~vgiyHy8xu$VRPktX5#9W^rRiRIIk}#rgVSa7UTOON$1Yms<4tUHBd^{F
z6|@E9F}0GucFnfmPkClNA;YLeyknhBJ=mNUH&Cmr<;c^8cmG3*2;@NF<vyP(_Pn#v
z+YS^HQ1bCk%G%)(1iVS^a8wL!ylYi?`C)_|gO}jz76E`~IvMZJS0yF$@UvMaU7<#G
z=s5tm2Ds|&wm%zXM2_{20C=VA)dbm;R+N`#qtZ|2jwYcmJ!;xU+Pj;`<gY;@&Kqc+
z^MI}WO(%GG=3ic%$lh2PC303L7zv<m)uxL@k(WM(25*l72yn<wFP5Z|Hh{{$22k?W
z%ig^h8w7qs?0V@?n$2_%S1lg{uvIjO7DuT?J|9z=Zbfyq1K0<)oy+jshx^a^N<RXB
zU>W9vjFMh#3MwI&Sef5qNe;T26s>1s2f+9zcv{pO|Dk3|$(oM;otjk#a-l@f)bSCE
zxr-^dzQ}n~l0E0cjkgypN1Cq5$aJL}Hm2%U=~chsH0dPhLVsSl7ZU%;0qN|KxcFS_
zNrF1GBx{NA^XDAq^=r>IMeHBTvluq3;b{p5{s&V8^XXYYqryAju8^l<p*-Fq2w4OW
zfm%*-AR{^GC2uMxYYN>hf0Z{`m?s&)eoVlEZ!{r|V6Lc6sCW2+0T4bjGjo^o<~k}-
zAr_fY(mz2x=X}fhFo{H_p^eUN+CsXNPiOP%Z;|z3nG%idzGsE(=AH{`a-BGFr~chv
zND7Ck`BzNPAOJq&P?$3p5bl7VMYpa$`gVq&-xh2Ozek!*)jR|S?E_wln<`%rWK-Hs
zI-@}_<b!X)T<ENSnzqppEdpG!h_7-B>CZ<<CKqikySbtAQY?IJC=2P+J|Z|T)uUTm
zV6Bjr1>l)&?BgAIEv3%rH#N&~h~U1Q!O*<rfwww~M?{>~YC|qQj@?Omck2y3e)n1m
z7*`p+bAEO_iF8Jl&|j`UL8x{X*iQoaWRq|C*H_)Tc)J|TJ8cIOZ>2>1{<7Hu!7#G2
zI_+3FIa9LZkBcneM$vOQiq_0Pd(-smgaL_6mH!~^{ezGMwj4Q+EjV$-EsqR$qF2B|
zs005`D#8DYi12^D{eRU~Uc2QXNNZfSY@^{dR{sfgYdOn)ae}DU*x7dm6}_aFN4M?R
zZ2-JU2xWa;;I93qBniP$MM9nVpOxPJlWJP!B`Ep_Hd{b;-1368LQ34JnKrZ_5T2Qc
z_;uC%3kC0!O<%=WIbF255{-i-iL(gI?Ww2$L=K?KvZ<!WKz;y&qq%V3W+XNm39SDk
z-H`a!#MW?&iw;Xu$E<(u>-+}nLh$G8k#L1pggWAVb#c-eT!RM`5RwSmi&01JyrQxO
zkK@W5U{6l#&a;0{p!t6I@S&c55mgKCrJb<Tg+WbjLW6WX2>q}Ry}K7}!8g%ymOxV`
znuQL_U<#_hOVEvkr#@|?py58_e=<D%EnvV*f7KT?CkOz_aSF`$245Ut+H*{RJ5>mU
z1{e>@Sj{lACoS&=8ekev)Q1mE5RWDqeewWa763cSM>W9Qdoo%yY4c>n#Y_7td?Y++
z{MZ=U3xRyb+m(~l`Ry0RfR?(<>x7(_zR*K%c{O^4-0{<8Uw{8hwBXIPL6P_`qE8>A
z_ur6_sk0;++WdIC^~&6Exc?C@y`QFK$MnM-txmL1kOs=OyrM*JEF8~l|B+=vf234w
z0jTgBE>8rXbNgP^`3FDJ>x15-7w=dex-9!+H2UL*$Vks4T&MFRu}MJMm7{#X#BHVB
z&(k51k+WytMoCFwN<f`A0VUH9v1~^D+)DTHVTaS|IGB7sn5M+bqx|SPkfx{}Ms)%(
z1}64*+E22TyO3TTKm^#P0ajO`ZY4Xf8^!d0Js@+GORt(*%OM48h*nQf--ZY=G^s>j
zD5YnR6%ZGLb<P{+oPN6_zxy)FZ>)B0KF#|1@C(KFc;&jXn;;dcMNW<<zKdN;$D2bU
zvB5i-E@*)wfHxIkfj{jyyDTj<-y|I(;F?t(CL+?}Y`fLP;T2Yi7k5fPuI5n$iv*qR
z-Tq3y_Yz*l?U@gttw7D|$~NttMZX!24%?W>n@lQP8A%8Q();q1g51XcsP8@e-FxuD
z?!)JfJHL;_64D9feiD;4!=EeQfz^#{AiOmw##?DWDm-7?M=RpwkfN~lllKj}otNmO
zLImdf1GH-GLIBOp2dcdpB>W-3-5W$_tbN1T3p_`Wu^WL%uXg%#awGV1TB4HhJY-7T
zjsqnher0nxY?Jw+hOa09b_oG@+MJMm31Amsq@yj#MGzbS@6s%7*5_LWs{;>wa)}g6
z)+~TR|3z(}HpIx7ndB;si^-Rhm;bWZd<0w>-!>~S>iPO2<f3)0T>;=p*XG8em?HgA
zl~4$~_t}&+ZFT?JQC(eqlv7*uFY$NJec9A5aP8sTV{S9oAOX-6oM;NTiUfav%_wm@
zgf<Pmsue4IYNitz!RNBY9l3ibh}UGJy=%j2X{sBPZ+YqqW!8gsrO4}kk33YMWnrf9
zEbH#wyI7};+kRUwS)p(C@Yv4uK@w#goi~gQv1-w<E-_cu*6#Um&FAv7i2_Jj07tBG
z`^m59=?&OV7PO#WBgK_LxR5f4`{iEx&1?jD4}0P}IXStFm6jr+ivrXjqBxL?fR1Vz
z_bjoD)xrF;(4lN;#!(w`(b`d-G!PhleSHU{`R;sj*)naCo(>L;`ak2wnq&Z3#>~tN
zHw-zg3CWEOb(;UKY`gxAtY5yWx;hvLsy_UR%*FBa_b&zFumngKNK^BPS)IAycKQAa
zmylp#C`g&55SghSwt~!XsAqq4dHiG*%QFeAf}(rDs5t?5gJ_MHkFJHJXEPibAiU73
z0~U4W*Al|l9vjgPjR6w1+acBq#y^+7Npe6=xF=YqtG7JYC_`^I6b=dh7?dtBQ5LO2
z2b3~3D8<QIMTG3VA<g<7Tek0q`0)=ACA@Q53FWO^$Es+l$33}dBlqggvnQQ5X2XiG
zE<nhF_;lV5i$;JT+TOk0OBztE+f+57HX%+PdF>``Y6)&q?Yy+hbaQ>NY>CwoVxri}
z97^Pcxr78Q=^ksq%_kGbz}K3Di`K_Y(gdJrF{enp#RR}09iJ$$CKt@bKxnoJ8@;8T
z<V2;#mGu*3-%e6a?%VTy446Zh$Hm<m@#^Z@SjcJ}BsIk;j#e;*Nl8k&=kvDb5ycGq
zGQeGn7OtHy2?#q`xN3|>#t@L>n%CpT>V(Nfu=ts;GHiWB)FZ~I-hV*pNK{EX9h#MU
zQ&jq;5$b!AwGhO4&cY{Wfy8Rc%ge2FK;KT<<C?zOG#e!sP}TLcVW-$JWhmvt(z@px
zq1C8pNfRYzPdgwv_ELBuZyL356w=yD4NXV|bPiJxCMXA1I;oOQ-w1jVuy1fJJ6I}+
zn|VfZ6uL!IE!J56?x6idPW08CyXE@2B5WP)p`z_urfxzZl@;ACQEBBS8I1hi7Qff!
z5tqmwmu!MrqhhHf<H}7-4k`ByFQCgk)vOB>DcVcUDhhp2cyn#mJgD0*G;{;HpoVpa
zrMFS`^}^}iGD<o@z6FJMs0Bc!A@0;1FNK<LjU2f{5%m`yc|&?It76OQv#^BxM1;`C
z+kfd*SDeO%tw2slLuGF*Id$+`^a5@Yos4Ip^B&%0qCWf!u6N)3$k!JK4s0rikLhpL
zUYxvZRADuyvsUT5EV5a&h}u{<Q^~q~mKE&w#2r*LOohfkq_KKt<M~LKm&Y#Rp9>32
zbcRDntY>>VFimQThpao&n$nB)kI=jUb_kBkk<<5@;O-0ZxCqNG$3?U!TeZyl74o*O
ztPBjjE2GIKke*N`QU?x)J}lk^g~S0oxg|lxXLOuk_IG=a=m+o1FmGBQ1}_MIje9Lg
z9k`VHp57v_NjLWr)oH&|rRK9e$5@mS-bA<8VFBt>D2WyTIzXP5ec23JFrf#7L?02f
znF1PGLqGfw6-9&{E+38bzET+axyP}KnKq-<@mM~D1<R-<swL)f3!it=L(5#py`nv%
zGKX;LltMh#g5ql<4zZ&Z)%6y@Ogp$^ua^jTcN-WXpansC>+`r#JSl~jjxX?`k<){w
z5ZoA;0x<{C-5!-o23k62*-@YG-5_zaU~WnC2iN4^3OlZ#Mg<*eh1{9syt#o`ZqKze
zQdRwd^+$cBR%OgZuSLz8XiZ^gi6fg>U+k)Dem$y?91JajDgCx!)lkEG%3}h)MGfcA
zpNEbpghf~~;RiZ{B1-_aB@^4J&pVl*E9TV!HorH17Jz|Khh<9%m*TlT+|2|sfHO%0
ziav?}HA=eQ>({RlvFOmr8hHz0KuM?pgxEu92<;VgZ4MATygjkZ4o$U8!4EGdF&tcR
ziInd1YL5_ecZ=LldU>nlq7bEvAam(_TsK$f1nR4U$3w8ZXW}*xBu`#ESZP;xw%gvM
zOjGA@;S49HsgUqC(KJrx0mPm;sO-m~wSlE#JuzRV1Cp(B#gAEj0HwoPy2A>i!TYVg
z!Xmu?_5yqaQkKxHTwwp6sKz_;EpaL}Ie(@yMg|h3Bvc{dmgJ8h9RU~Juk?ztW$zWq
zKwEmIBuZ!5Buj{TO>smIu3yQZHO@Yfdc|hNe71&naPE5v^9Y?g)Wpj9(wGxXScYO5
z?TYiJJzuA2d$p47F!_{N@MuJ%W7z4DtH~n`!3}sqKXVH`O3s=m^o@Z2etDKx_eT}Y
z710sU;~AqHA~v>F(3<2U>ZRYVpwcA{?bt2%xg{PXoxmY&J7U4U`z}wO)GlE8l>GOy
zDe0>&x=(mZ*)eN%VQ~XM`k{G1T=MX!lT%Q^^e(D!Q~{UKugsriV`H09&Uk%xuH6h7
zgXGT-&i$UCGK+FO)|#R-*0BG;0b=$EoGGrE+x?XG@@jTX$VJ|f-3b_G7opQ*1Ii88
z(2Y^wTzkTau8}f&?M!R*1;`x<hrCg0?U8Kes@}Y}Ke^OzIZezapgEc)#vS+P(4uRz
z`F4$|-~N;AdvV=%A;cd2PYkRz%i4hYp6ohsvK{e%4jB^$iR9Pk$ADALkVM~4tX@FS
zF4f(E=zSPjj6sqk;0}apBGbK(?K#T#>cRRdvf_Q{QX2v-kW1gVEtQdt?E$JVga}WM
z;7<HxvCvq=-to4yXbeb+F4PwZRVwCL3PB%6Zt`b5quQiGw8|pa(lW-7o*3en$n7Rc
zoC20M-KBMhX7!d!@RQBj1tb9V<r}W!Z**QBfy5%MjKSw&GEt)YBC~fWD*xrG9RPuv
z$|0ScBcBkQ_%$ew{WQgB33cLvvllV>Ao6=&prvC6`AqBs%peA;167hSoIJW}@d*8f
z$>{qaDlES0(@rgW90HDGraxfw4hot+`%gbIjDY(D^NHO@2@%ySvxUmVP&A9rDzF&-
zhJHgtFsuP4KhlRGyw75stk<5!j=GN9nl{nk#hzXBIZepoz(vrn94B%HA;auzV{6YX
z-*w>hSXYsVS{SB=gkS}(7mH;BNDSgOG(&U~FvQz_K_Zl1)$BubG{O7KWFJUZ-A`ka
ztIZIC%1wETMl6JrX8G|h((-AqZO7SBLBm^?-7SkJF~GDw(WKCd`QuxtGL7HQ>sG4U
zk7uYP%gf8_Os;E1a6@n(Eg|YP>JZh#bO1tu((hA>72B6Pc)06~A(^EF7$54BC;Fd3
zZZo7;>bwtt;ow;8qn1lnZYoZTd0nnb^nw_iraY@u>?>1%a#wD{DucRCwIW?U13vVm
z&fc@EXk$LPycF1#*I~iPALC6?p(nhxH>SZ;>2l6p7+LH@BRjZJMo2=bb*Qcko6Whu
z9+4L-i|a4{dQAC9${V8C4**}x;ha#@F&YH`A~1)VNr5FwqQrO++!^=XGcaFXnCPHI
zdW%qpCkm$R+P>&2eB1CK&j$O*ixkpY97Z?KC6`Rn_}4S3)Rrdoi0RCeAhLFJwgY)+
zSjR3WOEwJuO|4-e+w6{Pjl}z!<WGx-Ht*K+X>_YS3w};sz<xDppRn}M?XDQInOIlt
z41`Ui^hLIoxXZEfV>R>S?1CG$`z@FCLlr=U+jsa9J5#W|cN&CdeNbfc#eJ`!y#6LS
zTWK)`j20-5mF?RfxWc&BWs-)5>44pD6W}&#J}Y>75A{ZYcmGuP7w<bpvIgJdzJqIn
zQd|I2+Olm&Isos#U^6CznDot{>lZ;f_MW_S`EoIaaK4F*(UYFRFUO!MyZ@kDku0p=
zdDb(ca95<joj=}vsF?2Z*>ap<2v?G4vsiB~X%N_%SPx+E!+BShc4%Ug*n7;#YJgpw
zFt~7p9{>XjVx*ar-Qxy<81N$lH9$YuR2;ct9@M#C5wJ-|onn*mf?3<`Y?H#}0d(|D
zz@<WvH`IK79v+_Pt_7VzyY0(OPN+Z$)WgHW-G2IDICCc@WjV56svl5A(|<&H5p@o!
zLX2*k&O={DQRS*VcH+JCo_pZd(p~#y<v&92pci)FyZ7y25<0fGCuCG_&+a?*Xt@j*
z6V$TU%{1=q{yesi%92hi8(?{(ZgZ^;dQV!035M>t)|wz5@{GRpzb{D}jhNq~^pqdk
z<}WTw-CeqIk%Bbyb&Y^Q8?4qF%_pALnYo1w9}G};w%^^gm4lHn@_7|AH}HnzXkwOt
z^~BM8*REX)#8d-#hVTIaLWA=`U=@nK#@^PKuB`Ck{t=~e&|!AKH>5t)gaSmE2IOQQ
zDs!^l{Sh~)$dzBoun8(%JbJp<+13A@9xa!%yl5}=^69eW;jc=NbW(K6o@O0IVMJv$
z?JOju;L^h)Rp4GP?pK$W|Jmhno(cNR%lp+lzGr<OYezT*mS{uokZn)9QI~Kt{Bl$8
zP4ktDEYXW1bMF;?->ppgpCK)~N!yu-Lef2GIG_`&2QY+I5Hc^o$;mk~99+WQa*dmR
zyA-1q){L56oezc>gc!G4Ga-WjeMSREVJ+QrFW&0L_Ss_A(g`tT$pmwX=cXCQ!&E)?
zp6k+~aw~Ura`hu=0=kE5Or-)auwq=G7`|Kcc|;ObwWX3q`16p>9kJ+wvgRq}t(+rm
zdBBk`1jaCqW1t0BJW_ZyAAxL{`ck7dN+J2v_IDrP7TX&D?DkZXLnD_vIoh3ID%_u=
zVZG{(*P87{4sFj7V#-PSLZQWncdl^|iFrO!r-%t1(Yr8~D1dD4B{*{xUnw4Z31&SE
zis=Y%g<Fyf22gO1xDL=~+;5xlmY`F~e%Or`LCglxke&{_7wd>V8~4cJUijn}otONl
zxj5YH$Q1k9US6~sdmbWLI+;>NN4}|p0LRR%H-{63k;C%eA_^rjVi(I&FSvQS)4#M;
zXr$<2-`tghg}3>m_mMe%l4H8Jba=jAkyK7kbVz=<LcI6~-XyeO;PAUB8Hb(My*@O*
zomzcwhKvEHYEqhikIJM!WscsZ?jkGk&_d<0`LbB;5;rhq8h*3Be;f)a54(#oQ#my%
zhwFKe)j?$$t`z<TBO`RwR87jm=WZ|0^fO4j93uo8y&9s=7SeET^nXISz7!aVNuY-Q
z;U;h_DqECJ{S{{Ljbk73!B2EBrGN-i7e-&_^jx;(mGO^s=IE>JBk6gBI4~5IBbGSz
zh@8D#zI6qX&mmb`+YA(s7!eO3jk;ztR2B)VRg9Y7@1eDPG#jWNs`6o*zksZ@B*DeE
zas1)J;4iqx#fAb}<(%Ex()6^<455`7G%E(4*EP-?&y4A~{48l8S7iedL63t)=vjn-
z)iX?o_~ry=vTC%GMd;zAAZ$!%HSwU%Dr`_dYUUlwv+CP|2%8`5MB07lr;Egh$~617
z>uv$4FqXzCSZ4Pn-1SLX2AMrISp>_F3f!C(`FmQb(5UreHzrkk+k|yWJz~}gh*w?;
z&k#q_{@MPR_b=Vl4RSDLLaXI$u?MJF%L1@N3^50N+wdY1C{$(^gF4S2bjv)N_+{_t
z)UycT^Lj!yQ@5LRhXD<kzip8WDW{A9mgp`sAYIx5UH*9sAlw!2M%3kH^c4PD+Go3u
z><{X1fU1n1E_^SmUhZ8Y?yo(QExLgac#*HIGP}bJ4>=q2;ZhHWaB*>MPJ21uEs%!C
zOz>iA>nO+bx5l~5a{_C9p2dfPx(RPdt9?NchK!~!RbI&uZx=p3NLVEX4{Un~Iw`^a
zTr<Z^_l{>5iW+2s3-r8CqvW321AH<R=~J<*ALj9@y1<iMKMxb5nk1})P;j)z$g)r?
zF!~LJ*NYesH6_+V>NvoVj`q_fR*2O@{$cKO)k9$do#&;1d@`70cx>eff<UenbDcjA
zG!=T^(@S{8ps$Z`?S(>J5fiw5d0)P#7)yG$VL3F;P_~jCE{~?@50QEZCJgrfY=h6N
zkFWsr*|iXYe2}dI1hC&$qR{SD3Y0I_Xk)^bnUtmh&EZP2c}ODF45UxOoL4bC0=T(%
zr9THy&70{qmHqnmeNHTwn-b^9#fB`3K`2UY9a@FVSt6pAh<b-{0o37{gckUG-e4Xb
zOi6L6PcRiIAboEcJ(r%;nneH8pTJh$*6VAtbr>hU0a8O>z;>m>c?yBCk1+8Cn`5w7
z=BlSn<xO<JnZ4EEU2qO+qX|N$ki2n22nhaP(mUD!oR4r+OhvslX8D@UMF&^|R4E5G
zm1{A4{!Z}|xS#F;;i2mi(i$j*Jzx!lvKSvK!c%Ga2|}NElo&35w5w>7*dQQl=sQkd
zpYJ0)n<pndrP|!IKYi<m@v5TP3}l~Lk`cy@ne^}-q>@oMaJ{CLS1{?{7IVVkkQ`tw
zVL=1F5*yD!lO}4Z@)BHFrH6^pCcGCFBPs|CG%fK3^iQRwpD9nCJRwA~R5b&iRQw$%
z;H%9Lh)l6W{LS#<eB0TRc|^qy_l*OACA{0aVnyiSWyDxx(+Tt?SO88Mf*OP~IsHad
zPl_Z_H#tKO6$XzZ-)gLQ1o>!%>@1r05nZ^RyZS7q8rmLqWJ|G#6&8ypx+sHTU<!w|
zG%)2Tb|ufV9L+?tzaJ0^_u5S%VqHA`>X}16vvqG0fF#M<HDKR)?XS;cLI{1LytX^S
zso9inV=s7M+R2?n|GHZTydtrno~C!>ZXNKO?&!j&ZIpd^#hB1D%;d;gKOrnrN%RuZ
zGX;+|v1Mi}s~5PqW*6E`>dM|g#M!O)>9R8|m-*6DuSP6~8I2Y8(0eB@;a1<y%p*fg
z6b5=(uS-ksNw`cRs|9?S(W~@jhB{^1Jc=!#pcl@^I0C3)9XpT-mw~_k0xCu!Rug>g
z;_wF_f@md#LeH?pC@U#<B<?T+6BAPxrW7j40Qo;4L@~=ptwDB0X#D^jfnd-Mnh#DF
z9T|x^4zHCrQ;3D!&`>qLSqsMjt}%ywEq^SvY7d=YD+VEi4H9tp#*G^=Ro9`2*@9wQ
zKu=5W+Jvw|47K*(a+U_I?S$ZiX|o3S3t`d#>L<h{bYzHUSk?jfOAXpH$6>fm70`P4
zdAoO>{2_nBadYsM&|Li$-1jstFApLxZvFwpCRjl*_lPGHKg3I!h<lt%m?~EK@-t6N
zjVO+s|A0iUgB=Sp6Vc)EQ6N!}Er>cutPt7}`~))?py*Mp?+Ye)I_!hRYH*GcJ`<`Q
z26)mA*?)hifV%Op?|La8y*1n-1HGmBxgf^ZR|j-uVIjYbg=s5fE%rQiaRJ7JipDos
zvu_xgQMbB8mJs)|?m_Q3M+~QwOuujY<Q<#6?Jv^IyHnL((B1{%(Lx94-88x3$D*iR
z=Rw{EeH`;6ZLbUWnqsEDzmpvd(ACp>d6@HL>G5!8<!-J-=xKX-Qg;#3G&p5I@JjE#
zAFb@lAgAk4!)f_1->D(qnlN;jv-QFWvYRqgyUvSsDcV0Uke?W`oz1Zkt?gu)TQ-H%
z?EN+a*M4Dv%{GW0(iT7T^}||v(qZ#J*~a<PrAu8&*ezjf0>(<R60HCH7AQDjsbz@f
zE4c}0QrG%~HKN1)Xd>>57=0qGd`8PbnWn?uG~oBk14G4lZ!-tcQ!mR%0G6D+8e*&K
zbo3v2rcVVzIc6pPXH~v|&9@4n9A31U`URR=8hV6U4#gd%axndiHIli^i+;2I3S|jb
z6U4kOzyKS&ry}dWwXNB|AiX`Mq~GXq=O|s$$h8^eco&bzCpsw%T<&ws*mzn@yZRbD
zPayCB8F_W2wpK>{3R8v68&a+atUl-t=N8hCY>nuYFv%qU^>~M=zw%!ELUr?b%qoi=
zzka;)1vBC|EqR1l5=&eiDs<7TZsRHh=?#!wM2=syMS=^4bt0(L+sUcE`Y?Wvl^=wl
zq>j7csuPq)3i8o=Px3^AiK)!;LdT((8H8>D8}$Gth6WHjl0u*v6C)T5cbma+#k1h7
zhWo^M1vhAx7+$}#I_kKI&b~HeEn3QfPB{qQpp*xdg-AzZd)C7=NaaMnNcx{$2C3Yf
zPj+Tg&whwG1@@{PsuEb===XSQM2eP0gBbFtZZui{Mi93O_z7X`{$Z4?J8nwYETQ{b
zz!{DXLU0oEU|dK|n20liABnGt9|u6oM(t1!aJYce4d}HFP%3dbsQ|7BO-4q>W5@z(
zO`Gl;yd6!5i2a(*3NTY|hHYk`lMpH*Oup!|lF$ag4%*Nh&_D^9HneMjyo4C^K$Uht
zoREXJVD(U^FnwqQx`7%oH`ygZuM1IF^6FK-eqZ@ccmPr5A<Zokny=k^=wCN+AJ~5a
zV+;nHOGDaxe61&ya8Us<5L2|6eq!97qAdoHNH)3;8EN<H@0~N;9Xuc4A;P|`U~j3;
z-Xh$+4?sI{!SGomLT$tw$Ux!kS{vZSDsp_iHyZ~=h-lCR7ss7IFj=YL?IOT9#ybBR
zkwMXJM3RB;Xrkh^Sy@(}o{6bJn<TPnnSIPxUYaP~p~z(13UoySgnvAY?r=lhyEg)S
z5U&s`2YQOl2}U!6$Bhu=Jk|u{(0xpRr6DA25EE}`EkM(m=v25W7!Bd9+U)a@nP_-l
z78#&Pqq>j2)x~w=0+lz6JCki`O!@urYgY{4dPpkI`dS^)McVFCJcAA=Dpbv01wa5=
zOn+BM!Ub52z@F81A+KK(7t)jX%x29_ID-(2vO28PY^qR6HvnvNvPYIJ;ZFGp2KPk`
z0A~LWGjBdW|Ev_uV<GnbWmDBt;*RS-e3ycj*I01ijn0gnvYFWon74Zb9m(wx&YqAM
zn=trHOFSMbU>T>Cv+aWO#@dvK5*!^feHA5&Qp1WYouybQUGiRdjcFbp-uWr{37k{!
zo?w!=Mwmi7Uk3%P0HaYW3l&{P30s5uTdYQhz=Qxxr7RSwU}#IrB=zQ1qApbP61iRQ
z$bht6s<4D{Tkqn9J0L7@dT=D!IXytJ+P`7EIlx;~k@@J2jjvmlDJV6p1Am(23<Md0
zSSqW}!;njOP<_>jSj@ch6Y?vjHf=p|gDs~W9;YFTZUmlZEPt%+bvUx;0gMMyc6N3{
zUYKV%0SUOL(2-u=uT_RUtO1UXK2lQ8e!5daOTQiyV*^eqc{@LRD&%ZnJk|*o(ozQe
z1Efn9#9+uY;NhvM3tKN;XpSdN1O7gNwHCMDP3c^|riDT9nF+J!?lKf47dcG*@dN+2
zqgwyJaamnsnZWTtZp`{^0WsZSP&>YfQ!pAo8A&($;L4MD37B-ydKcEHulPTBc5qg4
z%XJ^58z5RDJXmgxkeUfN(P;*orWMQp*&|!7Kf-Y<;(h!6`j@{D21DY5QD~<8;R@dT
zdnetQC^HhNoa=upqZStO(SK48*?+N1djb4bqJd+yn_ckY5&}fO1rGy*4q#3<8x%}p
z#_q9&(0MbVI()3$MtbM++bu}i@B-1S3+CtA*Z`lxi(lgbS}WQN`FotuBzyL|-b#iE
zI}%Lx)_bTs9B_Jm6vtZ-=NU+hl9-U7uc+U8uVG%p<H~ao*7eIMB{zExDm`zGx<Px9
zD16+QKfCN7gbyVKhcZ8x<p6z?3;&6qmHhp)3h1loq1%t1;^dSizOogn2<dvnBMy~u
z=&vYj<dKJQJ>EneKQ+?uUV_l>&p+~kr}X-1qBdVsfU624-htHxDEP#w(7mHuNbf#k
zNI<$C^GNF9C>j>7l{|%bANPOy!PO|e6c*B@7mTp6o(TtPB>0iT&s(1|lHkH+{$2C^
ziVTgni;!<6xkM1rdL{?9HNv=@->FOd?lp{S@I}l6SLcR(qSc_VSz!Q7Eh~g~2qr#8
zv}*Kj7GMLIaqM$9T+=@brwVj|8O$MPXLr~yuk-!B32NuB@K6_EW>$379Y5;wdyO?f
zr||*3!?rN&FflS7vLxR8cTf9ZgzbVj+cRuPTulWuU(%WFJ-dDoqbI0vJ8$kLk+u^D
z0ncp5Tw4PWt<|wdtC@^+iIX@}<+~MIIte^7CQOA6;Y&tr)VC8>Ke2VhDb4Ll_<dp|
zMLKf~#0|d5$l#=)qu3$Ua$;qlA)quc@s{OaJbhZzasK<3mTja<8^6Ej8GQP)<xqV$
zOdAxLR3y?2aVY8LJy`VK<E^~Me}CY~Yp&-jiE}L&gIqoWSMXsPo_lkEcu%rmsnY`r
zgc_}H>h}j}inRfT%Y6*u2a7KSy52#vcfCoRc_;ZQfQEDUo~ro|R%L9;f35^yavtl@
zS3lU&`Dr>b{?C(JPqJM97!dQ<V@B<<4x`BzrBsq0Y_dAn8VtCbPl-PkqA5w*r_=KF
zDMj9`&u((tBh$#oX0<2NtkbyKZwGE)b`&YRRTs){u1CeEziCeP;mSYvOCr76GyVH)
zJz#*qS8U$z*YIBK?eAAbA)~>+y#W8=U;1D9!Iz1VtoX!NOiD^h#1SNT$JW=@92^~4
z6=Hs2V*cybtDqp92s(J^5EQ~hf)c@F51i|tbQZsNiMXGGs+yXbU@)t|ub}o!4lQlO
zD+l9Qyro<uM&Gwl#b5~W@bU`Cj566pOG_&vQUt<7{DWcc!ootaAh5Ea!_r6Z?eIqN
z2?)3q`uX`q-aK=VOiWx{T>JM+N~#*R%i=FzIN(EDU$Jz(?7hkgzx2nmk^CsErQcfc
zA8vbh?tDg=reGYPUiXK}fXh`|AtUnvGxVSnLh5y3Jg~#W1R{$E2S0cab=W((Se*VO
z>=h^lGm~+6pC|X469V%(-Kly|Ti?AR91OQ^-O|iuhjyPjCw7nq`M@>YdHsNweBQ(E
zM@2=>8ydoRdctY*lHS%jyurcK8yg!SMlp)<{*kPYwtnvF+C+9Exl&S7Q!O;GejZ}K
z7mx<DnlR_X>&B+B@8IzZ7cPLq!GV^S$i~XBjpS1Nw<b9b1RTBYe>TT{(OBqnc0xh|
zW-G_Awm|7ADJh1AhR~Hvt*uQkyn+(|b!cQ{ghki&L?k>p<^418VyWex#Ak}1t4dIA
zCL(S5LB(rEJ39smFFNqBxX>@|umIvr0Eic}ib(CjpwTj2eSIwTGTcHqrZx|bC@S)2
zS^|9w2H}c3)YR0htrt-v;A{Pbk(Q^IS4$Uj2%o9;#?Kx2q(F~#^n^>^a<OuZIqFIA
ztf@F}$S}zAqqWrm^$<!K&*`+RtVK|FC@2p>eQ{nQ^1}yMIYq_9TwPrW7cp4BlarHe
zUV?M1Vtg}*8@p!y5}eks3l0qpfq3F*OBr4s`ka1`A78PtS%OueInUap=JN~4!+Y1y
zR?~Vl%0-C2K%hlNM~|*tyDB4-H8u_mZo>O}Z~tuI;+p`i#j+0wX`oXR$4&kGbU#wV
zkw7fp;h$3Pa84!Snb%>?%cp&8?d&Gn(j;1#VNk-!6r%=hM@L6<GqWep&Z=kE|Kbt0
znYv{23m5O1FGlsEES5OUi->Kq%)4OHdEbF}=!Vqg-y46Zxw$#Fhl9*D!mPhiPcXpK
zvj(DGYFgTj5%Te_BCSpK19m;8$Rf8FFF0-7K*?3o`*4`%Pc8{@aaFa^_ROlTF0Hw<
zu+*QM5mz}!7aiv;F8*{Vh0L|Ms;X*903UU@PJNnh0gQ#TJtQnl3!1TL7mN&xIAB+8
zc~VWS9iws(Z#Cz0k};5VL95xc3iD(2=VoUwn)W<-_Uzf`;NCK?$sZZ<=QQS02*!)4
zVQe&agCpg6E-vz_U#T$aZD7C@Vbb>F2Ydo*JuCR!H&1M^`)7vG{USyaag^wb2h$2B
z))y^DeDicr4e~Lo*w1+VjYSi>P)KmFNA+0&a!ZJv=(_u(M%xUVnwv8*5h$JmfoD2(
z>afZYceQ+*6m@oL$9M$)eq{vNr&C1vy!4lHp4Ec}i%7X6`}b=nrQz`y=EwtE=5F}K
zFlrCY8DUUB-ePub#0XOoMrhaG<lt5zVo|e2Lpq<Bm?)kr&e(blH3r6X(XN@r#enSd
zxR28(&?-q}0wyLV4Rh2Sns1B*12Qt^u}>BSv<D9!+`IQzPmdf9Jb@2Zqxmr~T;U$X
z9hq|)@xb|yW2889B@ILbIFf1Be_ewg`Nbxso@bSKuA!phS_{klh%19AYBC`Lx)y%q
ze`p-CGB-E3vhx4^@udvEKRy@7eqThrH8s=4Q2<NN;<tjfGoaqUi%9mj9v&VaKYj$X
zBy1#HT=H*n0PTqU8g9b)!5G>N>_v1UPQihJ%i#O4A`g>QU}@}&zXC8me|g}bXC6O4
zKVhQS7nQ7%CLXI0q+ssu?%~li@1L}NC#o6Q*5A4E&ivuSy*Ff}r7yHF1ATD>ZQw{0
zhgzo88z)$PFcl6|@MQ-^gr@JW6p>UzhBFGsgV+VWr(?&8Lr#hgrm!4zgDa>bW`ycg
z1o>brYW;PU16XLE-|cUENRsX<<T1k}`RBFU{3`C+MZHW?&A4uHNrgDs8^Hdc&lYwR
z6pZ6iP_PS!6h~=TvT?8QOdBx%`0-=U;d7T*Wi9G|qU`HvYkS0*{-v#}t2<1bZ|d7Y
zwB4KayBKb;Bqb#c2)KHAc^RzWRG8(l85EW0&;PuWn~{-`w0-i)OL~!{n`fR3UynQw
zAp9oDs8t!0dyE=fhgBPRIXIGm!pYJOC50SqLXCPqn%LB&z-)x7a(n+jD`gMwk%1iy
znI7~nqMki|{MaCv6$dOYTu`d}TP9R+-wkVXbFP^tK-lJaB|E#^bFy{M&o#VJMFD~V
zH+*9f`<s}Uk>^Yp9jwOQbmvkD+VzB@lK7?)kMQ|dVCE7#7^SA7!t>w<_hf%wcJ}iq
z+M_y=tdEmaA}`xze%O1AH@Xagk)&C`BDD#9r72mADLx4f4ko6u59WvVLdV*B!^_>B
ztM|tvysw_b@7iQBSQpkZ?{DBzp^!W9h34mhOUgSmwv)Xu{ucMz{czyW-W;f~vEtt$
zqDBK!t4`|a>$|hxR-Ur8wZ);En{$aMnb*{h{P&&JFfcH%ut*<LL-m0|ZAMpBS6;sN
z*sxN*jcNDBhaig(0byZbFU86b$h>469}x0DlL71Y?}C3ogg(i@5E&Dk<A~9Lm6cV5
z$uWL2(Vj}Y{l6;>JoS$rJreCo&B$1X#Q5O;{TUlKh|}&-JOB7&NZ>0fVoQtin@}I0
zSr{Azo0+UZGa#;?_`NFVs88$@E6K*C_dp#58c7(!qN+g1_0L9M|MznLjUvp?$|^aR
z=;Gq?>C+{KD`w~$fIkPGnn?%qod|M<PL=CM18?ro1R12o106JqG!fGo&;ha4g%Dp=
zhh=4Dp>_Fw&>Uz2j>hz<wY}ZFFwb$->Vi2wYWvaOVfrU>{4P*NZs#$aWkmyrxEK-r
z2jC6<u7!e!?;(m+Cr5(1iASTL!R=}kffj|~sNCG%DrtW~=M9Hjx85j4xVuZ<zWwgp
zEF6)RV*KQ%B_zs-vqpIE1Z1#7HLtwAIqetjzc50VxXq#_sf>&GbVkd0-4c>11QV7O
zV*m<RfAgSL!m&*dxPL$gKA(<`4jr113?Tvk?C7vO<kVO0jd}#Zjl4@W*TN7H4a=%!
zZz)BN`o!oc;XZ<Z!^g+>Cl|~~2!-=4Ee2GGP;jC^rHHJoEG0L&SSZRN3VxN0fEm=(
z)V#b|km9*<w3KKGx8kDUejW<410}lYa;7HyRgxf5k1PP^#1}!MY$tQ8gLouw*`Tw{
zYNnps8-**TZdhXw_o%l_0RlWQz^ga8U0J!s!(;bjMPS+(?<|e<)2EjqkBUmd4_%>C
z5rlK^lnlwB#>cGz1`Yk4$e|*UE`^(-=ECb=;Iu9vLe>S<o{pXqDPV~TI1ne3gok1v
zzRGTAW{ug7Kyn7oyf>_hB|Yr=yCq6tvJrIQfgM`i+8GC!Sn&gu9Vw^@!zeGRszJ8#
zz881sJTNsig@-qq+pod|cp57|4wblDm7EgTQulY;J7JDn!Lk5%oMT{Mpnj$g=p-uo
zoWRcCk@5r#CX)O-<x#$0gM*8}it0^FBG6k!y1TpW&aYvjY?h;*y3!X0B*l$68}w-<
z7C-+az)hEw=U}BS#Y~VdaDI$9uAcqy4ieZ~!u8oR!k*sQ+4<LBf91~H)dz4BYVYcD
z#uk=}_Fukog*dQ=EPnFj4;`bhP2cPN2!{E)^_#|@1A>BR4jya*r-ewaj)%*9`;PW<
zj$##@*QJ={{CP8c7)0Bt#YK~bh!V_@m2uJs!W<tze_MMy2BARgVq#()5bDv-0bVJ{
z$+=Sd>g($P)okqTVdTrSogIKGfi`MxX^DD%1CKAddesK!hv2crlrs!J>`+juj#JFN
zfaK$2V^2&%;9fw>_@-QO3ayaL^>X-44D*;y6EATI37FvdPM+*RlK^Z`Jaf=Li2=C7
zI8ayldDeZQ+t^*aE|C*=@%!aU`9EZM|0`tef6N*G&pZAXSn!vLe+-@hyiQ9?`-5pK
zP6A)Qe%)@m$8|^z+Mq81_}=|;fAQkl)vJ4>)Y??q{QXbfBpuw!e_kyAWLTwtPs@EU
z62MizU!o`iH!z+VYDo?oQiBJ{`=8%_ixZp(QrF>y{;?`A4h~00m#t?;Et}J9>t<pf
z4*WYjzx_tFM5JDQMieFDb^I6sT!@#81(Eu{U;i8b;0!V0O5Eszs==laa&i#IMwB))
zEr#VJmC0<{cOMf_n9ZtgXy815{@XXdt*0(uF;U^3yL3mit~)V3eDUdXj+5ydITv%)
zay6qkR~7S*Wie{)^V-&V8O0ietX^t*ZVvM!kI0e7kKk$5MQz0Mt_<tn{T&bx066&U
znFPjDkSfzKEkaRI$^Z1o5N8ZCGc!jvqaq`3ga~|pG|$XeTwF|yo={L*;uNcDYHB`z
z{#;cRj}a|2RN^@dQc@m~tbZOf=AJvJC~G_f!k&X~qC8^m`wlLiJ$v@l*46@A0fG}x
z@k8HX)zuy1C*4fr364CM4cLJTQA~zgiQo@a%{xCP;Asvp-dy2=sK88+q?&vH7YrB4
z8^P30o;>-p!&3Nj3>-t?e(4ZI4``^XOC_J*9j~4<gXgT=5W0sR2@hwp&j}_beGFsq
ze88p=fwB_fh0<-0)nL5t`v&Kd;3Vn(_3QSOhaN|LexI2wC_H{~!f3D4gO4dG6_u6e
zZv`bM6KI(C!H*BpyU56b&fS6DQWQ9<VT0j$Hxx=iJa0_NPJwteVH^%ncSlEu3yvW|
z3}|d@Jn8ob*+rU_to;-PZ9<MK7SQ2AVr5zUf`-j<1o`=wD}0!67y?EhD}WWkkOHCT
zmWs*|EgPKPxOC}~l~v|c;;&!7FxY-hr}))=&(5buV}>;w)gx>8&g-lujt{U<yrS#v
z?WDu7RKd$(4<%)mh2?EdvmA$spE*IyczVTAezOm;{im3ijQk#F$=thl5Aqiwy=?MZ
zfmlGl>cqn?T3c_uAm`2_j*h{^30wky+X_bHfDAKByPcsht1?3ys+IiPRoTjF0sRa7
z`sD7?UFJJ>?7-kvta4D_<S*7`D)p_^kDk(%vOc0H_<{K{hSVE4mSC{G_X)Z)cKQ$%
zm7YU4eAQTMDvHu0qIFwb$fo&zJxIfl4DH-73JH*oX?D&TmUHK(AoJ_RQIiGZ=n1Cd
zp@O#G%GbcS2ve21dhp%tRAfM1=+~}#m9Qr5-tnPR7?KpO|5lD!q(CIpb>J5Y3JN^@
zig>ii#u~BWz~+-9BXO0`7=M7L;`~g-WyKw2tTc9qW6$u;5yxa8M+ae~BGd0h%R7Pi
z%em+ux{Y+G$MlmLBNLNEcJM6tthsWKI(r!WZK15LTY=P6^lO7ThhKphLGM#d-AYd0
zo!p9CXOlZX4!`c*hYvrFWF#hz6E-3M74YWe<z>VjP`EN;8ju*z)0qWJ!6TWHD+7Ie
zxJ-LQl>$i~5?3gD54AmYf?WVIUPaLi5IwQ1eUX3I?IGR$+Ut@fg*R_L6$PM8JS?SU
zJA9%Lwv23(uy{ZngoN`3T3<czF{DZCKZJVaVpj#?aIs03@Gi2mqEO`f`}#fv?mv1|
z{ln8>;u&_(UgKI&7!WczW0xBe5@MhS?d{#WcUbK>dGVJ=1Kk;YTq0gPr4wgKrC5j0
z0q~yuPOgtnBWyg<p@K(fX;}lRw183};6UNPS>M(VGhssZ{)>My2tCWHX37(u))<SQ
z=U~rxD6{47KRRDSy^Dd$sKgox8;LFR%~%ELw-?4CZP+sZvHZE5VyA`5$J>8DZ1BM#
z0O2b>|D&u-A3F=U7KDpbB2XCg9HAoeY9`bmV)_LU6}Sw+m}J-8+lvRT1dX1T^gYu6
z4msYwnHBL+TDlg|?p9tbh4ThMlHE9nUxeDY!wV*m66}bAPpD>*6V3!46tCD)`YxP9
zz+lLHFs)8lQbwlb`rP7T7o<tt;df>*6TDEzh-Vr=jKisRod*wWV&LGbB@XDU%o|y{
zK`~_4E1_x&y%Nw>l9ltS;%<kLrPepY7fKO=YMSPOZ41sMvY-e$uROc?d$(`jzIwGB
z*$1L2C?vFP>sI19L^vn;>4O;%E_MR&-5PZd%(=a-jd&0!S~c80i733nikdW#oq?CJ
z=LMFdN)8SVfcj}^=1}C|Txtcwg|&j3RjGRE+l<W@Sa}M;h)_0BhZ!Imgx|yvrnJ31
z4<ZZR))8hNsa)<fhkMr}XW)Lv!?Rl86?y;uJ=Brl;9w~6El^7_hRZZ;Ed9XJFRQNp
z6Hi~v*gxC3N?6L^U{VFcDl=;$jD)aJAVGxya&d$f2QZNMSUv32CJZ(>l)nU5-azD9
zYilbpq{8DTk&EDBrsn2G#G`gXg+x|{7-}d6=Ck3+{>PknVBfy*r0~2vd+0WJ+)+EE
z`=4x$@(;h3mBmL>#kIAzLN1eGHARPk-qWDE%*f3>hZP*rNSuBDY0XC_-Vg5PD{0DX
zJ{6dio!75D>SJ|}^#k%7YfSrn6vRg=RIu&Mdr<eS+qS`a^nQgV4Y3LtM@y^AA;F-t
zzMl5QQ(>)!`gr_J;V$h8s`DDO0o50Ug@vG9EE>7#N$;W2rRkLI(W%tJ^SBU-O`DE5
zeeUSskS;(HXQ>q0Y1s5IxgpEAYmv23E9q{|{sqR+0;gtYwau?{qBVj^jOyE1b~bS+
z7v<VpPReyk?~yJUbBnBKl<Q+|sA#b6ozQeNoiG2wVCa^<mh>QSzwS!Ycw^@EcDG0t
z0e(KlD5j?+PmbBz+?@<5Xyf0JN#Q-f!r~v-y4kT&!^+Dssowdy`2;)9g!r82Hfze@
z=X~M5M_SL9oSLutZlV;c@}A6-B}aipF~RqJfsUEpOnkbswf1DNqus*Lm-iKc6?%dp
zuFN%mtt*H&0YlW7Q3;{KM+$2|W)+=1tD1f_gb626cz7(~?#&&+3}z}J;jBh<BMXaw
zfT}V_V{@|=3@vtUD|o=h`$`o&x@o;7+k8;kG6j79n?qGYLqq-GS)Yo#5L%m^$-)sd
zqmwVQb931t3$h2$=1bb<bpI2iusSw$3D1wU&Eki0ub=CUj#_|z$ZJr4NyZOus_Cij
zD>}v*Du0Fr;(@lEzhvP)a}SlgIXydDRZ-C(aC7#^`+<V*=1EY5X}O}oAum9nXq1;m
z6?c^YDU4;07Jaclq6ClMT1AAeOqN2#(0y2BOHCaBes@<-k0w8B0HagJntO=lzT6l{
zINg>R1`U_FXEBF(5~yGO#yffYb*UxgrlxjUdU}qvZHH)Rf@BU+eVf%iUL>OdTmF0w
z;!*os3ZCF|lO7_u=$##s8LzqL?s3m?&#4*=%~Dc0c6wq~U^4=3=S}1kTy;k5B{%Va
z$f>C*$Y#6h4j(;w|0B2i;!zY19Bp^?3#hoer*qHbSv!dyXP?cWvk7<fl%hhfNDB!w
z3VwBtbgt3mvF`lA><~X}azk-vv!>>%`(rKjm-)f(n>O;8d|6~P91SfKbu-Ll4=~I+
z*sry7+XfA_(SH^C^VYHBc5)R-5tThGtQ-l|mHSM%IRx|SwJqJ%yQpfMzgSV`EzAUM
zklXr9>kYR>eh*Lz7-MR*kII^|{(xg}J0^2D-G){OV=9dFH#w{ri1Ts4ZHMI*@L&u~
z_ADPf_wn<~!0{Ix8#rVA_KsRJnk}AP6G=CJoCPO0fm(vRUWoQjKb4S>AX+AliCr(q
z^MkQ18qF8cx5CQ6)fi1!B4G9%(lYA@7RTlSFdTBUhRmh3S$CGTgHFKWDyx{$?CD-Z
z&(<Pn=@WG8v6$&#EE9nyfriWbW*P6jWy_Y?!Mc{s7v!GA;q|3yL&F|QXW;0qTeosY
zA2xVh)sH?*xEnb%^J6y80((HyN)}n0`Z6?R;1j2>8F%CW{Qjm84^W`Z&EtVzptKNP
za6B+)L}USKQZc5!v~GG-H;NpV%?hb*4WT@X^ZgWGVqIRah0=ieib66_vE9O$;kWJ}
ztzh6Ly;8TpeH20z1)+k!qeV~RN*FmFhtqH!T!Fhvp$$*cM0hYCB(to9i1#|+PKQ|L
z;H$3ti4T+)Z*}CoEEjcRZFAoC?(rz4W|OJxml=xh%yzyXSvgZ1${ims<{UKVG}q8Q
zA1JkOW7MS3aQ%j9UQfiNl4X2u;^qbYo%-Es-_%_*{N`uatIC$VQc_a))%?<Ga4!3F
zY%;r0M=nKm?EYAx;e1VQ=;%lQ*OX{k+h<$3v}=E@Kl*vBW7e#6YsB-?UYt|08K^#D
zZ@&tdx&{pqeGO+Gu(7~}sh%~?&b6q~<|}iCkHs9JwxA+sWM<OA;)bV`bmiOn1P6na
z^dmpx-N}IFaSljWQ%3mt^XG5)&7(##9zXtqlz{4ejG8+7<{9D{G+(|ve)LEfdR$LQ
z$>u`?&K3%jZyzcUWr_#HF@%Cm+_*D8=WfM*lgZ5;iNq64&HAoBp08MqM)!*%Fw)92
zL}TzTJV@Xn2$MQHlz$P-n{&INe2+q$b#Tc4tlHm!J9_V@J+<p2JJ=P59raQ;IN`5H
z{gqIK(04V^GpC;siWj;jKwBw28dETi$q&^L&+)~l?P~P9-jAb9cdP&0nt?dPfs)Ic
zi>(fk3M@d(N2#d%-@JL_=SScmR0DjTIm#>Q;Uu1--u$R52zuyAsRX%5H-iC)x-<cL
z&oSZp=l>(%?B}uVRN}>2yx)5aOn7TN7a$|v`}S2lPznpqHGk_dAn&PDr{1KQ>4B|E
z84~q~dk#@%5(i8o;=G~GngVqU!+vym%uJxr8ADJQA0NkH0*{5@LoMr`^kpZW*EjX+
z@0B7Pk|Iuu)3K^^+wWO^pz3Z&uW<gsIJ1s|W&Da!mR61C*wX1<5#y<p4Z;54${R)}
zX7t&p%}fvZh`jlk^WWHe>##1@uv^f*w~dMg24#WLNC=340Vp6P-6E(oN=SpPv~+`j
zbSd2k3J6G-fPm7Cbi=HdeZDj2I%ocwYv!A4=9-Uxwp;MW8_#n;_r315)@}1}-TDe8
zoLf4iOC|04)2=D7C(?%c*{y{=TYVq@O!3A0jcYh8U#F8B3Ug8u0$k*@@7}#bArOxq
z6$=Xswmw>Y3tqJ1oag^O-t=G-*d5pcsDPh6dqxy`Po6-KsDqjfNa+ynDBkuHnoH~G
zY0i$AB0M3yVw*P8N6MbxNlI;K7l;H5h$o%5vZ^ZNtyZXC&|^R+COtEgSvl=PYN`oH
z=lFz0wEAv`@HBLHYdGpz7yQOE!=Xb;LPD}<4|H*2=q7@PWnf}5)!$Djv{BCVpka?o
zNcfJ++o{@m#X%pYDAI!ctZ)lL;s~%3{p_jVzqKAdR0IzjhcOl^ENwFrWQGxlme4|-
zSEQdp<vBOn!kGM3Ew$<Q#6)uLYQ$|im$z@><83i12A`W&P!~-f|3Y>LZlaOSAduM;
z-bQG;VIQhE;r0tX$6o8&fEivv6I^e43*Ee(yLJ(WY#YHSmkJmj8KDo#MeSDl6##8b
zwAOcxyISYgEb>hUoa1i@?xm{Hx`w7R`~0I_yLQdn%QE;b&lxd$B%ed*Lh!oAId?cJ
z;u!7a89bdL@InZ(Hdxh2vVqMfOJ4w41B-YDIuz(S)6kX(GN5rnw=jG#JPknZTghKs
z@(ts>{n;s8d80Or9tN{q3&uZ7&->r-vN$29sZ*3L?V-XiM#fPW=@+B%YWUWTh`rGa
zxAzAG<ukY$u-D%GCiN{gqpnxd#mut&c@#ntGC**tM~8;a+Zn*&Wwb@)`HL5B){;4(
z>I13Y(}Ze27je4907L`_^$R}>tYzmTv035v8j)g%dI*ep<*QepZV<UQ&NM=GgP=dc
z&{J8dhl65NUea|$VsJn<n9njrKk{X0Xcw9=7H%zAL^(F<vk;gIiHMv`lGh+$lK{eV
zUF{g0w{9m3zaqR4yu9w|^I{8X=9$N{b-{iJ&kUVBuo&z>!osT*@_!Ks618Zs3!eNa
z(qL#iAemi(80ggr5DhY7%8=izgP>Cyf@e+4M*tMB#^MO_72?h-ZAbV$I`)3u)0cQY
z8oC-H%Q$Zg;s+WWS`NpKP5&OtYiv349IyJce0t2nh9+u%{DQPw_s<=npa7gy%a<FN
z@T;#c%0d0Eb{Y{lB4gi2C`LqNA3bTq969UKmaH1<Sl<pGf4ThB(W--FWYM{K&(RA<
z2YP<|fL1|dyV<+)rZ%bKD|7eLKR@8|Z&`a(^CoK1O@R8b8l%S*Rfj9ROWYO3BZZlk
z-trQoYjUw2Ui$Nn@jU~JEM`)*_MUUsmBv5YT=o~OaWUe~w)pu&M!w}e#kJCX2eJ<2
z?OYdqM5fHE%*dKzSeRnrcAqOKNuPao`1?yv&prG98QuS_jPefYO^=HkXR4K77Vq6$
z`#iM4xSq(GO#Sil-RZ!xK;;gK6WP;7P0ji7@ei4#oprB3a`nV!;maLX)=-$pfR(o{
zY;9}nK_jTqmXBtU^rAOXJujD<kIced6pRQp-Qpa5``gMI8f4Z3+J2o8vR-bEjEqEi
z6Jx^<4)?kXd0oK88(b%ZBBaMVv+L@G`zyho$h>1{WMsjg%_)M|5hJ<YChLy$lU6(<
z`!qLUEL|GD28AMo2ChhD&>BUI173$N$d6*j;zzN|s2NWbN7kTMA<uW|G7rzv($X5F
zYXGCh6H>T#K12}539Mobt0DX8-ZH>{zRaqvwY4eep+W@Sg2u)x($<%JE@YJGX+k&C
z9Apq>FzVL&mG;HRs41~1X!HK${y8?w9L4*E)3WoBzJt*b_jUT+y2EbQZb;o=WeRz%
z@i*sO(96CvZa!gQ0=o65YRHZ^Bfi46jx#tGWz(qjYb`Hg4O|^|sb?v+Y(J89{xYvh
z7_;NX;PR(G+ToMybN3s?<zM%I?<X5l<kV<cYAOuhc)admz?xbXQy#<EnC{z-iUk#u
zaY5dvRhq8MOWL3S>}NclsT3^g7p8t0Z+q7?shoVMg<)&uZqMDOp=lNwch2QI)9;F9
z;!W=t?L8u#b><S+lg=8Wf3ipqPPXbB7<FbFYxcRwpJ;ugbWwF;*J;u|nuVpP3eCk0
zXDOGus!AKrxqB^T@=*)!hmHF$r%dz6)+xO|+F+%IqhP*Z6*|UvO^US_JTOh>r$|IU
zGV#0zaJ(_i!#P){XJ+mc@<8|Ca_o$jg?GU8>#+!jHEJ4~&GB{)e%{%ccq%~$we_Qm
zQ62(~nzZOR;v7iSAn9C!6z(>E<uFEqG->)=EA^=C>QetP##M>jr$ulN15p!G9?*70
zYAPPUgYIn2wTlcmu*tiE>%K)B0ZdtjJGLIJ^Ts&r!tcx$N6H!^c|N!1PQ-@!m-Lyy
zVZ~3}V!)yWX_YV}M%j#bpQ}bqMMXH#;GlUsD0S;r{ZBJ#I&aK>a2{-)7mvvG4WOI%
zw@yE%bj!(sv36>jo!mcUt(bYp>{x5V6$X0x%6%5>+$U|c>+7p^hd13`3$y`)VP@@{
zLMHZRKLs7-=X{jwHM_Q<Elj{<^U4`VYr<|gh|7ME;9<LVoVhV$VleCekY8bCeSWBn
z%Jpwe2a|>T@3Gtu6^hXwQdvs;)5D^lePGw5(^}=sg^>u~Gd1!ruaF#SJ-;Tu^qP_4
zWuVzbm1JKNMy+0-8?rCMf3?Rxl;UsfR!@*A6M1DrmV4%;M-$i7)v_;rj)G%5ReMmz
zjMUX<jBGkwAZgh7oTy_8t!4)$(AoxH$)Hj0Z{`ODzeW6_x~HV%Gz7RTOgh0mLI~Yy
z5IyNSs%+NA#92(1__lWMMNy~2w6NZwNKXGA%Gb7Z7?z;%xeLibx|V8+d7k1E2TrgH
z6^dmC{i%LB-Xr(`#c?UE3<#B6#r*n>YwgmNwYB8-@AaRcyg=zAVQ4sx*a3xHii-CS
z>{pzZ!0x&MGJ#(5(()>uw^4hV@N99O1Dm--onkT@?T3w%+6#Y^&^N3~|9IiLQ_n#^
zuju1QthGx>sBji(I?w<3VveB}y)`!&*PN`n2vH*<KB7cKL<HdeBZGTLN!#+gk&=d7
z)PB8E?C$Q4Y_NsQ==&?|;cq66U8{_j+v7q#jgpMk7+0kB2js<61ZjV9f8|cHKlnKL
zwD8Ln&7qTLT!I2E{ag68Y}0%@`0NE8qY9awZqqf9SyW9W?k=KW3M$&1bkOR}AL!=N
z>krM3%1=#Ppew7>5F~NgbEL*`en39?C<SA%2-$6o;+$Xg%l&^Q1}9TY?xhcEiv0fd
zgc(*cP2tzf)IH+jE}Yd@JcZ-34+xH;k<kWj3HTOaeS=yNMIXREu+})@>Dk-%6V5gG
zCcX!iTp_AMz{O*uyN0JGhX!N`YR;&zf`Z?Jnihl9y_#|Ti&vO_%N=@AZ9<`q+6~N=
z7PZOCY;2EEx{9zIrCoWO3OZ|^!#N?o3XnI@;jvj!Vt_915Zz^7Cr<UCE}`E7YV3hb
zXU>#<*r6G%Jo4q&op!`o0>$qw>tGBPu%0=GZ?-KKCx@npzRDYH9}!dZ|MyLf@b0AW
zEKRu`$UfSZOA+wi+9xqF@x-VOIy!(j{Tu2H-bd6?om{!nC|~sEO$AY+7~fl4TO(RR
zp!NUGGX?%3I~5cJ8V6!jjRDkxfhEPvbo4>|Z`Pg+I>|K5c3ap;>0U0WZm)jS>c}yr
zn`FQ${JJH!N|^dlNBdjOs>5o+N=a6_<1(dZt+Rf14{Fv(@5w)0TNW};+TJ8r8y~{f
z&{@XU>$3Z-I8ahDad}Pwvum<K6%`{Y%~!wu+R44MldQ(~v1wu4vpGXULx}t86IwQ3
zd!T!c1_*&K;%R}=Fam)fa2s>}jYh9U3hG0L8Xy=2Kn@6z59b;XN~kYwo^&1*xY6q4
zFC~$l9h!W4Xfys{jIM8BzUSHUBi%Y@1*z8YJ79^C8W<^k;X!SPUjOgu>FH7-e$CFG
zKYzY-Lai2LAUAR+;`$D)$dW0B2g-sPbL`Cbuy`Uzf}$=_Qc}Jgldm@4eAZ|&+fsb<
z0qe(iht<ba3f=lT#b(=hEoTpM-Q*qd7*X20zwlyo0}<)C>g$JSPTKIB`wD`qn>@si
zk^mB=LxasfBj!<xYH4k{2oaJ@uwNb|j5v}JG0%2sG@|6nY)6gi4=pX0QYH4l?;TYk
zr+wmdJS(i6DZi+SRei|Y`+LTWfwzjXAtzXL^>dyd+m{zgZM{3D^-?^MqSU(X6HsL=
zP*up!&kyH$e1?Z72f(n3`e*cF04_sV{S~w(^M-@7d5`-<z(hjTp}Te(kHDt8raJb)
zcfb4?KRxnN`fY<N_v*n`neXLcKqXRaBX67WvCPSRnS1GD7t@?0vyWG5YO1b>@>u8F
zVF$a&e6LEKw19x)TKAA%aGiG#Ec{%0E0ZG1uT)@9+o_s$PP2>hvMm0vs1Tp;sHze(
zx0G~=iinNiKD^)FA&OYHV+7LHVBWB2KGF&AezcfbrMLlWHT`fDSDAb?Zew)c=(^TR
z=LdDCi)kO8Tr!$a%+}nroKLo9h<0W23^07+zOYXr;%^JGpbt|=c{}*T>_j-uQ2Q(H
zj2+Ugjr~?#{-eG9wIlQF4GiT$0=$X73sDJzusqxt?J{_F07fIo@kZNEU6}&w8j$e^
z2PS2ZnUMJZ#z_kr3}9@uq~ru+$kmI4i0D7x&jhk(H2utKWx}I3F-uEU>zazL*^ho-
zui0eqtpZz+!tDL9_9K3&1mP(!Utjr+ntL48Rw36Hjnp_tBgVLInH}~{IlNU=tz-(B
zk7Z<wqO@4CxOMAoe7;K<pQR-K=UB8tg3>i5d>#LztfJ<e+acRvIBdbJL!ZYJuN$eq
zFDWoFT6L>ZN}~M7Ai`bmib#Of^m5rDTakj|nRR{f#ATKR`SnhUEN;h9-JM?Q!Xv%C
zR1^k`1`QF9L<8x@Ykc25u+qu6M#n55U4OJ;zVG`m%9fS3L-mI5UfK@FRs5i^J8tlu
zZDUOq-3oF-(b@opGicx8FoDHuPuMjm7Hx`}2B#9ay^_Br{4HcX$cWMzgtC-f5#@D<
ztEyuaQhqu<%mpbKed_^x`uk>wQXYo*`jSymC1;wbOm5%4-5g^E%=$|K#kjUNNto-4
zg5{0Qq3)*AoSq$8ATv`V*}D7D-?GIXwY#icw3n3hSm>&B)_CzH8IW;NSr($dr<n5A
zkN)y-yiAm1a)>_8!Ltp*KA0HAWG#FSUL;(2aV?VmD1;$lYALRZ@k4nUuSwVi2Sl98
z07lC>Clbb448-~X8l$Q`_5O^Cwbj+)2A1l9?lt+sN7U8T=M8HNp9rT9-BkuCc2>=J
z-1ujg#}m7AKNTZfwI^wg3^LNEMrPeKYUm>Cer$i!V45fJwqMLG@hY~{)aUbN%aM(|
z8K=B=hn9r3ykejQ@onxQGQ@(F%#T!e1OqST?EC}FK_XSX2>opgvS~p*4@<d3F<kaV
zD;iUY!U4a+!gYWibLMTX#lk2t;DDTB<LT`kmAjr>RtP?MsnGA&_j0rDmi(RUkbAL?
zw6LH6W^N3(N%!tOTvLqpl2$K4+Lt3Ui$*z7N`D-~^a!F!i8iPwI1SgK2k)0^*&y~{
zxm^$vp~DhwdF4n6cPr|X?%8vQML0Y#kp8lKG9LW03IPv1h58^~Lw%L`lEa$S{1{jP
z6oTKz&=?+-Q3Q{jpg<1N@5X8_bbVcPck9Hm$&9S8UmS0j4LE-Cq`pq%aK_IySsmL0
z<oyHZX14WJxiFTpDU9?+Q~0vJmvV_Na8BXMcaCj4C|Z2NLYAIGi}SIe{&>{(i7wMM
z#wCAdWd+V+I^+r)QwubGtn88>51vLS0HR}4WP8E43LNtr^Btg6=S2J1Tr~yf%g`_>
z#S{vXM<16Yzvu^5w~KKM-A6lhv@M9lG}zzf6wxZ2PyHLP&4IJk->YL=!AzJB_yIWY
zxV6Y0G{J;<q(*Vf4l)!2^hT?^w(MCbpUe0sesrUL*L_D5O(=f^Hc(Y%)^2T}7$zUh
zRU>L6ewQ+^$KZq(fm)2fvxd*mqw*>x6G5}ddrS;a_HY*_<nY^Mgw?K~+(*BI<#wuS
zb_&>!51bL||0F<C0JFmgUw{~Y|9*NDk{O~iZ1612F6Yg%MB#y;4v^y>ejoNKH3r#z
z^j#AnQeGK3f0mRIA8X@t6Oc71qi4R$9#j!MX+w6>2I0Am>!gh@-U=aQ2p}fi80^0a
z#viJULxbO^XoQ&oIN!Y%neWp&n-rT_{46<h;`sp*=T5TvLaq-5&N1B*`OWnkU5*v=
zp<4Uj?BRHGh2&!9nzu(rHg8t?C)HN#z%cpW&h6x5>amN@PVXAb3?3T1koMsE=~5G0
z4Nx!UAQWktpG2H)F&YK=^(BS2Udo4$=qITUz#E=*Crz)O@uKmc(*?@9k}b8x?UbYZ
zukUGV3s(#!Qt~|dy!7S=_5+%y@O+H0u_yA1;7cd0@9f!4!)h>j*w+$yAD6=}{!9RJ
zBp!HR@<=y;TpU9vlD+_dns6O5A8VCZx<Cl=Wk~D4ork-}VdMMA$m0R|AO;TO>d<~h
zmWM+Nuy!{(L#T2K3JQQFJ}5W2Rt{Yf07&34`jEw-8Gv5eNgA4PL_7)#TL5ohgMzGj
z%be-O@RPf!!;U>!Lc^U0;u;{d=t2o#3jy^d{DuE0X$|oS?qxD(QOkMn+XxAgOS8~D
z5O%<<QP?D$`*C1$ST)0{r`a(>+?RoNrSO&l+rMvg<@wXs0+zbNE!``xi||WJTSx)E
z380bj<qavkWk9dh009H=!FehDn|~qg;k?mEU|ttVZXUn|D!GAqW3iLJn&)VH*oz%y
zAKt!f8?$Lu@4j!%?4eRN;mpU&lR0CQY4)r3QjyVz)G*I;Mb3b5Z*SZ+rS%ufYRxjH
z{khUIk)6Dmb@)j~;q#=BB&uvzd&8J40rSE~vLu_%8#e`cFQYx7LY+xDL1mO=s3xAH
zV3ydLFF934bu665%t7nTtCpX0$zn1}2R%4m07#$)vpc{6uLJ!BwfT<jZimvBdrGeW
zV>35jGZ_ZiEjQ9-_c*LF;10pAw7=$4jrCLrCTv>cUE3%O`c)6WcjnHfBF~x6YR%~!
z@T?!I2|gb*1T(crqX=p&Ab8p*YTS5;R1NRZM)HFk1L(Ky5jS@I{?dZoA4juuw4<>5
z9BAOpV!R*>b-T#VE-a9TO2=pdlRhsIbndo?v-3$9+vDsSpO`?i7<=_}US%^3jb6>3
z{ho*x2d?Hw3=TygkeMhZ@lHYE09_hq5*0fBNQ3vK91U}n6O=(7i5{n!rDb+{IwP=3
zAdU!|XM>87C1{tw-Pi8`L|SlloI`5`{|YFNI*x0oF^9lpIGS>|3#5mJ%2N+m*}`=<
zy5sjOUigkO^FfpAWj!YsmweFe74D(i{oV1e9gW}$v>fq7Py}3kj9JCUm$vA=sh!;#
ziYa7H4=t)-K55SEFME|(VJG$%q6p!f8x+0~eiKh>v^Y;6U^iUpqR<!vB?HWbCD@1V
zB@F>5D}_pB_Qs(pUZtwS99yT6Gvc04Qo4DU`hImDG1jDUw3qPUnzC_DB9(r0%Z?0d
zr|+Z(A8&8}dgUUkdm&Hp`Wic}Q0s@zCuXW@3AHtL=1!tcdkW4qezUo3EopL*A0HGP
z-ebo$P%`kzpL9Qe#>X$E)S0$Cd~XfQ@7y}%QJshw4Qrk!Fv|xCp8)XTz$Pk+H&lh)
zgZ&mY`2@7^MCkrYzktW~`T$je4b_n&gy9q32?3)-NQ0h01Ev<J`UKCidVm1SG`$H5
zmF4zev`|p4BOICWhYtYAK|^W!&!6Lv;UQ!YZAAPWuxRLU&?m%P!S1@cDKsP#zN70>
zQ`|z1W?;Mn9o}URG-lIy6`r1+dtaSlVVT6^N7G^o3hO&}yrxMNXP4qp`HC0r%vEE7
z<}xEZya(MNWB4oqZN;y^@Gt!M6jn4QFd*>b*5_d_cJ4>unBeBJN9)N9d~~Uw04u7a
zct>{6IFNDn4DlSd&O@(jsQ0R@JqRNvB`f!@!)Xu^rYVzs8aOiA3lHu*BX}Pm!kZTw
zAj9$merd4ou87H}xbAE@`wv(Fmn}YAjgTfLE?{w^$%UgyAL$(MXuyvsP{Ek`if3U0
z@WQvLqr(R0EMk9yyx}!nTD{@>#F~0tU$2Vl>Yay&6jI6fbO8*zn#FGzlJym8F#GtH
z@_sF_9@omicYcpnoOo%h0%ZOwaQ?>aNg~C}86x$a^>p!?g_KT1W#rR!`Posz#1@ek
z93hxva;v)$(|6|fHHHD;Kz<b_V9(%PsZn5cI&ZWDl0mSuSo3;c2SITM6KuK87DxzT
z(Rh{d3m&h8I{7(zdOfrOMp-2zan^8hf8*f8JQX~{`0#vcs4>7gD>X#x2C3MTtC}+1
zRCosnie!6V*-S@)-h^InYk=$!BPl7N>NiM^$j5rZ#tIrc3R(@~0*6K8Ga%>qIjo{=
ztrs%PZ*J42fBLkH&pQ{~b@%@LOuZ&LsElBrNU`~<s<w6#?G$j-9~8ARx1m2kh(#Li
zXUS7iQNjH1C%jGO?~SgsfcJo^Mujg2dBUr2-Gv>tm82&*<8Cqk)i*qx1H~QsYRMXf
zD!!(`HbI6TUEcB7-2CT9truZ;L!8T0vFEpT{b|lQ@@qFzt8qf3(njZ1`vrw+BF@bT
z*b04{dK{!OjBOrNZ4jhr+#UD4Nc}(~*+}^0s`*v^i6>dV-i<SmhbnXCw(C$B*sI>H
zh%VpX$<HKln*HyN<6?oPRT1<T<1dCU+25m0XSaKLo6fs3%``2m`^U=Ij!*;HqkGp%
zhX{T9#>T3Nx|B1pbEL0%*0T@zgI~b|2q%MVUMovxcD5$;fSw)(0OSaT{0Z+SHJq1b
z;Gj8^WoD~hN=1GKy{M8D1uwXu*k2UI{MOxqT!FU<ohu{{83t{Vo4HBJ$*6}N=Z0mZ
zq@>O!`X4JzwML1^K1GjK_wWF1TC9fy(p_Z2>!>APCNA&A$v+CkC_!pf%eznvT>|V;
z=djbu$7coy2QR@>7rT=f@Cad{NZ9pFDIbRfEh>=(T{Kz|+$oR_|DjNU{uZqUXoB`C
zDk}1~J3~i_r0~`2*C&&nwv#k_`P&9K;_!kU{)fPOMn-TWd?emdLG%*-`RAWH;E$LJ
zxXufq6X9e+O>N9}{5<2ft$MO>!)%N61l+<*{yn)F*BtI1*PmC?qlup7=g*(9NoCER
zN$#mP`*_Ma%kFE~p<Pon&Az|e#~uPW&smvC@X2KN)lvKeab!gPYXM2SB*rfT->U6j
ztcaWw>K|ZoHY_VB9<tBqi*l*k&syOb5SFu1wQ^~!!_Kld|7FJZ7t18<<rRU?Vq=G}
z(A+Wd1~Cd8A;`HQ(^J3!hTaQ0w`lNYl^iCfrlI4G!U%AkYNmN2&KD2~A*$qr?Ham<
z$CbSSve01g4AEzOGLeVG8YvPUj~TZ&5sTY4dO$Hl66E{t-5=~?5SD!b{{j0&h6w3Q
z_=)g}QEm{LQRHmjxibt<3ck3wT54*NaaaRdCnjmKBzP`LNzi!{4SB?8a7suI9N0i`
zk>h4Kd-e-*LU6C~OVGa?>n_aaq%!x#Z7r{?hzPAsk@n)B0Z)k62t7YVyiEjAcg$rt
zR<VF!h_Z{A3I{?8h<f2yti90c4hk9_ABPHN1m8!bDuDUYgQR344@LA(1W6Qm+?$s#
zfuNno<VN;)&V(9io2bw=Iq>k%{Wdl>h6)GJ+MeCJVb_;DDDvid>Hc&b_Q2r;Mv(-?
z)NY6Y{Fzms;Msza0;8%v^n%ccgoIO4O3KpQ-2Cz5coaJDiUK<Zk6UMrLxX}M7;(a$
zpt~HP+I!$Y5w;%6DUgt{DGB=#{4{~APISogi%X{cUE@GHTZgTIhC8P$44awqRCQ(Q
z;mm<#99T19dUY8G0?r9-ZEZ|}1m&;q_iy;=yTMQl$Ux->6J2r>91OHqZ<w3|&>0D0
z=YD&r(lIK^6k!&Nmw>+5+3#>w6hoD*p`n3$CWgQw6>=k;3qHiZt6I!6)|HfcbG4;s
z75uu7b~ik{sKGe+(fs4B;vvWG^;z)YW(;$q$8-a96c5cXefaYFKxm?+@kD2y`$MYi
z!^W3dFFlTb%pGaGy@y2Lki&I)jylu*L*b!<o8xx!*$K=H*;c7*w(?bSkLd?{$A2Y#
zlQ|c2zUa1%^~6A?xTk-u0Dt1QOZvkGB@gfWMYEve!kCxrQH29w*0$}O6Ej-;psLFi
zO5VPWhyKaX{;q4}Y<#j>?j3`Rg9McZWal(Fd*>K1JVE}K=ZHp9GU^*}*Y&zNDQ50O
zWPVsdc8u5bA?XWF4s&L=b0t9vCgG7yIt3EWwXq^MG!Gt)6iN8ZP__8+)EuP_dpqIb
zM^u)Uf1X?rEGqcw``6x>q&DTcLH;H&aCQd)>&$5I%*sE@CeFbOI&!$Eokol2&OCh(
zenrf(CZvL(94IBQzmn6+eS`hbHhL#puSsNuh3jeLi`PY`E-<;tT~pARI?cap8v{Xs
zb^~~UDla+ED<lN?w~Ba6>~0LTCOG;Js^Gtf1C>cWyfp4$*B3P2^`nVK${DV|q=k?W
zsGZjTus12vj&ua_1ZY$WUV&qKB)P>0*<46im~s5dQ7**Krh-mHinbu7?JxFL^SgrZ
zxp!~l?Gu=W4H$x1o@Sfy_WgzT$^1kTPBEq}z5`{Pi=d2uD=!x@pbCtOgC~Q6a)N-c
z+hF5)BC><j<Fi^(NQDmz!Jsd623Cq8Pgue%G-jp$qfdjNzFQKd;@v}*;tk{XY6Cf9
zv17@TZl9J{n*5Du1ZOo}i2g1(>kc!=edN5eL)o+N%^Vojkce#KfX}5@bfWa|W75C2
z?I3rP>&73R65}C>FG77YK*%B@)*Qis@k=@2S{l1^8{1(C5v{NI!?wML{{JP^`S-E>
z=gN}|X?bn}l7w3JUtyLrp?xQpVB5CsAjU}_lu%^wnK_=jwSUGxjN37(>m5J~wz7iZ
zzv=z_6wu8}7itVh-7>=YKclOvXl07Fg6d^;H8}bL?seERG?eP($%Mc=u-?GQr<6eG
zkA>fWv8G)0=?;VM1A3eS+Ya9n6T_+L^PmdWp0hY<RN@G~2euQ0NOkM$e>ar*FLB`i
z2gHH@@6mT|R&?8!3qQ^zrJ>1dJSzPc3Hdm!J{$S?4iW{<dnH09`7%euBb{jJo*ZFv
z5_usa!MX7KP-NVXBh7EiyFV#jl|D4`r{KqC`eHX<dUdN-L(gI#$rW4Bpt{<-3{nHl
zwoIR{3iN~5xb?OCj<BO4zWQ$nY<&}!M5N3A;gIy7@Pk=9x`;ip=0UW9fD{H%V>p{M
zqm#QklA^Z8UjY2jvzKV%#Sa~?VL%=BeieUwx=HxFLO><a66E4i4PcLoib|s86u#y9
z)e2)?w^cVn^>ZYv6@ZKvA*LjMWe0WT)BDz_`H=t`C$!!U?!wJJo&H8BER`r_X&c0R
zE*v^|@UB-3e*P}Q)>C1UM6U=U?&%50{h=i}(L+3z+i$l%tZaIIeja`9BC4H)oKxx4
z4Dp-%zAeeA0P0v<UzcbkTo(ckZaqP9Sbzw^aY<WcxTm-{ldFUpzrF0V5H8}Q5OPNx
zVr(%x{{1AntxqDa3V2D#bRar1D6PdOeU|(8V&fo~st398qtT7)*Kc=j{gS;~zl3C9
zaIh8V*_W(|rdSU>KH_!Cp2kSyZFh6`QKWUi5*DE<pnwB^&$j2^>rDjR0Fc;<ahVu#
zbAGN{H#cBvZVu1Xe2riwRaIh+(t}ax-66dq7RKku5!-ZCTvxRZy>`sjrCI&^yi@>T
zKu8o}Yya<^@cerxHMO;pi_@{7Q=v=nU}Ec&gtk7Zam1lo6}SVCcIB<hJ?lv<!jI<2
zD?rREVBtEEgh+mO`QKZ=%*dEX04J(7Wo1ZOSlH(Xi_Y>>#4Sqi9yaY$1=x<oaDNz%
z1lwG%5x?X8K`Xp<oD%VclBl)^M*fFyJsQSy^1DfBpHkHM+Ak9i?8Vj<T%bgn4S<J}
z|0Fa54<DX?=!oy#J-&5AuU?=-8y+4Wb5mKF48P&yzZamZE0zZgs1{~GgIiApbI|aQ
z_PU`)g3jh`HHuLaBMCL)VtIb(TaP2>K{eu=OqhRtQ$&i6_Isj&652qZ{vN$W{KiwT
z<(P)Kldc_)R60<VsvlDLTYUX8sz;ZEnU}kq>)*fFHna7niU0ZbhIom@*D0E<3&x+G
z+HE~T;xGT}2mkMUJaesBiC5^s0bm|ZiI8qwckTTtTb2%oiqpbae}4{GudihGn<&cX
z=H?Q<-_z4*XP}@Xq6AcYFs&lVo0=kLPHlz;C)K_a<+Zr6)l9#r^<`qC=n`A*>895G
z`vb$nO3<#r8#mLq%Le^3f*_091ayArn3I!|_9kefd`Z^{szh=8K2Ppcx$6m85GS8c
z?zqcdkong(t6jv}zIq;_wQiJ;a67_YA<1K%$~YtJ46HWly)6-skGm7k$7Z;@kg#?j
z$vZZcM~pN^na4P+K`fx1XRzU(I$#`qH){RolVyB0<>e1R!{5;b$BJ#Aul^g@_l5&*
zN7<P8-ga6Ne%gwVpaHZc$vgW_6-nM-l6|^*R+r#T^&v+P8};;s3*W?EL5~eI0^1@v
z-U03UkggYC=@VAbd-IOrv!9;c`dJH*sPKhC9tHFUoHWYF0KvgIZDD3Wyt1aNt1B_#
z0uX#aJNUE9QTkqZ=!J%^qrL|#7qsFqV90=u4jnoKy!g`W#dx&&&_&9GfgEZVgqDfP
zN#K+agv^1n4}fbLfF42Du%3;%m>&ny7!=9CKOpcM`u*DzS_(uS)W^hVLzs<}(_F_l
zK><V?>RMm$oxZ-lCRFCYL)sD*>ma1SRkgNW0y~7BeguCD4Xz+YcL1Teol^mXk{FSJ
z;t29M5W=uzP_q-9<w<6UV;5mL3XTpiT2`4zA&8cNnSgXbLrzZTkd>UAOpvB=Q)rUe
z*xJI11W-C)4)!Zo>`=177D(lZ+yzM!4(n?7!mG7r=?-u)k`B1VlA7`n|M2-F)wXRh
zbC6|Lre2YlErBFgZ^dtJdb%Ow7;{gu7i6CxRqF$xdn}7FD-GFP(O5gTo}_{@hd!D*
zu!kaMpyEVt&y-UQe2{b8&@RNC+6}P4Im7(}9w=~#j*p6BHCU`?yR)EF^zud9C*D5x
z`6*5e4#2?zd?#<5SYCPi$eK6QczJl%mij#KwDdh7G!V-4L{~sx?*jlJB7P>lgFU}i
z>GqbGQ4IaV7)OZ;3&>tHH8q*vcu#ReiByjyucqrL)qrU}(y1zsJl$>Cy-Uu-s4ING
z{7m>zxg+rmhB1>`quQUT4~aXQ5b*!_AV^G4qN@Px1|AQ(375D~+kQrfCse(#!m(rI
z)@$qs6k8t_dT13A&DwB5nxjWQ#Dn(!1&q>F80?R>B*oa{7_VM>vi%yE27wU30X~5L
zp^>(BRCNka{1vp1hG%Z;V~u_UVFKg~71hKyI(bri(UHyuUIYhJ^%)sHzI>3r8B8+A
z&~k&m02&x|iDZmg_yxoh7uEWBYT0XcE{l`gYJ7)jSoLxstWdPJ&OkR@5N$Quqxv&i
zx@djjZvG&3F-lJ$XnPLpHn8@YohU_{1Pu`2eO1|mXlY;tCn)w2(5U@tmdbBPqb)=k
zbsriSpp`WV{*BA*y$*|s4hRX;f#+Q`EEZg!lSKxwzf<A&3sQj}8G7CbObhy0CnM09
z6&G)!YvWALoKTnszd^9gB=W_c?u*SXOAFCveK-3vDI8wXM>U8;Q+f#a?a<IMa`J_;
zW16=Vl;G3GyD&C9yo6b)fqI!=v4(6Au?BUOTYwK*)GabAeY~1lo}QNW^+g8&H=_Gr
z*l}nTb8;X`1)~MTL2QgC=H|_oUx2%wz5#89Tt(Ae^bjsj+JIvNOthBUEVdqa3>3g)
zcVsOe@E0<NyrQ&y7Bci;{Wi||KU;C{AT@Q$<zO_3?tnIm4g{^EAW*UE{!R<f7Z4pn
zAZm~@5(IKIDWPbwFgGXWlHI*~7uJ<92DBoCY<fZ8YmcLsN)X5;x|2KsuQtF%qVN(i
z&k`ZMWiK?45qfD?)Cn6sT;)y^mffXb3J8MpV4vN{ZdXSMA2xN&;OiPQs3hA<{=}v_
z=^eH0wa_7(i@!Js&mghtpQcGe+qPL%59f$IC5Oa~9;0p-3JL8kBq(*32cr@1h$|w7
zUfpTE*z#fOyi{1T985cznV6DrrlBnbe3K9W$wZZ-`wp#3a=@gzDZ@nxvuwr!boMb1
z0T#xQ&~T$CKoB1!xX>m@SY@;TkN?_1JpD4#coyaNZ8itMPr(&8HWvbzfeB*)?D2%1
zj}nfOSVB#WjpqzNT+%yP6T6;URb9;oejk=>L17`3QP6Jopyyg&FP&MooilmpJ#>z*
zpc(~Y|M1}uEFt>)b(np@sv)8^#G#PI0SOWo61t^;Nt~JeaUw^;y7R7x@5l_hp|76+
z--=~-^}%_{y=zyvH;9LI{pPC80b=P*9n_wAj8H>IGnET1)U-PqO&64*cyja`4}fRL
zLLufji8Wn=2eEq7<!#ST5h?)qdEo933w{{LwnWeX6{UI9U}d=wjFw7ugUX6M&mWEd
zxS06{9vz@Ik$H^zdO#@c9rMo#L5G4I2T;Ock3l<TkY{nrHmn!@bT5a6O@BYTpV;vs
zp`r4Oj+jNEvcLlq*On@%_(3jwm!upevR{2A#z`YKn|J2%vZ;lI?``?mm>6hqI^Jn-
zZ1!6lkX#m|*kKzM+`n~;Z{~^6_LqUd7z!?8k|*A73}F!x+*=CaLr#8vu>P^<r_3jR
zDyOX>IYr+c$YD^-{wYlJy!(0E^x?f75@A=Ck)r6o3KCgD4xx}*PF95B6q$$7WCMEo
z^Jxq?HK&vvuwwBEF4YJh8(3~*Ek5LswgtcivPd=l1bp2(M$dr58^o{)H%esZGRzpH
zx0{sIKgsY%V$`|15hFIJg@F;GsS@uCzbRxo6WDF#oW^<;8Xt#p_iaUs%l|Fa*pHhc
zhDL(&e(KaOw1JmfaRxeoCkJX320zM!k_ILQz7R(3xXmpsrNqS8h5dJ6h5r5bxoQBL
zI3B5FCxn5!ikg0ML2DAUjgHRFMBgF4CB-`c-i3r-cHD7AA6648B?uKYBr?HhOCwa2
zL6?-#hDSrJ%!~xTxTmIunU`0&LVajp0BsM6imITmWo4il$tC|mc7WC8^@RaP7c)x%
z#9lgm9hy`myoLMw#>WeA=)sXI&NntT))IG&c-+CwYF?r7bSU@AzjsV5(t0UmkFerH
zRs|<72IZ3S@<ALn=&#yfukP=B6&xIVh)V7}?=QK7FVXc5e8jpJBzSP2dQHm$xTcd|
zA#Zv;p^Fv(uxEwPU*Fr?t<jn%I8quDx;AbE>a6FDv;<x;PO&ZNSy>Cn0r;iRKiD&{
zTkVDD0tYMW6yg+4i^GEFB)j)KhX*c6WjU=Tl{jDFe(XnLd^@&m!q%<8fC4)9rq{0)
ze~|pS;y<w!pI%J^Yu$~SP=R55d>qlN3+W_+j%aHtglhoJ&(UB}yE!|nq(5*gDgLqJ
z+M`F0*sN3^%L0(?ok4kqVm?Tpm576lbJ8$!$dNYPp-N?$q4i~FW+@CY<l>~SEsB6}
zbIx%MYe=uSPEwXS8RGQ3sdc1(d5Wz<rfMHf?e-G4d!VCpYvBqjt5n*d=|pY4*zCVT
zLPC(!HbK*w3H_EUY(9|S8FYO7Lv33UHk4*hBn>$PXmV8@CNMasp^Wvw5l5()y(~mr
zkDunHegrNhQeX~~%{dtuELP6otp{pbs+9MNcAwRDC0a6Y3eEfqb`6l3yzKWQl?~=Y
z+C^k@C-or~_@|4}0G62Im8O{wMTiaM70<1zs=n*CQpZR#E1Ciz)`Bk@M~oN6waQi(
z*k0EXSa{A~H?T&WSO<xk@aZK`tEy^h$oy?>Y>MDFs{ab`0d#`?(VBUUoJVDHPsZzs
zcKb!2TIM|2hr>@xOH2P1w5jkRYW=iBqB6<u=6Ah>_wm<6VTeyWGxn(%QFvwQ<hom)
zQ~ouQ><?p85g{QWFeYE&;2@}tm<QC5W)$(iUJHojNR_?q;?NS*r-cQ`C!HFCBoZzF
zC%+Z+<#9?>J^KX7%EiZ)!qD8En-#zjgz|xVvAf0ycd0M#qN1vb#;an*R@j`{#-NfB
z*WYg-79>&zru%4)LBv72vrK>HT0_sAAh`LN#99Gx02YvbPk}TI@&1q;9VSqdnTR~y
zrwr5YlM&6F6YuNjU^uNIehZRiaZynjDz*!CV$VT8^JALo%Fo_s%bkdTkKRx`8ICZu
zMRLw$Hb&~kI5{{hBIA5)ZEQS=Wn$4|5uqIzo6X{Uls@vfLyAHeSZS1$1<0K2R4)`%
zq<pqUSr`59>p>WXmyNgGt%)Lzhg)BdGqDZSi|mxsULlVk?S@#JveMMn)^>t|n&;8g
zG&?Zvkd?L-{^Li_$=Mu`Z-L2U7nB8{eUk`wJ;^0z`gEvxb0cvz()*Qru9V5_<42Ev
zfSrJfO2rya6U|#lUVTkS$X8oisfGZs`MoBJzLSU$XngDE2@(lj!Y1>b4{vq2B)kV;
z`XrZ1Gok+vhc9@(N%mimyvwHjM4V$_NWe)Obcsnmp_#C}1lPQakG>dg<z2htoUzKJ
z(h3j*5jKzca~rgdBVVXbjL&|7tKz^_A>@N@^J43&&>6=R_IUBL|I<~CqSanlSO`w?
zV;n08^S4?d;Bj|zlV=tc69X%Qyv?xjgVWNb3>3$3w-8O$&{K(dRPuZWv?%DUHfQ_>
zDF=jfn#woG4IRK-!^49esSkrZ{&j6jOD6sy-dYb3NBjq~CL`&o^mx&;GybzJ4-6!|
zyfVFpNMW2{y6_`+adAPqgWfYNwK5$y9iY#BJb(y+52a{*bG_sz4rCN+pan&tldM00
zc#qwL>apz;m5aT>nENaqwYc`kAr!ytcmQ>>FuI;c7XRnZqe{`Jqv+BnT+(Ia=2m2%
z7#JT%W!r@Xh!Ywgy1KF=8N^8sE1D)YB?->;5aasygj)(a@baYBnBA5zQBq$&3;8Gd
zx*f%in-DQU5e7RlYZSJGHfO3wb7oiWj;e3pmN3?PU*{!m?nU&5Tjz8Ej}gZ;ieyx2
z$hf3}FD2uHytoL~5k_2oS&_aFusV0?4c;o~f68jc+S=*xX)0K|AjKf=U@-P}>&CXw
z<HCV~0nv?|uAIkFI_s#CiAmy+l|WC6ibn;GML3~~`1q}7KJ-gm&f@`>2Yevp>C#I0
zKB}30olmd>0fn(Yn1T@&GOC1&%t!YkCzq&&Dh#K3jDF3%o{qe}E4zS5ZiQhJl*t}W
zPKW$cRWe7B16;mD^gi!blXqDLK|yiiM7*s%7cq$lR+Uy}B>u3Jf?MwkZer#Wp+NhQ
zg(GXCHMtYDz`kx<Ev?q3&Qj$W7_5gNTqqqNu(V;qNsRzjVs3ldaUyneRs07RfFm7O
zh48kVh3)ziWgMaF&s>D^S}$lt{TQaZ5kng<GApIQaR3YBM3xb@BGMD${(^%O+g^h#
z2DRSQx7`9};oW}UOM5yl_Y${fx^;W!u{lx0z#Kem;aNC%3WPbh!&;8tWiU&P;Aj`a
zMV`S8!+P5nZ7c`3YvJ4hfq@Vc;9brp+kvnH(ghNc69s&MdVzGTnpl`X&tX>IfcCgW
zrNY<pa%j|%cStnpteMcJhjo+J{`ZTxev7r~z_!nPEN9R55x!a)g~ZOls{*ET_RN`P
z@<;XkkTnqM1sr<V%dka&@9YTh0k|Q6^4kx%3*X^by2NtrAY6K7kUJnK@7lSOn9hhG
zh*Lc)%f>FEr`Qo1YA3?^7C*nfzD~iVFBfzB1WInCp2&>g2da^0PESYo00|y?ylgq)
zJ9<$EqWBy`Fezcijo^iLR#conb?QC%Vn`mDWg=e!W=>8vLV}LDdO(=5*kO`;D6_?)
zhw9&AuQ=190)hEzNmE*aPstG+p<7>5Nd$kaGWVhe;9lbEKesPJWO!R&|IfE%{_9u&
z^Y5eozt{uI+G)ZI`WdjHA1~Dbst%$kBgSJ&s!fiJSOXMXUd1nB`UyrXusc>3R+Y@S
zo7|-MTc!8kD*~q}`sO>n{;>;*jeQ9I&Ax6N7~krensDT{w)&-7Amj{Uu!OR)`qz%`
zCJnS3{OGy)6?-I(|3~Bdzj^>OkL#s$b>mt#3B%>*leK)rJOG7F1DxYKy6CqpALrD4
zH`>`+SxKBr_?xQ&+evr0{QB#H>u&09F#|z0ibd;?j$$uKT9Q1?m5a+V1yoWle;JbG
z{j<&77w+YlDi!Z%Wo-?oh=lhm|5PvH#6i6Z107HrL9-c0b%bJUc5W^SKqbOFL<i&W
z9ob_P^aa^a6buU@BVm|x0+NM*fMH-5AY%b@L%~b9(%?U777swAMAGrCwpL+n5fKZC
zd<RZcL#mHDkg9CJK^jQF5{g9Rq{z5XgTrb7z;QB;7aW;0b8}3?0)M|cUh>b}{TWu9
zGrPVVyt2)N-=($!ys>!vh;uRFH;Op^e1cT^BOy=UShH^N0%H?pg<RSUhK-cFQ!%nw
zK7I@iJEQWQzLwYcI{}-yTghy^yu|G970VOLiSQsrcWSOnAE*Khx$5tn=H|9U#{k!1
zzp<99Z;f`^_3Nc*m0%|C<nS;L5JLhkC@Hz-9~u-TVMYwe1XT@${CEETu_|FOy7KZe
zMWuiW2dOB0Q+@#!fl(3kgN=pLFz>+yVm8iO8t1of36)_#ftDkABUnUfIw1EF)BcbH
znfJ_h4)s@hV^VGzPLybc4g3#Kt`I>-2SClRjYq1zj!X!`lD!|iuWH5G{<+#O{r0He
zuV>$`J^NL%&GVnL20on1f16ho(W|l;Uz9g^8qcQD(zuQ~uPWci1mo=LyRJQhd@%$j
zMfy(Z>4b^doDl|72`5t(*dsNAOCj7L<H<J?a2VC#KEc>OPD7)43Tpw$QKEd}0-(Lb
zSIzfe=njBA&u%$w(v6^Ve}4Lx*R*+2xd%<E0y`uS|8!XM;kIF^nTN!pS0^Or0Z>Jh
z4+8~si{y5*MLXVX^JWvpE&Oq3TRz6``Qq=Af3``W`S<DzRuzz(R26nO(t?+HqFzo(
z=?8K-iDp)%5abi!0FHqAez~<6DoZ>Yco^y!H&H@TaS_B_SeS5R3PSLMb$mR_2v@0s
z2_!0}@G!7TmobUP2W;Q517l0nl<y$#K#QcTyj;L`feHBn@&dHHP6GD95gd6PwyT(+
zhNl0Uzem}nl>!;8PMKPb1+KY-eY@uL>#wJD{e4fbpF;k7Wqt;r<30ms4i3VfNyKt5
zxjz02T%yguo0w%p$$Mw;l7ulJ7_fjjHCNH|Sq9wXrgv>E48CPtF(uda(JDiq$GO<Z
z$Or-yOJKcd1fa12#I6d~Ip|yfX5RqG49%O?Y!gl48mO1Syu4_(<2~PzouonsO!&PW
z8;<Z^z~-rcZbvD-8}{yT`*~1HU}w+oGk%1ZiWDZbVf}fTlbhQ*KC5g1-7dUFQ#5Gs
zh+$^SI9+faj}9bpe?z07pGcep;5zj<7Ze7h{$Lj$MIgXv5b*zzl(C|u#WmQVzSywu
zb36w#qUy3T;Y`Tad)4y?bdcOt<;LNb@YvUo^LMnjiz=Ww(htamAnV<)#2VZez`n<d
zJMa?mcwPQ|yw869b#JqMgQiDXGt+TTylKygy(C%<W8nD_l|!=_R7*FsPkyhiCT5##
ztgpK1Z6CagPKDa7>TO7`k4Z_h!#5QnhJGVO=~FK#u|(Gm|1?29nB$=&ylndQn);C-
zd;9stDEt8B04O41%BI{4687X|H@NKZ978ut4}eE&>$0Xlpow)LCup)oFrgx3yKCqY
z6{UUm?k_*)%}_}m1QOirRPEQ>Kpzrhe}kx)jR(JMOxPMB<EG!qUuZRb4+-N>+Fiq?
z872RCzMxIdAa?I5D>fDXo)7-H2e(Jutf40P=U4*SgF&5EC8n{zbDmCj*V_FA3HVEC
zDdbv{J;mX{!D9mhEdVo!ekba?ro>Q01!zwY!B-$DLa_Y^X%L*Gu$rM_mP=G%=CIJ9
zhD;3rI#F!_qJa%F?A7`(_-SII3wH}BZ2>3GEzdJ~*$?`U5Pk_*5QHTj-ZXl9$m-FO
zfzb@Q0s9UdAUX*1mpQ!)^wFzBLc59}zllhLUMC$bZEjAEm8~t_c>#(BJeH<JaS<@%
z-a0$qVdNCJ!~5vhXY4g(%?3JrV%R1g?(V>*i!sp7+s6lOy)o?j-28k+91+-{I3jG&
z40hEM6cBL0{o?c?CgGx703|{ODLlLz9OfEsW@TNr_rVX3LR~?8sEdp2!bhD-FO&hK
zyLW^2$!;hhQ0*BI5J$N)P6wXP=uu>r610D{9BCszKe+6GB+j6ic}X*v&Jw}^@Tn8Y
zR#sL3Hnr8%NHi$`YU<9jNMxA+DVi9OUbTsn;3F<J!SmB3=7}_4f;!?0O2m2f_~ds4
zF&h_2<_IFJBUEX5hL|k_FJ%Q{97kOM%B!3l9-uKCPnvnQlU_b!#l~p1os`ala{KdL
zBzLzC!X3}wT>oHGIa*@zoWjpwcw&4)cgkwBY`rLWl`fW+kuiraFFUS2GR(42pEhy%
z=<MUikLd3y_Dc)~w*{t#$6P54dNe;CmA+~`Y3n^N^19oG$7U|2iSNDX)!cgnYLseT
zO<t^m`2|yhMSmE7^X4uyQJkrJ=EJCvGG%KYtf;F|p6U7``Nnnq(&gBpv#BD~u6zOK
z`_uKEHAb!53{;9ncHI}epjlw|=}cM3c<3GWJ7LAv4zwFpaLeg%G;<Uaao%J3klZ?G
zqAOpq`ZW@c=Cou{wE`=d+<b&2(bi?6BVhs4K3Q2=&`wh0>mZT>XwsU&ml*??ID!65
zMrLLxoC5t~&}dDGQ${y&q9O+!kNfF><K*nWRaVj$D@vh`#F5@%d77A-G!gM4KYtbJ
z6bA>#ti7MBb?NNP3`!#+xtZvph0P{BrE_D*Z!aa#V_HfM06mpFVyHC=>d_gw;Fn7)
zPhh<?(?y8dRSsM$sjFi}lsCH3BK|y~F4@n1;>V{O{?n{j*sEBznHf!BGfgX?J1j%#
z4pUv7bi8=1Q1^w0eb^P>Esd6NF~qn|%-!f2YR$l4@zjp};90#0-b+Gqghmk@30g*M
zTO~DXVBPE)dZsw-$)+rUEzs$$0Rm2lq7MxX#Tia^`0$x-Z3OnBWK3>!7Mw;|E72%V
z#jF<RAgX-LpOpNpJkaFad|)WUnM8f-)9FsZEvXjqzQ}qHwa;$qy|#Z&J?wvX^UqyM
z@dMOvDQCLNe_d<p%9Nbqm0v0H`o2lG#(VAO7uWH9%)yj}W^x6hQ3G!JOevoxdM{1i
z*brP*Tq(MfOq(gtGrK(UCf01WS#a^|9S&K=jr6p^-k=V?)cEt4TbJopg1G}t&hb4u
z`&-1tm-(Ef{>(oO&0ne_eEq}K!a^C>qY=0({TNkGu4Tk+P!wqYDfZZGaX7`=q{*^k
zAd@NFqQl1_?m^G&;6Ze}#n1O7EcXFtg}-oINPvhTl24>fjnVN4vS6C1H1%?!ACD<+
zOPE@;Pc<j+E)c>4Gy-bYZ=Pea=m)|<QlrsR2cZqm6U+snn>UB_)YS;O0f0aIh4H|J
z+TC<EEfAX!G^+`Y4^T!o_w)z?#F}XUex$taNEhg<5OP9MRaewfpZiBRuUYxWYI8A|
zm*aI!U@$l2`!mS|5QgB4KkGY|l_4)hANvQl$p_YEHMs2qlTG;YmmH*r=t9P#ts#IJ
zXA6a(ZuA1)GKRt*$LjoOm>#_dcXNt*Cny-O`av%Mb4W)p@f$;HU>hwl(MhD$y`>26
zw-o%C1npNc_Tjij54>3nhrw<34q7fQE+85hKct+7IR)C2-)H{(IoM+mwLaRW+J|l$
zQd#*mb^Kz4&(X}>QIl94Apsw?>4g4V{qsIr^7MY{+tj?fx061&D^v1?Z|je*-azAB
z_1>4J)$bOU${vO1r}o8h|5}yo;l9vxby{<Ra)R=WJ)2?MQ}r^JqulnIXk5KKVf663
zE?-ofx=iKGM4z~{+^GkOyA-Ey%uv3CyQmk7xCqzD``Y)<Do4oVim0xdB=>8Graox-
z(f*@GmqXj4u4TSpHaD_w^w;I0ktGTHOTOj|df%yQMw6>5(4n0~RszZ%T0$d;<On6`
z(1I<%38n=_@Fw4RNWCmDHpecQaK|IK1zXanKcKD^*e$cu(hguXz=%=)5;Sx%cxOS=
zR^Lytuowe|axSK#qy)DJ^2<-KTLU-`Ru0fIV7DV^W3i~ntw^Ml%Sl<G-T}{(>;)gz
zC*qm~n!f0Wt}HEGV8G}$V%GffLdBAz5x8^Vb~8?A2D8%BDR>^;#Q=hTaZen^D71Wf
znt;o}(eV}r%VjH;^O^R)RS_*xG~hyM;Q^9@ekF2pxi32k`v(VsfZ=Xb7A1fa#h^q&
z(%r>P&^8dqGTKnPm!N<s8^GyIV?XD~a%{~PSXKiR;}gZa1WDBd@e;)u8f<`;9dOtY
z*(LHDR2{^zi`&%qHcP~KSJ#}6(8PNA5(A~gNd|SBGN?f|LDMI*^y$-WpgN++krG(`
zfNOxj2(JPZ8tB=Na~p>&@kdvJCt&<DCt}S6QmtM4PwMErDVx$P`>t!k-|}h!{L}f9
zKQ8RPjzY+H@g%qiw9u3IgmN5uzDr%@tkngzk{5bsccHww!o*-@RYCkQ;!nU&Q`7XB
zm70gXsSLCUYt0kqo3i5BvMPUgj5#J1x@WsA_WK>CG1$rW>na^*aDA@I!wCy9USAbx
zE2Bc^LaUe44kwZCWLaBUw+q`PKQG?a(`@PO@<Mz=o}^UO*Q2w*+6;^=%!4CRf4qMj
zGzG|u2$VACQN*akYMxy&k%;47k+Gu<MVO0PSXe+|1(X|Liq)^rf8GDei{cspWbnu%
z(2+v@{j(u&x$o4^_6Dti;qF0Xz(BlrT>F@n<&NGV+HOeCk<e(NM~pt>*cg!b9<XyJ
z;F<wU9K$c$x`RBWa^N9@^p!=ki1Nr0LOF}i!PI%QfHBPtt=)%6G|*_o^7W21CC&-W
zmm)>_Kp&+w-YqUHxH~y56N44OT1b$i=i}o;F^*t}(T#)v(HWF5tUr)hbf>(g2?6oi
zvRUT!7emc|rl)1p<E0P2%*n|?O>nXm^-U96yZe7Uf+aj6T>YmRXEH|fcmhEJ%OSHA
zDeIU(&6`F%9Dxlb4=L;R{(%8d9Y8VG1TP`~iG<qlzyPgsJMbd403V+Zm)u-<J2;su
zgO5QYkK7Ov-zLcG3BeokRpg(9?GqYq1hE)8F{1x0>2yjDr+Na-eZ*&9$`?_r)2oX}
zzgD^zWsb*plh2jxq8ZN5#nzdL$3!=drc|!8#W)55ofQ2l$?$w}_0K83<ri-(c})-E
zWb{pG$=A2}7LdDqa+T9qx#4rkZIzeJ-p8ky$mKZVlnyoB){mchcB@<Of<l&3yC^f=
ziPq=_36Y`TnInnA3Ku!qnm_Ufbz%vD;6nf)Am1)UfL%!caHrGTxAg;Q-N>b(#w;7a
z{?bU(aq6A={{8kq&nqPD$BsHe)RB$eq%0Q@WZW^~1#D{|BJdlSGpvUotZHs9LK%i9
z<Kyi$e0c4|Yyz@j@r!XKwYA20I6;$%H0(NWO%KfA@V{cW#MYy6q7T2Zw-l|f6F+}p
zb+piJ3qp**vYa;>LJAJ+gTwx|*49ApMsY$CeGfP@LxI`f%#5hkXmI)?U?<c(7y0?u
zkwU@d@}MGSqAF69E+D=DCU62PLfH&U7iz_t@85~mA*!vi#Ssv*S5ZXdSxQNQ<8S~r
zF&^pPaJy1c$J$sm)K(PCBnA>*SblGasT4hMNTtvwb78C;h=*KKL`-}sun)Oiw^USB
z@ozP%DeLkQWp3C{r{b{3K$uQ&I@CaIn3)pu-FG5E8U&qzo(Jpk5B5FZ^?X;Q`q9x}
zC33{em42i{Lft(;lvo!+CJS>G%eq;EUUMaX$`MJFPZpUxd4lR#vdU?W^Bk91Y((Z<
zNx2<<=m>jv2sWqWebg_Ep3hrrz26$4?4iMbIJ><#rJVPUTgGzfpk(orJLVQm%M!hN
zVKy3MyyE@)YMNQPf4EBb6N}DKmqc#m2A$E3cu|vcJ2~ePJL{fa{16=?=B9t{VrW_N
zP_}^ET1LZ+NDH~V!KXCg;puu;!}(XPoy%FXi_>kE9_&_$GmlYywWU4rHDwq^&khU%
zeUmDT%)n6;L=7Ee9D7VVgxG(PC>}d{|4;w;xS^c^on0x{V{{EPd}}OAPibpp;%ORo
z7bs6Jf{~Ib#dZagFv@<o;T*&68?Zbmg5%Cz#R+807z5Wllp_J0BaSJzz{E5N>QQ&W
z5_GAvLwrDh$G{2DG&?c#BuR-F!hi-3IqN>h)di!j^Y01M>w6Z$m^Fw9K+s&(^KK<Q
zO?guM9cj?l>K_<I1R9YnH2ol|W}q_FELv04w9m9%G@ce_n`XG{cNTOPAPD+>_c-@^
z%TPS@G0D~E!%@@!HT%-=1<<_CofEP+7=1@WqZ_rI{1i(dr=FJ|=Sz)q=LL=xH}wc4
zoV~qY)m?i2VenEnjX2U55)@9yt^_b_?Q5Sa$X_UkROXfJc+?M;VvBZb>!Bu}*g)Pu
zQVju~M`NGg{yF5K$xpuhH0{~@?N0aI6gjC63G^<S6>qLC^(-|nWh7;U6omM=`XugA
zbmU((@aVCPOZl2m|2|2QBXh){LDMB>GTi3u^y|pPrUSS6j-K}vN|{hrO%}@fCotz>
zZ)qRN6kShERi(3(TaTTVSuyka*Tl>YmCl`Qds+*XUzP0NdF+#X@j+9L=?C>w*Bmug
zIi@A<JuzO0QuI&=QMq`<F30M0TZ7xpvHY7S3gnk_2x_UdoW;PZonQemdFK8ha{~i|
zfvX<CS0|blReTzwH%%%%kmg*mIp_Izo6DZ?N6bF{Rme05a4kCKVyX%}foGh@;6DDL
zP#l;C0=2>#^M)8T=1`eltw-gOR2XmPq#5HyEAD(&uFw@_gf-m-%ny{FZzWS-1KF&%
z6q!k4_vgrZ*oanI%$bM0&H^F~7g=AGi2mHgrD^S!O|b)SP_~6GE$X1JZ~Dl)5k1HL
zklp1N`;C}^sSVYLeweO_(-3gmNyMO|N6#q>g5YFffA!iy;F7ck`Zlt0EUVFC^M1R`
zP#u-4UB@_3SZRD*q7K12P270BI6E5*NB9z3nX}8@KYnsoD?Rt|*<|50QBt{SLCC2C
z-%G`a5DQ~h#KE1UbWTK-Et~%N*Y8j_{PE58VTsn@x-)`D9=iBk()>}dG*!mhl$mmW
zj*fm%!&OCV`b-(W`nmFZ6n(9;s)FHnHlMsLpquuTugdjNak{eWu$iF9&v4Z`lVtvM
z1-164`7IqwS%;2E{^XNc+q+>gl^T`ZqUgW!{^u@o9+#guBAPfwu13hkZ-)6VJIDf+
zk+a?5xVag2l|AV%9R`#Pa%<C2W8)h*n#U|Ui%%XA_<cDmN-bE*>tbpUJ9}1UCSkW#
zRCs5nyUJh%*Pa&5k9)n&^33%Dw$?7+msv;*wOk~}B;TaVQpsLA!%6XK6@ia2xhgJ1
zMpcwDhh)<>2qsH%SL~f26d7gZmrodaeO*%0J!ULCMjjvw;Ipy{VseZ1w|A$V>0e^>
zAx7B%94r9(f+OZ__+Qr!0swixqA79yb<FQ&=l1ts<s3}|q9nY%?M#m%2SIuphG!og
z9SxNA2|BiFLtqQ**Y)F-!`3`<EfonnQY^0ZyKGMZEVR;4lV6|bte<-!!PoMzvPJ&h
z^2fh#{`a?5e)GWJuqIVD{PV|H1b@8odu}LGFg&Kem_5xcQzFXLjV?1FgF(0DS;?+@
zH!C!#EqNzrc?ODWTWB9}ax60>^07{RuawmsSuGKt_Z8iKt=`9iees!)n%g_39(#-I
zoIm_ii*tFiDIa7-H7)0bou`U(9GBbgSFSvy372QY420$PN!_!@f`iM)p1qQ(^y=7o
zI?=(|$wWv|7VNzVGP$8&5Jfm0C`=~^V|Ye#Og9{my`jX)%gqIRl!E#X*)e#{O=AXM
z<fN96Oo2_ZjNOJ(VFJw<z=uIa>6vc^%L|A(60&PhcqVt44P1%u6+HT#|AE+*q_aKo
zX%-jZN<k2OBqgZG$qAb^s`D0J1(b?l{Yk{W)H$_^^BH^`=F^_nagu+1#lmt}lOy4D
zsHeDF`hRc%tlF8b9qjGxoeK()4^!ry32n_o7zRdcq8qou^Z^y3n~Tfr{Ccqoodjgj
zZca^dDIM;}(s(7@{Nb&EW;Ozc^XAI(GQf~T94o!;T2K^3t1U6{^B3Z?5_YAbvb7Ub
zojb#O>3ok#boSFXZ{Ez9e3_Jx?h;$BQaOrD|KOF~uQ;sO*ok0$yXqeDF}AI^o!n<R
zTNmc2wECrPO^|Uy<#JDu{%r@w)?yD@<3y<%pPbKn56@0VGktlc=I$@))+L!28tAQO
zGApQg!~KbSim+N{Pg-nMXPK(nu<XzCSP4|BfLPT&ndT|_Rj67xhQ#dI{dGp&m9A%M
zTZ^*czOWZp9nfh3zrm<uZBd~P9SpD%-KhAo(8NHPA>Fr+7&IFrF0b$ESlZXy3;O~!
z6iWz#H-&`Y6VHL5Rjp6aJ64$BYTY7twvUo5cQf4dsFO<Y>#l(azG?5IzQm|d7|9?<
zk{3sl2&G#>-yqEVdapAJeuBF1G|Tsz8u?HuF>e%d&Lp$z>RMj8-L$0XSBWk{n%evU
z5)gT&e>OfJB)uRZRW!Q#jnXJ6B&Hn2)d#<{PibjEJdY!DVn#0;5I!h&%f%%^s22z`
zvdFLvno6e>Ujv?^GC9FB`zYST+4y@egh*aHZB-x5g~0e|C#lH)#@>4XRkd|n!gviR
z22{XAQY1&k00@YPN)8ewtAGI|O9oMti-3Y;QL-RO2_`@!C?HV~m7FAJP>`JdvGDu)
zz1RJ^-rv>TRbBm>x>a%yaL(RquQkJ%V~$DA)tzA~RtE>lm2~UNyV-?jQ&cuh)!JXI
zZm$x-j(Pq<A^fUi0E^Xb0Un`c6YjOyTTY2BhMzPyNHuO#OtU@T%BX0$uP!&SSpTyt
zLK*g;G28TyIqz>Etpdum3qr17dA+e5L9Gz+BBQWnG%Qf}8s=xVNJL={FziJdnXQe@
z8!^+$lxGzIIdgBXZtrZKRHQ2+5`iNA4?cnjR~4y||0=UUwk;}Z%)=`1%My)}63S}<
zH7}Y3D9heVu<qPOv!(-yNasl*A)#7HgG;902>2$T%`d{P^zy*IeUh_{u~WWnD0T$X
z{yIS?x%C~o37$uXcj_x-&_3PsYD;Ct*5sk9dk&}*ED<Q4cRpH}Shf0iE!R*3v*nKa
zAa!BIl}K)-;&*VUXCWEt7rC-8Q|2RT3Ciu$uN57*U;V!B6@FbQb#A@pL3z=wY_oS*
zIGj1crXLB6TYVj8lzVz?>cgYc&uy&7-8P01byD!x7x7`>;$ZAm8v1;aL){>wNw{$%
zLyq($Z$m`nxFG}R210L9<l`ix$a4wv_m6oW$175MkJZTFNY4pw6l=s>AI)<oIvj~F
zDmpbH!FMAQ5`neLy*)i-I`+HkoQ`EW6&faGmoYuD++}~2g7ug}%zO`TSj?B-TT~fs
z;^NoB<?1}{jSUDlRCuXkMepO46Cc)XvAxXpTbm$=G%8Eb#d<n67>iLDnd9rC&>58M
zd$TrlmLQHuwdQ_&_(4MObanRSSPtt0p7uu_4x|k0nc2knnu^KkIShVCiV^iZs@Il2
z7@M3IljrT$5nA}UPMYT7$cYwjIm)ATnsou0A9KsU7@4KU3C*_eju6OYXMR|sd&Dh#
z1ISxa^N-wW1*m^H36;-xOypW7SMW{us>R9pW0Rn@Om*vBuw=p7<;MgoD=Sf;9}o@&
z_7?I!{SqmhaE=WTJr^|eKl2HB=huI*7Yc4m`P&8{m*g_h`{36#_D9|hqaDc0QQ%1C
zv$;g-x)6p*ZZknGW0iiLVIaf0afI2>OfNg9Io_EKX~x-9b@{0Sd-t9PW9?GYQ*FRM
z^48(*34P1F4iI5{ii|F?7TMT+<0k<^oUT7_&+#S@d<DG(c;qkMo`{UTu&7ve(%#;l
zFNNoEc=skbI&SrcFIKOOM?!~E3^Z#3R|ej^Nqd!a35aNUK1#Snbi5r;LVw);Az2df
z_{pq0<sEX4CGFPIiER~mgg{t~Hb`zXxX9*#zGBu~O`Ttn54VHx?wb$3v~-I-)QLBh
zO-)rP%MEDiwsPP5zRYQYTi?$myt|y<t#$s|kGHp-Y`=cBbPIo-zr3XD?12cAb9sl`
zWln*(Zc)QA`1Rjo5!xuFAbV#A9wsb&?(8@kCtRW^?{I~dpr^(VAWFy=x#@(`M_MnC
zL3&HzplgJ33s|Da6Pcy*ekSw-k!#*@X2?1_4FWDeKDUK}nJ4>l3AUlw1Ig;wB%gJ^
zps`0Cih@Qzp6qz*Q83<6V8(Gcq+Guw4Oqa|kx&BnX{XxVhcoWft_7r1^z9Rx&1AOl
z3;|Np@q#PMhHv)X2(1}g-XI2TqvM@gEsc%ZB5$Ue&5&%B=SxXR+2)rGA^9!ONrn0o
zSZsW$MjP7Suc3yaG&ykX=lOHpd$)=`^7WuZ5w`ayKPC0r*w;QqwLXXZs_xmGGe6yW
zoYTlE#g4noaj*Wh8^Qc({AX`=lzE6r`sH|asdZU!B&CIXOT4~c`Lc4^T=Mnj*XQ!;
z?bY{t&9B+WFe`i&MZ2%#qWi#PNAIQ9!boQ)i@}nt<HS!ySDjs5N8Vtf7Tvaex!mai
zgWq3`-&fv^z<AuEw(OOnyyG`plrmmdHvlowWNbd%t3jw|<4d&wOa`)~cdMuBX^TBW
zokRYaX+9u4up$neFRY4St1FB|I6V_C;_x%tScY-et}6vMZHa2>n*a<VdH8TEA5BhN
zd_2Y#z6In7T;O5&7gkYGT`_tYgaVsKyTO@jhvq0Kt)?5E1AofG#3d_7l;<GnqAm}8
zSgzxPs9<Bbdm<e=(hOZD?nOlz;*8W%eFi2F^$9tr_eP8k7kcA-Z+v`w2frR?>eW%T
zvk$DdACAXL7?$M7Q{@1j;89b;Sf$wxPv#pfmQZOd&k@6>IE(%oD@OU5UN#36r2m$e
zHZyJVC6T`UeiWKvghU)Tw+6@u)lDX+yv-uor3=0jK26%AANN=@-OKlL3mqIBTf{+G
zC$x(<l>za6`xJGnrr|(C)hBbSzG9tA9y}Lyy9=9`&uqCL{_UpH1`aO~8Y+gd3s(`g
zk(RE*;t-VD(=%;|%BRCNo?2~cKsV2lB0w+cb^f4edv(f1;L(%2gXyLd58kV!Qlhdh
z;3$xL(%3j|bZ7Vj$69X9=ey9THzRuNm=hM}*r@utPWca}KF4-n1?ciF%`Sn!+`lwY
zjhu$+F^c@C9RPb8ZOau!*bXGlD=l0rSzXrgkVn{k%qe+4{2gc_5F5fih!VBu$9Qx!
zn?NlJ8vwy6Jn?z#1xidm`gkLOAMptWp<zgBBQ6}VW`wx&Q4C;%_dy{RfD=BDYQqNq
z_md~A2T}>EnjxI9O<+*bFd}#6%qN&wcIC(g=#6-1(ovT8Ui<bMa&$aQ1v9V+@LiZv
z6CW36Y;27ASnQmf{a^M?3Rok&sbvl!humZ|0UOOq2+~_Ybg%Chq1GR#BMZvMr09>6
z6}7cT4j)#ZZN~>&<FwD3PlD^jmA@V!1pDkane_V>2JVz=-6`eI*ZugAzEWmqL~Xn+
ziG1roFYe6eubbATaalH$?TCCWuc_Ae;t%l;ES{JxX1<v)<8Yh4!7*{0caxrQOyk}0
z;)-US<`7HXngpe&&#pqEO-a&g8?zgA1qPiO6|O1{@A(zG<3ah6rsb%`!sEL%U#Tm?
z!_)K7+e=R;vnAg|2>V4ThT}}cxi40dwcDF6#dDskWKO6-_XwoBh#JD1aab`@67tSR
zBN>tJ{Ph+--41vTuL(TGa}=R{IN}<R_yao$wh&SulOm-f^oSWq$2>zu27HCo37&Cs
zBUC|MQfo=eWwK2@Xg?QHXiPLcbghopDgqBj=BCBf3zaEH@usMSa(2#NDEu248W}?P
z%@-#PRkH<1%Ppek&-9xmFD5gJuwN^>w+&?aR|cMY>%imw`Exl;Zvx$yXWk|eSPJeG
zP~z8XDJT_?IeG}=OnXoe^?%=;cnJ*m&WugZG?4J2y5KhPS;({pjx`(@j<&toEtIQJ
zKJzj<u85r`Vj5JWSdYXB)37T6r6+3UK|pduf)=#;eIj;+3TBV>z$Ahh`MzT-YF8H#
z<i56;!IQtLbNMsF_U+qK4RVt;ab?nt$5ehG?6pI!0#U$v_lGOJ4Pr1AKsx1>iUJ=!
z5{Axh?b-qC#Q3Z47&gwrRv-mwL%re8emUZ^85L3mpC6Jp>$j(}C9wJI<m}gbfy&<Y
z=~*4#FzTYZCc7J5HBaf~D|uo~Cp<oHNZPY=qNW?dco&xMy7s0m=@0jGPMXN;YwdV>
zxP_U66jp$6Kpqv<OHdWm(HQ_V7R6!0)I)-(l>Os3Tqj67{B148#U_}T>YX1Y?p%jL
z3c5?c#}MY>;W>EF3e6GdkHHr>Xdsl;w*?|7;L;j<j07FkJ@j&KW^%U1gMgxC2)72I
zXyJj2fZ79s3|bJ;$_E>f4GRUaXpAyE-Gg$@grYA|{qLEaK-6pxZB%>;9#th2`(<Pr
z@Duz$!2TJ`9BUQ0yJt4J_PMs2h@r!!OTEYi7Z(;#J+*=bL1pO@<R~~e08zN&z#;_(
z`R*i8{ppcrjzVr7G<Isu$IoGg=da9{2T2y>`ok{{!O*>u{ry=v1mnN88PX;7h7||a
zdM#j&khJnDO9wDZf|jJM0~_txVrCRP_zRr)zdbw#L@l2p1BFvOsus$6T3R~A^-2&}
zFxcJJQ!xcZ$#V7@2M^CR%;~^FLt1|nHwpF_fG1QvZoqSZ$^_6uG3-7%$$-zxqGomE
z$hCj?`a)3}v0Xa|invy~7rdPhch@s@hdv3-nX1&$H)YS2J!V%QlqvTpWGz!^VJP)Z
z2K|R_;#~4ZAxn;m7r&N$^qKF9BMc(q3vn|9UmNg=!2;=+CjxR;GB)l8IdP4o!|#6o
z?$h}=Wpyc9%GP$9*dVb7XUMG?s;r<e3~@po4^Uc$M14wtYZ3g#spR%Ka5yz~zB(i#
zARtgIzRi323~cM^(_SfWFhUdE$hB5Gc6AxP0afa<i3t}6_c0WwIHFu)D^fovL(z{+
zxp^{1SpJy2c%YWQrL`5{@$)eX(uRpWi(U_>5uJ$RJ*ZbfcodeQXSdhh_3jWoKwyE+
zwp@7kR=0(Q#zvTti$H!ZjXe2403h@=lyPOyA^rxBlw?=oGkQ5-#xROGsEjCW-i-hZ
zHXe6~!!F$ZhKmmGM_7hW<~27SG0jX_@4j{W_VLuu%bn=eL$V0}g|<7@v(~6zVu?&0
zzo9(<QHZ9zpi)f1pW@D4dv=EdyekmGy#+$F+OSQG_MS$9){(R?3wK^Ng$T=3$kAt9
zJC*-#(D3Ds({ESLF3n8SJXT82sC&D{4s@3vjCb0Na}QqM>3FiOD3>NG{`kCKyW8_M
zDGNed*GO&!)-M#<KJ!iXfsGpN-nLI)u><X9eqK=2vPBaR6rW$CTb*yCGs;)Ppb0(h
z-3|Rw>dl*Le+`!3-+c(Rv#0WkY6Gn~)2LaO2df$GX?#rzjS%wc;W5KdI^gSdu1Bo8
z439`|UAKh+m{J+x_P2+kPaC8zkg>;ldg6_y@9$fZeG7HhZ=)A6cOhP0{(1gWeNcz%
z(I1YNe^usOM75fXa0URck+54(2rb9(<TxEeFIjXkp9qvqH!LM$k`P{P)e`+H5>H>N
zwYbL^{T}G)A$|v;yo;Nnh^Q#$u?pzEt;FOh`8NReQJKsH4fNfO0tT~W!QQLtgP8P;
z(>d<CjLP?24-djGOt!gRwin}qagdVP&@u;lkN+#`9Uq}8!OJ^5mcLwO&kzqd1ep`a
zEx{Jh>AywXwmap>yi<icWg*F@!6En)vu$wqbPos%6uHs&MZA9cO0p%GP&akwB3xZG
zs|<FZc)PmM>S{^!$emHv_8Rq$--X{kJaO*|=)B)3uc=fQUTAZ?X|5)Ta#q><ad^o4
z-pa>zDb23VKbzx(WM}I_nWsyxqC#nlo;8?L)Yez1bj-&=K;i?H4jeaxMFj0ZXB-L`
zxKq-*i`MDl!zt(oL{}ByAqLQi98`BHLO;L>j8m8+GZaw)_5gyBQ&c>1{d*Z8zU9T`
zKeg2f5K~|(cW|i|L0A;X`O?x7-1xzc2U<;v;lc+wIXSURsG>oP1*$Ai8RSGq53DFq
z+$$%9?z`2lkxbOh97RxohTWapwppQ3q?jf+GIDfqkaV@+Fu_{BM$R(6<{K4a$cf3x
z2R`#-o@D;F07T@&WlJ2e9|*G;gQd(1-Xf+00k<9)eeBbxgwcns2u2?!e!at2b~fxI
zN!sAR*oR-_kt#3>&|tY?qH=Lj(Vq%Fsn?=s2$zepG5QN+5gVX?S$!uPEscOV+xm}>
z4jOG5J%`H)<<%zmAfgvopnlXU2~`e#2k?r^Mq^d0pC(t8SW#~fEzR4&`Q{!~md1;X
z+?>&z?(Q;YRnC5h*P(l^G1TO3YhpE@h{09fy20M+v?^z}?NncLHu=Xos=Kx6>rVDg
zOW%wbJA5ciA)qWmc3jL&u`1jdVwSR_qGP#zEwFA!!88WKhfA0lv2bgK6~=<HyU86D
z6`c=|=NFkDer4!9@&@LH|6KZc@WtWLuHn4lpJ{k)@W&>{+9d>^@|7M`%cFm4JMgs!
zj%#`NRkeX*PcRsapc}upd~ATF!n1QO+9nOPgjYaE%iOe-V<6sWg`d>$aYWouZPE46
zq`|^~rCNomZ~~}FID<&mdb+!>H@!~6(jzi%x*X-IB;0Wqml-lhuhZoBKERVghABY3
zrcBCF(DPw;j1w!hpyzm?SX|Askth5SP6ISi#}7K>fAD-Q8mCdHdF12~>!W?V`Z>Q1
z4Ht(VShO|d**L#5Qjj}u(Y?srF42)eqtrAXXTaz@(tN3X$E2?DWs|hF_j40;QOf#a
zwsis89V(_F*B;GxwPZ_FjeL()^D~?jPHdHQKHC0Dww3+C;NhmT=|+onlM;PP3ysxL
zuCEuKPTzE}D5;7|y{Q<wX9IsBX#;nk<OPYxzL-u+T|s@|AncUH;()S;AGSd|V&mVR
zqy&?z6Q&AQQ|nomE%QEbY%~dNwW$^#@J?`gH<X&=&D+-kCs>*+%bL_QN=A~<YAb`w
z)rk0lTKGN!z2#qiZuiYB>Qh@$PFcO>8CBwO&zR)6ITzCaPX!e-KPN8ri0KP|vr=PV
zhp+Q(vpb9QuaEP5(f;%><LzSlNZ-60|5pcy+vcNRy4dVu9=G}yz;e8T8~QU!m@2>u
z)e@`y2Adx1b{*u8{eEHrspq-V2D9mj!~<&-qIUbAvxik)$S?^?PKpi2Yrbjq44_PG
z`rZ|7od_A82SRPTT3;gYKppO|{x5YYFm`~07~H5p)|6Mm<Q&HnDVL0U!@cahy8F;2
z1s#@E$<$?1O&ZNl4X8BwTsa+YB`b<=;eLR1$t*1|0h<XM=-Ru^By(_=X>v0$X}g~#
z@RrP($rhp+1UjYljf&lu!>vGXN#J3Mo_i=3r}eL{Z5-G370n#{FMBKC*qB^?ez;Mr
z)_O`nO%-Ml2W72eS|;~Xr_oQC?A_hhCw=od)a)@JBkYsItwE{rhX%D1iJ>~RcFb~2
z(d9O78i@+X$NhgW@v0~G-}nhrGYW6C=T6f5<UVs3tS#EBf8CH?rLGp>c{n*zI`@0>
z+j*673XMo7h7+o|_@_5KRp=TXP^dHeI`v#;{!1hHkNoLYhRLkuzH*+lP0Lz@t-!cN
z_E*xXhiA4(1^X*+rjQYxT#*G_@9TudBnpZ((&7+sqM&ep%0x;2{cv_gO7PZ?L?W+8
zal-QZN=MsX;X$I2M}c?$zu?c6J_}kn_C?r}W#rnlmLl;Atd>u69+Cyn<km23qIfuq
z4q~QE*CDsJ{5jXO3qG8<L^ZZP)GqL4eHSt#(4LY7Ew)e?`(lb*THYLY>sbnl;4_j2
zl9Ec2@1EWQXskh&>pxuO@9RB%ayf<oB|s^N(fBY01uGf2jjK47iAE7@k(vQ-=m4C9
z10wBDR@HS1=_c69_RVIQRO>fT+!6!_pYQ9}#IyTF=qXOH;V_a9{!jT>|1)pt|LESo
zE7)UOEcglJ&9Fd?GwU)M5%l6rkRgG6DeJz9@NwZ>zOSLAIFVhu-m`GP7t%V*eISUs
zjb`_$9?{X%tRopH+^hQL^%N&W!qBhyP+JMgQMA5x-_J^OeOg6fObhdHtJ1pDivuCn
zVU4E|R-bk!?V+N$6{G^YQp$@NK_Zji)zk#lafmQbJUk!<O22gEukHsX`!o@_28wY(
zIFu)ni3u_eH}VyEhKgekGsheGnpG4Jw-ZaW8)7i?_WASrB9uw6vs?+`0Rawpr+ws|
z6x5fvQWC<g<se6=wjH98TyZdN552aC|Na6vH~8`zd(E{Jr3x!=lq{B#f5fcmH<D20
z3s5VL^M+wO@%J7~Z4b*tLS_kAMRgb6Bq^4>Nz4?2o7L1}>DQYdL*MdY2C<KVT8-tV
zxJGf<6E!CCfidKFv#W%=?mvI7%5fdS<W30wP8UIOWHyQ1R%)wV^GNA_Ki&~dc8HJ~
z3$84IW>aTVr1Pj_={`?&6<kbE6uFo)^tcSHSHPGAIxD<4uA{g$u11#g3JOiaqgh=6
zW6L@1i>aONt0<xuRxU_?TuqbuSbsDOiTV}tMuFswo)LmAnwtU!29#j+!<1qB_kM#8
zSu`?kHoJ=Q@?(p?zpcIpgELN^SK^w0?4KivPkl`RaZr4KXbKZqv)}?Rj>raNQ$?<r
z0JBGdyuT-Te>N4A>jAB>_p2Zx{-z|#myU<Rd5GLw`MzI4wZ~oO4qLQ|rJw&VuXZ#T
z`I#ZbwyJ*jiksqu30XK55n~gR%}ip#qiV?MC1$$=SNr0UUJ}=oc-T8<{b-Py(p4~s
z3&<rMUAd%BF`i6K2u2(10-=<|?Z2nPQn>Y?%1hvald(Sha`v<oC*~ls3Y_8KJ2Z);
zAc#a-FQX-chb`P6eTPJ)H>4%YOw%bCMgPeKcqU2_fRb|vnABXv#D^tDNB@AHl+w!E
z0Y*69y6++FAHC@RyIkadWKaGl!h`i5Cu6Ys9)amtSzsVYv-5s^qKCT8SlH#rMr2Qw
z_g#R@t)eFKuO+RP;MSS=tf7dOOAZ@_5#tVANyoI(@aOqUlWBgUqgg7PK~Yi5X>N-{
zP1n*0-u{gdYKXzNhHyH|<DplBT~Fi`vil;RjGm*w&L+o1H5xxbkOZYb%Xh0+;Vh7|
z8Hb$n$S8z1@o&87RvhogWOJ9}!g{}}DJ$aHQy3p6d-@b$h;v=}&@ODp*4xOR^#d~l
zIa8NqIU;C{vGdiBjS?gI9O)GL`_Og8hNjv@C?I1TAyab#f>K%EI;4sy3xSYcL&@x7
ziL4N%Voe~p1ZO}M31B7Co^YHTyskQjV=E&;K3i2c+q#EFS<DJ~LecSB?9%N*V~`h&
zTKb&;V5|-hdZ`+068;-mMagggG95sC#eNp7wr{cWOvVjqkXYe_w##I#mQ^z#(6I9<
zkgL_hIRF?+22@@@n0gO=XNGtMmvwBsbbkdC^woEtf!T=Wh3wt7OjkINVl&L}JOYu3
zoY8kFmp-djp1p#OjVBg)j!jn+f{&fT0tS!;)W*1rBj)p1qH`_m_N}({=npObsRP*Y
zCj7+|bP)xwZ-kK|$DH$xzJ2TD^H){g^WP6ui_v5*7SSl*&uP=+q<xG~9mOOJ2+%|1
z4O9FRN1w{G@bny5)XN?_4LEB(&}DDTPFBgd_Jyu}wfOY=pC30$%>1~6=Ur^~2zo5Y
zENGhzfH&)!XS0ny!kJz||15McTTy?*nz=%ACVzQx%x!tTy=1E8T3T+E*hHmu-dYNZ
zuT@-{c=iT2XM0$kF&)Q&fTU3bAdsYFq-|b;Bnb&DGe!u}aOCwj2uAA1Ee@un-9%U)
zcpsLmvU?8KM*9dp@~kxD8vvhK)qt{u%;y1Ud(;0!oy^y)1^MjF$rQt2t`k^653+)d
zbhNZkTOw-#$kWiO82kP^QoV*xK*0shrAKEe<{>5P_WX+?diy~`I*RDwj~$jbZ!Tev
zZMPpn0pPk3+NzAknpI+fjfC3K;GMl!&Imu)rf>ol`HOZ|(@L8?PQp`P1<syqkU<Dl
zdy<7s61T4t+e>jG{uwdR252k-fGxVbFwu3q8Ic2Ro+ir7YQj2t=os^DHoy)pq6b(j
zVD9WyPyH8#IuEhaUYi(%({ZlATt`Jz6wMXL<~8-Iokz1Tr{zw0n?c7%imD63f|^v)
zKGzWFS~WvLB*>w#^BAoVK7Eq7a+Q~s!l2YjR73a{7hSJsqu6pB5poV}h*2{d+q`o+
zc@KdKc-(v*?lkPb`s-sj0VNPK+BKboW5wV+oRKj?6i}~s0m3s4YTLjO{;+eJrVGPH
z?l-sL@UGE?t4KteXcYw`a+d1jUlfVl#DK_+ph2J=4J0iI=9r<>@Ch=opN@jN2K~Xe
zT)w<?ZP-v*Be)Gh_O;SQ4d*$*?DuS?h;~w4+2`1Jvw8EQKJy5n81Wd>SRP;@bgox}
zk}R=2TP>S+5w<J!-C#1k#rq6(tB}t?t{JC&6S0J3bD~hfJpUcGlQDDaZYqx*=*vL8
ziPm#Hg-RH)W<kYx>^{15qtBk0bAJ_tqUYL;JBp^>cL&JhIRwvmOyO_y7e_3}-e<IL
zse$2md0Lkn!$7-u(%x^MPI~HlY!I`v#HU+qNQ(p>EE=qaA(Dz2C5y0GU$AOfPiVEW
z)=(toeN<nE<EHn{O^qVj8{G<OEl+=9keGLKca`WE?`T6SN)ap(^OHkbdVPdyz3X7?
zAL1m!iA&g<C_uQtY=B>82aaWpIN;Ew?)}Pb0EEsOS7P+!`*5Qj5MU-bP{r~91x1<g
zaC)Jt^_qm(q!NuS*kdX+<tl#HV1Y=%$NUH|{e~r6y0VCxd%>VpjLH=-7<w3vqQC>D
zZO^O<OzBy_6y8d$wG^CB;f`=e2S(KZQTXN73<9Vo56Ih(=SXVa+#pi3;LCThEaSkk
zlr_XCC^Y_FIhe3!`lw<(m{%X>wqCn-EdstdIxW2n$TN6Tzj%)w;aJX<07&E+wCG+p
zrvjfu43mxNXFMzi5F9BHlRAvda`KB5b+KzQz{wH@8wLMQmcn8eKHf!Tn#k&aw<9~Q
zf$GLj?EWZ()@AK}N7%@lcRdhda_Wz%7MoBAI8p@H(PGzZ2Bej5)E5ry`}&fcF!8Wn
zu#?dYA@lAj#jWJ)fY(5}yl2KNZ21FS-rHB^DMIIje^h_trRqpV-2k5}fHGlYv3u?M
zV?0S>ivQpon}CF3bzy>*Q4{5ZAO*U+(1kJ!*S1|RPE)VD4C8icu&tkDDI*fZ)ILYx
zD$D>)0K5|L6Q*NgVnPVvEGkiCh5(-I0%UmGfS}{(1as`7h_2isI-E{@d}eR!Srx&_
zta8Z}q(VkPJoV)P*fxCqrvfehH%4R-=l)-f_Z<1gNYP{aO;IRb6p`BvR8;bT=QT{B
zM{w!he&6-SnJD%KA|6+IbYNfW*@tq2R>5tCOXHz8JdsRPJ@q@jCzGijyfU8DES4Du
z<aD5{VH0qF`u_>6{7=NE{~K0=xW0cMJmZeXS*n{!gJ*~n4IeZCOnYRh9{ro&7-?19
zZK^~#=BnSKT?*9%WqXya3}I4EvP{u}0b(jKrpQiil^dg1H`MT^;|+OGqHe+=0KMe)
zKzyx;#sL%?(fxZ4a1}J0$w(y}W!I_31i<VBSck*@CVGDDEbIc78ouziMcynsIAds`
zBL2VaJ#kIW4HrCNDH36)Z?Cl}+$<!9BoQJU9unq_FME<xQk{CkF8{T0Zy~Z#VliG_
z`V4syZ8TA$e1~A54ITy2^gN;#v0p_rFq09ZyJBAA7S;t&maWz>d{q7eak+dwM;4@k
zORtg7iZ%3QB{tQ++60A!fh`Gp9J~Fcy%#pO;l*OxJlLturN+^h75u1lZZL?4smsuv
zjpk8DAyCeKW)o73FKWu-!43_e+Z_M>Maqk&1)*%abp;<nVtMweAnJ3HM7s>Z<tz9>
zFZ&<+JQ>ET-IiX#K&#)W4m$*dL3H#M0(Qmp4&?m$am(-wDtSUWc2>)kQZZi;s{Q3_
z3|BABXsh0-%nd-|<3<5J^j#HX_@5N?JRBgVt`Xq*JwS|=?0WXLL)$zV?at}2#J#T6
z@`w|IfA!xOjABo~sHKb_I-C4<A7XzO4||AU^ybwPs7V|7TO|fStI`GWivEq7(_kYh
zeldfWB%T*h@8iHgnxD7mx?x)3#B&L(f48usNE})zP~C|};KnB5098w*Rq(Y{f@_Wf
zP-Ni~YO$2s{V7S1IF2&5PzI}_Z3JES66a6~&+3bGhDN&2hoT+H3({@nOcF^fj?wC@
z)B%*q)RV%uj4?JYsnjb8J2L?X+GLk0{1Q4c&AdBVXw0x5@RVJ^W(ge5KmWnq$WPo3
zaoO$7=OoSF4LsW@@csb6C-+bJh(55ha$cG_|F}&Fjg9^9OdveM780ABj4TB~;?LWZ
zO!r<7zo|Ds#1i&r6M=>U?&=h<as#J<c<fQ0;dIn~APC+9nExu(Zf>@+)n$QV>R)C8
zC<TFdoKS`8(8W*#CX5+wJkt#53U1qQG`dueEMU%9o^G-tB#<k7_}_7+uj2WLWWo~H
zuGKPw7^X#H>6hR5{jKvt1RERQbj~pXkpT7spNk&0c(ZxV2x7=A#M|&uKj6ACI}}4j
zr%;|4#&*v^2@g?IZu(_O$qyj5y1$p(!hNJ{bv%n>;(A<O2)R6PJamjd*5F<punr-P
z1&H~lUR59GvDY(>#jUmCsb-Ayh1*mi#zGhA9O@rSv(;|C`ulp~Lk%7an#fVC*1QY0
z*EkUW`)2JXq55b1m`LTa7qu~<Hi)uGC4tD)NG$wReXxwJ7_v<N5$Yy(s2lZ%AaT_!
z98$zZXOvohN95oFgext(_c1{3;*l2QmEb241t$LxB4>i^Z9R)hVmB)_tMd3(#qkqw
z9*(XZmhqk!iEG8v-qA=6jef;<g&Ev7dwwG}Ej4WC%WMg72<!|o_9a+Vr}-7x-XC<@
z0jQqmh&{z&=bWO!kV0dBbA*)6*E6NIh_K$m!$2vCetr-~rU6ANpJro(7$`>>ApZ8c
z15uP$rPuwz6)MWp>tqYtxOul=T&X6Ws_Q_TDR<njn_DomqsFFZ^!-GY@mp@t=uq1=
zPN7hBR}dW!(R@}cU?a1!OsbaMkyD~fYP_X7X3)x&fv%rCF{Fyqh_MOf9`*(BADL3U
zo4whyutTWL!3cdnuqoz<To6GaE?Xwmxp<x8LnPr?!lT@UDS8k$Bnc{FpN~%*qi|TN
zkIxSy7neJtBVs>;MP*lbWI33WfSqL(s&AJBqX_8CSGe=uwbblhTm|2Vu6$X-jxC-r
z%r^((LzjAm@*83MkVkyOUH}hm48a!YsF$uvar+@}fc?&u=|ZSoE;5Wb6N#=ym87@B
zbgR7uP96{j(y@>54Zgs%CbrnXd88JiH-JuzKt=0ekl^6x1A_&7q!8DF0R^o{NT<iz
zm+=^US~ldJB0(CS<X2n|&c-W5>qy+;08Bv52cv8X_$ld@?zw+p1UyWduywo8qqCTq
zzxYC?i+Opb1SfaLRY+AC{Pue2y(^u?1qrrA?xdza5rfE^YPwcT7YU;ct1yq?mPkDn
z{X#Z6h36)HQw$coVu2EoY5@zMvc%$zthm~zDFM<<j;!OJQ%~9#dcAp4WiAo?LID(k
zam8N98bN3B>S>R2<{H8_>*EhQJ`_<;EDsPfzS+AqK<t#im<&_mv9DTGakPTVb6G4L
z`mix#&`0XLNUR04JOiXKS#liUr?;pQYz?YBgKMNBtjoDP#HodU+$yq|0M>ACQx1Gq
zf5ILp(iIu1mf7?BP5vyT<}mo2i4~$@R{%HB*~>XeOjL|v3n`bPj;`cl^9#NS^B3L}
zKX~|2nXmIjX4~9ni<^Oa<52QhV^{nE@sCb-Z8KUMIyKoMJGXe8gzvE&UTmtc^PZ1d
zUbrl%b{?>TIDnxwU4YCv>{39hq0M^=A6LRiBvwg2+qoXewH>@S+m<v`D23WxY@kEn
zl_T^<)xp~tb~|G%1&M~0XcQ))XG9?FY@%XJXF*Sm;FPT)Ol;65($X)j9FI7?109ce
z&O7G4%fdSiFspAkKL-{np7nco{I9tz!^J@@|4^QVWB%_CRSEBZIRcY+Q8ccnRM%C|
z(^4dWGE9oIp>m(z?f}oqYKD&io(`<kWr<E@cZRx>nd(dYolxTC2)8_GeBtpqAaZj+
zUZN^haEM_}uUqw9MYb@O)#dxD>R8J4=~SgOL`?>kAJ%$R_aZ+xN13eVvTY}K-xe}K
z0FexacJEel_`45AlHgWaFLZ*1AE|gBe}i=-!&1DoWWvOq?Xb;}$LyT{js(lFD_`{`
z>j;n@&A9l|+Oe{V^0_Ft_bcTQcjo|D-(#5bf>7*pltg|}6+k)C&5cL*c!tK$7Dtpg
zC1hw`^TRlxQ0B>lE6=JXq!ySF(Sf)?2y_(|rM<+<NacS$GV;~N#U_1}e*u*w7=33v
zGc%l>j^MwD(ZReDl#gdW3$?ShO<_ZLn~8Nll<KteZ6*&ztN#{V2fUfU*RKF~#tqkP
zr}99gAciM})cKSS#$e~+V8Nx}28jxnjL~qrgT^G@&IPEdhu@RKSFK@yLQE(6$CSV;
zmAP;}N!8<!r^+CCRxOuPG{zs!hDphc?{(xvlossn#E>Ke6a-o;5N&9J(P{*R2Pt4a
z6k1VL>cj~o_3_~eR5f6AMcKC4!oj!#T?A6GQ_u1Pe%%e-HE5|NKp2+ms&OKGOS{w1
z3%Z$`+y$UprvXCzSc1@VHvkvWvG=cl^x}4|gI5CZ6#eh8qOXW@e&v;<rVeRDE&w+O
zqaGcDXx>3SLC7fhZQQ-{1y4|JNcPV5No`4aq>Ll#Jp!AJ-$votHUr;8tqJzI2~h!%
z)o034Od?81cAfrkLb(nmTRVZx<?}koA^EK^I9gmUZxmy}nIrgvG~v?6ZcgXmP_gBg
z0ZAj*O>dsHZ&s1OKw#=gSVaMy*2mpvjC5Ge0LYZ<z3WzlB>-pF|I+{mUQ9}eB#<<Y
zEH*bvAUObt0bm=Fk_Ey!SImvrsU0QS1WMxeW)hcu8fk&I-YzXOdy6X*3TpTSx~pRF
zsdy~Epc+NIG%nc?ehC(!75f6d)EHeg8E$IDWJTd08E37){Y6BfB(mCfao!c-PKC%N
zlWnd~b3u&<TAHc;gc-56f5m1J8R6N+3udV&Cgb?Vm))t{cdK$=(1$Xq-pgC|l}T~Z
z1`snk{?+~}Q@l!cZ_3{%V_@e$<+=ztnRSu+Pgsm;eX2N;{swcT7MEDN&}Kh{E-s__
zIO96m>)kz30|y{d_%Q~w`ZjEY&UfTE<B|ar`MPp@h`OMC1^pK?>8z#*WKnn8wVT7d
zfNSCV(nQqqsoo>QV7wr}v1&<V0)k$8x{KVJ|K+EzG7)U1XJv-OWnE*<#dvcz^Ibw-
z@8<@I#nX=?%jMUWuC9g=RDBX9c1m9f9w>UgMVmu58Q=`^B=1e<dYv0^{L5tB<a#uW
z%VoVF{WybAvzU?F{B{>SjBj&xmaQYiX1L7@f_#>CB2Xy0*Wa^j#0Y1J#7jmD9x<OG
zR~AIM{rTrjBCWS?@kHt}%0hn>dhd4QS{`!sF17tWkt8}6yl?qMlJfK!2Y@-=VoyT$
zTiBK|eC$zTGT!2K$kWnRj!96$X1zSq$(V=T`yvO!#U7>UOlBST{Z0nFAdj*6`GK<J
zxNVh4->DsxVaIGu1rsLC;X$#%KfPr}eCNOH1^pDi#pOq3EG489g3+e(V6p8rNcTX@
z78Wbg;kEVXPc$Qw2J{Ufr&<wC8aQ-#$OKFir{>F{o~LXENlA~1Js?0#0vylD8XA<o
zziXTJodtl_Xts0oIvd=G1cs?O3t()hV*v<BsilJwFgp|2kmU!6rWnuw=4LWs0uH<z
zzg$&OF^}M5c`7r%SU;Dfl-#wZ<-}%y?v-iQJcY2ATt@~YAxaoiHG_eJ@cV5;B%mip
z#H-*UW+U__U}2rE#REkKQnLtZA7jxVWoX#M2FD%b)^i>?!MSw$s_8i!FDaE%K3{Lu
zM~RGv>z88}iQ>}>x6JoRENkm|!kwO~_65h=2}Y90VMnTLv_-NfBadx}Kvj3i{}2%h
z>i$KFPJ;K-;ipb9!G0WDdoW%nxRT){mm(e<U4+P8L1D+zh01-bjxz#fQoKlby<kf1
zVFFRb$jQ3AUtSWkMu?sC@@xVIE0Fv9YiA@QB4XATj_7>=oGr(E-RV?70^Dzi+iTj<
zpPd-C^#kD_zl&a{yJXMocQDL<fmyfXc-{~9joZ@>x{SQ3-%-^H<5&V{PtJ+P1aV+H
z-PI%nC4dRa#m*!3PXnghPYQ6z<AN_FZlGZ=LX-(jN;pGi5$o>MoKYPFrQ%Z!T_~!#
z_5^ErKR-@G<&lTrvwLSWV$4cRRwHkWID8m!BDMH$6^N@{%%KZwJ%B0gILbQ)z9CFR
z>S2kn8ULROU>_+uIdu6*Su!^Ed}>?EYXia*R5V|gzJJvDtc*77vie15<8Y+t+dwBZ
ze7hNL`WPN!|L&k|9hMFlBmndF;I*D@t;Pn^t?}EZxdR&Ve{*v={e88w`7c_Ixof(=
zDTDSUoAz#%yE?wtCb9ot8LdJ$S*A{44<?BSI!ImK{qobk?4X(DYoHh-{ox&L1oBQ`
zfuFyS0??%Wcybog;B;rs<LQ$zSt8Q%9b1(5TEBHcROlrT!e5<m6xe5}Y(0V#9xEBV
zDhfxxk$I;G>v9LR+daK?3jlT5m=aE9M4ct&mqXuGVU5+h>!S&X5Syke`xUjti}V&*
z9e?+4JV%9$nP>FH?*Kiox5(NvzUg)@jV>Z6u8}|tnv&rf37ga~7Ffb)ulIZ5*~zi@
z6G{&^e!w;-i%f`Zm=TUcoQV?wzqs-@LfHbZao}0hQV}FDWL~y%qph)D^4>Aq88N<!
z=cHp*6!sv(4abc79}0lv=0q1D6QtLhEuYbOlP$I1$3(CQtFhm&LbJ}RjT@cA*yidQ
zpZ~GVV;)uRQ#V@)?M?=GE(eIwN{^qYjb@S_q!Ut%CS(8>KY1}#oOXXNQ4E;l8yCLi
zOc7s>N{h<O55QNLGS&xEsh7;N>?4EawGJ=Mo$chVPS_#qi#q5mLi}e%i7eSh?vqv(
z0&_#xVzLei%f*9bD%5&I&0ovw`R1TF_<9kiq^5gWjW$*zvZ?n)QS&T~ZQ7QR?cDF?
z_E{4~0!cXwBjoaSEt&}E{2n_KgH-O5_;Q#vz=$zIRD?P-4iYI#SZ?Uhd3G(J>7QJH
zH>Qx(n~MunLYNmCiLRsu@vT%I<ePqGpZQi!bEhSwOkHoK^Q_}s3cR=4SJO5jfmuQ%
zi;-wpd|MPW;LEMAN3qF=&^A7RgUCYOvtI&K+aG{vxJH{#?k;MDy5@`hHm8=^X{SB-
zGWr>75lPb)1*vKr`9~Uamm3L<iRnMWv#qBQp+%eWOexR?X=8WMh|w?l@lKum&4Tw%
zD`Jcd?I6Rtw%3^41v7&b2|P0Q5e~Dm_~dt(V2JILp>UoI)}`c$C-0X|wwu8dnf)p+
z0FNpS^-lv>0X%#k(MudvRTUFSi*N=W)Z-HqPxn)M78Z+hs>j$oyO>}LX}+9=Lag0I
z@WoNOa%15`2^5NfH^zC66^rk$hbO{r>STY(qxrlc+vc)+P~w+u(hS?0vnu#=ijJO1
zxRH_hnablNDs2xUbOKHs*l1HEPRC5h|G-5WNg?Ys`l+Y!bI45^b@&x4Qt+DV=vj^U
zTd5vDf11<$Cvcct*e3!GE&M`3ArcS?i1tRIf1~A7uKUwvz<>1M_WY&~J8<Eo*YY_z
zw$Y;S?bNawQZR|w-*#pve{1?v3)!gG!fJf&mh9b{JkwYg7WOZh<6Ty2^{c6V`zO71
zZ|&9eFXGhnD!J9KalQQgI?10eU%f*7R@?uClE48-5wm`OmG?tp4{fgBNU0Hq@{X!&
zZ0l+7z-@>lEkFZ^6Hrx!h+~r|niDlo_Z+jL$ngi7k;;6>MA~-8|Hx;dk@dc>{We7`
zfHuc7aN)}KimIx-$vWMqx&%Q0QpzUr5mD?#dK#Asr1z(<8=<@Rx07lyX;!*!<YmaG
z`m)ZsB{OWtPW>BaTc-Mbc!3SmO#5CcUcAF?-Y31}`;P@Az@AL*Ghm`_3th}c&WL0@
zYy5Wd>CnzvAv7k~q&Lgra<P8TF>C6QWWG7WQ;pA9&KULwWjznv@!#L1Dl+3IG~$`0
z&+z;bJLg50d!JAi;Eog{juD~c(}HgS{JqDXg}^MvuNbu)Ij%Yc#EBphqz=!KFN<4p
zzl{CIW>eAn)mQswM*Uy{x0eL?WcxQZ(mH{rb{xob(56ToD}fq;Uo8%DtV=s~#vu**
z^x53U#pa&#{QE4SZ5usK!XSB@fjWNmw^&*b^&@Mr-0w9r*(>R_kV?ArQU>j;4cR#l
z9K0)4xxND^k?}0uK9)gcQGD){M^$+lLrvy)-oopj0o$TuApNHms$!jO(T8%AEqk|?
zKf<ngxT#EP&mqvxCCT{|RPo#YG*~#8-DytBanvZ6ps^OZOu)9C;a5pT5AjbsYIb<V
z=lE*KFcJl4j;tvA*&g^t`lGzoR2lGknV^lKQi`?|5?K=QxGL@veKXe7aTim6FLjdk
zWF6-I&3k3IH3+F>;7rQtaU=-7{mg|OKabSw1$Mu+s@CBUPGPVYAcm#$Qh0oKQ<&Uz
zkBGTPY*v1BzlbzIiSboxsabClJNfx{WM=#n459VZc%Nk`)7d8UVsHRhQui<9IEGF1
z_p5-DTR%_+1{8ZaKSXmaqhpE-yskaGu8is>cIvPw5uBE@tkhogY}D#@gI30`w;#&x
zzau@&XTcI$6zVW~I4ot9M+w%E%n15?q#nn>7N-{o>oJ@Sswgl!LiMa=TT-5?T?9nz
zFQ0>W^A{=>hBlv%pj8v7mJw(iv3<0x5=jYSQB5RlnK=(xajR;{_|wBpg(y`NN1W$<
z0YisIJE0Y%S(g7Qtv!1wK^EnocSVY|16t!5HMVV8PuN7Nt|sqDX_oW-R~>Y`wMcw|
zJC$B`0F+WZ1v;8?r|<6RzvA1MrZVY{oi86j4szOjR*Q4=vryMj^d|6rXr)u*AElDZ
zpQ}wnW#j=d#fXz~fBqtT8W2ZR&5olCTH4j+E6V>?3cU)otg!9^0I68H4h0g31b&n=
zz_aK<03|ml+k#E#2HtM}2*6{v)$qan)-3TwY|cA*&#BDybDsjW5S_=w?6fbA+I#R}
zdvjY!1Zj6~(MI&e1R&CVn0r=VJyLlDLa_WUDk>^lT{FjjH^G=1(Ho5|Xj~%Lt7hHu
zf{@a<o*`_csO=za^Kp=RvKFHpd&3$&friN*R?qzeyqqI;?}|vT27=S=nXRZSpm^&7
zKC*_wmF1^6%VjRn2&T`sOtFkvq2VVZB^hz~d`5c^0Q^oR9(t6ZJ+_?NrUVDn@Cg_a
zTjMT!9mQg=#Via=*rlzNQ=S8xx!(^U>XQXq15#b+PEJ8oCeNSE`c|2&)&ZOKYoCaf
zG6V#L{WBS*EtBdz1E3<piD9=Hdc!v4aiADNjF^X?J&=e7(G#arnGGOGgs|PJi|iYi
zK(GB?$J^;EX@h}S2#gCm{Tlx69iBiKiJf%11Uz6Z;jK=n1L|4IA_%nCEg<b&%@mTr
z1`$Q%+ozL1wZq91ZqQ3J`s_U{O#oOJ=0(w0{5(tB)}LmWi$EkSe8@hSk6u6L?x9a$
ze2^X)f%xw<aJbO2RuP`mXy=+nAS)FN-N2vTVzmrW?%A+?=yIeZ`)Jt|Jeq*@o3ks)
zb>0x*BJx`@T}X?VoEM4Tb}3jzsosAphN{66of!AcT)?RE){~;$;)zz0oz1Ue-Kc%;
zJ-@ZX3K=Hd>?K_feSl~F$=^skf?5Y_cE{<R`n1{QMlkV&i*r5dHM3)iP48AJ)|EYJ
z^v^v>Vmn078nYw&o{~6VdD?ARs|9^K^X>af@&RoXx0P@^)!xCa#bj9#6l%M2*6H?$
z+#;J^u5Gz*ty<Pvsj*ngM<ef8L!)beJ=<L^656NMjOqcQ)6F=bK?B;U#sLC4H5ZlF
zO1IY}Ef7)qh1StneX!j!=n-bT1xkQS4n&Y10v}_+NIeE8(^uSPX~=N-nDxQaYj-2^
z7>-B)nV}UNLqIgwK&B&AP;6Pyweo=m5p)sx1SkjM2%}YJ;@Nhi+Ve=*>LN@Uz^5Zf
zV+h_1iv%-*TI2H4bUsvtNKTm*%2br=To0>C^CU_f<ve&U0)Zo8BegBeyqH67cy0J?
zhL`o}O^>W887xl>4;Lc?B6&Ad3EENNGcahwGY|x@jSTsfOeRvPQ*J3@y^o9n!hBXT
z*8?aL?HRDdVcWTLgp?rTda!-?TIo|O0e2sU|1b3e1L`wL^jZe-`F;%R_XwFYOsAaX
zT<!ZFVk+y`29To?`mc`4o$lMoytA@1aW$x%g{CSWm=h~NNyvN1u3q2%^gBYG`f#hJ
zUWRbyygU(-0h0?uZ_@Ql7z)vVD;PI2j~nzJ34HRT8GrckIPwhxRDG{UktbgoQNNR4
zZ&x$l+)^V3rYQeh7gz%kO^!&1i?A)lwxtynLtxl+(;>Lji<3&WbGR|*yf|53!q|b{
zAI_I|2TAdEZ>052Q!o#`TRf5hIg<3!-Cj#Y<p2&Biq+5cvIXBQVyS)B*mQX8WZ$(<
zmL!_}Zs6tS6}^ZpyqAS^smm@8(Bwr?X`azLN=r-B83agZST+B9fTr07?8OZ6e4s-O
zktlw1eH{^So0SjS{~qv%k)hD&J0(J)>6qX&2&?~qibMtWTBk4Rw+g;c0L5C`oyMk=
z3bV&Ic^0l1DMRxll9ZY_WN%j+L~m|)QCp=I(ngbeF=F1jtDFE#eps_rd|MKtRODKh
zt@0~D_~L(=NeoTr--;Ffna7N`bXaqsPwix}$(D}6pp~dIL@em<?tPBI#~fz>{RXy(
zkV5jq+h;F<H6iupMM(q_!Wb1!9-bjko#znsLOcfBTo}FV2&{RXepMl`%+3n+F=%~;
zJ0d9}HuY=dhkpu3u7CtRmd_zE-!rg7D5a{p+OMS`1?6KPuEb!X8uBqh>-opD3kY)n
zf{{)Eg2W<SCT0}PGbpD{;>B2j1>c;TX2>9b(8N|pMOl8_fQ^o+tBuHwFsK~~$5Cjs
zeWbZo(|3~Qa&6D@ckdRjh3Gq5oKOIToJkf;2oT@rIN!fIi%A(4I2%*;_k5YE(D5KA
z^ciLh6~lpsdS!BJ$mL>N6w$bO&qss5x?7Za-)?MQ{gzc~ks3Q`X@eX$1SlvafrT~!
z;C81Fkr(zA8!uJeOt7hvn-b5XSMjjKZ}BWdeM&-3)aofJ12EJAh1Y^b;(a4-z9B{h
zrN&JMgQS~=tbm*x1A)P3;o|~YBIjXk0R54oH0JpLcGAaX)d;BqGa|tQjUU6pzj~ZS
zVNzMg*jC~pHfq^Gu^_p4C8qDFeSp{$(^S_*Oc+uej$iptNc}|8E(Z$%5n@)R4n!9|
zk%8NVQgETbhHc#?FVcKo6Oey!3PsyrE7St{o|X<2n>Y{NC;S>5Ou3(z)aH{s3Omq8
zCT5)~x|>elZ=+{n#?XrvpsZd(TpDtWTRc&Jt6MX6@B?Kopkx>Vq1o2phzQpTzD!zL
z+9JjmEH4KkGJx*#l1MFuXlz=ynL=jV^F`^3yFpHX7I}V&)*>?uH2nPz!Rdza0Ya<b
zvK3~C1W9-Judl@-h?Z<@Y$~QsnkXfaUSZ=D6gZz<|FfU~bKv8F^GOCUYjV<>f^X{1
z)5YR0F|O%}TG1%NXq8hGlUbJQIR3eE55<Bs|FLpICH^%qbPHH|I(y9oR6MoxZEK0)
z_+i-qU<H-KotafvQ2W4#IOC>RX(|<kT8X^4vD341bP$6&lY>XB(Vv3Bo2J;x8@Akr
z`*cA_9+3aUVOHXe3g=;f2m~4={UHlFC}qG<5V{?bhd^TolN~voWG?2TrqEr0bXw-~
zLK9I#AnuT$@eRTDMi{a_0c0h6^xcqL@Q)BRay=Md2gt}+@FNtn(B?HKfovJjUp-PH
zR>RoJZ)Z>mC9WC!oM2yJRY1@Q-iFo~$m@k?T}EdMYJZN%g{2J*Xj8~XqH+M<ht@?3
zF6Q)m<HLpbP({W7r&+e8U!yRR$ix@w^OxvyKWS0`16n{tdO8z*7LX=jDcvO~4G#zU
zhcMn1PzBp)81})3wL8&@%sRDE_W5rn#}vnZMgs!>j0Pynq&5N1RCwtrD-#N&^*#1I
zDqEjm3BWo^M-qdabo6brIq7CZ!(Bb1(e2X6<_po)90g?64Jf`x#dCn+iI6*iJ<0`s
z*_3G!Wy}f?><L;a;nfVVkfcEztw>-K@i#Mm!}Ohr>ZKtR>0!-AfYlR&fPZiauaQ|H
znIJKy7v-?}ENaiBY79w%DqIU}INrh>2{QhUxOY5U!MkUdlrf)@(XwC_hWgIo(m-*Y
zLq4{1G!cOTq5k3@c<%y#X(o~L1-H(R52$e7KzRlwbt3HHA3{O|l+lqCxg{IKlBY3S
zVJK&tE%zGrLPQJwq!AQC9;j4J16@snbpzS}V~UYT1EkgipoQcn9;-q%5y>N~$;R|f
zDA$m{ndA%P=B_soAPqZa7>^URuT#K}Nw&gr?GW8IXSWm;&flIA3o(_q$(<mWoCc^5
z&9G&VL2jbhP5>DcXWjzsg|Ghaw0Bbp{g1VGC-J4PQp=<M_uIS0z$E)KNdoIP0-&b`
zM~|en%lhIT<f+0LoX!lt@m1Dpj7G;5xS$zy(=ibx91oIYxqP}Ifq}0;T0jjGD&VF7
zfkZm)(#%bb=q6iS84w@WECWV!fz+jBcSrsMppjt$Jab_n?}T4moK>Fb&wwap$=*cv
zD4pTfcK)2<mp3Ri7+bL!PanUB2Nf@xDEF*sv=GVRPRA}>pOQRl*a5>g<%#Si(cI5g
zTB0|dHKTUnQa4*^iSA9&6z@&>i}e~<|EDie03A0%l^VU*O^**6o%!vDr|kf$+E+RS
zHVyIV4(z6Od{}=7vgQ)@MhY%g$I`8}69s4mwm~a!BI&CwOa6Cyo|XTdp6BvJT<L$^
z^UR+3mNJiar91dUF?uIBJ^q`X=f78ao(qZJCn?l}U+QP@>F5o+fjp^(4_(inKp~4v
z{nM#z_pdHb1(z595VioG_*CX|jEzFFp@8=8FVIff?}t*X5c1et7Soz*)vH~FQOZL@
zY^fjGw*2%dDWh3rn7*buG0mu^IE_k`t)?qq)L12#fC?Px&N{1ZSTpbg`4XvW9$u#=
zt5NMb12apQ=L5cfB|c-N9?%e{c1glI#&qANif5ucVKQDL%^WX!SFZZ_nrI!XA3PHC
zTUD$6g%i5_zwLbH_mw1_&$mno$NE1zX;|%0>narkVx7J+{zaj33|rT#nFMd0k$+Jn
z^9@MWK4sDxSe#6AQ?y5urzwm%;VPT~Wt<}VB5C==*GiB0t!2O5dEaY;gUrHAdvAH)
zi(SwEy>o|FP3*Zhqa$4gSKf@C<uOgEtBJ7{w|!k_YE|c@qg5=fBVFu4FC=_c+ugg8
zVs_ELFO)-hRI&8^n#K2qOWTyI5bv!i6>4{02NLd$QnW9ue0?RuwpYGnJjtwU^0k+7
zO6xN-l+$uxDc_HxFQ?RbU;rjlxt<pmLlJq#M6v`0MS&egMRTX+P~2(Y3NC*?x*C7G
zX|ne-J!%-A0|>(iIcmjI(pXk-g){-Ie8sIyLMGon+(Jf;ldluU{>zBffrsebSo!*Y
zzxe-_=g&@{_&;M)7$vB4PrReXHO-`<Jf@#zDOvvtzy-v1zVdRRP@HkG`1j!%;P@Eg
zrjzW9Upx>fmWGyBw0qY85=L;a#{f$6_(Dk&5<Y3#ub#2Zk%Lijo~R4lGQ%k&qj%zU
zHg4L4*#)oEuaS3QUfBuFodW}nv8Ol6zd`L*LjDN(JD<Je{GPYr4L$`u?4vy7TsQ0?
zGr1{_CZPDVmWuX3egq-nw6Di&yguZR-}-E2V?%^F!yZ6E%by&$bjg(yc-?<iO%tN7
zkb=D8Uxhy?a9=SM4YBC%hD8he_Lry#UT$Dt`G~iPCvN|RK=H_|Wwu@$^x8H~5o;21
z@;LlK!I`i_1uu~PV4%p3VYF^j@OrF#`U63Zgippq#o@{6KfhSkGmd~E%rZL6hRg*%
zukq(%La!FKRJL#Y3%^;nVl@ivNXUo*c%PVxSKdJP%*q?sp=EAz;9E(els7HcSMUOP
zCdqa5?q6A)Z}%C8>mYZC+O1B~A-w*N^?X7m<6$6P;m{#%G`$pj`m++Ye<w6RFSHSJ
zc~b0PEh#Rsth~+8U#nL`lYUcDBgSE7>zR)%;b#R5e=b!Ys5l_jD>90t=bWP9_{#ds
z-rDE&7D)K~{Ji{^-T3uKcgPByIQ}Ka-U7mti?*Ah>R|Y9iJc<%v6uPh@88#JVm+3Q
zDWHv>qIBoV;&)TeeKSLRKQ%cS{Cws5ca!~aYas^i9xZ3zCwcF5;myZ4{`pWTf0omT
z^abXEA|murq+%N-dsbm(1^2dpIkKJ~^S@7eP<@>uJBWSd?jBlT&S@F+WzJQRg}vo+
zDzz^g9b;=KDdp=(M^JaNLYN@%);ivd%uG-b<`4ppC(Gj}UsTaR58c&47oDVL40&N2
z=D@4?`pJI<Jx!6}^nbdz@yh=h=lDO911g=sWM*vPj90I&K`;oJZ)aa$UssnJ07We=
zt;?7DFk}qO6fzJFa07-Xh=Ym=0CD2oEAkA<Mtf)x48#^WJQ*IGq#A#>8VX`iyJo@J
zHv1BNJDG^@BKad6rIRP!UyHj+3kwU+%+8{!kL;=r$rzB0?R6-aE~0;W)AgUI?P08a
ztTF~va?w(Gq!(?M`zIIRq(96Bt_g!h=;_HsAB;LCh4Ne{jTn_jH`#Daj`Az}Y~mf!
z-Czlih;Tz8_4{{qasT`L7><Wr_wx`>sWP`l@?lsD&=4Rf)-yK_?rzo35S)i=A4L<@
zx$zBQ*#dge9!rMN!DJvsdhrjNR_?eOB85Fjjh^*bsU!0Cuixo!LCdL9kXWzz*d@FP
zC^u$_y|ACf1w5;X@-vcvxp^3ciAqqpO+|s%v2|O+8aKCPP+2h-0~KZdw0W3?^PJ>_
z?rl%he$`9HA<^-K)HB}RjQOsQkivq2)orbL1}0~A_1TR${>)T6`7!)LNxUozB^qFN
z3`n?<_WY4&0|j&X=T9`FNCK`0b}>9U%5UPiFCI&UaerHO9oU;DN<Nc9+(uhl0_LT`
z7r5hqm##0A{Xvfv_uR>;avpyUJX7W6&G3VV>-|XW8#@7D2>Oh&A+lH_^pRz#1cYi0
z4N3k`X9ovNcLjuf6WRPVJnP7)Rq?7Apkf>miX|)%9`3`6jOW7SUA}Z_kV*&AJ7z!h
z$gCbTIJkh{_P095m_hu!6dMDi5hO{0K(C)AB;?~`vEazh$*>)avT_Wal)&RV2-pNo
z^Ac`LVi;%1&fbd>11gxw=Q3i~R$t`B2F3;fMEnAyKOWzqfee<woWZ%H%n9)j00nR-
zBQO7LGVozYNJyr@P8c?-mFGJ)u$uvTqg=4T6_Kg1X22$^%gevga3$$MFC51p+noJN
zk{U-%q333L_Q0}#D1r{rA`k}O?!lE;5JZRX|0o4G=Hbwhwj6t9hx4aS-MuYvHnO`*
z9o$SZ%uHXP9;a&a7*^$0@JeJm`y@9vH+8}1WK5z;jQiT^o*zF5=#Gg>;s*o1;2=$o
z3C3_R*8};76r!GT28JotgaFmze~w9yW)^7{&HQd|ZZ)qUu-sO#lbd?NDtRd(A;E_U
zC<T-;%^aBVVb3Mqj{71bq`3703|JnG<e+{v(*FyFY`hX{Xa(>M%r;Y4EaQy<$JP*o
zB4OP>Mb=bTkD;FhB$~=My}<dft9W@0-<Srcr{m{neX9TVxId2)in6mNmW@D;z&hSp
zlhAGL2xw)+&D=V>;<9w>p_-(x4A7{|onsCVG1{BYH)TLiz5$*u=A_5{V|cb_A!N8t
zb4P^7J47lFCZwvE96rn)T7e3}e%^MlST0_Sy&X|z1Ka{4;IyvEKsC18<9rlw60dvf
zMGE4lE7__^n>KEI^<5h@%i|Y|*4yaJpn3)cVxy#`^%~#}0`ArEz%|1^Z`CA<p1_k%
zH(zaj`vS8;()Dg6Mtt^u@F3T+!@XY(iirQ_6DFYg&gX_r8UM+qvAMY#LvHF{?c28x
zS3-L7@J_C-uK29L$Vh%28{86mrH-PK5<uE<Qc#LMaT{ic7vsORJ_B>1+K8RS$9u!p
zw)PFxlm0K+*t;kM>}fH15}i^=^cGd)?iEGgC>?N*eAClVjqe)V)6);soa;fGVaT!5
zpu5_hK81a~wa(&eoLB1Nnp2*o15MSEN$Rk2!0_c$mH@SJ>o@`)id_U2dOw};9XyWc
z>+56etw%7K()FDuBy(OQ5zwq$eF`|+8c+FQzzSZjd|<Z)Ml{CuV0<$^54qmy3Ls@;
z*c0n3&<V2$6k$(o5I=+=v#q-R){II0OjRDj$D_l4toAil<zocPslE(uT5kXq^xyS$
zS>9Pjdi(hJ1c{ToS~-2^hi>pSLze%7e8kBcOo<BY-wxT{K)pyVUiF*-o)_Y~-41O<
zMMdx(8!A9C;&H0QtjE|9bX_gtZFM}Xv){u8>3GBjYG5a0vh6h$KQk{xI|lZ|4AK7C
z9W<97t3z3q&(<3y6Yr3yW43bj`ygKK=k2{c1jEIKbbc!JqsW39+__YB?!bV6yFKsz
zc7^sNnr}{@OAU7A0geMS(W7>}UTKRd3~v1-6qLRH&Anja5w8xZ&lO6?4q#*h@8do>
zd3oINH=7>_I!{%QA&Wnf_Or0u5DyLsVdwpRh(+OWf+k`pFnyDjoAce=oSjF0jsbl_
zJoh;_?X9$O+XN1m>eQNl+3GpALm^kPaf_I0H9G>*kos$}`h%M9xceV#aK~MCJ6B_6
z;{-j-9@yTG<KUwuGy5oM>VpM@5ld(LUTy;lO~dEs=a>1T$sHDAhct;e8{Pqz8+4L8
z{K1%4f8v5oE6oY5wfdW>OpJ~DoyzYrg!OODE+{xrn|RF7kZ@72h(C#l;Cpu~VuUOW
z7&WJ(-r?N**|Pt5xE~7(G5O$B)F7$^9#nLd*CoWo^>4)K;;eYg2bzAh9mLd%LpjEe
zftaorhy#=jny~-k*a?w0=t4UqEiGRoB0mpN57@iYQO8_n(*ModBa}FK^5m(gfQlY8
z!P#Rp`UEXBqiOjtvLPu|3MO!*t7|`Rby*oG7G7bC2Vu4>-deQt^P(L;8e@$u(=@d6
zTFdrfQU)B-N)Uodlv*bV=Xb4SD5aDk{Q8fBgW2ZP0e2uB+pJI~A}oBLQB?C$xya3h
ztSgJKB15L8XpV`^;lZk(zHv?c>b>52wBr-;ddNgv4JIB9<KqH;R|KFWo_VdF9|F}i
zmqEJzbcv)#f6|TF0&EhPhG#>Kh>e(ViK_8(PCAx$c1&nf0m&3U>m40UoqV3U+ZQJ1
z(SZx@=?etP!wQ|EE^96efal9~oOq^r0pL3>)*~1cixi%Ut55Xw^hl=x9x+CCdTU~k
zB4!9bE6PPwaNx0OCRV%Gmbj*+!rIzNxE`mIJ3UTjAgU`UC@2r+$;r+Z!6*PHCn(-r
zLmd+ZRbLD_uMa<H^tUFNf{Ssa?{q-DoInv`7^B^XhxNc0!edW(_RPBuqZb9#p50HM
z0PSe|3+QVd(64uf^6CXwsXC{PR8(IlxRW3=s*VB34fc{|sHen!UWw7pwB@<-G|AvD
zH0a=3pc+g>>ydjPW6pP(Rb4*q#jcEV9#>ByHpz{DVpLEh=TUQ=J0$q`*4{2-L2X!u
z&lZkoZiy8S6=d!k9YtUnc}>fYLFD3(3jQXVrwb@V#PSW5mxNuyV_s*l4KC%)Ky?X`
zccJK}2fJhp^%UIo^O_kDBPv^3e&BEc2Fp};QwVwVC<}!-y_BAB5ezqG@j!^?T752-
z^T7I*D~f_B@!B<O5U-T%d4LY&Pu;Vt02K_P8OPK#*#rkb&&CX+?opslI2@7j7Smp^
zbkk@#b9><dbKG{Bn6VSbZD8|JP~kh5zE@nl^==`Q)G@vHR21gt47OxXpfLY1FwjW-
z34RT}o&-=E7Bl`vQR>v0GlNO#2yQS2vuw^t93Elg`t>s?S_H|N{D17dbx@FP*Da2Z
zEh;7|2v{JJ3IbB1D4mLQD@cQMgNX?$DIg`C(kUPqlyo;JC5?2Sb>sVf=X^6~=FIP(
z-^`iwGT*!(4<OueUHjU5?X}j{m`CyVS$5-3tl{*!3>pI7Ijn9@(9wPW9FAeDh0Ei-
zW@Z!c%T~7r_Y#TN?}-V*g7{jh1tx?-Z#Ry6!5JNvpLIF%W{$(GA^+hS0i)%cH*aE#
z;#?tka5%jCtcgth)#>UHtb3Ao^jK8j2jQ7d&(E7-+ymiTgt`JPI(Yu_aC0L97a%o9
zyu}K`GQecwQtaYA#@u{-pKyY9brIGvrZBs}(!!zHi_F)r0`dF6#<)RO$d7RQm}I9q
zOCdFf))BIk_S)JT3k`r*i1WnR?-l6C;j;9Hb_uDdsNj0>fhowKnc&e9#zTvLbe{;m
zZFWWkKJOEpPu=p@76bo9zDyMO#7E{sqL?Q>PEc)B-uUeQ2(A9-2eEUBkN-XL{D1%B
z5i$Kkl&~07M?MN@Cpi;|OiPQBc8(egoa9qyR&oEFEId3|=gLZ^6UGxg11FqaNhl+3
z(WiN)zT){h<J5tw2pwbnKfT!h4yb;EQ4%G0)YnggELk^?*wB)VDi1{MYf2i)hF7Z}
z1mmm8SAV~*4`FpQZ8-TKu$>4r_#j#gNA;j@w^^US;b7kxG+%{hJGRygrQCF`9ErsR
zyGgA|P1Peu{Dy{xK8+tbLG|$IC!tA_K9MbC6A{+@%|gz{B4?|=d^`PdS@-x$&2M>A
zsqd}?H&c;l#UNrMX*2S*v$uDr=I8miw7GpAWE}L{pSZbU3O7F|=N!%~<gQo{$jwKH
zv%rH2FVvLOMCA+0T1aysj1F6#pD6kEjrPo$OT4_mIy#!0o4dOUkqzRS{Sw%LB4FB|
zrGkR*a3OrzQ89F->b}&WO4bE((*X)Kt^&$j3f#o|_iBK0ASec|kpUzGNXz`p%#V@F
z|12IY+Wbm)ip=5E_WFH%BolmrRVBoHY?KVM+4xf+=asgvk=PT?V2HW(OsK<e5|t7o
zZ1@S8Xt9N&VslIkJsDY$EhSR^<<VC5#9~;=Lz7F<t=pFbND*}^=UfbQsDK}?$UFfP
z5u7~OrL5vH1oY#u!v|tI{uceMm>NiPke^8C>FHgQHgxAq?8SY6iqZ1PKJ8{~15-85
zEoo^NBw4_(Av;z^%8c437oi`tTsJOia`Gnp3Y5;ha$#X%M4G{m=`e%^3ram%zsHXs
z6Gd`&clV@Rp(E<t`9?*}RS@4*`eEgT?CcvrRb+<aD8WHGgnAHaIYt?toqe`$#B8la
zQJ0;V$8rl(B9qN(3Wll4CGJb}@|m=|oA?^;Nj@|X-<I3$ijKQDHztpL%zMP=<?qk!
zIA>hx32_~A(vB7z95WehMR{joX^(0K*-_mt*Nlj$sH%4teuYDMi~-wToQc|yV&Xw{
zg7~5!%6y2Lx(i7m(pHMD1^n-}-A7MQ$P5}{)Vc~S*22ZGAlqv;HghOlVBmH}^7~Gs
zZQHjKgO|!3Ehavno#4H<GkM3$Ex3h=oiAN$4syrfh1VdpeItP0{{S{X04<U~8%s;}
zu-KlS9<Ya$Vl<Pvv9_>Y)<-@T?oJ@4Gysn&DJhZYh~jX9_!ab>AEl+qH<w|q=)5o)
zW>7*Xp`bGepe5=S4$4_+%p4eNN^D1#)JTglYJ?8l&d$!**cii^kT((97;q$V^77&t
zux_BPA0#A-;VhdMeIg%jKlRi$_?y7c^WzOWs7Z#<Lnz&jWdE@QD|SgkdK2cR=>2%7
zZ4<2(4oJ>?X~rOYZw?+Ra*Q*k^*=!sy9|E#NJb$j5hz6c>mN?q+F<JsTzn@3Q8FK<
zH_a-s_*DmTv7S0b5Z=MPLse@9jVTs5$~=OXiW~`x4f?jezAam~rdVv%x<tB-c#wSm
z^Fac`e?#`z&CO8i5(KRT#wlfb4@^BsB`@URF(NmD{^YPPW=}5Q5dxAVKEu5~83`Um
z8INqlGsD0f|8fbO(kJ~uK~T^y;IU&iD((_+j6D`7(jT~Z)x^w&@Y&y9^Lq>fgY4W~
zLtp&m0<&w<c=*v8FI8ClFv$YBTdaqN#|IyNK0ah#xVLbYW0(!GQ-W^O#Mz3C8#w>(
z8@Mu*MisKiG*i2DB;?d=scg%Rz^>{|_sdbFWQe1HF_3;F9%%t*P!W&g7oztop`>`^
zaGLqx4@^zVzETETF~pWg2iWm^0}hHA4bYj$#}h1i$ZZ&;0?FqgA&8EQG&sDpyiC}M
z06H2zfT@DEw)5cV%RmE%=MM?R++0q#UwK*CBIrk$CKkzJZezo#Z7KnH2vITFq}|o;
zU{zTeWILV9Y3JoEEK&z<UIGbe48>QH2&%_X?CCf$89)XB7Yz=C_mNyv;>57EWZg`1
z%pNS=LH(o2Ab)}L2%CR`k$cih9;P~mA3v%R3a;1fxwmOLKX5S{-cNu4pFVlQZ~y1a
z*|Q}T74hhB6N)I5G{(1h#UuZ~rw;gB4$^3>U@9uA3LnO1lz4z|G4-TgJjhQNxud8k
z<`Id&0r%~jH*W$0RDh17jB7;Ag6<gpk6AYKes7}x;zI%FB`X{OAiWpZjH#%py%!9l
z*Z7p3m1PT16TTCxV2e;#8<5*KdyDt?etAmxgaA544snK#4%snAE5$(@4}jjRG0x|P
zI-8u6(+Ud5^j(#&ZxLg2pbtaE7|`m`Mzj)v8H1qml#GUsjt=X(sj{-NvojyK10n3j
z>3|BxM0y!`GCm>v1Cam&SW*9Q{W&TQj8Soh$iO1!s%tq$=D8ve;}Kx;My%sdZ|@B}
zcC10$k*3P@*#%<}pur9lp#hH`eZfjd-$hKN(XTy~D6axMz^o%XtgErj!9=A4A$JV}
ze-i~UDikq@oKZot`39z_0)XX;SOw~V)qZoe8-yC7&lm7Er58wXlywg_61=l?N2N^`
zab(5gbWL<1TuMPdqp%<vUH9%K6LY1Fet>U;Bpze`5?XFd@8diAtfwe0VLzV|pT(Q~
zd@UsZ-UfZA8~Ng9lNy)9Eogig)Z*UAg`NEvJk#06b&gp~y(s^UWBo3v5#Kn~#_+hT
zz^wQ*qfI6M#QMjYv3GwzQ_|s4=)Bn4|7D&mn&Bqxkja$bH^;`|#@w4D7N-4bLF!(0
zo~GZs_?*@!$5r#vEG=3&uLqEx8&pU#8!M1vxOh%3;#FD)_hs5Ol^+?-S@Tq{d6K;Q
zQ;lBS81{{8&vyE?O_$P}Ri!KOe&9<l2@Byr6%Bq_KTTJs7eY?YzdoaPMs?CkA*^8x
z{Sk+~YU%oKgYPgC{hQ(X*`eBN=~aLSVv@YjMDg^j1HPp0ObuZk@`t2)1CGO(p2~R2
zyP_WuUTu9t;o2;#d?m1*p~COr2VLtN#qoVH#KPRXY^>9(OeYopIV|v`2F2m*)4fP*
zC48C60LRYCvkh)lWbnb_GR0C{tM<f$C(3h0QgWx}2E616tfAk(Z(?DOcNdzKN|@{~
zUT)Vd_b!`_0r5_#nl=VKA|L?zR5A7vveOxjmPu&VP_V5>;w(h9Sz9j+`0<HIzTK3d
zH8-H7116cEqT*m}ST`K)E);0a?*)gQ2ix=2t5<58suc_vd~%oT`0?YCrVN+_YNt+s
zUWt4<rTet|v9HByqQ1cmPSUOmTT(e<3W=CuRTL8tf6N5b(h8k?g)|~3)A!Ar6Du!|
zP*a<}O`^aW9#bu6a9h#7Acsq6h;B!Aa-(E!q*=z<SXTB;vY>|Zp96d``}z9jd#;w3
zUKN|vqE_9Dyuo|o_GGA$RicE6ADPR_m7lH$=VcDfoj=WZt@`9yn&3m!KMOkwmpq4E
zI~J}j-1=;`iDqTtYGsI_i+xYU%i1HB-fgb)k8iTcDQGAu-x1rh8r0l)aPx?h?Py-Z
zk6X2SJ6;wxTj|Qxda0}kTFoZc6^mC?Ggp`IU$5bt?^8?jsg!h4zPP@pmAkKensJG>
z&l&?o)|rxWv*UCG{38WenOCUrNZf;^3*<&hfcK?i7+n5&_Zo0H;d5moDu<>ZZ0d57
zDvjVEfiB1Sdp%d`Ig_eXQPCg2aQm^t<A*VbOAxo~ch&9J9U!GXc(5Lk5$VCXSyPq*
zgcPjElJDP#Ctm#!oM;0G3rHZ+^@%qU*lk<3pgw<jIS<~0dFy&_<qe+rn<u*a?vCd8
zyarSusxwMXLP7$0oDwh-4A~1X>c=mJoxBxT|4ZHGVMrvfGFdp}SA;784uC-($4zxb
z4x`~<Xj%(-KGN$EfeOGk@C^xs>;(bzd(619uc6H&1d6=8yzuZlz?K*klZc>1@QM*^
zn_WjSQ}!`2@$5(<lZwmKPobS{>Dh>jMm|XMm31;D#Kpz2Rs_Rj{eZ2p4!ruofB=3k
z!F?@HbK(^;c>#wR(b;TJrx2vMhVNBQNFzPU&?;eiMhwZ7338(NS1eL`Ahnfj!oKgs
zX}eN^zPUd>oeFa^hs8MB%0_kATJPs>b}?SRv&uFrzux%q&7H8`J#G`nlB4w>eDIx!
zmqbshz+`bOi@A__NJ)04wVam0Y%DM1ndSaoxjMCH`9Z5PK`s<y0&PWASJNN5a<7lA
zX1VsPJ-AV6Xl`|HCG%UBQSQfBRR%v^9-;RyCH47EM#wyS`tm8aUV5A@0}VL-R>-i<
zo{c69XaG!!iHRYRLGBJ@#p>Qdb@f+7VZ!p?NRFUmBX7hUN$3a7fq()zBWCe^vNK27
zFdT)9WS}a@=-Wuj!<d*D982-kFaX@Xbt^J<hRc^V!envuiU<q4J$~#2EzVZb7weuz
zDGj7u$n5Yk;^=I#J%KU~AbY@m|1&qIAfJsq08_jy$6hZ^w86cBNO@l5;CP0Dh^VO0
zJ41{@1%P5KRv;SV2|-H^z~OB4@rUituR9Yx)h3#TVYx+v>Z8&V3~+rnQepHwh$(5<
z0=oq;UDgpm5|MAOyg`<Wd_&ST2nV;cKO33@w-~l#C|2nn*9;t+5GG2l+e*jZWblCq
z3YU5pIRU0RAiY4-lAu&$?kq8v0ePXyOG_h+%JNK3MuzYK0B@49C~1>^eJTbN;qS>M
z7UNWHZ*6_@<cT{>ZINrCH3EGH)DUqvn-Ov*ZN>dyr3m;9-2<#|q#}W>+PxDW4RzCP
zg9n5iH*MM^EM@Fnrh^Tl@R}nVvv`5h(emn@+(PD-E<E|B@Zt6f;k|4Rw~r*p_Vn_<
zKCZKcY_Y`quH0o7vknJ^oH(@)e?F?aWEclB{GxDQPVb6d8u7MZ=7{t!4dFQW?%v6R
z0>LdQh9axh;<?^x&3DvKicGz&$Vv5TWgbo*swcCwr@Gk4sLmQ;_6+Cdxcqc6ug~lD
zT<iF!PeqiJT9D9mAdN)bNKP+H7{%kQ)*VkbD;*pOvHZ=k{BJIR>)P+WzC#og6vvMf
zKYvQ-#UaR8kckoUZikr|o}Vj7moR?TD6+zSc`os0tQs6JbFW(ke1TirZZq~XXU?#*
z%h7qEb4P;%5LhtbNOI%DV>K2+Tm1SpF<=s_bsY;lH1q@(?JAL^;pwB}O_V>{+S+KY
z8+K%;pv8*e*YWX2c#Cw>?}?%E;2FvH*VNRg3GAqJYR2goU2swp;|-IOlTjAtqYNR+
z3o9!o6q`86pn$=&GsDJbQvmD<Q<9}TodT3zP$b`suj%99z#wi8TL!$g8#iuXSzJua
zz_J`|&#HdB86Czn5CJwbepE^+eq)HRC@L;q$3WcvZq0#fV*u|%TnxWz+LzJ$l;e_!
zlQ%a89p4?K`JX;rXLNLM*hPBq;K2-E^AzDgUs@i!qO!7BApKY&8t^c<e*G&J63|d%
z<539u2vR%l9;Q^HmSS6O08_4AkyAjAtCsuH_XBHd@P8|7F{fF4JCGa-68=%gY*5vf
zBIOu-6qYjY4xD8z+FbM{?+Bj~Cy{fBN-y>FI(uv^GVwvfAZvczsy3%R`IMaT`agB(
z`2QJs@xzvUSRp)6byEeCp~2{rXC4<1F^nkKSk#wASsNIKr{qMfJZ2z|&rbd#SYM@Y
zGI8rNdX%Q{WFYiY(Lml4G>)CQc(LJ4y92^LVEaeOM8JTlh}W?B-Mdjs<^}gwm#iBf
z-M@byN5OT)3?!G#%Bc}J0+&?Tr?&zm>41?Rs47+&6T`zxNdIGw&|WyRz&@xR$nThG
zc=J(s_%F28tlZ<VlwutA{K50VjKaY-MoYCk8!07N!ODA4Vwym1(?3^;WF8ALWsDE`
zK5DsQ3z%~vtKr|EG5o@gg-B(;daf=A5W*%~IALytz!4E0EdV$Q^!MUj;2EgMwP593
zfc2PmCG-r&ow*!H<1uX)^$)_|VUPM&Ov?jivS%XL8$A=sqetoA>%GN*Ry3m%G}B`i
zWG&VE1H5b#YNNhAm)@_EcGrDv`8!Hnxv{XQ_+q3#7te~ewYMWZzfnz5?G~@0?l}Ds
zm=d4e<mVM%6ZDVXFHLQydB-LnhUYb0qOclwcP}w6u~x2cq^IcS+efc@PELrD^L?t`
z_|(;5Wq3+)d!ga+j%;K$Aqh)54Z$}`K6n@?l>VsVBex!HztcU}QI}TxLbyO+SwZ<$
z<T$G}Z{ozo<&Uk^U3>>KTMIp^IBue%SZbOp%u=-G=u>%?{4lLUxj9r|C(9Yevqnf6
znVZuSG9y&N54%Uv_y@mZU<(V=4056SagzD>O^fuxQVa68tfvcG=AW!vtjE2K=TvXC
z>izxXm~tECv8+D^hkK~BMvu`>O85VQ(kNVSf?@V%4d?;jp)_@SXJyTU_+Z7)i++{0
zWEw*GLF7D)!#%NZBOJ)OTXaP22tH=jTh)&2W*by@#M?!{mVb@2FU<8pFBVknfGPc<
z7m%9t+3de-52Qrlxf)Be_3PSwDCtA^7G1|hzkd7nTcun5HPKaf{eu?)EDI_W;z$6g
zL>^OZacKzfhfQ~{_Y`gZawHU3N~WQ$5cn?0qt?*53gUS_dOwLn<1VqhOh%Ce+oOes
zGsMop0q8QJoCBf&UdqQwx_9@EoNs@N>8Yp%674VKMwRJTvX^J5rBNBB8<-b|?b(Kw
zH5jK2@2|&3;bhxG5y73SgSF{Uw@-M#`hwa*;M#cYdu_uuo5ul~?T_!c8gD9kbaDsv
zAN5Hs47kRT5*@LVx~qNTQ-LP}q!Z}{^+|zww>|H$rLl(;xHw$=GgR7nI;7B^#jDJd
z-P12fz3HCNm$!K$rn$Mlk)fKOGdFmVZmesAVrOijy-;&4e<Ug?fHx(sfz_x++~m8;
zO7+j*B2nEE+0nzpufo~aUK|iRwd}BD=p$-#(rd4zij_lcc!bCcV<9#1j;0aSWs{M!
z9f|U$tP|?-S-q8UyJH0+?yD^em##iu?&P#?chz$hG!(?7&NQxSh4-2lZqM)N6;8TC
z>A^Ag>VRGJA9J9mjnfGORr5J(Q^Vbbg~3uaNO2ku@&nVGW$iYS6Zc!I_N5_$`JCjL
zt)_=a3wj)4%zVzCIRo~`br~tkt_-BCu@l*Qk6)5?^#c_FaXZF=X+?Am8A0R0hj3))
zT9xV-U*NR2v)gg{RyZK1Sl0u$$eC3lxUIhJLo#`U9J(DjOOlu32zEmAwe0tfwE?Js
ztN=8K4+4&$Gc>P!9I|?cx@JeA3>5Gx{5fJ{7>RI`Ws4kKp9wWDhKnL~dj9-5LIVb>
zqwWV<6{x@Z_3QP@)lIr=%HTBUOQ{zg3TSsRs5|~g%bk<C>b>d^HU-@I_Ingz!pidj
zTRRy0BN6g<B4_@f`8<(d={xzEv+K4M_Vsl<0U~HYfB(lh_2R?r*Y@$1{?VS)dP}SZ
zEd$P@e4ny5!UK)Ce1u0gm&QBho9zzHBcALM8qp)274#ezM$O-$ef^IRzq-)C?ZDH2
zbo<N(GX$vSI`=Pn$*0~EOFUq}a%NQQU9tdncC)(5*#zTa{S5NU%#WXV?Ug=I@MxWH
zZLn|iSZ5@Kw~^MP_|km2#Y^gzTW*Q_u1r+xCWUz8pli}Rd7`IK;L3r~O*-eG6_n$*
z!<m5G!)Bt(q#@arncS&pkFiKTR|vgj7l*YDlV1g4h-mb&;X!!ttU0sbG-bkzq1}7K
z4>8;tf<ms5rviDy9<yGc76Bs-anIeVa?AIq9kT_eZuoFU<KqW;qpq$$Zr{HBxWaE?
zVL{)ezCL0xEzC10C~;&hT6b}2DM=-RfrZ5suLF6iJoCv>G$BCpM)iUU)uii_^1`lC
zom8AOMr(@AWNt~nF#<dE@x?+5rEsa22c6At0m`ukMuV6g5gGa9=~KA73MrjsW4k%x
z^9hL(;PHlaYjblr8oo?o-A%X8Sw3;;+ZSQ|wWC8z#Ze&x@wXzLmCk9~d-N3T+d`J=
zbd>KM4S}37mEV2zbe^3!zWZdBCq`e3RXOYFY<JwgZn5SSD6P?G(pY;^<V%TL_wL+d
zv%20(nk9y(7z31@{2JA{Y;_qkY?oY?R2=mV_)A7Hj_8b$?#!oRP$`(7)6mu)yO{xo
z<Y4d5-d{=iVpCp<GJr}BCAuyy>Lwo8&iizmuDeEIOPf^}*^k$Jd^1vlO)HFa$Dbud
ziu*7uG3qB3A>xs4cF;QrWM~eerP$1qg|-tK==XzANlfQ~7#Vbd?&|95nAdwG@_PUv
z0(NaiErHr`8t8OipN@J+>(Vt#Y*8pi<Ef$6G?-J%myVl!@FHVSa%Z%4U+%5UlaLAL
zV(5EiWGpbyvy$($dOk>h$ohrmNBoG0;Keu2C=YFHGK1fUM&1Y)E*4Z&eEa%X<s0d2
z6JJTXmoHwh8Ja5?DD;<FE02JKMHr~|3XaK(V*K}^!-xBSO^3+PsUIY<u6$lEpVaEX
z(UyKm-5s{&e_;Lwpe1g7X>oC^H7y#6Lvf=v3ajw01txy}d^9)$ju1nS)YdbP3R#&-
zCxwVl+($<%o+gl!^XfGHEU%H=4(i=kCr?y_6g~1MHeb;PV@^9eveVm-6<sT^Ub->f
zIvrfVU$Xrbt9pc05=VxT6Md7<;HfXea%@7T_jq`oInhhpJIVFytK1#312j5!^?0~f
zs!L=~mIw;%I#es5BO`aN`DU@;#D}gMn!?9g9!k<N4k}blme<)ue=qs1a{mf_bEUq;
z>h2n~Z{Pbkc0>&J^d!SPfSh@aw}JiKx!*vjj>0_!9T5yKZUcge05+5>ANg(N2p<DO
z98v2VH33PytFJ$V0m;Z+XGb$ne(H`ddv7bn&Q{g*W_rCLO*FmjqKWs@-p8z|4<<ZZ
z_U_$_#Hf+!2K^x*m0XjPU>a$M1%FfmwQe6dFn#@`vyl?m3{RfqxO_}5RhjtU=YJM_
zU0RhS?&9LsR^{x>bjf)ddUe)<fr+<&UkJc^B}UyldV0{~oza%6bZ#QR(CGo!N6*0u
z*^4%hApa^^?USTN$4doYF0NmH$4Pd>(%Wn8y|l(jyYmqo54S6??WX?mib#MxE4)x}
zZG5`ewCXoMtN5<|Zq-8@Qpr;7G95NYUQ#_3jWaJhzRUIX6-Y%&#*yx45s3B-|FOR^
z?^p&|g5ycjj%U`a`j)nq-LX#ZxAr^GFs+-ij`jAM1NlYf4@lhTi4i(Es4R}dIvI?w
zp?_$62}@*@$>0}#p1g*x$_$@TRo>lF-@Nac2K$S9mwb2qL@t@6Q7Wv9QpJze==Er&
z<%eI>miZS<_b(L|i<Om>h+NEVTpBo$Ya2X7cbm<hiIY=7Nhv|;LZz67f?<7XxgRSl
zSb%4x)6b)*U(FI4;c`H$!{+wmgYux`Anj~C+#vCZG&b{g!|lfB+?;}z0#XN~4Da5(
z+m&s8ZMLPVO4`b{ice(n%A{5(P?A1B;s80}RrZI)&aV5^#>(}1A2yMdA(MF7*=R&{
z^49I&%BQ$nM~{ykOAySoIcZq8U5G`R<v_><#lBx(`+0nXEyo{m2Pz?JT4U7N1H7RP
zcZo^|a%|+)pMYN&bgj92cpxsMt*w5l^!D*NeDr8c!4oB9@F*at1YG#o*!}<w!dW=1
zTK+-8{TY%zIhhCog`oJ-W_$HX-Z?aGht>_Xv!TV%UsH^|g<sqFX0wo)@<>iR(-K1~
z)|J_Jf1&SHeR*b8ACB!(FT5IH=`oW|oU;<mDz`q2ynId`=a!<Bt(yDX<?Q}Wt}&{5
z4EUM-1?uM=+avs7aXQ?WL2+_XuaKg`A>`8*GIXb%CbcXG@F-cHd_DB6MY7vGr(JSf
z!f|J#Fb~s9DoS>1zZbRh+=yAo3@_ssm}YLQwmVy9H7@CI@>Ca1dHC~zdwx*<kZd;B
zG2zilpLYSvr9*1RG8XP%?24WJ#WaMXa%phI$i;5QeBO=MPpfCY$DX`4o)`_y4{%i^
zoc@w@AdC}MMu<J~Tv<Ni{pd*&&0U8(l}<pt<X?{vz%!uvq(EP{>Y)pR>WuQk0_=6+
z;o%^oR(7)&-F$3Brl+fmVcz;CmUjaXx^0GPu3CB*p~%_O>MCVLA&L^($?20m6QjfQ
zC-Cx{^5mG?z98p@ljc`(z*VKCqP6zBf`ix7dfyfnt|QcE46Ef@iKR=`ZW7-^q5shW
z&1L3z{8N;+O>HM+qVB9q)e52?#h=Hh@_uNSIC#l#5;oft2MoAHaX~EL5+Nx_bPbVW
z#(mxpK$9;c`606tkY02Y>~0#RY5eQqzfsfJivD2JWY;z_`~J=IWRH#9J3R{d?=4^H
zjoS51U9wO_-{hu-Xr(8Ikgejv=he-L-w1YE+^Jcq-2ID}YJ<4e>{qHo1-2b)6jS}O
zaLJrNzBPN1SqzsHSMQ#62f#9t@3NXZm7lf?y(%P!zyv-5O^EX(7~-1Sy<)RUu%$sq
zej8j<KIvEb-Zl|EqPXoSB?44^26OO4NJ*ary80WVLQv68Kbf@*nxs08BW-(8(P#Rr
z1QU}K(;yE%%GLKo4YD&@;yrR$(9v?JLHNuEpAnluIofDF9P!odQy^Rh_c1r;e#dj!
zsVGlIN_vomGUV;slRSwD;0UYG0?+yhIyMW_ZZa~iWLc}b?z20(r_dFT_iJ*=LM|q~
zYKPXW8`&R+$({!e4L<q#t5`Zutc`oU{CbGAkurgT?;ZI@DpK<O(e_i@jqE}mJ>|R6
zCf_&yEmF_#F?Yi4tuHy~?^x#9#EIFT42riUXZ=Wb(`EIWb>xp{wO>(CkR8adt(amR
znNV(#OD~{9YEh7}>@j!b*x7f3r6tcvckZm6<_B}yv@Kn1wxzLAK6};6Ve%gO)hK+E
zYygMo$1Eg$HVlxc7H1dGg$3wvNx(}zRbPi1Yj_^cd~C^eUbflXesNIU2yAi`zo4+~
z<_q6~Cf%9$^@sNF*N&aTyetl_&m;%;j<(#;5g0`Rz0>us2l)IHH*JS%iXKl>Bw_M}
z3N3PTOu2U;w-J+&Kv!8EU0pO_&*g@Sp>K1KV;_WSUcC08Vd8DR260<Pi?9~KFf1`8
zjd%rSGc&Ww17u`mz4C*7eUg|_6MqjcySV5y5E8V9YZ=mo1Ib;F5o!h|wL+KtrlY4o
z*+X|jTgqsi-tY$N!>Bo;_joDyOd%!EeV%?R_%Facr}gt@5lm~O?yp4(g$=TEq@@@D
z&!b&M*Midw|FsM31kk?!*DE@iD`Y_Rkw+zvvUIF}y1kNhNI#F#NYwu1rmjsvaR%$-
zfT4@tEM0r-^CkH4vGz4hvUFP>8>@S2b>`25<$HIPlFrPG<$!R(Z8G3}Z@)78$?7Ry
zrg7jL5S_#)=@IPKwG~vWhfHrr<_s61l^GTvgpLwzGo@*OYTBT~myS~xqvci22A((w
zgl1=7s!f?;K+(g#;+DoeLBOdXQ%5;CLD7Vy1G{<W4zsOBCc`>WSn!ku?LlqGztDDa
zT4V!97@-a|)2=5!5s*L<7^#bVIa##`a(^hbq&6+V#5d)&MyU*hW(5*^&=bM920G<|
zllR#(CFGuva$vjhqYFhE#k&V?LJuE3GHQsa#bN;<fNnoG1O<$Yj4UkU;0puT+<|+e
zayS7V4(hlkMXfMjz6_lYRt5Srd`=6**iHgroK(^dHO@z9!+}%0^HoYp3gM}ZTnZH`
zI~yBZ2pYi9C|s*sJ7H^M)3kDX5f+(8Q|IK+zeuJ;uWNXh!`<7#yg;rau)n{n-z$(u
z=bFIW*WpsvKtLD{4G@Gh(rh6J_?qrInNwvFlc9o>TBQGNe?FaOAWDT#FK>6h=xZ{v
zZ(NG4m*Eg^q90PKX>84SBJ)yUX}z{Q+GnPd{r%S?sWD$ZJg1Ri(oFx<p`r0X_W9+E
zC1Z3i-h|uSdgJ4jvFa9<RoT$w_VKW(VXtP@da0g>8`n{daGgCz&K}v$yM118TweUv
z@89i~+RIP$L_*YSOt1PJSW<{ARZXCYt_OV%DkZ{NIPnj9O)gX4j>MI~nht(hN2t-i
zvjl+CsGtDgeht1Yv}$i|qa_cizI9S(&GDyB*f;m@lY}P;O5n@Tq9B!|QNM8uoh72J
zy7qB$JU=o<?<N>5aB9#)T3*b9v+nR}N8OkAwkG&*bOvO<&7)@|gANwAMMTm;=;I5a
zqUeUl87lD%-(ICmV}A$fCa?xv0Rn`xT#W(TkY~GI{WAa-DQsy{QaFf^1gn3tOSvI}
zE`cf7KY%uy8C<7+J{e*bVTQ-;zAD}11Weon!J;2h`+A1g9v)Dhq{_>f{1bU|5uWrr
z?_5?VL{tvxrJ*B)Nd?$Xnz>f73X4bvXIyW9juJoAv4;_Kl%k>{5O(X@u1un@hIUS~
zYY*PM^;p{7Js@on;248(b5f8FRjDQJOaSg}==JHa`3r^(p=E+h9dlPs<nEI}w+B6n
z0XHkWg1QVf4|n3wL`B*4sHZ6XUx)xD=ow1nMUQ62zjlj>OB^tjtw}K~@sJn#zSzd}
zob|dXD^2^o(H6H;BQNyZ9Gs1uUQ|^&Sv_~Aq;xuI=X8HgFU2pux#6zcIFpXiQJS+!
z=m#J_1DEb7b`Pp?AoT$D9TC6~HDGTdp5Hrz$t@@i>G(Pt8?6yV<4Ukyu>FaFLwK%-
zcL@m!7K3mB&<%aN2{eyTv5icKfwtF}XIqKJG>UW(CFrX8b6|}{>}g*n!m<WY5nRk!
z5ZvHk1k@HyM9n;##E_6~RPncbuxU;5zwO6im{j1p20O=ev+gx?u4-#*cRe|E`SKdJ
zJYGF)5j#Qk+FiKca3D`1TzJo{LoZGQlt-{IRQ-$4L++54n(FE4iH-*`4hcP=U56Rc
zja&cV_Z=Ah0Bea=iIG4<E35QLxz^)Yw#@=lA`N&okEi$V=2J&9E_JC!iIA^tk$59o
zT~R?$k#floAOe`8o9b?u3^-w_a9KHAssaML+sNBe4sxMsJB#a9ik0{PleVZ?k#Gne
zFjV)gw{xi_ksZ0Vqu#<kgFYFLX&g@i?KXCH_7Kp&FbV~ii3cC&We&vv;qndAh82eq
zDUrmoJ$yP(XLp`?17@IlP8{>3r|jNnc(CbpMq0ZyNJ>ia&6avzqh+et#(HMSdgh@Q
zsl~^$1AfVxbKU!ut=QIYJH5|FRtK1K0+kOeCjyhm+yH$1J|;_yaKTgJcUe)Q^Fk<-
zoQdkjX2xq|%;j+-h$`qQ-O8%~UWai?GV#S&_wMT7ei6-NF0-cN<|l69xmQ+>GJzH-
z!9gcD+YOKgi%e805&SYD06&QT{BAvI7H>REHs%OJF0?G~F5KD8ogwc7dnYu2R7N_-
zQNW`e<ID3TFK-#1ORs`XXqlN6;>xf-l2r9-H0}K*2YPxY;a()ce()-~RjsWiEy-ax
ze~JD@ucixxFh~Ws(MpJ)tJJ|RMQ95g@@xTg#DXh=3Lwp)*K*Ym5b(-^dE#4&13-0=
zOrRV_eqfN~C>MucrVGM-6SZNOkWuyvNjjXLPTfteSkXii0l630Yq1PDSpIlMvnzEu
z+1YT4la^tVrmYsR-$}iLT4zFPQcGo{TYb8Y;75{QpL}yxL0zpeLeOiQ?yQOVn<!Dr
z96g;vo;I3PPN7uN#i^Os2PtK&lAE)(+Bk`?ya~E?^Kfibkl;BvX`V~xIBDnD3#&(U
zY;tH&<msd)yiZ_m&RlVj9&9^g|4j3aInEr|sAeVQ@mS<I;fO_xq#-TE559<?Fj^<F
z5BK$@-|H3JJds536CsW{9}yKfGu$|W<QWPaGk0TU7#oAG&@la$K-ZyIpeP8;UYrP!
zq@#q<Kj>>j2Vh+dD-$mZ%I9Z(etr)+*`H1_wqmU_IljQ+t&0?z>#qHXnQH_|D&@}8
zd;RKR$im<ASR0~Y8&I)ZInc|i9wiXyR(DH1i~%-4rBjSBLdYYLlWmp>TR$r<(712+
z=5QG_f{<oFc{sRa5{I4w&KS@S{lP~FHbeor0^ZD*FJJcjzR%Itpi@;5(}d8AigjRV
zd1P#i>iF>$P$mtNQfME5-2(!Zp>IojI|!1u$}hL<$Wfy{e%uL+FSWrNJP<6wfDFvk
zW}Wvxwq4$Mhch7~RHW3BF9|4u&3|-t2Y4z>4|`621i-Vmon!QK=jL1nKKt2-?o1cI
z5KZ4yMY=bmf;PgV$3NYf{1WjrmqG1SK?`q$?XBD1+<G^37zbAP^?K$W42BF8&>P-V
zscVxkQcX0NU7eaZD0k2@k(D1{5WJMF@u0vTCD^Ap{}B72;6oH4yw_kSh7`-vBn+De
z$x|n~FWAZQ6$ymM9A6`%I`YwbH;$_BB4@G#2@QK0soO8W;0Rw5VqzCen#vG`lU54<
zc2n6dT$n<|4(khgm~z1F42Bl?`*(^SC@v8*_7IYB*a7#05Kvr94E^Bi$mSt<=GLz{
zaWGeVEf3@hFe)u!Uy3L1=H?cB$y{(&Ffuy&2(AfR3AvF0s13+OBX>s0qnrQp*RO+z
zXS)S7pkggRt>!>WHGP;#X%eS0kxcI_nOhK;8|N*EjdSg>%axcX5%(9on)ymW_<;ha
z_ny@KZzJjHugUsfX1Z}~?-s^Cq4W~MZy)|Rwd2s67i_!mzibi)VUIF@FRi}qMxeFu
zx*L3EY;g7GYYPilb9Z%|!ThOeZVYoddcKEn5d{MR2Oo%`5FGjj1wjz^YhpqZsVGh^
zAt53Elq|Pl%;ZMwfMN#jXh&#hTya>Yr>8?5nw_2Pjk`e1Lh|VG?AZ{)%kYQC0T+*C
zd6wyb{&6w)4j$to%>0Sr7{uh8%T5cRwqvi_q7DU~1IZ!!<sl)OfGh|%vS-hpfjGYE
z>S=9Z0h7cmVUOcEC}2^(;^-A{S~$<lT-n^rMNdyS{1LAgu)D?eKz0GkJ^$tclyX+O
zP{?!W*J!P;twJ(22<#1XSpYZTN23ODcaS~;6F^BtsJE9~6}iH(rC~4->x(6G%_6pj
z?Gzv13Nm}xr?$YV1kFucEXX_x3JO|UTBr&D_BZ2}aixz6!jxBuLX_m;c7A0ZyYsZf
z3FYnH^PsfIY73cyH*IymA>QB1Y29STeCHFym4rUN*vt0Rd(=CkL`*<)ZSix)=0Act
z8Ot$8w6`B{<t=E>HRT_k8#-E-awAf>pTC@!n}@&L#cF&lr-d=ay!qD)-2wlk-o7!f
zvt}U)as|?TmTxQLFMgy_YOn6BE~z|QNH4&Dc2(nhzQDxl;&M>;>vc=MLfMabnu*`8
zmYB4lxT(Jo(vN}yj<$yW8p<W0P=TtFw7?6=7;y01)d!c$tI{%YzGEO-NbChD=En#c
zhRW)>Kv=uWW&C~*+$QEd$4&=^wCD?HAeda91qogF`t=>--vOCp-dU8g1`ra0R4^`i
zPS)nKo48l%G?<kw580Mu^@Md7kGF|V5-hPYN@IS_$=wb@L-cn(-|b)Xay0Ab&nQMc
z0@j{pJKZbAt)pLX<TP&Egy+q2`t%p#GWJsQ#Vl#c$W#MbQi^Avi#HDbj*3r$Ie5Gw
z)@#9z;)*{}ca#rA2iyC!T8d7Rc$-4Lw%jv344t}KqbG7Um`*j}cRd1G1|@b9EXK<`
z0adBQ3)sZ{2>y=LB;E)1iJuwxH_6)U<5N=iUu|FyPj%Y6llp$0yVG$kE==#^$nbv+
zcmyLr?<B5y6W)!o{Ov@)h;qvoXYb1Tci!IUtI#!iVh7=gmaDA({bh=+D-G-)*&K;q
zcJnpy%b<T~7;wsQU0*}6N^A^C6}ArW1elVjo~O{KViqaf%16wqpr@yoU2FdG<rl`=
zr`&cW@uu2B)+Em<F#99^NAR_W+d~`fILm(yCiHwtkG7IWQtSl+U5KJjVf581Ltyk2
z!VLT&ClWeINpuetJv!`DDZ@X_kT&?HXA}J_1o`(hT(CEdY{M@0T>OWmeDCF1$rA1@
zvMLue2x}L7?0f3e;hK-%@ol|#a8Pe-{0b`{oQD{H-YKP?N`PW1e#1(K_|a$oWuJn7
z^l$y!|8(*?{$C2NON6#?<HLNz6M*>i=YP3-ef=-vvHur;;V<JEU+`EZB_%K3QG{wR
zlHX}THuY0z=>6B$u!B2y`t-HbPxM)$YqBia?DtI=L>bT91>PRRL=mG0(*KelMk-^G
zTJ?1t{c9cf|K-h8@lZo5YSfw<nVf{T5_%nEvz<G4#u1{8Se1;pjHS)^vJ?){=Bb!#
z=H@9IKlb@LiY!pI<7k0{EW=|Vv4ILdcJ$R$9R43Z?*HeNZ#;+ptn!V=@PEO^{$)Hy
zC<S|5iJ7Sc{dZ1ZLnBy)z|dgyZGG_)BpuK{Pg@Z`MQ2IR!jfr2_^V86qPix4lpGZ}
zgt!31Mna!M0k!b`=NtP`OBKCi!|zSVJFl_fAAgS8h<wS+s4hP$ehki0VK=iOY`LQT
zdX!oI@VU+iC=gw;%DO}N`j5X}X_bVG2hHc-W4kFXmVW#8ZN>)Q)Dqr^t;W)#qN4ix
zF~7&oD<P17-uBnPOeiY44iz^!2<*O9`03BDHsWu~2iz4@r~Y-tD147o?VQ8+jQ5_T
zz(q6f+PfF2-ucL_+cxaW{@S~7mU}-|0k#2k&F$|Sze(-yZ;HVbqgH51vP}r~f7k!|
zEqYhib);q#S%fyBvVFt!Y(t?>(yrU_rV-kdrdYX>QNrWozVhE^qK>+3?#~}KD@Fwi
zTvD>`-<$ITSrn?q-ECiMYo}mUo+;}}+&a(30mE5Y6W9rpGtEfx!=%=POWVI5WIeL+
zn-EZ~j<9b$lYg1I8K6{NdM#jWZ9UffF+^yYh;Ik~<@dJK{fGSU;lum(Y1MAr$@RZ~
z9}J8!U}s@6Qh%SxUrV>ls-hxoFfM2n@ZyQu{{Dsy+=k@{$>{ydTm^&?-Ip)duILe0
zyKCcWZ9)OB!oIdfT@S;JZ}Sj;UGDGAlc9!dn$m}T91zztf9>5U2>$?*<?}##w{PEW
z@_O&as~6JQxc|Qg?A%FkTcK!c4^K#d%2<o_?~B=EA_llC4nz>ral&!$)nD5@)7?<e
zm6eI5IpNQHcznefLo!#STyjSym@mg^G4)@k0J3^+$GHR*80PH3cMN=pTf5?gJRV}e
zjLT{XooESeojZuH;)8w<9`%RQgWt1>$B&OI|2g|Np2Po->CWyZfY`Us0VvL}Y1Z<a
zDg2N<cgvg;n@EmW#JZyB9rf8N%A@E5szmrsiiu&%uE7I>m<fu|W;CB~TstM{4B~xm
z*6j4OSH~op4!3!T*fd1AE8bu4g+vs3FUax$1!4k(3rb(`lL>wi5ESr#`+9n`5<Q&3
zN%J&_LW@W_UP9r5`A<S2pptjwe(Pl2Hj>cYgwNo84Pr(F+Kmv7KpCo($Eq6uSPWcU
zB6mfovLi10{1xnbPSXzkF%55P%uZH8FD&uz_eD3pFNyENs@e$OHvp~=geTpGm3a22
zxmEM9xmrR%$API!eOo?#J*wh;<m3?bXpMPXNe3#UYEB`d`lS_rP1@WQ8Eh(UggXXJ
z7g#TTBf7_p8$-~?uX+I^6rpj}8Pgzs_xGMPY|DnojgKJU-TeHkm^{>&fQQpNGJ-~d
zTv;WchG4}TfarkCZrQRW8?a7J&S4m&;R8GeFqTj#1DJg7?p|45o;rCUb{dwC0ITsd
z0GLCJa(v&up20zYT+4vpac5}6T7Z%Ptbtwk2x7?5%T9J^5fB0X=+Sq<!7czaH*eny
zux=cb?I;>Se3KZ%cjSnkgPdWU;>Rv{K%?%5bmrpu^XE^T_}tpc&Ck!z%#4!82<Q{x
zu7Mt!_IG|bmumr8!I(@48lNHjii?*+N=|6BS5@y>TN7e+0N(){R)32qZh%O`RObW_
zkIJ+$FAvY$?|MDd-vHuv?A=R%reb2;*2A}8tbxA*sk8NHDGQ4k={emJQ=gHJ4--4_
zrVf_%Hy$SP+Oe)pBwEtI&^>AF+Xfzy?%C5gcTPFq!P>;JcB+-(93!I<E*N-TfRxnx
zOSLsk-XK)c<|*P?gVBA4nK|(*Ibaw7n9ST1NnmAq1`>unbY945#oji(^$rX~>8kcC
z_#6jXF?j=-XpDOL$OQ33I!1%c!o)O+k(C%{U!&KI$JwA46FP2*smF8Qj%XetCy(P_
zYSjyg<%DzHJ}UQ%CT%mQi9KOYvxway9|-8R6%C}wv24&a4<0!1;nzKJK6XPZH5A*l
zzYMG3KT^{bw}hOT*>-fm!C=Id^VajiDk)bPtio|1N<t-1nSQ~q?YyvxN<dD)pP3nA
zxF4v40BwFWMsc1NewB?M12((5x+;;-frW(Ltg@|*a6M)u*Z~H?A~uNc!37gjQ}MO)
zG0G;(7<GOB_gJY88-zTBwi0t@aCTwKER8PVc9)>ko#pmw%g)Im76S}bJUoU^`mcFC
zoE`66#+?B?^KxqV_DvU@sd81qhQ=C*KIZM|8!_XFp@s+W)?_(Bd%)hlor03jCSGmu
zR5uQY<RipJts&-4c=iJn<6J(U4p$-cm7cn}1!O796W05%uIU*YGs#hhr1|@|Z=gM=
zd<}r!0>!f5RrhY-(tzX*(T-$kIGY$6n3gt;%mV0Uxc`y*M4&u)$%!{3Uzjhi!V3w9
z(d~}RU4&c-B$AGf4lBh#3>yMk1UMB>Q6))@yVD1$6fhpWF8P19sC98x*m!g@H0KCy
zT(Kr`#j=2sXrMTuqT&bKUbYLHEA6WbWYw4=;z*k4dkND}AURd}peqmaZnw3?0`vr1
zLcgP&;@F-6y9P6Oz-`M2|4oeW0Rhmo9n+;vv3aSfsohME9zM)*_s3C_*TeS5U`?P=
ztxk;XMz?|Gf@O3K`e_JjLvnJi<oTaW-n$>qlTWgIS#TD|dm(YWpLqHF`Hi5;lg!M_
zCr@gF?&aAXf@3<i`uZ}YSpx8OhbQZ)Q>PFb21Z6k;7-xoJB4{#2>wD4D)jdom^LPB
zUo-J)^#?5cM{|A2uQib42F`ZX3`5--I>76E7=fYEi*TGS<s1DP9WWR=bPZL4B8-v`
zreiuo3OsHT%7XAKhzHNo)tM?s(BECX>tXBf6x9(Pu4nr<HZ>!aWkpKxW|C0vT1y`i
zD)cUXj*P@RvI0sOLgM`V5k1}shB;ldO#zUiOJymw<N`NFlSUwNAHu_tcu^cW)CYKP
zWJDFF4;WT~My14s&b+=`x7!9x6=9qA4oexhac4&dklJY&I~c@sL;k-SLcqIdCIH2&
ze0s}|>AXJwb`g;hLh3ovh2_+qVKj(`Gd|vhBp6Ezq(IT8A7y2=oB=ZPIq0YW%bbLI
z2t9f@<`HdrWX<R>#WB-`n(%Lb0>Baq=*5Kvr~wg;>EUfA(vPuWVDiI(s=cw1Fg-<U
z(X$b%$ulAx0aJgP2APzN_@Cl^Oitdbpvvo+J)Yna6u1_Q6u7fCKX^0Jy(`CVi(gJk
zPOcgK#gD}rmlKX)C91Fn*8o~xh1i?4osRYxj+#!@fWA|r2^ERIqyjQ?#O)L)!~)2E
zFN_BT2ZKj<Cu6Ctzy(Z8b&F=C$^iKR2+KLrb8s|*!8a9ibuk+aP9xqoP^9TMnPFv`
znkMNun`G_%T_xv)IoEZ1>#n~`OM(PkyAZ)M{$RdQ?48$u0u>TR2L>?Im`OPev1VyT
zit-VQnMB_&@jm#i`Nt80obH1B3neQww>8?>@~hpfKU$`JB37@h65*c2e$UEk1;ls4
ztrcP`%M5!QN|nXn%HW`UXGEy<(MI>|$CCaH6AARZ(@pGvZ0<iTtj`h_+Kveyw!Ot$
zGXzaeJK?A_l>M~}#E1eFF)ghg0ANPf<&d%aFs;LFQZYd(OxXqD3DZLwLZhsZ^==&G
zLkfn7T@jhNr4YlX%U?zlyKsHgtkDzpp6}irEi3=d&nteL#Vp+3(QyJKD-di9`@y0@
zfwG2QD`n8p<k6YioO6F3mENdFS_Ztm?|0(9pohJV6CTwYy8IIdM!3)y#X1E1?SR4(
zK?eseKGfBqO~4U+0(kdvbzdw~v^pewA?hN$PY2pw3L^UV-9AEez0_E2t*zx5S<NRu
zMu0pH3!#k2qtTN7slX3mj{7KJxFscAXAzq{-6R!(8N?!kD(031Wp-3xGc#8Fw)ks&
za3kSoG=dV`YG&w0MlnJSIXSub-gros67Tk6U7w+BUd5cVev{*9OyNEW1wkSo7grYW
zc{B%exw7)^Wc{>e86f-{^ES$fFYtghL7N;+(8{VR--MpNzKh5q;9%PTM}%y3*x&}%
zT($PxEUnAd3QWE-`^XdY8Zd3s$d~bZf<c!GIPL}=9<X$24HqG6+*yUkk5O_M#b-v@
z;E_2vge6;_S%#}Yv-@InqotJ<Gz__G_Ak)<U=Zn{kV*|fDU8^Z1RL0aYXN1v{*rTj
zWe3Z?(}qx~=6!PS`Td({Fksk8j1NWxBQY67zoU&ay?f(kMv0qQF1C{=A<;qjvy|=q
z`4g77{a6*$mz|VSU2q!0Yvkl2=(E-MA~9*0a@Ad(&I=ys$MO^J%7L8`(2u!w#L?EQ
z;SI~Mn6a?;?~hjsmK`>p?WhI4qG1;@f?0%r>2+ZoM`sFGz|0ydsFqfxPDEd8X<`-h
zktM|@VGFeg$a&R%va_<vqz>Z7Jl*v#^Jd!EQx?JU38l6H$Ym<_8X7P?ScR4()a2SS
zzs-ooU~w05(z0%xTNvJU4`&^cf*e!>7*P|X6NJ``x&>bkUIA)Loz}Hu#nsg^sqoSL
zav-n%0=di|>SJ<0f_N2K{NjCKEdo$kbDcar=(vgPWnZPD&@db5VR3;2V8D3x8|n%7
zqV4J#%cZC1yj`-Tt=|MzK(CQci(J1R<5SU3*o=<6vz579U1HifqSogKoYLko0TJiI
zXE8h`Ml_9&o_+{(du18HOqK<D`FKV`v(4=kgW_op*0T20LuX{d&mnvsrlLwzxP&+i
zX7=yahnYGG&rc7e%hxo8M!*timC@+5zimAo7nfqHS~ahy@#G&)QM|m{ckXC9{HD#Y
z$WgltUIt+?P?BlmJld3BCA4vfiV%lrXazzU_9RW&Q;LiFp$cpkeWGOnh&iD0NP`Ok
z-w3qZeKy}APP>6|L(J3j`mv0jSVO#Gb)92YI>C-&W*#0I%7!w4pR2mE(&y9W7&J3I
zy*{2ZpgbHH5D+(*q-*o;RdP{<s_P3KHMH3p%8?*|7L@HcZ`v8yKAWOC&jTATPvVMX
z|7jH3e9K^O+lyDP8WMc{{0Msy{1qbNfC3V?dE2*yGj_ke?<pf}J$JR_2}MRk5SFru
z$q0B$S)nFRlq*lViz8s}1*4IPm4N}Y#LTt4Ox{d;`;jl5o$41>)Drmw1YD7g;cR6{
z*^2eEHOUh2&mID5u-dt6SDZp(K;JdP`5+9(!H0nYFVF|5hIL4h&es&41w!meIMw8G
zK^z#Wvj`>w0gkwCH?BvoNh(M6EDMWtsuNgBZJe0@{IHJ%?WGmDb)mCk9+9kO6__~C
zj>?B%ftSPl(QWs!R8UC|s3y1iWv*pcZ6ZnL*vJ^m0k5DH**5;#5vba*YbTQYc!NNu
zLPVQw&IXE<?_zI38q-M$>_sHCqr2M`#S{M2r_4Pdr7mD@ClnwO>(Vd${QYCsN#vy<
zqVi)Rj2v;r7b{Rn@glK%6u>~m<ZQ3{EHF?NAUsaNzgGpl2mB44DN_FXRml#Hj>1)<
z1Fd+1CieB&eoyONyLKHoaNw3iU0oeMy=X)P&S}ncl*okZaj1rYYSbiLAu$mPXw1{h
zJ@B3U11my_%eO#)L0=2JP<ha0hzJI=SV)n<PDj+p1o#3k6!=97tMb&@B&cMR4{fr-
zO*GrQtVhj3SOS6A2D&F=G-_{jvyjnm22~=!2N@NT0bMnk(T8~C5~$cZ!Gc1h#?6tE
zqV6v+Y3Jl`|6gt{lUjQ-$%!j?1Su~tUI;%6BDSbBMdUr=xVWl{%+b?;6(<AY_nu>H
z5h(7l20&MYO&74o8qVDclqQ<dUy6&RLoUNxHPK;*p}|)SbJbJ@1H&f}GzhF&#8i5}
z3ZnQxJps<uZg_C~A<fxSUQt1`5!~A|V`B7C!n9FSfu;dk3`h%rU!nOxhOdsiy(|EI
z_yWiIBy;CvQs074%-VR5pnBCXq2b(?sw-hG=6%G6xiEe9t2GddM54EJzX%m4ZD9ux
z+#4GKoeSO(SbjKbP`0|ROqIav*=>9VRgNdu@1b7^mdJSu%-h=t^9NwNW!WZ}LY8v3
zl0t6I1*|?o`d0T)J$b0##0JR@vM)A8BhKZzCYok$So&??wcAuN)EH#-yNl>L5<WNl
zlwvY6-B4ex@k&Wab>=%}t<5(6fCUu~PuozXiJG=H!Sr_TS|NrtL91!r4vZikM;E9K
z><$xi#UK2#+|&C1y+^2sJ#zjE{uN;|Q1$-YVvHIKob8$s)WnU?^7H2->wuu0OMe1y
zsD^Vk?{b&jFLzK^1M40#qkaI1DBZlB{LmrLWQ@ST#tYtNf~YhO=?EL!!|FP}EatQ+
zzs%r`ExGaMY9w4-)=+o_z+uC&XAlLS;I(U@x<rLn;rHUzrzfU_=1rlydo<tXfg~NS
zKpg-gPR2Eqs|OhI-gA+U0(laikUmdU-{%vPoO};;GGL$%5ScEwzJO1|)koV<bE>xM
ztVDVWv6@lp{bDuYHxJrOqIEjL&tN-=)NT5^|K<XOzN)|n{7*|u!4~<BECxFf;lGjm
zUJV`6wNs});0(l)H38)b%MdXHk}E45k9bTtUmAznv*ECT(`=+6mK`(fU`v7g3yu`X
z`*?XNCm*;V?}U;Uw7)F;Cg3}`)&!KAg@uJcL$TQm4Tr(q1+M}j!DV(>K785WFgRMR
zg|4m$Chn%#A_Pi|f)H2|DC%Rt&;d&m#8QY+W@cu<wE$%W{!F9bfEvb+_)`5X`mlye
zh&4R>TS>lrXeS+kJR!+siwDP%$HEk?^S@vDf4hJBpB^-^3QqiQ+wN7uX54qazk(lG
zKnf$^r8;OK*FpT&_r(H7cKLh)*lw~pL<ju$dhXw~&l#-PI=A;wH_hj5WfNs`oKZ*9
zpHD4jgZYNzh$0`cZ67N=y=ZDjYpef8cbbrbrFPVI)^e1}-_s*$zvZo}GenR&{<AVS
z*b#2ZXbgUFg@ER3Tjx1TxUvmT7DA72Iwrk|z6ho^w=^?h2nr-A#U`4>d<S*H%brci
z@EeV+%k^uf%@2&2`}k0AtI>SL`MUxy>daMM^$=~!8bySC-ic4)DmS5Ig6)!0yg;%u
z08_-Om+tPcmRJEXn?}H8bYz6`&><qtAc)CBwWrUYuOV+se~c&({0*Vl86gVTteBLP
zwWTG9pvJ(gATQs(`>1T8i?enRQsFARd!#jo)}3+kfjSK|o<3o7Cewur<~W&97@{;l
zI|2DAc&3Muc_HfmWlmEA>>j)PkIk!|O>S(<+xg!<!X4hfSsgWP4r@~wiKhIVz<_7t
z_RPxM)lCV6-6&3FIc9%O6JqB08AiC|S^$X*=tqr(i#*JrznT$-rWOuGc}AFu&reUY
zf2yEw#Vp2oBz{M88-XL9$KVI5!$92NE69%Gn^5)o!ifp)eIkUTD$bdwPfJj6p)yj(
zd^%?v!{?mLy(pJptT2b0MY14Y(Qjc7mOnU25PuQ$VK}*9Ky4^&cR;0G;-aRZq1nIx
zhLTduqAWy0co4K%yLRq8hNK0oMju*uP&o_+*@8(JyPhNwf*nR!p901=>_>08&~<$c
z{Sud@-|(4<Mwa6OAjF43K2`keCdX*k$t!(YBBQkMFO)rDcfNRB(YE=cR_2zWAFJ$H
zYAv^unn*0QKgA^yJRfh{bGW3=w$2BQrTu?E8jaboprz!hs<DB8WMkXa-(EpMo%my1
zU3S-)SLCOWV}Nv;1-TC`?KcJ!m12xUmjliO{OAUAM3eWMH!z#O4?loa%IeZmf^VtA
zr~UyDQH&3sS;;dS@u$Vb#aUer@i5#BD7#b{01hWI0rJH_wQ~!7%v+z5q@9<*EyxPL
zTYVH<di6oiru&>oR+f{|dcdNK_r<V@a`@Ulivqw0!x8N$m!bi{6A4(AShY=QYrv=R
zgX>tYMYurZHyS9PKYijO%~;1A;?7U@_1PwP!W74j#j03<!G9p_Mr0k@UwHKBc{1ph
zLEBIWmM)SB{KA3$e%iBVY0xaeT7n_ci(P1dq1Q@SFC;eTe%A13#G|I4{jqjcrZ4pR
zB*_)ZKj!zT=ouL5-aFW%VEvv!=U0LJ)5L->9EHilXv>91oH+pDF}Cq!wp1EG(?$&A
zVQyYSQ`==_N<>bj#;KAJ-FR;sYd~UxjO4H@oUU9kf);D=J?QZI`uZRm0$3#UDF7A-
zv>L{Ga9Oy~z{J21z-Q0Ku1K=PeusFJ!~XyEDEHz~^6sOgjOXl$L0*Ddh$FWMRDyK?
z_;mB6oM%s->;>ju!$gBo$><tDfdr8qW-D*5w3opg`xKZ5X`2Le5>$C?fJzJay5sF0
zzwC60mR5+xkC1cVMZ>R(u$8<zh8WR4!@3iU(n;uCnbECcHp+TW5W=~SWB`Xpwgeo~
z89JJ>a$=|p_;3VH$aIX$^!<K0x=p%X_8{{=z4>F5X6E|dU0cX%X4*2F1%2IORvo>W
zm#PTUQ4N@m76a*tSNH=TwZJ+pbMq+z<LmjnNsi(H?ImvRb4TjfiqU|&c$WSc5+$V6
zhg@O0X$k#5Xh<92Uj{s+X2;qUB?Iqrq2mYyCZiaygFqrttCV}f?*|{)$GBTKBb!6p
z<KaQO&LV&77J*rT?n*BKaOTy~_Mg4I`wt(^(5-NdKFGqrpn@#6)&Jc&_(xeYyhpA_
zcJzGc*F=RRHT(YZ!{0)q@lPSMbI<D|ael}3-&{u}o2DIq>#{)P*}cu#Pl#EL(fdgG
z$j5!TBrICFb^fn;0m^M?T2~N`;SvA@6jH${jE&APX>SU(K`4UeJR2JaU^8MxUUoJR
zw?w3ikb@XP8SsRMRoKB{3CjRz5GdqN8X6j~QUOY1TpP?xu}2dT=3ybQly&##P8n(G
zE<g&1I6xcy0I7w9&>TIggtGxSI`VD~m`I0SW3Nb4P*eMaAccT<918*_MQAG^2!kvQ
zJZuC346Vi8!=BG!GYS>U<qX04(>~ZzXx&(kw$y&L@kW~u>fJDKgyC1sjy?q=V+DA<
z7&nKk8+2UwR4tRTe$o<O#@PWaR3XklFe55KRmJ&OfC>vze^<;1e;+jKz=%S91{Z-+
z!`It8VpJRUi-hc$h@q=X$GB)h#zdv0<o9mA^)K<ON-vfnUVap(gapJqfKa^SRaBFt
z(g3A*z&JWi;pl+_sc?#Lb#*mS#U#NYw7}}3R{`B0=XbcZ*bP!Y5KK7^wkCwv)do(&
zbPPLuk9usSbRwwhz`<S=267m2qN(YeX>sq~GY=2bhM2cMjZi>KZrH#++sbQ50n#x1
zM!jhtZch^-T-GSH<_O#Etq-^J-ux(-f7N|^MoiPG7RHt@jlbr++OKhviB*d!D(slt
z@yw3pgTcM}+{42P^NHhQ17qVq*Cf}NB!Xv8Oc(Ww_*HlD=dHJ`=`B`onc=KCeNXGx
zgCHZ@qyg2+_DWv<v;0%qKi7<w@34G3MDCMp{OsN4*a20OQqpT}{LM)(WDaxbWr?)D
zuHwIxwftv1J0knG^3M3EqY?OvoE&YRlxDwCxNtaiI`gaZvrn_p2sk>O>#v(~^`ObN
zIl<0a|0*x-*#-wO{>~R5<G)*)m|({i+R=%2$(a$ty04rE0w}~;lz;V2gJ|1e*u8#x
zkcu4Ea}D`^9$wzU`7TGEfMHnEez=QHpB2q7pp!n88knUydlKAV&tQJ2%9o);wVDvD
z$0-%gYqev?4v3MSz_V?N$={+FXbDc!ZVcz@451~~%gRDZCi#zrgmbItmY^q8UmJ;q
zjqUQ=GOki^dVUNW%saagKx*u-eTPuIV$U^j-dYuto$VNIX@*Do?6LJabjh)9hxDWI
z4+B&?3tctMFQIAYS&7qq67ep#`g<_HqqrJnS9P@+!U#Yz+F;e`xEtYah6K^~l_>+!
z2hYaf3|dBbR;r}^K}6k}ZeW87vaw?Vd*esl0#d76B61G)_Ug#Apf3n;e1+JWl7aDc
zmA0Q?SCM8VFH_+gyB?7cT21-#d9dR+cmFspL8~Sc1+Ba(ul*mL%3V!MeYAG6r|gp9
zIjYkqD><n}vEe!TLgLP3=-9YQ^$|WJXBCOb!dS;B!@2VxPso(eiI`D0c0}~Y`OYdf
zbUSFr4_xcjeA>K}uJEOrqF}iqPjCGFIR*vS?s{+L>a~G+Gl~TL(7fq3hYxx|WMs|@
zl53Kwn-yu)-52NtEu!xV>@(p{$WNb(lZ)T=B0}JzOy}jpi&7D_(L%G`Lu29VpXQ^-
zH`@m+TP;rYTl+O1w$W7xR#Kkk09bg0Du_vbx}+bnkdY=S?hEcw$RwoSoxd6q1>gsy
zu8$^D0cfeAv%27;sByux*KkAy?Oh~NKDP9tu&N4V(|s{4!{aH**Nc@kgy4iG-3x0L
zmI6F$fd0_P%G&Db+;wQe0J=_MrLad}i(lm9<2!ls>YrLbr93N7mG4G2{&*G7oE{;N
zK{>j<@Qdu1OOzZ8sfoh|Yno`j9bkmp8$JB0apvTPm`h>JqLN42Q76y<Ls8J<WA<*B
zrS+>I2=$1I@PAX8*9gWrj+o=8;qV&)Y@2oUr$gbC5C|Uvw*#F9-~jO3of&JN2jLrP
z!hmrTLOW4e47-Q2a?gHbp8IQ}TE{?^PmF*Z|L(Cdcm{zYJh`&sq+3RG1j>Vyk$4Bp
zw+*e$#y1dW1+vyduFH#yP@{E+R>~LhQt{c<yrN=1P58A@S&v?{JMXyR7C{oYi@Mxy
z<5QX{kE-sO|H0l{M`hWr-=a1uDi#<B7=V;XNeF^s&|MNL2vQ>5pdcopNC-$M(xuV@
zA_ho<Gzcgq-66f_jo(_oz0cWW?=#LB=Z`baVti})eSq)#yw7vr*L_{{nsZ)r-CgIm
z$DXGMbb~Ca%JNHei}2?7?NNV{Z_*#7@6g(pVev&;*hYAKS_!%s-Esc%;CeGHz5sI{
z?>bg({qZMdLN|qis2=idy+Bb|VeicIvTMy@io-=4kHn|V8#KtItuowGldyN!;Q%|j
z?!>{obc0(P@_PPc&PW<?Ua$TsMJ>qw{N07eY)UhkKx7yMtV=PC2skc$C3G=-2&oKC
zUz|LBofAv-UvL88PG@Wnw#WbD%z$IkDI_mw454?<1#>)XIy@wKo;2i1_=}fEY&rI`
zvKpW+c$9~3F|$nQJiMC#9<m4v3*+~ZK9a1G&s-N}$1!v6ZvJ1mm2fzWy>{yYdM>c8
zM`jvu<TCoLR$}fTK6%jcCVg3azL<bPh>O7O6u<lcyD<5~Y~SDh5wK2}TR>J5J^upJ
z3*?YY;lCvY{Cd#{f2Azno}AXSiC(Y)(^k`TQC`LUauq@40x&lfp5P0-tQ;IK{+i!R
zwCN(!;4N!35A*TO<W9F%a~wN*)EOt_#7w?+FNDrdpFVw`XQ8f$<OyZaljqMt3DT04
z9l>S+%t@?AhjXX3P-G#`PUPXsh{L8jn?ka3A3va95i*qiH>z_W7{^!bMx2xbKYwmz
zU|0e}3^iXvTU#4i+E|0c3G=W_ouiZiC;UDD$FeO6067ECDzM>dS$*;~y;OM_tCzWG
zal5VKDb{_5ZOAgr4bqA{dYXI!`pj#nX5l@0MLXZkj<^uHU2jkvku6yM*9F^SW4Y<v
zn}RFi&RG@L&%{lubekw#mG8fFDA*I9y+Pnl?!|{0+=q6X?N?})9+b<tKg3Ucnmvj3
zP5EcB+48RKeBbu@iyfhD%(*)r9;eo&7ktln%zFH2Rvfo)oZ}CKZ+68P^0AOFqgW`0
zBnCd{fZ9>^D&!hyhgH-QtApBFZXO;Ue*X7xT*7=uJbip@VsDFG?6>VVQVO(pO)2T5
zp&7yr2MNcQ$>;OF^Thfu7^?^~$xa5AR#lndk)tNQfjS$+PT=CXIXTcpqv<gQm1i4_
zCZT2|&6LZWqjv>PokA$!g~@=BkTdj%Xdj>-V{T<NIW-0FdK3UP##J=nFNt+QF|}QI
zy_91e9FgPZMT9VC6%t}}zxzSoA(e%O2Bs7f{=|+Q<Jd!NL%Z|v;DN(h*%Vn~(5Ed9
zOZAsk7KVTaG_4{<10|7LO<)o4u5Ut-h#K4E%RN}4p*YD7^#jKyQ6>#=IMiJ6)Hv*A
zmkfk_@!tIAuU8I_ecHKe7yO-x`)#ub>_o&;h=HMxpPQ>I7iIPL?=?|oZ_fpHGue&1
z!Z;OFv*uKT6!ZW<#=z!g6c%3k_4z(Q;vs2-f8`uffq3bM`wf5x#F<{lGw|Ezro&QJ
zKTPOpJx&Z$7bnxhiMw-U8FsR7bW&z!gJ@fw<U9_;{kAMl|5{nM+THHI+ef@Z2E>!?
zb5l?27XNTejLUeGV9<GmneHC1-|pdrSZz1w4@2F7HQy^&Kh3F@RZwb+<2<OhyFM~Q
zJx}Rcv5bP$QjkQ1|K5xX66{Jdr;u5rq=FdGPzYmXS-ML1gXJ+hpO68G5h0t5A&+pG
zYSe#}*JYkHW||HA4ZE)lfQ~_@^GP>EvjHmtLb%G7pofRi2xVbrjzaR?)HFRcMKFJJ
z`Rz`w!U~~jsOMSMaiil%?4_HA;u$#gJj}O@p>F~tIr@uMU_1T{B1t}KK%~sEQY*NC
z{UjuCoNc}&EsZI>S$f3E1`*E!mb2)0fY`&Ko_f)|&mV@eyHmF|AUMGP_DOKC17zQ%
zOPnGK7D&_9bo7}>wHm|?dF&fx<`7f@VF0#sL@P&iTqpz$M<9s3ZKt1t-ExnLA&IYC
z^k@J&?1pvI<Ig_SYFy6h`$;E(q{rv%-8vX`%-M7R24Ss}tycbp{&3uCMtC%X0YX7%
zA=sbn>@@i`4`)7>`BnQ8o};KmXKWJH3pOdmV6x|!>0@Ul&b9E8b?ZDj02Q9f@}Wh4
zHUF&IvYYHwZ`k1_$u@ypDa+tl_P)H-?nqwi4`wXAPYl=Ey_}c*{?hJ#cW%yTS6%#N
z-r1b0@sd!}+3VGzZPOaJZ1Y<@vV%;v7`?LcQrux8l$SMU(?YG;VtMnK-dbLLUK8Gh
z4X2*CzH(KHR<Y7(e%t!UK2aqj)ar_I6X(N~o6iPs;=Z=YK&;}?4s~%!eY1tp$qbvk
z7IUSN_a*)39e2E)X!xG9YSP_9z}<3VGX2dS`n4=L;D8SFsAh}vQCn(hwL*<%#jAue
z3}!tB4wk6Wi=XQubw1CbVREGSXAycF6LqDwpVCGrCKMOhF1Ozmi4rs-xi2i}+lldO
zf9Mrk@Zlc4r3DoMv74iJzohwiw^Z78xLTX!o#+GfcdNi-X+{jVk682k5I-n(qT|I>
zn>KlPd*i4zhQuFR6G=02jx3BQEOY-whBYj$;eswrHX*5}yIUDuaWZyCU*;ioW$jRa
z8Wvdh$tkQynA=W(<&W(SF(taeqxlpJgs3%s;dq(SN8<}3BKfGGY6`yIpk<EWDUbPL
zUTe)!71e0a+7up*lo`7}e#<LGkgl9zhpaSwG!T;oy}0g*Fdi1FcZ>`WiaR9fdD%;L
z(kQq)S2+FlPZ5SSRMxu$pR*G=rmElbpD#<7ZLH^fZ}GnTJ$>4Same<`%o!~f8RsC!
zk=guu9j&|oGa2S**SoJL31+>k?Nd%J%vtQv7GAU)u7F(g#r%s;?Vn0>N=v+ty4Xx*
ze-NHh(tN0NAR$hRGi{F{a@UkCrSF66m8xnRDfAhneJ&p4-Dvn)`QX5rsD+azjc>1K
zd^hdd)g7%HZ7h^$*W>>(C_ar}fS<}x@Zypl_4)gr);7vZ2RhaJ(wv2-nk}x&^rh5b
zQDyJ3Q3=yJEu@#zEt+JSW?xcydZTRIm>Gbes?nIf;%TN2cDTWkg$H{u*QNLlJ%HBt
z(Zgx1e(Xz2J`BJfavW3+uD!XHGxQdnudp8mE+T+lz95D8fii*B#mosgf~5^BN^iu`
zM|s@BJ|N%4?Atm4Rzd&GxN|`=WSjxlAw@UIP(d|2CKe9(n@`qJJK+?#kV4vXalCI>
zuqfrs=}w=Heo1m2VtS*@VmAoRhs<Q2&*!V?EF{l3kw$)TzLYp|U-<U$3IYxcndfo(
zgbKR|9XTQecnPk?K!-uKd6p37?;rJED16yE%s(M^Q0(RVfdIDJ%k*Sr1di>n$Ibw1
z%RQX0hN7>}<_|iPNWf@wcX9h@KRn`;;&J3#>HM(IctNC<Vg_1($Iph9dM+=GlLD~W
zLn?s5FJ<f8?w_PxJ_0IJ=JkgMoG)-*ew{iq-%Wj3S+J}&%fBim^^Z$??Z$;43$8g#
z1=oVxXFK|}!<3gS8=S=2kHo)H5lh>Cf~a@3QiSUnw6&)T`OAvW|46Jm@Y+r9S88rG
zS6F_>(mrAPvrbmAyW(?ixTSpai*tODeqz-|SoBW6=NNtV4oKa5_Ph6&rdMr$v+hZj
zB?_L=LQo(f(cZdahfbx*k|B2Zr16C;o-js*Q10uxpb7qfPJX*X<feUArQSv~acx(q
z886)0{RMMXF>KS&LG8_zJBvrqEtatgRc9)yRJ)uN%GrVRGro9FnjhC{o5m$AJyLPM
z_yCc}xyj&AwpydJFW<b8Z5|HmCNmk4UBbP_7_SDP^AYP^qz$OIfR+Oz^1`~i5xO!^
z5=?}3wM&NC$5S^~sFbdYyfn#vpQ(vUpy*jdCD2V`xCC>1;8O<<yhPp*WR;|ymt@Th
zM_NoBmh5r?-wg>z%pj%PHpYrLu{_&`^|d8uns@QEC|3A*dw&_bs(fK2oXV9VkZv7S
z)G|<#@~-jcOXI&=I`wyPC`BMsvd2JYPt7~^l0Dlv>n1K++?vS0o3VZS;ls@6lg!sg
zgnseR)6>3gb)nts)3r&z56{Blk|hs5D@wf{!r`?4ZGa9_iH&A!q}m_R5z2Kfb}IQP
zq8D$^9{idS_Cwn7Ig8rTKyHd5lk^tpW6IpNi`OKoXC1;%$H|J+@7JxhIZ;MiRV#=j
z`aOG0dA`}wqSI;Ej-q0ye<eow^<*dVBNS=)BFGq+x4_5&FKnJ|I9|UY4@fa;eh6=;
zk-!T&&z}O6*sPyERQbbKe-PQG+`OEuY<QDJmtbs=3yZ3Ksc-iiukF#rb%pQ{0UVaZ
z&uwLCIRV<r#EY6sW4|*&^)t+7WL({dQ+gK-P2c*5JAqOrs`U&W89foP_q)N-U?!L=
z`}k}MdZARBu#ZXF=rmj}K*O=<m7i3*Nc;f69Na#(6eiiUZiON3j_5><wB^&fX)oa*
zqGh;h)vBFKn>TMp^FSY;iA(=}Ba#mDUnu}dkuJy5e}nzBxTEQb2b~QswhQ}--@$m`
zuC{`f;PCf;V}WU`$ny8wK55h+xpIx1*2md{H`5niSw6TggZBU8v`bc1$?)Agg~n*}
zhupQ#7nBo(hcEq54m%R?=T7z4OYLXXKHg!Bx)@z>^L5LUUB#Lqi@$Zgeg68nu;s$^
z5r;U}TsQ8b$5~_l;sQ{{&nd~+^Ioc;0p7~pW;m-t6QS6`5N)z*l@!y_^9t?yxBEDF
z&R2-+4~0nwsILN+e})r7fm0!?2|-gHSn0z2{Hs^5NH80WX5)?u8^Pv4Ezf1#K;j_M
zJkl@29Qn6R_qvC3<0S3s6c2TXQj0CU+$Z*`F3QZtu198^v43Wu2W;Y0(86kcK&C-z
z)|+LAQLlP0y^>XN*w=Fa;+GGR5f3I$lE(i}?k5h5TnYLprJT_MTA!{s8&weTLE=18
zy|^xNurvOH*Yas;<cR*9AuBC1P}H;rb&8t$gUIL~K3v+&3>A#P%^#;YCnuIJNn0jr
ze15xA!RhhwoBLk%X-dDh;I9)M370W~*~i=_ddX$m(NAhjwk&STmasQ99UP@>+{H9$
zoPG@I95PoW#Y=u^p#NOpcHxe$U4A)|k&-ocZb%D9%PMLx2HJ4mplet`#W}x~R<^v`
zx$FqiVotVEo8xnuTJJIa4{br*EF?7Yl=Z;M(h(I@w4JySgUns2#)(KDFgAGqyHMkz
z&oUarS`j29F|@ZSO6^8#l4;WZ4?v%b9~$pWH*PeaN|+7FRX%Lz;`~t9BiMZ?`PM=A
zyLT_1&oFG;$5>zfOye9#lSDS1QrXs~ZS^F;y@@Ls7|TU}39pCGILN5`v0Sl1>Rhxw
zC|HA56Ax0fM184!6>^N^IljgeyjpExe7Da&==`QyOhbY#gzWs_&!79V_@isL!RE22
zNZO6U;o)Dalu~;noy)c<*6f@z_3>7>I`YqIRdWA|y^^(dJ743cgOl45f{X{x_<o@f
z3rz}jomM?KX7*itpLhiSRKBlR0wtw{^9D6h^bBn<S^|YT>`)w$Y6EOnT$q^#GLQTA
zJvTJtA=<7av!37Z78?#eOL7_-KM-+1qX`NKh@QWsW*)cg5)#4UJBpx}9OJng>9V#~
zLyDD~TWMN_a={7pRNT@L<Wz_}lZ~=thmGj+*d#)kC@3gOJWp|!<Z@b1m#LjR`*M)_
zHGlWT3(EjNYMB@lU*2=wm+d?>oHh~yMZ=6f{ZW1=G4Q204>Xs}nwk^v&M2mJB&<pP
z*;-luxxZD{WWHO~@@i6EZdc`RFE$>zBDYrTeYa>xrJ<|duinmT9UBjmw{c&`OTwts
z_P+YF>)CFdw?aX8-mD%yc9^c=6#psl){OyA^*=Ug7*vRV;Cz2>os`i2^RYp~<sPKa
zbtuj2yMMpip0{}Jm&E)p#jwjmM+#e7PNQ;wQ3dlHtW^Z}$Hf5%zP`(Bwa7LpP7O2o
z^zNFb=2he<%XQ>PrRPrLC^}+x$w^#3mBU72$1cU=u|LG6%r2F1%Qdc9y;AM>bm4r^
zFxPbX@zNhzoFC^V>u+`c;M5;lG0hp~Kyv_45i_$M7<+xe&1`<e_6NXoK+ape5Ac*T
zDzK7)pDYK5WSf*khyKL#@mQt!sU`i_nCW?4PcJA>-&Ny_5fua&UkeNSjp1^Nr#%hJ
zDF@$tB=MebIW^yV5h=91ypJviyWMyWNQC}yIc@xmU4+4$rJ~6<<j}44c@Zs})|ba4
zXb5T&Fb%6djII)SCsnjkJUQo8c5y6IlP}Rjfx~La^KHk(Kpv|aCbeC?Vk6dgDJguS
z{qZrjx|K$;IJS`S*9vGxMMVu^zQ~ABFh8cWaeKpgg*E94rHU583=e77eCI&2y^nAG
z8mdPr%A`TPArcm|{sDqo+mpGc9KH`H4>>F4#EM?$tsdT{9^_;CTF&phPd0l_=crG4
zyP4~L-EO_8)5k=Gq|JYOh7B(X-mJ(xX!X7BxR7)c@b*nAr{HYc{q|4L%dDsJJ@&bX
z9%92GVn8)4hg%m=T40kX1|W&qX-_X1)kt5`*of|%-#StD4w<KXVd^-cYN)VZQZv{x
z-lHA%i1I6OIfYn>|7z-qj5*whKG=O7g%#U<-38KZqXaf$sJKf?K2e=})bb(VGW3dR
z?2SQj1F-NGXjO^jp!Pb9_9=~?0azsA-=9+m($;8PADUD&$7sVsLXPkyavVP{8~tUw
zN%2jzyV^gYPe-f9!k1ED{w2&7BnB=CqX1(<z!TiEc{8{3lX?(6xgwGV1_$vXN0W4x
z@9>?U`8=vXWsjI|T~FVs(~-aWsT-4T_s@Jb$6|>extSv)uBYyI4fQRGTLm|#DV#0h
zJ^AL>nclH=+q>R+>nMqCJHO+w9%JjJs+J?sQl30YB}#q?$z}QFR~4(}OjDokK77Nf
zv*R^}1b_ZGy~a;`5+F|u16X-lISbG9Ms5L*0=@TxHHemn4!*WoTUV)NuN*kT3P;OI
z!8$F$%UVGq^$0M$IoJAeY<e13zmQ2JdERI|fJNRhYHHbNX}ZxDQNRL=Yn_3)7I@Ja
zeY&$SCoyn87fgfEVM#dA>0WF}))wBORd0)CawRKhBLKHfzIh{o@kr>^K}#rvZd%b=
zDOk3AiNC`I%&}9N9;S&n>d@%s+^R|lD_E}WV}RaCdoKG6bSd*mAQ58=L@Q)sg4pQK
zr+6)bLq%8Op*Ziw^g+g>^H7CZ7Ho}Gj04{P$Vh!wMee@dHZD)J`F8WYf1OFX2+UoU
z=${gq$hXzR;$~0R3Egi`fX+XF&8Z(8fINTZ_i@0Ac6k>>9wLq2yWjT-s=9a5M3&_Q
z_Zr{)d1E*;q2i><;pP@qJf!a*^yXG>wp^Oh9<wzNjo;8-9KFP^Su-UA##C=ZQNDrI
zQtdirclHidJ=>076oqseKaAc+2Y(6Lpj_DYCKb8e(y+@?_4EKf9d~^mlrp2t%1!a~
zq%PAIK(@j@aW8nPZos|uc!2v<s$of4PUQfF@AKy@eEwQQbYIYRi|4mOJfNC0XKv4$
zDJs?gSMzj~Tl>ItZND*GPNDkku*t4|&TZuXCkCY?>E)s`0>G1(cNz+BFb(P89R8<#
zM&elR>phZJ4}cG}(^LS113P~I&cGxX{07Xhe*gZBxwe{s#et#%NE#m<CCqP+UTK}q
z?(Tc;?f@k~Khi<i1pNV{e2FAfQWDLX5x@c{I2iBKU%c7Ejl96@))W-}UbMVJNUqU9
z#5xPXn?iicD5SiP1x;K#)GU&McTS+#K!Xmb_uoz<PHnUYVb29i(DwFM3t!*irv(WM
zXgo3wp*wRo%Joj*b2P_mSs^OAT-AE(trCRHm6ft(-Y8qq=5u*@As5P%AW9M7r%ZW!
zH>8_&p|=KG?X@(UiB;MfSBjYi+Y#BggA{juzGM`)Wqx|uw7o^Q>^8Tv(!559MoG5s
z@KCc;#q-mP9~O&kDr`6AJ@s2qF7G=PuN<lFHRTn0{q*YKHADN-QWBosRD7cHBxCN;
z+l(?Up0GBK)|yvklqO!e&c<Wx&lckTWbtM5HSpQ`PR9IJGsOaV>2$Ut$4t*ItZPzm
zbE5cFO&FZkV=Z70LLg}xGKLTaUXH{<Lt`Vx!L`vaG$&pL!HIWLSs5f$wC8D#U3`oC
ziN<Sk<;@%3XN7RT31w{JsO7|Pn=mxN2xi(l{hmEbx%L%fMBB5c&8tdZqlKyp!UTlC
zndLa8n36BA!@-6W45SfNx;7{#_#nxJ_-<C~Vx;^6c-+K%7WrOe%BDg*MQ)3CFr#{p
z9Ohu|+P6S=hnlC8zj6cay};RALqH|8)01H~bMZp%$R|oXZ4KDqNzdRwWisXKQw7Ro
zd(<rl0%jp#Q|N8TMhjw82Kps5P6j(-6ywmXeV`-cgiISGr)!mj^sqD@wv+&GrD#X|
z$Gtq3EqhL$GatI=-sG#Ps;Wx*%{*Ik%)7)9Q8qLDf!wISS-)&MDkfr&6!HljB$m3B
z@xf4BqwoUdd5y@ovNAe8bK-A~i(t*#Dg&+oFte$9mc)!O1NNpGFxFe+dS8C4AndAf
zQkthU!X`{aYYkO#_EXD4-0Vs@A|gw(3v=}WyMG&{>~uba6;a(R9oDLRO>>)xfQ}9G
z^`4f}3t@NlH(brTo7<u_tv;aH>Tq+Hjt$K_ZV&HgkY$Gh6#(fW7j*)_aS0h27u*LB
z99o^{Ou%L+yH^dk4!?Hi&j>UE5JqV&oWf24<pKaSn$gARHAzbgp~5F~YQ6Ra!XOwz
z)X%45wevpcO(%f@F@U`1#EGiTtiw1X=rNiB)HFo&Y@oo{?qojhv17vkVu<$|LK8d-
z_fS(uCjEewTFDWw<CZ<LWeekI*=8bML&xWd$<}yNadCcZcR&tEnV$Lk>&<)0atd=J
zohq~c1*SG4JY0ynfd7L;U7NSGC;+hmH#Zr*2K6sUcf`BbVWP(jZM3!4ReJ?(-<0##
z!qRzEsc|1pU0%V_3*Yf@zK(D6aLfXp7<}uBoD)=x8#zQ?34-K=G_C*gMO#dMgLs`q
z-VKQ~?|e%I4vBcQz|h?h3(5R7JiIho$Z$DE!Utub<)T3j)+2g=%I~O@F$FYfCjA+-
z<H(I5A3dh_jw}Z75^?o#a01Cw_tq`U7MtWQI_2OnhbRTtMN#dP^jnA`IM~Jinr6al
z%=;5L;XT&V`>KSqHJD$0e1coyq=IwK;oq_w>>7X4KAaP{X;(V-ssECijeW~Y+U<{}
zX|(b9({Z6h+>3X|{Gdvkt1!7uy|c2GbWv<ae&Q3Z6|u#XsZOapJRVGf(J;5#v>C-B
zVC7hw0I+`uUq{-VV7M{ywwM!xWg+G8<&G*?J!!lt&q6XXGVCSq&K6?J!%N>R4YW}w
zlqTnsR_QE8^I+Kaj~^OkqGAZ*j^ky#o^}S0<Iv|pZ^5cAn?z7x)?M}+AO?d3aHo~)
zZJ4<uV?e6&;re5_rLMIixI=$>BD{SAAMRi83T7fAv6G`C8WcM70Sq8RA{T`CtnHN(
zry4dXJ}*-iV|oy@Il8)?5VvqshSAnx4fXWg2zk19V=8MlPDOmcdM$^<%3Ri3bWTYN
zOVvKceM`tMHrM@_qvV%9sSV5m1gYadsll@o)dVtLMv%lm5f%_5L6dPu_eD@`^71eR
z(bx_D#Cmo)Asq5gw&lAi8P8kkMDG&FmH%8xL!p+XrZ6o-*CIK6&`4pY!y$HO^%L=*
zIVc>?UcPepI-}GdYf-(MQ3l6eZdV-MXLtC(p<k2!$;}(A-5Kzi_jgivn&jmyxadzN
zNnI72-y;zZyahomd^pJqgXIv=i6m$>tph~xC%`Lc<Bx#G1Qy-rTQ$`(Z(Rj?>KV%#
z04*%+e7oIuS!SH~SHW$C`_CPTrq`FA;e%e8zuDgHV(7Xb4eBAhA_~fr^Xl+6g^*In
zdf+)8^HX@6{$X##A>ZZn5Uw2z3=bD<Hd#O@1imG{`;~q85yn0*VEo++h+(Eqp7rk@
z7{~$p7A-CvYh9RCkoO1k5Gr~UG}k*Nb}xFKlsvuW^S7Sr=4Js##!fJ%+?|8rWTAo7
z0|x=b^P3d$77JJbc*>@d6hJk2ETkg}@JP`RQS50mMSce%3DMXOa;NWn^7N@$=eKQx
z+p(b_wr6Xkp#9a;vzvy-8pn<3vb8C(yv8bWXYN=-oB4MtgLC3`ntx)j>waG^5`<~s
zqv!K1jI%fU(tT6<I+b6^(Jr2?+_UWwkHx2o*F4I7YMSgU;5_^%JJx1W=@0i%{ZoPS
zt<qU0<?)@}zmG0n9gWY{K0O^iA||JhBG0ScFu*XhBsXz^HIUW4PKCasV*+D97JH+^
z0QA9biJq30cGs@oJw1OuQ63`#8>rzR0cJ}j#wnU$uwWn?8QoS}lAs(T`x06$#0`>#
z+RE*A($LgI+$IPJ9vA2by@+P;TZ<~=heOYp=@OZEiVi$jW@NN7T)*Kh0x{Sd7zu+3
za&k+~$$=1?)(yypC@CL%dEvmd#z@uZXnh1$M9>pJiHNspH1m2<oV{dJIAnI{N8&Z(
zWNHvg(INlY(NPWT879Y@aB>D-Q8$5N5L8Rha@&NH5pW6$3JjIMp9qVO@4wIBeCm4f
z&K+JvvxR&P4A?^GXaH=^U?G9uhl>X_^p=ecrl32cQo~X>tYi8(4*h5}5b{8E!0skF
zfOcM}kIxwFQ-CFd0tSXhI+8eiTwo(Lt-i97Iaz-D*DuXfxJV9>xh#kqq>GGGDddz{
zJJH0MwmOdd6aiTTb~lI`XntbFP8PE#lVQPl_kuPbt6R5V88B;<oNJBwiul}UrGqI9
z6I}qBcoJ#o#S#P6Q@6tP&IkLDA|eg>D;<a;#8DNlmY{56EQEbfC_G&t*Fzx*Q*R;s
zpd&B7<OZT4u$}6oy}*Mzm0w1lLyAQteJEGakb(aKdLa8@ObZ$|s*xnbPq%fOaVS>f
z;~#W7n=(s^jg@Pt&cCx9xgy=Uf2-i~63%yo-RK0z??#UYd$cv@4bMH>hDLUreA1(k
zD*?}++RhaXqFPvN2rD!nvq)&U5WzBaUV6#ueTBc!LjUyeujYuiHaZpWLbpt9xwPj}
z0E4zy=C*B4?CMP4QZzmX=9Ryg>6g^3c(187@cotjV}8GWMK*<4R|)qYu0LuLYAzUf
zJ3CA?A2Z9FcU=1Y!!hD~p8DgEJ69W!f3K@c0a3pT&w`_xy?_1Y&Cp0M#XG+U3Q~t%
zLH`s*0n6#ICjBrc9;ewn{{lzYz7nHRqseGXbu@O$&C7{^FCl2nA-|^T2C%|HCTn!A
zT?0y(aEliYWU^df_yHN;OuuVa^euhy+utqruw4?bhz5s*QhZQJ=6vMLvf(J0|3xd)
zZ%VbqeTeJf(D|r$I&W#dYW4!ZI|3j-#>9R&`#gE$PAZ%%-vigAg1MNQ%0(`N-+4ZB
z${S)Xq;jds`j3*o+wDE3`Tq3|nYp<+b_SW%EK2L?jYWEM`ikOfmVG#NVb->j3W1ry
zapaEmrN129gxWB976E}dBr$bI)i5L#$11!&uf^4b7NGam>x*Tk8hLnyI9eU?>}e{e
z_xN=$yO$bwcHF+!fqq=Q+RP~~E-sK|;azG1`<_UrbSSo8b8c8)z!DNY3T6QqX_ZI!
zK*R>&V`44Cib#rlH!HvL_|P2jK|V&?72<+pW@p~QsVO|Ay96mT0sZkAQ>(P1zJS#t
zD5z1C+bZsXPk2^T6s0zoFbd|WfBB8l6TeaZWcx9;v}MoG(FSd57HPzwYgedfRDG>a
zk#W{wyQshA{HoF`=t<052n!9yNGy@iuf$ktO?E07RvgG*E!$39)glOYp>9a@(no_e
zD}3oSCQFImsQd}y13|uIx6EWNjYG;1w<w3zVq`;c{+Gh)6{_0;Uba7wf$6U*-LeN7
ze)S9=(L64Hb^Y5t)c4;4FuW2mOT*WHe(`^Ed%Tz#$N&&yOuH#a!VQyv2Iwh=%*Bf%
z<KqgMoREr5f!uM)U%#c){!ROxPa!LfpRuVZ)tr^AD8z^>d_8x>y?#Rn>MK(9>@)|s
zC_M6X9dLF1vdU<B6-yHK+47@U!SZiw*zo@{g)=8=_vNqRf1AIu6iP>a&pzCb^9M7R
zd<XF8S8PJ69T*lCW?cj>AIdI+#AP2k&;o0#s}<T(yw-VfC`6SKBKe)8c*T$Bz*FaE
z!eO&tZ#8`B&i~)IjQ`8m`=4Jl`}+)U|4$$1e{RqJliOpKY7JM26(h$8LSJ59H*Jg*
zPPz#39CWQK;Ue1FZBVroRYQ}x;(3{Lq`X=-@#*^`3}FZefB_nbm~;-EMjnha%(>U&
zN1V$c_hUNLSK<YA@P)hhKe&-1d9<>z35ZXFOs=Gs_>Ss|!1Hg#$8_+|U>d3DANAtJ
z3-B8cNGRZkPyRZ21|O;qPe+o66kp9h3g6$q>#r#%;m!j_?9MneEN@AEf9-N`4(|B*
z^EiAqIB8igrw;(4yxvPLTld`ZThRmxUndh;kSRTjL=;-_1Y^<g$K@KFIQS>1dX>Qc
zU`qO;w(Unti^F6ela4-q?_Y0j2rdINg8<S2$ddL5AQG&|`s=YN2cxDNtvOi+g(uZ;
z0DaW^Yy7Dh6-inp0DBo7mS2c+`BS7DK{YrvCAOsjcd-ja(nAt{J~}QRL&o(8JH$Hw
z75!IT*vG86Ogs`{?@G;`rHmvOiHlkc`4Gn${(f|f_Fj*?wxb{7TeYPB{wLj#N&r0R
z=!Y}Jck^XK)S2Z7*}&fsosfQ`3T`bGua>_b`1kwo(TD)Dr(^i;89v?Z!N1nA{U~#h
zxXRw(M!5t&E^8%msD186oLA1AxKjEt1Jw!2=d*W~-A(r`e~w3YP$1ab+U`ywIMDTF
zM^cOByT!8ZMMwy23^ys>GB@7~_t|R;h_SbemM>u^)xGEhxb<bjgy_+-?JD!~MaT+*
z*^-E)d^YsDK5onHcGu;%AH}E>VUnOlV`i!V#w)llJFiy%eJ7rY&91b9wzrT*?p(Gj
zm0Uj7!y9gZuZ(}bc1bEjRLpXIK#|QaZSM9bP}*;SAzL?u@4oo}3H`*xK<$*pzqkOF
zq+)b{H3md?MpndK@By%XNTfbC<(~PqFR^9iZYD6kQCR(b1+lej*8&`SN4z6%Pl6cp
zxk>Z3+<#7we_Y-FcSraC`?212|9TWv3VRM<nh<Iy<iTWG3Mwy{guv9nB;_|zPG;r<
z)svvlbXjt)wukAE>_I+t;sUzj>%;ivFIeiJlTm}sCbAW|FJOi_pqdN7{OX2?gZZPC
z1tc?h{uRt^l(JxZIKay^v1J#2wsZCJe$o920D_)g2y5BM2IgNY(!Iz6zby!07ChI>
zFGbh>{XP$-2;lq1Ik~y;{V`m&4*eLl9B_^^!hCax0R;%a#Ub!|hT8IylbvBp;Wt1}
zE`x50Xy}iFTn=ah{oeP~+qRjZe1}5xasQv$*}aWHz;OPujUX~c2c#G9ZhN~AVl*7S
zP|u?P2ZV?5#%gUO)k?q=0tA7l=sUZBVTsy#yD(%oN<F^-OU$BP6Ts$2bP9<^9K;t&
zZ;<{oF-W08XwDeab=@w!ium?!5`Uk!!`CH%13`)RmM=g&u+GVZEf{b-d}hXv{Y|9F
zTJRM^y&Gji&Jvz`%lCp6_lU6s!t{jUtq4)j2!e(Bsp{=|i=D8WH;17lcI2+C$M_p>
zE-l|}$*LH+Lq%hN0D1*<(Ut+g!JOj{-6`4&wHS=L!wdK6ug%K}SzYv(9PsIw!U$<y
z*AC!5Q+@D`HCM4#{xR4A#TD2e(2{WZSj#@&BbTkQti<-JlEb$7?<u5uz1Ww)kYL;6
zXZ^o^9Y}hE!kV~9Y24J;pT=U5Naus}0jjq7iE0_7oM5`;0h)mv7_hwQ1EV^)KN7>3
z54boKFd3?N1e+Q&Ov9dXgHw9~(}RnmfxuURJ7zY6Ck#Ux#-o^>NsAy04w>`}pf%#P
zdT2RDHh=j;PQg9<_$_0=%0s{9sbe~@44*E*2Q~FSq3HbLj)Tn>2)R_>e*CKO-_y1l
zcC#22fz4v`@%GADAGG1g1QZR7t=I_#@$eIX$)Ep_V_VkaI4Q_JTC(}E0o=!FbT1x5
z$sbXIH2!6dXJ;ImH)5yBYUgnuC3#Jh#jRVrp%jNsXL2xvZM`eBc)`KJ8p$wpV^vX@
zeg1?=F;fJO=Gxy~kOk8uC}wfnbc^)aF%+}INC`PZK^-Fo8D8{<uywakCFX9v%c&f7
z!|O+GTs}0<f`x`PgLXN}|5p4DE2=ks1~?mTv5dwdprSo<9i5$wIz~Rg!x&2dU0wNe
z#)*1@<UF$c%e#%`g?J>n+vr`v*x)#tUPb(nTp&ZC4gfp4pVmw-SYAPasRSw*o&VEc
z4gz0mxFuU<EEahVM!!jEI8P60n*D-OZQxN7W}KePV|a=loP<9N6<R;aq4+=*^~|vk
zY1m0iSQ&@E-VYNZtWJz1vDVXz>QsQLV*9T1Vqy;|0ps_;Gtr`=bEH~E_!4=E57_Oq
zUZ66VTU#HMU%nSR$i29I5ZGN6vLMDV`2FeNw6ffYeBoe{fvM?ItHbb~nMVLH$Rxv`
zZd8wIWU65oEN^~65&H}vEm1AaD*|L>+VRwNKWv*1LDbTH6q|Xc{@;zZZSUSVQ>9F_
zlKOlh`W3`XXAKL$QqcU~(~d#Q37{X0mxh3{VSw&6b<@Y4z%cn?9MY+PjhtgU+73{Y
z3?N0)1xG6M=!qj0&}TSY_}uT2(7>DdYO(F4K-7LWWs3Tc2`!@v#^g_g>umGwaG2w(
z;U*X(*@P4mhJeZpiaxaV2&Ra<U>_YF7UNzXV`2vfCNSzmqk1yGbo4ZDE7%EEO)!#3
z3^a*Dhznp<D43VQP2U{>+U+}y{Hq3?9giqTQl}cUVI~tYP<S+6q(>k7P|^}sQ_!C6
zEdJg*)*RsCswvfjdzSt5kV*M{T=Ap3jv8k9jwHd`!Nkn$jGmW|jMdN&wVbLxTp!x0
zCNs`OI$#GP$e)aMw}Gt$NXzDvc=zxx%TNBwd*F(?*RNj^V$u>z3J<>m5+4FE<b`h?
z>r4H@!T<>VeDdT8`rxnVgMEBX;y=+nQhlX^H~sn-{-{RTjT<2{77Lu*a<EA$S~gNX
z5mtyGN1%5FQuz4~#b6WTdH|-~=$O{=69bb4Jnc#>8Sq_b1J9!G4m8;S6~e^edU7ZT
z_|@o5z2PSPwN0CreY07Lz%%OZ?q)tpwSIjA%7;6vb8~ZnXtHM+iAYL%oMMXp1;bRB
zCn2cdTxx1=Ziajcj6Y(B2tS6{SX@CaiU~lQfUm4zh_u?RW@7S(EzVLSSbqHChYNEr
zwi4hgbPjt52Q}eg0^}7o4KUgT3>pnxgD^CcBqa_NoSC1oupmjPWF2TdSTFDqXW2KC
z*?xzdDEK<=@$UzO|N0N~^S^<9mNz@OZ~wo7(7ft~R+Oe;LZFci7XHpM&~Cq1SAXpS
zE^LTVrembAN#{6n!~{7ODuT{)LP-tFb8T91g3EFdL0@ShT*(~JaVVa~)^2zQLDUm(
z_N)J?5&mw&6{JQu^LFRny@u1?X!xpU3q*qA<I$3bj!JP;$ST<fIDd_n8sBGS0b5kd
zR=92cU?1^2V)0r}e?^Wwp%pt1qtSw6GTr)NM#ePKsjG>L;i|S%18NA@=gysj)eh_&
z(5GU<ygASt(eB-}?-a?-5N^_My3vdw^pe)~w{L*GQ7D|?<7@o-H4-%{6ni(pl|*|A
zM@WpSek*j==zb#yj&(tn3rm(Mgf!~@+nzX{nn+n65y2p6`;Uwk=MT<|8gys~2W@T&
zpD8YFZv9S`HB`q}LYB5#T!{LS8;DkF{X0M_dDNW-BO$%s0Jo>O(3dkBNG}pwLgaPP
zy61OsgZ!^j#SUeyk4rkT3mMFQ8ezi`wu|8_g$WrMz}D2Iuyu{dAIQ!(^9F~7<w1b)
zdF4&88A<#9cSRU&wM-*z>JuEto~Y>2h8WpkTpL%5{?uhzS=rYG=r;kc*xz^syr(;>
zwb4rrnFaNUd)PBbdaOcY>s{P0wja-csY2INHF6lgd>o?;;g!VxgZJBmZ~2eFS739c
zu-hiMdkV6^Jwlv&p+<cxP9>*Us<PH&44}W^I<~g=RweKrvg+*YL~G+lfG?&KL(eK^
z0{=nHXgc4EI1+GUUqAtjhqg(vq_kAy%@q*<1aNeuzF8>>)0@ZW2D*Suz~)zci_G7p
zqUC{d%&34j)Zd>6hBXkn@8IWwjwHkfRY%)hYjF<{eQ-=<7}XD<({eOb4~!lhH2@>x
zCEM<{yL25HTDY&Vo?e!54b>Vdmeas1ZodVq)#|3CC;F;jtub>^qlNstoGx&7?PcG<
zK&0kv$TW7V3G?uzquv6k;}Q3b17zj`MtfRYr@J?SbUlk^r4*c0$GdR^R~_YZp0|A&
z8{@-<l63$6pD7hW)xNNXJ@&WpjUifKuR(i*#^e{+W`R~$Lk*HLM7d=2ExXko7JsaJ
zL01aOG?|gVwTFkitn7X*dpp!YQEC2QVO_~zl_IdVISpP$u;`Dycpv$W6JQTq+=eix
zsm51gk2&gzBjbmeZdwcVtxIK4r3O`B8RbTK^5DVxNgWuT!|4Y!4y%0EwMYJ17pA0G
zpy)$dK=>rs76cL0OaYPN)G5^>IrB4b&yU<bc)<1z=xz-h$eFJA;-x}1TtFIW3hb<C
zGg_(RSZbWv+Hb_L7|D?1Oh>rT<A{eB<9ifqaPK*+b%2bbZblNZYP0$$6z0Tm%ghe7
zB#?aa88P_-<tk{0^H}H50~jJDc-pjtEV6Xz-f?6*NHvWDu<d<6jWQ!wsN%RV2(w+h
zJd9s3zUY!NW(74ts6-5hSOY=80RzUbheft@o8U*=HB?Pl0&J=dD+PADFYHUW32I`{
z_y|@ELQ_Jq@mi8&&h*b=Qu}6xSEiTrJdyf{(|_|REv{DcX++xlq7Ivwd2tOWq^}iB
zu=&~zvG3)>MI?Ia5`+KLLc_cR9m(M*D8oSYX~(K{=Lro9OXj@%<?C0mdGmV#y2>zB
ztKl=m%@=csH-FRayanYVls-Z;ojY{AFl3?!6P&wQ){3sWoIOV4mpB>rEuRg0kH?i_
z9DC1uVV^jCL1DftY^6_OgCeO`M9GX=6C^kEykBj2HmJl_i+g}|g<?V#C6<lNG%_Vv
z>MR}k&K8;uHNYg+ChbYx2g%8%D4|0ziZIv?3<GBn$WXZ+lQ+@}9l>!1XKj0sn9=1&
zP&%d%C5gijCi%R`#gO~cOePkM#ezbM;0p0@4*Gg1w=E!k$JmnZuZ!>yKm${%IU0Bf
zEq}1_qi=nCazKyA0Sg0k9ZuS+QIt4GG%_L5XoLM1P#aKytE*G+t{AhS@xjr~*f=3@
z@Gvv;QLIzMX0%L?;3}|I!NShGUP4nQJ>rh8@DFfOpm`exT#H#LrvwBX?CfxE0K#)Z
zGm<!_2cAxS59p(e>kMH(!TX_q=ZEof0jT)--LYWnKWw}9aNUt#SYN{PBW}_!laq%*
zhMQmtT_bV>I(wj+5CUy9altdk84|>RkYF;#a0M({=^|9yitdT@nm#bFlN-N3h4v<h
z3MJ1YP^Lol#nq-u>TS&NmawWaW;$iQRd+KPts9rQhAN$b&;r%hV}!D-x#}_M2M6CQ
zjGFQDYI2@DS>UlJBDn%VsI;Vn^Ia@_-htJMr+2{ALce_9YN^fim{FMmCDFaMZY)%;
zGA*d==U4#(A6oR;*p{U+Kqh_U1SQ|UUx=w07Ut#~PfK*axaB8q1;!khfuDmPV>s{x
z3ies`C!-40u$*Jia-O-qTC?wa<l}3HwjfBq9>-r*IW`d^YHCL5S_wPdQRl&^mUco_
z@eEvr(XlQV&airL0{*OyHE2fE4#$9F4`&=$DB-T%bFjq%!VB#lA_GDJxdI;2?CJFj
z@W{vnt>?=yF)#J;JB$JGNrI-Wf{%G_Y(O`Dd)^T0ro$2)pgzS<DHh|r#_7lougUpG
z1wYy`rG#cFTfk*8heMM)l)A7m7sp0WfPzcN@nwN%;W`S6$Ly+I#h2l&<hO^A>%xsY
z!`ZgvyA590f@2I?v;r2p{M8*DWH*U8BQm4ZsN#@31g_2^Ae-IQit6f+(6vmoM3<G6
zT!t?T#&#Gswbs^A@7>D}cFSj*aR3Z3+KqKmx!@Cc7$oTrD0SrI<k;M7Y{}3~aAnXK
z-96;7zdCvff&owmXtfTYW{g6@fK|6kIL&#%KHKHnJH|n*HcN5K^Lextt&7@Awxp-C
z|JTje2?+_H>Opon0COC7Wv@q%FwuJ)zGs_U+OGkghaB)cK@d_9X)!)%90ea!SXoh#
z9VivE5bO{A;QO2m{!BITDg?eUIyx%W7n%bw0mDqwKpS+hwT-#3u?Ev|u{Q}z7bG2n
zKMl*+RzsW;JMo+$&p?be^)8CMK3`i0hhX7N=eFMU@bG{SDCmDuML0Peot&g@z9DJV
zUheO#JC{5I6bTy?e5D$1pgUX`JA9BXX`zKU^_;YpJWBRKCnfd)W7rX;<)^ZWi23(m
zBXB(NcALkH4Hvi}!Vnh>=6=-GC?w%$iRM=bxapXh@V*>ZPYfbNgE6ftgRpb~Y{1Hm
zkkH`&CM_dF#y!BQ|D}N;;64d|j<-MO-rk0fAVQq>Lc`auOs7u$LFACcC?n8OWn}UY
zLb1olm{%1Q6&>|w{dbX49OdGgz!~8MyZ`cXX$cAVX~NIqI#J0`0}wrdgH?16WXN><
zZDO5BXDz_Thvf1o&UUb@0il8pis)w%A}uby7bC4f6MmU{vKsr&N_=CkQoPKSE8*qe
zsln2%+_LLY9N+-3kO1BmFu;vX81@Y$%lUH_dnXmk<+ByNpzU!)0JAKBSim+tvwkI;
zLFo?|K-Fyap?Ksh4$oxGZ2zDjh{C2JCquAug5b-dyA+h(MNDb3tovPOjd>YnSCJ(#
zvXjZil{C7hvew8>05#l^z6zXka9{w&-3!F*#o+rF1ZV7#=H_g~V4Q^Qd8|%&M8p`W
z%o_uUb_+Y_1k%wgEnpn>r&6BBl!KmAWCUN7WZ*%%)|Y1~%5J#<5{rKQ`dcF(pKr}K
zktf`sQp)Z(-)ely_RYdKk^Qz?*WCGe=jTeM(8+t<FV{b+cT^JD3c)zz{b9po6DR|2
z=>!1*6SC}ilGz06AJ?{MpuV`#U#xTktqu-l&2Z8|xKbm9t6T!*d^QeV_%OXI{PN{N
zn9u0l!ootTJtx0>^ohZNiw#(tdZ_X09^o|PFoxaSaM2q?TI)AZi=&dutpJH{@_2&v
z#Im28Y1u^|BSvz%R<34pBhxHOU+^hJ6KZc01&u**5y@_nj9+85Kce-0VRh{_ICv0>
z_Hf-8qn`qq>*TD$$$)a-XQIfm$;2NYsY}eKa4H;DQ)_!K`0(LFe<NAzrX^8sW(e_f
z=krvN_~_{9Y|hAm{$lVfanqk!%V}4Pr>mZ#=UILSnBrEM#dj=5Wh%Aw6J*l;4t4Nw
z+@suq2lomZ`LUZCJlLnGM?rON%e_MsN(LM0jfKr+w+XIyJu|XDl8b_>RON`>ZU%}O
zJXI4oBBJxq9$kPBt+tO5V&&(}_Q;O##KJOjuzZ4+9%2!<FL|d+T9g@~;K65@RI;$K
z(MBGLq+xT0UiStN_KC~UCXR>dh?dCmZ@A`wT>I&k%bIbaG$>0ZbJl^-3nSWoS`}fl
z_H$)>JMlSz6SJqPFlS@dVCEa3ip;#HToxTeYF)74Ck7fhcLdp^1C09o3vqo#b_)5X
zoVE27@OzqLS{zySB|sD(rv{CoYm%gqH3p6;>@Flj_?0w%fzr#b&_IvH3H6qll(29f
z{#HR@A*O90UoZvryZz%RSj6v!Z4LyW974wM@V6R@my1WEv_L*8vYzlW>K^{q8qRdF
zoN6N#6|ZC4AG{qZpBbEhlS;%hXKLUgA`hU?CTx-yNpal94YJV~IyML?Uw4M_9OkHV
zdz0}KJJe$|v(vlHpyk`-<a7C<$G$acu)?H}xUj`_GsQp2z!EmXA=@*TM1{7DVd{7Q
z(Ektm+!7+drOEu5d%<yV8$|gqaasUf11Z%M<V8p{UfI;~bLA_bEG;#aD%g?a9UR<D
zvS4S89#@fWM?++Htx#*vj^ZwDN<Kl3+Q}cDXb&Sp-ZjN6X#)s5Wc!5+7tkVoMw@{9
zh>w{jfon%6-#}CWNIT6^H~xJFhY1d7G7|!21Ndye%Yun7XGpF78WE)8UZa$iDE;ob
zyK_UYg;JZ86g<Xg4cyKQDj&Fq_<_mmy)_0HpM<O(St=6!zRpf!V+kF~FrG%cyZP7b
z+l^k>{OG9kls>NFl{&8<D)+K0$XeI5ayx2gDS_kc>?9p<SQmLuJaRAG1W-?MF#!(b
zdE&;hy-1F+J#S+BC5L;lzf4w6R4q0|Ev#&O;Io%+{kwJ86aHTN*3fPtdjdTCMUj#H
z4-t6Q&7AtSJfkbLHuSfA&r&kp{S-fcpF>@1E14M7SBS3KJ(ZC(RuCEbC`i!)Zhc^4
zk?p6=l|PPvP0u5!+Vn!^2pgNYMKwaRHun!EP~VXt#xeA3_z(%oje9Fi8{XvK5^&${
zavtet`}(wt2&CM3>kLWnhCbukYjiXoQW6W*!Id0_z;2)X5U=>=cLO((CA9V6Og7Nu
zLxT(m)htOVR{hGO(*&Y5tT2_ILnF#hV|^63m*|QCqI%@%DRt=*lIn-v-hh<DZJ#-$
zq0<w>s)91yhkJ5C?#A$Zq}2G=%Tw!)<iD8vjbH$c-&XH@4A)DWviU<)c!D;UB*)m_
zMpBeCJizRrGM(!0{7n}kBO{9{53Bk8E0S+L_tZ_+(3PUu8=|QDE|5r{9l7lTNVYTD
zH0ZWE1S4GDl)3V-0L{PUj>u<3&;Ewui3Z@^2<H(TaLAmT5Ld+Z&Lp2L*l|-=ASic+
zF>pNAz?D?2AQWa*Y8hO+u)4n^NM^1-KB#k(*FnM<f$x@;mRJhUlc!)Uzs^-myr(;n
zko@DDpIM<NZM2%=<+J;;;mpO|mh1iIUtEC4-rjj;9rvWyYHMo`9&>0)b>XcU8y{yH
zb}3Fm%Mu|5mYpzC`VCcct;Bj9=SnE^fGMF~Z6Uoi{WDwdV%r%k#SGh=ZJU^fwQP}j
znaUO*Fl&l9BkI7pN0f_>jKoS05q5C~Q@WiWx#V{~Psmu&Zmgb%c}_ZL*iLOUlKbtH
zkIW{1w!X2%+%xw{@A=fjIN6u3h!9GbJpYLh`JbM;U5WoEO>(0l5?o(n`{8pdnbDaz
zIg&6Xd~)=RB^}B$P7M)<JzhAyXsZN*w{G1Eu-|Bsr;{W~pr#-|VMZXr;}67BJY1UI
z;sdawK-48&ijre_=mT{iF~K^YO=`RcF@#VWREO~lotO|etr02T>kzJ2#i!72KnIAa
zn^yo(1~xPl0tyP-X=n&2)1hMNlXFNV;{0Z(yzM@B22c}VzRl*l=L``mcI<+Ag_cGP
z-S>wXW!u;M!a`_|Wm?CQa{PX0qOA$6>4{DaJ{eZU=&+HR36E&fdzN`cUocH{vPQMn
zz-D#V4+ZEqnU8f|qN1Xj^ah6U1CRo~Yz>|xni=s3IlvhYnQ>~{H&ic_3x)EGK07Qn
z<}&(i!ueY#$YWycIwcoGRd4<8L}V+6aB~xn+u!9&abyh@vQcp^QW?zB?B80w!Q;!u
z+;umSuj!2KT~mxs=PB%_scsP@rG$`(BpevwKX@RLslIHb$j|dT$fO7uTk)b4g0-9X
zg9m?LL#-e$ziA--z=y8*vKKN^0;kIB<wvN>&TS9jk)rLo$~d}*XDs4GDJI?EaDvt)
zD+>r1?+14juVydrn5v<5*8a)Wb=BbH>xo$Si)Oy5Os094FnM%_C-EA9AfK)#HxB5<
z8Ev?#r>Vr-it6X-ig0}O7!f>%e`c?MK(<Y*-35(tEbu08<cU!;W=Q=JJ^f>&m%=*P
z>Vr=2li(S2*%l=wRzu&Tdn!?fnu+_Ls&x$|&Z@+9|5dowAYqq(>g^`QJ|U#~a?j~t
z;HQ|;M7M3*3DDm|9vlIw2D&mL0iB%m)95&>1qs<r*!`lf?<xE`JpALw6dX?l##xtZ
zc*+jb#~)MKDog}pqCAGUr|)e?UvYSrNSHrxemPxm{pz2~FMLpHFhyJrG2P<R{QJ_)
zgAgr<in?Z2VwRCZXu!bc;v^8Qe7bxr+8HUq8|7M3wOo9G4wC6%F@66>O*mh;sktO_
z9>-=}tGQaG#+QE9K|*SWo)P>x5IDsa8j$6iyGi%6<zvrrlG5=6Xw17wp}veBZuwT2
zA4SdYciv&vjc-mBO4ne=1U|5=V0m~4ICm%{g5W+7FzLhUedv7RKCGUz3x;})%u2AJ
zd<!BLie66$F;Frf(39afS@9fMO%E6?gt*Jj#er)P_vq6r8<DTstZ*u)T{OI!0|drs
zDMmiFyFA3G^R16f;^rSp{(=(L*4F;PX3QlGf!C(Hh;zg6v5OAnPd_ZD^RsoYd^9v9
z-^rf#5bM6#dw9nEoOxTSPc#6tph6-y?^_oEV%m3)(b4ho>d*ysMVj|x{(ruSys;*V
ze#Yia|D2GiTLpI?WwbqGc0p(hPh#ao>ACg;tz+jo89!7yL>G9x&v=1^7bKJG00LnP
zffENF@Ystg!jbZ^QFtRhesvlIrTR^SC=lB5S&%d~qp=ItBBtTqs9-(rEbk4~KA`Rz
zDhzjI92QY?w;lWik9Q~D-H+pN#YncR3B;QDRATR?@-+`1Kd#=TD$y#Kj7+5Krn1D;
z8zgXIv&E^H%0p?D9D}Yo(@Y_n0!o7n0Ok%QS)k%6;$fm-$sH?Zr!kSW23l3!I3^=y
zJ}y-xideEbo4r`td~GDu{&H`(sLGLEP+dYS(JhcEH3#V+<4Lk3sz$T5@!ED0mpb>W
z9zzv#+{fg<+Yy^~Z5IsB1obT#k|fs8)Mx}p*l~Q&yK(dN&iA^e=vY)a=x>955dn_$
zd9gccnE(#Ehf+K<u>OV;2S85bou1-93cBUdb#SPibR!$Sr8qGjz1?=v(Q<79eg@_)
z<So+D+9g*M7DnU4wvVCdez6XRXm;isl==lZ{sxHh%%MlO!SHe9*Jn}hvWki&7(MtO
z)iexDS}4I*@SHh=zKBJUdC2tcty}kNDJ>wqYc~}~BH&Y9jO6S{Q#cbLdurIiv+xil
zQ*^DW*Po7~DoHQ&GzYhv0;CEmgcPm7dmJdQmtKt@1b%pluT}}pT}Rvr?DD@@NhYYv
z*L^q)!LI3tnez*oKr!3t!nr!`V@xhG&&f21D`~fFyPA2pY1+Xr@YTtypVpM=41P`N
zip-L^jMC#Xd-_moZu2*uvY8q*=o8h>>0z0i|L`iZHdf(+V016yzj<04&SL<gLDmU+
zTzAvq3jfeWAI%)52*VKCfp@gXOr>FtC@m}d<c_;0%nM;L^BtNuP=3UZ-tY+sIIOkx
zHSz*xn=w2q&<8?zOoHxwv`2LJ;cX7)GI+Je<TaTo;9QUTft>EFM6PT*e-R}VEW!1v
z)~%&{@fkSIL(06_KK76k=in0pN^5(N|KjPImuf4eB5+asZN?IBfM48rCk^HLAMl7g
z1qq`uNRA$K2waR;rWX{6k$mP`cWwZ1;-Q=JPeSlncs>1c93E|69c&HG*b?+Mv!OVZ
zQ@Uq&LG9jJDDF_}hKj7mNFOkod+V+0){xvvhwgWkTFt8FA{uHp($|Qi>2&;U*Ro51
z#sAbLKw?MR#=6$3B1ga)1;f_SCg^>(uMQKm`;M}2WMl*oB?=0(ppKv&LUJB~-^Acj
z@3^}=sh4xcG49zTEUXQ^eSj6dILg4}pIB<h9{=+^Y4J|SGhLqt7eyba9()<wl=13m
zOpFt&^ZK;3$QZ*~<s*iyRFQ)UuOENkNtI|gpvdty&A&;9QsmNruFSU&6~o?kp?2MY
z4*DlA50x^8O_wlqlxAVR;^Cr&^66F!Y4b6(8!-HejQs#gjOovH9@^>?nAZX6U>&^S
zv4IjHy9CMxtQwQA(C|UxLxl+*+=jkA5frcDZOTGVU)~*N0`duHHz+v1RaE35ZH3Bd
zan##oeFpWq;_Fr6jR;}lr|f|ekEYdYJ@)WGFR>KA2{sIn7c}drJPnaM^|i5)CY)2j
zK>%jr4W4MID7oWIH0mD&&4F8-2y7ZwHlGA`kJ3DLk!Xw&FaqA!f`z8EYZby(Rr2|h
z?3rWAo!>;5FJYI33-@oLPY@>lhcO3WdV0n#YH7ctY9;bKj>!#iZ4~P<0#JDn-4tad
z$i(Zb!e=?ZqpquK@s(t@u(tN{@YtygG0}6jN=tL|_jcn^Z=}I72Tk&w7n-j?|L$5L
z`;B@#HVdlL-P)S^3N5I$yjIs-K+qv*AELDhH8N>H@kvb`NQAi9p7-C4zn_OqvBTsS
z641R0F~H2^F1Sd}*I*WfV!SMHg#~n6_`Q-|zwSh{7C~41;zikGq}Wvov4cagWokH?
zM}2834>Uya7V%tKHo{g+oUgjTuz@|kdP<}}Z_$m6NWQyys;^c;l@m-R)p1ncpMvme
z-(%7Nk3=}2TcY^OS?sqJV!9e@*nR8s6>imd^6+J=;=%CLYv<g^aPPWH<2*N}EMjC?
z5Lx19pBX)Y&}8r1Z27Z>CU66|`<O43&!XUg&~wq|O_k`11VpXwTm2MVBqBmIoV$uZ
zQg^Bum`QD2f>=@x^#OC)=lcY=^062^-nWyQ)%MNeHxZGME9+e;78)(FX?~qOEhCLS
zf-jw0)wjaNdmDqjNko!lPRYm!!&!;dfMx(jw?w>p1vKrIklk-I`BDd^oved_yQ0-M
zg@he}2Zt`CcPJi}7O%pW;B{Qk4Mj~w#n<W|<3Uyjo(Px;3!7giRJl0$Dh5D+_Co+}
z9LC5=H&RkU`HWL8_ht+KBt@Rs?_}=!4c^|~DtRugc%JC-K)Db4AR`kKh(4%cuNKo9
z^-YUfU8s$6G2F}kG!OkB)bQ_7@1qLG>%rw0y9!e?{1Lq3=qKwR%#_m0qnoqw_v2jD
zUqgwTHpasJ+KOtvYDxqgG0MV2JvzuBUTqFK9e0$K^@RP#?M7Q-F%8iQm4Jzf-u3Hb
zm_+zGT5dzed1ux0oBhQuCd#rnFHEYOrML-Spj_mW51jVU$y46fM&BYCCct=7pW0_K
z0i>f}dt?41$G=tj=nw%FSwv2b94PL?lUF#p$W(#s>%GbU_1t>A{=3H(&g)9y)?hhu
zMks0i0E0d)o3YRkGJJrLbql}sF#D1JRWq^>ymIf&tN0{+I`Zct4BbUfVEi<l=la!B
zbQ9XUvqjw2!5*N-$sZo|I6EV)Bk*_7Nv$z^)4fo$bK_TZ85O7BPs-KJoN^aE@WS*u
z^2ztSCs<x9v+P9XSZn6NFVyqm#S4<iFIA&eKZ5Q;fUoa^bY&VjE3mw90=rLBTal^C
zKthoP-%V6Id$9!zM?OyhzEgd|y;j&mmS3v%Oy78j;tIFX(;t0tMm-qgm%yypopC7Q
zY#%rehQ{2ONYXKTW#i-D9XmIQVayZ;iBHVG(LEW6z~FVbL-BRqnDX6kB3I2x2oiC1
zeDUd#8wG?5JJ(RzV_K1~6a`ghM=?XUx0}*9&IsJ@w3gbZC0(EzpxTV^DtU<p6__At
zWt@P4Qqeua<R*=<a&QGn$6~gID*zXQO&5i};Pv+b$bnGuNft8QLP3M5SmnZPRxQjO
zD9MA^j+F=Z3x(+qAc#T*Vuj`ta?0Mu(-9l=Utr>8ww=uhlP0bzt2T+Q_V%%4mjxpn
z<DUX;+_7%$(Sk<Rzjx=)pl?M*2I!2!4x{Su!34R80^Cs&k<eHmqpvG8cb_&s$NWqa
zE-28Q1DBjtOj3%kgZ%@xDW(9t%GLwq0Z(^~WEP(+e0&!r8961+iHZ3&txVyK$@jI0
zISJk(Kp)@Pl75WkqM@0>4AA(Bw@i0{=OGlu1T-`^d(-kt7YnIUWogj=)wrW8-@;9|
zZy&Sda}51qpBFu-+sJOZ)_b3>&l&(*vnfqK^Jj^3I=U$<@>dt{aXR+wc4SSQQa?HY
zU?gJ8Yr??<WjQ8VAw1eo4e)_fp;s2<*W`-B85J_&d1KUR7gs!@gtN0V!1557_aGQ7
zkN4ukWuvtRhJ-Vc7<B7(_4R2pt~HTH;Zh+EDeP(NeK@=VLXCvlnj1hR3(F0Jj7Y6F
zW!I520uo1n5CAU`_XQ;k&On%P!JQ@pVk6qB&CAad!&>9vQG;xl>EOY3FkFjw_yq*~
z9vag6E2cp#Fhj}p{Y&-9Kx5FEVWSO~!8jl65NN71ki=<Z-!e0UMZ!xF@Aw!><k%ZF
z@By(vuZ_%a!1+!f8hw3;wp9?hP?n(`(udjJdjyPQ5Lgk(xS5%mPo5+VG#vOBB_v3-
z9}%I0>qQm1di82FqCmoZd4l)Zx?i)(n5h1lN|@`t4&b6$G)@3yU}D)AdQ2L)(^xyi
z@?TF+4~E*Houm9fQyE<nhUM|G+6$(?_)ndZ4C3kq213N+AUZ%;j$=|?JqUARGcs&Y
zM$aLUfv-kwI8V`Dt~ox%dK_WoE}>hFrVV1W)BGPG%<{Mu$V9L>ckJHXfb$w*s((qd
z#UtIcE!591Qfmx%;BnFpbu$=ZkdF<n0(Jr(Ne>Qs9CXNo;jg2GD1km3k>%rTSO+os
zfM13NTv*vj?`}f3E8lH$*`9S7lG5>aK0{^y_NT2<Z?ACZs=5eQN5fCGZ8)~B^-3$9
zTblnZ%q{c)hl1LB%o6+*`WXr;&()YGlqqI^ZF7fGiRVW1&20K?<wVHM#5Ukw;lVb*
zm+JHL&S+d0Rfp2bGul|U^}AQ15jN}Lu}NYw`V%~4-psmSY!;?J>BN>d#5}UwLdL|k
zf7vkUgUhCLGhT5VgD%+RP8@UWRyt7uA#yPh23H2a^b;ENC@r7BKup?D$E#g~`#+(Z
z)JTAAUjj0jYMgL-3@#LW-5b6fu-ZR(dZnxx99!HrI{e~{*yylj1HG&@;S;TFwfjxy
z8R!jOHGBq)Viy(9V42Db|H92hK=4^Rr}LPto6qC%V@fJ2m1BD_6&|~h1A1+0=(WEp
zl~RV;DV0L}efy*sWMi9tBC2B@#07jQEQB{@ll(&LZ89Umez3j$9nR>70S@dGctJ9`
zN>1)+Umi9tO}X3Z71!_l%x`DwcCX06Pf@c6UUO1faa<AwH}yvMNC!Ey*!ynao0DNt
zYXx*7@JnQt5sn;=7RnXWJE6e)KVHif3F6Jsd$d8Y{J@HE(O;kR`O6PnvHtf|$JdZ%
zDgVYmW99h^Sky;a=z82C)V{*1;by!oZ10M2*JU9X;h0|hw=ifw-Lj$%-@9$OtREvk
z{{C<A?|**re|>u{-0a=*Q4xH!R7*&aZeK-;c&-FJftbM~;HO^QM(-=!*<su&yf{iF
zK^U&j3vq~okf7VEh5H<Cd_oKA4)XQnzeMu<|E2=>{+lyQp6$PWu|K})9<(W#tn%Y~
zF20F*<N@FUX-r}=lLIpE5_!U!t&2e1^dDL^aV=Vwg`)-3*cCnOm@EJJ^VwWo(4NRO
z%i7?(SB@7D?GX9;&vXBWyT0N-+v5LQ!Qia|%TE6xs1a5y8X^BDT(T*!MT-D_)=D6O
z0T#_ox0%@TfGo+YGL2unaxg1W0Ds<I+O~UlCT4Xn^kaq!zKk5|D=Lz~=<4nv^1@QL
z(K!53cO8in;KFW06Uao6EOi?aCxocwn0iGMVnQS;hDmiGVO&ij(yc2r%lkONE6e8w
zGG*Ovq7^e*-p3#NFkC1T1~X9^oFG(G(_+8gw)-o2<_EfoOpp8RHb`WbgOBoqWsM&9
zNvCgy|Bb!(0IG7^x<v7)peSO(te8-;iHv|4P)U+=Fp`4;o1Bh_h!PY5iIOA-0m(rG
zOhm~D2&hPoA{l`m3-A5!-(6kR{a(H5dR_nBbpz+H*<pR_n`_Q7#~2eNVkB*qbvyA|
z0e?i;hG0L70EduQyUIlF^1(l9xm}QR=p!!^Gv$0%;kJI~_4msfX^rLUs=^|_?>5-J
z!4NC{(P5h}hy1J2l4bEHR1lNi82tG-a%Q>q$lk?_^uDVo<JEmxmOoQdMb3vWlG^J?
z;JCGXm^e?;VoWXk?IP6wyg<;O7tq2|;aEGRw*?pqYCm6DC;ZJ-fj?iZs>ZDyeFLPQ
z<$pBUe^fPzP3gBP;9&)~|6;QEN`IuBne`&?h%F(3t@L%t%fP`I|JQ^Of8>(PHNg2W
z2UF+<m%r=QKNlbr^S}{vC*sSlD&on50MGILsQaGkp(D0#&6>yI%bM>$(!AYcV4d;t
z@%5!dEu|{q=_(tS1#i2*4iC3y8Y;I_b|gU|b#=?~I)0{Z@6Q6T!vjmV^kwa~-tv1I
zor5VO7!_GoHhkpGl$HZ|DjgDe@Y^;yR8&`+V6@z3qFA7M|7Yn>WeBYTN~i!PvHm<+
z*F%36a3`QtRBRjlPc2`U`yVL<P`f3QL1cM&nzLMEvidX_76VN15d+>2?pe0NGyHvc
z1kh3dH_w~08Cv)9-9A&|_>VZ>Es7kK^5e^nf0}{(`<P$eM_9D~Q&|v69PoAlZ>y_Q
zaJ#=OoKgb`2VG&f?g70G2@QSq>J<>D?_FJJ4qE>4s9He((tm^<e|LuL;m=7Z7)YT4
zSFw0dTE;Z3mLH6aVd&_xNstk%DAN)BK*74yR&Q2PTuclnL1qBLcM*IyOcYB6yUYO-
z5hsm6<*0ztb0QfQiy+M60p`cV@&k-;fKRX%!xYGG&^o)iy1r+Gdf)jLO9qU$oTKfq
zk@uIt-xRpQC$yb5mFW>jMg%tr#=kjiAb`be6I8QacsaQ%rhtehO*h?rj`U?e)#dY|
z3NR|IcyL7~LI!d?uzKQwUf;56z!#0_Yk3vxb#+;$oO%K{wj99gJ2rcw!S+|3+ixTW
zI_r9Y?7AOa3f0erc5mD$W9bIximQ9tXZf07J3y5MNjDKYqG6C}aq%hxgsafMWy1X!
z@l|3PGNLou<eqh|(rZd;`FkY|P79rWuQv}71S-M%L-W9l|I{HIEp2BSaQ_&c1EPRR
zJ>t+#Wd(%{TsxdGVS-iw%2+_Fg24jlB!K2d0T81HfDs$fL&skb;w5B+z_se@_9EKX
zxOk-{RTBAW(g(1eDQ`Itp-3&@RYho;xIn;MSOP?<HxB&zz_5M0Hr5?HV({D5;V0Nf
zbIpIY3*_s#r2Wq-i29P(<n;Bo4;US14q%GVhfmIDBLXn2yCsAgD^QyH9G^bb1`q=9
z$=196C_@MMm1>KNzoAV24OQwyJO+{ja{i2>RvV--0N@~b!b1o=RUqiN^v}^-$NrQO
zOj@c%r3*9C5YQJO_5j9)avs2$W`>?DIht~|pW_!4#LO{Z?A}EC@5J{LjE1cX?3KaZ
z^)g^a;jhRzs`<-sZp|F($C}C<Br9R`+{>*?I)o1vSywv{Ynx3F<_J`JkSQA7caWqr
z8s4oU9(8%Mps(IKsfzmEa^y?A@pgPv$Swl)UIBbyVqz+hBBNI1yont_2_3{iDZMP|
zmxQI`0fIsp+-R8xY$eW#1mE&CB)ng315lOOtqn8;WINOM4*Yz4n8y{DlcRFc<8X%V
zr+tvH^nLu02_E2y6KM^a_S5ZFksOSeNvkpo&}?cd&wrS9&6<tv6}aq^r)9TQeqfmo
zL(eMD7%EWIo)e?E24_@jLcMRG4y3v|+zOdxzNK+(HkO@?EioUy#QBT}yg{1Cd#ZXC
zy9)D3<j^OgYrlb|hprf$!zwvL>F1|&L4qk6hZJ?qIxJBenb(=@o?`bnEX<UR{u#J+
zBY?2BJ4kRo1J&~UcXAl+xJQzRR7)_DT07>z$gRqW!gh%OobcXQ9n=$)4hkt6E#b-F
zK7th!qkF4XHV!F`xRexw|DcM#0$vQ`j#S9}PPnrP3EsGaTLav$jSh*p{F|+?)!yu1
zzLPRB-91n*_iO~8!(2x06p_uY-Et(Lt_1s_;AYuo^T~^GXv4wowT2lznI&kJ?)ky=
z3|YYWwtkT9@JQ8|kQGey3kV<$&)RI<wTFrSh_mSOxog-vILM~9_@iu#xzMLurlv)>
zosCEDySdTO&?qOXHG=O(OmY3$F`;WqGQg(uhs2VB_@D{=%XPa}oI-pZ>?SfG=rZQc
zrwkrp*ML+ewFdO#Hg=;(@9KB&--E5FG0LgF+IVjY_#MEe-iG47J?`w4+pY2mKdHFj
zpnde772phaGbbTXaU2X1cF0R5Mv589NB7S?2+5k<1F#mRhbZ*91_&EA($8w>^MTeS
zOk7*bHrkebMWw!~isR_fiz+tY4pWDos7UDz=fdYT*3ASEM!gqd-q|K?`35#dzbdjD
zisGKu3xFcdSCG+>u*tqQi_Qb_CUps_(?L@Oq|kcihAfr8ffVZH_Q}+vETmO_xN+|X
zH5C;C{0=s18Xgq5^V|o}v9A#MR)21S;S8xZ?PzfZ^Wn1=d*URI)Ss}h0GEl@^{h$<
z=(vCo`*S;C>4GS+t^zQuX>HUL5))%xY(kK^o0&1A10>`c=B;~^@~|f`tiI6JND442
zJCSQr-B`|onxq5t<;Eu_xSBcP`Lkcm5fG7aFb0)zadJwln3!U=Y>8k5EReTu+eTZ#
zK-MoWRio{OHd<6<epRZ*p?TutgCuZGVoQ_w238=R%~MKHgZ+131{~`BY9n1;QA<zt
z1RzFv)RPM%eSq*5M%K9BYR55&q%U_6_q(AT8)^cTxGwkOwIOQQ^9ZL&bED`o4-a#f
z7D0hS(#na~YK(hrK7p?x-m(SX#^ukq838UA*^T5=R8%x3TN^xw+HtwCv^pixwm7*1
zdk0G&3iHO665)<!-eGnpaO4$UUZ6=#QO`!F?QOV&syjMlkNm+;558j}ir0O$Tu~&o
zqu-TL$P{$-q;J;G)aaD8;mLQis9n+4%PhQg-b->~Q!X|LOIU&-W<;UB;p)IZk!_)G
zs^`#ik44H5X$sDp*7xfFbn`<LfnXhJO1Jam3qtYDA`xotMKChe+YWM+R?AQypX=A3
zdHllo)wT3*Q3YY2rY||jNOyyzfG&-Hq<PO24C+9~eE1rUP@?X{#Odlrl0Cg?#pt4Y
zZf=iWW7aO(kh-`Rw^-eQ?!9&#5#2Z#k?{lOz^G`+$&I7mwq^cC2*e8jn`tXLZ3j7U
z9Dh?wJQ=K6b^z)t7kXg?mbf@Qa1)l(<bT8#x*1rQU=-{5PFfUW=x&{2>V-xU(DpUf
zUSu_qlZ3KqELDj88JWli^^&TEzJmUPo@R{pByCn4(E7e0p^sv8ua9G(JouS7L;7J=
zF*kxtUZs-U+x*L4h#G`8CuCl&9eV*g1G&|<|9N3A2li6YY)>$m(*TAD3dZ;OtzvvY
zTPTPECF3wanS^WOruMh#E=-?7_@Stbk|m;8WbD_vofZbQThl|D$}m9}P@OQQYoG?x
zb499!9A4rJ8`)X^DXAJWg?R)p{-#ZNxF3k#=N}X1fka0|MFEUPYyp)23h)2o#bL!N
z#65j3+(k|ca};9^)7FfB5MU{2Dq32a;2+Q$fduKiA&3&F8<aoKVjs9SL337+)m2Ct
z^iO>)Y+i_5N$P{D95NbxQqYCnEeG?HrPXLTZ9B}yCXw+3i4rRt22bK9!axXX)ZEyE
zI4c?*WFpJt@g@2v(Ez$w1O7A;RS&GJu-%lTm6#Y~iG2u$7{eyKpTq6`{fAQ;<@O=)
zz(v*+uPFpo!Sh(njHA@EAkkh>Z}Rs(RSimaLhmMA7GG{8b<1E?@TR?c+p#JLGJY4j
zaNcp<x#NO{Lsh6?v&O;+NQPQ*S4kx6M6vPf5o$Il#435Dnzvmpwv5V<gn-Ti3|w)W
zrxV=B#k}d^PPFK}VkLATm2FI2zCnf(($b5_xTE_eRoOTp^t1j6gx1KA^E5J+C(Fed
zPY;Sz!a~O!3)mKUd_E0!8N%3d4vh^%y79<Is55iHP(WuKGgaR{So9_8030y`abt+a
zsQE56g{7OLVFn=&Tzw=XWK{CtcSSY#CoQ_7rPYB17R2KjTuS&&^yMlV84=2@gM$O`
zyv#-xPENFW9M7D2gM??YtQ<)b@u3Iz4ZV{3?CD!qB4mi^{>9bpqiRKn$7py1lc%Ak
z#-DZ6wUZ;j<4YYeS~^|_XdlZe1X{JTvoltaG?bwqh{S0EBbld9<<VLf6BDzy|BZ71
z=Ma%51_bO7Qim#P7o2b)W<;)n$^^}>L~IXhTifz(#QO=H9Jo>;kb|O$giv(L)<dif
zD~c@%rlPB9D1L+HjQ?v1YV16WSjMArSrK#vk4u+wtVXpkXbDCr2>l8wDti6G$Q^e`
zd{CoL`ez}#<Kd(tz72U8F5BdhpZ*_UtIM@M{^ZR63WWXl6UoZr@ju6?|9`tYhj?uz
zpsm1$I%SER#5WVW5_!tCTn$c;DIz8ouzQq7kkTX1DlxOuwWFb@>7QSza^ONlBy-`%
za4C;ZAF7Z=MtKXLv_*ZnzYsks=>N{Cj0_JubiJp;lTp-t0m%mOiId$*|Ht<Q|7^T&
zl1zt7K`dL?u8g&4q5WUEc>k#h@&DZgz7Ip@z;>t8c2kvg?eVw07)4B`i{Q?1tzUWt
z^E>dz8=Q>MSHuxzgnv^9{X6?znm5}vo%WR&bW8G{=acRVdzSbtdj8gqSBw{C`8dQL
ztLLqxno!ixI1R4Rqz{+}BpdJ_pik0<aoZoqia#A)jQ;yqM;Bv#(2PYs-@dDreI4pV
zPoyUv9tD_Aheocr_(KS7kPbtrwhxT{&7139T;5V9a0<#moR<s?3@l5Cj;A0>>vWu#
zO5dRJ3Uglo4-kKHpe<msVM`G<43K4<;meLXR~WGuC8jr}UV~qSZ_BZX^BdkDu%;>(
zO052%2@56g5zCuF^8?c-+sKKkLJfr)UZPO~l2f?K9aJN;w!j(R=AXap-eF^-U)TnV
z_+O}uJWCLBNeJ$^<(q<29$l&97Kz}<z>^?YfEBjtpv>uj9Rv3KKsRVhcz~J*4w)TZ
zw6`+w7mi>gWrMHk)Ae4lLy>qDnI_UDIp(bc-9QrXb;zV+paal)L61HSzorelxITio
zz4RW-2X+K1I(B=ug9jtwiWaFpJ3kL&ta0QLvVKSu;f2AyfB%^!&1WJwa29?%y#hMG
zG0$=E7xBPC;dE{U10Vbng#08ZB`yf8{VcFRv|78bvk^-8=3<vkQ(hfyZ!3(}*6`Ex
z8s%oa(@0*?Q(~=Xm`#SG%9lgw;Xks0JqCKtpWojNFaZ{~LGey!d=$ezqcE8l99vwW
z0H_MZW)h(j)y21bSuidRdDsQ)17Zq+%SFgdo{!a-Sy@>jCdT-OrT0MRI_!T*)mrB%
zpcHIo4#YTyO~bzQa8ZFW3uyS(BVG5%&;%`{jVz;OelBdL{CukdTvhMV9CV{C8QKn2
zG}B!gs#a7IMx|hp?Gv_-(>e*mKfw8if&fP3+fApy3L_CX9o`}gCL`xQ%>m;Du@h8T
z|D3AyR9H|zuXPNXI*-;cF8w-)A3fTL2o^Aa(lL-kZpXL8nE{&81r-!Inw($3M4beD
z18`h}Yl`~Y31}e3w?jdD6Jha|EAj(0q8;Xkv$qYLyqZ|a0NR|dola5B7XG-6r>Say
z#a&$~E-rpg{upR6)j3U&F(VDqHH*C(jsLfTBE*F!7uYj&J4`0^L+HH#Zi`GeBv={|
zF|%;2anePDUPs)8XQ17x6Rb;_*ZvjVNUz@^r+|{J8MzYJ_wbUKotXilofuZ2PHzHz
z+K*d8TpZQt`$`7q#mXK3`fSXb@Y$ery1Ei$Ub(3EWMcc^PsDi9+v!3175wFxN+{eH
zM_<oNs|e!;!UISvwy)z{e{6}(6_f$FcGD(htz7v;-Uz8qi5dLTkSUVcGq``T=W(xq
z6|9nW{Q~g1^#)7dO5I|dE`!l|$J|{0nEQhVaeCe%N!mqP+}krXI57qM7cw_wHy7C8
zh2ZMuMpDfLAE<nqAFw=ieppt3U3`vTXKI}g*~}~Nb+jycEuD1obSi6+6WK}2cCB%c
zpGpQbX|h<^;{)d;^#PGq25}GKc!}Y<n+|qDxZ0(PHPDGi-@faEV++wmY&H+nHe|z-
zV`BlB@hO%GxDIwaGC@35BS8F+Q#CTLhmBa;jf+?!MJ6zS@qjKU82uSsfbAS=H%+zY
z9OA)0ft|=aK&b>WuKdvZEI>rLfQ;0$ep?BA!>{pd<q-Ek@FE*yOH^TEcInckc=cqk
zZGcpki(Ww@8GPaf^_8*}bH_~y1-&YXV|jMV>42opPv0vvKHg0KTZ)ErG+!esH9RN$
za@>n`U;m0>3!BZs7XZv0ijRavMiPw*3MHRoZ>W`)!=$21t1R<Y3<lKt!=nOO2HZ~w
zHJcGPknD2-g7OeJ=7|u9To)~O6vP~Ok)ua3m9h}E629$Vd;T=aTHviOoD7dV9|LIz
z65Ivwvv3ArED|d_JDfN2kk5j&(P`b~iNMtoe(qF7Y>W)^1~l-n5ZKOah|e4xRaI5V
zcv&EQn1FN}L>;;TA9tUpPceU)n^wwOVFRNu1hci+lYyVT$^qCQ5}JV)1fUxlJ_yuo
zV3}cSOZ*y;<PYO+$IT+8FT%poue~(ajBFiBtZ4*VWYhp)b)dyHEu8e%NNGmt0K*(%
z`x$(2LcH2%Yh#lM*dJeqNS;lEg$JMb_(blaj{E-gYbxM3&l1>#uxr28xZ&@Q`vE|6
zyiz7-J~jtv`lwi9#UhG|mi{2&F~*6wr7@ORF>Hzy0w@j=7w0jygG;aGjL^xZPLQd&
zc)s8hfz-HOGrmeJ8M;_hvgOUBEW$jQiI(!}YECRA?!xkyCxEUzeDvto^z^Y(ub}&1
zMX#))zm<Rk{navkiS6n??gSS){h`Jg&uyj!a-)8&iU!J-3BSi$O1}lW#rDP|vRW%m
ztxsY6l4m(6rsZibWZxdr&dRS;&hoL0^`rPl$(%s@Uv*C<Xh!!`vQ$0^&u(N>W^BE#
z=e^yo;a6O5&yfG~;C)Ybn2%K#Yw}t$Jo8OAdcZ{^_bIAl%)eByaa5tXFy+-3etyAO
z(54D`r}udZCOq~_H+U$j)@;&V`|^fe*p#E<!f95U9D!V&9^IIp7af=Kem1)lZD{BB
z4GhzN-|n~8)skkzo|-h`WkNAB+y+oiewY(;b~CD>Z%<BmiaRbeR0x*a#t!>1$_w0~
ztag1azsHZmISv9x1Jqkr+px6Ixx`|a1uFaW38=NC8ixRjH*}g&C>+D}Ps?tm<2WDa
zaIL{~#211VV-pkW>F79^nVBcq1I<bpXx|mP&>+60%&QR$wxL16m?Pk)ygX^rK#CI;
zC4n>IHS{9Fv90X)rn3R#sShXtKPmJ&CqqnsE?;(ke}5o*RpX~*F9$gKo(ubVjL2mu
z1HN9C08*H~u!7@l!$&wz##zWFu<z>~cu<5Y544V<??jpcXgkFBHWk1dsv)01`xf#*
z#7nwhpz%tt8+s60woielF5cy-GQmM#vVb65&+dtSho}SBEwoy$-?(wna~IQ>8=&FE
zC!>+IX6ZZX+ra2VTea63zx>mEgq>6*N!PS-1O_KJYq9`A$`}iRUV3e{geFG6DPTm1
zkoDt}W-=H6_dbZE4;O6K@5>O<9Bi$dbbYD)c&78>yzu6!<3-HcAC7z&-Mignb1G~2
zWxc1y%~@+ZY3EM#;aPf9Qm?$Ho$SB;blvRQ(<u}&bayDMdZ~dcBYvdXaM1lZW!pa0
z0Qtv<TK{e<%XV?1-PRu8LVsSJURz`8d%~;4hvILK1?e(AAO0<`8+GN;Tv3|fdo4#>
z)#<-af3kc1{Uw8^5|_`=(O1=OlzT;o0LM4fmYH!oX<+?;yC`m{UjwLY1b9SqY&_CZ
z)WCY`1u=1PeK15J)*Bw`Ij!lW7Y%hz;}+V;y1F_@zET_|F8aD3J&jRwh-x6x%Jyk!
z*dw4pr386xL;KFaK;cZ}lmI9+r`NMU9fonO0N6G=SRoCxT)O73s_z<j$J5n+BKD{A
z0z7*>^QB^a(A;zUJp*%0!kkr`1y%|>3WV&&1^0DQ(B8pq3{C~S2B1wsSKa8gG7hbR
z0Laro>vM!1O1qs601xo~rdkbO0@K*hAoNSl%Y=Bq15x|K?IfO}Q1dx0{tAOswzhB>
zamCe@=vR~3)f^!sTZ^L_q)P)|Kn)e_QJU$mUSY2&s;I=V&VuS8WpgqSyR<~t*rIcR
zl@~k99Kdh)Ad+H)@6X{?jYWb8oS!}Ro&q!RF3nzw5r)0mkxA&xE0q{dD~x><1p`v>
z$JBTj^#a$6+)D|y(mF?<mUS2{^fWuXHc`$F9WTmLv`qSKGH+#7l$&o5ozOFyh^g0Q
z+bq;m!`uGaesO!P)$S&}9Ge*(D_yH^>UDb_6oxMFF1)r8`RYI;NK;F@V)CynuP<|V
zd)HBOCl=Qh-w|zIJxKl1P&40GP++8bq+U=O^KVDG2X2+`t5C=-|A1;^7_lBO3{03{
zf+2|pgNexqj$)CyzC$!_PD_%=?0b-XMe}gOGZ_U7>_D#s9RCLY7U1-sDk}?ae=EUD
zC`utE0WO`f-aB?Ztg!%}DbKcIf-m$==+|D7mPX))ZpOpYla8L=2^pcL##^T{Zm3Cs
z9l;Gsl0WAVCI#Y<6A-Y4ffy{R%4s2HutkO~G3?zR)_M%5MllT|^9>W|$594d6cYoN
zciYyj@a2F7azaWy$_}usB`;rQ$yAOZx)-s_@m7YJJ9&%(&;qwDLk`tYy4nTx3Ndj;
zjH-8f3TYx3;-EunFG&$Y7O=b9DNvr{*k`|n#*;PfhY1+6n4zfx2y1Y$S~UiQen9n@
zUj&26M-Lyuy#Ur8Fr}JC+~PK9&;xaO*HiPvd`MkW$KGw%-kJ<T<~av7=0`kLgZShC
zwd$CMYNOm>tc{#dWag_2-Wq-h@cD^Hj!W_<DALBP#Dj7Ag5<6~*$J|r*UnS#-)>mv
z<m7iM%Fk|K-7vDqom{E1N1lQ>EjyW1j1Z1nwX%i()B)?7SZ-+dY4=s`=Kr{6U-^B-
z)$|OfiIV8EmujeP;ehUEd!=Xtm8fs@D;<xU28%ykd9<j#$GDwhkl*;cQT4~CpiLfp
zkG-`P`X<F*4yMT0XZFlJp7h)ME=+^dFSOFcrnB3$NTb|Fra4ABc<kFZlWLP_<;;6#
ze%@_RslkLDUkp8#HZT+wkhO!M1b-cvF6GXho1fFr#%fui$Bofksi~=WP7WB`;Fcq|
zB9eThX^NYA;nA}QBY|hN-_Xiu+4>OMGEw}z8-D?+&bjQJ`}S$UOYdnJ`mNYC=gytO
z6eWcyF@Yo{!+&xC=yEk+QpMopM6zqr<AU34AXXMd&`1-s8tAoP!+5y6^MgwR3l5CT
z=o=m$hFwTcPZQD*Z{fK<q_Oy=tdiG*2h5Xa^}feiLb}pOHw$_-=nl3ZB!LMI3y&B?
z8r1aADBxgC!GBr9b~qt+5zpg%i5S#hCcp^9B5KVI;O?IAQi1tt88#m(hNE3-#C8vc
zxt^bILnMx!Tk~oh8PsI_6#ASmz$4r02>5sI%C87BP`a}_dA`0ZxQTus4l-<%jJ<Gs
z%D(4{w03hcR)d3G5?F1(`)Kx7u4A=s+o3BV<qHl-1ZR}b3&_He4?{hKA&<Y&2gtYR
z-@Ic-oG{bfIGo9eODq`fyJ@e>4Q33>M3km;7%ANrFS4xx_53+rD@#kIPyyGovSOW0
zk_F4t_<W1L78j7Q@S<jQH1rJo%iGL%(R1F=uEIBWR$#`m@5`Hy#gYOZ1{^0lRP|))
zMfdqto{Eo-pgB1ABI2>P+o#H}f$u5$a$jG_$vd8<OW;;%vI@Lo$8Olo%7}7d8eODJ
z<2I$4CV)5!r(E#8`^?oxu{c8@_oo)%wCJ?|<vC6f75#f4Q|f`2m(uGsw6wOvnavr?
zj6FR)Jq&VPzUGj2AagVTGbC%O{mU1ke+%+GANKW=?-gT|_2r^;US3{M5;RX~EWkTK
z!Td4le4rkRo+{spgO6bwGV70h)HwJOPpqiHE&;KAy>Tz@YnXm+^lwg3G{ja1Y5{R>
z5t4Fvd!3Ip35485blu-lh@IFuU>V?)bO5F59s?QBbw3rB)=uEIlmsCrL+5D@Dx0Wi
zF<SS}I5k=Ar3sFXxY1Z-Cs$rxj?CTncE|<v3AMDIr#dz!eD@AAL}pqbzs)!(v~Obs
zE_eTHQ+v3ocJDH1Wd-9rk^XT;c6KMx=kPzDY<}V1^IaRBA@~&Bg}2xsHF{yYP_pj*
zdPOo=c7PJ-sCa5MQJnN~8VOh*S0*P}>YlQ@ZV&g6cw$uVqDwE#KX&;;s+ZK+;DW54
zO!xew``iU1wIafEzFzwBWc9^u>t~HGeI0ny)5+tcKP!=3WZFL3>-cC%<@ez(_tdH<
z$r;s83?i#)gXfL<qJ|7+8gE})t(*HQ%Wd&6-{VN1NX1?yih6VM9lfqMjencdS4mZD
z+F5fvvT@C!_m7(_l4l*AKU!={i%*lEl3uIAP@{Jt>TB5DXQ3MY+Ane+-?3U65q&Ls
ze?;~#Y@uc-y!RR^$H&~Q#x;QsawG<74+ve7hTCyv6R(q1-<ftB%GA5DaD52q2-<li
zhX)Sw3Hj;K(mQUJ$8KX}#2@N(4}yQt_x!3g`v2u=yKx&KE9M;DFKLGEr03>bjg;Yi
zQNm=7v3bwZmnWCR#p`&A@;5*fRuc9bw=@YENKO$REX~XeYQl?QZiejTkj9Qh6lQZ*
z&b+ezdA)n!AsY0s`=0??Fh99Pb%kb^O(4+WHP6yaK`p|81$KWAB2`~sUvDo#%SJFT
z+<J>-4sF0BfTdkf3%<|UVO)BX>-|nYwUiEN)eL>hl8d`1e35M#G%!z!`m)lxUPl=@
zaOD}1@?P9kvOjW5*gB{B3dYGzG4DAp#w}(_%?x>S@3*djoS|wpeb2)^H&@YbW0D$^
zxkltjF2)aSH#g#v3f{0ha`VX6t=muAF819X`ZjG@IL*}+FL|V*@{@**&Mmjby&Y|n
z9WRv~3x7|T$yv&oTr)|F<U1wuOJ}@_?itVe-<0q<@g(tK`-!0VDjBv#wG-1dU)or*
z{WL?ZtUDO*9K8Iar=yBn{IKBTo%6cvBDudE=N3x7m%JEyf$ekAXy*ZqWJQx>PEqf6
z&+#U5Wwn16H>=uxz)Egp{T!v{wfelAncPG1htF@GR&_9$yr;S4-nv}rf#YR0LgL{X
z-Cg(T8B@A6C;dhx7P`1mM(uqR4$t#MYTKL8)jfaRoX`;eYm`ZZFg|{`U!wi8U+45U
zfjARCc%Q+<Oy;7Lyl^H+25}=fuP2+-^YVXS)=$Janr_2yOtSz<6zJB>;W|=y6<NQ1
zfb9yChYa0G)D(?Z?$)8CYh=`{{Q$Z3Jy$T3l%1R}#6xNtIoQ|FPiGA9kBm|DJX;%9
z2+%HJN!fz05_!<U-@+|X7wH?n9`S>n<R$)1k|=MmC5YQAwgegxV2Hu82?!ow)JqwT
z&>PLoIiCz<7Zg;_h{@8XoxD}l5HB(=L*fNXb`9vO=#I15Qy=BxQdkhQN5w{<n+Ywh
z2aZgfgKoEyM0I+co`vvoCP%$igob$M+)a~J^mF^IAGcf`yXwE3-I^%vIIZw``9~Ks
zTJxBZP}a(A@kU-9du&-JY@SzVdYIKoi|{3*g)OIO%KIekbIb!N&xM~MJ&o5!Ii859
zawj%4FhA)zBQ0{8W9W*J!O+~JL03OdYqjJ=C*^D3JXW5%7xq<~eo??t*_;wTHQB2E
zjUBGdjIqWZ&T>^jOAEQ|lj1D|i(<HYcX58@;o@Cqdg@>c*8K=HZY#9aQ9Dx$O}#Ak
zII8i<Qo-<-`^j#3`3vvOatjauqNl_e_yz(Po7o$Rido&gz8=4Xx@9}NN_g5cVldiS
z0xYGmH(qfQU1+A?ApfZmumluO86>V*Y1Oi{7=YSiP3*Cs)6>dJF|pDMy656S`f6La
z^2?y3U=|o8KrXO*Ph&P~_tp{|GB}CQke)+UNrr=*UJ20~Mr02xHBPxi7+b$K<6%B9
z`Et*#+L#vi=X;`&Ar5sGkpvlrpscJ0NHR+V07>uCJ&BANn=@f4FE5X@^;=&b-yu4Y
z<ln!Yo-JbWM)>f$V3PsevUbgzv&SDDTE3ktsa<I5mv3TKRpT+46MydKRx3m8vpW53
zhXg+hu4-5pu=BKeRXDfXhkaf_TNzGq92x22Yb<Ks{isq;XG&4Y>irPAl3vQhLSJrA
z`Qej+lIl(IN%V)m>)g|ueI9Ig&y~Y4o2#!(@R;&3rz=$#9x@8s?RdK5klIpRy~Hl2
z4VD{*9o9cD%1P2MWYl|9P+&oKR%wati2X**QV^<M<lmj!vj?O>+lgw{^}D$G{FJq1
zInNOT9v~4!+=r7Dbs|z_L(GH_eMV;<ItHv2gUx>d+Ct#~20>M2WrJl^)83%q@pvKY
zu`{!gadDRldlPKrT#xyd+gAGCR=F52e{1jtWUDy4Wx|34*x8NHPO+69IBY9e{o%vK
zHt&0S`*$-kYHMgD9Xc#Wp!jWhx$YV|4d(>DrK{qO;^eFjFH?;U^owfHPexTt+>dWn
zaOoZ>#E2rykS$s&dMVjTO=YlqOKh@6VQ9Ku{ZIxsk>R0rZ)0oAx`dmp4S5$SoB0GQ
zaSSM~3xny7=N^8!g)>kbN-VTp-&mW*AFbtA{x{B$M6KP^MtA3w60S5Qz?RG9tEf(J
zv5PxNz>1TSc&y``mw(g=Ezq`|vcL4}fV@jg)18;{22yl7PFn}Z<)3^Qc(v}l$!!6F
z2O7Rie2)VT|7`f&o{_)$-D>F#OXfQhhE!S#eey>xPBk=ICUUS(BoFMKz7z7C0}!k2
zzR8J~?4D+m_90no*W44=^oVINbe?195&I;%UkFybh&h(SbDbB3>g(&XjoWxozrT6&
z28A-PoN3^mwU&wEF8xPOb8?P=J4lW*T(3h152B{V%=nlanRMS4v?ww51_Mv-qy_mb
z$#@>D)vM2mJw0*sbeXNAKpr7fqHACHA~p3k{>at@1BDN=H~eHEs`!!;AD_*WvoA8R
zbvAADpDQe3H!pm&FXn<|2%l_*#=_clNGHF2sXICB6rK@kKiL`e=XwCe-{I-&iGmDe
zz&-B9W>pXjG$bpXT-*U)67N@?f&L<QX^UjF(zHGDOIxme;X6MW8)DIG!S4E1G^}aO
zzMEM)PY3KE+0kL$CtoA}f~N8=O)%x?q}K?oge~W6!nCs34<fTJ3=KTR@iXvX_D@;~
zaqSqnBaMHri4BuymGynl)4pDfrs-AsPLsE`c`EJgUr<k-ijo_3X&D)>1Fr=sAX*%K
zokc}rIdBg*hbjYkCv0dRNsnFALUuY_6HyAb$Ydueia+FI8Ryt@4fTR`4%lmSYKZ=Z
z8HnPR7G_voch@mh9kt_-o|yeDXX+8`I`Z<i-1zkLc25)T*ZdDsLXYM>2#`DifB&u9
zloFMs93Ct0%+%KcgcR8wdE@)}b*P~foz%G;qp)v{kBafhTsZP{a<o%hQIz{#{$n#1
zF}hm<$3KY2y^Pg=D^rE@h|rL=+ge)OQ>yjv5aT7=^xCh*N(V_Q0doY6F`3%spR39%
zn<v`@lHR3$<Df`cv!)dsd=g@G$GvBb(g8!kjST%R2bGU1yHBgIj^Ft%zE?a-c&ga<
zLeis05=#uM>mYGxgJvB6f${4UitmZR<|_+Rqkx-{)W5V7GTnj(E3xRHtV=_^PC7+c
zLKN^|XcK4eyl8fT#s|#u!rYB#SYcM?dZ2ucGwwEbG<^2jwK*oH0jkF@DWiD;_mD8U
z3#O`rIf?AH{)=ISg5KW!bBrRwElf`?+`IRHqe06set*{6&G`!KvJEL{TqQdsE5-R6
zlpf94nwhA_wC1{Ugr%b4u<GH%2kpk^4;6i~?>%L%d-7Bu|EVX<zrK0%nUCc5cI1{h
ze6kzi{{R}j^`YwZ<3o~Z^FP=FBV?v`zu@b+bZ*^yz53LKstZi~MFBUj72jZ|9`U#J
z3A=o_S<0?-W_GsBNf<C|O5!7cX>yCsSJ0_pxL7n$Zf&Intjdj7*t$lLlxzWPE}ebu
zdpsiEQqzn1J&o1pY`8P(CC_nser)KY=Aq!1CW}qu+4g50XcvcHXSL}$a8X^Ok$bNt
zkeZf}>TAE1lS$q!GuKCJ`{GTD2#X;5GakGKi+n5RP{(t5&ng$##HCmra@zO0=9qM*
z&5sUiOt)-A=)-@?Z>FPvhAtY&joa;WA+v#mcm^Gg%Z_PE7GO>KiZly>l@8t;MO)24
zBr=wHRLzxk$?5zY&=jpAN6iWDE$c5(M!uoeMXAKXVIISXKIWS@XWqrG2SOKDHD~Bk
zdIq?T@8ie&cUFiy+Gt^lTbF}8M@uQlNk9R1=zEJ1SBkPbwEOlgz4|H(HXKmwScP81
zCm&;R;qZxS!SYujyP~8@Vdr-iW&(cOK0b|Cb&^jfbl=i}htx;;UF4vbb)~#ffX8t%
z<PF=CECNG93Q+u44H~horKf+kv}fbSaJAw+BL(ChP#bCZ?6LiH-(C_o?wwL3jji=e
zJl4nk222_idw1nOx!tv3JetUvAk!qfG`hx9>c`&=Y}!}MD6V*d-w|&>_b*>*$KLFF
z?EO^3rt59d(fcNh3Gzp)=stb8ENC<SqK6ilqyJKp?Wf01PvrXS3&A#QDzp;2cI#Fn
zKqb%$R@2aI_XTFhV7qVPi1xmUrY1|cqVGD(Hhe`^wj&cz_}1T#pFNAsQOGST1f-MT
zn()Y}-xx@R__Yik?P*2W)x&Kdp&K4nPm)NL)L4Q<RKM{;p+cODC?)4Sn!~vN*0uBS
z@>-!bCOyVGc~iu?C!#8!<j9x&+4kN*ZKR;9NLm5|89zY@@F^{a<Mm60>mEFGDd><T
zxFX$A+a~UnbYfql6%1hO#^W<PjgSHRF73f+tA%q~h(&%Plw11xWO75JK+o-7CTp^@
zX`-m<=Pr<dTzcZy*S|j&?+`Z|ySkYn^ziFKzNMK0gAIuu4uMt$l_rhS{7-1=qNM5j
z$6GgdKMT{Ue%3$6b;n-qcl+V)Ml&@9gKx5}=h9yvWLuZG!jyhT^V#qGqE8Po9yLan
zZI|%TAHBUxV7aXpfyrb%=h9&(Yd%g+PV_c31wjwkw5cxfuL>4vAvPWRa<rhkMcAaD
z8>@`Fr9gVY_33jS4or?V_VWCuTyx#oUx~<b>dd_Vf+p=8MY1074PB!nT5;5H^8auj
zw?~qj5Uh0@4UY&2sGWR)5l;;_idqLTZ$R!c_$g41-FMdnNJ+>Tak5e9W?qukLn|>#
zVD}{i%E^%*I4jVW9R@aJuqKJjez^XQws9Ofd9-I$(_TP!pKvi0rqc+d%nY2W>E|cE
zJweNy2adS$H9L%V{pNy%T`rT%Tu`#T&nzNB4D}PZJsntBpIo_nq#48(jxbJ^5i#$5
z99Q^3BVGiQ(`Z(TH!Keu??c*yP^>T*P#XH|YXnuXOW98De?%i+kAGb!9|>Qc38T1-
zoWC=pCw(|2hKwt%uF*ZA`L5oQZD-uwxqy4Yjdyp7+H8Z1d|sB*Ov<?U2OcG!PK$Z5
z(xltbNfL$1ON9>Sra1;OKDNtVuU_@E%SxrS2wWa?K@#oA{L%4wZLHV8L8XVU_8E1)
z*3=Y4KZ1Q?9q+s7mURmyU-oZxQ`~L(!_{6Wd}hYFOt*g>u%|2Y$3h;Brx1(@fX`|X
zMga>wAcATm3RI0R@A_?{utu(4w@t68HS`N?dKI#57k}rX3v6vYg98dZ22h8IW&sF*
z{A4h3=C5FkIqrj=ZX%~3CYQ-0k0rA`3an3Xb8{o_KXBk1o&>Tac_Tz1w5?6vSj6(+
zrN9m$jvH851W`1-4nS@LD!8!lqPNH*z%bgiYoFLpoH6T9A;eH_zImVq_c!dd#XbNo
zk8D~B_a6u=32h0~<?jvaQ^^n=0p}B(oWwW-MnoVi;HE?&HCiA0>04<zuNf#k@96!J
zg4gSe#KI1`(FDNUM>f43?vNntjP^msH)#3h9cDS2Lu&N9h|Dt~3WD|l@Rj{L*MQS=
zSY%gOWmmPtn2ht+;ypWX@sE+CXopuI4LuVp`LV$LyVm6&O}njmRB9bAw2xDUjfX;q
zLIh?kJ`3yR7kL-WjObq(`<s2822b>)FKdco5=V0T-Y@gKeM^JR+Ix+Bqwa<jgcvJl
zEVN3RXbuT9Uf@u(&gL5{{#$m`BSHOjN>p-`pQ~T6OPl(LDcL49DG{+BLKhP%ludID
zgZxS|0z{;ffx{!PC!$jz-i?K2beNDW{*;xK1-47tKo&$=Ru&fQD&=Gf2;gA3(8t$N
zPXoWUiD;Lx-!Kd61{xXydtH5Ss3A)#a#~s(h>U}W5iW1}QsUfjVZ_IaaxbEz(Sp#!
zizHBRYBAe%)`c^VwnAc0Bv5x}91#fM<O)|++3YB^L4i8hV|b~^((hMwTzjwyeZ0$d
zVTb^(cwGH~I9PxTj<heG%I(3fc=$Oc5(g5f_l~Gg1dO|fhlP-vgV4e3C*m-tFwS<7
z34C+E0Tt>}K#~xk%%RKWR}N(o7to5R5jA=!w6NinJZBq#QH<o=*-WpAkYQhPZ!JLk
zP`Ozk5L`|&a5c;R@tN*Wrv7cTXMtkLsGS+ZJ7X6$7JP$sN|H$^_YJT+fUe$`y9Rvo
z9_7ss<nC}81#es_P|`m%6pXI|BuuyyB&6Ry4G7=^E&tu>eoS9YG|Ag@D)-ms(>O52
z&y0m@6KL`{*DErDWBsGAq6I};7Nw=HC~1a=t0$YZ8(co!YnZi)@c|;ojMWEkC*|u}
zn=TY-Epi`_8on+(@V4;by7^N>H?+I$(qf<X>KbQis($=3W*?G)oeG@~H8r)@Zr%}G
zK<s~Tlm$T2!*>`O9X+Of+ZIhBKzT~-U*IkTQTJ+*5AWU;qF8`A7g`Nv;IBbr`0ybS
z7;M;34d9kb?{Uv_^cW>x+ctx(?1Za_Xj1s$xds?7pg-f@#Yn0#;PqgM&mi+aLW349
zAR$2fDcFS|3$ex_g}7Q!wrOr^m~Cm%{#%JF*4FdkL!75XyNvl+goJhVjWasCr~~wt
zF6eFQ%*1P!_pbYd4<TqtfCeXw2LZ(eBE6G{6W~XIo`GZx_J<^gK@(yJ0|V1l223Ed
zWpR*mnCN>lpqa}Iwo`=7SML|#6UKX8tTD49vfzEY2MiFy{)*YZ=~I2ZIk4)1BbyPm
z?qEw=7^I@56(lmhlO#^~QAH&nWhL+Lg$2-`^j>gY764Pfj*-sPbQB-p(U40SI*An`
z!|))w63mkfx*?fL3Bxx$Fe?{%1mT41*S!E46Eqcne}9zjeeku<)2aG~KtfusCyHGh
zmup~)Rm3l!`ux~_EiK26xnF~s@qrTp{huAq3A0{P(Xrm|ttC9i+LKr^cv;oPpnKFw
zLQ+80{wJ<JxLqT-e`g_ogVpjy3h#Z?fd<y$wwUM&5MmyUsaq#o@M)NeCcsM)1F`S0
z?uW;(($fLH-Ygy%M|%ZXr9wjLoh~s=-Jh2(UvA0JfBJn3O#<RbOin;(F^d0fr?<$C
zb;646<Ht*l2B7=|@tQvOfAwbTg63{6JsAj*>?0#VS+vKsY8`5gD}Q#Z6PZEutjRQv
zj9BTt=Y~tFs*C|y<ZalPnvp@;84xr!zib|OS%9J{J)OK@z>_s611n2PuH$F|imU;T
zzbQsX_?ANLaq7nV^-)t&rnpZ(=TQ=W_<mju`V&wlxF8(}o6&fMwl~a|*o5K_giE==
zaw!ep5Hl|#4CS{$P}>VM^7_z44qs-Wl!H&paE-)Ue;)(5g#%?Tl6b&gwBjZm1^jp$
zx*`kw0*6=gV-L2g<ATRXe<A5DV8rFjY|AD_Q1EtaEOxn^B$N$?3eTuQUB+8cTg$ls
zp&QgZ3a`t87_QNzPt4S@B*&<z8mRpxc!1;DO!YS;z=PS3>{S<!&WpH9Jem*hr`dM$
zQ3>zkvYRJWrM0SiB!Y`XqAX2lw|s9_33~LX0)9~q4v9C6jfFrhLf6zl<#Pq5hk%RO
zemV+K9dNyxWBy497tos2ldqRS6VPgILf_Q1JNgNoA?Uhz#x4_m#9y%fUQQ*tZ(q<v
zCJuI(5W47)lyQw#)YTC>6{tHP(=?>pU5J=_VuVBCe)~JPm$EAPsyJOdg)swY)et@m
zrAJc{GDbdVIB`+7#_l}HqpsdY!hwLrrC^4ZFYWE<SJ=aM5fmAJOx~=})O#Jt#b&+g
zTty$C7IPdLNRM3|gIG25CXm?UPV0vFG9m~jcxZfs{F2%4e4<rB!MC277qNlNdzzb9
zhl&;*Si&>D<HfN8?c=G@&X+3JkwIhn^%IBVTbi1h5XSM|>=t{X{rnn%c_J6r)H^Ay
zL%m!;(p1gcmcsn-AB8;I@^Bte5M;!aOqx(PF6GR+<<Zh>o!*f{`RaOf`}rwx`jQPd
zdmBSbE!XEiv%jM0YVKXV+H~#frxvlyc4gI7S{65J=i53&=2OQm%Ra4t+8nCg%Qzx(
z%vP&z@Wt7|M<0IOF~H3OUFC1E?htEH(!hx8NoHmyR7_ZXAo^N})o0b?<f6*~-9mFk
zW{Ta_SAsJ$tj5vp<bzdN<K=4&Y1!GtvlqdhsAtgt1Z@Q-V*R`WKVTi|w9t$cl^_GY
zz7qiJ(Z=?Jqbp?JIL^_W<NfNQ9ON)wb`&k*wd>YBf&(---ay--@_hD88J8RJI^iPk
z!^IR(tZ+VJBrq)IPQd;Q>$v(TD{ORHT5>V34ikYe-cJ$yoOi40SHr`tzpD!$d1>Ki
z9vBwb2i>Row}#?c;!p+(l>8wJ-9J3=7J^WOV|)q*jE8~R;P2lvG^7ECF2qNW_l_Jt
zUgKQYsEE&jKzkd>8e$D12RPwEU<6`uczAeVCV^dr;e*jIxPW!#_sDw*3fr@N%5GwK
zKu!J)d^iM+chH=hV};;^uMIa-L~EEHN=iv#&<?U63koF}t1UQM(AC?!9qr@=<#f~_
zh>(mYZ+=9b0gR7R`!)Qt(QN#I&iA)(vS`Et?!@?mbHb5QH-o7~JRc&5-Oa!NWYB52
zJ)c)VU=HBS>e7jFq|f*SASGeoKm-<`)k%proD3y~yJ*@AoR2;duKpWwRj^Yrp0OPA
zpM99tL|6AuF2LRe)LUSK10?f<i3*&FadMB~8pFImI;=+05vAfr=dYp??>FpwxS5pW
z;$o=AWh_?FADLhsY<&B`T||2(SKGz0svyYkm}bXWolRleoO)+NJ57FDer@_Y=?lC3
zv@Bc6h94+GG@GpYl@ir2hIvg24dx5xsH>_my<pO?wPpD}`k=3^cJ^ecWb^I0V$5-p
zZnk*Yl4a0#L`%xRH~E`dd1Fg8zu+-JexcdR<xh*A|DNoP61no*?x3J@Vyw2*yR)S?
zJ|i99)$<Cb#4T7Ck9xNHHFR`1Jq(KBm{5kcWy~;rhOAi!VPknSE{VwBaGa>Uh?_h)
z9&>A8=BuDD(2I>zt}AudUwlZPj8(3)A3$OYeD7{P89sMKc=mkemx>C)SGg<ri<+4~
zJA(IenN9PWeU+l_g@Zd~E(^QqY(7okwec0|AF}rfb!-<X-*H@h=aSB8)yebmsn})E
zW)1LMD`1s5?SV2;e`5^-DZEJ~8^e710YyGsqF7iE!F0TUyuDv#JuG!EL|eB)+G&59
zF8urf>*F*`zfyEnn5On@K22rHgYlz3uNoS{I@jo(PeVfk@t=zsxUv@W9WiZi?K`yE
z{RxdmVSyzEUkj4WT)8r2&ysapsF={jEc5%E|9pkt516u16|_MA{q8-+Cx>(Wx&wj$
z#KBmdI;hIzei~)#*Nf?I+?XMI@=6z|jacYNm6KqQ5x%v;!U`EP#5bj~g2A0i?m}*>
z=-=%2w%;RTq@t?IQZ-A=1tEwp#;4HV8q0zR0}}rB+@`kOQ8%BzM?DCB$;YKZ*ui0&
z2d9cqiEp?|Q_mV`B~;3q*8BMD3hS=4*Q_!j8DKn=?qWaA`tJ5^Nckbs=enl*@THH<
zE)3El2D)Vpg;X+U0dL6s51)ieuN#(|T$Jq{t3t%6&Xg5vzU<?3qv}xj4~Mt62N<WF
z&|9qky#xv<H^nNPp@W;sdF`FutX|g5P>Po=s}hN}2u~&U26#O957o}bf4RQ>%WVwP
zVgEeOnQz&;hWx{wWsyDk>HD(Cnf&x$@deYG>?(3{a)5#FOocG_&1VF_!xhla<|ISp
zr3jItPq)oZ8`V3icLi4l#D=`s6PjJ7Yzo^CvA)A((Wt1#j$nBG)B7R2|3mZjzrDC0
z)JhEz6;NAlY$7gkTUmi6VETL7+lq=rVjeXBqZ29xVzFMDxD91q$+92NLGjC%!2_3N
zhk(2q=Z;hZ(RKEDEW23Yr+;5(ECL??wYT>FP2u5&@c=~RR9~I!!9WFKr#NkZVT*EQ
z$d}=_$i>A6ZVV$MW2pss)&-p>ut6CRCV#Mm@nhUvTn#eG@QwhZ8uMI=W#}Lp4u7z0
zeRWj}9y)q&nn<;D5`X+AX*t{-KPDB#MGTc9)J_>$lAG+pAJ4F%9jxv>BO{RiU46W4
z@y5IyGApHlLAttLsBK%jVMA708jCIQl{>zFIkM`&YakX7zy(vcWh*nCKW=XPthTIG
zAgo~E>V9G@;=E!xbRWNn@-rhnomG4t5aaZW!=Djojtkz?!wdGropq(fQGY@I@8wtO
zabNa@3uF~POQa`A_qeX=;(1s9@u?ax4iA0P*v>2{RPGPMJ%wt$>p%Vk%wbUBCfl+A
zY5jM<nyoYaXn-aP+}ZWVs_`dp_Aeg>0WXnkV`7C`bFxYqKJS~5Kc80{5DBoeI{hsY
z5)}}leY!*xv}bDn{MIl}c&G^kM5y8qQHU4OKNd0x>hLm5ikY8VuLKqDaq~ax806{c
zIn{Lq9^>;9{WltjZ%z1vKW{HM%xHs=y%y%?rwquKlsW(BPc%WLOi+lAl#-$*E|viz
z%ZuJqT~b0chptKp)&oG+otIyWdHJ=ru53L$(*&1f8XB4vWbMz$E?;(LY8b@IK~Gdg
z0hE3?<`mwBnRfa0-kn2C=PnpP$_)r9iTtK!H<7?Pe?M^e@LPD-w<lt6At^sd<iAu4
ze_n7OobrS4`9>6#$k$}%{PX_12YgIGBSG4WW8+`X<{_K$)~#qYVZECD5)$79$jcYw
z5ICk)Kv1{T6g)Rv2G<$%?YnZ<u<Y>eGWH=~?TjCsjFC$=ybgKUfwX<~)E4|(Jgm1^
z3^ZImSM7O+xE12@@9_U;9sUo7|Gp&PDzkE8UVHQhQX*9979W0Zm4DlUiy1&19+*V|
z)-lbtD9VB>0P6}kD<FA=V5|WBIS?J^XJ!KOMwuSnw=(Fjiw?@BP$+(1#=zyjE<j?F
zo)|1PoR?7ULtP^SeI9xZke(CN89-hhT7^KXp~2u_XP0bD1I8$OvI=B3{RGayw=j_h
zaq{MX78P~rOCBmJre(9kjB)Uhp!uRe&V$=KW<<hW3HKPJ1_8?+8qSMkABVf(V;grA
zv@{el_X~p!kJ2^p&+nfv|2`PMpB!m#!|!W>nQsGdh&RH+Hl2rp<doTwkl)~~B|jZ`
zejP0>83=(%)A53T)uIQXTcW?GicWVWmGkw7y<ISInC!3ra8?wJDk8wf1fUf7T{wRn
z5%qI*eFuh${-4*L`$}$0Ur7D%z%p_IBE@)X#bih&zrt#Myj!BI3Jl^7IT<k5QF3Cm
z7if4yqXs+`e}k#qxPO3eW1zLh`SWi9wtCo0mIGdc2zzvJ5IFHD)akVl{53XmA3D?p
zQ3~cU5-vJwGc*bmeKo=C=0#Y*Jngo&d^qW&yn|9&1eFLXDd_l=C(eycPme-?i3+?n
zSWXUuY(aNN&z1`_M)3#md;x-kKZEKN-6R3@)ltzwyJ7l92<>evOr`}U0K)g3+qWNh
zduylPECW`bX<CdfFxVvsML5CkSXn{OTZ9b&5EL{7g-=;f8UkTKocRoJ2@{{dm3;r<
zLm0m~QEs8GhwpZCYSi^RHoMDqTMiw`t@fPbU4bZc-ppq5FEBB9DWiziFHntN?ig>V
zqJitVJjIH<cVgS4`*w-sHvodkG-Doa=#VtkpF&>gdvhAQYyx!?7MP*61pY*TN?7=J
zrQp8s2_{A6k*5S#eL*>V>*h^D(I~PKCnJ#K^OE6Fb(0J!(CYv<6{^k_fTgf#$!MW~
z6Q8k@HEvb|En3JAtZ*g8U#6&}olSd!@d<i*FW`>^4C!&wV?RIWV79)Cfv=N2idD)7
z3KU4l80g&O`vPSnzl}F7viB=PxgVG}rBsWY+0$z1{#;|zq7n0Er_k*SrZ`#NNfUtZ
z%gvpz8gUi|D{y_IZ&Xwl;0gFMHo=1JTxNVVs$;m#fw<8CrPEfmT=>lq%c`Csj|2>f
z=ml;;R0+z(u>VBUj`Uxos}kCj;p$r3_N444kQdNNMNlZ?yWy6}Xqm*fMW>h7xcY?v
zngm#{Pzcqdqs=bGEWEgil;$*$OKb%u?aZ_^Vx;g}HXnQml*8%Ja1W@F1w#li#N^uU
zj&)4q<2<}N*5txoSd>Ei$=Q$i<E)1L;?s?#IzDqe1IDZ1XT3tka^F~~7FQPkz7J#X
z3N%SWNkAYEFoSDT2;ZISK=dsPexuog{=Fir2K@bb^jB9UB`0Gq2||d|SP-BCWj&k=
z@50c{aee`kET}zRbY`u_T7&}5`P|Z?p8~zwJuqqK=H}p6+1}BiW!{aQ2@Q2eYT~Lj
zIR#bzvwQL{r<g}Af8rSuF`d(~yyWA6_y-NG4@1*SZBvg0*!KU1TtrNb;|}^70Ktxn
zRXlv}wKvh>^(aJgb@4xkhKw3TjR7I;gStq1IBpIf3F8Bh2f^G7i%YqEyM|H<=G&%2
zVsUng6%|)OwpJUl1)~(4Ps517{J)3@4;FPi+?l!!6>XJ)Wof}b1b7EJS5Z+tJVug*
zkCvky$$Y`*;28mz9sq_+A|hzgss+1G!F<edw)+{R@u=2p?CkPEHC0y^<l}?bhWq&O
z<1oB|DFKCu*5ZRJQ<&iqC<yMU3$OZlY$;+}Kxw@RFd5NFjy2?zu2){!ct$OPJ%Ynv
zVJ_~Ex)){8HdwTamyQYDh*L?6gt^qzB=-;6TRN6D>y#wD+ZCE~r6*>BkuuzgzGlKt
zJTeQeWoR&+aRDhOFDFN@qXL%{f*8fV{o<GlS6M09sELqtLF<Fe)7J<UVw#yyRRewo
z1r|5@=g*%}BE#ED7amYpLXLnDO*7UL6!Yxu=$-@u-h#pfmI8?Gv!rdGNSX@6o$-h!
z6*tqt^=W*z^%;AOz7P!(NCex2!4A^W%Fyrn{#_2I4%UK0Bj>n~kY>6Lz@EHX`=C#X
zn5_bB5Dc7H72*IbsRQ5zQhM+HeO8Vm;U+?Ee*s$tufoF863sM=+qZ9<nWf(^M-c!^
zKr*f|IoTNUCiny#PQ)1{Zl(~m;+Uy%wNHGLIUJKPB9XfzZJpMq;vda6wM$2kZ4T*a
zeS)_DRro3vabZqg-sZ1eLRj~dJfaJ`th5NKIFn#ce}BBqG2D&vRx)U@Dr9fK{Gdc=
z;n8l9!<e+jXI%DEknNd3tRd$>VJw2GktNrVZk8`l5;*pW=x3O=2CRFXRwD~zGt7o5
zG$&!o&^U*FN7=`ZkCR@G)E=m^c}NDc%ndCaqnphklaG*;r;WQYI>ue|2FR<NWCZCA
z{ME{WuxDyxOb7ZbY<3@FRAxpkP^LgZf`LxhB!AD+=3j{%^>ao6Qv$pnm$&cUCgs&s
zZ>ZQgIEV%e9Tr<_>m3PIfNA5mK~#?Ty3=6$$-pFn+jb)<?W>Ef=yY02HYo+!3y#(*
zl(nm?pgF}n5{BrrltcE2{V(LZRtwJ$or?BBU?D~ewp@+qg_VtyF8Hr1-h^&I%>bt3
zrKR*!5p>8>H7ENL;-^042^wq^zbhepX*dJl`GD~NTwg+|X|(;x7IT|H2=7=<o=itm
zQ@k4ub`lLC1d!dun|*_Kn8NvWIc$4n#-kcLlXeZw5wr{tPpMa}`U16E%^bcVfs(<c
zfd5=Ff(~r$@M!(~d23!oaIiWYKn~=wqeHY}#R{KpXqBo!lU<+r>@S?kK94l<BBx+m
z+tj2KUr<wA<b(bK!=_CU8C$!Eg+Q_z10&;0d3m7SzU-LKotY=)h<+fX_i@!6pOMca
zgdIM5)G&7tlmI|A2CAw^S4QBp%glgDPsiX$Xr^+@+B(PJ8zgj0yPJQH1eA3rhv5T*
ziBN=-MZ6nQAbe>LFR!Kon>0boEn2<T7ngS{DL>BeqQ=0bM(meGZ{y}_HnzkA=kCKj
zA`zL|+;F61;K>Vp;U~*h#Dc?)CsKcr_#3@2Okzp&-|nF*?~qYbTPs_%*8vJAB!luH
zUz5Pz!cqJivVIGLi|uw6*tC>jBKAcrNT?rBRZ~*}18M&BJLm|z58a+T(((!44(t<R
zmN&`N_#;&Eh~3Zct~a}UM!j-1J8Ur;iJ0smQS~i{p9*ev%nCX{*}bjv2Jbft=g{&c
zTt~38Kl$I^7^Msb4dRT$%-FTEIS4mlbQ|<T0@546>qH$oza<|%)K{<S?bI7%(1=&4
zz(1IN$+;F+_))XaR<T_IdnLiuhwZaLZWKsSEDVrn#vuN;@?O`~Jv#eM1hM9qx#hsP
z9?HGeQPR6<RaUxartt#8UXAak5Tet25x{*s<6aB)SlSF0*Vg_7c|#9Vw&ZNUK0vhJ
zu!k?)r1ZPpP)uR9>jF7ytqJ1doQ@9!W*;H^V`KaVAbfZbokKCR(_fME!v5K_O++&%
zFHhVbI+YzataiQsep&dC{8kW;KtzzkxV=o!v0hJc#0l~bJhx}hxGQ~bnApZR+@fIi
z42x=tg$^x@>du~=)G<yTc3x|qUqHcRs#F-)LZ}-(j;sQQtwDg*6{QgG_KfD-k45lt
z-g<~s(=|7t`-R9Yp&R+NY0*NK+%<$wq&@d0B-97)A1qfs#LZszA)WTp^Y@W(%X0)N
zK<*kiA-IDwjYKt1Ce~~+A7G6$l#zpIa>}pqNo9Wg(<e_3sHSvpxvXPtGx)ZnLl|yA
zAU=q9!(kYmFtq92?i)XvObeb8*gi{=&FjlfwLx&%`HeQd&oN?C7=IJ9aecV8(S$=r
z0~s$!OzLWC<uva()iOT5X;*v_7|0ShjaDjwZPKPqjL?2pP*ZEDt20CIOU>jtKE>>+
z$AyK3KYkdR4T!yi2-zUlQ1yhQSiraL<ifi*HPCDGEyq^DWFJI!;i};6gCvXq7CKD8
zjPQ*QA5k;Q?acuDRlU&wq$*?r#10$HI80h%pjrP=^)hkQ^XbNNY)Q3>dWgiF^fImy
zwWq0QJMF84bGy1(!FzxysoHb!^S&0Z9JbON3h3=l_5`5IuwF!&^;2~<YA!?o(+G~O
zoi~oX!6$IeUEVxLa9MCq`y68tvQk1X51}AY2_D;HNN5hL6Y2bvP+3)#AA5FYCP+Jw
zTrgFjv8I7n9u~a%9HN9r;~CNx78y5Z5q}ztPeRmEw&FSHEYGGMNpBN;|GL2Q15#5k
zK@jo8yK!E^Lhv(Wco7@{!!NU@a;jHQ9p@!mMW&*p<RDZ>?Z58e+48?lQA{09j1Oin
z0;^8IK~H&C0E;q)Pgm_GGZ%ciAwzff@UVw0yTMom<Q38bLYu4k_0#XsF);=rXA~9D
z<GyKwi*N)2KeZ0<UQi|sG&BSke}0zz1exM6V4P(p7x`E;8@^;3KI8HHA3v6kv9V^v
zSy!lX8A0fPNs12z#RA&s3=&A$@uPwvx<Yc3Co)han<A@;)rwb-8AEmmxlb;}NafKX
zttI4V@z)9{h|q6O0exuy{&1wdfOimP(9zig_Y?3b_no#rj($6(|I)8C%&0A-sev&?
zyxm)UP_=N)4G<5&w+{`~MD&Cp3FY4k?Ayk2>>VUJO)u%+-Ay}YdLuP6vmHkwR8D^7
z@acpRGLnIaTY@_CiOE7K$j>T0b1#0TqFR%RuxGy-)ecxnEroX3KVHbLhQER@wK0Sz
zmtzXWpK&^%&x+#5+Qz1(wRLW0MlzU(11*F0<0_N}hY%A~1CSA8D%e(fQ5Me`_%Fm8
zHg$A3LP`V5111(!l$R47KjEg~FouoLWyk%z=kS5C@{Oi#a4E&#gL(-sDV*C^Fk$24
z#~gr<KqDY*WHMZY$1%ui|MA_3#}Eowg^J{)q%4#6AnhHq7j9J-wW@s=prLy6W%-=a
zC9#yz1Q`>%Y7rP^ASBx?59sOvLl2W$6LC`Z<<;Zhxg`7|$`gqLeA%^&j32p@4mW7z
zMaEExW#Usr4lTeH6I428@pbR+${sUyd6ktl0?9sr+%|3F;X9!#sjez5gA=cag}DGl
zJkuV3inne^e}V=dZfS#}dYr*1n)S7{%b@c_;XKjtr4p<RsI}xrt3-Z}m4ZVsm_3aD
z#T>HPxk<yqX)TQ%ljE4X{`;SojE2Av7LOs^1Rn+VnP<lv5qeRYA+Bw0mBvm)-iCrF
zxxOP-Dmcbuz{(iYP%SYO;(~s02=@9s$ZK=PaFnY@iV4q47`FCjd_(L8ps03MY;5QJ
zIFGK*Vq!AW@)lWszYL$y&_YyWSVaAOGd4I4$u5PF^|?^c^noVDtKSec?*dL(XE*#;
zpkx+!S|q>Le*GI<1y?0lv=Wck(J#4C(<7_uk4tGqz%p64DyR`f1wNhL)8q%Rg4nPD
z!qwQkvYF0HVub}&A%RN<hVFx~7e95NP!h=&lZK8?4|X^zAPn!cLpVf{fXS-StM@e6
z%#1~Vt^fiAHr!{BsxXDtpH-0+99iTH=w$<CX`Vm9iHsQhsGxU|51IbS1>iQE8f-px
z{CFz7!x0I|gwG~{FExaD2faQ}gM4VS0IH&4Ha0%aDj(wkOcYL~*ROv^sDayDCN?q1
zVHZaogWjTI!iYd%JV{pTaBSc+B2hOxbLI?gTySXnaP;@|Aew~Y!1ndEz)})U1pMCh
zDeXR`anG^?PyPDmj<VgQ<X2~R7w0H>gug{e`v0!u_wW8JS->;@Wm?i+nH}oV*l<X0
zsE06B=6Dt&8sP>nXJGL8`%bzEm4t;(5=u3bC4YyO=eOQX(;PpidL{ik?}xz*_r+O0
zdU|i%#{Gr-R8$*3VZc}Vm)6#xHuMi<y5YU9YNA?R*Ho%>TbJsW@3SYdgGc@|q2RwZ
zOCWD89bM+kOIT_I<Ownn-tW65H6(5Ef~&(ff_N$aI}>CVIJZ5)zj<(O?5f%-`ZIUV
zkBy0`Ep^NZhIPo!DOab3#Lq|Xc42%Md`E5KDrr{MD8Z12kfv3>_xJN7&61y_=@~^7
zLXEO%(<W3yUXLExK~@0gbvL(*ID$ajti|brB01<>@cS`TkjwnikEpd_GJ$Emu;n2P
zC`4?;Hz)%VFo`_18r7U3Siwk%Rxddqh{1>rYGwAMl|4hy*cL4<+F{5jP=T(lE+nZC
zyTHt47UdrWZwXoh!f4FMAsLTNr_U%;7u9az?{gzWmOzRy@qz=plzE09BSR6b!A-iV
z1{Nr0Qutga_9)j)O^>PCfcyp}j{=7kCRKw&muqHFWK5<jAXxg*BKd)FD`oT~tuinm
zn8zqLCu8siADl$#h=B%p;euerqda8uKZ<kl@L|m7q~Ohv0275D*fIv_Usg$b*2bJ5
z#Q-5R0C9G7e0|*pSgC=*0M6j);Wl0r1vuRtV7Kz$*n87>EZcT%yzXXcCd$-AWGqR9
zQb|ae=P@KhWD1$ONh%~`hD-^WWynlQ#zdKiC{t0H=kdQC?&n$WTK^C0>-&CqJip(%
zTZZep&htEu{n+<y-?nWBzWO>ny}o!1JSOB7IJOgusdnx9T;yO|m4bF0I(u0#2DPsu
zj^yCyL!f=L<QG3I$P7Uc#OT>AvEwMvz@wi;dJSgvKiuFx6oJCR!qQ(PqH-HV;9$sl
z>9G_cnt^+^W8uMYlKdx~E%n?lid*j*Bt@&OqjVoj92mw0_??Y)gg#B)b`y(#U|_aR
z>0clG@WMWjGPc}-2T@>qY1mOtPR;`dbbhzKmi(UP2iXr?<1=BG2Q@5<Q3Ea7BJwDZ
zwVmh0*j_48)6vmUQCVXq>K{7fU;ALBt}mqd3YZ;xAXbBv&z&wxAT@XZNN;0}Ph8Eu
zp`xM!z0O7Fqca;j?Yr+?xF<?`gj~WwO6QoPmaOMS*PO!?$a7;m85x&Rl6>$3bexx)
z`|~qQwp)R?Y-{8W4#&qrmNf;uTdldoagJEp?Ag70eU1?&aC4v<SJ0C8ZHG=_5g^HA
z#~Xm}Pzfq>0CdrbZhOW51&aq=T`-f4?W@$yi*PGk|G^Fc>p-A@BGwLrf6@8aKn+4G
zG-NCA%n5x}bI!ux6+}hfy$wg{;F3vSx^9gPQn(N2;G#0rSVIO*q-7(w!V}z^C(k2~
zMIwYzrdPv|8am;gx^|q*#m+<Xq}%Qfx>{3c7O_cgO)SX-9azjLJs7?$3Xs5*)}iHs
zo7|6IM$QJ_cLTVi9qsL;_l0~AX#H29UBY1Dx`v`khRwtE0g7`D8!4e-=(1bgK2!(n
z)<mtHI~_#}k0hD>w*kF?V!nUy>o?*V0D2^Ko;J@J<QoL~H=cv}Ib=FP5fRRS%n7Pz
z)8&ua7^COKRl+vfVn4&!Pa&J=6yQmOoAkuT2T}m&k>=ng62z^&jAqnO<1#$Z9xorn
z*aXZ=mUw*x74}PE%M9nGyK@?ixG22!FF38;(9=7tnujb03HBWF^nPZ1wl8|sN22sk
zm|`uDZseWyj;Ms7Yn_5qdhfbO8Kzj5Ek-XQDcwVOeofJ@p_;98S&mT{?{XNSjx7bS
zgN%*o^zA5C@3*vg0DW7H26dq?OxzQImX0MN%!aej{1bm3i0R0D9=U8_b!Z8yUkTgD
z57>Wl(mM7Xu}FaDAI3Rr^$d_NtoKYTXJBSFxX_>=cNcC99OefEfG}Mt1uS`zwgX<4
zJ2AG;=GTh!(dW;e<sdHl@gA24?$D!f8dckN3W{UaNhn6i%&ZM=5zsY^A#9BX&_`{1
zzY+3+fsdWA;f0L(g0Ibr>fMm7Ay=9U%tj3Ut_h&Xw9FiGJZ!yIBkE_@M@Ql~k17E0
z-W2*fg@r=sszIi@Lu57zG;qOe|Hw!eRGRP^1sZ$Sakd_Fve5TKvWWu(X>f$EAonFp
z%W*^;oZ`39OF)5v?SxuT`ISsG<jlaR2eHpkLuZ+%JUqGxY#U0B)9Cu4b&CmLKqW}%
z2J4%^mzy^68-2MLe}T8MASg1D3A|f`#(&MI2QrDgnEIH)O`Hm_at5MgGbFbQCKoVU
zJ4U@%SdQ#RbR7|8JzJl2qBn%OO4e9is4da`1R7oq|6-V#f{BjvXc%H7X#X?=rg4I_
z;2Y5C;x~&N=f1#46^#sR3@3P~rKDuf%ilrQ1xt}IynzRfGlKA*ieCYifN1*JTtsBT
zH#nG;X(OwlfdLu775ff+Ref}HB{Q?iFk39oQmi;Jd#wE=@qT3xu)LU1Ur}KQ<^|9L
z*&Ib|2UrutNiP7zhfc$HB?mDQ2avp%S9VMMHKy2YaO5}%Wu>OFF?$>imj&!!*GTwF
z2%#Bw>rF;_`sS@$lRyh8tq8{Lh9(2toQj7GocmojaYDWpcpPKpSA9r%bx5qau-4V%
z`pY|c?WNmX!yNe2Wb$3zQr^4K?AW*{ctAi{RjahNv^FhKJBGsBI50`EEjGhHT%IjV
zE-h25r26g@$5fm5eL5OO?T7i)Q-TZG_qu{?SUb{0OYZGo+r`y&_0H9*!K#*?RI#s*
z+nt<j&9n~IJa{Ou)0}1m64jgX+1}4GpM_^@o>zMpYxr#XYb%Y@Q9*5`%0xCnBZX+$
zq+=;v36oc;TaIzXYS7<ekmr?;EOB-gUY>vCa*aOM<^UgW)4c-T$Bt7(8=(6hfU!|)
z^8xS!4s~nF$UM~#CnC0AfS^H~$4P%#Gg91())z8UGh0b%z{*!NFW!XK!fD=mn=~xa
zlA>;4IrA~~0;lY+CC_S%8J2$Ux`acnuynwoYC!W922^<u)R)%Q{s}WLuBxs+HpxNV
zfB(miAL3$SF#k3wel`g^2{g=y)OfxXu#h%3usoBWayVxZnPY1cPHk=)$$_et;;S+C
z$^Pxp=dg|^#>WFrDJ)9LjCC5fwqR*x?_22^8G%om#I`@y!=0)&NjS3i(mw%yqo0Y-
z3U{Z7fy^;%3!3~>g-J=jAjtx3yHz;nfg-@L4`!hJ4t~Yy7CdA@9Lh2<kLJB1JF)Wm
z3#$)8lf+?L;yPorh;#pO3>(%X7WyBemK45p@gh~b+V$%%hO`S?TUFew6{Tu|*j!>Y
z&V`-j5L3VapgCeh`m*5-P^tj)Gtj^_QN-mF?b|LPs+n)1`h8EYjmn%Hpoeg#ujEmw
zTYkI=L-(EJ;;j5{F7nQiFWTRKe12tHSXrm1?AMnchW6SNCuA;F-)Va0=rWm7oOQQ|
zdP-!ew*mT#{(knAukwLGfo}%t_-slNlJX{o?!GVx&+`~4{&bhK?QfUe%VX-jNAu*}
zm>%XcHg8UOJ^yj&xzJ$PK*%P;_Ee*!86(+6e#al9V64Z^ozuEesNsCwKBYhVzGevZ
zQ3Zu1hCx!yUM-fLs4%#3<Idn0F_GimiHLRB*hUQsumQ(?Y2wsEt~UXX1!A;6&#7@s
z4QDHGi=ea72gno|#Fxj~gO`u>Y;l3R8JK4XIss`G7GpT=kQyGt$P_C>{f&29&?4F7
z;^I=v`w}q{j(Z5DlEwM>4bPPmmC0F&Ar`UjL0R#W_m;4VrWQi4atRANVHp8Q#^Jz`
zH4aW74yL`_TFG+^%6ynI4~<2!Gv|_6E%IIbgvw(bE4<j|(ZT@4h=`?H>tK%)3TO+l
zu>al08=h~dvLS|Us3mO#2!NhmFDBi?_XR)((ft9;`cn9ZP!rTzlT~@Es{Q(LIeY5k
zN`d{c1;WC4zkZe!nrwi@Ap|Sb$1)a!4=FmRiFNHam#2SuTL4Tz_mgJv_s|f=&TEIC
zQeOORjbTdb?tHwrnQ=4Y$ws!gQn%7od!JpwilZ{`$UiTP>E1FbimMBq8IpRrpHqJ?
zx;D3CN~H?8ZVfage=7+UXg%$9*RV07)ImckZ+~$b%zaqG8Z16Ne`59f)^BF*ltHbD
zmbE2<9Rf+kM+VMuaB?VHYadVjaGHH9dldy|p}HwuR!Y9QO?W6wcMW*s-#;HM{_cH@
zYTWxtX81&(l7G%*x7?@K8nV+j70eYx%pj-&Bm%5?2{#i=Gy;_%mBIdl*hWQlX~pC^
zasV7O>{$uKlM!1Tx<9N6l3e4cL_ty)euf{z&g|uplBgEiz24<hu^H?G$lJVJBmeyV
z%_1F&9@A@>V&T+>bO7KGvRVVw%hS_eO8O)zZQkK8_wC=W#*=#(JDi!hA2A<iB>{6p
z-AVm}6wG}({^v+L$VMW-?p%XA%FAy6CPgU*nl!o)<Sg;^9bd494i4xS<~42oqxczP
z8x#^=)|;C&4WiwKuI%+HfSlCtH-y6t4CV_S9v&to5-9V~zCrB~yTk(O2F8B{4hTYE
z^Bj}ljP22;Mhr@;rCGKlO&^!vUu`w0TH4=^MSHfg9=h&X`k3!?tHb-$R_d5?v%*V_
z7yWPcc&i+4wu>p-zsJOD_v@$^pJ~-!T#{`>D_>^)hP4R`%cXFoxO3xKO6JzZVx|?9
zx(ES>J9~o-BL!lSZnu|Mg$ACFMxT)hLp*_aLt>7mi3QS2WVQy#tbkq<sc(BvXChkT
zN4S!eQbP6n5#E1&aTRDknSv*Z<69*q9q9}3ZXEYZNEKnM{$prp2rgw98-?Bh+V*23
z0W26Mh*pWxvMo4A$R$r?yzlAmE)4ukA>8CW4M}u#OW7sSooa7P6X;ED_m+EO(2D$>
zPZu{cS{r$Pi`Ln&BhH(Cz3Zll^u%w`55>3tReGWHc+#I_CB;egx?|c+!ig*KGcH#;
zZbY!v#(UWr-k%oimkNq(QtGJ-T}x?qxItZdL^ji7S}=kuB5c8&Yxsb$rGWM+*V>HX
znoEUYb@nn-PiD@XdGMQcP~pT%Vvx+MaSOCqIn>`^w;Pf<fyguoQ3qO7WOrCL@(fvN
zHB+QQu&}nK+`1LYNh91eBxaHUte<h=s12B4oV?Eh-m@!EE1hcSuwR<bJ|7K2Rc6LT
z%h2~YNo>rmu<8MtyB=t5Z^u-^G!QY6Dx%uPh+byNfYGtBeUMi{jk<zDA2(B6T>MYU
zKKpmKOidvVhLp+%sT9^#7QzNrG6a}_tCN+}x*$>k!Ke#~SYb67e{))!y%!q^j9hq-
z4DLjm5PGl=b#)N*Hh|Uh_3PKVItc%qpi`@)JC1>gm`pZz(czgFP~sv)%vaR=3*#;!
zMk(4JqgsCu$}LP<;7!g3*AJ=c+k5NKw;)5ReyP}FFXt<C<lC5_Ew1(*C#sG1z=c6%
z1BWmD98=ahMd5ykgX1?w(SDwc0}YOuYjS+tFIfZBI162zmhC6F25vTF9ofW<E;||=
zuaS@dfha=aj#z*y8%-KwMT|x)UK5DChK<B}qS#DysYcI?RbZ32HdYr-;=wb+F$d%r
z_D-S>8Fa|S$ur)_Mn%Evq|Q5DjiP*1W){kvOIS{Bzx#GDKDO*Tuv3VNwa9qt%VfZD
zlL@}0&g5H)>yGjo?W&i#UYr?RlIAjaJx;YzuABdmz!#b1>w}MZJ3gLFmhum~_KCm$
zs8Un5@$DI%_gbGjDHKIMA$;J>Lq)y&EZWbk{nZEXB=S3()&JROANtX@&<3Ngs~6mv
z{@%Y1f}}no3Qjd3m-vKGq$9w*(P@n-;yZlJ4oN+(7xOHW5|*HZTApmURxKuBy=B>J
zt`YuA<1Bm_L`x4jX+!QjtUy>>4s@dd3G43Sq7&s<jvw5yrBCLTlD-V*nv*aCL22=;
z;}ZY?z)W0RGvniGV3WZF5B+ttCvyNnRK==e6lOxpB^~@$2^uG$RN_Y{$<|xA<JP~D
z)<tKd4WNG+A7+e!rZR`;jXE94J%>iFjL=n*l_ONN6ggPl#h^JbG3_G6fcoQw%BE@p
zd>M^X6EtkbpCF_NRI!~$4&Lb8O_cHW4zG+7M{&$HZSjL<7u^)MAifA2<$^c!1KNZq
zIsBEk*npjeh9=$-F1HV^D85q5P8@%&T9UZW?c3kqM8}yatVeYcD@6~$+O*Bp;|0CI
z=Jh)^G`xSd<E+4<M9tHnY~@C?oZ_!#^^bOa-jzsw!s^7~7l#KfePl>?IyaGB`ztk3
zT~F2fLHEhub53<pdql7KMLz7HZ$8(%KjBjQzP!wiATxW-uNF2Qxvu_2{&Au%CklOk
zcm6i#cI>dwt59P%=F{$Qoc`SQ$Y}ST{SQvj8o#BNyGBzdA<Z0=nY++4EWSy^c>Y#H
zXpPC6xm4-V!EausQ{G-@=B$ln^W2NmSl!MC`uzG5>lsv(6d)yPsILzqiM)e6w44zq
z3?9cBB6cKs8KKg<@wU7v=mlj2Vvnw)35(G8P%y1nN2&)3be<lMY{zED$dZs5J6Myp
z)6g!1`Zh6O2G?cx2M-2jSiCG@cAmZ4!NEa#hSui$@G#6x&kkRPWss3tx27i~<quuP
zNq~qDMzPt7bMuU#AgaD8Y<E~)l-TLbqS6~IULy^9V%P9v)RicwHNlU^wAV9HN#4&?
z0hHR&Enw6e$bY&hfe}ZnZr>*IUBC^H7C4Vd^NexnG#@^5#sNkP#uL|1O8d)(83vf=
z8J|Hk=qtNGXB`j`0YOES;3GMSXc@u<wXr5AZT8E9lZWy~^Je*rMrFfwja%>BPt@)x
z({o8)Tk6KOnUO-{->ZPt-qqs}-9M|~!%-Rs=WZ+W#;zZ1b-Y$5m`d(C{vLk4Wr0OV
z+WFY#^OBdUL#<3gQ*NEo?Y_^Hwm93;P|_pU`z7;>6_o(nJ~sYdbF;j{*byHmMceDO
z4Y@C_vS$yP%#G9-GKD*x`F=-B`3HTN)giS*k2G>LG_!ZNbya#@;0+4bnm#Zns&rLJ
z!?^QPW%vc#ZPX*p#^xqxUv(Y$8dl4jKlbzb^br5~;Uk@6E1y@srMPV|+R~QShNYKW
zlzwNic&x`;<0yYv@(6o%lnl&GDfhkTS8WO(nSi;eX)6zBSQLP$V@K~2V+d`So7y?7
ztOPM3^K=-$cxXc^&7O$coB=u2K;+P3d>I>eJW}oA#rxw=ev$m6ND8PL9Gq$#IEs&~
z{L)^LySiB|$1Hq0Zk%~JxEfP@pU6xRFX=^z%1qO?6HU?V`2cC7az`<8T881RUpyLX
zehWVbb184(6gUeySN8d;f6Psnl`RE7f|C59`zVgq<ma6)CxI7K<?X!kXg4z)j|zb$
z0?9@;kMIGc7kJTIC`B`zNddKd&-K8|xH!q?#!0E@{6Ei*H67eRmW5ZII|v)cA@&}l
zeU63t%Gtxb@zrN=mMNX7`TVpW)e5VW?|=Dn(Y(g~{TIPRam;lIi@FwfX*6o9>!#>6
z8+WigB8i23=DTmvmE<J2ur?ql@YK{>1&cSAjSTxfQ5z~1-i@*0j>#&^+)<ZE9mJ*-
zry=nuueF@);QXya${y=-jx6ij2aPJQ`~L8L?qsdh8Ck$RY*w75s?m1uvoV9C;KQwk
z^brS_*SomPE+jY%6~M1(P?%jqEEP}iWbLlX-od$&$QsAA0O0nGTG3*jD@$vt(l<R#
zAn5H#DWD;Uy9qO==)qFb@$Xw&UNiJ}Ir<^+2l&2*XdJ3+fAh3}zhi=?ETqp*(U~$<
z?W4cT28$6?rA=DQtgM%3_C|;}CFWmE@zejAoS;z=XaO_J1Pxhr3HscvD~rb+&||Qh
z+YPRBllk351(#Ff=yKo8UolccJ_0ql69y`xpsw)R=bv*w-OcR+jw2Y7yheI}bn;Q%
zy#t$3o69ey#W7JzH%v?@x$Tx;s@8pSO8P2ZDFGP?3NCn4A>9BWuobKoi*QumiQ(V{
zR1!pUP8-s39Ec|K?{wBwWg0hgGb){*syNp`bLx*mom_(4?hX%C`qdqM|1Vl0S|+lO
zSq!pEFMRgGBlK_`#Ad0JAMf7R_Nh75C^il!v&Klta_dh$nxooY6kd;)Ppi$PiE}=p
zuDw!RGt~8}NPuE0yqKv(t~=!7&h^r&N3BYH)5CQA`WMWHX=s|cwYAfHZ4ET5=W`Rx
z<raVT$GmD~t(u2tm=gQz%-0TSFS>rwgoTwa{JlZ8Eu)>~VA3-6o*fsJF$8SeA8Lh`
zZ+6qaL$()XQef0LNcL474;FEL1Dmq4vH1fUEa@MEGW|;Sa1j?>oZxr-8@FWyx(ibG
zA>|@lIRA#pj#@bK0Q3%?IRGwG_K#{Cxeq(WgB?Bv*r|yZOBwvSj;VskjFdG)1&<sd
z%q&5!p(FNZQeis~>U`wLx;EtI{dkyL9*yJF*U(VOcv)s}iSOke^DySk=fU#P^31Hx
zJSJXTqlUrVzX3o~Y**5Dp9C!kokbCm6&HgqC)dg}JU=bnpZF&@M%urjF*iQ`2cTEz
zBgU6IfL+(u*CR20r0jt<nWCBFG?=;A&k4o8*pz(>Ibz&?R#xLh$Eg28yS%#Mfw#1b
zu;g5o8nx>t&7DG<H}Kl=oUq)vnej@sv75&s+SL%<Jh*Ajux?W$15fq`g%P=~xn>2u
z@n{tn;XqoY$WS&5w}#TIJuf)k(>;3Zf7Sb|RN=D_|G}FxJ~c99IoA{Z%>}UV=!tzC
z_(eqjn&O;a{|EQL8iglV2o-zRi@s9G_ZtrEmwi=UVWf8N97lKU&GDj*uTrn;jqYDR
z4k-1^?QiEiwy)3)g}@#H+7RiG6J%OAK*mAPE)dK3z9!1)EpG97@&w!waj+$dXL~$B
zct(K3<TvBCTNfNllmm}Ph8pWXk2y3`5|DYhvHhfJfX8szQPqlT;sACFfEgzS;ViC&
zn4#i5;&Pe08oE|?b`Q6L<D=N(|K%kfWY0UiQm}rCZB>rZE@Fwtq0FHB9kYnF)Ya-2
zH%T^BDK8115D#OCPZ&cdAt4s_FQzc_IhdM|U)PTJc=2MH<_bVHTQGQ7h6Y#sHGZn)
zdFUOCn?L$4-Tq->%=4^&yiT9N>-4_sJ|WtD<SAriULKWMe|Xd1-=g2<PxL&Qrr4U#
zq4LG%(OLlx(VeEZt~@ug4?1K)$IZ^aa_iWJysn3Kb9q8}k2R_a!?YIO=2P>RdaLjo
zCKZIKqo!gB7;#E>Ehw@^X^!%daqnKDuFJlSw$>k%Lm=!-j*a<he0!e-E>$~fF0f1o
zoiCf2r3M*d6Ii!dDtDifJ};!vtsv@%Ts2^8TVEUCn*N%h!|*z_baXD}FPA>QpnO_j
zZg%$ar5xjK`Xe)+{5vmR*?AujoJBN5B11>Gd9#hwFy7CX<Jt2)M^$M-zg_HjF-^y?
zf4}izF(Fy8o%e|=5<BZNJDQq~$Fkn4M+a24G%cOn7MzG;jTAVA6{var3ar%7Wh!-h
zLUD6cM&=OilIYI$H5+z_agxPMksLsIt0cwk*EU8u?GD%Y6Au2IzW%jUijk$qZrn1z
zD6t$>`QfwJfbvJ0xzk2-S7T+v7j!<dG((8ALJ@KS=@zs=D@eC6w-M~cSJ=jhIWCw>
zf);+P(^>PwAmDCVzn<0nU^7VdFfKx#O(qiHxIiHhkQ2cCY`?Gm$N0GCmlaWExQl#r
zqERRelRJOjbvkWtzIulKqahc}enRsxvF2~ToaF3I#ULRmT-ZRE#SOpLyc>qYGW8#v
zZ!mv4^!zX<Cw6|rm#n>kee>7--0k11T<PSjJyPO-c{og|5<Rd>bF+krG|UCttbP4!
zm|(8j)Rs1XqC4;z((&l}O*=M{%H1vd#GHFXWhwn*(@`Czw!HS{^WlxAez~LXdbjG$
zJh?S~%YT23$?W=tzNHSD>^f7YlWW_$9wud4>*$&2u?tXf30&|g%G=#A&a^L_yJ#=d
zQ}JlEv!Cvh($?uy$<T)jh27qkLXO4!S8euLnu~w~3<qgqL%Kl}O)mjZ-El`J%*@!S
z9wUGebCU7$&@m{o;^<q1v10(;O7nBtM1I!$7_U^kxKel7=g}hxDJfp(y7ltA(uALf
zhc7M8S!Lh)L<z(>4AoH5lN0du3FcZzleoGL7||mipJ!eM%;pg&tGXO*6PX=a(S!q>
z6*@>=JA^;*o1m%=T&Q>;4pQ+~xJIdkqga)A!zOTUuxO@HS%6PGEKaf5(H<i|v$u_}
z-?m^b30LYoqc+XYeMgjK$P6d_U-x23&k(va`!kv>%*;4BuPum#8KZ6f$`Pl&p~VK1
z+|}_NKWASI+$jE5`i)dVPs3wLGs}uKqtq?7@Xz!cyOh^qWBlFSOT~8b8*FLbkL(H!
zdv5;ni6KMPSL&`i4(hkYPK)mRlJUi_`=_suf2V!#j7N#jqR~sO@T^l_KK^&`u>Vja
zGl+NOD89jJ{<5z?%Uoao<#G}4Rzx?5EdUSbfCXsJjBe&!#3nbkoC-_Wf<S^4R4+Lr
zj|JKcfVFTFdeMpMf$ADKAwRhH)Rs`b!K{RA9<^s3PF-DHC}3%xK?rvtV;QID2Qu8L
zA;EJan|5)bU49uDpb<j$ej~5JZy@bg6r>*;vR!^cf5R0yfO49-rl~8Yq@=`7aSGMY
z<7eJ=kV!mOI)xrr_HL^zBTKyN9vbxiWGl{Bh42j=UM`agV2##LE4UqoT5BBgx%6ST
zLOi^9dK&|Hlt6A_!LWLV5ha^TlS|!necjHD>If<^0<90eAiJHB<6Xhavag;}*SH>8
zsib~N9Z^ir*E|#0=Xkitc*c(_jaim)jzjL?&*Pz|x%h^)Ud)>bNVA{fRDSVX-%0V#
zsXG^jc7!x4Tz?+*0XP1QUd49K?md`I1Oj>ROs;;FU(cV&05=a<8$$qdG8?1;>b)#Q
z?CE|(Uoyu(UfU;pH*5U$P)}Q2EWI_Tt{eWazX8ckz2X!iDe&!pUgef#)tFw?5ukbW
z2uK4B;8$D#C`%{en7ek<^v{E@nWm9Rm!0KBV0lKFfaJvN?XRI!#35s5#6h&{)b`$y
zxOS}@15MEKIeI1dKwgvtz-&9dr+B3MJF7q_!OOY;Tvz++7nW$6MxLAo@opkQz#yc%
zN{gvD<-j7f%6<oawC5%tB5VYm)YzTm_wg3Me|p{lKhwHJ>3cp}1*;}EF3*LkiIqlF
zo(zw&q1TkF_hve)6=F$<&fp1J3*E;1Rm(D>5rgJrXzmj}4Gec1-zlW>T-9xp+v;I<
zb3N(o$^^J=@_gF)BI|GEb=kGse&6~-1D5q~ogzAu%DGEw3TB@(gXTw;JIu^|%!Jf5
z=6=Uy&cAW#iI~4y0fT$?yUFz#$25*FTwgb|{WQ<`TUV~FLK;V!ucV)L_J|b*-TLgx
zoKKt~nvE|Dmp80m|B(N+oaWRw#M>{8pWwFSKeCO#DWz?qNY`w4Vnbpr|HSz`CK;H@
z_o>Vak)eP~4>D1op>v7u>;$9u#fzp;jaJYE{ynRd-X3@L5^O)eIOUv)MSI=T=Pm?`
zz>W|A2z-JzJHcRZY#>ALIrD?C^euVZL%{QuE+pw^JVvB|SxOXZu$2NL5A!52F8~YK
zqfLSxhk$iTH(#Cxz&C6GK$irs6P7!3Ab=uE0bYz{hwa^1;<9q@-kPfDO@8XOPSEZ2
zd@}_=inF{B)_9jLm4kT$V(1ilK978;H&qXqG-Sai@j%2|S~|K8uo}=J!Uup0B7z`#
zf03u7oezQ~B)4?E4YVK4W@n1cn|InChA3H8wC2i3faghpS?1kZCHxvX5ab>^eOhI%
z8}x^h`09=UHz%=hX8@#5<4L|c49Z%r9zxilZ~w;zrAi>Nkk#wDwp3s9FZ(9-HmlgQ
z__RznD~WhN)N6|;f5%@H3GW+;7}j?_T$9fo)@6J*@Pr~3i11?v%2a;4A5X(SFf_)v
z0`B?RVqJ7AE^DxdWh`#5G&`U0eCcP45$}nqNgE4m6be@C!%FXC%J1c`b1m2q8D?!J
z(Zb9pl$9c>tD4Ld{}~7dNJ)b~e#9#|gAa=S2kXdd=tc2}K2*Fh@9u(gojH3_R8$nY
zArEO@f@NO^&?Ly5U!me4j9^hM?!8;Il@9_+S`b`8Jm3?P#qPx2R9seeoas^s;AI1U
z2>9$yDyk!?R*>RU{R-M!?r1D3Dhe8XwdY8MH{G*{2oa!q(%dg)STVvF{H^7AKmy9@
z5A*;CR12s=C8+`^K+9Z|$hsD$m-H-ZVL4S>c}h_+RRhYND%2n?zh1%i8^;>qav1j`
zh4C-!e{};~kDffrI7JZdL1KU>@rn{VPE|5ANmaXj4TU@DzV!rO+XES)g#VF^n&2(;
zk!hxAH#AkFla{#Tc^%+1^AEG&1ZjVO?I<WC+BdqQ^@>_4Y<eBECy7Rpd{)pnlL0_#
zy7-B~uX`Veatn@$pvvrCe~_Ks1P9U9y&S7bpz}gowroLTB9ZP(igt-sgQb9yWj{>9
zHZwj(99JP&aByf7VSvPmjUFgbAItIk;$KEbHTep|S)?CX6e~#$@LvaXlh@W)RePkF
zS~i>4($F!nyw>sLBPRyBaq-H=y+bY6)BLm|-!$169yEDBd)f$`r%#oYKT$uxCXh=b
z_o<)XIJDk)TGHShgNs(ZEgmDa!Q<i_`wlFh9-O1DUhtpLI=a=FJBF=}4T(>V@_irF
z1sJP}nYtd^=;Dfw0^d~_?#hOt90X$JA3p38n(4zGBx&{JBs~VdB40sUe;m%F=#l}x
zC3ymp7$DS-A3m&1BLD9M=ac$0?X&$$6X=`T!n_az8$q_ldxB(!)k4f6ac_a8+m!4w
z?u#D)9<9MG#LH`iXNVWV1mZuCp@EYVh*lWtCiE`QK{&(1Gn$gqHPrgL7fo~hR<R^d
z=Zw1*kj<7U8|V=w_d*$y1fI>#I`R6k{*9EhI&c4;pNH4Bfr{%zXBZv;;Sb%r8$udg
zgqJDfbli~kl7$c?rHMGeV60NShaU(d;dSuZNXP-zYG?AsYUBe4=97p@C%TG(-FGm}
z3y8KV))a~&2%!W-L_~ObCy)w407FShiJ~L!#6Ck4lLDkx#1quGd4DRr+<^@FZegVK
zBxq5MuK-waJO^Dj&GjF%<kxObmZ~kwD(hwpwtRKpNcB72BYW!1zDB>rUlMZzzxg+r
zOK=FP-aamJwH;4`G+D4_fvOng^H5@ZUd_EBn1&%s`UW<S>j8-5E{?}_Bb7!ghBZ&{
z0jT<4p^lii^X%H?%d~>luhiUOKPK9(C<Hi#@LU9Y&LyK6k{H$&0)mrrSMbBa74(wu
z<(XuIk=R7hY(Y)ido~*-IR*%`(HtY%Y$61$kqQgM6wS!U$o9KAxWTBUPBls8=H+3}
zOLcGg3&1;0eWX_71GR!nfi&wONFT=Q`5Cgf9@vaZ0ij@OdKx5;*kNsFdU|@6E5SU_
z24qgx>L8Oqqx@N)U#mTwpx?cJ52<jh#;isG9uhsI2De^5pu82TfsOz!@Zqk*<6py0
zT*Fc!eymnIvSngUX`m2;d;g0!%V59Oz-$4JXc1$;`vVuj)j}L>9+&#j(~|*9dd{j`
zp}i%m`R-;h7CRmeQh;$>TqY*t_wldmOOe0SOPc139=E?aeYs9Da;Ko~&2>|r$%?*z
zyg7u6tnbzg-g(ph=IxVf-b_mqKeLigzSlZ`ko5+A#ICUY3!*VQOxK4k?%26rbZ3TD
zW{&pP<;aQsEph1woV{ZMxikr5ugO}u0ty*;1FZHcyjN8=Lyc_Gzr3Io1sUs?ovi_S
z)?iW9$HpaK&h{uaT;Q<;b8k~M#36Ly)VPcQpne6WK3wpKZY5!hE01MgB~CnB&?y#7
zh*D>64*ljvaNsJ2_gz>FG)0Jwb9#i%pt<>@CEF~kJ9oNJi@}1YjD6%WB$D4Tmpcrz
zy@yU*7#R5ZGZzmY0t?MM`V*M>4fY8-i%Yfq-6R#Ocvio>Cswm-CnZD7HE!-S968{}
z4J_PBy!dE0$jCURFuc4$uC}dh4r?GE^Dns=c*MgJy#%}^FFc?GA}Cp0QiArx_AOhW
zUe{J=Y4sBY_ZI6~QOfkzEr5rI4VjYf7Zo{OzFY|!=6JZck)a_!=xbPI17Ti!lE1C?
z<)W6dDOCRzgEza%oFpZSKiW6}AD@4G?NGruzwk7U0^Yv+^G$yC$3N@;9k^EC;V@==
z+BZhX&iD_TB#s(>jC8w0`S_~WsR6|d%Z$B?`QMhlj%eFGUwQFLVo)NC>()@#>jveD
z63Hu4A9IQqX_?}E{~2ouSyw(dD}W_~gK6R5iV%WbWB}SVYR9S^cQ{m_)7p>iGduSq
zaw15T6Ho#Yk$l?`5sW1md%+WW`pg+Le>qN`w8rCs=Lb$l2@no-Gp_>Im~U|br;JQq
zM5%~W7(L5i=uc44sAt_gj)G**_QEe_@a_d}h{OxIH}Iva($W~x4ieybG<^{Ijn46V
ztu(F|#Q{)?fQqQTd+**mID`xLOaXgC8yc<w{P-!Fr^tf?j|P8g0OrcX6#4TfF&zaj
z7e$&SGefVynKLUmkio?(Tw4JK7}B|AsJhkRXo9pGG+#GB@?ccpCG+z03FQ)@3m3qz
zYd%FLOdtikC?*Ckk4R9#5Myys+}g#E(j%{Xg&PGI0%-IbaI`c@Z~0IbroP2>G|gqH
z=Gzm#oAo-#KG9RpMXUw=hTI$+7oZiK2E$NT`0IxcvCtml_S(aS8o*`)$k|`btH5w4
zlY?O+0H*79__88-gxvSa6$}ot#-l+nY(jJ(7>;39Ra;!uwHS<g>Bc#@2x}2AYVZzr
zXpDB96joMN{>Pcx7DNwtlH|K6^GD-C3Z0kKvR^{7@2Bp*cNd8Cr$JsubcAg?>7wDG
z5G!!!rFQhi8X7Xu4I^}12oJTo>p7f}Mm$CH>B&b1A=*6zMiV<s4h+Upw+%&~9EF)p
zt7o2-FsYd9^ip^WOWCz=K4p<$@tOODOWs_~=QHAHM{#}?53iiQ^hrdp!^WzGk<Y<;
za!{ouDOKEJ#~Jm4*~ZzVy3_|>Db6_0<eulLwcPfDG5pNTPYoNNp<CU?3uf-NNqNfb
z^2t}&;^kEHRI%W%`A05ewA0z^%d<=0M^ps_y`8hE=s4e}zkmNeJJ~P~`~jpEVe!gA
zyXff`L4fTOzKY5A&_)?7tka|lzxD^+nsA{xCaG1kQP9jY6kHz1ff|^|`J)r!z5Yar
zdRbp#Z^GqPzqW+*6=np!d34AMBaiDn$Ss5?84`&TwsAXI1)oK~d->DR7>RjQrPkDy
zSFDj2*tamkZiZFhZF6j~RNjn1xB;a~3jdC7A<YSelyu}rSoNQ+H%}=3?W&8S(AfVT
z^b06kOcR!=q_HW`dcYxpREo1H#5L(=L^(L6uRb&Yl=V9su~FtF%0IUk_D#Ggha9$V
zVZ8Uw8;+hlU>FdejYMW}&v4wP<C3DVm=F{-v-Od}qK}!a;6m~0ki&j}IjKs!si^dQ
zBi67A;z&vT_a}GV6m|PB(<UlfCdbjg+p8-O$RQjL8oDM+2=7gq5OtblOPs40`z+)Q
zKpj0)8P0te@BpqDM-$T;F*uhRP}e3!7*V$o=TfS>eZBYodJl8?TeZ(nYp1|9MsA7C
zSm7C)1)@6$l+PWCsP~IaoZPr!10X>574QkhkgvD9!nxEM*QxZma41B{mpGT6i>_Es
zJSX}EOaM-Y^tZ_h?@NJTE2e3Ebq#0BeLP^)2_Xp>!yI3?t`Vcc`uzWWkTvIy65(}h
zfeh{K{cG`Am4$mPwy&j9`UnSz)vvw_BEv2aNa%fVfp>EGm#=?coo!sRW^b9ns{=jq
zP$lhMwZtR8`*a7<I+Cw*B19X!`Zc%efJXj_SAvF`eEsJa|M%PDQ&cKWiZm=)9$jTv
zwdKLwTr#Mlyd35@)t&ob?Y9KI?~$5@m8&j=ldnI9uTAzpBo#e!@WGE>&}Z<qO$0Jb
z6(7_AU}C5dH?F9+m9>?%-Zk0Q-1l4`&n5ozDDF8D73}KQ_<vbJR{kITA=9eu9e#e#
z(Ugk9+aaUaeu70rMC(%*zIUzw=no8g5`zFcARHJ4tvZpG0YeoQ5;`etlGa~|<^}?M
z6{RMulkWYEl(tC{C$88>m;lP$<IyI5ruh2Lium7G#D8|y|6B1y&wP-?vTsbO+9hBp
zL=Uz#HY#NMA)F&ck&tpj)rmIZap@oUC7qLW?&oy}Mab{FfY6a)(!wR5eZ$0L0X;D=
zz(Ny6@gv$tt6Q>}a9UYRQ*bgJA(xy?u^Ll<eSwo4k{S8zQ`SFa<oQT&_9vcQYjiYm
zNa9#D^ov7%Mw}!4;_ySAW2*})`ANdYY{=j83JStrvf=x||GZvO1F$#rA9_2E|B<uC
z_j$IgZgeGL`vB14(kVF{IzXg?U!EmDJ-GVQ=E_;NBye`wWMABu5Jc=>vxm5Ft<mu&
zhMAD*=&9)#D&pI}E|O6HR&rnFd>C?H;E-2Z5w-sP?(Brh;^xBu$q;~sEvw#_@~hED
zCzd34>ADx*KQYKShKSMh&l?m`N+luVMOok8x9XBAv3jheYau~}$Hf^Px)b^j0Z0hU
zlTX#VYC4*(g{<8e+P7ly<En@=Z$tn2w(&d*rhF$g=UV9Tu6mJ{<8;SGG~wp`{Pr!%
zVF`HW#Q*q-kdP2jh&wC+xeiGquAe-NtIx$Gej%WVxQgQ|dG&+&uEtZ9Ga#>nyQXQJ
z;g3&rZ|a}7_=8Frj&~vEe3%Xa51z_;CcGA{8!=-&w>GNgZ4NR`3}lmi@Co04OS}3E
zDei$qiSCI}8ZT~}{px+!TfKbNHE<&YcATvMdlp(AoqMLlbM;>7>PMiWY>!VyJ&8wO
z(@btYt=H<?l-9#L03Js-?*@g#PG=3pKhCBb+)`5hU>*Z=<T12KP^TLV6H`gb)k8b#
zZT-p3sep9AS$&$fdi|8-`tSV}0}cLa?}nG$;;1#RWm&0$hS!Y~CNNU|)syG#IKn=B
z;}uAe2u%lC*Bnhy^Xo8op_c>vSUsPnGO{T4%$jhFP{LTFvM)t=>t<Gd{uv~#LBm+>
z>4n%SYWZ91w@i=^O}<t$i&D+hh<|edR=@k7EaU%55_WFqdSXvS!GI?S-56%hz$3vL
z7!g`Twl+4|pCTa7RUX)gp44G`9n2Hc6~<#^4fxIZl`L7;vNrNX$`o`m?tssRcJ8N7
z%Rm`KRu(5f_gBxi#q(udL%X_HsZh8BFoK-OX|!}x(C}9LjAG4dkYdp>H!nef4{%Q{
z|1iFverI(XU3m|_E<O+};xeqo@qL}9)qN7|1yBuS1Ck;mW#U_Xk@nezg^opKlnCgy
zqwJqWC2E}5jj7!N1&9^*{_<EN%g{OyItMUSQSU!+cL$9M<)Wqrv}EEH&9F!T64r=D
z8HAtUdjNGq_YwlUvWg0@f|+4Dhp=++VChR(X|7olw3<erGlx7C(wPy+1h||NaXaQ1
zwE%t-eH@6B&_f$vLmb2xihmt6a*T|O$OJ$ThAXS_fv8wk6L+4n+Aw}*Vo;w&Z=k0)
z0arRgM8@y;=syo7ldBfTDlpvve6*8=F?jd%6RY=#QkO_Wm%+-F_5fpj7`P<Ng{G(P
zew4)+93iM^?yX%nIeZ9zk-U3#on1g7D4wwhx0TTCs_4)GPuOP12|XCVg<?>;H`0^4
zYO)$d&Y7dP1UzbnIkLUjsX*Tz;rvKT*q-&esPB+IvVrxDVRL9TLSPs#!`e7R#jUkS
z+G!>_#JWCt&+1!hf^(D%NEjW9(=N+}VPRpp=WpKp31ZbjuAJAeLtu-g9|n;DvgcNW
z51cThfr%%NPz9ovghnPF2$m)PI4558ZBOn83Pi42L`CalVwHkvYmy>O_`itj${NLh
zLtw`6um*NKK&|T&<d0^LqsxIAykZs0<OKweuO4>mfgKPZ6EM3ji*yi~`8bi<8Q%ss
zq30bwZxP1JZ?xHoKWTPdb>IFDPZgBs1$YNoD@j+O5f4H<E0>p#kB^sEqje=Zsy62M
zaBDW8F2>WP#7!YWjuPjTqEP%&ZaNl7j}zUvV!@qTPCLj(Qq=wDrIGmKpdrANv_m|F
z>o>6ddU~!WDo>7%@+iuO(mrJl$P*<C8Fe!L^BTX8GLeY#s|;%b*!14DSlR*uGVw!e
ziMb~H`SUEMpTt5UVq}*5*`a^@!%qRj0!j=R8A{WiZ~+9TjIT)p(l@jU%K8Bmoz!(f
zT5_c|6gsYpRpUUJjLyo)Z;&SD>r~RKi*6iO1@huuRQ<p;i8Y{~4JrtrHJn^%<JX_;
z`6vjyNFoRtV(EpyLyz4w#!Za6xtbY7MWTrUFkO6{4N_cC2Rv(ladC5DY+J@>ZFDlE
z7o9I+e$YYHt`}lkImL;^Z%d2vc*3h!tb44nHlxy4)2OH%v`ph-V<klRV6<tWi}pfp
zItcM-l(GzmutCP7mQjdS9L~BHYXyMXF)=Y(|Lr0sr=iF&FP^(Jo`56V=69=X$|%yi
zo>1HyqIiCi_Oqww1ONl#YnbM|e*g*$zcVk|+uP0Bvl1{lr6(T7m~0^rsn&l7^9)gd
z-+~4Yl;Xr{JbJKs`}XFnW1T>hn<PPI*}HEaTdDvaFOcCs+zNmi&|@G5dT7N(L_|E8
zQsM^A*$$++PaX!Bq#uA$DKgMJ!rT}{Oi+w!LwLz-Txe1JJZ?&AkmsRELA4$*SH|Qu
zEIRj2w2PV3-T_{0vaSRGo-)|}HTuG)>}ud=jS6%4C{Sm?X6+6LzQ6zcAut2}7^YL}
zH*JE}NH%)_o)p$2dBez&Q=4z#GkTfw)&cDX<^0BCJklcT&oFWD*C^FB>+7Z1NG>!u
z45ImK*Xoc>SH*yF$edaN&EMe*f3=NUxE{a@)3up`fPjX^aVC7m`mJ-huCA+z^Tx-X
zo_PP1K7vc_GS3pke-M|$O0?sS1KEZgd4Iz@IB_`kRqlcb2OfO=KlyVmFs``xcqtKP
zEio8++9LmkMxyTH$CEBCUV&h=A`}BE@9flY@&N|<8{;zpYSw-}j&0!j<e%M!nj0bO
z=$rzfB0=1t^h$;m-q5l*R-Qh6nw52uLAxKZ561)}8a%@buR||hz4{62qBe91vCDr2
zU$}DlvP#xXS>RbG|7^yCUqk2Kcw;*K6!{xiJH$GyU2{%%8Xi-~DlPI#N?dS;2CYLj
zg~3tWc_yV;zGZ9z%*x6wSRQXMd=b$OT{X-N!GQ;%qa-FhNlNB}iw`3ZAe}I=vVvfP
z)2;F3l!h(N0z{!EtqOphNU6~tf?Fg2bLEf|25=KYX1xSj6)-Y%8AlQ9LEt`%)dy_B
zYY_7UcvR44#J9``wGO!a`&r@|&Yc;z*s$S$W^DNOf3Tqc;2o=>pDcyBR`vXKJk1Cv
zoL`aXM@3?t!SnKAO#%2mIJoDC^-;1ad=pRsAV~|}J(v3zh?C1Xu?6_^$xOJoR5?D6
zjO<1P?jJ$Xl70i^Qx+!W%KyMi{|%45_ZQpg)2#y~Xo?#R4pVU=tIU{-$IVvHmtQk}
z90Il`ZFL+C=!!|%Qas(^>t?V^Xw8OQrRzRccYR?%u!BW_At)?~%!n&UrO+UdIe<or
zAktVER)dpz>eMvaUWlh$Xo}(3!70;(l-btSw&v7L7L<?MFiLC3juIFIp%4a76}eeK
zfdc?^0w1C?1twoja|m4Wt|O459<X1LmXZpC1`!84=pKQdj~@$y0|T=Ph*QZdIzPW2
zFwO@Dk&xA*N=6JzK~E|ORz@<rgx;<pGul}fsaQATBsWO7Hj3A;ZW)4%H&1)Jpc>4E
z$E_qJB{QsTA3sI$lC-oPQdkA~{?&85<qbG^5C*Gh#iF&*6EhoCn+W7`UQ=X)=+673
zMemS_gdXqM!1xV1b4PfTfL--DHj>4>{tB^#bez2nkDfnAF+;TcEGTf`Ics^tIJ+F!
ze@1C~IuGZjC$z`Ur!0et;q4Ifz<-gCX1y-B(<f$#%@A(jmt_gu1?ga(?@;xfnAyg1
zu}?>I{SJ5PQa7VrLN~!~J^aV;QUV+r_YNMM{NSi~?kHr?vfCAk9cEHv3)a<Ce*8#D
zK_SNp#HI|Y)yZ2h-WN+@ppwKmT#!g2PFwV#2b`rKh1<+)cok0-ldbg4>aSpaY5@)b
zGPsz+{k%ghehbaRv&W7}&XnV+p&W|v_t!gD0m@07(m3Xe)7D>&hG_&ZP6ae5NeMI8
zObH^7Qc63ry1v0MnbEYu4Mi(F&;<N+2V;1x7_lm>Z$hHpWOfDlVj}kY#*@^$cUz8h
z3GbQO+RVzy34N>C(W}qvUpwpT>qBe_^AQ7+`YQt<q{24F%ZrjXo18=F`O)1mMd5&D
zlcZ*eOwv(a)?E>9$0a2z=#VE7v+qkXGWA%V0LLMKL_gFKj1~ygQJ3{AUY3?-0J{q5
zJ4d5I)VV`;_l7YkuTi9Q-D=TM4}Ev4-C#d-m$c1lkT>G+^ctp0Zoz*ccu#3L2z5s=
z4EasBg}%r85wrG*>=MXYQI&iS-dk3ZiLO!kNtN3mEOBA8j1XB5Sgc5mMBp(*D#W{Y
zK_fuY@EQphicV5@JUw_}+!#hUm||QlAi15jeluefjq4_>19t!5VuBc2DIa#b5%xh4
zweRozhTF0-|BE|jVS|Y^YF8wyo;Bj$`>Cm^Nudf*hYSxdJYNkDg9{ff7~HI<M`}EV
z>d><WUMd1enV>y@Fc1z8#mGsCZU;PN?5#=T6(}IeT`v4Y0~tA{qSU*0?;d}%^GBV@
z{w(HAP7X{|exOtv`u6RBDk<a92AhRoJdhccm;<~X=@~XqoW`Sn@=cVwblv0iQrtS5
z!qX$&K45Z)sy!o!6MBQg>BllROky<dpB!`SruH45vyNoS0SsW?p)YY`E`hTOmga_6
zOaK|txVVB!p_oKS#1`3DLpAkL0)#fWI0Sl{_oGLjtt*k`0Mf%8<UsRY>{R1SF+ind
z-0@Bl1476LmChI=37Riblu!TWyd9F$M63Q{nEPu!sfKY$eYD#+>%Q}UnG}o<)wIuF
z>(kkVjYn7NHU`yth4XfISgGk;LPiw9qkEos#{=ZV^bIh8eD?8{->aGi>kXd&cpox=
z9VxPgrX~{V6|`P|W<QR%fnmlx%%8EQf5DOs<D=muP*_`tr$yW;LEie=6zY3)5uxs1
z1Co7wuHnWLxh80K*kWXgAl|-tCMBd#UY?Z-gH$oh3G)<eF=luK^&Td4f_|{$>>|&h
zL+87>_8kT&EhMCo0-;&)rBe>rBoO`5wLJ_a>2CO$9FWuS8%P*|GN3Aeq#69`79b0U
z4j=Y5J<G*qh9!l}5RF$dP87U`R;*L90Bl|BT0plNz;O8=KOm<kv+e-_p!JA;6<pS_
z9Eh3LrRn;f-Me=OG^`0kxw)4^B}(ZuN(lT~A~m6APhEN_u>T~PY7L*NG-G#s8pN6A
z0K47mqY>PZ2~ST=)#fZ=6Gy_ab*>xbJniDvpCcn_N*T90PKh}uiPNy)CMz$$D4O_m
z2G`Zn&q_xZ2F`y!e+c#!V5zVq&r9rOzHT~S2zMx%LUL4R$rsG-C|44b+&%i`{;qeM
z^6`-`7z_;!6$Iic?ZgOIGgKW<q=kVu0Gapu)7sHd2MU#xJkao}q_!y0b$`q!u>9pt
zIY83a6p!Q0dRDsH<%i{KjH14}U|bC4B_1T=@y8jCeOB+F%K)zf0MSb)t1;jtMX>6Y
z-^#Ms7(g0-m=Pi_^$b%V*gGD&0eyH?bl07NdR+Kew|b}R0Z^zf7it&i=H$Y})KD|-
zEhak+f@iAMJ~lD2hiMAsaG|XNMCesFFmuof>fb-Y2yQya(X(ySz_<WsqXzt!naR+g
zE#?*;_wPUKRz~DxW=G8k8Z>mj{3dP#Ja>Mpj9~(OLCX>E6;NRk<=McQCl(UG$H&-*
zEIrR0JSe*OV;Fc6%V6R#$7zvaZ~*0sHM&~1eBB1>o4qNfv*x;rr51`EwUnXV1?K$b
z@k`quyP}Eh(ia7}!uioC&v!IJJR+&104}a@A3hv|pjqv1R7Ri(IOb@EN@7kzcX3hN
zLKcofL@9d!NZUFJ+A}c92t19nB{MA;yMjETFfw%c#Q6QQF3X*Db=+9<(No*&5b209
z2pwHg5wGhbk!-|-JD7vDK*5Im$N2M$4_HheuBE7EmJgI%$vE-{T&5<;#!Dr?J~dA>
zx!A#o;3f0-)EE#f0SMs)TtV$Or}X6EL-=adq7ce<>A8+=N6I9eHhq~Wq!2~z=f-3E
zGDAW_n4BhMaZ^DAjapJf=0#{zZ!owPC;nZVx-0-d@LH0El$xJEr_lE=Y?6_VoDme;
zPCq^|{HY-0BzoQ<x=ASoajf9E)V4<nz*6mE3PwNH%<3LkmCa&@L14xvRXjI92Z5&>
zowcoP8eAdOqGxRj(_zQ>2~lXA@fi*~Ku32OQs=)80@-8v9h^YbQz^=$3D36!bEdu&
zjLBySuWmD=*SXPqe__nU6}aq<BCUq+1g_*WLOiTQzBDwD**5Tlf&ipYkOVat^o`#d
zit!Q22RRpg1W<AjQDzVfLH#DF4+a2N#}Ts!OoHbvX5xML<O#BZJ6(Ekox+qAK$x7^
z?d<IAFmAz$!{I@D{Q7-p^te(&hVCO83Y|VZ3+|B1@}fOR=WtEd*Y88%IJUy%V9Q9`
z&;-iQ4e>+wTXzR0CNl40K6z1gm%@HVhn>{fwqlVUa*C3kB5$S#U$h%1WhbertFtOf
zZOI-vn8KWK`SQi-k0U;F!V}?!!+vM{L(cg3EzO%)4s`Mw4j1K3gw%CfU3J?yPl5Qp
zA93K(65EpO0@538`+jPt);|LSd+F$i#~UIpWEWXkS;(vQi^Rj!1}=yGK|xB_(<F^n
zmX{#nSyo^_im?*<Lyxh+D^6?hD8u;(3||=dr8@h=s0DCkUGVV>ODv!_5d(<8U$6w2
zz^2;$?3Z;ujkcp~?T+P*V9E?)G7p1C<WK8-B%;KF0}~+J-Q5WkjZ_y%Oczu{2#=E(
zyQS}JriUDACymHlpTA!qEEqJdZ&fm{A0nfNCI+Pa-yyi-d}KgvAE>Nl^!bH5u*HA!
za!|W?;mb3&67U?|9PmItjQ&8Ah%DOb?Vc+x1k|rSRYTG1K8=Kcj1X%c00EMx(+n(S
zci8%w-1)S>+NAG*3Lh9!^#?3>(x0}wy;uo){?7G(lZe|Nv1C^3wy(O`PEm9xnt#mp
z3#ntB<?x%AX$UANSU?WKkqVx?$c3`O;bG^&FU%HjMG)LIQ#g5;0c6Y=^J?%Fv10<0
z;GZZ0h#GEN4<j>rwnHmt2LDc~xQn$PKO!c^M>AAyS08(&S?t)xlNt^B7}34_Zo=Q*
z$~AHX-HfGfzqmQ_6@LelV-lZykxmrv<`1AeM5clOP6?qR;AA;nC|@6u;r3J?TS9u=
zvyoXm99E}tvAA6SjhtHViy_HK^pP_Pp%h>vq(hV?N5cHB+%|2`lE}#DEEED8>5WPK
zkQaCKM&gzev8<hp907TtteHK6ba_}~ns)bXS5=|Di}4NEcj7E`cO4u&H@UP06qP6K
z@18t~VNYAt$ef92cLH(=nLNAJ^?0YiPe}&ESXS}v`H!iDPCNHS9bUc7qLO0U8EMnh
zmT$YzCu7F65$ZBf4}0P-dC}?y)9WMl8#_0LopnK${W7-@_1goJ>o8ZHo&Q-$8EFq!
zlRDv<=Nf#V7Xoc1a|H9U<%UObdXY=gm25=bY}T4yy0gy-0j_#!m!xeYc{EHd-v$Pz
zYq(1)zbc<af%%@k55zLu4y8v$83ddE;0{3kbuQ*;$A^7f=7rY?4t%xxMkE!)#fvKq
z#)Hb*PBusZ(N006aY47oFEa=6_Cs=80NMt`o}{yrK$Sjp{8u#&Vj_?JW(;fraxhWG
z2Yjj+`vIB5z>gpB&P9H#On(~7y=mx|rb-Q(0ED34%P~GN!F*I6*;2m#8O#0$V`s2E
z_8zI&1VeO<Ion=HWga|ufK*vF^*l^(#(rKlH2i^$SBi<ACpDV@FNR1+pejHODb=0P
zD<vUuT*$fS(NT_@(a6jnRG`>W`stq?SW_vQZj4~7^Q%(y_P1}PV??X;y<{_E*=9zW
zh^IboQ#L4laB>Z+zE`M+wJ2_af>|P(&Y*wgcV-gF&V4@I!s?nD-jd8|-OUlWEt1`#
z)V8_wKuwjk?8R1c%eJ^lS~Vh2!}&1DMF^*OzrX`T*jr1j-Y}BEy?r*x!eQH6zRiC(
z;6gkOwN4%6b@5k1u4@phC^#~14ON4DboOJKYUTirL=Yu4)3k(bZD&wt6k{e3$}+wX
zr%OBU<HjdN-UH=c0QfIv_l=K^-fT_(j!?CWnwmNF#)*7gWQ~X}EH;0wTepJ!0`Dj+
z98_IUOcB};3h16kmLHPt2|Pgv0pm2(%hQ*RMeoF?S7nwLR-Xi*1&$6LAlED`<#y|1
z|GpaU>Fo`F`-o_L94r(9n=>zWi5*(L#{JAoJop&*)&mu@CShZ1EqC_wS?)~Uk1f$E
z_}?PWEVt_>2>=89l@+D)u%<{JO&&Y_Y<HSAkWVsG(V-d01G;VV6_7<#^UexVVPeXX
z8~CN2Xj=uOs9Jci^DL@G1O>%8W?CCPQG=xGyLbx;&v{Zvn{T$HhA+Ra5+657E4q=p
zZI+m4n8jc0(Zg_aPK*1-K?tS$udXXXJ7HgU&WkT3#Bn_biE?Zk$6W-1Kn=-j{a3Q^
zNcQY7_Z6BUwb9}aB8#*shLPFWOchq}J|FWhn1b7#^?Y-AO$|E?n|`I0z}e->-Me>T
zH-@MHM=ma{J}qt#ZS^Y^k-zFo=ebby29|j3E4EWF5>;od9r=MphU=p0cFWYS+TA-c
zim<TPM{~FTqNM4sFa!FFN=iZBUmF-P^rk?Dl2U&(=~x<a=O*ONqC3U5pLgHFNWYph
zEPH+v-Cnw`degW?x=5bxUw4scof6_2^UuIdt)|A4rS#!bSy!nR;j(PMgnWeYzK4f7
z##tqwS&>$;@JKSPNIw<%@(VrRh1<jq4-sXzgTX@v>VXPQ;oh{gyMBjHe?<2v5VZ2p
z!pFL12Lj1$5mHR~TYr*6Io0w;XKOOI@BM#s0a{hwzii*w)2cb!-S8WyB2MoU3H2~z
z=;d}O_o~}g^y0;#s_wHXT&%1=L3YQeY*n0`P~1++Z!XBo!z}F|?lfAnn9GB94s+K!
z*8^xdR?l9`v$|krhuOQu`gz~kJ(}pUHdG16w4joQtr|0o=degcM92ZqBZ%e`Z<{xw
zx;>uLHWxe>5TMnDJ6ty$nMsr`u{piHmsfOnYuJfihW<Z)Zn=x7D%!&aEs02yT)hu(
zDwTEHRN+GeDTh~atZsdmM}KiBVGX$(dUNad(0!H`nrf7<ruK^eepymd?(-RsdOUN(
z2>VMVlyjizgBA3nw|5UruhAR;zy|lc;HfIz6b5IU%@{yWl0e)S5Zkrzm^%&MSXyg^
z#!TZp9Au)}yLbPLMtnn+aDC=rFxy!Ac5MDH{AZxvM8*t@u{N0ulyx{26CyuLjML=d
z&1%AT`~7W-VGpvhzUME$Y{9%=3tgxiKe(4447Y28p#ZACvI|mFH+heH^j#Y(f3Tf_
zOjtp7T>nxC<l%z{qXE;*pY7kcrh1#FKRbVx-qglhqN!uR?Z~Vvz$Ei$(T4>yn10u;
zgAx00pptpuUWYUqJayy6OF<l}ay{-JIQ-UGNiJSK#)Oudk)3H<(Ux-cecMxeS^sP8
zrZ3$zsi6GV+Vz4(1;Rl~?%D0lNBP(O9W0@9M5)K{-WbQJy>+5d!oV5PqoLzw4Dj`R
zjew3DR7ARBNy=!`E+Q4|cTPMZJI@gD^y%okt=#K3Zu}H{Tn~pTTA0$;t|_bBy0GY<
z_8M!u_gik|<uR9i8~uW_AopfG_Gl)(`a;Y{ATC(V<;)v4Y(UYf+<hLkM6S$t^`Nx0
zwA}RE+*UggHUXS2G&Di8P4A9b6-gCK7tn&zub($2y_dA`*h8x{qx$X?e_yE_+Scx>
z(F;3lo~oA90s>1Dkt<X!i@iQ)y_h1;x^#^iA}NCWKrc38*~KqYgRc2DN4FREuApsV
zCyPHz2&CxI{8!lAeCi4Pdhu=l=RW=1B@|VN#;b=?&xj85jY)d!j_hgdFVMn~P=#DH
zC2dF0#li7GYIEerY<eoLC<i4Ex?Qp*7)by6U1qF+v2jo*Urz0pJ>5y>DF=#CmOOZ5
zI!Pv)qGO$O(oEAY=jXH8j$B1n7taPc@=7u4{^r~LYJNy?9b(yh9%KA+;R#$}_xu_5
zU_J`cD$A<l7JlwHY9DO%tSZ662n&B1>?QXZmW9uvU3DMZ9$%94VlfpwR$)QQonTaY
zvli((^6Vj|GXeq<T=hoU{iktAupm#Bar;*Kgn$C%fA=!(InvXsJ-T4z<#9IP2E#8z
z4$mA;?^my00iEnZ&lSl%-VWIGiRBHmxCWp>aw>HNlN4LGBE6tIdl2~DC!`|4-w+cV
zVDJgY^6$z9Rc)7!1P;8~w^0sKiF1bA!<H$IX-Ttja!Su7x!*;85j_?u7;Xm0{a8-m
zVv`#A6O66u(=XU%sFY_L_tD8d=%Y;K(x!br0)8f?L)+BCJ2e#i{5>W14w>t{w#@01
za6ag(!Jo27yj#DI;UKTo)S0Z0zVN`oF_hojOlZDI<J5c~=L|yn2np%T|9wxe((D$B
zN(OPwL+Sbv!^Q6hwZ_YDOF2F&2~W)X)pZBSej^G(gch@oH*om4h0EhE9Ccl03%@!u
zl$0jf#c+!n>Dlqlc40&ftO874;S&-n#1=<ay<Cr?amNr_tt{Z4h~-k;$|Ejr8C%CX
z-=mJPJj)ZB1Uw3MoaM}T4OxL$-@eOFk#$sUXa?WE1vc(?&MHYwr{h~UoD7JIt7)UK
z6%J47@yOR`9GANwV|)Ay<4MnFNLSGX0824x*qDL{e3wOPc~F9@m`zm$PCo8U)G1dj
zH^!nftdY6B^|Smn;U=ItOTCZ@07wkU#t@{MZ9d>=u%znDT<fCPEwFidD~#S1EI<A$
zfh*ly>SlQmXP+Q!1a2Y_$)(;#lli^gnBjx(9s#@Ke+yd;c+C}MZb2Y{Y6h`n2W*zL
z{hmM9=y6UD)^+DVY(+A_c2RlARu=gL^4h5#=x2K?v*)CJ`=(<NZP(B3SqtO(wmFy7
z@?$Q)CMG6KI&)v=IwDE%n_8txI1cCiXe<meK+%1ak;?-7JFupn<=&E&{fw<AfD*dH
z`j5-xP?J>cT98x|^z*75enYZvr!<8;*t3L<%#Pa5?)l`&8zZ+jL?mjV>*!cmeZESU
z@6BP6PBeWU*k{)iA^LqCwZjQAG%}Izyp;bvl^xkIvUEj_is8sYY97m-S=;`X8-#3G
zb~YlaVGt7n6I=*aWQ;=Q!GZ_63g8I#wX?CZw!XO)W_iuaWu)oG!)gdrxA}M<t$V}u
zek2QglWN{KK+Uxb!VpCuisrd;Xac8nqNA9)!+6!v4|<_U@(oj_iP#F@UCkT5&TYD;
zs1katZeWG_$!1sW_^m`kO>H3O2h;RDdt!KAOQ9JL)P4aKH>n~l&--JL``G;h5K!;6
zwZzi^cn_>pc{7|hwK(VW9H7Iew51!w*(SA^3h^SzuKnKu)bSi)f_B#kq*G}+D%12G
zNSryjfR?&f317rkhKUOpAtF1ZX>Xk4CFNTce}|<hV37yO4M+-!%p#EP3DeoG+t0)i
zr6U~2B`g>UT*j69Dn;wbs0NuT-G3zo9xI3J>JQtUv;WYx&V!uUPSb6MmUYe;rTL()
z1Wmj#L8bCcM$^`G(Nsbods)WR>7{JQJ;ra9#1qk>VtHO8iHBoLfS=zBo^%7i&_ga(
zT(z<$$a|q+o$`VDpCp@s+{Pw7k*?m#*2jv_hkK>{=Tj>@z1g}(fvll01Sf#h*|<{x
z;ZSqk#o*WF{>4W}BX~Nm=pIHxX?L2C4HhHfF}F(kH_*k9RFm;&2??m`765^xA+xl&
zNQR<+;N~=@qN0K?YchZdY-UZ6kRjd0oHE2xtPUdP0qw7%tn6~v-2-`dTZoE&tPMmh
z8ag@$tVxtZz&8rPw1T26!cN^H%G49P9z?93&z^wzQO6RSGq{Xm?8w#Ur^(D*9OWSj
zkW2C`E6sIU)g|onq}sFR@NhY(9F){34>(#nP<Vv~pgr}WkY6JR{5x)BETcd0=0K)2
zdX&h7VaJ`w?xkT$@NLci$KHEDRk>~3nz*+S0|qb>Oh^)p2#BJX5ZMYSh(s|b6hRRb
z6vT}fL4sly1A?Fg6#)S;pd{1A04gdHOdudil2qTsb5Fa~+I#h0Rco(RdsUZn?%n&~
za;^3M|D0ov(R&|#unBaWdyenICV0N-KShZ@i4o&!c2zAgpys)j*K2%>TquW!^9!JG
z`-mgR$e)IOpPWniZN6bn&-M)S0ZQ5z_3eGy`<|<BCMOF~jg8W{ad@mPSPXnQkx!S-
zdlkE6S5|fz(sLP$aduOfs(D}1(c~cbuJGpi_%Yi&Gbd;No;{;eieP5{QeCvpcaU#d
zWy%k~<y|R^CmfqrL$9OBXJZxr*R0^~Q6?ycg#)@|WprYqPy&nD&G3m%4Px5M=ZEl!
z(pDy4DC?zLEbvt^D#%V8)X_1u=E(9=TrU>bygFL2b75M`u6&Bc^zOA-@Ysw~Q8Bdn
zL1-#EF81uD35<o_@2-s_!S}K`kISU{_g{4P`=M3eCuvijj#5@W<nZI5*1OoyJHU*Y
zF1a_mTWwePy7?y6<{*uvm7uNrI*hVdI$K@PnJGs_&-uvD)7NA^l1Qa^T|d!&Qcva|
zeiI^rf~=FrkFUL&K&Q&6yvw$FTh*qOgsQnvg@-Ej)A_7AzpAi4V3%Lv9ohC$Uk`sN
z{D5$AVbUgMB*vHbA2~}e+94vgA(6%2)w{5~oPi?sECyzKu3WK#c|2*~9}IdB<ocNY
z_KMO2s}jD~)j`0;034k{qjdbZ0ACTJejRn0GDy=b=s+Z1!ri;NHT?P&ZFy~kj!sf+
ztcla7tX$9TKWc0J_U^R=BR=EVatz!MZ~ma5O>|Jg(GPlGudd^HC4P&Wz#l4^4c^+s
zI)-mP#^>NM05*(waGB?#aV_d*I*}U9rsFL4robWcW(|<Xq|b|aLs&oro%|Y3`Oe0u
zG{c9hY7R}*(XpJgxEuy2IAN^X1*(_R)wdOr{<JkS8!qo3`Tzh}anSW`X0|pq4&SAR
z`^hqx*SZcM+<85aVJug+QZxU8VW01H(+6p;-hap{{245KZo$q$bQRNl&^P7C{UPyg
z^)HtBPhMzdwp?Ff&b^9;M4%qR^0Sh^8wz(xu@?it9zsI!B)GEJv)SeD-(4PbRhg)(
z>k0mqt-h``%evRlPPUhJ(JNe<sc#YBEsCSf?WH1nO*$I(S0|~G;%u6gR!sMKltH%?
z<&If{^SI@iJ$*jc5;8Ee$yWdCjNV(fzxFLRyYlP{0I9mqtn{51|ETG@Ov@p^N<8;>
zMqvdX;22cm22nt#(d>KoM0|zmvoD*)oR$xO9uBAc`Cmm13!$K+QubO7iJupKLf19=
zuDhA`Q*z&JU&2D3S88z5gI1?8*88fa^^6z4r^NJbjgGj@vW(BO10w2`@Uim{7fzBn
zP4f;PJ{-0qw+u0#kK1eVcjXH&zij59>}5PyJ)B|^CgDixq5DH>z+UIxI6l!Oa>aJ#
z?#Z8uH&3+H*S|v@XD&~9HEHr>nTiABp0)%f44u*A=KM}nk|V}vW_<Y<=xR_Eia}RW
z_U5XodeKcwTC}x=%W^Xd);aAf!)E6l8!P&rP3pCA1AooBE%`XW#nlxTk`KZ@^x`;1
z>U5KmvRLrU3S!j_IdO)(`@@SdV?kY6J*pn<{kQ`>z^@j4Mm3xEX+H!-2qY-?(qe}?
zSpt7rP7^Kgg=+GuG2ahdo6=H$-evJpnE1f5TOkqj35)5<SRg5%-fA&&(xfKQ-j0R3
zWxEY3ge9ycwHXmGcgE+kGUh2<;Pmn2&}VYM{D$6rr3UqC{{g2F!f^!j$MI~|`6*Nl
zOnZXD4qnITYNYAH!{$LkzrDDcG^EzO+wbo`D(+A4z8Pz3&)z}4OqOb1G{tdO$L_%B
zH&*2IE!Md){t)vNacC;;%|(#Eu6fld<HOglH>rSG{Uod^d7GV_20&W?d!#R(onGFU
zWsPsYevrtjC;R?K1=U#|&vTQ<5zR;2mxpl|fI$jy3?ap2($l80m+Ihg+|tXqyn)JK
zS1~fq2ZsbQVPU>5%#)3;ef{=Lu4m7l0|w~8r%-iAl^YTx7+4@CMIkr%&auR)#(>fr
z4)?E<cDFVxCitaM<!{^e3*?2wy`;GK8J*r(8G^?3o)UYd^h|5hD$e~7E}8Y#=CkjB
z!yCm7^fkj5tezgsRnTnna8H7v;~P=&t*7$&C3pD5i85w>e)^OLAw#=3;lkjAn^n)h
z37T4<zDV3noM~NW@gU%|E>l#sh@M3JzqY|*%u7T+?9Nuw0Hii)BpmcLf|P((F|Z!W
zn*;()bpHVZ*4*!LLJg%6YXZ6h2gXS#5m|JwQN%VTG~iV71}+7d%$16Xx?senamRYb
zI4X9&D{wfc4D9C{{xMKcx4c@yEh6L5g}f!UiXh4c3<@enM?~$$4d>{GC1j@f{AS}s
zW){WV`t=nxuPd1vm(7KUX({312$HXG3=6@*zI_~SAxlh7tQ1Gpz;*9Z8n|)7ikZrW
z`nZm|5ZIr@2c8T3NE885QcQvi`BnLK;ZLe_Lg~}Q5<sUzHcHov7*M0T`SVjEH}U%u
zYXq)FOf{f}!v;ps1rP-ywlf+1+O4y%>Vgt%X7%vi)qyRqD(0JWsQs&cb$<9x>$AR$
zuAjw=)4#q;Z<LiOiP&r5>0W$LtNiVyeV@fCnYHut>-eZtUGoQK`P=i+xCd$NU`lQB
z&-7joqK*ubDH#`W@{+c}!NNQGX8BTjYLT&xPxZY!@n=&G&?y{0%U^!|*grngLk`z+
z-NIA#G9uyrt@KGZU<~ytYTOF5`vtz_EWdy(4;az=kgh*LVo|R<yLLq!Rq%LeKBkTe
z%0cv<MwC`mh;dc}Lzb^tu^lqJ_N(#zV1^_Uf(C^!@ih$yhTmD5Evfqe^*nbDRKv26
z(-INdo#kcqeVPleos&$K>LPEii^lMFw^6A-1G64xFbDc!R#qHeEtD-gpt1^?EKfCR
zOF8c}rf$i!X?M{(9WbInFPGjp_nJ;q@9S6DyUP67_2j1NN4?>0UUNnN*f^}WJ?Nfz
zpvZ_NQnh&Tj){rd_lh|X5xm<Jm%hv1xi3jdxcn?yw7UKpqOqc_8zpY{?JQp&y>v&F
zr3JU5y;Q8qOYYncF}5cB*{kG!G9gad+SIzRy>H@v-!EZxM{c{S(oP~Gl)oMIpZ_*;
zX|Jm#g#Ox!X~#13+Dn)zdsH2kZ|oU6@j(=7Zt?S)raIpr*fAK@Ee5ViaCHrpH0Um2
zKgy%y;hD0U5>q?;yLtO*kAL?KiTn2Z=i9%Vx3B-ZdHcU!_<wzS{MbfbxvZrlsY0Ln
zFo=odmbd8nObSO3?K9KOq~(ym+7D(HOm=rg=)EX|@IcI6&r_zG<VOD((ec60>VJ1(
zhccD#_wT&6wDl~B{~@03?_aL=tufEpjrwF8^<`~aw_F;W7^K>ukj>=p6rr`EF?*vd
zyx%p~exVrZhF^0Pt%sot$z(Y`4;TpwhoG$MBHnc6vJ}<d@BKggXd!?XKmYZ@|K^kV
zuiNu`dHi=4?*Gc$qjYNCs>-&m>(*`q#)?Sk{G@NcevpU*Loh?+M|pfA$Utc5;B_vd
z*EOtUZDm_mw<pa14h(y~mNArppUlaF$hi9VUwv=~8}|0Zwk=w?=-m19lth<bSTQXY
z^eWy8q59M>ZB51-N7DM&MM^4R;X0g?;ubF8?PGs8+V;q>GQ19`x6+T7ZyetiOl0TL
zDMDUm_G=4(gUzp5yxiXQUw_2*M1g1YYEMwm?*_O|zq_k@9o~_-mXV%lKd<D92H!=!
z+CsR2w2X|o5D#C!Jx5%Q-(GQ-y@oNpaz3Ea$ae)VUbx)$YPYalT=eDNom{(AA9deA
zR&j;IS*EQadtF<6DGZ$xv|w2*(ZK6oTf<dhC_+ZeD_7sy9VS~QLe+`izfT;Q{OHQh
zb^83~_U|aA9ZfPuY4x<=O8?8evM{r4i)d?*JbnJW`r3r=wY6e^_(9RM?UmcsB3=0O
zVd|?ZB!%!-H%wRYW+wQh@DEuldYq|XfBovVkT!vh8-M@y)Gujko)o<TOP`6J>CoWc
zpK%ie3@%PnlN1|1ac!As)eEH-;sI*I7kD8HxO6GY-ke`&v>k$>-Asa3ckSG{bNB8q
zLPd+Sa`(2VyW%!kC3?Q#vIvEj;(B%4uGgd^Nu&~b%A_8<%z4x7_P<}f%lW1{gIciU
zzt+~ny|nFnX747;MK@_#VYxv(cV)G<ZxY=7-Me>WO2Ho4s4_5j99Sp9?~%4!Z?ZD;
zZTo(G`_@95uX}@ohs2^i;z>?D2DQ!u)^yVO{mDHo++8I~(>X<OCY$bF_lJa6^^^DS
zTWQ97phz761$k`zn!wbX?K$xOAuZ+R^QzLFzmEYE`IA#TPiwyA10*C$^TAt@m2RlC
z3!P;DU#CFZO_BJ2A$p0`ROa^eQn($Ur>H*!o`B;j=QZVnogKwhUGF}9{Au>0YGy<f
zE?pp#Q4ZtK=qm|3vkBWr`4#l3$zFP9r=(Hyl##8kB_x9XjIOO<BRE9i#RW+Tqxu5m
zCQB080_68B=H64*)hJS@AdY$u)W^qyFP&SmYRLo|EMc)G>gKd~<?$z!foonV2HP^_
z40)v4f^oXKjVw8Get=!dYXANFsj#kJN$zCuu}BDib18)=dgC!Y6>ZYtK$g^TX74ZF
z?P5tBW4a9cn|5AUAAU}r^Z;DsL*&r*vDe^6``+p!A<<r6^zFsssa6IMEi~&?UMJ$u
z?%!`Dv~Th2w8WUAHQO9E{)ZOe$EP&9876!kGN+P81+&=?@ZF;~!*B5A$3Ut4AQxr0
zHQUd7FwTV@WKja3b<GRC1mZcCDY|@b<kcT{ybRur!#s`_RNSelseHNh)$}9Qf)of<
z3}wJ0HktA&aX`eTQ`GI?IiSFf9l!Vx85xSzEBR`R-cm{$U8W`X!rWCR0nD8gUz?A>
z^y5fs;~xjjBF(;VO%xa3@?y>$$x7@^a{z_4aGN8S>&~QdSDN8hhz@M8R}63w(~Z$M
zySui(@q*G7s9i0uOXtp?5eZIcZbuhR{HiDNC_{Mx4u=G|$&{7=1`U>fNZ4zLmG)!&
z`q!IL4;Gx8--`*XR&fLEWq<;fKUbCY#deF)cGRz*zGlyuDj_khDz2Tx=v_>Rp!oxa
z{gTNiVg_W>l#PYA-+Q)hC*`DzgbWxrxpC+=kF7cl0V#%#YnF{3`LUuR`b<L93>1ED
zKw<PVuYapJ6{XjP*7}SoHymCqt@O*!;N-wt>>pedM^BmJ2`N!q(ODw&VPOlGJb>Ie
zmCvRTmC_FZ!{0tWkC2-}Y=Ve0qurEgd=%JXdNzAQ>z!LM3=w^{U$|zV*()oaEQzN7
zw=_L!=kMNshqnsB%isZA#xkaTfUK;5Cwzuivc4{@kXqsjYJ@u~ovZ$P_S6fi@t4)^
zKDo@s<-tL~4Ptr=G6wo9Ci`3gzZfy^<mzXB?Ig1MR8TNRz$fnFJ4~;51-8uuj=vtw
zoiF+Zpi}7b;eozo2S(L?Z^iC)FpG`82JDVE^TzaGf-)UvGl1KYy93_P?b!;%wb0`}
zcL*MYO9x@a@M|&}t06;{Kt!{4A+#30yTRr*9&ib6on%04?@RL);S`YKAVSAK8s9(A
zAsVSX*z#=R>q|e~A8v3~Z`)60B_)HCLSxm;{6L4-3N>F}UWuCW6KxPS(L@SJ%-M7d
z&7C{h<*V86!7)%Yb6Tfyf!IF$-1KB}9?^t^h8ZVR?5zx&8xwv0PQbT<q#rc3aK8#G
zD=R;I2nY_|%qFKP2KnHS$&*umIxcv+G2|=n@tJ2ufOt!w0hzTE$K5ILD)t2(D2tO+
zPS|z2o|c(En+t|y2T+iAFOTDvtz5pG*&H~bHv&d5n0WWEEthww!X1EVQM&U(;N(T0
zbKrDW>aXjojPo?Tv2*99Lh$egBaBZ^Lh1*#&Ae~Jt1XaPVo|=F_osxJ&BWUS1-)3g
zX@J)wTmxrU4BgmN?*x()mA2H(EVv}+APV{=QJ~M47pL9g`=+nULpTmv8nLn@#M$`x
zg!S7#KOK@}k@fJQ2iV4`Q->=IN5w^rU;s=aP1-2q7HzK-#G=}iDT`CS@qL&%($#4z
zx^fW7;U~m%L_^^gB%(_C2AbVwrkuU2G|%3?5Troq)uv^r<gwYUs;P;0tHgGq>C==}
z6;;)-TNiq~@pCp#e=cje3tpz;b-dBpOK@vt#+7tW?N`H<BBZ|e!h$d?YCY;roPP-B
zW&KzNDJS)#ZO_6@-P)#;!|TLfo{O%v#2;1@6aXb>V3`M-jah+FD}7gQR2_(j^j1mN
z$#{0Gj9+%Ai%REdI7LEZmAt!L+%&e`?OE`ADy|(Q9!}Hi@ls!}hjQJ?Q>RLA?IdTN
zjF{<$o9U);b3oj9_I+0w2CwJ1NG}~Ivn>~3l{MRY@d`q(q>9e9i4KR*-&HkV8{d%F
zrO$}j8{Qx!0F0x7A7y&dOGGJZ(Vjr6rLrq%l`iQD<DqZ!Dl6;D;?`dl`uaO>P7Vjj
zhZ7{Q#@dTpl^G^47}%gK&+9UdRtLOTM?)c4<ledK(Y8K%^vHYoo)@l{>`i|S!;`$C
z6r^Zq(%x^;e@<9$)97KtCZY&1DsS>^`JqsH`pqvh>1&R9^LqZx>>QfMv(}@IL~-))
zYe%oE<lL8i8c1s+S(VSDER3r!u?NH~<>Hn-*l==6x**ZXwv>-UuyAV9*4+2+-+T8X
z^1p=|CuZ%{WYLopsCBiQrcb1x{j7iQ-b>%E>~QQ(BwBDzLeVJ6ft&uMhk~9>v}Tr8
zMP=ptFr$7xdb1xNI2{tAbgLQDm96|`mdogM+wTdH6;YhZ;H%a@jpf&#APn13v2dlz
z_e|FFc!9gOtmaQa=np2zJ#`CFrEBi6`<&`;Z+=d+57~d<pn#1TEJ##Td?{8pj=+j_
zm#-Ml%=n4E;^=|Dv^HV|Q;|jjB1A+ICSk-}XJg2s`8lUE8~8{H9rZW0bUnPsR1vR+
zhi^A>f>H}Cv=p}Wyj-(RG1&al{1puTcxme}Mp-$6hMb~mZr|4o^=Rvac}?YLI22t2
zg*4e&ezQ+}XIooa+!_=c2cO=v_KtUlz{5r*6&1HIrlSh0Il(-1zk!-_DlFCHJNpbP
z!*6)$eM!v$JyU@h4d>mtI5o#M@~<IFmLHw_@aWu_Ker@h=4aC10Q4J)nESJH_jxmY
zg(r7YlSfv#!>O1~62Wj$JX&;MbB-X}=SZ|Ie!K_ymLunM*qXD@62eL{{@1Woka5m^
zc6RlKn7hSMg-|02Xx)Eu?INl^3au!!OVy<R2D4^qSl<{qYLsz`xxKyp-t$op$1Y9p
zoK+2|cVg1$AzLj@4ekud#7CtjrcgaOzj_GQf-c+0M&s5CG19@=Xg!SMMc$p@!c^R9
z1|D5G)A0DhJZ>{iDWNSIqjSbazMOnDCZ#%i+>=Sm^tH@#dC2IR0^R4N{I!pZUr{=A
zSyyo|p>Rzsco}|f7$=P$!%<2W;4VGTwWEZ|h*z$OBeT1I^#3zEm5BngTN-g+dA4O#
zr1SCNkSFk=2-Oy1P)S&Xe4^%6p~nQjXLRWSmyZPn4buw!D_UurSkbhAz-}fdC)3=G
zgEUcjh($wKh+XdlPwZShn~D{P6xq7NX{hAZ6bQeGk?-6`P^>{Erdz*duDjcTcR>92
zU62r$Ts<hX&*cXSr0oT_W^rp<O=MQ{Qsx=b9%0ft0h@d!OUskqW_m?vHDE*NGmJr<
zL@NY&v71{x{XUF3Sdr`Z!vFHMYd;7{CEMcJtF-ct5<fBgOV-=fgLT$~eA{x#zGXAy
zoPOZ5gI_k^M0hRI)basuY&(^^1;qW+<g^V|kI_Gzd}n=q4QM+M^YoIA7j+rBe!Hq_
zY$LA_COQSW)wQ(~qLd`Edw%ra6^}~X1ynXW7154SOCtrn)rLMe!)f+Agu_ZwU$Y(G
zI61>@iiPw->kU%8q)UmV6?Ufe^MqD@pyt}6&xNWRs$a28LC<5+k|`mpfe?x&C@O8f
zHd3J<&LT-K==wVl#e+7w`kL&e#OjERcP%MS-#k>XUm?KL4-2*U{Bda2uMVQG?NHeR
z`gzi8Am}2TE5E>q%w3{jHLEn>8}|J+F@TFI8)_K|8xS?>So^rQ8Zu_#Q=oxlrKLqH
zgU{jlv{pIhCW`%)l9{ie5E2kT7Tz`Jm$OO|Tl@I)>ou1mZtGV&A;SrAH=t`^Oxqnn
zZQNI&-Dcmbst%2OlAWEKsI-*q;n2udtbpfUoxP3v-zL_6r{!Fg{+erp-DX{bTg&Rc
zUJMOd^YPG4mO$du_wN^<>~sA<!O;alV<JEHu#{Q8&;<+sWH;~8Sx%QJJc@!36)d-d
zs|_rB!0t)+;R3fq0+deZl`tA*)#1Z07HnN!+V>9g3C@gL<(JsBOG@w1AakWGjpWUu
zqvOLDZW<~9U6EKGo}b?OdO~WC{JNpbhM}8yNv1Pbvh;{b1UZA@UJ);qbkr}NhG9B_
zOIGmAws4X;*XGK~HKx{O5H-tb$Me)k$jS{fayhp(2A}8`k%xaJWp&Wk-dcU3K<gnC
z$Ehivb~Y7AsquzQ>S!ImR+CbRSSiXWl6ek^Uy9P*N-7>+oud~`jX+w|H$-Xpa4MjJ
z#1hC1+}Ne_%Ktg<TQKo=>qX&CLor~>Ei<xYq7?bM%#(g2wtg1VyHDy2d(~<C6UCg;
z$3y!}JXOJyHeK-S?|8q<1?L7($Ea&)94U{zJH`8GZSwO!VEROtdV}-Aq#pU-xd)Vt
zD5$O^HbI02TDX6o*zJa%jNBQQe!1-=%-RsL$UWm!w!OTm9yNX-uKAWNVzlIXXX?zq
z|8DP;glQDX(et$H)mD9-r!TlKx5de6D?~gxu^kYPq;gbG*Lh~h$B3A&n>H<Y)+fs2
z9>&{8PD;#k-Z3$28jL-VQPu<U*WrXy!`eqro)}l(8kF^>j6K7AY>KnHPrbOYhFjP_
zD+5bM_{r*=8-3U6;a7uN^iGE^U)*>@ZIa@}W)00eK5}^8j`%MwxBdDJ7_@i)el=y~
zp7d}{%<@@xU(4gwFSE@a)JMVYs?7Y$ISqU?RFq-nG`^0?a`yD}{Jy=96}=RPWCsM3
zRZxJ(p0!zcALpXxsGM(oOG8E*E~O=?l6;yXC7|(f%<clOrPm^4FGq6xT4lZQd?<JI
zH8m3d9oWLu;E3;cdu<bQ>)ED%NtX6VU128g8r0bS@T;QC>=m~i39Wh47{yuAJSQ+F
zj7!LYAG^8bo2vYf4bRHFwvQS+;oRsSj`QaEC7O7QIy%?*{#gAXG27@;RCoVFLV3S{
zv&@n<e(+u{p2ng@M<1LTv-0YLbNYJ|lrqgntEEOXsUnXE92vtD-|uJe=GwDok4fvf
z3AxxXX0=<5$<C3JCt6vUY*^`#P8~WBSgjaK+VsOX%QPBGh;tv}O6*?{%=f1~$dTKY
zqg?QH>BGw%eAYWHSL`z#e?%|~=M$f~#ohKFwcT=Q4&>-6T)=5*S}ytUVqUvOyX6iu
zPCgrdD8+H!!;9fdk3Ge-JgXY3A%mls7Zw(_U4xEu$)17*MGa{}zI&?00;%o`-4R-c
zZlu5QC{fMSTzw}h^U53hCiC^!WC~$J>%JSqTo+qY-IjJ#g*DE)e6js8>k{g~pRVuQ
za~%$^a!X!`c7wgq!mFoFwXccJHnZkDhe2|$(7^2LI(gEhbS;^IXtnp|WgJ=_VypUv
zW_#$xei}B50EUBA7Er*wAR|gLpV5BM;!fr60dc*2qT;fx^BuOSPSgXKAw!$3Xa4xe
z6$y#_&H@)urDBCuI_$9)T!Vz-=!51lnU5YFPU#XxqEY0ddoWAA<tQn1%Zoq%_+zBo
z%lgN`kjq9dlSm(^_xVJ+626s}J7lieLD$vS<Zv7xg1*mP8#W#b(j=RQoM-nl)CLa@
zrC`0=`c8ROEdN4S*qigCy88N=3x>_BbqC2=W+kPbr0qUu@xb;Bv%4_+z^fG@lHPPG
zEDkD2CB&UMvzU|yQzGvUUCH0W3cNW5YEv_{N1iyn&(D$l>uiw(GQ9@h<EfTr^YnVP
zlen?a#%83BXWHP-zuT%~ixWmBec4x1@ve9KJM3HMcfUX&(pp=rn~gYCuSq`1K3Jkz
z5nHk+9EBj`b!tgCrck`n_warE`UwNSyd*PLdpxGHtZWbmV1pUmTxcuqmx_hQWtXHL
z2~2qY#ZdkiD1h#<O5(3YbBkHrKLLkrhE6>+<yHIpp6$mNybJ65(<U~zZP)v*C|mGP
z0NRk$>TI<@W0}5vMFUweIP~wm9^WE_`F;E1HHR*b&$;{Jg?X{Ymv3*AEv`1vCw$~h
z3IBgf@43bro$o^Rt89Jddy);pOpt}um#8fvE}Ur2JG*4AL|ghX%IOpdvWWXvUx_Y|
z?RC!=^k1^JrlaDIb)YQ>aeV7%<xrB-t1>hC5@?mv5`-}nC1!QcM>x1HFP-J*Pbz%1
zAS=sI!)b8V64y0*9k$8E7SDXJM1-tG012pP&J?_Tt9XV1&~p2ABqW{-IL<_E%$V<F
zl!x87Fs+p8W9_??q$9SuL4#za<>mEg<aTPf_C!yNRl9b4ek|h&fTMz@(NG8hWB8II
zDRIduUzl>kmjO8rTRuO%)3?nB4Ak5@;SpF((#+xhvN}T_8v=FMI{R-&=IU%|uiq05
z%n*&uTeh5@u>N@V4yrYwVS+R_w_L3FM(!2&!|(`$wOG)zY+6v6`0m~vxyJIze_UDR
z%m-t{l}*Jojd?_Ag!K90^W?H5i6`S$K75*J0m(jR!|s*0M{?L)t4r!OwJT?6s&nbS
zyDvf`B385Q$=ueFtZ;8g*fHd4YHmgd#6eslRKz0~nBY7Ly|H40DmlP}=#S;)9u);T
zNgDixudh~};UAIYf8ZD>27ch^l69O9=cHO$Goz9<$XGp~AEu~L-@XGmWH?ueYTD0S
zqX&~gEqmr=WoP$B)|%eOSWHcEezecvab$tJ_h4W=)fC<RsT?WJ23t^At+i6d!+~tm
zW|;S2g3T#XN9TLUj#V9$J5M8B)T=Mt6&x&migb8<)O+_pRs_a>Naj@&l5aIt=MZDu
zxWn7F(T*$ND9L0koWOHzRXg7$8C7)81x?a)H$V0}i`!$Yb8X(du8&JN|M!;d|Flu0
zaQ4`G1b#cJJL6`0pWGKOUgYH^U|s`7)__u`lIeR@8k;AcTJGF;s?Ip)lG&w3?Lt90
z7vU5WT7(7OEzj0lGNT5EJ;~_ejY2&ww+@dU#gDb=HiSCOq)lIe00TAY<D~et=M+U(
z4+zFwSEfzao-NHSDF!uDvfRXw)V1^WB5<3tA-YN~nb&a3_a)B(5!h1s&bBB@P4+kf
ztOJ<RaMati*{BhL0Xtt|fZ12WhMA&{HqX8`Xx}+vP>aGv-6`+V5LYjU>qZ?0eJXB?
z(Jlc_`5q_j8yA>Hk6YPLw0~5|z!2Rb>L8MHlKTD8*6#Tq$-HVXyBRZ^b+%J42m1GE
z{{e07$hlf*B297Z+vL798i$Bu*B!2my59FwG1Ju;A~!M`f#_hZ!sj!n-%4)1q{k$C
zeV9kufp=&GUu-S}e4emw@!nTM6;>sSs<xU6JowfMpW%DdDNP#rX2oS*wi;voPWnG9
zG#jbbkVt|0W?e^fiGdJt>ktQmEdw3MGIxZg>!|A&$5}1CI^K@1_pD1X&u8O=N2$NW
zN^4NV#}ke;lPLw=6*L7YoT@eAV=+w3#a$DeJ<IYFZQ>yuH5YwAJJ-LzWm~PX$--sy
z5%jU>>onz#!cLVOzf!v*d6>1CnKWh%JLY;(^y|4*O<l2Kaz;Ve+Wemr3+ww0IJjn`
z>cm@g$^)LxFfdSixVejjSATKdDQYlwn%rFYozMhaG)o@K3xEVXlJ2S6s*W64j+qG(
zqQ!_<U+xCykm}MUX?x?14x?_`-|@!YDDqMg^(a;8z8tzMfG!RnKW??OD|xaI+fwU2
ztOz?NPRJuMw#f^(4|7hs078(Im^fQ69#_56I&g!1fB-B%8yh~mP-vl7Y+s-MGt=(o
zW)&)>x$?^?v{+Zce<i6lg=>G*FuzA$l5DeF_9RJvCs&{j76Le4p4F6*?Icuth@7x%
z>Bo<pi}9F9(Gv85l;ee?8=^TLmlz1p*l2uG`#b6S%dhm^{X?Kbx{M0YKyz_bQE_pn
z?gO8rE}=C&$4<P8G98Uz+%&tfOiM#!%f=n0sv3BCBl{VvLLm=KdCJD1zg<j{%lW0E
zG;-u+KD(%zbAlB5lH)2GA}lQ}i>fIagp$~~4-G$5TWZ%YHC;}Or$Zc>-ze)Fe>JPm
z{<XjfKgF;8q%`->^)Cl*#83cePd-T!n3Cf6<@+1OrKO6|p(m$1J+zJ7P3U;Nc)9Eu
zaXW(1TrvcVJdsUn-F&s1M8x2uA0dQQVUf_lex(SI8#wSaxi+Wb6wD2{B_h3~UZ<vq
zDB}=^20*MY%m_d}z$5<B7`H-()CiLR2NF2m33*}u)fGkzkRi`U1PeBNAv~M`XElgH
zfkw4lO{W-gy*MLLNZw(0K{og9-yaF8HN$A=#{%{OSYTC5KuDbw07pg|cnOSzxb_7f
zAu=2qS20p0kMAL?E!>)f?)9gwOwb_Xo7AIxN|d7#-jkdrtdKraG?%9D0W_Vnp@yRj
zqt1*+kLET#hg^FhRXaWroEQKqeC~;ttV$LPXZ=M`2!w_2-o2w*<<_!l0Z?Yoo-Gt*
z^i&Jbx@^!Uhi%J=nu0p~blQF1`^2=gG>#rRHc`L9Ib7$P@pZa*o8TwAYSrE|_Ad&O
zKY1maDT20-zm=)22iSE`1_w&IG^C|PD_YwR8deL>NSK5sTuu1`wH3o8LJwRkD|b<z
zvgirqc!sn{s$OBA<KgbwU8ZaJP%me%8d-aFAs&=aej>UoEN@}*aM0ab(r-a835h?&
z(w8Wu+w}@Ei~FUp8%!-aQQNHVbN^Cj39m72f4<@`FbaZ;I9_!Ih3NBYdPB=ADl~R{
zr5#%1QBkIsd^mQ+N1Srww6!h^rwkyOHjR{wH_eqhB%5$3f$0)$oX8V}ULVwljGXbA
zbejHRm=LUtg~i_c$&8j`g_L*&eLny2gIRkM=F;mp(;mjVW&bI(d=ko0YxAPTkYRkn
zBHmx1dx9Okd@2_G`>;0k`5wt7$@t6*7t+g@F^e55{X)+bo1xs>SRk}qCPTytj$X0%
zI?KBi+A{4V+8jzv9+Ulk<49npuI0*=7gHO(B@c>wH2Gimpv~sY6L|yxh1K=d-$@0`
zI0MPxSq~&OtOqIiNyOwtJiQ2TBstEIWJjvJ)i<ib01o*5J~+0(oF-kvoe!c=Mo!$f
zzV?%aAc3E$|D_6q%xnAk-$DynjFu^Mn&9-Ylr0zD=Jvbc{W$oH^j^Tr9z^8hb;UMy
zTYPNn(|4&;XF|qU^+$vZ^!Klq6He(VG17DSaJR$nm%oTvytu+w`S#cUCnClFm0GX1
z70ko=ZwfVDoI4!+A6kI_8=}g#@7to_UMHtuept(I%Fdn_>V}4Dc%QS^|0k?`n5y?>
zW!0hQ3ZpOYiCq}GbG_3Yjk{Lz7G`~4cFecftS+6LxbNV~^D4@x22bv?WQS)-<3q<S
zE5{xi9(1DTyUKo^e?30w{&8I0h?AS9)ScecPNH+0m9T_F`+q5Nz1!qa5)yaX^kfoV
z0-Xq3LE`$QbrrS33c1I6pJ-n&YIOt>9aR?UHLcsg@ygXt1nWx}R4oe~tld#+MNZ*P
zuP-(EAz2yi%%;K+2sf4GhCq_75Q8D+lgecf4Aj!KZnOEHnHjR1{-Pj^r1=Bdb(sI8
zs{m;94weXiy_7rinbw&G=-#R=y2o@r9z|$?+gm(*4-u4funuu|3)2SM4|s@Y5xjM;
z7AKMgt%8ly548(kMR>Yd_s(6w(0nK)q-Y)|zvhfa!ruzafn$mF%XW9)Jt|zHa5isO
z_0aa0VmZ6pr>zuu$~Orb<$AUx;?#bmHY4BwqK;7>CWPhei6^(`knqaata3v$Eo0Vj
z)>nRCV~903f(%8mKA3Nsv7vNP$mv}@CPs9YYiiw!vPN>f&Ztq9v;+t!$9qepMGQ{*
z36-zo)sC4gcxSI-iOU^&{G)_WHwAgu0w;>}BJi_)&+82owwrPICw8aPf7qSff{Hl*
zRkWfgHlDjbp1uO_TEp7`4-qAJwtVvJt#FalOA4ZqEUR6{thECs1rOQfD`^~K5s{d7
z##Dp%y(;N1`Jma!#0hU3MUQKBY{Tq{ve|#pnu>C9m=U#y&hpGd=WdAkd=XnB4vE>p
z6qFp<ino@m=Qbx4IOjNeHooD{I~3TfY%Pz=AdCA!Q>*W^v+0_j9-YWbaC}dnvEuuB
z1S=XTRd)eHmbRZjvJnwz?U)+h|C?Di)(9-rR;pdZWcQ}6qX>!P9qH_zHaafMe9!5G
zKRi0jzgu|r?z1xett?WLfBrf?9#CM%zL5;9H6hyU`{y|l1zJgtoJ&;pC%tldy(XM2
zMk!06Cn8a!A4Or&uCV1Cj{vHV9zWIyt7U0XJB0Led_%f-#;s^?fFg|+BgYRiAM!0f
z?mX_X{*$U_Czd>&E=3+!4lU1WIYEvHoJBQYv$gav0qaRjEHyu{!IY}of3A=oTYjaR
zyy}Eh`QYh<2$^i2-qH{|N`sBxt#Y!ml1h(Q&ykfb%~MY(sJ_!lL3&LP?iOe?M#gqD
zy2l0>%VD0o&hyvLFG7-GXtk&LgdF}^h64+Xx(^t;9FDdj1~<`)n`P1TqP%m?QRgAA
zW|xwCvfSsfcko@WB2yKV_>XnS9aq&?eb9fW*vVaYmoJ=Jr_vw~*{MEbF+z;!V!&2F
zpt3z~&`FAIo`Yo(caX(fM~mEpBeacU>^+lXnpEHw4U+wnpFT7qVaQgCY5&xI%_7=m
zUGi-ES=+Jjdkvd~C*UfWWO6427EREr572bcus<}uRbVxfQuFyV(uA2uM9Q)_`TcUx
zHY`FNTFPjDr!W_;tbj63YV`4}CZ`Rt-C~)@DjY)ox~%8UyYHS9p8bP$kfg{uI6H~V
z!&H&Xa)aT(Za&i}bVeF>4Gt7k!nOJ3?4(5|T`6u{%jb^@v2o5KSA>K~2Kfgft$wUq
znNk*=yfh)T*iZM5)v@DkobJnTr3X|YO^d0e#?g9-jwFc_4bl)<3$aB`i?vJsudygL
zlu&h8RYq^^Yw8OHq(r3TzD9&Hs{?}ta>D;b9D&lXIdHLdJ}Biu8`IBxCLQpOXRRl*
z)VWbPo~iUu(Jgf2deN?Z_y2uu8Mi9%@L@LZ=KuqFOVUK{nVfUx5UM4O2i_gdntUek
z31tp)89M?fg^`@gb5qfj9k6hk`WZr${MM2H0I+1hBb33Ih0TGn@Y62ZgSit`)q0E}
zX0E;Rx2^~B1~J&;)5%SI2(P|^t}KRcMRF#O^hTHrJ&!pOc)qWs&pTS>)L?6mF7tcE
zSPfqLn%aU@TYWW0<AmnAg{IS0&$1`dsL)M)Ks~o~IdO`u0zYV!Kl_4|UcEgVkb=??
z9+rBCyPC@zOqZH`9XO?qUTD%~qx=>eqAKtMrU+%$@^aE}0e2zt;%=r;E)2J>f8yD)
ziO?yU(1;ac8$NvKfKr9$i&R*fQCPHe#{b>xNGE{*&QIERgMfB+uq%JPE@nM1GyA}9
z9;)DAJFxqa-B?}yf<M};>cAwZL5?W_g#l{~WS!E5sEA`@+8ZI}dyfLF%0I37o8$iG
zPWg}r5rLu#dO9;v?EpZ>q$b}j565KKY<i@b<$(3Au!XC7Rmky`g2*RVa*Rwj`M~8>
zCXwCbtIad-?6Q-}j?Z;lt=1UIE*?>@IY`gLZB3`1Tl<QLndl<)EgMfEKVf@H^&cG?
zdK8jET)MZOM&u=UoTewOO~wUVAeoW391Fd4?eRplw2z!Dr|NaZ{F)Cd#d!9Hh3qHE
zMr&iVQ|_EUy*+1KrrU4i_xZQDWoPh_rEERxerdSwBxy-8Erds7+~ife>9Ni#K*kR7
zhwh&5Tg#%A-SluwmuVB(Zl3iYe3x%=nUik!Uby+Ol7Bv(-sRS`TM%C9dbN5Lvo$bP
zyt68`zG#2eL(ebBqBy#OLr)P1A8Zf?-AyqIc=U;naHZTpF|)Vs0fo~$s)Y0pFR)`F
zFi<}DRJ?gU@pI^|q4#E2w<LQ1by9mwZPY1Zw3FImkMH;64m;lba&i88Hm!4(V%(X=
zT%%E87rqloMmkSeL5UVJ(sC@~z<lGANKVgQK~FQJ){|%cF>Ltr;d9UEO`cHi5!U=q
zTEip9yJrr2&N-6ORU>h5%KCy{vn^{<I!Z0+DAiHQw43~#y>>AXy)UTT6DOwE`E%z2
z{fBAR-t;Jd`Jbo#;qUa;7d-}yy%6fZ-{EW9GN`)ON6Y0-hknxI_0*nB|C$+dET5p*
z$I<St9PIzs;lp&wWA7nTT<|$h<%4NXUFI-<GluOxJw&DV6L9wa`XSCEMI+wd2E$b1
zek3sg#^m!tC*S9BLDCoJof;UwBw+V*P9x>q?57QN=J)-If5?o#aI0vaT5m1)+wD3@
zOD{nlimT!=kH6W$j}_88jZjqxIeogtcS_S)>N*kvtXRdIcVv4{CYD`x$HoJgYSi`k
zMBife$3bhI>UJSZ;Ix%ra3;0uB-^v*ED=NfbKd3I?1wF%&-%`~!36IT6J@jVaX}`^
zo6~z4%vi6@xf|$vJmqS3#&#T2KS)K=hX@R?|N7PvbsYq!`>&YDS37l-TGLUg%YaEG
z<J^>oq}rKgW%L^U!~B&&cLhDcVdGbu@-;GwX6F-!4+~>r_J<w2>VtkR`Bn=>bHCTm
z%`I$e$ohnAl?N}XNMTn;AYcKMb2N3+(j?{)->*rh8LKQ}N)WF;*?JSWjkHb-1R@_e
zbK(SoyqBOyf1R3G<L}wD48S#Tjk>vR3LnezZ=?g3=}|$P_;3cu7ZAHNcanFr-{B)h
zMsX<lJXlbDzP2RD*lx?K6%9r!kO`HZ+pn<=u7L1*A}EMa0lURX;*s0}ur9tG-n!BQ
z&9hPoDW-}(c30GAgRGrjC?OI9T&4Esy=J3`qY0xc#WX`?*L-!4vMUKA(y}-*<pJc<
zT`)2k;5ze!iSkb#>X2hm%}e;kli8g)8!Of-OK)SVQ<w*H0|^<Z88qlXMTC|oRI|Wr
zACIk?rT$p1looz}bq9{2f^1(|=h<h~FQnWF77!B$=y^%^pPx~I3Pae%T%CdZ*5zzl
zJlQxqwiP&K@Vq9~Iz|rWhWbAoN;q+3Nhx;IuB~*Xu}jfmQ%ui>J@3%UmeWx^^L3MJ
zTBB|Kk%MGAn6RlFz7w|R&h)WY22ASDe}exe9wv`{+-bgPxl3I}SXdYbk8*|j(*Psi
z)2z>|@rREcYoKd9W1%pOcS}de(!dG$;H6=fVnu$Yk=`RQ!na{A=kymsxk`j!qIXLv
zv>@EQf9o)N4_uGa(4BLngHIff^VvLI^@4c?UJKlw3+MT&3%fx=(n8hpjVFDdQU9Qk
zQ}X`8+20f^5_5)=Nrpe|UhGGHAv+-H=(bz5tZO^pD;eeV-mUmxoMzBZ%9->D{p)(W
z2EzEHu~}wnYLmzn#c$zW_bpXEJo|aGlMm4TM|zdJu2tXQR}9nUXPp%4vK1)NVq5o%
zHYKg+mT<oJn9^==G$$YCJ5!uQ`QuzN0+6M1tKZ?0c{{}cWIBjER9CT1bNHpp75<Ho
zda^B{@VEWW``~vlq2R87DQh<TpX1mryO{n33I>M?6f*QwCr@rB)ZjQiz1*6tKr*JA
z$Oe~3e|by)r>TGYY*z{k-^RnxsQG#qCv`oe&b;{23qkW#Bdk~iBmO%26d*AD%HfT^
zT5ECzX*e(TvIDghUV9B2rGDiv^H=m!y0YA0Jx_&Tg?2Y;epq%|YjY*JmPUm^H~C50
zDIFUSeO+TuPCKu<mkm7eMl5&k{)Q^xeLpLS)tm$z=W1(p@&866;^82yN=3i9aK6*k
zb!Tn?C6YlwnDvIl54%f3=Fc-SN`|K5<ckFh<soAUJ^sdB`cIGp4Y0gQ{7p1&a(Hv&
z8|KZ#U})BBK<I3?aT7{^;vjW#OX^o?<STE$T!jT|egeL~MbY7Qptgx~Wz6JvT(HU>
zG_pHluD$9fnLt^TQ+Wo<9BLtPDtW73h6r-rd72DXXiEvTYK9RFOIJN$cGn92tLT_?
zl1$LHzJ}D_=0IN>#0TtP==U$_bTi3r^4;W%cKm&3y526|KvG;CB_7q+=L?LP0`9I~
zxjV$0&YBkPO>Zri!kwK)ESgXhXr$4_f6eKwACF;gey)@CguBDSIhc;5#7kFHd6VZ9
zlb7Xq%4yo~GY>p}gX<^7E={||P3kjhw8yi#TBXJXaP-?t4_in!_Ex+K&>Im*CeH5Z
zn|!BZ_kjf)?{yDQ<fKSuz{395kN@t{XQ}6pLE~2oreiT@FDHj5r3W~&IFTv#HMe{@
z#;sROG*rr_jzs9@#_?dr;^4RbdcQsoSaq_Owfb7S5+G&ZzX9gqoJy??PdtG>-!q*E
z5x4B-r}{!fk5A&j<ou20Kc?sSlbVnT4c=n+m;SgF=ROXQv6A1QR7(0rmQ=JpkpKvZ
zFl*kqZe4w*^|Rgl4+z3NMF9kOkDn2co;pXNF~sisjK`zrHZAh<o-x8>nE$+nB~^#I
z?dr74H~&F*`M~hJUC#6L{I)-bYClu>Hgl_nPc5^hx&dP<e}2m&wRQX8(-6Z__{D<n
ztN8Hha~utt`tixl{1u~oD@j+kKRy_hl5q+f$KKCJW|@klxPD)mLQ9oz`VhdZG)RGl
zm}S127B)^Z>$jAW8;y5b;InVCm`AZs?rh&I2-kaQT^cIVT3a%$EdsxbsMlARD+g%B
zmgtrSFDTM8sH75`{c%?h3g-MIbtlg{VN;Z;u-N@ueEvk>ljY?N%hA3JRnjVT=djyl
zGc%g_2h8k=M=YlKX)FWwHhE-9?dx@b2YanO*;{crQPhHyH<Ws}Xbe>d?W^KZajLqb
z(R9NNw{jjoCIQ;Ntn$t=vc?h}4_wFgP2R4u5amklEz66O@)3*FyL```7+}p}SL`q*
z>^Skgn|y9J`IZxRrU!hk2|XF)zf8gWsq9V<j-(!G8rQ>{-d64iNt}iy<lPvdnizgg
zt>2E;#`mAn>Y;ZyoofMr`hB_4s;pA*!SIIItnD?8oS)wvCfw4N`RQDI8gci}jP(6w
zAvPrg9Bw6!)82Q1RSrtw_ba%ndTLfZk7{U0WB2{Kn>8XPWxenjb@TJNeBYP%dlWot
zY0PsVRTNfQW<Hc{Gf#h+xv%MMW=Kd#zq*0W*>X<ly+}FgNi5h86Cl`{Rvbkio691l
znr%iGx{dH&q<5o|hLgRjDxY1?hYt8f@#~axZ<U;iW9<0hZWljL!8KHRwz{)LNK{VJ
zFv6<`Sg(RdHg$J~p4NK1c|m;!`QIdNL5Hw``-fE6slNzbq0_5jory`LR$OMP8F;*O
zdhptgQdy(jW{-2LckC!N^BfB{`nK96lQ?l&-boHPrC)yVv*|hc8aXxR?tbXblzQ1F
zJ*8H8-%^`^{pK1gtJic_c|O+pgakZ>+<=`oxgxQcI@9FBJ1uiv54@bH?sEZGgl_`+
z5hwP3hg08_<d27xL<slkhEWjKas#$1ORIjKc`NtyI2qk5!^VxfR`Q5(gl<js-@(Id
z6PH8(V+c9h*)%-SLKAB=)2A^r3U-zw;X$j@yPMh%7bFQcV!7>cJ^@*w`3C-+`s%hf
z$8T9{_{W`*8+zBx_qf3J69)F=V*6DWgN>m+L{=!Hp|ZM%z5cTzNmZ$YAhAI855hgw
z!(3gVCU5Zv5pg4;p(dA4zWbPhwfBx41)Kup(mfkya(bD3ZdISz+jXJi+0^+RNLzMQ
zl@+~x-z&+yKJ$p$f866e_Ejck+a@c25=UlaXxRRxGY{@tDzmM&f6~6Wad}Ri?Z}hF
zmdorkdIh90RTp;1E5Nb1WuiBLR<5g|Cv46Id-&q&{}WiuF)#WpIORkRTbR`)Y`HF{
zZ}f6c)tYbUXH}A^J?if1);_0a_I|PQdqwqZyYhMSsy`23NJ1u#sH?-7PD}Wo=lp{t
zGQIBz?fYwKQ*!t~!Z<W9NOd;w&(f}4C&$drvZ{1STY0Si1cyZh)e*WQ!6XLOIYo46
zX7{y7wW@@r+{raDDdyzgH|Le09=V7(%~leL8Mo|sRP`AmRU*Gt+4r+`Z{wZ=RpVsU
zVl!KZ?$eSQQ>_u6u5;?@(yBv$t!vBWdv%lFHwRSdq=#&<+1b21l8-(gwkDb#r_B^{
z7=Jc7O&esv8I;ye|IF-1maSUxhr*>Awm8NqU*5O2xK^K_kXO2stf;d#Ym#$9#QKXf
z%4ejRbuW6N^XX*Ov<VFtjnA~uN_XMHcj<k)kIp{rTC5UKFRd|~V`bloe)}hH+S6E{
zlK5mPg)TGs);laAO&R)WD_0t<OBq>oB;!<{ehC|C{rS*peXivkUiycBx-C`e8Xgu_
zQ}I^UW=Kl9`{gEsF|$-&sHB$8DETpd{}^-3byuO^Z-}8~uUPNYwTB{I9iCNq_PRM#
zEjg=BgSPUeY2mpu9&cC9x^|nTy)SjW;Yp_l1B5aG7)=;zBm>qR825NjN>eh)4u`pt
z+lXyh;#9|_?t{@D(M=2xwerd7RKz9V7Mi-cx*mo)74wHV9WIIRQJOF3R=D%*IpuFr
zp>a}SKrYjZ7VIy13qBDrW%o4UV5@j-NrBPJF43wPZ;jBlhG`X#RIGbkju)^~kaUOU
zUNax-s7Ck%YaMm&aAl^9>-M-rpN53E3)+{x?>}1kV0)8Ihi4>oRb_;`#0cYl&9t1|
z)vCYFh!f}9b*6{yo~)~~=Rb$M?o6NaJCg4TRNdUa-`3Xy8uLS|nyY;5wC8?2(wKXV
zXe!Ed3{;Q9NdmcZS;7G6Q<{vVj|PZqqU+J{bV}lT2^nyN_0;~LOlF*_0(~OPWp?S1
z_2jckmgyU+tsje#f2&rwq+IQtTw8^0Mc13Umm*M!zGi=A27)$@xOTy`V9j8lDAm<w
za&_4aVo<_1i^$?x`J4hnPmLmB_EZb2+P9W?>(tu+#P~J6&2w3e`D;<QZ7o<?{O<f_
zgw-Kycc~<h;}}e?9<{jOlj8$*`l)t!hpSz*DUb=?@#Dn%_I;;X>Hu~~O{%)xlm|aA
za`S)*XGjC<T{vx*ss;O=&MUlAc=qIIH|5i*ORGZLGKwB^#GHybHcOy7Kjg)&I`Lt@
zsZVsCE|VqB8Ca+yN-scSmuKhxE#=4KoKco6-?NQp^Y8j+WK#?CmHKqM68L^5TSMvO
z#%V44*;X_3zNbDEfM1+XoK&mo4bY<G`C+YJH3O5Dza}Y}-a}{j4G&O1r!sM@#`!3n
z+B|>$kiI$G90lEfs-bZjNm;tSod6vXmmx2f-yEgut+@y{miv8VN7f~l#=VaN0*~t&
zMaG>Gw`vZkjMJ(kH4rCeq`<(1^Be?<7C7>Nu2RC!bouXOp0o5q&i|2&{P123y$vBv
zbL}|wy1sXzFb<=ZnRH{`*3VnJi=-f%w5#Zqbw?>8?30q>&<lVY#|tM7k(Qj?{PS5u
z%(@qmGL7}&Rg?2c<x&3G>VLypP>>%^!vp%c?>ctI8OK21#ReLuPg#A~e;x4EI3>86
zJrT0}1ow)&)GcQ44@!8YhP~}@zH})+c@Zq(Vq~+xD=FAUKG^*sdC1U%1ICT0bzKQX
z1ju4)k+%Rpv&VZ>Y10~@z0uy=$94+`6<`M(0es5%iutR#Nq)9wrF(|?DQ`LCnDYIB
zsev5x<zw528wx19VE;E!EZ1*19}6>ff5&&eO2Ou5b4`8Ppy=jd98cQ##~Q`oxjsMt
zAi%Vt!-pv(z!C0yrk#CC8LD7??YgJT`M>+=ch#-UUv~bngIXk@`}mX2zYwh{5DbQE
z$r9Y_k3<QC7Z4-PC3`x1^bqJcF-rQ4-59s0kR=~O2F($$RMY#`=J($3nlq-f^wQfZ
z&vlhgzdSV0Xd6yXUDtksyHfq>ZC!HyRo7Je)4c<_4$83YuBCX_-<(!oTkpN<F{!my
z)%{MAc~P;vBB5x?6QDCiVo=jliF7+jKabH1vkuDXPG|^!>thRC6KB01w^)RDsQj$1
zt(;5C9>Wv`YeCI7uJ&pJ1(58mW9Nbo^dB>{_KPj%nRXO=m8@hU=UPrsXd%Qk!U+!8
zUSw@#u(lLgDwCEO&f@JLeE19Q*(%`&6sEL*mU9-~BSMHaLDH>cGFc=y2i)K5>l;rU
zM2IOL?TUB5C=Kv%V$cg6v3kxWnQ1LGF>!Amtb8@+JZA;ZWtq}wa)nx|U4i2en2s}x
zl-<H)czg|XOkarObnC*cxrR!L<K)&5!{e4vst&XHz{44({F?lGcw`BnkjCqN@7BG*
zf0-=mg93B5YJjGsZ@xgiCs)y#-Pxz^=h!PAKY<$?*bsk;4MFz7hR6#%^S(!y;T0A@
z7<RAQ&9IyI1=gToBIZns6Xz{lxX}CM&{Y=A&6S?U1!h)OR-oy^#V&s>qn|dog$i^8
z)F}NWC8rT5adp{vV05yTI2Tc%M`T1}j5_#+Rmr$adhX8@b{%mZn_0onx_*0YZsJ2R
z$&K`#@*N=IVC11lIeDGhRPd>y8&$v?2I|ursY(_1g#T@@9z5lP!6FJk3cK@NZoMTJ
zA>hA-5uv!4ky=Bz^eG+yL*+dy@F%c9SaSl%l=MyW<^7icryQ_uIKD1=?-f~<inw{s
z%^wrR_sj1HVLQND4A73#6nJxfTPhK`<<0s88Qr@Sg%fKx6>EHb&SRP;WuU(=wevid
z0)o}9YilE;qE%;|5Y)ry?iW3@I36iGeOmL>0|#wLJlkKv;`1>>eIS>9jiWQ*arzG1
zy%f8@ZTjxRj!Fy-A8i4G2WF)HIb<ujf@$)vq)(^SB3@nnO4OPrFyRiP-7>@T_P-uo
zXg5k)Qt?Oq9+sYVA>Q02#j_7?*HLh#Bw(yFNYolu<QWg!)7^0>87yHqL*J}A3mWtW
z3NRetqM~oAI=6eyAu0XSDIn_2c*l3b#I~N5#uRnLI<4e-wPksu(l!koZ9g^9s$d9-
zO7cD9O0vEWbhi^0lS$$4Zn5)dNHryi5X1)-h2mI3(}lj2azm($l;$3n-ulC%pl02I
z_<KBPxvkAjbzm{{aR{uBLXZd|RJ#<2Ptp{Xs;8|k8sd`WZ}t3H0XqJC5zWYgP?<e`
z^yn?niI_J?(I_6oB;(OrZiiG4siBv1S;$-LbDfH=>>GdK9ZeQRd#5=0yP#z)zwWoZ
zfc5EgZ+63>ZFmaYw@bM^=~FSi$-fA9N~^c<_|h8RR_=Sg5_ECt6DgxhcXGAIb@7=d
zNKn&HX_!h)#>1um8FK*V{4W-rWH(ObjmIoizA^Sm1`c`Ou0NZ%|JV8?;3YwSa6v0p
ztwNgj6*Mpj)3sq|&Mo2$67bnxYmK0`bDEQSFkA2EwgB(SnrYQe?Tz&Op67@BaH73T
z8<f6xYs_|#{Nepy!*n!+?&>k8;4a%rR%eJ}#i5DPk_TmV#_A7?JO8kKpl6~R2b~y@
zm+pQEP7R%-&u%ARSmBP`<(9)*e=B9_<BaU8pXNIDkN)%F$Mb6zR&d~_pO9WNTTaqP
z{CoSw*Fu-!`sg>M`HOK@Wi&7>>C`uN!`?Sm*}l)unkWN%F*YbJZ)Kkd05us4n>ciM
z`7g7l>s7!cjQshxs<b36%$f}|V;Bn^GI&4;_0gAy!=)wXLZ;Pz>C-{y$HQ?64zK2?
zNU7EyGo;z$+zm=kv+uuerO&jp*hYxMCTrMXEMdvx+PgmRIc&)3(le`Yh@sd6+foM$
z)V?f3-%yOL5P<9F9fzpM*<v3qg`b|YQNtF_66HlqXuK6OD#!&~Ve6FoJftOn^XkZ7
zkIcU0#k|LjKuWvNYkqSuGFObSHSUZSnAX;s7ZyRDL)}?|RID@m+5TjP;Ym>Ky;Fb@
z=Ww2S@NTS><ICL?dnIqI>^dst&})CRL~a~Q`)-<3Vzhks_e5{dHEvyP24(_p1Wv`X
z=(onpdK&Q&3CkIvnXyS1NKu)ITExNQ4za<l2re;@c7H5tWnr<vR{`Q2ek>aV;9n{L
z0lQ37gBjQahKTxiV~O0XIE28pd1);!aAH0W4pnzNAG};^_bH;;;_Wr&XFb#44GKB3
zT5DoFXGX`#ctPw5E8hS`#MA>(2OJs`+y@OMn$ISZZy1Zf#Fa#BQH;Y-q@R9%U-d`Z
zsvYuuRU3IS8;J1!VTZauhd*UYLai23u%9N-L!qLPIUEq+1RW&q8Vs%ziM8bV53FJK
zLd@ykzZ~Ovu$FDB0N6&gbQ4gJIz3?da84630ap;cq)8s!0p!R($ba_NC0-X~X-?}k
zd`JOtgu{h!XNL}<uy-X?gA<2&wzg!Q{1%>$>+BnH1@!*(A-eMioMDc{M8rhKI;Fw;
zHEu@qUlQCDLE$gQ%qh{ADDXpyV-h7TPXj_)6q!B{jX0J)>yqD}!gBmhw~Iecfp}8+
zF;90*VhQp0=oHTulCZ;({t-~p>l5BCyr8@A?k~PFY^p#!cu+#hObUX^;W~N^t{Y4L
z7#^xfs?jda@4sFQBqtN!KpibkevqHy-)zMizVYvP{$Xb)`FxHIks#1Gv3AKt<+vxE
zDLRuO;P%PoL%=pCq&2T+|MT2mV6l3OxGGL%o+WQhgYYU2WEQ0RQ~geIWY$7y2~txX
zatA{TY5{?Fw6b<RP;3faCshJpl^cBtPJk{{K_5QWE<Ql@=&Kap;iY}H>#_?V*0*49
z!Np5OP$pmq9^X1agAQTZ`MHB-Y&AZfV*FH7(l*Idb_Owb|41E@?P5LE%?kg}5(GNC
znm{ieG37YsQAq?Bf-A!!LJ?&9%5BLfvuflWWmFQZKTeLhGhO00tv#)DOI1%pK!ACc
z??Vkbn11|W{^yaV<&xR{+doCz*K7VJj^(n@S{-77<8xDOV9e}^5e$b-9>nzzthq1f
z^}>6~Qa{Yq0=I_`ikw}0>io(o6Zy|I**EgK9*db1`_Q68#A=PClE4VVspz=hk=j_^
zzvg&cXVc0Oe*iEys+1hnqp5kfk_M%*%6aTlu3q4PjQgZ2p#t^JdsQP$GeFj6Hl7gR
z*^sk%s1c{}FXXGi=93yj$$NyD|7r>J`v!6+-aq8^D%WCw>hp6R_{mw12#{#(vCDUC
zcwX0MWpVy=MMrqG{0yHm^NNRX@PLRs!uuTOJ}Nsqp4oV}ix4&qc)+`8xfTOn5flen
zI&-El(d^RYqIR`%2<D7pLX2ZFZee#B`bQ*f@_$NA>e<Fs2)bm2%#&LyS-l_f#IB^&
zJoIE+)acqrZw1>5B<|}8<*=>Xd74BR$C<;In%tk;%9l-AEEBp^u=Uf=nphgd`R|jR
z&y3iQho<*Xxo2N%*Z6V6$+1WzRCoxI_6jSiOm_d=78`@^&r4d2B*>Zg_rBX3+S|WJ
z5x4zdgAfr*^ppka@S{%D`02^=wSM{GkhDQ}udx>fy&j~qRw7tf1M%}eBP!%&Q-7PO
zNetHht)mq0{=ffUpEt|0(``Kg>k&d!%akR&hT^JVuJ8EeWxEJB(dYAWNUT(V|FF(N
z;)t}!@};SS#3f*6gJ74=5)$rXeh6c?|M6WZZiu`vi}V<3Z4qxV^*)d_xN=HnAdpi6
ziY?Zb*LDM9K=#D;s(XGy)V~}fA#vz3*{3j(yDD^W32(Pp5AgX@Vrsy#-v*%nS@`+?
z*=O`WA$t9<YsXCFKdr)(2d?7k=+#<%`%fvN&CO6(s^{}_sK+^(qT6>D0vOJlOI><M
zc=a}t@|m_m4&8cd8qnSjO5aOsuYmGHnO-7NMGkB)5&Ac=Tta!VMq;72^o=bqEms*n
zz;r-Ogo4r>;3)U}CB9*i@y}P*(cu8ZD&gg<WNvt!OidiRigVjZ98zt2@nd*epjo4L
zkC{SkLCZ^E0_#~4>9ItxE(wC@&dYi;hF<GX*+9hf%Kmds&(SUP`cvdrf%Xy-vxkUx
zum0@aRU-m{n4wjFO0BRt9o_jRenSRIOx=1N0FJ6Sez^Kk93@Tk+TMGCqG1Gw;S$)p
z&62z5nlKlJ3E7uGpAvw0@4wjsDsZWLM+q-8A*NOiCp*fcCN5h4Ce5?;c}Kovr0D$)
zUVQ(=Rfbn(2*9rN4m-4W+jZr^5Ft~3bZ!F!a`-C8P=-%!5Q;f;UQx482;mE2IrT8N
zf8U@h5n9QoJgT$6!A6B~W&gTw->B3PKP{=QnTv^~@~5`%84txed`jzhSQ`+&uz|R8
zDh?MtweJ)?bF9f2z5={N@^wt8ium}qT}YD1R}>dKTW-$nGS=&XpN^5SaV4)i`+_TW
zc`#sA`<~)`gpp)yzu6xjB#5sH7uT*mS8FJ)w!DKM)g-yfi0E<lx4rDW+ew5D7hkBh
z2bHIB)ds3our>Kbqa`GyPjJb%-!0)jf@gO~b$_?@6Ck2urnq8x@z7GI6W#=I^)&f3
zPbM}&I7jENlMdeeuLs<Fqj(>G@jiCN|9Zd<4Rlkgnvad<rMk<5T_l>OfHMhd%+i3z
z;mQO3DJDsp=ab=)=N&Pe#J4E&`m=|S!CD&VQ;=9~gZe7#FZ{?Fp)f?wD8e|6HqWST
z9FxD@xa#M9ef+;d`D}3qe;=RXpZ~u;AOCx?H%qyUTQf(9EpZ#`Rd<EN<`e$<A3tIm
zhMJI{b!<T@Y-<MIi)MNtG@2AU%BpkMPxFLhE3={LGc!s5t#~DsiEK>;>Axtq=ANFk
zl?-Y{Z+TuED6*OFFCt8r;VXsOyNP3%e!N95Kd{>M#VVB+c6eg;G-@lX{&ObwWr1!J
zPtix17zz3g?BhVUobCLLHV=&~w#)n9$rLZ$z)NrIMY6G2xcrF1p&;=)LR~!;y}(kS
zGrQavsJ*ces{hZfy(PT%W&GV$!hUfA$tpT3L;}KX1X=r6me&-dfUtj?5W()ZZjDZ^
zpl7@#9-YSUH|VI=;Gk%X3=#3=co1lmBn0BO0HLVOL=sF8Nkxl`B0;&K-x^>N(M;%C
zxr543c>i%<akZj)s|8~XI<SR*DMLsEnIS|l@ZO?KqIGA~C`WPg8E3#ZVw91XN{n*e
z6Fibk=NH$8bS-F&PV6a6hx>Il)pMX3|FUO4iO{9Ik#)i2I;7z9VqZ|B)j$JCKX;mf
zJSNLyGs!=r%mWePxb+t@dX!i(Nang6pICfgT+s&iEX9xuh{4vs{t4Nx8qrL()Lc5s
z(@=ITotCKC_%6MD2bSJ=v!ScA647T2Hhx20eg?}oBv#}1CNanGd$rS=HEXEc3*y&V
z6bO(INyZWTHyDVDWL8T{tKrUVE>o13GSBV3Jqi%Eq)~QGdm~2wc3qpolG^m~p`qUM
z*z|_P;7ZOff>yRv8EYlI?hL4DTkFTdS+gF16<R;tO+hbuMIrtyF8Ibx*ihgGYTV`c
z!6+IJI0FQyepXXhUQxnjY%c={pF}da|7Vhqq_~FiHWy~e@w^UXF46H6#d?SJE)a==
zdg-qa0=SF%`A5Wgzd1XCxLFhiYFXKcr;QyxEb*0_K(SUbWsISIMSFteesOZ_L@(Be
zBIUvMDAu-9j<N2y>+7PEMpRg%Mq#_GW37Q-yrj)V=uxC4XTIR9{)Ien@dfwnz_z-c
z;S`ahD^lJH?J=LP2N6d7d^#Y&6bOmq>jDC9A=!cCjpIS4E0~>j{X=5vE48vN*P5oh
zq_mhnzr3cn`YnJ~2bcG(E5FE-$P0lxGp?L$E%Z_5_8>j2-5u}!(my8w)Rz3=7XM&H
zPcAt$y`|2xH6ipkV#rNqISVfSwfBl>Ms44CL5QtDgjkX>Fs=EnMvh`7ENM=xo~{?2
z!l}R~-}NY9Q%HWjh5WIPc@i7VX`OJx7#YzFhxp*7z*^5mDrROt<jWzPH?8S9Cre0}
zO*KUwpTCx#ij8A+m7;UsbZI-y4l8`b1ZwzI%3|+#kub1khMb#4F9l!Mb!~~zgQ!oU
zR2RhIaGAwcuUV`@k!`-{X^=84lfBHZf;Qx@1&4A(PZHQ}hDuP|<`MbA?BSCqPj1K#
zKj}Qj;R8GQ(~~ocpN>F%y1jbteK!3wcI&bnp-N9QKF8phd1i2%MoXhkxxBQ5Va)-O
znGDQ9UI8Lp1?ql<%xmnB4r!R#)@m%w>L4-u0{X@dsWOd=cCozgFZ@5+`|@xs`}a+W
ziZ&s#lPz0{EQJutzKgOIN<Fq@-?C&&3sICpkv&^kB3n@*YnG5Dp(M(_M9g{je!s`e
z{O0)0AM?k|F>{PR-h*DA=YH<{^SM6fbzbLnUhw_Y+xPW0WylfqCBzWJvYvtl?4Hrx
z2NCIRpGeaQIX|M(Ww|yJ*UbR<;M$>gLgnZ;tQF`K*HKUOf#6~QYd)yba7tza*_mHo
zh<<+M9VnIt4G@B<z5XcWlI)AKyJ4twPqPTMIEJ|{=!31dT01SSAP%E&kkDwGnx0NX
zUt@DOf<6Ih69!+%+OB_YkH_EJGlGQuhG`|DUB=yyNJbyLZqd4AY4HSX&hELBeJfwT
zl!4zMDhE75-%v>G1fNOoV{Lwa;=gVQ6f51KVtwQH15%!wf@+iPA0X|C)+_=E5Hx_6
z4xQ+vB0yY3R(ynw1>%?JR(@En_oA7lE?TUI3fUSX0Yk{m;c-Nn8cZm(GNK6?-q+_1
zk>Fn7WKfRq>;JnwDG{_4Mx*hR_I&~;#<DzYQGq8BG~@+rH|+*xB~*IXSGvIN6Dq=P
zqA7_3Nugc{f@_3`a1v^6|D@|_2!^mDslW#%ntTvCe33gMm$Qb-E3*Fy7$!YlpA~Ij
zBnRTeawc^9pb+&@py~4f7soz-AdDDI6w!2q4YwA=&~?))SwfrvrUJ~9<=6Flw{K4a
zG!oRMUMB$rK>^qKd=dOWyy|vh)vv~ZB{lQtcZ2fH0gJbf8ezN){L^sd{96{MS_asd
ze#bYOYwb=DW_Ew$xl@W0T^^-O=fNy2#gWL}p4g*Lf{Bwmk5HqdqJhxm4rWAlrjA=4
zq!bZ8LE{FE52AF6sY{V7T@VdmCvcb&7Yg@SS?4`i<OJg%D{6TKh9pGDRJDA}gB`O7
z;*wBdQ1)-z<Bqnpw+!%S^})e=o7ZO;A7l%G1FDaX;}h?Hd$L7#wh_WcYHule>VDKI
z;KLFwsH3ApG)O?5O|*tV7~u&D4<8gwdlmqyVAnrlEyQRa$wc6K6P`XoG%^0|rCRxk
z{vk-&VQ#Xs5~nXg|Dl+Khl9SeT!b?)*SJOx5UcMTE8zu^vAOi;O>7#@zhJSRsLH}c
z;m(U@GzKl;MJ(JpjS8HIE*|XlRN7O5M6)#<77_M@KM+^1X+JdZ5=)BwOr;>1!Vo46
zf~_J_4cvCd(_W|;;P#l2>D?{Gt~-h??mXD*;g0YO^*s>J7f=zPXWtXM8#`X}4_^2o
zp8rQP7cDf1Tzq#j6h%8%L|9u}+jo8*9K`Z}L$r$13<6|<)#^oqKejSa08#0cM6u!C
zfgv1F@zTy;;OZ6KM*@S2OX#CyG>ckIQGH(vG~o!O3{DfzT8ApjXadns*B<8t*@8Lv
zOAOP)7eSYz+IjM17q%;&awU3c2oXIfU#M0^)K&<24FbC(LX&~^VNjU+i3$(y?`hm2
z9Be?5uvL@;fLtF&E9Vq3jmhlAf$^R#8$2<+l4v*teCI%C6(N8Jlm~J&`>_sTCxLk|
zu?a=5HNYDNQ43jI6?*B2X3hGA#Ryz3&67EvbaUO_D=;|&R*H3_2Z~jAa|jsZljF;r
zrzxBwO}>QPKb9L-I==pAt+x2z|8KHl{~78hLOE!Kzb0t^=Rdnx!x6IR%c$647;5Fy
z6tXiMm!U}@Xu5*Y6Oh_KXu;wuOGe`k!2yP0Nfwv?RCBXqHx^&=e8k!->GbKVhg6&!
zJ-zBt$6qkBm;{2niv=;Egb>P6QjVjqjEAfNC~;N1r=dvDR<eGZ^qzf5f*-TYy11gk
z|0~H#tdsxiv&1`=h_xE)^4LE6e_UEPOvpQa`?EmAe`>P|r4%4)on<Lb5kP{hr&ohM
zk1aLC-ioYs)!*GR;=(Ul;drs5>bL~+T{kn8*cnm%vkTh@4n%<!$@+inC%=<OoLofz
zB{(1t^x&b8UJgnE@tNT8(KSfqH%^tye066WQDQ=waw-zUDNZJqply$2QS)j8e)$Aw
zj+_U+wE6gB9E^78ibe2;>*zB)%V#@i795XMQriPA!^6DlYrLH{P&9%EJ`5*_6hH+~
zFGS5K4C+k8YNRT=dv9q|bB2gf=e@2x3qzO_j>?6W|JDK^5nTHL#9axfa*k!^og8@8
z5O#^Tc_9UJz|r^RBmsViM4@{I%Lm%nLcJ97yeCB$+P<JNc}uXdp#d#cj%`Y5byraQ
zFm)^)1@1)A{5D};g8yR<$t;c#cXSUC&E$aR)^Kk{QXAv6ie@S#*LFm8-W2B@DoV(=
zkV<4jb4%nSL(w>FEP#EHI%%$@2qhAfHMHpKP*kNf;0y#AHBMBDkn|9FF)wsX&}R^q
zpPE6z73MQ+Nfu5{(|8-icsGDny%Um_knP~S@W=U$PCPC2d857rl`wo#VSjgCar@8V
zgbq8#3U|+x8iUSWEvNoNey`>Xi#AQ%gk8ycm4HE7@ULpiXEti5!j$e43N0CTdmA-8
ztgvlN`W^15<uG^d7=4E>m@X)`A!0rv$i!%ek}NV3a|l~!ZaVcm8wEDF@!*C~2Mm6b
z)rloxRcLT<WOURV>j)*BTUu}5bYYCQ6^?_e<%xzHyl+xZ?N+^Z$JN)nVkQmeE^#CK
zY|NhGfg@G^4N<g?@5=%_YJvt#Mv?A6ZBAFckMx?5k;%xsywgiyb*PXm${kRrB&7Mr
zV;eC+pw-?mq!5W3a4&&|hN96Q>xxDioS}ra743Zx2Ee+_1&|v$5&5duxT$dTSqDFR
zN3SlwMBV6J`o~Y7fXu6~3Q*@0$_V<`)vdh;YOspS`;QIjnRAd2K&_iL*NY|c8<^NU
zn!(C{5M30;(VG_aVJvhXos@suw2@`mVKt-sUjtnL%u<dI@89o9R=ExoLCCsRp%Ees
ztXAW3Vlo}|ePP7N=_N3+H0Qm$8p<oW!EYHc<f>H5(gEv#d(hQP|CVY0`yO;vqEXBe
z7_(p$TY>qYRV|VM?6x?x>>%eNbas%MV`T~IuMlt%AV7VO4Yn+y8$3n4uHR&A47L=Z
zfBf4qPe}N|E+KcR&FO<n39-D<*EST*r4+h~N2o|a?Nim^PCVo9__2ymlHi5^QIf!1
zY62fk_`<8EFm~??==Y(NNR7!-U7LHIbo!><?s=y_3-jmL7$@jtTQ+>I`XBq!SJ!)A
z&c5KwE!z2+&RuaqMM33uv7l;aM6#~pE=uRpFS!{eD)x4I9T62Dx=vDe8Z+b!2zuW>
zX;<K=`-*adkRYcZgWeD7{Q(m%x9}(o$Y1_-NJ_jXhN|w@%5>x<X>k`mIxjKXX0K-4
zwb34`eGMSVx;}i^t2+koKj22hnP|+YCB6XYz%f-JT}Qa<=%|%{H9j#x82$M9$(>5h
z$q}h}Y6Aaq7$rh`BGcpo$cJ&JEozM&_~oSJeS7y>fzPli=>?BOcx0HDDbpzX`TfC4
zIjFuvfi_~hZ@9O46CM{z?L7G-K{+`Ny1K1UiW0yR28DrI;84-HjOw=H%Nll;Cu!@N
zRSa~BG|wk&RhE^h=%2}SD_PSQuWckN_iN?YJX7Yk;%;wGOg+JFfizk~M5Mp3j}V-p
zTC1D&=Td23VPPTWEWq;m%NHvxttPlJ_;`7VSXTDEdZnm#)Al$%5Dceqb8sA(nw$Um
z+QhPP<3_+jOl)#;`h`60``DOjrPSfUDcE?Tyi9~Q09s7xWZm%&pBw{8`}_UStU>sw
z8>Xj4WxpN0Oc)VkJK;zG;_mG|?dRtwB$Pcp{0{&62_v(!`06vWvc_TJO&)>+BHE<{
zR1`!W&_qVN$Ugypl9rxsi`D_3g{Q^E3*i1jLqqY1&;_`LFbiD$8}hC9&(|9o8jxi0
z^70bD`S|fG^t!{F9w!!j-ZDc&l>@oKV~L1NqRasMPn)fP&S+JHYSji4ReNTp>%SA=
ze04fZ;c-GjKRQ`&-n<#&KxkIMKWIE##|E@Le?D6N7L)kfuCCbiTPl|^cY~Y-j_$4q
zANo1dy?$)6p81a-2QGT!oxvwY*$6RQ-_61zgOT?Z92a0d3t!NLgald^K@N(eR9sU6
zcg4t)UZ=1JR`bUo=y4ozt$XuE#e`Br<2|A=91=r*XM<^^`Vq<P?HgrPO*6|i%l0ku
zyZqa%lDoOFKi~h3+ZrkI)Gm`-=-ycTi|wu~XC<uim(Q>$vG_Uov1uhpO-SYS<h_5;
zz+ou&!rWVCUTvJ^R;%+ZYMU&(BQ4GYBOZK<QWJT3IhG&wJ*vK38IS3xt)i7V^5&_-
zo5pK%QWI5Ee?o>g-}+_OYHX_XGOKi4dpY6GW&x%UCbMIWWfuBOXC+>{1^gJhlCpWv
z_5&RUJK8_kk?gX(3}%f-Ek9bAy<8fxGBAxH-?_P;ow01w+BCf(GGZh5jmR}M*d4J;
z3f&+cR_n^d{2oU~i*nC`e91hs(q|PF=G{XS?nV|C`zK6`LPOu8YO69a4h*w7?<+`Q
zm84;aX6pKpvc|0&!~XKiS{}yez(h-Vjw=J44(IP5ao4V?X=-kJ(8RE^zkW8<rL+&W
z!pK?M24ldZm>U>y+wY_<Cp`fovBneI8;=PDYa{h^Y6GN@N4Iv3eJaW8&Ot`Q{e|8W
zGUd*m9)WzW^#I|u#gnxKII?-`S@-Jbv^;$(ylZ88W+vIu=;X<_GHlpLz5Pd|rI8J3
z&b@SAm^c*}Hf4DH_;D-~J3BkNl>xz%3JPRJ2SGNRtZN1otCxT948aC``26f_3j0Ag
z)=dm#WoKuun>zKErADQx?~|6kw=i>9Mn-6*6_Q<^ov<a;%Phu$y{pvW(xn4ML0QS0
z$T_G2^ImQ~%9p~<7^qW!L`LRfl4?kBaCmo?5QRW*B;rS!xaQ_b*4nA6?Y2_7rrliw
zd{!}yw)Z=<9{RhoGmFpF_S1D(mQHvu6}m*#><?>c@w@VT<UGIe)H<V)_{-($n;RDn
zE{GjcnjsB7@t+O-6I#3>GMoJ++ocq_{nq?5+9NXZ7V8BSF>8^t^Kl>hcfNQ(+Z?ri
z^5N;}Z`txHNvcYAN^?bHuiCCMt=PmjA4px}=qYo{?p|cQ7r5%onNnOV>Ez@jR=U#&
z18mS6w6{K&nS}-Yk^Ah=rY$Ti0B8vV<E1Dyw%^^q4*~MPfhg8d#q8o@cd!Sm)}#yd
z%b_UrBeH=bb16+`-vC3^UU>M<9a)s6E+rvdot>MbPdLXXF5dp;%_xq`!YVKTqCcue
zbb3{2YO(Pt)hQTu#XPK@ds|*E+w)fKHexY3IXR{p#Uv$(7GL)EE_?96R9QKAOh|k8
zU(Y=rVY^0@=MZs&<>b#i??K5o0r|C~WA0c=$H$KZWQBdmK2g!QEU(MzT3WmJ6%-cU
z@n7Xuq(K<M<S^D9VZun9BW1atW9&_)wL?ZyOUVbs*hG2%W*-L|n?pVe_L-93ez|FB
zXc(NSq2Cg?Gf+o4!I621o_&4`#nI#WCc?eKml2U^3ygGj=&PqbPIInatnNE|i0<tv
zitC)=yxg%956jv!^XPt^XmgctGrnnYw~(h^^_}Xp1Mij~&SLG}GPe(17V&dQu^vqW
za>Y~k0*g1y8#g`-XV!4|IA@+D?VsGNpPuu%Z*Y0(%kWjrO=?A^8{)^~P34N;Hr--5
z%loxHFtGTQ_s4mvt+b8vjiZP6mP_Dp5~JR+gP7_N8`}jm4dL$i_&9DUsJxQ1HJjY#
z*H<fK&^QQd(b}IR9297I(KRsGx@{Xq17Eyw;R1<N<~4g>SC^(V8>Wuo;ahkUBWdB$
zjDujt0vl@c=FOI+ms7K{B5H;XNl3`~{XT@5OUP@ItA0k&&xaDCNinf#DaEv$oE#YU
z!l7AEu;K07P~#y;Dp4V&{lvbU*Ng9gk_-m?uBWA?RaLEn1`-q$?6!AzbDPKabr#WC
zS%mSdk+Jc0hr6r$Eh~j}_byy3Ie<wmtzBJTMVHZtxx8Zc<U>!-VMx+mypXP!!jXx|
zUVk7c<QcTV{s*hR+dLf}gaE1?m-olPwNhVglU5XZsNvtNGFT#{3nPBa5>g-}{ilRa
z;zS2VAIES&d;6!Uspl?U{5~>5&&sO#M|KMxD=Pq=^~^hD3OLBknl~B9`;<3caH_zu
z8J&kQm`~!nRcV5=vxN@pP9qV?%E>XZ8K$QQ)QPU(nsUplPVde*!8vs><LT2Ir={V}
z+h#i7&6#TWpkK;gTA}Qryj6wW!q2MnYZ?VBBUH?ch4v$2ZEjEJawK22uP)vW+{XWG
zsAhZnCLKEScgI@xUePYk<@>YcRDlVFPy;`yXS3D7<}2JeC-{UlTR(O&TJ9cjAB)_`
zgiYMo*Cz{^E-G`t@DgHTAKKcSFI@_ci1?DU#98lg2Ry#HhK3MQG)YN_xC-E-3--+Y
z@@1lq_Tj^aI;Tm3)<PNXj_VN=1HWD_g?*?E)uiEVC_z;#TQGq@@_`mjaD03(o*hnJ
zm@GM+KcDvaaT}a<kXhhgi8<iVRla)l3ac@Bk?$KzvsD3dR5zrwAgQ6<EGsKpz}$3H
zv0+4tO9ZEhOMU=EoreeZtlM$R;k;hJgL=j7AmGsag9Q|W@B@H>HI$i|iJ?GNR<Jz(
z<1}C&j_U@#zUCOc8W9meB9SPl+1S|5eR^>7=Js_`B!wcDG}=xSiYxq({9!D)nEf7h
z_A4kVQ3+uLkWMrSzVx}-*>w9>p}phLeD?MiU=)nHr?($xFTBbKe8;C1KQcaEhPDpe
zTG%+dNI%3|!A}ek4u-VOD66fuFJA;mnqHTh0B?;owYoj+$^x+26Sl=Rw&|T)ndQ6-
z3m;%x#W=CWy>EK|-oWJjQXP{7%@WPhK;^$z40v9ljy8X0W@f9|JqyH|5f9zc)Pun(
zl?9U4{SN0IPH_bdZi#=wYFN2Y8G0l_EpgxEwbomQo{Zaei`I%L7xU#Xl^nN^<G7ns
z;P6D}iT#P-5~|<B3de;RexNqPCPAW~HwKpFf1Zu8?iw=}ToJlqrrI%sV;@=h^~Lz6
z^D%iZLxO_H2ex8P1jbEm$v$sy&q_x({PioJx6|dyArzqj018V=7>oS@zabosx-ZSM
zOfJ%5v~}y&VXbX#9UZdhH^vOtNiG<MsC{*o_ny0;rgp9SWc@JkV8^k7mM|6U;N7ap
z==3YbEz@|J1o(^nNGeTD9fkWVOldp2x=z9(nVntx!5SYA&l7-`nHtJXn*dHqii(Qh
zNDd)d2Q|VEc{|zrhK9=ZH6N0ekIxEL|J}H_=RFf3?O}+n`|b#=p!m4Cm1e%y-*AtB
zqx8eEZ_9YsZ#k8MPmi5C#k5YLCLvLTwvccWwquu-!l?H*bai)|7#i}dC%$~q(4amu
z|48wfCRagG5wE4Ok&(b7E%kuvu<8>&;f`!e^t*|}A`RI9$ez&CM=>SE#pU4acW-;3
z%j>@xA2d5%zUjusEbjXH6%_}+-$RRg&6a`vue3!q2jeT<e&)Hnn`Du)U(8=j9$+c3
zt({9s4>PA0au81a=`wxkajEpE>(kHO9Z6|3@$r}V15YcODw%dalHa8Lq~iIXcYPfE
zZ?u?c#cld3ogXtQ*$uXCduVyc_x3gyRgpR2qt0n+52gyge($|g)GU8w=PmXL_iB+D
zl8vd)0|ho6868HJ$96diI|J9GquL-PP?nYAZoi28D+fdBxm;vF$ZdlENOE!ptXjb1
zRc|26tDk(ClCaI1`K1-;^x^<u;z*1`fm+;1DG60fq^QFcJGO13Pk7>m+I{)5DC-V>
z0Rgo>wYpK9#^k(M-i2yLJT@c}GwqDL&L=@3A=n!|>DK|OoR4uLk<j+UMniMv+&RRN
zb9QzwQHq6NVvEfb*Ougr4AmK)qi4^aE#EDOMd0Az0PhBQj6oww1Rgxm*4Bm`jogiu
zl@+=Gls(ad&VjTH)Y2-7uCgv)x6N;G^7MQW+!V9f?ZVYy#3*<@FY)U|zc9<Ks*+zH
zi^g=xUHvrrkDCm&{@khbTe7{^P|W)HcH7|cxeROnGIENeYMfJUFRRlt{zmyx==;<6
z%Q&7nqwuM7-=lY%1kSxQ^0xck(ylWoZ$4ny{PPT>6n(m!mTAI^-$uhiL7efu@e5xk
z*j_qKjMsKnjjYcA#0fnRGL;r#>${a!oVM9$=asTQ?N2SwPv_M(40q@xYdU<0sAORC
zJ~-z*Z*5^vrMWVr&(&nzeAX|N>dpc8y(}#bH_Djj%YPnDziQB9=7rUgJjkGDY}cvV
ze(#sG<xKy~(@Uoxr*L%-tv&gD?U!h!Ek|O`6@&hP9*bU+GEoo9Ggh2#TpuN@KdSmD
zd%k}pzr&{R($KM8Rew+fs{3fQaTffEiZD-JTH0~#*o!0*Z320zMnQ4$*NVFOdI6DC
zuX66)l$Aq!qiVJTVOv)E3M8k=8jHxuCHRssY*4>VOYGo5kY>WaB#M+qKxuRH8%+*Y
zS<dCwCR^}98!=wun+Y%Nm?Vlsj6hf$l8=U+vvAQAROMedC~C#g3DH}$g;%pc7P1tG
z5R%A43X>|p22QW5%A}4Vcj@q%0qYV&SE~yB_*u^Tx5myrRYAlxEwG|=+op}hwJx^`
z`UT_z->e=*>j15gPrjycYe$E2Wo(XVn1A-$x5p_9)!Jy5cmk_WxYkssNBw=mR!FLC
zUZ0@7{w5+alH<UcfbJ$A`zZY^sfSWO^t<$YuB*PIzV*iD&7r~wGx2X)4hz@q1CMdD
zr5ye3N8jGC?tOaJ)j#*M<>NbHht?&8y-l~=IpDIlNA!8-q}S=$Lq|muPoD~HBeA=)
zcf7v&O-euSa}nc5vFjAZ2buf*4?U}!zeF7GR1a)lS)ZL=95gE!l*I4_72`|J%PWz?
zU;=qBn8$sdoNV%GYi<3EL6i6H?d$BI+Cc4F!`rvBw3182E0@{J18(RCF=n73XK$4+
z_KTtYIj^Zj@~PJ5pxY6-xjzBfDR0p-Fz_w2gMoNcrAu~gVL{2HH24SY(#lHAsd}}J
z2OMAg<<F0fj-mugpK%5SIscNz$o1#1+S}R`WqxZGWTYRz?COfpZ)R<6jcgmWXxQJV
z<;=o*2%usn4zM2W8zHNDgn-Y|pVyc+ievJfPVs%q{;li%_AV~ZxA}bbkK4BGLadv?
z;lCSgSO0;jdXB$foM^p44v-p)5pn7-vfLuK>#Msh4E4<LO$$wBWoODPdhBrCdU(v|
zfP+eJjzr1N&4Z$=<x#6r4qLR2i-+Vr^Zvc|eMj!+h!1bJ2ie)~qTqV^QLTqRdX@8j
z@oM8$@s8JvIRkN|k%ZF3(u8XX3PT_Q7KMJcccI6mSmD|c4L?`;%-NCGpOnW8j*_)0
z^WXg(+1l}Hl^kt*a&5bpX79D;l66Wf|Nad&<11Jx3Psr~tDZ=8!QsGI74us_fuX1!
zJzi<9=d^@=dpEdkqVR_)l=*+5jhac2d|Qdsj9C$*sKt~8&e++Khm9LF{a&R^c-eMj
z9<*w|sbK|}fIzqUy#luxr<mwjE8*S~O-R+~7H&VPVA>}n^kZy{rj6^3xMQC*>2c-J
zCgXh;4espY$*F(wkosNEdx1B^U!OmE+Qh`H(zge*RTqb`H84s}R!%O{YZTwq85x+G
zn&L7V*Y9Jl$F7z2V}j&DyJy1xT>s5*Kth7o^1Wo%ZGWMe26urizR?Pwxyx)n$B2*b
z+TyF;)z${c$K2}{hGC_3b0JoL@vm{x^<Sw8GWI`@#JP2C>awz<)?LzlDsS1gZP4+2
zyR2%vVfylLOZBzByeGeyg|bAfJo1Y=*1xqR<fi2Y-U=M!@>vB)&Ipk_E*X-<sQzMe
z5a(}}g}8FDa=!ydci1SiT)lYB{DPXjH|LtBf|zZ9RbpA0x4L>n*t@}+dq2j<Veh&N
zL)dUnxVgE(GJ->#^{=2)T$3u#&%cC1!Hyj}h@Pp-muHY8lI-kMo)#<mop07j%zo?d
z(e^5RwyN&FH2KOMhbK2iGMt_T8=p9YEC#958tslSr39|Mnr?0qMY^>PS@-WV73DvF
z9wW~mcL2{Ng(0Fs%0RW!x3}F;_$)^+WENXpd-6@50uAfi&E&PlKX~A?zBXt%1E37n
zS&*`5PX4s1X=ca>^DU=?Ymp`Pvoy;m4qww!%caH5p4c83IhGix(@exC0Rua`A0Rn1
z5}bt@gpM461k89YHzP!jOI4sVCZ0c8C|HMTaHv8l^r&zUXNL)O<)WxvvX@!yclxS3
z9IraA>9)5_yIz*LxYGaR^yV&Q;|)7kHND;5q=#1TnR^#Fe$6ko+b>QVJYP#)ovOBW
z;z8@sYU2T@{xM5^0K+<)n{P>~MOYCfb0igTA&K#gX>QiuRdx4UmfEwC+`<*ZEu>~f
zmHsVFiE<A#H`UHUWY}LRB26+Mz3?qDoXzYJO|sL~-N#HMS$hvV7ls%a8I7!O4b-vL
z*QdRHtD4?85i`G|cGuDWGAvKsL%EMf>k8w$s>HNy1`<8?3qavsd8dRKCdVHNh5?iD
zKSusurqRJYuu#qasIu}3+D)*4$wz@ZP&GJRb)~}4=SP8}dS-mDvY#M>^nR12%8c6h
zZFUan-(x1n=pVjP@9fkLuJxVjZ2w@r6yBN&aYKd$SZe<T2EnyuZfs{Uj*Ju~Ur*i8
z@2JTo9jz!TlE4iDr;ELZbbQ1$3tI?RO;mG0$ElMPe0(t4kz;vJ*Q-}&A(}#+fC)zl
zdS!3j+Of>KgrkdA`$f#RtlKed09|zgvgRAyQuC42<5nJybN$>?U>)}IrOYyS-1|S6
zu)@<5wNOwM&Ock&^68Xt?9BN|0h`rb8!ZEZN0XDAM30-+vAj)=YPrhO@cyD$U2CrG
zVyv@^3%4amb$~myquLRi;h@+q2Sip>$gNKpUlb4}aZVw9-PN$wSL}owd7#c=bq$S*
ziVBRX;nzim<nyJL3X_a3dURV~y`x{`5z$b(T;Hve5mNf5bh8t?@IKKDryTy3*L%~C
z|30!Ln0hUuy8GImMX|3r8bS&;Q;*Bg=N4#1NIW79hI0f8#Tv%+4NYEMlh&<3C;0U>
z!N#PZ;Ide{<-4aRYHZcgMD}4~-7S3JUoS7hnH@zsSJy*M0c3IS%|U2GiRS7`!Ps~*
zys+@lWvhPVO^Agda!Yr8Pu`I=ZfY2p?_p3^S4RPkdp({i;#kvQ-@pJ$j{EZsAKF7;
zhQh+q2@*t2YL{!2WbUS6QWThug)0H>WI>?bXA&3Mi!Z?;81c{W!>)7Iz_ktzBKZpJ
zTsD8SdXLf3(e26>5@dDQlnU=Ju`V(rFBF>b+&qQMyE;2j@QARmHBc}!V_o*ft8`GC
zbN#)$YM#313@kp!9R3?7r>2IKFwJV5`ZYctQd79BiA_8Qf;3{X5etvEn^5e$n))tI
zpV8<*XUv~dcb*;e+Pi~cM`5<mHZdliDA`+_;lLm_bM)ulwdIyOPri{@r(4sPp!F6*
zn?oluB3g4Ue*Q3D<es0I4~-mPC^#I6$-GaT67)0hJUu=C8CjiVjA<%d3w}{drL~xn
zhMen8cwTyRl?sj_<ifKSm^p>May3uoO@zp><JdSkN$gXl-^=Qp@Uc=o*{-fG2#xj^
z$#QW87O#O2Q5Si5;B1hx)}3pR@cA>iB*KjU!w3C>PwuX+tWX~VF&k^QQ?%tWd0=QU
z)z-}POv8>A3sguZCn$_(yoE4V<8~f`G%v@9k)a`4)&vIZke@*Ts{6w=Bm?LS2z<Nx
zcHd_z#6oQYZ2(A=nc!SpTnu}ZXKrtI$;qf@1{WRG67{mU$X>Mb{foHiu}X7~PHjFN
zo42h@bE8TZ8n!l2zX+yJISKH6okVGT&iUuJzH|?cf=^k`I~Xggrs!<ror79afQAi4
zxwsT-D;Ti%Al&lIo1dT0)VOea<FF9OEFqzg$*`Is9{%<`PpvOHgYO$zlOVc_%3G$3
z5@I;~_z{?c<aRyG3^9A{*Wl6#=Af>i2VBD4sqk%{kPc+bzRnPrvH3bS_QIx~mPzuV
z<jwOri%5+1<i%06DJ<a?YgD@302D>3-fdruIkDj*(#Uf_&Y*^EQsM2UqC)AAu(04x
zI$nIXCbYh<IUs<HG=y5H@PW|q_oK&ERP^iwR~+aS_!PPQ-0Sx4AtT#GNB9Cj$#<hE
z)+ndCTH)Dy0Vt2a#Lzk}L5UnA@D>AL=V1u?uY3OPYIXW$+;aqn<n&EPLr%yAablbb
z0T`6dow%{*c0z(k%d^jNhgCBvCTUY^M8dWTJ3e}xR?zR`BifYvGJjJH{}-!aF~9K7
z{+)_^S5HdkJUZEY<-W+|n&@tG+I?+?A8*`e;NC0r@K$JZ-&q%fRfE^*!4JN-HKOuK
z>3>jh9(9cAJpVr$-A*WED<Ad8?4p$OyS}t^`XMx(T_LF6?y%U^_n)q_xjv#<71(%3
z$|uC^2WA&XyB`!z#JsP1j^6h!o%Rk6etzpQRD0n@iNt}H{mzDtci&evNDZPGke;P!
zB8xK9BS#|xgW)e<Xc-xeEi9UcE`c;&gJ%E+fW{p8eFolLBLVchr>d^Vxv3R3Hfn-*
zKnefrw{KDs5^k=phDJsN0ggE}I3as`y`grep7?pE7Vnoyw+G7xW?;mq9>6{<T2SKP
z!ZhM2s?J=rvzr|>(Y|*eP-ug^ygW%98ufbWz2~hgs;-znBHYbBBwQ*vcjgQY6O-E7
z60E6sczLh&li)Km!f{YZNvT)cDA`VVJH=6UippGt%s`#uzbHY~*S9<`Pt<LQawlw4
z8#TwxJTvD;I8$kN=Ux1LiA{vfLu>Gy)7e3d=W&(Am0=0*uD#)xos_-rlsK~I=<R(;
z8QVD=_nbMTprlo!`O!XbILLZy<aZMF@uY?3g<%g~KUs{r|3TxXxAogvGoyUj3(}mR
z&sHBf5jnTmvTM2ZKIeZmu}Ml@CfPLeAC@WiD7WSFF_tdZ0+Vfi48wa!B*7wCad8(g
z)#V?Nl|urN(sf_jIXQ8_pItFHB&3b+xikj8@(8_{jVy>;e?NvZ07}=st}ec(<PVw#
zN?rJux!ov&lcS#N0Z!G_)Li_5_}=Eu8qWEbnBsQXgvKmENy(G7fiKK{Lb6w0QIWYb
z1}k3Ur~>3Ok*Os492DoN?+~Nr*ZQLo&oA(|UhAsz_wO~rz}*JaE8zJEN)RguP6o$b
zcK1MJcu3+RyNaMLBmlrfnj)eC0#sB~OouN&2@DJbn-1DCC1n^5;J8}N^X|4L_V&Wd
zj8|RpzIZccOjjBk!-%>+D7bcPhldQmeaOjk8ChB0Q=bDB9+#9zS#Q@Zg?3ZKBtKv=
zzGkQ~N!}kIYjSe#8cyYG_)X8$?39n<C5DN{zRkE!JbDDh7bSdC^$bMfMA-xB`D9WP
zG>S~j%pdk|)uxpjVpNRq+L_JXRn$82@}Z>X>6vFEK0khrZXV^BlUu6fZQ_~RA2LMO
zrrT~J?%tFdpb;r4+wkX*)N7&FhH-rS{RKhRMoq^UZ|+L_;-)gOJzoB6?C)617yh@)
zg-g#kyc)h4>2$?=c6DYJk9z6I67z-jr+YTZF@TDkM@rZOVt;se7TyYQR+5yEfD!2t
z#2zeRq)8(mpAtqQFoEl6LY#$6rP;?^PwyZYYgF2i@jFXD1S~>n995;3B>AMK^P(2z
zCaxNotfZM6z_4qVGhiq<yZhgMq4)e9_TTH*R)Aape+?}yd6}6PQ4K=~z`=}hQ7A#e
zHN80-w$e7EscG?YK9Yww4Z)NFgf!@6lA0P-SPhElM?8M^p`H#v#KOYD>wi!9o6+-<
z5`LFl@Qj2;7dm!usgNdEU=RBE_?Vj$L^?Em?5wP^Fa(FIa%x(dSUzo!{Mz)+z~DdN
zi#wbi&JZ-HoL9t2Pfuw#(v9G-2QnVH+gQ|ppa+4^N2g58)btaW$cy|Pm`8X8Z7=F1
z+8m5H3*6}oLjBzhrWzX-LyQK$H6if%k~9|=7ufXc;o&c!Tre`){h(>Th)4xSP+q!(
z(WqCU8EVMRM8rWFjp_!~CkNwj<}JPy<d%_C|D2_Hc1lrJ7W^gRM@227<BOa~d9ubM
z!!~cmgZTA7=sx3i3A6w7vwpy3Vd&JG;?3SWp5{?!En@j0(Op8R2U?)PDt~#01dQH*
z8K+7&`|Fw-1ATo&Zbcg9ix&kPA~ITATQRSp_52<#RHrvnP~c%*<qXFY#otgFN@FH%
z;eB1yCvmar1N%NaD=Hebq|qG`77@XouK(7ENwKhIA}{I%TUq&M!IVn;d!2tLbh3>@
zm+EYEAcS6nu$T96q5dfo6HhO%>DYB{-0SUZM-27C4jeq#Q||RV+#?hd`FeqF&}2l%
zA=)Oj2dPnKC-i!z7kBcAGhBCDnmUWT3R*g~D;rUBD23>(wN*E@SStYMZ&}&zXXoD6
z=4#c8qeBNWK$RXP5egnYKI%J5%gghH<+30};n{@3RucfCCoet>H6HYIU|W0p^~gxZ
zHl}weGXOH6kmv4I^P6^X(=3%qQV86Y_MH8^6gYfhk*=s#aclZHw-4hgHI(c3JOpke
z)HP?zNWE~>HKr@-xNPLo8DStaaOuMFUiBc2OHzr0CT(WZZ5fd!6dJ3WC~&)dY*+F;
zuKfAv_V_T(qp)fYEnPV!r5cpwYdSmcE|ef0!g7IHV^@K6{Nu-GQOC*mEkWqLefu`;
zEeE2492OSFBeDA&P&8^re^$2cU&%2+VF$mk_Xs_LAu=28sbDjT<veCZ8jSG-8IdG*
zE@I3YHw=PFc=G5qEA!JP4|JguPQM@wV=aq|(@vSH7lLDg8QB=Bamd0MO?zr;hJoS1
zQLSdi-$6ANdF&C)v5YpF1a>4WiO}E(3Cg_(4>YdOw{Zwk5n^h2dCLjgmbY*DBqbMM
zy6~-~;7tHp_Yf6k<1#WcbIeMk>m^-x;e(<+30#P6l57Wha`UE5P?{Oq+p@M%67_xZ
zq%rG?bN@OhcTqfkl#y<-ML+BH&oi4J?vP#_dzeox`onVnZ-=%?<C6RKcem}4xoxnh
zEh1mKda7Ui>~6={zK2N%s`86<6H_i7ksk677L}_!RAS=W8&(!ursNeSJ1kq(v{bxs
z^YbyG*7h^LtgckCv9Snc6`s>pW@cuX(}wCuXy|clX+rjC^@bYd*8(4+C-};hwcFe$
z2q`1N%lJ5H<HoCCvzL~ZC^l_^Nt4d&*g2Rz0}ToYe8=rvUS0-kj{5hrCIr84%}MJ(
z(1y?pA&#`RK1xZkwzmg`s*BnV<yMA$mp<RU00L@mejdV{Cio2t3)?$7N{WmBz@Q?y
z&0@goxeKAI#y2^`vAxsN(g1Lf<$~b7zj(Z;xERNmiK!_F!M@H;H}pC-_C-xWN!X5d
zgZ!yq)xNl9l#;|)Zm(kh5YG9y>W2JDXtYrHcL9-wZa4%ZOv@_=3@R<{sjjZ>?nd4E
z6$*fh)6;rq>{3UEh%aZ{o2`udZES6cardD9aW=%q$0Ku+l$3<#m*}sP^%7fueXB6T
zu4EU*G<?Pdo&XdtU@@Yl;2YYlAZ<WW=-EDZ<K|72jbIM+ZD<HSSljA<b?zn>BT@7e
zy)Ws&JgNU!F8KO&4Kx4%#(@BSfPOPPsZ*zxf5XiRh4ii0A>Z=^>RVa@Fh4QlmnDl7
z?;u&A&as9azB9^l$^4W@Q;*A09*fd{{Bh!=V!fjL;F?$BuxxszHbaGo2}Sm^(r5W;
zQ;VCcEPXHE?tgXCjfMAh_cfcU#)VspiISeZJTKL1gv0KCkI&4`G?p;#7kC-7df@Qz
z#Z7U#nx_LE#vMK^>M|hV^U?3o%4R=#&v$cd+-%$%9X4w7r{|{rG#bv9xBTOkGV&ID
zvNRiQQ%gW%z#E0r)KSVxN>{Om^9NCjGp;2k_3PQ#j6kL6nFp3k`Js#}o7}f7!f}^;
zIsG56FGr)I^mmv(X{)Nb#lIfwqHe^%>7a*)N1*RG3{AjnZj-VG(FOT=ax8>$263Gm
z3%r|?T1(VgN=aLM%@sYad7Ja7LTDE0PIa-JDJ|6T*%M(3&y_=F7c<lt%w+ul_~0aW
zQ!svG&;iTcWD2?p3IWy1Sy<?x<%zck8!^02FvmE^vPj#kXJ%(9f4U(3xL?(>A%LAS
zFi&WVL2subHJVF+D$&!Tp+TIYUqw-`_zl~!oUDa8O%#mqk%mQkkiTLHWcAkEL)9@b
zIzRuw-9id7?;R9J@BY~T-@J^%Fg!GA>)6?tn0h{ZNXBa@Ae5{+Ey0_G@~eKuUCd6$
zoZ=-7G^!jzKd({I=D)AfIsUJU2_!B?K?(XLPABbCr|`xrhMkG*;qunn2mq*h-ZK-^
zF?#*_b?>K7^&D<)ZtK6+*K|TAHv}YZ4AdzqQ|VBOA}7=3#8FG8(9++}CNz*ivz}0o
zItJR_%y&zM9wc<A@GE;Vev(XKXGnGWvHrc|T&f$$_R;)Cllvo~GUwsYtOt<ySJNoV
zkp(1f`lm&l5d4$fBRH6wNEIKhSVB#e<+wwY!gxEG!uch-UbdkuxPk;kBLn{XyZBGQ
zVN@gj{ll3Z#Owd-i+G`be^2)H|NqIq@5TSFJI0*li2h9piB1q*ojG_?>`#)DgNn-^
zJtE^m*OZ}#NmN#q+2FlTb(gHk%F>!+wT+}UC@cD#EusWA_xaUEqY`rb)rE@}6;;W}
z0&+7`GE-7Am57(3fjbfw!araCE6=&+%jME6$BmE@qDUScP5bw){=VFRHc})AQL-E<
zX(;~N)fD_sW6IxOYE<<9+r{+v2Nz;L{p+imbq;N0|N1(x`Oja+gTFuiul&if|N3>y
W(9&Er*f)@osi|ly=N>x~^j`oicdRu4

diff --git a/vignettes/examples/vision/autoencoder.Rmd b/vignettes/examples/vision/autoencoder.Rmd
index 926ebfd28..ca7dc5769 100644
--- a/vignettes/examples/vision/autoencoder.Rmd
+++ b/vignettes/examples/vision/autoencoder.Rmd
@@ -156,101 +156,101 @@ autoencoder |> fit(
 ## Epoch 2/50
 ## 469/469 - 2s - 3ms/step - loss: 0.0712 - val_loss: 0.0692
 ## Epoch 3/50
-## 469/469 - 2s - 3ms/step - loss: 0.0691 - val_loss: 0.0679
+## 469/469 - 2s - 3ms/step - loss: 0.0690 - val_loss: 0.0678
 ## Epoch 4/50
-## 469/469 - 2s - 4ms/step - loss: 0.0680 - val_loss: 0.0671
+## 469/469 - 2s - 3ms/step - loss: 0.0679 - val_loss: 0.0671
 ## Epoch 5/50
-## 469/469 - 2s - 4ms/step - loss: 0.0672 - val_loss: 0.0665
+## 469/469 - 2s - 3ms/step - loss: 0.0672 - val_loss: 0.0665
 ## Epoch 6/50
-## 469/469 - 2s - 4ms/step - loss: 0.0667 - val_loss: 0.0660
+## 469/469 - 2s - 3ms/step - loss: 0.0666 - val_loss: 0.0660
 ## Epoch 7/50
 ## 469/469 - 2s - 3ms/step - loss: 0.0662 - val_loss: 0.0656
 ## Epoch 8/50
-## 469/469 - 2s - 3ms/step - loss: 0.0659 - val_loss: 0.0653
+## 469/469 - 2s - 3ms/step - loss: 0.0659 - val_loss: 0.0652
 ## Epoch 9/50
-## 469/469 - 2s - 4ms/step - loss: 0.0656 - val_loss: 0.0650
+## 469/469 - 2s - 3ms/step - loss: 0.0656 - val_loss: 0.0650
 ## Epoch 10/50
-## 469/469 - 2s - 4ms/step - loss: 0.0654 - val_loss: 0.0648
+## 469/469 - 1s - 3ms/step - loss: 0.0653 - val_loss: 0.0648
 ## Epoch 11/50
-## 469/469 - 2s - 4ms/step - loss: 0.0652 - val_loss: 0.0646
+## 469/469 - 2s - 3ms/step - loss: 0.0651 - val_loss: 0.0646
 ## Epoch 12/50
-## 469/469 - 2s - 4ms/step - loss: 0.0650 - val_loss: 0.0644
+## 469/469 - 1s - 3ms/step - loss: 0.0649 - val_loss: 0.0644
 ## Epoch 13/50
-## 469/469 - 2s - 4ms/step - loss: 0.0648 - val_loss: 0.0643
+## 469/469 - 2s - 3ms/step - loss: 0.0648 - val_loss: 0.0643
 ## Epoch 14/50
-## 469/469 - 2s - 4ms/step - loss: 0.0646 - val_loss: 0.0641
+## 469/469 - 1s - 3ms/step - loss: 0.0646 - val_loss: 0.0641
 ## Epoch 15/50
-## 469/469 - 2s - 4ms/step - loss: 0.0645 - val_loss: 0.0640
+## 469/469 - 1s - 3ms/step - loss: 0.0645 - val_loss: 0.0640
 ## Epoch 16/50
-## 469/469 - 2s - 4ms/step - loss: 0.0644 - val_loss: 0.0639
+## 469/469 - 1s - 3ms/step - loss: 0.0644 - val_loss: 0.0639
 ## Epoch 17/50
-## 469/469 - 2s - 4ms/step - loss: 0.0643 - val_loss: 0.0638
+## 469/469 - 2s - 3ms/step - loss: 0.0642 - val_loss: 0.0638
 ## Epoch 18/50
-## 469/469 - 2s - 4ms/step - loss: 0.0642 - val_loss: 0.0637
+## 469/469 - 2s - 3ms/step - loss: 0.0641 - val_loss: 0.0637
 ## Epoch 19/50
-## 469/469 - 2s - 4ms/step - loss: 0.0641 - val_loss: 0.0636
+## 469/469 - 1s - 3ms/step - loss: 0.0640 - val_loss: 0.0636
 ## Epoch 20/50
-## 469/469 - 2s - 4ms/step - loss: 0.0640 - val_loss: 0.0635
+## 469/469 - 2s - 3ms/step - loss: 0.0639 - val_loss: 0.0635
 ## Epoch 21/50
-## 469/469 - 2s - 4ms/step - loss: 0.0639 - val_loss: 0.0635
+## 469/469 - 2s - 3ms/step - loss: 0.0639 - val_loss: 0.0635
 ## Epoch 22/50
-## 469/469 - 2s - 4ms/step - loss: 0.0638 - val_loss: 0.0634
+## 469/469 - 1s - 3ms/step - loss: 0.0638 - val_loss: 0.0634
 ## Epoch 23/50
-## 469/469 - 2s - 4ms/step - loss: 0.0638 - val_loss: 0.0634
+## 469/469 - 2s - 3ms/step - loss: 0.0637 - val_loss: 0.0634
 ## Epoch 24/50
-## 469/469 - 2s - 4ms/step - loss: 0.0637 - val_loss: 0.0633
+## 469/469 - 2s - 3ms/step - loss: 0.0637 - val_loss: 0.0633
 ## Epoch 25/50
-## 469/469 - 2s - 4ms/step - loss: 0.0637 - val_loss: 0.0633
+## 469/469 - 2s - 3ms/step - loss: 0.0636 - val_loss: 0.0632
 ## Epoch 26/50
-## 469/469 - 2s - 4ms/step - loss: 0.0636 - val_loss: 0.0632
+## 469/469 - 2s - 3ms/step - loss: 0.0636 - val_loss: 0.0632
 ## Epoch 27/50
-## 469/469 - 2s - 4ms/step - loss: 0.0636 - val_loss: 0.0632
+## 469/469 - 2s - 3ms/step - loss: 0.0635 - val_loss: 0.0631
 ## Epoch 28/50
-## 469/469 - 2s - 4ms/step - loss: 0.0635 - val_loss: 0.0631
+## 469/469 - 2s - 3ms/step - loss: 0.0635 - val_loss: 0.0631
 ## Epoch 29/50
-## 469/469 - 2s - 4ms/step - loss: 0.0635 - val_loss: 0.0631
+## 469/469 - 2s - 3ms/step - loss: 0.0634 - val_loss: 0.0631
 ## Epoch 30/50
-## 469/469 - 2s - 4ms/step - loss: 0.0634 - val_loss: 0.0630
+## 469/469 - 1s - 3ms/step - loss: 0.0634 - val_loss: 0.0630
 ## Epoch 31/50
-## 469/469 - 2s - 4ms/step - loss: 0.0634 - val_loss: 0.0630
+## 469/469 - 2s - 3ms/step - loss: 0.0633 - val_loss: 0.0630
 ## Epoch 32/50
-## 469/469 - 2s - 4ms/step - loss: 0.0633 - val_loss: 0.0629
+## 469/469 - 2s - 3ms/step - loss: 0.0633 - val_loss: 0.0630
 ## Epoch 33/50
-## 469/469 - 2s - 4ms/step - loss: 0.0633 - val_loss: 0.0629
+## 469/469 - 2s - 3ms/step - loss: 0.0633 - val_loss: 0.0629
 ## Epoch 34/50
-## 469/469 - 2s - 4ms/step - loss: 0.0633 - val_loss: 0.0629
+## 469/469 - 2s - 3ms/step - loss: 0.0632 - val_loss: 0.0629
 ## Epoch 35/50
-## 469/469 - 2s - 4ms/step - loss: 0.0632 - val_loss: 0.0628
+## 469/469 - 1s - 3ms/step - loss: 0.0632 - val_loss: 0.0628
 ## Epoch 36/50
-## 469/469 - 2s - 4ms/step - loss: 0.0632 - val_loss: 0.0628
+## 469/469 - 2s - 3ms/step - loss: 0.0632 - val_loss: 0.0628
 ## Epoch 37/50
-## 469/469 - 2s - 4ms/step - loss: 0.0632 - val_loss: 0.0628
+## 469/469 - 1s - 3ms/step - loss: 0.0631 - val_loss: 0.0628
 ## Epoch 38/50
-## 469/469 - 2s - 4ms/step - loss: 0.0631 - val_loss: 0.0628
+## 469/469 - 2s - 3ms/step - loss: 0.0631 - val_loss: 0.0628
 ## Epoch 39/50
-## 469/469 - 2s - 4ms/step - loss: 0.0631 - val_loss: 0.0627
+## 469/469 - 1s - 3ms/step - loss: 0.0631 - val_loss: 0.0627
 ## Epoch 40/50
-## 469/469 - 2s - 4ms/step - loss: 0.0631 - val_loss: 0.0627
+## 469/469 - 2s - 3ms/step - loss: 0.0630 - val_loss: 0.0627
 ## Epoch 41/50
-## 469/469 - 2s - 4ms/step - loss: 0.0630 - val_loss: 0.0627
+## 469/469 - 2s - 3ms/step - loss: 0.0630 - val_loss: 0.0627
 ## Epoch 42/50
-## 469/469 - 2s - 4ms/step - loss: 0.0630 - val_loss: 0.0627
+## 469/469 - 2s - 3ms/step - loss: 0.0630 - val_loss: 0.0626
 ## Epoch 43/50
-## 469/469 - 2s - 4ms/step - loss: 0.0630 - val_loss: 0.0626
+## 469/469 - 2s - 3ms/step - loss: 0.0630 - val_loss: 0.0626
 ## Epoch 44/50
-## 469/469 - 2s - 4ms/step - loss: 0.0630 - val_loss: 0.0626
+## 469/469 - 2s - 3ms/step - loss: 0.0629 - val_loss: 0.0626
 ## Epoch 45/50
-## 469/469 - 2s - 4ms/step - loss: 0.0629 - val_loss: 0.0626
+## 469/469 - 2s - 3ms/step - loss: 0.0629 - val_loss: 0.0626
 ## Epoch 46/50
-## 469/469 - 2s - 4ms/step - loss: 0.0629 - val_loss: 0.0626
+## 469/469 - 2s - 3ms/step - loss: 0.0629 - val_loss: 0.0625
 ## Epoch 47/50
-## 469/469 - 2s - 4ms/step - loss: 0.0629 - val_loss: 0.0625
+## 469/469 - 2s - 3ms/step - loss: 0.0629 - val_loss: 0.0625
 ## Epoch 48/50
-## 469/469 - 2s - 4ms/step - loss: 0.0629 - val_loss: 0.0625
+## 469/469 - 1s - 3ms/step - loss: 0.0628 - val_loss: 0.0625
 ## Epoch 49/50
-## 469/469 - 2s - 4ms/step - loss: 0.0628 - val_loss: 0.0625
+## 469/469 - 2s - 3ms/step - loss: 0.0628 - val_loss: 0.0625
 ## Epoch 50/50
-## 469/469 - 2s - 4ms/step - loss: 0.0628 - val_loss: 0.0625
+## 469/469 - 2s - 3ms/step - loss: 0.0628 - val_loss: 0.0625
 ```
 
 Let's predict on our test dataset and display the original image together with
@@ -265,7 +265,7 @@ predictions <- autoencoder |> predict(test_data)
 ```
 
 ```
-## 313/313 - 1s - 3ms/step
+## 313/313 - 1s - 2ms/step
 ```
 
 ``` r
@@ -292,205 +292,205 @@ autoencoder |> fit(
 
 ```
 ## Epoch 1/100
-## 469/469 - 2s - 5ms/step - loss: 0.1009 - val_loss: 0.0939
+## 469/469 - 2s - 3ms/step - loss: 0.1006 - val_loss: 0.0939
 ## Epoch 2/100
-## 469/469 - 2s - 4ms/step - loss: 0.0936 - val_loss: 0.0919
+## 469/469 - 2s - 3ms/step - loss: 0.0936 - val_loss: 0.0919
 ## Epoch 3/100
-## 469/469 - 2s - 4ms/step - loss: 0.0921 - val_loss: 0.0908
+## 469/469 - 2s - 3ms/step - loss: 0.0921 - val_loss: 0.0908
 ## Epoch 4/100
-## 469/469 - 2s - 4ms/step - loss: 0.0911 - val_loss: 0.0900
+## 469/469 - 2s - 3ms/step - loss: 0.0912 - val_loss: 0.0902
 ## Epoch 5/100
-## 469/469 - 2s - 4ms/step - loss: 0.0903 - val_loss: 0.0895
+## 469/469 - 2s - 3ms/step - loss: 0.0905 - val_loss: 0.0896
 ## Epoch 6/100
-## 469/469 - 2s - 4ms/step - loss: 0.0897 - val_loss: 0.0889
+## 469/469 - 2s - 3ms/step - loss: 0.0900 - val_loss: 0.0891
 ## Epoch 7/100
-## 469/469 - 2s - 4ms/step - loss: 0.0893 - val_loss: 0.0885
+## 469/469 - 2s - 3ms/step - loss: 0.0895 - val_loss: 0.0887
 ## Epoch 8/100
-## 469/469 - 2s - 4ms/step - loss: 0.0889 - val_loss: 0.0881
+## 469/469 - 2s - 3ms/step - loss: 0.0891 - val_loss: 0.0883
 ## Epoch 9/100
-## 469/469 - 2s - 4ms/step - loss: 0.0886 - val_loss: 0.0879
+## 469/469 - 2s - 3ms/step - loss: 0.0887 - val_loss: 0.0880
 ## Epoch 10/100
-## 469/469 - 2s - 4ms/step - loss: 0.0883 - val_loss: 0.0876
+## 469/469 - 2s - 3ms/step - loss: 0.0884 - val_loss: 0.0878
 ## Epoch 11/100
-## 469/469 - 2s - 4ms/step - loss: 0.0880 - val_loss: 0.0874
+## 469/469 - 2s - 3ms/step - loss: 0.0882 - val_loss: 0.0875
 ## Epoch 12/100
-## 469/469 - 2s - 4ms/step - loss: 0.0878 - val_loss: 0.0872
+## 469/469 - 2s - 3ms/step - loss: 0.0879 - val_loss: 0.0873
 ## Epoch 13/100
-## 469/469 - 2s - 4ms/step - loss: 0.0876 - val_loss: 0.0870
+## 469/469 - 2s - 3ms/step - loss: 0.0877 - val_loss: 0.0871
 ## Epoch 14/100
-## 469/469 - 2s - 4ms/step - loss: 0.0875 - val_loss: 0.0868
+## 469/469 - 2s - 3ms/step - loss: 0.0875 - val_loss: 0.0869
 ## Epoch 15/100
-## 469/469 - 2s - 4ms/step - loss: 0.0873 - val_loss: 0.0867
+## 469/469 - 2s - 3ms/step - loss: 0.0873 - val_loss: 0.0867
 ## Epoch 16/100
-## 469/469 - 2s - 4ms/step - loss: 0.0872 - val_loss: 0.0866
+## 469/469 - 2s - 3ms/step - loss: 0.0872 - val_loss: 0.0866
 ## Epoch 17/100
-## 469/469 - 2s - 4ms/step - loss: 0.0870 - val_loss: 0.0865
+## 469/469 - 2s - 3ms/step - loss: 0.0870 - val_loss: 0.0864
 ## Epoch 18/100
-## 469/469 - 2s - 4ms/step - loss: 0.0869 - val_loss: 0.0864
+## 469/469 - 2s - 3ms/step - loss: 0.0869 - val_loss: 0.0863
 ## Epoch 19/100
-## 469/469 - 2s - 4ms/step - loss: 0.0868 - val_loss: 0.0863
+## 469/469 - 2s - 3ms/step - loss: 0.0868 - val_loss: 0.0862
 ## Epoch 20/100
-## 469/469 - 2s - 4ms/step - loss: 0.0867 - val_loss: 0.0862
+## 469/469 - 2s - 3ms/step - loss: 0.0866 - val_loss: 0.0861
 ## Epoch 21/100
-## 469/469 - 2s - 4ms/step - loss: 0.0866 - val_loss: 0.0861
+## 469/469 - 2s - 3ms/step - loss: 0.0865 - val_loss: 0.0860
 ## Epoch 22/100
-## 469/469 - 2s - 4ms/step - loss: 0.0865 - val_loss: 0.0860
+## 469/469 - 2s - 3ms/step - loss: 0.0864 - val_loss: 0.0859
 ## Epoch 23/100
-## 469/469 - 2s - 4ms/step - loss: 0.0865 - val_loss: 0.0859
+## 469/469 - 2s - 3ms/step - loss: 0.0863 - val_loss: 0.0858
 ## Epoch 24/100
-## 469/469 - 2s - 4ms/step - loss: 0.0864 - val_loss: 0.0859
+## 469/469 - 1s - 3ms/step - loss: 0.0862 - val_loss: 0.0857
 ## Epoch 25/100
-## 469/469 - 2s - 4ms/step - loss: 0.0863 - val_loss: 0.0858
+## 469/469 - 2s - 3ms/step - loss: 0.0862 - val_loss: 0.0856
 ## Epoch 26/100
-## 469/469 - 2s - 4ms/step - loss: 0.0862 - val_loss: 0.0857
+## 469/469 - 2s - 3ms/step - loss: 0.0861 - val_loss: 0.0856
 ## Epoch 27/100
-## 469/469 - 2s - 4ms/step - loss: 0.0862 - val_loss: 0.0857
+## 469/469 - 2s - 3ms/step - loss: 0.0860 - val_loss: 0.0855
 ## Epoch 28/100
-## 469/469 - 2s - 4ms/step - loss: 0.0861 - val_loss: 0.0856
+## 469/469 - 2s - 3ms/step - loss: 0.0859 - val_loss: 0.0855
 ## Epoch 29/100
-## 469/469 - 2s - 4ms/step - loss: 0.0860 - val_loss: 0.0856
+## 469/469 - 2s - 3ms/step - loss: 0.0859 - val_loss: 0.0854
 ## Epoch 30/100
-## 469/469 - 2s - 4ms/step - loss: 0.0860 - val_loss: 0.0855
+## 469/469 - 2s - 3ms/step - loss: 0.0858 - val_loss: 0.0853
 ## Epoch 31/100
-## 469/469 - 2s - 4ms/step - loss: 0.0859 - val_loss: 0.0854
+## 469/469 - 2s - 3ms/step - loss: 0.0858 - val_loss: 0.0853
 ## Epoch 32/100
-## 469/469 - 2s - 4ms/step - loss: 0.0859 - val_loss: 0.0854
+## 469/469 - 2s - 3ms/step - loss: 0.0857 - val_loss: 0.0853
 ## Epoch 33/100
-## 469/469 - 2s - 4ms/step - loss: 0.0858 - val_loss: 0.0854
+## 469/469 - 2s - 3ms/step - loss: 0.0857 - val_loss: 0.0852
 ## Epoch 34/100
-## 469/469 - 2s - 4ms/step - loss: 0.0858 - val_loss: 0.0853
+## 469/469 - 2s - 3ms/step - loss: 0.0856 - val_loss: 0.0852
 ## Epoch 35/100
-## 469/469 - 2s - 4ms/step - loss: 0.0857 - val_loss: 0.0853
+## 469/469 - 2s - 3ms/step - loss: 0.0856 - val_loss: 0.0851
 ## Epoch 36/100
-## 469/469 - 2s - 4ms/step - loss: 0.0857 - val_loss: 0.0852
+## 469/469 - 2s - 3ms/step - loss: 0.0855 - val_loss: 0.0851
 ## Epoch 37/100
-## 469/469 - 2s - 4ms/step - loss: 0.0856 - val_loss: 0.0852
+## 469/469 - 2s - 3ms/step - loss: 0.0855 - val_loss: 0.0851
 ## Epoch 38/100
-## 469/469 - 2s - 4ms/step - loss: 0.0856 - val_loss: 0.0851
+## 469/469 - 2s - 3ms/step - loss: 0.0855 - val_loss: 0.0850
 ## Epoch 39/100
-## 469/469 - 2s - 4ms/step - loss: 0.0856 - val_loss: 0.0851
+## 469/469 - 2s - 3ms/step - loss: 0.0854 - val_loss: 0.0850
 ## Epoch 40/100
-## 469/469 - 2s - 4ms/step - loss: 0.0855 - val_loss: 0.0851
+## 469/469 - 2s - 3ms/step - loss: 0.0854 - val_loss: 0.0850
 ## Epoch 41/100
-## 469/469 - 2s - 4ms/step - loss: 0.0855 - val_loss: 0.0850
+## 469/469 - 2s - 3ms/step - loss: 0.0854 - val_loss: 0.0850
 ## Epoch 42/100
-## 469/469 - 2s - 4ms/step - loss: 0.0854 - val_loss: 0.0850
+## 469/469 - 2s - 3ms/step - loss: 0.0853 - val_loss: 0.0849
 ## Epoch 43/100
-## 469/469 - 2s - 4ms/step - loss: 0.0854 - val_loss: 0.0850
+## 469/469 - 2s - 3ms/step - loss: 0.0853 - val_loss: 0.0849
 ## Epoch 44/100
-## 469/469 - 2s - 4ms/step - loss: 0.0854 - val_loss: 0.0850
+## 469/469 - 2s - 3ms/step - loss: 0.0853 - val_loss: 0.0849
 ## Epoch 45/100
-## 469/469 - 2s - 4ms/step - loss: 0.0853 - val_loss: 0.0849
+## 469/469 - 2s - 3ms/step - loss: 0.0853 - val_loss: 0.0849
 ## Epoch 46/100
-## 469/469 - 2s - 4ms/step - loss: 0.0853 - val_loss: 0.0849
+## 469/469 - 2s - 3ms/step - loss: 0.0852 - val_loss: 0.0848
 ## Epoch 47/100
-## 469/469 - 2s - 4ms/step - loss: 0.0853 - val_loss: 0.0849
+## 469/469 - 2s - 3ms/step - loss: 0.0852 - val_loss: 0.0848
 ## Epoch 48/100
-## 469/469 - 2s - 4ms/step - loss: 0.0853 - val_loss: 0.0849
+## 469/469 - 2s - 3ms/step - loss: 0.0852 - val_loss: 0.0848
 ## Epoch 49/100
-## 469/469 - 2s - 4ms/step - loss: 0.0852 - val_loss: 0.0849
+## 469/469 - 2s - 3ms/step - loss: 0.0852 - val_loss: 0.0848
 ## Epoch 50/100
-## 469/469 - 2s - 4ms/step - loss: 0.0852 - val_loss: 0.0848
+## 469/469 - 2s - 3ms/step - loss: 0.0851 - val_loss: 0.0848
 ## Epoch 51/100
-## 469/469 - 2s - 4ms/step - loss: 0.0852 - val_loss: 0.0848
+## 469/469 - 2s - 3ms/step - loss: 0.0851 - val_loss: 0.0847
 ## Epoch 52/100
-## 469/469 - 2s - 4ms/step - loss: 0.0852 - val_loss: 0.0848
+## 469/469 - 2s - 3ms/step - loss: 0.0851 - val_loss: 0.0847
 ## Epoch 53/100
-## 469/469 - 2s - 4ms/step - loss: 0.0851 - val_loss: 0.0848
+## 469/469 - 2s - 3ms/step - loss: 0.0851 - val_loss: 0.0847
 ## Epoch 54/100
-## 469/469 - 2s - 4ms/step - loss: 0.0851 - val_loss: 0.0848
+## 469/469 - 2s - 3ms/step - loss: 0.0850 - val_loss: 0.0847
 ## Epoch 55/100
-## 469/469 - 2s - 4ms/step - loss: 0.0851 - val_loss: 0.0847
+## 469/469 - 2s - 3ms/step - loss: 0.0850 - val_loss: 0.0847
 ## Epoch 56/100
-## 469/469 - 2s - 4ms/step - loss: 0.0851 - val_loss: 0.0847
+## 469/469 - 2s - 3ms/step - loss: 0.0850 - val_loss: 0.0846
 ## Epoch 57/100
-## 469/469 - 2s - 4ms/step - loss: 0.0850 - val_loss: 0.0847
+## 469/469 - 2s - 3ms/step - loss: 0.0850 - val_loss: 0.0846
 ## Epoch 58/100
-## 469/469 - 2s - 4ms/step - loss: 0.0850 - val_loss: 0.0847
+## 469/469 - 2s - 3ms/step - loss: 0.0850 - val_loss: 0.0846
 ## Epoch 59/100
-## 469/469 - 2s - 4ms/step - loss: 0.0850 - val_loss: 0.0847
+## 469/469 - 2s - 3ms/step - loss: 0.0849 - val_loss: 0.0846
 ## Epoch 60/100
-## 469/469 - 2s - 4ms/step - loss: 0.0850 - val_loss: 0.0847
+## 469/469 - 2s - 3ms/step - loss: 0.0849 - val_loss: 0.0846
 ## Epoch 61/100
-## 469/469 - 2s - 4ms/step - loss: 0.0850 - val_loss: 0.0847
+## 469/469 - 2s - 3ms/step - loss: 0.0849 - val_loss: 0.0846
 ## Epoch 62/100
-## 469/469 - 2s - 4ms/step - loss: 0.0850 - val_loss: 0.0846
+## 469/469 - 2s - 3ms/step - loss: 0.0849 - val_loss: 0.0846
 ## Epoch 63/100
-## 469/469 - 2s - 4ms/step - loss: 0.0849 - val_loss: 0.0846
+## 469/469 - 1s - 3ms/step - loss: 0.0849 - val_loss: 0.0845
 ## Epoch 64/100
-## 469/469 - 2s - 4ms/step - loss: 0.0849 - val_loss: 0.0846
+## 469/469 - 2s - 3ms/step - loss: 0.0849 - val_loss: 0.0845
 ## Epoch 65/100
-## 469/469 - 2s - 4ms/step - loss: 0.0849 - val_loss: 0.0846
+## 469/469 - 2s - 3ms/step - loss: 0.0848 - val_loss: 0.0845
 ## Epoch 66/100
-## 469/469 - 2s - 4ms/step - loss: 0.0849 - val_loss: 0.0846
+## 469/469 - 2s - 3ms/step - loss: 0.0848 - val_loss: 0.0845
 ## Epoch 67/100
-## 469/469 - 2s - 4ms/step - loss: 0.0849 - val_loss: 0.0846
+## 469/469 - 2s - 3ms/step - loss: 0.0848 - val_loss: 0.0845
 ## Epoch 68/100
-## 469/469 - 2s - 4ms/step - loss: 0.0849 - val_loss: 0.0845
+## 469/469 - 2s - 3ms/step - loss: 0.0848 - val_loss: 0.0845
 ## Epoch 69/100
-## 469/469 - 2s - 4ms/step - loss: 0.0848 - val_loss: 0.0845
+## 469/469 - 2s - 3ms/step - loss: 0.0848 - val_loss: 0.0845
 ## Epoch 70/100
-## 469/469 - 2s - 4ms/step - loss: 0.0848 - val_loss: 0.0845
+## 469/469 - 2s - 3ms/step - loss: 0.0848 - val_loss: 0.0845
 ## Epoch 71/100
-## 469/469 - 2s - 4ms/step - loss: 0.0848 - val_loss: 0.0845
+## 469/469 - 2s - 3ms/step - loss: 0.0848 - val_loss: 0.0844
 ## Epoch 72/100
-## 469/469 - 2s - 4ms/step - loss: 0.0848 - val_loss: 0.0845
+## 469/469 - 2s - 3ms/step - loss: 0.0847 - val_loss: 0.0844
 ## Epoch 73/100
-## 469/469 - 2s - 4ms/step - loss: 0.0848 - val_loss: 0.0845
+## 469/469 - 2s - 3ms/step - loss: 0.0847 - val_loss: 0.0844
 ## Epoch 74/100
-## 469/469 - 2s - 4ms/step - loss: 0.0848 - val_loss: 0.0845
+## 469/469 - 2s - 3ms/step - loss: 0.0847 - val_loss: 0.0844
 ## Epoch 75/100
-## 469/469 - 2s - 4ms/step - loss: 0.0848 - val_loss: 0.0845
+## 469/469 - 1s - 3ms/step - loss: 0.0847 - val_loss: 0.0844
 ## Epoch 76/100
-## 469/469 - 2s - 4ms/step - loss: 0.0848 - val_loss: 0.0845
+## 469/469 - 2s - 3ms/step - loss: 0.0847 - val_loss: 0.0844
 ## Epoch 77/100
-## 469/469 - 2s - 4ms/step - loss: 0.0847 - val_loss: 0.0845
+## 469/469 - 2s - 3ms/step - loss: 0.0847 - val_loss: 0.0844
 ## Epoch 78/100
-## 469/469 - 2s - 4ms/step - loss: 0.0847 - val_loss: 0.0844
+## 469/469 - 2s - 3ms/step - loss: 0.0847 - val_loss: 0.0844
 ## Epoch 79/100
-## 469/469 - 2s - 4ms/step - loss: 0.0847 - val_loss: 0.0844
+## 469/469 - 2s - 3ms/step - loss: 0.0847 - val_loss: 0.0844
 ## Epoch 80/100
-## 469/469 - 2s - 4ms/step - loss: 0.0847 - val_loss: 0.0844
+## 469/469 - 2s - 3ms/step - loss: 0.0847 - val_loss: 0.0844
 ## Epoch 81/100
-## 469/469 - 2s - 4ms/step - loss: 0.0847 - val_loss: 0.0844
+## 469/469 - 2s - 3ms/step - loss: 0.0846 - val_loss: 0.0844
 ## Epoch 82/100
-## 469/469 - 2s - 4ms/step - loss: 0.0847 - val_loss: 0.0844
+## 469/469 - 2s - 3ms/step - loss: 0.0846 - val_loss: 0.0844
 ## Epoch 83/100
-## 469/469 - 2s - 4ms/step - loss: 0.0847 - val_loss: 0.0844
+## 469/469 - 2s - 3ms/step - loss: 0.0846 - val_loss: 0.0844
 ## Epoch 84/100
-## 469/469 - 2s - 4ms/step - loss: 0.0847 - val_loss: 0.0844
+## 469/469 - 2s - 3ms/step - loss: 0.0846 - val_loss: 0.0843
 ## Epoch 85/100
-## 469/469 - 2s - 4ms/step - loss: 0.0847 - val_loss: 0.0844
+## 469/469 - 2s - 3ms/step - loss: 0.0846 - val_loss: 0.0843
 ## Epoch 86/100
-## 469/469 - 2s - 4ms/step - loss: 0.0846 - val_loss: 0.0844
+## 469/469 - 2s - 3ms/step - loss: 0.0846 - val_loss: 0.0843
 ## Epoch 87/100
-## 469/469 - 2s - 4ms/step - loss: 0.0846 - val_loss: 0.0844
+## 469/469 - 1s - 3ms/step - loss: 0.0846 - val_loss: 0.0843
 ## Epoch 88/100
-## 469/469 - 2s - 4ms/step - loss: 0.0846 - val_loss: 0.0844
+## 469/469 - 2s - 3ms/step - loss: 0.0846 - val_loss: 0.0843
 ## Epoch 89/100
-## 469/469 - 2s - 4ms/step - loss: 0.0846 - val_loss: 0.0843
+## 469/469 - 2s - 3ms/step - loss: 0.0846 - val_loss: 0.0843
 ## Epoch 90/100
-## 469/469 - 2s - 4ms/step - loss: 0.0846 - val_loss: 0.0843
+## 469/469 - 2s - 3ms/step - loss: 0.0846 - val_loss: 0.0843
 ## Epoch 91/100
-## 469/469 - 2s - 4ms/step - loss: 0.0846 - val_loss: 0.0843
+## 469/469 - 2s - 3ms/step - loss: 0.0845 - val_loss: 0.0843
 ## Epoch 92/100
-## 469/469 - 2s - 4ms/step - loss: 0.0846 - val_loss: 0.0843
+## 469/469 - 2s - 3ms/step - loss: 0.0845 - val_loss: 0.0843
 ## Epoch 93/100
-## 469/469 - 2s - 4ms/step - loss: 0.0846 - val_loss: 0.0843
+## 469/469 - 2s - 3ms/step - loss: 0.0845 - val_loss: 0.0843
 ## Epoch 94/100
-## 469/469 - 2s - 4ms/step - loss: 0.0846 - val_loss: 0.0843
+## 469/469 - 2s - 3ms/step - loss: 0.0845 - val_loss: 0.0843
 ## Epoch 95/100
-## 469/469 - 2s - 4ms/step - loss: 0.0846 - val_loss: 0.0843
+## 469/469 - 2s - 3ms/step - loss: 0.0845 - val_loss: 0.0843
 ## Epoch 96/100
-## 469/469 - 2s - 4ms/step - loss: 0.0846 - val_loss: 0.0843
+## 469/469 - 1s - 3ms/step - loss: 0.0845 - val_loss: 0.0842
 ## Epoch 97/100
-## 469/469 - 2s - 4ms/step - loss: 0.0845 - val_loss: 0.0843
+## 469/469 - 2s - 3ms/step - loss: 0.0845 - val_loss: 0.0842
 ## Epoch 98/100
-## 469/469 - 2s - 4ms/step - loss: 0.0845 - val_loss: 0.0843
+## 469/469 - 2s - 3ms/step - loss: 0.0845 - val_loss: 0.0842
 ## Epoch 99/100
-## 469/469 - 2s - 4ms/step - loss: 0.0845 - val_loss: 0.0843
+## 469/469 - 2s - 3ms/step - loss: 0.0845 - val_loss: 0.0842
 ## Epoch 100/100
-## 469/469 - 2s - 4ms/step - loss: 0.0845 - val_loss: 0.0843
+## 469/469 - 1s - 3ms/step - loss: 0.0845 - val_loss: 0.0842
 ```
 
 Let's now predict on the noisy data and display the results of our autoencoder.
@@ -504,7 +504,7 @@ predictions <- autoencoder |> predict(noisy_test_data)
 ```
 
 ```
-## 313/313 - 0s - 1ms/step
+## 313/313 - 0s - 775us/step
 ```
 
 ``` r
diff --git a/vignettes/examples/vision/autoencoder/unnamed-chunk-5-1.png b/vignettes/examples/vision/autoencoder/unnamed-chunk-5-1.png
index 1aa75cd9346d10667b7685ff13d5757b23f96959..095cb80e36063dd419dca38332e7c74ee8cf1a93 100644
GIT binary patch
delta 53605
zcma%?Wl$Ymu&xu_H9&9(?(QxL65O5O?!je(yA#|=aCdityX(fCy>XYrcdO3td#dK=
zRIOEOW_7>a{k+@7(8Gn$)w0lluczR`D(>AGXDsDaT%2e2D28$&!dJS#Nl`KpY9W%e
zU+8ht7$nT|$I=cw#{?%`{v37YxI<pNv!(8`!JZJl;tAi1x5*>zh1au=5Xht7+eNII
zZsY&?0sil={{P>61MCJOKz_O{&hw=WCp`AlE3MI)2?-Gq5$WmaAt4a~nMpw&qlKs^
zixv7E)_V>Nez$<4{?ERaPJ#mgd{`<f=bVUitX743PYOrh@wZY4gJy+6o!j~&6XovQ
zwOqNL&odNMfOv23m;8WsJzD|yDV2EtLkGPsw|NJVrzYmcihEQSU+4*U1p{GWzr)U4
z3{oBejy^Z*OS|*UD&4EDgC~*q-Fcn0q7w2rkQW&J*a0wcQ<RkibIYBT)a75QBif^a
z5J+h_gqWz77nleaDxa6TTx-sn)NQ((I<1h|_?RdNW8+fm+XDjfu^bKea`R8K^`X|v
zqw3YjZZj6D9!Rfz2O@A7F7AlV6zsQzE=zS<Ev_f<KNDo68nE*I{aaY=-t+<f3QY$o
zf$2o^<rK(iD-Q87Y%-d~bsU&wg9h!PkxcVdc<+~Q@cBYhtnY-P56L39)Ui$}$74F^
z+=(1KMX1{>IIKbCxKU@X*Om_6jFsD+of6ygf!sPy9LMF}`pIGGj11hU`6DVa7M@Q*
zZK>}Iv>9=*A>vb^a#Sq|&zo4tR~ghb)^r#_fXzcw-{HeKOfzjZVaH>up)j8E*X1sE
zHq^&X;~RbqcNHXSeza-d<083AexY-}(-S<5=hLsvIVrO@$QCgqaVrCpJ(AxW&CVuY
zdm=4)K|Gjt6h(z5n;RX5`h)io1x1U{5Gh5UoJV-!b`$quQC+(?dolZDVzpK7Hx^<q
z;PBC*tHPiwJT)o6T!5mhzH|b1wM~)2WE(3{b%q_ZPc0>QQ{CmG@LlFxR8E$Eg2(#g
z>p-MMD#S)fqbV!?tMs>*IatC)8Y9g3>6(jhnJU$MD2RY|+F|i%1hNyyY&`Z$Eb$|V
z^?=S&RJQH=#l2^nx7bAs?10(t22_X+AW&MnBLNoBIkX~*TOJsdYqNX#-vxNE5>@Oh
zO!pF$gq=1PD9ET81gdh2|FqY(a*Obt4xgP+(<I%=AULzD*u9Wm5<{M;joFXSa-$OY
z%Yku<dvzllDfhRjrQNZ$l)S>;-2-%@F-0p{8AO+<M7#I@%Hsm%$`@ES$3oJ9(_r?~
zmJd7u`(Ly~Lv&W#ydL4D{;iJ8Q22uNxL_d9_QJR;c7%SnEC$6DWY>(M4@eQbU)Ya{
z8TFv@6KV+_HJWn1nW_D}K|_g^?|eHyI<d6Te*A8@coIT`J{*3<s%-nXEaI0JqJLiO
zAvKW*1!{qH|2|C2Rt>bCU9mURHx!f=IXHU!G%hWproq&6PnonX847b|i+fu~7s}HQ
zJW*4gX0N>p2QOFo+&53Fr%2@uKe%>4OLSkPwI~`6t}uZ9!Flqf<_HurGIfhLJXpLv
z!mjw_pe2DRyB~;c+rK2FLMAp?uv|(7+Pt5{)&^;cya|z1zW~{-t7%z6`SqbW4-m+c
z>{6ZR00SQmlMn?5A&A)umaAem1kX;NR77!z-kawGF(4vv_i*B2ArCH_qN!YzS0r6w
z=~bxPdtWbj8R@f|$FUK<giKx5$4`)iC$Wo42XmJ}0ZQhAKbS$jmZrAG&R*u5`q~=L
zS5offZ#+JsKnML&9a+dQBQ%EqPuDq#XM>4_9HXs1_$ide9Cy(*D#{;P<Z5f@pbC54
zHJG)-wVrcI1#2gt;&Hjl&!`c~ec6s|C8SoHxpSQzp7SRPvw{2dxhF}6hSEAeuOK_W
zFdLT;hrpah)*ChbVcw`Zg`F89_zZ9Xy<MyxHxP0I*#0&#$^L19nwCv}qAr#76B%wR
z(V_cxRGno{be{gv6m3lUq&{6goR~VOTA7O%SGBDNtCp&?3xol;Rfci?=zcc@7`O5n
zGPFmc)GTE5<VBUf;O!e-5@GUqnY7^faUrtfMAg};m+8uo+9ny6K$4AGUT>|C9HW!T
z*9T;v{`Y${`SRsslOP2{Q+K6MZA9tdWZLSQAI3d9pHNeCX-$Ru=OEwPzO_L`!%l@8
z9?kvEa5T?dQ%?nnZ@(@er-E`P-FEQKgykHQ42+$cTh%e~!5ynRajAc<+h<nxE6G3I
zETeb7gm)&V_l{}Ci{07E$BZIFC?A&57^gIV%3vX6dVV+!67rf3BzUd{R3`h@rw}Uh
z-<pd&O?N0!3KRL&z~l4Y$@P2aP~I;~@dtBL5AMD$OjO;w(Tk0LNNn|mN6TTaY2Di+
zjZaclGgzJ5;NTR7&dPbZznDEJ$(tw~2^jlBONxiAZ^ouElXKPsWjD{S)EL;z;erf=
zkJ1zKyRDo@6j+F=SsVE(iqM;E0QU$5uMbyv825P-|1R6>gB%^CT!{bbGp_pGR=u5s
zK%vhN2JtS%kPfHnx%goaC2Gtqt!=FHaB_1&>kaaWaI7qDC~05cj=Lxl_DOgz4f{31
zBiE~+7)=_(bA<j%J@xOSxt`GF-#`Z<#_1i1U!U>$o}+zzMP_DIR1*>jAP{TU+g#pN
z(A`|seTvg3=+NB$*FY&nLPJ--&scAb`8br9st^k!;qTSxiG3Y*(er;FYC4;kiGFY8
z-!C&iYkoVNDL3daA5U+vF<S4my2AakT|ri2vbBx(5t4DNO#FhJ%-z%u^#ufcnsbYp
zd_plsoA|){m<UabdL9|;@pJTf&cACMFSfyo_09k4g!==?|D@Q@%-HBa9QKq=zu^0#
zJ_+2{BAf+RqX$Mg8w79{>5*Mo6|JOUdk<(YlIEusx9wUpB8(5RxSI0p3<qn?6;im&
zD6x`=m&#=S^D$~s8#<H$wOI<lbEDvc?SXL<OM0*`9CgE~iD%yKg@~1THMsZnmBc|%
z_%QV?!?AOAn&Z@TXBX!aPD;XWO(AC$NPBQYV41tp?n!2Zdj~7Hjy4C@2wsZBLu4AX
zp_HBTN`D!CwKqG`ru~ZBM0`4v`;3$6l3@48bl33`;dvkP*N2Qif<Z-qFx8;b+<kQZ
zD&A09O+~{=f3<tA$^Y%z3EX!&_LLX%ICHu<J(=CCb!0Pw^vYEGh;uks^{-WJlY<7j
z6S6csGTg_~?j0NTo)z{lR1YQ$TiyB|z|bIz=_*k2Tm$-#V3aY7>AcX1*q>bLyMG5l
z;ny0t;li0_LEBRW%&35ziS#fXZAr1!(I!d=I-j4C#8XuKG7jP3D)^4{2{!cGzampS
zO&<5y_8fG<pTt4FUZq`sTgjNf1GC<@c#35*S1(M*9=&$(y)im7tN7Y{p22Q1;NJqd
zZj!p1pxtepj0|QldNz!pcUoM!s><w|P!z)t<xP=14iTX#9aSKegmnMFbyiRF#bNtD
zb%eUDfa1VQ!2d`bB{#P8>6^0p&OLrXq=uV<JE<_L4=yhL-Ld@RYVT$1er3<avI7+Q
zj<66H+s3-q?X9|Gx32|5!47=?8_7DuQnmygDyHOuEvN@Zg+Lh*-}ASG4!_qwyb4=l
z$eQ=z4k;Ec4v>?SuCJxM2$l8rYa0y;9qlD_`EE85TPXUkcDRuugiq^`RQ=P=v-eLX
z%}Q<bZH14d+|=e8+-S@sMu_-dZwmMix<~g9Z@YOYzX%`6>zFI}DQlapx7}v<)Rdru
z;9mPs9DQy76g09jQSZi=tq*VzIJfayzM;KX3%Y%BEdktEYn;y!dgL^&U<5B4o7TWj
zFd-dLGSUj;V|27VnWR~dtfAIgN<-ZeZ+_O!KJ^9EZ{yzqMXB!+cIx=#8pmA&aprp~
z(UN4~Rebl_S{LW5OXP^tc-3hbnP(gUi&&|TMm1G+5waunWhVq14fXCfXfaWWlYW`#
z&oqZm{lL=s4;wQD=Vnkx*YB$(R;YHDV%?sJar8k40-A|*y;iW};FXM$maVh7n}Lyx
zr>nk+i>?Kvd+2_SaiR4-0P4*M?p~OEP9+<M^tpR@<UFV~a|vWd*<roDU$Kj!_p#Fz
z<3@iw!MSva?B6WSy_vX<fwr%=lqNcN1VK)wAArujG6{n7Gbi}hQB}gpA>9Ae*iceg
zm`@2(80iYUYQ7~xq@yB+M)q2JSEfQ{rhe-^XFanA%2hL0qbC{Z<ZJm3KR(Yx97%#5
zjJhR6g!tY!1JX*u+S>72m(<laP>Hh>I08Y1)mXnI+gYOZFpyEwQ`1q^8Q!FB1lz;_
z;8=+-hGw7CxLi>uH?tcYR=BSG-_HH4eN3yniaJ@TW*{5mqh5n(BjCr3UzlXZ1U`^I
zyPRZTpI&2?C-sJjRIQC>Ul8Q$*h$j>7n9ItdEWz}*Zf<B6|7HA8UZQsF{EN#9^s;%
zN8oR*p?Fwg{NrMXL4}ro0~5FX-`rJTxQFHo9TP)QalYIAP)*$d)0w@u|0NY2MOS-X
z2SC(e&6@FWun1ZnOe?9`dIh67t_}nXHoFd-rylz`ebi_*e3&bsNBeUHb5BVmBzp$^
zz}S4uyvpk-h=daBebrc8TH=BcOdOr(=P-3ge&}s={5@w?15WBz2I;EWwulUP{Ow;j
z<qlIxp10BiXT(K+OrWz;G;v6np|ir@B7%ev?U|KksDYetAKt>J$9<L6J=sA@3WiwQ
zl#VK*-L;tbzV7zCBy>HN4C;ZZov;yMi75!1x|wCPVdI8!1Whchtz1UWwq`9@uvIs3
zHRhUbQ53ODn?S3fM;CdW6AM7$N@mcHDnfI9cq&Pc<<C~qJZ~eky0*_S$C~tTXdj)l
z&7u=q52ED8nW~mI+I=DMZ&yP>Nlmb3T=K-e6sj)GKli=;-&nc$r{-6*RI@R$53L;h
z_Rp)(*s^olF#|a!k*q&!Q>oSE@N^mb&JOfFSxHOUI)AfwGV}eu*u4Zifx28^GO~z{
zAhsn{0i@5w=RKo{7rAB$(OvA7uTAP0a*eKpd_rxxLG3FSb$HDBHJwf;xB4uQqMkfu
zL<jaOrH``$oUbILcMv#|48^29nowc48ulrbC93XWImNG+W%e`ACm%wxNMmy9Z=%K-
zD?!bBUX!alhhI&7oV@@AWo=i9@0y@<7;dkZ>jw}tI_Y!TMGrXnam4t(2gE7h>WGfV
zKtVj;$kcUBjS5#wR`ACRCoV4Zb0A?8IQMleGNq<!!*usklEwm8gWD}ywEXxsO%(|F
z`Sf1&pe=poIqL8Oy1;BAeL`2`l_hnZ*ZQ|ze%`AD3{4X^R9*mp<T-3!rLcup&DLE-
zGFTJpuCzUc4i?pfKXt*AP<iEmOlFsC<7PmerGRM{>nxg%?2Mt0J6aNiwN4CKEc9no
z1yPR`x(^BF6Fn6NO;>YyX=6=G)~~F8Gs}?odtF~u(b-PM!8f-$e0gf5Ee6ML?VP|w
z2#nil_v#>M04-NCL!|s>ww_num%&%^t?6(F_Bx}C(#DUU&V*hh3b1F3t>3O89F~mW
zoTAXSjIzi&)tx}0GnQkbq}{?iN(%DFfhIG}iq-nxD5mx0;@;|tpLN$;(ZS8lyVs|^
zwdr2XbW`DEJG_*87J7(xS>0WvA1ClSe6a~P?Y&={fo*$k=a~x04DNFM7Y1;ciIZWt
zoAKdiZC>{?%|$+7{K4-B?PW(UijC@)PtGr)NRXz}eKYXL$rqL(DbnIQ|H0>cY_$=~
zZsR@@GPpD9ZtwJ!$#+!uE^udaR6}`jZCQks^IeCBd_=Uk@I;Ncbcc%l=%BLGN5ZF_
zsh+YAtam!~JvP{Ayrq|8G%pLRh5KhA`;`RJL{+W(T??{NF`!erj1&7pdmNsUQeHK`
zYf;W=kd7)$o?CSYJgMM%)v3USh65#aD6$w7`*ACtgBR>T2NPqJ9T>*y^Z?Rr<9M6Z
zt%zDbk845ob$jr<i}MvK@@eEo(vcw~(NGO|Eq=vG8O$4o-Q$wC6AIjcTxg#z0>-z|
z1#bi%pYIfRgdp_JpngGbDC_%z%|II67A!ql2EUSOYz+95>Wa_THOOJIBeX=*ZKlAa
z2>ajScHG_bzC@7Rh9g-dI;r{AsgswrwKI}0(z&iEQ`RJg-l|Gg4(?s=9a{6*Eh51F
z8c<*v06`FvkyU8P3Mbw+-%DzUF3%(pj+~^gva>!R`TU*&fr%~*4C*P*!<TFCH73#O
z@Xr0x=lXjkk*1SxI+tJG<@ITo#@a=og`QzHHrkpQjyf8#kd^7t<`&*7eH#?y*=N(K
z?zUt&yk>eU&;PYETNt5MQpo0<2l!R<#=pF%7k|Znj2Dsw><Ek<y45-%p03m>%E+tg
z-u1es!1qj-z&6NN<o+bg$<BBybvjS5+I=26pCVgH4zO%%;rn??zJk(rYzXJMl)b2d
zVMhK03mvN)vLYvoS~Iiljf2s;Tb_4BMYyIR_tMtLtkl(>^F)1axyvui4+t)SR@bK&
zE-OERLvL@<qjTYI;dv|4PVMLe>nvjl&5P#$`0=4Yi4PSlpA<6vQM(r4yAK-pdvSHk
zzylFR3-?RZ{ip{aB+ANs2x}V_Q#ugt#ugFgh7#Xcot>Xvh=?FcJ)GSB-ahPq+nR8?
z$TkUC&B#()8ITRxe*F%c4OF9i?dzw4ky0wUL7B|ZHK=oOya>u5*Zw+RZT~fQ>!G_~
zc=hmH)X-h(q%CBn)R{NO*81oJEc5fJ*#6zoiKM%Plh<jTlzU}Hrc(TF-dh9In*)NU
zTc^H?Gki}G+O#Q8<{HUk=)m9o<azg>cSa3_Zhd_WB&ZouA#jhPQ-E$p27z_i3FyDj
z6(hH~!H;oI7dM}>SNJbf#F<m%Z#yB(t7hcWGy5lsJLa^&!3l$Ad+*t!n6{Lr*tf`H
zO{<o_T;uD_`7asWKqjmcDU6YWRe1$H!-#j}iCw+hZBLyD*VWeS;zea^FHihOgBaXL
zortcL5hCkFQ3c^Jpj=$kicg(MAKL|{;=+%B2=k7*thzQ2Rn5T!d)Dy)FHkX{eqSiF
z4fR)KWkX9%Wov%@X9?r(K}Q1g)(jGGI;cPi(}F_xViY#J8fAd_OPLo|^4W&4pV(yI
z7ZW9i#?rh7LSdtG*r#jxO$%zVwYHUwdG+78`=)=|Bjjbrff1I9QU~&Sp3MYX1C>;W
zzHZ&SLd@^NJ%?jhMQj(cg-SposFphNxqD6gb-<$hQP<$EvH(4Bqz_dT%Hz{Sx)YDS
znYNRkf`$$^k$(Lhiu{i!XHpTs^SzYQQ_al6I)GUJEu!yHcpV+cszzxDR6O34Q6=O7
zrA#}850~Hr4p&Iu%Z}4meppLr0hFIJ+izSOxaT!K9hi5@L?HsWm>Egg(DWM%G1;$1
z*UV9_YI~G|ckc;KxP+Qv?3Td}VP}O4LkG^&`?E*?9+#5v0pTYCSlPShnGv0(-*8n=
z*WRauuLp}P&E;qYuKH(8(XLDCchq!XS91fa@wN%Tul{rN6fH+VM}@1I=TTL_deu?c
zcjve_XQv#)5G-|GbK4yfjEqD>Efs^{A)C%&ZQ(EHv^|@<6>X^f9kc>Wbllr%BAF{x
zh30sgT#Y}kQGOlRgHJ4VzvAvnG+&=wkZ-PV+3;Y+i(2AGR+m50ap>~ooG<;Uw?!a4
zJM9qxw3}bLs%Xzy!Io1S3u{WM%9%M?uheYZ?40~`6jkMw4HkZh=5FP0tK)Q^?{m(s
zx(JQ%WmWVikL`bm`*9%MV>GXK*-gHDZ-P_U`1Z)ZG4J18ME$J`!DHKX5IMg_#YJ{X
zKm_KPbm&(C=AQfF?XwjsY`7t^Q?SiqE(f|3FuM(3FK-Ke{QCC*SF^XtE|b;csnFyK
zKV>o&<+`3{Q8NW^&pdzzycd~UG+B_qj?9q3j~H$H{EMDN=(z{g1l39*B8S*ra0qG}
zAzY?0K3+8+*1bnR!5jTVoByd@jse~XLP{i7@&FGdOEhYsj0k_FeO|Me6xj6QD!v;D
zh~~h%XPs3EY}}wBM1CKpE3>yeI}ZK&W@+R3Evc*R%#&yiYyn)n$%4dX6=b4UvE!1I
z1z<_|a?W?#XG%kondAOqZq_!3pq>M6UOuL|O?7AbLe1MQbu$Ur#$OCzo`gIdCvlQL
zSb5?^ue0Jv){AULLM)$CSYBxa-UzdSW>^)EQ`5c8{lGtP&hTnn^(YmY9fs{A6|gC5
z7~qNC9Iv$#ZXqu2rPXC9mX`MNxj)TAQ$M<-49JcbG?~-2mexBVq_{cBhC1s@3j&;M
z$d!+mQI^2F;^F9mdl_}O-=`efC;O33xQfb`TbSnxo<Zfw!9&#kJCDjl?27?dh<+fz
za5JV3@mdA>H!;z0aPg5746Ln=zQiO3uHWe1`D70k?yGY*xf_kvo<}=KIQU?A?JNV2
z>Rm`NdRy(DG=w}|Va;>GusE*cCj1sCa$~k&K`-1K?Uf)%_qZQC)yCp947e(4(J9`D
z{8bno(!vL-@9=sExZqpIegyWpnbDv7u<I~rZGI5J3Ulyx@B9lkq&#OJzn~r4!Y8-I
z&2s08Hldk3%t+ODzw{wUj#Ji}T?$}t#qaILA@=gvr-5(Vm`~dM(-DCL;ND+f1bW*u
z$J{n9R_S$GTH9IycWk?&VsCt2`{tVgtRu$|J-#w}|B;rQ13pU78c=OU*t-58Nm+fv
z6IR;lygt#hJj%jIURT`c`x*ZC>p7|{QNJ%1Zq?b1;~FmJ^<ifyL)Zh?XygIvjY7`U
zjGE^3Oq-?4labO~MpG~j>Y1+z$9p@Lp4|irvYe46<RvQmE7)sPal%$}r)?%K%~nM3
zK8UTU+1h*dVFO#z7*M%EtAzjc6L$-xdYa-`xi?1MUDeD_N8E$#aV!tAk2gVxe|VJ5
z?<Dt@khPk4aDh)%oq0x@_I!Tx;<zZFH7NDSYq+3}`PAYc`geC_oHA+Ez4kq%fhA5p
zrc{9(e>%0@ZC=cFpEoC+&a5nSIw>Xo&?f$9c5&4AG+eXg4|G<B%rwbhe!SB!uhPkl
z%MmQ}lvdYPevhXr7em^T9ymD(?_G--j?32*ErFq9XdwrKxmlHDT+^@qZ)t;m<Nqp_
z=A3x+z`}nWtUcE!ruQmSnjAzeugiWa#dGb9rnWrbQ=Td-#6+XqG^y=(gd<&cLQB|N
z{{%-(I8{&(0HJee^bB;CP>Ybqpw1Dx(sBp?lC6}Zyor#@wYhs%hrvjGc(=NQS4ogE
z#F7}3Z(Tvbe_bUawEl)!YZ=t?c5fvrO{)j@#c}cZq#K{hQp}hG1uJ7=BvaE!*xn6q
zGIM5OMoCi{3KK+8qpPO%D^^PHqIhF3gTycXy@X}}(py7}HW}(1uifa$&N(GT>bzmm
zUvO0jBz5Wx_lD>snhCa!&alrJ8R?nn&g?$Q$+xnx$1b-xV9b&_g_5wjKX~(N9YY_)
zl4z)FEYY-S>d2~y+ukwrMRV-mlDx(jXWBssvmV&FmJBp6Z(XlBstEBEXd`N$`@2nS
zUJ0dvVthWBe$ymQ#lV!|ae+fK+b?TxBP`5S_Ld>YKN3M2?4a8tQJqDtYwERwTdsng
zLqZVKm!ZM&;hujZWBtv6*}CFKsqw$B`VQEt>b{={z!hOu8R?z-4t{?eL<i#-7mz=O
z2O{00SJ4#yvDm{|t}{I2ZN$Ywxf4=lcR~6NFxWVqh0<OBGpeB1+q5X$%!5EdG;GN-
zdHmxbAmmKE&8@Bd?cL2E7A83HvvYOrH)ND-sBLG3U2^Z5`jZ*uH<;$tRH9Im<s5>0
zkl&LX1sySRzS%5S$No=|&ApSg-kFH>M|Zm{EmZhYB`Evb9GIaVsimS5Tf1^k!W{_U
zrfw*a9GemNJ1sstCOsfCD(WZUu!ga**M<<hEAvt-d{7@HBJ;=&G_WlClpDoL$TY1;
zB;VWh1cyZ+Cn2`FnUGcH?)V8RMq|Fi<^=pT7$u$FuZ<K6S_EFeb+gunUe4>a`?Yr%
zx-A&igI7DvZAwL1Y!6=+2?<F<*^>z<uyOa9vRho@<GeAe?fQ$<_eGCGu+`AxpfSgN
z)yN+t*V9e+#aHwyspRHNiUn&UIWRD$fFK(()ouAknwd8=DIbfPKO?jkoKle&!imR6
zhX_ZmD^^{B0>w_?m8@WyuD!g*bK^=hts&BEX5=BRu92FRe18B-^3oe^y1ER&Q0}V5
z;_EfK0}*}rKW}2?#-73DJ7nDFikC(N+(oPD#~a_C_%lvu!`WwMl@jdM+l`p~Db1Ov
z+Lrti&2J4R^AkbhZehbLi5(dJ*u<AZ%a@t^!eWJ_q#masgV6>}@nTEWy?NMh`)-x7
z;>NQP$NY$DeL*cdQbegA(0PH>%+JSS&qCUCM(D)M*-9$|;+Ni5#9zD&=o)76&N`m^
z3_F-?j<cTmr`0pw__s@*P+dAGB8=CrDH+IL58+$iq9N5jMJ9XLtL{oCQN!wk-ko`f
z7#6?!WB}@16DPo*Z_FjE;WPNmva*bEVl=|f>)^PhkKJc8onpP!YzJtaFJ4Cv=(6VG
zi!APS!X~`y4(pHVJx=ZnpTzbZ|J_2zHc8T^bws*MQ}42h{@&S`%6H8Olc6Jjq3*i(
zQ$O$3cXb4?u(A?#FDgStx#WH&i)Q$L6V<5~eaxQgFUH$wV!TEhMqvHb{NJ4VX2d~|
zd}V^E#JRVhV2;ZPAD}`N!ZxJwCXk)k*%W+zmQknk`!=Iq=4ZNl;49U(fil+KW||b6
z$=zz}{nTk_>7ygO;DDdyBNF<NQK?;u$UpqU2eJ<`l45Gpq#xjXd_F-jJ$A=e(GQP%
z5W^oeMaMFPJ1HrPR-ZpMlvRtp_Z){(zwXwfRIPhHeqrXSDg|2bU_0qqxNc!h8ACN+
zVnYnJG(If$s<Tpalk}|D(o4C@Gdiq!Zjur_Q5V%$`es|-Bgga``S|!i)lKKUW>4HE
z7d~DFlrPq7^#~ipBtPmP8E5{a1>0!&65ccY%rutOfz!+c!+-lL3RhgEisAK}2p@sj
z+i@MSF0RS|l>&GRT%(CwvR#RA8jh4^u*&4*Y;wEwz#@u|%hueOnqKGOg`ycUWc{#p
ziqZ=DN1lt6eK&EtuwPiZojDgYbl|tayeOwYl7#z%I#k-qalJxMEKKl7A<sAcpprMx
z0_>{rRdtdVLC|YQBB#}WXC|SK)m?=3G6>8$WTEXLdIq48xEFkHoXGB+#Qc}v_}4r2
z{Tqs^VwQKcb^NVqtSv_f5@I~B39ZX|;A7eof*Ml;eMEN1+gqZl@?0Pi%)uOf;L0)y
zo2q4{U|R#N@3%M3*t?Bz*;Eg`{Q%K5a=0KyoWZiJ@o~hBi69}x`etb>8Gs_9^WR5I
z(m%@6-hipL6<rY5oVsT6-e0P3A@?5ZjqoKA>E*OELf|yr4y-{Pbn&ieWzqff<?m>5
zk0*B8CU;9T8285k*FlbNZ43Y2=Oo%g8n2t(RlWCa!;OJ=@*3KzgNF&xiA9rjv}?zP
zcbm`US-%R8PX@H#yVkDwnPTrl2w~?gGv8rpDL_J5HY_XqTA(#NqhvYAY~EF_Qi9b>
zLH%iI@ve_od~{GDWwKb-S-I*Z-Z&Bli?H^LvAyk<o@7SY!fk#FOmU9Z=P6LU;vI5M
z8_AO66vc(19PE3Hp{Svu7x%=?_n1~yRCi`ppqdVU8{y2MY`5I*1g$pqjQSnl=uqbK
zJ@98<A!EJyj+2wT;7_N`HFkbA?4BMOj4pj|_nVRjS4@@dEN*F5yYbU}{mpoVgQp-`
zjL&anEe(Cp+D9211hu|up-^V`AH+lNW!-zx)Gia|wZC&@>w?%^*{-OEF@jj}^3v26
z9`RLFXRo<4c;{tvsXU@l8RlCt)42S11i-n8doFJ4uRc7*TY<1AR5&UQ_9U}5P6fw7
z#i>3iOwd5lK91&G_c9wCtABGi%V(QWUcp;lGJW~4Rz~(3j~SXQ&>i{?TAy=fls0I{
z`-dg#P^)l}I=Ye4L5<)K4X49b3-+-&c|rJaB>A?9U*0=sa$?+LvacrCy1ff5#6W0R
z`HIXA!)87-4N2pI&J;9{FqcFw!7Hx-m+%^Bb{eW2f2D2ZGN6erTsg1X^@)MHRVwN3
z9c`$aQH{5KR!DCQcc#htlV1pi6>xtmj;dXVe@Oco)YKPj;taJ46nD~-XB3&>ZFrwI
zzf!fW-sN<L$q&=nc)%}$elSTm0@+h3XRt=VqL!wrwoYndTKXqa;w{KtlvNS=M>gvj
zrf&E{m<{C5p4(?@SR+d5J4Q-svSRXsLsA&`E8C2Yx*p3J{xdnz&!PJbya!@beceq`
z4?!2|c#v*K$L&!<eoU+JPexeKWlOOf*c8xY;Am7-)SK?(;}K+bWd(N-Fz`?Bt@*ff
zUGP-k$ZDw3Ugs%m+mwQjqG`$R2D>^QGhSy#Flyn?YKHYFGWIh>>yQ=^@+-gB#!qb%
zk&+vwJwbL5g^@_q82g2v@Dc-y*9L)uH>v^Fz}*f*R9ib_JgcBr{8`9BJ1tWUL^IOU
z8)o%&yK~(kiwr8fVq4i4h}6w~Wl6j$T@4jcxNe#k_qd6K=V3XfjjU0AUW$NQ{RzJ(
zP{%qq?Bt2U+`3j964O%xVsZr@_<4kcVODFoxaQ~Q)z#I1{|?q3X3Kls05?^p9g`@p
zQFI5C{pf1DZ1J~+FDj0@oUX%dYL@_g>L3iU%OO{S;ZZ9J_A(FwydCkdr4hwc>W`&h
zNu-SXzO*Y!K-fKgy<;5_*1AtL!drMg2RwNgmY@G+{mLEl$ig7T-M+hjtzh`JGNSS<
zFr3SH{s?jKnV@OxG^%o@*A4x#c_QngvMt_0Jg!(4R`-_M-G_vm?BsZmLQv~^mMr27
zj+pJUK*V51m9lsPTH%6D)atv+=vc-SzJHgOmzR<8yrLHVvU?QR)?~{huRzL2#?(9f
zfj;f+@EMNX3(d%iYJ7Gd4LzF%j~jIynw_H$vYM&7J`Q8u{TE&E()Y$X0EMI|wHLEv
z6fOIs@^BjLZ}}_<)gXg$i*kpLKBV}E-z@3F(&QCo6rv*0008&b7Aj+nsh_}_+xjo(
zR?w<BftxZPs{T8VkuxR9cn*DbV@B2y{vo^qEQW|SGGFE=N*DQkpQjfImrh!mgxCDX
z(IL*~tYlr4^|@&g?yKN#9$DHSI$K{vPG<yx*XQNBxVX5@B6J~pP`ka6wIh}cZ0*z*
z{St(~%m1eU&^ARf(8q6ndzV9H<e&5TC={3~4}))fv^Eyt*_zq<d!0>2RQo#sZ8cD!
z*v@jxagqh9zoWgf!QGS~lp^ElS3y2BB5?@icOVM`8z1wLaR`L>k%{QEQaAl1M=1sS
zz@(6TC5mCY^C~u=JMvV-=P{t-BoQxs`FBMhDGfY7K)`UxVes{#Su0!p_wFAZXo|O0
zEJL@Tz0ZR2$i=ngWL5Vkg5^<ZsW0N)c%m@GjD`d~<1r&Q-XtVousfu+rG;<Vu!)mn
zW>!pR%EuN(tv@&Ep(%v8M|5);hH7tKF?Hh{#Z!HXH^|(J)@mr`y3;6^c_OMTu=6R0
zB^k33Fv+D(Elzswp|Fsv;>#0eH#)#`rQIqCT@-*C(!@FVLo*Nr|0I0d2{{6LtqyO6
zH2fYIxcYW5VOcUFgHR30T|)83d)pdxC^W)ec0Wui8P6J=#I9xdu?a7bu}RAL4-zxa
z=Ss(X42&bB3?UHJqntt+$F(CQMuy8xq=8<5{v~*FWa8vWwQfARvA4X8zWFi@4b4g|
z9UZOIw8N9^IsWoeN``RWjG;!`m2(KHKtKiPTPQ*ApqwoPmTI2|ZGuIA_nS~Hk01=x
z8~oeD_CqJ`pE}Avp9GeiM%gq^^*aLJIRhhIa33~Dv2Gxke<?qJ_Xuk#6M8Z|qo4!G
z8s}n&{EjCdje7mlwZ4H9oo+Y`(Yc%6zOVJIc@xe1^V6K`L>~#<C&tgOyj*iPm94x2
zRKne7>96Eg_UR2vG+p)nhArlO=EY%fH0+n53WtX6;GdHR+G81O&QOOhqdxqzFrdbO
z%rQ+&t!1~z>k?Idej3GW-j6l75zg#@cYn|WZ#=^7KN0ECMTY{jE$l>fK5Nr6lGM0G
z*8AVxH_3QcyM@z6h{+I6uaxe)n~d(XqI?tVwSr>LmE95frqI_sC#BEG%%LdJ^1h<$
z1Rl$%cdm_`x$?RFDLsl{zEz#n&&Mb4C47nRY4JaDbsQ>KzU2;MaY0)YPE$#M^|b7*
zkEbmyUHx;?DD|RVdrn?AkWO%HUrG0#A5ku=x#yNfp`qTO>9si%{ROa|v5*LgCUyC|
z1QGqx!oCf3yw8Xds|BL-XBO8tW<K<0r~JO<C%-N^s3^n0EGsU~C1YWsX)^q#qavp!
z{VOHXjwYt{0R8y<hLZDL+TH>rQ=HBFxEMLe&?1p!C|P!}PQqya2of02^R$}g{0FZf
zm$E;8R7?3Av2A)*Sz7jac$@DF13^xqJ4F-wtr7W|*SpPAMFz~=DJZUJWo4yAN=YD=
zHchYCpe1HZBQZZ_?!tE@Z}ncvoeQBoF6tPOf>s#|{zh9^9`&w%?l-V!K`nAs&$S>q
zUPg-P5j7eV?9xOu#9Qv*9V%%Yb*wK@kn0snyu#8vcB1dAZuPp#iSYZ@si{d+3tvp^
zqo1F!umOp^f}4gpHJleefvnP)?M5>=wvh3Jd`wJI=VFSFL9?rQ?NLmPB9Eyq-vgve
zP2jC%WbmAGS@89W_yT~;aMbwQo<z002)z~ezYkX9RD6F`gMXdK-?=r-{l+ux2#@u7
z?SI+FG0EyfB3sHQTk;Xq%OpB2XB-p2$8Vff>LqYHBb_(NDAs~8!txwA5_7cG@Ekoa
zvTFY_9Qni}7c+7>y#>`(A0Mhm&8fukO-KEpJ-SR$%b9xUI33`B{N_ZInHv|p6hZR&
zZo7Sq=R!G8tkeB$Y|?rp(rJ2ndS*sRRaIF<CC=oK;ERV;fzLBK9O_A<3Wt+e$8cS8
zQ_jErIrL5P^bUQha6g3uuDj|NW#e|sP6x6te^(~@|Dj%p-e!wAEeD&;M_UdA{-Vmv
z=<(C^miQ{B!7c|_>c6i<EL1J^f#{0(`e*%|+NWIv9hi<m>Z!6<%iI{KD(AHibnIsN
z-Lba~JON2;Os9DrB@2G9q_Yirb>Tq?*=dPDN<i|Kx+W;4gi0M1;}+Su)gc+gdOVP@
z7PY%{j-k}D-dyahr4bek)$y&Y8vJSEH3Rs$u0j7zp%@tXW3#nQn5fV5i-rw+!tA>T
zvGtOSh(?HMc$HAD)(S!J5k+NaCZ!}K5hGhB`@;oOKL8(8cx)Y*saIuUa4+<;j%#*)
z2_4Ui$0*B#SGi{yxPtHA|1@~bMARN&th5P=y)|J#U?Js8B~HQ4HO%iese@|&Ii<lF
z1Z%eDBmpp3@G=BGks}<w!4_q*a<R>!p13LX_kR<-^4>l28U=4JQy>|y2{gV8Xh1U_
zyjSaUF*!{gc+d1*m}NU}e={Y3g}<IV7k%!65n4Ln;;s4BPL^3&UH+Q0bv_h_4<(Fn
z(`z;Dv@9>`b1tc)x`w<ilF9)NG_Zaju4rV28-Z^A*C{Kli4N(p5?jItl*&g97!wb$
znT0cq-pzL`4<fQ3T-2Eo;<9at-=(3p_oJ)$*U@8S7^pKJClMG?BAao2XkJTr!XBe-
za2+wonTEc#-AwY}P1mf-4fjLS{CX!_6?;{cs{E|$yO5uQzb6JlbdKmF*|$lfsCaK2
zQGk|@zaepvYTu49#HQPzo2o?aJkbn?I2SdOcg+eZ{{_JbU$pN0@O@|p$3xHb5?%ix
zeEsR6@k4@u*@;RT72ccbfs42xZWH6Ti&J%`0OPnsourdgbC)Zj@`XZO7S5KX;&Fy{
zn;=4zSC%d!ooonMEtn$+?TgRXuVUuToXx-<xRrOwpu^O<*j-jaMkgsU-g(q8T!7a@
zlG9e)5ecqJz?M-!KzM0h4>`tnH&GX_VGs5w#Z)>FdZFjomw^rzXN=<$n$<~KZzeZ$
zL35tPhs#URsA+hm!E3ih?uBF=)4!CGy~Ku1x^5c#|Ga3pl5XF>sf<J|Sm0{^tru{|
zf)l1wCsxg9Q_LRxi?`?=Ek^$-?5lK?$EmHq5oU63%-~MhrUucreSp%;j#r_^`_vIB
z#wSaE<yUF5(~u&2V0YlA)ubOb;qhJfP2oIM{X^m3Jo+xFpwP7BpR<rq#WxlZ=Bp)x
zP)h0-O6jX^3dbqIPDU0Xwxzxmgd~9TyYu=B;lM|VltY8Pg4TnowV%KxBXvV=D^AMT
z<b-cJKIy5IO*x$ekA7TiN?G7Ij!U9TQH%{y(kewS^78J*)`J5;K5=I!7f|bnam~ug
z&eujHUPTSRavTAomI{5Z4~}TxnE__8gP8NizLI`4b2JjMB{rVg=B-DJGk_^&q2UIb
zR=P+^v{73>0ltTPDT&x#j94upG5(z`Z%K^F^mxhXN7!PE{v!wFJ+GLt>?!)e#>!+8
z#<WoqM$~AdiTQ{mY4eXxf~rW@5t_CZ-&Zqq5~=Ek+_3)ZfD!lg0!u2CTAf^Xyd*6;
zC!>$CYYxkL+<fkj{nHZkNI<POS=FtZY=V5*+nPY=T~^drwb#^|h73T<$Lpi`vcj!c
zOxOKXk^N$HEL~Ww3S&=PNG`(%8(vxq=8|Hk$LD$7yKM2EOrJ0&LVUtOH_3f!IDw4P
zh>gPjuJewApJ%<=OiSsB2}?|j-J?D2fgwZ_D{n+-pS>+Ls;Dh6<AFg6`KvOWN185$
zn<foB(F1+9gW;53&$S#sAxu+7lY@^{(T)C8nZsn{?e5Z2P*RMsNeWw=l)J-1G&U|0
z#PVEb!M-zCZ9yJ<TAwZHq`9jMZw@L2+9988jp2VXZ}^;lxkV+oSM<w1^woSHva=uN
zVfW7?b0jkLXyHySGy<N4H)y!_@9^y<L6Lk_iy%To?==h=4b`?m5IFXxaqd$e!Tzwm
zH?QMb6LJty#lUIq-!Bx)wNC4$B8Z6nK8AH62@-mH^({LR{a6bhG>ciY+Z?;z$wXK5
zGn^k^-n|jaU1i16-F&+hPhxCwtNA3(x?^J#z15TBzC;ka5a7d5Sa+zqy%yPh?maWP
z9|_t*R-A@6;YHz%jo*{YxqCZAEi<9ocZ)DFDvT|pP&n3|C)u1j=+SE_Elb0Y9&{m|
zyZduYPkftQ_PA)m(EY*>H$^+*&srV{!G^h1E5T^oN)e&0Hm%I%nPfDdvkI2&<?X@I
zZzq;<9c%leJ^*fUJ#cfy%8p?zIrygqo$~vHIvH*-slVl@iMfXxk;Jn)Eh)sGc%=AO
z(ZZIP^z$lxGt)=<G<;hz7#KhM`m#LE6ZPgd--8oJ7hRpOtl)4X0VW*Sqo(fVu5AI-
zNdr^~o7m0WC?CmSju-rUgCqj%=c&EYn#Kr92wDQw2Vi*{E?!9nMuQz5%|;?>{wnBH
zR(d3CcZyX5o;Q-nx{6LN+V^(1@*P}Ee+L&2!?;--rDGJ?4)5<sIb3{*(&XbZpE1id
z?;Y%sAKY+BXd{bE-1<0p$&1|ll!M6Z1=kF+-uT%!J7?7sRYPwf+xehxazPm+&jZz)
zzD0YhK#`f(tiTtJuKL1~I^gu>pk^Zdc@fi(<Tm9xhCaf{j>aJ1D*<ODMZupui$L|Y
zE$joiig)&Run=cnf|p7s_A$U!5&o(50vXJ0ZV~%_%p9urI%HY5NIp+L4w(05HT9se
z_VHt<XDHXQIHrV~H63Ms!qt$UYN%YEmAB^uGf37tMJy{VPhbDy@StwuWwRmjF12}R
zW@o_+2v_~rR*BoR5L-gB)w^HJXNR;M`m5qLy=-58$7I^PA3K=7EzRu5i*+y*K%YpZ
zg+Eg%55sPq&7X-l&i&Ql+;^kwIG3>&VGhMW(aB+`nU;ce^&H;Tr>?qQFybA|=N<>-
zHM5Y-|HCv%P;rzc=B~c+if5nnc!oZl*pv=IK|($clcg&r6)^Exe#g57`I<^mn`i@~
z3qq!OOL-|s3b}FL4HgWt$S&+;wy+_vPc~xYaRJV<#4}BucnNws!K4qd_2QtdLRKR0
zni)n|x_$y_k+FXaQ9Wpkr+R|EHN?QptypjG2T6V&^^{$6%Ag03drgOFai9+F7$@a@
z^V_^(@c643EEAdgGgR|Y-TO&OIjPa&k>iZaIBs^A^t(xmv`ZPFTK`z#^?V4XG5^i|
zGn(80y_t@kLwIg>MM*zB2aEXBcW~i(&4~+^PMO4<e@32AiN<Nq(ASr|1P$mFLKq_E
za6ONamihz@U65Fp6QV0SOYyRHCg;k9`8KIC)m={UO`0$bI-S3HoOOstQ)hKXLaFol
z(yzI!<co;Vg;{RX>Nvx}q@@uP^5aADfZ@yH#d8>=F@0<DW+6+eg<!}}i?AE(@T9Ng
zJ@9O>LJYBCyMB)Q(Rn1BCelEgp5E+=NEIcU(AunuZh8jV3_r_(O*ID-1Nm5=xMGVd
zbe{6v;@TiM<1DKCZWh=5+1P~;Gm>MdR2a2`K>;zemXc=fF7kp^+1>yum8xGb?`Vjx
z8Rm)iaIdWXG#{y%<SXyBA@$Y*_@sDK%AL3}{m80};S_UkPzM+J@);n#LVv%d^6#ZQ
zi^|Bs`|;|I&QRoRG_dixYKaB7PeaExqmfMprZVaUbeIOxMqy3Mk`=;*IeE)nH!a|r
zQGE>-bgpaSoZ(|+A@3dRY0t7=-ETnF+Yjti?F081$V=EvS`1UAe#6%5BeG?b_IvHV
zm^q?=zZLhmREZO?2mbun#Y|KcvErQMgttk1r<Oh036=CewTGje|7>@g{=ije)V9C+
zeyXpF%CM+g<q<T@{KDZQrCAn&Bvy3tfujT8{<HY7w8pWIBy%gXUz~0fc|EUvm9)FO
zREPVc|IJZFtK-YoFE*_Ks$Tku$8Ekf9fWh=)nwG5iz~xt0I@kVL{su2D6MZN1z$a`
zigcrvR2XgUfC_HCjzg$2q&{dkSzG~c%WqDBKv%2=bhyZ877Zg0LHB@FdUmm~&n+ln
zWcWAP7?>R4LzvK~^TW5r{EC~gPf5ULk04d++};lv8%7wXl=aA_><lzM-rVdzXwLC2
zu6o%^X;7GeGkwy>P?|q@zbHTp>muwIew{v~_GKgTNWihvRmAWSh3vcP8*<+sCtP?P
z7G8&-)X>#jMM#%sbvO@-FgJZf<jO;{0`wI!yH>fNqoLW`JkzY2n1VCGxctl)(1Y(Z
z1LP@~%JkEGBWg_4_Cf&ygLoeVDKTfg%c`&-j}c&C?)Hb4t?E-}M}Amb?Cr#CitFzz
z_~&MjN4J-C6*WjPPDN)Fn^naa9>&R?xhy-$^wFnl8wPtNsXc;P4>q!anf3a4vn`=s
z8i*yto3tu>w_DXs*?*PJZ79=x)g+4)sdkn8w7PZs@LKS%F?-T5rJh`$WR4VE;YoP5
zV+=UKjT||%F-2!Gvq5)_83`fNI1oLI?ehFo_aH1Ejq%$y^;5Be*!$fQC-p}bS)r(Y
zM?=^ab{!t3d9+*aeiq@M(R-$zdrBU_SK^Cs25Uuh_tm8C^L?j*Y!>gH)`nDK5>=yA
zN^Q{!3t!LgQNek?Qvv+tw6MbNm-3PG-vR9!691U~yOIweQYmwdyIevzZAJBe-yPuV
zu#slk?w4?H&|_ZBb_MC`!UAsp^Uks6RM-%vK8jEEi3_e`h;FDPecCp%E2gW{{pCW#
zw459~`nauQZWHbg5l_fX?d>@VeGghs$IjTBe)nwiFIZ6Jt9YtHnXLH6>wk-Vfp=rw
zxtZ}dF?E;8C3qs6`9;9*Aq>|n1U>`FFjOd@4$JZ6vSupYUe9q;bMu=-2Hj`g&Fi-y
zy89EF-Gw_7J^H#j4vwrnx;%r%e^RFXFfFM^LwH;0uV*5$HzjnulxW#Cju-x!nZF^e
zAw^F9@F~wkj>JX^^fO;b1kz7vfk;Hhn$z5sMn`SK_MmB8Bsc{lx`93#)N*?fM5*t!
zF#mQYBBF+9n3FB$asEg)yU)ZiZo7s;e!qi7rP(c~*Svz{OiksT9Y8>?{macPqxek|
zPhjM0$%!lW88o}`W_$tla`jXppTB8vP>j5itjyK5r;R2kY;L<yOweium{ZerQaN8Q
zJD8&i@?8F0jHf9?<l={#iD5~w>?Q1-6)bHkWudf*=|6iqN86vmwa^|NoEB18{(%Nr
zCrYcj9O58d8l4&xl`SDHE3=ctrt-%J*`Mcq+xz*u2L?QH4R~*@6mt1T_okdE73+am
zGpyJ5yMpK^L`f>$l4H;SSd6&5<Gitef<~ulDl%L2MWLskA+OkTKY+`~1xer{5qoF+
z;o`N4NrCvo?;Qm-E|!FYx9&mW=c`rN-G=d=r*u`l`KgH8?Q@h9Qc&zi5Min8CALKS
zDNCjr?!i(^ARLYp^2=hEa%OT6GK0oD#>N8rf1V=_lEs65SfJTpUWBp35`w&ekYI%(
zemOW!)ZRCHC0k0Ka(GPx^>luE{{ik<oG8*O-8nzktEdQO{z;tF18Nchwe7I8uNBXB
zKb$o#;iH~}-B9?g6Q=-ksIHif+(*0K&)Ldqdt+R&77p~s)!V-I*){gg#ch8Z8AT`_
zf}0Wtw1-qfLI5x!ZskD~Q~A^s$3?^_X-G9X%UhVH!zg(NvHwLgJ8?S;o+8PwjxUd2
zg>HiUDgD@k5dG1jCw;yY=XMTWXR2;XknrlWy?X)=LcZ+(ktEb$;jDQ=ML!LKq2({+
zwpWzcA=CDS@ct`i=jIpG%D}&vnvhl{ut~AizE9zHU;sAxSv^d{wJ6c>S41L4JQp{j
zO@;^eG+kVFEQLh4grA4cY`*CrwJn!yej5vMMP78FMo13>s|r{-&kr-OZmUz%Qo$F#
z->}HU9M6@{#PPv`1as#souBO~Bml1=?IG%tN{1bsbd7L|>j&1ZCt<EB4Na($ND}_O
zwB}DDvw)j*gq<WIUgaj*013A^C&!-*)^@K!8g@3JcLRip<0bZM?Ma!O@0O>&qPp$D
z_UJ!sx4h-QvFR-^lqq-l+C--1PZ2p?v2Hx-yM_Io9>?Y;r3JO2?t=>;Z0uJq-46fM
zD7aFHEFCFN^Y$SQMi$D^Mq4Ool7r-6;Gs?D7$Es%VYH_3wb7L=YcN%~(VpU9&3m_f
zCoxs<bZRF|yQbS@v*<l>GcQh@6b~cRsBA5C<NB?V3vzfIEDeZ}yN?Pma<vuL<`LU(
zP{8o$P0)K2`z&78C!D)Uf2sBCGamU@aE%lZ<p>e6@?>EwbAqb!459neBWcpy@$Lx%
z11+&_S!rN$ZF1a^;E8R7Z(0lSvw}8$SJ{9mfixR|z~i9ByDH$IAQdp+Wu`&MqA^fA
zK=j^he`;W9ful;<`hoYEjUjn6%5?#!@Eg20sMFr-cuH8=P3KEgoXJ)wNe)PKE!uy)
zBkoVa2>%+V8ybtOX67OAlz;1IFfz~z7#~ElqU>w(+g$NhHb0FDzA(pR{dCK1dA93~
zq>1}sAYs2~Ec;rqF8*z_zo+XvLC<hbrySkKCZ@HS^e38s9w;+bwEYEuB|1Dl;svhb
z<g=C1BEa37W1u8+SE}iu(f66U<l+d~;Phd(D86yzH0U7K-Y`@tOh0qAi5YhVa2hCW
zAZ*OFD$gz5XX^i7n4lNYC8fy8lj*$Ef3DbYQ-g#yM?mMw#Ft2(^Y2oZ#ABNh1+64f
z+2QI#f6C7t0yk+jBixZ6Xy}b<rZm0H87v!;M2l@p2zy#$YdA64b&ft5SSjk^f0I>7
zJ12T?${$CzGqt)_lZ(`Wel0wZr?_>Ct~_W#fW6Ji2dOepJlnJPQXtHGJ95yyTwc*e
ze1n=GUmZDAz!*B9A)_%t;(v&HtF|~AE?66v;1Dc$&_J-@!6A6?;10pvoyOhW-6goY
zySux?;BMbM?|1Ek{R{RS%;{KnPp?%~cM)<|j$x1!$cgR-+9|TwXBktA!L?))>WtjD
zFSwcQeD4dE(wAR^Mf?=gIv-*6OqF==(QVQ&t*}Kga3vq}Mtd8QcHUnrmSFe`;KI$*
zh`dNgR|M(qL)!+mX#MNhuDl_8hZ?a29<ROX(5w)Q+g-1=4qz=|V=^}Nc(k)Ce7Rxj
zz@p=f*u>^-$gp`Psra6!Kdqh|Vht_)Bgo8to0TGePd|UH%|=0ft#l=@BAP-;xAB&Y
zR|<D?C|pw9IjWuYV(^(;5Je4mWM3J4nvtNyD5O-yrKlniH%oWEbyr%sNs;TA{2hKP
z9O(F49HvhT%MO=$)=I<XB3z*Cm*n6W{a4D$Qdo%uUwsxyGA!2Et0x{dm}8wEad~f=
z@{=ZmJ6L+z393!Pq(Bci1CdjOJvsS=50XH+7l^E1qd00!0qOO(M>hf?4zogs(f-QE
zM&hxwv=!O^%t1CE%VljXe6J_GqW<!+af#@AdIlbUvmo9u(LFIfmA0!JO+ZQ*@@92z
zm2oyft|K3d=j2xs1-G`v(=gKU_V>6Ot3slfA#!GTr|z~GHT;w9iM1=^b|=ln{mM;t
z2^sTqqtnz&-4P9K-5C#v!Zk5xx6E7V?xAmb*}|u5+{)j%KgNHRvoZ2Ozbn6v?ajF$
zZCoZF8Ex>hI}L<o8Q2Y(3odcK7Wi_>a=^j!nl~%8ff5lOX`o6iZVlf!tIgxkmS5Hq
zl$HnqQfJ*f+$MRE)oj6fZE1U8K^1+4l%-4+y!3p~ua`r}P!|V6+hq5LObzIYUIeR;
zDI=QbD;{b|23yib>^FY#yL-HpPBRCrpUs7J{(x5;&4`Xs$Kt60PYORb^D+>JX1Ksp
zi#oi!GR)?*i?73IOTeU&%_QMB-l8KuHcRL*u*(ml@}`t~V1Sycn%lU%wVLO}@y`%B
zE}by_la|GwQyicyJhgFN?sr&BgofNze8Flh2R73so(nj*Q&fP1$wlTE%<|>Bl`91w
zrwF(uu%b)~jX|2}JjK&)*#;r;J(jXJ6$A_?MmWr27Yw#{nTg|*8C~<Y`aFNIC+k;T
zU2Gu5Fm7NRC#<t@){PHF#RaROKZ1}v;GNg47;r>?N&*VGDCRhMB2iACzd3ypCz@k#
zf%3^n#v7btWe>_Cu&VqfF37g$N~C1!lkr!oicX?9Uui=7&Lu{D`5}dr*@tx!tTexo
zWs*EjNQebkBG9)v`y9sW?<YkoE`l8*>-%~(zujzwCqFw7yYkPeC!`T>A0F%K*_?U9
zF!na|>i`6Pb#!ZAgPr>OIewP4JV1Z(?d^(s=qMYLia@%vhwl(Bgm>%HtJOO%4S0X?
zA4X?;VQ?)Tx<rUAPMC8<9v|!CrEe->TY@?{Jj_mP%0}woU~r42R81kzE}g%t;O(6$
zeYW{hd{MMrqfe%(d}~+5>HJ49JS@_r`lwL3M+>k(cZOq#X&(`rOmDPnJH&<LdtAuj
zs>^}31akt6EQW*+%cy>yy!-;MUT}3H4d2**;&@$o!q<matV8eEpFSTTh?h;v1ZSb$
zwHpJ+Yw{wtAR-F9n~9z<2_b!{LGl8pgPPr5xu&m(*k3T-p_(^0RnwoZkF_)u*U2S|
z0|8Aay-&9jw!Y4w;HwCD4E*`GEdE^^TpP73OZ}x;_y5Hum_T9oXjolyub9{1yw~@$
zi?4>`1DS|IizbDYIhWG@m1*M&Ii(*yhvU57F7{cJ*sQ2tul}>>sRTl~!5ttIXT$-|
zX+AC}7`>qE^o1vahpeKu{_$?ajG1L%qeV0Lz8OJ{3Vp>GQB{Yu9WRO|AGbAhI?VI^
z6+>*|)5QvSpNE<l*4*l!ihJ7NZ4PHv>sxvLJ(krCCoWcHeun=}c8D0BZQvy>R59;$
z&02F#WN3#-3X*9ZfqMHoNH#^a^{L}Gohfk-(nn;J<J?|*qzibqt9iGja1sOBwmML(
zpn>_8h+oClHHxZW3B!@2^QaMazy9JU{D2YZ68wxm>K(^tiXdLQj@-(U@Ep%XD1(q`
z|1y(|dYc}T9)2wG#nUi_JSZN$HKhkZm(#DW$F?Y~()o4V%>k4TWrVzz9?OYUwX@8e
zZ?gU_J}`v^$IL#N9E_)?I}kvjls8Ks)+cRSFFdFe3Nb#<zWd#0w-Y4A9Q0yX|NGD@
zQ2t8IR0j(2MUB-cJ8f_Q4vxxCBZs^DxT?LLA8X(7W=tZG1!AXiKL=BDaM=ogZBKqN
zW)e(^E>>(bu$b^%#(Y@G6m%Ox8J%=+gQy^^L*oJs7z;)(0{p}(xjsN@@v-KP{y{9-
zmb?}3I(xDU=S@Jx+(24Q)Z)^!Sh=9-s7uDzGG~9S!sd3(5MK5p32zx)1`U&t`*Cu!
zhLy9BmYQrxMz#a|HG%>+n^NKv5!=dtgo1f3WX6_4$|Cf8lYVXiES!!lv2`>+69#|)
zUr(+MwMka0{`(kD9AF@WhKhfA7Rtt@>-*NlKwOw8*8|#Gsh2(X&nc*gdlCM?m9PA9
zEot86$ieqAiH5~ln#UQvq=rwdlK6Ss&^TXBvk8A9_{jCXScq3qyj5Xliue7gckW1+
zW9<{HM1AAUf6R-^aoZo$(chb_g`+ur7*98#v;9R<q&3|zfHQ@M9=*%K<CqPnA@2+v
z6b5y<-$NHYRS{C#FZ{td`G;Imhl7_1BE@EPm`EuR;Ip{tt`E{y&`SAnl95n7B*<01
z24UD6%liD}j)IBxSxJ53RnE$&uCbtyf*N*!=yh8uKM}&wpJS`Fys%^5G7Gml+Nr?$
z`sRCKpjvDa@cDjkf-Me<Rw37e@ORx-nudvl;8LH_9++SAMuE_J+6(rBj^(B{Ojs=y
zWTrlGGEYs*#bm|!6RUdCQ-<^pW{wdc))vQ3h=^TW`q#Bx4@~47GX|Gq+BLN!x;r2_
zBCFva(==MDb#4Kfa>G-05;R$?^^RyR#P}ayXp?h*yD1EtqmL)A;=B&l7soLc>m>6-
z*-{X;Ce$V3hPYEGO?s0q@WJ(28U}x0E-9t@9rYL4CzU^p=?bfKUBf*BQXP)7zAx<l
z7))uw<<F*yr>jw5dWb@{nFHfOg$F-3$HJ^;LRKa^@z$|R|7BBXT^Hu_JDLK!gd?>1
z<l#49Swtz~T+RV1(`~LW63wBJeJXxGNR@nNJZ!3wtP9B*;f&kYWOx%Ca7|ofZP?B)
zKj3?_2O|h*(MFDh_WV-AnNBfv(Jvb!M1Jpw?-TJy>&4Hp3KxLw)WkNMe~bFy-MKXR
z?7`wI{zH)tQxB^2<}i&Zb3GW|^nqtHjSVQV)!jvg3kgzlq7fgL(@lsY1mq)$QtL7c
zN=t|e)!`A<Uf^*hjWOHfe)PF!s(VE7-R$pb98iB2Ne;#hqfcc@^Vt#|<$k|{El2=o
z1htYK2!sS)<FYBtE#01{_%PU0DR-*<MMoH1ho<hmXcW|a@COzjgfk)xn@eo<q5#cu
z<jA408U>;DQ<N}>x*i8tQ$cL%N<)iNjL%srn)0jux)rnH{?1NLIP7S#Jy{#QFc!Z!
zqrJzUKS*_J<?{9vr|P}XQ&v~zxxF#g$Fc=o1Q$-xqfVbsV$c4D-BVK+%h#SyGda0n
z#x1V+VDx!WoJqaex7AXi$#m(hJpf@<jEK7^8$+BA#BBC%UvyL<nPs=8gt=x@du|_@
z_&Ln6EGd~$PxxhJo=Gg&cF^)ShZ%ATjU73Tl*m3;^qL4={phQGlP1kVYjVDceZTS}
z1>LUV06fgr@gAWq3ut-Qs|X`a7eTEXO5K7@QMCd8YVhceeQo=i4p;-TivZ)lM$i*F
zcL)*;$6W?!q|t;RlZc20m3^v!=cZ*anJ1K!G|~IRS9eddW^l|Jo+6=V&<Nt|NYOKC
zue3jZ)6pI4Ls*ppL-e+x7lrUI$#%bhFr7aNN)Iy?gBYg!qy@q-+$=Y_d=dRjS@3M-
z4K0Kei_0zXGs5w&p~e<%c>vOiEk!#?=WnzIVFBAY7Flk^V2|a7Y$xn0C3B|lbd-OX
zb}`T~R`Z+&dX}kw<ki4au-jmJU@o#BTlmpb%K2Z<D5$-zF99reAkP)Bko8RokM7~6
z8I0?>6EVeWUkdS4c>TS1iZgIqz4PEJ({jMkE`(`k=JpIHr@(RN1=u0a{wQUXj6l8h
zb<JpbCDxM>mCP;pOpv#y^3&jyNO8%>IQ!9erYt_$SXT4j5f}pU!%9n_^;1sxg6<Xr
zoDr(F9&?GOWa$?-oJQgf@sY|45<J2$Ujmm=1ShG<-=DBlTBQ!_a4O7ivm$)OKzrCx
zxcH9aoD8j%k3j<mzQ8{?3LR>#R&Q%mqUU;=l%i}Sp_i&u>N|Qu9_ZZJPy(Nep_`E5
z7$2)(d)i?D3m;+!_4+5*5d3ek2G?)<u!38YD}15!Ctqz+rEj;EKZCCe2|Me8uZtI4
zgo0;A#Wp=7{hj8Me3W2PhTDLg10}JUHS%@^t_0el;a4Lv0^r7eFdM`~O+LR+V4}Vq
z;j8jK$5wgSD7}O_d^T8VaE`48{;#}0a8svs4_H1HUwFN<MwO;i7FPay*}*K>oB&>x
zftg2M=y*?5m8F(0E-run{&jG0u*fbeAU0k3_2~0+;wb+M#iHK2ph%7FH%foKt#i`y
z7Qh51G#l{s0H_0w&eZ;Y#>c5J7mwiCKt=p&3Ky)C#Bhm8<`he_$9i|~p@;|E-aauA
zKJgdAql;5Z;rW}g<f@`fnMLVILM822>E_MV#wV1%C(q}fD+;k+-k0`Ans-a9L}aDw
zo*4oACCf%9^1mWR`un~&=)AS3GMOEW`WSaTOLzdn^U%=crJOU%SEW#|ZTrQI9svy{
zW&f-={1enl9m#qZS|4aPj?BKK+pQlFvO1)3e|K)c+lb!|bGUsJA)9=g$dEEZJ2mjE
zGDj)e)%HGnLK15I^_Fig0RvX)<=Q4)<Q&DCg(Vd9eZ3RE#Ht;4zk-6^Cv1W`ijjz(
zQ<{NE9kC5;ssNXj4E}#q)ZbwCAwRC=PzU``jep$QP}QpJH@)u@SjlStqBg|&!|sGE
z)AI)>{l(|Kp`}JsJmB1O5Ox11`I70|^uR?317m7V*9%4by%0NhpFZb`awg(&|JVef
zkL#TdiT|#vPzVh^r$5C#iT|x@Lu0!~)*uC7yqC1tco(^I4+0y}%|dHpzhV&Gio9V3
z?NP!m<C>E_poL2cJf*i=>#OmL*4<v+Ean8RadvX6Oq33vV;pQqv|8@!WR7n!&-Lv^
zW2^nShKNtSx{dM=pQ<0vcr##`joZ!Zy<8G5mpwz$v6)XfHvM*iy`5i_^Xu1CSrP+q
z#(C(8Ir^NPzr^~NxzBb&HB*D;5IHj~$v-gw%}nehMXPmx8(P9hlJ-=i<yowVzb7Jw
zFL@!s5Ipj`wr*50E%3V8WZ{XQl49KI*CBm$6~ibY6TZck1rFN6MC!nwZiw==n7Uc;
zi|UEBy)y!Kj}aJ{W__XGi%OiI3RM6(3nLL}11V#Y&-XsH4|=L2bAtbVlc2#%xy(x-
zR?k}ZvRNh0IDdXj>!=)oN4QPsm4>bKy;#nKpG<FpIE9jfXzHq@E;#%g!|U?8M!Cw3
z=9jy%rwsN(*re{FuthUo`-Igmrx=)J$NTW`|N2k^oT|gs#pj2<(bQXcY1shTKY4FA
z9C?-esRIqF0}eRShL4h6w+_0F!|pSyLjC?2?FK8GS{a#HxIn~LRcRfbS^-alF1meB
zswNVNHP*sRHYayPud%d212#Y0bOOKJV39?We4f$w_ztvW3;i53+06`2_K#1G4Gj#?
zQP{rxW-yv&wZW;WhfAZ6kw5?tH)q8_)U;?UGNkjChC*L_ASelbe(VxK=7CGmsfA<M
z#`OCd85$U(2x0o=Si5ceofXU&fK;-r;`$&abmfGmrx%Yvblv-u<;?b{LlNTxeb~fd
z{i*XGv7pI4*Y`_AALmMf_x^VL*Vh=|P$%wQ85}1WFt-rCn+<qNERqGp_vo5n*Jo#E
zS63GoH>2P|g8}e74)u+!)4Zi^Kokv0YUc;Uu!+akAnv!z;J;8mEV?4v&VPR=<bi5^
zpe-hz=Q1|4on7ZRQ*siy*&?;j>QX=rY27QN9^?z6B{6av>`uYm)?1e8*bfTsQ$!+S
zZNpAh;0{#Og!Hi<3VQ&YTT1~|dG+|i$v6~@KWb`A&5tl5{S*(xQ#_UyuDy3;+8Omk
z@z%WSS{X!^CdS)f8&zkJ73w0xN!A=^m5OF^5_G{0kloUz3g<Bh#UD^HMo)988iiTW
z7N<a3(WB-57kfT3bB?9i;XJT5Nk>)4NZyj0Mn~}N(>qi<eLYaX_3a8nP?jg%f`K3d
zt_0=t>V{yu<*dTJB?6pmyS<#*8%*en*k_mIW%c9k4l)j`yDStQp9GwHBg1{PVzVQ>
zWDRcjXPz*hv!(>_ylb=HnFo%^6A!2H0B#=>D0D1-ibtuqTSf*#vmYec1y4FK@15^}
z-g&B{Tu)f65x|YaL8Fr}S1)s%)#)+5@dWa8SvC|H37Xy(j2?1!tZQ8OOtx>?n6@|;
z;Sg0q2U@2e8XtjRrj}Z6B*Uw(O0AX>cO%zkxF1dGeokL1YNVem1|^~I{C%CZ2sI%)
z|EPH&5>PT)UYMP={DmH!?iXu6K}#dyE6EB0Mf+7;4uByi4fe(S4D%;C+uPflG&Y8i
z@-W?A6kTNX>GWgb3Dmw+aXx$8Wq9(hRK;4mJ*8x$+<vn0fzQ1%Al*47!+oyK0S!Mb
ztESUq{lSaEZWLDq0{c0~mPvYfeM3`6>HS2+Le)E|)~{e$o|hhbs&4zQaUr|{Il<I^
zOedRLJn&dP8B79}4z608gkGJUF_cqhb<e*XE;)BUJ=1q~MGqArGTMjI60E$4cB6vi
zbYzVQ*06jN`HaM(mz+m~uk0<23-i9N)__TPt6Sda*85KsmAmvU--9u;ZZ8L#4N{BE
zj-%M6IV=i{c3|#_zrZ7hkI`FB`q_?LF2$ei0I6_IFucq+MrJTSt#J>16JQ^Z77E=~
z%-lcc|9a$9!#ghiHZ0z8&iSUI7vOTH{(2Go0iqYQXoai2W*}`BLsG8v@<Hp(+sW!2
z8)5XBH6)Ld+k{PjK6#C&;k2|95EhXqUURU})qsz$0*p7|;N3!?7qsPJwJvOCtOOcw
z0iU4uD}c)&1VnlMM+DT2b(wjC@(I4fu`;dYrR#wI2(1UQD75Dcx69#mZMZkJ^)IDJ
zmp48y_QLdGRukww+J!#;@j?f!zs`JX_Y_r|43W~Z3i<ek^jYnC^!r|6ndpRvGk>y6
zmzzf$gG89mOZLeR%&)#jB0*+e<sbVbKsFINa5;9K+>>eeX8{FMCmP*j9aK!Kpkn*>
zsK9Cii=!mRYvQo>yk^*_;d~(maly{H0%LjFmE~du-jJ1o&Di%j9I6t361I0&CPu`i
zL^={oSPpjIIg%?qAd#=60hBB8<qrI&q^nO-v?hdn`$v##qo2TNO!j)mJCrCO_wE#G
zt<$PRn0CoOS|dpYpC(KgiH`PDO@)T8&`Ym!-v{LprK3fgLD-8jMc82Kbrx*%GNE!J
zE-r*u3comy^27V>vXoULlu$0_jGa_zT5C`dj*3IbX1F6q;$cTHtp1dMj4Av)`;S<9
zprm8`VT)yl8GGD1_~Sjm#-bH4LXvaBh+Qcu2d>-1<Q&vZS@1H5voe9OI7%<HfkA?@
zQ2SZp7cJh3PO;>WBa$os52z!htj%Wsw?4pk^D%AnD&hmNtmkF4#qpHg0)~aXk3!>^
z)89wD;0|^%%}Y>kI)u^3SjT*FYLBNS6p@%3Zq+=LB`%Za(vjD>M-Ct=@9`B~*P#R5
z5IwfVdxnwmJsf9@QlkxpAllT-yy>X%vU}0QH6?@-o=FBn40Z_1ZIOR_a1?VUF^c@m
zVC1vm0dP-Ad0>1zg4npRq+R)*=eLOzC-zDA2~!t7o=tNrC}3NZf!fHyd=|Vi*RNkO
z6_^Z!2fO;{JEVFqn}GHGv<@G#sNL9Eor>Ssz8Ou`G2QPQg(DvMvCYn-`|14{mi-AO
zJ-|_-MhFoZ-*+ssca%aumW4;!g^$t#kc<SGmDBUbt~_2@Tg04=KznEiwt1@+YW|nr
zOvCuOhg&rL!lImGH^3U##%TO^WJJ@2^LcqQh!{~OHKMqF7~q$FwI-q6pa))Am$egd
z5HqPdDX}8oD)Q8;G$g>IDi}r7Xn%bvbuNgRX3x;*D?~C#)jq~RQ})l??9_ijD4S({
zeR>6feX`qp&XDvOZL?kCy%hG01S;@xz?5@695V|SnCu@BkcU&EeW$wGTJlpw;{cX9
zT?6{|?#)gB{ERAFhx-;>;w)vk`Jew|6`=-(Z}I6$2}>K23AwvOUqqBS9Xy-WKv~2S
zyu;1=T|j`;aE5ueP6=7zym#o_UYsN@U%OLjm4{%7X!5dijdb-6_pmVIpjL~Dh`NhK
zJnYn+_^$1(tt>Cjj)i6TTq41K&Wi8!;s6f51Mrd>S6`BYW7pJ_T=EWQn{@#Z`N&0h
zCGLe4+%a(laCbs-5Odz^s;9?hVzZxYnIu%il^tdT$zskT(Gs)@YPeY-*O><C(dwv2
zXJF!DoSoYCZ`XzVOZ7#dhW;u4)aY!|*<~F_-+eS+bo{S0y$N0$SB6N$D*_7Gg7nKE
z21K-=OpkpObAsdBazg`in$HW;-t}xvmt944tvV`Niy9j%0bLuDrpES?!Zs!jUM{LV
z(D>**ju=|;c_PNPQ31J5#swiMafjlVUI5CVfXFT??U(Rh&m?cqKCf^Vh+PJ<lLO@6
z>lH<$CYO3ZJ@p!pU~ZVWk>dOu7bmk1a6A=x+raTBVb3tXPi*7iuE}HJI=SP3SPjDp
z$1E1e7kr_dYOZbS0b62F#xi^a!<ih=h2PQOm~9k$kJQnxU7>NtB+XXb#J*4bbcYq8
zGD`gOXP&LW>Gl>44Fv@i75Vnz&Sy!dfIlh%`QE8j*V3rC1OfK(_?!4Us?_5LK#;{N
zr+3dakcJdT?2^tVxQ<{l%>LHRkKyEoCt=wjKLf#{&7|YdGcY_)N6PvWQ40y0A%{gV
zTZEg3b7~r4&p(qFvHA4agBX=)NJ<y}^gTp$9oIXqkZ-zK+S^sP73Q}G(ze%a33=DT
zb(tGNQ=<x|ouv9+;vs~XwwlxhQ2b;%XwLVL7iT}3rq12s1|RJXgXrmiH+<A0h#hzH
z^<5&kv*?CP%7BOsR(`4u^2=74{Rtq<bBh5B=9F*{{T6IZMQc;^c*{blFlLf$szW9e
zngNG?3zZMGwmjrT1u0cMK_n?%d$r=Xa_<D;Dr90C2ah>#BX&I==@wl;v*j-TOwkMG
zpI3L(tXAxp>aU7uqT#cdLTq9g?Xhb*O8@#x*5bPBNg4QPWL<~mX5MoLGp33d!KPW}
z<1@ma9DH*@e;wH<xfrAS1_p`vgg`MEP+A0%KVrvERrbdWg2k{_VUe$npyvg#=im%)
zNHC*bu`0pD=j^S%9=d9P;HhU>^ZK&&gOiMi^G~h_;q^XD>Weu0=B%1ZY8)ok$pxZ^
zn&ThrzNVBTQ(f!lq<P&p9UJF~DUj$<*r&7H>!RlFgX`}OBll<p);2~rtKOVn!9bEe
zhI&7-EhSvMm}q&0MYfh;rI6QKFFm+1!aDYNpG*{wIOb`k&2NDKMI_M7aD@hmegQ}O
zrE13KFBr2K+4!}f{6PeXSGN$sW1~al)vXFnB5dwQ&-Kak;Q2o{SdqTYu`-i0;*3F8
z)((xXa^$uN*$CKLDqUYyjAYTCj&c4E*Ir0;>lgg13qHu{JP#%nw3ZUNayO(EaV@lT
zd-hx<FmIiWXekhYN*@LNQ4lO$++7ejgkbqQ%k2Y#gvuIt3dsQ_u3Yz=#jEa>+IRl7
z)+-M~b$vz29pi^`c%~s@@q`(i*lOFB9*h%1Ev5;+u9CyBUKWC>a)s@h@%12}#7<5;
z1oK)G;+6PJ97Z6Mtp|$^66YGgV6$AyA)PJy7Hb_X<!As7317@y5n6h$boXz*YC#7I
zYP1WQfCo#U%Ni9*)7f?pEeiHNyGbRyKGkaAXY1RgWM`24{8FX!5cpV_7=8+zY)8KY
zX)13ra|Zl+!$I`2)zj0@ZvS_tq}~%3?Y#<0novt2;Vb?}Lq3M``su?=Ylj4if{&uY
z%HxaF>u<oCD0Xx0#s;capq27z*|fN%qy$|~NDd;KIx{XvM^#hZtFfkKp*xTIoI&eB
z<ezKus56R8haeL_KL-al_rjvKj(qanjv9qRuT}%UVQx{i80yK&D~f&sKTP~1x;T%K
zfkRs*Y^p8HM%!P@HXVs{b8Z3_Xz<DOq`Z`F@E0&LJj6i{s9%MA>m4wnmR-%XZ#e-u
z!rxo`*oqz>8pC9rf<jrzxli2*?;GsiAfAx9-`dwwbS;WUK;wqq*Qu0x70$+<=X9l~
zsBr41CTitAbnSoN!=OOISeuv_7#J8C@%6^{4GBRB|06L(LqmHpyTHujcJ}z~MGDji
zO#<|j<|HYz)jdS~&a?G1jOkDw8*6*G8$FasBjPKDbFkF4Hc|SNf~^}Y`)v@Tlap;p
zUMDjv6AyG@yL~HfDPpS!wroGD{+1pL{(HIR;frY(S}J~Qq-B(I^MAYqW_p%~Hrt`6
zj<VJtTT|fs>2JqeA~r)t`oydUhk#Gs#4l%}=;B*@erxyVrs{1$X>2AP*aga=54ZpQ
zRPWoP>SO*_*J5jy4_z-SPqSTps&WF}Sy@q0NlEbD1Q6-2luF}%aAHvzRdrmaF3bCc
z)>C;${(mTLztnNH2U~@b_WveuvRWISm^rJ89MFE6d~H<8(P{ME#Nht`K(nhU(;Evd
z*HBfWj_X%-(_WOYrnY2!GGKsy6)1GgM#0Yg{+N)xwyt%{R}LJx^fKv-Tk_vaN@<3r
zTp~+@J<4ZAk!((^32%aLWY8eMKQSw#yGTsJ#NO*J-uVY|MkS(`t-#Od!|dcl<d<U1
z3V7OVGV%Paq9h|Ks;&-5DkumE2@(2$)h8|5)RrJ=dW#7zE!(@MeF7W)@3eei26r(O
zUl*#+-6%(GvPjIzSCd~J@A1bNyq0W~u83sXs5IYeC9%&w+4L>y2Y=iV#S4{`{Yv=T
zjhXelk?}*{muEF)x%ttb@^AFTvyZS*(WF*02<wsQ%+|pJ3}*n8t^Nb%I%<&{4H><k
z|M+g`T`hS|m(2x?$f?;DmB%MnL7>5Nvp|2}eUE{><{wB(V--NjCmQRZ;)P>iEo`A3
zaep|ISNOBm?!Nd>iyMQ6rVB;DJZKb@$@eZuhB195QIb;mY1I4J4eg#D6=8r3+rnK+
zjVhX|o{hISCl|;kg?+}c0eSnI^iBdB$2+pyEIZtCJ2{PCnf$)F;1ir6B`<DPcEC61
zx-{4$g-Z1D8t*FJC+W?HzK+>nhmDq<d`ShPQd``V^HKTQpi^zhq1!X|O{3lzvSw5;
z_YSvrbd6!3p!V78A~n_(W$9`z{7?OGk8m(Zz4{TT`2kAWrJL&%gP-vC7mEusb8}0L
z)>$nrE#Sh66QAA#<--i?$93Qq&ujIQBwg8QH}@RG0U0Wnu(eUPkFp{dZ<~BE&mx+7
zt)MUDp(!wiz~WYAmGI2Y=|KhWi_RZ{j^@$1Wm_>K>-mhX_g=S*0S?h1f<MM6NO^vS
z*JEr1xLP^i8U&M!4J*yNDYQN=hn6^7j`@O!jJp%60uZ}F3xSeQtl&`G>d>&bL~x^S
z*wQ@8do4UI@|B#Xorep=&uIwwue*6-QG|SdO-!;YD~pQC%F6QcpzU1HI^kt86`5zG
z*W2HYe1~{sTnL|I8#v~b3!x1X@?P?}meufp<D7U5hDtT*g^xS79eIh_tC*aI($^&2
z?tS0&5`#dSURcSWc&X_;_S3tdyvN(z6`64P)oT@=L0iwXGwO}Qh<I32Wwwb*$l7nn
zxT4$2vjCsxuXEl*_%Bj3xqiHi>S~!aoUS`f%_%7*$JGW@RN!k;A}YgBj_lZXA16Ja
zH_qO=TKS=KXYF9~cf_qD>%4MT49>1+ERf+nrvoR@Hhvd0Y)DE*l$oycd<X56QCZgU
zor8<&79^d7B#s^Udh!7BtATTpQQ>_f=`Vzm63;GTLW22I#b?>o@@X9!Hvd0_TYG5x
zp`E%jHQKyc&@(Jkk1XZ-6uj}(`XKP((W737*4RD)8%vYKz{$mx>CEX2I#Y9ZcjxBj
z)>sV#@eLU!X?tI8OtWyZvvvC4)C*Jy7Eukzm-wZ#>h^d@%1R^$?2)J<KQoi|ovt?V
zyJ&yPcZk_-^>2yv@o1~8FT~^f{^-`C#b@n4a#nmSVeVc;XT{O(*R!j=mk2C0poQzg
z{$UEFy4a-8b~Y3<hbHAc-M@L%y<DRluePigeuLWb85!m{z-GcBO3MUZAcG6@s89?9
zEE&nZPL19iL=X@!R<^d1HNkD2?2}xBo%Ee;#N6LeSK=<R8~ZhuQZ7yd&1{c{ZKH&(
zb*lMVWFw>WMo)15=|ECmFaU2Xpoh57vM(!c*&(da^zhh^b8;$*O13g5wbW4{_(zu(
zt#<FZ@_&o%9?-tqek`iCBtBmG0y2<?=FHs8b73Cqw4pZ;$I}=&f*ccBQ%ab6<QnI4
zoe<caGuyS=7{_c`Pc#NS$!FFT3}fORcKn>2G&GkR?cUxH05b8f0f5RFCmxz@!bdOr
zb6~x)s(NbH7a|xZ+sZIIZ7R_q*v+e!zu+~7e7IkrNdiHgj&H1SQf9q|jsCQGx%;Eo
zpE^&}5{sW{rBXL6Iurw<0sP1<r@Q-}@N;MK)rchmTAj`2HU)=w7-sE^3~ecRXJ-Gt
zjgE?Y;b7r^IlI_JqyUQNyA3A)J1DBMnrBiABej=$yOxeS+&Ng*)?q$78e|Rh{Sp%s
z5#jA7LoHbji|JHGJ4wDX!nj6#o<*81*3u+<toPrzxo>jYm^tdw{^q4T!cUiWzvvfh
zT$FiNpY7i>(t_hJF2EBgwD*YWK#uFAwq161i#wUsFDk+l0))u^mXV(9`4kf=z3XIQ
zJ7@^=6o&nmLPU6GJ)QCAQufG~I>4{LFAD~?(Raf1(0fUxU|u={dbd^z&p>`Jjx8#4
z^K~pdJpA_dmX&Zu@a_J~7kz!g+6`X<D8eU>n_V!`KOtb5afgKewiDKQ2bnb?AGTpO
z&Qdj@xy2fi4A5R2VPopZhI;okPEuyD40m%?)Ve#pj?H~tI}0yQZ}Z)m30<7O8$Vxv
z)72$39uIyIxNZD{-4>44dFZ&lKjXSOOC3qXmHm}*<PKgqt?bh$hUPbFhG!m`Nyah}
zJ`EOYl{Wuw+_Iwf>cZT%HstKOy3g+y*VlGTsxU!f(*WB;<HZ&UQp2HG%hneY2b%2l
ziJo6N=9O)n{OAlX7b5ZR3=liAt!rVUbd{!oMa1q0n>k8E$%r73g|(%XS#eoOnYD$L
z>FJUGs9cpSG^V=hu<nnQVcR+uU~qkUW2NV;xLZ(qUMu;jejfD>$znHpfywa_Sz_dS
zvB?o(I-uUSRQ45slADb3HQ0Q{UjaW>!Tl@SMErg@=h^Y@K^u-NiqHZxlw?1a2H#@e
zz8NqnQB!sG4+u0n&Q1a=+1HhDZ6M95Q&d&hwzqO7=%0B*Tmz=DC}>gP`Y)`-8`sxA
zG@JcU`%Eyvf=o)><<R6^dGi&S7JWT0x8Xa0_*VLXy0=O>A8xS@>dyzW{u`l&oiwqa
zxCR9*?e!gCb=94qj+lF>=65hthoE<9xA?>VR4HQdq7-!5t&KjdnLdIT#`-6~sB2Ww
z@XN3}Ag52`=L`S12=7;;4X*<18{V#Z&o5!xJRQ4c&c6v_8HQ^80(=rNLp*yoY4?E}
z9iMQm-}vfLKHVvWnbjOjPU%5ES+)Bh?R%;}fJ^Ic^$3Ge=dI+#Yl^9xYy2DG755yb
z!xNcQ52QWr{JDtpO8EZvAqKbgIX2e%V2$_>xbaT?8E@Efbpn_0V~~kuY2~mx@$rQd
zsEN-d5X*1U21t<J)`_eb^W}aZsRbObkV0f8zL_5p3*(ycdA#0|uyzKGF&cQUQw{73
zY`KYf9qbAm<iJ0DVQ2H6;olmdBw!J?G=yc|07QF^J7K<+Rpo-NZn`d**c!|Kc|s5g
zK+}rrY!e{aRQuhpaiTjP9iNz(o0gZA($J5M4hI8!hX*#-(SdDsO-%d`sDQGGEF#Xv
z)x}&K)1`RHQA(EM9jaGsd;}9(nIifNb$S7Nq@#Bf#3ekcA3R#6+*01ZmYQ-HL_YlD
zL@u?o6LXMKv^POfSq<5(7OCD|f;<b4>&E&+cU*q?PU?O4v~769F$p@FuUJ7#AiNv*
zYAS&l#z@$;{SR#_>OK)+NT3KSh}Ag^3F*?N#?PI2&AE(2{t3k*XPK<cJ_|BA4mtc+
zvh|rry>wC{^YL3y<u-4;AxyV(J=@}98w1_M<ham@7R6(9D@XU|ihnI%<?>)Pl7Oty
zo-9}dZAiEaS4JMPyvMW)2D}x14}ndh*44Dr(vdKX3B@da_h*W)d%$};_r_v+c{gbZ
z1qlZ{_gtgue#nS5gbPUGY+;I{zWL1c;_4<n6i2R5dV;=VFynF(LQTx;wJQVV)-Weg
zN?zMZT~c3jY_}?U%bT0y-%nM+0zOhux1_8{%Sv9|Cd3?bv<jEY+|2ZwjH-vIgs{2F
z2a}xkEA3VCi?yv2A;4}#R~@k&O$uyk5HJ4I=CQU!-CuZGQcK2Ts@pwv48po#s_t-M
zxwO5e3+Eqwof)iTK|2@21v`GDSvLQ`_cbWS?PU7%BD9_E2EYK_p!nmQw=S9}G>OCG
z;-TQHfAXzR=?M+pA@+-)K<a}(3-!99wjDA&9?l*tuu0tl06zyA)y_7rj;!;cROG#}
z3=s$OQf26`iC$4LVd>@w<kyv}`#EdUoWK6>S1%D2jIOX^MPQCTyE7_j0oYK<>4W~A
zDW(&eN()-6;!er%NsSRRx0mG6ee;(uzqPTYUL-AN$o07{crq*`rQB#+@pIhl`uX*T
z8CBI&gUZDt!1drg2o90g)0-e&(DfPo(AxU61ZhZwY0z`A*}m_!-&Qd%F`F0oQ~gBJ
zMmZ=XFx-I-ky&{ehS60_pSnO_cYhB{Fqz-}?!LDv{PlVz5wu+bYXSA)YUgKaPCu?!
z|9d7@zsK*wMN3#5zMH_CS%C|YER$%gZtnA5exJYu0Jr(<NGEWIaGbI%@^F-C(}qu&
zrlu+nKOXVn?b7u!GbS!8<tQO@!ZH=_)ar20x{{};5G_i}i8@9~*>d$UYKE5d#%xm^
zMgu1C9G>7N`ZMV)Zt(L(9ZPoS65(=tw5#=~&9-~o@8xtZ)`#5}jfHoeF-8Bw+}eN?
zX|yE=M9a9f@94J@b5wubA7!-c5#95YRbj4nzQ($QPu|VxifNfznQ8uto^-6D%0}Bs
z#DrY%chRT^5YzRL(}H&d?e{v|p>T-PTEq?#fz9%#Nx){G+vOizTICf+cgLcjIZouH
z<fj|k><Zc{)W($<Pk5WtsPNa~mtAsF4Miv*u{CH4?)(<nR}IF#iW?SJ2OQq6QPd~*
zeY4`^UP)(~nO$pDREElD1uOoa<qECn(W9QXFyf+r+>Dk&u73ft=i^nhH~GA4KM67O
zM|}nrhf6uzOPgxD&CFiDkUp|dN1m9`TqMT%k5+C?VOeifi3Y<XO{TG9P1}Xi5%B;<
z_WOw+&~q2{HKh$5)ddxKt>BGkWhbDkYlDuA%rV!&z&IQLRmVrHGP$Zs$n7AYB_SjE
z8Qa{iMJtVC3!Dv<+Opzyld0BH52BxFc3byE$47HZX1kNFkZVTJ#mnw2s;&Qzn$5~c
z&c$%8=iNA{B&HE&5C}hqy<gD31vK%{CudWxG*MZ#VuwJ}vP0ry6JgO&AD%XX;P~Az
zlhRXTGE`L6<+W5cr>2(X*4AcL<WS(uXJWHNct2g=v^gE8-yy~c|NJu2`nS*ju_;C~
zG4jjz<!$5ncgm?%E39c8Gj=ZtN_t^?0x@H>tgfkI|36IxP=66I>Y(e`=7A0du%PG?
zbIW?Y&6`E=4_EIJo-N*aM?F4Sc|Nu2TsfKAc<Xbb;s)^OpOCS;Z_d$n;cMvAl#h$w
z3>zTQPg!8Nk1bA>7ITeUq6^h#-K71~`CzZQw`JwT-FY||>B{b0*C0@ZsCe_TmirCX
zt7in)(qM^<v|dB&kVY=m31~kiz<Yi)?FmC>vQ|)3jL0r6Ev+McF;qB<6YF-ic5l+;
zG>?F(Y$QR+j7oz>%!G^5P||@`9tieNO~D!>Ae`d2uv?os_>Nb4i|B$*OqBKd_0CXT
z)JWM0G}@QUv{w|Fe$P@g79sGKTZa`K<2(OT)a{GSxPimLRkhm-Ca@&jnGk9VpOsvm
zyW-_ZZqFp*@pixdqBo7mSaXeZ4S9Xt1^a=gdC7n~xlHQX)-8ylMB##n&cls_#oIyy
z$(H`EH{I}oLr0}I_nX@#py2c8g!y_0WK@E*jE11JNi_1TUxR%EykM1<82Vv#Qk|GP
zH|n#syAu??@^AA*IG}i{hXe7|QrQ#I<p|>>WN6AwTY-4wF<CvG3%F}}XrC%JE@4zu
zRMa6A4+-<IS)9G|C8F7I*vj_tr+KUZXYaDRckRng%DRqNtRU<xe_McW8Ncq#*l{Ps
z&!ZBv3p+|c`+@k8JwFpaJwgspe&RZ{JCZ>sK*EBix0tr11O~o`Y^~ZXH+pm35ioCj
z5&%@48m~Lvo~d*nmbsbEz6l`XH6f#1+BxzVcs)Wa`TA+(bMql({zaOWEHjIM`Cu$>
z?V<X51R*uq*E7-G53ZL1C)MHKAr`G&Hm>j6cLBzuhA_W^X-Gr0=;3bJpIm2?IidK`
zH#(Qr14pZm03vpI@6laOWb)1r?s<oB8E~%j#tOkV2#ti#<+AJHS!8uLsP18?Jl`(a
zkxBFNakccB%0@pSnIDQ0De6(ikrzt}z9=lIvZaOWRi~P5RwV)ZuZH1yYA4z$J4PIs
z7mtnM%C>H>AY;{m0H$WPwYO^#$>5)Fl8;Y*3ZHHO{>3niKS1{Ki2jSQAOQD##qaR(
z8idU*us0YJO>k?~F>}l<QBL&w4`PV`s7D2|lYIl!)$UpE)w8G*3=<O*Dj1qv=6Vit
zBHI2)b;S$6lMY_IuOPJHQ@?14H^fduZS!qq82<GbJ9|V#l424prhpyjQ&DL>xaegh
zV8?<6w5X}ln}aJVP<&a8CkXlN1RSYt`ur+}>!;)1Wwu~Sa(*x}BT_<hM+M(9U{}1s
zJ2>2-pc3)r=KZ6g@9Y2FKC;3Fkaz`#p!V?C9r)N~`5<~Td-JhA9Ow#-urH;)l6-w<
z`DT}G+7OT)7?S~C-bZ)7TvlFM*nyVXV_Q7~SoU^ecXc9s1YXzuU8TWaYJ$xXJ$-*$
zflya3Z8hwF<We>Hpysz{>hBM(vR;FkLsE0TL<kL)5|@-59U2<xrl$EhQej@I`hoFE
zVEz+=C%`2j`j$9>UyziZwrhOimy}O^*K5{SV)LI@NVmiOJrM$t>0x1MK}ktbB=OA1
z0PE)|0t>c!^LM`2Tb6^0*v|=O``CSu#^C$53uhrsl=8=jS16I@MtOzV>DgHprXN3;
znRhpcu3T#NWFgm_<vOu)ya>P$S8^mI<sH<C2;WNX0rU%i(dADZfydc)8PPP3zCI>q
zYz2b2L{)|hPv-Y0`>y7*>=<jxen=X3phxEQ<0ysN8v+V}^#8#V@PEDLyRr9O{>xL6
zV^@UTZsB-ZNCSGD1tsOCsDy|#(ikFu_EX)AcEob;Z@xyO6(t!N85I@P3NmlVVxbiE
zLa047>lWpPq@-v^1XM^yU}7XH79TV+fp;QYNCp%Nafk;J2D0b*a_!FR;{r6fj6Zm_
z`U#LUQ+E?JS~ySV@$mhsAFCtu`&V>UkUcyB5(^d>7zurin^Q=-*<v^wsI!o6Nyc-J
z$03ByjrkJ`Y1Ii_0>%ePhQVwLgNZ;9^C9?Nker+x91TH&hCArQ4n}f`K4J+7QB$qd
zy(iW0F9%l>YNMTj=c^gh=&oh1CRf;B6u@66+LE)`g(u9Fmp}1NBa$JAh@UQRuC8tD
z1(THRz#`HI3O$_)cku<B+v=O|FEpYeQ@rG5&Zeh;UzCP>&fmV3nVOhiY<7NtW90B?
z#GF+mOLFF>=ZfoPO^Pz=Lcdj`vy#&6;koIsc)|RItAjiK%7QK6nd5{N1tsZQF<^3p
zmvhB{+3Xveng}9`+f8ghXV2=rZW`9pvb<|=bs8BRmXwkR4MViEHZEKjtwD4=)g<pW
z&{|luq!2PG$~Dwo?xrWD{nlt+nR)=(E-YF|{arSzp{Ay$fzN(mWo6`~u3*1JaWIOA
zZEn4?Byft6k^Ecany2P!%5lk12*6SC&3z*)d-MCXDY>9YkqGLmlNBU}MDx1)eEn@y
zYF0z%mR;}a;Naj(<E2fr%fg9&^){p>J&Ge+Lc9BC!+sb}HIv~GFEz!nyv^n;TZYv&
z-<=<=@ocDH@KV%NSGl#F?wV|G3X6)87ArKgwKF-a*Djm9{#zVym&Fx|0bpL&d_scL
zumpH0s%Dmp!Km<?7&E81&+>YQz3d1`4);yX46%gA%?i$M%PmcuATvPHpvIl#ejwN1
z-O&f*Q(v6hG#OVkQd9AWxc+7r*$ZeG7Uh$|V__pAc#RXfTX@yRB!|aFhD7*@shSnl
z6oN2Vq*iZ|DqWv9-0?5~{HMpOxZ>iYqvB%fD01?i-X744`=(9%p00Vw`z_m_hao8h
z+uCVG^t{43bBe*2<3mSpXH11pU)+qGWcFf{mJ4!A+?VEh4tnZWAA0n`wVjeuq@gNc
zF`R&S&;G5Z{XAv1NG3Hcm2h5um9l$KMyiFl_tk9hPV4cQnza&GZ%gn^49N)7Q5BVc
z+tRBAH+5;<v|th9WJITC#J!v-X={543VLmnt)-`*m5;?gQb$2~@>Y`Zdzd4!<O#P7
z<s@P5{4FF~uku2nd`RvKj86~Ebmivgtnv@fU4iU1z-fq7X@~0BnYq?!ycNFnj7(gF
zJ)ZpP*H+oA*?a`-=~|mxn&y@jl8*F?qavl%WAZeFlYh<A*WKO0O;A`XGXJaRJbiu-
zz7g=RQz>AhuduW{CnCzw0{h|4Ei1qf-H6Jt8pYi#e;Crvm(fTkW$-$PFS|uC7U1T6
z60IxMAm}FVBX=c=9rq&#_mfgar9Y7$ETF3Jcn}Y?4FP3l1gS4Dm#5ZD9BbN&(KxtH
ziw_NpC>)DrN$D~KB!3bQj0={k5+&6wM1<QS5(E5n{<tNz{0k}Yt1bG-OFm2qP#WjW
zv%j&%!NJ{RXm{_tb+X+aUppJSl`}Rl6ckk$%gg3&{Ku`sY3_M_B#7-Q&p;I_O|U3*
znS8N2Ko8`{*M$!qO3VFL7nD%C$?musW<gc?w|NiXWuz-EEU|PvexSd?Z(*S#?ea=v
zyMi+VWN5xnq%TPpW4H5r0S$`azD`UmOU;KHv%v=xUH)?dc1uB8d^2@8c8ZcX5ljLt
zgS|r#2<tO(TC9@l@)8$kvv+i4)w|bI@AZqb6o7DdV{h<weuoX4GHfeLy#DaeeC*Ue
zy)q^zB_XN%A4>=yaeCnHk&ydeIC)OC!Tx9TxaE2kEsepcq1*g~H&|md>ks&C=Hcaj
zaN;n|`r`)nA`6Eae#KnMUU4^5ls`mZ39aMMb^hHyXMlm(La-ARU~$)kl|t^c9xBoT
zB=w@6z25SSiq06dZ?1ty&UX<fJrOO9iy_uXXK|j_d0AA)#)U*&qavz)5%7AoZP#;m
zWLZ%ZUI{FTH`aPDaa3rWBXmNl_FMj{`S~gE{B`Hki2AcwBoV*kLHRBvZj^?zuMl|_
zRol?Fe`Y7dnQG5yBQJ-2of-JYrX}A1&xFL(Yo06oZ2rfFGQa+qH+Kq=M^4rMP*(5$
zjh5dv$%F+5p(b9fu)!2rD(dpHj4qJ}(jWJ(&t*F!&kLaLE495?_%`xB=z~qj1LFa>
z7WEU_Fdf4JO6op7L^Q-}r&b%d_anWtRdBPllm8}!_<^2+SfiGJq{Qgd*a{dBOKIFl
z(D()g;tvgXO-xTrP*Go><6`1_0P*Wwz8|lzJ?g61Te=Fqm!+SLIrx7h2YG>Hl&M;1
zTl4*!)l7#ucCB%dk|t-WuR=%8r>EF{@Eh*h4A1Z0uWyOT?qnA2|1UV>h!!q0o>xcS
z2EdH<lDw_$4gD|WB?3r5_PmAVKPJn4q^ZP^v`5l@;a;ZVK_P^Ik^R+{knS6x<8~p4
zMOSpH&F2hF!V?-EDfQbmB}GF{Q#&C2$*1}_;6dm5{rIitc1c%O!$5lebmY2bW+%6&
z?(xz)P324m&x~#~#I9BjW7qoQ`p}Vp!*8lY&9&e8d{%gd>lR2w@`2~$?##!`X)m)j
zu`sj35Pk`YB7wZ|es)Hxjmz`%)UuSmW2+pM`gCLW12b#w!LifC{uLdw6#<up5@SQ#
z3E_&=WyUg|m2gKAK5|*R{SB-w{`Qy&t1DOeuVs6_9%Y>L{_6Otk)&hqZT5Qg>G*2<
z-RgJ1telQYERgWXiO>v+smusXpF(Y^;NP_bzsoyu1u+>3;mL0eQoB=#FJ;>Q^EmEe
z%-n>IXfLwYIe^_Y5i-9g>AwpuK%Pgt`akbKNFmwJV&w_H@oJq{jH5WtSB~lw3<Scu
z{Onyagl(LS-MTPFMIm-5dfH=)u+Vz#iHpf{^|A%HFr5^oG{jV7Hz6>e=g!g^FZjKN
zUy9H!Hab!=GVut&i336bv)7G{!t1s8qad$EM%6UJKZ!erUq4{2xfeSvNWYT76RAUT
zISPl}fSfUAvN~Zp9c<RPTVCLT2B`{-4mQ3sv=7Ec##gCcZ;}tct)1QOJsw?d3FyeG
zh-m=Qwnmtb(>hf@_^QNwgJNzsA75nPS7w=9V%p&nRo=cTc1u`WU7QdQ5)ou2r>AD7
zWTb&z32u$ASBs1$AGgQ5L58YpW8>rF6BA=&zh%+tl(9({-*_jb7DQLL1D3yCHaa6o
zs($(cmw2x4?huvrfDrDRzsTl%iwe$6lg9&-zB@#7pivUi{my=|<gLztg7vZ6xxGGB
zSp$)cuq83bOx%h{Gg^e2=-qP_Q0-O$tMue1sx1A55JTf|%2WRXiLDqRMAa+hMOMe7
zfk!&YaXI6#8?oc&P0u%BZ1Cd9@$~YpUw@qn9J--!uu_vY;C|>z9$UPlM>yVPK?0aO
z;j014vB}&)!`99o#*dej?)%i-u|Dc<j*RcO522PtT}{{|I{(K^AhyHY%bwPtFjH%H
z9VUJ*tdABtT5?W84BghLp&f$mT!*{yN_+|L^GdAz{vB!8lU_+AMMD1JA2;EmJf<Nz
zbT1;o!ke!c3s>^MxRC;_DBwg)hR7wI$1>rCF${}Rks(eCKZC#a@JilbC6<>Z=KTvL
z)g?J?OXbqtTkO3z=xI9hU<8U{5XpwyytG}>+Z1WKFyb_3>MGcMsZF)ft5=zp3f!9c
z)c!@3oxv-HF*Fr@qv#M^ocRE@@9vp5FrjtqA&3HDSHt+1*4o-+fS0XVRc%fU>5Xes
zR#-)KqAvuvYm!}IN7`j>K!|`t{A_M4-8Yt7sV-ttvzZ7pOKsy*hfQ29k`xIO#J4q#
zsccw|Cr->aaE4_F;eebIiSK=g8oeF8i_DdUdFnr}M$GNdVdy*=+c%jJ%fA%5gopqt
z@AGqqEei%LR#6pjMl9`T;55OPgUFR|j&M=Sl%_`J<$64Rwznh)a0o0~PjPepV2<#@
zW%}=ueQsV>`F$)Ba5;_LCu75}ix6o$PMVLzN9y8E=giUF!Ax8)votYszPo=*p6gN1
z_rTA9SR_hq>J$rYXhgbUQ53%p&Hv}gyS^1rm{$vS_W^(|`a#>JS>N@$t*u*rWBtEe
zCOY=V0atq>5j`WfFQ0yZ$xoL@Q+{4V$<?{5WEB4&P2a#CSNFZ$IBmnGNt(vC8#`%|
zHnwfsI<c+Bwr$(CZB6X_=lQ+Y`vK-Ud-lxPd#!b=Fh}3wc<TW3=RaAEO(?IPuhs%|
z531F?T#Rdr^D9fAWIwQ?t#=6Q*yy%ve5A$g(S&#`5fw>Ir+x5{$ZzjMKikzZ%f+FV
zYo8}oe?sH}*i}SZh5jg^hrGSAWWr5TR#K_MZaFqq-QulXnfP6dpUpiE>3~_^?&jk1
z`ktDSj+UB2^7iKLaCTf>LLXM7Sdi7ujSHQ0e?2`G@5ZJ*nVMqKz#9TrXIh_Nf&H~J
zJTxQud|q7ub#f-mY$YgK!U=TV`hd?%{J#FWXJZYx>pQ9*W4a(>r!a@%SB()D8W?#L
zRo3Qr5RhJ9u(;3qwHp3*ZDf{<-IDzH_KRA#H!@jTO?7eXK6J6Ys%fq*binEn8@AAk
zR6jHQC-%vym32jBc|{2gJuTfBJq;Z#psA_88GLL_;|ZQ4A@yynPRsrK_rvW4GIlfM
z%034m<|xHwM*4uQ_QRw^V$An5y07rkt)>K56Xl+}))=;`Uop1st*nb^Muvxn-*LZC
zo2k167H(t>qY!vKou?*;Om}Zol&ty7Tv8i{|9Th-xvrUxwNwM|?&AY^!hjkRZI!!5
zSK`t&hw;?cPXbQ*WpzT51e0K675!bhxB$5#!nOyehF=QGpC;^UnP=r^neDaEzZ35H
zwOG68IiXq}dE&Vn8yZPurM@xJa@a!9DRF+`He2~{Ul$?j;=-kmompww2=R!~R8ier
zVB@^G$IW`{9h^K#?5DiC)ksQ+O-p-yZ5$e+uQb2J&(t1ZiZbxE#yl`08x&O=WdiiT
z90ir1W|s-*;7GuwwiQNwA5$f{X`9g^d=C40wlh}|EmcYThbfYaDFCvsw+!iN!132&
zI~}b{LVC>oj;r45>|@{F+-=#~YfdN_KkkSrX3M%467P)4N`=$2rW5=8&|zWPq^u?A
z9T(zuIspD}1Z6H-iv*qeC)`SF0-y*9+EObEsDJm6YTaJ1Z91AsN9N^pBW`$Th$wd)
z|Iuv0vuz{W56Zgy8z<2CE5DHJ$?m7T0<OG#lKw$4;ky5pd^|V91b;TeVGFv?-OCwS
zIhgSB7rY(nrHFR$LN&O_eM_gUNAIs>(kdX~asP>qh4Jm3os)H&qX+O5aFw)Q-0H)*
z2)f31k@a43H@DmI<(^13qZ|#psh-`Qw0x+{WJ9rB7m*z7Bdea=OIys1pOMOo7WlT(
zc-c8q?83aVE|VCCgLaM5?}rZViZK=&kfrE%Iq{l49vz$|!5o>)()8x;j`SWblMrPt
zBbu#jxVW1fDp3m_Jb(nF!QtLFD5-HY(~;z#052aG569fBnzm|u>OD>sxNwu{O83^e
zN<;mZ6sssB8?+Wv+vJ8fJ0qf(X+kX-an)*OSQKE~xZc^tO$nY{a}}<=?_zoo%~H9;
ziVjRizK#SBkKOL|EgfCSfT0Abky%j)a2MJs+LgOB*EVth@aDn#t4;?cbgoNSYj)rK
z0Um0}kxehY?Pv93?*Pk90pYQUArV3%qqnc+Rq->C8$L0%GV$-vx>#CQ%tyQ8#_IO^
zil$ylW;?S((O_MkS0fM4n#P)yi|giRc(+FiPE@4bI|6_`4_ZU*z3DohaalG7Zyy+6
z_yZSi$RH2wpn_8WG)yzp*j}H{@HjqMDHu88KY#a&&%Y`n&Zm~^b8xQnWgF}nJ7mm;
zXspoW-?~+Adg2Zi`wugPNSHcH&EMbrmI-<BEH7H;6lS}F3~Ewc(QLu^=Ye0S*maHj
z-3Wh!c0!VPUY*{>yu5;f(j(!HiIR@$Ou)OcFbHTpofpTP$gJmbfA29oG+(sdxJGup
zJ~N~f^dy&?ks7D<!g%>Eaaobc%rdn^+db;bm&rlmA}8jigv7hu9=JgFkZ~}F6zM&I
zXlJuYoJ@JNa7%!$>L_gcA26t~_r{e`fq5pS%AfX%U|i`2BLXH!&-tBx8%Ejil=AT0
z2ExdF*l=j0%&2$JLB9>=!Ywo)^{9f%G$C`COSg*CA*?II{xB_mjce{vPA;C@)h812
zgYsyvXw|UeHn!HcG_~zHOIi+*FPvP;aI~?zYkNV-epfVg)OVk}+FP43?sHZ&99(~J
z;B-}SkNU-j&GL@Y>GhQZarVD^-;tOLc-S&_Mcjh;)?N`4YS4twab1rt+<ZI?lGS~<
zbZge+HTLXT;{W)Xw5)M^g&ZbLUr{FX<vC>1FY~w*tQ!7u4Q(ASwxe-N0zO4CPD2{I
zL4vB1akYznyLRR;K{}^cA)a1(RoG)_hMgWrUPlie?c5rb8a7q+H4KC0;YUXR77nML
z19&=%C5p-K2@YqcltV+1rme^mnO&c6o)omw=Cv%#C88dZsx_?FAAo?JBqg9Lv^=n3
zCMm8mb<neOGeY`heGo0QxuGC08=wA6Lv@=G6;vDdcDySB3wM9MegkulU(XFr^81#0
zUy}FpvzHsleg`{Ldtj~8@pufV_H!p~0tG!HlEjB|J^nUCca-LI4ItZRLx|MNTn8sU
zr)Xgskq(LkLO;8<XmVJaJDxNZPZPfRt)#rnDJDEnop02un^uOFv;-#J6EdIP&}F<6
z{_$0RY-x73z*}wmt#?@r6H*>l-rUgw$xGTm;1YP=Ml{pvD6Q3mn;{3xa7K2tb*ieQ
z!LFy2#I}X^6`=t1%H*kNWI+z~^Y;{g!a}@!G=Bz3acbPS!msenmwTdprS6-&XG_fE
z@w9mfEiaC&Az_2sK<SK$4-61jacf-BJuU4(rR*Ok-aRVTR!KWDFQ)qAj)8S{=z9LG
zZO;K&r$KKzK}Hsp5LFgnwEatorOL7Y7Bps=R<Vv0?vkhl@0NjYYsfHU9UIek|3B3e
zh_d=^F|d42A`SA}myy82%4Z%g)dfZtX67&yzK?_PJssoOQ*a^`bQQs_iD5})c1{ri
zEnVN>@Wed~^!_}eRGK5`GnnT<SAO|3{R0t{e_Tz-I{}te15|R{F2A|4aJe#@d`G-y
zYpo!rO${ye<Cf(4R#esF@$e6T-sS0oyy}jI5h)HVN?Tb%2wp2XsYu+_#@gEIn(A8X
z&F=2uep;$ZyRD<Yt)qDF+mlS<*e?Z2KM86*5x1C~XOF?G*0$#{ANk1<TB6*h$)Woc
zKtjY2^Ba2?@cR0yVr?#II}z3t?4F^lgGfg;G}@dV5R({-1F>w~Jw~5f0<3NHKd`N`
zQh=Z&4Y5GkUKPesr=?;yU?n{Go?ohq@DchW@>hPdXZXfTL&=bQ3zz3!h!%bMv##?x
zZuxi=j>0M}DRjd|a`KNO??)I?n10qhWPYg*OKcbbhLjZjE6y~x)AOt2M=*F!p@}xS
zv~WIRX{=vTi<xQFnjmlOG(k*3SJ;9n!;9REe{4npB>cq0bGgZ>XJoGgqp}9-jTUOi
z!gO9i<>BB=)aQfyiVi~C<NSy>=1jCrjf^Lnvx?N-zfde#+(cf?FwBifzA2G0<BFN0
ze=7kqx9Ci+&}QB|6xXxM3!9xQh`djShPlgPZ2>0C8o|%|#_TMiFxNozRMpp^vuR$B
z9*~HPgNW<t(a6%o8o#|N884oj%{;Tc4<U4~JSI23w?6U|I5lBIOF1$1Cs+_;zdk(j
zTU1rVJHib67+(qvW!5&7(hR@e;kxO8Js-%bFn$)_S;v+7Su%XJEq#?bbnFZ-cCNI@
z(&!9wy`?G5sYeVON$b1fvO1(iBK+3|{ec7<N}F2kZgcb<Jt?;b@0nJChI>b6bYkM}
zV)q{@E%8wo28kESfY;#oHnbP68%b$lmhmBKtilwxnd9{7`|PhKNKFk51qJ`Tx@|z9
z15X);Cc<&93yEOM^Jd^qo7C&;xD%l(3y=KXAX-Ya`Z`NJY%ts~E?ewcF*Z)4^Su{J
zoSHWgA0HpfL3Y$BM_8}g$d_0_4Vt>W9~;1vYk)J(&F^>#4I+a6+Lj%(47s^quXwX3
zaD9#O#-#q6j(ot33~r`gkR*c4U>HDs&eqW4va4k<bCZ|X*VpIoZ!9f%;o7RrYwU=l
z;k=_WkGt(T8wmcCrA>HnXo9rZ-_xm(8#dnzFvKcR0kI<>+>4Zl{0m{sl5X<~_>PGm
zB*x2)*zUetL+hd#8;7!Iq@T6c@|or>hk7z|a;O(*#~Mb5&PRVB445vT0`hzZwjYw5
zIsbSFJL&m<IEsb|oz!?pxS(-aEcZ0z-)=yOZo`lGy*&8&`T0UZd0IDVRMr0CDg(!p
zONAqc%UB4Yk9CN=8hs(7OIlxPT5}C28A3Dnb{WgY{9R$n69<OrG9Vxz-rt=sH=6Aa
z2NLWP$=rU1$`QS4ScK2*0evz%TVFRIyq-Tl8F+L5Rb3t19(eO_dnR-!{<d~U0g8FB
zVo&F6lTO5s3Wlq+|2}s+b_OXYmJ(=WoL14SY47N0Yun*AKRq$?T9naHl2gWRWb#GB
z8K^psfbP`fxBHhe@p*5K9J4XJ5@IJtNLs45Yowc+tWkM%D`|gu6!=H4*#e(IsDoJ8
ztgMAI@<RYr;ke`(O+T2&7CTepD;<~;16D$6Sgmgz_BARUknkY2xprR%2jcLeBEvBb
zQJNlgg_iwCm0nm-DwI~w;54+Ixed-}NQrIBgAdM#47C_-Y<RWZ6eoT;?fwiSsSn<p
zCLuxL(V?DzJ@yd*jQ$KprKYAPr{FO&vT|p2h~lkXOX!Hjf(haDCR88u+m8|I#oPYZ
zctX^eABD>-P&)oF?Wc!<YSi|be4l+&(d{6|@;*G`flq5W6tvX;qUWU0Ej>MXG&OMv
zb{eEz8_WC|@X4vaag#1xek2B>RX`OvuW%6DEA}4!B)|i>oJ9JEBBuqWz|`-;ih`1o
z8g(f-|ERbKA8;6GJhTOrPFzdn+ryR}3~tz{mu&L<+YuyB;NxOtY_6}hW&>{tWt4Q^
z?k|WgUKq?Np{06x0Rb&3B@q#gq^L05*SSVL-iu8&IEW~ovH#HdU*q(TB8H0*We_{I
z&j|)H7l0_gf}9-WFa3Z3XxIM;*(MZ!s{3D*D_v-^Y`&0p7KJx7$35)cmUNJw&HWVg
zSUpvyoPMy@=Frc}_Xa<>%8J<~NtD-xdrhN?dUyhXc(duERSD=?WyNKVr99_xTc9S&
zHF7~j0R!b3+s~5fe;Q?H8Ea}7m>HR-W@kPDPZ-~bEJ{y4p6PuiTZi%q3#@jw(BErn
zFinhT==FmuN8C|wx?V~msvCR>c)vw0r|S#<Fg6x?ha8j0l7%zi=se|nf1=+oj(FW6
z{8Za%;Vty94&hcsKyR#NQTb~U{U<j<i#5U6YdC4GYt7~pKAnnZjkdNE+b+w|9Hzco
z;MV)$67xbeY|{n)&+D_KOm{piv5+K^%SW!#{J*7O&2CPQgO&EBw3W%Y)@}@v(kO5b
zh6k6%{wbQE5C(vpO4@AGU3Z3!GP(;r8Mw!LKwA`1ih?A>t5vB+?C6M&iLoN}X{&sZ
z8o!g=b2tO17g4>8Pi}fdCv7s^BDER-U-c)&;0JqQk&p=b1MCRq6pmidRBm40<)aor
zt@GWr@44~yvV-~{w1zqXcF{jUu{Wu#Rw@s7|E*4<yD!9a`8lxgJ4;j<jWRevcg#5I
zqA8tAH!Dy%>KFUg3Fp<8*q2vWhQjw&ze5apnR&(r$MzSJ@-k0ahYK5im`)(L;nRrE
zc-H|w8d+amDdIS81)NCXzxqg8eP=V?@q4y}g28~trma<)o-a1B+I<9WKlp;*;d~GO
zZPb1Q2h5JYRI65UEeclR!Htbv6^z60gO|4!d_D>k#zWluoQx@6Mn9%4;_gzP{(gMH
zi-M}i3d-)?0)9uJeb!PDSO>b+Ci>}VK#GzNYA)%$5PvX*M3G30{I4C0oCcl4oOLqu
zaCso{$PX7Cq<0#~yW4bS%^a!@h0Wuj+)1uXA&p^FRHbRWB=SO&BE>b1y{qR*C$ow$
zh40$Uw|Tr^RFE?=l>=$K_?!LxeMH5CC&85j8wqQBUbeHFp)!uOC@}iL6wPoYR{ArI
z{m|=B*2743c9G2HD(R6APR<Czbv#$;Uz;x;-(;}~R#Nxp9-ScCN-h<gg%#RozTezR
z3VO`6g?(C@_U?;Zgwpai+he#^&PW;0Mf}`2?63A+o;aq>B#L0KP98{NG^S89uE0Zv
zmLKqBttECD0c#z4hCog!6?qeG)WOLINi>lSTA6oa*x3d!c2MGMA|WQF3YApTkIFZ%
znVyPrVn6?!geOz`p7|M18Ls-vC9jSdaBkICTlYFrkTLzF%?Rm&pE$TW%Ir-L+?-vI
zL&9tM<wJEV4+3VNC6eIZy62e2+hY$H!dM;*iTpz1;s8i|11j$a-*KiAF~MrQsId;>
z1?4s&w3+LX(-k-}44G->taN6u!8cLQ<kQ$TSTo7rg$E>(eIP`?!ttInRM}k-;6KJm
zu|ei)K?a;wWEb~Kvr}e)7Gg2CdH>Pgu_PGyv;Sy(vxbgxUH`;PlLH<Ab8WzTrO3)?
zt*$ULwLek=Qk=#`gnY-@|7Jt{%vwqQ<A(1}dc9_TI<m7S7%1By#DBEEc#jzl*k>94
zAMKp|xDq*65+>w<XCkGDdL;O&Y9@9kimETONBXFS7Mnb?EwTEX<zw)_EGA1HAvAmO
z&YU*-c1@h3@sO|Jv?RGzTH3L3!5cnQE|2OppeZ2~wPXB@t=jP(QG6`mWJ7<}dLvs6
z*b--x%fvptv?y+r7fT9ChluKndPnHWhI=W{@fTX&n(JT94smrnD!58@rjGsFu81Y;
zrA%|@^f&Cv>SHq)*U?}Nfpo(#;U1U6bNEzOLrGZF8O%Iw^7{BZ`f@E}d4OguI_v2I
zkUQU#e?sQ@(FRdjy)t;V4wVV6EAAmsm)NYcl`yE1=QKe~uZg+IiSh88NJeVcv>#B_
zxje{4VH9AV5m>W7*nU(Odw^PoN0#`wzwcA>PZzHtFm#Pe<%hR4+`CEp+ry_mYodIJ
zRtj+6koeEDYhwQhdWjw`Hd^I#+Hxc?vlo5pBpM&!;#4&yA8R#$-(36glLFK?SLPAh
z`f!}O(!3G!hqCB0D5pV&{zTcrYE?VNW3ykc^E+C}MX+dW!3WEftsA={ZDpCE+XOK?
zZ)3wHB2~iO_HLyNlc}zkQ$aZc;ZVPvvQJ!kNMvZZ%kB#2ZIe{L5z(Av+ILOh2lR_C
zx6RYpoGO^$_gEIXuq~X%i5D%GoSOU-zkJed2ke8Y2wkzU-AX!pS^bTRVF9+=Zz(&M
z{VVvUwW@~-l9+5EOxCoWo!a?!Tk6``)>g;B#=^@zwW@kgbe7o5h)Bt5<aT1I9>Q8i
zyJMUF<qH*<@c~PQp~Tum5WWnc@Y?m<=d@~@oQ>wJOjI{cmBEc=(zJtH1#s@X38f3h
zo{?eJ5VA*qNT+)ZlyucME>admTpnVyuh=h0U;>DiSCx82v9UwXO81+Ui{8cQ{dA+a
z%9wNnk`okd%q_Wdx-|9~I>h~q`DKF;7rH_nf<)OFjz-(7AW_cFw^%@KhN<S7A8ji<
z?zG%}+m`-Di|s+uTPUjK7p>lSaxA_pM~dmFovgP{!DYr-7n}!g<X~S<hhJQ(H2)t3
z4LLO_-G4J$i0i~gQ~DeP+=R7t*~UA#(JVk~Q&STI%gpM=Rm+~D+bz#S?O@&hlb1y#
zqP91Pz+k<5oScG+kQRu7E-%X|0Kd8cQ;{X!V3(&o<zG&S(w%;7elJ#D9$C5BH4HDp
z{+Ei>Lg=<mGx%6KeQ@5$Pc4#t1`JMgS`fO7Iz*kK!kYS>ksaG$Y6BO!an$NNdjwxN
zHdYnEnRz)a7-9W^ezff(C#~umXNIl9psnAC`y;&|Pk*_a7@!~vV3cEHcX%uUG0NYd
zu}W2Fy%WlYU`?f1*yz3UATbN2-|0;M9GXuxp@e7f3CXz3G*<KM@^ysz(E?W;CN@Gv
zL0M{!JI7~X;I<QApWaKln9BQrC&Dc$F-l!(d5WE?J&*?uh4DMDCsm!y6}oK6Y-G=)
zBkzKAMq@3@2Ud}E-Z1RcFpzj<=WOS;ijIfn3(GCfCOFx(nqW<(CrgV=EiSQq1&d_X
zak~w)<%O;6ZlUloxtP~^zNOWrF8WVXA(<>ghO@;i3FqyhCp?@l@)%ec=^7r|fKKYM
zy8=MA7Biz|uQewHk?+&T{m3flW1x({nzOvY?nccCAcMQCSv)>2;}P63KZYTaD1fIR
zePGWXu&oR}m#N{U!*OVBY3iV4<?4fAF52wP=qVW43mPcM%7;4XZ>amVpelbwMD@6U
zuFlM(s0yJJdS^bY^ph*KIypWi#1B726Nckj#Hes0t=(q+yx?IyEBZe>rrWigYcm3M
zUT%Q#=3sqr-#N_s0A)jTrSv)$AI?pSS3mFzwp<0G^Y$i`h2M4X2$ks(4yTdsuFX@*
zp=Bo5yH7W=P;Vx$Xvg^-E|iLWh2haU<Vu1yc%sdmvGLI2u<~13Q7Z}Rabi20coKhy
z+<_>>1e=G;LK5%SFJ7;e2~bJvKYF`P;ykeb7e*r3gVEu>+@;CGrI8)^S1DYo*i0s@
zyH*`1m<?FSjw1K8utNB*hsM#1Bw#9xDAStOdggU{;Wu?PR8$E1JgIGf*&c{7N?o9V
z&!_V|=Z6S0iSMdi&6(G(Eah)M-u@@2?V~LqHbnkMFdhy|V;$7N2*g(DI}^5Bq!aL8
z;yhg$d?r8XTu4lJ$4XnC=*GWtq~1m>?#3T)U-*x7U}M1H1}akfTe4qMZ<JA1{Cacr
zA&%*1YwbcAR_>{zsxCK)g1jtTvM@jMt}QMP{057Me&A#CQaS-}Q_ZWAsUZ}}u6^GX
zkbkN!bs5lY3H%<bjSpAuPpBWq<pH3GhY&xw^g{)*{8K)-cOpyItD&we17p5K8Ju*&
zH5tmKoD15Ov<#NfH|c0N#i7oC;fa0B9E?%^uJ4P<+G)7>CzjSF<zrGVdxv^adydyT
zkT@>+(;v9rp4dC(2CGRfea>LUSbFo8-CLrz4ty35Phr;gCb?S#ged88TA;1>InML-
z(aET+Gq<>fhL3}3i*xER7NZ|vJEr$BSJMSc*2>r`4i5g4(kCb@%{MyTMM?RHE=B!t
z-fsrIM+PqTbI|MKxwI}NpOy}X{^#0aNvtK0EZ=FL%5&;*Dz)WTMP403&t&&SjLPh%
z8kf86^Tm=tDY7*g?8+$sj-SmCg<WgXaaAz?2M=zPs&Z82tA(l&zD?EdeBZR6np;?Y
z(ftiR>dZu}1WaA~PfJo(o&yUStZtbfK6ANO`Oj53=5x^R^gvl@HLyT^sqnxlyH&_@
z+3UT*<$uFIN?s25evhTbu5t4frecMDCVSXMQb^q0nQy?+!W0X5d2MUs=2}~pfz5+>
z<DpZ*#$TfD@9iGpuZ=K>;2H6GQ#IKlXp$dY8dyG*6%<ojDD9aUNeLV%I7n=`kDQ=0
zF$Nu+Uz4w$b*m_Qf*nKqcqU5^z5ms;!IQ84^hgpkd{yOeF;oCg;G7?#TryH(-b$+S
za!M+yo6FJM#PtEq!0O59%S|xFK^c<VP@8RVmJBWdEHAQVHm7#7C`^nHlIx$kX8OTM
zNd@E6Sm*5caC?7;jDVykiO8FEk(;IcxsQK4aqCzgHbso7CoSAz&c%q~30Fn*w@822
zKtFwn!_-+>DqtX{`}gV8jO4yy;sWe<qcDrFKMBicUj+o>UY5{G2=m%bT#b!>wx5}m
ziG`w_VnKfibq(n8Vh&10&S>u$;Nj+EW##2%<ltZlU?KIllAMxksjMKtN(ns?g335Y
zSY23K9qZ(ct65~=-1Q(Q$}g&}Z?A1_1W-b7OKl?TjeDQ|M%5`A^ar?T*K;5hm<cx}
zZMDu7umBibb|-&h<R^#ea}W@4GK>~)q2w)5SQt>XJ_WW=cLRM%l@kt%M!#QLS8x>N
z<I#5ybqWYq5zZYJRc61emMP5CHrbp`-}DS@SA#+*DfT+OMIvsBpKmjzb*>;A=)l6W
z-)jO$?HRCki?)pN;Pm=QGz$K(jATFQ-)ggCAV8^aG&F-vW$;K^dnpjAkxXZng?U};
zm3wbMRo{^jf<d=Ysc@_+X9L{aT?kBS%_Fz>x{bUYif8o^Z#KWLfY~`FtOW#wRemwC
z29~U6HZ`n23l33tsNFX_ba|Q~5kR%xk3oy)kws<026%8sYKVV%{y&#Mmw==lP(LrC
zRbV^T?mNNC?*0gT)aK*2Zb>el2n|2-Hg_8M>Au(Z2!>z3>Gv}zbEV-!O;i|dGQw3}
z&4pSejlB|r2p(M#Kn?#DaEsS2gbS)T9{U#}#vK>Q>Rq0TQ(xbiFZk$$w1*k1SkaCT
zE%fW8|0*F2TgmJ!ZqR!P_m2XPYvWaz`<geIn*UASbiJ$!+S!k@TgSKCeb>MgO^SvJ
z)t%xb_Q(GHhaW0hAkWvDHhnXU=UbX~br5PlY7{ed>9{y_C7S3Qx}mV_&SJUj&UCEE
z{Lxv&RD2EH-W>hgg4T0mUuQqcN7wqn6rLBI6WL0=IQNM>Np6v`EdX+I5wjg#+M7d=
z>!Q`UZb2W^<bRz>{g0z_aUVMNpI;9xaw+9V@O4e4=1z5@1C{6*w_vO|7wS&8<_F?V
zOq2(lVB0E|5Gk(|>p=ZtT@PfToZD4Xt}(n!%}?Q)=j7$5N*4>jm(I~sv%JfTY^ry<
zN5-)+9G(QP)TJF**?#@e)rX$M`EA90>HfL*kQBFetLp<5)PxSn-&WskrB$ciY-;%}
z$_(R+*pPW@#!VA`ruuA<!qskEl{HAX;O2e&_6SU-@>H~vd(_Ob&WF_6U-$*$&yQEv
znIy!1fcx8j&V2K=sQ4Ld{jzok4|=FnJ-t+0b!X0WMO^`!VFnC3E_~lOrwY;sH{97<
z_nmLz{^JZ(if1h*rZ;CWIZ2|NT2}qHP8&SfP}l{l2edGh9RaP6!9M!!;KrHN;s2>G
zMcR}?2Jac>qB3kjK&;Zj!aS5(;%?kkOVA+xA$jvJu7%BM70r}?uOvgKOjC`JeYMlX
zeU(75DKiZNtvb@i`m<6+_h(_t*($!5cjREvMd(9LvDk~&j0MF{*39wd=a#T=8kD79
z6uuJ6g{G`>wz|W4!%!|oL0L0X3$j@oi@2$qjb1bF^j^4WfZ0S%$^vr1?ucl}8V$$Z
zqYZ(>%c;oUuCdiuR=8SY;`rd{c$)BosjUE;f?b{jdSWZCWYh9-ENB`Xz0n#!Ws`pm
zArWDhxqTW8-ZH9+@|#Ov$gK+wrxOUT%0&T?V!kWA9^P{^#3Wr_G@RTg0~to^aYp3L
zyke9C;rRcWOoYhsG1mlT83heBE!o+nm3cmH##-;vvu&>aZx-qYeK0Mn*u>Rg*Fep5
zjP*O-wsjkn4Q;lO&U!a-ewn_R``yCZ(UNh<q`cfLQ_YQdY1h*9kH?i>4a}E_N~&=;
zyt}T!w;n8cq4M^c9u0~3)K1+4-_A#@ktX?R#_N%ylS~_%Ry2$!5-Kv*3spTiH7n5R
zz<tNG>e&7V>J47h+P#qEkx)QPB(Nf7s|5yc@n^$((R6;Vsdh~zPcTxGJ8!l^fQu|a
zya3SwQezB-x%xX_IM=4{h&9jEK44RZlb-NC@`pEWy4Cbg72A+!zmh~#SVUes7cHUP
z_2X^!q$Y32gricm8rwYIE@0Qbx~HTmf`jhhf>5_=X)Yw)%G#L@Rg0p+fbwP^SQSzN
za2~+?kJmOYW=60t504pz^J5dD25Hkl<m=~Gq1;Sk2%p02=@3%}yNBbqowL0^fV294
z>{h+G-qwr>n<&BXunhksli2~5Mo}MZc7r{Xc%_*sw(2@geSPQqHP^zgi<kZb9W+H&
z_;{(Gi1#n~bQ-reJ2TG!Q;NS)fCC&2@_8JZ&iehRngJfq?OyyZaNb>Pt3n;3gv5AM
zCb#3tN%(m__IuRM?3dD2cbW63ed$r_Z9N-nV<AJ`gPX$b?v4s$=^sTS*UXj8jsoxm
z=54lUs9@ztbaTvdd5I^#sPg936e{s9^S^dqeJzanhhBMPG8+b93{c6yvH-_i0vM=A
zw(oiGS|D5n^Xq^wY$l-x_h2`flb2uh5BPJ_P!q!&GA(t8c^tK2mwego-}QT;Y({>u
z5K_$zF5fS`uM?f^ZC)MpX>a|4m=eIzBDDQ$ip3F--kJc`7234f^0AhF8(Wi{Xj4^t
zzx_<?bPF0+u@(m$ZJkDDyH1UxOh-Ud6~h<kKWKx*4vI9tvUM^H&}eB1%)*<j@Qk{y
zPpz;paK0V96-JGUhglmHk`Kairka7juQs>4%V|!&Pbf6pK(AQ@YE%NtK#Ifk3^P-0
zmE~sl@Hlbj<{D@r_fWW8(^E)aeyhdrlOipJnffW!ms&vYHNE^c0s5G&Y*-DWRl?Wn
ztSE~Nk{QK+D_ViaV*d>DDd$IUDx&m6=gTeiQX=&EiS4iE$w|n<jo<Q#FM}GqSExnR
zWK%br5@5P$ypYpJ2(`1m-BbyZhBI}<Da^0_a<QnUbCVphfjV8<imw%)s4Wn!KM@4=
zP!-tqE${#o7yqH<gI!B!GvsR)Pp@-FpJI3C<2%D|+dbrrG5HX|WHh~;2=gSD??+aM
zf;^>7DWK-m?BV<-`Hv@g^w=GzdX@v;cK4^IhFhl{7S)+oSQ~k~(tuuqCiTIco%5=_
zvYf>Tncf>TrC~c=*p3;N1%k!&QBx?=?0OEv7l70T+fbs%9idF~HMjOlOen_1L^;Zb
zlb4%CQRUYo9!;+D(yn$^w6x7nLfj+VxRq%IA?@n?b2npPF+c|>K%4LF<GzAB?bOnJ
z(_Y#S_Ag;AJQh<FDkYInY<xW7-#mXA1tL;;F|62FiZ7(`$tcN8h8c&+66gufJe}=)
zXu#pzaoWYr+N!eWCNMdy@A_bTS%0r}>IoWnX3G`wOfQe9uVl)FnpD3~#PGwg8a+(A
z9sfs=$P?E!fD0$>0-ak6O82ui3jVmV(le;4EN$oDdp{7S<{|CoD0aOi&tjgvIq2O|
zE>DnK*1y<7b)aWusmv)hM!>}kKp$7v04^!XSs!yEd`kMvj9^#LM=|KrPW#Ztn6HM5
zuyO1=zQG2xl&ZfB8ly%nvXX8^H!>@r>l=N7TTk1(@anB?hmpzl44P&nzr;obCHL2-
z=lYvm2K`moFAI?2P-iAroalh`Sf921>m56B;ro%Ely_)|$fIAIaa^#`j)jf&6F72D
zz&k#5k+J7V%ku4>wA}f8nD}H9sDQ3;W9+HJqP-mWB%4LNvr8;w^Sh0ma%`~Mpp^!b
z>nqgxb-w*=h#kh&k#fz~QS%dg+go_}C`tOVW1M7R?injCP9}|1G&Gm{E2$?A4z35%
z;4?1{=BP)q@%Es|R;~6M_R>1nV8A6a9ro^Z=`{O2W1^-mxU>-2bae8(z<;?5S}?Kp
zN}T%i2R_#>`uC?_4~nZaa4Dv*o(+_I7Y70Q_U%tr7+$>!8pn(9BNo#u&PRDKcW}kR
zh~Ta}G8|d{k}P_?f{A(LqpY6}rh{W0xF_#=q{M0#W-B>OQEAm^emh#m2gU}fW{k{g
z-VAqerPX`O?cHm3qP(l=vkmuA{pi_5C$&DiU*|M@KyCQ3FssU@-cTH+?c?<~ubaK1
z^IHp33QM;8-COy@Q0mN265)wj3#;`R!o^v}g3ay9ZtWctPsC+pm~F1YMlGj*HP|CZ
zju&6_2M5(E{5jn2do<Y%1DaB;7GC1PBYq|3iRI$hN}Xx3#QGfU28+$Grs4yohp{BH
zZ%5^SqMFW3awzZOP@|BSi_W;5=m@B4bIiZoA-_Wh9qB-=q=r-%ygsU9S{^*JPb>2a
zj5isj`?rNIK(vMlFedMV?b`nE@qv<bTYBhjm^)C5^UtjdtA<c_0A1`E7kdH0Az^G!
zba3_g=T+KvjIc|oy)ATOx>lJuHY|1U>hDTs<|B-_GE<orH341i`d*&1ik|E#G}RQJ
z<d}G8vI(Wfw;iunauP2HQbc07sA!UFF=sB`O0OU0obNEfq9R{7T%h|E5hGiJ2;vNv
zQiA#be%O8Wu#7b)fX}K*(I20rfe#H7Yy|nD2EmxnKB!SSXrS0#q5m7@0FwZZG1GS!
z)}2Pv4-eIz48M>MO^n}CSXzk30@14^6t=rfIG$RGvHV~!VUgG2+3dF()>-#Yfie>>
zw;fQ$O}*X9fz<o7DvRHc726#IF&@*VQW|`4P(DpO*Got+AnJyiO_1mhRyi-G_vzKa
zneVK>WM;Zhz&7|s`WV*Fv7(8BGrLB-*!3W|r1;Qi3x{w6)3ZyCCc&5ep@6vRNrJ;E
z>#>>^e5CvI%*H$$JJb6;c+D{%xdp2I6ghf2&XnBsZN{HC4s^Wy{ys@L#*+lNE@Ksk
zekt4ch71Du5JCPc$0>KK$mw@c6Q(BJ=LcS1JBPGcDuRFOeFK3TJnE>azF4{<Egq27
z$u4=$VpA;%$Jpq`pL24OcW3{n{xi80xXDLh!$r2n{~T>Qo?wDM&YhS1n=Ck3CI{9N
ztr|1F*Mp5v+eL1$HBQ0cHOJbjS7%BR0rKGF4p1HJeKe#KH^;!?c`l%YJLC8?iqZ+s
zG)Y^iZ~HEyAs}vyyrCa?es|d_Cf{S6Z2tbMdXiz1D%)#oZvJI5tHAoQQt-76=4&@)
zW4OH1>e{}Wg%(=urbJtB@JDcKNd|WoAt<h&cQ&S*sK5w8ZgngrFmYm|Tieg2Z69yg
z8i3TZJ7HXZey9zRZa>_$*(!9;N;&?mhoQKBX~RL3`%{FE{n{KQYM(7DgUbaL$DZXX
z30oxSYt;o;ybDSzJRjz&^!Jy}U@aVxnDRh!&x?HJdA3HV9da*-9b<g^fk<V3LSzh{
zwR4*th@16&Q-XWH)qG*p81GE-dtkf#4}j1!0!vP}E?N2;ZrBTT2lb$vvCzz6=FrMa
znARa%(3z`j8{-?WA78xS#t88ncc(A}7iKF%%OtF$-Pz@Yn&fe@6BeYhOBfd%>eKG5
zVrQx-L|wN=9Q^T^-AcrBJ#s9>WpI$gHOL2<?mfe+URm|sq!-37^A#p|(3#6k0LX72
z<gwzT^-^42n}#bS|9s&73Di+IUdQ$a{<?~JCOQpo%(`>?@5(F8NAlUp8X4Y)lfd34
z%Wwo*ry;(W2d)73d;dQ_`uBLs#~qkImtlj?0U+Hj(vKL8(KND*ra;J;T)*@WnMz|;
zP<9ACHR=xeyTd^59+JE_QSnj;7GM(EFHTvU*!!7Hp>4ahkNSWP6wO^m02d>4xcJnW
zYsUlNzk?gx3<^*>G)P4$)9sRp;xO-TWtyM7_th7p?HtnhQRS(r>Dlr;t%mOtR{-XX
zeWHuQ3ARY5#nTrvDlzuO&jG{lF=k)ge*5-e5OzS%*{6F}8ly?~V6szO0$(xcd{f{`
ze<Z1bLN+eZic?Dai08)Dt4uy;r8}Y1-j4gAwg126JI9KZd;-0J)0&hU27Z1MY%^E+
zds>0Vx0y)e*WgxmmUc!H?l7ct=*-9`8Y22|^7LJtqF8x`k%2h}t^NCr?ZL-;=5C++
z`d_NIFQ9zN?G5t#(l8)y?Gz_wQASRPqmjk@sTuY2O#b%|8V4NvI2W2qVTV^G#+oX$
zRooK#kB856J=`K5cBXe{oM03#;1Wc#6zn60BSRDD6*^YR;kpo_Lu?h+A)Ri4g!b`r
z`L)|gbZ}^7qQ86Kx3myqLn~0iDv4tKAi5Ls@EDuNAk8I*gAeoyOD4+AG71>l$x<%i
z!Fmn$-*leFWrBABW4kGvXj9E0(vz}ZP$B0O|NL&KJRfYvV7IIs(z$DDo<%e!Y8LnF
z27_oMBD_jcMM2G~c7zojX0O5LbA}Ue854vFGp1Maf?M(1fFNUK=oE(#ZAvpSl!RS{
z8S*&IzEgiE0MIGj-9h^vw^8+1N=+jPKhnOp*6UbLX2qj4Z;<p`^b);jp7H{l-%^a|
z&z!>xL%pKb@ebNNmS3NR)T*FmjStXbqh)0lma<C$olQE)govq&w20)4z!Phur;s>b
zO|{v%VS3Kij^_4OYZnAUJT(9A=ZQb3X+nBI5g?gJO#p0&pBsC^YPp`CQX+<JNYZWF
z2rKCj-Eu*X)W*j%Dl+bsD9l4%NVXc5>EEP!OzAro5}+`UH{EA6eXTfUVchOJ(va>Z
zQXpy}rH0SCQHppT2kpZCzVP&u+NZc^apb&?9rJtY;@SdC0+Ot=uEJac?tLOY8$muJ
z5secSCx9X4g=)&*QV^Mu9n!q73ziG;cjUCi!}F0-bS938(%W{r75W5sbU$Z`I|Vh1
zZ}qMPE`ZOk7!cT}Rli%h!<<^I$#FaBf=p{-UZPYf+MfY3agiopcN-&CNoE~Fh0pBt
z$ZmRYeE9vS+mj1*^cNilXNh$O^(!_Sj|OQLP~?2JQ|>Q#L=|yAkJd0#9w049r*^qY
z^oNPpz~!ut;a<{g#T`Ks3BT5!+kLw%w>%%50cBNQj-CJCb|iM^;!WCpjB#f>jkMY>
zPJivTeWFe0dHXZDK1L#GF5pphI<1NQ_YWd;YjM8b*7tbXl?vypqH{{jC~6<V2P-Y$
zhTHiX?EEZMFDd7%AwiAdd0MCwdVoNqr;d2u)*CDKZa~eG8=M_Pp~n#IPXvp{Mf~py
z7Tprcp1-9wwb6R3oL_tL##l;4J?Y4DlZbk3Tq0s^{R8IaPbhyqB#m2s`vP8LX2;pl
zF<f<L`ZK$bhq%{Tggpk=KD)A?eH;n!%jRm4-AsuWY!UL-doN}6@6~!eLh4UMEB?;m
zvE5El$VJ1r-0PWxc*mVS3bLP~goSw?izJRG@eK7AG&<D=uf9}V(26X`wmL458Wr#5
zOM~O7qC<F8I#_pnhVMast=rQZyeiP4<^9qN%hmAbYt<~BS`$j;aVPunc1|JS+o+YJ
zGp+qan({Qz3fCq}&&Z_aN!TK~_6bQ9Nl8w&y@y%uCsL7?!L|)(<ZkLEDz5g+h<4tf
z@J8Fog~OVHw2(l2s?W7LpAjAmsT|z((B&__Q~Y#AMA~HrKHD|Bnv192Q}Az|w=3p9
zC^bsi;CcvwjafCh9uX=&;-UrI;)qEFL2H3*xXp^{|Awb}F~am~Gw~rOe$*k1S;YQ?
z!wE*E@^C_0zeu^pgUMi-Rg>oAbbdtV$)3T)NQh2+sVQj&$4b)Q$+|>Gm6ji0l%GDS
z7Ii3&5~1Z*y(_QjAAd<7M){EJE7{q7`P*k}S476e*)^=8jzrRcFq8mneb5Ybe!}D8
zsSFRyP8h9kWdu=9W)WDbdkX7HiFqifR4r15d73f3N{D07rs%#;_U#^^GpPklJc=4G
zB1I7$Xy6-vCjNE6i@>g5LJ7m`^3yC<rlGIG43p_<)H28nV_KT|KhO-8tYEMOTg?7j
zuC=y3j~Bmt96Z+@TL>U(VeAT>omW11P7X8fzk5Nh@nM`rb<<rx8&qfT_aLoT-}u{F
z+vv{Od~p1kZ%xWUg-><ybO$*8=5?%3l#K@$>3of5OspB&Hty+U{msKf+3(DW6N4@*
z%hW%yRemZyB3cm3+68addE~cHTCzf&MGPF}Bf=<ce#_!@nKiH!7vA>msu{UG8Sm1(
zf@Vm<#b$MIc!pmHqrxpTfcgA$XSQ2KQ9SLw7%XdrW6sRA`aKSZI}1~NUWAQWM&QJV
z%t?7Y1t_xd&CBtx-15cMbRoLN%t>GarZqP>k)hno<P<B#g^zV=`l^J2oR)CzFudQ2
zl+mdqzcxv<m>f_`NUX>*-rfVBqw8K%+j1&I2NjVNdJ>(P{yle4p6I5|o@WS(zq5)C
zndD?D2Sh!Vh0z@)q}^Z#4ffj0%B(Nl-<;g^z=p@a=y?dKhI+d-lvQI>k$}=yHi)Z#
zP){`HRCE5eHw)*j;9za50V73+9j#XU!fXL?Yf*6_N8rAf;E0MA(v>_wF^bI40bxuq
z)p1NpyW@tWv#zEhzx3lS^<mdfb9|nZK7(|#^}wQ8Sy&J8j!CRf6#nA=$KA(3U+KRJ
zgS{>En2n;G=FNq3HwO{bLF_e)`<uaGwxR-4f!Fe|tSr>76dW&ZkbjEbs}r%ga%Zb_
zH>&dD48X(iZ1sF@!v@bwPYlY~bX|qwVht%1`E5dpN$($%+hzY$@B0n=`@GW`Bd0?T
zEdkN7uq5&b*RSnV9PIy)w_*yZ7We(|<VfF7OXrEb<dgWgVDP{!Bf-zTy~XG5sSEBn
zU-51nw_PaLnok1Zp>9FZ8|PAB%M}}n#k2w)7%hfBU_3b-7d<E_D#BU>cWTtuoB9^J
zZnUp|tsEcJEh=AbpL$oP39F7yOEZL@xKc0U4@o=H@o~85)`-b+8!m^A#R;RBVjA~4
zwn4+k&ctuH5%c*^%TYq~ne}sUsXWgz;>{+!KK%-HU~z+o;EuPGW&7KK68ZXT51uj5
zVR4j7QU+I@DIfhtv9#{hSwJ(%*ZTT#>34;qux5DRa_qEbzvl@t!RzUA(^A=9{pBtF
z{N6fFRU9fwJFtOpXHNeRdYsD1>od%qrbv*bW@RAn$fxQ~p?|N1(wT*1f!jIs^kN9D
zdpbFky^^=JZ(B>K^;CGe#Sv8QV^S>u+`2SfQSaqaAvx+zU5hg5F=MxuPOWmx_rMVz
zhdFV9s~;4mKbBqnyA~@d+7}<IB67omfTi5I*O7&8O2MNG#2WjyA&EBk`DHm|i&UA4
zjPJgK9=@bT9=_S(gQwMv^7xyPNLd9$pBvk8CF;P+S0$oN?#Fh_ow;5uQhqLg_-kd8
zZ{UW&DKj2~$(hiS5?pX5&iSQ3zO<6W6b!bf;%B5i(P-bdr5tn#4Z)ezIY*=wUr#E+
z!W`VtMQ^Ixw_jEaGTPDKu|Bc7pG%rpGmJBZ2C!!hC0Et;D3$ZPjF&^4cd%Uy6znC=
zJGYCN2K7lJIL!GWJeiImM#U8XE2;jf0%za9{^`nwe7!F-hauf>JW9#DMhY4CnqOXx
z+GvyL#Kt}`wJxI!8k_}J^K2Elg#nw~w+mVPMR05KG(L^2V-`3)c4jRF4RpNSF}9|z
z`(QrEkrZ4A@X%72U+|sg`HEVK;O}=`Qmn$^{<_j967)Ie%SMq`Ic-+pR%np_ck>~X
z#GjyN_f{(VolAn$_8r*I0#>Zg`(g0jhQ#pyBD%Bn$<ui}{4Rcz<{c)?HQYrImJ#<w
zJ=80p8<GzAKXKwS&qeR+=MkVv{}ha$etdNO7|Qc~hEQmQ?c?PDwRG0jXSZ;_b!67|
zUyS>PmzpT&cn*e95IlGT&%*I@bqH0)!e(KS!yx;P?5M4J%p(6mK99aW3%*X=6AKDa
zjP};{ki=I`vRKfrs@NnYO<~LE(slSl1l;h`yymV8E^EZkuBfPpx_G0$wJQgP)ppCU
zw8s_}3F?0|0|KK1aFdO17*57(blS+CD4B`!D*v(1USYR!x>W#aeOIfI+^eoIa+FPu
zmCMup=SzqLUB!i!uQjRa)vEiJ!D3i)-qP(@xMvW5RUMVa9F@g+NOg^1>mbn1lz!YX
zROTCNje~ukDotI+mLH@voOD$k<wZY+)CC9cZc=?3Qg+Vrf^4R-i^>mG?>seWKZIqN
zK!2!Au~jn$_6GoCl0Tn8+ueaZd+<Qj@+r|olSQ5QXZ~8*CJP+R_BWx6!F*nNfwslK
z0dnK_r$5G)wniR(5#q))h%Ea7ONk#xNw)mkhNi;f?FHm?up+)^1=gW5n{QW0H{XZd
zJ-`};j?f>}Xan3Dw-gUm%AZZwmU7%5<B`8#Sz4T*L9qbc?iAYLdTouxMD#4Iube8k
zt?_+}Wv?AtcOOKFOg3SbD-DZd7q2Y84S=iOvBFUOrhJ-HP<m-aLr}zBox&CuDTJ=-
zKJK&8;AE4Z<~zjgLi%x253UF~T*`<YIBBPCLIvycKfSuveXjn14~~86FDJkBjY};(
zuX6-YCV(m(({XX1xC5?uWrnR2zN4dl+d9-7t%vo7@=j03wxO#R&<!iu&=sBBDw@j)
zRLUdkdZtJ5Y0YjJd7TS#ex>8Y*(j_Qh>m#9j68T<6tK9H`e!9lh{<*<o&9&&V&^|j
zi@ctdHV?2eF*9fLa9^JK-KOYbj?duS<z0>+7-~;~Hb^D^E_0>)2#yVTjfd;rFs6xc
zE*=4c;q-6QjJqlly#v&Al`XSfBl~7cacwCZSUY@Dy1bm|%eDla97au&l8T05J4P^1
zh$uBE*yxILjhA!l#{Mg%2l0(vhfB#A&_CRzHDw2KxxgW3>}8et6y3cN!c=X3=aE8y
zPW!i=icbx#uPYnQe{k>N4!~8)h+V9n;G5^#WO_~QFXoL#%t45+tFwpnPn#{6xc}l~
z)D^U#Xho(Kk@5Gl<X%n}YK*HJxACnDPeg3BSXZohFW>0}(m<l=bU|c$;0a#5F8($T
zvko50zk7zg5f1fPr~wh!XZMGr3!*9jJYDJ=+ewOe<H*ED{X7|J&gCARqjTN+>*vdl
z;C60u!nI4^w`@-}bpz-{FhTIk<%i%bbG_z^<%Ex7ruzIFhAj36L+T^wy&>fv2W2&C
zb~l7_+gh6RG8+5&$uYP%=<D^5A3KB@dXj=y`)19Yg%f^dU%y;gj36&0a$*gzI=s?X
zn~*X$GC1@H&TFjQlw|z%b&;8AOZV>d7%j!rDMh++;M{k)_6Lu`=NTbc?~w{8Gwg4!
zkl-WSuFI-wuo^PFOFm)zmM?0w06cX^{7x^(k<Vvj;!zKbvQ-7C+K5ZFemjlc^I#Ox
z2XIf693^$dJ!7@u!uFe}dMN>Rf))-2Dnk+8rO=Spfx^5Je<oZqv^A7MEN-fB{JQ~2
zpanFoTcxRFG`MlgZfg_N7SC|`lPtxMg!v~@sH~I#C5)owY+MGLel2w?cuL+z>gTvR
zuOb@_A~})D>fBV^9Od26%Aja#YxAhk_}YYrV@J--NqGJ=T2er#0uSWQ6I=+)-BEF|
z_Ki+^G=8J;S5>Ahe4J)P^X~rIwS~OW3fFHC-w@IM<oh&S-JP)zSTfVomS%iRDZM*)
zuRiqY6LQJl-vSEy`taOdv1u9L{#;I0adm19_JQ-);{!fl_qd%)R%0pdig`x;l;U@y
zWF;<ytjI$Za=OAYQwsnt^FYf{GUwk|=GCDB|C^4n{QPL!Sl?0*qb%};=vpf_1_mCU
z7v`4Mp5KvZ)NUqOKDUsEVk_+6K(!r;Ev_Gy;OS`mpCs&(;IG%wNo~0Oc5Y6*vRJ~7
zLZ$Xz5Axa#{2M6McUU`x8<B_)&W&trb4toe@!<EojX>tH1_9V_%PlRm{-lfR85Y>*
zx8>ftlJfPNmHW!ilD}1qv$Jw>UFI|pKe#H3#*7?=Qos64C!<4yrsSZ_6NRj=sI)qU
zM#qOI+!B*B5)(5rGGbzqlamwTUUIUEn)1rNg>nc!DfPuey~96}!I}9UJhwffNY5EZ
z|5ihRr^(CF-^T!=qvdDn+3okO^mKR4>+BESXm&zoTRJNn3tLNTb0hdgjoAo1Y##51
zj(+rT;62~s(->+ZUB;`6GhL#w8@UfZiW<f{cl6@Vf4Wim8`xyvYgwA#TwI-9(NNG*
zQqs~=P|(oOP|;CcoL}FbTZ)}_Ub!8y2?u@EVrTby_h}D!Lyoc&|FbO{*##-~lMEbS
z_O}^~>$nAwwH4v_q>~yY55I~g2}<OFeRF3npVAm)FmgVov%P!*MvqjQp!?V;PVhL~
z&%M-^X?8x42g9$}%O(Z}MMOnJL<a{32Fb~SnNgc_g?z?)_3yV84Tp6%oSsK|5J>E;
zg+4VzRoMa8pT$c~)YHGZL@TXqtg}7-bS-+6%3-X=EK6{+Fvw~xuWo5$r>ExZ+8?5W
z@oK@xJ;5Z{z-Yv;dVmiZEywM0JTR`<;N2CHiDSR{acaFWae&)D2qrl$ZZ69`l#uX|
zzwr)PIedRqju~r5gxC}IWWmh5GD;UhI!bk*rV<98)BARv6^m;1texR3+K0v^{suSO
zPCIoDxLKwA_b~n!Xx9LehKqWKdp{&G+;tg!eD&Mfj&0;eHsm5vm7u!L)V|F1YGM#^
z;-r)FFQtFz2YKYYkzd+u*8b?fsj0>8dJf3_R_dJ^ndc*Fu4*f&E;lx@xONX2Wc{5q
z;R}RUbhMNaaGekL22T+@panTmhW=jwNDR00aZvy0P`?1|K4OHO!3v<J@eF?W420ea
zvb6Dc9|^}meD_8SMiA;D>45cF*$lqfAs77cXop=VsR=RnLatx5HwCe;SwV8VV}lTk
ztQ=xWyUZNDZa&O}?ZxeVBX7Ttcw$ZpZ1L@lwXaGFf3j2Jqr-3dUAl0>OixXA9V^w3
z4)qHlvjB{&Yu2t;+jsP=Z&*TZaYb!YXW#2L_|5I-S!o_d{X2wnz&hGl2LC_eKf|}&
zmio%l!t7KKdoYOI9>y-fx}4;ApP2$#w_%%sm9y`i__XK6l}%kkZ@!y&_=r0(HrU--
zUs;l$f0Y^^6?Vhd!^OeML|0v2gzMY$7(WC^qbYn+7v8kvfZc_Fh@@vPN~)SV2gYWs
zB2ngNxSfKGL(<>lH`||PvVhTn&gL4p$tM8XuX>(yus*1(zEPB$k@RApX|LuKSJE-H
zcMFJkocjD_O<Vu?cM^~EkAs}9uf6_N(ew1De-G~lU-NW^vv=Ama$>x!G>d<9s9yl_
zl7Rz%yz5o>SU7s#h<cKqTT;`~GlCTK)6+A#I-yth0c9L7+>j$-8p~<lqU|=%l42un
z`FR7{&EWHd1Rn`KPs{)n*<ZYrl4acnb^Rl!y>3LrrM;+V>K*%D;*mM2k>T#9s^Xl~
zf4B#s0he4ItpM#?6{LlER@0NvZ6yw3%9UM2VY`vlS)W_?<J0oW>N|(tVIHBMXJz-F
z2!)4_L~{|^aX}n&gB|rqyG?k|Ww+Bd=7zh}6=g*E*cnLYd14lRL&wJ!(s7Ax);6|w
zzI^lklV>mMx<<c`cyvx`WO%T>t~59Ge{tlkYwk|AhYfYqHj=Ix2O;Lb$Rnk^>!9s9
z-&^++G7GC(2i{_r)}Lk-^w^h=ef>D;81Ylc{Vwb=d-gQ?PN4U>6Gso|Zc|*3ij<R<
zfa#yGjG1r!*4?Hc`jF`4{HpfX-%b1*_$JfeQdy7@7ZH5rg5xohJsO)N`AKDNf97kz
zq;FQN7Tln&XJ&ui_tt~N=Vgt(GZPX2_XR$G0qhla9L}6QPl^up_i#FHroTg3R+x)a
zhQ+=Cq7~;H_~sLH${PCyhK5JboBJu^p%y#l@OvzHk@b3bcxZ62ue-AiUgp0l$b1rY
z+uzH@{>Xtnno2Ul+{hFL3CzuWe+`&0!g6L_3B~RD=Ju{WHzMLQi>lkZ;p6i7+bO&i
zYgTqY6a6R_{Xkz&XM1aNBiQRwV($iBzJMNa^5XpLEDUrjmMtO4MUMmT$S|{Y_P&1a
zae6^jU1JMs7dbY8&b{J#E@*!RjJLb9qph{Msj<GU`c-LhL0(Q~TKt31f51!UPl3c{
z_ckS2F+L8GnHKvBFnPOW44gtp51FmAS3uaq<lM3<XtU7&di)(xtj}<coM>q9?QQT5
z9|T|avhc<8tR#@p`d)OhvoO}v+PYCvkc)+c7d;N79YuD_c0E(;)9!w^qn>8HC@Mv4
zCT1G0@rEL_>$<(AvA(A2e^q(u%fkG;?99}pgxHw-cW?T6yPmQ+Y`kZO>SkH7b(}1;
zfOkX^_eV$gHK5EdrDNk2lT+5(YifPU?dr|Q`1I$6Wz~)Cy+cIL@kb)|(cyu<?hdfM
z)zwtN$I9YDxZaWALjx}Z+K(RGr=zh&UQ%f7YGmY$L>D~{iscOKe|(~{iW<7cM^Ct3
zx*qm0A>~<KaYb!2^2$zh6d@7^1HGNCjdfM!B}MtU*_mm{Pvaj&MIbF2K5l1ETAA(F
z(^gZGLxh9qVK+KSOpARL$jI;F6>$7*gA$1Sh`ozfKxkBaN_IhcU2E@aWV?fpH|Ur<
zh#h%quCIX~d46urf3u9#<b>F$h&#7}{H}PoI3Bk+uuo^3GNu#<hXhGsSnOLspXr1p
z<y3a;GqrL&?-Ljr8Jn2-qNKVB*RPs+pY3UHs;ew5%*{$qej4}i0p#4@*ZZQYlcU`+
zGvj?ankt)QByed=IB`Y_LqlH&WS)YGmU#`Iu$27P9s3Sie>u2%2fzt|ywbY%0VwQ&
zyjG&O6vHrf`0Q5vA_raY(f99#2K!&}a&tLlZ)JW!f7f<ZML8)^MBs^*<X)n`0^+r1
z{bu!DhGy2M&R+_+4eHw5k{Y~m77&j#qV@N}>7t^%XDLr&qa(uLTK9HyIdjtHsJV&3
zo}JsaZr&gzf4Yu`6HRSMW@zXyfNK2FF|Fq27nV`f0I}PgI)C}bz1XxD<&9W(31A*u
z@_k5OM)8a6v_!DWMTXxF4)nX^e(tor)nO33Hi%tLT1=3Sn*%CiE=Sfp2@H#U3)sXn
zD+j;W2BmGg4G&qJx^U%2#N)KQa!jCeavEt%fEou)fAC^5?dhZYVYjYdyX5Zb<Z#^5
z?0~+mwuZ`P`SlV|6q1Dr7re(rL4S0B-@rVIrRcOb*zdP!>Kd6_J9!3%Kh7wu?&$9A
zA4Hf(i`Aa4j;5OO!si)Dad3OT?swVC4R#(MGSbu5*rKpOT3m>K?HV>_21NaO38@T=
zeGAYBe;i(5(z0-_gU8*@edcy9mxAuaWfYcIWAAmiwnbZAd0`HGhqxJV#q+$=Nt+|4
zMtVCnRW@&s6kf;6#m>q|w-OE&Bje<lT+ok>?_2oyeCUA=W9Jf(kXP2+ZFuyQM_|O0
zoN`dt^gwIqakOz1lx;1wAa?ldcst1NvgZZoe^d6>=EfkUQQROdz{|Y`)eu-li4BI5
z$h_1yfhhqkSO(8K5$TP}ItQ$ryl+H4eGc2@JNrl9O}~FTj^uSsmGD;fe(3ef?k<kD
zM@<d)>`+sZmqv0q5)nQ4KFq`pm%Xf_+FlEEIHtb2trO`?nLy_&TAQIl3i1GT)!WVa
zf0W%ZaJuPeC~XkqU&FMLq|?~H1g`zHbOk*sXQQrbVs*yD|IWiFDbI3?z-JHYx7w--
zn8~L<eHalEaQULM!||hs_V3lv1V<al;h|C~DHm^l2_p8DjH}_m{AMli&3j%4-+W#{
z5$q?e0flQt*-H>Q^0h-Dv~w`2F*^W~e;RcqISIkFZ1k|Nj?`1RuK<;wo`=Xpuyd~y
zk=hKZ8_ScRvI_~1jEPT7%gTG6m64kCBrX=V7(nSqkkK48+N-l&O=+Wy1XN96UPao)
z+g|~XeF-%kGY5~LxQy~nqazL%u7*TCjEzf3OioSDOixXIngBwNx(j0Wx^Nare`SsK
z?%n}&SqUK?4klX2I|<h>e+Arf4Qh8WvvctwatQ`z){d_3-dFsBZr!;T6?rc_G$c4E
zz|Y6s^~?!taIzzTD{YjO1oaCSJG_`C5w(l&!%XZe=@H%b4Jzm?=vDu~>o;zOgofP>
z3k|t><9dMKRglA<IRU<PV*|Zif7+UA${?E+gMK6QD<JO+Nj!)D8n71&LeEG@3g$Tm
zrj|B#4yT=6FM9fTd%C-wKj-Xp`UG6=hI%?$8e5eVHmsKr5#ZxqgP1yyjMT^XVaBL{
zhJl5hi&s<@R5nKDHjZbUoLyYcySaOS$X#8|I-NdckALc#>MBZ`<YlBle<BaH$Z4ql
zGo!vAj~QPUL<5YCi(f=aUP(=Bx1PR{@j){SOY38nM-H1GI%r~Suy;3D>?(>IH>{Tw
z6BgiG%Za$SKuN5*89Wg?CF)7TBP1!Oq^_;E|A5IsQ#13!7Dp`(o12+}(DxhYf!G1(
z3i7hjl42r)pp1cr2DqKYf1C^bU$~VLTGq|BYf#*xqN<^(rLD75d&hRoZ5nE-%8Hvn
z*b<^b0{lGOoE&T{Ot5f8j`uklJP|w8UZ;ZsaDrm%<rTN8sjF*j)7-ALLkmW(uBN(G
zSrMgNSa2O5oR?=us&;r|fnHe5z1aVSneM@s9H`$0ZMfJtxOn&le}sfZ#Ka|~q@|@K
zCB((WFpU7f_-a-bCPsQXC=Y{*f&GVuw4?wmc8bMID5;jy&@-`daPtX>h>D7diAzXG
zg2*Mr#X#sVc0Rb;SF^D)Gchu(qC<r;khE?NRPBEayd?+mTv$%Ml9p~2Jp%(H6Eial
z3o|nl6C)!d13d_rJZ9wz>g7~aQ1KC}eNy0tP2b~ANNPaEZDCgq?8sV4y9xx($jHP5
zf(MZ!p%bx#;Fp2mq15SYsn;JJ0+Q58lQ7f;7cBlCIQCY27#+W?00000NkvXXu0mjf
DZDRMx

delta 53889
zcma%?Ra6~a)2$)6ySuwP8+~!N;GW>_4vkB2cPBs~xckNl8r+@W?#|)6IQM^#zU<Xw
zb=R^v>#0AbP&?&NwenCvm%Bmg>H@6KeKQWjQ9{BwthG3L0y+%jpco;tDmpYI4!RVK
zxSHyJdeDhU2}$Rlmnz5P+q%G}XI0n>agXZ*2K%J>>s_ws{V<`(O|I(zSKd`T@O!nr
z>i={D|HtS4uaCWbEWSVPPKol_Ewwt2HIsl4B0Un-R8`g0)z#G0G&GWu6Vb7DcI`Kt
z-N#B}iQa+NUCWO-O}>GLSRy`CKf0d<Vx4g38VQN#l#~OTZQoxOShD%vRGv|VHXFUi
z7VIAf%)Yw$>a;yR!azfZg@#E<!g#kWWgJ@5%RxSan(db#52oI$D(xdLz;|b0Pn#_e
z9T}PEt&ys1WPWytJDM;H4gmrN<i<Tq_WOa~$1spj6nL`gc{)J%)52sIBP~HgU8KCC
zwAxwrCUON*Kn_Dh*+P&IQJwPjOrQ7lbhCXW{5h^AIWQ$WF;-jWTYge~{iyh1cl&h!
z0fMQBvR3%|SF;XykJp-~_hB-?4!q&Yt+>`V1IO)#GKo)+ou#Xj2Xge0e@K|=?)pMJ
zehYB#4Lq*$zMtlItlZQ2ZC)Zo$ysN%Nd;aAmVLG$B*@Ol%FGw2t_m>8Z$;-@&09~Q
zhF;}+`|L>ttkcK$U5hP{-^#<~Y3d@sg<J3RpQja+mh2Xxi;Y&7UdfNzq!@O#7g9YJ
z%FYSVmxt<Rdx&l<^uz=Wj+-AB5#1nuz{uUB&EqY+@<lVdA$%CMa^wtW>sw*j+Fj2-
z(7;=~l0M`)*i#R6`Yo%B`D~r+Q8)k-pO}#JN$BtJ`t~b9ZV2bMS7Ki)*{hb<Q?kyL
zX6G>vKT35(Lc!`T#0Zt#CeH`|!-o;kEo&3i$j#gn56h`T>(g}XeT!u!>DoI>;A26P
z8~!C9a99bKk^N7(D?8CxYj@APtV8AdE>@BT7pLn6Zpo0)UAxC09c?M)oc!FN;7j0-
z_W-X0Ot;40zg$S8)_Ls^egFpE2P2074G#;I$mul6fkNps;g6X<1K;w<mjC^&AK_c?
z2NvivN&N|0icd?rmKT;$FoWhmDi6wpeXekmca&*&HLtWD+`j>@OlyxfHi7u1Qv1h_
z5y*35Ot`zLHnB~lXoWGRO4}2{A6z@Z7ac~tq8^8VDdejTNoS%jE!9PZV8>o6>Ypnp
zGyIuUwA!8jX~+xuK1U3fTHuNLk=nAz#d<mQKc}Sqh+&5#If~1Fi`HHRcrPHjXE|Ii
zN3F93#Wq3hKF_e>(nGYRy}witeYZT8_gaVEyVhd%db%!_Tu_p_gQ?>JQ-0)yDTsZQ
zHyRr_Uzlw@J?`#Jy!y2q_CVI7=%HoN<+XEU;;~a(1U*IS(wYCZt8H_*3!CJr(*j|=
z#wsspp$y4!^tOV(>H7i)L_nTuaJfP$(GakG>8jQxI;j{%tW`aKUGDJ@3ujs^NM&@9
zCzSROQL~1bry3qV6DF1ViOYi;7o#*e&JytWXkO8=e^zNJ!uLD&?j0}3SmeeERYqBE
zYLtqZp{D4UrG@7#+p)W8Kl$UzFdgLijYL&yTNFmTaE%c=e-ZsCpd9ki0K6fdp?R(J
znnoZH)1Z)0aS`Qb{GOp$pS93Z&Woo?a$#DP9-=F=#M|(Hyftrc=_aQc-JIQCJL*TI
zJe})lxEUEnRom<&2*0Mro^y*~Uh7jwv{|8xnG)EzMp7Av?2eCBQB<0moM4OdSimhE
z6!twdbqp}jO!DOedZ#>0!tM`-CiEdsLq8Xn4uXv3wx_V4Xv8z1@7B9}+eEiQ0%gDJ
z;Cw`S9s`bM(=*?Pl44rXjkWbP3~`)US2h>CsT~ImJ!H8RrUoWi$SG<n{AcDy--o&a
zBSU}Bgy%*5`DK0C#>U)DbYaanB^N-t=exU;+5hrL5E;-6pfNRy^4TwUCqo#xawC(M
zVyrOHlvR8;@qWrW=ubF^B6z}y&6U^!f42X)yqZm4?p^qiQ&LX%kTaD(ld%1h?lV&F
zAe1nYUJ_VduYtWr-zD$R`%!wbX}&rar;Y<M!t5jsy-f~oE>Hyc{k?xqOy6G3V&<lx
zBM!IyM<3uD2{dCdHKJs<Y-6B@=;tM-5=ru0j}i{YRMl+wpkGt+R2KYh{%PyvhwAKi
zw0pqj+JD9abHvt9+O|q)`v-yjy|j!p<Ru^e9F#EnN{YX9sx6E-`Fht8|KF~9Jt8Ba
zAnXw2A^7QmOoVwN#*I%Cw^PpNePsT5Pn;voP!w2=##sqxsl}3#8v~D0QFp7z8NRe;
zPTcvb6eYV>9%rp|z)90uZc@|&ueVQQud><_*2<&N3BKAaci3T>0aF_BK_pTCAkQB|
zB~y?yMuskf`;|7gG8VmQH1Dd`Lv|bB_Z9xWJYBuVJ^u(E-%IGYA0kY$D-m6d#UcEy
znFxGY%WMmy-qS=q$>r@h^%3PiSCX3rUmo&D?&EDx1pv>t1Q4PNtLmDcA1LZ{tQoMV
z5kE-cQJioY9-5XZ^HuBV(wWM(Y&E{m*gTcD$^8x`R%N2=C3MNsBzl&r%l<A^N`M2+
zkh#f>|4%?O5mRIAss|=j6C<SaQ;5uMEHJs!{w(5~sy{Cf=SpB{j7tkEcszEqNe%10
zLQtVloA`Qa{9zdWXPwUu9NI)}c1!%<jC{nM((l}hN12~{^KN}Y7Vqm3@%bN7b!^5W
z<4VX6&3p?fG;rQJCx2X%KkWG4)3!Mml*oOTJI%*%_w4)W$f>`GlOW;sL=K41pa8ka
zv<LPz?N89drFeX|%0Mpxm^ddWo=kKy?v}fsO=LeY`|j7~es2F3a?U0)iHwo^f0W-a
zKj)X;voNz=&d5Qvqu<8)is46O=xKeaNJ1+cSZJRsSYYR)Z_Y`@)?Z4E_i=|MAe}8h
zz<Ie23JoGIp1q>n>Jb0F_69L(3}7Ukn-efKv$nt333w2{j9&ECtP0^WhD1{})_cC(
zew!2XHI(xhjkerSwu~o1e`4A2;<gRha2zIAmRFYR%FauSN)5?2mR0;uPg(qop8v*!
zm2^Xe(#RZpz7Z`L+2Lv(4qhmfB<JZTT0?P{y}P-$BWB3O5`pzAZ41fo834RO{5lZj
znV682s5Z^X9#qT<fxvsLH>{kU*Wos}JaO}046bcmY%`BFq;fTeLL&w~9Gp#mm;It-
zJ2P+XiO7v=?%Z<3n1@<Zn}a;u(kd$`3D;y)Hi9(ekXwF}w~1(U>&@)(`uAM*5+72G
zwTb<e*F9crZIjJE6kI&n0|0yoMs4*i)-{x520c#y>83ojw4Nm}qA>qSN=VdDRiFOb
zM~a+(X=L37_1UYb4C7Y6`MU+-&EcRzwYn$*Z*%vvH!P$vu*4$d>5XeYiThwCLZV7n
zjc%?8_1J^{-W9qhv<GFNAP)-(6&4X+&!y2X{~wHmGH!P$ip*?F0KdUQ>%--rhQsul
z1|%jT5F9N#0y6~ZAPK+d57Otzz}5RwnFl1!chr`3(f2iIC%VuU&}B}ZGaA=QnA*Z$
zL64!e(aGhv^Zqi5?dduMh9g}^g$lye7wNK1e;A<w{{RPz_MV^D_VS9g7fPU;?{G8$
z@%1!9xAgk>RkE(v0Q?WRKMcP}76hD!Zr7xI%ZRo{Uf{xnfy|sdLmi{lwPm!N2wh@D
zlU}%Z=YZO-Qg&gHwU6rAr)0vJPPd-^!U#>mK#%|qi`QJH?lOO?S18TD6dreF!u%#@
zx2~HlIUU8FliTr}*P@xxtfA)fpy^+255`a*0nPy)7B)p?Kp{1k$Z29(!#n|t6i3`<
zky;%jsn3^hm@XyM6N7Y-IgiE9mgAGF{01G~i;JTCocreLHWm)z*5-z;zEfLEbFXkE
zPLM$#w-`GjAmRDNuW7mX1Wp$#QpLscz5+nJ%Uw7N2p;0RLtLzX^3TK=8$D%Jc`3@5
ztg%VHE>aj7phtd;g{9{AFIzi@lf!c&;@pfpprsL6=+fWR!NSI`v4K|f&by$cvTE5G
zMSw!9VzUqwkwz=`he9JKSzGn1`h8$~b9dq3Ak*U38|kyqyStWyxU;vW;dfgZm5;xN
zd*iPi@k^sH`+rqb;ymYZO9r8RHfuRR)!i3-S8rrQK+8EbWyl`lWfj!iAaw~g&tK{B
zx4^%6a+zV|1xH6j_{5|n5`vsa>_u;ckAvQkFRHdu%HJ|JS%zI<S4P$S6tR0GL<6Ww
zADVs9h!pkB@0o0m2{QDJ+X7bkwshV6Y&_J=C0uP~v@gJa=@mB*a82Ihsto*|pizYf
z{7b4h0XmA?ppM*Sgo(|F=vrr@cUD9_X)RVGND4J)1_A*X6yf3<ZG#J8)M%wYtkw2s
zL^6ysm%93_i_D)fG<ER#=7#Tr@eQi0#+Y+PBOPz|8~F=~4n0xJPS9$c58tU&r2HQe
zRbjui!<?!ymVCPHx<8q<l$)YKJ1G5LC1!X3fUzE8pTHMYS2-WWz_MN4w|@DV|4!XX
zdbsZp_iJnI8bTs-(xO6CB;~b?)O64OCKmgITiWLqUXaJJRM?!8=E0V;QPix6_Pl1Z
zuuLk&qCGsgef>u(cAuD;7akl6!E{zL4wA@rBv0(8qAV_rh?Q~9d}AbB8CE99Z*X^k
z1JDexTX&I0PgZ?o5&}L?+{Vg6uVmndqpQTAjme4}?WuF)9xE>$U9FUesONNA{r8}4
z>@)Jx63dWALLMcS_1&_`3>~q}3-Qfa7qAjs@FM|t&eN;XmMF7gJlcbM6EE&h5wzAb
zulvyU;1qN^DxB$>W1$&a8NF&Rkf+)-(BOphV~nxF=J+t5L)aICmIEy>{Bs~21d@0E
zv^7EA>JGdo>wzVGBfB^#8;*ExUkSXJekZ-(A5M8V`)679(L^wGsESM~LO}xsw})?)
z%g)X#YH9B~c=XJ2AtR1aCr98I&2*yg8&12=ZhFcW1c>qp%q%Xe8>f<xxqM^+cbQbz
znE$PfH=05B(AJP1weRc)l@ka+4?Hl~9_Mj$i#oav!c?gbcKYOIYOCum&MfnB3dGB;
ziU!<`6n-q0sa#mlg%@6=E?@O-u;MA(j-KOenLQ6$B5sMH_x&pRj+Y|jbeOUI6rE0D
zxaP2acjZe+h?5Q-FEqXZ`?=)-wXFcw<7eaQwD!@Wx$<FBkJ5JeZ(nl8-BJd6_PG&O
zLY5oNL8OX1^aKK7C0oQ9VX0NSP1&A)q!gMJbK@;iI6SdchW`D#T($q+x7i>k@sLmR
zMOc*M?>%`J)9||E{M>Hpf5j!yi(cmL;N6jY&rl2x4%ZaB1}7%(^RfaCEs4(td(qkq
zN6SxxfA1!Z5?Uv_`v0+QLcEjAB>tu!4InuE=XV~Og^kBV#qy8!)z~EfdDa)I8ta_0
zxa8*ndYamTpmo5XKX=WG{d5!<Yzd5-YPy<+8v2$OV0cBtO)8AZ;huZS#Phx3us?<8
zX)E=?<E>y{<pbq%g%4Dq!Ij)UqIj|qdCeCGHNUae_3j=jZ@5$b&&VG(#>(IJR@N6o
z&SiCwXMBOiImE{{6^xT|vDdCGJVKOgBVr0t(lH;l$}vT;vW+f~-@I`l8O-fNT<;w=
z;`2nJe<&!9{a5kzPN^za_zmot(8LxXX7cg4NPA1yNFl(tv1w!ss3oKlU;Uj}kW~=R
zi2C1@L3cc<5n@N=B%oAhOc;SboBi8T>w4plOOTzA1+&roT$<KA>0o#KgPlRt)TqS~
zqKAEY(@6W7_f7LsxUAuGSk}MwaqTR@-Ztk-I`!qo+ScqO&K{5mlZN+Y*(saqDt%WV
z8d-UPix@`VG+hDW7Hr?Y2*E!5T%OKLJ1#m+Sps)xPmyY;?iH1i)6jxfo1e6WZgu)g
zFV-+e+i9B}x(gxhUT2Pn?l(+jiSlx?GNm#?mkv<GpNPQ41KE60)A4Ke$UNrWo)=5q
zXWE)!V0k?|I`YD`u83HDll&nBh2hnBs^rZ*+|)&0&s!iiCQkhP^mI+A8t3LBe9+(8
zqP(H-)c$zv67ExuZq=`b&<ESCzbZukXSX6QE8Pzsr6Dh~HgR7+(Cu4IstCW|Nc<J1
zvRSZba!LfAign{j`PUsc*t`w=i=)<SUuv|5!yrJRB5K((qA}C}#Sg3fk|GbiI3~+A
zdVGxKL;+Z1J`la2AAbG{YR>B8Lr_z>;MXJ7#yBWgflXBnU?IRiWr;YFPNXCQ{nlb2
z;c54p0MYXnH&zs)sKfUhcbdxLUTpS+#4bLy2_Wa=2Oud}F`OXT4da^AN{T$WgOV~X
z{J2&UD2hwMJ1Nx|ylLG1kzTR`wh;tU?_`g8AAoaPH_;+!tizWtJ0RAEm&l!NZs8>(
zjZhdPZ#K|Bwn$K#MaZu)*S_$MPR>)&tksQ6M8CWw?e)Ja^JqCA?Kk1~D+d;qrL@Dt
z(%})X;pBfXw@G_#7BEX@(VDam`WZ0!i2A8qdUlw5$?$&&3GR(US6(N*pqBS7oWKn0
zqX81e-3+j;52&$9Q#~KYlrl>+wVOgb0vsZ|f}Al(-t6~RR1n?G^jsW*m#nalEZKm-
z7DvA-NUoe*Tkq5UsVKd#(+IaXU^N<492C1gynMXvi>L`4_Q;oxfP3+67hK5_pId~Z
z%tt6mcp~+%wSd!Sa!oGI_h~cW_m|z*2KJ1%_JioH9zIXQ3Shx?HGinth6NA4b7559
zOe59#m3a`qAcAFpbN;ih)0Sl^9Ua)4jQH^6?EDx^D?lb<qBBos87_GIqhX79P!rk<
zaWCZKXW+1aN>_w}th}1xL1W^wuI(z*B%jbZT<Bo;BumxFVK~&3Ycycp_ALt?coWXG
zpFnBSTzcPQzg`X*x!C<wxcl|LwYTFPKsv{ja#ydx#Gxpc-c-;vL?le{k&VGwxQNg5
zCwv!=Ffw%HeiiS9(@_LV4ntiIwTsBAWXBh0=XQ?v4mRZyR`a#LgRYOn`Pq^3+LBI7
z4tgKNRM9XA5d%SwTL8mWUB4<2v-Zv9*eCpeB&WJ#v-@Bd4;`BzJxNzpZF3`DZl&i-
zz!3@99~-<eJ)^3xtM|yuR8#)D9daX@{ZFpG-^bOMoAGx!AG_N4%Gjvt>l%K}2RP}z
zP#->`PqFT_H^H_HG95AFKsainnV&HH>z>g74AdRs9+c7zhr>O$_woRXI>s8BMtWb3
zW}!3jXJuocpaLNLEdAv?Toh~5vAzgB(cXr9gar<xNYq7-*yaRUcNmLpHAEgRd6n=+
zJN*D-h2Lg>_R<-53$+8i35VusnSi0jD5SHjsQFj1r-_#x=96Tp?+KEzjA+}5+jdM^
z(IJ~mXHdH^c`e1ylAnO(PO|FU9ADR=+t^7EpsM3$K&faVF1yObjcS>Tk=JX<9N}c~
zCFk)xZlUx!YqWkRxHD&5i295m&j4)h<GQp>0qw;BAvEx|_=S`US!#-|IACXO9;9N_
zsVJ-LpZwNbTT+Cp^)q59eVEJ{7UU@5pIe18Zw@1H^yg}-|2tsRH95*wX?=W3fKMv3
z5%w!@<+TJ}6b%%O6|on#98h;;wUo5nnQMXv!vcjgXDeuGeX);4{QhdLuBT<>ETtr&
z0C(L^MZ%8^U>2%OD#rz3ZEhGC8>;DLq~Fl8&`_|HwN}&>1H`M<5<xN=x(BY$qu+j;
z^$?$r^{W=i`vADw7cf1eY>W&w6;cDX&wqEQvq~ta|NGrLsSn_RySUkI?kY1b$(NcG
z0+O}0R%fQU*f}|%D0}}@9~+cMI9;XBNr(jwLs^5Mkm!1lWjBc(gyR$^KQ}LNdbc-x
zL%7q|&2)frPc87}_R$H@*2j7Vn|CVvJy9}>@RYTd0zWOG1IRBD1g4*+&!yr0=yAOw
zU)W%UJn-e^XGhK*WYyI`Umo#W*Lzr}-jF)>-%bZqRrIOAZ%Wr=BPKxpsz^fY9c@o1
zWb%h!AR_&;bC755WZr^$((63Fwk(zh663c%HXpI!xpNOOMQL3n30t}u@0a7>--iop
z_ok<VfOC(_KAZz{%TDD;U(%~n>w&Sx7ITzb39gYhh@glD=zoqHyZ>Eukb{$pT0exg
zK6G#$D9EWdq~xW8kXw;{$en?{nfY>UNm<|o`g^&3L73oZtY909Lgv4e6<eLxGlto?
zZ!B3FNeG1VbTo2x>Oz9rj`6$f3Iqf*ZdE+A0P|s(byvaY!j-pEwgBC37&I~xT(RU<
z{~r~@ET^AlAML^XH_k<G;s$tvDDl20slB)h2gmA4MwSu(8$k>cQbO#@k;SoDHfma$
zmfv+m6&nS5=p%UX{=2D?x&eZh<l&pJu6+2ToM3Bm5qg^fFI5QzyyF(hQfzK=8>L!I
zAS6~5^3rPhTY!U&TWt$er-OMI_$G9z{4ZEW1j2oIOe>9dvc)f?2#YJ4@mQ}NcGzMg
zsIr^!#XKSXrj1KRv*algc-D;1Mx!Vm;Ad;q_URd=NNCIV?EHs!|9O6C>mu4SPbQ}_
z?&CyBC?i>@ppAGD%(7+r;leeWwDX-3$cGULINberwC%CJyj`mD_h!+NNgp<1TM9uV
z@F}s{Cu34v7`h9JQk3j5w)P()5B|0ZJas(dFKe5VT%qqa-rL7#GTTuCI8{rCYaoLb
z0{``Wi!!>1EeI1+5552F+;mjI5*Hx88ql(_prff>f#Wy-`Cak9H+u%5DLkC907U)`
z=hE1}SGF%gvKw1o@H8lM4jAYd>51Cm-%W<n7SbKpZqR_c(sEo1c@@Q}F8#bKSUjOV
z%zXS)n+n!RC6&;O7tzb^UFY4b<dCj2sY+LJoOUwoY%Lu)Wt5~_;FXha2B9pXD9_7?
zV^I^^^t}z-4{ZT@XWjESz)F+^hBwJ8kLyEjVzWhprsV{FqO9-YSKRuZ7JkIgAjQYJ
z@%4J>k8^qTK#tce>E-fp30BoqR{biqHFLjP=(lt}auk1F73OaG^noh{=|l!zR$wGP
za{^zOzdnlI#r9J>ej2K#0;K#k9+DR+iGwh*wQVPO*Tid@^DO>{d>xlS0K#H&Z{h%m
zX!K@ghYG6jgfEAL)mX^V+QG)o%1QnsW7(W<)ACC9pM3`BCg!Qe9B#Ka&NeVM`Bg?d
zBNH0s7>N6@;+IgY%SPz5Q-9HHx<z4O6KHMqS-T3je6$|9361GQL@!kAM~8_zc6uw1
zzreF78<Gis&SKi=jza+M@0=Q2#K}bq%lvgDG<DUCwB(=4DJZ>CGZKxBHaXbqF3l^5
zB=0C5YV-+?KE3gJO7I8sULt+~oY`c(6J@=f)BV(O{q9}1r#~57SeNW51cIC`+I`|n
z?V44@Pey8INBS|jIMmy=Bd4bs7zK@fjbZmV8<K@NRc+z&-?o6y^7?1!li6#fY*t+*
zBA*q?^E;~Q?gJ=F^AUtFhaG6hi_Leimzbc!h$mMKJI-Dnt(|n{q904l&$4=Wa}vLt
zbv%8F#omp{tk;yoK+vR?r83Fl{LEJk$Vdf+a~vGpt=&Y`1&(&NbtZORn#%I_CP(5g
z#>vr%EgDc@4FK!)YBoW}e>ma{{<3zAMCgm$wKCtuvk)qBbew|x7%Tnf-T=YDP;C}u
zE81UxD#@GsY2eExzI*0W^`exORsE_iqHj^rlGc+~|Ie?cJ}<04x6D~r^NuCMVc}F&
z-#x&_ZUkykL0a!RnEq0fE?0uapczX;{JA{m$GtVuDq!pRn%Cu(n)3{+UQr=-zak95
zTJbOAyvAJ+eW59(R1~e%eef0kI`#qwV{dnV&*Z!LM!&4eFw3tZO9#&zQ%OHPBPork
zL!lzu4D{m@pXt4fXAufsn=gCgA3~cg7)i5rPDl!)l}GqGOTjv~48Ny;n)i;ANZ!1$
zXMK1k4FWGuM;wQOzyEvQbg31BjPHmk$zT^n<Kx4lql1G3b+jHw-7NZ38KrAc4R-57
zyqf6av=WDWYI3jr$egXbd?caAS&*!i?jtyHk(1wZaGmc_N>5x)%H1E!FDHQ*m70ow
z<Fr0+81wh+ok#OrPb^ydcgG@{so{y<(Q#QB1wcw!T3)_BIwLY6Ax>NG?KAV`+dGw9
zuW-CwAcXs<P&z#SFLc=oK0wbe<hk1)?^HtWTzi&B2pD{OpDiC~JblHyO!erqnSd@D
zilmul;bPvHGL4+kty8%OlvR?$CiX;46Tan6;onUWF|+anv6E`Cf}Ul>r5n~V8#({3
z0dx>SAG)caxu@Jc<Q&R+xF(fc*Ume7&LewGnJEZq8?Uzipr+pe_HgnK(*PT6xpgi|
z|C~>?-Zml|iQsX*)b71Y#!vMPJz<u!-d(0Wljlsr)wmR6`54ZD*W#J*y6==bRFuF=
zXlDKQ>(0x(7`w~eLUHZmIOgemjKWF}Ab6_dG0~A(C%)xl?kvCLPlseEMd9PLCKu`9
zuo*n6@37qNPK2@<u<p7@Dl*0TyWIAS1UBewTucEbC#mbi?R}0yH(<z6l-TPwt-sO8
zGjXO4NuGgZfv?Hq-Vghpv7xaJkB;XRX4Ogh7xY+6$Z}>~P<5JMLa&9SXs78<poVhj
z8U>aBlB8z5a8_$kOg{U%x!&FBeoWC=LQQXc)|YibS|)X{?I;pxyaBFq@xhaDrlRH+
zyDkS9LV5G>;sM_*Fl4eN-yzx8vYwvWmNSoz5;yL{M|aj@N>uZ0wdm%5w*F!m@xTby
z=s}=Dg9exFB?;KM{8MKCZLr`8+{RA<;o#z|I6P%4<5#rm`R~@6kqIK$$p`7Ge?L>Y
z8>H0Lrfgh@6|1mL>izuxFs-8hIw?lSl^!RxIEFrA+dx2&L&!;re@)t@d|R`NK^;S2
zhgkEpTdlY5*N#I!-A_}WZE@>_3JV;~hO_7%N99hpqyY!KE@KgLzQ#}kL_qn&ShvIw
z_a%vtbHz&8_-~T6uD}~>$1$QhdpBm&<9_Y6tgE`|?ZTk<Zr*bB^lMJWrI!_{ekw#t
zI|xroKcQAdq$dEau*@V~734oXe<6?}X?;$R`@)S92&e5|@W5jm(tjEIF%2$Gsb`6N
zrX`BsGQwZdS!rJjK*t8|7=?R^>U?!=9&coHDj-Z$RHh1GS0}q_KXf#6vhR<*tpejK
ztD>rt_~}@JU-n~-q2g7>7^=VA5N;;4#v~1OTW8D^Bvxgk$XbW$js)1L5dKtg*ocvl
zR33x(hfAD5(5Pgp_IxY2aU44~$#igYvDkjQ_CytzX*CPM&yoVDu<2q@fSr>gGxPFd
z=09Vy?MS}s6pDWE!UWOfGaL%T9pv4A7^_e4vx2}9I0@wWM0nehkh5s8q}sW(B?aHN
zA=^M+YZGjO>zyCeU$%9)18ouLcw)Y5Y}>W=bQJ$AEUm`|lagF7os@EYHB84fDSd@i
zBg_w%I1OwrylVn{Rlci7AN!V}i;pFp?j49Hgca4N=YG41FCk-jQSQ!$srUi4q{M2_
z=6paQ&&~i>!y!+i^LlBWDZ|IZ5gv%Di|an0YEC0iv)h;dxaQJk=dGtNrI!$!2H#6!
z<O!G*(k)Zb*O<HgxicL>(rG)(e~yIsc$R`U)3yx~%5wvD(UkT+9<d~7B{%r!>+Axq
zDqh|KFI&HIyPlC5o;U3-eU%8oKRL3hu4HBz`6KH`dVQFey(@K-tT)Q}X_QiLk|SM|
ze8Wmx6tX^YI#H8!39x^dOAm97DfsvOLBW!{q_w>HRkoCYQNPI^H1ZuxgXRyHF0>-|
zrPt3Ig%}F}^J><nuOo9#qOrAE$kx-3PECw@;Se19jx7((h-n*&t0#m6eOdEBe~1Nk
zN?7^OXK&;~EYC;nd;X?7$`ks~rj<VHS6-`sPy8~e24iY5L-sme=wt5nn4;LZgBTH)
ztpjTJ7XFm?f5Ly+`l4Z1^~$02UlY5uL7j;IuBrptf0KiR8CZq3-X|*YCRD5|&(qOE
zPeQxJNd}6|FcLID7T<P{iJlo-+HJ4LcJ@pxp;{S|^VVItXOnh+E+Zy>sO1aANg_TJ
z27!l`5W(;0*5qkl7kHb*yVkA0CS2|w#FMeNtdE$Gwf+OQ7#SHPWM`)3pyNG1o}+X$
zSh@l8*d<a=OZerHo?FVoTJw)8|LIM3Z8bNzv=O1FCK=BPghS-#-*X}>M<2f;Z%zL)
z{!b4T<4J;!nYClX7aKDfS;2P6V}MfG#_MQ(?e_6fOa-pi!^$2-bI_41X}9RVSD_li
zW^Vq9c9s7kvs2@eyoUMKz{MszAvYEWxfd|Npt-h(vPlFjQp2}bhnz(X?4>t}2*h`=
z&E!mN^a#leUrQHc_dPv?uapFmbRC@Up(urmuq@ZNf|iCV1h7$3ehBw!zXa);OS_fJ
zYeJuqIO)A`BlZBO#n1hi;}^3CPvu{cyYHpV)glyE%Q*JMWDuA0Fh^GykS(VZZa7ka
z&PV|Br4^%uSYvdMik{|X-v$&f)dsc}&jV!<k87(kGP5o-%cNoq#F~Cqc6N3k53d%^
z0Xau*&9J6z@^`|QhwECdX|-g`ZTlx_6bCp0s+QK^((b3!RY(rU(beHZj-&I1@QS38
zf>mBR&CsK5_3a5!L@xtEnJFgsOa9*g1FTHHS977Ol^d}};_#hOFBc!&H(i}n2m9Z@
z=o!EsxnwUJZv23>zOH;AK?Akh2_M=UsZkVEgo+;j+C7kGG{gBCfcXV2FRCmLb>9b0
z5)itSeelO)XsF#$<Hy8~9pihqzdjj*v$L=7uCA_bZf-6urRL)iVH~;0TTKOAVP#T(
z!swX%`vMFvI4Zx})BQZaPSu#_Xmq}bmS?jkiZIWw>&?60ZyzszK@C6t<|`)K>dxgy
z48&Yy_nOPdD)v#A6xh2w?NePg9i;ZdAo45K|8qepMdFF9g&&LFu=q(cC>17h*Ku*=
z;OKpP8Ne~QE(r}v)CuV41Fr(g^1~F&l|}w5&+RHYF5<i7(A}_cD1AXczRM@LK1c)(
zE5nLkv*WNpalLFn$;7gCh$5>5yIC&@zQkZ<4bC%y50<rzHIvdapY|uRV`5@%ZwE4J
z%h1l-1(x;lM0`}WLX#nrdjnk8+?St-&?wav^oG+AG6kmmYY1&m*P#Fxa1$IU(^kk?
zugj^XppD(hJvs99hTREKwqf6bZ7p&1QyatHh{X6!J(oJ>PEmgd6sgaqN6)_3^3EIv
zjhu)Dqu=eK<74y1E%E2+C7w{n_yI+MAry?(IFD;dGaTwAkUOjWq;`8OTC0E^IWmoc
zsSbCS@NhtrG<b`w!UF8H_8lt1|JHQlT@-^gG%%Btl=Sk_(a|}ajWBAd1})`q2?zZ~
zG%H+(iF^Ctxo*of<LgLhyiQEUt9}aakq*kwgdi~W)o>VxqDz+8QEYYcQt}^$e?;#)
zl1b<qD<|P<aKAx_@=ey(n_c7K=V-ab03C;B!ycbmRj{y){{dtJaEM4quWzrQK2@S4
zzj0lipsmjiysQ~N!3f2<zDekHtY|ybng`{}SNBN+7k~6T^*Qd_0Grw_REuC!kww-V
zA^P_ab=|d_@YQJdnHkjMA7wcI8O?#0#yh5D7vAS(W%ZF-mlNIASI15^x;i{B*4lOh
zCVKCK+ot8r@Br>eD+Yus&v)BiR7(giUh!X0S<eRs=Mg`#TVg1Lg{iGZqVHTe=2J{I
zqP~1geE2mX(`|)h$BD11)mg**tGyBAYzj;{Oz@t1A{e)VhPQN?#WD_sdwzNH^4@TB
zZ~H+-Oe{!F)i<HoU<)G53czY7F+(;ea<OTb=5;`$G6E#F%0B04qgSitRbv)49nPcU
zqDb|i^<U0fq)3S?N;gAAJDUI;kJl?Md1mpMg=&wCp09hB)RJ(*ov!^G8<XeP=KA_X
z*+^-$uyk(mR^=lKQJ3``>dBvs;m2Yy$z=0ckrZN|Pnz<^R8q8_bb16U;tr`H7YI4<
z1MTCf0(gqUx-`W~P*q(N@0cod{^gx|H$l=RvC9=r#_@AG%$qyf+PXZ<QiSm_adCo(
zHSWF{Z%RLlfG9_(y7=d>YD%;g$k{}!_p%h|FKo0m)s}WA=XjZVYeKK4KIN4dPk8=&
z9)H9nendY<P;(RUI&MbUNq&dR-!g3_|FaGv7Xn1O3J_&{WFl!~7dDnxzG`b~<>V)0
zhLqZuspJc~U++a>JwGSd7XGHEV__X1{-da(B;UF#8v1Y5W4iV-YOL}o1GIr8$7n|?
z<#s+)FllJo`Qd&;mBhSBe_I`N!4FBsKvSiGejg#uURq`2aB&_{OnSlIKEwW7+-E}b
zXBglJ&IGo+C(d)ieqo3iTpT7!*ubS?@0tIk3H27p@aSho%?)_jXX*;cN%Xx4h}|vq
zAP-k$f&@<?dw#M!7PLK=54AjVDchw_6u+2fl7qM~Uz6nZ%BEJ@e4Zk@1DAaj*!EH;
zCVTXsV+y{|1!X~=h9w9P-raU*3!4ovG675sB>V;C<>YisG^~u&;IYAQO{XU`(EuV2
z9<G|D?vRyI?Q>{4T(K=vQwTFz!$AZ{KIX;}oh^wv_o|>_#K%indh7fk<QeqxHhuNk
zU<L!`&f&D9QrhiRBMsP%`}FPi2ZyCx9)2H54MWW*aDV^-LsVB$U+`<^+9w|B0(^Tt
z%Vu|eYI&z86`(>>G@m#X-+`WKOGM+M5@wknT&~&8+i2zzVIEV6rscLp5vnln>x)8c
zTf0VzmLd#4^Z%Zp56#q=&yo1agV{?>uPSCV!mf2#ggBW)!@xW~Jw1APw+bxlW<Ie0
zomSUKD9RoAK+pQJmDws~mWoES3V6fPVspm^^w!V&$YbTV>e|RPV1i`=ZdxQlHE&3m
zwy5eqB7>TuwN30?hE|h~hvMF+5!@o*pqdr=I2JaSzZxbz(Ec+o-?XO-C8VdJprxb+
z4?{&sF)RAwAA0T%*uF!B!N&Wc#fR=}W8rF>kQ%QD)^meaR`sCnfZ%fjfMY&scl9lw
zbxf11+&tI{G-8t@<U7wrF31DN3GEY3d%*FtLM5M~<r~BfDypB)H-otbI?qoFxd|=O
zb$Vvzud0Tcn!0L2a^WfA7e7VWm;4ND{iyPLr(XZ$@aC4h?omfVUFEp3WT^Ei8-<3#
z%u7P3`1#FLb&&JMAklFvASV1D{*s9_VLG~|J$1%~L(mp=H@xrRw$pvdegK9CJTL#@
zSNzSOo5!WuXELkrwSiEn+sq_F+x*HWhcl--#`=c3??gJ)yrPU#DiIG<Hu%|BkqldA
z2^mS%nOz6Tc#Z`=Di@c~kw@Kx0oh{78NsJ>+Ee{v`}^O_og|GSfiU7}ho#8-B=&pH
z_d_j8d1+;FIT&{+qIkM1?^r4wjNQWbA1MXd2XUjd7$&QFwuc~Pu;qsu7hH`c`4Tvc
z2Op(fC10d7J3lP+DCqwAy7M;lfFj<}ux?H}b>Ey^+Ib^*?PP@Sy~QfH)*!aC!fF3j
zNq6;dX3Yzbk&rS4a7)f~j_Kk5)~GSQEasL;%3|S`I1i7DHDV;@7{hgvh2Ge)`$wRy
zts%GWi*4lP!_bjJk*z<txi)hORZveV^&t6=zn`c>DIP#m<Dij%PWDK$)aQctuK>A&
zzl0-MomUzAFKH}~CEULJkAo#Raak+th+bEvUncGQkJixuD<lU855g$>m2CsSbm1gD
zQ9Qti?~FyoW;D8X7JaJ_<sepP<f8{*Y}R5c3s@Upn0Q#O%&Ca0i+8=OfNT5DkNm#E
z7-^7A@jHGCe0(^93*x5*$8x&v|G9*YIhB!`U2Hvv$r->kW1^iTIk0NmP3BJv&@+ul
zOWbba;Uxk9#S%;iiI@fm?c;zzwu&@Yu%c2HrOVc`fN?a^pIr;j>uU&Z!+T<2PFyP9
z^1GGC@*Bk|+Jni#o#-}ZX57Ay+bfI6vu@x8318V5E?qov!j^f_QX3Re=$q2o^gHoq
z0%8fWiVH920(@>@zc?_woOo_w)^PaUlc-;UQwU&hU9Df~C25w88u3X`Q@yZ|@@k}0
zWaak}f{#*QDi6Y>`JSNrfKb6m_4?y@Cuc4_F$nF#XU(a;$7^!@@Dt|!O^-(f^Zu}f
z`ko74zHAaPI<Ox;#ns62;^5O*(^Xi(&$qBSar_7=-p*q^W5Q#_(Xn<U08NwVj;QS*
z0uO}EIzi!oIh!eaSQ@p{lrvg53B|*pc<=UHH>@@gvi&l@2eW)6TgRepH+6UMc7-B&
zE2LiW$>9o-3Pq36I~-=d@`KivRnxK&nLM@1=BV4s7b$c5uPe8{64R`q!8#bkewRKt
z-O5{CYV7>6Z#2%m`4S(^wHqRQwq-q~001~RkKmj3bt$uJC2fV>z(#m9E);|fKmn&W
zG6-|pIf_Xlh?|WW3hrn)<b+#bZfx2>KQ7%vDM#i*aC{@NOPu&W&QZOxar+#Ke9q$X
zW>;vszBc~Y1ZgG?sdkx-76<-i$T^faGEI$y(BR{NL$<X?Pv_z5XAh3+i&-K2e}Kz}
zwVS%Kx9cW0w&;JFHR!n=O^KlXuCozAR|`~9aUBl67UwHTYdD-(!!aL&8!U-+7WaQ)
z%YQOlVj**agTGUJ^5;2<jb@>%?+TdQX|uBRG7$Vg@A8S<S9N8nF);Gi=FmmMZ;kn@
zQc&)4%JQe8$2Ni=JlQGX0ce#4<A6Acgx7p;g_4?*t*DuP!3*?09V{rZZ8-E`>a|O_
zdRbpL1OM=ZSWP>K7|BB?Qxd6?=>2>T?!G3O6N%|CkdKvINTj%+)ZXcNcHD$0#Ruum
zwa2cgn7&WG8@B1WW(9MOJqIUM9ro&HSN6l~q4qBUDNUu^5UfK&--Pk;K%k(w`PN?g
z+a8wV1JePDyQ#IZv`bAb1RETeFb~0L(}(9StB#T^CvGBCyM?xcXslx3_p<&5<NfgM
z0jqCw8Y*x|!Zd1H^%uxQACtzzHd&YQefn^mrY7^H%}t(ey8h_LV_G=oKL^5L*B+ee
z>x7Eon354+O70OQ4t~iZI0KqlJx@e+@yrC389&|yEFC4Oe;2xX?a#Ua3R220w;GP+
zvut!DlI^hT0!z^nT9lLuIH3a%B!30{&k?WJ$X}ofXla~C*R4c&BAG>E;%|F=Cl(Yx
z{swSB2yo&`s7T1k$Mk$Z8Tv@Vh@@#ycakm7ZT9ewEDz=B$6$2J+5@~l4I-wH{zL9R
z74aA>8rF%_amHW8Ph#eIYQRg#k`YagW{8YxT{5=^r5r1+y=V>{VC9C2oJMG;{D|aO
zg0jl4ZAV(CrJlI%dTipPW7zaHvi@r;O{D*7&TSbd{BZ8!Ck=j$T~;A95x2mn#H3{^
zd{!Aq-OTC2q0tZP1_8fa$Z~Uh<C{FO{8>^*nx{aM3qyj8pJ1D;X<SSw{#obS17k-<
zx&;zW%PEFNgzwZR)XYS<wIC&V&P>I+G(5lXRYfW|JI)FvoVE@i$%oGIVGZiyB&3@H
zdBb##_cfZ8;-xw9&Mz83qJB$=@92l2GUO3ZW97x&5<$Uv0yw`v_q;?cQ70i1l(!Nk
z<>&^Ye)Y>bwx?H)M`;K8ha|P7k|OfFQ1*8@1_WE>LUWCYm4p9_)09*<Qqm@~7$q|`
zBrl1sJAY2%tLb~Y5Y*3|*23N`pjm3`8~wsNQ1DOtZ^za%KQCH~%e73ok?Dt^1-y<H
zMkjAx5L-iKBEWoJgY_TQy1SG$Td@DyDYFWR8uoz&InB^tn_8I4i`NtVa_@ifdG6lm
zxtgj*t212OZI{*+W!2;~j9ZiQNU(Z8zU4HA+Oo6#xZROuC_h><!S$5fd?ZfM1yXfa
zpgS-}yyh$$eFO#P;z^u(e0nd-e_)Gp!E7KRA@M6yB?B49M`L-ta$`S+NB&ah<Z=mD
zcdIxld7Rlp2r!@as?RiB|M+|@sepy^zlGhyxuPM+oXH)T5>-A9!8NVabj-K8lklo1
z1{QM$NG(EmkbLD+Hy`{nY7`o@ucW53(X|(5JgUA#=0#Ms&0z<~OPMDjXqu<>%!%-%
zsD3nS7>J!T^tkfFCQLU{T^Hg0p=LgywNGS=$hAZDda+(BYFSX+u9A_Qht2ZUU-S~+
zR1wofIg_eXStOC*v&*7l=XO|%=2nN9pYWC$8d&Mx371=lqov*+f8u+B2X;O3XReH!
ziDKu9BwZolD-~1Upe%x@r=sD%j`MBNX`Yf2U@vDTB9`PltTaWlxY7q#G|uzue{W>z
zHS=Ox9VTrAqi79Rh#`B1i{P1|<=uBIH|Z{4Y^IA#0?u6hlToDl<f0Hi;T+bi+&T4S
zx}knWrI^7tUjS$2g|2LarNHY{sP)J-9VecSL!Dmh?-S2ousF1cY88|-`Z9SmdCmGA
z5bu?$tG*e#{@xmT3KP7(7~HsIiRNh7#JK<}p{5$`|4-q^+)+Z(%tKcXk*d^JLP2q<
zu7*W2Y0HJJQug7M(TxG<weD}9)h3kcm<AyfSyiF%v(o3}IXKR45nT@@d`~ziBvA0@
zwQg75rkyoS?et7fQvWLiN}tM`*3?b`aJS71spxd99b)HY%rLaMYGL^Nuj+R=Qa}=4
zYZfdT=wHK|F?w;VA?~S$KQnv1Qp3GdpSfV$71%98l&^oy13G)Us+lj8)aNuD`JCoS
zgz4EB#{1<}qT-|rU{)@k$@#wh+!&lkv2zfPo$8Mym||s(ZB8la#hB$1WRElh$cOcz
zvS{{%DAqB)^FzNpE=Q+~ch1<DiIM$SCilr{P)(v?VZEJRC%VUriL+&t>vz}b*cSYa
zv!A{c5pJ2^{G_=#5>sJ5xcf($p4RE1f@bg6bv5s$Y$|blOq`*t-Vr;zon5YQCz+^k
zUc!6xZOsv=h#VD33(Y~mm~(y+5F%$7AMRC@kqhp@S&e+1g!7t3{ZCzob2nMkX8(}n
zzP%9(^s}CO#vp^(XY=2y^8}Q-Q^TR2o#r~^5!sI{y7BbX{RD)<>0joiXbHMOr<9M~
zCpb*ASeI!LI7oypJGrY`Sw@t;vW3b8_1&qJfPq7Q{mzx7jMRQK-KSv%T(;|<`L|o8
z66Ysn8q4dx-0H%E>JRZ*r^I(Er4+yPa4)ZZO@ra{4jCCzl~FQ^iO5BG?2a}8G0&Qw
zTyNeG@$O<!^<{j6Q_*P?`SGXXoa{eBsGAEBWS)IG=yhlIS*_I7vElzZY+0Hs&P%@G
zB!S%e`cPS4{4>*xmxEW2fH%?(NbHw2z3jX>tyH`N%i^*^>eVoG1+5GthmAI*Sawu`
z74L9rycQM}(=S__1CN{aGO!7)1ll-Wik{H+(F6vqKC`Q$+_?2)s#5d6yi2S?mb@(8
zjnwzPPgO3ES7;>R?i7S{WBn%=jl#sl&`ZiLOF3vpaY9c|(PNFw2tZR3RluHXb|YWb
z3KtjB(~LJo*eGd@?##N#5zlcobbKPvxhaM#e|g$F_wg%rcsI0?2+w`r`s7$Ny%8W$
zk<|!uk#v_&8ekS;Tf?XyPIUNRB+b9xu`-dWG^SfhigtT@v!uy>1f~6nB6u`M%N4vp
zHfqh-JKf1z5Q#B(2jIRQL(G)R=*e^C!L51J^80_)B!z9Xy6!fg4ewcX`RGBG)V{Uz
z`&KVMJlXyzS(@rTK2FJo+-eDD(_k#*(j-_&%{0`a#bk~^@c1tdf4TL=bimT$!(9!t
zc6z8Y{r!A_?yHu66ALnn0D08QoYX^Lx+9TH+Q*@!udWX^Ca_h0j`gp^*nBIol$fDN
z*6Gg3Nzp~i*?twgw@BC+xz_vg5Q5^3Anr*r;XZ$_aM^H9yI!Z)lp?B{Rd@@jB%vQ4
z*qXBvU=Vie58P#28CxQe2sn$gTIkhl8Zks~cv6p3$`4No`ie&T=jtyk@o+v6lr2FE
z&_#|&$Q+%p18g!}_HCN__R&>Zm__<JXr@QE-#=GYiC7pp)d6i4332s6_XisXv|)!A
zX+mf`!?z-=L{{pY=>GCrN<p*{u&lz>iR|WyRv9;TNoec6>cvX;w=Iq`2PzR5s`SH8
z|DddDYsA2G7oB*<tEIKIwf3(@BJTM=6Bix%(Dd9yAj|B)=1cr~qRb@4uYjQ;D)M`G
zWjQ7N&Bxbk2YokKH?M*gL9YF}$~!qtXGm6DpHV|5WBce_kzqQ^E$lQk5tqKX^Ry3F
zGg;g+uQmN#lIXvqsb-R`+V1Xx)mTasRK80C8(fDtylEeS$)<zK{+?Q$-WXe^mM&k#
z5viTs0PN6JEj_N`#5Iqzee{j83R|z<CnJ;7?D~boe{utpGbxsN=~Ly=L4Z^H{3G1Q
z^92<UaD!`;At$horZOfaX-0>Kv4sIs=J`n#h6t0e+&N-o)iifP&>?Z+*X_(VFX?_&
zAMsNS3PEzB`lm13*~>uMl3ptP*qQ@E-4tjBH2sa8<jLG#U)(1JUnIpgvjyRX4x9vk
z=W}4%qmWmi+_(E^8RJl2*lq3hZ??}wlg_U!wfn=tVD;RJe-;_!kbbd%cJ;7mucGI;
z+#gIH4c>B`Ny0kQk=jyQ3wr2)K_za1y|4|?wRffC#Cw2gxOTEOwSm^PFiaudcRhOr
z5Ht}`%>*H+otfTFjM07vx{c0E-Ua-PzTFAmIQm>xOJ74xRD6@EDtiP;78A^%bxWE+
zBb3uW*)@>H2pcaFf(_lRjlN}E#70}9jv=|!an2-QZQC0iL7_QE*gJh8MCFjOm$1dF
zp-YTfHdpFOc0$FO9?!Cx1PHc#j8uYv^pL2yA8wlIvb!snF-10()m~v`{em2d8=}k)
z#TLWQAVS9IdJU?#PpG$0nPA3kv-{eAGNWnCO%;6$zQq3#da$IkD)`!ie{eU8`LbIl
zm%sJ7a<)#MOp<x$*z)v3L?I}NSNKy5U!?IBuqq<EiBY8O!H2@QiXs0HOuzzm9akG?
zz^Hu5qwwlKW<M(*C^#aQ=C^KT+&us6Jz7beBZ*of+3PkM1ovFC*)P9~O_Gp`Kt}JG
zHrU4RT(7pjM6j=Ji+m4#E{)he+{hj{cyB4dKQQu;=f`O9KN)=x*u~9?JQsEAo7q^W
z`-HUMN|LBtKPQ;^ukDALEHWT9LXhxlWxF7WzZHC>+|p09Y{4gK%X!MzD+V(%>Y;sU
z-z>%(F3ciw>-W|QY&KyY7zPafTc(~fUi;;^ar5Am{}%hM98COd9OO_U+4s|Acx<?e
z{fL#gX)+0{8054)efiJV>c)-@F5#7{2qk@-RqM`?nm^*O_F7%<*5m*lYa^c`0FU-n
zWVMk*i)ovXZ2o&`x}krtap``_T=QK#8#5Cr9xb%@3SqfIE&F*XHI1IqfvEyIS~GOp
zp4((Q>;Rsh`(a(;*{Q0pz3EV8B*awZfeJ|ymKkkC1!`s80C6?L039Vo@m|Uddu`j%
z$Cq3U!aRe=##_{WqZ&XN?V%o2%qXwYA+LQ}Pf<q7`J)X-dC5oSpn04;?1?6$KS0`(
z^0Ri+f&PzTBz)_EU+-3JCpyxgystBVKgtPM=Dgig)s7aPL(eC?Y?FnF^k6{5Ll{t0
zbK=7hKxnnpa&~{t|0KmjbCOilK6cI$7IElz&93_CVxU=CuYj_b?6C$cDvZ&nv|xJB
zU4_ztBiOFJ;9v4$ozqpUVuav`e?L~kMVe&-lac66r<=N5g2mn9&@Qj!btS%$e~#+H
zhqvwwOtJE_LrK>$3uQg3pEkDsL@KZpg`*prd{DjN|4a$RsvoMMNA?~SKFe~ad4%y9
zdE&JFPrQ8q*o34J(>|-||BG2d0*3qlFo7;7Wx31BB#5!Ylz0`8zKy`xqU$(8+-jGw
zWRkG|*=5iXr4{{pJ3Lec+6tG}gEdXnNOypvO9#)_!_Oj2Mk%Yrpx(Qu)52_Y7PRC?
z<b+0=XjW&M_uEHeYjHWF$}(RJW1pI=!@dA>ZeI^*sCJ$!7cqv}e|Vl2eG5GOf5e?-
zR~&8Br6IUBuEB!@3GVI$CqS^^7Tg`Wa0u@1?(VKZf;++8U4u?P@2r_M|6usQSE{-f
zb)VX2?`tDalNb{(A{ZjIa{m0u`|$RMpC@ZxO*QGA{xa#zg41Eaj*hf?%072b56yT8
zfwffuoBi-3;}4Wj+t<u9J4QtS^U&oiuQuzi5XE;(|3n5kBO+Lx%grJb8dp+Yo11c>
z*rP<BCBw|g3kqNhZi2}~@|iE4BzB9Of@ZbVc*XLmtAoO^y6E2>?FcRmL8OcQl1gFU
zTh_#JM*`hDzjC0ux?+M8XW8D;UpNXlL57NNWPC2?We4#G7LqayMCWHu9W1<a^AKq^
zo%G-4)4v^AO8E_C0lnNUF*z-paI5H@Yn>wdObTa8M<c+xrBUF-l;smYC7+EKohul?
zM?c{YmPdPzmA5aA4(_FX@LkH#cK&t$0g7}#CUczA6t4P%-3n33FF<bJt1wtqIA(_L
zgec`3>m^ch7xRQX>VKi1yl@Gq^S{dNp{ZtzVOHimTGUwDX118AFhp(2YDdfEr^pfH
zmoLvsPJpq!9t-WAX%6i-_B?$b3RLT#BgO77{wuSAc}*53W*^u<NqD(L{UUZ6F1ALG
zFb{vsWO`E~qj2mv?L$nBmXeIb4##SgNz>G(c>Gtmn^{<wmNnHAZunM*_A=I?w*;+N
z$Rqggo}HOjQNg;rmq+Ye;Q`oFXUb7H&%a4xFo8YOLMnZaGgMk8Sn0g235Ng_#jLmK
zYLsJ71Osol)J*onrKLai8z&3JoCO{~H2`HCV(qZ9tfo$qxF0jD;rXN;A={qh4c2tt
zBzvIZY{-a$?0B9_x!$!MPO3voBNYgR!|6PJYWHu*B%v|)<S+Rpd5CqD#IBd!Um}wA
zAwaqr-cu>&NlX$!j|dK9DY!@-4czyIXGr#!S+Nk{KhO~7rBNQ}Up{9f@vfw^)cOnw
z>#DE&r_OwMAfenzdt-aKV)AklEbKg0D;H*K_Mk=LL}ZdIj85T!93)9@`P&5;kAtbT
zz2{BA$8{0-N*Dg0zKeg4$PV}w9AU7R2x!X};zUPe!5F@#=mj5p*jcsZ$Q%lY-J}s(
zhv*DZX1nugl4`_m7U%`=s$9;AP;=Xg?x{Ap_05)3ToMQtkN8CeU%zmF9d-?6*G9N^
zWMi*Td_7bellekTl^j|Tw(zw+{L4d@FQ}^tMbRSH>JN=4kz73%<@CpIilM2e#6Z{(
zLjx16a@Y?ES<#D%aJ#wmvUNDr6|<wAm)-yJyQV7V<>nCcWB;cl;?krY8$E|!+kTn~
zN{;LSzkb>=8@p9d*f^cVseX}%!X0NK1@Tvwf9(H=#$>nIEI7rzs)xBb#k1`hXnuR9
zmm{-b<@t-qTNR+$7Y8N8&ksWs6Av6FhVFvAeTN+sbW)J`&KAq{>Z*X5yj4Yg6FDah
z#(;W}Cjzwxp=;&a?bLTMS$K3&B#(L31>0ynlf_%+uAzvpOyiLh-*A?zsa|dMxOmIx
zMn_xH!y<kjG^roT8D{ZnN6^=mKg=F{_E10TryECk`-hYwE`+eLB56#53w$Mok5fp|
zik!bBGUoZ-SoCY>^>FNOt`glj%8k8dEi5|2(`pUtf(q7w{|;L@m874gtY1tV3Z|og
z)xj|d0gRCQQi^A^bN)sru|@fq2kt$FO6n4GpQv2Rlg<!1jQ@WL@zC0}v*Ihe69fl=
za>!n3ZQP!X?fi&D`j0UJ&}$QZcv01ttN&ecKUQrahz}~}N^N0@Ju273&zO5bCH23U
z+;7QvWAYG_e2O0d4x$N{)O1}JB|(1zQcl~pl(JbmP<SdH@f0oF{-zw#i?3>&fp&)H
zUg8kYvpJgmIPM2duao=;f6V`MC~5s9xIdCvZZ%LA9886m?Q-o2B&khtyg;wf!JtlK
zjVt~_>jcJ+8`?M~ZC})}vPHMGO-eE2b29b`PpsBs&eZ)GDSOWww1`AC=g*rES5RkA
z7e(&iypAE8to&=EZ2Jss_eCMlZI@wtTD=@%CSh}hLw8#hMcoV?L?x|hGIl@DmdW%a
zqM0(a+x~FTs!0fZ6l_rnj0ZEN3@TIlRQFN%s6OjN=g-RNNGB)71Wnen&%)SnN|LvE
zdcHqV=jVqRoz5|2V)lO?w-%MUgs>VUaia({c{hBp;uEx#m*)i8o-scY_L_2??WNW#
zZK^#|NXCet>tTQAq{pr*vii9bH5v$rsTchYMo5$07=QydLRzd&Lv9z;A8%6pNl1~<
zu$yZV{L>~Z!sQfOQZuur{{dic1YGRrw!idsJ*Eh6%Z;GXGu_1P-fm)2>hMjWyPYB~
zg4*0GYRx9UkG{G0$Nsl1Q_^l#x`8K7kEz3M74}l7(l>oYF*67*krle<kJ2>VBh1pf
zHNC<D+PT0*+2}9Om1C>DHbi`9c_6jiZZuG*nkXk>4=NYI%Gn>I&4@F((`h-ls4LCl
zE5NUUFD2$jRFoMUprR@Eoa@<+%U|h~R(7DOMm^49?!$UTtczuRRxe>*{a)B{JfN7d
zHEVyWKEv7t+jop_A$cblkBDG^idAWeGP@&2>JISevCOHj(DsQ+*Ufa{eDR^plYmcB
zi^^!I=>#*AQ$l~)k=&~>9IWUrudIcC>u>){V8I~s>#{<2YnM1!=?`^Pp1J2T2|am_
zi+;AlwGnj+T21*->7b{Q*d%q+drVhqW&E?CNxG`5p`?$_#qe!A(3RaQLqhG_Gcj?R
zatvr+qOt}vatpZh4nA8rJ1R;#z0XOjEtS+a=JHT|8CWENP12@+mS{~a3H{DT6|7?k
zJL@+5g`?g5u1Lv1P0Y+eQrF3RV|e!kYnTFZnH<8r*Rdg)sq3Oh6r^16Wt@<ZbD(`;
zaqZdo&}E_|BrR1#;=ig9vswlyi^ZfT0$GlW6N#z}wH(aIcoe+jss|qheO|=#^omUO
znYs7yXc71!I0km|b4cQK-h`!ZsC<O4j|ekA#=okye~tV!t2z3@Y$(L{gmX{pYSLSm
zIlzgQ`SZZIgj}0bA-fSz8a{-d5m8oE+gLyJ>RYpva0>aClE9h>nTWb;TujCXObR80
zy3PzrtE29GI7?l5sLb%xc`f#i9|Rx{qOw9aYf>#>1tn7{8ocW|Eh3(lo*0{Cg#R|G
zs`&#Fmz4W<H^2*V3}e>*8a}sa@$)n>a&<V$Mllvt!1b`W+5(emOk_<h(T3s*=}BHA
zErg~HqGz=ozCB9Ha;ce-rUw92wLY8a3?p*L2><xiHla2?QlSxU-6!H<+H)|Q=r@&C
zo&?Yu+S4~yR8^u(D`x0EvIx#CYi1?KV(fvm$kML+7$X`I3MBX!`yPOpny(gVeCyG}
z!wi)+N5OKm>*En<B9^9xM#F;Asx}^PYihggh;CJ4$l*H1Qp`2)bd-QxR@AxSc^@hg
zge5cCx`lq()fHIBe4q#!;LQp(nbH=!iIIpa^xQ*=JDEx?&i#>bP~yec#^<T+X1B&7
zTC>7<XxU+yc-;7#DjJAUX5B*hk>aPoOf4%ke58tn3mh)!1~hwkDn;0<yl|-?Y4RIC
z$N7KwJXAYzv-S6gvVh(~>EObllb(QU{elMGvUrQO<#fwBYC>mCCrN_l50%-ezevwD
zDO#12?*le*E&^^(mtah;<HqS)iHHQJjp>^z=;|=~6P<Q=iduK*)F5zr7RL6TH@bl1
zM>Tr@{WP$;!98%MBUEO1nR;%M2|GS9WSBBwh-4m1$NK@?0TkzQ+^7e>>zZ3Zhyhru
zw-T`B>yd+maM^KwI{{1nij1#2H<UAo%o7SIH;ydR2}ZQKjez4SmnD}4Q#kS4RZym2
ziO4CIt=@6=Z0B9j=gq2s7KYH8>R;Pv3&S2bia)yWW9NNw4Ic%B7iJ@r8OYiw%SGs#
ztG)+7iQiD>1KY9umJVlz@~$v#R|cMMFt!HBh8+@cJ&K1arbv*pao$dekr*z!&Tpr`
zg>tt^@{kN=$H%6z29wl&_P2<7Mxx1{4U(D&$aHF!m973ODO=qGvK-UJ5!rxsE;xFv
zYgJe0k`H;+1s~N#tzc2)ecOuaKwZG%fFyHvb(x<9U_(kQ?G;zmNJz!_H+K`(niWUD
z@Gl7$Z^S|CW7}9*V!VbT3bGD0UPFUhZu=tsRvk>sutiNHCGF>J)|P|FySd#!URP{8
zy;m&33A)F`$sc!KTC0&euVc}!+)3REktp34-so)lu9ce?l*R?b-1#TEber((q|lfS
zN8;wzfGzxh*&-?`dbHSmFC!%b2glS(P=-g~USWCDaeeK(<!%RpqVCnN5v%?(J8vn@
zyS!}AcFo?#HTePt<b4%&lLofviY+{>WECDuqXSHgc$M|(wH1)LV==3G{QZ7#+}I;a
z_ln{|uqV;;phs#c;-bhXra8_Fy?MsVwwz}TfQ*wJdr40vw7RiV2SJIDQ%%d3CDh2b
z5>mUdvN&BLhyOX%n4a>tN}|-q=R|{O%8`*uz8dJVl+E0W8p-CgQjcDI`Wy?ojb+pC
zHp4a#<W)7;`7kxGiDa@f#wrS(0pB`M+qnRMBbwARVN_Ohe0YKuIa!N>(lK`AUt=fW
z10#O9qy$+u-8*y4t0(OicOio9XSe*fR2pBgKAhg}K3cWvVL;g(-X5x3kSWeuo8`d$
z;8juT_`m$2H!pM&gtazo-TNM9nofO5)q{fApx5yb!(XlHBs<2M#1h{@{f%F8!Ohj$
z#3#L-{TUGn5>_e=eSf>jo`BMG?N9Ob%{-;8`U%F|(?yA!=pez4ozIY+N=slNWOi3f
z*W&jQijo=oY)1$Ik3k=Ry}gY}+8faaW-K72lMdZ2DPmJz2j~gGx%Nx0iY@K0h#l+&
z>f%LYKWgi}k8R|g69##^yFZ7_@2`nU)TXDZLV|cI5QT5}I<6lLsD?6;t+2CnH1f7~
zHgW%1Q(oA_D9lbR^mOFo7UE>?tfm2cJKu=>wwV(f4jAz&j-9iYH9HMWUneBR;RvH}
z_Wr&&3jr$>78)BJNbTn^S!dV@B7(CrR2QZqNgq`Vl<^48k5Q!NTogYFkyDKQ=?V#s
zuwNlM)Z9MZ$PAGM;zlr|1*&>nP#9F=2IyffJ_Rv=KJ9`mkBBqdPfxebK<b`<h}wc#
z(~%LLSiK1H{K?k*(*HE5n*?pn?9FF_Ub*velsUWZKeZxb){8&h?${Js3_A)l;kKY8
zdLOvG&l=md+rWNO`;UP_>1gRG^*xHTt9Kgd$YT#JGhh~IQQgqePm8~%yuIoqzCS?g
z#ipGI!1mXJp^&Sf=t20TjvA$CpoU7@Ugnt7u@uBEZTRE$HPXI<jRhJGCM2f7pD$t3
zscrA(6QvMmdr4WTl|2@_+s5LWihkNNAJ^E%%G&g(#QP=mXVLq4qTyGN)aMd2Pb!v1
zGiyuu#JqPvVBuQ|4r;Tnh#@VWb{xsZv(kI=3VPU1*>UQRHrI=<&a!4;Ed{y7P1r9<
z7j~s%UL@ErJ^j;vPYdwo<V{3m&D?dkD@jgMRLNp~@WN_@jrXfYPiNF}>Xq)Je$Pux
z2@V3CM{%^-4~{dCQwrr{EYI)?@OC%wgxk7WRE^02kv{teM@GhmIWQ9fh$kh(--~}d
zl10eMtc@}8bb_(sKE~G8PXBIZCTrI^Vi@Vs@|MlJ_%7Tts<0WWe6fYI)DGf;uHFE7
zq-sTx5Q=_xj_Mz*sG-6+-LVDfYQ$`sub>s?Y|ZR7-jYDSJ;f)KVP<&mxm!8NmNd^U
z8xtV|SRX1HHkbwdp~1vRB*mc}PpmuPip^v~;d0C+T2(6su2I~J_<kkmuA-BNBz)v(
za9N><Zx^ri<)&(4MEl&?rd;FOx2wS?b$J*4S85cb-{I-8UYHm{?PAQ)vm>Gs?Jk7&
z`pv)SJs`=AmFp~A=-2`Wd1rRIXp)~PFs878D$_N1^UXH)R?J7DA8xZPCslO+QWw=x
zVr0UBkR~F<Lql3<3RWaSIBHIm?u40GRlKIfi$_7PE57EA+G0?tQb2sL>A5+rU&+MT
zsE0TS8mh;;v^DMv&85mZ`tOhdxK_=>QrEK=X1^i&r<_v$&%D%+t8mntLbNSb)GS;;
zet`uBMNGe)wd;rFTD4!!zH497PsCG|Nt4!09<8*RdkMV{egw=M(_nBtBj7w1#I&QE
zZSZmjFvzOToSGRJ86NB|GgPBjc_45Xpdl6KAi%@JBZLG8htv+d1|IZ+1gn>|i4IGQ
z7QA9!LKCOe`iStL9`~z#tH&d5+l)$pD0vL!H@BXC9@6@OFnfI7RbB`uE+mvPn@tcP
zbhOYvP;bm9{8%w5zAkEtIlN8z;as<wRXDemtZhQ15yg(_go{Ufy)`BxxKzvJ0MQ;?
z(aOSl7}Ys~X{x1?^{!`)m?<IgvcHXugPSaNO`wB68A36ARmwX!KfXrpLwpEaoSYpU
zTURiSD*g&b^mae^jmr>^t&Q-|d)6+))9rI|%&hs0FYXI^Ww^AXyjadw!Ijj2+IjWa
ztmPaffpYoQU`;{a3ZkhAPQtJLF=XNpxwf<Pckrj3oC2%8(94Jq#xxO`a6zbv!_-}4
z#mXjX=lMFvIC%>*MGyWb*!3oWeMSON|GhDdQZW&4!E6uoac+vw!t%`h1}wr;1wztm
zg6LlF$(_QWA;+_~p(}IYb`{es)2sun*v~ieVsxJ{V{_wVV(Aj#kw!FeC<0A}Q#`T`
zJhE`}w*HZ9oODz@cP+R7z4#lya^Tke<`uQG40&|&U7Y-|%Czd#(>((!YdXK4=8Fh}
z@IzG1F3t5qlLH9BukZNQ@!d4w1|iz}r0~`@I<nj99P+oTJER0J!qEZbD_5228DUGq
zokPdLgN4@`+OMe)OrCCk2@yKqKaxi3W#%k_jvb$)igAaj3hZ4`G@+MxA7a!JNf*Tp
z_l(7;rtsGnkGw{ny;z|72FYeXF8q-8*G5x%U>LNF?;pAn%Ok>6ZB2gN>n;)^7@s{k
zdS7J0gc-#s^n2$94z=WNze?SzjAmX3d3`xT4x6RaBMUkiL{|za#rIRe5RB}_ds~ah
z$5UGF%yBitAgXJ!jdamwtDI<(pR*2#{@xRh14s2$S&I}4`Pl%q`F^4rrwh7#PiA;*
zu3s0#$Wb9dm+IEYZ;=Pmy-+p_K!43XU8z9P8(Mhy3O>Nn9>c_Vr#yXX-AN0zJeuTE
z=+(pG&rWHG`<+WggU5wFVUWQ_PpVnXop;-^p%C3yZ%XZ9q11o=qy49ORdr<<d8dA8
zNTy#>ntuf1do(~&{Qji^vpa&O>-^66{icZCUt}^e!+tRuJK9naTXl-9&LJSEhf3sj
z<G~ffKv1wfr=biwwIV&+4|(Y=D{A5TbRcvk>wpmrf;}p{M-p>GvCn|degE&{{yBgS
zu|?py0zESA&}%axLW)Iz<8Q1b(VMW5l8<a0^$q+8K<490uS!pIRat8>U=Mo)g1i?s
zbE`p}J%m7K9aKbcnu1QTKWs~J)MR232t9HOx%oa4vwECe_}1ddzsRhe5Og~YCBXVl
zVJ(BOqiq*`8?9=XYxG4|mS#p?r3?n*0}~=L`syslQbRal{FF=uHUHe$3sg}#iO9z!
zXQoC0$hc1n-HfO&;yIYJf=)LUDi`!pn^t*+EJzkcr#lCa|CaSOkr^2)^Q;_a&R40K
z-@S6Dkt~)WCNACnP#Y4Kvvjf(o60XEU&BG|ds`;C)s(&cy?rte^OU@AilFt{b%^T<
zfZrwQ{k`F0IX`R)7lbM5`@lW`Qwd`v>&XQiCdJQl!27g?D9zT#K3Ccis5#_>>cHh!
zh_KXBo>)wAZW51<{Fd*=f2-Wdy-ovotYJBx(N{LYn2+@18A<KF$e`ZWaet4)3M!LD
z_Rs!ATImHmRir1sze?ABO@-)*+MZHH_>0cJHX<K55)V2fqM?d*+O8=p40<dcHp&3I
z8>&YIFihw>$G2ztr{Mfr0T3}&f*I!KYfcKuzPi6Nvs#PXFa^JZ%1CP>fkiT|1Sv#I
zqfRg~x(b&T*_QNHINQ#CPpn3`E@eK{meRk{-{gP$4O+eGe(`@Buvu$!c}8b3u1CoB
zO+%Z0=|1$>QM5Pt;V4C#l6OXur45LDXT{$?=sETpv7*a|)5ypr(hP!&Yy*C_oIXGL
zI6hp>B03H+pg894r@hBS=r6o9``HtG<|ScR0?kbEo9&#!Er^7l=<s>MOVOC&DPt8v
z6eB91Gqgl%k5pz#+qPu_I%M|TPkL{AORQ_<9%|a}|NfXfuQUv&SG`M<L<7h~rkvd<
z$REd+lrIW*a1!TPU_mO)WCETUF;B<iE3%I|+^o-}rvrZb&iYQKqEcdNG2KJ0s?v>n
z11*PBXn!NgN!2HMH=bgjPM4tYozPgp6YuR&(#(>%oq;LKB8ugv29~B5`!K{X8coj5
z4-`a}H$6=@?lslLjd)8lG=Koud3J3=LOd)Qoor9{o*5hDcHXDYDc@@|04mS!l6Ofk
z{k_`x!Sik-8A<wm!SL)z5H?VGcdj9V@$^TBvXRs6953j7^K$5lS;R<Hbz^lM4#W+u
zlA1NqRP@x$Tk3Gw6_f`H8C2gk8hu|?H5dCu=AH7<TJ8Vb`yyVv18e2dWY4;E8(b7Q
z`%b$pjrk|@=^)D9I4N=3gem_MA6KyF2kXNI4k*(ld?hOJ^HuwLnEacMa|9Q40Fy02
ziuv-$k=CONIv0E>N}KBf1pV<n;)LI->~J>9xe$W(^MzxN_8kU)-Rn0iX7*JHKPXM*
zg`1k}EUe5gZw7t>CaA8uc4d%op?4wyw)ah0cxG!Z9^+?;sE<#;a~jORv&3FQ&+Itj
zRiww2pN91<=!25A=Hsr{JJV{8q?+%QDE%a4xmR9c?7j*wYh$|(idL8iOiu_9r^?^K
zR-LVZb%ogSuhPeMTS9Ym*3S<rqe(t|XV`VOKYYJUAJYIH$7`rh3wdXCGt)1pF7obJ
z4;P0YNI(R)w{+WBGLkUakFVbxM@8J$&ZG=X#LN`ItSRYs7e{9&CuIL|kdID&RxvS;
zdL=&HPInR~XDE{J7kDa*g~nsrw7l6NC^6;1VP>xhgiMnYT8#p4oAff)Dl1_d&j@}m
z@Mj~k^1%225+4^|umRE>Efl@JzR5nW24}-VlrO?Bq^T^jBb)LN<fVv$CFzIP>NY;@
zbm#yIgXIw9*A1FCxA>wz<@#p5<Wis5LMiStMN!A*SOnM<#D@@3nnubkPsjva?!)x$
z<*!egap&(8Y%)7xsQUZG#RCFhUSACwPRGj=9)Xk?9X<RrMFPvwuVL3tEHg3N?=VT1
z!ce<{q#ddq;+z#jX(;hxJ{j`!d!IH3To2sdBugd)cpfAQc5grWi}a0*Ju$XvscRxP
zK|f2=6i>l?hdC2zKO3AyeAL=bDD7(VFF<5wU?P0WZYp@S%Gp$9YsJ7wUPMjF&eHz@
zM+0<8Uc%JgsZ<6~ZY+rRbaZZG$E%3ydvJZ}ePxr?ei~v_)NdieAWBZtSYKU0qUM@%
zGx0=@NJYK9bLLb}-tF>!+f~WANOa;V3W3B~L}%gW{=ARY4`}qD8j*X-k&$ofgo(8O
z6_ZA4OnEy4FuzS&i+&|nt-lOs_(`1E;0AQMqO`5dI?7rSiE;@G2nj|L)*M@uF%HW|
zAfa8t2XJlfsX~s+*x(4Li{gs?ccGXYdV=hm=n3ukQiOKK)XARp^{f7{XP)lTAK|SV
zp)y<R93Vy|q;{vLRWu+;MR|=Y(8EKjGQ+KD{=-pI$)>MW6*_HXBkm_ykwy5QzJT2`
zciOBz3<gN}>m+kd0@{#{+>8M>dd_bsGt4eI0rxo3PcfwyypkYGA{}1c`*yM`4r=11
z^?(R?Qqur>LX(P`{3afhF#=s^cBv`pEV5rdlM^<;bpP?Be}?qL)RpJsALQG->*RmA
zN!QjUqxbDI+$n|nSWDsfvOi7LU4UrqzwPc%f}x6|o;n+XO`#JB-4#%^J#SS_zuBhh
z{WeU-Zs$xSat5yK7OCGpSBz8UjByy64;}JroJJkfeMN>ri_>R$RCqj`wh#+-<C6k^
zZKO&%5P%^aEbRDFWkH<sLt;e3=Owi1hS2)+qz@+cuGa{uwOr^;{fuwZxG*4SbSn)t
zz+XZ2){pO=4m&m<U`x;V9kNU*u!Z3~K`*dGpe3;uhn@gIM8yCKZ2lZ4U!uCB^uWRP
zBtP&{-+wzQT@MvSZSUOxoyFA*#nfX8ang5RiIYhv)ltfO@Dww+uAH70dKfuPlsb}*
zn@Y|0^lwl|NHaE!K+JD!04r7`2o+5bn;5@3=MX$7PO~7gM)`%hye6;K>L*bcNl?(y
z6-5rRj&2Mo8F+DMe*Gh=CXy5DH_E-|Zr$E0Xqt5dj17NE()zD3Lid+_^whkBNY>TO
zU(C#0ZNXbU2qrjc9L~3q@?*X2H+Q}{(!*82!PG6*X?A1F<J;>lU?A<YoIX}S8-2df
zvF|#u(5I=PpePoD{fPUv);1s<)c9HFw&`4+OhEhmWZ!;9Mh+cOF*ZInHZBGO8ygc7
z8#_KGCdPl7(vt?y^-GI;V$k*?SCca=NqWoNVo*)&EMw@$6d9%-hysTz*-F?GvfRH6
zc}T0O$l`o*dEdkXE<w-~Z_{!EQE(XV;Jog}vP<>rcm(d%Q+bzNNnXvA!v{!e2@AVt
zoEI`km?;Vwwz%?}7+8Du9XIszCjVJ{IqwrpXKVqJWa>)se;yZm(E0lPdIu}UHQ@hv
z31W1FG=60C{An+W8v|2(+3z#~y1!g0L1%G$LrlWLeyF{b72q^t>&h`X53FsFcYxst
zExe~E-``G-IkmYt4i2w}nVA_;VJ9I9R)^96SxfCU`<JLHy=D<M&_@eDl$(!;Boab`
zoDf-B3o(gWgzdt><+O%gZRE0cS{3shhQ3f34(G~Zk6;_WIE0UQ6h)631~X-Ls9_J%
z9RU3A@&Hjv7}LuPffu7VM|$NWGw|rd)r+K?lHG=mAH^2yWZaCNXm0>T`Uu=qihfFV
zW@_4LDFRlaTNN*VhLDA2NqMO^2B(R{%fkuQI}XI#UQs19T9V!)Prw}E$_<rTU*lUo
z!9|nP+c!QqGVyz~Z){*-tZ#Jm_rwIy%g~^NC3GV_&RB7lk)d1tqx+S7`?DrqiIfC9
zUXtiHZkV{X^WeIU2J<byE`#60b0TtIku`@|mE*Cj<xh;zx~afh{mF&PwOvh5*J3qN
zdp=tYzoI%CT1ucm{Oly0nr?8(X$AgIhCA3?QTT%7qZhD2!P3INlw`-<Gc_IoJVBEi
z(w2{7Fyfi93w0HNA#cg2@~C$W^9ks(XOPKEtaRJf`cgkSsXLqKw1mMc9$L!Sx1eYn
z%&KBY0Sy)+wI$Wl69PkvQW!dDQAhUw9;%$k=(Hc{UtuO^;n^O!nX7ndR=>>W*YrRT
z^J5<#6Bs@seQf2|&gLmAhV}uDHibX2;QdWm7$jmDSj-K5e}PUCE3KADFkyz&I+T1K
zl?jD9VgD}*!isdLCv4)peaq{1;fr7x8@-c7TWo4}fQyr?zA(pKnY!M8Z-|{Q{_0}q
zyWi268<#QW_e$yS{bLiF=*!MqBJ}lz@hdW}kye)gB$5N&^ca_ap$~NbL?L+!VEYz_
z<qO7CsrA{$Lip5svZ8{rQg5qax7zkh&PdNi!AnBdHpkIq=)IBq1MF()ZBkR+P9uzK
z`-^~V;ahAx4)&&*-=Ybk>&1(Pl*K#LN$7<{sJ;g`J9>IAj@dxS2K*Q?<X(FEMIj>G
z_GMOmY2z$)Hh*q(FEA<4oKw)z)>Yfu*jeB1rf>|EbdpX$a(Y5P>Fs?G*XSo8%#4pf
z%uY3oKP(f&XDj~1MWBpST5W^L^rc!XkrwCKz)eT|E<GWIoD4tRSo^qaG|cxE=GzP#
z!rm@!VuDMKIIzBU;A-zaHY@Qcn3q?S+uGR<5EQm{HFSL72Y#a6yEY}6{T93w0=_9h
zdvmt-vQYDWPN40lk^<TV&XoKjo;PDTw~NWQ{Vo<E2sI@HHGct=fD~agpF$BK8VG64
zb2BF_DeKB%1x3RySU_=<{!4io3|R~GC61qUTYj*$?}D)9NhjUs-L%YH0WxYh-p@#^
zkcbm0PySjHu(t@9LPmWkkXjZ<|9f;Eqo~j;h^sx|ZF)1A2>bBNdvtj7lEOxT6PFrJ
z(iE4|MS+Qgyy*;CDY~e0lz~0~e7NS*qX&(;)S)jSc$rs#tV4Vq-{+mlgg$cZP{%es
ztsH~ZS9sOU2{~sT)7&LyoQ<4YI@`!$7+ZcgHwXc?h;}WZvZLiBkoQr(ont5F?SJLl
zZC{p3(4MM*ey5MK%_AO;f0iYFWjn&;zi=t1_g(hUox<G+F9{HRGV64KE1iDmn9xc~
zCuQf$D=tS<sVXl91`ZEdPn<MfNF`{^a5d&2oX<%4iF|2iOc1TX#gzkQ35jzfmzud?
zs}q2v`7_qey8M_JYGB1Hq%f&hMV^Tfcnht#QRpGGIG(-q;uiQKrcQ8MuAvrUiuO_W
z%QL*xKrxJ*jQc3m&#&H6yvqistcEkhkig!?0D?!xLi`rx#-Vce2jv0#kUL;Ac_dK@
z@k;p{AOF;krzc(vdsjyuV=Eh%nbonEfv-ThDp!;1J$~+ArY8?6uq4S}+H;th!skFC
z&iEa1k%Yzyx(9XZoZ?rNd7p|;#4BY_@L-02oSfYeEm3~n9o>~ppJ?y2b^}(r^pZ13
zDwIC5H`h5iIy|&Nt5+g1-^iZu!AnnTW6_DnCxR||$Do}%8=)Pdxgn6mNpbIqv{E3v
zKGxc~L?60CUax~#e=3T|_Z1QI9h3Ez<h%s>^MtM)^CZK$Ek1LGu-C-DKFLM7a}5FS
z3yQD=m+><}7^=<g{v~3@-Vb9_XA8=AO`9&B*Yzo08ku7*=*VnuHtQJx@o<YN+QbhO
zAC+%wOY11<GE@hyV7#-U^v6$>w_w1xdT}pY`E~o~%<UO0<8J*;id55qEa+9nFXO}S
zMI&3s77%J|(k{%*4um~W&98Gjg2IYv1G!>9f2FIGn@U>VNQ{7d)tm70cbKZX8qIfi
zJJrGT+rW~Xo_4nnwom``9=gE&i|=MVXZ#Oh`Pqdgd&%=nu3{b0N|W6~!$4whTFzP}
zntI{CYJHkPNkLRtJt6y*bOU1@^$jS(<ch9L4H-uV+`ob^*nbB-y`ik{a67=MAIXSu
z@w}OPE*5l^>xX{R7MkOWeHDF}(msK?M1MG@32D=@{BvFZ4LkdHU<fSy;|C;5&da6x
z>0B_JBdG^fF>x&6RO$sGQ8wVy{b2<X8FsTojP(xp`G`$Ictyv^@#%Ie|1xy=FlFfK
zbE-@@`W|X)DBHZI_J`*W_N*zfFC0&Xn+n+sUqn>*poCHO^po4)+eLp<=i9k)@|2d;
zkvbDYU(q4=64`+?cbGimZ;nl6Aq$q<_~OZU)meSh{2-Yja~A%llpTQW@9Yotlsigv
znzfU#6Av#5t$<^)nv}<IkuFz|J`}rt%!c<P*>6Ss60bgED#9b?*>8HEXsKzayJfyA
zc%}M{#dEsd`uDjI;l)F7Rpju$+dAfSR&g-DddM%m>Ifw>^Ju1%O{<Bxj&rvtJ#dn5
z-&qTi2{2dq0=-UutjGYH1Wykq9|iS)EB}!-D^6Em`r>B7%F^I(msqR`;J#^(!Ld$h
z>!w@@&<{_p^p7uZHSAsGha?mR@^-N|wC0lUB?x){rJ~zHOIUVAQyc>xKM;TJ&+=k7
zwindDZz_QBZ)(Wj6~Ll~Dh?wn-_)S_dD#4}eg(X1Z$eUu<v=b6E$8{P?nZf)xhQT$
zE|i|h;V=?ci~1cJ)VAp5zr98#<x7^U;Zo_cwpv3L{@Lk;nN>*Yd1GpB=I{D49J6?@
z578|$GCeg!yDVS_9{$47wyV-ckxb94#lqE5zK72C=)-;vDTqoW#$c@~vHK6KAvJG|
z@$fS{VqqZ(*xxTI`u#ijcTv%P(n8by{TH=lOLSxWU~Phw0Nws?*@3IKExvkhxVwhN
zfs%{dO^^#CFMb2Rq|c<88Q|75RIxn!3a}Y+j7a%%5tk_ij8#<REr<&lhxOFnkFFMk
z@@JVd$w%HcsNtG?dP93H6Gp&tMPdp^dBe2eg)mtGK`M5+pZt51*&4B~&jW{ah+YIx
zGQ2&b&cpNszw>Qzui~Jq**fnA2^}cj&hL{T_O)w~5|Hg*U)J<U8+9KTU-xtdzsYNi
z=+Naty9OgLY)X@o>eZj2p0=;VG@04qiG~*&ww~|YhM9UkT0TRYbKA)-c*iAnZ$wk(
zAF<hhV_N^u8^tvI3{F`g|Mcqwi^`G9&0^1fuY}fKk$u4FNGpwZv+w(pv0b)*zGV?1
z(?gc&J6Xc9`TD^#1Eb>2wTY8^ZohZ{))V@qDB#<(<feH`_n)FmI|R)58vowXaIm2R
z^JZ-6XZJZg+rC^>SB=FRbW9Lln96627pm+aK#YuxnDq2NX0DZm=QI5qg+SMb-@jW^
zN`AT7T;@NM`9ea;iWdd-O7^!mw*rr=#u6J{z1&2b{pb@sOUNHYo13}jH8c|N_72Ph
zqg!@d)<<Q;Wq+u}B*!OWHWddpcX9L0Rf~uWF_d~hvfH4saNO_BX!4V}KYXSa*N*Lg
zD@@m5aZj(ydWO6c6@49b742^c*~!<BejC(YPCmF=4iuz#e8_!~VCQ$3%~?me5+ZG0
z{r1<(k?_;FzfrlhNC{0xZlMSvq&MeF^XoI32rB<EjuK<j5-=Wk2xKsbEl6>2cXy1;
z;Ws?6`p}jFUolT}_c0QMO)L*QU-bY9Z!<%MXO`{FC=SXk0fYd~dzEwR%N+<zL0!uM
ze14*FR66qO%E7R?TDL*4s+zu2DJDHKD-DOt%j130R`$E7#T^wb3G*|&B(-Vc%U1gz
z^II}lS<hJAny2Gytk9d=gaLD<JhCA8wN97$!Sex3v{oO14C}F)u0ePuwKo8Iig=9+
zzvEm$c9^m3<T%7Ar2XFNk>r7aK@O!})mUUUkMO(hgp9`<BNxrJ%3(kczmS2TjE1Ck
zAPDpOQlyXiNBEj3zl6GT)^wD`b19)o&y}SGK^8XN2k>6{K)#S~xD&R$aTC^BiVx05
z_mz!<V^a-tev}>Y8Yv2xH9Y|5jF?~X$>;Rb@*nu}B;V!3OIW)5`}4QhU#V_?$b@}@
zqK6aTp?Us9e=z^|6t>+lhJO`JCfMTkTvYuq&tUuI{yWWoH!60Ch7i2Z?fG{bR1**r
z2fc6%ot8fQ1z@nmIanF!zv>omXc86XKC*DuoU+-l(pde%lUXZd;R;Ad@8z_RCoC&}
zi;2bPo!wbV+q9<5$uU3Jpv||ozpgk~8~&7I40WHipA^G*K6}9FEf)6zB1^(Al1A8E
zR|J0_tuL8s2Wv35>+&2JSozGX%4q09XB7rw{}ps!sjc$V?_bcBz;QOXQBZl`ZMEv@
zB^CVq#I!xVHK~N^$^}3)98((MNaW4~V0;ZyZZ!ImvSrA`nGge_#-xCLBW<Wd`hJbB
zQ_^3{aFi`@fi|e8rh*VWX1<;e=Le4e+q3+_rrdC7QCZFC5g+^b>9nePD2SaaX}?Gc
zzD?T_d1kw>!+;f;>b8c0SWFmJB045GOG{I9^13fF=5bLn2Z-e4yQ%ai{Y8olxHdF5
z?%UXUZtv8H0L;zx#j&KB9SuBcy|uaUpSMU*P1=z8l>|rU(tat(Bb6O{i|_RQ5SZHX
z3i-U88%B!FUly22cqTq@mo1GTPB@+#|NAL~Lmk(II`=q{m}f8+>s+k5!<gaZU8@;&
zkh~zr?#z{e4WMo>!CHs^M{_ld{rmb>+1v+`hY)jS>45!dmQ%_|+o+GIiEU&v|Hk=-
zl0f%>ffZkOrpj990t%J)(AAutVr3BmO{$9wyxXB=V|v{UKB$W8DyR=m<whf4x1XB0
zKq#sEz9|@Lt!xHDmzsa>`3x4wEIcqDh7iehZkpO!0jr+`7E=>rVkrSE(U=D(7FAy+
zzJ^Dmw>rI!!g<MI`zT0~{~hdy1QZoSk^juuU~OX>B1nm9*|3K&KyNvGJ!&0Pk0gN+
znS^>AWHewwNw$CO)E(!v+c4>J$0PAvSs9<k5)eJZ#}pZHg^q6{S*c-(ekCHP*P(Y+
z@mSlr2hQL8K$zW&qsMK*MGN;6SzTRkZ+?Egy%d_s)wNY+FE*r%BE*|c%gb+b4*BAi
z5O#aeKSOm{7Kt`uJAS+-t1fs(8>p1xz4hu6^YhK~ko&gEic&jkXm~^M{+~Uy;->sl
zOkoHYRf}trE0pvnhf#19Ye5g7ld*T*9+{6ZfH-I6wIRDsGQ;}YgBWGjr-CB$y<Pmo
z?+_+82NxG-b3;QD=Ynr<+SuH{Qs^dsm&V$}vQ`=)$%1lRqVI%&TEB0sgNZu<O8I>?
zbQP~=e@K|R$T!K!vHK69f*fpLSkKfa(jnb5e}8{Rv|%B#QmO5a=Uf#0F}a*2P*j{t
z2T&tVPxK+0e6F>tIgj==%W`LK&nbwDsx7SC36j&O-e|*rq2v3qHTb7OV2<C}11@32
ztkAC~cjx@b^*mzU!B`ODj@3d^fwo2ePE!!Nym_{f4ZUwAEj392Yt+nh=H9!8vOp0e
zHaR@O{H5Ag8A&<r{mGQF%(?S)+r-?|7Vy?o)xwkASxzB5y8q_4CidxGL0o{+I1n`7
zAhNz0<5#=<mH4^dpoDGnt-qw@pp<!*ovmKa{9xDglW=4p_&BTTU_w(TMZ}|{t^*Z=
z3a;Fxfn&G|Dvv!na(+dl$NMalj*tiT{tuni2{Bo7or_p38H~hFC-Pk}s9NcuRp9QX
zZTA$BUIRlQV^YjxT+dQgA`Dl{cMT^5+-ZXjN5as&35HAxoR$635eiBfuKwKF{YF^>
zf{{ux@}9BRIO}%8R!%P^&}to@hap7Lt9<Qx`$LgR^zfBgeQ2BHz^Vgk5WS!2@J^D5
z>(lSd&CsQRf`Z-MU3_Md)81wuV0CguD@ANC9U`%C;0W+sIU6t|26hochWMyBx5vB1
zvz(%!HKiblL2A?8M2t0?9JwcQ9u7tUO8WgP@dUYqu#SjNaieUkzSSRpPK(A;{i!CV
z*t(_l6S`FvRFVxM$T^P}C@CrOd0hJ1ev|C;!JjHWFE2me*oaiTR^Lq+Kxru3to|rt
z%DS@i=%qY^nf}?8fWQr|21-eMsAueV?>H@_Q7+fc>}+@6XKnF);jG<Hw3l1ZHz;@p
zI%cF-gO7DJTf8rLXPulnTNo`N0<+}64~)x6g5iy^%v@Sv^V9akd++y}m!IFm1p=I)
zyZ!9!GTwK4%+0+(+osM0-~>2Gnc`MG{b?{juaXz^ADoNL^2v<rK=%i;?)|O*qnjI#
zVUDA3wnadr)f;9{yS0vA1=^k-yjj=lby_f<`Hzk&53lFPAKl=c;P#!b;x2@SuH^Je
zHN-rlcDLP+b}(Tc3EAkFBYIqKua3u0Ma7CM%lsLF7pL2nc7`PY1;eOhbL(*u8La$E
zlrF4DPh^(rA(Xw0fY18!zdktp<Z%4lB|i`u8T$bLR@V4T;Xl?KYrH0ZymMRc50vx}
zTn`d2r>-&f><+D>c2*s)4Ojv@EHeoH8PER>K4&}|>FaRmPfd4;Pq>_S3XGMCnRO5X
zUoEfC3*u$-ehw}0%)JWru4@2W>PLEaF8FBTmD(COV4B$Ij83w^&SPZaeOr^3(E^1a
zI0$dG_#zHPU%28*&h_oqx`Ps7SIs4gfQX2YkdRoswj74gjmGjcli}ZuW;`WAnwfH>
z_@6Pj@7sI0IEiX%^K*u(Jm{Md!7eDG07&_|0d}BZ?p7JFXjhWeQ&(3|P|(%Y6&F`A
z>n6u%XW}hWy|N2ec(3<$`jMcu9hE84855fruCJ+Py&>d)^b01=K-kwW&ALCx>y6bH
zpJA$#*0+ow5^sW5L87xpjEf_8b^YCmBhfzH2T%NJ_p0N^mbvtyji}~L2&2sOU~_YK
z8~#{0qap(E`ux?<j>VdU`q}Cc&Sx4kCI;r!2i=wwN)Pfg#xwt(9|9h?Gx>9$*~1tg
z;hz>943F3`U8#U$cqKEf-^cuO7Hx?-F(WP*ZvYJ@j6{G3B7sHdbHhW{zip?O??G!`
z_nSLA*dpWu(r(+ewiJgKAD)(|qPg3*YB0%b`vHi0Tq_p<gU++*`~Tw?@c;hdmv_)q
zmUFl5GV&@Nq!PA&rO^u{^gr$=UN~G9+84o+QaT6&HS`yvpBJ&vTb7`g>sD{pK>Bo+
z%Xl>APMd<f7;=!}SIJ0p@P!@*CoDa^ES>W=41NlF&zH4OmyPOfYx)x*=eZg}D)Nk&
zl!OTXFd#T7D>f)IJVal{WMdHnbNgV=$lTN@r#L@9=#6e}4yRxMw%yHkqps9QpWPrR
zlE95Jc*pA~RdboQ)%_7RP(otRK!hUv*RQa!P(&0Iq{}Nur@DGJb~cu&+1Y8at$Z{F
zKr#b`o3DrrcGS_Gc;Jcjg9sm_WD|y+n3g*`21tSM(d*jox2kh1s~DM`47x+1gvppe
z5dJj+Kf53r9IMcJ)41v1(5JaxnuN`<_;1q?N5~b*MSM(LqK2Bv;tVfr4$9UP1&NIf
zlL8u=I<<$^XQabZl)h>4t$NmJ6`1h9l{HEyG{4^PZyiE@A}o~8HJ~znTqSq5@K9$G
z1gPrcxzJfo7yI`MQfsrC(()=I5_X}lUY{i-A9=M7OJy$IDA60!sa(ev(fRJ$uX@L&
z`iCbN7{65J6cv`+nK-@_G**%Xva_68E3If|*VS=wvvN(%e^ZU)r-KwGd5DO4ce6lf
zI!tg^S!bbN&-M}DlHVy0eDH-N_#a1W0sCvDv;TQQo7#_UH_S>#pEQ5?@WJ*KPl&Y2
z{J&cQSy)+=l+rWM50hYQ8Jp)&t~Z*pvqm>9ulpgO?vIrIOlgH3G*xdvlV^e`_+MH)
z+@GIBFGK8SCm(*hkX=}52r}#Pc>Vs{Hz%X?5EGSD#D~*W(ZT{^Ac5N|as&$B_7<6j
z-J(Jh{DY$)Rff_(vlBAH77VFoRnfnYcDac|*4S9vK*%9v$j+)#<k~l*k&cWyBoivA
zWUP<Qj*R+9W;)jK*(gNVZPP2HIo+4wheJe$7y=Fl8U*F6yV_|By9$c*#v&#rM%X%<
zZJ~|M&!_As3xZp_zjN%~1b$jsK#Jx(#|V)BTM8wFm@<F1R_Sn=#puSpMAx-T;J6(-
zos!lQ`>v%rx%86oI#|vfK)EyPxRKNqq*520(ePNZ33VZ-h;FW^qazCV8HcEw4!Wn5
zB625Eb3bLRz+IvHoTV(oqSILtXFJ<>?gPx9D7jeAT)?8DqK1Zmp^A!%Az07${zN17
z=&OTD?!!xUt3D5cq@9)1=uDrcw3cj2xZekxvChx_@61pBRK>aGTf4Y8@iD4`Chji@
z{Xu`g4^-G}C}r9R2DjDHR_a^j11Xz28iK33x)c1oOUo%orLrhaQ&|0E?c<Jz)Bnav
zDcMFl$(xsHt8Ki2<5k@E&Eb}Uhz>}X&jbX}dVYSMm6dfqalf!%6x^0J!<{<ffZ5wz
zt>1p^uza>1bfZ`tMe>c0yhm06DkTJbolfZ<>P}Z3Q0GC<XNE4c1+fcQwn;OcPP9M6
z1`7J5>aPh2yu)qY*s%!DsOY4mVt8EKD|DaeWhJluP)rAo>3`j@#o2DiFRlW;g3ZRX
zq|iukXaoG!ISp;G^Z&}5$p(w}a5OmGK$DD2%)sNNp{c3qUH{B;P-|-GuiEt_qRYW>
zbfh-7gQRyfGUjNf;wZ~MzZ|Bq#KtaGqs|8M+5VEWBaM{&OwA#ZlKAl-Eq}3R?syE+
zPnGUx=m3<qLQ?VA;qStT?=X-4z0bea_rqRp1BazS0T=zJ?aSFL%86J8x#UzdtMyhx
z3;NuAUBXa&sT)yMD_T76u#(gTdkLWQWxW>HgX#L))$Ow16MnxuLa*lMqcMa;$C$g)
zA?q?CNN?DGT@~I~B!r}R*PCx^D|(QH7$5R8&~o&XPUhGBrp?g)5^FZE_cu#O*)S%O
zmL%s{KcqA6dy^u5_2VkCH^ChTXK_K3kG<s)pnvsd>R?%u50Uu<x3oUuK+Y%%(XQl7
zMl?<n$tk$k@2BIZi(QukI9`OyExA<8)Wro=mFDiJb6O9?ep{0>*Ru4)e?6s()+@<@
z21u{01*8@9=(YV{YwC1Y)bFx5$V>Lu$(?K=$Wztb`5^6&MZ9>}w!o=;gW4xpcO7oN
zK4GXIgQ8@uYXx3f?vjOx4X9CK>L;o7%1Apq05^}ozShj2)wtvRtk~tfcZ!Z`TCl}`
z@0NINXEU?k;;^A#DYykEv-yxSy(j=kNQwU23(3+-_W%ytRO-Pdeqw^^{EXm-AUmi1
zQ|b%fn8U!^4oKfvQ&#TP^25)+!ah$ePtYvgEZn@SE9;uN+A*1lsaI^cdBPMgRS%Ln
z?1yF8`=-V@G|Tlq`BX13z9yuK$0b^i_kXMivud0iUE8-^UiUMfb1_+!uvG!}`Wrc`
zw03yfec_i2hV^dmlx%`y+foK*M)&x1m~U$h#fDR@&78BUNo&lIvrjT2q?1WNUPx|y
zjES+d<d@o+^5dfID$*~{q_{-=?~AL`>^xk8-GN3VG+;O*@YQR3>Q(M~)PCD&Gqog~
zOwqf*b=IV0`;MpbSB9Qmk*`3D)BjxLrkh_C987gfUR7#kc(Z;_CjyAdXd5eTmQC*f
zHz`4byWtKsYO+KBU)lpR!-eSMnL^{ddLkq|41pV`PUfF)*t<1w%zcl;**78IxM5ge
zd*7(E-%^zA9cH2_uY!EX9zFjmy+D1GJ6QT%UwpymB!aK~JF`x}g>a7M`|UsWF@YQ~
z*we@aa`G1Db~P2X)#Y`Zo}CmN<`5^JDjix4hojU?$DQ~6E>U^BSRf&<r^q+KP84=t
z-MSvH86Hx~?siX2eJ!dfsHiKdsx5@9*xZfG43BmeI{+<R?G(R$J3A|z|4L0tfb`54
z9)hP;o%BO`IlzBre^I}-G4|%9DO09DSXbhE|M1LF9oaE^a$Ibt{x_ACxpj6PHX&a2
zcd*AEc2Nv;_rH1ELyhhh7Jg14wxz|j`CrK09Q?dA!;5Rveafm}VG#D9?Ww>e%jTx2
z)E}xsi|tJZT6)~4RQ!Yljd@P4<|a4ShkFouw1VPfC!nMK@0RmIUkI$}$<58qE5s(i
z!GSpc=;JfK=V9Ss=+?pfS2TGIY|FRJq*IglPrrF*D<BvRCc7XlciyAAJ+}60gt^`k
zMXubESf?l2#~M<Oy0x}_Jv!t<&G@(~zG{5jc2%oF_tM+z_qKn%nlRK0P9{+jJE~+M
z<`h~e0=Q57@xKWeIsH?V_av0?kdt<MpIUkx*qwu%I}k1Yd2kQJ-0iLXf6N5s*+PKi
zb277i8O)m7&CQLIQ+<6sJ9~i5e9+W^(D35;goKV%Oe&6-S94AvH5?i-CY)gS&AO=4
zO#KE<@*+o*%N-~(LP4aUTjz)NTTLDCbTxax+-+l2(SDkHh>w?ZnY)RgO@I9@=k`7<
zKd7sYhns`~4?7wO++EaqGhT9geGYw?G5LkD%HRtp0r_dQ$Ga{+!iy-LR6EhSS@A&d
zlU@Ut+WL@UtqVg_mx?X4%SSXAlmloqRfOtihidxcv=z}W%{4F3Ma1I|c9d_x_`BE!
zcvve1;(zdtCnEOz^BPgw6ahuW%JrS21phD-_7b~b3hbgfWYvq*>H<Tx<NE+Ax56~S
zT7I|yM?Y-RLW41XdV2^@Z)oejJV{!|Jp79f*$YYyjp16rV@w+n9a5C(O8zhO@@;l^
zY;Ps@Gu$0hW@mu0s#AE0EgCQ&@K_dr;^Tqz(v&OQ^Wo$D^F#Q?BLO4CxX>~F^_S1#
zf#+XM1^;w-P{*vH#yccvX1&&w388UDDyhk;^bf4>D31NvHVb`7{pTh}CRBy_?qEu|
zY8JnP@f#fet5oX}(9@*?J3wH_wRM*+vU4OL1XU&rA#F=XtUf{hX24L1808gDT10em
zqFYHyYvp!~`n~3{g01Iy_*24A6{T;B{y*OaD@es(F%4~xj}RZ1kPs^?FOvSd9NGV(
z=`8%Bc-*cn-QBUIq@;p$3kcFkm$Y<A$4EB_NOyO4cXvp4cXz(?{XL)eFWA}LiF@vI
zopWaO-U`IhvF|$i7nI~h{QO~49>E3E{seXT8Q;0ne@!4{3KRVH%NNqZE=tzM{oEe4
z-oH$k9re7pzyNms37Cl&faDr2Toj<c*A>PBucFztRRxW|SNtPuqkUghd}92)1AJV@
z$CybVpNRLz_;L9H*P+C%x^W`nU&J<b@n@ZG&?4ky#H9GUo{;^FC(=^WNXa4I(Q~h>
zX{X>Qk+!<p_6oHx{p{J*&MmOK@?9k*72k7i8jJV3#l0{?tepfXMi<zh^CtRWt_lS*
zl6jx^T2KAxjAA*iebUdLX#NbBewf{(m3(YoN+Q<hC=s(TU7NTc>H0`Za$m(C`KxBt
zM0D@9CATdLQHTzlaVN{H{yWqReFXUMbCF}lKv{V{Jv0}><lDR~xJ`R#TsCq+Dm;?!
zW$k(iscD1|VqhO^68x>Vb9h8ToSdYkxwM$6jG=>a7B7@#@OEkJYVDCxUHEvCoFmD#
zA{@{?1cJ>H4ekiDq;B7*L%@Q2Z#7YkH@|M<xApxVwcc7tcZLTPd4*#Spq@YsvXMm$
zA6}4E(5wW+aH>PeCr8BcjhqY`+KX#5*_d}%Rp=am6<_SGWu_XVd-dt|9**PvsQHx_
zAF3^|mC@!n;AddO%jPAcc9CzWr=7luy=4JdsZrYqvMxng$5w+&Q&pECLq$o8dH*YR
ze&|)<CVJOX-jYm2vTt2lJ`{n`=rtyKM?p@9BEZS;ne!wn4-bn0E;*>>6bZ~{jJh!6
z0cdKiUU1sWf3$gcx35CN!KRnSMxJ&tMR6Sq(5=julZ}+6r{l(j@>2TPRD2cPm3ck;
zMOj2CAvrh%(c&@wZKcX{&)S0u+g=r`QEYcT#Ql+~IHT%6cI;tMB%f?LJ~fpTa(rTG
z1)CF(yR)}r)cSul(2OjFB{m1d0AaFgy^jHzwW;j?>Uvvc0gt<o6~$UlpXCxzyJ)W5
zwLl<tYJPV%v#cO}wC>8aYb*y=D!aeUA7g6|dVmkMaF_i<aJ~W^B;UX-!UJ;^^3V|<
zJT1N|_;V+&Pdz;>6b4V$Rh8Uhu7!RMSPVgcT`-fn_{?j<A$)ufP&k#r9QKcCdrv)-
zJWnmgon|s9<WK~@;tU)L0|suEm>k>1(xE7;+eXGYx$nVAs%>Pz@1B*h+~R;JDW(6p
z_4Uro_+W*!fr$7rhWtUdf}iOFOV4N(L1>%P+k+zs*+*?UdYd07UVOBzC3%GqkD9ge
zG6S!cpDpRfHpL9Un4Bwez}Tk29QA{WzOlpjKIv||s3z~X&@+5^lF(`LwLo)TYUYO*
z{;=)l|Bl&frOp9^5($p8^mj^H?9u4J%;pg9emx)W%PEYDhq!p<CDhn1UBLfJ-lTub
z#-s~{hP#qJaR^dkWxYA-yL!5>%`+XsAYr{d;QTZd0p3*P{~hIfr``5{#|S~O|EPd{
zI=`R5e((6i0!{D4OUv$9?cjxFmaFq3W*AeA^6i5~FH*q0v!Ot^abux8xj&^#L%=*i
z`L{#f025WU$u+(aKyX?)`Hku_B0swq%MtX16$$)zO)4D!_f)G1%POLcm?Nhps0HsG
zk@CPeYiXIqw1?l-1JC~0sd*s?6OV%Jgcl1j`{#k3lo;gVUwZk2-Qymc3aXG$_dL`!
zG~h<b8scH<?fbO8DN#SOY915~oo*RsQuZZToC;|<9aB=}=NX#B&+xdiQ_xmdIsg0x
zeK62HN3t*G!w@rgQ5m%{_Bgd%V~Tb6@(g$Ykxy6`2M@8#FBt`JwH^Jn?JJggp8C=<
z>$TOY5=j|$;ha6$0pWxiG#sO;FVxf0d?Jd-<d5<S#}lJ%Y#6z}x?x&%PM95IY5pS^
zghGSCFwI}!D~?|JQQIpQ%;|D7Iqn6=j-#TA!}vhY%rW$LXs<4g3qkq1_&muqz)8px
zyq-kC!WvkMvc@;dg<yG6so|b6>S9yR1Ks1JB(H4-lM9&3+jop}nV=V`ZZ}}DC~J;E
zzWeM+uNAR#r~cdrIdf*Ys&BP3>Y}jGdAPWpMi9;8s5H_Xf~((8wYd>Y0<X$)AQrbi
zp>~amO1&b@Ed(Vm_f=3$L-QJdrxJLz`?BvT+!)_)%oTq{psl^KT<)UX>U1Qs{fVBs
z%v+F9Y-<`EaNfFW&T09EK9a9pRS3C#1>Z}3(cLeT-z<Bn-)HL^ZDD=qsJTLPi__xb
zK<4E_2_07O`iRoR*{UF}s;n5D9+Mb{iQ_gmca@cW!$7EdWP4k+=9>U)&)JZ4>|qaK
z3YO5dW$y39Ak4AOlTOEV({9#pV*OE?thT)zJXmTF3v6ow2p<<kdRPs~5<6YQMdVD(
zH>VTM|8%_5vjO)tdk$bX!Dm%EW_gYDy4q@ITl;2wOw_zz8P^5bHb&+fFadYt%Bi5B
zD>65hbWl7PV$c3ZUlt6LOSftcW;jr9fbE%)AuacRzsnU7CND>wZc*SvM6~~3*(3cJ
z$hY%<{&=01K9}p@6qhH7+SY~Su9n_0uveb3n1^Q6FbpDZHveVT-nRa<aQmwJ*B2hh
zHk2?tViCP;UH23n92omVFIaDF`^y760#el7kI=kCTs({nR841&yMvk}kX{_mn&w<l
z%j>$?aKLUT1|#6<OD)kS8JU=%k?wfc_;fBcRs3AFIJ-KTn9G_VSzK5)bZpu=ms-hE
z`7&N7@O~zY)FUf;k#20w<vOdt)zaSF#EFc<W|vNK{YxD;odEl1e)^8RkkILa7+Oj6
z!ViodaH)pC=P_MpvR>0RSFPzOP}yimyYGt*%g=nZHjOx946!r7xERRX!(T>#xSc8N
zPejVe$c>MA_wO|Zcb4rYO+A<Vy-%p<NL~eR^JDYsNy&r<<Wpc(Em-5A^e0|T;k%Nm
zqKfhyS2d4wFBjXhnYojlqs3cKb8U-DQ#0?~gBT#BW)UMMjqNdl+swslf?Br~b7Bno
z$%P6`9WPA2FD}&~ZSO#ej7){=^B%3dRPF!s%E=SBe*^0YtW@C4?65Y<DT<+X$6>mJ
zpYrocfy)=-3kMikn@{NiEeEGRywuIe_jHdb$w=Lt1fyUyclT`k@^Nd!<VDE&?kC1k
z0cS2XyuB7bRxS(z_Yca3zTJz&j3Edcr30P@SyZB3R9RUDPq(4p=zkKWaXU;Nsh)2c
z71SJ%Ccd?lWfi){;F8%coHhs&a_09?Z6eX{&p-6!#G=aUwOw&%mpHgNIiVwaZLBmG
zW|r+1MK52!ub^<~oLv?c(~#5D{{A~K6-YezaQ>|jQptaWbxzO1nO~FXZn?qL6W<mB
z;M(>Mit4!v5!_;8pbLD8U-u%?`L;jN&SaGiZ3=#&9Q_Mk8W#SExh<bm|EP$M^M3d^
z`0H>a>&Kv<aztiigo&xd((;&8R0%BZ{(;U%Ps@|vV*JDDq}C*_zHD85W~EpTpaXXX
ze)ARie_zJYMZxrcN+cdfoWNpl`D(^yk$p^jeEhzrMe}d=dbD4<rwtX*SVqQ2$GZM2
zK1cQ5ug}<vRbb~n*tUD(TKbH5;X7)qG2VUVk|to5K{Y1(tzwcryAqO~5*MrXMd>vt
z{3(?#5l^T3>-t0{`H?j}+}Qz8U*5PVc315KD}+JEheG|LOYtlMHUf%5zaV}6`c+D5
zaA1I_p%>kpl*g`b$892}rmJc7ptcdOAdA6=k(k`HAgi;e=T?X>s36hxOKz;AT*?)a
z+(7ApFLhue`H3|X5R-;^vmQ{KT8REPP7y@8v5r?8(eNAZA&5Ug?G`{~O5wK=l6Fy9
z<X~$~yN|n^3yuPTMlFb9YzOUs|0oYA-W8DwY!TgGJ36xRYe8-55n7kCd<}IcTz6ds
z^QonL(i5Y@{15@I)!WkZ0ea$zw|mE7{B2|9xt_-?o1Y#|#(Ud~J=lNOaP#~9@0%7A
zAxfCtSv~kjWT=z?q3GLjAkspt&b!pqR0!k-+{7SFQbMh=)R_ck&iDnV_+gR<C3{#T
z{lMUC4obOYCm~4B=r{AfY5k;!1Dv$PW){I>;A2G&*%)D@a_>Q&!7?M92vX$M?7V#y
zgBj1YVbjuxgQ5;RZY(!6bhs}Za5Cr5&KPji<88jg@VkI%>uJA9Tx}%qrCqUXhwFAX
z5U-M<-5Z-!iP(h2+jA<Z-%{sAG~KxC!vqaYZ*T8H;e{fo8-lW70+}jYC(P<a+b4;t
z<Y;kWX_d9*S9|bHLs$0hnwElW=&*2q!Nu>kn6%RV7AZuQ6<*&LZof-Sik-5)g=I0F
zLokPm8W`%9LAz)p(eQYI`y_OS67slWbe+<rG!z(iQnnoI^-$+%c@m60XwwB<#+^3N
zUGxlJWE@N{3;`!FL-Qjy0s39>p-_Us!9gjhV0kjQl_sZy6lO$SP27?uu@=DQ&Vhi3
zdfdyEnflw_QRlcWSoUQFy}7ZA2SZZ-yCz@~U@$Fj#tOzv6O#Y>^|$lON=nHnDo1Qi
z@BKW;VzCvU9`(^Gj+{~D8slKbE6?%m9Ae-hjAmQN|8;;F%~ela>j2jR7^5G`To4(1
z%i;MwKs^i_oAi0T^W=z(0AeLaNUj<mkGlayjoXX`S8an8DJk6#CHKPa9@!k3IlcmH
zgTK@#?GKE_`aci$Pu+#~UtgX;t@3o`s`>EH$a7f%<Wu|*SxI2<22XYN?FMhAs}QeT
zdN>r)-sd8!_=A%a9!Eq50*8?A{BXwnnSqWMO#4`r(!|84tCRRSzaho=VqG->0RjFp
z?KY3j$-d2mwz`mID8-qH8S=1@Q{a7`{>BnrvD^G*7T{tob<_hMPp!1@a0d-lsr~%+
z-|5_M*4EZ#Wdy8Ezb9^Le#fDKZxawa2B=l%8OwJUd`JvL!qEjDGp~ofek}nTcq2G8
zK>~C<k8R`r$`a-XF8dX<PDYr%31QL1B?CpIAL}Zrs|7q4`xYPn#y%)|0#+)ng`T1<
zq)gW{+h3aBz78TVq7ZSz^<oqya=Qk;$NL;<1HCD7{DYi~@h4oS&+WNlXbe6+-N<Mp
zIoGGzX%IeNrW-(Gb&bJ1_7PosgowxMIbjMLQ$|+wh<GLs-=*!&pp|AMzI_y$(TsYc
zqTTNkCox2m(zT=_6L6O31rQ;cM8;MyFrEUf-NHSxItS}e>>^}N61>FHAE};o_+G!(
zqD@212zl;+%wJ!ntT2DcEzFmd3?_uf{9$^tpFazViwld2z}}^ALCCh%pTLrvXe^xL
zPMD9eYmpG$@?9LB42LPs^}(YgZ;YcLo8tdm9>;qo?lQ_Dk!k0^TEs<j$4}_rp2+AK
zffS%(!%PEeHlWKg(3j|W21z0QiQKB$r;SVCKM+7RG_>B(;ZTwb=3;;TLW=(GW2FO1
zhjgr{sE-dce@cnWuNmBo2q=PO)$hk^=w3_KRLGj3Wlxc_`s#HuQRRhPRRM=zZ#QtM
zcr~ZjfAyD;A&^BAF56>Rfrr6A@~H|+MH;<<PO3dV`mgO>M5$2C6HFR+;ekfOJziw$
zP3Tw=WC6CP``s_=1$2`$db0`hlJ9@FGjw1y@r9}${X$Y>^d{#o+`hOOlTQqI_n*DH
z`1in?E_##FnplS_1cQ$qxtMC`M`&td0Gm}ZQ{gp?aSFUEdFw^6jMn3aZ?|3w9^g?@
zR010h)0o&PUn7h0t(Ig(dpr85-<KJ0*dN)c-gG?bd1&FOXI-@$nP&N`q`RTAEmZ}f
zvZJA`Lz?D+Klh_9{f6fs%==PErvu-7RsYTXo*Q3JYcG|gNH7+maZcaZF8g@byM>@&
z?~L0V%?F^eZN3xsAr*YcV8<}#6vF3cERI+H<amiByGp7d08h_}NhQpKg@s9uVAVvv
zdbTX`WweLdXXYYP0n%QU;v2*|E*FA)?G!M?gJj9L$M!B6PZ-JH2io&A*lTx&+VNS0
z^Sp#Ss*dN-1@*NVF6U0c$`^kr@MafKw<M4J8wcp@|IVt5Ds>nBt|>LV&uEelZ@tbR
z31-8l(8M{OvUSWju6N?8F26CTNboE1%1ywtJD#`Py{|4Y{PgRT?H+FERNu-AvDxXQ
ztGnCx(*ya6xkP=jJns{K&@iJ7dH*qf2lJBJg?>E?U)qFnSCOs&JmDZK;sE)2KCDaT
zx+~C&_bVdJz7dN!UFTmdk=qU7k3(KWJ_^gVHmjw(Mb5PjYetYU<b**%h+M*wa{Z?h
zW06b`4p={qN5)UWh@8SxOZ9cm`DPK`e2br(88kB1Bh|UxmV;%yl$WZBboXY@Q5FN>
zLWuh0iBQ+4ZG#v{KUO<{TO&`6?`wRub$~a=+?x9H$J<V{_-R^Rb<Uqn=!E3>xL$HO
z<`|rT1<#bqT3nbOc4BKLE|Gc&e^Jm-jrum}a=sz{bbUXAD$nDNAVJf{VsV<TvBaNz
zVngraj4VfF0m1FdLrQ5!xpxPj@e%52%}!jKJh;#Gw6+zA_wBHv@LoizTj*rMm%!)t
zCeNRzI3^dCb>(doG{hvPr>|6O_T%4=4MN=T<<gup*HP9J`9?Y@TNJ!^(wl`~Ms~jT
zGL)HnERG5to0(giiQVl(OTQv@eOFVFgI0}DdlMHHUzaPurJyv#;w^Ake!VO`&o^r{
z4*xOaWL4LaPm94P8TN+l)j1;~(*;}xZiIPRm664DJT)}&uIk1;Y`4YZ+*(aNi~nZf
zT0Yx~pW-$@+9e1{Do%Haq#}Dnk@4ZoLRj4!iTp{r+&}tTeCa(v+)e@U`s<BU|GtC}
zFEvS7``-*FN9#S&>;N<|WA<ooVN^Yf{>Id{9uzMBb4_)8R5(i0dZ~7qP!KTs4tZ{+
zdpcB)ZLK-DuAVg;nq`6T=g>FA=XOkf^G)-kF@cJY6xBgl|6_TB8*Pe-`F`iXO7A(j
zx|$Vi$Qtb5T+#iPfYz~1c3UF#2KRfv{!FBFRfOQV(ZZX(P#h$<b$oFghV>>|08$GU
z8T1<-i#%F7%AVe@{^Fa!I{<xuWAWraiS8hg$0fHml1~RE#9c*qZfZdYIgWM`nept=
zqADHUZJvlhI`pD)S{c*wb{=Zo5F<c-y`BgHf_U}RF;_o2*DeyB$rXqrCuF%6{}7P7
z^h)Uc4<=C^^nT{acu{j+k5QPcwnqvA#V>8-AZcdv!+kJ+5w0l60=irPFOj`oo7pk9
z=bk6==hsTjhF1jd6nB0Db6V>Sn9D;+`Q`_I8Y8t|x2R$7{EuQ;6HrYXPH$qg2uA)@
z%AFpMLybO_yZGVO{Mb>_?0-&i23uCpN*LJRydJ1A6*P3++S!DbG?IwDUP=m+5ZB&t
zDq$!tx|@Ok>`P+!A8Z0FS60LHniy9o5-YA1ywp?uoeB{#vHGkae{hRP`U)O@T>UXs
zrpY$$GHbS_5lWA*^b(zdIYoNgyCrOO%*s;GUR1P##j~y`D&-R?Zf^<@BVG((2RC^)
zX78=+*K`Fysn6j5eAn62U9<{=SC~nb|Lb;oQUQ4FF+!Myp59m%lNcUC@KcaaX(vm%
z&-1dqfYBsFC&cR%O&GP=gkjY<{Q5geR>?<4{k`oyb;mEpb0<pcO}{VaeH8Nq+t^X6
z0+t4zvsX^{SH~fBQJhkG6uHp7g6y+@Ufy>s$cQ8AwmxkPqBC`}(oZXH-=I4wNj?Nm
zvw{8}kWV@wPUm(%d5sC1nw{LiyLLGVymL?%Mz3tW7oT^`H)N!`9?0ys<|``yeenHH
zw6enFajxMHBT!y4HZ<M)%@r7-$=jgYj6;^Ls6EHoRs;LNQ*4Aan-dBJ$G6ooa^M15
z??OO`f598_dz{~6+Z&S#ui~GH@z(nG?goH^xy)KzV*JCv%z5lc_xok?;TY1oMBusr
zQ^BNy4)VWBYgIRd2eXa}7FDx_SHwRXr1nLvzrsc=NtauoMZ1EiI-nS6Ak?9sSkhia
z6p@G?RG434==lVL9(pzC?`I!}qGu4>QRr$gZ2uUf`dqz1{b|o)P73<r$|R>N66FbO
zBIFpMgGMfkgKn~6HII%Z88gE*+!3CMR-H7GqqCm{Ok+}j$N2O{LKEzs3IFwxQ75<S
zy?<|ycT`h!RZ&tv9{!i+)apQ+_3Q=$4LCg673xpqm?RQGkN&}(pw}`hC4RD=bqP#y
z|7e50)XNDbZB?wmGMgc1pwiOQx&jJPDmvfxD!exor(^O{waZ+d)=*qiQs8<3Q>?(G
z++2{T^@ZHzx*m)hp1P*iPQ()Qi7s4{cdGJ3e>4a8N?uVf(N&;dB=9<tkcj+~fVH*0
z+1cg+!9D$F9+Wy-k|b@AfmAgKY$oo=;`~$0H66X1EmjJ##@~!O#w$nzy4t1`HtO^w
z6m{m#6BdxZk?;jpkHwd*A~w9SeY|ybTU}bLd%b`QJ7s3uqaTHDLjIwlL<9!ZI$Lq^
z7NtY0)0zC(!jd4BjQr1eC1_=O=5qV&AF8ImsJ8<e&m6&!bJ2q0fxFZ0?|_52uK9BJ
zJB)rDH)Mn3?`v8@-Eu(2(2bSL{rZRTjxLxcyufhzlsad7r&f=DOLwyZ^Zp>esw<@e
zuABEqsF(oc?^vb?T8x*v{9fyOr$6kz$V&f-wI`VWy1?YMN!Srh$^Rb|S9={6skOk{
z*wCnVaedji;rxl7pXEVC#8r%L0zL&k>4e_<ZfEDBY_Ti85&&Ht=ycaA=;!Rw)=OXN
zrV{KSmL$UUq$KebIa#w&Tc1r?U7^Rw`NKR=qtX(U&+Dq{8nc}dIX(z=H`Sv!R8O}$
z`m?6<H*B~q%E+t0)1z3@n(lBOQ{)(I`=Gb<>1S`{ACK!b9lC#7w_f-CQ(?Q_rLHhJ
z!NVQA-weYr2SgSmTi2BGA88b%li;?5{G}i{PwiIi?zDkYDLd<tJT`%Qk6RNix2SjB
zghvRXxTmhtk4IeMxhjsb9bp)D1?n%)NM=hivcJBY(D_mMtaMmL8DPxvmcLLsVV@cA
z5X4VUAk+@$TD;zo1xWVx2)d{Jy~K76N^y&1FBFw=2Rc-^X4Dg~?S@xo5s`I+AjvcQ
z0POx}wCGh^lpx3q2mgV~fy9q5mR+G?mivLaYs<6$5Ma~SS`kW7CK?`Ud<ZqdYX9bl
zBWUwl<-BL^8<&%>Tv*LjXbrXc!eQwXJ=D}Mx4_pdRulN8uNP+Oldka|es;~t9GSja
zoI+cuBk;hQWvy4qHLdg)dqaQGh&5V)7lSFz`2_wl^4UNM*}TIAmqn}7rnR9ZFS8bh
z5KC^1g@XCRM75^I7lzC@Qvpxglaj`WYe6VE6XGe^TNXGU1XTgyRS6~WvxXIMjP&IH
zqpC%7+Q&SV6<}Lxv7nS?OhK!yQ+e@-R0z<pwE?^1sFYO^H;~7BldzK1%EaHH`4mL-
ztJ!@GJ=`_zmAAa0rV0!qEhey|bVH~wS&v(pdY;*Tryx2i>#==iZkQ9|*@r6xEz=zV
zWwl_xC!+p+`D$h)rRBM6v)*_o_8*$n3XI|OI@T@DP=b}0K}L0D(R;3~gKjOd2z&s(
z2;s~9JW-+4t!2?VJBo41^V1FoYQ$2C!(AI5E|hWALoY~!2EpwimUuKbgq`Y?v?ar<
zUjEB{U{G`xfJBn>(szvZ35!WISU9wfc5pI*3naYeg8$-SW3}`rCtl*xV%5|&brn@~
zwB{z}msdZr@&4xx88)r48VC7LNl&xWVZVYWE*72?HiZ~`&}WrHxn|T!{ffqME#t9?
zyeL_kvg~;VE7?N}Hbbj`Tzak3uRAoV)$Da1bl=8{*NA|SEK3SAm(IhW5Lki_TeKd3
zV4teL82xoWcrdP!IVp$zBE0kM7Cf(gt?C!~c$=H_`9ht)73&FR2g(6`%_OTPG~x&k
zx`GHEIk|rGiTeeE!MB!Wfee2}yQz(zaM2Uzn_SjHgFMq789_j!uCX+`+4%?9mp#iD
z5Z<KVxwu2~qf$LBJrS0XXUR~rO7tfrRu+%1NqM-Z4kIjK8C4!c&r1se#@#X47D-5n
zMU#m=z668|&zU#WQUP@An`Lv2)BTbE&0JN}(E3Ov9}}LTGH^#-{fufOQUl_@{7>Wq
zBYvLX0|F%V&F)-#)EtfRj0Kt+xF_e7RQ@Cq;vBmVu4=2<fvds?)K^5dUA!sY|MtKM
z_wbu7t;PA|j{}{nNy^I8)0s8~G5?<>_^I5*3J})ky*_q*pa1@!vHm~)?JUnynOYN}
zrE9ZDLb%{sLsi33N123;EHwf4autQ`U^wU!E>DGTRz(lm^JG0D)w}4u&?h2Ib;Inf
z2ZoOPG^ELkyM{NWe5W5SxXTup3l+(yPrZ&_O+R*F8#qUu)J%Fi+rHT&d~sRWX9zwA
zoKELGbUyqY^X(DQ3B3}e{DD$qgV`WI#v<&Ot(yk5Lq7R=^Y`7uM%@rs2iGd(zHb<@
zdir^}kG<|)={-Z4>!uIzOG5mNug35?zTC3AJ>$8~#FSTBiL6vM(W*`wJ{&@Sz#E%=
zE4<cFWJ=aHswU+5y5`nlp_D!uNQO8Ew7K1$?;5+Kh{^J>v$##k-tc-uNz6)gFB4I<
zRo$(q&Bj;=laxYjcDFzx8EuA)$Y8=cYbsOwTe*dw-ElAU+Du-j8H>c%lo7P!J+$Q#
z_ltxzn3>t7AvQSJ7~><w<U(fzvAUC2=vM9bvYR65ZcySa-HJ3&)~X6!)>9XNXjn1t
zxsA(ZQ56YcT#56dPX80Ze3G@biFsBvBP-8lU49JDSBnP?v!Ks&&A7zEYlH*!!3BwF
z%~%grE$;d&`LD@6z6662(FFQ)$KiMnm17cjroUbPa8$JYda{#nB^WRuh>GK{LTc`N
z;{EjRat;aOW!W<FuxKC%>ffFf@MH2-@&>O09`i!MzQ6NBiGlsCn~SE#9Y62&8D_g5
z)$0bQJ2jjxx`{fBQ#E#Ebk$t7eJft7?ko>l5!d!|cHA0!3-EBSk99$4!!j6zcIi|%
zHK)K$(K9G4QsH<sj>z-_g)Q#akR-&DEH}=nbj4wfnyhJ?k8-nhjBv;XIF+)e*c$I)
zsE~i0kRo^+sHs0uwnain=vS98tY|`AM(PiEJVH^*-f*Q>KT3(S#5hD~KM10u$kzX{
zgRQ{fbdLVbb%_Uo>W%E;{2J-kV0X_FglAy1uYrKdYtMv+Y-?!um?TPb>NAh;^+BAP
zg2_fZ<?9M*g^Ax>=gW#5KvGPKOR#B@N4e5!tlZX}9n1TG8+KeS;GsAL6{3D#`6PHT
zQBOuCcwLn0eZ0GxEW%~`o+C0H#>s;j=O>ro%=!Ie`t#uwG{4lU$^)<9cGH!VKqE|+
z$MN;|p61UqXBK4e_L9dfqA{VAn<DFLbY4VaEv0R!V7(<8&t#e`@Qq(-=v0|Y;OL62
zvBDLV<9N4oP)}l^r@R+rZ-fNET-+7iI3oAAqU^-DoP>9DSGxA1{OSs(%GmOtOxWi+
z(J}k!ePM$q{ONd^N1)3(K8DU^s*RqXTiR?4yv--0<P$4+nE6ULM5SUfhZg)K^$A!R
z`nlRY;3`H>BPAK|@j8bv_nJ*QFI$DH^kitprG?-Ed*wp;S$;J0Rfb$xM4S+uWWRug
z1rm%H<Rr}{J}`>@31Oq2p`c1fBJ{@V)rTE&Ryq>Yc3?01O*;WW-OX~U_lHXss7{Bz
zWt90}g-`4?a^9fHj+R<MQrAWBD^vP{rVN>ahO5)yo&avpQT|Od-7P@&sW5OYXOME&
zfBQyGw*d?LjrUF#bdO4zxr$pUYu$AWY#+PJ7-I+x`^1NMH!b@i%>5V2gn{VrNfBr3
zCFq=M7V3m;ELipND-au-8grMIQ2{=*3tJIn{$+VRrAKXBTQ;=k?~aBRERjX*@-kL|
zWq13)zy%=3ilOx`PXUo~*Q}A|J1Wz-`$~A2EB1eyIs1dS2{`{LOTNvM9&WuM-H>5;
zo&+X`>Pf8K*s}4IFj5bVeyucB_B#1dOh(~AMPF+D({bz*1ryarV!t;g${R-!S1|T{
zAN5!e6^|z)0%2$QT-Z=*w;Hqko2ksr3_5VSWt>O*+?I~?d8=K^dHsa)J*BVY>GkM%
z!{hS#RlDQUSnHEW)VNahH}j>(jY9K#joVLdWo|9mAGz51JGwhJg91v}_|P_XlZqsu
zc(2_NrN9zOZgHa-)R7bImKIOz`^Pop1_!QVDSJ?g`?hvJGeVj`{|3wJd<LS_VE@N&
zG^vr7qq4gGn}0@MxDUR1$w6?n-i<GNdy8p)_Of$D14;R%Pqz`K<2=pU*KH)=TbBAX
zRNffHu9{$0{f3%qo*T(pn;V(kUD6c;wWE}vL^N}9l<VI(E+ROX{0S&HS<mVJWh|h}
znU3?h!|{i?C+hu!-)2=1qc1K1>>kqs>dSGcVal7+*yogaI{Lq~IVcN`LuqV5_?%BV
z>9YOP^HgduKsO2-2d}w-+itYpz;xIqywMmgY^1^K*5R>&q}60`&pv|Sm@0foR>^j9
zmHQ3KvCeP#uV19b;h%H=*8cj1aYRB$OwL*9*m^;6YpgFoIHZmv=uKV#z%Lb{e8Q%J
zv1wRLIwZx_&tCppjo5y1nCai@b{qWdtMoAGU$u$)7F)>uo=;+P;Q}bI>$PR{L<qK$
zoLQ7I3`)7*c03*1$qv4ix3Wg=x@o<&@1;2;Az&vO$GOrf?~XVSVIvjh;NnGP`61$s
zhI|QrkCrE6%qW()=Q_j>aPl11d%3sM@j5p5>m0dzaHr1!rm3+z7mH`Dch%{8uB|Qy
zyT+N&N5y*Z$>RwNjSu2~=6ARl7>bkTA-Ao+xN<AXh*X{)!f>~v=8DtuLa2UrlkU3r
zbjlb~vvwcjc1Xd`b0D2AiS&2$RHcRc#<;_jw|JjKRbyf3fr&1F(=B?(I#x&5zo*R8
zxFjy|MzczS7Sj46M0TdM_*^NE1X$i_Z%8`XkVr-nA~HN527<QyNMV76H5hT8c>fF+
zU+XS3q4-uuaV6z3<7%@9c|`UZY4HEYkKzTdAlt6jNp7%k8C%TIV1#xoFy3#Hp#j0Y
zPx2c5ZH_6+xzh`bQ^7>4TZDWXA9odjZYblV`t#13-=bj$^;kgkEGJg=Uwy~Jy1d@G
z!|Bz2b6pdc{{jOBHFV_3Y8GRY*N(i5#vabZ0<o|_v0q+n=vsKr^XBE^E(`qu*NZJ$
zdgUUq#fP*tfiwO5_9$Q5t<<A6uyL?zpPfALpCxuu9D+Qc*OsoLpcyR9<JGwr)c3@f
zTqKQBrDVY3MqV)!y_=NUigEy*^4Q*e9{(w{AnBCPj*^zSejkmOkE_~%J&0d>k5>C~
z)gr4x-zh@wJ<@j)7#?sAF~U;)Y-fuR=8}-~R79#+Msv!A<10|=h}Zj|aiR8tel`=o
zmz}FCR%s6`M7!qwz7z;LBJ=H_6>Fw`ots#=k9Z9>+oTf;%jO*Nk(?VRWmCgTNxx)W
zgT{XQ(8mwco)Yoz6^F8ZGu?zh<?~&HESB5oSs>WqWIh~v6+`VS;AEqSrJt3b2z-JQ
zf*#|Cl!txBwTyu7Y%el*<@)UFG9LlmZ~U!;^2{8_W%w)<i2RDWT@}dWrfDrz%Ul=O
zB1XOWCqFETA#3M;YhPI2-B8w5O4Bmo?$wGo+H&N|E_xS1xcki$3oRe^>w&g>pm?8<
zyQ1ov0B^Pog8kwonAEg-88)xr%-0%j>w^q0)E?Y#X@SqI*%BxFRc>iyb;syDyutJV
z&q@sN<LH>lw^PrvC<5DKY**I)Nyww$?53929uRfxKo_X{<6SdBR$jVo=uB``oKeTY
zxXCM0vXa5@_U+n!b$#8jF;5Ipu%Ok<*Dt!P@kQipLMvD!R~BYk;(hv>m{og3TtF9{
zVrM_8%->z-tqqwUgJ^I#=`~Hkqh90HY6d`h3zJRihVUo<Ia3mp=(6-xp|#|*Jnucn
z;-3q!IkAqiR@!oby*ar5ZZ`VFK1DQEg}Lj~r83QkKQw9(?VZ?mroSt1_;Aqph082?
zkvAQ+@qyxqq?$cP6K0sxs#l&GM!SS#_s*zSCL@w)xPXwyKL?L9{CH(-^{boW3STgQ
z^>%ZF%c#HfR>sS9@t7z!%X_6vxIy=OHAJbPelvG|M(yU)F<dZA`JR=VOhs~OSq|vt
zxq?O}GiwaQ?=pFk3+OeDbC$;rzfI77nEz@x8B;*P<u1k}MVSUir5tI%fQM6GkdoBq
zpKZ#vk@uKiwv|>-#17UmxhbK9xs(7`no6q!oz(9^I_(MvpPRx9nUkK-%IgM~MprZ_
zFxCqStKBa;EmcoczZ$A+J{<jz*R)T=kKn-SoRsIRsV2TmEf1+D$&w>{8nQadaL#r&
zC%%EvF$GmoOiHiJo``W=M$6A@dyay%JC4)zf>O=r!Ts=Iu)Q3m+RruuuK$1#3|!I8
zN|UX)Im8S2E2XzR(|1b{0g}H^zA8=0k(Y<?8#SE>^C8ve!E}9SOz8KkF5l-lBE~TR
z8eR;oP{mE-9@qx`{sYlqh#E2$=8|RQ+Md0K(+*z=hMD~aPUN-z$CAC}3bWAcyMH|`
z0g48{lN6xMKgdD#EFIc_u~mSkLXTmmgQWu7;*K5JvMuHgYfp&D1`fuKl^M~T!vHa_
zi$7gU0STEHLP2FK*~c|iA1tuyIe|4{9{l67j6YzBFd;Ep%vDFp>(y_#m7%1&h@#3U
z&E3xXHyjgl|M2^@wAlkZ&g({FqXo^yfH4(DnOfYHZ%1qGBHuKm83AH18s+@yx#-K}
zC0_TZ!2Z=Yn4rK*Am}!;U^jxAvNBNxCgE-dSCqfYL(<Ir*MuVGlc;+K<&4|{*WuYE
za&x-YTG-Mkv&ZIzIBSOa0XA-`?qndcv%KhLJBVP=U+_6XU7#@DnnjYSG6?XEhuZNt
z%9eusGd0c|;$~$T;Gk>lVEQ)jx60b3HSzwQ=wWk`;7CsRnTJbj1@6Aru-6P<3ZfN>
za#v*$RW=O3UQ}GUG2n9VF4X)RyzASVI}xbpe0|9mGQre~gAdko(Guw7ulBngGu%5f
zp`1E-Bix*fQ)c1WY86xi$49;nE+O8F3$4vFrth1R5va5PJ2T1M#QezN<1YHV#j`vG
z*%`NSJi9Ho9m9sP04dAwGA8be7}$%cEDw|;RrwkkHlN~BE!&Z+q9&7(uMgzH{FV*k
zr(8`=Y3m8a0sW*YCHIfUv=^PHPLC9+TLv0mUHK<N#aOfS9j}*D!-ksjJ_|qJ3;n)7
zFqG3?>SKNn&}NgG*LM$wkmIyxklDWaFF8yi7260wJ0xiuRc(@s^1F?&hnTBF(8Hu3
zI*s5c1mUqG3&ogH<`v(an)%N#VIZu^eJD4x^cO{XgWU{Xk^5@3MERB+sn=R+W_d<L
znk@C4{0pod^>Htz6Q<5+8r`w{Ztm(R_CfoG{0*rGFqBn4g8e(1fv)W(C9Z?=S<Y`1
z)fsA2EMlSikW>o$ZJ71T*?!iWY^|X#PNa-fN8Eu1f{<*>*-ehm)X?Ls_0JKp5hbji
zP!!v8x>`5>-OS+8KnjP{P!oY=?(|?u_^x&O>!E}7x1!%L(k&s4SBWHp{x(I#8PqsU
z|Kvw0fK^OvdGm0>9#oExu+mU;%L?Ut8o_EKTAS>l=eZNc7R(=45q#f<-2}KV5MHP{
zi1b$G15z*a5H}6Q_??KNe;|yp?i(RqJ|c?dXXbY5aG3Nyd(e<Ue6?Fp(c7n|a5~?N
z9Y2P-pa?(5%<XW>p0xUIrMG#d(h5?4#>I}j0hXf_s%buB!NxtC57i#+6N0Uog>xmk
zmA@vBzZC5<T<o(4B9pSSRadf6b+RV*Dn??Kn?qUO$cKI~Pv&zA%nN=qtU6+TzuFlD
z+iZEEhByLAjrENUPBR%TG7dGR>*E#*?l0p~=*Os=I=40G92ZSR#b!k`z-YKB$-#C2
z_fNs;UBP#-Ju%vP0*bm?Qo#d_c2{<k*f}5KM&!cm+OQ)v{lu`uRONIj1+71u@S2C<
zeKX0F3%~+CV|#WSkfaF*N!~r4))P68yYmtbe?zvS;Mt@+8_r0l(!P48nh#o<S}2Se
z)!WPzzdx)kQ0BnmFIv1LNYtY%s2j!W0bb#IR{gQxVkusED7?5RbQ89NB)`dgpf@A@
zQCv}gn{sBDk*R+BR3iA`D#b5&g;J`<?INDDog1}Y$PhOw9fytyQq(SdXpcuucabgW
zw67R`_gl@f8Oa<ZQjJpGD9kk9t&PB9me@G8)skinEGb^146okH+&9@qSb1bZpqInB
z!M}&Uyr{p}p&|9&2!ukxn+yrWj&fg2qA}%r(A$oh_E1QoulyHH5!odLNeC?KKzG1T
z?h@^~`}*{<r^L$U?!4xJICS;Y<P_Ipx?X;=JgU$*Wr|ZAQS6T)AQzk$$5(GPxzu8o
zO;3){uvrGjwVWPNjY9`lT$2X8T!h{^940L`37;cg6r|qgM@0WEs17XU6o+eAAT*<w
zGSI{y+{712)T9N=VF#Kbq8e|?Ao#RR$+7S5CWMhvZbGBFPldK?{pgV``CuydW_>u<
zxHxX1Xvi#jzS1<JlU_?S|22T%(6!G?^o}<5?atmtAVl(>RaUZ1Da8SJIO`%Vj|XSc
zHRN`~Bxg5m(Ft=ii%7BPiq87qXX;LFrnO!J>uhKtAFPf)w*U2kU)lIq6F%a6Y;P~H
z+qBl%2L_hvU@>{f?K&oOcC-T{0;}HDA<APvBky*4o+T=F5hVI9&`CWe5y)0E(dw|2
z7EUTIUhymv8`l5SQ~m%v(?g#IKka!$lCPT9kE&~ltEQyK1e@H*N9F>CS%K#l=5OF7
zD?Xt@d{RtjnX45-CYfTlnn(NfqoC3JLYoh94@+)={ttKqXMx=la_sf3jl-JZ#pSQc
zikd;8iEcC_dq}5m*UCStckFW(lqP2e?|6eChjUv3&I47FHBi8_PMyV{v8;YyJ=uOr
z`s^Yb>-zDXd8r<fmq-71XW36D%;DgE9hQe^mg!|dz7u!!Afbn&L>neME%td()lCzt
zn^dtUgBng8_1g0F#2ML5j#2}p1>EW44OgKu<8AtuTS$r5Jx7ewTm8M0O4#UNHVN)X
zc0XL2?gB5KgTSQ<g^Q+otN;m;0YXbRrx2%r)bt}ViyI2|<&W-SJFtvCx(Ghb9i99R
z4Y?W1Q;3K8z`CdRcZ}riLee4Qw~NXtrniMikyXK2_KJ#cM+;4XeMT+*INv8K4eSGW
zWXa4Wdxr<5iOasYz`)_5w1cM7+UzB%_Vx21zKcP#LEysK+Ao!F%kft^nAybf=Em4V
zjQe!>Ho~7mm6f3>ISza3B<XXO(++1!I%A!ll9ZNe!64mc=Oz@E`kge<Hh8wyQla%{
z`BLJa;rdfSnL!qhnDi)I_m`Hn3$fc>-jFQqUmt*5Locl=>_He;+%RW%|CRAFV#2OI
zsXS&n3qbF^wqWC&-!JW8FA-NgnQqEFqo1v7pUyXHm+T^ys!@iJtCXWpC3JbnZ7tJd
zy`u>cMzQHR-kz={g|7Oa>tI>j+)QlCYx$fubX8S^h#yfm>5{-X_JP}4ah;0DknB0w
z3sOI0@@YSe3@<+AwW^`2uy{jOchPrAZS0|89<cd0KhR(EXi9`eikIj6&>oT)gYk=2
zEVXfco8qv7?<0pos)Nw;y*0HfOT{#`ybqxD=MBp5>l!=}^RQ?C_)3_pM^d7i(!%_t
zy2kUz7dww@$DhRcjdn0AQ#7FjI3sVt_iOP!;uxiF+hI=bI*%qlv6CYKq1C_XE;~JB
z)PP3CZsyp^a(@IudqRY~E_sp0G7znCo>$gt<6&M?+{STt@RCg6fqufFgsq?Z%y1EO
z!W2pIx8|n(p`jw!Hd+$m1a;;an|R0Ka{TenjylGrSV$vXnoF(;uF}T^&tq*P`@Z2h
zJhT(f?1X#U+3Fq=(-kM84@m6HdQGqo`v71>h=?y>GG=`E>rhjvv;OL$;J2&cs_(ac
z=YwzMra9RX$P^BE9O$;x;=>i%N=f)~x|xcR>Ao~;Gn{PfoU=>c6%!IY&UPEO<dz(b
z9th)bZjoVa8J1$d7Z3*dBqe6gJ<QictESuu<z=kVUTavn-f&=>a75=|^V^Z^MFX>o
zryh*Y){miTHs+uWwU4g<F)H%)A!doPA2K@Aqpnu2?JYfOh#-(brpEVVMFi(@h7%Xf
zwd}~m&70fcxVc3ac3&t};Kn(>rj>?;<MAsnj(f1ctqW_7bN%BOW08RYz_G6ReATiU
z(p0yad8hneds(h4(V(5$^NLe4Q1s1yQbDmyH@E-;r{jfpU!xHY6nr?Z>v<beThkI=
z>R>J-urXxk-4>INitVw>gw@Qj2A<5Z5y|Qk^|4hyFTjZhIfM=s5rgTSw6t{ais3KU
z?i=|j^%hbNN{S`A{^p}N<N{7SnljwO)O;qQ_(71Zm*F`fLAl@JYyVaO?Z2cO2p=`{
z2cC*{-cuNarHq~@s*0>FF6<dx$2>=eEbf-IMPVqPAv7lyNnfT3wK`n<9hMWLiMr{5
z*)8WjZwB*Y{Zc~WG81pgsN2A-;>+S@#nnuB%oztOixRJ=M~z=W`M$wf?+WR;sR@+a
z&WCMgR=qm*3}l9ixm3`A((|a7$UCvt6EK3(zcG03CGvPCeoLp`2~GWGJv-g5zq{{C
ze7W&1ew50;PqjST4p7ff#m$wa=$JweUPkYj9-D?r8qQ<f>*+5QVOFC~<9H$<wK6;S
zm@$)!yZczU7ZlYsRb@nO1)eF7Y|rTMdVj9YRV(6SXy%W>M6W4e4QZXlWAnEY^%bNU
zgY>4A@P@6rq3w=ZUm)lur7Hzt?iu}TB4+2oBJ%C+j$))zrS90C&wWeec8NHrg5&lg
z5`+3hhuIM?S0Q+&yYq|GAMZH$m)d-ryL*QEkO;A_5-Sr^8-L{YOA!)#(ar=t`)<Yr
z!!d4aIKQ`pXW}7{(p=ic#5MS}uG=$2lJHpB1YPo_Y+06}q<3vTUob+%G<fkcI_%!U
zw9}+rLn#46T1lzwU8kXXBdF_1B6^%GO5lHXM{i2M#t3!$qpC1D^%jiTlq~EN@jQ}0
zxhKz^XcPCN|7l@&vEcXp$xTUWD{CMBhZ0{WQQJK7;8qyGqagRdxx3{qu{acbAXAzj
z4R4%~nT>aZC(_TdT}c%aX`DF-Ik11EEX}BL#M}c<ok;2*lWK9>roDQCo)aK#iN?eF
zwKLP;fEZqm(u~FmhmRl~r9Q@Ctxn`3N%8*Jj|6oFPcRe>A-OdQTT@fB85uSbip-{K
z>b%mS-x@F#cHjTE-!l_$?3gM>e{zn&wmPS)1|P0b_DEjKrG#yw>#fN?+9(F+#{G@4
zAo<h@f!t#@S#^Z*7j_eL_HG3aMp+uqPtb-v*7c>`nwh2{iK$cBt%O@de3YfAz@~Pb
zdi`VAi>&9|z&3e9%}!wEsfmlWU@-E<a<huKpDVENHq-PB{5GCBa;3qmtv|ID#3H>|
zceQ(IhSDC>zufAXKksNx8w**Mj2bY(fG9Pp|73{6em5+E$&!51d`m8{LGDKLk%fo4
zJSVT#8CM@d=~YtSc&TzbeP1wv^HloF$n`*~*9|S?BeJlfge|vTn5(7&jGks@2Gmy(
zAz|R(!EIcC&$gEL`k(k6eZ5FGR3_gMY6bzBK-KQD?__3Z)dHtbYFlJN<E+J2Fo|q=
zhxa!}Pu{mpTyW#$Ywh-Zj`?l;LIOzyBc8`^nj*uLOy!NeeS@FQ4_p8G6pgWWkrHwK
zCMSVdvv`sLoE@Y3#brq?+(LXX|1OQW3p4;18r<6Y<N2e*0}_vi*;dJeMsXW3;T(Yu
za}4W053XzC<N9sxCrQjT9&;u3S#e|ae`rh{C-5HU#?0M@TW6U+CuhLBiO9bL&>Iy!
z!<R{h-j{sexG>4|>!B)l-VnV+y42gM5IPyvnM}}uhGlE#$o>x|Wf}+KiC%j|J77@=
z9s2rOy1w;PXD011{U<E^UK*ZEN`=~j4ny18L(}IkCi!VOt8_3DOmb^sT2cuj2FeXG
z^n@iY3yE`jM@3Mxlo_k^(8fF=DHY}GgRSJ~(TF6wnlFE9$u+K@I{hxM38{jMhNk#<
z*DKq|7i4oc!7MYxB3=RS1V2@y6`=i>eR^UM8#&kIz68|NgYmvoyU_#Vd}G%DRi5?f
zHpB_Uy(Iea0xHQW_LZ@-y5yS|i9Uo@DUCUfz^YEzfzBmO?uN7`%m=pSnnJ+-QMs3W
zUOVoE?Xo3un2f~1cQiT2uK^Kpm}*W#UZ`St?VI!{L(*wO7eqqnmNRY#pzYV@Idbjj
zQ`a$v*(z#`AVRPoVKB4t=kvUtem+byLt88ylt1gwES0c&-75GoE|6y~efcpuOVop(
zl=S%6(z2xFXs+7M{t6i}So*7(CbwI_7eO~gq*@C2i9QmU<D&Llej%FIUH@rG(ve!D
zheWk;@W_KoUtK0{o`}kUqcu-aR^v~><krPZU&Gs7U;hRc$F%-a^+cl)Cr#>$Yff2_
zhoj_VHcF<$RXAsSb}@AnIc&A?PLunr#j)eh^HXiaafCar)OeMmEJMHLr6YKvpr5t4
zN$AG|*I$yr^yh&IVZ)0nzaww5h$WeDv|qI|D!4m4+n}G**x1Mj*v9?)eTK8<*G=gi
zIYtv!`fE|d&lc}a>e&6U!>nt+<jj0+XPaqd?7h4<v&<u#UVGxt-r+^?OHPoRX53`x
zx@L^8WVdKn**Z7IjpLg({;5}2g1d*?KdMKe`^Q>Y$nt%<td|?pua6&w)oh9~dw-Ho
zxxj?lD-?Pb<T8D@08A?8ygnF?%5IXV7XX^8FMQsq7mvFUX?xZ_7<ET5kK2COkxkuo
zD%%3h&ddYD3ufiMsp0=VSC?wNlcA_f(ool2nqTB=a6?!N`ZoB%#(RouLXE(h@TX^1
zA;$-X%V!2PJ_i9jv)4DJ#aAtF-SL{bWA1%VvXaN4vb&(gdth5;vwyW#m0&NAwr{x!
zpUi&tuBWmamfzz~o4?=o%m{)1z=Is3CUz~+AM`%$Yn}!%d5A9B0Ryl&ae*Iv4+axp
zc1wK``R)wffF<Pg_d{Bj1kU@qQ$oftK;0&vJ=HZw0gu3LW27)fm&{sH8~v1S-D6_s
zT+umli*76cknGEEp)VB^4Gia+0L(m$3fk+NdEd-r63+D9ajcBZDCHy=qk|zDH&j(%
zEStLkST$kNZbjbDdNb`JG{2&a_HAY`_|7n!sw^O#&i26VyQQf8&=mn(`7q-jE3!P_
ztqBQIR{38l+a1pGd}8w&BXZKIy%a<{n17aYpwO;>GQr<A$7=)OKmdY@o|f#va81s9
zkcAqHe<}tK!TuTf?@i0~#y=A%`Wx}(t8J0Hc3)}3YA9s9NXd!(92x##=v)akS%ijw
z5ovT+fSKNtmNzVGZh@JR^Zc3g^A!j*2A?IoaWAf|DCucv{*4EdVIQ{@4fwqSBb7he
zUxonEtEymlf2}BtBPQP>02ONQ>q$Y)=x^QSF011AHSZsKH?}nB`M2ZYY=yA2#bY_Z
zEo<sG8Jjoc*PfmfFJJdd1tVS9IU8nyv3k~PZ4d~=laBUQSKC8$vs<!I<AEj?jUuvJ
z6EhFfVbod2O4boh-a63jlvDZb8(J~&0f<wW_>V6s_^AasZ*$=)Z6LY~AVihHc|$_U
zZp$m$w9*(X?(%>Z?=`01J4sz>^+Y_vIg8@Mrtq>PH0mmq^&jK{SKKx4*D0uIkiHut
z{hpt*!CWXPsa;v1<7de(vD)muri-i<kHly_vIMP^%!~s-VscDuVicG%6m<bSyPBA&
zI?0_5w><k~spTr1A&g<Jr;o(TH>6rSb>Zn<Z$!>`z!Q|`>F;9y`J=|UazupvQd6Lh
zs8g42iYcEmd*<d2S_%0UwUBC@e?fIUf#rvi5?Es!Xi%-h<28Q;8vE7Lk`wCFj}q)K
z<?Ha}=vbCzSyRDX%JuER3n5@YK07rtub`%-tfQf*p`oIqqpYl@wl*<8vpls3U&eU`
zA0SC5xZaB-%&)myKra=Jeo?{ocP}$bTb8(%2)ZCD;!jRX=E`Jsee~54ysc?4k;%t5
zoC%e+?^=m=d&^NLhoUtOkV9wnmJs3y>;Xob6$gF}+S&$35BLkhb7p{iH`g$m8f+!9
z4eu5gWM8rS?JnmjPB0SBx^C;&+ZyeYbJRAr%ji>6JxzGeSMjr#SknYuS6@nY7-n_j
zuA@p41&>^CJkI5p0<ujm8RYcH+^WFsikJ1t>t3p#ImCnGe;-~G9RiMP3ZimregDoL
zeN#eC+rYGJ!g^)kLAs^K<NpQn3=Q){UL+9e5!%7#He+|2!ki_ENA7p+jn!pExfxF)
z&$;h1l;@@SPeV9!2c)9Do|6|gp|}OzOnOmO--1e!&XRacvlZkq@WHF6s~xspyv$5Z
zj=y@wWrxm2uC;R=>UV&Bf?&}LNoZI&9uK|!qyQ_+g4hYiKx7gI9e)JEs2@YV_V*t?
zpdANzF$w~^&vv1e*px^2qtBn%Z>cH8wU*>~pSBc{)v-F@106EqjBoeA=*QU-k2*2H
z8yz^=o9e2{3UksQC&b*k;&%k3Gh%F{^GuBcH4)<E6;d*?bv<)AF1@%M8*`oBzrSG2
z<J3=Jti#X!8{WbB4u4B}$Xy=Lp7|{CZsgSq-UnfyD%%>8;~g7>pl4<iQ{Cd=e)f7o
zW`0RUJxER7&yILvP6}-CZ4EW$C51VukM7^Teldvjo~dyVi(W>{!qM|w#Dgp-PzP51
z_-EWJkq|aOsfXO-tOMHdo&(4__^(gnL%khMAohZ+XAke*x_=zvfAWwW)|pOfyd$Ge
zh^+>{oGyH`j(n8<;&nx1=fK3QiAUOq@mo!H+M~FrtD*iU-458AYAc9v{(K%|PCzVr
zQDsA$gWl&OlhSgE-qyADj(q%#H_f5Uqgr(+_c-a0^-26|NB<YrZj+q~X%9Z*br>m6
zkYw!B?bRIOs(&U(eez-2i{k3mcVn|89(m#f_nO|ey7E^qGM*$vU%L=+0?ytUZI&0?
zz`S<ZT!;D{&_z!}$H*t6Y3gvqH!S+Wv+SbshR#9!3F`}fPsh23A2v$)%)=e>4zGW(
z$E^5e#^d{7$@_U7ak8^Cgt3!Qd14yBK*en+>Bqz`r+;nf=yCRHOj3IO+s2;J*%FV;
zNevHnHCDdPO-s6WD?G@@{Q#g{Ls43oXFV;&lDQ7`JK$b2)EuIUdfOe{{VzquKg|KF
zzHj1BdEFO`{HZ^F2=tVj@xFWUthd{NU6#gL$}*xGSm{WoJTU{9i05J&PBCR88|M?j
z*Y7^gDu1r+9GM;Q=$zE>V1HXJi2Y&gjSGIqUG`X+Xl=suFGxZP;9H<`?JHKUX5y7n
zGv2Y^BOp8~AvN!9V=q#$_%@*@_@C}DAs)D}`tw*%Q+Yu~Qp~kbU(Z8(theYWV=i+f
zqJQyYtYYNdsIk@F#XB_Oeo8@Q+t93upMW))cYn<lh0j6kA*bE<+nVVpN%E1(+{_Pv
z&rngb3Ce10vDxq8AAb8`=Ih$7G4$OT5dZ(a!#B%-ARc+&>8vfzPL916?0wY9)?80j
zMwpA03DM3XotN70LA2s{gKz$6ZdpTb-@xE7`f~q?DAZ!-9JazoM@NQ-hXw}+`ulsk
zI)7T>V}5yI=Hs}VVP{VscHC)(#7;6pL*E0R8>C|1AfcpVYP;XVFD&vwdO-zJZbGK}
z@jmHk;W8rlkr6cZzTTd$j<%MjhPu)hDe;kE0bVY9ZOru5<i+{enCWTY6FmvWP5|%7
zaBw*hc=^u5jKa#=hGx_*a&#OW{+z;db$_J4VxX_LyQ`zU6@A67DldIqn3tQG{^;JV
zOF^fOIXhTw)m4=h<7Fe6p`q`AO$slfq~{RasG@6X<K*spJ|gaEZfONFS}{0=y~+ae
zF-a2`JUr3f^>lT%x3xCI;A<-1mK7Hjyv%u)aQn)+Q{ZN^eT%M|yo3N3GyOU$ihq?P
zx#$TX?I^Np`sVh|9)1^Z#y!c-N6SuK(+$^{Gx9jvb=}t7P*+`9{-zWJotKlD_AEIu
z{%-X3i-A7J4(_orGt^X;731e%BAKC~?}4e5|GsM7dL9uO6>XCpjz@fguYzTtTZk!-
z<Hbh&odPU*q}JThSclXqON(C>f`8!CpFK`Wh>N@&?CWu8&kj?45WBb__j-CLiA$P`
zp1=|cI#ym$Srr{qn|()of-XlTB&TNOy?$HMgjCt#n86ewanRS((b7;``KII*Sm&AP
zDNi0fh>Jm5G|u^Y9X;T%!`w(qRUUIyB$c6|?}6)eub^UJ;}zMcxY>C7ZhsfA0I=*I
zrDhi*ZQ{eDSj+Wi^cj9+sK2M9wGoNEFz-cnW=85WqCX|T+Z`-<bFk#)rQo%WfrJ-5
z0kljfEGfU)aGU)B_fzLC-GVX!`6X42xPH}ieHN)TmlnOu&PaKZlyDDn9u^vS#^<;@
zl#AMCqPJN|MgmudB$c6|AAbNFbY8lent_c+P+Ue;&upjDQSZQTI3bW%TGI;6(2?!#
zGnTCm^mVm0)|4Z$r#^f9Fya2)JGZWdBf+~J+-qyOMQ^j>MhRgMJDf8_rtL_m68!@Z
z)tZgU+FPs~oR6If3cq>xQTof0YP@k4={W|xzeDE$^RiMOC*F^Vx_@!yVo0FxNzbDP
z_wBabZf2~jp)3a$JP!xB(n29%QW+Zh5fHhve=lA^Nlnkf&LbeDsAaqz#t!!T#I(G!
z`mUaL{pdl5E&1MVSaW>!62$&6{@$I)h^v<_ob&VXJaW(xisR{Ps=$qo8!AA<;~g2S
zp8H|3p8=bAW@h6PlYdjy-D<V#z_HUIS7RQg=e@x^l|I6us?i~+aex#Z)1N%J8+GII
zg&;q$*bg~5?6fpB(%G!60M<Jn_|_S4&39568u}p+b%)=VtXK^dyVh?Il~vW-YQ6iQ
z*V#+A;RUy~v%9x{5YAP-2fTwTU1xg(T7G&Ed;8|KOXtr8_<wqP9Nxcs=e8~SnyQ;*
z(Aeo|Y1UFA6PB2e#9W8@6ToLIDB4BM#KA8vr>bMTZMW;mpeuKho);lg<FH<bYg@F|
zzA4I0gEg4&kbqOij~;Ziw>CG{+pH`nDa^ltlNCu~R$&qnBr`PhA3=ySQZaJyODL%8
zn_4@01YC+q&VMOwY;Es?%#VK<9|gR3ceFKEmlbBGCdJ(b`#V_jUhapT_w3weqN~11
zT0~$27n;C9`3z(*bXNM$a|ArMSw+pr&MP9lN!?`Uf#ZQ!W1qZ$?eZP(Mm|h^em{oX
z>l!QIt1K);_?>ij*=KKMW~imAAdTGRNJR8tcEH7GSASA7aSBL)WpC+tEZ|yPYGF-d
zOFP<`0yCLz_*U_@C?_Q`_Gb7wbfepCYqeDmju&ySpP#W4x@Rla(4ulS+NQP#JpIEW
z6CS5#y@X;BAb$fXE4;~PJV}U&2oFAU!tKyLc(mzhD8r*|p4?-<2d-<f0=DD`Ny@`%
z!earK?teVWeu?*!)>M_36~8LT%S}s;kG>uj<m=_;dcbj)&2}>r9r(5LvNO)l*a?|O
zTqZ(F8Dw@&M?q#6aXU8t(bM#7ki9%l0|{$l{GEu)p#i5n4<FcLXAN#NTIk`%1$7Ha
zH+lOBaM|)@6f}(N8-yfeHH@tt-A;$zxEufAVSn<Il(gqr&(l+$CMP8(#6{n@6yom#
z&h@>!c3PQlG1NuxbDS)6RP!=+BKW1NFqZ~o?%l!J-N!F5<l>bZk$2)^@7%r>aqUX@
zMdTjq3{G~7t%kZ9s+(jbMFqjx4pq!ZMD1dh!1KCQRJ6?OJc45K8b+252R+V&UJSo{
z^?!QAt*GdzTM;*|UJk!_KIja%%^lclXT8nT#6VY5U3rs?xDYQV6uF+CvE#uLdPYK0
z3hH`Wx7#}%banGM;d3V7Tws8o?<wyS$32d~%+1Q&#6U+wRZ(uEgopqy7ds0B?OMu}
zFm@8n+J6GvlV%0QS~{p$hh#R^4$f{Ko_}5^eNOrMfyht$oCKjCJM6sAVJFz&^>sAi
z45f?|OxoF5=xNr>)7bIg@$Li`PCgMS1ywDBt>%{2Hg*nsot*bOIquoL%g)wjyTw)`
zJ*~|uo8&f1iU|wwg1i-)R-u9sq?@(>1b=d&f(_@Y+D2y9HaqR?9d_?=+~>G=w||4Z
zo$byYmT-I50w1~@;9N{tkRPsjbXH`Z#*X=mqG{b4>UE6kIe7)eq~(-U)HZAC=ouK9
z7#kVt>*?xffu*jbAS*2)DkQ+i!^OeI!o&cjfsiAPL}X584Wt43H{5)UnPUS8U0y{)
zQ%hS%S5M!-&_G{LPe)rzQv>_VVSntrAb56GsE$Lu1}Xr~*Vr+)Mc9(Fe8npGq`}C-
z#>vAcAS5gzCN3!@EiEM}AucW^DguJuzzrB*&&<R?PfJ5hwFcAw`yUNyNdjo>NM{D+
zT3QAcHm(f<!Xlz#V&W1Kk|1&kaWN1&8apRDD~O#D1iubMPl=uG^EY+~yjCR#)$*dG
zqNZ6#OGih~z{tqN#K_3NKu=Fk2g6-UwT6;n_3BkCS1ezSii4mYZ6yC-@e|;k8GwDT
z=}}Ry1A)`iGcbVQLF7p2-^31rN2ABgs^)X-lkd_)lG6$q@qdl~4{ok<Ao=>@_y7O^
M07*qoM6N<$f|@hZbN~PV

diff --git a/vignettes/examples/vision/autoencoder/unnamed-chunk-7-1.png b/vignettes/examples/vision/autoencoder/unnamed-chunk-7-1.png
index ba436d5fe1b46f6d47be16a09e8bf10cc0e65c1a..e86c38b7a166d775ba4176bc7676b7f3199bc144 100644
GIT binary patch
delta 102577
zcmV*pKt{j2r3TfZ2Cz>9e|S_`wsm76?(V51RjGI?S`w?cySuwP5kiO(NU#7QK=1$o
zg1fuByITVdP5aDy^Phe0ty`h#nMXH0-Tc$Dzwf=FlT@X0_c>eFUi*tc5Fd9Jt{uzF
z$Us+9RZ&(_bo3XBOsCN7`svGeu9(q~8}4kZJmL$1jFFI4*3@H|f3eIMG)-kWi7{XN
zDiM`aqnTLq+ygv#HYPN6IWgh4Mu~_^$tkL6=xC@a%1Mcj9rb^{kBy^8ke|1^t0UWv
zWx+JiQCE}^ANF}fWmR>VR%};aH=eD9k)FDul!UnG*fFC=j`-*A`&W_wD*i7j|5f~7
zRQ{{@zo`6I@qbYXe+KdMbm7`sni%S8swv4!!xLAavjb8p+WXed>!?f%<T4b6k7~4-
zw1S$pzLA-czP1{E&?82U8Z9zbR7_k<bganeQRuP4SH|G4pf8OT88c?|s8Q(aBhb5y
zl~LBFo7+0NJKC8uv{ht8g`Z`Vh?t~|yppPhs*=2nq?pKXf5qc>M85>R8U79EeN?m=
z=5|h=Zmv#Tww<+=InzK_Ls?F8?8uSCZz2?LWa;4I!{^zWGj!GDC6I`Nh=Kf3&-k;H
zFpvN*SFVk@5e@WCJaIXFTd&yshN+9EjW3Avwlh>1rg%{)Ic0TieTKe{y0V;<*yt}t
zjvgx}DJ`p@f21HMEh#p3%t-td329k*MFn{|8EJIS#6bO^`!7VK6x8*MENppp7DhA;
zMX9mE&oXk%SaB&?q<JNISt)T*^gfh~7&QjHBl;!Mk`m}QjDq)3)Swwz*g5hz4tBQI
zRu<+=bZ}J^q{Z<u8%Z286&*tHTpKgGj;frvgrt<Te~hf0yaLD{_?SOGi3ADqcI8^b
zlcAottge-7SbBMLPisX+xEqTmCsh2Xu@ce>Nb`Ey>M9B{5@ScBuZT;_E2(MfYN;v9
zOT$-2qp!%JtE;1>p{9b)gACR^QpgD8Ddd#3Xh!DN=0^HjDsmDc!p|~tw1}7l(!3)2
z6cS=0f1`&fe&iTYbS4y(Rh1QFr6u4KqW4kJ(r236+E}x!EG;d}&6tLIT53wjQw_HY
zSn=Mj9BWfUZ53Hj2`L#lMP)S&Es#I(F@Js%4HDwR=deulH5KJ$$R{qNZN~GD%dKf2
zmlq%4Xs#nWOz~il6_nL9)sPafc|gWrUR6_%e{M{tX{jn8<3C~y`ihFCt|60QK-1Aw
zS3z$qEh#=0-WfcFs+JzzgaJP+EjG+06W1E420jJ4*25H!&z7>9mL7SwMv?DBW0(?R
zW`-0`(^gkTXHgtf51Pk{x3G6cif5S^XemqM^N3EU0R!X@e9WJp#DRqQ@i|sZJym2#
zf5izyDy3m$=N_6=P+OQB=3&p&loq}M8+N!iigGeYndtI~NXn~f8yH*KS(zB>s3}N|
z9f7{0qOH%gu(2{XVHoIYsSAG?84MW(6?JWzwgzc1C>a5vf|$6Z6nqL1Fb|ZB7L$@!
z(a<$C26HYiEjF5b9}OLS14BB)h-qwMf6AoOP*|uW2i`ywzZ?|LK#KQrVY7_&kzW}j
zE~AL!6BfuH_mn?FNd$@TcjH(xw3W~U7Zt%lucWGhrK3+|YGGQWualLbnw0Prf_o`T
z-OI7kN}2{HEDm8%m1RUnh@!9P8(TRzbM37yOi(l+jMtFSmXJoVyPArk92o6ke-RuJ
zz~LZKV5|iKsf30e!<1#p(9=|wMX7|~ebhCzb!i54BV(j^eRRao+X_Dzp?D`R7Y8e2
zn!1AIXmOO0paW^k1Nq~g@@FV1AW;GC91FUpywE#}E9sipx%x+@MEUdWP4tu{gs(vA
zf$&-!yf$G`MP=3Wj9FYaKX;xrf6CD0#Yc+Csp&x;<>TSPvu8oPtqE5V@`az-^Wlyh
z-Wtd5B1CvL>^mZo3TnCxGi#2G8ADH9K}tmUK1laky83h@6DHk2Pa8QPta(vzD}>@v
z<nQcY$<$LrA52_MO&1w|Cr^++>KT8Qk`59R<jyfSP?Lejjl<zD#N;&%e=QvNei8nB
zwk2ItL44RtMiBZz!SiVNarDCqnsiH!TR>!>JJ$*s4vCTC=qnZuu72S`zV6O!8w;ke
zLLfLeV!~7Q-dm3x37H@HF(?@;t*m8eYRz>+;klufvdma2zY~*`RaDi~(Px0-X}Tzy
z6uK2E<#V#PFw#|(6B{8ef3IO+X6xjIv)Di4`F^&N2@)6V!8W6*ND7}d6%V_hyP~Tq
zEBY_q0?dQ4ovR-*4*`4!Q<{p@NJ(W~V_Rq6@WdFDD!OoOp}3-<EDyQgZxdXKVR86=
zZXA}eo|@dS&oc(1JQP}*Fd2sWdOC#W2`56-4D1{|o$M?aI#4qZe^bz6u(%#Uu^_+k
zv;1cW4(>v{9IcJC6^C8xu>vQ}M&YoglJu}_o4B!J;u6T#i^8=XBcrBoX7B13kre6A
zcQDgOmP1NOn_<Oq^9v6N0Bi4H1Jxl7CE4L+wGYcRV7*1fQL-*8D<>x>FE6hE(I#QE
z;l`WUaNT&eW^`@1fAN$+eyB+^GPAZcGiDH89@0VdvJsTW;d5=x>AJ|)D`*&)+B$g#
z;bi9D|8zfN!8Xs!$<9n)Lw;B(3j%c=LlY}|mMNVG)TwXbG7Hj}iZV*jWuZEwplN81
z+(LLlxSy-NnZCO0SSb|kA;Zk~_4ah<yEyUK_7L!^%1es8fB!4A;a5=K8bxIPn!37r
zG@8Bv3fIZRd^Az^VVK(>1>(}{FvU|9AqN`^6L<yEJ3cRCMoB1W=$qQ`+@U~a3k53j
zG<ya_fcycE_wyAG5Jx+6`uo*Wl=mp9qwEPiN?i>UMT@*wz>$H1I+#^;Rb>S^>=oF2
z->|q)9~XNwe*<)aq!p2x%&gfiNSr~OP*|=1zU2@@?$+Gi#V5?y)xpA06CFFjFSWE{
zS=-p!**mb&vC~ykKxYCs2;f>H_5G~uBe3TlE*u-8D1=WVRF5T<wCIq=hxmEA@=&r#
z*QJ|S+VNaH{X;?ifXDm!3LC`Q44$vlu%}7XWp(t8e~@oRd9(PK5mc~+)uV&Xgqbm!
zKWXS2BlqhY5)<O>Y;R_$sfbPnuB}+w+OxTi&Mtg+p_iouXN<m)sYPi~WX34)FD|Zp
zcMnf*A76hrt~KG&M++P|{e!ZPpj#6N**Y%MNJ)}jNE#hU>~Vtqy!cLRTT2rQYkRJ<
zo0ne@f5;#5fIn9;1EJb2lnAV@4oZ*YWl*Fvg3`Q*lp<kT*|t_hnH1$&He9|>aAXkj
z8Rm2?B`B_<9Ls=VY|gUdIPpDw13Vmm-iwpbHg^n6D5#rKUzil+WT7jIRH&|RX6Nh`
z5*8jA9TS(3m>l8fVrQ(YBq1o*Xw!5)DEpvze~oU%aSx0R_HwpEA-^K@VZ_8h^RoCz
zAdemq65#8}cjlsV=;Xrp@bd8o`D32&XDT$14~uAmeu;__)W*fejvh&Ao(lNgow;^a
zrs&W>(T>OW4h#?Q0L7z=Ehn$2tO}(SI`ZL;ZazW3kQT}t+WRIHH}$NX)m)O~=Rj8=
zf3gD?lpUmHWaZ=)6qS@^#s;`p)77Pf<r=8|O1#&&6Iaq9UKJJS&Vk<{3WD2jkh>vg
z5*i;985R=g=k39Fb3?Yo)63fj<PUnppQosRK#d+JzLend2n5)KyBve_dqVT*aT*TG
z(ao9q$n`on^1TB>{cx{UR}DgP1qDS#e<dXd#U0)K!{bA}ootvoO2ZVdWa1K8&@^Z5
z;i0+Bg;9J{6=eJkt(^U$(o4!JtH;%kpU~V|l@sg7HPe=X{tb@z<*7KI5(yQWIoC5Z
zAvY=9$BD(zlp7<0(tD(NadAa0s4w_OCMU+nMn{B&1O@o}`TF{JdwKr1fFlV)e^qLQ
zLiNANZo06^4L8iz#V;Z`HayVV6>@Y_D_f2ea-6|FDBHqOx4fLJj5PRKMQtM%&oekC
zEhf;-!SuZfpsJNmYF*FTBlnLD&2C8dvsV8?TGQCUD>A(lX}n`{*R+|ldfThBBHgWN
z^1~YZf|kGVFHkdNx%kIqlx4*RfAVdOG-b)cj<|%Rk`5F9_Vm=0<iv#d*chyRe_t;T
zH;_N*5&wTGaM4F%Ar&v7qHn?T4o}X{OoFx^Qni(h1B#fCBljk=CS_>MBNH+TU1EI;
zu2)!MZc>;x&%!`eQuxLi_F;vSR~)?iKlcx>>?#gtYkeW3W8oZ}T--Qqf9CAD^A|2z
zzH)VMYgvjvo1r-TgN%5;rbqUZk_+kzQ^GuPKZ$DgsOXv4I{SnrWMyS$W@MzNr==yu
zMTG%o2ht6fKkgy^{~;rW3t2s<pf4l7kWi+Xu|0z0vr=Ni{5_oPEt#g4C{}jjd-%9H
z;-nm9(gOH}ye8d>=NS^0e*w1M8rgaY>dLCI{4yGQH=KHUdQ)FhW`K<b^nx7&;`3`e
zr*u!9KC@?b@7zf>*-;)g`to3b;LeZ1<_Fh_DI1_-VCCc&ol)9YnjY!RHqnwn-x>{$
zN<vCSmuc+?_9i1UJ0~~4pr|-MJ3To*Iz0Hl!QIaVU#n3<A=+gUe^SC;C)vM=N{x#M
z^um=rh6xJZ*^aK*<)aI)A}1${MMg$d&)Ayh5g3&m9gI>vbGkMFDI_6S{6g8xJ+`!C
z;nuU;7Il`#dzh;tQ*XldiOMW*Zf<F5X>Dz5YpcnP^XGw&7ofX@=LVxTkkePvL&q>Q
zF}HGJMP{@w$5fzre^K<Il~Fdwbq|P4O3Td2FD|RBsjI0dDagr4Nr(aYv!C_nBv|dE
zp#w8!G!~?GXu--0F&g5?Z(BQhhsFbNhXcSJ8l7ooW$VBL;GQu?JYsZ{qLQ+T4&BVw
z(cLc$#RFch&{`A1`V@_L!5KAE7HwV9U7Ho+$W%f`n`!3}f0kTWUspS>rn;)KqC7h;
z(A9>4qIwi;LUbnucn(pF7{JI(NI2mXMyD5zYpq6+AbCo#EkHhA8(Kv^p>fF>xkVM@
z#y5AgOsGe?FU-X)ia+~de@;T0cA-ElEvKLac@)|C5cJ}rf&sj0&d^iW(lazNwM32{
zWp@lcbm1ZVf5reOT=m82T!_CH-;s@rs+vRvMvi9V5uIB*xxcF}FUHH(Kpur@bZghZ
zxU7n@lA?m#?9B9(D1TRb6I~P%LR61?u+;g(&<ssoLt`6fpYWud@`g#ZBv=S59_*B)
zqNbslos(x^RANSc>9`4PT|LvgCbc#-jH@gK`LiGPf9E8EV4jq*L|I-{T}yaV$uf&~
z2qbS7OqvGS6}7Migf>wMA}Tm~1`Ko_*p7TpKM1AWU7dJrJ7g2Iab;dw%Y@?_lU3PT
zl}+YsP|Jg`u^=}SMPbq5Awmc?&g0RWs{I^uXMv)((D=;anu${y3X_7Ip%6s1krXua
zO{}?Ye|}-{=>?VJ+o$y|>F=98qkD3Db0f%~{jfhHp&}*<f(#8bHBB9wAx%q7Ne=oY
zqO!<wvp8;k9=HInrHo5z`gCJ+KwRjkqn{cpA*-mSMWdTouxuT;POk1=K9D?nxV!OP
zsdgU$@`*?(OvNA{6!1$Z>zO*Z2gPQl#7Bn(fBLvPW8Z*b7o$YP&|y+hhYGQ*l(@(!
zVSU#pG&ZfEqM>c-gp!mH7fV9%qcK!OUR{rge!?p#CcUVpxx07i(BO)t3+K(5Ic*Ba
zpZ&ByBcY6>q@)BE44>7~H82ulBPnzP3hYb`wN>Q+2C1%1Lz-ujh`2H0vPv3ybW<yP
ze>R8f=;Y!msF(Tr`nhq*hC0T2u=)Oxfcv6rDTt%!L5GPAcTA|irweFZUmFnIq9no&
zehUTdC59z9e7~@`w7jy~iJdbimZgQcTItCNasF~@x(st$o||7(dU0LHti@}0ZQHbN
z&B|qq=J$g9*-!g35<vzmEd>I#9e}s!e;K0S&6G}p-AIHT+XV-9YVwj}#88x_h8#W9
z5czLKq?VClGRj&8CRPs40ztRl)5kwBBrGyOpk@q=XIk2$gwR-D18jgWdhl`m8R~bm
zs3iIw+B62zyb?YN6h<gMJrDWs_U_rOl^GF$1Q%lW<W#f`P0=UyibyLN-#veDfBUgR
z`*&~OyngMf<sg6d)BcQv44$QBK%|in*jA=FOV~9h;cqt31SMMINGotYbPcKYxd7gZ
ztUKS;*@?#mNADXH79F1)>E~*P<287d5;z1(mk>Re0xk-=Ee0A2;=)=D1<IznbGU%$
z>>Hk(Sx{Ef*fwRx+>Yw3D0gdpe|aiUA&`}pc09MRw6cjamu@<A`TXhQM-J@Sv2`QJ
zpZci(Y#9qefg33d3|2ebdEt#mwq8YE3f*Q|6>U1S<0$wWIZ&YmFlvC>j~OX0ucl`V
z&=)TcH&<szF58~t<m%z=9~1_-PAh!lg|8Ao<=<2M`@j&=d6?jjp{gJye>zH1NymtV
zlJuy=^xTr_hL+ChO~}-{SsN(gbQEe%qSEL!7#0v1ms8a`bJ5VAom)4pU$bgp$wH7n
z?V*1T0@qra!m(4m85vw^m!y0=!-Vv=VxuLASeft3$M82<Q4FGjh+9)#1yEGtW5=LG
zO5fB5#nV2X?yk-}j=ilte~0JHM?s#yJJ;HT20os86*ZVawPn5WjsH0!NZY^|5J3zY
ziX$Y)N@Ms1%GDy`QnHIG>Y6%g3lf7}tqqjNQX-|GMkJIz;mLW`6T9aQEM2sqZ%)tj
z?#Up3-c$b!L<vNmY`qFc)R5w(Bmq??jOw+?8CK}Z!UQV7ZOpNHf5`bJB~@i&T1t%6
zilaw>ua^hkiR)m?vaz#gb9pGv1MEG_v<!Qd0zl+YhKAdyC{Yx?GbI25z>{*Ic|8qf
zS#env97Q;J_=QF%W#%JuUzQac;LI{q#qD@VMPaN1GCmuq-X-Uj)wOg_nbh7gv9Y1H
z8syJ=>YssVg1iTJe-x9DLW(D0EI5ycsRe*p5xse&9aBi-V}k_{O={+i7?DKwLk>@=
zI(r5A`+(x@Z7i)=ENdIUQ&F+Phm&#?80$mDQ}Ak{8=$U8&QDp{@%TPLkqK$J#nt&K
z5x$Nrx`yDRi%?S-fU*mZOU)^+t}HJtF38Kt$^iNEp897XfBGN<tObQv!YLVedhqaw
zF1Vtys*18Afvr@6POC>yOlouh-yU~bDF;D#2;?FZ2sD;YNT9DL--%<#GB-CfGcz?c
zH4#Q<!j`{a0G3p|FNwla7^Zk?;2AUzB_IsS)Q2ts(yD)0Y;sm%dTg*K*NUM9r;l)6
zVls+q020Gpe`EiUh}firxY+2(h_I0V;_8h-P#Q)IqzT7d1b9UvyOWWVmxs;^cmP$J
zDa_a;3p=gDvVX_|QP2R_)0XG%7Z&8_31BxXQxm2UgKnTt6HYx+uaXd;0RZA*?%?Vh
z?#KT?@d6qIZ$5Aw&@`1))HJkUc8ZDYzngDJbYezAf4H9;+sa5sS&?vEV@8QeA-e!L
z5nFpUax-ooUO3Hgad!Og2$dBG%F#7&jy_EBWH*iiHDHaRBu`eu(30yF8lP24;i807
zE#&+;x}Gqy<{3bg=pAe<jE!K}K@+2L$r<_&xeort_zng8fnJZbDP2cZO<hAnQ%hS%
z7iAb6e^>9Ihy-XBI#@DkRERlptfZVW%0^8rDX~D7$JE4_X$11;J@wB(96(s6hERza
zHjpMO=z>9ymPn6qa~55FZKjQje{@>WxPs(R4?Cs?4AK(4+|i`%mQzrtF|F7x-a%2J
z0p4zo4%TKyhI-m~l2tgvOug#;xh<lW_krSJe*%W*85|oH;OUH^+8P?_YHF&;0}+)X
z<irD!HNdR^+)xr9J9>nOU^vzU=CxtG3cWx@Sy_pYKk1Qw9)jy?6zKhZ#lKHY03;U>
zi|La6@1}~(Xdmc*zc*_^P-AGpI0Z7mF-RNWl}+im)GsAA=3lYT;L07_&0kpEqlCbF
zfB8lv$AtR1Lr+jhxuB?_XJl#T=<Y|g|A;Ah;Ag;N`b61BO;JWd<UNeZUz-RA<Laro
z1wjEvMpht1l#XsGRgs5wS9pASetAxOu)Dpnwmfw1N8w3F5sZP>H83)>vg0_rdHcfz
zl`Bk8nKGblCo3T$MC(!jDIA2uufStcf3_$I{E!!kg#E;2Cr5|)xUyNsG%C%2CLM<P
zI`e=6$&^V<(5WI<51bq{Ln9LlDvF_~c>eN&83L%Gb4nRFkk`=C!G0N@uO!(7<+^zI
z1cXG!CZ=W*XeyM?5lOSi7^+js;W;_OP=uE^QNMApwT6Z*MDwGmNmy!jPF)pUe_eT5
z8LHg!3j=b3R$ftNVq~C)qpg_%avf^ugGnK4j>kJ}I1Y9;EK3Vy){Tr93_6`|Xh=6=
znwXheT3MSL8|rGP$V-V*R6KuqF#(~#5xA2LF&bH^HD_T)3+TnG=|MW>6%>U8j~qq^
zd%HN`@fdt(LH#l+I#5mX2=MpwfARM8aO1l;J2^S>xLgj~n!sd9z-%?aOhl4g7?AE4
z-B6UE832|?*o~!x#Asw!7No~Rwr5G#*Tu6QlElalGBPYHD@zLtlsB51TUc7Lpk_b`
z3mS(ou1itz{N=>}gc@G|puIs19?|u*RP~H)9KAwf(+YAjQxjt&Lj!%?f1MmynEM2S
zsiDxPsKvl@^2za0VS&D2<Q>=?F3-`)8CrQrnTpa7!V@t(RfV;&GB;+R8?25Di~)mb
zVr~fodrGpy8T2SrR#|RJbf5>%nrR5G91=qbfe$h^F*P+ax3I9Zwj)3QzTRGN_?#TM
z_AE0zs4YiP@%-flSMI24f4M;OV`WwKm@KZFe|V6On<EaDnYJ!|QK^MB&Gi*<>-^kE
zmJ@Vnptb|bmr^sZ;CO~6WS19YBt!<lyquMd9h>Kjl;DB_2?K2vSs@<+)z>E~JjQew
zk<~Xq23pv!qlCg?jw{JZ2=j5a!~81BFx3YmHppRU>**V!Q)6byf3k4^%ujH5Ok!+Q
zILuDDxe%I1cb?)*_{$5<YbfL^X1jpw4QJ~f!Lb=hxH)cZY6uwS%(BKQbGs*2<t2o8
zu+5BtiGZf7t%VXRtb184V|(||r2LxJy3(xp5Ko>J-PqjPp6BB314)>rk)FDOFrCA7
zIw;o!fo(vVAuBvOf7-aoC-vT(6oI_XtL>anm7f|F;O0Q+5vUe07!M$RnmYP)V>3$|
zdnb3_;3%l#=4YiQ5Lg}{?u1wT0(IhFREX}J06I<O*0NemJGbD3oXWz?gb3VGV|zzt
zR5Z_6yl&C72_*n-)6v$}(b3h_)6)apt3ii`X6YQ5m|r`oe|PfulC%gfwy8EI8sd8g
zg!#KW+L{__DG5nM1w0z5@lhdu9xfbP3!pE8DNO~6#D)@CT}zjcl#&V4+v-cR;zB&R
z;D0nBTpLM-qcXBe>X5Qq*#XFoK;_g|7U!hKM+AAh!I+<!k%10H#q*aJvL8Q;jRucV
z$m*E${9|*fe_I>M^DsA(m1{_9X>-rAEk`%?cUGkYunke*ucD@@t!F?d8Aw!>>GodH
zIpZev4bJPV&5ic9rz>NA1UJ9X*ic_S+tNr^1*i^zvj7uoSvmPerWWU<#YYCg^cB%;
zSCql5O7G7}$?CD(A~Gu5db^sc@{=QcoPbGJOGQp{f2?riLt0T?mku2n@6fo6qUxs3
zSzRs2FJ+*Ui!{%{nio>>{8xTa(1YfOXH>S%?wVL#kP6gS-f;zuGnQ{Var?-cp87OD
zYxR+1#AOInjDUPb)x<F<rJ}8O^`6ypnoHunZS|y;V59*LR0MfB*_i073%JODN{?yp
z5tdTae^6PJokG-QFv;m~@&roc^ldz2a_gonoY`Jqk{%P_YGc6!4g|^f84{p_$KeOW
zq!-n;PMf!)zo)aQsxUn+*xyT_`A?(b`419o>M`VkWV9h}8pUim9wkt*!Cp?b9ub+9
z?Y%2^oZQmiQ5xrNt}=qGiy0al(G6(0lE#1=e_u0c?ka#mMYvnj<cLX{fXJjs2*r){
zRE0NQN}XZD4@$_bF3CxagJCQo@|qf$!UEGBiux?q(3Ilw-7U4nnZVg2;La8y%Y||n
zib==J2OdE&X@%8IlV^9gH`btcj|~j~$S$7g7Vz)><poFe>T0SgO1SGP=#UDuQdNv@
ze`5$>A6<eH^Xoe2uH3qGYF$>S6H{q~FcCM@*VWckQ&G~lat=tyt7+?LAD5pH=)%$m
zDjXX}kAR4TaDO)rbk0a0k4-%cMMtNX<fg^LWWBwGv7xRyCP<=Ypztg&pnX+<JtHcu
zxV|DkJucMOl})lKA;p8?QG?#R8RYN4f1=_DCMKt}wt4)x@}ivd#2CmeT|n`S5A%h;
zs^GRPTv?<#b>X;zgtDF)+dVigGbIk1A6#1to_9oQas8y(OJ__diubWMR2YHjQZ#h*
z(W}V<pN@&WTX1}Kd1Ga6QkWOd(h$R&9X<TRVncmg9W3Ztijq`RFQKAu&h-jSf6Pmd
zj|jv;zX7hUVLEIPQE@35IRym;dFThr$w30(77&$^pOF|vjObzrKPX-lTX-9;n+wpF
z+psLmY<MWVi%rcfEy~NzNKK53jtEAI=Mn`85)S&87u=n*wYD@jK_RuUSq_adOE%v>
zA~e98u#)!t0GLo~A?FtK<gnsVe<V#~7?PDYISoS#wmVW!UK-{>LW<TmwzS9SC=xj<
zFHX4=7@xpznh**7R4Zd4W}f1|P{R#46%}P|L#)ow*a%FnZ)XKKf0)@386mEyjj8Km
zaC;5_IuPJ__=iR(B__nj#l}QOMTQWJ{Kgpn{a?j{c!9XPI&$o7EX{-zf5Dha%FLGM
z?urp}DAQtb-LV0!D-w*MllW*obYQ^HE6U5L5ctOcbi>@89juWXW|&zcCHMw|i?=bR
zN$nak001BWNkl<ZsmY2`83a-o)7r^9G$AsG;F=XC*%Vi`uC5Ns8nv{58ruSfvG7N=
zv$ix9kkgBdkW#0!oc)o@f5=Y4Z9N8tWZCmvJ^ca#gF-^XBO;?i{ekez!uWTJ>i-kk
zJecqeQv*T$jOx}<^kOVa3v&#g!4oN=u_@UJp`KiGeHBR|oE~3zXqYHK4c5Yz<ATnS
zt(7^-%S|k8InM6BLEz$9D4UZOp|%JJny^tp-q3D?@d;UKe1hDCf54#A4Gj$R^?{|>
z#LU9V8kuq{3o}z=BMkbL2gT#eGcKpPy|ySd(uZfM4_x{H2)AYPT-?0;0z)HW@G1sd
zln8vto%E~aKcuUh29gvT83HDqOHl+<<l-2Ur*B|jz(h9=-Q8g{MwsvpkZ&Y)u(YHM
zzNQ8Y-1))LgE`Bde+z{>aPgLOZ6zro?1DnkLuC-=>4a?M0(Nl%E@Lcgw;2dU7dQCD
zMMf$Z*m}a$?SiiHMYynpR}pAx>*=GU9*^pVlW~f<5lLwO2?d=0n1+4{2$VK4SVl+@
zJZj8XaX@4!D5(H_l@3foqg09GB*aGx7~|1}SI1UbTmm}zfAR`S%4$$Lr!!5=S++o!
z;O!4C-rPV_K|<J%r#1!gxi+SFivZQXA^N*U<~Fc>!V2(G#te+_R+d*(2C$vBE=GGp
zU3a81(=j-sX7a*Ki@WP`!ub}O6jlfi0Px)X!=QMF6)y~8extmP<oN$t^W#Crl^13s
z)rWjQ5j<kVf9SDf4@8${V8~z^o0?P0o<@<4F+E&ng>*nblr%y_Qcgt^xmv-nEatg0
zeIHvvAf|zy)&_UqhjAZL1g0FegM+=DtqrlqL|$?%iYO(BH7M{qAtD&5YVICe+CG2d
z$<6)krLpejs)CO`VuXZ}4%5cTJ1ik87}nC552I!He{C}2=kgoh=Q9w@5P;I2^yGw?
zNF0g-GbKTb0d-K47eP$}Zaq!kkd9+`z%$AcbMyLiGM&SecqCwpqQZqC3TIco8|KPE
zUPO|B>i`%-TppQv$hkeynb3dlObAw9V4btz$_tV=;vg!bgQyA6E5OszY<=QOn&%Ae
z8=N&Ee?P*-L|OPg5I@=w@uN2}pN&7bG>RlZe$8_tsF%p%u$Z_wF-$BWNlb4LI27_P
zBbebn_fvuS>eEt_lThXp6CDJdZe0~A41AU#VATqWN=nKqfNBLytDZiMM$<##wyusI
zItE6_1OVuTDB7ZQjOeOSE20F#1Kb_vuO{n*f1^i9D(hfsqS#<h9_xMXY;v`oDb_g(
z<(Rn04$hfM^iZU)si~o%p{{Pi3rHwxoVIw=!YMTw!91g37)s3Af)$SvMte&mJvCrn
zmWS~GkdJ>#0%jl*m?lS%_$ezZqxVx3z!8Wcj?n}Vjne<`rm5dq@D{45=$P2J`1piG
ze*yipkO2iH(ioQtgf^<`0BF{w5shyPVmijw&dwIa{^n*T#==UtVDJTx%5vEDb~cu#
zWQRmBsWC=I-N1t55geNt8{`gb`NJsisNfEe&<qsV(}=bj3dcpDJPW}aK848r&~5l3
zNriQjdRO#JEJ^eQhD=HXbUg_X0RY!Bf7F5=v?3)R|C9*!uhA&(k(9x`DfH{%_v_Jg
zb+j}AJBEqAByrQ3xcn41;Ag6OZxFx0knrfZ#1xoC;aM>><%wNiFf*qofV0ZOA%~l<
zg&|pX9GH41ga^JDtg@LASwAOfPo2Fy_|9Y<fbs`KVOUf|UYlvl4~WSq$%yuMe__#w
zuf3s)!A78TX#1+;&1e#$@aW-0A}6CmGh`T>TCiXuGQYaHd!CRRO(2xw!~J1vioQCw
zU)XwqeDp&S?Boy;$Fv;Udej0W(%5Ok!~i)sh)o)4_(%PG)5b^t!hd@xfl;<j?mj^g
zvB?>E$&r38w#K?j;+X0{*TBe>e`!d=8$V%~LRJno^+DKW3lnQWf&PBJJ}{gD6eq4O
z!??-GJ)#}~J_O+{Lmz4}+FF_#>Z-~lpS%*yg6k8RR@_jW8t&y_JbclWph94vk9++X
zjVVX0$Pyz{4eI=%d@v>@ue`1WvbkW)fkO$Qc(2g-a6jl2=&DHLUNVf}e}H`SLlU%p
z@yEiy*uoNwoxKCw&c+IzYS^-dQNOSPi)?oZ7(xGq-u}tAJ_sTI9jeDEw>5|F8ycIE
zQ-YV2Ky?Ol*f2~iZ7fa59ytnUB=InfiY}9d8*YhlF;Nj=!GZq%etv$wNPJw`R>U3?
z!T4infG3HzLta8t3z=sOe|*Nomnw$V$VeB|P7<uP6MiUaV*w7~@X(&Xivr{VQzYYb
zkL~R48yuOCmIHIr<(W}F93VKPgj9S8#%DsOK%58+X@*RYkA6thq!2Z#=^&tZ!Ie!b
za}y&X=0!I|iegO{YW|lL_fK{DOM3tPHIG*=GOW1nK{06s7~|k*e`TnS|3J)5Y%ka)
z2SfwF361#;am%|PCo>Il0lcL&9E6&A!&Ks2_dpDSw?g5(JXmMED2%2<RC^@UnGW93
zS(Pn4E#>Lqe3qV^@Iz6^Yc;SBU_hy*utdbEpacUnEEt`Nj7>;R&(13<t;~rJ;$u=g
zN`#7s4GJi85VA6Xe|+>q8X=ev!{tnKYqqnin}?^juYX{Wzqbc2B3qc?VIidXVUeDo
zO!*6%`{zm}zBF=}{981SNl!Ew*3SOX871|lnX!ScHb$C2KY{XSj*F)=PQ&CSMniBj
z8sK>Hx@H`o$n=u&wN+(Bd0DB6v2hR>CL|^%h5KShAtA`4f5H;bSzzO{fNTiVdg#V@
z{~$#FT4s)cNk!wQF6wT`kM*)M80MQt;wTEmaHQK56~RdqWKnoF4?MLN9hnsG;F#33
z^z@9(tn8fJ#85A;h5m55TuSkw-p;lrdMeVRV0Q)#)7gQ1^g|-Kz$AbKh8WdDIu8yF
zkBr7}V+t;cfBD~WkWXk{Km+@s=KlG0{y^>jru+cErHqy-$0s_oVq$e(VyL@=i4Mxb
z>894`4-E8lvPDVu@KJSH9Si3mq_M6^EseF6CHa{t=-L15(SupA=LbZkMh5_CwTk2j
z;4)FyF)+43(LJu8$>>_~!!xTo=B}CBR-O{TVGIkhe?KHCA2EvP&=WJ65NU#$@&v?1
zer|SFMp{Y|fNy~reOPutDIT^h;cvZ{XMXfUBBu=`Vbu=!MW5iXi0Ih(q?EMGl!WLo
zj24BBIMD7_7P15&%@eiA57hfF4EZnV{oj!t;PBSag6$R<osw6PmmKcn$YN+>x(QHx
zkk<!_f0v_KyGLc$OzN9Gt+S<}x->s4JFlRqw7jyq7AWg@mIkVVj%!+BMjXtn(>3LZ
zsa0E!Gtm}D!9R`V76#z1mAxG`Inmw@MuG@VRAA_}@th-`kdu+2R#V~ryquuWACsDr
zln7JrQNkty<&qV&j95tTaWM4b2(<#rN#ik#e~*4h_#i@5Ni0_M506hy&B)FxEH0}o
zEzCvAkBJEO_jbqYwxFX$C!$gmpr}IOfB+R>ZeO9u5`U0T1Vd{S*T147fk%*)k&-0P
zD<AIT`|X0CfFs|_KQt;KH7gMrYdce#DkcvEp0J<~Z@j#Ly$@`b7+Bmlb4u&@n$o<&
zf6|JY`o`wA&W4ioXkSN5eHGAqAPNJ<l@M=frjN-hTs<jHdksZ>YqzlUvWYX6&Yo0X
zoEGWBGbd2`N^%1BG&2@a8v+QoCQ1nKaI6w~9AHghXvf~kGcYnDJS;RMBseIL+I~VA
z3}h8}p!brj_)sq%i=ibC<hS~CrYWA8fBxu)M2wQ6j6e(b6C)C{P)t!#Q#Zb;wW$F-
zUOujTMj*5Ag~^`qIF}XC*TnrrsChw$SsR+nNVEZ>tfPxXj}YQNpg4ht3m6Q+3!z5=
zvtmTd@Ox>VjLagSiW40Z?CoL?OJyYSptFOESD>e(?a%6|3Jh*gdiCV~jcb-If0)(P
zHomI3yr!Y4ee$$feO--ZnNhwx3mV*a(0faLd0uJ^K=#;9=x~KZ#sQ7Bi#<$U8QQpo
zrj|5z_xDV0s>XB{1Xe;vo!t9t&*KAO7EHu}{skx@T&R^H7US6Q+)y&%<L&L`<pt{`
z(c6+_OkgrJQGn@*^j=(7Lc;u^f4!rpZ$LNtby58&OxGi?q@rhH=i(clm{V3&+tAcH
zscRbWD7CjB?N^tV6y|0j^~XjL%MiRh-Qf)II1a+XGr<aCY62raX6EKF^#g)A3as$>
zzJ*}=%K^$>#Oq8kG)X~*sC)fhgMURJd08ZbKio#Tj;b79!D?ym{7Z^ge=>Fs&ui>i
zwRhXPfqp<Rmyc^~ojjwrf7#&t?g?e-Q9fKVjI1fHZJjo0LQQdYQdFR)Z%}wNGAVh*
zSqWiaRrJ*uwr(NGMfII?k=<ylF3n39A~G;XnX`vqXi$J3F?Y$f!}|ouA`sMaa8cKc
z<v<pZh7Ws>7UbKO9FJh6fA`wfI-IC-ETNx^`^>*Cj~ARIb-E=iLCG#_nAq0YJ#)^y
zMavfV&!0Pc=CrO(r2g@BHI-$>g?Txd=_yI^vC)y?p(xpfQinSpBXs2PxLhuW!(nr{
zTprI6e`?s6;0i))YQs6Ar<aefaQu*CV_|Hdt)?iA!`%0_tAB!Ee^S@x+R~h)$N+at
zJFcy!AT3I+VfOq5#Veb+#g?}BZ$5Qs=Y|#Nx=yTVZ10}6VELNOyVmwkYbr~R^g<Cs
zOnOOO+w}RfyW1M8inEhr5>hgAi^{9(CsgKPSUf|MY3CM_lwZ@<GaWs(<_Y!VDhtx%
zLLjEVe3w91l$Mehe;*SSj`WWmC^537h1+t1^=PJ0H@3F19Zst>22aqUKoVDkOX^yu
zw~?ARH#NpBC9Pjqxf?xJTuKgQCECVz{J_}E(uPUhGkX^-8Cbo3^Tzc<t5>a9wghO@
zX3dz|)!ETHv9W$!RYhr0L0)!ddMe(BgK2jH1EAso7RtxRf7j2?KOitLC^$GcBqSs>
zG%OU{Rd8?+K!#AV5T8iB4m!#<upjOh_VIjztOP+e8rC~-%?P6+EiNLZq-BU%<UVYj
zsaSX=RZm^9@7meJJJ&CtH@&@~wR`r$l^b>*Jhf-2e`;e%YM49Hdoj}cf|X0=^>nq4
zuP(_gMAosff4O7Iw3h1pq!163IT+i!1t;WIPMkV*YB!RuuC9sI`6*C3#9Ki_V^ed>
zN(9#@D#YK*6*@h55gsyoP(1}z=;7k22|#QqkS@_1G-a{z=RkI6{<PNG;`Eq6H+ySA
z&)`?p>PL$bOuBmJT<`GIqPq5;zJ<#N*KOXp@9?1mfBW|A+OciRrVT@D23IU!x~P9%
z@2nZ!lR^Itbv0E2mMxTaB*w?Z#>T`%N5i-=G)Eu<PD)NrNl8sjOHEBlNls2mN<?8}
zRt}0AON$Hh;dRN0ktnr-jgU$ti_Gsc26Px?!@{n*>_|5&!J7B6qa_tJ4G4~c5Blh8
zEWh;ee{<HKe01~N;T>z2^ma|0JZs^qO?!@AxOIBZ+67Y@3KN6idAH42uyXUz%Ej|~
zy4#!T#y6uVpr?2K;+e2o%HPRC&(y&!C_bxfd}kNzpqbadc-frJhSKyXUq=fz-!D8d
ztGH(3goawedS)a>hx)lYIgrej(qfn^4Uf!Wf0zjPZJ45zBU(h5l7oZuig7Jd`v&Gs
zYaLgZ79BvYQb1-C<X0#jqYHHnST2Eaxm7JQ77nc0ux-zw<7Y2lx^VvNnNuf^A3GvA
z0vmuLd(i@L07(CB%}wL$$5mIBmmmv}o0F4`_v7H?H!r`Su&}78xTK`Cw5+tWq@=j0
zf2a^;3Kg(|tBH7B1v)p#4tj$f#1&5o(cfwC?>@+hjq_UaBb<%oi1rC`BjWOSE9);Q
zUfm`rr*+}Z%Wt0CICF5zVE+tc(N}ETd-BTtm$y&vTHV(<E;knL8R&iA&MoU!EnUz%
zV+tsJ(ehPmH!Pk3EYof_2IlBO#%7=ce==qIoCQl)t=ob;#nh(Kv<Oc-7r%(4-13I@
z8Q6D%=`71j$KZQAOI%2l78UTT6Qz4x6%<mV3W(LCQVXh^r}i!1Fwi%(xw;@F5~F=E
z#t7u2AJP~xDR~SexAO=?@o3ll!HwJZ9zJ#9+MP#_9zM8#@9v%3H*Z|Ka{1!<e{*L}
z9Y4H(_x3Ft)(#G!003Qo2mm@-n<tE~gS8t~l@%4`W#tu>)ivX4>*^aC#y2)jXlk0+
z)HI>7aePBVedENIc6=A8z@aoN0p}zb?t+5*-v=W6KOpxHuj(p^;Ajia64_;O{{>bO
z4g0X7$tw=s`}^C+*G}!(u&j6Tf1IW3cb~la@Xg;}-8#8r#q74qjH24s>7e&h#}4k^
zwsGysrTvRR>05T}Ik0|VcSByRj|0PkgDi1cK~+=d%=ycQw(UQ0>EOo2(;JJEf*ifW
zl5#62OrE_&KtS3tp}IILF#@8H50(m#5`h*y-eDjoEnvvR6j>3;`BhC_e{+|u-@ShM
z+^(j|yd(jG-mu>P#|SYjMx&c>ePeQJI_3;)-GkJA{od18-+c4#-Pdp5zIpT2>z7}?
zc=q(kqlb5IUcGqc#E}Dgc5d6eVQBSAVf@$8-qzaE(u|ZmvALzSy<<}6<SAX<-BYJc
zpD|;`^l8ZBbxoN(b;hjTe|fNiYxxp%Y|tOy*;t+(ALQ;}s)OM}pN13_%0ECJ9bY@M
zDxS}fq*nci%Bm5U2DVTzbWxlyCRjnI<q(?RzIgZ5x6f{0IJ#^7@;+orH|{}-fBn<T
z+h_I;EkprcecSYTD>m#tcHzv4!~1t_+c>mp&AN?Sw(r_|;Lw(Ve>ok{mT~g%N6}ke
z>A0pz)8{T8+_>}L$vx|r&T6kMObLt6D6DGinBKQ+$>N3m^XJX&o!#BqP>~O4eOI;>
zfWyRul`Oo&Kv6{<lU?B*2E>Qw!luF8it&?XE?Bv6c2`SXS#D}{Fwr|u!zGiCS3C@h
zY3LXj*?UH$!^rUZf9?B@p1E@S(U)Jp`x?nxByYZYg;f9i+0%#jZe7289;yD|ex&;?
z8`pvFU$SsPU+=7*o|!YIBMpP9XZFmRJ*RhWU*Ej>^B45@_b-?~e_mhT++O7KmJYz<
zTDN8uLA^V9LPaiyHUpmv(Dx5p0{DrFe|&Poocc6hOBHHRe_nuu^a0unQ@XCY5NM#|
z5S%k{-lo$}?p!`~VEfSW`RKuJ!j}HK7q`zH+`OW%Yr>=%^9R=NJap#TrE{l_9on~R
z+omnscI?`-@4%rWJJu|k*#=1H*rfEF!m{f6i5*jCFIcvE<Bt7X*DRge-C9?gU06{!
zaT0tL$x<wRe=|CpYD%(`qtK;i8R;lV3jcNi=2OVeN%5chhbLqf*ECI<*@dEvilUsf
z1X!fwXlKd9Io+>NJXuq+;iC++VM^bsjXMq;zi|Ejvo~)Ag6{lltokPp@7=y}<>I;1
zCys&ccWvLgdBeK3gDVHnF<!K2AyzTc^uk4pmn>bje|-7Cz>1YCSFKvLa^;GFf#u7V
z4Xj!{w0_f;?K`(`*|Z+LZ26KoQ<|#slEQs(!wiNC1zo>SSNxMRo99o=3+3ohSj1rq
zWI48$Fmong7ZcZW2+C~eS$*)vrPGJ^Y*{<dKWEY4=6xqF-+lfLY{1TS%X+(J^{-gJ
z<KW3lf46U3y?73Jy#ssq?%RLh;Gx4uj_%&La$!&Bges`=R@KxsOqkfwHFNI5WrORt
zqHtjT?5UHQ>nF5M>Y6cU-oj<eSFBt$xMuCp`V|Z3z@;xpj}7wRSQx0w2w~dPBo~GM
zK~cS$F5T1`o!j8(<m{p<lwhOuDnB~|^CWn?e>t&jKtB2*;iw+K=9bO@ak(f1U9x8L
z?!%`q-+uh+^;chg_2$hRta+^Zm(QO(x_{@^^(&XopFMr@_|d~i`MY-_?W4qR?b<bK
zR<9mhwQ6v1HTtWD)~#Q^VZ+9an>KCUym`~6jT<+tU%zg{=B?Xz?%uoq(Eh!<k-^xs
ze|~6bZ&x!OS$21{#p~Y5=Fg{T9xMLox$TQPN@LwjDHbGR(*Vln++28UTNds&ll@eE
zhrqOP-AlKhJ#lF7_Kj->7WFM%vvuEzi?^SA_4M|oll!-=T{dqCHr*F*K0t^6DtZE^
zjw3mSW&idKgG>5mPHt%yNIQ!B=JqdHf3bR{fK`0vtT}V%_b*zqY{ek5d7HQG*u8np
z(s?sFC)PsykZ%VZYt&!{h0~&TVpDtQ?Ht{Ff};}Bb5S^*lMMlNT2eeLIr8~+yJbfc
z_y7$ZGp=`JM%jdJ6qIb;cl`W~2VcH?1-i!<AL$>Rg4Zu!JbV1`{@vR*u3bgZe;x|?
zPaHc6Ss&8l&K*0pZ{N0U+tzK{w(r=nbLXzzyZ7wbvlrcpef#$A-Lq%+?p?d~?A?Fx
z@X=!@(93Y5uzSb4WqniekT7QTh6n|7z<y#04=eup#oYtlHE99%)SxmsLgff$M;@*`
zWBMxyoDJFj$>r^RLkGZu4-F11f9W4sheERpHy^%yeCNv9qx-gRSharJp2Mdv-MIhk
z$)g8%Z{NIj<q`^b&z?Db>g0*TDAHfQX2lZF@QkTbCbc!UK@dBqZvnP_%a$Y4w_*h{
zdjrdt3j)F;ClBu2uxe57w2sEI%xFJHOPbPf5I%It8D;?d(!#(mYPJh^e}z1JgTkX?
z;uBNSGjqs{Iw2O*c760iqVNGsJNJ;}g4*_3P`5aA`pVsBFOceABgGSr9t8RN7tfv`
z-QT-&`__%?SJB@NDc|XnC*bHGJ$mHG;loFc965UQ*s<fsPn<Y$63MAkr%s+k|3<gt
z#L3fV&Yr(;2_=f>&Yp%-f3thziUl(}##iLx?H<tUq$Z9(Rq-z_?;GrC%#Y$zE20FT
zcRqosa&i>59T{xDgyIP^mTlc4s9LSsuw(!6^Vjb^dwA#C#nTYT?>>0^?3G&&o}p0p
z`P0V_AKbfh3#B_3!T7;B-n$#+yXyv*FYN1?+S$_9(K)4i+Kisre|-y=u2{W(^L8R8
z0{=F+e#@@?M^WB%_UPWNLr@_e5A`8ix~g#HDGB9fK`f$!X&R{&2!I{4w6W*n382{I
zj9kc+3kve{a<Tjh#pC5?N~(q|mw>qJil*+~C9AjWJAU!jlc&#Ld<mL|81L(MAYZ@z
z3Y<L(^PW6@MC5w6e{SBme(ma&%a<=-x^(H{#fulv|Dk`O|Gs<$<SG*MKluBXuUx%$
z{l?8(ci=_nHJ2_N+q-#i(d;SBH3hJ*i^b5OHhh1w;=jClXkA}hd2#@?B1*7v$dwN(
z*WEolJPCdxWKEfDpV++G&c4CbgM&m!vS!oH1IN!@y@Pz)e@()tpE!Hz`rSv*U!$i$
zM7z%(Ke&s6-3w>Y**<j|3j_$z=4okdYwy6R&CGdA2RDI>ICK2a-W{9K$yl=q9jUV@
ze7<$*)S+FQRxj=Atjmx0<C<s)r*6rmMs|SQCkQo&?i|YRfum7Z-_XcJz{XioQCU@u
zGU#!&ARn)Ie*sl1%>vnaDBa=U_3+v24<A2$_TuF$Q2e`h-+c4Uw@7e?2Ob|ee2DYW
z2|z~xCHxPdW^xb7-FsN>lkxyW@V$HJKOa7N{20CDA^IXZH<wNxK<=||IwU;do@`Sc
zIf7l{(`vU^@h@*2-Mpx~E;s7E6;ZI&4KPN~F$fIyfAw%?<AD|vj(1dM`NU}paa4*7
zGt^!V96x*cCK$Jy*P)Pc_2%7&&tAUy=7%4?|NcAtC|<vK^5D*mD;Lh3yKwp1jobGg
zJidAE$j)`k`=+%*Zrj{EapJ_T-lao3kDR~t?EZ}lClBu0ymtB8t$UB2yLRu{+voSM
zpFh5TfBX8Cb6P9XLtHGhgzQ<vCvpriA11}tM~-wUD(jirxk4Ea1(S_U&8_W|e%;<V
zs?Vzp1ON#+6-`rSFJ8TM|EVkYkmjGGEC1%}ccAs}zW*Nn<nO;jg8por<$sOcKX`wT
zS0FE6zWP93f_#Ytf0Ou8I5;n!-@AI|@Xie=e<f`nmzNOeY(-NL@X36#;$Pi5xqV<(
zb6G02B1&+L9P#{9a9Ct444eYxq@EeaD=ek3wjH@;yz6^#?Z)kU4jw;q@!IV>ckbSW
z(|rHY(-*Jae)Ij`{ucz-h<<qY8hL{2m(E_ie&^w{S6_es_csr(9N)Qi(ahGCiA{~;
zf9q?<Rk!sl-EiRC-B<tf&5Qe2PaoQ`e)+l`2TxzU_u}3Ee)syx-79C1y<a+Qe15d2
zwXP(^o<$`fE)JGP80tA(2!M->rm=%(cxq8YQ%lF>sWWEJT>$bc6ptq<q%cO!FDk95
zwr%>nfsK2PUA%K2Y5v7atazmMAAbDtf2Y4;`8$@sy)Qor<VQ+=5Xko*2yuG8{r2U<
zn-`An+d8ytR%>Nuq^BKSiQ3Zs34#yWt<$^L&hKg{N(%RLbuc&3kQ1eNWnK6lUcUZ8
zAz^`dMF3M@3$vC(Eq>DEuBpf~_x3GXv3ApreMe4RAUqUuQFk6ZdGY$|?|vfne~uo(
zcW=IYeE0grGsjL}xON+9{@o8R?_NBzZFT>QmX;PW!k@NaaNCiK_g??~>*x2boI1E;
z{Qwm7F5Y_l`une*-o1L}2yzY!rqmb2`*TgyVICi94kJYIRxDedJIR16Dr8cj4hO}E
zT3y@FII*?8vkT<o6;G}#l~Sdde>u1X#vs!%5$^n!eJ4(zyLjdLtvmN0A#eXBIsk8}
zn=jPqdoo<Dgnaug<scw4Kp6s#7QP`o$9L$=-?;*R`<nUaZ;wNMPE7@P8pT9DDZ0A>
za%#`I#WPx}a^irL%nGgzBuGqiYX?VX7e2Iy{m^~q;X*mgzqkj-Wfa%ee>XHvnAqIX
z*87NUv;Y7g07*naRNg&n{!%Ew96NsU6bebtoxgbX*8L}6zWMe?G@bKXWna9<vu)e9
z?V3C_O}356wX-JMo^0E8P19r>ldXHc_qq4{0q5C2tiASHZ@fNL9(+%tezz|t?}TYH
zL>s{V;&WvyiICs@{cl1`6gvZc;8d#2fC6<(_{U~nm@sY4I`F)gRqDZ#CsgOR(TB!y
zZ&|zPhu|&h1C`jZLfi~M%ac2}BlVZP%CT!FHsqd;es%q}NMQz#?T6Ua+*0;mz`vrT
zg#@Gl<gS<06x*8(-}ccphqYL{rd6#pm_B|;jw<JNI~}Y}uX&IP4Y-pA`I>?CEq)eu
zVUGQk@OllS`xNr|AfiQ(Dgm?D@2t%w7xmcPu2K48!IeS-D+uUXp>r#^9gB)z%Zj_#
z=nGXA)$j=ugeg=jAvYJG!5{__2ZSx9pb|>x2Lez$+I>S=T5@_a#?I_bDHD^7r@AF~
zQ-&>|^J4caDp3&8-M$yx$$a&Qx2(<Q^=NIcuiNXyis`Gzxv=ledW!C2UQZYBJVc@=
zNt=byeVbq|3V`kr@2tH(_s2`an6C-De;qb|T<t?Av!Z#+_Un7NDjC}(=OG4Qx~Z<R
z@MI1N4MIq{w2{{Sf{Mklk)((Jh*qBMi>c3E7y4fZY)0MdhaV<ku8)|MiG`Y`>1fmD
zFl4YKVe7wjpR40>&LdX+?#Nn^eb3rn3J#w@RcH}FqKAMc<nj~aC;(N}?(vv?2HQZQ
zCV=bnJqs7rQD4O8yl+Oc`$97V{oj&zI;V{KT@qF=zL_R4TO4+iJ#+<TG9FA`*OjI7
zeXZUrHZ~TjFVyr)qP)$2tyxn&R|a~ogL~!;SDfmgLQJ)EVdsz23t7V+=;2S|#gD;9
z$19*-xDIm|GF&-ZGH^Rdx2;mp(#7<P>|ZyzgKsrkJ)cjZgs28xKG)kYH#Tca$W<$C
zrdyy*&T+ld{b4LusWNRUXKC}hW;#0Q;Zam+uG&M`PA49_5;pmW+!(!Wzc$;bo+n)+
zY#TFETZAOWm+@NlH$3!W??n08KZ1M${FOkNjV~h>9xgMfUaR%%zxVLCJQ3jerMmNx
z#Hf=MX~iFBGQn^#frvW(ez*7$rh({~Sam8w61MoM)lj6}gZS6@>)(5iKav2V<K0_t
zi^!W{1Du(%2P1gwU}dmTHQ4XP^@Y`?Ku9BUl<;YHr@9fl&kBUL21v#OiUIHgQ2rVV
z&Q2D4Pf#_i(dK^j2Vug42eA-%=y`1q?cE1t)VZN>j~~9J*;%VZBJe!Y%}D!buiac{
zx$u-aW8HYP*TXdPK`H{=o^Lk0(#7lY)_D&uZNQQ-1-y#1wRzaAg{(#uo-aPhZ7a4y
z&A16s79blb*U2jEKj;>cvL5%VR|2z=J9{s$FP^5Sd2w2Aa4vS)cv~%X2fI0zCtkm}
zk*A6EIelK!YKT_b>>mRuX=n92yx$M)=iVEQZ;1Axqvq=2dS2Ypg~BL~nkR8*^xNDH
zR%3KMR9A2Bo+m+CiQnsBr`9aVW>45aQT3#yGLtgG_6~6m!oqG|u_w_%6TqFw6ue2|
zYe|B{uhURve?hdYM}iv+U9J5MuN}fbQ&|(%5%2!th6-oY#E1~A`jA<#%g*ReW&v`R
zyEysl_}OLMuK4Aa@ALM^v#C#OjXqBVw(sFWjH|4>JzlzKcXy9D2h|m!2a%BX>G}a7
z+O230GU(e^5cuberzm>o7%0CTeXuh?_O2z>iY$HpiZS4)Mr2b)&WhK_2@w%SAcUgI
zIlVYC{}m3sq$SO({Z>EQ$Q5F2RFvA~U<S|G8d~ur!cqxhz$B65Mxo-A*4^##j#&rA
zk|0N&!ch9-gI!@0-Ya?^p8m@5lu)7_I67?pax|MQ<i2~if_Qu-26RQJ7uEKBJcK*;
zyxfexul;4o+35Cm@R9Z|Lqv0{80?5sQS&91Gj}$S!I%s6N}L_u-0TmGArBndgmwKY
zL=*IE8n6p@y6t%#g8hP`;OtBJ8)39r`7&^Tkqf)^T-g0%WH4=uVz>zLbEEaL$KU^P
z!WyfzmW&J+Ha4eU6L_a~Ep1t9cAGY6v)QW|w4BErRK6Jw&T0ea3L&CPwo5?YXf>Z+
zXqUS*|KG}|s?V3$fO=wLYLfu$bPu|v31dhuWi}IX7Cj!71|N2|m>G~bF@7QRjm?)<
z?C0Prr`si=Zs}Ay=%7Z>XGtU~{YGx)7sj+G7IU~GS)l<?mMy{Ue_Whtv1Idm-Ocp`
z%f{|F?TpwLKeHMF?-y}fJ%0CZ*YAS7pe0yWSII*=F^J%*F)r_)xa%s$VMGP?Fbe9p
zJ=d_P7X27-gApQ$4mLp%3IN}R|Di^2(vRxz@2QHW{(6->;B23xFZT#mNU-XEwfN_i
zwdZ;3GY~+GyZbmCzxX*?aK~r)k(eU+{T8%z^qQR>k`J9hJ6Brun%C1h=;IRa;S;gG
z)_sPnOtKn-rP6#|;Kp5DT_D6;n_t&fj3j;NHNLZ{hX|=zn2LZc5H+tL5o$4WjU+!-
zK;uIfg&h4?N)#>_|K{`Y_jAT6{jPM^{5n*SA>fg(K1E++W0k|v=y*NxL-&!0&-Z*$
zd+gc)HA%2-C~zMoV!N`{K~2<+7)M_c(&uOs^66Z!-TCNa?+A3%d)@Q6-0TXmxU9xp
zOP#f1iny((F3PCSEA(9E!oVgWK;pCD)he6Yg<uwdsKG-+D#{;0>Vgf?vj7uTNDux^
z3-Il!Mvzy7ye0fpbq(Hvc0U*1WZ7>H>#pYuIY*cAyYr!-lP#~Ia7Um?`wuyv#zSW9
z$ji5$YRSo?>2tdZ>kJ;?AQjTmfg&%Qd8bIWwKiAK-{|ilIT4p#fWb${!L4Nzuf{%g
zspS!Fn$}<L7L5ZMs9XD(#c}EM(@akj03DJYPGsqc=#Em{R~N5{O289Fk^I+$W2QU-
z`SM=C8=}K+f?Z}?^7-+V;Xg4K8?G;~VJ+wDuG?eqaJOOb3K@<dedP@21>kjt@Y{Qt
z?eTklD>sFGJYVi|z}$zERn8XA9Q`xv2)uo))oyjY+^i&pA3Y#Fv?SBO)<xO_r0SrJ
z2tX4xbTombR2SMpj=J@A!2xUc!I+7~?F-nwmT`nDqF*m!Qkzp?W~wd=J4uWkZ%nJ&
zh^@K$UEAyZ{B|AO$y=*0;^+C83fcmv{^g3;{)31f)akAQW}0RW!z$o&gx|rIx;()>
zu9WixyzaI)K`k~EWDH*awHZM8RV}19-|}Y2nfliOwkG%9>FH>i`Di#OS#LeTU~?)q
zS5zshmSNnWF*`T3zAp##bmJ?f^vO_igV%{Ef$OyPDnac}n6f)Vv4M$euQ9Swh#({R
z>uSw*QA1WP)odZ*dZZHxd0x#GNQR@57VdA)hP(x*-LdM|TMRDgzXEFMp3jHfqpUqI
zhqPwwpdKQq{a1wO^LNUg`3pQ1hL-e+ih36z&WF$LslvIhEX(?UhM3!zM8y~k4cCjD
zh~QGH;72Hzb%?^m-;FM}KjR<3jXQ150y%keg#w)Jo(j4Jd24lr>#RPq@%1-;m&|5!
zceXliW@mgAT+1XFQ3K2aggZU{Ew}ub1ZiYT$kqCd-cBGLQrv1%qOb{g4b%L2q<m?Y
zImg#i(BrPB^urh?{`(r<*(t)eOjV)gLH;<s<!>V|1UdqBG0FWyYd_eww|N;1#A3*4
z#L*hITk=6ZFr0y@q2`uQ<iJ$*2nSryhX-<E^W|jbP7t}fE8txMPG|mnicI&7EQ-9g
z_xU8;ajW|sq2mqrvHHT6Q3ZJ*+K(V=p)e1kcOV*Xe}GJ@%2j9{oN>$8Wjw+d4l+Ps
zb+DxivJlxxsqO~-#aKP6PYrpfKVjY(9uL=GWpSg`<o4rt+qYKJ-<g5Zh=<cTzd22H
z9+Ss&a*_lDR)MwE%q#-J(-DwY^=CW8VbcUjS*Oj&!0>Y2>-H6w*`UJ?3}asyHa1u2
zzhf)2BqPrKbhbcyx4DF$lC-=s0ve_3cSQv{6tgISgzu_-97X<8V=FIN*eVG-a(n;s
zSP<OL+iG$@$kndPwV{9gfl_pzSi~0|G{;F?`F_|%3k?0Wn9Y<1kIE9XoYD_53GK$4
zdEWoinb)V~e}6i8q{4DxC|KJ~!JSU_pan7f={~);!<-sJtgw=!2PcE+bb<EUm{4VI
z+{PLnTgCnr>Hfg=`v}($FKj*Loi<NZ(>?rOd$hB}%T1)K43d+^Lp9aa-QA`O#WUmk
z#qwT}Vn9UX7dVu;`eQpsY1kmJ1E%jn0UX8pu(5?g9|I>vu<rKSq7mzZZt3&P<4P*g
zDEE(U#=4E)ztwRrtw&32tG)}Q>+U+QjG2OSGqM%<pB6#{9+oq&^(_XU2nzc`3>NNv
z<#0iaQZRC9BjJRLb?k%qw2|Tfm;dbpVVDun0|^Y`Vw+QkJ|EoEO@gFASf`oER#HEC
zI+B$aToqPQyv!}N3>=93ewA-R#37NDZn62EMDxypyZu;nB#VJGaC8^Vbi-fSU0pQt
zVbg{ehq+2*N>Y~|?7FPyrio>}b6is$*L&^ZhFvd^54@nM$3A(L?XngZ_-YF2c!lX{
zSO9i@W=>Gx2k-UbG;>wJ<cf}0kh0fK@9XFy|2v<kfXnIby$@akQou&@<l*$ORTWd;
zpnsqIHe6<NW3w^57UkEUsMoMAJDtV|G+hVL47(o|!;(&-2!{vY3#>o+^^mVgXMecg
z-XJ4UALgn5=4cS(knjIY$U*+It5;~CC<m}aB23Oqk?|dw{6?%G5VU0ZUT-`ah^adA
z_Tos_A=yF_>Y;^Fqsvy#AUrg`iP3?<QJBZU9fyl;L_&TyZamzQM42|&0^7Y;ms8yK
z8_OZA(Bv?vDV@xtC6xSyhJ7N4HiOAnGeB2PU0mPtvhT1zRG9d_E|Gylgt@e>!2vLl
z{oS_Ewb;fc>fwl&3ten4?k~b_nwSf*ep4V0_LIfYDo@1mIa8?<C6^BXxp1<N><+hg
z*BfVP(>WXk6tp5d1h<z@o?dMU=<v-|W=%$QC_I=um2f}H_<9iyg~xTpTF$_70>}H2
zVic#U?auj+Z`ypkU!R^~VM4c!+kkB{w62<5yu{%7ePav~5*(Od*Z#JW<D)ao++^L&
z<&{RK`;raW8NJTOZ1uOvBQ4bG%RG%7^iwm`3%Uc9uU!xk3{ATM@?*$ayYbddR4txk
zNs5v-1QItA*36i(j&9$~Z`%s*tN<-ROk$H`-6SL4fqI-tZoc|Df6$Ha7AWPG+&r+j
zJSPr9aXol?xW@LPr(1SsYu&l=18MA>H?ADN)iK2O0{u7B#Wq%lMQMuc-u5>SY+KLy
zT@EL5Xbro)ug}LR#n^IJk^A0nqUGMkj27AEW?`udmGySQ^p0nYBZ#KchYw#vhKh&6
zU_AvOYtB9LnQ7Y;;L5`30Gb_&26=tO&u~<cX#PfHrc_AuWXLSI;`n&Z29_bSDo%@N
z*Wxg%ol@fU;Jrph8pRWRo~Dbw@kx)|Zs{mm`}gKS!WThXjzLvBbCdl0$f(h12{8~G
z$NVycwxeCOl9Lj{Kpw&c3XMkLx(=R+tGUTqAb*e27nE!hyM<W>Y`Y9d_~!TTtMq~r
ze-d;Ys@OA3M%<ZXS*gK=trj-Pk;*bcYAWVp<E12KB@p4l`y6`N=`wQC*b$mFbQd*L
z>iDpw-VaR6*7m$U5MoTa-ww3|<{mC2=WoZe85PDw#vq`Me1Z0I_e@Jz{B9#z(^Mfc
zFG9<+aAT(!y8NmFU<GYy)EVf5-Kmiui^97=x=>}wAc#YtQ0B@gnku9#Do)v3!TI2v
zAi)fd;b{L}uQ;@RXXO1S;=Qw!#0v~6li0oAQxf7T*Nh{01X4z0u|AA1@)Zs($jX{=
zgE$cpM={WZhBD};<)^9zNy7)|AUxn~de_UpP|h`*8_Zx}<6Lfyj7LW9gd&`aG*g|!
zSm&#@4ql#g5-EM9>B#L*>D8*4$XgGtzX^o&<p?Y8azY5R#*8^R&}!<C2;R_=K6y_A
zBPWy2&{vy39Y4~7ZiISWjO-o1$V1SWM@K0s%B#f2p^QpVVW}r$SSx*ZgZ?S+nn=O+
zca-*bYA-DW(23fmY8nEigprtzhXUeXqzGVmwxiHoB@AjLK!CfW@@wPcKf$O9&{yK=
zbizQw>s>AsT!hbTc)o>|`SnO^eT>X2{Va)|jB?u9UTQz5r8hv5l9888GoEni`v&Xn
zJfQ8mlZw`LqphPZF+H|PNzv^#u$Y<|l<ef_w#U8&(6L4ADZ;L%DR6y}?M<I*fH%ii
z6FKggj7B?$v#%J7dKlI{Dh~~I9z!vWHC13laBBe+ymYtbTHc=qfLGLt71z)68qLOl
zPOL5~JOwfn3d!KQpM5?*xs%O?LNwU-Z91EFnYy4SM?z%K?7zsrbrg&P)vpP~|3-~A
z;;7|-ogsODo;yc1_P~fQXcMF~az{==aNlU>gxloPmoj=y23gv!9Y2-nbK9;ZzOp{h
zXUi>*ny!aN@+0bD?%&1o`OcBeF_{h<6XD|wq7_<(y{YPb)0XsN>5!QL<-CK~l@<A@
zz`u|hi_^!)gB~f~?bkx)2#Nu^HT9k`8Yw9NT_H<CvMr6rI*~<3a4o@M+nHv!1)Q({
z@lol0`9}(T!Xu=CbR_?a<5Y9bOoc>92<Y8#-6%Uj9JRQVBrJ)4?6i56qRuT>tAzTB
zA{BIT&*A>G!%3N>HQrS|s@yN7bl2EwZ<-SKgHX2m5`BANT(yqVs7jKtSdg78<e>l0
z-F+2Qa|*ZYBdhqt8a8DhA$B3J<v#~I6lm}07csJO|7j9H&mM(DP7+<nS^Xb$<H>Cq
z@-UbeX@?fRO@*v$B*pbOutP!v`{2?hZaypMKIq4A`A%+nAC`sAwZn=St5z(xs)*`d
zjlu{E`ZK0QHOMAfPg#pI=z9Mk1qtvUS-h(6-LCGJ9ibaB!}R@PWl`f2WHpVAFfQ_4
z%@EHBHg`!c*y3~*vN(a|Dv&0c%b|!jK6D>`(@C8-beenIx%*mTay321LCBEbp`F!l
zUzQwp@Ug2KVU{lWJWmlo9j1@paS+2r+YaiQI&5il+rF8*_?MB9q0Nw}0JYe_$uT<A
z9%e#|QM5p@T&{nAg7XgBl)22`5fL#MxpHUjFG`YxX^oz$GeHHUkjP%_$xs9<9pO}m
zf3c`Sy_ea!i$jx^>sVlPt!>P{t}HT_6#Mm;e8sKV^uId4?<D=a{@v<mu={ekQ_c8F
z!;4iKRuC~mh#9pTV@jj=0yqh1yJ*%pp-pA1`!SR0{{?^6$6__3!Ya#E?{P9Ly!!dH
zii$bFV1(i;am!`<7*TZ^+g{0g$?x*;@{W7?p?v6CYiJ<Y<u5KPZ9^76BvJj5gf$Bh
z=r3rIJ=&+ko0q?n?M!8Y&H0BNWMQo6$?6r8tLn^=4$Z??RVD{+0O$f-0w9rUg1|1E
zZ5zYvGbMg~g@AnctL>0JAI~wAZTXn`nye<15N2gZ+GV@<?Pd{3!1aIpE$aXDa**W1
z%aX(2;u;YED-(tQml%kK?19NO)*X36dj17$J0KL_jrc0n1uj?IpL7^?kBpex21^=D
zVd8ivs96QyKTxR;C=Ff^njWEKXmwZk^rqWh_kFs_`Yd4J`-@VYwPlb?BsRkeWbS&A
z^-YGG+7NfNS0~AewV@4&B|uLw*q(ASpL?pVyY7Rl@XtKwLm&hZ<1jH4^h3}oC|<ck
zM6!c9S+qk9eTW`z3gkyszmoPC)GB#^euD)dq;98VygGn|j-^{y--&sS(<xXt{BS5}
zJ3A#JyK}~fAkuqZ>dZHlnbEzAdCd>s|8J?Ij$t?(8FL@=OLAiVQRv*9#H#ZAFU}<r
z7_^DWO^>{fm&?5~+Q*t4Q7@-inRssc@n}Q}=MMsfg?v0zg`hV+xoSM8Mu=xuP&})`
zHZd(Ni57r%<$^77ee^4*g{BWjwJiCJ+cAXp`Vwsh|MD-tn3JE+{O&wheX^8!v*VGi
zaIeRb=HsUS{`S;`LJy5WJw5Hc6eG^6=$jL?`4CiFL|ScbXpxqaV!WVS0){1Lb0iUe
zqNJDviGT<%ZCFYMxte^S1uK09XILtM^b-^}vman8Np(1VY6KTnh{FP2|6i3LwQPM5
zL)z&<0q)v`?Zj!I!{J)fwKd!8WmofZ4K?rM?X?$`m2|Vq>87tHhH+J=R0G?5B>W|i
zEh;Ri7)(@XKc94RT&aA1PbZK{&_oMTn}5MLu*@ueb~m3om-_jAETI8rQt2A@tovGU
z2Iy+am2~nxxI5_Vi1jdRzkQ;%-9@d+(#<0Nrzk`>^fNWhZ*i821R0YZK2%FW{_nyH
zE5<f+zS6`4VHj~eyd1+xCc6+sclMlvOfoJ#b%61*Ttmg4j7tHTBAk#Qcd@QQcQ+6|
zqcFKX&WeEk*>n)8BIDy{4b43?Ch6l>1^i<vyv)#RwVE|2EA{=jt?m7MzI<==(NZQ|
zZ#MQ?n<{yU3YwBCo++uO>Q}*$CBqa6YhmlNg-+k^bBYNroUd2>Mh?S%F?CG&b&d9A
zo|Z%=w`>kW{!B0tM^F~zC;4BFSE4dSUg`*z3UDr}!Fu)w35lk8hHySK@&49K1hxrz
zqlpd1YQ(QsuKOLnbs$t>Z0|u8qjM2=BQ2Ukt)#DJ_WsV6P`gTCZ>W>B6;!W-`t+EP
z=iwh1oma$kC>Is`j>D#h05{u@5DS)(Lr)9~%2kg7m$~a5PHwX$v0l|mn6$}<mQ>QP
z9V>VkmsB{_Hdpi#Fq*r5RV8)B0Q7;RRd_VN|5YNg{9H3uR#VT{@5&qN_u<NJhVA#;
z+Ly8xyT<O^<dj1OSy+%3ujNBU#A+e6l<MWAGb~h<fntrtZ^hAUwC><U;j0c|vS<$|
z0~!(}>Ck;d3Q`Jzid{MF1w3kx@C#kayS<o?$W4n=FsO~o=3vw)x325rK-qFAu9Bh8
z`3JxkF_b6TVM&?rRDQ2X-EIvJs!A<|*>D7%GiB0Y31+sd3<U10$IrGkbfo4clv6`w
zBT4kYFozX{k`fA*W+&Kch)b>=?pd-J^xhsn6Uqs-$6vPBIsTcNXc7~|`yMeM^lnMx
z5y6^Ga&Tr;$TBRLC@H4{NTYHU_^qE;N4sepP}l98@foP8_L$Chn6aZ3MB9d(v;-L2
zqX;C+^%1W?pjhsuX)Lgtx50X|DK=vdKaZ%l{E#{vmM#+r;Ap^1<9Sf7$!vYzA7sS<
zc8)idh;Ji@v~U7H%HuG-EE7u%uU?ZRW#W4aGrsZ}T_Ih4PoygZOpT&#R>jAc$)H_4
zLrUfP7dcxzPj@dsk%jh9zOc`?`;);nP^S4_=ybEu#9O+d7g;Uh3*7BHn@fB&OhiMp
zoFkzybU(Ws=4wFMSx&({b(G%^Bz?TmV4?nLJ9>v*I*bm4jFN)NF&!N0&HkB@1)}Z0
z#Uk*|{*BiO>KTdvu^W?NK%CjRz#(43-#X8umC>rmUca5?O(n*Np}#S#Q;Ja_IAs%8
z07yIEkK{%3nnF}Xh-bQUX0j$HXD%%(tobF#Rup82)=N1o!$Klrj+hs6b9m4>tKS12
zFGz6>_<N%+56eEzp6e{>Wggqgv+It;zGV<k3q_14uYv)Db92eu_li|$L#ylSQ&^Zu
zBhHll=r}v@B^)@<Cb+xcPr*3z*Ze`*O~S#(u${QnQdXvW(qh72N=CQ{V1jMQtuW$3
zDTpX7A=F1_zc*hFB-LW{I^H4_r{&}&fOU)w5mSXj8JF?v8~Q*Vn*Z7_JwZ_Mb6RT`
zhZ7>uYJ31x%I)-<NvU8+M1l8zx1xZ*@88$mlRa1u6??-44Wdi)nRSbIsa^C1$IaqG
zC{bNvdga9k#xk2z0=$Cp>LkSKCa4Y-Dr0EM?7Gj0ysVszny6?t-VXb_$RWYbw)Vlz
z#jP0ex0oijzjW>-QDK2;rEa-sD})E|1w1i!dT#(mRB4mFfbRkRm)*S09#6Zi)D(M8
zLh&~SmSiHriJypU)e+?cxiK1E9}*NWqAEdbTFK-c9pK}spFw?-Y)YN^cbWWWC>UO_
z70*5t*KprIuQsdqba9pGBd*!ZU5@-s=sRIe4yrQG*(@db&DiN!=$T=;32uy0V17~m
z!0P}O>eLTWked%@z}Yjd$#WlQbZ#i5d}xY_3mk1Ows-%dcj4eCp&edWi6;kjnU0O;
zMZr|f$x%?wVCL2Z)rTfj%R2e`Z&YCms;qlnoA1V5t~DTFpmOyZv04_NFwuwTozHc}
z9X7=dUV;&6tjH0hMGiQahzWS~l3n$e?g40-!p{9Arc#@ELcsgO;1M7Kc;4RNAJdR*
zEH1_49#}x33iZkRGhCOFyEEA{#60q79ZKc~1vDqBLI2Yhqgpkk3?}C(Ts=G_{~(cj
zHr;^=RjM=<({s@D$JphGaK0gDG{Urn<-O{ZQ=DW`l=j8d7T3@X<PSmkrZLDV4)`jV
zs%lN&4HI^DTy~i#z2*EIeRR|hrDA9V|0jFuDi?V*@R!{1oU0if-pQXFCaeHo>J*%n
zGRpm{XkJY}qMrY$pzNWNLDR(|B*kUzX4mV}ktkRA>1hG-?V0O94s!ex9_aC5LIdf9
zWqD^Rr{Nkjaf`!^&IfLEEXMlT0cM0iyO5QP$zt~8C#1CGmSz=2hx@yZfnp=euiVZk
zW(&o(CI{Of&J?Jft;b(aCA_eP^TaZ?Egs@)d=;+>hfps7VT<h&QJK*JMT@<pWZ35L
zSU-vC-76<lXRF0Zg<+VOg7@zTB2nlxio>Mp7ZLQ&W7{jf=p`YUR#qc(0LEUBS1z!+
z9eKk9AuILY=|8WRY$Dx{+vkn5BVp8kdcD3z8`-N<s<OiH`(V#W18@;3fEO^@G$2Q7
zkG_EHyMJYzjd4l`+dOQIzUdA#Ve<WNrrIG!-v4I_o(<_;GfyTh6N98M&aGf7Z9&Qa
zJ{EE&x{@n_g1RK2KfO52PxqTmoZR1_BKeF^Z$W^k$N9Or5vO*|tL$}Tuho#c$93&A
zw2`3lCuhIjSgIA?{1hrU<IAU*#T;w!JeI1Bkxol5nUka4F23PvgUOb3apf(@9qIiL
zcUu8jf(Glo-rs(B*LCxx)_-ISZ3XZ(SoaTQJJAZ$gKUG0#GYy`PX%<mFB!Izh>d&5
z2q+Ht!ZJqc*Za>g2Peig7B(hQyfPh8-}yYDWN6H>HI*2`@$;K|II3p)65%uWi_}<c
z31u|*5^Y8vdely+F+l~+Rrl?wd)a)bo0Ho-;8Qa1TFRNXxkgBonFXoksU&}xDrg>z
z3nJ*C$Qe*Y<9CY>;%5q2-v@RZkiAuF2qp+x0<AS0m)!LIpDP}x1=GKOzF$v+)}@01
zzlY6e!{H4TK&X_^Ki3CdJ#V^}+M-bs%7;H;ipG#zN71UC-gIA`4))^&J4OGGUJfp)
zkz*N^P-<}QQW3JDq6^OgvCM<2!cBwv7(x02DqzZDT%M(`L!u}z-riByVQny)I`jK)
zJXc>QYTl;>{&wFO534Xc1Kbl4h!DCh^ixreA8i41B!Z3gO#vphCAdU`?FZVwds>z?
zqtQCGrgf~bu$e8VV}=EB@yO4smy_VEwAmb?rh0o9wd~ek1m3oVxVI{AL)B=L7E8D^
zFn~Y?8@>$Ju+UKWlFG#^-VS9uANTri9W4VVB*N8M^b4{kHxhsH`AGI1akvkY!rB<P
zpKn|xXzo^e`zXkc)AG1Yb7djx*Ixs|T@jkOYBznKFk|e*Dg9=mOU8esn1}M`hX)~<
z_NOUAdF3D`^30nk!r@Ma{=#FkquQ99K?Zg~Po59cwD;Y7IPKUR*Ph?|o%^~kNwC6$
zHIUc;W^Pln(bko6V>%gnEXaLGh1#E8>TC!U=L8BPzJf0j+#O?y&&;G4qjahi#$CR?
znBi)27<tN;qy^Y!mxpnYCcZ$Et2Gr$YTrHHUrUhD&nZB*@*{+?%%~rz1(Ok40$RKt
zsH~2pqMzwiH(@K2JU?>Vo$i-{X<0#OY1EK<)JwfqD63H9AAeE~j`LwZ<#1Se<?C>g
zw2sqQBrW{;!sDYe3z~ldSVJm$E&WJM*@TlId`F?F$sSRv+De1(Xk=D*u?tDUQ|m10
zdKa7o`j<>bPKlfttn#9ZBaFCWK(*p$BZh=uHdBUKazb%!Ru6u4f-(diX#3b0wHyMS
zELdickA)JoB@`?&v^j~g(Xv>`#(vK)sOEUOMoaR{8{YjeeWpseXUg^+kPLbQS358|
zBF^~6VZGF#q@N!d%;w=py2a<?G@2>RS;DK5waV9SY~pAAO|Pyqrl5ifm@`M$O&$3j
zAA}z8@pN-4F+58zv&ceH5c!NI>bwHOKIX`Sk^yQwK8@fm&T4lb-4QFWC(t=>3U&Kn
zQw1)~<mqK&8P&3e%O){(Zi0V)Zz>*F`WXiYR?exx9B6`94PB$`q8j=3HeV`Fam4&j
zyuLp-IZTL}d=m8lb84s=xK~6A<>L}yUmu;9(2~#g;3BNc3b2asW_JKPQpJm@GONpV
zZQ_}4zD81E&f&7<A1?BG4wQSl=Y(ykws9mk5g3Yo`LDVP`xTiUUz<#<vJAopF~mpl
zrJnUU21eTDi*;saXbF}|(9?J{va+fNm7u*vx*B`4y`<%1cdHygi;l6iCRy>pY+F74
z^Bz<uJS=oVtBNFb`iGvyN<#r!r${tXvAMKtt*(9-Y-sjNSUAlnhu#zF5O&Getb(w#
zztqT?5htuR$?L75>324_`A0YtX}A_n@Sh2h+iO`AV1KevBIDaL#$DYtgz+=-Uhw-K
zo_1Sae)`-Vubu$P@sd(H4}FC)LD3GWs~j!}3Jf*`Tzj}f9+AK0(MzWMSg+YgwEcXJ
zrC?~R1(L<5N%%qzxB?9lRaO6XxSbUsqLN+GUFR7JcR8cH5#uOUk3~4O&czQ^-~UC6
z3w+bte~q0nc9I$@vq^WBh#QWCt?(1Y?Kiuh!mnddOTz{Zh+hS(aoIDWOwmKLi!0Cv
z1u2@#O5(>7+e@%!uh{c5?}t<4zUZut(DJmFH21)D*)$b<6*tSl;G}8<4&|-(wwpSb
z-Xg^lfA4W&=3AHcT05O!0IPtZdS(&}`8_{eO)9EK<O_-evp**tI)*%5+aa5%n$xDG
zC+0Tms@efGo_ZGo?$7AO@01E}lc4aUhrMB=6%9dhMp+bjk<GW@8`5#10WuK#o}+JR
zVR%U+Et&Z<7mOcNMhe6XN=`Z>bW^unJRE*rJsopG9qd4Gx9F(g7+5#;-%H?iL7A;^
z(C4nC5nTB5J@ArOSzhVO?AX-x^sSbd*5mOs;0TZy2)8@#HLPqKb)t<g*Jg5-)3NvU
ziy6pcGcQ-GE(mwOLj<vF6gys3>cAFj&1%O@alklbgfShvLx$m@SVe$tDuW+i|N6vl
zGe>Os`#V1_Tz3<Kwp;-b#*s8*v8HKH^}u;ONEP9)=#UGJC9=YyOj}4XnzDs?cjlTZ
zg}@Rr^F_7NeKXj{-&v7lt7paRV2s<yN0+*v<sDE*Y!j~oQz-R6hM?O<;6P+1C>or_
z=}e7O^n*b*b#&<RlV!WzO<1+3#=j(8Z+CAqp+%ib%(?vV?U*x;h7G<v66nkLbQR*T
zJI9B_cxc;~ah{{equ&7)q1vEGUAK>+PrxJBQ%5<{@Y~vgFjsqB%P;%;2g!}r+(dy<
zGRJk-y(w_2m#j3oY<mHhAivw0p?`iJhv+p1xX-wl&!aANQ>-$zykp)$&B1*={wOUt
zQh7Bsi>UA!sWCHO5bo>dDb@aT)lJOvLKm}C5K(KyoOca=MX;T4`Dz*z7;Jomp9*NO
z6<Cwt@cZ39y{rVU{bjQI)wP;g!u>bJYy`SPn~?0D9QBheUsqN(Bw$a`uvUnuPwPsB
zE<3^qPBgx>AEhdhvC7Qqk~B@wn85zzJ`(E%xGj0?|7F^_)P%{cq!7~DUEjg=Y2}T4
zPwnk$l%XXJzEq{NDlnx$2)k<L>I7V0VaDC^L46w)|1#}~MvNzKIB3(rm!8rao>Siu
z-N!c}AtSG!mOPu3h?er3Ou)7`LTVoq8=Hh^8)Q&@6|weSn_%t~h|BJ9x|k_C@?gzb
z?lm`GbE`YAltGa58dBPYrz0x#y+eppVl89(&q77G{^3!_wWyx>rqh;-1l%Ae#~UJE
zQ)UqC=jSoHJO6NPy=vkZN})1zAr%GVe!HK<ZSCyjVa?&M=c8VjT~XK44oaI6RDSy#
zuyOf19x=iUdxt>{qQX1I1Vb>xb86N)sfNYQ^oVRmj32JYX@Y(5vha-TpFPI06A6LU
zP`B@hxqw-S4E_<M<6r>!pCau1+{{I2(Pw>r7;b^AfiLed|0p|onew*uU|W7e^JiGg
zvlo2iS}~aoTp=k<(dWV0SY^N)_pnHbmeJ6%^vxSii{Ds6nsAW&(`g11;u$hjo1S5&
z8S*wlPGzuy61u0iqmN|aSGy}owzTfG8C!B1cmDab<g^3OUZ19XK+bW`_lwJyaXAEO
zJv|8s81e<blfC7c+DluaRepP`qoAPRU~$|`f?xIDG?y3Qa0BqJ5H4C1qbdl%Hk6p{
z@M3F3`qd&cpbxm^C|w30{K8d|5_|qSZuYn}*uJh(yJp&qrtO;t1;wHk4rLLZ+3I${
zQ15GUPaFPOQrFGK0Zz`W79RUd-e(Y2OWgm7qlu<giX%_Sr|`NVC|9Ppc;4np(#{(8
zel|UF(mOuhT@U^YZ(+(Y$RRuno`Rx+=9h^F`<gmLX~C!b?(uGx=tcz=>_Bcz0}Bl+
zl%H=@v7kkcB3VljA<G@;sqvd#8v=GjWmhs}01qDS=a+E?FyP<(99pugxM122cCH?q
z@o5*fE$<~*<sstVI)wcdN*~@@Jw96-C2fCi-?rZHU}`xod_scbvlVYyOHE~4!&cyZ
z;1GUAYNN7LzB1(z*-irf+O(Zu5v+`uP_aH`dcEl8^r!}$WVpFA&amtKb`O-5$%S0}
zci!^@ktU4_7T_l<Y^7r*EZp+Vc--N;-^%hz{1wBc*w&UlgiFM=^(&8`Zde30BI=96
z^w>Bn194pY<<pjtFZeynb8uu3hS*K|3k?}lRV`#RBW3CK;!QasFNuKn%{eH27|Ht<
z?BfAy5qdxGdhKh~=l?)p+5?~Yg_<OwWKUek@B>MK1+cc$-TF~r`yMnNdfmelK|A}m
z*O$7EZ%zthSGJKmB9IjFhwau-LkaboN_Nq@Fqp|R#*P2&{!Hu={7S;V5;iqj7dCf3
z>h~-2-EdT<>msW|p79N;*f>_&8>7GU@MuE%RLnyk<~t=0c1P$Kd5IZXdh7-M3a*;&
zO|2k69l(40xyO~%iUl1nAxXj_4&M~((jZ+@u1?|0@D1FW`<92K6Vc{Uu7tF!GQQ`M
ze4m@Y+ZaM^^<_EBY7EZEAlg>S{UW<L2RB!5xzohkE3x*Vb?E%N8_7qW*mgb!!V=+X
z^mcUSift#$7O2*F-Kga#N+BIUqFOD|;5EOP2uK0CO5bxqgA1Ay3BzucOa1!>*FJ)I
z^3P<AVWJHbiG3;58xR?7Xn1FBbXV0g&=1Y%W?%J1Gs<4~=U_-&kI`Lih6_3sw~_0L
z91Vp51U*E7RZEpD4QboTIr{<<D@(<*W!IXcUfI%CmNMmgZep{7Eb^G0Vz2V_;+`*N
zz(=z5n#u-%6wla4=Ii@2fcG!UkLn!VG~yi-I!1a1sPklWL&bABVw|OQRJz-zC1ds1
z@kcP&>q4!CIZkLs3T!vGAb-e)=3n2k{qvXd40SX`fBh2RotafOG%(taU|nIZ>Jj5(
zW7<j$(Ytk}TM5B#s^a&KJF&g4I3lqg1sbfmJtx1V9a2GcYE!C{3X8aasb44Bk2-Yq
ztTx+tTW5!(C+>@{w;NNzV^h&B%X9En3k6YG!-&Q~#-d|+O8=Q)E%)^v&HX2<k{d`q
z9zF|n1MheJ%vTqy4Q=H@{P6{dy|_k9^<!9^_M+LIRTD^2$a1alc$ZFBh-4C#Kmq^v
z)V@v`4Yc6WHfb}Xhtpk5ERv-3RDD&QjI4mR)B(F?<an$Ms(;x%(6}6^0Uq#$M}v2=
zv&=mv4<neUAMaVo*S^>1AaiMB6!l-|s2`3|&BET8SHL#k-GC}@olDR{jrC60yX9r>
z^6F+h;ZD^k)^X4p3UgFwsI*)RpiGrZ);Ako41|#-K*l6sU5$)|^3@&ITy!np5|!;-
ziu%6XGB=zGueLXuTG%@@rTv1zv<dHQ>4_X2P}#!9J5jW~if6;`8}HkY(=ftm`9`LX
z%HgI?r4pOvi+h}<BiP~n3>_*vN(;J7_^f8x&__6%=7U{e>?*S;e=oHL(gV%&ODn5?
z71UIi5T~pCZMnD`1U_HxKBgxvg;6i{d!1g!K8QixrqI1^K1TSrJ*B1uT$+uqXfa|x
zeg|o;q2#W0HZ(JEM-;^1Mp2a^MuDsLs|JaOf(Atp!R@?>g4fb4gel%~-A$iEBq|W#
z+Kc$!;}5KMe?FbeZsrOCX)3|@OY32Y$wro@=3;uz(S)2gs1}WzDxlYSKVj6HHr_2H
zT7sbj45S&DkPVVc+hR*Es1fna^u@DrH8%Wka(;Y3M7VWrlr@14!t?n|(GDTAErn$S
zmZxkfur>T3DO=MEN6b4rNIzeX!g&LzdcosfL9bsfXEg}Pi3}{JR)O$E^U-jCi?X);
z?{#Y1R?_diU{nQFiy6h^=zC?nIM>f?DLhd9ME0sv;!&c(c@yB!#g0s>YRF8)LPf4N
zI)d&Xa~i{IDbVtuGW>5R>gDEgz0v7Ym6fBeuoqr#L%~sG#`lsLlCPG=U~xj)K9F-M
z^bXXNjHpjAWB?vPmiB6VDt11e9a!wHkv2I3L$1Pqi2E0@wuVNJhDHwVId$l-WSOw?
zadNf@80FN+9|ZuLeFz==c6>BEr`Bcf-0kI2D{J0*_${(mB5N*hQGE%OApOAB^b+2<
zskp1SJvhdV0<EH5X1mczg%V;G!-Ax~Vsd|QBJqYh@ZBET>u81joEVSi8&1hcn75TQ
z`*H@+_xy=yB{2o*t%Y$~rtU81{uj^+zU%S(!mrr_X4(@od8jOH5?FHj?hXGa|3(Vw
zy`Pb3A;y+SCjG9t7$KvrwS~qNZFe4Th5#Ald6a}?rQV52;{>|f9Os>$*;ohQBp|sA
z1=r*Ny5D9+=-S;B))pI*jbQ!wNm6n0Eb1omcjrOlQwls?)<z`L@;+Y6)A%-6J(7C3
zE*i6FmB6ks9cjP#DM%Yn!7*jSy3;BBww$)r<>(e*ok=UG?bDc*U^F8Bstb1O<?Z$Q
z^x{Qnl@@dGnM<S8^NT;^plOPzYcMJ5)RJWlaDUnjZXMIa!uY&9T!GXTIOM@d`5MvL
zoLfj&2i051W0Y1blJVKYDn1)0XMhiFO`hanVQd}Rav!Ta!*UGKF&DeLkZeKp9u;47
z{~bFeVkVd?WcMq_qXM11PKT+st9D{sD1zuMp)sHE)xZ9JHkwlJkp|+?TG7wP!Aj8j
zAF}PSrFHI@&71$OrIwEJ#PZdinkFdqtHj_sGgEVj&Qh~to|<kt5+4daO?ec_91Ki&
zNLZ+>jP$5f7&K(fEd0i3aMpO>FdG4684_M!ao6eL!w(-7gR|!V;ZUENoaM%n>T(8J
z^2H?MVix9r9b7{IN%;-#knQoc2il<n7$Uov0AUBs3eBeRkbHb#VP(ZyJfEMSp+cX0
z1UUOrrQ~D3i3Y)dVTTk4wH6}$<`h#M6k!4;7B-F^S+aEScJisGIMxan74X0YAXshR
z7F(KMmAAaFFMY1?gy)-G2cx82bqHvZ4@+cVA!xoRhxo3EufU<9Zh;D!8Js#m2{(wu
zP~HPVj6!VdV@y3$b`wVbR8b5%P_g*t?IJ|C@k()$*ROQY7c_y_THI0YYUHbYu;|6u
zN-W^<%74O!JozM>hT^C%P{c#3HAVR<L6I(oQdJ-JgQABvqLt-jlSUdMBfjTX3NxQ_
zfM&wapN*WJk+4wV(!(_Q)1@guBnxLJp9~pBccMTN24mq-g2r0U+3wk-56+)bPyC4c
z+nD3DhbMYhEjvNCkAIbbHK8-r4=4dRwc=j>Prk#?-PAuGaLUY%It@k-cjZYoBWe4z
zvQGk;){GGxjwn3FUc_Ji_VLv*N}|Z5fM%p*(8EjpBljmW493A)&(e|s<xd$o8VjBO
z-6*{(8+cv4I8G$q4~D}Y3I3gRq*K20Xy%^(mb`6w)+%c$FO=Eq`$!o#v?RyL4Ngur
zT4{BRmmUwpy-kgD3S;z5>U9h@^u>pKpfC)B=V>sozTHSAf^Iz#riPt*n)sFF-<cId
zl+)X8WGQbVhRq1d2fH>l0IE^0wLL;zPksicyXb@J54meX<qc?55gXfSxJQJtx+1Ye
z)cN_dS8<b-x${ClZiX@|qSUuN1hUTPB7&B7`M4~1Q&efByEqzdqZCF}%5pUHzk$`z
zfE%C%rw)U|`mez=5(+}o;;Kl@_-v4Azu|VrK?gpZU^Y-};2nAcHVuVcdoMZDs=RYr
zDsLWCB+f?i&w7PFF@t{~d7iQL`NVUSvBLt|rkonBoo&GlUHk%j_5GwMcE|-}jLhUT
ziphU(!e)%MF?F=gPp!)39-1}Kj~jc@)0Vhg5h5mQ(m6a<Ui8B|9fkaFXV++{2z&f2
zOn=CKGYKIL0N0!cSU+K_(O4A{UQcZ*X*FVM=~$V^BjBCt7d5e@1!<eyFvRH$Yq$Ty
zNfh3|-^VZ`YYwl-Cj{m~-lh`Y^$FM|>U;x{!T<bDt67`>akF5rTJ!!?gBzkkl6|Jv
z8sNDN8DH30jP|(fkITk{sg?Y4$#UkU4PjP^^#{Fo1G*ik%M4D43+&0wt%+@6$xj3C
zva@$6E|tHt!=T_A#DHz?N{dej98i(aQ`89cOy7%~2(@ZnkRU?d|332MX=-~yfR(4a
z!sjtOq7SNz_PpL*h&uX+?b&=CeM^nR_7dBYWEf495Za3`AvqPL(!8tIgDm_V6&3aK
zi-N-B#6%;o8?;_UcI}Br{tGHzj91j`^QYcU=GlacST(pW*EB3+3?ynr%<fJSuM?aC
zd1mgtpn!Pa-O5_8X!oVx_1%2;+TS>PLy=(5ZQ&^-t`zeRm1>)=Uj-bW8<DtSWTojs
z>v|IEQJG;lL%(id@<<UeA&KpomHtC3VHd+i)|~+`Qk>x-g|94^_1o!C<y7>HrSgI(
zxYXfonSS%5MSt2lazPAsD|HdD^Z1A(X0&VH+ozHC!OkVtoZ$6c$PGKXGw&Mbd~?V$
zKV$Yn>^i3_q-cp66=-SdgT3(<S@~qFxv&o?Dj%lvY!Cq(ljCOvLCLHxkI%$Ywr86t
z84N)GQxwgUu+ruqdWtTm>u?QZa_qqt=~j>L&DG5P<k3T|Za_MxQ&|+^bzR)4$BqJa
zq+<8fWzL4YctC`pXuKjc6?U_toxa~7ggC=eblWD|esWq;iZ~?kHl1}DL%}$);!zLH
zFO%<qb?Hzxs9bjm*GShx;yb_SN2BJ&qIiK39v1)klVPfsJE*o9QNQ;c(va%)dUrKj
zNoi$?i12FiUlr0sOk=y~Vnlu8b<S>&$dM6BijxzpHR<sxnXyqpwNx(nG?UnbK9-O9
z>34;v<*un<-za&Pw)(Buq@w{cG(+fA7Xf#0dGoKJ!<Wm;+aFMfr2FIQF?`b!)e$i4
z7Rw6j%t}JTQ{St&q-DocDf#Gp<g#*yN3uAMOYi`5GTR$~6v_eXq*3WWj*1d%RAv}Z
zP($IG+Rb=oKE4p7n69uo4e~5QhP0C+N@x&f91qkvf*rL?1I^`YazEO+{k~oKIN!{v
zFP2`xsQmNBnQcBE0~ecOrMb@$HwBR5F6*gp=N?3#>e~D28pROd&aw^~IR>KAzg!b|
zzU*&fX6Qii{*8^zX!c^?T+)fK;m4zYtMB{zg);4`YIUXZF{jgI@BeTiukV$)pl))W
zK$jzP@*OymunWNmPX~Y>Qp<C*mFEd1tp;NvB%~s=tI(0k@}Xvp1rKN^$TifVmEn^e
z^6RX}o4L8GV;nHXkgulxES5jXP-Sr+k$DGms^6Us8R7rJ`;XsyXXag2#gME$Z!=s=
z<)SoyzrQAR)kQ9v=v$C!q_?Vuy0ScSA8(ipP01n70|_|#_6cGGqlkN$)Gjz9jL24r
zg~fJ@u`ztaec{W}uPmTrX*w&N^uo9#8p+sSBv6z9D__ZActEhG%z1vvl=o8j+zZCy
z8tC@<xH=sE`X7`>A^C0yvR0;sq{qyUbFXrUv!DMHFce3X0L{vlNKZ4Ja$;e35=W4q
zg8s>73(1W?5fb3pVqJvI<3cW)STgHh&Mhe)TX!2dUU8ba4Dhkf-5H2?kjAqp)z&&%
z>$d;8-9PYQHN5Zfd))aAGV#3Lw-K%^4i(GCq94rh7$Tzw`-kJ`(mKkH7pbF^nK1<)
z`#{sd73x^YKI$9r1ivj)Q7<VeMH4|LMbS)ax8LqBg(e|V=pz*;$oHvOT+#*|`$TR@
z8c6?UVXJmI1S<S+=Xh}5ZyfVBAsx<_xNixflHTyW8<@6#`4_6}_Tu`RBLo_kHwDV!
zSH#~Z4+Rk3LMqxJ=<mHTt0R&{z3(EP9xX7Ri}D$4_4@kGzEE@0VwZ+=WZ25N)IU+f
zMFz{}My6!c_n<5%pdCR6aNSmxK<uOHsaX9s4a-XE2AIiS?wai1F`{T#zi<7pzI1U3
z;HxvVoP+c+?vOfEve|NZdk{>Y0Y2W241w;GcE4+>qn76XTpY^vDGlitH5w!2CCOas
z%dN+u;25UtTL^WAj>FhZa_2g2OFY|&zzFvH5KwaDdX}n`9?vlt7BiqUI4CEv{$JXf
z7>x+f>+q2HLl8BmTbYVvl50!A!}I9$;pBeUIzi-tEreJ(Qx`oPjtuNif92fF0*G#$
zQHi(mDrK3>J>@{B6n9m(lpr~2pd@UHXcmS7Q|#Tcsg{wVWN$03K|wq4N9r0t9KwU~
zckH8VM3$p>H0Zg#+ju0ks#yQ&lcwVCp#KLD>84@b+}adVi$AGeyp;7zcm&^O<|bP_
z5!;9>1+3iS!@DzL8;ZdMe{;iry+2+B&pZAO7899IO)GA(wadfCU9<F$F9JgazGI`y
zyTWxc%1h}_D3ynle0Dg;9hovA3A(j8S}YlIC^%S!S%?fY^B0Rn94Z)rtP1iKXk>dp
zm=aGYFe5XefZ2MB4;Ihk1l^$JWla6CH1V-0u%+JAYdC&4PRq{!J-hjJR`@9(-9;?L
z$7S!l-<}bdixQ_TG?fVuO~f+vUHv=z`Y3#5rWv<=vHf$x$q9m8-4QfekyDt9VX)=<
zIu|i%9Mmg?)pY}GCVl!d;Yr!}7X$!M`@KUJ+2qB|iCev*O!HxS*(f_VPPEePKVCJy
z=bEG%3XdR^>`0j4<ABcpV>ikx{H22<9y>80sKC_7WqjYub}f^kv8c=bJ{}cXhxPP?
zFg{yMkoR+JKYJmwyxE+8JmJH^-RRi*CW-zn0WZuPO|H!Z)JZ6u4rdVw>gojmSPlN6
zWZSH7IUv5LwPhJY|Lt@$;^ruj^7Q%XNP%1cR}u0cWr}rtw08h*D!2fr9B+$skjgAq
zSZ%Zs=_rPzc~}~Vn(husg<t$7NczVMOm*CHhts|L9M9SN(cq+PD@zUy!fJ{=GYC{H
zl@Z8%TE6pIrnX*)pi_c0YXmI12QXolU4)uxv@mVdn<z`lD}b^LnGD3I0T^KIP=dJi
z{W?4UA10^`xLQuG8pw8Y;zGXqFpMOfxAO#n#|nL-lv%Sl9RJkpY59owQ|W6M_I7P-
zG+STaW%|xVa{o!h-FZ8_$}yMa%*Xs_-I5|Riwqwd<`p7m1gKdk=GH=ujiLVGngj`L
zGb0+oqdOlLCroiw9L*;LdwAomM<~(`nPdGiS`0r2dUiIasDGJX5a)1$OI>u#==p#^
zzrRe{EUv4@Q|fG8Id~VS^`XS=Fmt#m+lG_xs`SJ?X_p}@(9luvA&;)IJKE1s9JB?B
zXn2?63c~f8fn?CE;QbWv|Dov|gDdO8bsO8ZZQHhO+fD~Nwr$(&pwm&uwv&!+b)37u
zbMDFCU6snNnk#G0`99AWTjplLQm{LL!KA3DsL9F6s;a8<D8iY4h+*|S<E7QMhvtzr
zsV2F+^MAiyE><Fs>0b}GUW|G-WN8{F?Ly1*Qtmf>Vg;tmyu%z{9{b^9R!t<Uaf?ob
zcpi1hqV~?foRX1-K%|WJmj(MoV5;0f2hkZ=?w{zZ0TiN;sSu{jFhw5@Q8*G$YMGFr
zQ^sRw8tiQK+uJWVD2UGT9Splb94hufm0~NyOlU_Ekno&bF~y-6tQSf?`njZC^3sCk
zEe3%1c(a%Kt1CJgSjSLRhMtz4ClnUxOCJ5Qd7z!}Ym-5qz8EYbB!P61!WbMm*~D=Y
z<O}ln04qk_zW24^U(7;)5`t1cCc>?@re+RG7z6F}vM}F~Kzkzb+CegMdaRIXZ+}b(
zZfOETkpRWoiTHos9IBYuSY;JZX*$VhwhsteXgX&o*?!S-<Fw|UJ;@%_mT^5B!(l$}
za+JFyPpHx2P6Kije)I1)7k=0Y4#vEVs95Y{fF2FBX8{?s)I<y0#Fu<|vFRaN5~fNz
zX3{lGyFyj8w+7a7Mj(-yaC4rumP>eMc4ZKzdJE3Y{L{ar9^78ckP%(KBd98E>;6Yn
zq3mAbFbmu;Q<cprN}2|T>+^yqf3o{+VWCjZRlw`V_yP2`pZ0Xk44gMBh@bH-&bt;B
z;EWq)**)Oq77Fa@Kys<x9}Rd%<zysKPsTZBq#O}TF!jmmzedAF1QE9?3D3)tC_2br
z<S19|QUkJ8;De72&AT#I^pVi@o-8>edXcXn{)1YZ`1$Mo{y96q9-AYhau@<N@mEC;
z4Ynf<6<HL@V(Ayo2xYbD9#MuK*m74TApIx8jSFi)$ID$SHvz!qtA8Y1oqpkuOniVW
zX#=aln9S_f45kEs(GH5%QU^KTulw<Da#M)zZA?{8@5G5}C&OOT)dBU)8v);k)e~fA
zW5AV^$PW*|`vVR819SJ)4RJS2)h3knYKJ3lSABq*5{$Tb)T^;{%N<;)VLhk}Kpkuj
z6(4*To1bWOK@GMbfchZ-9PI`*S#_SH(fr{{U1I;+&-HWdC*}0PqIeI%lxK1jqM?+`
zE*n^87^iP@FBjj!R$ugK7xi-670>tEO7(WF)hdv~vR{q*8ZdPwNGMVOdzh4jA!qd2
zgNX7sAOMP<<x9+9az(hS&h~Bv0EEIAcR$?@|LPM42G7oMRgPpRUHK8IBtG#ArR5)=
z`K?~Pil9V9h6m|;JAqrGVRd=5o~iU#UG&69DbF|$)I_PE+zOI-DkM+~ma3X+C=SWB
zd{^}EG#ivU!n!Ck1H63YI^(eLEm;EDk9#O27TNp`hd<ZkCr-~jpEpN<Y4u;)GX**_
z7;OkelTD^}*%+E4316X^JcPf=*e6&|v*ZV+=!<(Shj(HlJ3OaqnB|aMe^QRW5Ri98
z?I05{2Su8~rK9C$<z&m-*qt2$Qxkdp#E1@Q#WwqR$?EL(m`Qt$+u@FvmB1Ui635Ge
zgp3#pAo+0L#T@#1h+7lT#Y$qki!2aw0Yihw8(x;d`BRZm9?^@%1XA>mIU{LhNqM8<
zj}(`9^7tlR-mWA7%X||(eK2EvXA}z9-zt0jEFF<=r5=cp<_WTWxHL1n=&Ec#2}|wS
zQxYTZ%EqmL=j%Rg?-?&Aoi&5>_~}vG3{KTO)s*Rw`|cGYQ(8&Dlcnw*sYqqYO7l}Z
zeN}F|+3Ln~#zx2?3$p!!Oe7C+xW#h(r9g-o`+yQ+Ppk~%43Ibc9p6vtNgOL@xA{t)
z!S&Nw0Osd-Q}h>nT8y(HulN1!2qYrp{B=dlCyneA;#+d%gAfQg-%tWp5_!zj?NUlK
zW%Z%MN9+)IU<wc5K$KcCDKnJH2nG&x9qPHw&8?+ncYz@^Xnkxz9p9e{QWtm}a^W#M
z8PJiD8)YKNe!1cXg3j*urz4q#Kcq3dBRlG|_{$0O<HZj^X0Y5A%G5cqBdbaZ(*0b-
zkVNhV2N0x32PsLYTH6BO260m-+|k4Yu59QEPzuZFrY-?!ZX+j^Qow9>woS1sE)3I`
zGlin`oIEf4e^rl_gna5?#P1j1o5;SM-##Ai_v_~Q{F$>=6EL1^N#=SD{^U9vU7*LG
z+-((TNm|@y{g{IW8pqz0;iwv-OEET8UF$ad)UdX+3-S75VrqCYm@G}j;@QuWrdZw#
z9|9@W<p2|fo@+Z`r2}yIID7%*(|o=B&5hJnx?FF9@CdBhM&P1ME2Z^An+WQRk!7Cw
z{}tSyo_9zwauW<yRpR2j+WO{)`0+cUF+WzWi)DdO)pkl<*kU6jtBs=iOW@ca3;Yfp
zcw)kdrn9tE6jWKwYcUBpd8OwzMf^S{2Xv+JcRO!x_##S12YT@>0L3ZJkRy6X95)vH
z)Fj1KwhmX9({(j2CZGWj5PJs%+-?le$ii&#5vZ_WH%a|4jsX)B`7iE4dMU@89`GW^
zgxR4kDGH37C8@ys&z8@nE_&`8^3<0OLGEtU+X^b?1ez_2PTjfOQx4L=j5s_sfWP9@
zxX8U#v1nlMKryN+>s|CX>#mJ+MI#fl^6$_x&S{{W3*Jr-YdX&$I}z%;AJ)yq&DAsV
zWe?)Zgv8N}f(ry>a^+|<(ePKm&3r0$<KKXf+JAS`@T1R;y1#8UUZ!Nc;XPfCp$>D2
z$M;9rSj+^1R8db=4OiC|kT@p#02;x=e|Yq)%^*HNMXG|us3ZLiAcyxDSbl;CdlORl
z#l>y;q8NpCJF;r#3s8qi4!2*4v)LWj@w=q=oQb9t+!O0AX%xVXDG*^1yO?yz$;(Ee
zGuz8kH}0M;rKKH~l@0wTlRu4W+7%HdR<^cKFQp)-VhVno#7sxH`ZRL~P`5E`box7~
zE2{qR2#QTD{hc45Ue3EA^=<MM$P(2Sw}(ZQ4l1rbS<zPi_iw;I)7eGfckKCoJgJ5L
zFDs+}yQ%#r3F+LtjBc55Vf>dcCYx~o=qI<Gn>3qAn#Q|o;(!inQ_#_Y89t{!0<|0N
zB6~0?V$FlEq(xeLd_4g`fFPMND~;t!Y+(9O-n^BSg;Gp`QGE3@WQ3AmN%+$D8O?<o
zh%MV24!zhOeRkGouVmt$n3hOR^xYhQ!7=?#YmZ|f1bUA4(NfeF*jn7~;n-43#QhC$
zv7q?y=*AW?zWJ&Q3I4XZh_9|_Ix>1v0xK>%P9|2RrC%4U-cbO2_>DuVpAAj%1?$b9
zr|HyLh~M|u!{)zM8J+SqQX&k9A>Qs5-h_p{LBRP5z&wIz5EdU%p##?^=C31~<RdKk
zZ3RKzfCUY<8i{|`lDSKF3E}7QTW78<spYg(hh(MBCU)8H6q-{8?z?)-NBwuR0ODXa
zX$+p-h>TOm;z1xsNv*av^#B0LNoZacMnsgFBqw*1W+jSVR7nAK&2g$!DnZruNSI_f
z)BSmZbYt+#W@M<PZVhE&MQndnype>}VN7>@SeVV;U;RB|rq$a>z2>p-+duHvdw<4J
zD|Ke*1^CNHpfGj}Z4zse$gSNa+5-3KeM}c{GgP5-hXeRS<@&9-*>j(|W75=IPMb1z
ztga=hyUfK?8@Tt(p4h9^L#mha<Zj0X=C}kgzMv*&q^U2jj&1^OXD3rOFdskWZ=mO7
z>nT2IFvn(fI)2a6$9~l!(iSW6)M<CK4xOodx!q}@B@WrD`mr3W2HxOwynC*-!D4+h
zWB%V>9Z+$w9<S~<Z^p|thdBnC2hP_95V?jCyb3V%ajj&5^SFSF#v|gh^&u^C&5y|l
z?yAS-#QDruRDq%tVj6-!4Ja`nAHc!I#WFd=&+}Pm7#rJynp1HELZg1XUGi3mpv(#m
zW)=vvd2jAlmq!a}*@^v;US#67n`5=Hv9UYH1!U{8<^g?T!4#?ny274ThN2<D;>HDq
zUA+?BGIDrtRt==I!-($M*8-poZPT7ZzetBR=f#vI<xouJp#s|jSx$N3t2#IRq|2~Y
zrgeHuzt8ZMg#+Fs825<{c-kz_5cY$RMH`R@2|_E4mjOlWM14uRee~-A$vnRMw3z7S
zfJgH%+7QAF!}m~e9i=~EGX4zbaA}mUvJLP$m_eLGw9?|#MiNqzn?obx@Z+jXI=>Ib
zhUtNnur6fHnNt1C8Ipk7mGe9rWOvPERp8h+5~1Xx*52-{h_Y5N*{3h*ltN{|0c4mG
zaNBRDCEljgl8x>lL_KZL(riBP7;*Osuw@n9u}NyA2{-Z(_txw~*cykMxX|eeO|=4s
zuCF3zw*!$j*Swv782|QneOM~se4xs#2v7E|0ND(xZ3l&_Ip=-z*(A}yTn2uevOxo*
z!%Q2_NLr=&+L>%XcArCqkVuk5*DSWPlO`UXhH$*PHxx^2m%HH?K16Oe`C$pL@3<27
zc|2T?)ppwQ^*I<E2M3~l40!y^T-~C8ibtFyaE6F-?ne|QnA6<ef&bLfltbjoJ%pcx
zahk(f*Mn_ncpzM6>@hX?*4)d9K=?26gS#gMAF_A}+(&t@yH2%^!tfd?NkK}Lknigb
zJf6kc!|B2vqr5)Vp@bhl8DN#U*Ht{vTWuZ9m@VT*j!(mdbcdgrm{^!<Hnq5<q(r1y
zU&=ielx~EaCW0p4A+8OB5|ZY`$hU$N!=J>Th}%c&MWO<8m6R~{nUb)U+hKd#_t4^B
zIsQo^5KR5c!)UT>A&$m4BDsc#+7_CQLQLTCe&Ss3M?^&sGPpQcA)qL#GYwb<IESK%
zB#3s4;yozCmapKNF=E-GB`{9QN5elqxi+yhb+QARc^lv5%T?wSk)1^sd`rRdxSt8m
zdcSU$t&<ul76QL#vVYqjb7N3L@`XJ9OhfsXMLP5D3{$tWs({k^1|%e?x=SI=MQIDb
zkOc2`5)>3BmJi1j1L|AYmo-z-(V>G8d{~p<1xtj7Km=_a>1V5^?i3vz;g4r?`TbU=
zR(xZ%fjqW<bK?&z{W?6Y&D}hLRCCjM#*%2Lyssh-+7K+58*fzo8xRIXk=`OY3TjeD
z<1gL&9*22UqhtZ%R7R^q3p4Z6{3H}?Ok``Tf5w3asb-cWz!db5ijnrpO8|uCIUNoU
z1O=vY&M&*k|M<IqeHY$OMmrk&xu1`hfool<aVpM*8jiY2_XJ09$oD&5i|w<<@`G?K
zdhQ480tFTk>Dn?*7EnrjVOzW{z~a!)k`iw$COaCphl)eEe-RNxz91t3?O&{MR4(x<
z3W`l_w9*e?<#F_XNU6uyX0|WfW-D)S+izQk*LGT)_x1nI(7a-gYUQD%oBKoLe-a%7
zmdxALyj8%6u#;!FEr60j#rdnqK}-}#=u3rEg)vWOPxEo%u5hr+6FuEnQpr4hP<1u4
zy#NV8ZV>-rs%!QllS5{EOq48WI?{6Wo|xC`@h|{*=#bfg{|K4&Zpanxq8XS~f(Bs#
z+d7a!a@?rC+KLilBwx{?%rS3y7B7MVcZUk$48<noRwZz8b8~ZPf|sK$9g+@z>Eb8z
zcBaHTj`2iv-N#fAFpI5ZS`R$mO=y?5Qp{G#$}v{`E8u*#;TS|!$l5FboW^3SWav$>
zvE&1wFA+C+oJe7POTs(AXo*;;fB-Z2geb+CF~*uP$WElSq8QkR2NF?vwmZA-krAng
z2&lKWA4TI5A)RGK$E}{vaA7$6@{>JTB2N&P%~<mjX$quzxJ8jWfcx{$LgF8Vg57=x
zZro^eGYH<Cswtw82OW3Vp{NSY7ma=N&}#tnB1s<ZG1?Y*8L=PR;7#L;S<MtLIjN@?
zp3_&V`-1s<;V};19#04evWNW4`11Sd4B!hFSPp_uy9>#VgzOh^=-{P`aJ6{Wj7F{I
z(_%O+Ly%p(8%Zqku&h6ZK=K%F1>L+{T|GU?(Qfw8aj^1}a?+J^Vm5JXC*&7n-r4}S
zEzQ9R6*(zAZDCO?(=X_0YV++NS~No{#`=6;IPny3!{tweqVtYN1=o|SHv<2&@$cK|
zBMUG`cD8e7kiBi4H=`T=bOu}GZ59TNO2gj(0zHE`^K#nyd90WuEi|8^hM*-PUi`Yq
z6d4T9892GaPBIeBF}_9(YT}|w0Js+k+utW=!=JVK^y=V_{{z{;vjTP)J3ZBOtrUXT
z%>+?rtH9JfLVxl-xK#ex6s92KqGimjTQSWV3i~3xg-P}yeEaz)SksjxaCw-36$++j
zf8n0$niWzV%ZWGe;;WQ;7FF{8b8?jB7f%l|L+7Aj+#&QI!N*@<nBHRmGNgXTHgaw%
zDVh0%(9+$mh({=3xzQ727?ml8gR%@KMu{lU$SI-UVL=OgRXQzzkRn+k@BZ*zTuXJB
zkHm3s3sKss_ctJcXEg{1ho~?^TZ`TIt;e>V5|@=Ew5YbeSSl@XA>|M!$j2{nA0tg=
ziKVIE&JV$Q-;x+NGcOwu3Bg<pol@~Kq!SHI0Z9?&vo8Iu6w97?h?c&}*8UG6IuhsH
z9v*b~RVL@$_MlK>rN?2E){XliL*`@i1UZW8dH0&JDGRP_F2Oi?Hj*mIg+JM(t#n?k
zRQjGGKIC**a*uIbJnufJaG9&64VsZ?;Q-cI1;uN7V`hGMAmbE}0-8}>Rwsr>5DZeo
zQB)^ceyG_NRYM@5U&*lLaXPV>Yt6^aXj{*$Bs)7n59EK<Wa}hn;vt%~AKGb4spXX$
zg%&oJ*_tRt;%?ahYbs}5wDY0`&&S<#LxOVkA=M}HxLDyqgt+np!XpXh2jUlkK+dy;
zqL`D4l)K7@;R&Eg!Dq1$((yj?w=*zmbupdFp9$pe{<b~3?Ec^(#&;YK)VnB#)rDCA
zTZ;+C7)suS^Loc3yM_m)X^6IlSsOoA@W$=-;o^ea<U^TYgU-j`(X_On1&~upLaM2V
zM>Eub4KB)2&FJ+W?REiwwU?9r^3DcJZ%<tt(0<ZH0swZBmRA2W)XJhd{*?^RCd7lh
zTq2?4q)S?o@-eVea<d4^t#8f<!SPK&XM~~QBpVRqwcQ0nVGd^HK#!9FQGG=1xM4P3
zP~l%>H3-?iui+cX*lC7DKWV(^5$QQi=6bCZ_9Ol)0PwguPQNlCKDTOMnt3*~e-VXI
zgb6?g&_~2(0&Fm{gz#;~VjuQ)sU6oPK@2M;7WOn~`3SV)tk>3cbhENSobpq6sI2{q
zyy+T6guYB9@lcYJba_lC>Ll%ddgn{<l*ZR$&E^VuntA%FD*XVqd<>`MQFXCVODwqO
z6e7mwNyRlaSC>}Sse#l_e%7tkb-1dJAw)QUHw{8=n1sYIWx0h-4;J4fii&>TieUtB
z(aGQNkK7IBZLCeO@?cshBN+yQZAN#Hs7@RFIZ3sDasAs7@%<`N(lGZ|VyN;c!(uqn
zhgE^&NV16lji`4YBSeZdXIzDht0B+`I9kV=+g@b0K@q@4qjJXP5%YKimOnpB)$Rc!
z$~i*bd-t5V-q*tZ=WBZnzxg|y41Trz_H@$}#Z%?r3z3b^3E|k~8MDsJkLSB(v5yS-
zH$GN%E*6WLoRyApeq~cpN-^_C{%-||QsUB?E2r`oo=uL-OfDe$1vNg({3KIljstiu
ziW#rBy#kqE$!zrdOXlcihbY`y%)tm4oRrcwYHFMsLkTAjvj&b2y~e{L&Scz2c61cc
z7(0qXj}JuyZ{a57+Tc=n_7_ngu{D8yy~q%QuHuT6wlE%DwEnzI==kk9HsFzVaD_aK
zW?S^a>JLzf0Yp--`tFKoYg?aueZK#5)ZzF|B}_WDj!}fs^d;#{Kn-1TtPZ#locc3N
zTI*zTc8HEwD69U;D}!pa&t%Dz_A)+B&fV>28o|)mk0u`AXj=dunZyyTEkkRY)nPSz
zrqAqmCiwBNxeeXnHRVW9NQiyvP(eG*eXd`2I)HFUMOcuesmAX4T(VpgPb9nL>t^wd
zWG24F!lbFOtX3&S0)}&{{0itDwV(=n<ta`_o4{E}FWVC74-<ntW(xAHb8<3neK<`P
zhyL|^v)O8YQV?JmnNYy99Z>9kwe_||<IS!IT^O4$n=q`R*zGw)T2o{>cngy-9fe5t
z2BJwDSma$I6;^}%qI)gO^I$qJ;zkQf1H-zsngNAXN{I#Y^H9OYs0~nkv1PdL3XsnE
zBmB$t$KSVbM$T8+;sP%yI<zfvC<?I;bX3^n6k}f8Gjs=|ErV>LW4oHT_kXN;A7pB9
z5|VPv2U*L|rxk1OWZP*X+VF(Fq`;xoXq?nKp6p1Ks=**k-T`i!hR$lY!{orjKPNDs
z9{1IiCw*^kLVowVk>7xV>-<>ild!ew#S%G^CiawL(66uZr4bwg{}j5nl?KrHpt9Bl
zO|%26^S_gmBIt5kKPj40m4h_SG_@8bP8q_!!~c6!-dV|IX;risvpIg5zbDHh|20j#
zrCv?A(c`Twrr;c0oj1@o3Ue2MLKjB2MUiZucC7U#0OqF-69DDYma<=5IXXo+4<zwz
znihQ?#bA3b|C-tsB3d3@kl_`zIE0Yog!y=w`kb(ESbAPgK3jijD-o`B1+D`T<w{1t
z%*i(3Ut(QO&XKi%NF~Z`Kb!kwyh1FHoV5xDDcB8IEz-?fa*_~<s!{NKe8)XQFr8DH
z_gAx%LTF?u11R=l*`@+YN>A?h>gDxnT1q|)>Y`RNKsFo4cw;c4dSap(;rY0(rHskQ
zx0W>^*9BS+XmRd_z<nR&f4B5v?Y%T+ZF-k{fQQ(*sTy9SDCS!`2?3&@uyVD48<P5m
zp>qCj?fq&cm^&vCa7ZoL(kwlFg;jnCT~Qxq2mi290P56&hLI|+9AAHe42;`%&XBo8
zS5aLC@2z(bVqf*eRD30IG7GOws&LO-d$FUOtsDwotn6+4yJfU;$QT`g@n4Qw$JdSL
ziV259KnQg^#kry=rp@eO0U|Ubx6%Bn30j}m^3${|VfVsY#!p~eHm}!f?=g8MKX;YE
z*|^IR@V6Phqp;_S`L3-F%{EQE&8(^TK%!+yroM3OLnlxc6PzAfxNK5Nfc_xs1zy}x
zk&{Kam;j?Gy1pVq$+&N-7+$T~eu{?+<?#VQp0-Z(E&s$x@I4CRY=G*+=wC0V`}u?U
z@8Wo<taE00e=-Q4NX=!%f)&1M4)TrF0f~5C0DXD#)x`b65(NSDuP9&_*V;!;)h;YE
zKRAU3`k_%uP2a!*osb|sD_L7r*l{L>Q~w{KV5jry3Sw<0bI#vp*V*}pjvsHz0*D_a
zhfBl5Hk&MQKSk*Btey)L(8g$x0$q7I4bk!8$)5C<2w<Oz=3Z;b=XSWaq@e{PK%-zm
z066<Kz$9ApI;Q$D!)DV7a1u4U7fem*e4V$k%vwsvJ_V_eQ5N1c`q45}TI{JN_X8{I
zPsRyFgfICKCPIqEgi(?We#X$B!9Nq6G(!Bn$aiqLus>d)q=(8ajx#lul^ucf^rJV&
zT<$u<F8}dC$fCJGoMel4Dx!NN+e&OGKnvWx1Tm}vUWFT54jBmM3TFJ3&uB?-N($CI
zGrI$WV*Fz}r$+FSq`-;Z2FWmj-c(IMz(OQbU}gEwa*ioO@s#LuulaLF!%84&9Tr7i
z?f&@q&QU&RA_e?X;c4lR5`stiyJcuJ5Vb6+I)q2~_|$BYM1X5lG*g>4*~2#+0Ism-
z!3dOQmF%Z1wA{x4p*mDbwmG0GwZrh6*i(i4K*^G1k(<}SUG5iJqKMtW8jp_+ysiZ<
zm9mGK&zTbc#vUW<^h}3zp!f=w#P!n?ViXGp2Zu;E&ye)koJx~1N)T&g&%^lnnvZ1g
z-NXdYEtdyAgh}zm7{t&>xV{A7swNrvgJ7$0-~6u<o|_|hEi`$LRP?LYh^PU<r6vOt
z+azHMr5wM8LM=87Gdh;eE3D?ZjePHsXD#fojWO@j9GLrqi;<_NtGzg<R>~15Z?@HQ
zcZWw^jJBap_+Vcpek%0QL%=1`KHq?<$jXqeaZ9=lq^0z!Hlglj?J)zu>5cqs0KLuS
zSx4`Z2Bjz5XvAA@b$ezW;W%Z@Te;U_Y5J#3Q<dB0|7QlAntA*B>=V691;rN-c+}tB
z#i?tm^y9oOd$KWxCTb3(2rD?cimp649x?Iu&uV55>`{7ed8;!-8yTH3&IpQUVRI?K
zK<Oy#9B<6`&tZh@)0jM<q@||&Cs0%{Qe6f%(H@AdyjhI8jk{3Q2JwdlU7YNn#5@)0
z6nrEHrMEK_hb$zl4xZ{x>W=B5;pOq=>8^=$Vx8t#IjKwK8TJS>OjzeWnH+{C`bkn-
z%LwGfPj8}&Uay<8Twz&{D<I30`P!J%-QO-Ko)3!pt0pQ4=FtPt!aA>-F6l(JyFEPk
z7dK*<s~2m9Ks=N|6jJ>TK7lcXEp|K`M49U@h#UB$R8vpK%rrJWHp5JZZ)~^<^2}cb
zCMO(SS%_KI23Fv~lrS{lXyp(jlMEtT3XK4Tg>wPwOcE3`BqbdY2?dQd{_oMn+uhUE
z3oI>2sJ{GT@)p4OlBNNibxbuBG&|sjYzX>)VcB+&#n-y~@#V<x|5m_s!GEeHBN!YV
zS0fI7fet$>59t5hjLd_iLWnR~m(%S3x&U0M$Fuo@_8VpAEgw^5S#wn{X>}hd6B`k_
z^4vjBD&Nu6^U-7+5=>)29-G$xWeH|Cx5Ifl0n>_2ixh$;EQ8=s8!G~wysSb@RCHtv
z96U7re^(>NgpX92Um080ilB$Z&`?*~o}=mWIGnG(N-LR-`c@zA1(RR?M*z3+aC$jI
z8ULC*^8>Aq!1V}x&n|LF6mouE=v<yzoZw|%`axs=m3Wk%?!R*ORhZTzT8M!H6Fh)X
zmBV}lVL#c~sbGbptGBy^LYQBWSFo}vP3C>+?5g$-x38O)x?dya%uMNhh6EEdiT=UG
z)5|327bE+0nptQLmzF6?lE#A6e)*nhF{x-UXke1NH}`%tbWl_>Q+?{4JyD(MpD&zg
zDZa-bBrr6gV%7{mz}Ia81~>ojfO){-mWwoDb8q_tN+deOCly9zT+RFYUDk<0gGv+~
z$Yq%*^5jHXijb)q9?y_Qd_LH|du@%q9nb}MenGvmtWZ_=UrUDu*Vs_lcO!27fuBp1
z7#TdT)V|^fN_UJT_<Wut3|&-IhR_aOK7!HVLopsPI|BMJ!bndPv+BX-1`vX>tcBzR
ztP2`F>hH{hX77`=1r;>>JZ8Q)h}J%dV!f2<lb&DW<p==Y(lC~Nk1ej=$hxvOO$v3j
z2dH^f{mvL`^v5g~nF8ByIR&3llq72JGM9dQ=%t2jGUz$XDM(EX&ByN{OXKuFVN*kW
zR~mGbXS_CJcfW^&i<O@o1!!w3X0a4^=X4@dfrVN*ZSGrMcoXq>;&LPs5$JEG5SvlX
z?wct=(^i(5gU~IN5o7T@6rE=hAfZhYwcBcHXi#p;T@pl>%z&Cch;$HivX5wFp-_jD
z`Klqfl2j31A_X<ZC%mj2@dM#j-GTq+TX&FY?*7*!_HqpQqzfQLP$Z4xeA5gzQFyZ$
zwYQO}rcF3Msj4Kb5V1qEvJs$&#>yb>H<aG|K}C_Ntfo5)EJO(KF|kq#(svYfI(&rX
zyhV%vk+bU$QFV<vh?Lh|7eHY`<d8BmOuK4X)DM3!4w$f1jUF`>)zyJu>js>2*cADi
z{@O+8ayxVM1wc|5#Bp#yj?i6Ms@jMaS6puju{c^3hG0xQz7I)_9+yy&Qm@^=Gk9(1
z?HgZFimye#GCS`Zu4+5$TBfM@B!DK3c$MvWlF=SkS;ux(pjN2Q2NONwwS>ey<e4f@
z{Isfeh4+uc!<nIx^*`J!9CS!X&%pn9wA1V-jCw8pM}Turd@kqhdEdv&BmP9=fR~$r
zJ5TnQ!7Ghj9cg(?wPP3)ZGxrkVF~9N5<8?ja+-$3prlR&SI4KgY1)^}0wjfV4-e0$
zAlc&ja@nnn2@`2c+l$kUMCyek2r%6Qvy<?ycB%#ItQM2)FDDoMA7E<Ig**Rx{oh^&
z@^huRrU2DS@J(7rH}+?oL!jN7DfdzMN4w{WGgoirc5Jg6SPY>waAOE;Iiu;=8Io;H
z?)SIw331BfwADF0j$T0bSbT1W)88(?o393QYo=|V+qQo|e$`X`aPqTVlRWu&1`bq6
zj_*0LYzk9v$1NBNNTG@N^<EswC*H(kd{qM|9RM$e<yuqB$dV|qc&Gp1tK@+orf8tu
zpG*dIhmxb-f1^YZn%U@==gmB@xfeaucM)(0VE~C-pDRc@Jgb)dFL#7p9JOEO5C~W2
z28NGeqHevLxinI-EQoiwAJSOM_!c0v%Ot9Tgh!$Ac!$I*oy}m${@GY3dUA-IER6pV
z1~72_ZnAT;e+RyqpFO<@c|2ZDo6)y3owlxg@17R{$jS~`x_DPrqA_J-W3*z~OoOwE
zOL(O;=?s?^(#<TRaVFQjAahp0DQzf-$q>FgT)(HxR}_zb|GgS8F|MUK8@G3iJ+rlk
zWSbOC434Kww;_d9pt`OJl!ys}ZCD8cfZf)235u_$2cp}3o+c|R%ap_}vKFuxOaVB8
zC>2ntoggdO_<Qf4bE>D<&>;g$BAMZPv;vtHoX|)db`lYUcaIo?3Ag81>6(k|t?p0o
z@)NZVd9D77z37!n`9khL-?v_YYj?-*<KyXF_}kA17~N~pv35jbDV8A+P_Y2O-8Lms
zfie`uLyvahRMnVvBV>S_V5u*$dOyM{$J~NX+S-av!3L#~uhB5Co$>yNr1~JE#XwVF
ziO`4PNSk&DIJm#CYFzW(AIXPKEbP4uP^K#IS^V3%&}Anj8IoBoaMBpE2*AB8k%RKu
z0a8EDRC|%eJ}Q~>oiL#Lvho0Q$wnL^p-bl6XU+<4P5BlMNLip)LatA5PZ!)yTb;hC
zJ@3)0Kz@g!JKg@a2PziPGJb&KxeOu|?0O(Wo$Yt`;312$e!v>iVyG~Xvx8I%?ry^d
zOoNEyS?rJA!^nr?WtpqEx&Ha!^qbupnW|0%wlw`bZdTgm-?iG(h5^1flP%_={l=IA
zC}uF7I2fr|=%;W!9zs^ulB!*S(lKKFmc>@bHhZLFky!0Mu(5?c1m>^h8X_mObiX?r
zEcbHqBWI0!JdV!Wfq(CIuiy7TX+h<zQBQYIUuiYTplYmzC=Kc|*jQBn>|6h)X-PMU
zo$!9xfPATINDE%YGhk<L9K#_#wBRu`DKd!#rcsW@4irHEn9O5Z+g=foO!3ATQZZDA
z6XZzHMgUdaB4?wk`00FFrg0S{@_@QF-_dA%zSCY;%`-AB6M+cZ^9`rNN$7?mfoWxX
zbB>OIyF0_OsG5OB5$`8SS(eo%Y0zl80g<`}Wuk}TFk~%&1lZSI3|V%xGhI&0j$LuM
z;Pb6`A1ntB*tWXGN3YCtk5^F{<MD)}3P5i2w=e>lIY*~+(>S7Ut$QEtoJLmHCsY)r
zve0*tB`o`=#XTYDfV)g9E~IOc64d=Go9aqB#;`Dvi1QpZyHtrMQ0}}~X9p+G$&_j6
znZJM!2-Zp=!1q(}zgJ)@nyHY#-ElrGga#Yj_Gj+~hB!e43JHVx&Eavh2bgG2!cM3?
zE9|w-v=06wOBA>f1{U9e3@I29Y4n$??~or(u?Ji&Rr?fBT?rhWj4I~~0?y8XUbO42
z?*9b3@Fz3ZK>S+9u4BX0q%wWgqi4%N=MOpG@gV?OsBkvftCmy|*&t_3+T_u(C|n+1
zw-naT=Y$!j|CCz6Ml)O_>!b;G8^zrfY5fc?El*Q~0->p~=hj{h{Qjs!FTlr3bl+Gw
zz~}w^by1p|F)IuEx{Ud05{82(Z_I^4JIm$`u7Rk`Ij<TMg%X3(bU*l$CMu^uvK9+<
z6JUco*bA<T<f@5&(jWG5K0e*Lkx{?XX8v~ocyDdc>2~lOP}0`r@_aZi{=Pa@JsIu7
znQ|iKq^$z<ap@K;4Eh3J{B$DQ9+usxMW1y!z_vzLaR%I0`5`SNyD{*2Q-fKYT^+?o
zbR~3{LYQ}kVbxeoGn;dP8Lz_C`}HObuu4<T?yx&)rcW#&o7futufF?q_!Td2ok(oJ
zUCG72B&vO{uSFz|Zi*t>9|9&wBO#xenG8nBFAg6O>s<sLs@G-as6OO}BI-ZlYOZ>*
zo~e)+%2klP!dK~dcfWkW?fiEWQcM~+I_L?w0ut>Fe|0n%vvg667Z%e`M|T1?z)U2>
z`sA}#QzERQChT$6(XiQhw?+pv)U~4GZPP$w-o|D^(-u$m_o?iZcAX?CMkT=x`^OXL
z6r%O9&7zoSr}OqPl2;|bj6rIRpqtLEW%|=(K7M5;-WUi1?&o^b>osvj5genNHFf}o
z`|PA><e@!9Z0f~6YAsEkt*(G)qy(jeTUQqGRqT%AmMHdF`HhvhAf!Y5TLvdwLn#q7
z)uo?6FR1@;4JYsc|LeEAmL0o#6C&_F|GBy9-NsG&+X@lZ>j&qbzutk-zb(TtJs1o_
z_gm;&%svQI0um&(_aY<_si?6r&48VUhT!Jx@~-OUVj-rV6?}X_z3KpM^7^xg)0Bm1
zXm?(C>%J;rJPSrIFw|!rTpA1Qdw-8{fAwaotFD}kvtLX8GE?=~MMG5OO;k?Ajk35J
z#PmK<W`YV&fY?vm@(`CLBdnh^0L^J3KIWsvCu(I?BUpN8j#t25%zjjkOiR>ZtNG@(
z8h-WWd8^mwKRiRZJD^hYXy|nlTot2-dDr%8c0B8molp=s4C&tyX6pFy!_z4)G5#k9
zH{aL1qcpf{atIsRY4L9TL8TTp*v+6AbQqeZ={<j%n{}HbL`P?fP|7>bU!D&0&4KI9
zDEe0-T`6I3!7y>ZpG-M=IfwT3q<n9=P%Yx9>?C&ZDIxj6XaLc3nF;!VbqtUV4)qCk
zG64q8^3I4o%6ORQ1>E6%PPvENHzJLmm<l3S$yA`V`+sef^cg^rm+xvY@Rohs&+S4$
zCwitpEgN3SY{M-i3O76s{f*Z3Y*a>Md<fIakY2C0mCPUx!>8Gw1p0?@KRoymk;w}h
z9@ceo_5mOlt6k${_?;i8R%NG>@6AS+TOu><jIPRV=r%!yw$9Pk`1iwD_%9A=EZ=bB
z3!#-R^FJXGB|gsbp2_nvHE-HffshaZsF9`~q2a9n4<M^Vp5Lpw7(wzC@}15AK>ni1
z@To?xukD=@cWN1n#h~BeC`>Y2uzfDn<oTL6jtS6p6}{%F>L8Le73h^Dpb(nl?lOCe
z`K;BWSO2WjfXo6&0>p91AIQ2HeflUe6%E@w@7nH{L?Rk6_~!#fcumX@S!uAFoaY=+
zjA#+nD}&bRV#Z+{JYI#=Yu|u?Xzdw-omR`I%kCZ=U_@fiSSn?B9!@ovIME|W8blHe
z7BI2y8&WQPkaxsx|8U^%;91tyJVZHeOJX$iU3qoH@`2(ecd_<Li6;G_x9<u>7;3%W
zdu~1x3i=%>&BkAg^Y_#D0PnJp(`a8$>%K+xvzkbuC}Rg8?<z~|unJha+FL~!X#Ru$
zNcs^BWQ!8xdv!-yGFDg}LqS=`R%ogLej<}dkvEo9bC)R#jHeTfM-s5@yVK*JFl@Hn
z9Hv804gGsgf}^RmT0WLikN#Flz8{FOi~1Ns;g)yg4Q3?(;u$1%fVB@G07Z^^MyIpF
z@%%-><r)0?7tDSB?mmSh^;^1jMTC|`J2e9oR*lOT*G#@JKvZAf?|m?we#IH^H<IOl
zJL66MM(&2Ha#g>|e<BcPd+d%49wH(jzM=L}F;&>hsF^Y@dLu!9ai7{kd;Lx->8)4R
zu<|vR)M{%4Ue(>|AQ_3O()wD)lL{WtP7wFiWrXX?m<2wy`4Z;xE6+zqPW4@P1)LoY
z7~*SB)s08HajT=^y>6dB0zxr$QDkh4#I>Rbnbm&}rrc92Q1;&5>!Qo64Q|zgY(!iV
z2Yo9cWxV?mFH*B!(UgF4nrX-x`F<W&PqSN1=K~GsM-f*Rp5E>6r@i&|8B_KJN$3~$
zwbY}Dg2|>rli((tk_&qor^sg`po|v&<S7qywu3~I$dymtd$!VnfPWtg=Tu0&$<@Oq
z681Hu<RZ`cDJ?PLdduq0)S$eC^ocAy`<q%$S9}3aq!a~*!+nAN7yHo-gctl)3S$W#
z?4QOIL<6;-+c;L;bZmn%F-Dtng9(pT_J(avtKH!y$5`j0KRiOR@w>}PrRRH=nS1FO
zA%j-$OsGIhT3;y-0QUGfR#T3~-9LaYx=5vhUv3w2|C$>NfxrVpmU0ftwRx3bz>CWU
zL;we0Y&ypyhfNMBlwS2k?s^yS+*c;`7E;MV9{}CqpXB*SC~Am%>Ul@HD^gA~A}+5W
zhipnbOQzOfe}$Q>CDy|Os{P~2lsVgn?4glpyT_%+{k_`~uvptdW}ui3DgEru8p&lq
zk`$X*(nix)71vS`+?5{-L?aG_**<7H2{t({ha8^|rolK{<D(N|;-VT+Xo^k{KQc;;
z8~>olki5UWzk`Q?gN=<EjN9VO_4)u@26!82jjnjcHJ6lxc-SuZCGYs%{QkECDWM`F
zA{hgDX{dn#P(_tcK@qJ6qXR36`fYl{N!1nDdXffR%4)O3{O{k+=j`ujh{BUPc6PrT
zUBCLb;{^5q>{(CmpG39w8^&;(`RNfj_-B5|*g1!ijY2w0O4hj?5%b&-dt^WrDL=5I
z?`CoRr;i{7Tus6Ju}apVhwRimYeSS<J1x->t;vf7AbRwckk9{vFNep*Os)A?Vlq26
z6vn4*wUqMv`0h&n_thv;`W4Mo)ZI@NZ1MhY5>f;{4brqjOA3<)oY(+!+Ond~wk=Nn
ziNv4W8W|2?(3K$C_Trw8Wt9n(U``!E=e%>w{FJeWS4W_!h|fP)8orWS+3&R+*MEPT
zC<ds%1I~87TDL1V^b?4IHF>=pG_yv=0VUho1RT~tI&PmqymcJ=0fVhHSA=G{-6d1U
z_C)5e)4o6iQPA(B%)8We7;ePQ)A0JxF|ghJ>-VSg@#tAsg*>wfsRU5hhs*|MrbfrB
zUI`H7vCs>(+31Lx-<$Bq(tf&C(aqm1kQjU{0CCJ%yt{|H5Zavony&DAxA1UWH@r*!
z74_F5<0FH^)ptPf`I{VCkGklI1;R7TXQy3dFZ&o7ZRm%shfYyP?eFyH0xE*`@mM(S
z8`?&({*HWs#Z|xK$MNT%8B{b=C3#O}YtnNA9P!!;s*^)A4+p3=yDRj;G<T!iJ|3U|
z!59Usz=&T4Hk6?fP|yZ!Ukbi|14cJ_Hx7d34KRNwW<%zB5mwn>ba_OAeJx)|b3=D;
zCJ=3@w+S##EE*`LIWCP6R|Jbe{iF(p;-ylG887}^-x#N7VPbSfWD@$NGOyZV=^3ac
z)>PXS!SNRZiaw&C5&u4a9D0(puE-G<uqkC$)`#ErKCBr>@M&$yu-Vg5Qgdu$Za;cf
zaLsPrtyC!F+0tq!kMy(2>Z1VmGKB41%B#SXrHzq2Gyv=m=(vtHp;92xPvsoO)P~`u
z6;?M4rDvTnG>=CaI%wQa!kdF?4Rf%Lu$*f9vCDA!j?0ER=bPXD(0d6%(1*l3fKC@E
z!L@Du%nGR=S=9^ufI{WR)q$e1<ZAVyyzubcX1bZzhM#XW0mJy(R>ROdr{nMnxS|}&
zs@b2cwiZ#)&_C9T<0chgTwCW-&@CV@PH($jb0_F(?GuTdoh~R~bplS_64Nm#-g71p
zo1=<lIAu)>xZc<$yQmfvh^&9$0ltj!AH}nn%1y74N|Yu2YVi1GjJn|a!S9)kg*tZt
z-LEo|u<2`G01%>oveEQdMxdm!Do`EX|KR0i^bI;;HCz*xHPW0mo7e!PU;}#Epxj!b
z7+!-CWWTBTkPjm(tZFI;KHBoNZ3aJhSw#!7lFM!Phze3#y|g{2lS>gt0l9h`8~l9y
z%3z?4wO0B@A(hFONtm+`^gB4|X*<t{m;|AYrp1j=61$|1hIWzP29gp_vjCB0i9`ZK
z&@Yn!zuEGEO65W!^8J~k9WNBN;d=Hz*<y8lbF;V7knPLpcts>VwpJJC=xC{F1-~7p
zkMk`lH@O;&GE&oH>UG;O0mBu_`60}`{Ut$3rTCLLW#w0h5*C`Ob|>h3@Pzk5*E3!j
zR0)Ce?(Qp>RaMSBULQW@=IZM1?(*{T_V)Vv`oklV%?(fsmvP9DY}BW59seNYsa$()
z4{*<QaeR4IjYbfO@^$<phQz2pgo92)FjYoa70ZH*+u+8A9k;<d07QZn`^VvnyNgXu
zx<Bd~^9w>qCM!2h|FT!04@^#uj*iYULviBZ_(%;=?N9i5iUxWe_mt_X77NX+Nae6o
zTD!Q;xm8X{gvOcLinAY3G*=a;T8o&71oidx6&4y>s2u~_ozKL!j*hlYb^?KyI^cxm
zwa3!_Yo4Qa*d)0Y0I_OA6S&OmRN`uP2IiS8k5TO>mo&hLckcC7(!apJsl7__e{aEV
zi)CX~XhVO_G<@MB6AAeT?E^>p+Ot1bOWmqj{FJgH(=!;pmxz&j#^&0H*826NT*jay
zB}b;@@={ZB@^G**u?|m;pt^WxYDG-5QbXfAzw+YdD>=}E0CfWc<RoCzE{p-_^7_==
z%zeAJr@F1`Eg5xEf4^lOs#6J*##dLDSC*Ctq+Czuo-N(QFWZJ9HMEodz=x7T()=+g
zzl6;r1C|ef%urQPBc)v*3enp2Iq`tD+M`j?j=Si9{a*=Hp9(z|Fg~_mFX{R~2u!YW
zexDce@b)in0Pt|(ey89*-CMHkd*-@ZL>X>7EsXWA@Q~D^TxQf$gVh!%YxCFZtGG!z
z1~am@v>(2_3dVv4C4^)c%%-dsN{r+{XGCQrQ<}ySN1MkZLR`I}{~xPq!=ygD#RNfU
zyTrm%zSYg;^_3+R?-vuxqhN5%qoempE(WE?QLI4^z+8+c#DVjmjhN)(05g@Nmq$}a
zYkM<SXviln9t`1OzTo>w;IaPj>=F`0e`xIM#PqKT`y8cb!!CDgm)qwn{SK!cfxqi2
zUG9Hc>dIJ{^}#J&iH?;On*2}t12^c{kIIH6AznN8k|OzbpeLi(HvbFL;Gmk1w6!Mo
zaB%nmD%|J2hT<k-egZ;XHBJ$#aW>v1Dl8fqHjFhl=e7d7{|%<YTPt^Pb1xBW8UJ*?
zDOlL1{N>d!75BuZjcH}-PZJ}_EZ2HU%Q`IiAE~csw^*9n#U^6RLrKl>^}s~u5=Hny
z@Z_L4CzL#pjTkEMc@Q(#3mA3jDgB2A_IV7L#$RnT{<X2GrLPsnOl8R6fg#aJ_35$&
zU<>OfkXK=)g!i>(=9b2X$B|SYk~7^sVqm0ign`#pSClovqelp0<E6F<TlBUWnhOUb
z>IV5LNbBq_*H$==**{08{srT7byfQ*m_Pj#IU$dFIK0_`=_pC5*qPhBprUP(qb?6%
zR<RPob;Q3pT`FHcbKf8Q*k(<PY2u)=R~3z5S{S_x=DhcQ_9@YrLAlKV`3S+pkDaVZ
z%s(aS3>NKkd%Z7j7^z7?lxIg1GcAL68V!7u`d(CCTN~9dPF>@x7AMk&sv3)iPYO*I
z5D-Wk>`QK-;+clI+Y{d36U`<S3uFUuB@N4IxYcx5M39tGkWS4~T3_<gd&_H{Jn45p
z-)#N*-rXzR%zRgBC#c_m_I}76aCLbq86uGeha4cquJfQ(K*mx<!>`R`oRP`KL_NK7
zuq<MtX3Mt}I5H(6d&2!BNui|^v*i!4-i0-mMerZHgbc^`zO1crzBuoLsxk(2^SBiE
zwf6eI*9_CDn&d@LIsp71pmJN+vhi(pFHqi+IyhG}`hTl+BTM=$6d4fmbIHZRfgG<z
ziUnbR2xq}u@rLX>RY>og!d(2Iz`C~obvFw5dim_|F1mTX9SLv9=m!5KifcVxg9$mA
zQ6Y(b*)SV0MH|8%-;WRoPg(&$M?bzXtF8=5SD;;)J2{lI<#*edxp{Z2$6z#p>q7u3
z%+hQ!eJ===)>q0WR}}^{G}c?qR#@WRiSG83GxL2y%dIV#f}7Sf7KcGT_Xp_fPWEpl
z6oiobPN!0}KEMUn$23#c$Bra|DeFEY!O`swkOuul5=2yMQr1~yb=w0}gi@<YxgS6U
z9?W`iEo)uwbX%=2P|Dvw+`GBEtC0!Rw^;!7$>UyvugFV^*p*pukH1^B4w-Wb1?btv
zMwd5a#EEiE=Nl-qUKDeZKWKjwa}q7BhEpDvxoog_qFw6;6jH~``gFqWqiTmdo10lV
z?eT>G#4=hCpP*w_8D#;9$gYuaxU03oL_Q8j<)unH{^~2YR^lf{qo%34OqbhL=3(^}
z{qkaA-Rr>!g)XIKuma~crK9K1Vs=_z?im=G-d<j9g*KtiM&2&QcNLoQ*yZJwvp~q~
zl%7bZy38F|c?iYroFwC2hjPJ03@&D%D(yP1J9?@3z`}x#Jr%&!^&fl|t;;<e5Xur0
z9UUt#<G=qE!HP(Rqz+2s8T2BgoYq-tn!}Q;E~B-*Hp6bAb^mcmS<B_H`MXfgrZ}V!
z8~@PBys8=kjxZIFmqJU(Q>&c;Z7PYRg%gFE0g(*xMwW+CAhL{>rX6Fq5n(j~)!C2=
z8m~Ht0H$CSd<1}U!tyuyiv0_(f}GDMK0d@r=6R-7wKcaY_1AB-f^*WH4b{K770t>#
zfB-5pg6wwAJMD2SkH_gSd}unC5TEGu$jo{Gg%g)IB05b7GZn%Txn6l|Eu}wJ-Sgs~
zM-`4J7c8~H0eHG4KKFN1R#;k`<ZHP7YMt(HyQ3J)a7e(qEiCGPo{^;yp5syQ#Bu8|
zGn@+x6vw?$0|`BfsAB=xT|lFwiJe6eyml#^4+*5-M04;He49JWu&FAnEi7x#MN8{U
z(rWu%LQdu{B!|-GN8K;7HXHLnu%1saahS<28)0a2h3!hba|G2NQ+xZz*n^9}$U$#$
z^PZ@X697qTN%ww$8e$t>F@POkYDXIS%9MnWmXs;NT1v&gkXo(<Y$U#LLUN`&$*Sss
z?YOeCvOHvcqs|53?dKlAS%(#IpF&(|As0yl@zUD-hc!L}iMtR*uXPtv(gvXz&IeAT
z|JFHusqHCVR8CG_ToG~&pDZh4R17CwvDC3o1Ynng9g7AtQH5Kw%_M|4b=35WV7s>i
zT;D`Rq4Hz{wmNjaw3!&xsp>!}p(qQXDko>1r`s3$2lyomi1fwDqPgd8fW5;7R{w2X
z5%@ibgAlB!Zk8-?(aQzKvYJyhyM=BSL;wLR?V+)e$|5#0GA=489?gRJ4hna__(T8A
z0uH`}>0Bxcazo=Whj?x)#)s)i7?k)mMBeicsY?Ph4CQC%M2wZSrS;Xt36X0-@l)8S
z4=E5!R6ZlWu`yr4(BSYQG7^k#13o`L{R4Wo4GA1n3W{p0|C3i-;}h`GuT9HotA^Px
zP(!iPm}3wMile*J>-$k_Lj%*+j8$|1D;2&aiMQy5)3g1Fw|=-;IRr0G^-R2pKV@hl
zb!G|a$t1mt<mJzd6K5Fk^0+vqv(EPB=8le*mbP^fd5ZrRCa^7Ol<aIoOTSH=U-@LY
zH5*Q|O{yi5KU0_|GWS47oPxR&Xz%U)kFwOjGy!8d92^*3t*{0brUoD?Hii($D+cS9
z(@|Sl8p63y-``V^oO-&0O?dX^TFt;d&Xc+25$Y5##RIppIzGl)lfe^Y)AX^3I|gNR
znxzB|XRtDQe75(2!U*x<;&@QL2WV&aE#-(8r9=;i?3&%*Xv$R4ax&u+vl<aj6)`a_
zEo>7>6e_+0e+FWX_5gsg30%U*F*GCdG#{2f$U#`Yz@)J?JrxNBA$maJ#Kj#oePv0F
z5@@r)6Fp;ZF|II<VTZzD__ef1dfQ9L;^+r^N+2vr=)!LP8U#ekdy^(a3nzP5D(KMQ
zs|&~^W4c50*Z9@RzG0wWkNPMj<{0cH&9%U0{r35!&iRgW4hH-+A`CE?^6#sRSx&41
zHU9CY4+5zQX09(>$8}%&gz|L`u5R_WEt|6;$n*t53$6-+40tUS9|sg0!s>G!H-~9*
z{6#>8LM%9qS3Vg%BI4e7gnbuwt3u=<I1qiu%OAg1SvKi<N7W!5Vp%B<Lkrl-zhqEE
zXj>NX)kNX`D+I9mP)uF&OqkHqUeW6fG>xn6T}pxvv>VY!<G@rYFiY6P7Tym=YXmu~
zmT1<1jkr%dng<AdL4R}v7JY+%0s6m@U%o-Vg?GQ9Kl;dpDtZy9r4op^UD@FIu}l9S
zP3OQKSsQKJq+{E5(y?vZb~@?UW<?#NJGO1xwr$&XPJQRz^9!o>uBWQ@yVjgz4CKL=
zk%IAGOuDL@isuXsVPTGIRBg62dOUYUl-VwTOnP7f>_{xl2;z-t;D1$xmORKw9LIQR
z3+4_Bw`{)+7fny=)ll<N*^kyeU>oWZv`%k>e;GyrUk|_rd<);RFGM%rlKa3nkqxs^
z4uuQRlr%8PNUiOU0*;ADggBes!H0o~h0&+KZmHV(+vr&lOGQvo@2{+u!T7|gAdi{y
zRviGtn!|2Wxz7F5WajhV&Wjl-UUdl{wVp1HC5<%9Z4C=uzBKMUI%XSo%76&!vl(^e
zqGDL3a>NlGaBfzl@Wou3@(9`xJD}uFYx{~b-&G!V3lKN5Mn#eNhy6g>;tBDmRvl!0
zL3&@X(C<chW-V@|S6xIdxFM9S*|CA6w*e4#dJ#BR7Scr6Td%^3bx3-`X|vizO13nt
zA{yw-#v1Am-a%>igc+hE!;<%b-#~#jnM1+H%(bE=9d0d+(?W!6qTb8*hZr<^NK`!m
zm$XR?HY^oNvyWzU)cHv4jXTwvCd>h#iL(D4(eqSp#X5QnZD6lQOfo5$nZl2qGz9ob
zZ&oeqAyE2Lw71&a$W#RHv-Tze*ihqs3#K9mdys`&wi!vs+cmCY=!L*yo^wFiKQytn
zE2%6a9z|xR40}_Fco9IdJ~4T<Vj{||HnS8|d0Pwl;`Bis^U|maV0KnLnox*}jSP>0
z2f;1P;}Xn*dQqN!F~kYp%|WqT7zOY|t9cxUEQ)U}#k}%@e%7jTK~P|@BUDyY(9zRP
zOioV7MMcBcJSLZsYH(z6q{>Q3EcT2o3dY?eNz9m2K(?op$T+ap=4Bf6&x{RD&#z9Q
zj(ngqLdZpPL6D=*5~bmsT9$%>3ZYw9!NRf(QBfNZiuPs|9m0*50h=mNLx2Kuim8Zj
zcsRrd=Xy>~|1Pv5l8+MekLoBUXhBW9UzMqGff@K*E32!kQnoXr@swYO<fH^fEN}-{
z?nuyz=nPLw!I31ESI``P_h4%>-$TU)$R|=c0lj;zvt5WGKT-MOS7sHeZLN+`fG5B+
z3u{Z0p1<V9K8PQkQtQ8F`T)dY)}hq7pPp_pM!SykSimF`2e9=KkqZSvU^@}*(yOP8
zy(h`P4=^hbCI{xUfn39gph|3u)mph;IyLMlycl{+v~>RlTpgWWl3e3mAUHzVqYJ^z
zYOb#VZdWa%HGQqs8;ba3KmMqQD%fM?gf>!9GCiOnBI{4QqLza{AhsAtVM6#N6OxkX
zf&?0umCVWm(_k@d{T5`rGfOg<a*zicYMZ${z{aGa5fR|-Tpk$#6K*-O&|0GZd!OJP
zoDKi+bFOWVs;MFpUH0_$G?Dk@8xzHV?SU&8i>UAzyQw&hbu2tYiB<e$pc;TVhEQmt
zs=W^%D?LkH<p#8SzxRXKsuX0`|Cy3=G{Bt09u)jZif%h3U~pBn6w;Gwj)O$z3|!V2
zJ7by`3`jzHaj!FBzYpMmBR;3_vm_?oDqAX4Ml{v3JNU$z^n-QkU(9&tmkMiZ_z9jZ
z0wfcJDy9`9ADx-hDB+is-4u^T=eb4*9`E-H)mk%>{s9`9csT)<y%7-MBvnOcaeQbr
zyTSkDdbt0D%hB=z&*T0AGhhYL1FuD(?zopDy^giB$|$?WB)tK4OORzi358sam)!|I
z3zhl)*hfX{Ey~x`lM+G1Z9Qu&B|4_$U(}`^&^&Rs)1P2?%I_g!(=mv>K}kR-XE0G=
z8|6T(7hvdnj0?8bem<^&nvs^?Qdzi8;)e+DYP@kO->v;oM5<>YdnF&4ijy?dtnDBz
zi%G|Xy0{h&9>i1SKP4rwe>{8~L<ff!mLRZD7lA5cZJ1MCyW!z5Ft@k$zXt~)+H_8+
zLEvJ=#U|$%Nxb{NGSfAV%qGHNir83EumQZb+yH5;fPAXD9;v^mD@wecFQF1t5|AD_
zniA?7V)#BNi)^j|e7|ot3C{qc38Ls2zbU|ZaLOZ8w?rG%*pFrqya&x0MyB+C>>QJa
zwx{Q+L@OiAQDo<arxr$M=EMCS?gx(r;i#!|enfzSlf?N^0L}Ipr!Ln|>7ekHR5nLX
z>kOFR>+eu5t(NA>Vr9*j@!1m#X$SpfgXc{^3&h&t=&o>k`y2a~Ag$2R$n+f)2iAqe
z1i%3A8}PPK2Wr(a(ZG*m_j`cL3=Xm2Z-n{|>9#-)QK%gCS~><W{KjSZQQ-m`lKD%B
zAEoMUlAf-z;+8HcBCnzV-(#CNjIYz@1MmSBB-%&ngD+H-;Jg=Rxu@es;+$*?m3n@G
zE+q!<db1IWOvHHoaLU)Se{?{6{npD7$wa57)aPSF=4|bag|yu>QzE1YbyI@^kypv*
zM@H<o!1|>rK{C+C{ZGJ6qAbp5qhN4G%jN)_khy>2OgMFfo{5d`Cnj9UK&TgQ3Lub!
z#8)NC`1dhJqVf`?yFD41sHiBIuRZ}if~z_Pg*t*wp!fw?mcUm4NtWA?hm{X_dT@-W
zmd<Wf{`>4WH(qDPiWs3gbZuy3OAB^Gc(}AQ2On#Bo|#ctI4N+=rr{P-MJVvq8|9UV
zG*Q<8KjDXtvcHX|jVz5q3`55_0T_&_bJ%D$|3O975YAG<l?|gd963^sr_=7`@%jcu
z^6*Q<N8eW7Ma>lFll^b=VE%p2za`LZb9PrHs>Qjb+1bU##rgU9Wlj#!)a>$NMsjzn
zIby6HhSGEN)Y`HKkHMzCiWd+`QBqDYbycYuM?VF{o<IlsH#nP3J^c&<0>H(kP30wi
z<|Z&602XLjNk@o1J;PBq;9E}g81<cqf)ubs1q6KOnEm1Br<@bkl+T7)W7`-TBE+n0
zu7+9-`ZZB8FZj!(q`0su@TcWS-Hi$-^W5^H1e|A48j#UV!IS7)+gwZhK9J2_@Ti}M
zRY7ab-O|ZKwl=OJs+X3s4`6Aot!rRnW#?cSNJD(-f_x$n#e2?aISKnKB*^HF1x019
z`JbT35WuM`A_#ZHYO3xkXQTK3KV{<Jc7n0QS&5N}FzMi<_ZqLqg`D`LsxB0n%?eur
zn@akLD1?Uw_5`In$oU+X`^C8h5FW3%ypx7MwyO(cy=xf{fcAI|wRsMfX4n6Gn96bB
zs@4iPqR%bjct{PJ*VIz64(uuiVB-$Hth!kCj7XfC4K<fGa#GpCGDHHNj=#?`UP@*_
z7&U<wC3hr6Mn-!oYT|*H6-v2u#P+~c{wwCM|NhJ1si0*=m1_30A~0wjlcFm5h3XR+
zk_FL%w}Js-1$<eJ0qZzC4GCsqw@j=_Rq3T>HnzY0lc@Dd?;=L1E%CIuJ-KzsE`Zzu
zp{N--cTuQ1QtZr_g9O&6D4-9+wXCSc$RI4JX@Jduc>^CYDh3`20THNsdU9c!S09?n
zv&s8ebcN!4&snXla8f9Aj}7AHUR?DytT3qIf(eMxD*A@v-vgLqiy~{X(V;<O2#p+R
zQ*-X_Lf6cO1d*Dc21~@X6~%PrhPM_*GZ!jB|D<U#x2~$LqGk?Ccm2U;my<}icW6{q
zPEIqzjN)^V*f7&>WSNuezjQoc*m5A>VSa31O)}}O4nABDH)<IWEwa{eP%xkkR1Fz#
zDFJ>!J2QVzP*#~m5>!9g=r;Q(TZsf`S2rgYhw`rBRKT<Mu(;1L#&elVmE&r)8gVxh
z-3Nac26UQPDz0SHP<%7irRA0T&EaaGSM4_<OjZ!+*+#LU*BHU@LrCfb#Sk&0U<EJ)
zI0{$ooKCTazu-JL2~|Uwnb^ep?P@UHX9DVNj%T~x^v920{AX@W1+;;qxRM15&qypY
zGmwvm1orfVE?pa7y=Cf`dSN<T{qZ0K#7lr(dL-IMA<IoNOBDuj*M0t{dni8m(YinR
zjMRzuemX;iLob`%g_cU42P*R@ml`}eD{;<RvBa-kQbR3x#gH;9i*vkOScF5hIlw>!
zRmITbBtm0?*&ukUcCVAB^7_CO*~sRSl)gq-q$fEV99C|TER~H&8KwZtD&APy*piue
zJBT`ATW}KL?kIO8%lz@AbW|keM!$GVs$`w4#X3YW9evqW^Dkdv4(cMZ75z+ti;7vq
zJgJR0e;I}hjBnn33gTdhBIg=~-vC_E69z@oEIy2O_CP)6mCm}LG=2()ft9D)7x)$4
zAo8Ctaj>wxrWJj1O8xUovx~qC?UDv1f4DRZA3Hyga|5zZso>(w`*S~w1(Mu_xPN|8
zO<WL+j@Xjk`a0#|P?whDm;z$WB0>=l$dM|vyTZ+AG5YE#cuBc20x(KC1Nbp4NglH6
zBTG}gC5jj}+nR_ut?fkAYMMjSSr1Z+B-W@GcgR!`M3Uk5zW0B9Lox5IQ@a^!64Jel
zOt2+A*{8@w5;)x6+`;wj%KsdQ{{2x%eyyymua4tfH6TtArqF^2KXZNpeD1USnY$QW
z?=_b#WgzZZl-EHgCo4NT2~ZpSV(IxI+$so{(q~KpdNMOIJ|-m{L5sKI6`(zoR9K2~
zRW#ODL6%Z1Tb9Q4I$TstJ`@a<+cvwTrYH|;7DBhrE%y6t{Z*{!fR6ZkW^-^B(P(+O
zi1B5ZDc_0+HT;^F@^jN<;O~ohcR1FMD-$Y3Q(V~*RR4J*@xHxmP{0R=KiWn_ixyS1
z*@0Jz2dnD*$k+z>!{6UyWGm`Ito;KSm}CDiF~Id(l^@h;Gz$D>St!iZ`wSt&P;&lo
zA%IWU2FJjY=rXHHYY$Isv2t;6!(}Mgx*kFle@#q~RoPeoZD?`G1ZUT~reFsQ5Hu)$
z53|&Xr7tTbH#;DP)PMuqO5iNrmCb)@a2Q4!e(>-uD<o)}Sz-ng{~Oy&u9d`YQW-P0
z^LJ0UKk^VrvbMIknX4_osH#4@rz7Z8={`<b`3_$YzeT4$L4`a*gZiyvUi>B|nm${C
zaGyVFJ_RcP^d(v=ZA^#=-O0HAZz3uhrvK3un~0zg@`d4HB0!DZ6@J1dgYQEmZ!ezA
z7e7j!!m~Ka^av%r)c%q%(9MZRlh+wcp2E}85oJcJpfQ9v*4@#~$pw58L)68Nni<@g
z8l&J_ffTY7QgX%hQ}lp&@%NbN?>U7SSRb!x<it2tmH9bVj^?IejDOhI4KPt}i11=H
z2_7~nf)+VvWq{qx^`%)1>vw+>T#F95UO8`_UpmMNjs#?IydpHp3Myh2As~VdfR$P<
z2B{OFX?+Zyv!@t_E7RvK@MRma)K+n2`kf0Rw7q?kNN-MbFn3(c3l)`R=HPGzXE<8x
znYCGEs7rQZP{OCucqwTjOpb&UAs;d>KdqUTSaNycF96H~8gUA74JB+7Dn38#c`TS%
znHm=R%;u1kzC@TaYLKNO`3edRcj~Ak&^g3G$@4aZ$pP)QO}xh<XZqHFlv17YIfanm
zO-5^OkA<gtvJUN_5-1H%<e=4#3yFzHcx7>$w|Fk6{uJ+Le1w`b&sO^k^dA2yP|^t_
zX&8v+0Kloq1a83m{P_?iM=$r#2op|e6w3-`n_P4*<ijiFC-=7{9#i)l`E{1-22s$D
zZ!$TT{@W#*{!*{wLo|O+l&|73aln)ETZHkICNM&RA(zGzPxELho8a`?%GSc_&lbuI
z)4hA+oMsZb(XYJBG7E894FMPmePGlM8h}OX3&5}xMkLMl$Lw)o2z+LbfmqZN8iRqk
zxg{2>BbGflHc4fQNR@>yskaLtY?-9%0sW-ZaF|V->IX}1$iUayk`We{&1Tj_pn%cn
zE$Zwg?joTSB&YryRxSWP5M1V=8o*Xoku)5g;aMR6V`c}r5^FhjqiDvpUmI4U`$~0W
z1EAXm7C8gSvDJZLjK+^1e%Q31*)LjXt%C{h7ae3XwYK))xSt2N)03fQIZT;;q{U%*
zRf)O|BBp<v>!V?07?nVwH;9XrCo)+5(@syo+NM!IY|#AP{-`wt&jmxYfRhd7HEV#w
zH@bv%{Cl4_F#C2D-i_Hz9?d$_BooYn1(>uGaYClCkztSB7cIqr-#L;PJoWtBhmhdG
zc5#x;Y5U0}@ay$R-C|JkrB!P$b88Y$SP`##t|=1PBefPzru0O&^+hreo}_3M88-r-
z4?aB=2Hyjml8FIzq*kE`bLb7cQ^TPxtRjM8Z2L$4k4<K?*lhHb{iD>WL5Z!;dB6g#
zy0|vi7B@DU7LoU%@9q3I$Kpb<A7kymeS+}t4%UD?Twl?QFC~`S4q_sBp1K#<%og_V
zJ4MlnDdQiX;6L0&v>p1?@-AlCo17egaBi6}#Qzn1&Glv#l+Bzhvb*2F$E3#pOHRed
z3uNqR2p8QNk6l79HlvyirHbPOfI6;7+35RZ&_S2Fg;3H^naP4IiR%1T@H>!?%0@!I
zLV2?|9_sQ4_y=>pNA|I~^V6TCbX<b@Qo61Xp8O-2y%${$)7;DqUf+j+P0S}K<86Oi
zPO8Z`HfWR@YIguNSA^R9-`@n25th84TEHJKX=TFLvdNv}U6AcRvxc{x0O(CehK&sK
zyelI~CDNO?J0ob@>mxjT3_{eC<ICMMQ=|Y-_?)?L?z`GtKMyXIfR>-b6#i{|8u5j`
zqTmqb6UeBjm?;0m{shJZlw!2ZqR4yKM22!O2T{UxIpYpJZ3)fA#)L<wV-TUC?8Pp1
zWflgt8(K_UEI@dYpD*bt0?-uCb3QY$ONbj^9VRJ3V<DSR9;jhL**%ovyCkIJKwpt*
z#~HuqvtK7oVZr~fg^@;t$&QZD{siNdp6;1}UZlZH($LUYTH@f~XwvAhnpsjB#NExP
zh#!jna9Zc#W21>mEJ{tqMs#ZMs=?PwSZjX+4-EzyH2z2$ahR&P09;(;;3-8zL%@y4
z+5(ayGJ@I4x*Br|SQnHM&<DCm5()|$;8)vneIlY7a?;0{&F~4GLwvHCKP2=EgN?w-
z8z#hCBUx1A(SDJMg7yDGJqDsf+MGTkIv`5JXL|*d@OM1*h8pP1#7KWEabMI#GwqWb
zg+S+5y~DadcSs`A1B$yImOmctDK~kk6iAO^%+S>wo(KBY2rpXrJAFUFb`5C2dy#Qr
zMTHgIG=6*B$%7nrbo;bYF=x^BFp$nuvPKtCcv?J?D&rUxueW(X1?LseQ}A0mDk#|5
z85sC|#fhJe)|37)?irNQ1gkR#C6VFMBcL?R6_;_f8fD0g0aRsGAu=!K$ib^B;L6Gp
zDi=ABoh&H{d2D1nh|6l&tu%Rlp7l|1!?CKc*5AWv*stj5wfWAU9df7Mqviwo`|yjV
z%DK|*IlG>|8g2{*gbFL8GXg>#z83Yuvcbu|XH3Umf<mDI-1>^Xa66~@nM^bpe2HL`
zKYr&qMJNd%0)VsRm`Fu%-{&a!l!vK_i7^?`NMh2J>2ZjZVALP@6Y!na#-WJd4Q#?d
z(bCpm(??5uw!=YAHa7(ch2A#Xb@vPEtyXKz(JOffxOhigD9xW9m|&DCgSpF+-a*Dr
zl?!qRU~T0lF|k%$zBHlcR`KaV>5G9YiE%Nl`HEUTfU~?ngh4Z)j^N!53(Hmqst%S{
zNF%q%i73U0VOsw$`jusVF?HYQUnq*uGOg93S~-#}%7jalWx;GFqVFFrAWlEK5bhrA
zi##X!Sji96U85?MlJG&+w(kAwv$`uB4$~f~IfQbmGn<(R(w4Rmi`HOG=8IXH&WuBG
zz;}sR0g2(WiTM<{xf#PF-XW9|(}wN&9aq#X!5kdy&G;LTSI0?bdwr}g;ckL0k$b1E
zeOMM7)OFd_W~R2*=GP|&duJiGF*lhog~_sN)J973^&L`?%DmGv0~3XK?wagb+@5)_
zTARB;hu+UbB|JU_Umo}2MQ>?AX6^1a`q}BIfGfMApF_eHVaD~^W9N>j+@ts7;2Kvr
zJL%TB)rhK|@xy*Jd}_zR)J0)lXX_w}Uq`cQDv~icEyCC2+`h&Jy=jl>O?KPrZPguh
zb$7%lX6+Fi6?$>}g4|=%vugSh4r>n3P5{`9A{1K=jUtj^%tTwJ{s?uoWlm?f%=k_>
zfTpXp;rE|_$Ych?R<F+eoudnEPB0pdq|*4^J3qduS*G?(%<d3j9Zv)@JkA}yHwbV@
z;M^h&Og$n9N433%iuu<!c{YXoI)eO|LFc*Bd225aC6zU-@rRy@M<GjK-ucOA`ZQ)~
zU(&j4jPsBZV%1980e;nY6iE<LmbJecU}loZKnj(Im-J@%OM>4-3L@I%!9u0@DmxNs
zzB?7y2vpaAKjk4x9LL74ca47VeGG}PtkT}bQqc1b9jEfYSVP1T#?e&dQzNGm>Mr*4
z;fS{=SIQ~1v=39V*f2J#Cx7uwt8BAov!U1B$6)!gy`S&%`t`D|O!5$S>3s~KKsNke
z!W;W0Q!dSOZ|?5DS!Tyum|Vi8|1x4!(b!IQU);*LoQb;FUEQiuGG9kvlgHPex98=@
zvDZjj9WotGoOWAn(FTA1OEL23jC1;D<UW|=)$AAFO82bp!m$^Ce^Ygzng<T~vJq@s
zi~_GI<0N}3R-iaF-`&Dz4bb<Ih}U$@t+c+f=Gb<;(UXW5<f3=VyQt#sa2<o-c~8nu
z&qm(b-AA{I()YtzX_cxMQNomg4fCO}JyB6_DQy$T?`?Oyf7;1xf4&ktDP2t?I4mLI
ziV-j>c@>9rj)W9|E-uK;HSVfPNg^8%Z9dXA{(w<`eF=edzckhj0ciLN`7A|TLZy4#
z16zSN3?c;d&JLM^Kza#Zibv+<sldGJ+szY&^9NnRtKjX@{3RLYSxC<v*>aq5ujq~_
z{WBwI7Tuzs@rDN)nW5UR5VRCozIjjOQIZ&GI<FSqxQx_plg%Z4Gj-Q!g5tL6C&Lb}
zUpE2-oD5G{HdW<o0L>07VUe89u0RknMnbm~QEY!4)Ss1Mx@olonA~w}P$bpX@R=e<
z2X-sE+l!m?-c3k^zWZB?_oA2V;*|5JZ1m#7<(MEhKi@Q}KtYWZclR$3!yA!q%xGr+
zAZL-#ky0|y(33$eKg{GRDbI2M8;p;L5h;mT-F}+tzm^N%0l)&4NC`0BvvZM$v2m=J
zMY<IjGM9L29K(Z}jeS_WF1hjJAd{Hi^P9Z~(6WdG=HaADwuu%l3o&6Myp6-dI>*K<
ztSv99JKeJ~(YLlrMgDv0V{m-gmS#)U@x(&)9j7nP`DRO$$IkHd+RnbL{O5XtAY<R8
zn)m}Pt}0hKV3^((nXb2MN2s`1*F%kjq15seC$-IVUZ4+z@M0UuW;7iwOhMBv1Rdh4
zHy1v(3%&Ob4oK|r7Vu1MV`F1`d65HQYHj(upt!OsPijI5C$&$A;G>lP>P1u}{a>?}
zwdG@NOqVs*`;<!)p`ePsW6qt?V{Sitqbq{Lnd@T~0MCMEJB8rbtZW2XH$TXZNZGZB
zP~Y5-<R*b1GyL;JxnKIc^gBXsW5zjPRUn6Eb`m~sN!V~>sy7u|EFvm9Jv9hOGxDY_
zi^aFU<a562`wz~DxaBR>J}ubgy+7Al^;DF<I7nOn`&GwUnptD5MifRQ%Vmt%W|s;r
z!%GApky4~E%~o?CEpOOnX|KKd^?7)RfdSQoVLI@-ax!u=C-2lpAC;M&JTsH$4^<a}
zq;uxh6G~#|8PpemcX(=JZf52GQzIqy83Dd80#|vK`L<qFg%$JYfKhV0yP~b@QEfXl
zCMhIstS-)0NhT@bYPG%k&cRpJdwDg$Ix`f|#3+kcH*dm)JkMO<8=QnmN7pv>p>lw;
zh@ZLmO>omtia9E_1Az&9GwE_O7xK!W2$Yo@euKCmhluyA7^~dzohaB+uQgr+V@kui
zJZ)anh9WiFUC!6L`^7GNEeN!r{Bd>Bo};}Akg$fc{{~EDnNcj<@9$=?_SEjPwFUrx
z%M_E<*xlg6WXB~74~C-gICHhK*m)-ie1e(Oo%@t#t#neM{WT?e-aT7+w9*3(=_r?m
z$juMUrxm5c0^vX!x`N1lp{dTXvS86o7-Xb3n`-w)s9P`DZ*(+zxUU9?x}JctEfWdU
zG<%G<VbE$;8noMehRCP8C#e(J^7{g=#*7sEv)EH-2`RH{unX5E$7CX4oqG1(S0Wu!
zdeBDV|9t#iubuvsOaR`APh^y2WwpgQ&e|GRr~3y4Xqeat*_2E+Dzhu#hUYaEuHN}B
ziGfkC)$aZ&u5PSf5>IzsF*Bh5^Qf}7{h5dl=p}_9hy_|ar=K%;l=VAE$pF9*GlpFr
zv@h4J(f@@BphF0Jh`=4M`vszzMsqQK&oFb#D{0Ixa1;MFP+-7tpg6WEEhddFb5poD
zihQNNaOjnZTKyI&#|q#i=h#41WodYhmY&o#RB7m9;&it25d-2FNq2`sbGB3(Iz5)J
zHVeIH(zZ04x_{VS079+L&texft~OiEY>f#!aJXAvWs?{TyYnv`=1cFFi3}|GYbZEr
zW)Q0`!X4>(!k$oM&~N!T*iwsBvHT6A_!g@^YXa^)cSe^M#QGNG(^BSLU7jImQdXVk
z>+9}kuxP0DTCEV?t_C!>>$~q}{)H4ZO|kO(equ596UHh503Qc9KZ{O0o=&zd&yyJ+
zTsbfFySaLa2%7so4A&yEIHGSN_jms&hv5W%_+WeU&=U!>qqT}1G)`prki{SENXhiM
zvug22)liw?P7*K8inuCC(wysXy|(p2M+XkT+ga+r=NxoQV#vED7K&RZF=Sw6$NlSc
zc^tYd#2_Du2JG~}VpbqLN}k6oEa40Ilz0htP<g_s%)&|!(USi<ealQ7R(Jx6wI59{
zt1#~KH*nPKc&G_DtkL*W_RtQDsek!0->}`0m5v=>2T(b`jt1|2%K?cq){=_fH7l$4
zM~oBtO-`4TE>5{Y*4JI1hfpVsfV_K`=hb_#<Z6RZ00ANWWAQZd8KW|S{P4JW?A3Ow
z=jENydav#HQ3wwna!8foL0mNB*sUuUFz9fBOffPoTZU%>fiCkK@H4igqbNMn+yVOM
zL||ax4=`W8jLXUicvtMYI%3X0tzHi<Yz#TBlbJQ75LqUub1u-+I);vWXRYzkp-*<Y
z?QFEo0EbVG5=~x3*T)Qg!0ue)xh4YwaCYtHo@r%oSy9ce)(W^rw2@3lM?=LWM1jF_
zyTuuAzTjQsc6h!ty1uPLh%|><o9oAOLB1ptwdT}S1vW#B2{kDBz~9)PZ^;CL{kWFd
z8cLVc5DjWB7K1}F?6od39xf>USabHq7&wCiYV52m$<|aleUBOSzz4!NKbiaRNxsh5
z`#vt$4RLOd4)3XuuD4r#^ULj?&nB>uR6}~slLx8K3KNnNh({t_+&B(M?u72g=SE>(
z$`PO3+&Gq!j($^0guDeqaNlDOWfFGYD`ld+J^6yimWj&wNzX5CXKfO4ZEESa6*}|-
zVtf}7sM?_Haw85$>yz!bP%hJWOB}Axd1gHd$|?#h^o129ob*z%)7wL-ra}Wzu_|dw
zFI!%21d*T=)a-%0S*PO4>C+r5*oq=>N^HXyy%0!W!O(w1gf|B%KWGUFXW5wQpP{4V
z<i-t}-0l(c)I{xh!XKk`gDgDr;jdN!&L*DSgG$}r&+Also9oMr*@C`zC%vyn5%Q5}
zpB0Pfh2qe%D~xbs2AOJk5Id{i@O-VdxZ2vf`?AL9OB%tk#O%d#vEwIY(+QZn`x#?w
zRxE!_CNn=sN0n_r5k2qyJ*{bZ?7nXAwi36{B#{zq1};mVSU(-I_<ip79;>^7s@1na
z)s9sC6aH1F2c_35TRaW=@llyff=K(5-9Q23-UwTa;b}eJk1t`sc)+Bmb0Bo9DfL{q
z3cAgCNeJ_PW{!$&7^7rX?TKJ4@7Js$Q?)ZLZ?b<%25=>vNEXQNcNGYH!gEi!sxISS
z3Y2N|6C&_3alcZLe3k7eM=Jq<U33HRp<jT{%>15Fcj;r5ucN()QQ`lpK?vnYX5+Ai
zdjNW9K?Q$GDk5T-wLRi?)^nD7q^ye)h`GE}`Q5Pa&}_|UNbHhsZT{<%EBMXQ-0pr|
zx&1!DV1A02-qdjZeG{vRRFy5bs8-+I($LXXbvtmn6)?VRzX@@$-3jo0+&kTe`mZSM
zQ&oPcB`<#JcBbQIV1?TfYfi7#VshTK_T<c&-PH}Ks<pWc)KcUchG}%oBNv9^_7yj+
zC=~OEi6wqEoZxte-8mKi=0-(aZ3bl^=5_e_AK<(qvm=lc?Fg9B5)wsRw2HQ+YM<8c
z07p3aFWkG;{(S>Lp>ub#2E23jbPM+g`hGa%R9daqdG0bm7LYDraTJ9;WdQ9Wff?|S
zS+Tu_zd#gE6HoDQQ&UoPb>=g%CE-j!-0+L!EYo@4hi8KRGCiAIqa}*aqNjNcm{zX3
zcU*4~RTFKtt@!S(<oaSPHa=2EU-GO2PS@mx2s!G^Eyw`vfakS8S206miQew$wyO<(
zmuDCdvdOu0*o!XZ8usg&^FBv&na9`HnYPsW|1gh0bedWAjKajM@Bc-};tn_m^h4Vb
zFi+y`!DY`W9SF*Hk>YWZBS}g8b6Ad+k<(9sYG&|8T3Fz05()Q1G2|5j`{)(*$rJqE
zk^1JlnMMKJV67uz<f~4hp)Jl9y=#|#EB+At6D_GEhnrRse8RWsl^^tbH{dRywzM@d
zI<qq<KfXSacHaSlM8>&!ra+uV*)N9jZ0<)zpwpU0*6!$PdD1b1R5}_K$^lw>Yu<~}
zdr!aM7GJCL<HqUI+gZQlSuA^Fo@o;lKTN|>cBdEMb?>#lSdmcB%i(-(RdrIb{0ZZF
zrPlYo$#xwWh@}<~61TlC{Q@#oy1Lg@H9N>hr$tbPE*-pN(Bri}aF=gk%V1Sb4vYi1
z9n>;SzE#5*ePoF4s=v}4q%$eyP8hNNsVISwEc=fPEQpV5*sioWUkCI*PM)w|bGw6H
zIvfBRhqH~SR9m|tzm_?xZ%6>O`!`jI+Kycsz;;KWfSgSz9G|fqP&Fx@+NY|`zi&RG
zFAP&eS}T<ZY~ZK27jy29S&v1f4pr?sC)+TLL?Jn`XrSz)sxmtwW_3Co##G{IU>%s7
z7f}Ee56Y_xzL%p!_Su`=p0^jJ8h38pjetw=wRgy>&9>*4!TaOsY<};j2O2N-DWLZg
zf33UT^f{}uq(ac4rmU>(ZDzNhoqd>uw&nid@txNk5sN;?bsh~33oH5W--Lv?zO8zt
z^_fiGx9uL=9~tdT-MycKJxIb|tYW%jl%P#`ILKZOw+QL%_PcY9;w=c?6@~e5R)7cE
zWyk0*<RFCu7e<<uQi$*69<=nm!l|`QYs^W+x?**xE=UhYZ3=O2%L-V*f{gsYr^7^b
zHPAmoa4IJwjKnKJOfkRS!qw&GIjB9Ql^>_n;m|%VUCdrh8!CNGj<L|yFShHghnJ$Q
zH8E}IxIHcw{`4Myd%l0{?lcC}7X$sLI=VV(xh|oq*K?1(P|z+9_*vZ^w|l+O0a;J`
zq2unCH032cddo8lYfEl>V&2}8lEXL;`R;_B9`AlPYKub?OuXs_0XU>4aWl}xuj&m^
z>@JC^x$#m@n4T{rUHC7?(d8--H>L?jJz`C|MrtY&TV?bVP>2NFOFnyQ?tpEOpW5LZ
z7ic~B^sXQTxnF5QnbezKxG&q(JB-<7>+J%Z+!KRqN}|FE$%&_Mp8hoycd)rC_s%`4
z^4=<Ja~ngwIBKVXB4ier{OJt=?-d#Ji^VY)2ALT9mPK@I^CR5k6!yQgML+-bFD4$E
zdCMT;TM8iPJ^=iDUk|=6tpQ-Z<_hjj&5dk?<<-1fZgs`qca!Z4P0w$W`@Q4lyy;f7
z4rui-6#iHYde}maw|j%-od1u4-2U9mi&T+`WkfkVCOvbu(4c-m{@Xz`;{9Io10*8!
z70Gn`J4s8?^`a&}FFaGyVPoUZ_!8E2wPhbK-$6}~d-&WX1aS<2m<rfI{3QVXtOyo6
zM?7*>l~KjG-ayU{OWWD!F^W}Jak}ez;|jGSn5!?xiftY?j@(6t$UsOeMj&pL>Y(ru
zuG&k=<ZCN_TzS=||L0!hqH%DMuJadG1w$!6u(<arW-Y6_U)wZq&bZa_{4fdF>h$}<
z=866E{om5%GGKXc<LjcQu{Ts{sl(^<Xn8wpyWQ<eO;-Q#Dxx2uviO~0)5kxsCC%pp
z{c~j99ayVcG-zTORW)?^K<9fbEV9naC#4#KIy{7q=#W8*_?r|@8i#BM#Rv0g_)W7|
znmNP>;YLSNKZjYp((!I2I;`E$ZP>pFdpGDOUvb?`1aL9$0)HvjA0BJQG@S^v#H2%h
zR_~FeDfvlbv{MY&ON&KsGZr`hv4-m~zq&Zj2n{}BiGgc}5o)oKfpfvrL2sTE`Y$R*
zlxgh^sVSG8hsi(5T5xQ`e$)Hm;<5ev<LPPf9_WSNYV$kQb2mS)C0jgPba?l<x|@85
zUc>@wUpn4D1*-|S@VVyQ9*isxXMqjX^Jl}4-mq`v+ppj2S!Rw?XB~pk=`v;dJ{x`c
zV7x_laI!`aDU2Q>0LW)&fJB|(5HMF@4GbPl%1rC(2h04e3Rri))rWig4G46xun&po
z1|uwfTyDdoZLZei;#~n#SJr}aSsB^?J;3z<U?a9y{nKfXp?4>G?(qO@WA>((R@i`a
zL^S1ckNoj^hX@IdPwTfR551B*JNu-nawk}ABt=L}uJei^p^=5z0rdj^Iy`T_7W9kp
zr&?nJUTT-yt={u!(@;KmI4aK~7Q=SS{mp5)`?aAUrKN<cd}e2|9luXN7kj{~H?aSD
zqhK{J?TgpY+*c9Ofc%U;L_K#fs?};ws>kNEd;SWj+Y=-bRE^R$pkbf6l7dhXUaXC&
z*S!AM^8L0x?C%%F2k}9bL&f7BOSIMPQRer(6gUAc$1K^Si2Egl@U)xe-y~RagnVNv
zENLz(sQd|ChbTg8AcgwUo%DD9*QyjCq~n84nm_pj;{3T;4Xav*X?$BjF$TN~BYVPQ
zl5FQ1rKj8b{_Hhg^7(qJFQY6bX@592*FtF4>+E6>?}W?a`{=Y24{WB>`s_piaQM`J
zJkt+;Aqc@oVT%L*W;tdtiFVK0<AIVh5#8!hbbkv6T$lvB2=v4bOt`3VL(>AzRy`z(
z=nOht-OUKrYV1I+22h@u)3U*`z8uepkguk9MiH{?t2f&nPd~U)_%sqFdXNA!V9hlt
z+T68soHW15NQPPr@};2<Jl@&d3^o$jYu%jO{v)nCJ4?FqonYki_?6X^wJL4|vbS^6
zPs_nO)-NP2yZ_js)qQ7FVg`VFc-Uw2c{v;di9|0s_O>=$PA^7-U2|hVS_m8`VTTt|
z>h~gCXR!|Ne@{*VFE=<3EGQ4z;8}!?@VBI?9UJKT;=+ig?gQd9iIf4=Q)i)|AJ`KT
zyL*XqOwI_YojEln4R=!W|2^CP+A!?&d@m*Vxc2|vQM{#?1frG!RnJXo$ra<W(NB(j
zq&3hl#?vasU>ulkeFKW>5YTWeb8%6+m~8~YaH;#Pp8TqK=H}*9drcv3VWyBMJlQ7=
zg>b4Sv$ek}N{x2#)9PHT&BkA21HSh;^AG~2Ue`IRNKd?Zbk7_O{SKG-L(|$A5Cs6h
zE2=C9^KBV-J--3~JqOYu*F@PMex%JQb-A7P3mw|ITQS5DV~k=nbhCA|PIGZ49`HNo
zXF%nU{ulLGd<KP!lIw$Biyzj0jNtch`*V%|u(@Ewvj6+eq;+z8j3#EaK?dq6#9dY#
z^vmwFrx6Xh!xhROb8xE}*gK~>7MI~5JINggMG&OA`y+r`6@vnz8%NSn(a_pY2b7pP
z3#-Y&5GW6xfaRP#638#1`&Pf7*gg9K;A6^0^KHKOYl)GYwyscawmP-g7PyP_oy*xB
zp%dR<JJI6~nsBE#`{s-WdsNRfJJMk31vVn=`reeI9<>e5jt$MVES$_NGmA>fAhq6A
z)#n=sR1$!k50o|u(m33sD`Oh&a~UOJr@G@lpdxt0y4i5s_ov}I@6PtW5>W(@=UXPL
znQi6f?h*%Km6Ub@84F<wlNU6B7Z@R9S_XPSRgAxyX{9AS4u1KeXknuaOO_Wt#8|&A
zc0o)`h&>8rX)t4iJ=tSx=%fbMdJE8J45IY<jz|GJ>2Fh_AUDa#H?hz=0PpAXb<US8
zn76y2=BY*^n)Sdg$&-u>*Fs4J1Fr`{7Jx_Vsdp5WVz}s?YLDVk@_gfc<T$ln1`jQ|
zm^1WHWPGTh#oz8&DvK>`obw)J49556hqF+K_b`-vD!E>I1y5h^Z*6VbCe0u0H+_q`
z&sl)K%XZfSsU-l+7k5!3$;YoqDx{scHy``+UH<f>fZq`8Vw^s0f>hv0m^ks8q>$T*
ziHSIqj6)v`VH<fiFrY}y@)`=j6&lBp1XH1$b=vHw54*>o9Re0N@gDn93m)&Q4elRr
zvddCCD0d+IeS}{3Gik-Re-VY9?>LAoPJ!pYtT98PnF*6NpKojr2L>k?`Kk5bL%B`m
zEV$G#b0H5sr@e|4U>MMlk&`##ALy;V0}=CYQVlg#C3a^x#nJ7C>aPdwGui$1B}LnZ
zY^bkI<c;kP&(Ak@rD`REh4n=bhvm1Ah87_zkqRtO#n{d-eWP*lxXJ2*G6@LZT1CM2
zFOyYKOb|i_bNPzQ&f0Q#y+5$G&h?5AUbMEf(TBG=-d20KG5(D97nkQ`Ipzzu(653Z
z@7r6(53_UKeq}`-s~gBr^zSnZeQ|GF?bkT%efQ>yM1o!~JA~?cfFCcVB0ql{FsIVW
za$wM<;i2tVK}2D7*R&)|Y>1~?($oXY#Y3UIS8%My!W|S}Bi>(je+VYwAPJONtiMf;
z=``NYgu|Vrw7u`#-fP;A0;RP(U)(!_dRNtk-CmCmy_tB)@6&Fs`1IRdUbh>-rWc&A
zYhQ*nmq8DggBP5<of7jv6$)-znvTxciX-~v78!Q@x<zlj9GU+za0LD6+7Fl#0`?k+
z9K0o!?D?~6f-~w$@gToErnRyAKn_AdzK^kifAZ|SOX`Yp!xUb#NoVuT+w*{j_3Dy9
z&nOn{z=HepVt;cK^=3q=u}!^J>-{?jBO>^y<dT9{?mu`-=nRwE(4-~3DVP)Nx9h(Y
z0MCx`+%IBg35_*O3rBbRoq*=6ElHxc!E?pm_x<36&v#NQhAN#`Q%h-`x*BvDXuHv=
zh$cRw`m0|IHxb8Z=%^@?v<VvP#F_q7F=kN#aziGA_!)XX@>p2edsfD$=EMwDA<D6x
zRD<2(f|&e%Ci}N^lvUA0@mJAvFSvCP8OJPrw06NIS8d^Ud|OEQDF7Hgw*l?*&Mxl8
zIw|>0U==U<{O%|H{zeBIZGI3Q#WNt@Z$GX<p078B?PZniGTH;sBWvnPz17z4Y@uX1
z=*H#XA8SZLkm6bIH8AM3JBWSlRQL|<AAP~A!x*Or`pC#DD{M{nv65T!TFM?jfQ-Hy
z4snq0bxEafZr^vN5`ftk>Qw)5Ea8UeC~}$eluH9X?~k9Gkt%s$dJ_n2y-dlrb2kCQ
z=6og5K?aox32fAwBV&PMpk>FDF@siGOrNZYNhOeERAx+0EbNe*Dv7CoVVqP`iIsma
zsbZ<-)?-zzXlyy0?F|ONd29e9yMUd$MZD%KCJx)Z8cE$ZBY>gm;L%5`aS5uw_NcRa
z_`75eHpcrhBT~jW1kX`VA#h>WtLt#u*Sf`89D89<&onhGqp!BQ!fo-Cr#`4oah_Z-
z48fIJhNH#f89FqnU&H3EEb)^g<*|-)>2K<7?1+2zu<yZ0xXs62gbroX)h<ozxr{Y0
z<kkK_u+@t32_RHt^5xrfN&%v`gn}8$uMbSGPyb@LWLYEen#Q2pXIIat6Wa=<XvPqP
zkHadzz)eg{Ok8|$U|>LcGADe0gvL|q3_9J?fOvn5ni{yfiQ^f>J!2bB8Uv<H@1!Td
zRb2nWSm-`2H9Q%#d%f-neYe*0Gu?S~0e1GxQXoYE=U<cdQ9(X8AK`0r2(P8erLTC;
z3$BP80LkVOJH6gtb5v=%>%5BDr~J#Y>AscGkL=(R0aL$mi>5#wPi2)TwCWh_4HBB8
z2|d(!-$RC{brWF52@c&)c9yBqqP8JC<T_FZXvl*SUr<@#nvys(x}vIg4Lbea&xg;`
zYx92rHfU<32MobNzbH~)^ENLzyq@3NW|t+bJ3AUY&fnBMQ=NfY;0yUi>9Wy*+5V|3
z2LdE-E-cO;7@5yMe*6dwB!@EJ^Judb&5QZcg0|~4J2d}UZsO|1?<)#fdund+kmKX8
zDh9fIe~($cq2g6?yYNR>srWhj{8C`k73l5&C@R`B=94I5Jp_Rj;~78eXtj7w?<QrH
zBl#6Jzku$+UZ3oeNyf4_HMnM6SlL$%mtp@`VJP&kD{Ays+UUTy_)4gllty%ARmIYr
z0iRwA8+d4SMl~Vp5BxAD{!o-i+{S%Yj`tiucclk@O;zugF3aq$S|MBVaUU*=%bWmr
zm{$Z_w_ypHJ$RyDcY~2w(*c%2(`+66K*UaUdDTk2^Eo6j`=o3bG|0NTnszwMU5#QY
zlh2OhaZD;Q4G~Mp#Hd1B{NH`Q0bLT1IrTmZLm}@#@F<eNp48Sv6s<PnpJc9ry;PNU
zXQ=4ewKktX$`LN_d)1yi?B#?$l_J2+&CSK7p?SX*c$&=a_Wg$HrS*tl+{UvWljTUp
zFR|u!-aX3aWFqv1`wRshRh!rsmo?Nc=*vqgL{2!_2)SGAkCPf9S}mA=bqD_SFIhe1
zrscrA7SyYw^mfb15BSj?oR?Hv=`EzD;$x$xRnr{{)nE0}4iYO_T6`y6iVNU<kpRv{
z=zf-i2~e}yIedN<=Y1?UvkgY{5ZIrqfgGiTJiL&waRZecbKA~!jlg6EW4M=&laJ|N
zED{hJ1Om<q6E?k_ROnM-E30&fg&nCwUOE`56X%;7SmYeI$gl`#IJjLm&V{lZi3^d)
z^v@_L(bVEPSQ?RM(tIQ9Qvhh4yf4w^64>#~lTRK()nP2>LUYCiK@TUhY3<iZuPyOG
zhDAlWz}m*D^X^*aaEZ|$+6-qP9{O(7uc6bHaZA0;9WdB?R<|c#-eZYfME7fd56C*1
z!=%@4b@V<BNQ+y9+xM+(B<CicWdZ|xmFmmez`43Kb#Q;>1yMZW1ONwTI!RZ+A1ZX4
zHT^{oFQXf7;rQ>z&EXnWDGjba;w`L@2T}_p+3jF8hHW$Rt7@SJ&y&`$cbHSa{TpTF
z;;aB8%oNgvDpfR*z|D&rBJFD69r+>0FjH0#o(BVA!aK}<VoVax2~4048(wu-6pL2N
zE6=_$Oaox<h|S~-fSEOWIv$C*KqxrtdUaO`HBGeL?rOUOJi(#S%Mat7Ss`f>Ral2`
z?%ke0zoa1=&TO?r+%?v35`#ECQu09M*aax~N;-?uXISYjRqNE-?H}b`0%b0`%XFhM
zF)()xz<oD-tgMXg-)5i1x5~l7L-j_A@;Vd#Nwr`~=YS7DX=^@bzO#8rxa&$6(oOIe
zX!?zuuXp##L*aQO?U}-ZLF;pz-3I*T{^&)tNLJq`HzIL%t_xz!Qz&&|fB3~~Jym)P
z{tA%x<>@xmv1$}`Wo2(>m6H#dXBNl=1KM$7Y~7$VZv<Znb*A3qz8Wk51&!N#71J8!
z;2NraTXYMMomdc#WuI8z5)*ZU90rdV8|kI5JV#E*=x)s`-Iv=Qv<^bb>hybZxOp3N
zHs{gB1>TY}#Eb2AJlo^qcY8jM*Q~CT;-C};A<6xG-etxo*GA4og1}$4KsB029(sPa
ziQ^gF4pE75Y_I3)pY&#o<cr=lDGFNFJUxWN;#dHLI7Y%6?|UFiKG?~Rvy!-jjB7AH
z(<XeApOjU(&Y0Dk%5E*UYTwbJDXj0sc^Sc(zZXzf=}oU|rW|`r053aVzFZW_ehbU<
zz{G~pbCUXNs`GDg$oDp0o%FMy4u3j2cR~fZOdLt1E5K#*OKM9@`T%lWT%ZmThBwQe
z!!RH~As*##->FU;#AIGO>4Q&4V7TN&VB>qtbASt+(!QVf;{hS_<hH|sl9~2t)upbR
z=2AA7_u&>?9tqMp0mez}E%f5`^@zk(kZ$s{dt?56N7Mp`Y-V<=e+rdB0{-E-Q657-
zW3lx)Ub3>13vm07Aw`K_N~`^k@T!Dt^a(KUq2UxaG>B3aFBgS9Z?G*uy~9nxM(1n7
z3fp{LF0i!a?Rf(;S@@b`<t_T#iTn=z$|5yBz_{D}`HCRn06s&|`$$a^EXC@$fGUT6
zYY@@?mW>9t&@{MfUl4*2X*AGe1)Xa*cWl^|P$}gU4=7xpBU?f95M)Y~Xh#{rX#<Fx
zRjztp&_R4$S7-BMcZ9gvVh4%MBwrV5Ctq)Odc7Il3>|)Lbvb{6CErZ!T=w$z@cH??
zv#f1f(YoaicJaBBXiW|bQc+S&%gaw^5bU0t7F7~QG$sV)xRAzvkDb}dss>&*3=|tD
zi>4IR0#9J<95eH?D%mvh8d=+XHvk*DhNj~2vo(mRDFa=vnySNv*K+sl&o|>qiAGZc
zwD5OX(m%uZD8T7MtMB>JWcq(lV!zL`#PQ^u*MKA*9^!y9r6#bmrxs*JEN@yget8T}
zX+Ch~%yeUB7P`Wi?swc@QU|MCN{NinCJn@rv;&HhJN>4(TAs|5^HSPk01iY;-{Y_-
z^}@%_&*W;PEvTLC9Mvz%1v092-Sn8nMK#rf;N_mCmS~oxDNOWAWZTSwiTO|KW1d*e
zYkHmd?99ZZ3=pdrP+^jsYz*GWCHbIqqvoBLS6Nt%fVbT${&!9Wg@<MhEHjFVQ2#Z=
zoFD{tHU9b4E}t?NF2yKRz%rT=lM7;cyQ!?_<?Yi|03Pt<`n}-*5ZACC2uHj;(dDc%
zkWku}SYF+_s+y<bU0>Hpk0q)Ur!_M?RZwgFJrRz0Cnm|OVVYbp$*qNa6^uG72t~*W
zIm|lC@ngQ$7KNHnHPh4T4S$F$%qd1g`R|M)yu7Zm%Chg*UgP>RfT-K!{asfrwO5@q
z;VzR3`RQ?zy1YB)Ro$bYs4}<hY$qV%?(_ZWs+23r=_?CPGq<K15<;H3AJ~!O*C)u(
zqs}(sRZYl^lruk!S6ixgLI_i^1{GfVzbt{Y;oo|6IGKvEh<grA3}cdq@!b9j#58Os
zOHpb@zfpX^t3y!J4zdczeOOjEa9tw0ekw3x9lEE`uaJir&MPpnOHRe}uUNqiwQ%}2
zBwx}BQC&UH$Ktq&9&f2?%+NmqTaZ&14cSZS?>}S|5N5E34ecUio}gf>9YK}VLi=B-
z&~FfUN%+HN0`fg(6GJKyA`8jpD2oknS(-|E>x-NK2Mmc{TVIG3C^-uTAz%9L%cY2M
z($Qxxnn?C~T$@e@LejTQmFY{>+UhMfw#uiy`L-o-*Vg5$a9C5qep`eLF*`J7CRTQ!
z&fZU$kS~E@&albI3Y(oGd1yi=#)oftW|sf1D91!pPmPu0JSiwi<;t6U?q5>>n*2_s
zEV4TUpmkeZX64^p8Q}Rm>}`go@<H%c&5goeZEfm8Kh@6^SXuEcIW$yZlKACx*i2t#
z@<wpjYt`2oN!z(CZ3K@9DN53k(X&%PDADg7@9vpGtx%y!^azoB!xEe$1>r)0Zo{hW
zD;C1~XlUqYM{0=VV=$Ne6Z=iVe(s-8xQ>(!;PKuCGM%7b1h<T327UGZ8FYE<`Va{~
z5c2sxp1)(cGd?#Eu6F;r96R*1kX1}2NKUISE2(H|JhsvleO)rU|K+PJVrNwlJ7s|(
zgP^S_Dvik2u1%?s&Opwtie`HI(QghE(O_0gr=soIlu&aeqFNz?S8c+}wS0JBKj;nu
zMwEkVs6rE(^o^Xpk|K`cH;);&JDv{?sW?1gpQvI{&fKrzw?b5F@Y2+vZ-St2I}0fW
z*GJ9%c6%60=*Xyu*raq9qbW%Pfrx{_5U(CBH3qY$I<1K*h~J@gYw(9YWi|N-LsUIk
zBG4xCsY9c*&|qcrHD3^kIC98pwGXm@F~+j1i_>nO{j2<Ad*)um5{{EY3B!9x8U6os
zqsY%^+1ppw71b-Xn9q56A|HGiKY)<FZ56i$wBJ8wyDqKsrPFZ(Z3LHUeu(4Qxi#4K
zuTxR7suE1}g5}2_9aoH`9okn@F*5z>MM_ZsY9?9#ho*Djj_ZB9aBMq`8{4)UJB`sa
zXly4ZPGdH9W81df*tX4ge(zfU`2=&$ta+XrdtaO5yvt)+)voqs()brNKMLUHn%xy-
zcrLsfoy-sZYH~EN(HTrU)UAD_2?AclkZ!gRg#^y98K$LofU(5HL<J50E$g5R&{xq3
z8a*1OB0zp+-s~+7aUJb`>D?U!iHPdkhK3wN?kJHr<=HR$*BD{>Td#u%AU+;9jT#>*
zS)n7w=3ry8goc`&l<e*r)qzW$<WPN4eQQuy4}-wpS%gN7#tgg4r3`L<5bszXD?Krn
z99h>%<S~Wt_DjrtlKWF|f4;^YW9B=#x%u8y#7*Nu(tdhm3~@$wZMA%+*ikd9=*TT4
zLTq1@AzUD~t&pV2<5Bs)f_uRf&#R|rt9Z+sZ6>9KV+r<$6Vp6h6+qvh^{>U)po^7e
zo=a983%iXihQrX-F1P@<nh(o&ENqcOK7O)s<dhgl(NVh}YU0qm__EyYQphUqfrg3(
zBl1SScitteG5-!-4X3*gGH<AR<dt21wH^LEj1l%V*OS<)LVd*P>h*cUtpA&AV~~b}
z{jE7EOfi|;F=A9=suF0*b^iBePpvPrGz$kUwB+u`EbU25js~uq<AZr3@2ld3Gl}9-
zHyyiUN@XY!D~cCN$zO7Q#lIE(AnXn~(~@U>A!v0ycw?6HXJf!?CA*WSVQr+6P<Ef$
zfJfSy-ql9G$#cNbIg1%=`r0Qi29^x%HzwaN2eK`p={EkxZ~{OwY}%J#sf!6D=~vVi
zZ(5QhpwlTqy1?`%pxY2`omZ)IGS$pLoi~q%IKu={?Y?+W?cop!snrPz&x~nBI}PXT
zNV7-OYciDiMTVq0iCF8Dcl`}3g__DId6|%o5`TuO*7S#|&o6G>rPFlv_&8yN7I3Kh
z5LJcco%$ySSqx}iT9ooPH@V-Qn1W1?Xw-a(ti((x3kQQi{`8|eHF3lF*ZFyLkSAT)
zTkkwn=>C7?Q{nnVPdLA1HO_(1Oz#q6W<S{Wz-d};l?+<Xr$;V2eFGfC#~f5o^s9zD
z|D|)gy(kefX}9t4Vi~5%Ue&Q))26xiGwC%|fts;A9zaP=QmA*-`>W1HVlq|<^($h5
zaEQ<~7I94~XaEmh1I~oQx3n2<28H~LgXA)6m%bA7MK}yPaf71%T>OA()8T2aVb$$Y
z*z*OXCK31hWWSkzm#}kty-o{X&yk2epTXS8>kEAESkF9%UlTopbTi}4ArrQoXBL0B
zuhQF_1K_z?ZJAKe8h$=G&lD!~UA5;7O+S-4NQGfFmHQ*wZr%AUu`V<XIY2|>+(P&l
zNykjDUhRz}6md^BE+pFOCHZ@?@yW(RSm#+D5YHS#ua-9APR2a2e8;BzW@yhXklD(b
zCE}hJ-J?3o(;cq}VrLAHNJ&$&t_EOWn|dSC0m%MObbg<7QtWY~#xP9Vzvx|wczq!E
z%(FYZ-%l^a{mkbt`wf5?s1KwhV5!~n_-n@aK+gWfVX%Eu4GZVI_(0^>rP4kHB)C7`
zMm$dq>1|j+6g0o#;CFn%sE>3ou9S>;lzboD=6>Mc)Pj}?EG7B2`^hP7C3yv1p8~%l
z0JlsbnUb??p}9FXVnoQbR+m?hl$qb{(%{6lWchnN9W|#A(kh%Ryq~Y^Pua;8eXIYV
zWDE^IyMJlnaa5WX^utGxmA4^O(VU0Bj7e}AsCaPtAW9?ps3wG7NKJ1BTznJrCh7%1
z^XJ&ouT;b{_+@S+f=;)SKJ{~utanYo=EjWt$aiMp;fo98jec>7BHb9x^YzYL7H)D*
z{5gL}Zvp+xGaH87@PEn!mVNoFuR%3`g|6`4N5%Ca#gMV(Zdf>^b(<Cbb{yGbe0prr
z?&ABDvX&YheTmc5@@~EzE%>N2=UYwt5=~el?tnE=bI=&AW<HrEy$Y6sqSpZftbA{H
zjo`*IGubnSlm9opz|O#GGV)S@A7*5oLqSExSd^2WR~g(ObA8Uvb5W!0>Vc<4W#m3u
za)fuPoFM4@nc>=&Tk;jn@V@)whD_PnQ!ZN@aq-;ds6}CcQQRL;V7D5ELgiS$+A_M0
z8=bANN@o08>~F-re`aAG<q?$zSdNC*R&PBw=l&Yycu+$8I2Edb4oVHn?TjvL4UZ2f
z=YSw7m0&{ov5zVZo=u!n)6~#QqOr<F%IekEX(HHEWpH$CT`8<?(ArjC?>v<y==GRH
zBB1A1{@g#puJuT)7;y10xB^+Qx`4~1C>MF<(#WAa5Y@c2H}0L2Xe=!Qm~txULz;k?
zREQ>m<7bp<^Y*+w0&((}L#%a<xzfy42w27pDhhW~)bdaSSb5=peZomQFT&m>BU3q|
zykI;p>#ZTpm^Yd~pO$-GS+1d>Ptmqy46xCTukxs=q^7cXm(_JUt>#AC$A7l++L%XO
zV!{iRaTr|nU%jND;=X+cd=1>-aH)CZrwaoNn4sSruT9R36<8C@Y_s8eLwW}VX}Xhh
z_)|P23Ya%l{<&`VO!!|kz65$TR*3s6By!m@GOYvEAYAS${HWE|&(oR}kh7!7PA4uQ
z3vc`I2ePZVfj_!~6KI9i_qI;NfZTBR4RGA~qNAp*NZ676Ha!3sk-Z&EkMV~5VxxzZ
zEyF)>1pTCq`G4^tH@z+5?Jq@Fv((`3&eJB1v7rR+`*`+BNYmUbW+3a3=Dx*{Fwq+r
z#%3n@$?*)IFpLmL*3ZxnV^r|P!#*?Gf6^jBWYe&~OCz<9$j%g=S^K?hHnetsGuRwt
zx8|Y~*tT(TLQn*B+I=r)<`%P%4bCk2>dj9_2O??=^+DdH9#DRUtcVS9e`WQe1bt`m
zdKYZR^-|Kx#LC9<`l>Ju146s0&)O%6gCUW!X=8`rMykV^rmR?N+kYS3SSc|ybqXjf
zQBhH-ngK9LNd^`)Nm}fFks5xg_S;pfE(Z{@^Kdj3xYPyK`PmT;eJ0x;!_5U_0vlT5
zQaO!v<aGX7(e5^Ux6rMP72L}X^PPRG;ezj<wwUx5D5t8xDxB!I!TEjZW3~%p+*h7&
zx`)x&UVSkA+&OgDx`^A2iKKH<0DRy<0{c59<T<D5nS{tmX`nq3&R__PC*Mlii6xy@
zMZh><4QKk*4C`Dg+!Mc5bIV;cW=Ts)PD4dg6{$-_&jd;L^Xo9@zy6H!yTm8F=G-JR
z96NsJ%E#>?>ZYptiz?a)8K{}l{QSMWM0vEGs^ywTd$rq|Bur}ERYf(mVo3hdg+o&+
ziKn(!V{m{(7%;R`;>KMXb#cB$L5hf08u<o9f;5|rTnjKQ#42J9(DEmu%fAht@vZ)d
zH6Qk*vk~<-i)s-u2t1Q@zUD^Zs$G|>Dom}h)781Z#zp%VrD(igFS2&4jsI$`_Y$X@
zGHkKGnVzgTy`6p})-B1aYGpE>D9(!v4hteHphjM$VlU86K#cp7MHsWe1niv8{!<19
zd+`E-w$>MALpKLAC~I3Ez^{2tLGn~hv;3ScI+yc<b!tA?Y`|B@y~Wi0xwN0OH;jgZ
zgMo^cyvbhN^xRf8>cavNW%&IIL8EoIyn`@Qb7*=QI}NkZF*J8}&Q(f&{TZiZ%P)s*
z-<*gDIYN0za{j?xEN%|gAe`%fnh8+;wyhB9A2{Gsjexpva(63Pd7iEDI%;C0diQ8R
zcZ%jYLik87i(;9z^})^<A0a_TiQ}SJ;@C!>rI(0+2Gch~7Nl3Nb*2>$A7L+$Nrv$s
z0_y=6Ix^Qu-N6^;`v%Qggn>PAYi}!&>~a?;O9E5ZLNPzFE|#FV|I?H~NfFR$<Y24F
zv78#A7iSjgBJ6Z+Mc?z-*~P@T;?IstH`Gzl@4{sfc72mLEtvy>mi>9W9+o~@CusU~
zaS%cpI)x0fE@;;5pP(eWr#ZW@I`RCcSD*p8Z<k@6oGd+Er2@t#`1L{}Td~A_n2zRp
z?ea-tqQX`y-qnqB>~3(fT7X7p+5QbF-%%K6_!Ka$pc}3!H+Afn&*EJ7_=V1TtFKBi
zCbVv&4}aUpS|a1Vw0;PpVr4M$z?wnA`KQ*moUWMUwuA~^d(WS_)#QpG+i!XX@#xH)
zR@#Kv+BI=52n6IX_QT{yl2j3{?egkMupjWAgPQaxx2;E3`J8v34wRm{adUz0`^hY;
zORxqptYjvAORE)q$hGC+-}b!~c)H&`Je%8$$!`;}%MiHt%!?S1AH+(!3d7s$X&5{i
zd%JeM^B{6>l=DouO&`Qi4A5zzS*o@M4zTO>pw(hL!Hi$Tn2dQ{#H|*q6Wb@QDgc=Z
z74JfW=bBzKiaV`G0NnkMbXV(pgRhON4%;?C#B#0#Z)hwov=X*w<8Wl5H;V)*VWk1_
zMJ38rZ`rUM#*l!$XE!#=YVS-ID6+Jg5gpptz$e^YJod7D9b!gX@yXX5$7bEF^mH;}
z%1#=#DwDs_ZFR6ViXg1tSI!2~xMkL#v0qsebq0CHVTY6txYR7jwNu1`5Eejc8|o{$
z+#$lkPflwvBT{c26Z6SnM4oPuNU<<k<l731jI$sQl?&R;;Cm~jD0`6}C4~v#El$`C
z#E)>VX6rJ?dRe#>c5%Ub*cSpz&)R?n#gosdtu$5TcyK$HU0M#&9xZEP9acdkX{%fK
zRTUA$+3#`#=$YCn%3&pFnDR#s?jqg(CnqO65CoUX^kF@xidw8w=}KRcrz7dpOyE~6
zwcbfi+|P``i241=&}VAs!^>vOrMofDzBqebF3n8)S$V8DL7%wm4L_=le$zLP$*cPI
zfxyG_JAs^yfoFPZwQpjXo`nrGsWosIXw^8rwjY@TzV0X1V{?l)B`%z&4vf+l7MuP!
z1@Dddx`qfZ>{Y#MRQQC>+Mm(WOJ7-TwS5*L<gKuKeBo8iEIm#lvj_ZbGQ*n=44<K~
z#-G{c6}xrw_+hPXk;B{^oK9dFScO1x_;rQ$H+=J_5v2dX5D_6^CMbo-XqLG?DCa7-
z4AYSu7_SQPAk`F3Y0y$1SYjpRa?;L82rI=qJd;V8qsh&wwEJOtxwB95MY0Xf`{T*c
zxtWBIV<<X~R_`?R7MCe7jCj43fVfjm;<7+3^ow$MLVSEiMq*-ON=h{Y9|YINS_art
z{h#N8Ka^H1o70mL$!sLOvo55^Jy#(`Nje7(0K%A6I`1=E<6r!FB&_kC%yvz)+W!_K
zAUGW5p7oON(^}Lk)I}G1_b02Fg5%jJKSS`3gt#8i34=Bki}yM{ACv;ud|@+zT#X7&
zow&fh&1M2~A&e)6m!6`nrL(%itCf0`i%a~DR=it%`ywxfFN2f`AwY<DnDYjQB~1tf
zCM;Nw%Aq)L#U7?}hc(}W(7E%@JuZ|qbiSs?je~QfD(|PdXVGrR>s!f8yGO0Akh(|f
zCLE_Fc<rvl>j<V+a}HbO?6$uoSW-6KVWb9&_OYy&YK?6)9i3x~0#iFf^77G8PuG6c
z!Es}Su|Ak2+D-*fs-A$D?s9WUFT4O+pbp=Z`47lYFkhTDP!AW@;ifg3c0)9A-a&pj
z+)*Mdq&6H=hx{3*?Cm8Gt}38~Cg4;1ytNUs!nj|sH5>=VcR9&bSdG#>6_%NZb|uVa
zdJ^q*pSI!|GH8~x;EEHnQkoOcJ1KU5CNctR>PY&`)I9;?GyM$0_mZmkqyTwUN?-Mc
zrUz&%b4)=TRkTzQtAx+!U{7MCO)3+6``;Wp?hj}9k%tD;h)~Rif=6B6C)3drlFL8K
zp{RVEE&Q6&A;GdcOD#<92YdgN)R1i`;=rVZ$5;R5_ZSpV-bhM>ofMeDU0qSrhQhu(
z^HwQMCEmBGpqyY2rI9g`g8~BZi{}zHmkGrjR}LmWKpj-NTZ7&a@RZe0<oEd&i~+F=
z%)Wk3Z5MVeyAc_0y0(`%xcN*RAO|n4lYs`wv&w3sVtxtkm}EU>Fsl#|t)qcxEx9GZ
zQK>>nzEq0X*1SB(gnI)&HL7*6Tm`i$axZb(3R{myXVThv+#qQe02!9&ZV7AWTFai1
zwE9x(*If2>IQOlU4#-L(q9C4E-FE423)5|P&!Sg>@nh0=LuCEV_CLo^h*KNnXiLtg
zbd9-kLX1q~ivy$I<Ww*OBk_e5)5k^`X~?Pum*R6>yp{&$#|6G7;abyvLv~_S-O83e
z(Vo)nhUkHs2Xc5V0Vf<p`XikdXvQ&}CKu)<6Ccvx=x@jC*U-vG|4Ndlw!2A)`*m>L
z)vC?ZWDGP<YVAQFEHxt1__-98;Gco_ZIT6hplng3iE?@8e^}FAeS;RZN~wLl|FPCg
zCHPTlMp9wTZ27!R?@r=I%Z=oqQKYs7tZIJe`B%6}+@wVTE}>Jdn?5qM$iT;2#miOG
zwb{=tM=KNkft9znfZpCbm(%du+BuB89Jy2f=(urC%Je2Xsbp_4P-sO`Rj*goG(C2l
z$_%(6e;lQ?iDVgXf%6#z=A!jRlZQ)tRYBS`#E<O9HX^z~<l2CgKHAcsMyJ?GDQ&*1
z{%YTCg`b}QdNSw!r>$M9O*C%-G2ml4VdydRrOM4UgLkIAsuKab>c6zE>SzW*9Cx#~
z6<IX8leDc9E7{TRTy^|93}a$`7Zk)IKc3jb{py<e$ZTG028h|Qhlb4-`y2uGVcD=(
z+_Y$JkU3msYDF+b_9E+3^zYjtx9Et$JNVn`_i8|fZ5|P0zs+*=nxm+;w$yUBl~cRK
zAx^Y7a;P(?AU*uMIfX<{GMKKS%EI@hZG-O$AE(?|;(C3v1c?^1Y<~sh6QK^+twYFC
zuHW#wnQ>I6>l{&|$CyI6eYemZ=40qBA1SYR`a(5C8BQTRH5aS&1l&r08NMeWL({9=
zpX>uh2#zS@&>KN;YG3a-unK}bcyZVY*X9n4P;IqMc`bDv!{UPhO9yc)rPayVdJ!w^
z&eXY)l*ZW5E2J2C=HfM{yxRx%ot#(HZh%Tx=d#*!u{)|#?ocV)LI|{)erRodc_C7M
zw<!xsXEN39pEV6D!Cx+0z}YVxGFfXlE(2is*AQigqnw87!irOxL#Qe$wKB2J@?&nM
zNoYbz;o-~#gjZZ~i0*IraG1d5*r$nr^1pU_jFje2*#Yq)y~DeZBzW3NBE6q5t*5%+
z&$Yd4^-CaxlX95Z+V%KTM=t#diW8a@-sg#*F(K-z_tqx*9{t)jREHWoi#x;#DsXj-
zacdQ=XIaT84_~c`n}(VXBj42bdSr%T&?8eeMj-M43v719<C}vi88|SLZ1!z%Oejc4
z6AC#WZS@Qe2Ew8e@p(U8#rfn|>&B*wu7sPJ9x@ohXtalM9XX<RPAjWzxuk-|t+WwP
zdn%4i97LBtEyNwAWm!glunL<A0K#9MmI(uD(+Z=7&`Uh_`O_vhn;+9KScrtb9sPPH
zu+)9PPij3r?{p-N-E6I~(2z~6`dibAMH2Uq9r5Yg?^rN9H)>ruOd7oH{A{^|qauHL
zXjt6R-7rx`f-x~;PPy!6jr51c^D4cBE%U#$gW)$6^440zlr#nTAX(4}0g$fW+ELO0
z+<m`EB0zNyu#6G9NN9W4Kl;7d;-dRLDF}<g*^psvV<YHNT2T>YpIS~cGNql`@Rgn1
z$HP+Ax|rD*kFHmVA7vtAs5V8~D6-~)dk5=Ul&LVqUZGrt&T@+`34baET%gUbzBMk>
z{M%Q$Unrt7w`mKNM4JFaEMR^%uGR~u)pI|2?w>_}UNP~So1C19wqu*iNDo%wDe%`A
z;fkL0w+4*WxAXj<<eUUzxt~T+k=pL&j;y$-+}yah)oj}vci|&+>%RHEkjFeh$7xqp
z?vg*1zgQe^5yu&*`&?O1^~L0OzBPBN(c_RqE?L%7G#QESKM01#WCN6Np5K+7>&cEz
z(eHw-zMGooQ!z!L{HGJ0SB^DeIG<hQ8s;2-$#VkziIFJvvW!W=i(>>ij6$fwIx}3+
z#@&6@RX@pb+SsZck+qUE^nA9|7fHE5fL*b!DCTu>;BU&~B|JPZeCgH3w-U{DQM#@d
z+d03eX($Sp0?Y0I5KTpHwqbo`W@n}ACNj*(7hPpWa)T|8kocWxyiQNcC^t1+|2Dd=
z5sWYAV|~9@4I#kHJet(KC#JSJzrxDHTZO?vB#d@E0f2pc@8z;<em#pL!nT^rQ*{I}
z#89$Z<?rl!mbRqKy;f@iy-u2v{x%5-(ozX@_knwrU$avI70PMiuP89X?gFon6S-8;
zhOpUiB(Rvs*2x?;;7z~VlfpW1q1dxFDZaRC8o-<CJ5gsB;0J4XDd~0I2XTB7t5vb>
zSnUBt1(xs%heN0&NcZ{!`fIJVYyC#lCZ_0`Yh~s*S1xGj2z8i!?dJ*{_KVS`lN|Bk
zd{!t_lskxkH?i+^@|FK(fR~o4;xFXlV$bC-G!fyyL9!7<czF15nUOM!qc5r2uS{c%
zn?5%u?~jR9-8|iMY0^zh4QF(QnOHkmSlM-*7gsF=LByltSbavS_9sW8q&bx#^^^Z5
z<&_jP6O#i~>{Z`JcukxPYNKTDrZT-V8J+)Bx34<`I6}0X<os1E{~a`6o>Cd)jF^lL
z{h6K>+M@OG@O)m*c*EX8?E;hsp4qO+aZ74ekxI{`F77bB2P%R=tF;5mJSbEXx&%LA
z1n`wb){p4H(&<8TLl!Fp1<gPry?_ZGlSywJP>)Nd@?T>0mMOj!_ksCfsOJiPiJY8_
zEFdfU?Kh|~9-%9>0Hd8Hw;Xs0Nuqb#^nF@T2>=%;@tF5|=uT?y7xA)b%lVneVbJ`Z
z#*yW9dw^ff$xMFqxRAJCFa*=4OJIdL*<(_8sZJ;T5y*P$&JdheYPT&>mZ8E>1NGmt
z!CYxg&rYvUQvA|bjxK!I2>q*+g`R~4nAD#Z<CTTp6$~HS1?AkfHBFJAY__9NZ$#Jr
z8S^gZ81Q!pK;<n2Lsf9T8w5G#>sSM``4(!Bsj*@O0(a|+Ia-JH-u}kd8nF=vRiI<-
z?eFjXG&MEd!?KqX`MycZVYB}5H~F&N1C&U;3qq)e&E&eCKSjsy_qe9@u4uIZ%}zlx
zv(I-gwBFau*>)Q(v?Q8;kvOnrPVP409RteEE2~0=YSnx}{ek>IKdn8!Ej;_~^Gx@Y
z)^Kp=PtOAsa0hz1CK4g9UGT8CRX2_`By|4Td6VHU0P*T!Pt_;Cz<#qQdIyeG0_3?f
z#q!ANIW_26HiRo)n{Z5}yjFpu;c)m~3K$KEy{zMQ=-V7Cu(h8ofA?_zXwFV^*SXxo
zM~52{<8EKtT9y9&$|%`!6NLI3g!6ZEmQuO*Z?tmz#YwP)=6&;Yee_ovl_c6I7Z-0<
z!28L<x&H8OL!qT+17|epLnhc|sEvem-Ui-M)zIxQtV1_v+qREp^AiI=zCjY96?ni3
zDa@j=J(>84t2y0tjiBUjG<f^hKfBY2wJRtf#J@mV>!vgcR4X}e9-R|(yUw^>qNq7M
zUUYaJCBz7LBGnq$Rd-SB4h|WDf$sFr6WWF1&ZYVe9>MVe<!CR~Rk2}%R7h{~{>FP=
zc+p9!(k;0jK44)!dkA14R_5namtG6n>V^rCf&q~}2~LZL^Z2vc_b8QZVZ%d-bFZ}}
z|L~_z8k~2|%pvlt2R7NF+}-~d#A%7U@X|az;|otPu?E~Z=j?v7agfRnEg72qv+t$!
zC^Y!B=&6!ilEbqq+qT>wnfUmZ=){jt4ELR6Y3_et6@}1$NdhkVQt2^V3<03$O_v}O
zwjm~{nKZs6_><ODbFoDmVCW1_hlMBC>y0rt(pI&3M9}PWNPM<ntfnyPZ6E0d4e`;J
ze)#t>6vnev@r+?C`hi0&M9hrLTA3C4=^V|pObwJ^0)i19eF;mt$%V|IP2@mkB+AUt
zhx@|7ON+A$3usVeEQpU0rbG5KP0^MvHNJkcBDg<hcFmxc)Q=|QbBCuZ@BN>cz-V)0
zz3od(ql?~o+imRKh&-}e#5cpH^KU~FMF3mmKPeKSW<;z0^P{#n;$z!V2}3CWy&eX$
z{-`@?Fol>y-mpJnL`?L*JL8TXwZq1jRofI`bTa=e_+<sk6LABBT8ic%5_G+@aB^}|
zn?rxv9P<C}-MGv4Trs3M8&p6ivzn6|+H%n61@}Dn%2fp(2UgAvGcH;P)`kAn&Z9m%
zFs>Ae8}>Ayh27sw+)5V>n~3fhgh6+38k`y$12z9M8w)G*B61oAvOy&O9)D6Y(^LSZ
z=Er9Q#9abbE@W4Sj~CJ;A8=?T>8OGBYCbj-Nuvt;rl)cX*qZYf{H_-#6LH@kZ8NM&
zVSA4EpN;ld?=o6Hq)DRv9&q_+F7<?T4AaoJErSL;Khv$T^_K%@P>#;O{KwJX9XDEm
zU_$p$FyUAF?=X7#lg);Bc|A;b3TR^lMEuch)LgV))okmijCK~S+Gy<PH0JiC3XRSZ
znA`BEcwAs<jG>-z-S9n@he;BT@sC~fZ7yV?ToO&rWf_j#^g`~xDFB}|ORI&jMm$XD
z=etSDkR3K))aoA{q|Pml;!f{{=vXIYISH?gzheA3z^-Vqel1)2B~O)>4B&U0Sn1uF
ztqq^h(47w5G-&#3?4|Ipur@xgq0Fh4ZB8|QCz(-!d8z%@Li#eWDaO{dk*K_Cc#39e
z4Xk7GJMf|S#IW8K$Rz1aDRd?@Q!PZkrcDC4_Y$89!(Qx*ex)NrZu_p&%F-rHyM0Fc
zDFPYlz5jWS_&|z*m7w{r32=Uh_%^I3j*1@<j(~`8vv{z-kNc;5v$@3bv@ygdRCi^C
zzY28?!WY6}AFvUVk#R$HBR26q<`vCNqwQu6)%vn!SyDzb%9$;yUeF)&`<Qd}hb=m<
zyGnpb(hxf44^t4fQq&-4XChrr<8{e|*?u`Z&qXpg82b6B?f@670-$HoHylLc3P*x8
z)#f%2&FSY%f6O}R{M<gapBhMjbyE*GvLpEYfl+Bf?gl&K)FS?5^VoRKGzRB1kKb4I
zG9~JK8QgEWkNro(;ZJTZDeW+H_pT3ab1W&SAvyZ)dvmvZTeIBAQ2LdyBOuS%i-MBV
z<@u(5zl3%~=08k1AX9Rz&0B{4hW(HK!MNlgOX!9@8YA5`8!zm`E_=hIU@>%2>AXY>
zf7cWx?am164pGf7CIoDZ%Dy;bWGSkkHA}5J50x6VE3wo+O4{Cf<>YRO)Qvs`XIsO?
zBHynN0tJPYliiJ{hMo%x*9bOee*c$yW597>9NC9KWwfg%aNvHpI?K%3)!1^oQN>Jx
zV$fxOGMr`OmA1lkk0MOgz;wBBad|<6N9cQXbR5CmG<m+7f^S(x#xn?UD-4sh2rn~?
z{-U+Q*SgdUNq-V3L68lK4(0l<VnW5b5eC6eCK}0RX1HoxA3~rpuyS)nuu=Q<uKhqM
zZ>C{z@9jXm!m4Xv<n14;Ak!iJ>a%XSe~IP-jJWsppL)AVobVr9U&2+tvhZg1Vr{*X
zW-#7s6hQGDe6TI_9hEWH21EzT86l18Xi^5Zq+5Nv-nw^I%?MDZHUb$U?!xr77nIvP
zd_)yJcT{t9r6q^@2M3su5&eRtK{B;wGLe7cGi(4$$6I%>7trqY@SoDoi5exFXK|VR
z?rI=Z*uTULK>K%j^o7FtCWM)=co?Zh1Dft!VD#=m7`mo0>cmY@pUuJ%MZG|;m`)=I
zbb`P7&mz2pVO_vTN<yoD^TVXZ|NrfVW)jZ90?4%5>2Iu3Q$mk8TG}Jiu&YY_u!;vr
z3}e3Fcd6VVm)3d+y4RF6vT*jRNdI}l8VBD1@6}U+#v`5^d?o-2BhCLs5-DfadP4qI
zZ%(IyXTmc#_w}!Cbun@jMfUXK#wq_6dFABl;PiNRbA5|?=P1hQm&8W(yN)qM(C4O$
zN8{`-;+M(wcWxU}ujiNUO!QcRpKl4k#Yb*jQI+tASN`qv!MKrCK^LWZC@*hqX{%Ww
z6=+xV;@QgQe2;qTuDMk4-=!SbJ*;+r>8&pJ`x~e-JtjC3KgzAjZcHj{O)IU%;Ca^K
zZ}xnVq^i_vGd`UZr&h}Ddsz122KVUAi##}2mXd$0BDT}Y#y^Jf`czJrHw6%`+=Yo(
zHsx|^2A&uaJGYSnq-aU=a%?X4P@NL7^TXFD*dU=YomAssIOGMsvM<=I)}wIcf_RmT
z6*kuAM9R>=o6>VZu{@Tp&}GwWC%YYX;6Kmkz22N(_b)nEAFJ1~j+_lULq0Z-ZkxJ3
zcpaaOBt3J21qc`-cd$o*ci@R!+%XePx5%>@Ykd_=w6X#|J+wY*Lm1)>73xaj9Wj8W
z`&mQv`}ZMvO${x7HK<p`Umm2*sXEKNuYM{haPAj^_Ql6bVW6G1;TgohP#iUL@!1D^
zht+dVn~RdiBpZo9=8j1=_r~jU*HAJw*{(_K&OfK>SVo$-Z@u+>)PR5xWZDZIh<;@X
zYZg~a+im%;=}|XLrI}8UjBdKGM{m<_nzW&sgUr+|*)Ic@IIDIf4fdxxVY>idWCbqD
zWyNX($v@trRc8HyCaV|QO(Kt%6C$upt*iq(s=D76l#1`rW2Yvj?<ISLak=e09tRLh
z8LB~x^3bdMvj!=jq`=j@5u((WrHK?cUS8g_1$r_j7_%(8FMKdmAbPC`?i3?KS@BwH
zMWyj$=imXRIbY?P2=&HM!XdNhXABf^#|=5R;$_vi96~LK-kyZ9r^gaS`%_L{l5AEo
zk6tXFF3}!}fH!J?2PX-}1OE|h&=F!ntKbjkGAPV&n3;}}3HUmVc<>r|Co89*q6rZ$
z%&-e#&RJGp<K}ohIoD+8ocHQur~fY`y!7N8*$ahG$n$MJ^&#TB2x|mK-n-M@tbeV#
zp$V<nGt3w<dZhpH5;$i1-x)cB1oLr|;0Qf<YsVROTwKh4xdgRO+ARm6J91jEh)``1
zlp|$bQWhx;4nTxQK)6B-^N*n~Wmd4s=(tkVx=4Ey)7!DxTz9`yfBi$yBtm{RaQfKg
zTPVOIEO^4}Wp8Dy6I9j^7edHG%NnTv9*7lG42H^Ykd#B9Qf(jLq|WdYx>QTf=yY?s
zw<ohEI>-iIi?>;X4St^ezya<X-EqpY7tC{OpKVA;JP;XKp3xqV7aw065D7JU)XaNi
z;~HML?p?8~e(O6<uNM^#xWjQ4FYx*v>@;-ddL^D62S&VdQ+N@CPl5Nadw2SM;13h+
zP%<|%C$%dZ1V9((Q`R`w#DWDjT%&>uOWor?BK1^!M-x?mGT!c+m}m&fu6O)5&)wi~
zbA=cc4FqG@4UB4d3}AEe6U1b8g6R`|GZ&t<`kpQIx)L*=WH#`^lXglI7KlR1?l;u6
zobn}%0;5Njmf#RWRK@sAB=CgGl~|NjyV|_)bwg_mhav#_soEQcK;Gf!bSPV0F4VuC
zbrA`5b$LR5S|e=LR*h<9`Bl~*1A~I3uWC`rWWX+Fk3T8Dxr>bHyZ`QC)dqzCrSRJe
zEQIgoi-eLO`U)qk8y2ffQ1FdhescR2h5$b|iAc0Br!CW9k}#F-*R53P#{7WX7o@rh
z$s=D1zO<k;LRfwR0{ZeN#%)Y2oM0r8q3N~3nTfd}VH%FFd|XU|O~u};7k$)IZU&bm
zp1{@KKiIUW$J1SZMCqEpX`xZMX;{Li@`ViQB1JBdFW{7)u*@*dt)}o{v`Np^dhHS*
zCrQ!_V`vZw2EGiuN_-sbuBVK)qz1BvBrh$i@du@rJ5HiGJR{?%H29cC4n{W>OM(6B
zjE~D7N|Ce&L^!0uELGNHEBX`E>`p$oT0kYp?Jf|}OKxY~1**mzL+=R53~f}To8Wsu
z&7NeFwa2t^e@_VRkqj-?L7jC;jz)HoJY`#^GlO2UQP11+?coG@;-|_W1q~f7tQ>=s
z8P0+$h>mU0N#mhY=PTDII_r1(+$L;cIhdfE9CKO8@J%|oJT<AD^p&9yEQV+Z4?vLa
zulmF!?#(Ne5LHk~Bm6g+fA;HW*S}R3dta0RkqCK9ZipGI(z%{(24bn!0FH!sbCP{<
z7<U|3o%Jz#f_Ag6qMZgd=HdLD=r_8u*2&6fM}>Z$ol)Q0;^xQG+u?h(;DAGAnztd|
zdTn(8;#Zqs1!j+#lTqt@GjegleIOe$i;sRp@exVq+>0xhF#LjAimw;s>0q%<{;vAZ
zFA9cT4)^mYacyUm2v&DeuhsYTW3K7Wcim12wkl)AP(Pjs_sq7SjC7neowi#k7;nss
zET7bTTspZmF`zXo<m<TT;Z#nGR-_$#oPZqtFN_y~Cr#4mx8Rg!$U=BD3gDO+5k2}^
zK6R|fXb(Y5)bzP{57srYvXU5G(fD}ao>BjiRR+2_4cM^T)W<r_pA`Q-_M|VZO+AI-
zYg{d)u3{c}0{?;b5_5pw)jw|t=XNghQy0XQ&&qvy!_?l3MxNcWLU8!$e-2Cdc9*P)
zePk3yzo@}QDB#;zXZLh%1;AWQg;Zwpsy8_xF9j3OML13xkIwEbcJY%53*vhkRzF!w
zIew@4+gKxfo4m}RCl()`2IIN%C5Qu4-uhrFjqBO&{CX0E?X0;qY;+VikkTtI%B+DY
z_4J$bS2Y$~EGR2YuVN|YN&L=(D?SB>n8CEY9|}*{^O<8_m0!8-45(&0FlZc;K!4al
zagLGSbmKoIY%LlZ42=qM7@J-vWrixcI)kP$;oxt#+oy-=Zkjk1S~y46rBpqF$+X>{
z5g6)ZzRbm&{(U0Rc$~)VWc?F-<|kb=j*w&|C@uE_nz2`jYBNyBlE0MA@-lu8%?TEE
zs@AVxGbplZs2%1s1Al1cL_4)~Rpbc?;qq0;jCP2+8XuPaNh4+9Xiv&a!up^D+#pId
z**B1Gg`H?gYoHiBrzV7ngPUq0esjne=~Bj_?^nb56YHOHRQKB1y);J&W_m;uw&9We
zJ3nKmG&595|Dx?PpvLWb$5u&tY$NTyMJD$GXPQ2{8W+4@2@sCrWXcQPR<32Dp$f)8
zi_O$(x0qBPj7(>F7J(+#B#(_v`W~N5iQ<W77SM9B=YZrur6P4#-~5imCCW>LxR}#y
zFRT@&n$&AHP<taww=``w6n3Eka~0E?)_5Sn`wL#R0sVnnsyPPS05^R;#|mBu4Dsi5
zetRE-<8+X(K5(}mA5evJQ6Yb~>|+?)osc9P6ig3G?JSk~o)_;V+ZgcrkZGQ{dUe^=
z)xPO=yyt(uZ<Q?Vscj|4oXYKhi-5J~9h@=YXA4H*kg?YOCiZnFp3U6^#jaerQ%H6T
z{VOFIRe&SUEVTs0Yc|MI4!y(?fE7Xf+J@d(cceC(1aReq!l%1_5Thi4<lILogYQf{
z4VmO@+-H0)yR48;LyJlmTFDp|6?rsDbTy3N!W=>TknTqWzi*ssdh>%Y&0m6K@k+_G
zuk!UoHFV_?;?c!fd!!#bM5!Po&O_rx%J+MjgBjWX#(%ddwh1!T9c-N_gMH5|Lg7}g
z&_&<10*dIzxPiPzcAi4CL1YTnRGi>E#=p5KeLGJ^yp&8ug%QDAGsh@81Boq|If_SN
z2`_n@D~kX<ZzwN1sjk(p{J&JpwButv<TO-)2zum*m!>V)uakE|#$ccImuDXnWl7Iy
z5Rq&cU$q($xXCQG;Fqa{m^4w;y+kG@etj~T0sA2?Iz;4h(}7&961-t_{ZJ(9kFey>
z6i6FrcWYlAPozLV6!fu0+m&TU2cyKh4CpOsTi48A8)|iA)I!rLhFJi(n%@vUcm6%k
zX#s)mkEL=%qB5-=wE6slA1UK!J%xYVz3bySa8($Xa|SL(>}Iw0;Qt|^*sfuv2oh-o
z0(YKuA#7;P$`j2gO6r=DtHZObJe0s`&+WSsez$O!&#}dO$e9+&_ouAlQ8@&7<C0d^
z?({DffgYhO!rX}?-B*ib(%s%`VzWmCW~T?Z(UPVNk@p;AR8L1#XzFc2(+eRtiJ3}z
z26NM+%U;I_Xwc^#!#NwBzRqp0Z`Y9bK#t(<k9?ZyE4voyx2A4=mf12Cj<bUwO4Ph7
z+u?r*Bjk05$qS*JYmea2)L?OW)K07^B`3qoPLT3I6*m1JuR%+4$=uu~y&f^-C0~(%
zWR1Bd`>V)Fh7y}r`&)cScZNhQwRmI*^nNyE7O!0JNw5v&?`j(0L@-C$!*r#Vz%hQ$
zbxz2$u_~P>D7y<Co?aCcVphG4-rt|KCK2$%;zOdvyGM>RHz(K5mL#RwY;Hqs9rfER
zC&b6Z(R37IMW~x-*T#f|_HW+qT^z`Y69rv%x*wYa50XLPXdFY@(0lca>jArUY^oSP
zdUzm1?A$ZtNTZ_^ewX^d<dS~`1EYeI=vxf9xO%<*$*D4ih)ihneUM4Vh^$>>3q<u~
z1UgKczLQ{CXtENij0L}#MI(iRpy0PSU62)`L{{`gHC5$x3?$X2mU;)_yH@ECW`D?n
zbkm|W$5x{VaL`0@0(99+3F%tOE=2oRFyZX>MEGrzW=R(oEsn_>4bQvh!0z_OSW(Qy
zszK}5l{tA0_Zg@Oim+~L-XGIp7)&=$8gm-3X;V9rbw#SWb?mqiC27r2)O+Pc6i>bF
zHOi2SZzXpkSs6+bKkrPB{kTY){8Etqc1fw>S_nh+st=AZ({)r^+MWK%#YNiM6eMol
z@O=}}mlV|$OG(bG;mdOB1*Q=ptHS+<T2Ezf*upI($k&F}M+OP+z)d+O^aFNk7c|aG
zHe!=&82aZX;ePy3vh8Ip+`MH^vflwsh?VOeltwofoR23#C@Ms7RIj~5CBl^!R%%ds
zqe75ruWdZq&oS2nXVvSF=>%3XO{D|HfjOzQy4|FOn){eB+Eo=@K*K<_n$F7OM{#Ha
z@fk6@Zy=zem~dy2fnsqBzN~;qu{v1U8~F?}&~sVEO!u$d2U*}iKA}#LthFvUl>VAp
z*!crbc9Fzq6bLaK1HQ4OZ1`PQO+ivdd}?WS;~zho-a;5{Zmu}ka2-dMP`inmsFb`P
z`@(0+Qi6#l+7+Y>fcv{g6~=ak_fHg2;3aKpyYPl!8-Pfm^k`#{uw_?4Hu|#L?;RLM
z`c#v+HxUUP0ZoC;8C4$1flcwWbf{q3o-CmHAG<`RY*ZEjF3hss3&~6=!52!D?25n~
zx=1ZgN%mpzI2W5w_Gw^?2J!$;_%Kgf;`ISBFeg}!BF6>5&q|DtO`Fpa3&l_c-)dvd
zuv}?t;NfSQUREcfTL1AQ@2C_vxtGz7QQu}3Pcr0-&vP?xyj^=EXd*s(66GTOg$vho
zAhcQ6<O5#KKHYtbJNt@l2RCloDNLuOr6v>n?eKA`N8$bKT3q)wA<(*^Uz^$jPXUTm
zDn-U(BL^4og)f9%O+PM0^A)PsCXWdU2fs_GW2arkl4chm&I;?SxUEJ--Fs@+1C18h
zGOR76Fn(<sS(=ne|HfkNB)oHrP}5Qz8=s;6O6&IC2+B3?DMAuNej6m;dbFk;jl8ep
zR{t8Va76x1W^8nXp01>rj-D2c5$2bpHUY;{7{389FucLdXJf<r`-n2HtoP{Cj`vX`
zwwu`~)Sxy??&re+q#@Yn&UWTT<%8NpgXEa_Sp(&g?~g%BrKl$Jt?pMgok2sb-UO>U
zfgcDuR3hau2&hz0t~1HvI=`*9kyPom`nr43URdOo4p~pcUt;<m^7zhlS5lA5O!j^)
z!g;0uRA357xSzumVytHiyGn^UFdp+w#LH^hvBc*#735=M%%l`wsmMTk$WRW4Cqz~{
zFGwO}?2N-P%T#{YL9_A-nsM1kk(My34@mi5n_f{cRB<TvZ>N0HW_B+SPpHhS{O4!a
zL<RWU(XH`knD@VjL?b7^aSPw|CO~83b%!|u%o&rh0eW9SauLutV{d<3md?_uZ20$0
z?Ns*190QjxEoZF5eKq?+qWC86U^A)~pOz$S0?EYr`rF$11pMAF3^4s2IU39Kr50Ez
zvMsBJCWB7i(WOe@^Xx+b%=hu2VT^ojf|H^e#6QokCsaPMSY^-49;c`Iio(w!f=CK1
z2q{xq?xV=P(_{3`$P`qNBJ@zKQ4!*~i|H6D%Ma0hJ*D3=6zp^>JCuDbSvS>=oAgv2
z2PbyBAjKSFw^bBfVy<<$b*c&a6DB!#9KHHO-C>LdLKu%)#VB69Sb8e+<5qxJ#)bV`
z`flJl1BdAoj$-bm-3lI9p?eYvF?bM=5*88=)Q#&*BfzRQ%~9`ogJO`Dp{rAO*<bJx
zS}heV%n$Qn`8v|q#<fow+KNYy$edRMKZ?MDiL*1rwY|idR%m9_1`V>zXtty55Mw1s
z>8`RW9Y({8f(d71(<p61d|uzr!HT~xGG(rI(VCZ(`Ev|Sje`+Cp2*xsNR|Vn<VHqe
z#GOSIpuUe}3@Khqnho}fDGfCLhM<c6as39OLn;_h<o`VTl2+^L<ZFrN3^E|h4fgHf
zsY7ksT7aweJH+|JdAJ`2(?C#k?B)1JL%DTL<3V}fZ|vf}5^rIv%J-A{3)dr#Zxh|^
zKiQ*1u3lPd4P_@|t_DFa1_Se~huix_f}=daKj)Du)*<phdBO4xSOd=~#inCHBHeQ+
z1uN0p+SspMU>_58p?%z5txSaFI@qVS6m{+yD#M?|?tIe9mfA}=eNyEmnQZ`+_#r`)
z_7YE6>Tf@=8LIuze-^}}C=vQ%LqZe{_N#F#dGl!eHVzBf$6}a)2!J}VR`nppJbavg
z*7Qr<U)rgk1x5L$Mz*IrM0n^C{&vSlEwHviBz`6+-Kcs44c4~pT@{U3i+3&MdT&42
zWYb=B=p5sF+at8AD;QlzG9_7@weTbdzuFpWyFk*L8{%52SN*=NG0_$}$Qg|6?iM#s
zFq9Z$^_Cp?xHa+nyaOJdX@>P(xik(3Jid%Enc;aVkzNT9bM1k!f`O`@K6kyL{ZOD3
z&tdr5seGICr(#|tg+iSmX*QONgLv+HMTTNQ*M<Tv;hwo~A(^>V@+xbbNN5+LD)7fG
z$8a8~UhIO?t7J)pzNyHLEi^?S{QbinYM4%l-)8Hf)@zjUEHD%X)p`0B7Km(0Z{oxF
zZ(zwctHp>lJB6X1aNdYD+7u;=Yuf;Wsuc_w*K038?#3F2*ce|!klpKPFt`6F(`|<-
zHBN$W7Q)Tx+FP#<im@U1=P7@0?U9b0jc8s|3E|<CFgv(A;hnM~1lR9%$gh~a0yg$e
zAr3Moh^MJ)0ALmiN6O-06;`V6KACi2rPP{URx4hPS39_;8j&9mB9w8f0uM?Y)g;@4
zTsxO{@#pG+&&;58T2?VA`t@^n>tr97u!cj37Z<glEfi$S74A8PP)8%+`pJCHriQNP
z_kp$KQ*zqp8-~YRPx$e*Fy}HUHZh)RLu|+FJspGZ6JXvoAH^4$6Jn_I{(N3F@uOyp
z@Kgkq{AX5$-S@Yf4xS&n;D@|C4GicBu<KNt)uuhRNqjeAM^QtBF4F?>Sk&c5ICV3X
zO6p)&ExIwT!fn3JW24OeMeZJ7q_`?SNJaEpso2SS7neuDI*P>PmhkP`Ya7FYD-zn2
zjI^BBfbf9u*1(G77wKWi-p$XqD~L2abDb|u<q9AwebQdHRn-rc@xS=r0rabNjPD|w
zjPfX<Hmp<~*9my6HC?7`aY#{Wlke$ZACnUs3q+@?U{779*Y8aMI#F9VgRKf#qB9?p
z2RgL@*6hE+jM^RkMnVv@Io6J+^9EC4+!QZ`0N__jUC>#IYtG|@KE1o_i?&+n_)5;t
zzcA@NU6#A&S9&J+Ng4QDu=!47vf8+u-G1IJmw~X2pSQbZGi}ws{?(eihg)!)KyPRx
zE;Xw-78Off#C|jo&elWg^*SDn9V(9Q)v%AKyU$Wkk;Z(o<d%_AFq!NTC+NCVn^-Vh
z0ae*i8(mIiO%1s`1B)3g-AV(=0*V*PoQEWea-utbOq8S9SlC!Oo9Y{D*yl9<NLDRb
zTMByK{IoS)t<GTFg;a95)Ihco6x(IL9YfRvWvVXW6@QyHu)iBjXl1%v>VW9Qs6_xx
zo3h%M+^n8@jzCCv-@y0fxZIBmxx+7Izzs&Ov0w-x)$`_lEDdA<sgktZk9)Kqy;C;-
zePgA&lcMnqCGUiZ_T2?*LGNv=EfUf#C9<ve6a;~P{Nhc8?X6@#-Le{`BzUFMqn7;&
zssLrJ#O%73=ytfn_K|vd!Q^-an%3PR$gGXVxWyWK<Y--$<`#QNz3Q^yM#6a@K!Qk#
z)sPZ|<v?aU)9ZjVNvbk1>}Y#F`<IpOMl9g*e6c935^(#}Gu<6U&v;*ux;+U~n<3B+
zVZ_8#L>hIh2{492{rIk9Hu=FyCN!yObO-LK;Em~HozNM5`L3(%D~Qc-6I5Wt7h2GB
z=?(tobSC(w1g6BL8TRj+`cw;`b6ocAs#gBf_{5vx*U#;}ZCniO<oML|Y#iutcW*@B
zk1zy9%G2pTpf`QFC{yfTxfPyQ>Z0wiWk5M8Y1fJw3@OVv075;3d1PPpGdPct@c3?z
zMy2<m8k-RJtB6dseXjYBaykg%l)BrAW4X8FL9Cr2en!wu%A=rDLh=LIG_%?2a&h=~
zCwBIqOnW0O0)nw_aEgp88{W^lTXvtA-D-{@psEwAA+kKvCu7_NYrq{x;nR3U%vMFx
zhF$tHwm=5IZ1IoXx=JJ4;HsE{`;3yXYK_(m9YtH|^9oOQ-**JH-_Y1dXYVF;;bn5Q
zNzh9KG;p=bjbTW9#LU1#SY5RnYPfc1sq<JRc@clBr`;xS^Rk=W1$cfQ49v~Md3}%m
zhs})YFzvsbkwPPX_~gN(bv3$X2(?5U2!C>dG<)C&AC9-X^&vRS{`00gQfF?OQlO*x
zL+kL0r@QH#LB^En>dred6PF}NE+gxIT7uIDVpl4o5TFLF>_fH`f7vsHNlsToPrK=J
zY1s#x{;%F0#d386BMJ(1+16QU;k6&N+mZRiLEOTB<`}vX#);_rx|z7`RB<@7fJ1i5
zjImA$D9T4EO~p?1qd&}5>x+&rI}IP(Gsz?}`)$cn*$<h^O~3Y4Q89V7)y;n+Ai}eJ
z^9InJfa#c~`zURH%kchUu%LDvP}{lubWH@H9vJoHAdVML2>h*E;3n*He3aWFA>Vra
zgx|U*9vIhFKF@0%4PniMd}wP7wo<kpT#0IUjd5GUyI3+zy0Bv!5^dvF`jOm(G+9cB
z>+ZQQ34#qA6D2fsqTnTHe%U!wqXHPk1YDQ^0<t)ZQ}8NpnSUs;uqXJ09@pDSqZ2#p
zSNnxF&obZ(AZtI*$o?nu4vmw?K5~Y*>VmCym#ZWOO+A6Q{88L{(bMMVSNXP#8Z=a?
zsB)j=ed{%syi=>K?X*`|zar~&9EJlB7>cEuRNVn5aE-l{2-!F#vnvcO_m#V4qJIQ<
zYO!|25&eXBr#pbBN%v%$*qE+HU&lJ23qTxvwxgLaMx@@n3c|J;D}5vsTK2s^csIYq
zSI?jpv}xlm|7{PhW=v}>E2_=#`42YV_oecr_U$acSj0{DcbEVx-I2*9e_YX1BRg(f
zgELwzqi)s4E6x*mAw<m3RbndIg2^91Wrhm|hS6*t{8OcGX+e;MPY}4B{CnPkx`C9j
z*2uRe^^0})u=d-R=Anvp*Rz$dmz{YqdJSJy@TmHR9_M!Gw;Pyfe~@LXETM00e4euT
z3WF9UZI9muoX6objo&?Vu;8#_v-S4M!FlX)t&<zUz$NGTCCPKwW<Jqhla>MK9)p%^
zVPVCy(~>E<t45$FXbYh_v|$?2jl&`W1&f1Uzbi%g^Zq$xVaw{;y-@u0I7@;qYhWRn
zW2tw12-8;4b2>VUg5|h9x#gA?RMTlJM~lFj%#Pq*Jgm0w+tQnw|7NHQiC0GYy~5n`
z{OF>M<8QU2(%68%w0<bJIkYw~8skxMUFhYw-B|d7Oh3b0N!!x{!@v;lqQG*Fdp%~M
z-5lI+PA&byoBC@6(F9qY@N5a@3(<nssp0DU20I0pjm9sHzMh1lDZ?WuNi4KxD6+E|
z>W-pE&2A!QrTB{=sOM%Gk3Us!yFd<mWNIDnfI1S<TT&tJUOuG+v<E++t8`XeWJF${
z>>CiD1k$gIUgSsdLJe9TC8X>Mzp$&T)m<GhWb-tWv_VVsiVc9mbVBokvEkov_X$CB
zF9O;e4S&JrFygfMLVy3xtgOoD=*GrMP*`zlK|@6#a2D?J?^V%tL+Jl#It!*a+HQ;D
z5Zv7f79bGZoe&5b+%32V*TxC%65QS0-QC^Y2Y0#iepUAmOifpHb@y}5-fK0QdX-cb
zGZO6{{^Bw-z8dLcL9-N)FvJ6lK5>OR>3`uhy_ZoHl){0wZ0t&A5t0c{!papx7=Un4
z)J63cOrFAsOKGjB>9}&`GvM&fKh=DQ$msA*1w-_~$z`1+!m}(Y5qL%!A4EhWhA33c
z{9v_T5}N}8+_<f&uDq+svVV1DcjOEu%f#c>Ee)juQjdk+ALmI}4r~DO&)8F$DX1F@
zEH^>MIPjBO`Shh+bEMq!Zb)$obX$5c>8&LF^uH^;ce;$i6lD^$SWKUp*@oZC8Y+Ba
zLU^{UO)agB4Qwn7_wdn^bXv~0jxCw$pRP!QewBr1B6(EOOo-So7grl}Ot3uWHI(Y-
zqrza~X1R;VpniAMwUGmyRSSYz2){TE`Ac$+wh_b~+WUp))B=tpBRQchZ%l_3&Q7{5
zOZM+;{~Fj1dB5s*{Hbp(v<4NZRBseT;)$xX<<*rDQ&d1>eGVdJz}c+JNtzLD!x}y*
zc1gt4qzXtb%Cc?D^v>)4SWGIydO)-Kg+50`@r@owz(Eo;WjqQv{0w=T!&<4hyr~2Y
zQOGio&I~!Q8Z(wcW`CxFTdzEJjR`W#CP>!cYjK0$fj2hJsVufS11C<kR9*Jgd}SGO
z)10W)@r7J?k&5tEU18uKUAe*F9-`A@^Z+$CjXxB@D9=-v*Cm>8<fWz-7KVm?Xj(?m
z@%<K-S?sM4>J7mC2FCDmU{NBUbHl00Wc?_Ksb&-jkKNJG^e*k{L`J7wp0kc-QCtTy
zEv|U9k}|AndJ_xwO*|CoZtt|0*llio1mn4m7r*wIuWvC2M$JB6Q=Gfytm%{pBc=gU
zQ073}<8SmC&%T1Q@PB(fDd9OOz6oKr;-WMoR^Xz)6WDIDQ7d0JN3lJ`%|7K&(?Pc!
ztkGq}p{Jvw@l7!DiOo9&u@=o+kscOB>?it;?AVXL$ab~gX5#Ub{u==fi#^DD*=zny
zPduU=m`wNEFtvBPl`JbOvtROY%Z7|zOV}4-hHQ_9QK&)ua-!)TM3eQpBMy5DPSf%n
z$Q%3ecL5x9Ul{<S@r(nu^tT7w2Oo^>_n?$>KIe05{<XE4mR3rx|A>$Alv`#5W~B_n
zzM-5(Byp$~zaW^rWCf_$82BvPZVM||W8hFf@n?H{gS8d*g}|fIpw{7~ReBKE9WR6>
zsQ&s`w(OvxZ-AnUV5#;olvnmzbD|S04_@pL0&XIys-Ew+r~#+&@HS~h%+lB2j4?0s
z@`_844@PI|ydCG=0~mkm)fzZ~-$d!YHTZ>1L4lIL-M16H5-fM#80P~9=ZxwMBfq$}
zkT0DaADsPtN`~VlL~vIx`>c%o2#b(}WJ!zm0YM1`Eg~!{7jHW4$VmY!z?WIHZ^@+$
z0(fVfn;$M}J>JwNNLKaqG}^LrFxfBf$x7&@kz6?A57T93ZtIXQ6R{PSeH{kts#zQB
zq@!v=Q@|dfNKz~9qS7x6L$c!dkkFId6;g@-v8$9X_D=4dAV7>I{^bY#)0kFQhqyW>
z^(@VT4WyzJ1JY>YxwVBg=!A^aOCn?ii0#T{eSZ#V5rt_Zrd29Z6GWFgXi*nF*JHo=
zxjfX(45$6ech{tnw)Hn^V|+ust(B@dlkpT^1BLv(_7gQWM}$EJ+Q{BTm5ODk?hgr#
zPX>bV#~3HvEMy$jHwa+!XoOA>RUb5s=&RL#(_4H#`2I@mPUFMug)jk^j5ac$9gKne
zA!e+oYNM-o{5{}pB7$f1wTz>G8R%w!5H6{PI`)L6UmEyaD2Tj8pFoP@#L<RRSwt>>
zBqJw$7}->IoX@o~YOvu+c_eCrm~i+e-ua-+`Zx6-U238Fp>u_A^pe#|$H&q!lA1^G
z_l)+Zh{WoF34Ryb2Pd*k-5)G~U9w0+xF0!zGaG?2u>{rO)Z8o6%j&(=Sg73`h90~5
z1g3^Sa96-u#8K*25%VCQZCWOhs*pQphJn;owc>X7yC7-BJ-1M_*pLI(zZbMx+^`aS
zKDa&;lbD!vh*tGb$}m$0DGR?OgYLv2F<Ry|eD_z}#NVr^bJ(W+G<hb#!5&UI^JZ)&
z;fW!F+~wZi^#$nr-MV?~yaDy_Usp&p8*~<!r}0zAxuW<xFB$kK{9A$v$_p?dxU9HS
zDL?f@KW62TmvuzD%qftFY-0R-j(^epc4p5wYWm})f7Bh91Tr6lf_X*ii7htv03XOd
z#rZ@9BiJ|2^m`JC5=Mss@;|1#Cb(#Ft1KQ0=08C;ti-0xYQXF{mI?I6srh<EG%#+8
z4)ABZ?>K~hIrjJRv{BW>ESmLb;a<{A>B%o)WS|hY8lKoqRN!#F4Aaek{1xQ(6nOe4
zHup7ctRCkQf(bm_ln^sQQuTJf84PRXn&1MJFltt_x}X?<qLVNIaDbk&=N{>pTOUzK
zSW*o3D|TqZ&#;DZ2;4%j@us??p@nth26BXsWD||Q{WCY-PXee^EpeE5;{Lv@M2x*)
zO<bk+Pdm)^*ytmkSzG=*U4A_|JBTOG2g1C1i5(@b4(snmro1AfkJGh_$Xg$ej|We=
z9!G(lo_V~?6TLqG=Wj}`cjoBR9jM<&a>7g@`$<NF{`P$FRIqZ_T|SAolmBT=&;hHR
zlcje~6eA8An|5Q|78>myW~wkWJH-0e;x0CAai(^9zH_8+3ca88?NJ%Yu6825;jfzX
zIDY<xlsUtn1?;TR))GhA@19MSwbUDKLySd5d~<&e3?IflfGE9>=U3g$`~@+TgwxQU
z($b5Sn+gr~POL@yKmX+Nkw#TIg<nOM<)k6=s{BYba3#BH#7L{%LgWa9<!!QS`nyH5
z^I}h1?|sPZHyeX)Hw`hZYfYk~(~=Vt#Dr!b0x7uSmyjecj1Kera~@roAQgjTl+Ryd
zdJW4(wfS>f8=%#-{p0>}Xcx~`kY&tV(C2BR>u%zHU}O!SOUxX#03s7Q2K~&UfTo+n
z1G5|sj?prztLSr^n_px06>ENdt;^rW+`x>`_$|X*6dGnubiqD9>;IKYlPc*;%Ncc2
z6OhZ`dCMZ|6S}|Hae@n`E!F9!2pQ)|s#fcf=VELCW-iopd#9(0)PAHt?WAto3ZLu8
z)hOB|XtQm>dj||N&?@1yEm=};=wPR%AM2Htl#3387O-z4sH`Z+DWhYi?N!+8@BN<q
zfI(QQ33u?F_arpSbNF4tllk(a#~T}lnLrwegFIq+X$9jsTn@ECe;76N+aCDlVjTIL
zpPvtWThmX>$uG4vv$C);Haa}PA|~kM_ZV*pJqd!AT>Yvo(IiBEWP-e<C)feBHlFqz
z3lP<U4`=yH|7zGZAYyQjjln=LHd9P$J)dg|Xf`dp%C5f<;30yTEHyoay+)h2)uEwf
z5XY435?KFyr3+>C?sNY)C9&B-SY}}5x&`p(Yb=}Wt*j194K9qT1oW%$o?D7~N=Vkm
znw~P_o85JIJl6(v>JyfxD2<%13RVOpHJKfqA2$3zZ>y$ga2}PyB8^0?1lEi?@O)7X
za9?)R+i`VCMDd9|zjkxjVr(b=uy7G6hY5x<+h^FC?(LB<GKA`qFVJK@0!d1Q<^sWb
z!ukpb2sPg&0^s1EpIlp7d6!o;w`)}4HM2A|HNi#R-+E&kp9JukZ?$`j;#mJ&{~T#t
z013?bUAJ^}x1xQ|>~MX1$Le*veG8Lxb-56SuI3vYz~15#E(=GhT1?na=ljIVou;7i
zrPP0_B%z#2JG&)|nyRuW2d<MRTM+mhS;IlhPj_V7F8Sj%G-o7;sxZ~rntohXM2*Wp
z79T4K8)wS`<(>BP>ij%fpVm*2w))-Ms{-iO2HRq~b>-Qr&f~r2QT-(rPcr-ld2n~-
zq}l!WwK%c&#d$&1nsHcCGP=<}q<}akV>?wvMjP{IvXo(5!Eqdc$^a(CCmsNY$xF#n
zNx%vAlZfN2C&p&fOOz7AB)ua260PGsYNVWuob1m@ar2)SBRI#Saaf7{;(@_1IV%JG
z(+dj&11l>dBQrAt6Vv_RfE6>#XGY2|CK{fGf4)zw>_n*4FD0CKn!0>Uy@8u%?XGVZ
zy_I#4>tvveDC9d`b6aaaFC4(b+Wz2dP)(#v_LCE*@)o!Q&gf|zAFrZ3Up2%3Tw-Bi
zA<P4#iX>duKT~f=DVoI{2>W;I7)oQJpMl>T98g&wH?2&!wpfbKB;5ra9bjRTrY0x0
z^D@y4jsKFD3YYvLEj2m>UOgdrhJYm@8`BQuwYygTIbk;Z$7v_E0iQDG$BnlbNpKxV
z5V$^izuXcA-Bm;?jl&a`j4qbTNgn>iQa^$~fu>c=f<;}8L%t>{9}Zr8bBtiL8o%vw
z_q^))r`HGLlNm})4xR%!P9@-Dr^nYfR24Nj_6~%OA8BjqX(Mo29&fpjVPI~Z?w+6T
z?;js;Z%2@$MTJHH^5Fj`{B&~C<`)~GvPyUf4<bi#twJL>!({OR?zismekN;XV4H!f
zizC<)$Y<<QNe9^fG;vI6!5i(J*t`=6>FgmXx~min0!GA7pOiyR{;jk;!bs*<XzH~$
z=D$6sTNa|(FbwPBZXI2ak*3Dh7c}t;QE{TvtwJRu*&YL17me%!!0g1q4HtSH*9(|)
z{O8jdo`)^!I*A+d70eE}*AoN(07Bh7Zgy;Oled;oocN<ZzMNvlFf1jO>kO{euy)28
z4Ow9QB(HKuY;}SwP}Tc7C2UDD=mPygrLf2COI2iU0m30nsy2C3Wx-mjHwAXHp*>u{
z9A`C^cpw4W4@2$O+RV{W(a_M}%gd^w<j1`IfB$u4vk+mmpGr=hFM2QhlPWik;17ps
ztN6v~xMr{JkK27qt0+6nRnjsOTIi6J*zN9xmG{2hR4d?UvV3yk56H*nP{z*NW<TsA
z;kjMFH<G}3m5bT&jb&xEtIa`DC6&gU>>3Ji1jvumnx;HJ{R)RyI{T%VP`){tY3X_m
z^#%?i-@`N8vuo&n?|ir87taUtLpE2O;M-<0A-W!_KCY2&1A&|8{jfuMjI&&XkHX0e
zApmmO%V@noRGBdV-j!RE(6@kuGd(RR;j?)DUgx(A4D6&J-JI&ch}^)AlrjXg(Xz}A
z;4cW8)J3;YNv@_L1KczUOg%y4>}G*9kvcb(Pv>^ev)`;}Ng0VicugQUJ~kgsDlEwR
zJR2x;*U&=3_i)B%SM(&l+8m~)AkD@ej_T&vtRy&#TsaL^o{?Uh@~_S`8HqWEH}d=?
z`zCgGf*>v{@+T`IXsAc3YPAau^tl`c5O5N-N+DF=wif%Q+rZN~<^#vJ=RPtra&)Gq
z0_*RL&$y4R80q`r&1pv(z~@KO#*)<GtT%cw@>+)=|C7o5(w&2~(cTV4o}q+iO_1gL
zm`zzPcN1eIE2p{cCdeFIXlo`H_y!kmea1u(+aLcciC4C&P=y;OJJ`y^8f0h;WYIA6
zHS4eRW#GW_CkGjU%w)xWK12tc#g=<$ebv4=8E75`Yg(RdwA<J08_M^GeAk`kQ#s2R
zD97*?$)`p+t=BiI8n5hTt9Q$-O($KlxJW$4F!df?$OPY{oz4W=e8@;nnq9$3fd;3o
zwlT)$TEf3Mb#`1v67Ozte^v2-mH?3zC{QxSs3c*K>VPI){h=<qB`f`~2}HT8fHjIE
zq?y0B)K2JYM7>8;Ml-A3M?Y<kG6ff8JBR@ZTrTI#6x7zsXyhLmoah>wn;F>nl+XQ5
zBfnm9<%KNnV)^5ptGfs$tesY$8#QuQIY;bh&6+`nppl(&X>)XZcwq)WcRu668+-he
z*&>Xu7>m2Tug<byB%_&O2VEulQO&&xELXme`5ubjF-fcPf9lUzn6T6*3*K182=D~Z
zTAEkn!SWol&50HZ<-e-&wu}s`RUO1B6COlS#J#~{FM*?5tq!LX0*`6TdYx|Y_eze1
ziI?&3Z;hL}k!xqX9*%(C`G87@2-wam^Q+<WNHbkcnvoTyMHkJ^S~0TqA0}f0*9tZu
zV|;?S7;2+C3&Xr9w}7tPW@n2ip{AtHeK8yBu=N=v==Nf_H8cFiaq1AKwwGX2vAVeC
zeD!<x{<{9STBM?=<K`ZgEwpc~=DFa$BbY3t`Q{LSb|ZBN{{gJZI$rxr^uI)PzFpC(
z%Yll=^hilD_dx!?jA=c{9rP#Jf7IID%bl8Un=>AT{*}%`mXK8%6!yPPG$y=6;(^VN
zuZys5qf%Oo2TOie*JoXBIG^*$A|n-~MgtwJX;aY7H`{YbTM-wP)fO+7>OuOm)p6FQ
z7$-@*ty4*XmjD(JTHze<0&j@@6H>kfL!fgtj8t!(2Hm$BJY6=9OtfYdZ%Qif$Uzun
zoJzv!daf~P83}mA5d4u%s49-fztEXHO8p;IfglkdrN7y?CG9_n4|ARcN*&wdn=gZD
z%Kcn?In#1o#N`rXOpB}pgHldzk$Oi%PIhzqA>=(UIZ%U_xxr6DaMJ3GamWkp|3E63
z&8i*EF3^z6NIQX+Kct&MdSO$+seYMaNij#^$?cGpbaa&NUA9kNn)TYaQ1&jl)eKWD
z^JPkb3&gwPi!R=e&;%_6cX+c3jqp5y1V$0?FF00Rb@~2>5niwL*z5)i1sXGWqJ%sD
z-~e%<HYn@;E-wsc6j97BB@$>&{-5RtX~h;NIL=GHOhlqNb0{V4WsG9qod_dv4QwD6
zMwhi@*Ox|s<L?D({+D!&H`_t9bv;1JBtx$CE=fXSW}vi%hY<<^q?;FCr?Cb#M}(x-
zq)s0xSc6Q5V!fi)Y5Y<#aIoNz)cR6g0*L&_Ez$T#U|tOF$$^`eDy@+Xn@yt%LXVf8
zd-r!?){+k-Joe!1`#T?1ILEOm7hj~6Rubv<$;kog#+~kl+|g@da31ulKlSef4!hz=
zl;7r0xL-*Hmuh2Uaqu*+^~MK3)V2bv$WrxWOr{swX}@klshOpv`5nJ`v9yDL!Tgct
zhB#U|Vq0@ayeHf$+_NTepAFz<VGG^hA3y2^F!enSBx&oM!K=_};PfY;G2J(o?vS|P
zXns|=OAbQ_z6Gj~X%9nX?wd=Wzvj28uIxe1*S?(|W}&a>%A9w9M;MoN8@}+2p=g_k
zJhKQkdYb-R<uPknCx&_pVjwXE<k-pjE@JeHIi|d>#;W~WEx2;OG?Ig5J6H?5_r%b~
z&V)Le9e;@GKt2(dh!+Cr-yuxbOWI%nOD%X>H5X>G4{~`KNW2}&@PRC=S><@j35Og9
z_;g2SmZBXjbV^fypxNY32vWW(`;N&cTD7LU`h4Zm3+i62Qm@_h{`MPyP46-lvN`Kj
zf3wq%L_%_YIGM?ZbjSzQ?gGKm=|8@ktp#aL@^9?pJC{YVtltCZ`!jFk`C3{gDJ-MP
zNVr)~m(dAM1AS1O-7}@`r|3^mg$ycEqR_#1S-UWPqh-fR-tUW)icGE1`~6OX&NgEq
zH}wnHhv<tnck&~pymk%1F6FcKR%G~0>Zuyv^x~W7VLegPn<vN_G#_Rv&&hRB5&Wf4
z-^tma{~kOHty~{i2n7k|AV*@CF^$t#i0+G+e)y}2cli5&g;Z|Nb$nH3X>09o{#36z
z*5M-aFj2U6#a&f2+3V~FV7+iHiyZki6m)8F*QlSif`aHoPzykH=uXP-E@N_FyV>R*
zQY8s`^Y`c#YqY(8xL<3sZ*P5BsV`k$thRncr$G_&3<#Ys{H^-nOQ9+|V2b(BXnUb2
z<k95(Hu4Ce#sCe27<V=(PJVP(OHlSYLr4?Z5#uel7m%S_ni_0;?EP+!SsL%_p{+YU
zT}ta5p9i~Y83KI8R&D(BgTBnVW6kMK^Wl!D-RsS9jaa-H>C1=@KRdh6NZP&9ASzgE
zXKZAzYd#<Ooh8Xbc+u+iRv;IG`&Hm!*b`poBg(e!f>8>odU1}e`j6esr%(ub!?m};
z<<^{%M8RJ4sI-Y`;*h7V0qFXT&b5Xo@CyYaMv1HL&nG}iQ>WwRDA0FMJ~OU83GIm5
z>*avMMxNjM7X4Yi*X05k1$Cn%f&yT<w!8f%{QlY>mPA)bh33{0Nd(P5@OF1Ouk=!a
zm?+t_jIJ6r_Kmo24RVD8YBf;31$J;6%Uy=Y2ZL5hEAtk*m3aFT-^_fQlS+JC;w7=S
z%4yd=H40EJ&@LcwdEODZ=^lTQRU73>3=-<_h;ShV>Dy2E^I3igWoX!@%(BI&hzf0x
z?*u!Occxor%JRX}G6^jrd6&)LhxBMZqH{rAwdf<uuay<1CZ{JvHQ1HC_?Gm4MS6+j
z8Bqq+sb)#-uwfH-*4WToaY+^3l6eC1_pDbOasY=*9xA3m0)n1QpO5|<)5#1$>XTVT
zxQ#ydArH?UrRLY<0u;FS*+DP>zs;LxZW$`Lv@@c*A@=h~)4yG&7vr%#bVc?mF>1t;
zCX@tm+)2X2PH>n!Y{Zt^<$;Y8(s33yc_7ycCtk4u^g6Xw4&ITRF1GNtzQBPZI$U2)
zfzIO$WNnSd0e99lkLGglXUcW?-AiN||K>8=->-g!%p&(toRkcI4BKVab5+Z1xi_3Z
zR7`e{M<>%M^WSXT)k|UJT`v*z^i&8v9*QkbQ`IpIxg9=A)H5Hka((qEOa7xuk;bT2
z83U{J9=2w~guF%WXKJq;t{?9thym<?AECiph%+gLL$mG9gPu?}XacNZx6v^x(Mck`
zkbZ1r8U5}#b|^w-#u*%4mPkN+hfNTMn&LrjzExeO{}MtpX;U{>7U`&Pn2y+Gnn>1U
zhxbRrv2gupU>D8~`W>gEF8U&inOAUeL#Tam)2U&<E3)Lg)z{9xoNVLqu#X4;P7QDF
zV{xWQW6T4!Q0tsFsYr3$Rmm>2&DP>^F0$z9LKd6a#SBHgbgZwABfVDV>KyH>Cpf&T
z<(T&jCt;(f0+YM^Q>xLaT`cIDk)*W)YgT?XX2wJ`Zg#?Ch-U<N^DlLFr@QY1FSn%K
zkrn|NZ14?x)UYeN&R-5GO&kF#e6rv=iIlJrR~!K(T3bmAp$ybtz^@<Lt6y1~v!3k+
z%N;zp^58YUT*+*!)}T(TI38UOA+$^e*=!OXbxJHjcD_B{&({u8xdhwXulM;MB-vIW
zbI1@|xR9FJSa;4iq_*t}b&~+~*1%m0P802v=lekWVx?}o`_uJ)7#_g<yyBct0H;~Z
z*v+?xdw=mxD#vL8r#Vuy6AP0rp`xQGo%}<w6D8$mQ$Uz@Mc`uvtwh^(z!=vv>T_Qi
zU9o^?#RG-(xp3-vWAz}%IU3Rs)lp}Fm*<-Kmi;PW#US`WlTaCc*8n4iOXEa!Bro<S
z;fUIrPl=?)R}7vh$Ot&aO4n7dFWldWDYq21jRG^}s86HMMyPah_ZY1aj2+Rk_`Tll
z`*&Z!5jsA;DxT-dq1i<;J$~Y|X2*jiUpVD*BeHqsnr9>?kH*9@#uaU#>h2dvz)`5D
z=wP}xrXqJ0Y#Ec?$Ab>j(pl^qnI7or6V}qvAWb{@t*<W52<(OHmCX{$u*PQ{1Qa-Y
zJlj8eI_rMZP|;CpteSf*@hVpfJ_do*y5c6fGFk3FpXhbJ@AJsN4a;j31@QiVn1GJS
zuln{pOm_9>43&1oe!lo)llptu(dfscdC}V=>}MBWMXXtod*?E+0x4FJUbPiurX%H^
z>|gO2c=@df4IhdhFasUmykq|yrk3kstJ~>#--ppO(fWMB?|yeYe`E)!y$n-f@9shN
z@;sg&=|4OrzO9a04s9^ciBml&DzYzaH1TkasOTYTUSifr6SfS%<>4ikAy4IZSNe%G
zywcD~a2V-uxRIlc4{CRV+?DhG3~Q^}P)JDtC|<!nV8$MbB=bf<<cmWU{|BYiOFE>0
zzI^Vx1na{S(bl#eSom>$Or`#hZ?%b9h<->HFKufH532vKxrm-GISM0;=C50;)JOs=
z^;bT9b3@JhlpK^)%ggvFQQ3ffQ#tC?s<o5Ej3n;hK$rOBLLo0i(Y$k7)bkSh&4xzc
zX)M7Pp4r>$W^aQj3x}B3xum`JaG_oC`Oohr3}Op&v81l@*4(K#;<mt=poR4O<Dz7&
zZckc$F2CE9*x1;d9Pnhe8)<u=l2>LI=QMupACZ}}$aJvrbLXjsEM<KtF4F;)#A0NE
zn&Wr-@oqLQI5<!wznkw&%4WyNB>foxykI(h?|ygF-U41AC{4w<JSsBQR5HTk+?_^f
z(~G-i%PabuW~r~4B)vG30(ggp$~0>mvg~y1kDqJS9)%^L-QHn^?7u7rRc<)vkm1nT
z>Wabo|HS7vBs!x>y*({)=E&?6e0^<nnfx9k?EFFJ_y_)i-xIE$xrfC4t}+u4#D`I(
z(2WeDb&H}jXP@7reby|wG}7@vSBC3I2xr8{rW?YH+y{e0;_btnj^vAo_#VX{y8~3e
z|HZ<R0AC|UmwdL2dW4Q>+@j5f!_G*=dpG!?*x~-+Bp_$>!3TZwZT`7YBp(i>CrHrj
zZx<ZE#nv;f9`kDQQ4hX9ae+D&2t-uuCB^qrfx2eT>!68j`X`kH9OL{`#D$6pDFk~C
zwWPg&FS~}+tA9CG<=3Bu6PRojDL-D1jc$AH^?2Q~8mweje*HiKZ`r-?UtaKXUEnM9
z1mTr69jAocBwR1hXPXx%B|*7pKX0o+Ywa$FoHE?Z4PyWBc5N63Wq~G@uZGmGved@d
z@mB}LLTH1OLHRR{BFFA|%t-fDzRfT;dY+55lt=SR@zCm@TAbfU|8UN6VT7Sa>9xIy
zg9N>B^wdsv^(P=m)6qb>Th5314iQ>6N`AzS`FOL4A2LX5N$Hb46V1u#34seV?|&9O
zJEBumgnxP8ZF}*RFaTh>ozkw~dYR*Xceq(~tj=%Ux8rQRW32Pu)kWqEL2uU&n}_Az
zyZ`bX0+h3h)_iM%mwQoBl^I!!&nNwCcr9C5z}1Dz3IL<VQgq*YyypM%Z~>lzwr)5T
zzsXqs6d0lHO7qp#f+FPZ{2SjEnixuaX2F;y5Xk~XQrnyl)Rnd8vMm_X2DE6M3?Uf`
z`~`Wa<Ga~mb_Zde;F87=NX)~_=^Z_H-dFIEmoS{dd*J0gY<Li@O!0}C9*Kb{brcx0
zpj2~;6cvraa;hIHsvmGY!fD_v$7nK(Q}<Ta&s-*(LF)uC>$b2)4QZXuaYc_49uFDU
z3~B5{scp`HeLyGwdd&xkH5BmpJp9JD_%$Q(TKq#OhxtRB1&JFy(_cac?;|3#9+2J-
zPw@DfzSL^p@98r|BLT4+S?6h$*E_%g_zGT(O6qw!Uri0<vsoa;iT)ZT#I+W5zFFS%
ze%nqhJO@A1Vl0-NQ13^Dn!TT1i*w;0cKm2YdSw8q=bKA8^LKLHa+IIVYJCV)B0nso
z-qsodX9sYpgWc4#u?J5D7`0<=Nw}3o2pAJ#yF&u&*^;dIU3b<78;2n8*C&6aFpzb=
zv2|m92sA%*B^iFKJkTV?#BHEIen9iNdpL1HBuc-NYU-fN{o{>FNr-a&z1~(2-%!tF
zZC(a2jZ5u3<h(iQ<RBI@F57Q!@6v)Nv2U^Q&S!D?MWZLL0X_Dh3W!Ao^PO!Zw-t)3
zJzLq271up69fzl$<yqQDZ%4b$l*CVg_8W;MO@N3RtW%Z{wCqY4rEQQi{ii;Z6>Joe
zxBli_B0L!~y+jhS9fMVpU__Z$ABw%db92DAs)vXPx=Jwa{{~zN29wlZwn{;HNDFd$
zu6X7bj-M~Fo)4Q=1zUiZ=LK^PH+%1Uch(JE0i3ZQQLIMwnnr|Ncnfv+?<OjoU)AB`
z3r}{+@`lJ^_RZ+g*z7Qj7sfQ3U-*j>rvy;6XC69TFOL?Mp;Hsb|7j<{M>@fOT?7AY
zLdc<7%;wuCCW47fe*2xrD<jD$Ql1)v?Y&rWAtI;SN)hey{Br-U^;s@*7~&8}V=2L4
zymO|1k1;jb>1yH@2C}nF*Pc=`Ra21v;wqo&$gjMf#T4@jijDdioV}Z@KdNsQF9hQu
z72ooET|Q3wYRxAnSt?(GZmTml+W^PwEx8%=abh(_j8Elt!3r@&8PPnu{w`>s98diY
z5ZDbjWKhteScUT7=u{`tky1_mx!JYn3}@B(Vx`vhqg#xOgvafEW?n0E19-f>*f(lj
zWO8$vIihz^v*vrgdi#ixdL75Z`rx{5`GBf?S&SM!cix`EcupnZ^ZsQq2tX-&Cifw<
zw|Tum`}v7UK<ZYYCJyjg=={iLA>cyqh8&JO!L$gO6_8T;!5@4##o}-u5sy93UdO0W
z>oz%kvk00le|->oe@eoWe^sN+=+$_;o_(0?q?w<_%UtIMtTq=K9u6nd6jBIIK~@DK
z<0-9Q4RzUgJv?jmnK~i^fi5A!v@{Eb(3)t(!{4?>=7>E~CVoE2V>4H55cD0_{v4J<
zYgI-c(`-q@2@i$LHfelq?bgF*^bU#EYc2-|{<c7z)ayEPhW9MR74VeM2Fjyyb^nMf
zJ&Yt<`=p)^9UPU;A0|G1s~|}t?w><~Yp0Y$Nkv5^`WgV$=jAU1oM%DS)T1f2|Kh3w
z?^{Bo#z(XIthprrE&}?0+J`|%Z-FJ*<sJ&!y!s({<tud$OMzN?{CnAcw>3%QucMgq
zNN#<ruTLKMi|e#1=it>Yr`Hs3tG%|ls%O#Eg2J83{HZ0J_2Ce0<xDY-NsppCZloy7
zHwdl+v)iz|g^KS$5S%Jhrh>h@!9Mu$6ESPTKKU=K^qx%(B<TJAcriZ<-&pAZkc#Pc
zQeHVSi<CUy^QldetT#HXUvm&OoC~_1jpd&PFV=^Te!@FKfv9U_;ilmb;tFFE#e#iq
z@<(&Sd%y-IB2<a|w-~gum&rpos#kU|){+n3ZKA?KcLYdJOc>Rh%{N0vGrcz^WNm>1
zn`a=|)#{$RX*_DO(|yJ}3cLO=@@%DQr=8#~H;lCpY{xd?4Hd|A(X~+5sK1e<DrI2=
z%NF}-`XylEXk|6&prOU6uwx2lZ(OA~(F4AeTGGyOH8z}cgOL)upl-B>x$JZzRqG|M
zJuuoBnDo{U5!#g4F(-PYrUc#Fn0o*lv1kjnLUw0s^L(_NOhuDX&TRO3hTxQDT(Uh!
z0=0zOzu3H{p(zO4Dba5;r+I(7-*q*ghf1!XFt~&wdzP4(v=nGyM85)wd3&ldY1MXj
z3;w3Io_|2^y|h)fXKh92yjR+?!EUsRYqR%&M`ML9x0mBX(Ed%KY|)VXHUSEG^%R?Q
z0V{xF^QZ6!Z@(`Iuf$%qNH-cg^{wsg75b7E)Hf9!B!S^0&AI<k4fr+Fl2c;Rg=(C?
zp9a57ta)D_9?F9@JN_IDZIr)6X0$p#OfL#ug?!*+Zf<<LY7=*QIsZv}Z)>f^>jGGe
z#?GfckaW7<ggNjfvWJIb5u#1AIDEoBRGypsCul4xtDzYAnF7vBJl!gP;&z5C1F&?u
zMe)@)MT-+elmA5mBkH&tGJJQyPX7<VO3rP)tl!dwq}7vz026SRD}4dd9y|ZQ3ospY
zdfi{vu99U6x#6dpDEK2otFVYM<pbFQlwPynhGKq58gA_fR*8A_;ik}-95~8Yh-I+2
zA|u1aQHrrKc=)n?8C_cSZHku{{^2U)DRN$Byehxu`Ft9+N#6f3Ct3Ja_v5y^))w$K
z-l_Hi)+P?xb_*t8FPNGsE6#MjwX=7xIwnU(xG0<S?=_buO=R-G(mf9Em;s7aWV}O!
zr{>OU2KtRTWd-@Ag@y8OvqMox3$xu>Ei@+sSVolGy()Dv?%2ecVThFQ6FMDm1B8<u
zkTN*l*B{(hc^MxG{;TBF&#xh3tz4Dvw$G0<jP`%|_Ctr!PB1Ed59;G+ANDj&c4bp3
zbKHe*3(_t&gL9HWC9(Xv`M{Vavf~dK<e|~5?;)W!x;r^7Pp3jH@B82vHwJF4ONCy?
z`|u%JCL~QT&AX~v$l*G|g*I=4$6HOaY&~_JzSXMJ>+a;o;lXyp!3eexk{fj4)?TdC
z&C0#4Mr~fcxu-<O=Wk!7vClx~+cV20eq7AP6aJwm?gex)(>fSNoWRD`vgX$;h~?Xj
zl=Q(#-4^E;-Iq8a$luR_^6x4kA5W9f54Pa8Fj{&pSGfka5kLJQw2tDul>OtDw}4;C
z1^!yhj};z;PXX2ByBlo0vH7rSquJg^Ro!nd$S}YcECvs}6Uy>zzb{|R#J@~brq>$|
z4zEr|tNnS1(^?ejHU+M*r1<IeNV~7M2WT^6fZNwWy=KIt#?FuHbv5!gyL)_kQ#=lN
z0;O0TUhfqTgdV|`;`(B3%$+jJU3zJHLO!Q%!suy;Mc?G1Dt+{2-=wIQou}nq@{6=Z
zHukAQ<R1p7;c#^5>+UAn2Ty}WAS`-<kb));v`lU!Hbh}!N`N!Bv>v6iWDt$Su2~Pb
zyPXDqKMJ;c!3J}ZG5+In(x(6WizoU<MTekj`+wK#n$Zh^DM2TKl%f(XLp3$9LKtfs
z8OEZ%on`U)>_>3{omHf;Z0|_-e3;zuigRX*8xho2!@~+074B){$6JUNS1cvM52MNB
zkRK?&$<~^IKhIMWcMN#rHYo}mtPJ$6zt7^A!_;h-I}OZl`eJYC$h021isIX@?PZx}
zX(O1J$7DiLQ7`XucJ62{jk^RhG%^`Vb0(rYltDvda0!?>3LBMLli{hKex1vRDKRv5
zsh6NtIp0h`4^C)z*sC?PnYxa2UZ(OQj&`rZkXLsCO)iHA^I%>vnNXAE&B}h5rin}i
zK|%uEHIDjKRIBaeq0R|T;EU{(pUFK2{|&Z^wiU(O)`UzA<77w|{=XjP96X;KIF%k0
zOw%OUzF&uP%ScI;P+{?0m@g5K$Uda-@l;z_U8YxyogbIC7YVx_oQ#=#4kM2xJK9p7
z;-cRHgZ{?>7rY8}7KdMVLu0NOcV+a9<rxY(n$iD=84zI>3ZPpMYfCZ?p?pR#v@e2|
zynKRD)JLK1d2y*ZTJOfV#o?M|z8CpXB|;ycu!czqb#A@Q6G@#+dW8o-{n{UfDr)x0
zqe3R)?v7d!>U23ANVqACB3Ws57;Q-iI~=zF%7~P7qiJxTT2@jnzgKv<3xbLuiY=|o
z<tKhB6DcV9k<jYiE7#AA4Ge$kHz6h=wh}JF%(y~`ZV?6P^3x7o$$R6KXn(SRj7!BU
z->5n5npVQA+;~UE7R&%XW|==<b-s#gHWWs!tdfDLm`7LOuiN3yC)8}U%?3T8W)ExN
zG~7P-CsK{D=%Nr<DpFfjR!~t)`mWLNnF2AH-JD?AEpYDx;%H(V%q|@18C#0}Vuuze
z0d4#d-8(dgon{vijvOI7&WN6NxU(%=GHGVdD*ix9NE-yrCJ(L!dHpnHo|(i`4!1S)
z2E8?mx1wdX*afzZ<BNpCF1C2Vm(l^hW}|K2#;(LX8m!H<s$F$&l(+m2^J%}s`ZE_~
zq}1fRvc!gO*g~kFq<#fwr(86N{o^?fFbLhTB`H7wNO0V!GfDnyd|$Gb3W~uWsrit%
z(Bv*?-Z_-^n!3ly&&N#VWL8UVM|1nD4+Nk)FoEm&CJKVo>lx;r8;z{4aTM4(VLsDq
z%&Ptphiw&;?b{*hqb|`$U3iF#gM+(wcyRFB)Fe;pS1jgke>FxnWL~vn6`U4f{CEJo
zcrZHg<|JVik18E^kSJ;-6jyd58NzXF0ai|`Jjz3@*K;on8th?rGZRH_mZ7%TO%D#4
z4Q=IXV3h4T^}mk;o_*Zco_=65*;%-(D_>IZ9tnJ7EH#^-UoSMX5z#~z(@n_AhoYcS
zGS}lUQeeA>qs-954V4~i`3Ra-Wfn5bP%UGJhvYVSxVj=p`DoVF^C%OaSZEi~*`If@
zu*BlH!b*fHb>Xx$WOUO=trQDf+B$wkVth1}C!=bP*1ppdpj{N&IRWtG%ViGpPtVHE
z)3pQhL|`_QrQMx%6{Xzlb!pODTKw@5kEkpy&5KGEgN-VvDbC@-PtpX|x523mu5av+
zV?jhu<$j{QxrY6FQ$9x*M*d?D1#i)m(x*8?NM_<bn%fHk!E$M>e@-``vl93vIhQ7h
z7jSu<N9x=jpJ>5nO@PPQZZffc6mKF!Q%y-V*<UlggA^HmM5k4tDaiDjT>w~~U7Gm=
zS^C08j`M)rFDHPG%=IsMB<1a`D5Q{Hd{@s{FSMEj?4&`O6vQN9RHS_D@4dY!#$sWy
zxY|ZiuSqmEeQ3#_%ZGYGxs*^Yg23hNsV(rl*$KGaZKWKf0O1Ok8ClQKS(X77KDv+J
z5+c&s3qEsE^DfN}r+FT^^QP>TiQFx1lauj9Yw6m9+rdkFBBLS}%x0`_eu@^16!a3Y
zkkfDQ@R|HvJ-Iq0^j5jC4x_@TIo4Pkw~a!BqHqtKK*82?rX2<3oOFO@+Oyf_eMG6I
zR)H+Bz?a}QVBUZ`B;c1c5;OCw<=t6PVw8w<%YocRw*B!MKkjAP-_z6HaJ_af&$m}-
zC@B9eb40t)+|vB)&<8XlIuupP5SKfYFtPm3cx5R~bFa~obXr@%?ZSml#XQJ(t~8pP
z2nH_n+S}bq0PwUwjD3DZpq-l7Ma8=AL`N|>IKvABepUsY^u*iQdQi!+CbrqnJqCpi
z%!JUaDESv@a`SX6Ok<iQ=z<x2JFeFb`O-6N3a!nwq$HLk-LTq|Lu)JDY|3do7FNA!
z)J4d4?<<Ap#8LMi(gnVc8KwJVYwg}wbKC1$3+3u8me=^=Fwd8nye^UVpwYVpfvedL
z{P52}DpN>Q4=h(4{J!s3@93N8DHcUngk~)rV(TW{$oT1*0y1m*ZcE5;CQ2&tY>L?H
z0acg{$Dn1$pq<NiRnV3TyZ+QxwiR@_{t<D!HorZplC|5sUa$s&m=_)PQcA8*9AKs_
zs;TJ?J|df_I-`nD^7B_z8qFH~O!7IZi{Sv24(cjI6*p3a{b>Kujw2xr@##7?e$5nY
zvz$%IzbTY2WBd_;>eyHh0aEGT_J6)R$Jn1qf<x>mwqpdOD2wTjD#8vYSzB%X481%(
z#YK?@W=(tf5H8<>diNTO$jpVsDKnp{<`l6^E7$CI26kzq$b`JbLwSn!GWj8sC#V2@
z%`_fov&n9wLS|j6z>nIfhw@}j=`({L)G@wCyu`Y0O%288<4q>|5z;+t<MWq)k6qut
z88Rn)K|_uT>q{uludav*B>8I{IM_p*Nhw7&OgqgcAHBN}EtIN#_r)TG=DugE={9Hj
zZuSry`?WdUh>}y9k4tR?U0w~Br!E4OVBzcg*)f$zW4T_v>&95n3abTzLH7ZNEhh++
zRm2dkv(^SJ;mX-UX^{X5`M1Y6MC>l=6l~#Az%sDi@GwIboirpZExk2n^$XE5W#dJE
zHRfm1SKT%rC`i*uf7GFQORkDEKDXZ=6-=jUK$F{HM=$M!ZcAj_DK|5l!VsYH3+hP`
zj#-(-4*3TiKA(~KO+YL-+CxOrMbWM=5TKJ*P!Youium#fAEuBE59hws?nBZ>M~QeR
zK1<<#@aYeV09(A@z*lT`t-(1Cj!C_mKkJFs;QGR{%hr@61a;5j(Pj6WhYM(>+Ilp&
zG_m91Vos`Fg{oFqO-A1u`36YAP|&?|gRXV^Cv2eblL+2a&CKalVJq_XxQy0>kXz<x
zGxFhIHVq5(v-Espa;!9K($mo1+V$1!0XWg(u?Q{=G+HmWnNt<Db{ld3(SK-&ux3d5
zYNb*2yFcYl%lSf}&Hm~(B@Lg9NER0vpRy9PMLSWNvBK-%DfJz!5&&Z2SH}G+#hO{;
z)8uuVB&HT)zmUGf;!?0C&Sjtr+#;ACJUIo&az5XDr#3!_Dk0WF+lyPE1~F?@pMYlr
ziw)E2weIsH`ygg8^u5YhRmto8Xr?Dk*`B*b@NTKCz$L0ErJX(P8%&;ri-W_1aQhD<
zjXzME6V1(tX(Z<&A;9Wb-`~<xGyi4D=bLXmh$gzkL|CwGs`%!N9>DO91kwiB;5pF-
zTmCZ%Y`1q7`0QnKa~sq`uP;>Xx(t7+X(~r28+Gr%$|{;|>*bY<t8iJ!e^kDw%)yu}
zSn-*SbE-cFX-{au5egK&RP>i+2IJXP6a#S%okN1$0N;UD1Q0e8m+?eHw#%c))faMX
zf)eV3Vk>NmG0wF-rI9r^OXIa37&|!7U^kga7OXEe7}<MnS(!HOf6|EU<8Cs9q^<Bx
zMo3<4TW;&zMcY1=jnJ$*gl$wk_D&=_Y3A-NjPy2ioQ4hGrBLv=OSFk#xzeSm#+0&P
zmC@KS#G+n?1*lbwP6=ri<%#w93R+5X+|L4}B4$m;{|R}R%<g?MDwUQ@R9y)7Kbtfs
zR!dAG#@RA48m|o4>gXj3hTKiH_*v6OA=>&?A5S5t1@^mGND4X$oB}`5Wu8AZl#IG?
z^z2&DtlHW8kSb?Y_95=A7>)?a(!w$c?%ed%u(^ivBH*rUuEFyO`}GSGfeL?35xEdE
zmr^V?+O52tyq=oazX&RK@zs-k-yO-a^Xb97FEeQ~8UnWLd-|3Vn}|p2l5`3;5$_0=
z#zJJ4NFq^;A@Qz;a^=lcO@9Za^Ycvc!oQH|@KxjXBeE%BDJgEZ<g_%@a7$v6VeFk#
z+U<m&E&&_HO%)Mp*h}D}H`BOClmnwTpTg&0i<D1LMr=Rr+p!4f96y^&c0Jl<c;`y!
z8u0lR>izirki<?RI#34l!mH!RY-j)QIpMpy_;2h&6M3i){clg74TmJ~Av4qkxlMQQ
zbgr?*$Gzlizk)#e3D7H5UmFVCtE%A`(z|D}AtM1H6q#}oYo2j^8}%LnE2q8ypXIh%
zmt<}#DjldnFGHwzrha>S`)Su#cck?;C{e22*jb0J8aSCP(WCNI#^Isi9}<f#!enw=
zdIKK%O_BAc0eP;kWy3~49hACG330}FyT1OG3sn4P-%Z-^{8|yi<@HOsAkf06X!Ye$
zuA3Mb&dVY*Qb%9nn+cmsM->#{FG6i9MZ%V8L+AFi*B<X9Pq1Y@{TxbBBMjdyt>;Ad
z2r(OQJz+A|;<Klt1|7r;ot&ymi({<(^9F<Z_aDlD;Se92_^)!1aY%*mBluzyUfo>_
z%k%!&I(wsk5zx@!`Fg6hgaI<<SdV0iUn4I7i!}q9dY^nHnNkXE2-cIm!&x?K6;jsd
z4{6$QOD%3|8G~PcQv-S1nTHeRgL$D*31LX@j#7}PYvp4fNX~x}99|=C3HzsLy(UJp
z(M>Nd&d$!yPfsrtRB)Jeht<-3hHG2?8VOryMSw-lGJ7U{6=ZGYn-)Qpt_CC3mfyt%
z@PA7^mLHISd7)&*N^I|iGfEYPKNjVJoYIcCQw*G%SwVH7Gy6VNRdpc~ozjwgpvEOz
zM$a?^k<1*YVUqp(f<TMgz-Bf34vFZ?OpkW>(|ORC*dB)#6@~i|MhD|>Vzc5?{Gd6P
zriavy0X>3Ta@PLr4Cn9USp%EOzsTV_fy8<apm+l&RqM;KX3L{w{$4Jhn4F2GG$nnF
z!nrmP&JSU8n=U&X60GQ^!$V-j9gbYc{BY}R=j-kD9U8KT5FT@OYglc0U`$2JOqFl@
zzEjznDCVCpG~lD^h;-zqySUK!-efev8j2Opjp=2T2&4t3G|hCPQ`1VWw9L&L08<?r
zj=Y>iC$1H~;q@aRt!$s&Dea(kf-kCL+v8QmcNNh-If_kPV8jV-(|`}&mHN?&gO|9R
zu_%kYS-gH6)_XcO=Y$?hs$MrfE`u=$EyI!CeTodqr!i_x7}jYC<~f@UTP|N}M^?lo
zel5`ELewtc_WrX@P}gwxM5qHFa8X!fx{phoq`}$|>X7SP#R*&QZkKh;=Ywq<`lV<N
zTqw99EJcbHU1O>bV$D>OyOqm8brq&T+CdT0<S{?^x7sx4b;)#?u6mc<=0o^J%PDJa
ziaMgq&+AZF5KhpY_WDRN(mai?6eDtPY!81%B?fInfrt+_z!_N;EPAvDu0BNs&gZhF
zQME+7GH7|w;?p7MpMrMI0@$E!t}0m^vuXs7=L}~;Eu?Q|c6y-$<o;gtm1JxWSgpTy
z#Wa^xP%_i6jdrVBvkdbs@uBrK#?G7!lfkvRbtQ1OADc3qlC7t@NJdwuc%`}l@DX#}
zD$;7Q-uleHKcJFRoQB%~D;{3O<B}1V{Q7?N1!vuDdI_VHgG$Xc)^Zzm^OBq?Thxi)
zPus-%-hV^C&Ig1e!P7hbK{_JD|Cb`R1G(7?*;d9_9Y@=rl}LzvU=g%X0jWsuh7<Tt
zN?+89>FrZCXiGH*(%@T&Igi91%#AbaqKc>ksC(;HCIV85c2ZFoV4P}tj}X!>c5-Fb
z@Ym2Vz2bUjYh#>JU7qMsSJMIaD0Sc9bEku4=()h0v$06h_dN!OI(7xnFTv<#Y`?JA
z<YFXM(4*b*TCp%3LNMCKn!?Y2tu|5sugrMY2&~|Eyn1w6BkPPPAH|VJmP=zD3ma@*
zJccmUa@7yr-k||tUWoEr7@J|cVITwq@~M=#h;m-_x0q`_gH)d|BRv*}u)m=cT)wUT
z%^fLibI&;CSIFVx$NC6ndg!+A$0I8PV%Psto7S}C__#`gC*ff*gjXbV(7@x%SF<ey
zSt?8Jf7|xCJx+&N?H&K>a{+BR<KEYKz9`U>-SOG;$XXwupoezdeOeNgC09B{i@M0t
zfCV>L&|Mg&^@-_#$~ZVq>^dfHDoFopdT7<*16Igvbit{OZ9!>S5yVPM9%8rJtZ!S2
zK{kbm@4L4kxi>s;fZnnaIDjagE;Ckso^MpN$VvQ4D<)TAbnP)LwfpdRqetk7m*eN!
z(a{b+QhEa5Yqf7l85x2|(f^4{__x$K8HA->*3GmeCagy3@imuPizqpHi?Y;42rF3^
zS>V)-rFr&XIgDJpVbfoxW@YY5gezlr=cq~G7~q4u8YQ0{Wvz9*M2%u=<8uA@NtkZu
z^9-3DvD~9G(KQ?|CY-0}xAO2T<(5(*t}(-<*Z6_r?WTee%a@6E)gh<Q&GRU4bA?h5
zZ)GmmfFMZvW-qDM-)O=65WInRh!H=rg5CMFj=rE>>^vMrk)+hJfU$ti2Q^HR6@pF9
zj;A&M>S3=SMQ`(nzcr!hE!-19{s>G4pD}#6Ael6ocy}p{1+0|WRiyfZ4%sx>OQ&8f
z6-I%=p~t!=7BHuXN;dEv+l~sJ%H4|rT$0`64^iJcG%*DOC_mqRu2T&>CCM8n(GKVk
z*`Pxz`SP&>75rf{-R;`VB~xNGEVQt;_XAK<ZaXF(8>XqPv7ZhSQ(FPdI_o;ijYSpP
z#*zaOr5{cU^K5r9Qlgb~xmxH1zV^*_-vY2?pYnR_OvYK>eu?~oNhTLPPsApGfRD-<
z>+VImRN1GMeB@3s8#)4A`=PDap*6?oq4g=9X4q3dt3|!Py(qO2P9XmKMSJ%|t``J#
z2NMxQnTBgoHXBznDU}P_yC5R|r^;%u@Ir)^*6QHQoPr-TE1Qc+Pvl{56z+|<asYst
zi^I+Pb2bYbGQrq7*m9#Am_bH4E0&+Ti2HHctf2+c%Xp=gcT1(dyuZ$wv8%^ml>Kf4
ziqz8eWw)5Im%wq4;<k>i?arS+qTKQ4r};LfNlY1p1637zKS1Cv87Uz#gp#s`nB+t~
zjIDfUR#nlQ-*nuDPMh<f<f`gKI-ngxata#q@TkqrNKuCsr#bww(_OwaAwxFx?P7zy
zSlOiwjhco)_^rIMu(Vu)7v+yC5`SCv4g2(QUcGesm0tkrQEqPz9<acM5Xpsx7x4I&
zc~3^Ca2(Z%o$m+U>^g{+=nXNjgmuI7F!HnV2DQR|N8U?+PZMbKe1-B810*mMhnM?v
zPYWh>YTT#CLX0(b^j$33AV4C$U9!b5x7h`xkQiP5BCcdn0pUq~L9RTp_gVvxpNbW$
ze$T>r3r<xjb~pDz#8`4eQ}S}9IigEVeCA+sZ70?VB(?bpbt`us2{+Lxu_Y2Z`jdR(
zmR66@F2}&n(8k2jDxemJ9LVl+{7hItWjt0N;D+)2CPw((%WXU$h`>oD@VL5L)T1k?
zba!4jm8Xbo*xJwlbU=iOP|CB#8(TA_hn*BFz)Ip=9AqJu_kRIk2%q;rKX-t2l4ett
z`I?e*NVDTM_7$^5W+bl!nKvhA1};e5aY6dVN49_LFoMtn=JU{jeS0z8J`)pvQ!_f?
zvf{$UFFY~3WXzQ2iKQ7)Uf@p1>OOQ!Pl}I=0g4JbCg-M-gTc+tRF}?Qh7aYi_i&p4
z^}x)=i4Nnz6WZS0KO9VGYe%JXOTq#2YI^4OZhk=TaKxd!nmXqv1$1nMp|Zdmh8VC$
zuUXkO5`}4#CV^@@fIPFnO$ge59T5<}|I_{h1<o#U?mc`^KRINADN>OWoThQPF)=c*
z^Cq?1_?e48T-aEVA#ycQlpCsH>g*q#Ry?-4d{kjhMoK~~3$zdb4KliHmPfqK06@jk
z;UFk!Si|cCf4VbZb_y#eOdDI28ZNZf8BUo!FnNy(@o|GefD$9>aXtfoGEh6_9c`ea
zuB^}jjK!9XK%UivDiW%RKv8r;Pl7G~H3J->NgoB@%8-82?VPF>DTi#xr3f5JK;5dE
zdq(GsnYnD+iLFZ-OB4O9RON>2*my!A-#l&7xT;b48Od=`Je0`Mmb^^xZ9>RHuz^b-
zQfPw)gRZ`Tk+G?{g_X5`TL7sFK{cwWI#=xHps&E$--!~klB2;AkMxx+7>$<&Q(*2w
zy&7RLq?H(moNg^$Lugq<aj7|oKl$As_8%!w_lrZb^p}1j@+#CGot(7}R#?Rwy2lhv
zS-kD+{j*ya)#ZvDHRLF}Wy0*0tu52)YfADYfV!XyH#%S^IijO~a(M;mo|nB_piN^y
zjuks$XhL>5XhzMQR4NH^F;;wMu&SZ0Fhrb*oFbU<ke3GmUwro=OQc}58_M77sQ4+v
zb|CMDx)#*`(W!Z5h(GzgpY<Oqd?7lQML3gj@(hfmKFp+4W=h38(Me9pOc;%N)|*x>
zoHeO>WQJJA(1e?R2=Q`oN-{8b?q7k32$}`@Mx_;xo4#<(&K0x9=0*w4l-?PtVdm@~
zlU_WoG&@n`?P#XeUHx7r>t$dogz3>lL^=1s<NMbQ{Nj0CB`FJznMgB_5;gm<oQVs!
zo^3zBy|q3soP6;kj0B;QvdQz;E?>~xFs=dyGNhxj&{vm#=PLL<eY!|IkJ_N?Uo%o~
z!Pu>`epYMS!l^X{2|+HVyqRA%I*hs58f(fk5EZM}NYH)zKpB+a3s^PYYe#B5sAuC5
zl8{?b*9`Htp7|dl@Q$ZL@)5(Nh%0QUfR!S!<J?D;EqxP98<%W7wS8$*d6J*Cs+_#O
zy`MO@dg{V|mKjsVk1om2NJ0$<cuyGSiBG(g_(BDf84A_b-q9;moKajexw*A@a&-}S
zoAT7w<ctIy>KJM&NN);$7^e=~8x>{Bnnj>uXW>U)>RMR|$kZc7C@APyxs$|K)}wp+
zTYKn#gutpEAdeu0FpOQ39Gsqq+70RV&In`Iu#Ad-#zpH}o2oLy-At6^c;xG}hB^*u
z6&vO+bh0+qQRP1HzIdT`Ogz(xvbF&ODF;U<&*12k+>$YqW;Tqk8kH*%M}RhQSa?KK
zbc`s#Q($YRuOTfFSPgFsTXn#zYHRY0;*RRooW?Xve#w7BtBzq>mTti$@snF1-s*#Y
z+kt(5ycF=jo-dysD#`%Jc3oSqsEpD&u&<Bywj+H!Y3H0sJ2!Gnb!Ay`VO~~xsyHIp
z2a7@GKujw+l6XaVZhg);S#A_t?(8HC2p6a2l~hlxt*I;>S&);Fnw%hx!5J|s1Q_Y4
zE6Ql{a9z6tm`?$)!US}bQ7EQE9RSPEeuIX8fyTKpFe|i?bERSK8kCS%HDw;e+kNtH
zIdF2PrKzbw&Q4Vd=?8WyFAZp*%Ue4QQ8Tvp5+&xA=Ou-EJD8{sWt1hL0_6an8s&(*
zTw&w_s&~T%_3PEUQ{tHuSWp1a+uU57J$yqV<5RN>%gV|tDk>|hs>rngf>z{fcK|tm
z3jkTM0|VxCASLPR<K^k$?&>0-BottIc97*n>_h*dBh<ja#MzFA=&6~xki?H}m=E!m
zAN|`6oYkwSs>0G+hMoFCb4>mTlS>Krb0ZCw;vgwTBRM%GHC2+9mYyt*3<ZiR?~vAy
zRB|sSo*&M{$3}@l0(?B(oWTh?GA>bnl9gXvHhS#DDN{*{P@D<4IT#%2QC%a3x50Kp
z7#JBY3Jne<?~b=SQX}=XrQ%2LFq1H@E5-Sr9+)($m<R&F^L+ut+kW<MHE=t}G0SwM
z)+DU<hPtUg$DSldw3VY2&<=xD{B}i=(Qye$scD(1iLntO0p1=0PF=LW9HWhYv&;eM
z!o@El3BcD;VW9j0L)74~h!~K{s;QseviO7fjWq>{A?}vsAkx4pPD+&Lnb@hxiScnU
zQ4u1Fx&bn@jLQPYWf;WaN<atL5_Ew3C>c8kCgj&lT?FyApZ!}6+{X_ZhQ=iNM*2K-
zwKKDVIaNuuY{{#j3z7<6enFytXmL_nR(eW8Y(!X)AGL{r!9cIRw1Ji;o?RKmx#@|~
zVg8;1S9c)_N(aYfj;ftHf5oQ#Yv)fcO$l~3;N0KIiNy5Ox%(xxFfS)7Lz0q|5EmKZ
z?+Hs}DW_WM0ISSNeE>AaR7ZJ8FGWMgfQ0-p(-uR#?Pve?gDDkv_8EMC)vw|dD9li=
z9#(UiX(<VD(cz&%to!FDC(^_Z)3R_4j?by62Srj8@#6*b+(J`|$G^LL>(Ofm*SuSk
z7G$diHx6AMaG*e`!+lLvc`2pfCdEQw;b3VjL!9P>W)acrD~ivI7x@S*^pppKz=D5#
z{+Q`YA>Qmm{a1+pCjLJsH2$0T|Iqkv;{QYAzlr~ki~j=_hl#n-Y<?2}0000<MNUMn
GLSTYEW_U0F

delta 103067
zcmV*bKvcigp$5C92Cz>9e^{L7wr*p%yUPr20|NsL3^usCySo$ILeP*vNJ5ObKoa8a
z?(Xi5Thk_$y-&~C_gmjDLz>=mnw_3D_qOYK&e>%G!|=bKEP2<vz6t{K^>B5xWt$u6
z>1e1Z$ViBc{Yt^uIV`7P+TgAg(;M?5TuhWkeMQ+gaTz5IT|;xWf4Lz;LrGSA+*hAu
zBC@IsQ)`|_peN79l%XaoD)^@{LSmA#3d-u*>M9Ddl48PR{>S&Raq<lI_wjIZ;@YxV
z#(LUn3esXDK97iuijJ`r*UitJ$6*=iswqf{i-`!28#{W`|NVXcD)u+<e^L7z_`j(A
z4g6o!{s#UpY9V0$e_pPR97|Ju9Sv1QIVpJJ@&??%)bX7IYv*-UrUyA1DhM9cSWzi?
zRV}8Gxe-%K6+h@vqsNRD5*85^6BQ8_8aoC(R(NF`eg(ZGEHrN1*fC?!>!Z-S2umwz
z8CY<fJe+JT47HS{MFgK^jF6~=w49=fx{9Klw1lY8NZ|21f1+Q4-VFZ+^gha3h8DKY
zUhZzrj$B)7D+^;i9d#vH3E|PBiQhy3Z)9oj>dWVGEDUv2<;0PRfr*0sQP22`v~aLM
zZ#PF93nK>LPCRj0CdWIjpmFMwzNW%>A6tF-5x|Q`$||X8F%6m8YD%(_qGP`rJyuv$
zLP|znQC?O`e?nAv+-UrYxRi{Xg1nrpv=rKBq5wZ&|CNxWyqc~Ni^H>J88OrqB!vZ^
zW%M{<F-aMOyrP_pq?ibLAKFHZ8He5x{Sqk&ar7I;!22kuGmKcaPCN&DTaL9A%fc8f
zTxEGFF}%!16N^k)n*iR?#@s+#MOI8)LQ+avMpjN9f9wx@%wL{Gfd%`xIa<S$p`W;n
zj+I+@Mn!9X`}oWVcQ!*-0Q?wXaVdF(ysnm-vb?mo@L2SUn3SBNs)mlHs*;=(yfPNO
zB8#@Jwx+tOGFlJP7<&ZBD0EZEDrqu|EUYbzn3~G6;zEMYGJ33#s5nAi0euQ_QK7LT
zfFC_he*~=wc_kGkc^N5j_=M<vlr@>g791OEww0wN%fj4PUsqFA5#3ZHPX!FTkDG(F
znZA~?jEK0Tw5)=Xs=6lFANZKRJc|Jf_2oOTO_>@Ba?<1zm)0`p1;ppobk^o41Ugx0
z%ZvaX4rF;HRSi`H0zMwl;V-A6p=)4bz|d5Ye@BP^sB!2OWepvDV?#ZLwuYKAdSfXG
zF=2RTxG7XLbq!1n;hUvIM_kFownosvr$F0!1n_umDXD7el51-W`92IoGs4Wx5%3Hx
zH6^qb#Q=Ih9s|#^b3wqfP4zUDr0{w~D^$-A><@g*U!KK-h5Pdztc-P4&_OCj9Hf%!
ze@3<*VabJcMJeH)cE%b~f}g+#JM0?;S!o0&+I&J1aw=MSCYH8Vruy2d@{+=%&@0MX
zOk<Xfm4&IHo{pxP;G5BbAuX@0rp3@wCl3tTMuDp!DkdQbpF#+Z2inGpO3Ep#>*$-n
zaV{q%I+lDNb#11ezJZ~Uv5Bdfu>k|Qe}#&&a2tr=9|zzK5%AuwT($`l-7Dk7q!qA#
z!Up@}p7IxHNnnuy?hclQT8ij_iwI$-S3*V4(#ba}ttdUp&)G^}RZ{R1)V`FV_oc9u
zqK2L++krSxm83;RiJ(`QCRX+?j&{~8Q{)W@{55oFi%TKjT~%2@77pzZ9vtDoe_<z)
zI;<&!P(oeT(2Q+nsH>qQgF*@Febh9xbQpRDMkWY&CR$?XZ3Q2U0N&Z#)!xd4p(Zae
zRt!ZXXhCv#V1L|G{sJu(EIQD`fn}f}C%7HO6m?8(-2$RgqXYPMrn*Yvf}cR}fO{<l
zw>EL2ipZ$yny?++{XKZrC_<AHe;X|-tEvlel&_~N&yEfLwgzlPh!=iwo{zl9;jOXn
zE=0IzBmPB5LS9wJ(A?U=#@tX>O<qz+@IDB8O&z9zk*TqPo~{<UfH3kRaIFx)BhTN(
z-qKiC6@4%<Sydf$_&a-n{ZY^Oi?j@|*kBI_3q4h7c-+_>{z_C%L!V{OfA^0J;Bzev
zG~~rb{KzN*ALKlbg>OgSEU#f;>EIq173AS)g$@q!(PHQomc5&QM6jQS3)hBaEXWX0
z3rAG2%077O(W4>qBR>Xh!ct0_`exRS?#MmY*Hn@grsF$N2^j?y4Q-|&0MF1t-lX7K
zp+i1rJC>1-imd1;F*$WTe{+trH;!Wei0AvoS{7J*h$q*ap)4VI*0ewDiuQ_*ij2sA
z^A>PCnAp1cqvIivZ*Rs>mK-giq+`Ny@ry``MWLdrBL|Wz>Pm7D`~7={Q!#7@zQ4Nz
z+eBAYcEsly2VNd>ElrIL^_jZb1oFfsLgWl=oxGfFS%%t>GZ2;6e>7w}dIraV{hOcV
z{}02?U8uK{wUL&>h;1!QUD90S4r?e%jfl319V;p(j?Q`!*tX-ORhi~?ZvK(UQ2~5=
zb0#`DBo(y`tsLC_BSHh=w70i`?2x*m%*eFb$LSh4y+y=Our4DbD=RA}CnpcyCUI!P
zjyJb)bmwu*4YXj#f71r>p$5aq+}hII#E`h<Asj?MHj2J+_>LS4108hM%d6{|ah$z^
zaWM10|8&1#!)Kniv#mK(U2a4w3mkQAeN!ttwwVFpsMG(1(<}%vWhE4#%RqKUUPIpk
zT?-M35&mv==1esiVM*lep@W(4=i}wUcXj4*?ZDwzk&_VmfACjm!LOkIX$%qlYv}0c
zG8jxf<gSx}`B)<DV`yQ60K}=+5x~<KA$uE^Df|S&9j}*hW5ngvnPxUT4@gjPAVEd0
zX0O0Vus`7Oe!1oe=45MO@L~28#XX8@D0)JVQb!$m(Lx_2aHJuj4#%pRijur6z7@E9
zzwr1lUspSGe?7E<q!bWL=GI(SWG-ON$gO66cyfrM>(;`~)i>PF&7P&NftDThOD(O~
z);1hlJ9{o#b~>u^XieY(0c>jo-!IZW)Oqgd>R>}8h45;G?6HKRrU8WUq5fWOJQQpi
z=opw<+Vb4I0>Z%lfXDmg8W+sk9G<V_h^I;9Wwn_`f9P&TakJRCQM9v#;n7BG!rX+6
zpVXNq==${wjScm2u`}1#P(UjK=T<B^c3ej%7gxTA;FhHgM~r?^X~pT$WW-3_FRpHU
z4^J;2U%vo%M{D9nA4^^21|Ow;sA^3RMC&+BBPl`NLQ-f+;u|N#-<$8u<ye}stnD0K
z+`avSf5HBc2mGa)IT&4Tp^Z9qwNZE^CyhL%Q8e;Gk_yDp%H>!QX;Ktt**NljL!yGw
zonc|1sR+qc6l3Wbnpm)H9h~`Ieu17&zpTYcYgsr2B^K6CX(&n#c4q0wAPCi%=C&^0
zq2UoxF|qN9Nhy*3uC^vRisDqdMvI~SQQ8Oje`^L-4jw@<A>J;w$mLgnI*h0&ATNWL
z1iH~9Lj(Q1_%4oU9Xh-6J-vMc!2Xyg{Dm3=?BgUFRWDIigxt8O@YvBb^0dS6;o@j(
zWrh|FB<*;7pP-09PXHcmY*{%4B^5}m7@#}c$=x^jSHeO$eLKIzl9v9JGh0iO{p}6p
ze~IY86-5W>nc2Ddg~g@iS#g1`)&^>lf^-dJf5ks2+=(e_6F(Il<lz9nLj(+$-yn8F
zmq}PcY*ct?kiU;7-`yRZC0^b>zF>dQBmOc?84PmtIPj$nr$@lSCa%kIIKC&4M~_o~
zM2v1>%tTkOy%XOjFw7s<T6I*xC6||1e^5|V1XtY2BOoFn%*WZrSX*%f@QS9cQH3qD
z*B)6nx3wsmZ>Ee6e|;+#|LBa;^2+MkhNcOv?NzyP{*LBa(ony_{=OXT=hG&x%&>6u
z3QNpOj_`G68*0do6GGuVLS9TvK@;)|0Z}PQ32`xz;i18S0sek|zCPYwe<yGxf57NW
zt-$E|H(5;=6uDuCaa{c)Q{p0meB2;LH?!h6IHQX*#1}<d*z1;)m64W$yH-KVh|Ti~
ziA|3Ua<@18AOomk<(pRDzxL>Z<LhQMX82pHeI=z~V(%T5QHBujn%vVjW9FRB>g*^F
zYlhs20zXyq7yJdP`fS&L*v#_mf4Cq%$4EnlOzen>ODJj^<KLc<mYR~3m=G6>(GT$R
z_H+mPgC6n!qXrv&G&T}=ab+fp=M#}qkd+K&Jp{FtjXm<1&_(V;Momgkm`6v*7_^C*
zEJyF~q`c&CA0A6jMMCh#>UQBplUE$N_n!}rtn4X?;A(y)t<7=?NhxXWf15FD?)*hd
z2Uo71(_Wq$z%^7D`9Vf~nA0QcNhyW(MXBLlxSm88dz5ueIWE58iP_m%S(%v`8R_ZC
z@zLSH*@1Aw^^bdq|8Ll+kw#|EY3j?Uuf&xY=3LK^gzVI~@BmL|J4<6TOXMp%^F4jt
zop4Z&B58_#A*W$r#q$b{f6s)o-Wr|t;`EkPW&39~&)InT>6y&~Em?s!>QD=I3QQ=d
z>z>j(b=r*nS##!2s>zA=v|-A@2?Tq796o+<o|rxZl=ZBf{bMrAn#(exe7L5X(&(SY
z!bK%6sjOpc?F8pdW>!vaUO{1TNkL9VN<vIT$luQ0FAZ<2F#;pXe`Vs5f?6k8zllzZ
zj|}q0nLR^O<h*m8-0+o;HoUT|tPD13X%$@)Yo2FNbV^JJ3iT`uw17w<0nXxAO6DGM
zWnGK5o!h>+yCT8ULIoZ5rd;3XtcupwwzjtR_KuE@n!NY`9^CO1yGyWcFlz%{`ii<}
z8HOe0RZbkA731e%e?|c>f<Ckoisl?W0;7`CvvLbc$}4N?YsQxr=4PfQ#)AFX&-zO;
z4EtE9z>FJ<4Z#j2SUCYlLk!*9)=oZQ2|(O&0OAhAz}Vc1W6uNPo(X0=Vs?{)qLQ+<
zfjP&?!#^DP1Kw^>S`*Ov6pVNwnKe@uZ(G`1mmTV4tcVV6e`8zE@RXv4`nuYh>Z;1|
z6*=)iZZ?L<t4Gczcz2S(=Mce+0nE&VfD=|>Oh$2Sdo}U|$yI{S0(8e~L8-_$EIuVO
zuXucIQ)^e-ga(9tQ64T){MirtOEU6k7Z~``vhs=$N0F5esumX=0_0T-LtQmZU40`n
zOLWnr=+00Ve{FbhzcIlHXMJ%v7aHKrcjDrtss@pPk!9F;#^lvaUf5HgAM4H0lS6Kr
zfwfyueD?VA(&ECroUDx0=m0l6Qyt_If>)1gu=M)F)C>(BeG?lO--zVgipEKGBv}Xp
z59gGGf`-1it+Q89bW&zPS?z?5p8jb)liFJvYb(pZfBx);{UsUY%#$Weq~%rAGzB}A
zOtbieLh!~iW~h@@QI<6@w24#@k-^c`Geqlw>%{l+2Ups|&6&rwMQ4H*&df_`nmYK!
zW>>aX<&ZHO<nq96EX>P7URX>-sDOfv<9PI@s=wsiv5@x`mXK9aGjU2|QF5>gB!cKN
zlDs<8f7II1-9J1bqp-56vv1DQg#)vu_fGC?Z3g?ZANChyw8unKkfDI4s-ex$XK1P_
z%0j(FL<U{lYzKFLPn>|)RKh7Wrh$nCFfO#!(02)o%P6R7G7L;vY>vI7vzv#vF9gq?
z9`1ZMy4**Qd?Hhe(lE&fIsD>Ex@Pts!Esrse+e<+LB1X?_-??oi!nl?XfY|PL55gH
zQcP%!AiwJy7MET)zOiHKgwoVdS4#r;v6w0%r>1L+zQa2>Hlw(vwRg_4b;B!`Et)rb
zM&A^$Kl^EaK}H{v5)$IrFnv~2N6$#WjilKP$gwli*HV!M8l;*Q10iorGUCRG$tbGp
ze;Sxs*>N2lot$0WsC=29pTD~!Sy0DZ4=z6-3V2@zn(|`Ed(bw<2X|~(fR`(v&eQ@1
zw+P9wgWp0^dx>TV2flxJe0qL)-Nf!06U)=X-K=zF1w4OQRUJbM4$s{`I-{h%Yvz)*
zySHy%zh>p~#q;NY{n=0Z3o<GKmXZX6f7}kxTXgl2^JZp1lHEv#9oH2*cB*m`<3y2{
zrHU?kV|{dgD<D`#i%KhL>X};EyHJL9y_au5P-u8mAO$lf#v5DOp@7hYsSam=z<cm@
z`vvQFtcV2q9a;=SguEhN3N%M3AtN8%-<`d)+AA|7fe9|)?#U`^>6@WX>K&P0f7;YL
ze|X37!w2^4*s@{mszI<n`)PkcMmo<@(qK}^2yUyf1zS)xC+TlCPy{7P;|LX4A3FMU
z`JAG+qSKx4=HkqAgp1xUI6NjHCCcB;7W-@PQ;KjQ&{%@^U<SM>sJ7^-%ZmweH8d%k
zuFl~EqKjWdN>*WcO>@VT>2tfPf3u@KteJAOqe3t%Ep2)3;pycQXDr)%_{xPdCypN6
zyK~znus`)t|Jf!CMw1(94h)7J_Pk)nqqAOFP7>{A8D%X4D96$CH?pBZ6KK>xwI4TH
zOioqT1gJ0Gp6+fgPL5nV2WK}=pMc<S;B{Kz9WVGPiYosB@E;OG$jieNe}4=Wc}bBm
z5{lYJY!swNCuQW7RyVeFPisL(y}PxZ0uDzZ=OiM9eg@M5g5q<l+Gi|Yw|CdJ%^TLN
z8d|yt>`!~>UxLB5mZEv=bZtf&r`jdxJKoTgyl+LvN)Wy>-;a;!Z!#j7L<JtVhMF?4
zsKkWFp+JghW`q1`UoQ_ge;1yE9mmdr=fX!$UVw+AwJ8Jcc>1U4#tgbF>w|avFBw5v
zdM3aKGGrh>LPA&y(=Sl078#$KQ#`)DrK_$mDa6fMPf3_INqJQwp!AJM$*-Q+J9lW=
z;spb<`=|9z2K)1#`WIk|U~**XRnVe_fR~g2R-M4B*CKmZp(+a<f2csWvB2<<{Y{E0
zN<_DmD2a-bXP}?AC*Rr8p2N1Wwc|SSke>(Kd+2Ey@l*0ZkwXz0E~BDAQSi>R0SN$i
z$^r7a>Pj+VGRoMCaQ5^Mi%HHZK*xP~c3hwfTVDm2;~^A<whoB+Y#@7=l2=~e);ncV
zXWPW)#=2^-Kkuo3e*vZe_5s~dR9q4PPtsU$91mR!K(iuh^9UU?2;<{ID2gWC^G38t
zqVq!*cd5E~2M729@OCzqR&2Jl4e+UGU*Y3UIhu_1G4M3Kny3b-DUkhBR<=C8Z*Wv%
zdR|F&L29I*6Wc(Y`shM*7Y4BGBI46>OR6g?%1R3JbF(wSfBw9u{skBljG(n3_e#(u
z15Xbg9#I8XP*PD=QXsgMico3w4315U3FO=1N-KRq5H|$6BIF4+mTzc~pBLZR!Io`d
zVQy}2W@ct8@XiD!f2skN1m2Hi;mMBxo^E&s<RJxwNtsNj5+GCq!sAl1i!$Ouyd13z
zHDUD;mzSute}XEI#BkL(AT%;AIWayiCMq&K^lx6h2^b2)h=w#ln+wHPB%(WMSvfhV
zyub~h!Z3p#n-oE%bwu<JQ6O>}V0&_S9{%CM{$4<KvobR^HZnBOV=@F?kMvK8Q)~dB
zcv#rG`9=8iKLVbTLGa;&#sNb^QCU@86MCnN(fRN0e-|2)l$jXe@6NR{(pFL+uC8%o
zL?qF<06Y<n9T#0Q?w;N_%y4ya`n!e73Jk^Q>NrLp0X$iaqe%@IQ559KsOwugdWR)s
zm(jc^L01dee~z{%w5)jr5-ECn8<vR?G&^WuRxa5?|FPD=f78B0(|(}VV{K-jt)i-?
zuCAe}f2FO1A`AyNpWw(uC>GjV8Z&4YbF{F8tP+YwO)Y6-p_9kV)Wp~b?9Y4ZUx3+z
zvCZ@$6EmVAO@^w1L5-FOk8p7oZGA0c8`prC^y1pWlrT?QV|8exC2F~2$+KHlUX5XF
z#dY-wjt&d-ad)z}HaF7O)xw>uf*xl2r#|f4e<E^u9{~>?Fg&l2xbQ$P7fjVwS65S2
zRY5lpktsr#cn~@ba47&6l*EL`juN7pV@;u78``VT4=5`sDH8T4J@PNZa9)ijy?+S&
zhtdQfasj@W4q5+h8J`v73-#|0dMyZP3{7aKKnHLvLIZAPGXtFJmlPfM->J{w%pKP~
ze?XAkqmAHu`9-G0hWUFyO;8}YprEd6WNGW<;ZK+Uh%R~1XTWXxMA}DHL0Vks1CGgG
zTL=g9>gm1(DuE*{LmA<vqg_g8<e}UZk&sbPk(&_WVP~Qx2bKFVxYJPxbD(wfjLfZU
z9bDXf0-%G+4LYdI454f%BQ7Lh>(T@%f9!<AufT0m925k8tcyg_e&TadVnTi0xNH*!
z9cDn04%2*Hcp!mfW=wR@si3PKG&vahMy4#<i=m}>{_=(u0_di58Vu~nt7~fGdl{au
z1X%=iboKNN42_CQO3NbHR4AS!f@Yy{bfwgR=j;Ma5#ByT{>GkT4Fy~9=Eu^Vf3S4#
zoSF*Sx^gnobh_nN8sw-_UU61ZRFJ0=$6OCx9jfSqNutvnw|Ce$*xTB$Em`PTH!?Cb
zG%zsG*EcXSHZ`}fw6eA^(brK|mXj2rrFj1GW(r1=BXA`fd^9psYr)2f7Ep^<)rD}%
zJ2)B{ZaItz@o}}s?J;=I0{qg-f7*~u^9&5|_x17ebmzOeI6FJ>9336F)&wU@9D1t>
zWg=4KLXW(D(GEocnjUa@1l3sDNRCEMWno4FM0=J7OdZ_oAwjhKphJdjWo5}?p}5h^
zf@NvNhMWOyY$zN;yDlxo^OrY0FuHmDqw)sPcx0ffsiJFQ<K!J0mtL5gf0dRL7Zn!d
z=izM6#@Z*CObv-X1x-WTC!dlK9UkNdhrB)4!I9_W>;k1c1g3%%xbTDzPiJ9mtSn3n
z(GFHa2aKMfv8jb6H0&wLj8xF0Sy|<IsWCyGJZocpxXK|g6sPVW6H_xYa|;&B(%P0F
z0r>fN!{T#xa<pTc<3?>+e_D#?FK;+=M`z0^<b`EabdA}L?g0_OzV1%gRW|0h`bVb~
z)wDK@hh69IPO6-sMFY7V0AEs7kLBPMmY7pfn3)(A2>o(aHnv=z3j)CvITCtW$}$2y
z1iG$IWOz&rphcFchYn~#y^c1Thgn;iofz)xVvF@vl%T5*T5QmTf1#zz)JLnv+>&i$
z51gNnh}fjK=m_YYa(5+=M|+;uO!&(ij%#S<D^|OJ=nY5fo*{9W$+$RfZKe+#=B)DO
zDRX-#R^=y#dUDN8K#72%qos)gD~!F2riq<LSaLy4dwp4ULZ}zd%D}|J+K%Vy;R`{S
zrID_hyda#zc{)hfe+0wXfRG_GJla~g$S3(hpA^Bo&adm9P*so?9q4XP;0RI+n2ZM$
zKMifBfr+`Ljh(ZHUr01$aSO835(zF32zSCy{EBqqU(|@|97Ua`V`~{rV_Wx-#N5iF
zti(uMQRDhVWsYy1zGVI4z6qs3ZqwG%($?0|(bd%j>{X#cf5WhJ2}&xcn>1%~Q)zmn
zH`h!H3k~so0>cA5oH%Csnu-EZQA$T6Eg?G8-_zBB!vgst=+cy@MQmu3(Xn(5O)Z@;
zt)rnVJ3iFQ5$+!iaMwnY?x?hkq8fzkR<=O0BUm{Nl_j}p36a4*?$G9EZltG8OY!{W
zjjYFyP@}<Ze-tv>7QBGC-0JqmihQh%WaSo`R@U0TeCx5z3%jdQ1G)Oh@mE&W(9+d2
zAQebdlnm^=V{&UJ4GhoguFH$@u`^J@`Uvj+VR2!8e6FRDjxtCcf@T30*0OT;i%KiW
zO;3mlhVCn(*sdUrRh2&MlakS8yGLeLcFgH%sVYc`fAn<*C0$KrSqWi5%ZHSLnvMZf
zWPHNnGmEQRx@Y#Zp?fJ4tz3k>14dpT#q+oQs-XwXkI1ZSpVc$5x-bo-uYBSQo2L(M
zKY8cqn*N4#e`~eT<HV#1RtzORqhjh5oI1W^&g#9ZXSbFn_;7Tk6rrU7H&g_BJKLBt
z)hI18e~{8Mw(|^6EpDtV&PgTmGFaqvq<8{tvP>JV*u45Fi)M5-lxD;Ry4kRdK?6bJ
zLxlvW;5qOEV>62D+WY3MSlHj)QdN`@9}?hAA^&+&JpUgVKI$>$f>g92b{eg0*@Ob9
zxDan=j%Q?6W#^ogJ5Oz0*i{zqVWB*V%!}!pe;66)F>of$5O#dcq`9kr1{LXH&5$KJ
zX#%5?qreq6(Nz)bcu6%w8-8$NUUg}1T0Ask37FS(!xT2??oeQ|-NI5!ntI#nO0qz+
zhtkd#BGZMkn2KqDl@C0FW7CVOTPDxy?QE_=?;aNx2$Wsi(@p8`{^bpO^=hgr%8Iz^
ze@azIDO9RT&~6L`?xSmPQbB$9+?CswO|8!kb2e5SB?!dznL1h;s>+H?E0@5;{F;vb
z&f0>+AXhdMq;PDUJOd*WBLdtVpmIju@%X5Rrs$ZA(!BHp=&ZM6nds}NVSyyN2MYJ{
z0^3&^)H9;fOB%)(WW<O0xp7G~B?LSie>|#Cn>UB}9aL1D;E2gBt7~nlttigTNQ#Bn
z(iMO={5W3ts~RrL!j?tQsR`N@#Fcc-xgH_$S*h_*{BY#3cs`M7B@L5iEt@`}IKkIW
zUw#ypOHtQmqMs%ceA=dV?jZ>|70s1-$>H8SOMOglcJd4ej|=m4vu7D-DoD^?f4#Ue
z)56g^EGa)DAu<R%{dzdNhUKt@M8qVeW##4N<)9ucD+>XDdth{GL1t1k(V~kf`~bWN
zKH+T~-CaSx+=k6Ex8Wi8E-o#vtT;a>Gc74TCNcy8??@yhNIK|W-f(q}V{K_+id<?z
zu^b9zmRx>7WLTgNaZ1|p1EE8$e~s*0(3Qo&BTt%Ps843zWYzUqTn_|Kemd4cLO?T3
zEbTBmie%2piP6^yv`^qSO^kwis+Ea=Gf(SZsNw>gva*tvK87<aE)t9D+gbt7A9{9#
zMu{nCVd=V9T%H4h4mfz80bw!8Nr?&Zaj`McQK5t)zX|4l{|!8tH<*W;f0KiqjitFj
zA{a|anR9p^ZkRELA}zL~2R=aSi>Wqrk{_*$77QHp3UboQ1phG*?Jy4)duwzJ8=6}q
zEgn_p001BWNkl<Z5d1>minlRgsLF`Y5d;Fv*xK1AEHNsW(3%wl*|b)*j*d2p8Z|XR
z8k+^pSoovbT3eb?;`BnJe<alm*e(I+%E(E^Wj#Yo$+F|QdHDwh1&4-3L`KDg1%Tig
z%j9pbo;ED_hNXcZe@0hpXn8TVCCdWSXK+VKSX^pOVwjhs1yfl<K&QtW9ttM%kb`A$
z99+>l;#gUrxZKo|<KW`q7YtWC8%1+cLi7{?st6k$>;vUCXrGXwf7>U>SqO#(2Kst>
zOeUxlo0_w%tkF?!#WFWDF~X!@IRG9<p7FWWopr@&QNBD&CTQscA)LeIxw?D%2Zcq(
z;!zA76bO8*o%C0=bg<;Os8Bf49chVRT3j4+@|b#hdd6tSp}jjo#s~}E0rQQN4wjOT
z#@kfS5LbS%_h7-ce{+Py9bEC223m@e0@?-5qKC{N^wSB{$|-emf-YksD7WbeOa~YE
z#e_!7>v6oG>vlm;Q!!2~;ZXz{TDnXW)Z<p&2+~inFd_x*|4!h1^;Ibtq&6{GMj#P9
zW}L7XFf!y7l|jBr8#<v;sN~=*;73!6@o2-V;Zs^n94h&8fAWe-s*pN2FgCSdb3ib`
zCjhQ^3q1{aaX~$vo)pA)v@yd|1nBw=QQtMPuz~3lR=}4sF~t0CB{>BpAlqr_V752p
zbw?{1JB4J{OkT8kNpF2_1fQirbA@mN0M8>J9Flh!c!3l1m$xRc+KQqa1byg762YTJ
zjTI(qAUX^^e|<w^6Eh2X*wYxYFs6&stPl=RLP?{9BxIE}&{az{%VM27vk$oy1Y;WH
zX>D-zeT4QQEnw=vwYRsk<=7B&Oynelkw+;`j6s3l2_C^{6$_8Jvd;ONPHkD(Sr+GE
zp+bH1QKQ5awT*3@eZmu?Ltre8#Ryr3zr10E02KCQf21VFMqyVRlqm^W49J6$x(KT3
zu<IF2eFN;n1D{cb=$mI6kl`Gb#3Kn?v=lBhQMkDA-LX~<x<w=ix(<*r#N^OX4>7kF
zS`*9<)&w>30%OjGEl&k+#6na+3sD28SHP!baC{R=TW1gNAD%g(Akx)TN$@`4KiUxf
zqYu%ae~mx56!Ij%KIuUbia-RXR9u2KO3GP?&=A3Y8llkig_FCmUVVC6N-~OkVq=1#
z(ygN`iHXnR1g%<LK~Ygj8Cb2rY1L&i7z|zHZtH04qGe!&jsPIN5J_7UjuBNgdPEdu
zJmB47{c18lIChMLk~WqmiVN}Lu|L$#CP&+ue__m#D`$+8>~J~LfgbYoH8j-K)z#EY
zd4Y+=&3#KYFPc)58NxFf!J)*eEf{zdFxpuf>8gVAvK+JrfPKPaB8BwF;Wh)b5on{{
zPmuZ~7r>Yi_<xBj08gQcj){$nPe@2iqU5Iq3MeR$#=KMzv{6w5LbDEoD15Vs?ih}(
ze=P_3{ubt@CW1^j)%b#2WgWP7wl<b#WQBz4)EFnNrpI#d42jE%3-$oD{1Fm(v~vec
zXhY=KGl;Spa>s=rJqykoUWMrTF|gr>CKuIDnzN#RVrh~eC}h$`$@L^g1_E76UlVH3
z3bg(9r$nfKjYX>t$YL7WSPv9CzGQhye}N*S&@?oXhByNJFKY?-fcXc7M#RJ?r9v+X
z&&p6kj+pfYJ#)GgomCE$Dq-hqVoH{+19ZI;#)G>UPGxf=GJj6Wp1OE@@?FS00DT`2
ziD3~TIW1!jKQK14G&3f^m2EI`><yg^HUiM0?5l<+qlt?kmlLl@bjfHl^bJkSe^_kj
zh%BgX?VTskMx%@dJ|Y07rZCm;`Grp}u;2caMv+$^4r3rzw4lrk586WEEZjS|(}+I%
z1snet=>LA`eTHF{t+R)3aAaIcW`0VPzbnT?M^Ow*9q8y8nHlRd@WfAOrjU_^NqyjU
zaiC)@I4Hp1&lj3gK;p#Bb%ZuKe>q3gGtie1yftJ(E=EgJLtRZpiPV!<WUw54qtZ(n
zOVT2|?M+4wx}q`!dQ4pF$81blVnmiG8EVk$55<GAsreQ4Z4k|cU=1AF2;jZL5+eMe
zQlO(Og=@*sf&=!O9}?X#B83hj9N0Qg2E;K_BYg(7@daInc-sqZ{g3hgf4|r-fA{~=
zLct-owFBQTEG{*-6b~tZ><rehF*LQbu{0%X<YMB`O@r`P)-h(|f?HC2Y;<IJNKin4
zzrVj9GG8~Y6*0$zYJUt1^di}I=$6pXM8`8GK4alaWqoUONEg;kqDI>ZJ`_E%0K0Iw
zX^(oIK)JvY$vE8Ox_J17e?%pw=R%)!MOL)00|*?_Mgkv-`I%5D5F;EzhQ2Y_Z+=L$
z%APFpIiW?@!;>^GXbHuMM5t4#{TlRyUf{on{r}=2fk!SHS~+?I$EFuzj)Rkxz8daD
z#oEMn)GRp=8URgbtZ#@*-i5hY=@1LxDWwr$bk7@>5_j|n!X$Vrf8@@~!Rd?#g)y{=
zY>&8_vAs`Bc4b?CTSZ0$pRFq^_)s+Snz%isCB^BXu!P4cuLun^Y-pW|ic3t%$jL7*
ztISOZ=3`Mj+62JE1O*g12vnKCe)B`3tA$vapX=rWEtjF#(3$4Sv$tlM8j+c6MS7|#
z>Fkdcs6~Dv^#7~(e*+dhQ8%=935dxoZ79o%3v#nD(g67h6h}L_db!{*Oip|(I5%T~
zjwh#M?%*4hQQA~jRbHH*ot6|A4~}7CQc`k+AHFEWsW>V;5v>I-UJK|9fn1M)37$U)
z-oK`~Q&4ho)6~VijRkSuwt6G(W)MxsGY!dpdl4asEJ2T&f9TDFn_ANV9VtE`v1#cU
z8JStxIk|aBVcw1`=193*8u&0D7mlf}veX!uodL~swqU>cAwjzV2t7)XvpB-gA50pi
z^g-1KSGgdSh1rMFkcOlNDvupY?fWsGg8cvO82Eh*zo)dOnS*ak*7%9l`AK0O_NLk>
z3O6vbMjI-~f6JMJg6xs4>N46amtcfg&!o2Iy2{dmtW@;ufAQ$ySg_*<MyJIDf@rm}
z#3;}*QPb8lVIl7x=g*{dtoRXGRb6w}%<ZU14RkOZ;bea-Qa)-7vFC`MOz<?}nDPR~
zML}Lpc4m5NGLUaU8GS@_Km!lcmheA)5NH1OheXK+e}H0#t6z9xRyJWdIcbS8VS(O!
zCz>QqrWd5?F&UIbC<*>=M*i1Nf#1!rd#lglx(CIi=8w-$iSTt|8){&=2>?FW`y=3G
z8P*=rIW?08X7zQqHCC4uWas1;7ME32R@Z@K9nVrvg{ruw7iGpn&$@wz9MQGPad08Z
z;>h`Df3V%df!wunPFGECjE}t$<-v(i2fY^VbHp8T($e&3DqNqJr4s$IX{pIc(DfcI
zC?e3ln7pPD8{r-gO+QYME1>4lTH!W}-~5oMg$VD*LqS4zX?c0Y`0<rhRn_H1xfxLE
z2=MX5JzUn77G{`!3~s3g<{t}UNSa?j^`}V_e}9<Ziw}*YN0BlQRFLrdECT!locP`W
zVbO_c*-7ZIwl!m@VDUiE2@C#s$II#2`NCw0p(O(|rnEQJl;szdjjw5FZtdu9EX|1V
zbFySA1MWe{4X7<8-qM_j#Vg#rXia-{1*Wxoct-ie>C0wKYA8vM^5t0&EPX{;N<Gb-
ze@*0ufWoZ-4-uMU7185^pc$HW?3}%Vq9P;0!$Lztf`jPkC-i}VP6Zy|UYeZ{=FMXp
zYT_WAX<%%Id!|1Dcw#ELq7M4Ig5xqv>l>SzCroT@YwwuYR9jhEn46K32qb=(f<dZ1
zL2?7?h{jN1)F<joOeU55f@(9=n~`aee+K`J`B2o0=u1dYQRBEp>Cpkd3-Y9A771CL
znAi{>S34LgBY_*8?OnZtyqq|{$fwF1ItFJ{PhPlb&GH2^dpeq`N-Am^TRJEA%^c`y
zF3*bg<FOd9-vRfwhKl^OSfK22ozdb7jfw{uYgaqyywbOE4@)a;?p@eFxuqJ*e_0S*
z32ilU?ynt>504N!;sAe&B}AU2crcDF&m9F5zCJ$Q-rg`y61^=c#)K0s$iehNxR=zI
zk~DuP@8~l142(X_t0yO}DJU~7c|K8TMYSDW-BWs|PMh98YkF@t0>8exqO=I;J_!Gq
zsEDwTK!0CvPds_e!QKwJ6C4i5f5yhf8rPgGSr#A=g7ra=k>jw<a6J=!DiafUjjBr;
z8t5|!%DEJ^@yGlw2Y)eGem1G#54+JoTSXR+V70V!`8D7bO<W@KoBLPo+rEBiAuyOL
zYMa|9PoJ}J`SASS3FR5lzK-UYSyNKi-ZyDNO-W92bdZ-{a6}9`Qu0f(e-p#uRAH(b
za@<2xiW|D;qI09Ux-36Kz{tQFWiFonVZnj^MBgQxq2_RhY7^;pNg}Ch&bB9$NF$rQ
zi8KH@Gb|lELlEwD?e#cNb+9D19TRrmr^NANh(xiHs)3DLP(of+YhVAYIRo<-Em=0W
zbm6=?{nL88JKI{C8fvS?f0q>(=I3N(pcMeYZWs!7eZ6543JU!o-NAQrb#-xea^gWl
zp#x5$KsJE*!rtD_&eqP(9=+!1$a91xgU$^raz45oF#$km<d~SxFf8iYT342v92Mw+
zWyiHt<)uW(G0a}S0=$yBdt6!P!Y!u{@7lNmZP$r4&7Hk77Ywf1f3kb+!oHUBj3{s9
zF~nw+)^|*sKdZN+xvC^5B{nfNE3desx?w_PKBmPRY8czPhb9-)bo5W_>F#Wu&`?`h
zm=PZeJ_Xjh1i7O0)TD&i=m><r4;eJUC`nwFqsF6|LEhNfhBH!DX&mmLMUEuS2$$Bk
zPwOC&NB1u-DQSLMf98%3FHP*dA~VXGd*?5P`59|BY~Hqc!`k5$gG(39pF6vM`qU}i
zo$V7RG}YHuk1t0@0E&0vFNleWiH?qnibR(Zx|PtS6ciZX?++)Hw~r6ZMe@VP7Fr-^
zb$EKA7k&Nw10XjB@{rK<Pf(*EF9?i<Um4T!>9!IKozXDff5Fil^BsZEB&n#Wk5%M8
zE}bc}ypyY^uGoM5+>u=y2Io!dY;5nHwP@wWU58HZUAJ&*b7@+*2g1Ds;l5zy(s}(o
z?M>CCc}3+_b<M3^Q~KJf3z9=UQRHA^=N^)nS2=O&)TzD5dU|>$Ru`l~>JU!_4U0?5
zBOYExYH~t!e`tWW8&rDmAUt_#aUn5D2oSY&@$`68pxDwRU7|K<#^&PBfzF-zeeHE6
z8L>g`kZU1o2A`CxAA?SMs0?YD^ZeuUYrAF-tzNfr>yF*~4jnywaNnMt+qZ7oux|CL
zq2)^!%^#ROb9!IT<gSjk361qN<I744^Kx@?a`1GNe~k3B)D+~45>&vL=%`32ln~v=
z=!*#OR>Hy~B5~~mvSXp>$n%BfRM2>%uKIDmHlXE`?FiV$MLqR7QSMgMn0Mi^5(*l6
zghs(fb#zs>e@4^n4W}O6I)7y6+NE=PCQhEYXw~Mu$1mPKvv=)+DUC%*!SK90rY~5z
zW!=gpfAji#J6r0TT9FsfKWF}u88BKZz?r3MX73)HkX_!?-2*dd<}F;Ze0FzZSw^&<
z6N}6Dk4VZcshK#Tv96}Fyrd{UGbtv_A3Xt5*-}arYo+0qIZP9QdmFkaWr-3ImgI22
zaYb$0)PbRSeeJbH=`n%iC<Rpo>@WZ3hXl1?e@p_fbPY)<YnirW<F;M<4jw&u_TrVx
z7tfzPed@%qBZv0y-MwSmmW}Jz46j(ebP?83#79|UZEa0;byZbm<@k#7veJ^`qQZiF
zv>38sAwUu$Gb<}An_d+uC_0AVI1ggfq=dNWNErD9BOz6YmQRs!V}1|dAAp_QG_S27
zf6~QBmMEW~cNUYwQ(1ovcr}~g-1bGguDpG6^X#Fm!waXQ6Me;|eW$KIcy;H@?$ra`
zwRv%{&j9!RySA=hwQRwh=~Dpo#e=KXZd@`QRHof+^eoVZjLSp|WXiPJ3zn@~zZKmS
zQ(MZ?BfV^0{Uej}DjGY{*69X8!iL)NfBXzgzPGi+i9{(8N?)Bw-Q%pFKpK@2R*z0A
ztZtn;Ft~AOU}|f1VQLg+`(Ta{*l!13gdBHh?Gc%U0@>Aj4jwyo?(+59_a8rc_~8EC
zJGXD%xON4W09pZX1Z>;9aoy^b%NH-0I}_dilO}a_c6M~Mx3x}enb6eOP+v#Pf8RjM
zptP(EMTRIeq}N1YQAufeMP(IQ@Z@<H8|LTk2;?*(<0&fiD`VC_-To8o!I4!xrI8L=
z0%8hTW%2kGRT6c(@Z!lU4&VRz+sD^W@7=h3&g9w4Htacd?d!KczrKBH=ZaY!m6^qL
z?b87F)5j0(*}iG*%4G|e0O(tHf9^fFVNq{mew?qpA<F@s#OZ}qE!{Kb53bvO;N;~)
zo0d##E=dk{@(EAQtDG=-)>2A<v};0jNp?~scp)DR6&@o5o)?~BAS*>FWMYY|$drPr
zmY%uGH|*IkIJc*zGC!G8&>K<v|82mNMI&3ExPs<sOSc?5d-2-s2TxvpfAjwRyYIez
z`}UhRuU|fY`uO4fySHy%yL{o?=@Ul}?FWtY)u4|*Fn7+J*|TQN?4L2cZ))$9$=#DW
z(JDZTpt-53xp~5bmX?Y53p|pgw!WchLQ89V2l5u{;kd(1>3q;}!$c=ZG2t;ZZ}Bti
zUto_;tesJnz&DhjNBxM%f2b0h2BuIL>L5R#%0G_Lv=1xjT(aldx6kfeJhpqo-~c*G
zH|<5hzxn5@J7@Q;TZA0EhK_0TR&3mJ{NmY@M-J@XzG>a6HS0IQrFii0)}h&5P?mA_
z3_#vnepzkHq-k@P3~$<X=+xc~%Vu`g6{UtJWENF5cTF2uzI4f=e}(hs&7Cu=x4m(E
z0kHMmxK=<86BQhFc!mL7z#w*oXBZG4-Vr7Z=8bQfG-JWaMYDR^>dW)eVnT@8fhx`c
zeFE@=TE4EeM^tw0<oRp&A3J^V+MS2bUcY<y-FM%8i=lt@;@Oi&Uq85qphuhk%qc99
zzIEf;;i197<;#~Xe_gs{@uGzb;IEmD*1+^>XbtpCnbOnK+dFk?AN~TZh^~&VNt37a
zPVJjM1J-*xx+&_==Z?jroN!x!ytEj>P5ewf7YzL4QyXVDr2APa(~a^J6O!;O40P1+
zL@^2If6}%O$(=ZF^O+}iube)(W8L6<^k6sRlm3Sncg`Q$f3jkrXTqfE^M^L<I(+u}
z<@2YHABMHRb^FfUd-oqae01lU#WOm92_2W5ky}(=-7v9h>Z}FJS8v*RVB4Byb9>wC
z%W{gw*H4@TuOeH9ZD4wLOHFA`N;KN^Y$I(&3Bff-aXtn5oV5OFKty6zNlnY789m6$
z7+;*5o(O|<f1GSBjd4u(+kwZ<or<~^%QZBuqHX5T_I*cAUATVt(aUcMdy9~N{ptnU
z^p9ZIqjUe#g|nxQA33;p*Y+(NH>_X3Zr$27YgP}hTDf9q5G?|74GhekJ1{VB-u(Fs
z@E3SR%$z-E?!deS3z3&Fhj0_oX_te056RhXQe;ocf9D~Ofq!y#%lwJ?VGg=97jeV@
zSqF|K^qf)ZVq&`X!C8&{s}J40eCEjBt!sxC&R#scW&g=5_nyClGho;H<#T#wE?lu;
z=b=-V@7%n0>HO&v=-%76|G>dRhmRaRwrA7IMg83qsvyf-Ra4(MVPaR$jJb=J4{z9p
z+=2PCf2K}qZJ5wLsb~7^d5e}0u2{Khc+J{%8&)iu4V%6&BQDs}fu*M=Eud-Bom@2k
z2YL0XItFIeXx)ayq~sJ=HGz9wT#%EAbrO8sow+t(zxg52;hY&SAhD>icj3BidykyH
zbo0TpH{X1NKz|F^BjjJcc=q%OhW_StwC&HGe?EEa@PU21ckSA_bH|SD+qZ4qx@Gew
zv<?KTV8sd)RIOSyjJ_b(#IogsL-1Eakbr!|dHA#g>7NL6`f@DD)AyHc<o_n{PtWgI
z(p47cZbqvh5t9Z`Jm>Dp<8s&p*MRUYnD#;GwY|%BoI81V-;Padh87PjTeEHd$xC;h
ze|+=w&gD}Fwy#}2Zz(>yFWq{0{|<TQAv=!kz>bZ>O9y65Zfm8i6M24f7cO0~dL^YQ
zK4a$Wx$_q;Ub=k6FuEePY~Q(O%bI2Lrgu-QgYqHY7BtrA#tfRLMbE^h=g`|ax%-Ah
zCuZd3<>uz%kSaYn0frp;e%fqVNZ??he{CaspXi+G&RIk2x9vN6_UfI-uU@}-^9=xx
zZhCalBj}&w10Okjw{Bd$eDU0wQzwodJ#ysm;lqax9Xxno|Gs^D_w3%iYbRa@TQ+am
zvUTg$ZQIZn@YN=?DmHG~v>Cm)ZTr^E8`iBEUOBXU-i*nu_`DCL@~XgH`&<bpe+>Nd
zOM8ZTYtjSl=tgC-h04jp9m|#3+7R+r;5h4Z15zqF2i6^g6MWt9(9(rN>yc}A@z&R`
z9^bus?%4hv8&_@EzW2zP%V<ZS^?&cqt?O4WU%GJq+}SgyPn|r1JpB!8RxAaCr%#<S
zsiU<60=n4)3l=Xy%N?)x6)RQ@e+{9lk#Y!+o;tK^<Eq7T`nsCSvts<6EE$R;N%+w8
zV`vV<FHKDRqI<h=RmjsXI3hYWAt^N@D;IRDp^z7cWxIa!L!yNn^{n|g&Aed6`tAEq
zT)6S@#Y+VJH*dfF4&C(#dW8LJz#bj>U*nCB5B&@0&!0bc?%dh4XU?2Pe^1~f+5^Xq
z!D={!3@wQxM~)I-9XxPg|Na994jzP;4;|RIXBP?(H=%tpeNszpX>M|)zniT&Q&mcm
zjJ-Y^_!n395BE10MDytpQ55N&PjIT7op|;(EK}$*Rx;%JCzec@zI+>zwLn4Zs*O7j
zpr>&E+1GckUpfPh{GLN6f6iUK{qWhFSK#at&*1viOBc`vI8Nm})(;OZ8t9+e-3E#8
z-oEMmvj!F|Td{h>mL2F~+`4J~+ToRh!yC5lK7bsl>*tQ`+qMoe#7&SN;uxq1MxK&X
zZZ`NL+E}KM9)SSdF-sdeN8AAvmy(%RQj8)d2&1w6cHpV1j}FT@e;^^Rreh`~tPY>K
zdhgkDgggNM{vG&v_{c}#zoO8`vEHp)w{9Z4apU^+>({Pby?W)!<;#~YUAzbj;=+Xs
z7cXAKU!gD2;y8O2t&wxERL-0_dHm>+LkD(jSh;xi)XwHA&}DbFHq@Xqc%KdY%WH?%
z4|G(d1kxj-sEI>ve|#9Z?&0a_Md%Zu)6|&j8<$_#JutkQyqSj9Y~FS7#QAG?A3eBx
zi@4KIp1XYG-lOMlAl4!7zQ+&m-M#_W9<npoz(IgAPg{FOXIIyx?#WYT%v(0R8Lo)4
zCl2r1xdp9^HJj0rI(PNXquZBHAKtxr^|FEP`ho<1M^kk{f7dNJ)W{Z?`-GqdffI`F
zL8DQJsc&RTsc}LnwYs{dwzdxJH$Nm|RFAl%w5BN+#$7k}%w96QW#5U5w;u!YZ{B|U
z?)&fG6Gq_wb_D)Uo;-Ps>=ClBzyA8+!-o$ZJh*@V-o3kb@7}qC{@*=h)R(sfcI(FV
zt5+^vJdfA>e;|5L?KK4{;a*%bonM)y{%->R>gKU6i+k(yqCXfB1ykKz+}u6Aef<K0
zLi{{kxVWLk)WIh@t72l`q9v3gvVQCCgD1{ixkcP;@Nuu*y7%?7S8w0{^wW<&{(zpz
zyLWG1Jb8Hc=GBX5&tJT9{pOwf4<Fw;e{|RS!GXT^fA%)G?<Y>2*fVF@x?M*v+<x}p
z=EYNoc5hibxOUsVW9P5mfA;P32RAO9fVDrneSAi!D@#+Lo;9)~M<459(rSHVOP7L@
zu9>YHr16k5*^FZ3&Pks(caBzAS2eV83rWncZtELZzHaB?b2q+z^6Uiy{=0YI|A6gB
zWbnV=e}4}vfm#IM^kaKN*lV;VUcGwx^5qNke=pI0zM{T<@!~nyGi1bdaqsS(D`$@E
z*}QuB{2A@lc?p3oR!q!R1p$lC2LAQ!Q#*!cww9;SBciBn<b?a5Lc*ispy3oGCw0vo
zyu(wA>N?T2jAwlhuidm`@1YZCFI~TbSMUA%fA=3edivt^x9@-akN*P0IijE5y+Jp@
zjmzgQ-MIVpv)AAK`19MZub$YocJT~IWi&T6)YVpZ^e@|Z@cg~k|M~vKgKK9F@7yrB
ze&?Yx*Y3Y~_g_D}d2;XSS+w?-^)(g5cv<U6(CS%q0OD$IX@sesA7DmMj5JK_y&}?z
ze;Zrcx+YJZK5Om*uulM<WOgVqSUkVzjFN_~84Fi!*^gq_M^B!;eErRL?-Arb{q&E2
z{Nta<{^KM2=PwL><3E0)489P&0t<&&HI%>j^{q?D$JsQzWO{2wW~8UBp%RQ`gVgQk
z01wOU)$KET*3R!~EKZK_ceA(9Q<oK?e|2SD`JUc>0l}f+L3l)fF;f$(mP0On(&V10
z)1f}Oc*WYyJNF+weev?;D_5^wzkcK9-G@(Jy!q~jf0DTW{PWL0ynXrj-i=FVkDs}C
z{SHF@-A}LXT{^mb^}^|GZEd87-?w0R`_W7H-~9aD^ZQp%AKJNL2oidiZa;qWf8#e#
z?_E246kQGrrZf~L1UQ<iK|em^97c)YsaPDI2dRK7BH%{Si-T4~t*&cqp4i^m-2?X9
zpAv<<ytWA@nwHeH_0CzgcE_PJ=PzBoar-V>>Cc|OcuD$s@7|;R|09L#k0XsZ7rrMB
z1Tgd;f`fv7^&R0cz=iSV^^?2Tf6gBVmwE1_+Pqj_uCWTysVMYWZasJ%H&5?fzhrt_
zRc<_pl3BsFfdI+a!rI=+#g*^j<?ZW__B#(J%Ax<oBP2evq^_Y6vrsxZduPsH1__ws
zCr+I{bN1Z%3zu*{^X>Qlz_|bXGg|WR-#mSA^YYna2acY;aP9WPr?0<ze|q!m!7VH2
zPbDgQVE_Oi07*naRGrw?-r3a+>8g1{8}~e&16SpLw1=~8Ta#_qWY<jAG}(5)jLDcL
z8<TBYlWp5}-SfX|-Fx0ar*+nM?+>1x)&8<IFR~%r;k`fCLwp*xkemOp*5IuS{^%@C
z{x^O8n`R2piJi4fLu)L~Z*uC>V=uvV^RKAlIGSd)S92>n3GBl`i(LM`1Xup;j0?U{
zMTXyQ3-E7Y7cZCUKCgh9L~~}nPPg5KOm9?@JV<}H#|=c$f^&gh@B(Fk2)<Vh-;oK8
zkNoC!>M<w%a^{%QyQ#u(ZdD@*uRG`~OC&axFzl0wQQHD@>hd?k1awHn!UeMzojACl
zfbIt}iPOO__*}aj5r{;_oZWE&PT{)01yFKG*x0*-D@AvchMj=fSe7s7r4@EMdm2&4
z?JksVBiM~G`WzDJ^nN~?n0FpRJzI0%n|m&6A{F$#-x?k1;0@bYh(L64o^oFEd^)>_
zWdS_ij~)sWVbQLvo6dw?PAGoi!mY(SFRtpSC*ho5TJ`mQ<F=fx`JvU@0|7&pF#hhr
zNNrI+g()X{V-6rC_0~E&Ip5y&2t#{7cv^}NAQp(pS<@<Iu>sGeTy#jw(bFRC*P{*T
zO_xGXFKmw2UBnSA&)E)LyQ@U2h*-$tyx|M*H7(nv0CEZV32nE5V5{@_+en{xBl1I2
za&d68mZ*7P6qLe`C#0<={|cQYk&+q;o-Y(}csDty^uVPVqqQ#c<j6<_Su?ML44bU0
z5JwE+0atN+vf=d90%%b#jn-{EC&Q?tWzlQ9J=$*qdz%D%RTk;=4PC2Cgmotewys8Q
zhtYkU2sK^JmF?_KPf}rW*G7>AP%@f5Znq=dSx7e8J$`&X+)G$<pRWOD{|eI|tFi@J
z-EU6{0q0j04fUi{w2RBowk*}V(BrlgHbq+U!H6+>%U+%8ZXFN0TG&=5q~>sG9O+eB
zFy)5{S$ecz(SBOG@S{i^usF=b1O`o>uazCJu|y)i7d!K(P2H}(FFV+l1+P+XkBAPp
zJ(INOUpgtkZ0~%BYrLg=-$S*StM3hkT7c6VpanDR0crBQd=$>TZjP!HBOCVZNA+^Y
zITRmCEJ|Pv#Gn~Y@&b`J>@SL+_V(ix)=+WF0r5^y_GDDNaC0$(Jg$n_BK21HEBEPW
zFKwuq?H;>5b@qN#4xjLz4^^a_ErFD^$g5<-T~4<j^Xjinl{#}GB_$PY_6t6Ge+H&M
zfyK#9;Wpr1&}nZ0wPK^$c2dj%(WncRC6%MWM11{e2Slo%n^I_k8h2Tv%!U4`#2bq>
za@-^-lvUa@lRw$FwF~nEz6!leD@cRk$);_pD1<49O3BF0Roq;gTiD@!62*s@Om@X%
zD<MEy-R1QD&=D4!ZM>Ml`nCynis>*9kotehFZETC3i&}8QA(41T%&)y@Wtc4z+8Z1
zp$P%O?LV+~FzVq^xn1JK!C&h9hYXaYIZD{W`Ff}JEJhj+mR6jlS-V)Bq3ynDW7`TE
z&xIMm@ZGZ3Acudy#bMGGuD2R*jZSwb<#XPjNaPVQ+5&%kWFg&XcktP4Y#0jz*t*SG
zHU$Bf$HCm^S$@J58VRs2%fnog;z`36c;U}o0nK<?&`vgWa<8k>?{Ib(vg_Sc4zosR
z#o)*TSp5hly^*+|-~aoo1YLy_q`3DFg8xI}Af#mh|C*9zaG;1m#%GibBUmpRDT9%(
z#3=TInu?yBcvK-8Q=nKOoy#7$-YRuXcRz#n8w_FKxmW}4!lTyWdjK&y)SS&WM0{d)
ztuE(B5^&)Lii>q=Lx?*KsjK%#dqZJgk3-P%OWVYhy`YD7Y8!<4S0(0!XzR3xUjBw7
zQH=RtBt_`Ta$=ImkBgv?)Ch11dE(O{|9<qnp6e}mhC7zf1!h0h<F7!&`Roh%bt{r_
zm+R&=B2Sm!y$}S6?GtGy@VI{KtvN|1;tMU-!t3$IIB~*~DbUc-Sz*4A_buIxG@!x>
zk#ATTw2+4&ik?n!2PT~dP|`lw>(|*Z_v(up%SHq@65D=d`%&3mnMGcq!<QE@h_A!g
z57l=d>&ha*n1B0a(h|T5HGp$Z>=lGgMy3jqpx9NfLy#*w(d92z$nG#Y9&NyTQ9oqX
z@A4&G?eLZiM${#e-fyzoX!g55nYRvJZ}yDrCaDSssgCRIzK_B;7;t+%U*A5$E<wNe
zo2`jxhQr75w0WI`BlEC06_}N6bMS1$<kU1gLik$~3zENe96&P1)^&3ux1@VqV<`P}
zu2Pkkb1ktPxcF=MT;58|1xc__+~0+BUYD1ZAl%E(QxW4n%C$}57U$#sPNu&9X3AD4
zVyTa#3&k>=@w6G8)dd0f=NEnHi9SmI`;VvNcS2lc;*B8x;<2aIMP|~YQ;YK9v0^;>
z=P&5OF^odkfa7rPE)sEt1@(XR;VT5ie}w|x77@t*79}^~qVb(RZ)cy;o>lXQw^M$N
z67av!Fk;;`KSO=(FZ_v})T6cAsa;*(1bQK!`n9o^(4L!g`yV<XW%+|&?yWVM4JmHw
z<4QYx_E%iHN%CWo`y0v@(eLm)DJY*l!m>Aj*OzBtH_Y8$U!dO4z}!;~><@<Rt39p2
zyE06oR301p+R)fKGJ8F*SY$HT-cjaM3u=X!2MQeDKQJx$DcB(@zmB73RW$+{H-qhx
zfqP6H0oq4#ub=xxAhmzf_X7nymiz7Xa>MWTh~{R0f%495Lbny;^7(yr0eK>QdR=XH
zfd>Hg7lO>~jjLb}x#MxYceXB=POE3D(XF0OVGorrZHTh76XMlDWgPv@weP?G;PoEj
zmBa_J&T~$2lSSYKlUR#0za$$8!>-UpAY^6Lh^p;xQqUnDoT{p`G1?HJKg3^KI&3z%
zDW}(2SRmiXqY#UDAM+O+RPFluzQ05h^4I|z?I1O99wbp`2*?*!OlAqWZ_n(PEvnLm
ztAe`x6i}h=m%#hea^#<?uIGpKN0E>JzA(cskC*#*0oq2!EVbVxXW>H)pURXSAGC;I
zw~x2vW~>xj&lhnSusSS6&quAa$7dd3NMNzFUO|M4w<M-H#~6c}Nzql6_EwgcK$@WS
z@d~j8qR9J2x~4_VMw=_~OpNX#3UkW5fbGE=bV^u**K_Z(?|Mb_H_X;oTnO#knXh_1
z0?sdUH4?9GwyX7aDNEPG#=`!BZ_w|Fv<u0}-=eB|?+R`*f>o3&p(SLszac|;xwu$<
z;b2TY!A?&4gMr=M#9{sy>?;nBGyqv6`i(@c=`%|eQc#l0KOXWCaT?6O%G+Z8;_NuP
z&0!hz@;6@cg`a~p2B|m*xIAqI_<FlvuC(|PF1h^_aJyKka~FZ?35&=(-w=ZB7HM}m
zKhV$k@OcHz#~)80t2`{ZOO95nGaG6DRLuWzCMUjt;%GJ3BCI?-rHI4708=UW^#lra
zlfO2VnH;!0oeXljd;usI0n=abmp!2e;CzjiDp%aD&c<nn89<GFecgpB5*<O{;VuiL
zJ9C&#q;as?KSY(R6Ej6OxDXg!l6cVQ^4PDyDTK`TF=!$oOK7{Hw(g>yEj3$>tXZy$
zDq7q8EvqPD9)&`;Cd4ELh^-3XE7S$lZijzcoAzof&V^bQgbsSNqR<e%t#(7+i1!wl
z=LQM*<v=oV@{=>Q#a-RKnEmkTNt-w7GrA(RUO)q8w`vRJ<Cvf+BZ<ovjm%&wJ#!&V
z9s}}=Cn!I!dUg-1y56(EKAsb+F0(f}%d41adw-`)MkZe#f+=$X&S;Q9Ype&AM=?W`
zo4L*-!z!ddztlHKDR{z45niPk?NQG4IXFay1m*wO!Bn4NuXAS5^@9Cw!{-2Vc;aSf
zz)|nfBY4Dn1X*hQSWsCx5fihJd@%U3ax!{DKr}V3Tk8bJ>xOuM{4Z6&{-bw5KADM>
zzyNh(rbhixnPjB{a9#g30E$z9D!=JdDjc1hJlMhU#7l$q(*(oyE);Az{<ca$q-fp!
zCRCQfc5WHTt-v%J#0k2dAQUY-H=j^(b$M-LO`$JF>$TZ<8vH`64UO?T2VE&~+&Sxd
zWtIKP9sWCqIr=8A$skB}g#pRdHJ3-nh?@56z}rsxBN!ma&zmjm<zRNWRUg_;Cp>&n
z)`QkH2~OgUT5D^gI(%^^N0-(;k0MMgybI5#t*8iobyq*hD$)tL3gYk-N7YfIRAvdQ
z3?fE`{~BSmP7`#uxDBY&kgdtj&#$Yqo&gCW`sbp5Dog1KqmG)P4#AirFDBgT$>7vb
z5Mos4x`5X$Hmo5-LFcoriweTf>3ddWFBkb+b`Wa}qFt|L!G?P}#!QLB;mC_PN@SPM
z>HHXFaFSR<G+J#yO>k>3AdXi1*8LG4E+SmEn;v=pcLOpeje^DC0u^h{cat<<S7f}T
zu+vjQ0zzpYd2dwykMH&yu5gP2*`0o2{+}1=i-5|KaFg{wR8b(FvS6Y7F-HIo7{{PC
zKP?!Ew@VC4;{qc`3Tzn+yI6BbB23mps(N0KAo21bzucaoz3tFIBFwr_b@8H*x5I3}
z`DvWU;<8q}q1La9&w~i!#FaGbA>2b#s-Z#J!UFUCJ>29i@3z*W$}9_^v9S?Gx+>e_
zG=SfGdyAm`fn~iJI9;I*tZs9AZa>x^{~gl2j2fCG0AAZjMA&G=$u9JS29RlX3M3o#
zBA>ZnJl4exncy3*v}QmEUmn9AZJ*;}5ug+Q3IC|q>vfXpd%x38ostyW6A+WfsXt(k
zh(TDh%}z|z{Vw3>^ck3YFYrJr;JrVQ5B%ADza6^l@_q$D7-CffdFl%3Hej8jjrfNS
z!D0!WYhuuJ>RErRE{{wHSi?4(G-W0=&!Q=W=J_7<-hqHd%<`n~l+&Yu0;!C~8vfoE
zPR72E8M&aB`K67(hzi?7g!=bAW}A-H*v;%t-@A}cjwen^V+iz=c0zh$W<0_QaAVvf
z8d+&^mcecBFb-Pnl?<94Ni&^OnmiBsvyjIQl~P%SdB~YzPA{LlW_?HQsuOB7Oo$a|
zoIoYF-k`rkez+Z8?1>!7^p6rOs9EMQ*v&R);l|dT;BcSe4!yl`#>+`@adBH)IXO9^
z)nwI1E228ojEsVU6km$VJM--UBZv2g_o2y6keZ?k^#-#3rnEnwb~d(`(T1Ie_v)_>
z!B|j>APRX^Fr0=`d0jYlFw_P2WD3denx=O`)Y6Jj{MeR%S2PUZvw_<+7OhA+H^JoE
z1^Y5J>dL68_U*NRCZGoKwSBw89ZZ5hs#?o)%bda@*nF!GbZ9#uT^^t{#SyhOrXVBP
zy9|P!tI&u1!7-LjGl{R)=6X3@r+FzHkbg0f5BqqrCfETR{)(O2Jj+Ow1!AwTAV_9{
z0zyvGiC!)zNSNGzE(r<J32@Pm_95D9C;u>!hN$yNYQe4{eX-Q`+<~`DOuW?k-rw#;
zfh%4hK}3G5=rXi|DFtwks#V3sB-cd(8dk=Nr*q)-hp1VtPLIdBU>~|~V9e)+98DdJ
zeBD`3^3)=T9_4gU5O_4--6lvpZbd66GT=O@;S$JIQY^NNF|@RFw8uD&hNNF^81G#g
z;J^h04M;er_gpIUf<WOO?*nDqFOyL>2WeJ2{wy8MDDNs>NkCQN?<@qINrYcY<6{aD
zv4MQIg3LpsYj{P-1kS#B&ZR4_u#2{<z|KyEbQSPK;sGxz-|Xu2ssdPRTRF&B@X}2W
z4Stml-pUOW)D;|)QA^2+V<_z9Y;s5arASLNb7S)lYVCEhR`pJwCAiR2G#C1`uKPOt
zhdeRkizQUvDqvJ_#qC_MJx6{`k0WoC(+t%Ftat{~9B4K?Flp5|K29SxkpL19pEu95
zW3UV3r$pH}XBI)L5e34C9yS#HQ_{39R{XckTa)Q3n1+LBwk1|^xrsf;=pYgx19w-h
z`*c)HT3`X1yY7vRs4=x|E!=bU+zh-g)Bf?c4>w4Eu8(Tun-oFG)y^a&-TIl!B%mV>
zop(M3wU^bW!LN%GuLcS4<>@Q18Ju@u8N>_e@j3ANnO7n>Sw*5JP+Y-@9?32>L|tfB
zK-Div_PO5o+w9j2O(A$tr4cL0Dxg|;dGa+KlSW_7ZhpAFqB35agjK()wn8jfrGPA@
zNl3r|FtlW`d2A9W=w`G|B=r=qK9LxSg&M6@Ky4@Q&%Esl*a^MQmvIqC7$gGMN)nwe
zBeic*e-kqc*;r8}g<(BZSuMZ15=(G23fV1%>15<A>akKo#rh;CplNc|-bVeSOJ{AU
zC}QR6g+2Eu=RrW&F$vd_){=}jKHeiu|0bLZ2y(38p1U8b&P7cnUoW%yd7KCeIk1vs
z_G#^=NPXS#CydjXW~;V)1#jQ(`oz^g^K`I#EKSYPCqToBdJtT>Q8n&)`qk|mIgefV
z6<R1{&!X)v!DcPuhsh|}>tw%3h2;%qH7$m>RzXBubvy+bYy)@PIm7@rYq>w8tUt;+
zV437n4beABO;?lC==?ClwThbAacu#0rHN9Jy$Rhd!4UUi4iS6l@vn`w!Dw{_UBW*z
z<DBeVtXf+J8%?Etpvc<p5*;I3e?u!itT7&ONHWaofbw_d^bB+!LU>$~*<gZGA!tN@
z*oZ$aRGbQr{R5#-84Gte*!ImKoeu9B5U)G&`g6hS@b=d5s@(OwZ75>0`7l0qxEO1z
zFWgbc8R8Jr{U*Ci&K)27`!|}(MGZ~1E?7<D)e@mx2%0tGJ4tt9fiQyS&K+&epQI!#
z89hpmukg8|)A3IxOJblz;qssO-{v{G>@9zDy*IXAGNmUrC8aySo`T?>MqQu;fw7e9
zmTr@G;!&m=fBf#43A+ncRJ>p76e*<&^%M*xXe;}J-%)%%uY!?MAP*vGk+x~!HSDz-
zc%4zWRCumR^B~vGSiKVoShNfiaJECB6>^KQCzFwH+?@SC?8Vy3@*=E%_5`Pvp=R;|
zAO{3@^W&-N^Xd9sclcvNsLB2v$lEwbm_tL;7^oZVsj6#$F;*39fLO<NIi-DP2#nW^
zRf001feij87Kw1?4sPHt2;G6^d5x%n&(=!ulmwnC;{_hBRvnu^*K67>ga9HMooF{Z
z&1RY4+KW7cJ-OZrFw9DLP+eua4iWkp6){?SlOXs*txwnY6WXbO#qb9MG6o;ys@LeO
zEDLL*h@0R!r&9mk)xRyyu(38kMVI-jfFNmzLauzOSWpCxg5rq?RtSlGg0|7~g;1GG
ztjsXkGcIFA38~&TmWP4klJY0%?w?PKBA~Gb;Sh+m(dA=psi9&b@^lgQ(7?#zHM5Ee
zBk4ur)b9DKCq|iq$oC4QVdyWyy!!eVLmkT;!T(WC8T_6Ep&!m}yp1mJpVUgToO@Kg
zIMAo3`;uI_(R$fmD4@Kd)=hbMLB{J2=V3*4cnCHL6XYfEJ!7a}$9;1vKFx8@yjRY3
z5NF{(URp*0xt0!^O2R}W6emvLPD4OVYRGS{3}T3|$X_1}<jue?9{JWW+fF44H{-C(
z@YJmIZ*2qFZyiD2ATQ_hX!+D0cslIzgMgSBA2W}=)RCTo%<rWO_=zUjHxK3NWS6%l
zeT(7G+sz(U3Yn!A^PRVOO{xp?id0lYXX{{EDtfG67|e2xK4<UzczT%syPfedb_HD?
zd2fsTCnV%&fF=<8IARVl30Vi*i2XFT0)s<jKG_Wi9uiO~IZ9{A8m-+ZD09JgEm}-8
z<P!&8sQ*o`EErs+K6U39YG>Df>z^BP)ENC_3@A1qs`h2QZrlH`pF%U?2wGCQ{?~7N
z9pBHDe3cxp+w<4#k6lBMX8Cb-P&ga*Xal94*!tPGrU}4m{C4$?<uGAbU)O7G+nrel
zeVE`V5CTKs-HhDxmh_jF)-yS7|G6&+AD($*lNf4TJuhVELoT*WEPD^Iu>XLR%@CPh
zl#vDDZpB9(yX9k*P^bWuGG}V)Y7UB4T2D1r>D^bCix+eWugs^jo+7IkT5qKH#0Bo5
z1vE7sU}2-ivw=g?W~t`0e;^&z@clmG_tSixHEe07M7^r~6O3s;RC9Qr=Hglp2Oa+u
za)C@lTWYN*F?W=Ns5l9~u%>iTVF|dZ?1;R)N&+9HWr?bvF+Bd)Z;C>ckortN5hmC_
z!6j6k%|BD3QowK#H7<-JD4M;ggFX1PTzHL%fyL(1<#Z*X(CE9@*zC`b$M;+B#n#x%
z)f&I&x8uZ81dbhXA2GbI`eJJSqy|oLO)!oo1yZrr&<Pt+`p7@`|AI$W3a8~FX)TmI
zo&5Wx@cU%%d<i0-I||@Ei=VOX30^Maz65FFGoa^NbtgBr3qZlh#ynQqi2&HIbNbGJ
z09m?r3?yPi+D$nBCV%eZl+CO9A5s2P5+FVe$<}eW8%JKd%op%s@4n&E7{ZZ|Kvwdr
zGmUt=FRosnD4D;DQuFP9S|w#Cwj>K)Cr%p}>WE>p-Aox|7qnx35}*sk7OX(GF?aet
ze)JVj0`KpWU6)yFoi0wFF|P85lNO}_!z64Wn5Hf!Qz(>3Ws0n|aW!kVH8i8J!t~M4
zaEuT-Bjl)YMsgUsO+#rQ$;$h9?&AYic7}kV#B?`ZWsVh<Y{(qU>1~m=zR)v}#y!r1
zN7CYDxt}tlCM>fAMIzAtF0RIbtPN_=bO5?V=Y+0W-ai&6PHM`<1@G?&T#PYLQf-7&
zL-#a{m$sVMK>S53@Y8)%mix=vYL5maNr{M<77nYPMccm~lo+toOxdxA%{NqHQ0T&?
z<MpZ%dF(*JK8>}KibaTxzI%SfpEq%-zO;t>w+BmENiJoL0{snn{ntSvk?RIzkEw<^
zJXt2)u7ZS-b3*qE5x>gA3rv3C`fiw8pck14EDmjIl8S^4vDcR<PVSx5B;@Bjx<ztQ
zb`ttQKeK<n>gu=)YX62<ir_!UIxw^t5QB~k)85)F=j)tr;97w-@7ra+Gb9sNSx+CM
z8$K9D&l{0D+GwM3h`J_e%Pk8E#q}Y|rH(RHXH8`(FD;jF@8a^P=r<|Afiay1VIaor
zJkf!y1q&)!O>K6Li?c@5KQa5?yg$sdJ#D9=woq73{b(}v>LZD;>-I}WvHkX602&l4
z;%F1lv99gapZ+cmWp*zT@7a&56nR&}^F48gc#xc`q8u@8YNOIEIdp~o24m^Um9vVD
zZnMTxzZO4SWss}aMYkKMXNM$~6Foqn(Cro(ypm_HS4^=3zfizr`s&!Jnq*5H@9BtM
zijos`)?XOio=|#VBj|S-+3ybW@r{@DD6W5p<Nzn<GQz1cbNHFUdvWM{RTc+WD0qTF
z;mERB(+fP}T<4R;)>}L7I?c+{!_<OpW(T7akkv7MS40?}p?`sMl8rWBuf<=f^Qp!*
z@@d&BZRv2TX(ZdG6=h_+7?$Fvn64d|^_`M6RiU|v`JQpK%_YbKibP7{n&h*bF4)>*
zdHt)aI=U&~Z)tP1<QJE0J_Q;y^035XxLXYPZ83bohe~?Q?{|<7{hLR^wcdSB6xR*d
z(D)ImVp%9SFINEAk1WBN6sbh0+EUU{DbgVdZj2eHU!zuO{F~+)SBQZgR?_5em9?qT
zE`IN84X?3TFis<p?trh)<I6rM%z3fF-(vE7N8hbETS+dGZzc$v7obLn2g%9lAEiqX
z-jJGF0#8E|eIOM`Y150zXv%nFgu4U&6ht$9&Bv?KTMkrv8<UBz;u%2A)frhT<Qua#
z;{?)N)6w;^ISq;w{#Y5ExR5ra?sVRDD)^K6hXl2CxF3^>c5p0h_P&wl=)PeFJ8Q+i
z$ytQ(X<DP{CW6IMulg9yLVnHr4TJgv8bX(3v(rV#R3|)NSnaR?^D`*b5Sp#HlXYWJ
zO7~YH6f#g|adbqOH&Q{VrlS^zwg*+nlTaOquCxU7FU~J$NNY=_5DIa7N$&gyC^ND@
zIUJ?>kkm=Be$pcO6&moTSnEAIeF9V5mOw|~br4rEt!sO$kk%#w@Y#KfpHO5tm`p+1
z(X1igTT(I5^EdsfFn?tt6OqDFAK+IEc5Ng^4Tu#OT)+Mt1H`62<P-Sa0>1oK2N~0@
z7c&j3s9Bv}pNrBvzrt`oR%$Bbb?DWCi0?R)!WX05&%!qWirY4_@Bhk2D3s9nezMUq
ze_?{<BDgU|fml@K!fQ=vUH_3#PE24mp81x~u@fkC7R9yyHT*##G`G_Hhw0_n#Xc?`
z0f0_Sda`e%loZ%bIWm%6%U?bvLqR!#nN=NF6OtS4t)gde{%?OZ_FnLn{aRXpry{r?
za-3xtb53?1iz*~8k~r<rxQfmPW-Dh@YNi=UjHXlm@B$Y~l{MZ5+P8QrDlL^bbN!E(
z-MaVP+x=^oFj)T1M<$_=bR(k`%qNMy6L2L)^+C&yY9#=&Qtap>ZrRm$B(ehi8xmB(
zFH!-h6%)!Jatv$HH$JtQ5JD{O3Y3o;zHgop;j3hWBVpzK(G_R<J(m>ZTxDd?*ZM!9
zRc)aU&_=aXktd<9nWDw?xzuqhFZrEsp}lwPvqW0WZ{?^|Q=uJ1)Ut!<?Bc6g0RAOQ
z90822H*shb2@aX&sBpR!MP!u;7^)8<cn7wfW-hnLe-+tnb|Tf6B*MPe2apZft*-B*
z+lq+ANi?ua$A8(3X5b?}-@`Bg;+?A(zTH_!+c~IX@+B=0WKDf<D}^9Jl)KmD7@CqP
zS!9m~kHK5|P(!(bNe0P@T^0Kf0AXCzRA$*4>)mgWqh&{!OEXk5aJEe$eQ-6J!{cCK
zP~k&zj2X4c;w1Fu`}sgyiksDITBbDfhVT8?odEsSX(g&oUJmkqi}+NYb6BSROY(eP
zLM!qou5cBeQ6z?<qeyFRc1Ayr=$uM(Z7#5Nc~hHr0#vT|_>M+=JZ^V@<GwZ?Z)Res
z;yWs!h+9P)YNY?u;UFGwazJRT2uqUJD#9olSqYviDd_q?7HH#(+z+kAwEHf#15Win
z6J*d4RD*f=<uUPvnbJlYPIlV1v*i(Y_6Ml{mn9hZLe5-dwL_F-nUaGh!t!nO9);P`
z);fEs{h9+8)PM2YQkV?lF{JjpOpbX-O)QIK@75a*wZNPHhYHT{@)1y;g9=&y1vKRT
zyPU3l2U%e*UrKaesefbWzl{ixRo>o=x5dUi(V||gHG83Z%nZQ$HTu)y<YRv3+hK<<
zyIb_N!!H}hw5r+MafD_?n`=3d<H@I5jv&Sk5MmLnu>-CD{QL7lTN1}ab0xk-9SzE8
zqC0y(`iJ-b>n(j@PXX6C{?c=zpar)U;VG-S7Jp`o_bpn!lwkQ;vItu_Wu?(+tW#a{
zGAQNWkTPg|3hUgB*r}{(WMtvu#%g?`v2cl)2CJv0SdASNlh+Z2{X$1Uj{G2cBa&Vo
zfFQGeRRcKaVHEy$#`DO#6m~|{;4(mu2#dZ80(n*L*(>vG*5U?HO~3iUcEPAWLSon5
zKh&lOglbJ4<xQ=s^8*B7a(CbvoYF&NbuWLxtNh7|*Zwv>4+7)5!60p`65Z8<jKsXe
z+PDkW+OQwPlsv+I+(ZN&!G$nlGdD(Z2&41R^MImrvh9lZ&1jHxf7d%Gya?1GI~e^;
zPfrG;zgQULr?1@-j8u@}Lz}@I3}<C|lY@y#uFuh6Z9v~}B}?E23&0Vej1Nw%jA*d;
z+3xqWZfhkK%#%&D&lYKQMy8`q2f1&5@VD(^(56_=-bUMR+Wbz<!l^k<5HKK!qP3AI
zfwUhaNCz1wIaeF6T)mo3s&Xk|zhm0BhS@Acr0avGW?CS_IQbdTdJLoP>(PwkNdr{T
zwX)m4ht}GT%i`9@4L@c4lI`x}bG#qSe2q11_xr?;wh^QBU8pY#T><q-ZKfo|LTk3<
zX#(3Ppk>o7*l>a)-^ZwV_Dy73|1bqM0;mcbe%!;pB)k%T(MbPMGj{wK%~OPB!SDn=
zZtis{uI;lnx;&rPu7fhI50sfH?UXQ0b6~p>SH_y|1pJtGE)=s>($M#%u^dPJb%IDT
zDTbF@S;vhYW$&&#Vkc|zkbH|YC3V=E2UGvb-{yYD`UyGp3&nJLa%dGZpg@!X8CW|Q
z40NVT@vkctv_kuNzPEF7aB&Pu_~2%0bJw_zgmx6SvKV%~9zCxre+IHX^qgDcjAwA0
z-t_BSz4}LzyirnX!Pq4jG0{{_(dvihbvwX~{3X!hd;Kboq_E@1PS?GG=Lhfb<a!@D
zS(V=6vpUTX_o&&g|G2$dHstQr3Mdnut<~C?=^Kmt2cZWQ!TL0;yjT9C{E~E%_wNbK
zUK$Yx-xTlov?z>oW_qu(iC1`isCWpbmy3Bq2+2}VqE1=PELDdsKz(O>e>XPiVLm+3
zgp;qx=<-zRT9*DeO~5M*8ukS|Z**gUfuf<=?;rtLKPMuH)!m-tE@Pw17MKsTR8KGF
zRI*Fq&8?*3;2s`d5&f!$z<u7GT*!vgzf8dt+N{yhWvIlK=jMniI~;_UGxx5ik_mS$
zS&FaI_VhInb2Dw7wn`{JSVh<eN;XFwwB3EhL7Ns<@O7Z1JO&3I<b1mT7M+>Bg29m2
zpgH`~WKp)L1qe+_zQ2-wApYw-Rs1bZQ+Em(s+>etLfn4X4ZPWoP7dx464FjEt39kU
zaVYO)I9s<JMX9FkT%;o;Z&<IzPnv-HjG*5&F%8mMa<cFL3bA4jp1moqHI>g-Yilb)
z!b{L*Pkznj`J5@v1^mj%`CCv9amIvp!FJKtxyO?R-p5pEd_a((2WX(OiI%*!<S>&f
zsX{YxNM~Q|<+b8R@|+8JCyu_hJ(6Z^b}jgpTd%ejR@_rtI+O$|K>sC;ep;z~@;315
z-;YE=g+L-qgpSW&)Gj~Qvp=MiOowg}LG|X)r~VSa1{e*K6{?amA&y!Y^stpYj~E+l
z@eOiB(QwWj<3A8011(j|3XqYk60V|nVTHeyG8%sLRR$F^Z~EO_Y%V@lWvn4Oh0BbQ
zXwAvQP0x$VODP59mX|7V_=)-aY0ktl9$h<SAYl)d{yvGqAAgz1N!*_kiE8I)>Pr|K
z7rf@Ee&A{&;`4!`HY^zdxIK=h2DNdQTUb|4$ium?EkYe(0;CCrJ=gE4(w&z_Brrh&
zQKg~yw0Bw=9yu$ov=k}jmH3zo)K;~5?fHJvbWAvf{&ag(-;h!bw!egMK}#n8<1Fy_
zz=JQ<Hne$eE@&w4FmbXfTEq_2j4tAgt*063?oFVNd&2bQW=<Ds@zgoZXbh3pEQbke
zlVKqw)`k|H2mGnB`NAFedD|Yg!y<!}r~8@stNbp!!m|}uM4{I0u4GD5$($cT7pPAk
zGny6ZYh^C))?Xi$?eztx8%*bWQ!(W^@_f&u=c^f}vwh7vfi67bxbkw_6t{;Mr3r2t
z{sJ3M0Ag>i`vuw#PED!OeM&ZyTbO5INJ=-<8*bqnP|Y>38jpTJ@=O@po{Kr%u4Kqp
zN5jxNAQN!Oqa=%r_}^WfgO;(}vbN6`&_9Hk^ir>@0Gr5WH9qQnRQo;z{^Q~01_`^~
zs7X@x@#^LtwO1v6qo?Y}e)e}EYbN%wqfH*pW-%JSBNbt7-UGE?v<imRUaLJIxO&IX
z^^+9v7_YyT!;K=Agq$?cWBMBtkMYc6;<7%??8^_)7e&{SrnvLO_!H8<e?MzWX;n{Q
zaNNGrc%@adoS2Zx)%y%81;Bguw-D{w<k9r-1>bbCHhi@qb_B1C7e|tK^$&hqNxd&T
zj^9v_Fm<@O1OMjXz{&6MGP-_9n!Rx*+-@=s^y1$SXFAxe<`q`JeKjhp`inpu98`vD
zyL&9*9d&;{cuSAhCbh5l^m!0HRQJ<(NB$~ytODlS*@VfVy|Ab9Xh)ad`^m;WrZ*3B
zx<Dbp_~dv!wtudi8rsKC<@ZNZlI;@^4^id;D>=y2m))yFY=?f?J#%Ssa$_*#Xm%bC
z@QsVJTs~UsM{~E<$M`mmSVkm-DSDDOswzG(wZOpGl2gP9x|{;Xi`Ph!8}YoGc3>1g
zyNFr|dKFy1y}JCg0114Sa?ZAnQa1i&zlFi-&)46@1MG;gNswYVIy!C~9LD^Ew0fYR
zB%XSHBL6f(gpzf2Zo1;!ano(|I}Dfw{#*(|=4}yD>G|ER<(?DYrVrs3NWnI1LvF;g
z6B0{NXZ0WE|LOcBNun?K-H^sV4P#kHZ%bs6h+InAe=0~2Fz+~}#&IDDD>buNoyLgr
zegD$Vx`2j~HLW(z`0sJtji(w$p3{k>`8i>hzfT@TU*H=E5>hvIvc0@n^Eed&Trlul
zigdVcFJ*zY6=mq+POAd{BzZw968~QEQR&I8l=`>P|3YZPMx;F6h>(h*dyBpIKjUas
zd^SA7Nfj1EVa)!H9JBlGhO65zgW3-<0rW><kCGcb;}_HOnBEIP<c4w@Ic>x#RoB-u
zT7x1b^{B-u_?z*bd<lOmz}5g8>hJc^LE@osbsAL;Dxon-ATWRU$iZCLH?R+Zi3O>>
z>(@gW6q5Q-aYk4&J}{2#;~i`~bbMmMKXFM3X$D|}u3N&MTf-3?xT5c!hyQXjPm?EX
zE1_49-_D}u`j5~=P&skf%`gTW+?JtKB*wx~?;UAg#J=%4g2!}Pp(Fqzr=*$S*ZIo?
z-DJ69?Qi08gA?SI`-<nM?k>ba&d&?^e^Nx8AA7fL8O)XG$HL*r#=fgcYDrvS_Ai6?
z;*9s7biL9gh@<2FiSri=cd>urZM;48jgk;3hCbIbQc*!W5?9aXd?-aoC)kESgEYu}
zH=`^23o*6czm|5qwL=VGrN+e@e&?_}IfEWsv^^K}xVc-tK+Jjb*zLWptSG}Bw<0i9
z%}57F;NJ+xum?q|O}8=dvUTS<-S^*;K_fZ3Fv?&D3**CNd5-CiQyjSe{ZmA_Q^_)l
zjQ6n<4SB5hnH|{M9WRehl`l8_+t=R(cqY&EyK+4B+@fL;fy6y}2<7;@tbD#_%Fbb*
zE)3;_7$~DDCaQlsIa~g?SXc_by&3wAkG}C<mkwf9SrO34Z*@P(`b4M%dxIOJAw#U?
z2Y}NZ2I9fZaxi`Y^md5~;MzWiOD-6G>UH`&-{r0oYSvHMlz$nQk3_t6ZfXLHprM5H
z71AfH`ulJ63t(&gP;P5a@(268Z*D$ke$*S5J2&`g+6LKZYf*HOnYE=HUe9_`pJHh4
za<kp%<>=#_P-w`lJVW0w1@Fkea79`@w2F!gx4DJ$xbZXfFy|h>Iot@{&|%8wkMKbb
z*6W%GJ2xA(kU&*+0fh7B@~1H<SCmZ+)T*dJ$dIT{0(ik8_w@merbl>5Y70x9AKlx#
z#gqE&av4`!D<0VGt8s<{?Z=5m_^1xdLMd~W7M7UbN+gWbu0dtTw$`?V32VCW+ZriB
z&(6+$hJb+pngleFtSWHC6?L(mC2B!yAM2qD4&15@u)L$T$;)G|w;shv9eR%5b%mZU
z5MShC04hTVZY6Gp+jqq`+zkDtex!FN3RJhuglgIa@|rb696Kwi#)}@u5&V{~`#J0S
zzp?jC3#rG)Up^hvMx=j1$mEiv{>g#5zQEVzTS@5=7nKWEUY-$YFq?Rx9NvaZ6$o&j
zZx}nA%;5EJrzut<`jbs`VQXH_OzH2*b`?FY19UXKsy<%qtB+y$GImm%4Hey9eL_RP
zPDk{cVrek8?;y}N*$evI9}tfYAHSa-#$K$SwYb<_rg<WRg??44>Ls%Y*KVwb!d%ca
zszLJOwt#FE#xm<Z8tC5;5acV%X)nt4Ja`*m`hA=m19Q{<{?Z}2P8-I3l_C+FqCSSY
z2K)fG_b2w<y=lSR(y1=5aFz4LuBZ@i3Te&BWHVi^tV8|m0Gw{yPq0W6!Esvg=em2l
z|8>ue9y7RQcc7vQ+H8M*T)p5QuQoV8yP)xLP(=%OrOf?MK}UmgJm?y-^BBgpBGIUT
zI|PGuoZ=A8#~;2rI-%my4c`>4;Pve@0TMZG<9-M%nJk}-AR-aIP`8Dw<HEs)?ytth
zS^J@cx?1>+mG(7hcjACj%olvXiNj(HGVvCZu6_bB>7G@HjPKpkurO<(&1fEzw3mUw
zg$`c>)xtWO><!t)8d{pK!U#&VX<|P44>XMD4zYI4yjn7(GyFsz<BJtvT#%puDD|>@
zc)h-DS1_#8qMU1UH09E5On#UC1w}yi=|L~^)zo8Gx@O|@QuYnaYcvKWMm1loCj?SV
z2MR5Bz8i^Ao)hebGQ3B9je?!k+w}nj2@-knEnzgnqrU-*BxoH>HePmT<XGQ<{1%K?
z9!-d}NTjvWalhg;cKv;FGrPnLIM}$CH;=j4M6P;O=+-M7NVo;-nLg@PnQd|jo~bPH
zUC)|rim|mzPr<w>7{fSv@uh2^!@<EJ=i-m|2m})$P58i%Qt*@RjzR_^a*wgcBY8+W
z--3q)x><z%{E#OErP~^-3|R}=Io8(o^fP?C!;{mPu9LP1Jm&!S9c_N#jA}YFJ$M96
zNs36r>pp7SmQk#g<mXX=OcoK3C(co-7GJZ+3ry(O0qVjb7|*3t3;J+J6D{m>j2$Uv
zrKMualz_sjvi!o5%A!0Q>^K7fU~I(u6Z{5F?WA+X_hu@~mXIH(@7^FO74GADP-hr1
z@#0IkR8Fhq#wgS4<b?si)<I7lR}zBzi|Q}pFL0F}qwbu*+`nY7;=G^U9$YOnB;(4e
zs)@f&%w5j}9nbMs20K5mu4l(yK_<z?a;?|&Tw1#a04qUlEVg11_uJ#;H|=fASlfre
zjn*YpLQI^h%m5Y{Q;HVSPu>EHigFqLUyuZNco;B6$=nYUb_Sh5KyGKHn@qVVn#pAs
zdnDS=%R8fe)O1heg22*Z`y&E<UeA#j4%Cm^=l4@{(2wsVV9^)%@At_xHO;I)2v+jn
zrF7&hq=G>W-s7jD<;d^$>rHMO{efC6@?1EUPDd6e{S<z}3Wdm`tkQ^B&T%`FTR-ao
zvGocfP9eT3vy)x`S>N2Q!v|T{qng3T?{Egb8(X=INdd*RZY~)U<d*k+{>0SGiLp$c
zm~f}bIykzi?yvayb02&Y3`&Ah9npHG76L2izDJd)8Y~ZkEVQ7!D7OMOm_b)ST2~&^
zR}q^PI?priF1(&bBtU0o%kYJEOe);sK#|erENFIU{AdZ7eZLaP-Vg*N=eQ1)D3otP
zoQo7Izf8y>JvukAiQi~V>8a@YKdmJ=Vy1zLb?r_P-s+w@;dM0_&p*0N!sfrnxo0uW
zPGnhvnvp+#=0twS3QC&wgNBCQ#ZaHAb3|1d#^*8h8;YUCz3h|{@L*;1b=p1au_d-;
z8kSG7ET9KQBwA%q3o$<hLWqiJc}3zpl8f-sqNrq40n=};YQN{9m|DqG(E3xA*>k9=
zWVdyKgGGUF%F^n@_2G0DXwwbOV<#)nG`|(nHnq}-epb;A^)2koa*E-@m|me9&andT
z0g<tvx7XN?ZH9spFEjNG+>5c0*O)BQh(dB-jrRZ*#tc|DIz@H!iPSC&D_6U6RCVQm
z!mN<*-77TQU<bEXcQ+5$_u^B`){&oFifMy}pp7J*AG&{Eg9}tUDmk5-4`X!Rk=9gp
zsnln6^%0YUS{aGJu%xvNJo`nzcV-8bK05QLcQdsfv%Pmf>1m2p5Tejp(h@?wvLL|A
z&j;kQOr^SE{VC$yJ)#6XNC5>&6cXRnk~o8ROj0}t|8O#GG#0dQPOs{vX5UA!W{`M1
zZW-Za>I>P9H0En&e;mF?hVew0bXp5IwoPVow=}ksAz$g!=pf+)Caa6D|Ks8;Own0r
zP{<7Zmpsm8s#QSmB}#~fMT9|sgN~0)j1CY$H497M5uvK7y<&2H1S)?YN~2R|Ac*fc
zc}`ar@hK3pAk`CWchx=(rbrv0@jq}akV7l1{@Qt?^}~7zl`6}swV&)9L|_8@nfG%%
zML|Ho+Z$H2(ZlQY1s3Mjlb5Y0Ihb6!j8qKLA3?lZdvutAzD^JlYW`gOmWFA=(++q{
z7z}jqRy1y0M$uR81w+H}0OQbG_q=(j`)r4doi@>5U1Tz3Ei>I|PRL053)cN}BqC1t
z3mqZ&&NgPEfZt$!@)yNKf|Jv5N;aY`eT?#F$_~P)XnC37ePt;#9W^4l^;fdvY%w`2
zAdtobj*fsIiW-;aUm<m*2)ZxhF&*IB<h}XQI9RJeMF}5reYNfKeZ&Y?(`3M%nxqP$
zK!yx+rm4Ojd<edp-pS3)34+D(GiPcz#iRbYo^!$26n98PGSB>P4P6aI>qW%ws#^x&
z>9`o!`cv7rMpzi#*FsYvt;1(5z`4obEvyk;T1`492u=1S{WjL@-Eghe>j4n^VyYN<
z(4b^nUR6RjA&>Cr$U3ivW-3Hv61?BQqzCb%rxJ+|?^`m~-cPC_%$BgHmbG_{sRQm_
zdb5k^mWB!jo^xTBmz(n;`;Q`Nyi2?6jRq$v*YC}d6(yCW8F-s`+y&4K_lBl^9$hS+
zcZ<?dfrtj8;VyGC3p_N)R^5O+-W+mwdH6#-D$@0>Wbjv481AROfSMK~<#3`}s(bys
z$n-z1S(kbX8h9611BA*98!~?hmS4Y`Co5a&*ZMuy$h#YU1f@9X>uR!Otm_$SgJJYn
zau1T_u%}VuGUwzB-9?U9WG$>MvE~+)wnJTgQ95GY%V`Mkq)w7Suet-u*Lo9vIqsg8
z$=~+O_mkqoBFV|UV$c1woWp}>A_N^v1G%Mc7a|+$&X32dT_S)u`~GCMfp9%MZvwcC
zp153!Ij7WoO>o0-|12fX1JB2&g5XYfuPidOQ2Dk-!<}L|+2&HtAkv#@w1VpT#Tm@#
zjTtCv3<L01a|H*M*8{<LcbPXZO1Y`KMJ$^&<%NPHL*m*wiyu;YI(>7~BHU#@GiTJ+
z?nHvKw_Eoj_9XE(PmTB>nVbW&*B2b=F9pM>06}N<cK8#WT>PumkY;RIDIE*L7p2$-
z_r@tztbw#GJqoz}Q+%w%lw^G^?n*nSvHnsJ8a*5=_VPZN7})ZEz^`E4Ln4x3-+EpC
z-*OgeR@dt*rY*m#nfp-rHJ>U;QO-9enr^NHV#VishEHgWV#A0-Q^dFmH4TY3H}kz-
zD;O!^7q8Y<{uLd5e*>7S+TW?Ul`zwFvf)!m@<ekt2I5jM2qK{KYpSL@rZHm`b~HK_
zb4IYfe2r{?@Wzr(Gt}U|hSGl})0lT!mJP4BgKQpDv0baa+7>3vQRjfpFpU`M5DzJJ
zGgiq@@{)4%;~mo9a@sFR$;qaJFs(Y`JJ+n@zrf=}H8-bH_R4#It&Geba4sgB1?fOK
z#o$FHJ+K^E>G*$dUv@wHH(*_`gKWy^xqoFw6|@E9gdxp(@<~<izw;!!-TYxs>wD=)
z!44%WP7z!KUCI4L0CUi<xeqNY@?WSM#0)$658N$-j$yobA7O?9|Mk&e!=#4|UQJf*
zyHb!F5wik<8{2zctaO<pmp&#)1nU<BTe98o1r^CpAG11M5ATDr>R{c=^#u`I24I_Y
z0QH4?y?pTMKp&(2@bBGg8<b@1U$}%gXuGPL%PDAum&WD(kjo)hNB|HCLI;gX7k<Xq
z_#xjgizl>a_)91RP_j_N!6(hTARB5`5UtT@eV3C4bYn9+mYfmcvwb|?-=;zP7fZ&m
z9+x-BO8A4#jd%>5etAENO1$zJ-HrtXgf~-yW&0Q871e-Xu@+!51HBM<p&2($wd@u-
zkVMq3*f42pJuVm4YV>|_Ap=w@-Z_<FQRd-1C<rXbd~Q)NYVB;bZV%CN3<R6(T_&gC
z9w(Q=e$OAI!4+$bUOjga`^(oo7u+yOr8tf?QQ~#(c?i%-p$$Rl$_^e1;#OvWfhr#a
zsqQJ8WRE6Bi)h0BuZGM-T_YDAW7pmCum1DH5!ZD*L|W*(moXw=-a_iLOe~}nPfLZ?
z{aL&rUT`lBg*)9r%R}N-uVzsQFMeQr9vBZd*z?ZeRFPKspAY3^;iD$Vac0xw0n~6j
z2x)Haf17>i)z&u!t2uw+I8O@zigK)d(LVYt?d+h9PRg0`+EA46FddE_sebS;t96M)
ze_j1l^uBAq-$DP^0Clf{C3@KKc{`kcXD9|8ef_4JL~v|tU(QU;OfOk$9k9tvA7P~;
zM9INW6jbJJ<e?|$X3RG|5#qUW#1Iiv=-0uYCA{n{U4Yfw*I>IGb=?9?8+=a-x8gJM
ze7ndAQN~hS%fBUD;U>Ary<-vnxW9M~($(}QY_oFn6)TTzz`6PN)$Y1uc7*>-yL9+n
zK`=0n4xEi-gO<cG6y-)NPiFX+WI{-Y=r|bL<n&G`Reqv+K^0uRe~oAZu598ASCQUX
z6w)i$){lX7$3$}`gRWD+;StoYj!!))2EgPxVd2zwoiGFS80+`q@c7E1dd4+1)M9cG
zwg@5r!$dm^xPyX#QGp0?`qzzIWEiMgY+vcr6cu}H`__7UnW##t?Ot|}^{o#Sh9jML
zpc_bqJHbbHP@$GThUrRmo<Vnv8P`2uF8YQh4ch%)L4hp$EMTL<?W4bEJssQ)^W;Z4
zRGx$d7f4_G;l027h3j3WojJHQAp=TPIO;7(J}TE7Ip3rju5L#2cdi0lxH}=Am47Ni
zOWb9feGU`v&`rch=@Fi;0Q>ZvfmkbQ9J69om33;@kJ*h~T>CzXw5MX=<uv<`I}5nI
z*+Kybxkp+A4oLjuM3G~agvHL}mz*xKv^m8=%S<zbe}L{c{Ek5wiOnq=;fiS@5$50D
z`=ZUzNV>VU5l6aR&JyiI>BsFWLL$S3MxxZVb$6m|{h$wO^3R|0kbPb+o(?rJXA94@
z;By{f_}DfS9w#;^SNt)kdLw=Rs+BHkq(yH_ruK{eC!imS=U_4nx*7%@Qt9}DQxYPt
zvpUGb+-1j|GL(JX@_9l^#B!-8*lK;AhkyTdguK%1Btg%{4_<s{E8g73(yE{s*^R2-
z7o2Roy;REoE#MU1*6W{(YkeKKWOik8w7rjUUweZ-t{V$AZ20EmB>okil1K(~yeH-=
z4{6);1<)53ODf{~^7edN2D{h>+AwO<nzuQ6^%asrUbk3l22g29I<Ac<5g*(AlWJp9
zgESo&Z5C3c2%LJU5^zgKt1D4o!e-i87Rtj(`XI7U5ZvSjTb1n~Q=i&b7QR|HlJffg
zmU%PGri*MiHAl6(v`E*{Od{B}g_T$mq0<1xc&7cqpa#%XmZzkZ4n{-WHdK!bF?DH^
z3qsm-Eyr?At3oT|`Pe03A=N;F(Qwpg(Q}!5l?`z=@x_nwClSaWaY9QncH|aZ<L7~D
z3`NYVo+l5ls4jvXuGdc+;|#0C(UXz4b+ejiZx3k6{HI+SgO{j<;>6<7D0A;1`Y@O|
z(5Im*tLFcpN+2QbHC?FmY2=e{0uG8wCTBpU`bDmG&ot|F*ow%#6}wOMZnU~R_ireU
z9ddrdrILQ)*FIwH_xw9g>9~3oF2i6+s4Ea3+6x9|ikGJ<HkN0(dy5H?p`a0e7TB6x
zNsg4&tuQ_Xh91X2I}S0ZuhU!{O;S4tyq6$-XsNK!&2+Igx;;d&{LQX2yKM2cU<n#_
z0?EuPU8ehQB@H1X<aRhZ>^_@=%0EJQ+#-TMs7Suu!)+*GZE-7jOQTWV-vofO5&Tw-
zBO;+Yl6fJz!1SiHxUPjw`;ziZXjL{Uj{vC7Zj(jSw{UUbl&{FqeDsykKHBR67s8$v
zhbvz>2<e?LR72nq1Q|6+?a=wSAmyd2(9#UZa*Vdn;xyJ+&F|leM$19AiT<8Mz(QoJ
z*9#Tx)WSRq#h12%B1a!fH?`pGvI@5T$^9$oK!Z`oU&AGYVzvqDLax^bLAazoPoim-
zoj#5rrGvO^rRN_Tk&=LHmOVFcAn_vS)5WVXgZ3*&&O(|~3cM3^IF7ZEGI|TDMJk2@
zc{cqp{r|%RP_p&d<pS3uLAj%EoLAtpc6(TK`WTF*vjP^r{m19Tx?-zm^CFkon_Yg}
zRH{>zDVu7V31Uh3OL-)Ow~mqY7HO<Pfhw!uwF)_nxETh^wSdR!jW%XvVvxSVw{J(H
z+*M?Mj84yvNpTW0wBs^iVv0TtJ}Iv$)4cvs=9b%^<6!;KUgor)uiVun^?iHFYTh>x
z1|IfSN;(K}6ss8IP#zsQXVuX5Pb8}1&td&CpFTRTx*|k}F<9d3l(l@!oNPS&|A(e?
z@T&Cv`f#>w+nj9MwrjE`cd~8UHMu6+6DHfXt@nI?YrXvg^{mr*?tAb3+1HiMKEI)?
z7fVV`H3q%~Xaq?^K@C^L8<#KkCR)VvED!IG#gNrk!IbnH`#Q$In$lOz(lk_A-4|Y{
zJzq$Dulo|K#HrQgZYo6kWgll5##<6i5@DZoZtcJtR$B@Wa}CihhDkO+T6~R>2IN{y
zeJMnjFCGlNu(>BzV~GI`sD`k80wYJ(DLaSwC<H_RA|g~&R9xK1m@WWaxDbXLA_Ymc
zO&HW-tZSO%7T{@QUWu4?KNC7V&Q6cDx`&$*XI8QVV$}bh@Qd!Y_nGhdpvF*k3WpYc
zDeaIV?46smAUXubZaFLt!WZKfiN0VD8AL<-8?JXd$yC%ANR+hOh%4}Nfyo-Z*T4=3
zA3p#{SZK6N8nw5dQ&13{=GqyuBHNYifhb0ohjd|FSmQ{C;jozSB;%dJxAFuPmp<f@
z-%&HJc*7m^$pTsFpe@y95HmR;3}PDiIYXHNodMkl+UZhY;1GLOsxM+9hkge%CK7&S
zj&4|q&ldQ+@u;~M$MHLuNS&Fk2Itru8YO`6PPQpGjakDpy%dk9q3>xl!Shio5hqt?
zPWELM5-X+3iO73b7EM&_vc#?wZPegqFDQN?O<wUM!y;7|S#3f`oJ|>A(z;(jgY%Xu
zEVs1Z(B#CC7k$FTaP@ar*QTHKc@$5w32X4A9Hq%jFa_(YblTIu?RE<FSi(=UrfPu8
zez5)|r^Rcy+)2B5dUMT+&*50B1^ejn@Y`AjK3S@sh(k@jQnS0ULPfES-ToPIW{#RB
z^esB=M?>FXeMMEJoHgB?x5%#~W}#r3Pc^b@E~P69-uMk`=Dj3xCOyz;)NNS``1rtE
z?N3ZeVLp9SPPyxsJz!CTtw^f`+YFdb-7VTR5DSan73uRNY)$~b$0@OaUn^trC(DDi
z2v|oiXLU|Ai$##jc#?&2c_xjH2wvjL40=#x+V*+OKju<&ug&9XXW-;b@Z97)kZ89S
zV=egv<?bNWFCan5Qbky74=+taxF$CBu64%zeIEB;>fg`D%uAQ1vVoa)2*8;Hj|;R<
zcLZ2N>$=dMt)k@`=nQ^K;(#iVQ*iw5mWgjrL<8I|k9ySQ2}!OUWMeDPD%@EO=)zf-
z`7I9D5J#YYzdutEg$86PC~IzhD2)DTUbKQ{Ukug$KCfr(8nD)?MxILnaT^)|c`+W{
z*Vd)$FMc-f@etnI9Gw4~4LD+0Z##xad;#PeKuSDCH5)7WuoF_4ilw8XwZ-&Nn*4|5
zZfWIsb|mX?h~&Kglk4O4331!Fa5(z9Q;GWOHF?2*A(97skdS~O>#&*Qj`I2C4@Fl`
zm=%*J#!|Bj4BGyBV$&T2_`WBlRBZMBQZ)G0)+7RmFlK;QdOjuueA<IrxX!c{jl=(7
zwquBU=1|+SQsSS(U;ME!OH%4;12+e=<2BHwBO!)q9aYb_PxVDNOQOB1hc?{S%raBK
zECjRO|8SfSxchs)K1ks{o#f63++caSnd+UX%Ux+Nau*fRuZKoIVKb|D493t9j(bVg
z(p;ZlC+n_3<w?l~RLHo{W_XXJw=e)##z~U=f%Qux>G%u5U|Yl%G68cyxJl7)0tOCw
z+0QeAB-J?<?7eXGm*T*8lCCfzZ@i@&=I!s#-N$+8Ki*7v0zLLJauO~eYI>s0*c*jf
z_4OVX9c7<#{A-7>z|IF$d5U5luuX<$UN?gj6t=7@s;C1{Q;L3|*abUxO)?Dq2#yRh
zhy-nI-Xjx*@hvVZb+kkJ<a36Jv9!9J8<fbZq%8V*M3ku^V87Zy_^xEw@py3A%j}T1
z)mhu!P6wYIKhI!WG=MI`5Stiu$E$f&I3DU=n(5pxT%NSldK)=am)q_;de)kD`zlj-
zrCE{~$8rtOCe`5-(|hjAK&V#(A0n$9#!Xha>qX@*$um=tu`dIl0by{0fXAax>bkA(
z&#z&zA^f#YZ$&?mnDkQKXyk`&voY*U;yVGaZ^VXT$b;tJ{KB@?+@U$KfJDzeB{UEI
zlM<Lamw!X>BBUhieR-OSikevZSr#ZP_&lCTQD=Y)%_|%&+0JTLX`A&o>Qs3LaI5b7
z<tqWeH?ztH;YK5Yx{!6Xvo%5As;FX?Th!4WipD~V8+_CZ6pbNH#;$cX-@BAm%7TJ@
zzNS$cA@PNYoZ4;33x$TUC3e0+B5>Dy=cjE=Bt)Z#(=CA9M%KZM(gHKqV6bUqM!)_1
zArw%}btUNi^)|Ten7`GzglKmYI#r*-Zc-fj@oZekw2G4P`N3eVKuPl3axBCV##O~x
zc>sov<fp_>3$<U0XmhG1_<uq|N;neejE%*hf0h*xnqzv{eefq$mV!Xp?(6YA5%79_
z0ilA-LLcvMTcu{5ON?2ppx2i5w1or;tH}UmuptdMT@^)LQT2%Y@Z{JgEn{<2Nq;X!
zyovS|9MWxTHz?XO-S$w5BMr`2=II#ZTOfHJ%-%w)F5sVaME>|L@^JlZ1jg?g(7l7?
zBOolS@c3vGTFI#A^=3I6cnQztuXi1cTIvXb5aJVf6?9_iHkBrx-JS_r;{k=0Km?G}
zQ!{I7ga*@XrY<C6!4V&ym=tq+z9iak#7LufhC@hNpQ>nH4FRAlC~XYI{(zWoNhN5m
zMpTx{x}i}jM7QQ~`vCa?=4jtXCNsDYU6tdplyF?Q4nH2KIus(dDT}KdPY=-35_01X
zrbNN~z=K2w$cy*Ul8X9gx73vs3;}t(!6lrNIpFAbwz^o8))kXt!GIlzRX*O%Hjxi;
zH~`Jc28`%DeL3UiuZy)yALRTl&(EuUqdH)GN}$>Hv#~vSVs2oQk)!mj3Hx)lL<|H6
zGmvP}Er!8Pr1(P)Q=31g*qthD0008RseE#f5MN!uy#XnCv~ye=YaXAN2;f5=uvJD<
z%cxS^LLVsFtB<bDY_Z%vwEc4>;2$S`O8cD%sbE1HFnM5d4+@=D-JBA_?X0cG?F}VS
zN<mFcNkv65G&(FE8V&{d;_bstnvEA0lhFn-$E;<Jv+_D}5M!$Y(wM**YNPTs9f$O*
z^*;-NHU4Mt>w7;58hel$K$xIX-gERwDWl1IMkk~fgg&g-aH|_)CE06g-(~OPGB#y+
z<e?Sf^jzvocw$^QnJ)=+Q2A6$APh|?Nec>ITd<aO@Ed15Qh&rP)<in^%-0YVmy(fO
z&c_4V9F~ewzEjz$>1A%k4naN=MeT1NB^}_d$eudV>WgQ`;p^)e@Y&R<HAAr4WU*MY
zmLsiYCLy?F22E}p@~Jk*(OmCw4<8$=B(JD2GdVpqBOL?x^vfjc+_svYTuJR?M&!~|
zLhZeyc8dP^&jOa!5QJxYP-blFTHWY|h_|E*yc{`W#Lj{X3UATMO+c@wfA#y#lZLpL
zu5e0u+RoceEGo<wAXDhp3*rPs1t;y49LaY@baZ5p^(W5lO48q7bKW2i<vgM0>Je9)
z$v-Oh2_YBp+vmX-Q1sO^5~5>z0xK<!a<v7`7JKMHI$r}~tOP+Ut{xJ*cf!(kENUlb
zid3sB6ZZjoIUJ(0CnmGEv?ULUGE^L(9SE#W<swRWuk{tyfU{Ndzd4*f(?lCR-u=z6
zrJ1b#CBH+5>cS00_M~p3#!;!$g00+`RMsx}z3wq(BEKG9>#VpGJ?ypl%|xAlhQxO#
zR$({-a9D+sJt68wm`J;Wnd-F9v1-htFymS6(4M!Or|5-F5uxz)vdALs1T8RA)a`=U
z&e#6rb$qw%0j5^>p_iwtWRUAMF^AAtfgLtH5PW3~2{J^aTJa+@HCNdx>efjpMzgY7
z2IO=0c_w)APHFXcdXM}bpWBtI$7oAM1}n~}UfzF!{8>gxg=Ut=s41#zE3GEquh7Io
z^S33(FSST}K}yi$-1d;DabGc}{%^6@mjlu3SGxsz0A}fhn}0HcfdM1#6IpMTKgt%2
zAD&r**l2DW`twL4NGCY$UT9K&{W!(X@tmt2`-cFx4n<=}ECjlWoe=1Gi*i#64P+Ir
zXeaI1SJFL<!mJ%iV)8<{yCWhRh^fhy6<3!NCs0Juo#R<89F4xSJ|4X4_}G?24=FFR
zyTd9n0kXJ#hM&xKWY_iOauDn9=QPsL#8;B?U)Fk<!<4%j(j&UcD!a9$eML4WE!N}o
z{U1*6%a!s4(QbG9hsE@;l&Z%iqCSWIvOhuBUHL<I?B51XKIv3mMGuXmJ62}UloFH6
zqh84Ojfu+#*e>J~47i+)i0_#xYZXw8E4di-0jfYS9%Xp&cyGzrCUi%Z;_{^9TntBB
z#eguo_FwnY0!5SUfX@Ni5=YK%UcR|ac?t6uYnLt3;FGtXcv)-K<3FV+wz#<IP}Nze
zgZ7e$J&hZfBy_1)7iPKWx|{t+iXQ(SDdcJ=rsENwghIVWsh$+z57j+JlVFBLFefX?
z0u~caakDb?Hs^X;@6Bh>lknexK<+;t(^m$3j;b<j*0tv>5p_-p_lla21a7b+PdeXy
zQ`lV0m9aqqk4dU4XhFKE@kEmRV|XL-&%z0G4dO9|6NRF(Q1|Cfd&3wl@)!KU`zS3I
zS4oR{PZvVI9}ll`0$kURZp7Pv{63?e086kyAig3MooNbkJx<!zKM7z7<M;*y{FEPI
zpx60YXvXAsCr07?HOSBu1@DuT_cnPvSSsrPsy*72jj*AM?pa8x`i#!_xI!#HU};aZ
zgHk*+<~K0)kn=iUFW6bkP#=y)#&hW{7Y@X|0?xrxWP)=%R9Dde7Bw@nJnBvW(k(t_
zVqzhxndHKnni}B-JxQl%P`VKTmMogw8bnS^N=TYxL$5Saj1S=#uZ<m@UBX0N^=LPS
zH7v$+?LS_}TVkU#`X@XdZr4kT(Z|gT=_OKcEQlIIJTiRBq^NtP>019SRIz~i0;kTN
z&|#xgR+uRoN(_G^o`6+NvMHeQa6Xz6-IiJj=L$m|!;*3|<fUU(FDp9}2mkE+^h1?4
zW!uKy#o2r$+W^w1a(W;8y6fcw_wMTr_}jGN?F<*x(+I4gd+3}3kzBR_V#beKmO&}p
zaLB~O%*;Qh$3XF%hk`-SVB-GnzP{4vPU5W5>7f&9@US|gj0FS->H|(tKS`lZWNdZ3
z#Vyr2rt*axuX0v@vs=jn1tP<1zdMl$ggi=I(KALhu}dX@4f9i4IWjV8;)u&AkTOxZ
z8655G0C>0OJLFhJ`5DIA3PxG6X&V>2IWF@(_ZUW?cDef_nA~(Q*mr}_ei9KC9(>$a
z7~zWR{V+XJkq>uo8$gt(vc>KAI<gr)W8iah#;m%0`8rRH{alyLN6Iyk`w;xLCS!vV
zq{;SKWj5rp7Q_)5Dh2b4^Q}zNQnDk<nhO~LpCyJsOy(D)IOKcVS|+|VTS;OB(u-*2
zm|WytFc_0!S51$Y@=uBTqN#JfwC0vp`KsSJETt`e!~fB#0avTdjrMrX*wj)`ZR>&8
z{3<FWY}8z7>rH|E*~!t>e4=wD=7;HVf0ZI1+&Pw2%<Vaq_xd+88Y`W_3T@HB35#Z=
zh9g5`1z-f*5S~2#zOhvjN%e!^O-45H{g`Ki3Fz{A{5ai*KIYY$u8>DG|EaRQ!nJ$@
zj*78(b)JSg58#PB%BcB7sSw$Yi8P+tkr-9LfX>OuAs{e=EKO0~BM<i5BuMDw3Jz;G
zH4w}3gfx53C>oK_Q@Yr0^`4!S*@siTn>f!pW-#owxV9v_MxSWh7xM+NsH~&_pZDt9
ziou;?q##vlzDXk)1|JD=AnFm>WtBPQ2|eTiSq8;d2ONw;2Bn$rX}HJ4ARr<<JU+OI
zBu7Fx%ZN-_?H}NRu&!(~p8q9a%>qx$oG5eH)mfZRcR1&DdE1M9pV|KWd<7?QBob=#
z`T3qiE8(117IlzRtdz1B0A*YHd}u|8MDAtcf;x>DCS?XmLE7!`wQ^u{K!fHxCG+qo
z?h%D{4cLW%rfdA1pIzi4NxbedTS1$tzfr2r{qs7s2_N3dkgcMZfP#{$tIX+jG@!Ar
z*)o_w`QzVWe^!W&Y{V=%)Q)-Jbz*W;Q*<<aqNm4oz`u%@lu7<M^^G-UF~cm)h4Rq!
z*oPKQb=<o-G_nDXeI;dColyeqLX_ao%{^w|_b)!$M*E$lV7@SUGu?5+=k9D!C?7~P
zeT*678*^E*3g<bAcs9f5%LtS>sTTr`NWs_cgc27kuB2mR$x4+&)xk6%yZu7#_s?UJ
zMwbikkefo-=|0@YiTh?thhy(@TFWB$`111gR~mhoNC64JwMtdWX>ht6IDIU|o&8db
z06dC!6~i&iBCefHZpI=p_wL+e3@tD;>V})3!M}y~k&ExOoQ@(vYYqp#SQt?i!1Woa
zEt5h>&-%<5bvqyL&Zl1sqDBJl?@kAO{?roda#iyVFN;QoH)(})atB87nWIAQ{8&rG
zM<*|_9_`<~F%<I#-^ip|i`s+1mdHX=1t3UBc^D~aK^y<DsSG+NpXm9KBu#e3aQBg*
zq279@M9Af{S!nARG3`FJ5h)SrZ{l3pUdoB6b}Wx;ltYBb2E(g+Rwe#Q!zdF68Zqy~
zU0x7U)X~GExo*ab8tx}<rz**W)&+5FW>ZqjWeK6V2bxO+OS;uSUT-(V!ldG*06^!9
z-{Fqv|8f8nL;V_*biRWVv9Y_tRr|*R80`PM?6?j1_`pnjQ34|WL?$kB|LRTb2`FH;
zc_+_6Qztm_B^oyu&#IS7-BH8_oeWFtrW}wuc=gTv#MNB^y-0uX9Kgp2G5FKy?$r8+
zw8K2Wf-MDj0+6E!r$VK0))aC^0;)Eagc>|m-^0MuiOi0KbuAr4WV|${Lhd$FXi%yt
zwL007_Us<<KK#SNdT<HLZVDi!C`%40q#(UlGEZsHXJ}gBpn~{So3zM~7(}e;!c_Jq
z&0hX_&-CwO>m(=KWS|ms4I~Aay1*zhI_fk&p0wy$1CLnr(6$bzqm`p4fDEKcZCGPY
zx88X_tPac^*h*3)#!#yH&<jt)`X7{AeHdY9g)Xr^#UEzvfBy|PCBm*O1uCcEo|N?c
zrX!bo#b?F9lommR(oug6FU3;IXm^|cYT{xh+U~TlSbvDb1cnudB|;%zy($|>1*exr
z)(|9`{jf1D-K{a|)p9Eq02RWj_$Cdst1_rxwLc`Y;7kt`6q`v379t>?4@v^zg$cmO
zZKyI>oFPweP~)$eP3Dvfo>ewVsDn#A<P9%t;x@}ku4fGUCGS3?r92W?-p_RQn?wm8
zOGZs021&x9Lh>3Y=U3;MDbVEeft<+USErlg)(f(LTC)T<Zp<hoKse@pCa@|wDjNyV
zE2oGX$0a_=ouE&s6Vgt63m>(@h{@(~?XUW6Fv^@K*y!>8mI{9?Z=a)VpqY`IhC(Q>
zb+0PEBoRWhipY^{;_No2tTM~m;CzP^B|9MtbMM>;K__$@Fn|s>NJWaO;w(dpi&qzV
z#Rjo}Rqmcn3iaLz7{-v9Xx^ga_}$gu_u1nx_8z_LDCGP0&X}b}ywzmy><YvJs>zvh
z{5IrS$w!Rr5CxzUrDv(-;9_5lb1Wrb_q#}YzC@|P;0;?-iAAooy?A$UuVjdMC=vf7
z@noWq7i4k@wLT{ezc|>4daG<&Pp@5T^PA~MmqPk=xz*zZ6iZ;*&*N9!sYMpeg38OP
zW0ru9iu81vjVfakSzapN2;u@`I@YWM=xqC9T?FzixO@Cv4`y(C57N0rTK5M+;0oyD
zUiZwXLPfMm1meBXg7%`Io`b~gA24=WTS55!z#6<Rho=iDDByX7bT&*oGi{z6E1=#2
z9Uy06W32{wZVXjQyN*mtYoIayDGWY35b?i(lSR;llp(mai3p0Wy8kQ|6|UV74kl$J
zRZN3mIX)CI<7$>)W@Q7K)Tm@V{&nd5F_~Yw<l}q3T3=pl@wm<t@Znu%1d(_WrbZ7?
z?R&!(aLyr!$WlVBXGO*iugU5OE%uPshf+8^CXoPSM3$$tDDP2I$8VgRhT=AMf$0Iy
zQMrA&cJTv_1P3Y2#gx_-lC`KuW<!B4=P&K$B~H0AO$*7kEIRo3mokI_5Ga>qr05xj
zS{plqf2?&4Wb6a;x>AZ!jbbDUN#>Xu9E%mG5g<!N<-Xg%HC(Z7LX%b60JbnUN?I-y
zmO_AXSUbEiQFZpZroE8wN#9qkKyK$RyY0nQ_njLc8@c1THqRg({nS$Se8)A`BGHUh
zJ=uZKIfBj_k9{-Pbj}PcQczeT3iwZ3fm(P5?%#TheC4ank)SUN0w)T5km&Y?Jz3$0
z7+Lz<l@7bV{a5RCeIx{+EQVXX-bXvr?f|)Cc?Lx|7Z!C4f*-~XqTf>D!W73EKS=z7
zV&Svt=Rr22-=7n6KFCzz#Cs)~_eXaAptz}B_$D~W21EnXUjK0BP{Uu$bamXh(^<fd
zJt3&D78H#dZNA%qD~!1U{!Z6%Q?^{!7rZWKzB8BMR;|o-V@RzU#S*E)##SrJc7U|E
z(1j6c0^g(wZ_OKf$ycC^ie;!x=k589?OzC5JR|JI0Xm6>cd9xk3NL)@2np~zQ!4o2
zBnWgcP;_bV+4rw)0((oI(GaI1ZwHWVsCK%ny%zevn&V^R2)6S-TcbY4r9){5w)uO_
z)&rZk;?N}|ZHn?+ps;90_@%`d0T!#w1{K2G8;l$+=B75f8J+D=e0aprrNo37*;;(A
zt50SEGhbfM2TP^s??6dMqvd)JaGEUO|9+h-Mvu?#7uz#W3&CG6Pv}dQFxW^%b29tc
z$vd@IB6$ENs%}yBdwEpa3I#m5=&+eFf}FWG>?Q3F5x>X&DK$cu*RSe#z)jpDd>OfT
zdwUqhGjVWtcsV&P*&km~_!S)DjjoDPI)TYyaokn@_+d4cz1U^4s8{C432(iSeIZl(
z?Z`W1m@1jjmU(RZ6`!*;wUv(R?RJyH0S&z;qE4s#Udp_^?<5DlizWPK2JlO7Y~qW4
z4oOdDJKvB|)S5YULlszA0B>KMG>pr4*3dl0II9y|5byKBs@&4Gp!e=llTtp-&eZ7W
zGO*g$Ut3eoL%p<hEFtTE^eY+7j}VWT42tR~%e+sob+E0P1ZucuQ%93^(H_eRz?!XG
zX2OEg@R_geSt8>NfO!_x@Y<4Ei!Wg4R?!eN>)%ssbz1B1l?5+A$@@q#mmg(bng&^{
zON1uE54RlYv@9sX_`P)Xr9W;@5*>DQs$j)BFYw@mh#D+Pg#;M&k-x*remV9|_D64@
z(zcXzWXegq%!UAQ^^Th!cXyk|G?&s(N4tk+otjhmY)sUGV^gwPDk;UiOgln^uqZo(
zni}gY7T$g`-YM~b?glzrO<D5w)PvGW8Sc0SG6Vtq`1*r^6elG;zc3o;^Y}0oJ$-X@
zLg1oCo}M9>#eeW!KIH4`jRr4+(K$%h^xxa=FB^~He$;tT{wgkKd#26yIFfoKn2KtA
z$XKxE+|a#kIaP5DEj)?d3P0mQyp>Nr7E!FP@vn-(i2?xJkkG#bc?|HcPcEuxFA1}G
zk@L^nzxU7~+q|Ejwt(@EwCM_Zc{s=>DxSiI-UI(|VX^a}gJyY_aK^uiFprJwvnf`A
z6grLA6l-COxAiymZm=<bqs}F|1}uZF1Vwfm#&NxdHQZz8kAa$=Uw=p3tF;&m76cua
z_FDLm$dUkd*F$1kq?Zo^!kQv^C**-LgQ|*>Y7hkj9R&@9*EAf;rq+16yllT2<q;m6
z7c>GEB)X2XbV-I0bSLS${pO&Vzi@y}<<OKEg2e8XsMt=V93TRF_pee3X8$@@on!u`
z?lKe@PJx){gM^vC#ugy@l}wsc4bmB+oDh~7s}+E7*3DG%xB394qiX=P%!&61$*?f7
zA+j>FJ+w5H6P=1v0lZ>y;HlNi)oSKvZUcz=v^P=L75y!M#OscgfWJAEp%EdOnwPc9
zt(@@kCN0DAc%8JDl-`x>+1vX~90v#Igxr8&PipL+a=j5s{Set56Irdp6;YzbVx#qZ
zEjVCw#-RKDg>!-rt-S*j49DH8fukI-P&e`J7$JazibZg8ynlK%&>XO5pdY}=Mn1uD
z*G1MMy^`2gVJcPe6Il;UG2SdDAt$Xp*~raX)4^Xw#>>Lpelz4POeQ(IjDzbB3TtY_
zkG%;Y_ecNj$nxO?P?u8RU?a%JtAq|3_8O3G+E*Bqdm?!kMQe?>+V(QoJ2V?NCIF{>
z`p<u{(J9>diE)Hu2NVsMgc4a_;;OCpf2!Jqj?eG*z5kl$qPb0YJ{FMr!(>w{4e#8{
zu@C|n9y3aNk&ZwC6kr8LR=_nz9uN6T)ucNjOa%>s9;W7nCM&hoHI??<VTZAbZvw0i
zcCfPzcxpSKTm&0TO#KM&k}7Jdh>WzhS=tlTh1=(>_H6h#iulCvQn8G2EbW93@@Up5
zLcDg9$_}m+0;wuVK;_6MeCFs=e4`9<2`(-#FA`ed+BqPuIQ)!1g8K9^z(<WCe(?F;
zZCK3V=Y?qtcKbSs;jd8W;#kv-M*;4hD<OEeQISx0cCGDeev?ISi1&3yD<~u;QCgaH
z%(Esz=>)-*?O_w>QCMK*3LtIS?Lj&^IOORI^;Ff+0DJoMb@k0f9l!Q~uPv?wI2u2S
zyrpg_S^6EB;|pQ^jWWr%2#m-SlM2w_9-a|<*ebmL&k}g|!9pk}0{&-3EdU{VCAxB`
zWtvuu3qbo<+>+AnkJ)2E^AAacU}AxhtuyT|toQ!rwAv$|Mx(P_iOKDEo<=h6UP1K1
z;TQ67HfSshjH6Uz<4X@!(i7<OfZz_dusaJ!@z1rn(rE0~8K6be{V3(_Y_2P2=p<pR
z?O?KNZ{il+ipdZFy(H#GO{8!HDQ+F1Ur?8x&F<`+jcZme<K1T*0gcf7+e=neRR2Ha
zwzaf!L2rIMS0Z4d*ziuzya5y$0j2?nJo%|#58JhSEhIW{fzkSOTkP7IuEeNt^F!<G
z@`7V*7jf4#dO-woc2V%7m)3l~PB5z+?fTo=PzBe*-V8u5)m)U0a$xGLV;xy!Ghc>U
z=j(c(QWn;6cDL76(_GSg5rjr|8LPI=dqcQAn{w$$HtfP$?q`+IiZ+(o-pRen-oKfa
zYAM?wq5)sc0wc}VoZxArZ7uC&9MH>rurZbDxerQWS|N6B{(;>;3>rr&CT;JshrR}!
zbt&pJzXD*|P!pm)Qnc>}->rTz*)x@~ru7~LtL(yv<cEx*18+iFBu6~S^cY2OAbd9e
z1{*mFOO&)zppaNG$)&!5FV?fHtlO3rcX#xkwj&y8=~&oD*jnn#obs7VydJAiC;KJ=
zuAYl2Il`{E6v_fYVNPT;FqLBVcKywPvAvOrpn$(O0UQ71=qo_ni7pLLiq__~7Dw0r
zCJP=CUL?{oiNc?pS=)Rx3g&V1h7YoUU(vmjKy>l^LPY!iSX@WW7gBZ!^H8!k+R(<?
zcVJlV)FV0&o&?*oodwUyZsh;zDusBt-$e~uuitf$m6x1s#=zw%K~rf<ZeBw6lpBJd
zq6^p<Ha9;$!p2BV(NNa_60vJLvOJNfK!PowK0@V|9JW37$gGJr1sXdjnO9WvriOEn
zR260Ph}A1(gc#hRLYJw+gT+t9?R7CRHKhNb=|LA&G)b-S|6_!{ib2I38B!>e-ogkb
zaI$JVH$wTd^|Ze%@*O104_npaXZklt<`t001hq3L&(sOfAi+r{TD`?fcqX}C+4a+Z
z#35&ZyMg-GDNLcO?GNHXXn(Ggpf4yxX#g*EU0<;ha?m`P^HDSx<;>}hPR-RK@0crQ
z?&md{^&9A1>-K(L<&*wc_&~9hh^iQ+77iBXmKmmpwBWQS@H~nt%p++p0qf!wPzF@}
zV535mkv1k;DL+MZ<fXN+vzfYJd}XBxWJz<Pm_T9fs?-zLe2h8dQ`{h_eJN)Kw!ieQ
zcDvWP;vIheZ!3v9bhWgXE`>y$vstFj$<+^0LWcAz(N8|+_QgOGZzD3|0Di962`{_2
zO#J#`Y&^H2CAm{cSlXOH|Dsd_(2|^GHK|i?eHs<VdH%y;tM&Qj6u86G<@tTRS!`Q|
z8MaC{*b<jUS2Bs(SIb>d9v!-+E-8+_r=_e<3`**Z3yG;<K+X>LRa%Xh%xh0bXke^r
zZSxZ)z7%amMQ4S(?O`{PgC;PgqW&gQOWxtKCbd&>lS1*W9HgBL?fY&Oun+CwV93{H
zVrZaQ5!~wOc%CML^zV04iqr;;4>q%l!zfor{9T@Vd7}xgxQei4!YqGnS3N^f`|sjT
zM`dw+mtW3{A+*2k|1Q;Rw|SjE<#s;15%Rbf9$x_yN)bbKxrP`f<#3%GNp_JYBo&$E
zU*s}#!!uZhAYeP_c4M2l0F)9VgGhgr#RU-wG1QT6gaeJEEBrl3{wrqQnSK~mUk8fx
z1PV9G)}C8P^rM<Hen0rOo&SIo0skI$=sX|$)>ik~gEF#;JSD%X!Eho!_U?YZ9X(xr
z?fOq^LRHO?q;AsRr?8gb%|U3Dh*ty%jY8w`49S&S*@2QhTU*5gtgs7QPfxC}@sc%_
z7g@UPhr<kVeL9~Afiu{#26k+32La!wqy66JxZl|-Xom%gQ6*zzw4#|z12YN>c!RX5
z3>W574J=YfrD=KME#u6EAV~&WFd&jS+NvP6Vs}zQhu~@Hd!|K=bRA{#z%y+L$wN6<
z{)0xjYnG$A8Cf1c=aBKoawZcfP!%MNdh;NT9O?y{3$CW}{=RGe(JY5rser%#-mdxu
zjvt{6D!CP4DHBQI&dvSqR6k(+v*7Q*4aSySW7a;|?4Tx+r+1X06nBS&1!~s<7YFMf
zm%Ex$VCX29+xc$cl-+8^-Kgi+d;fnevu~g4m$#)k9BKd~8cU(HKEJX#+)WIj2wA!?
zK60%qtFp$R8?LN_@d=9qtAsr}^mYA^lTm$68rDz_h%`@7>|Ic9Ua*$*2Xm4rX+K1<
zr?@!?!UaK+?vHkRgXMNUWFi5-FF3UCugCjQpsb?T#l&$mrpjE1UL%7!=GBGli{7+`
zMr?bDo*&S{%{Vtol==NoHjj9`gPp8KIu{LPo#4&y`Z7WRy)H9i9*p~M+AO`!uzPKO
z-t`hM0WP7LIPnFsX<}dB9R-VM5x-mEOd632c2$s}#^$>ts6-hVUtk4kK3FL0ln+Z2
z+VY=yJqJ59=h#sE5oR*EqAK5WM9c&&JlDl_&pRNYZO^FND<19pWb1dH+~$8rSNJR9
zHrK-@G_@zm5Q-Bk3oj>cAEf80Z_RX6k;`WsMrgoa(W%MV0kH(+mOM|`=xtAe;+N8t
z$KN@hWb;u!_13iPg*>l-ug~;4{mXWLpjrrs(GUcxzlMQKqpZ6@LJk=$h~Nq=VaZE5
zz(43MB=ki{m3LGw1Vg-7tP~BdhI=^hcTb>C9TmxY4q22)bbzFbuCM}@<);2+a)pEo
z-_Js3YZoQ>F;|rz&WV7`zplvxfqTY|*J<~wn7X8T=lHoUPq54N6UYqaOOvOKyyvjE
z3F;)cdwkU(g%(ja!TQCt6^Q!fAg^W*prjR<YtvE}jgrtCdc-SJsE@D*DGR;B+`i?G
z@K8Y4{6h<=k!;DJ-BNn~QpvsvbU>cA`zOcYOZ}Mqt$^IDatJ+(4$_mHUBuja)aI3(
zM1}fQ!%lo{>6^oVsBRI$BjO)afzakv5htvpAh*4+wuLCp5!6k<;HFU$nt(C}F!n3Q
zh&QbP37YSOrodR3(r$hHCs9-e%`cGn;=!E1(d+&=kh!mirN|WJ%%!p;qNvmxz!p9T
z0v2L%L0}+|phyz2JnQ46fYnb}jcXf71m)ZxJs@huK`Y)zPBF{H6we)wDqCJq)noDW
zpS6UL-+S+&9X{R1VWY>Z`*;FS3cqO;S>Dsmu?56Gr{`-bp9=I2wRy{Wjjw<O3+a%(
zriEo8?DgR)P8_=2lRzp0VZ*NYe9+Wdzz$b0h!VAJ!Uc*H=Ou|>X|p_6ctQ-E@i%BZ
zukGOzd3vBw@Fa&&ug~qvmAezt*2)i7p}9rrJoxK=fChKcu5c~O>;nK(UqE`B8(oe>
zmPU492>e|Ig;O9~laZ`-mMx?kx`<$pZbT9E)@u=Q3h4JbU8(==`VHI=?_hr1Xmd?&
z+fyVK@-?2R(rH=UWeU677`ud{bGbGAtES7XB&@HATanGgzy;Qvud&G4z{>tlJ*+&X
ztJL_2o@|OFl*P}ls~hkSWqE6D0pt^&6zY_In8*w`poOu?(wXIsr*W+12Q%OzQT6o)
zre-hm;LGC)wfV}Ks5e%H(~_Hiy@y63hr&Q$VVsDN>UMO+V9>1MX0YdWgKcCkOkmn(
zlWhgV?6IpZt2(j?Agy^?PCV!UrJ<MFozCZ1Oenk^-uGLv^ngujWRFXrxYbb1xrkUz
zDhu7Y1+p(9L{LuKhzfBEB~^>7+4(kF8t%T8Yc>HBZ;)*QM8dD=VINm+;?&jI8J&~b
z>CZs2DLIJCj<p<S*-$Y&tF#d<)jwc|Dl4c@%fX$!(&6a>(o_*RC3xCdO1%R5mVPxa
znTK(o$+&{#0(P{VD4orPDa0YXMJ>wYZ%fQg&MhwwFfmdwuIwG7dW<%*{*q1eGu!B~
zV-U?b_BFLxxMh4rN}wONGHN%wy-sM@E<0{^UyoJZ1Izzj=i7%}0lj4=TidF!2q$)}
zbR4-l%NL*LA1hN5+9K-%lq^50fyn7)qQzrvH%uaJz=E-hiwfz;zXm7Rh;f>6nMl~%
zU#+aOOHdEFS#&lET)>BsD8PkU+Bk~qK^>QHhtc)VZQB(YOp*S`E(Zpvd0pMMYY*qE
z=YlWXOGZqyChDT-U}S|NKq<7)vfxmtdT)jpiu&(c*_B1}K#Qee2<E7FsWvOdC>e-~
z9<O8oo(G6I18k!TMy){+sVL`v+==*S{T`PF@sHy@><zj-pHiwWw1xP@C{YjmKpkz3
zqBctc=OYqDquze`H;8BI=b~TQ|E8RZtpz-+%nIw56vRMy4{5|)e3~}W;Z?&2c$m~8
zibm40yb{W{H<4lx255=u%uH~SP<7W7x}P2aKCK*-Q+T95fG#?w9Dea+^A-T14Av+|
z#O-UOh@x;Li95w1ZWR>Ej0~mhZR3NJ2RtHeEvcsrxoDAFSd&-m+(+5golaL2)m1|%
zqCN2bYSh_bHhfna&G+MK#p7Zlbv?c|ADn)6NhX!Yn4*l5IwC`5jJxp92VQLk1kMP6
z3H85`o>JX3g2xZ_ozgy&q2Taocy1QuE6_0JLWe7?M3vd*f>K_t>6tC*%sauwC9%I9
zB5$(c#)NX1;CevHP)^lIJAEr5zelw1F6nJu{zkj^)ytPbHWo)3#oL;76yR^It&=Ga
z#3v<E{|S^Ub#KMP-dfs^3so`P(;EQ%lV;nO<PE*vOPRXj^@<!@Q947$AM>d+0jX*I
z_RoX2Ouy{(>7O(+F^FzWlHHFiaQmze_zqP=$PY4f`5nG96Q`-PJ!y1cH9|!~q%nzJ
zsGFLqf0vjmo1OcQLe(ZBu%o$EDiPZoFoyzs1G|G2`Foo?8oH5{kIGu>w!8ob1G(u1
zE74;~b%BI_=GU$6_jqvo-HeM8QeS6X-#-Y!2&zS(J!zCF5(@}!OQ?=C6!Kc!MqawK
z9r1xH&kW8oE_h*Ix7*6PEEQ!&qKd*9R47^>Klj{?ptx&VOKn@-fRKUHqrp6O=+;0j
zPhiW7nMUnyR_8mpl?9g_D;>c9qKvA+`5ZBgE981OnN<UWO%D0v(mAUy;8#b41tT!_
z2?u2bU#2dKTSz6C6>TK0YN#XTfwJvsM@<V%ik}@|=c=d*`R%S5NUfrOpUqCmk*s%m
zPltSHE5UQlq8iVC$)Ot|CZ?lY{4AuTW=RTGf)Ih81R^Qt<*4LsjR2gz4P@Xsp8@Yo
zLB7;&+%oHtf(G)m6Zkrpu+cJAxO{z-d|po*?e|ABj<&XrzzdY^abYrc?Pl>y{}e;7
zU5eYz*uViI0^%D=Z`u-zE36^|&dgozVBj~B_j;fKb4i8OK4GTHk9L0ciBZjbQvRr5
zEoF6CLm5AmYtJe?TmYvo3`8r~ej`0q3zN>a*9Dp0o{?ae^X);P0X`5e>sl-@sUacx
zFw%(RVlGNj=;f|c*x;FY2F$E*)jCf%0=1KhG>X7=c?fI#MgS05Q#n63>LtHD)Lb0F
zyAY8$-frQJ7&|p@Ii0uO>NMQ(+uYsbb#ixl3aG2HAxb9%?3)$Sk1A0Sf`Oa{6``?3
zl?*b@63@kg>yEvIP#<c}g$T47DV|>stjB#^y_PQiGtq;PY^FM612f;IrLE?U&=5G1
z4Z=?Hkv8?`Y*gp>KV?$<7)?j6HRA4Z9DO@v|Jhh9T>hSZO1dL_^HWHxtTL`Kh9#+J
znT_QSW%p_sfLXiAOo1<>J+v*EDA)KXtY;2&&}QolJ3rOI<d&Ccl0w++M_rrg$55F~
z(>j(%&Gwqt&daLYGIK7^T#K35s42zLrEx*uOrtna;FW1uM=h_|5q&{wR$k@rXeDVk
z*`aT<>+1c+iKW(fGio?#_#iir|NThQ&=rtH!2wnxGa{nMq6$n&3JZ#8qKu(`#q+R_
zI&Rutb`uleN?9#9PG9G~=RXrZ(#*JCwmcCo`x(^k&~AA6h10%*xPPz*b+FYehrS1Z
zr-p#=$%2ib`Mh6ijk0IMe5jsLO@lh?yhb<J-%;XuhJ4I#(I2nlo&B=cNlK3iWeMBM
z{sKsWIf4fo_)>(tvT8^h%q)zI3sDKwc`8UuHTbPBrA<yoz&f@5Skg?x+THI{#T5VA
zna@w{+74rBltxlyav(Ak^amB0l)Q>N>ky0mZN=-ulw)rz*a%cn!#ylOubxV~G$5#+
ziqx@cFDUNdv8OC@LF|&L$>~vZ>6!hsxB~102$H~M<;})+SoMSU;EW-`HuYgiYgzl&
z0wGeb`F>qNA(VtcqMLI}Z3dy2PXVb>Quby(&xBz71h3^DO!<7y5St#9Jdt0mysG?S
znUB}Rm{}-4W$)(dwPJ>Q8ZcC~=bGF9{<x|Va&GJJMdaoV^<2yTS?rkQhXb48;0H*G
zV57xhj6S$N2G2&Ej(f`MZ%B4MVx-;da8=qYG^19`Kj#HJEku?TKf3$;#WI)RK9_TC
zx<HevzYnIa$qa5Mx)_edpGK>OP$XQ^e0L)m`nVT^h4lSbfNItrkqw35y?aKY1A=hX
zV^o*-lPwN(+ZKg(MyKuPeC_9yl{Wz9#%a0N%cb+!?r)*M`b1P8gwFdn^i4)#lte5Q
z94wfG{QB73EI8py=CJLbOI!YV7blO`vm~BX+TE?Ek)Lxx7X_|t7(OBXy56dKwSyv9
zGK2?8`^BQ40=btAF>cNV?49P9tp+}*H;mLXZh1g&*Fj`jnVE9Na6A6EOA}zSSuzeg
zMSZr(>F(1F60hy?CyDIGDMY+Y5^`BmXufe8{BQGwv7JEDeWm8-O^=-E*@=$}^h0<1
zE^_Lbfh=@JD<5xAPnTt(o=v|pT)(E$`zABL8V<_#aNoeke}53|6u=N&<a1Kg#h;tH
zvyH6LItq;E4^4gM`_Jy?N<#r&mAiRMye^{q+BF$U*$Ut%%vhAoZ>>)DCBslYBArIE
zgOi-xJXX7Fml{6jXZTrqN**}^u4SJOr)t#ePia`j@U~0a>`{8+hU=5#TySZvht$sT
zraUB>_&*1;jRH8y&hWvtu{ztT2x5{2kg6aE*%GEj!hA=)_y+~iR5$>@kI~<LwY583
zF{e`=JU;GYUoU+P|M_~Y%iB=1IlVLtVi|d!|BYK4gNM#NYCJ=u5vzJn-_DVz*<-Y9
zG5Q|ZoQz2c8AT*HgrjQ`S~M;>f1U-K%Q|X{`Ku6n(>SY!u?um2azYzwOIL0Cv`lPo
zXHyKLor~QOmx65tr3;{YYKSP5;XGC8K^mz-^DV}uisx%W{EkQLGD66sOHPR;0N-nZ
z*9xNWS%0PXK)pGr1$>@ZE>{BSp0PXBXEwD=aRJ}^d5h1m8`PBKSRy7WxXMr-Prw0u
z?vHYF#ELzF%IYfZ-k;993=l!xg=`KrK{OW?vlQLLvJ<>_?0|W`#eEaN+Bh?tyQHLO
z2D7Dkp&m1Dwef3ORtNZFcs!BqPoCMerC(?;o|o?D$E9ysv=2PD<hw)J#YuD9nRxvT
zD_B29bcG?V=9fwxUuFpL)}YUz2^hfm@j(SyOx8X$U|z_7A2V%x-94N(S<U2=zg!!h
zTJvbPc)2?(x&Z2<8=`2AkF!#@sZXl&tOI1}ZE$sYyrIM?N|a`d+UzcmChi`q)0r5?
z{*ElwywN40uZ-^YV{2oe@N}5eNa&7i3G0iXm;M4x96zBX-nqOox823V5GplZ9jOuv
zk0%gV`{#6_ZDyvXrsnD@CnvXfwi=m7<+4#j3}>$Rj}Nf5iaj|{5;78RTxn%;fJq6V
zgU(OYa8f>~1f6^XIGXXTHvZLRv!k`o<SsSyTPj|L-^=BEPX>iYPd^}AD~gOD7{GD$
zohdA;tSl{Uy(cbH(cjm{PQ{9-we>&XS9uxrs01+DmVH_^A-}h8a-g^>9UNm}Gsv=U
zUSClN$aa*rmX>yMa#B`Sw)4LFz{K|;vUYIz$-m_EFqSwdug;K(dhfhUSZxqsmjwA6
z2*}pgbP!3IZb-tgohP)Ky`lR?@QFOVRCevXma#XZ0f6`_?>#`;w(i^xm1ZNKv;kU*
z5BsJO|H4~QgnMfjPDXx)kFNu_wYYv`qoqp$Brv)T28&J6ljhkP8`<ZR_KF6AKDtP_
zM!hg;5bp1s5^s;SV+4O^f|w%`ftom5hn?f<t?+huy+VdcjZ!O3Ph|HtyPaQw$CfL$
zic$eicD_0~I&?AKpX{}{DF<Yfs+8OzdAe|LR8J}<&`K{<y7uTKcQ!DwkB*_@q~>A(
zS%^&wqiwqWc=Gu98|d4qJ<}9GuSa>o5H9>O+kT|a^VL@O`nu2Ybg7!ub6Oo!Rwi&4
z-n@!bz{Ktw@#Y#=76q;`WuLby!%e;>w>v0^v|ctUKfb)WDVUK4?(`>`2&#%H-#Dyj
z1C6$sr{$@rf|#7tOyAfLkTk$T=QRPC9H6xQ;jrHQeRBH*f4sJ?8%HRp-C5q@{d`qr
z(D_TVHY*t`hA|}57~>pf-*Ew@sOS_eg~i^*x2?CcyIn9k=o1$Y27O^N@7)z}qzB2#
zCafa>lFw0G(d@h2Gst*}T}L+NxZUG^e-wLutcRqY#+b>@Njfw*BCHeC+baPGyCzss
z(i78AQIypek03U@Qyzsj=0UZB=EJ9lXK%Ee<bJ0a7iqErBQ}$otg69zr(RJColXd=
zbQQMK{Mu{G^?4?C;0IMDNuSOaob_yTn3<j+T>qnC$GF|-<?H|)^_Ss5pNALeJfI;&
z=V7Vk2rvqeqG9mIWI!PeqPzm1pZ2e5_($e-d+a0|zE~eL^&v*xI$U4=RTo&rcANj@
zS?b?QM4!J}jAJsx*#dV(U1o1O25-|QaDhB&1*Wo?cUS&~KzmAkR9R5!K;K^-MG+ZD
zq{iB`tRw<W91sf^!+$VVme%H{qPWZkAOIyz+4Uu6QjgiPtbOWr*gt@$>6Y5K))Ame
zQiEiS%U;{4)#37^a@O$Zr=LL}t|FtW@M_4xBjLw>6ZV19jC2&-(+EGEsg3+%5gCI)
ziPAo~t@#6m-^}NWU>`moB=vDx3g`_>`ap%z)oIVdqHeG4=ibMpc8v?w{h}_#Hd%c`
zd}#wyJMU>ae=EX2<rDx~85TJt1cY@cfL+YMDGL>IpqG*)fK3c|f$!RjPOZS;L}XMD
zk(SZ5Bg-#pcf8w}>$c}|*goy{`R*T-ioe!B<zH0q^?;r{S`OOZnOn6nsim4A^Mk69
z7YevNEuj37#*~wsoJxRxcI{wQ^qWR7&<x_zgk+kH>w|-)tOCIO`|_bUzynEUoKeR}
zCG-=6*<^*;-Qa|OSzWKeTX5oH!+y3rfkdxr`&R0Vxt=FFxg`>bD4d1aOUl*2VuRiy
zEM_Pf>hNr(K}As%hh<a{V<2efn>ERL1~sZP6%|;5us_Jqug<#M|M~uUI%hU6d3ysE
zDVGY+9*&R}IRFbeEKFv77RJ2#PQhohICK?GaP}8fva)0p%nK9SlE74Xo27rp2eP<)
zE-1oamnGLHi<2lpus2@OKkUGF)~A0Nbc5CBUvtm<n?H|~HvI>KRdDa$3I80e%wgRH
zF|nyBrb3=h+v)iYr0!o1+4ZxMLpzA9JpN$$W8pSS1t{v%TLMc7nkYyl@Ev$LWtmeM
z25eii8Z55>fdV;0GE?rJy}6Ejt-U|bJ1pR_C-aTC9k({xwcB)*3?snuelDwh2aYmT
zd>y4OfW%pB6%DMivn87ig7IHYxuw+OV(se&MQ*3yj$hBc_md0z8OZCZV3i|Ym^p2q
zz55o+15RmZs?82@6VtRBeJ@9%41ACYBvS1LX9piE3RVqu+3+RvERF3MhXj7D#6uJ5
zdEr{CB|EO+ZB<=w(VvRlnp{hQv)ge(8lvqzm_xJq?n#cLcs9>U#4lb-v|I$SAEhrg
zx3uJOZ_Nb>zV!?6{rIiZX#CWCI&XDwaC9_90l30w3=*y^fZ><7A8Zz4EheL3q8GA~
z)%B&&*3CMvOzPhxVNu(`#7G2M{8CbqlG3iM{0G;d5`%t&r0@#+<~ycjj^4{u;H?3v
z8gud?QMsu!xxlZ!&3>3gC{J?nrvby`0tQ8c<~s1Nw|YFIF%V;xR@9C_O$JSbutZS=
zgitHDKpdt<rS(*%w!_w*ABdQxWQYWeGt_VNO^L!x0&ZPF&k8S+7Z~`>8nJ7FClC*B
zX%e^FEOk^wn1YyDdED8SMopY;tih1yd(AbEN}R}<$$(|?(224TGBti$#kg?&{UE(j
z5?jGV;quH-8kk{JtzK`r7$ke;9#$e?o~7X#PC<HVY&uiGqrb>PE}g+pT|eZiqiZt6
ziI`f>Pj=2~A%Pf~pDG`9m}R=5c*F#l8PQ;rq=?w%hut??=4#+zXAPOHU^A5N&vVf?
z{sg-uDB@-yWu#_|bTn|7mXQrZ`f^q~t_4}QX7JPG#wT10sXzfzq+m>Wu!#W&reT%u
zZeqEVvhh6n1*jDz7S^XHxHy?OgjhR=rc%9ff^V=Q&!WpK6M?DEto@!iW8>0tGJCnG
z7~9AA7Hfk9;ufo^FmGj+^0}<JzjT8wLvk}~>13n<30Y60q@bW*b{jWYl2YMD(ipiq
zH$h~0VijOS(3mVARat304Lks8nbT8i8xS8cg83Wn2$faOktg_)GROv+z=*E2%4A3J
z367Vs7k2hbAL@EYs9`tDx?E{O1l*uc7$-484;OKYLy^UWiT=gELJP#k;gH;tLisw?
z;MaPtKzu=(6(o8BwDE5dn6&}<CqBPY4&snlJ!OP@-HTj(fv!(5u>ru~pr}|-a4;A+
zf3CfJzkezN`M0zeAC*&iUVb=y>Rz$rj$E>mvbKDbb49=t->{PSb46=jVr67QVt8%5
z_5X(n3Ils;<9--wg8!hfR~(bW6)${4{x(uSfS02`&sght3!Noi-CiFY20erg>ljIU
z0`~ub$(&uoUw|;WJzwC7GFcF-f)FFmSaV5`yUA~G!JzO^P?aHmuAKaUPN{}ovJe}H
znU{`rWmC$yo*phk9iTl-Ku<2>TV9uvc80?lhrcOmJXYaoZE9&_e{SsPXk}|_W{)@!
zRI+gxcC`GB4R7MB&1!+S<Q6{5s}?hgGcOL_B+{j153sYdGdH)kh7sz-_{Yqe3ac0J
zZ-RU>7)UAL5Y~c~o6yzGF!Ph`%Bc)$7X;k}vV4|+28gcW@%7tcT)a)3`o_lA)_%!3
ztPv1CZ+5mRaLJ84CEdRh^mMM%Q9P%+S7mg5_~7zK$XiEk`3SeUJ*vEF{U6HyZssNC
z2M3s`0AOdFV_~uIxf1SIPEp(Z%U~!=hOuVa4Q}ypu#P`S%nBxVhxN~MJfNmbK{+Gq
z%$9CE0`4wmkifG>FXWH)F6sz4szXf;2^A@0bwNoH*m+ca9gPBhMAY7EwiO9(NAz9H
z7xJSEL1btn>u@ldPQx$fqR!Ed6eU(?p%Eioz@^y$@OqBI<~2Rx|9?cCV|ZO%xV2;3
zPTIJ!Z8dHh8x0!USg~!hvE7)BZQHi-?f3f5xz7Hv|E{(7T64~4K68wFq<>Z@`sL_B
z{?ZxqYGz8pqNX=d%OCv`SgtQ3(gWWwfF1DB$MySTi@L~K1&_^A2fXhFJ`?K)y!cQ?
zYI>Ts#!rkU9jm@j)rKhwdmQBKy8&vs{(Tam*1NEG(#AO$qG<G7@2HL84Z{r-@e2}V
z&l-zwz-&!H5C&xXdZrbF++Cxvd(xg*j*rQ%3DWxRjhYJQr4|oe&#Xq%q;Jl~buzd-
zbYii_?!&*w4_5Q9cwZn;jxhsZ-$72N!(IKYuKko}rmIhg&;R?H1piPz8BOW${sLO7
z_qPsyC22H>s9Kf8)Hmh@`*Y7N0VthH<}HZ0O!^%*vkSzko{O`H?Bv`}68m$=^qt=g
z@f-yx_Ea>p_+AXpi`l<kH1~vortXQ%EPZBBobfTdUo=d3P7c(Eek9=cDMSFeIKch^
z{qgzx4e{+Wt-He`SIIzHKJ(2#9vV-?VEc+`+f%tunS~Z!X|0iXa)JwhZv#Ajytx1T
z>SZM3p?1-8tyzD_pwnP=-x{^Qsy-o&Xg`&3fX|*T`$H5StLl90Zq8T&np-n@8c8*c
z7G}WiN58kbmSrPpEr*9(FN>1G>I?(-7s7f$--kGyW9#B-cJnt^Y?hB4I4k+D3Un9&
zdQe@tp?FOQTSla8`9k+VfTI6&az9GdO#Duwz@Nr|@($=C3ArttxLJ1RJ9x#3i7WwJ
zk*+eQzWmI7Fl}P<jW1UGy)##-Zao!0<@JodqVA#$GXD;vhC#)prnF}uN4?DxRKvE~
zg@gYMr&pFgk5)_h^Lwuf9=w@}si|>4OpE3zZNx30j=uXD*8NZpXqC-9!eK%S#>QrI
z`<;<^O=J1R)p1c|E2f|PxFSqT)nt|gi2QSJjtD?pBKRpL8qFq)jYK?BOKWpMy#Apc
zvVqCI=@k`yWzAqpBe~F*x^HgTAskM%UL2`0%)c~E)uZrNnY+GHR7PXN6dRk!DacC8
zgvKC)6O@L}iMOKyR|TEBfH7L12-`?WLP9phb>DG*Q6o@R%WFkc9U2*4RhpHRRf!y#
zQAs~GE?0A(QiPz!Ng|Z2sG>L9Kf%VG>?TWobUV{4K+j(IEai}&p*J`;K|Zv&K8!Q;
zw)78WRya6nA3Ir1q|4j{M6ZU10)ww;`S!I}D<{|8ogNzS3{GY$%C$b(B2JG@%1+Ki
zBg7^l*xR?VhVZ7A6MItHycS~3C-dhg9=lgD5qER9vvYKGR9eaXH5mEfNbJm<27MFJ
zj2g&<fNJKsWAr_Z&mZ|uBCF0}4>}8cw`e+#nT%I%L}_qCH=d2t?5Q^Ib0El!&<Y~e
zrPfdko96%l20|kbcn!><zpY6r8#XT|&~ZpMp$zdkyk*TcYoJWQL1v$oF|;x4e`G#$
zu5U4s=4ho#;8kib-+mg<!DzV<?J^@`<Lq|l?@XpbkhcWxH%v_<{0Hri4vUKDd$p!f
zPQd)71Gj8{U~{~8PU@elD4oM9z|lL{qoHd7(F!2#u}501LFD&}L3nknWfE&>vl^?<
z@Iid{@VPjW&mH8}V-v!IA?f&|k)ab=!SW}v%4ZuMpY}TL%ZDE0gW3UAuy{AYj+FP{
z=MWgzqr|iFPQW_7Gxl{kF*n{)GL##;M%LTpTAbu{IJ)%l-0*RlrgDAp;U7a)dkG~y
z#&E!v*%+xgHZZ>^E-9g{8DOTLDjI}oS;Z@nfp_>ZK+|K-5_iOka&`OAzy&?QKM1Fh
zHd}igJ(3El=`jdT4llZm&JtYFxYvF?!0}O^N||ss5`gp{-Ph8Jktp|%cA}F}RVdb)
z&+jGbpM**%wRLt{MzIm4yet~9gJSX2F~DJ-&D>-{?~#MMmcmBGTT2jPIPSPy#8tyw
zGCH0-A~h1x@Prg6>v`r{`y1HH57Jo9cnkv6HFGEO5NgT+ko8<jA~3{bIpg;Tf@2ia
zyrZ(Kowc@3Di<Xs)zT_{nb>gHnw1!ygfP7!A`zLSaY`v`4Z5>|_}ktST>J0RN}xjY
zji%B0%m0729-+jbl$ZaAQ3(lXc@`984ax%e(X|DifyhHtG{u&Ogb5re4rsbv=1L0w
zpq(W*^=W>0C&$c)7TM2`#e=>#9cy{vZC{XJX?+T0;pAmt;NYO7<l&*EotqsH#@BEo
z+R<tfDSa*n1;^LB=_yExg7!agEb;f%Y{?07=-kHauoY0He#ng=Jf5+amQ^UKhcP>L
ze_=PSlI$Yd&@ZaZbJ=(OP+c?=O83${n}hV94UcH#s!=>n`d&MZW(Sj5s9U(klAhXm
zKciHUSt)Z$gOVx(63Sss2m4@Rm^o`4Zjd6w!zJ+{{=gx<$1DFF1ir8r)L<El9Vmd@
zic*E0jQ0Fyzr>QBj;85dAs1FSA9&TKr#L`X3x`MWzLz1e8qJ?^e&dQJ&cu+Ih@qtS
z#Gm8^NJi5YcQo|BkWSHQun7lt;C-(7HU`+(X`?t%vy-7PNN;*G<q(3ie)*c96nI&J
zFj=1M5rujAWCd^=02ka>{Vw0!t*NWf5>P;8EZsh_fI)j=wsL=B5TWn^uMi>zx-w+@
zBl$O|ObPHU``eoz^*%M@UxR~2EK*UeOsy}^wlVPt@NrL0AxJnegd0?Q6`}fLUZOwa
zlpt8>$T_=*<&-5ubL#L2f}=9K$qx|gJ$_K#AVft*%l1Qj18C!F@QS|D4z<$;*V!Y=
z43(b-v^9fL>q<c-UY=iebW>3<DCC>7=j)@dt+3ti!0Ln+&fniqx8gbN!qV013Vo<O
zxTy<5Ff)dBQNBFXB*(`Ga<VKZ?)CE36L?S+rP3Ign^+(2og8DL5lxh-2X)iSPWGQj
zp|N}{$HB&t0Fd$BGDco&G>C)VuW?&@fs9Hx*oSO>CvNtX*2<nBjYIbqbD)o?cy4h)
zkY6xFc^Ax0lbbUM_YBO|YegaQ$sdmw7?|MTU`a{Ifq?@xn%mnu<iI@zSrqV3%;v~E
zZL5j+@B-CS*&9%!R$kf$rmQG7clRxG*%x|S>kgG0@NoC|+t$j>$FsuM((3vA1REU4
zrlko{lbvU48u6+obi(%`C5ZsVoAYA;(JvYM;SO&cN}0o(3UAhq57vyk;loc*1vR>{
ze&P)uQJGx88d*hJEIBBqH7Kn%Pmp)0OzUL;r|GDYD6}I=;;+<aKe^c%x{#xU$!s<<
zCC{}F06MmCFpe2r=>eI;MrPK=uq=7E+>zJ#icmHTrLS4g^F(|$NEmG{vNr^MKKy$0
zqQ!*}AK0+ik?D!}|7S-20cWeRCXV}skZu5N{E_Koel|kEBY41+#<||r$-bevmX(WT
zW>HCOnDAR`2<x~yh)jKSH2gicnL%e1NOoa-SV-K=$^`8I(;5rfs_T>d(vecNZbvda
zsK@m>Ud6`;c8*!HO}~j6;M-X3A8lVA?(T=7G$jN;@3V17%Bn0+^)OL6D%iM4%IFw}
z<`fsFG&jdo=j3z|Vxh&0>J$tp`fB%#;6LI4|E(WLTyWZGrm6jt(b#nM$&o;%fN=et
zEYw}tM&RE_jEzRWV<b$ElN$`TG<?~`(v!4R?X+BS1aegodv^|RHa;#PSI<D=NSh$a
zFw<6Ck1yrE)5pSk5{{dR{xcoLC!E-dQFhZQBuq468GU)zwbOCic|Cp~os4S1Crg@%
zvZ8%6e5NwnD<%@+Owj3rU^PH9m%s_>Ky>Ail}@ePC+bW1XPP03h$K5vL1-eR@hwq9
z##PFZ9ZPV)PW+5GxwQ6T_tISN%0f@38Vytu)>?fd=xi)5yMK}gfq-jDCg#eqmQy7G
zeH$NH@WI04NVA4xT1oQI@s3Q867h35i11{BWJxCx+o<>ptG~925-MPYRU|(yrzoos
z854CAf*~9kb56}jD<E~|)@c16p1-_Y$|B<T;nOL&ab#M<HV;0!#u6}ar4T>MUVGc^
zU?X%u_z?p0kZi(l2;?Wdb&Xhrm%h+iYQW<So@Qt*x4H4A9@G5l!7xRPu7e?FF<@te
z5D-o1<?orzvPF7kI|8<S^$<t;{|<m+8k-QtH2$`9-cBV4;e(So@vD6o+3XQw^!#Y&
zslRmq^D!Mwmt9z*V*%uySK~UcEoE<tse*!_{A@Bp2;T~Bu43v25YRTC%iqmHzCs<W
z`cj}#Hcp7p>O>L=TH#S@@&{Y7K43$dP!Ju@YvAogNcTqnDFce4J8SXFrhX_;N-6Ny
z5^yzf&MObzkQTI1%ak8!b3X^V^QD{9h2GR66AkxB%{iq7%B##$k#PX3`?Vw3?$;#3
zNg<4Rm5NTLJ!fmj%50!IHq7%cx6rDvjk$O6RRu_L%g0$oKLPvC397BQEh0gnRYYZI
z4pzlsTle~s56~kJm2-nYJzR<tgGKQV3ZAz5wOp%BK7<{xPRCWUJPRRJVM!V>>D$5f
zJ>Q?HG?d0_;9I#rL3XEK9>}H9(#g9oEhoKYj(=<(2N9XmK3$o06zRG4BX-IUY4V3U
z<~zOt`0(-aE43VxmU^R$o<~VQ)!Y-fm4DsJLiV;f8(=0UuP7agg8Jat$T<^>De>uO
zBh6sBxpW47nv6R+o)J%gdvUqDkK$fbnAX|K|IC7?%NY6@qyMTO22MJ-`L<d^hl=B6
zOvn>?ZHqS=N_+(mr-0>`!c~aI6?m`svt0_?THblAB|biuD$5R1BbSSLYH@ZJ+MYCX
zUim{@1|T>o`zC`>!@y8pQWo}Z;tTEX0H81jLCzzX98J;C4vmh0c<bCRewvcU{%3f$
zvd%0M;9GY_+c=rr0(1;tc#-C#qLU@HMrhdf@OsFrEY7gdGE|upUgT|2)v0~gHFEH;
zP))?<1t~B#o#kIvVWA%sh3@YSpR|DjGN>^lU@)45CvphGca~JXH=P^)$r_CMFZQ*4
zH2%QIJ@d7s>f+eM3dCKkN)p-`L2jC%;W*s>kaVbT)tZ5+N;o#H!S|w~D(N>9Up(*~
zv|W)aBH5!M%w+^s4rfR@v(2miF@8H<XLc0~!_P2zVVU*0VIfXB4nfwX!Ex{*-I2N-
z0CPHKfUYdp8l{yk%zk?Ni{FnSz7@?hLc$}kEZI=pc)aw{djST?F7G>$Ec=s}ww2~o
z2eG09nutL?wk>htK$dP3H4zD2?aiUNnIE9NMQVyS)Tt)qA^w@j>E*GJt$BWR*e8WY
zJ)cp2L^SVryeXpwl>;UufIPW~l?5gU(D_S`WlZ@mFt3F#`|n_c?|XTCdQp{MSP}>k
z4AM8tm|B5+hDDKEfBmc=;>!lezT)75Vxf(#^UI4Us3}nhBA6flku1dydj4%|Y4tw{
zr}eNcsfBX<js547_A5*jFPUKIjq%%R*<h*#n+-jo3AGfgkEdch$Xz}=!^YYSG`S97
z$NkxQQmO#Zk;H2f-HcS2f9DcdY-g-a&(AK7_h8T!!S_0$vYq2cnWMW!5T-73qZf)x
z;M-42_hS0J_C@`(7ZHW^xJaK;a&@`CDr$Yi`D;ozG{EF%0sgoZ7z&0ewbqu0|97|6
zASJ!YkJidbYjtTkU?JL%^LKR^0FP)_ACpJn&HeX>5yW+r-SdS==_mRUEuk91a~H}x
zO=ujN1}3yNJPLl~{UG>P`PK-6lXKxy`q`lbC_bukGA=gSZ&Q6ez1>S|1K>R*c#^?}
zJD3ANzt_JrSD)OqjOCVBPi_s-P07?0966(B-@fH}kU{zSS~3JTOV70xs3T<y1rd&v
z7JHY4z71i<i}ENXPhkkP(*K}^cc3mQVP>swsAXQD#*ptO8Ql4aqzk*VDsrnE(391z
zP?7S<d1GviyLm+}$?Szca5FizwlLhs#Ldb`**mzQAunTaMs_e1gQ>)F8#<QW>~Xz@
zpYSU{;9Q>Je2NW%z_ceDa2YsZ?bV^zc9NV9SN^Uw4yv1JsFwOBG{h7FL-sJIAX2-f
z-1Pf+evCu^$qbDJBmEi=t+^vE7_Ahr{3Pzjdx{-?4{j3Zp1au(&!~k71!eGkbTnYU
zGLhphQqyz(N4$YOtp&Cxl+Q-j7w%0qnl<yKuw7CU^#?Atc5p~qpwP&aqVMR3GFCDz
zkAYATolsm><I?z-Dr?+KwHlAbgCs<=qP{Z!j)-D?J8`QSW4jptA{il}h2K8X9GKLf
zcy}u<oXRN}2Ct~(sFW1G;WFGQ1t7y6(s_HY@~^(Pbz=G-9)1YITQE^XV*`BfC23XQ
zHVeTgGng%s?`CL)fE9cp{O<u|5@sZU#vOwa6X6@lkB^R#4|V_PSBjcnZBip%FScYt
z{wNOX0WmxgAPY*sLlxBXA2=bits!^_a$+Rj=Rzf2?DTl$1{ZkVYZeibrl+FE8>uO_
zf$9cjw2#$mmR+2<dFG~DxM*m6QMvjr1mXM;^ciVN{BF=x06Q(tP_0cg=fOar9v!QQ
zjz}NczHOd;-bQL2R~*LDE^cb~pv3qG{z+0%I7It~y9S{~j>o?DFs`E1rq)RY_S|P)
zJ(rB2A<aDdFQo{wXFpVK5wwZ`Dg`W@bTZ*ff$kB+Ca2GpoPe_lLx^`8e?BD~qJIh_
zBw~(yxR>3bz~8y2P+?nKZ8pd==`#2zJuyqr>l7H7>E4nShVBMdq^Nh}BNAD}P0lwQ
zv{B0Kj+yKq(LYVzJwEV*nEgdF{ZD%X1Gon2v1SXjQbKAj7}*;6-Sl{r_qFIR<>I1#
zD9;`mU&_)^3AS_Ex<Y=6xBhekBB}^Shq`3OP84hez#;kkTI&98(?9!Vl@?#6<j<fe
zQ-dO`gru4R$%wWJc)8{8-xe0;r2z_LB?B|ojeh#o)gn%V&vuSJ6aoRn!V%r0G=5(i
zK-fAr9VhYMv0X968EjAoEN)LwXwhCq#%OxcDpUmyIuO6<3^_ChxWvVcXDF1ESnTcb
zRcP>10gV+uLSplk0A2ouFim#NtE^cT*9X{mA1_fO5MNYOPMn$dRiTM-LnID9q26qw
zK%;p<veiu+cKy&e&wu8ik0YbuiNYZ&(&Uf=#S(&lcW_&FzuXu9%ASGar~r2B@{8ic
zi||+H=o#X-0HYy$J~`yV(qKs{(}QF+P8XL4-~k0CT0vprVc`hvcX|0x!fT{;|5mjt
z%|F=$BMtL(Yl=~kXeCs&K#|gVGhsK--K(fLA(@H|{2LskrZ8Z2G8V@xs>nDpr;!=W
zJJCj!;Z-Ka#z#LR6qS`38CI?&Tqz!sLv3y0-wod7FdKT*9JRfpp@bmRkRQzwaCNZ(
zTuil$tc-s<iLNBB@gU+<>#*CJx@dX1fnmYAk)|{<$P&hd1~{W|>5yA-!>3YAl$~}$
z;2{vSCKAZCThcy#GTzL+dNSJ&??5Si-<IC_!#>TyTw7zG`~Fv~x3{;C{401+jW8)H
zO~^falF7fIC+@+!UsrEteq4w(6(>)%0J-KuQ=~^+Ab-s)vA~<ySEZR`4gp&Qk)HoF
zhuAn1IRXM~JXBaXI%7!AWS%%Ug+YCdhGIBcf1HYwPiF8zJ5PexN*L!lIUZ9ya(D5J
zUStC;u`koo__Jn{`t26)D@n4l2e@M+!xQ67^m+(8mxK%S4Bv*b(sBOF-7W(PeG%jb
zpph?69UX4r5w3dWgD=2u?(#+@MYL36Gu{y&KUkO{kJienz0e^)yStX9^Hh1Ztwa8G
z1x?M;7MGOe$)EiG4?C0k1abx6Rg}&aPm&>gvK5<93<kq`VE%{3Yu|lgar>xe_VrDs
z`khe6>JnZdm>rVW4h4D*fITd%YK<<#-2wTGW-Hpw)*?4GDurExKmM*P1_7D~9^R2$
zcvOocWhbf3y4aR)Wx>r%=T~y`?&Ojzrj2ASg{B7Q1ZYX|0&Q;dMI)^ps(ej0-?Zx?
zjIlZWD@1kgQJ8z^)A`s0Ru+4{OJt_(dwyq7wRea*#FNUVN<5Fd1{8%3b^!>adl(qI
zU{(>o&&N3GzXe_H6ce^V;{Ao@hKy2Nq_J)ali6`xo)hiR{~}Ttv^$Odji{EZRfC(6
za65IfC)JSVYziUL!R?Us3C7>!<4OOfIb^x%j4NeB05Nr(Ye;lN38OhU&o%Y=$wA>-
z+A>x%;xiYpu1X;Ny@HotFly1uvxn(_ft^)0B-wgP2XK>ezdP3a?wUtDENCN?s5gfi
z)<FFK3u|u28@JuKfJRRj>&+HV+jBoLDZ&_1K9%NThtmOPx<gY5Pi=FVn<+|^kL{3d
z4}Og=7$SR0nQsZ*T#Ex5giSXCD6d8aN06vOU&?EF66;C;(lN5n`_zQ49G|nE7(<BZ
z;6=`EWc04sqy9T{w!<WP%}$H+7y12Fmy0!3)y6iD$?T4;Yb9<+i*Mfwb8%V2LiShg
z{`t+a&zSb<eZTumMAr?$EYG*L+Q&|d>++J7cQjipyvQC^uhMIqZZ)&^@lOe($AvAU
z8xTeXb6_6>Y|Q$r4dc|L2giC4xno4W<ZZ`&R4oW6?HKmWF`h~D$I4`;R>$U@OBNk6
zt=CmQi(Zaf3gFhN8qQuXBot3NWp*vuw>Cci_Fmrrf&Ys_cuu5m??ilGwnI<~vG*k%
z6wuV6rjfOGtJw-nf7(Q%VB&esh`f6GJ3}6wMn?qzgoNI-mM8cHV~ev)+`P1o1DW|I
zmR#y8YUh;#ZciGrqt}>~m0SI5ZU&DFGljQirj&thO9Um~5o)6HMp~IZQ{KPmR3Od$
z5n7QoGNK7;kJH{ugeJ65(D8ZM*)LzT-RSt-UnG>rMvPx;fQDsV9v9YnPd86xPJF!@
zI$C-H0{<k=eJAz&r3(MS;vK)Q$T+%v_GW~TF3D+wt{G(M%B9!-x!R`Al+AzC;T*;9
zz-7DM89)+Lb!(>B2s)jCb_jOJvF?t>W1>L5j215WQb_&qD*cKK;}*#`d)K|Rdw62|
zbu!WZS6JkqZ;>oSdFtu0g5C<#;dSBbuxNYM!1*ELgHD?dvA4tBv?r5m!<>3bwXO~`
z$CssXQEdgdPzQ41!L89x6LmPkF_C=|vHf!IT7%#V7jS9!(IygZ@ArdBz7E&Rb>OO|
zz(&_6vpjmpt_Uah5m$4J0mdycPSg+$ANwfGoF^Yb&N3TwA?HU@3^n^-OVmZyhLc)%
zU@?2ZvDwGx;fO?(nAp)SOK8)5+i3A0V%8xaC{;^Hx6RN;GB)a^-v@)J$h%)e5zb1;
zp5fWc%WH>a=Pu2R_*tmW8|=0UPl3afc@rG4+xiQK03XrKkF<WR=E9i$@kQyHHf`37
zNu$nm_CWwPy0IZ!aCO=FnZTdO`}S%c(C>0<wFj_8eOx3PNK@M$f3@G686q3zms2rC
z%h0hFeu-g~s4^MgD6&DvcLW)`=6srV*Pa!gP6(Trve1t!=me9ci}7g@Bpbne7ePf1
z5<IwPI2a(LU*YW=Qus}VER~#7T23ea`!ij6NpGf}nAj#r<2x%vPv!2hJ;DHZ60R@_
z{dL@B6@3@qHO$my4|shWR^FrK)~R*#NG$hr5-NX_dkxi3O^0XAc+^MXw@zeBY&k7m
z<iCyoM7sJ*@Fx<>Zdj%W12afMcp~U!t*oSfo{h8F-s%1ZF+5aSZZuu6-k@hk)nuRV
zk%WK2;{=K2_)OvuIqzd@yFeQdAg+|MbhXxOW2WZmpO+Jh*TZ`c|A3k}v6m%;lC5Fl
zo;ps$A<+#TJz9AgFrCz|55VgVcz?Z^amUX)Iv*L2&+iuCPTXb@L)S?KzjC`=U;j7U
z(}S*sHo?dk-2{Tvy{r&?GY%OU!@xP&a?P?I6k_D@wwP!yJpE=7-QWV~T^{q#E_mGu
zBU5m3yjq+-kn*ohIYD^Iwq>|fPAJ~gs~Cn=vh*~>9P!SEma=uBxDh<Z)bqhS3PEM-
z-~}AXi<9{Lrpo=QtCU6qd5TYvk*=+^y27`UXR1O}mRh{OKS`(6?lt~$Gjh_}Ctqnc
z;p}p+)8SK&8}M;8y*mgfdQ|Iy)TBCX<zDXKBUaW<#?V)(JLU8Wb7aCAaFP`LI6m~-
za|!fuc?#v}+|d5`FApO!iF%?wEuGxdwB0?zda>s5n9a;KJ|g*f5|F6IQD9DUS-z-Z
zMeT;u-BCBru66YwD`rzj)^qn1b43X{ZcZ#K4;`PGnuy`_%?!Yh#1WFN=nM*w<JD2u
z)$AmZ8NLa(UUgMjB;j?*fA@O4Z}KL@F%9N9t|bahJDL^z`9lRl236V2hvVkXAX0Vu
z-TkLxqfc;mBcz8erUa$2?-GQYw#M=j4{u*|ioSu}Z|>PLtsm{i{|I6O${|P(c8hr#
z^-fwG-Ve?<UQq#3@FT@+LGOK?Ajtm}6S#buDjt>|kUOiI!1)VHiAl7QKVuh3rdrX`
zqMX}^fSZyj)SnE030Cp;k1T4j{65nlN<*Putrw6_W9yK}^#o^Fc2LrpKOf184ptrQ
zi+XeGRND+Oswr0?KQu-+mJ^pm01FA`)Y<oI+Xobvn8iZ}Osu)z;^HuA)miPIG@f!=
zSFUwjZcP_042U(hHQRpEt^Iu+UpT77(q;<Mqo%(c%;Hu(xDahL(Q|DK=95M!6?y*w
z1u#~?(RV6#z;%JS9gHcno#l<4zTcYg|59L!cS4aU+)LzeLK?3)E4`np2|WTaf-^;8
zCjlnCIxpwr;BjrGaV=%nE4V!;V%*khy)OIHiFHp@62q|^0pMfjU%|%d`k^AJE%#sR
z{k1VTS2hj}+ru|rF1oxJ2rKeh1Qe%>5H<w1BMu%tagoIt0SZHl9vxl-ibGHu@|Tb@
z7rnECh-Zd1W&=vghXB0$uUEkhQ23bA9Dwp+ci@>j-RP)V^4NG*JO(C4QfzV(TU(7W
zkAG`}RbLT<cC(t5n3;#v{=_$@+QkbYx1*ttlk*JOl$5am>H-6s?Kfx%Ixe*Z5<cha
zfdbj_Hrd0X5Z<1l><UX3krgQ#rT*8YDdTj|dfSEIF6ogu-*GzI`Al^6ROb7L43Iv#
z(UzVPXbM*b3p|2EsAyZ_KMjTveA|r%t4=@?<dA3*OTi777kZz@&A(B4=SLkMd$a>@
z|3*`UKHu(ovQF2pZdU8_+$1xr8*wP{J~<edcLy=MxW6+{hsk54(FE3-&D?-v?Ry~Q
z{Eg2}(l)U2fQi0y;?L+gG3?KF0`|~wYNOH10(fOWM^+3|h-$wR3#qS1@p(f`CbYGk
z&kUR;mg}JFi;9f*_n0(VJ;Wrip&i7OcWZ-#3#3p;1yz8n?vK`$6>ko{ZJsnPHctl+
z$%DkwWspjMUZbVuWoWpHkEtqat;=LOlG_$1DFa?BWPGC+uIWh}N@GL=0N~w=v^Nja
zKjk)~Ek==h8VQJ8YYv>R5s+e_bN*$fIU+>hFzB|J67%f+DCe584oB2oPpK9Kcg9`6
zvg;D^jQ;ZZJYTL}4O&MWNKM*w4_8l{D~A-}9i{>}X&y<r%|Au9M0m_`6)*--l)hp@
z(>jY~(j(=D<GX;vgHQ$l_2w0vN@@`36Abh8mv2=wjiAQZM8pQc@jmtZw~jEOR-2&&
zLi<08vQvz#=DL%oMnr1D2W$N<FOiM4_@vY3^Nr0Zm$HuCw|^kN@y1_|mbUWSwdDt8
z>U08>^G)>*$5%-!sf=tiP-ZnIB4k@)X8sKEz4iLz#zQurRtsP;vG+sD)I9)xMv%4f
z(V>BTDi#-WizqBb__LS4^U^KaX72MiH28u%%Jk5WwzdrCT025ED;y&gT721=(7BQ;
zM_B5OpLa*SYTN&r<Uyx6nnxJLx63DB=X|vpl$qe}_V{p5my%xopLgRg0s1e-CUpJm
zwpiaU^cSQ<!@zgUo39C3_s<y6xz8r22QY7{xMtMUu%6nMAo;EGFC&o&&P0yc09Z8C
zo}&qPHg8h0+~fEk8a`($Op}*RIW{GwGz^-J_EO{9yJl@lA2&zGw>g{b)aU7KC3crq
zLK}7lwkGb&;XD4BV}DA>m~2V)KYhEmVrv?fM@Jr;9{~QB%Yutt4#M!_L3g%=L%Asq
zXf<UB%*g6Dz29K3GdjaPpBQVJ>yy`Sz413?E1+$;UeENf^34G@gQjL{snKC}>kE0*
zo_zG>-76|--|)xtyk+F&<t}l*Czh>#hu6c?cE^VHIqYc4Sm-@#06hLi(94LT<Z8LH
zOamDIBXH?gTqVy2N{z%r6hqq-r<3(iOw&wZnNv%`#N9C=<#D5rHL6_qH!4-NG9;lq
zjV&E5*Rp+oxP31_j9$EC)b9X~ehp*OEin=q>hgNOSahwPh7}vq?{c~@R(|i))n2UD
zs;ePm@^tu`J=Nzc=zNfWuPQaE<9V}fs6Knq2;8#vhU&0l9sj|CUgd2y9P^;UD?<#g
zmdb|q&WSrC?bw2imk?K|g3q6kNry3QVifF|me%BfmKa@3w3e({Tp+dPjW!6m7(-U6
z%6mKK$+Y$*hJ3q7@YVPJ99>-$5&~Y1dOG-sLz);SWv?2)#!s9TbS@;N=j#rA%OhtB
z0;zTys9>ZK;h~Ws9r5axSEjf4atFKR6jjxrh&ql*+Cl<?I$J@Yok2K8bZBDP=G>=>
z#3KV+YP!47)Yrl>U47T0$7Xd@TOEGq3Y){Ca9s{q!sr~qMeqHAb5CD>pZpyDcHr?L
z3Gtkv#G-UbIL4&dR1xy=x;?EPr*gFx`0NkTo)qN&JN2W*%QT;<eK2f=msWZzA`>4y
z?X(nq(th9?cW1OKchi>r4gO*e6%{{P&ntFJY;bj^XPt>1XP?axhfy~myixb4fiy^=
zfB6=-ty-t}=E00<O9plgzW?$y;5|iuqvPWln)l}MGMlgN*M7wwn6?bF+uc+(U`4$9
zkXaL0QJ2pTqZj&5{};I8X%c=$x~7Ka>Iz#Ho;qP27$1>x(Ou<w0A{Mh_a#qzl+aNl
zECdE`jGN?h?iKgX=EWG>$4mhq<IU)@^!JclC!S5`=NATH5`ihEY$5NrtDcV2l!+!z
zqLPiy+6t%fk$qWQY$gp26&-&Ipp{Z~@v_8{-|_h(0`+l{v`T}WVFbFtX1|cE%cS9%
z<{u?7W<?@2QvrD339Bi6zq*&0H;*g*!#L?q^ygg7G?76kV{zn#o0MYg>gWU=jaI8^
z--50u2%rIlT*7w9%M`CoLU|FB199<`Tx|b|=SlDg`ZL_`Z=Nxo_V@?^m#XXpoHH=^
zCT5|-R6~6pG7^3VBHV{ARexk|rSQ>vd+JcbZND48mcquYfQ0&p@{G0+>eGD$(ZmJ4
zxa0cz9e*8$hW$Vu39S2buL8E)6s++q9tLt+Dta;&kjmYgx|Q_(;ox}P5?6tXw5(LX
z?g*|*pBJ>}5i|REF3$n<JKa9cjuQq^L4Z||>d5CZTAq;0W~s*K{iUvsH68UwW7oCH
zVpNC_@ORPPZuaetj1XI+x$l^kma6uNqiQ#aMJXr+@}V0Ry$q4jwDW+D;EM0)6Ux<%
zG?HkLJkUI%A57qv<cGfmoeY<?7%tJD%h)jy$AcAlh*!)zk5Awk+wYwi=E+;=8WEf%
znqTz8=$Gp8S|qcXH*!1C&;zb7^#IlgLR|wb|Ldie?^tCnjCP#HAB68B8=W5&af^SY
z%MP-ql@F>JCgkBA*%o8>66R{`P;>Y#uR*A_WDt(%P+@dTzqR&dY57zU@O*vism#7D
zPA@(7aU8u)FawyJ^t-%2hh<f&1c8sYWwyKD!nO{Z6XW~Z&DK&T?-v4|j|+2W<&~A?
zd<dF7>&4YVYXZvJoYd2!-+#);$Y^OvN~$Hx*u&QIx{g(Mvx9R!8v+5JmYrX5mLC+&
zlbsw%%UR6vG`PXe(ppN~G(#|3`=V$9LL{volPtN=P5{?Sp>9=P!3(T2s4c~A>aK@V
zPbccmK)&mrx7l^i%3s%`oK;NzLA4B`e992u6@ZuJ(}bvQza~05F?Y5jd?VmZMUo7R
z2E|%<huW_VecH*K@CKVqFqJR28>?B=B;@(Wws(39??mp@m1V6qS?M%;Jn!tx?|VjP
zw$8oobOE-dFe<XLj>^j9i5CoT&!b!(q+h3`%XjwsGS-Qy4Tx%^(xthY8;FE_cEl|$
zPx0@h{>ztq5W6hxD0eK#WcMqlqJ@N#2-g2Mq#>rBhRhqLVyk5EwSN)(Gbbkc-7Iw;
zvLtUpqf=TuM2P>3x1`8$8S+$+1XE*z8Sk}p2p~vbwemeF+K~G3naO(>a|w9A7g)Pz
z$`<fOf|gfuiA+fols9<e-P)GH!b(z8`6Rk>n$5<DYH+c$AG97E!V8v|q%_;f()pp~
zc+$O47;9A)QI_5!WjwB}>;5$5w~~IMo3uLOO8T20S^W;!oJKC0wWbE897GfGx;5+@
z15%4g+K(Fg{R#hhQ=`wTzg1UMRSE@TbjzkNIu{%rAGf%iUi1|+ZMJ{f6<a^vlzveQ
z&ukA=D=w_gqY(%rrD75C?nTXYPX#oayo7U}3WP*e?TP$<mX+kgM@nZT>o*v^HwoM$
zk5emTowqwdXE~P4m5LINIH~+eP=93tkXpMwUSx$te0GTiZzi*KY%14vSjjgQ<G=lu
ze1qP)zTv%U1b^~rh`@&-w~_F@42=jug;N?-(xPfF`0OjU!sPXKS6w}?hiy)ji?Ult
znW?*1%xr1!`YcVbrjF*~BDUn7ad`2XtbCo^czYBu7uG#K;rBcdpPd$awYLF)&Q}j6
zUWlU&&-a6MCZ-Ee8CP~}aKBKt(l=hr$F%7Yx`U1~<on+kISpB*{)SupN+27F7t-9>
zfW+YIM+3D_sh3K?>jh=06j+yumE!kA*Ii%p*|%BbPPi+v2TCx8{Ka;QkdkauDY&$T
zBA}tuAefh9J(FbEu>id5-T|<DFY}MKo9z&L)!A$MofJ&8-3uz}5~diWuc4D4#sE}%
zMO{%kD@<n$G8ks!6f(63b#6Jhnx?LRcR;VOZ>Cb-_M%<b;CR}`!^%DMu%PJ;4CD6E
z$>yy8#BWzF-YQ)hN6@`r-!5K$zK*W`*GFo1eI3|GG+RxNAu3r0RF#3V{xn2&8xh^x
zoB52#u8+t00_}A(t&>L;l|`;**LAn68!olCl>QKHhE*J@7E8nrGj6^@muqvx82GbP
zUL`pKx^6G%r7qY<$1bT_A<3)Iay*`wf5r#6)hiA4C<h2r`67icr_FPgZ{(4{EB8xJ
z_Py-obR6tnb&epQQETO5!JIhhkPksb-bLi5la3^{3A1ONZnr3_B-Lq4jB4fo#fc9B
zygJX;0@)$&LnIG<I`M$H^(2okp~LC$n(!R9i9Q|Y+%Vr@k07qm7YucZ{Hl8G@9oww
zA{4**Ea>+hdQ1k@Q)?5Pz=vVPugj|2Y<S<D&A*~n=iC8$=i*X6lRdwzK^94$yLc}V
zUoK2MH?B(f#LlRo1k^K&p|G2nUvNy!76Wxr%)j^gcAvZKcRneGp1kS#RM?vYm5xD$
z#BzK6!|&6UQP=MgFak4U#~<f0J24ad!zZQwk4leWE8`Kj@esAZB85+Moc_ekVEw+*
z(Mx~@3OJyM?Ig!}m(ytlA@`g^_O0L#{49*z-6JZJ>F{Vw3d>DsPK$$?@^}#?TS7oq
zBo^&f#&rgMjyu$wH^kd3SG69$_uJNn^;oHeHLu4VTzDT%x?<L4OPC%k=>~rrP*2$E
zI;aJTJM}o6M*NGmeR_iGrnSOKbT>OUokluZQ3$}z9t?Ov2m@&p-%zf-eM#A|5-^^i
zOd1&kdX*g`WJX=j`Mikc$y6d>I&wEYZ``*&(OiBm&zgAIn*F+e$v?jK;iTd{DjVzu
zXcq~aYnHg(YiGD9P01OC394(Cpo}}b`J2gXMzogNTG~9_+&tXf3wTNH!j*_88H>z`
zV*z%@<pmCj7ISpHrWct38!#_t$3;rs7ckUGhBE1kjZPo@uA>Y>6w;D)<;H^i7X#&2
zkgHPP+js19rO>fBeRV?bsAc-~h{wkC<HoY9o2@(<-3KcHw_LN2D!%UP(8YJ@sc-VV
z07Lobt?=dxsO2?`rM5mtNn`%<HjmwQ3Pd+F=>K&fv~Z@G-(>mwi28PDAT`{rU8V{g
zF5Ex#C0QaC4?Jn$|MKUD`9O>tgs~DnhsG3=M<xRqVux#r=;a(U&{+pPZ>RnP3y{we
z419@47C(a6DP~AbO2WH<^d`o99IAjmEz4l^L3p#B^7R*ldV%Ar{&K{(l8&+sbiVC4
zuRai4ULJd2gR_3FF1GTq4BtI;NKN__;dOT1L%xAddYwUCg1@E1pw2j1I0CzKeog;S
z70t--_Z<%3>gwGtv`26;$*5ibJEh^N4JMZnQ2Ym(v-#zj$iG8no$@a4K@US<oifG_
zczqZr^oX}FY;y+wjrd1?SEL7!zsYfuccQt`2l<o!c#0NA*3y(#63ISxW3GyN>c-p>
zaoGT+S@(^Dx_S#MGdlW^FxxZ{MUy&4PNCI=ZpdY+Mq8jB^-+ERG=E}2wUPkBqJW1-
z^+TpL&tK2cbO}e_BBfcUi-B3x*k_@m4tHqNM&|Z_FEEmwLR++u8FzqJetK0kf@>;?
zT?fBT7V%jH!<ZbBGkc=QS@vZ)pR0c4t76|k<RNZcqI#cz+;4RbFaV(rCv$L@E<$=v
zFz6L?>6OzF>~XPLPXKrWvn9y1T?^PKH8;F4g@XVcJ;6OqvVaL8ydeu4(oq)wDT5y>
zFD;{~&)iJSV~u5DmIYMc_Il;4b%f=RpsJC~|Mlk4RO6s2gOzGfFC{{(M5Ia0VL`ps
zPDG%Ay_rD0{es}9C<YTa9(~;CwA&FjF&`eg8;bPD#)ehU^LYb3zv7sS$N1LA9`dz`
zVBTt>tw9A9b|(ax5!$qbEF$=P9`Uq|acUY?dhTvq_RF&|WWXRyCid0R^Ikuk?`0(<
zW*0pk3|<IS#%1o*>4qrwIR~|&RNKWA+46o$bpvldi)%uvPmzsp6-hTboEu+72am=0
z7O@X0pxeOYwfw<83n>6SIB<H_qCb*2oAdMYn&Z{r74uCjZeh6KOw}kmeVy5iC<~@F
zg=4?dS>C-E0Jp#`eAY+t?i4j*K47~S8sdHb6zdAk-S>l-QCeTv^21^EqjJ<#5ycUH
zG_~2qRe3Q0l>$}RHbBaWdH=PzJHRnc%J`t2Lt|oYwFk@WYbWUJG0l7!9;BS~^U+ga
zCx_f=+Y+>+2=S%iB{+BhIl(}fWWFg6_;oc`Qhp!cA^v#!PPxw;#|6Q|L18N89*?qr
z9d{y|S5Ljx!pFtSHzBSmJo7Ia>gy2G=0$gMME5ha7^*x2@6;w+9IhwFtfL&g=MZ>z
zpWQ<5#yV-+NH>j&(^nxbzA0^dZFBk<*mY&Z$=^_KRl$x~Sx_Eo#4)iUlM%MeH#D2v
zs?FxWb|M_Ez~RtY{EgV*g}1>v0%*(RdGj>g_v#e3iF$RGPoU-l-x4vph&7oVkG5kD
z+aHo!@~2|0rQWgFL|!2}1{Ar<x~PhENc!k}`&13)v(>GnBU)0%UhvRmk<OPF?unqf
z*7lggZvFz<DYSI`8Q=T<Tp!Xg^(IVdcV7gUII{loI!6W_4iB%K;*!VAYX5c!2>;>|
zH8njyKdU?#U2ybZnyz24-R&4j-M#CVfa?Mysy-%V0r@wG*@+#tIyM!ApDo0N`S?BP
zuF(s&_X_s;dSqR#o04;6PGP=<F4Wd=aF>NTX4?ZtpvwGsi4i~1_ThUNJJp0Lozn(1
zd5oUqrx}o1LC4enmG7)D7U`i!6G0wqFzB4##B)4J*u61BiV4jus;!hl)46Ml$?mv*
zqntR}0A`X;ai4^IL36m&{b}|_mv>|Nu8M45&xkW6Tb>t$J%rMhfDg~rQ$J!@Xt-=1
zW?6-(i{{*M&j<5bkpHd%6z^X*w`U-ba!=K{dueodp;tp4vK;SbHqhmym9S@mn!fOl
z`EU4`_nRAU2HZBZgQGG1qjsm~l}bIQ>)IQ~lcUd<(e>z;O3K-MoqOG_{zca{6FnG;
zm=5n<3sgIt3jy&*s!ni{wXTTl<hmB=4XQdKqm$$f``Pe?OIrO#W-btTum^y%q5=hr
zA2QHp3z;3<(H_$_bj?4awG#*<n+?E!FEN;zo@=2SdE{{2pi*xB@VoS!xE_bqe9c&-
z(PO#3ofPU4zutlVQoPpg(|Ed8x{G}O>SD{pkd}qNdubJ1Df~1TLwvp8z3a+f5*4r#
zUQt0rqfSEu7J@kMSP+72Bn>!CVayxs{XvmbNIuBbF3UqmBNNJkkH_$hQWu9wFV>3#
z<)mwMonDn@XJ5#=tw+g`+irA@wl)lplIR|HWDS-r;3$NNCV`#F==}}3_2G{a{`V0@
z2O2@(#F%WjTeN?@j(c0!iczg}Iq7J-6|_ay*95F?&L}e;lB+X?1OVFp@t>pp3*$np
z<|AX!j%J6(zXCLc%p8fIrGX@`NB6B#t5C=MilVf7;BHTo^7PxOcAJMjs3q17tO11|
z852w3CaE$)zjlr>??Qz<p0<3LRP2t(Ax|3HT&*;_THVytywE(7lUP>cA?G*sk@z_+
z$X|7imLg=ofB!Bc1IQ?)ixh+?Yv07qq3=_QknN7pm75>q|K$D0yMi>D{BcIk9rQ=l
zPlY&vrTmYRr|xab>0;HLldCTGCOBOTZd<D#Fy%>_^~%*%qqo<A8*^E$p_)pZ5*42x
zV7-pI@yh|f?7_oU@1Nz&AYsNe{i`Uh%>xG)Q5iXNNUQd-2GANpA)gcTt>*^&wrJp)
z56@4k(bK?bWNx{|ZcWK2MDK0j&+t=+9jPElE-d{YvC?F~AH=t3&%9`nDmufudUrNB
zj?6;)0XK!JdA;?~vBB4DyWVcU|MO6j9;*iRhkkv#v5$=;(aU$L^dI>*jw0Fj|AvXH
zcDGRuccpUkfya)Tnwn5JS;a7$)gL7Vn>azG&HnLatyF5%T&$cX=ovo#lBgrFZvT?(
z1%#X@c4or8k&7AEgLykndt2&@S0^)hR;TSRh|0fAV%>GD7ZY~>GEf^{v2K(nUL@)K
zRrO@U+FM=?0^R<}D=oY?ehnKC)4={UzT@t$nOk9g2@of<H-e=L?T;d|l6O=z6#j_b
z-@+zJ;PjS}v@o(#(O{+z{G2GcSp`=#2X3WBM;X6(Zqy<3JvL{yz4*!;pkH}!ZgidY
z_09#&aoROquAiIyJJ$wL*n%7#FPtoAx8FTotyguLJT!;jc-*&o!6n;A`s)d`DnW0;
z#ab;oAkZinG19)4$OlitWp^E4EQ<H1yuaUYTv>)-0j<_FBnw}o)$rS{a9IyEMX~iU
zdUAZd-F`6fE5QnH^|mF#`q(+OVpC&d6DSDN?{Nc^C3^kPXUm=h7kr8A+%~e9d?~o(
zbTCsnXlo!q@U)v{rqD#h*8tegKY%199dPeJA2RI2*^*9cyBj0#*itpwR|=b(JAa9C
zuJZ~8{bV7Yxp{H-{D$!Cmj5z4JY<Y?RNU-*QAewuRv{HR6(tQ{(p*&}KW6IAIiwWz
z+Mh!Q-ccPmF|djc(0!w09-Vn@az}4wBsAE#?JGK)ELMwVup@9MGNOLY@BL-gS7!zc
zW6{xfiZRZZV0<XzNdC;-k>Tz0(Zh2S$g;r0Az)7Z>!<Ah*Dr40Mwl2Z615b7&h=_O
zh1Q*wrL6LAi3iW5@${PuDYwhbTN2-tMK43h!Y%`IGH4wBm>qUpcf%0FK8+#W$m8s_
zI3Zl+Wz_9@-Me=z(ePw+%IqDX=A#2RD_3jGW&b^7sd`#`6G5S{l|iu5q@pVPQ&gy!
zApl&TAcLypk(3D;_=d%VZ9nlo1f;!RVG=ETW9NJJuFrg6iXAxApf(H?)dY=}dhOwl
z(dBGDMZvGiwMPj^2Txrav4S`{CvbO`>Wq&+?4MahJ0R;-qvy^yn(W7C@_7KVp&UFy
zbP)EJ{OJ2O53lKjm;E7%;$clZ1C8R0jU)XoPiprqeD0beQjU|;oQNXYTz(7Gzy0~1
zUWO@RCfVHA1eW5sxu)x=96gmL&(G;yH~#CL;rG?WHl=g-_T#1LKG?^_g)^QdIK}>i
zMbfKO6h&eiJ+bGN=lWtkMkheOEocL>d)iOI+md|E4m(3eol&#N=zgURIb(!wfgvL&
zB@ON1bnlpyAPdTfMMBnhuE~<g4im$>l3kqTad>EDK4-oKvezH(^6w=Wf58mDH1;1T
z1T_`V6`f9Y8a*dBYSV4$mjF_B<?PM{Ceh7Zhzsvbp;ez7amU?hF@S@wIgwF2RR~g*
zy2j#c+sHOdUPm0d<=SQ`)KK4)Bq(45&EOnzCI%wgB<2f&&Q~L_7WM<y-XoOqov0j|
zN(<l41scE%kvK@2zu@;ba@4NOJXsCX?Uk}$|G~hj9tFKVUY-j!2nBE_8aOzJ1T~sn
zA1BVWr=5LTPY?QX03qf8cdCn@6|*TZmm5dN;_7QX-m0sd)!Hq_hCS`j5|nuF5{`8V
zecnXl4K;&Xb8i@5(V$ko??*W9@XIGrAwf!DC$6-?`m>Vzudnn9Cwx8hLfsnnRMMFp
zq4x2(4i>nVr=HaByubQ4I3~@n_v^uEWBL;wvKjX3NWWt(3W$Dsj%K65^*=fiuGzJ%
zG#wn6k2p9mrZB3MlG4^Hz&~NKgCh152>2+9epfUm;tHlTrbKKc`S!3^XfVLyvKI>;
z+l#;A5k$$Ri*72Ux7<r@cyL23#n)E7aRzO%##9ZeOx&L#dathhyyug!wK)?(ofxOL
zxA|8lX)%%sfbZK>&K7o={xQ+x_BkEys$%Y=J7uADvH-cOsh}iC9;lNW5!A%rs!0r)
zkgNW)!;I-W6%n^a=WmaOazCU+K+^7K20KX=vqBy7FKbWlTLbomKTc<m%D3uPLUUdx
zI*(CpiasAS5x?xHx7a>6Cnrm3DNDUq6dM~GvjVyQ0R-x{I+}?&Ss27e7seouZV6)W
z7)JhIZ%&W7?^^H&wh%GO*guf^5Z{07@A9J3jE(gt&*a>}KRUC|CWnl(4@3!1&(Sg3
zE<cbTnr{K#?{9{e&Q6wn1Qh;0i!Q;=&)e_iW~iRFddIwr@O=5|RqJ5hIc=#v9)}~(
z34;&nfVD(O$L|aJnsjLyY1zrlcmV-d2f5sgq|Q31g`X(0;|xXmpYt52xW++z$I?-Q
zj3G7By9Q<;^rMk+ZgJ%|Db?&vewZ0nW;bFyo3^HU-0HQ~WgDdP|J4$F-pya4@XhDu
z`7^HNOi51-FYJy-2@6Ol>FWbWXd48;x>~3G`iVq?MNMI89LWL#7B09nl+$bUljPro
z-bm5x2w2tUuVj*l4h?eKzFf22BN2^hlEYseu7~j32KVSoxt*#`N1`@h!QKe(lDmR7
z7R{zkPnOV~Y*m#tH!Tgc*IHKQrxybA@$(CjH*SvN-0j{)SEtj@O-2!JZ;MQTBC3Hv
z*q}fOUmrCi)`AR}1Ou9k+N$o>DD9^0eGE@)cm*Q?is~X*!2u5gvC6O1w5;1q^te7E
z%*|QxrJgD-(=Q!^O3M!C6MbgYuO!ZqyyfJNY<CuR$gYnE^RkDl+8Jm&ua=K$>gQ{<
z1$7wxQZD&y{$a-wxmLJ@;Wuu8vX}savk5f5^p;pcohNst@rNtN8STJUXk27eH<Xn&
zRQRBi8OVIr%r;v%!XqgmveM1ru#$fhV8-;I(*<gBjW&06N+jW*b&1k$x)2!xCFmyi
zT+Cev)pHF*Rm*K=uJct{d`-6HWfDBVV4>xYrA1$Vs>JO$5Hcd0(FP!CP^?Ly9GUt@
zZHF*iLb&W6+>iM$7u%8x>~d~XOF1*(C`BwplaQUoU^bYB$!DD8Ic0#19-)VtCf{_M
zIHlgraeX*CBE=9+(HD@S!NPMcgN=BU%ohAYNi<?Ik@=fnMZ2Ti;T1HKawb+XFfiyg
z+vsX)E~jJ5h9e%1WdbbNe>{6aq@<MPHAU29yfF521Wot2`m#8ph^3Ekbu=7^Ql62(
zQyK|n=@6#V7(xQ8l%M8y<X=}oBfQ6AOgpi=$CK!#>WDa4&#)3Jfnff+nq<A*sr9(~
z;zs^-u~M_eN_x0=-vrz9MEeKj=JL;G@A2Nuxf(Kn6jl8y5vZ$aX!7e2Fe9{p5r|gA
zD!%YRzb+-Kt#LD1?lrIde>9zCP#j&jZgD5LYl6GGyIXJw?h@Rs@!-LNySuwXkl;>m
zx8Q?2cfND(`9D+DyQ-^u?|z?WtxepMg0l&B+M%vQY&<g!{T^>)JYiMkr3R2nMzf#(
zTRO7Paid58D4YwREK`r)t2s-OLA%@v!&Txr!_Id>^@q39uGZ;jGdEFFb9Q!?lN<dq
z=Oq1Lpc9xHUEQ9a5(N|p4pHn<aet-g$+m(m{1Ir;u{_)`4=YSVL-904o6i=7e=Gkb
zvn<75b}1OOxpc~?!{?L~>b><L?JEdk8edidY7+`A7wkLOIp}5}nS}RnkZfHMbSt(}
zQXXU0`Ks04+rFpwvSV<HJ0*)EiP#PnEc?jV&Dj2t=C>`23j)OE9uW%~Dmuuc@rbjt
zehIXCS%^u8c9usW2rw;>Lw)O%OLKbq$9~)z)t^=`qNb=fjbSg~!Ci-`Z##3FA(nmi
zL(C4hKaRf5PTOZH+g`BEYNVvO4qs0vUWCDOIEH|auf4t9+k1YN2a|{ZbNl?z_`79Z
zbqNHa^MH7aHYCsjLG`Q;ka?z5u~(1rgtOt7E^>@=Fw=@7*5#Q=TEP`ca@vQ?A&=Xx
z@^{p%cRf4~t!)M5>tVNn{7w&J?X(@&+dmK<PK?(7*$QYhJ3p<b&$FxVehKzl7f*7H
zl4m$w^U<?w&GQwQ5ldy`f>{ZuGJ%caekK2aR3ZEGDG=cEvi5t936}C6?VpmCAK24*
z4MdbwPRJ~_`Iro0<K}Qdo&Hu=)A2O7nCWRu2ud+k?W65xIr(XF?b7>jkNRg{K!uC6
zsz#@`z-y|ii(tOQTv=I3US3&QPEJWlNg*Z@;oeM#1!-<~VA(xdmJ`drp?_-tj0x9!
zv|yz=9|00?iZ+CVe{q<HbBb4Jz1z4dJT&DD5jh}u*}jCxTaXHSL7bqmT)ADWOu8y@
z>1$2=uE`^viEaSJW@>fybo?_lTU_-{k|4gLq^grkpMin2pMl&kvz@eq31~)A1ZzF!
zSzFxE;<B^UdAX*x42{IB83XRNUz0;Z{E26sDS*Wbp2vUyGKs6N<!x1o=`wB}qb2^;
z+IfQh&%29j!V3iK9=`I^YsT9id$uatRx-Y>{#>_3mdcPn&)Sav)NRMd&&;UKl)(?N
z3MgPfXU)h324V#&PXPby-}QCS*~OR;0;*=G*Xcg;W!MX!WohQ=*P-Nbd+1!3G{_t;
z&wzf^ComUltrkTtdEHy&cOB*-NTn81!K)waH{R8wW)mBoU|{y7qeuFr4JPrOThc;f
z+DitZF_UhOtL2O!NcQ1cZ%Jm;(K4GY?aQ(dk?4wXpNJI|$wo#_Ya>4cHt}x~bg*0s
zxsO@J`e-d%XWc*kT5i@mjnuls-M>6s073Ce4@0h2%FU#koz?a>Cd&NIP(j><m6b#$
zb)C!lK@z!L$-KXY9@v%%=8ozBiscVI)y|zRq!Po(Mr}>5d4g{uenkAjf<5bc@Gs<$
zs=1>Gu0IL;3XKml`CF^*6*s~MIKh46O<~`t+$zR7n27wWfop!Pi_eksm>~vC0B14m
z@av4qry`08o+eb;oMxAi1Xod{aDnp;847LDH#Ux-PH4)IPxw54H63&e(~@wNnp4!|
z%`6=pOiXOR-ELcmvS_5leu6HWNFHOd#p+&&CwxquQMY}B6k3jPU}xG9-Ii9yoO)QC
zG%&Gqdjj_KA)E4~vm1?6Bp><|;8i;fmK^@)@E5dtec;MY0h-b9`SMrxGY)d@pe}<4
zr$$Xy_P$;D7s;v#nZms(p9O+2F})x5S%F0r`PEjkrV&44A@5sF@x}awVa*P?j@L@=
zB}SaCqh;yJPWi%%CWF>5E`E=M+KfP`?7bg<jupre%n{mH^|Rl)Zz#U81jx6`4;!dj
zgt%o)Op1_Uh>}Hq9u?^|Qbv;9whB5^qw}x@2$P~Wn?Jd+l_W`KKQ#jIjp4!=m{!TM
zAci!rZ8&B42of`YUt(=FQL`Gn%2o{jsnThO9LJhUoo-h62XmAK^U!w6rmFmn?tshi
zL;{fTQU+$|Eo|cs;_aXpfDMu)8RHEzR^V3xt5}WE^vzHSCR*~ikaTi`l_kjEG5Y}Q
zytALt`l(kj-gcFQEbD;e`6MetoWIcp%!B@47SC)EXvGjhxB2PpoTxyWr6aAwLrV?(
zlQM>#IKg16v$QSf5A91YT?7JEbPSyS0L!i`V~j`L_TNVdR)g$7GWPy~xv{$38Z+(p
zi;Rkr8s;>ftnVVRnN#TCbbH|0xF(E9@qqv3^-4(xiG>1EgL*+-Y+e=Xap|BX4&zos
z7dgZ5h?WZJxAZ~LQ-TMkK2wC#ngNQjwt`QDlT@_%MHcpZwq}Yp*Q^@|hw7{(auK}<
z&L3<sv2DZ|Z8Azgx)FFBau>)5*#*13gY)xdm1e(9y=@DQ?*8VyuVSV1`vxY8V8_X*
zaX)b$4;NR^Lnl;=l6jz2_MhI(`q`iv@u5Ua#z{6@hdlxo{;WZO)N9piIL#r3KGSDd
zkzqw+%>#$@xFyBx2O4pn&oucNIDkCy6n#txF&||2SL_9_{g_g!l)q-X=9VUG$N;yI
z)ol1~w3+|NL7PEy$=UEpS#Y6#7phXCr(lzd$W1M`Iv#PQWEI5WLYY84o+J&&`K{B!
zs(}&F8-7Iw13Fw_t-FfL)@m|w954RCYG@Amk;rU)#ig>N%r-WI*Hl{ICI<<*`FnDu
ze`*->vtu}5%wzdf>AX8V4)H<bV+UYzqif-o48wJ-655U-xghvqgbI<N@jb<GELa8W
z<~e&b{l-@Gzs(spYTlC_$+k=7UnN94U(>8_E+{ZvYx%y`dbz6ZdnLOE&Q{c22z4P&
z<ybHGcACD8CGySBnL=%RSK@T0U}ocJW#ysxj3NNYSJ?d{1Pkcb_*T=1Oe|TP<4e|Y
z<$Of4Vw2E{wdFHJF@`Z02eT{}tnL3qRPtJe!-4*yzZbHY5ZAIUr>#9w?dREENKO})
zPJfWO8j+-GE`<q9#hCHP|M*|fHIl3xL100kd^v4?1XKY&K5cLw0&Wx}Jw1ICRFnY+
zaB{n0cj-D6&7r&wdJA}{>FvRn;XTdj))s`cB#rlqA`lh!I!Ti`dYzF#%5rJBku-Z-
zN?as}_i}982b2F*4<{94V^L6>7Qro;mv8)_r1$v8kJ}p`i9P{K=t4GwW3YmEP;@RO
zbppXs)zSgL8Vn!k7ht4}UntTNu1^Psny;47HZFubcdv3zT(EI)^8A!w0^>Y$P7Au>
zm~(`@&zvOul@7ue(?@5;Nbik35Ag`I<5Lm{$#BljP3<bIB%hMteL693gRp%>!XEpf
zA|EI7Q(Xs=(}};m-VEs+x}fj|*B6<gL)+6Rnd-u63c0@Y4W7kz>UCW1qLBh+F1(<v
zUJ6(5#0QMWb+t{l0L+@l7KzRlhd&aLXhr64=T%_xkdjgiD(Y>M29DZynm!3(Nz((4
zolcO^@{p7m1T*6PKHWfOFr(R+FdF-ZN+C0AB9F&ath`)Qbhu_zs=k-4;n3z@D*TpU
zhwH(@tk9bL^>v-mMn<#U-|xWc*KXv6L$Na?U!|;}=+!r04=wQ4ipAH;)A7SN{+uRB
zUPd3;$q`~+Z>(Zd6l>lvte+w&<MU6Nn<755@0fq)tPhD(T;VtK^SjKdFFtPj-s!0)
z4{?6|Wo>f#q%=jhvEc*SWRM>|k(mh0*D;-z+mu@+6ZT>atPyt`YHbBbVOnUF1uHIQ
zW+TT0z&RK{Z`*Af5)mGQrB^o4$NACF*M(?I(?a8iFw**BM1H;G1B)v>D%Rn;JICOH
z@{$8{D`q*lm{=s}L(|C}4jxi}FP{gOm!OlxEKc-pWfh|?uhDC;0eW(=6C64DF1+OP
zD(Nd@CG!GWN-2~fL<25XJDkt27nW7Q$xM+C2-a$(Pqz;sD6N`K6=N?qzp=!srpG^`
zDl!TZ@uvyY6->ro(Z-}v4y>v)k({SP5Z|!T_ux$2ooaBy9vxNASWMZg?}!pw2o0ZK
zYUw^#vZ17<UiC|CmN$t67LXc%vT5Q-%GMfklm#UuhHo8A=K!AEpV82r{yWjJgn#`<
zm#OiBSD`}_YsYm|u#n9nu=+bi=t<uBET^=&Jx~=!={8iGEH#%fZ)^23dMz$N3eYS8
z{%lV5?w;O;dN9sk^y<j@*@PJIG=v_$+TJo%^XA<{t5i2TRsYBfeR;~`nEw46OM&GP
zMrx3<#H#9>dKQqGK#T>|(FQI_M8UvGE5-D_@0lMl;If8dfP$MR1r5b8Ab!cdx6<6>
zv879hiWAHTBd6>;+E`x__(>q9z1?CZd69o$pfkz6Y?D`<N5m4AJE-lxfvC_H6MwbS
z8OvA;$S`-T1WSjfoW5P|b)ZBHm4mS+2_|CiU~25rGT=@^QnZV&jyOrdR#~#fe<*A5
zDTDyZrK3d{dG5xgp)rL?*&w-&wb1Fb2P^fm`}J<o*zqVoBT<u%j;_4?F>`v1iMD*X
zrxDU6pz7S~Wprw4N*%pag`-voe+dpA_VK0QsviRz16kp3Fs2vVfJqbbdy49&1de?C
zHgGH`1{6Zz(vUm-soIHLPY+~WD^?v=>gV6ks(E)#ms<PeaVzYnz&R1{8j0;$l<lj0
zjNvp=I2~Xe(K0WvvQTF>{=7IozZ((c5eQ}uXC`&W!Ww_?6D7&wvi$kov}Zv?81YMh
ziJOL(wQ6APk47c}Y`J-ww{Jd*Q;Y{`-x)(JKLeg!Tfqwp$@@}rkzn;ox|AmBxwh)=
z7RpxFObg(Pn8mQ4!Lr%r{W8`4BjfWysd8>VwGaBn#s-PZgBG-hhlxFkGsEzb=6{5c
z<v)I-QC$T}uneh`l5lx*WbUm|bF52gKY(GrBIx{6RrY%5ekT*~c9gXiefSV3Mvp8L
z`1$O-S0WyU+@i3|e!kHBISLgm@+0Bb6W&7nJIHJz-FW>*nUibb`-VnZ2#;$d7^!kR
z05if&gpB3CFeMj-;FLh{wp{S8aCnn|zT?ye_xLn0oT?ZOVne(7#47W<#VN*WXfVCZ
z5$pwT<|+>2f$=*!#)_MA*?sBm@>+WblC&>+U+*@KMK9CMP`NCMEi8)aYAdSYn+K*E
z4VJ~vkYcH%X5G7-wW*%>XEU*WB$8pp8U^s)bwg3$QnO4NHr7<p!e{>w%egma1NTam
zzMVqQuuB)j>M>*V@)@Y>!)eA&svm+75G3k2F84}YBLCQB3Oq@d95__i@z3)BuW>14
zR)~c03wsBYQ8cu)Idap*jpvhjP>n%nyK2wsSf*FpOPgboDiVkR>PyC91Z4sJhQnx3
zpEMjDSy+s_Bx)(4a|P;6bA*1dGn=4mqYTfD%qkL9en^o0JSAk=>WjZBTYfCUIc+85
zT?xTdnc+IV-)yezA$_=hA^K?wTuQEw>`A}lR>!!(*3u(s_4SJ>$Ay8pHJ{5iK`01D
zX<haQt9Ll3*X-)siB^(hHZ}}mDGqdnYaH1MI?@|`V=L1Hp=vNkm?FM)d&Z5A7-M~$
zU}stv$+wP{f}Y~j;Xe<T<!Avh$*@DQzCqA#q{d`u`_te&u6q-y!*V;|?@9nW1TlWR
zd?o65b)CQe0sb-6_~8}uy|Xg9j^Fy^<i_vO;R#U%DP<@Q>Z?V>W{hF&`Kqqkk{dZY
z-HsnrOh(ujCdEv@Ma+nYImgQXW$5v>`M`!UN0@zs33$7*b-Th<w2mVwlyAFNw-T#)
zkrv_TAR_h?-IabA{e1`gI(1ED)a~LG5)xu!a#!<XWZm==BnJ^xBFWQ3>h!Qq{2dxg
z<0c!Jb0`1Te;r;b*V*_)rHT4oFj)7@8kZ9qR^Ue1f&XV)RDQO$*Y5RhweK49y3gRQ
z{M;^AUEHuLN)&7~R2#={v`aLIgYC1?B?|DoWmT=Q8gnkU7ik#4|J+mNONGSKn5Mhf
zTi^Ox8u<3Jz4%vbc6VW8bY{peL{$#SM;lH&rVXkPWysIu#y>}*C@vEI*>Tqz5Fgl-
z(6jpzx8G46Xt>2xXU&uor#4g+9jemEYJ5Ci92`r;*A$uT=1l;FAD{5gS%3TR!}&}X
z=;w2aV8Pe$Ct;Za6#cKmB7P?F(#ks}6-rMl(U7M0W)-E(ET0Ft?5^E*!@uwh2gEF$
z@1)h{*&UE3#wVcfO5-*wW=}6ceo_{!Ey{?bpm%eTuXxC9^YXogmwq&7Tuy{-hS7mT
zz<_?XecTZhnP?GEY2qXiUal+csexP9(JG>0M!5F?YdwYm`DPdy_MJ|^iSN}j$m;Gn
zocb_@`#z4@<1jq0<jdf1x#=^}PGn6N@+anS<;?Oee=ZsE5lG64pEFl?&Ra7O>bhLe
z3RbywVdXL0Pt#<g<yKx-%s9}{gZehN|6EOT;a~XZ4biqbysocmn1&e2bzo(mA7M>`
zgdO4^!4_Dc;oL&1GKE?S)@M2WU*$No%tL2+BVYX@n(<+nu!)_@F19v@Goo~{CHO;l
zVNk)0;#?wx0!*-h_ef-JKTWmC1txZ@o6<_01WD?pptp)E6&ug<beRuS{V*m=*(Q;<
z6WZ|yUgwls3z>)qwncAE17+?mWTu(ci6Wfuz^60-2iIK__Opwtod6AyI^)}fh~ua4
ze7pq2`xeQ~b$ZxjqBcjP>CcqaD*?AI3U{mKMjWurPDbg0LCeSHT!BekLbYxqJL=kM
zyO4H2Z1NPlV8DUDc(TSs8qnGMVq;xt$Th&F__S0rMoBBE-RmSeUtL)ymg*<%;#QX$
z5*7q>bm5DPEI3k!kOgK{e2S*}eSH5v#oQrghAR4Rgy$Cd)cl!qjX9J3i2Wr^U}q%?
zr0#cy3G~gcTW%Fr+yCIHFMCzX$$n=QB62ED!rMD-&#rfUx^m4^?AM?Gq<jR|?Ig=w
z`E4qD-yZ100W6(feLXQL_31zB!~J7AGJw)2x5l(3>s>`Kw_QpJ8hO|`LVt+c@9B^t
zRmL}izLDa5(Mvb>#!Z;co#Uu)dB|?;BrdDxc^jvi^EEPf%gEGuGw#;NUPMDX!AMnA
zM?<5_VdJ@(f#d6*vHa<exP${cZu=F;v-*Xwd|{m)q#HvMCh&020;b2~Sd^GS516bq
z-p9;ou^fMAs#Fg?cc<h<_<i2XtMLj>R7_(h&o|%2iPOBbbTe7iZUWrgmq2{$Qz((j
z1%&DF^*)F~k+`@|XN|iCR$B&5087bRE!pcN#0oy`o&gns$-v*~uDQf~TeV!#Aqlh<
zAY?;L)0@T^c%<oM!RId6n_-{>SpIu=1=kK3jYxO$KJx|z?M-6FYw<YW#ui0);`OrX
zzQ3u6jT^o}J+A#Fw|BdQlTE%XXC7J08XK$DA782-HO%N@g=z*Zr`DO@JTK~&gQ;&*
z%PvtZJp-Q}mWy}hS#q?srokeNp>PtSl#wT0n}|u16%>g&iht<%nY`+NsMg3Thag{P
z8)nZVcvACt6-4E5uPu&$ixcQvR%<A1)z9KP)%QwzsH^tQoYEn`+qEF~an57k|5rDm
zj@qQ4w1$>$LLilkJVOx6*NI^-o7GZb>TeygK}SzdcdePHoOD}Z%AY1W4aFP=Nx|QE
zmbh9M0s;RnZgPQq5zsV1NqUuf6%Bkmub9y~)VSrI3vRBr0S|E<9oYh&Vh`Ht?d7zN
zPl;0Xh|zN?mO)%*4wgt$ou6IrjDPq#^ZlA$2d#hA)7nIy)7Gw1{&~wP<xYFQ(noHM
z;6X_eOqpW@$0d*-sS9fy<PFC9nDMcNvp=;DVvSPRI=%$MecJ-gPA_;@Rn^t98aag*
z=9g486VYe_png5Fyyu(HBXt5gkox-DacU91pSCX%9+6ad#V)b&eW1FsTrHt;n5*+T
z$~)UztmV=QYX*rSxQyFK4hz@XU?odxnQKKncJ9@yUasNOowau9&NH=DNKG9x4hxl(
zz{NcWdqKCddJu56Lm?B#c@A%}Hn7e(rp4)NG1VY7Go()t%!~&L3(tX?w|LiBRLP<#
ziHhnBRh$@Q4c(1PG}ah44Y2>i2Lnkly`8$m=^3``pgtM2hv)EVRw08e{41~^O7JqO
zu8R6?Uxc{JGxhiuoXYu@-qE);Drt}b47(9dO?(nO7Y8C}^Hv1Fkh_0&^!;)o(U7w$
zaN0<QkP9P8QcOHY-{eSM^7gwXQAyzG{}zSw{u_77=SPGlIt+e_%m0<+=`=T%NFp=6
zK}bbIV*WlINu%nku`cNEbc)2k>*-jz((bz@=5%Q5=qTddP*@n!rl@64`$zXzx2d<*
zVMDP(A}#=f^;<PC`)i!UJVKa7%oY3Mh${la$%>CmNtI&cRl{(gop_<f-`@sjsUkwe
z8oT{AlcOU1Lh!_(DFXna=Ido$2~lO$i{eH3#}B%jI^w4*@Vwjc8<YyJ#`cb-hJv&)
z+;BZ~d;|&a2@ZPb*fv6OcJRKR+niY>N>u_Ii7P<;1QQAa3&!I9LVU0kHT4*cE6L8J
z9r=v%H}pe6M;`H4(7i)XefrQEk;;S6v*^EZfE&a)WLiX~AM5akX<eh;Y%8H<9wyI|
zKuQ~4{!Kag`O5^MqCPemSg3AXQ%XJZ<(&O5_a83B4|5kD1ySkW6DuP8OsiX?*uK^?
z`mO*T0=J+r0zIhg=YMBr$<)`B=fYCoOA|6al>Vnin{Aa`%AAvHkl&Ave`F$rJ+9~C
zGB@wp`aFk4(%$-~V~FgLR^uuLQ1u?(df(DmdU)wVS`ZmM;j5x02dC;Tegy}0e+<L~
zk9F%uH(&;5DQVO}b=9#}r1b5JgJq<%LVycsbqt#1Z}?Wy?4l86ju%}wahFerjkZ*P
z^m<~}-LE?tWkX7iZ>-#ba)<T9{sOJP3TuTqgxIJjrcwX2HxyJUF*F^r_JqK)8%_rd
z8JAgIMKORQQm{i{((*r(&Z5!(x)5+$HIV4`-hsE`*8H_FiWdWz>^I}5p9-Y|0POMk
z-WQW9g)^P=OcnO&I(nO;sd*YShdZ}so_voUax)VM){sAm*f@9gPvxiN+<4WB(VyI!
z=YFdTOr<B$(VpC<<0nV3T&+C*&Sm*hV!FQ_{i!MOM9EtL7mhd6MqWWq4qVfbm5msu
zEwlXxcE|9-naM@u<daz&1x-`t0JDM}^BWzde(pZWGLrIB!yOb>HCRH}I)<pha{Z$z
z!`)nESw9-4)`o^9Wpwq@QWu@(9!Jv`_DzTx2<CaiOw0~3&TOozf)370oT6xtb=qJg
zCF%|!*&~mFQYVDGcEEEJ+fO-d%OA&x-(^8EX4}1rZzJ{NSuzg}--MwpfoLiQyf9uO
zIz$Xvp-#*#sQS_dYj4xC&SLJEf8M+$D&~E*ftM6myX725PS;3Lva)t*ypNL1V`D4~
zIynt}TSMa7YGQ_lhRVuXT3TYd1`24R2?5Rw+)J&70aq7qxg=CNk)@nJ9kPUVnz@<Q
z7iXwU+2{$UxBMjtrn~)L0OFYpJhc;|3WdLx)Kbs<dW4vT)og2?9m$37KiFQtvFL*D
zwnzPCS+lhD8VD_dqST<eqIx$(o%nF(r|^~=o<6EHReBu|Uiyay2Q=`w)eD_D8ExFy
zZlUB;lv#fLN?=d*-j7b=?~G$NS18D&`?yda_q{Im4KB*cDJw+_Sm1WOLW)(G4xjW2
zB~$#hprNU$k^Jk|FO7vOSupC-Ol-J=gbDAmmuk6inuhcgv+%yS_50s&iv;H9#BiO?
z$2kPMq8rd;W5P?T>Pe=jBQu08w)4{7YwC&A0^G{rk_LRO1x)}~wu<!l7=<4ODPrRR
zWCJEkMV{HkK2bs!P|)V0GQMJ$%*1<q`q4Q7=w0~xg=A#g#Xn`Z6R;T`Rt%ge33`1r
zku7U=2@21Q%IZoJbF-CE6nK?1@>xz7ybH4yf{RHDGaSv%VCz+kf^G^i(aEX(_oCu5
zQQ&$IesFFH!)>#L?CjXTuWSW9gFK9O#8^USPSEGp%dH8hZ8ULp_?nkp;&1JvX0Wz6
z!sx!{xh>~)1fHIc5LFm6r*7N-UzlL~V$gQ$gBz#G`9%=&j1EOY$VPA>bC{ys@&p^b
z5hj=kEnfL`!CO4A&l#j2-%#K*Uh|tI)yI0nP%GZ(aa5;LEN1>`P&b+XaZ8w}-Lvt?
z&k8uLQCdL^SjsSM9QmTu!~02WA#aRrU|ycckFM=dF_$v{1r&I=FoWgH5FCVRRLA<I
zZyw|JZiYm~;M%3E1$s#0cH9b28=O1qI6ArOA?nK~Y;*1W0LzB{`uNs->v5}p8F;F3
z-O#2z3V*LDS-sOmeye)J)cVEaw&@itr3`>%LojFCoU3nU?%mUo-_DN>&bFv?W(l)K
zSiTgO)!0oC-mryB{rB*2+bwmvie|7T)a`s^k)0Z2skX7^`?S=L$9!iTKdOnslb;4-
z%=&C>z0Y<En^#Q@n93OiFA-_9(hzYUQNgdA%1>HQ(k){MkjrbcMBaIFYT=mq0F>fV
zqUlypS#+GYljj>kg53xxOY<u!obP#2q2oUI>1M8hMHYNRI9$G&s{#le6+}&0${4)g
z=XZufUWo25_!j177hP}O3sWDMbFIr5bqM#cGTUoP%1`e(hSQ*0n{?)Gh<$k2@t*bU
zSW{=WU<&(TDERA&>&<ad)ejNPfI3we7=2Rm?8ni@cF?8czg@iaHWWp;(+AVVZmi_;
z)krKF*44AK_8Mfo+HYa=#_<Ih!HtKxB|=-hKJ1)MSO%`KYe=|1_=x_vu#wYxND>UL
zT7W(~;I6s^cYPynEP<Q|pKT#0A+i1UIJ6IPqNYR$41{=tGx($lOQ*K51*m6twH35P
zc3U~2;1}aT2K4Oo<q!W>f+%tjp7S!^y2x%`)cln>uE67<qphLV`LUdyvqn)%W+>w7
zMuVWt!cYbiaftl<OhzTT4<8cpz|wNYI2v`=0+C0yB{HhTmMw&T9=KTpMkfbT)}T<k
zBFL7X=yyE+_lTVHf2AY`l)%EZ6a4}!n1+s1o)pLOfr4mM)R+77wN6hkY6%to6E~@i
zt*}Sep5osjQn%e%qj%(tfc}oRt=@opgf9HfOwX$E2p1l;$C#Z<yg|YU3q+F~?&mSn
zb=M=+OKggPSc_WTLucpcKjlQGmg;U^c|VD!IyxEz%Ql^NNN8e!i_nASNwv!44t7ta
zfZ6B7xhN+6$`<XvXJu=ey@fufhjXYyPnE6>saWMdCkB{0^*bC;dVT#~zRK)PMFm$4
zFx|;a8v>8d54j=%XwnK25Ag70g2YfOos}b}JeEUV8M|SS^&`)xeT{;&k_Xs9z81zP
zoxU2FQ)!PP#M_EM8h0l0%f(5?Dp*o7M9nfT{xjccW4K{5lYxG0e0)q^VSE;P#n3>%
z)ov@l<Uit*QTXLMv(_a|j@zY3-C2=Hmp=$5)X!JtY%{qVz552dGriJnpaYGQV-9qk
z$1WDb9@dS)>%B(U*S0^w6l-ith?VNND3+mB{Nu#2umSwLb=_K7$iWUG#KWl5sOTsp
zWTflAxY)S0`bG`)$yD}Dtc1xH{(>7{=9cVjc)i_^-x~XSkZA}#HESxXMCP2I?h9K7
z-R<}LAzbl<duQugIT=<rtWkPSYC{w+smy<Y1tZPQsrjYUEX)v*BMZ`P@#M?b5dQ`N
z9*bdOW<c2OPatY(cBP7whW%^70#!PLi&pK_`1s_gZnKMp@@59qt(ZOinx6ia)1%Hy
zI@nxZyA=rK6zY=y+PB~pBf37R{=#!a{j6X4lR&BA^X})q203-*P&>kCA_Wy8nu5VX
zb#++{F|eGGg`LjK-STT_EwopZX$60BMqYVf2%s%JGK{Bu7NvBGTn{}nXF=fAn%HTo
zt2Z}LHUr`4i3eXbQTClIQA>5r@88bPT2j<+0ziX}(PU2{$p#Nx?pM83w2RFSpF+Uc
z=@HbJ>Uf#8V$|XnRGIULk37+f(p8%jum1U2CtP}%x}(CyR7Z`Ba#eCB$R43T1^yB2
zU-3SX1DR+$*!vzH;9(OIgf!%4Bq|yo$<h5&U{A3++Q<36{?(wjx4im>k@bh;M!O87
zfOW6QpBEO9;G^rk{QgdZAFT(aFl%;=B@1T}=m;vl>fd46Hg&OR69k|CK4k<Xgh>`J
z#6ZZU)mvgMldhc5A+QND%fEWPK7z95I)HmIxk~v{yH<O~iBUxbh&K0ce*XhLbi+p{
ztya9<36ICc!&PlXn`RX8df#E$5+NgP3jj@kC`P8gope4NAG~Zqmh*51K0g>%;XGV#
zjP&2YGo;Ay*+D0J+{XZ3r-<K)?k=f^HCVTakL6)iwJSdT@5&z?>;GL>0er6B5&D%m
z3GL0@Shj(|0|;W|tPuLFBf~UQr4~EggPX%<=cQ8yZBXnv%lRlpiPn$9wIcp`x8GTs
z7jU8fpe|F~)OFws44@7yNh)B#9Eorc;;Rr%pXhYJj`Z_l#++1h4)O{APCtrYa7Quv
zrW4z<*&uG8xpT00N`{V&2@sNQhxtG_jV7WW9v!B!9dy<{HGkVfKAjePjAe`O_NxAP
zN023Lcuxi6=VAqYZ+2P}6n?Za{&ZV-)E0NgPlArIuwzizP9q{>+YmY1YY)_i{-5=Q
z_2r*G(LR2DY|ZPK8ypyhwAUdaIXk!iewGbBAVCHH{(?1^T5AXxsdc~1%yPRe8Wphr
zq{^bZf(u-VNPgO@q&ZPw$895!=a}RcfV$Z;c@-jq)34eJTjI%*4?tt1<C4(-sk&TP
zm8Y>ktcF8=oTVPcnP;-w%4%mYdFb93igaDq>gnd<;VwZ!ZpkCTF`Tt1X!{A-rL}d#
z3tn!xr{R0{QN$<Un6((vqaN?Abr$b3;MITM)J)2<Et{T)23{)R^*Pf~C-GF$K09Y;
z&dHG@*+s%%$bcjzcJqT0w4X%Nw;74;=i5^j8WW7vgl(x6dJ-3aua0adAcN=0vNdx&
zIK-fZm#VpVu?n6Dk4H{5ZfMtOk8>>V{-%Gxi_r>Axt9ZsMmE_sHqsf`_qaR;OKCpe
z*v;%Na@5B&(I)))-9kZ2O@$o(tOrqQWe#7W-ar><%jJ$L_XqrJRE$M_aBbw`q982t
zxYtTcgePWos1V=LYC#>Zu<qjJkC+RnIG2^!$btSb%fb71lD4_9b-Vvh<KHcbU9?|T
zp{ul$6orF;99CVIu+~zp<T-}!SoD?>1;6BT2Ut(C%$y1ucr~KV855j+-^0ViO-+<C
z2;*+!&TmBvz%}rvP<Lg}_oYGmr~{J%C$e7MiN+aPFPAr*%AN-QAq(-$*uA5WfVaaR
zj~o<;Pbiw6`Ywr3_H|VKB9pA|?#8rH4UFL%ej~suW{eXIPjf5|da?$LH^CMwE?<yX
zmFW(4+Ld#35WJe~dgA`_6~yvo;}`=piMs`#^}Fh(9LHF(H9?!#+r?DpUzYqAk~A|M
z{npE}tsgysElI~GohVVd$r$IC6dUjgsc`!A=ux~WLSG_+=$0X|G7H&mP`~ieRF^?=
zw?_d;*A48uX`FojfL~s#=9swv`5p?hFw{*wP?}f)$9ATvru9|KAUr21A+kL->4&52
z{(#g!0tS~Q>>8Qgr#Sx~`YP^5=Odm%xaXzWJ~<D$LK@A~TnlR2O7FR)DIfgU<V8cF
zUaY#ov+&RZh(3<65%D-=WGvHu=w58Ct~Y?iY(kwX4_+;VEVW6!lp&ipDg}}DM})D_
zwj~&93RY~FRg{AfuoZNp`BBR8sV0r{tmlWDHA&#4pVoK!Z_R{UA7$qodfmu4U8T&o
zn=opsUwrxfcKS+&f({!v#I8_sS@Hllmmg{SLvM(q$ct2EgP2KL$s=<^n5I=ej{uPD
z1qUg*<{*JQ5#8g%hDfK|H4>kYkOBvy)-kuSB})4bV*d>LLxx3BQ8%mJbn)aa)9!(*
z+4Jr;WZ^WLN8~jE6HWb;r`BlmAOBZ_Hp;mT*-YYzGqux!s}nXMOer3iuXM6(X5HUn
z+IBH8Q`N$S!_s}tnDpZjaIyNY0fO}`b2kgyJ##BLWe+Ew_{5ZygkP{2&{g7NF-V7F
z_#GoPwBX}dv-5+8Vz{ClG<@(czDgdA3&CZNmLqOZvfsxZ3oYn{r{pP9maT@*s~|$`
zOIDN77m4cuN-t6|3|YSCSr}LyMgXNnGSZ4+855oY3dvBzxJ&5A^Ju6PAjPI(@$UE8
z1gz_kvmxN~9_vM3A`Jj*e9Ml;|E#IDpU!|>%YG7s3LiojLq@ZU%xrvxZ*XJ`6%c@;
zt~OghvwjkJWHp|vR-`Q9J%VWW>cOYi)SY*seRRJ3LZkHg+SmZCH#w)J18He?WS}um
zP@$=bl_6R@N*s9`yQioO07sAw4&We^euRd}n5vWyPP|5C%kcZc{n5heD*Q3cV=oKh
z!W_wwbINaw7b}B;bo|DLI~q-}f|UO|A~iEw=WhFM_yfiw+H9Sd|4#BcGp(V341Imw
z%gd_`9K4M1-10NQaCrk~Owx38B+4&|z!oQiHt(Z+teD7YmSlNgwUg(8rTKibiJdL}
z8rC=|4A+C$jE*?r%YlAZC{Dsx)p@m2yHT4aAyikv8SU~8`Us1lK>p`nl0$vJ;?$^j
zAyV$XL70B6=wyT#B6en4>nqB-_UAen;`3_0JUNlp5}j4JVIW$qMPbLexRfw>S)%bh
zjrKEj8;{JTGz*&q;=AzOEga+<6R6^@dldw;pKC{K1JpC)R?OQXjkK;^Zx9QRASBl{
zIGfaltZeootLZ#R{s{sH!v?IJPhkmt{>!eWMnfkHPK>wUAQbg{mmIp;ETbfn%L%bt
zIYZgMjN|g(q;69haM43P3bLT8vi#b@;+ld8b&*`+C!I6^IhsL2HP2VRJq#qCMfG2%
zxx1>Y13QZylElK^UVMOp-}P;3i9w>R6t?Q{QsU?ym&W@TX?AUw?@?c7>*tK0np~am
z!kk6*31R@kn?&O%8e1f8JU#{UyNIkH%xPz7h9a%FoTB+>oq7%)_^WBQN$pQKfrn*8
zdWIV94?4h6*rxsiS@+{Rp7cS$quxkRHLOCR5XnRx?8HYOby7m^`zK2yQNNb_I)+;i
zd&(f230!GlXjey1I-*xm-ojq&WY)i=-j_n$Nh`CG_{0Q-;{r{cfyJ$%uPVB77_1~^
znCu|sKWm$Q>99lj-2<>XKGHhd*amrZK@^W@he$w?8)~WQNsQTNhL6+153_g~i8#$$
zQ^gY9O?qnP{=DMZWsi=R)r}pi%Yb~kmP@j`Y~gR+4&H*eez4_6EwALcC)@18={of2
z<c&VI*O<ZIIB5KpLWm;#co&u$+aIMU4>X%@-9Mjir-%t)0C96Ny<d|f;H@o2t@VAS
zDQf|}M2PF5@DuiiZPC04LdeR#<kwX*p`W_EOcmKlxzNwVFxSY^txnB(ga6e6({=MJ
z@Pwaiwh|;vxZ!2RFrj}SHVq-2h#f``@WB1I@yHEkF1Fr=s13}@saBat<ifd&HCE?+
z+-DlsB9_eYOgSu;GHzwTs42JaH}qG~04nHOxF}HIC1eO-aZnir`gt4=DVO3h)BN(_
zeiExkRS;H;?dQSA>MoxKJP;6G@~%eYV6^6rG1ZuwACdf*-AJZYg0<JGs;bBjL;dbS
z(#t<4|GLK@(|R!IjUS<YUlP2$g@53$xEot9huV>{jkhg{O3HM}nytr~PlneGd_%gj
zEf=rbdwOflGa$FVp=KtfQ*v#s15d0gJjjG+-R~|6|1us`(6n~0-D}}m4hPY<Ht<e!
zwovA*)~a?A%%BE-i8E5w!RA+`nn?Q{gR}Zq23k-Yd*8fB6;<Xbv=7Ml;wE0^+~tTu
zjlrM@A%hZ9G+qsRZeP*Ix6U&I^fVI1&%KA4x0;Uj?`Wz9McO?~|8`LA#Oa+>BxA1m
zT}eoITT$g$4<R@O*+7ca+(vG*oqXndI*cBtt+3f6$cSr6p+1ai_-_IDGFCb4jv3>K
zW$qtoL*AJc;M%O=1zEt5Q-z3dx9`(_e@1M3Ir1y9QXuX?%AgA6Fe-q}nvM}>%2=46
zp2<VBv1ZIiRasybQ~fgY?~~q8Q#+P<_Hm^8bHRl7)L<^v5i2#6x!pNajGM**RhW0!
zphYsLTP))l6yAZ>nTY!y9Rmm){S>aFa#TadFxE&+8feh;Lg=T=`NJKF-~=kzghujT
zP!ATYBqvKHpRob|6WGWiv9F#`Qz}61k2`uG3z-sRa}-zl@bflc_^G<^C*_jY>s5R!
z1doyZ?dGMsH1mW9uOa;4>=Dsv2-QNU#wWc$1%>-me1txQ7lx}`msg3q#&}+yl&p}F
zNi@SkI?$25uy^FyH^LUKmYfu6A}zQ>EPS)76cK`Xf#W(-7GSU=0JAY<nCX_6yzZAc
zw04cw&mC6UZ7jaV^K{XyR$OCZ>Qp!=?p9BtxQlM;Xv9|y&wNN%2TH+EVb2mBo?!)t
z6=j`+b)BD=L9g$2s@-J2zwIP_<?*<UmDhzNU*yof6fw)MGZH*LwcRl4wz=E2HWCo>
z?oYRu6U+(50GI`r>N3NGvNe+T>B@DOZJF4;YvN<VBfC9y{LX@};vt-Z8cU|N<oB~s
z)JqeWz7LRTT<9)H6N>+?`*u*gqqDOMf!@^+SDu{-yk4U=%S>+tOvwx4%2#dHxu}7U
z*`Mbj8nmI<rztnrb&QC{0r^q7N;{%7ojy=r2*?Bp(4+Rq>8~$mNTD1GHTg|};33%i
zLA<BKAY}d2E{WqV#=#)us6N@BlzJ&_LG`|fE<3oOU8A2PdMUMnjxRP|=EJcRQ>TUz
z)-A#f5xm2jniiGctzs_kv4mMy{|PPpfXMJDF~-acYpz*^Y78~EY<-Uqa>_*)^wgC$
zq4=Z+AOz--8rCVMM}rKedN9hp5ALe*E_;lrauBGr9L+h;53ce<dwdhU@B{VI@K0>Z
z>BhnLz3snDomFt1)scuel`H1Fn<A0@Z!#_>EQzoBe+ytiTvB%2rDUDpY`H;-9cQds
zUt99`O@?g(-j0>0)6HHylS6VzwK7ddjDR>9dGpywB<7RHMYbhP?Rd<?Odc~#4%h9Z
z*_P`l#mSYn9&)r8!@UCNjVe@)VwgikQ{3xhu^rB(Zz+iQH<0v^Tt;SjFEM0LY`;JV
zk>iZ3t1|&ln+jTBww24}YI{qFJ5Jb2`79)Ocv-fU`b#XGCUPXn5xJ_fCcmPn1~4;s
zINiU%+CSYfM}EGSe{M0v7+pDSxq#loi%+WyG#<Hm8OXzM*t5EhqB}_wcnU2&w675G
zxj5YWiGpwn#(|#VDRZA!Gfzd_)i0oD{!STGr&zO5ryU|HpfQswqiLiz;Un-vMx-Y+
zk-9Kw)*t7=KD1y_cyUd5;yULX1*(SA1KG#|H&8f#6-f<$H5qR5Pg_1bZY5{(q`;Uo
zrFwsh-Vx|Fp!s>R))Js;n_SzCv79G*I6|3ICUk0TbDXZT+UWxyJ`yNQzKL0&g}AT#
z$(?isR_bdS&C<tzSAR`zC5-N@vD2n6i+@PLI9cUMPVNMzGHw)CVnTlhnkjjgABH#*
zdDQv91pCE1CSu0h#eGR*vOd35aU`FXp4ori#Y9Mvl3a(RAuIE3j-sjvufy%UN6xCQ
zUv0UkM7?L!zR-uz+bQTUM_o^6Di62j;=@d4vy^_{!9-8W=`6ojrAe8|57QtAX;0@0
zuGW2j8{2nMu`}4V-qoZ5biKh)vA_}$Zt0+@KtGW1#R8tA^(9`nxx21``diBGSu(AW
zOpJ)h$c>rA_Jm4019^Fs-Qjk-u}%L{NcLZ8TKB<R;OXw=NTStuI@i90GK=!4K?*;p
zEnJx*gXcITm!-JM%=FaOtV-(D55H0RGanO2YT$9xukqAk8k#_Wwp(C%Wa236{tkwb
z>xaC!#vU@V({%QbOM?kMU$R{+SK4Q@gfC0u{ayO6iS%&X7iE2;8#=CKdt-eZDLHWO
zyK1`+nm8;fNX}e;$tS5NFu<5HO^rpaCUDzy;ZKBE!WX*`IDz)PX2x$JdmPu}Vv;PV
z1~|AMwQKDvP&7F}h|xDSFn@sQi~o7b8+Z<<8~L#_bhUB1QL0B&c|AWsd{=Z>FNG9l
z=%_Lj^u!LAlB+YI-lY4PAP?qQhI)8NGI*?vri(ZV6Uw#pt6sm4rN<%bkMwDd%tbPG
zfOt8&qp#Jio^5d`-Rc`fngt<!j}l3jIG*^fnuYJtsEi(<8p40FmU*uUezU^!6Vf3^
zN6E>K{uyPU=g61+1R!czIa(IUn7{u*wKsYDWu%szjT>!?<^Kxb^ZKuzh8&c1bQOsa
z5}n)?6`rL1gJ6;|*(+}>*G@!Fy@_%_*T{!dfAHs@lNgs92{OS&)gAx9?R7dvx<5~e
zQ5oSWaiowyz%tXw6kOWL+<>5KiQhPD=-W5OJI$Gw0lf!EdHBj><Ez5zN)-hBt-T2j
za)^?lRCS&v7ZB=CiO5>vnE=o|ZM4A$G_fZw$hUQ$CHk&;G3obn2m5stJw&;g{q!Hf
zbn#9bNXEb!VO5%wZRDbNoN~6~B6f|WaAS$7WlUZG)Hpj*SYqWs)<j!se;K*?nI3f9
z418RJwyQUhv#Z@tIdIzu$DA6ZE>Ki1@Yf-Fc>_-U#JkRd*x4MBIGle6w4WSbH`SMP
ze~s|Dt)knrLLNO`6<x`~{AX5LOUdr}#TDQfhNv5A^t?f)Fs1gfburED%fA{{RN7ka
zr(gxxO`IL%?%ww{T1owXHiz^9xR_h67lR2xnSYrLw=K%)_<pWXTUYce>xg5og_`}f
zhRX;##|?$?YG)-G<oUuDQr}`$I<<iZKSp6;YGJyU8{cj70LuAssT)~p(eL4?N4n=g
zIWmMBBq3N?YHWJbCMw-GK`P`2fDsq47*jwwc3oVOo^8KpQF;D<2<f5-5|_5aiL}=Z
zK}1vqN2h&S9l~?7V!Hk(5;H-?YO!L~&B=4oC|++9-Z4SH8%$1!#Ls-k`4s&;=3-yu
zU+=?Pka>R4Ez)Z3qFq9|dhw{^B<S9Fg$-UkW~bdD+?ZuvwG)?hQ1Ty&4-~}65CFZQ
ziE-_PDms2`xzWu*9aRXv&Gx;V<e=pi<@uQbtDS4iI?58Ad0G&7<Um5&w&J>@A@MZE
ztAgcS!_q*R5AAq_P?VBHWW=$=+>>9=nC&QNxpPP(fWgMbP7RBl9)TWiuaqemBs2Ag
z%hpCsN&P!$juuZsw+0l`#;!~+3U~(^$5NH4<AuNcK9&g!5>(I52@guI{?8|4kE!fu
zr@i(2ucIi}NolV0?u&0;G2OTlqC2++dK!DQf~bbH92jR0&L$duP5X(pwrZNX{tqVq
z_oyxVA!1@8-|{%d9YezxL7kMJBz$HnF+@Dh)TM!KT^Tv$G=Zvd1_7+p0LT%cdw)Hj
zi4SqJ1Zf?-DcKx>zC`wL5icv^ks|ZcFl}qiPKW|;c!yi$LN|Z28q=?WT;?PpHBbAR
z2EZqU>SoyTcU*Zv#&apc-5RU#Yf_BVjP%S%_RAw%GbkT{(0x`DxIYI48FN!Z^CFz|
z4jA%+DN#f)7p!TGiGv})h>!ojiXvlQ1A+_CJb1Efo@}z82R+C;mF~3`Y9$BJGQvH$
zOo7+cVd;b<N1yMyfV0V~kv3u4+j^Ed#A~V){cO-$Ry-u2$1FsCj~oB^)6qyU!0W-%
z@gbpNQLg7OO7N^P$|}XhrH}j`RX>gJ{4(?hN0ovezK-2Ib*vR|E6!GhWEOR3*EzVa
z!Sx{Az1rn&kqoV{l=h{$FB;{@@0+cbp)={2Us`Q+Gg}|^qe&u%GcH=i=`9mjNeeCh
z#SH5!?vX>Ec;5r<Vp(p$6G;}cC~SSe{8%Ui5tH+$+tRi$3k^4=4;*kj-KUij8$9P@
zQ_C?B)$|w+{pSj>#VJkm`8k_>zZvR(aI<(h9r0wzMcL}z@GLIo7lG|zXera)3%nuD
z(mNb4J4qMtft6v2x~=E=`b*SEaJVewFrkaFzF=?9@>}k4;2;=x-v0cmr?6O|f~B%|
zeao2DB_S5rVLk~?_#3OhbKGXjPaK)}gEGRkJf=u{#CCvL&?o~E!wKP9r(>*OK-A8n
z$jAHp!Jodn=u$m2%>$M=C${9Hp(bsx?pr(0%&XOWai%k@f2R-k<>Oy8naqWk7^1fu
zyU*~9jL^p1>*HHxY5Zf385b<wdy0J5$JpAN5v)=ydg#pz7;yvT=8Bp;)LzPl*`MZI
z7zstxc@4Nu<%Cab5{v6i&Bi}k0m_2vZ3HWA@Iyt6(_6{Cf?<kJ1Xv8=9rl*z$A?(>
zxFo%tXxVp6Rz8<P_yqp%&l^8-fsZSnTS(SwJ<spoFfPr?sRgq`^wDuVPx!GXw=WOk
z{~P*M<fTrhCXViz^p;-%pBnPA1v6zN;UsaO08H(ai_e;GX@frlxrf=@lHhdcZyJ7=
z2EtsdKnN*WA_(WR#b8sG5Qn%Vr-%J&uI?^OijJ*sVSgU9oVh)|?7>=ai<>Hc_S>~N
z>uzap+zL)UfISFf_=@|;2gY!*89Jft$^FH*JcBfuEnkts5g=0e*5+;H<aIlg`5*!b
z03Q3B)s-7opS~Mp;+~Yz2*DBGvqVTW1$_<e?G!>Vi@Fl6+ZpxuR*jD-cQtXbvfm@j
zY4LIWJ-Ye1BbzAcacNbY8gz!yRBUB+yaayj|2Uc2S_y7)jaf;<(s>S-mK+K5S|u!(
z2Ti}nC6wdMIOzpmJWf;Pn5>l%u6WB(1GTiPUz@)0&8<zT!up8Kjm-bi_SXFVzjm&x
zL5oTrhpjcgj?&Bfx1G}3o)3`c?YIa*&tKrp<#a0X0PIB_6IB^#b87}9s9@?Dpoyku
z2Fu7*;9Zv`{{MxyjoV~-$?ff9WvUtMLYV(wmO#h7?q>Ch48Q`wJ^dvTT9DS5-<IFa
zkd~wdYlx<JtmIo(Ye;!Y7VA-RE%ju7qO=ETz-Kk*QH7KvT=T)8FF6P!kB<Mg|2loE
zJGqIpdUSW;&tqsS@H*`{Gk2;@#4%30C~_7pE=JM9a0Y#fk{g$9eeG|#Nv+)?^@EA=
zMwjtszc$`p0@6c%*(HXYHL*EL5IVKbnFb5Geu&{%5!NP$rz@xuAE?;oy?>w>zK!I8
zS~%LI4w@CuyTx&<8Ow!SN*qJ4t0*Gj{o^yW*F3OGU<(F;@t(nXmvBE*mlHVLxE!#H
zP5Oi#VkbvXTUGCWl+g5_h3xJ+uD^F0^I4Zt@Ld9Hj>p^To)LQO&Fv2ykw3JN-iNYw
zZ7pl^<yKBx9sd|&AKI2sPsv#;8Q&`W2IV2ayy0&}3|e?oY5HdHcLr)*-%kG60^Sh@
z$O|dZSjZ~%dl|<j5j<Sk=jyVtaCQ#{liPa{>xAI278Z4rCn5E5*m@|fW8M@E5iR@~
z{sscaRCZ&Y&2AfhlI5HAn82x?z5#}O7ZMTH^x244D5(D8<jB+lB3Y;EnW5Hf=d9OQ
zO?vZrtOI&P8BBV@IkNaN{Wkg6?W%%LvS!`q)-|$}_jcY2fsT+L(k!pV+~R9QQbOX4
zqI6yo#?(PN3v43;{LgKj+tJj&7ml!cqu>roC0g<|$>(TC!(ZE0#RW0XC^mA4Y`77Z
zj;7Slr`aJ`+Icx&n3z#4HX(vhRW71KN|+~c!3outhv63Ym(`9JnGx#uyG9u_cn1Rk
zL+xZz>sZX`mI-yiD4Y5GH7WmJSat&N$DR@bmk~u(uMMlv-wpo4cYaE?3}3CdV!r?;
z_q;UDDH<6>GQBMSPB9hQk)@?|*MCA!`sC9fhz0(zhrkw7$%Ztzf+qZgs_#kSAmCr`
zXuJ;|JKFF>U&6DHR@2F%xh~}U4bm}wh(r`lU}|yQ#ONg&HM0Ry{IpH_;(kvdFV}hf
z+ODDdO^je1EuPuLMzMfO+5vI)<2!IVavC)sQl<I0lIFGA5*ebRp`|oE5m&YgA%f;0
z<EmqcXHWgbtyj&SI$&vQyxJ5Gq#apm){<7z3=E7Z-C5vEe|hJ_N37<vKf)kR>iE*@
z_R>Q^Gj+F4>hlhb9Cg!FYj#<JQ*UTYUlpiC$7^Xq`>}N*GzYxooEO6~%maG^{@I^<
z9^3t;)C-1(KMrzBDK!p*)5Em~3vilzS7fwy0k5}L(Ep?99JuRhqb(e#ahjyDZ5xdm
zyRn@#wy|S0wr$&PoW^c!+s-}TxcB~oGsfQU*k`?S%{iZbuXi6RS4<}4dbNF`spx@S
z4t^WTAES=AS`Y!41s7nSrRK+Hc!W8cwWe-=+wl7BUKQpmqa9|qjNiL5Z_K(!pE-s-
zV=i@i*vEyn6MqU7Slck;$B3!`3*!R(6eZ3eS%e5rm&YGx?$|)6JAnneZ+VfyUfA(q
zC2G}`$U?NmdviUsU`zG!fZ^B2PyNn>Xv$t%5H=LrFA)JRaLNa^Q5q!$)S2`cFbLWu
zYjltbd?)Q)N8X7b#Q#`QmgT1G?t^|@`W)KK5cy|RiQ8@hZPG-)Cb9#tE#^Bleaosy
z^upOv)Np%4b{=jxEj<iK<#L;(AJ@qlrTG%!gl>fZ>Io654A5x`2Qi}U>_ya+_PZAm
z-eYR;|HTs+*6<!Q49PwYig8G|*=hw(I{5JD*$RPn7*+v|IaK8)28ZZL^^Vsf1`uV^
zMo237{L3)71dDUrhkpS-?_xmZr?}qe;>!?v7Kj-KA%=n43Po}@Q^LiIN4piRUL@p(
z1}aoY>>ML{+wC!;jkI8UGHo1R7eQl1aAN(rO9~-C@XukMO>-DtmOHm?FtO3D`{U&j
z84NRp@JLMP2?Ce}bDl2XeE2o|Uv=yBqA7E|&%JXh&X$ns6c9aNEDrxm)d0noRdjBe
zi;azgg@u!qm7SfFqouyi&Dj+}POHxL^#eWSBKu#9gGFr<E7$2|k=^vzeZ_44JHFgJ
z9F}o6aixm^9k0_6JL6Xm1~!SDtAQg_#7OlxLlKBYum(ll?}4yna!N`)>~6aIlcF|<
zVGB?DbY1E-kHGp_kB{4yubb0(dr_RNxT7qXA%u#eMkPNbb`qiCJT1?&IHjQ*I-me@
zQT0|FccO!1pqWUqXOjj3?`<5ox5LbFo@h-3=t``$_u@H&FXii@F*&DQY&Y>U9sLb2
zd0y{JHSGPB8}V}I_|4dZiagP2%9OGN2Mr)VIHp=N5Nn5JE&<E;AMWC@b}xy71C)FL
z&9pSW2k<(TJ=pFwYS~{$=~5`gV7>nAB$%m#7a>ETaVz!^7eLUpgAt)%!CE;P3ouZE
z1W~ak9&?3vTTa0dY7l;$NmErv$u0L@b}C6`9jkpZn`gn{M0P1SBE_rv!G+y=QU?Bq
zPbUC%va|m!Nj0n$**f-v{_vwj$r*xd;B`<{;I<8r8K|~LCFCC!a5)1UALl;PFs1WT
zjN|K9mK&<=WqgcRk7c+n1byWYbUxr$+mux1^}C#%Usp^75h(PU$@Ne~M5SPg^_7|9
zLL#Tft?}1=Ao{NZ9cx(^mKyKEBu5#ltN>mo``dDE1!ErhMScEY^!bRaH`4xqH-23L
z8wR<^FI%drX{3AhQ@a^VBqilG1~B!Wi;Ezq=TmJHl$2&=GmLZtzF&Rtn5*g6ll|zd
zZpV(AN28^(2V*%{6t1SmtRrrp2XE!`_DE>LV(B*xRPgzM#`#1OioCRvZ&w;U_y8N%
z51=V8xU0e~HYbISm>H=X+j?}2PkxI$849Wijxy(+s?2;=zKh{VLCM;O;=d-|;J=H>
z3&Lbh=abWOErLoPg`fuW?ep7VlwR}9O4B6wf2?i&mfb@uW2~f|PQ|`_aIQr8v+!Ua
zjwoSa3V)?_u@81l@@i-cY|gt3fYFtSj+k?uoUi*v!~!w*SPDNj8!BmE$Us<3?#$lK
zX+$AVD?o0F-*?G)YEzDBAC?w~G(i+zgAiJ}F+OuNs(J7YkYERGy6~E=pxe&j+=3GM
z1!X?)#SJp7r{%J0L=M?;I;j|(8^$c89hWvY{~brw&QlqT%!icqI7|uv+O_ogd-EcA
zwWiRpMxmOesZEUK-vmvvIvMvA45_T4@3dB4872SHlzq3~86T{84~)_4Ecos&`Id1Q
z17oM(<Jeut$}}@98yCvZ;3++;q7AoeN+mlOOKH(7|02dPrqf_?KN~E}XqL|UR`2QR
zi@+99{-9nh1y{Eehd&nsu>9sB-nKYBI@rJ5J&wLkH52!wz89(_r-jKk8rc6c)6c{n
z`NN(2OKF+kEM!Yidbo_@TK${c>La_0PUrR3bH8@V6h2>VgX4SgZ|Uh?KKs^oZn7bC
zM6D6=sb0i}LP_&s0XNQMKH~tT;~9`yagK|f6?`bTy16)gECrqns2$^--VnkWc!vvC
zrFy7D|8QSwzy&)|kjsLRaaYV_8TvjysQpNI+V0&&^nfn-_Qqx^8Hx634Mk>;3;{7L
z>mSOn+D7~ZMce*xC-de0X>_jcp}CRrgkfu}51UIJegBya`WqraU_!l&rA6RI9i)fl
z0WEDwwH@56w<6^u0IgQV@j|?iv`;RkSR^s;U?6A-&4?@|B^6=+<GD{<GSLe+Co;JR
zlsqhtD*a>8Gwt?yj8foMlWAFALq9pGtF7}B1~=A}Cqe}O-qeHA03OWG5?_c{Uk-dZ
z6~J&aQ}MiWc44N|BOpgf0t<Z)=5=%4D0_wt!xuJig-`GVPRdKP1QriJkQmTdj481$
z@?~u@XI{ITjnwCkE4VHL*8Z(NI&Xj!#`n{b{zuyf*PMU)b+j9QRs1s;UZ1$Xz#DQ!
zK@TdmfJsX%_(%Jg{;RtXufH9d$gHv6!OTwAhM5;o{<b}(^w@hc3}wx!9KT~kDjN9K
zsq5gA8^Fp0NYXz3UzD-nkuT$SQtDL=;~<x)kRDMdB0|LpN)9*hzLDT26o1aG(e<5=
z#^v}XvuP0TZv}iZ;}t6`mAmEO#To1yw<rNeFbigZvutlKo{Nu2Nstr~K|Jc`qrxeQ
z`2$zbR<u~2&?@}xj2U`A=Rv_|2*+x%wokDakTXFCK~1WyzIk-F7S$9guh%Uz5D7eu
zq{b8s2;#JR6&&&-HJd2xC==&<6GXCXgh)cdhe?)Zv_VmseY08b@NltAZOG&Q#F)0W
zhnj$?tzl8YO#8bs%+{RvHj7(qK$Me_6^DFs_n~JCg527dS<?QMLQOui7Ur>$X^DLT
zOf`v3Tbpx#=}zu|N{+QfIXZ<RrbcQ?1?N3+PkY~PZu;&UCcJ1P)GNnyl@&V;jsG1q
z$>eucH5Rr~DMmvsUSVM$Tl58s^t+}_466t^&|2YWrsJMi8fBvhiKl@w@sVXySLe|k
zVyIihk}7x7;SqVeQ8_W8X(mF7HPA?a#sKx!GpW&)TKd1X=MeFdDlPq{mi@CPVbtp3
z{Ak&?s~eV)_iWqMCK9%l3CZLL2Cz)TuqOj91UZZDV=#x|dXib8;26m<9qo2csmPwL
zy!bDne(Xo;u*1b{nQdjxsGpj8zQURePASB-TJ(BR@0O5VYwJ3hxde?+IhHJdaJ6Q_
zAspS2M*eIT#jaL8B^?Fj@*JTPVKSE5Jb|r%%l656sKUnE)6Vw9g==m|&1P+$35XV3
zWSWa?P(j5McH!Pi;1WsCBB7=$2A4&fIKe!AdbRpxQuBs61_FXWhZ}+%O3GVmheix?
z9I6kE(eea0T`PCJqAJ`MTZ0$~kJz>di6u@x?fZu8!9h+zW20<ySHR;56*NrFP`svU
zhb#cmA8-j8@?WvYAWKICW>j>)$l~J2H<N#dCFPi16_=1V7~&R3LR4Jz-F7%m^$7Sk
zoho~LgoFL<$ksf+h<K9AoYSY>|B6HnXTHf1MM|T&xUspOT+oXpmoOEuZ)Bf`s_4bx
zH<E5oX8cfcZ_)o*X9L@eG9ev-0Q*RZ`d~dEG6^ZFpw|VkICo&JHMXeQtk<{YS2`O@
z$$K{D1cKBtDy9=(7B)R~2^nQwRk84*k%*M2+DkIr3J|dgQ@s5o*L(C<%3FA6+pbp!
zYMxyZvVC*>&+ORH;v@j>!p`}_8v3u9r~tT@-j6w~A3jZujZH3hh*5IaxdZ)7Cc@q)
z^=Atd+QQR!C+iXl*<B#aR49sI9opvwPAnxKvjyvbu(~Co_SQl$Dh>i9J|aetS#-fZ
zf}((l+3wjlR6@LiOIy3@nml5Xipb+(M!iW+4M@4LjED$?TQs2cIs73k+P}(N!${57
zLwEP0Jj&|Y{A_>skO%`Gmk`HK56dQZrEelK^z-<ydJWyC9#FeKoxLJ${VPsP>N&vi
zGpjF-ifK~-<jZux!ENtao+zy}A+RURnrJks^JMVm-g1QFOc^6?6MEalIcG$D^b-Yd
zyji$90pI;|`~(CH0Ivjjq<)c582izLdLZBc_*F%~c63ETR#RJASzTNncsZTRJrd|~
z$WAcOGgMXDOrQF<YjP<N_4IDl9z7)%(cQpvCaPXqSo3*7uiGKkrzSm71#&UvSO}<~
zv==w8pgH~fQ}_?6j+SS9K||Ahkvnzks-UE(9FDbu6F{E4un)!!OpYvXPmb=+3&&9+
ziWghJMJB<RRy+k$nMJW`o#%2PTvt|&?|{->A4q3r;SpdQU;X-~K;2m3CMheum>ysL
zGPJ9kWg~%d@lQ!rslKiKqgQEuM6U-f+1jrb8nU`O`vSzjh&kKT?P#(V1?LTbSg*WO
zmGq#b0iZKAHE*OwA-`ZuXUL|o=!?jU{p|%MqgH;X%gN)fv*!$?ko)dn+;Q388a^O2
zuS95K=d@B&*s!=~4gc`dh|H*nNGK3!Kb#ic+S@s~I6e=2e}RC+iHN7rtXRabsVpt8
z5}aQ1y%>w~BHI!`=B*a|ls<enPi7V|qkn*T22f;r1&7tb-Y5|GCl-E~#T4M`Q+Vqd
z7I`TCTUrWqm6XuNgS4r0u8K^AqLw!VCv|TJOTL8<lL|#EdUn1%F1XJ?U_}>Q_X!|1
zil1dwtDcmMysG1xiTxU{IL%yZfAu8{On-*}7Gy8?WOB)}wG;3x&8vGJC<7@m_nUa$
z05|)M``vK^mI4t967t<<t83Az$!*`u`TnVzdA`#CB)#($11&8-(RUvj1BLotw@6B&
z;#W|Nif1*|@1A>e%_3(2i`@6E(4yfW;3OxYI4&v!F!^Idt>&bDh=qB0SUn=Ck3mF0
zh>3=Oba}XYetdpPgpG#z^&0_t2Q%Y$fP}?DiS}i<Ay1Ei287;OZ6JC-DCzLBLq-&U
ze&vzHeoGpA1GS8MIJZ1nul8kSEYU}cSfFL~M|Yx^NELIG^v$ohZ<H4jQ_(4uA?Kp~
zUI~hylKqd$1e0#a!^5MhQo!v6EjGk3vD(^|cfZ#3+LaZ>fv3g@@AQZiC8Y(#cbWNk
z2&?_%7HS@3#@yJL;36k!Fu1)qXjoFR1aWCLyxKZWcap3OCRuNwc4!6?W`@c7MNwcb
zA6?uyFDt63CpBS8dCciVk-mvx&)+yqQ8d_!eM_$7B=OHh_~LXrUP<tsVw(A;UO)Ef
zQjui<8i4GbWwynCJyTUvy<7&GJ&dolVxu(@Q&F+-x`<i%X+23^I+d}QCPv1!x=FqE
z%kd>WECv_Pn>ro{x!2YfeNWZ~H}G2O^ms%uDAHh&GUGn}W`WXVo2|5<vSuh;3s(c?
z2Vs>xM)>-<<Spz_D#Cd96HhJ6-}dRZJK^{*mLR$M%SSQwwU7q|MJu2=1C8*@p6l}}
zxGW89mdE#Fs2>z%X*#Ewl0kwE``Af>E?PZl!7!r;X-&717NxKS9~RXyQ(EwUpnKHs
z@w|ScA%EJwBD2@`x<6Q71p!&Q-9u)-|GxP=dyc?j%4-)?X=(eyfxu$ud+~soYY{3(
zrKAc0oeVlFk>&%Y-vGnpUlxaZw&vvp^b`X;@2d(s2xur|^7QGY5qycNr?f3g7a$g@
zkg@XL?GRlKZRw>d-J~vlI^*M=#zd1hpPDgkZ4H%9WDcqf>Ax-3pu}m(wxTkK>{T+E
z-7_Dk_v8=TUofTLWg2IaHqe7B(&N25IT4)Z;2$0G`_{Y;8n9i0QqukMp!U;g7`#Oo
zqgzTG&T|&uAU&POpgAv+`BC<c&glkDw5p^mTv|n?AVAn_D?se8<8O=M-p&8quANl|
z4u#raL$e@3ZCzV4_UoygUG5jX8q8*S+YJ_+!3BxQ`<Sl{c&E%C7a&c)fSz(9Bd^<O
zq673yizxzt`lt#4nfg?<6C)lsX4=dFn6a6KzkRY2ngKb13;~<u*K-H<fv_RG8p=$Q
zs~SFAS2K$snPR%k<wHe#?&Bq$Awzl*3+td}^si=`%=fD?8&KTBh0pDLBT(=*p`@d|
zM(4c4<MqO4hda9KAkFm5j{KJTno0PuI?vNs?*#%N(r!LpQx(N9A9&IzkC-Hs5Z66I
zgRwvuXCd1o6L5l+fmB&Rn9TSf_+mT?{{8O|nsdbrG+2qQ^~&#n`jwDm<&@V@Jzwux
zpxabk)0XySL;aiTJSV#w;Xlo8S52MR2psC$4$r42SPHY0dCshk`yef`w+B37m4Nq2
zWoCQ;cG#i4go4+NgQ6lH-H1$7X?xGw4jJFeH@a+7K3nQJ!-`rO?$$B#9$Q&uVY*L@
z_9fI*Y?=HzX7wq2j>rA5fK+36On{xF!r->L*m#&vJ}_?5CGi=3BJ|;6O;0tPS26L!
z1&=eyt}FoEaB3su9Tu4ogsYN&YyjoP^m>4QfVn}eMr1eX0)JOzQPn)~_GFz~a%->a
z@lpR|pbi@IIBB%wT^XY-4>e?e2pEt~rVq?ga!soQ9P-+{ArdNiFkB(SL`gc45x6KG
zAVh1&tJnKXfnc09Yo1qk>jfVd=&<En2aI}k%_cJI9|zG0RLJMlQ&Y0Puf*SV4FKj;
zrNaH>3w}>z`vDAJ{N7upY($ieE~!efnNbwXZ_{%h$jNiBv<S(AKg5Dypr1TFS|L&)
zy^8&3Ijc8t1Q<G__AyZ@rLTBj#LYY#K}$K<v`tjUih9c2-#^uDmN%wd(OAPL&puAo
z8FLv6kGD3~@A*zGD_YcT*44U<4}puW*FJ@(UrvAP7Mphsnu9(KlPj3_2)?^}Qa$2D
zF}-g)0B)<@<t}vb%)y!gp#oaX5hO+snQ{w7?j?cUmDr0ZO)QJ#I}%UC!M51rHMJju
z^xzgqZ04UwjqblDC!AGH*e7%Y^HeI#Gy6hcm~wdAYued?{)6=O&}f(JO8`2qePXa8
zd0=>^j=m?q`&nm`IgH4ApdO3pe=~}Y5VS6yy}{=*%G@8!+7E;47segZPVs&9EWMjk
zOT(*h2zm(P!QX#CsUstXM{nO5Gr*d1OaNjHFDb$>%|W}R#Ryh+MOJ^^2!+)-9$<Q$
zE+3*cWDXT_tgUQ$XWfCL-VYzI$FsmpA{z?<-~4wIiNmrv^GR11{UHmg_rK^m|AaOK
zKO_&|mp1J;`tpogFRhhzE%ZDSiPS2#jAA@pxPwq4xcZs$Dy=S#&Q0=40=5^3`Nv>k
zx|pwml%86gf|3xkAa+^d>y>2AEz3&Ci!D#k>Z!d%@AHx$*sy?FiLI@w?COVtS2wlu
zWv9}%caX}^Qd{2lP*3{vhF(fRZj^DOy<~sw9+v(dl&=%N(Ce6<sqZ;{qPeobHk>KU
z*}SkU*G$1VJwM+?=&~2eh2`y0%{v17G~h9n6Id%ZhSU`A788<$a#w>Nt4F*2obi6!
zgytyAkrnP^xCVHeEcH*HWw_5F14RSf?C$UNy6qoShDRUJZK{T~(m^nB`OGtZmjinL
zc-_vRdStg(fx9|E+`*6QnmT}Ac$O`Em{Y}0ALYRrzmj$=^46`16RH5>;b?(R7_3Ax
z5lh!U;SgS`0`Fh@u2>b~uPtUPJS7pG)GMUW*5W4pHQ)w`Px=;9;??gjy(CfbUh<{w
zIk)H;EBHImb~=B3Abv&&H!1zD&Xmc!*jnN$^Ka-r*vd(r5!Q39+t|Ej-P|-KuBOue
zdVPV9jA`^bbR*on#g8!!_vGm8VEY^o6$5uWhGYv3V&E6z3a5_?v|~0PH`+S|iH9J(
z9IV(O3~;)$|MY4bZJbWI47gj9<HO~fhK=lg=*yERh>-+5@Bf2^b}(gL@`m0<#RY;S
zQ{A2ow_O;Ql)OsOshgX7yi8Z=FaCQ36ofjZy}s64Ms#2NgIKcJL7V&#J@EK+Ecrh)
zw?X?tl+?3RqdMBzzqU|ed40iJWph{^-j;5GLL{C1<f<CJ<_>x)o@FKdWc9^f9(JGe
z#lVt|=|Sk^BS<&fd)?Ev`SoDEppDa(3uNS&@13A)`5z0NdlmI$8)}`d`ck7||HXzb
zH;F#k$5)infC4Bh8OKFCvGcwn?SAgdCuT5bo98iieW4lmCm2u^EyR;gGoTz!k6{Z?
z^EZ|I(`qrE#upQj0g`TE`DzrN9QpJ-N%(>T6=cCv2ZbGQp0V9FP>7gSD5kRc32|^c
z-5y;(t1`Ik?sR)x&vB(6CbZw*TQ1ky_?Yc4H{OM*4Yq;+|8AG~`$u;FfWj>D4*}Qo
zvrEfzUD%Lrj1~Y*8;uFNZ+c4ThbAyJCg5%u2HTWp{>!DIj(+{gZpeWq^_jA!%J!J2
z+e%ctu)_aNM@qM(mO`v)0;}e(moLg!in5lA_X%``yt~r-yw8p}lqa-W>nJF<e%I2F
z@oQUSo@-!PeXJC5PxcYD1D%TMPpYL=H@lHA#|2oxHYYbsBbIb%4$$eU00M)FjKd^s
z4oAwdwFM_AExYtDI^_#!iYLDb=~kCzc|GNy-7mndFl|ya@$eRmAR66WkBnSRPxr&3
zz$o#%UtMiaPhVY)j1)-35cRw7;AA<)J12*;6ZCww)4U^G=LU6Qq&G5SmFg+Q2pE%U
zqX}XKI7Icjl01Bm8yyev1@NO#3JgwnyLXgGvTMoI(_8ijJ+LJwF&11A^_%m$yDr7`
z^*3Gae(%=!UTu#YDSOA}aiup8bPL+Q?;JM|tuK5#rL(k0w|7D{CFHOn0>r<q1t+)1
zp9T4~j_=rn^41`4JHjsc6sWJBz|RHPsJh5mXLbb^pfM8+%?inx0o~k)c6X5z!y07D
z%xXh0zdyD=lWlJ8ZspMg6(prqH~);93w!DvPZqN+g&I78xTKGYv0YxT2luBS<XD&Y
zvyp74_wx;i0Dko84!MRJe;v~(dC$}nDaA)l`28sbbYR*SyIZVHJ9EB`faV%c2yL`a
z3)$VIpIlDD!L%np>jX#PQ!#Lmv%O#+PFE^;yQOXoab!%*sVO-}LR#Vc`Ogc|?Zxgv
z_q5CL_+-LAw18>z{RxjP;rSaMRlfO+^VQH`biu3TSj+j^rt{8OQ8&2rSAH+(daT~C
z$`L8}+J`9Zy07Ft*c`E&M30RsHS%{XeOe6e{ekLF#Vfyoo!J{&-1d-SwTB;w=8Kyg
zr}{4=*iZB9lh5_I;+(er*C5vcXw7NY&}=DtQrfpVMemjt!#yP!EyY1M1;7)3>GAy7
zyS_?VY`>S;{`IY&b0=ZAf$bV2{ufM5P`y7A{yjT(1l*Xp`y{^yTaQ>UU$j2p+tuWG
zFT$m}Efotq?eD`*3A)<@TOcY0h#S-8aQMOO3i)x}kFt#Yh~ty>es#88MOUJY0&0ph
zW&&gMv_nF}(|y{x)cLhhXDHUW`6n{JJdi-0$TwyulxRBT`s3qSG^4?I?QU?NC5n%k
z)<L+oh3)%#-M`#ue-_(C=<l1AY(Y9_ZD45asGq<-%lmZ%NYgI%c9U7E8lE~@yiLeO
zS1`|(e<iy8D0$jXAp+cQcScA0y;`O?)vDy+T&KU5gi)OV7Fjy~ZzibwTH;Fre=LB|
zbPe(|6Z&7U3uePf13yI;rzyN0wjaoJ-6q+s#}HeVP}8YPx3SZ@p05Vu;__fif%3Mw
zm=kSl*td)Rj-{!bNd8v0SITTLL9ctvzlwp>Hs=?C6X%DFkLyJjPcinxkIwkZ-L}x#
zaO^o<AF=o;N1fgt4aN$iv-YqIwVaEv@IIwEp$SoG{>Al`{4fo_EKbYmB;(QUr}_-w
z4iOahw8B1ba*uY9FWha=`Ens`fl4dhRj!*LLe6)&gS1BA;MXVl@zo)S=EcGbl`b*t
zmXe>ap#`Z%1qdbhT1;zDD3T?Ess!uu_(EBRDak^1@iB5V8(s8IgNtpf;@=sPo~ge4
z;uf!}v~FjWKA*kF`;DAy%(Q86GgW1k(k>5f?QXyKF9T#0K)Q3u;dQMA2whG~I%*zJ
ze19(F*x4*M8>(I>XGNW_a}joXU*oHB#1KX4Sd`I7!ZoVxLt?a4*ChpZBxEI}fy{ce
z+zT{)^<k)v|GiOMS?R{y17K}(hWjs#hSPHIEDg28;7j_D@>0n3<?9L38*Yx&+Z4Td
zUd{*0<sQoW-P~hMHX;BH9IS1h?3g0(kE{4mhv3q2-ceTShaL<&DK0~+DU3sLtU6r5
zM`%ZC)vdfUxcI%ZB_BRV&-8jAFPxDrH^jNT{9P(QvDH8Q?$ZfZeM`r6Dd_uteSfr7
zMRy1)Zp61e=ihX4G5~Gc-)x<?PW~-I@{LuL{bPDr62Qgr3<EX_okE-UhCDK8d%x((
zYl==S&qm(-`Kbosj{c-t+nWr)Ybj04YDj$R6?I}(yAl+jjej0(@JBB;wmgMR=YW-e
z_@0|p1AOc|m6fng6=p5e0F>7QR#)e@phghSnf_JfxD>P<YwD>-^P!Ox7$@ie_FAZ4
zQY;-dl*Nw5*T7|MP-A40)o(7FvW;MZXqQ6n;ae+iu|r$CbFcdsP<kfb=cU1%_=o4k
z)3(6EJ_pi73F9M5&)fC+T&P>MOOB<0_t{>XjE@^(bT`CN-!RejUbG___u7g-qdy;t
zI1!TQKfm&jhjogy(zJv>v`c<AP{$ldNJuU&F5f5u5P+|k#OTFSKB$drX(+#3h37)c
zP(F_}<Hi-zp=TR#VGgoERe#NAe7?`$bIJ9)-nVzBfH8i$d*{VZmdh<#1PHra&2^0~
zM!Lx<HHma9#`eNuSrbv^v4K=2ETU{@sdt=$@1jETSSBVni+fm@RsV(h*|{jz)@4-_
zceQ<yfWy=Uj$ym5V8(+9t0Ti~K`Odt|BCro-}Oa-3G*7H?x*9|Lo!fT4UbzrG-ccS
zBar~?0Nfbg_{k<e9!PQA=O>YPMB&HwkPKvxR>LZUUbPB0()B`Bf$2<&XT`=Sr9mHd
z{L}pL=F~^2kZdblM^c-o8oxCxH1qjTYrjJUOr}!#b)D&eU<G5YEqeI5*H@LDz89UJ
z&p9_Abd!7<V*!O^FG9!*DK+vDcg*TopMvMQZU*zJ&6q{O=^I{`HnWsnEomk{DSduQ
zGLaX?3Vk6L5<$_`ij@~aJ(f&iBqp8}M*RV8O@|5P;C8?6baClsE<*a64Hwb9ewzXV
zNU}bzIgYt?D^xEm8xIvigdKks6?-<f{q~Km<!XL5B<*iX6N9JEMdhMJf<#hbgbzWK
zs@~kAEOpL?@6*99RqVjW-O)@8dF<!vl72j-*ZX+K>LO$Mh6c1@U&o=X&+TJQHwEAG
z>_2p{A_}Z`^-UJio=snmH-RYgYTeHx;9wXuIKAH?S9#?GvDau}(lF;+CZ%U0HWYyR
z-{dFL@J=npD3W;brcP(r>$%woU(htVp8_j_qa|Q|QlV@vXO@Q}ES~fDphv%Lc6;6Z
zm}#17@9rmh*W;dgCXfV`^I!HxBhej1*5G2o)*DYf`vKFnj1|R25yoJV8LhiOWm|eu
zOku9?LAkE>hRy<~o%rA7l^sN>TD-8Y$6l0b;~K!-gEHoTeNpJtg&u#m)qB2$afrDZ
z=vLQox))XR;ZMG%F6jNae0*y^1H6WT(l&x0B}7seq1$VRIX<3`2_PeLmY~zka^Cx?
z6n~uT4qWj#GPAuC9u=3NpCSO(fQu{IK`d^-IX^-}%vj?1wwA0^RKHBjC9@a!)PFh%
z$#SN{U%)n&aS*5A`feN@e4A1&@EK#?{l;heNo6(k{pDmbi7nKvtI2XZafae}=KBMT
zE-IMwQYsXF?36HBaYRqSF>1V?fXzPNVf4AL)u6ZEbn#H{0D;uAh7=H*Wc7jUv4)ic
z++6>Iv}4Fp)6c|BW6AfJS(T{vex523{5bDh&rqNI%X7wy_j<U*<a6~A-v8=m94NkH
zT*dh5<6hK=N?gO~SG?vVyRb$U@BYh{uhIxtQ><|3Xz2I7n5j6Lc}P(r`?yxy`;)5U
z9+CI^)$ni+zgK++>j7Y<R&A)i{r<5Vs12&2Sno7*pxaUxS*vn};Wsz`6o%9&hxZHS
zL&;=?`EK(nWMQ$}>GIm#%#=0vQ$H?c>{09$$3yxzRkMxYEslP8GLh!*_DTj>MnX0x
z=+cle0wA0FLl}0re|UHqrsi#xJotTUyzDWWz}M|FyUA$?lL~-Ywiu<taW;6v87h3I
zxiPcXmmh2Gr`n~Km>-8CnSG06Kq%&n9FhBd=WtRj0?p-TbB>D2RoNbFazizx+X;fB
zzt<3ce>pB0R#E%G;`>R^;}&_s9Mzw^CVHc|(e3W|iMjK5%@J&21YkK$9O_yC)2WET
zgsycMIj%!u#emfz7DA4LKqUV{GLlRbl7wP(_YkVGf|C4tWKu`%qvR|5mE`90-rQ78
z_DYMzzcV-v=v<AK%ke6^>{IuT?AMRe*Bo!>{kK;TXp+zKZe+VZR4A!{pJG`F_TCCr
z-Rf}fWcwV{tO56Xz^|ShYrEklIr2^NzZN#if>zsEAazVc(>th!PF~*Cc<`mSGA+BV
zEsB}8s?~QT9*JB{AW#vCUePE)wfft0<R`$D8oES4*in<$;`Kb=XJdsmmC0PMfr)is
zYFD{&>uY|p{US>ohBN20)N(t7V>X5dy>2ha<WDwPq3QmkI4O$a_$)f>lK5n6`dayr
zQRh7j(DCIG<gT~!Sp`*UP)+JK-Cu>eb*tLHC7Ba^P62Nsb+F`tT@Kgd;i7jL1U#N^
z1B*q>xEGb71y73M*Y_x20=5&Wl}dY3Dzbv}T(MNBBZrVA?fvQtnlq($oHJ$<lOUk`
zuK$tYYNyz)bstW99WwejV;QJW*Mc2>m_siCpAon6POZZ+fcw(~)*}cw9X_(stch+J
z(SSzwt}8`48!YbTR@E@X#g@E-x8fNz;2*E-qdLlr4+0!IG&Vr&kXyw}KV|+7v4~jN
z(CBu&UvaSm{nF#rC`z^O+w)S6&UYVsZLihJ@P?GGro+7?g`QW{zH2T)kE<5s=uhBv
zjtY$D`Jpchl%q;Obe&5k_+rHG0>bgJSvKTDg@nO~@l>{QhyOss!^1<zIXm@lwVN&>
znAVSj53bWYPF-n({V`;YVKUcelLuCbN4JP)#yPV4Yi|oZO-E6{_h7ck?tIzp>^*|(
z;%6+k5FB9x->3eMTl>Ga3a|Y&M6rPS#fg}s5oae$M_V;J8w>CB+=i?~+uxQ-FY!4Z
z$2|_dqCb(`W6Nq=2&6E%L-?VdaVjedYvc19WX#g(syge_?BF3%W>bS|eD+)doo=^(
zyk1+X0AoT?J;C8?>;*s}|D}UVQPbi5rU|d*Vg=X(4GXYKQ#c7!56l-RzW}?<u&2mB
zZ-$wm$(Vi`&v~T7y~MrghaVEJ$4h8<(RCduc1_W|%(@24N9mCk9K_&p@PZs0CcSox
z>D2Zdmf;IRg}YzP)A4mew9nG!E#FpUcCYE<I~NLSkMqmZMN6_MTD;g~YDq#--oX~(
zHpkg7{Ds;OwL{T@5$3aPB7jy}QZ6!_m`?5|b?TcKnuV5tLi|2vG9r<ZuC`!vnTx~g
zgQ&UP+AFw;k^=s6ox{SAZm7I|_w9bn`1PgRwtG5^um4ELddBKz>Qb|3f3uH-ynd^L
z^;|&lAMU(kFgZ3VqSArLvI<5fnvu97$|3P!7%;5gZ4Ol8h!1V{4?y4gDv{(`>|j+#
zr8@p*OL}p5pn4d20f+^d$YW#0dHMb3WB-)PZc{7_{Jc#tt^!hU_;N4qKSFWLCh&-w
zj)yhX?Pl_Mpjx9)28&$@8vdK=qEC9WA3Hi7ONv<&S*6*$qB`d6W#R7eAbt3C&J37U
zC5Sjc@#}deW^#T&8lOsTbYzl&zAC5rX+nUiW*vCD>dUe^R3Kc53f(?9+<Y70`w+E%
zTaS)Dq$$7D<8!mVOnxYGqDmO__L`b$Q2)-;)m7ETKC@&LTis-y%3f1$@h~55d@7!}
zHyXdIBr!iBRV~Z(x&N$d^~(qb(v?y1`1YD?Mi^V8MJ*C2@L6nden3tr+*h6Fn#y;A
z#;V!K&60#EGQLXR6ucx)2O`1vFfw^%{$!k6VaZX2XR-$@(0Izn{n$3XE23agbWlG*
zC90zyngxmRQUs&$fkVAQs5Gn%&2HEB5@tA9=EN!Rh*22BfkG$sHpA8Y?k`7nBM0_7
z>*(f8K6k)z%BFxv^)8D~k6X%u$FOWRLPN`5ICqcN;Qa4BunLlpXz=*VayX|Y`Yb^o
z(#Wk6sofu6`$~ze3@b~(jc3Q1wP&Umxmb(mq0CGk^Y5IYYOBNB>x}r>AiiXfK+$zn
z+U|@uKUBej;UX_&GJO+Doa#Mpm)_0(X|T_w@8b%b-n2b#tasf{6tNAfA5uv?b8BSa
z9mh{C)JPN;@Zm;m@xl&uCBcYp7=d5%cuay=8!k=rN!K){vqrt5Mi$bXft&;?7%wt1
zxQBM`kF!{x-Y;J?ahQDUk6+HGx}ju|UEUU|@uw=ju5n91WSBZpEu<v))%hIGpx6Ok
z`v!oQm5XO>RehQE53T1Xk8>}s8k-EZvb=J|<ap|@VnI+(?ykw@TZ1xbH@WH2b>DuP
zIA;u`QV|_U#E{qTtiUy-bhq7(c$rW3EA~SESYVlQ>TJxUS+ptL^-m&rW0WYKvZ-c}
zMt|l>!!}T@0xq%qT6d~{#7~0D5ZenkJO%+#WdHs-*5TiTifC)V-IdXw>++ZVZmg!P
zFU>_QYpnGh`h7lJ<MTY&HTB*9VWkKp0B(G~bmU}@aU^Z&tbM+0lSLOk=k$C$+&-H0
z=nDWRBYYWZOqN;P<@_!AhGVB+{(SGWxGt2|EbNFY$f52sx!Z%b@)w6xf!#4KAoBxE
zU^$80cAi)I!7j47IN3P1p(OK8BO#~_)-_oP+y4yIqBY2zQch~kcHd{c4?Ox&e<sFU
z;<#$Dq#fdN>Ck`ni~b<rXphdM%2;C1=e9Dkkt!x1m|OJAZp^msXX}5(FYby9LTk;{
ztn+h#;^pP`yS1tP3d4YJ5*0*9vyTA;XH(Z&EO&T5&(UhWR->y|`-87NVCwh^@_+0v
z>HEHX;L-op?QuC&{$=_hNhPT$aW{u|n8#G-I8!NQY$Y99bdVfF8PR5TyAdU$J)C4N
z1(7upn69VDV>X*vwkx5Id003ETAVy9-+0Od71G?t$>do5I#x_3F<XotNhAh@J$o+b
zaf$>_Af5Kb_t+31^FdBEtKlh)TF3R`bFU2nc}8~pnpRkDbl)T0nQT(*2(RTHO04y&
z(-c8|&{X@8m9f}eXK=9Rz%j_ovumB#@5KL@*zzsXLd+0AZ_Q2fh7$BR5!tXer?M9;
zWkYA8bGqv|P`ipbg3K+JNdxTri`vMBCN%185982FCcW*!u^HV->hwoNk;_g9g3DNV
zDPCeWeZB2(-)7u68FTo2j6~%$yXnLEvj$a>7}&6?HJE+8Z7k&D!z>pu)aRqCs;I}L
zWKe_q=8Ngd;TuY`bD+VJ$=uf_m|ww=@M4og`;MUs=g%hNuZ1@JnSoKrCB0b5bk&~%
zt~Z%te;2LSnqRjZuycPP;^r(0deFAy0KNfLvLG`B1pdcF1=i%&$a=_v&21jET6PK{
z);3iw%TSn?kb*%!A0DcqP)+z+$_Qs}6*7a6_Zf{<QBy$%Ui`!t9bE$dQnW>lZ>80W
z9axue^$e>1H7KccMS#hp!Y|!Mv)f6ZP=!qBa#+G*uq~zddV#063tge~%?`I5!(`vr
zi;JS&jzxxSPBRm+8Rpz+r0{5k2^vDBCGIk~?qoWd<X&<)Ggn(-mS-ZcbvS57`G@93
zOtVCPFJ^Y!qdSzg$jV9~xTUqV@jqiLn0+4E-@(aWA`^=JT>$gZCrp?kxQ1F6P`j#p
z>^GHp4!N|ZttbvKYJT67v**8idXRe@Vn&V+K}|J=*qQW$=NU^(v0&~~%`Gb_OzCVc
ze^M>$&xLJu4a|#|#I_cUVX_#Vy!^S4qW#Bk-@m{K#1{yCenrB7Jw9Nbo|&GXU7T5%
z<6<6|2D<^z?Eqi}l;YSMa*b5x=6NDOG(U%}0obLB+O{SA&+EI<f=i-!8{g0GTn@Q@
z6@m?J?*Yl__-M6*;=$0*5CgQ7EDUr#g@vu9%z}bUTvU9@$=T`g<Z@x*@D$-i6_6vd
z#5R-=Dz#twCSG~KIAJY@spA5BMUU`M5r4FkA48okrUMSyhkqpIr4C&Zs3vr3t<Q@I
zyif=!NIsHch^tC6@C1-UQlb4ejM1ESPX<xsAxxR>Ix>bAqkB8(F%G&7-%>h!^n-R^
z>8zBmZ^8@&nebywRGG!2cg}ZrcV!Tl^9iga#Y+kbCen&|Tj)alV(IDq*a=kfPBXBr
z(cm??p@B!bs=r2`;W`d}m81&q>L{bPF|98Y2^9_QC%opcSX@l{@x*`DR6Nyeg{3@9
zHHDF@+C3q0;IGK=j!sX0DF+*0n8Z3mO1oV6Zd^UA{?ka;gfJ78>{fM8)QBghNP*F4
zS&ZCmHc5i$-39x?-Vkg1oP@`DxrSi(Zd$s(E&(ixIF9!b&iF@!qVyCgZ>M~B|Lh9n
z<EpaA;Ia|%QMK?jt0!k2$=EMhPCEUED_|#G{(>p5*GEdj+=j4@Z-Y7-F)LpiCKSDM
zdU#2Mij9pCgrRZdp<&OkWefEce&CCrPJV*9fr-w~a(cd++Qv?JXgPV^_cvHyOk7@j
z@^~OP|Bg#Pp8Mh8o{q~Fc-hLh^nKZHVPZVfU)4$>fL=jw<|OxFht?z8gSeC6JpWoS
z978)5+A=i}g70iBD&V-Jh^Dj^qY4p>UN58*y&oe*Scq??rsz$2%%^w-`JPCRSLqcY
zF0hGr%3Xevj^CsOHAG>Yu8k$s^_&&x#|5-GAaHkfzZ2>p%a_<nxRcUKf|yfxThUQ;
zSWTQu>X>FF@%<N<#{uAM`DC$JO`$TbnMjIU^$rMO(&IZPb<vy$;-2Z;a>EEcFi<ei
zp+O-b!Tpw=2H%Dy_0=PEYD8h`;fnlE2Cnz&WMrk~rQ~F$VogEffo9Y0Q_`#8Ex@x{
z!#~$QRoZxSegBvIwCnvjW%Ik|%W78x-%2{X@tPlBOo!|UK!Q8IBA`J^S7wzxr=y+8
zh-}ZYlsk51GOV<X7tMn0g-xx1RA_1B7)Vk;5vlqT`eRT!VdOH2<nNgH#er=#t|@P!
zvS*;wEW%tywO;O6rn(-N@5)=VDsakm$!mXh7wT-Ci0^UtjDnFOsZ>Ud;=6OzorR~?
z`Mt}*M5-AB-_QE{-SgRI%b_y`A0Ft7mP8>{+c$9}?2#+O{DlAXJP@~MbC1F8NP`bo
zurDPz_7h%oV3gSUD<0YJ--)Wt^x|7`=BaY#)W|h<gC|CrvR@sFf2Vo5YXSdS|3x<K
zdk}?6eG&CO0<};CLdlcZqnMkKOzGGXYADbXVC~)?Ln2yvikqpGmM7@`WV0PR3;w&h
zx#6<)grha~da_D1eauV&VJp!%W&@@us?uWgkyw|zf&1l~ma*<V2~+OtMqlie+}t!g
zw2dhZ+>ZUq>CU0KW^jPcO&`!lvNuikmAxEAktD;n3kTp`AOrcLwgrf?SiILNlU8GN
zyWKuFy1a4{43J5mUmwAXZ+T?3GWi|B;0eu{;(I^;qC8fnvg(`|i_^pzPP<ThG{5#W
z$eG#=H8SRKRO?`#+Z3#eIi({#{4Bp=b}*uEwmzqibxqXi-+|)35dwtMAQ#b3hb10;
z5_21ovWJt#>B*;N_tI*Zbng$6bwMAayxX0n8vPE}^Apw7tZw&hQS4lalOZIl6sGKd
zIv^1Q6OR_dUWDQO7uG{C522K9R2|e;%0C@52uc;+^5fK-{s+@2`FtiF4FRmkOI;~$
zQSQGh=jTo2$0@s7ZGqX-iWWj%w+s4us7zjulXtem?@w2M2LB!^5D7NA9UdTAZFqYP
zX9o_;XLSTQErow9_QoQxPo<~JD$5V5pr-4vd#=nb*6pXU_(q51rrjgnpf+#|kU8Ob
znTNrNlKvQ(hJ+TF*6T&o!+@;Q)e^_lP^ctI$UEV(4}*uhi3DV&67lh~Vzae>TNgFx
z*=g6Atq--dfYU8@ToQuTi@pAkuaBTIT=Pf14%dt2PQ&xx6g5-X(f8$v%ZfzmV}q6a
zSJ2LzeAsA#D?IJ(Uc%+*kx*{r3g$d4yS;`#J|{aT5x@B#Z%S-%7|GIFWF8vN-O%9`
zkL!2)dk>wjHUqZBLCk!c{~9iP^L#p*yBE40kN@yU9=CZ6Uj*7Qq-mvK5gnBdM}mtc
z%IO3aK{q8}DcDg@B}2iDbMOX^KXWG{h>pJR#pptz`|#{Kph|r;LYF*)`|p(|((Y{w
zC~R|A)>hUdOUn4$`;5-+S1eZRc9>11gBFH0T1}3Fz>TS0XI-PsHi%oRHN+KJZn&eP
zqG3^srrP2zY{$pq22z#(!9J5RV1%MSIm;jIZ!_9ExH{K1G_xP>@WS6+FkE8R`+vLZ
zWGB&saH>L@sH$P|DzF*Nhpw5*;PcuD7_5ak<@B37Y?imN&HQqME|XlP*W>=O{JXN%
zfXfL~&cLRuTv~&7wWVOXr9E0<p~vhZ>me_aT4IAW0z@A3Ly)4Av&@i_A>IWNV~C43
zE*O0259ojsDKOeeXxHsE9P;W`0Lrsvk1xFg$&AtB=NfK)kI_mjE^x6!^l#rC&y~n!
zaJrn#(<oCaq|eW;@jHm@rnaYlH5Gg2u?XJ*m;oj%zS7uV<su^Y5G$9zF(EK9go3r0
z522gXJb+}T6OZ<OB?$qRm7{?0U13zOE=oJB{M~mlbd&l>_R@NC%EA%Hh%qa1E>4AC
zxJ!zH<;}(Au6N7V49e!?*@9Ltg>c3aODc(KDoat|ms7SlZ281;Xtd7GTiPO&1Gi7W
z4zs~q;meKTtfc-SqX8CbK79WU@ejG^2pH;YvR|{p=-{=iLL(NtBiD4U9VVdvDBol|
z`|?v_30RXD!#E=;5yy0TJ}R!B(pG18=(?Mm?Ez3{v5fTZh=|`ep{AyOK@r&4_lEh(
zIY|dzA`VA9A$t<^=Pbe{<w`lTS^`V}``(F(f*bVSMLDV~QqUTKk?kU-nFYyJTv0r#
zLb`1BoS{eXpeY5v8m$0b%P;r+kr#)1Nwri^sue_^S8cjqF!iV}OuYhAK;e>ajlS0n
z!R2|85o#{<a26x9%+yZj=Og9Om|7u2JQ~jJLdQFxBes?3eNyy&6VI_WG)Dn?wnmn1
zxhyauqm>{!QNr9p>Q4t32sHgf$w&*tSn3?E!+(~sv#_v`koeQ{z~W$tDK#;wecQd5
z>F)9(beWl99s{@4P@tkSKGF!Jq&7RyaYG)<DON*<po(T|Em)SnL|*zDHqikaYfZ@~
z1G6OH;PVshz16|}9p(xCU9<o&LZ7R@5vqszu|;}ZAu}mPh}xCF6a@WqfmiI@KQ9E?
zm!-Zw1P5JQeXk^Nk%KkW#ZQ-d4ZhzAjm$e!_DPSd<gK8*SQv6aQ7I-IURiX?DL^GK
zFI?Chm7&V!oFKDQ5*#(a>1~tw9BwyzFKQ6U7M6c)Z`?2sUs)qe%Zm_jo#`9v4FN|9
z=n7vnC{JE$so}3Tn1y}-gYqve5Ui<O=tw}stS^H89q+`A_NNg;h-(b5#BVfs71AQI
z?@*Z7;Phf&G@v@{C>6U-1JN;P=BpuzJv@J2h@`|uDE9KJ5Sg##T<LEJ&3tMN>e@r<
zh_t@e$V#KP&WB)S@oy!7IJ}Jus_2*D(BHzeNZ4Sw_oeBGiv<Xyf4}UsQFX|Q$z>H%
zG-l=585l^Gy+=hxB7$~jrN^hIr^5Hj%7Ghs^M?W``4_X?W`Ca)QLMVEvF$X57ek&1
zWO%1ocHH@5h#jOEcXMPx9BL60${>1C7^8F{gyU`z6l|v^W?@V~gb&vIEH4+Er2mg6
zn}}t(k(LIZg^sS65@)YB&IBu)3tmsqPMy;#inYeFP7=<!MV}lFZLv4MxfQ(z3Pau+
zwlX|x)!oII-YK>8eL&2;6QW4eHqR`zpt;j|U)S<lBz{<5i)og%Z;D-ju-<&s8NVnw
zNWvovNx?YSHz@{6Vu=y{O(MnmLm7}W8(70(JbJ%R^)6!^QC&;j-J`(6MMOp;!Y)GE
zDY#K-gG`3^B+@lBcq<5G3Ag1#G1c@q$(Tgn=&5KzS&0bQAcy^0*V2=rL0W3{gu)kB
z7Q(TmOAq+UkG;^_^3ioj?T=`dDdi<P4cL~)69a)yt2e+7vEv)Q`a!Q=R-aqyB)A@2
zbd_yrMG_7L2KiEkf0TrMW@Vx44mVm=%ARSF)ON2i(oY!z@XjtTnzEx*gQ&n5<GyRP
zUOwp!;k*_mY7v;HIQstniG_dJ%6u0UX4XuRlH)XXTj+Y~uyh6$u+Zt>--Ul`qo=Ky
zP!wn#G|dB;MdF^T$DTb*3t(W#Gg3c<Rc<YF-rZ;UOtrSa)9Guo^(XWkr`#L`ngk;@
zEmVlU9UX`1q>WfF7#4<bT43Ou@!?GxZ_ReZ6)_Pku)2vO=#27MT$5Lj!fuR%u|Ws9
zDn6w4mO{n%J%PDo_<!Zz*ccr|f`=!ZzJGLf>rV&3m!8ydDhmxoSeu^4<J*4+jFW@7
zM|k;sEam2{EL4WI+&m?X#dYPxsvrYXUKD)K^gy4k)bPnDO5g_nmh_rluj9i{XCJDp
zzG%k+riIIz+*saHX&ehWx~|_p>A;H3G+7n2qoq~EwZ?Sg!s}?P%mD}9dUvKYwju{|
zsd9k4lnd6a{zlPu&o>rio(F4nMjJF7S&;NLv^9l{hV@2)y{d2NmnAgPwfQpMG=bkK
zLU9bOI>=K-EOJKtiR5vYP>}-sa`u)s(}Qd4gIKeUoKH$D{c}r<8aX>ZQ&8`h9M%Wn
zMS`$$>gMxliDe5|_yW+TNm(|a62bY9I{;|f6vB8>{IdT-myn^kg&x37tenMo44EcN
zQaTco-f1xLz?CwQ#SepK%Y6k#ww=A&?W2bWf`VMaNw5ikOD8@Yyiik_Tps-?;Z*ZW
zW`+GlPKQZzn$G7LJWz<Z3uS@Nuz#C1E!R8bp!mz(g#F9+aFI@38`;LBGzd!MARxX0
zRJ|+8`~dwBkld)QzSwulbOigEgJ^Yc@147k@&|!;NU_9tPeQuf^8neNt>2NX-^IM3
zt-C3Zn4G1i-Qy#jKw{Iw>A14^<Snt3n8<M}%2{Q*s;F#CQ5V5{Gb~7!yZ!42Au{HR
zti%`&UFcrysVKyc@mK}7`P?ZKIe-a&?%uVIbC#IqlDKkVchQ+bn~j&2m7kuR8+GJz
z%N*ILy~a}L#KdBEi<C+9-J!4VGs@-oHN|&?5Bog$7esbGkh7GS*o4S%mNNq)qg*3L
z$rTj}>yEY8ibr!d<U34`*+?)EZvN{w_-SM42$c|Z6#BE#=jBdmro|Nj11Q;xq8?BD
z9nhIce{a|gcf*rCAgL^eq~)abrGI*Hago%_WIyP|w#_2mUeLcPyg9QH8W1JA!P~*e
zYUyAzf!8=5Ohqp*_3MPnL(3!jhQBm-7ILvhR4<q`HKX{LnREhEej=jH5>_a~Xs`KD
zqN?#(Iv{JI;@43XH`L464M+%%Pp|oiCPbW+Fs*un-Xa!?ohD~jH#N;X76*+gtP!i|
z+D1uDd0v+JFK;so0%k>UEC%lEF*qR!F@=a9*6xCkr|ii>VV@vlnV&X2Ck<$aGBVev
zqYw;BE*!owut7)8StUgI4H{IY6~!DIlo>Qse$`3FJv7Hf%@5`ZH2Nnya_tR&05}pj
z5Y}sxe{z5EG*I+ZK3mk0=#99~{Y#CPlXv->9|b$P`F?%T)^+YQluS;Os7a07WvB0g
zE|F8Z*Eu#6N1+)W=Uc5dbtJh(jkxeHL`F*&J(N@k>Z?%;2(H`eK99jjnC(|n%*;Tk
zXi!$s+w&LwJ9aN0Ku-QiSlEc~DV=mwWApxUU*OJBIg6|^$MHV_cnXL0wSy#^>U@xs
zp+Yr5<FIXrff~#7<kROQ(RcF>iipiDA3NiXxAvXBacVb;7>Xf>Hvt?G1uH-qMa96z
zJt!tEzmm$$x>y^Mpo_w3*)Cz&Rm#0a48&2qK-)IV&&|FgMKx|s67hc*X2hYuiUMIJ
zoBM}&)eYJNA-*2YHm3TTf1Kd(|BK15dMMS=H(ub4wG|kJR?)`)4~&QvM@57}6;)Bt
zky9<f*E2XgT9Q#Ta{SCC>$bnUb>-X%Rk;Zv0$$sjUG-GNDl|4NPnHB~#m?3+jo{`T
zK%Gi?$5BQKNH$6QZDM~6V(h>(1hU1Dd0HBF;##H3c;UXTsH4N|RsODL5S#!WE(rGT
zU+q7UA=4lKD87nXUXT*IBh3kke!;IWSsW4;5fv?l<|2IsRF!z8KX)I06vY<BNYZl4
zMmA2J+qQ-q?e=!C2JjwYgz;7(*UDf`BWqU}ERN@>EO;bjk}Q8A(qBN*Zv_P;9F2k|
zn*q${!x*-k<Ff@sF5wJ~>6~!VLL|A9v~k!);N|Nh^z=aO)NXR%{PhhlQfywr;7mVq
z(qJYUDkwJAmKNq%TY(`IsE&ZzdC3T-TEesl<9JIZe_mNPws~4(O>tI2gs+PwK(z2_
zpvfG>kK{G61^s`fFlf!etRo2vLqv(`xusdqOEuFTBnSF=Lz|i;n_5iQl-xsl9lNp5
zsVy=gd3Er-4Y@Nsu{(;BAupn2`xi2EFk_f}9<0>oRVI>bX=!Cm3taY`c?OG9a?3|e
zY#mpTE%66+LKgSs4a<@N9ZfupsO01omJH9&AQyws*~)*2)?bDW=CJqhnhe!IjYe9G
z2Tf=f?+`JF(AE#n;+lm0m9>p+T!g_e-XXxEytWRfCq>En6@fPlG2nn&vx;pb3ezS@
z0_Ao9dS(xyFK~Cnfc^Vd`%h$8yTrQp&;fl({xbmk9YAt|)ilmGmgc6;0VH;7n7-uA
zg)LRtA}@bSwO)gDt=vQ6GRsESRt_)D%T7;DV1^O`z(IzSHh|q2#Ni8QaS)g^9AG=a
zPj?5*PI2|vDWgj=#J&!OL+NA>RNf_#fkG$*sIx&mPG>*?YR9yrO$~K4R62mMShEp^
zXZ0Y8glHm66qnp%V9UR(fg-f4J$AYV^_6$$wDf;TIOIqkMVOHU)UB3{UtHd(>C3hr
z+tSujkrHgL)oZAsqaOtFty9L2sTp3Fot7vOM}>z3k^oTPMA4#HCs77t1()6=&_<##
z6H{{wD;ryT2Vp3Q3acBZp-gtLtBDFHe<w;FmKFz+cqFf6#>RM=F$U%?*sJ*|b|swY
z);E7LgOo*-n30G0lYjeS|CtPVzc@5YKlu?+2I*86FMo2^y4qtDZ{{6eGI7b)Q+G~n
zS=5*>66p4#(=B6Xtz0~3N>g2VfecU=wBbez?8Ha3RIaQd@AI;~1>7{I<XUm|jY`g~
z95-#@8W2~D^t4cWWuTUs6Y6sV%Q)0{$mV}Rz!#r=C<0X2*iBa)zrARIqA%SWI{Fa*
z$7K|ZK>W$Sd#(Re76jHgEW(<Mn_pNg)nO)~GGi*{2c3GUTl>Z#pY^6y3ulb4EzOoH
z2%2yeA&whLO_6Jz`y(Z8+)NN8$t)W)b>Z3_D`t$&m-t$%zcN_Y+C3ybt87d~Zi;^>
zKwzy8+vXW8&Hu?5t(U2jFI11l%>e6x<@@Uzo_IG$h>Mg5$fw|#h&0nE(RPW>8@F)F
zsk`rQTijG2CMRCi+#^ahV#54&%NMjZkEw!!49Tc$O>~sG2)=jkE*#IJHfa0Tn#5aB
zcB^ihvABKV<hr8da8D~uW7@dQMTLJcp_7H4G6PYuc#RnCw<{^21fRhp^IAJ{=|N*h
zpUC9=s>W8Zm$l}93d1*^7RgnI$PrhVP{CA+hcl->s$mzDQqj`3<;1q7lPXh#9khBW
zo45o^^J^zBoHK1=!^rZ&>{JPYb;&+qm?z%x^1&BMn2gkO7S@2j5co$)v&(<#CbTYY
zolsi>+NS(yYg%?P7In<@ROBxOUyRd%?hU{cVQLnx`7Wjmd8%s-br_~rRaH?jwD%^C
zuWmy1^q02ie+I*-9w3iEg)oFYlN_9$huRJ5`--ZCS9Eq&%c2d7C)MPLg_i2Qc;xGp
z=0*-_l@J}`>*io#sKp)dKKOs2cXT|X(Q<GE0V!93n_om+dVcw+@za|dYKG^_q%puv
z94(HK#Knt{*3m>)9wM+9-U7PnCdS4_26{X*Lr3;%^1Q=98&rPD-=R^*5Pdsg1abU?
zIbbh!;a|349xn$xu=VBLL#3Sn*>2?IFUhWGoYhpB6Bpo2@_3TYxsiW#u5?sw^@y_K
zf?-)1Fx-v7AX6ZwksNWnnljfuXERwsp_iv8kR$qrilvza<+bDL>#8eCi}JEF(vqd|
zSR)3mQX6wa9W@0`9xiKl1@S2WRsbYjPaA)LVjKWn(7pqP0LQrnOja1+%$2T<S9o$k
z&BS?NFL&i%vSH;;Ur&Eem)xCNbPN)vQ+aAYQzKs5X^^&si@zu(zp@}z9N=oHGnkDm
z6%`ib=K-D?>4^NjpyUFBe?tcJef70Yj%P|>;h`|zCiHao35twO%E&DqF`}}ns=B(S
zhCCaT1O@=uR%QUQViyL~>CEk1g+YP-em>q_o*r(<wgyw49cX_!5%tn<u&Oo)n7BLh
z5It>cPvZEI&GW%t^3uOt!&<$jmKHR<6{b_4Nsh@^7+s1uSV$5qwE=RBMp{~WMusdi
zGb>FR8wDe(h}_ap9@w`R%o28VJYSqiN|1;mLj(PU?w|x6o0tOAr6Wd;9yf6^NfF9&
z05=DPBV)>I#PEMMm~Qxn#foA2i0qvJZyb#@F_4oVy~2+&@>yvB$2KsypQfcpSaLya
z%L1^Mz4kBFusg?1mKn;WNm%R+aZ^8TdXikx_5wMe9SW;>c15vqiOH!MnK>CL2{DnO
z0X`layJ$c7Hk{*m&%w}zXK+j^fUhOdK#2@R)QISqc%XmEs%x4$XUXdMEp<gHk=}OX
zBGSbuPCh8lPhw}Jr6eWBOJYP6bpylD3MvcK@2?Hua3#^fJ_0*4ef7b;)h*n^Ko)cn
z*vnq~mulFLA20-kNleU5c<5^9$qJ@aCFimwTLD=e{e#0ranjVxVOi<P2{F;(!Bi#&
z3Inh9p$UJq-0|$mD9g`EiHi>L^YHTaMMCL_#GK*vljpD4w0GV72^HxP?xvjjJGqe<
zpE~!xWE2<V4a=6LrzR)HMuzx76Inh}&FVH3S3+(nfaX{kY7BZ+%}fxQTsUgV60n!O
z_AlQUQ*mdV!RP%ej)B4&`RbuHmy?;EoERsL3TJ=WKVLbKJAR11tye@+UR4t?lBPfc
z$P?%ZqteS7W-i}y_~QPxGwU+LowVV_q0Ivh6ew4?udAu7pd+}c2@qJg+F2+JPICk$
zi0B2e?<7&6hpn;3Kp<EMNh%yQbt%}3UDUsV{nzpTq4r<L|A*Rt9seI{|8@L-oc#}U
Xcp?$V#o9Rk00000NkvXXu0mjfT7n3q

diff --git a/vignettes/examples/vision/mnist_convnet.Rmd b/vignettes/examples/vision/mnist_convnet.Rmd
index 0ff1cd933..8400a2519 100644
--- a/vignettes/examples/vision/mnist_convnet.Rmd
+++ b/vignettes/examples/vision/mnist_convnet.Rmd
@@ -123,35 +123,35 @@ model |> fit(
 
 ```
 ## Epoch 1/15
-## 422/422 - 5s - 13ms/step - accuracy: 0.8894 - loss: 0.3635 - val_accuracy: 0.9787 - val_loss: 0.0791
+## 422/422 - 5s - 11ms/step - accuracy: 0.8896 - loss: 0.3635 - val_accuracy: 0.9783 - val_loss: 0.0791
 ## Epoch 2/15
-## 422/422 - 1s - 3ms/step - accuracy: 0.9665 - loss: 0.1111 - val_accuracy: 0.9848 - val_loss: 0.0550
+## 422/422 - 1s - 2ms/step - accuracy: 0.9666 - loss: 0.1111 - val_accuracy: 0.9850 - val_loss: 0.0548
 ## Epoch 3/15
-## 422/422 - 1s - 2ms/step - accuracy: 0.9743 - loss: 0.0824 - val_accuracy: 0.9883 - val_loss: 0.0441
+## 422/422 - 1s - 2ms/step - accuracy: 0.9744 - loss: 0.0824 - val_accuracy: 0.9883 - val_loss: 0.0441
 ## Epoch 4/15
-## 422/422 - 1s - 2ms/step - accuracy: 0.9785 - loss: 0.0695 - val_accuracy: 0.9895 - val_loss: 0.0397
+## 422/422 - 1s - 2ms/step - accuracy: 0.9786 - loss: 0.0694 - val_accuracy: 0.9895 - val_loss: 0.0401
 ## Epoch 5/15
-## 422/422 - 1s - 2ms/step - accuracy: 0.9803 - loss: 0.0626 - val_accuracy: 0.9903 - val_loss: 0.0352
+## 422/422 - 1s - 2ms/step - accuracy: 0.9804 - loss: 0.0625 - val_accuracy: 0.9903 - val_loss: 0.0353
 ## Epoch 6/15
-## 422/422 - 1s - 2ms/step - accuracy: 0.9823 - loss: 0.0559 - val_accuracy: 0.9912 - val_loss: 0.0332
+## 422/422 - 1s - 2ms/step - accuracy: 0.9824 - loss: 0.0557 - val_accuracy: 0.9907 - val_loss: 0.0332
 ## Epoch 7/15
-## 422/422 - 1s - 2ms/step - accuracy: 0.9837 - loss: 0.0501 - val_accuracy: 0.9917 - val_loss: 0.0311
+## 422/422 - 1s - 2ms/step - accuracy: 0.9838 - loss: 0.0500 - val_accuracy: 0.9917 - val_loss: 0.0312
 ## Epoch 8/15
-## 422/422 - 1s - 2ms/step - accuracy: 0.9848 - loss: 0.0481 - val_accuracy: 0.9923 - val_loss: 0.0307
+## 422/422 - 1s - 2ms/step - accuracy: 0.9850 - loss: 0.0480 - val_accuracy: 0.9918 - val_loss: 0.0304
 ## Epoch 9/15
-## 422/422 - 1s - 2ms/step - accuracy: 0.9861 - loss: 0.0444 - val_accuracy: 0.9917 - val_loss: 0.0303
+## 422/422 - 1s - 2ms/step - accuracy: 0.9860 - loss: 0.0445 - val_accuracy: 0.9918 - val_loss: 0.0303
 ## Epoch 10/15
-## 422/422 - 1s - 2ms/step - accuracy: 0.9862 - loss: 0.0440 - val_accuracy: 0.9913 - val_loss: 0.0298
+## 422/422 - 1s - 2ms/step - accuracy: 0.9861 - loss: 0.0442 - val_accuracy: 0.9912 - val_loss: 0.0293
 ## Epoch 11/15
-## 422/422 - 1s - 2ms/step - accuracy: 0.9873 - loss: 0.0394 - val_accuracy: 0.9920 - val_loss: 0.0304
+## 422/422 - 1s - 2ms/step - accuracy: 0.9876 - loss: 0.0390 - val_accuracy: 0.9915 - val_loss: 0.0307
 ## Epoch 12/15
-## 422/422 - 1s - 2ms/step - accuracy: 0.9875 - loss: 0.0375 - val_accuracy: 0.9925 - val_loss: 0.0295
+## 422/422 - 1s - 2ms/step - accuracy: 0.9876 - loss: 0.0374 - val_accuracy: 0.9925 - val_loss: 0.0288
 ## Epoch 13/15
-## 422/422 - 1s - 2ms/step - accuracy: 0.9887 - loss: 0.0346 - val_accuracy: 0.9920 - val_loss: 0.0297
+## 422/422 - 1s - 2ms/step - accuracy: 0.9887 - loss: 0.0346 - val_accuracy: 0.9920 - val_loss: 0.0291
 ## Epoch 14/15
-## 422/422 - 1s - 2ms/step - accuracy: 0.9889 - loss: 0.0343 - val_accuracy: 0.9923 - val_loss: 0.0285
+## 422/422 - 1s - 2ms/step - accuracy: 0.9887 - loss: 0.0343 - val_accuracy: 0.9923 - val_loss: 0.0276
 ## Epoch 15/15
-## 422/422 - 1s - 3ms/step - accuracy: 0.9898 - loss: 0.0320 - val_accuracy: 0.9925 - val_loss: 0.0279
+## 422/422 - 1s - 2ms/step - accuracy: 0.9897 - loss: 0.0323 - val_accuracy: 0.9922 - val_loss: 0.0286
 ```
 
 ## Evaluate the trained model
@@ -164,8 +164,8 @@ score
 
 ```
 ## $accuracy
-## [1] 0.9913
+## [1] 0.9915
 ##
 ## $loss
-## [1] 0.02430356
+## [1] 0.02471612
 ```
diff --git a/vignettes/functional_api.Rmd b/vignettes/functional_api.Rmd
index 51a3d0b2f..7fe7b8ea8 100644
--- a/vignettes/functional_api.Rmd
+++ b/vignettes/functional_api.Rmd
@@ -208,9 +208,9 @@ history <- model |> fit(
 
 ```
 ## Epoch 1/2
-## 750/750 - 2s - 3ms/step - accuracy: 0.8979 - loss: 0.3540 - val_accuracy: 0.9448 - val_loss: 0.1903
+## 750/750 - 2s - 3ms/step - accuracy: 0.8980 - loss: 0.3540 - val_accuracy: 0.9444 - val_loss: 0.1898
 ## Epoch 2/2
-## 750/750 - 1s - 1ms/step - accuracy: 0.9511 - loss: 0.1634 - val_accuracy: 0.9605 - val_loss: 0.1386
+## 750/750 - 1s - 1ms/step - accuracy: 0.9512 - loss: 0.1633 - val_accuracy: 0.9605 - val_loss: 0.1387
 ```
 
 ``` r
@@ -218,7 +218,7 @@ test_scores <- model |> evaluate(x_test, y_test, verbose=2)
 ```
 
 ```
-## 313/313 - 1s - 2ms/step - accuracy: 0.9593 - loss: 0.1323
+## 313/313 - 1s - 2ms/step - accuracy: 0.9598 - loss: 0.1322
 ```
 
 
@@ -229,8 +229,8 @@ cat("Test accuracy:", test_scores$accuracy, "\n")
 ```
 
 ```
-## Test loss: 0.1323339
-## Test accuracy: 0.9593
+## Test loss: 0.1321879
+## Test accuracy: 0.9598
 ```
 
 For further reading, see the [training and evaluation](training_with_built_in_methods.html) guide.
@@ -643,9 +643,9 @@ model |> fit(
 
 ```
 ## Epoch 1/2
-## 40/40 - 4s - 99ms/step - department_loss: 2.8465 - loss: 0.7669 - priority_loss: 0.1976
+## 40/40 - 3s - 64ms/step - department_loss: 2.8465 - loss: 0.7669 - priority_loss: 0.1976
 ## Epoch 2/2
-## 40/40 - 0s - 6ms/step - department_loss: 2.8554 - loss: 0.7538 - priority_loss: 0.1828
+## 40/40 - 0s - 5ms/step - department_loss: 2.8554 - loss: 0.7538 - priority_loss: 0.1828
 ```
 
 When calling fit with a `Dataset` object, it should yield either a
@@ -776,7 +776,7 @@ model |> fit(
 ```
 
 ```
-## 13/13 - 5s - 418ms/step - acc: 0.1225 - loss: 2.3000 - val_acc: 0.1150 - val_loss: 2.2942
+## 13/13 - 6s - 430ms/step - acc: 0.1250 - loss: 2.2998 - val_acc: 0.1250 - val_loss: 2.2939
 ```
 
 ## Shared layers
@@ -1184,5 +1184,3 @@ outputs <- inputs |>
 model <- keras_model(inputs, outputs)
 out <- model(op_zeros(c(1, 10, 5)))
 ```
-
-
diff --git a/vignettes/functional_api/unnamed-chunk-10-1.png b/vignettes/functional_api/unnamed-chunk-10-1.png
index 94a54466e96a38b0121a288b4101d6206d368bbb..754a2d2bc8d0ac882023373243f201bb53c5008e 100644
GIT binary patch
literal 20488
zcmd_S2T+u0yRD5m4WeT}LBJe{269$ZL?lVhBBBx+L~=G_Kwu1jN>p-|)T9O!R1_2>
zG>tSta<a)eG-ti@?Z4_f|33fOr%rvl&i?DvOw~Bjbie)f^W68kueF|jq^_#Sx}IY_
z3kwUY(z(+YSy-0Yv9PQVUAr28Be8e+a{OoYWo5<FEK9^cQAP33Sy*<nD4jm3<r?<8
z#l=-y`z>=+Oy;k%f9}37cz*dGx=(g*dU1UBKI6w19q(1TYJ6Chzt8B?o#+nnja7FU
zkI!$G+PQk`zKts%^xW!Oh<Ug=ZSR4Vx9_dIc<@{WbFuwCU)OK1Dkas5^=;MTVRD%?
zlkOLkpZeE^a~An4Ggq=4{c5@}=)RJLWq5Nd%cb3=oGgEwc3pAn_F6lZqnp}USyu0*
zEPF41PmE>9!;#f2D~Jb0$*VaxWm`VCEm@p;&AoxesA2Eky)Rz9>aQv0zCZml>snBF
zxN+*qWh`N5PM>zA2-O|@Uwq>KrC0yYdXV}hW48Ap2UWuoHgZWe2K!9?yf#N~W9s<r
zJ7uh-qC$;2dDo^gkVh`<jNf)@$3vDjs~hbyj(zI%vEB%efRVKLp`vLM5mHmVg}$wA
znu+J=ne--Q{2s+^+@{iHxK+iq>^dti#VpNZxJXw?NlCDBYPfCK$#n%Qm&3om^QL+_
znpdvGxK9tI>ZBSR5f#;$ot>3*?E8ys^Yg9G;-INWlS}7kf1!yMikm~Yte1hlzJ61t
zh2Q!O8?3YK^4NFoOiWHIIKOV)I;*Da#Mi=>Z~yvh)2cOVl4D{{-@kuf(xK-GUOwHf
zc=oIubH?15?dPFCw`>V|_Uvt(N+|EagJ<{b*;DcTySA!ofHR{hR^h=$YK8czQ>W?+
zUF@8GH)^Q77S=d<@^<mUnDSyL&r(7V)0(4w@dTHo<ByMzwvZ}mDf-buB?}jyIer(R
z^?g{wQ$NyPwvB_sz_KmtTJ^wpv9|hsfzofc98%3=NQsN<IrLWL2;fJ7c((_X&YwSD
ztH34`qY-`fh_rOGMH`Rj$Zc-dU&{i$HBKogOuk>ihEHhxyV!YUWe1zixKP`R)e$nJ
zN_Ih`gt~Y&)0x=@Lp$ww)o{TD8<*F=W;4sVJ-(3_`f8#LD6ScuGgb<P&Wwp43T%mZ
zV*@|l-0kS-kWMqy(TVe6jz_w8`io5r6%LzB7%VoIEYeFXC;Dp0jF{Ni{!qK((Y!YO
ztgA<Oc`GdRot?=h9t@GdCd;<sqN`P5iF*`$6TIf90_M5JZ^$R>q}JY9wZU<$ODV8v
zx@UTyw6qbM%*0=~!*O4}eDRy27kT6sI5P$`A}1rdZwJaT8=q+n^Yr)k2UMH4m);MU
ze89@tqqkHgHJrV$u#nQW#GKY~bIU&@6h2wTCZ`uJ<>D+gH8OIe+_}|#P(b;a({ziA
zAwJG%MKAVk+w?W#F4)suo*z`%Mc2>28A})4E@;%(c$Q8k2{rk2R|Ij(tWopCG7tLf
zu(UWW)$lXNI;2|8XK`+F{OkDF@85Gi2l5PO1~xY8Vby#j4~=zK%5dAwPcgX4M`t<%
z3rYJ99{lXy8}1^_^qR?k`*!c(Q$_BKg~dg$sri~TV=3z;<|28X`f+}!Qt|ZZd+|c1
z91g7}YuB#*LANVORpjx~VlpF&97jh-o8Fn-eH$H}+_H3+O*SFK)LZ|<hY#Egp6%P?
zKKZc^;<HH8DG9dFPL`!?;p9wqY4g&>jZ}}#)Qnf{4|i$JNcE@`$W2gsevo_8VK6mU
zGFZs8xH*HO7|7j}?+}0R!pmQy!)4W2`2C~hJhU}Dp6dN;^Lby3SjYKuib<MSFnW7+
z8QudMO6ePjPl)&hr=pn|SDD3$I_f~7&l1ypATgej*4x_~zr-9~(utJyN@^BsHx`R<
z?O4akDVkC{#ctyIfvg&R*57fSk?T><%(L&bsU)`RiZvS>gI)S-&+Oc})2gK)N6|}S
zr>M<K<{YhrclYin0|sUEGpAKuoJvl)`T8D8(T^nUWL{cPXMog%>C&Vr+2PgD{(}c~
z^6lj0<OGZhw3epqmULo;G}a9#W$-IM*vM7eTI7LEd2HME?E}AmW?9uIJjG^qV&u>Y
zWX^ly!Y67Jd5kBxe>deQ8|K<lS;d-vWwzEQYPOg>ZF-}y#@K6;O75-<7RVCt^Y_<w
zcP~Iex0y~%Nce%Zn(Z-di>0QDn#CdQ`VK|FXh`nHFZu70a&4o9?1xI`sOh^(#3a)9
zaJg-1)qW8gnw(=z8)(chC*JJYfL)32-W{bcDkvz3-0V7+yY$$7ubiBz&pf@&ac0C!
z#I{*a#$&49iDulZ7mvL&H|?|JQZ!MkY%;Cr>QR-+pK4sFV`*tQ*B2e=I8cA2e6i4J
zFs{2Y#B`#YQSF4=x?=SoKYUp^1)G?q8-yBtT|xwn>$9$@d0}aj7bUrzuoo<`+{xW2
zGEUMRtyt#xfTQqQ#@R(Gp`r#w?m21sJscsELnTYz0{WS!+q~yfNpxpH)*<uqfd1!d
zvNajz<y*IHizSojNxJDK_olmT`)gz87<N7$H5xW3CfW9#XUl!)1unxTj1g}vL1C{s
z$8B4;8fAO^dVhPa(v;aD-X`HSH^F<~<;<D;G^b`IW#u}2ZbA#5!$s4)qM{Am#VA-e
zrJqT=Pb5^A>uziFnhuwA8A|!~?Ly$-!g%CT{X4U=2}W*PlPc%JSV&3A*4AX*^oE*f
zMQWyna)?u$YWTp8R@LzqKHW4e4UMOjnfLCk#M}E$w$;%IUcGTw-&kNnd)Wh#D_=Hw
zEl$!NKYEm?#l2}$)B|q!bI%W{kx6ag&z}AH{pB$Q<q*NtiG0@e>ysKL%!)d{(q4y=
z#(ZSw`u0^;S65psq744b7{b~}S|Alz=r?@!=j1AR#Up(EFS@9t^KZSxh{7zDR?l&j
z-1MT{YoSJSjk$J_q!!$9w?+K~zbuuc@kV?K3_XUd7*Y9CHI0mH&YV44kV4BaS1hN+
zsYS|?JHuqX3d~!4mZp7N=8YoV2aeF^4jee3bN%}DAaA*f%F5JbD^}OfdgzBg5W5z7
zql*!Xy8Kgh-@@Xsk4uVS%L!F(?-{f5=wFkAG;1vG07dr1#6%OWo#!t%#c*(l*`^R3
zQrFli=(4#vwocr;ckj+C{!9xvU2R?S_I@QJy?8E(+pTMJ2yLe@R%vdFbEt{;Z1;H!
zRn3EIHtzb-kZMS(3~x9m#K-60A#-a#sg<J7COZ>Vnfc>~dTd-=!J_d^i_wlxb6wBX
zC~4PtQCuQ@nEA|JA7(k35)~yMuNFxTFL+nDqjZaegao(o&M7oXabK77=N||2>vw3k
zi5w%<{`KkO$3C=x`iGnM)#C%_4L|E*>5<me(=$*s-I^7CbtZ6WhApI>A#9Q9IeW8Q
zu0h6q;++5=KYu-H0jaXuVY=I6xV0#GYikHS^TWMd_OPB!)%bBX>+t*cNhQsuUXwv;
zQ{!EnZdxthUY(enZ(Axh@JSHrcHB6zL5OmRvz2D#;9r1EM>I!vGO^{jq?{e|ODyyQ
z+P==r&PsTDqk41M-D*$7l1m>9lXk1^Y<uc^XPi=W1A9DoK1|9b?d)7tW##x+Z`GxU
zX!i(@AqfGaydUu&1SBF`du&q!5;fyZCbH^Yd_5HD`F5XTKx43?qN47+k&T*GH%n))
zskf=%T00@v%tlQWm1C}w!@q~L5|X}|dW_lmKnNAt=I+@k>zQ};+e?i@UQ)){C(KIk
z`8AZ}cLy8VOgl;D7K%!Q%h?{1lPg($_F~B~QpYN_JLz$y^daE_<^1`J*}ObA)db7!
zS}FN7hUor4>9L1&uYg?=Z&M9(2cR1yJ*Ix2u|P*n96NJ~&|lEFj4{9O=)Njp;u^$K
zK_|EJ&`^ge(n&XAG^%-f7h{oHFZ31i_Fgufb2r(z)=t5(zxLCi`R<+`ou-~4wt(cI
z4MKm{>6Z96R#<eXAGU@jCMIbfL-~q+>_TK2mL1ZrH-gFvQ&KMYZ5OEBq9)y*xgf{v
zl$#hxPM1`DMVW5(;5~B0{&#sMKc2QJE%+s_>^YUIWPYg7D%W0OH-)t(MgPR)Pz#;0
z^U%JtPj2wbmBcWrw{6?@HItvdkK1KX*T`palFaVsTG3UvIb6^_#QSx?jofT2C)!%K
zNviJTNEurpWr2d$QPA_++Vi-@*lcg4^~^}SU?qo$RZKaLPgvG-t91^otKG)F9^zVx
z-F5WRt+^@2VB=>LOkPrQLc)!(>fy$WXvR*K+qZAK_q@8cYkuNho&VZ_`JKmyKIfvv
zl#iJdxgU|0EjBP0cN>e*NzpSiWY#JR{Y=X$vTKtKtcrQ&T@d_U^}>ZGz8kp)G^En*
z-M^oRYtl~h$=^d^txY$T6F}GRW|v%?pLX%MefsLvIL0c;`RU=@RQ4S^5@MA@bPNg)
zNJ<)@n+#BsL)JvQIQMrm<=8nn2VJ8d6zCWk{noS1wS6>E*J45}(8woc{eH^v(|waP
z)9JiGYO<~r^Nn%djW;wOn|=HCiCnAN#f6S&5c1j7_NC~tZ|OdZ)~sBT6L0S)x=-}A
zYvf6!S`)WQs*2v~Sf?^KMYqej`kkY@w>OzYClk%ITGh`%@{Gz{|3e`e8Dq5aBuB}f
z$fbpA<s&W!WMx}Nhon^niwyniY7?JLgtjf5C6*qa@gfZszX+-!AiYqGIa}>x1yyFU
zcuYV@h=$tSKg#_1^QS@f^{3FJ{h14o`O<z&G3b-TI5XIgniam$@5vJ_NEk;tDc7>=
z%Tq!zl$*a-TWhxh)keHE?RyYJi+GiFno<5m2x=FbeS7wtbZK!4(Oq;Os1z!xC*`(W
zHRRsB*{m>luH%zm(Z=DtygXi0^-N4Uex)8zesj$EVW&^qbn}|!pzk?FZGQB855;OS
zP0Y*F3zu}NB?hXHbE!#$Zb}|8$+Zm+;F9Xm<{9s;ra?TAgy<z>6?~HrY<zrt%Jf`2
zU9_>P2$|%*zU!E?w4h0J$^*EPy}J9m14cwYqD3cB;&LtY=f2x{cjg68qd!gH_1rvl
zDq}M<Gm~;!XQBL}R&Ch6qwq>^By-|WNXXkIqqbr%WA8&65dRAco+8!_DNUVvkjoKX
zQ~H7F7bmV(bWy^>cHwi?!eVohK2)fktV5l<@P4Py+#eZe`ysiXZesJPM9Lb6irG0A
z4%>N;a|VTkc(jy92g<s)_oQn({d&KwXo)#bwTc!LFrnntW~N5G<5!+;b_nczEzx>d
zK!E!8jGuIqn2?YfE+FxuT%pTw18$IX!N3lYyCb4sG}UBc`|IL8Jl8Qxn3F;+ZoT2Q
zforShJN$&uo5ah7VgyM(6)%pEd)Yk((+Y=@vTv0fsi>$BzWQxPdDbyO?TZ)RLLSd{
zHTOk&bO-55&pn}xS4nY~(9(zzV4x`{8Jal_w~05`p|#j+ewVKB$4_hR$`&BUh#@5j
zqMPQ-(S@F3ka77wAxyBL+^I~{r8!-v!3GtKE#%(8?OV5AhcX&y&d*X4O04?&RbzUn
zW%ApJYFA3Ive!j*^{5Yb*ARlfVY<!7uCVRZiE9&cgXzgo9421+eTf(dU9*Lgg<1=-
z?5xMSzFN`J!kHrv<oX8FOQvSR7V&`VlLOSbmSJzmCCNkxw19Z2j?1i8Rr8nx@~^MT
zQO=hNne;)Moj@yVB{x%735K|jeqtx71&vK9i%7kC_3E{_{oB&gQbD8Lb^kO!Q{1u?
z2j4aREObftW#dVE`Eti3gRU!JNY=vmq8cHcSoVO;8XG7`P?3qf)Z^--sy5uBucZ~U
zG?%_qi<vA%(6}H0?*-aQ1HZ#yH%c?1VXsRRhfqhyh4_TDH8kGnnz)@I8U}F>hgu4g
zAzL!}l@F<gsbhgNM+1GdB#)aEI3_Z(m9+)k-I%kzTz01yh?P9yHdPP3C*m?>^gUck
zS3^T%ptY#5qqB1x7gy?+FXw`SgOjkk2vp!Y618!ccocM<dowfNp?6nXHDvZohvmyp
z&kaRi-aN$hoM&ig>lg-Q>g*1;my<I+p>{TMYFSsmCykGeCM=Bfs7hW4WnonZhF~Q6
z#m2_EW3I7wJ3S&&%6a`3K*@EFjk6e`D6cc+7r~5JFJL!vRkPeIc*5OtUENe~c}S>P
znXj?lw?xeGm<OUegF>;5r~ATu>V58sU4oqTb)#IljU9_mQ@3p2DUw;!%DfBG7`mE@
zy*!NBR>+=mT@i8{3vHnN0Z-bX+4~1CMH3I@tz9N!h7#l6@#v7a$h&2GDT9w7BemVU
zERBk5>*|~qW=1#283xBf&W207U0fPoT9m_#qw6FtS--!O^Qaz196WB&eMiEIZ*Ko{
z+V!8soBluG+W+p$jb-sy-ps8d=-cz>nw*`>lDczfCUwvY8kX}YtS|ohOMT70<<h;w
z!|B@v3|un@O_$63<P#RwN;SwyYE8Ih@FPZPO9*|-mMs`3Dh;?<)g4}nF8c@Z``^8&
zf6os8Z+WKWgYGQ1r1JCgoAT_V73YUqi|E}`6BCY~K7TGq<>BB+!XV}$A{!+*qaSQR
zo!rfG=+mc9j{^gVJ?%g31e$@SkotLkSyI2yh0v(h4I4Hj=R~)Uc9zr5%gH-!ZarFh
zZ`~FvN?=05Vb-0;WAI4N+atc4Z43Gj2KJ;Rn}2(GY&|P$YK<(5=^l#Ny{M?&{@gMJ
zXXg|Aqdng1uC?0{eXvYKP*4?GDLrFqndv@?*%OrgB#X*mk=m)5(avp~HmUC;tq{Sp
z9s|>H!c5z6j(*|Qd*ORxPcClR<AL7ZS1_11?p7$TvX^4^!nUP=z-h81hmMbr#>U6T
zEkos1@U**3dAuCVHeNXGS|(on$8d6bu>o55l)i^N30D>t6SF@yCT1o^pg7@PBChKz
z`<p>f7W>Z6vzLVB8F+%2M_Ue6w3WE?v|(N*lvDA}P3o0+=#gW`G{l;3DC9I<s#tjt
zk4zc3XLspR)aNHVJIiWVa{NYoGi>v_w}X4?CJw#7y<!_XdvE9FWjPP<gUgpMI|B6S
zBnoOCJj&J18m)cc-ai<{7mgjQ>K=H021=rit81RF6VVP55)wE#HmghF2Lk;37s`BD
z$CEaOltu5^O)-<Ui&NdZZ(qLJ0gjcWoS!)%?aHTfcS$+(PbZ$fEALw7+S=UqEbGD#
z^-ZnUH%YJSh89XTE%6qS+Yz;$_$#R8WNb_EA<Qi+SFK*%Q2X%K>ocy=Cr75IT?$X_
z);xIq+*&)~9jfi$ym)JVjL3Bzz@VLg9#+7;I_dDyqv{qG7D43RuC4^g6Q`qlXNdJN
z5ihrxnD#&;eN$`5m#V6R6wav9(o?Ovce}g0sgs)qxZ+e;{$Ftazv2dsdSAfJLJ4{O
zI_c!y)e=LR=%KZ>wHpFYT&p?^l;q~-CV8{#VF0&G(%9IT0mm^GtY+31c(~hmk9M+u
zZS6${Jy#9fQ}gUuU)24{&Oo`j>-Si!IO-0DBw}k2BQ2qBb<&Io1=7dnGjkW6O^2qi
z;@7fL&Sx>$Wili0S2UIQlpIuftwk$a_G@*!zu##^#X-ylG`@7o+xPNUski1HkN7GB
zN>Ty&<8R%%RlvQTg<U+`>XvGl_!}4p#xoLtz%?~Bz<vIh5#C^a3~btY;_8#_{CZIk
z0@Ql-Wm=otTUfZ{yo~~7re0QN-nzwNjQI@L6MXi`jy-^Y<zxsbS&J)It}~i)s2rM?
zFTcHb@nU_pr6$z-;r;thqwZ9E{R#$K=-Bt;jaANY;`{gVAQDN&h0f>Com02B&+wjU
zUqkwtdFGH>M2a&Vk|g%o^XJ=mc+zo2_kDen5)#g>TD6Ms3D&P)@3=5y543*{09Dpa
z_2kKu=arN+LPbei0b`-Yv90(H9a02srJ~b_KTkkM7CM17)&xZ*YS)@#x;UWg4qV?7
zVK%lfUd^{}p91uR$s`qB{(G2&rh@Mu8U$CKmbvlk794<BXmr<IZvlZ{100ZMBV;_X
zt*E?87mbX<xIITsV&tyr9?#)dCdP^|35RTJnqc6Z+^{dF?T;3hw$wKY4|oCSJKcd*
z=CFw;@F7{xP4t^g?!+L7(qajzJn%Cs9u&x?&`wF|5!tqd7(`A&#hOgsb^lg+1_+d2
zX>!h<K;yCxaO2#Xb#C6wbefy6%uTQ4a%pb24?TA5ScB9s#hg(4i*p2XC%{4X@bEA>
zlHl^>e{gvbtNO!ya&m3aOl4)|RMc>f329Eea;jcN>0oc)W3k~S3_Ph&m%53OKq~~<
zPaCg(mm7FQGj68K$yUC0?Xa?uZM(o%4jK1sI<31-O>QzFm0QMLOG3|k2*NWP)rV_-
zY(FJlCG@JcmX<5k1x^7KZcTad;^JZ|IG}d#5tEP4X`+V@e<9Rxa*Hldg~;WP4<(m&
zZE2;Ray1>e`G=iS1G}kbd*f9=N?2D8<)Q_S{g(kUXMPvKvv{|B`SPS)tuUKR)=axW
zNCzk`0YKOVHRW`E^~kpQc|39wW7^e4f=`ZDA8$9du>9tx@GSd!&0Y*XjpXL)YBf+e
zcidIu;PTI(c`^2G-n^Mi($dnp4g}r=OQb4XDjr>dm0Ma5E|lOjkNAzKp6bZ-HQcW6
z3lEK#7?5rBW8I&<2n$Q$w4-Q1kUQwhIt^&i3y39G;FMg<oU_J|&4;-Y?UbO<gwzJE
z^N@*vKqAJ9fcY!eu34joql&(K`}VEyrS~j=i}}rByLZd;>7=NG!C@GpaI+2EE?|yp
zv>Bx-wxWu-%r|`t6{{vQv#MnW^;~+Z!m%iB#=4W5vNH%XjP8y)+@P&%tR1^=IFj$k
zkqp2f5@91WU~IOrxOS~DH4C>XWbyOIkBtp*S)8Qeo55$Sp;;TA9Z*i8P^2p}!vZOt
z9UUnh9hd2{l^~rG4n2ARC+t+T)hsZs_L-J4*fGGuzrWtyH+=r3XnuV+N9(f`ltgiI
z;gzp{hD`qX=btGMz;t)TfL$3toTPdGz`$J0OcsVq*bgrH)paAS8RieDKzn`zzW49z
zThr2<QgVT?Ll&7G>?Q9~vs_0nr<;t!I<$rb7BJ5bVrL%)1J5~S2~e}CL4c1ht0~vc
zv^Ubyg{`#knQ+>>kE;lb&2Lv5^Qe*$tr!Sin?Nc;C6*w2p^Ow^rYCoU*g4p=eZ|NQ
zE+W?d<>@|ll8~TaL)X{mTxvcSR8^hb?kAR%ET+RfG>S%r4lJKW(KZneui1tIsPgjY
z2lB%Fa0%^iRw5hW*0Ym7PFb;h`QP4)s9rtaJu$(<hY#Q2&)uLNpYy4*qT)qxa1VL-
zd#ISyROsv117ucRZjax`uuKa1rNEnt=LY!b1<RH#d;H{yqkgq2K7X*QIF#lDAPp;U
z!RfoqSE#6x<fnKhXXB|mca}4n^Xtf-?{6*B=H?mLXsXw-+=cyRn2bk5_Z0SKW5bbu
zK-iY4@Ru+9^jfhS7YR5`ud2qRX7%auwt)Fv5;spDIdX*Djg@Ls;RPf9A6bB)(xruM
z&meZ3^y9%D1{zZ7_alMDQ(*^KT)k=qzmxELorhafA3t87Tm7LwCnyHHhMsKcBM3Jk
zBM^PVw#sZrDJQ!yT<MI3iT=7EGDbs*2ETxSv?>36`;Y9|^I=dDyV!Oq^oGl$f|iQM
zc9c<HZm$WAz|S=p2QX+W7r9Rsyb}-;)3qyEB%8SRAMnB$NOo%B$`9laIS;#Q?D{U}
zJ9qDH<K#4CbT3mcd?|W+KOqEd*;6DvXKuuj(C%_F$D#`5_|x(@hGVh4b%WZkf4H-X
zSUEuh=jCJ~fdQ<*s9&GB7SQRP^M*Wkdw5gpU;9Dss~NIDuZgp_8*fPZGzT2oAm?`6
z$&5DTFV>nsOy-G9H?)a4!h<dxE-o}+Xz6O3nAQsrT^kU!JcIJ~?J2bW<ZKzRU!9)w
zEveppkKxfzCtA5G*Angtz@h&xiHkfw3-&oWL!oj$#g$8*Hys?3o8!ci#^&DI9_wX=
zO{j_<;M9B#&n4J<TA=bGTz=TbMh%r!RnC-~b(-*NjC1G->w+yTzrPkS$hzuJz)@fe
z6J1lU97(5v7{YHr$0qtU{l*g@0?e?J7;H_lTUw+iV2U!H(|Q1#skSSiU5mdt4b&^8
z7frpUFNRIcqHl`ec7|BsX={nSkh;%E^Tmr7htW)7EV7f>Hf+!YY;nSM5KNCeiC=9x
z-@Zry$f?;+?52JBy%EWCQD*#cDuU){xi!H8hRszjgjGFUJdZXzH#tbEgf13_YTyk#
z#?SxzZQalXEhD28v=71r4^aG?RxI2MOBF@WW^e%xW?DldaFZljT!H~9$`(TN6&Bv)
zaGNPWmImhR&e@V)cuHlX=RXnvwQ5=rMPx8*CmPnj0n{IiI~SbHzWlW(KurEd?Ay1c
z#~v!5KhI0TL_4}S-a5TVXK6Nash=voKM~%WL7{W1^Y6wtFcs4Wqle&-uzV^nH+5`3
ztEcx87f&gE5W8OnXwBB{c5rxjDpZ$c&$m~CmFamM>q%|J#wCjkyBymVNxIOVfVk9n
zHJ%n{I*CK#=9MDZy`G>gMBl?ue14FYP9w|+C_?du{KY9c6|N$EOj)R1Za6VMqV~lR
zO&im|Ymr_$sQxu@2`%T#KcK(CC@q-n33V#xa!DZYl=+!OMOQI-dwYA?{G(T?N!kSN
z=&dRKH_!?6w;`mPP++o1VxWCkLJ9%h)uL4qT!*mPKsWvdC<S{aRlt>?L1RGkl;$kC
zf-t}h{`zp2_wZqLn11NbnDnE7(tiq8yGOL}d?g%Acq;V^v*V<}Dwul&btK!hdDmRb
z&dA8H`k5I^xDJJ)tlu!9#nMbXRM3aVXN2GX6ZH&Ju%jjg8RS@XC!00!J$m%${Dliz
zSiMDQ^RTS?Fj|qj)nq1LJbU)c5pTP3Xv@~EF|uBB38+{m(`@4}9bVowcp4d*j;ppD
zs88hO;~P&J0{jxW{$qDwn$5>02>`31f<eLDd4g@a_s51<nVmkmtesV<mSBVvb!y3|
zDMU0QsksXGJPxJUYIdyK<Ut7^?pGbUdvc4aSC03h`{eJRbqvq^iMmsU&?e~+&2R3m
zVbdubMmHoVELJU3n?Y8eN7uX%6cmK;h8K8KA)X>WP<HN%GgaP>h}3|cH4seCvM$c@
z<HOCk?B|NQqv7KAL_MjCkMvGLWJ3?-Q+DU`bkw2?Fy8WR><MT|Z8rK^Sy|VVW1ZvJ
zf05LjS2I{Fw=kmM1s|RRCe;vST$PaY+~2Q-;N@v)#&hLz%s0d)L-=B4?uP|_wX{9O
zPfnjb>z;k^tnqmj6(=JF!@2(K?99yLxZ-|nLE-~9F^Ry$<Jk3OGnNf?;LoW4m<ax&
zRyhB&uHI$mxLXN2$U!=A0hK}*PifA{WHJfW)P%B$$p<N}!+ZChYHx2}ziyod#PZ|E
zk6&PC!OrbR51#HnqwMYNP23)0gusf$NO1q&y#$a*3=DH9a1#*a{aB>hCJ&R-R9+oF
zqa4hiIA~UX$Y&2l$c$iooDdHsM|=N=d9(kDBk8~Hnuy_6-s}0RSBbc-#4z9YKj3zE
zgzi0fKqDsh()m9H46+jj_e60Z)zxv+1I>gG#Ff#@tTz7tBWd?9CZGQj$VjCBSZ*AG
z%)wwbalNgj(50|BA|fIU>8IRDf(xPdrhi^jze=t$Go^rg<*yhlZ$y_cXC{pn6O;)P
zE!9;{US#pU<(7I6Gc&UsXJKjYg&BkXH3@>HoM&GiH9O@$y%{MIq{Ql7<!%XW0>x7i
zYw4;9b8>pJ^TgRVZ{CPBt@#D*`eYO8C$tIeI!}(hBiqdSc2@KHD|w$keM&?YNu)vc
zmo*+#1O1d^^E2~o6Stf<1*P9Ah7@%W&+1!JQes;)*}!2LdSuH>-w|I8tG!o<{$~uz
zA#2B;poHJ9O*IsA>ih8_o&-3dfgLPM+LB<2A9M~z+X;TV{Rp6|JfxlA=E>+kves+~
z0{H3k7cXL7yx1C<{fRy3=(cv&eY1NV7Z(-?7mDY}>I5%5MA*LL^z6Lua3>v#H-|)T
z5L>a_Y9}6g;NZal%;*U^oZyI8@JQ|O5!OHcsDF9P+*6z*${AR*K_IvJ?XO?Ic&S_%
z`yjA9rT0cn;XUcV__ak+ikE6pu`&*@FFWbp4a^n@80oLT;%5WGYw;ZazMhS3I;o0C
z8*baCt*~#o)f)T&b;z>uotX)>{idt%O1yCI_j;u(SN8uubPoMtYK9q%UfC@hbbS-v
zd3ET#;mzQ69P3LtS??Tr(+o`jHJaYDW<~vS{M9M+45D$Ddl6Bh_wU~u01ak=(5*%7
zwAEdErD7%Px_1B7Wk<pYD9FKlKsgP2L|ac!k5(qX4TLW#B?X}|^w=CLYSPBDD^0OJ
z?l3{@sFOzrwqcvY!oBCRLQ7BY%e>V)3WZX@eX6J|d)+?;asPEi=wJLyTlHuwZcSg*
z)Fj$_xfk|`mVtqLyM52Moy1%YmlHE-4c50dvJT4oR%12i-X<Q7{DoDHhJis6dW$8x
zY{i!^>d=th3!{oW{+r}g%$vH714hKaX6u9OtgQ&*bL#lGZn2%sM;*@9fcte8X?a(L
zXGB&DxIiA}KMhK0z+IfBc{j!WDLPXTb8gTb4(Qe$Us5>KO*gmRAOwj!LRectylD!{
zel35X!}<fv8PMYG2Dt|h9`GSE0Ewp&$St##n|nw@%KY<_8YDY~A#q=G@ZRFx3$WT*
z9ypj$&h4PT9g!+R_2YVKk=f}&X0WlrYq}-=$&){jWg{Dy6XFT;<!ez=5<Et<%EPET
zKvfV7AqXF9rB2Cg1E;F5*Fum^6&YS$jK*wo3sDdmeHf9cw(}Ky=~|#sg!RYGorb&-
zRy#<h{I#kuB9|M0oHFDRl6OB2RUbKg_#8$|0yz>%Cp2hqh3>AdK8zYo8Rl!Q{V5`U
zZ`vY`LNS7Y-9Xs%$)=0vkqwTcpSTd=!*l~`OvuizPcFv9mQKuS5bIWmyml57IVi*x
zvoC#E9;hJv?}EIK{ru{PNGI-QfoCx$BPb1G$U~;MTVtuW<whB60dguL$Wfm;bH>d`
z2pF-=bKIXGI55;(l4Vz{Sb>P3I3XI~UuppykfsHW9I1g=n?&62u#8MgvbTDme%?9g
z;sc2$azuu&y!RbJ1nK8hRmrN6vWzLGe55I*ZBP7LU^QiI>}I;@(mzF3Tc)<)lWl|>
zT4bS=dDfr90t?ln-pA51QN(+}H8A2DUO9CepN{L|F5}UjMII^V6elELcqg;W%Qq7B
z(TTZpYU-Fp#nwnl73(>UcCSa30+a*#ARx~31)g*L@!UC7{15M%%4)Bf_6It!Rn1K!
z?oDS~)}>&APoEPM63Rv1o6(qy>YW48IUxPX&vmS6^{Q3+Ld7$ebLJuW-a*Z3^hTPO
z78(Ev8V3Lg&LIP^i=L8@@b0yk-B9CM#2-_@CL)Uv0^g}4FE1ZdCdJ2BIrvLf=){S-
z(Y~ctkKxpdh0fA$ffxXLvY41PQcRg+Ax=#FgOp@`e+j#D&)`MGAbLv=o}ZmPA|j$a
zT(VSLIfiB!8z0ZNcW)`#AJ>dMp+p1~U@v13je)bzw}1b~ZYHuL#JEAkv|XpKFm(~W
zOV(0*26gBJ{SIa6Bv=VZWM!C(B?q`crJ4GqB76a#hL4x`JR(;*0Bhh<*6HOZJf>_^
zBBX!XNB;h8Lu!0?5>-@FOY6nzKXya`cZAHZUb*r*ND4uuVV~1FPs*s%d}IegD<{7c
zqKep>hJro)Y#{RdT!TXZHdqvCTImqm4G0=r8m6bDd@C48N=|a@T66#cY(O+972>|h
zeYQI!6)&3WMBJv#Hf&Y8B*?mL?~70o>#P1ea;*lLR~~`=UegaAf9e7@t&JEX{Tje4
zG1>ARJv#E%9euR{y?s}ShhDL#TewezjN6Sa<8?3|3XIAOpE;o<mOF<a?U!kC>eL;?
zG%gzs1BdC@*rbq3pf4Jbi?bOlLD3Q%bmM51a{m3!AlKGQqXqM?HDE%(<kV2hU}l@?
zLI<LG+l5SiQ?0rG{PQy0z9##~`y1FL101G){|uOn^c+*77hq1%!MHTIaDp*((pC4_
z=WW4xzmNnD=tiR7KHm#NK5WCBR580&>2fl0+yR5>C$eXAo_zpAN&>Kya2UBHZg#z8
zz_ENS?UvcuANcj_SHbk7lhe6NIr&J4ZQZ)HYjg%hyL?0ZiGmKedF!?uL|`A#cMdt$
z-;L!bs-?|6B3$I<Z{ZM$bM<OJ_xmHC;Ww-b=yqN<kWF^g*M9|zr+EtC@M^iYrlw{K
zxxK<XP=?6pgjm3|s0UQBFeGU0)2B}b2Y+Jdt^u$UHvjwwqHc!ZeAC`P<mz&)b<0JG
zXG*IyLHGn<y(yS&$I+;YPKB2C4a!EphL%4E&n#$ejaFo%OF}2Smhm9di(!LvReIp`
zMak0Quq}NyA0`rk>+K^EO+0+yz$daagx|%P$dXf;fOLX`j)=fuVHb{!Bc>AsH4je5
z_?e|>NVciLoQvE?`;>%u5&0Q*$`ub0Mm3yZ$*IkoHogBY@rQzI=N*jslEKJi71W9n
z7LPlq8MWB{-u?so_q!S&!p>#*Y#9>N$Tid-1^fDeXwg88IB~E8vp&H^*vW|h4Im!e
zpe$4r16bEUPAl9ujp&7CeS!wJf0^oy8}Cs6lZ&KTtk~-g{<WVFe~dfY58!QP3rI>9
z=;&}ePCtOH+gVhPEuD0Hmw?_od90S?i^l!&y(b)cE-+dObsc)X**B+)*6YZPa1oTv
zdvPwm<r;XTh}<Sl)!sn*=#}<EYH@<Un~IT8%YmWHOO+EsYw%g<<ROkI=q85`BvlOh
z6`Vn*cC40}77x}1{YSfpf!W(jt1wC|q_P=j@K-WqwOQ?yBoT-Pnu>qAPjLX?E3Kiz
z3+TjUHd4*j);5SfarSTE*F$*&V<!;|J7C_1PX68<mSCYz9AUw}taOK%3edH69{?jv
zcIFiMfLOVH)>#5k5rKWAp$8GC)Im{lq-yIX5;oa|3yz!tFn9*qN=jrGM#}uig`p3;
z!9Kmlns&j3X(ADNH6$NEy&n^sLt$i{ED`U8RXmhQJgqy{f3nj(>&L(jXrSP9oo^W!
z3y8=L;opHd){>d{Rl_3`xupe%a@mrZjtvC4(&iC<@+tePg$>#)0eR`ORVb-_7@~6o
zU>X;Z^#2ipH0VACY;Ww?6#fW}7A@l-vFk?<|D=ZaO1TVa;qVSA9QiC7m4l9aeLUKl
zb~0-{Yw7&)Ef6Wp#d&Lx%JXV!F49ZW6rgT#lSg^(lUC)kttcGsUZM(UQebD%kjDka
ze<ce3vGe%H(ba3yVAhDQw#Sx>B?VhZm~=VfsKJ2T(zsj^$AkR@G{&buSb<x&Z|`yP
zs+WmV;fR`<neh{BUJFQxi>FP$dl7?kQ&kLsADLNMIk1P|1J+s0Zqqe-1Dj<qw`4ID
zE&kg4naj8aSa;EM+d>uWy%d__y=u873VK}{oNfY`5cz6HsK9c`+&1q*_o?5OhTi=L
z7J6kS>os2=w|LAL9Q~bDorH?-j(93*NgR=7=piMsHqWe~yv~g{I#FYvlCZRPc=cgH
z!Ky>-pHo2pO6FUJrJG!s3$q;P^$1fUcMH{3kF+b4J3m1ZVS^zToPz8#72d1UP_sUf
zi#8IADSq_uA>uU8g)M9ME6JrsXZkF8BPptm0tYR85oHhN48q((^lts?toeOzIJ_Y<
z+x7f4*fXgGA4DJGr!|Od3K={W#n~_DJ}sbK5-slbtF?A{6HV6XmN<dO9l84D^XJiU
zJFQ@X1vvD9T3*DdBTtcU?gx;K0H-5XFU?7A=GO|0Q}4+O;Fhq8pAm1H?sezRoqFKU
z$^N?XL8ckE!$t>nJ_I*JaMdc|lG^AMCe#)fa;sT6x@v;_XXMtfao1yn>~=1J>!~SJ
zG;vYj$=ZXvDSMO%qd3BQ&Js1{Bs_$(LG)SN^LmtDB9l8f;$lHSHgxnHdSZG*x@kHr
z-GXqT5O0I)gU};X2xZ5SUmvWdjS*BtG>}-eB}pyYt#}BF9@Oh^2S3-22D6Gd&GIvi
zqc8Q}B7*LaQwW3#`|mh*3c=E`KHg#?C5U=52W<FvfE4BPkog!<ixS~dvkQLAC=3ea
zLvVng$-#ogN#OhLEz=Kpyc0kG&%ueMAxxME8Q&T``<r-q9GzL3DVKv<JD6%(67fs)
zWn}kPk&huy5xWa)_?Vy!S2Mu+y0Tn&unwZ6_!aaTRea#cfNKJ(5m|Dq8w2Cz!jL$?
z3%Z`36s(+WIH`x;LRjdCG7vZT_KGzWgP(%`tff2sH|{6q*MQ!qs1)OPN(Y;=wl-5j
zDdBBG`^P?fxR&5_NEg5}MIam`@F}sg5S{iy-<2)8{uO8%x91|p2I6%w2Ef<D>n6W@
zrw%%Kw|IU?iik5;Y_O;1z<9p?_1OU@nJ05yLieT*Tv%)O>!yOM?ihQ|zdH~5FCHHM
zqvNOlCti*iXcO#?pE!X&u0ag&uvV;z_#rQ;4HhbP^T4IkBlkx}Mu;h0!POLT-tA5@
zE437G{LaW$4FexUu(g}8{C7kg|KlUJ|0mw#f5rR$-(S=El)+^zLpb?jg<*`FQHWCm
zA@sVsI#`c`WN)w_qW4l~ZWa83BG;zfWeMAgZ6zyfg4t4}l}3=I?Cflg;WY_yB}Ykj
z+=cl0QyKz%%L92#`4(BIXI#tHZaH`%VN{j~1YG&@7alpX3gj-YKQ2`IbyixyG4s!d
zg@v=v&a-&lN7xSaQy3v04o0+Sz|b-xXZPgwAGEwYaoArapPSZr`Xf6>B)c#`AUg*_
zpo?+LKby6l2#F!tMN7?XHP}ZwVX?dpgzLP@z28LtLw>uem$6*r5x5<3>;_0+Ci^=^
z=n5RZiYF}>!PDA%dQE)4FL88pJ8S36aSGzZR_H%JuQ8V8;L7ab!@!H_cut;q`)3@&
zdHYT8rA0<Y;+$6RT?FF-M|__mz<bzdX~4^qP`Oafb64f%1&FlW$B(CpxtwZs7wqr}
zl5$B~ijh6wKuC#AaYa2)=H~*Os@b}0mr-m1bP4>yut3F?_1yRYa?dqTr+Tpk0DO+%
zY&t=kHnOr(G0|8PX+GSdY#dO~hnHYNCC=;MPFr0+EzdxPE*ldKPEvkLmNeQJ=zD9#
zca-u<^3jtgUH?Yp?9d;$+uCmaZp>(o|C4n&-npBDk8;)VlL{w^pxpi!pciRSqtt~<
zw+!U*SAF!{bfR(RRU-?xb;k}pW#GW=pmHDD+S(ZFZ(n#7wPz2d=92jR699n<WlMjB
z^dY8<Gt@*f_$@M1@LIH+BG6+IN}_&v_-g~!#}Awwh%ewiTZF{SUMZ>25!+k%4L4KP
z;E)hqnnHrhza%#MuI@x73kzE;@x1{5{qg(%jM_lNRYy#$tdfXUpQncK4^CDKNmqqQ
zL}TDWtcHka5IUSV(Pa6N1uI)r9S;X9(oGe`V~IY4IY<S_0eX@KW2`TF^YR4C3!z4F
zgv^LlW^aK6HNe3NA_^zB(9UMnkfLwQ#(HbbTJQ#}g9+e*>+lpY(_A!WTXTyO5sp96
zr2q$v(<x(*m#=uGy<&i`lv9upC*XAvAl?J+hgpn%{YY9SY2R@V(89637s0DDii*jQ
zdjmDm4@%3*WUQ_s4Quf3((S9?Uu$bBtWrM%uu5dwWTqO;1uOA1!~}q0K%e8-<$E}4
zy^CG|NBqUpr*8n@bgVBBh9~#}e9>)NwnU?CV$|XVs=~p~TBxriYH|GX6)S>}I0V>2
zs^TO@W5V~!wy4~W0D5K<$<(bo2;e~nO&?c;gYZd!Pav6nIB^H8^8=wUNT)qGhB((H
z0o8Eiz=1Oaas&oIv&IaQ0#eL{ycYn#uM2NiKSVs>MM#MCc+dAHgzUUQ?D}!2-qFMN
z>v*+XI#44LChk;SOm8BX+Yk{xO{ABar(xB?#tlN;@R$9=)oa#R1BsBk5wWNN#@nw1
zliGdc!x|9~{E%u|CO<)3Fz^vkLg|4e3a4$}NgM>p#R)7V1P=!e5*QI1x`rS)#Q7BJ
zH6nlO&lwK^k$|DlsyQ!%Q4Fj?<g!d$+gE`fc~IE~j%{FvMt--SI#|9q-8MY4iy(sb
zh5!i-c$$`CL6>Ix2jwJ$r1XX<;{o8rLz9E1T3T0R#=h(e1|Xw*<34iH-4RRXS87Pm
zh)%&m4V)7a5cvDpdd;{EUODITr}gQP(R{6xC7}JBjXbhOvEB#|n4wnYSIL#N%En!I
z)$yE`zf5T#MaVw?e(ruE=wxdeq9(UkA5BJa*XX`9n-BK-<1unUzbl`QM2M1J6VtqQ
z8@7cGLMvSJQ`6JAqQr@$c-PHpCUnG=1`)SNaW=&V*#Kobu;6gPRd++XS7PNcAVE{l
z(fj715N$g6^tkAKwXuHM>gvC;rV)uKTfcT~!cEs}78Y(?&AkZvNOCW>Axx@mMGH7I
zT!Uo3XM-u=$Rth_XI9Bo-g60xMLb2R7J?%iX)XFqzyl;b2XI$i^tqAGMCcsCu#pVm
z^0&i`IRa_8FmwoE*O5zr&w0&0IHF999>g~+3`cxhLMFf>Xh4Zjif~-cLS<H2%mNaT
z22tXO12A(qvJ(V1|JSceXswAj{zDx20heM-x?Zi0Nd5Z!kkrs~Sj4F3sHo;}HfwJ!
zTLJXZb+Zle5RjBpn7{xY$K0h8)(p&5D|$KPbGzl*wY4&=Pk=qq=c;ArQ_Nc+q7ASp
zf(9_%ex4hv6e10xChOzWu69rW#U~A4Oh6ppo`ldh#P>~P62~&hRk%$?Q=D72m>a^;
zqcOs6g2ks`iy6Xq&VXZ%qY|=X0e*g27@JA+`~a_*a?N4Nm}sHJR=+-G?%V0YRv99c
zK3<)5h%JD^XV%npF~r(cITIzZvvCMPJ);KYbLZaSn<{Jufx|vN_V>?Wh~dZ{f<7-u
zi`Wk>7BXsO9>A5vnLD8*&h6W;0{=H;v?9fBJT>oUx=>24BF@gae^)qlDg|zc)KJ!q
z_EY8dIOImpU=o26!(7lwkan+dpSTv8wZrFY^Gg3@50iAN$e<ufA(#tG!T`1*c@mYl
zw{Z&ZaS}snV{jm+SXGACA-p>BbuBR%oQ<-J;8-Hl;4Hm(OhV#k_qI4DbE-s?INXa0
zEzRAAqV3x5cvcq$P#7t6`WT?E3329^Yu*(nHf^fMH{!b+K$wPpULqppc~MbO?=zYO
z<-;4c;|#Q>fx&BfF~H&PMmw|yx~P%c?~L*nk&(lhzhPR;0kGH%g4P~Be!SK~Mhc%l
z@&%y?Grl2W{eZNsY;)I$ufu3OOFWa)&kgI>XFK$&(#LSlTg%NeLCVwY(S~N@OE;%}
z>xa-iCi=__<1xDsaYVWcY;bco?8eFIlBE`sl$<V1B;b1SAqS)*aKHe^iO-zTUcWT1
zdk$i!!6KYEt^oGT?v0pAHk3WV?BFn5hfXI#sF>56W!J>hbg)sL*`&uU<^z70%c580
zNjxxHYiINWBY*iAMo`_R9#AViArQ1Cv4Fo~{yqYI<?7XfMZ40L0|U-{d;|w`$J8pr
zp@?ySnESy<2207!%|=Ct6%k>Vzt8nIIJtDmA}fr9S4Fga<GPI-fq`vuLe|JLE;p@w
z76T!eXZ&se-ewRPEsj=Ead(LMF0aFi5`rylr0pAx-1K#<7VRXy#R*U13rUJRrme7K
zmr9BX*FDO#{&kco*0y!0+zvnC@5Koh4EH~b-SgJZ;JVf6A8%_nCBDc!e0_83so3o&
z*Pq<IZB<RjGCq|v2mTa@x6#P2k>@-oaJuIB_k<5_+iT{FgHCkPI$gQb+ing8N_^|M
zDcREdgg)L==2^tnvvJgSKU@Q9rv!t+SXz)bx-q?w$EW#o#*$Z)J#DB6pQ?zMSasiA
zB-R>^L>~Y*vC)}+mHNuFcto&+Rp}x;EP5v(tQCp0*f|Ng==O?lGJC%LN|<VVxx-Zs
zJEZ{zLmSHsGp@F-u2^#TJhtM&#gJ-v$4`bOxK7M^wi_dQ3<ppb0+sGyU}$(Ke2bH0
zILEGCFjX64Xp@tZ`NsJY+Q!DlS;ggYANn1Wl9JAAYipwhtE2W2nQ(mQ$nOvtV@>!=
z<>J1+zPthgDpV@9gfe8wJ%=!(!`RP>si~?t?w6(?J)RO@paJ<*KHeN&*U%96?b|n{
zz*Q?(_Lbh>kTvIP{djcLUPM$hDI;SvgeI`v50fMDr4~dkm?A4H%M5hQ&B@94s%955
zNk~bt>ACBh@lsU(BMvQ4o11^z7iSd~8X}MC<VJb(=0kS|j6a~oZ>~?a`*nwu4_y!z
z6^)CFI|ximW8`*yW!8=0i!wNN?6~3f9XjQz2$nwi$nM>jbf#~)Y#KiEnMTvaVQ9SV
zS|E=W*`;}$t$0cgdWa_BAl2r9Z_Tij`@U({t`uyt3fo+d=D)sj;#yjmETvZt9z59W
z>+0#b5J<y^LO-Jlw?Y!%aFP9a56GZWQfev{nARdgFV0Jb5@`9{2ZtR-Ot;(W@o&cg
z1J!vk{sW1qZk=jT_#pb>eM-8Gb;Xp^(oKazzsG^#Bor67cKhsyNa2GTg43w(>^yac
zc~%eS<FU09GBYo-va+H_Kb@POZ!vKg`E?Qjg92&h{GZH7Xi_*cy<_8fp0k-ycG2)W
zLwt65dU|e!ro=`+4E+YYYg_*KexU2p29W>nUrS$BRGbf&aJYKDH~7U1Vke>NhfU1j
z+eNf=b>kx<xPRJ+PEJhpAqkLakZMrunNNJd3!I_cd0&2E9E>1<GlHQpJo5?o;-|CD
zg~rGja&U14rKhJ8r%MRSik`*n8VKyd$|Q6h=x_*a`#;=U(NIQ6>h%x-!#JQ#BAwga
zDtvUy=;s*1LD$n0?=ELqvwAh5OM`>A=Q{MEr=zvLPBs%fj1c|fhY!^a44Sgjy+(fB
zE+{NC=7t*5c67`}SkovX3sE%GL3kGyO~Xj!jf<=qzwGL&2S<xIJ+o_Ld3KuD;#>}d
zm`E&OA5URR8vIIkX(4ba_zU3=Hh)<85xgIfzJ9c=zrO*jtyrdWUfN^o0(2ur)rRa&
zxmQ(HsyW%&-{xdDXe%h(#TU`YRD~$Zi8e`kFXSOoe<hcGZLzhpv$Mcwskqp4_V}QO
z8S@w^^PI%4jZV(a6<4zoi;GP}L_`E#Si{obKoRv$(uYaGkcHMC+t}E6B(-C3cv#ok
zIypkx?QKDUq?4oLx34dIn&A56!1G%Y{AfR^Y5RO92ggMcDb=<)?+s>rwfUmt<YaA4
z&1hINL<~Vf!tMO{&y5ch(oKqLQOK$F^)bG_t4m2od3v=PHVN%TPENCFOIausW4kjO
zslVLZ5|75eSQvF*zC1)R5Ed4W!G27PkH2P;%fgC$99%poQ;QsfZP+&80M<4(7s|@Y
zGU<$gp`nO|DrY!?gaE7TL~XS7>63qk3^!v+0<z>LmUDhlImAUaC_Kc14p<(J-{9Ev
zUzjg&t|T4h!#^M(DLq{+JUrYT>OM@+U~F`hm=kc&hCi9gi+_E5eEj|W`?34xM%wQd
z6ciBmoU%D2ml)*Wq_lE>J~hTZ%Jg41x_Ucj$Kr#Mm*d2DyAfacHu`_V9~R5$EA*~Q
zN=gbJJJ!xHW0{cy;ArJ8y`w<!*|KwIKVuKeEkzuDzIFQ{@x^z<_u>6d{=i)A_$fOn
Vhj#cQzS@sP>5S^>v{RSv{5PwYeqjIr

literal 17031
zcmeI4XH=8v+V6voZ5$LuWTXx%D59VOf^-WiDxlIt2oaIq!B7GOEF+2vDgr7^rME<C
zD1kUqB1VbS0D%BTq=XiFNqMi#DId<->+F3#oVDL`UY4w7fRH@TeP8AO|6Mmcykv4#
z_*aQv5eS5^(Ye!C5C|b}1Y+&6jqBkn$9Jz;1OKhRapCM~1Yhv)O?6f*0`UjJ==2Hm
zprrBMuv<Ik-tyVh`|_JM8U0ZwvSxO-sa(Y6O>6F~IWKY|d;jTkQ<fR?Q5eIFh{)yG
zwE9TOllYy_9$9S877==@=cBxT>*$Fc83#5#IO4MZfV^<gOOcIxZ(_q|KUj3W3j5@B
ztnUX`%b7pi6zW{_s=q3+tIImU(K;T1P_Pyg8`*_GtT{^*`e0BeiTLgGKY8F4Zi!jw
z4Mfj;)rbsHMtn<`-@Dgkl2MH<U|A9Ka&u?73w<<KxPRwOn>Jzb6jxEiVG;Iv#M*7%
z|4ARH=LX4Y2V9JI?9yw9mX;h*e3vriE%XGL7Nzl})TMbhxid2(!w?f?aQEatUYm(?
zrXOzVhRnC68^ykejZKS|*3LCfJT@Jq(@hE|F|aG1hwO+~<>lq=yA}o$P&lJl1=s!$
zx6hlJTIE?5m|we=izo4};j8CgTjpPV+tSkVeoS)yd2DQZb;v@huGX#LH=C8q4H_F8
zqn<pmxOnlgo12@togFSJ>eul;5=uL8qHTVhKKC=xd$hCA;TiA9o#*lKnMp}f$l$@a
z&rMC1=g&vMtBHw<rqxDn&CAcfs2wo&t>KAeDVB?o)(h42@E_?a&hjF01B9iu%#oG1
zF5B8B@s_CJqa+@SG&|dB8tF-z`#fD+iDn%%66X1ibsiz7I*fG|Ml;XG$c$_9=IP#q
zz1FU-t{tz{-9u#vxw!}2=f}FRUwiw)7KWQ^yDQOo)Ubqv1e=8({}?ebu|fa7a1K*e
zn$TVq=uKu9Tr2gHBHe6!KGVP%3g=}}Q{_Yo3k&7A1d*{c9IU5P_0-4Bmh>`D&jsJP
zb}MZ!htakiELB5Gt6K{*n>}8}2%Rxbr4P5~C6wN5+|FvtF~jnc;^N{gPn@_N_x$<x
z7QWt%8#nToW`=CCOjGA88M6FE@`J$7BhQ{b?VpshZFG%AAOpG_yDR*&yvhgGr?MLA
z>V}`4m1*WI&73Nlw6DGIx-vg5q3QF6iWc3zT`im!TD|!FQB>4-nn%>)Vvuo)j&;cV
zSaT`Oj)hMCnyBHvFwt8^U;OHgWiIe=)$`XN0J@Y>#Gu0QqY0FZk=87Ca<p{iV7yHm
zG1#B9ASo`MCHNvWMOQ-2<!K)?sSd8-i3a1F?m~~liZ^4>@6VekVIjW7hCj5mZ4!@p
zUV%U`U`|vF$voMmnrX5!(MKW))}hp+>*yQbSOw!2f2yukR+vq^MHYLbE&N0YX3ovg
z**OGRbN#P1^g@WQc^1E4ek3!|s|XpeIAvdetm<GDWoPdf+qpCM%^L%+z$s^xnVDH>
z@Ql+FS$(}>&m{I2RqF6-cU;a)z{L~9wTB%)?|>5Ui+aFBPX)SrVVuy1+j;03d_Mi7
z5zEcE9vO7@48&8sGIDgT{rd%#>b)^!2knkjt3KG_<w2Ej*Q(H^=|L^@=nI<Tx^?Sb
zAGW_k4O`r!rFHE>g0h9BrR6vJ@9J(Zuv`{u#Wu!@vGJ;tQ|9{Leitd^@~M1I%nb*J
zyAl_OYXg6L+E&e(rc}-}X{!fKT}O{~UN$l^@~i5Ndi2OF%OqLNt!=;V%2=^h@O1ik
z0A_A<a8B<aIYyR-;!TUweKmS-#WF`kj3kcxr86IGt1?sAb2hrI#I;q5fV_G&`>(qj
zloS<>4qYqxCZAyIH@|ZuqprT*wLRD31Stwnc_eDKZSoIub90xUUtW#&R20KyK>`OJ
zvMv8MR1V?O9^ALGFiG<y^kzc3Dk&(O8K)Ee{*=42pdIZ_FDsayp1#xV?CI%wiTV27
zR=q_72XeH9n_DiA+s8*f+IgrY{*cXVZ(v^zk!TB1-0@I!-|Pa3k7RWew7yU@y9Jp+
zulX1^X6mUk^F_6clvrM_*&90ZHQFDI86-N8w~TidGKoV-<pi0q*~{c{-cJqEWh}O)
z%%khw^t7Lag@x--Q?g+F&_l_A-O(&Ybrtpa3Nty7(Y(lJ#-*g-UcUStJ=$hedtbQK
z%v3i+utLEr!uR#P-Tg`%`RpK_4nuL|4>U)Ws>9R_m>X?}kia3U@_+oeiD%MFTph-{
zipjp54YHyA=zM&hab>{xXi(3citZmh<adUTqzKRA<1O+0ouey-ZtVx?6<FvsWlXzp
zE}HHOW$V4PxVZRG(Bm=ORj7x)IyeZCxus&S&i8)>a2VAgcX^YwMF-tG@~yY=ksX77
zwJX@qeptKJg@!}gB`j59OHM^@fh)p?bKU7yTl6_ocX>>GUac?3F#(xbUS1xtKv9~U
zQc_kfLw3}Jg~;#Q7lQnGOTM*U9F>pdE(aq6deJmRJv|1iXmr%=5UQq!RdhM+OxJ|z
z^AyaeB6;w62_zrgQ5Q3N2_i1ktJbT^REK@~l7sPS;#wbwQ1XSWsHi9@K4%WSrKK?|
z6`LEQ&+GI}F0diELF$)Ldc6qBRbxdF3!c7B+WQmiFgZ?NUuQT{;sy0*d7^6a2RrB%
zRKFRoU~Tt~gJjRH`jFO4<K^#<R6NLCT-nN23kC0*rOfdSE=eX_D1jq0RDA*x+hBzX
z8!Xp(+4?BnafHt$@x`Y!W4BErJbKD6x3{;KQ4Uf3h-*Gf#6{~f!uesOj+OqqTcr|E
zK?8pZO(q7d;H`_DD#y#{vtKp(hFeYb+nF)?I1VRtZk%ofkt}FsYbs@yA9obk(3Slw
zJ$$~`Jshf5+$AfEXW^=v0v@++exZ!4>{PDmzq46tvJv~U5Y4ird!FNC{Pi!%Vi(_1
z7s4bYze^*sww_7y{a&|;?uf3qI5HW|uR^hwmN$4T{JI65@bYE$9E-+M317N!iN9}T
zuYy8DK-VLcn(VZ*F&N}dcsCA*ld7xX^yPOt#bm3rRysVL-t<;k#egGd1XC~JVk7Mp
zgeBQqBJ?NCwX2WfAASCOp46d@<&-n0uj9=ETk@@opL(~RhFiK~Vq#J~Nhm66VYg3L
zWSJ!4js`CcCRmu6G11oW{X2=*Bf3jmMW<eRleouzQ+4fp5Qs>^k#+ajdQU8^M3HHd
z&!OK5P10$lgoh6w3V9b7RzSVoy*yI0Zu8Dg9-Y;iMeQ<mQ`oz=6FX-2JpQ2djjAsb
z?P1yL{I4m#45wx8VccINgs%{u=8Jzp<7)J}UOjuZZR(h0B8od|6-|ke(anvN!JHFB
z%63svcr*C~3VNR1=@8TI%I^!EdBR%==hFg~UYXiL9?g}>b}TP4W2F10^q>y*g$s2M
zGJ_0vu#eS=S=kQIFf|jd1LHeOYk2Rb&?!1QWc7~rK9xJ=k&vtte8tMv_fJ9TfI4yt
z-e{<LlWl_1vnHQPkHh-<C^nlNZS_jCdQ`7=WG5M#;Zhf0ZrzH{G6lhp43d0jnyRO#
zG9w8s6tghlSw+muY+y-h>+=@Vm;ox4!@KE@paWVC5qiFBGGw8?1=z|+OCyKE?p>h7
zACmU|y$qTV*{bBZ2cEp8?!K^ON#LX%p0XX9G4!A(zaRO?ng-zkool&=4D!Y^!O_X-
zfUa&YsqtOY1HuWrpn-Mt3k}x@UIosnVI964D8tn?Z-a-w$&r0YOJBW-cH{j4RdXLV
zs|?@Stn4E-xfO*n_*Uvp)0*648``VdKnq!acfIhhMnw4`%6i$5u_JW8Xy|w;jiZ%S
z?bWMAcf5T0vL4=wAPQ0VO4Zp7B+!W`-cj)0%5ZyO(~bN7ZudB(wz@aNcD%c^6_1@$
z;*NWV`-iq4YMk$QpNd`1VV39TAFQ71u&IP*CMGGF2N~r)JKT)na~IIeCmQRyGn^NA
zf42WnlLZOOy|aAJrCj>BU)}t#!Xv0Y%*_S6(@@2^n0gjo92MF;ucDYm5l+TgF69VI
z9@E5%3_%U)2wUhKqLrnGSEHsOP;%1KPs7UzF9uMIU+c@h8)7K%{eIH5l4MZWrHt{Q
ztPu0XkjeYkr9q4my%Uem_*C^;T3cIp&W}^8jYp?vAk4lB@ohGCO!QV&H}%m7l49cG
zHlq7XO1cT&-iBw-+JLMqbZSa+6<jHvYv)@phUEwLv+UG|mO(vomm4w+uvUU9Fvmji
zyFY*VQgW@#!?mK<iv@e}{-~;I0h>v-G1Xg2pCgdC<3!)hI}dK8hbJoNtf?Su>a*`U
z-Ginfb8~a2EAulm&#2TazN6AZ@3)qD(#=7@WJ#`^ZJ8dVNcSC_=q`0XhqBkUet-RT
zrb+VXLgYugP<DL;uH$M|K*3LHDn7?7s~e|!{X@h9#+nMn{*s4sf5*<^@^k`;V+h^F
z*LJ9JU+OK63n`Y*o8)hC>oGkZ=Kg3?O31`<$cuIA9Za@MeUzA7BBds5MT%;iN=ePT
z*o1~1bRYa2ClqsPSj7932ffUg#2Hjs%X4BweP1Hd-|XcwW%=KJG{#C(dsFht$}Cl#
zn||v`?h&lqm_}Br7W%wH!{f`a^ae~FnHYhrgmR{mKEqxJG3$e}P1cs^&M7EpuX`w}
z?%AzIPArrQh4?$>(fJ1Y*AlVn%`DzDsW)V@dbu5s9h^BzRJ98kSLec=wSmG3!&Q0r
z=^_2Usvp?Df6ATC+pnzLp-!_+Tf1S)J&(k>obvLkU0q$>MUIb9EYboaWRy<8qexVP
zYVJH8%as@Tk~g7xkkUd7o_P@$mm$crp@LmwQo0y(B`X}b7Q1g&d1{7EDc3zd&QJ8_
z39Vf}9HoT2WM`MMev4H0?CdR2$vGt@=BRLPso=t@r#~wlICMxjXaz)$8NsXf7Ud9O
zxRQl4v$TxAa^(u_ZQ5kbiV0|gOZN6}JMbvk_?7a|#evAtj)Fu#e}8oleRCtNXFcg<
z7AH^M*)~*5;?CO$%jjGM4Pz#FUtx)Kc;EVlhFq{l;0p9-86lGbUAH`(7W47gR!UI`
zsAY1<7g>F*&fY9NXy1Z@^Yyj*1%V9uaC>b#9;Kg=k&%ODb)3I=(HxB~J)opC$PO~U
za^;=C+EoRNw?Gn?hAd1dENyDkni=vmHp3k|veazOFdn)b9&_*BFD$ZbcnMcykV;@o
zW%K)9k*pGZx_@y=KXBlH2g&?IE?rUdv<!Q_M9$cl2YRBXi&X^8ta@dlYEE13-VXSN
zVita#O^M64zNa@ezYDAF2BCT>RWEE7bb%XKH9J~PetsKB0TfmBRJJ;YFe#*Yt;FTo
zL^*Q{i53pE>24)nTVkhO@LJ<@P{+>cZF}@-55quo0OUrivBo;Cy@cAqI!Tu`HH+Uj
zgHdc_2jC8G9wJIUfC{-BCa(D*0rGt~mCwo2Tlyhdy*!$S0YO?j{qXAPXv#Db1DQ>8
z|MBVB+!0xXOA(Z92!{!3Zl0#+8pQhD48-+6z6<|9{6GLy@NrOp%)zqA&IpbEs2ECV
zO*-!16v9i3(Ypdcu>0_y58<%3ZhbY?`gu6Nm10W|k9@6wv8x1Yp}98}d6ssmra%QE
zw@8l$A2V2l0Ql!X=3jv}|8<@NECw+(I5kj<0SCjm76_ftE)@lJ9Yh+_KNOa`;qflf
zZ=8s8KGyM$hM4UUu^|2+c}IYq*!Nj^dEqQ^w!7xJ2y{tQtAfkacP5iH-aZBOl;x;z
zfZXQ&yPS%O%Gk=e^XFT#FQ<L1k^N)G1C7T%AOn`sO_j<ZJhD1FJKs;N-6-~5c#j%1
zdJRU9!_|e$viOPHm*g0S&uB8c%RFt6=ynTxx}Jzzon*uBM{hoW)+!S3X_};Y87j+W
zO4)6nKj69M<}bV|M>5}gB$_E47G~eS{Bc`XQ?hm*iiDNxQx-9)lRPs5CAczhQl`c8
z=g<%ZD8a2ngL1geeQo%frR!TRo+jESU20f?I$(*nt3Fw#`=&+&zPo47sj%hQO^szi
zY1=b*GY((b|BmHXp=ST_p2(N=+2QcQpQ%(UbMs!_(iRD|bIDrhU%%dqz?>o43(BQ{
zEup*DXI;F0^t-S_yL^oo!G_uJWb=SW#IhmLKEJ2O7P|fh^<(??2}-Xp`*FnbDfl}0
zKw()uo$SJ-qqj$3-F9BQ=iuOg=FZU~?T;C7VR^b?Bif4`DG{(ys%OB!+rQp+Nnh|(
zP*u%oXE&(bkEu;N0*hn!N0tdF-ldfDYQMrFN`i0i>+?R2<pkAaZGW<ceV@H%33~x|
z^n}9+V)On*(ZRMhb0}PDoLHezA@-TH!#6*PlzR6ue3OYpor>NG%<F9;c6E|ZPQs#7
zwSV#Z{L7O~@U(#)&Vtty4nLK$M^)7f`unq#^>=;*RDlWbXUsS{ItsYkFHZL^eWFmz
zPMkQAp4I$DWYRwrx_w>Mw%qLO{sp)9;FM2eeih!5WsLaOK<mFPXMaS`{vwhD=t1Cf
zeUo3mmawq0%B!rr27CaRCVW6cZrA7wO$%F@7nsuSLWjuQygZicPsGXFK^iw86yQV9
ze{S4(cg4)C8fRr~okK%~BOi&YbU=czg5wc){&E%Fucv3Xw6xUy-u^Dgv`ij%<t*kr
zBJ>Z&;wyOU4jq4)I~lH$KoJ`mLA*af^i2UPJn}YDrX@;D=|_OSld#K!2O1uOKYzM4
zK9`4GL8v?k+=H$a*7!8&*Er~x6Fl?tys@zb@S^FNnNpvDliaD=tyoXzmUN>^Pz@h-
zY!J7<dw*AW1Bwy$5xRD30t%b+@#D#_FOQrG?D3713Fr8NG|-9;|Nd~_>@o|Vm!7`2
zKP)RFLlGj|$}TvgsJIw><vR%Mdf;_Xc?7Ql$g;$jd8PnGvWwg*yJem{o8wGPd9CL9
z+|SRit<d3YscY*QnV7QzB1HEl71Tcx-!P}z(9qzvI5i-l7g~;)AIkx5a^T33%ZF^r
z+5ik%0P+P$!v*gKTcG9BA94QtdEiayK%TP!^bB86muz7h>K0F_0(p;5U(eSJ7<<X2
z9)b2kDvOPabClKz`ZnYnJd%;H<~0gRLB9Nj_^UuErdK+xEG@4AdlR5y)XML=m=fl^
zfCpcA9v2_)RA61)FvJ74pWjvNyuwvI19E<=&Qt^_91d?+bufsnwLC73%=SDN(-^yt
zHuAg1Nu`Gi$6sF~%3A~60vwzxMX*95Ju|YhR(j$ogGHI9sa_CO8@Ss04;?B63HHMu
z#j9H>q9_-izi?q~Hy@~L*o3y6++W=QN5APaP1kLoT-K=(9yvjz7C-lAd^;-}mJal!
z+gF-E7gz&r99Z8B*g3HsJK9DL#6hlmGKgJ6ywJv`rjAVGL=XL>n?oSbn_nJrkfOcP
z53j*hj^<WgRJkrZzPP*SfyQ1Vfx9qC)*38t0?ErPCjtN!Rk*mZG4x$pTIsEkD-a^<
zs-GyDW$u@glglmy$5e{!NISc=7sDzz(R3i_f4&g(lfOMS&3F07;HPJM{+8nGKmYtw
zLMQ0`^XJdo-WhI2`ZdZ61lU2!bR`x5sb#@6v1jgIRaI5huYW+`DKLw0*Pi(H_LP9f
z2}ojneSPD2fYR~f*1eVJR=5i_z(0VEOn~tSea$MAM%^<=H|<eVGiE0(yds0<7@b)J
zK7!HRx&m%RGg^D*=UW0<I<RLzbH2mIdrqr~c6?IxGlf#HOV#PBxLwd&m=QRDi2aKB
z`N}yoI%Mjv4b>BYkXmI9>$l2`(F(m7#0vErbsJeQ#wde5oGAB7kM=%=@o8GoA=(Gy
zGNmSAbryB>cJ10V^n?P0PO2EzEp>$fDL%dkycr|SG$rcF#GsoE#e=0MmHP9n@bg*|
zfVtmGw^sNMJ>}!}phG2F?gphsY8+&TbC&v}3G{E#66&9QzW{L@^sfRKQ2h6@1Bh%K
zIz|lczrCTvou-VZgjO5T-=5ll7qqR;=%rd&-W#YZ%laqu)qXG^eUf8nDs@U;1{eG#
z036gSJwC@jmA*Kdhrh#!OojsO!v0=gMggWT)t7F3{`~YZ2i%(n*%qjYR?yTNAlx~S
z@rM3Czcj<Br!C7=7WQOGe<{@^n=JmUURz8v59U64H1_-%n*?Q~2aN=%%cJk`=F?72
zP9GeOdG!o0HX2fjfZ%zc6zO4I^)OLU3z_$zkDoC#R8&%85tWXEdm2g&$APyp6^w)!
zVGBr7wDBq3ng<8%f|>a!&UbM=+VP;Nekv~zhCx*C*-d}Gc<~~FJjSS|4{7Z)NqUcT
zT1fZ=cMj}2Gh&G8S8<D7FYXS^5R$1FV7Q<{YDT+$f{QjJ`gQ=`kxEF2i%SlyA%F%Q
zNNTa>aV&grup*1L`#M9SdcfTuP*&dJHnt0Fd^5~=q)_!5f+gL}c)AyoI?ubWCYlvg
zwS_a>oHp>fg$lAqr~k-4)0Ei?6X4m0TbtjWihPlrT%gY#F%?1;hvCl31lP^Kp1vO%
zy3M*y@_9Oh;38-1VgZ4nHCYk6&$Q#XT_&{7ZvZN?KyUO0&^@o!gjUQ?_Q|##AMWGL
z6YNsGKa<6qf@-%5rZN+iEM+npCL)3v2c818u?<E`cnUBWhpPO-{VFQ7=kgbtu)c<n
zYe$bBZL9Ei1|MY&G1&n{yB(km-P1N~4nXMZW%i$ebyW|mW?L?lI(?M@A)}2W+^CP5
zVdU#LfXXQoYW(uW9D}JG?`#+bFupuiOr7m=PE8JL8V_J)B-pipT&#{ZGBL3X4JFd7
zx};D+wXU%W2cX}zW|`6=WwFCEr{tlZVus&FdL-a&2u%UjpgN$viOotmHzkcM-cM14
zo0%Qb=Xd!^SFt`EubH|Zq#*>2RZ&rKv^gyjh6N?i(WEtf44`sHJ$%SM&I=^1Br`R>
zcmNJ(-`yS97ZzxM%7}mV>@8U)<RUoh*V=(z9|*UE2_Vk|22BDgJ(YA7n+s^Ji~{fH
zp+C^w7Old8t35L+pycPuiUDZDIE(5lO(~iMBs>cIceb);(9e}Y3>FM)Qo?z`VNfar
z90zVG1Dvjsyu4vGX09bdXsz0<;Y%=bio7{709q~0m)YRjQ|_w;tW_{|fZiq;9)ro%
z3Nadl<`MVw>6^!oe}k4j3NE~~_a`(#4}7Sr8(Ewu$9P`~LjlsHO?ANdwXdoAD46P$
zBAq`(2$dz&y@Cie*>mRcw+GwyjFw1eg~XgQUReAnlV$9a0*mh6lyqEhwQ!RhnZW-J
zV(gQwnI$0$UWcx}|1gA+LUDGm8>j*-1GuAYtgCo!ja>ym81S`%IZ!B8X)DhWAblkS
z8;57GjV$gg;+<VobOkE==f=insEQLalIWzX@fn~zAfGQ#^;YtuW%Xs+0-^!a#wl4H
z0P`@MrzlYbtz4HgRV#7$y2kZj{Qo!$5V1u{vwuM|TBWH!DR7UajZGem8K6<JAfW!6
z_5z#m6oPwpj)rm<3<O|0d$)!hnf&hEZsS)+4U3(=w!F8mRrBdL0K07pBU!iJ%2HT)
z3na1toKCP#O7Torse3}1S8r}_Rp170^PG<jB$j~rqsO}~ptlGxOIxYCRYqp!0XeyW
zu{3~FT4Bp2AZ*aH!<T{Vu>iP#gPsbDq5xaIPB*e4G;}~VK`d&;=-6(++|e5J4VU-|
zrYZT*R^(b*UCvqh{ueZ};6~rP?WzjIyScmX(G~t><R2k{dnciz?a}>nePmFoKvF+T
zSu?uI0{-z)i~l22w8{cjS->g__!rN0{o~A9WdW-!V3h@|vVc_<u*w2fS->g_SY-kK
zcUi!y|NsAL;Jxbq|DD^p>i<{$|Em9A_5Z790anifteyo}JqxgU7GU)(!0K6m)w2Nq
ziO&N3c4uA=ej9(U;O_#gvVec5dHl!ouz&W~5&kzn^}nzHgnly|az1DJ_AfcckV{_;
loMuEI^#0{T60>|E>c@c&k&98K@RubJMusM*aVM|e`49g4#<u_f

diff --git a/vignettes/functional_api/unnamed-chunk-11-1.png b/vignettes/functional_api/unnamed-chunk-11-1.png
index 2e9ecf2cfe0e2fb5f0e94e9831dbabf71a1ce7e0..42e568aa6f776d04ea8641d02d8a3e2758a53d45 100644
GIT binary patch
literal 92394
zcmeFZS5%eTwl%oaZ4PX+sDJ?^gAzot0TDz&at4(o8Oa&90TdM!36fQk<S0=<MMSbB
zi3?PcWF<<jKH2v^)I<GM54WmSYyaW4dkzO-t#8da<`{kS-p71!MfSqhP5U=dD3q<z
zQW6Rj%3rzo|DPMy;V1jponGR<)?JsmAVFCs|M#LW?iq!0h$1a<M#(X3nCkSrdTM$3
zciHo`&(aUHDzpZ@XVhmX`0U4ITIY55>|GTGzfI~U{(m!M1WIJc-1$tO;miNXjn?qE
z;fem43*r;uyf&Rip|oY1GwpVTsg+K}+5N4teS)c_mDkhN{=!@1>$%b~x%K~kONm~)
zk?Y@YpYQ)4{w~BfZVlzv7~i!Ie_eil{C&eUyw4%J01kD9YuBztY+LiL_;0);{r26v
zV{H0soP@UE-=m|3TQgmR{-P`^aLE7rxVPe;ep|iri2wPHLivAl4Q3sMu+H?#=e>8y
zr(ZAKQfW!ekI^o$7kzjC8|$i7s}i*GGN?kK0+uRg&wg0ry@5jcR(JOHX3DKC^)Jt@
zcUzk6_w4mpoC;!7PMD+|y`7n$lFa7e;Sp-!FxpY)Lno{?n5q=7pi%5zNEOPn?yY}(
zKtOA7FWz$b(EGI%%99V+d*uWr)=-1*-@mhI=TENj!Fy1q)7Fz;!q=t8fBi4NyWF>m
z;`s5;{CuL-_YW1%n0=2NJ-U1Q_LrJD7NSqmGxf_~-Cnzy?fCI40u~)#@@)Ed(a|lQ
z{^V(sVNe<S)@g#xs3xo;OfYj{VPV(Kol!r3-k`p9Qryd}GiiH&fXk+@sv=Y%)w;hX
z`PMB8HTUK@)fBC|H>S=0&!7J}-OZx&HrKkZD*p9r7CK@3OAce5Zdbbz|13xG+1{OF
zlZz19v}qF$2MrBP^S#Y`85f^54t5q6<l7Fs+RtyI)+8e(_2p}_Cci<&E>5+y(|aD4
zesAgQ>`c%u$i6lkBNJr%=hro9zsShQeEX3EF7134Mn;KXp2e=eZayr1#i^NnW3(eL
zM9eLJ$BrEbeMY*wqcq-_&<5(}S)X4|L!Z@t;rEYH<y@<4CFADp+3HMtDgxN!TxZ9b
zM5kwF1k8rU#!{a=`MYEsPe0OG*lyuUb2ddQuko#u4bCgi0Gs@)yI3T{9BOr_z#Gdh
zvBlqI3?ZT}_Vx2-4~vV7vC}4M?tN9X;UZ4;1J7mrSzeyKw~00od5po^*ViE-wLii|
zcBDN=b)@ah?t=$87$O^&+Yivv#!n43oAy=&ga|vNT-VbJA8Zd8^J3&Tu9sM|ew)(b
z+*I$-P^hg(ZG@<EU$9Zw{+$%csjI6fH2cI{bI<SC+kDh*E4@g|=*-oz&O-b4;$I&&
z+qcgQw>BIpuTQ&nuCg`LD12^oak||y5PRm8Zd4gqZ6S_q^F}H|WTJ{U@YgS^xKB$%
zsl^-tY|7TB>}$dVN2W7x)E)71bNyYq*M8n{yqr0CkXs|`bA11|M_9WU@BUV!NV^s;
z1k1qO?B<OdrGxr2EvP~=I5K8F-$zEw1#S8s$)w&Z9I6#x<{KXRSr;WcC7-GzS{Zve
zRLebL`}XaDI)%=H{GORzz8vZqYJJtAMryTnwG=KcULAu9|C_V(>nQ{Oh@xy9Pwy;r
zF|7{aw>M&EQZ06OOV%mWKv)>l7hEgxY|Ao!b}^7+KxuK)ZnhgGo?GpW7QTINcyVS8
zmRjy|2%ogF{aB}{+eEd1r+K1sqM^f{>(-^-yEBDDjlU+T+qb*8xR^G+mY*BR=^Cz~
z@F~2$a`<w%kaCyDav_!0wPyVH@4EUJnWj7&qY@z<9UWtOl#PWu<PaVsKO`@#BC|SN
zK1y=y*8O}em-T`;HC4oy7cwg!+&U;|eFMAArjhl!T~9Mi?fQesUW6Hg$AUaITt@k%
zc^hBBywucSQ)T=KA={r<)dHVCe~leG5a{CMWQ?7d`*w$>K8$bh%gZ}>TD>>^U+5y;
zri1qjScJujAr($YS*@a6RlqCQ6pVa(f2Uz`v#XU=8okT#iJZ2+&ri=a#K}3hnC;Lq
z-iaE6t;XZXAbov}^jHec3JMiPPTMbRZ<a9MoTA-;)Uc)9Zlq0FJ=4&xrMU2~=G5Y)
zg!}h5l#D+;dLcnEPS$>2?$V{km*>2)3@Z18i@CK~P;tmu`S=>$-`W4Qe!XyjgyA}@
z&43qKA>aFWd(-x8b0LS(u#)_L{`toQ`<4IL?fZwdh@;E%(<3$}JAZj@J&>TBsABQ>
z{OwA9)l{9Ps$kxx*vkiYb809FTK8NTsE<*OxwyM2MO(90Y(8nz9*);|Qj^b54(Geg
zXJ6ArY-Hy3e9_@yK|q>x6%|tV$po;)ARzVywrfW0+qEm2M(oeKw2tfPHx~Z<u80&b
zCTBl{-*m8?Ia2A3Nt5ljcdJV53teXHg&$~To3$$Ovg5oZB7}xMKRudn)9<6Xhytt^
z_G$y=R{03EK()|$s=g*%m}!xhJsS04ZnR)*xF#%KJu^0tQ!_9&o#x=F7h-O6HzxY3
zrDrmXYEurM^E{=+^)*#joKri$vG>FnwV+=mY5Jgu;}xfgzRS*9Z%mX+nz@c0lN7ce
zK0ZA3<>k2n`bgmvox%k2IwV$u{_b?Z%Asb+7C4N3wJcsVR2!S^k8ECAoKGGlPlcU5
z*)gAhHM9@f*A`l6*Zpbh;=)9&XK$3`<BWO1p@ul$SlRH9!6nopr8rqSWI9bbvDxVA
z>=G}U=DDe%t@HMuPzNTFwb(^OL<S;)f`Ss1;$PvP;^Hk)p&l8{<gWi^S<33Iwd~0r
zB`2Sy#u11=*0k>Vd@XbF)~Yqb+1hVR0=~ES$Uex#zdd7?(?G=-nV!hO|H8(_n)LtN
zRXWey*jk_^dj3SOUC-z3CD|8z4&I#2{PsX`SA#XGN7eBfgUY~GvxXu!mx9_<v(^mH
zZn{%e@6xC}<<GXZ?`FC3ZuI$3bE?B1EnTPH-34xctzDZpwU~ytnCWz>?H*&=+kB_!
z)cNhV;?g9xQz$#}yO9>1Z;dm}zJIu9-@b0$x>)03hb}kO=2Tq)j>zNJZu`)Sn&kMR
z66K0mT)+OR()=o?W#?N|a{maYK3WoehYufC2`Z%3Eq%B*$H}AIkI8<foTtsK2v2uX
zAhp=CuPWHnz~|)6$dl$FRBLICBG)|XIHFr<xFoR4ef}~zx|-&zD7S{xNFYgvpSpQ=
z=xZ`BN3qrS+ZJ6#=DD=4J_iLm=y5RG?SmdmFEhS!|MPwQ)&nQ2(qs{*SxC!4pR&`|
z2UVipBp}oh&fHlCoREdOJ7AAe*PG!ae)#Yi#n?-AujQl60*e>ME1mnp9d;c!U~1RZ
zIsWmS*ES~3!Z$G%)O>rrsqt|IDXIHZA-`RF_cDu%i-#KEGlrbH`Km-rVLz|oqe^om
znfZT{%mD!lTebdAj}LfmOhL}h6`d|}o5$bvr*)=i7YqmP<xu<5n=Wl7m}=6LuwUH$
z?e|W#(X^n0p;!$pLM%>o99w0$a8BDB(>TPuI@Mz!CTPoirhto#9}cU2j!3WC=<&A5
z$_Sv9R%#j0aI|<J%W5PzpTK45i~O3$3bo!k*2OyqacO60TJRm%VB8GYW!{#>s{KsU
zucupYVWy5<HRb30Sg`<yT;8R0lwh(t)%hDYZX^Yks^-VA<%xU;t={s|B1g<x%C}Vn
zoZOP89VraZk_<b=QXb!`>DT>NsBq@z;qd1Uk#*LY9uHCYO{Rxi>7D4#o;$a-a<DNz
zN+ISVlW4Ypiz#yNww>|D!S{Z0IQXRnaqHIHlER{i67c>mnqZNy<*(Hlk=5mboS!~@
z%H=yUG&GbZ3$UU+IwQ~HJb5F|PqV)!Tv0+oBH?oHojZ5J-F+3?3SDxLiLN`|dS{wi
z82kRtdYOyu;}X*LG=fsx!900OGvBaE2B#S*ly~$ywD6w><woD5V<M9O3W^fVURYeD
z^Et3<S5I0^6r0QRut|ON1F8_RQNyqAC3ZhPZVc_!%r=w9!&7|@BYhu#-;}H=P`JD_
z&!(CZcF~{ZT4qgCqAFd4g=2AqXw9%<|NYF&-3VQyl~#nyWo{r&`K{wkI&t?7i$ZK~
z8>hJAud4|z0J`bz$2^u6>?<GcW{<_nwIxrmY~HjfL)c}gS=U1ug=S=W`@Vh8WB0kw
z4KRm`dvx|4UX6HAMxX#3GnGu^F>Tn=Nm71eL2XN0Xj0W}idT#a+q@s0+kO|6JW5W7
zuAyPHTKaXwS^!lDOL5tbTTD#n#EBE0=4?u@UnHq#?hmBF!5MvClNO{^j<XUnUn}La
z<5p!LXQt*-&oj9M{JF6=0-hX~6C7hri9u~SmX>N_re8VYKB*VjCsVC2M2cS+4&S#C
zO#v4p(8w)K=c&P%2vL_dj!Hmb_l4gV3~It!(-s5RRVVu+J=!?t5qw+QXXob+97|uf
zb$>Jt8twX#Xh7+bjMZy5y}t7DtTFx3s8d2}@4hA_^hiG5AHa3<Qn*mluP6WDwFd^`
zBD2-A9maH!(3|5=WLr=Z5Yqbsi|0oRDuTEp72TLrqouZAju25VUYriCB>BW>CCBBK
zR#wU&&m;p%ya2!+oE&HvpP5lb;+SkRO=SiCwr`h<6x06w`}fJprG*K(09M6MX(KLU
zg>mD(m8MGzv)+#$H9nIQi$`folnoa$c9xxK6<*7Y_0h;S3y!@M{Hl0)Nw~=@HCZ!9
zNJO+g5Y<2Ljo>Ug3ESb86NzQhwnJYp>$=YxB<$S8YfuplTof-8#Kl#1wX(M|=#AC)
zvxUe=TyJa)ZD!|jIQS6$j~T=}4W01rZ?2WBvm0)apPw1aX%7O>M~RLN(sftoe*5b-
zS`ss^muQk`m`S)Fy1f6S*-xv~WPdDh3~)+YnkzunwwVVeb+7JeWZn2)pIg&b;E+l!
z48MKn4!_-?Qm6rG>m(%ZR_Y092;Jxb=KUIKSrk0!h+>aVt^~Yq_Lcj}i(5#)apljL
zO_3xB#Ha}5OdO>a#Lf3DynOkRiz)ZVj~{#<ORhnmo{KnTbh=Jl7M*HTJXwjzN)&aO
zzCq2mBY2Xm$oUHw*bf~#?KIv)`pL--o7(t`0c<pj<BAl@)kg#*S*dGjg;5_@(b0-H
zrcr|>AMWM;ieT!{&^acZo}O+WbU|nwPipe<{&tI;nP``pQR7dKXzgdDDiI8xGz=DM
z5npF!iU2~4fwY`Q^ZLUQid|=|y_k>2+Rlt>7osDW>-H7jGE+4^J+1QW*yV!}T!`nY
zhXh^Us@%h^Gd%q$3Oh)^nf=hmPoGS_yf|YqRqQ-vQvLiyNpkj#d27ab)Fc^2hKNbT
zCkZTzH{mDrJnct2f-TbTdU^_JHwW}SD5n#$mEzRMN-JQ%Cv(jDH&+0xMBrehA|&%0
zEs#e^*T=EvVF`d$P19ao*YDQL$r^8XI@&9DQwtn&Ttd1_ICyvzfa@A&C;GKb1oS>_
zDO{Qx8rWa?>ik1nU?8t4E}gfpk)R4hYtTC=zdXA(Sv~X9h?s)))YKXZCE^^Ka!M}K
z$&)I0q$}88x=}&9r3ELupI^>kdu$vlbgfB$m8w%{J5bM5FeMvy>S?1##pN#Q8>3o!
z(U}gLwsgG<`I9p*$Y*;0{{7t2{8+&A=Qnajk%}5)WcE&>$%>gNqs3_o;W6+oamJ@n
zK700T$ftbA->*VWnkS&pnby5J&&4dHcm4V#^1@K@@?z@k+qWBYZ+7#Q;yLLLPJ`Jw
zbSF&IMFsVhk3-yjA(uLh9GThP8LWN+SWwDNqtk8RO_(WmJw(tZmK2|3mqT8ozu&ZF
zOC!*r?ewrZKD{J3B@1TRqQz-Vw5W3U2Nj@HC3L)N|N84E4!DrpT$W0b+Do*`YHtL|
z8#Rw2tYse;qMZc*zg#Oa@dzJa7XWido(+lHV4l_`;g`ufcSS`-jeL5xr(EvJr2)JY
z9^zT2$Fdv0Wv3SS3i%?}S^ac1vXbHWB*<5dI6}#7BdwWb$6I@9=1<}Hx&3;--hO5Z
zXF9NQaewPur_7U<og83Cl#?}-v6ZS=ssLW2n&|2o%dVni^x@%xHj-t&49#dhval&}
ze%EJ4shRWc^kS~3s1ev;E1#g!VE)yVCpD6f_Pz8KoBIrk?f4T;&67BH2LfwD1!}*u
zdY!gLK~Y47H4hwVNk7+=pzQK)`>M|y&}Y`6l@Xy^cgy22!5Nf8B><h(X;swfUeuUT
z0u4az1PEB_pdbHesDA0vqau{Z^#_3{9UGqzQtq_peo^Nv9-|twOta3nPWd;xKdo>!
zA|@^R^dM{oj(}{zs|GO6&dm)zV2-%rwT%u9=h4uqojZ5(-~91$J?#;^n#9UkIi3Gn
zx438zhuSyW#b92e3|f8)MQZN~5BP4)Uhp`0US~oR{<^<H7~NVw2xM+T&3}`kTRb{&
z1u)>~Z6<Ggk^qt@x<5Vk?De8yaKBaeO+Mphw<PMl%mrWiDO(mWSKA45MTy=4q6f@}
zoY$&ndB__HA8(kMC4n}2<FV|HV|fwBclwGgeqUaa{60Pr2gjv<`Af9Io74ZFQd9qH
z;_5&D`hQ35{g13rby)-Of~ctT<Q4QmqI6*7-xb&UiZ8uJV?YRKtNVZLz2&9hOE0Qa
z=xm3eK&++O{`z(XMfcJ_B(%uaaeh*R|M&gwc?(*ep5ER7Thd-fY}-il1&vX1=kk2>
z-Hkg%==R^*1#qMEO5%Z1SM#@TSDB6-a~2Yz92aFEvV-AnI=W=Lp|8dOECix(=@uoS
znpf7=%7G(c0{}5;PDw^xO9Z?^(Vn*bOrbxF+}D8iT7l}Z@Vl%jPL6>_;xGLzpguuU
zn9h#(qQpj`dwhjX7fq$+#Y>lz&_MVfy|9f44nSYc6Sd;0`F2AU&CSgXFEvrO>e8;2
zGy?Z0&Ldkip{gSdbJLt(NAX=}XtnCrL-Z(7hz+%Sn|3Pzo%6ZP*%AT?G)o-{^=?o~
z-DSR}4Y8N!CL83GGYu~TP!a$$ZTk}sj9lKR>#-OxxYTZ0oD4wTX5OBnEgZ^QJgSQ~
za$xhsUmuv32j<U-*h?=9e2_gKe6!zqhU!4)Q{edfI(a$#W;1i7>nr5YftGYVaK9m-
zNp~|aq<;P^wYWIbr2(LR?9%fWK+^Fy>RzE1U(Bz5M7r0Kew(4MoL-BlOXuZ8r<4-r
z8QlN8R2*4F`Gl7C*5(xmjI~)_9P_}IjMx_FmHEVjnIhfNt5>ghj-PF|1A92&QL}k3
z*YJQ9eJ<c{``}*xfrdC2A;xzvb8}C*VK?GOL91R)Eu7W}neWOrKOpVo=wEuh?%h9Y
znzGF`?|FF@bS!<iw|Nls3jspNsIjbpSV7a4v}?wL-C~xOmQ!P0;!TUFoGg5NeA=HV
zw>;Q!#J)wNF2@7mf<e=&7iC~zup4ZYez=FT{&#no-Tbuqc}Yo-gXiY@LM;iOlPLBz
zMLQ9=Lvv~sB_#w14J6{TL3hMXz|jjnwzS<?ux1cl^)cM<l(9K)ivIj~$TyRT#V!PK
zVv$n}vI}pP^ykkXzA|fF<bhp(|NUfe35Z2&uA7$fj%Z|7AJF<Yf}*1qxd{*GfH=-_
znX#yj6xYS#DFf*CfaVMnv`M%W%ri(W$T?}z@%sG3J*>J#t}1SC%vu$0-n<#%$VR!u
zb>c*w)y`RZQRnFPQ;<CH#FF__>rb(tJb998cu88}#*NoCk>X=2?ch1``0rFjfW#*3
zd($FzzsQF2WF5M)cmD`GjK0R+smL58CkKo=I}6L_YxCLW?IvK*Rism}*hAqCo#js)
zSBbQt7mP*d#pUb9E5z7#%?w8Qiq(e;JJi3C`&I<Pr{yifo;|U^WBal}Zr7p0+D#2A
z6D0%)TQN~3dCL&<v12CSCArFGD!Geq01k2G*hpfP%{wFcjCU}zt-cjXLTnA*ruln`
z*E4#sBUjKLH0b$?_SrJv;5Xo4v2b&zM!V>`{=PRo))kqOzS?n;c4Q~eUx6*d(8BD*
zK*psleJJ~fAC~^8HyF!Qd?m555F2)~5^(ie3;GpnjY8))z-<1^a`XnZ5!wJZ4L?dB
zDw=P8(t!>p+#|z}AZ0YH8q|sL@ryanXib|EV$g_2=X;FzNGIzSi>*Z40Yux?GAz>+
zx@~pp{9s~gM_YF+LT|3tV@c1>vLpAVvG>?Df(o#{a@cc)2oa~r7Cqm1i3d9dzb1#e
zFHWh-dG%v&Z)&`COxFP~w|4DXRxkqNQ&Y-FO-e~>C*GL1tDuR>x9+_d@<|Z9FhOou
za~8lP^$dOar$^2==G&Pfmb#MFSIf)GhwvKt<U5R|0ClpHmztPRyn6KoK=kC?P^xKb
zhCzlJhp{&m+%8fRA<8TKS);(nu|k_lxbpJ0=N8&mch=Ly#-qPiP1G%R-?VL;%0H{u
z`U~0^_Ev|go;h=7ustUg88ayM4=;#e)hskVp$6!1Yo44a^Px*LZB8+e>SDz)HvRGO
zfk{K`<BLID$?x8so}Ha#1#js*7$2vUW1&sh6Ak=#h`3&MN1=;qrcte|&^>OVnt^f#
zQ4u32zSuaLbi@U%b&T}0W1zou6z~ZG#68@}Yay=^nJGlTGUdTeW(1Yuz``Sue&`?e
zJGJl`RAMh>d+binVRc0$1wdm_njt+lWQ2X$HU?2ns0E}=oud`8`gVHlx^+TSUr}fl
zo|!d$HQ}LLdMNciYbhdE|NQ#CM|xW-`T$418gTvUrHUE)<$P0n>FF(+B2!aSd$HJ_
zQ#L)HpDH=LyR}NzcjD6lOU5^q;BphjaZrS&=HlMgM2MQUl=(5qaG;U|;vt2H<B+5S
z(e4Yn((h!J%`(|bOFPgwQEc7otz3Z58<WrgIV~vC16d_(e^<oi(39$*n!YzW78Ml*
zBBf}g2sT}L1N6G$1<)xpX^Bjpc@NqClMkKg?}>@bb`J!2xHAK~$6lnpKviA0CO>Rs
z+;!-xeg=@hw#}1@tXNnE(K4$sLAlamSxihU(4o_*%<svQZylSLml5-wI}RQkZnSx=
znUm6sqQW&5Sr;o8X~FvygkfHz9FDp6+%P!&FNWfX0A4f2^y}y2y20W3#vqUF;pXO!
z(_;g%Xn<~3c4`@N80~cC)%U9(mX6mNMM*t9VqaDjz#e3P7FJd~6<eo)oNV-adfI+a
zaQ4q1H66Dp<x2Fn;q=|t-r(;_`hyVzGUzX^wWCJ8Pa7#1D>7~2>h<9D(?w#l6Po5O
z`tzZ$U-0C~uSBCNIq@ngO56D{q0mAXd#ZvPdLzwg<G})OExu(@jC~Y$Zrv~L(k)z+
zG(DUP6<@`gch<_Ftd$GwN21HHV{N%N^+M)cpA*6c{LW3aF)%Ii$d(Nv=gQ3xQ_(1>
zq~gjp`TF`cgLX%=A`$e7R>=0X=Qg^*Ux)61r+*C%&inCWyO#ck+(oa^vIr@w`c02g
zNewbZ+a=pn%+1DjkqA^RTwvPtK0T^K#vGoM_ULk*cnl8bH5ijq${uJ{66B*i?MK=S
zayqfCOe8fXLPM6F8uLLBY<;VnV?nh&Q1Ue?1c_}3U}fN>q&4Ua1@K#15wi}E5emmE
z*hx7Ik&DnwyZR5f{Ae7)wq(R%{2oq?8^A@|<`=q_m*R{f-BeIUt><_{o;-Ph=T5Nh
zt!T?O4^=d$SD7|snHlTSMLIgck&1q-T3w@F+~rH_oFIhm*I@q-OwS_7+8<0WA3SAs
zdaP(ZrKu43#snM|@{>&hjS=|DGLxa(^A;Vs-y4m#P_7c)z9rqb;aFlBNuM^6L`?A2
zzDyBGGOP}XK`hAl?dQAkg1{V{KJNz)>gu8{@Y@fop;oZRFM$BYYck8-n|}u|nSdrv
zrBWY+O%%ws^G3B1O-S8#pnJ|sNZf$FxgTWGcz^BK_$l#)Uw6co7i^@LIeOKhND#0I
z*g|@1<lzdOW`)JY`5Duy=O<)mbV2F5{rT>tk!Kx)^PSk9W62IlidinA5#^lFS4>Th
zeSB2VOxr=ViklznDlXo}AeMwe9}n57E#EE?DoZ@r&XJBhMMTrHoE*Ht+Rb}@Hpq$F
zwT!{w(3E4@6)kZGc?MuK7Fl@+3aj_ShXj4TLGDeAaYyc!$7YeT8y9&_vMue}o$7fU
zeVLpSC@TR<Gk}0o*hc~Cr(Db7mh{1{Vvi8n0PS3>mw?PMfBrb;w0YgTHxQtzBY*bn
zZN;hXA3r7|mlw)LcXWX()FAy24x8o){KXmUhl3F9e409J83YoB8bB2#HGw)!Xz612
zMYBLU=ycGs-<~vUc>#dUE+nJ@ZsrO=vBILjd=ccd1b7Lyb^APitW3CkPNRd+Xh9qf
zr4=TQ_f;pxMY^{cs;jA~<@>Iu_=?EM9bWTmOy85qn`jx{rM0!SI}RKO?Q{2-?cJN`
z4oUi&nm|+_%duv_^PE6IT27mVUnMkK+aVIDdGBU7R9^O2o{zk?_(<!OV;2~yq(S%j
zksR5_-xEkyeavg!Bjsfdo|1!rJOk7%Pdl?hn$FJ7Km_~#?ABtB;@<Q!dYOdfKi$5X
zKf1d|KGlQps)Q1I@KYFgF4~i3&y_`>><J`AT0i))IM#-@-O_&d-aYjU!)g=dWhkSb
zQza!OoJ&hf(o>svv0QD<vx%ejpFo$8*lUe=6i5;_u4hdAgGaC{a4Q)4_+Y1}ID5&5
z4_bqjojXcVYBh-3Fz60qnFIV{<{&et%r1n{NWGuL1!(<?TgozWB<iS@Kj;76v!`#8
z{1hw^Cp0^UfkL-^BFhUCp0&oHD`ls)L*a6QVWDARKQM6nj@`R^8=tRPw?(R{2)n3a
zS8wy+Rq6HgEGP<wx`a@!bTilm*5~+U((3$!{!3GoOX>BMd=5EB6!au|@r#v{V7=pk
z(^}Hrg0|Plw~cpob@dyiMw(B)#?jENFLrZt<A?6yW$x<chPc)(ocgNC^>_+g;Q{(H
z+|s1$LKpCi9tW%LyY+LNgV)9lvD?a~$H(ijfXOH2A^_554z!^pYM@&1a-^!JMl|&`
z0YupgWG-QI?D}iyq?g~u5>_p<R?Jku6L9$kKC`rMe|2aa<awjWvz~vz5pSskw;W*6
z?dasx8rBr_{P}Z}fxrUrJLC~dm>yu~zoO&YMNgm7(xRrFZz~OKQnyo1<XTHV^1I#S
zPx&0Z1k^+l<&abc2AdLz82Ft>a3C_~E2}!ko<cZDJY{y+^j5@d+<ELL{)!x2>?9u=
zl-*ofEryFk^Z#&n<EZFi8bQZjFTS7Ifa<Oy?lO!5jvTNU&8iqnR1mJRrvR;w@wq`J
zZQ8KGO3eccuxaboH~B#*hSgz$4ZZ2w^0Aj_LuY`jLtl2UeisFHg0e5yeKEgD&p}o;
ziwTfP#h0ORK%eF|Z#l`~1AwoWon2uS0%^^3w8WXEBj1jFpbO7n?{QQ1At1A0z+I_4
z5x&Bpvs>we_y)S*MiFxTlcqQl;rL@?NsM$kVRqT#c`Z9dLg&Y|kRjj(ajWl!V+H-f
zMcLAEY%g=69E#{1<o9rsAo)kKEfZ*`L9<7s=<(n9a{BRp{)=`VNZkV&mfw)i?TEu*
zoH#$whAE*AKa(98ZhMo64*eCtc#D}0`plf2eQ>PQAuAF%DLUPvM;u(8&O@AT@Zflw
zn{mF^L}LE#Fo2WWFjl;r^3R&}h}8%`<WiDkV2Zi{?BW?WanI9JV{yT;1uSsQ@tZi9
zMkV?YoB)H)i&M=+hEYyZs~fP#c4UEPh$HZ5kcLf?*bNA#VzU6m!44>bSFmx`a@fZ-
zhBhoRHl~2`%TfX~;Fjo)tW!`B2kI>=+|er6$LPwrKzmTi(3g%IjP2;|pXu_bz5j%s
z4IzJNZU$v$i@;{NM6_NNU0q$*3X$^JO}{2d`#$z#9)*}4f&=u7{=i7tGX><&@VR?@
z040M!5%t|DCZP}RfiF=XNuC_9<QCrT#n@XD-fkjyB5Ki%$S>*~fzg%0clLDL9PJPc
zotfyZG_ukAlSMdy>aA&APW|D0GlkbnH<e@n<|hKdovb8+vqRu(;VyPG)kvAuef=o_
zTAm~1a}IW5%eD^r9u%y%Vt7q_Ts4D_yuEp<uQB%5M#{B4JOOyZy_J^dW6aWrk`j~%
zx27cZhS4pb{AD9V;$d=1z)2($Zpj?rO(U3%q*0Wa+JVOd4Ifwp!;xPnc-K*m6GnAQ
z^d$+>M{Cbro(=kg`WPou9H9a-C2mM1S)++<HXXvkqYVQu>)~oj><2S%<Tl$I#AW_&
zbjt;qvC&F@)@RuK(8*JU-M^-%v)ihc5E}Lg;p~x7#8)SpJtcTC!XHj1d02xQ7G=K5
zc>6Mlk1DiJ8rmwn;N5uH0nVNFOTWyzinPTSCly2HbWjPesbe2Tvfa?6ByHO-1uDpf
zt5^<J1U{{~8<xiT$zzWS`<sY8<7V`Nh;aK@Ne_S>UzpvBWsY67W}t|){lsZAkT(X_
z2-Q6&FKn_1#fr4fDl<?s$LhZyFupc6ZyFoyLiu%^aN6s}d&>2u{(w$fi7h|oRaI53
zt(pK}iMu#8_y%rP6o-j>Z`#H30=5@sQ6F#bdQHp17_?*jCXG^&uHz4o{zLm{-BLK>
zq?aBBWX2V0zjc&HG|9)whARQcqcdVd=#8TtwnGwRjStV<$fqiHfpsQ4V@ePziEVxH
z&K*0zVQ@1rL`GM(w+j!TsZ^}DTNbPe;b)I8j0@V?u5p`{ReO3JACE~ROX#8p#0}X<
zp&Jhi7psU}q`-X5>eb-?Hs;J_$3WJ#uY7Dc)^@Q6LeJN<NahH~4}CK|OTe|9CD}k%
zSqHjH4gpiZV#UwlV9?hEkV!01CE^@*Ag@qsn!)lBd^gfntP8l7BiMbMiwk|ET}JSq
zEi^PaZKiT_M$0+Ck?b;o9I|Dhd!_>u3tToaQx-G@bs#k@E3ocdOQ4CCmR5*I@Fl@#
zOi?p4j#sp(Q5>^6`fw%;(AAtXp8Q#NQA+B{jT<-Wb+}0<pJNe@L&H1}87zl1uY&fW
zvCzdKNStAGbi^RbsKyPjydK(CD$+WOm{@1;(qlu0n4XG$mR@)JY^>@t1<6nk;iH9y
zH}flh7FK_DbB=JIGbVao2%oWE(;{4HkQvXxJ!h3y`~3O7g5q4^iC{Ug`QuEgG^18o
z%>J3;6A;k+w(&|?0vuF7ac^kI>2f!H{u${ipZb^1MY!3V65NiD9pJn1<dZLhrCln(
z*c?*D)W@B2En0(4D4~xJ3M!x($j)B|1s3OA3pp+7#fw96dY2Qy$L$Yf(v{psv3f)c
z8sre{3o6BbK9-iMhu`f_@vnRxrmZlYmN^w<<8qKmiT*q~%DEp8xwD|slUmU?3`T9#
z7@gvNG)EE6gRH1(Z81AH@P2?2w|mEqS7?6o0y_M9CT*;GZF;JVK2sg3blgX2s~IbP
zer!igHj6krQ+I(OFW}4X$)V?dQA`n%m4_88KKJ*U-EL6?mpmMyKR8%qj-DmMsjIB)
z0<4a_X3fPOq{dVlwxSJJV9GTwG_ow53Zw?xOS*iQUasQ0q%#9H$kT2tbB69>YN7b5
zZ(X-ZBO+g5Z=P6&x@$aYgI7}V9fF|OC#HC@l4G|+UjAL)S>3tE?uTy%F-H`$6FUXW
zW-7$&3nFQ9s#!M?O@93JDk6?|x{Qf#MW=l3>5|^hFE3R>KdxW5j+pZZTq3Ojfy?M2
z9z1xU5}M51rPigM;_9PNvnL&Csi^S}B(XH_+NR2e%o~qnH4A|MJ&&DP$>1tg$eY8*
z_#eT}+#VkMqiPF>%h<ibR&Cnyt}gMC>2oeEez#fOu%Dilym+8w6c-_?s*;jFNKd`Y
zVJOwI{5HITy1Ke0j~C$Np1Q&{#dq<xlamu|#J)|Gw3VI=KB=RLW#FnWDSBWBGutNH
znwwQoZTOQSU5y{H^U6O8wAYW0y5K9MEx^T9V`9k|r5LTiwbgo>|JTO}hYJ9&#I`au
z(ymJMY*54h#gWwxO5}X`awSf7b>#*D%g(GeFSb*s)Ib>3p{s7nFpvr0(nh1zs}Kkd
zy1K~PXdoUb8^<BB4CkiY<IJP`_dlmG>C>DY-#YdT?r1BMSSgJy`}w4B%H`~qK@r#_
zin~;<HV^j<?f_p!3<o)LH8&*njH*MF0Z%NLRe_x5Z$&;ZRp6q%WPx4bs;;^;WXA8J
zC;Eu@`=jYc>`)upWoi%@-Xt*U&;ukcUIkg_H<-nw>Iz4@ipPU7L<e=z&@VFWXT`ea
zfpq_RWrz@74R#>pj02Jq6k@!qomu6c+pa8$95-OEDwx}%g>jk0d8DJ?QX;?#>22Ku
zqzN?nL<M_IZ9CK&P+N!Gl1$$|caSOaUND+#v<gkV&*}H<W>Y%OA&e%*(<dBqZm3L2
zN4Vgg9Xqb8orLkCz}y42xDxdc2oo6vtK1|L0h+7lVg9jj^Fn`t3^l_i8IVpG$&$ac
z-rru^^u|;*XfD&?R|U$?FumGSbp1%91x-0|``jGZRg(VPI=YH7K@?wmeJq~a{Pdgr
z8L(ctcJ)-$b0U2ty5JOKQ%{x$F$#ue3EKXAB2zqkERrc8Fvbn2D^ob}NS^<J#r@US
zW%3v+Tbr-Ka4y^QOrds%i&~(7{m|EHld(vDubm<+tVZhsI5oe<+e22cHGK<+bZUxT
zjbl%BxUj&$&C{oERX5BaIbP?F>P}28vSv^5I)m5R`t0HQZ3jR9`t1S>5{`A8j4slP
z0EcFV{sV-*VdgL>WBZw_&e*hugPIy|;r#G4AT1F8&=<JrU-6rMeGR$~_Ktr_S~ZCl
z^4(ot(nUQc4y|oR+Hi7Cr}gaBKw_O*`pyhDRiyh|{Izz(a7l2KySuw*<~`AJmcvGq
zUJR65Cx|@p?#bbEfPCfAWu&rz8Q0PnBsP6e^qnTtUx^#E=JS(ifB?;@n-CGb^oYtJ
zTgFswGV$~C=g)j2z1>xkA0Sa@YYeIb>NmZXFEOM-F-z?Eh{o$@+A>I{pJ<+~kF<G2
zcZjqH1Y6UxDjpkJmN07U6!VGhU}v4JS2#)}DM%C@`^P{G&4C%F`#KMPz$oDJNPW!1
z>Ru7(KmIx}v`3eLSC9#~%D*!zSPpfy_E$RT;Y=D|+YS(9NE*^>PH0B?UH@3oiMwl|
z2If0YC4fOqL_0JD8RulXBTVBsO9MM13qaOdoSNvvkBL2>@5Yxyi|7uBN`<6t3yuWO
ze+fsR4r`$Vc7@-(O$qDd{I0G5j!!qw%p)jDDKN%Slif#{*a+VY4%ZaR2?|Ia)}U>0
z66bU}SB}o>!a#3O?pHe4d4Ky+EgdeCti2JFwL3od*wHaHg4Qw1<$hjgcKW}!21-LL
z;SvXf`hnVByD!ekGKB?lX(teu8Q{0{4Dm&QVDd6otXcOH{@5n#^d2y)>uBhcz$%qq
zD=y#7uEH_U1sf7pv{=g|vIBKy5_O6;FjCw@2kw9jjffj#d;LHTY0z;l<8&K?ZS<U?
zp%aRR_>#}J87`7{CfTPYnKlt39$w>*1Anz)9w0fwE4fvg=4=Jy@C61p{Xew+(O{Pc
z(N8&;E+@}}VN5`{?hAzOAdiD8=RDlsAGd1Fx_w^>VUg^f?l$tKnGd#-IMBl!n>KG&
zgbbNnR8sB+0R!mSd1iD(scZE3T<J3DMAA+?a{-dA@9k*@Rb>n@!*iag6jmp2+=O$7
zfg=f+VW2li_hg5wmfzgufZkNu2DxA8)1l~Yk<)n#VeFV}*lToWZ14pWNdsj@8U4;1
z;|4;d(ez6GlpL<^gvuwEV9-}XgfD7gR?+;ZFmbqmC8@w<4G}KzCT)vrE=#&|Qw=5y
zw{F;Q1&uF*6(8lQJ5jSXLR+ba(J9HWjzDsFj|LILC=#y#4_Q%4QA`@)iQ&We06?`I
zGJp15=_wW#DaZmgCasdN{UpiW)c}YmW+9-qL7-_2p{T<9)>j*u3XaEw7~vyC3#h%~
ztTADtO>n20x1^nc-oh2$VYlp`wEd{$Yj{}*5lEUs@CD>BK+aOVa^*=A&nJjtLuj?%
zv|El#c$Qw5M@;iw|G0rp)L989#v7;{oERT?2hsWAOz->muOOnseKa%Nw<1-=!pDEf
zV|g)0=D@tLHT+zJb95%McC_?PH;_1Oa0e0H3PfdTS=kVJ4}vD(>U)XLee6v&tZy(m
zKCwv8eFdopijJ)F0z~G4V55jS#G)+>uGn1iu?ZtdA`j#hVTKdn(AQ+XMKnl6waM3{
zVC-}hur*=y2$BVqj7OLeUR#Bi1?1^DROz-+o99p_>$~whQ=?Q(Ea_K)F4rhH$bO>p
zB7j+6KN_9jiop>SkTd%ACbVm?W%a#0Rh|N%ud@E{4&vYeqh|Ey4?1ylr<d|4o3e}*
z-KJZO2I^Je#D<Rjgv%{P8N})C{rd+8j`L_?&xi*+)WFBD#s_{i>PV|@b{3u=+b)~4
zRR6GUI0s7#QGaIDg*6mk%4Uk=JD7Yq{SNA&=>E2St&(loWx<L%ubAGRkI@mcyqn#g
zm3nitvw}^<U=9qx7Y7P-IvyR+fueWNM?P7DKW<LkW!g+36n(o!0H=D!$9lA#ak>w|
zu&5)|3^$s7fl3!QpNzuWi|tXFfhSErXBi(<u*rewJKoL6HTk(y!Q~<Ri3TJqEVel_
zguyKqA8ij$-SMOxyCD^1M)feCJSZlCAhKjIGj*`Y9($0d++GZ+T_NArIIfJi&6o}!
zejnd23~O#s#D$8LG&~=U{z18r<j#1<jwVP;p}*bWOO~CfgX&(=pJ&p<<~0Htclg&7
zyuPm#SB#2_yft4*+bRQ`NIfAoKg?S)y?GBuTKmb_{!_D1)eH+p3~R!S;TR5Wua)tO
zLB$LN9B8HMze6K-+-w;nPn&t@9)bnF=c2p6WU~x+d7!F}6Vv6?NJbgtAx_{TX)mJ%
znyA`4j54;iSq+OF1rD|2$XxBiMi4^=_E<XKXif>2Np%Y5F%Y5Sx#hu1XvodP)(@yj
z+@2thLmNh*kG(`$GDe1|czV>UQZGd<{loVGIY7W>lvU6Uyg_q8ORbZI@IG+m$P5rv
zIG8$07|stLJ`6`33huzrMI%13*Ay9?LrTwm8-Ei|IfbE>{KmdBuQAm4UwV%dpAQ*R
zLmnX_Ye_y*a3g|icvMkO4wx{@ebL#g(gq|bpY6{p)xo@FR)T65;iqgsvWKZrFRr^B
zYy_aoD@+B7@f}bilW2X5>(gKUaf9@eUU0W+H<6gguM;V7jsxQrME?NNk8P6~yY>aW
zC0v6SAY|DM7%5<)q6vr;4u5H2v@1wG?I%vnQ>~f+e5PQrZR3EbT~7v|L_BkHa!9US
zgcZdi{E<2t<bu~3tm&mULElQgEp6%ak8mBA<~M8kvS$0LTeC3{-P<T>>^LL)#awfV
zRU6N2f_943I$~wi#^fMS4_uZ?(2JGOGGRF`k&N5Gy&omD5h5+0)3|=JR-OX%m?m&-
zyY}sS4P=%GCJEeCqKtG-52n-NiN1njL6Rvc=Rm82a9DG3R+8~DFbk6anl~Wk%%L^4
zgPW5I^%$T~2`pkG0j@i?`1v!*?BUePy@H;N3>*2mp8(gR0<~NTqdn0~Mt-(%gCG*3
zNos{w_=>EBIm!n}4+UFW+k8&NFbrZ4wv4PVMgr=PN^K#qXcxO{qx%QL@r(>AT`%<>
zfS%bz^t>lep4iG9iB?97NXTK*rehnvfOsL3MX;5=zWV+)LH|1Lv%R(tW-&F?46vMK
zKcYc?7bprt4R;YPN^q?6TXd)`&(?YjTPRpW<6Sf|4IdMw5Q>*E-ayD{JQ~J_STy5g
zV1Z37?FE}k(reg76VQy&T)sY3gh(WweW9B_?lvt#UoZuL+D4NqMMKO8_?+-$FW3%n
zqsaoBuj=TS8_2Dj0&Z^r>5w2YqD%qYvf*?(|9HTx1dgpKRaX~V>3XAxLDG9KcM3%H
z8vqA$DA8F+Ph>udw7u9`TM)Fp<<BkvmFAL(DCkTW_cGkJ{Nia?9joh%Kw|9%Lf?p7
z&P5W0qd_VEF`;k>Iw1TF3J~E7=mf1Lu-e`b(J*ARdiCo6HfZ5oDnZSJZ_IBn&Bsf6
zyhOzht+TEt)3H_G<ogY7_6qE!NHHIke3Vu|JL`t9CDR)WrzNbKuxv>_Cv<_s&*hbE
zq_=>UnkjVS5G>UAIFnZ;`i>(?3x9t)5I;(EO<=p2^+#?qGcq!=U?w2m?=q?pk=T%D
zyI+33Du;wchU}mKs8ee(wIdrTHZt9}KpTTJ$W{F}C*<XAbVC0B;1rqDj77quO1Y{C
zkF|2N)KjGyKyl+!nd^KbfP7nAi-~XCdg$;;9Dk*9PH6~}zs?19tfufDChpp+JUC%<
zV;;+)gYI}+i14HT-2b+bV01f$%Qs8@+Y6A2>_&{X>tFrD3w&%e;a?577BkQzO`)s|
z?JGJ1_Yv!n(oEUo199MyXe92guU5G|1YBU}A}*=_78ZBnsfZuNkeBz&R!Th)6DYSf
zI=Q&CcUY2=LTpqbe{ZLc#)gr(MU@%n$)DeUE2>@_hQ9%1DlBA(H69VV-IQ-4vSd8}
zcsu}MzHL1#Jsm1dmQe2-NE`bx{j~n*?W2&}gzSf($CXfSUD&honD6uv(2~|WDYx{=
z4A{z-BI!D<DxvMLKP>gLB{MG#hu_}?3^$t+tibxmorBT%*2e!_0`NaI`oA*fpJoIK
zkR%~&4~6nal>o}Pa}-J(eBHmUghG}h{|jFtJ^$yeTKVD1ru|=k*S+gJg}x4G7N#sJ
zOzEdR)uBxjAZEu6%rl4aBavO(Df(|koW>njE_={HhyG*W5MGj=A5Ov$;zzOLMkdU6
znuzr}f(UT5O(&y_<SX`HEB61dzFT5&nW3XU{{|xrJN=$WY~Xx+gf2#vY5QMigKq6w
zYw@%2v$E1;;)eU**2Xw)Kk#b9hSl|nhYl%+Om6#deuMI*vpaUI9tiGYSny*G9j>a$
zX587C+Amo9tg4E&K)AqVnzuh}6=mQ4RsVHY$A7-VRVx2k4a&X{O&tN8gLf>LMPh}x
z+GuB{udnZ84*C<$AZ6{@0^0$?v1BK>;n(EYRukj85J7T~Eg4~_id`5O7_c5`<JT_B
zzESrfmfQdmA#sT$_b=rQ4T?YxgOkp{;c)C6mQ96!8?wPWQoHUE(_5hr1XE=ZQPFT6
z8jORSIdcm`67M#_2pa&eYTVbS2hQ*zd*8lGA;o5QM@L6Az@jWWftd9P%It9g__Hg2
zZzsbfX$qJ}3RAHMcbc1@FO=*`{_yj8`CUw@2Vlg%%IN({3iYQ4&G$C$d7zpdnDG2%
zjepw2BT2?0)yhjt&tpcFPR9bj$9Vksm+!rJ-Cefi0v?Llw@3HB%})*-<>KPn#*N2$
z@j}o02;So{>f+J8Tv}FYPS7`lpt=etiQwnVYmct{UC-yzK41|!EAr7K|9!2V&#d=q
zo{yu3{fho~IgVc7+%%SSehAMIv1{d*hQHl^s!G79Qi4+X(kDEE$jc4nIa^-B+4ZBZ
z&(MJuKTM-rIn~18`DM_0nAeQ;^&JDgbI`g&_Tu!vuccqRAqn;{Y0#r<G}qv5dk(Mc
zLLBT6_v7N?E;|KB;AyUkuk5frz~l3~ckil&*OEQ79r$r~58hZredYa6-Qql&jvpR@
z&gO8|l)MYa%DWtV0S;efed2TSQs4Id`?GidMnS^-pB+pWrM$M$g|(1XV_v`V*z3~`
zCBA~DHJSJ@i5EJ4ekB0Tz)Z#tA5p%KfiwB#?!U<fTm3B{ATUrD<yG1Z4=E=Iw2FHx
zk45|MV?Bv@stzjY$KYVd=43psBhSBo_5!H-V$8(=KHqCAuco&WE>?&3?%lg_>sDo|
zF78LzxMfRKDS5G7t`^SQ@ZtPiA&t~nV9!qf^EVuD`Lz%>Iw6K@nQg@r-6WHNA$w_f
zb6LD!sR<F|04U2JPdxDw<jZ*rEsq37k*@we6-$Pr@b&J?%47cLyZ;YYqhmZD8}hCt
z0ve|vz+`E+f`S5%qO0rMj~_ou!Nj(JrjevC3~yx|6mwMrJrlM$8G}8i&zvQp*t5yO
z!QrNZ!`|&r(VEIYaH^trfRe0^ZnO%6g0Q^Rc9<bXDO<c*$>^9al-x&eVf3-az$CE@
z`a^+94Mje@<?2JS*ykX)Cm6vp`a!?=%%1wzueln}cv?(y<^4>VF8V(~FI@rT9o~F-
z<zH3~MUM*-4-a@HS}|8g9RZ|J%Ix3d41Kd@k_p6B0((InZilmiRWmz@4{AOfs7z?!
z{d;=!wR>@Yz=hRo)`WLx;P3Wrbsz~kkL|<>`O(focBGR4bYJDT(SaFreNSN#tprQX
zH<4dZPzhuogZTl!kPs6)DrSJ6LUfXWIhA;@3B$w+5(1QdCE9c*G)=J8AA!v6hZJ58
zrq9gn`5s7y(y-+-@$o5vGh`*UndrDU8IXRs<3$2?sY5UnJe{1hO{|=rnhHR3FOz9l
z9Svqf0fqxsTmZ9+RY3wyEHaS_R~{p@dyLrif}q^7XU{QsmHo+5g4IB>PaH~CP7KU4
zX}So)^DruvKU_!BupTjrifW_zVaA|@_0KPdiQNt!A{o4FatadSJC5^SAO+cuoIUbX
zigCvS|D&St$L*O%((?bJtO9d%Kx3oQ!(O<a8WT=}{=+_Bl1CMsv~}BcUfv&%PYlZ!
z!PVI{+bsOK*)$G`BHT#}W6%MRB_En&jMHBMH{DRD*8lYv3s$gU>Y#~nYetN6ttHZ;
zKUf=e<ub6T0U+hHSz(=uh%CXANE~@OZn>XF|9(o*+ErXD|69tQ>B72P2??j6<Q$Gb
za<2)s7pjrYgZuZtH1wWe7Zg;d<u~z9$>&&`tEH_SKrW0>_5hoZa!3`lh(E46X<Imz
zS%2Ugxn3v3c0i$EP{_LHz5*OoBI3T%EAQG{8!hmzbU!NoA>CQn0f^T-qztZb8Av%s
z=}B<dRw5RN`?7C37m)&60XsG>BX7&ovdz8RI&WJB?~N}mEa*diLsCv~e&prlm84s&
z1MiBZ>Vf3%LqqJS<^k}ow<*hlGCK&89QUt`z6iva9rJb)+)u__K}zoNHO9>}eM{)o
zD=_kW_}(VL*n)`~;jzy|{jbr@Hd7D1a|YvI^<co77VoUv>W6-_rHwDq$<gsM;v@#a
zw2d_t>O}@vIASH2a>B_eKeuw`gwe6)I2x4#nl!mVY9I?z{p~9#9x^fh16CF=Y=IyQ
zE3pK<FcT*@70vb?aDryVAx>f5hrJmr0)=_rt_#|14E{!;$kplhuU~)TwyxhZGePKb
zF9Md3k75mhPg+K%ff~5H&@WyO0=sDu3@9k?NGT+RkGO>ZWXCa(c27Z%zU~7}66*@b
zQhqxQ?$xukB9dRf{Trdtfd>oMsY(i2eY<5C3>%D6wJ@?n^;_|!KT%k2larPpxBQ_F
zCiH1%-FRwJs9|bl@J6p}TMQ_ni`bLhn4F3MPjOLA?YUvZ-MfF4R##V>6_Pt$LYZ;Y
zuuNa4T=%l8T%tW?^&*4(B?2j-j0~1JP5yjpA`Am_DpJj7vrgE&j8j?zLP8QC&~et*
z8vh%ov&`wAMMwloMQcgdY_dvLOFx@g1(qtLS~cs2Jb39d{9Q;B9G!WVustTsDk>=W
zgC-BIj$?ku53xGaiCsA%`eRv{X&+8!O6yxBIYZxKbaf@VW5f)V4Q2Fw$ZTO@VF0Xd
zO$$IE)e|-$<SiK$mZUj&01&ru!bNZ60lu%w0#1fEj?T^ree+-{Ib}U?V^Q#+7=VQ_
z7-d6ueo%ztAQYcd9Tw;8$egv|5^*07+rKdmfHEJ1toH!d`q*u4`T}mX$qjTAsG}X#
z3pDg1Pe7N&kJ4=1n4X8LU%r+9o3~|<w@>zB(C=~I5>W&=x`9AL&K#GqBbI+3Mi|mZ
z#WBlcxP%H5N?pR~kM9e{-3Sur&zEClSs$id9*d4sux_hHLw*GEi%tE$9ErgKD5j_F
zmI0)h`1zZz3ob*oRhof0Q65aViE;rVCTXY%|JkzxKE3Ki8U``G0;+;l57A+BSnnh|
z&@8}@!avh4WMU512)fzkCRgxT_2|8gF~(Ou9UaXm2z%!OWZ4XO0B11;$v>0|&fD-?
zI*jwLSiB8sog0(X6`^yPI0%wY{eA;Ex7^CFubi=2hWXRnz5qH2Z-|})9V;tq$k8AP
z`YyLC4>hceVT;>yN3YM((NPNKI#+0;*1vz8Z^ck&BY5JeKOK%1Hkp_Mzhbol<Yr~D
zgFSiu5e*`e=68MD_{1(`@a0#9L0%&#O5_+5Q)%opu3)GDpiB@-xSR?poK*e6{d#%7
zJG@(_7gi&$9QZe{0N&m&C@APVffU4FAc|gw_A<}4c=J9+CMI=}W1^zez&=<=kes67
zUCqGoK|uwQavFzb_NBLlg*L;LkNZ_VX7{6{i?kACcgxiX<mVpW)*`op))81oY;0||
zbQ2RMSm=0-HwkZX2?@E4;vy<vQ}3VI*>cF*)@B}lk(yYrVw@D91J$Lm<s~^XeBo3D
ze62ioY9Gewf}3k=Q{SyO!Js-95qOBT<j6SgVL&#IMuAg+>SWxK_A(0tVs>@T+ewkG
zgqCORpUC1psBslm<);v-N^a7i?w@8R?anRND9m&5z?$)}tU5we)2<7JmwzJLxZ%^F
zJ*_qb9&)`FSmR?b-lmg57pVIjFx)Ft>sdTqMv^N>_$p0^JgDWwb2F?gfe83sK}_Qm
z;f^B-R33icLfp>8p^+tJw~XS;gu-OEj3kOoGk0bf2V*?dV^Ie8DqOScTwnowfN3)(
ze;KY^o&q;SWP%gC1&7hi<iv}T(?^_C7_c-`bVm!Lf0VQ^)&Zz0{xAb!9EOE(KS>z7
zhj^n8Kh>Uz*yw0@4bCz0^D9q1$4!QxaX*(5ZY*rMLWX~SL;q;f!PQN)U`N9s%B!Sk
zv0_NHE`VEraPTC{L+(x=oIGQb6?@y$%M0fd%E?8r&n86>8Hi|X*EIIy#}QoI@dQ<(
z0v;SD<lP;}{Mt@HZyDFOu3$9SYqqiWKNje`Qs{DB1!yY(btEz6L{ou-C2`&x;4oyE
zHgf@CdV}RXjSz!)djSWTji^y?F}PR{(J)lfwf7hfGZ`d$3m4!H%r_#Pl1s$M9WlL_
z&+r3%>-6i}KhZHU50E1oLDCZ_$X&dUVG<lC7kRQ^5%)_#vp?vBS>ftY6zW_sw3@gj
z?I@`KAPk>L0V=R~A3@LLhc?`}@%2+;hpC4Q77r=#Tm)$#^((sy33<I;e%JX<w;Ao~
zr|RmM@~ebn-vGlV3a_8ct{Q?9=?N_Rf*HZHw{g$oByLbZ1~X6<B$xaR7C{5L2*kjM
zdDCHcAlL!;cPAlwrJTeI|11R!NO_V<E%wkQGmLp@Zv`<;7;r-G)BRU0vL@>9r%GtV
zKcnI*&w!4Xf=1p8D^69Z3YPu#N^FeB0G1IDPEZT3kF!*y)peT+$O2RX?j8E}b+;De
zq(G?A4L@!olkb26mG?~?N5o>KJsuCh+E!q+ixr4G5Ld~y(1o=DID!bj06W7C8};(?
z@(fJxlK8Rx`Q?4ebdM^a?lV}+Nx|<sGOmtYtc1oEgW7=moc4rXs3fzn<f0K=Ig@PG
zlJ-892k`URV}7s&r&FXZUnW3PT3Y%k265o~dy29K^4$lsBjwm(-o26zlUX0|!(bu1
z1k4?s!<Yptp$YhK>m~Yj7Sce#Y2tnVGlGH%SQ*L<E4f!y3o{FtiJ&1VarA|pjw9Uc
z2H5rSD&ATt%z^d50Pqy{hKq36m$tVH^X>-h_Ve>&gqexpZP>c4p|7>V6mU?~xv{U`
z4M~oC`gn{FW<jaJwJmILqW+EeWL3|IhT~WXO)<|9a7_X!G}f4zB~uMZlO^wK!4Wvt
z=duVn^PqWt2DPl?25L3=I&|d7InOOmalX_<d`MSCzK-*)P!s?8?tlB$5V){s&)<I^
zLUyv&e~Q)cv;tI<hK+=(i`#!UZrJd7g~{1>4hAr|E6P%M4(|Wv3_7P5L+m&I{S%}!
zZj2CJP@19Hcl@+rZAAD(PS7;RJt03z<K$`^96$eGa)IYcV@_86f4lGc0-Lu|tSiy^
zkn$9~4C|J<_a4p=FV3?TaE(2~y_`tJ$J*C#*>?yhwmeE@)$x9GhdjuI2is9@Ibias
zy0d51XHL++7$L^V#`NG5;+OH&yxW14#W)KF=5n@^vvUQyq#jiM3cSjv-`~9$Dv(7l
zWs@YNlB%;P@z!gD)%OQCaVf<!?yAPdMiH~E*6lECumg(lxAqMTu%f8gbn<Pm@&o13
ziVHo$aGertG>mn1bpXKJV5$|+bg-h{iNIF{(VNZ~m8xjVq$(_CaQ0ij1*8>aCzTR6
zIP_dD*_8Y7qXfaDWKtE7ivbPx6J%!$V?9G*bi&L|G-O=`bO3KH*fyns1F1yPJueXr
z#>A;(e@`szY?Y`MOn{U6^SCc<R$}4!g{7A)-eI`Dl92AOP5HsRljpHq1QtaG{45{9
z33Se5$jB8#v3*}~>&Gk<FKe?%sA{h8c{zP~)uH@twN`7?66y7D^O_)y=3j?Wi8TR5
zE>e$>+cB_qg*a7fOINYM+h$nx{95`}NcT^HXG5s?y!-^EmGV4UHZH80evYfL{(|wG
z-UdDhcxt?p(5HcEu?-md{{1kTmOxNKGN3l->Xd?UNwgX6pSAOVg5HbSg&egSOtQtm
ziEym_`t|GjXdcF^1-e?=FX?XwJ;UHO`7{Y`D{@VRQ&M7LB8CVxaF43V@36lprf=c*
z61=TRPftJoem%pDzRid#03%a+UW~+%-naCmD|%znd}hw$uh_t`WU8%u#E!ZKk&kPl
zzcvV$Sv)L5)z!pp2#2x!dqIK~3>KkgZaH}Bip}4jP@AIBw}td#REKcnHd>4e!<n`E
zxIlz7>IH4VW2{1E)O_Xb+aHpPjh(%15`b1ivj&hpoT$;O^#DaLT3J~c!n1F$BT&vp
zE+Z|^+rD$>v*ycZxtzY=+iUHr1F$2N2d6~o+g5SK3})b&UM$R>N&@LjXvv`L#}Wy#
z7gIv*&&b7u>X_>>4Ty_yuIL-zyqhh$Dnhh<w5%BQGzKboh#k2`Ex*zQVQ13=5>k~I
zk}MK=aaXIc*fh8n;ndI@M>shz(?Ty-z(r-DHK9N#Ea*c&4|!et@Y8Mgg$-M`R=X}J
z(mfWD`fo2loNV<_pIS0JW5N9`nEeT8u0F#hKVOKteD$GIZMd+}?}-G!to*h*Ok_6k
z5Y#k=$Ib>PQi{88V#i8w8AGeGgR^r!0i+M7a3vvke#JaYA60z}Fav3i*k+@pPY|0%
zdico}_3rVv+Y_Kk#g83T=lObpGM)*Htc>A?Fg+LS8XZr*qWmseVf&;7+_P;Tw6^6s
z4q-h+r-Q16$xkC9xL|(R4yq#4RCfR?eiWYUv_!hr$F#i1%I@d-`|lzF%b}L`?%I^d
z1JHRfXlpWHqIos~>sJ7#e5BD_!Y3rTX)hNCp*qQ#B)JBIUNs`qxObA=x=;?}%m&FJ
zZ}IP{SE)w`O+%c1_wS3q6NRIEI+EOwHcS@2{ctZa*22ZX2vBDy<;0>UhRLsBT!ULZ
zi&>lDgk^G79WZr!3Qq>{-;s1Pj-S=)&y$Nw7+in-gQ<toYj(W?9p|^7{^umRu*~)g
zkM?~}gXN*+FbGs@HK$uT-mgA`hdsDQiQGa97}gp479yQL=qll8I$D5YE)>CvbT<Gt
zYmm}jU#A#>#+|2!jdyfd+t~%c`Yf}wG;gV%e*NAf3{~P&AA_;PAFM(JXa-J_e2|kQ
z!8ip5k0DrbvxC`$(;5nCDPY6uZK2<_i;>(vpe9N#QczQ8IR!DrP}QnZ2c1M9jrhWs
zmj?N{xrB$aHj@K(C1&`S_*@WwcBwJThCM+}>~GOQjv@rhxgm04wvRC1YM)mpH+7=|
zkFU@j6X1PuLd%i&^kKjuU1{}pI&eLRtztE55t;o&5iqwClflKx(bz@7>^QV`XmJnf
zbUnL{n<O&8O;s&ZJ391{)5lBq>fVQ^&7=s#R5^NXE;4d~mfOGr*E{k!<gdA!i3=%0
zsTRiCoTlV%<k+!OyR<M~B2OpEGXBSf8`tB5Gi1WSe@1Q}MvHnFjS;ciJ$U%=q4pH%
zI3mT|7zG7eQ>?27Fv*{r*Bh3WgS&u8E-x9X<bM8Y!;5#sZ&qb7hQ@}7l36Kuk<Y-`
zjJP)TW$y<ylvZZg6w-ydV7)hh-N$f-T=IgT(X;OY>{qIs#L^&QiS)peL+05eeGams
z?e|MdJDxd<flJHtmFK~}uwf%(m7S96a0^xD`y^N#E6W%>hJVdipK*XujTRI2<=MDx
zL6%&fK{X&95gZj<xWNSw|Hu(v;i~_Gz4r|2@^05fHMYc9f;A{;L`CTeg3?5zg3>$E
zRhm)-1Ox<3j13W_cd&rck=~0jN(brEQJQq6N<Y_4-Zisl&iQb@?LG5uX4cw!8A1Nf
z^DFm#wHtV&54MY7YTk<V>A(dc;?%S<wZl5X*?l56AByG+5Q;^}?|W}A0{*oo!ewbO
z=212PLtW>^8KVl}gYZ6TL+sB1-XS%5BV{|xLzFJilNJHO7K<xH&+h`r7s60<;A?+X
z7u@*c^=s!PCB1-264=!ZMX(q4JSG<SA#~zA@P>$+-#*(mS}rX*Q71=tI&yUmj*KjW
ztoG2QQ-(Mlfa^rI=sm&(4I-cn&|o+4jIqu)tL$O&{RiY70wR8c9Nc!Zu5E*f(?A^?
z+4!-5#93Hafccz*(k=q41-~UH{k0CuGyi1uDIm^os5OgkJpCK0h|j~rTttxImz2W0
zibQsgt=L)#M2{#lE7Txt;8mR<#+BGZP4%v=t&Oe6X!1+-e#EdpAS?w3_yGu|5f>1N
zkEDoNH5wMYV<6K3_jIRD5&uMN2p~wr8F0XKO+yI!A%z5{TJe&1`#DcscV!R|_Q0zG
zkN!b`3S)Ijau1RAN=Hl$l)Lcr|Iwya5%Q+2jH4)YiIX}=6k$;#jX{df(t6K;$dLI{
zlaUyCgx0ArZ?dQm+7cZdU4yI+rrKI4IS)`mmbSYeHo7Z%c4Q_5m;%l4$H77LEt6uE
z_h+2u$Fme$0SE*)Xd{U0SuWyxSTk6#BOLzNc%8;OP-0>haN4sH5>(Yt%Z!7G4a|D!
zU~iA$IR$+P5R(yzlm)Vt*f>c^S08w}2xY>etU=jB)Q|S0R{kw@+J0~^5|KikC6wIS
zP)!;THX~#39*7^9fo>QhmZ>~!KO&6aiH<~qE<F@N>W7T&)*6LLKw#$yj)lMyqN@~@
zlt5s}5y}Kv^%Ke)?>>5>84iyM_uD9Uvrq_P=@Y*Mbe8#~kaPv00DX`;R*=bOA)FPT
z?-8mRA9fW<2oeEhoTuWA#xO`OKqa4yR7gQEKQPiSA6*dM)j-&YY?v1nPX8dRw+!VD
z#ql8Ko%x}=yB`o&s4qP68;}^9uzEv(#$p~?7y#Ff<|JZ^(qpSos9<5i)AVBGGwja1
z9#MovNmV*)3&g1Vab_#!eYqBotrDm$4&%$gH3Ef`|8$(QgkGK1uB9Y%_Xj{>$3~xb
zV%EnzWWU+X`U&o+Z4F><a0S9`%uJ$6GJ$E^NCzC*moks-IZ*oIuxm1unBM7uIk8$q
zTLEJwk1p%*&l@%zD+%VOX&KoqPgz-5erJTElynN;T1~Jjlc6ZD7^@q3{-wB41F8&Q
z@&sX!&g9YXCipn=X9qZ_P`;9tWx2<I2Cq<c=n8nRSMM;E0x!A+QnC*Sg<OLU+OS0v
zkyt>!RqO=)MxF;O&#5phv#itJhU<CxdnEy#+oT<$ELQwWTM!%=I}Zz$+@upT;`hMZ
z6XQfxyz<};8ygjztkHnu2b{H_!B-^kPkVH?XOUg3Iuf~55)GUw2)Vwf38u%2T*%(#
zE9S7!V|#(?jw+ASybQY*TzLWeUBK@70!xeav#8MC#y2tA(P65NU6`89X;eCDr{L!9
zUTK+80YgZOmJZHCoGWIE?bMEXRE;|My79tX<V#U2xXo$tka@knu@3SS3KmHa=0jNh
zy9gWv(}T{Nuw{J$N0Gu-nj~hjci-T2I(F?q*C_JxMrPCEv~yIWf!DvQ8fH1?dKd~=
zc|c#W>S0+qhug6x`%#`4&QQw(DDhbipw>}UlnI7O;rm&jve5QHZPd@+*wm+Bi#Ys8
zipW-Y<naucUMZ5<;<-DT1{xDB((@>;Q2Sn3&S66t>@2n645825uGg+zE5%*s0|IW)
zTnP>hNTl~@_vlh=j8o`0M-k#FQN8vu`@sPdlT*2Y^w#7l8JrHpp5lcetOxC+YQ3i>
z<$BFJr}pJ2c(^RiNT35NQy2w~$zOFk&rYHII<C0-<gqU1XlBcu?+%?TG~oX74X<Tj
z@d#xPbEUI6GUEW3L}_W$_-yKqU@T4y!#x%i9zD>cYu?IFwmmlS5=8nJ^E-mD2tjib
zeZ?k^&#+vci|@G3AOdxR!xCZUBM3WXsP%*gFB80JY>2T^@py*Pg+P77ENXU2{J-L7
z-^eJ@KreE<r4m*i4Hfsal?k_R-!4C8uj>Beb&}Z&gzB<9_u){JrB3yokd8zD!FpbL
zHPm}GdaPtnfCwg*V}%V;8Hm0V5KRE6&jHA-c5=Q)0_S?rXd1TY)!s&-MEa{iIKiD`
zW+w7^&;pPIDyW+a<}53S%`0r1wsZ!e;TxEfsao@At5?S=ttMf2V0Eu^%7?<!$eAQ6
zP%y2MyiOHmZN%J_-_h)GtyRNWW8ft*>R|we8{LXg?;yw~_l2gp9~~4?X4y0*igTPw
z^aEJUU$8s5U=z^;6!;xG1ci71`Yn{8F<|R|SmAeS^SS(nhsF<|F*c4FL!`R;<gcx5
zZNB8s@N<L&!#iIC$Er<Q(GK*f3TA`gmu>v@%x+7a6}@d}aT^mgw5WuvA1*nM2b*!T
z`a5cA`3f=kTuB&k4awyRv#1VlY?KCwszP?NZS2a$14W1$jYP#!euX)U4M%=j9+X2?
z4huje1P7Rra`F6dbKldnmE8D@MD>}fQw;ozy?$*>44qe8uLyur^o3ac;e+!amQlDV
z(`^wJ7G{Mtrw(Qnhlt4dcQgi?a3SmmfK;ztgorFZK^vg=8a1$)M?;c;-{m}k5!g=@
z)U3Zp4C(Ep)yPHe)+Ps^nGRe6J$T50Oa53D(4j@bMiuyTOM8x;9Q+UaQ6`q$lA*4=
z@snGl5Y{20oDZ{jykaMnyw1%}Yod_^LRTmXI#Rh&(~N*Tsou~k;x!<u_k{LB*={b>
zW)Jz%aLDxFAqTb)$w0*)Za>0)2{U%zE5=Q67QI|uuncbQM?&EcBK3#<yvO2k|E^J<
zlj9W9`%<YbZ$Sj(x%>4vYJaAsT#}AUIuJS4(m}^cE>Avq0Mh-OHG{Mi?xy{C@?Own
zd=xonjU$p)!37PCpoA{GaW>+$a&bDx6cO5js81g7;sR;cu;QWAI{+ek2h8ZfrNvUv
z0DIyPC>o-?$%|Sp5m;=vFLyS9O?P)foTU#)figG`A*p8pO=dYoXDJ+2#_74BLN4I2
z;Y=6Cr<|P{4(`$lz-$X-7oE2L<fs{yXMt3v9t4RfBA4WTgPOcSLBBYtg<<X#kTzpr
z9P}3x`P&-Kk8~99;I#AyN>8*PQ5&RMgM5H0Gm^mIAT)BerI0LSOm+j4iRt)bO$Dxd
zN|Xv*%E`<;k25jVq)`@!4G!mRo^Z%zAr*iLq3J*f*ik-E0o!)m07Q;IW%*Mx2hz$5
zj3j-E1Gw3+Ai81F^aon5pTCL3FiCnvrTCLXb#qu(dOKjxeS&p(0j$(N9UZ#4xw!{$
z^&)D84L4L(uJ9195m3uIR9p^*$@Z3yM`6wZZtIxkb}9m305$s2h=a_`$|}K@*2lXG
zJnnlQ90fkjBUGO-&sUpf5x?9YtAgUJQ7`SThHeGK3{=`a5hC3W_!TSIK)i=h)_CLc
z5WW%k81L#ld>h?3*#!K*5uF3h(*UTz=TDz5+bjadBM8QQF**ok5tS1dVWMDj!4I?!
z2?2>c8JKz?qH`W8?D<>v>>;cmDSCs1Fy!b1`Pgm`in|)E0r_CS4#Q6<VpCI(eM3Wq
z4@al~YVY!wRdcjD@Vt$aF&)c+7sk-ovYIY*kjHYwKZ0>&#gns=47Hb(+ctDz*PEV^
zkp!Hs>_9QnnqvP3s#uC?G16nX@QC21IvQb(Of^Y-xp&pp)cD|#`glO@)TvXw6Tex9
z*sI(?*$eZiBfX{|Cg^Jwp>h{BtOa=aB4=aM0{SJ^NzknopMBkMP8`O@R6YqOWxbwN
zS6aT~bOIz+f;~`&sUXIU-$NA6pOVA;YHDf~pGHwPz5C~%0BOGKz>5%0`}09Au8N^=
zRL!j2@arYTBETTZWmSudsHwBw^XcaRdAzSZ^I_MUGyo<kYQ7|+!1ndF8e)8thRQWI
z(JwlZ>IQWCs_{}LtJM&E9iLEzb!2nnGb-V2;Ck+hBw&3pnrB}DCPWIS`l4fC7qxpR
z$CJ*aUK1UZ4OEPi>_o-m=O2IkAiAcg^~cw7YBS8De{!IPvtdsUt4_;0Sm4oFdNS=M
z)`+dYJ<v#RUQtm|m5xU<d5K;&-7@d(Oi{r4VyOn3HDZJL8im=vkI08L`s}3T3G3C3
zX#V*2s?p<4f&@*)X?{vpZLYCk(O@iFGiYGq>!&k>S*oE%)=Tg3+BZ1ZIWiJ9lVu6B
zc}h{Zuw7+$&OeaBToJp0oOvo2d;xr1+v6V+61UvrBx)8oFB}^`F(;z9?c5A<Or+g6
z!{Y`jmZ!c8N0mu3(|yPt25QzcNt~grkis?feY#+2_{r!`@eEq%Y)WysRE#)+kFfSJ
z82m8-`U5t&s8VBjy!j~yBaIt>FMXo^>{<Hp%7eRcMV;B>je6*eUy|itLmk`IYk{)F
zZTR&_;pb!JKoB^vA5)AE%%JDrE+;4FZ3RKK%*nk>qJIuz9?4NnRUl+>++|Gx&#&Ig
zTglC!0KOwW@DLkYiGJv>qzu{GqQ&f>^Awjsa5qe5i!U7F_=F!bI%7bKw;t3U&vW*e
z=!d#ec4q9K27qN^=CBxLWo1<f)y>JH#w2j}5(HTn>zyUSC7SLJ@nBHbl^JXf*0-iK
zPz(~DU{QgLAG82osV#&g+cK|+U(twt?+X~NeJFqEjlAY6Ep$HAa~hMLhz-cP3gHR_
zi<CmL!nNtK1ul7GCykW}ZWP$u17JrLZpfFM@b=j(xVb_8Adq5%mPv+BmugoK^6WI{
zB^kje20(j*LLcK9X-+oxrr{WiIk^MWf9y#qkej~P`E1Z5WalTQyg~_dm_|IR_-EZV
zL5n4Q(uJA16Ytw*Aqdv)kIG?u-P0@7rutuNh&_n+6ullKS{S~M$|pJNiLtTdUnlYq
zvQ{FbkEN=#u<3^3ZH`z#Y=(PcC-*n$sK3gahft>>w5s*kVjPMl6mh91hL6lS{6*w+
z?>9g0-wpk3*v+xk`2F>-PuXM6q#18d5;%n&;Yz4d&`AVHb1b2Vtk~lpyfOjq;eds>
zR9lUR$Q5deie8G4J2K7Tp!Wfa6X;`La@z!6(4cTK7^{E^cm{6Z94<?5UVs&m`XydL
z_L%c$md+i&LQUMt+FFSkA!-~@Jv`zyEdRCQWZlm^xtG!ta0OnQ;=<;6Asuo;p()ey
zMKT_F4?w!Q+?#4yRt6O-?9542ZKgN1!Y7g@eN=QjWdthgiypD3INXVmQDR*tMdWF;
zlQzi4zVhttvJH-_v`BWb{cb3*+zTh;_f|9w9)<=zVNE!~C#4h8eRy<rY!1jcj~>&&
z5zok4<@=)hy@}Z)qVBuT=sr%VdI7qLxbI?-5kS!m3%^pR!&uL&^tGDiYclcq^``hr
z(Lr61luVA{lv<ep3W@2_u16yI10j=IN&JP8^_S_h)7&V54p0Wi#1iX;Wc+^6YRk4M
zJIcZVt(x^8n%1Z55w&34@M4En8<YS8^(FH)*J`B9ES1vwq&@JTG&Y<tl%cU6WY5f*
zhp#Q*mh1zOd=^KjPQANA7&-@9vr?MIP%G*w*;hWHQoKXhU88{8<kY#dKDY*Ihb-x|
z6Fu^bpIsz&C_Dp3`a4Zs>osn33`Ug_z<RmI(1n@s(4%q)=)CZusCrNw*`)!i*I^({
zoI@O*#?WAx`mVcvDnMj-$D`xl_+gz4Ju)>q5T7<*p0tftGa_0aHZ3%{9BbLq39FuG
z@5WgArKX<O10dmVWKzxf?HXHQPfw45ZC)nmo#<Xj8jlEn&PNFwV^5OmlsBIt8;ryW
z88bs-_bJPsm*~k%PTp^L3Acgz(4C;&KQ#i2V=599Dx}F-^<!~aO&83spS>OXr0^do
zb8fUb<fDCJUDAa*!H${AP>2^aL@HZaG|>rq&n!Xr*NqQdr|qF&5c>08%uGuy3fY~Q
zwnyE<5>exZ$=rYKz?p;c*b{Jy(c%)S%6{Yf4~i-8F?p?fbM6xE;ERBWy#tX_5TCQj
zq&3&RcF2}w3aH%93<4_W&#~1200p4q&Bz$)kJxMrgn7l7T)l;CvKqLp>vI7Gr(|=%
zahVJ|B~yma)Wk%ch%S)VAU)0ZvzEPd|JQ&sNtvTIU}<r>pkjpREI86*{DQ-wbv{t!
zaGue18{3u1_+ZR0=@DC+@xVfRuRkm*g_e;Ofn;6;PwWx7`PAMNmni`U-PXUK)knc7
zc=AvvZCt~Q*u0L*130U_LPAWFWz6!nn;7sQQvLnDp5B8XouCiTOwhZ(f$>W6rbjQ#
zN5eV{h7*}j-ulR9Lm*Y-u2l3cKostjFEO{NjgOD>sc#)Xx@h*8V3$bg>;gqVmjU2Q
zjJ?I+>9R1W8|b2&=fH2xTptH)&a_5t4%tvQ{4E}X8-3{Jb}UD3c)WcN4r~4#oi2c6
zJ@_A6-v!*VeHXhp-YUt7qOInHGCh%Lj^J)wR!UrUX!T;MJc&V&hfv#7&k?}nLBySy
zhN03Uy-P_tD*jTO?&W(%dwQ0Ec=(=sKK2pO3)*vF*A>{QUr+oJQd!-hggn^~T4zj<
z07KA!fay3?w>bTUQki?N=7q1nve$Kmnt;R3rwOQHOO~31=QXHhp8E*ve!#p7io-Vh
z$Lw<^A6`go7R02V+Cc@qDibrQ83$nl_mdvU+=tm&3=4iu18UAOJ_>FsNR$?M4HP})
zh`O^-VRk@dMo`<x@bIwDEFtVatzQpf28MVmAkD#Lj~_qg&#FFcOCGE8#55Ol6sj3U
zorQnIh&@1o9UoNHI`~Xw2`wOhOEsd3>z&X+)RYJ7-a@=Z%p5v1R!}`P+6_1~xo~&+
zVb3(xU3UEb_1QeC3tzN*Cfu9EoMtDO-V6e{Exo_j`KnE5I^kwb!^Wd_0b^gk7F#ON
zrN7U>h^NT!uIG3#x>SLp>6BeAXi;%h^e)N$t}F9T>RwRK?8K@83=>Lqhb-FC#-`rJ
zw4t%A+9_?QXE5*kz?bpE$n)xE5l0!Fkj$q)9Q!a}G!bO|KM{#Le*N_-p2#&PTj3W4
zy%5KNzo<$Q1y@P=a%xQ)PC+QmueSR<1%F2Bqky(ohh2#;i%H+ct##4n({-$4Vt<X>
zhZ0LrAcS@Oef}O!7AZt(s+7@OwvG9Fz}j5y5KEVokx9ELji;jz>Xxn}x|Wt3Nhb!+
zA}3+%Rofxc*GJ~7pm{<PhCser*ikAOTd^i-^Q3-j@4NMQNwqdwQ(F&0=VZr>dnjq<
zBMe)El_tt0`+_p5JV%Ne4^kKsG>tS5((RM-dC2z{lM{1kvT}0%fYVE`D|^t1QU^#p
z?6-5C3nP{eVpo}*_81JuWo2jgW?l)JJk!+LL0VnjjC1GC^_vIt>F2G|SH-TBjfGG%
zP*5lc#2-n`fhnX!mt%cIw<o;@W4L3~w2l<xPQBtP>d$TLMB{UQ37mW5J$b8S1hT{i
z)B6pMU%?Fo*^l3Wu&orUqGH6_+S(ftr4n}P5S7uS?fRkxJaeM#B7pBeDvoWf<%!)A
zv@!SI-MbgC@{oznCzw#Y2PgrnT$pJxblX=*72hUyav0Ov<O18mEc-OBTHDx=h}=78
zW6}HfhV7SY_cY!43Hg4u{(+5gGwu~9PIft=8smaIw=8X1TnF|kJAGWIfp&i_etF@c
zk29IXu>|Xoz>(}8IRhVB)V3ZSPE@bu`48(rJ>fx&O~SGG8(NvtTUw0CQ{ExfYGo&M
z21T;r+__M22@M15&Of*&>yxKXk0%$rY*5>3spArlmqI>4aD5&nQ`k490I`qF$!Nf)
z@;qCvYRx2SrzpS6!`ct=#uEkKz71LoQq>*C$H(55sz{I0PL(@<{s~^ewY4oF6*WKh
zQusO2Jg$u0A~KFB4M-CBc7Jz-<dW*D+38V>WG57xg5px{$md1xKXh~TZ1iOK+4YA1
z9z((oO^JURE^Vp1QsFx@6#CDfM}~Ag5Epg<<nLh|VP&nhNLDJ3RZmDrAPVRjH0!vH
z&QSb}4RfVKyf*+IXq>%(N?kxwL;Y|~%QdSBd!I?yGk{aH_g@~HHNM@<<v6JopQHjh
zCjvJ#DUX9pJ92-Vs{J1=K$6~la9c_MQS@yBjo-GNo3NQpb0_H`E>cY!$04-I(Id6H
zk~|k&Rd74@3VJymd&Zv{mNHKaklF%MKgK?e#4fO~5M&mD?rA6=$6^T%AZh2rxf3-~
z6!1yz0pM?(5@C48W#BojrE?aL$9)ibZT%K7a1l`aXSWz94ow;rpETGp047fQ2Q84i
zX^!mjR#sN-91Uys6&h67#htG#x`mtLhi@WRNuPcT0@g9;_>J<#z`K)B9(OYfgT09B
z{;s)iz%-HugeA_p(3w{7@|CXh?8G(VK}l$!I@xtDq*iTf(ycuPE{T*@!k2mFg3eT9
z-K;T7<!~b8zi)@4Qh}rBB6Z}FQlN&@X3_8@jiQUf=M-WSS+8ij$i$A|1GVsfWl5Yf
zL7hmMa7|UQK)a==>JsZpZo#s<aC;EYPco)6dh21n3Kx3%)ck4z#X+%$+saRFQ`$Cu
zo8M%GftWzzed}IAxCCM7xCI8$@c!`2$jA~vSq3;1AWfyS#|mx-k*m=(CjO^QoBZ(W
zb-e8Kd$7o3gA%AgYdH0vX7}eWqHP|ec=%bTB2E#hK!RebV&6OOFomuaO?~O`#eM<Q
zB!6RI%GAt<Al^$A=hsNcT`plS&|Uq&)m3qL+rtNRc&hpmm0j-y@RQ$hdux5Y3ehw5
zvLmRa0`xSe8q3?-w2_Bp!Tb22iK6BineJ~t{uNiW32M5@k8yRCY7%az92E7#CFA(l
zXe82&c_WMiS(t*%e;<z1PW*OZIC10NlUg`{4#Kl(D5eA!7TJOPy!yqz*uy7d0FkKd
z!ln-<Kn!XSj?;wk0y0iJj3(;KnZp-Etn_td4p_{YXoR;_4kc1L3RSRrbA+zpC=l?y
zX5k{k@uj;sKj_DWv<;w>8(I_j7t~C;HKuIJ%SG=)xVwB7o~#2tuUxj5=<y_nm#>XO
z&){Ns_hW(xp7bhUMRf0R#IrQNl{C9t<}uPrOf$)S-LSX+1Iv2S(T{`0>efvNW0irc
z8el0Bqbg$jt!vHiDG7l<8Msl%`XOnTB-BtMJcR4E#J>&VDpTW>7gaCfS^%7z*xtCW
zWADU&%50^KkuDd-iBAkR@)Z3UM^Xx?YBaRN8hmv{)wJ~h{&aLTxO0D4@7JGo`OI3J
z!&J=<WN(@(9_)-ch?_(q)Q@4y78Y=Uwb~SzGVmoHZKx!GdzEoNl!D~r;*}J5i=JaY
z#dm3G?#Gtw^|VDB&T7vklUDK6_MrOkf_n(q7jBrwD(9&}+0i(b(v4+MtZ@HD&pF+9
zH8MMs9lhpE=(0pFTMRsrhKVF&(hzw^pmh7Y+j<d*AIUnFeRsI&uMJ+&u!2pFgzP2C
z1xS3p02y#3*2A)?5WOF4Apu&c6!s9OQ&3agr!cjId4@b7CQ@SD<LVlmOE;475tl^D
zn@dL1jikY8!fEM#s^|N+E}4eYwPo##K@CBc?nANtO%`%(4aI|t8VJkbpEikq9GNC3
z_e*n_r+<@w#DfP9UI$2+9MCyEbcUg`bQSgfqVfCZLO0f(yqYZDRo^gl^ibj}uYSs^
zjt0xp+bjZ-p5)w6(K+{C2?o=}ivTys34K*e1WuG<)5;{wLsC}?GWeLw0oMl=kDdt6
zieSJ++58+h6ly8nChUDam5Uln5@rjfg4eI416d%PD6hU;tD&s*xWQi*#w_H_TunIl
z%0}%eoQNmp-Mt#VqO=8w()V!$Bh>VQUh$8O7ki=>pP?~Us0A12RDE}X>TEt;XMIj~
zFbguFuy<7nt$quCi?7vuxpXo*S=R=DTqrKNqJ(LgsoVJtDi-@ERWNSja{V{;0wW-9
z6$$oU3&Q*bV(S_x;g9SeBpzb%5F1Asc8Cm{c=-D6`X?|76J673oDQ)f*C+{}FWeE;
zTOdGnw_lI`{dw~eE?-S;e_G;B{GLS}fqo6AOA{ftB(;U1%a`E6y(bjk&-!#YO^23D
z5z~&162~nUTkx`P(_@XIZi5xS$Bs1WILd$%G2?D&p6q9;sA-eHC8*r^XYH<Mv36fi
zoByxpX%<JXakAWA`9}gqdF%Za20V6{?Bp)zA!UI@?zej(LN<~psQt1;(b6O3DD03~
zGtnd+z<-FY*B1LF5#*qf6IK2RdVV1xL#dk1eQV>-ahR!Vwe2EP15`_*MYbmG{)`|`
zVjO%A18BYS+ZUPw)7A+0RBTJvdE=k63|9}=-7|}g#Iq{}@)Vz~?O?8nc4t5(6(>Qg
zOT^VRNi<m9a{c25BROZL<OJI^zB%nTPq{JC3CwEvj0@=DULSR-<&tAT5h0j#=MH=m
zABd&TF@5aVIfZU?Z-GmL{jm?fxNL&F#7t-EHy$WzB>|inV6LVWqHVo{)rjR+UGtr3
z`QDwag84`vWk$Am+Q%AQ>j`^Gbn?EPDr%zIyUuul-)I^LB<0dg<y88iC<>R1<zXE=
zaWfCvgWytIf%tRNIN;1}bmIXHj6voCwZ_%3b${ZwP85_;7)+Sbuv17gWUg<0-rxdY
zrL0yVr1JOM9Gl~G@-(KCz+RNjx8PctRLxUQ`~oS7<}@<{6iEp=Tuq50y^|jlJ&XQ`
z0MQ(1nuEywu+)Qt+rz=nX;6L?hi*e*06$gLdq103d4%Pv0${F><u+EzX1WF=;f|e&
zb#DRoSP#6xid1Qo$b=GwJkb0WcqYT<;iAiYs}lu2_qO>aBQDevZYD`IKKhtL6I4W%
zTN-$vY6d`<L266!%y|dLP6Rx)H3|XYQ*rG*4y#%w^-I^xTw8ZAGlzC<I?Fg_Ezc)L
z^Y>&H!!h^)=y&6Q6$B>Mt**|9qzkI8bgNT?daTxIT`St<z|~LdoSzsjowl19GtllC
zZOy8UIhT=ga^1|cMAosJsp1WH{c*~No|G&KAx!?7H#|lkd`D4V2((2uR`+?rzE*O$
z1aZB)W;YEQhj(OEt8svkc4Opwzt7;><TVJO4gsrCX%mBTqCQ^>+}`+va01BWn3cgm
z$4kIA+W6ZbT9)j!H!vJ;JkNzdn*3%g<ZF|9@?FT10NSYBmYDwj8hG2@!E04MmyHx0
z#b`!Pv|T_I$Lh?nLQokuY#K0zjc3UtPuSy8%*fKx(iIg?Cx+8c%C=qK{En-&RMNUc
zk<W^UC#5V-i+wyMHuc>^3}<6aId@Jjdj$u-HQVHDrk){pWUpV;ZI(y3e|vi5>*7MS
znzMV#+YN84ixzB!UCIUf7EUGx3YIg4Sy#>>-ZSsBCD;@h_x|hSJjnp6RGk8Gh$)OV
z{O#B3@Y9|6uJtD*5qqb^la~Fd_uEkfYsT&3>B5RI6gyCnH0@jmkrAl1#GARMO}wOl
zjWo6vfYz4OIn@TomxKlm!&D4@Ht}O*B7~yt-N`_6BvIXMs5JiwUAV~C!d0wG;t0Ww
z-}A9W0XU~j*la#7K9W^H^<4Sogf^zy`+@T;CLjci`x6+2-wl1=ew<i7h&?ag;I+HY
z==dFC1?hHkVF<)P)yPLXSA<UTJI~)#hd;Gl?7bwLIwmzXorYScYqB!+7Y4G9LWW&2
zP}16`fgpqS><pdHhHlXXnrdo|k{t~TvJ~Hb9?O(tPhU?@2@NW-iMsU`Lg)$Zq|F;Q
z`ugiQl$h?0XO7K<lUmiuA3%AFgNMxH*#6<+_~QcYDB?M%v~_g2$*9bgU^i%M9>rE<
z%))}Juaf&s=R&H0zEjuZ<Lr4dj-wLCJ%V4=_HX#zgzG53p)R@*6}_KT>cKY+S+Ob*
z#rN#~9^R~5Ff5T>Hc#&{^`{VpRooMR`rx?3`11k%krT&|NPtN2E64P?czBjOYLF$Q
zfg*Z0<<6^)C&9rcV;AGu?KbfD0)`S0H2(^?iF8j6MnS#!oFF`|jd|5Tm1N)f$GVB$
zckkXc-Iyn0;Lb_&DN=;Qw*cgMR8?Dd)Mx+u@;C;c?Erxy&i?cH80^y~+163O+578n
z024|Xq}$YCu&=jQBP|r1W1cOINT_f5g@M#bq`ge8KyzBJZHEnM2%3?#5gJ$(v~Al5
zVWbq-E^uz^o+VKW)ZfRW%323${9>L>0jN3eLDhHoRP_*=$LlN$hIO<_Fj+mxw4hT3
zUzbH5;QGMfX)qr#I0n~_zJUQNKBj|Vt{O+cWW&bIo7Di33CxL@`}zC7LhNQVdL=h*
z7rYZ?Fi{*YX*8$mv{@|LM8<^>8tpE#&^jAvH4e~`)q5D*K^fhnN_Z%OA2OPD?XmA>
zc;p2CYz#nM%qKOlY@_Vt?sT&3C2IqZEU_j3U{eC5c;YQ&Mc82yubq=_3^0U1ZA`OK
zLc9V{_DID<JwRMC=7%^q%1m2gvJs3aZSII#-gjggjcEG!U5HSK2GJ6Q!6y=H)~(aj
zS_1tWH#W|nAqK9frM1H(G0!kH5!HrTma5|bc4SkdEpX6EZUwf<zB%lJ7iz$)+Gsuq
zXF43-iY{|49TUw^BS%3lTz#b_$T^PWN_Wg?T!LZC{CN0MohC+fRvI?$h1|S-*3W%A
zQmxP334(-5KV1ejD@TJF%#KYq9D`U@1BlyJ0yL6jyx`@v1#yJIuGtvX7z=9GEx~bo
z?g-eESQARFW^cyGn4q|~;JYpM9#bjWNUh)-#}DXRya&~aoNo*oRX~oQeC=9vlfdzo
zG!y@lt_qObd3tH!lP<O;#^YyhIQb}Up(V2kGB9c95u6OwrY(pdyg09udz9w0>x+>9
z3pZ)Xz}yKyRY)#u8Olaa$hxG~gtMr`mPU{_^$DAPJmPIeROD-COUopcB0s#77_=zt
zD0eX0o5R9lR5H&U8Hh)l%pvwO4A(S}#tfICam*2t2L`Q3#3|4y!^Y6#7Gc`PywiJV
zPJqCq%EgOO5ZJg)rJ<Y6sWXv7wR&u(6$GZ4%HGvlvOi|h$fvVs@7}!|jYi;$Q<9&b
zUkvnBu_(6&((Dl%$=OgxgPsI7pZxw*2taGQ5N0<TgXv;N!SnEeF)WewiE0W8p6z?5
z+Uz>wp)!uUnPzI;<&h1nZ~%X#0?{GwY@`z|%}H2BzA#6jQHa9Be6b)jn|Bgg+jg;b
z9UUD=4%FGj#B?q4@LNU^L5z3A&Y%GH(qTb!_jN&IF&d$v6HHl@u(!klS#A3=%gD%R
zj5d@qWVf$ld->-R&83qCi?i@KC^w+0f9x!qg$0INRlQBlU=??yj|OQ2!$bZ43?#uH
ze3`Mc$;U(|KAe$!RU6o<YO_{Ll8ql4r5q6+6CDYoHCj;E31qcFN@jyrMsgY_Yy$Fz
zu4FVGfd}2R^AeZiI<1#K{0+$NRcTtYx9Sh{kDy$CBX$wc+5oII%1C=Sc2SIvPl*zH
zG5Jq?ZFqD#(^7r6o^SA-hr2tHSf0ZRbNRZdr<d3B<$)pDcyD#kG8o$KMz@!u6D(vi
z5QVhU^568ykx8ZI76#DOq6l8T<0|0y1Es2KTkJ+Tu54gnpqjU~A^Y3$5p249oA)Mw
zdWz@e<=xQ0!Na2lyeI}1EC+r#GT!TVxNpexQ8Y&L?2v}kPi+*NjB$4OKV9`!AY~!X
z>F?&-WeRf&R#wUD*RQWJX@opC4f9M@<mE5DcD``oAx)DCLXhye4_DcL%;M*C1#+$_
zrGmMAyB2b3E?m8n^1;q*5UjIhx{Fup?IN-+iL*sQ{c7YtdIjAe^z#DVEzy;?rPw0j
zGBM?S5Ld43{X+t<npiqbW(4W&2;432{m9oj3jycjDLPDW8V)HYiZWbhe0KM?3C;qC
zy;~r`j=^SN$8MXMo7<h6fC@~$si|p)2?egkD7|t&J%=~$i+KnXukGInKvM}=>EACK
z2OD_y;qy|&a6Rs>JY7cti+0oc!K-dbVl<g47O4d`Rie{TIBO}mJ5Toy3uz$T`6bS=
zU3t#%JZ&@{o0w1rW~mAtR1Dk=rEP6_SDJJe$9AScN0~PnzcG_3ZEyb#Awpys7j|!6
zzn*3T$KZsc)K_sA^_|g-EBzHE(U@0WbppT`z5MKI`frF=)#TVDZ!13>Sw|**K!kC#
zoC5qn3d%kw4%ahZ_x16Cs6||*-~QqJ%;rwiKIT(df*jxpt0iR28|xTfrDFntJoH>K
zV>F2Qj7@Jjqp*nH+*&OZ%;I9B+FDw5pmFd>^RBXWDh^bAofLm?V{qH9T`^EBDPt2W
zqZNM~7TEde)3Ze!%^y=Mf^NlGnY<>#InSA5kRi<7vwiDUy$NYW>!NkXcLQ}ftqFvT
z5~%9|Wao{l()&i#Y+jX%)XGLYK680p6nu1Bw2$uvw4gG%OK-qJ1n4mefoPCyF<_K-
ziSD7Wgr+VAiFI1I;&ZB<jv)6IPVHErm(jTH`neoO^QI@<<>(n_?WILrXRqhtfvEy;
zWqBYB{SHkSK)b+uG4bVUEv&8M**u_NqsGRpVrQcy3nP6LaUNt0IMyiA|9|;rALTST
z1ek_qFE$Pd#0YmPN=8FfA0RT4VSpiw1x+b;WATRrRUGo=q1P}b3g1<S<8xTmLv&RF
z>=%r~eK1HSYvxUzEj0LR#}5guty*}w+}`VakIUiY=AC^Mzy`*-PWVMh-2@=kSX5b1
z;HoxFI7CK7D2LeVoAYk3x%jqqZ(1}>#-@|ka4z0`dKOD^0G`YAm|hAB#&LS*&RLUH
z+!7-al5A-JRqyDo<B~x!Jb>JKe*3lOZPsbwY&fQ2gq;#U@`9b#nG=R6?$AAMf)i^|
zq~bf0I`|x(8k-YcL%Bc$r;@T)CG4iOhkCZJ8L}vb<#8mXSL?=E^tjy~L|x9h`G(W2
zTXBm_$p?6N<dzedcs!2q11W5M$+w#;$*6D=K{>w;zrvkk8{4Rpy5l-qq%x|_U053x
zbF%nlxc2Y`C@OiVm~Ujor5Y#1in*@oL^0M^6+}f9cS;NBc|)M1fq;KS$);`i{yNH^
zOLDQ^_xND%;`<62?d%&#H^oC=;B;SXU5DD9t6>nCD^y_lDEe-qO@|7kBQa~#OsN(%
zLM(Q@UTW3Arc6mGDcq3()k4$5Ek-JPRW@dpmX&Qxh&m#mzLowVtUYDWpJvXw00dsr
z80sV!qt#lH3-+v|;KEH8^RGjJfZmWltE#PhdGBiHIjH0TNZ-M8Cf6QQw-l*mZL4~c
zrdFt;s%mJJ=>r6@InB^9Nuu!GZ{xC5C<{7FL`jDV)qgovhSQ4_wj~gpQH+V|bPMt(
z=)c1bR{G`CXIN)^0V|UCmfdFecUUQBVASuSNfI9~Jwx5@h4Y~lsqNfQPqSLI4WS_<
z5honHXOL{ng)1cP%s|pP{238D1fTR{b5#Ga4~*<#N)91=U=7X^cC#vlInUzoI{{(B
zcCsgDzw9RkUrCNB3?}CRPR1b!9L=%;)E)d+KDaO(ow0k3Lr=1eKaO|pCZ@w!-pNwB
zxY&6S3q-T8a{+iQN2@?wTWy176M?dn7q(F%FsR9jX?ZFvg`Z3wAKX}6D1rFrGBE0r
zgIH|?SO_SiZKg<U@$!+^sYemvctofG4`j<PzsUDDR|JYWrkCUdJZ2Vd=ze_0w5P9+
zsB3R*4N|w>7;df~3KhxA&5gj%B@&_xX%_n6Ct24m*a;}v@dh2a{5!vzl_O1z;ndDn
z2Bg4F(ns7u!Sits9zR}oyy3-=<BgbPNi!vmB6hOEhA+)aU=SMzRY`9KKjUA91knrJ
z@t2MBW*6`z=vj~9B}u{Je!2kLlDrZ0S|=dgP)wut0g}9f_*;B}53E!uCy0_<wz5h;
z!LDa)4ft#buOsQ}XV4PqL-0TrgPcaaSYVT7q<N%bnj$bSWkHYS#w@G~7UVa`f8uM>
zvHo!>K$G}(Ghw_lgStz`V>v0rhvQHVL{K`R;8^RbdgdQK0r;i{FT1?1uON>xhr>~&
z6=$mz64BMM1vh{L06Qz4&;nkQY&zY5c)ZAJ1mjiP!0U`BPkxyhh)smgPQCqR*T#Oh
zq0luJeaS{GG}d`%b16_AvO`g>SL~FawcW)V-OTHu)!?AZlyZy4@a>X-o7@seI)L5c
z2geM#jVM1wwt(R_8<oNi>juyUw&t%YfArvyqMyYesZr0<d15PozL4^mwAaVRfk?Zc
zs;c^~i{7H>32@IO=2jLN!H|*6oE*(oS(Gqvlc|Wf15yHP^J35$3X(Bw0yUA!ii&#S
zr%I9+#KrGXIt%cdv*O}}P`um4$;QD!YpbNPl1QM37?}9|&=C#}DbN$D-#*q1+QgQZ
zOJ^ZXXWNb)18vdN)Y;A%U3gV{K~~niySw|mxjD&FU2(P%G5u0AShdK}DRzGTmZ0NQ
z9KHz#=GRyVg`JaJ+}t#jK}AGhvRTkXEc;nR#MzqK+U0-`2=PfEOTYQLdoNpOr;&kW
z9en4wrVJe@%uxW34C+HSS;Ws60;WwLn0XZ$$=KTe)v&$2-8X;Z4Samc89^r+@%ptq
zJXjEQoPzUh!hE|$JUJ<sdsM0O=Z`Szv}f<_;6&Yz>ekGf3o(ISzP<<X|0t@evZAjc
z1vZ$7q^v~fJ<ecc_r_jF#JN!_z@B2|dT818$EG?^@!0;HDV=rm+w<<;Udhv^|M2nm
ze}v#0a4Z*)V{>M6ary9OtQH45Q~o|6r8OwY?#$pggM;hgE<5=q$Su69G?zs-V;Ik)
zqIb^M8i^7m<>fwj<kkH*5MsgtlFD~|FlY3IB6fVz?RNFSu=^6Cm>B;2Im-pCgOCs=
zOmPl_eg6aZr)u(-LGVDb9s+&vJnuL;g1X$EcK5Hp&QrlPQjnUe?B(r^X_#{OYpRwK
z1t`h`a=We@Y}>X?nJIBQ-WhgqB+hMtbsAvhV=}GP@rjU{@K1~_P=%Q9MwPX-wF4_8
z*bXAOz&9?A6H%4^__rXz<G;C?MEvMySgtIlSvNS}IAd#@u5Vy)0Sg5QXp#urL&BK4
zleGNBi-$q@Cf>Z&wT?l8_;mUVo#qM9vcrpM{_W;cLy$%9;9#s>-$iRE1>>g~En{(>
zsLm!lB6MqLXvlTR0P&Las~Wz&tOS}vroA66hZX_>q|@{4#qw2vM42T>ys9k$7kA+J
zaW(uMgv@AyECaV6r`S;qQ-?o%NHv!hfmQ7zb#=vp$~%-lq^R%;4dsI6lUPtZ&LcBj
z$$0^e9%$VRCYS6eEC&7NujRUsHJ9Y13bjBMxE~G~aevGbTo(RkZ8eb{uc<lc%CBI+
z^LUl{duL_Az`#K9(W~14blASOY1NU|?j;|#k^fpglBrBM#g-a;b&q%_we8zd>`<g-
zVZ}!eb|YvX#l?>MbE<zHs*!)}_IMl<3O|3%(e5zk*2=szG4c?eO#a!Wz1&FH{F!wB
zcX#R>32>n>&^(8<t?YmDK_u|*k`7L&9+Xlxzw^Hq*U<?hf666iR7n(DRj>QYio(l;
zgH3E0)~veyc6bm@X)8?CxHALWS=2<JwOtHg4@uw&Z{|Vrstv5)%(buchwCklQ1ZL!
ztOssT?kdH9eU2S48ft+43TrG3y7o3P2Ctnu3crPG*{xUkr91vd3t&Elqp~qmYU|1s
zDqCT<z<}(IjlIB0q;U}q{^cs)Mhf{O7KOK0(j*q)XS;^8k!wI8(Asg(Q9N=Y-~QUl
z9;3sw?}yA`H1#Ebp*QrR20xERybC`dlx~o$tSpCdvTnu}b)TBd!B42x9p3Bk<(cH>
z?2Rj;nxHn}M9U)(04dWA=O?+VT6bN?sEl7X8eeM1Jx*1jMHmd&$X-np@3>y7!|CyQ
z<BE(wAiSYbF>h8X;#v<?W{4?^Unh+|-kZ{=qIX?<k(yd}^*eNb)`yf=t%%x$%H4}g
zGrbt3%tk+g>0Sm)oz_JCW&qgAVH||}TT6$7xJ}h4dg^$hkogV;rC+YE%-oMNcGs<s
zU6n^J6|O*q>&U(bbm7NOc7><w5ie1OdT?+Efll>t<UNH`(8&J}W84_5fES$J=~03)
zuCT2fH3_BdfeW^$5n+}YPjk%>o-{z;xZ=`T-~uz3f5v}P*$6~=(n}dl!y(mG@mIut
zSW#$+Yw7@6auyaC_C9qRKUEo$z&wT3TJ<!u0#fEs?X}0;7m%)0^vUYikE`YwR-D+p
zocD^KefuQ?B}n7zRPB$HPK~dp*H$CrlfFJ2`vF;H^mTRDO61zn*RjVGkx5@c;uN<=
zUo!hU|BW>MulRA+8Ae7pMTtanwEYRs>nF65hW>rs-Pe&|g4d;?q2W7sT|tsfii2b=
zH2`?nEjIl7s9ra~m27~CaXT8jwjj=ozW#qd+$_x$FH~P-uBlGLH?|5?BAy1vmcuwD
zp`%oljlp|9<w(MydIxSvww_MP_Z>Mw8M~w$bQs4C41F8N@?Y$ndUx<YX<GmHRsYwo
z>;JPiv9+Bo9aI`L_>z<jaT+8UUOP?+m}o=NVw})R^#BY=jij)BVZTR;l7czd4J)`Z
zQ3_IsQfX-^Q+_-^s8X00I=>Fn4%(DSQ$PwO0f|%Zc;BL^qCy74Vut?x&)mz0qUjNc
zknUS_VG5KtQa0{eU~EC>@=?9w(j9WVLArA$<0z(h-Je2a-Sr$?y$IqzjhF-20Lv~f
zG$LFBEMDj9MSKQxiRJTEz^dE875ZETh;8pr(zgk-EPq?|eh(2A)Rde0bA;Z1#}CSE
z!$pIb|6<Syq$Zqprsfv16WdW7%AK|nbPmh^<bI;o7`HLSSmjLPjJ}?;{6?exLJ@WW
zJ%#pM_t;rHn$qRpE^0NPGIG$h&Hnt;PoC@c+$=O-z`xRZVfiDomxLK<i6R~F^T>#K
z#v&ae2bO<r-#WxvAol9Dl&%)V`pet@`^VjojiSO*8F>YlDw4e}-KUdfM0)wZs$BBm
zOT-jPb|@-rqmZWFS=s&EN3oI{pYBu{iHV@(NxU(KB2R^i|DxsZ<qV;rM)Kz%ExH-m
zB4E<sMk5oRx-S2MF;~BT{p4#-<c6>+e#~{F<);_*91?}|5DZ2(H|srdpq0Q(SJ!Q3
zWy9AqP!mBXEyE;_$o|66yL6KkZo)uA7ISlRfed{5Q}xUD;m6b9!KkAjn4|z1pw*k=
z@L4-6>3Uvq=}>xzXFW*6pML(?`VwVc+HG3CD1{kCZoWn-t&zs<e+a+R$IZ+Cbm;2U
ztIuN=1!sa9vXB{%Tsp<CjEmyB(QnJ|d%sC#ImTRiU|@sVndP6h_TQi8(CsPlPfJV7
zkkImzzNEYSENx+5J_t$G@vK(lfkM%%_INwJx5ZZG`fn(EBA-4jRA@sYXWd0JghnUk
z&z62khP=8Mf~(jp7CefeeMk-(2p;?R_c^x6q4wAZ#Pp+o?R$`bAB3-`rxDD%g^}^v
zv~$+~49WR#z3u-uor-owRBxq{$%x{cY(s$~vvG072T-0qie~hQ`|*;z{QOu^`_(HS
z#ozw-V0l`D_EaC16O{oJxyk?)_W=MSZ5rV#S}1L0w_|N~$`;byUl@(9ECPry$?l6$
z<m;9c7e68C3#`Z~QdmZ}#p(39y^CT|C~ys2lMhN+!5k+1lUG;J3Xk$c(uqFIVabq)
zTG|;YW0R~1SS}BZQ^|b{l_8lZn<DY&Pw##1lgGb*1T_>nXIzVL)&XMs4dYU45TT8f
z@${H*aPB4R_TKspyVXiL8sGNe2yKeaBe3y#p~Km&-3#Oq$gC$e^ab1msKMTu>QW>V
zmGL0NW0jS#i75b?T1ub_5HBADS3{TrXmUTWA}Y}?CNs0Mk|^;h`vucvsSzcW41CBw
zfT7qYhq2i}H2Q*_!AOv8D7-Bppa1jEA1K7FyRWYdVCez$_2B+_zzW#f%L1>Xh#3Gj
zUNA}^ee%iQfB#-mTIvO>G5qj*?TWGYH!U9|A<yIDq$%GE*yt0S(j_4Dx!MISA<?4u
zh4-CyO%vUZ-%ndvDO@T7kiTK0sCBO!X{Jto?_32;A~j|S0l&+#$sS6ly0dPx>x1*b
z0%qM91eYTMR8A4Y7Vp?V!o~^)!k>6NaCLFb`41kzL%M-kNSSi-AX5qlaeA>{mhB)G
z7FD<82&5df0W;SJz_JKBC6OyYi35OLSsMv9Ld{tMw2i=10W(~Sg`7RV0j*PJ`5O^F
z_`0!Mjwf%ztEuY`wDX#6YPh6T(B8h&a|5ld(PjkD0Ybq<Wp(wyiXvVBwkj}RXTyM%
zmKo~2FdH~*0)+_K^zv}&t2P3DNC0+!xqdu%Fd?T7N1B<ldqy)7wq-__^f03t??E@(
z44`M#Kwr^+euX({d?1yRhDYz6?p?X!)Z^=rG%<o7csKdpo!Lai1;4KkDRzOl|5IN-
zbta6|d5FS0ND9R7v$M^ui-x!$#$Y2X-EEP3*@g@14S%3rlDOcTCMib>JIo4CP%T<q
z6A!T%GeRkK3|jrSl9SZ{L5Yti^Hya&wvsTz@}@aT7&KA%O*YuEgvJ{-l3)~gKxECQ
zvI-pShX9b7`Hp=44htLodZy-Tz`zJ@ZnB|REdrsq;uT5hirkc2zH^im1cXEiwumA5
z?I<0*kFKQ+s(|dQ1Ir<WJ-C7?coPl={%jGX6Cu!#2WS}U++jcicyMKS0qj`H6gdSc
znIA9}M(w3P*6n!$n6wyJtC=J3k{qT&MK(Wt;Kn4VsAJ!=y`-#+@8S+5Wg!$Vsw92Q
znl;H_U1z`(L_|fMhx3$@20-C0$6{k!1P|I{h5?mp$8suxT<BxvQr0^KX~HO`3oHBM
z1duGO;Oc#V97{rLRxycV;K73<Yj?schcBhbLyqomZ1ju~`t+{LZCM#eb9PhAXka$l
zXX6jt2sU6Ax%oj~I8^o(0K>=TwKczL9WN^_RYXX|UC_!eXMx1l<B@Tt(}axbVSo}J
z^n>9!+y2Whm!{I(F<KIEyb^vtB|MZJMV;&6W`&@EHA7uh^(8Qw=LjvCY$g%FPMw=?
zhQ903Pp+LdV*U54NX~uNokzDHp*F%-bdC4^A7Hp9gLKO%aIDAaummdM>|$*JAs2;5
z<J}UkTaZtvTgem@M10Ma&9_a^FoHRfN;CuRPc&+Iy5{Rke^d{j1pCO^0?@qmWupkr
zxJiU%4+#P6IjY$Q6Rv6cg79drUUIzbD<zyGwzDTI{@oMZA0N_QC}i%!kz?pT2QoWU
z?0V-Z;>JpBu}S$_rh&`Rqz+bpcnO1sJ3{aSN-?;ZUZ}+)G`y5rVb}tDY|WQj)~g!q
z5O@7@7HBS|L9qHG%_4wVf)g-3g%zZ<+ysA0Atd7dxV=i?fQaV+*3`L!YMQ6-ORB1l
z=d>yp*I{(Fx7iJ$&|Bjb$9H7pE$4~IoIn5i<Nc$!3bsW0R2wW!yDS-{$43jZ85$Y}
znNpPG@~9D_M999!>|MSKAh6-O=_T|D5pi+lrh9=u(U2GKGLQg^l8rd1Moc%*C<tRF
zSiV!rNZ)(I8CDyabhg8XpFr){^f?{D9~In?VWzx0HK0l=2?%UUyT^c_({i^r42o{l
z10i6TA&e&nQ_H1yU5|G1kk*?7=T@DCE1Te`2@!Q{a5cKLn6{naAvhk~U7ABht%XxK
zb+!Thd2|{>y9(o}rakp|E8-Syvyz0@L$nlNwiA~kUPY{x85DK`88gtgaV9817s1<U
zWp~`U*cKon$>%8?B1Q-UvkwAb9hHW?R<;BLJvV+Cf!XZ#OCv*E44xRnT?~b9T+G{T
zhy;Kek=N017jE~XS~NM+*avqAZeD?$N<zST0kJ&@4i0`%8gq&57T^1ec^T|%Y|+w!
zWDsWqJgcnfDW*FL%Ao`#SGmq-^}s%D#`-j7RJD%F?FD}+fG-ovP#j<t-|hy;2PVZq
z-hp2qH%Fiqg`bRlII~%O`}4r9yLPeTF!IH#Q;rUgi5a+f8SA3*Y$1vWV;|gjfl$Qg
zaN~qsI}Ynwticc~<mACE0lTyo4ShFa>J1QRE{fYHCl+~g>ZjcB4BX3u_{+W|1B;ey
zG<sB;n~&4OjT#DLH!`X|dc>C_3vYb&(I6j1LX&ZpinmZW3#M47us423SfX;)W{Ja`
zoIWHLIyuw56}k&gAoD~XMc}h9PsM;d46=TWp+20Dzo3=7^DUF}@uNxH70o?q@VUiW
z?WK=%X8Gg%1yxEJ?97}AM&_Z&w^85Y{QdHaG8z~M8due~(x=81pgkeObi-RXqX;9Z
zJtf<stU1=)0D&~*68aD=JB%ap#weww6{&^gp~4xGHYnDVclgtEu~&oWCt&5O{rp{u
z0`736H+PRA3L{txUd+Sbdg5<9h9txn9GyzGlgty;MC90l7prK<;r&_%G32Fy8VEGy
zyYkNi_WA}~VYIA#XYmm2xR)!jZMYJxa6^{hw$#V<k4zr@qzZJ_zePqyvic{gr--9B
zH@BJrOlz7i6%HPXl79%X!t;!fF=MVk|FY-S*VC3ZX~R#rg@s#&ZW`buTs!WgNP)ot
zrdtux3<rIB^d<&Rk_8Hww;skP_r|~2+=@9Mr~Z=KhWm6C%^k)g()pGNGLtBQ|D}}7
zPNaq(^<YJU^{2nrP(Blxy1$I&jATo}lFtDvdG0(AY}R#p(;&L=6H0JxqxISoon*9w
zfNQ)oN|hr56#_$FFou$sFFz|0_)kbNZIlul%n@^|bU_VZ0B0Vnt_KcC8j4O@TGpy5
zD@#jHl#upN%NRgoPm~j1)i2oiLUjHWUWSvwfJ*ienrA+Sm=BG#5#Z(BNkx0SAx8XF
zyB(4ss>dQ;zAObLOTQ0MNg!=0pORKTh`6R(sy=+@#KlXYoG=*cE+O6wwXHX0jN+1w
zc>44dR80|h>10TSsf*^SRO#zdy$AmFMH#nF?%~ngad*^QZ}sPJpPj~oI|MP8FY%oJ
z;fVkx5Jis{@vftzptK6-SEu3fF)pr&1OW=3G~|iS!S(1zcsc$x16%1tC<$N=PgM)#
zzlY~_3<fQLlv>eK%L!?GEoT=+9&sSG3rQ2#te62!P{I_Je!L=e#2+Pv#bka!<s^yU
z0ZIi)6mA?sp_MPwA;pWuga{?neWOnY%;k0qm_C6@KwZm%l*QLS{cRkD<L%$a{8XHZ
zb{q?+WUq90EG^na&e5St^TQ|{CwLXF9<@uvtXYZid&<7@rnQ44J$7ixcQo>Farpw^
zmjwpH19~lfYzoAK0l;}hWo6|aoG*o;50@9D5;RzXmY@GZLY5k3roe@CM=Qs+g4>Cd
z*H9$0EqyN%>&9SkUu5>t@H8y@cUM~6+%_PM!V4)<We}Hk;}WrrE(rrMwrn563ZOhl
zJjmepo1|oA{jh;ca7uAPgPU$Y2R<PW`A4VeNI|9MhJe*~G2hZ~mogzo6q1zvU}gMq
z25dFQo?*9C=v}KG^(PRK5=4MARpT*7?V~yrpUkk4jfZDvsO%x3P+4eZ4nQ@Dd;P$n
zLlRi!7@bl-@GWiV4f;U}<)T{n_U#o!64%ry;&SkT2#F|Wl3DQb@x_nuv9gK-{4utg
z!N%>z><h!Wn~-uHIC0|Rmt@?gGXTlC8WezyWTB`(fS^d=HxsOgFruGEgf-M1l*|ZD
znrrIX&p7_CA&K)CE(}W81fNW;9cnT~dHLU<HXFl3DaQUoLG@2U3(%(KCthBkpIFNO
z^dE*po{MSw_2t4CYwQ0q8vkAN&Hopm(;Axg3H>OzT&0jS41!znf_O(590jsy1i%=C
zaH@Le@aN56AyzNN{ScCD_G*n5O_)Q>osk(Xqr$_GX*?<uyax>Jh);O<XsVg@L%?nH
z_1{IBRkSs&S;4OW<pqNKo_@AO7jqDM{o(d!Lcad~Ww^J_vPB+8Y}PD|yE~+^=IVVM
zwK(;5m*e2~M<seldG*t59IR4+?9ZO169K*B=WcMuR6by;?(u*Qs5Gcs*~<PyDvB*<
zPbeD)D#nq82NS`GXcjYot8OQt%58<2VY^2oh8a<Bx|QR1n^|-WJ!BRMSVBm`mg{fW
z!1Er2v<9bDWM<~Y`T6<fL&(V~R69rkDuXff5v2e~CwuSTyH|=r$k?<kBN!fl|Cc1>
zGcJmun(OPnwB=+#VuP7GkooJw7A%|aV|@?WpJj;>;I)ZfSQRuNM;?lZx&DfHZ+$~U
zNeD71hZ{gt1W<GodJq-3HRJY>4nL{A%yUCZM48(mN<M2_t5-JSR_{j9w%y~!ix)JN
zQ#S6$+bLvT*Kr)?zcYh2h!?bUikSzqFsr({I;;P2D8JAMb2Z3oX>sBT=gTgK=mKI-
zqomDOJ4mqx+=7Oi*{60|{*M-bVjLl&8RW7<83TPYIcW~SnS<lp(0|#K4q@MAt1O6Y
zVAaqZ>O=MB10Ty^MH#?E3c_r6Sz2Ik{Hcpl2C1c|UJ_T|#;M&5p1A7BK}CTa046&B
zVqe)>Nj>A8>r!X+5&2{I_Ael4O(=7l@`1~7qU_pXLN-%i5Wi}__hq!ZhgU&0vnjk@
z&fh<+dr|a-!m^<XCW2$b9+8jcbMNr`oPz>is~zzV)c&*-LHlEXSL5aHuXa?9?I4Na
zNE7{jt+nhcCW=UqJS#1W+UZ8!%Xk5Z4D*8EZyz1<GT)8c8qNf>?X@ubglc=q!0DEl
zVG##YM$^G(LG!TL9G?wL!+g9)X6s-jnX={qytsYmPCOdCE!MeABm@tCJEm|4rE)Yd
zdk)-nDRZ^)N|5i70zEc$n1s662`Me`qh(|l0F>OFIt~3d>bY3OQ1_{idLDsJ8}~R4
z9X0a5+DYydtY}r@F@}$WRZ>87y`}BUT3nfv?)y>og=j0L3By13Zd^`M(#lp)AQrXk
zJ7)njbVmMnjL`wb<JR49(gX(Om%th-(nVpNcvlriQ<qeVaO6o(y|G*vj?c~2f};KR
zHk(Fm0jIqUi18*_Zp_oCPbFI@f{e?2kcNQ7B*LLvah0+ZvnBb3G|+0jYy1s%@_j9a
zPwY?p19YJqd|GUPfa%B1A+HsWl9Q9kYPoiN|Ni|3SPLp6VyHB(E){O1;R7hXsi!rc
zDBo?3UBihNb<H=Oe2mCLR`#9EOe@fYM!lHejtsL?%c?<y$x*Y#{$3>e8PGKX=qRQA
zAdV9)?a9px)J^~}8-wgd#i(QlarF*G3^cjh_CmcmR4Uu6TtNL)a<6+`3W?$&pjvCk
z;k&=;&nMdsVD66P!`r(_A_69AjLdcLFvyruO^DB71q+&_#S&h(xQ7$F34Ar5`}?mQ
zalwLK+cEupD<HsmF~lGUID9}ce8SQ$qiTB_q=R+B^}l9*!V~ld{4Ec6PiXQ2B;BR>
zAVbZ4=-_cQ{e5W>_^{iilN(yhu;h_+un)lY_Ho4SQWrr<<b>h_aV<l4!-+I^iqMC~
z%quD?%5_poQc|COWoX+3yVzD_%21*@#dRI&Kut$R5q~r6tzua1hHG}3^I)~typr4d
zYT?q0+I||6vR!X;^BK}L?`^W!AhF^MivQnS+~+P04ebP?!6;}I(U$HV9Lzz{7Md=Z
zqR<5G#RwpI8ivv|NwN1ZP!Noau?5L3-+v~DfJK}<`Q*fxoNov*8}T)t#>1qU+3=8Y
z^ir}4iPw-D<c~J$jsUtOZzMg6J1;HR<xOj`9i)M01e|IZue&}+@!XJcF_!PeBvw*~
znzxX6x9O;yr4f02FxsVlNQhFep#Q<3D2hXEntXYj%;4{icC&ZUyg{Mxy^(HIEHKQ#
zL~k}x9*CFHF>*XSK+I*4**<HdTW$Xh`c&TjTJVi(5;uYn2|_UeZR+X+8A2czCteH<
z%IwGW7Wo#=9qL|2H~i*^98{m`8Tf|yXFYz2qX&z*gkIs{x<?biOwbAZ!l<H+K)+>l
z7roLu5qvk3pqC5N%8$~}&A|uwm^-C!^IaAetjsZFn6#wHU2UPE&4m19<)Y0WgF%YM
zE_Wk6h{CpKaN!Zhv1yOUVe03<e*G$GLE%$lNO?@D8Ll`wSx^ko9Sh)=O-(QT+N$jo
zZv2I_2yFn-9nfh$AT{Zw+($9^-Mo<xPcQYa`}eK;B=#d=j}zPH3Fr|DrG@8`TE0O6
zrb!(mBcn1LmXzL2^K${V>=;eH%n|8B*B-!Kpyo7WP&>oZ-~SMpA>y}yN@Je0;dI|6
zi>2Ad0=LS&C@0!7Fqw%Hu$VWwl}4O`U--n_)P)e9vWuwqr+x&OwcxEe9InKVfic<w
zU<QQD78(RJ5gHiCMk+<LDfC-HP>cwNd_>tcMY>36V2*c*?mUEg1Y>U5xzp%|Jva&q
zWdQv_5#PWJOvb1BkRN!xfdUyYy1J#+A1c42R0`58z4rOsToHBHbHT0DakpgEl9b&U
zb9-EikQPVK28|+-S)h606wM4qX1k|Y(T91+mzivz>TLaTLx_aE_{onT_aR9zWS^nh
zs*j)7_ID%y)(d+JAC5>fgy;&4>h}8h_?(C90lW4*I0?u&n|e6Iacc^?EarkQzJTwi
z5^P^lDH5jWp$G~JZUOMdz`&r|)>gF(jUH|as)5iiViM}dG#81F7<P-4tLjV6M1~L4
zm5j7#Vq;Oz<4E=(ch0mK0zcjLjf~KXy5sR(z;mKGt++;F%;r#}`6nmyLU@|ivUD6r
z$o8TK7!2Z1oY=;V+kv{~OpKnFWTRNIk{(i*)9EjMe>bAV2n$?hu&9op*W5cj2F(&B
zYTQXu8n)r=f!Khmx#|`6U>#dG&HzRhI1J#Hyy+tOGdofM(RrabDur5lqM33xU}K|V
zLN)U5m+6*J^HSI$r%vugsPJPNS-}%fq*`L`n5)Vw^>&A!zk@Nq$e%9{MJ0~GSD!7H
z#e{FsBy<BYC+TJ03^Ab~Om+_+=j1H^Vz4b8N16`^vr{D1$DzBn><$V!T(cB<)Q=Az
zx%4fq(b4IQy?qu<m!c4Eiv4nPmja!!9~hc9JlGd-K5|Y4Vsf&bWAE<KZv?}G5F~Z~
z?pC2J5N>0*Wy^rXvv04{MhY3=sY8H0j(n6jcNUj7`SX1+6k7qkvFS*MC$@zF@1-xW
z6P-!Vf~}o~i$(&y+|<8x54bCFj>O*4+(sN<M=uuD)+*6R5;$sjZHF$BF-pY=vvmTP
z3R8frD8duSz(D5+m9@!m>gDm-B43uKTrcN$!)R*epo8T(Hrcqe*KwMe`P}45=lKhF
zxBdPT|FP&lR@6xS_+<0>DA!~CALZp4*nU|5=V6sK45v@8=BQY^B|_4hJ8$R8$c)uL
z{_^|Jmz`a{RxH}u+8G*ow;fnD+9>EK%j?+Jo|BW3Vi%TW5Jtm&0Z}2LypkI<!Gn2l
zo@A)NxvHIP`g#zlRE@Vr&>mE8FmH-M)*o`c-$W+Ej&leiVFfcJ4?h7SpELQ9=0IyB
zTb?q=R%g}}+inWN%YNBEl0v{`uz*bcsW41B+pt?eI>plk*uIoU0S>rmg%|gpCF6Oi
z6rF~=bod7;LZC2&iiW%x&!GgN?C?$FT)O$W+YZYYj=30X{*&%Z;gle*OK{Xa-__6$
zLkFE71uiZ$!Q@%vmAuurFx!}h<lLJAIUEqVE)l8P*w@iv8RqLWG9qa(K$H6x1{|=`
z(~n(g>J?Z)vS5rM1FB_X;1rm^_1oE(w^5|Oj$%buuPcPVCGh*xMx*5IGJMn&DbOa}
zcQ&7Vf~Aw~(+6y^ha49Ob18<hX_))i&)W|FMjSK<1R4sup(oIR(=|%03-@p!hG@{)
zw(VR}L{%W)E7-`z+Z{(m-xl482aQmMA8&x0+IDN&It*H)e~318KtRB;p3w`aeQYxq
zFk;lQ<+_6jR*g4W9{XGQ2Qdi-n|iEPPOJobl1cUya%~PmKJSnAsSKnRC%NoGmhgWu
zbF*(nv|tJS*1Xoe(qs&wQ7kbxaMk#u;4j0j*V<z~jVh=ZaO$!n3G4j6x1SpF-HDGR
zh7<FWvwU2{)w6Fz_x>PAqm)Q?{vqu#4qpoK<n~`8I}g^0N(=eJVX*LnpgJ=od_R(h
zNwWkR&}-lE7@4>GP59VdiEnQ+*f%#h*cbhLV=xZ57+Iy4($z8rm=>3F&=8y57ixJS
z?X@V7Lhkw9v{inO$Aw_wm=jztfLS8=Ni)Rn8}PyCDRM?kIKzr4>{Q^EyE7fkuEVeF
zx>e{ax8YWMYoHpVe4z;hxS=!SH+2rz=eJ<ZmIPm@t&u;40fUx7qree|Pj(?v`Y^6t
z*|G<2Zpt{_^uOn!y8VOBCd%3TU{VYcA^FmqA70JCgQS{vb{>?MUZ!lqEWmN*27Z*C
zk+;w&_SGaUVOX9}c^_(upgBw%Y`X50KR+gQ5Ndnw`Vvf5B5v0g7D(X=7o>It%|vkP
zir}BYhbX$1MK+vIjguJvzyVr|J1>FPp!6_$!ed(E;!SYE8-{A88TVAfNt7S29Cxug
zp%o4Hz@9tcXzjA%@V|x2%Q`y5+R7{)e@8WM<M9gn-N<;s)a3>GW*@j6BRHxksBSC-
z^AQc3DbrG>FM?i^%6s&j!!?+gmo?G0gMnt<#{56*y?Hd(d-p!9@pLLmGojKzgff&&
z4aQIslCdOZEM+E|PH2=lGL#{+$~>!7W<tnJ$V?fN(sOO+{;u`>_k5q<TF>*>eb&0y
zeNOTDyoc9ozxKZNbzNH`{#u;cz&?l3b2QD)M4#-{VdXzEo(j*ynj{tYi7@8>g=La9
zxSo~ObSh`)g4y_Gh+7M6EUEN%Z49hyYk>xZ2L_KH%W8fmD?59TCfeS?q3<i`Bav4T
za9#Uzk<n3w&Y!DECDF6W&uOz*^lDkQL;b>k0H7M-bo$Wx<Bt{kl#!T9;Ci}1s8YhP
zS3DnslQeYtv7Gs9OG((iecQHo271?~yVn;?54PpEMOIcLO{8a6kTWhdr&~Ynj58>X
zf?fBBA&#yNTQf9j*}gJ!a3s>08;tAixRg2+4zFkGS2FrKpqEG^iI3&qt6KpeD-u);
zMbvE1P`KhO6&}d7>N)1&_ZuDkMqe$As=}-()u|1?@ZY*TJqUAw^;zggMY-wgHCnXh
zH1zGUfw`WCPdS@z-y@{Kq!C0+!3^T;r!n4j4YvNdP;93dSG`TV;4PGXsm+gCV_0}H
zT^%i8rYs?BrzN-%x$qGrdy;~gdM8PZLEeG={7->2kasOL&fw2U!;#_~FX367UdOL%
zoUvTRxsICl6VH_r>=2bB(Sxv+G`ZN8$4Apfi4teoA}a$6CE_EzY_4KLV-K=eMFgwu
zXwAo_Kmq1SlNb$DHtCQv92`E$Mz{ZB22KO=o`&xOFC-@FHZf!ei4|zZ^tbW^Dug&c
zcQX6o6=>KFwvAmzORc`FBU_8cy$25_iat*RgI7X-fhN}RS4LW4h}kp}cN&<4{`52J
z+Hj0o5WM=sQo3}52iyMYzCD)_sQX%Tv&j#W$?Z@tC73iKEInQQ{Ko;$DzGt2+x~~J
z1AFd4pk@$a4t;z0f)_ki^ZKSy#8}nx;x2yVicD^G%@`j-3XI874x6;Xqku7LBqcuB
zCJHy{mz>`VxunBDLi#Sr#HH(Yag&1^xLb`a6OhKeAQh<QM47`UWRptpJ~cK0QsB$$
zua0S{$m_YNt!Em01BpN}qLZ%na~D*Z6kkf6&=@*`qsj*-Qgov++)NRfVy2tShXB?4
zAY+e9TMM68nma;WNeGs*TeB(02Uf_MXcvq49_MICqfF?HW0scc#PKW)#-Je^c2b+o
z5NrGaMp3q>-h^2I0tgfSh*e5&ZZ6WOr0HX%9jj%#`jEO)v1vY@X?oh`7+#yq5A~3x
z5(~yJ;i3Q@bo%|Wz-f03&8fkcQ?G=GQH6Ue1;?pI!wnU+5{6f^J-q*N0{xaFKsJN<
zPbOb%5rf0}T@n}a2lvn*5Gqt*$ynZ+5^zX$O}5p+SS)D`k8&)DQrwIYY73e`5!JTo
z0zqPzFsje=36>UFuq6~^%5qF5*7vhyAYDuMGjink`^ioZ`4x#+qCA{oNYA-}N~z=7
zJXg$z;F6{R#X6ILm4AuCgU=vy@Io4@4kzkZ#eC3kBGJ>dRW;1*&Ej&h%mwtMh=|bA
zb2`mjYrB|OV@D?q1yIl1ikdGrI~ud@sDH?w_Z*>=8<pfqKnmeMe*EADgCpd8-oK{|
zDn4HNNu0;=@!F}1G}%or5a$^;_-7|u95BD@W3JG-rdzxdqCBTwJ|!J)Eqe~47l9*+
zNN&_;sHQOajbk#ZvF8p~$Z^0y;@S0{j~_>4RLl|BS*Asl`|&-Yy0tvVB0m&Q{)%>J
zio&sDb%V}(ao+onRu8-(FAcY*0azFMt2jDZO4@%rNE5u4w|Uk$t>Ha6Cy_j_<tUi}
zKA$eX0F+KyJ&AylTKQg-rlrfDuO8TmG;2F0DvDsu`M{T3^lCot+55S;9bRy9wiUQG
zhiC|n>YCq%yZIzD*2t=cTpc~)HqFQ;y8Gxf^ob*HjGWRP@k@ObR@hOo5_oQj0d?Mz
z83d<cBpfe!uB0GysWzAY?DAK<lO<UrS09VU#6Oyhr_qQt%u8-UEhDcrPdi7>XiL=I
zBAJ(8y~fjn^OUr4*q7IDg7g3M@bmY`{?%VdqOV3am<6p`G^kR}gCk*$fW9s>GtHzD
zIz7N~oNVVozFLF2Hxdztplfv#Ok#*2Ck_n{BtD27JF{#?U?GQ@V+kqd4JW80$~Jir
zr;e*2MxSKs4_3G=iyyeua_c#=>Jk9CQHoCRu<i!Pak6`)rzdK#DP1M~K9(dm+NCFl
zrqI3?`*jgRPo!zoqPj_NYOVR{<km}$j-U@Vc4Q`D3J-sGc>OJMS5O2j#U|Ag?PiOY
z*N}Fo=I6JuF*O>W(UO$nXo;@gPMG*xW|o0RYU-D2H=^%0^ta|F#F}${eZ@9N#LFG?
z@Z$(Xnm&6BKX4;cOm}L<fk;!%sAMD=E59$amd<V#EyOyk&Pe{tKa86we%-rY0Eo#5
z@{ck)zk+|h?n#>~BCofPew!~n!il12K=IYt_q+NQELdRUHY>_gfZT@a#;;O1xs$~^
zR-hOV%H>+18-up-GsU=lpSDPXXcxg#SoCq9tiy&!+8LJ5zK;F7W2kNhmb*{A2j>Pi
zfB?T=81JA_o|1;q1Di_l_>70GFx#t}w!%;{4aXZtTBIJC01>HPdu1IG0F49jNaH>!
zKT(~Rh(yIWBnl$D{uBC+^AcgYXQD3*+vV8NBP!R;z)d-JjV4U`pkEw`A%n-VqYl6q
zB+#gDQAYeGsA(0E=q|T;^#^)yHa`f?grjEGVY|e@V)Vqb!;46ZDHNQ{a9Iv%sZW8}
z&zTu<#-vjR5ci{l4qR(J9zfP2%Px4jJ;=d#YgOBGzOKb#<TedUp{Sh)R7(2iCF01B
z?GNAA)fx>9H`c{aM#kz#G(9ic>KcuJ*je}stE}harqqF#hG$k{yQr@Z#3(L$pTadx
zJg<$0pqP9a!3QtX4nbWbNr_aTJFc!H0Mt}Kdd#`VP*P=C1B0bQ``VC)#8r!L)RscD
zY3VF#M9S_9m`rDsA9J-XI?7Lw1<0@zzE>mC&U$sJHbTlS6R0VCh>Ww(86X#tFJenh
z!C&Y6&-+U)8(c2Ncpx>&_VGw&sHaZ(Y#vgh7XHmo>KN*A-UJ;>cqDY;7{y?63-m$F
zGw_^|k+C+nt{T5CrRcXS1`wn?9EChCioxd}xlWo{Pm>hFS|XjC*pm-yOpZ41d4FG=
zxCzM*Q3@)e;%Cgw$7|wm`8<oNZfG<b%b@>P-!MZx4hi<J-6JC}W@h|YHiQ|UJ?%0b
zi9B3uY#Lxz?ZP}7<BLNqOI;4dP$cXg0xM_0V?lpHU9S*qfkf7%spzB-1<C(tt&PcU
zwHV_O22N;&{D_D4_mWoh>i|+xApPb0Q~6|HrOCc?W|)5Wh%3bY@Ypm|Fw;2gG@8&S
zjSpybboh$lk&2M$*Z86yi3tnb;DGiWVG6(RIz4XPos)8G6_{I}Uoow>-hg+YR3MQ8
z@ueEf9m1!tVw0Ui_yuIXAL{aCGQy)%5nP4(W+KF3TuZ-jy$KQ|0CHSJV}Sml6#KzO
zGx>@-UV3^L#QC3^qk!Wk!Ty9sB$_v-93g=z`Q+l=>{q^qVqsTuj(xvOOB#Nai{`*|
zTqURm9WOOThq`JgI<jSVv^oy(l~0xGrR1r|x#}O}v&nFX9p7aUQUV4mI5puq54E7!
zE&>c^!_{E{bfkaZFF+7;lfYyq1cPmK$Xm;GT^Hh*H|m0{6z-!K8$`)d{ZS!6{k4B6
zQNH*SEgdBcgkkhOjUn{4FW<mPFzeF}FcH&;WxcN)d|&DU`u%~PG`gXJ0L9o}<UcKD
z-TU-OJ|IMLg+37`^LOdqrbS$Ww@&Jmv<yY#RmQlEL^XmK_zR8Bi#rX0*auo90Ua+!
zy+Wq;Ak^+4hC7@qTh{)9qQEP$G8%uM1m~P-d;|6~-xZLhXo~UqpTLP@@@is$Z}(f?
zyz1#z&`(@AqS8>ViL!zdkzf?{oLS=T8T!9tgCU+NjjEhvUi-<pm0CYE(~jIFqD-(h
zP3s?G2}jX<^YCS?CeESsI>5agJC%7}`Z_N=#!T|T)8a~Ch3$5pru|IgP{x#IkK_#q
z>xD=-s2g5~J(CfhoNMVG-hutbfQlv8eJXXYY;;F$d}=lisYu5!vro~_Hf}q*7RI-M
zcBV8C8%_q<)))$-{=jb2?lae0ZhQI>OqP203(5!f=e!6oky5Y$IDx!KdOBNipil~R
z@i#NUSn$^7JRdC%z#2=3Kf>i_A6NPQbP_!f%NwHJ+DXi!ID+**%wvpF|A?A1S(IQh
zjVEkfOVbcSdlw&LXRjd^>DM4z3@7lyIi-Z()@V{Y?O(U_A7q49rPbwq-@%WuuKaUm
z2rWGVl(T))<58C#g%oQ=Z5r)Uq|Ae9j>c6>&Gpgf1%ND}KO4@8&f0Ls<$A&4v}t?O
zypa2KVA5vqzt1*TFr7VSrja54z(%J;txXmh22u;uSVITujpF;%teu^BEZ%vn;TvCu
zdQdYN6Bhr#?v9(S7kh};ZaOUvoFePcDl_0R-bi0{W-)Ix5_}+W<&Bt|YN|wLSUd%5
z?}mx$&eN5d$qD5E*mGq=JO`$eew{h<2_>dYkINI#t%<885_JyN6QmR$t<$+=GCYM$
z;<-s{nQrElV&vKyRs#6EQlL;gR8f3&d6>7)(c%SgU}aF@B1~U6I60LP&yX041<s%(
zox^D-r|&9{!Z<@DA7Y8#3Y)7K;6=)6qhp2YSiwifN|NOIoI_LcZh35-uiN(#k5>AC
zH4uo{J2`rp(L^HexM$v!Ek))Zx(~Iq1T>$ZZY9pVd`U4aE|}6E0l41%2M{bvV`T&T
zp@bA<Pf5tIvxU>V52wRb%W@V6du(^_K8wb#-FhU#K(U`BLBrNEmTMX>Iu`FFlT|}A
z+*=tp?m7F9lH@2`)iTp5nEBgOohe8AEnU6_h0BmjQE)N#2u+(|3QcBA(z3FSI{M<(
zepFK$uRjF>nSu{^Wrr0mpszpRD-*cou3Yx0W}~6c)5PH!<;0Ka0gtc3IE*P&9AIP)
z>^x&FK#Q_#On$F7?ihF8_GOhaa6~h^=8S<c)ZW?s*9P88mzDKiIzygW=Re|AcO$xn
z40N4;|36%S*pq$+Eq*`2hyiV`V#tUfns4t<-(I?Q*EOdv&1amY-x-VZU|Kp6@VPBe
zY0SZ%`QeipVcfCrIZ3@S_1o&CTrA?a+2ktJ#Tu!R&CZ?2WV&$Ub7Si(S=b}n?_x*A
z%)c5l`+Ms2G^)=U-*kQPSo|TAlSkxF9El<a3gDHNExYjk27m&X@ciy#Hs%z^Mm^<g
zO8W$9h{%G?Sk<XdxLdSgFD5L1U39=8fcbtacNwY*jL<#``+bY_*M13TnQ12fWd#z#
zpkonS_%z6WFaB-cLGh9eGp_`9$0Jd8t%2a8se1JXU-3z}w+aWPA*mwnC5FSUVJ;qj
zcABLFZ9^{!htX5Rf-BIuL~6!^{K5-ZPUg@t4Y{jzG2qvAAs_L2{Fobc$gAu}&=jTU
zT5r(vt?}(F+B<|m7ztQ7)&f~}diN#ZFf<H1_K>;W1^H__mr4hHPo)f>)cI>wOU#)!
zH(@A$Ap6-r<n*yq?fHg&)&yCBXp5%ok0;^(3eBNt&eL3E@ic1O&OI>!prWa*StSZ~
z^h3mQEUm|bt8`2RApx=5emJ*v34rU57V(aOp7#P^Mi>CAM`#63rLn`i2P-MMA-Mx+
zx<8(49C)sPYk?ttcFv=crK3Z#FLEk4>@WYzU__z6_oDyRRg9TU*Ug{-@Dur5{H{WQ
z0I-#M5|QAQU{COHu2|54M1!hyr&vd=w^o&s7xsq+Nja0BGbdV?zGz%zCdl868;XWc
zrhE$wp~+wh3;hZvh-8Nd)wQvQ4@ILU77IH28smfnXnK02TUPia0M9~=L+*g*H7*HW
z7~*bfZNx*W)io%u8LwyyF;!xO=!wQTXIs^w!s<qXsW7(Gw$DT!AyEn5hXQnRl{M8?
zDhP<asDDy8=@ef7Fcv=j7sVB2wgKm4l9%`eu!j!Go$Aq#hG#?0Met3K5`jnxf19`D
zUpq|{GW24s?ZdyLk`OffusmB;owAM=Yko4vh~4*`tLmYkD}$z(@dT^oyPt=p2l(B#
zCJOID$wre(2_ay1u>M-{5xRg~$WSv;wzVW#z@rV!*25sNcp<r`u{dZ99KlknmbhQw
z9Xf#u{2YD;!j}9Zpd!onrO6Bm$faalVMFDiJdq#XS@e_t4N|C@)P>gPSKO%Tueg}9
z<;gnyo*y#irWONK{y{m=WtmiHjwCvvFFTxvW-J|yBxEY+vFI_s^1-&~X#%==Zu9u%
z$Mk!1`89j0lOai%79JU(F|*!CDEdOUaDK!ty?93=rzs29jCFdypmQIvLy#|``*%*k
zvGcRuIgv}Px`02uD>My)TJ3mp3lur_Q7!4rwR+X6=2iET*u7@C7iIuG>Z&g1-<p5J
zBM_-LjXom|(`rw*{O@F_mji6$#B*g9(%xa65zGOm@`nDZi={OdUy&x3AO$8%E@DU&
z6ugnBdbmn;lRUPfC%SYouGbnS(@<Hv68$++Fy83Aj^fC#p_nvvW8-MBh)z*&qjoYQ
zLv$J)I3qZMnv`h1Na&1b_YIWeecf#;&I8v#J0_uoC+?h+7CUp3*6%vC83rwuF&FDV
zP!5fm$A6*T7B9gEj#?*0_W1$v!PLfs03yStb9K;v;CXEW{y-l(jyw;5aUw<1K`XlY
z`_lX2>_*GyC0c*$+IA%d3!~$N<AA{9@Jw$GS)g&=cn^4U9yD<!Npk1TBg9n)Y`e<S
z&F(TLp%C2Y`ZYZ1YfR-&3~WHYv%dW4O(8kAV^|aoArb8C^26o|k#?u_9kup}CO9PN
zqXHYn?KT`;6^MC9UpH%_`#tRO#Irq7-`_%0U=%fV)J)FAueFYFB*7fVkg7ZKk~@K&
z1~OJ+-<iO-wV>zjdI5f_SJ^l$QKxHMFP$7tKorRN_I6pIokc=rWEy^1V`kCf#jIJp
z3HzFl4md3-M9r#RSdK8GQBYR>h0Koc{l-#~Ts1FoJSAL)!y`@IROK^{rc#(%MD$l#
zLa9igXk`6isFfn|B5CMaRF8vMh^c%_#${=8m!Yu?fs<1*;)?ISUq0s?Vr6%bN2uw9
zW6vd(x;QB5N!!=h5)X^Eo8$s!Qld6uRorW<jxx;VR<T@@s>#l#wE-=DSN1l)#5V-?
zgi@zg=(SNcLj@p(!LKvqri_fpEG=?;{yhXX%?aWA?(UC(8=5^||BT&e&Ah~26}8L6
z2gX_UL@A!EdVX{xk(kS%S9F;f+NXtrv*idkfE$b@<^j;+i=w3%6l%fL?=IlxlarL5
zWG-g)UYq@O0)P++OYvA@T-s?s;dIc0inh*@%jcfdRb4*!Tbg;phTiIx3Od7qYz-~l
z14lm%F%h}tDo6;ndU<GE$7jbrTX#V(<pJ9QN4pX|_Vew7zYJ6Uy3T9mmBGVXCa?yL
zf(8a2xW^B-WZ`9-EzU?;w^0G6R_4YX%jKzS9yqgALZZ3a>XZkzY;xs-^J*|x*6_$7
z4I!F$!HE(mbrXN)V=ZzF6$h$T*F0S1aJBISHntz}>Pclm@GlbKQtalzMivkoy#V19
zk$eG_A=JQy%CX-I+Tzg%6bk}NE>ZUcpg5A@(5;_zl7KgC<b{`9jj9i;n>ggt);ELD
zz`tm@g~M4vyYO+x7xqbfYrzS@lzY<G#-z37I0%*VZXZb9U)G4LNohI(TrR7U25gUI
z(rrrFp-jlU_9t(6q@75O>6RX<Y4<LIs!?f@ec#q+&(}+t5?^$43k-wx<HYT?v;6L5
z%yx!Up_=DMUg!%@_HG#Wl^@>>sFEc<J_br8vVKj05n}1KuO`k1ss|qIYdyXw9U27k
zaDWmc(#eY;($bb7m;{*Gvp6&vR^Tm47C%eg|0pRdO!l5wvrBi$+Dr`YkUz<`%0YCY
zA%@Y$cBFEy7_ogGX_kvk1ytC$LX!*(qvq-5ZgJ`}v5#`YpHmmBWA_{Z2CZVN)?xiT
z2|f%gf2QsEDKPm%LhG2;_b}aoA2f;NQwQUDbpYklI6C<L2!I_yrheAlCfQ*YK*L08
z0%O=cLu`F8L)q^x7B?su3S`_!6d-ud9|dh=8(xwRBCWRTEF5}K$?0uV36ZcZh!!fr
zsjTrP0mQWuRJI`0$n9&d0f~1Sg|QwFZ*HNdg}~~j;xM!i_-L&v8J$}Td5hP+OZsk-
z4ib`U52E!+gAYd$_M_9#C(*OBG3Imhw%_Q$r}c`xJAM$K{d{2G=he&p&Iki!05WbQ
zOZ+i2o#MhmE@?Tl%)=u;lGyeCh;DuD`@X@}<C7I~6f|=0cVA0bD_|b}LoR!T(^(C6
z6`}<JuZPP8ne)q%=mF9cvKv&gG$yVDCZA#8n~`h%2%w$Vwauu=&qTkPK<OrIBP4}+
zO&xb4roajP*@pL+%ZHL1$x?*&o+9NG*9q>soG7P`OugJBNh=0X?9;_0%Y^H0LTGGw
zLR}<w^&+altispWCDdXJtaKW6wnUFvkV$isBFrp-=tP<f-@++<?sL%Jj^p#KD=IP_
zE5!%ChOoq63yoh36u35feg%y|O67isHF*vR+p(U!8ZRH8ZTM_8i<&3L8?x9tN>)ac
z8Ox*ak|pU6sGh<;twoofx@WfDN}<*ctmR~=O-}yhr01X$kcbBE66W4uHwo9w*QF|~
zo20Cz=z=i@*9T8n+Pw3f%y9oH%7b#G<#y80Wh1BNfTJ{oc5%jd<Jw^iE!YYSE>)*o
z96Y3>am8i{d-+CRvPY&jK6RhnHWA>Uc67uBXqqC7*G8^{+znH}I|$?&xAO<UMXv>^
zir7{Ck)XIKa2KZq6ITDTF2UBvT>*8z(90cK8~I9hRrneW0;x;5t8bR@)38ars5dDL
z(7ceYf4uGCZ(v6N1u4NjsW++h^pqH<c}vp31J_A%V~3`$t$S40Avmot2-%Y3#ZAog
z&@*+T;=_PG23E3*Sr>&HIXO5KXI$=V_B)Mkqe$ossuUZyv&7XN%S79ZHBWSWVZyyA
zkTd)>6G_=a=nXXlo(zL~92;4iu*oD-w0dr`9LEzx<xe@H$JfZB<cbP-(Rw*pD}{e$
z@>#v)Eac-39k<u3H<@$k*uLiqy&Bg4n+1U*`^wXaJp(0gxdUn+DhymYgJ&Tuz#8AS
zK-2HSJ}aGA*eeh&6}`?6N>ushWoiK2MM#;{{RcgU+T)hI+qcV($vp}I_SCPhs|}k(
z%|-ES&i=@>B3Q*zVdqe=y7Pjp_8S|O^l%v&8fcGx>t!W<$DwEtKw8FHa63dUWC{iK
zKFLh{LIn8X<;;J>xlS<W^Wj)vcuK%>tSb``wF8H^6Q*`1pt=g14${{-a%rLBN%JSh
zso|qwi^UZ9B(u)f)oMoaPNFK;>c(SUG&;wv&wQhq+(w<W(y;zlL)-l{vbyBu<>Pi3
zmBow4r0LZ^SiGV->um*@nMcoqRol^8y<ZeF*+My1uMQ6^v&8%tlchOEi@>p{AFO-A
zY^VMN?;x;}The^wjCK=%G7?aCSY7));Xm9Wo^Yo^=x*x(E6Vf8eK$fNK%{^bA#-OH
z^)jPYFS`b9GL7MLTao+0D-I}71iI8rp=xVn$v9~D?)^q3Yii#Ir@{q*I?5VNRM7xf
za^2+U{Q*pvjmON64$NSNPt%hY5PH?~F92_|>a!E8rGguBvlAA{daNyeHsW7UIM^24
zV#{=#h*jDr-Q4hS_s^zG01l;!H#r-JxS1!MUP2-*ZMC0F#v{)>o^X1k(WnZ&>?kb2
zW70B(dM~MKH>r}#K<WB1&J~q~XweD#Oae_)OTB{<y?=M`1kY<KA%P5J>%NFTJRpI#
zt$|zI*6b!in*6RpF0fr#_)~`!!OEym#Rp#zTcBkTjrIwRkzRA_)-4f_0^}@n5VG}8
zZ=^q^xOs~4Xu1XI9{{$mmdrkZNFOzwDoPN$DHB(a(1nM3h4Ve#dRFFUHk$XOmLcFM
z4Mk%CwX}{SOBN2=>pL;xEz+oJx?ENG2J|Au*cvumOYTEQO+AO%OHkjQxBG-h73cSy
zIvJ#1%d%T~4I%*lte?+Q{U67n=GN7A{E455_5CVn)|_DxY&ZD1!+LNWrKd*AcRgYQ
zl5x+~-xKv&eMH?bji%{UOm0jN>&cUnA^;tIcZr<jhWu;Sh_I+M0X{69sc3>C=RRiR
z#sx=l&~^q+%*wrg_pS(jlZNT!NX4<UvJQ`N7dq*CB^na^s4PJM*65jUM(7~2V+Dlf
zNQ~3Gg2PBF?`kG-xB}-<WKx7grds30cr5B#>RZE+!E6dsgr>UiApHpsgI5wCO`oD4
z2Fm*G{rh4d&V)I6p}dQl`90x7*alL`?P%`WR?;Y6LfIYqBK;!}9LNZQXyG_$OutSM
zgE+X7>u6vX1O}T9Bu;N!#lbN=n&Sj)wkGG<BdvlbWr;Y21RckXX>#YTCYp8$AlM79
zWn)PvN@)`GXndpVvlBDOwJ`kIbSdA)@dj}m#3eqYey-56M#+o_zB?yB6@C9wg3(7t
z+tABF{TJSNUfdy-`a4<`sYD<}YB<2ZlInf9CS6yWNOM_%u6Dnt!xpfsc|o|bG?{PS
zv{&zZN+erqI8C5#ai)Xp)?!9Qs{j&oV7(47iHPHn2mkBC=rwQzzaG8+$yTkO(H~?W
zicPu~2YOR)1F?%Rox+XOZ?(E-8e9na5g&*eiuGq@<j5`=lPq>(Z1E9IDKbjNXrNk$
z+D3p65kRD><1__mieE%TL=is02qcB9v6M)oiyC$_c$-G3OPHDt6O{nRB+);%fp3br
z&zw-Htb)dwwjVHFqPq>Q`%uDz5~wS+USF;JXaLcER<X0sWkjg*?*RCNsOlh+0ZwGg
zS010fF@F(JOk*w3>%E*gRD%{=c9nDtIubpjUv{VQo{b7)H<JGT2#cMUGi<sSTQ=(M
zUUNv@V#SU^kG=0Vj%|%T08VxVMz`JEbS4L2mfs-=pTM=us<7rt)XjCQ?T|u<QUkwg
z*9>?~JJGAIPf#Jx{%&#r5gNiOivmi|3-pse_cR)|FC!Abj2y+t#FlXrBB?||$xjWU
zD?JtLEhzxa)M*Zdbm+6F0|Q`e@FHgUS>qRv19(1AE<x<3?dWR98XlJM0jw{od;sDV
zx56rvufB*cO_3@`j;K=ms4?o`jrrK!n!*Ftm;pjtDb^WG+z8kX7Z;01aKl<mlXf+#
z3VEVc7YL`dRiYZ$U%oE^s)zP(60mGO26K=1mQpx(Y9x9fwO6)lBdPCHP{7Y<^pP)0
zQHTA?*D@1sfvq0cU)ss>xFM5a+=xVrBUneDJ6w<vT0NdcnyU!Bph@39dFne%5NevI
zK=7_@*~bS!Xb|1}MBV*SJ%_iVclcaO4(Eej#U<Pa(-@C#hGo`3|KF&e>YYP@j)pd9
zsDo<2f$87E07x|UVL-VN9NvhDKJ0?#v9A}eCVy=nl`u1rStweFF5ggzMwM2osc%6F
zfDP5wt7AuHWd<9W@zCmbbyB{XLMHoT`?~(1-Cvz_x_{&}h@c|ZN-@4Iktm!R`bcMN
zN=4xT!1Ej4yU&2iH5xR+-k-?z=}s%4R<|kr0SH*#h@>4KS{c?8GrkxetfP2+ym;bv
zfiCFKD55r99;}qm^c;Ymh;9^~y4ED`@u%0%iw#WrVJ(w1qC5p`SXCN!ImH4}SIP0;
zY98SjZpe*ZjrHSMBp0z_cibuHwA7n6>4QQON;Jh5=QZ6OQ#?q_iDlqJ^8kf{hen>C
zd{W9j$Ij>}MJHfGg&}&Nfjuo)|Kct6a8vtuDN;Xs7Tc8OUbc)r*~Lj!iuHiSZ|5Y*
zO#@Xw_0}@euu-O|zNYjSbFc9-V8OtdiQyKVsnkeWJbfCwm!6XdDq}034W@U*oo7i5
zshnsu0%_Vt$+^V@A^up~g@m@U@vy%%*VWUpfg=Q_zRFmsTmm`YZj#OqIL;Q70^|%p
zk`+?`&_8<<UAA#js|=7uUS!iZpo{|0UjP-zDk2|b0;|C8DZ&~w4QQT5ws1wv{Ear=
z7+=AB@{%Gi0IV%VAb=W@wgj13-_hg)y9vRH+Eq9YIdSaxfcKyPn&@_vXDF1mqlcG-
zRDxjN*cl7_TupAgqc#o~1+ZGS0>oPB={EM;2Zb^q#T~@b;*`BxgtgjD(`@lRN-umw
z%Zca_ghEA*QUi>{{p1x7J0IeQ!=QEx<t9YOMW0+_bueYnSduoHEz2;03@J9jm|f_%
zb8^lRfC2wR!kmoQTRTxqfVp`DkDtytq)No1N&+E7vrW*$6j(~Z+9g=R{OGz|`RRPc
zjgSPqp^)KcA=n|KIEjd_RU;Wednik9Sm*>Q5}FYV5USJK{u#5D#r|#_+-hhWx2O0=
zh9HG0LBLM#T6#B)cRUh9L<|?#cU95TLd5b%|5nFQ^f_Wa2??=<iI=a0qbK6{v%@5E
zSuddP1C|E4L3QDFV{`BZb|6g>5jJ=UaBn%QumV{LaISDVK!(3!&^97n_D5f#82gNj
zhXpTGuu*{uFu#Zd46u%H)z?A0JOanCUq}wkV}MFo5#6?J!ounzY{c{kTow%C=OG*-
z#G|ezpb4Xq$q0?`I7;GBW0s&4d7&N5gXMS3O#~Q)er@tNA|#&4c7h+Ni-d6Bh5Pqw
zHAM1&RDcMCu5Mls>>Y>FjL%9a7eF2<!(vV*1me=9Yt@9{ELPt3n`myx2FU7S6!ThX
zo++l4%Ps|#9K85Pa97$7{pPOIdAhT|##Czaw58jG4&zAgPDpQE<@x8@^1Y}#kp^rB
zk9$3keQM~(Bf}Le)IjTWSH5>*FE-sTq#1<d($|xPArKY9=Fz8H3w^@_^q32izTGM+
zXf4#1ym10tt>ZZAF0hy)j0%7X?~4+jCJW;ec}NR`RvPt)>B^|F4UVLtN%FbdYj4PM
zgjBLKKGOz1+mU&`CY(-uoM_VseKo!+XbDmfB)dfWRuMkz3>(qIo0Ey|1A&BL=B&aB
z0f860UY$~j1u&`wMiAO6qPpK+M`NN>;l+og@dxABH4?N%T4Wb;C=u0fygkXcj$>{R
zjglSF|9fA$J=}^CZEw&B68AW0Y_rE**pUW|CH^xZ9k<?#+0pZIt~@eRYbRx+cNd(I
z;`t8?<9oOP{Ax^G-P3q$0G-uiPTMj@8?<3pbR95xfUq+YD13pr?*x;G52Xlvg{|%~
zHizq&g31bhSyc+E<$X?H<9k$NvD4!mRkKEm&^sG_zy7o+-`1^8mXTuUh@U_TD4Yvd
zE1c~K)Nm}p58IR$_j2gQ5C!!IQ_4g*a?lcZFAbzQf_+sFgh*X-OZ+B8{>wVSEycSe
zGfu#+Va|L6TPO7Sh%A;r<3w{3UowonTro0cN)u7UmuTFndM|Ljy;H|ux)5O|HPmKB
zBEl*!w4An*W+@z4V#jH?Qm>>a<WG7k=M&_?H}t{O2{JTGGq<X{rKEa4-HKjMbVb^D
z-arNQtE<bKY9uqjCLrjQCGirnHeax^qFdF%B{P0b4#g2+6s*L(>L<|=dB!nxcl2Af
zFa&@al`fhbkdsuEiQ|zVqb%t#`-VZ$Y`4$yO5(-_&U7^vC}hu~qpyfKUGwX=JzUv?
z(nQv5H*~VxSyIr6A#<v+%v%zru!L!TOMjg}`2RdyFlYTwd*iGe1u-wn=AEawlcfm!
z_zk4U8&Su(Wv7sn<(}f4zkw9VBXUML*qxBUq>ke|lpk+~6IfNyxym|{uM$h>qKYes
zIy57rCZ$>jP1`Qg#stMzu_HDl1`aj{fUiKgHil{HDf)1CymDk=Ubk)sVrq5J_#O>N
zyhxPkw@b_GtHWyasYsecBf480_5$(PfeiU_*7BrO96Ln>JAv`z%JmT&J?B11wRqs+
zO-j4A=1w-6F{P8Jf+a1f3f=H++}uTc4z<y$(oom&;zjx6aM{qvh2=rhI!XW!Qw!r_
zj9WHlDir~_IR{5}tUIHE8Tu_6MNTn-LaN;{X=c@Ot+S=xx%pDFKMYA5LGuefE;S|E
z1v*egP@E?XuBy8B>$_V)ravsw^?ojYRt~}&ftfU-n!Hm!UR6ZTi$;x4;|!*WlKcV_
z!$U)SuU@?(p;KR)IE^Sr##jwSx&P>TW(T}glG3a^-8RlBT7+zf@LymiaWlEG!0!nn
z#g{R5>qH;lLS{ZSwN+GBP_vg3pGe1h?lgEWe3>%^e4O?j)lxY9<_7WrEtgQ+d5i43
z5xV{HMEW_l!a1`#QBS-f%O<Y$MXmO^a}n{U^Yw{<74QKJND>C)%?S#dSeyRF0kxH0
zV{l=)HeO@Or8a$HS+ZdA;WK9m;0I^t=q5wJE$uQ3bshWL6DO5WU7SN6M`v-Su-hr1
zIxMSRN&gQQU~WD;`kGM|56ar{WH7+X8#}Co<|o<D<^rpuqlA_ch54ws<>ya|^FVzK
zhlXvy|3s`GbPNpeV!iv(i3BJlsktARc8?@nN^@H#u@s}MR!BYM7}#<bT0TDj1GRq~
zku*J194Uzi1U{G)|IN$q$O&;siSZu1kv|jzG3mzgrZNNeykr$ctX|zfbIDqGH*Uf1
z^g_DTka0O2+D27q6{!xUv5++KHnKFEy3`~7?h5Fh0Xi2VM^k#d7+QXqnpM(K)nXE=
zT^%nI;elEVI?q;8r&B_PT9wqTR8r%$P?kl@jj}Aj0E~-P3ic;sBE(BN^nk)a3$e?S
zSZn|WwxPB_=Dq{80`fYH<(I%J;amMCd$bkqq$Nl@j37fN1~nNmz)$WZRzV~#d(MCi
z=~>`SV+3;Ah7OpxWrRX=6eUM7))g08CZ3>rDWLBC=l1O*NMvayhikvZ%<u6}3&&#6
z3D|-f#2a+M*tAd@jFg}qnH_jC7CYG3_TJvntweq<Jub_axLg)Ty4!Lbqm;Fff(>aT
zxeDV$d9MIOp_UZr4+_XnsK6zO1VZudplkR8^-xcd3ye+!ou30eJE=mUwMk+Otb<}O
z6KLcRmBRQxZ{Veni24$IM&b?NKKufq@u|^rpfwb7jm6`*M;})XPjfS9)-<{12z(PT
z7+nFhm494a6%n34!4eF6ZrE~*`#8o}e*9&wCwYPFxtP+Um@eI#ftH|JP}vl#hNRXs
zEQXRwybw|`QGbod1aKG<K035`j`oC8s2C|E?S7&dX2PElEf)o9IbzwxwyOYpRWJ1Q
zZ0p1X&my?Q@S|RuOaB0_Y(q;{BuHmu@l%@yN*yspRt2G&NN?V_1?|QqI1v3%qGbPI
zatD(|hO9x6*!e;HSpt8EZv2%OT9`Z0kE2Wn5B$QB)97@UI-Tj_6u|nK8-j&`jRs2*
z?U4k|)YHD07hz+67PUY(l2lro^xh~b1>8r=^~yxQ1|idEAQy)o$OoH8HN>%hC!rb7
za!}@9VoB;Lxul6il5fN_xvc_uY$-_y+h5=C+sO{yBUDqGNHrtSL!fC-wU-W}zs@(E
z{j=$Ks;Ndpu2VW4n`A#q-91v0gCY<KLd|HiRb}M#Vp9Yyx@a<Xp(*l6%#L8hIo_r_
z&OEBwVQd(QTo5RTMepKENF_~c6;zaWUrM<4XO@Y}kzl{&M;%tKSrakcWH|n_4>dPM
zq5fmjXHxV=5SHMy=!1%g(hpuzXHu}iZsn_IaTx9LM2A)h%YaY%tzH#oUb(3$it1d2
zRP=D*h;Op9EJt1_a*k9Qo<Q_SJcxVUx|YD|kO}w^=@oyn9jhH1siXRPc<n<Tl(&U$
z@tFR5l$^zhRDsP^f@e)iP$IA*(ex6xyGY>){0$R~11mh4ks?+9`AX!1BAA``T8dPI
znYgDVztdxz$M`a}vOO|;coi!@if;BDTKOK!t}M}OC&bI2JEmjNLJH6mIrkKvhS|k0
zWjUB)B%x(HuUV-M3T+Hz8=rIi(O23d5hNw_*utIjleruBd=t;QEq)%^03*dDE)6%O
z3na(6E?>jK()UgD2b8^n*{074x^Yu`1N&Q6?G4+Kcf$uf={h#5A|mKunt*P`aeRKj
z_qB(g6IZ$*ae`*ML4F{y+Vq@0Fmqe``tLXkcS{6uFr4AgUUmeUj-9v6Zk7?wNoR7)
zBmpc~*#3Wu2);m~ud`0))nv7jK%~j%na=AB>lsdwjaf??!U5}f4^Ns}XmJk)Xb5eR
zt4$5{A*i;?a8yT>jR3(f;P_UC>rn`y%JAbolkSXM16B<0m7|#j8}TJWuI{ez?!jaA
zJc6&AYr5$iJb3VV;+ZG4%y4<&CEX_)n4c|Bg=YbZ*|OU7pZn)Gwj+9wA2ysW<&%>@
z&Wic!a)gj<r*o~l84lL%h6PFX*e|frYf_BYmK9b#cnhlcMt%EvE3l`J0cW_7?hHGV
z6VSRTM|<v<sPiscaD6Ko&CnVLCvFH;sW;NF{obax?^{BOvLE@mX%psz=ml<OS>Yr(
z+>l&&@d`@vI7OVWLNO;9uHUE4#R7I^%lm1H5BM7XgzHRWuNW;Q$I66IwfQT@K5$6T
z&8$vjUyJcQaP>R3qqc}`iRHZO4*;$d;h6|!vlcA`Vxb|jgTP2MHcQ{J26&@19X5T+
za9zI1ZIjF6291L|(BpDlv+LByq^tz|C9#eJba|~<B^|BLYrgTpM>2f7Myx3`P&ChL
z!iMs-HCQVdNA4@jUgy16j8h|a>L3zDp;vo%$DuV7`*7>&(|P?2*9)cb_WmvRmkY=R
zl8+;a^f=k%8P2TJojAV-h5I<i!7Q7>&#J^`Y^X)*QLx8Tq}c%5E<kS)!}WWoKEU9Q
z{&_C<Ka*+x`)~h8Pvh$otwnzz_5~<+w~U7@&7MJ|-7l37kPeW2(bGme<;R66?t~Hs
zm6uuSVH73c_&}}qc;yiU%)n_-G&zoj2=Fe)yes<WdJOHZ|INQS?7SOh&)ErjRi@Ka
zB3BZO`-(<e9eUUalQPu-Wf;N$Zf{e3?xSfOIVgh*ho*mKNf!R`#@VQ1grN?Px?zRV
zU@<m1>4$Z8{;&nUn8V`#{k8mWulc_}_rHI?|9w{fPtR)lye<5R8Ok`JH6(C?dkAUD
zP#S=dB7lvILz_SYMMUJUg>iNJ<?cxCyIu{x_=VlKaMaMjEEo}w{vR=Igd%p~FOm)2
z;Xru(5K1L4q+CV{TF8aU9$lebv1lQ=VJ?Srju$@MDJKZ*8PbGF=rdvGhu_2j%y4}f
z$<x0JqGSK%JIx~F*gp4Ly2sJPAbp*DY44{PZ}~A8YTJKWNgLhLlgn)<(%0ZxOpasF
zm6@BdB8x!{#L)HV#vWy;|Mv$~0lgw~DjAw~baGNcGpRX5fWG-5ee*$9`?HVF5||1i
z?8Qp_NxP~%eb#2i5U1h&@QJrF3keBo!k@IO&ISI>jNor^A=R?gK^Mw}>f-cnT-P_z
zE#$rYNN#xWGELt`X<?(|g3It&Js;k?3?`qlY|OyJwpP+#u7cR2ApY+k`)rkGB9c$m
z82-`+W>X&+MR0%i5_9WV;V--K98-9PV8&bav%U0jTywwI_c6&x2^oM&*60~~Jp6+{
z=k8YFFLaAdK-CPKs)Wx1jcj}{_i?}e{p+al@z*p(8cnq9S92e?>EDk7GPD($qEb!X
zzgKtQ-#_j`UW;krL8UWLND^o+f&2#t>auw>v2CA%l1e-}2JSen|1w2%{$@Y-5ZV|2
zXC-8&sG#<cRNwf2`<-tXF%q+06FmX4AK-|mxk#P2=9a+|nuTN82pzG*%N=yD>*ucG
z;)(w(gJ&sNe500g(RS0^3TZAX+Wnr2Ul=<r{)AB8{{7(GQh0#1#EB+AmPvKe7d#(^
zl)2|=8IE{OIPupp0yUmVyDgu)fN+n0FQ9MYBhAf4ZM1EH8QoCt7;Hzr%X}bX0p!3A
zoMyO{T|@9JAyT$mB{8`HU<+o*;XAYnfc!=xr@vTPUq7^eHWY^pnaV~bVS((9BpV#m
z;ey#PL}v*2g1#_EvHzC;Oh(%+GIy`|oB#gX|F_Sett^qA@(WrCXw6vHNI=UCJeJP`
zg{acu!#B~DlU^fR{4@~z@ISjgLJRBI2=00&jr+i5Le$&}|C3PCJOBdTMHMC9^}hO5
z_+$?;SD2$;SMC7)j0pHJGE)HSU@O{4-l)n;5kDeP<_G9W;)fqEp`G-|2o)3oI7X;y
zlMD(L|Gi>G<sc6t+v|mt_g>!xoWPM3>ex0)=QSyOYq9<Nn?{u(dO#_+E?B(0XfPIk
zv3Z%H7tCeb_n@6Zy|&UnKyCh~5=3pdv7!e!n=h`H-0HYuF<X$qdn*MmiLVIhqxpY+
zS6(E5inJ@J-FBb{m__Ou6Mga<pt2TdOpV?P^ckoZOT{yS`VPVZAoT-FQ#)(SRu<qN
zK@4c(ZpU;Xs=Lv|w$_;i8cgaH1teN@(@5t4!|7dY&e#G0;v+O;GZNy}0f|zt1V%D+
z;e;IElOb#Nhq6E2AG>MuVcO>BZsNZQHzS`bLRB5`R*c|qN1V(|6<h+jq7X?bQ#S-b
zis&r_Kavw02Kdw&WTB)hrjF&Cn~R8USuhRjRYg<_!l8d4vFQ5q%C1dSRhMz&>rIb#
zXdGnTILVRC{O%^F->Kt<D?ZFNzIwOHP}px<-8)w~!x2`Iu(KoE@3M1-aoegMSB+5e
zoSrTE%k6e>L&M^DseG%iY>zW7Y%gGrKqlq@t$KZ}XV00}40s5yTUzwC=sG%zVV3PH
zD6#=(FSxn6h<^bCJU}=AGRR@|slvGW^XJd*u4@<0Lx;?MwCis`7R;=ycL896aC5-4
z`)>x&m1IHZ`3S>7*gbB*0MN|0po_F))remHvJgX9JU~{A30?sLvnTf<6<s|PH(2lJ
z7i>&dv44W@@j2(skQdl$12tQTHk*3g9s-ZWkHRn00MgjEFsG<ZECpLb1JK~@_^T+G
zxC81yU0cX-6S}BxzoxCUrbESS6koo2<>{M`bK@oaCAPpxerT+080*h8_YeBLE7LBu
z@khXy)K>*d;3Yb{Z%Rsd+Flr7CD7&Qv|faqwI()?)786dH8Zp7jhB%5mIdusGs-|I
zTB=z2=FPmnnAl!JuyO||Vzgce?37jk{do)E=+ncvGBk`YgS79%ucv*|9>XZ$Mze+%
z-|`Cv1~&lyLuSAJ`Sa&l+_Ae*R?8nhz7o510S2EgUUW)Z`z%n;4C}sCz`&wH=cA=u
zb>TLnE_5F1XF#Uk1lIDS<zuk)+{r%QUk`mBO9(13F;g{ebi>AthFDJUZ_GQ{M*~Jq
zWnS5S|3Ph>V`g@CobI73fXI#O;sf@+S8GC3f^O?`W233>F~CEYfs<ijo|NCdm6J29
za_z<pcX#*b*9u3Dyg}I5Ua*E=cLVGz!wQccJ^BhgY1*QK!Po?7S$E)4?J;uU+-Kwk
z;dsqn1`*`6(3yeu(wd<FBLjnZka`vW7-`EdL<jgg`s3|@2kpmunc>PO!mk1Zo8Fag
zM~IX&ttzml<gf4pR=xYCt!)#q@wyCUJOLT=cC;kI<`dS(Q$VbxacF-Fwr%u*lkjd#
z%KO~Z6b?Am(#q<VxA$sCFs(E1egQJ28lk}KYvkPt4M_%o&5~AAhW4(me}Gf(l>FM!
zv8C||WunF{St4v~YydMBz=38toKp^>H(+35atH7nEy2$Kb4nKHDX&|%jzFQVNAIA7
zeuiX%TQiXjV{%lJ&mYam$T)*eZ`fsgts6CCcoxu-Kj9eApwzx??B6xk)bq0Miq~{m
zIiPyjefTL)0Tw&XnhtNQx9Ba)c7j{J6g~3w<=KGYy;i4IV)>a!Xn2=hz?6|yaHA@X
zx0jQXJB*g*Kj=0W+at!S1&)9`B-=?AZj)vgO-*?SN_{NhkaR8soKqz~D`<&sVh2}e
z;8$+~eKc-HGh~^qt?jhSQE<hPTV8A)&%W&CfT#R(7t9tni-~CqD&sx)3}a4=5E^D%
zSTA9TWk7=5`yF8CSzz&Dh4K$r7h#tcC*46NCk?`#C010t^Ucj?!2Kg0&9+YR(d~0F
zGc&6aJyCcCrn1rp4>EwAw3+21USU{aYyLG4@ILM`$*);B4|`<2<i!gY+=vp%FNEi(
zYWts8--8|FRrvm^8T^S8&gB3Pi!Pf1FFmaA?mE7^=!V29OS53H!8ngcG7Ug>aRyu*
z6{nOK`GQ2u)}gUrv-d{!y?e{Zi^uUQ;CL@-iFxf<xEb)<>Aj33{z0icmRjh(i8|{Y
zcn~EE{QUg93TvvXeTJvLeOvhek9s4z{fj_I^Hk2Z|Nh0RP$$Ls1|iJ5_e);=u@Xe*
zHIf`GEO&reR!yg1LW|FEzRiH2aUjCtsWfvIz>Kaq1>rH%37SWS(-c2ss`v2p={k}s
zp8;_9Dl{}S)Nq|Q`=t5weK)ti?%%(^jVqE1*K@-4-)PV-Dk`eLfwM(3D<3vjVVBXL
zbOrm>Sc9Jd>oKVDGRmGP!7}ioWH)B)!lZ%W6--PY&}K)TxTrYA!omW0@GV{MF$|Sj
z4w&pcT1#L8tt}P?VPLz%Rm7pkFq1vEK?8Si96Jd8c4J>AybWg;Q5jFfhox({xVU^7
zbt^*;RHjyb_^=o=>wfZ_BxoX*y$5L~PXW$==-|Z7OGs~B`_RAm{rg4OGw-q6E<^Q$
z&EoD_+m$2YsX4t8kuK^#k?v^`%U5hNKeNzN0$h5@>FJu{4?j2SG&ukIq3ZqR4C3<-
zpyH1(EKT*7G7$Cm=t0Q18L#WA_s3<q|NDTR;(&`@ZiU1~cLsgu(R0H)3=JpZRIadw
z;&>6>$;5C50IKolms_Y`C3YZy4LNlIVpIJ1<JdcYY-GLv{WL2}T7WlHVGKzdO0W#)
z;z|yV5VLDIIMK7G;1c;A|H9117H`86xx=<n1q~1ZC^Qy1IXR`6)-FZRlSd+dJlE02
zcXiK&dI-1xJbz^sWPyBN@R~*(cM0?Q^-(sgSa{>p1`F2X8`mzu#{8UV#YL~az@i~Z
zcKkD-sVXo~BMZ9{M=$i^rd}d5!^y^F7V7JZvn~2MjNzWa#S(Uy2h7dQeLeul-i=7)
zALrkUU*0~3LUktyE>C~|jpxm5a^3kgkT32wUw%DSECE?~^ldAEnW+IDMHpyq2lyS!
z@XA2?sLi;~Amik=)*1rUC4c|@2E0u)H}9d<so<(v*ExXG@DM1v#z9zhW*Tz5yu6Pe
ze{0~SFZp-D`dl)%u=ozY-Yue{8{p&3Q&0(V!xjE$Xh;=4kW;;IU?=P*<wVh-?){}c
z{l2Vhd0JYUX(9GlSXabkiGDbs^5VrZ5cofEEP~y;0sJcLfBks%X76U4ld7|whJx52
z)f*9JZp*E}C18s@g_6qB&Q2b!-7^*ze7p(RB4;3@@G_W%@EUz+_lizH&%Ahwe_>^6
z?2q!v&6Nc4_YQ*3o0b-KhL?Z`wT#Mn5l1qv{gFZk{~o5YECZ=k7LA+|d-UgIpB=s}
z<B@-L`b6I>9s~<g)stfPKNS@%!R{+W<F_4};iTobYw%My(ZT>t-nRpf-aX>wjeGdN
z@#i?~L_vkB>q8&&89qGeBrE1|@u0oFObEXHg`m~m1q^No#}Y~pLJnMexMGvk>(*8d
zgbP_T5<9+rbp<bU!@EbBpp_LA7XF1~RzXFj5OLjdX3_$m+krL&YG#Z%S%rXb9C}yc
zLtYUzK!36K+%M$G@9_GnKzWKRM0}uw9`%3c>l+J}Ffz(2D=#lAD{JrVbw^egjyHvK
zR?X~-6P`Jz#!{Dr2e=O@GkCi6f?-?2p&0<I`vLHpxkF-gosN!9I9SKMU>)+`3mj2a
zR&MX_UkYdV2Q{C8DI~64le`fCC<60cM524HF>vas8w<t?Vvf1>pG}yG5{2C1tc!vk
zV?7%}$*Q2FRHIfbbrg&B1E_Y}3($6WimvODrAte~Pt5)HhsFH3L>^1{H!nwk2a93}
zRxxYjD~Q(616@PThDQPx9y*Bn{8%-QJ2Jr=;81Y(3yPu?-^_4C=8!rdE2Gb7XTb>A
z0KiU?>w^+zcJ}hcM`*=u_%~fTxManOLZnMoAXQKnjGVAqtQ^NAskM4*qE;Zyu6_7t
za!3d3?hMFd+#0cKT)+fQD2-4;*K7$oW^5;-6)i1y!wF&`9!;uaN#yhAXQA)_OIHxI
zl$*#HQb)^+ie97JcMollFl64Awzen58F3+D%0ahiH3?5587AVspt77CdE|SBggk_o
zHG04uI6&>EhS_07=!y;A3E+ToH~o^w_z|qzv*0ChYo~A{ov8veBPuRlfF*`~i03SQ
z_ADoY&bJ@m4WUe22@Nf-_}1nnJY)0`T>~pwSYiYp;C}B;u5=pv`k;3fp{L;U=M9+2
zNY@XWkbkgr3O;<0GZ5OlR}LSC#O+&G@oKRCL4?zP2RvTg?`R1QsokJMgJZ@F+`=_G
zTO6g*Vgyqpcm>VPY<PRLjL)6h#vx&U9d>$D$(_8k2akYr|NVvgQ};6BJyBlzQy~z~
zWA2~j;BQtQM^iaCs&va;N+;=`b00sy2IuDAe-=NOJ38k6<G+6~_Za^BNa!*A|FamB
zM|lk~U;tFsz4=LVxt=K!=nSi#hrQo{H9o}&?Qn|LIA;M)?;ALkyOmm}8()in2KTEZ
z1gl&j22wg?WbyO^%-tA<_xKW{E?XG9=l}iZzps;?!~fyyjChBd%;4NPpH!SZ0{Qo3
z9C}fyE*-h{H4Tu&FMd|fLpX2AWW%aO8muE})LAa;J={mHr>5G}<sQJf@{6CTdyk4r
zrEc52ii!#aMa7blFlLs2U_RZI<rW_wkJskeCd)e)lzhq&p!b0Dt~(PP9J~jU3a@fA
zJ0~KXUj_uM94xk$@=By^l$<|)@}!`+_-{~_-k~H$Z9hDJ>eicg?-l|&t4uwBoC_#$
zx!WpdDf)fr+rY411W}GGFsCzSX3Lb6lu-Y#Lxbckuz{5;R~~Oq+{pP;nDQc!sEbqo
zfF<Y_8EGU18R|9AzbN&@$lpzLpU9_@`%QtpQ$hODVeC9+Gn`DYh7}tV%%E>O0}!M-
zl>gqJKIU=EAl6{moQ169*lDfOuP|6VzT?1w{9pX5)~-Z7rk1hy*YDqB*$O`LpeTor
z3z>o4!_#StF_EKImF49}PqF}hlC6n`uUuM`!XnrzyV6&_!4Tz9$C5lP9i3&UWmjXD
zq*VL?RnMy|4!*=m)kjoR)|WtB!Rvy8_8-7{38gVAI1T-;GG;+}-X>eNed|`~&oG>%
z?oUJgQGhG|Kr*~!lhoCskuNn|MZk#$v^R>|n!Bub8&$wmhw92%)H+LXdtR5N4<A-{
zxyZi%ip<IVpxw#ToXMT7@%HlK1Y>LI?Ck7YkZJxwHnbE2mu_O5QLoK$&?FN^f$fIB
z1ol=ek#q@_=3BB%ud}-dFwN(=4Fo?4C5e|H;5aeuDB(1zEZ~s**ZRKq=)bJVQX9P^
z$#N6NI{BbwP9neR#DMeAx;{|6Sg-(U#n^vZ!-kj@-$;2Uj{F_s`iBqCM;;NDGWARh
zgP<(o!IDs21Va552*`HS&msprgRYQR@vYg_?CcK!(@-V}5rrrF2<T&YjZle3-vH;X
z_r-PA&)^vzmi&UQ#I#<7guk;rIdhL?i^?R$>!dRwC96oi1|Jb_5CiM_RIuW1ZZ-~P
za@GbCbqfe@7}n%tL8GOpNUlRT5Yu9&n24e|<zfpf&_4fQN>q`ExdLM8z}8Fpa(~kS
z{XQ3?+^)U@1PJ`R2K+&-F?u)P%+C!CecL?XHuwebf!L9IQhZ434q_!-Mg@4a_yF_+
zJuR((d6#e8_(XT+8cZqBVIL5FQCW+0qkVYzA#Cg94f0WJV{pM8+~DRNI~F6;u>aOR
z4+5fFDAjqFI}vCFUFlr_>OD5!@AR(r3Jkmp29-#?i<+9+eW=B_9LPF{*3q$Q&T1Ll
z0?~KL`6v7b)Cq?{hxrZ&!PfxSzHG$`58WA+AW_c&r_k8~%2OFI9VE=V2p50q!Zgf`
zSK!BLv>eAxDC0p9@D0Qn+|Ksxi`uS^F-l$i@wCR;!rc5LZbEkkUMTjPdJp0q({}=3
zxG-d8YwMbye*if6J)q~|$V9nML@e3G$EW)0tSlx_GJ=5r8V^?$D=yw1-i+vYs^#h3
zk~#OocDOmldj244yw?p4S{J6F{AS182QeiN1$Oid3INQdD0ZS^*dTQ^BmU*td}*@!
zNGSOhbCycj>u~*!ls*HV<UIg0{lXZTdj9g23nz9%W{mvt_;|*XCmRVF##-GXCB=b#
z_8QUZJNUnba8`bqo?g%M2g>|;cxu%=jC;^%IG%+OJ|4V0rY7hcq8f9l`-3nl>NL5F
zMAw2B(WbMU9MnScXEcOCMbbdxJpKHh`;V@win2hIQk=Tyyu!F2FbjAf2do2kc(T#Z
zj-c}fkOfeUbLEO4)Oi2Uw<35e7NE>s57<mC-~ukfXVP`Wk|piZJGWELh{|zweXf&T
zaL>eOw^EhjW@i_e;CMh>$1QpO?Ae=;>~Zeiy%OcE3}Q5jPezRQ1>(^T^5h4krp!Qy
zpaOo1Ky2D`4aN_;_2HNuE!PvKjEAclCUZx3W|O32gtaf9R}U&W%qy{~`}5($hi_9;
z@+eu2q6`cTZ-IXo25z&~4Cv^t;OnYI8Ep?!X+v}cW6;MeKQ&bZRH&%2UoIdoA^-1_
z&=XV!P6K0JF(`8G<53u*ZJS})F&|!5j3}CeyM-Y7tBR2Wz3(Q@lc(5yFJHbS-s2wV
zbuf!pQ&`eS+R-X0B}E3$#%ib%<p{K$UF&Cs-Yn*4cliEg(fNv?I|PfPXr|wVbmJpT
zd59{_r4oZs*$(Ugi{fon6*E9FDh@!p^#ESn+rDeL4zNoqvr&OR17FwD((-yt%ub+C
zNv=kuF{`t`u=A(=i_R}!UK*9IgToL5W|A)gbbIFfdB4=0K^%ZMz@AP-1m4`-j?PkV
zZQOd0Wto5r5JC*N2f)@UG>B3aXRw=~WspUO`8b>=af*iHD38{NTZc)lcMxVxHC&tl
zs4$Wv%=qJ=kd=X_Fqyn{M_w6R%biZ;*zbWmuNmNQM%?OdA)#Br!R$C?Fz9auozP@@
zQ2H)`mNr6Z6<YCh%HsKR>*t749D)3)9dVU4at2k^kLhVD{i9XG7elYWgt{M{tc6%e
za-a{q1XVlIyJg)rFlyIHBIDrH()(OqUcLtg(Zwk~4l|Q{@Zozj;2a9!xptqz9znsw
zDCT@Qn$V}J7@cn31yW<j9~#Ft{=uuXs%lUDw?O~`*SxG2?byEE0HKl)aJqQRY)9c2
z4mRK^=y~cUZmGOMJl<ngw+rgnXHXBkOiH>L5WqqgiG=k&yZAa>c5Qz)=awz=Q2)sx
z#&vx8G9L?`3dn+j1t1H{pu^c{@;oZ46O8W-Lg$w^H#aBc^$ZRTsY2)gtMg?z4X^=|
zK6F=Ly47u9dT-0i8Nnn6a<K^dNdlYBU~6bI20LMiK{9w3)ho%BUZ$qrj*Q$+KLdjK
zS-^Vk4<FY3WE(#;c<3l%ruEN{+tjSBg#kD}9kYHvs-5p53H_cscE*L=5iCG%wdc#>
z{Ob--!vN3Bao`z-MzP~WTuk*RUTJ(n0>w*8c&Mu9E*~GaMiX`=G%APjKgt-;OL>dh
z6xS;Qs^K2!IcYrP>uX$~|E__o@eMQ$yD?^u;0GjL8ur)26F~U~6u>Qc66bH#hgb+Y
z-TgX&egI)p4X{Ht<U8D!X0dU-UeOT&Goht7ftc`Q?&?CLYTP%~95j%nMliPVUVaV2
zu1KXaehlZq8uXTE!=C*cZaXKmb{ILJ+4VPGpJwK#=P*@`mslsDAHPxn<^^>F6Lb*0
z(8TIGC!i9N$%sSe<m?Gzo|E71Nf)%K-hvf>8`oJiKKdnh-NHjuj_k%j_&1m@b+Q46
z6E7g5IAuFzX_>8^Fol1H9<x{9lgE#j(8ykZwEQEt|6D^3$qY1rYk*BcLgF8k?5WlD
z>3Fz2d}io{e^QJBo7+@x24}Zs;4utjG6wk7U&OBh%KJyS=)3{+VjDmLL!w)s{0T>j
zwMQ?YJ9%^SFFOm&HI)-o0Y1=yrIl34i?hBLGX*HY(&{+}n1NXR_S2)0gA$`9IARt7
zT(6puM7sB4nhcRN^C;cOS-er$_^GWF=~b#zKp732>^Z7WAJipM3()r}UmFhDHN$53
z?F7FjD51lTXIY!vwrYlK!T1LOpW^{3=&Ie806p}|?N>=jCkCw6T&Ey)TR5p+R<Lya
zh7Hl9$whD89HRUn-s)9ya_QZDwu7e{UcAAsPx`W7Ozf*zNb>9!JEz%SnS+AdltBV3
zE}7L$fG)uDTl=A6;pBY(fEQKR$hRT6QhGi#IJk>OK=Dk0QvVN_w?_crM8{td7Z+cK
zG{(m_Eoqd-l~WdWjsq=|H5gpZ&+DNK%Mu_<s~2Ls`}Uv+;3bED)5y^E?}lGs!a(V<
zt?;=_7-+@ZV;ZUw=!b5DaZZ_IE)SO0Dg?-wQOU|YT=G&lC$9vP!Ki!7<i@AKpl@zo
zau_@L8(R7*MtS7m7Tkkk<~9JUr2vDHD&s)c(y|ek-a46k0&gj`dK!&n+0&a4Hj6Vr
zi(2&nq8D{v2RNbJtLv%#om0Gwl3<6UCTf>F-*V_`2O<%L;5>fRq#T#p(cOI;;Be{~
z0pBF?5Nt-O-QL$X-5_}*2WK0cBocNPC;FtM2$P!8bjtc;p;Jq$nI=h?CT#%S--jwL
zCf8I>SC<Wa=NLhKp!=AI=|NU@liq;{!fzotGD){K!i)aP&d!c7C&6aGeQW*P{l%HX
zcaY7Ub$*hDlCY6^&`jqfdU}-4*!aoYVy9Z#*l1+;oIZ7G1NOHYSQRF@B`u)sf5e#}
zacx#tQ1@$B9qc7}YVg656pTp5kSAADQ}c>YPe`WQPI9lY8tpn#r5_5YdTns}q4g+{
z7Jzwm+ZhjGy`K4@jVK8<9r@8|^`FS|4hzcVYr=Gcs&Ai^kvMEXJbIg_5_7>GJ!gp9
z)LqtLSu+794sB_cy1Dr&_TV>!li!)F>(;sA+BhX8^=zepVasEueebX3Ew;qydE=Vr
zjL3_k1k1704UifqIf-EG+A~;Iu^wIh(s9C+##E^_EZZe4yarPUGhn40Z;b{axP3Pe
z6ZGm1Z-N~A@ot0pkWa>s&gA0AU1}49*r8=~q8_!i?8pl!DtfoAFq())dy|^)Tzx&(
z$n2ooT~E3;ZOK2adCvEcQNBSrY4A|yD3bYtnwr&M#Hg}NLKiKEE|G@oz5DlNAa&Y}
zEf&e^cFzI1Jeuvd;5+S{HYSX>pfP?5r1R~aitPDt`6|g_XJ@B2Ze-y*m%^)wpI=Ab
z2H4MQ=6*dreIKR@x+7iODk8ELA6@|NNqc+yqqS1owk<?Cpvmijtm`Z?B8KZ&f?GCk
zW`JB<4jCmTW_X~G<W}Zq18agr@a==k0S+vt+dsbL2ZVx`VIYWAz($FJ+loyD@8A_1
zAS*<okkL5ZJCmEJ9y3=AAyjq3>PJCYxhycq5-07_f@2V>e8=pm&2S5_{R!ihH>hh3
z%+2qEXGS?FL_Me;3K4E8?>c;YsnPAWSB7r!TNDv7J|Ms+^Snc-r_F;r=Pp|KpiQiW
zrH`oNc=X%j`6vpXqCQ*n{K;?VYJ7+7Y;0<^IEYZ6Du6=d!0&PIiqWDJa4B$P%EcM1
z!kJ6Che+Z()IF^DLGRz4T=+=rY%jJi+32W{-(=^O9cV{?8z1*++3;jN8Jjr9qnqdc
z;K6%5+mD!U3)$dOBuhBoWx=4J6cYL0aug}1Jx+gSov9o&2t^`-Q{NRSa$Il-4|PNE
zvpyU()&WD6S{1cVFug7Hm{5<$M-`*#?}!rP!z~+8*MTPT1}W-I2;4Yf(={-&v!{6n
z0{jOg9-}eu94_H(6NM2<-5=O?yoOo}3H3jK-!ZnZ6OT8;W?&s&^CC19Qa85Yy}9Fj
z$A)3YQ~rRW_1j>B5HVn|AwMTy@ObNy4#?sDhCPREe=;#Kp<)sxc_K=*t$X*bLsf+c
z`wp+`BhsgM!DTnEV^*=^?5T3fGAUWQ{-@s@G_gWp*x;A^m1F;XKHeRUd6A^Ih40@#
z1D}W|iW7hm`LGa=CEghg>Kmb<n~>)+Z`$Mmt_F30Fs)cEx&Q}&{!rW6pPYL`&wgF0
zs%H%DVfCgTRBC2z*%^gO_%FN&S!kC&m^5eEEW=D<o{e?yum1YMj9{)du;n<~_vFUl
zD1jmwmJBNsrb=F6=U;_DkF{{$xrULNAi(|GSTy^k9x3h<f7P}<UcUROzl*`!o;@<m
zf7((nW7~yY#zDCzOI8dgm5GUoY4^N?kM~*x8V5nT{tS8D(hh7l0ak>&HA0duzfNr#
z1ZZ%s`&lA+b#w_iO`7{&KB1;|8)eiI%8D7i*N>#)=t0?cd#n7z=v`I-hjwGE((O@n
zill|FR)F=m5NUM>m=a-^;kTRcaZyuQnKJqqs~N9h6w#aNYC+zscUEox2g&G>kV0gL
zYqRQ^U}7q4UNK_@u7a0|Sw9NtuuwI?TPJz>Vz%7HhFu3$&LgvYY&1)_`P{ZnVF?3L
zpqlEsEw&M)1KBYz*0AWl9kvGu2blcGqM-wj<u)|GF%zx*O6cDC`c>gNzzzsFV`vyY
zdKC%hZA7<Xj(-2ZKpqHucWeG%<$Y;5)ouH=l9Gr9DV0)YNrSl%LMXE>G8P${NoB4?
z#*`sLBEu$ZgeEd8Wr)ZScZE`7E8{k0c-PhY<vot~)Bp4T%X2(+XWM)KuIn7uTIV_y
z@TnWvQZX%x^a>)^Al?dLSz22!iEWna4T>xAD6H+=Z=Nv&>fNH6PZ`*N6iw;2$Z4`-
zTlKhGz;p2jyOYsxR^gZ*{@yotFxxmY;x1mF*UDOLeIdQO+?R6P?RG<UDyEUlW-vr(
z7#JLExQVzwLI2{{*MFzLR)y?zJp3IB3pOBrhOAgTaIV&c#Ax)P{&yoHQVJS~4gJQ|
z>CQ*)6^x;AHpjU%<Xu4Ix`_%@bsCMG*3Z4TTmB;$OpiDigNP&ADhNtYV3-3b-+$sw
zz_=3~^N1$1W-yv`eb#c1H{}<KU$O&h(xD(-ZM}8LwR?28kEgNcIk(Sy$JO|?MGvyG
zvmLFFaTMwuz|lFEzL?Q6Ra;LD;L7Hm^uTke?HC8rAvSIe7mt>gAhkS_T2w8!q&QZP
z?Y2b|zcD|RJQ8&f^KQR;H$td8jX@~&@EAlPB*EK3L0au)36V`cSp|}3E;u^}<Bn#9
zp2Giubmb(ZJ<dO<;Zq)XC<T=JXuqleo^lI0TR|&izOD;fCiDi_w#1R@;)6YwKbsV+
zT7=vnLHb@g+FNXWrWJVpftH1k<o5IBU;zMCzf)u>odzPJ9-djq8z)9UiKM5e?d)ve
zaHLR+q{>W0R$B{f&Y!;nc@q(o5YZN*2q`;>z#@GH23?K}yH>PegtP`Y1k1C0bd$BU
zr#ZLC-93}EofXOW=C2Q%P(trAf659JR~0=tMw*E7J_QY*cD1titg&SEh!nq!g#aQs
zarcrwDusRzwcjBKxb9-F(Y`sve;1B5=i9Ql$#{O=2VP?C)iMtusK@&$qlsC@8UMvM
z9&a<N8yj_#g<bG*5+!aWUqEs4>D5(9^q!Kur}4?y6nnOT*+X^mQ+;&Ia<aIU*n==G
z=b1kfLl%Dg;qM7<l!`QLW|F#-lXEV2V5f2E3cQmFBs1I7r_=W>l4cPfz-%!IJZSr#
zcSiwqSE2fl=&El+n?}s=s^(wfOGYSFb(%&;;?Mno@cz)I=p=?fKZdR6oB&ms7C9;<
zS-Txhx`|r0r(R$5Sx2_XbS$7ed)&5DuMRn6$4qAjNOR1HvJh%AB%~TABkI1W5tv-e
z+@)P|2i~?9&bBl@KQaN_Fbw>ssB9;b4cPn$MQ`|PUPl1uM<yTz02$<*dF23cEiYrc
ztTEYdtri%95w=a%ZI|R%?G9>7CA{gc=AF@_$~<R#T5gt>meRJR%im_tI&e@!_ml|&
z3ir<@qAViDoD`)1k-yppkl{Vds4VYZO_@gn_HmfsiYf!yuP(SBy*flw(Y=6tQuH@<
zqBm)osl=3aKKd-$BHBYDt*R5N2Q*mtH3LO7*@;9s=)LBIZ4)>KU~u5M5i<ZFxaH0x
z9=6wBj0H6H4GntATZFkV)D5Z76`5sU#$WNEmRCiPjjEYJ6IH|UlW2JGmQWxrA+ikx
z1#+7rbasf-)4k!=*X&taSH}j&w6{QW49dLRW8>pfP&C5kvJ#K0x34cv>MKeHDf3L7
z4k=3&!#LF4sra=K9NqY-rnV1*P%WV89&js@>?z-NyRHOUjqKj1Nt;!;Tr6dwv{BGH
z;J`9x<kU0x<<*P26?>RStM2da1Ia~3(@E})lIr8oEB{|DK&SGmoSYoS{U$*AQH`ll
z4>K|{RGA;-(hXq5t1qYNM!A}}_tbI-a0IH4A3rW~FpBhwj^AsrFQ`B%4-!iICXN$W
zG6<>OF8U&hiT=N0&cN6WimiuKB4u6&*kkeclz9e<Q3>~MGS>a6%=-W+z-6FmpvA4I
zuV01i69Cz(V+qO?XX8W^btI_~d>;3Q8ORHA1{Wr$*W|odQb#v5bjw3l_>S;`daJ=)
zJs&6-vKh0_+^8n?5{^?3h?oVfO(G^n)sJU6vIbu!WIDVaG`K!J9M`mF^UJ3&K;Xb{
zkM2bio<8za^WmxwKxc>uj$5XGi!!mZdl;T927{#bZV)0K59}l*zn+nY(AU=4Ii6Ae
zdGCJ!tix0}mG|PdLj`i7eG!f(+Tt<5$c$r8Xg$oIMq$L}8vG+cN9ZGPIeu2vb|Hsk
zmScg6W)c)yY+dk6A#`MUPx)1NxC>Dp2l3HcsH=z<+Qe~y2m+o#&Mww*v9UJbq_p#1
zArywxc!EO7_3rEssC7c8f+d5JQGDXq_~e+^h~VgzbV?ZRwf*(y5pK9pP*FUFfW}S3
z&(Br*dJdZAf|fM#lN>bw<+Z!78t#Tlqr)vVTO%l8fPkq2U*^)!u0D(bWKln;`cBXL
zLXG>dJC(%%Aq(|jYrLFQE)J%c*Hj1+g5forbh<}WQIY&V3FqRE_EacpQ1VHd*`m!j
z<L{|OQ-U3KGp!s;La8H4k0VMoeE?ix955;X!@-otIs>znTSvOD#(i?|R~mD@dewAe
z^W6a?8uP`~P+H&m#ms|U<jpfqM5v4H#my_!K_yh+F1GG&!qbn|ZMksnTo|HbU`_qZ
z!&x+S&1XuHxg*?$kUzv5rV=b+`Quok7na30mgVDMzc`;Bns7VUaLULhJW%uqS|S#C
zJ?0yWyd02R;=w9J`pLr1=*lM)qoSV=x=`%)T!z@qP_F+%+01gZ4JLHy`#Z^5fLc0j
z&#l8lL1wyx7MftC*cQcK-6@dO#xr|F<_tcrc(mu};Ydj@%e9-A3T3Dmt;3Tv>Z)*s
zEO8F2cg(30GPU9(-KG3gWU`cM4XcuC$&-Jv0bEdFA|Im66tz83U-ldJgVN!wb?GZf
z{E_~Er;q*jUKtr84jLWlRZB}x-ytT(MkXhQ26sPCX$9Tsk`pThcL_`ZndG>f(H=|`
zV=WzE2c&49dq%VIgR#6z8E0JeBlfs^_ts!@;;sW;lTD#$Oo>i;+Z!?c0=?*qIA=tL
zJ;`?6r8Y-%iT*G5Y`TYW;{={D01?5zKET2CAM=Q2YnCGR*d77Y7-mO(DB6GVzd4$Y
z76-M%)Y2I)EZXI28Asz%+9$TkYqhN+;Bu{%90&wM$Aj;+B7?rH;NTVTIO_b)t?@4s
z#Qv1hIha*p&kKM2eVi+lp^qj$q32+@gl{;Um1}PjvM+Nmb9cWAH{D?D$qk)VkTgeb
zmHEq(*~n(&;(1}oqN1f3cEYqTcG2c+MIiCL_biKJr2FGfG!d0xlu-cGU=Sl1$lRIv
z_2%8fZpOyz(fIhIk9~w478hMyRDu0*-@nrMxj-^cVoFE-Azh^3Yh_(-Dq_^>`G<*#
zs}VS`7TN9`8ne1C*qatUZ*T>7c#1MN$G0@~EUOTc*|@voAg06(;unCMJZ_70&CHNR
zu@(~wxNT=qRY<X3fbsmJ^?!GmQkL8`zwr)q?|^qik=A>od)?#Qc3));Pd6oofr;aW
z5R&i(0H^Qps|6fc{o5O>T65iVVAhwihqiHg*j2?jQPDDT*7ngrU2GC+s8kD%gks*o
z`27!uIE~q(3kwU)Iow>zn3bi$08oA<6FEOPrDV8|VY`<IQu-|bYmU<Fy50W(ihtt4
z0QMUYD&Y%ULiC5lnm;Eo$#4JWqoyL8e#spKqZs-t2~S0~={&1Px{6sS7T89|3wH}r
z!k12%#{8?3W|A?2=8t6Pl#naK5HYcB&C!gDi_=OMmbQ@TX9$^u1IK!3r&tFx^s>kW
zR8P5QW{jxLVoaHT4>ewrH{y80C&-Cl{b;~TZ_t>l^`YrbPwJFw-1G&MLK|U)qxzx;
zgmJ<__q71h)!-N1+tb4Yp&7UVF8mvV1DxKa9l!-i%V}X@u?PDFhdvjrLk(Yr=qMn(
zuR*|Jsq>*yJYh~@`woM~;8x&z+|MCWrAc^pT(iI2Hh#$H<ZzBs|1z?VylQS!9Kni4
z6C|-`Hp2S6i;sofuy<@^@l;rryKEoDBQi3waKru&&j2x(re&|r^!$i7c=^vkf&>b`
zk@vKwS(huBcHyo1MMp>TQrqsdI>diUd!J^}_}7f5`X#A^@Ed50VWFYWCO&p6o;miw
zWBrJGD-|E$BItFCFaU)IE73X<p174}whpX^CTtVi;$MhGD$yGtxxWobI4oO;riLGY
z2~aNJE5@6ww=qNxG2EgG`J(ULOTMJY_O81)s?{^QsTeO)80B-7m|@$Wy16maL&d5i
z6-REjJnh)S8nt6Sf8UvK4H~)=n7kn4LDq2aP<LM4yO~g1MQuKri`*9nb!Q)RM;`sT
z0t$&;W(CVoB5&Mh7uX`)xJ}vTHQ>Cx@VOvkIMADa%vx2r8x0r41{(m-8Fl8Egno9_
z?^lR$<!~~$IyU}MI6v|5Z0;^}w3Y)OU!ZYxYJC*hop(w%tJtO9(p@L1UvTPnn3Dal
z+XEx$Qv^K|k8sS?e*L)a-Mm6NmD-D4D(k(?#oHbXV+?@J7PXuZugrl=x_Q#N6TZPN
zt9hE=L@mR3fc{I}vgvQDYU@gpZ4|W=!TRmkwTm65__xq|wRHR@Xa|fCa7@GSLfOxj
zCYKQ~&<~qHY7}E948oThTWc26G;3@1lG2pQ8D{9MvZAU)?i>Yr_y!&u$uGRdgL4NB
zW+~*`Q+GDe&o{~6|AQ7|?{MLQWU+|h75s4Z^4z7mUu;)&zA_Iv#`-w(rcyJpy2(HN
zo);zAa!62-&{&|taCy&+ee8jR&PDuc%7|ESfM%F&osxpQye$OopUQoeiS{0)B0n&o
zj*kc2g!td5HR6fCDl4Fi{5A6C;o<Qg%Y5}?Dso$X?K22B7+tL9wBviJ>mF%pXgEuq
zZ2&wzHV~GOan*3xztu>~8@y=ie_QxidT-M{m29{^I1`?Bhkh!;@m!F}0ughY`lWm(
z#<L^ktu4bYKN7YHIq((V-Ns-&H)v4NZxXXlR`xBKAwotutf@)-J;@9OMytjxO4XrZ
z4aMTZKKf@pLfiwNC|m#qM~-to{9C}q>=Y9dJ7W+q#8_?OYABXUr;9iYze6NFV-mxw
z=<&3xKTRB+Z%Xx#G@n*_6{wS_uc;_LvNb+NI{wF;y}dn^i4nk>oxQ!4Iu>mF@HfiF
zh)YyH)+5vh4~=)4TkYClC-SMnq$0D}tU%}3F_VoeB%ZgZHfjfd&Nz})Oxds=<=x{G
zFLo0G70KfpstV9qH*f-~ykr4aR^yC}Q$qoju#M~EoO2?1fm2<AG-d#sKqB$1s@l<E
zq;@Po=dtV7A9GBUD_iXs_wue;ll+3;&RykfL?hU@XuVTU8R(Qncim3ZI;(ZbBrSj?
z%iLKJ9VyxfUTV&TlkPm2S^}=bySV+Gyq<~AUEb4=vs`w&Su{pglnBWz9zhEt#aA{_
zgajR3b+sB(GU}$cqLC2PdJn8*BJR_?N}h_U4--XZ48jQ#Lcg7xFnp-ui52rPXv-`9
z`)@O(ca^BXM<D8j{}~6^l?JwP{NZTq9{%pS*m>K#U=Oi^@FuX~Dky1SFRucA(g1|J
zh4>EoD?SzxE6vHt$)drp@|zi}R4>S#>)Y}*UaMRD6*lNDQ=S`9%bka8I@-d_JQO*B
zDWY7Q?Ig(}d-WgGw4uH0>y7J^L<ny_l|`|O9(0m0;c3tv^jZ6Ag|g<^9<d&Rs~#8v
zD!E?F?4+ztW5$BZE4OZ}xV~b>7B5*gy_7r?$a1+Sv01ra$DMSHdn7C2xoxK?=rhuD
zLnD$k%XKTZ*oe*L4JkN-a%k^HJxr?2E_*Q%Y#EdAIC{48oMry==QVKYbN$`%v_H0h
znzsKO`2Kx=kdp<w3Ll|7i%@Hy$bn#4dkTXiXD)NXipG4P@99h>yy-5d%%BIa-S!iZ
zACIC?TbNRL;1@l7UA5mMHkM(yB?UrzvqfJ(kYveJ&A@M%uG{**2LY1`xpzu$VaLwa
zO}zJUkI>V3=V{gQwYRjam3D5`=WV=e#ccTm?+*5-egFzn2>bLGC`Y_x<Yr+(!Cru8
zW{Z&1TTD7=!Ah)ahOEFmFET&U>!*4uu)xC@F_W8DFC3o6SK`>#<`+km(G8dgn2rU5
zeO-B4mfBf-WO;9GQ7I(J+dBjKqo0>){UBTApIK)EH9nena1->KZ3T7#WF9+ZJOv1z
z)@6eT0MfO~J5c@B)>#kv6g+>ffeOu@Ix26&Y`WwsVnD4GcPCcPoAvr4B~p7`uGmJM
z=<LFtD%!<Ef2>ZjP<4E8`MGOo^_G0*9WImXYqJmaKgH-%*w}vr<exgkwG3mb(sbX=
zn>SO%UlOO(<f#gLZ_;9>D*7-)ZLWEPc8spL2>e-W?)V3yHaHxPC51b;Z$C7UkNH}$
zbB^^B7&>&alsR`~{sSK>37T*>)D_@c9lOgEF(5}Rm}<u~!QuXlF2LPg{oB(q8kN!?
zbN~KwWo2dRTd-!r=J%c#6tvWj>^m1U|2XPzU&Q(!TfPmU@ocuR0A^b+ZHLD@8ZrLL
zE>E+>gTV&=gc7G`ZXW;3;qE?U4a{<=Q!tsxio+>HUql=XlsV+!%hA`Kipn{)opWI*
zPemnDv$L}^8KNSy#f<dyZvY6~t?neKb>m-=vu%)<f(oP4Z4Y=e%nI1eOH#I7KzY&L
zjTG#x7`n^xwbI-16h4eC-asN95XeEj7bs(MSVQ9*kThctU^yIU^tra?nxgCU^432N
z<w&#bMd;zlz!WVx0Oi9pqr;l$8i?q^rDYs-WN>@xnCp?8e?F4t4)xH?pd*RqLs^yR
z*whTczvCA<EX<j=P&83?@DgONAQra!8h{Rf{|lfk0@7Oz+L3CDUEmoX<nlo@M-Nz+
zdT<lII<#d$xc~UvoJE=3s|4BdvY(R%g8dd+F84oqPY%V^=>5V~!yTeLW`E%`@8Mtf
z0#q(s2=Lu#0+1?PP)4zlyPTr7W71B%U;FZ`4M0s8YN`u6;Rb5iP9M?5SZ)Q-Jvj9-
zlNL%rffsC{I*11KBn4F5tWZI;Xw68q!6dC&auEe8Eb6v(drxg8q9|-iBGd?CK1L*N
zLylH+XKM)Hpk!DZ$U(ot7i0B|h#z!q<}f1Oi?8`LAnLpK?*||noI+}kP+MAt^(7Wl
zzX2q*@w#kh*ifPBWV8J-XX86RD-F?B+C=PePQ{3Jj9Fvy%Wi&g^t&dIOZ7>$y*=g0
zwUCAxIzDp7xEIXQcW;vH=M<D@u;OSw3osZg-~-a7YgQfNgUt&50cj94h6KD{;3abj
zm<2CrG1Z*xfF4k&o9At9)cClV@}OaSYibLI0vo+WJ5c|d=*A~TMBD`XE=pen7?bR~
zxS(C_1L%e=?*~9hZr&R9pilkR;PGdkq!m91s*My<IUkAdbDuX+*Ak&69)=niL+EQS
zVh8y(4-f6!-3r*ZkyD>8kzWJPu&0h|pqqJumFk2#l9%6v{03!%AdHHFQP!pcM!b0G
zQYCKUM6uW@6<Pdjb4?nOPYT`;v>U6jX@(4NcmZIF;CNy<tfPzjO({-dJl?WQ@sgVx
zH&ftwJk^h;?FBP4Gon$oSlYR7j--z&<cIpkY$QvwQOF1d4}>?ILkpU630*ip(TniU
zf!Y?V|A^9<_5{sI6&73_u7)oe8+iXl6Csm4Q%G+Ms)!#-KwB%Oc>T#4gI4E0_X75z
z8p_=UhA7IBF+8OJ2?zf?+RT+eTvD7MbJT}ELL+y|1_w~PnvAuX9tX(P3!Es$op?q*
zdHS@0?ZWBPLU?ATQeLpAx3RTN=CT@MG>9Uh%fUYN!}D_|_<`*^#U@PkLnLNL%?UyQ
zB3?QSi}W0d0>I04^9^K~sAEZJD0?!lT?E+06F52o^+YL4R45d4m*<dPg#46-VP*3*
z6k&IFIUo*1)$Es*HRWBvrB=l_*ySr$Synj5a|Tj>(4Gi>cZPm@je@)+5ani1?X9R$
zQryW1^Y01F<>yFAAUKm*b^6$#IYj@0*8TXY{;O+PRJd+oRx}wIi_Ko#AmtWWH5Yy9
z!k6M`x+~V9Ye1fQgXWb(qZNE3dy%d|HsUyqpt0)o>C;74Z(&l}EGJ(s%!9q{1r~8Y
zn(vC|0g+-uOSC2|a@?~JuN=``4^@*PYd7Ywnn#e^i7dn=y0*F5VC*1;c08GD0qV6e
zD<)~8+eE-Gy@-~YA788eG@<ISJI2f&&|jKk$xz2TV(1y;1aOh9<NiLp*v-gDBZ?eQ
zDSmjPO1;NYQ)+iJV$u4r$kMsE6X1$(ft;lF8W|eif{Ib>)^JB&QeV*|y+e=R3t>fI
zm^uW;!XJ}$X0b4@bKLlkB0`yFPt94GKg_`4tox@`nQ-lZ8ercaXg}<HO|{$T|1U{|
zXd)D!1chbGHJq~OgWagr0)N6L#e#Z?a59DCh75&7_@GLC#ZePzI<`*USJv?r7PIS1
z<-u(Imq0ku$2+BRgrt!6VP<A#x6!qOtvK4oiHo3RCn97$(bs`BV!94Py%fOB7-igp
zcYY6n0gPA=0y-H3U7vUd7yBeECL;z;TT@KAmR<hyc@yRmUjur)2RSjGH*5~LLG0l6
zaA1%lc`|E%r{`T3OpPJW!GR%2H%8+Q2xDCHo}r;oB6@?mbsEkQTu{y-y+QLuhW$u?
z1AO!rf~Lb8f}sxWZ7QauYH+jybEGdKSR{js!}#4QJ`mU_MM``Z`5;0e`bcc~@B_;6
z2r59{9GRcLe~00}!&a9Mxvv6p7%q^?c$NU_nNffF1B<S*8zra$SV4g%TU@4p`S&EG
zyr7qt5K%R%48nE~3=9MY1yzFGd4e(p&dIA$Ua3K*fnA6BOY>*7AsvWbt%mW)3LIoc
zVmV+p0{likMZ0qw)%Ku34)h^5;P@1@bWVWq18-{P!=rYA2zJnuI!6Zp00P2Iz1(|q
zPDn;C3AE#{i(H29F{Dg3UBr6sgb@ZF6~OS<DE{C99e^SY3l6`d34p-%c6c}^EdTQM
zgr6WD3y@uzjb8)uF4Ayn3XmzlAo2l8;6)DLf>0h`0g-P0!Z0ov>9#CQ;~aAnTWS&X
zoi_M5*XC!RnO8G_od=-d?Lkk3`GwV3xf_By1#~henaR^lkh=jyO?1Lo8ss59t1Cxj
zLoVuNHbo6Mzxd$}en(??ZE-=7{8q3x(8`cO%xPeww$N|MM;**aP#nDt;Vup&vDtFy
zen?7noI|J}4Dc-xttM_<6^`1P&6^>QVTW*Lr-%qEm>!~nUGFh>hgVwr>542R{GjNB
zO=xhiD*k>C?g=Ith`<vR7fB?L5r%@2=z-QIz|W7Flz>-!kgENl#~>;aoI4vN5>$aK
z_(q7@8Ne1fZfxVK8?dPL$JJk8!3+Q`(a0{oK;#u$%tWFbU<LC;?g63#DY_u+IF7Mp
z^6cU4XaoIDK>%yNKR`+Ifd#bn{aOnoxySCL7!k23u~nUyj43SKPmE#}TIKUlGiVl^
z*Bb*|y5d;8RI^wZp2JB<s0*r-4!Ro-^!pHfGNv7K>c(oDbqeDu_$=g|y8P<l<2(KH
z!3Db)%Q|RgM&udN_PL>?xKa$2_j&?8x6|E_7tEYO7^)G--<qi##_@Jn0TbgnCM)@s
z4l@gXKK-ocXcJg*k=gX#^-cD8-~#KT)$Md=z|$NfLose;nEp4WA9Oi?Qg7Wj?CQ6R
z_<rL&6bmKqkd)N#%IHi)6ESN^6UFd-WP6KK_{yRa=wC=`s69AVu);JQN^REm_IBPR
z9;Re;N>ymR1;%UXd=?N3pzcz|&6HNifbTdf6iOe>%r1BO(edTo1lZZ~G|eOWr?A>U
zW%K~@p(>u5beS+AWAVhZqWi&p`y*x4HGW4zYiNtq%fcIEtwK1H6#q^1?yu_?+^sTn
zxAp?_(f0fvcn*kK41<^K-J9&Efxo=?>jS?nB>QH6jTt>_m$fkl8jhXMOG-%l<7Pq<
zfPiQEufl&XiV)$C)}(%S9)Qn_*nZ$&ocfv10@sV1m~3~tcYgZx?vKeRE@5SLwc7V7
zl`NmA7{!;bHIi{Xk4y$M=sqByIj@6td!Ci=&xTF@UxNqFBPOjSIOM*hlA%gL)|(hr
z6@`}B$;*oA78&6oTcW(Im4E#Ex<E^*;xDAMUqKV-JoS9IIrVnSQ*0DSd0(F{J5;Go
zj8t!-I>==2z}`7XI(P}H9{q>N`M^9x_rgv;1O!#)yg(`{T$Ym!?z*rOsldR5K+K`0
zRTu<kWo5k%=(Fw#4rC>Ah>HZ*_833~xaX;3PJTf_D*;-D;>$gvyC=aC4IY7m5biau
zxC1|LMkTXF*~jAY%E=NLFc;LnZt}@5#h^m#=#0DwD%S!dKv?3ZFaDk!bv#gbmzd_t
zUu3(9QM2v1?l>YQ^!>G9!V!jR@W#j^B1>Do?|WA&+iHL2Ob9sCQ-CTkE)qzSBb;4(
z&SqGSeJFBuD2>uN5m)N$DhA^d2nolM4=3@VrHAGC8lp;qw#ELfIL1s!JVp+AmzZ`S
zl6*KgQn-BG3<1s@7&hMb(sFQDM=s3TDw%=O@jEI)5m-Rm$w7yJmX(``L!!^)f61F8
zeis4akj)dN#FH_q#b8;6du3JCQssr|o!C-0_RJv|aR>oaP{IAm6>xHuh~>{_8#V#5
zhhrZHWUjRrfd^C|`eImKNZb-{nLyzr^dXnn$7CQ_#mc*;H~t)Lr6Bckd!*V2hle*6
zn(V;++$FC`Rlt7t8}OgPM!ka_2^S2vY~4CaIu=B-$A;8k-MSPVGLYgkCr3`}GgFIP
zC>F3jDG;3aN%Xe7Q!XD(6~!bmrS={ERmaP`it2{1AI0}ygR1!=T*jYy{3k3VB=%6P
z9SO;$@nw@q<u6IC2xw!-lo%o|8ydkH2mnVh<Lxy)vV_Du5D#R71BID+#G1t(O@y5H
ztP~Du6k{jZ)<Wd%80anxNLZpTpP#;ud|>>ed{~1M%>w4MG+<@W7$Ax+f)|aK2K1qL
z(p<=jfaOb|5ryP#kBm$}PL6~ag%*euOGdZAz^j*C&M<`+|0+1iyoE-9JSZ4AlaU7K
z>r_ybjAHfhHz;}9wBj(uNG2C?aJN8dR*eXS9JU1bH$cec=enL&LgXPJEXsqjWGfaf
zv7;ZyHK{=HG77B&W(tWq;yT`g_WfN!NH3%``32<ahSR4y7zT!+!Tp9@%JvsbMi9Ew
zpxrlebb#oqqxpT}{rAelH0X>JiEJMgb~SoD@@`0r2YgNa?AcvxJiFOftXKhFn7lGd
zgIh5%rZ<X|UmIr`uLMU{2`z8{=6NcsPkv%zV%<!90k(rJAifDW_9F}Po<t#c=gu8W
z<kx`A3W5%hz*Ys~+pkY+OUFgF(L+#}AWGgw^V*ACPCg&LhY<*4V6RdEl>>en<>zv+
z6#*cF&`Gd)h==dCvt|-JU~S#4mlAP8f8(Z2z4%7S#~LACN_d@@)4spISUTQkY+f;P
z=A2T`DjmUX$#?k!gE!`|)BK{|MW(Of=i*+|`dQ`Xl|4teR1@wlTCG-aQjum+Ut{c{
z#<|vdWd?<PZSvvt6h{S_6Ae7+w8$;*`#Y5#{`)gOF@9s~knGFhGfg#<o@G<sUphFI
zU!wvf)=HO9+-#IRZviQ$(&>iSW<`U`NaCE{jlsdejsOwx;c5_gzoRt0gYiZYm}KI=
zJkH7rj*pM`<+$&$A7_}j!+gi6W-x-KI>zX*+DueIL1Ffw$p}u*9Wq{rS+q^a4dJ9r
z#;-1(VnwhE2Y#VEqF&{RHgp}TuQ2#tOJQ<oGZ4ZPL`FR}<3)VAaRE1CAwdvICC0$Z
zDPVE(uw97x1nKqAmshd~8Qc;Q5=1rW<7AW|<zxu&gj~;UZLO{A;YtvSwCV`X98YGW
zyo+7m_zb!bj}k2T$hWDyxuRH3cB7DP!p5+$Fkmu&VA<j@hv~9QD1}7C#Y527DnMPv
z3-MkkZtaw4+OPMB7OBO>3cS3$-yU2j3&JPS1ZRjBz6KliQ!rS9I_sNK`KTm*p2$h@
zJ)eT&7$V^tKwS@8yt*RM)z!5lrv43jty2&mj8?6Ze2uv>zPysLy1Rcj&o&+yo^Z4-
za{zf{U}>o+B3%MYno*dnOxE``dns*WFc=1=rabs&-C$29<%2yzj!sSkK1pQW_+9__
zUf|lXik??WMr$_5apZgGV}G21k<mIZJ)uZ!Z*Y0Y7*8|u2Qd<YMFgUBa7Lns7}D0~
zQcbT9A7oOdyp4Cqgy1<Z%u@E5n49yW#1BTUtAR42c<OBV;)3T>fU=Q#N}EuLklD=|
zkQ^z+#l@QRc|X*9rp-&%BUb%>p*0j9wWLUrCLXroP4chur!|#|Q!hlGKYM0xX}I=}
z@1x<;a1<^Mmb`wCQO{%}CG|`_g8G(@t6>%s69s@hMS_aR_GcDze+9eqOf#Q}w0xYG
z2p-wP_x#P+r7I|kQo^@ylsS96oc$30RudBw^+jKJQ}NSVBKAWpkdpU9|7)bP!Lp4@
zzk1MfpqEpvoJB#N3bz=);CEFJd<8F*U|I0*Sm`sA<=j*{D79T4ZNS83>gwzFKW4pt
zBclGtNpuY?bOb^JYwIKxb-QGgshArud}3v0CNHqfdViEO#C6^h5OSW<>sqyhr=r`X
z$f9KGt0xN`s$2(RE4az{@CInpte;Q}4Aww-SNACX@)s;3_|;dBBm4%DV;}i#&mo|V
zmAiDZ(!<-^8zNqDzu=8$p-mKj`O<_!*}%cUAtXFya3S(IeQ<DjQN(4|mrM)9c6=0z
z3%F^%6U~~JaC+FqPCnsY^-^hZZv3`SK;w7#Zk@wZw|G(p2r5soVy75RA&k_IKk<Q9
zV|u8tY2V|(O3K-73^+8^V5lQWMS4I1N?o*esW#ro7~URyUEeh-SVlOhRO)E0aMF4j
z29`U2Gs17+hB=-+JCsk}dP7ET`cFg-bzx7u-ARS~_1874aa%U!mCZU=&LXmYLmb^?
z{Mx2ixd)SU=FO{DE|lZ!pf}MCe}*j;uUBmIx@(~Bv`2kSj}rQ!KQRNwY5h16is>av
zaFHLyAc(_pgLt@+eMf6x7PI7#23m%@ZY%8`;i<y^zjJ;DTWZ$etMX3VrTrQTpOYX<
z%wq0sS;E6Kmy?yH{c<JK2p68Or-jI^%5-&j$}(L~EAl%N=#B^VMX|z{a^l41XV0GD
z#hLu@jg-|FFXLM>rFVnb=9lXV*#0sxX|03NCagH6@S6WC{wQ8Q_?iSuw1PBUPA2_U
z)MJYb`s5xaCV!Asv7JRac#Or%zg0+<P8gzE@Y=qb_$nW|iNQ23Y(4!o^&J-xf3w|b
zDX{})iVz~!%>Vx%?U~(rfbzUXnpiP7LjI)ya@1aCm(LL}Gt6BgPduNQ*q2KD^THK4
zM!ZOw4z(;JzR67gvp?qS+OZJhMxH;@$diSoJtkHT>mGF`lRs;z>mJTkwYm9!0B7Xg
A1ONa4

literal 66248
zcmeFa2T+w+w=KHOt)jFFDh5z7ppuj%*$5&6A~}l)NX|i`ZNh+{s32KIHW`$hQBhD4
z5H>-whzOe~VUxoh3(tS+ovKrH>%4pG)~k08fBn_;2EMSuoMVnL=IT4=6lK?}+Omp5
zp{$WReeyhovNRk2@BV8!KG}Tm%5(f<`9*ozlazV#zh?z8k0_KK6uFbfR9(aRTHHM<
z7#;INT_z?aC1I7K(c3upP`Q4d{k&hb;>E7fPdlF|t^5?nyq@FM<G(({#D&S*pE0YL
zJQB#7c1XC`Z?KIcGG$h`iBagRSD)NJ{L8MWr8P^ZewyL6J-OeL$9;g|R^Z80%)b;5
zfiKO+DOoH2d`r2sV$r5Q-*Vr-y86#|fiiEF{rOIw^ZVk3@2*C#vt6|C-J9ya@M7fa
zg4t5tKi|C)T+I9DJIyWs^K#{tO-m@YE+L1EZkc>}+TL%7-*|I9UM+!|q?6_5zL_#`
z9Dkp`i<ei?dzKND<wLH|ylU;*v9Z4?&C!}K{`zyt|D9*}Km1DnRW2v`v!^!UFjMhp
zcTL=*ET7xD?`JU=4;&U|{_s-~n}}lmc-6S`t?aw-x%ILC^m385Va6Z2%G}=~RFj~=
zmvOo3cxPFVM%9Z`tXsDx2$)t}?kKsho$vZuF;Xg=(^JH}=5&D@<NW2zv7aCBk9v1=
z^}z!N&g@f)sA<fwV0_qQz;)oj0okcrIlE5Xjn&RD7naSj?^MlqwPW3X{^X)Xi#`YO
zMV>!@o@LXfn28D3a8U=D>G8q()&h4{cJ|~1jTE<OtL$jJk9pOLo9ym=hiX#{3_jdt
zRe8AU)MGvasSoepbBT#*uUogSG5<<hU5Y`<)2C0jZI#E<3to6bk(`_8h(8~Fckz-X
zPVMiO70k~~he*0R`!QBi-spXxQ1~`Wd*$!EbG^e)Dp|jv=eNVxIF)@qjHWD`!B=xv
z7cF6ulZiVY?VD0Kb)c-W<^EO?->f#TrnuMQP2;oUZ8m-1UUQZecKrC^`}~B@mcqh9
zMaF!)4~Lj#ih7dv+TKmPy59_E_s9o+)1Ci(Fy+!DyY9Nb|NcHS{WhF45BD>_aq}h{
zek|A@rIlu~wrR7l<x^S*V|-kM*Pzg{QQg(mHPEUiQH!RTdg;!?$B%!GPM<w}nxCo|
zZq9O0{YzA&zsKY-TbfDv;qxjg{rcm$wcBhL%L<<ue=HL5*@h()rHgm44mOVcE@rlv
zpPO}?qQ|7=3mX)AxOIiQ2wK!}`;WHdUvcPf$<uiL{JEC?Ru-0*cos8yb@VW!Gm=AH
z`PtD$jNW=BA7&2q-nyiI{WPau_3dJgAJgyuvt)%>ivd1Vi#@lKHvHk<rqIC2iHUBG
z`3aw0yLQ#&x!81;ht|`*mMmGq@b{iML9T(t6Q104y3k{Clu>VB(f#!~bv)nS-(PkL
z(X8+7GRjqBGQ~QXmdbb5@6nAHaqK?l)ctj*j;HkSN6uBu%o=a6Ez^sy$I|4{e6>Tz
zQ$E0ObA$Dgii6tev4u0k^7!jE>&Y-thqJwP@oHspf2#)MI`x`pZ``mU%Bi=mfoFul
zs2%Ps6LRWNot(@mS|W@4q~oQW=+z(kg{`N$qx@@9jms>vmQW~RC$N96o?62$p?U1s
zG3(m!!^USar`Y)~TpjMJ98_=dnw=6A>iDv5adPvf&6{JDq$b}FBt|Jo&#JzD{hI0&
z9{hpz@4uh<Z5Iy-^q!mSI&<bss*W&j>@?UaStb?MJ~)_yzd7mDQ)8ERKhvry-g|yd
zOTUx}AzQ)+G5>1bIYv)y?JL6*DUA7{QjUW=cD(7ITB@6z<46yXn(h-=!*=+jLWrR9
z<ZvgMg~`du{rvprSP!bJHf34~dCz(5*}FF+Qzgwr0r%KXo3yA;Hp5n?GB75;Z0XN;
zXPD7S3@=z@O1xGG72YCh|FSqUh|eH-uR=)dx#!0}M@kp<)FmZc8Tl35D*Mkr|6m6{
zX-a#Wx|JlKdXji9*bNO0Z4okmd_W^P@#<BIT7tTQLsvz4s{Y8Y^0b!mU;)#bEE|I-
zJUZ%lNR@_E<68VHHhPI+LBJmQcv*jT>+#UQz+cWSd7mo6#cQzxpKvK3uaut8|5Y9u
z)8CYJ@bKaDc6N4W6clQ)6d9!)-raeFg|zQ)uh%AO3DAPi2TUG*=Pxz&addh;W}z#>
zy`o)nd-<EYw{M?2dGaK6L`q7^ZLnZ`Tc)_}_oIxy1{G{@4PD(Z>*nm|v(w`iA0O=8
za`@71vjX$lcw<o$Q;+%C@t*4FdnBS^i^X01aM!#wKZmLmnJgc~Yn<mvq5SsF&nuO@
zn#Zc7{4;y2Uzpe37lxZ<EG-k2^PH{n7}8VSoIXxPGs9(;4A0-+k42q+xXZbB6OVSy
z5s!&i#Z%%g1A_h12{Bv<6t(U36}2&3CK7nvu2)_NIQFw%AIptu<JG7gzP`SKN#l|8
zGo|VJCr{omD|qwfsE+|h+P*`Xm#f~j*N;9@lKM1UVBMJhDSfUgO15RrWAiOT-|gac
zy>0Sq{hFm*@9v`57%m5r*eL8adU?Bq>sRZBzJ|2zbvzo$x|vf0ZAC?NrpNa-@tJ>p
zc5H30^vv(#RZL8d2{tCSorxW#{-J?n?oQohd9CxSdu2GY#P6$H1TSC6O4Q49K5*{Y
z)t1zsKQDP=bvu_eW?ES^M($RKlxi`|+0CPq`6az?ZhBAy0XxmIII-o^qkVa5c#>>C
zf@kjKViD5|PX;*leLLhMU!K~y;nPFU=`ng&q_?wcId+ZX#BW2Iy|lEnBp%9|Ve@K5
zP#Gdy%xj}kb1bCBJeQPRoSY;ty?nWw?Utbfd2d4<lSTD`JUWg)Kdz&>nLLk)irII|
z$LHYQy(QMcQO227RaIiP->;fYD$LIgdT$TROM6_;&r~q^E0nh$TYiAxsy8o$ZCVrY
zUNrk+{%ukX6_zEw0$x+}eKE;RISw(!!)Klx$f=cH6yz}c^HWrz=X6g^oJwPsO(gcs
z%LqyLp65R7=kbJU*|vrx26j51IDY(W*{?6bM%rV2X_Y38rwfn%v)H^oIoxc!>`v<^
z2h`4%-QC0&u|GS_1zDvyeGT24`I2pIvR+<lv(H*2<c$QqZA2)1&YhCuHJfV|jHub<
zNuq+HUQjLHq=~>k4?N$#eXCvMRj`pq`=|4R%X<;%sm55cwMgOpfvjBTj!L?ZrHj@-
zKe3{-GO}%e(N!5)KgH-(D18wApiT?(<d9Q=<m+%@HxS`E0v)+O2k(0&8q)q``^uFo
z74=;D8-;v2*-S@V&qPREv45#|PzxbJJN@#*-r2pocOMCJdSVy+0z>XK)?j)@PVTji
zI>umpdH0JKr|#wIc46g-@~@{*zW=rGK2MOYJB36u*z(c@Oj8?AVyCty&CSltYzthm
zVntr<K==!tENe`5bp2GMfT<t7VEp@a=6J;2fVoY!nbs|h>EXzrPXmL?*N+Ig1ls1h
z8(<R07a(y^i$kR(drn2L*5$FMJZ5?H!s!8rgck7+tNHT3znVc3aO+LVQccm9T()9G
zS$(|*?oKa#e`iBl?R=a{%xHgBTSy@Dwj<{PMr%}*_9ONV2uaV+q!s(H?B2UqO!5(b
z*3e@m3dNp-yo+r^l)qF@jT7?d+A^g`DXVc0^)y;)V_sSKzP1**5Me8G?ZBr`V`Vu!
zRS}FLkRn+)I3mqxbA1wqBUEJF;6M_amaSs3ofthSo0esLIL~`G{daq@@bmCIpC~pH
z4BOZOqSGD2DD)uTJG&|Xd6BTC9T5<4F_&%p$w7^6nb?|Ev{Znx1ed|qwY{c$goJ7h
zbBc<*GaD=AmaX2XiX^H~mSx@2V{7twx6tLUJDNJarJFtSW0y3~;^yKyeSHP9dV+dV
zUtn9d-HEawzRFbp@=%e`K(!QoJ>$=h?W|v6eGC)Oc~$4#yLT+>*I&%y-o5+eNLQtc
zruSfL0X6h%q_lUdVGB?~qh~-QzVLel`1vr>>=`|3KAp%E<?7R-pDaJMDQNW@)MeWl
z`7hP2o*(;hSitVbN!l~p{>BW&{@08am$u-*$*C!G><lgDYQM$HR$Z=%Rd$>1&uWPW
zTK%{2m&dO&dQ*zx@y-YIa?^)@l_yk3%lF>gr2mz#Xy%9QSm178hBxV!ZH1m|d+Xv<
zWVZ@io{N&X&w?cV;KAP+Hf{RSv!jn9uKZ%lYOPF4ZQN>!6hPu)U}(2{akPBU*+>a4
zIe^Q<EnAc9J3lLyhuPQR9tx%%4Xe3q>_=HjbgUZF<yNfS#?RwD*%>6{F<~3i7%-RR
z(u8&8G&O3b5PbO5o->babFsI6^OkzQ!*aEKc0T%yV*iuJ%Brf5i+6N>eSYHZojb~}
zFMc?iV%C@*=RD9%w{NTXXuh)T1heEz?BQ=b?ufXI2Ghu#=B^MS3*BVSS)0VHTX)RN
z%%&#0A{EA^4{E-8Mxt0&tD8K@caDzf0NW`@u@+yR9-^iVH``$vc(hVa0^Y{coDfO3
z^GfDFG2FaG#5(L<adCg(^u)wv#H^myg1n8qy0PKnE<M=4xpp15jvqg+ny7gQ`B!Dl
zR^gfsf9Z%hM_Ck}lBDq5QC~1xEl+;W?8=!lcLRC#;vaI#{`wI+SJaZ{a&XU{J!PF9
zSqVt-&F#zQu?`sBXL)3CrzpeXTc1ixOXGTjc=dX!RsO0^(upB>^xG>#YQ3Q4w{y6g
zW3B?Z5w|jGX5Q7SS37JzCz^V>YL_RYRzpi)#Hvy7LrF<)ZRrXQkMHUQp%*V+Ot)^y
zwajQ&SiO<^D{@D3{8pq4zQT!jMn<A`KW-Ekq#m|6mHB>q9cQ|I=b511MU<=J3&po>
z!$3=3E&haC#HjKz0y;9^1C{y+aTmqeDTYIKq^b-CoS>h*N)dh4*?nyIwrp_F8EF0G
z+U;VfO&X)!fCbH4ax)aej<i<|eF+gNi|Zx1WV^IibLSYXx_T5D!TjgPhk%FnIZ_nL
zD=89qzXt-5#T~o8oXjUcyOfkZnJ9hN_GQsjs#r@Y3M@RD>?|q~Eq(VN?^kA6Hk?<9
zQFO}nl=b6~PDT#w!4Qw0cF*ab80v^OC@RdfX&b0v02<P;j(*mR)6}+#SUZ}_1dEes
z!*l}Woocw4lOTX2ZJ(n_Cqkrlj(xP-Xm|UV%H~CsqXTKbN=Wq1v>okM7Q%c0cE#ey
z&%b^>_hi3{%*|D-=dp$rBP8N7Eb7gXQ10HnYoFjhSU9aX-s+y>Ha?K#+}{|d7%p}|
zJ@F~2{MgvyZr}ck$DmNRw?0L~-ab{ftqp0<5iv3zaMipiGl|{f_pvA3TIYem1HVP~
zVhx&Czq}J7Vw21+=@x6(QF1_q>brRHVwUaOlYl+V@PY)W(y5aX7dGD_>#8BmG*sUK
zxyzqhK*G82>2hZFL`-WufbFU^Yt*hUU!xkYy8q^l8(c?@Xny+i>CEZV=K$tV<0d`Y
zr_=+e+4$PX2Q&J!n*9O^!{3cI-)iqaM#-L@8Ht>)5g*L&0FA<5Rp{wX+eeeDe)G?g
z4BPg1y4%(z@*<u-eE1ObNv2MrBmF#MphY}LcG(+Ryhchgr>vjr0^zc-oR9k5^Y_;{
zd)C+NigVyTK<vK>wX>s?ltc*ZnL5shAhRgua>REaY{>2;F#M>W`$)NnLK)!WKId{2
z?+pt%;_9}C(%xQ@6W_0E=Q_rTIrUr`54mw;>6iloPJ{Eb^c>-t0uR}=N)$}|xTk0!
zjg6z1>$rb~EcVsF@-O7lS2ebYJD=t7`2EiU2Q)xgpp`y4VMU`X2*ou0>VJ-36Q?Y2
zOZK(c=A#oG{=InpR(3FM*(YADKyLhasU_EG7=^+ZzpYO;&<n3__A>+TSv6GnbwG4a
z!ovFc4av4!fCUr@li%OpZjo?(_2;&S7X}B9`~QOL`maI^VRR_}r+LEv{MdQ5vQsEW
zs*Ak6v0N*fit$j{2a~ij&dB<0!{k$T$YTXrTcGIDK;F%lWWt>0Rks)WtjwlpE|zB@
zMC1RGzWmQW*21#Ih5th~^#5a*<F;%}H^W|$lLiaa*S~Y;Cab2*@?X>5BTvECX8A0C
z<0s`gb!jSpBj7yZN-O&oie^M4p^ep@P%wES7&N9D->Zm_Ojx~%mkW90)V+;rU^yJy
zJjdgZRLp7A`)2~V>HY1Cvx{F+sXetC2IY>0VA)6&yh&#1&oGpO2P7nPrROH1fC=M)
zdR4G{bDjG`5N1_hT{uR9y^wj$x#C-E7y!p{D3e>+o%h-lFanWxsE;2%9(AZs)}zHL
zM~xzXx!YZ$T<|0@WmA7^fsViA*i(SkXcVj#V1&fnMq_#P^J7uP0b?i<2JrKfgM=p_
zr6;gx-Q-ADfutkF<pEeGLSzM36pb}n5PG6G*WuTf*Wx3+zt(Nrme}2WIi^H8Syz<i
z{8ZW1R5RQ5dGYjz7Tja2*4TDd*66p_S0vy?dz!LRHuC8w0*>?!7R{@Bc;tgad-CAH
zv)EX!*|8=Y?M%zyTWhz+uVOtAivmi_YbIN^a5>8M$3U{h<7q$?+)XIe@4Aik1=C3_
zkzrcN?(fS=8BoV+`~I?~MzsLtzj;mULz3)+ddefE4FGj&5$+spZwzRG(<_($ssvEb
zzq2zwZ+5mgBp@I_(D2=jvt0lxu>j5{d9xrz^ZJ+M(2;@Mx}W=y%pA(5R53MN#>e=2
zg4qltOT!jfa2XNq1ERnMCJEu8GQOJ)$7ghG3OXN-E&6VK3LKVVHl{AZYs#!VT)h65
zM}~PVZnSh3S3!P0fAQiAvRXj?o1?xwaQ@}#sSH$m1ozD`QoK9IelKb^1biPwWo{zY
z7Nf+$dO-Cfr{@>EQjB8w@M)c^2|qeI$U-o7w~5crKH@ac+%VAfK?}G$l@P^AcFpk^
zGIK0#>PV=tl|7yLGKHc;M(PG$6l+alHUlFI7?pU!XDz*4q)inFOYOzUm6Oxc3A#D<
zZa?nqBZTEF2n9FPmbBjJ=xDQv-|bnQXL%`g4*~;Ca_@LxX-L*7oj)JY>M^3=pRpqQ
z;y`l_`|2k@e*6$LtKOZ}<2CeQ(-iQQ#9=?3U{CcJh0tF^g_Q3LM7fqGXzlWn6fZ9?
z?bjD?miV$o<zE>QOVmNtLhuP+!DvAFBp$cNUwZE2!=c65X*Yd*n)J?~MAfTJ(y5Q~
zOwxRPyl8eTRJLj9?l=IQRe$}(&okeXkaD(c73Orj&D(%*<Nu<ufSc6`%I{6oN-OC+
z1V|rAvR)Y@tSZOBblI|H#$p`o>=j?+D7G7rVUPSx3dR&)xkr2BfOB{=fY-JOkwIag
zz-KgCq+Ge=OBvwBwoi92#Ij`{EaPXo&o0^6(f>|+In(AH#r(Qrg!Mwb5u7<W)te&T
zH}dn-BdRgh5v{u_N-f9Uc-hjWvBz(&A`haTsA;I(g*p;==;KuRTx5E>V8QQqtAp7e
zKLGn4(`36FbR*(ZM$_Pr5BDUiS5TU*P!+!U2SsK#_AouA7kmv+L2k1j_=;bZkv1dD
z9eCs;Gb{4d*ls-srylP<J`k^8kV^&#RTv0B*_5KW+_fB1h<<K*fMLHX!V`cP<G{Kc
zP>p>+3Old~@!-ySXfghH$o85ICEFUnzvz1modRj1LvZN1a}Ndz#`PJ4ZEgLY`ecEF
zi|h$DjZg;N!UCcjY$b^=DCepXTGMUX3N7#Jd~TTj$j_8yP^6D_s0L|6HP_J$qpL(z
z2U1~H%kvq#tMr7e2iU|sHu3a=`|%siF{ACZYDnE9+gNdVdaQp7gaxo7&kv}@bD=H<
z&Q!xht0rj+fx<baLOolBb!7i?9#ho~0Iz{Ks+M|5j*LAPd(YSB*Abrj>SVeC(h*k*
zfSnO9pWv0_A*6h4YcoI~)WDMyjRrN``(N`S>f<N}gQX<1F?l^zQ9i{}CergBo}P>^
z0+ru*EI?qQYSXo*MUYhph@2)`|LoWjf3FG>7xKD-cy#s_^Gnn*Q<{aCD8H`_b(C%m
zNR=S9XRRH^xXMJLERd&6-Cfdc^p%+ncHY_)@1D4r>J-oK|EvfbFa=*JJiAT8m7gX=
zIE)^5+uk91OeD2FQp_n{K@0O}H#c*oqe_;;>Dw*r^+TW|IY})zZYN&k?J_&vHZP2D
z`%~2$WaA-|iZHX?d-eopnhs7sc<|uDXkUXs*%(s8V9oejekLOfn9RM6JWL!M95E$x
zc*V_|`1HS|wlrt^TQe}CTS1&V^hgR?HP#$K@RslIO24^!Q{#3ERB?Ldrel2#ye-q1
zKauL{>b^|1OiL{qDby72ca`xK2?E{rRzJlA8WJ`xZcHb`+_<qK$DymbXfJ5+wbR(J
zR6Bt2m?=S~;)!3OHomg#SKiadRl#<3fUpo+zhOi8K!@{V#3e9HEu)?lVWI+M!e4r(
zr>3?bxKbIbHt||!>4S;6k6aN`QsgqI&uP<|AKVjp#O_^pZ9))z8hVGzl*OgZx7Kag
zU^FGb2r}@pi%SZtiI(3KBo$hlpfNu7U{Q1QQ|KmpNy&$J@UjTl`W4;aez*_>u+Vx0
z+t2FmS&tBuC5PT`(5hepD_8Y+zsl&$NTqw<!(R160RC&X2!5n%Z9&ytjVkzX`7Ybt
zhYl$tA!Ig)RGih+e0qzCEuhm0bfE@lq4X;wCNZY09w@lYo3mfPc<~}Eca9pC?=Lks
zVFbnS9bGaSY6cZWAxSZ7Q$j)Ad3bnK`-e5A$p(Oi6VL1FmwZ^RS6CQ4e`NxsP0u~P
z!rz)4B6>w$dP|lwo$IOyXI;0h{d@~(rv${i5J9u^nWlM6+r?C|FKba;Zqs)G+7Uuj
zr=@22lGJJWjUS&Yx{vkQxov}rQP*TslmxY)*>UU<Yobn;BD97KosH|)2hitlj-^BS
z7??yrqs`_BTQy#qDlkKl-r_DEg(?Y)M;>LM*z$H#e=T?cgu6^DBa*P6FKM;05=2!V
zBGexk>{+*n@{5GL_hbD{R{B1-ZizMjUC)NNmf7$WS%MQR_xa4md_AVE!e{XSPf}7+
zmap0J5)>UB0rIZDKT$h{Aaiws9-o~;sL7jEJ$O*;%J55|mY_K?khoi<JhHuJhGb94
z${rLI)dCa~18)%sp=Oi5>oI^Mmz+p|XbS|M;qmeMSBAyu7WKTKL3%*$iaB;akn!7Q
zHsY!I>cT(dGBHgyfF)FgP->*tlpL8<AONID$M0_zfxOp4c0XM-hA>f&I!<qkw_=Md
zA(7?z3LdlH0-vo>;67f5E3*Pn0kc7by^LfnEdSs|R20!~@Im&dl>KLB>6r`TGo1bs
zKVD1ax(r&oL_Qa3^AMo%%{Dv?45R`O?C0kG(k~3?C1l-vq<~SUW1TfLkT+<K&7x@F
zJ=Uk?GaY~|0o<XVW?Jd&em2j!|2&PVz>{~Ws_GHNz-!W&m0Od)Oc*a1hdkB9vlS3l
zYU2A6&=_a#+_@v>K9-2AlLAt{v7oiBt&Q$}388Ch(0ks(AsPGv$sj5WJ>#(kl&BVd
zqbn#bf_#-bme>ZRzFYrc=ZRb2e2r%&hS<%<->%?DH**1$q}KoXxxOG20ou|bDNR20
zFfY%eOB2)6TiZNF?wf^U*KEO*WaZ3Z1VswwoW16gL--A^m9Gy4)RZa{!!A2~i<!gP
z61aYJi+h_7Qm43XYrv4C3pB{~BNdU-HjX_tpCA?41{(R@yC(qHWJD{B8V}hzUIp61
zGBDx^uU_Lf=pV)7uWj3phen_9oEkk;AlzYJw8_Au%CKpGtRH&iXFN=9O<>fXyHT?K
z!UPy*Sz>I18|_{0{P_0TNKbq0aqVk2hB>Wp8;`PM*I&RZyn4BpvQk{yJnM~SA<a}p
z5BBz7sEm+YTV$=QtgM)oVO+M6lZ%T>$KZsww|56Nu5Ek~Qu?KM0|4*P!Hj)DTw87#
zdQ7zYI5&ay5Z2om)&$SUn*$4R3gPL(NG`}zcC#$N5aE=ls3^(mf&T9Z+F8@8v7r&J
zKUR9;kLf9o0s<`c+df&hu!su^ssePm209^!2|D~bi+%r|?uD2_s*42Z0o3|YI!zXI
z)a9w(-@k2r1RPM}#Nn}GVP()7<3PR#`i`?~NTqkpLHH3DothsS^(1(EG};vTxfarX
zf)Bgf$M(%r<AXLMNADPy1@1q5xVj<h0q59Yn~k<J>ZdE|Dd3ypppBvgnI6c?#n@#w
zNn@44Nn@z34;Bd=^HJ~l*(_uLBDyMs9Z^QgpgQ3>I>+|@4iE(04jJy8h;=jU1~H>Q
z?(v1_yLXhV$&>$587>}!Fc4aopd57zN%uJj^@IEO%b8Y$RlSsZbk=kQ3_0~VnQx0?
z&W4NCrY8cl98i97gJy*{JqsHIQTRJMIy#0mBRq!Xh^~)iT{heZxM-{`M(^!SfI3~%
zUczVqwsFeqf_VW;EpA)yCd$A`!0`*aI;xMk7xm?Eb_Co*73lU-&nDsA^J}PuLOPug
znO{vBdmx3LIeYfYr{^sjt=B9I(~9UIWMe#nqL#i(?}jgUbA|CTJK290F9Q$E7hU9R
zoFb~!CX1j>yUTj8+JJeB;Ax2QPUhS;`9m^O=L;aUCF$kW#kq-gVj7(*rza=lP^Pi>
zCtR+20e)4LIs!MvSNtrkW$sI`fP(Zm?p6yjn`Aj`2JN}IIrv@zT3tKVcw&gC(_DxE
zpZh)E?LGVS8MY(RmNqe)Ab`KbY{x-pq&j)wH_vuq=AMIlY$_-;DGzB(GxbNTN`^Ll
z0NFSfYO<hnpXS0f7gd17Z*fy<wZ*(90kT_bK?EBG?#^j8>HPd8!*%Fl0jkI0kr9V@
zf_HDN*-FJKB6O#gzRl)apg+Uu^QE9Z?0fvzHQ4oXEIE95`pXhZA5k4QAg&9iNV1-}
zWVw%*w}P|ZO$Xa9e5$h;+Ek_Fxb9Rh%E1_s+04|##-Cr#kAr_3tvB%QJS|-*HJzwZ
zPsl!8VD?^3lE8cL-lSd=Q>&znSAiR}GGR!}&x|PEKM`=3W&aEEXhjv-d~FIT`NMaa
zicd=u`Pq>pnLp1IO5{MzuZCojcFI9$V%P=BZJlZnyrEy#>DU;D(A`u}GAN8|1vEE3
z&2wY)Y!P=(Y&k3&8B;VrCr-=3>;~%>A@$Mb{Us)UKB9)k&x05YY`ABEx{Au|+1;Fz
zRu-BbgCY*mEnf5Wi`FY$+ha_-5$?hG6V06nJZa71)*x_$uT3D|^a5BpndJ}IvM9Q{
z*Ktua*PtkR`f=jk>Ep+*$u^c}=kGgm<Qo|`Vlyce78I=c#XxqIlniT9gi(byKaG@Y
zUY8hLm#Fn9X^-02rHs)5s&Cm1iP_bTC1W<8yW%(_3jM9uM9<935MeN;DabO2g#fbX
z+=#$KjzyoUQVaV2Vq&TRqu%dl^zOz=>#S+SdDF)C)9QYU6B9p|3%9$xgajqb%$xm0
z>+4?Qxi4YT-uf7pycP>^4*@U^a&^plsF<6_n`OI+=eQ2NAgM&#+D7IXCqM^r9_VA6
zb4O~M{Q}8PatG*a<Ea~8maJIYeP5*Ym8b@6IYe_8X1@7pzn@f5gyvc%ipvJ*2K>tD
zmz6{72j-TDNBQ=%>_7jWj0Xz-utMPuuVdJPBcry~;)nw2M9i3*Dw>~7A&M6<T_9G_
zV4X=vMP?}-o$NaF1KRRX(1A2#HMZ@d_9uyMDdyZq1jjgROkj;5tn%`Qb_Q&wO9ZjQ
zy@4*KyklOz<U(DDI+`wv4N}c)V9++&@oO4BCOJ1x8@_$QI@Mi+gONs_h^uV_bNWO!
zbobsNn|2=a6Vxf$$HgV)G7v^*+9g25HlEe1RvqS%hkNFVColocLMcKbm@W`7<VuL|
zUL8X|Dxi7Dt<wODF4=s~Bw<?+(HqdDo@L|FPCrdNi~)}x5eHb3Zceo$UoP^nez6a8
zrQ;kRRu9AhYGoawir$pg<LzS+nr@HMq=Obw=85}f@8P+C&bU3gY^JG4ww`Vt&~1p~
z?vT2KQXzu9!V1N_gm#v7I3zd|2Pt7;BAQl1zN?K@=49p7)kJAr3w=;vc5y$GY!SEU
z#m%>hhoLCM1OCOMglKi^<pGD#i`7Kj5Gv2Z8v3GqB|JHdnWBM;to+i`(?>h%jBky(
zns1EB`%;)S(7dnW^P_$Ka2h<Odq6J18+56M9y)Z0YFsfR!3|Z#OiNo^n>xE8?K~mE
z4r{zC4)eHhJ1V82|2uv;1X$cGCm}5hU`>JJ_}}Z-zre%9kwgL&g{*tf9Zvm?Y6NPb
zj_#||0gHy}!rH+a;YbYrmGT?}G_&t-{^7kZA`UviP&-iCdro&Ns{fS)3ZE<~5_Z4H
zrUxEoExW|ixkgBOokiwt$+}nGsi_-&4v)vmA{-@D##(q0^8Wb%;694o9fyrW412`|
zckRpOW~Yhu$rE9~ItzgDSB(=kXR0`7MeUq@K*OAZ==OZckuL(m?hji=VKuxAU01Zx
z#o9Lg<QGf&MQaD)9b<vSljUfT3BOZ3A~A0UL~k(~+dVMQB5Pg~FoCKe&-j*McS1_x
zXGh*BJcwABS(atPVVa#w_t&?-8g)2q4NENRLLa9w?_P9Od>x57m!UtE_kyf6nKttl
zckX)?Y^98-epY^`<lEXkViDgzudnbetCg7j%h@L1TLavJKID!JFuI8xUK7W#0@Q)k
z1<Fs~_4QSSt~A;iWZ-7@a$cv7INL(mb|anFfS4gzEBuJvaaeVffzivExMbbAMSswz
zyXz`7Fgcbb-97F7T7o7@I`{6}u@gJ$SdY@~aJT@|38NBUyBr=_fmdO{%R0NKzXKFx
z7`KUvZmn->kLQGJNli=IU2ke5mpXakoQzh#@vqB~1qd+N5`kgPDAZ|(WtB9SQZ5n4
z@m_d?O-3|+eEX(Jn>`xFx^bh)$keBOk&dpjNo7X5!FPW2tlzF=yXWf>XlvybX)&j@
z=(Bm8J9Zoe-CPTDg#~WTr%yM7AUOpNMZTy6jvKI}YlB|@e2HIa;BKeGhlFU5a>%G8
z7V)1ff@{~W3-EN>yEMeb3`SuW#mu{0y7aQRVSSc(+L}t0^XIqC1@x^X_BXH@0^w~i
zFqFA>i)ydm%*GaM{IT<LohE#Y1;V|btESzCKkZ9nP4)t=%$ZQs9fA2U)XVxHzyPWc
zWN&AaMA49CZsHX;9Mbpc5hIMv4vOuH9Z1V7fPKtI+>oWLE7w9tdt82Yn}LDB!t?hw
zPu)1BA)Gx{Vh{)8PWxyEc1`LPBGEM!RGb-X+dvFLfEB1MorH^6WKNyR8D%hD>O`JF
z08F9NV#HZ@WKT=}9`zjkX5=3|@1m^@r%2<^x@_weK#zxb1l-*s_YvdjFr$1%Nmj$u
zpHS~aOErU^u55c9a(bmg8N2t4T~NsnId~L=TYh?AKtS#l>6-Hi38~G`jTZ@B8NQ^%
z#8<k28Z99<TQBqTGig6{^5l`kY1GL%waHmeJ)H^h421M^_%h{9;{uTf0lMJ^chTHb
z0-=I)YU{v6TbZ;2G|B-_J85>+b*@BNKy?pmK7i7+y}e!0|6s?Ltu9OQQOb&|!CU1C
zx|W`D{o1v&L1CKoS$8;;W-awiJ&qi+&T%q12Q(Spy3`B!$(qqcpN_?4g$2a6S>5o=
zOVG}Ur`xxp;_q~w$Ih&Sp{MI%rgeDq(<)d1E!>K18d5(PzA?55)Uq~jD=cUkk&}}H
z!ycp}a68$cC>bnS0w&lEtS^dHV>c_<5$O8LNQd^*o<~sr={3U7)shOdVLT=JE75%*
zr!?KN{oPG!TEs;oBkV%%tVYJhw9co@QWMWwj<QiM_(g_Lz8|%<eTDkj#K9U~3hg}S
zB>DdQsh$L30a-9uFEOpAT6`VLN0ALar&d6()(hGHHw(+9k;W&%=d-3j`;LMJGrrOS
z8q;#_Vhi>>Ev6WI#$CI_SlCbaU4>kh)Q4Uaw2C}Wbdg`01fr1?)Z$g;khmSEdXeZX
zu(qx5_d8z@K4FHqoKro2W95csuk&3=q}Xq)z^k)v*%JTi6@NJ32O%>Ye)$r9o!C(;
zHiJ`7)hTP2FnfP{9aX6i3nmm=#d8!S2X^f`4kQY@r5ZdV(7a!gQAA)S`2i>_=-_i>
zXVH)X20YE#vknQ^zf@z?w8R50HTiP`k#P3!-=7Oyqn@IlgtCW@(#<xvg-|h_oiD(t
z9=4K}kZAtZXd3A$TsEHTK5m`Y73uop4tMWjiY2^TM>j5CyDb*pi<lxsMMZVQY(cnY
z!8II7+zKBrFgY*H;PEDI&F?j>Ta+Y*ZmeSAdNAh(G`4oUSF^w^qr`7}sBvD_=j^4s
z#EZE6w+Nkq)rX&_4R2Y#A$ewYmH|#aMBgT3eDE}OY&w8y47=(B$X4`CrHR7o75lSz
z@5!G&{S|<JyM7V4U7sR~STT?^1{roC+zjb2MB2Mh#pn>z2?zjcz?TZ9)j$!YCcnO5
zlZ2Z$Z8s^JAgEWgt}BTU*PYeorf{w6bISPg^GE~WfRcZPz*Yu67o2qUR-62GTjq@Y
zDpW3%9u4d|&X2)QlggdP-`5wGr>K9K)ha8MU?Mt)SPE-^G>=|xP4o&LR~k552Jwno
zYhAeTO#ezJYgSV-5GN3dm9`g<69XkO>|4sM`B&6n-$9laq7`~%!H5vSe&y%7tkTCD
zfC`-8?Z`9sYn7(FAx;$A<KX|w`2w+oh|`7Wb|B=;kZkYz`Kcqv&rS96k%&ji1&}vV
zp|W>Da`qH6ORHjoh4+ZS00DYJ&YK8us5=gDbDt#zGB6n{W)qg4SPURB6oAF6Ni$W7
zX>0NvZ!sNtnL}=bV4Q2)eoelNL7?aR_a`%~n$%C+VyXhC0|#9SJOn+cf@+b#<6(UK
zHgjzXd~6SNq9V>#Z*`JXPy-~KytCeIAg3!GP9a3SZg4Jy2nA2)I5S~KU?RxidelI)
zUM%8ARQKth>824^N=!DExDOV@ASPc1<xJ8wFc$`i0e4^z2zN5?teZA<jhzh%%l5|>
zG90<;15%85#>o}2=ob_%09)Ko87w&V9~J7jIcO+S_7EEjhM;+)QmQ-X^q*Z_L39t~
zIXd6x2w|{dF+uTShlw9<k7qJ*W&tPuc$k^mi~DCzMtux7pBkl8)9BzVBK!9(tqc)5
zNMq7*fof&}>1(aXv+-L97EQA`=L<cq!0Gp9HRoc=Jpsy_)g}HMHm5%ie?tJw@(JM;
z6>Zo|+)W*sE-YN<-e|)2Q#Pw#=+Tgk?FM?kIoqxTFu<~cScnZduqvJ7`w@)<3o0Oh
zSbYYqWYmmlRrF5-%_)OGYur|p@HvpDx6V@~?);t(c##CHL^{TI3Jcf84TXujR8U(n
zI+kwz1bT0AfVyxWR~2JqlqC@j#^_VP9+{&@uQo-tgS}|5asg*AT)$vexnHV>rJp8V
zi1i`bB2#jx&CQoxQmkz1(Ow1n9*e-*SywFWC}3J^ZzW`4qkJSx8Kl-?FH};U!{fF5
zOdY`7R0b)mwD}Gls@i<$;*EA-q%rrhP?3}&#J0^~9Qgb9VO-L^!1ZqTQb_GUM3$MG
znd~3y?dmcOlk#ljob78ZXyQ8m@^+KK2{{>TyCF=B+a-xxhMP&C>pzG<7o7>^+&n8B
z*v8c|KjJH)?+btZJNjM`oB?fSawNm3bnQ)W$a;LJEJ%SyTw|kOl5_|siy&QQUwi?q
z8)B{kDaOh?eL1A{@pKM3tw|*Hkje&Y+SxT63JXGSG@*OqVl1>eXS;5IFM#D(?|?yJ
zzDDUGu#31%d0eLfXq~UvK~^q-4+ZRdqyk=IzgUb9v=HAdtz(G6KuS(Zx9gBWkort_
z6E%Btb&+#rihhAueF;A<14Od7ssCqXqyZF}YNTJrn<dQIs#(BsQ}8E)+s$sWHt{Pe
zDk4-1mSf(`rBb9^l6$bHpPGPJ$9~KcTJ92|-<>Gx@T(hn#<`3%Db%d@FqJ^-2k{7y
z<U{{whZYzzf7FMGFi#z6rj_ZeU%#Fd55{7H;I~jUxEr2?+hAa}cLQbMBK)`u-Ys^C
zSewF08FZFZq2MFMpk9HS+P|z3aQr8LvmkCW@P@%jQiC~N>?BUml%#7RyAZx;|K7ba
zFjx~xs3y}&2b#%gfEI^@S<IUSNNIutNXCPm4ZM;8G*er%V11Yb(mki0t(voy4Gj&6
z^&C#i&aSQ)ptl5+Bt7*hk)R~#)f0)sbK4t=yo<b0@Qij~Q;`%3HumK~jbtuh>76@w
z5}ow;i4(O2?k=$6ws3m7zq3dwpvFXcX`_Z-u)Poel-sj=cdX{?iwATv6G?QzY`S0i
zBI?8;SG|N;;u(Ss7g9Ws+7ukb)kT{?M4khlA>9ru*j<le`KUwHw+_9~OGT`Thw4hm
z<nl@4&4(Z$JWcI<j92bOrip=+@$=yrnGA4oz1Yq*2%!uN8XcROl(d)|78o-i>|%yI
z-TGdaV(gl9uS7PG<_*GHqAiGs2^kJuiiBez8WfNtZQMT|ydyHsOO(3Zq}UMqa>#vg
z<V(x`e>y)Rv<kTl=+cH!c3nv?TEFKE7qSX1MyVd9csjWh7*+^2YH)U7*;AqQLQao!
zq*tT#&cw!gMriKo`2tZhXg!3*rH&w^lnt*eqfn+jm^y^Nom>1yuZz4Rh^cVy6k_}*
z`ENDnwFDxaV8ckxIw77z$LssB?(N-d|7HO^QeJ2QGo+sRsrsUP{#~xcn8*I*XRzqA
z8cLb7BJrn<o=4te3~FA=ZQ%3$sEZw`7qU|ubSz$fH?iN1y*{}P{_2Z>2NzuEJ=hu~
z!`xi=b}Q1*W4bPQ)$|K*7xVcpuRxE>tF6}1+<s)P&Z4|I1I!K|SP1}yP0_3~bp$|8
z8|C{ZnH$Z-d=9+_T|h(t##SDFcl&!Hn}EqTNola)y6h_{&02r{J{FPFS;Qn#*eZm+
z{Cl<oL{LuJ_~h$S_Ov`l)+~S2ujZ&ef-A<mt4?4&jlR20`ChSLJG^l0%J9$Ed2MxI
zWJa<1u@^rou#JBsUL9CStn>Drc~=Jz^%!<_2VSs5#9bI9vy|epHiGn(UQk9*Vy#@A
zD{jn`-Re26*Mjt`zhzBW#-&f|iP8;I*;VoND8li(hhR)f%ksfkS>4=o7oXs3!Q*vR
zXKu*f`y;>aXq0XHyH@t~+_$Gm(`v*~RI>|T|J}Q~Fx41~|3$O#e`Jpv85zTvKSXU|
z_3hQ$f3x{v^}cU9L7~LJw3?lC^XAPf?hAdwSDUu|r;U9jDlCg!Pa*3dcEA}5kE<5&
z2mpYQ11}=shJA{o)wd}-x>(Usg?rit4Ig4E{ynw>uQMPB5~Bu*+NuI=9lgL^`di3u
z3cvU7R>;C4_<!_7|3|O+ztLj--<Ko)VSN4n#<Kg-+=k;{UH}q&@b!PgZ@aM7D3rvD
zkXs0i5uQ6X|JtQ(NLh7yzO*gwM!?tOvM*y#tFWZ7q{OnUV|nTGqhQC53&)lmcy`-u
z(d)0=x-84yUQ_+Bn(MEnr=suNJg`gc*p8)t?Os~_ac1tfY@uapns{D=f^$Iq$j}l0
z72*yR!D)Gqet&}EC<oJ5Mn5uryh@B;^ceKCYM3S@wqO{JDxbQB7QQzxyyr$D#XP6d
zk^Q^WrV|snNp*0(1#+O96EXL`xUEA>z1+xCUm@@gj(32cltEMH__Q5HaMkMd<e&of
z?v}i)Hvg&&3k_hdiP<Cg?B=qS=z^dy<CX+>4)-1!=Df3<LOJ*J0a;qVkAiN1P&be@
z2Y8g2H;`LqGYe}wi_C740wstss)(^2^(CN35d&>P=Qb}xk08tF4*a<9JwFrii2-|z
zU1<vyu~7$p^K9kAn;Djn_FD=w-~s?&N!M&0eh*DF36;}G9bA2<i6!6lEywD&<EW<5
zk{~sR<?!rP$Hr6mGd|K3k>@fvrsYleb2Jus)GdtH+l8z0P1?VnM4A8a=SB=p(4T94
z3d&-c>Ijr2!*y*bE_OA@BZDiyXk!vQy<7(te(C2wd?xxW&P`TO4NO{B;6=}!d`QOO
z<8{EhgMxyBmFP;Pym4Q6`{8Fm*SRAV;sw?m;DEq@O<Ettk5Bzg#`LPh=Q5^5(ryg>
zU25AJymZwGGMzgXluy9dltz2^UHG{5&n=>0ibZo&YlkxwW~jxFIFPR?=hFX7#3CD$
zi=ZkA^*B}xS10A}#)V0hLa&Xi|0A{L2&94UTNfV8V+W8Y;73vC6nKa#$?@in(sdY5
zmhFFj{QeP}7Ssrao1q&nbYvHdO_cAPXsJ?;KJ$dTwsY(xK5pCj=VKsyVo_~!Lp5Ee
z=hE_o8Ko_SYs<obO+0uKtOZqM)ykEhe5&wQENlMUA{LOJNH1k#e;!Qd&&Mz{oF*d5
zG)TcDcxQz;NFT(xiJRJSqu@u)AVSWuV~mbICvLP{KAbuL8s}pvx%4*;G~w&&OXPE-
zY%SJ{CPo8*Whu7J_#wV7e85=(g6?CMS&Zyc$VKEUQ-+NAe}DIXaSS$Uf5pC`jJZJF
zKs!;jarl=nXN@mi;^O2y5f>K+OM@!$L3MOEp<}zht;hhZR4um2W4P><II$Ox8XhH+
z_VXT$Z79--ibd?xAgIFY<UuBZXqPSz7LdgfD(_OpT72WTFyBiblTLT^gleam+$TMG
zC;~Q)KESj5{hch<t5<8&%~SxTjzPCpId|^dF=pJ2vgj@e27g^f-AB}Q(t`K(oVo^n
za@3kE+N(UHfZyOoc)AVo;SS2eZuP1My+DE&7BmV@LNTEtk3rACWhn0?R!4hHjrAu$
z%aQg+6rwm6OL*f5k(9#%0>l6h_Zu<56K<0jL&*sPfLp3)J}(<1c5a9X^K?j@>>PP8
z@sM5uYg8KW<D5iiH4Xs?gTsT0HnjaacOC;_L4$$~y8QrgV>D&a)nODP#(tbTAfut@
z^n|n|g3{YA<)KM>fgnETyZHyOtXub-#3#t7q-`xZ=hv1~r%n+^2q`J?1)}yPUVJ}H
zia1yyhLF_oq?v=WAYIsp4<DA0vfklYh`4<F5I3Q+cHTMu{2Uj{st;zO-gEh7nlJFz
zfm1o{x3>BkmHOq}SCB%-zJCdE#2SG&lW}v)ee$&wyZN-7+#wX8mrlyc>WGzqmh4I`
zf+oTi;0t8AHPUk|xTzzCO5Fl>D_?d{0Bp}@6p0U?KD|8YDK-Lc&+_Ij`Gs3YUk+Tf
zWVz;Dp3}C<c#@mPKRFDkU6T)}!rXJBLD$%ujg3tPF4~qcR%ithu+z)P$lQv4^D!8<
zkB-bHYX^7kgDaE0XF51!P}a(pdPDx^@wCSBECE%L?-i+$8uI;fWo6L9yPm1yR^OnT
zNMhw;&~Q4VFmB`|Za_W&v7VQ>0aZDCF&5P?<*v9*Az!d&KpM>|b%+6}cK6;r6?`IT
zE&_Sf2_`VB0+>HkTw0uhvWiDEpDhgSWzL0bS3swH5)m<wT7e3`-+mrGC6)~vIvx$(
zfUtd?S?cQ=k>>AC9K=oyT2sTyz>}dCG!2Wl8?k056c(^vjzr?+YW%YmHh|KFM*7w`
z9J?fBzyhicd~5Bggk`BsHI^r4A7XySR$)|#4<h2mFytR|J;!HVW|;w5WU)hV3mzDR
z!-QUlwc8Dvqt2;bFNUxLFp`0^p=H$TCxF}_cJL}p<P%h*S0rxsk6^HFUVSpXS?l_}
zg+aWV@S{v9n>T3XhP~PMmEm-?&md-mg_e$fUcQNeL?*Vw?~R+lI*D6#S+7Yh!|L<}
zAJYgOD^M9Edk5@6LUS-T=UR!FeA!Gp1dmsJx*ph09?w>6P#NF^UkD`MT*3aH?Paa`
zpUY+A#^4yTW^V@AeBfA_j5B6(?<*6UJV6hxUfs+|G=Vp>yMz6s_K){9bfn*ZvXnSE
zt6)Be>2KMSBQ-m!LL1^G=P(ey)N@L2)4#hqf=njqD9myqoR@_TglOH&;KRmuo4mTp
zLrrrBp&@GE@-1WSUm!jZS|w(>nJ3WzTm{V&dH{7AhtUv%NNek}W&g}fE(U_Rw-B^W
z0~qONSin(BFGNJ?hl|>ZgLq)-OZm~T!-g<f>NpSRqQ{$P0Ea!z6V`5#78mevz3?SA
z7JBAw<kox!K_dZNL@rKeAiW?K*Ncun^CH{&vJtBi!si3TkHJj8w3}6M;Tvfe!TL_U
z7v}y#zrc+>OACA)Lc!Vbt+$>d#?TtnD@+pfGST8I3DJ}~ef|1%r{XmtKi$qeH{47(
zLUMi(_OUMUaW+-%J11>GP|l(2i1|xjg8Ru%0uyq`rye`QA^5N1C+IfvZ|VbUWEBpp
zT*!cfP?Qy5IyTRU>sbsy?oi2A2J{~cgA<8M2woSpr62sc>_V_!ME4UPu+5Q1h_7!{
z0VEpgdXTCWtQSZ_uR?R3oltREIN{`A78uDb7#;6RgwyMbqp;=<?}btLvjY_oG@QzK
zva%CLAiDf)bcWR9z2?&~=r*NR@V7VDefs=)*yKpSOYDYNY;U)Y`yyejx%4~pGhNak
zAGMPE|84f}FU;N%Vh?KCU8-T=`NTp8Zjy*c;?99#VTni@YDwB^L<}WTAHw#x1gF=q
z8h+Akx;xc{fm*!j9{g-quU5Y}wOSIqEpP876O$egf=;=E$QtI~UOlII0w4>wXgr0#
z-weSvMBF8LcO3MYc%7_d5Y$OX842hFCuA%<eHR;LK?H<ZcahEc&+UZ=+$_@K_=p^U
z1M^H?Viq_a-PK<VGM)RiX}je~(?*858ZGl?EiC_|@bh<r!Tr>_5upzTQF30DeIpfZ
z76;ITmCUB&fuY>-x7|G;UM^dlCV@64$!M@pU~bRw_EM=-a_|6k8XgR@Tt4&%S4xcd
z)*EfZVL9~0ea?@9gX4%<);JvD<!58b!XT_8IQu~gs9ILEi9m_hCzfenHUVPEL-Vp6
z>3(h+Bep%{u1mSz*kxn|(&xzW5HJ@7mXf9-(ne40+xiHZ5bUb#n`Ww`WSL3R{&<T^
zWK)BU2H6@4<Hi_cxb$)|H+`rCW>_j^O}gW~vY=~5jtZd1moWXt+l%=H5rWXm)J&gy
z_G||}Q~>4$V;TshDHe1(7wiJyq>1oLI8ww>I%2FUWEGFY&fesqk!Xl<n+}3y3SuAO
z-HH56&SC-8`xt8kmikjNGNkArT`0uOOOtXz7s+Ux->IR6IrJxU`06H+mA&WZ0(+9t
zH3e>-N=ytiF$@I`EGqX9%zdMbJx?tCgqc;5#?}&O(RfU@&)-k~0IPKP(#>GcCdVb1
zm{2>N;p(b~s6KNf^&L_7L!>-&%fuEY_`yQ8vd98gA%}@Yc+Xv7yVb7TgqoGO{Q>6B
zA?ZiTXih-U(IEy4>>S;?j)f<_v9Rj6AZrs`1S?1sXzpGg4zJ(2d>B-8P_)<<uL%)#
zpu*n#T}>Lyb1fc&oPm`Rxqfpuk^=GhlHzS@q*9tEqkZAB*MtD$juw9W{P{dGJz|yP
z(2r9%vgrVD%|692>=z&6$+v3BRU=F%e!;`7j{qIC>AC$1Hm!kWot)4{%2RUw63(r{
zacgIQBZ$@r2k4k9`gh{M&|b!1C~neaW{LC|zEe0q=dR~aE|iNNN~ive`koMrWMu?3
zR`_Q05Y17^@$TQV=M;Gh9Bwm;^?Lwk0sc9IiP1B0b9UkB`98klZ}Ud%f%C%C%a=e_
zd_CB`W;6d$;?~#kU#R`@*#(P*BJ01uBLnnbdNiIF_9Y^$k~)De4<YK+P3+iSC>+%3
zvL^>)u=qMVZexu4cEa@z<EBQ$)NiwE7BDPrfk69je?t6Jz)#d^a)OfucH`s`#BPea
zCXOM(q93kVk;R4T3tw;Q7RZ4AqTfCL_*CLb>a}avQp+HDu68^9Q5HUTmx(#PAozSX
zj((*asVefG*Og8oSr5(FtKBYBW^QOEUFK9@GMpS+s)TLJ;zks?4N4LZE!=^e88CWm
zdUKmHe*~2zY0N}T6AME<S`$M6k6DmoF~4yr4M}DqLhxl66vXU*o<7FhPy=WOQ&P{X
zexwhaHg{>!{n+141@2k;?)^d}&_mWE2i1@g5?+INv|1ptS2gE2uySx1d_A%3+{eEY
zc0M47Qb6b~MDKN6sCBL!LXZl&D_LMKfu0|Sf{u8mp%$ooJy9Nw*SN|17FXx>u{1`7
zh;g8-`bKQd;AnK45D?{9`ShEOn}{2)AjCE2<f&6AUf)(9fiK3oky#^dwGZNUj)Xxz
z4`>2j|E;J{5Siogj&8r-t`I^Ku2pNkR;k~16?j94VEgzn%ZJj^UZ9RDbPyAZDGb3U
z&}?3haOyF}W(hGE>=kcYCEm1o4At^>SabH(kCB25D_d$S-%BWly{N-UXDAx&xsvn?
zw9(mCg@;73l?;b7>GmQAZ;{hcz&I(8zPLLZ_N{X3rmR|dQE2Hkbnl-aGNdZt`WhT#
z1&msUs+t_P79A5q8p(+CiYFiq4&;cD`H52LQFvND@%u__51sfVvP$pT>Nz56`f$90
zV|Eibi>zuIO~`9zLbpjCi30sHT$Qu?_U#kz=lIBn_V^l}tmYTMt_K7JXdPpa8{@2V
zIya)XL%7mN?#!750UFt*Glx%(*oA2Qm@J0e{HYe?oKYOTNKRy#f^`R#C%Sw{lF~+_
zJcn$z=tWkkCEu7|Z&~~n&c6h}kRI6R#<;W?_y>Tabt9Hhed%7W;q71_S7l(q_X5j*
z$tVECIDq4zl>83iiiDot=ea22w4!1i_~^(kCsyR<nj{^O80mu%uCV2iBi1AscG0j(
z_X3=M%<u#tVc&D<K@-3v>AWm|dZ^fNn$(l1PKe~KieemCfgCu8b}L0pm<kR#P3@(^
z1`o>sX>svl<~X|Zc;c;VWA9e;UB-EXXXND%KIqsWJ-HT0@us@EdLaGXO9)q^4R#n{
zMzcfZWTNz&c&M`z+&geEP&JHb1=dCA_41q<DwSd@d-BCg9q#kK5hK<;%#x!I&<q$c
zD~9Tpa5U&VkA_P@k$7WCUWfY@WpQtE-avJzO<`P{*R%pa@Mr6wCr_fkefxGb^wY98
z?6+&reI@=3Y&zC&Lw6!QP|00AZULQF(Wd+|`aI!2>Z2Kv_9h%`cu#2>E$&d?!R*`S
z4yIK~30P^M`>9~~^Fd;AAm=%d1qc~BlQ_+rxM5OD_|xJp<`}DC)j~%l3oC?!6&f5x
zH986ev5!1q(``sOapFV<^ldb*YokEsZBmatcZ}E@NGIW|S)8p`&EY-M9yGL=D<2R2
z09`y9l3+`y)0~`~<iss93m6+!;5l-F5B}90tTm}~BK(wQeiDwH1yo$QeQpZ}b0k00
z30Sv>uN1us0<rUnpUM9WS~HNLEE~H}w^-#4!t_9$ChaL`MmU2&q7E{Qm?aSe%+bF~
zwIb&oVb5FGjLnyQ{#=cgib36bgBuUxkXElRkI`6x6H{&7nqh32___I%oV~q$a)#($
z#jsd_<SR9mK<)@(34QehaN$5l`lCaOxx~n;kt1*Q@?ERwg&;%BfwM!<OFfX=%Zt-K
zk6~;`9wi%W4TqF2IDu-&FtsQUmaT9P)ZXbe%)DNgcLx4oa;^t)Qow6S4%4jHcTa-B
zvip&{#+u@<SQ*5IhaMh6uo2UUb$0u5$=Olb746vMSEtXOtwG^Zg98ajYd-d=I_gTo
z+ri354lLD1S83Be-<%Hr!722+!oC~=yHh-z(p<O~6a$hPhz|t`rWfT;1uc8ee1z;R
z!^?Q-EjaVmI+wmzE$;j&tW*_Jc;G=vGxfgJD$_VN@fsSpIvowrxH4=11%44Ov_|oH
zc9Ft^j-%bq;k8u}KS?J)1_YMG1*(T|J2=*)7XO^FpCM;BJ<-rt#+>67_dsKc#q0Kb
zdmRN@zoYgdy6j=JS9r>Q43!2=s9rY;v3)-dK!NM(+`loFI5|Uubk7WlA2@>L_nJ=Y
zCTjo`Mrfwcbjv2tX)v0IwbTQX<6-JLq{{QL7szpp#hDNkYGb5LalVpWt|#1=SG>%X
zl}B3L2SsOQnjL8h1l*L)E?u^a=nQNBnZQKp?bYkQ>snqgbG!a_?U(7j>S@VbAI#Se
zg&f3bZuM}V^g9}WR3#^tA^MkFNdx8Apyp65^78a2%Y5IhWl+WhcjxfnxPjNz;RgN9
zIW)tw;36EfyitO#Wo+26;Xama6qsr2HnPg^gAmU$vUm0cPq3MA1A1f7kIa3y2z{<T
zGywXh&sn6^HLA1HV(7<*wWY|*rpaQ!4(X$_23mwcEi>XDK~=Z~t0dDgAA`sST}=QH
zO<}ssf3Oj*JGe9t1F;$5UA4Bjqbj)%(@tK$h;&8NLDFPbq9D?8i7we}SGv6f$x@=H
ztqpLm(=ic?Wh6Edl}=fdLlHO(8N^IlA-0W4BQN}W)vko?ibC*WzfTUiDZ%RGX5b_-
zaheyJS0ZMEu@B#M9@>htOSj?t;_5?diunsCeyol%qfrCfR6u+A;*Z(T7!l~d?sjTw
zsxOvEaA75WoMUNICydTA!UbDtpsOgRA3)VuH`npg@C3}s<RItH9wQt+B7;gup{*X)
zE*_3zye-T>)FKtq73DuQyw9b}*3#1Q7;_am9H_-lQ9=gA%%LaHYEp8!TjKHjU)v2z
zEZ&>HH(&ocg!R3}6%llb((JrppZAg!9(v?}!y*Jd2T2oV&5D)D+veAyHRN43*m|D%
zuTrC5Pgs{e9@&hW`IJdtwhzS0%$sFSdA01UsI)+WF|t_T^k}JzC+w~%C1j|#qQv~&
zEihDi#jyc9+!sQrf9nZ!mH3jb;XV;l%Hc~N|Ca9<lEFKM`K4i*uroRiTt-i?bFL?B
zkc!iv<-a<@NoNEzw0^ij0J%g*bddhxHQl@poA@Qiz8MAjU#V5o$niA5z?PsfJ>Sa!
z$fk1b&pvXV9(6DY9l0`qO=V-m6>ViR=I)AeoN7h-#jo_CH(dl-D_7T(7^^|Y$1DoL
zIlD-);bTA{uKv@*K}VHr$)PZ7JK&r4wZp=cV6ee1_0<&O>)GBv11oA!`mce_(=6w|
zUGLEzpO9=XRtN4^XKHQ!shLX^n^y8#D}RAIRULi4;b=`-pzg4KAq>@ks~`yS{^mo+
z!POmXRvr467odUB<~}G%nv~eLcfjyePi)Bu1~_Smm;y;38;s5wPQB{bpkLZIBav0o
z9wqht!hYK#U~+b>7u@@Pa#yeEBtoG)eE1yguta|%IS;wShQQN~j*00!YaHT1Ed0df
ziOD+$M=Q`vIy|G~SV_RdkIAH&gN_tAe0gS0%wKBe_X)zn5!qxnn(fabn9;Cjh;Mpy
z)DiLH47fG>r^iE~{168p7A$)Kvp1l-O_;R}&YrCF!`TxZqf(OuPk>mE5?92=Bgh*F
z%e|!(JDqS!_sNOFQou|x0H`|2-oQdZ7J8d+J@=mNH6X{V=CflkLLE!DyDS}Lan;b3
z{PYVg5D5Fxx|tk^esCXlIJSYlOhk54Si#J$ikI`-CQ860k<Ju*eSCb}9lhS;Py-0a
zF5l{q=ZJt{&k?~t-Jpji2?IZk%@Al@)~4v-;EpNEk_79#eqo=RbW%s+#Ob|Cp37JL
z5UL;M)aO)p@Y5{MeX%h??!ARNsJxRnpv*2e1%lJ6RjaD;9*mEi{@}c2NIt|`AP3{L
z*q8C#y?<Xw(z^w@Rp6l3s~3XT<>%*nZYAt@2dQlrJ2-GRE6oVH$5sQ6p_^u*aFnuP
z<+#MoV^)Ie#=BVGmf08Dcq=%t<9wK6wIHY<jaE1vqhd2SzY@406?%VSi{dcHC<s&~
z5NcZdMECFigx0o#6Z6Tf7G7bpXwGzc`dK&n3~QFg6p4-ej0~h<s?$hs{CAuxbJkl4
z`?n&hP*;z28?-25^o^Ou;nfidd^Rboq|h;^Z*a&#J&;@Lq8qn{#?pfQNYW)3=P-5)
zxV$mLuzyNS=`VQ;)kxRN8Xm&HsXI7^0BCJwbP|;7R&v6S#2DuOR}TRWYuhqXQjQj)
z5kPSois*3f6LqDfQ%uPq$i5j$@a6x$rZh!PSl&Jb&us?H?H~4pzf-Q1a13pEz2$Yt
z$;7=mezi>3U3pUM=M)Q4nYJDI3T*HsdYlG}C<=t_f&TI{Keq^E&+LGxNf;~~2Xaw6
z25&vpB|RTJ>jsd@Zm`I;^NB{7RPFavL=1T%Fpc$Mk^9u&Y<RMo2j^uUcWJ#6(<BUn
z6M8ERr@YITl@67<=m8Rt7K-X!*CC~m|F%K#kDEBLGWkQ*e0hCW6^!D<E-dG5um(MA
zd<#tbwqk7G?Hjow@LyaYr{*$vCO+T*5QTAa4pdAjHn#7XAl`6`>1?|WM}$g=ak3>8
z(6jI8O{!?5emaU_lmRpZ3Ua6rA;76t=-r_+Uc3kpbp3q+`C0`9r<m*SXE=W4qP7Ug
zS&O>FPZ_g#A~&2atI$NMHgc*d>b}RQ3c;mV{CIzdnx=r63$(dcGWD698Zb^8!-)_a
zP3-T{Smbf2!LZPWUAK`A)w(2|TC`orGyHLe3^1^$<?6L->%c_+v>%$vC*2{me)Lt{
zoU*GgbcDayryj9zOs<HuQMrbLSbgNRWYJJa&H_}i=TsljRramj{X^(~u=gflJ?H)Z
zuQ6k0Zo3(~l)(%pd$v*-h7dx?(qbpHY157wJ7p|IQiwtnqJ3c^YqnI{kxDzQD%E+s
zzTEf!_dDl0*ZH06oa=Xf*L7~!b>DY)%lG^Fem?K_>-~B?U(e_BrT<8|tltcvz);O7
zaVo?9>w=upVK#ewE93+#i)}EY6P#cgqw2fiD3%~t+b*5}ZGE)V%wg5SE`Hy}g@lP4
z72>266<IN{r(IaV+o>*2sbxp71a;S}m|nASUGvprAFAyYN)Dz1G*qL%NyblDXO%sN
zXGnnSy49=SfxCvicx(*b{?^(N@({T*<;+ph<7m4*y3t4*(+hupp9Tf(^@%fU>IR&5
zp?M>`nzv<3xJX}+EG?o!S93P_z=Ol%b8>Kk8YfYrx4Ky2E)bWez(qr1Am2o?hGu1a
zs!Gz02$d&ab@dT1_LN6Mf7Fi*LIN@^mK!3!7Ps_Kdn#E7_SHVE|CZMlQBngE)vC9L
zbsb(9lbQ<DAiXO*OK-gRg}p%EFq||ZpqE+rt@*v;hrd7#bP`#@iUu<AA&r(Nx&`^H
zUR|TsL0jFtw{^{TzGF9M|8k=8(-qZ!#lhb%<)8c<2g`7^uD=uvOmLB&TZG3~;Xi!P
zYx?Qd(J4^<2Pa3#1~zisyqR+4?sJtsP~jbLe!6B%`7cn%Ri4j9)>lp3Skv^gC<Y%6
zyL+FcZ?9^wh7SjgH0saU#!hgDP*rSk?wxUSz2`a0>GP?TUtp(HlDanTl?wGv57&vA
zlM&>ew9CHg)K>S{Nj0+Y;qCfY^hG%GIyy!?nIqQD+9H-g!?VWPmb^4j8!%MSYFw&w
zNb+I#yYtcRi!^vnZA8625`)D-rmBlB4ZkeoIlP*#3VmM)O@*<Pr(>3<*jlWkKup^h
z@K%rj%AC8Kjh8H0;sts})p$WNVUOUCJ$H(IjUF|~3)Ygk$uAW|k<E>ksSvp#%|EW#
z5|dZd!DqYAp5@(Wj<yeWK6>O{nz+&F)(=^Kdt*?ZcJ#DM`MVb<I!^t0(c#^xm&V~6
zqT8TD5P%>wYt^7P0iUL9SN+aT&uQ30`{LaLCY}w;vOSR(i9E?-OAIN*L(|eCoc``N
z)k3d+adDw4ack6*Bx|b&+~^Zc9aJW4x^Oc1iJ$7%C*i1MQCWC2cq*CC6Ip?_sq>Hm
z6-&<uuhu@@P>($o8+2<Wkfhqk9HMOA(myL^25s+a*ATO$dx&wIOZ$DVjqVTLa(Hv|
zBX2aw;%k=_3DfI-^{$z7)SnbQ1nPQQQMBZg0t&?2J~Ah5;aWA41fgXKc_jFc8=MsD
zY0!x7q{ejEV>)S5r0W$Q3MHn9TGw*%>P%yU#!vNcx*k5$r$D7IJ(dZDV+#sYx^&4q
z?9!&^2d|BFoq{_Bs`p(~X*6$-#1+*?6MR+4cBbS1bB+W^x+abES%sQr(g;w6*_nMr
z=#LZghW%imzTo`aTk~VQ2kcFIF?$`!q{c%VtL#63uq?To7YT%s_zr~_Be5+nO1}|E
zlj}SwoXDz}DP!g1>-46)yj}!mu+~YB2Ea745eW@h0X&Vjb2n#qGaWs!vf921ENFG}
zcr`UW(dMmRzup#~ulqyUCuw;V4adgYnb|gMkJY-f$&1M@;W=*VOrr?cIPH<QdQx#i
zhN0oY3#dCXtX?jBWy!@Q_4s`)RkK;-Fg$3ijlnYF5mu*6cb#}twcyG0#}!~i$3w&7
zm$p<}`)K&={;)UF^OmZqi$9-#m<CnzLcvHyoCnA09nnp|!ou`lM2L>I^<<e23Qyk&
z&YK%INq5jMXR7_{l0l0-?JHa@<}E@kwktI@a{I{;U&^?;vpL(RMdvSdxlJ1FwmZ4P
ze;`%L2d^$=X3v5-SYqq)dx?EI=ci2FpkJu)<)JY3FWsD7uk-0h^K^^6S4H1g+!@~X
zp^es(%Dx4WU1Ot-$IZobc|DbXgo{1lX9MZGKmr06;Bj&uOtT)q*{X4J=-<XU)nqpc
zUue!KVMK`mmWWcQ4Lr7MhNv!8udCO5(BqvHW~W!LS+mAdH6UnwZs3xyvPU+U#i;9E
zf(>X{zrK9`{<gdBo_G8^c~$h~ZyMn(`8QDTZN7XrcMiqJ>{|OD4%S<JRW!-ID&jci
zY@Y6AHLGhDfkhOa36<=DOtwnq5O5%feKaN03$jsherHu$q?>^>(eF17DiyU6NnGzx
z3^A7oOqcf|e<A2nVJt0=f{bw`gA2;r8D}LNwHmW8`Rwnf`WaabnfPY(%F)`lH`wcS
zUO9Ttmc-U}6Hl<F?HsNW@$cnnT;N6TS9^UN<oegEUt9QS<~=xkd&!FVrzf|%bLWnT
z!UfkzYW`*z9?LKzAyBypf%joUPUS;V6p`IHlROyJ{&n=|nhQQei|EI@E339NtqxV*
z8uv$ppct3TEz^MLtBG?B)i3LAC3jw@suxrFNAg6C)(<dNUmlsF4FSi7Y<pSRMjoHP
z{!L8p*L&TTd|CRq0S#^ApK@5{TRYD2b#01xs2cXzztA`ZPUx;k{h?l)5|T>~H#zm0
zvEq=G?WtjAemc6ksZ<~##^1GSWk0ITh=ZMiBi7yBa$s>{{fAF`7ALyyPR)!$V{KSw
zXXAK1(3!o}K2Bw~Zw1ttmapy)`AR)tQbT>GGug8o6LqCK&N=H(?okBd`3w%gCES{Z
zr7jFYVtA2g=b`+pXxQ=N$HfUYzRMmmjdyr3EW$}u1f#+&8!{yE^+m71h8jajE%Gi)
z#Ifk^7jW)W@PlJoL*5b#JpPz(9>X<7gyLEQyyckb85roWDClBDy)tc=)+J;5x;`!&
z_-<>~QQFL&?wx8^GlXE**~xRtQtnvXFnb6*rP)*4#j<+3xxViAe>a<;xQj35Fv+ho
zaYeFham;caYI%dO$@vR^yYuVqjVWnAQ@r%!^hmz+>!S}|QPqtGmwztYkdxQr$hNcR
z@6fSny>&qI3>IDL>l<l$9l-~S?kD#)_mAoNC(Pa~aFw1~4%cPt2uY8&F0F2JTc`m7
zzejrSEYu~(v^=&Q5O<PkPOKsKOI%F^V}pa2=hd^M)U++9xVSAfmh;odM>XGRgtQu^
zrlw}tbf|U*_x&RKHX<|_hos!mOAKa@n22~U<L3S3qxUcu==^e_rha)KiKl(*mjt6~
zC#L~EOYqw*d}4}Kv<QIR>mNQ$*3}Pb-Maqb#UA>_-XsWW&EKwEm1<~rGX7Dluz5yj
zUh>j%42>_QrQwxwW0~M55tG~j%UXVSH9e>0=tYNH+eI&;ObXCVj((J!wAHW_Jk0yt
zfPnmfGhHXj=HU)sF|eaBZ6Dl>UTEB9yEVV09iH0n>*|>?<Eiu6-$%t>^u6{9A>8=}
z4>PPgdl-8$%I!PPU1w&ye8-)D|5#Jvx^B&yJclEfPi8^{@NanJ^KGt<I4dP-8#p>l
z4d9ie+djMYiSNXmH4O?^k{mLe!e@(}#Mjin&B<Wnxa?f*MA%R>r1&tx+s?S{6LTRS
zMw+9PRQ~R2P*9N>rg3YQt#kJcdv5;Q$cD5b3ojm#bMJ}Y)nGU?>d_ePTuTl41LKu<
z!+*pNX|Y)Jn>a_w>ap*(o|0?nt3HbwmYqu2F#olCswh(<%Q?BKHu$C$EIhb(>8Ibf
z2e!c4H)sBElkPjQbj&SDC9eC2oitF78P0$f>jR#u94YGwHC?BD3iH+vX}Koy6B&$>
z(K2W2kYGdm8yCm}J-T%@<iydx%hC7(9B$W=-a%deZf~7xV0ECd^e1R<6%8MH0ro<0
zx;env&^~#500`W1pK^!>L;v_?o&M0uBq3Z{J979zS>0&T^%>Q)B*92Z7b5|m$5S!M
z0<=uJ(Vk=DU3J412&uEv!!7-ioC)g@F78WSZ)n_2`|h~WRUL2~f<fWJxK{T|JRb37
zTA|7y3p2dt+t2V2D@0l~JN2F0_11W?qU#*MUYo?xsmD^n6l-hQwvJ>Wt=g}=ann8d
z?BBhE$>r}^=g5=^>-<YAx>GysHHn(^@$u$gH+;~0?WjHGJ$LmT_)TzzgPg=u5;!4a
z{dJY@N1T$5cUDa^j7<9@_D|{-r)_tMe!oWD{D$R?D`zGe)^*+sWWChOA#UZJ*1^HC
z9BY6!9)jXDI#375nHbY^S)buiMMo<Oe`y*?0TS%nWPPqvziG#6tf2G>D}>;eB84G<
zk~G;RyiwTC3Z6^?o{^jJ+Q}3Ny1%a35+G_+sKJ1$GCE-<tuZS5BOpRh@lDWmi$qnd
zXKD4m)#>A{AGFq7pgp6;Z1LPEj0L|CR((B3qLg4?$Vk$-;~X!0f8JiTQ8bjwIWA8*
z6_ZvChJbY1jG`hlHU}_pe3lFQI>K~c^N)0J5%-SY)Jx03I!OCORORg-b;o^bUI;_b
z^S8Lwaf^3sNepZG6Ctq8;M*Iv^$wh|W4uumG`rd6fjCj6vZF4LbPS@m3Eqeop09N5
z1P(Dd0nC;+)9Z(cU0TF%=UnW0DANPPaB<<|VDY#eF07`AOB3-E-Bpiqq=%aoA7z)V
z^sn+(O&>{{bnT_3Hy&Kan%Zd<SaRPe7qYmkxbr5}Qyfo2tsYWL2kid1&2M*`G_p{8
zA+e1}5M6817#MH1=f`7WkrZkF49lZjh_S<Qc*;tdt)nCUj=95w000wB3|(DQ6@#gY
z5Jy-IZ4khuQ<pABi^qQ?J`X;Sqf#Il#Q|dZuKM!mN;m9tmw?E1>eg)?rk{_;VGNq2
zAwKN!A9sNoTrf9}=H#2>sw;wFKbOsJbq_|;Pxds`#tpL2E+DGX5*oQj!`1oXLgsLZ
z-r@1k@mS`XM3Gw^4~0f`GHDFcQqJh+q(TE&p%F{^lw`~DA8uOEhNeJNCq#NxFG+fb
z;|rB%lf#9(>rRZPHGdy!Q^oy;G7i~7OU|6KdE$RPinGzXhQj|B-~FlGXc!e$R312R
zV92|l){2W!_hmGgK0uqLlISrs12yxwMzd~=w5mDwe*zpO>PxtLKVE&<8kiyzdg#sD
z?>kg%s|p=e7}j@g!m(QmThNd?3pY*tmoX;261pkv3^=QpqbP>{z*7?_V#Vuwf^?Ga
z$C&sM_KI@iTx-O4kFsx;>?r&<Wk+#%7ShX@Pft#;6B_D)pp+hW0U`}|1%%DQ^-+Ks
zR<pB4BR8^CHj&g8Lg$olz_H~|>qC`t0gy-Nq1mp@&cdS?`2b<wG^1ryZ3O?Py?(y|
zm>80~74weI$uo{}0kCQ%jiY<#iLHe)M#}ciHVG;P^MH7JcWpa|bV`iKs7=L^5K5&d
z*;N3s5`NhRCNxPkKpC)**?^P|ONEuaZDNn9hiLP-1nkpfvdRMZ2KPNj&>BdnDD4mw
z4H?m2@3z)x;9dq8W({!8OC~>*iJRM%cjm~ESk<}|Sgf+@b_AP5Ep<~nbHG9|{8QFS
znGkmghsVP5qk%z@Gil>&{A&NbCY1lB!<P+1fB^=xK&TElT{P0MlmRS6eW_lYVYvgl
zE~IhQTX}^t&^iyt5Ep6->|o_o3rZWI$cY}3CC2=%6ilZ`hmiEB^0EUdxj42sjruib
z(4h~9j}Gpx7QXr6!?uEy^^#|tx5<Kg@wx9MUa^*V8ZRre!__B&5Zya8BC_xL?mI`Y
zVRW2N7^Yo2!@*{>CUGMZw?5%K>fe}u*G>!ipL>E6IcG2K*ww1Eyi|KUn@0$5BJdBN
zt|6-0b@mIN(hdUkv(bnpCNC}PH%z}39VnW>>kHMa>?Rn6Tin_Ln<S=b4v^1&;hUd4
zdDvI~n|8V7f%RPO*qF%G=VW523{S8Gg-goq{Gx3T=R{9|m>n`H_;Y#U_7f67gbq-&
z**DhBaDCFh0nD<pX=m`OhpGqVuBAW0=-pW(Kl)ly+oACv8wyr(02=v@l|J4mtKym4
zl>P%ao1QoMQa&}fIpYg*zHif}vdCEwfax=5R-`T7yYiED54G@37m?N1!xgv5Fb#{i
z)C~#{j_*Z`L%UY>rzfq<jVwvCZm-&O=#WYo8uMUe95t%sR(DH3UB)bg@OL;Uvgl&V
z-rDHtrNM!+ILzidY?7(A>>CFzaXqPhNU$y8*+8}p%lq)nci#naJRzIEQJDLz@<iI3
zzjQnX-wzu>gCNqW?bU_clvFHWgxd5jp_h5pc~i1jnd!EHhx^so@dIe@vWb(~8Cot&
z=hAq_lQ#%*WzprZY%#uz?)k?=m0McQ_12#|@PGpKZ$CVuR!-(i#>;Ls*ryZqBmQ}u
z9EB_<n+SjX-^5ggrjnIu2Ov~7L!0Hl@6ZNq#LrxKm~u{fWJt}+O-~1vId^=S{xv;G
zUL6$)MyJ3$5ILh2?y2DF0@d1ARW9(G)=u!nW>f@DV@_my?%nX|ZI=1Qw@U;{rV%b4
z29gYeg`FlN{}mdDh-(C16*4BF`u+JFV#iXVN)ov4S@_)xgi%VoK8Lu^kl9({IWJ?8
z2q=!D%UJvy=@{0t6-9W@b<7qUMFuuGjVYCIR!=%{?IAFw3^kLvy`)Qe@FGx}1ajpU
z_?$w@g=XTkd6V%Lcpb^Yo#9iE-3dWNW<SA=W*J<dO(M}zH=!iBa(cbv>pKj(nkkn`
zAGMnEcC~9aLZnEo*xEL5*NYIt=<$YM|FSsp5)V=8+xAN1UR`N~UywpkWia7Px;{3h
zLAW8$V}MA_3|GR39(_#33Xl{ExJSmN$fW~0$&1sETuAiJ!heICF(o$oNwh3dfy|+9
zvE>&^VkvAB`6Drx;GBZnvXI9!+C19;2v^3g4jEwm$fr-g#+Z9i6Ahxb%oBJfYa#?V
z-`;`bAo862XB|I7inW0fTKvbP&Bak<7;f<!ofz>LLhy>+Q@<=aXM&}PC0mH)$9OO^
zX(kwC;QeCcsax`g4aneAI^818a7G=B^dO17Cm_8XZ1`x%@J$<<y~V(W^ALDcSZBW+
zwCE=zwRHv5)<Y~h2Br<5Qd$>nyM$dz^fN9L+8Rt^naRvzTAPz?h&#T}{>jz!*+eg>
z;e!w7)SbSUpu5Q5KCQ6pok?#?tv9}!>8+hja$AyD0T|b<>ghpu;q?U+{Dwyo)ItkC
zcJrLLC`f+%aP!rSA$u)uH;3F)ef(2-$H3Pk-;B<=sz*eDyt%a(&D+fms<)x(q4iqO
zqTl4-ENXVw3TVhM1l6t2Zi+apkhWUljTp8hRyAwB+W&ovSKw3O`?qjnY^IKBGwF0!
zxKAvK38^*QVODfti)nv;@}Skg8O%5JjqIn2JZ)>ITNmL(YGh3S7<9r4<EHWM<UXB(
z>*mjM^Myk?uu5ol0sEYx`lMNZWzkug+VNhs7CBp~1FW$&>wmxUhV2eL+$E@#D6$By
zN$cgvRb78t^fmFbbM?nP!lxR>GzejY1(n`)S4rji4f$*?VjUANMZrc`i(l0p>HEnS
zav8Zn(u!qF?h%XOE>GdbHch_tEw^c4(FeU(LmN7@m>>;RY%aDdf7^P}<E<vN@1)FZ
zvwqZ-4WE@2X1{g#y2U}=-N}2$eE!29A)HM*T|=j}b5%!(v9DTw=;^fWKQgr|ZXtm9
zIez#x9Y_(a9j0IT>2bdnEk5mPUiX*2_mmcMul)a+Q;S&nKeLIObUdT~$p!d-#5SJn
zoVJ8|*0wym^x;t6X47ld!_Zn^b*r7@{sEPF51TPR!krl|ZS9T%@17$eET~(qxSkeU
z7mA4!PJ>w~pM~gr`(;m2p~%1gi?5cRKRQ!v;V7LOwM3XcSrQ2`{fQ5S(II{BR-=Sr
z-M|)cJLm8l=2k8WG4x-RPZ>f4+}S<@@~<^ZuBDi5OnQB1$Up5^Mk~bN{PC54fA#Ns
zKpdG^Q#M}n;>gD4j{|zQ{6|c;sScCA$(?z#RhL_rzW@IDo~TpeUJ<#qZ&8PiZNBfC
z(xuC9fAw9yWYUG-=9-+D-sP`fuUv0&>(4_jF0SEkA0EWt`0(+=(Z;FO2KxEl0musP
zpKKV_!sB1+`u|a1buB7qC%0I9hmN~2Ts@1-7#Wmvje;7oYx=nwI^WL$(lU=Lvx%;D
zp^A82K12TE7RLdO%%*v7-+pwuByC8sy?k<)I5DtwxWX=KDu#0Uupyl$%$YuYHw>M^
zN=yE9Z)?Q{G<!!4{^V{reQA#J(Z23tA+bQAiL#hN5erASENKJ`l!BxtUUYBI&wn)#
z#1uumbBZgu9i;305XDdXe*S4i-kQC=$J;L8@O4g^$R*F?%q3LK7W^IlL+RnzHo3~^
zwOUMVr-(&kMHxoR&8U1VIwj~w6%KLo)82}ocKG;ame{pkSbhJHk-}Buyh7mkY8&-8
zjMrO%&YSItLtED=PM|G)=%16B)h=}<Uz$%+8l&*qu$-*8h$-2cxo{VZ=+IOzb>&^|
z{NwYxh<_y2eAP^21BEq(y5hqqLu*?^E!ZYgm-H|h7U!@>7+0LyqA;PD+ah~hrL<Cn
zHSzZ*xvv2diCuLOn%sPV>O|%=xo3dz8IomGzS%Fo((Umpemw9La#Z`wP{7UpVSKx_
zLc=)eE3Nl8zg7;LJ$?H7H?!oo2Pm$im3VwIy2ar|O0TgO5LFE<`}}FQ4H$dE9bLAh
z)Wdsm7vAg;Rod2WRizpFfYoMUJhw@=6uQ#)+ezcanRRGthkyKI9?#c#9B25}H-9Mp
z#53G#+9Ckh1*`;>$C#>|`q)R|sqr%ej0M12IyMoS#POSspWk{0J$fi(<%${ftb#?~
ztI6k5{`}LwRM)LVBR-GpZpfc$e3H%`6mqZVcTD=x!=|=(%IgX;p%&$z{XC`$tb)mX
zVLhb>)#Qba>revCt=dt4F_)4~8ZY`;EYEK)VduG?GsQNw5$|EKVrXqYjn~;G7Ls|p
zD#H7T-$rrFOi6nwdYzw*=BNA{{0{nqAf~Aec|T13{Jq8hIt~B*6#?bH<8CZHA2)!T
ze@TptDPfMpYow*6JwWUTUI`}{(ft|d$3jAc@Am=;z1W>gsR{RD$0#fQwDZrhlD6vN
zwjo6=wfq7Kb>ZfRWmDjb*{wEE&kI(28+#A_;L6CMqI~`8G(2yqDJ4&08(d&*Hu*5Q
zxW!<F!_N2PDAryO?PTr{W*K#k8vN{g3wgImljssZgPc_jeEjREBT>Kdqnk<+J32lU
z;v?OJI>(;!`^yzq_r$t5Wj1neLc@G2Lay!-1j0aqg+swfJ8Q()p7nSs%|)8;zK-zN
zBCz#H+{&ijr4oi>5Hc*19r)FoFE^>MUf={7LT0GKY-A=7t%uSjLtf&qa?^yWPj}6J
zR4VKHB9*u}a0rffb>&pyGEqQ^(X&uaO*6)bYKZaesgV0Ld*1DW*jIoxA|)XC5YZ4&
z`fA|lzbMGy0>9%>H9@Qtph!vQ{Y(8WRk{cV1`N=q3da^Y8T$Ka9vp^mzf(~@hiH5Q
z3ujE7ba`-Q?!BgGkVJkH3Z!?Q`=4F@^8H^l1$!P+xL%y@{CYENx?LY8K?yM|lJVXN
z9BwBQx`D3RI=Z1*83C(t4YKh_rMnc+&yI~UIm9sZsh$b%XmrIl(W_lOeIyNu4NgsU
z1#|A|>gtXaIj?8VDngfFb&Qp}lUKX`2@Ij`78xF2>@a?X<2h(<^va#zblai$mixZ2
z9}Eudmuo95ENtI^c4H(bM%jE<o@Rn*KOhm<D?RbXyCl3cG{f61Z|3&mVXk%iEExb3
zmf#An)!PkM<25qBF}HThX#F1zc0`yC5_gZ&oiD!${H9y2oB&^~fc+-pOnRsM1HOQ}
zvX*yE7l2H}P9`w%TRF5`1(I`PZDLQ)1nd5AfoUjQ6oc+kYI}ch+wqqw=@p3<d+@!g
z%pcav4W9Jomr4qct+RYIm(a*=Uz|y^Y)->*?iQT#(xd8Nuf>3H^QlBN%T^(45^Co8
zQgEORkwH*fdjgH`3aOZN%5)ivlIoNd-G>g*bKvn;SLP7Sdk4h<hL2i#`}<usLNg-*
zUz~M(B)a*t!q5PT)8BbN#13Al47+Z5^6J|QtbEm>Lm!t#LL=}B9ZwA*&e9kU%dA=<
zOVBl3h(tjM5hA3QE<KF1NKMZ(p{8H8nAGXb;WjGm%5F^+&y<;#1GA&QOFcT0VWk0<
z)7bi1q4H>euijoMJ=!ZaLzOO|<x5*Jq|H!|`YyE#OEh6(bDgdq(AG3^7TV{|xk5Ed
zvSlK|)$wIKs#zJ=O*2x?mG)+i!R@<udkLfT12nx8DV3?A=M<J3^RKra`^A0Z2a7x1
z6z&@u2h?aXt`9S1|MU)RpdI7w76*2aR3or&Mgr~|xE@=APo<ld558ZGk}l0OjF~`5
zW5#A${GmT_qpRsMSs8N&P7Sm;Y|oiD*jvjn#pEQ4$uz@kp+=N4`Nbx{`H9$L7$Qru
z2D%wv8t=p7G6C?iO;wMI5I5q|p}O>qOx4yr=nRMYPW3-DI^B~+vGuc1otvZ_7K7kx
zR_f0K25f{vJE!3z*r%p){3DoSGC?b;T8PA3tQz6jyfln;ZFWf-BWat=5o9ZG1ZR>V
zmN3$-0pO7hJ+Hpstb58#*2;`GW<D&vQ$npW=^fI`FBR77_1SoI%`&$xb7KueWvvC1
zB-W2KWbP4I->fI$Bgd~|0K^&xHtp<&c4S*D%nGe@0OQz$OaUWG0(p|T@fulUwnefS
zTJ%@sPha`dlbTbkG|@#f=Nf&lq~8yY6*q*^F{Ch?mll8$`WHWXydeV^88%p^k1nG>
zMN}jM55;ncLOFn-dOW4WTT|k1pU>7C`KT-H!YAPGTow}`KGlJP(k$}YhHS*Z{;^4K
z?R8t!D@4AEq(k{ejDWm?hR%NSZJYo2DrW&ZBlA3#DW%lGEi4(zAe87~{zr5<2r^6&
z?0cu&PfV;{ts&JkVY(V(y_ZpB;l<|K<|bce?o|x=PZ#x;6D=UA|6SURB}1RwyEo8c
z``wF(3y)yl${|<G5V{x0ZUnh>FWZOcO*e1ZBJZ#voP6J!jl|`y<z#MTHEUDj(&3XQ
zFDvXq8WF0!z-1_^7;wQfGU)ttf;js?=IPFuk0}q!`aBfU-(;E6j=Lxa0I;XOvvd^8
zD&EA(&Rnj@#Xki}r<WTc1;dqWoNhMMep{)3$tWg$)U)SsZl%q*qTx(FM@Bsz5}^<k
z3~AG@T{7TAGNXf|Ohr4!iq9LrYM(rgmOVU0tsOUr3_(bP;*<>G)^hP!EiFTsIyQ^f
zv1<H_83!VxOt)x00paw@{xfUlwU7`YQmbG35#s{$9*R?Eq$JU&=KhI27ho$2t-r+V
z3gWjm(O+h$;Vt$0dLw$pK#n-|OCfAvai_6vT;;Ci53$6vAd{gp|H*kH$RI5*bjN8B
zUul}rM|@jV5*C={RGH{s%r)46**QJjml29TNy^{iGJ12JD2sxtNKKT4`D>!}#d!n&
z+Om3a-Z-hTP#i&E4~QT0J32Bmb#?Vgvmp>ql_ns$L1eDm_|Cn06(6l@m?_m4Q_1N3
z7R!zO-#Kq^+6#M{GE!`^pn)*cC_o0lDV#S<YWKL^p_|yE#YTm5Udwb+4VjzfSv<@z
za)I=U6)qhbyr+2A{9V#gn?*s{C#vgHE=bZ=MFxe-&hPk9QEPs4*4ta@0M6qAblTd6
zE;B_*9K_7ukL%Y?<yemrM~q!IFGJaE9Y623b%0A7umk$yyJ6dqnQ0*72*h_|?OOj1
zYZq7j(|3biqm;6W>)kp@QBHpICkMm^D4oJK5HucpsMC`gp<Dq+iKT|Pd<YkH-bD05
zsoPt%c>&G9-M4f1dvw^V@!b&j3K4i*Xs1{sJ<Funar$-aSbY)}g4A<>@_%Ndn)dT0
zsoD1RccKXt>kWR#NX3=XuqVP*;c90g{*}*S7E`@QaB<FV5iUX@6ecQTD&8hauomeb
zk!ww@fAO~ElrhqBV0UN<X<vly6mkoYl?bcq=$8bG8MZIbfI}3SAd^!O{t`M5k@jiz
zj2mdq16V%fx19qD8l(E~iLkA#nVL#C=HHc5U$QNua?k(ewyP+^_~~+!e}DDwdm!uJ
zbW_idD$O5viOwWghTE2i^py19B(jO1l7c&HYfGuIQ?6iI5B1{XnoV&(hfBzYi1u;g
z7Jd{(&+BB+fSvsHJ?0A~o3FCk)0gk_Z?M8Y>i@4_Ex${98&3EUeMY#Jmk&DKQJ6>h
zm2vZF_`NB=@Ri31y2eg9GJajuKTSS8w@#ikc%f4#?XOyCE9Ldz?6NpYXw_l@Nu7s2
zp#^=*Zk4rvz9~ZTBdjklzG1@cf{QwrGeNqK5Hls2{_&StPw7j4(pnhB^qWA~04HQB
z&vTHF#!-H6R^fc%fY*;G|Cq!@{xpwLQRWFqCQk_m>vIW5+mX0{)~9y=<Tkm_)y;#a
zD&-wtdGeKzM`lmEPN!aK=!qw=ISMt;@iDIxqbyo-8k@!w@1wK_2Oq%_jf695lsl$U
zUN!Oa-k&K&t(Th9++;__q4&j_TT0X36I~1_TSNiH=wcDOpdnmDg<Aj6xao}8uwmA_
z<rAkPo_#M~$3}}BrV)LHK7GoEVNRR}!q4AUu;&x5_-Fagd8E3C$B#%c3y$@UgyX@Z
zM=zjT$N&cMEfC+bFT(vI9#xZ%#@4>u$F<&e)Awp;%OkE`iofaRw2u2Q#`q14>juY}
zTlKg^(qO09Jo>dMG|dcNH6<}IyrzS}2T~6?YwMBcWVXH{gO#Dh7{c@#`t&1Zh=^m3
z96qdf$|a~fRu@@$<TBMWn?LH!nd2t9e%g*Rb6lHamX)H-3nGm)d3I*%1n<)gyZgQT
z_KNbGf#0-rO_+asUXwO(X45(`@1%(wn@6K(E{(b9kqHQK#Huv7YR8#8r0`D<{xk(@
z((5pFP1t$j8<tuxW8a0J>hzG+l((qwxA`2DGV`oZ<TXpO@w=3B9H#~;n9h+w^oIDu
zT6l(5cR{REU34U8(gF@Y$y~eERDIR?r`z-1k9EEK&nxSnm!I#I{Ry9{ZAQHYz-D4n
zsdsAqlxOd$R<Vt+bU7JMmUvMpuf>gC&aUt~N~*8!^+O!ESSTBgy&;T{`D52t`kVn@
z0b(*2WFVu&#HY6`x^j+8hOTF%2Kl3Jv~5I01d`mg;c5{r7JqHo`BP#JPjA6;K`;-A
zsIj)fonkeR;@YUUlrfTq#-tCpr0`==I66!Ke1=jZvm&gaF9>0g#Uc|4r86HM9^UfF
zYxWK2{rYVUR|6=ND4idKIl=K$+A-`DNQrb;?Zv2Kv}=<S*u5B4w9i<?fG#TiB&zNW
zA)mx{k}h-E;kuRsBX|3~(rt1(x7)AfcOXdzt4QQu8kX(brHe+E;?6$CcySU=CPAPz
z8HTAy6HC)Vv=Qz<6H4P3tr%HKs`#tOsP8Y-0l}kQEw639RFI!vL}1sBJyAy4yQRt{
z2sf__P$ANL+mXN-_3#_VR(oh??#6z&$R{$ZR%6qOsJXnFltW4h`!#MY_I7wP@Z;f$
zXpzp}oV@*^PN-SdMKCC#&@5opiN_AG*+zNV5ff5|w`HQ{e8~d29${%hteiu6nnVfy
zq-D<H@PTONpN%h%_9H12OE{S!#u$hNs1-(=rZ14DCBB|G+8CX(bo#JFg~cbN9IRP_
z!I{O(Otdt#MMUf*_Zg>`Ay1=!&M6ZE39tlFi;$I{hN&?KZ|N?oi(W7x^}N{VVJJKF
zz2o~GH2%f%P60{OuYM%_P$abjoQ;cdl1V}ZCDc0UJBXWxT%4rf02b7NDS#yx>2Q2p
z{RysKv3N+J5P<rA2D88Oz*I$;j0&}<q-2wOeHLJzj2os~L7dhV_62|x$mdxiE@$a_
zLW>N4i~i&my-qkZY^jo<z-3`FK3PK7xY7V(m>6o%P{{O&R}GFn134yba1<LLCbHbF
zI3=;H0vnMndH3D#JcHh39%3HGaX=Pr5#=j|h3|x{Uph9E^NMj?@FTT@9d*>T2j_LS
zJmNf6%k9wabGy$y>ZIMOd%X72&`l=OM-GAO8rw5{4c<N1kLz(r^iJ^wL7LqNz?Vw{
zdK&TrrC9wL36L8O@F@eVu8mEA)p#I|d4TO3I<#vyS_WG`?+$igY!oG&&%9ERMV^5F
zmU%pQ4hnB?vwa&MvcU{}{y2`>;K@K1fle-HW%e=bO#V84+$7z2g;NS$WfAE?g*3+<
zB4Si6pD#xwa!D?2R!xjd7ve4ByZ$~`YXOwjiyLSa=xW>m>$v3c4KBj0x{{JU>PGq9
zgoK_13EPH?Owl%p;*)shSAxd$fWTt6aSrIlSgbzF=QpEl_HRVn*4O;R@6)DfD1|_Q
zPM}ye=$Z7{s$z5f*!-gT&|g*&$i$8J;i}FasE>|6ZlJEMt^0@;X9P=TQvyP~>sK?^
zU*1Q_(_+T&zbybz>4%GywHYzaa-%n;;ISs31`Gn6tI9M}kqhl=SO5_CNo#5@)VZ+Z
zNl;x{zc2=EFs--h9MH-^>9@goJys67Yc;@ar2c~3f$@DjBIoZuSJKbznmJ6#Ikgu^
z`9H4S^n-zZ;{B{|Qstp=3Wk5$_nOFKs}ZpA?nm~~0YPIl#hBt`)4U-YeoMQN@hJHk
zRj+45)tz-BHFwX?f@_q^foNv>qj8D0el$(OY4i}6l}o0Qz~WS@aWp-sh8u)ue(oNM
z(2SGoSQnW^cjqnds-(2Zt`3{4lA47FZ<bKV$WqZtkL%xqrc{E?u$a|7M;W~yJkSy2
zB17d65aBsF`rOJ9NPOLMgiutm{m!I$0Mb!b=szwdl^V~L(OWY8%DjYF%P{#Zqlc@t
zjbD#`I4hR9W#xHxi!8SJys8J%+LW)y^T`DSYE`7hzV?4YqIxW)UDdeliyt`Op6^yL
zM=5XdynTLc|5*IsQb(Oel{W*enr2y8?pgU2inAhOfF8S4Dquk#2n~@=bqWV1$n*pI
z_wOG}*%lWJ4?N}a2>C4u%xL<0YHqnAu9;$T1lL>+55U^BY6;@_1&cK}=I$gr_TYgY
zYQL1*kgG8jU53WVluZ6f-b2R&-k4@%u1fP<j+QLtyQ#i5+>cQV>vYBmS1==r;UivN
z5`*4x0P544tSo{R8|=U&(=4yJ;2J+PzzU5Au0s81a16m%j{B#)hvQgA$8&Q`=OoQ$
zlp@8s4feZX*}H;;j$2m2!*wv-PyUw?)3rIGz(<eXOlks{6_skts8tDV-5c)RzyFLG
zuHq8r-q4*9bjaWGH#b5@(TMhC)7C>v3<}F}$;efgJIGg#H@(-oGDVPF;zd~`Xp*?=
zLzsHX!FZAJ2lJ$2=BA6NV3}FORAP{SLy)T6AsA6<gNuPL7=HB1(%0>nvoBE-H|f3C
zP{lmXCmdeJ3V2V;YJ1qGX_?2Ggsujr+4{x0Q++%(?_c~t{msC;d*>_t+(}1$^5AM$
zj@%=<&(}`-Fb=bZ@!Q|LyMB{JmHxuc_C;!Tce0EGG}vp>3Ff1P#BW;E%<c8BcD8@<
zVeiJAu;4l8#H=Cluaj7mCYYQ+8rFmB702ya#hnw@Im5N{gQYEI4JRL~QxD0&yiN6X
zRYpWD@kLE6d=7We%3;(8NdH&VV+5`Cw`o&+FBgzO+!|;eT#i-hZU0ZVhNxohut)GP
zU3X1!Yfx4>8lq1tbhU<upP!r05ssejzzZScQ!p2Z9y5n5KeOJzIWIlug1#p&*^@<L
zRoHm@&Yd2NCbrmE0(rl$@kCnQ*VCG^CscgoEPq+5s-clyur4q{_34vW4sZX`-jrX7
zH`$hRIiid4&*30XzufJe<u_N4IPN}YaALe1AVb_nwp}?$f2O57iHuTi+oAjDc|8|<
zzFxgHE?vjxsxLNf_NsN@KANel5kH2K3F@!+k^txl=`Fh${`ra2scFe$H*W$!DGSeS
zA7BbyHb9UZI)6jmbN4-oB;Bd!1T%>IYsG>ENASI0Z}rD*IXLd5HQ&!@P#7{4KL2@!
ztFq-Mp&es=JkDDT8BEQ#YXZ5kSso>az4{TOq_Kvnpn1V=LRyzGbvYGu;l;eRRarBj
zn}~J_>*=Hw5h+t?LkY7UzGx99HUf;AoB(TdaLS!OL$<qksbVN0qs3@DjJSLP%LTHD
zyk#y#QJDwqU&^^Xij43iC=!)}wAO^-?J(^jM`@etgzIbJ51K@pk<1TO`AYz6PT>mz
z(w?gq!`D;4{jte|u|+}ylx#%DdCM6t*wzz^`HEv3$DTynpI<G8>PKC~@w!q2_*%8`
z+m<T_S>2u=^po4Q;DLAM{cyGO)gzUQ_m2DZ3M#j)M7U*~d3H;KVJFy6#0z0Pgn*&$
z^Wc+EUOh62;Y3UO)QMf;Mz}FhzI!pKY1JaKIo|_%-F<JUf$KkK=X%wqXR8cT%g%OZ
zDB)>x#)24S?uux9o*Vcu<u_XIQOLDxGRr#h#bbT8{c1aVsL$;3qU5U#Kt09H%wOh0
z#vMZlYgOh-@83VCfw_e;TT|jaPr@I0ofwUYN?a5v_0!+V10ip(UOxe`daAgk0cMMd
z7bS}Wc?Z8fAZZLL4!zhlq|76ntE^BUU$?JZTV0m5Ow1cV<3;(c#cmRR5A-J`p^k;;
zJT)D3vzbKp=0@yLMsxr9=buiVL52^9x5Ze-7`XSwk_nVto=q2qe6OuLd}M?2-?1Bq
z^}b%)!FK=+gN>E->5aE9?mnl~&ttROxCaLF_WJc-Ifx@$>XBFd(3~*#)bc_$6y{Bp
z=_h+%5&lTKb~a1JMY@CL`$^fw2vS7uFw}k3LQ=*j4ff>G_2sV3#fL3h=Z4tcN!A|y
z#9Q5ulRo^#W0pQW08_W$8}U80Gf_T7fo6~BH!o&PLqFl%KKxkuTlbM|r-**=;i~ZX
zO}<#V9WpsJzatTAOhaiGF-3_D8JQ_O-{V2O#@lX+4N(gUr%@#A`lL{LepZe{(VwQM
zJ<(icchaw1X3AC$Sr%!1!K6`K>Z??)@nGM<CcYX~MzF4KXzV9wM{*NO2M?+a>Y5oV
zc&>F#r}rH=fC)!#v_>~arAEKpa<z_^eu3MOe<SSB52-bWy}XK!x-=eG{9we&K?PEO
zN1W-bJ>&NL)V?0qrYd#$INhSl){vy#!RKj&CT_HI>6FR%o1v*7vNsZoY2Sxurd({Q
z35&~X{Djc%Jj@XZ8m+B+UeVg|VTt4OX38YP4F#o*?~0ZSc|c4QB3$p+j66ndAC;Vv
z5!9&tp|y5^SGV8U=|)>#w~DIlPNcuvNW9m~IWROhf8e{#q1lH|m?H_!$=?~Juhyph
zxlcLA#ZC^s{|#ypChIhHicGoDP}xOYU470I+ltX4rB*lOj1!$b__Ix2#Ttb<0gRW=
z<^(9}Y*VG>B}+W<8Yz~zppO~bt+tDH&LDHcrSlC88r<)}Z-{wbe#3VP(6WAg^;p)`
z?sM(On@&+$6LkJIoBRmsu8Ezrmwb%?`aI8Br#dLBN8+3gHv-ttLsMgFG-~m2liS2u
z<u5aB%Bx9Z{7Gkw3Ga-mj1in3jm(9cy80ig*EPaASkDk;A?JwwHyd*xDhZm2WAb1i
z8@8*Dj!_Y>ErQ6pb>bw5iK#t>P5)rz7zIaF%U5w$+m|*v|6_le$vckL3B9bS6m83O
zMj)eb4?b;Y{nnp@c24aI?OTZyq7^>x-x3~{9uL~Lckf6;PfCHx^pgAIp8M)HMb?bp
zY-*3TX;wx<gG;N&LseyFHQL2MrgUpFb6hwP(o!SdT;%EdmzEd?9U>&VG}h;3<j}}2
zN-5v9{Yq_BplW7a)b5GH4o7%}xh*%&(4}sSc^(``f-r4xo~B1%NwD7x+_QM^;-Y4Q
zUOL=U+-X``TtBS7vir`5HdztD2{&pVzt10if7`F{w;`}4*0%W-hiu#}UI3IxaViFc
z96T>ri{9m)v8qnhOH{^LOcEn;+=|X18{5FCC_1ml%*vYI$HewU$x>ozP&<Dzw@ow6
z%*>1;>Z`UmM%(0X3wShPnQqOG`Upi{Iy-G4Crk88&wBlC-s_-GdcuWBvZ27KE33<$
zMcOhuHR8vr`ifstn_QwJL(21D5SIMXogi05QGYx%gR(Sh4Y*Y3k`yLvX?=PhKlE|v
zD{<vZWe(1@r}^_b8#J7(N%`3z)#v`34U_vm{Xg892S48Vnfr9D{SO9TJudNgsR=(k
zPIab~w#+&fKYEJg(1u+#r2m{~$C*Yf^}9xNZ0wayP~zw;i9krJtl!wNsplN{*sUEp
zr)GU{x?ke+b!_|Yz$7jw9M>CsfO4~e3~xg3sNv%@4x}GVy5HuHz<;bXj==O%Dag)R
z-QCXyZ<RWiu`e5ZME};_{iT;-(ZCt<{udmp6AvXC*e^Y%6&rokW}bDa-n(ea9)@ez
z`g3$2uO$t=_d2QU(x2prvYY&<gzMU0Jz}+^%a>=~qTunG?^1nMmX_W<66v_qbOnd(
zVNRh-Sg~y+trFKIPMZKH?dL!%!pw-O>b}_c{^Q5u4v{#|o!Ud;BS6<id<3L8K*Tsk
zyG$vo+s#hq<=Xnj%6go;C)6XCeR6(#w`5O$5_C)TZSO9Sg^tb{Nu$uAYJFJ4X6qME
zkM^^3o3M+jKF;vs+lOk?KQ-!Z3u$;ZF0%PAKO23w`tN#byI0s&t?u(S6+@d7sbts#
z-zPOu+nQS?Or14rR;<oXZ5|j)4p*!w9KS5{2S47Pzhl?92c!FkxAY$G`X^QYMz-CI
zHu}ELiyn9CZ>Jkr?MasTj&eGcEJ+kVB6)$863Ke5ECRE-tL*sb^#z~U>r|2_k_<3^
zI4`ULq@-BAu=6elzI*r1DL}>1;liI`wSK@ed5umoJwv;5%adudH9|r{dN=*_VW&_^
z07mOPf)Ur2HpKGO8-d2>e66OfJw9+@V18J?9d6nBwB%Y>>E4pKy1TqTtc#0j&Of~9
z;^wowV7(#R+0Y?fe}$E``C+{3W8eQW|C><&IDlJ^9Z$kC6ZPsO=f@cL=G*W}vx=7v
zaidmHxq4&>#o*MRKjR`zx_tz14f^R;msf%hY|i2!Opk5?k6sy;ot(D8&TLi$UVEO!
z-yVV%&#TWN5=fm?;o%EK=yGZCZ2gHdekNpc&PM9ER%tf2C1Q%uwY(rRGt<fa;f+It
zzvDq_j&Jsl6{}oZ6~EH5d43%0frKspV&<+CTxleuo}DUI4Q7dVPLLC=t(|u5&~+Kl
zqq5HT@L@k5g0A@dj`i1H4!j;$7zrionV-$tjE6aoVs(M47a@0QV?QQE=J%XP<-n@>
zSaOOyD&2dJ4E~X)cRe=w&G#!WY*#J&eE9w#hi_0{j|vdL)Pt{DTYzL*4O%|B?G&-d
z&Mt<mnKZ!Oz`1oU8$e*wfJQtIObBX2AAQtKRUXpjh+c(?!@HkOqy}mBOz##n;fyGw
zJk5<Fp*p)KRMxoAd@#3~xEGAeepK+Jx%5Jlbmr0`&8z$%<$(}ljS7p4`W2_jr1Qci
zpCd=!4j&j|XMKIRs1D02M84X;f4Ft?E^yL~;F`3BBQQrh188_WCGqj&F)_ZA=`?G(
zsMNF-eSQ`|;Og2<9k90`{+tM{tSwlRU(_((uoUbqwBgA=lzNfE&OVd}L@nOPh4YM|
z1>8_?6F^BP&Sxghfp<FHo}ax?2}qQ+GDy~n(oY^0!?;UVer}mb=n4{(!#Tyyl;mTU
zWFgE=3h**;OnQl7v_@$&HNUNDlxwNq#>U3)b5(<4W}NpsF}D7BvTmdpC+P&#(uIGK
z+s}!pNaEy2PLUS_WB3Nv3#;7oyMF)3?P8i^&y&L#^_`s?pkru?83eGEmg6GYOEOeM
z<X*Vl6uy5QuLw{G?UMSP6f#HRGY3~`J`^UcfN~xH3<<q%1|l^*+!Z-?WL^hHr?}1v
zY`_N&!qWTv-0h}zVoU4qxn0<qcA1vD1W_Pa6)Pcb)qFCa>(JGrO#AljTk>ta{6>an
zZ!&>nj~!VNV{96#3UjMs%+jy()F}UwOqRXv29XeuDr2yy`(w3FjIX3~lK#n5d^_V$
z8DBFJ#LMu`2+o<!mlPQeWlSSw%!A2oRm2(Pr~B8%qHM!@PqDIHI|uH!-D0t_9aF70
zf4)J_0!*^Uls1tZM``Zt->+}qzZb_kMVOpmr}l(>W?vm2AAf+((6JQV*|SGeWggX!
z<G&M7RL?HPF6Zg@9XdQ28_ZGS6`F>f`0Ok$?O4i?-NgEy;C6rg#h9z>W?)y(Vx1jN
z*f+AIY>VsbLz+_Kn%3wsGTg`NWjW>S0f~LSw*>O^iW^qr$A$I+$G&o*nI5Z#lZyV^
z{QRC`8i3wMq`Xl!*)9#HC+a#dOX^&#mUNcV8nr~49abM<`&i$T4uSfn@xsH4Q!~s`
zp%YiSd1BBJ7MPbCBn|WQGCsP0DLD4J^r2XEn4LiQ_$;uzn7P@bP8yRxm_!LI7FziN
z$)>5xfNp2^E2%%<SP30;c<*&mu2+kYvmYm<nQ^bNW>x9lG3R)j_F+~Ekn)ZuAb;rb
ze<|}pu-a$P>C0%`-)->XSJsavx%b<BE_2G@qZ8%v=~4;_-j+OhaNc4|H@`8PUcRy(
zufl}j0jsRO)!zz<Wgqo|C<(=$R9LX$pbjq-+i6_~QH#nQKA|@OF99L(nD~<LL0F_@
zVyla74YJCoQm|uT4z?;5QUp;?5zP`LpjSM}vnTjH&Hu*;lHeCK8RYW?`>c$k7bxW&
z8F0$ij6;X)I9If$*qI)MFhzaOIi@QV6LQNNyk2xD?%nS*s*@$sCU$y)G+WA_XHc}n
zN}r7<b`z5AV;SZb*H~%}N4EOoJ9i#D8}7u3AJ(m1D-&A~u_ue2D|v*Dxb#9D64D7b
zR$RozWfp`lSiCEN!QVF|>Gv;$Kp*1R?>+~l4B6Q+a@UAe3E@umJNgZasXA0>@~%vb
zCLOBYy*?V-x`;NHK8?CYt#-cW(P`0DG<ucv5D+AoOWtC#D4S2SAu6XD3QZww((3sA
zd)+XYmk)E_@g(<=^86yP*=wBT8EN~_H<8YmIG9M*4L6G6rcd)b<%5381SxUq30U>{
zDE&{c+md<n1R%<Mb@=-6TRjZ)RqypMfFLx64#(Tg2@#+W(;y8csUuO%4kxiQ_$;^w
zl`-{D*C_fTxc96dGrrCJE_Axz4^%Uz%e`w!?%y9TDe`)6DqOQwf?}9P{hjXKHhwtA
z>KdzWo@cGJwZ1sqT@%bWlfmJ(1!I9?JsSqz`%621_<kym$m@yTrS3pxBSp~ZKJhRo
z`6_CjX=Y|Y@s_PC&-<<TWS@VNP&(UvdBl9M{rzP0nG&fB$L(+i6Qr-$1DI7gCYV-g
zcN)D$C*8-oYl7G$nN0)~tts-3xifQ}MUMpEh7yb@wVI5M{JQYLIkm8$uJ^v+Y2LoK
z=&cK2AiHGVPad0RvrESP(Z(NhLqCAVX=3~r??8RUn2fVxlkoh<)$%j&PxYAl#(<~N
zN1gXegLt6RIn6XH*hpx50opIz?zqSOm?6ZuFJ|pYk*H~(gFtDy7+emn={G}Sy)?~1
zhBET%sNGa2OfV9&aOZCvSExn(mWPjcYt#0C2JKaBN-m1(me#kpi*qQ9jYtlaVw>a;
zRPC5k?UD~+pDmH5(s8qGF!^1YaNNKfcI8ImJt0m}+E(xL`o(f6_<{08R2x#Tg-Y97
ztQG6ZQhyvM^8z60=PeV12SCjF71O5b0v^dPv6}<=+(X0KK+NZkVt4D}`8YGu>nI7#
zpmJ3+NnpMl%_SVoBrvsGYH~)lUI^~0(8q%}dyQDNQ}prdr9q$pbl;z&oWIO*H=Q%l
zGB5`aWR3&^qP9?XO(;WzNe{)s2Evz26%h;&Kghyo(e!l0XGUoHa#x@ek7;BsX0p`m
zwi_CW!wTNxQ9-VFJu!jCN2$Y6Mga))4ag$`BSEJ~F>4;Y%zjU$b^$y=sBGeBNaFz4
zfqC4?1vEdc@yEG<+pW>@5FTIQlM~}*WR_ekkLEfFRP>X+kR@?E5<-ViUu9G<ZiX9g
zezRI1cZR9lzGUv|CJ%E6ZGsAa?7qo-5Ik-prw>!xDQmt|Z4~kX2URNgeIQs=F)%?Y
z5DMH9b-5H?To6g2iVD|)+p(i!LK`W(CE<rydYOLbMsPfF20#fayW5C;BEGZy#$~Gf
z3p{&;vVOp!<0Bkr9_X*dP04oXWqarLZ3*DgsJ{VU8ZQ%}$Jl%Jmv%DOqEMY<A;q<%
z<=(biu>=|>gSHNu*C7Tkem}$6!kDLZ!v%9BJae%!A+8>>rU3z7x|CkhT<jc#YDX-Y
z?%I%%Bj=9LrgJ}B>a5*S=Q#LXJIf=hrjGbr2z+uM_h9w}^<j^$B>cR0+r6Zuh;4ya
zRqXTAd!-l+65U-~z|P5{$xh4aAj243brTPzhnjA*a}0v-lcT4$=7vt^75n104IEhW
z)02+cuUoT9udY<O12?o|3z09>Zr%sl`YX+X4r<#czI6Q4i8g5!$V$VCF*2iNjQEMy
zd4yr9$UP%cw`dVNS!svr?sL(9jC<e=o9~sYIFth&7A*6D`U$h;W(T#&D<0mv_mp$W
z*St=8$MAVRu9%iG29|i#L-gjx_td^>@v*06XJATmj%gLB2;mRtO02J4dxl_<!Xs=P
z-mkMiJFcL?8+_VNDg|A&Ye6t~)<e0e^Qc8_Mmg_ttW%4dsk{9q4bJO*d!COFOZBnj
zdqxg@+`NuJTLp@lA(tc;9f`#()D58CU)6T*uPP`Tho7kjYQQ2UaaJbD=bNU~irdu3
ztzy=z321FTy(PeSuIS$6^X;cs>8BW(V=1#>eb=rsg*W~j`tm2Ob}c6T)X(kL?9aDF
zO-PO(ePwuPjvgGeD}z@K`uL-QT+IKw`Fy%Xt7GJ@OLwk#dlY|zddHbp`qcDmG0E~y
z&ZN$r8#=aFEZTU*zyB6_)PIKr>;J3wK>GD9o7?>c24=49;H5851p!kEGQPTU2R)M3
zCS$nIDVAWPaR*AnzDcElCox0utFQi|(4fnDHz!G++d>89SRTuNS<Bm1gtPnHFl4$6
zqsiEs5$7G>!w@*f*_}36l~%n=U0%0|iRe3THcQb%5)fS5tEgJJIRN9F*do+jeO@6d
z{Wa&>t5?f8MILqtf$lgk{l9)KH<i^k{*wy;%l-e>_5aUQFaP_h{_897f6~4Bm+G9+
z3^p8!o<%+kL&&)X(NwCfp*O%Eu5+%xOZmhq-5dD7bj_`0SUGgK$vuhzW>Zj4v|clq
z-wDiBsA{{3KaI@ZI=w#`czJRU)XJv@epV8n0k*6z%}6cC6(lY!bMwB>V)b7Yn&EJ$
zA-4)vRZ*rDtl|Yf>pWXzJ{8j&a?A<6Whl*rQWakCyh60_>K9`p|L(ndJxuzMjhOjG
zXATvVdFcnA)T5x>`dO?Tet}FQ6184*wz-AH6$-e#Y8U=wuf}JY_xn3*_Hw2!1UHY#
z9@gaQXr~;Ym(^_fNo!TtwIvNklWY2xwViqHuWM6nBhB~DJ@c2D&id15cXtgNe{|rU
zjxGbEriV_qz2Ea%yPZA$Sn{Z@@tLV|(D7n}41-jI)E>bFxgVT0^EC>bJc|tqyy_km
zI6YA(3Iu%7u?rOfHfx}1eMPnrpIg0$;>79DS#{AY5-g*Q!$?hq<FV9{pN9ST`I0O0
z_Xkph7);&U`NMzufoMMgV4|OQ^&Eh_V~d+Y-MtICM}Y7JB2J=~BGG-&rXN2I*HS1k
z?zJ?36fJ*Pk)K$+;~8(0>7#fitAE?qb;yEni`Rp1+{)X-_m@l&9etlR)Icx%XZLb1
z?)K|~>Eb@tO&f`=MT6qG9Gtr3__k#7sU16Z*jtX0t5#85^^zA@&oQh!-7iuhvp;B~
z&>&nJNbgg|VMQ6C443(UAH^l7MOZh*Um3l+KTGB)k!l)D=uX_eS^7ogp6S}$D7oY^
z8xi36DK6aTvjF`!mcX(Q1<{VjU1y<qJ!W6=vPRD_boUH%#tPDT9Qxm_8Eq7IeAoq;
zMrGon6iUh%;;xO}+@j(0(w!P);J+Abk__?ZUen|Uj$eL2vb&J?iZ|7?e<O_dN1w&^
zXE-xt6q<iABc-_)cQ+{^o}ctsh`TLrK2%D4x`FX&bc|0N@M9(3tJ_Y&>h#P)a#qj?
z<56-xJrw{1h$5=Eb=Se?@7=q%Nm*WG`O8R~KtQUzYTGH%*>9S?$J@;jC$3;GHrti6
z0MOh<ohD`30LEwi{p-&fDiqLM-PUu8%lr3N+=2hDyRqY=Y}}iXK?V_L?%kWdD}zTO
zjY6T7xim&1_!C<#l+H!)i-n>sKnrcdfz&)FrnToXy%n>D-(Mglq5C&b*GKb#@5DH~
z%*^bffM6PPnj3wg2l-QeuBjZza$3H|;0%dO%gJ=HW$+p`(*~e>NtWYwu~A#>Q>@;W
za|o6<R>XfxJvoz>Zj&<a)}rNSUHd^+u)se>bo3Wa|K4t)+%SbNMxRq+m<jo{wstt*
z?08eL<$O(;%}s-`bLY-yQ5tj^W@JE;qzAifG~6|e%x;~tm8LWJmx?>G)E0{rUMBl?
z`kgr=vMR-rZ+V+b?cAe>Nn|Vmd8ycS))TNq6^2xI9zF5iH8zd)Hgs4Ffg-EdU;3i|
z7vzRHEBYuA7ry>`hm1F*Kc|Xrn#%Sm{gUgO(nbon<mY051Dy@U@9Zz->)y_T;CGMX
zj?V&Qd&&R@O$T3+&qXv1gjFK%C1xVf#*dCQH_8gPoB~O7FV?b@MqPMxFi?tViHU<<
zO<7hT1wN1WPN~b+COHV!ZrWlR%cpwW(5}~n`Ef4JQ+`nBeUB&^up-_JS6)xx)hWfg
zUn_wwig!XcU`z|g{S>CdK|0L(qn%K?3&`CQh(K*`*C9poRiHXpzcixD)K+0KdjR6i
z_|j%b*a;hhWAZB;+qP+=Ci7w7dqsRwF0}`lYiX7<Um>=?za$Ug$dbMKj}E>%qYyT~
zvHY-TyUrnnESlG7pP|E0GqGx!?<mS@s>YSHbhjXSN141H<TWYQV1|WI6o)_d(JEfl
zEI4>lY-QfVbzem9s)*jxBFsOF-rrvO52AP2B=yt)z0JfV5lm#fsVc!=syApM*Q&k|
zi%oGE99msCoqm<TldPo7%MK7Uq~upJt7Jx0;a}Ry_aD4R_1?fr`lsrhW{8mt*$A^Q
zxq8<-R1F^oX%VTeYTc%dLPrcq;BM|15q~<>RoE|pCX_SUP@lQ*)$g9g$Ir9><!Q|j
z7Wrp|;X&Dv5D6{AXR}Ls2f4o4GoT6q-uTS5Fd8I!x9&Xfl%{>fdB7GCe5^jE={Nws
zIc&hh=lg%wWjD?Sp|(vG583P46!%rs=EcDMGAwfP<jEYYD@4@*!}U3gf$OMQFk+3=
z^zaTfM_k+fKg!-8l+KN|?+|n%4=2jaJ=eW`;+tcZMLv})CX<V6^V{W=9-ZJ^KkiZn
zG~%dg@2o$H{o$PFqOu8_2eZDPre>zFW<DS$zH1T#h?7xj2RQ9-BO^SYH?@bfjr3uy
z6XJ5k6OTL@S2N`ELiJu4eqZzRms?e1r&~x4XW~5%8}%p~KR>_TDPkvjvL=o(D@pYY
zwdG;U?xK5#6t5ugR-gPi0?c}RRJG(EOA8=fRTQl&4R+QMIe+V)*L6{dl+EBx7+6Hk
zCSqmf2`PJVIs7}j^cp5P%uW?arA%r{8dK7P!Ih#}wmTVeqS%wEie(kjUeiGAG#6hy
z1C>)hZj0vrwUDT^b?Xnz<FJ&L-wSI)oJ6Iwc>Mr9`1srjkHm_RO|E9>3bwXuZ0fMN
zr?zJ0(G4DXS^7|P-t@$a5vcVt{UAvBW-`A=)WY?zng`6SmBrHEa;z>fa4u2gjdd?{
z@A}Blug_uTk3#oOS31IQ9f1%N&=I10C*03kXK*}mvGAwF3o*%-48aD4%xZ_3r&n|r
zVU_gd=;Lf4>zh+M`&1bs&lY=nkh}D`c?5=h;TQn#2iFJlNJn*gJDZSa6U9b!^ft39
zO&^_Pvo<7Ml5{|_C@mk>NEWEK7(1|R{RmD%s`StJt|At}>p{g$Y9W?w>INz>Hfyu>
z(Ck(vAKVl^GJdchDqD@LglBDsX^RY7Cgu7?iqL)@!FBA(c3tl&_MgJTr{#3A4q~O&
zt_*08!)#&vS>NwA<w1Xv+fofWTXVu@=;&0g9w!K@kOAnjJgs^@TVJMtiI(QN=0U|B
z{j&XBu{ND^Sc#@?Sd=Db<->m{2Z~e4tXVQf>+(7u`)N5&l?Uj<<v(-23W9dTB&}2)
z{MOM5BZZC~1()n*wB^?8F(C4ykDiq(ViF;}Mw!tS@h{%Impy_$!jQwvv%k;>Nvk43
zeHG{`{R0_K#kba^PlG*JGBhqruJHR!lD2Nm5lSNWK_56hzBIP6W?9tOV{s-tYyUkS
zJogKLtnG^nM&UP=EF3F*A%{3GfkQ77b)ku5T-GB1fOoC+x_U<*#<u+;)7_eb?m$M9
zKI2GVaOU@a?8-2Tb1{VOXC0nMUN<{c$e=H;tWb8h!Z6l<?252M3?z|swtd*1-MjBy
z-|tM0*-}wP92_EX=KA`1vVj;i`V2<D4d)RM^BlNXbBf~_RiVKshImtv$dSw*4=Tqa
z2Rz0=*@{!I3dBtgH{nz+jNW4DS-h70p&_^*{+j2K(#*pYGp$n3`TG~qtvT+IV<u~2
zaT^85=DP&pihlZW<bjZ0-&v1`%fjrrAx6<JKa1Y+#HOA696Qb+uBnKYrQ}4g%+7+L
zUj(0FWjvP=fTRu%Yd*#Pz`@QQmMe{PsJ;j%BN$+EBXc(8_@zB}6pwoNM;y(%!t&%|
z@u(hlJTwA3nkZfC=qw5GSGr0i(C3s4a25Y!`v$QDhR-TaYqP@k;QR{9xZAXD<0z$M
z&;)B{uukbC#`0Kal}1W*ex;~)J03L443{w<fwUm$@!15H!8w+31S$sx7o>-*6ZX$r
zi(c}k-~TSlX41_xI-9aIWvJoK72$>xhYXobe>p`4lF(mM*+0A>4T`uZnD`*OGOo(o
z_4&NVCSw@HvOI^TQshVsUDokPS8F>(`k9!+`_Y_!N6%9kviI|!alX^$nSb~=W81dD
zbaUWS>=a8zp1)_&6?xTDua&tEHv_~?;<NJ-_jUGX&3_UHmf+>RCpax8@+EUv1EIgE
zZt9>k?i6EKH?Rc7<}?{Hbryw~IKaqrqphk+c1xoO;{Y9%q_V6iNihYyE1RAA^(`bj
ze!i|vOC+VD@gY+bc~CQGljx%@e6GXFvdK+o$4Ln1P)8=+@y-DRvyHfi$l}m$EO;v`
z{L;_&-mlJ%Yr{wSSN=$#Y|%yor+PZ>1=Y3@;({mt{>w1o|0Iy+-`Dd0{&Va4CfJRF
zL-w?*jRJ#)SS8*HN!8L+B3xXv6s#b{0uF}wbPRKD7=5c6dBn#+DdVEZ@QX*?n%Wi&
z;9wdN`jf}2H~;pNljg`xH@Fw7$a-RA*jCPz$zzr>fi<zfHDuPY;kR<8d_9;>xbpGX
zDe!JY#uT^f+djQI>SV?4_S<@?pIb#+={()x6p(<uW~tG&)iY$Up-cBpX8K}51%2f-
zr~O43{tC%xeEd<ImBq145@@*0=M)0%^R@?+)aL^xj(Hel6mt=73+?AHcPJHN6Rk(7
zv4Hxo%aElDH~@bI=9Eq(uW3RyEO!5KyIS|z^^@D*rTN4VgbY4nvI4$zp5wfAhYtFT
zBO|jPA|<eN73n#?S`<aQE8_Bx24ebje>Xz-?3^hrUx~j-5(mB75EXO`6>sn7UMT}F
zE%CV;y`-YxNid>j&xCm>)sk$-webmQq}3_2OE1ufx`@)iYrI{Mbf<;f0-J<R8dD38
zp=LfKqzq84`nwaGM04?Q`2NAQam=Lz92lqR;W>E8o9ic`qFjWpcHHBM*JvFX4^}bm
zuQW8!)iR`vBShxEGsoyO_e{L_WWozruuLYAV?%7XgjztHFCqbNQr6h2get!2AmlI%
z=1eHDB_pUV33-B`a@wy;ub(fpJUGJstq{Uw@G!~xI)@*`Wl95@1ZFhIfdU~ToE;D+
z#t}4>myr?<y`hVpK!(PxRyoJ%awiTu2-B|4ar#z0>bRaz!u<1S&m>bCd*;<pF_wgO
z$gBc(zDQ0c<l(HDi%4c?lP4!3R||RJ3VBKrRciEpl+>e`R~7XTmzTn$H6OyDFr7zj
zvEk{dZ7xl9>%xlUIyr@&qt0_0H@UWx_LUCHwNSEqGSucgsmBHYjh*d#3;)H0+S}}j
z*XD(szv70(eTmGxOT-ur0ZUFlY3J+MKydl2%%^nV6x!mSM|LrHTgz6_&+lTRVa>eP
z;OjfQ^F+4Cm_36DAq8zHE|W+gVp5Fw7}X@=YnH}RNrn9|Vz^y((L!-1r41Xp#D)7U
zTw1!yC&=~#$qktGJsDm8JhGpKfV5SnOm7$Ai?9yUw47R~e{b2@>^sRn8Pal^9A?V`
zSMV5-b{Tjbn68vJfi&n;TtJT#S)cGK^)V_Qz(PYf{C=t9y&f`jfoEH1)0WKjGE)OL
z28V5|Y3xC#HnkzLx6GnJIAd8~3@jzHO(olzw!iynGPR3aKD?pm*2APph9)aAD<MU)
z83Iz7ZK($l`DWtVmdo4So)>Pop9$K<sliX?i+)!RRjOgS$G8VtWJ{BTp;BtpB(kJr
zCKLraC;DhDlc|&X8GqUcDNE*Q71mSEn;<F>V=s9NvM4dmm#l?JAL!GM=8eEAEHiJ?
zLql$|9(l&dA;!^LMyz`sf`HUSMAy%acQJth#*R4pOrA2Ocgm1OmySuUB_+h#I&mRj
zJVMfd+*SvzI%&VQy&^4n@*n?V$EtUyCA?6&VaKpr{Q0Q5qB79H7*o|+rw_0^B8`Zx
zXR)xzEZc0X>Gb#Tw(TJH6_@_)C2R|_uEerP5z+G~y@s#YE(UoJJ!V1n`PCr&3MpLx
zR1L)-D@cfdgr^h=|A2Gn3Nf<N_dij#lTp}JE+b{sB1m6pjo5DlS6PmyWV&Z<?gmLr
z<S=G8)R<<3nN<JAYAC_<VfqB`?{u^D5B7FvQftbVLim9%VAPCeGSOw@I3r|*dKBfU
z^*tD`Sm->{5jq_^;sOEa!O8JqqT%AjWi;yu({drUe)+ie_sQGD-HU=qr&a6DbGhq6
zxTuB{rJ8z|$*FumNNIBrdw9`!2%Nm5b)k^x;+h-F2j?svH()mwuGA}xR_rAur@RL2
zAA##YDlT%CcriC2$T-gDoC`hyzHA0bgT;V^;Kw*lC%Ha}Vibq#A={<HAM{8VQdct$
zJrqEcp6nY)WW<1(auyK8`26y`e*<ZHP=QYXAsVdhQCTGXmXBwj2#!h?I>BVTA5}-O
zY-5x-GQx+87YY%zc(PE4=r<MezpU9Pr{~RpiK3JhBQ7R7E3S%>82pY2i%*z==d*a1
zcQRx|jwgByU4|MghA6`;ACq1*{9ieqM1|~`^H*&Nj25xy$g<%Uz936p=s&ARKq*O$
z9y7=1afMmR@vvnW;-hh5jc^q8-5a5hq#I7a<u6#X?t^U=bfz?=4Dc}-gjixb9;8o{
zkFzIRK|B<ZqJBb;>ATN)!-LsB?m?`#vQWXdDe263j!DO{!a;ixt8L{z=iivXIybk{
z*}9Bu{^6)4)FpkAEH@FBWx!SXSj!{5+NxXtd4{Tc`}XaR7dLvAX2)c#oyAVH3U37V
z+Z2k2u^LL>1Kh6vt-UjiYWm#PxV3h6Xiu$`ia1v-iZTz1R%=BCL6CV6X@nF66ofDa
zdi1o`ieQyVf<%i11qC4@vuIIh8I>@Fv6wPMAfSW<gTQ?@c+c%UXRW)|UH9wtldD)D
z|6#v-Kl|CgZ4jDyAp&B?XQ<gt%a3G6!O6LpepadMD=fhQcOtL897|#^9U7w5IYHs5
z;hK20a49icJ_nIdeD60~6xdxa7FUI$B;JkS>-a1aD9SR_mgf)|1hQ-qUuSM9`G7xK
zy3}i$D<Dfv3a&s1Jt<)EMS2Ad7;F0$>Q8x!d>9c~2`pN_bhh(2^t-{x9N;d!b4D<>
zU+6zTl*2t(MBQo)5{5}XChzjKl4vR!Z`)l5#!r!vI%IUU7>A~*>D7xtMe+D*xR70b
z8~SkvO3-O80GG60tPFo<L9^2YfDx3a=f;9s_8EAO8Za(7N>&w0pEXqawLx%@2TdWg
zX(LE^DWH^P`t&_sk&OSsV>n=t(P1JHGPhLTh8rGATmd}wr^_u7S|<#5i2z`^qa?tQ
z&aXcN$D?P<f8z=;z!iu)IKvgNo2?(~TQ(5j0hPMVgNgTOeo0*9*6{&8&mW18RVv$I
zV<c@q5ip2Ev`x&J)3lM{tNVb4$wgq9_`UqJ6j8I+&qQf_3}}RNq=mH^;io1}npeB;
zuq$Ktc-ZCprc`<<p7J4H-Z2topathc!|1Xzn<A4rID$O?#8Y)ke&D99Is?tI`K{H9
z|HM@+PkA2r3j!LeT!{Il+!mjO@sLGYowlJ+FZ6O<v$+uJz!SJpqartnhqjN=&NlbQ
zffxB#1|EIt?3Z7Q?uI_wnGDy60)}DEm^IzAV9}djgSR&jiRqRLrt)xWr+nW>d03UE
zIZ6wi!1m7kJP&<382;RINfHcsdD+$J%Gx|K@^PLo++h1Sk2Po?+u{g+(|!)!)pni7
zA{Y|on4VI>yfj09W?q{Aw;!{jH9Phl*Edi>bo%qQ1c7dUKcY;1%H*wuMlUrVz!_%u
zA;`W6G^&7_6BY&K2DOTi){*s>eEkSW8PdV+;{%%1w)Y+dOlsC(XNfcHxX;=PbIH&N
z0Znzq3OWp}@JYQC@!CgY;kpv#$QSp`U$*Tgq~O$&`ABRbtc5AWYrz?(?i8h)=ta>1
zNTUoD7uQC$PK+t|-bkN-Tgapm9%D-8m}&wQkeD&WSeHyuS8=xl3*ZpR^>fRH(nWo=
zpbSKn^I;o<5xyy#hN)8w3nb;7NMCc%nc=K|mX(!7Eod4XZ5$tt!FOzNg7C%89DRb6
zrcka$R5GCNiW<!@sDM=XI=UsnZ7{d$lB9&BzfPn>h8_Sr@{1;5H&gB~!!4T_;ZD&6
z-oW#Cx*RZft~<;V(rJO|>f41GILl~a9p7gZ!|p^U#*ucOm@g<mkl~quKgBE0P=zE$
z+nk#>^kXX$`Kx_6L?YPAu96LI5D7RW>7bTm-l)be0SG}D5}5)x7h_xIE(CG{4*+mp
zwx-kbz(%t-##WeOA9}i&CfoIII^bZ8YAOE2C;DvY1szTAC?!Wlr_w?r!R*xi95;S3
zpMOuLz^*OLdsCAk0s|y{>Zm6e{0U@L-UY>*Y7IJ~pF@-R0-cOxubP2Z=&_jg6Dj$}
zGiTJHCo>E7vqBqj!_;x)Fb1#+Rva%9iO#J}3jCGFHX7325}6D?U$Y6Om(rpI{4=<}
z`PI@E?0k#&<~T7`Xpgue!~<c>JEyy^N(8V}1X7X~Ctcu*O}v7FdwfWf<bq3pDudY)
zm&^y302gElU{QBko35KD<Idp?ZU>iOjcBOe!5Gi*Imlyq(lAIk5qO#N8{~+28aYJo
z;7kk+Dmg9-9sYa!T@*`0BSH_!`%)!Gq)<iGfrP}(yc5#NmJ9C`!cSJ%Is->h`DT4n
znW?g?v*`@nBj#`GvZ7Usv`?vmJi{Aj;B1@UFKNlC?3&z)j9ozQn)<2AUuZBABiYw+
z(DGW5WPe4Y*ouCLQibeT7i=hJaTAXVRz+?71+L2F6aK!w+134t1C2}c5#0|!$-XTx
zfT6SwXAI_>8FAy!A2FSv-GVT7KZUci7iuKhLOJ<DDcpw7p7u@k42r(R&mj%-T1l<m
zBsgv~T2!hoMS9fp`ENM&>M<DO#QtDg7Hy04her0?!x@As8=Pnw8<zoUhQNHGTV{sA
zdzks-Gd4kh+lEL$<2i_Hl#0>_2~_R)MkUyk%XQUat`XG@vwfta(TXWV-FWX1G*rQC
z(h$6W48A3F_I@M~+95Jen(sp$MpY*Q2E!oP6*MwoFvA|dT%x*@E)vAJ+TUpZ8e&uv
z$j`lXGSIkah<Jurfd7l$C1bBZ+L6bKD=#k}Az}ewx;>-Vncf}4QnkE(c*Fqge$~Oh
z5DPY~C1OEz!#6huE+i%n40nBori7nur%(}LmVI|-Ofmb9ta24%YzhWx3F_pDnfb~g
z-(}E#Z4l}%dpCiM$JxFTzpf>|SUF+aa4WouJTCk&<Nz$B3l^J-hIcBChJ+VY1&-@c
zsfuRL<w*po4zi;gGlDP{FG!6~y>{(e)Vyj9BQX#$TAWnSy@Y)=I45)E*FVwTwo9>G
zxG?ZGhQ+obw)`TN-`n%^N;)2qC#PVZdp%<KCtp}TCVCXI#Pxd3liX7f_4{aG{@bni
zt^``N#C*^9r%zKZOfAKFK)(bAGM>9x{N^2~-1voTwCYLiMB=uWFj$DFM9cuQ!30(v
z3pi2QgEO|q3Y*YpDr&@RZxir%xh8QLuJAun>_tJ+ykQw02ji)Fx9^ivcO<YExHy*?
zzcV;d+IuGAjleR|U-<<y@hbh1h47JZ%_W%{*PmIe8W;LEqx>J{*errSQ<yS{&dN4U
zlSKJDyuw!KZwLm7%Zu|UwFGNN9^>fk@!f6>6qs1b)kO$K+^bwWaip9%_>)7eM-n{Y
z@F54>NXm^$hO#0|JVPpoxK3~N4Ww<*|M!J^$rii>xP6^y=p7gAcVJ@}NKM`~pU>dg
zW_`o^`#x(Vmd(n!x^x6xa&yfiqq?q>uJ50}nea8er&VfX34#JxWBfxlpdb;T-#U<S
zLdu4)W65)ccVE4I61PTU;`B`%2*K#}4rGi&X?PVj^{!1#WB#i3>-W-ZeyMOiii|4h
z0w;8*alo(3jC`bGG}Sj2^7hFF%(^uf$*e6_GPb}WaLB!>a|Q&Vf>Gj2m3w`ncWsC<
zzfx`?T%)f3q$`+Nvdp`<JjF1h>dV!UzVqJ8*l=9u+!zlfZ)*{&_tPE7{Tw^)0w<`G
ziP^emzE0`E;3WydQAlOe{5al;Bkn*O2nlU436GIuTJNB!ukkPqq~ihc$!8PV0Fp|+
zRa2|RWOCre<QqM9U_4$$*SKAxw;MBKK84gw^eY{9(0{)m@s|A0-95;qbMw0-x2jHl
zwSI(1aheLWVh6n)sHr*iu50x3z<Xkxm!`}uFB625`RFa>NYSHNhaLkSQUMardW<p%
z1{W+fxP~r5i&HM82k12ERDugn>-+TP6%{>r5>Vj~7u%+$F}dpLhGCXfS?!AO$yfip
zYz94d=7Ru{I6Z{whHSJ#ttOJm%w3MWftA-0^RX2MbA7_Tn9ZXyq>iPuU^icURn`XN
zz1lI(a|1b|S)pT78U&Mc;v1k@a(1;Cnf+rb5vgD^FtA7VU`^0nT;L58Z}#@r2IJUH
z*dBh5+`z;&C_qsF;CbR?wx4$2h|EWdRuyk5F(DyA%wF<Cm$aAjiqHmHQtYSXj$&fW
zY(L&r%s$_?+Xj#QJY~Jd$>U-)J?T(OEx(=oM-S510w2*2#_wI?0I6`e0GyFfUcbiM
zv)s&-q9?;%#0l>N*^JHb4O9;m<<X=SB)Zg1goOk8<~q7Rf)Kxrj(M8}7eB;~@C?*V
zZU!C5XZnCQxOrU5u=8@s*|BkrOAbLtbq<%YybWbHF^=_5s>aMUO&Si^5aX2i$g@U`
z8F05F#$n!vc&(Kj?fw|Bx7U<S>Rs~bz4w{Y(B-b`@xnQ2j`_Yc+3Qhv?M(Fx7-pz>
z2KCSegG83M5AUFLha`v!<`x*<Q`|(pK)YH<N275i8knca7d+f*h7tD%Zrbhc*cPer
z_Ca0mw%h9g%My1til55@kG=*SM1E{T<)lPZiZM=4l?KUtqp0@-y1750Etrr238~Vr
zb{Uh1D+M)|uQbK!=F8-whRf_t5Mx|)AAATC5;55tTfyLJcLm!p$F31N!t?^=NS%J=
z_C#&isDwI^iOWR+Ee>(VT@(Kea}als01y=)gVC66jOi-zJo%d5aMeV%EaO$=is?6Y
zR<G&q_RA0LOf68K&}I!Z%&k$`4@_RgQ<WC0=9Gs17wveQpP5;!2BMxN8jE3LUGGG-
zqHH|jzUA05BgwY+w9+SDUJynfc77+q%y2Zi+EB}zhN;L3$dz4Qs=Pm$)jN=~DL-Mk
zEM|)&ue2-JC~id0{?jE(IO2d)#X~|oAE3-0v9D6|kG_T+R_Gbh5AZ=-=xgwSY4r2;
zejRxcY7qEPmz#z!g)d^*J-`~%n95nvjECkuKPHEsw*BI^Uho&!lh-ng>+me^Z^Dta
zPLLvy(UUR9=DQQU*{327-YlNTlS57o%rQK66kTB4XL237qoSwvCOEY!O_+7S8j<f-
zjES-l!Jet4$PV}}IeZmF&Bug)L~e~#`YRGOP~m#kwCNXY`fe4T3h&~rJSciP48Pp5
z@v11{$9t(L@SXkSAdUF(TtRt+n&X|@-FNe<DpR7&o)7UNtkClJxPRdx=++Vs;Q=>!
z?VBt32w3p{A~}oY%5u?=<S2cL-=NlS{SoVzsx<n)G%K2o%iz0@FkwNzc4k?!eEsUm
zg+nnBae{&t$*mug9ltuMdOv@}NKf=>IoJrfrtbc|b1xJMJSNs4zYOlXdRoh`?*O{n
zwnQC=YS)0rRq8dj^Z+3fO-H3EDA-yY)o4tT#*z@4tzax5x@31Pce_B!FxwO}Ec3C%
ze4Lbv2UI*9vGa;%;a;YPN1A}IZb!)_@!<q7@Qn)Cou3*Bn(3cuZ9d_5mk;GAY|FvE
zJ~lX-TE<fgL$N$qIA*G$ij=;6?T+?8``%+kHbPr!PDy3E>Ke&)@}b-Ek1f}~Pofoh
zgi9hJx*-YAuJ3{V;dKb-b-ka%^t`)d6+w7h)l#QHGoLVS3|OK(6ZlgZQ(6$|W<Lc-
z*zp_TuTcRYPaF?w#iAWyk<NH_TX0Y3<s}Cx_3rk1b1za?@&XejgnW0-VX^_cTv?`W
zyJB?=0x6y2Y7<XJq?))GQgkd3aWT<^&m63k@#S%tW#VUe3E%@aN*H(A4g2lDrh=JD
z=r;imnsVb5Bwdpx;+lA$d&D6Nhryaq9NFI6YbNm|D3G0uLJJ9G1lq}nh?0THIgU}k
z^8Tyxwfv!jXhLIq&LdBibE6mx?(7+b!eN-SlmvXvB|WOHVUiHbtj}V?Ia|V_(blT&
zXuW+aV|}fzAX%a<y}Si2&B*`UEwq6k>{`=~yxjMnR|h(u?=(mOL&E-wemw4%w*4+@
zTps}$z5#k+;$1Ep@>G9pep=$X)nU=PgD<#jncZuUg12EZX?_4H7{F@~rNF^NEltk2
zxk-Z7@5w%Ioy14Jbm(L@Hjr`U&gfSeLLfMV0Y=S!C8UrQ(m-cu6s1?HI)qOKD9tRa
zym;{<9F)3olBwgc5#h$|NA_rqHidG>2P-g^CcQ`28R;0iS%AD+;O`vvGNoOp@`Fns
zvIE^)y+gH?Rk!2;1r0cWnR4#OdX0qv$2hAi0lr8huTSz_IPZ>Xet?Qrhvq+DU%sM#
zR>bt>yEc9KRv*BFMbT6+5szCA0!o9V0J%*<jWQD1jA?(iVl9Gi$jbZE9kbf-Ns(V6
za2~&Yp5BVzAE1RJMKyRG^G#c)PzJYGNSL91w@Q||gdQy8{KKFiD0x%6;@Vl%WBFRe
z;W*J$xLvq@#GfSy^~5OLqEf)Hg0${_d6;Eg-67};bwRDBvfb&KH6b?Kx6u3`Z6-(Z
zNJSFdNmL|t1obY3<k})N=|yqldg5-!vbX0ef2;6|huf6&aw1qzM-P5Aa*;yupq>di
zcj=Hoe=%4lWb)TEv&zw_JQ}>d86kKSl_6rRD_NgV#L%i?Y6(FB$a28G>Nf8yQ}`E!
zY~c>1Fec^2P#id3IA;bxTIeVeN?sVL1U+<jIueUsDX`8RZbHOdg0bed1c7$uIeB;C
zx&!xJW?Q{xE#xu?@oj5*hPY*_Wfp)oJ-*8O{TiBs={QCSR)lCx#t<|xc3~`;=N{s=
zS3Ly5@yOU7G_&UI<-xmAX5&=oi9%C{L8G9hgXQOzF!^kF0*xmU0j%S7z~1E?H-7%e
zKTe-s0xe$i79YwP_zWxA;|Kr@)Nj02)Qd~dEEc!Mtuy~s&*$ks#4_A9ZD=zZ=ctP1
z`J)?uls||jk4bLzPtYBf&_n5idN#H7^EGRf;PBNE%_Oh={j~}Zh@+}m1jcQaAZLt#
zRfVaTOKNf57~4mA5UR*;Xc<(?X!}p8+=qx5S#Y2x(?TjFvIE|v`nE=tU=T$uXYdSm
z4%JJBs9#&qU|Oc$X%Els)7ugf)Q1Cj>E*OC)yca&Mk7GtD7Z)VO8X%sf4tH8DqFv`
z&%}3S^%p%aTXKl)JUzTWMoV$}8?Kv|-bqSDdH&7F7JXG$TE)J^hVJgT9qGGXR6oD_
z{`=Xj$Y?l4I&R0@u>_-Zq76cC_yTn&<nx|oi$43zqX<lx-B>?V8jL56wOs^Au->5N
zj2`)+u`7|M+b&*wN)%b&oH-U7y8D`+UyE3xTXQE99nip(DClK*uO@FUG6KW%iFkZF
zd^KhYH?SIx0=R~!pdPSwYS9zul$HW@SA0VJi(oug3!$u^I8gy!D)m%&UOiAi&<ouY
z&PJq1BgCCLq(IB#6JvdK5Ma!XI0!7!F9z(MracO_N`R7T-ns&}N1YlyXzcYitd~&=
zg<DTvGzw|Nce?`s5B2-dF(6$77xy~C!A_cQ#;aU6t*RcArv*K`k?L6!cn{oUs_*WE
zzVsy~BVJyjFd#x*!H?+24YolSxttju>Z1%hV|x$pAw-a8Sd~ep#i>bo3B9z$288Pl
zdcPQkuLNKcp%3~Y`FRF=asy7%H=l}C!8gCw&hh2qr@$6l8>>Ed*|xUsN>$|`jPmYB
zb#ioUxXS}K&!(j!4R)~V?vRDH`=KLgVfF3?vrSQvk$^H3cc+*KHQ|#Cgf&%yi-}sM
zbHS<#eV*y%G{|==C;V}bX-zvGFj{1fHdi4T;cy*sG#f)Lwt}>ai2Dm3VP4Y*SVB)b
zX5iw}8x(jDeSq=MqUIwb{a8E;XKde(1fCXIyLN@e5{OPu)^bCizK}gM_a5HcQg!(4
zrd!z>r~1)c+Kr3rR5K?)T=xjOgZvG6$N-fH`cgRvS-CA9?uBX%Bj(b1rJL!suxNGT
zpLaCJp%takgRJs+ymn1wZ7}xOd7pa;B}Wr%tf9Jll^G%w_zX0=bePNAH@Y!Xvq!3u
zT^8?#x-k%W3?EM|_vaYRDD+7%5a8p^)!(85y~$%MK~cj$hkCEk5!}CV6`+DH+l4`m
zPc}~`>2HP)K%i-x5skS75#)a@+-PQt8H?**M2m`dLqsteBAfuhS8U2f)k=;SN4niw
zgZH=C!0lL8-CVWOvGI5Wh}`ndvl!xtbMtQh;{3F-51950xdBqiG`ZouhqhmA0qfxh
z7w7?tNgFisg8W(P*OQ`@{>Z%QkCQiMzx93}v}C`8HS=fi1=^tM#w9I3c;(ACaLUz)
z_fI0oX~97d<@Qz)z?f8YwT_0~Lc$3B_9D64ntUL7eEVwbInPX#hX=J79PE{a%yP%B
zbkUv*B=w*-)Q>BBB4{fT7W6DG!SCUscvm*y#^MjbSK->a7!R!_beRpguo$S&ws<nU
zRO)6F-GGXY_8Aw_`mhP%91xVLutja-3qK9@Iq}^4!#B-_CP95lBL?9d@Os|Fi<~#=
zgjPTiq4TH4OQtIQN|L_(8T+cxb9S>QoC7+13YT5_W{`{MA3?vm4xvMh9`HD<s-q8@
zuXUo|+WIGCSVcpuvj|p+GZKnx)F-41ViW_q!#<?X7EtD#(&!VAm6-ey`=Zx%`8q!|
zmpmURaXpewgXTDy>@jFeFAZcWG1@Iy>1dAUMcT}W5~CYxg-5>4Lms(B@)#~7!W<)y
z380q=)kh|dihg{Fdh3{LJO;8~W3o^0?-z~GfT>6EMM*jF8)*&;oy=mLoG>)b2}WI4
z3A+wqY>1c;x5?QWnF&lwD`-3bAKMb)ns%;cN(bkmlD0q|stw(MYDw@>tl>Ao$VVC$
zvh^XbSRwKiYq&WtHLB~>tvG}ZCZ!X|5(z{n=01WBOP(l<*x3_g5O5V*f-%SBOV&Xs
z@3apRZq8=6m*~t>$L4g@16eF482W(6KLA5?aor(cL+U|Qpl%W*+ZZy+P=ZbiAMmS?
zYYD;$S)zkcTQtBq;~B)YYd|Vb3#yPGG~h7211l1FGiY-WZumm%av(D=a$NG>APKsO
zbIUDw<s~J4YW<!%2zzLKyVyW=&OuBD9po78)PjtB3-^@gd;u%vR?tU6i|zn!H<1;i
zKJFB{B1tcRQ-;?~7|*e5+wwB;2({dHIYs0NNU?{&9%)1$h^ENV$@?0!lIG;z!oUYn
zCWsT}u!FUG`Vp)8$T<H8y^5^F#nk7vjzGmvH3|r*XnsxAp~p@X3S4DlO3-n|t?;T6
zy92k;3i%6yp(>}c#tFI8=iI6fG;e|>zWjxsf$nkD$W^sRid+V(HPD+2UocPDhvbeQ
zqP$hnmprMGum<Z!n<bNuu*#L3cB}A+Tledt@Hc<9QvNClCba#)j4>YR{RCt9@>$dX
zmUJ(ItLz}+)Uw@k!f(speTirU&Mq@-@wps&010`^>qg>36Ds&OvxYx;x~^^4G?GC=
zBd1dzFIvRIVY5*{a}#WA-h;y~+ktXR{N1DyM{qiPykjoA8g;N^5&o$tmCMzu{v8tW
zh<LDoL=b=%Pt&W-E03h&fj**0MAv2q9>rR-RoNNnSRhxmyA+Fq;%kh<;D`MyBqH}W
zy!6{ps+m;Mcs?Dm^eYK)<ar*VcMK+xd3mYJ?H-`H6_b$Q!QX%}o1tQP5|=)o_S1DA
z8V9}O6V04G{=|6JAMY$Ml%B^jRr%&W-cdT04GYr^G<(B7JD2m?=|4TTCXCQ~CIjm8
zKS>q$r)X+nQUz17-?@Hd{DAw{v3}M2MF8_@%w1JXJ>Jf*PQK-%{KhPAn^*I8S^La=
z<2&vD+f-%R$b)e>_LtI*0Q`LX`Tsdhx(45mrHkkgLwFjtk>j$k5LRl%!QU*#SF4~e
ziF5Crnwgu0s&>~?`1C9fgFpClcn4Yvv^%e;L+!7$kD0F$_v3&1+s@F2VJs=Q1_60I
z4LdpMkFCkoo!^Jnd7pP#{DYZ2Mr-L$_8b5EZ~j03=>Pia&eg^JLpx@E&a5Ow_`RP`
W7k-~~cMJWOy<eN|%Gi1M<bMEKba=x6

diff --git a/vignettes/functional_api/unnamed-chunk-20-1.png b/vignettes/functional_api/unnamed-chunk-20-1.png
index 9b1a6efeb080d13ccd367ef5fe668bd22248b0ea..b5579e26b447ad5b20b115a17719554e823ef559 100644
GIT binary patch
literal 197539
zcmeFacTkmS);&sh4{eW(Iv_!mV1tT?fCLGG(A|OvC`b^IXoHAI5>Rrm?Xe{ul%OI>
ziJ}6c<ftt{K|q3nL?s#s5+q2vYaeIko4WT_-CMu<RsH@Mrs|W1a}MwOyw9`u+H0-7
zzo5f98f#W>T+PD5vPScOnm+z9U}5=w_YdFWC()+a-uUNxD{T!mmKFNn&9d}J7M9H{
znrc5Add7`(dqf@i{BqZuy?;{|WqkYXw==<kNr7d*{wQR8;dapNw?{S~5YpSFm9+nU
z<rX>4ob`ug=kEN)>Sy8V_P-z2`-b(KpHu!SY<~Rs@ge_!fk^qMWga_Z9{DBYHJB9H
zuGE&xcszAYENOhEwS^-HSMfhz?T?RSjQr(){u|%zxNvgEfBs;z@Ts(K|MLgZ|0iBx
zn2F15Hwzco+ZTP*B4b3Bg{A$Q`b5==2Uug4<%;E{g_`xc%6hNuSXl0V_vKx3xK;Sx
zncMU_mgBLhF{_VndU*DIv+@1oKii~#L$~wDDA%fLb>49nmi*wbP4Y(W^x*Cvx~u>C
zuk`DGe;NLz-tgc5fp31=|KcF)fBxXae_y~{j{mO1*In`dy$=7l6CIC@WEbZ<`Rq!3
ze7yAX{BUAT#8&McyM%XWhj5E0S(JH;_D>BrhG&<5x$~_I^ZeC@nCwzL%tq0Kr8WeI
z^mxg1hkdqsrb(WDb-<UF#Om96$@qle<ez{4-sRn`OKG9H@q2c=y18lR4w-v&bB%Tu
z=<)`<`E6An-X`93gFfg{{?Fo`3x;LhpB3KzW4(SgkA&F<nUl}AH{Ce>n1270R}*&o
zPZt%JrdzD%KD`y~S6!KJ>@ONlxcexSUcj>5`P|1>VF4=(y`ud%a>mo|nv(x{#dE4h
zb|^6q8}pU;Kj(vQ#?U$Y_t(=OZe;xXKk&`@(ErB^j)>ti&X1}a85zkswC}nh=eY6W
z#fuZ&9=%niUQ;_9#=Sp(P><YpC<*`O9DjVlIQNtuCpWjz(`8O^B_*W~FYoev{@50N
zW4HbO&mUe2i;0PquPiSZ=bb(>Ff>%%k!LTaRPTDUz^VI?#;?Ery7SnB6#X9i{_3z~
z&B$$dI{Z^Z^&zZm+=lO#=|V@0-8&r%PIctkwcb1CW+3kQ<c~F64`wAwMmP{f7+B}s
z+>#c2Y&;U7_^U^s_Hc^hFS>W1YrAb(;oq8JAac6(Hv5sgMqw}S9<8}^<gV502R1pr
z^OLnY54#-2^4`S7#kD(EEG3<+zi}ltcGJU?^{$x|q0e));7OcFG0HSikKc1HqWV>~
zr64Y|bDE+ZBg?#`TAjz76Az=?VdT$?Z2f{Oa~g{i;Q^uZQ^P#=<4d!>sxw_~jZqVS
zUB6Au!oosLjnBRP*QPxCw!(o3?IgYA{jp)D4^C)?ZjciCel6byOWz6goEeoj=f6v`
zYe^IB7xbIfyLas2Hk;DZFYmZCYl@_XN}A+uFJGLjJCtqz@(x#~X+h8Qmlyli9g6L}
zbxB}jdc3<y?cBh#NNe1pUEA#+MB>i;lIzsHXYHPMHxFsFq#a7uif&!!FTQhU=&^?<
ztI`wwHzw~73%(+(QC)oQqdr4*`o%ul%$n9gYC=6e<7HNso3n8Lah`oaqWx^#I|E~6
zY`N_-_niAsX>-k-8z10JXR7J`uTN6u58aRz?eU$e-yr8;d@w<|$k<*ZYGgbaN2@7a
zH?eT!L0$fJH=Mj0ZOJmj5kK0e?NZSxJ*~NRE$?1snw}l&Qk>}xSgEhhv?$wC9m=~t
zV0kLh`p<v-wW_0pJ|#&#Bx>S)OL|Q~um2`?_B9)21Mg-P@7Uup##bLJ-|NtYo9N4H
zvj`DM@D#G}8O^gAHLvhrQW&}aM8Dd$A@10~njnVCmv^_#OWd2=@nWmRxrx|#rHRV*
z$tHQH+1X87n*;8-^jF*2KXw~#2z&j|Hr+f~i0{m!Z<^8$#nzvGyf9i|X_#%XceL~(
z8&7qLW~9~g>rxXV4;%M8IXTtl(1!ed?Vjy6r77noa7P*l71%fTj;*I>ne*y(N8X!=
zML)y4M&en;AMU*CDf8j;8n_Z0R)PbP>@nU`)x$%3UPwsjn@4h~2I)FMfoBx>)cNin
z&DL2cIQy8DCnbJmqDC-1PAOn{QB7?Nc1H9DWzRE7_wV0-TG74R{?!(V;&(|_gZ_<s
zXH(2Sp8sy`0mDp_-4oX;7Psc$)`Vso8XI-HTx$4(b{x6GSu}h8o9~*l&b9w~DK)<0
zV^&F8yz1F^FY2oWu~pTFEIDiw{bslo%O-2Z?1ulgVb{$r)c_n8_n6+gU@kEZgfXoX
zHa1(klYM5#wk{t_+0(Fhwky~E+*nu9)7_`rvf}4j5|^WlbF9>ZdTepp`!4cW92mj#
z8=RSCXAhETP!PtUi*+^ATlrprFNPpnC&NCwEWdeW)8n`80*nW*E`%63*Vfi*nR#@d
z3&P>yo-Z&pHojiCFjf*0?Q>}@e{<GL{1}UYjbJuX+!@F&%z;oK={n9PQMh?@PsQ@*
z>=3j5hm9)QiUAFAil(8KICVKa`&<^^>jrdKE!Ey;o&DIB{e0X8u_Z`Dz-x<Mx=wtG
z51;R+$9CEKwk+VjKD^3|eTtPUdPS&bU?bNNHi@y`8XS`N>k2MbwJxUpvtwsQeyHgx
za?_rc{~K;3X4~h?sc%@Mjeq}sO&_8`xL9_n^n(Wvg83{FWHzYydSaz-J@;w5ttY&n
z!N*&%H0ms@q*U^B;^mzqA+b)yq6z+A-skG>x9W_Dj9kA?`O{CK0(-q%x=mI(Wg3;&
z*Ikp;Rje>AaH{J07@;kp$!m%bySOwSpsQ4&9>P6=eNq!9-?KqcQL(B=UPuUG5eJu(
zwql&uP|VgFv9&R_3B1U7r$@>c#_VPZWt~qtVQX7spu3lAw%fyUq;#r5ozYlZ8~ic5
zLS=bzSoK=na>dHhct74L>fW7cUBj4oW$LWH^yPD%O+bD*;_H53MyttzXPSu-a(-W?
zk=+i~&YKsX)il|slZMyZ)vV=}-ZHxAn91FR6Xc1%(|<dep;aP@KhF{s<=pu#%i>cV
zrBP1j(uw`K4zC}mO+9${uphT0Il)>j=reFdL+fSK%Obbo!r305x?B&TE8AOqwZ?D^
z85)9qXG5m!+p@wQ|Gp|c&ZEa`Se-(gSpb4#uv_C^d7DM#(vZ^9(x?eUwyHvx{=PTA
zuaj`iK8r`ZDZ66XyTdHl8mA|8WvO>1Y~$|JQFi=vWr2a7b8j#8Pu59FPL$GeF!s*0
zOON_haUHFU_vks+=Ou2E8`x&yBYp=T{GsgS1+gBLg^~5yrIT8%caGSHwEw<N<+`f{
z_E<lbRqT}jkEO3mZF?xMOdo!8fVcP6>Bp~Mzn<s|k*r4aI+Avd{_JL6g2Sr%*%oCA
zeSyMSyFHNf11%T(cx*aW1|oGOK6mvl4#~UJil#?-m$SV~J!)*cfufhFL0V;dj<q#n
z+{EBD%QQ#BFOLD`c`0l3E#Cb3cf^cc@8j^qs#g{pS2&B@OWa4Tu9>@Q^6GksG$Y_u
z*GX2yP4_JaP?kD)eb<jF^8=B_1y1I*F1Su9hxYpl?{5Ac{9Gw|m4N4FET8Fsf`>5e
zqMHgjs!P-Q^>_7CwYog}!)!WAcILfF4p-UpWoFd$!~E&THV-`~`mXOecbY!8n!&#*
zQbd=*C1%8_IR5F$y6}X(K29$FfHo5^k6Nb2wUznI3bkb+p(9spv+?!0%ewO6Xhm>8
zLn-U2%U9$S^#>r@d7**9!72nT?nK{7O<duvrxP}gX<wW-$vFf*ixjDuo9MTG`{%{4
zn;MsWvvr~9yoI!&@28)qM?2+B3Y_@QeP|6To9$6rtGf6};eASE-&}uqIM6{AVGhTR
z+@OLUZ?RF6XXSIdkW@V92d^14ya%kg!V}!m9d!1~^m{4)#kr`7Z@&A!Qov{Aiu>|0
z+juU@R=8XVBpNP)M<}4u9lou5_AE#w(O25B;eGC@CaZ^SUS3`k1Wqpr`fmU*o#Z9>
zd9!sj8=L&@r}xZ?9z31k-*a|50AR5F(DN86r>?^5if$)!s(X=noA6ZiM^hO_(lWLU
zp*~|pBJ>z;AI-LqJ>9zfhP<;t4Znw3+}UPLmr8bxN~e-h?o0gc#>xORb^+x$tgrhr
ztdRP@U$ZgFwSlgBVdSB12cKl=r$3`~#l(Ocetq&sYFx?Q)(dQGY}Uxp%nOjQ)p6)5
z%xSg%4j22=q}E0j7QMrV|Bj##LW=-IC@ho&Tp@*kcd9MRyw*$HG(Qvok<g%a84hru
z(s-%N7@`nh;>)xtk@&OvbX83NoVg2$xZsK9h4E{iy%#QA&>xJg#asJu=3Y-1<5Qae
zlA`+455;m96%~p0*IgACYg^Q^A9#AjrG9r?Wt6BvFkcB$pC__+KTvEKJwKhjpIcM<
zJMtYjVJn+1t|b6AoEYHWtVlUN&ZU}nVkp*mTi7^`YSXQKm!gVNOV+=?q9MS`ujGD^
zH&M>fB<{?ge>LrLJXe2XcT=`yg>jLqO_xWnpXP7^AdyRlU3yigb49%CQqoSd!d=++
zn<UHU`cCSu+(F1bJJOzGTz?76#q#NQ{9P6A*)e_sSb>4@?j52|<?};@@6%%S8!`|}
z_}rdLO?>&7ea&thpizCir>r0Gr{}B_a&r(CIOX1&{c5eqBE=r#0v=s18bPx=$4$<t
z(eW1+5fKp*x^pyJphhsjw>GEOf58!$McBr7X5{r?f~DO-pte0{-~ByVBV5AhyAH%8
z_M?`5=X?>CUFyUOw(z$A+$nsxo!s|iood&};^iOr_04oThvUgPONk(-6uXaZl<)O*
zt1Xi(8i<I>k?--eZn0gc$^UHfi2zwv$yhMuR6N|0L&Np}<F@hVUZpki(alHwgAOgH
z_sj@yxa#uU$zfDa$-q5QZbQ#cbrrfadmXuB7{o1ZQu&X+uGcC9L@Kj>bp9LDxPX<V
z+5T1o-7sC1`7JZ3ucD?TD;6j9cN?S@|9-;GZ|QRzJpxDXrH6@2JO5ZGnBcnTSkTRl
zz;0GFbboi$(0fa&YAV<3rZ}YC0U|`Ek=~{gSNM89VCkdfL=}%^io?y1fM~TR@63#L
zij;j4m`Ie=PQf;i44ALW)h#Rj^5Jelq33e1fajav=3Zv|b1jTl1T@d>ZZk9Z!1nH;
zZ6gDTboWqWVud==n2k+wrzqWYj4cYgCgiXB8#V}<JhXbU&OOQ!g>NoIxrHOlOYb%a
z<5QGIy3SR1bO&Z>%CWB5u=})O91gYUu%-X}VCw0#9rTbLUd_Ek9Xf|+V!-fwpL#7x
zBV1rGVV~C~Mb{G<L&#CqqOlo5%Pl9S`%m797CTzqqN~ck*JtM7p5a=t>`?!uFI)YD
z3<#q{8R%#hSr%`ycdW~MgBVq#I2^ZrWqGzDvf9FHFj_y8jjo<qVxD(w!saSCmGb2=
zHbO}0Pi+)>#+PW-9A6zPK5N%ea%;aD6%MH#KK5omHEtnL<2TzazgB7Fz5=d_`Se<J
zlpI@{aQ(Sc>lEIcPZ<{Ues?R%twD8Vu^lgttd_u;iJDrTZwyHD^ET+rcQmLz`TTm$
z>-9Aeq6kM#$r|aQ1;`0^3`M&G7+{>X+6i;fWkw<A-tD_#pQSn9dW%lj=;DLc3-V4~
z8`{j=4%HSr0F8Eds#X5Hia+Xs%5{dxnb$zFRM+;Qq6))vsS!<&dvHvA?()91no1S2
zd2g_m!Z<3IWnX0&q(5zF(Aj8O?(4OP!tX%QSiTusIpTYSoIVvcKOOUu=3@DYg9`MH
zx#ux5E!A$Wu3BZYpWZ$(uqa%443OIQOe8TB_+m>K-84bYIRrWFTo1j42SwfPo#KS0
zPxY4j<(_`IW9GGOVieE4kP-`N0Em#1v0`S8D{h6VHeABIxDtV_?Dagdi~7RMsKMB7
zgxaW_vY8IC8T=+{!nu6@<U<igYQ%>r-igFnZ;JmY-%JMftc(JE1OiEAtGUs_ZH`6^
zkh&^ienYx^xBX*6`#X*AZyvplkJv_;8u`k#R;q^2?O-Vmz}5~^@xl{WYi`HR{IIqB
zdn$uO5;m;jJM$BP$5ac+BjTPY-YJ?^12o3S_5grjE4OkFGvkHN8bK9hl-7x|`QmLC
zrY0w?-aPtk>kY~zELGqUSvCiuTu^fKtqSI{>sK1hHOQD<Mgbno7kTX(4?;zlaJaIx
zoiZX3)-omLv{(j$<l>k|MF`HW1}gVj$L<sL`t0oXukHm=EznfrF`hEv)Lrb@bnqoY
z`)$<1%@ar)A=fOtMQsF7$7;q7Q}#okGU|}3%4t+th?@J>yIAQq5ko&|I)2Q16Bvkq
zrQj%Y1eAbwX@SX^33KoFsV>9u9@Rj722S>AcUU#ll;__3j^o#WAGM>bu|i!G+c26_
zBfzZZ&77ZJ{!!Gr(^Y(Kw4j%P!+F)qnQr_*+Sx|^RV?kl4Fb&|2R1V_&+*hu-aplp
zWEEyw>gm$qc}K}qUk=cIqA_4u#;NS{D_6&q;K17^yFZt$MgHbQaMQE~z&Q@Ybn>Cw
z`!TgI1r<x;qqc=O@#-f`RJ0y80w?8jNz;DaY*3G+;AkeU@UR=9l}g&H_QFThzL<Qv
zn_a#Up-g8ZfjP5tFOS5=m%K{~Nz#fQuw0s3oNDav4_DPGx@NMAt~^FdhKuD<_&t-n
zJ5QxpSy@dTR~8o+t-{1U2>IKM1>T5`j+R3$|CuAZD4AFG)ZsYq4;ftL<0G#h_5<l1
z;vIjL*_pOtj>EtN3OV~Otvyt7^w?OMDM#vjgocyTn9)-vJ9F?zYoxPu9yi}Dfi9v{
zkORbgd;k3N&kgw0+BrhH$n3EeU{$I_YwoZzp^0mTv&UapF5i-K>M|a<Zm9dij&&s?
z544Ly=^zKq64%k;*<E~2yL5;#fTAS0Azq0`Q9@Eu68I13^j`Z&9e2rSzA2yo+}qTi
z0Blv4v7+H^VdqBLrHJ^_z9tbi)u1FfsQ!uYcV*D}ma&J3{rbbk92sL*Z#3XV1;1>g
zl{#@J>BHj~6E?=v5$`iVJ$<?XUktAFuU@^n580j3)z#IG%$qiK?AWod;C_WRuM5a9
zVj%MacRO~-BKGo7eIbkQbwDVl*aL@2C7`;?UAWD><;xke2H&taSg%!?dou@Q9kN>@
z{shm9^+9SKieL*4ZgK{#a>Z-(X><2mnLbsI+7Sd4M5P_?yR8syH?5%N(=J=NE@f?n
zQ%q1x+zN-6S%A3o`pu@+yI~Kjk3Br`EFzbp%r>A>gMmHjF-jgzF>Q3C=SPp|?e6Jg
zbZ*Rh^X2nL4T@hnPOHvU^V+ImC+JtpJGB3OX<;^;P=eh)eUOGBzDdS=u-t)wo9d`1
zBdT#`1?W!A&NiuwCQ8eBProy!bKuu{f%flHL!veFr6>jwMCEV6l@frcovV!&vwnPm
z)ilns#NDyuTu8BkKdQpiu<>^YmeF#{ASh(9IKM|t5bRaFbV<N_JAEgM6i`6i%Hn{C
zs^-St`@il}1{_u&BWwaHARLcUB&-5Ydm_YgF=<Yd8L~_jrk=|X=jqsQ;O_2D<;v(D
z+JoG~wI*yTD~nSt)gEWnUtU=F{E;UBxx=8kE}2{T%$rA1^@hW7ZaT7d&08c0NKJ*f
zS&Nfg!onhb<XK|sS|}(lk$BpHF<Tg*!E`(3P|ZYIoVNt>?C`1)@NuYZf4oZ2Y#=fw
zI3h7IQO*JtopfeJUh2Yy#}91kn&tG>7l)nFY(7u7=%z336>xp}qsw^N?0!_a5!HYV
z`~cAEV>#}l9iD(!2GyN;_SYu+yj_qOsrb?VFi%%=UzkE!v)mIkwMFa>q_}p#NuXdU
zugL-NhHI(ZW+=G}-~O?wxk0j)VaCP?q#~(>d7~(<JU?a@L{y`sPYzgFc6lyUvn5RO
z*n=$=exF|o&3@(EHyhYN18c7y$!*b*vZ@UHiUn{t8;@Fe*>vo-2w3`@GCce|Cepnp
z+yC?3c2xES3ca8cHURa+NZi=3=8HS9>&GLMX!&Al`t#2}*Ba~S=zv3(v00=94-zu>
zx#B%AjidbP*<j<Ma{9W0rdLUpP3zaTkCluIR4jCl2baBC2g;@oH8}$ZS`9}+TUn4+
z!CiKp>f#wbzv(7TYhoQJ(%TJJ%n!wCe!Pn3tBX{XqFdN&mmY70gY5F+=T)ikmQhKk
zA8YWRGc?S2FzRpqqq(?(P!dWLQELQ%;9g*@!3d4w-jyRq&AD*yHLV+!rw&pTYwYvS
z;~aFy$Q?<@;r&zL@K7LL=>$Nm4XXaW_smP=W<W<3A~yHkd)T0*<T2WjM-^8}PSARh
zwD?XSG@`1yD89c>PpocG8s`I3YIXMGt4y;ofn%)(SBRA)5Mqt;B*&^MxIQ84Gcr^4
z^!pa=Qd<zn{dgc^6QzNHz@tv?kG*iPny|A{tRI4*L5ULH=~UtvVhnz(58+(gO{Qa=
z0Km;V1eKt2hs)sCsm_?0BZihrBJ4b^jTN6?nF?PyPzf&o$bkC%(oCl$cH4u(zF;vc
zJVi<OwajgojL_F@g)=?zT&A&dlXEG%_SKz1h$uCbtFKx5dl_clSdLgmof0?g3qtmJ
zqN*y?iyN{D(a~I3D_3Q}PAS00?c28pqNY$oDO#8MM*#u`F_a_4AG0+3v5!M|EIfbQ
zAG(1(gh#UHo|h=X|F(nI5BKe@E8&i7_}pfhn0rMWbyY&Z^1QTs^3og7D-1A4dkAWA
z)Qdd}%ll_kQe#FXJrZniy*ZUr-~)2#Z#B0gK3Rb(aQXBYJXHm1qZ;T3H5}une1vI`
zGoucvBY_ch{;nmxkvj*3mIf|ZyL1>F>IPimuR+5&&bB#-yoa)Ty+39WR;-<%NJ>Nl
z-2f?C&^HcQdIMtb&u~3(5@hrQs-@KW!5S@Bk><b^*}-nSb)wxmd~2930PxE45|O!d
z>S(t>ve}RF9EIF|BCaSW(uc2A{`AOdoLLXVq=AHPO4Yvht~pgZXy{O!f)sdxNNE*>
zXkxf)Jt?`hRV<%d{L)q-(Ntdb8CW*e(dh#!2-LwYAye2c!QZV_xLLUn?IErZ4{}px
z+o2nr09c;SZ|oK$_V2niBM6X99d(LU^gd}S-H^>{u;At+ip2Gh;?Umz@@qEuy6#f%
zZ-fO)1NIdTUNvoIcr8xU^y7qxzx*Ow%RTTfF94WdNdtoyf2`R=l!EBwdy~#P07i0O
zX2%5qEOZppP#iWv?n%jcu;>)vK=AZ*n%3vs2WJHW7AH=juIZ38qp%G<>_}f86Hc{6
zr&?T3OcbPW1rI=!HN7-9uPiAlPe2(vP+wK_&I7;6XCf0#8iMLLfCdb^o>DKY>UWyf
zgU8qekawoXdsR>4dEJ}%N^l*{-9a)YOKb^Xw>418)`SAr!P>dkjsD!YNlhiE7dL;F
zLV)GSuE0`};G(5GULyd7K{{LS7^!3db58`pAP&|>yF6JVA7v^*6zjo6l`>;%5uLaV
z25ARHU#g%sB$_Asv;-Zy%N<lUTLvth3mV?a5mY#I!#>#e%L^bI>rZbUr`C>MHFXNY
z;d)@`7O)G^M*n4T0T<GGFt#v9qxKdKlxX3ub=LEfgLG~=?H&C6{TD}_1H@jce7&y|
z9K>*TUPoS&?<jB<1g|YQ6K7L*wMy52GI9>=TL@nXK<hh*d7x76w|^YQrbHy}ub1!L
zITL3{<T=&#F>M(;XL|zzNT*P9qx&T|0M;d`Yx2MrjtPW)F8dJ>7+PhzNC;*wRa~@h
znN?R=LjY(Q^B|R8W~>o50k|M+N@N_5j}BfwJ@Q(Ho}yJ)#aQvjJ}`u0Ghc{Sw9hKq
zHdffruxiw_0k!A?=`q?XzE&C}U?H?RYtUd-2@kQNbQ+4}x31w+xNqEVfoj1Rb#lhc
zKT}>L**@C82a-3Xkh($2N{y3~GllQkfh~PRSrY>Z5+t~68K9!?AM14j2|n1>3=R+1
zP)QHQ&@)J)kZ15gokR$jr@cnm7UoX&Mk!ErBqYkFo99r1H8VPaV`%w9sBBVM{Pe`N
z(Hb&D4i0gyHA$RLF0GTCve4c8*>D6}+C*pXa=xKQT|79xzipD+N%c4htBW5kS9ZAR
zFc+&VR^Hiy50T9Uq!G_C$t9F$Nr-t9_?O#fX}n@Z1rIW@5=&-!!r(>8c=;ycA#<Gj
z4Ee%UzML|1eYRD~xu?|C^r6sRAS9Q*AdzZ7n-D%F_mOr5a5ViWm3Pi66MUMRtynQ1
zC`8?M)l2f9@aX54wYZX-vLsoydkx1MRA1Y9EW&*mo7BMK`ZAJvaF?pQygYDUJWAn;
zrTnM6gQY#jx@b$0#w`JYlVn25#d<*3#)W5oJs2asm0~DZx$Q5%fL=XhcIGiFAz+HL
zQ~>BfTx>6Nv8FtO0uPLK6%R?Tu|tHyM;Df%6@=t@VDBuxT^A6D`&McIrJ?|0(;Bdg
zfyhi@xmgidPS@MNHcC{FDELuReQc`(yqlm9;kwwG@*O)*A@`2p1Fk>_!ube?zVk2n
zwqqnKmZYGCDX}|>A8YW``B5~5p~y<*n`+jUbevJz8G_oS5^Tcy*)BKTkqW@BKmRJC
zhLV@m*&O6-bzlWy<T-jbP`x>Y-ExGPIVcRlOIvbLBBw6#q8g2^^PvEDeBtmblQYLY
z9-5z@w;R@tl@lLb>@x<y?mF{ijSB!?HMZ;~(w}9Wx)fYpU9Beis-zI5h{OWrm|^&F
z7^N+8sUVKMN3L_P3IqJ#srLJC(@dsF5&@GGiofb8H;(d>lz}bgZYWi%dRA~E!`X%R
zSMu7%u@k9DG99f8%9+r6_wI>Tr{D29Y9#L2R~h7jkY59;Vm$?dq54=2L`}L)B!5n#
z$VvG|=_y~YxrM_{RX%mzd@Y~CE)+VE(#wQ<P>5Odor}x}H2H+&R`BH<H*NR>Tg9#%
z*s??EuPn=rWC1{gv4Y9TL7i3?P)WpQ@zHndLcdx`tbE+z06|8S(XVdA)At<BYlCUz
zxM>=dv7HP$o!|d$!(kYyO!c5<_5n=jFw0;guV8kiw2_gKIRGnCqz-Oesk2<>AcB?E
zmet|zP(ys|WG^M*%}H7|BKSATqF5<M0(5f^M3>dinbwu_nLnZ&A)#1W1@SppoB7jx
zH-aX|m0p2{jLJz*@ZN#KbH}bt_z=`le?Tl%BFR%0m2%;Akln1dbh#{@k`$}&$L|@q
zypM=RO>5Nnbk2>=?5E$eL=~W_szM@-kybhN;^sfEAwPg=Dq*>KifAJO;Q-a7H3gx_
z(r8TBn~cg^v!(*OT?)r?bpr=|*7*mdFVUG~XU{>Mr2(9fgedLu^v6qE_*L;j&$kzO
zc1Ww!i`&PjWH>+ih+^R?LEkeu)kBDkhvd6+qrdujZkB!?>sIogue0R#S;Wsdv3E_S
zgqd%iFl5^D&QG43s+X_9mJm}~L%(3jPXZghJHc<}b*&Dwz{^)+F5D_<tYzVFg3peB
zzff0oNj3|h6{#jVo@u&q@c4?>W0<fDcsnH&yZ*>=xsmN;iFwzSZAoZcB7r{nk=*h3
z&8e^+5Jk^=@(Cm9=FQE2{Ee3bx=axKD@WY-&;{K$Z-y=9@MB8ro{&QE_w!{0{pWt&
zX_9-Y<s%3X<1+7a&tv5!tHU<&koF1eebnj?dhervP4Z4=aqb;<(edAR#_-CVba?d-
z<QE|v2@vnrI5ASL15dB32KaA1`2)S9Q<eF=`3Z~fQ2J`a%YOj0Dv6+4{?8?GKlQ)E
zc0|kiKVSbZzNG)YfR^X~hk@(A3-!OaP>1s#tY^7-mI_mFTud|T?JY!rgUMFHk-_PU
z{tOuCjuc6I|GX9ud}1YJax?P5vOHR=F2nW6{f=`lC29lu*<no-sOjXvlm3a~oY&M4
z^Mj%X@Xl7Iv32m?*WgVT{q^fGJCSz(@x}WySk?ddy8Pc4FjwRM*gD+bSB%&XOO1UC
z@g%i6L93?`-yLQaR^AqlmoMy4+I@V}OB5I4%k{Srya5s6oTf0}e3-4l;q>F5G&7j|
zEISTs`)Y2uK!0Dgj-eC3r;1&Jj~!TAg9P0VZP|jbwuOU{tAZTH1L4ocCVlnDd18s>
zE9UFDhp|&RNwD2|;x7*CoQejo-ZshmaB}!nmiZ?6t~2|#+8}S=A2ne6en+IquEQId
zCD%`+y4I4oo&>@=2Y0d`%z|}gAUjcb<mVhf>ICAk{?M91%CknIieD1Yn9ER%Z8gb8
zuQR?=LLv1>B`CLig<g;pgu7cBMggdsORg`U;X|cVRr?m=fmYFe08Q=B<?V8QaR1q<
z%7EFT59P7Xtdl6&IY=JFj@t2mkYSSQ6!C_2kc5bF`719@fkR2_h2I>1Gug$Ji6Rgv
z&2M2w{M8=lz9(vTXF=i0T$K1)LJ8s29M++fxr+bnVdc-Q271QE#v@)U;2o0i5M27h
z<UwY%2lHO-BE0_o&nmbWjb6Vuq;kT^ePRfgduJr-<^doB7x)02UCOcYO&}4avsF>f
z%`Nn<SPY$ex)YbsJ+!RK7=XB4oHFc;WnbNZg+AVy%3azIlVromyf@Iv9y~C(Ae6!S
zP~00d53$fV`&Bo&$sLAj@8)x-ilv!Cc<ZQ|k8MDrmIQ~u<y`jJh!QweW1yfIBs9UR
zk*@%DKtAjcDp_*#k?j%J89l*$p1<<V<KL~JEt$rR<Eae{3`ErlRKTW{eDmf_vsXeg
zB=Ud7Fuz-!LWM$=t!bPftXZYNA<>uF$(XDzzF{r^eG%6UU%qDf<RQwIwE2f00?&W1
z&B?_Xx_5i<hQPpI4sx#9ZoEFlH<&Bu$K&5b-_F^5b;H5Rzdx2py1@V4_6w}6hJT4#
zt97i_H(<c!$$g~~nK}u!k?~pODP^C=y~BIQ<BOEo#^c>=O>c2O&p!pGzo{Z%rCH~9
z*{}2_x4}2})dhg(3zxrqjO?IlX%13UAHW7<y^g|<<fn!3$^h-Nm+UO$Fd;*zctID(
zVBNJwetEzbqRU>X{N?y6@;h#jK5?LY0rZuT!;~+!pOo+1C#w*R*DtayRRuXZbyB=5
zP`WGSyKciX>EzBkew=w}BcNr%0O)daTQ;v@o|(h>#}2aobcJLxJfrB=S5PECy^16O
zxw0}cLF$>oaY}xIE#_(!kW>3X=XwMUu#L2nXi6tHVFoq4I+0!HvOot);hjhBNxjag
zkC71}I7RCKLQD$&u^Z`rpkS$B(tipdc5@ohqLU-<c)krW4%nu}|Mv^~{H{qD)mwZ?
zH>j#8k8ckSf&{L0lWzgh-CE$m!s`R~+LYrs4ror}ZdBWas62xAN+{)%MF(b1Eu}T9
z^-{H>Pk~uUn1PkG8XC3MkHE+0o71`f=Ahkr;q|M%I5raqbW!dbIe!t~fHkX?3&~*F
z9zkjsN_MvbvfLG7Q-yGTymL2;XuZuIKFf>q6R0GeTo|1<)P&ec$XSH1Ae?^d{@Ugm
zmBD<ZMv$3~$Y!M~{c9-c=;Zc=2P|4sjmNB0p#odCnR|o*+-lS?`1=!l#)9C*b!t*3
z16hbf@pe+W+RwF_C!^%aK`7o2r+HsyLC+5T=Sbypf2`%-XdCCUpXyR9^hSuT4C0XW
zH2ArWnx8uY9?>gglCk~u)OJwPSUk?X-qV5m*6o#r-oq2U<47<SIFMQRpcdPO0#Jv9
z-`YYJA7XTTXU~C5Q5z~9&F`d=cPodF+j<e5P->Ua6p7~~hcIMnHQWZ3JY>hI%1BSx
zNWmY+yijdu<?SUwD-d}u!1nv0UMGR#qX=D1FB9G$8VF+skLVam6l;<dC?Nl~N?-$-
z)=(<mMm<kF0r8a^_MAQSJYjFi_3+zgQ5<Zi5Rxf<6e<X8q<bjPKyW)j&M@XW;LdPN
zs^N#=Bpr&VNFrr8p8bmA)ddf!>c>lb9Ef-JAQmN>V4Nhvw-maodT<dUvnPa5R0&yu
z)(pjM<jA>=8#IJ31EhO6kt=;rs@+frZ<BLqKRnQd(>3ke3F`J43MO?1%GmpR_1PqD
z!23>;AEg^ToE6F|*1%g~WOBg9H|cmEy(jO@pBb#fx-cHtKgJme0u7){);-7x!CKKf
z$gkgrzZz!hT(<kFxT#S}p-m<?AG^z7^iij1FZWh5N5QKjSkUFFO%ffsE84+x3ZZ`i
zu4R7G&B=}~0C~q4((sjGW`;R0z4PC`?CsMzr0do{gxOSco~R*H&T%}61Ho;eh@$+`
z55`2adceVG*Vuw2+2sP?*z<=gu<7W2obR2c_50R(r+Kj~wNO;T5ny^eZZ+4AEm@_L
zwUyYAY<n3n$B28Ff%$1sSw_;`W7mMS8GQc`VAOVzPZ_bsZuoL7NUrrL$ec~4j@&br
za)ASA8|aNFX$$}CCs^Ys1sqwnuwEpwT#HL;LRUzOTTcc%G6ZLKH7fLnR`ykeu!WWE
zZb&{<hQJ#{yFJsI5``7iT;3{4JJPz!i;$31lNIPj-gS+4mw1%BoNS1TaXq-L;vorD
z^vHMzVb&Ab+T*`4n%hUl9Gf__Lx^w$sgVlC^qec$!eqlimV$-CG~L7mBH@zIkYUGf
zaB(#k2OJa@R^Y<z8gudvf>=s!OUENEY8)Lx#*1?`D-QSwK8XV-MyC}Cb2JX}z<PR9
zU#xSvrac2M9^7OL-AEV2J9!V+^O&9a`PLMuM5!F?`rM%|P@t4jbcp&J?e75F&mpYA
zI4HI}L@PJxw*(1kr~dPZ7f4Q_4HZ9ZMVxEDI{o3LSY|*?05E^*jdYVwD*~}?7eo%-
zh!Uv)*}0XN(O<0dz+_DT$?$mdk#%8X0W3+mi~#V$n*ER%uBPjfChNZzNA2pXMOHmF
z3FUyiH?6G~R%V~3eAR(n{IhN`9|PGq6l}4lIY_+fsL2z4c>%&bb*k`|?T0DFq1=Vk
zwJ2-&=VFW`lunB)fS4j#3ickpD>E)f(5v9ma!7tF;9%?wpK7PY>Bp?`G8qEz%NuY{
zhmGa|1cL`)meA}NgY$yBO6lXhhp=|+DC#5`3N)jm-g8GpD=}i6w@*u4gIp30DXIqi
zUWs~dTF%}ZP>HC?gtmgw3RZ%NRO#k@7|>6u;iBT8ZY8!+CFEXeeOWzPfIUx?Cc>x0
zXh{qFb}PXYHi#9;{jzdzOJg#Qr>f4L_bJz<ZM005$lXf3X+J%g`<AeW(C(d`fCDg<
zt75wa1a{v}(|m_x4sKCm*JcwN$i40h6JV8GO(T(-*sb9Vex+*cHJOb{AhqT`X9w`Z
z3Hp({cXtk$ERrdxN9ngE>`4K#TC^04y$4CoLAGw>r&K=%49F@(2@@nohL_Bg>b$Vn
z5CdpgwMM?{u+Wqr*rN&D%pN+$M5A={6ZZ*VP5oijY?7t-)fM65l|~NXTDN!l@GZYY
zm^H{Vz&0w6mJTAjlWiLlCLnF3#MeR8<RudlxgK($>Eza^s}Xy4L)I>0yxeeH9uHAz
zN8X#~WPDpjvgGb?hP$%~b}aGbq14eZlQVDgYfL^_E>B)dG5G{eHO^=;=`h=JV&6U1
za5xMpXcFJ_+Y$_48$kn}E=4T4SiMsl4yVP&SZEt*YFYE6x0!uXBq=394I?vddVC#J
ziayeOLC~sj?=-$VQ}_dukl$XnGws{W)&z7=?hZPQBVV`xeK_kq$A#@je^+G3>XO5S
zW%PM{;OoMW<Elv2M7nm`X<4pplx4<?AQ6mq2Q>&sZVPm+WOCAf?)k9;lshzQaY$p4
zo$hdvRKlIo2gM@{3~(i^!Q`@<fHA2CRH^_H<1^0$q{S)(O0-ufk+e)r2i9QQC>{{S
zour$nqM7{!agTHTwu1+hmI)1zleQFU470}#)b>5&93hqEx&A`L-(<)+MSiLXt?+EJ
zN!M1SG$d0JqWYyP#0zwFDCH}H59ny}10t_P_flW#HH%<a<OUd~5Ewup{u(=fvkp-3
zit_>vte%4#^^ZMo$w$>e2sp%%5xBCfL){vLTZL*6LiZ8OB)_px4LOAS;6>H$SVA4q
zNreR4h#MI?0C}tcuGMSpfV^lo<f@M7bpSCZjH|c5eR8P=7C*gvZcC6*Nox}Sa{e(=
zKr+gj31}8Pc8dsANz|+1ce&3N|Av7mPwT3{E>pk)7%Vf=`+<N$sh^_XC8I~RI$Y3l
zzzmx)U;gyt7_aFj4L6=oWd4*w)Wwq*q^luIZ6%3`;8vcFs)E*3K(Nsg>(0~>SJ;cd
zWWc6l#FzJGqQ8czg&SA+`<X1kHra<~84l!>He!Gl3HJF!c?c%qSng8e;n#ZhkKtG+
zt0eWEU2zQ#q=%af`H-6YLaUi&_D{S=?(~MR{v-fkprxjED;&R$Heo<;SB-GBL!Qg2
z9K~nqcafIY>kl1N^MgF}>)hc}%_*8=g@_#7?F9$|MSe<1LNk1@B>E(@iD5g@MoZ_D
zL3LP(o|UN973wp3y#k*^-DoGUMXsCAm-p$BlQ1Hf)RqIl9r&OG15xhqGjfMT4i(`c
zuI7)#7j?6&r?~lAq@wK>wm=ZF?ykY>_2ET`#jHcL57nR$aNtTMWrs-;@t=zA-BJZJ
zOq)z?MPXxdqw0z>Vfe0PthYAlERXuBYh^u3?jmbAwQoFd25OF+6ogIYx15j<RG-gV
zDB|ivtsE1k(!$C5K*WIZC`4jyU+5>A%p@<Z^J~RJ;u%Z=RyF&PL2S?ry%3_Wigp~$
ziZe!oNV!{>psI>qx$nGHOgffIbog}#nn<onHwrTpgve1iR3ce1dkvPl^w1+OdW-Qy
zw=oolu3P!=-#v2o>h<0BMz!;3N;AASTu<rCmF#$2oyz<E_}jC0d1WY9^i@=Wx1Vn}
z(tO`r5m4O-KIQ%>c}q-I7}N`+C9aY8<^uI0_O_9mLtF7ALO~E}1(DAIz|;COg22T3
zz2IfiqC3&E8Ckt%lXL&u_J4T+<aXNhqU%rOvme5B^9^Bx(~sW|SD=;W5>j{63dA16
z<0DBxsdIXV#XaE!sM>Q>Z3{E|40n9piG*3BiW(C3Dv@L{Z08ItPrG09YpQmkRPDOV
zU&QT;yo|d;lW^m^bKL^}Ow6P++NG2B&?(!Cg0mVr<E4~Q>i4PQ@W9#D>99cT$f@k+
z{@X^`?Py?#bzS+ZNZjAi9d~rh9pJushB1PO+^|#xV#)<zk7rLBeFtE;A5BEf31r$|
zM)fsZ^l>ze-23Dl8=IkCA<z(_3vK9`MH3ej1mBuEita(`LnG1GsgqimPR^_x=^AcA
zhT^+xlo=ejqLPjVkzdfilJtlVxP}`i`&qX?U=mRh?*)Ddnu9>3Kh_DUBg8QQ5*@kf
zzLuCUIvv^R<_oSx>crE_Q44TFY@)o#q(4AmQ4T?Us)o8tpYgu4z`4VNv%Zm11<*qX
z#e0D5zN8Qdt=ArK?o41q57`0uO$4mW1{8Ueb(;Y`S;3)5rSwxY+fd(T@A6DRJ7U+=
z=n4c#o;hVlZTJonr^_cFV&7IlJUYD7O&JZ?6F}?fUA>6ho}dog#vdu$)J9ROYV9(a
z6H(OhC?;aZ68+K-^ch4K1h7Bpx7L3KIdXWuKwYhph?sB37hj_!Bk|7)2`vW@jus^e
z#e?WVFErB}!inMi=+}H&;0L@#*PGjqZQ#(@dO@bG+;5=})-r=J1b%<AA?n@w<ziPp
zlAfP^iqVJG3oYr~ajNJp&15Y=9|OsMb7+{ehGI$<u-k?i`nAjG*Ll`rbk+VbKwAB5
z*$2G4JIUF!qm^O8mH@EwH)x-Oh2O1H-jx7;h(<S1-ilLn>qCoQ7<`n&odMMEO8y}~
zzOSr^9I+zRS0T6O(h23(?gHW#ag?+uEi~@CnvXm}{eY`Sse>ye#E)1zIMwpE7<6u1
zgtMba^(fNnBQ`hFKAc}hq|Tm<7yg8J#u}-Kj~JMCS1JMfp=8&0H#j$G?x^w;z#s}$
zy2h?iK&o9x`B#l3$ksph5lFO{ECTVb3L~U_W=4EF#SCx%9XBlgsz?VV(<yB6gc*u*
z)T%7bpl;WwidoW*!CZ(?2MrEu2ukU8m<jsPKM*8kU9DE+I=|B5Pv$AG5y=PUyYS`-
zh`iK3z)7S4HH;Aet9A}p=!~*P<MPUYoe$(mE6DouGrd&BqLZ1t=k&qJIFWidJ@3#i
zT8g-z1JpIFGzK<9(|e`;>fr3-kibCni#{;!TZdylSWdlj=uX){LN-=d1JOAd<rdQf
zMUFJru1H-KUWy*oMNyKLDpUl#!+DWC^<pzM`VxpGUq9Y^&DG3*xE_7K`ZL59Dnqp<
zi14p+eXZiaWip=-&X{K$gIR6Aph^4mce=6Qa4F&@>#;SM?O{s>2xS&)V>2jDH=<EW
zM>8$&&|x;6aFa8qPe0yzl+Ed%5Z@%<ffo&4ZUkL{xQ-kTL&nd>p570y#boX}>^oHH
z`Bf7httOv9D^t9jpK(U&C3<oKtg8x$==FrE;8m@HvydNqIS7%&w}$KNn3K*|CTbg)
zsHg4Nr;+&{;47?xO1$Fm9!~A9W_MKhBHcE!(S$U?Dn^!+;kziFIQK2YwU0GYB*XgT
zs140O5nR4(RPki3(oVaQLdggRZ;f))L8*Sf-9d&0rMR{G)+BhyLt^29fz+nKP@a6+
zH42aAP^U9UdsJYqg73?}kNJWUbJ8&zc)2#D*3=4asd<YyF~&tHWNF`8KWdqEYa2?8
z2TXFo<)!nPM)lq$Ml1`?3*T9;X_E$!z`C7?RM#G?kYG);cKHHCfyBwRRfs^_udWk!
zSVQhp(saE#dU5Q<{TpL(mVXNnBT6yKV|WBAGdX$Y$2tL=QJ0AE4!3{}AS5@HOW2&q
zqNsj3Rc9){d?oK_EY5l?WCtwS*BTSS3chNf#2Iv?XpLjm&2?|0X3VdHm(S-$-9hc^
zL}I1BoF$sBd4}4wgi;-YO+GP9+2|C+8L~;-c7M3F^@4U9w;7_AC#dkLz2k^#26+0u
z&NI0Pecx7iiSkqZn)GIB2!Ts?emtC7aGZf~d}j}#NA#+sWrFlr8q-*L)8WDO)#pPU
zE$%PF$QrzLY{(-T56~p1QrwXY1|5E|6}?Lt#Xt!zk3F4nXxWz9VRY|$31oa*@(jk$
zurki(ooW)w^pEYGzh5}juDpym*Szs45?dNi8O03vz)y8d2n)UcWejClOjCPb{`@6x
ze<Q#vj+k+H$?9jnog@;8sE;fC^sK-DghVZ{75`+>z6_U!aKCTy+4oFh+Qae=+dTXH
zN<%4Nz39RU+Ec`Kh1WL@pY3<DNAKbF@hPvDM~+qjs)X<r+#G&G1`?pBv@T058P$gd
zW0JLAG(I7{>)JD~DH3Ibt~@Pprfwhq?VE*?(49fR4Xc%)tT;OR`9jaW{A)^Y$t*02
zCv`&q$pTu`q78Y_xen*uv(^hCS@&8m=%U7k?dDhJz7mAvbbic$t)Ul{%8l_qQ_*f_
z8^S#_g8C05IV!=FTy^fnMGhaED=sn|u19>=mtg6RV>W+89Rh!UIS1i*o1>&@{ypP9
zCjr}nZg;V(g{k7*N>}!)k@ZNk6RuRrv9tM#6lh5S^6){zYr>2JvUw*XR!bd!blwwO
zUH;ryE?F?|_73qJBy#&zAv_C}#P$O=g_BVj21#CA^cD1#OrwG9yNK#kpWn)HP|1%w
zI}EZ^1K~O8my4^5!Y2pohE)(At}V>XD-gBUH34}=N2ws<KG3#rr-At!@wG$=mQQ3u
zS4t@<J(+Yf<TkCrt>H62^KcF~CJKUof4V&}nXzZbAa%Q^b%CdjOdmxqJ@mQ)RiT%^
zuAHBP_p07{;1RLGa>x2_R`LBEL7={685_F>VfxrWA#1$+Me`F9_jJ>EW?-FW3&k@v
zc%dkRD(~CI7=lPJYLr`}UG4hx^3ls6WKz4XjrBr-K0g|7a^|FT##<X(TMZ~@t_3eI
zyu1*Q#vOOV<WqoSJ*dqO&RY%H8Z6O%KxxXW1BB#yxPm2JyB-*`EdwI0&SbR&UKFiK
z_vv8@FF(CLA;rbQ+qo#z0Ae|W4wTKo*$cz@PT<-Pl{n)`Mdn|VFgNLX?+{6}6nR8(
zm0kJpu+gQO;p1pV29c=fQfCBWJ@!J?ygtFlYW{=hfjiFo=U=0I;DXOA1m0dXSm&%r
z;Xo-w#lt<*f~*&F8Dl9$MSUvaaux5&u#_n^*lBKpDDP&kB=!{xyXd8QJ5H5cHZ}7{
z%S9#7cvSk*RHJG*Dq0@VA;2{KG40`{`-lXhvyQOf5iHSDq32iY7vrn&D{dnPszB(F
z&e5M6U+2H8-sBug?Y6o;xEy`|zo+cl>eX}Ni-jqTe<;y*mWZ56+de+^UTd}#x%N=4
z-+UQ^7#E?h=PBesR(}R7TG0=RE9}mDv(j+`q{;M0@6(T0T41a4tq}bHUN;0=mQys%
z<jfE=cAdXqOo)kU@!X}vl|>P9SXH?YkC@wyTM%ayiMLNDo%wZjiE&Btox`_&>4KO+
zwqCN2mU|?1g9CsXvZ|@TxtE}XH42O>*B-|p#Jug4i^voe%qQG!y?tzIdwj@3m7=Al
z*14n4Bg*gh0zUeVXCY0<Uu|dY&$OFTnsT~B+-r^Kvnk>3bvGTvD7~DW9ZY1l(1Li6
z{$+pF-J5RPgdR?%ZAta%;hgq~`{1_<Vy8wjcU;WD)1K_9mPf^)l3hdre+ZmB=TUQA
z@wou9c;VW-I;g*gN=MNXhNWNEfd98DLzWW`)V~ZoFzA#GN{qay;=%JMU50{OFrWm)
z<07>_v|DqZM1^iQYYrm5*_|l<-PgcF93;VrJiS#Y<g9_Rn5`_s4e`XyXfB~-NiT$h
zk;bZ|M%fXio{X^wnjU%fz!d~_X+q2#=nk?bQ_o9EiI+b(3iTV=2p<SyL(3qFpQP}~
z_(jd|{|MeQ(Rl$6NNai@H-Zuat`BA1g~j=)eu%)e3O!eQxE|bWBd=ZWx-^G1sAEN^
z|2(Z1J#$GuxDQfT?W~6@T9?T#5ZWI-;AL=lB*G}dV~Z{}7oc{nUq+7yHkZ3B7?fuz
zfACx}t1)0RVq1=3qNx=K%YF)jQ3k}rK5bEfhujmSw=@H|Uy;eW-Q0}mdw2^nyH1Bc
zirxI}J?K7;KHHX!Vk32Df~~Esh@-9R(W<T5S2@KWi(h`m+CqGRT?rub{7mPxV*9J`
zitUJxJk$dL>Pi-MQbkppWdJ`FTg1hIpLbp81|T7yfc`TY&JYIub=b)R02?Vcx$wiC
zyCk)fyg`66Vd@-cwy#TiNcAr=@a&C#8H>TT^(Y@D=l!Q|ja?wv>VlUQ(Os|(-+R^N
z9=IHt!E&4+Zf~jFo4psYMWR2%Z#LlXq^Te%5yd7i$*1YW7e|Mms3VHIS1DY=qEvw}
zZ)iW)*&bMTQ(sIV#<EsP*s<2{MLH4x@<JEfNptP5s7Xa<QHNQ?2*73d$X#1p!7rxu
z6L?zcyKCsyN$2Lfvg~$~+swE1LIFPxGnRgdO4jnxD!(2|g+M0Gdg2P*RgC{Ag04X8
z5w=^Pre4sPSqF0PDjuS9PJ?i))1S%4ax~9?W^1`I4A>A3&)$zye<^r@NEz?mOXhdG
z1@Li=1-)sFKNr^sveN*U^llGRH?$5_VNKSPQ&xWg<G!@FCK9}u9Y&c#Y-CIEPsdu@
zZerUW8lPc~I{5tZXh>9B{u}CDE*b|VCDv3#n3ybw2{TDutmkNo2(@DmE`!_SR#+Yn
zSn;X#BR_5RJZJjiSUE2{v7<suYr{e7fF>xucq}@BHUU~NCUbcPOGV%1#jE&v`-1Hg
zokd_#A(}vlx#COU<!u(t#6jgU2OprPs>52pEZ|?tVdxOzM|(mBYr16Q#vWA65#0Mz
z4cHh5Rum62(Mji8#8Z8?9)I##H=*lIc@z)A0Q^NeScIwWS$Nq`E;=*TY?7-4C$ZtX
zRcjCWg#Hf-IlP4m9uyTrPAe!gU7l^d#wq%W$VUuYd)l3akr<@!*{q<0QUmD2t)QiB
z^E@!GSaPXcS;zozjr7Kn45B@Et3U@P%Aji$`M&zsOZ<R9h$h3!abv5#qlXNF^pJ4>
zO2?wa4bw?J&U7%pTg-1Kz%idOkn`SL0{dK$bW@Xl>SO1&D@Tfh$CydVbz+o&WiTGq
zaSDb^s67sdAUZKmLpRz1xPrGDK{C=GE1`K13FAN}Ccp8#fJk4Gn&EhbE=+WmOlgh$
zXQv|3D?rUiaEw=@4cqu#`~hmudy4r<vpL{rLc?z<9LX|MbSj;wwhGDv@uLQYMSpAs
zO^}=@xa&Q7iPnKO*WiJ!!oj3Nv2s-r)_NqgG1=pCD3H+T&DbGBy-=Wm733bkXopS2
zh=PmRHp)k&pYsCNiF>eiK3f(tllyJfbWm44A#lc|y__X{-_bLz*wdPZgc;$r7Slx1
zxJyS5BiuDKqLH2vSP0%dvfM8FHdCwpFtrWsR%Cb)JDRmqwi-n&nU%<k&un+Yankki
z|EfL2E@lg!7XnH%HpGp3_yA;3rVO=OL@`1-yhvU*!jG3kkJBAMX6nlubRr^O)8o_g
zMv&o@vv>!M(6rQ*C^J8kUe8Q~xP0!8am@eFES@?>Ne1topXpxwZiuz}(3mZHqw3y`
zImQYER9@Y?s>*IEgqspTv@+_l^sdAJka)Q&H&;+La?NKYg25QYa=0Kx4LgNH18ZER
zkS~%kc7s|`uFZ8(p=;_?lxl4C8E*2y@g+o!e$Yh>qV6xSvRQ$|kizVl4GL$DlXZSD
zerEt7BJ|fE+d0p51p9?X$jy&YCCD5)2ZPr2UBCXa6H$^}MbtuoXHkt<LTVE=xL1-+
zM-_}wL-`4klioYOzDnU-f6Rvlr@$o>LIeWpFae@w2yGBsBHw-1pw<QZ)E#Q-ed9^7
zJH(+6mXkS*#--qOr=G_t%8&?$&>aFrh#0oNBQl23bc~I14jTY6&B;O0Ki`W-10ddQ
zjyY(@H2`YMi3E9}lIZ`%j^^ZfEPx>kBr>G8oC>t&U#_1${g|frD5FNQJF#HN)De2}
zZv&S(J5xg*!KR!_VrM}?>3C;%w}PqTC568WA~gXkn%shG|B+mH)JjHwhzd7i2I7!Y
z)J=>tn4T@0NWd1(*JS<xDXn?B1Qkd!7K2*wzOBhZuPI^UcqI>=;wPl_+&HTz8GUH{
zupHTK&=0FWLl}lRTn5LCbEAg=IPiug<kDvsEm9;8ps%E#?Egh8_;o0=G=jpQM3nIZ
zCdo@A%GeAYQ9zkPuLD3gxFFQqMop#F2<o}=2xY^ly<yk!pE`KntwIm;*@l!aGg%St
zFw74uFG6Xn0`0SUWCb+<x97F)vz)VDXCYaN`28?+flVS1i?TV>3|f6CLcWgTU{sXH
z>_=qgX8z5IVB2!7c&eBMd8Y`>c~)*tAJEVh3d!8TB@Q>PhT$v*V~N;uXUBW?;*9d3
zFv|hCRfB;j>rtO;DV_ugqUmu)Vh!{NB79Sn{^|@seuOy~2p|tLJH$|^ZvdswtnM0M
zP9T%L6___u#g7q)T}2ccLbkr5SrU`q9c5=Hoe=SkW=$r8)1jb1v(;*s@wsC8&J7(L
zWcT@v=dLO`bQM5TSwSod0Vk%00;v^_1_lPM5CkD2Ontet?9AIg=g=TM)$M_qTAZRV
zhCP}A!uO-j)lVxgX*?x4Gwc{*S%pG2E4a}b!2H;fC2C8>c8z$j0+$QPXS{wThPaAk
zfXkvzX35Q#&AKuu8K8NGm3qSMsw6NFH}B1;9=a|gvp#aY1QSbBu6_%-f*7+2dR!QJ
zF_3q-xpo)^lI@M`O$z9PqIqZ}V{zFzeGPmi5M7UlP%o7W^dONun*2i440B!(=D-C4
zl}L~p4KD)0H`*qAXJ7nG8>j~P%&lgLx|HCAOxr!q%42r<?AvkTeM#h_{&*6{@1b3q
zoRefNqK-%E!#)MC22J#{6YNam5vHy?6uZpukXfSRbO-l9;F?@zbZsOZ*8pSo18IcP
zaDgTy8=A5eA#oo4zvm{<;UaCr)QMq9bt?d2u%IycRG4O=k{yZ0kCB<l0wZvU7U`!g
zh$56+E-ei{uzk?r(IF@5hb4DF6*RQVO&90>q$Vu@_2S!TAR)^Sdib)a6Wt1~OzIw@
z*hDKtHc16oM<d4)jv|iEVV<6)L(z%8P??4U#p^#1_W<9mGc(NsqSD|@TG5F=*(56k
zmp@%1j1Gv;rCu&}LNiW-;!Lx8F0QZ;<Ph1vU2u&x@P=|w5T+?R;9&cyilu3nOcn&)
z^?mRWhJtod#9_R`$WTC)YDbj6@+cjAFZO#1Ht5NR?0}PiBORf#C3h=%VaZE_V5bje
zEf{rgdTvX58ks$r1(i!25es|0&k`MI%6AuaAJN&uR82d$-0Rx~&-=H%HCnX*?~{Jc
z_DF<_n6lQvZ2Na$LzX5TA~29b8I!Ij8Xd=<kV6<OD$--{$tz&1Si?0`WKA9i^5y*I
zB5y8B$OM{b$R?-4ABiR-Ee2_>l2QTg(+ld65cM+wRtwE5ZcuV}pb`zA^H80wP7Xzm
z0F*tZ=gYIx2=ex~JdLUL8d<DpO#PfXsgnIQJ{x~~-cbw1I|sYoJL<$sjXxbh^BnXt
z^iy+Pdl!i&B39fALtA+3y^wcO_W}fsT)7UgmIS)6$Pw<Qox7H4*R;bNygIBKi;?0&
z=%#sD(QRl8^$(@Re?-f9GzrwVYAR?#A&8h}2I|SSZ3Kc-@{xVH6l&nBi7^^MuY1xb
zo}CwImdN|k*BA)EK?ca$!q|$nuwNe%OC*OBa|Wov42HX%+D{G5&w?O+$Yu^Z*~e>u
z0ihnNZID{cQrdnL;sArzDDOl{?iZR9mFXKy6RhTkJg}kD_^qen<o!yq$QK!^nApeW
zSAGNH^Qk!0Wn1ChEETHQ2D7YgmusE)#qw-i>$d*3-Jq?RW-hZ>Imgv;Zh%*IWd+>R
zbZ+}llS`aQAcM}}90W9O1-bR82(=X($+Ylpj)ap;Ym^3OZ{^^LMbezY;KkCh{7$DJ
z1n;aDUiwRAut6-b@y()MPmS||aw=LwA`XqgacojFKz67E@y~&sBtAWdP2cBHv0z^A
zLIVkMtDVI`Up3AZp*+s#M{lv#$KDA_)OP7N)Ic<VPl_5|87B<VQ6z`3Gy#`T%4a@E
zJ3o#VQWIhysBrDm1PXtfpun4JMFnf7?^H7E+OL@EZtFOyh+>IQcb9vGyG=el`t3V|
z>h1U35ssI9Yk=l59=Rx)FdbGP9b;UxkfeAN+;P-{Xs3HW2Pd{uHk?0?yA<(`k|6<n
zb(MUWr?%1P%=D$%b3PfYJkxP$-r(irQ{&}G-)r$jZ%0)UdhVHn|EO<rur_sNc#uK>
zMMx>nPj3jppMt3DQ;jJOlV#tbOEUKqJGG{ee=~WG2nj?~<@bz8^&IGM)VU}?RQPX8
zJHTaABB7BBwbvvQjk+*zOom1Rf4cw<5uLbQ>_mF*_tH?KmUG-p^h9CAD{#p)0DuHF
z!zmKkX=tn-=!;(#+>zg53N?5)e?v<%UW1(-<2{4O5T%0&k-1Z%*rpU>X=oUQ+m3!M
z{n@eZDMvv#KaA>bZ6CLpr(@6ud>919RUMm2W@9I^9AQ+_ovI?VT|S~TqaIu`fzz-p
zA2TrTed_||_8r8C0jl*W)P%(C+ZK?n)j@_G8jN_=m<S?>F|<(9U&@}tEly4x#KmMB
z5&B@7zeh3%Q@=wxN^EMU)&uh5J0HD)P4@NwN5C9FQWD4`yLdM^RSoDod3Dr*9Rc{R
zJLL~1mj8;$na4EIfr^mTqqCj8%W`nSFsBBHjXk|0+)tbM))L(T3CJ>cO4QWiTx&&L
zK=s?mp0tNnc1T=7GVhIJh{>4;R9%AVJGwK6iXMQIzhO=)7*HN6%J*B-#AhH_25~kh
zPp`h}J^D5AcLp0hW<Bpq8-e=MsVJbf;Rz_FaNPK^(`amPK5B$S&|pHtnn?xCwQrN4
zVCK{mM^g$2mC`_;Dr6sSvV>7FNBt)xIuQycl_*zoi_;jeQ8>}?<c*`TVq~i)FD#~h
z9NO=KuF*~8c*bXL#n?Ro!GJ}V=kZGNv<+znCW>1{8t+ENn*A=+y+TGq+MG1D=(@6(
zE2VcrJoF%`kR`RBINbd%0r>!4Rq(8Aq^We)K)(dv=OA?TL4phh8NewzG(`Oab13%s
zVB-lQUpfK<H)d%DV-YHmY?IB3T&c;M#sg8>p&6MZh84uH{SMc_3J*TZo9xddbXO5+
zP0yd8DJ=)C)QZZZwNM+x6Yj$I6&Q$$mu96hCwE{l3TWk0c5*IMh49n>jU?en0U`RR
zVV^SxS-=A3Aj8uT@CxaauEG}(OmmbB?9}Q(0xU=4Eile5ua4PPp^aeHDyR*g`wkWj
z=D1Ys31*K7`UXI8bI|3{OlcZ)j#4;)56(4e8qai%Gd)eT0<A7hD4<!q<nLp4`M^c8
z(=>l4%-DY?^_#rkFlq^q86?-SQ;w#t=k_^+MF-V*WxSj^f=3lh>kt-cI!x7Mxd~zc
zdCrJooj_$<MRgS*6m#wh*$D|!0Ugs6Y$`_ygH++zb5mU<#i#k~8RiO;eou1`iIVPv
zkR47$*vZBOeww6300DMA4jPU_Vh_26Xx;{?j=YW3qDLl&soVl=WwLPhDPXAN`OyS0
zrUaKOVB(xeS6vdN9YLYRDU1Fz&Fpc(nLLqueX9`J>zO0e=-|?{zJEFXwp*6_l6X!7
z!~X@{*e80|1=^mbsb47FYrs}=W4M7oO!i9TSj7F-DH!=P%@}682-qh&I_JI7mMGdn
zqfM_PL2%blA<DjDzP+8o9q6;t>1OQ-WaUcK3|7w~w$e$ujcyX!Z&Vx8v>+Os!T{4`
zTYA~}tJIx;4^Hmwqgz<!5_^}%JnXs^odI_!@;hA&ecDPf{ivCX`q&KZchGpL*V#LT
zhxM9hKq+%cbQfrVJ28r)C)Dh&%i8G^)T?cwDV;S`z7T1AnX5Ad99lI*-N{@UZA9MI
zVoqtT(LXjh1yT!3vXmBb7))63*OE?gab^KNj^Nzm7b0e{n`lQRl%Rp)oz=qkr)xp#
zCF!wk`$hb^_`|&~FMt7k%d}5!L^kXtUQug1+D%y6mmQUQ?bfoenBGQKoVVDp5q1RF
zM3OkAlVoY;;Cw#u1QH1bH^B@J`AjUPMCx=5eQl$TvN`=&?KZ0cXaRiGc@*p{BfDsV
z3>4GGmU299GQ?^BD6I8_U~Vo=Zo`~S!_}|v@8zXtp;4sADI@CUVmW`|HtYIZ+y|S!
z_UeIOXSu&m5hWyTXRnTN94TtH7Clk<Di@1PgnUE?9UQXqX^RTy^H}<QJ`xIqrn^8h
zim2a0^aOqKPd~R_sFTY(9<0TDo`LtTuSPyO2qmH)(X63G1zuq4qQQ0Mr%^UTGK}n?
zU344aQq-n$rVk7#1LcEA-jNLbt<>A0UrR09bj{ZsPyb)+y$Lked;2b|QM*YhN{O8&
zQv*fFRFOz#A!DYJp^zz4a}t%1gfx&b8A^#rnMF~VhcZ@32$|pO{<Zh>-{)QDS?7J%
zIcL4^S?l?&=h=Ike!uVc^Zg9>eO=dm-MLGH8EEJtR`t7&m*CqiRlv4TX%QVUJK6+$
zj$%DgagOEUF+r(^-1`w;764W`SVwiNOxLD8`INeE<4puZpd6WaC`3V<mkd<3@C58_
zlZH+uqf+G$`qfjW4;Zoc%NukL|9J5(K|*@U)ENr24UK<1qDdIBxPEmfYRwS!)FXZg
z(0bkV?(S}@m^|mbjA39h=}P<WyvaY_l&NGUL>~Wmof6J9<^RXmOm~$3d)M*L-}uj4
z{MTswpF0-(LGVFaT81c(ln<EIOia;jfH`=3mSR#q{Dh@)8H*;w8D`W$*J1q~^1I>C
z>y(>wO?HUDo1#($Di-|iVk+kjX8aUi?g(;-CYPoEdcCYN0XhDUm)d$zxc~T?Dd-Q{
z8vgOJI79o+f4yWw)o2y|sRs_d1s+=bJ=>Vl^#}kUn}1bzh&V{G98HCoOtKy1$BbBs
zmwwBNLXwb}hxl;?!Uc_@*(PI%nf-rXeY%w%l0)@6Mf}-w%Ab3g{V5A4JnQ#P1G`d6
z9^O|`4-Rle`&KhNK1>wUlW1S~$E#)q7VJO1{?A(b$6)-QyHNh~Q2&06j1%eaFX-R@
zj~?o`soKneKLrx3?Q0UkF4z@Va%szMxhI=9b5ZW4KljY!kuzT`v{-(Tdf&;`pZoZO
zkG%5<_L*_?rC9#7OAc#P)P?4XG2LT5&J>V)XR)05b-v@-GT&o#qTh;3Ri23KjH}Zw
zuC?xY(eyqdzT8Xix1-+Qm-s(!Z1tZ`XpFJ?9}m<yM!ek1gHmemsY!~rqx{EEkGcdh
z>)Or%VlQAzcX#L{v@@h%kl)A%Q%)_G0Mu><tNio6R4g+*A}AvxvjPsovEw=8?VHIO
zfafE)Y$hfxFY$Im{2N`KVAoIbZPU(UAeE3`T@(F|J->1<<IGS~K{U*YAa0}cB=Sa6
z<b+nO;C+cU+B76hZzbCvY}kZfmX*sxu8u;tgg3(+zy(8l8HbE)t{0k*?J+Vk3hmtr
z5~ZlP&qI&?&cv9UU&Q_->;;7V!3XZ<78c}+gcq*@GUZD+KyHjkGaTNe=)0$U6)Cb~
zX88Sm9i>WVs1$;~a?uZ&q7GCY&ZSNbfPnt`o-D(X`_zmAfOS+RXpx@%lwo+z6qJKE
z--AI6eJBGm4QCF2+8|x;Dc}9SKLdBW2No!$<~=~emtzb=6B!;$vY*N28hK@@%>)R}
zklH%>aM^rvO*u8A1Vd0_-Ut#;0ZdPcxh?&W$@f8G`I)N_J5f&j*ZXKKWZeE8s*C9F
z?^p6{OnolILcc9r0dgoond_7YO7S#|8OL6%0eP$(mqiAlf{ODPt?D<WrI#aN_8a{i
zln-9{IslwV_Ys#A3IkCPmK#5cFz%>CfpJISm0p*%z}&2nk;&1jrm&vrj=9p9;4=VB
z709PG41a%;2ji1o^7VAYRA#UKeMQB=&I5sICXBz?^84q-_<T@wFSM%S|GuUw#<RQg
z6VEaD;0!G&8A(PQoytysr<Os__R}e?gWKPM>|2V~A%FdZDcTzNZFH&%^b&+>|KI-L
z1WGePO8k2(kCy!P{;SXrgjx(Ne$*;RMg2^?g5>+ed5u&tU_N*&RzL~h5B%|lhCKb8
zDd_xl(OrK%w|#Mlv2gFH)G+Pkw1lYb?=HfGg9z~Gqx2Id-g%gFSHV7*b`GP<O8>yX
zz7(uQ{EhHme?t|R$cRllLmqQ5Y(-C{RwH4dMgf;2djGnpxr~cSCSe&$I`T`Ts9zZ%
z<7flMYGtyx`)q+a&2zv+4}%gItA&3n<B#6?y3g<ghXe*dyW}H%G?N-TS%tttUA}w_
zji6Ql;@o`Jmws}pgj~#-m>-kO6Tz5TTNe2Z?i8UP;?^NTm&efhhU+viJ1{;abRXjn
zvhIoQS>^OWogy{icuJjMVy5<?AeLn7SG|7T|LdEviN7EDN~~%;TQ)w%f;;7F0cas0
zBt+-SFVN?5DCXffz?fEAdl*yu)04kec5F>zBI60+1K8BaB><K~=dW;s_vAPLjXt)c
z1N0ZMi+~g!h<!$~9YE|Q>5PZVR3dli?HN?dasUPtXk|R2Qzd)p4%lANj<aDyjG8^Y
z3zL>ILC2WO5B#Yzm<(@Zw0>Wb=<g3$vE7c7s>w@k{$ADZCixi7VW%`gLPD0!m_2{*
z%Ff(_^j@;|G+*z$+Uw$&1;~gbViQ`wUDhmCH|qT$wE!U88r9#AC)gjg$xQry=<p(s
zy{^(tm4~r`#{NPo$wnsq5;9oM6S|DIP^)cQy8a7kOh9m?C0e6{;bX$M0THtdjYvX1
zLUKQ{wh`CD+f2n^jS>Uhzv;fNG)jGl(<;e?F^dm?k9^$l+;H-Q2i$f@Mh7?Cj0);Z
z0xKB5<hz<204rxnfq^84-U;>JtNOesFe$Rja|=-yBm@gcUyut2-8~a?mZpmMI-(c*
zH^Dl}v}jW=WA!n9NdFjOrkVO*um3bY{XG`{8HfM5aR`0~fgNC*wX0A(w6S+Fix~Wg
zlt&@}q$a!>$_D<Z`MnYPy{HVYD*B{r8Bha+^n3h39w|V-zE1q=#yBts(GBhZ?G+s-
z5^2x7^OFo~7rK(%#qDhJ+SWT~7xMR(GIpO&Q%@n|7Z#t#1?0}nm~lKXG0*v*p5N!f
z`#0gEd9z7;y9xb%Xer@!N_ZR9_aWx!6{fH6Xi(2!t+_b%FGJIGFRPTk%XDWxymz;N
z=G1GlV0=UEv_*&uZYEfsJgUPT%P&zgE^2dD^TOT5rYTFD=xr!qJ;~v~%8p1GyaPD|
zU_152trx&mw3QmDkamXTD#Gq9=6EwxNPQW_S?DrLY;y|6UKjoZ3=47)lSQSa0)5hS
zsFIZ_{4zO+ra|PHk2f@f_=DlX4?4Gn9B%MCa*&wq%W9@g5<TW@1<{g)_AP`z3Z%+F
z?!S^W*60-zpXa`oEJp~nrqGpSob+oRz^wbgP2`b&2#g(H+KbYX?T@TbN`|BV3e}bi
z!iS&A>`60qsfwzZ$a#RFZJ@&$gfqDixB{--XW>aB$h)7q!H|h1>6#gz0=$nfZ0gSl
z#S9mpBUBVCh>AhgZ2?#bQ-B>*|LFQHE{5FPsbE$(VeD_a77X~$hmP-2g<A;rij#_Y
z`?!dpAPXIcRL6*_CyxLkF7VdA<Sjwflhg){bW7A<XmPL3?amZZ5#ih}Ae^4C7CZtf
zurUM5-iKph4~X4+7%q!+Z@pjNZV`9p3YniNVEh#5G{~l%O#K=3Yx$;08Ij_$GHNSG
z@HnHNBZB_|dOs+f`(VePhY)tn_U%uXK4KhzbFcB91Cu1d%6j3@a{RAdnEXqIdV_Je
z+$e4$WpG9>$^t`teB?;7K+H)=4LHhm<WS%_M0OWTh~}xYHh?~yq?qU9X#w(89|bAT
z!?PGi^L~cpvI(_ddw+g!%+zf@gwbWV{F*{OV>LWdf)Qsp#@>S}UcTuOmhmO3%}Bus
z9ORB-j}1boX_bTGdkJd*4PnVIuvO8Gq&uTFc6q@6D~($5i%UpZE+^LlT5&%|J@$*m
zQ;9h{Fn+<p4Bs0$IpVo40jyr{{JkM`=oVGj?DAA?<^&B+MU-nrZR>)ZCor_@vWH*v
z2LIZzb^hxu^I!9J3E|Xy>|M_p3pPvRdvhdXyh5eY?HX_O_Xa~@8dRK(+sAn6bXtJr
zQzirJePb>kQLB0lD`~8X3uK`Z^)<8o4$yB2Dv<fK8d1EqFQZE5a1D^|eN<80(*&n#
z6`CaIGi=z_n+zcB5-Dge;Z36&I|V>D?5B63_>HGmhpRKCnUhq5ARl;onn-4N56a<7
zK2tA3u9aYkDUGLXIzq}tt0p3w9<@@VpFBL^N~8B+!$PTtO@Oy0#Yfb3*s9tX>V@j5
zmAu6+mNX|Ws8_X8*R8Rq2x6GmAH#;qdzW$$q8{ZjyEO{5QGads{Obequ+N@MMD|KD
zC~C-7`JJjvXgPqdOV1NgkTU?`P7Y*wP$-6r4nb?-5AIWxLDdhYp^XKq9j;RMJ7T1H
zq&(9QmQsZ{Ns!)4=MRQa9!mN)><14#19h<7VXm*I&KyS|SiU>F_W^P~hJXoOhRMLn
zsueZ193WTuD+hay5n~IDOdd7g!RfMZr3_$JV!nY_t~h^C26GBYAUkaapj@K-p^<Q>
zjf`5??0U!up`yRiRWqbeFyDEcLbbreuaPSgD=Y5PA7Yt8SU(4dKE+JxlbITjp>nas
z)tsy(o}#MdyG@hS`GmM+;=ANej=dvsE5&0|!&8&S5dG!6bHl6z<$XAhSW9ew8&4Ry
zMPh^7A<9sI{e0KqIau+x7GU@L0+Pv4)jV~O+-K37s)po{EYw;EXvQuO<4@LVSulLJ
z>b>mo9*$<F1RI=OSxdHARQGHEr?c}$jOHP}yAW*4{y{u48Tb^Hm2EsxPFB(~6JJ%S
z{Q}8-NYOYiq8=-p*zE9hbrkb#&K5%zux^wSH6=lH`c+b>`<FrFKr$zibwgOi=NtiD
z=@r^S0gqQw1LCbls0xBJ+3tWL%9u&&c>ZXIVmlOZzKrUr80_nPVD<}9J3X<u$B>=$
zn~>&FpKEr$UTVKU3+22o`O~+8?cG>8N|_d!9I7qwp|e2kAo1g59FJypS?|P&NoLfA
zA*CSaL@w~a{3q-%G0doy5SEXr)|fYOy5yrb_v7)O`{`+qBNyY>>c%j)qPX)4W)53S
zp3kr5t{0$yTOWkk@oRaw-7dUBjxb=ocp@I5vZWC7@v;Zka~Wh)`N#rRQZL7Vewxpq
z;*O#GeVuJLZl4S-<+C@CCsID(p8=kajGW^9U0@A&W};Nx@XOO59`8xz2MO5mfLqDg
z4n$kyN{rgFj`4qAvW3<W=_0W)KJd%Lfoev%XkBY2PDpn8u)38E6fwZbFGbg#G)U0z
z>LoV#_)xcF+>Ul)1NJEHYun#Zj~Jfq4?Nqceec*+CA6F{j@nS>MTM~~hh@g}QW)9g
z9WgWtSwjB)0RyB5p&=$&PEiK<zpH409be?ud;-3EY+p|sHA4pyOB#@pn*bA}CSh^H
z{&nMlIGTO-)cDAc@9jVhE}lIj;9D4(R+;u19)P%?iv~Z%0So8WkgYYTK1lGBKUw~}
zp)`ejvi!*TPAMHiJcDr3cN;a)bV#6eNNr4(?Z}ag%!3Jh4_OrimJdU_PDfdgwIuR%
zhT}ZoA!?tE#lMu?bzsmny_Gsbl1RMrslmjLi!y$wux6X&QBue3Z-s(PAmjy$7N7*h
zk=beJiA83OLuT?c-?Z4FmT(Ptdj_SRqYeq%vAaOEt<gPB(Go&1rIdU3Lg5~Tj<9TE
z)yOu{rIA@fRD-;=Xywb{HmL76K{bqCGq5RDbfoZy-0J3r3-k`ukP5%(*hqYLrAT@s
zL|XxokvWm^G!~~`RUpJvn@fUM%8}9Tg8GV+W{32RyV}3wY_&sLNpHr3<^y=<%Mj`)
zg}%qK(r~NnB*h4|Y$pve@TwNzi&x30DFvZ(kQ=pe!GYO5;CA4^0n$YAIYRj}PSR$y
z6WjnvIdw<yp`DUUc7l%ak2VcwFN47h@A2=9`J`2f2O|$nc}{i}6lwX4MoH<m`XyN>
zfC{iv2F-yBy7dc1DO^-8gW*Muh&-9QK&@|h@5M0u;#g_7M${mW=W!pjx6q+i=csB<
zvjg7{_}YL)^*%e}D3AoY3^Dk^)6J4fnbal}bLKw>m5)$yL8SMj8>5E;07{(Nml3cO
zlJi^ymFQ4!8#*SahWQ!9xo8Up^#w8Wm8oXfGpNs+SC=1E-uoj=j()=^<69}3_%ef<
zz+0GAP>B>FnAmp&smK9V3s)$hrl66bFupC+(a}*r-8nxQU<aWcoiZKa4m=4zV|Y!H
z#EC8p-&Mf4k$uU!I(Kh3I>B)6t3!86^gE>hJ-tPc)!h8K^9aJCD+p1d-ZVe4)*)#c
z^)LYzM>;DpMCfLpU!Bi?L6)yfdpE1kfvWQa4-X^o*-~PGBBYEh-l<D#;qG|LxtGM(
zhMnj<paS`abz}n>hN(GGlgLOsoQ6D<`UeU3qP}+T!C~y5mETG5yaDQlK(9^!8v8@1
zGMgvtBC|+jWl@fOFl=atHwNx{2Xz=4%Oh(SiuHrW$`R#|Q_%Eb(%H3O1e8M~_pY5g
z&Y>Vjt(*3T!bgOmYn}VX7O4UCJK2}%f{@vFBFzr*=V>=uVwuN~ScU`BnAJf>W_$w#
zcR%jH#EX7o2t631xAvB&@J}EM`{;)|KK?84A^Fz%QT;C2$Ecogn3`k^Nl|yecY{>1
zB&_JiW0?NidL)2TDPk8r6xjg5S*n^79qF^aOCiJya>K3cKS&o<`8^Z;q~#Ft&3nUe
z+(AsvH(-o-$g`h9)S7Hh_sVdnUN>NuzR|xU2W@PwcOd@QfeC-v)1hQQ(i!ged;{sn
zns1ZC%jpahFIah(gIyJ}s*Ez7N_-1oDJlb5nsVkaPWpLoY%weArVKIbQ(iV864+$X
zgc-er5~9-?xZC~YKy1~eG8n<Z$N+){NgG}{`C>ZL%hSbSfajEspgH%f?BcH|W>5@+
zvs43RO~WG|$p8AVCSu-5p#j2N<2ILZoS#=nD>d}JQUB98RlL&WR@1y58lg66`$8RY
zs6{{QAw#gr3@mtgq^Q}0dQe^>E7@*UPHY2e6jM;}Grf~a*~uTe?)h8|YyxEESppli
z0*3WuIK=dCTTl_gTh%nia7ClGI;R~`JGXb_Ml3B4EdJq<ELyLSo(7FwK@k(xe@<r@
zPg8TM5)a0rpGV~x5Z-kUj9~l%e@b=<AO^pJ>e9zAGjODdS=cF-pIO7>g?+3LhqlqD
z25e4roqPkCsKJ5A>sQrbnbC@?)TF?woK9w(FiIJA@oArcI6`<d{8XB@YJk%Z!gFvQ
zw?f@6aNG_LR|r%KHE2m0-n<W$V}7uQjeSXvCLckSM#qM+Rr`k{$S#4@q|-YgYh4As
zvfb%nW4iV`Q3TQky(IHP6!Y*Ow?n3^xY9LS7{*soj*tPVlr&)5|2b<tLrRJ>Rii*T
zLPEqUO<OLH#p}me&s#T<dk1fWJL(|AOX#63vH-r$oJ>Z=i0~v_hWC%M%*u9v5?vlE
zNx$C3si#f^TI1^{CJ9wQ3>Rb2O15!$0y#zQy)1FJnjw!qJ$1Y-Z0_i)9y~Xg^sFtT
z`t-V$1GKx4zn}$UzX?h|!{s*^jA<Y9#uHXvfwwhpwCnxAj5mcZ9G@AW2I^S3Um`Lv
zFimxdCWh2!Z{42*`!RErYL5>#N?~1yV)Bp(E=Y(hrb<SDr|<v(F>1r{fX*=+G5-Tj
zc){zIZ(!c&sxsur+~c2^t%mno(3t?AmLMS-O*Kx!i<PKXnu-K^PM8AE;2Q`OF_j<h
z_5Y|vlIrKrmXd@))DnM4a6;4o`Vmq+&oR#`Ox~zI$}gaMbEaI*!cc+JKJV$IJJNN{
zf*AeHJT2Ju7`u%BSW;!p4xSdIA$<K!7Clf<x*#%M$!AtM;kE4N4n*7+AK(x@tWsBy
zjuQ?Z1d1FC94VT>5WZ$<Mu)ODGS}XoSyLIQ?lmm9?V}znV=>tCoVW}Sokdv(<>Nbj
z<!1WOF{M@kYAB=(P`K9Jn}7|C@@1WZ&#(=SNnYy-BQtht(NR~Y(Rh&&JhUiH?sa84
zH8%>zgjUDILe8I1UD=r7c(^1(Yk}m~6-o&2HOgaBF?)99Q`k(PM$`bpy{XX^=+v?A
zgl0}M)PcVqP8orS2<ns1s5-H|H78-)u=Kkq_bx5DJhZIVEXU0r#O@d*em=v^C`gQ+
z1AoZr8O#N^`{A+db>h^U5tzsI@>RTtbK^XQ4SF*>A6wiNe0%0{yl~_!+8{BlZwMFY
zF}k}3`)_9~2|Nvr^4_{9IRBwA4(MO+rlL|i_~KZ7`a?B)vbruA!Kvas5s8PRfA7}Z
z1c3O8*1UV4Tk{@mIGyUEHwM?d_XXWeFEWd|s-Sdi_c-Leecnn2lNx<@?crQjRye&W
zOXyV4<^aCvvZ3Z7@p(-$I1`p4<y7SOO;nvK*~7@BqB3{hE<y69Q6AD}b^q(qp#>V)
zIrfU@KsNJeYGCr@R$c7YfmLePPE@OXcs2tG+>P&w&skZKYi;a*{xv&|@|~j*6MgL4
zXq0{_(XONs$*l7jYvN~_MRcy-kIByr9_o~G6tyLu@dO8?R!jEC&^_RQH%gcAXi8Ox
z-20g{C|jR-PiZ(}75oX#AB;gX7p=~SfV7TU*u#am4@y<G%m*zF>;@@f41L`v*$0ew
zglyB^%8VElB)a=0a!~6-?v&YswPb2kuo-T_m>5cu6nF3H<zOYx(#O?xSRtm3*EagU
zSi^&gOe{=+cqnJ1!Z&xjPy?ipJ>qmrYH@5Gf?_-q84WIPcfF7BC#oXva4sSh`MNa0
zw`PdH!uos6iSGpl7iAceEOKBr$hG|=RS9^C)O@Eh$q{{{f-D(q^CD_pz$9Er!bS=O
zI2@5($U8gt%Pd=lX!5Q9JL<Y^HI5n)GECik@b38^kkmg2=LYpl1;rL`(8-CLY9yM-
zAe_#nr~%6G;R43jzJlsDpixgMa@WzKLSraQH$=#c|9Jg=^6VKZz*`9VuW}QE{o#+G
zJ%^gP1qqSUD`=Os5mUtP<1cZ3#L+CmV$%M_GcQi^tr<>Q4w9~y+(hA(OuGm2lKEVc
z&*-CYxcE|Wy~P`(@EYaYQBG2h+R{ov1}n+KNZj`v1q?^kqlzni82VsnQZ<vPmdY&h
zEu#Ra577gu{Mq8lACc=7!?1(`aL*2^iU?ADnuvI<0q~#V%GXrbOlsYfZ?N;Lm!9(Z
zlu2rBa^*SgNSgT_-=E#r#kHYw=_*o7DT_DMVCnI+B*azJryxY$it&W`0FrI|>ZT8n
zsRa@T<<J~hMpCezGgu1*Pcu92D;ad-*yO23CuTNs9jCN1GNWJmswccJG|Q)2mgV?R
z&My`c5kd1&p^n*GTNEjainup96yeA+{A7T|L_$@SF%!n7!n=^5eU5Jk8?InO5TXf6
zl^>)^r>%>OI2flG)bQ)e@Pv-R+1WPoZ6tK0uDF4~4geO6CUv{2NuFB>De*Eov8h}P
z>mf_^4Z-3_gZZk(sp=2CG8Mqucj{N4EEB2D11>@>#d%NcZyqGUcaI2xa$ZAekz}_|
zATJ#wV(>MmT0M~63QOI4eZK+yP!dO=Kc)*YJY>54UA>7IQVJUgM|x~i>mZ{&mxh+j
z7|;+T>P<dS%y#M@4uW5_A4;JfcnE0A<sQC6L8@E13#<?dsZR${$y)w&T#=b6C;%#T
zQ9P)LtOKc(2oCVd5o|crXsQ*x94x3tn@SRB%&1(i5B+8fkXc=R;UX^rn570IRyl&f
zD<DT%;QWKt&3bss_x^?hdmg_ZGJ8T-YJUBm%-p$5pO_vakv46IL7A<6h5WK$2NAYH
z>da2{3jqVvH5LGOzH}t4xv8fk5SZJ%TQ-w5B$9Bd#WJPdA0+~hDCt;AsV3HsJPvJg
zi6z|&g`O=JWyTNP{g#1zp9=C7n;)~WGJ38-mT&xQp5PfmrM!kSo8=J(UN69a3R5Li
z$Y>MWM?5}?VJZHO5~5n}257p+Qo#A*kUE2!mn>a33Qp>`X@x!k`+_v)Hcux$coRWp
z(>tl$Q4Ba%p!7{^tQ6V{$t#*lU&#5jCe>o!ha|%t5bG>LyvArA3MaxA1({LxyG`cW
z;0MUV5pff>JPj;htWwyi=23;SMsp6&K`6T!Yz;8XC?P7aq8Oh_NGEg|jBcn3G3N5A
z*x<?4`g-|W7(T66z0HXgU|)F?0F68nMx}=ju$|m*Ska2qf(wuX`2lqgV~-~G2qE`%
z(MAdhn3yk6e2-{K{Zj)F2?iSlU5Wxg*waE6IUV*Qw@mr)jSrzN!GHjjII!*VhfE0E
z`(w#^sy8DYO4dF`eDR3Q<K7=kOtT>P7J|PJsnkoJ;n%3S!my_HR2LqsAIejf2@GQ^
zxr8fDsd1lzCAL2*<uY|Mkl>XTZQnu`Ru7azH4+;z?(AKB&*?&CKR05IQ^(k7zY!D!
ztbk>*j~q`*WITWKKn+FLz*>AgU%>fGKV>*67GaThK5o=^B7$1Y{w3BR@~~(x{V6sr
zHone;J@LqWHrpa5aHocpjE)1?h<x=e18l%sV)n72-!GpAHF;b{g-|$DLXJ$~xfdc*
z){P3s$t8!y+)BM3Dloywt+nkzEPjoW`BuuRIi7;{SggemW(2>+a%MPA2Mhpp$;pEm
z!aTMesElUV>>!xr>zB@_c|u;|#DmEnU8uBnF#|A10=Vz7i~Jxkn?jso#%;YeaCfv?
z*%{90>hn9;6@G_JhdS)^J<XYTsMbdz;5hKOO_gEiumv!hLJ~M(kGg>Z6v65x9?SuB
zL{%ZwYLjXogHp(E9rBGl?8WM_VX2gek-&`1VU{JNAFE3{tV;PeIh^F-n0{Y?)VYh2
zGFKSBDrLFIKfx_i*F^&$z~_%zQm;<3KA{3?0y9c#u=U<J5K2#0?GVOLkPuly@Q_@u
zvT6kkusFoPL6NE$0BCD|@uXylpyY83nxM1hM?55gx`{2K!fDNd8^dnh7;P?@$-b*(
z*si=1hicugVpXrj<Q5UvZKzHM6+KbVa;hsc0YZJj**1<7^aP`rKsL09kJJ&o45tdh
zF$pq+n8XKJ-X3Hp_YhD*%ELz0Wn^kWTobWZHqF^DbjQdwlFF7CZbSfXDllEXARyfa
zK!M^s0<Lig6%reX6`6vf5Mul2jTk8rrdeg+y6u42Fb4j>0N0gEbS3`6<#D23`S^#2
z`r8R$;`aQhK92hEtN7-YCUBsphMK|%DE*-}FQn=?^0*-PTAXzx!%6{sM(u1{$u^00
zMik3xM#UUIZ~!O+mf2$D*HnLB03z`|0wGe!)MU9ik_$7DKIEKehH87`P7y>U?a#?S
zsjHIK`<-31-5H<(3H3NC0Gz-<4sFN*6lllBNE2^M_efS8T&TODI{K8M5rpBbpql*Y
zoF%t?tN{h;<VvpKbQn>b#>0bgVyCPfG%TZF8ySNU)!~pM^)3+)&*-eTW^|O(tU&YK
zKWDRWaiWgekEApNP|{5^SXW|6u_k;^7zWqUnN7b$1~Q~3#84CV2jYFWdU!3Py&+i$
z0}5gNEACDW2<FYZ^cn4;iQuc{9>kqTiJ(%3(2LWKNDD3@9k0RFdL`im^v3zik4)u2
zlf}|f{KkDej&?(=eOWZF2a|JkYO;4qMS_}FxuT5<167{t1RZw?q**#i2JS3REzQJ-
zoVrKN3dV6TDoe<CPkQ+O*6FYrkRAYP!>{82eh4ib`>kr@pGV=4?0I4Pmb{{Jl}T?4
z$q`2Qc&BA0D%(j*@O2#bN>uZI`zINUwgf-}49QMetq<z7KAf?uDFlrYqWBMb#Ty4M
z;i4c!Q^Uzc9j*OQx%2#;IBD#tp1|S_P!f&GO8T85wcoPTYoSz3M#m2vDugKpb*0h8
z>Dz}mjzME%xdYzjro)m9GpQ<%Zo5Qe1I{!A;*F|0^o~rj5=hIbPvml#h%=TN5@>36
zK<kIr1%e?eiAE%~k=k|krA1B%Qg_!N>umIep@X9*Mu6VwPAXyOOhF0ou@Xbb+k&W9
zphAMWmJ`_&t%dUyujn3v#^JVcPtF?B($(cadNtx1#ZJmW45chn&<C7b48IJln_w3{
z#@_`6{Cy+<ycFowpl`;$Bv6Nmtqx`rY{2XNVSSVM`_N>%<44Fk?KEH%PP&DVJWF;~
zh$d-qnU*s4IjmUw(v45^8ZsP9Gf`}aS4k1RApbdr+ykfAUq9fV#+U>dzaCAEJ|ayR
zwRXye7qkLZdmuPd2kT$-_U%>7Ssc@}In+G1(nRI|h7?$kPyot^WdI)aXEP*sOqH(P
zDTNH40I9B*6zSXW22-Pw&oJDiSzx>YDj^vajU9maHp6)fRi24in;Bt&5@pLwYMVgN
zGxDD-Z1e4gj0l&>!W<z4qmCCcm2sTe8+#q-{{H;E2IxP3zfMj3ryC(-B>vNt@%M!I
ze>)B@rwW@q3+3&{r)GGs`qRhvPo<QVmxOM_*7(%;oA`^|$YZzmy;N{@MAvlNsJfY=
zAm6R!B9}I<$Sg3r=Oc17*#+&R<kWmCJBBaHd@pV6`QEdtTFmHeVJjDlor|;OkX=TE
z=i6>U*7kW>p}s!<?$tl5Nm*X*-@QhrAA7*_UoX`iyKeoh1^BP;(tq9aIK5qy_20eJ
zvtNeA{ktph;WhqLruFZxKrU$W4@>_4_QU@1Rmrjj!C-Zum7iN)xu?-bbg!0>SYc&l
zmD;s0h<6&w6joL-KDo%PG23%xadGib(uGE?L|s-cF0P@Z*|=o&Z1cp8-&$MGB0;T7
z-ov<(@-FDC_U_%A>>@&cT{KtFdFw||MxT0mY~#8a-<nAwK5_?(YL-<JOHu3p*zi2#
zlfrconI)qZ+g<0i9SBN4<lVKIE9iYkuBW6QLGfF%YoF=At_xzkMTk}tUL@?K-#PG7
zLGj-WB(C*;-$=S{W&5|E`@e7!{p0EVe>!pgbG!d~+W$X&y2sO8X8G700<kFO&^CwZ
z6Y_<8z$D!UYf_`pRemn6U!rs_Fi{W!_u&xi9M8ro9{=Ub5iry5OxzF^qCt?%jt@RQ
zHqpqm_U_+5iwR=h9Rw!Rn*{s$&Hve5yTsn!UZNEGi_>s1Xi|B?!{cCQ<FX{+$rkqT
z9FV@#P!}7K+Yb(k9L$_!Sy{zX@t5>Y*dgBOadC3`LGD!(cR<d3p5Ls_W1KRwa_0JY
z3tV8p{gyJVpv~b@?gv)Rf1k5uO{ep>4}36mnFYjGq7)5P3S@a%717A{%%1zX*mRt3
zM2b7khOf5*xTGFUp$)_1BaYq8Io!-FTzNn$fBg8o+uYpz<=eL`)22;B>bmgN>(}au
zx<c+c!Qr~*=4-D922NuFB$DcATXX}#&s6|39D=Bcc(&^3ZaMQ?pgdcVSxNN7e)yo}
z>h4ZHAoMoSzFc|vQr2BZo4LMG&U_>8#F&qd5AB4OBc#8dmW4EC&ccQ1nh$-70ucJH
z786^JUGIY)7c1*&AkB9Wcn!tf3^WtG$vMp|cVt;1%OZ878#-pWe~*Mid{~?W5VX8J
zIi^!jpKb?<VcxldHQ>bpZs~?U7k&Hu`80?!UOSi8f(5YPF%~KVk9Q6hb;}>V2GWq?
zK5SR)1Rcgs2qgSEcDkKyc=NQ@7mg-K^ItVKHXbnoWA+*V*xG6kjI-o@mkMRxHg7<G
zA;6wnBP>*JOXdPJJdbR-LOR$%1GeHX;pvd<HE}X3DvI}klamt#>su^JmO_luyhmg$
z&*5d<+{fOz&B*fL*G8PW^l5uW<eSwZ1GXxwSFZ+`bD_fHd((@{psy5@S~bP`FJktg
zSYjGjn{9Ph9PI6PB6UF1s^g&Ur>*c`uVaB0TS)i1uNlHv)B>u%mq(;+Tr};ruq(@r
zB3D0jd<`2D-XZr(#^%mrzCkR~znCD2O{$QvX?)HGO9uZo3F*OC89z?GbMrum3*F<D
z++#0)e12_@dG+ksvzj=|@{1`ZJPtE44b$^<x9It9%{x>3eLd~F3T;~3HBnqL2WGaj
znaYi<4~cK~L*SikS+-(FwNKGATpDU`X3Q>yvxHO-WT^J_ZEfc0)yw3E-cNK_2d}~`
zzNCX~*|NhY+)H-g?yY}53zhMfhU9bWeYaKO4l{wY1RI0|1ROdrvG)Zs&wEI^#q3+o
zmG^@an+5rcI9zZ{B7MaJP`(2hDi?fqHda<vYS{*ch7uxS2C|;`xMY*u=@(Zhr<xTm
zRF0H39P@(96tCg)$E&Q(-I3VSa$sUQbm$PQ*TW;M<;)w9xb(FZFFD%|TzXqwZl0q|
zXe+F#eGALL_Pql}bYE2!+dy#HMb*#8?LtFCfirF1a=4r`DkdhOZoPoOJR~KrfrM{?
z=myyNyY7AzYF$K@!ZBa61gk+&+vxYxvXSzerD?Qu*|Lk+7gh?g%$+MNp*a1{SQt+`
zpcRM<?^ndIHRO03r~4Pg7>O-2Vv%4ucI+6^<T=D3>3NYD?%<QTxXarSM>`~!&vRCx
zcg*XXVQ!88uoKr5yxC(%56wX?D>_mu11bP~u?UL|H&>G>As3HnveO~sR{1^=tfX^D
zd-d0DE|GyFv@de`y37?(uh6=7{*CCE7&BOFI>-7ux=oIYgB!bo#O`3sK-S&>9v~ge
zv`IkuYl1W!l|ih~{VxgP*H+p9UD>Z0*f`~FY!V+PZ0>^SyKYEBPHv`#re=@fy09H5
zf5tjbp~uaU-fxRIsPisdpJ&dVeGe2CJKxz(FX7-t)3F@WT||m@Mn8B!ZGM1GJ@{7c
z=&7;&+O1Su0FrDw-tNzL-X73;NbqL=J+RSA3qK)8zbS9|G#kFE#DsTw@Opn9G-(pD
zsoC=E`STTmk!ZOhymf1j+M3Q>-Q)pN{3Xv7?mrb?>u%w*FVNkB{R^+QBXq=zN9RmU
zt}MLDy<L}U@!~Vk5a}jz&Q34V!mdvjgQ$!DL{A-Cc6Rpc(vXe%7siHr>+kL<-#O<r
zrtKl@D0LP|Nl6f#i2>cXh@kH?V1`}T4BkORhLdIy$haA%rlvn&4|fk}MP1M2#IP|i
zHmyM$9XYvCpiInbH*8ob*b7-`Lu2DcjoEJ}%&*WiN2ZAoRDAjBnNaMV!ARFWTnKgb
zH0-x#6_T`O?=>hXE9XJ+W|N!+lnL<Qqx_OkTz1R*1fQRMC%5$S>LhU@?X^fA2RRA?
z;OwfSZJf*Jt>;mA)i<Lp7XQUpzG9wbW*Eh$oC_$PkJ*M9U)&C9-P0rDbgaNV(4s=J
zcmy4o@ueh!-l91yEG*=aG`sZu`}aWIBzJr$636ml2C7k!?c(V88q-|J<-38an_ID6
zXG_Z&Aa*S<`}#33@P1Q{#|V{=EJ>&e^$C9uAY&b~!Sox)w)QuucEEXY+qp$_-`Bu8
z90J^u+C@blSaoDV`3sh9$yo5H!I%lVTR!sEGzgi3SpN`U_>6J|ZhE>w`q-#nfsUJh
z!{ytDKcuCl*#$NT39;f%p5*u6A3t4v*6pH!;3?JUt=wk0a&3FXyiRd)nB_*j)icXw
z`fGmZGUi9b&6~@yw|+FqNJtO>cf0Cx$ad#$-OJKdx5AB;l;*-#&%F2OwYk`Ef()`@
zlqGH1G{2&v0_Kc~<*C@sI8a2Q9-#m(FvYP`Wht+kY^DurE3W5^$;l}Ji@u2R4-Uac
zkj28d(552Q_4@|La7nktkZChkhC!zD0w7X>hRxWZ8jz#_-_HdU_zq$%)UnJ3(UcFo
zoA%unh?XJCT4`XQ%)-1!L4kT7XVlJ;rxRA)4yTE%Lq@~My6u2(F9Idgmo#=T%Hn#n
z16sVR5KFBTM4zWcFkn1<aYehW->fA8Yd-&)=5PZ0jY>J!zb470uonWKV!33z{J@`o
z0<7b6>C)+#m<^a<7n75>0JmF#6_K&26$&Vr2CoD1O{?a0VIlL*ojc*Zxde0L@aY#t
z_2(9e9sk6{vu2GwB;Jy4{WHIQ{mKhtTL5ot=gyt`Q&4a|PU&K8@v2`*m}mB%UQXkb
zI=OV&CWDu)g;zg!bX-`xPU#MC=_9B?+O=!for)t4+Ad5K4$_u~TU;Y8%|&q#YJb2M
zT|n#LHvV3?eh^^#46_iN!8rtLzeYwr!&XZHTO2DVXL#4B{%f>BS$DX{3eH}>%QoBf
z^_NG=T!lhQ*Q1JG9D9S9>yJOLa7p>%#^^JUQPPDg1@WH7YP*XQft`D1_{KGBW&jZ`
zK-^{zCq;8hOP)XfWCGy23)A{@SC{pH>_7haLkxqykXzahN=zQS6*f#+#CEL9m;1bX
zrv`xYRrmO+G6c3iBB#cgQQVw*QV&Oym}AF$8oAG}X5|$Yu35c$+U?u7e+&-F!;t(P
zij?5WC2KEQ)UX#D1fC2l8{2uzWLH<$I}l`$r7P`E5#?%{G^J>jgM;lu@JNqgh5u}D
z{k&i~DC&dK3^C;E#PQv`cYp5fy?T*_`<viZfB$@Jcu)mfr}dXTf)|<p<Hy^2#(w-b
zPT_T%$C<NdQLw@brI&W54Jc_8c|^#6`+6kNEy-neTwENqhC84b<u=IiT6O8trQ=Rc
z=N~=V3{BS=+%>u8R`(2p_aITL&y_2CY-|LfFuQ;|A!gO`Y2%V5OHiF-WZSr8@nSU)
z(?PuI(-G+HfjoJm!|lDQs)d;4tQ;JdaMzOFlbld-?mBXWm#T{rLhIgFRvw4uyt^@T
z8EnrBa0-3;{(T|f0I(LzfbH&t$CTp>ma1e(rHk89I-IKHmM%FwWAmf!YcS~jZHjsM
z`KzU+jmC0V7cDw<?b;#;%I9EyJXRWU8*J!30tNa-I4a(pa0}&Gw{8}m_iOBc#}T^H
z)8ga%qp7Kho))l)idgkE;9%>I4raIwtUr*bdm5ZROlz5c^!4q;Wz&N?cj3aV;Fv|z
zF>Mjvuf-a@h%>7d6bdgK0}-PA06%dTGM-onH|3Km9<~GNeQ%}EPO|B7*f;JOKx;y2
z2$ucNotqh0QeIRfi{m9AC@3mz9KMb^+DR*HTmoMRv3D=uWbo`1FyGIh-F+@!-i<wu
zf@q*06sXOaH*YsoqmJ5pjIPeo9}x4sbSW1w>ks5cEhv$21>cJGas{hr4OTDI(<zE!
zd0GlrICH`FFSxOcUu(|i&!5>h_W~MOF5@MMS|-!JtY|T7!C&}}nAf-%CN=A-RsO$z
zxk|bZ&BxY_BV+gR<6B<6dKJF@du~^5mc0~1cYu8c>vgN{Q;!h`#D9&69s>deXUbmd
zCJXEPa~GASn4D^5juIRDHSMs}aOcy$8R>qrcD(vEt>8?%xck4NC*ED>kxk8G@_mGX
z5z*7r#@c%_$Y+%Gtl}qRd8{~g+35f1ov?J*oPb-m_$iXbJ*>`rgt@r{k3mO!LmcfS
zYhfqBzioxT4F||VcHlS~+Izt1J+^=xhb2Cpb(aq+k&Zrvo7dUgQ4pS1)yy+z9_i}%
z{yoWX)8NTpKkj~$DAIds7eQT}{J31mL>3O}ti!#{I&Y{$TAeKffPCkY$6Yp_TLk-1
zR&Z+pm_;H7hoO4#1yvqjLM5fMo9(r<f@mt(Os??XgQ#*ZmbIMIL>fG=ja$IzZh6jf
z_wL;mgm>#ZkJ*B;3QFw7@vYrykHGW(Smb~<;BC6uP!NMJ%%9dHa`+~(O6sH*s&V9z
zN5sazhqK*kJU*Vv_wTjax!Kv*kH|E}sB+uh!m;23x2fVAUjc_z_Fh}G`D!~#K2)D&
zWN<WY(9zK$d+W<cg#t^GgF*M^MzwZ;Uk@ltz?-Mm$4%rQ{0&8?(;L`=65~gg3hJt%
zVn7a=sJ!+h64Vf>Wp8eY?3=S%?%=^&6r;<*^x3?0!Q8ov5uj^oTc+cQQA5*&y3O3&
z+>j=(4Y+yphz3V3Hlj75#~}85Q}Xs8Hk*6D#K~6D`Ju6gM^>)=@s_+l(gyY^DJ@1V
zy~GI89c9r8@Qd^7%#R|;Htwv+_L_LU^H%r32`pTvZ<_EnuAK_NqfzWQ^w4nvbKI%q
zs-K_M1P4(*bv^5qZ<-`~VV=|tiq~sTPEMwdO@WGv+hnYQOTQ`M6wiC`aco8NX)te>
z!EIc^-3fH)VL$5;MZ~^r*BsU*OYXnF5p)tQ)0?71f50GkBe<8SnuTRvPG8>b{OGi^
zJl(y#*A<^3Ve2-bgOMQeP;Sqj3*ScPt<FPOkOwXJhVh9Bwyxfm7M_#P@*y5oE@`f7
zX=yo^kdR;=F@=rz<A_*YhRoKjnkbY#2u4%YDFOvZ_eUkejmsyhsvz4Byb^}oSY0?0
zp-G8dpl4%30R#_uNG|0Ac;vM{Zus+0SwI)^*x?dNypfEh7M6kOIlpL^r0Wk^G^5_%
zcQ6JpKnobx_+f!{>*T4qB@Zv}-j7}{K$McP<N7JCx?QHG_qvCdEMLwGu;SstNfbFP
zC_$&H6r9+p-JN<dDysrZiv@bV+h7^`7Jh)WcekN@8m52wm#H^cW8H@P3(iFFvU?!s
z)$Vj#wrm-?ZmpzUJAb_Yq+ktup}Go^5H>gir&oBw!!ip+$Dzd|`t`V%bzP8k=m+)}
zkA&Up(3N!T@H3Z202<V|X-q<M!D0Vd&6zH$eq2%qfBpIuTSsk!E@5xDsdeEYl5X)w
z%*!?#368D&vX}G4#uD8vX1U_VH+*KzJ2cWa!%duBaAqOP%@U467O{bh=FkEbv4es&
zb*;?H=as)0Up%jz?g{+8n&bnVxIKpQl-~(!W@m>g<!FkCi|aXUX(}z<r~OVR$B&}M
znL5EG2zGobgtK0}cv63}l4Cn;IP%|iVZ_!OemI=tutUuOoZI@4ukG#8t^wAUdrxA4
z#P1MIGseDv1z&iiX*9P1GcAGZ3knrQK2B>yIMfG2%3F_FX3br?2-#6)Mb?fJU+yg7
zbsPVc5-$<@4d;HUlZb^$VqC78taQz-dXpoV%-}hr^ro)9EWPvOmTaKGq8!9!yXLH>
z-SVS_%B^k1N8C6#I834CUq98%v<T4(8Nlx;O%~DI@ow74!kh-@JKQo}1_0kH63q%;
zzupZ2^V=gFI6DH$sCax4!UPfs#Fu{PRSW9>)n`_5cQ<p|5IMB-@bi}(;TVM^b8V=x
ztn4WixeJTSSlj;$i<=xDP>mQRqtI%%g<O)l8{R^jy-U;<NAi=e%3rl$$m_4df$1Q!
zH*i4m;ZU>;Q(0Z-XlyKRgspN$=u^aVbv@YGk0wu|2vW#x>g<^_S|OX)ZQNLPBqses
z0=638k#F8CwBK`z-5AhD2LSDyw;<|8GMaFF;Suf<GM-2uLmZWdK>g8kKdeU`r<e*1
zyFBbXmFq47Si67L2nCm$cKE!lL<q22Pt~lNZIttC1NX|60yz4VlT?SS;cAl_F;?4O
zn$cXR+#|~}XU-y61#K-VT)gTF&J&R?$=O+1>e^*5tFV<6cR;bBWDgNbeydDp0;0Ya
zgbjO2ajGU(qz5!40L<dqv`OV-=V1Y4-oXza-LihvfME1mZdnYjy9K$4Z{Z#BRbSkq
z8zagpn3Fp$q%|q$>n58OoB@WjQBq23KmYsueEGh#lCBgJeNKg1uZck&^*gfD0*6|W
z1q;_%rfUo8&RGqNZmpiG=InyGhw93-PK}ox%y<!6f>di)md*RN@daDk2NByAex5DS
zfAYtaQTYieSLpj|B{v&9GjZ^$YWZTTUU0_A7?9&8cENK_1uSjty2)A-19iWq>0_bW
zkDp4rJx2dplYhom$m>+fr-staDW9?`zn2&MW|qQ3R_?Wi*?^Cq-%KVH0<#l>iauS}
z0s>A$-g5|<<umu;O-!Ev*P#0o_q}`fluzb40D!O`>9=5#MN|p3kd9WY+B9%PK&xir
z<U51l-?|ho(IP65(*6i}^j(jwdXF$~i`TeokGub#v&cdyiNS#=2SsJy&##(DmzWVi
z7hns-k>rDErzm~H@QO%w7s3yeWlm$lM$O93egVnTeH@2BhK34|$AapfWz+DCb%RPc
z6Vsg4TTJujpjvYl#E=@=!}v>rC=?YHkGr@iU`9ZfvhUZ}SSvz$Xd*Ymb@vj$T&B(3
ztDCV8QVL~fS2EQPaES{M;X}_}WKsy?kYlhDUVs{^<49L9V`d?ppT-3A>pmou6o+AJ
zv>1>^!h-ZK(R1_g-fs<cFWjHDIjE|tQVh#uL6-$7)-D5s@bZ34!$X*~2xL?CioYyX
zID%lC3n7v4j;omb_X+qzu7+MFM-Ffy*viG|XvYwIgGF~9L3QyUU~n>wO}1~{llCp}
z)-7LPsjEdr*-^o_MSw*g<!yf~<P<YW_CPI64cb%7mqP2s0u|HYp{ZJ&eOs_0Qi=f#
z@u8kuRCM%bXs&KF9L^BNN$P3v?D#YU;@7RXq&+Cxl2cF!(<alKyYK<1K`A1jrr47^
zG9IU-q)@_z@XP9eXMb4tfL%J89qg!fOSi6Gh|}Riwv6i!lca$YGpi)4cU*>QdN$tW
zWA<i5=Y*)?y@U8PXJbc|_3)5Pf-6!{(O$|vuqC>V{*3!3eEs@$%IF(1U0AR+enD9z
z<zA;wF-dxkE(KGdo8_7^<o@jV=jo`w^+%gg-j0>r+>a*4C~rkIx_YLwEuQ)7_wSbg
zy)D4LeK}Y_hn75(5!Vt@QQ}}!Lla5uK^c!<Tad6VV4^3ADbBih@pP08<>lp>0BC?R
zC`@z$i57CbV?RA2uU$KZovQ^odz}Tg1wsU&7CnV6E6cTSI(EHRC>C49#x{!y4*b(_
zZV}Rqo82Sa3v3ub+mU93s(K%@Z|zPh`vpJVS06uq<guvfZ}1-eZiqCMNk~Y@vJ+~K
zUGzAh3_E}E;>F_n9q%)mZ%17TtEtgNus08C-eCFCTas?V6jKciVF!}OZNmWM;E8V^
ziW4O<na)F|G9QmBzOoSaf`Vg-P9{&}wsNkn&xVF;wIDv*4w&Hh&oAmo7hfex3id#Y
zmD&aKDmcR7`9fEE8tA~9O`8^i|K(M`Hxu&z(>PV`poVLG$QTyzPn4S@{gx>89{su@
zNI+*HDgw5OtFA-#^iYW1#v2Je+5m)<A9F^|f^5?do8L=7A8#XN6p+MT;g-QlbEf8a
zIHPuHXk7p1(C;9)Jh{n!8#Gh%S4n6c)`J=iMV33L3ofrl)ErYYGfaTf*hP-yz$;_)
z4+N(~F4#1s`_5{9?zFG7ZtaGB=~GdR{xQ3Oz@n4I(G-Vp5y%7qC%;08f!zU5CKKon
z5zNlUXEXcnF>{tqf7P|q;);jV@}*19VWt&Z2O*NZ3-bcYj~j9h3F=A}obgY~|8+Wx
zYhL-~R%XuC?BO~HUAi(NTbYeg4+z$<7O-q#PgL=`G%KxfKOVK*!rf)7bVfcGYn_nV
zvgmIuKyBJ))c+bdKZpdiqmGa{Vgk+s55Ebga_{V+M9Og2t~0rU)`04LEp)9#D}hh@
zBFn!-|0nx|_cwUecV!xU8wA(&(diw!q`TOPV+m{<eMWN(0=wzznArihZ*K}KXKx<F
z<dp|9OC~XFUDqQ-JN*(+-K8V7^}vw5R<?&*t=_54#@RMt+rS&Kb!`Jq8SxV=;W9kS
z!2NJk@KV1g_eVr4;20y(7n5&pQwElJHA8yxm)Yev=k9Gw)7rb&hq#rt=P<|MXbTPt
zjaH8My6>xYv=SR_^R=mMmTR0`T(&1<LCRiy1D1MoE_Vka*pAmrw2`+Gt%QWI>&=Cm
z15e;0e>Uq)g#S;Z0Iv!Q3r!2lfW=vqYTOEV)6i-Ou=l#f3)j9~q(Wk^jlsIJ$GZ90
zY>f!NTd)?T8)^|(pNNkJ#zBgHUo#R)-cE_PD3kosMjbP22jWByuaP?b=@e~`Kvy5D
z)L}46(?hr<UGD^5eDZd%D%}4k&ZD}CxpU^cdjM`Mx_govsh-owC<<xh5S2e^*NWFF
z{*qXafuKfyY><p;DTfTW7ujvCiqljOox&OQ?Xf*hj``X8o|KjXhYUkH|IG9`a3Mm1
zB6=TU&8B`Gj9h!olM`#aw0wGwQ`Z6YShdwXq6kNkvhY_WpqfXbo#?%EPe@nSq1m>3
zgkE>k$*zsHS%dY)wOjA$??$e(&RYP1p_PX8mE+%46%|=?b933n6{atO-2eJUft>KM
zhx;+e$c0g%)ZN~`4r|D%zqrgM3fRiy8X-+&ShP>orUqHA5gC3KAkd2gcfE}EuJ0eA
zCB5i2x)V>>e3hZ0VLqOewv=r&PJ}&Xf}=>EY_F-7hu+&LwbIJS%H^=*&zNa#d%RqQ
z+01gUu+F<J<DFfe4I0l-$n(H~8-TeoL7!fMVDqWXb1bCveEXjO?7xIxoH1hto0$zx
zL~4{;^`QF5HeXfcWNkrLRE=KGQOAe~EsTFqE^dr?fqjL4fr%R~X{U6Cn{aSi!5OfV
zydxu_11n^_vPL(5EOC{(IR>@dBj)zoKS`{3E9k0-_(}bs6qAB88`<WuzdG5<jEzP&
zXVMLU(nhn~@@-x(_1;eG66^gnO~GeYy~8+@S#Cy%v1#tEHh*UqQE7%OsK~9)GM;lx
z4ME-gt9{u>z{*Wt60ik%IQ)b(@@;X$o6yc8A&p2h^Y{7*Go)gm#>1jTCuAp%2mG23
z{L;fg7U>Q;Yh2AJFUA7J9xIQ*`6ls5Hw9dY(yBki&)!9R5M=!Pgrd+CFw?m>ka|QB
zkWg52-Kw><m2%9-R_O0x@pCKNE+3G4-#qpZpl@?f-{}s@;AQ)_^LNT;uP8Wk7Z~ay
z^fuU{+xhh?AClb@{leRqk7C$)C%>O`bGt;1F{reXT$W7B!^QU;djJc&Gd6AxtG<XE
z1OH=1IL!tgb`epkI3@%4cXf2EcY7l*FVE+EI|R%ph)E-xy2px;AMEyk_MO@?gzTR}
z#lZ4ak4*&%B;HZ<`a|~w(><IWJjkmzY<Ap19VrujtwQ@9IfCQ68!|M<p^P^Lt6HbR
z9vAZT=~EkpiyDihI>{@7+XoZ)5^`)`m^JX;pmv^0H=>&l-JhDfa3MSNq^y9rmmx-D
z1#=n=ASihMvBV`x#s~}74#1}0M{z_q@P4y0+w$eHn%vO~w`;c9Fi+d^#h}QtY~fEE
zv~meksY5a!zp1dU^Q5)&ole>vZ4XAPJZpFchl}!T!U6`<3ueFn>diu1=y(~19zb1}
z%>YnCfd{I}%36lTtIxpzG+FeRi;GJi(R?RHQ2@CJV8}8A%f2IzV2N}%45w((15|Iu
zsMJB09#_XaQJ#h%aNmP8ZRV3oO1{VK?E}Ui8X(M3bQ6D-;SCX~Z_X4L15-HjkOtTk
z$30nL-Gf5gx1LbT>lW=XjyWQHPIqAJXvp9<8c|q&FYs!{sxO!~ubP{Uo%JdGhd>*p
z0Ny@`gM2Y(w9_@Wj(v<$TvFQ~B=IiIc%4$vB8U~1(oj+^w956?#EPw!e&%|Bel;4B
zpowllNf{eL*Tt|<V@1}cAt{XlKJ%s!n?vHvvF#oi&!6*G8Gp;vu3sPY{Bsg=TZv8$
z>l1f6&tokqaKF5@a^$lm|EUy{bvU#;Ib$`ECVs59u(T{O8k-X^Y?hlYc4gP@-A}#(
zY~=^paO?y(AjPO9IXWwN%@cHgfZAE-eZS<!Mr#KnrL#Z4vE~Ef706JTTEE@#*aL?u
z-QyU<Hl~X1keHa5UqBr55trqGe8~ego8)>0@&&e#&Bmf5V^3xqu_$pfmmin<g(6GE
zjsalDq5s-0Ec|AL6&(D%S;SAe*<!Q6+dt@=QGW8fOGn>~p({W$>icFO8#D64t<e4b
z774axxo*|I;)%9*vGESHWA}&_U@tE{@*y@LUijyXB}&uw9{wi#R(y9Su;q<69&OCF
zc7(sd?l5E*6Bh5kGVg|2-9GJ+E@u8VOCPwB_^L>lrJ1$=1Y~CQbl%*#*Y_VzSa>31
zVIxxbS5Y2*vv!|fM3zz4y9Rf}y;*R9f7TPg9c$1W&@Q=*kT39p8#5!|vDw}GBxT`G
z@W(4`9NSCSA~qcSk`p!C_RHI#umNqT{EM%gM~amH_VtI;Z>vlbXBf^|J$oT1Z-$Y#
znjcc&g4u?CE4QmSWFjQt^O|y)vDt4{g_M-ky2E@`+_FIf?FeHaQ8-Zv;k$w8kvXy+
zsLz`{W6`-qJnPr*iP^fjv-IOfZ2~YIa-xbRjQOMv<yUb|Sn^NQ76644y#8p?#iOr9
zzW9XO!x@d0ohLC)GS1yM%d^MPtA@Q)f6AS3iV9micPEM9{XM5dIsmszM63ZD7@)kt
zY3oPSd}sv+3>446!7((P0xqEL#Oc$g<0amnU&JFNW!PB*zneSe64)%!ebQ*?Y2^nF
zfgVpfJu5<cw7*FPU%zge+=vw-2VBPDX<(+%dYOo^hsC<gClPAI8f2vAKA%2ot`C4)
zTj>txI|2)yA9r+If;=a<p<GwtS;NbDX7z;QwE*;+UEj_r8#Q(PdEWpMEp5TWAXKgo
z)~6l5+MO7CMfk_AGLyraMp$!Ng1v}GXQzF94(h&PzP10tpIPs|mv?-bq3gINr-ASW
zN&WmY^?)Kb#$X+?mu`Plc;=xe=PF58wGEBP!V*q$qe^F4gl+rn29pHc<0vOv>wO<<
zX@7ZXO_>(B*z2}L9CbZj_U1S2fMDgu{@&KsHSW?Q%~O*bG9#iNWtQc4?Q9I$YRFO!
zXjzyWds%U6Pi{s1J-=B3W`Z@_s#eHc3jt)&HUTCf3-ATdyRHo5Ih|ZGCcV1J4+{|r
zhr5sPv4=0KYpne(9}F09m}Uh5r<(S2m}eiEx^n5#POQ>z!Z3cBq5ZYVV_-kc|1mEF
zp7Etcr5+w0kG&CkeK^`)(~^1MKDPh59O#nbj;;in07@u9X@|*6CIIPH&Am7~gc}jH
zJn^muX@y_a4whj;Q&Gb)mlG$p0i@dyY6p1{g?&TqU0n&!=Rd(&eFeawVv>M%d{BOA
z7gcj3rA_ugjEB^6S5iNacMdo_Y|QjR2@UZO^KWEkX0~FO&(hhHjq(wp3~5B_zfhB9
zoty=npP0RFXAHAMZ(v{`RIm$+A0vfD-#;nrWk-{fsJs?g_b0584C)RY6yK``+I%K3
zgR||ZZIeBC@L*EE5r~iO=fJj2phmggMK!P^6PBeTD-ZdiYvN*1NkY1l2y8&1$~uGr
zW&MO(lqdzfwg?bes5JCum}yc52J^$V!K^QrK<^AgOUv~z_A;|s_PEV!w_S7gLy4>7
zg6r1hfBJP}EJ8TgOB$n_vNwCk+~4uV*5X>DnTgewKf%j^X!Cy{0M|9BbE7@I!B1bV
zdJRz>g=*P4^_KJJ&V8kS>_s%cZMpT*)#uutD43e^Oiqp+>lwrBzYpn5d_QCm??(h#
zIK}=9{SKFf3aiZsoV2tjA83sBr0)r<xYar7<1+_DBAy3u0$2#F*K;PDM3f>Ib*$vz
z;HytM(4ySw<?i6*L>=aBlYzOdLXKYQG^u1;0QuWeD&#QOZuw^UHoFF0XMM+dKWqp)
z@i-UU@P_^?wf>9oVJ5Y4FSIyZg4A`NLnE&GJVj#;jbN#MASNp6u$~9)5}aYq%MXN$
z+289vW`E!SJE)SgK$arfLA6(33ecTb-)O$iZ&tl=dAg^TmR3^zp5KY<xP8z-c+dCm
zWbYH3v=1}csqRofK*05{Hmvhzx4mVlI#R8ZsH@(eNrXER!SIBv2xO+bPb&}-oWuAk
zCanY$2*P+>^+V$vE{Fh9M<f;-v2>Jm?aXMthWvr}11ISbMA#1w3D$_|`*9a_0pAnz
zdU+tl<nRiaPl=tA<I-L)Z>N~J`*E8H>O#x0Ipudsnw|#?IEP2xckgVGC?xDEoi8Tk
z&tJ!A$E$=n{%8gFdwgQl>0^C|HdIa@CXqoLRlY6BfrS0(5sviHr2S=6@L!5gcSGaK
z`QO$RHaBx0%j&x@G=Zr;zmnh?<f~QL$Lb!(#b>3#iBP+90te07_agA#gec{GHe@~f
z`4#+;{nk1mcW2+MdcVR1<2$=_>(;F(C^v>E0W+a?;b$zTyEtM}>&Y++DQuwOF%{-f
zuo4#4#ZIm>E;K93$(^&=VDIWGhEPBtEF>o<$L7u?*kX1Te!YE|Jfwm;+m5QdJSf{7
zgP&Hq>;N)#WZR0`L8TdvB<^rXE@^CBMLAlV1a=kR6Ge6j-VMAxI>3Smy{#QY1RT(9
zz<df-W+iqps^2e6i*8;uDg#Vpc41`n<l(`AfUa%D`vg6cTtuNrID@Mgnt&4k%$#P=
zSuJ=_W$)fO28M>((Pbl^de2|PM&{Is-sh`c1;>MTBiz9fS3vS2yNmXW3=?fdoVc!@
zI%ZZ`ZAM{KbCmFhRHv{Cqs~dlM)x`^S-L~5payTcVy9hmBEmX998Y)5dExZ6au!6E
zvrEx>Kvhm&J~ZMOR-j6_*y_xt8WY>-!48ANMCOC$yxs+skW+2*p|~lt=zjkN_jFb)
zwdrkLzh=#D?dJ&24}tO}un$nfdBBT&=~Lha1TsA62}N|_4agkeqmIO~n(yoNG5>_x
z{AK}Qx&!bwb;g&n;A*qnQ|(vL;ciY(GiJtY2o)YkcpDoVL+p06xqd7A{Gv$i#EK+l
zqlLVIC1g&T@qDBl@Heu=;8lJ=_Ur*+DW-5aY#)IIx4x6G^u87d+OoK`Zb@F~(P&*h
zPjrD<Zb7Z-xGY%r&V0mTE9RwlFo<o(@l%7Qxu?whG3{%_3eHTi{FX173)mF|wy_`r
zNS)^FIqE+W6}48oo((um%>9a+K=Zn?av(PV_8NO~EbEKR)Cda(|Iw;g@k6g>l9CH3
zm7K2bav-2SfVMm1=j(e^W1ApAnc!}u4Lnd(A<CY;EE}*u^<>2D+gA`n71y3|sFc~{
z7iaY1x#-KqOO~8{^F}%1&pf0L0imIBp7>PDa0w@+G3m~i=5aZLGxUL1ipF`}i288I
z5P(HOP%T6yFh}G>a`s*?8~$UQ(jK~^%gx_lrdU*?DbdR>LYymB_#!)0OhQP0OVOzU
z$@5N!i4Eu4qMAeP5#ZH6#-iK<@M~!o&-T~)N00I&C`?*@(`hZ?%s5a?w^+9T{edec
zQ*rH+4{8#24$h*AXoNT)p~8k8_m7=8@mR4l8OIR+!h)0!_G#wDi-Ewar)+XK{Td(z
zpIwM1xYo5Hcw!0VYh)ZDV1bqUZE4SV!y7>Mf#GyY9GfUF!QRm+q3WLYGoz^%%Mca9
zi?QX*{*mfg6}@j6m6{;r+5^4XqQ#4^cbVM<=pX;S!)9x|?dIvSKn!sj`vYPA1Uj44
z6&e~EC+rSRKe40iqchh{2TnIgv_ZN;gEj3(|MT6EUWF5Ic_iswkGq*|0f}{Nrofd2
zN}KIVRS2tKeLL~8)H&LK>p?n4L{CHS92h-Yy<50!mkAJbOPQ25L=+O$-r^&pqtx}!
zw_d=od2iXGqsqv>EUGh}J$nfT(!u2VhdX@Ht$VC2HDyPh);AUU<K7K$_LeUYhthKf
zP?itkL+(GnoW|RgF1#QO$eHr+*-<!PrjssQQ3o#ba42rDw4IlpS$qHr*ZHVd&Bet(
z*{(*qdIXm8*ylh(pGCBA2e4$u=mKJztT`3L>xU5LRc7ezOEQcc@GN&slfokOsBb#h
za}p{Z407cs4@mqy(Nwd#ds-^i%YmcrmX?+)Ja6@5yv?ie50$>SxGU!%1?)Ykmb8Cv
zvSIw}e;hEXDWAH3a>{%%Xk`u`Y>133V9{N}K3}Yg1SI}ug1@Gc|2kh@;e3(Noc?U|
zv}^R`o8cawp8Mkiy_j@!?&ou0vNFj`jc>iMa8-Q7r9S9&ej#t|Lrg!zRXkG}eh;MX
zL8ZB#(4sZ@x0;>sz+qkL1CJJX4O@T+uUPoAflC3XvP5Z1YpYWgym7UK?wI;X9cH$7
z44hKb+;R4@IF#Y5#r3PO10VnKX=hr$-{s4Jh8u=st)EySi8R+RoDQt%@HzEcBdXr-
z&TG`-cnIRmdU&JqaK{nEobXf7g?*mVVK)N<PRxf(SOz!451%~wT%|Kg`bnX`!Ilpn
zT02l+p>1$1u66z4PX9Pu%<6M$;n93y4c#x5w%)y-go)n8d7q$CCj)-PefzeGT@~NW
z{5<;YVEGpq$P{7&=!1ImXow;B7=ooCQ((&SIHQdqwE~uV=Pq6h8=pFA7p7kJBo_s|
zbXDE(Bp!SsC;G&8qvMtP-C=Q2NjrW{Rkv9#(-9C7s<7LxS$n_ih`*!v(2#>oqvZ*l
z3~P$Azc+%zDDNi?E!ePFwYz7E-S}5+JvzA|xDsR*#B#Y&)+ZdVuADGikIblVSg3P}
zn6gE0*L?pyGXOl*PP{*|?IvWjZT1uJG&KbV3*7udTRL_mB21q2Ql5I&xfj*RKY;kO
znX)LqcHUk)7YN{8XnawQzO@O>VSCDkAq+&(`Mh+S+UcL#U)}-_bng85yC6B%M^5#_
z)^|!wY2Uv0an@KnNtR9N!^#y7w(iyiX8>+q0@PppiSH<U=(mjqsfu6ukn~hrPz;Bj
z>eC8k5<<C1@m8;&RkKO^a*D|y4vF%BLZU{Gt8mJgxRoOwI)JbUFy9IUNgOgSb(6o3
zjxU-w8^;(43+mOkuUn&cE;G|Tsj@RzL9Me2VFc*}F=h4|34$eB1S}*U>I$_7${Y~*
zA8=T|SpTxG?=A#2eR!f_AKezl&a2TI6%)275EM~fFkSz;#+@A^A|H>oS0(7M!29h=
zZa;FiL)bek+v83sCy37unjZAPiycvrZ`c0N;DTFI32wI6DnM(*ZyiugN1@D@H(H4|
zQr_Mu^uG%!0@!M0?SmN;PxEYUjgbHRiKa*~d!A~y+v1LUHWVF9G7!(7CgV1%-md_A
zlL?*V7;eb8*M@fBkR!o5sax;d`Ewjbz@kafk0H{#gHs-O%0<1N^-or1Wr$yiy0SW*
z!|`opr`IXlJ({5o-(8jhq73m^{C~0cmSI({UDxok?=30@1}I{n0xF1vgoLpz5J6Hw
zP`VM2E;|50L68Q67LaZ*XbI_90!m6tH+<t<y6@+C-uM0Wecy4se{POD_QvH}*SfCr
zoadZljyc9h^jX#jJ+Cbq;Fd*Zl>hP3gphH-ywEAVhxH(nhM=PN=yMgh(ne~Lp>eT7
z6v)>xF0KoqwhkPTstGo5NWF5vsNBt#w1{|aWU2-tB$BZMLYU|f|Ni0y1JICFI1|oD
zOOs|{>{Jqya&dJ0b&1pm{f)7!ND6IOzm@&+a%AGil!9=$V0ZX=9iWbfncwI4W49L3
zU2HGd2)^6E<>YO23hqW{EJk|Kw!yPTip+%nE_6HY8cJp~Vx(I+GHBY~U%2MKCCHx+
z$TmKUWcHfjx$h%%&x5;5#gKXdxB5MVQ&?G9uLGPRJY7O}h5K&r@hzf)f{eI7g@7D>
zfzhx%dn&~n$L(u8YN14#nW?D|@Zdo|+j8M0=UU`JV4eOfbxZ=-Zx0!x{a#u<?dKNf
zA*XT&d|}U$BKXc<&^GZ`2JUbVy+6&uJIOlFreQ@IkGAi<kO}fi$isZ@n0ukUUY5u1
zWleiUqOZ`Yz<6J#i@uGmeu?dkcE27*l3_0pL~fwbFZ4zB^5r8SetC|Lj*OhVvvh@~
z&d_;DN#3UTqV_XE-%7877VQ1bP1XRsB1>&kih%vp`$isq9bxckPB40FyH9MSy7r>n
zTL)eAQ<9QvpegBsrY((t%%xfrX@%ddT(j*q+95yCU`tVUxD+682N*cPdNbdhe6U)#
zF=1bnVjLNpVt_#cvS%S?ZLkgUao;344jwCfUc>!1*AQ?&5N4i8*8oGo#6Z!|*1@zi
zf#oY!TnSZ?2%cS%!dUjazR8s&mgqe^#mUfnNX%x0mGD|TN52ekd;0i9SgXFTH*N3i
zTmjf3(E3iOz$flam^%Z3Rw8zEa#Hb|f?js~M;Vk>+7%ssg!O3z0-L;1q2)2bO(}h~
z2hr-xnV(R$2SE^b;t0aS&5oiC__A`{b1%HS9)KeF9XKMPGRr_FC$r`lSK;04?Ce*S
zx=DqE7BCDTrT*<vvyw8tJNuR|U;Z77dgYfGX1bk6HUEV%4Hv|co<l&Ma;^mbx0UOY
z>b5;J{1DFGbs8Ug<I5scRkw>hQ&j&*DWq4TE2NKT|5vx;*&97d?-SifVHcE`cXvrj
z<`YaS6j@sHc&%;Dbz(t~jnY|J<s_Hz?SXE2qoCQ!cGvjs_L|xeFlfmr22K|Wmrem$
z#ZVb^0XC+mQ$koAgr<wH)xtSG2g$`-+NYV1T&FnB4WG5PwdFA0x_Psf83>upN7)_9
z+arKufNlG%+;)DQ39t*8X_R)@3NA$rg+%EGqRq;_cOn`_pzM<QpnFT2K;}CuDd}1~
z&AwtKShd%0*@f~>tPx`Kng_w^HGmlHZlky*yy;{Pk@UFn`uZz582(M%sXU18`;j-&
zDM<Z7XtYQqFKtLo6Z)9P)27igkYCmQHX<<n@}PMP4!B0G?qf6E-e_VH?UQsr5xHER
z6M88zSv9H&W9d7O1eBFkiY=4%f}%CPuPAg_94Lf}fRV)qxE(j}1e<Q5mne>m3)3d*
zFKtmu$;>ty@G{Lo@O&GZ36<+!BRu=4c(WHiUw<+@hSj3^RjW1zG2Zsee%ujQ^;89d
zainclTaq%CE+&@3Xk;;H#fp$ofH4TYBR|^-e@c)mv|<9c9dlOlS#I(|+Q>g7gZ^st
z0PkgA+$;bNi}Dp=X7@G1(9l;(*7xcF1pMq^3>R9mvAtSrX+iW*!>z%=!KN#&cV6D}
z0axceiUb&<pPL=?1|OGLZ9jH)KC1UDD74||7Q>!)NXM@QccG~<MJ>zvxz6uYu_^vL
ziPiv&!L|r59|5G~Q+xI~3w0dyjr|x*)3$5@ki>1LZ1-b4M9A$XIc>+s$CIfS8EsIH
z^`hOMk#*GZgA8Uy3h6c5r<PdQC~q&M|Cyk9Xxq&e0+CCYd;c?Uq$vT=@FB>urZZ!D
zgkLE?gFIMw(bB@gD=@I&%NM%g;bF=eoeh^?iy5e^N~07)Mzr_%@f~PXox*>UXS2|7
zBc#2zXCEq;05wf^0<c1y>tEQ<rQ|^9#$QE2PLAQ|(WAGt`Mmbt|A2fmdFbA~duyu*
zEeEp{mE|Wlf=Yn;rwU8g8MXhpm8agkc@v{LK}H|35_W@TsFTjV__7f!k6oFAcyuQ)
z;H7@E7gJ$J51JUztzGMN&1$skV&c+ateShM{ZFlQa>Y!<M<`Z<Pq8Yn6Reh%i7z56
zLryGvcP%qnbq+!=yC~;RpO8~9YbITJou@P?Sfd7Yk24@3UlkUKzt*nJCsP;o`&e0h
z?7tyVYe!n}3G;^X^WICJ&0Sk^8{nazoTZ}zI0R54+k`V`<u9{`QNaC}gY8G2T%j>`
z!(%$YJ3jIlZU##f&zf*unNgwtu<QG~37&HoE>!s%iNq1&J)ucqpytrE3<l`VR>u+^
z&43w?ysfPeDy|rZ)>IR<TAmTVYVG6bV)rJ2=u+kTT&h=~D)%9xPf_AASueQCV^w+Z
za9qtKN$4U9f7W_Ju_nac{uJ;P!8emDrK;}u@onfFHt!>l!-xu$(3i6G@_%8dH#b)d
z)Fa*ZoIE_6a5GN<J=03Hs_9$vz0))iK-9g?#$%6tZQpd@=qGf(L@I}O4a#GCj81s;
zp}7;*V4P+>{Ms-F;XLu&{GlfU#%BA`nhA1op#vY_9`Mxf@G9W)<JG&1qh|?))b7}l
zgv-w#_Y!^JIsrF?u18>gEU~sRns3Xe!*T7Sa_lU;8#WhrqR)JL{imp7R14_-5*Agt
z&N=|yvN)?T`LK&oG0m)VHApqCE%zgaQ1lWQIcUdMKz9~9e%L-Us`{ouBa)a`Y$&8}
z5ZY2YysRWZkH*;R>?j;w=Iy#-S-A)}d)H54;s`vK4Zh{p^ek6uP<hLRN=Ql7^}L9=
zk{H%edinNXskoW@Ozk9kjHE3RRkbvOM@jIL#?w156N{kPc6L@2ML}=ks^Vfnb7^cD
zd5a(&EiK~qvj62cXc<O2sobm9?fCkat;C9ts~Olrc<iWTI0viQv#*n_W2H1EvFm!Z
zR;^qK)3gm+ilae9OGt$3izaacPoP32ZUf@B*EN=%JA?XWe4|m7CY<mCR&pI^UDPO;
zMC(C;BJ7>~#d!c91MdoE;_wmvS@hl$$Uupt4c->{Q9>UuzC+txFJM#!d|j<nqn51n
zYpZr19H@t#h@92UFl+$8v!{_CkP%_p3@RrEgfRW;siNb4LfNJf19cWHfKOC~6VuqS
z?SDacvggot1nagn@`6+e1Z3`?X&#9fJgXfC6Q3`&OtZ&a<hlUWr-{Fvq#;aS+aDVh
zV7RXL*3X|66r+QnPn|<Wl4DZ!_DR$LAyonF@KtHSi<1(|w!dlp!D_Li7WzdkEiFxC
zLg$#Cg7VqdmAcX(pcPxB=<+o6esjVULE{qL4`1Bo`h9K3!CWeh?G~^Lm<8I-PSu_R
zj4#=;N&*AJ0p*J!>6n>NUp0+EY%O;ebhOC{mOyP96<Q`OYW$F%#SJO4{4^O%LMfCN
z9hXCpANDY)#$VEM-!4(!bq&nr#w8SBQD{13cVqxLxPdawqYoG))UwQ7YAZLQ=+Vhj
zdf$nGM8>?&{k18nT3}rPyiPp)?8OUK;rE<e&xFM=iRsU`zL5bFP>>mWe@+#^L7gsd
zLMMnh8URS5MxEW<3eZU=YV+SwFD3MBqjP*xf3K_X?vm*S3iT}jY9(e+F22YV{Cl};
z(U$uFx8ttdyt{Yi1RXzIjsK1=`j?%@u-^CAs+|XeYT2JTcn!#%@K2xx5aCy;M`Fbt
zV(a-OjBJ95A#B|?5K_*JE`kgo*(*>q&DIVaGP0$->4Ntmy&(>YwctHj{^TXosB?To
zsFh&i-70y(V>6-|IB8ZWQ||j7>tv0Pz{tjQw<55F&XwmF>tSL(p<&Hg6r7R$8MrVq
zI4E>UuBfy$4-t8@6BoYieQbp*ZymcXH^reqypFz=$C7DGXajq_fffs}hn0BMTb^kD
z9lMt!e`Hk)JD&X6?-1WFGVq&Fs0IuUMCLV12?_KJ5Z29NbfH6~TYx}5Miz2D)^4QZ
zFTyN@b&Z7RjSl=DT30LuZp<!7uU_Fj5NVvxcXkK7;yQqyetu(Bku;*Xh4FRhEQsTu
z<X0snKHbK>R7OUoX%gvu0?#e2FO913^}bvip9Y@Qz`Jh!4nwXb=>|MF6x80H+kiA%
zuQi%6@1>WQ*2vH%s_V#D?xkd-0*FBR$g5tx%0$pH-e%l&GQX?eIy*gm-tr`R{WQkL
zB;kB#MSiL*PG;<M96elixqAV}=w>iBysyF4hs=mecQ<ii+zY6qOh3a;(i%VUu}<CH
zd=c-dNp&@2E=lRYBw*Eb*?kv`GO^cQp{28O8H4){!Mo^Dlq{rxX25`+K=Qfy(%Ls?
z=^GF5!_#67$x)yl^Nz=dok|V|An#yJv7kIC07Y}3JC_4)Lqg$GcJ8fwK%Y1=2k`~?
z)A%zGu0cyYr0kASy{Va*-b0Yo=LG#Ld?u9lTsKtVtItDH{_>mIBGd+Z$x7bPG=6(P
z<D<8z8xZj~=AN|(SBGsky+2$Fd>u5r7b-d~DGoTv4_PL9fSLr=iv5{`kj#~Km1s#b
zT?U@X_N%GJPZ2AB;6y(v6nPfJfl=G9Lh+>fRRQP6EHKpg{ueG%yvASuVs@YZO$;rm
zVl(xo5f4@wiaG8Q%fB0FF?U-;{UZ*gXClTmdD@cCS=9d#kodfw1Hy*aDAM*wAR>xO
zqA-XS8*>Sez|$}j^eFiqfLqAFOvEu7=#Yy;GwNu98i;HRU<*BK1|)Kl3>x`khWB+F
zn9~V8Fx;RIoO=Z5PH7RXp?_h~1sb_Js3J0aJ?{jDbpBXZn$=JcL5eVDDd#16`0&N$
zt5zBG$$?(>+}nG<Z(UQXg_4qwl!$RlR-})Krsi8xL41tIRL@@D9I(1q>LFo`Q+l{y
zRfnO?{4t<ALOlR&Igu!CPOWSWM@m*4JA{!jcuPf2gx-?uK7(P4Un3(PbAeYu3Fx&i
z4-z>zYqxdd#@A^BAW%PD(ALg{!Ne~AOJ-&XY-sehO(0o*jqXMpJnd?x*S@gT<7GnN
z*{vLeYP-V{JYj`zt!TiMcnd=N@!lB5Ghbt<;*6L-3iBs`D_nqE$)E!xgDsONS~xa#
ztC2Kv!_rK`W>Cs#eqjQyuq}m<WgNbdQ8Rw1lyq%t&OzluL@0^p*~vGRY+*)~uOAp*
zAtN2lgBa<M{?-a8rr6jHwdTDso$`PU#>U3Z78a(a8=&A41x?vI5sGm^1i=Hf{aBxQ
zDF$*)J*f3apKB*laCMvWE4pzaBv@_6n2}1hZS8lf3_t(`k2#Zi5yU{um%aLELPg{2
zsquC)OSSOVx^)VM?$^M@K!1kG-VKq(tGNq!kdPjTUFu1yjfSSNEI31@Y&E*`VI9R(
z6N%#085D%-GyW6k<jP1I#G*Ckj2q_+j7q)-^rM*S!PWHiFE9F!+CdB;tyjna1g@rm
zrAwE>W-v6gr)IB!fKtkUpAka(%GMLo($6%Tuy41&JuDq^n}AV<3#iU%5DJx0P59@*
zaJ#kO)C@xtOkKT7qo3)F0b6c|(5-W3^QiMInp}XF%LJxI0+`{LwVTH9E@rqsr7z+j
z1;<+6)DDUggl-E7J?21zi;;@h3R+*xyUZ7HceGM@ET0-hA%!gQPIj2JY#cym!Bv$-
zGuQRnBzVx3x;ydpwxIhJ)ien6z(vh3U={oITLRjbd#8g=<G$3<U%#G7g7ISi#^SKj
z`w&c7XXrM(e?<ll$+|NHce@)W$*D#&oEUG@Vj>Bhl=NgKV=!E>hnH6lWy#<1fn+QV
zRh-$ia6o1y-sF0LZ2U9e)mr#J2E3~N8Ss)13g5wpd64ypq+9Ca)D#03AV*snSr1~&
zzUt0#rT2&3-hiIl)6oj1*M_ZI!_!y<|2}(`NDCJcC>4__2up{jg|+mzOR=Z;VcaqE
z8U`mTr>;&&^axdx0hND271ioZ@KbkpcNiC?5Gqr|)#RwQ1$GCw>BCz}`%Zi*yxSSO
z7LQ48QmyHvs-DZyEr*^0k9{ZTxC<D`c%Mm3dy!xJDnmv(%ik=j)bXi1b6e<AI?=#$
zTk$2I^WywrnZ06pcSkQ|3RCfA8FJ~-WW^b}$ILLx_d@62GE(5%yacBnf;@DT*PvIf
z49rINa5G8BPZahdO?;;mjON=tw@^?8w=EX|WWD!{+YWYQKo8Em*#I=}^1RJJbEe{f
zMNx;D$6!tG4Su1Qz?Fp`qKcf4T~35NPGsOnH=X)*qY^5MfkMz#KgVxk;8YI;xB5JW
z3s)4M$N^uUSdY9Gzq54d?Q$zhM5b>P#8ZV@?=9TEhy2Y2M3ue{W#FZzU8Riq4%wKQ
z{Adczc%xweWzN`djBhVVNNfT!vk};T^~zAMyV^50t&l^B7fnZYLtoq)3A3wn7CY1b
zdI&t^B<5~D{v6sLvLkO(o0g6%2yRhXuGtWSeXZnuSorIuK%srxllZ$k*Bq0*mUi=_
z?l0ll(3)X?@P}!N4fe5!qVjWU$y)2d>q6RX7_U*#(Cs4&9$)E9a6)pH8IRZclW;O^
z3|Ur){+1+MpXS6S21QCsGNdR|R>msJ&#YXo`QgOr=Nn8yR9MAR#l_jZ+V&)*`@fio
znEt8ZpW)x3t(P*OAF1b&-jkA^q!k<HZxp88GO&juOlc3t9Tmw?8P1`FkIT3NczARk
zn%^j&ojzVMTbT8-bykF@J799s(J6OGm@Dho$$|y!0;!zw7?XVfRE;bpK{_2tj&nHN
zJvu0~+Xz)%F;rn241>|Lttf^D{7aNR2c9%o9{x7%Q@8Rf4#x(=sscC)JSQe6;SJ8F
zF>viH^k=R{(9%Q@V7!+G?#Qj9V34xz@j!KQY<rfj0cCzPc+o!5oVd{yBlR$wVeZhM
zwa~!;PciVO;5mri7d0vB>FXoYyyC%4v7#EKCO25_UiH<J^*jbjnh$wvxQp=jE&qyB
z-nNjUEH5vgIC(lYUykbDgOQw(P%a1dcHtq&z(}IW)zI&=Y@I+zndkyko1b&D@qk&`
zac9-eBB5sssFV)mPUUlX&~c{iVuZW=#>$b~jj{7_DKon=HOP$ry3^>e3<#zs9#EOZ
z$LB1J$8~o$Dap!u5}(z>LE!;$^MN-gBIi9i;O1J|lbmAFie7P32Ml;iz6@C)m5JGs
zw#A(G3Ml1Ux3uusUFMeeU@a!LJ8#?*ge7uj;-`0t4;`8x91!}lyv|KpvNk~F<_)ut
z=Mx6(a-yO*d?8qB21B8JsDmmpTJaZOaPOYNCtM(|oED4aL3Er2vk{Sz?Co||;P@8j
zF7)MUm#r|u%$N<7r#+Bs9{`emDJm-J>OkC9o~?6bCXPIPlX4mw8ktQ!y0q&du2<E3
zX)&{<W@Rf(5MUb}1_n7hr6oHZf5m+Xx+#{HNVLxTh9Q1c+%u)#Bk3TM{$0}SWBS34
zxjW?nbS-=f<pNG2KIz|C+2f{{b=nSab<N#L>b&bB=oAuQ{xSbh*aeS~iWOGgj+2(o
zkx*-WsUEe%#M!d0a>>|UH-SCr11VHLX7EjC_k(>TLCgx?-BW~ci;1bUwNC5&pcRi>
zOl0JFH-WatnvnP|1`)*1XtvvFp}V6$v7&r>G8uEfuG{Nz@Eky~aRCDG-})m@4A^+g
zmS4>nW`Yx+S;zCU`_T}+fcd6C*t8031V4QEP$4+r%FkR2Sy@>#oToZ@wcze)b`(7?
zMh?p3v4DcG^qgbIWl0Nc^nF0W3@yJ+gbzV6?0Da%vpLgcCS7B;PUF#P`ygjiAwZrP
zqo}>LSt)lRD}eips4j!X=QF|?WSg;(uCz+v&@Jy7eZiXfa?M2#)^yl7nqN>!^1{Cu
zwSEMJR`9?n%WE7Q93I3asce*ite+GS?qgKXg{DmVj=d~;<LnOuwP>r#m8ZO@hcA`y
zS0;*nP1H!4<95vahD!dP?aqOxFL@Z{w$}w6@%vJRE<*=2rYm}SYyk70DcF>i>J-~#
zJ8~=DB1}Ds^}Gi$ygR)A_Nq18p6#&dcwUTr!4o&<ou`DpzW&w2&c$61wyG8hvG<Ur
zhxw>yXn>m)vZg*ZI$uGw)5|7|jBrUuS9M^ac@z{BXu?yVY(|0@Y5FrOzB6!0Ic;KO
zWONjbWKA&I%W_^C=<9m~&-d#T17omyXMpGcXMrz|<B;kUqZ#U#y?_F(o56&?F;>{i
z%c~f`pC@KMg?t4})y-fXr0vkw2CPzffR4fT_urk_`3;iM9&g-)I0RvW#WSO>c+EmS
zai~<qC6ZX(?nA{e<ra+MBjJsjpOJV}N(#RX;+DB>Hu2y}tSFh#ii4fO@cmqz5eEPO
zIn@n)6?x^ufj$0-j;nNrxh80`JIHL{k_YQFWY+`6Cj<McuwXh?iA|k|?q=&YaMEEE
zZv`T76z~qenmOlWHg>Mb1as?iE$|GCME4~$Ec*$7+|dk{d_#48&JFxe`cY{c^W~^-
zmEBjI5<W)1?MTSyxIV}=8@8wycT+&swKA#q-TyuXo2IMdZtCIb@)ZZpPYg|?Y#;h5
zOH98Eme`ZgBB+^1<XZ~cai%E^y#WNY4^?6D45Nvz?m2MWJt5e&TM*WOmJGKBvijNc
z<&bq6#pe=NP|<En%W$7w^eD<BEBwAVXs37G#Z;Xyy0R&s;|#hhPD?C$ah|X8rIXq?
zc`xShKrt@ii(uWvT>UHcpN5DB*VDhdar_Pk<y)L|{KcgERc@x2v9Q$}IK_9-zvbSl
z@(!UPfR57=LFa`Dr@(okOSyK!`G9UaCAj5<jX6`PflJ0%8z<iSs00b5qIgjV9?XjV
z!kPK5j31U5p7^Ij=RMwDTlVvO!r_lEjH)_fB_!^BFwz?+h)&kRiZc@JiQeYa2N8ck
zvJ0iSo%%IcB>0WL`EdX8G2pJnBW=}Rsy|zW0-8VgWJWYH(EZ2|3?^HDK=fK-C&aS0
zP!D%eW`NFnpw*`RbTqzj=a<N*%@-d6s^V+cOKkL3@pyIKgVkxSoy8N*5UI06W*?TU
zVkNd*Qr&D56k88NyHDRqOdXXKNaeyARSZ{;hF&}4%yQn_IlvkMf<*oMWO>P2d#je~
zI55+!ps0AjY0}E+Wp%C;*WVk-p;Di_VBvBp#>A)xBf}$5?N7Mo$*pXgF^skLF@VbE
zU9KSM#Y$~~u6y2rrg`Qgc-H)@5|l?ggXdV=pENgm(>qOlE+%$wq=nKrlTx0Ep@9I8
zCRh&}B(^(S#I(Yctw8ol;^E}vWZ_tr0Y-5w>~xrDG#Tb5pZK1VWZA5z7a0+8PINr-
z(mo<LA8oYNmIS2iRR#tIRM+&VDE&f<q`moa#2U^63{nxnw#R#$j@QHS_G+N3ciAa=
zkr&mwTFNCZUc5Mx>NJ~rc*4d-<aSHd>0zwi1Ng0Aa(wtTJeb7=tju+KlRBE9=N3cw
z9v4B5*x#WpcX2kCYCn(;b4{p%6~j2Wx-DB=LIRnkCy`PQFW#TOQ@nK$OWCo&=|HI#
zBkYYv^0^$j`!;!7$ve}coEW?wiq9Vitck6cS5pffo!5yS5xm{vJJU?HNcrpCz$>39
z!SwHS6_u4LH$9T(L|U7wzJ7g1+{a^5sEs(r>{<Bu`1TQVIw_~Vz?Y~44y1lFlW-Z>
z!m~Ai)pv3F#_Wh!oc+FTerzTZHDHrzG&<?sN83-wV7yJt0R)>j4P^k|;~>$WI_G{s
zuY5%&i($?!)~$1h(mF=bBrstp*~~b6n~;W@9OqPA$6a&9qUmLUf?ER)FO>5;g*2pp
zHwzGxj;&u%FJ(DrfZeoOjCR(BvXQg4bcI#eKWD9?mh>%h&BtL%Jq-7R<KO_a7%Jni
zfvSxNpwgVQo-qH{IJ&wmp;TT)3|fip3^8)jrlMK55vbAAvpa2AIq|3K@*Ut(Q-Dy+
zgvjTH^4cQeX&uquH%{;L762zh9S=}OS()0FgO3=lba&)+#9k@HQeZ$nTw-^8fDdw%
zG>LueRGS?)KP+LS;y7q~>nY`o^~yuCzhRInP%xz-F=%rW9wdx}GssgD2jDwm72r72
zGhE73kq3)~#@LiqYnf>N!o(+rW+siZB5@F;^;C$Y8KQ5;hEXvl;@N=+`08^UdvH{A
zbUVo#GT=%#-|DjQ6y93JG=H;=feEI1O)Equ^Guv(sl?c(ov7ZIl=Q3pKx>X05P}jW
zL21SXM4sQov_lzcFKu?Mv#P{(+7wAw0X#vk&KZIYeGp=ZOvF<vYoz3$Df|RF$AXlh
zyXitDLj3CrRsV$>;#^_h#baYIgzIrrRJGYh?4yDBXxG%%YBk4&h3zkJXJH}UA0V?G
z&lwFIN}6F-(Z}P^oa<H0bp}>^ocNt1GunGKOBKWFM?@#$-SfegqB$j*IP@uVYZg5V
zsHm;7D!{U;f7PI8{59mnFx=aP$&Dc`G6<h)s#K>4W39Ji4Z7o~6mUc+$$Rx^XVWWl
z;!{dqO7d=6>NJb*NOQg>&kIIPm4g$Ln1x|(t%X)ezqIRy2bpzPWg>%+sU<EZ?eLr^
z)lEf7X^PGgaXJ^apP#CFnZp1&`4hAe&mt5&YSJx3snNc@b?S_L+os^hD6X}}?5yg(
zeqB#&k%^Nhan?x5Fo`Gr@Wj$6pBSG3rzmb{fxg*PO}OZ0V$4j8Vlpja5yA^lGHg+U
zt!$#BHWL$58`ON<up2muzlvqD2)rSz!DA=B7~H52hyhC7d~a`W8<b<*4dSx-A(pX$
zB%m57Q<!Z<rukvHKpc(=W&qjcL00I>i)|wg17a{vPzPv1o^t-(6?9j#y8spMA&wE&
z@S|!+-SHTJlmJ;U+_Cd9<BX_In|?IV5p0{7wqzv%8nw$mmNWcx3gyDKm)9(>piE?k
zGKilAL9y}y54HIj#6Cf{Bd=R@GJQ)N)$T69iUL6B?Zg7AG*~P*-;Vm&XdaoNz+Tw*
z#*zt(w%~=HV1khw1rbxuFBuE4H8+kQ3Zy3~4zZUW{(g_1*p)P@OTEQYBPP*##NrbN
z2umB(iaC`eGbM8U&Gv8omOwm87KT|Cxrtr8!cZ`R$xfQbID`_1ktO9}Wvt$K$&P_%
z>kgT;Ng3oN?XV}ZC99WsSQ8VRx&RFvi@f~&1(*+GCS_(j3P5)_%s8Z~%8Dl6%4SpG
z6xh>7>ctk(gPneK5d3o^$3LVd7;&47h=B?oTkyNtx|*6iq^s@tXjd#xA!7VS<hxvQ
z9}QFm&{>KprK$XaQ+?&D<Gag>SENdErfCH_EQoWPf9#@9irvvo|DrEBw`O!zi6AaG
zXLObGb2{T52U%rh|N7j@*s2$FbwevDAq*TItmm7WdWg>l@SUWL3Sy9oGqGIxIy234
z0x82Db?VdRRieM{C`>xR%S0W_o(&t&1+EZ4K58yhfj($EvRvx$D6Uuo=YL5CQN<!2
zZf=j^CoHfBu2-$RRB)2RtT7FsjzMHZ((P;4t`%&4o2-S}Xt+>pfg9-Sdg4qCD|Rl8
zlQk&L(lpvGapuMa?mNFRE^FihndJtzsh|O(gl<NKJIp0rJmZ@BFeKxKIi7rCY}IVl
z`V^?yX^^j|fpdonT5T1f8}E#pdl;%B3^-*?YU=BCBn_>x<Jyq^Hw>Fg<YuBxSAY{^
z`v&UDhk#IfwW)vuw&9=j-@y|M*6A#`SeuJVx#S9<U4(|213=n+D2F#eYfP{H!(4fA
z#pB0lfH2%TIf|<N&8t_tk)kyiI^Zy;g)C%R$+mOkQ!X7JK3!*o^n{qf1+-0dayc<#
zc-Po;mc*sD+osAW%L1pWgBKWxLldNO=Ff*N{KEzK8*iu)>@e{Jc$P&(HvEC!H|p4A
zBJDbet`D`%G&9HD$`Z-jX($h7wt-b?4ozGRVxB_$a^7Ie#6$5R4-5h7jzbmNKsJVv
zh#l3KNBP-il{<T_Jh(?Ci@2#Es%SSOPCc_QD~l!9M0{P8n{|+s%3QnFJUHdSN=#An
z@qLLimNV+qv%#Vc>d^K!8s&Lj26>upYm@|u#?f(c?no?Bt^2`%WO16Q4o)!ME@FEt
z(=#6JOjl6U3jw8RU@6EbDsBa}pzl}r?mc8kv9h7TCw>7HnlsQSi=XpK=ZIg~am+6q
zHdX@}+WrTf8k?#(GnsbG$5PZl<CoZ#K#xWZenBC-o>Ze-k%!LU4m5wXYwGF>AOm`I
z<_6*-G2^i;tU!*br+5hPksDbjEifR{uD^$JZ`+aUAKZc^EMn9E!VypiN4Gl87#XBl
z!!3?T7%cN}TXR84)GQysBnvw+z9P^qFCSn2gt9yCM<J>{OJPhb;oL63=CT}PB;HRL
zmvIK(muk0XmShl^panYuzbHgR^{F-m_h6_Uj0pF~Mw=APnRiKfu+psa(A|HXks;LE
z*QWut9;4H}z+4{XnP!pc4o~7qg`o5EtXRo5wkuel#6+pGriOtW@MsBSZ9d6*ej_QA
z$u<&>ZN*JCP|S6xUr<zbcxJELqiTiKOp=5oo4Fvz0r!RpL8sBOV;Xp!%#|yQIE|hF
z^Rm+Yo`?jNm}V|cxi1zeiNQQ-s$y}jVqJ2h2ks|7BEP2v75$AMvT8_ZBStYURH}Jv
zW;D_jF|<^Y+eNdG4&p>29l?Q&Ro{%kj2Onih)%O*(e6iN|0*A!fU)5dNZ@g4IB02U
zVY(>2Ii7g*;1f1&$AHBj2ObS9w}G#D-7i&b*@<n9>p1e+;TqXX`8OUh*A4Pp6=Ho_
z+3&|@+klQ>neDZT@Z7mhE@Bp>)eOpQoIr2H8#Ddz#PQ5j*lAsz8w(dRGMzmjI{aj6
zA~p9~-s6#OL}erMPrW$eskK~(mZyKu<#?%b695aCI{Y9NZ8garSAbaCVe*||oNiLj
z7m?U~S7KbWX%NS0<8!eK9<1cg(AG0hTo;2(_wi%&%XG3iMG!^>o1F%kM#};41XPdf
zl4&-S&naoyB9p$E9Q7*j2M%5s3n#`}<OTq~u<W=iJkq>p>$iZ}kKZ6ZvG3s;MvvhS
z_wf<dn2plFB@Er9K9J`8Q^Xe4(=aW!pL3(DE<C4``aE~XEPJBcZsE4dFSSm)1#y(d
zds8;c=p2m5`0;Ri)3vu5nL_5AY2=;=x4ZJRk#gR27USS7C|VYk^B%+h>16^(SktCt
zzF&W(>mFphU>Q97t-CEMI<`c~3GiM%cr6m^H-sL+3Q0py_m7E=h<Hq5f+>z?ul;O7
z4`6tCfVdvIj+62Yqv#>aHzdcZI7e2MBnLHnNpNp&B<7tv0mN$e0Y4%EA!gmbWK}U`
zybXa}KLztCj>9NOEUZVCb`Ct;Zv6NXq6=eR?ym=2d5~thuwJISwh^_E7obQ(Aqno<
zpmUr)_P<1XePAkffE@R(<KcmpSHHo`a^K^}m!Hvvy@9<pOCFe*JaxgyyyG$BxAAmM
zhS{J?*d3K4)88FxxVz}r@7^jon>=<TB6r$9_X)WtU#R!!szb;PlZEq?bk0QS+03OK
zAI0WAi)lJUxoVo(Vgr)20B>ticV%UTUM-Ob@>Vu3R%&f%*al#b%Tb)dP`VWuuB*j)
zk37D571_JGY!7sTZ1!vXI=-h3Bj<e9QZA&rakW4;Ypr5dyl>y*u+gfamAljs0XC3R
zv0!F#CT_9Bx{a^%Jf~0du*u?t31c5AdC1L|SihF?Kz)H?_VL!~6qTD$ol?gUusX8o
zeC~G2ZIP$y93h6|1Jq-jIa6(}<2We{Niv07qdp6ibrZs-UP>%Zw;U9OFWB}7xfsNz
zYZq>sx16YrbL8VVXc0^)DJJU#IlO(r%#eMzF#NGiUhLZAJfI^sd!-;lR!VRmv8TJM
z`*RHtUs?^5XaRkr;lLsL16bS_805R(wyqTyo&S<MCsB}r6EI&fccymZ=~MB=E2?Fn
z-+!Hy2_;I4z7_WSvGgzBeN16v7+r;@w>Kf%BnR8OaLKA2(sy-*xaN&1l!?>Y`zTri
zu$_MlMG~!di=VdUtUvETE{a|Ga$lIgamh>`cf%k~v67kP89+dYp{Wvi*CaGbg-}}g
z$IO0FTi8enpC@}>`&bU})YaDJBS@>g#`g$<vnt675-kY)M@TRGoFmN1V`&AlxSp)A
zf*ELkauez8ji*m@lkb5Vr9|l%W~nX}u40n~oaC=^WFQ&m94S`)VrD<A5?v2pS*a&`
zdb7qdmrxou$){%1iFelm`b-0?OkBu|8zyk_(cD1SZ_ZCJI65{2@43W<?!Ouuk-X77
zNegC9trL8^ypkfl=IPU?!zcm@G0$3Sn4P0OjsC1G0RzhFt~rd?#^YVR@lfww8{%Qw
zzCCf>@@gU6ztulT^#WeT5smCcYo&#byFDcOk_~orW)vq3@;T+#5fnDo(N$FvoIY+1
z7X8HH&lCqpW})E$+9Qy2j=Y@!sl~65I?#>$9I~7<vp*pXLZnu6_BAqRy^b&mxxj7m
z01D%@%a4|bKZoRVRLySbrQ~#6oDN=W95zRSjq)M#{)S4%0J9QyQy|j)Q#Lqqi7VyC
zGc|%eR?DRSg$g~Ei+$f#$+L3u^3&y%bGva_*a~jslS!jP*l`dg(Ha)ftos`*q-L5a
zl>SHt*(K5`s7lp(A;Hve4;f_{W#<4tAkS^w>Zc<=Qz+LDn0=&BlFpFthdJr_$)j+Q
z)Nqbm-A|m4iNoriJ+Mx_(x&_K9dVQ;zKN$#6Z7Z1N(yDV>uSZI5bm_U2;<P#Q}l;#
z@KPu}tXm~Ttyz|c+j!FLXUTFWrdO81>QPJ4EBJhYd8yWi+%2nakC4BL+~(y0j-`|r
zKC~FLoYHxb7RZQ`T!VvGD9r}^w`mssH)I>W&%gf-{}37C^V^H<&o9rdAQ<I;{fKk0
zg^{=V_s``M+`ldI{`}_IYSQEQub<m;(!^lzzkc4{Ri_#3{_~q%oBwkONZS7|+=Y`C
zIZRg}MGPMv9<{P1l&|Y(FekMhi4zIv&tEU0+#Z?0G2$(B-{WeIy-nmP3gui(%FJuz
zRnOoXfBSeN(pd7_5b+TONj{YF_IxWS;z8FCzDz;ZeJQjCf3_yzR@M2?b$GWe#KF>)
z1zBq7x@BYqTz6Nv()-U+`2GLR|KC?slzC$$YI)*2e6kSB_3ug4C<Une{9^i$n*o1}
zI=um3xDhbMyPDm%8dU8D_HA855&uY(g@;kH3u!o@E`r+n`{j6{cp)D@zqb%95QQT4
z0EMqP0b9mMRIdjB*HsxdqglEYR?ciPr)giIQxAmUc+dxMf;83VY}u-E^Tc0!XxpYL
zBBADgY&Z(Tz5nb7n5h0IDq!jS=Pr=_^Pju$|I=NNdntr>DoQ2o3(|(@P8x!9v1dHV
zU!BooXA5t4KR#0A13&J}Hd;7lJ3!jTK(k`Y<_QJlT^};NK_&D#q`HNWW2yrP9&m>9
z6C$_HW^@8W*CDKp^Z+7*-_aR{P8HG<A}xx)1-Ft{eF*xAUWF?G;#Y`WBHaRlSM=&v
zJf$TCLb0@PD4xS?|LY^cx$XuNg;-MnNL%NQlQ*E)CCFY9NoMQz6|qY>Ny{9rr?Mjl
z2|2V`<JUsorc17{*3JChO^d1xo)?3T3&Q+a<<}@fv@Xr@n9YP_By18WoY2eoWNC#-
zFtx7}wATyW1A*B`*vj}sMJ0fCmZ=Y`R@>52q+QG;Es^#nAPPh!Rv6?0i9qQj^Zw7@
z^gOv(AAvUgS+q}1RM6h-EiIInDC}M}q*La%86(nMO0%>t5m+p9mU%O>YP`?a{Hx@5
z;<hA{3Nb7T5)nqxb@^G!+HbUHz<Ughmb*z*)Ow_LJJAHcXbPq3EXlL9qR~a*hIx2a
zDmB+>aq9ajF`r4x-(MvWY+;7C&DB!EF}nt6hfU90By+bD6do_`;={B;L~Lxi9Wj%p
zPy!o)AZ<I#Y$4Y|#jw6{3l(P@J_MySK4bu}H+YY>9m!=U8FXpy{^Fi}_a(4tSyt_|
zzOMS`)fZ{6j={{v@DF+?0h58Xj7M3K8waOOkat6YhfXjPl5ra&+LGB7OuMHOBZ3W%
zO?u(Daeuco{aJSef6}XZo0TLF6p3ZFp>U*X3we(ZG|mj09Wf@yBt=)~;Y%lIOGo!w
z&PW8y%-2V+b8<wvjA&PB7*6}bin6ksz=UiXuq0#}fw8ZrM%Qm3ZzewNCt_y_NLDMA
zI(+lD0rYxuP0v<OZ@YEuChaP}ra&zVZplr?$)sEm=zJkN!a)1$AJoXg+qKzEWTIgk
zdaRG!;{a6vMeO9>L*D-ONOC59xX_RLn_eUPuyLk6xbJblKL9)q(m`9t;)H4>4_&&^
zZGWx_rstldMZqlUXyp=5-iUJbHTfbn4S^r6mQB)rlTJlLrfo0F^tnfF0{}uv6QBaA
z34RRU&|bWaM_ZEAD$Rws<N3fP=>GZAy+2PyNeOno_=&c2&>Py{X*ffkUF1=KE|x4j
zn8X8$DZT^=AAZ=>Knrle%}$w3bK7!*_PdUgwAWWXKVkjnf(FQ4W|{wamTDo1eXWjB
zzZK5J5t|}TUdqlxN+Mvl<+E16yQE-Awe$>akxAz?&AyL9WB)7)usgp$lGF&|tD&yS
z)$M?r#7G*QOMYJ|n;IynZU8;IvlcPfO}A|=`Hb5nX}bI~sandVq~u%1rvV53-&KVZ
zSwHC6mg%*E!~@~UlXv$1&b*{(-#}JPjRm8DhUU3VO-&d;KeOR?V#cPHE}EpjFgFcm
zzsz@C+GivYghrgB^Qg5LX^-LVZTRO7k*$LsPD;93?B9yy#gxobwD*U<R{$}CkDlF@
zwu=0|$Q^9$q>SwNc<yzuE8FuoZNb0wrG0_OKihdrDx@96M5jK|Z~6C9!bGe>YOrOK
z-rvr&`*tp~_wuUunGh9c2Cs(8VjG2W>*QG$hg;tOMPEr}w(E~allu~bAl70@BS29`
z5`ZkW{Y@M}3K2x*dyk+7{OpR>92kh+7Q~RD9j{|IL)zM3Jb_6+sMuYP{M>Mu^d7+2
zOs~0t<G28b|7)kk1*1X5=d9;1-?v`<Pd6Ze{5gJ-^ockUnuC`g$uN+92rn=13zGUH
z`QJ4APYO)l810N<@DE8{aJz=I-qoSqjyJw?Ux~c>-#-^<PwU_RiVgaovXb(jyFgnI
z|1Z`;lms*WYEtOy*L%Q9-1&&vQi<?-2_{B1D{w%kCxDO(NN0%DZXgR`JX${TGwrDP
znFN7B22d-VA;Xr3zXPwHvBzOyVGSS=k=7q+U(nisO7VLDRbK7ANeh}yv;&e6DMY($
z@Qhsatcp#i7n&_{@3#S^Zac*?BTc|Ef-XbWTv4hJ3T>i<YLGiWVA77>h#N?4Y^Y6>
zUFX5a-iWwWYSmzr0j`AxIOAn?7eQ(xY>EK4Tr|AZF@iH7I)q^{!X_!vYXt$e5QDK)
z2U<oGGS3Nq6ZyI>RMI&r1kY$S(`Qf+HLwREYV?ileyKvzl`^jo*;;XYR1(?Y-eI6n
z`J^9r-h-Dq^kEsjD@io=^eZ$23m>kM;Nnw$DKTi53A9-_$hz%`WpHY2EN3~$5djnP
zjTMetNEd~64YyA=Y#O}O$b(+J7#>4Q+XTWF?;!HGJ!h6bb0RyuNE-1OfTzAiFaXtz
z>-ln+G<6|g1n6@8phh2hBS^<csvv-Tbk81j%rlxo_~IDV1SimNdW8cg%a4>VfINMZ
zeWGGoCoR!4bwgh-tGPQHH?IBuCaEGlSgd9%l6Xy_Fcj~5sj?9dRM&LY1TE;h7Zo7g
zsHkFrArJJi3GpV=Z2-L+*wP4rKFdsaAg51%yZL$TaL*;Dj7CS25|vL^*xQ2Zkhn#*
zlj#eFL^N<Uco=AIn2VkTahint>+Uiov$8_5IO|P+JlL9&HV?M8D;gWo*NmV_D<%b&
z(w0;;^D;hQpi-B}_1V2ZCp#Yj{^h9C>yp1g6lrPVNKPylI%sf0!C>9-UmwUBw&cyy
zDW{3S<54AM1MO4vJ2ouW*>Up|Yh}QvB|HNHq25=!xQka_O#HZeKxtk0(K6Y{$k2#L
z0j)S@zVHCPi{=GOFUsurkazR?zZg!)J)2)JRp5LsvSd5|``qB<se7{-kLai$vxDbV
zCTT|(Wwk$s&3ZszlEJev;@6SM*g}N}1VpVkjvxQl%Uxbx&IAN?r#5Z#h-+i)NShuL
zBiB!)xI|JuY!ILdlkM@{giN_YOz0wAq_ZE!8zYc3H8SzoPyI3s=AfnSI6i)I$h^Lv
zV$y0{eKP8`?$oazjlHc99InL&nt0H58SX0&NwFv)AIPv+2{@-iysNWNjwm(Pmz7-v
zDW^13lRRqtFbLtAJ%fXTv#Z&DVQ2BBXJ#rliwV~o|9P*5OQv=Q$Scoz#Kpx0*$oON
zI-o%t4#NygU|ql@eumv-EVS4=e*Aa^M5&~{UJ-(v(=X!ZlML#m(1WN2pAQTZq-*=>
z>D}#%R+D!_R&?WgN5@;p8eZ+kRug=1|Nc+x%?ohEb#ZZlh@t$OGiz}Rqd=oKv*c{*
zPR&S9??JHyG%oSb1T?!|!N1vu{>q>0EILd(5T$nky{k5~hlH~&l>8W&nT=9{F~ZTr
zm=O%z`L`}Ws!3dUknIJyXv(3f>N-J9Hh2lSP9u3^#I&Qz&p@u(|KQ=nt9*)3rIMdN
z1pa6mx+UuJV!tn)GNp{E0OGylIgBc*k7!lhy!pP-1_Lm)kl2EKmMU_MdJI7tC#gY1
z5t3q>^2QjP$s%S6M;&1fg~oPyR63?^-u3Db20O3agdyXam5KA#9JFmnp-5ZPkmB1*
zKeo2EYW5=(_g06y%jaLhNDLW-8~aQzMiziR7!m;Xo`<$#z6$r8GumuB#CIFjqY%OM
zp%0``rgab^3<=;lL_=iL(}9d+s;RZwm;Uo8PUs9?eaDLSl_IE@nfW%D&Pu`<HzxdK
ziF6|jpE^4`y+UbUMJhm7-WU}Zx3Z)K9L%D-6THAXhz5PQ3G(yBqEMc(h0EC}H?5pt
zV}jtqKB%PG*w{ouI?ma}r5bG_3%Fun&b_0j2}B6|Cq@n>H3+y;_->eTt3i4T34rY5
z*S34x*xE*c@ia0vCI_(-IpDJpfxLhSsECO}7@53y_ACU-K7XG%(>2|<FBem$WC=o$
zT^g1<FP=Yt1#`dryu6=?>+r8FeR-Uy5T+PW$(s_nc!>2U!`m?sw$W;Jt45a5@O`%R
z&<43!KrDu$gZGRdkq8gpqQJ??$pq=~9LRBbIuO2teJL%!Cu{lo;8jETYsf#4D}b@c
z3Alz9>0MpX1Q4Kfi|krBL4mY1?L#YY7>;8eCJ86QU!KFr`dh_#rvMjx#r4%<l3!3z
zP^*WSt>B4WM$uM<a`aH_KzsWeGT85G0#!ha$xV(zEhmY1v*BdM>oIRew0c0Al85AP
ze)4u+-H_Y2DW1bvX1y5E)$X>&nQ&3`?-@L+8HgS6B!jth26fJJR`{p;24-T)U7LQt
zAgmi|S6%K7#Ry{`Hrq>#qCqpIV{p)TyaD6011;HP=~qKm!^_)SJ4p>Q6k}jpLBI_J
z5+R3~q}sk5a1Zv=|0MGLckkYP=NAsO2PW(wY}W_`O5c9ialgWh=V7%R6%0`%^)9Yu
zw_9|Aex{<D{qO_ODc-c@^StyUu6txwT>oOok>6dI6?fB*eZVwgMWN*CATX(Vp+Hbp
zB?{M_G9T{GKZ`lgeX32Y5H+}!gISFT%!2Y)Z2G-r#XpiI4t3!ug_tO;)lQ8wD;`A}
z#ABN(f>px6$mqMTtH1vpG&7h2-3@3}=Z@mmzb}N%ffX8zG4?SKrXaASs@)AMKM#;O
z$pY#DQBDTWy>S8B95a9kBnu#LvU$svFmUmDiT9zfBJw9bXy&e1zB~f+3|r`4Veb68
z<p7ptv~#4SNTeXK6XTW#lcTJB`WUv#G^`d}F=m!5&LB%TD%y08rKT1Z@-Rn)lnYnj
zahuDC;aO@s#zIi|7+{2g14iVPHpCbK37Jx|Hog(Bd6%Ro(EwBqATkiESFgT|m>6@l
zy0P&Jlu@OTQ}E)dvcc`T3LS%d)4I9E4NOc?;9Kc|Y1YW0xokC$&Y#e~DExOm1GKde
z%`ZZ}zNpZ3wHp=;qr*`}90D~MqI;u(raZLz-cjeV8RtPs;jf^rc5IFx*c|W1a{a0A
zvUf*E2txA;5*`UjSn7IcL7piDEs}@aS?>v`?;eKJd6yumP{nV^gC0B3n%k<ryY?Oq
z#%QD=-L?ZDvl>AEVhdg5PPu4>!z4%{ikH+jt-QBevHiISqFx;aX>0H(55?y$%t)Y)
z%Px)whXwD~`R-~rG;n#hFQV!yD*#*g?f383qw!u{aIFKpp^t3pOz@!hHSqPS@tsPy
zTy9in4%zhzm;!~e;%@vV5tcEyC9rXDT)+}kK$L<CZ?b(!=>k*=lQHl8CLTwvrrtrs
z7Np6VNlw?Vhg>xOohneI)0BQ^2S}sz@W`2tT)&MCwxd^@QMO(hm&2;4Sbu1sAhz;C
z6=+Mh#(U-cyJ)z{gGLNNj}h{XoeFrA<480+^#{o0BI!+~6xvu@>w*ZDE_Ca@deodE
zJY{08Why}X?uvTXV4#_)X(hPv>a|QDCh;($jT_w9#|8VpuehC^ZEfncSHlGGKrl}=
zgGRSG-z8|MV~~>r!IndDl=5KhpSTFU8l(72_!C~Z%pkf*VKSr=x;>HDdt_c(pPCYb
zkv(5}-=;^A$B6u;0(i|wle58k=72n%cMce4^&na=cAj_qbDXYzgg0nZXMtPiILuf`
zfI(pNLSTnT#CiN#dXU4!6@0;tQ(+TCnI=C!UrTTD*RQY0iTxr65C-I-j^JAU`J)1P
z%V$@R`I{e~0IjjCwuw)_rf66R8d|UvKTz1t9#-NDb}+}GLy|ams-UXKYtbu*4!RDo
zuoNQ4h10@^8JB89NHh|&pc^H_rN^_KD5$%nPs523l+69@!~~Dg0ACkTuCzQjAmAOj
z1>nm2A?s?i-=_z@pDi-x(gQ{XNJ`%cHGF%%j(SIf&3K50U-USI--1BSHTfvHcw1N4
z24Fh2CezB~{AGRpC?LiM$OJZMel@U<8|7%W--_Ld8O<IrEQ&U6Cn*W$$$IQGFk*_D
z%eeE3+dsE62bZRYf~dFh&7)ckWL4n^=fv#B<1Hr0iLokeOSHZD)QdkrXs9Cf&!Kow
z4Gn<!_~sn5+$#fZEf__=0|~5KFP6Nsv{ImmodRxM9dN1IcUO3ziu`wZd06m<wxOXo
zyzU^1xO#&{kXZ09FR>fsF~yHe;XLwM?Lx@35#e8VSQL2;jI)I$sN!paQ2h!$M_tIw
zXb;VsSz0O(#Y<Q$c|OVk13(Ew>-G3Fo@9vRMByAcQ%I!8VQngeC{mnUCgQ~b2`PY4
z+>F7naD^rIjiZ>Vxv&=2R!~IpZ6aopnDqMiQ?@D~Ab<hhBv_FgA|lNl-@;%s#9loK
zd3gxwJcIoe7ZQ<+IL167l$0;WW=8Ami{E+X=NAP6o*H6p#qbU!0fbV35#&$|r|V+8
z;cD8=S2!OGeIn<{jQ`;<vZoj3hnx<wv!7=cx+RTWnK)KlRP+^VZDjAoKTm1JH}HPP
z-JRx6p&k%q8tU&q<mBWuJv(cFCA50Yni`BR$Kvrf2*}P7Co>YQH**C|dpko#IuZ*3
z-gEBAllee|dDCw5J8RrV&Q=nB(jSGAj4aLc4K|~mnIZO+xxA#-k!`#v?65lyR%f3k
zbqSyTi?B<<#wkjg$RuqsXOfU@tokjL7t#z~7y<X_=;{i?0GxmSc1S+pN%*_62;BV8
zHQkv9S@#Gad2FJhEl~^5;DYLc7x}Ca$3z16;izazH`fW9z<GU8Na!j8(V4<*a45B1
zB312&|MrZB${(AcV54q7=-w|;ffUE1GF1SD@#&WD1C5DiA*uorY_O}__{2mVV%KXh
ze0&31J&-f}By|7)*@Qe)H*rX5x~+{vb{)O;iBw4)iRjTt0VGW+R>6r;f2`I14FXjZ
zL?-xst6QMS7Y*{FwjNQ;2b*`_d1CJ;WQjYJ0XZqvFd9mU>gkyP(iv6To_+qui(BA4
zNmNy^DJt&;t$ssnDH`5?)a@TKc#pGK4g|ZzJopA7{+Aptqrv8@qY*f^i>$GR`pgWT
zU#Xize|(`?@&rVOksqsRu^X3QO`vJcr&Yp`*bbT2z&92{&6zLXn086xpe4e^D4h8m
zW^N@qI3azTb$IH4Q|Sx2`>31ValW<v3a(OV!Bb`x=wTaAjDuDyh5~N;Qxr~q4u|0Y
zP@~z8IX-?yo^5N$vu-@#i+7NK<_ACHF^iurd~gShf*#BE!*4y_LNXq)fwypU*@_Ln
zqrEgpSR{|c&MVtLYBA6xk405xc31bt3vcf*G<&u`#qO2Fu?ZA5+qFROuRpcRhkHaW
z23dA57CFB!{?HbyAshyGH$12N%WC*)@ta7Lc2NBGshk2TU;z_^EJ+%cy*d<rsAU0y
z)tdGsE);fm9qELixJ&+JauhV4I*Z(#BtwRAMzI1byW}IJPaq=LPItb<ERi&k!tevk
z9j5-?X}(WQyZeYmAOTe~fZtat!RJMCizvw<JFdbvpeffWx6&suF%ikBu9+c<IEckx
zELbFxNItH~(BQy+-_m_V8_cu}Sbw_}EPWvVBlish*4f|XIFOUHzHA}E2BW;*?(Uaj
zyNKa7;e79rkxoJ!bAUC=m7NBI$lp=kM8QqyN<je5hYtaJ{$vCp--&Y>T-6!YuMe?m
zu{lc~ajiXKQ7w8FwJ^EfuB9tR<jJ3GI#7dw3dsTF4b_P1PgPS>ZYA)Mcp{L1*Rvz6
zY0hEuX&2hzth640QWzTqat(u_z(6v06ui8R-H6YrQQtdH(2M;=*9FaJaa_9s_>47}
z(^IJ+vI}npoW!o!f?|B89W;IJnVpLyWlFDuF^55dbvfjzG&wy;^t>1#P^md*;^^q8
z-3=w12gP?eWl>L!qcYVrSR>8wr+{Tgj3(L(sH?}2T#%Ef7q<yRu%2&R_L2$M=WJmx
zcdw)uo4Y<Xwh3|YjRvTWpV^O*FNYu3qubWZV7NG%m}y19;tU<a*EmTcp<gsIGEy>O
z54C~1z3#_fL#y+65x>)ll`A>e*w`d8k>i7$BYo@EE!A)u{_>y-$zf>$8a;aas#(_M
zCUy`weOo(Ux>*({F?*rlhvU%RL5P$fd5f*jEh+3Me^<Pr-=Q#l^Z>3V>AnJ&Wx!Dx
zfgN5a21&UeuYV=?#w-RJhdyPA=zx(IHj36ofVG`c2=C2$!&iHy<ph5tJs&(!{*YXn
zR3FH*Y5|*)&^QBX5yPfU8nru+Umcy+PI+G}C8LD$!kbGi=uzYETR~yt;8e|ufXFbZ
zC@@ulRp^Zn7*2vQ@TMq19CS6P29KBXI=w>*0Xc7lNv_qaYza~#uKEOGmD_XeCXIt)
zj0Wt4CFV;-uKY_-5vt2%iQsw1>+SX$%k5(byZ`WEFI30&x04Bgu%X%UKAnUiU_WJ|
z*AC~d9yUvmQc_adskxNF{x{qt&An~SYC1Yu$juNW30w3UzF=za#cQZzM|OYC+PY)M
zI~ZcsM&%&bCR&@(;v_B#9Nu6jeVn{Z-0Jrc5h8$>`~(j+Ngy^t#RV8&Et1rjUOT`I
z$c4Dvc}RJ1`_q@I`GYukY*JD_UVk4Ioc_tbzjSn$j?S0VKFF8sr;cvfhf_cS-1Pp0
zl=OQtB$lJtZC=bl9kzd#M8`0jPFn01*$?1>{^*j`IRRorATm<eC^*uSAXNh#s25Qf
z@B?pSDw~MD@R1{5+ct%-S$PjSv;LPD|G-ady;5+_*{W)N0y1ZE80V6lIC!C6y}Qgm
z9LJ|}*7Vcx%jBiEZM;`MA4%|uLfx(~eh_Zpjn-1mQsm%_zl2;nLBvEx20GK27yG21
z3)KL2R@T#y$%AlrX6qmdR2;hkg?({I?cuMb0Ip|39>BMnTqh?~PSRMmo+#NX1p$mC
zInLWA_Q4;$TcEK2z1(`;Qj$pS+7$Ag^hXiDPC|kO=zJ7Lb(=MV1G26hBFi-t@Qi=9
zPXerUkScMqB+|NtNdIa8*yv)l;sAH9N^XGTl|DALEhw9k6A(MDfgt6`Z_r2F5?NZq
z{(0BUc}g5h?S2)XDImc4$gf}T15`GQg8&h6`pL%q?YQRz(t=U8F7IGln@6;I6x5!|
zg2hOSlb?ymB5mPRiNM5;ZvbgS0377Yl)_avP5~gosl)l@FXfwvQz{ip1b#Qfq<1Ch
z*Iqzm2gZ?WcKj^NJ{zr}ckb5BsXN$`9Sz~BUg8$r_{dZG&!anDN~J2Js{M+Z!~FQ8
zz1Lj47OY}wRk&-x(>>I}ON$!sgNb+%>W+T9WB9RvW>Nof05MmZAhxAvDTxxWv|!oN
zrKJUj<Hz#i?hc|R;oHsvkf!(~$9a;v#+pyf?tBLoDXk~Tx|Ju9OA!l{i*9fsjLhc)
z+#QGW*YlDX(9dctxR`QA@j+jwbs~6D&n{t>Do$*1P8JSzh%7cKg)f1Dk#J7!#z7!>
zF;3hMV#kp2dRl-33^t!B_T~RzQzz=7Nc7;*Ba|veq#l6_M7&|eaD~%CwP9AoHJ4_%
zMWPNd;`K)ENSgQNXYiSn+x<|%s>d8o=>(&&<vT!|7YiU1vKdH8D48_U?&ebrAs}rw
zD1zFB%4Xf=%!$JQcslhLvb5QAQDQV3W&y%ZFNcS+0&1<JjKRWIvB;ivcrBsT!z^f)
z*T#esz8b)&;n9{D<zw=(Dvf3d`ysblt?5Q_{k=|CJj5+&v2I^Wwjw&-ez0mIXuwge
z5+7zo%ow3dqzd?D|LohaFb)BMhB|g?E%g7Bjhltu#3L#)%y{aMZ-k$YwQ$ftnRXH6
zXc?#_!xZQ(h#e8wjoO9KD^`GPX)Q#NN++n9R&i?+4{yUUq><!+g`NSa<c0mTL%S%2
zl-<ClAa>bB?G6kK<TL9yH;;L9eql!(-)E@s^NB;L<=Kv;3rF^TSQE0#W&%JbbsY*>
zrRh!EUcPvt$-e;1sLS?g)(iNd@wb3H>d6fuEnss6`)yE7_>ml>KobI1S_*Nb^7V|z
z`ECFQ@kY@xU{FHvZ4M&RO}B0felK$VeY`M8;b5x9W=|9u1oqdIVG*702x8M?0std%
z>_y`q@?nZFD&Dwp;|m`jwtF8#uzRC$q53dn%|GUWwPx-%a`>m{{Fp43j*ebx*Kp<U
zTS!@*gNzW#`ypyO&iWWz^~=ad_~83PU`~%)FqR_-6&eE<&Yk;;tAJhIEA+MGE{RUR
z&Ho$+lNO7gUtbw^D}N;jAkN%kvr#G*)W8cGV6;m(ATW?5$a)Y({hx(oanyks;<2G?
z2j~~dr?n9>dp$J{AV<adp-L)KWC1leOtC#>2u?LzF`A0Ua1E;XKmm?0O^%J#Ad5;u
zY$K2`?!%-8jz<*!mMQgPlauw>c;+vLkf&fKDQ32HyIoWLJ)-=y_pBRf6#6z1YdDg~
zL5p9or2PhQwM3O{m~ylC23g7Ax9eCeA~KP@cfU&zMATz&3k)D4vp@LR;q$0R6bX3K
z8y(lnvN(0t+h8;09gQeKF(X<if=tI?rUG1WCw3_}Xf+zPq-aPAV<kMULJhhEEYKcU
z$xKN7Ngu^`_5e_ED*z5x5~UvX2#uh#sJV+P{X7V+5gHoWX>ET3c6+b!bQ^_f3AQ*V
zj9mJl&sSRKgD>NkEe3iPAL2S|Bv}GLz*6dx6?AnYiq#RjkY11}sZbG|h-dTVP)aG_
z<|K+p&9r(3vm6Q|us@E&;=3Ve|BwyZl%VB;C?t<>fIQ{`f>BxJOcz$OcC?Q^sv)RC
zI@ivxU3m|MY$!5>Xlv7UtsOXQ;O-lK3x_;<L2L*aEAO2EjQ<SX8%#2p0&%GB<y+!I
zBkE9`v(N!Jh|S)j_h(s(Ki>k_^-7}lpD#%{%01b!Ujj>%_#r1ckgbl%_M-g!ebiB6
zWDVg#l8?KMc)mpu6M-hA;7*fEV7pw+woPvQtcf!@5-Vi;Q=;C72bX%D6F@^H(aVJd
zZ1fD5a>^m9K|eM15{B5kiO^Syn(CI;B~vVaVAGCJ07$)njZ>_c{#>N=Tl6V<QPtLA
zRSX3F?!Wlj1AM|RHApFJPBT3P@b<F3eO99*T<&HdIFKmf1f+zlq@O>38oE1t#S<|T
zc_e~~-K)p}$|ur*68IcSD0u?-6o+SPR1VfON`DQvx+@8qPY^nvN?!W8^x5JcK;u2i
zT$ER`&>&ALwnu2P1<#!zPod%AR86Trm&(=;`G64Auzg5v6N~NfG>G^&3U$<s`ksQc
z)AM~OK68AqSs9c2@wTPNI>#WCOP)*A0x?l1*uwZ6?&%Chm0#fzUll2Mfj$DXBP1?d
zxKJ{#GI4WyM2l3GII|Vp({J^7kUjzOWB;xM)K55^>eX{mBB-KCSUfC*tV03Nb&}KK
z!j6hzvbmWBZi=IJ{4%a`d+#5R_7hSVKHwaX3{}KATaV^;IN6gpDitbV+#L)53w><K
zDDXOS7G?}cC+bGsKaVI4DU$)Z)g%<^*T=@7ji3ib03kH!Qewl;oId>%GuF=#%ZckK
z9@t*`L>fT5%ZelzNKE=L^;k>Jv=Nb;c#x!Dgy?e)w7{}DCKmf&cQEY|NsdR&U;uSJ
zXxh|5m%w{qx{S16PzjE;E-utZsfXO{!cHF<9j%70ZmFsUWag_dDjEhsdurCR3m)(|
zN`h>+%RDF&etpR#Am%vOB^x$g#@m+FRp*nW0-%W=NUCU7WFVM^161e`I`H8+L#h?Y
z>)5PgRnkdh=q&zLr?Q+>qA-n*h5Gtc57KIb_Ocq#yC`U4BFmIRkpNTOvlzv#A)=5A
zb2bkiJm~Drqn9PO5uv19^LVOBDLu20=0xi(RI}3k7H<I6sl(luET9|J$o-vj-Tu@C
z0)-;(s*$J(^0|CbqIiw~JJd=cbtS@6Gyno5-7oz8{gHYAxWck|3lspjc@NOI3eUDk
z?Ao6w4EDFDXR{YoC7UR#xvZt5<H>A+(t}~fPVziE9)?bagMr2^?1(jTK}w3B@zS|-
zM}%Q2(%5guro7qNusK5>=qZ}HwM6Tj6fNjrq8OEdekwB4h_SJ;9Tmh!43gPba4(Sb
zJ|zGI0Co}>@V3>+@Nt1hpZX#PLw`yC7LmtS&bCD77l|7e@<p_CdXN!%K>`&kYwv*r
z;fLlg&L5iw6U}Jibl)TNGrnl{<G~V#Z34Q0Ul9$(3v%O$3+q4=Um$!`A=v6$psA_p
z<Is#Q<8G=T<XDw4<L_KNg-(yr)uM6eZs=jw+JhC&LVPAID$s|JL(bj_7!N#~eFqL4
z<m8mXRwGSTY>k+p#c%26+KW#|UbC%bTNAK`evzM3lZ49Mrye%^J}qz?j~oPAPHGJ+
zrs+1UY|=LG52USnB#I{uAt*HijUuffUbFn>W)bCMS<8xikyq&=Z`w~V1azTYM!5Gf
zSVBeL8(XKcA$MV%3zKt-(*YYO1}j#q=s~mWeM5yHi`k#FoIJiiKWtxtb9O>!pVu3(
z6TVUhQq+;9|8o+H_Sz3WNCN?V;-gGdV4n#7qk?-WHg^cy<iCDG?O!|m!v&!I_WM^D
zY4Jg}+H?5BhY$9pE5%hg{-E^#^|#1BjYtrn<n`Zl=RJ^=;xCOO-1*yc)B-vm2WHfv
z<4hS@|9`r$8$ggwk%L0sft-AhohxKqQgo?&;-7f?-##d07s+u+UNOW=Y4|_A9fh7m
zcF~!id+oxPd?@&n`2DxHz(1O`;n6~h-=o-CS65e5``+s|07`zxZ4!XG-x!jkbS(Lw
z975V(e?TNpc}?3++lAjbsf{K=22i^BAzU-o;pXkzqY3fjxJ{K?<B}o+THXIYKb7(b
zG6H0;G~#4_zv<6W_`ke{GX1|ikw1UF?mxf&U+)y<KlkK+yCwg*C;$5``M+{cUZrJk
zDl!1IW%asszmr_?Sa2{;pg!PXG(l^0%ORy_?gS(IZh}LHU`73bXI_nhG4c~>Pckhp
z7LP@xeid$?``f`LxeO%~oD>A#y8)zw)VJBy@3&7p!B;Nm92bo*+A1q3EKCl^-!)JX
z>ACnm&CJM1m;_WK{_67o!`_#;bGdeHE0t!6gh<GcDP_o%B100H$`DGKhfEO}i!!DN
zWtPlSW(^`q8A2puk|J};lp(z5wfFNJ-+O%j!uK7=`|M}G&))I--S>5^b*}R~*IJ;@
z=JN5c^Xx<=+4vS>PtzAP-@FO{{_Bxi|Kd_2sF8~XMw7IS7aIRlM*Z&x>cjqjHQ)W`
zAO8QF_`jdhzkmP#_A?>ppMtp;&MdgqFsKnC{$95M8Yy$Xzcb*k;Su3gd4ZipSWRc^
zA3~92p>pO-2heF@G3+je;H+F~Hy-zrg8o3{XXS8!U8oDjLm~GPJH07yDFHwxKYm<P
zd<=~Yycpcba(M0605aYG-o1NEeaAo#5HAvou$&i>67ewz3@y>YR%p!9&eu9ep<eu3
zhQi)vywYL3^AM?3w9cb5VG!%S(nEY+`!-jtp)(z)t&*Xsm6!|sp>cYBryghwum$YO
zOxPGmP(ArFgxe!(KkjgJcdUqIGIaT#1AG*<2`K+SiXK3<j&d3V?>#&MfVGME7!4HY
zU>@%mg91D~0ZS-chXywqUt~ZDS4UX{Z03wrIA94hzyWE%nlm2H#QX|H14cQ}D{BHe
zc3qR$G1q4=g3^<a-B?gczI!;QN_}<p37C}b<9Ri9fcHrNmrsfV;M;7B9dUGW(gFq?
z7#WG4SP;BWk%QgNCYFi7RA9iYgCo|wN``gEY|)TV<VP_R$>sgV1ybWN9X2@`HHTtF
z^Yr?p4^N%PqVOqq?cU7|f((PzvH^wGmkI^`S+o8GY<A?I0TAKyWP$d~(o78^{)5Rj
z^*05pE;gt-G(dVr;G1W9g39{5KiK0wXc$PxhLv81c+B^+v$F}X9UL^BuJpxYI*T>H
zaB@CEfsCym|8Pu3!Oc}+texDijXJp{#pVL-MaF5Tch_hy?BnEAfVpXGc@j2+955+e
z2%77wASV}nu05cPIxph>{rhze)T?TT(m=tSTW;A3&JUbt3j?<}od6^XO#_3tI#<AY
zZA{<@p8gMeY7vs|E*=X1+%UTBurRiV4ysK^NXTgiP^Y9GL=}&nSj%mwHDh!3u(0G}
z7xn$$FwfUFH#dKiMhtI7b4_HX02h}cc>H!4qoc>9jXy*ANkxfh1?{IB<SO7~(adkS
zI<fMu4a>2}dw@0bA<N<Q>e>aXin8`gbG8cd^2A_Z@frMSgG`M+ww5izE>L3Lymvhn
z)fn(BIH#x<QCO&EFuI8NgPTA3Z$+v72|RPjzZ!-b9<6Bi6Hs<Ul_Sm#5STD+efstp
z;{hY!{zw9qZeq2Hsf!|TQ)K_3$Z-ZMfr@Ko<qry79Xz9Pm^a-qTlu!k-(S`I6X46V
zVEt+g9h78t{05Z_0u53gCie5{<k>hRAnA<a(<xqt@8LRzO*_VMY`zP+(F%5J;_-&>
z<^iV>rbmjx&q9pI$B+B$^_<bb6Q!p&n|c|#AAFY<Br#zrs!7W6Aw0wud{wg6@^MFh
zFqVF!LL@}PS^Poxb0_X``Xt+6PmbP|djfC+qLegu`CL6Y!un%$Gz;xb`jOSBQJ}K3
zJTBl)^w6vcc8WFBaQgms8x9Ad!&}HdhEC3ZhC6EI8f8{lSs9j@KEZfIHE?vzb3+et
z`@5kPV5h`mB20k}Y_K*8$I(y4Z_7pzP290CuUwvG^4K0){z`_%=Vieed0Bh>5>mZ_
zX;S<SVqOPV+ezfelDw2Kog);D4a!`VSTJurNw<yrOv3vJEkJ<rc+D_JYzC;tjJd7u
zbK{5NWIVgSjF`k`XOAs+yoFhIB7XanGUCW1kmGt@=5RbP;S<4SF@G(n;0b7d-N<*S
z$ZQ-#@dkUoDJiUJftFtMBpS9lCLBJ2CMWP0_nnB<h1F|krxq@sF-qqR=H}*@7r6JY
z71pX6gRXq052s?+*YDpiyRg8WnLR{=l&IfWc_d;`?s`ba?MDeHqUhx#WM<n${3Uus
z!rNGC%pff-T{!z39>+(<_{MW;$Aa71+Lqj#@wyp6yIh)#m@;&AonKB|Mfc;J8g~zI
zxvjWJ?3?S$_s7j)g2HG`7Pmgav%QVCum4#I#V>!)SdP)l?hhr_IwvrKsA#rqZt@8t
z90PQK!jc~dyW7YMYq?JL33_0r^=*i1!DD2zqT-M%O!0-wlM$@gj>Nz67v`rqHQ_#N
z#i`5ccmo5#czhO{kdTm;8g{wQsb$vkj4uu0;zm|>Y`6M5|69@9JlEh^>o?uJYBp}%
zR*O03X(A+LxxM$iQg5!V&<8GL8yUJ;Iya=Irgj_0r^z1sAq;XoVR}D-<kOf(sJZMP
zyf%}Rdrii>t9>uv49uw=3Pxpl0?)f>_W1$yI2a>l-hTR2+&w)}!MM`Ja?7oSl}&Sf
zq1O0U6utY(F&c~hoy^~amHvZxW7VMbbLt_<FmVk*jC?<FH*oC$xaix+woK6f3TKb}
znmh;#Yqrd4h0{=|G0X?I;bI=LU_n@`#tYbnjb7(gZEnpv{en{QW}vbfOz(KSQC`8B
z;$w4x<Fp#$1!w$@Vo7v^5G*;>7>BnNB<>L`@6`XhrxVn*O>UJ;%k>{~fu)@nB)0Gj
zU>yrgs8mopWHER!`sK$pIfrrBda<xsWeHtL?-PRTa>3ig=J;#~SpXrni-EynQbgT{
zP@2TWip)|d9atZZY-^31`AmNmtwF-qgRHLnaPP!Bo=sf+`_9bzt2ei<G<?EL*l<P~
z0S^55`BM;ImDpDhLA0>(_HYdvD|2y|Mu{hK&RXyw6v>H|FGa*h2D<*GR{b>$^nbQG
z{b!=5<I4!eIpfuB?hl2{Lt%j>q`D;xD2wQ4m>RI_2C~V4wS?hld_5DcrUuk~j3;8t
zUc|lChZ6_!?R;qa3j}5iX3HS~1j$agrwq0aD>y`>=5Td}p<~y!EK*GEUNWasgjl5b
z<UWabWP0oDUjzRWYH0I#`;FGAzZ2`k0a%_~@LJ$J{~baDv1m?o7I3_??gJNz(U%=~
z4NlVUd|g@ZdH-lmd;$@B1cFdg8}xF1@L+@yh1=pR<n_ZfA6r|A3lP4kQ9>g&M+kiD
z=KOOe$)0$IqQr0*?l4On?r<_UG12v~TEpB7cNBiW;ncJ=tS^UTmf-e-QNXCY)20%=
z6vRZ2^VrZL35)`=L<^M<FoE?a1{o@Uf&4T4F`eAWLMS-g8DMn!f<>sHx-6mlf?zCI
zU>c|0vK<`EzC^OeF-F*2Ey{(On))S15Q^z>AbpaxaCHXu%)}Ruk!@FSI4Eqpi<J$Z
zU)2B_GHmb5H}-}7)JD=A$>^UR4NpjgK|w)<Bl@AEuvHPZ5^n3CW#}vTr+Wq<F%mm`
zu?GT9r4vB`0qXcc#1n__X&%EZB&2?@Vm`ShHh0Y{y170dQv_}3g)eG6)v*bRI4E#%
z1Ujs*bQLNfs1<eE!8fQljd4^59ph0;y~QlvYKupI(d`Boe{he<HWDca?3$!gKTh0G
zE=TWKK<SbM=v5sW@I3w{R@@<(IHpc_7~<(=qnU0PO!Xq}1L+cBIp$>mp8)jA4!=#b
z1EZtYD_4Oc2>QTlZ{B-stq=Pp98OfAO|4&2Rd&PC^oq+Gw618+`gcB@M3uuAg(W{~
z$Qn{6$y4W8GHQmt)(W7g+`M@%&<7CM=`JY3T5}9&`~ZQgu(%Qe1w7LQU6ObQ>HtcO
zKTfnfc=)iEDy$Z9_H~w)0`@9$5x4%*QZ`KUVs7>R`wd@SKT8}mbq3V%vbLk;DJu2o
z(WCJnAK89|ab^^@>$Llq@CVc2xv~P_ypF*RgX3`vxdEUPja>WJldU;@9@BRq$GRLV
zb(zqI#p+OHE0y}2{MFl!dC!;T5PuAs@rhM~eu<ltlkw=CQCVapVh?*im#JHjj1qzk
z?LGWw5<EJwA<p#o>ni{WEvq_}OFyM*-eEV(7K-C8sVON8p3hy-+8JV`RC|EDCS>;e
z-_w1Qhl!U{1`=?QC%i=3PmGS@OOC-erUTGzTe<@9A__Gmw(-xAFH{4$(1I<X8Af;G
zEwYfNF!+Vi0E6O8Ps2hbg4V>PGkh!O;bU4^X!afjFmp>d;*)F%X<e}gmUokn&OE36
zIpMUk!VWId7=sJfI^0+75A{a}Cl$F2ulyYD7k)n<{C-uJR)Ih)!4J?c{gx1S^z~%-
zx#{Q0&b({s&0hud2@X6aMiL?wzmH;ltI`Upo6x(lO@YY=al}zgfUe-#2gAfCIA<QG
zA5zgoY62D$$4VbrVrjRlTu!U@ryj!{3c7o@1JuDMz&GaI=g_3JjX`K%vkWsUe7-Gn
z$hUi5j_hrR2p1=J;>4u)<j6=SQhWwN4*fexTstU$?Wi7vU%HTIs^+`qp6z(0#)k-4
zI3QuX*VMM*gU<N@BWaUxEJh)Pqp>Ff89fC)$arxFIXu(DQY#3%+R#es=0jPePQOJ-
znM-6bgw90-53HE6#JZ_5V)m5a!lV3{P{n}#+yqtYb;IvuJ7abYobY9c0g8$5d(kck
z^}IUG+JR-e1V&D~;%kBNUHc6p3~8J@{nolIgSbHs$jBIBeVZgUmLrakegyJ_5b{ZZ
z=2Y$l1cs@S54oQx<0!;KcrulEWdm)E5I1N2o7%F@sz|wvtD_S7S%yk+1j7E{A6g$9
zDVS?vhDhh1NL$%Hpa|zB0Im5)35|Ie?1!;ii+F9}Vxrr-H;+6O7#V`@-1!9iLK~K3
z6*<{pZtVFnRru#gp@rvyXTJ-T3fgM0@a8N$9->k*JUq2o&hyO75~y&G%f6POS$_vD
zKqwZbY`o?I=d?b5*|g7`sCFpIU<k;7MHp6{Qq?nUQ7MQM2jE+VluBeMl(nR|`zX5k
z1x}V>xvK{3<-q(}c%EmA%v9kk>$161zGxSZ5)ZG9lz8C9NO|6PY-`lioHGi+JX8g|
z&|k27jTCn`yyJn4v~b$$TA_B$o=$h_tjD-;5vw*HC}2GvSDfI_phGVhG7W&=Y0}h&
zBnDxwOZr03MWN5$IaiU~2hh2Kc+qY06A}-;1`ivXEYwmB*fIu3Mh+mO#J<0<1LfT)
zSy@Ju7-vJ@D*M&GIUF?vISn+*X``7jIK8V!4pDw3@#XaP;BT(q=N|~3)RL!Bs9LL;
z6R=+0i#H>nC8>aIP7VXZm_+AvFOB!=UOI9~L7mQBhrcsl(T0w1ZR<MA=_OzPhKb&$
zFFzfw(fyg0UXm_*Aw2p1OX*~})U=wM+}bad-qcP`PHlOHdlVHFc?*_*e^=Vi$;rma
zxgP(2esQs1WaO^P?eVSk@Oax}T1LAkgdcI|Q98&Oub)?f@EiV^PK{16UsQT*hYxGo
zFQ5vJux22%ifLHz_{2p2_;?8Z;h|FRmA{NIYHi%OQC(T0j|pBn#}?j|l_fiIaq*3X
zv^#eA;H{iuH>abcy9hUbj+X0#-LRDY`q8+j-2Kn?A3uKN7Zk|iS*kqOPu;9KJbClx
z+Q29w5fQ&TcecUrscrKDy^XH3qq}fgld`k<TKclu^O9H0rgDcTv6R>2?>r5XXl-4c
zEI0{uY!M`T=r{&}G&sb>=%EZElr!1b*l1rW!+0U!txG85^2^HneSImJnVBzITW`U(
z9ELr6IE00@OilIl^)F$RFJ?0GoP9@zf1_rShkR;3+IFB_1n>g3brEL@Z9M0a?I`2r
zQL8ib;2_>aq1TT$4uWL8wTw)Bt^3UI4KzPJo4^Hq1<woJkKHr}Hlc{xv3s{ah#g&h
z{nt%RYJ-#qPzZ<MVwnJEcX#UxRJ<%y)70FF@av1qz`)4(2t^>T`<Y#j@|>12Mxfkt
zi5|vGp(96>AggNs_|AKC$4zVwX2(L|>)3Ta`*_W*4vUr-pWnP$M;tK=hcaUD;R>!y
zqG6@JVdt`aGOHTn&%VmL-YW}Ks6?0q%2Ilm(UODh;QXlFH1HHVySg$5%F(se4+wZt
zO-&KYwBI*28e3Wh!XS~A)IMsC4~Wh0roA1=8t9jdxE?w_KK={9;yd8sTUZ++t$L_g
zWel2hvQMq%c5YXvMvAG%I3&(7w8$kXE~Ob;Sp}hStP1x3)=~s@ek`hglwa+pr#OD>
z(ZhomD!!t)#P5jo;3zFAxs{hU`kX7$ba`7w84I?&a{lhatb^*-)}7G*)g^}?adlRB
zN{!8dt;$e{Rf`oPg<40348P@aZqo{F;xTW<OPAQue4CV-%GuIKu6v-T^CdgGyYNH6
zk#uKzFOumdvhx|O746%5k3aI378brg9czTjN634{<LX3rWQnwZfB@QHzX3(rJ3E^$
z=dGc@x0J_$KuMeVYyQpTS9UoX=j5pXNsU9vYT6+gw9)w6wm{Cnva%!4`*;_cU_!f#
zM9sm=yBX~%yaFotr+394W<@xH7#SEEqNKpj3Ivz3h60>iAky<Upe!8nu#_k2kJf@Y
zAY)#BRcKh6`rW!^48tHfmK?l3(b0^5mzNKrf>PEOBxh~cE>68`4i5dx8iU9HJ7sMx
zEjOW{d%J=Ir_3EUK5kPthmZdii8x0L^W<AF9@J>iDJ(JZE}$qW(4TrOOFs6X_|VdF
zqo_!j{B)Dr9SAPxGaa?v@`9gwugc8*><tED=tp%5SNO$o)Cf#qO@4zQNJ4c}Zj-en
zdFea8LFY^NBM{YtajxIW!0uRV*S8mqA2(pC)tviO2N0)M=KZJlniWkt5H;GmHV`_*
zv8Q#1+z#xz2SoUWRPVaHSYp70*B~w_nG762E-S~*O}i~EE!$2Gwzs#3w|rrh@puI)
zA|GUDx%X;<#ub?M4!Ml$lC~#%juH4nYEYj#z)trWD9G`p%6CLGq<*pu5kq40_ARA`
z0Rdk|mzn+5tJ>BQU0q$BCDyxfIvfV;cVW47LuyM$hd&|`c^fdjldY`fHt%X``a#0v
zECJf>XRSuND)AM<#&q{kp%I2nfUtTf<0*+|v~?^xSFQ-&^!25or@xK(ny)x_1hX(;
z!N=U5YHCDOo)Q*G{7{A7GVfKd(c`>&5L^bo3<d>YZ#4ka+fFkJivU!z$w;O*Z5)0t
z|LJQG-R|>xXtu0VTzx%$ro53+f=O>*A1mJF+4^?8<O2Mtx^WYDOy0mw1OoDW`wWt>
z>&o9}CcPg&-hVJHaOB8t1p3Gv7~d{59h&`GMTzWj=*yww{q8r@juE|wLH?WxB~Kea
za32-L=NH$g@vGjT^a;bIRDP14PQ}W~iv2HZzI-tkr3?A57vRh0yOb1QX4n^2Q%j=R
z;jh^r-!(fk^R>UfsxvVmAr$zXlH&WZ`Ndhj+4(g|02wtdQx>PZ$--_xQ{3(s1JY}5
z>0=}6E-<k2DD>8Mc})M<hT_T}X!1M$J7jtb1(La6dOA0DW^VHK_BOh%-q!b;=gHz8
z9*?xYct%fMf3xqEcHP%eLAm+yy)w&>8}Ps>aj@O2`+=|ux`huP4j-P6i(@ALHhO3&
z@o&F6NxZO_-A@5pnD4Zz>UtI(x7I#bo!WIJ(4#aUT6<bai3$(G@BV#SQB_1zXVwnS
z?;|61&kfI<p(YV;d)hZA!^OqL*RNhtU_!X54R)(D;(ph?@I1ht%$rg|^J>O@ued!0
z3bp&OXNmGQ+*QO4AWw-MP^k^%!JzRc1h4!X=)@3dQ0a=RE6az_;bhd@*P53YIS7}v
zE=owiH5yjdgJpfMU#~^%uL1W&7_HQvfh@@m1tjCIpt8wMZrsL(Md{MAG9?%%G3R7I
z>Jh_7s+PwjfAvXtt*#tJ>(L$5Y~S#FL!|M--=XPV2s`XkCHIg-gyZlQsBs`qg97(N
z9{vp>xEmtZ38-^ojrZdXks7H7<(OKixflB*jPXr|dtV%xnVtQlA1c|y3=3XP2OGuH
z7cW#o1*D~!(f(8eria1S5cmZVc^h;Ddgk}+5)xz;#asB!9I#%ZeIFG9z7K*ajFas1
z&)ew0EzKk6=>7riybBwN`<D$%@g{FJVC@zMs5pimW)_x1z<p%t)6Izjwo*#%W{j5`
z^x#DMg63$vowao^e!B36QUSd}zoPp#mVXR`mboLoWBvr@+q2+cz3Zl`!puH|YuwD^
z$olJ+f`WLqES{&G5ZA*p+<Z*6@bmNAclhv5pw*D&bLY;%)4GLlG~shmya#xAk`^@}
zrsPE(#u)qypjFqy_NpDc%-)t+sq%dF4U8fA=o<lYU5|>~aS1mAUPQk6g@u09Cj2Eg
z3Six<mCD6nK=5!1ET`AgeF?Vl&D^?mYd(Y)tO*R)Isc}<eltqySD1kyuS$!#1^M^+
z_jyt6rn&v!ER?LquO}bRGx{dn3|?xhgD2VoQ#We9?E-DK=D6YM)2B~84;p~YE<|#6
zyi3bC0nqnU>)ky^LU+j3;>-j_MbQuawz9Mg!{M9CdSJtdK%D}=VBFlA9hhNT=<r?f
z)*R@PhT3Z&n2l!H{dOSVG_6{r9^X86-~|Q|Jh-+X^6lHVlh86;A7nmxas~}4^5~Pz
z5EFICL+9BkP+WY&t1x%d)ipF|S)A|f2onLo;llA7^4yBe3M|ShD&LEv!@^`?!aJx3
zT8=G4O|G`FkrQK*a-4fBy^q;R6*ki;1mol4!EVnM$aAXslW^2~Mf!Wf;0sqbw=hiH
z)_JsCTYEC=4!V8hHnJhtaRC2o>ZUz;a<K0G`vGSIRCb)8+%``tj40^nJiv<i--0LZ
zV+M-WhlP<ZFQu~MA3kKTN&rDWo!1NsIRJ*>B*~G{(QvfKhofG&HHTRhCrfCVnNM$I
zGXY|fS5i_keue_?o2LQd9XGPo&CqQ0U<_bFRC&V(I+PZV+Mgk!TM0<<@dZD9`m{Fk
z1>SFy^#?41w%3D^*0dxQ4~<mU_E&2z!|ff{p@k2xQtSn&{O0}p5RG(@fx8-hV93Sm
z{0@}hR^5n+Lak{jDZh9*4jl>$vxCCoXbCgqy51g&S`0|Z5DQ9(#%1QhG{VicAnC#%
zh}~aV4KC(ng_}yC&fM6Tt!xz#0!Wi0@b29qLn^VhN0nVoCoC*@a9%YmDiPB5A3369
z{1Tm-$`>#0d*XDy81!v8-W+eSVf*Y6Ha{HhTaT)6d3ZC7JPtfrl){{2V5!X&tO$83
z$gOUX#uqQXvZ}7Cx`FeC>#y7QsZa$KdM$3ETcqBXnKJg2Nz)M&<4PQZ9T^(Fi$tb-
z(&hn1KkbK<>#u#H7yP%gWGu`VZ`s6_0(6kGkggibH9!BU1|e*hQ5nz<|A>`U2*_`v
zKT`lDD$lb_@Hj#w3r1uuO-hJTqYSiru+hTIOc8x4Y58h4rigV>55%%G9&ZQ?3DMS7
zwCrfE&&M3i<R?#>U%W?I5sp`;F=6@5fL3$(3^twboxHw>i76Zd>3NFz#+1$|E3<(Y
z;ev~LbAtuwGwk@LoLsF)O7eLhd*m(y6VsIBZQ%Sc(Db)lXm{>p=nk87KfwsI;2V7%
zR5P+J^2a`V2M_yyr%qhbyslSpX?-?J-ej7`^12ucr>moFfmTDnyZgb1b-Xme8cKkf
zx7qKp2__sQI4uVUhwM9IV4a#8R53EO2nR^f8Y@TBWl7ZX%!t;rRkA<^Sm?ntQyeJ6
zuZ8t;QpcGTR#I{49#|}!ln@py*mM$vkz+R>7uW5WisB`~nibmDK%!eS^-Fjs4+D}B
z43~*zODpa2Qfe^()@rPGXOgqCRWQu7YD9`#zG#^n*Ht$6$-UA5n^>ckOSZPEv)qo$
zi5@-gN89^2KY4#c6_tV+05?M$K7WqGikt8U%dJglhs}JahxSRMzF+gNKhdvyb$&If
zWb!M{NZQ_M=%c5eU$0SY3rtoUko)Ja`ttF$2J-PImD&PF{l{LmLMteFaBx)Kc*_OP
zvwtPFDu}#BJUe9fMY?X|94H}zH|$+p%tB7?_qzz5Rovm5-`B%9{EYz@&tZ)@I-fS9
z95I*Q+f-jKr>?#Oo0N1zIC$MbYph2otp++}=xJ<hB<r?u?vvr639;Pn<mgDwN&Cre
zkv4}s-0nbRrbDHmhbVD4Z@qXSg(`GFv2+DA({NwqvB~SZALWgab_g^GnadkfY<@fj
zxtg?R=NA@U8kf30!vl527ZqeyFL2r;K)Y~LbZM23KHiEP{<!^?SoLjn1oDAW8aW|A
zgUHSYrqDNj&CQvyq~-*`<4ph*E(z~BmbuBkyP#LW4Ja6QLum%}5Tw%!9^fV}i!i%C
z5BtG`)ZpIy{QarGmbeb_S#~DtpyudD`NMk<t3tP+ZhHgzC=87*Ja$G#d(rc@8KWko
zVRqJ?#a2Lmv!=MZu?1=P6Kr})?+B}^scqk|mB9x&Mk`%2m4=qq7mTM<)7rB;?G^er
zZQUC2x2d^V>GbKX=yGs?qWY$$ra&=s*K2fk53=WZpgbu5dErqp*~)x<7AP%F>Y5tn
zClP-*BtY%D>#)!M!pMaux?-R|-{?NL!=)VsTtw})vS6+pA%G8EY}2(9ES_VL1o-|<
zfNyj&M!aKynu*eGJAnp$P2Q$rveF|=VX=TtX+T$nN*-t8#G^(ayo+GHL-e}A!EQw_
ztE;PP5?(bZi--mS_c9H;c&q_7L+Ckkt1Me;gH_zx`nc8Njw9Wnh+z|-#2fepp{TC$
zFHZ#Mg;#*x&S{(Uj4C`ff$@IEIB2raK6A~|N8pqdjX{wgwgARHw?l(oPIDiEOg}21
z95<*@{DC*qHg<<l)PJUCXV2mSY=_Ue8eoiq0zur~(eWnOxX(6>5KM8Lf<K3-2G!Qr
z--JI{eh)aZs=B&Oh)xl~SqmUzNC1(nR&yhZ&mM|c?sRl?+~0$?UUh8uAtxWtR1PE>
zgVzTt>b?8$U9T~YN_L=S!Q}nUbJ}(=VdI%<3Qn1vJ4cUbi2oD&l_y0Z?YT$?IVJ4<
z8*ESsf-FRi9!Q^%V#wXdT$BfMqhe!&k^cxd4C&Ivkk<a;;f>%{k@<wW`(Rvys=-RP
zXaK|R`wyWQ6S3Q|_GZ#T_gM+PJBWrq&d*jKxH<7$FaP3=uX7x%r%uEo)=@%18<?F)
z8kY2)8rWq19MvvR=4A-FA<-@KkW3|>K6@4^nMl{@ot>N87_cukIyw-D@H>vR>#whC
z#<pG|a6$p-dT<5J74wv;SP&RxEB404$13dQ6%)@Vas~PMK8cB}m(N?i&<XPG^u8Ip
z^k?!hFsx(qF5vZIi<Y(Ux&sya;k%k#0QW#_NMB4NoWV!9d<N%@dYQV;hFPJryE_>j
z<`y7bPn-bXXpAu-P_!7*4OQn7lSb^2?CQqTYHq12y#(aO3Dgesv^*8Oa!zSA5Dd>_
z1d^v16PNVd18f}ROeJLB$GJxI&rrHMllg3?%wG*o^{)NAa^CVEdSZujvJk@KoebJt
z9Ha7A<m2aAlhx?L5I(Ksj$VEx@~%us_$<zA0N_(gYb!g-94=6MA;l=G_hGEf#Rv_Y
zi}1w6<`-K>0$MRk!jYyOhMhq(xep&dasbzBFO4!Es=PL7^lflZ0o5)8O%X_H-Y&rx
zm{jshF!2tAD4|ePcK)Bs-R?Wf&mphWK~YvlB;7>OG&~z4-T~oS9=id7fWr`53qzaK
zR8%;y>fwRGOf{mOL-)Bq&hi`CfEdH~Y&(h;xhd4D&x+9x<4E%aqLoUCfV(%ng1!zF
zz=OG`8c{&+vE<dx<Jj-VsWk3=NAFN19`^fI=;w3+P-g=uklP%eD!xZ~Y@lA;pF9YX
zrfmvTS+lj-W(6`Y>(p64k360P;49PKpP+xu)5U2Vwrt*f5}!ibu@{7VFnG;SeR>^v
z15m?NgP<@2r#;Z@qt5v*bvQa(8H<mptd1V@m_Lr%O9f@8-sA%$ej1djjvyJU2GIjk
z+j!F0)RZ$iSK*Ve7L6QfhSZ(F2~<AdKAKY8e429)4O#G6QyQI^c<@?0E@qHut|Z48
z##%)ZYzs%q;nwb+9<A9S$b^hyDUaF7m)~jJ@aa<|0byZcqN0)fM&-JKSxdO9;&&Ac
zHC`mhc=Af*A?_)+f>_hK+t<(#+;Lh(q%E~Fwb>VZEZFePYmS52sRJlrv@9wttVNz;
zq-n#g-PKe~P%gzi^vs9hP!=ibH`LX!fh6Wcy=umsp3KMySimJO^G&it>ZN$Zk!4Rf
zSH1`z<<W<a^_3pItr;(_%_Vbl?x*FralZQJ+&=XM@zh0UR(LI+;?>SLzI<?3BbN~9
zCo3b=6or_dv~%xXHRP}}TxxN(ZaHYV8Zev$f;uOKam(dqH6Pb~d#`=EA~iZQi2Y}D
zObkIi_t8Bn+=?qji=d^LfKH5!ypM-pA}pSPBSa<q{(Z&~j4x1zcZ`t?{c2dS;h$L4
z+ZY`ky;T52@@Z7Wjn_|uY}zj^9XG$UM9ad$au|$TmZ55#wcJJrd;5K4Ejyyy0072*
z4vrIosvaKF(b8^*FVAVAMrw+VG2v*YH&jB6Xo4y^gi#LlZXHU7lfa>jZgiGXavK?s
zc}a8J&CM0)f}iQLLrQx3eb9hAL7FM+Q&Lho!V#He+qG20b2BZguCX!IDz|cV<@`N?
z7V1BJ?2G;L)uO_?CY~F+d+jne<5ew{cxslW`Qv&2R3N1GsTe(-7Vs}}dyjd&pY*e$
z(BFR<4+LNR1ZH2V!C9kg-U#F?wrTTb_A?36GzLG>BQ-D%*+k{9XjVdW7Jo5i%7I4V
z?cgeD@!1tUBc5$A(Su*7`Uj|n3`><4lbYkl*tUWQdiy(sMJlS?Ygu2=5UC&Vc|Qp!
zDAe9mSNmEO<&4>>7JuJ(AkOl*<lV3^nrL(}$C)*tmV>U~Sj1=VY5u`vbrUGlc5_n<
z``RaEuuM;2nX4e3fSH-u070as5%CVk?=PPJxXcatCOP~&8HtxPtsnE$WrFFDsA%fm
z9uU%J^&w$D$S@rl9{v>*og7d&!(tE@eX1=`VT6>#@?!t|^*Di#{BEbNoAGLJZ$3Xi
z%I6>yxq;{B#jIpglv#0Z6{{=miq}g+&j;0Ga66NE0y(W=2@{kTXh!MWTIrl0t~-v-
zB*ex?BG&DvHPuk=xvI9r;qhXD%WH{FW!j<Zc<(tZtX)v1FPG0xOiW08%wCd2v5ZMt
zZz;V&ZN-gDPfypu`^IScLDIrfbD%DvEV%~57w@qs#7lr!3@j}z{VW5h^{nY#y~~e~
zprA`vdCkthfB!xpHg=C?dk&+y($S=RSryty1NJlLwu;zv-b2>Ykn%={--hc*)CDCa
zgYVA+Q?~-Lwt{SEqW<jjt?5IIUp6h=vFn__p4Qc6s8(~3bN~$3>~||JE#(3k#1$GE
z>PTs2vRiOX{inba*CC6uLoSwha(4tp!es#r2UyX3SO)I$O>b|!xRg}Umqdl&<hLbU
znHXPe&4TIs5>)zif{67TD6t36{vc!wQ<lEI{(g3Lxo|}m-%R*Y^d?Zx9xy!<A2C;@
zT{wFlJvpbwVVR6F{sX|7Gt$<@SR4Qde5iG^gH4g@Ax!9$3XhLhZFE9@f9u}Z))oQg
zQsZNGLkoW+j^%z!sIFCbD;X-@Me+Wq^PGN!y)1ny`FW@2yzw2E_<H>5Mth#I;o*Cz
z4y_go^C~yBe7=}9l~O4zEF4%;B03~>{LGoVn7{B4a3@?l&mdYjmrx8?V^<=raQ(Rj
z=&S9R@E7@d3eAm;`f6_NB!pFIWByR;ia!qh01!;v0!rXhFfh3qauh>S^kbHGRabur
zg+i_58gl=ozZ_X7m`~RXoM!^d<FYcuv?Dww<`kLrscdIr@@7H9H+QR{p`jye$5sh|
zz|UkW!1^Sm9VCGY1ekFID#+J-7p`6v*|cq2>d%fh&$v>AgcN{5o=5`Yh7XU7D39M|
zICL}SoQu#d^<no%>ofNG-9z51eVN!CCvig68<Mx8K2GDarY(h}E^G<RK+<h1j$t}Y
z?AW7Dqo6%*+5~!u&ne7QJ5;ASYiVi*VClqX7XYwla}kBUnIgSEzA~NWe2I>>NTA4|
z-v+6Xf4<dhX)lDjSRJi{>@H&Q!3WTc&nA)l-N3*A&$aNgClN18LNP(h@y7Y5nC#S9
z5K>iQKt_N5tSHRS--q}dmZC9xF{|*y_Iqg}*TN$|4RLXC<;Vv6IpW>lHWWPO`7ebx
z4^!~#`E%g?j!oD9ffCxZJ;v0~+d7$r*iD9i8KK$umaiwwyo{3@@RCQpQ|eY|3)j31
ziURoGR9O8<*#$j?COE1&OK2V3XRgwvRv6RP7c+j0hJ~f^=}K1=Dcg~8!yzc}jqld%
zUdwb6_<+R`2FQQ@gY9cO>FE{G7bXkFMZCEKyiD@c^S>C@(>4XUd|el>N?%n?l{Ui8
zDeh5u4emlV^4GP=YU||Vbz385F8&jV-c)JNpZOdqlErYEOhmZlmLTo+x^^13Z8Ps<
z)Q9DM_OP8XEI{_Ij>(e_l$Lv#yiEEDqe*6Zgc?qG94cRQ%sCKhHr2$THRQY%!bV8I
z2ug>`s_DvM=5}_?i*ZaCNwH5VNYZJi+}$QdX6E<#sc>AQOw*cXWMH_5gnn3tCk-hH
z)Pj@D6U$49cWV)?IG`kiOS2yH+P{iv4H+YrMKt{9OG9Fmz?}hXJf!JoJ9aG5rmrHt
zh{MEkVc(Ms+INKG;n0?pk`k7D!q*eX0seyN04>k>fIS|YVnS!9?zgeA2DhJ<@ow{1
zzma(R-2;8&4~qJ$>3i8B&*|#vX-orb%l%xS-?OK|je66a<VS)jeeN}eIXOAH=B8nT
zm=%7%fp3gQ>)pjHuod~;S_3wGn@%lV%mP~>Jc_fin(-#jCj9{vpQ1FriFjF4s^1S|
zVy2kn+1>BJ;vN@MS}YZFD@kx!z<n?Lj&v?e^$iV|uks@IY3X@xWE+4=y7ed_6Rf)b
zSJZ$#<V@4hDugVCW`JPnLJU>OQd|G@ST~Pj4Dgw-<Sy^!#Yku+H-xEsLNH~}n=d(K
zgBDuFotgyGhQB9@5CDZDRb#mAf#)CS&vlJ&8vXI3tkkI$C@+ZI5`P#}S0?t*3i)Jn
zLm11tSF?{Ia<|IJyn<o>?Thx-*6=LnQ!l*eZ_X4zO3qBFrecec=%^P0)gNdJG!Y&d
zSrZd#E3_-|IquhuFn^UEWva>yAU*S6RF<$pz2WgqJ0mLk&u?fOqU*cND>qYJV7;}4
z`rWpUi&;52{DrVhw|f@;@%#72tBX9WM@)l1+n;8Bzqt6r*)Iq5JI%sgcrJ;h8?lI3
z?6|yd2ivVRnie<GECwx@y6p064AJwOIT_Aw;k#(kQ<$~)^jZ4uHLVdp)CvnbR3UGk
zy`(euq6BD2)b>C3n!}wR0+mi+FCMpOOK7a9HGTC8=FhmYRD%j+Xx8e!wn23^(RFD~
zU{=o!21?mdI_L04NT{*SYk0;x_DQMSJ{)f%np-K_<$ZIa$F}8?si^`?I1JYXa0dSp
zQc>E)#RZh+gATV(B68+Rx#Xdwdz(IU|08JrZUD0)HWlhbRrm(`X5yMgs1e*?0zUoU
zYg-&3Suj9CKQ!Q+7Sr7oi>LOmr|yN>fQ+9revav-NvM0nXOS8<@tULj{m|N~`S1$d
z2O+ay^0x4%=V+>{?^}qEj?RoaEaIHEtAA*S3hYJQB9Nf#(AimM(qU0&hp>e`#L+rO
zAq5D*{ad#PEh8*8mhV9xCa?UOnhNKK{PflI0*~L0H*en-{zQA%?fM2tDtqxK21=VB
z9`c>_-tih$k^|;LjIIDl?twVrhvo{CSZopj3F(L7aM3Y#2ooVoAqRm0-XeW6u0Dp8
z4hfUd`5qS=+ZyOFjy*`ossiS4bj*X%GBz>U-<6h|dj}qFV{p7VZfLkCifDzuW^x<0
ze}iUv(6ch!!hsn7xN{E7N-zfXw0mNkia2;hXsgYgMwmkj%8B@;v|nIIghrMOYSY*7
z@-Pe)p}Rh=gQBO%1(3-VdX@DeczCF`w4|E8?B3!aYT4>1dcoda2vOtRkWibVtn8ZB
z)>dsy#9aDABpMLunU<t9Cnko1T&!Z%Ew%`&AC>i;V^%wLO0!3*vYdU%Mil;#Ktnw#
zxJ`Hq&bJc1<<T<>EdJaCZapF=c=3D*-U1`^y*Nt{Ctrt#xb?h2CEyF6J?Br@Bv1q4
zvr{>hYhD5PKC0TayjriGFmr?g+nS@E+d*j5Ei&5%lLaRsUUo1r)Gm}i#V`*|TV;fq
zxTEe-EiSDc9b}Qh2aIO1k|`~$FkLp>NeBZh-X<QIF)w#__tS|o@nuJW=S8s78z<#Q
z{JW~EHK;63z%JrCbU;o?$$#-nO7A2)drUH%QNulDPhMXo2_R>S+x9VRpduqOKVn)C
zBeeOj9BwYIZ{Xg_8xCZjw0`&bvjLKHMk+H>ip*AwC!CVv;P5S5M)G?MW^?n$T_eTp
z;GS=z%xXx535EuKjQx76u+F(w?#8aH5Tr}xc8CH^F?kTOwoX_e<up{N8&!HWKJNu%
z{NY6?H5@84)d#8p+P=bNT;_?c7FXnhf{6!dVpQs`S3nU{*-os=n8HxK8`W|04`V)S
zXvF*c^3x`vnWBu@j$-od92AT1P*y^$&&41^{+3u*0v}PbraNq5RlwE~AED{6$;c!W
z*uoG$kAVSa^>^!m<Qd)JS@xP{>CA_)<+N<URa?D|zN{<nXyhQ3Y^b}AGJx@Ud@!!?
zD**CBwU2!<qsL@F_W6u1mw`ukh)D&X=Rmz*!mX|;z0f^VmbG}M{QB&vpFOrTFrK`E
zU*PNHWb+4WREoMNo6Z)YhTn_<1aj1wZHMf?9+!5X+Y0`Jw`6H$*?ng8Q}CiHItO;3
zsqLmsP+yh8{Jv57{s76a&<t9IcwO+^C|_vY<l{4pfD%k8z)QlS5v!v=LVZ9^h61H;
zGQw4P>vcD`Ly|A$6&0rh)A_Z);nYonDf<e}i1*+L+M$?(t*k7QfDe3kvz~CsZ4@+k
zx=(KscLde<s|GI(Mg<d7(|#mPM@x8p?E7D@uPekP`!931el1L!dmxoLUmzAbBfzQ>
zhIqYT;)X&F(dO?lU{R30^!UGC0DB}+obZ#qt4ouWTFu6%PHj$3PF~kx4wW$H5(xIq
zXm4oqfaU#C4h9JNE%rBl`I68g6qbr^qx_N*)81>~N~|hc-y+R0#?Pjffe_f;AlBh5
zY||<9mO{f4gC(%zTiWUiq8i7+gJ;wYQ5XrCH!&kqZKJa^dBR8$N4W(P<lYU_?Z1Pi
zUf3ko{QTgtW2{ivQoBE`14Y`VwxO`Iv-4GNFAF;UzFwsVR=)@iF=q)>Ne~>thf86X
zT?3cC@x;SCtEUOorl}`XEoJ8n)3c_zmwtk7)eP(nPs3*}Z$&F!SWEwJ*d*qs27{kx
zowfSlw}Sk9X>0g5Iel+@{vsDHf$g}b_x}nuFYcir!Znk@r+BR!Q>m4hbp~~LUArf!
zUF^14BRUE6b{D8q^AF$Sv)rmNesH_g+jg|J$^iprEXXe#7Gk<AJI)Ns=$xg-_wVhY
z$6EPo=$%I^!VS3JfJdvCGH$*_FyBE(w+@lx?ey(~#Y;MEHtd|L@VvQtN|%+Enq8IO
z?KpAW-F@22yngh(U6nUqE(-z@tj(KCA|3}iHTsi#fQwt$q;qV_*0(mTvAOvj7^1;e
zQ1hV$U(kN`WsVp<&^V!2d3~1GS`uW}ZI{0!COlwU*S(Kr%a$!A5%37eAB$5jE-5L%
z44r4B`n$Q|O_4@-H!~)Kw3R;pFn0S~1<EaAiYYNQH&^mBlOc!0+j4Wo4Hz4Xzy3=d
zx4XoNfd6~^g_?Np@WdH=;92}MR}raTLzS_i<r?4Cg4K0LkljN<9Z-+y@LHs?gPc0&
zEcFv<ZR@dTJC?TzEti3}%^5nawajg?xJUi&YAN0$EMJj5D*%iL9#l1jhA}folAf>7
z8Z5d)g0XjUdi)cOkei(^xc%w_45!5@*2TP|`UFo%;=ZsfBpmZ@dR_^2D@1L|gQn6q
z_?K*3;OiaE>_<SrI$v^taSS*5Aq->7!%<h)%hP?t)Uq#@smJIyMQbGE&wcnbcum5>
z(y|I_@Zq)5?Xr30Ef`y0SNj_!E8O7gK;WN1E_9>z(e&eLa;DMK(*wzH0*WSf2C#Gc
zq&zNj+(c)o?a2Z)JnlEUn3$NE{K$i!!_9N&U9_)Zq+hq<8@`*?DS5?7=Qv?QuEr_D
zl;-+Q8Av!=nM5`sW}WI9wCb!=bh=mb`7<-%X%!GhX6das4g_5YBI<u2%~E2?n0YU^
zd!NAzmpfK4MkyF89aB+JX|LUQ(5U<syb$VGOVRW<bqz302<tJUpWj{Kq67Z%MNiaO
z%buENS5O^Yw6Lfh)rH_+4ch1f<}Q74{yjZ?c_L>m9ejC;iaEY?oe3od9jbsF5>NKz
z4ZQ}*Cqc!PwMT|!NPwSz2g;WKt6$hUj6C4aWNC^tfz2&kmp6o?at0e+8&6gjRpay3
zaqQU8;WU3fxLDmFC0^afk?Gd9u4~bf8;mTN&}bmrEC|Yw83vu>#^z6;gV?*A<E6Hc
z^TwX>>52pssUML4H;w9ol)a0W0ZT9xtP@SoI43?sj+6KDl0jY<w+&j1Z7NHyJdZ)A
z_O}zEzV(wr+V<nuQNPAr+fKLjaVjY)Zoqa*SzCYjJlXj9Lp~}S$pf|!OD+JD)<>9h
zSit@eLlCI}v%Y`-_U%K#WhKlMor+;g-Ppa%XC)REu9$J=iG_OUj+gIee=E}i5fGaz
z#F1BJ{go3CU!Y*|2ih-Tstx%F8#iy3!>u_hMw>>?4U`wxKR;Q3j>6*O1$NK8fmLEE
zWN@r<mm^DE8VGqo;z#Lpfg`8R4rBxt>D+i^y=z2i#Fq1@(2Y4kjR!7t|9VZa)KdzI
zXnmuZ$FHxlApEsP=5*X_oyuK*-(2)4E-E?!wGVt`#+(I>+l|*_X|NKsEsx&5yNYf0
zyMT`skRA8;U<PO)dP}Y3=h5Pngx;zUTX;m@U=|!>55j#u+Ez!GbFGL(L$dddr8D;E
z&f-M(l<YTaahGuhxaLE#t>F=VqGLyFp@+(1>=`LFee{CMn6s^%ABo-uDE%8s^UfFM
zq7#*$T3cVk&;4~`qOsFRUEK%V&NuwBQ$2#Ji{}v)h~M*rcXoREJ0^)$sB&=&^_wUA
zD&IYI_-+pzjd;Cr^J(q|py^=D64da$3tF7IzrX*K<-y-lPdBn6X>Zlwc7y_>!A-v#
z|6xZC2m2T!$9($ih9rfLmrpD{782NeWRIK_M?0+kQ&Lh=dr-mm?|S)^Ul{Ea@D)-+
zY@yh!WQ5<a3n^X!#S(tS8XQyz0nqpYJ8G}#wRz@YwaE?eH~m1y1Sw8xtQKM(QdUK%
ziBcInXvAPeuYFsDQ^HWl_PPQ^`5s*&aJ?d;YqBObZPNgPRU_F(7B2NyQ;ovmJPC_q
zM*RF!n<BQOzD#=z*hz7JtqUNgCPr}vrL<|9@DTod0W3k!0@gw+1jOQqC~Xjp4M-kS
zgj;TIZce)@FLdm7F0KE#D5X8C4)Fi_Pbx8Gc743@I%wEqz`!7%>E%MyInPFnL6g!j
zF$Dqq70LX;!6Z&h#dTdM>1#!2JUu-JWj<un<_!sTOrMwEt7jfmT%9Uj+eD$gmpK=Y
zes|j=E&U^Fc8F{_sh27;X6KW*;qIL~HSl&h4Ajy{U!OUtz6TBlRpVv(WmToNzQ3&-
zKWQM=g@n2m1}wgq`3?36%RJZP!<qGhhVw&-B*fn2P=G!nxHIu?pl_>QiFytx&lp$L
zQ6>OjnOFuP)!ag%^kmBN&y$QU`BO2iKgi*dL$3b~?*&EQZ5M2Jd<$c>5E`WGN|8%(
zlao=r5vK@7gTW%X22|X`=;#AyLnWJ?5ewk(BVJDQ$cFv#zjplS%6S>v9hS=vA3X2}
zMLJb_P(|5F$!CS!!gcb;nD-I?9l8_1Ib-7DO@Jak(;|@)35uU%if3&@VNG#t%E=B0
zued_@@*NC%4gd>)K0>9yS({LE$wCR$n<mXEO*#N>0dVjtK1V@TplEw;_a&+0q5&1>
zjRTgs>55R?f$BRo6#_|85O{wp<Pp%|TOqp1qXO(79ITOGhzhn-_MWzUSGQ>AjhUTa
z=DOf97c6iu-T$`?L;ZbV2sXHI3$XM%^sDma%Y_l5%c>^=q5yKDlCCRjwMquvp7|Q^
z_pb-$b?rkF&7(7%?CfigEzi)BnY-H0z-0PCoE_uOKHh)&GscMdVi%<5Tn)_G9{;MJ
z84McSyih!^qIM5J&INXsDBi_QbHI{LfkS6K=){|?u`w}++f8fi+){W7jltOtNLsxG
z1Vof%z$kZM*o)@o{*1zAwF+iBe$Shn_4S#Eyhj9ceAQ|Q{p(Z{>*o&L-*)zbQr3!3
zLVSF#rn}r%t96-0M)-E-o_eT`BD8WGs+^7Hugdoa39%JB-A&$p%^-{;_<ZROi<@a8
zm#nOKJJvApT#dr|KJ^2!0H$$kvvZ7&;BP|o{%R{*J~ex~;sh=S-eq6s3!o|`q}ZiB
z=I<2<?#Y#d`4oL(mridJc=;ju^V(ZO@{$t(XPK6%{V{44<x&OOcgCfLK;-cjV5Z#V
zCdYdA<dt)$_1gRVc0j+P+P-}|I#ss=&2VHCiM%@F8@w1Bno+rWB2tnJW{9KQt*~tP
z+;w4!6S`3bWhM$(2VnDOwbLT-^yOo&>p!3slf&;VJw!4OKo})U^>P-VH<8$2G)Q^1
zmy6`9wBjYZ%JmZ)>0IWvZ`u^LM3D$p9!(Y(VP8SlA|Jni04Nadz>R$Coa4u!89au&
zoLFRWhq;F-k0$M16-X2d;GMnvpR;fN5`c2pUuy&W1npIq;l(?Sp&y!JDq69WUn{P>
z5mRCdVt2!ZHoCI!EDds6n#;Wfl2n1xI=S{JJi|x|0_qm+h-cI4CAMxbS|%QKf9uAq
z5HLaUi2xYI*cCiZ-e~6}>BTE+TD%*m;MPKgRj=>~6C2(EUFRb}%uM#NqLa@L$_%<n
z_H`;P3XIqtcaTz>>7$1>LSL|WBps7bBVJBrRYo3&J&44AGjV;)MTCf*cSu(NDEQqd
zhh77&VtJxL)AqNJTJEpgKqm*G1~kc;Pf_q(yaD#fd>%^u?V3G}3_v8qzYPdP<(EuW
z`%A6C?ezlxb5Q`e3<{sGchw6=wr$<2trxGUg7z7bLE&<t#E@;I3GN*%ECNTI&0%GR
zK<k4DO4dN4HR#NyPfveUE1}M?2hIJ=lQi)Z7T(OvJODZ=DR;SVz8<!glafnyds)fB
zFHG)09=cX+B^GhVBx$qB9>vyL#ThM5fVYQ7-Nn#h%g_VQH7VCpZ5&m>m@<9cDJ_?N
z>Fmrn&EXRY4rzuu>mK~#xkH-lu_$vFp$a-HFr1Ni2TRk3D+RJURNX4g;H1h3+YinD
z9b^2fmKGXJ;keYBZjpo?AcBi?6AhKgny9No4&7_0s*>FqeQQhJP<~-cedSqby`-F}
zTg2N6vK59l-V%5MJ>HMQ>^|Mfs9eyXpFKhW8IvBj`hL$99Af;E$M9#Png61+N*|b|
zVfi;o|GgTIMMim&{#s{-|3InyYj!qu^cntr4g@GL=Ra(xQ98K7wmw#Yb4#p_F@NCj
zP1*iN((A7{@yTUx|F;V3TU+CXMJL@f6tD^OkBtQb^Yo+efP+2RK%^Jhh548lGwy*B
z0^DyDB@q@TpHaS6ef+4EmjE>_1GYS5l?@P`kxPZXl%S#spU&C-&owFb5`ocuY#*Cn
zxL4H2dYyEVD9CLI1&zusEo~skJB|sN*pqHiHFxU&+!)Niy@7gp_Ae1^VxT|nMjQNN
zMX_@Ol~8{MS?<Y3fmyqZgb_XqXRS?VzO1oK0ZS4S3UDkX^aG>4hNq1hw@$E~-<AR`
zDmA*ep9iB}w%#1B9P~$*32AzPFmMT^wm8~?h_j>>qbb?YFN5>{4pRr#EpTY7T7K#)
z9N9Hum-Nevwsk~<KCg;x^!6BIp4zrA5HtZnc2`wZg-jf|^7$3S^>sabXaZ2LjN;{M
zRz|jdA16VJ-3qYTr6)W=8`2}VxXLGfb{0?qV=KYQHs%Ef2S`Ef5AXSG0{%m~7=v_y
zS!Uqh%t3rJf5lL={h*=3joKk|SWoO|&qiG~Y8}0{UB>EXvGu7RFXj#C(?Bp{-$9c-
zMFotB;5Z6bPMUe1I&((p_;CtiCSzqy+?hFsu>=g1nW$HP1D999AXDUY8TDL&U7w7s
zY38CNCgE(p?1@lclW?;*1-3JJFfO`Eo3q`o*Mqw>M$4pQ3kAM&0Djk&Qf_ynlzu0E
z1IwU~dSy@TBW?d}J>s*C9XkdC71%zwlCh4P;<CGoi<nt+f<Px|-@WK?$PBaoQsJ4M
z|54Mj{0i{(MKiOvk_tt5_4;L3a*HR>p0q37ikHx`PoaN~$tc{R&s$C{s5<eY0{?oD
z^Q3n$`rd9AfaP}fSe(h!ZLWgF@OE`3A#&4w3Kxs^7n?gc)8=XzM6FP=MQ|>ABGA1(
z&a0)RbsBIM&TBO4nXePNXNI{h3|O#|RQO+L20-TF2u2Y83oVu;<S5uR*V)+GuJ4Mx
zqkixX`?A%)jeVHW`)s^k?nMIF!J3}i0f2tulgoChQs+S2;Fv1mn96Rff@ENiDWC=w
z9*nT+U3^dP1^|zr{T+^=nkn<${#cb2`1{6}ckif>wlQav>K`VaZ!}4d350O@V~S}@
z-;<S2!Ddkc*&o=&xzKOP%}oU|*y@1JO!e=_<xBm3--d?+aP=<(Nry}Twv(QPbv;Y;
zJ<LFix1i{^mZ~g1CtXFs!CS2NDSYku-4`@L(T|6y*s4rV0e^IGuktyU(ldIj1tO-(
z-e^(qYk{w|?sx+XZU&WJEQHoXDrb}_gfV@7etw^%BokaFoSHhSV|Jz=`C9uZ@-aXX
zr*aeT<KkDlWS4={;d(uHz6+4>9d60M&urB4T2lP2Xqh>$(M7h|S$*)JF_4u~;Ugq6
zJf@lPnU<8Fx(IPO#J?L_g_6C!eZk6ic498`pmApR{<pL#{5S2Gq&;FAeKQr9L924x
zZYhg4?ANdU6S5<$=)qZl1XV3*n>IeI#iqr7%KL-h#KYKFG5}{7Gbp@Pm#<-5@MVB$
z(tAabl$1mPLV~ZLDGo-+$iI28S)+cgVN89wQ6X%oQ+?KXe@}smq^DjA{^y){36#Zc
z`gx(D+gIklQ^U4@zu+irk_dt~CopR%@%bd|epqC`)!ZxZu*wwW{tgG2fVW}VGL6ms
z9XHWlih_A}x)rA!PDugARG`=CZb7`*Q$-p$f#2cL>HR@XwGCF%wsL6F{pDxKmO$-z
zZCR9IgUKTlx3siaB^>=arq4yeu$wxS$&{zY6ZGYy!4lkXSy$uR6rgmE&EaEcQ5_A)
z$ErP1;u{2dU8g?$6q{3blX11hjX)T}WafvWYIe>F|E7uPhzLK>E7LCP>PPqKQ@M_;
z9tUsmxU$T)hif4>J>4Hv{O_P7&QZHh(*nbj7?|t3<Ml$=AbwGsXYz`mH>n)qdrk3@
z#>Lo$lZ9C|Y@K2qRL}6Oo5~>y=TK(Sbw&Qbv8b)Bz1XV*3$&4#ynOv(xi3pFKZUq#
zXPwv>ab-D&ABl#zP0{yso0>Hgn4I{7v9~zf<7)*114bTihp+Wc$3ZmYB0ELqpvsHY
z@ySN3FZ>ys1Si3{#mO*O&KJw3lyO3BK>S75Zp_!~ksB;lIOR41c6}Q4M&s&Vo@Os`
z(|NRI6#%L{V+@aYcBS4rwz{clyQHL~qp<6mTfX)YEEM|sd9h&rwWzcrx&LWB6(I7)
z`wgWas|ko6k*!9>V|LXCwwA+0dK_s8Cx)d*(qO|#Lo(ca6ciYp^%~Oc{=RLPs!v{j
zs>iXK)crF@F3cBy@BCY4T5>-pKAuJNSLpNQGGs+rn8LwcRd)*RVGceBE%)O?lq;IH
z!(b^$9XrdYNB^}|*~6@UpWj8S0whpIRF(Xp-1F&$t0(_%7kmvGU<!`)dXCXIt6&9L
z{$T~)*lp3^j^giYZwXR&L(kak+k2eo;$WN7FSV(QQF9TQ1DWR9aZBL&zB}PT45zB!
zzTE)k^EeP9XAgLFLeAgm7{seY=^trA=B1U({DOB&>%upK45SpJmy%eUxla%IF5RVq
zcMGOAMK~`Z(tR>=K}#3uLyGL2nwv}e38J<dHrLFh=6ET8#F%zZynPFjpT^Sp(cGxb
z{R+mhbw?lg$W`^=k6|0r?HEME^}ya24J}shr$Oa>ji`Fo?C#j}u3}5tA(3~;gegCr
z92^c51nj5CV3~!L07>JzZ0TWP;X_!3aP{ZsXyIOTr(|^8In46g5{^&}j%--UV!AJK
z_<co*sQm(=NlSTIdsCyb2Gf{6{XDp3hB4e^4Sgz%RND@KbU$Yg=r10gfEXjaWxav8
z@4mbx+#OPK^p@}9QJt3*1<ZKA`ZIqce2H3H))JADfh8JVN4IS$Yd5F+?+(hSe0JdA
zVh>D_OgBbwA;sau7qxSAJd8QkCwrSyTiR5WhwbDxo>jtOQ&D$I4MSLeM5wGPWkiV@
zSkNPxPOi7VMo-e^(Dcph$Gm+_F$Szr^1WPB0{e2Q7E&Uwm}hr;pN}eejGzSyknM<5
z@DAI&Q}DJo=q8Fgus`uY3jq!DCR+81ZEN#^tt-kwLu|~`vb-^7n0#fK2i8B0gk}4B
zb?goy?QXmUXN`K8;giD#sr6$O7@mMUVR-K)2ug6D7n|ZF+C*Y(obdPRe;l@Dxi2q@
z{-eY-fBszFwoQF!&zh`MxLHEBJCo`Irsc>&g^gqo9%B#GyBbE8XZ#DK4J50&v2h!q
zs5$d0$j!68q4(q2QuxR0z&CdG^c4O4=(mcsB+)`>6$;F?Q}w2Ng;=z{i{0-cUJcUx
z5cAcP1Iq$e#daj+7fC;Ot2yxfJ(~Ifjm)VC!*dZ3fOkTkAI)t2!5Op*NG+e>l3B5Y
z{Dty;3yt%ok*SY_MvIdEc9EBc;uduZn73G+ZuCDMT6Y|^fbK{12S(Cti+SQ289LBJ
zH4?UvYTB(yx>jaV6e8YGo^N0C>d*S;pjE0^Iw)ohl#tpU{8c-7@)dCW`i|`4<4YEQ
z2I^?rYXh>(K&J2(e22Sv!sE|i!*1|AKB97O(bIJZ3lXw<p?XCO&pZZy)He1D0S)Ug
zUc5el6@^nfcJ5pYUnFR99*>{B6`p%guwOQO`X(%G7$c<%(kBGvT+B9DVr90#(8&Ye
za984JlyDX_#-Hz^%KMIW8l@&tbb@$Gcd#gH8&r&kQ1U_UzmE0-v}hZ<ZN1~Z`d0LE
zV_#jnr{4}@4k*zmGizjed)&wprkJmzqxGHb*jodN@VAjgvu4sSnIEy|n7ZHdc}xcj
zEx>TkK%)-5+#jaNEv5Gp5(JOfW8qhTbN3RviIn@VPl@=RLxu;tGSq8Y5(|s2v^;g2
z=E5*d&N0Y|DQS@_)CfD`@dp*+u&m^smVR5`=abg<6^l}Wg4;2qoZcYnKeq`+k?9JE
zTO!s<PYGMbc-T)e|HGR%De_WcQGP4v1W6ywfJc|dlRxza`ML;7=`w%((P99W433rq
zhPP)fs^~$-z6Ij8=%<RB+FRHUvmy*FQ~GYT!o~9%71o1oG+~AB5da-l{<;}inmTh$
z9Mx=}DdPP-HFcBN9A2idKjj^(9PEj@H;4+W{N{E-EQyQj7GAl0nG)3bv#ZcvL#{Qz
z@EkcRt<FSBD~S3v&(F{bmj-!>*F=~fiz_6s6_UGPY5Y|84~osp+9E*7HGKJGkij8~
z5BqQxIN*p84)EdXVgC8^XHK&j1<^>y_EjSr!b}ioRRk8-k#fv${>sbp@(h*)iX=Fy
z%RkbSrd86;h>p_tonjC&rQN3F^K()33nkow1%dG364Mc!eW{(^awXUuCM0R%$+vj^
z(}Xzm6~&=0r2DTBI!=Ac6*)fLX9d|pU*EBU_LjQQ>@{SH%;h5(zi<O@{5#-6oEQc{
z5M~qZ&C8hEN9tAc>#8G@;LlNB(jwvf_;h3x)fj_}hd74O%d1wWs(_OPLE}%{`ae5b
zTB=Zi9KtB0n#kP&^DG8_2T?cO&7eJt`+MsFF7GS|Z#=j73`=luk{79kxu#SL(JVzc
zQ&Ll=q0#UiY##f^V4Q3VZHK9w=x@>)p?i-8!6wpFwG%#qx2Ubxe>(sCMDC6o{C-f~
zlhITsEO!&WBpB9%va(a}fu}mu{x~^V2+DesG&uU9@qv_$peAU!PLnYK`&mV?=f(%R
z;_t;*?5dbP`gwH?onlf%=mrZcDky!h##Y`1hFOqclwj6NBF@X>iIJ}gq`Ipi^wA?W
zyp|A*CCvYM47L@bZbMRq3XdcUL2i$3rPon&KUU}RU$>^9cyquxUZUZJ=>u3fO^psJ
zc=M<rCv1X_vY(5K9bOvPXlg*_goTG|t`*f`#a0eETy8fB-lrg>e9@tbIa#`Y4(jGp
z0Li@dSwo2-KmsYLsf6IiNuVmB4TWP}lTt)G5Hkv%Uz-fY=ICdh=+<Qrx8Dct_Xg%0
z?S;isNLcuUnHiUi*WUmvt|Z|G4KT6=s~4OPVZ#ju@@>J-aRlYr;}2MRvGyJt0Z0(+
zLzyLR;-jliV2GB6VC$U^7K*5F)75s5fBpJOVr<PGw1sM$Z`IkqjRxRMvh$e}H#~6r
z|4hOilJ`lp=f7S6bqvQL?!X6kuy2G+*1$auw}s&T3gJ*<8e^jvHf-1suDAeRTL`|#
zrj2Cg(Q!D7o?Ks3uXO9ucihQN^mm!>?}wr%F?6_N+yG_EJ!xy;l3+{<fJx&)%nHt`
zV^QD%XzgS;3^%vG<cpBPh_lPv=&XJvTbOuAnhp*PT?RHfC6&_<v$BCqu3<+*Ya#Aw
z-jwx5o@40hYO5zXuJ^&cdp>xr-@)7Xp@k3j2p@#COL*0Btk2*iI1_(MQ!C=yGHfw>
zMnj5<*LW&~jdx+NgdjQy21?O`%8a3)kJ=xDci6afE1~Un=`h`DUO+2!t?d)EnPM$R
zW^*Yp1NnAj9uJD!u>1FY5qK}-*AQP@Nja*V-&q}C=Jd+04S!hE-F>N?C(O>5++ln=
zN6LtlO<T5HE`EtWbjJ2Xr0DYd^$+GQavV4i_(<uLrqn6gUAq)mK4`tc6EA>?FSTAp
zRkemlHfUyVx(epiWF0rdp-bz=ib_Ft&0|Jc0Xjx0ZnfA4&!p@CI(vlnSB9_W{Kl8s
zFuEpRY%uulUFtEZ$hqyBqRFhGu~R=mzQQc<7yH0#7Spukq57f6a5KsiQ_t1?`}Ze-
z`$XCkhR{n|%@EN4hrPE9t1@fb$G47-j$<nZ5+*8wNGTv?kcx<uAR!<PA|NF>ItDEQ
z(x8HLNQ;z?(j~bC2_>baq~UiibmsAS-*5lp{qX-V`#7F?r1rk=b+2{Bd7amJ?cwF!
zNz^+4#SwxDy8nEIE2atNNoX(@ZHUM>?%e52rqH&>eRySYWFjE)`SSu=n=8@z6WG&{
zzc%z^xiGbbxi}Q6GXk)C?|>NXUvL3e(c0FQ<n(eahsyB1PE1w<=L!&6VL(#+QXF$r
z&~afc%7CX*cZpNGu2h55^U)NTr{B~i2p?~Ppb@;t?`xi54x<zHEBs68VC~ICX9St0
z#`&flzY@L|two4C08~w6fop!@JNzuCZ+Hz2^Hub1`XAF+-oJpr@H0A(ztqQV$8#p4
zI$(6-;3JaVPUI<wO!13-#uCD--{9ezLKX~!?Kf}V{(>m?4)8G+7M81;nzwO37<&9G
z8lG1F_S-3F21Q9Vr3$gKu0ure78ew4Q{=wuK0MG!gDT3!;LXVUUqF_zckA~o1Er5*
zMwx>(OQeHyfdJ-`f4#8)Vo8h*WOAZ!Rf1!1y^s?j)n4(Px&VJyP%Ln$5BD-ZLiPJt
z9}yP5YXhOuLlQ!$Mz(5$pDHnmn{&6?37*OdN^)xIKI9LYQjA-+oB=w!qlg1CI+Yxl
zq+2E4201}OxDWh>y&8dFas{Qjwzgy5aotXivyU=P-(9mGNe*W&$Gdm$9zqK0A;5kw
z$Oj!ckDfjWP4u|MC}Nw(=-1mPKCcP@F2!pzx}SRx-seGXZf-L#Ic`Zaas7PCDc646
zmx~x5{N&H&XIAe)db=NGbySA`AS~>7{cFS0t5&VLYw3E1v|5G9V}EuTpUJ4KIR~Rl
z(_P=t&~0~p8!HTCWD@|k-}ON*WIJj1n3SWHfG5=1fvgo=^pW)1x1vzqz>h1FZ8SYR
z!-(RaRK*E^sFhk<Bmro~XhU}SM)x3|eUet!t%gmfi?O|&VPs1eKBALPrVJym_bf3M
zbp>2oMN(Y<_+y#j3~qc2k}qF+zh{f407U9?R;{P|R35h$R!X776uOK~vCID4h;?Rt
zi^ujtAn<WK0L|q7*!B0;)0a*QS%G005qId1LBRnKx@=Id(S=lNWr_Yupu=X13-)SV
z(D4cKLVl4|mj?H?59rm}mS3D86?!jqEdW!ACiru2#3^bP$k4nykJLzk%l0Ji2toyC
zO!W@G4w_D|HadU(y;>U@9+Sr1NPBPyKwDY2%)hDXufJ}m0!6nxR!VZGq>k;k`6e>Q
zZ7z<N`!XRaGHph$k%rX4gj(0cQ%&5D&`})YIbgAak+CW;3T;BCA2i#JAtYa8eB_)>
zHOf<?kW7v33{e?ZyC5a?{P6*MAqh5@rgJKSfk3cRx~n4=D$yC$ZEk7V&3{}Q>`He@
zPh-L`>x#bQx_md%)jBOc>#t8F%U}Vv5y^)MM<npx^}i}+3)iB8cJVVi8uea8#Ko{f
zu)<n(6##hF-q_NXB$x3|Fi_S)w%5EI*d{?NvDzszY>$!wNG0Wg2lwv1{geiTq+;|N
z@$}0EPNZ%`Lkkgrf<c!3@U6VKcsFvwE{pZpyzo-z@Vona!Kvf<cpm9BD3?rbhRZ2C
z(FAIEOG#DHM;EpOQ-`FF+Ji=dUe{&&q`~GC$C*zDmTbQlUubsrVtgxf1@7E-jD8=*
zVcvcIQkI!^ER#>J5Y6;2u4lx>SNEHjLLUM_$ymm|na>MXP#Jc7>4EsrH4x3*uBiBa
z<x>-Bd&1S&{iQBu3pbnMF0Zl^kl#Jnz>mVdo363x_QXY*TI2PnSNB3rOwz*;<FA-w
zHYjF;^5A@X(D$yT6PL)aep3oS-baah_e8uucwn|o0bP(r5YLs314YIQ{)sfriME28
zHO8FzT}NP!NRXJ`yty9*vQKp`e{KZ%Sr{F8W&KLZ4HKD>O{OV>0q7teow@hO$LB1n
zXu_o4!zblEe-QXgOkOQ=+g8h$b~nIp`&^!l`IN20r4W0MrI+e6K_e|;%jjc_a;e9;
z>c!~8^A!_9hf;4$Sbd(!dosX*FjcKcsUhzUj%d)Wavb|3y?r#h{^4h^elMT7@Oc@-
zT+Ti7Sa_r&=%D==KxBz8Nknh-o=v=)l($(oa`Q3@Dfc1c{0oQkeY_GUDMcWC07dOA
zo*4Ql?mI$y4a}@Z{E#uM3($8VK>(xU+#lfhQC?$xKDW2u#-#bz1qCYsu%N95Mq>rS
z-ZiIJV<VYP4yji5dOaWc_U-X`rNEGobr9kOBY87`a5tHcAWyr+u-qMmSR4jAZpSGX
zfms8I;17fsPar>vEc_H=K9o5<5FUL(mB%e7t#=3a8}wBj*u;qD@Ar7*HwQj!0+ewE
z-sC{fCxiv7H*63C^#B^Ft6?pdnwFIMUU}e6|AfRN2^A?1{rw*x8+eW8a_nyr@U*XL
zX#4?8khCTvJS>7+BJJ0RRF_e7_pg?!fowQLQOb;kQ2a7{CnOKd_V(-7!VO-tr%~K^
zhaPE2vm2I_|6bZYG*r1$M75ShJ&z7j_3(LzGmyNqHi)m&CjG>pzJ7jN_h6OX;aiD7
zh-hqIeSxa+UD}pQwGf@zj<zVGGj?s8XZyvPYGrMd?bk90+C?8T;NA|19K+doP`;GA
z%vmq&`tl>LY8^FVj)cHCE#N`#d}NHs?|yEs0C0=@c!UIG1FRwrW*68?pktWAhTbDg
zMwE=ULesuoeh7)E3rvQD&K!`yZAAGD4g=lZd%@!&{ksT3Q2ts;nL-N+`KZ>S?!iIU
z07(H9y3n9}3MZT3IHLQJ4aJd4Vf{UL|M-LiXofBWhmt%Yfm`FkctRt<AesRtAk@<>
z@6^!IiM$bLjl5MfVhDW_;iwGl;5FNiLZ?i!rU<xNr#Ox%U<=>~y~pyBo*e2(rXh7b
zax7@LQ7DAyhvY^Cy^9Dn{(8Vsk6(K}Jf9MqFXm2r7x>q?g)n&JJv=-+pdcy=jTZdg
zUfxxNu7x*tb9W!X5So$D+=SIKjz=PMzL-2YX4ZxvyerY_DZ~#^Va%!fA}8bM$BK!4
z&UoYlfJi=M;2{vzHnmM71W(92K$yJmLH*ODikyvLIi~T~3iAB10DAAAECciOPjn}|
zhSVY15x@U_#@d=^UlkId>PV?w7qYcQgpNQR5s@qrygJ6sErP1``y3l{l6N6dzlVgH
zWIiw*D^{)I+$#y#W-B(Xd~9?3B^38j0y@`v2WaqVa45igA;JW8@k(28bgdyQfOe7n
zJy)T}O4wznqwR~%K|jJ$!H*nfWLzYb1qqaZo`eSm@6`cZ<<U3}HOUE4_B)@hj}rnA
z4>}NAQ1+)dA%NM-#kG|jJ;+mC)zZ2H5vIKY0`%ytdWV{oDJoT2-_#{K`ul%JsmwW&
zbuTwJ4H8CqYoVCs=*y9^nS6Wr%-zLQt0)--dELL9+G1~Sj%NPra=D(|)@33a<sLp=
zxsnEnopXp<^#}ICcs<PlrZc-f#(3}0s$pkm)%T0@lhx#s`ID!^>Fj0Z9q()uHo6?!
zcfmRzGb^@ow=fUAP8j1iY+)R*trTRls7kms$qVrrsi2q8b?%5yt!(PgQ||NEg5alV
z1vj<s<dy5PgN8j7$mcpBeBn;S{j)tWF7zE*M8e)gbG+r*ZHt-nn>_`-T@JPzEy>oA
zC)0ZE1`X9+%Z_HVRPuDV@nFVvI5yV^vXCMihmPbT23I(l0?<<p#wV&4=c<q-G^cr7
z5vLv%5UPHOap8L^1tyz$(1oQ1(NELyYO!JE`sawTKszbKa8i+G(;|8l{#{gO<>ll$
zjhQA?dfFOki(2f)D%j0Bi&cBxrB(^<!uIw+q#<~L=kuRD<L}RO{H{KA8Ye~oVJstx
zyLK$9ueczr?3`Q4$r*#t5sJT4)}Q`ZWsrLU`lvj8(cvi!Z+Uxyw+JmxS9L(6w<XR&
z*<e8xbRh)Qt{omH6ANA5RLBEi0B=5CjOPB@otyI)CX&YJMMIibMe%kzhj||?-O}Ye
zeLy-sh~MkH*TJDZu6M8UYG}s>24-{AXXZ^L)IonXfEh7DoU#e}LG&wF0wkYb+8{7~
zF*j9tc$yD2Fj<lauA}2QHW)@lLNB|D{^rGS`x9xeE@{yQ_9iA%wf$5ZuC_HU#0x?3
zzE0v`h>v9s@CwI1%nyu~(GMOfm22epv>f>0HPbqoM;&PVrKdbVRyRLOEub$Uvo&^X
z5w(K2Te>^-Sh`!(-7DPkIq%)f^X@JG^cB4oQDg;m9EI{`>XVN1#OW~m&ick&im9r~
z#6ylREZ-z$AA)6!LG9?(o2fP0+OcxXxk?hXsx4YXKU(KGianSKd~Uu^G2dW(M^^AA
zt0>e^1u-7^lTH{XBFpphod<>V$4`3o2W+!^1>N*oH6^V{Djb7sQtmh9!uI0MQaKC>
zJ>{OSm7R~B^Ela`Gv=W}WK-|sE7v73h3C0an=6dvbG2K&&zTs<M$6cwXLl@qGg*8b
z3s+s;@S%L>h(f%0ze#Us>iWT~y_cYUkgzkR&lWo<A{2W!ZDEnr_P9Ci?2~ZRqtq~m
zvF=3Ph`Xw>dqtD4<Z~fIb^Ev@+)P2!RvC(slFtJz*JmK#=Ehc0*7|r;A(Hvs+)3Zk
z!BRO^)<rxtPg(3_jnsU`eWGsBR!sRVT?p#rn60KWePe2Dxzjl0nSCYgA@&?IR(aAS
zNoNub8|V7i9H-AU36-5=%T23vh>eX69QC>u;u~X-!Xe*6c8sc3OMQgz9o=$&nV!-!
z7LAHyJ&}4t@ij<U4R745R<Ucq*jhCM@Sdbkc5jj9XP>}?&H9%PUf2S_2c@M#__znP
zxW6?HT?g?x1~q=}e)B_YY)Mv;O)81!ai5nkP1Q1l-C_1)+lA~6)XxjUU3@At_-uLR
z2^&EHm~YX=w%+DDrS<;%{Eo;Djz<a2$23x-_X;v@Xu-i;$FPs_i0t!AVO7h8H2(Ti
zXFfk~u8S=S{r!)UFLPvZ$Uk-a#ZvLmk@W-60(w=a9~IINZyp7-Kjui{%V#A;&3;L$
zw;cA`y9-3gw#5N@dSe&;er#C9qXXwQjIYZFVL9*C?c1rdyqz=C&WrPHiwT^CD1k?|
z6t%|lJj5Mjwpd<K7RtP*Q`&ZT`?Qr{Xc)S*&NRh+X_x@^E8l0RKC2u)xqOYjb}aDj
zI=}q6t~#JpwEPNr3Ur6w&3;mde6)0KN8in;kG@;9E!on4P`W($aSn(4xw0+0ZaoiD
zr&%@Uh$s7?h{2Y{gMYO6R_0-5W*69<lW=TXeSLlPCr$znc!@diyMhDku>W0uKl*Yv
zC|MW7O9kjbzIP^85SV>y#C+g0+`sATWvjUK!+@H<`(*?FMJ?`oc6zwlW(%t7NP>xd
z+?&!CWHbI8+LHVeb+S?m`9nvq^vFQ!T*b72CK+`i9vuGKrYQJ91SNHsU*6DAvFY;F
zAM`c`T;(o%f%mz>EY4_Cr|;`;O!rLeI(F^sCCdM%kmf5e&TQF+dWW=azV`*b!wld<
zzWjV|SKwSo)m?n*)SC?s8xaI}npg53B%^v798gC0be#5cbVjaJh|(@2A|~S-v(gx7
z-a7d$y^)nItMk#JBS*@_)lma-1yoif?Ko2lB}psgS#TqmDq&i!G5n6hazUpFNT?=K
zJmLy=@LS(#ngu}4k_eGFGMlA6yKVk@c$>3b$O)l^5l>=OOfkZR>hBdLNNnL8)p1^t
zMLF%zdj7Ak;|tPH;9WyMKqNKc8-j7JUd{*&#6vDrHCxua$=qQPa3N}&bkX@?TE?3B
z4$PhQhiBI7kFrC3YPbl?P5or`jQ9gw%*(#mxWMMc*MI!dF`2(;kCJ9q$?VY1w{R>B
zy1}NRciPaEDO!N{&;$gjKk*SE5Bd;g+;*Kvd^3m;d#mGozk0`$JL?TDHE-IKKV4yx
z!uQLKJw0uk!ou#HXufd<XvWhwgC2Z+(bl-9+DeoDy^Gf`pDWA9lcA@om`ed<<pE4h
z2c3A;hMin`7n?D=dgOzarwHOqj?L$oO7ed{haOH9s}_r;ZGhcj7|&emkK<J!vR!p6
z`qs_PV`9SK{B>+YzK_I3MYXI-ThTVxeWIh`iitaZvp!o2F;+WfuzLiV<ndw-bxmL9
z#nGO70|<ch6K-b8k{^F;Td*W<GjFPZO@}q{Cde6_*za{;h6}*CEI`?~z14Qu9gUhg
z^l^yJuIElPfQEYEj-v^n=v|YowkE;Iq4XyOZhU%e-5sFYUV0?yyzPM12aWp$@`|c7
z!#QIY?!>8g)!+$U(Vb;n_<CQJbGvR({|$G;Uh3z-SQJ3Dv|c>s>wfTHs>pG!=rVHU
zxi4ZuE~xcIr9-bzBG|G?f22ofkv9~dM2^dyk3IdtG11WvW{0xb@P<MW_N>N;EYSCz
z^RhQG=owHv!{otr9Zmxuv;HD#LRz3x>f85}m6o<2QJ<E~nHwz|$sM*&Dk)-#F!T!z
zi&odx)&{Bkz~CT8RD{RIGGhQr08ej(cgi+JKC=Kvv1=xDRA_p=#Gse<?jREnG}1m#
zmj*SVJ*aI?CkFwS%_5y;AfJUnNKXxC6~DLF&Y@ZeK#xN}>aysJLCmxjq~6-eo@=&h
zvp2;^j3|pFp7ycit^<B5n~7Icfu@ps_tmV@=j(YsIq4)lYXU`gD#~#t${M57J0gRw
z8`PvfP%&c+&OIP)<j<e%F#2x1WXmolX`h2bi5C-Vw=K8PG;TOs_6KxOq4+kC=(vH7
z=S8(WR9bQkW2){w3KyU^E8dgsQ6C)Si3uLg@C?U^uY8Gx>_FV$g7TOcR)P-Jf*lr3
zr$vQ<Q|CNkXJF(EBZ*;h;M_)F<U|pSYoZ`!if7(1_4Wi(8P&xLJw2E-v3NPp&Ia_3
zV;J>SZQc9akKpl3f5YsHQ7kACa`O&_5Zana&WI>@iF*(LY(u8+iiJ5rf;l)gPZ&@U
zg!gg^3WaDt)ts9e(LcC1YbyRDxC%J<#ws_fMY137Oh`=hAbn_n3&I^27s!g!3WdL2
zf6_>CQEI+XyPw}>YHs^CNLnq0yqNaJ1M!#5_4W112rmw?vbv*Vqu}X%23nXqcRb<a
zw?1oD4eJ673ZhYzRk8kj*AyPO!61L6a96GJJ@YDms}omWJEdt(EllP!!i1XKuxYrt
z>Nv(eAd+?iEH0Y!t;ui%0YecIFM3c_U19sq!mdNy?J)C^BOU%)9?tk(qS9N_cd~tK
z)&Bn3#p9N;HzvnkCOYim#M@5|FD?vEVUP#|88ZcoVND&rjbu*@&3I&=uIKH1$TRe^
zLlcfrigrJ%TF%SLZ^S-W<&A};%}Nx(z-b|KvB>kQY;(m?Dxy(T*#v{4nER2d%jz;%
zoMZt_Ki}P6h(_ji1oNIGP{$~^{34(`#D7&|`=Ke>gwCqjl!W2Y(b3}Bix^hWFxz;-
z(u>$f85GLm1VXAZz^bYgoE_kq9Z2o)RaA2ZfL?@Cu!XEMf_x-`VlcJ5sDClQDR}}-
zI5BNyWZzTad@yv!!BA^W?-_$w`251R8-(b9)_4$KYF;I<9nISAWcjca$QwYha3d_T
zN4N(c^Wc5fB|kgika3^kmY(-aQ<mLy=+T_7e=!w_yagAVQBlQIM8&dEJNa74Bh~3o
zKn%3-I3O$E6CEG#2Gpyl<X~8i;@y^$Ay%z6?B?C&Q)6L^bkh7IXLI1^WdIUkKRKgI
zxDeQk7V-HHsu;I6&F4t=*xDe}E5ZbzEhKk{3cbbL3N!uzy6x`TS+%?eVO?n*=Q<cW
zkXw0PG0EjtgYULR97k$RrteR!5|Njer!y~Sms(f^5$hzwtuIflVHLC>JTroDw+Lu_
zcuES#dcmnr=#$d{l;ueMhOqGC1gCh<=`(nOb0eOP&AuH;ko^I5Y5P(<y=tx1f?E)C
z(DWn@DFa-N`-8~&LC8bIn}%{d)^(VOQ+EekP=FiWiDIS;EI6AW;tcR2Z0T;iUM7R9
z>R4`zt!Xgc<`bs6r>kw#rXqSP6^tU$r{+&~?=|OBR$idrWIB*sR7l<hj~IhHrQe}w
zq6K_}9IL5LZ|P?^Pt^z|i1&A65a3&Xa^OYUtjCny+Wgt!)DYTTY=broC$!rh%yqTl
zUPi(er^ECoy%S*`>SxFFJZ2aqTgFbNXVSi!&di3_e}@!u3o#6+_3?+=4rN(mt~z1D
zqhi)T;wm{gxwGJ~Xdw)MH2y;#%nhy0YfP=hm_{0C^~Ki<;#{HbbJ~06_RPSvy)g@8
zJhpVQy{yB1Mp&xs>ua9RT3xdONI@#d?~ozWBN6_P*dLGbsvFq^z}^TaQfi1e(wPA|
z6JqyAFi^43l`U@vIq5K<!><I>Ig1!z_KA;9(DYTYZ$RIi3xZO%k#f$ecNlNB%MsBy
zF^i}fZ6fau>Rr)J-C&sxjmefIb*?U5Omm2PIa+$812d;Qu3XPLuyo51-3&)v2Q&V{
z$erK=l9hOsVH1F=9K^{Ven;*WEnaR^bw`1v!wieJgM@Y%i&h9m1<Anm3vzoVQ+c;s
z=PE!SJsHG79$=lGV^4@x`<qpa=le}O`)@23S-c$Q_Njet?mn1Sd%-oyuItDM<T$9=
zQnHmnMxMMLJnavrAd%m1m-MtM|Eb{}F*?(;;Y9$!3{}iZ4rCYVA$Gh*oCmS#AeXse
zmx`gG`5DBfz|MI9B<z|uLvithzm_!E7VfY&g@9Nz-`rj~g1DuSoL3Y;M((oN?ZLnX
zS}}SHIn*sOBzn%sa&a9sDt+rwxGaksX|Odg1!C@!ufDlWybo$6OtEos;V?@uxG)$D
zpwEASPc%wRnjVx}LMCJ~VV(u>TOHWNORjL?TV{R%Xb6_7`SWOT08wEtDm+^V?m%Mb
zgxR;O?28`7tvrbcT0qgJCqXEoeC+e&PwBU7vtdnuxdOlB3D3-zaAM~5xsr+}#<q~;
z0LJXZ(fqkl<>40C34I#36F1i#4f&1H)x~8gNIQ2lE8@){ZD5YIRbyrT*!0{xOajvc
zJrV4WrX|CGa5@?34oQ9NzP?#i4QEAWh|!t8+r(~A7Y*{baMOl5(axibd5CX1U_(4J
zTW=vtzPt(h-wkaq?@&0>0&^F2<w#<ZeSCe1k0n9hdI1kp<%5570mk3&KwP?(Q7}h>
zdSchB{P!xo&0-de!2moOq(d*CuzAMMF5Brm`mUvG#-`+6;>Zax)v$1WQ^rRX+UaIb
zlePC8VP+nsZx=MI>Jod(8!T4Ai2#AmQ5T2Ue0se*9WA@&*^r($^j6JY!|y+c88@aJ
z0I=Dw+;bkAxv%0xTe+3>$}ar>d0!u&x*4T!{ZzKW=gDKAfod5Xn+(C`&pQ*b47hgF
zgwi?&exA0^7oT*tRqR`2Ldpw*n{1EGjfLr+o|9a(HxNBhr?Xma)=EvUPk4+6#^oO;
za@GL(@-$}0v^vZViaIiE^5W}{PaUQ26VxXlhr<QpEc4Dhqs${APB3LKMx4^NhvrW>
zruykw1Bg+P`!J*T=swn69DJH~)j%S0&o|AJ_jX+`wtB$z+Tz5w$EGS{*ou*blO{cJ
z*n}~jf-{HOK6~vO9WFmyH~Yy?#Y`fXx30?JTdLjtOlqg|U8&IG)<*}Ad-8k9Nrw7M
zu(C^+pLOt^?pjXKt3z<?Sl8$<m`?5H**vOEOdmo4<v~19R~W3$J$u)@3HKCK?ny~s
z%BtXNs?MCf+mCcuo5;E-%z8SEhchKTro7IRUXVJwl$FK}I{mV~xMb}JmoyNRW*~0q
zO{jb^s?XH;jc(jTkvN87EX0-?rZ4tNJGz``v*Vmx=CwWXZ8cpXU4RB1<sl1_V^gd(
z)j%CkZ8MhB=D`EIhmN`)v!5twTSw3C?Q{;<55P>5*@>UT)S8IK)0;3qt(*A6v5lLu
zz7_BDe1}w4IXnSXl94#|d1mQNGOHP|&y|f~aov-Ub=f?Kyn&N+*kLlet=@o4=g`7~
zwt2AH$;jI=6>f**9|q}Q#GWWEEycz6z8reh-;RhPzTTlx+wZ{4?W2@Ns{$IuI)jhN
zTK6^QR+!RzL~?W&({)gsI<o*%g~Ms3Hzj3t2lAgiBUqfz`PTjhMVZ!S@94YohuHe;
zG{T`Owu9+1Dk@{tuVRe$TNPC;<U{xF_3M)yb(k(w7e(~U@a*Zu0h`t|<B!KtH`@j0
zX5$ilOcj4d&YkjLK5{LCd9`1kbUk!D_>lxW7@Vo(o<UM^a8|;i9_W=L1Z&@(KP8gz
zjli=AKCns@2@U`d{`@v(3E$datCA5y^PfGNn6+L^cBlEZd!`g&4&VfbkLHeJSCDu*
zd+61X4V%0SvE6Pk(m6dA4Sm4R&e`caVdy(dp*%7GiD={w<J@i%sM8LTv~tO?fH}-B
zB6@Tnpfd!i7noxM{32)epl&M_kHLGQAHnHmz#O+5p?jInBh7dKMVA7Y6T`F(J301y
z2rW$Ui>wqrTAc4~Tr~%RTB|8Mb<bhinvvJPY<LGBWSCB!xO#wc?Je2(#GFN3ilH6E
zRKf?Tm3&8}WxRg>X?Ll>DXH;e112N9r5*I-9~BSTLjT_E?}x)|qiI&Mh4W14hhqhD
z6dO6kdVbb6*g1kxSEDxk;K3pAY#`1iSS6F3WINsEXShEN(7p#jZ#*~AL_T`7nbVuH
z-+%68mLOn0+N?Tcz0VM&*|<F@JRibWUp_|ZtjEJBu*QhvQTktPKJn32Xq0>HEL=t@
zLEsVh638+6{_e%fwUcj_>BV$9mmH8TUu&cJcB||%VUKi^w$?j4xEfK_(?Sd7^!i^Y
zPj=4_S>=~mt)@`6uBLw0Td%dA;?(Ws_zUIYcALFfPyCtRAZw#X^zHfJMJIhlH~(^q
zc-7w_-D;aH6$cl8b-IT5C7S)C8hq;Lj&*lqQOn}43|B~K-n^U=?b&~QfVG2%q8HHV
zJiuzR_jO)bNa~6!FRq@-JM(1a4Tfcuh@RNE#8Ru(X_QeDiuk=V8wBc%YzNX&k$Z=!
zu(%_N@TnUu?z!4@t0=7A{9Z^qSA4Ld={ZUH`Q@Z?@%+l4zp?)|rcD0y+Pw*@@YB~6
z=O_Mae*Ri~1)khbFX6q1BA5U4HD&WDOgsGP^>8)jiT?C*I{WV>P$>VuSOq;;mJ}q9
z#_Wi;?e62b+ZrP)SeKWtSG_ljjUpUshgfZT{nOnon1-4R((R95!k?_mTO?^Osa16z
zqNO<fh4eCHc;l#K=xWNvOrsuW);#|{AhNC`Q*G0rP=p)sNm4@?^m(cgiX>n*uz&w{
zf?DpY>fcf+6mbTdy%b@Gt^fCX_RnW>%)X^o8*NF5N_y*-?t?;E*$<$`h2UK2O(;H&
zd~jTxAKNHXR);8*q$zoCi=9y+c?Y{eP8em@AJnoU_98mjxcU1+a~i>L1Mq1SCwnEx
zTZIG4YSS3-T}soCT-)8Renu9~^v%lcbp}u>bwGjm(1uORDARBfQGjO+Z)V@lH&<6s
zc--K)m7J02D|nxNqhaLSP+mD$RQ-OlkB<)z=43`chxw@dh8dj2jLRr`UhpTr!yA=z
zP*tZkwkBLT(Xp2DM6F<Y{ZB^?5#siLk26-`U!#Uy@$V{-qw()5{C`}9v{G0hyo9?k
z&6b;#debuOX9qL;jnu&kfH_aDo(CSUg*a8pWdntx$NQK~odG?-`r#iFFv7O-8$pai
zF=02&!=iZ4{j-R*z1uOMdrcY2>ROo9p>I4}xXYxjapAAu+0LRq8|Mj-Q52|4)5tQ)
z-K0al*CUs%V5=L^MFE^orfe$-HB60kbN~q{pD~4BwXZ!K6}%boY4MCHdamgSJ_#GC
zJAWCMzF>zc4{WO^cqH}>p4`B*FTc8}-+&=|16N$e=$GF9p$!cU4aT}f1k)KJXW`|A
zNqu30itAOd_0}Mk*}V*7c1hjgV38UA@{z>%fG6!DAcoQzcKlEa8K;sE`McpjqyQwl
z=r1ZM*s+R2`Ea6h?;_jLqaKLduFdlxy%YQU&ygUzz`kFb*l{q-&I~*y_BhNXp?ovO
z^_cd>MCnZI0AAxtkdW^`pa&@B3akh}0GjO#R8V(f2j20Y3&P`3M<)|+T+uO3ehOuy
z*L?CRV1}3-Onwo+84hp+>C5AKt(Qdzke`2#^^{*2wXVjTde7PmCs_^W@r>?v_{d%;
z=_`E+z^z=ljr{51ph=#;Z?&J1@*(r^QXJ^C4O_vq8bN#fxT?H74X(IM5`Q49(TMqr
z?QWcv#gc9CscztU>pVB1P_DLFl(QF+nXoA}lW4>UM^dTOyNmK+^6xi{hb(WTbXcfw
zUMlat{S?YY>lAKLOoY7U^Z4<Pw>Xabk00-6SVh@?MFv}^>m-78JSp3S8ESqZB$POs
zbP;%K!l#27*YEc2cf|6ERf4lD1T+Hopj0oPAaQ_cbkd|Llqbh>kYSplFs7YWMT%=c
znn<N8e119Q?%UF(2>AV9^tJ=Zra*Ue)2fgx9*#g9Y;XvV7GL1c2#6Og6VA~Dq~;<I
z#O)aYPF3oK;Q?S5?f(4TGYc<=_NYMkJXq>S!sii!v7U)6S_2^m4~TBMeftg9`S`pi
zXLARI;&LfQj=AmB(k5-3z#(~u!YxOQA}W#XKwsS2HO;y42I-`3G=kFJwTL97>3pCv
z2YTh<X#6!^d2;w2I3<h$)}2|%)JO1dG4v(>y7w2#8sDg;?Z0~&0e^XruyfW>fro6p
zgtwdsM5RP@hIi`I9UAgA!8A!lgQJFuFok^UdyK#PF#4|2r6;gs(#{i9f}^G?KsPkN
z%Yl4+4<64$36Yrd@iw?E3T0r7z&R+N{i11d^f!#85W>9qyK6;}proA@W3agf(wbe;
zjP*#t_Kx>wHdI_Zi*2+g4n-v3&ILFlH3Nd6!R^GpvP6`0?HC8;$wsAsrF3QU-?(LN
zpBCn!?sP%)rdRUKGGh|h!_t$o>;)bdeB-a*Z0n1l(M4`z)6?(!f3vm;qaG4VoqKz6
zv&t#xC=r7hN}X)`n2e3{eso}v=MB#<ln*h70AP*MuON5xa2LS!YCO0pz>z^Y!=Qf9
zlX}p$XV@0X&tP?mKnbbtoOf|`U8>)a<GKgQ*OQ9`Q6tqBQWWr>6u>Hzx=hJwIoxgv
z$5J-6UmSIFEfm%&ny7B>SO-}ytF`Pp<vZ?TFAB5%uvj{Swt!W1cxWk49{xBzf9hBM
z;!f&aKx0_Wk_RJ4FesqACv1ea@Z!QJ^oOj;B%z%iD=3GqiC~ZU4(}#+<kSp$vH491
z*FzdM^qU~s2t5HM9CMYys<2l7u(RT}vv@PYE?YXeZdm?9atNkPA-A!QutvTK5()yU
z3p(t~lqc^m|9C)=$l8gOWg65f6vZm=2J#us;ruUcunA@q`@q0q8tlNPd2#T2&iPPS
z%H~NjDPMux#v;0OGdNAvz(CS6Au(jREu1{B+mX#3<QNhQ%Z{bk_+u$n5dI1&b?6)0
zqIe?)vT+LCCOps0gol*Nah(*|fu5I{k&X5gAh-%aTN8|n+;(n(L94{~g`diNPoRIR
zgWZgdjg>U~7+L-m*5*aj&7!Ti8W}aU&@Mf!-f2Y@6*{c-Im0WHJQSx_-}7XrKP7v9
z)dY|me5)^p_W{|KF<>hAG#F$um|-K;_*PCmMU?~tN`2_?ytdN*xc7@>`B5A^@v?%H
zHqgf+hD?IODuV+Yexw7yIy-a^9)>?4A8qx+q|!EbM|fXwfdlM0v=^J=?`vQN%mORL
z{n9<)^D>Ilc|9W|QjJB4wL*pqH-GLJ4C=3KR&a08(mxwip>1qxbRb^FkL5yzizG(R
zaIw4|c<OrkC@Q8Bt)qD5^w`Qtbg%D#k*P2xJ9LjNImIvDu8$kb{mGsdVJqDg{_X%*
z_5h820P+=An?~9WbR{_~0im{}ng}g&k~d-3VbwKJyh(35Cn@<1un|eWD2Hy6(jzJu
z5el}*@WN4`uz7HZ7nY9}L5Ee|TZ)s8K7rBjdJ+RR)Mi4|%6qFf0_M>+{6(1++@bLA
znOa5bwcsdshhxWI5BAa~PaU|=d!v67E3$U~qDB8ltMGR)jkil%+tt-o4d@m5{e$^$
zAKA#3gtDXusR6;Vj?pua52gGAvpumPu>?H(t+}^Cb7$x*>SSbaMtUq^xk*|3a)yyo
zGk^@!DuGFIK+Wh_|Hm6R!?G^Xy3qc>kuXTAT0nc<Kzut!EbCN-iNh8W4H+h8F%6*t
zD*pVEl83*&4DfPrv!Aq1LNaG3PF#h?*$9;F(%uX&;B(pmHCJlDbi$%?{e$_5m+iSv
zw|Q|)XmJZF)6HrVvT|};Q5LqN1Ku){v^bMA;vF`(|IDUK;wYkTM-j|0ceo0{(Dm%W
z3_8)Xd>Avil>o`P!;2{Sxs%e|ePRkI!nOZd5*gytg08<hOz3gd4p7YHY+_uxYpw{M
zHuBr#^}^&NN?IN0tE{r#^Btt@4D>}rzx0ZxVqjA{HGjckD-F%_+$wl9Ez%^g_#u1j
zMQq>1;v$>Q#udv}oi?lF-M6*ia8I3C^tfJRH-A?$S8ZuuoQ^@aPd$4@vZoklk(s!(
za$kv(d1$GP_*oeUMR-ljcBssFf7oXv+)`C!G1&iT!XbZnR?Sv(@W7-R<dD{^U$5Y`
zS7hss9saKJe*>ylcLn}^b>tG6e2=OsBRo!z@q*mEEJjW~YDiR#BaNXu*{)>7_feCQ
zkU&JE>}6(QQQ`jRpBsOmgFxf&my-lm%R4_mzkv6ih8`-)v_cL!;MQqbwnE0p0z+&n
z!i3b}up3Zuu7l<n9rX*^dH0IApvyrVgX<(OTxft4*VbLT;-G6!S5)Qk`nB-gyLV%R
z|5%RNZ{Sh29KDgelB1ADS-EP}5&Y5JH8eFSY+FFpis)h9)9B>M^Yimtv;Am1<jyc7
ztE{Z-9JCoF#l<gbYCc03gKuc)kdr{kNvMrYPGY3-In2@{?+ux=fG7(h1i{562lB_C
z2$Q$c(<fl)$^CbhrlvAaPv4aT`;rD+Pc*BUSy@q0SnzrH@Cp1mqJqjjS)(pNK|z?n
zKto3tj|)F=-~b5)@#e~jIT_y<>8o}4_b}!Amk&kc<b04#8e6C#tOx7ZE?X1vh5|TI
z_b{q622BguhH)34-UcE68KhQ`P$)zCrIMrQ*jjpCv$G(h1QC%-)Fs~x@uLJ&58k=H
zr5$2ANeKzB{bcP>59n*oN-cfN8l(?AFEb6Q`BMJJP9Z2s$@W-PpQw5TyHOTBBXX$R
z+1$FN6R?GmF$siI{l*k+<vhDA6r41;hmIuWJcF>V?kk$Zm)^ApV!9{cc%v><4bn$C
zgsfbpRPyrjT-@Ap7R`}$vlb?)WTl`#NzZZ){j~~{7{@9FsJ8xhwwx#AUNEbbsV79p
zQ4I@NhY~u%TOlR4>B%YZCr+GX*s>)Yk;%Ckb#GZX(g~0b`U|$Jmydp+hYB26D>Y!7
zUpL=(HOs6ETP^&&=am2s?el_c3^yTMMvoy*=v{jc+MfkFc3sEd+K*t{NT8AJVL=vT
z*taoe!W4Z0XJVU{_$8Fpmp%amMOmyySATzy_5}ljSh(v%Fpu=Gi(Q<ZlWDhqe{MI}
z=|(J;k~NqMhCmK&7m^IMX_KFNrlXA-FzkhGJ$CaG)LR|yCq|?9#z6LfO!5JNW*?*o
zvKtt9ZU`gZ!H~i@BQ;J;fN>2I2))&JI7HB18TniTYH3{^9a`+68at5cr(iIkj`_Mt
zB0t9HFr^Wxv;PY8ntH+NtOmJ5FZoo_CP%&-DEzJ4aZDHs%EhaOR(^WuK@?$LPh-k+
zM9d~;<q+0JI981zJaHV_r5|Xlu<girHjJ`2vPcXI3q$u_g$EmCnN(jD@-u%z$U>1g
zdrI!63>J5M*m3a+MvZ6Cj$vI5@TO1?(6Jc8>T^L0Aa0CBLRuPq2&vqdNuN5q8OQk3
zr%!ZUS-M~_+(Kkj?%{V>Ry53R)ER_d$KsZjZ^Nt6kkbI}Dw{ilfHHWV9*2nibXtiX
z${IwMW7%sdl#~MQe56PJy!chnp^h+lm#aC{sPJK5`D4F0D!y(?)%Ap(HniI3yZ`R}
z`~C0+b$;p&`V+AN8R!?U&1`dsLyi2ham1qq>L+j*hxKq_$9td(S&yrp0R6YAMcl$<
z{^6c8TWM(xCY5c`mUMyaBUE{P`m<OtgAb$G;<(VxnY}A!|6vY}OJH-H%Lr^e@9g4I
zgUsB>AmUNP<^z&?D7|h=a_JU_U_oD7em=~nk1lV8ep_BH64E5_4IAsl4@j(Bv-LfU
zVVeD{C6oi1t=251e3aK(fBbA%g_%%2@y1ld0$${S@ZFiC5usoU8A37WI5&&ZSe4^~
zELezsafX$P*^u|^?Q2SpYw2&b>7fTQQP!8UeC3)iSPB^sHdAEP@}@ydOtr}8ISr#m
zKqRa-vzRH`-zOS%!y4<SbR@xxNQdkbn#dc|6Vdxm5_J!=k8yG_mz2@FTTgYkzVp~s
zC(_yP=(tE%5Ce@c%i-1paHjd9IlezC{_A=_m+-D#Lh(4x9QNuyfArZ~D8f$kMuqWl
z>A<RSVb~2_kv<Tfj0Q|t?gQ^Cov#PoE}1dC%$aCV>;)dsl<$~N6oo-H???O6D~wsK
zPlsBd!XL|5jy@?}41*RRh%QfIOO%8CFj@rtS%@Z7UpdUbD#!xmP!UCtR8?6lCe6l;
z4-?Yaa=zZi<9wWuj!t?gsQV*>cn%ZQ3ws4Y<;Z#qtX}OdlN_&vVBQf3IQ4_#^@490
zMA9a32eA8@-X1HjJaQ@GfkL86@55iu|G067|8e7BBe`}UpK8jsGD)P$JP}@I57S^~
zYMRV9i91<(Gm!0MQ>(D`#=#}ydv;#plBj6me77&u1mni8q#J0D7>MA`GpG9`;n4Un
z5t|q%2r7I*KyjQp{{$8pB`;;977Eah?d<Q5gHFU@%<zLfRYWh791zKszy2Bv)5K>#
zlh$}y#AR#A*bxK#SQ;&o94ABJ%uvHReC*gowDh^SxLm4!CNcpMma-Ao4EWA+(HHmJ
zud{})g^4@|ZN?Y+4nRwd0<E9{KqjKXcPreu@dBNAa?d4cfWyWBP`L%<0MSP@7TN%<
zo7;AI!B#S(cQ0s^?{;QpCJ+KG_8vG&d62$w2qZyYkRgLho}K+I;d1g{1A$;DE0DbE
zx(|#2UQS{vtswx+{(13l#(WMI5A|0!B@k6qPVk`>F#@e*O2K1j<|JLwP;;G{H--Y~
z47h4&QM#Ji4_%mXFka(Gnwe-02XG6hCU}~T5;N-N#W1ST8w96b575GEka?|O9aeFm
zBQyb>0ZJ9Nb5o}7J2^$Mb)n91DKZ)S*;wqHKFsfd0LfME1_;lO!(Sa@Vk+!whCUA1
zxtG!T5Qo0uSXj@XU@BOm_BaHSju>oI1NQUKq6&HpFr&7_C={OWiqYC^ux95SXQ!t0
zK(p29VLEt_^d*4QsvbFr?!+0uY!=Yn1a2;m>BdS8GqbbgW{(Rnbz`=a0Z5Q=-_!Ft
z`D6tV&B?=|S0ti4B(dNnL?h&I^-E6I=;5m#@DgZopaFSAf}m8~8I1KKn+FL*UOw*3
zcrxilS8s0&9#l{TrtKvNEzZ>rOa*Z2a?JBzeNAy4=0(Rz2%ptZu=XNS_y~aShOi>Q
z^Zm{z+N0rv3-ix|Xz9Mkj-U8d)?TTM{1HfQ@Pb{|L(jytDu4^h*AMs<KWOmiTh2iN
zLlL{vEY}13R>_D9bm#@J8<Y<V2()xg{ct_5yy*F*p{F-CGlH76f{?Ah;rPP{81F#i
z(T5SJn@U4ms5k`^o4+Z)x*1cD1Aa<45`YM_q`Qv6?$tv*1I2`f0f+8)04M8kPQA6z
z#R}DI9K(AhBZ?SQPwubr0|KQQL_Ef^Z<XWmH1q*SZPP+)w<4N<kMUycT^!~V=-33$
z&&~Px`|Gh^hk_BJ>TsOYYde46uz?E7F>4~mZl=e`lrn%LO+`dp?vW3FL>zrdGG^ww
zRz1hU#)8X!oN5O3)6{j2(EE`CCyiB&Hy$>a&%E2M@yLhV>F%I_z!`E7ctml;(-6eZ
z9sPykWc2wvmi+yQIGt~9VO>--Q&aJc1zT*}t{4su4lX>jRCl>Y<Dfr7o#Q%GJ$xE@
z(U4gWm8TSWEes#QF}jvft&Ov>Ehxri8>@ktE!-Mww}8!Pgp|8n&EilLAj~Tvx*mue
zaOz=mj;O1vlNgeA^4quL*~fS97KW8c?ln;4tq11k+JRC=3au7EX`P@LMCnH9BPGFj
z_Wnk2o5Yr(g;Nfq9#S6I^&je9OjtqBbIrPS^1{MSb{Hzq(XmfakviLP{rdH7RKFvN
zm5FwZ@-P&xiWkU8(gIOt39<trhx81IY*VK-G7|kCPDq%E>t_lIM{+}iZ1lyaDoLLa
z^`-LUL3AfXz_y`17<&4UED}c}*s+sQFo=!UBO&e>kf@%2-Es~X**5B`6)XIkNms0E
zKd>pzAljIea4am@>vhjWeuuQ3JVH!fm+i>94pzuPKk3*>_!e$oHuYAcw>1Jd7pXkt
z@3Vtq*1mlYKAo0nCA4B$8@6rScr`ts^zGX-Scx3S13X7rAr@nzM@L4I;~G}Mfa};M
z2*e9eJ32ZND!o|pbs;w*j8$5n|LR9{%Aoo`ZnSRqehFgiCurljN6^{*F%nA710^yH
z=SbXV+T28AQLb8FqAUe(na&eRk83BX8<LJky=z2yeYd+?J?ff`7x&tW@Kx5s{PL^6
z0(hYwS`TRtmdcW$vCdF*V@MHDPQkvrv%W(i$J*4j1MqAuf*UiReb@%b)Goo~dbEGv
zlAlISasEW-+3iiLGjwdqY0;8Chcwb9_8%m>!<3(ouPKtJp<Ag;s|2#m`lkHc-1S5L
zNZz&u>B8_DwOU;vtBvqp_VQ)_GaZtTvTcVJl}>{`4EV#h-zH^DCHWhjSIQ)RmM)GV
zEtu#T)FAKeo<TmT)UaHezdiM90gqX}!{0BC{P(pMQTJBpfw1;&MZ*`C+@ZjSgW4|#
z^&l!YmvY?B$Y^3Z2i(-MC`~~cXS{_yq3a4<SR&V@E*irs8nNHju1&LbPw#o)liTCG
ze**M=McSCrnzwTwewe_Kf`2dq@`(XPXxQotMsI}yu0BiQE!-)K$L#Fv$E^o<?|$7*
z@*?BJ-eCy4v3lrUyY_*849Et@+B10gOhv!bFz<&`jmt9T#=(Z^tL)1`5e_?A0Y3VX
z#*WEM6snXDNc!9mXKkd}zFl`j(0r?+;%t6W!{egEC#$x=(UHzog`SpNTSfA8!AY?I
znSS=8DqLl=c~xTwwK_hxlK#L+&+ipY3WNH8@3d*STMGcj@`);q1Hqt@#GoqDgoB@0
zw{h1lL(`l~&n1r+U_`N|$7p9dTnqEjqZgpo!&HPG5*m!eL`uoCUmZ<SEFKsgCx!}4
z24JhlG0uIV@P)nTNVLM$BZp5!SOmeF;!bNjyBth(f7ro~zbZ9dD+cJ_8A*7BM_ZP;
z7)r7YaOuV=8YR$SkBN!NQhFv8pSu8=&}d*`-8|ATTOL|u`6-nDxhGxGBVqu5f<z9X
z!I}*l0KBVYjBN!1;Oy+IlmRv2ckQnLvKJ$>1;sI)CkY72wj$Cukz34fiLo6xo396~
z5p~lwoTu5eNyC!m_;CeTKUrLLBQFv~3vqg03yeI$T32;m#&?l0`)97~N<mO}g^m$A
z+jz#hmU9}nnppyl8wY~a=qFbH{d!+3-Jm!B{!j8>;cYOk95y0Bt9%d1oJI~PCocY=
zW#pXCp{MXLgip=In}=Zkc|?T2>LSSg#OiM|m&5Ua>qz0zMb)>#wL%Mh@s2DWR{af9
z*nk<lp@^!{u$yAEVa=LH_;;fP8Y&%Gk6%KZ9t#I}oH}BK&~{r99G;QK1a?3W{SldG
ze?rID*hAYTD^9J;kHlhhAO@!wj9|I0oy^?Fn29WgnU}Y5z#6dy`9-8R%`MZR8sBf|
zn&^L-2UlpZ*rOt<&;#uiJ<RWdwsVk*YQp07n;0Si8h!#i`WYZWexM4JkSP)k$e39Q
zXArbcVCx3ZK?|f7z1X;JmE2qM1pH7oSDQrusJ-J}qbW|#JaVGGn&+;Kv(k?+<^{dO
zS1|oVoU|`IHZ@e7fEP;VA@kFG56c#|)15o7>NRi*a#P2PV5NNvMj$rPI6gO*jf!p`
zd`}IUzy{QUg3LxgOhb7b0bFCH@t~>eg9k6B^B|Fw{$Vv0H3uVA`EracL~=rfD`k8o
z%JcpHJB9KbP}TqYb+raSMLkm9bfd<a_8i0?hG-F*L5mF;J382!dxc0&n3<(ufP)-~
zB9qiqk><!E;4`iivK+Xau3uX?U=6#1e=xDCzLLXu7H_~|UBkmk<ecKqnJ>S1QZR);
z>Ef+huPP=wePPe_;rshQ8*k)=$yI)F^^aD^Mf-Wg&_t>UEySH29pNNUL?R5Ad93lQ
zg*5gsNV?kJBu`^;3_#y-K&r!{#g;q%2rUWcX03*Zo+6eKlhPnCZGgv*TuUeN*Hq}%
z%;SMfItt>=6_V5vkm&}w)k7N<yo~dR-Q<mo;tO*A_~Vbh_fPMrInG<u$H<C38V4`D
ztf8fJ3tA`r3P#YxC5FvkUUL&k-V?}ML<4AQthC2?jFudmS|e}RAm217Te7O>sUYtQ
zFc6oJD5-gkX%9QnjXySA(|UqE<S;iPhSA!pXqre<z5<kvd>l-h46K#q(9ZAAC3t1U
z4-jbSYwxvhFT%#AK`{j<s=jXy@Ey$l6=Y<V;tR;#4DmP&fB-b|LcV}3YeC^Y*~&1X
zBxuv{MIGo4M8+YEb3qbXmdZH5XV4@G9^{Ae2**b!j1Y`OvZ4=dlbONHwws721kkjF
z+2$ejzn;882Q&`naCCf}E65R{G%epvy~oK=>%C?Z>b`^hG#Up_>>37^t^hMJ4#v42
zST3(Hy#nI434kkUj;dTn@*#tJMc?aAS4A${9sUM4Y%d~&vVaPu{aD2bk~iXP$RUxR
z=&col{i-lsK-hLUN3)LLgNkZ42g<&oQip3d3&_BEl*Zg01r$Wj7miL&QjwY8{PcPm
z%#9>EA>>t5f+6@#;8Ls<NwQE3hl1T@4??3LugcG#1!sRc2E~e)r9dud0$3rklf6}A
zlYRB)kS$)q|LC@jVaFKMM!F+1pxahk4K=O=B<X!f<{kzGT}kW;;5CnI)M1vuBFTXm
zf}yuU-W`tvpT@(|d!T&x3^W66!}&IXts<O%kKS<;Z$;im3sxdY$Z4XHL#;0fP)BML
zg@&y|92_+R!HAnN-d3Nf$ZYhbL=CYz$7LJi(A^XZ+_qrS|LfNq0HaQ_fUQAZ!G4Ag
zKjO5dudQEVkmVOdpah-@>dC^ecwmA@LU>I7+A~7L))Q4v!VZCXVO9&%0`;U!72_G@
zY|8{BLB0v;$m7aK>T~JrEneVp^uo&5LDq?mc2Z*;DLcyD6+;&HG@_A8@Rnl8pdCbx
zPl~ZK-0W{)1?nS-HV>=IKVXa3bpKTqPym1(gGwj?`qncF4m|Wn(|9&cvtryBpt&eo
zZUa3S6yD11n>IZ~aK<ZUI$DF&Rvs!z^w2f^$c~+5oIKK%PF1VLgBf(mf>1m+CXgZ^
zDYaC6l!ozCNQ-&I*cXQ4lJcPOGeNpz@4w?*dj^#tQ*jotrY!R)0O+y6HiD=?O2l9V
zDn<-@0YuYsTod=uoKwAdGYvgAB)QbSc8zDh4s3)hSHt%Z>$Tv2#V1A?FqQ&i<(%Y}
z0A0l*FVAj(Uy?`HrwqbXa_Poxc`cC^fFf_fZoy|ig}P;B;;+uw%>7_b$UuV@u$z`*
z?uI)skTEcxWXB!9-mGUHUIIY`Zm1#=7z37o<0RYR!xG3|i8F{73`IgY11}>Fl2}0@
zI@=h#ku->)B3@#u)o_QvXF!PwbW8xDmVU_RCL%8Ahu+IYOgf!FqQ(<{Tnq||&TxNm
zc+AQn*>LFf30D+>3tEZ~fte7Eiz27`OuNJ-W#vF%EwSiJmBjPO!fqN=4)gMQ3VhiB
zTNeFdUI>2%JiafMQ&E_!#tu=3?ccwod?0KyVnrPM$ajPvMMXv7$Zf@v5MYUg<veu0
zeDF=oj`9U`1oKVK4K`#QCTq}})dv7|#Owg&SO5-)>Yt(_vvS3XL%h67IK}UAXX&<+
z*HM`?d4=&o8D=N}3M2mGI@y|Ku8Xx-*D1Pl`*s>;0i|{j3>Jv^VB!&^5`dubuIX|}
zG6<e^@#*PT_OlrnOhW=GJU&&-O!CQf+<Xyu71`r3n(}&j(a?_vaK$YmCT55%c4oRK
z%=htQOGgJpXLad@im4+c-5v8SU$c!R8szd*+y-?Q;r`%7c0<w4+Z;$@9LQx&IqFDz
z6iB@UX7KKxe;xu}hVdw`U_waEX#4gkJd8fzm3)vZgp9LBtvzyva9l+IeP$OlSOznR
zCr|BY2-Tg0fR!xC^kU^=AeFouvDp{sy7M8y2BH`c<Mke)B_L+k#>?dJRfJ_@(oh{d
zA}QM-{=JC2c+J*BySH!mV=-UE+-!@X=ElgElbVWg3Q`0}L#Bz*ebKP*=&8dPNeM)*
zE*>5Y*kw0?+6}g3D?*6#co8wO5ACs9DZQ=;?yYxJ|LEs%5?lY1lM`Neiu%>pL*^C3
zKjCjnhK&4YH{1LFZUg!Llh=p;{q{d=^y44@uF3yg5#-$e?vwwuDF3d>|5}p&t9|ma
z$Lfcx6TaR{O60yr{Jz(4u@&9WVy}-;Lpnup7I6svIyFRaB`-;(A`l#5;uM|U{$nMJ
z6yWcH{kanOb?K#paiG-(ob`aHM-zABTg_IO8$lv0&jZxCqu+n&zyCYB{3lcUpC1MP
z`|bZM#*csfyC(l@Nhm+9$-@l*B@TdSkoEHAA%1@T{IXL=*Zlr__TuF$*+}$03{3&s
zsoZ>U|NgV!2xYxEK{3+=punU)i-fm)xMm;&Nv!uQI08O%N7q^J7p<fj{hJFA%x57<
z%1vr#OmZg_9Q+=$(s@#NYmw!$B<%tG53$seL>_#ZF`bI(X|9%-7<A5d#qK22(2^`u
zmS6e#7^2M2P$>L;&12dB+{HryoHJy0K`BO#P`HqPBK(J4^y(Wqm}_L*nxh23rUp{H
z7Km9uIzEiaHJB8#X7%ce2<&1NlM)fTeL-&fI5X2KV@wrOb<m&D;J7%ePA}x30>&s|
z{~%87f}|_vU){ZTFBUN@DRB~R31F0_t=Ry-X-`}a#$@XN{aI9i>rvCAs}J$RydT4q
z&-<`czx=wk_`b`h^Q4A{aIP1%DJmpEZesLtK~Yf|LNyRz)~sDy2LTR%49qbQSs_6-
z;In~B^T{Dq5^iDEKsEUzz@(~qc1no8B&DQ$!J2}Wusq79Fc?JTR~8f+V-NL26<!3b
z^62v~O9OMGmlPNgD{5L*OyBL;#B--vkYV~uxM$jp&!?Y9MYXj(4i1hd%m7qwvy#u<
zcme>_0Pi+~wYiBA7plm05CIZW7=A<k>FJdKlr8~Yk%RQ+XtoN)bEiiErHBZHA!AEN
zNraGmQ0je;pc`bT-4bNY37?FCArV_F2H6HPK>LajM!%(Jy6t<3mhx#6r_JcKGmq*&
zqJA-&Ef)?xQGfv=;u{Wrf8KwGI=c0u4N*erwBjumK@}{sSK!>EO>}fd=FI@wNi{`D
zb1QZ#W)Ww+{07m;^|l)ZVWb;(5J<+x@vEM=loU6fgaO1wY6v$T43b+r5c9GV^jB~L
za0)Cs<bnvhy%0J?T4Vh^tqI5$3Q_BHM@E@#rj}(^@P(Qb_IYUNx>Y4AG&`X#7A0ka
zmVHBxLZEmClk-061Qal1WV#TnXEE63>Av6qVq;3oJunXf-QNt?7gYgmcLO}V6Q?D4
zc4Ft&t=HTbg<1hCB?lfv0#3xvN16<ib|Oo#?1@#+&m#jQgC_e?OS%^qXZV;0*jaH&
zi*FPU!zl-Gaq(jie>a49*83?(G{f#BNfFT0bvRq8Rd)cOeh4|3<S9lFDeq*QE`4)1
z)Y%*5dw_Ke<f)sT0jY<zVH(h9kCiji;D_Y5(TYi-6_c5ur;#~C`!E3{|6GyQgR%3K
z&RXo9_j7noTpP+cbHSura68bYTK|()iPfm|yu~ZgA7@<@xsN+eV$kHU%*hisk!GdN
zs-t>5cXR`4jplzBa`<mUA&d>V$ZRH(*e)vx;m|-%T@Pq22WiiurvzxBcgGY;`Jrpy
z7Ms%XtF_vVNy*1t5`pSK3MA$PfKf@K-JOa`2Yp}^!1P<th9KZmHWk+;3%-;b(r-+s
zH&QkSx=xM?X&kyKvib@^^N_K<2V9;|X%Bsj^J-qCu}bp%`LvT(YG7{GA_-41nj9OG
zMPrK@og+#c%IGlj==e0kfmt*JBh#%6PPQF{X&gssJr1=C!VE0P0<{MDR05^}KAUah
zyY7kEM^a$X)}bh@T{{Ou5(8_Xl!1I*4dx{Eq@%+p>lM>`0E&_v5}z{xShF8&7Ai!h
z(NWU*p5MlvpW|BbM(>Y8Ghg;)kL|M}MHe7xayYEy*kw0V1oILbPzx($mV9~^+5XX5
z-p2rmd#ogE5`oK1P6}c6M1I*aqrZ}OjGj>o*;RTDSVbaQS7LAreZXie6LOK2LZ0#o
ztWo0}?~shpi4~OB9k>6(E*gCUl`!ocqLEhsoJbkg*3@u;^3U>7fS-RHU2{AUc?2zh
zV!?Xnm7Vv|<Pd|Mjx+tZzy^i{=xqXGBcE_P*fD}c9#wgH)ie|oOmK<Av}f@k0r;fR
z)j!-$cpheqsQDy1IZ^I+VwB0Bu&n(LFTWOmaleG{$xJ{R$TQ(?;K?^~IfBZ-l<7&w
zTjQnOwZ=n?(r+Zd;1vNp4ONxhW#7Qd`_TM?6O~MsFm`A~U$bgC#s+YKY$npyrEiNu
zGB;3NVs!!Z!pzpg&`dxE!T_RmDDA%zI)RKB$IEWN$2^5UhiBU_3`YsV2v=l*)_^#3
z#)&1XcJ%KmA&NwsQZYu`+ntduZ|WJ)6S~Je&|4u;%iH#4HaKirOi9nipMdZhG{-bN
zfYbaQRFL?}ekk;#e3W9e2ttidU|`=vWwSG&7GMIrrdd1?@%3e6Agt*DTUzhBs(FdH
z0x=_-M}*Pw%B;(!-Uc@;2Lw6QXpvZRusfO$RcP=qEHlz78z~m(b_lx=$W`KXV9zS}
ziEzLxZHtX9EvlfB5GOSb-H2F#Rm55o&~@9kNR0XDg&}?qP-7ey)vE3XYzWYe&`rhY
z<zK&iy1dqUYCjVb43@)j)#^bRfu+6k=YRGN13Q70O0d3GR(heR;A&47TI?3Wz-<Gx
zSTyC@W<tq}V0R(B=A!C$BX{eCHqSzm$Tv9H$kGnbgig`i+}t==U<57!W3?K?Zzh2J
z#DQ_9LnVg2Z^4H}=@KBc(i8zECQ+ct8LtHM<H2(RsEqGnH1g-qpVwD#W2lF#=dNH`
zK<9N&&#Xd??Hr-FU{iy;8ID??0nQ;Qu7NwqQi$%%n2Kpkat3yGf>kNyEZ*%BinxHt
zNPz^-uCl!eHKz)H_P3Gzcn+Ld^BRkyW*%anjLgQYJ44|sVxTjHLyt>=*(kvPZZ~XO
z3^fp$bhHFJ7&V#+*t8PZ(Do*9oWFo{S!R}w{=%6rAdqEUVsk3mRix=;kJeRvFqJ=Y
zWkR8jkdV^B&{fT`mO;4^q|<6ttw`z6bPxwR9bs9J8NW-<JzzA!|3Ot=@9QMN2&g5g
zOBN3NSh`RXJOmJ*Z&<7F2O<OkZR7w7915h?geJ6cj5O{=m2q@p!ooTa_oI&mBF{|J
zZO2oLO##jKEMhEX)mLfcNP<fjloMMf01Sx)6XhN-qD=!Ipb;`Ylsz1KK7a$nWCiEV
z2kJ0)#nrLS476JS$Vx4lL;#N|6F%u<U(=jsVg}Ijll)X5r4<XNyMv`swvW6J5)1EQ
z4`biglAQ^rSmdWqSF^ISm(V^iLBU^&M37PpHn&kcDt&Ja@<<5`^?brnfYV7LAHoc(
zIzT>B(gF+*j8{95$MjhN&jz)ucbm<JSG74V=7Qg9?BYcwjur(8MbyYpLpq54B}pj7
zXbwsQ1ExzdFNcRAD+;vpTM$P~F~ZnY2-T07r7;!<05ir=oc*;Ku!Bm*iSyY>c{7r*
zw|R?Wa9`Xvzg)auxB9is4VHUzKHGcl-I^R?eS2)+6>#^Dns;WwhluY6hwL7KpHV0=
zgf2U<*ccGlI>D|&0vQX4o50@agg)SEYBr<vRYUN<58kIuBz_>YJ!zr&wB^f}%R$dY
z00OaintmWi26jx{)A#~<v;z2!aIXiZr>6<2kB}Mh`8Sb<t{Nx^XGQHDS3pH)R0Nz1
zGmurddJ-Dq7wdCvtq5g}@aZf)FkC4?DOEHnXVj(CkBtoz6G03;jI&259;lt9&{#n$
z6MAQ!z%w~%*wj>7TU#-FNftxj>QMeCyn<JlR+Z78cMujDQ=1e~gd&7QZ2TjLtl>ZF
z(W6I*g>(gKs564a6Azzt7*qeS32(JKA0*8aV5W&<@<Y{#ApZco+Ep`KJT;bu5o=4}
z=|mO8+*-vv3(;E~VnIv=G(geT7aT)She(hg3yVj~bK?nm&4LXT<9WgOT#7(|m7@EA
zTrc(vu?Rs6lm;KrfXGeb7G+@oP#QZtJv&SK2|$m5gz0+-$V-z-7ApT1Um18Cz|AH>
zJ*kcc9@Dr@+%T!Ef;F?g0>cDiK%ygzdO|oY$ie`Z8Bp3;E0gQ^HHHK}VoVTO-I8R4
zcL0>%5BYB(zIDlCw4oH8(bjKX3$sMZz1~_43l9!u6Xp>z9RWqn#TF1HB4OlX(R)Wu
z0aA&s&!3;77(-b01}(4B$>?G@?=Hy&#8>?gSXm@b$``g3%?zly`D5=y5sC!}28}N8
zsY-;KNb0O8LB^xROZXF{+k&Jv+tux45a87#7@^M*^OyjXPP`NrFqQ8Fq8<{ko@4;U
z#In4oNDSSle?U`d0~Fdmbk=JykZ$2<l!$eTjP;uzMW7}aOZWj9Q4tZBb8e|Qv|I0=
z^bi9r3ZWcfpT=VP6!XO=uji|@p;ucCmlKfi(zAA?804FDt6{Yzt`;Igm~hl4G-_@C
z6c`<~?mR`}fRaIw{CzZb94~O{AT9wFm`O(mXu(aO?C_Dk2z+=FK#n63=#q3@x8NDu
z%rEryB@gPHkY2K1^qjB<>kJ2z6uFt1C6Vzy24^ZboVpP15ZVjGh!CYve;Qi6a8xNJ
zLd|YOgoqcQfxXOf0DRALNNH<H#{*7SX<M5*2;ZcrK*(@VZI?&$32*ED=C3@At{BFn
zk9ZFRr5AWVxs+=XKUz)J{Pvp+IguFAg_y8*PZhf?;&mm&dnr^>pj1T?(|MU#URbhJ
zLrjn&jU<E<K==@d`5Uuvu1O$XQ01DzFezZL<UIz^nMhu*_vyd}-P<3Ri9^_B6(fph
zjHCW0!fV$tB!2$-4k*)lp$(cS+A2(DB2*DYm{uQnzUw8z5>UzoTt`}3THMhQNsXk#
zF3-&$LUVTV^a9~p*3Xtvsnp1##Wx?pD>O=c6AaC_!qux+zc{RpUJNp6Wgl`Rqa181
z_A>jM=n~quckf@CMkXdEsEqAAULPD7XiyIC5iJ+<NXo+2)^-=!QDp2+T(VSsK3Xd~
zP#H_R)kZR=$Qed1t-`g)*5CCBzRb$niJ~G0;UidiG+#o&$U7Zmp<`D@%3F%s?8e+H
z*iqRuq8diS)Qh5$SaKP0_VMiNco4_yw`|#>l>^1!dl*$y!HtfiT}VK<2<xo$b61x(
z#|4~+wQrBPfkYaPE*pih41^8c!LSG1z%g`u4$1u0Jx?n5BcQW4Ze7&S8k%pcs9>d<
zgT1u_GzOhvDsuH8Rl^~H8Z>!}OG<KNCn7;*qQYV~jtd&>xHm$YzTMEzygz&RPV;e0
zM)FRF`D#w&aTGe9V!4as+Jh3FgVLc~*@dB6o^YzW5Dr<^*CBf&nncG7FsCG^k!Mn^
zA0&d5nORQQmMKtZwZ<IKM^lGs(Yc?yyLD;>;ca&j2J2DvJR@_D`$VU#V#pANk3Nw2
z*=@rgJ3FrhwL@vBxN`>OH{LWgvCr0>jwftA#E>v(9KV2qI#~@E_g&aY=`*+Jfu>YU
zrgGAIqPwDKK#&`EK^^Pq#M#8Yzekul|Bt=*j>_`hwnoX;XtE<38wyhFy#vw&G%8pC
z3r)I!NRuuv2m)rCV4<p@bfWa$rGwcB2ntB=AimO@bm`w*sM$H+H|}5e-f_nrV`q$W
zvLpDG=lMOW%r)m+>(R%gWKz!cj?qW9Vidk%#biDCv)?V4365l*3_{i7ZBSg=*|3ys
z?PRl1pZfYwPqgyGOZiB98pKthmRK#M^F(Nc{fpGG8n8-|4w?7Pyu8z?-D1<b&$nj1
zqvkT)^Ge(CRwbL(_#2s+j+XFyJ;Je~Z_}Zjf)K=ESQDMFe&nKf+35SuB3rj1uz;iI
z)MK%<w0Y141M9<>q#9sjp-S^oEc>B#DvbQHpWK)USfAr_2s2+MT#Vz>x1T?M9+(Sv
z0cM2Pi~mf2*7Q`P(-Cyry*UDS;H%q&cxJ@ib6bhrY~d%_*7Ocx`|W@s63K8T6vBxv
zd%lrWS0tcvPSGp>1vxHJ*b0pS5wx3yR)HZ9j-A_zsb?X#fk7yym$i528V_+9fvg6A
z$RyO{*^7^eo*0qe1FJ#H!WBU-9GVkQ#{=_;9i;$d+sH-+R(`)M0Cfh^M43YKQjOXf
zbM(68jsR4AQOGAWOwPYaT@mU-e!Hm~z8^54rFG1Tc##tOq^>*Y<Dx(`2}i@MkSm&o
z0tV8CbO)6z>jY{3_Aop(UfFIK;dZ%m#>Ohm_ijWDfF)Swr)P2-vu-8<+o0FQ?6WRw
z0Lm~9QJZ9VoO<@};cI&A<rR)^g1e)n(229D3NA#v@Bv|AWe5(+VW}K~0AHv22J=h`
ze^eB~;6LG}e-d_>UjzZJ-VD`Bk~XSGzI4$oTk`tWYGPg4mGQI3iLf*&d^iLy?%Px2
zR8mVr$8Hv>N_UKJ5|L^)p~?u|2-g#40G@B6@t`)TME~BW3|hVfd6)t%RCSH}c^FSO
zA$<xC4h8x71O?FezS;}@kUR>@WGt)qb*l$!li#G}sQf%+pMzcD4;<Z?xuZXCT>Qhc
zpLU)~c(>-~#alOO>0R=CetYAJ;u9;^tmOVV_{YT;F05R8Ve#=0#g#uaUy%EGP50(~
zzx?u#XWM`6o4$M3E~%)c#5l2Nd;MUipXiQ>(kAbEb6wpXvZ$Dm8~Zc1?+}4ZM6(Y5
z5N%aiD5rGRceJ%>PL#4pOk4tQ4kzib+>7X6Y(LHmC5aBB{q(Z>C9xfS25G`jmHQa`
zBugeLUTfI0YwbQ|xfS9w@zwEubrPmaQH7<}ac$nY^R?+xSzBD5#t+uu4|<<Hu%jq)
z?$@k;Up*XwbKnEQO|wx7PCz)+i8LZ_p8Pm&9}0&QDb&pd8Le+1*S!n&ehRoDnH4TT
zWN|Y1kQFi+Yh2yjY9WZheyF$`jdM_Mu+2)1k810jRKM;-EhQS_Ds(Y1*h!EXcTP$Q
zrRInnC>-F(?3dN~!(^uwHj27vHbb}B#@CuzdZjn@kL-T2T~=eP-?6yY9{eW7eElbN
zIH+1)gNg@`_X0pBklTnOGYcSyQZo9dw?mLTSP(tr*-==K|AGIKMQz$s8RoWm%NCLu
z0^^hhsY({l#lB%IjChEu>wR*kSEWBTZxfxIXL@o{AJ{`fj$I~cQ9V3@fjW1i@>Rxf
zt0Rovom1L9JMA`)eykfU4pnAY6A_J`trwNq16&ZaPpAg=I2i(Ai37tovtnXwj0laR
zD8qU|B9fJoawZ}Iln6*><nM{<>9)`u`p7ZZ8P%!Hp|Q<|4NZm_(~-yyRQ7>`K-8i&
zP+dY_cwAFcldX1UB6mK;FtOfYZiEMma}+z6Xk$r=f@(gHULLIw@rz-KguTJlZ-?yX
zCUxOEdd{2sV?gEg`o#R6Nh(1>xRWU&&4$8KlTvt8SKj$ewh<jp031*6E-4OX0N?{q
zwzK(H8wm;}_=b)A$J+t`VGxe(*oIJVgZBh3R<q3W%#_vpI`k|aZN1G8W6DS>v@CDQ
z3T_hoMrmS)f)8CVl&8BkZ8}agW+We~ouY<N%JD;|nE6IEOol6}6LBJf2tbuZ%OwJJ
zgF4+;lKH8R1&}OIn>;Fq-RJyw)*`5s;4|mxHvyo-Jv8l_mDvYf3{{uSceGF=jKLD8
zj`Q9WiHQiukzR6!N=0-D5Ue>k#+Lm^*ToJ)4TU@)FC;Yy>#)K)-Mn~0<wVmfy3|Q+
z)_mtq(og*llMgp?lb8_Q3@Nbd?Mh%XGzt<kyJID{nTzua$xFPEt3kx!+`sy;334V|
zA*lm%a~(n}qBU}JfN#7l#Bx*xY&Q<Ac>7~wqQl0x%V?<(Lbb)HNTut8;NFXppPon}
z&Je#3SC~3cOp7>+x1>(+lFk+ngZozb=*t%o%&H+txiLL%M3P<V;BX2{;E=_n(}+n%
z{k05eRH^{92O%Hl---n;N{>I4{`6oVhgKlh(cl0F+m~4a$qbyS5~qc(hGf;SWulR%
z%N}ux3t>JQ(+5LzUhEkGInSC796j1}5Z}$7&GJ!F-PFD=XNHDM--ds%Y%DYDFSz`r
zj=jx9z%qE`F`x~9{Mh}G6n@xrgl}}Ymwd(+$}=NJk>;-v?T0naTfijX)C@G4f_`_w
zJJv^NcdR{m@}w6DoLA1jez?VP2s<XUT~%U~`F-2ZbdwTsdhG0M7IvH)TZy^CMLb$)
z1#*2Qgj*Ha+m9TnA)!De$m+|Oz==1POn(By+6z$ivbb*S(mORns;X;3U^i+~THXZ=
zzWFeZ2;;3B#DNwK=nD>Fh9n|4hep8%sanUn1Yu#{2gT^i<)kB8?*OhH+fX1R!j*_T
z+Uv!4?J63@(YvfV0z8JJ;Jlr~evA<g8ej9+uhw9|RSl&7ZY4=lqC=}8HMv_MglnPM
ztc1f~iN+<yQ7K9T__l#p4Zbz&8`HU&o3eP^`XkHwZ_34~H{SqbHV^pRjNg;m?AoH!
zn6y<669@goV9cKOvcT3Ji%-MD#lu;tMvd1#bJ*D0`T`2RIx+<q#20bJD`KJ;rT-|n
zs={&uz_!<i>h@-eXCfF@#QQ;ExIDx&V?cWHa;mXE<fF>^rM5OUK49kSJuJnjkG1kl
zT>lVyMZpxY%b>hzsz>Bb`wbm5U`pUdAs+iZl0?@gg;H{Pc53V+bSMM{>7(imMUkkI
ztRyQAD64IB25jK#UagovaGoV#QK9n36)UW4(s}QU-y6zPu-yt8XZP5jNF+v|o0<|t
z69KN|b-oKe^LTHWcvAmE4=i6;6xF?W$r2w-5V>~O;+8La1GPeTSLNGP5e|hR_I!pl
zZ$sK)?p(XKcm!~m@U~`fHI;*>At_dN5kPU{WQ`!k2;qT7&8!(*ZdT?Ta85QeN-3Fv
zl8n(6YoKP}v;wNjcp8JF-qA~fD_-Mbv8M>!jsACoV9TIY8isoK$h6Rakom`Z5g>dX
zaTw_sx8H7<yCV>!-AY{gn1Woe;iwE@8=}J4SBGGsQ!@}Uc_ZO3V`_kx@L>Y`Z&OR=
z3agABAba|@7lmh4v7M8VIf7%;s&DiJgfU;s6?P|-%GTPQ032}WRzJ{E>2(LxZ`cH2
ziDhcN8x=O0mBp-fCqPZJbsPZi!CaW}Rcys#evtGe;0zPqq3H)arYj;W;4ly2r65->
zZy!Ns4O3zl3_*FP61~ya%d6|%gT(ZD+`AHo+9*W=w{B$AVg`5_wU{D;L4WBEz(I(y
zG`nPFl1(*Rl-_=o7jLimV90RM+ByxRzsizaBphlzZ>BJz+H6!+<LAediV9E)jJbo(
zyw~dv5>4*bEhy{qI!)0?LeMbJ%Y#SlA}OFQuNCA+_gR9TQ;WY0vaHUC<_OUC1fn{*
zl=`c<ktbh-D-I~(P-0Nm>v0s{K-Z$W3w4trU`|S_gd?L$jMB;I#CU}*(@TipHH5pv
zag1yl{!R9yCF&Vi31|NJbnRm(@smM<r|~Qm;soc=G_!yQrEqi9oy<Lti-O_wwZRda
zqlwd(<Sl{65v}YWh&aMoJ(*jH)=K3)RE;0sS%OIZhh`nBRnGmA3Wp+Xw|3Tu0|Zb|
zv^@{KJwQOeM}<IdFq`4a`vkdq7i1f?Xj#b50;y9UBU21|4`y8hk(@EOUq>26Su(~_
zG+8$}zv<aGE&#?VAj4!JNGxN<DW~i%5yA(>va>GKp9*oiRMX<o35QZL#>Bq7yqqKi
zV3lg~&jE}z8nHAF0ax=iAo7_{Lqk-qBLxJb=}>C+1Z<!PF#=6b{FL#?y;vX1HwYvK
z`Y2(KfQ$a_k9!u~A3v?|LK49&jJVGTV|497oYc1tJX_O?AK%*}7$o|r;=~Hkt7Axv
zSqQA{dPDAKA{rqja>$jfke;0ug2rrDcZN^W^wMD;k69_g;Y#dXZHbbJDc|Kj&cJ}C
zmNG&sA)Kf(ov;FffIuF$30hxoiMSaxguD<7_}|-21gcVkDZvDA)WKTZa(p^4Fn4yh
zGW2BhwseNPsKYyxy#~7@kZdL~xxdz${jIs+B99$*M#mCIrj~i_&rrjXy@cjv9IUmg
zR(%*fii<@w)+kM26tWmp^~>x@9i@79-J$=~vxq*Sbmcc3XlL<2xi^cn$LOS;a1jpk
zbKM2KfZ6lK+d+&7$4@gyiegCJI49E1vRW6I!s(1fdQg&(=*!z}Vz70v>M-EK$`2y|
zy(Kc^?BKaI1|oCV1`~9g!=p3vvEAtZF1w^6B*4I;;nq>J4<|?7uT=pYPnF%s5O9t$
z^cTUlYQ_(T`Og&9Y;1H+Jo4=LBeLp~3gZIQ^10<)1tfQ4ps<Id{3W6RQeHKX3}}Ap
zqOly=&+FN+_05}<uE9$Vho#9L+zL^;in;dLXP@lVw{AvV_Nb!?)}5?O05NK4953QY
zAwaaVGYotEq(G&pL`4;?PA(@XX@uv3n_eT;(r6`06Q~TsVgEqxygkjLaq1r-PO8<y
z7e)_zJeAW-pK;sv9m;%j%MFs(wqvgV@x5sqO?+@s315|zs9_x>SfQs;zZ6ZS4ARk?
z8&im20@ribpob#e4Q2vllz{B0h=}wK<OY<_yPXE(Ja9fk@bDiw(S8#kC5@tR`A(w`
zRYb_6v8S)){gez{z%q{g|H2f>5bXvzjL3n>G5PuPPQ*u?tZF=ctV^HJ@aj32#|FUG
zhIo$2GJqT*!a}(KrNgU&L#AN>))Z}5f7k+n*b5u69o0)$^*@X@aD!oy1k^h=xX<#z
zr&oft9gcj0Y5DJM`~WvnYyvB|i|k7c+?urIcuot&$(Sxh8fGg5AZC6x0}>gw!WY}E
zI<}V%@Vqwc(oHNQ!x6xL4bDsqXJCfu;|2|vp<u-1ofTpiAY0m61fUofDK(0d{DG+B
z7#YHIGB|w7cKED{)Rfh)2ezQhtpvqk4Mvx!_3^EufUi$sH_qOvxl1azHGoK%rb=BK
ziv8(20t{1O9T^_3Knv*?(1}2KJ*N=w@f2`gb9eiboJ}k&!<)B%gIk2^FbmaF%%a52
zDKdQi1&WzEe!V8(-ZcN?(a2<!bl=%){v6rGdmbYZKGn7(Nr#8k0V2~JO=h2DakWAW
z>ym@R7;!ePjY9)O@~25_kS7$Z%3*uK^VVT02<y0%jHe;ci?NB=U__T^kOh^}>#K*o
zdB$^=#+u{dMkT1iYB7FwNq6h<JxFb5EbO6h#UZx~t!POz3&IpD#*tKh2$1m{9blIO
ze_$Wryn&2H%8|C=(VprzJ+T%9HzT(!i_0`^`&a%D#^0=izQD4(p`YQGGQ_sz-H6fE
z@o2R+sCN#)b4XvX4)OQ|o)TG9R~6DUuK}dETO%h6Xsq_c{uA3DpW(gs_5uyp@o3+*
za`bQ>Y(;+{<fou;Xlei%&8uy3&|5FKNL5cau`e2jmz3f9Bj~*HxYnSEd5#OaUfuVh
zO?hLZ3btN&S2Mg~27{|C-L_)hQgCq2b}%~(Hl42wE<^A>1{I4ktCO3XRq^5@ibV9M
zxd&nASS9RMS&<ER%IAntBx;aG18cDIuP<N5|7b(cqJ}L>wcOrN)K~wSfye6-vSy6>
zW6+~8Lss@fR+#mx=0+fz^?2N~*y^E8=a0J5x6k;yUYzx56wE}=bf&%x;)4fL{iX)z
zj0bcvCnf8mD~e4BPD)MZfU~7?9MBP$MONS2a(DE%-+#xkz`PbsPIQ42+=9AviyyqW
zsBYi{Cye?&l{rkHJ`N6=@L#9Pj_X++4@i|{hs&SdMVF%wNrq!97eM=3IHtX?ngOan
zy;a^^AQCybODqrb-C+OIKmKQ^?^7yLyyV06JtzC;3pgN|OhBMzAFMHtV`%uQ0CQ@Z
znZWH`%fS(eCql$B{v1_LOId=ScuzExk7C7{Nv1;M*r!kL$UDl4dzZK*^>GN;m9NeS
zeMg)DeDxF&JA_)NU&Gni-}<Z$;ePq_(y?R5o_rKSpog*`L`WP})yRWVcFkmi+|j{f
z;U16ME=6He2zD|5nD+@Kldx(4<YD;--|A`o?lC^sn5g5_CR&L_=6T1DRpbUXVp^xf
zI*p>(-#)JMdqDr?sMwbw+vQcml;TF^nX_jvLE<}#)s{4OVc-3R#;uSxiozX|`sWk1
z9L>2x`jQ)irglX21fY_*%r!%jBV=v9yl!>)gP~jIU6#tPv^)EgbCoE09iT1O<&Os2
z%(2z3re~4Lm;rvr=*bq#>C%=P7Ea)rbwMguh5z$=cHA?9fE-|qsSWlIgVDU}<Nh4!
zy|pDhUCdq2E%>9JKlYM`1HagSUcD%t4p_-w2dttJQ)0hI#yt}i0nJ_(8PQz`j+e%H
zX;OporV0_+&qvaDP}c^_5&Cm@P;ZkSX(S;fr<km&fNyp%!9H&E92bqqy{lGV_Hn;a
z0?O-1Tsp_LmX;Ra^p7<K?M@7|p%n5#kE{6^JVt+?R`7LbB9J#7k5feZDIF_HS#!Vt
zj^(wJ*Ehk&$17A<?osb?c!7BLX5c6$7uT(tAaA?ODTxDYXJlNBL8KF6yDJy%4PCQp
z)pNwzZqQ!{ZsbkgP$W8@Q+BDKr5EQKZzPRz9U16B4K9RvaS<MO)Szr`@CqpTAZ$CQ
zse*iYcZps7h!FW%Ob!wP*Vx!NgSPUpIG~b!+!FzXVA7P-!mzh~AhG=_uo*{9SE(Zw
zuTatjqeKoC0!`@nJ}jEd>uyk%=O2@?buBQvk*@%<oNqTKXzvd09W!PF4#e*)0-grn
zgh~(XHjGyH1qmf>`o0L+2u+RmM^SlQ;Elm8u%zpXHpHZY(YP@`*=_iM5Jx&q7)FEL
zdIn{bnvYFi0PY(0d>v=B+q#usv6IYsljL)8_OUzy8SZqc!4%-4*Pnj7rZRQ{?pYPH
zVlU(s;Z81;_4pT?MJ_Ik?D!XtU1O#?bEXH9lqQ0k_H9h@xFchc3Cp<@<7Q=yMoLUM
zP+&ySY!+-oWmW{aX>dUvUb~Q(ZEj(=v79?Wz{n6-xz5NKBHD;xrsyM|n4()k5i+Lh
zP!gvImUb<ksk#&<i!2@%+<R83T-?2O8JBRCR;M@6QyRmPOpypfL}LJK^un;>;X{<W
zA5Uw9T%8np;Nl`Z((E-gZ3u5B#u#t2?ed5nq3MZ%_g{F~nRRV2UqN<WZT~s~IR22I
z*OQfKIYSW0w7oP?o2Gy&6}5Ur|HJ<?Y+YylBQgKsck56xB#Ic5E1MbSjLo$j3K&R}
z#S}2%eWjFJfxXbAfd=@yVfhDW(ns7Or}}v5i>sq{PbwVZ1dJMjVbCQ<YzQiP5f^&V
zVnbCHsdgvFD8|~>ZA;*$U9)B7cy#wIx$7mh9c;VW8Bj})h)ZpMoS(;`8^bYR@Iq{d
zt9=J2xgR4^9w?baByZq+r43}+S<_4OP7b<D);!?0T)WRfk4N@w>jz$6*(y20_${XP
zt$$OFoeGX{6#&N=M_W{kxDGbhT-0}#<`sAlk~kFLAb3hbA`<73{8p%DjJ|w`bj2vQ
z97X;~m9?<_Oq$*K)AM3G45OQ^VzuPvuj({gfE&|Gj02w$6kyfh?;Q5Hrcfga_BL*j
zW%yCzp0XR8%vu1nzH4<dNl_-+7B9aqXBvu9GBPB=DsVK@#Gui^_p*k8du*}I@eExv
zLT&i7aEbs++hz<#?fWa7`My*dJ1#NII4KjoTJ_KR{*^&Toj(Q(|EU%nyLt|4v#?-a
z^OZjvPJWc2O3w+%hZyBE{q0?Tu(89WJ{=qPo<V+l8I^}3He9mqP^kMw{Be8d&asbS
zh!UAo-p*-snCj+(OkQV$B%^{gj1M)_c&$Q+`$=GF-tTJg<_bq$c+KD!ZR5}TeqcGK
z{_5&i<H?sLH?6JfxCD5kK6G3|1giv0o&r4~QhssI%ioFk1w60^5FEGl`|rr(=lg9b
zd>D`;Gpgdemzr|m8g>_$edOI5GA~s8cE1LD(Sg*14bM&%H~e~(s!`;j3cq<ALsDQ~
zRm!Q~HnzJdLg&UQCEX<5mrCkZ;D4i0d~3RH7r_R=HOarRSw9xljOlwo!r9AqND=G4
z$8OWCO6Ohx76XUZOaCjuvP@SXDDmpp#((3aDXV>V3p>%2j@;ZSA`~@OR6Lm8bYbH2
z)YNVi^_cM0po}HIVyM!a56+<?7)8%6;!z075P8nz)dnFL(_jJ0)BXz?5$zpQ$DWHm
zVo#PdcZ=3@yR6-7yMIUZj{}>>O4-mgyt&~6{wBerlXb|k5h8kN52yo<M3<&2>B?D-
z0VpUHTv2f<n2Qs|N;lx91X&@2bKUc=N6OCwi*jZXR~aQlA6MBsF$to^vUM5Z*tvrO
z`bqauJXcGWbLv_A>lTP&$yo6tHMs2Q{~iWREuiyF5E(#yvd<McuCFSwF6mz}dx(?{
z!fgeJ)r2}o`w3no_$KKe10WlbJenHuQ`<bj)h9eaIV%+8j!JyuHQ?D`*4c$AA~@`c
zmn|oV6eM^+QtZT~slXt#4Nk}Zxd8KXg>&BAtCaYvt<z&QRHVxOavRS8-FjP*br)8P
zWA-#R2a3Zbb08^dX`}M}T7?FseR~uiS=AnQobUq#4pX0Z=tiO)VFsj!Uc3hc&lBCD
zoG*Qp@nMYARgk;$;z6EmP;&&Uys&5TtxU1$_MpU9uU^&huRk0FI69Oz>6J_Zfg^K^
z<zY`fE!|+T@nFra52oIJn>KFT*!-shyXS^pZa@Nd*fDWrsXxkv#(_}~Tw_rdam0@D
zpdGq*?+481s@f7mpDa|q^KRq!ri&uS@9|5F{pyi~`nH_RikR4%u`5p`p()9*cE8i$
z6VwjE?U`G*DUK#k#hYYqF8&y8EoDQDclUsAiTbAs>urO;rwe8G$c0`P_@D+lanp<X
z7FOc2J+_9RS9<s8P60vd{bXUz)}S@-_QGo#R<1{ybX3htW94wn$bj_PTQAM^otTx<
zK&HLhds4G5y)AVk3(L)wkH4AG3IN_Zl*tTL|33BP{Z$Ig5YExI*FACQEg6ND+c0;h
zZWu^lB{Je#jH(dnqmbr01ksb=DX&EBtahn@hKn?@=P5%AkFoOw)WY(q#%iAXR20xk
zKSf8MZwWYddFL3$UV?mjo3qOv(q&(Fz<7PDKbbW3OcI+_KL@R>f}&<NsB4Z4%{{WA
z&YE@LSXO4wVIEU*p+Tucp#}WfvZ^aFpxcRU2s~4Y9m!<oavrHVv-%Tgi~uio?%n&g
zEtxkdGAX9IqF;If1tQpRiPZ;#rkeo(gkKq(!KCi`@^OYEA=kHdG>ud%Y6$r~iLo7p
z=)HS+U6T9+c(4^AQJxz_Q$gtpVSm^2nh9vJen!e~i^JRu(M&YcMDOrQZ+{F(#fZpO
zrth?X8l`~Q6efL*cS@}6%-F9dZ9j51Y#G?Lr*TV7T}Gg}vj6+#ROlsW>m&WP4O|^x
zAPNnUUShq#*eed(X(iS2!Vi6OzuFGlh|5CyQH7eF3}k_9R3cBNKzbsIwGJJEPY-BN
zkWy7B=<Cxp501N(IKzpsCs-mX3pP{U=k%3E2@d6K7Ni_F9K8cy{11dIo!zrE1O4iN
zV94JDh8?+Ff`G#7StgnuVuC$bS@(**I&IwcZG<;mv)Ps{TlTpIqPd6~0=QG+uwiz~
zI}S<4@ar=e4ZyuKlAybpq;bremS^XbQWzO&1%)KR)^gN2Yw7mKyGnHPo_Lro0(}y>
zg3pYDPw&dyG;SlK%fjO+cK&KBl)>R=>zfe*Nyt98<CX*?gf>32<^hkqX5l#jObQGd
zJNC08QT4^Ps0o(mDj?ohqv{?*%WyUEYPuq3$>g7msC1b%4z(7##}I(DS9jW}i2N8L
zaQ{`!`~9|7I}<Gj6`~YMV)+v9Tx{N=V>c@cmq-`*XqQZ)Re_oau!HBaymkAOF<~JJ
z)xoz27F+p^o9cQdVdSHu7jK1;zAQt*%o9nEqh`Y+@@w?1B^xC2vVhoxm5FVxsyNxw
zyjKKyotKPv1XGUKT+VGKpw!HrP|#&hqhd;wVeN~i#|B$lKL}zTn3u_o#=2N*haqru
z)KFyy?Tf|OistOu5IL@%n!o+gepZU0268_^XhBU=erpbCKi=z|hks&HAQ@^h&KGB*
zd(SuFHMH}@&^FO4JQYOeyKg)mr=3vga}%F?2juJ8dV7qWhzU#=wVdsZ4Q6+TmTiD-
zS16lsI+}e#h!O9i+5kY1qfa$<hAiuRsBGL#1Hl7YU-nvnsNcbcYgNm#5r{O53cTDf
zrVb}?$3+E5?no<weHF5Mg)|0fXE0QF5%g;QY<BT=iKT~x8ewLlWd<Wv{ihyV+GRFL
z+<AB@6E32(cH-Gg&fJt?<_A2nGSG4z;=weewP)i;)E&O_IcO%tus3*_=bpGjJ}>o>
zf63t`_zj6&=1deo8nSWYU<3ZPa*M=eR_q$6FpSH-mUiMaN*5y;kJko(g5b!M675Ll
zLqs>!te2g_spPYtu?7eA3@%4PdR{Z?+>fAgj19HU&k5v}ZQDU2Bj6wgX+F1rlaW74
z1O~9|rFk`_{6xKPD{eu^BO6!83Te6@3T3d^3l_NowycQb;4Ib`Ybnl+N1|mX=QaMl
zB0WglQ<s-*E^+qAJRLd?E<3xuiAhUVeaF?^y8R~~QA*}pwgG%vwWvW~sqHYKCf)!!
zW@W7YCk-m|%C4Yv&MW)#G~gsWk9#w%Ze?QP4EDqo-W&us3{FsXdZM~8;U}OC>pc>%
z{nSM9W*-?_60_03t9-)YZ2g#h-EQ;|4&~g>L_aK(F!~L5vbgf`GqKorVdr_AeiS8>
z1Tlo6Wg*q@$8@so0pgunUvs9uaC5!5kPe8qjHgB3c6(8%y6@FkWE(gSgoDdR_0cf@
z>s@wIXhR5Hcw2~Ul>w~mT=66SSt<O-6pGru2bstKn0++gY9i*<5HS}<T9+>8G_C(2
zS?aFXphXWTt~+ZUBrm#x>2|>NK<~|sc7z$Ca#m}7$y%d?zCIq)6fo;)ue3rQjjBdV
zL?BUG4FO|7+Xo?7^wAm$S^%=FQ&ald6iEKZI^N}dby`SEM@~_p4Ja8yBv3o_LkI#V
zk#0Ez#>AbCV%FG{q3yvuV)6#FSr(jdC#GMb1!cB;Gx$WDD}H!DLu>?tmsud%SjXOA
z>VCxrWWwtfPoLNlGVFBx4CW7s9B~K{Y<EIHMGnS@V#^7gH|Lxg-1nWFDlwN$Iy#ST
zUqt6;30S?h*&aB9>5e=gPAoVT18{H-k3%h<1;U#Eu&}EOU>TOw(|t*fD|n38L#D$A
zg0&)biLMx#AaIc>1?CX%u(pO+V2H(ML6r!c1Nnr8<QvWfB)vr*2cA%kbRXk**U=Ho
zXd|{LC*4;t?blsTmYV>ny2(NY=0zm;K=)<{maPr8u^83Fyr_`YNAICfhN)uNCy^R?
zXN6ia@_N$Q59Ac=7olx2$h#YIULb=h;0o~RmU7Obx+WPU;0JzSTxiG}OurvO6VR%q
zdsY!ju;1G@Mxk&c+aeTsHOT3fFdlD4w51Gy3zdR1!^$am0lTK<aQ=CZCJ0B(VCYG{
zGSv0KoLwJ{cLRVh!WJ;4rl}cqtu)ueWP<4Gl#MlQz)lBY(7QSwDhX&;Ibx><fTKgL
zpbqD#4A}scv#9eRA%!hW@950oJL5m}ue5ro*1yuwxo2_E^a`Z|p#KBKW$UCu4j)L+
zqZGvhjVMRV-~mm1E8kkM3iF=F{TU$k1e`$SK6_7Y-Ek@mk0ti$PT_L&_wZPwovm<r
ze5ts0BU}ehi1c6m-l|`hyuEKQ3s`;4Fkt4Rc~1fqVAwol%s+sM&OY@wD3_YGRw)=i
z!lUx2pHYd@wd#-Ki!kU`$JLUD*C@?biDgIXz&1;uEaiaaIz5JU)}X`CHdKt9$BAH%
zMXfWbOZ64l-iaLcKw?O?f(J<5u_ASw!2d?<x=r>sz&X}CngEUhnFS0in6Y8V?l^?&
z9gPW!uN&Yj!?eu1TOk8_)pmUUild)y=bI_8b{OLxQ)Eh8WpEaOK!5I;VY02F0(Ns1
zzI8AGyGbkl2|vK!vsNrm_hHr~k5g&1JRIbL)VgM(%oBmXIt|8YWc)t)8PFZwL<R(#
ziZ|&(p$Vm}zz1c>?@+-2lAi=M{mcq&H_<28<cDo(Ia2#1I1vVLDVI630M<jirTH06
zLl9n!#4$*0w06lTXCBu?fRoSDA3xBwaPP&_M7=YhHzkrbfFTAfT^{`$3d74Z6a$U<
z$dkH5)NQvJ6iO^=CYiIfbkBDY6Io+5Dnty@6aveLiio@$vf|F${~;fK;ZLm{Syj+;
zF6b)&j3J#}L>NGSu4nNy5UH71rjBj9@U)in+*HvqUmNtJR{h0^)0c3+Mga}z3j2fL
z(gRDQgUj4N_QM!R9>pOn@`&b7w8OSzHr=g*7lN9N5}gpb5Dd4&ekZDpKxwGNp!wVa
zD)t-9i%^_L!XpjeI1I)|>z8x9*@eqW3JLm>;9u=V!A|Zt=*odVMv(gB;FV(;dxj^l
z1GVJ#kS&5=G|oDQx-EE3bT0-n3NidZR7&ZcWUpQd`ex$#EWg$7DC-zRG!9L0zhIaV
zXSL0ao(Prf?U@Z*B48=hi%ue$<H_pA9|k_6c3+qjh{4#IzW0Lhap2nS)*mw|Kqh5&
zBcI0tqHWAN0zeDPygVkYaeIhC#>(UoyLB0TjFp~M2qtF8$66iCw+jv0v9p;!=>iG!
z1RaP!w&-dF-2WqBQz0ZbfU)(&b|b+?qQL?Tz~Zm`Sbu#AD`<8rZmD9Hz1RDtzFXgb
zuCVQRG}`_vX&cT%6HhuDRIy%Vm?>AsyA0Q9yMZyJewc>`8x#<4Q|mRnD^TU5&J=Vi
zQi;H&SE-ubWi?_R3c^V5d8^98Z(IO?&m~2>i2n#l7pBLUb;O}hYg8571k$;_h^~%=
zx`~sHV(SG1ut#ebC`gzrUNw-c!lbXY<5dc=lu-{XpJhu{H145tE>815<s9~8#9&yB
zu-Q$HVt~uZ0*@F8l=}fTgyXCzFWWqak|+%O#NHh&p$iDOtb>V}Pb9ETil$Jg{s9~p
zu4{VeBndJnBWMr-ExRl72h0^CF6C>zXM$D@&_av644!>=qI(h~rYcCHVE06>nS=);
zNMC8GZbRHBzzzvHiY6G*m*ydA565gkW{=Yr;O^Jv8bO5x`IVIWpo>Hygm2rrl@?DB
z83Ga8Q2oajkkO+XeSc>K*%C}4%fLX8pI{FH`*0f8BF6z!M#HXP11s&fG?4DXywV-)
zHO&k_dvYk&-WHP40cO^;8-c`IWQd5pCK33$<a&jpLl=OAWhBu;i(mFQQ;D>WcvSCn
z`_Wk<Wo&5WE?|=cevj~8iKAr(7<16tksHdT@mGsh`C4u@e?m{3Oth&7K7iQd$JLoj
zN(dAo0GZ>O;pXrtsG$maL=o(S^1f>=MXmY_15YGcB*rjyci)S^#xEf5`8`s&hKv@<
z{s-$hgs@svLuzWvqMVm(?er*9G)R`LcrN-6-TGwx$JJ-pa24e-`&YZLIhazk)0qx1
ziHHfnee>f=YDoz!CtfW`<6tNnVrGi2&Yg%H;I{b0jf{>~;b2^uF!|*1Qxjld4A>>e
zA{G;AaO}e2DN9;mbXTF=D6>%Emp&6?JHxh`gWM<~lO>tza@MXgFr10C)c0EwNCVjR
zc>(WK39t&<?qJkT-aGz87?cN>zyPy0$ADaN|5d@ckp~4uZFOMZLyRFY1>S{KGeC|i
zk&4FvjC6y4Pw5{lGh%oU9R!c29D)$EZ#p?SV1+GbeP<Pl?%JRwE1g7bXfjFWHid~Z
z@lfEWxf|3|AZc>Ipr`;D;U`QKH{*J!2nulcf;QjELW@Sr0I3k6B@Y-+5<Lfy3Dr*!
zb36mln;w2e)KJ8jrKg~+^YL0t;@=`XqPjY7btys=jW&_{dQj7p;b<IFc{@oI%rwI*
zfIP47f8;LP{$gRs&hgif-PjmIWE)^zkN`&mdzQmhkiT-}527Ix@QA^3{AEpme5FVQ
z)*0QX5!G>pczLkpS{;#E0bvr$S)!0#BS5#A!dvSc_U6-i^@DHJ&tV<y?IB2fzHZuD
zT2(NS`zl0ai#isn+AkPHq%43HDAUgYM>;sz<BocmG9(qfbw8pG`sWeFvn2)?Rw_U(
zv(O9ph}ynDj3ATKBFMMsZh?DzRo%=6EghxvyqnPKkm|QMP&}99&S)!oL9+&}bCR4`
z(e_A@wj}Ku1n@8extO63#(N=K)C1W08{)wtY>fCOU{q?5eF)TExOdfXpx<{n{ul}y
z%g;AIvLTEoR*!(rCZOuwSntRG>=|SW4e^?Hz<Tj+<Kwm)y@;GEjVHwNthe7Yvg$@%
ziH6W8;s!bmB6=Zk3PoIv4+W6W01g2_#zzr1pdKZMJ=Em2v7gCVLwIKr4TX@~LyI4+
zg}|NGQ7E&jz$xK{$0j~Lz7wso5ol_9K<SVLyu|D0h+2;ra0yOYGM~`&7{aOnFv79W
ziXY**i^{J*_)_~as4d=sf`Ozk*z-Kt4DvFJO7@oAba0qo#(J!Dd9vi*a_(PwL;N`1
zFQJ=g?=CvXqxc5b6wFYagim+35*?X>2AX3`9yene4I^F&1jQNv4dIxXIq@y@GKGUx
zNN;~2#kSnKg|tp<8qhdD1hDzfwY!56?DcR%5t*Nm+sn1izR}VC#@d-w<WmrF$Pxv`
zi1KJcaG%qeeNj_OfqyInn;#WHgS-eb1#N+%dIkIrTC+mrZcuN~5u92s{f+?e{N+g$
zi16~!I8qQ!R({}*w-VT*LWL9JX4miQ)Gq)5S+i!%R+Kj6VF-&6s(teJ@4{nOi|(A)
z-$X)3S6mjqM%!l2R3JoEp`fGhjEa(mN0eL=tYyHFLeCU=W5*?|1EZA27MP0KRu6P<
zIR20V_am53@La%4oh8M@AII-t2b3dc>bVQxCLp8fA$l<eKV6pwxCc+pV$PJuD~t|x
zl1(5A<QPY2bV=S>ue!POwUHg1dp`1yCVnAZvlx?JjgA^HE)(7WK&w=8qbpIy7>^M`
z_LNSZcVdQ6fP$WhW72sceOotz9XT@Q)#{`GLf;JnZu^~GfVP6s7l5;dX)kGs(Mg0a
z>y%HrCT-M|c5z|A6$kz<rA^@bx{4VsEiJ?_Cn+^DV}n`%kWq3nc3ia5A_?K7No!lX
zExjbZ_SJLj0k?+FIQ@Fdn>2O3h`E%JQHAMFSOP;PpRQz(1<1O5myTH|ks|RDG|b~_
zc@~R=PCPssI0&>@Z^_HlsgqS?Y5-6y$CxXEm<?^UjestRzJU<1kFo1?88}|i0S-`3
za22<PODX-S_ddGZqUfLK8O)0fW(dCmx|@QGio{#;jFg)xSOf$kl7H)nEUK#rl3YPK
z05zjU*GM03BI#~L7=7<h@s8p8TuWOfq#y5qjEDoUo3zTvII;U(<s~LAps7K*{!A(?
zaP8pI1>EEePyfn73~^5&h@7`*0osDFXMX(t6&Ktza?l2GETq2VH5-m3nfn6dfBo)i
zJuXeXr5bWWGNg8Q2OA0IvRYm29Nohcj_upv_V^BL{sH5?TAatjd_z|NKd}9@sC_Za
zZaCoafy~)I+`J+wDUddiO(B?6`2V1R-Bd*C%ML#U&L=>DU?f+}x#|Xn*Pt>%N-n)H
zucPqV&?o`tk%WOX3?zD8X50vl61sUy_z2%+Jb$x4Ip2>QI4TgeqaATSigsoCrvIWa
z4oc+h(RSzA_OBjqX>yRrCPOVs6H2Os$O9gUhG+>?Sdc3+==5EnX)$&-<88ZfL`b8Z
z0<Y*<{2N{j7m=!nZA!+@Xd#-vJ0gJt6ow#VfHX_Stkk-r^ytM&2eH7|u|f(N(!LTy
zgAz9YCB#w1yq2u>$)*7G(cWiwxyku<6nmF3D&;mm+v1H0gct>X60=ZYi96GcClPSO
zC42`G42XjKL}ulP&xie6LEMSJHR}O|y_%Xsj3jz-D(TY_mP@~E)9$G#2hiber7>Ju
zuq7)BK+;7FN8SPcvxFE03JxVqQBWr-;9wH*1Z7zd?L)&2<(>62k3_<WKsl;kN@Ep7
z;o>KUrBI;)I%O1Rvw;i(cvBy^mvXv!2o({)cus*IjJ3p%VG@gU^BkK^$gvv*%_W2!
z7^}YjP4|%{0&Ep9yzHr-sD=#7D6(NDkPJ*9{8ci_M+^4{1}T)O0X{1ta5Ud}x%DfC
zIW-lkw2$Wng)z|oz-@^{;TnrU*BMlo72r5iDb)<~Ae={XI7hyWAhH0;UIqLHnaJK9
zCSPkZPUyG=8!Vv0>oO}2+6-<}>g(#1!nvveAL|byVEE1BQ;}GP#sf-^OwQsyDgu8X
z&Z^+HZ1e(A<lcZGES+i8k5n+y{l*EQ9Y~~xLL@<v{VNO~65DTHQ^2_)$CEuUQ@~i{
z5~c;@uuL5#4jOS&AOnb`U<7&TI}+d1PCzR?Q1xpNW?~%CwQQ>t$qu2>FhRz^2s7>H
zXRS!x+q66r<s<^&c5w06?!0h}FnX~1Myb(2H;n4^cbeIUd2ZW-YM3-$3VD#iT~*|$
zpR^A?Y5UvwYtdv$KV8&;SfXX)GlDFB0iPHZc-<ukzY%)Q(Pc(Jq78r~2Ez!T0(VIs
zr4Q~&7ga(4zhF1hVVZe}MtCHIBKriGCxy@z!a;qA3IYQbLh&NVI2($AV22%@LIwaw
z2}H@6>e&Onyd?EA_%<D%Vvj0Si@yV-iTGGVK!IE+3_TPP!>E9wiwa&SjE9ZcTSAiX
zFw>x$aC_WM8M;Ie=G++Rkd6|$D@g7GbO6%b=!vKGF;GO6^q1j@BWnt`qr>-zwTOyd
zD5R`_Jux$|8y!AEb_}1@fGQq@vxKPP?|`5EoQ@V|C1hvB8F2Lj_C+Dp>JxPs8)N{d
zB*U-(H+K)bH&E>D#^x~UP5@>vJZg8Au$ngodq55~GBi|9(nchk6DS|cQ999V8-OF(
z-vKfO_(4`3i2MYkQ6mEg+Wg<VMZ&Syoz5uLio5%xjtjxS{?)-I&&+WhN8T@ZvbC4r
z(76>~()Vi#vM0dIuEHf1_m>2!8_s%U{)XjwF{v$wc|gC}1&QbH+kP>w!n!J?2!%(O
zPVjRHn!e26R-6the-NTwm{2RkMs}dBv7hBdCxLd}!RLnEty?#+B3+90-HCw;fD*X$
z`%5Ni85887H;{PTz;Rau-ly&};+kbbgAL&;;ph@Un~@EDqX)i^SVD#^uH7OsFS@3b
z^I8c{SQz0wLOi(E5Jg6z?ApImet<mpqeNmHfequdF+$oRCjf#D0JS-aXj6f|R_WJ@
z6|Q6e^o$SLED`})h1JQ~`G#IRL!lcA83bCx^)28KTyNC>-ZjX9r4%*ncO#2&CYJVq
zrfh0<-(0LgE3Hs|cF}82-BDJZrvS%K(DA5_r>YW32_+I7-t#J}v1<OyKQFmxY(M<z
zQ@r)!rAr$vW6&={mGrRV)}>32Q1!L<J>Ps6fjv>U?gfrNeZHU0F1Fpeb*mn(oqp*$
zxbQ$al`P_q5E3f%RNTSd4$64%-5NH;A_tZhDkP5E2F9?ie#>j2XuG*!hulW5)9ven
z&fKe6Qq68wJ$>AE@v8ony27UKZ&%kQLSDf*AKZiX-=}rG#A?jm(>E9$YGOvteRtI_
z>B}McOB%~^TNUPU>SVI*GlCRZK_m7Ho3ExKMAJazgCD8~T)I6VD6I4RXSWh)*jUr*
z)j>Nyp?lC%<N#-{0nM1t2=MBhYMVd3(YT`j3g_#D3L+FXTAR;}NR(<CS&mTrgh|Q%
zozerAU@~Kyop8w{3XUjfml$EpV!Ee#Zh~cKXQKn4ox6VI>0%alm3M4+_DnBzH!JM|
zWW?fK_>CF_Jrem*N_!vPRx$}GC6S&jkKcE1j%E#TfPRCL+&YgX<7tSpaQE}-xP>VY
zr}9lAY*T-bpN?1<j;`B#e~>8IxuDDr5#r(I4hU6!ZA(uV(oE3@Rg}%NNiH+|WO^-x
z2jp&LR*I^EoEXB|dcJ<@LI{bV=6}xD3CT+|UO(fHlPtHJZogQs+APjeC=`V`H_KJ-
zzqk-AuS1~ciNiS!69VmTko3R%cdks7?fLT!f=^0@_Rr?7RA{46J8dC`{6Y4DVUMS>
z%bp$xM6<bN#R@m6r{&xz#A)kYr62It3Y_o6kY%R+XufuC+Vtm#CTgBABc~5vNuLm=
zR(zcrw*CAx)5puLZdn1ofXnzj3ui771=7|FuESmp{3`wmKL5X7c-dN5&a-fKMpLiv
zezPU^Ki;-C>(R<5^3C46cW<vWj}Rqu++$s_aR4zWebLGH{`=A<QH-ZAU}?@_f0>>?
zFVO&i+MDhhT;;%ey`@60m#h{#(Nm^@-U}LKfpwnSCv4vTN=AD4CG%n&?^mK2VF#c`
zzJ#JfC$w@B!JoG8UClX$1N=MqD8)a{E;4ogdb<A-$^g22iZx9)iZ|Z+C<DHwrRB=^
z1(3(SmDwfq!{AcY`{TGLw++k1@i4oa1^n%seS5y~@2gAy61_FOB}f(}SY>tovHhvn
zN;nkPeJMZF*5myhQF>=);3Z)OyiDYq@ZfJR#bWxC)Hea$-~aeM+g~)@zyI;eegAn*
z|Mus%RjzzPP5<_%EG(%n|MO)0?FYuce&bgB?GNz@|M8!r__rUVeg7Aj^S3|!?+5?y
z761EjSoq)nmq+D4A;Mjr%qfwIGBdu~5gm+on78+BgQZOs7(hv32k}zzKXY;Q_aL>p
zZ~OKf+&*duPs-T+2>A5R;Wp4)IC!r)M>&0Y-iCiZulV;hD_?#M|37ij|C`6!Iy2+z
zJKYW<Uq7;Zsm@23Q6nH9Hg-=sh!08YQP+7a1OoTR(WoFXIy`3)Fn6Bin2aLEVH7oP
zFB>`NEwx>7-Pf;(7db$HAMyG-W`NNCftLfBb>YV!w~{p$@?wRb?&XCA^;fz6Ke+(^
zzY0jV=d}lAf}(eq&1kmDA6t7N|ANQHjXx~Dc;Vz?-{VWxnqFTjRrLOy*=gU?Rf5|c
zLlPb{PCGlTuXuh@isSCR%@Qgud8;Dtzqoh%!AibY-<{lS@ZGk@Z~Z1t<eFK|)d<O3
z_N@87uUFa7p}uUSX<BTl7$^6*iOHAaaeA@lSN`&s|K7Orw{INCg+I7`^5_4V2UZ>T
z`sde^weYY1z5ol$f3LxRuK_&_|M$<qZ&`A?Jq^0He|f^j*$|GMf?S1n1I3Z+ZvrG=
z2uvb}k%Q)hc<~0c>(;K#2be>I@+fs=oVr<5=j%EHY>izyK>ObgJYphcr1f)wfI+f7
z9CKiESEW3LB|Gjn01SRfQMa?Hi66yYP;x7_bcy_jg1k_tpT%N|1mlMt99Rk|VOYA5
z?UKZZF|6pK<M3DA<tAC1zkJe{Egb+<TlUqmi`&l_r!U8!rdL8Rzn=G+RJKMKIz2ON
z->#Xqq*vts_g4yS%8iOMJ5FI&T()`;M#C~khhEuaFTt;jX<HrrjS2U97w}vEkH1v~
zM~t0&_C(bhq7yg*`{bT64*aDmH*lvU(Nec))*t)E1;CFodt-|5<7tksk?!{1stIXB
zK$CH`w6M4;5aYx3aL~%GHo+9k&%Aiom;y4_6^-p*6>09j1K^Sbf1)RZ0R&jPE?#r&
z(gxJy%|T*3{~jtavFF8nOW9-$*JNDh+m};%T~-#C8s;2mQ7VM~A;>0z6&{BJ<zR_+
zq0Y#ziXBYd2CIs`H{T*|qMU|6T>36tyeq92c;<6Js#D{%YBCdw!xlRCHGjQuPiw9N
zr}^tmLNPbr9bbA1c6a8=tMPFh%?8ptCD3Kb7-C#0MpLCqAAG|zo4<aM3rmS3uAe<p
zDsmuf3_sBm#6<NzaD|TK&{TGOm@1cpUp&;0aeWk?@#r;W!*8+fdD<2>^@A?H-q!#8
zob$%J<K?*y^L|yrJ>fwg+)peoI<pVIJ-!TY;zk3O;F_DA{_7yh6p8>0;V$>>3s`zY
zg*kT)>$l^#``Ce^tSjWQQk5~d>3aw~!{#5q_9Ip0^}2yvKuscLUEu2gr<?1p!YDGX
zoPXqWR~LBZ4ge4ScZ~5WkEFjQr9WK309ucyBSzNc>rFGST8ZDVeUzA(NTVo!1M7vi
zG5+D}+k8BxW@2C=%ZZzl69xR;3|u|kY%vTU@ExA!xzfv@`}*=5mln8kZ2*NwDTXZW
zJUl(y)3)Fj#<yDDyjM^z$-{*i$s0`p{87=iZ>MJw+=tEn{o{_SDdWM9Wgb*iRB&1?
z#=8Z4qytV<#gv*xXIxx!$!~8XnV6kkB5j73na%;1g!Tj&X2*gcXjY@7TYW;3UVrPi
z*WUuf-4)Ia$}w<zdQd4N5)<(CGL_`<0bE@R7KelA4cuLKsc*MHr~(dU%1eVv1)0vu
zOXKl+bs{>eorjNT;~X`u`kL}zKWk|Dcmx~kZF=y5Z;O)oh8D)n`~j_~@Yajk+z@5p
z2kuAzdf9Iev`fbm506(siZ7RO;hV+ph$uNBC<AL?J@8*e8T48H^OwFo&*e>+ElMC1
ziGI00S261=2Uy2Z64T%4hL;Hs_;xy7t4FU12;|d=JdXp9EeAObH$ek-22^)b;0ArD
zy?=eEBey<(BGGDr5Wog!z@Yu1bXV&hSAA^b(9zMMNZR}kPv^13zYfJ^%r>1j2<&J2
z;ql!uJTj|DKVQXV|GBLk*7433StaNb{;>bdBaD~A#j&~2ENx=^0N?EN3#s}2*S|iy
znF$mqKI%2m>$Ak`W&lCaCf##0;#^-f0JrBjn8hRTNx5<qR0!60YgwwAaKD*e+OGOt
zM{sD%pQqt3ZOC9{IZZa0`h|4F#XtYKf}xFr7H@*tuc4}0{#v}Gt=E6@N8Jc4vA07=
zU;oaCKH<CDpwX@Y*ll?t^`R%GzpoIVIq<6ZhvOZ<MQx1B7hYq=k)Eml{By<Xzct{0
z>tGz?|L1#wXJ#iTSmM#jOM|N~6K2R|f;&9V<^!F}6-M#0&ePK~0ZJ9H4^Q2Kb43EI
z>9G~W_!iVJnGp+1r|$_O|7ejzOX+m(v<fH(&c(&W;nC5~m^o}EDn3m1^+DY}fcbpJ
zh73G-96nzIi9U9-pF+d|0|KsVm<0v_o|P{J1J-N?G+#H+7z;2HVd(nOhjdrK#QZr1
z11ndSqYe{%e|u4+Un;;@YfK!N&?S)du$9#d^wWJ3f7M3Qj-Ub+mf4RdUB<w)>DEZe
z#mkrL-t9y1VmanQ>m6dVhKn!bEu{j*MDcKc`Pta9FWKO>A9j<wg2FEb8Oa(9Xbrj;
z^kM!`0OOuk%Frbdgw4x_4D3tXSopP?2ww*>o--bIe(>E}-o?;gITi!<xG)%HLutds
zdp2Ys)o;A}?e_&RKzaaa<#tdSYQOzHE>Av}Rntlg{LN0JBKdy2_w~xj6a8=Vfdhk4
zL72>&0R}CAo6`qL@u`bo6oyGy!box@v^3Vqz^CmuP*rz(UJ?tA6{l4|D{=zeLIRbG
z7AVC-pb!*bu4Evmi#g7rD=%-`4503)<KhsYtc8F2xUklAJ>6NnqTpPo!`u$z%>EZ$
zu)dg?s?BvpOxIC`9IN&-R)%vwRW(1OP>Y?fcFr99yFoN(gUO&9_In<{KxZ}BM+q2k
zgolUU!w1ht?Ee6Th8qT`x|aSvbr_*(gPd>(df_gdKbmHu_+En{+XFZYy#mjT6SMk~
zNCt3Xt<kP>DuKE%-A#t-0Au1Hl)4E>)R*Ca5sf($ggQI`=k@n$<Q#>4+a2PqYhj9s
z1;ZD64LdqJ3o-Q4`uwTw5Z<W^Dk6_3Pb5Glgd`f-e76>OUWecoU=2Ao6T`leo+LUp
ze96dluW$slP_Yidn54kOe#95nIwxj6?hwB}c&k%{z4i!Pj_oHhTTkIaU?j<vlEc88
zI<G&pnJx*W0<L(-B*8C(aF-3QuK`PdQ&|{`9Rk0|8nYc{_C(8@acu>X3z-$L>r%pa
zju}ijo{BmwDwUIydj-)D8y>T(Z~cUL@tDf@HejBi;m2suU-&|}s<5=!Mi&MXLrm0%
zt6OStz=EDX-z)}d&1L)%4g{=P?#b<z@9pKb>3U|?WLB6D22>Ee6{;USj0XD}Quctr
z!3CEVrMLbf0vK4Wg*ML>m!WD3#ed(4OSmi-eorbD-m`Y)^1FF$rh?;bOzb);`D}Xy
z`IbMj5;b@gz`3Ydy9C_9Dhz~GqqtXy^u@!GqLS8cMGXylK`9bqoI5*S1sQ;1kRg_+
z>PAnzJFy~U^JaN?15DMI**^g}z%WJhC?@X_D7!cuN}>D>#V3+ymU+n-<urnFTxFFr
z;$MZ^X=v#JYJ9MaP`k&H`RQI$LC)oOwI?_cM<Pc|JRFB{^pAX==v?^oAbU6TdXWvS
zi&6B%UW|caSR@5^&Y;(By0JA@-i?TL%{MBgBf#aW`fT+UBj`(@Dob%O>NNRtO3?>a
z)~H~tt~fH+t}(Z54E3{oV;&{}-aR;>E^(MQPlpZQu5#Gj_2z_rAZ5z_yl{6gaf|Rl
z!5a<T58G5Z1jIY@TnzfGc^BskQ5*D*PXA^nYfsN2%Q3FLrgUD4S=QKwaJckMS()3O
z*tF*o{O~yXfJ^Q|&lR329#}C%n2Donb;B_^@?@{lDu9C#Lm5(_eQdyx#C)bpQv3lf
zmqBNsWsTAy%$N00grr#HPL`MyphDGxK~1_|iUbrK+~m>=8!%BbkTz=#`Y%dyt)*QX
z%+%ajBbOZ7(|tSpPjm%LAH&!PMhhMmHKxxFKhs|zuHI^#*5*j($fcG7zLJts)?|k-
z%(8#uh<g)!F-XD;!=3_=D}Dx%Ac|p+Cx!{~`2$e96+}|uk&u9CR2rOWgYdiQda25o
zeGlPO<m2<K0~av7x?$0B!Hol1Bc;~ZRuF`nHdeI>UB}_CJ!j8e<TR_?>E_QM$nutM
zKwF$5AO}8NA$?+GJ!+VAo5{`Oc~;iZQsLg{Xv32yF<2r~8!ROLL!lSM^7R7EH&Dic
zI&69rc@25w@>4d$b{ryCB7@<<QVbJbr*`9_$J3{h7?PS-(SrD6c|J#Pn(w_Fzme&8
zVudIlv_7&%F>em!9u2S^NY}&3ubJgTuo%ebjfGj#r5gUta{_*rAUYKyDyT+T;@r8Q
zi4bSnt3y|ZE`4=5#%aQ1mp9PE{x(otyc9On!qw~$HNgA+;Bujl?wpoOSYEdL@-2r^
zxJB5w=10z?g2S)3@Rp_{B2%ap>W%<<0p{r(NAvqfrMtrXR)Jt!YS6z{QT5?a_Y3ja
z?(<fL?mZh?KT90Np80ajn*FNC94^J7gASzaW7uz}GzihnE*Ro172IM!7Nv2@-rm@_
zJPTwVrpEl3Mv{6fx{K&Lk`UkAL&=@YT&HS!+@(d`a}P4Y5dBIy7({+my@;X<UlC!^
zuWAv?mBEc)I;~@+M<JN2E@c-qz0#0?MnfnTd=TeeH1ynH0Jx!(JB4fh9K?|!-`1J!
zmm=gmN1%Y^b`u@H?B8PU8G~jlKQ(78W_KZY9INJM%I5tT{m?tEf*suTWH1!bvUAeo
z0z=qWc<PwXo%TQ2B^&yR0!eila*yAT{%eq9od@j(ZK*YapxYT?Hy`vUT1IfZfuEqC
zhB89*<5O>9XN(qZFMH?JhE^=aQad{O?gaZCIncS3(Yv0a(}qir5inb=nw1*#ahv*m
zpt9#h^1C>@G<+!?rG^AZuzCCI2xW?n-V0g^CoY&qzX0_@hm4!j(xB3z-hrYj$Hed;
zc6ykh9tz`X#cqiv^Edkl@}Sn6d^#muIoaoDTj%{5(%@!m!M&b;03(L%z^`S65HBXX
zpr1<?2mvIC8@`3^j!Gu>T5Pn)H!BtDe1(I{iTR})r9j;HwXhz$_v7mA!D?2B426Ik
z6la&d-iYT7<W4TIpj;4|PeBdBjEstXfT(krnE{szg`0D?7cD<M{i<<zcvuk5P0aHS
zQ?l6SBn+tlJ8qjRSmAt-UGdCM2F|lTRbwcP_SN$)Kl<rq8XlAnVAGY>6|(F4Jsgq<
zOl`5*<0|;!UfT3PrZ!k>33y`Y#ABu{2#x8_gIOao4Jmj;weS2^O-I(;2<#9FD$W)a
z{U)bQg1<)mv`W|WzDM>sE@NNi@HvkM+c!2td&Lc!yPL+$hjnHUe1$d}?#4YwIhb>e
zXY(Co=6sk0(@t-CET{~m_h+x=w0L_+f5N^Z+ml>J4r#r*hZmgtna&PAn_WsPIbmQA
zorEg5OoyyLy{G7?YWnB*6oj4sjHD9?H{SrpEwr4J$`L>}(3hr^`w9gFg((6Tu=r!J
zGIq8mHD~Qot{WE`09uUi``UVcd<!sA8`U0rCtl%M@d)@f&Y)0)s7<_FwBi*ASWg)<
zh(Y)9HA1*@aaj+K%lkvlB%U*oJ(y277sBTIuu<4cPjvqG>r~y%)H2FQn3`-jH=deX
zofbzGq5Y7P#z~B`4wp{<xx(I^aUHk&5DMwbnG7ROCU7(tDv9)KA&wJ8Z(vYE!@QT=
zv|WR}u(ID|?Yeb^+n%aEVl*Sn@wCnkSXC}@Y_Vp#(Alnx6^F+@6Q|{Moyzvy*?toy
zh?{OWHZTlccxb@qEfO3^V`||V>eHS%J`|Yp?Uyq|V{>cc?xUVAgv%!nSiHBgW`^JB
zcdEe+nNWvhrlOjqJgxSScSYve1Dp1r(eTK-3P8IUo8IBWG4>Op==pC^G;+U`JgTnt
zNm&`8ZOjaGBJ#HVQs<w7lcoy|Qc44dyf;<?6ObV9iXAPyBmqx0goCtfyEM1A@%CMN
z_Bg>`N)k_1zi-ShXn3@tv(YbvXRr6?UX6L5gaRneS3!(-v0;cpT<;tm*YU0O116vY
zZzdi;quF7m(BT#0_~Dw;WZVKXrXZXzd1KG(0^BjH*Lf;0^a2EEYaqUdV+R4ecy?Cy
zo5AT;Xs*9?ZlCt&h!=<Zc@*U2f}3-vGeL=d1v;T8qiK$f<(T<b1jb#qe#vbF3vWg=
zo`SiVPH{z*pp0V|LJhAz+V$G=Y={<24%lnaeDxT1is9|+P?UWMx95)zoutwfs-*Q+
z&6bLUjS&8J68uQRSuP$Rs)WZTGre27#vF8}?2~4Vsi+?38%xMtdofAW3o<j^WvV)v
zOL@Wdt^?n!VF=t=5$06nc<Sc42_u#6g)$pe*ej4nw9!=@GW+m~9zG^u(ECW`wb4WK
z8UVJz=oje!43i`+6yVi0MYvp$`gfrrjG6ieQj1)sMi5<KmJ2Sx1)Z6w#v_Msg>G0&
z6;gy;q>SCElW<UssxcG;ch(wQRNm5}4*h(3ET|m;!sf+L!y8ejgU4}fMdzM9dzg4~
zU2!A|DYt;uc(b5a(Q5T;m}zRE!D8=}3j+iAqN*w<Nl#_bH3B<U;F(rgx$@kTy+s*e
zxo@sH?AiPQ<ztCpK!0Nzl?m%1gK4m&&J8GgA;h%vZ8PYPDDV%3iO=qW^68;97?r&0
zxkPq6VGQHEqWZZu)G8M`fsysqXMJ&nz3qp_CUX25PS@l|!%j#8DqGwio})@4rdDpE
zCgR>bq(EKH^HxEH`%;@O_1bUV44CW#xbkiw`xJs%$wVckR#5h&q_ebOAY(Z`|1mq^
zPcIkox+<pRu1&BVmq<)6AP$#ZknM|U6>7>;jz>qrue0D#A?Ip}!>H6msIlsT4`;*V
zcq$9a>uq0aHS=rOR6r-h*&@;=k+fUAQ6UZ&)ZRtyMkI-0l;V|>^mkxMLOWp=Ny^h!
zD0J_`lJOump^g#K=xG5hlzf<BmE4im1Ne9^1A@{1&?ZVVnGAt-{jQ>3*T*#o-T(#b
zt>BKv*%(Cdd;!$G$tJy#5!$Xo+rdZpnIeWK^M8K24qS*hMkz?4xr_mv@o+arPADQl
zih2@W!=RH|C11wo%~}Ac4}slg4eRTq*p)-M=cRJsB^Sww+oR@hQ2W<M`v`txEW<4*
zayI5(bJEhJ9Rnt`JAtycbx%ZCk6ajP1pE2vR!Q)la*kYc_z6^|uMqLZ9pkMI^J!hr
zoer0R6B>S)5B?@K!s6Ks6RZg8!%ABXT6trfGuc6?&2`w2)>zV<N3btGFNa|d2~@fs
zW*0110tN608FXfS`u9)r0q2Lg-y(Oe*Yz`#xPh2Wq0kyWv}hZj*I3fC(K>~U!w@TK
z-G@n2ZQ3)Stn?~_An>lk;ijCg<2L~jX+CJ>QJi>QlrZ4REKY!}MKD+6Qz}|1xr@&Y
zxo!F2(0ZgAC^|L#Xz905r49Xx6IMc&o!xLq>m*u;k36Nwe0!Oeh0g4&(R4*mV~7Tj
z!buHo>sV%+>udq%0)K0q<0#H3;}xMuv(vYB>LD8sz0sIEY(M!%1Ag%o6HrFL?w*6v
zB`2O5x=A}BmQ&HD`5YonT(xw_B~XzS+kL3`(|M})By+_-SYC>KopRh3)<@Ya6>wBY
zMx;O4c?1RET9CqJM8uH0lYfoC4&GA;H%vTX5wdZ%zL43!g2&TbTN4<pS|F$E-!WVY
zK7MVf=hLT!Xd){d;}L3iEtI)i8)#?)gSWCMe=E#og1LP84HXKGm7&aFV!}}pnVVe=
zNbG~BqWXQPP1>U@s6BY9ZnF!_M5T5g#7OGY;TWQT=$Bf|kEH)rIuA(wFH@3=3nY<a
z{ODBHS#-R;Pjzf{DR{~TGh=ZM(js%{@Itt#ql6CQTD3L+i<wV%#ho{pWF3kcv-N>%
zQX#ao6|N9r&lq;@My!Jb_grQNo@XbrkE<@jccjQg@ie-I;3K}^h29J@F>4fv)WNZL
zV*A8a6cJAtnEf3^CVRq|LG$BO;9@wxY=-867&aY6my)8Mxg!#V9N_X`t~dik#gREp
zh|Q=Hj<Tm(m@-Vt3MT~f)=gjg1Q)0zPa)Z8qvjbx)pc-gG)z1nB^x)jzl?sOW*e>V
zL2u<u*IB>t0G>BO<P4hzH{Io2mD*%}4&H2g3F6X2)$UR|upg@KTTP~UtmOc?IO28z
z5d;%hu>Ug#vkL8#c>HyuClMGVDaTcvKpmnD0mKlb9194|j=)>yAKTlj_3*T?96PZK
z6(oXK;X_Gob^^;V(%FJRD_bi{Drh`VGt(1IDNDbTngW6hl9_PAy^890k$$V-+&tLu
zMyiN_2`D28{jAKX&fe6_Xf!edz*QhVTPa>`;6RjB{X!JD&6c%l37D-V_u2!=C`c&e
zRtf6u1hxp6XCatp0hyq79Q{mmxiWIGeTu^?O3ExO_f21|lsviHGt$BzrNT+{?s5tW
z*ZiKI4fy~qj2m`iXA}?NmZdRWMYQl-khgd~(lQe(52#o@2{f$_>D#PM^KL#QXCDw0
zTp)v~O`n)zygR-$A31cPul@`Af{nEn<edr7)e($`pmhj_0&5^@xJS7U-(?*R_IVWx
z;7alVB%;&?1<b7KqNKbKdZN39IBFXx2!#dM>?YI$CEdjf7k<WKJ%Y~Gqf5rVubnbL
zUP#a^lS8NU_LN+6fSj}%qh99_Y_tqIo6+@{p@EaYGvLX@eP%4fw$A~0bb5)}GU&r5
z0QMxcA3=uTd{ZTN=;~{CEHpl;V3OG2kKKRF#@0E8Dyq#P;4zYb628LV9H2nk<2n|g
zlGnnSxu~qnA<1cWQp)S#CK}Tv^P?(`QB6OLt-z~lRV}1FnFJ4`=JgV?xJEcNrMl19
z?7hYF>?u?&T0jfjP}(nC;onxvyq=QoN7A1i@6)JXxQ%_)l<si1Y_ee3$nQhy^jC2X
zv%Q~zvj;Ltac8vf1=0=7+)$vpVFD)UZ-5ecySjDT@5xNi5oD;B#|Kqzx5gz=z1i-W
zVzY?IdM|SbUE!})eLUnMJ70@S#Rk!(J)P>|P1aH<ur{)sIeA=eWF*SZf`Dqm;#FUy
zOVY!I(V-Zeel)^_!EP{ZNZ^_OTuA-SdUW}oOk6RzaC#Aq!QY1F@(N=IEEm2{co518
z0a&zh;_1B<Q=?{CEq@3#k4=w3>p0dph?9(mF@q<^32bjston-IkV(<6PK3NQFDfXo
zp%9gfsUg>Rvub(8_Kz`xXLv!+xC~im!w_*j^uzqwkg^Ai`~Jyt+wR-wT?lR|1!>)(
zz8_w1aeLbn>IhX<KIq;|I3Kzq7FVCRI~^dmd#(6P+X`N>l!-yG;x|wq47*z+fg6Ox
z$4}G)zopx_Ty*pu!FY+0Ln!}+Wdz%PyNuGsHM7x*q($tlm*=h4u3bx@ycakQvjX2E
z51skek1Px=(7IAOqpGSvV5>Y$Z)jv-H#K@6b>og4bUe{LXx`V+<)Kbn8)_P*iKHF2
zFnG`cnZgV?v(~zV5<VvPE>uWi)ZtNWDH|b279c~FfpdKD0^9(6rZkeuy_W$5d>^_w
z{E_O2517$LSo{nTv=Yr`KM%P%_*ur_N}R&&qabAlnn`s;plCI{1sx5V>;ied6vR(6
z1$slmB?wMTvj3?Lx$;A7=D-7FihSfyZ+TV(B(cBfGUKWda=X!7u7m>e9^NG%$xIu~
z*S9ehg>DFdISH^tW4J>omo{E8AD!EOA4#o%@+bjz-z{2l5^X9GUzA^5XU8bR4dC-B
z%%kZA7G?#wE=)?Z5H(VoHg8dkrGnwx^Rd9dpl1aGVC<>GLT$a3&_&$?C{4H!oglO#
zOcRG`m6`Sf4)-CcE5KRgBA^JyLjYN%fHoivF3d+jmnb9#1C5QawQJ|DUH2gkt3&HZ
zDFEQsT8bGIcZPuRQiKgkOgvoJN>J0Wfmu5TS^+?#51PCoOBGNy0lu%a7p~VtO2TBV
zMA+kTDLyR|BMa9?<IH`oIArUA<SpYqDu|^ys{m=a;?3{fqxZ$4B{8fiT1oG_@C9g^
zg|hep^Y+CPX(&;(QR2?W=VC?h1U~NM|6=bwqpCc&u3=M-rdVPPpnwHbKtQU}O)PYf
zvgx3LpmafsfIt!xuu!Bofhe$%E?t@>AV?FXh?EUcDbkfHAn?v*a>n?c_uu#L{qZ<X
z#<9TWF4wiLwdR^@P8(m+VARsp-GUT15MwGQ(9octPUT|_LA4kWYFh9DwS%g6Jhl=2
zMSt2E$Q9^*@!)c1+t7A-gTwVIhKCdg1Gt2V#Wxu2tov{XNm$QL;fRq_R27eIhJ*m6
zf_CpD@yAu?$yVh$#GguUcRj5`^PNtk&wYMSXwF0GM+BN>ehydU)O`MJdeV2%uX{58
z<sJO<=T+DKg&hC=<0_ut{)NZ>{fnF5{FmJD&tDwb{l6dZ|L!wb;r5#2<Uamk#px60
z&)@$b{q-?vwvEZxjvYCs^3MFo56Mb3Li<%-2&*62c>D+5BU)eo^uthd@^McQrHvbt
zRgS3aym_+aRWG-!^Fx#ISrzA~Ba@!;TyxL6;^SlL$Hsi3J7WGxl>WO|&VIM{e^Pe;
z`G=c)2mcA%|NXjumk#sM>Hon1{`a3_zP$23h`|5;_5XhO|ND>SrbLCn=ht=rAtW96
zRksZc<Bu_<9<Fig@au!>Xc2~^Dico!Gww(o=HIFDJY*W^q5~2F<UbTDGO(m>4uD0N
z4LysNB^t)Ya9Q)c6Zjy$w<^7yH@E!z9MbY2;zBCr|9lUV)VN{w{^U$7{xXAx=5_ZV
zyA9ti<(v$FDx#-N=@1T}U(tjL>me*$=^{-oF2ntyWr^m@KZAZ-ktr$d(q9<=B{^Ce
z)V%-u*OC8Swg0=B|L;YCCHwy~#{K+qI{Tkb`K1_U1iHxR;4*pl<WG?QQ5xjfB?!#>
zXH++NosnYaG)Tz`TSq0;=!<*jQ1OxDk6jUT00+~N9!a-;bHwN)3`t4DuznGs%YMxS
z*&%~H$teJ5NElFbZKcg3A$9<0QgpI@X_bOr7dWi;`4wm}LmA$SKI6oZhVYV&VqeT{
zF#l=i6ivp1K*OtS1A}a;xm>b0HmhKK5re+T$fykM;dYjeNx@qhg<op#lb1I`Y1$6j
zMOIP!U9F>#ZBoPK_B{SK#@{?c23(Pk#cQ`O-pr*c7CHF~OuuIqJw!)*ZSt4eBw3CW
zEeai=WVAH}U@;26w9sUdG=Z8=I&@{okqaqoX@ol@_mG&|Ud$qCVp2}j^!}RX;-`;J
z7@0MqNh74LFev21xIK9{LCl9{#lh+Ad;2!5hjD!MRxU0FiT6_S^75RvR|+IA{Cy<O
z_A(%<8C35~%|BLc*bjdi7^c}h8^iQ5nLk0T`Q(D`+DG1j(R>&8SbZrgfSV@oUXsy3
z{V;?^vJRBuF<2jd$kB|{20(1U3o6tF@Cm(~(p!gePcOh6x`-~=o>j^iY)WjW3xy9C
zjAi#j&JZozX!JVLzx>a4$?lAhU!h}n9M02N91t6b$CXZlUqE6@l3j~Hb=MbuA-f0Q
zTe?<~AL5sFp*Ij1n-*pujZw>Q{i^(z$Db+hzby8Un<}tiWq0lYGbG2-?cW^n9KpkD
zJd(ac%K1E?=k>ud9+Sxc(}oT<0!v0P<xO=ZpHCFf=iGbIw3`NEE8Y0jTq4?e?||=<
z)6apNKCw-E$r}hIy~#5#w6+%TWwH#$4J70I0At<e;K=kd`i?Ci4i!KXB#OHeQgP{I
z-wUEcsrV23pm+1x+B+x)5kDQ7u~5YKgQR1MX2J;G2q`^qGC5%zN}BPgC)LR#?26VG
zm((DJ0jR?~AEts2Xef3Em#yQ>i)30V7)h4Ns>DZ-`)%2xL7OBwR^LLoW>KmDJy1aK
z?y-a~&H%wOwl8z;zmCcDGw!?^JGCZ9x<DsmpjdOs-1y7%iH6^&onb5iB5y;wH<@cl
zj%8f~Nb%-nl{5gFlTe;cO&3rmEi0^JJTAQZ0%&bPr(YK!Fu`Xgb@>3T$SoOD)qRuK
zf2sP-dXT2}pnHV6Efc+8X*CQ4XW+Q928fW3`kJHdha)F`WM{W-N!`U7<HjVp)1^_U
z>X6lO+&FYda<&PXobPlG3r||)&*-!=0UVOZ$aF&=-4w&=4xME9hDP-Tffdw!=TQJh
z4qn&<UF%aQnso5sv`yUqHv8+JUHZryBBAyq%+4c{pXkr)9i54dw}2njQgY&#i&$qb
zP*2$xp;K;%W$S)v6t&I{`SZJ#vHaU`l*hk(i<(KYvEJ^&y75zR!)^g4GVt=s&hZ92
z0cF5F7d?HLzjs_Y2<pmnG-z(#k=+2f7zCgZ^gr8yHD><u%c|6xa<n(-O@T^hU+9fW
z9UKBqPBE?vki5Z8R9^$#;rVD3RkrvyQam9u0IX84T)*)V=ov!xQFv(lyK_{)Nw@F+
zY}gB|H?Q0mjUp8A8F;6E$D*QavNFaqAV(!Bxmk>oJ6}@hhPEZeI3RC_8wT;A%=vri
z_5+)t=1U@t;zFTk7mlbn!~EbRwl-QjR)tG|SYMJ!fT(3%GBf?@qtH{mo6Z#}4wE7g
zo<e?kBIFWP!uN*+NqlK?wBh%HB4|yJ#1L8?k!RS1xT8MpGCoI=G2oV!8q4{UXH;1i
zqOLyaT0s5ZVbB9@%s?(VBT|MZmv1PUrq=xMy=O0is5q>r$h}y|HwP;OGLiAP?o@K)
zM7A^6`uCUSTRT5q@a2G`L$*nYdA>J_LXCYfG7!RSSxz9|8<0~j7;R9TW9St&UP7NE
z1HI$2ZVg_Rsdb$xS_wL6a@(!`9K^|c1_1P^Tf(D?t3BsFA7iJ70jQ!K8Ahrv*dxjA
zuB)OTwcb7MrZ2t21c;_6*@o}&*!A|X>u`9@xmI;YNHr?Ca$#=(?g%vDI709fUT&k+
zA~w4O)|hTPZ+;i%JwFR;-Ao7Fm39h%&NU=PgxgJ=#(xNXC-lBT8aJUp5;(XBYg1Q$
z(wM!*pue5Y)`$hNT+~112)7rkLDdQVFEsG#RL}w|=vc=1kmuR|Y^yvd4l;u7;FX5A
zX!McqALsgH=co$*?zeB=h<i?-503cakt+gr@%0#-ixEAHFK_L(`q}e79Cw$O=1vu_
z)@F_1;fCJ%RabcltNF9i?O12Owx{_|6D7ES1oT6_fzs;r+9t?)v;qYoxfzm7f?sDE
zsiHvzr`WJ|0?jvF=%C0p&kpfb@fUn}y}Wtj#xw3u9z&+ZP~9hj11v8F4jRQtu#T`M
z%Oq~{OY~fxQZL(lG>Ah&1xvtMd(ERRwtw0Jn4`78B3u9yz`ke=RABKSxdi<E)mOD6
za6L^i%;zP|8Mk6Mc=m)b>Rdqm$OaB(;99<V^=cb@D_p!EA||B(3Q7mf)Dww^1P)A+
zJBLpS$=~)zlcPF12_&!Xd3o(B_G8T*MPYbzi8Xa(Zwwgn@7zx}bE&~cR#%EsmoYP>
z3yJQyF@;a>zxwLb7psJJS$#P!2xwFX|H0hit+-(BpYCjaw~-ML+#G2S!5GXSJ*0Qt
zn5b}yd`hAJM8Yg2FA4n=a=?A?@y+-BjdAVH-4J-XjZmB1QTc9_XaZVVuT0e;bunI-
zAXlbXhc~y$dykSn373ODWmmKU0tjA*GLoV9vaA1{(%D%Vec0&^YYgbr^m!@ZJ8fX{
zym=w}>-0$o{6xvjB1h1LKwAYG33?f96^LAQVf%tBEe0E6k`NZ91mEHq(#NT2Uqp(s
zAP~8yCERlZPon}YM{sZQ?J{6m4y}_U9H7#&#9$yxR5K`0WQ>s#r)uxMRXiXAa*?R5
zl@|oCGFLl#{)JY!SZlz-`ydEq4+C7Bz4QQ{$j`~4FJBJoI;1dOQ~%Yfvj?iwep!U-
z(~xr~0@OPs5N!a73P9*amqLvc1tF2QgNvL;NSa7vYTvf$ARvaM=oL8!icA|iU&Pac
zj(;3i6tC(wy35X73$kf3@9=@&M+K@cvn;V+0*27uA+83?J7GNd7M>hJ@x%++FWjT!
zTYyrjyp3Je6Y}3Oan-6zKjVdwU2Q%KR-i}==Pr5Rp^>yuIPdpUTVJn`DC7pY>6@^)
z5L_Vu+Z6|w?jQ>mIFRQm_3ozo?d+HZuqQ3&4<=X^JkiLhlw7?Um+@T8$jA&c<Jm|m
z;|B5}HYbM$o5l#Fg7=}Z(_pm-DkNE!rq)z;g8b$R(}Vg&Ahrf1M};KB^qB?N3#5sU
zvv$8S38Rp*w>$5VRTrv8e+t9$d<vWIBui3gg2w_$&&kCKiw2Pj8FaCxjh5-~S-dxj
z$4=tl!HH9=ehIh|{q2w-m;iL4ba(b9Xx)(0pKRy~PQuaV?=+R^0KqEGWqFfu6?!e4
zT#0!8WUfjycoIH}TY^323v@9^83dKWEAd2}%_D$sos3SI|I{RnTpx;rHih^yYn-F;
z^&Zq<TtKkdx@F7R7#Cv54L`qq1<Zl|7YkoLeTBVG{t|Ixr0oR-HN|=|7o-4yCX%Q4
zsqPi>iD~=Mn$3?i^Mj+v!?6EZe3un3I+{g8SeWVBDc5W2zi!_2Z?UfQ2C{FTHG`2W
zzkh!LilJ%$=H$o=<(eOp+6{1HO+305P&~sGe0`t+XSQqCE>E<%<lgg*JNlhDkFO=A
z#r53snaDu*W0BKi0r>}ak_>^g)Qmq5zYUy_#y=pLdP%}=pwW}C8&)v`l@s)M?qZjb
ziyAWjwDdBbu~)fA;nUUvm3!0+X-Gei_a3K=4vux&5p!tR&ER++MwST8u9GN$NXLy<
z2QF711%<Y$uVXaC^jI-2*xNbKyYwh8hx(-|V#?&l0~f8(9Zm+kY=ff*$L1|~T9TGA
z68yW6a-^yTQF7u#Q0%b}A3hj8xLiWT0UaPJN=heTT0okOWbg>-w$s-qF@}+jhs?xQ
z83B3vdO(0K&=r#PBtcjlOz7!rU7ykvpo|O|)}J7gc8*T9N@)XOff{UgmIaxSO5fy1
zjp&K<M?JVPl=lXil#ypMcKk&H4fG(P)3X~2%nY0*ZY}shH7U~mzyJP<Q*8z;$af$H
zk=U#1{E-EuSkvH#8uY+R<2-O_hu))*y_yCQh)xb%UpNm~o7P}8kQJ=r$YtTHpc`T}
zklXMdkl8}OptPN8q(bz=FP?11@-2iSD+v*!JZ(ef94??QRKIi=`g`eErRjB7VKNHI
z9@2`8hs_#t1ue+dO=GMv4Xunim>682Bo-&K_d6KD7{S*f(CR?nRySRj4=%z-kiOka
zOgxo7EX1mKNDB`s<w?XMl9NJuosRW_g3--`al{MM4^mYBw_YPR8}g;Z<0K;}{J4sJ
zO)Y4m#AUC7D(jmFQSef{am^Z7mQyOr%A~vpzH2kyn#*LLGQ97GQ6S(N6G`&-(W4`B
zpcdojG24eP4!81=c^Vj3!K7RcstxJ<+H~ZlKqWiwj_kV1=V%1eljFouLT*E-GU(xK
zg-8}fNs?*9%kk>#m9pkVd5Q|64Vhe4(cHTiM{!{ligi%${;a4+Cb0Go5!5JEL|X5a
zyyFp<BTF%Ca-c#eX{VKQP;~7!k+3-nCK5~=ED328hj<nw{tBtkqaa6SvXsx7glL&F
zvGquY4ku||%;lP}43C#-IyGY!7)+9D@kV-?URe}`o3=&UgzmIb6!2UM;+hr|S)0MF
z8e=*Ydfza3o=?uUPpYepdf;fNi|&e>3`y#f#UwchSayN(CpvnX-*I-2<<}UvZpOV9
zqM)Su60KPh`NudG6|Q8l0a<z7^hLkm9a48d-xxusl43Y%DdMTqK+=<?zwlioDUvb{
zn1q2?&!D@5p#_Llq@kz7aI7sGOGW~YoH!0i%%7+k37J4qsSjXMH$zVlTjghP>ojK0
zz4yakcd^E^4D3nd8C^8F$@o(8H=(FSCp>@?K4`s8c6|eHZ;W-&<}O~$FhQp?1!~UL
zFa!8*!v_7J9a6eTL?AhDKu5Gwz$KhXv;axC58`5X5&fP*{~vxSTA1l+gS8oAwI+`L
z_wV;jX5wnn;T|TLA?o+XA08>?paG`i)6u32X?y_o-g)$g%=ms(5~G&vs#R_nA?mr=
zH|+{i)O2>b7W@uWy%AFS#FsE56o3qh4cp})0Y+?oXSy+H3?{)ZDE-&BG(qap@-1>u
z_cV>!I*42y1Z7XEvZkh{AOJ6XJJitK(gJ{i7}iBYJXW;^AD|X1Uh8ZEaLaBMO8>9&
z6n5@>q+N^{V_ekbg#ox1{nF8fxKu5m6sd!zcqxJC@3T|&7){1m2Ws%t8xO8+@@@A!
z!$y1M15``dFzw=EG>6=lour3V=Oy9$J0R18D4<p0>DuxMQ7Hg+K5X#Yb1GTBhVJZ-
zL@RP#U?&B7OqR5aeX2lNVB<C-1E-;=sgFnpO-pBaA0TzIrn`19!_c~joCBV?P&=7!
zoWQ;g07dJ*ZYroXt<kXk+xX*;gWAiFA3t{UZ2^HfKRE`KU-$1a5C6*r_y7WObYu+l
zO^dUS`Z>*+;QsPMz;sGYtVI$5`hmX_V65W0`}%!2Ql~F^7$a{|TgfITq)LmVq$HM5
z4)_=Ab2zg}B#b4CN!uZ}`0{=nyBbX_j8`#jIsMnNn==w=XY)TqePr0*4*9{B0sv!$
zrA*LEp7c8-<e!mpzyO%!BB~n4j>KsnGi0<ahf|P5>3Q%3x`)1Wgz~B02hB-5vyuQ%
z-Pv$NX(-enJF#O_6QZ&33Hzo{+5wzcN$X7(jUeo0ETFT0l<E7G`?$Xq*6Mi>+bUCR
z-Lbf6BI$6$1X|m<(tAaG=2683g54iCV2UMYf&C%WzKD~d8c0(B{$D1`9p@jK?PuTO
zB-fZy_Y~UeQ-e#WhClYXkMr{}ky;qAv_L^G)UNOX#|F`BWCARB+0mdEAt;9<DH-`{
z<f|-nKs`(s)OkHob(mbgw&@f(La{Ex$<fAB?af0@lMgt^IB-@R!eqc(ZEvW82Y@!f
z22)a@eg&vpw)D;*aadb<FaWT_cH{>a*n|-eo*#y8W_=<_(^X?u@pbdWiBK&lXj>qu
za`28CvIamQnRpuuxEdQqP2sIYu-t%~6fKQ7vICV{xvgyt`W-XT8v}_$rWIQjQPY}8
z`x7dP&Dc0IhNtZk-P7a!^q{O4xB3k)qKZ4#D&($f+=_XzI0g&7l`~;`b4__+GLi0s
z*CJU_gf#Lc_R%h@yb}$5T;|O1Cu)=i4^1rK!D8E-fJ4Ox5Rzgdhek$h%Tbk8Bc(81
zHer{uQsK8Y_L_{6SsBp0hQuW{G9DQ8Tf8V{4KvwWz++KyQXff&>U_6w?rFsQ9X>tq
z@=CT{LTVFeQsQjb2A<b}C-VeAy`P&au5&~^#~hlNF?iCoq%`Nb{gg^obv0G75|aas
zBh#g(kTrgbMz^!4qf+OZ=MQ~-D-W*S1FP28g*v5OW+NIa!{8GHLOeIsDTpTtXRgqU
zRO)dVYi+3ORwKoC*aGjzGfT+r&A~=*`)0!icL86=GL+`8ad*YfK+bTJN<;y8e~S#y
z1BePb#_FStQVYIV`0y}90Fhe#`T^YfGLuv!zicp05$a!oZhOpf_U5h~5H<}YtY{!M
zahici+NgRpnkY%aOBM2F;*j1B07Dc<V>ka>v-fyAaX#6&-c<exfRryC9v~dgs`bk~
z)#^1x%=gsaFt_vH>T<O1>6Nt54fjf%-ul(AI}iCnA=hRe;XS)||HTxm(qczj;DXT!
z=;+&V444x8LF$v2l+F2z(<=5Bv)CowLM^?$y=V4#8(oI8Q;wU*P1QH>g{h&EwWU1F
z{ec}om;N>6Z#e3_Mh(g92aJK0&9EcQvAPVXP-DsgU@^=8N7*q1V5X~U>eOGX$GV~O
zr?XnuGojln0ZMB0o!%rD_rI?TY-oseEk$RK0}F@%qml;17AZktdHEhT;LTkRw=4TW
zWL^Z@CDpRqw%PVz8zvNYEc>xM{_JFct`u5%v6+FKq%f`q!FA$;bZYC;!D#;Z%7aM3
zRT(~j!-^Khtm-2hanSp*fEACsE|x~!MY<6AmQI$qrCRR2uKn~DGGs$*WI!7iK@Vco
z*+;ni{ckO6{sR@u^R59Iob4MoZq%#k*q%8&`qNr3VRx>Nxq~S4PN0p_Ei?y2S+cRE
z=gQ;p?nOX{;{`p)r6eoX%Xq?w_{G)0K-XyZ$pys|Cpc)Mm#ocSqc=7W{3H<BSjh}K
z)blMcwSBn39dXZ%+w8p`ZnHD%9BzVDSHoI!1(u4V>#e4%t~>@>r}1K*m*u_F7R8H3
z{YSJf)TW}LVCDo7W&aJ0Jl+wowLA<+q(u~GnP%z+-Pj7`JZDW{Ig#@4lxGJLkH#yS
z$Y2f;&g4IMTxGRRW3))+bh(S=s-KO_54J5Gisly_PqKE(@d8(w^<bH!n)U0*vt={B
zk`pWMD>n$WKjt#uJ2>-@+ky#f&G?DhwXb3;evPhO=(lNT%`_Bikwt(#2WoO}%9<u#
zb|gPeLyzAFaEb9oC0t1Z#G+P@^D;`#3{iAyhmY^GY_DtMQ3mhLxYZ(L6Ee_NN}p2m
z$@0c9qgqSPNbdFG#gOqCNqKkVa?HD?jW3e5{mv}P?Z9#W`fFF^Q`C4>9rHtt;#$^2
zcLvz7?_h7i`F-jxAKn~Kg3_>qw;vuhJK6_M(Gvh8GR+H<!~Wyh3Hs02%ND~#SLgBd
z%+Q84K1{cRZD<+<AO|fT3HwNr_ATFj2Q2lB=O;3%z;H1;84RiR_dgA(Oi?~4bn1Z}
z25<KsPfWvza5cc^_(&(=4+E#u5s_jByiqZ3@cy#t^uev&5sc3X(phJeUCkh>@+c!6
z(*-e~!Bk_1;H)j?HDl$yOJQpupbRcrEO3Erq^+fezki~B8+zSJp_G$jI8Flu>ldVg
z4YyiU7ely_NKt|%aOh;<Lb#R-=7_sZCr0EPs#EH~!MFmV-%=>9jnNAu;EcdI<az?&
zZ6m6N0X&XE#k+t+m<+3U(g2iErwg<&5{pMGToYf`a{H)%0YuzUi7AJ!t$dCFC&X?3
zG+c*J+Y-K)jzyQH1y{?a1XB`kM#@3f9aK4!LOR~M3z{Fw&8K{qQV=1}Bl)R|ZFAau
z62F09>X-G4U`!6v*e^mM)mWC8N(Lg90~*zBbR?BRL2)AY1}aqW48<G0{_QmMd)wJ%
zswTFZ;;H7Bv6`ARG2Jq)#E!~98$A`GRu?YeY|zG|jsPkU9KJ^xDS!>Dw2+kd-^kJf
zLDohR`3Q3+*C#>ix(daSKIa8QC%Q!lHRYA9V1-jF6H*MZcNUP&sUn$sfrUfhf6D&g
z2K21tMObrSF5iW&0hh2CKYyJ|qRbp{vSDj91|_@NvyAsu9;621odwIiB=e2UBsc*F
z945GIOn!1ZAe0*COJZ`_+;0eUX~@*l>gLSup;Gn6-V9AfpYI(A*t$smQAVN*nT_KC
z-^7TVMu5-(%*iS}(KA8P#^|9M^hE8~`Wb*adEn^6!d}_EsSFE7KwP|a)7&vd#WIs>
zwCW_z{Pndmxi7W>RVLZmIw&C!i2_MYPiPL$I<|vHu=GgRUVSRyA2oI!R^a#Ov1IN`
zxb!nDcu)TN<$t!oP#74LHL??uu;<ODTtpY^k6>$|_7yy#7sy5qD0}vWC=%R6Kw2H(
z0|I2Iu7JN%<TP1-ljuIG`Il!NkRK+*-eJOI-JyrD4M656Xdi$6aqwh^mV0A@oH2E+
zc<2>EyZi;2pmJ=W5oqdwm;2&}n=Y4+i{5C*N*&A<O@OhcJzO|#y72nMaqHw4VbQPD
zjO1AYk~>?teK8O~rbJ`|q@<*Deexqp6p7Q1R+-+v@Hk}}vc!Z;6Ydymg>`4sS&-J>
zyLG%1rZDIm7F-<EAlKgG9+#+}3wq!kpxYT^cr++2Q(S<P21lNA^Eq|@!cPPnK!IG=
zh)eX4W4B`1)XNH?J-e)&$9)1~h>?;JJp}MU$k)t$qnP2;u?Di5sQ@uma?V!_a1n8a
znc(xbi>sdqjv42rYKCm5+)tq`B@T=|IBMvwcLv&qLUZdje(!n0JW=BgX=Su|6u~C4
z_wrO={}R%Co%q3uM{!F^>`%bBH=l8QgEVXieLH=;N;1{rN1Fv$Oe+|igdPxbk4@z8
z#0GOR`gj5}zy@lu_h|0}s(~@VpYPtC2V$t)>$8~E?{ygB@t_=M0wt4aI7Aif3Td*T
z%S?u{=?FT|1R*cv88k0wl%7IWa@pW9>~eAoAmk}!!7C%4X3EIylM?TaNbF}ExLV*I
z6MjR>4;Za2ijGc{Z9L@oK(7V}(PuX^haPmo?_F;LSZcbE$#-U3h8!lsQS{F!vCo=I
zD09eOy7%+ZqjK6O?%>!>$034RTGNS(o)VD?0QT-jzrul=8~7U@V)gjBqJ6OVCU0N_
z$Z*tTW!6+(1K}M8THw&GSlTs!nG!k3zQ0Uz`nZ%1Z2r4%zPZ=6x&&Sz@YLbswE52^
zCFu<VpT|qOlSJ$5bb<Ti?>HwZ2APU*EzE&tG*zGv9991`RwBbPk|+NyJd$k0#I(^i
zsT3_Q6s1<qMh*^}-~IfI8s8ZLfT(x#$^oKOX5!yRO(Rt)>@9KJ87K20(jpOfgeGA{
zoPb@mb!oxJ?7_ykbFhT3hssMyWhoDH28O7yZFMgWSSp{(Q<V)?i4w9eZLUTA5rFQw
zQJXI&G}ez0pr0{K)dCbztYVijsUSwax2|T#PMnBkYA!5>k3ai*WF5gbEZgw1{2Cdf
z7yzeFts+9I+JQIQO{z53>5gY_8F)vC%N}Xz&VmE^h8}qK33nzxCk=#A%TAKqxoNEH
zxLF97fDZ7wPG^?VGJF%Ar0cy^p!3d#vq*HQ2QhTO*$fLttZ@)%D_4$}O81C>&Jr?K
zJy*nWrsCundgqJ>V2<R*J&VhKnSjsQ<O|augJx)sREcXqY&(^~ReR1_?Nf!geP^r7
z)+YeSIFR(^EeJ}*dUwSM=x{>0|C@E|3S$Hp<K25imbOz?k}^3qkHp<Rd*+$db{5)Q
zFK!=Q^VWA|G0{4<PwWR|c8%|6W+c$>H+PR5N8^x<dQWX4Xc?DQdftLf+J3D&zkfh$
zg&l~}`j%z%eDnmAKQ1gte=6h6@qOcg?&9e(r!(;R#MssT3=0&PJF>ey`s=nPp&r-s
zO~NP*`)9Nj)9WoHyOLGMCg)jTy<B_lu{Q|oJ3jrPJ&styY_CIS=Sf!ij;y?UKxxaw
z6c!g3x1*bSuWp}2#(~3eX+81!gqnax5fj;e-ZDMB+SaNc7};NYeKdzMa3nc2KE0(k
zn&lJYQf;G|qXzzo<Ba2@WMO^1>&mK`YOX6eYW;zW=Z(5Y%1b&cQQHPSe}2*es%U!F
zXlq$;h{Y@T_Vi2hC^qyNElt!!^UY6O#HL@r{4<|oL-=HL?Tb2HzH2cs?yxCcIB9h?
zW>h&RA`O7hIU@(SWt3|(liXO?pCz}NmNtFggSY(IZG6TM2SoamEMf;Ef!6$z;Qu$7
zM8;IJ->m$8$}KEo4%7;6mm1J32*BRupp-RQXlU-{EADwJYX??P78<cs^Q;<&#n*d-
zNgaR?z()94r3>J^YAoQ)2gN9pvD9z4X``+OKx-7yf*hsgK?c2Hkf(}_7@aTbq-z&v
z5py_S@=B1+URiq`gbPa}f|DZB5EJcO$CMY4fJCU9@tMX9<6)swA|bn}MM)GVhq+NC
zzD&FGfdsZ&nI$|I*)S&jyP~`f0$u>NO6d%X`J-dDnkAdx(6A*{F>7olq%zh1<1X?K
z(eAXq6}wu*tGh@m9sFxR)u~E<T!hrZA!1i826BojEr=yS`+qnZKz1B+?&az9WNJ_;
z8zTdZs=v?<Qu%qlK`c(!`TLk8mFiC1*6WI>!&;-?08yLvbd4bC<<Jd}-cL<kT-Z5{
zF3(!<rs_q28A@cIdgyS-9-w!~K20vwPK0lnr8pFTcC6C2VMns1h-Ip0Jk~`F&-Dsa
zSvZ|6aBCeCPC+8*nOWpF;TOvB;yQtpD`{>MJ!9`nM<ym5%b{0LjdUphsqA*-IWC>A
z<t~4`&@$PuT8Gygi0WT-Yq^k*r(r@7>NCblf`8)HRD$Qgf$W%BLuU$=okT96S^v*K
z0I8EOhfs)(xU70xv0HAnjuglJmuME|gwLbula_VlTvx6pgxpR%2LSX7ff@Jq3i)Up
zGuC>pPOWFggZjSu9m9oN%ZAG5|2(dZ<g6&!wbd~0Rz=Vg;BC<B?7ncR8(o#mAnVq<
zA=XBY>zOJ9>}M48c>Zu1xNoFr-r3qWFOOCSt)AZNdrvPEGup@e=jqksfGkcH*1dbP
z(%^uk<jebLtJTlo__^Cwxf5AWk@3s~d^lP}!Tw*DVS!PMuUZOFB!Q{`v&YnV7}?k0
zsQsfq8V%Xl>MPa~(ZfI~Br5c{6nwf75#-p)(bcs8>BDV~*Y~Py%7)dLgk$Z8Y3@+2
zVcuC0T9KasbW|g7*<41mm$_j(&jU)RS|)d_W2#(6-t&qLj9IMC?`XJ;xuKiq0dyhE
zi!%Yg_}74rQvDH*IW%a&p|J8PAJe+rEswiG(b}}SKg05jt5{Q7A0YrqtI6rp{Mc4l
zF<p%qQ}sz$IM4O?bo!$2!H7eji`A)tuj;M=y|j*)MdpS&X1)2|_7UHQPt<T@*Q5FR
zMZ%IE>t4&D$By2<2=TV)nol6%+QB>pyLVQesAt33y?D!%*gZwkYhChe^_FBH?)!EW
zMZsMAS=Og<|IdXg8BS*`9FAT4j8l2>6;H?7z3MJGHT*r}ilyycAG@89)nzp0p_YjX
zcHNS}?l(4f65wL0{jZb(wTU1axJCDL{4&=^yu(^hXGSqq`#a0U`h`ZUY&8<2VAaH3
z9&t~Rt&A~Mu_G`kDvp(4K4(PUsihH7`HyuG{PKoxs9#2Ft<CDMU&7ZjiIY<~;)~2Y
z5P}k~!1GMWg=^(opbdx6s`&eTG!pkt!TJ}c)2sCH1Zd#3)&TH0e8E)S!lTBS?I@)|
zZuKhiA?GKlBN>3Z=KvfeiQ?<brwQK`F?5>F$M)!W3-k<(J=O(ub1!c0=mxpIUpu|l
zUrGBr`@rY)WoH;N(nP4VIr&>d4{MhkG#nyfLcUyPY)Sez)Pt%NCyoII`ZJ5v6XjW1
z&g1%kPKrMH!Z^f4UI<M^hX~Z8op6f8CKYPG4KNu!yY&7g>~<4*HxQ4$gWW_l6BRF+
zW*?XUqD`(lW!4YjaYT^e<T8Y#oQ52Yka!UsGr0^KAXm?TVw6D*<Foi2&w^V(t84)*
zOLjTJgaqj0rQ`tQH?iYhn6xeQUGWKoZ6=z>HgE}r4$7Z63Ws9OB>=ofH|F~F>nd8_
zn7zvZ&mawy4x;K%CmYo_QPlI|aOt0!osF~j=%)wQmenA>s-l_kl3J<&EAFDK^u&C7
zA<T>l?K+yrMM$n67o{^DBU%cYB>^-pD}dzX%m&aDY6GU(vV<vIEtFX%&x8R-PXZY&
z!^{}cOkpt<PsfO4q`O``y?yh@zdU(*y&gktSS`o?8aT13m?}5yf$#e;HcC&nQK_n@
ze1C12B8;ttxNT>Ri&9a|ug~FaW_dY59xij(4FRcT8kuMq@VI(c_~8A9pkU!Kp^+I0
zZYpubG+F{VW%N)%6_u}mJmE6<j5E%Ktm{!v3QpjuWTEyx+}MjY6VQOXs=I)~9lX~&
z9=y>8$_p82c@|^>Jn0II!BL!zAf7C}LR7LJh0O}egb}P4&;%yqNaDFr9!~}`2phww
z=0zJXqT2m#-8ywGZ}4jW1jY6|P;k`25d`xQX%eqYLe2^hxnvia1|ty1a$)9a2Q|b%
z#?eLkOztos$eh5>(n9xOj7!XI)Bvzd*~;zmzk&9V<y&SEn71~l-yKX-wf`FJ<*5>X
zFaEhwn?{=xmD)OyFe&rB6HnBzTo!OdOrtA7!gEkAO9hLJtkPPiqZ`=7ShmD0Vm<Kg
z8h?NzJ`G(hE<s^X4amF&csOebZ6BNF<ZH$ro4#k7O&0CzyuV}0MM3*SXn93z{IZh!
z3X5pk(sKGpFYMa(4Q48cVj(!Rok2oghw6ZbPf}739WLj21I^+vAB^xtt_bU8Mg!D*
zr!7Y-G(yO<b5g!yaSt|{^}+;)=A5Elx^69hf~E_ok?0*CYMbu`wgKkJAy)5wCktX-
zlLIg&{nAkKoFLXUh`VV(RAaaX?*_MQR8UZ$auB>>YEu)Jihv4Kb&*3Nz+V}tGTMq@
z+3f$J53T=96zqg15cvUDO2jxgH1-agsA8*m#N65Nh&}(}CoG|uH@%Bb@Bd3ddpZp~
zV@$ggyh?**1sL@pHoBAqP$)}rp?s+ANE}v1sY&(H0c-!i0M%6JX#msCMk!g+zcM%4
z@wQn6n{qw(RW%~CN3P$bOazuOrZFtz`&Zy1+pjk`1x(p-euY>!2wM4Z4p{$KVuj@h
zfA5^xvr9q3g8|v-xEJ^avSwyLaLCFlZVXI~W+ahdd*C7<8!6qgMC2(Lvf1Wisg07w
zCfX+FVkeNova&U?^m@GKOykCYY@rvKr53EZeDD$CbO92UGV7&JpU648jPqCIu&tm9
zN1#<&R)dayuhxuM78rCx*wb!v6t>er{Zg#I0cHtx`F#D+XjOW#N+`c~0w)UVnSdDC
zA-8$uk6bAW9lVHYA-FHD2jqbqG#JXlC9O@)v7KOEZ4a%s0X|5!*%jIv(6=(-NKO#f
z3nL!AR{@u@MInx_(oaViIDbqy4iCe45zpAT2k!D+OL6cFGg^wo9a}~X?_$O8+-Y5&
z2QRy_A@sa$w`Np8E=?%v_4<5UcMvGlxcSxj;3-9rkYpFQk<(z~rAn`005WFW3lr>W
zYku3nE?V_k?y$Apcn~&H8fsO^t~G1fPJ0GatT}|Sg>x;_R5%SC9l|qviNQr1_g-QF
z$p3!pB?^}$&{#UUxBIW#Zgo-a!)kMm<cNa%!JZ3dQ7?WFv+RyW-le4{qC>1vp-Rpw
zwH{O&;N$@7GNanNGXd=?|3Ou&8!_V&SZJ-7-n1<B1&09Jb=R(+kUL9MtS75jSVTDt
zsvd}z##iYAhHR7Uyx-uF#Lc%yS^j>l@lpu-mAWWM<C*nHEwJz$JSK(=vS=6q<!FCJ
z#!^Uv!@fY6O5tq?PGI{CPo9mzEf+2r$C@S~3&rE}u-z@M*zVvstq5*r-T3co@J@<L
zBuj0v<B*Ppvb=HRGy_dJT|7!IhaekCHoD_`ncl~P==5%ZNReg!5-f8M#wYIvUV-|$
zu(JpVHYGg{3q+7W)3Pq~iQBs>r22OO(RGhIp_W{4@u=ai(H+46`~b4I+?V`i@ZV9q
zn`t;cYR3Qk#tR3B<&6CN6@#GrV#a5hR}X{v72NQ(XATmGg#MKuVNL_kDDK(e7;Wnr
z)3UgS85>@bnw#W|s16X~3`2(dSf^#!UTJ9~3?vknI|A%e_1H%UVP+3EYHbIs;Nc;}
zy`pm2rN8}o`~`$OEFwMNmWG%A%*qQDahqKbUoP$^&aicp9E2VS+#SkAMxl-Msy3H;
z_FzO(Og1<VcBR7e?yzs$+Y%HHxIp?=ZVvKc=2uIYdAE_)v9ls(6;@K*_}p|l7}D{u
zEuB%9@#ID}Lqjy+k2(z6zq}pW%y=mp*LURyxfx^Sgy4?THH(2Vxkgw~nOtUt?um`(
zv7tg+A9#DADcUsBAFW((pzgeMCNjMGQ|QtZNtWqNjLDt+TdDa{;{iPZ)c%ajS?DsB
zF#IwavROSw8(|IWM!<8&GA5si*TMJf5HR~Q5%+$U`!5&Z%o*lly!xHj=Y3Vo3sFK6
ze+yxON|?D*UcLTFlkw^*ug~9HLg#r0SKKzNvU5|97=-CT{Y+TBZXfd}Bjp>5cT*=^
z+1H8aR0qZY0=L^LwiWHrEK!r=4SkpK(M;&i#I1jDA=SIyaDL9K!6j!N8N8idh%$x0
zjJA{HMQThffC8kyWs3+iDKtX|3W>v@I<^N=zI*IP)V4=SY0mAj6i3u6DD~ffNQL%W
z$Qd>+!F_;*`e)xK_eP~!6!y)Rb#f04xi#aYHpnsxi(P(-S@F&>1%#YpMy4JhvEsT5
z45K;4NMPGUaBZ9gv?{gZ!svI3B53Z~k7{&>-Nit6x<T$8mbc#-gLMfPkK=&1*W0WK
z_&o`0Hij$ilZ?}J3B>u1$331=rjl{|h`==ib<tpJo;Muby?N4_pR?K{)s*>zMBf*B
z=m*L#Q{Mhdmn$(zMQLGBp#VlYINyVlALf#Pf*FGuRV2>bz%6frq^6bTX3$$AehqRE
z4F<dQ2U9>aI@A{d-M|%-7MS)P!pU$wIJj5^Tr{cr71%}&;hDOfn^WPR_js*_ZpMKP
za&VOtK({%4Y5@2a=zaIfG51o^_LaP{Oz`NE@8qu{R-9edk6+ZC!ND=D5<`b9p|Jo5
z3%wEA$X0deXx6hGa5i;01+4Z~LM0gU#O56xN3S^A*cGU*)2p<I%iC&_L*1>Ga3V6T
z=O%h{pnuVM#<YR!DX~{+81vp9-u!u>RO<TIzhAqS?s6nIkGy*UU>KXWgz}$yP!8?m
z=r>!3kSkcaFJiZ9ho7<v`!q40_AIk6egXS)Qt81^WBj5WI@FX9^5bN0W*G=3Wtf!Y
zfo*gf4^1_cpjX7G{oq_S$xa23DHIr1E!MJc12Q%+Q^fVk(72*54j6kQ7$czh6W}sd
zo7;(WGhC7JklX#)9xoAJXJu|@*k~8Vw7oJckDOkPOUR8q%a!9HfP?PC+nTN_QBXty
zQt_E-jiUn86LHa|01gIg<y+1tS7-I)FE9aqF&9yl?FPV6w!xyq>-|%$@`3Xa0<|9;
z@vT?Z2hS8C#R=RY)of#oj79<k-7V{8HNM3uwAHa!k+-RLM3lpPuQ{`Iw^gZtpv7MI
z&f?^N?Z^%{ZQ3M`O21_(!6afB>xss&C@#4_L=H~v(=7_on%GSULdgB~fxlQ910r@t
z3vIYu%zS%P>SBl}2jb$o9fzNbzJ3MX9@xM#05COBm{4O4V&UP#7azy}8m0VJ!u`!8
z+8gH)49MA)Y`FpL4B_qRU5VrwIb6FvH&$RQxps{c4UJF+U+%swgXmSmA2BqWdj3Mj
zzGOw;J}|tdkx6xBCLBWYj24nAP#<-z2SE#aOk>J#1Egbm-0$KxM5OQ+X!lH4a7Gbo
zRO*seSsI<SN-fiht3|0w;{L;b(W;wiB`vQ*%Re0OxKhCG&b?c+M=7z~NqkSbPHGS}
z$uPRrn8)9Qo0>+2OH)5lF7A=#5Uf$5)+!ELPRU!HZyP5Tz&GIl5CLEAv3t8%YIssD
zYpk`sxJ)ZN+iZ*hNVP2^tVr7(cHb*^?iRX{<<|t}k4S1R0NxmLyMyrbv|x%A1I8Wl
z5+r=LxF+_iqhdF9<wHtcX;{6Zu{GxMUPVRAL}^!k#)dYjg*Zz|zccR;uO(P3h>)>v
zUzdy^An3rMd&rRn#t<h}8L7+;e)}M+^;IjXKHLYlPY-1G&7b{!je2r8>)dAUO;6Y6
z#_~{43-{BCoW!A{2!qP?0r`0EG^U5GkPM7q1bffDY^^?Cq(vdpYxM+KL(FMCM?uz|
z9hKn}Br_NyAmZD6Nf31jL+nrH+~}WHA7T$#6ju4lz%Z@OW)WMNh^Y~LI@M#coKHBj
zyQg-fTm!on+m5y1u0dX&<A-jvht}9QIXSIw3h5K@xxc5vq7cfNzLFwOqE7jw5#QC2
z^Q!ety>eG>^}3e1(txYcwJ3-#R;fK}xg>4noa!yGr$%iqmPqB;Rc>?mc_)x!lq=RD
zN(E2{8-%<1@nu<$$@2(Op9#qIJGZG-fC_E6j!B$nRr~%RP<}e6L(cv#DMW|JF({J;
zf#!GztfXgoWBB0MgJ*t&4CPMDZRsHkK8c5pL4x7g35k@YknE~{(EeM{rYS_@rTERI
zTAAF~zVdkSc}IJchz|XUCQx211%eg4{1kPxEflaMzq%$2>NO&REr#j%(1jfiR<mA=
z>@)K{YooU=2-IYpT_YA{+%ZqYH#w~vcg$9=S^K%+uYCYnf0_z6W>9K}yYqqgWB_&~
zet_Hd<HG0=;Vn@e;t8}t!sXq`vivK|Li!3E;<#v|>#MbuvnB3%Eyy~b^OMoP-(~$)
zIY^or+7vpf43y_SF_7}TvDm@)_P9Rj;-XRJy#Da;$qRFiuO}uZE<BVG)AgJxjy2?Z
zdhrPWG<EJ}0>1IW2SS`Ngm4$zJkITbJcM${FdPKMm~`icXNH;$vT^Ka*mewigWsh%
z$;v+$TnUIKt)$5+(jW_bbNX@+W-@`GcaCa;VUDjOBm@PC00J~FN!35rs0<+(RU<I+
z)hX|8pRS$p-_<P@Y;tX0jy)&kT;|@GQ)LzG8OqII>5OKNN}YZBL!7^$Nk}KLRqfVu
zmn!9465LZ`Q>%bKc8FV4uO%qS?G(``8%J>V=~;t*cRXtNh}xBas{gU(+*2$YjxD~d
zsC{wvCqhIk{X0yF!w)`hdx2mUOnA|xwMfVFt@<tD??B0HAO{9=Tg=dk^WS`fCHyYf
zMF?5R0+g)_0bvri+N}lWr7i}mY5Nj$i1bMVZLNDjS(rw!umyI^cipgjV+OD0#F&A4
z-L0h%>S42%BLSo_FKv=o*u&b45naSr58jEzKawUQ?ysH`)B8v4;!%CKy@9H9WkTPj
zp@ncI<u#X#(s2Ai<BwA4VZ@IXysbj?lxaK!t%pb#V93w$=svT`yYIiS-jh*`R(=Lj
zAH5AReHeJb73m-p6N6(bWM?S<8vff9&|u~IGhVXZT)e(3;8yb;KKv?6G&FDSmy<?(
zC;$A#x$o(j;;`zj{8ADFqZqBeuN73u&*apxPdgev4%TJ<ie)Ro8XOM@eP*S9@!YJ0
z<1?$6o(<A0)g7k#fXe=M+jm!vs$Uu?YkBB}S{H<mzL1f;j<vr5Z%TV6BZfKhQjiw}
z=Yqk79Ki%@n$fjANNlVIo+AxPO;4CjR?G~IU9_{h-j;W+DH{q+N9l>qA`w9&36p@P
zXm7Ry7t3lM@2N1xc7KireGX22rcPiA=tPEkzug>{kM4l|EzUq=Bwi!~5stzx!?q9K
zr=vTFw~>YgEtKNQJD^!Q_Ujt1=V%bIvUToECvdioc8-l1jDg?F0R2V6Yt}Af3|smI
zq?P;Je9{8jz}DNbT-_pIHie@Ysb$u3V?W=VQRcQEe~fy=*h5_pS36`5sx=#1cJBEA
ztVrT&jO>l*At7^S!;fMmEe|4M%$d^#sFPb5s+3Ry3QH{wpt%eHjk$(`yjmC_AwVGN
zt(@Z3|LEj^E@+f%{zv~3qW?^@s&EQ_6A;Z;;jgXdj29!rOWqU$QAz_OiLP}!GJbOO
z9_udA@m*OeW>i#GRth>blfPvaaVmycWyMH14s@Z*k_5KPUF5|4IM+a^%s`LGVBJkF
z8r;q~HKAvS8dd#)J&5sSwr*_66A>w(zHjhdr#B1|Fl5ss(2qix%4rzdS7a7Za*IEx
zXeCT5mI~+B*aqn>?aqF@6!6a|`Kn7~X}o3o_L0e>qeR&AbbR?P%4HifIdJgV{pMA%
zJ(Gh?IcOGO8dIrM=0+>6el)<BKtzi)Oo6l$AqKXnSckNm00S5y!^$nEz6Zf;7z~n<
zbwIh~5H*9tvou0}$b()x=I82dd?$mx16)PH#G!q}>~E9;m^l~%CMvlY>Z8ze5?(Es
zs;yye9>WJ~W1ORFUdqN!9{EwrTh0{aCO<nx%ZFfJ3`6Mq`cyW)`Fe1p5?+52mSx!@
zJv<zWQf-!Nw@a!4DidM#VsaeuqGiv9L|z`A=x2Ur=mF~vCbFZoJ6yXju@%u-unS2_
z7LelP!%ZMB8MsDAps)d_=AfWLnwOlO)1bU9n_t_~lL#sz9WN+U*?jhp$E#Kc7Jrk6
zW+zbHwjz%&0CuhHu5b|!C9Pof<AZiLj<C^?vC&y@y68cs3Jj+3=!K0ZC}zt)f34Dx
zC~ihfSDGSvsg^mnDgR+TR-x<sq-Bo$yMqAe@VGQ3znV~(@7r8F<-Dck7JeLtC!9Yz
zexaCa1h$2o=zFOYYStcdn+1&n^U>P398W~E<O)MK+)_HGc4aXH6{dEzLvJWrxx1`H
z6&qZB6ww(Ncz4SS%E0!WB`g{Xx{LFXnIQl~OdXL=58Ox{hQ@C5$E~V8j*x3K_eKt!
zw!q@ymhHscT1&y4PUf2-AD*FQ+qW0>X^bIBal;JeR6@645Q~6fOj+G4&DKnmH$`Y<
z&=5^Z0w{F_pT<DuC$$Tt$W22#!w{xRO(Rpgl*8r}F-YI-!no%ZpTsr`=nW<Kq)`Vd
zi}v5Y-E60)`)PKivux^P^t0s;D-*Rp-~9U8&Ks}7{9Y+n>6z{?o_#4d?&Es&#db#N
zbM^BX)`D3=M%T0D8Etn;d)mrugwvfi8#8Ra{pse;-Fc=D6+Vdus}0<W5erk^TU(wG
zT;m}tdamS|>t3<J@<EgKMQPQ@!)<%tUK<O+P~d+3I|ZpVww@yDV19M2e()cY#|!L@
zpR<ieDY(jp_&ot1krFrhdU<(`$g`N!>S=jNH64>^@Ufc}r=vF;hPHDZGG<{?3&9A>
zR34O!t8Q0shg?nBM1yZJ7(s~DXAgy_9ODrbY<%sVlALTHx1<!TtwQtAD1B9b%y5N1
z#Q}!}X&GrR#zru1q*hRKZGA8SIk+@HK5ds3A4KsYw75i3oIX^NTm?7px4Io$vv6{h
z)+1-z4Fa0`rMkw(ttg@@lMLHkilk%Vi1_kN`Osa6X0uD!gf|{bCt?|7z;&Vm-^52o
zGQAu?1CLjoef=#p(QmY)%6>U^;A|D<<}E&87~w_fNTj{2n5RZVZEdbscE&fj%Y)_P
zxM!Yy`)WQJjYtc;`Q&wy(?TM|HvU`WEW&(ze4^k|Ont)Md5T1~TRXKmH@5a{C<H${
z#<m=IW0JY|&8xw?E)BwWmZXL;SusEH>C<KE#@$NSXZ-nN#x>o`<=dbVf=RXf*a@}F
zml$M(#BJ_C9&V$I^L&O)=H5HpIT0>yZsL0IXSkBgWU3(F6qr{^=gRtt``A_MlX4{s
zM*H3v&J`cK8iW7Y?XoKMi>|#$G?O>Cbd6A-S;8x+pWhcN+!?yI?|QbEK+^s!3VR{d
z<W0M_x%_ee{^2307R<Y9Xarq`rm03$v3-~1jS*F#SC*iImX7GB8a!zmQ`+4&!F9?5
zHh{ITJnDnmXBhPi%{$PQ`a4n@Ly%9A$2I0~gMEi~Y{2c*nZuiNN*3g;hYZw{J&P5A
z^bX5E&bnhE0Vz6V4UMAMQw9d_TIGC=MIy2c)cTC>^Z{dv=P$d{VH4c&?3vqPy%8KV
zqd24ShIYxzCoFy9Fz0C1<@ski`sBCNVmx1gg3?%go7-@!vU~EA<~<&mPJ4cGJ|-q+
z5_B*&BmFAps*?1{cMH&TDgi6BF-O?s{>j&x+|4=*wMUURZxa??EH=W3o?yZz^Hx+E
z{lF_k3&wE@9^gnTa4=e2STG7G`Uq^a4sE_2+4_vz?XOQm4;Kp;Wn~>&WhUBsC=py0
zZwnnb5Keqs_b${yU84s<qkH7s-eibbflUZ3C8M~ea5SpwE*N4+G?F$IieNjRxf|E7
zyKGnB^7(kiKlnDkjC0@B%ZOO9_2^3R5=WtQlolE!CU+sjd|~1s8q<f$ISN$}n@zcd
z%0!v1XiPF1XF^??>W5Gj+}Nj>lA2nNMpQii2vF`&A{=}wMM*sgSl8HnDeqd)35y7{
zV>&LRZrIrVgr6GfXDuz+jpo1}zIQ3B{jiB!D<f&*r=huyguJA$^#8cjcvjR<Vl;|V
zq%bc$s15CmWE>-YMJ*%c=Yn5EpQ@U?*f%yNh<Z<N1nw=5-ne>5-zYH#j#>qJ-$g^q
zD%&Qxxm&Gh$anWgtL@ELQMJZM+aZPH#hD1-wWyt3i(Qf8?}mRy@jIb7OcL}B3~=!8
z6A(~E7m^p8z!1EkE)#dmtCW5)a%FqmeHCe7Jz!Ufs2A28Z6O&suDc^ebUZyvO&(nC
zI5`igVc)gwe?-?xnlsa{L|=<mzOk=LjTcGB9u>YuOAMgzZ~s#Ty}fV{c8AvYghfa1
zS6KdNQL>tcA47qzB!+RLov1Nq&z?POqH*TTZCO}2ZJRm~%w<1yjiZfE&&gFLRB5KH
zW>95=?E7D6p6yeoX>KQaD7*>^y`M@!KkzG(0nq{*Sh`$6zjt3C2Jf$uG23v_Ponkp
zXGN7MQH)}o$b^<~u`<}krrMEN(Ky-V^1z&C^+V0KG}i>XmgO0k?`lGdVXu$1|MuIL
zAT5=kxUEACoAT$MN8DMMmmIFPAz-tz<|JdtyZV~-Nmj_e8qK_yz;#_0_-GbwMq}|R
zXk?hUvzqsOpiECxOl&Mt0m*9Ip~X7>rM^1=FW7ss@02*7eLA_Hb<eioi!pR4-8^0L
zkfs<Gi<A^l3++u|&)|8KxC0Nr|Ni?d4xWt771dX!r(Lix)&7`R0rc+Hu{jjA^`B*S
z2tt$KM=86x6?vFxX@x+gvUOno4$j!-l8fh}TAK`uf&@fFv<RhZ{Qbe3H|DtS5F~sm
zAm^sw%eEyQ9M)2)MV>FQPw|9a=6d3)u&QftE}cSdcZ+`xOUN0m@sXgPF+eXk4}A@j
zNHqhaAuTP{8|XxE&To8KUS8e@`Rx4YIXs1Wq)?pv=fEA-<5XvA#M?}j4^B)7WAaGy
zp=79%iHV6F<CK@l)(p57)uUz`&5nwQu>A0QHgZhX>Q03GtGRQ3qvbdWGE%2$&at7l
zKc-24Fhb=*tSDC7y{mD~7cZ`RqP1oNS3!s2qe!IPY{LhjW_w~D8@MqC({gt_97cL7
z%8q|Ignn<|^GDw>t_24NSApU$)FLxjrj0{eEk^VSuVu$|Yi%8!eyH{OGim~Gk#(5x
z&s^kOw{~qfw9Oy!&Ro<pH@^dR<*|zx-RMe2w+19MjvMPucmaU#L%T4PrpN!Pti(V7
z-?k6w3E|;eSeF_@db5fklB@~X_}rzunWf<C*YgbKXjZPwx1y;S`tpr0LVr!KYipO@
zSRb8s|Ng6!YyPI57gX3SAO5wXV7|BOYD_vfH0FB+BQ}aL{)7HUvs}(4FRyx(Xz~2T
z_wO6fu}Rp?xHJM<IL^tEPq)+#ypzVV&R079^x3mW0Ho|1rw@pLu7)=+^3VYdZ6be2
z7N<GnUeD+XN$--A*U$KFEW~mp@kH02RIoN%mrcAATf4f9NrK7;;xb>vU29>)n|YnJ
ztg_kt=DoYBCQ$ve=)jfMD+Y4#Le4+$s)Vz%^I1&NNN)a}`WolozqsBEo)$aU`vZAU
zIagLvx(Q~qzS!#e`uZ@y+_kU++5-}TD8z2&0c?+riy<Z*(rfXG)o~4rixU7B`?<+L
z`qv7NOl22iK3g=d(hUVj`V<*XKTgXQ`<tXE0|POEf%ad2{dE$U^>Cy|i~6CM#4Xwe
z2KxaL+`DgW`VuSB5{=09ZDrG&ahcA2s255#qnz|kobbbr(FM!P;(TijHvf&(SFIWP
zO+&|iousZ+a!Sgzr-fRpLBik%df#s9i&l{+#z|A>id{$alvPwhF(s(w9*-trE$F4E
zrm*U~u+2@y+bhP3x_Cf5!J^r_dpC}na2!-76eZ>5s~xrz8JhsJuIG|;wrteKumc|q
z)BNh(j-g)Sg?Lzgh&^7UGbmDJmlIQ<d-fEMNf;@@rILDr*GSX2G%4}{8oGHXB&?B-
zVT!UAt?xdR&ZYe^YI|d#vRyL5FhdhP%Du-!k$BpLnRo9hArK0Jt{4e${Myy4KLDn#
zZD<G&2v}QNTkCW*kTV}Tz^v8Dxa{-KggrbXBe9QkH*MZLHhxGzpngCnY1DP~Jxj1}
zkf?2A_2a^cYh{SRyVs(s?!V_-BG-wXvX>bDsSC7RJ``Ue0NJiTEsTB*s3QWu=kQeN
z!VFFy6q56lI>$j(YTVZ!L3-2;-&SG5FA~Bm%>(m9u_&7VT}}`z|2%LeH}b8#y9X9h
z1e!G?dlk2C+g1na6A!kfMALQ<Y#r==+_-TUSi1wO+c7Ho6sbfqn%S`<gMEEZDa``s
z)^DU66$_=uC7X{Q54^bu4?iUU_;7aUdcJ^rE_f%)wsk9G*Gyij%eh4Sv|4I9Pl;Uu
z4~Oi=>f3fXp=(WTbyl?0o;^p2Uv<L+dGiB|UQP!4emZ(A+k3TR*_h-!s^hHCnQ0SD
z7vRq2D3OD{({me-j*)iA_G_lf@O6rTP24!Oi2R8Etgj_0c~3r*nRr)ia02v~dcdUp
zBuQoR09*#^uA(fUOR5Nm&|~>oY?*Hib!+BLwvLT}n8J|uvr^ZO1Db+J2tUk+o0;m{
z)4)o=*=-2g%pP`jc0cKnk}FAHefNsv_F*iuLvAfeKek{XsB!pPzay@%hiy#}yq{p3
ze){@197xLG3*OmoLE=Jp#F!f`6IWHh|L8l9H-aKzPHlo#FjdJroQja34oCMBM=fQp
zM=NuS=sa1(8R9yz^_AzFY7}eYi8hfpZ=RmnA}>dy&n6X~jEfymH5UNJD6Xqv$nEp-
zcB953u9$>CDWzT;B?ThcQK!ARe~zF$uLETx1t`&g<S$_UKSguL67aNJ*HsKN?<R8Z
z{a|aPaEP!W2tZxkMWs*pvmF#SrTNT#oCZiB>^#uG4u3PR%!Jj<eh{^`&DF5BX24}h
zmnOzFxIO{VumpAI4J6yj8sQXi=KQK3<gy=VE}-D92Sv1@x0i|_{OUWkvJiE90kdCq
z*#EmAmi=7qgrL_FW-<=6f9Mj$>x>bx@N4^o(n{6J%Ic5A-p!jf^#R3nzw}645Gy+`
z)3i+Q;=Ko`tm^=H^rxjJgknSkmNubCpoir{RQftkZ)BYL`&nP#StMM`N%Jn{*syf~
z8qXqCqu2^cK*(Qz{Y)Dd_iwYeN?*d#H!l2M#8soo=>vj@l;QZuxW#U$6%0%$V2_tz
z3fW$t?e6fF&y^|%TmT$+ce|yIj@r5FGybz?P29-l9XssYmv$ZXi*ASIQrPPysE9Yb
zd4;B<<9TjOVWtE3QaPpz8WD*gJShb|OcxFuT8QX{`+b2*NsbV&<%Y$N8;K(g%L2<>
zGvtn*5Qh&hE-n_0QA@{$^6Sl6fnfi(eoU$?4dd0i_3+i~kakhJb`=vDCqb&v$3E00
zK*}fwCmjhVnnFNrn=2YF;SL>JmmVfSxa^51p8ZL%D|9dmw{3(|!N){BM)Wv1V*DF(
zKXPBaN^&_?^Vk_;Ih(g{xBcL&q^g>4Foy$R7_Jq$`!S=hKh{&xHIU0pNip;e!qi_n
zCU&xgXiyD{@c}BZC>0G2J%HU>yYJwnNHX^6$P*J4wR@Q=i-0K6qywduyRSuONBf=6
zz@X?KkARLOuLn2KKrR_W)xy9DxfVhouxQWEMDs5V#3?rZ`kra5%C2sG06`MPH-dtS
zN?2HBE^{%rZe^A<0zK8myu4~qeb1u7$&-JUnW7%D=ZX{S@maKw;*7d_7r^QE&Iw|1
z8CN4l)Vr#|L?lWzSh?!e*gMqY)V{EGcnE@$GiR7U3X|C>*VEdn>#ncE=PGDLkv4PK
zy>i{5yST&*JZ*!_Qzj;vB>9F}``c=J!=6Z@$Qni>>^BR^HMbk~5aUZCX+um03BFuJ
z?%E2*YWlHbq-^nrzkghD$w+pY-S}IOTkgH8LFULnPRc5UC{Lw%S?Guz?z9akPS%?=
zymnK<_RQmel6C~%z<Fd}*ocnEftP&y_n)L09~$r5RN}@dRM7~=sQpE5q7dc~-F&NU
z>0c|0zAKv7t`0Y++#p_{u8PWaOtaiGn!wSg=aiL`lf&rNzB9V0Gwa<R77}tMmqpNd
zNe|}G4#-oRok5c@rI$Y3dE$v-i<2O)CJ@_iQ*KZj!5l|tgy(lak}93(H*fOI@CHcN
z4KwkAMeLxdv);ywTZcZ81hA!t-ABc0i^&R#LmK`YM$kWW%6-QCUS7Mfv6HC>NOh;C
zOc0-t*QTzGvKUz)m7Q1JZs275W$VxS$YAJ1gGTIl+pE#0F<kQ5FDNBt=-x%(N9MVq
zDeI1=n+cyY(i+P??@^TT<3*^*C5yB6^FGupWP)oVWnu>Qjg{f7xr@7!ib}KGEE4_F
zGEWU-lr*=SMzFZLO&J+eU3dm|*{5E#Yh`)swiaejFKyqb1?2K1%E~;0K^%}aKoLn)
z>6$@}T2W6<celxnUN#yg8yXr~-6L4@gIHU4blT`#OOM2T*{d<PZp1j0^lxL?+p~fT
zT*n^vM;f~nXQYXQWqBn%Qv*#_q@efKtsSIm!gAKu*1jdbSHCxZ^Tv$c8P9ap18kc&
zTTERF3k~fyJ)RH5bw5Di8w;fKio^0Q6flgkd=a~6@K7_5F&Nodh{o)=?v&x2sm(s7
z#F*zw+@sPdj@r2uhkYxc;`EO7kx@~pG^~n#rh`^08V<verVA?v(a^7S<-if_Q3jO#
zUA8-_XM@&$lr8P$Yd#`Di?&$h=;(N(i9Ea|>a#N!lM$hdO4~2PjwT)Wq-}KrvO8_k
zB#D*v$RiD@-3uZH9V1Z{vssg{U(yDeA3InN!p(M}nwK-yC<-KtSr{RmoSl_)Y0R7J
zdQB|v+QJtIJp(gN_&KFron_pHlQOEd)Xp1_1RUX1>%{KCjS@DajD?7Ahiz}dax1o|
zSzoqz8Gd-;#T5wUraqVZEi=ZRnHej{Zm#{T47*C5N5}A@WDFYg8l>{Gtqc2E7g<CG
zcpK;UigqIhPe-xg5D(Y1&5dyp^j0wrt&P&B$Y2&q#OYw_u=_lOpIVmm-)(6+impS=
z$8qPH6n+PWuzme{>zk)JGLIfT$^hAcf!dn7G7M-OX1=@z1uNuYZGvTYR@z~k!H+Sr
zbTcR+kP2)%{zWg)1umBv4F9k9)as&Y(M4|}?^QMsVSbSLK)o@PICAyX_z#RuylPz_
zc@R6Aoz+7^NB<7LT-bx{Z)~z<C^M-mD50Q=L&@)frb>ixj0#`M=dC-a;Z>`v!Z$-(
zZ1mV^N1Qn}C{5hvbYkf53(;k7gD*sCaK_n%L4_4-vD=lKL(^d8VD})ZU;XX{ebERr
z|8@kXh>=SsS<<`|FWKFotr9ykJvAkEN(YC+IgKxcWk~Zg9<x)omx3^}ntlr?E}`DA
zfjswteHQXXgaApUDOLF>-oX7EntpW!6X4KoIDX3+_Ckhs`bA-nXl4{{?dN?=QD09L
znh3D(0yly5HOD`|+sN8Y1*}(6?n*~tu}ztY`yHlHwkM^tp=p#-W=sP3$_6dZ@&i{x
ztVI2?f||eHl|s-|t_KC_(KiC>O~(_iLr!;l%Fxo~OW$|5{WFJ2?x+nIDdN8{aZ~!=
zzO^uW32B)^B(^tZ0^Qd3U_m}v5ySPk!jv0PQ(Y_)S60zzP;?BqjVBtD5)Y$5l%_L1
z9cdvo)l1v|NWZC;60S`+eP%DxK*#$5lP+t4<mtj6Hl8(A9MZcU^!uX2r=-MKK&FPl
z4w0PG>8RC3!)O@U_<?q(YbGOy=kVKCnWxk>nRQQ}K9?1aRom+dE9dW|zZ?uvJ1!a#
zLh)-HsbK6^*Wr7sM|_t;phdDr0<Yc$68JfI-=51DU3H+Rr-vV8OFDCQTano4+KOPK
zt1&hX4Z5@_Hi|{W2UMv=#H%rMKmy{`LAj?{9DvV=9Et9ZP-DcKKOPRJn?mU;y6-?l
zJm~;{43UXRcwNv}tnz-vnbVleCdiP|OUraiPWKFM+Gu3YVP<BQjv$^6@^v}@7k&&1
z@t@J`GwMpna${cCVA6Q`z`1+C%F+Qg(wFItT?U0Y@2Y0tUL8iOw{}Xn5$dWZ?kQ*S
z71>bnU?4@*10j(?iKnc;-XCZAQ^GyLFPCWgDd@6%eHdLQ3G=%&icWUDhIZhK%MSxJ
zunUKgEfkal*J&`dhdaG84_!!n0VA<a9DK^L-}>^(?ut<n*87MqI)#?wqFZ<1KS~vk
zAiTV+!{@dZhQtR1Y{JO*{<tK&NC}Vu$}BxSJV@|#e@ggKJW5m5lP^khP2HMRzsak?
zI?PJLZM!KBWCX0Z8X3i6TtifM5aqjZ|G<Ee#49#+W?fTLu}9_ovq&rQ7tiMBHMg|b
zyGcKfBSmTGK4ls^SnkyU9DWyVy+%)muA;`1?EUIYej%YEju=B*(NMZyk}gFEa{jnN
zx$z9T0vw%+miw@G?T<Z&Kh!C||I^-;Mm2eE-8gX_QEU-KPz#}G6uugb2nwx;K?adV
z5}AYw1WRxXauIQ$tyQmn3<(4T6hWDhDah>9D;5cZ7K~g3)GC9DgbFGmpm!g%YxVoq
z{ds@fbyxk-YJiaUeV^w%=j^@DKGyY%<1SsktQ>`hJ8kz_vSm5%*T4x^anN36o7NDb
zXa!E|tnwD8K8vyEKSH%#Vq08ng1k7JIB$|F{W?OQI_!Qn2gWJ!x7@<_UUan0#1_80
zb5OWyN^}o1AC+?;>O_Q2!wMm?qE*osPQ}FdJ~0ObL<(3Qj_q0fJqDv!glxLm=rGpw
z(Qfw`$(!@fozN!3IsEQGLWh7`^$=6#0)b!7$qIm;8FRP_Pi(}ltXC*_7&vsN=MH~A
z^)+-hI`5#;o4;MS3)~Zryxdd+CGx0Dg++6b4>eN%%2qTkbzYQ;Nz;ofdJ}$mppHHt
z0bqfqu=JN{@C3_1g{yS@`(fNM8O`I$;r_aUjZdFV6=1;LUb$#LTPBcaAJ0H#RPepX
z(}KB4VlUh~Uo$Xgh-%!#d$}AM+{Q9Aw9auZ_<Qm|H?r(bLsoPA%iJOEhm-*y>3Cf3
zhVTzN?OfUod)4|h3?jZbIPV>PX@?nYJ$|}#emMo*9CI3FsQ$on9<HG6)S8X_Fg`0=
zH#6)0<A5VMLr?cdRr~Ou8Y>R*ym&4uGrIisoT*c%UUb`rBRpVk*s^LT&ET_6i&7zc
z{C1&V#mjb7c9;)-sO|L~?0^RUxp1QU4nc956E`?26RVE%$#aDR7lJIe5(gx9^}qJ6
zC*~GW;CW1MIPEf98=GGa@=|iD(Gf*<2}frp`Qygij|vz!ZX6$kPfbMLrS)mwpT9JI
zw8U~c0iUg#XvuDeq9KwRCYJQScp*WGSB*nN2!Du2Jv}>0Taly_2rGosDGyQkq!3MU
z(#lG!W&~hNSE|e04;8)@c}Y?U%Y?^TCVTs&>C<D$<`E4g3xT&p;rM9_gphS1)d2@U
z<;sDH;+ukk59%n<2}0N}A<d`@ghj2Wd<Y{zw~Yi6VCRuuUAQYP#u+2Sc~FD#hn3(a
zmVjBfLUVv2{Fq1B#abb{3~B&EErhrz3ha$7sxMAcKln`WUWACD#BhN{@XKe)ny21k
z1C*U};QiexC#4kc**EMsrm1POumfFU?qvQv_~89JLrCQ!ag$EO#=bVX3{0vF22Da3
z_VuCdm*^aK!Su~jQ0`)7+hfR*5BfVBN_Kg+^>R@nff)Hp8Wq)nI=YXx{o7jbG~LE?
zSME+~49sIi^Vo1OYU@DHVP@sM(t2P|?SNB2mO;9F?3VxS7!VW?yT{U-gL$=>q^^F~
zJr4=B82#=FlnsVBt74EGwzBHKI0wFm0N_D@jVU<v1WmbcNsB`*KwQMZ`_x==!MVT^
z57Q75K?m7qdNZ}PS#-0I_t?RFDkla`L<&MbQ8PrrisE`lOyy+aL(nPaY6Q>r93)xR
z$dglWW~(=~pk)i*D2s6>ND-@DE?4D#L2#~vLkvFKW^(tgn2r_y6oe+0$_6BQOscK7
zcxYLZ+6Dqo{j-POW>(&~;pbuf`0?Wo3^bQf0cN7crXxujAJ0O%C__f<Y}9+ND0_)>
z{#S^*OP~?{YsPc<ra|Ln-5zFgKN5TY>*3+y<xOaDiNx|GvP1_(H<{Mm87R@CmC^ug
zEqn0Gk(W{6cbsf>RTx_S6j91fLC4o6z-Lf4X%Iewxt^P0VVhI76Hk#Qy!WAVO;O-o
z<%6tQv+rP_48Owkz?=LF?qxiBG6@gDfN0Etj`RTB7{$<P2gAJj228V57EnPan_k87
z*?=-5z1xHTWJE1)<sL7Crt!OTPL$uBHgB~vp#H!-O=My+<mWYLib{bvNxi88@NMUV
zyYLB9-?eKzP_0V)VvS`;2?UMaY5#NOjU`~ytTF_JonnB?d~ulj56<~s2#z)AUSx7O
ze2f}Pff-+U*9m+)S_|9nUvjzf{#{62hy~Xv);aauJxNxhxrI3nC@E>uQDA=tM4=$K
zEdshlT(JUVf)K?ZUynN*!Hy`}pTl(*16LK)%w7KN0z|nO5K#)$k2OVkC(1Xn#763^
zu+^ZoMt8Vvs1dMOG(wdE6uds@9Ux(YR;Y?jo6Xw{XQRul2m0#-{e^{vF!SNac*o$|
z8;nWt?g9_MeFnG+0LgB1EDn?|$Ic9b`+0=};0}E<rNaa@k8|hGZ^16?hQA>k4%R{8
zZH(+Z3ftl7>+5h}kkOG2a9{}##y%VfEeGb?{zDX@18+&%g)hD0v53q9FOSR8j^f#t
zRm`OL7@CD>rjiiUtJ>e(+{{C`wat)RLH1?0X3clU-Tj$QQ0msbLeH2!iVZlYL!|H&
z)K2mUY%sadUTr(1tEFX#)Kuv{0Pi0e6sQhXR*1ArC`{zeDBKqqI-HKg0=Izx3|mZF
zw6e(1#DrWq_3$g9di3@~$0+~$OCaa9)AG6Z^BSUgw75)Rs1LT(cF&%IU>B6uJM~M?
z0~lh$U@?|BQY<1;x6wNsi%MrbKD-4+sp;(iOh7O#z~+@9lx1^c0IQqs`z>j(=za)M
zVK{kbN&Leupk=bqIT1`pRm6OHKRet{1DLU-PP0d$B&)5puU06e2-Y^BtME`-lYOo>
zJlLoq!SCdMd!CGynQS%>WSCK>2~2R=Z1&z&x1&p;_5_T%HoyhfOa6I|=BNS5wzs^-
z+uJ)8ndVUiYcY;#hqD$&n1f0m7Mi_CbU}ipIlid_4TL{Es&GGt5I9pyE4^Y%^Q}F8
z@X~CspPiqYpYMogZh)T?!`th{z(6eh9L9F+92;i@)F}q@T`NV0P}d3}tvGtMkxO>2
z2ukhpK8!Lj!s2V(k{BCnNDHfz9aYW5(X}{e+j%6lw!liu6brKtzkjzPmP7^E1O6OJ
zkK_;>#okT5gx*pAlKyofQ5IO?D^gLrPHOb&FAsZwYwZBuG;_v`1@MlL;3Jv7YAfsl
z7$i+hG)5shbKX2-l;(n=W*D6hzSV-a@9hxaY6~z)S$gJ-5x#}jFrJ4o3}6aP@&-_~
z*n#$`%>!y@fGu0G4p;0VhL>kG_$xeDI#+EBuONl0!9%YEU50q#8~Dz6;lfBP-vm#H
zKJF&Zrjb|_@5-R*0^^y`g|^@F3tKDl^Ka{9V@TuQzyyw0Q=?O~$HU(aUZthvU@}~e
z!2AgHSWpEtE&#eu4hBLqf8@s>=iKWeMFptB80a+YP(CwZt(f)80D2VU(3$jiLw;W8
za0Q^O1Uh`huo(!Ob8BThTYB68aPY4B^bmR$Jo;a}clFJikvPX}JUkNVb3j1sdraG0
znCFSRHB(cwLvALD!$=OA;UVbvJot*gw)$Fe?naQV%l8iB?MTt|)HtZUWUUK|H!(bq
z`*+156M^wrGCXk1KDCZriCz{^X@16l@6Xh!cOsC=0IRsx4fOR%kpY{!#{z*S92@^5
z2pkPMe&{gX^ux|!FuL$scFP{bVJe5}mQK0r!g9;z-ot*-uQtH8YHf$k%H3Z<2X&zn
z_gsLk7st<<{?SKHZlKZE1~g+82q93Ex$8H*0m8vty7U*-eqgdvYz4k$*h`uDRS4Ct
z`&OkWO2f;$U19aZ#Ig{>c4`uMU??8l+DvNj@vubWJzGMqD##Cs=od1ddoi@7x4d_%
zSDL@bvexPKkYBI%^U0O=;mwyrN@!^Y#N`Z0igQO0%NyY5b*!6r#`|-(0qz+MK<y>r
zpgAcx;pV^!;v)Lr+yi&NGG8U}*DWlP`q?^_IOn6(lS4>2Dq3J;9;zZcG77FkEJ!US
z&{?cUC4L0zOmu~?QqBkQPw-on;ZA(j3z;<{m(R9s36G1w6HJ;oF%sQ?$D;TUEH4Do
zq9-=m+VUjS1}&WFBqFrRofhf0_czPb=wT?=q;YYJ(4#wZ)-2=&%M0Gle+~?jmXO}t
z-ZQIuqhfUIK%bsA54qdAf}FB%kK6{>8PemH`x}5^1&Z>VPHHNn9Iv9KrQBj2mn%|#
z=psOpJdE>UMbpx^Eo*8J#M;ql-}!(|_Xm+QVchv=Wq@Vj6F)z`=6PGhk`YtmY^)sh
zCW_j`VPPXm5jN?(BNNdU7vyKg?*|UQ-^Mu%eU%&zBI$~YH<UYqAlGUP+XN+03ff=<
z)g!ndF?tEaReqX330k%QiP_o0TCMdvcBWhum&KCD+E19hHBV#eJ!nJqs1lF&)HJI~
z?jsc`xp^(U<-@MVk>zEQ{1+9x`2J$EVMEcWE+o%TBrF-Y*_A8UB+XnclQ@hGHGgfU
zw${zpHw_k}4LV&h)B#QVJF~tp;}B+$E+FziISgfs9gioSD{6Nosut~?ZguZLI|%_g
z?aYuIWNvy9^7WJ^v(m9=^}A*)Ta(gRWzXRUo}A!s%a)~|)1B0D;~=k#$U>ZmMx9`O
zP2o27<8;B1EPZ&(v6il8GOh5Xu9w&1`P3=0XrMJ9LOGnZlHoN;n)xszi&Oh?`lF)*
zRMi(dC%P`z+pmtL0FFY?k95LLkFQybZjEd!mWBk9hf;+BZw%u!0;wKNxl#@zMNuFI
zwK?jSUPo<{(@9=ni3eZ1nB4ifNM2aj0iytUJen}{!DxGSY@Wz)G}6GO_~KGD3l=c;
zg_HV4l*qFst-8q_HB~-pjyl8bSNiK$DZz+YQ3{$Lk4AN)U<|Zwb)1{z75V#66m5f*
zE1eUAJLhB%A_t@rrXrb}u^l;p6bI?aU?>gx;J8|r$^C*x&KJaM&6siRW(mV_91%uv
z&fCuU5`cR&Z)b)EYZLYj(!F}GM5sm%0OSHE*q!EDRu3x+zgM4XXh^`loQ&0~>t`??
zLrCtnAtFx1@>;|AO$QAOl!Up~kyEFEzW()T982L4+V*og+xArHior9KC$GpY+In4$
zqPlXJCOQG7RvpzIqxr)`-O;L%!KsJFBNa)FpdS>N!?sVRCn@(8z#gj3FeX`qY2ufM
zfmsXs&%#|{(zIzg!;{A_SWcubgup=1zlq250S!Ty(ul;;04IRgee@k0Ad&=b3UluQ
z-WI>HQWs7F7zSQBjWd?vw5`%UG3WE&pG6|}3rS{fTmmt)UM!X;h>j?nip}K^l=LAL
zFV9?&D~pcy;pD;SYPD=KCp=>0gS#dp^jRv(6gifPDOusj5$R6=i<~R^QNb&I@d&D(
zyE?iv_}0*KE5EJL(*yb}Ub}<AkZnb=KN{<;1|R|~&*Ww-5so%f)v0SMzeYOf+w~gn
zT$a3o5gH!~QKqu}RA!pqtnrLnpI}v?2O$=mr*mz(uDlc7FOi53&XC~j@(4TLh!DVB
zw5Xc(dg_!ZR41D#Cuhw5D0^fiamkaPc4(V2bS|A4w2=C{k^htxtoY&_GC88rgEo(2
zIR5Pt#7?;PK|!2{YM8uz431}~JfXjKMJ3bovv!#SPUI9MFsfWNMv~fRzJ2Z-Q{Sx2
z;}gbQTLvioMjJnL?x&iX)sW$OAL@a)59LUpcbI!sOH0BjwRu1I{60mMV+*lb+m;Df
z5l>20<=yS&-MIll9Q-qDx!32EJJ;O06>&sANo_&gWLh?%lOs=`UV^M2<38RD4ONt_
z$jGJWixYLGh6JJCq)82T4$f|_q)IxZ3(^12_mawO?d_wfr`Yq8fpH9xc5dMA2^Q82
zhLg@ciK9oW;RMD$w)Ddc*9-<3=mu#zjzj6z)P@J1#06dZ&7JkpZU~eqsH5nz8Jb7^
zb=|+ag|VQT$NtKU)*7%G!$_>HX&c27>7)MPPpIOn4;bSgUZcN8{Q2_#yCZ+zlYg~-
z{)5)Z!^f)8jmXF<$tt`tg~1rR(#C3y6OM%ASFc{JpkcU<nJ`y<E)B$^&H5e(Q_$w0
za?tF%2xnH{=-y_v;pp07e5<yW>41iJ<=7ta?oU7DC!u13Q7W?WvyKMSbj|M224sYe
z0jwvDj&4_p@GmD*@`Wcqw$Y8qcqsNzeH=}Sum0hWliJ5U{fF1^*L~yvEicax+C2Kc
zjvVBfHb}EZFP>g90htvTq<MOlj!x}}(daiZrfVS*saQ#q*^5%6D;9rDL_U1&<9{{l
t{@bF*Lw!9zRqh9r`2Sog!|CGt>Bp}vUzM;Sz@ENhv(CXP*Ycb1{|T(lBlQ3P

literal 121898
zcmeFac{tU3`!>Axz3*!8>}ofPLR+PzNRlB_C8dEfBqfSO#t<@hYqvv&A}N_lG9;3W
z8Ma+0ie$>%CS=aE3=8WyuciC9_wRlGd57mXp7(i=d)>!zr)BkB-|y#heXij=&+FRW
zN7VK$n9n<(!C)*<-LI^HKQtMPxspH3#XG?#lb!L$T*HI=lo@RLUsP^v5QDLXp{l$~
z>rz;6lgphd;|b{*7n?I@WFl`}x$UFt!|TcG8ULezr25Sz8y}heL-6YT(K)K`K2-gr
zF7VLv@|yXBT9b<pELr#CB4M+*8@xaKux`z3!3}40@3aU^KRkBsoQb1JopI-AcG8%=
z_(12y^8T0E%>!8iLrc7s1#vHbz09&?tNH%=D}MStFX|3G$bY?F|L7C{Uw^?5m&5*7
zd|-lZGiCw5ob<W3caAolVK9zxJvot<ki7@DG~=N-GZ<hrH9lC)hb2#YKQ%ENZP}2{
z`5U9^I3B?4&LYLBtuv#^9@>s8Igg`t-*1&v$(-*QN7f%Wa3EplKk;r{MYonC=ePg$
ze=JYKP5<>5{IF)j#z+77`y00Z`2c4*{<99VR`LIE9he7Crd!~6-W=+#eyn4E?X`lT
z7+p}QZ#}zsL+>2=dCkdXn%^;=mizMU+|l>4e}`qgmZKN_f}s_tx0U`p5+wXT-8y%>
z+s|V#o+?UDeu-}Ta8S%ILf)n5must+Y&n_k`{2QY<9V)5a?>M~;^#j+Jf3c$8=UAi
z^kzeGOo%>jQ?~tehrar~spsAq_SL5i_P-b?TB<I$7-#W9;;9$u2Wo_}R%|)(>gGOv
z-EP_OPgkPtdul2k?7bdmkng^1*)nBSRaJ}n=fXRzn^%{Hhy~Odc*u8@M+#KNst4fP
zwZ<Cao?;c0%Om95N)2XgFLr%ce<ICn`IZyUy*xcVkLSC)C7#R>U9tK2&#5Nm{&yuz
ziY7;U9&f%K9iSjP_otOV$40t4U;Og^m-$>=ZC4g4cty^P#18)c{UWWpY+BALHQRzI
zsjSmkr|EBI@&9B<9vEoOUAp*B+pLnzbc?z`=72`Be(nc<#etmR5B>sMPb&!uF0rhC
zt~&MSB9D#31O3dthNX)5i-eDSWm^8%t77XSJZ2`g)iysruW9(@{e5p)`|ev=%;~X~
zNQWdNe;J{^@pJD&WB3%EhCXXd=MRSu+~V<Ys!BFc>~Xm-sAFtd_iVkmP3x;yyXNuT
zatN0wlU~~=SF7)0StMTjSVt_yrmg6f-Qh@u+_y5a4!w8o$T`_Q$jp>(8_UMhzKS~<
zy##Kvq`M^m>+|YMMfB1a4t)pe2WD6U$;?PiQJup%aPgr=AD^D7o?-tWT03*$%^dbM
zy{CE?ghKS5mbUwNz4BQuvTcUNg%!z)Q)4i0RC#ZRSecwZ<Lo-!V<47ly=&L5sqtb(
z|BC2+CdPgI^3JcwAD_&yGG<yfX4?4j%ga7Gba#n;_xPxl$NQ&eUXIsf<SoCxVvA2|
zxx!B4d>OlrSASgP(oA2pT6$}nWVr&X*R=lm`4XmTkWh?8X=&;80S`7S<#d7PmzI2w
zswXF<+U&Wofvgs``{onRH)cByL=0wLoIihl>zfUczW4WBd(=4W<m9BQOZR-aUS0G~
zf*`k^jBNq!s@8IzV~oq|vM+WmDVG~QwEEJgf3@>U!6>BJQ^O)@%#gg1m^%t?&fCqZ
z;`G`?b4Na`RCITZsZm?EB$!uy;<-iLHg`6wWnYNsDGer%#JNKy>9mc}i*%KhN4hFE
z7&pM=jEa1DR2#ah;+IxFI(+?Tw8~z0cXt)@^>&~BHLoa4#_oaRK=bm5wxx>fiSo9$
z66J{_a&&KNxZmcj;noxX_nj7MF&L^pAqetHS?S$Racr{JiK*d-(QP+<AGO;kfAop9
z@kM8|P5e;<E?2K!9lX~vCDh(^H-Ea%Iyx?HWBd3awRKN$wDc|%%n@!7aQ#}aQB+i|
zeg_`yOGAeBUGuXK^mCm#|IB}9f9Cvhn=+ZsSK2zKw9hHV-lVX1<HxMi%;j01hHi3E
zauc<Q{LSlU(f+cVvK{(PSoW}oLh^m_#5$&l^S-pFo_YD(c+-G8cdee?;V|d!c$?S1
zTvNsxtW{K0q(k?9&mup&GMNoUeyg~5I`r<BQ7`uAi`GfjS8YhQY*>+xh^^`0hx5MO
zvVOZ|n89?nPNn;rvmYL+3KyU9m>xaf8L`9i1YNn}`T0t$!mf+mRl-I7eD`d{makY*
zgoA$K&!EF1!{05Y1$`})^{Md}?#<jk;V(*k50L7t@q2eqs`%@lFWbJpToiyiFNQbb
z$saEiD_LC|y)WQEa<1#dQ07pHNVFS^dGyGUpZgjz0z{H?w%ryph;V4cIb0T^XRCU(
zrZLyW@oG_fD7XHlUz|s~m(SzBbXw1$&RE=`_v0NYYyHF%X`A4!mf^(t1Srl(H@ghv
zUh0mIHT?Riz@~`D<e}W4L~F~UKQt#M$~_{6Tr`g#U)eI%Ayrv)uYY^R9Ih3wJm+vR
zvm--7xD_YfsDCKidEwLVUuu&L-rkcEqYGmTIW96Ot0OeCBh*~Of9KHawWYo?OFbq^
z4amkSl1{zYxN)P`o<;K8okzYb9&5ByxA^$<{)gf3u^=~ge)d#V5gl*&?D-x38fQu2
z=HXMglrPP>F21~yn{ID8xdPkdRd}7dI0I(2P2hoJe`A)0xtLz&RysmK8`S;BT09he
zu3vA1PYZYxw4wON4PigC;Sk^-r7xVmTGNzfu5-00W*aiO*QKQmcF4m-E!o)a+uiIg
za~88t*Dv`24lzG_%!93ylw8Xtmh-c^!o=AUrWhVXxPk6M;@sQ)brJonQ&^<Va7X<T
zrq+g=%r6yd`xGL!86P-wXX`J<tW29WLuPNCaKk>|l{cSlS*kc*u&C8nqRgv>Ju{Yq
zJ%aiBk2Sl9b-ca1l-_IGr#N{urUUVgvBvTB9GwWguO6@Ggb2;r%i)J)FIN|@Re5#K
zI`8iDzJJZ*-@lXj>B^#{s(PJM_Sz+t%M_=+zFfK8^ho{ShKNgs%T}zo?Q$yi=CLPO
zu~-Z9mOQuTrWJz5EO=hcbhp9l2EIXRvcl&ZvN|JHEm*bVP<^^xXW6B%1@nT^!*MQz
zYp;7{c1E~-NZ+S?C|p+3xPd$G>s9HI_j^ji)>h{@HanSIEuwSzd+eJzL6)`!>|g$V
ztKPWeL%A4n!PaocX7AS4)?0mt@9lW<wLUqQc_L?OsPsHb4qh(BwBo_7zKMwmn^x})
zw_W;_1&{WNXZ4S_1q9q&yRWD!Uejo1Y9hs~I$`6cO+L8H{&h=YlJKiX;UZ1nkC;R_
z3H1p-k5|yDh*gigW%NirlE0@mS-rjn$LCgC&6AVd);WVJB~10H;YjwOh6pP^EmR_n
z7b_39^V3o@-n^a@B%JlBYS!*Pwor<_pLQ+<t8uFjX_1eQk4Ac|?6mmOa``bmm1+G;
ze_nmwca6{CcvdZwAulm+^#vt8r}p3!qyU%t(rR1bzr5V1Ckx5==)fYV?nOFMDsMg9
z@vbsGfJZ{exS^2SU^Q8veKP!%O@CJBqt!wEdj$DpFYd3aLBPyaSbd|{d1j)sGUuKb
z`8Z3LZDM)jhMNuYavIy{&e!l{o&Io>>-MZSbkI07hjD{~S84C?V6i|fMXSeO>j!ej
z^i`%yoW1V!pFOxK32{u^uH)@Es*zLo7YRSqJaXhlx?9ODVT#@lEwO37xb@&{M||_X
zD{~hFG7VDC9ex`oBVoKcc;e^%kC5YN_lpmAmM44er@MHqZCmhAkULC%Hc(V;r;;rY
zF?`Qmw-4dRPVgZ3zKd(Y5&MwtF?qJcLi*F?@1H4j&U94l6QF$VSZO^za@fW5OVq9_
z2Xg!2NlNnaSxo~E9zI-m{v`6ZO)E~zq*>HGYcx8Rq^CRQCVlOV$Dcieg7ENVLf<kZ
zUQUGS{g@K{R^f%?KvH0ug#4UmPoHkJWOcl|SDNd9uvgC%Ivje-rL6v;dZdC)$6b?S
z=071ms%DMJI`*$Vm3`6M%fsVxYj0|~noRlUH$j@|F|vc(8@gdf{o>2(YB}HRqj$m9
zw!pTaZrwL~kH*oA`fZ}uMDbp*``xg($@2?%x9@M5fh)~xw5t&cF>rUPXJWI9-buBr
zE*9``-G2VVI*Q213|x{Ltn+g}h;KfXy&V~#VvGfODrIHYK68WX2niedhKEYUijkOV
zl%*II`}?XFHq|C{UlE~ud7AuYPOYA8LAAOlRhE|%$V){1CewaPb6a=qlam?G`$}(b
zY3R|{jZI9);kR@N;Fm91n7CqR-}6e1^uCWNYt``K@i*%3OSo2S_Ch$?CwOyXb?o`d
zN4nldBVQ_nDTUxBvrurGeApY*zwU$jx(3C_<Ihcx#xI(yHFCi51gA)?a<DCs4AJ|V
z{|0p}s)3Ufo#W2Xx{^1lSD=KPvJ+z$WY%vZ6+Mz!0&XKm>zxARnICUiz6+JSJ~PdV
z<ukZ_>V-p$WikSBOh@toKVC^66+hk>i%1C*wfdzBYX#M0yB;RK+Gbvpc&X<}Mqs+6
zRipIWMf|Hx!tIZJK+c;{jBs3{XT<To{<7V1rnO0@&J}PfXH~4%p}$;}Wely?+)!y7
z!|m41a_3LtfEpoBJYQTMsi+8>O|PAnG_QHcCxvi_;PtTwM&y1dOj`UzhE=$Ot1ji?
z>C;|o{o|^pwBvDd&NmDX4_Brm-pQLUdcRf>$yd;`K4ycp`ox)Eu@HhgtQwDZOsUI{
zYL%oT|9<uQ=GxMQ+q{2Uu&Rg>foqCWXRa2}3MyG5Q5=3v88D~5N(%-WaGCeZN235g
z9RoMX$*}<e+@K$Pg$fduHdsl3j=t0SKGlbd7q=A$s5`UMYV{8ymumCBPRX4ON~3+#
zlcUj2!yPuwPMw8z<?`G}7|vX6Dc`We;^U3<>7*C^J`@y?9*N{0OSk;^^cMgW8tIX)
zLvIG_%H4OIZ}{_?Kt#olQSIW!CAV(HmI|?cezk!PRmBues*hCjrM1Q^3ePKdSR7Ze
z+pFw4GaMNJ6PKsRNm(OhxjX?>R9>z>AbMOhbMvg{xZ&0naOfY5sQT3oAJ2%R1hEY`
z`NF}n<XRdRx2W%EPwl?P$DU~TE5dSHuX5{e9Lv5-Mp(~WswVx5zH?X1L(7m;4!TqQ
zj`<>Ep-6HrWfdz=?W|$XME%j<^rEdj#K7OGDSLfi3o4O2@-7aIuEX!dt($X>XIPzD
zZTtEcn~J^Mg$tw_Zwy_m^QnJkeDF|&oJ<}Z(K)$u)d||-F3euvNued)nS({Ej3!6A
zR2znehFbGj{nzu{SVh17xRh?oD_BIQ2u4VYT#=yt_{+x>ql<m@l4^?6BVRtm94xXS
zJ5&<A$t8MNG;{l`?|Q93zANqRjbHW{A_q}zz@i(%C)-Sq)(*7A&GcD&uuk5(bxYZO
zW@f_t3#!FCL*F?ozgoO+_03VueU+5@>$&u$-3$o{Im#B+eDn$rED_-W;}b`!mC39(
z*JO6pr<v28w8>C{b-DKjSCXu4flJVA_}CGz*@X9x)W7mxs;R&bD^2#8?iqh6Fmofl
zyvebJZ|~l{%R|{FiY(o$MQ(<$B60Z51MiN87X=6ipa507TC`&M^1_#k6vL`7-QKLb
zge<&hTEsTH&%X8)cYxf`{;Xp9Vn)}j%c38TyqI+oGt7+q!a3Y}KLdvluFZ_sj4eVe
zDCx~Wn#3v4)~{dBQ=aQG*1uB0^-`l#XL!4h@ZCROUJGI6Rxc|oEVMziD#k%5kxW9Z
zQHHAR__MRBHAcMh&KFgtBV78kj^|vm$?9xDnylwKSV#-I&&I|ktC&t2<IU_XeT!gL
z6v51*bW_h2s%{8hCYC>Cd)G}0FhI+6za!_E!9f{sXA_#ht#@Hp4ts`W%shoJ40GzZ
zD~^K3w?nkFweUJsQ@+bZj-55mbF&<5Ev$$+7(&Uo%5-O0_^J(I7j~<(l=@tn?246;
z{UuFs!%cU=oBLn6jWux>^EvAIA~}EuI(FX)E@LGvpq}lU_2RalU(cVt>$IzS7v8P^
zuYw}jfHv?(<4Nn><+wpVO*NVIqFJ8<zaV%$H?I{9-C@aVUYq3S*y1J?ajDh&(wF<!
z2pMUQV8_Qs_>{Z#+g0Ryu%(fgw~@;@_T<D~cxtK~2*jY`kwA>H@SUlX&H7lx@y1RC
z?w0W)sn#&NGH<+JO!WxDod48RC|jhvBv_=aN>kmBs_^w8dP|yKI0g_R1LHA5FrXax
z&Q3=wIbb`3z;?9HZ$`2OlU=$<?yFLX0({!_ef&-jKbYqZm+ch4^!3w98QZf6No%pu
zD>j5*)PS283YD_vKODN<JA9{O5gwFEO>#5?c`U1(iH>`NQ^V!O$cc=``kR9NIh$>^
zH0x97K{y^x#H<$<ySaZ|!p<KUGi|F{##Q8)AB}+B9=Te!b0CLm?-J&}zMn^@n|VgL
zcL2rjYPcaEc)WF%kx2GO23kA>_RXmx<ehQEP7fZ=hCMx|<G503c_5ci!)MQb9eQ#+
zMR1|a=NsJ%=?2UaXL0W%GWVpc{o&VQEh6QdhVnS)ka2m_Y!=O^yN}SsVA@eqf>R}T
zzTug1X=|NHWK+g>oO5);{+r+ABYQoq=vaAJD#l{fP6vy1OT67yE+E5ZXz$`YCga&2
z@#PFg?={}-raxn;{NW>%0UiZ1>$2_5QLvm5*he)rLn-|=Cwwb0+jR(yRyv1zb``>n
zebF@5w4d_2<tl_%<D;{YUa5LW47HllmJux3t8#Z{>g;pkQDze)MtdwhirHNNc3)6V
zTQsCU$=<_xhA5wJUsp(J_vI6x1=!>J!rUjj;9{+{95>SUFtirWddaxVsqTeHdu!j`
z5nnqrG?cLOCo^v<W)*<mT^g;{X-g^Qzh2Dmve}`(QMc&~eS=b25SP%s+1!4O3)Pgr
zyHm|e=JbEOP3^~Ogf{-|^;GOXfIBV6e~5=!qxj!s9X`|1)q0Cm=I1$c=5Uy(>d){T
zepz;%$B2SbTE1%A+5OLv4Q0OskQSPQ1VP+wa^#zXT8?al6~D?V^B2U76CA?pu=_}~
z00ryItGAwgeS@+Z`)fG#l6SYBnI)lKKl|T=&LeO6*Xw@`rTG7md$@d}3$%!TPEL+|
zEl|+MJJ&Jq!$xjT&Un~10U@mB<2|zHluhAsm+7~raQ(;yz<N>;`*4G+43LkK(`*b_
zj0cMQ2f^r()4e7(HM%B+sP6dy_xk|rr}~Sa(Me|CDgx=!%xjN1#wl$<R+2Q1bp1u`
zG$6!RlC^r83l@LJx&nYpW@vc$U=vb+lRpV4<b(<$X^xvfEbT@}(fXBB5dDk*d3p0>
zL(9wxgPEa_wmU}t-UN?Tsw?tY=Z+q89sl#%Q9ZrYk9Cs$iM-)=`h4@>P%63q<G?lS
zuPv++3=%r*gAceRvszfqV-I#+nR~w29~e!S@S{jQ$}PZN7%wX>m?scs%OBfT!_Pd_
zR{>-n@MnwuSnT0@E5vd~G$fS~6_40qJtY67x0Mwrujw)y+h}cGdm<w%q_RkQ;&V_N
z5I2!p0E9(_$lw>F{yn~2<+wYRMs)O(;hVtq)qEcmI?aEdi359L^`b95w_nPvYUA96
ztB-e0AQ26_=)S)&qAs734b_1blrb#nB}nN*<9Bp;uKm2*ZhAV^BarLSqZAXS7$-@%
zjM8zFF_v=>xR1O$u)kZ`9!oDS>w$0QXaPRo+*f3o(X=8tXXpUfu^oVo7mMg5`4S)U
z{{Eimz1LT`Ou}DC(jKx|`Ro_TFXx%v^<KK62$T#n+oN!~4q@x5&x#26?D`O+d-r<d
z>GEDh)VRxH!yW4N$4uUZY9}rF5nEZM0@FDx$bIPWS$twPI>YB^mHhs%@2RqK(eLc+
zECz1KL#5w6t88JycL~$30?)J%@qWBSCvOQLKL3ms_no47<E`>-4g<}6FMnI`_-1V2
zKUmk0Y!>H^HQ#&Qm+&~Lq+XVa1SCktmI`KyF6Y>R-JOLfYjOkDB1?$Vg=K~A7QGW%
zf>q`0nv#{~p*c!dvHQaOpep2(QPsEIzsB!FUJ-E6JqL@PnkPn}hQ#3xHOA&GT)mQz
zw%}J>RaQh1u0MFY*FIk}R;>cnOk&a~$c92fc@xW2`_AE^MYF#=Fm22ngaN5X_G{^4
zBTUG4y!Ez9-jz>f-=K8xa^oY|`24yYCr)A+)8AxM%tvKvuz3XPuU9-KVJV%XIJs>u
zEiC<GhX1rJMWPzmJb<UKsXef6iMZh_uf+0$ea84kqCAkO*Y(VxI5Yx_QT;P`38m3M
zuDNEsS*OjmVOeH#&#c2p7ME{xpBf+k5*@(IQJ5KTYn$vz78{cTU?pzVm`PBG-2hO=
z0IDWLY--jab&o*aMpVvf`>LZAsIg6(bDUN?HrfobXIj|TVL{rCUe#2ek^NEj`C!kr
zg+!gi0skW|%U`DBPG+l*NGmNga2y@8aIlM8lJX|IrID6L1C*+Ob#dwrg1qHWsw@w+
zYcC15X>l7XMD@KA@TnKpg9CGbG9k#LnBpLT&M~e2arJJu_OCMp<StGu8=i}_8bqv+
zt1=3Z-|^c|JWIgWCXKkn&BqQ?$s=Uh16n*dQHBsFV1@~2`_+-U>o?k#ZmGa$!KCgX
z^ay6LiY~J@KyF)^B6}*cC+S72pHy>P;HV-p)bTEL4_+$Ft}K+@*j<|}Fi*hsLKp~H
zBCj_J3&+3DAwM$pd4r;NYMHEvzjUh?!129g@<5yT_6fLM)C0^x(1->*w=S};RDS81
z7OitF#C@tedjL>GDYHoq-2Lw}`$aRSK6{o5aqI2K`t<p;XZr|qo^1iEJdZ=3U0!{g
z5v9@vGZ7Rl$stJUi4DK=m-{iTcV+F(Vme-Y4m^0^ofPPqTW<NxwB)|7u|hF-@is|*
zw~-1BbE>RdM$aQ6s2H%>dG4%+mfN95RdGjS>aY9wEWtXdhS`<~pYMEAUas!DQe5z;
zhK5FmOMI8X6`%3yWcOg=R8@M?fJoBLtp{VHSr*dEdJ0dgobNte62_)e`-OD_Xuk&l
zq^0sM%tN2s+l`m8u*-MGL6?<=IU4jvvza0JQ|oP}28&Hb_|<@FzCztyD(RtpZ|j-e
zACAKs4U(MrfjHUrO^cq&UI!-;elZF7OzdhMEv;Ke<?`9n153kY?KkNhqVFI|g6EwF
z0$5C#$!K@AOP&zO%=5K?m5mJRo*DBzxl6Y+)h4jiVDdxAHg|R4mIoW1d!Bgn?Q}Ss
zQ<SGs$vh-@rsb?rOQJ!(LVNirFu9n<VeR0PR#`E6EuB9IBYeiG0`I$BF)#@KmRKYG
zSb{rje^swuoZf{(P(FP5dKb(9Qf7S~VLk~>H(Bx{OK27AoX49v0RoD7xQ5sM#N#Os
zvR{&m;C+`5PnESrKSvs*U|y#^V7S&lxWp2x!oHvT)5cNfnnc~;&}5-Hkp8aiA_l_4
zw@L#813AX|tW@+mEV3{9i`j242J&{!UBG*5@Qux?9Tr}&UD1xR@Oy*2#JLeUPJVD3
zOAXj3aeUiYPjpj-%4B;I2}&!9t(=DFzzd$G7%-DWH`_a*abNyayI8bIdBosYK3hI|
z27HPiLP#O_%fR%OJK{!yVq(_ik_g4V9m!s?5zd>srYqHUZZS8niccL&E=$gvG?N3R
zrF*h%ses3n%}*=0t|7O1{ORdkSw&?L@;lY!N2>r;?A+vL)`~-Dgp-(p^Za}o!j5IQ
zpqWKurnqH3SR;bK+8XW3{X+Deh1;YFQ6XhS+eoE)007>za!J{<$I11CG4n7m^JqZr
zstp>`ITRoKaFX_kW?I-57zc5wsf&&)ejYJfnDuGev1K+|^n-j+$^v?xT-Ad|B#1*q
z(gc`8H2&tI&u$Z7djUh&GAj#htv65h%V>5MA^aZ%ix0PftGEI%kdN^4fW;MmTgb-?
zO6m8A8G>8@-$wJUZO&*0EJVQLSdJ(-R8U5R2JFdNf-ZX-vm{8Qa`5(MZ+OT;;GF#6
zA2yE3A{t%?47oSgd6e_5^i3eMTH68?{jYLy7md~$7<99?lArG}V22J)ZHRQWZNwK1
zVx<a0CCv>%h^OG1m)iWKT7jZ@m%DaIxJ>3vGaJMKh}#w#XtPX#Q&J3+U(^79xoVm<
z7C<l&tRNcMOHHzYfP_hzFR1-Ko(wF;7r0U!j6A)8=ma@dZ>s8mN7q4vl6qcvYd_?Q
zx&1BR=7@Vse3A;lj{pPU?J0lQ_133}Smg{BWy1N0$Q3`yXyye3`6iPMoXO~w*J9y@
zrQXWKd2>WG(vRqA$s_nGT4**gt0TOhT7QWDlBYU4Y*ZE|v)msD^l6W^+<FSO1%M<?
zR79oavp#*m#?H<9^sM@`+gc90Ba@Ilqq><9*z-4EX(_I6_f9g#FsuFc$!FWF0p=Zi
z$ATqo=o?E*cC%EE6hnmLDQ6>i-VSAh^Ep)Kh<F<_DT+hdW)C(at#<4&mc*O1gTlpc
z?@DZ_0N|-H-Qv#byESb-SsNusd7nH^g64)nxVV_N;y9K2e_FL;vp87qTW*1F0z^Sv
zY7f@Y@ShlQ!O3&1m&RESq7^no+%PW6!PdRf$)Bv&7>;#GW*G_)kmmizAAiVCBNe=N
zN72K*eoXw#AHOQ=>o-S;>*(3+Ey*MBEq)wlulu>3P8Gbuy&;o!$V5scYh#UO2F$V2
z-sBw4H{b*_Lmy+`ggZBF+H`>DC8!LE$pNKDwFsoQZJ$sOdu(0}`A`b%^`Utayrk~Q
zPERh;jkRNGo2q}658D4xN;}26Kc7AGaF{r;6zsU|J8f+%EUm2vWS6sYAf9Rd{0o2n
zVRk(w+pfS{`CZx7k~tv(dKWqdOXqCZz#TRdSo!#b<eBnyOPGhXd+$yE!Yv1*#s!Wg
z9*1BP1&fkjhry$NL^Tprq>&5_*w@6jE+5Za3wF+hQ6|1UxpWHL8%GXe%+kpl3eE(m
zi?fE;X!Z^l*gtbB3u%fCz}gU*N5r5|$^qasbWj#a(IAL;6^;xiixQYynrd~!&d6lX
ztN$eTCF)8mrAp9)WWOC<1q3lbVbb)ub&Gp>E{RHbRM^&a&HFN%98Lln+afsQHouF+
zVA6m7y7WWQg=%e`LY&n6&QT)6K}6imySix*=~%1<YpPQ*vO0}U1tMyf{ut_>nvY;E
zVmb}7Y>gXH!U;DtJU_n_bWi|EBPgvu7$WLNd<^nIO@AKX=^z%gq#vQ|DMaDLLqZPV
z)J4@g2I0>J5GjpdR4z>hxz6?kybBleAq5zubQTVR<Xwn5{K)e5l>{XbO+*<|3Hw~a
zF71K4Kd>GpFy9^B#}_;@^XTwBDhqhZu^UengSbQ$Ii$M;wl7TYX_)?DA?_S@|0)jD
z_xdnuri~f!n75q6dkd8IZbwQ{)k88`k>O(*<b^1I+s0b3a%)u8%RC|s{P}h!Y<tf8
zXTHE1NkH%=H8R6*B`EqBGbx*4PC3Du4c45rHs10Zk!|2wm^o_9-!}$Z3`FC|6hoKt
zymJ)C@By-Rp2DFk0#j?C;}R4!!~q|CwN(A=_N0W(C(}1~O;1ftnZh|At2IMPfCAMO
zRHp0KUGSRn!4Yi0qz5O)@{CJw4<Zh~y8PYu4=Bk*!YkeT^kwTj#qnKor~;41stp<R
z!De_#l6bW>t1rF55Zt59Tk&FL=-4<k7&c5A?23lJ6Y$M32vc&$ia;QTJ@>%Vv_a?S
zL);VL?ns^d2z(`i;1bf_wPIpoU75r!?;bdB@VJ|kD^}-C3@nm!JCP8%6In=_pWcOu
z*K=aA1w5S1Xdf}#_P3~J#ffT`%{f_BJr?pKQzLTsc375_<W!T6>t<x6!MS{5Fs1S*
z-pMBuK>}%9%VTW9eIykwGgGbC#ftKwqS~a}i66`249HXFmNEAPvK9I+B1Wbd7W()W
z8M_F)i;#a&rFw<<I~5ycWzwlY(v{sgn&UJa$Yhl{=E1icmP=wed}}ubXt_MX;|dEb
zJXQn=-+{gcby|9t@ESdRO3gIWn@|(Ggx61bgL|NaQp4X3Vy6<mQrP11wfp=J)!*(P
zsALV~=hx<fJbT!yNNgsXZz7UC5jozilOOuL4+NOG>qx~u;m#+j3AW|RAEhF1KXD7{
z|6aP<ZN3Dl)xpos29MG?I=4ACyn%bln$_?v$8H1}f7zCG@egMNM%}w(`RXygUZw~H
z$2y)^!CjYUS|cJJ$c=y>PJbD1L$sB=#|)wdM<;}B&Z&CYWI=9VTel<Bdbgd!`{rGY
zjPRHkYUlAQWhzV#Tx%xj45hzv9?G|z*JPZ16KL!AE?wOtu5&3eSM6u|XU?3VN=3CT
z69%?PZU9d$z=q?{OpkSWuX~Jz-0o!mC1R6F$nqgwYth-Y%$c1?BSSB-`Z9u;eedtT
z)E780@mzhNQ=Ou4`!y-E(1d!4#mvK->OM+!x&P)m>9_*U8hmc1>u1l(ATIrg3hAs*
z?10>z0f7l}!}DgAH9hw$vI^e2Z21o=KOYIav+_ptMlY^4i}&u3x%Ol0w&KjwzOqI_
z*I#^pu2JRM#*zx@yGJG3ym_x&juu)cz3GZ4Z`(H0t(i}`=5IfquQ(M`{OCpY<AlfA
z*(`U3&YaGXT>Al+$OQLQgIUwLXCHAlj`XMwbOnd%f9<TV&#vIq(r1@E`o|g*kRM5@
zxrj%eZ(sCK0;0aFV@ukeg)-}Uj6l7fG;u-Je^oTLng=q+)qsrB8ayT|BU_(Ws$EB<
zC?uK}Y(CZZ%7+ghPED!=nw99^F#6jY2zo=SohWP7LwZ2UJL%6i_B=o;bM)xZ#qnAR
zWh9>J2@C04Fu6N98Um_VW)G@{SHR%-+uPgIq7xGn9pKmQJ4}tbL0Ff}oWjM=ojNH_
zlu5z#^5x6-zUa_33D+L5^Y^*~Jwy&NF{gKRA1fCuf~-cU_JWbrGUz7wb~^UIjgR<c
z&u?d4Eg@QZJ6;b*-eUm~fJfP-UND5(114u{2g${?O-#7g#2oRJTzaNynO#$FdV>{@
zK`Aqz_#?}3{R>tjo~vTH6S*7poC_i=E6tlmJR6$-+mgQ6(MD>yj5WjklUhX<)|b*9
zs;jGY1zw<duddUF5c+=UdHa%*l2+npz}nW<;Ydyr|JaITzzeteO5de#d9;e353B!0
za<kuUi|r0;ABuNf%&w?yLDsMmG%jWTGk*%q{6ZxtC8UPn5OteI%173{JGI=Iug|&C
z1^&gp$RyJ3*v=1K7yC^2-;dZImVBtC!L>#;oxVT2C+TrU{-Z}iNjXC$IURaH0E=p$
zO$<LpOyn_uF_hCqWvcCd|8Cw{*DWCiY6O7TW?PB!vfiG!XP9{RhaMq;x}<wh)|_*F
zf;e$?6Bwf??MkjPKm7}BBc7w4_YR7U+c`Twxu<`j;iQyv0haN`p+B@3pL<Cqk~$4S
zQp%-_R4PO5rREu$@mje(&$Z*VHuV_Y%hqksw+BT}<qPEqR75Xm{(1UykUG?H5njmW
z_C_kWQ8ppAE5;sr;Ddx@l9(%eY;~j?Q}<XgmO^r@D=AAq`3}^>Vv?EpAElV;$b(nP
zUAmAv&7K)x^W$;1bf=yCBT2nu`oXhj&sHR^OSSO$E+)-%z+C8o+mm}xJg#dqzdO^2
z>c`Z2#M3UCyK#H&r$=3LOdfITt?rADtv&I4Omb;OUX7nVsZ-VkZC%cg+1YW&sM`W4
zSAGB@o}_shl!vuWwx`)M)6qBv??UZ#V40i0xYpH!N31CZl9W=$gckMG`x6I7dCr=H
z*&BRpFue(Cl%LfVr!VS3W-AV@rn7pj9&~549)l3uKT{4?K}Ei{Om6tCta{oisF3NC
zq$RvG{7%ZKt;mn+Kw0(SsmjQiTeYL}bN%zihwb0*npYh!UVF5A!*0kmqYI{yR=k<y
zUN&S8d#Y&&QHyQ!5lMROlvETk|L(gWo>eS`hZ^x8VNte8@jv@aobA$J52+~3&pIcV
zc+d6h3#(=i_Ox%Z@uVGDLGJm-xu3i~jO-ruOpWCZyQ_cU-0YEAHR73C$@|||3Z9`g
z>dL}|CORKeogOnLsU2Zp!^8WaZ9FHJ4z85hqZbjE38&M@C30EM>Q3IwWRHD1q;1Wg
zY%KdEUTYg08&g4$1kCIb0+kV<2wDO<EqypVvIIhX1i!0lriYRlCNGlKKq9Fe*bP8^
zomUVXMMDU8R9joS(`EVKMEOAltQ6c^etojelM~fnG{;W7fc#b}lhO^pW`@YGq~JmU
zqixNGAlMMTQs$jK=d@TNGd<k=lV$Ju+lU$lZzBF#_mRa5#y-{EsjcIlwN-(m#>%UD
z5{qjNi`Tw`GG<#|4WN0w79iJ3_q4k8FDN&3U+tdLsV0`Gko)QFh^Jy~Z0(LOb4+3-
z6IbP)hLP<TI<tu}WLq$YyQ&VtL5F7bY*&0&kDT^1Ac>>_F_FtlNJwDzklS<%hwe@$
zc{0=-(hFSy7L--@%b1j`%ndL%kKAP!92{Iok}L8`KQ0xj>SrT~x1Dt5>gtM(Oq$Mq
z@IcTiyX#S+)%JFl{KgSkSrqB!`7`7t3OwgHRaf(_b|3!}egoD0@sXaE3ojRd7MRXI
z);YPZ3Ib$Blo82iU8lM8$M;5evU~Dp%!_LMA$AKh6MhKCEe0&T;gdNj|8pP&mU)+f
zGcpd9)e2%Gekl?0-u;9e&dG_ydxt#ST~R0dK!o@Cq8mbOdg3h@TY5EDHJ6n_?&;i5
zMPKH4sHrFSz2~wzVSfDYj;9@{Z@k^oF(^)b0Ya@$&LVAttokjv(JIY0JagG4SI7d*
z@@MdKE3^g{L%fobl7n!4=T227so!`z7V^EILcY_@cN4HGci?fgRs-NLh)l>WFlwkb
zIkF#RdU9&6ZF@;z)nH+)9;u-d9zRZ%9RO|ew#WIZ{ytIx;xyIwQSp(Sec}paOwx4t
z=Ng<$G2B~`?16f$5H$|hwp#AU%omRzCuMJHb2fm+D_UyzD(6Y#gS^l=U14Qdx9~&v
zd(IVU&*8E`Y*rDF{#8^;aUh;W3gc_qQV6!v+Eo2%CDfG71JOrj4{g!sXI2hr5Y^8e
zgTBeY`YZPu7*_pb4W{SX`eR~RVwq1l@lG!UsY75*WhK{);<tl)V->r0k&-^_i`2pP
zOyb1`U#|`LyuExc_bIJTVJ}2I;us{I%RuWKHDLk9HXD+qx(=9812WVei_W1}<)^3&
zA1r52l^^DJM0Bol8mhZ%62XIl|Kye8^tAG*8+n&1r34MfnjFPc`2bEPOiyB45{D12
zxl8&KC&kg!_dRiN1cb%c_Ysx@)Fwp7^2x48u|f)f#kbf_KTgh*Z7VkyB@K}8Fp)Z>
zg(Srd36_p(YAz*$rzbgIfm#*@Rc{lE)+1M#vFuz~m~N*@$Mp!vnW)X>ZK$LOiE0Rh
zBeaRos^i!)sxQUSaP$#gkZ7ANLkAacxqrl5=#OmSnrvMNqOPz40APi6g7YygJy6Hx
zZ+bHKNbaXUK5$uu>0dbIT!26stbd{MKrYe@#F{mCe+60A0Ir&E=?1mLrlk0IZye26
ziMpwTMfpX%KvsGO5H%sL_RJPY%PQbRRLbR{ax?^y7S{ye;xSdpJOLgc0NVy;`UiN!
z?R3Cy$EP+m-n&3G2ElA?J#_H&pyf_$Kw?F#im_bY7WcV#sYwVaS>8vmk3q#&EqU_z
zQf(81M;p>tko-}fHHy_BM9p$!oLVEWd3A~AE1daGxs6sI)z#H~;l`{pDF$frEMa}P
zTCt|O{BqJO6~(fe0PxsH{KE<IOJhk#1vQ0<&vVBVk$E7o<L0~geiwN$DVK=v7s+kK
zHryQ?rP9=wN5U|Fnh9*(=L-slsp%HB8%pgBJmpz{1PQ8qm-Ix2nkq<YTvQ9+y_}wt
zqalXDZWI#>pyc(;HNF)@P9e(@Lx_Fvxt!bJe$Am@?KOSvhe6mDnkVPhd@`qd)wjc*
zJvWZ&p0n?Hu6`URmenE6IV5&mJ^4ummsKby?jM<t)Mk5WW0qY{V`c(&t5o#GW)PJt
zYfm|zB1tm`+01iuArChQjI5Y+?}YV%&5L5@Eaa1YMWSevFdLPiTB_pOtY<o`%`%8F
zp^W8i_)aO<Wj-a_umT}!!CNYH$cwCkzzLv_y7|c{f>n_GnP()w=+7R{@L**SMPP)t
zbjz16rc|6Zle$P|nAYqaBnTWwi-8g3zJpYy;s7rrU3T~?9_rJ0fcmztL&Eg!L9yIr
zgu>F|!85fH28z1A5a=croi7Dd=zn3OYyG7KSBUBzcj*TX7H!><H|OzsZZMxG!Z+L}
zypY31p&N~ilTIIrpQk2k^Zg;o=Lij{xMp>^7are7AS!tH8Y^5yo}4wPptY^|fz*Vr
zz;AKBoN{RB7~9}seXon<NmJj?S`}ZfiQ&$Dvf*Vt^2<LE<BZKu&hVfUVePrRK3GSe
z2Y%{ZL?_|UAQ6{EJt#NlMO)df4znbInXlLE-Mf7LpuqF!Js#)dr$|d8Ua1t?qQ3~S
z+Wd&mzhkwOnwa!X=K{Y<E~`5c`aP(-wsQh?RAJ0_rrJ=3M*4dR6$5QKaLVjSXlUq&
zOlMe-cGChs1i`h_h&CM;-vQ)#2xaZvQAtY5NA@g|*D`5AYP>492uOBBbt7;qkMLnS
zf}Nuf*FIv6_Jr6!0a>m#0ImOFkGWY$<@Vv*;Salrj%>?$;VuB+IZ|65z9Gr38AS5k
z!H*u3u^uJy9i$eX#<Lrb)}>X+9*C{%0+<&gA|f(8!vjdx{>pH(LxZKg|4w+()PxNN
zAepV6<`LrABjfO7;yMXHt%s)QjmPDQ@uTDD^Lt3WlAovLg+)XT8eprt$`x3N6+K~4
zS;}}ARCZmQ1P%E5QZsV+JA*mMOk+%Nw$)-Zoboe(TRoy~hey=C<69daAOGgk3@{_!
z=|i^FCu$DKS~hKWY;vst*IV&PjxgF|of)?TroZfy@j%V3)MGPP$Zaw@HV)Icm0fir
z=F6Pe$|o7z3EW}GnsPtwuHc%<zc~^Sq5srAs62b(LHY+SdK}xC&Acq@4Tyd>)L0>1
zs!@sm)&j&cOw%E+qxy0)0BedGiRD<J>(ngQ3}huSa`&VTjA#)0)>okHSIH+KS2|Z7
z!AGrjYQy`Kr@rh{^g*T9>DruxG@vbirYD~|PW||G^!544H8mn;96fgIc6E+ryTr!l
z)JqE6Q+$QCrsE%EiP0xs=_0cHH%2CJ?<@w2oy_ZMcy5PJn!lhRz{C~AchF|`ILluA
z@-t-@gsN3QC5*d$1XO2xCL7+`DutJ>MltH1yn3u!IETo%Cg3hjY*%I0%WI1Qkl^ni
z>)Rir@h&Om2GH!*SjFk{o6%Ut$=AyIpxmY&7=2V31fx8Fa>c3oMql~fgThsz-cPM~
z1Ix341s;?GP&mqTL|ZRWV_``+Fz=ghJ>Xtl;oyAVx!~$-UEt<i@Z_5Q1^jWik|X;(
zq2bP_$1we!;Hm4}V*c%=A;n;6_(@Si()~d^ND;bNsEVpU@kyO`VHZEBlGykoY<@UT
zAAQJ1xC0WO7E`O*@$8G|$zj6%KL83!pOo$lvunNl{leGgGiu?og-GCA(QoKYYr)aX
zLcf46oAppn-mTO_oC#=wFnbWiXA%56M?bIq_}E^c-6Y|+`Qyh8uh3*_^!w%SWZpI0
zR=9GAMEK)`pw$neh9Dx4qw|H%B`z`17dhI_h4LfsrMu&z4O?G%+o1km@wxu$*W3X&
zG59oYv?X~%B~VBLW@<<WU#pC~6cUETlx28U=_T4PTP%)7OJXhviRC~6K;+Hyxq;35
z;ZZrM2s$x2^ghztlkfb7^Fd)+P&;fhk_K58A+Dt9rsREZ@_iKXgGk_|(QO*$`PLCv
zD@I1ij}vS``|m#XqNR`+CIAB|aQ$%!2|n~Bq=O+w9x)|hZ?lQr0s#fN-xdk*lYj<&
zllk5x_89DboLmI8vLTqb3N-If1HftodE(2YQ7apMeti?z(9P;8beNZq*E_-y?9TP`
z+&DruHFQeKV-#Vm7#rWyCxEJMH7MAt&IH|Bo;Wy$j8zR8tF%P!r{1nP@Wsc~2a-oT
zpUf6ZYYgHq%7%;DFJ??d(&>7<p_AG<a2iiW3EE1+<&L9HH+YeYF!B*`|G@FZmN)Hh
zMIv8YONFB|ODyN>ADN)I4jfq=vlW$cX797o?mhu=LuD+?(|<0nxJ}h(1aci`RwPnI
zdZGhIj-De@t8IhggsNJnDRMfc9%?uv*6yC1RiV#vr#V6!73AgS7s!?b=x<%@T(DjJ
zWcM7fDNQ!WnmR9*A@w>r<WWaWvnNs)#)#t2`Nq(NFKAwwjV!_l0eO9#mj=<mxo$4K
zDf=3c2Ck3$E&VL)e?haHB4iyWo8H;s-(6l;$tAQ~?9NmdN7am6h4Fk9vb{kxHau;f
z34nOGA8`E=E#-rxHKX2<#~JM3R#|zW>J5n!tPdn2W<h^GlFE~rs8ksY-e_)L{jKlr
zbJ)QjASiyH<l4Ej#T)!db%P=;E3%w}M;326kFs+#ubhm8R37USywY1iTKX7FwQ`R#
z)zMpmrX@X3AG)Ziul|K+U2}rOIN@{6Zs?^iJ3LBE+)PB6Gm9WS>bLxH(}VR}u=ZsR
z@MDH(e!Y1h=782>%R-Pf01|uSQms)l1b9ppdu%?smoB_rMNGPG#PdxECtLe@@lpM5
z<aJT+5tCN{UaK-b@{bqIO8MOhk6T3yCJuNf+~0B+(iQ?AtBmOaH@?sP^!WpqkO2qD
zQu-Az597W66!1RB_=EHTM$~Ig>dr9rpWQ3Yo3p=JkM&Xd*D1J#n`*J<#Q^!Vy8}(H
zB(9}jn>LMn-T2R*AL!?(_8+m5e|veaK{w;yUh6#Ri~oAv_<{EFua}Y_ZN=YS|G5Bz
z@&AW4I6?;dx|}pzR3FEe!dR3^X{Hf^-F|+V+!b~IcW+}bHr@nrc{`iB6HFaZNY=+e
zt#`j_J;yALoF$U-o6Ryzb!2a4|IJ+BDsKP3HaX6{#+m<H6d3Z|doxQ2P?EB{6P#Hn
z7giw*i0tB=%dDM*y<pnMW_upSllrl|v0P^s9ZF>OTXsn@EQ<*h?Iw>CrF}34DP;V0
z+@l5bWeEia+W|}6YYjXoDm;5N#<D$|f)g?Z@QydslJWnn<~&;8-fJ4%|Lq~)FTt|>
z?R8oCXK1?r_FA%wS~dQ9?f)krx4*sCEdI|0{<8+C6#ioe|5*e0g*A&c012P!7Gy9!
zLG<E{PO-E%ZNOU=QwLt|H1(QwcXuZyLF@T|lu1eJIsWGTA=F$B1VHQZu~013Ht3u>
z#fMLqbx#3qScB9kOcOXa(nppKCudnB*m|;M@)!sIjLb(L;c+wv2NhGN{SA)Gic%)!
z9Eu^w=1yz4U1<D1ezt)U!Oz;YZIzP&r?ty8f7b7NdC_ztz|jv3Ou)$R&F6H#F&0o?
zJTYQlYVV>a{8ord-e$UqBaxJTJlRd;8zA4!-I5ydw`MaEr60kb6as>{?jDY=x@FV6
zoV#vVhXh9HF^P2_lqrJLrH4P7%*j!d+I*V0`s-3nFis~ku9iN}s23$id%_QjdKDmu
z6&+*A8k`T`YR0vrv$L~Jhbd>&h}Wv%%m`uB@0k4;q`8DFT<2C620qXry1jQ-(G@TE
zY9BjBYJ8s^>|4T`K0Bo7-S<q&wB9G6w>fL$u>WeAZA|FkAH3zX3o2RpRCDzM@g?oa
zx67CckjLksIRaoz=0s*3m+Ii$+3O~+#cO?r7>SVItZ>CNCd4YY#fT?S+VQB8g47kV
zqU?_09-y7&k|m^p!I&T%0odKLi)C7ymZS4x)0Z%MUx~GrvtF8CJoa{zA;D-o%YNki
zYIs>Tw;X6rf<AY3a(WNz1vt+hzWZ9_e78OP^s&nwA{WP1#W(}@UjN3~E;SHjuFxin
zzShGY5&p(~mZ}NV6B~jLEZ@w)0<8d;xbd|fy{W_!;@pE)+^R1mm;z($`IeJ~DxEnT
zE}OPyn54|KT;JY)Zp|qu4iIv<%gU8|)2-Q0|H%g_wGG$|Nf&-jLXYWbv!4XZPYN<A
zbN_$uT9glcHDlv%9dqw*5)q-{19K8M#%MagS@^V2m+#$BE(0t%7G?JB!_>aHDY%3+
z1>|d=y}kW+BAioIFTq)sCumMlf!t_FAwTO3$9@?;$%Fx*?93^zoozI={e|<e!3Scf
z7<US1wTpSk_{$Bw{)f(Nd#Cv4RbHs5eK0X<c_zoGqvV%zcDHnF?j$Tq5NkO53x>}b
zx<h^ah+!u}JkjhE|HE>Qk@N*~wxS-oD#K5||4SxP3Tk&jE$Sg+D2Y=)x(epw3_DFI
z+~A`3_RpS!T)ittNgVs#_j8xUbH1G6^Goe%Ep!qYjeq@f@3$pN<}A@Dx^b6*L62?Y
ze5L70LK9%Cx&n%rRtEf2%#f3GDY?uZ%38EsNMkxts_c&71K$dd=euAc)GVJEKyAH?
zA9F5MFG34i|0_lbnb<c{BMeHh!Y!Qrx%}XDI(H-V{{&Ot{Py{^|MmHOz)6nHcD%pe
zfgWgCD3Ftd0(L@`wl?@cn$ovB;+Xxt_rP?h#bK5FqnsQ$0H-b6=5TJ{$OSxy4YW%;
zoWkarbKaVOj)f8R&a}H@oPcHkA6v9$n|%x4gTL}%&yY5E#kUo*=B$teb?M_uJO*Ry
zG-#PGr&|M7Nr96-=jw*t+H#z831C$wcJI-ITTJL*3cm2N5Z{I0zZ(RBThm|w?2b@3
z%Ce)dF5`+<<}LDv$GV;En?t0M4QTY+*_dzTMf4_VE$14Lc9I${w&<atuYwvhpE}GB
zp28YB?We?6bKW%JCFCyFch7EMC5`a`)V0N)#|eXVbjWgIL^(DSRlO5>C_~HyqBkzf
zUEFBT`8V`XoY$8BWU!3?WbusuNox?rN1XG+$}V3TqZcEZQe6k-N>O<a+E7fP;?5;k
zaQX8Gfw$otgoQyx<ntfV9%MBlbSBRTM=QqNoMV_%9vldT!B}>0HglM@S3oTPx3P@f
zPVwX{hsH1qX78rpO{gL1we05XJuLIb7rYmx0TR^os1d?BAxev9K`_R9J-!2{;C)?4
zmPYIr;+NbpKujS0N$pR_aU+QtAT{yrbO4r3{&bPT_~}HwtZf)cu>wq$k5r4R1W5r1
zE!D5uptg`_+d3<dl=QF37jU>SOh;>_;S!jlwv__sIfG~@yv6Oit2pYluYYi*Kr-+V
zttq5@rlE8+#gk^rJyHwj#pr|c%uzyB6<KxV(26ck0TD;~W(b~FktFvO#Bj@r&dO6U
z&vU;&I?$q+l*_EYgmGG}0KpG(t_tLzrh(|$S8Mwv4OoD1+C0;4FtZ-QWqsD&T$&O=
z7&nbUBR$}E&78Djd%(X1T@Pl*JiUoY{fe)Axb}~&KfT=gQ}kJ45B+A`+A-r|ONsr4
z!=Y7Q>_JuB^!FMe(ftUKsb5l{r&c$>;EvIxK{N+!$pykL#8zWM1{TX2=>i)@O$-VB
zKw)FJ#q=9Lr%74X-^%;s$@KxfY_BIIRzuw6j83{PNXDd`!t-e`4fXVN0UrPCnoA;z
zT;$e?EtLl44P=`}2H>=hxlCK+*G@##f({uWJaGGjd%9Iq=nT4a?$nm+Vv=YjW_Vc-
z@xoNqNVlsVs}b+%UQUnpa_f9~c;M#GEtV%96Rk8gEJ5$Dt=l7ZTui#*J~!kMPtUCU
z?aWG*(g1KU_uu3<Ahnd82tleCdLE=r3UsJMQbZTq_kI*784b}UW@F({aHU)tY^5=E
zLA4V_QhAF(QCgD<#meE`V)XC&dUZy6C=lr+6wpL`FeKZ4L|9`r`jc-U%+nl9GLu91
zc5I<JqxP0Sj9sWTZm>ionc7uT6$YqQ$Qf_FF77)1RVZM($Hf@}+6M@(J?76ldbsOJ
z2Xv_`CWI8%hxuPLSiLAzcOMGM9IZKNmt9oL7H}EkGuCzz^=I=WcwY8euxmkJ+!|-?
z&2iq##d#$iw0!5h@_b<GUb|$}Ipny};qAEZ&^7-w91-E=^2t3_>i3U=xer#J7yRMC
zwQ1JNLcN6&vG2m}+MO%wOs&y3cW`ML3me|b`cr1OKes1|X49&mw0Oqf0J2~l(|%Tw
zQHVs)p^o~w%JXsT3O)8TcjyVuAgK07+ePd#=bNJ&jSaIsK)sWlb_CjE=eZ4G9K?9%
zlwrJ=(Z5sH{ecgK_V{}V<b9c*9`F!=?89BZ6T?Tui1|UI*y~Gr$i`d2Gx>u|ues#A
zcrlQq^EAtl27q}Wjo!xJ05bFvKzeGLbQ<Ll+UWi@lXx5WCI7uMvzLdoE;N-Td6))r
zQ4$UE$5`G0Dy;yF5)`e^u#VIbpy@WG4M--4o)NX<|MY;bgJ4Hye<r4u`Ctl7yBjgM
zJoxUip3-4ScgS&zh~dg9KiL53#oe)xPIHW5QS50Fw$7XAIN{-xrvl=vm3VP99t-!0
z(`>P8{LUK(#FA^Jslr25D9LA7!Aj-=C8pMF5RuGj4wioe_eZT99n<X-ytK4$a!u$S
zx9gh~RBLi2waoN*e%plCXE!mTj*{+ft!iM-?N!z}DW+{rhS#?kmTt*N9Gs}@pL?Qq
z2BRm7z&h#l<$>7VPDG?{X8>df<P51TJ~4l0GG(}IXR-WPV-R!UW9lCTQKXW3u3GW<
zD5+ylIn=Gs=RT@b*q>i@v8QGWSpML_JJhYg@6vlzeJv&aQ7K2bk_^lWVyZpjk8;xc
zsQcUQVzL=SNG)t7%@d_3t^jLF{kfZuKV1`RM;Z~Sri+?W)NcwSC_)u3pe{G4Tx7Qp
zT4LzbnemNa&Dv;YCFsJv)D#D@jHZPW?<WzFQFrbh2^X}CT`(Pr=1=jI^FeJDM#IAB
zB+}d+w0sO=z;7{mS6F~m{cw7dLxVI1)FoJ0qw=PSvPO}M7gKl{>8=*p@$vK(>aFO-
zKcMdA=LYqi2_L6x(})?QDMAevsfQEMKbHc3QiN3uAI+hL+P*v#0k5bzG>udZub`!0
z3ewP<1`E*7WIg6nw3kQrn-wTxz1TB7-wu@V`Zop8=mabOT4<PZeDjKf6}OQXi-tU3
zFNS$$Mm6qOmRE^=Am8S_nY+s3NP8oezkRzu#?|S_WPlWW1A)%A2+q8{HU-f5smz@y
zJkCb@?MPrwLE^FwYK5{(GdOhm?KxnUOyoLA7B?AzeolqH3C%}`;%%oe);e%Bjl!s<
z2lDDN=NSk;f|yf=W08HS3Md=QSFeT3R(EK$DK;x-(`c^2Mhr+=*LMet_bJiM0P3Uj
z<@u0#rba||$*1B;t`{1{_r#zi&>arekVNJ6Am4-DOgUcSa+E9Sz?SiAF(1=D8jdBe
z(34L)&H2sdqx1DpoXpMzDAa?lD@fAdJ4V!Yr)|qI<%j0rgED)usUZ?5d}1f*tvC~p
z>hciU=sZY7G(f!4I8WCP)!QrN37UO*&{z~ARtFDGVawz(Lo)bT-_w^1OD~my5(~B0
zqn?Zx*5oEg*&y6NCg|Iji)R6};rOu01DQk-L@syi^}3e5U}GFWOh^$%l|629#Ba{Y
zNol&Cb18?aDtx8vMbr9ek{?Y$Vy9lTgZ4b?eU&FW;KYULF@umQ&VbpCQbNE%s{SBV
z587xF99jTKt-xrfl1noXg;w^nYV&7;2O9x284H}|rQZG;u{8!TasrO(r^%(zqZl{P
zFl^>Hp124N-8bwlNF`S^-NQ_Q{M}g29Zk}@theoVF8VZioI<W)O?>wGF%K+M3Ur9i
zSrUaN>0`Yh?Tixjb7lKx)~U849+Z9L6q(BcPdq(wA%&GJVpc#8a%=6ZYlJjL?S+_X
zAdM3%8m@Gk;9pEbseL;U<Yk9HZM%eaAv@WA^o^Bg=0ol++ye)2*riwKp2ZT?z_X`7
z#ijYWIl!N^t*i-tlIzu*ED>ZlW}rjTDAM-TPpbyf?Ci?^@@0{9=8EzpreBnFa`2Un
z%hFD#eWH(lTg-j$eB_Y_@8-;Dho1KFn7bmx7{08KEhlf=w7^|d_Nxc3Ga{uO&zQ;3
zL}z5@MZHIZdzCGoSEvLlX25ig8FY7=p;%PYh3d(z-Y96DH|CuRn%^^U9o$Px6T1Bk
zTmO6b!K6bN%g`f^QdISYn5JtIpH%e(RJqTX@=*B&x|GvG#krUjOdRD>i&>!;{LN$U
zPy3M(wSt;gabGk)xdjS{ZSJbgiMmbiT#&L3*vVRfwB;GgKo}V61n2bhm{E!LwS{~d
zH4<*;^Kp~Mp%7B*J2$yanzLF6H}?{Y&t{)IFLbv#mD=VYY<`x0bDiqn0*6@v-E-XR
zSWN#zsNUnQ9x}LLl8Sk;zASUQj0y|eTys;{-nM>HySA5O6lYxZ9ouSjKiwKV<nhyc
zjQ$E6IF<zgc&(`%UF~dkP(N?hjLq-abQYhpY~9MK2eV(EisYz44X$Hcq7H2Py<0d#
z&TP-p(dDT}B;vAdtanV*L;6(aD6{1uv|_PMI;aCqTBd2&>_|taP+0yK(X^RW4=e_B
zp`S}lPeDo)QtiSqN6Gsu!VYZ7bdR5Xm;^0=cO~!D(IFG&-+=60qMwb8*p`%RIrRAr
z%8R2t?VXBjr!qGcWxTJTt8Z*o{7|zOR|jTzt}${;zaR%t+r5B;&hl<a`!p4x8vAit
zO8J|;Zz6jyd5^)Ayp-h?8IfX`1Yh;jBwwZRAI2oD4PlOxNN>eSi<>P<c3&|*>};bV
zaF9gH&$fr+nJqsXyOd+<{tX)s#Hycox5wSsLy_~cis}|=w$Xt!_HX-V*Gj{f*6J=V
z8Y9u7nOE<IygLB>LdJ6JzKj-s&MajY05Tu=TOO(gWnz%=uz;yDW+};%&~`kTJ(;|T
zMI(ZJJNt2W)SvO3HE#j$GE&j`6=~1MvjKB_otY248V@_}6Eaaww7bd=ITuom1GC}D
z<G(FH6=n*rBuG8Xq5eEcNocqR)KTKp0)GiEa9gwYcEMKcgl4cM<y{B~Yx-pH3;mxE
zQ98w~keb;v+1D15o1?gIft<XWDT5Q^+ef+wYRoBy<KYJs<&Q~voS&g4Fd%h>OhBwG
z%<h*^OjU3NfXNfUgYllE|4l*Wtes~9ccMeR`Zkz<TvNH93eKoMHOUHBq@AY6I-s3u
z2r*RD&!pMm4VMBgo@lU4>Ku)?Bfx)PnDk?40ZMo#uNp{GKqxb>jv(8_GP;cPHAvy^
z@<{PWr8o^+Hie(CKq;8=a^X&zCt?WgQvezR&heGvKV;EHdP!)BB>TkSb!fC9X_@=Q
zw(X-QIS{yx2BZ3V$Q{8qd_QmD+e<neWUK^X;$7UFTKyE3wbwWIrcJTpMg}EaQ1Co%
z8l%#a{zCwnRHsF>KxRnrenHVN{yP8)G*S1XamcCpHRtwp=l_z8w5OHkAVTyQ{c(2g
z-G?JwqEBX434U9fqd~Mbmw#)A4+_Y~L<*jOtS=9)8=@U%bo&D@8Yzk@nwN$}R7@aL
z`da5trx(DekuGkc9(NK%`IWIt4Q4bP1~_0q)rPcBAA0hC(U?sKm)!*}apX5EIIKH>
zH=8^Cfc%q0EwT6sT&o%qoJ>?pM(8Y*<S}d}KoyqphIO!)GK@0mHs84|9(zVm9;l#n
z9G+R?h9ZyFG%hY~zsp-JxA{&v+Go?Rvypc2o4+$7f46ZDM><X|120;pMQ@9q3LPD%
zgbw-IN}AI^LndPt0@ry{ZvZSjmZrx-e3+?XbP6&bq*p>T^_NbY1V0NEVIkV%S3sb&
zyT%~|!$wK=yxL=GkQz#P{hm}!T6+<g@DU@848e%#225jR+Z5TuiX_bCU=F9REcn2G
zvhqb3xdA1k2P+a+=xzG8-|IQyjNxO1HhKchNxalXBgwyCD7^;M$hD79{Tjg%13=;w
zQwl*nBydW_y>+SPRiyb@Muml*`$W49nvM!_Gkl~}vcURRpuRGN+6W^&1oIGeXzyH6
zRB=!dkWQRoY)I8|`ZrHNTS|(^mE=1}??*YsUTA+FAn7Dk)tw!dCx{WF9w}1x`{`>6
zkG1@aLF-$Pw=G4hrw8YsLCy9s#{}8CA6eJG|DyoP%H#wp9;!~d?73u3XZF)PvjUV6
zyT7@@qX=YyeNkv^lqKcf<Vskn(lMHRPEyPBV>GJ4rp`E|x()v+m>G4nc@a2Y$gko@
zqjLeKAMHJZt1ADB2z;cT?uaz^DKt24!NYtvx|+8va2$#LNH`XGKMwJ5Is3s1K*$WG
zQ(IBOFiekNq(cbRE%nm`K4Yu1`1j3H6!*WyAssR@h7ZjXqHzY&a759nL5(x!7;y69
z6{ZoOq3$`%6dE&b>V2!)N)OA3oAd4b9p?BZ#zu8&1VA^_tMBHm{+Z;6>@oI?K@Kw9
z!E%oo2}*TIbw>5}j^;~kU$FbDv8^P9R8BqD#a@+4?N>-Z<@0docy0SujaA*j@nbA&
zdUVf<7rw?9BSmX6{YctRjUR+`e|pXN2b5XS4+nNxm8&BrZajlDj3(p4=Ls{9a6S{a
zb~ZFD^V~q1rv>$xbg`HO^eEQve{(QV{@>2uMsi(@1s|}7yb#Y>lVp!n>b-w*u61PI
zH<RB^arW|qam*E<DlY|tSPCiTKwo2Ik@{{&%AdcjQo*PgYGtn=l!6WvPOBd>h+>dI
ztsXO-9yIO9i-5tvb;8V2&KIot{@Vljy@6m$slpYB=^*54g&5Ozs`vivP3-&kxjRg0
z<lkPGCr=|9|Lt{T1!t@OdZ)ybuKU+ZNta_af4#Ahi?afMy<uE{%KdMzsLh<s`Rkoq
zb7(F8dR>0@pPQInjsL8~zpdi`!b<E;-7WbXSjXNVp^kI^gT42R>gwFOM~R7v(HJ#B
zQBdpxDj)(XB7y}F#731KL8S=@3P@L@Mx`l8Zz{cu(h;!G1O%i?6QziNfOMqZxi-mr
z&dGZ|+&k{LW85+R;Y*_OD|_$fS<hN?&NbIJEUYhaPSE*kFp7y3`H8PK`Bzccyx%{L
z4wM#zYW|UJx6k?;rw|&m%od6J86I`M4BB_lj&++I1j2ghHALueV?IuC_4p6-cldUZ
zq!dN!(r6J70_A{v>3pxBqRfv`X-+MK!je6aI^0RBtFtIXpA#n~4zX3Y#eQPB{P-R(
zn;2=45xIwVtwX?lO4F&R>w*c?tG;ma{B_mz-+#8tUw7mGzjuOrL29G<VMYA#p_Jqv
zxf3Z()^bH1IcLdb<hGd5{eyu(Z<HMLsBp<fTBWmSVdVOOyn#*H-*NwJz4JI*+erHn
zlSOTZkIrwGDLt9p=ERb>w(onE`;<uv8IlJ!NS#MTBk11RQwsEQo%1~Xc_ekz2mV}?
zf;y=HaGiYU5@9m7bfs1TW&jgRv2;=8yAtGxYHbNXt_zghlSv!3d?fxcUugS)<o&f1
z5YM+zLQG8_Ob&nzL`QFZiZKV34b*T&Y!5B?#6Q${t#`6q-;nx);r^p)GgViw=-lO>
zvrzLu{13&M->`5?x}4o!je<?SSK~YYw57I(Bz{~DDLYYBN-40?%5eo)BHm+7BMOqO
zLe-um88_@82eiuNm|+o+vdT)tNF(=1>l^eIlZj`{xz)mkice7-OD;drIt*YZpX^tt
zG-ap=)VujO7j;ml5zgZya^Wm}7G-M4rUl7lhM91M9*x^#7w_MVG{|25q5jzGn;i2$
z43>$zQ9J7!pi0ZZ|IpAqnkJGDdDT)HMn}0bVY6fkKy9)#v53m!NnrbK8)OWtM}akR
z0;4KkUO<@cQR7#cQ9Q}N-QR8ayo$@L@=~Epx0~9w<LB_lon{9f2;H!G^pu0;vV4{P
zCWRBJ^epbpd}ohJFOTz=)6_3ZS&e58$GJ!Qep<Qfk+LzrT2f?v%GzJeDj#WFd?Q(H
zLK^Ze>kTfIKtNia4wgFM$!yUAB>A)y&Q&Lwscb~`SnPT+`O5q04?=q^eEFr)Mic~F
zZk+2yOT}BH^bBdIrF|_2kht0B#DaXxU@>YO1(4iIq9C-NRHwOEyvlqqRM>5R>WoLY
zCH&M~j~d_x`LApx4umX|<N0N0+e3ls-`ksjma%(8U6wLJWJ8aXwqhNYe1tTT7{jJ%
z_{Qi!sGDr^$G|!hc2^J#A^^a1Ha8TYYps?oTQM0d6$59X>O5MPO=vTirB?jnElT=<
z$mpO9rF4_@9}J9D;D7@34ce1w?3haP%OBY9W|Xq6*O-V}VRm3QnWzUK0w)4SdN|}r
z12|BNE0^!UH)RE@5O%0hXr*~YM#!?C?#)mBP1ZAQtw+A{A}=e938gBEJ&f{2R5Z#x
z*(8ezb)zmJ^z*6SN#S0N`yewq(c-0U64h~6h>wy?P%ldHJB@0#nxuWW;Ui=qPeY2x
zu<|X^c5-y3<wetYTHw!TPj<O9gGo7vrY<apKlH06PgecqUF)@-&-X-}&0ScX4ydWs
z&nTWpZHAhsTZv%}OycePo+un1rP)WZ5UY$ea;`#A?!r!?nm?D7&i19{+?Wi&w->`2
zH#eY{q2n#E+66Ub*e|V)un$>K!`Z$MHvsnS=fEI^7meWSryn@E&;^<+PPFP24Wk8+
zbat)eT8Neyy**YQM|RwTJSmU7)0l~Udtu+Wdj)^_rD~PGc2ms=jC{6C=gjoY9L~%^
zWyFm6fQ^d|T;Q0U2?&*ZRLmNTQzd~S7*@|vLQVBsTz%oecMBzqnaeryaxE9{9$cLw
z{@9hfu-mX?_&3BHmayE@o9~~<06G)h<Up|{DHs?oCK~YS`wKa=gn!US^Cc<jlv42a
zyP<bFaczn4IqL0l;>+MZ(}2LUnyy=`hp-ki5S7aD{2*x@aFf|iJ4<|xOt2GpksR!|
zWM^y-h+{IJQOq<&>$o$``oNM{M%F#&9>3c329V)f{SX%&z&%d%?kH1Sd{9+Vb&N{S
zq5T=HA_h%Zw)IfmJFH{&oFHteZzA#JWPQ^NcD03G55pS#1%C`oJEiH8&bgHW|2-c$
z&wLcsY^2=$4D5skx)D{wFh~UTqvKqydjz;Geq#qUNTFxSY2l8yYe4(021=p8_*{_k
zkR{>aDSy<xWHxA4bwXMaR9Fa6RQUR2r%F>EXkrRkS%4`_Ql6^H(-T7tl!gbzI6dRD
zr65Dlf#I1GHM5a{4D}UKpK&s~)xoCx6Yd+BhU53of7ivo+(M(J6gVvA@L6TmD1i1q
zR@_33IPC+P%p3|V&FWrJbf?XK*F4tW+6s`G|59cr`U1C@p=yx^@DYN(Kc;?1RA-PU
z3Jc`nOo$rsLlzY?X!~vf{79CEc}VA$VMdYG#ok5{@^hsV2O^5cQa#x?I0!6brkOG$
zguxcwMxB~V5Ir^DK@pILHt@yp1T?;)eZs6tukogDeO?Lrv^9H1;1VL`z9HE#vvCr>
zva(<I&<RNp&Y(#InlaHz;3fYEP@6PKQ0A`C<GCw%R%NwFh@F6a!$4kR`bh0`{pz)P
zv4>rRP1p~V(9uM0Wr~>zxWs<IRRl9UPN@WMwj2tWi+%6J5~e9}CDL3zr3Zk=+!=eG
zzHpzat4_ywcZsIk4_94n*I&3#q<whFLQxZ}qL5_);v%}1=J_RUPSr-Ali%NL(VpQL
zNY1%;Pob3i?Dj2x%!UW|yKV2WB4yXDYkdq7L&Gmx5Z(WBE6VU8{=)%e2i_ilI+Lr}
zamL;pGEP1FEKt^mXc}0^UW5ftGVgo+)&ZVX?r7N-irDLbp@YgnqpN57SBqq6%Yb?)
za6Qs_0V+i$R|71|no*smcGUD2N_7u)0G&C4W<_rWM-Jnp6Xsx{xNyiGYtXWXagh54
z!YnGFg$c!djYwR5p+w*nmg(aZ_0?ucDKu~qcV5%nW+7gujV~A6R_&mP!+0Wkl@act
zxd{Q0C0lx_$r&_8Mm>#vx;OF^7u2yQoZz)Ogu+2~6|650;5)n#Ba1LBB&i1tr%wj#
z4G=u?_gW!eZv%o}{MAAX$;T_uG`@pI^-$?}7{#Y=!W=5qQu``zc_$751s|WUsNyT9
zwTLt`tU9Yn6JYX3KOmpt8(HkZ5Nx{S%o0~G+8OF0Ur)mT>U9r`X}b&7$gRB~_4T2h
zz`7he0U3>nTCJ{u`rr-bQf3O{BD$KpK|@x$P5}C}4IIH%H06Tl>9NiWzc1Wzh(dcP
z8s`lPFpw+l>4YJ-^(Z@mh|3kw&(1%x*ytSiUe%e`U~i%#brly(f7oMfKioEAJ~1!(
z{kb;%WEDd*g<1iR4ed*IF5JjvJTE;`?aK_W;dAr+{T<E&d6PmlYUkYxSJZAb&#!p3
zB2Vtry#r!t589d-_Fq?SxU(|i?Sbzw|4c4itvdkodp448$b7o28coWBy{$;A%Ek*y
zjrhsnphibocgv0ViJ<ljXRxpxJGxt*1R+Ut$<M-tabYPHb&Ewg?NV!{1OS5&{A!Kf
zp7uE6PLzYOfA_MR<kuFEFRwtMJGqq*u?r^a6;8yUpTF9KlEY-<>Xb;zBTl17R`_<p
z%|rAy<l1iFWMXuB%eg@G1&WN4jp$i>W5lMz)DhTPi?Lm{O3NB$C8Kg(e0c0i5Di1>
zcYuac3mNI!A9XBz99X16{u7jjn*fPLDfyR)&)CWOi++A|;LqpyF`*oLJu&U8j(e49
z>B*aaJ<OKbhw0(iD3tG02V1z9RozHkgNCOPT(D&M>d^ZhGb9R*63E}xW8t#yH||;`
zyjeVbIXS%O%yv(;^Q|S_F-8g|!9k7L<Nfhjk#qI5i-e`Oc&0@q_{g{q-15J1$HU8p
zFM3%(M&#V$6@!|O+nHyV!aTc5Ui18+T)CIGd^hj}+nhg8m|07B&HaAEUliG~3@O+G
zbaf(Nok5cbUUJ`fSO-go8r`LF-fVH3fvXcUmr2W>#>vtsGV;|6u&7amFw8U;kBEbg
zS9(;`Nl;M$JMtECFn?>b)q-HqE%9j&&9ewa1-yIdYrD5Dx+w`grjcUAa4`MSXip%N
zw=6R_V>)&e`dw;=Z2@}aLc^gb1^n>hdOxWATymo&3aveHwZqscq{!P6`fHU|0yI59
z4a+BCUUl#wnS>Kse+lnT)9EPtAi%8^61t<-c7xPq9U0<!8}p2JC)A!$p&?6#spPaX
z*B}SD^&5QR<_2^a7ZF3dZ~UY#{Fj(MY%tbFjl?~IH;Jc;AUfZeOz6zOgSXpF0zapG
z<VdEtj63;nFQsNumx3X>S1=?|8F2eeJ0!bBsw7m!-v?==?u%<b3UeMyw{E|t{H5nF
zEdZQmYjC>uorKG%Vf9B+RgmMc#tfM9rI<o+B(8qzQUv?1^?LXCm5*inh}KS*Jl338
zf7@}2|9dkJ)u;|<i&K;|;~+O2U4`?K6o@pv8H5}+=^BX9A@j!_csmBXFTAtXz_H_u
zAf6H#b||E*LLNgB`r61Wjl#^RV<1-}mIZL6+n+Uj9@VBF_MmZK{AT{ro#qww3JXfz
z0t5tc0uAJquq*KHdU!E@#-+1fduzLw;BDW=dkSJF-z{X3kSRQquav&WcXP`hnQf<S
zG(MxP0dwX+4-NEEn*0jVWZ#hhbS-Qk*;G+KJ>vS-?txNyqu7az?yfhV0X!c`Ur>yS
z_|9+uL`a)fU@Z^emWPRj@7>!?p-6g_h2nN%y>zwG&Yv<z^LuzwaA$Ij^h!FVdE6>`
zq1LJqm4c|ip2tm-qG^m+{NBE;!ca>l<wR_f^SN_D)VFsrNgp<`X<j^$Jkbk7yUoVc
zAAjCQ1c*!DLxrXRSbp#tYx|<7+2ah7qr*f|8%|_kB=7v1GPL{q4tA*3%C0rZm21i^
zK^}C%SRIgVTxF(sdq`EVoCTzlHLosNf-1ac2}ij)L$@u%b~2?t!5M`s+MV)Vvw4J0
zBOtdsRl8XC<R#}=)=qEWnI6uGzv-Xh<e$G>uv_boS}mel$t6G}D+qbeMsi0jogyz7
zrTR};=0&#JY{$Y4=i!71n-;oU&+W~nAQ=hciaUej8}2rA{32~WQPQdjjY_F8PwGlu
z8QJwFU4m!bL+x6^*S3X*1vy2})!U?3g6Er>Sj-kv`_}Ne%sRc2Gj|2Ue-T``<!n(D
zXXE05Ug14Va2YrqoeVG}7bZ$8$>rU)gj)Q?r#g>qsV#8f@E%>Yk34kAp0%P^oGKzf
z-;o;Hm@Z#rqdS5Pt~gDJWdY!b;NbxH&J=(P=0rJUT1uH|7Ac`%@5_$7t*qXW@=~J5
zz;PED6X@<5GPCbNt3KJ2<21NA_90%Lh7wbfUcBoDrOqf%HPZLOx^)cpA{Xp`v}l2W
z*s)Ag^t^64DNe2*G%TfPgbsUR{I3qtCL&E`MK28iX(i^Duyv{~i4~$#hUO0@b<or}
zV!YVhI!RCmxqd!KOd4H5{q_-JmXK%(z%F@n158O)>}))Lpnj0>1Y~}$6z1T?Lgr31
zGA|;_=mXhO&~Y=(A3m>ZH!Y()IoOy+H)X@2nLb&|*fR!WLqcOsK^x39NM+O~L~$I-
z43jAJ(XwyDc@W*^^z}kdLS3@>L!tPYJe1%jr6<`am9uO{FuRa_xP#1yAx6<dOUimx
zw?G`t0WcZhY48&=U}gf)NNkZ|6@>wlE)TWy{ZVj^Gtq{KNkoljOTp8<56yC20+L3m
zt`2hP$sH^8;rOaE%me|Ugc+9)_1bdnJ_rIOSSU<&!|EfK>;-qVgUG_3a9UN!a4Ahc
zrF#i*snq^la<wus`LP9o^d0j2k$&eR8k_e<JW$JA{^<Gl9962~!~LpymEcB5+}duF
zA-*p5D4seEn*bG7H2eg9*p{e#1M&wj96fyCP3sor*qg{6wnhwg6%iKDGeR{k&SR}d
z4rFRVv%Nigh=>A@oxH}r5y9GErituD$cbbZIUFOgr2~_!B^cic5_2QLny#Hd8QnrA
zwl4XC=szPymC7o$lyfTxJ;SZ#`Wc=xNgYqIo|H(HXH2SHL$_u?!;Dh_h{}AbU=Zcq
zaKP>6%b6Uw1k<$;m=o3yLXwfeo)tc>Tx+)eY!~0zn&|;U%}$5DJ6(8kGO7u^PFMWi
zB0*KC>3Hq0a>vr?NU4u1^Y9f1^8C#H*TaXillG$&8XXB|U?*mhsQ}0?gNQ)ojS(5J
za5f+RJ-Oc9N+Ec*;@D|EF+>$dSS%CrfBk#NA7Y;N5JG-CJ#7m4pa0`&V}7GRL@N{N
zhV6bCLzEYRrM1^^8Yu~RA_C;ovq9&LHP1{n2TT}5ro7d0Dhh5!Jv|1zM3ZZ(sz5K$
zDY;K$81;JrAcMP=*4JG_NaosJDnNcU=Iqi3bEP<pSz*8*t&z|x{sGcvhYd3F>11>z
zk{X<FQQ!%9fKc9%sw((V%j;OW(1R}095rk~At9I9hcOt@=09W$|GfLteL@{8gC<a$
zwTpR?*aadmt)6VTnIIx2H#g8w$D{&rq#Oo0^F+{mLlf@fESrn@b5>FjS#Nn*Ml;{V
zM@Fg8=he`4*~Gr++T&Z*lDN-g9oM#rZzu)KwxGf!@zn~!7`K2E**?+ngLyZ{i&u0D
z5(MN{I8(p+1bgj_&M!L`3hKo!AZS_N7HKOpw>Rs;2x6jEyIV2|fyoGHU%_41r~c!i
z%QWne!%x)9FYpTS%rJk|y`EjAu~5bcRiil7hR$^;Idp7XqL!^Zp;kMSf3b%@#hmhc
zlmB>J=iD}G>%v~|h-w+ki%F?pT0poVPdC&MmZcp#6v-)Pi{muP66&mX1`D60>~=H$
zD9Sit14qO#kwBGCyVNpo)&X=Sg@(C__&WLdTOnjL(CS!LpP^kId$&`dO&BFtOd9TK
z%&M0FBaEs?2zMn3L;aNRe3iP99Msv?ZXM(a3{9<aCcowt-iKW|z7#IurgO3<5m-jt
zz}(!78kY`XNSHnQO?HQc#v{X0OgJv!(P>p!ZzZ`Q%fhLKO4@z(nxj*ukOb_j=iMQq
zdv^G)U=f?bD>kFjWbqs$$Ag_3Gfpy%LL~u-e4)0EH7S^Qkx!7pNhA_mRUdC4*(3F{
zRdUYYT+2sSSj;){vTPw~QBewsdFbTBHZF7gBD|(Ad~eiTwqUg{Ljk284G~D6>~mJe
zr4?;VaMRZ?ThA3I1{GG}3PBISqh&U(b^=-X@(WzKv$`F1Ge+3d)7zYNHtlRzoCPN5
zqtq$}V&PtC*1c)4lXI0uAmFGS9lEs+xi-0r+!RbdyW-eKpNiP#S~6x~Y;W)sT?o~u
z{Dll#E8hvMSFxB<oxvE2+i08EL4AmC5s*TVSW$E%m5C}MtCzV}d|u7|f;ZSmT3-j@
z%M$`j_zfOsQ5Zl&&%@MF>Mfkm3eiO59>*qW@!HgjZrrALkHDdwzdgIWxzFri?w{DN
zu*uF!O=SZ2g>MKhy$QR$6{P<(+CyTddV8pcWim}c*zt4=o-H|K`x*2@%p`}ZRXS>v
z${=n3(ah{>NnkP=F1{c$!kfI0aUL&4*uL{vD>BnCuAvlLU-55ouimVuYzo_GH@4BQ
zI~Vd}8({=%ahfCD62uSz(peCwMiOVjYA*o=Z^3Govweqq==C@ox+jKym~<u3axf#j
z%IiNj+4Pl_aY>#Bv{t`KYmqvtMmk7fKS-8>u&OGJwgQH!$ei98Zsf?r{YSuwlD6=G
zuIB?L&>UQOrD*i(UZ|e?2AR@=vlGY4>UW?I=-x0P6?P^Es&Swo#BfDV{$Ps!J5rGl
z^!*v-DyakV$z5=f*wFE0M?3E$n)WqRXOY<r$+L2mR5X{ume=VRoP{YR4Va(S+3TDX
zk7t=7S%QZcww27<G~72tX+$eTW)@c~W|XCDGIOUL#!M<XGl5<Ms&69AGPND2Eb8sk
zSqGudZcUn?<|QgXv|2lnT-MuvW%Y-2(O)I!YvE9kr(ecR4grrGn1><LF&K@5<o^2d
zqamp#D;<yk%{8!U%G~~O{JLJ0hu~Rd1|svcOSIIO7@TOmCZ9gd=AUM83D`&B$iyV9
zeG*0%&D#QlFcdu&^@C(7wYi(j=;|ziW9AXiu-v!@Y|CtAvNbgU+{Eg_H?ry?ZNp1k
zwg4AP#*SN!ZEl|bnee!O-(KEV%?tk)_>@T4X})(yF#ksm;N5i_f&dyUCv$SM+$Gh^
z(w6aSt)yAYj4%+~6nXzUZL3i1%b9T?K|SJxV-=${f#UW-Yh;R5S5azCMiNpM#mp?e
z1<3^~FUa8+V~R#{;ML!c0<T&%XPBq5HTi;d$tC;3;YG(~C3W5W^Op(2nhdJrroL73
z!=LK}&rBiU$6NRlA%(0`IUW<06gwjxxC@@GT}&apcd9x}r_E_e{8ZTf(vxcJ`x{wB
z&!*s(IZs#tY-jKBhh(WpI4q<2HO;`w&rG9!bmIx>m1YNv5=pT?w!7wzRpr~l?k7e!
zt?jQh?hT&qa#uUwbSGqKP@2KXe+8>2cy^n#<&N>D^$KBSN9lh?Qev)f^iPET+39GT
z6hMmsZknTad!Xp#|8#P|zRlcT6jGs98qPsJp3s*sr~H?k!IXNvhH$<z;4aNxzAH8U
zvbBf22hF|RS6RMYOQgyrn02;7$l`(%Xa{0|m$_9qJf+g4rk3t@MI<=>bx>K|*mHZn
zowrVNR2ViywaE+${tMIJUiV9bhfDOi!|ogSGCx0NPxkJ_M2toQuF4Jw5>NC58g?Df
ze7Et@+q^M9?$d}elkL9>0+f=Sz5hJ`7<o{QPTRl-Emqk$8{Ne`R*OEPxu8fmZt-nT
z{#Pq5)E+lyu)G(^;woY0Q~XO?Hf83o6!gtzal1(hG?$UM^xrQ=-DU1DetTeiUV2ue
z7Vj<<7ZK^0AFb_8P$&M+dhKm<uJ+&$uRUq_e2L5Gxc2<oX?c`f@rea;B7f3tb3CHD
zRQKj3z`LCLj6DK-!bn>3YO8=U&8u{`6PdmZq#}7ZdzVo(Af$uVGdffpPFTyiZ=kvK
zXY9T^uSt%SMXhc>K#D*sndbTsF(O4o0qRvH&LbF=Fv&iQ1HB7{b9<S#!`CzUnU41|
z&=;qeX^)|BD`wVfsZ)(M%e_#q1zD<E+4^#tP2<^~<uy5zA$S7f3MWgwV7Y{D9GUS-
z0@GCc6{x_h+8JPKs=y$AHN0swWc3txu(x6gCCYCNW8yZ~cEF01T-xL6S3)dYG>$Sc
zVC3x$FWmY;l`ewcF0yW+%&AN%N@M2m^|V*<0tIsTI9(^+ltY=0z>b9LQ{PB&NaDE~
z3a4b?WKO(W+mTQ;hzDWMb7BF7@>6$eJHlE270ExMmXXc&(bA2YD=qY{+mT0O;_$5i
zfy{hP!GAlHqxERwU1>o!@GzckqK1<T;_UTqp%0_Iahe(x1{1?Vab(g>&8JD`G>*J|
z1k`57@bt0@Y=x6wFHjd1#Fnq4&L6KVyI*cL^dxvHcA$e8BoHJ<Lh<sCpc4yg8ig*j
z#-N#+Wk53y*BTz_?R^X_b}qHlB~3`Xu%q#VYzBL>$rrn?pH-G_1B%-i%VzReCCm6h
zLsR5kW;Edh%KPItKP;7efP|?nJY2w|ceDVFH;a)~y^e`9-+@Z(-Z0Hm%eTr1K*?F8
zxzQh0A!%OfTRGISH|WJ?$rl5z2n+G~s&mcAkx#aLTw6Bx$XOS$vovKeX@YVl_MXF|
z4om%?`sYh&Y1@W*)k+uk9KNOYWl@{{!attw&6K~%A!2YU5`|^yN!#{+9iREOQ~cKv
zWN@r1|16-u_isUF2W%Dt=I5bqAvfisVQN1@p`_N6Xe3^53CvFbMRsgY0<@oKs`Y+(
z)A3R7s=J12pZSu|O%xFeH})*<){_(QlCXUyT~GVPcbKzeSf%}-_0ydLKi`4{aM3U-
zuNbM;&2uRUq%yd=7nd&|IFdz~AyqDqMryKc>oaIjTT=ftC_}xi=aj6tLdGwG<yFbF
zM^^I&o%jTH*6g5?i~25!11C(IR2cQTD)AaCAcw5x6SEnL?C<8jx1UM}{$QV|Ql{^Q
zJVce%OTul)QNGAK4d}56<DYjzov72IRU6&V!#Rf2CtzfZ*%S+6SP_qOI=RmhHygY~
zM0c~9ho-KpuVC38!Jcd^%18C0e{jkCg4D;Te(?5`b=Z1lN3bdL;OsP?kbk&rtF*fJ
z)9D`;ikw|{TYHh0$Fdu0=Mk8$E4j`drHZhaep>5@w+8Hohg}r*XL(<^r@IbGBd6oo
z%hiuLtw7XAG+ucCx`IGeL$@9*{GxFLk|!y|ntZg_>54}e8ryN<hnS2c%O(8l9+LOv
ziULG$?XQF2_~hr0caqzuaJG{Ij#Z!>X29`O;>F=^K;x0B_0NT^Y0^B;m`?WD0{bzT
ziTl_Mt=G|p8t_LtSm>^#xkvI!t<5xl8Al2S`OC=JsysAKsoT`<dM@ugjH$yW{Z$)i
zpAaudbQmKBL0+tb9YgcLh~1}7PePYbrt&^S%h#}cTl}Kyw-Ljv1O|aAC}&Q=`!n){
zZF@8Vs2RExL(k#lDW<6$#W8`@qTSA34_?(d@59X--iTs<J}?8G*$3t%pEm8y4jdPc
zY-Y@D<|cZN)9US!q`K0sCQ~+AP-M(W)9Oe(K|peKP$l8kj&~BlF-p5{u;xX5e^~S*
z`2Dbzw#X+x4ym;_3GEmdf3f-c)4lsmv<(>|(FHqDrvMnvwyLWaxb(KhnwMZl`wDxh
zoj=%`HO)3ruYJ4C*;jbPhsXO*9`VlqJsz>6sum;H6iUgs-{->WC3i_xWDQWcGH4Lb
zXA_yM?a3zO7XqFNX7pGnoSr=kOjGL4Qm<bHr#O<|Q2d5czA!rdk~&E5-2N855WGE!
z$z=Vqks2BwBUz6v;G>yQcydK))Ej8%7S~-izQ}RxK#>_hjx+fT${I>%+TP!(f)#Pe
z;cL-H|I91omfq%*)EU}u5t|5AMjI&UV?To&JPnPN(nvF67>_QwhhH#E4X0Yz(D9fB
zsOF9xHLC{ei_zES*WZXxzN@>hL_@b9!inPaI1MkfkA(MhBbvzAzQc{E3>qa5W3fK`
zP1HyLO$g!p)SZFm{7P3-tW1Nl2Bh}evuD0;zMi&3Fxk!5;x#7LMWHB52}I0a-(lH>
zS#(X?vp+%skn#3FY!x&x_Y#FccYH-Nvp*@_Tde^B9ziRyN_(;k<org#VL^0IDzN=G
zo@%~*&#_-bbS-j^d`vTl1tlo69p%`)?n}mAhxyV&*XTy`@vA=}gM+>icG-7X)IGgD
zyX@b2##Um4Bv*x~J;Ivey<}`Y5uFo33#*%NV7xw0&RytTnyUJNS27&(AgeSy0|Wp@
zL{k(I2NDsm!8~7sO{8^Z38Bd~`k#m2-*5*JhK>j4J~MpUZ87LE4biCFRR>Ug1)(4)
zbGDKa?h+V(8o<{NC_fKJF9$f{Q$xK?qDa*>tV_v(8%iVl`W-Zx4Z3Wb_Hncgkh2Um
ztB--ErFhU!s1hRMuXHd6oPY5nu|-t*UEmx=Wy}V5fJhCLGTpjB_2^Uy;9L<L1KFlH
z69gdfVI)(~m5MXbf<g|GBI=tNqj@r<tV6F%K9DFC>t5t8E0L0GKF@b|qY)qN@yZ|}
zjaS0Rh;exazp#SPN+^aH=wTNkUn{kCZG|0azLgP((k|qlG_2d4P8X849fT!uEA=%}
zyB@XOoy1H!vfd{ifJ6n<nM8Bt^Jsn<vFw1xLTVKcF>^zKr6{F1f|(<BZ+G7e5nH5l
z4UNOW!`HLEXwZU2n2__;ofN}va+rPyjttEQGCYdNmtO$j?1-LfN&#h;Pp#6>+}M%X
z^8w6xGq=YbHhZ<vn+%4D7<2<n&kwa92Ng49hrrmv1C!H1lHvrkS2?Jb+P!eVwt$B!
ztJgbF>`l=p7eE=ewuZNnBXkbYlwpXRrD4@HdvntfLzN;L97(;=bJ>+ny~;FjZX3pn
zD9rJv9t&F*ps8E>#!yx(STm7Qyi119>j1x@nBmeciG7RQB|v};?>igk3_j0h8XMPp
zuTTk+^NikMwLWmKz0G6X+YURcn;u7Z^Qht;*9h=JJyeJAalvF1_w5i3czDcP85oAj
zp4<}x>W^AUzt$j|CMl~o1RW26fCx2C1y@BKI0*uyuuJv@^nJ8X(>@PMnpaSw+%fl#
zvh(ih-K5Y=#PEi%l(hGZ#bXf$lqS4hbrtW<{NY`KhK?B;DPO^}{6uRRrJ1h)68KXS
zr>^+$ROl5DuN#TNs?U7g_Lyi<5eT<Dh&AYmc+jZ2S!O=lPMvT|fhX~x8?x(Ds6+~A
zh+MR=^Od5^z{{N1z|v##gU?{F+J%e-o%OOJ&@Oz8;G_J)Y%p&_)J#BP3b^mIk&c1%
zO){?&roaxxx~B2C1SBrET7SfS!wT^871$H(S@ATlSrt&)^L+VFTnfVXp4z(H7!tZ{
zXT4o%bMpI=!%^+SGvc;ZoTrtnKQ5Z=3^({BwXE)-<?s8{SlKQfTwZRtc<D}!liw*U
zoA1ner)W8sZ%OFmPYGwYZ&uEqcbntjue`+@H=8v+b6$|PE#&^Q+~utI^bcMwe0cf&
z)YRSWH!e0eq-OT@G}xDQC=5@3ur9MM`*!8~4_OX<GGRj_HWLXqvfd93>ATcw1o?;C
zoRyDiu{W7kJKQOwuEuZ3|5oc!kU-{L<FZKY*c_dD^t%o@``!pjv<&j~8Z4QN(`i~{
z!ec+`{gimHOV)x)zjh<ZVTb~FZJged0FU8pnAhhbR;|LhS7g{k;w`e@aPIK*_AWpa
zYQAq8I0oO4<uF<q&gN?d6S)|jxjGP_M%=fs>sD{RjY@PcjsVS$bQ&)~z-x`EM3Gc>
zP{_@jB*-CW@a1sft{J}P<&{t58$lLCM|5`&7yy|~=5$0M;@39aJYJJZe&r&nF_7Ez
zLP!c0@rM>4UDv4EWPAZ;G%am*DH$1=qg}5&;Z#p)6U~++pRAg0N?{PATE+8;K_Ugn
zLwIB@j9%N&%ywj>N|}pZc=(OIf{=mU23PVlzrUiQf*Q-ZmA?G}4s1>|O@%lws76(_
zYk>rlbaxU3N?Disp&7NqUEV{^>9s3uq^2G^^qYRo%rUXa_uI46cj<(nAoyLP`m?5g
z>LsHE_=l4L53$<s42^!W-<c<?9c=shL4sR(LwC=j)BI^BdGA7muVrVO9GPnTBr{k*
zb2joD53AAhRDUp$H&0A=$B?sd=L*5YF9a2wC!UW3sPpr<VQ(WHLbs2K=S#S3A}nuW
zY>*hTR#w>Fy@ACU-EgX#iMX<fYoL4|-;J5?i%K^iXEi(7tizvxMkG?ZCAE~1%iix@
z7}I`}+P+A6Ne_S$O?81szyGBL=*KjrwG@sgM*B?5{Tz*;MCgY=V(JW`lKpVc=Rqer
zI|hQvkdFFi7!;6mI_0|vqYtB_H*<}aUN?oU2n{?bAg2hJcPEk;d9slT*}~B@3!)%t
z`EUt$bc@`+N!T&*BF;#~^B@Xo$(fnRnHeymsj-Y|5TF!Qccij|rYF1Z+%<<zP%VrP
z0aA}eyQXot=3}vP9=P$-tL`bn@ubDaPe5rUsAoh?yJj#@p(Byn#9T<*MU%*|d|Xg^
zN<x9tf;;C*!gDFtl`DW?6i>JbvcXdw293LG;x#!)Y)$D7Fn7k19%-E<f)t}0L{dcO
zH_b^mLyIFB=__~Vpg=e107vK%vs&tC+jx3^iO^@?uSPV67qfPr<o6S8NJO2+<oRDr
zu_!4IV1|xR&mjV79%(Y+;^PYUOd3J^kQ|nvIU$ufGf%W|?H@v84!lv?6@uZl?zU1>
zI)f)ZIS9OHo*I6jH+j2blW5*#$x5tK#s&%S`!Yf`D2=DZMDL8es(>;CaOciMli`9}
zF;vRn%gnQ8#i1rvL3?`}-gH{kJGrj+_VmovOYpHx8n39;ONbizeK_>19{(;eF<l#*
z{Li1*1h-twS^hvTc9B@C%8+v%&xl_5Tl^B_+Cp3Eo@RFejY}A;egu4~cvL+}E%Zh<
zC+D<ZN8~9Y2)lz-ni0JgXtE#qgyj}DEJ6~|2ZuxEE=NF*D=VXEmY*sNvDk{<#!O+N
z#MX<qPQUN7VJ{Y}+BPnk9jX_ay+<&;+I_s2-vzRI-5Gnx_m*bley@&cfPuEpVRX-P
z(f74`kh8{@mBP;Nf)r)lq*{|ggOYq~-)#6|CFat;!f^T2I^0cp?I7PqLm7luI-i`U
zMiE*UXrhH?p&~nrq#?o@ZA10~HjMK_IN+C%WPL$xZ>_K{?P<ZF#Vr!TNgJq@0;FMl
zen0t^K&UkT_!5@%aP79kHd0;1f*zvS^5*+Yk6q(*xl!pchgJMvpZx~f@bX4DpG(HT
z*~sC`@2jhJWrGkVmF$A!OO6;L10z*u)vjre;K>a?Lm)N>R`=#F7w)XGnS_$tphcFY
zkZo$+5`0vdpojYGN4asj<d@HW(!c!!iyz0S&G#dzx0E&+djY_^@iDYH;QJmiK+Y%Z
z5GajZfqIbdxMUJ*cWr_jmS$>xKUE+Qq$*#&&3xFuU6zb#@HrbHcG+X|@2nh`9CEJm
z*4tz?L`y;Xg`&g>BSalWGwjkz3X(<$9p*;f3{OH1`|WfloexOJxOsSZ=<|qj{M?Y*
zcUhM#6Ctf#7P*dPKr-iM(B1KYk~Y_Dj>{Sq4P5@XjTIhqHvHkRL~Q+VyYH)2_Rw&@
zKB4+x0ZU=3!cmcpuvmRie4l07Qnwr7!W9&H_uS|kUZ;`KkbGmm_cs#1zTw~G|8MVG
z*}&!f=leQ22-N-ULReViQ0ax0Z$t-IN#}|v{<5D~BnCsIzh$}4$J`8<Hsr-6m7_)U
z-+qMVo{HX+iLMIxY)+k_T&;ZLn>p65EcesCW(MUC{DhveAZC?*BVjhktV?QTE?h_!
zl>hqq$ElfIipd4@UqAn6+MXAgCzsQSy9)!!&68KMuna#!qUAhDgQzebF6CA<>^@sk
zpflR<ef&7KFaO0KF+@N~g+BvFxb1T|6|P5A-a>!dCR6%C5j9S6HB1xQ-oi+QW+GlE
zk52A6gu^~KjN9%)rdyn-A+cxZ?w7}UF8iD7e)#`;*L|<oG6a@V_0yb;9%Bz%D!_5-
z{;+Zj2A`p1i<f{IZF$8)ObajHD0K47nJ}{D9XJcE4<#+tyWY{+4B-TVCSnh{1=IJx
ze?=CtAi5lGND>d^0(J?|p!#oJp)=+pfNPEt(}jK<2y{KUB}AFop)J4+ex=Lq;}F30
z-;!ORN&oXdKN1T6`0f9nx%vJHK>q~c|4$Ag?kyJ)TkjZAMb1bi11BKsuG%$%LMfLV
zPxLINP$_UgqY$rR&qSiUHUvEv`25sZze87)q1ZI^IeBQQdy-KLX5%4@iH4(L&H+D=
z1W$P6D?V_7H-v@_jaJ5Nts~|M)ZGhoxqibar5qUj!%XHjf!_FOa|FsIviy@)_3d+E
zQX2MnCbyA@Uj@km4Xf{%+Ka--rN5Q_kKTagbsYex#toys+fH3w!#F;a>yu?VYgqW$
zmub?i;){O#pI=6nOyANv=a|Cv6ng#T7RI~%gdZ@gdtM(?n*#AaTD6R;l>U#hhY#xl
z4YtmbMkTHb|269@=3nmoH2WEZgF-bmnBqVs9;VurG>p@`vhd1eH?s=^fJBn{s#}cy
z>hfs{;rI312VYJs;7eNrb;dvna}g4Uw?k)1l}E$)*Pmz2x?eFa>u>yEZ*tGuyW@=U
z>PB+tVp*wArt7MyD1>kPw8Gsxna;<{ZH#-qK0){d;szu%pr1c7KECwp?5W9Hhn8x0
zZws^)`FS(G*OzIkJ<I*b_SS{+&_3P9=z@TR1@sB!?yFF{(;T#b1X0+@1SH_t#-MI*
zG9Q*ShAR&B?*+vBx?NsI{~#eyI6U16X;ceFj+J%#F@N`%+|iF$p*f0;NWIb;WChw}
zA)?pK9O$ebl-rY?U*#~pfW=&fMpr%bE?(4tFUh`f<3<s6W+mPC8A5fAs@c~g8IPm&
zp%Y)xD>ddd+RP<<B4zCxSf~ZWwyPg!-hF#<_1-S@zJxG%5OY=U4>M0Ji!^PpMJfMe
zLGm&hDQ8YcH!h?9&wIVoW|g)Hv;5kAr%S$d$7xZM{dwlg98V?NTxu|siX&IdT|gf%
zC^D~qrb3bW&^0Oq06|p0?>;Oms{=MhBo;mj$O$e9j}%9dAu@7~q7%aOcP!fU+}13i
zWroMPFnbyRyDQNFe<1I`4{HMYnfXb>`Mq}Y+ikP=dpBLl1}WXWcB9Xkl_fC(_=zSM
z8<!`O=@?2`(Sq4GnDk{Xf%<dvAu;N<*F@E*19Hte@~>|-bU2zEU_r_hn$3PIM`lYr
zq<+hHwPc}0%+a1%>;2<T>FNpj%}OxRpU2$WH5|-0l0C5Z+F##jDw%xEFbB>?vl8)w
zK>MK0%$BE3u;vu?Py$4=tLaEZibw-L3vSZ|v)op(&xtj}7o-`F5r)j)F*o);S?SyN
z(u7}3EGyBS{i=CJ07b}eI<}15_c>#j+s1sx6>@?O3#&_hF+5c>YxR;XucpS1)n3%q
zND^h}#Rth;9#QZja|K6<c<2L<)x9Ty27`a})!XssprpJl4g7ITkhpiREuEcSw5F5!
zVxi~uKFvYb5wy(@e?%sXA8swYnx+cwE-T(~ky%US%b8E800ka`lDk!8ZyFOx3Tbf1
z*?<IKTx$>95Sm>V67IBp?i>Oz;|iSwZ;u_!4``IfkF(<%xV1>&o&Obm0y|D@q)(Kn
z@LZa4Ilum?^@UjNob2*5*vE4MLH_h3LRas$D&mRAh3L^g4f*BmItuHc=&$pI$b4?f
z+&PV*Y^lskN~oX40hYZryLZ9^?-ClPn3woHfykf$W2x0XZD&72Y!Da|%FXXAM?d9F
z%%rHb){@!mDJ#5B%=!ZQ8ENwHe6s4tSCaxVa_jl;Kd@yuZz7eLBx31;*?%j)3T-Jr
zP0-x${e~ON)J;5FRty5ToJ%bVIzU1!f>i3iOQtvapC8r$TGapg`G4+%{?AZ2mVXB0
zA&BlL@P_5G9EBt5PXB4uZXr4ssMUx}5{rt9TM<p2j)_kR?J#|}lKkswQZS*kw5ka>
zPdQg7|7n2P_iEqSCepH$kf))7$72Xs1!7lG8Vt0(AID=y&ornAY6>qvuUJJ;C0V**
zJFGM>TuUmDS|3N`O4LB^h6!|QL8wpnT~7Dbvf&{YyMWrjSER{D`5a^RG(NbSKaL3U
zGU?qE^?|Pu(A>3$NO^cK?+ST0%Ip#Le64f6VGT|$k5=Pi?;L$FV=p`SL6yb*K9gNN
z(<Zcxyq#gMSATy|^h`Z9rV}57aWEevGhUFTCD{NBXZNBr?RD4hw)DPhL_X)8Ju8kg
z>9QBuN8D1}>}d&VA*6ic93fGPc0wFD#uL#B;UjFMeDB^p){KC&256TY-klQ~x`CkU
z*#Ljpt5EBl<>vxAA0*6}L%oi;6HH%f=DxV%bv?{H2nbgVb_D#Lhi`Xx_O!U{g-iR-
z7pl_4SW**|pAoX9ToGK8p1p&8jtCPnxJ^cMp6Jn+MjZ5@F!K5Zu}a`Bj)~6Tz86qK
zCMgWTC`f`2x##K0_~t?ILo87~L^sUELe_x+3WhJb{Gq6LLK+{7AkvAG!1h-~roK;N
z9ntA7aA4->um(Tg&xX%j;yBGB9!U1IWQ}+U2aI8fJ%qRYlynp255v9B+OAgc*Cj<j
z8kso)Y!HqoD->l5no~dmH0KUGSUk8jL;j{3=e+<*E*DA^2r+T)cpW4rDtp-8kvVR^
zEVWG>*8TH>-DWSiHhUyF=>k~!$l;e0azv^#^fZ_t{c{GdE3ozj9WFH^$)m2UUzKMs
z<~v+W8C{G-?<zq4WI9NvI=DP8qOGaP;T*=q2HA8)k(eh?JLAo9u3$TJxxn4RG>L&I
z0P<Uanr}HFyrk+wgLL*7w#A^iX+<|a7VWQJF4JjFjpt;2K#o0R#*GWiqoM5(|0Tjw
zmCSTwF;apAXrflVTZa~f?4u#}=xn=14OZyAV2Bfl4oDJ@Zmf-TmBeC^eLVHM&@hO1
zh9YQwV0zT((PDNsCU|Y9eJ=j>)v*>m@d?#;y$g=;RllGfr>L9E_=cC;%=gfs<HSWA
zP?{d?qwe0YtXkNBD?JMS1IJ`-od=O$))VRPp6DH`PY-e3y}8GaI`QIaA+r^2b!&(F
z4)@p-J(`1X<3y}3>~<_hyz*K=rSguIrOwVrIJ{y+XJ(I|4u48~u<1l1{pzXZuZ?KH
zYAI)8x)Gwrb4SjF>l@y@bqpqht`!MTm)LAxClB4-A?hEtn4Yf7vbE~^L40hVa<ZOn
zA8bg&Xm&+M7Er4oKWM0<lai83ZD+^?Ax$~l@tD-XXlRXBBxpaqWaeUo9jegb_Fylb
zf`6sng9jmrbm1Am#pM53X@02QR#8PoPoot*X=iJOy}Z36Fyh0@G7csqRq6R~&rl%Q
zcMlSb?$sSwP4eWe9gTe~GBriXa}`YgeCl%?9pgHt@C|s!f)#GueNE^7<MFJvRf7m1
z3&0W{6ScwyDS%S+-L-0=)r9BoLE4>MN?M{Gi_VX<iUO*bEGtQFM5poz6XvvOur~<@
zD7oETW=wt~Xq|M0-@X9XC@07tT0vNuS{>;WKI!kNt%RF*IHw_%70ne+@POP(+#FdX
z#&zsL(_A9j?2?S&uDZT^A`pi5WT=}o2(_!MsEWM&FDjZAO5{jNQVD&=EFzAg5qy^D
zZ6K8_9M6zL|KzX4la&OD?(7lNmX~+;u-$jz!!axHv)7RhMFffhXKg|AwiC34($FpY
zKZ^diwyFCrKYW5ep_#;`mH16$RJFUNQT^ruL^ao1O(Ct^{H@dL5@we=&{;-?a0qP)
zwH6RjxnK-?#EH?pux`+K<-U>pMXS3rkVEF9^QI=!7IQqLG>=VLbXIKSM2j^UH^mK-
zKS{k2DKcZ{SS?vX_8Ow(Tt7CP;HcTHVPIfDX1-()EQ#rJ%A?UokgB%Aobnb-WFns%
z*%|!V@gbG)E$^+___HP9sk_-Uq>e6_6v0E<HrmicdQ+nyIS3^gR!858-s((VR>T-T
z$nOVdoQFOV!<E5Uw{hKju%8p51Pquh)st3^=m^d}_|;6)EArEQwbKQTlhPr*<nE2~
zM!!6nj9!#YD`cg8qPJ1aL_(xEQ-utOy;~s|IcjUX!!*JWSakqIKn5MuLH{<>$~5i*
zEnTvoD78IJ={qyO1|lE!r!*(81+BLUgL_QFAHJn?zkKUtmr71i*Hh>N=wgT@W(M5g
zBF{Bvi;>=uMrbqREP}Qe+f0n~I6-odUXzXozWR${JI<c4sU1NVn<MAbQ{C`K*4q;1
zcoY(+VPlB1SDR*_%yh+u+wn>5RP=a5hJ_J2W^!&)CmDep&;!%r<c2N*<LJo1m_%<d
z@LUc%5t+<VGPgk~FYM|F`Jx$s_wXzel%Vt)XW<}J)GJ<Ip>=KB1lgIdSq~oaPAM1m
zrO3~nAtk1nW34#!WhS=Lut3N-o8PBu2yfatkC&-9U%wzDWjM~DBn7KR=4EN<ByXwS
z|D<OaxdV&L!F`IbL%=Z#PnipJPYaZmk*B{}xA-)P_x124w5KOb#Y<)?(7@G#CH2b4
zg>Hw$pdNpU$%rmet^qxq2>pS5c3e2p5+w29bTRRZ4DEL(aA1pv^kw0=pks_WT4?8?
zqw+RnMR}kVw`*B};ieAi#{0GoD0LQQ@ZD05J-LKzuxZTR&g^~{#<+p<F{D?_5d^Q=
zeiWg3sN|Xc4;VtM1Fc<|{mtf<vgeWYq*LojID*uH`lnYIy&gInMY03vK@3;&^YF}C
z!{*b#pCy>za0}Yvy(y;6oH+GdNgq$DUZfNT2^^S+5h$$JO1*gGltxWmbehq`5-B&%
zRTv`VgwlG4jlKR+Dn#{n;3hADI9MyBmc}SjKqCz?@!LOd-tv@sD=1W|b&y<b$r2QR
z2HNSLXCafz{B{2mIB50}yr!1Wnum-=G!ZBYBTa+~-QxRbS9Ie^HQD0OB8=!HP73k>
zN(8EQ3F2s7ke!UP1fJP;EpeR4-#1y|p5fkl&~BHNm7U&>W$4pC2|eA=<-r+CIutvE
zcK}^m-vUWwRBc%u8n(7O7?c27GZLGI3r;R^sY$hG#)sn(Pzm4YvCBx6r|EdD7wqsw
z^UzTgWa)s;2+}Dg4K`2pT2+>wY1PKSriYMNP(VXt>mo{~C~Q#X)q)<`+hl6vqPChS
zYQ$d`#NPsZD*0oGs3stYFm3S|+;FsJOVY*F&)mTLfB*B!<<z298Yo)pGeeR{+}2Vn
zYi4aOIQ7@sH1C6DHMs+DB$rkX=dl8jGpWkrM}$i<t@%nuB!YUSC6}g<3@pbM8O~1F
zsnFhxHeOH5ct|>vrzgbLksp+tZIeL3%3mEvtsGUl8)<G8&ZZcPUh?`wl!~ZHCqZ<U
z_Q71J(zcgMADkTU&XE({2_*+@`#_^m^-*d?=D5h4q{yV@TR3)_I1}qN-5EU1!<Fzo
zAEE>_$$1D07mjP5nxgp^PlqZF3%h#XyO#@%>CAvjq+>VvqC?Kop2-+?-=CYv=I=Xo
z68V2)J;G1@&XI!$pQ1yLmO|1X0>!y6Su|SY5{RwsK~JDQp~<WDUja3e3bfn4PjRzS
zpV7@-AmdwXn<upBi#wBNIg*1Ggkx9SGO5j;yMMFmvu!R;Q_k2{4MF(!?7LevpHZ+f
zNKnOiE@XFMHR*3cq$9F=VWn`JYO<b5Z->#$<}(3h7u_xbhqi+-Ghx)Cyn>lw`WaWm
zMS{am*l?7eGI^xsXZ<fWR58#LLy_1==PT?Nl+kX4m58t_4oC9a&Ms!MYPB`CiN1Ai
z$T<oFX?O!RaggO+PuV<RS^<`OoBS_7Ci%o^i+8k=OIO3gZ8$r1o6-mn&!yPIPGm4c
z9*1M8juWY56+lAWH)ID-UX|$4e%DP))}VtXU?p83BTVGKtIvaOwKS_y)eFnor10B$
zxC0q<V52Fo-=N%Z5~dUN!t2%ncG_NA`w?Ny857H@CU)yHIF){iW#7A1=rK4(XB%S>
ziIRe*oVuccV%s~%nJJq%3ltaCV+F>CFJdbf(2Za(s5EAstaY4H0j{*S`3zI%^7>@G
zt^f1#g^g;etV>Zmds)VZhK4>LeV1mYtK3n&lv`%>#(H+b(eSX3<)TTX?tZ!2ChCqA
zwW2MH!^<!b>7b^Z@Ame~x`Y#>53tH+nm(8MMJsB`eI$NMp(*WsM)2(N{X_QtS7wr^
z<l9tJ?WfS*nL&|~E6Juwo$@sEGnY)%(03I-IF4nu9EIzzihkNCze#Vrts4VRE?}SI
z3^D$k=a4sB&YWw?NG`pwJu5w007-^twKcVM=YO@cvpxi{)I9s_?A<^&jo)N8%K4(P
z9Z|l5dYoUH0VT@X_VPRQJJI)vy8vX@@Z^=@Dl7>b?Lk0Y?GJv@pJsw{;yYL{&<Y|D
z91{yPJHAptxEeiRilC)x%!z=N;{;3l=pScs(%Y)?qdM!EQ@^bUJVV!>4k!L;aZDFm
zt>ig2GZVOY|H8$7LJK5S{SKREJUq07^}C0W@8@5g*LC%o1KYk6%EylzWIlU1?~p2w
z&K2RC+>(n8bmp$(R=WEA{vUst7y7Pw^<*K}fsgOf1%3Uj#a6pbT-r9;Xj|%Aa_XZP
zm&oYHu~6eotM4|eJ`O!&ZD<(XU@R>usp{p;qmbXTdx1YB+cAiN-xxXKtP0mnz>#WL
z6YCqg#i<kOY0gR;gm`_mvzDV<Oa(LjQ<fgf7e2oH5i=tG`U`BoWV2SOV%;Jy-tLHZ
zW6H>Zq4WFJy8wsE;TxyM9!S$O!p{rIU9a<GOJ<RQQ|Dp-YSX?8t0zPG)-M~k7K5tq
zwhX-fTGvhNtf~X*#VqvK8&-tgD!RuLQdoFM^5DTXlZ5PX!z=aTc4Lq5Rdv<&>X)yP
zsX)S#(30mQD)tegKIQ%UuNtiC?CebGyYtaBTP=6qccROX$YJ`=7()Bj;+e75MdHJ`
zqv!%OP!00)i$<{g8U^(a^YcRH-@asICL~nGt7=n>)r$u(aiOiDqsvtthMTEdbC6|f
zTE*VMk`FXcwXw-SQfNt&ipEjf-1GDE>(klvE-kVc(5YN8kk1hLFxe=?((r<lO`ctw
z^;$ckuU-4@kGVQQCdyW&bKc*~BUF3dd^q&FqnmkQgUVI2kC3FNZWXs5Pb#qZ!5?)E
z$+CrcF&*Y^X|Q--0VlJ6@v6Bcc^PhB5xnvL#>j1yknC-BGk~EF`;K$^YC)c!VHD~c
z`ns;Jb72?W&WoWb0gyuKS!b`RtE=;BfAi+eT$7h7vfg}gAOrlpFgg3pfa@{$Dun1c
zqq?YmXl7gGV;`U83s-DE2q?@_&Ef078D902k~17Y0Cw^s+yx;mI-Br><>}QwF9vrC
z?f(6XfCV2y(4Ij@h<0i`dquXrEzacP0rZ%SuqOHM%lm8?dE?Y4^u4$R+cL};vP#x%
zY~3q9Qew!mCh2vsxOQS<;?ZIqU0q?l(v`)7Rz3A$D0uBAjs@&GW&IB3pNf!N+D*Wg
zM96X4PAOXP)qtHSpGMkms#+kS=A?eu0=_bJtK03{kGJNaE`9pgN7fxB;2E_M2)z9G
zWYs}nAr0twJPm}fDrxKX?efTY%$j7lWIZ>B!C@4ZOIib;?AB-`GHZWd*rB`Y(C?55
z8HnaaE-oY7Ccj-~4_jJV6muNxm7?S&(O;l&@nTZ6a$Gdt*$BdraY*G90jX5lEaKb~
ziB5!xe&4ZoP#Ob*iL#pjv6cXx<q!A&+BEy;UP$acwU@!T_N!wu$`F#HXaHEzXitJu
zWHj!17bt*A81Cy@t8A755)fZG)oec+*lvv$x)?xEg2@JzQn+}PbLV2Q1@_{oQsLuQ
z)~Q8>!dpyT^vQrdQRxrK+P=Fu@>v!6uCEuyALS4A)+ZZQKemwO&P3bFIhzOpy#+1t
zYL9-q5;u-PXAvkz{k_CdO9NF*i_6#$g~QGU?{3jnxFic*DbaAb?78sGp_;?O^c;W1
zx*PiR6bs+mW08uav~&y(l&<mzze^jp(`fiPwM))bR91fdg(g4Xht<3!@B1JATmwVZ
zv8oCfesXc*DNay7zWO@U!bgMm#^F@<3j)F=iu3+;k7bYSp+k?hn!bCQ4uGRG=`*s;
z7<xjf^@ATZB#$0-&7K}NN;Yn&=@da|ufaALM!T0GdUwc?5t)QdUS6I7u+MNSj=@~$
zz?05@*e{0ms+2Mjhsm^>1Z~sqlP|XoAtm76yZ4=e^}&PJFeft}DyKt8<129z-t+N^
zL<`B59XpgUI8zZLtgDbyH393RxZ2||r_hbrA>2dm?(RZ3L?s~Moc)<E{85M5*4N)a
zBpQcz5#=o|Dyl$y6E<%<BtF_T15>F3D9>--p8q=R8z_{8HRU$SVEOcSt%{5-LC^)H
ztbi{S*NRc1u>k=q8;0RHs|NduauB<6Zg9R#hD_f$^C(Nut)DJ3%$0V=rG8qt5auuY
z6#wUUk$8}&!2q6XPGbKxUlh2vrSIt=LF(&^b;IR+9|8q->vBXO-?N)P0T?sszvsqt
z9Y;+PgJZfb<br~{{1A>#>p1=&7I9WVOKv!E=~(Z@9Xh%50C&IW8DzP{=7f5OWiczL
z{qT|!xr7>D+&xsNDiF(#8fN|ZT^2|CI5DWuHGd?dkx(~hod0lDy@OZv@j<vrgci5b
zt^Dm73)XI3yeJo!H1-PEevig34`1xv^r^|Kl2TGPkYI)5tV_4hl9jzpr`xJMm&#+N
zd<ArG#I?dfXUnQx=b!bAUobZx0+wmiR(Q9lu<-nZ9m~7tI+SdDcO9o=C!nQD<h1b^
zqvFV^9VhAT?%^?vXdQ!MAGF;QHd}^f$L~LINoV4+#4g<OK-extK`>k2o){GsWdq-d
zVK90;qM{ATQ^-!_fwl~zRV#DP2$FZi1CLF|g3iU5i2l5C*L}dPolq8Q#Z(5<9}&0=
zuw+wQoAGge7&P2He6bdpt4a4!)u2Y4FfX?W*^ggrLe1zWsOMT>HNF=XCn$&?z-#m%
z;#z;2xWum8^O?tgIYwSyo|ZYUM%wWuOO_ax-@k(TIt0{r6h0Rr#;eGiKLJNW47{Oa
zj~!D%5`lmg0fXsJz<$fI=E*`nHGYW6sBzsNuKji0baLR@<8XFq!b9OE0LU01LQzQd
zcUupHqVXUUB{fNh>5mP=gkfZ$^1zUA@g}T;40z-!!RZhWwiEs4biPujW}IebO}CYc
zS16r}fKO^b=0^gUjK=fH0QSfF>#re^k?*y4+`)3f{E`EpdQ#Dcv4r4X1Z2P4Z2a+9
z3D6D8K;2%0`<{SvZ9I1MjZ<V^?<aC0yi)*>m>NF&3bY&@!NR8t7o%%Yir$H?0)SjJ
zj)vPVE=uT<y65W~1&nm_#*LB?K)(UqlB8G0Y21{dgvF|K^ytwm6W?45nfG#=vxL>d
zHPG2a0K%P{!SUw;|IGl^EN=Y1`<WluZr!>?BOm58&5m?uVOe_)%Yocl?`NJ^10~7{
zLqo&+ncuM5U^Z89OKa<qVhk4xMPL8dMCo}J0s;cN5vq_lwI<o2hVQCP2oAQL`^*&>
zdKXI6^VS-ntH=L(RDXT}{Aclyyn=$o_>e*60OH<wf*_0C>{0%&Ut#{w9S&LHsD4Cv
zMa|o8LipI${PQ<ob~9<#-+uKFyQpQXK^-dLjW4%NzU5%ae#Y#h|NOPZDYEytAD~h+
zq;LBs<c~^SKCP#bbUo>MEOu)*m#_NI7to)+MBscoj3JpR_LFGQrGsnr$h^xr=k=JX
z;6GpRD}duCWP^=Rd)a$7dHv7dcy<fnY`C{c9nbtq?|c^iQv&}%Of$dz41;GoxhRs?
z1C76*Ip>+v!#PcV=3)Q#etXcUYlw{Xy9^HQ4e+7}ZEHJi4=Q~eu*DFXD7Xa$1<RY}
zyg#!Q3W~q~oH~Olm3&a~gx9dF`}V)QKd%0V|8~1s<}LVd@5r)v*-4r`^!J}NTmRdg
zV39ca+kd%smdlU++p}X~Idz3O&Hn!8FXk|j{qKK$ZSFr`{O|YVpF94~Qu$wNV$Ifr
zb2GlpR*qKW>1j+yxS-#d{~1L{1$rw_t2v-m+W_MW96pomDV{vXNSZETwZ07i`C$Ox
zi!M@GdwC8t0iKAb`fAf4H@5KcmAZ)`>dadR5V{8S`?pOlm)COPZ+3Nb2b}!Grhr5#
z8i&)BUef#={|yqMCtLZIqFL*R^<EpMk#2sPk{Z_eckkZyYKP!p{RF-GM3j7(gw@*n
zg}3mJrC)b*sg-^F-=UF3V(n}yg<txQpMS#0|2_)-3EclBTbMm@BrYHE-7bl2HJt`c
z&uQ8?_6kYWDy@kaI4V@mX%c4Pt{NZh3r6#qVw61cp-cBOzkjx<u=#__Pd{J#c{|7D
z-y|%5GCeKL@<lxR2bK>mldIkpHcLFe8^*$_+?M-A>gad#T?6#_DEGL0{$5~U9Q8yu
zPj6O={)9ON^CjF^o{3?<Wr7Zlf#q5kB$={DkJdn=q7-{lKq>P0K0vmL$m?0z*dmZh
z&&|mE8D`#+RI;FcOP>IxC}dd0%?lO7Ww9-jzx?vc5R@bODBon|<P^=#A0r!6da!@4
z5+)pgTT?<?C(uYXO_=&#>E*k;|He<Q=g$wc$;_g#y;aa3U3~xSo3PMOZed|1U^<)O
z3Wnr89K}Ku0HChAClSbcd(CWN6mi1N?kPq-9Y*?m47{5HGJ1R;MKC~~e41+5(+V(X
zsiv|OU~&d(dE@FMXTN3n?w7hH65P7I%Yt;Kmr2~ele6Ry8=pows%U@=n|j`5siV+S
zL~TV$54v-s!Qe!n2)WT!6BmQI7LkyJN2@*wmp1-tGwro?t~TP%JXZ5{FXoB}Ty|c(
z)2dhFfP3e%Uw{4e6CXRsO|wr58X6k+Gq0_IYxe0s`tQ2C|MH3N`J$3{nW&9E4>SV1
z&9G}eeDr8awHsJU|C*O4w`wfk`Na#YIy+dhNGy<3BH;rjAcJ;-oj$Ohiz`CH*t995
znp-~f<{9<={{ECM8~7^*1m!@IGH3WrEo`fQ`#LFV(Z%|SZ+Ei&gk>wtL{r7i-e%>|
z!KC`Phh$~vU#vYb?{3aZ?}!xvE9V{yi-_R(@y8$Yp~`fDW5#Rp$8|wb#J+U1;-kr*
zF4lt4aKS`?6^_$`!N!g?T21&(J;J=D03E6u2kkO@0T;zb;PKim{msnG@<A=>!~`BM
zK8NDL;_D<TLyk`-<FCQ!o^!bcZ2Y^7n3CoftaSo*0*(tDbsL{?_+&183Vg0O_&GS(
zZ<{0|d~zfot#SN=e_-b%FBezA7!06XV8guv3nNAR+XNHT(L%t;&&2RLmNnln^kUy>
z63|(83Wdg%Uac#4f4y>bE!Pcx^L29`Jem&5+4k|EX<N*pZ`PUs;{G``Q1sH5-|lCA
zcM|1sIp*jL^zdUs3#507@5dDs6as4&ojld}%oWs#I+kuVsLCxbx19?f#h@^5KE9*4
zS)8f%)05U>p#F<WOGnaNpMs|RaB&7w@=u=QW|EecDIkqk5Of9J$y2=p3K40b9B)8y
zyno4=;r97e8OwuL-gN&33GpD51sWaDh-&bprIPgG$C4EZDV6CL^r&hB(;3px5hQAL
z6pAnfAfy+ub<{g>{z+qPP|L&0c#Si)C}$%PM~txg`P?=>`R0M!=bOiKCwZ?fY5hKa
zo7IA2+Z@%$u1yR|>#uWMzZiQK91j<nw=+1wBrY}z2&@fAcn!|(zOU+!S)X1BIQi!*
zxYiw?kH36%tCS$b$#alX?6&-TGJ6V`p9)&vvnulOJuZ%YIh9a#{mb*TE1sYO4Zgg8
zj+%SXp$WL7IKv#VC3zgrDi+VKy&;Fn?oHpG;^XB#0yMP>m$Pc586zDafWHchO^(1l
z*788Z5mBnZatp(b(34+#@90~Us$I|kSLDo0nL1jnmRQ@`{nJs`{gG%kNJYt>0Hp0C
zr#Ikn6vo;ZhsA%~7FKsE8X5tWGv3yB>sVN>Ji}dC+WP%jpUmrAK5A|%`G08va^^1H
zTG%{K2~mz+NdFE*l`VrQpG%u%PxG*@6j>tiXbsn#_`AO@a7EqlURp#%geGECKE^y$
zo47z{=bzA$U1K!7cF#7G-_C-kCKteG^^Rb}hX7yNWNAGaZNkW-OIdN}g+Ku06%=&r
zb-jI?(4Hi#J}taj8Qg;c(w3#GR>iRk8-9ZMS_P2!B*z&CF(_2hJKEdY^wCAX&#HYh
z%Y&Dzv52m}cn^5W+<G_}sHkm^+lI1W@dU`ZGxoTae9!WQ>!azClI1(k-$+-#bo;A1
z%k#^VF2}t+JVIzVY_N9yV?jsTrK<Di?mPP3;-3m{UAAw_y-Ee-7{%GjhPE9hjVE0=
zc8(aT1JbL7K~G=x?h|7ICg<ZoP`xsrvuo2PDQJqy@r*nJ0%CC!;C0UEO>vk3J?x&7
z><A4j(DzUY3!hs~P6~2za@kXeUK$pmmoW_TKETY6%xwUeENnBRzm!+4YM>GLgaH<o
z4M3GuBa7ID4uL6+IX1rF{`o<b@42McLf-dgKUkG$LPY^$ZxIvI{A1DT8r&_#FxjQW
z7jRTWB_#<}4j`BoCdDG0+n}o&Vu7{r50{~Ldk4=|oOo}B0zmg>*zG;NgXOZgfTPoI
z1!?=VN}66Mlr+zox2(<dyWeq27y-Rhdvex=&_2AaX11NGv>RtNpe4}3hl>aChDND-
zzqJ=(ZfDRs91qLd4qjVtuJap}s*^PF{OG9omuKH<nHF#R39`xyS_y2|Ia<H39dBr!
z4hJ48!b|xs7}f}a>hS16y?7Glj)kv9TOHbjBt&(%y-6?HJ9dBF`X;9!EJ8X2NT>M6
z6kz+ov9E1PAfzpN>Z=Vh5m3I!a(tj;6%XnB+nr`D2Z0i&D-owm=Wb&GtY#}A;j~H^
zS|V|&pYH7kHG==aWfK2Ntqp$Sj@j4?4}k@EZdNMl?jfKVXYV+W!(l|{7%)-Mic#(l
zLqe2bL$GkwZbcYFXBpPU8><!I@g|*m`F-U8Aah?Y^jN9e0KP<e!MhZOQZOw29W6u;
z!MOMbd0+uyd#ml`-_F7EESSy#-NrTiA5A{|_|f#6?|vQFIcFot4Scf-0A}~@+_{gb
z7UW$=$Gp<RgFgbamV>2N2*G6l)sN6!I%5y9gH!R2DS_E1gDpI7Q`^Ds<WEhC%_z0y
zPM(xodKE(nL>F@IISG<U9;<FCFgR$TQ~?uK!8}{({rJL{<p3!etQ=Ue0_<9ra&W`|
z9Z~{KERL`8{<_|hdZ3X)wj<|;KfO7C#7Y_Oq=d)!S4IT^3T6?SW9g(z^o+G6A4O<`
zauxJHe|5a3I>4{Zm>C`oA7LO<`T+9B5jEylnQFVPeYq_O{4?FyBYr|BkmN9#TI|Ua
z_`(&0w-0yMvPI5hn|D?55Wr}l3TG(0ijUZT=j<hH0Jy3PnKbUQ_*A}+z8LrvqF_Kk
zbU{9K0`6x+c$7vstoIs-i$HUdHR4CmC%meT690lvsfttI3121&ghtN)i@o;_>iUYl
zMM;b?QG-crAOb21Dos#8q!~rAASxm#RRKl1bO8ZN6hVqu0F`0|>C!u>NRuYgyV8q*
z^bU9JuQAE*y}2`Q=Dz=Km`tJ&;8V`oXYaMwUaKP2qC<gthTV|(z_I%jJC}FI4pl4;
zMM}QVEh9h{WNl}kz@w+eE$X+Ny<%TD4e2Rzi}mxS+eP>Y!M*my1b_u)fE9=4P=;D>
zCZu2qiO$NvHQS9yq0|dd7!yS5dHwW-2><iK$h4fJ8=mc73J>@Xy>{!?t=opanzPI+
z=yN@9<1};LJeX+MYFd@|v3Nq_>YdDdMcWy)@G8+8BfPc?l0a1;*;)eB3d07|G!Dk<
z1$88hn!jioy&J5iOd02WoMb!RxhVfu`@r<c+y%v?Jp;_tDz!K*&*-NOg$+M))Rc<E
zJhiyyFzfx9rM{+=-Hk2B{cSye(>%%C=?MG0rbw?$4K61Mxmj+x`SfT;00M^s9Tf27
zM-m4Y1gioaEE@TOX>q5AD^D)$TgQ{)?AN(Uyi~;eiAAfdx%Q*wIZZ<ydlb^H|JIE5
zgyBBn25oKa?;G~$U`&YF!S&`%3)^B))7gumC?rJ1GL}^C;KTu8Foe$Oo;hP%o`ciC
z4JdX)%<PL78(TmpjIAKDT%dhp;M}yHnv5W@T}1Y#$3-}aTP1=}b_h7Fs4}50f$`Yu
zLVU0BU4FGnWqWK>G4zZ~bai!oD?2=^YPT$=(S)t2;ZMoR%2s#VqhsR+O3{hRE>EGm
zRnP=Ky-okaTjjaBctgF_Xi-E=Iv7(ZoKgi5F|qqT&}y}o3=jtQqQg~|f)!z6M=g$C
zX$OdX?nTPe(5-@r0ynVCYC|)@APpK>zhOh$toi|OKK+`r$e!HL;jVVngj$aaJVSWv
zN<vWR=KykdA0D=>Gaf(q(w!|w{);CoZWDamH*x*Y4DdQ!*9g<Jqu8I$q6i_ztI<z^
zPkWSpCSd6XgAP2#pg?!%a=Uv5_;*lhEgpnM8&d04prhW|9SFXT;-g6lHO`Y@@S4~e
zz0^BAn5c)hK6Us<u@3je97f{rw&W?QB%Uf!R4!~hc5~S#<P&JoxK&r5C=2WXMN+;b
z0{!PQi*;OF?g)Jk0w7!}9^8RarZ%s**t(|UNIoNz7B~5KCOxCY0bkWesokT~{1%D2
za#(}9xx82SWp1-a8NCi*XBBf}vk|gklWdwd=<4DUz4{oQ+b@<Y??Za5tthb7wB%N%
z##rFAvB&g`6Px(vaikS2z(j&dY|snVG%hHcs7u?$e-H<M@v!(|kdxjw@NO|*CtiAB
zi;UA@uci2AzjGdb<M?Kw@jtu!Cx$l5i<cV2K8!jc-?pIfguGSr+n>o{^JedNTafA`
z){SG|9h%#iQ!@&M(4^^G0!DxI*9P8iDC#~KE2G(iQ4G(?83-@sZ#QfHvo~W(d$EQH
zLg`bqs!c{Ru~Bmn^7(qwMYkYMS1UV7{<d;uM_eFCwPs-)VYl=|ZDY|~3AS3kt@Ziy
z=d6g_!{;vUSNWc79Sv6PiZWk53=iMT8j~SJI-7}lv}@hb>ad$mL&N9}+s><dUSjjg
zce>Fk0Ue`J-ziKfq<K(e;|*k#EaXU>VZD;?#3`2RpDMKS*sh=6W>o!NUt=k^{FlM4
zyh*3D<VP?Tvwb+279S41g-RU=Y^%GC(39$lejOl{-_*zDl-lR?;>>xccGIjc3O_FK
zUVVl|cT+6?iP*dMb}GcIx<2t_<9;IZ%9pT-n?LKaJ=;Hleq?f&I0Th0^-4SMI;$ge
z@9B9}`RscBn&>rAA~jwUYA&`(VQSm;<Z^y#$<aVmDjB<!c4fv>_kN~TefaD?@9ExE
z6LnT|`AwC1O&)mLNuV|44aUa~IS==Jv78-MeSG)%33-GVJN1`s>&>4`<+)%5E5z+p
zu&mE>dA&i&sl+Mb&F!TTqNx*`6@U5}&HpL{=Vi!I!E4nAarnUg{WSZ^NMcuuEADY!
zmM;Y+O!6dPy9NWD>X)afnT9n!0-~Y;zF`tgNRUlJQQYf8`-<5ZdQI}6br08CEVjmO
z|3F?;Q%NuiqXs>d`(QO4xPSltfkTHXz&S_XUsww4-_1v#{8`sKJ{fOP7UWKyU1Wp@
zcoh7OpgINBBno|gLmW`llQQ>g$aKm?p{JW_qUHV=$660&Zj@23tq(y9pIS`>h@eHk
znTu;Tl#ePgH_UF*Vv8e{4s?7bVP^G3q-KW{#!^$Ho=RZQWvz4FdMy!DLTApNwToci
z|1LlODB58nAaZ#2U;|g;J{}<mmbUD`cxe>ET#dA}kkC;$clzlkpJthaWke{T0K58N
z;2B|f9Ba)ED&{-2To=+2V9Ax&)_(H)(xv|1R-dbn0uZ1X2oSgRMGVUKfmPz_;h~H(
zw;c19zi>*|4?C3HgG5;m)Gp1c$I!3|b()*ysOWI5nfikC+~I@f>4Y-{E1YwY8XZT!
zYLQY0IQKyi;%%3n#j`*os^<F&R#ZS(D5JMtM@;O%`Khs}if{R6G^ynafQ1;UilUp3
ze7RV^2Gb_bk!|fp+O3S|jSD4)3-m>Dy9Q9Jo4D^0X1l^_k$73oI?nm?xri-AO}E*L
zu4!BA)}L6EYnH`0C$py`99m6rVB@r;^_h3{HLrw(8XCPq{Qbav(7)}k5S8-JOkZ(w
zp|#i2UM?P0E}q)mjO8h-AgkFb%lV~Z^A@L{eI$-`&1uaTT1~}44uAXV{_gIK;@TS?
z9-`6{txL-+P|2LTyYc9oPva<BA9C~;93N8Am`Zc7D#%9~xjp>AEOHa|01LF^NIfqH
z#q=T1fm}legff6K9D6~~Dv%D8acaci@y*>cXG<zLbvltDSSLm3A|=y^yb>s88j!{R
z6RFDCg=Z=ks%bnwixbSG;!M2uEY97n1Qo;?p-Vf_y*bZy0+Lx5(n9NNM$uXGO*j*8
z(0C^T8cjK<<opvvfc!<pm!VyksHvv4@VZ%+@6QoJ7QK3AjmF@Y@5MGUmI}s<5*t|C
z2@QQfW`-)j82H6GFf%dAWD@+50(}kiTnD<2t4sa%*(L~6QNgPxTZ)$phJ(CO{<slN
zSoK|IUt`l+r@DTpjfhQ}(eKn)C0Wexp1$?`v(f9j&((i!T^@#kB%{f#ZmzDvpbYSr
zG`&21wIJ+VN-c7Jcc_n+L^vWiwG=CPyy+@d`rRV&Y@Ua%oStvy8IEFuMOB~kl1|qr
zT}`h0ILUEK<z<uEr3p2QR@1RtnY}oMdRI@>*>TCuZ^Wi_C;r8Wr1Os<LA$7|<2Y)$
z7NQRrr#>1=w7sV2-CV5H`Zg^02Q>vou5>PQu2r@p&!rM4(U)BlW7D?Fz5UL`o9Tt_
z;$F9I7fyDRBt`O`T}_#pH2+Yai_+61>XrkF!E%W8l?_6U{2)1~n{(>v>TZ`5C<H~-
zf2toiOgFg3mbMDWmWd~WUVN6dM8<`PsMNk45$ef`$hmE59OdQRKwEV<9;GQ|(?pq4
z>{uh`ZX63oR8=25<K=Cu3P1T2Ye{YNG~ODu!uwi4S?WZ7VjXzpFeSx7XZpf_OlY~b
zQu7E(hM`_g8OJmbz&_p+r9EX_iy{^&={X9hNZalO1VkbP_rS>UI3{Qgl|T&iE_kMr
zePvL$3qcCkiF|wqsD?0)G_w1QoDFmEDHo^ifM}xsVSU*UVqFwI+k-K1LcxIXK?Yj3
zEG+sW;J;eD*{Q?aOc$k0ShhYEJF5VinyM%QTAj*Tya;Es&z~=mnSS3EjfB$<1liT2
zXuEU+j)?-t%{acp1(30L1MCfK*`+U5>-DbKo1dT0?1|d8?VcUxRlB~71A5X75o5WP
zS>%=3%XL+vBWkh2Ecd}{zk{9xTidd@X>16T@%Q+8FqoQ-LAC|3>^O!9XfpN~U?<t#
z+Ku(~R@&-{{;;*(WS-~ebDTA$$#1qf9kk~2tMM|+O3J^Ptd_BWFb}LU#_#zn4}S<O
zax=;X8{XrgcM0CbxEv9H6-uBp;evG@PRGCD%$J;LV@=&n)q4~ymx!0H7B7|lnv$>c
z<KmfQz8SJh7WBOG$GaE!<Vs<JFL_Sk@Zl<y)JBQK2PV5b0Z$SLUoCwXsp1pLMSK4Y
zP2=>yNjZ8_V)jCW#wJ_=;(1qnZn5;Q{sW^d7)<zpmuJB#1F1!W`lRTH3>q^_PW)6@
zSXd57+{nL!7c3346q(JN4}dE;V4w!5_B}Gm!QR6F@$qDV#UXb+G}oj8Bp<^lW$;IM
zjBnkH3XOQupzKG%=HrBySFwy8GpBSoZQrM}^1fUCfPm&vK$28k5@`)incl>^E3+W|
z<BL<^IWTYvsSGsPYLhL3=>#NHf3s@b3k(_TCAbnZn9*IFzmcgIhZK*~&`8qh!ao~*
zu~-IFHP8uA#LVDO!zyVM9pzxk04>TnJ%{TzIuyA~wlUw`@P4{eRx--I9g^0fT_0WG
z{U<J&VeR)P>N=6ZL@C|#7PId@wwlH$ZM?rt|8~ZB^P=E0h;q?><HhJ%s1nH!f~_0&
z?AcSZg9jl1H&g~d-x9J~wS7a!|5-6hYD(2lKgir5v7(p}pIWo)dzW8aa-MT2anvg_
z0!wqP(?c$&J5P7jRAhHjC18@=wj!#sH)5hLEKl*}O}}%T;-#q7ILx1jBRT%9*lVeS
zbGFjjW>k3zR_e$U*f~<SKAIZt(&<@piqfCXr=IQ73dmtlZkeX~A2|-u+IlBUcxWIr
zJv>~C<HlI7@%RL#A6Ta90WG~kagH-~&d1BivtFap;W+q4=x+4X=bp{{fsi9^%z`F@
zP*Pyc*0+UC5cz=cFVgd&#eLpc$i4#WVMyc9+oe<->DK_G_U(VSPNe~rY<F3(XIC;+
z;dKrns6pUt+7egTJy!<9k^s{t^dt^>VgfU|%O?$(S11ChLIEXWGf=n;8u@XlmIqso
z+tVW~TLB;q88`r{UA5)ZrLN<eh?N1R+K-Xz0RK!=nfl~b;nTZQ%WOmWv!I|IlB1Qa
z)$t>KI7WF#enetQx#_I;F2So)yiNyai$1pY^X4hOHpbYe9n?(yyg^1e2XVF(jId}_
z-T`n41hq5hp~(ki+eTB}dSE@_(lMa$5VInJ$qsTHe`g(t1V^=JE0^lZ$z9tdA#B1T
zrXplxKfh^SJ*xAsS;wHwA6}XHM;{yEE*?nQ##GH!*Q?>6@p^u0C1b~n;KfFxGu$fM
zDyJ~azv9@npHvdn(n7H;nLQh9Et0{=HR8lJQ*+Keo48Ft9^h6$z0(|?k^w>&nS4J|
z<KEGNTKUc-K$-5r)ux`0(_Zso{{qX2*2rwV5cUydpcJdfM2)HELz~0Y%vZ=Fu)~LX
zd+>}ew0ACd&m!(qdEx7=HeXz&4GI9}sWQM5j(M%y<L7;UzQU%tx%<eU&jDjn@B{*6
zogt`Fm0~Tgb<f}5*mh6uspc@`4`u?@xo$Nj8>28prw^!khtq5t^}FEtx)-UknA9xz
zd#XSWGiH!eR3cLJzl;JJpqgSC0>G%-v7<qEMDw$Iov8C2!^~f?JW6>-G^cie#gy~y
zZBD(27|Q$DK7Idnev`_r=U1w5+2N=Y?t33b+Y{KG`N_j)&jxW^`1sun91OmUFWFv(
ze&XpjKR`I>Bsfe7C0wy<eNnw$(9M0CP|PLqfR}FwMQ+E!gUB4Wm_%y#jV{bskmNa`
zHpR#p<&6F1W<}*@tu+)p7Me&{e?9`RpME|eZLdab4N>AS;&_~=-#K@`3+0uzt+D`D
zs@l*k-K8)rg_n8*cyrB;zS`Hu*1S@BPzN~Qs3vzn#boKpYC3s{FYYjnJ9&YjC?W^r
z^ohU<Clu5~>*R!LhGPWye>@CO49sc@6a@Pr4R*b8<9s!CQ^)xL|218`Lc=&++Gizr
zc@H8Bz^I8*pYpSPUiJLK!ooJGldEYe&ED~m=~5%>)Z)(0bAj>V&_KB;qqm>vkO<eG
z&C->_JOImnGGDAVKL~~M89lw!ia=n7zTVky-}-u=p)>~I;x?&K%%D%qda9>VV`6yv
z%5>UH(Mr@7OCRXuR2bKR)f|qVPYCulP=cc<2@v(SA|xF^F(M|_U(0KWM|cu7cK(>O
zI9Qp9(kcs^yL&_Z&3F!edT0D%Xt46Qc~nF&rmAqBzyDe|aGU4j_50)X>xP`7_A?}9
zIsr5VcL`yImSeEZxIzqT8!6NW{gXi4hLL7<qTzN7%L2(|ymt1&;5otdU5Hlip^jr5
z&+Aq5YdzY8k}+I0(T^4uOr!Sw!l-xmVl8a`_gZM}zp~Cr>kp5G4zM|XTp)8jp}BXW
z&hY=WomFlv{X=1xleTl%ABNdm7T4vbJh)I2hT@&s6Xai`K(<h{O`iaq?U>rI8-Nrg
zc?7_sX?--=yp3{4e$BM)O$%JuGwdQTD)5ZBzrU>7_8!;$Vto7p2)R1}UdAt1C{n)I
z(bg+uX}qNXm`=4hJ`&Rsf3?a{Qz`s%x2C5%pCvjBKnF<p+8Hyba)S~t#<dUtb;dg5
z1;ztP<M==f#t*X!lR@pbh2MflHV^_%5!MZ-N472!h4Ggg>)3p3@R5K<%(4eB)j+N^
z3E0!bt`U6KAt{A@`_5NaJ(XL63(RwGT;Wzx@dnhzPl=Apeyo}e9IfR45S4D{+29g=
zN^Uk`@2s_d<_6)U<xYC%2gOcqu2#n~ATS+Kxg|~#hbqptGRz_wRoI|B#u_TpKZ!IM
zBczcSK2xVIypqygB(2ZT^*jPXbQMgIJ`Q6xu)Pa(gM>g&R$;d9-IMEHaY@`}UljVX
z5T#+mN&`mb=H;83?%MxdpU!NhG@bck6RD5#5KEq3$)%y~DJO0`3~Jt^?}q**XBWl@
zV$--fr~u+WkeF3Blz7(IckRKog62;gT4fChM*YKUY1A5?>tg0loY28cM*3&O*;6by
zx#$ixnvf#6dU~n=2lsP4AN>4z7a;ZHP!(6gs2~YreAgNB6nV{25}gBuR5B9H_}n(3
zMvmR{)4#u*+B@+fK5=)(yQ)vI2oDB;&@yJw^b4S_LSRC#pmO5FZK)z5nKd&{mhml|
zLPAt+0rI)3w_1a6T_`y-G6!xdZVXh3P(5RN&lPdOx3UR%yR(jupq9ym<TKvDhO>o@
znV-~A%B<$FZxJ0~Wo4~J)3~qY#*G^ndnq^iig~wx@AZ0-y~pi5#mCKPrvv-A)1~nY
zqENKS&SAPWp+`O4F21YcBf-Ndr#6G?kGr=}J<D>P`0;AJcWQp^%kzlzV%I?A#VZiY
z^T>Fe^mE!yPqucGsE_uSrhGHwJsFr1(lPPDMAKVO)3~rCkgh8NrJ-InziQk*9Er!k
zE|xF3kD^%3cpL{YYs|9G>v(GA6fE5t7&mfiu{@W@SN%Gj_=0u`*qD^q_4aPm(a}k-
zU1?+bN2e)~gr#T4b--^{z?nylQcFyTw5iR9<FgrB)`!T<{F0qrTY^8-Ijl~^*FE}U
z9*))hP%3#3oM;5Aaif$}!|93%5EGiM)$qz}P1jyTg1&vYq=El^m^bEs`Yh)_eld1F
zcbH~BCPrDWKSoVtuM}7jY9qm05}|n$Px(^AhZ-LhnV*M!79YMq9gX^xi*<6keY$V+
z@!X7f7pd8`-q#Owh`7SfGi=2TaQ0yKgsj-vNx3R(kf|#^E!;{4W8P1VzrOgqe>TAr
z=<8w=fg`Mb)5eESmndEyM)BB1kSVpD<I+G_pG03S7}ZJYannXEJIyd)Pu?A(fYalQ
zfCeYj^2DyK_hRqGd;0s39J^+KLo!L(H}@BdM3pa6uN(@SYO$l~x6wZi^Tjt?!m>gg
znE5vTCe(AiDHlbYdKP~JSMYYWG5$%r;>Qw@=W-dGw^s9xt1|?2Tu?_j9-P~jJ`*Sr
z7TAzJ|Ma6}XyRcMxD1jz%Q-n2!xUIgDX1wdJcSnTrp|vG|G-7eZdhTyCC@?f{KuU$
zsCidO*~L-29;I9uMlUB!DmwFx&!kITCyM)2KS+RTk$?VTje^B8MTVERU~qRn8%A`Q
z(bp7dn#HRJ|2@BVGTdas7E8hBbcw-pM}(bu2_G7|Z2^}>@6-~8A*9p9!2Ju(V!6NQ
zaBs_%r^Z=?sz}-3P)CWqk}_shQ6(&YVcKYopHUX`aKSP;Xd!;p1u#j;cVgf41D)i~
zsfrjX_0`;z6$vv@2}yX}KkbTs?EMwe?U4IXY~>3SunEC0^+I;m0A(bSJq9ax@WUzX
z32>)QOScroTVa<^YMd@HN!JzqQzGW+RG!C#7kkliZ%{3C#&zX#+zqeU)^GWDemdrr
zImAUJ2pwqrh}>KFB`n(AB4oadwoOND{g%^oN}FptQbzNZH_zphvmeWL?50TSmq=R!
zBx3edi#zveI4Mm<jJ~^fvRCuo_UdJu6lHPBS0OX0K(1Fm!*e*mvZtd5vE2CGj17jr
zpj>CrbG7_V)vG$4POQs=eVJ%oFvcaOmhnL?Q@Jb#NZM4|446ZQyu2ow2Bl6$XyN%b
zK<PZF4Nb>bNhfqlgU;AA+6LFn40xjo4FwdTP(W2SoD_>9CSzh_N5E1t#NKY~)ZBT&
z5IgJh&pAS6=!%xbATiB(DjXYla?SnpB`6+M(Y+kCFx}OJ#D<*%31xN*?>d4RK6UR#
z|H2Lf_41xe;<3L3lFpYj{mxo{V9nO;6Ls<D^Au+j&^xCg3NTlE%LWn04KTWiiH>`6
zBWfC-K7&B>QgRHek+Bzz)Znwx&AmdWQ^YcQMNg%!Z9oYTzI^l_;8F_N2I|Aih+COa
zYvitZBLv;26l^>DF<a2YDBjiop%CZncj7$GpeM`Co0l^(&c%Eyj;*c2h@#Q_Q$RdW
zVuJvI1Pq6&KgYu{WmCKhqR|yhf^CmoW%@K%MLo4v;)IFT4onI;=K_uVWuMGoWKN`J
zdJ5)O4SUki`c7)_PRu=mLWVt3Q_8TCFA=W^2Y`-Um}cy~w~c7za}jGvGfc4m-r7il
z7t01R(3&rJ)DdyKEFc9Qcz%5)d3i@rp&M%m&4AlL#WYcN5K4Q{WDLqZd*;lhb?e@A
zRlYpE+cVHPkr4?C=MVYX0_l)108_4Yc(WL4GW_D8j`@=!78o;@QWz5h-q`j`Y|tL3
zDLeNR@B}sL%8tH>M0U7c-S!1;393IHT?345)wtYaLT%kuwV!o=<Z-GyRvmj+CIX{i
zXk2VbUw9gjD(2us8^9LgNHoJq`oJ6Xh?3#%;|K~5FcBp;(!S^H?j$Y=yOlMit7=M*
zzF<Jl%M(!Nn?-D)lWc<H<EC~~RQUEQ88zr{6H~4`odl*N<o3@NC?53YD@BXKBsp(d
zm8c9rKF+H48F-Wd8z{c@{3EmZ6RB3&{&j>!?^ws9`G>;d;$&4c#^wp81@QCDpvN=`
zAgg6C2}?is^d4<IQbJXNb43j+3qK8DmG#HW5(hAThX^-!&ZlUJupx&{rrW)Ud-&rs
z&J73d=FQk(4MeCt#!Ie!TG5503(-&skJEw^ZYY{)7h0!ky?$e~XB;Ws%!gU9+2rS<
zQfT%o$ck7v!1*htCx)mOhyDa<kv;%ol48?ul5!vuPzkfj!=L_yk<Rw%z<fiD08$4Y
z?8I;!s()b43-q>12=h=^hdTdMZX#Q!S+&eRZJfST60p<mJ24bEXJi*Pg2oBqrKGlc
z51k#rT{JgB)J+P2!+vlDYnyt50Ti1Q2Q@s7sRyIt7(@J2jeS?$HdK$JsB8vUl23t!
zmlhI-ZD-ysb9MmmUWuqtAKC%A6RSKfi2lS+>nzzc9K$nFm-<G?>7ed;mG7z6j9fm-
z*Cq(U%Mm=Gd*0sdacS6(L?r+<Me&g}^ue@M;bD0aOGg<jI<(zy12rYCox_*c3k=Xe
za-11a@$4?TS$O@tob|9k%u_V?pB{M#gy^~oH7$4xTR@}w3M!&q(YS|^Da)d{KT%8V
zFASn#8gv&P(g)Nn)T5e#<?;>$UHU~BJKn&}ZGf_yxMeh^hNkTO${Hl)Wx`HHO*?jb
zUy;aj8NIbM9mo<gq#J(w6!?ttT>g0TlNtV3k@fdYz!ACJ_XJuf9M00j<B`;ZwkvZ7
zPN!-EUL1&@(lAb2X3cin4Ud}@sHTXN9-_)#zzG(fo@j-DU8HA+HPPXYmTDV^6C{&Z
zF|-ww3qtVJUQ0{M3`fw5a)<qB$Xh2E%ihlD4;b21fN$kxhFkE-4FYEr8}WQ+6;cy_
zj-H;XNf``4XPE(*jSn9KXds3Gt~>oA<NZ!^Ju&?mEW5#)+>7dx-NOkW9A9A8aj=$h
z&>eNd!ev9d?z~+goc!1)Wx&l;v$Ai-sIc{<^l|hu61$TVN^BG1Fj72N4^=hQ484@C
zD-bM<Qw1dIu#HS>l@aj{l|biKSgIejMNkKDgqki`a&^e0Po_bX>}1(2mj8o*D&&R*
zftjgSPUDQ(Wmu?g1-_zl$>0?uj9d;L2%H@4HLTqW+MAjc2xSnb`AwmxHp3xRWwPfG
z1V&#@vr@71UZZdXs!`&@5$GC1871ws?%A`d0raKdr~y!F)Du`g*W8)z<ruV?o$d{v
z%V3eqDEv5E*mV7q|6-2C3`&4?;;DCv$~w7sjPd>y>Ozh@zM@j6kCFT?ixk{zTCFlv
z9o#Si>myU}7Xx6z@81jn_p4vIv?e7@!ZtnCf-yRdx^nHNMU6wg&Qme~yoiaygPIxL
zMj>yR4lvDsOu3otn`tFGYrSOR;dJjWt)@f#t1a%H9v3{AJgK(ZCQ<mzI~L#>Ta8Ns
zsnw$Tl~`*L06mFEsVb*XNxk;FljqX6mgn};^vizdGQ5@!nNA1H%_W>KbWwiTa539I
z<W?AXzcgqGKz*eA*kOIV9O~I(ERP1jyRnOpqrbaac_%<cWv9uqK=~dw9DTs8EI0!}
z`J@sH>yySwdADvoMnEX&>p>*-02KnXD~>>5&S)$ZX+-Fe$GiGkKyEPrCJF}oXm{^E
zaE|1VW*<g}vJyHhp$NZna38vwb$mZfpVD8k0n0$>vnc6B8VCHpCq5&I><gKBN40F*
zmP2`LPoTr-eGg8VNR0JT-x*Dd(yS#i;gBqm1DAB{+(?XI&Jn~z^m(W2b5V<7=<Nm)
ziXiO1V|V}+C>9@whtGMY+k1Yn&lqBwy>EE?zPv8?>sIb%YrQHpWFPz~duQKUW})PY
z$Sm!XwO2RW8+0l9!E~xNl<y>;N33zVYwWV<<7OJp+*j5p9oXlxS77hD+mcgqOYRku
z!ndT4ZTpy=|IqX&n~7QJt(j4U9b=LuQ)6|j$PXh8B&va%TGQ)DzOfo$F9|!7tt+WT
zj+NXob{Z0MTH4muR^%N8r8vm(1^BEI0F>^=xx3CAMRrVuoOMDK)LT}QfSqzqcQ_eY
zfsTK?Q#j?WQ4yO((Dq)G-9i{wSCGjdT-$>*s16=KQ-Xf7>NJ~>Zr-i!pI>GzJaTkQ
z2#<+9U>ifR3QaB+)E*2yI<l-f0m2|jBVy9M;R1+^&E$9mAweFJ!4n+i<>z+~oNnbQ
zt*)+a7rGfSR_+uWJXh?-A(_5w?BisAg4LL2XzE1QVW-;9eb?B*++`&n=h#%{1lf`w
z?Q#=3*|$Emk&k+xuJh;Er`jhrM)z&+XxapbHKsn_n&$K{U{*qE#^&sBQ&(4}T9D^v
z0f>c~l^qhxhOXK|jRRZlHtjeN=o=s{B;}xecG#t>YpGh$;q_|6E>0VGetRi7s2j>|
z)!r8bBu$=7%~^RyNl_GZGQ)==Sjqz6Ex&imB}I^9<O^wY!dpNj@W69LvJc#=ljuEN
z{1yrjJMJ3QLq6m~<K^|79mJ|MRVgA_7k;gbw4x@BjP@f?HKx>6Vtq&5lmY)q|9**<
z!w*FxvL_=`x<lP^JLR<@{PUdg7e~GOh1nP_UvCAyWy|q2h=)FpOlo!&?UxNGDZjXR
zq(ef>4;vvOC~B+iaV4c^FuU|C2Tt?8+sw5sUnxXswB23&n50E}6!i&0AhgCjW0cEo
zqi%r8?qt}8hc;ryMn*<veom9YPVZFHYn0^SsQ7$VQm+JZqRewnM>RD3OZ`JL!g<F6
zCi_QAo%F1$LbZEd4>dRiEBg1pPCdYR??Bh{59uk|f)yJ5Cq}TMpLKV4@33-%ZcsUy
zH4gO0oUju;98~KCPs5J%x?#a|owHHvM?6CAj+vIFr!DdEYb;HhSTq!LrZKg5hijM1
zJo-vz{qC;JI=Q)sx#ATwV_KT~`=7#c&|diXp+kqdG0$&2bm5v^^UmlI#Q^!|hd+lp
z+?I}Z7<QR_YgEcJ+lJ4m_E%W9Rvee^m{YguURstedEeE#PCwYcnFP?lE#~gE#<pSj
ztg+kF|AyIaR(Gfu&5DvXymcy6?nq&7hh;}B4!<d(({KK??)NctcXuB^Ca`jvc|D@L
ziMmCR$kC%mOWPBI{jF7@3G^~s*GA1;#ChF11#lHe7EBxU5Np*jQVNIa5>lk?-90^9
z8wGtxS$~C(AcS&FMMy(Q2<ULQ8(_*Pcp8s;o_KlckIy*EduweVUp|0BI~Tjyuhc(p
zC<L;iAQh*OFe>?__0}eWA3~P(+2ooAChqRfP+DUTU0XIMHn>4;D92Af@<e+fnE5)$
zA9^QkE+jTs-vor0jb-5z&%>gNfr(?1P&6y@_JJ{C6ZE}SK6=s@3>`f*yjNlk>aK=o
z0&BmDvhoVjuOuBZ)`pPo0@o01WIe49ll?zQUEp5WaJ471&ihZzPzf@&<kQNKRFE1e
zhls9GRR$h0XcBIu-({T?duE57==j!jozX6KwO|{gNS3C|3JK9Fe|xzh?({xs(=?ld
za<+2`R+Igo%bXGw<{JEFQ+qlWIw;9jb-zX<ZYlClO}dQI_wSE%nyWtZc@jJ95>du;
zHc&d+YVvLSu#1MXboBPFr4M_mGg~dyMn*=+HtPb&h&q^hU>K_+A$@{-1aJ9w>{3_o
zU_?;DW?H*TkQN=<C-p#o%)s$^3;NHm1W^IY<+;wxz6x{u?hr-hLTA4IaSFs_mP1)!
z-!!DpZ5Z0ekdh@Q@@*m_XL^O60hf0t4LPkeN2t)`2F?x#lXw|vVA+j+W{r=JC(GUK
z%{RD6tdIU2zFLBfVTg(L6DMi&Vc>Ebc<x2#3=CY6uH_=wXyYs}Lm6h1(ym_g_{I&i
zmSjk?mvP}vb2B8??qJ423ba7}T>!G?zz>5H@4)5y_3Hv<^$z4)b`vMdR{F{8-kI=P
zW;_>?^!2NHPixV~X~Ksr{9;7gv1_!^p<}E?oDHZG18(4W&OxE@G$dpbG=`SqPNJbX
z{1*HI9TbkX6DK7Mv3zO!T!4peOI7Bql_b)Mfp@%7Z}EL3?1K&T8{h{|ghhT*B1BPm
z(-dOOtWJ;5gGCV$DFH37m=PV_JqaBMXMPrEfqv2%nB5OmHZV+1Gt@g#{|j<VX{f2O
zfima@*e@4l6iL3bJGQLh#C*c9Bq|)+hAjGK`e<z!({d{<xz%Q}l4#JAEZ=zG+(h67
zz|bww7IOhcz6OA0ODFVjU`{R&jklSD?wB_2DjIBR>+uWO<@Vi~YB<CEzMWx(bcRX5
z$4^5WD~!C~a$ttdMjoRKH*(ZW)sQp0h<8Tvi!VY#XpriyO6Ft*ov_UiHZvM)x*)|i
zeXTLt$}YFCi@l83vY+LXLy~+KyNzz6ZjeL?TYtj`*2YZZ`?IOo9EDw(Wo1r1Pc!(m
z(vi3N{jLQDObaG-C!0~dgr0y55a}pGGlr8K6fu3Z$I8lzL?4m0uK@eRlLUpyF`;y2
z@1p&JdlEY7U^McV7P6RpJ4_K(kT+q>AXV5wdy)CHZiJs<Pid-%h{!d8=z{Lfr-Gj>
z+K=Q;f7X*R9B2wNg{{{$0e8u%0Z*s;W_4e*LEwpaQ&=y?5VoHDv(or_(~6WhWq&Z3
z^zEB4khVNU_QQJ#uV31a=61e@)vRK6sIOfDX_OmBt@teje+YShro9NevZ1}7Yn>uA
z%?utN?7cl*#KTtj)T315_JJ<;E|wu~d)KZV9}A#L@nJLr*_*}VgU^;va>u-u>}Ht}
z3oJW$=CIJHB5>F-wcf{(SOB|xs`Mw{Hn;Ul)STk5a+CHd=~@~aoI2uC(;0PO-leo1
zy>Vr#7x{Wg!5bj-Mz|)vgGKKLqmy`m_1G`-UtHHdRjgZrI=>ma)HGq5k|&HyEr;!v
z`Tive2!kwLHB1|SNay;^PHsm|T0V2tcTD+D=GranLO#Dg_}zPr&ZQH6(n1o$F8mOp
z%ZsH`2hHz?{7qOU^*}{nGq`IU2Tlw2+PQ>jXxE-bYQ=$1M;80#M`9sHX1B?Vzg{w^
z{xj3WqXlw;$Nh$ln3i2>{4Je}{dA+QXAhiWN}ZNI_IbyhzKp5inzYwX4%(u~e&EV+
zgeln1`~aOYMc!IC%A$cw!<aEc3R|M6cm$ic1!7TfQM9ybmmeiDMl%5UVrtSD$=Qx6
zCbB8RFFh=pt~5{@``wX!cDcG`5+8nRKV+T6%(RbtM(nTqZ0KBC%5xUs*b0h~^N-*V
znVXE7t2LScXvt8`{ciKRCO}ScSx#e6R)xbRC(^h1p^Uo>;GF)rAR0h1yEo)&v*G03
z0?V_V(w(pG91=W!k#6^PBg_sxEdR8<!yY-0>Gu9Pcg@Wb!-`CnVw1ANE;Wx28)?`=
zntxv?tbXJ+-`i}-x-3si%ohAg+t1jcuvKKdNsuU4;AB*LT)^~*Y4~IQbDz!9cA`uR
z5B2-1;|0F?>jgfaQ@|ZtcRpw8^cZ%@4?0}P?4aEPT|SsKLSjGa{e~mDg{(}MJjxEj
z(qa?iMKCd0`f0+F#=CQAe*6c9c>6!2=eL%;QTo~X_578xo9|6FW9p)5myfKMkTVmL
zaInofwn8qZeLp%4<^MSJ6VsZlj5YoJrM}!I8!YbcJ5xvy@Sl(WY|WrIAWN1l|N8j-
z4~Aaa-=Ep{!~X^5lp6&tze}zxFJFd%G58)p&RZV`x<cwKM!d@*m#{1k<A$o8zq@{j
zavTY}fcZRzjf$lza>3dW1YUi0pm?^SKf-JXLuNtmH!Qm={TtJxGs%5RQyqMX{Y@Ug
z3{OSzs`Bz@WAotv6^>d7R92NvVf0@*l%q5o+JA|KlM#o*I;Nec@BTl17cc%Fzp0J;
zpDTv!T3O19${-pOsegGq?&(1$=Xp>d`B}{%*r>VaGb};{3()h@MnOSAOIF)xpsE+p
z-DMcdXcQt8ktH%BA%vFA@HAGEdtypVZEY=iQ9mx|yh~mk=!|QkbdD!2aWhbaMAc?r
z&0F#oWC)6eNA}@xK-JXHZoOwN-vNKwE3=W7-Fx~uD}0u)Y*|A_TO9_S%n0^{5EXg=
zs9WF67C`EVC!)kb%)xB7{JB9BU29eK3`F>pVkRcl<EowiV?{F!{@=Ky|Bn~Te?Q>A
z*!bW1KmqBJES(SZr=6n?V2hIh?U}e<%CnmXDg8zeuWFyg2$tE$%L@|1&`y6ebt1)5
zpY)B3dNE^pb%hUMjsY+ua`O;Dm4SLjpR65sn$^VVAA<`QFRSrXrJ8Ac5{`*BuLIA8
zSG7F4%It~}6t|vvOs-n{H<u|eE@9&;(hvs$Sp?*_yF)sMU~-Fe@a|9xu1}*3yqXs9
z$O|XT;Bu&qvE5R%&HTuhW1}AkeQz-ZstxLbxJ>0*=heTXTvsploX2Fz|NNQ&zW?Yc
z|ApVU08rS@fHW>i_-lG#EbVjJYN$_kJ{Yt*v<31We?Yfw-z5a*bLY<G6+~amLAt*F
zt87_AIdl=s9o_In(53d$FJQ9NAk#VW0_e!i&lO=tm=|e%IwuII6O_WX>FrEAtGK>>
zJErKLfFCR_l|%aLD3ryeAa3+NzM{PREmHc2Wjad8Zh<%Wl<Cl2Gs;&C8sHl+N2U!4
z2Kt8gBBVKvXDn8Mj@ouqs1|mMZLZ@c)|opQzkmP!SI`xTyM*w9BcEcteH9xU{)a#_
zrk&>BzR|m2$@hS^*&&x%QZtSpHQ<ZJfI^^~1fGYVI|TnM4<%2#Jq5}H3jnX!f$uIw
z9{QR;r9@-EM%KZ>K_SQF+_@XT`25`P|1a}_0m|E-dHinOi8eGe)FGQfR>%r=rjW-2
z*mO1uIRQ|;AMU9V<k6YOq!IYlam`YPFd=Z&{oMQuBzEC|%jQw3`Uz8~^q3hx+R_+I
zsdxEF(k-ZCc)uR?j3v`rytPDfDV{-bw=baw;Hd~=b;RtzD?gHkQ<`J<^~Emq*{~LA
z444UlXcX=NQOsq|7z8H|p@0gZu+Y$EpeCMZO6XasBhR@Y%uw93v;a_d;d6YI=1KP0
z=;))juz;%n0Ht-CsOVX%LPAo2WQ=_Md7c2##75Cv4)>^e|9;KB`Ak3ae0}LZZ^6BV
zdx)W23irS(vRqP=q3jF85g@r|tmLtyl60qbWa+JD+DUGA%TpL8ifa-9<8vTs%X@c%
z?g0?XUi}bz(yuLoo(~@+G`;Rc)3zfA?HFMw^B(a7_~%a+E4nSbC0JVnxiYgtuq25+
z!pd|c{9+yBk^O!O(@+ByQH~<KI;3SncXbX#(byQ_Vom$r4Y+cqw}8vF;Yf3mmHc|H
zC^237`l{AvWBI3)Z^v>O8!j?++{)YHXZd0ONmks}Mn}9IrkIfR_uTa~8Ci-(BjN}<
zV03%ScK+A&e5f;#9g0=i*Li6k^9?jvtqt7hML>cykdGX!m&KwfP8fPK=8${7un>wv
zy7pKJF(+^rZj$$Ly-ZBMYZG`0e8<MX4b8B#A#hK8B+>J^p>c_L>#wwfKQ5B!PqSDJ
zNfiH8I+(S=UF?)4D-~S)6U)&>?!aD(6*u5-Hlj9t_3%{f`ZDXEZjJvTC9qhNCkY9c
zw90LO`_}{*x0B~dmxh+44tOTLjbksfm~WC4(}LspmWEQQlc_!*b40Wxf0?yF1^uql
z-D*RbJm*DJlqV^~Hca(lMMa~byIN^J(=x8Zj3VXI4>fpq)YoCJtzu_`_xWgNr^s9E
z`5#c&T{;UyLmNG2_S&&$cNE^4i@+h3MC<XTKfCf)t~z+|;M3@6p#>HAOv7<d?m$G2
zvzHMw$KO@UN_(zh1Whmbccw?ae_QC{D}#FI036i<C<Kzwg+R34a)9WWpTO#W&wdDF
z!ekfVmKgQo#SI|Ra>zv)Upo>GO55+`(Fpp^*|d^x;g&h<dnQ+Rp0AJi4Pm6<SyV<k
z)JH=YuJgudqlhD70^S4D&M>UXr;i^$)|eW1o5wV62tWWIw|4eU?4oEu#s%Wx2#lSP
zt5&Ug0$<veUUVKGW*N}hc_bBR27zF*O*2373)aNf+IJ1ZoD(F0qH}dR9^%~SK$w#g
z=X-VpSKVpddITz8xLJ9S-=%8<;erUPti}>1^+^ytI!~W>b^mLj6YBsMxb~@WorSX;
zeru+8m&l&oyLS)wHAv*rW6qBJ8d$g1R?^lz_}4`|7M@^=hB=4_=2;o`rj{ze8Vfr0
z6Ej+|v2{1!>2XF2hE5L-7<>Nx3(LvGGb2N~SCmc@m|%5;tfM6z3e<ckKZ(F$%ri+c
z_DCwm0<+Qn@V9=5=?Pi9mqGZ&9+Cfc3D_)8uo`+r`xe7tR(H2r)r!BOspN4wuofVi
z9!PDZ%QpPmR35i<jk+4Su)-^CP9by|F?%H(U2G$<<vLCN6-}eEM?9rQPq=Z)NSDy+
z&;E?*HFpfFCMHGVWr15gL)B6MEr`ospXi<r@)m;oRrvcemTW+8ICVfNE%H`TQ6V2>
zhS&Xj_6T6%1h}3pfY|htjEoGX<@DqGAgX%<h;n$%AYqpp8XDhiKHzbcdB;ZZ>rbG_
ze6;)BU*`64Rb-+=u4k644Voxo0lqJX`6tt(hxmOD4Pi5~CXA$FcW~umgQ|C8Q)C_Z
z>zip2(#KBeuYuz0-|%mjHUCq#^gp9ZI7t0xc>4e08Qfi&@@*Bl9>dh!YP4p|TD|0Q
zkjkz`!%q)NZrc&vpMV5Y3YZ2UGX>jR&<V*<Z2;=ux#;I9cYeS|Kwi4>G|o20N9`Ge
zFEgB+xrF#59G|mFm3Ekv1)+lSM6=Gv&W_R3&}r~QmynlCgU0d0hK>S~!3X3$^vS1`
z@$qOo<Y~C5s|;mgF@|qW?V95|a5^yj>L{&%M}oC?*s~M$mSu;K#H%_~R8_4^kyZUv
zPuF#%@UL?jky4b>lRf~dEJE6{l{x4)>M?RCHip|ZQ|-EMTWO~%h&rxh6k}j$xJUy*
z1|OCO{&iVBwWgUfbO7R`Q}mx-u>$(Y5;U3Rc0~zFe(o3$e~UKruLu!FII<Ry^Cg8v
zZL|kHb}y0ImgB+BM&_fiJC;=XdL2byGhzJX_fN=_k5UlPM0>oNX29JZAYg7&OT_Fe
z*~q1#;ZLlGPTUKW?Rf=~K3-wL=&!&#oKd)C<$r!wAt0s}wWdkshjkFEmnoJUDy#;5
zU~h{_W12WkEGG>WOSF)OJm|WKsR`8YA}M<oSR1WA|LQv`YF>ne-5eYo97iXCDjLca
z$P!Bu3G4v$nG&a1*oCy;-WYEe%Vb4I&~h=W@On^SRuWev90r7B6`2HOrWSN9y32*&
zB-ZlcsfR0<$M$v1V9&6?1gyNGf-IAD&{Q0NOvfe&K&*hIjwRZbkFqbR9mL_O4V?#x
zG{nf&lt|Gx(3yy3N+>+Ow%e#aNeBaRtkl;;A6$S`U&qm+mKcd`JK*3jvvQ~nss$c6
zN#qc*X@xOt`M?n2I~lVcmEAIA&rvO);uN0buL<BPR&y+7$dET2{8u7;&J*aD0Chph
z!<nhGlG--gF-2Qu5{phQxvgs>vfGAq-P+CUf)4E#h3%jIghyS$g-p7tNQ|yh8$GUr
zYY?OiQMyXyDhx_?nHBISQW;Z^MQ?`UH<-O2gCg!^tw^4MZieIan4fC_9l+Ypx>o<F
zn%ctwM|w!pUz}y!z-W@`EorQYV#W!R?KD&6ls%R)mQH?UYW3}Y7R4@b+kJd2cOnKy
zRyIwHYh;2Woo$mj)j-)lGBn+7?Qk>woft0DopLhoHu@dw6u<2`w8E>&!pCz6Ys<0`
zqqJmGw+Ybg`?X(7S;mr!gzLyQBBxkC1TiJ2X{Mp6ZNt9fXsS>@LlQm_#IcE7b6oZ<
z<n>l%j%N!tV}Qj_<=0JR(V$+}p54d2j)IOEI+fXnLEW-uZ$WP9s-kJTA!DkcKOuFN
z+NVg{S5lZt9W{)?qS%hMZ411RJdk*_4xcavZ0WK{AZZP`U6C_u&Sk$FR1so>EF#Y&
z4>bFAp9fql^ztxoqzm7-?(YC*eI8pi7iIeg6AS9DW8lX^^PsIQ!&XTEzLXpJd+X)u
zY2U8Yys#EpA9wH7G6Qp<{;Upc7xk+gn<gt0slCf;Y)`IDMB+eOl6{j<KWzwdrdo}F
zFkjpeWg7`TkxjB3euZ-`HE5fi=Ej^_&~W{%BsJm7Rr4{8g4wP0U(-?i@$)GxTVoRo
z#F6&Lw{HZ_eAXkA+Hm(o`0nJsxwhyj?zK$EAKr%S%;K<plQFP(NZcwS;V`>v?L)81
z47^!6ME*ukGCUzKz8GO;B)Ro4ZyJb?a)`*%X|?mvy+iN&FPCThfr;tmRZM9M_D(8o
zUQWH^;vOvPrOhy5&YMhHz;1a&-L~!CNzOzmGQy=oHLM5nh@9j`$-z4ri_|;^Eo@dj
ze=B8jX(V<7xF*N$`uND<qCR6S(4F{-nQC4H%pKtoi^WL^E(fpY_I6*LW(1nUsYOVn
z7_E4J>)K5ShRYYMKELzZ5m29jT+eZ$K+C#Ye?NOg<?0FH4r94%lBRU^mJa`Vh^W;L
zrq1GLx#P0h<E}&DH3_1lb@hEiw<rt|Ur)%qH7^Nh5rriEF$tKt=)i^GWQY}*01SeT
zO2*(l3czG`MW>nQ*h6%d!+L~mH<x02wGdxL2UJxmXyAQUiUt~;K(uwuuqkt)03kfd
zN?yXib#7=_*KC*_HJuYBGc<~><gqGhRQ)Sr$xTd6WOL4balU{B+5$F{8vu{D2$8Dn
zDCO_7vP-aGvcK7tH*#NU5jj1B&g)L?1$5K3artByxV<Mlwi#jGz$z2dvf2B2{L4co
z&4uLF`6a3cGF+Zgc>dW5FUu4po@)g5&~^nezwY9$N*F!N?a9*yEioD#9R_cLxCuI7
z66puNDns=;4kyxFg{ow5C&@Kg9m{I~KwCB%I;&vBd;=A9YSGKWKNhf;HxHs|Eessy
zo^=y>*)dCgc{T3rH12!>Spoz6z|qM^`~b*8dV)`!3kYhZ*2FuNW^jz?s0wB`1t4IC
zaji3}GhD|lZs|dPH6oG|^SR8B?wlJnJ*ucUm`&b7`njB2PQ7BTn;YpVxSWitGGTJc
z?Z!9^087aEQdL6+G((~-h4l!gbUulT54v+xd@%W<U*e5=|J2Bfj5yJm@wJ-4QmXy~
zgVrA+sT*Ew=nDj?b|iD6$h4V=byhQQq-lY1QeU}{uvzJzmDxy~RMl+|44yUEAPwRd
z2n5VmI|CqdC0SzvZHO<LB;&O0@T_ef%bas~SZDnfL|;`6vNy!pTaa%yR|~YUERxjg
z>Q=Yl5g1rCV8^f#m|<VwdZWJ3tN5>Y7Bl#FA9p7Wfq2LD09W%LTE(s4=WFetc6iw3
z#`r{u69;sI%wSj^mqe{1H#CCieBYte0s>46UYoJi4Dq(q5{z6>zpbcB9R1-X5n%K^
z?Po@+a-mIl&-QVzC16+v0|iq7<k`vs;V=}7a_BKp$!{2r3@1%8pReH}u*XA`w-|d2
z;6WwYy5X%RhDE9ud3kR)jo^Hm-a)1wm-_w!^_ZB(oksg4cDUNj&5iT=d;j@9b?3(K
zo1Y&jUF2%Hr|ZDc<Yn`d*L*zu!*5?)FJ07ky?8Wo&AWR~e!4JU_SwE<#UGw8+GqOX
z?`w8$3`xzD>07HM3O}y~quaL9#WfbfLm_wWJ8Zou99Y7H>4BtATcji!RgZ-zZgB2G
z`evn?3ze7<Fylfoy=h%GZ{9qToA%^6W?cR951cU9yOhI@L+6Ez;~IwHOkCTbm27^-
zxmN&l+#?Dp8#iv;2mA9U$ntuIVR;H!yr_z{eHeN=tqPv{nRwOMdaQksdV!WO!&Z*s
zhZX^Te;KW{8q5E?NnE|%el$F57-|4#y4)3t|Gaih+)vhgtWP65CnrUz3KJfSd5-N@
zL`GKM*CYkzv-yJ#xVm1~A8qtqzG6ktmoMkDgmJ|?;74M}Zg~{;N5H%y5jyPRHM=r%
zTMUaMSDFsAst#m}m@nxl5;-~Hfr=a-YfKjo+W<!U`$N0{$_;&6q}Ht2iEuP>atfSl
zYhTByQDdJTb8j(W;SX&o!mu%XP$ucoctF=PRwvVTC`fH2SsQdI>HD|tX!y+(r#%O&
z-NfhHZ~NcgHYvfBNdhz;JvoJ8?(@EV#dSu0ZB~`uTsXNDhEDLUNbF8?oJq+VzH#HH
z#BPfqtZ_`n#;f!cniM<gH9~;6*_u^Lo$L3GBKH6NuZh+36wpX%-EtUwax&8ij0^Pj
z?fl}dh{dS_`#ts0f|Hz4>dDOtRTl-|h=WOk>s@QtuC+rTPOoevQRXd(h&|-yLRv4l
znE5~cxOvT*H8WeGc@s5~O*NKDdIX=~*RvdtGeGAnocs03zPg^RT?(?!>h_<RnJcj;
zA3u7uCUE!@;40>_ha!gUaULMfZ``q?c1weHbSH*xO~#6<;6Bv0B^_&3(=ZIt__{uR
z3+c3ysgynrFw8>UsxXMpU&_kb6`!Gd?P*UNMn*)Jx=7B%VYG?OGf${kt?%o}i0%Wc
z@3$I=E-FtKbU(1_t$c&AX~{TEZfWS1D_6E*d?FpXE7oOrC7my)fr|Pi2tj^t{trK}
z1obzhs$fqj^yHhs_|C>x8WJH=(4#$R5)2~!W~CRmPEP`VBN1n3D@iEYa3ls|jAo0}
zwr%BYDLCm)`6^=~;~umg+EUDSI0cI3+jkVpGok`6N_@ece=|z%1NhyGGspS?>J!IF
zFbGM?$*235MT?F>AuS$5MMVKpYLJSHeh5)>E9M^k(kItQN6~i{1jH`bvY$WU94k6~
z27G<yGPvUnBkz9!^&y2GQhmb#9*q1??*Pu9z^s_puU|8lp`0n@JM=Eus>eIaQYC8B
ze$`a#h|{X4PCWuau-9B;j2FcUh`J#_S8bIVS57-LuJ8Ak*y{gnZ?FEJdz%zE*kv5+
z^)ql7^q^O-3Oy7gHP1;+2ST)1XB1(ea~pv=niPIXP=lsUL%=9t?oo$xR#{>Xj=|j+
zIQoEMg7*8hYu7@+q$izgM@%c=udLpOp!<Ud=Ew0+#;7svBq`QrNz%YQsR4^yw7-CI
zDc$w1UAvb3_8Y1BghCBKl?)Z(W(D)EO<T4MwS0$7Vlh4%&3)(rOU8F$2ujfMaIs%-
zHrPRq#PP^MYyQGz=T)n)BdnmfpEZoZK0}md0%H-luLm#zBx5?H>Mvwnei!h;fJ0!S
zy~Fvp6&89yjC2zRfM3>tx$M`y!T5I>*76bj=6G1ZLJ4X|A=~QJ;pptpNS~&Dop$s&
zkoZ&~#gTOp+Eg^Z)(|;|;krsBKvKN{C5W>JNYKQW2L&^YMj>x+J<7J?<{o+s4@8X(
z#L}@qNtX)oIt=uSO%P&-ew7NU67nIy46ki@pHwk-fylY{JK>EQ^KcX<cb$;ykZ(#*
zjaaqytQ)4?&WJhf+<6J^nIWk16p(e5;o(|ArV3M8XaOA{(PL;|lp|}90-{L6a2@dQ
zo0T@bJyz_0|7)em#AMFQ%>b<bdIF0{<^kzW6~;f5lFolJGPcJLnglK?a_SXRROJ*}
z8v0lPnh|AnWWkh<I)DW057(W<V*2!K_5M)B=dLP^y1FHJl{7i`@l?NL%1hWVVc{Qy
zXtGFQ87>KHv6{Dc;*Lzo!`6m>zm2u;n?MKo`=iweY72xQTcHp!GMxg>T7iTiUAYNo
z(J)fePW&2}v8j+AAsQi*aqg&4l`!`+^uma&6<{UjS(jnn<_Z!{ubHhqwqzt(zlB+~
z%g##TS(Nyn{QFt__&=V7ADv<pWwC%Z3JI0+T5)r8cOyMz9)k2wDDK4$#2MY_6EWqD
z;LSi`!9gSKg3!Sc-C*>_Leb`gu#xxSZU|W^<EVuFM2B)Rlr(5!4^v`!ic^>z8Saig
z$+toZ_53O!?LVOxlCEs3(9%D|Z97(fzl~9k;c(PD^*>_x2F$oYPyH^2%R_N-5b!(5
zQDK3($aH`!m?OG^a<qAy6BOHMiQ(w)g8Z|Yy<!DCZ_A;;Cxz;Qj!rp;(O_UlFR+{*
zV9u0M6XhDH?Sx?VjhQC`!9pF=NAO*(QxCPk05loWIZm{#Oq|p%w?-ETqiM-}RaTxA
z1{D~18v(p#h|h3jUi*w4NrN9~&n|%zK&|1-&<dx3MEhR*p?~h`)!dbte|u`}>@)xL
zW<Kp7QtHC}JwsFs1A0J93c)d$qRsW|-FV|EIO?GrO!DWwO_^HcDL`5>M^n+fCH-e5
z<WFG!9w3#e5ueOyh65<Xg_dD2;|<^z520A38;i3YI!hyh1(}yWqYWlbk_oh+11=8k
zhgO{uE{7qZy?ps|l)JnTyM&}I!&MIB*BEF~Ct2u&BhXRQ!~$nX)*)DQLdSz)MhA_5
z3^s0psyq_gQl!UmhE)Pl9Ca8MQ;Ih@PMQzcZ7R?aQUcNbnzJ+S_UfVBRV2%fHDrKA
z-<26lfsW)Tpg{tqkGr>TKf$Kpm6q0|akBh7JXT*m{xO4k6|%)gyqm7K8`8oyE%s;q
zwy>vpUm3)!&NQI$OCP1xiIv~D0I%4|kP;$3ef=h1ze@iOqUmb59hEgTFfqO2x;)$W
z&)5F>lkS55v0uJ@VaxpgKK<=keEaWz&*a;a_~*a>J(GW)#kW8G_e{P$iGTk4|IahI
zcFM`-ub=tl9uAcd)Y`KQXEl6w@kLPjYhh%3+74AS)X@Ch$(rqz6g+@Ps4O!TjD#wo
zuiOhJCR=6X{1p(%bNXf&{3AJmTMAd8RdC0F9pAb$1!E;!@eQ|l=1-=6t#SXkCHm7X
z_Ww}@ef#38W&eHp+q3xg-~XP;w<qz>fB*eV9JfGK7gg~yr${Xv<N~kdu!M~*BK;j;
z>)Wb4gWMxptpt;67cFI(BrA|M$Tw90N9smOGu!u|?uOLP%C>vHJ0+Q7iq!@_qSCPJ
znd1QdlHuv`(@tw0zx?_n&UxmZX@he+L7Tbq&sJh$y1)&~BRFiR2F)Nx`G6vQ`aRdH
zDFLmFaHyD&5-p|CG=7f6Ys32W8@Fv!g0{`(V?k~t+)F7Q#t5N{9hlpzk#2{|FyaI1
zQ?f26oh%?ADHa_bP>~JZb!}k?w3W!Wj6__(JyJ&5WE0W{X;A~{D+59})MV39%nmCJ
zMaYSaCY98<SI={z0vh%HBV0}o=u11G$PZ5hE1+7-Hy+t}ZQ*i72;eE9%?bME5lBHC
z!F@=YlvEAy2)_NH`Mbg^7ntNoSQhlBa{%M1D26ofJ@~bNEfk8%G8D2kC`y7+C5J%g
zY8Iw&lGmq%UZp1W&trNe`S*Sv&x1riMA&+Vdw~LVA}1fv@nmWhA)hRC3Agq-0qMLa
z<K&oS4yd6LB(%k4Sy@?DFvx<rOuarQCBe9{+Cw)X?*{zYzw&ZD7Yt5NO}wWWeWdL*
z7oo3WkSKrT$gQGz>mmuyBe`Bs&1%zg`T^UNBHtW=`jr7HSHfBrt`rCdbA^m?7XI>!
zRdNv!2*c^|0dmPXO<JxINH%-${3$%5Y55EaH)K*n@}j6ex?j3O){#Ud;VTeVlc}5k
zJNlD`2IVij3T8%l4(oWzPG6o1R}P~$0REROrZ1Es1k-R4Z<b1J2u?$xIs$rTiVU9?
z=rLm;E(DUi0`Wfnf4zh0wStYFt=BG`+ACqxR}Phm$E1g0U8bY0O$O7TKK3AmRmM%j
zYA}jq5%3P*wKsmB7skT2%euhBo|*{yuP%Mp*lnHyfv8&l@wX0~Oib;|I{TUb+EkZ#
zr01qB7~2Rc(VzMSzN^`chYRiNytqu1>?a+o*Q{Op+N9wXdfZs2{OBL;Cwp`3Br-C}
zay}zj5p;-+3Z75%`1WEiMq<<|4E>LkixV*i&oFaHz5p6P%GV|`=lJ*INvmMcAr7KN
z!M2c{H?f%m5ZPc-#d(p{etq*2VF{>SlW`52@hl3@%yj|%89>xodg}95pFusF>HH$0
z%NLlO^25fslEpcE524^gzdD=%6NP8Oo5{xSI8f$X{)8-+MVa$j72JM#%^B09$tPjf
zw1DZ7_}>Q_6O#-u8Sijq!=MuI1zHs+07&)#{JTFo)>S5l{d%&m34g^MhwxR+8!td{
zQVvCPAYBEgGa#-ag@!Af;CFe~rlK?>ahfrE+e{dYka_g|QGjNR5zb~a*;UbU>t<g~
z!BLU$1b`p>YPA+3v-^@IxPM*3)Da`W0fJdw-<;4#DO6)eAxTYkI1gtabyfz(8d-Uh
zh9}lBZA0*w7|Ri2lU=s7R0#^90cC4Asc~dg|9Z>9w{7k+6I1Jkf1t26KG7%msN8LU
z6!E0(b9eYarK8oT8M5e$4+yLcT?cB><|>7TUiybd1Pu|Wj0aD+Wds9;7&;(xaf5_5
zEoxJM!ZVrh9gB&OQfV%cN*1D9sA?Oza7q+xl=ncjJ|tefdQ}9JYDATNpmHQo#QS~2
z0g8dJMPo==1M=*Gl#$)+L`PWxpa5cz<fcDfxU=syu7Bt}UewOMH@e?gFHGk5vGlsY
zmb*V%6>^NM_y$xZupcWx-%VHM{1}2JoB?>0ljuG#gQ*h~RY~*%ls85A!mjq7LD$6+
zK{9KYR#gz>w)at>R|t>{!M^AQU1bs+22yv*5{3?hrJW=U#*SXNa3N?0q4`~Dn4_M`
zXRjMFFQb6^*`c>P0`4#^6$;Z3c(;NA5tKwvU=sXB)*xYC#Q{=1$`3Q#WX4)^aK>C>
z{cXw1Q?I`#xlsp+?b}b##iPx?3=sNDBlIicDTV;|6`fkJnC0VW<{_pvLZUw;g(1As
zF>%l~nZoMozr!liPJykajgqNTh^*Ze)`&G^g%(xmQqRxFCl7(q_vi$ML2btX^T^~m
zf#`i4jEJbjD%g0U%7vE2;LS$69)3Ci0OP5gJNJUbJ`w58A<=^`g=lyS+8nh>W=~H~
zqTv3ej64|9F$x+A@mxBa7B1g>4s;|^7CwC+b~p_g4t1Z*=zKxAEs1LaYzWn7=7Py%
z*4Z2`<5%aFvaq}WkU=I@q;1bF@B4_356lR>3zW%x^XAPkx@3eol4K8Xr1qB?KsmCW
zTgP(E*56^%`F?<D_U_7lxxRe&fmeKJKWqd3{m=me{EEO4>qGV6enE%g<R}oF5kN8a
z&pnw6uSKsNKWJTlbELmXOF66};xHUJvRo8Q(*~}WSu-B1`rnVW^>-5S)`W&k1^lfL
z0v0Y=qJn)`0h#le4w<=p^BJ_I2zm*y?-R!x+)YO*FbdFbPNzeYfbMIh$2uDvR%Gi~
z9<3v{M^29SFQge7>!i01e+@Y-fa5@OylV%tlclk(3Dm^1ArG!kpFYt$hbUYaqS`qe
zQj^f#i77YDPpt9+>;&olM|2>e2fP`^mw0Z_mldK5CNn*pIk(z$^17Ovn+a~OKGXw^
zoT@stOvm(tOPDSc0RPzr-8e{FDU!6YtE=L^eK*hw4Z_fP2=KiJ!e?Bev-9!JC@94v
z0R6)NwG&28uN-(Z{Xe8Bhj1npYJ*gV0Ksw`0f@SL&<i+<g#@S9NPwwug3_D>MuT?#
zFM^z2He?4y_6H-l9RqajyW!QuVgFf(4*IZOnDHEunVXI!=SVEzywi=)x9S4QZ0*~D
zBjURLUZJVRADHb4r2<}@tkE%pSgUI@hm1rwUe({b#H5DZ{@G`DnYRIcZ1x6~;UP$s
z!f>c64E;ENzCL~-^U&MeTnYJ-;&F@2`ejQ8?IdsA_*q<>d!Ox)W2<BGCJx_xG|~-*
z8sHThNk$*mQr~^|9c2u7$REZ??)cLWbM>xf+#M`><{oC;Q!H%FnuUQlUJ5m|94DS5
z#nBauK+@J-`qcUK$-XHjrY)Ozzn-<X3giQ&8#d>6>4D$n*xLcz-gvSMyD|6-=`<eW
zhLoBeT78B{Qv5kgL1hbvf*VJ?S=WaRJ!m|%znuN}@l+2os4hh|V!hcVlOB!0Oa^wp
z6&iq|#vGCtwj2w<(VSw5dE{w0)0ihO**l9)v?>=Dmxe(U4vk8H$f^)w|D~tE!@@XT
zH3FDsEXds-R{i?xN5v-O8b(0oLd)!8TXv*00a(@r*C_H}?t56&^x$Z32yOB~qv1?q
zAi&1K7a2I~Y9gW_qaaW<00c03gq>jb3hn6Xh%kKidK>aijlMFho(RNkvoI~p?ZnMu
zRR~lxVbx^WVHbRG5~QX^^;v-T;mFjjCz8+}(=n!mDi@Ia=^ixU4bfpSPj>fvvr&4%
z{P`1U#y7LI-C<@joep2p4=;^?T7LHadGX@7M#Q0)>PQuPy?N0R6XEAC(ho#~QJdxA
z*yoy-(*D;zX}{TrSNg4Vx*?S;GpY+XJN)wH%hz*Lpwl0f*r7k2tYzZ*v#kiniHNs<
zrv19qyk{nF=HhQl+$Oi-VDs!AN`Uy?jx=sG2YdSkR6)ErS4JLmXF@UngjSb}8C7q|
zN)vZ*yJp5+vZc5+ZO&~gtBz6UoOJ!oHon=Em9Z!IuD3|s4hKbh*v?*jB%@>{sutY~
z;kb!Gn=vM)FD$0tder7Gahx#}@1Pd=0D@!_)M2SJu1~W)c35)Y&$4^xKzSB{)?PYP
zJPsT@Xp!n)Z-l`CJLqm`Xf7W{nobr>q)BVema)AGu~HG;8cM-@RASVlwXbo2M2Hll
z=K1v4_}~sY3&|0)w;{Fa`84V%QnidH!MX`3K}KA<bZNK*HHAiKQxoi#3~ijyh)Gv;
zI(4dDhJ6}Q7spssnxfNQh<zQ0C|Cs@I5-d@V!(4ff-@NoCeQeyDtGAS{R&S}S>!xs
zzWr#OV#nU6T5gLLa%D)ZghKJ*0M0k_WKL*tRADY+7+1y?8FaJo`|rP3f;WBEFm%S)
zf<Q1T?sL4E{RF3b^&yIMpG087tCrgI)QwCM(Q&Lfv~@CI>Q0o7=iT%H=V~2Z!R!E#
z%nGnf4=q}_kPIo)YBiOcOJ>ZE$>$jk>|DpKd)G&Mm$k5t<d&O|vh4xb-&Kcxyy(TN
zAz{e4O=HuMO7T_r+`$W8-6CJymS7h2nL|e#ulc;-kNf`@d+!-m_0@fgk~hV~M2!VS
z5V3-Ypn!m&fKgFY5T%0@r3wfH5Rh(SVxdcsDyUSc(mQCBUR0_yi69^#5ITYz?wH&1
zFZZ7NJm=o~;XLQQen=vL&ED&`)|zvUIp!Fe&}hcS)Dz=JXR4v3K%<m$Z-=ixw^*-i
z0F~oBh2}$t50f`8w9F2pxts4alY+ev4N*sM9Ck=9;zmlq5zKEYLkdnMBD&v$NOYP3
zm|w!piu&>bA}Z;Z$#anEB_x&$?L5Uh5wt}x$2B?p@Zlb0SaR5{Dv;HOY9je=lGOmn
z#*Sm9>$Dg22N=#SDPem}?n4M-fV2Q{lR=VjTPHTzlI0toqg4)m#-Edgh|W!*?{y58
zq^3k+IR??4l6Tw0y=~i(-9?jr<nuzwD*R&d(SV}ALdjRK2u|AJ5N(Y|c9m<pA@4Ct
z;NcNVoNu*3H;{VQtX+HCWD$L~<bZ%RshL(jrFv)ZnLe6u$(}q(Lgjnm(d9FA8K4!p
z$^}hO`vKyzvFGy<6Un5DTw;5qXc>{f+P{A-Aq(NWYbIb;i^*}I(S)k+qL?`dW^*7_
zaZ!T!DW!%fV<NglZM9tN?1Rl&7bc53VJKH^Yp_nYd?W9C_lmS~bqgO!&<WWgd3@dx
z8M<<`?8?-fA(Ic4J92PEtXZ*5@As{FlH9XbTf2YLxY(XuYgB-xL>$+>J?ISXBI~ME
ze#HzBNQZD7H4=p5&V^SBmW60N+2^nCxvaHY=O-L8VL1*saU-vKGbKuy$)*+mwD6^U
zD%BL^|8K%a=)gV%U5ieTEzXCaz4h|VXj@idyZIm2u4%%(NMGFuFHfp`?Qf(VC*#jA
z&=iy8sP4k}y0~(eXK2z~oTHY?lfv~r^YAqi$m#J00WLacB^%oX2Ka8_xAV|Ev&EKe
z#s$vSSLtozg1}^ku46LFIJ>rfsLW`RMWj+mlzy%E&|Adp=6`*=G`gSMxq&=%F89$#
zf4!OR4<X1B3D(RbUbIdw2CC^;sppVQy`k7qBwN+P`N+&o*d}OX!&-8^4`bmOqMivv
zf^$1F!3H!=U`7^OxG(R2SB^qL+r2-<mGXus*|_(5nz9jXjEnGoJF_85r=6OG-ruua
zXB|fJxyQfT`E1^?qoUv3vNgT*AA7b98!kfI%U0bLFh}!0|NOJ-P9k)rhQJx?RV|T*
zp=g0E{7cmN*Pc{$-LBJ{{8B%w&vW+hyUMd^{pdb71Aov!+G1k{=KM0B2M{W41~t&3
zh>=tlIp2FN*tezox26-gQv^4Qb@_BWYae?QuM6IjU4ek^TO>eF%pRg6<~rO36bbfR
zNX=3k&%%i+%#nl;3*KALvHjR1H*!9_a%!PJ7FP_re!Av*4(sXHIyEI9tafLkxpuvu
z{GxO$g?YMW-v0%_<tLvU%nL1yY-C~C*i*|Kv;sW4eC^La-$kyUh>Po*0~oYX<r|%L
zxj13XEmD6bf=41osjdxBRG~57$zha<VZ(**aN7?^L;qJ9rbD?6LeOmrh3O|Uyguvw
z2q;jyZv;$3#=&vJ*RNkULo%`xabvESJm&?b7fK3UIOSnZgscA6WLMztpkq6s(f{Z?
z7kZ{~xap!TcSCO;h5)}#{cK$bXoDj*FQfZ`FchoKz^F8q0THGY<h#9rqT~s<Uf75&
zHfX2H3G%){E};O;m;tc&i2bm1&5m>GMxe?88k@>h0LGF|7{t@3N9e?-7LL20e}%u$
z)KS9%OzY(%4zxbhLLW2h(tslQO5^Gxy;r2(2D#}BC_lu(Lo?rP54T>H<`m>{*<$U{
z!NfA!s~-a=B-64>ia-%0ig30<o!SN$a|rZ~sOF)a`?3?!5o2|>Rl}~cCr}4f&LI5w
z&+9*<%Z`qM0%{}`0C7|lpcu_Q8;j4`mTu^QyJNJe_%ceL-D01vm_s^9SN~k!^IOQ%
zva+^|=*qV8O{pd*R~qdX5UAsu>idTBjPOcZPvO?`mp{+wJruDWRDP0K!WgzB?ulA^
zPw@(tD}`4pzif$S3@V@&s5ha^C3-PDARrnY`rMAAC~Wp3LdU(QSU0icJ<=PH+$wwR
ztP|(CmhV7f^{fq-z_1+N{}dvAtYy0s+oKyfsZm%|e(n6dF7pUds`%(ygVkU%#vlqE
zSA+`(2}cK2^?rvuFS|e*wJ>0-an1dB{U-2pgok~+yqXYh;DLpZyG&me1zi@9CMw^6
z=&63HQ(>2M62#Oy|N44((V+*=p0bVugU<k%>;(iQIr;qXQt*`U+LIUzd5arNbk_v%
zGNU#z{j#4ri=0|47LXS}uvMFBW=);wKmK_8+memXaiL6sx4iyTcL?^=v2ug8IIhi<
z35)SVIQ#Y2ax*X)NhjlTlt#S!_PvY|!qaD$xD-!VAyW)ls>^2a>YwY^yL2%&7wGxC
zna7PGsP@hT5s%1$V9^i{N+-4J>=?X;eqKGUNTrC8#}CB@WtY!!<6UL6W{c%LT8$8!
z-CB%<Ty1bltSn>$o~{TpnY`g7_#qM~vI#&Hy9Q9l$p6`?NS<nriv9ml(Vl}tVBty9
zTQbfVuH$Nj?A$>8;U46>gBaJFvx6n#^hoEB_5M6+-q*Xm0xb|8#BcrW(tETDx))e)
zB6YrxD=kmCuD#_2kd}+8w_M=sq&0Yb8!IH}aXL<Qy+R(-=?uNZE-)_}q{*wQ+I-gI
z#grOo)=#d>9sta+Cv0dj0tl0_W=Q912(k*wabN%@R;6v!0lH{SFr9dk4(!~vU^n3x
zcnZU@;&T?jh$2I#AsybfXA#i-GOpWXv5p{WEM|_<CEu#&+W8`*X@nBu=-4$NHOvIk
zLbSM3Lxg&XnuqRGO^T=SLxEI1R6HWw%8#->a{(Z-4&3JqE`OW(T3_#tmnLRR`)8`_
zUT%Fay705p#onox8Mt9_gqwI)6gu+#=hJx90Ri7#1IKaOwrwvjI+^`TuFMfLKLde|
z3@<D0ayH|tupYTq@v^6M7H6fwIv_{E&P*g%#5PPECP?z&t1n2`rX^-zo1A<PTbJ|n
zuvBTfCy3%;qY=oLodQ9%$&@iJOkmLoi=>LQcge7JAZI_ksh)+4_b7^6>K>bs4qgS;
z2OyMut&(kF*ZB$CRc#I5)pOW+X3dV%kA}<%ra@Yp(|#Rc-r&IqLYNZKI?m@uEzt{)
z0b9Fuc%cxIWJ{MWO%F4g{M$R={k%OQKBuH(qs6PYZ{NzTcwiVXR$!%okMuhFpzu5<
z&+j4ju*WKdiOQpoVW+8p;||VE0$GeAg`bJvjR^1pN0jh_&ODp)UTg2-f|*qHQOnl#
zP9sf29ylgxyHm>}@Z;oHHi#q!_*#)-Tu!&b(Y(#vZvroGTDjWu%~bH;=0@I2jQqy#
z%>b)+ih1SPFNDL%H#Nu6Bj+txcd0DKnt4tt93*D+`hJ}nRDlBeqdlFk!kW^5pNFW$
z_{J}k4RLiqaN1QtV=F9BwAki(0Sn0sr%5o;cI1$>?0_1g!AaTXGLINY3jJhOZ#()*
zo4OVNZaQ5#jvp<SYC&i#06|oe=tks&LG=gn2_Wn*ozWmTuT<A`{Q|<_9Z4rqMt@_Y
zGCc@T9|t1u-n~1iaPHi>x&BvH&e-(>y)fb)wC=J-BEGF)9%CA?e+4QabZ7T*f+UEa
zYE|T|s=}|965cR+HEIcvExn)AqD3$+u=+_R@scx{0UTBt9DS0gTw_PimQmm_SC`@|
z87-tP7KZH`(ZJOYHpWj89qZ(j7nigS^7PCw)mXXp(D60SzdUvyRRbZbL?auSIpjw?
zIMft1F?QQ|ckQa#Q+vm)+J(PB-{(yj-E{WvG{6>YSR!<1%nm6s#jjPABDd-vQeC;C
zh8W7C?GdLQJ=>u#3Cw8hMA1SSCBnjjbpa}#%NXls^n%fXe*L>B+9%OKkVgaUsIjzZ
zRPQq>%=fDJ+~LrzK=gVG0n`cf!>|5ftsVGp)rK~SAo!Bo0EFcpwsOJL1RQ;n3g6At
zNucHr5;rpd1R)DIMwaT{8N&SzU)2ajzz<Yq;9n<*DuudI%W+Tng{*c@JTXJ(J=*~#
z^y#Om*GsD=iZJwja#WW+`@u71wL*%x+Bblhu!JGganaEia1(88aq6vNeX>_=<1DV9
z;Pav2Amar)1{KhG`>CPBfejqUI@mwGers>U4RRY;cVT1?FGlolA}CQ}o3I;1N&Mwu
z*jli$IQQGVh0<aepsXlhLBP$l(2ZyyVim|=1-+=M;&7-7oDUow?|(L`$l1F2^CPj+
zJ>@A0q)un9WFEXs@Wtu@=48RBi<FI*{_um~t^#X}@B!T3m^Cy!tKWF-R`{yb<0mLC
z!9+&zG1rYHeg1;n;ZiPbd{aiwhkX+Lb*&|sA?c=7Y!d3;lH7H|ImoTR3U1%idDb$f
zo0)BRea<$nRN@%qhPT`~uIX@`KyRy_qqNrO6g-@N{uj=Crroe+5iT9T&*akd;knXT
zIPt3Cb|%E#N$Yy4o)vE-a*U*uNAE10@?=nxz8jh8y?OvYtcq_F9$rKqKt4S0lRDo8
zSZyDf|MuH&4fPc~^X49D=-4slY~iHZ_7A@z&fFY^J_$k&Vj3#FPI!f0lW_PRjK9a=
z9&3}&&`j0@I&yNDBl+&G^=@5^$6RU0s^%$yvqiw)2`Vy-3GU+{GL6o#saRL8Jg;tq
zD#s$H$75Ds!Q#dbPrvTMwYB`|*EUtNMji;c#(8x(m7IX;T=s9;^vKNe`9BEk;Rx*F
zWg(t|qHj0=5((%!kNpYEemJ%X4G)z6b=DjmGY}P=`O?4IlLroXLHG=2-$mPzlIskg
z#7(R=G?7i}Zj>8QGmiSvP+)Mxyq9qD6(^7lZQnls1zorqC50GVKhV^vHY%;BR0!Wj
zXg^}bJ&5u^9HY{xw37h%sA+6}R({RBjDFOnFDGAgD(GltA!|&xo54^<Y3~xh1A_4H
zl%t*}n3uh0T)b6R*cpDSd#ryiHZz^@cwLR&>cBS)#>3ksV2qw2mLfnc(SiXWf6q%W
zPDHNL$&RCLS45(4@rv&PT%r9W6ZcbY;{p=DpBg-iM;HE1eYw>^41u{Lh(liijl{ir
zPrrXMiKm~h5}DfVBl#sp9Js}y|GOA2n?!V*MB5Q0#KL_?6EJns+Q%=1zINt&w})c0
zW`xnT2-X>RT5+Qgr<<qGFiriZE+GMj{$yoZ^K;9;G}7ei?p!CLqEUu6;HQH3fNmPx
z?2LO`khkQ8w?q;~jzdU|Wvpmm3Nj}|YPwraJkv+!0d_yA%5yP3cyul3KsV&fqX0s^
zLqCDm#h>tErafenvW*vTFmvGp1+*m?{pd9B)4JCnhaQcAy_>0N!V~F2WAa6A1QaPZ
zkLa3;ygO$I7kBviHe`+AQhR$YH1&WcJ8xck?pQgByu@9SNmH0XH|V+bW*2}Q6<`9`
z4Vw!Fz<0)fhrQE@sxvTDF+Ro@Q=mDD7x}3Hch_5sAm=jOv}5BIA2v@1^Gw6Uwt5e`
zPKg5227}cfce%s%jj~U_5ib44a?{4(vp|L~zEWq0?al1%rD~Xo?lLMsWpvHKrkz>L
z>WA|RGxz$WZih!m3u-K5wyo5CLc_`Ufh!X>U-2b6bCV%OLAX7j2GMo|oy@S^poB8|
z(N#x!lV~|!VS@CRnGPXK?DNS2=wdfC10N(Y%TtTu=sN=Gvme;}FNJ5n;p(6P<N@ik
zyVjxvd==%|D}sIsfH(}ky0lFLph<o7sK!n8AA_bI428ra6Sy`2RLx4=6*ty36RLp8
z2MOz%h;3@u09f@t_zC$cx;T9VwVUD`!3QZNQ@?yUne|B6kA!zlf_XIn1_H(}ybdIE
zy*SYX5k(tyumFlSHaSU0jhf>oTf@mYj63^rB-UC7Yn{uz`)LHKF}7HcWZR1(gvM<@
zr15+SritT+<SKS{f3=Mwn`k>8;KXe5F@RI)4o^;FVghaPi^0-4UTJ&k0#FeO*1h>&
z|L|gIQLqDLvwO@TrmPVvTh62ed`Uqcy&)cT{nrQ4=R!xraL(@X<y<Nr#DJF=Hn<g|
ztB-r>PIDwExp<~|&wCcDa_f`hRtP$fCyaSEWuZZ?9VM&G@dK<Fj#12`tbw_gvrHPw
z`jU4eXPS=Jb&!YC?HX2AL$TTmS<03Vv<tGy5jtybMECsuC~S(Be+n`XBCfBfxooDQ
zt{@337P}c99v(b+9f6{>x7aPzi9&xq3KPsSc<iDzVW8h%xx!Kqd3~c}MNdz~58%Kj
zV-|}&g<Ci76sI|SdT|maa=n&py!>d7Ij-t!iC>qOm!lrhz>_F}jVmV2fWO4~OQuhB
z%nMJ;zCuT7qrSM{^e$i&q+?7S`Jn1|7{)<l-Xmw<J0HS1R=!aI4a!G0Prt#pa?c*N
z@L2NC9n5978XbCu&!0es+fn&s**9T#4_64yy<L0%@j$#PMe<{`!C3F&4W~5E@$n+{
zqnNRLj({>83ru^LQq83$GtgHHUa?pYo*n(+wucW6gG6T=oXP8=cIQoC*ox}v!JUQZ
ze&)35WM2$vn3IMfV2*0vPb*gBwAvXP>gnb$KK|+W&eM0V=+51ec3YT?M3>KeW1~<y
zrWJhD`lhCBfnaS1puy|n1Vl>{Htxq<s?9uATRP#`{j5C#GcT`aUTH(6YERM-Hil_i
z<^#y~rXpMC-M(E0idQc%E<k<&Y$=!F#c3F(QouM5{VCC|l6h<i8C<=MsMoWAn-Db$
zIVcE+pmF<BK@I)6<A@|@u#+-+?@flA0$srK^#^@KU#tii0U;w;xxg4ZDglGhKNca%
zIx}4A=>_2FV|NS!E<A&_@FlE~&tOp(^7G*lH^`Q2?Tozh0X&n@QZ7A(5qC*s`%O5N
zDX4164wzl^VlXj>z=+Vs9#g?JOWl18n7Ck?n9PykzH}dVH%D_TI=k1+jnr@;2>_<-
z_K=0g53OF8ugn84yZ{_hbRMxV1{8?qWJjoMiG(fMA3J)~6u8^+>NnYfYs*3kJy0dq
zNYpXwL!;ie$;wngOM2g8hMiMi!p>c4<RjC&HUL!Z1=;`&IVcCFAl$dUk3N<%xCN8L
zDCF3RI}iI25b6a!PO{XoW5?QIuzdbE>ZBloD8n$Rp%4I9$iiK-%!Kq*NoY9_NMo_j
zq8)H4*bNB~9;w>xw#jK*^7BOlb+?NXC`43noODUkOvdN(M+s_{Ydr+ogt@9IhG*`c
zw4D!ovCRu<%3R?v(E>1ptMeccjeTl7op%J^i8xBQGx=~E<J8v!5Hg%SeRZ2W423a+
zr85v2NBwTHKH9@Tna?h69sPOOZ^@EGBAnrRfW1#5qLM`TH-B5MFA<w@v*;=|&ymB$
zdp^W-Mf9r)nEC7w5ovlz>Q;&d2GQq>)=M8<5ek>}<KW$hT<iw|POjOMK>PasjI$+2
z2{g!bKUFtEyB6&?A-NJB<lJy*PYdgdS6JwNhKyr3G7f9wcekPuNruo>rx7Sze*E+T
zc>6?O!1Uc+T4prEY%V`rixDq6{O9+#vbqUX$qQ~yefssyVBfR)hja?KP!W}Wov|MK
z2{eMd$ze1-sFVbnV|zJSwjR1~jer8>EXBxtie@IS8`|_x<SawmS=)l$xN1a@yX;!-
zx{0`V`Rl6R0<fRMRI|BQ^1<K!H3w6+kui*I{};B?o;>3ShofkCkfk*US<#pW;oZFX
z7b*nkdLjcr>de01Iz`ANd5Ju%LkWqjjK^gjc@0_F(>W)g06-vhcR*Q;!yU}@3&Ok)
z)<k>69ves{^TLbCVKCZEU>)f0<)wlL<W%y_lIhW|R65{rMmvqbHwE<-8l)7pJiO2K
zLoPAAz_ZXrmr!<hdx8p|3M7QhT&#|v*fv+~b|#uZiuQhNvQq$xPXk&YiUSdGfDox+
zgx?lOtB;z+RH>~pkZ4k3aoG<p8!|{`?CU8gwGZLw5MTuE0y$j*x*q@}2qU~F%2;F)
zKL_*d4wBeo<pD}W?^?Qid0i&KPS@O=XJVmiu>&)>e6~=jJOey1Q*Dq2(|~Xc2Ii+@
zfviMVXlyaYM5$`5%3DcRcJ>4e{uyXWzKTHF^3qEI$EgKF`Q3pFm;FmnD;H6%9EWg4
z4z)Ys6jM^5r$(J;UdpE493V!*3v5|)Zb3BTE&$YqlF)VZPq%@)W$EtWp@dE;z`A(K
z^3j?hQYIkC03CrndrosX|MMG>XfkU%M9cnSR@ZsgaNm4zkuEC=w(0P2=mN~TtV<RW
z0McP>5od{Zb;a#J)fYt5Lb@1$$`v{v_fcr@!6X*+6q<<w6Us1Sr#{B?Co#z`o@@{P
zl_MtA>vMK5V5(fQDq*~kYgSk-IAl*p*;m|br9TGUcz%Qb#>#Kj_@4Wk^L%cyp=V=r
zFnghx;h&}ty?BFmWqOny*}61<?Iydwz`;Gs*q3E#yKg$qyGcNVd+k?wo>%TRNV}F8
zl-9WPg^L>-xk;G~OLqj>OJ_W*3>)rQ7?~K-Ee9p5$3OCe0s4t$uzXpITg{Vt93-WO
z#yP29aJSqE3R(v|lfjTK!6YB4P`<^T_olz!?VY?Y_+F;txl`gO3zKw;BvL@^oUJjI
znn=?<{I;vaP0MLGYYY|Ug6krhC@$~5^2%<;QPs5k0nm(#9MnrsNi!yQ?AnDWK0e#g
zFSQO`$qUo~BpJ$F$l7RLD!%06gVn>`ku>`<?5eKgK+J4?Q1KcBBG*ep&HO;zNTh{q
zVd4`K+yPybg8lj8eo2%h1gw2<%{S%^1JstAfA550slVH8!?ucviMjrBmFuZUi&jIo
zAbxi@EGfNlSH?!iDpyoi-X!dbWG}e?G#ykrefno~{({I2g*VAt+72DqGCW-p`blPw
z%gx;fj!h*AY4}f5cp21^#*A4-ZEf-5A3+0oJX>rheDdPXb6Snye4jjkNlRKpxgf>V
z>AgIcqTg@&EBtw|_dXtjg6X|n+s@$9A?J@Q*MVPglu1*GP4uEViIN~!C1q4Wbg9*|
zFKTSq_v_#K`pyV}mjA4~w4m-ZnD$m!#7j!)$z%xo?SXm089gtX-M2N7I(y*rMAw<W
zfB)WUqE;-&O6I|9U=2dmi|v#j9sG+tW)~P;i;`rC_1#SMvu8fI)+9-~2ud*W?Dn{(
z1uy^e_jfaVcvfiUlmqeaOv>N7WdY0?DU1-dzw4=oQMkz%nQ--#IPW0=3xL^c)~wl^
zKS~=7Rb9eFa<=@fl+iOso}v%ViRgS2sESx#L`O$Q@SgLB4&VZRf&IbF)uQv^_18LX
z7r^~7w_X5g1{5Zl6J_8AxR1JwI+1=ksd#1}Gqk%%Qa-d!s*=He%Q0O{;vz4htgD5s
zQY(?2l(dT;<;y%9;|`?BR&6G6S@G9m7c4z*pr$M*QTN&IbtP>0$yy0fXr4oPu?G`@
zz2&YkcQ-dRHKn1&q&Z|`W7C5a!o2zYP$Dm<kmSsfXp05Gna1Xo{Sl<rG7uZ42hf>W
z*#{}5UaI?{?`o_d5B>i?rZ;uYd>?(()z!6w{ANJsDM#}r%vnoEM~B4Y(*Rd<GqyBw
znJP%Gb%||kFiz0*o0khI=Caw{yl+PD!30TzSi$61A_skKNz%;9=HKxx5}$*8S?{`3
zDaXCCJO&Bc80^GKA=1Q&Vd`_}tn-7lm|TGeimypx1eQzpL-)EK*dCcR=aNGaWQs-T
zIZJ<dy@LEu+_6MrpvAz-rtSuUQmJi=_B!-A&Chh=5u~Aoa~Zd<K8bFj5j%^~Z%!=H
z*z4<D*%xZ0r^&Ad!vP=Wdm-WJfl|F3Jm78wk{1}geG3BWPwablU?P1z6m~GEe>Wg}
zKPw!EZ<Dkxb8V0bKLxwsJ_I5m^3hG#6N&SXMpZy#ICJmd)%1~UGk-q}R5nP@w?1=S
z)Yu^HdH`=6gY#R+d5FYBFblBVY@@~T>(Yx8(mvpL?g!^_B5exX3q@KRl9+FvfP!Nw
zgn%@L%-<!Sfu9<ramIjkm7&D%28+?55Cfl9pl-&eq<G^=Gd~6N=}CVnc&~|K+hUzE
zX_I+!?iaOt_i`l?UZxro>Hd__r-#s+2TE=TZh0uU3@R}-SV**Lyn45#^<86}&nb)f
z@t~6M1nI3StE#I>!`TyEVyG@1Sh>6KpF<bBiY3W;XrJ{P*+SS>GS?D!2?7#N9WspB
z4C(od+`)vz`$)rUC0dZ{_W%yfYIYm3iS)3k6OTKtZveU2$&S^x=H8cg_;=bo!_~K%
z#F~0=5*dtf1f^1-c-<ftS-pLJ4;jW~)S5E5jG)0Q16~qI4bjcRCh2&|f8%s*W-~VU
z*E3ta26osSS{2>B@waSY=o4KA6t9m@G)8(iWArdlT&!Q5FInWkb?JpCdlWayPaF2_
zf%pp>iVo9hm!41Cer`Zm<PNScAs6XBouq*dkE}YRntIU0o+4Se_V$ONjg+T_1cAY&
z<Ug>5cKu!P+XC9`IUUv~qvnz92;gO3e#e@%8z6&syG6n9%%U^W>$gb0zT>+uIWB3M
zk0j$M(sr|C>CwW8ZhU<6v0NnXJ|srgJcrmJhyu8|+sZ^2$$@hJ;jI8lN$<$;?v_wJ
zL=f5=&6<!B>0^U;mw#DBqp{`4=D#}pO_T4eu32IbD!$oP(aPs^(46ao9*(ZrqaTL5
z7R)`$<pZH!BMJeu(WU|yj8CM2^)H%qqwhJ_CRS%3%`yht%8}Bx#7mtsF;8%>4KfTX
z=zhYRY0KThBJ0&5w~jtA3(Ij}CEkcq53=@(oD)HhXdmuPvt&0CD#G)7AP}^R<a9NM
zNP?-Li(&Cy!aZI%Eo$~Z5hC%cjE%P|KfmV_4}w3vXAwm~E+{-D)VY<j+|{?1KKh4K
z+Vl>UqRlLKZ9fSO-R0bg>mPiXMol+lSC^5$2%TPOJ`Ae)7))^so9B?9^MIrf^qczy
zcc75Xd9LUi$Zy>0A6b;Q#74n1%kI=3E?<+rKN^phVtlT$+x4dzy<vYem@1ZkQp0+F
zN*fkCf*9P^99p+qX9IAo>Vq<wF^KoY&Hc`m+k>7i6YKE3=YF9EyGf6z2+aUdgh4=k
z@y1Gs(HiFD$KTLZ;=PU4=1{Ekiixj67M9zsd60jMWs^gP&-O5UEJAS%R(ukdx)6s1
zfFxnEniruLcoRlmZk6;K@yuP`MS8fqchi9k4-fxI<%mAyRM`-S=U(O`pPr3_BcPF&
z|G=|X9sU<X78Y!`VLgQy4#6ubChZ?a$u`l4>~Y=Iip?wypl*}X8{<!Emv1XDQ_qS7
z>O}i2QVDO!twWh$dah@QQ86&La%eY(-7~!auCXdpdr9(?sPlAR#G9&Xqq@g#xr=Y+
zkXkYEOS<6``-fZcd;S7{TCXl#JbCx`SL{j|>ehMd%-j@haa=MZKI+S|9Na(I?XlY^
z{~MvF#tPi;7Yx@$E(z(||2>PZ_(&;KkOjmVplto0{$zgpafxJuQGkjdD|E%wS2KUZ
z_7}_FZzLVKxv!&tA0+|d<}XrI%pd=^c>==m|BmK5>02H{Mtl>?@gu(||4$pMR6P4=
zq~I%nUGI+a*?{CKSy%?g5DCR{T~HBQK}IlJUXX>wxduSJkoLb=qW}E3dpVnNXA9#C
z_>=q)S};@{2x*fd;Pc<!7=4kDrT@=zCHw!i8nJKW0P}a<%@aV2jjRG=t68ob#Vm+0
zDQ)5ggp+hBxShBG@f%y;Ais4-VcDXYy7<Gv>>)rOu_sxstgQws{Xyj=`X2n@Bu0oH
zYQHR$ktI@?>G1u{9bd2DrU<qR)W2p&pD7%zXa$Sd&)T!hpJci4;}^L=<`)ib`L_b|
zzt;fO+rRh0zt;da!vBMPAPyl|k&utx4J>zyz+qj{4IL>@sMsEzkp#8l1M;B<S;Y#G
zQfHKv8{dwBI69`gIHh|=UH#>KO<YqKoT!Kg$ym&!i;6vz5T?n}5cUkc^Z|_K9?Cv5
z2BB>roB!c+`uaSA@FwU6>mzsL-x!s0MMWC`57SoTdEJMRZ#N(>Kjl$=8NQ&QnO&IA
z(R9WD?v?1O+K=&rg=M%FWXJY52hdQu;6H&6?<)apLl6K&W@H?|DnhO@BcT;|H)`dl
z!H^5Fkl8$n5Yyo34HTAKW9l7guD3plU*|7Dhw(SmTbXyjS0va!WrkObcJA1517&Dt
zgg*KZ_|(D8HuT#zlAATTv?TXmNYs@_5PEyza|WcjsdGhgZLr|h<Q{x(Uny~M@iQta
zf!Sl2&0t5tgLDrs!5Q*7Df`irU1TSq`moz+d^>k``<)axUi}i^k?D1*_>Z353*Ebk
zPH4h`afrKHKRQ}q-G2jQl38Tvj6U40prD}JY4k8S$OZ+2QgMdwE-h+8#r)tgxL4w}
z5F-x_I9{$jY=!^iJS?%mNP(R%JZk;Nt?+|>hd$Z<-Tf2cX)Q)mWe0*`CVNWj*T>)T
za<0bW*O0pTn>TNwFV%ofd<00wiE@~c$gcXlRI23HaH2d+*!a|ma(wVjHg&MWWs`Eg
ze4>2PtPH<5`IE&D%k!lVmP^M-gyWqZG>Q@W5a9hQyjc#`e~xFFe?b~}H7fj*mFkc+
zKK~Pn?d|$_pT3+ZKv&{P^R72Ts(TA_S$=7FWxo@j-+!Ru)a^0l;^Yl@%kNI>=xhWf
z`go<fYb_gAZ|ch<GJ!@dmaQd0(Avg^9xZPIoF}ea^m7j~CaG}rg2R6|rC0u}{dqSK
z*i&~#B;#UY8I^sL!|h@+<h<Z3PevW_@$oA9vm*L<k;c#7X|lgHOEvc0gB8pb??gyZ
zlX(1Eu1J6X)xbV@lkf{S<$ZofiswG=ub21rF<)~ZmO`Z;UtDuyl^5DsN5N!Mgp<b?
zdm>~A2Z1)}qmJa4utP#YM!9R$@JJ4dhCl%J5&#;R^0|-i=<^wny>(lv<I_7zrDKBL
zp%GV5WF(iGwyg^LP9!|C+@Bxl8t_|BICrm(ywEI29`UfFjwQnlmh|H%bz(^qTLz4~
z`qT8Lgg&dgN6Vr+nQ`?@B!=(9I3M8!@4A>C-&2*k9$O;%_UApnRAgpz4)#&>gH`l`
z@_@nQVD4Tyk78CF!hL>aloa_L`}9~h{YhW}Sil$Si$Mt@OgB33<6E)$kF%>Uf}g4U
zyvckIfyo3kr#0FI)ai`?_e@fl&)6CgxH4GI&`e7kHwN~_m9j7IUR{aHZsIgH&%tr*
z^{0@-VDC)*^yoz6>4|B4d14pAK_La@?`9sr{e=t^y=Vb9h3)O_We0>rH|BR_gR;%z
z^T&4Y?dv;=hktdRB=W6o7;#9ZuWXZmw3|Laie)1`EuPp?^-tXHVY%=}_G;Ni$B%C)
zU1W0fPdKp?1|v$wFPp^2g|+FMS_c5Wc?0_}qICK~3)u7<mte-<hJh&qeej*v$1i0G
zwLlY3&~fsuT+oXx;0k$>!+L`u%fTW{PbrUPW3n}OIJ>>=A>Ps3V8Rz?tJI49281ev
z{v%3Uc_)l3mK?c}3**nGt7!kOA`{`*>I=8XewJ>!0j7OI#obTf1#)x^?*CC`Nv}eG
zHm&@;t{?yYzh&|CKV2|=zDYi=3CsU4-C6&q*WlBsU||Wd0$2xLp2@F7{4d3PUk|7j
zR=6?ufmpTyj6!nlpe%!M`!P(pMG~aE56oW<6{h0HqR&P^ZqKzo6<&>e($do76BD@Y
zcM#BASXfxcAnS-@(sO4HV1yzKGi;(MbfvKj#+4(yeYp1-v^zMjrl`0I1T=Rr+{%FI
z>4D=BBEXc}3zhLfspt1;dt3<s3*hnr+zDd<*Cnm3t%bCi?`Cfpy_-v)-%T_%51fb1
zbeWd-cw55%<=f;wz3J#$80x_ugEd5ed8cT~(Jzx&`{|k=+>2C+I$nUcFh6+%P#D|Z
z3#aLRB6l^=NiFO5A+&e{?u^OPq>mOaPp>em?`n^zjr;2;hE{fhe6x-kO^fIUd6Nqm
zc|^NSskyN(&iX8}fpKW@&n$^`k414Yc3^~}{r&`2eZN4#yY$>QekaLF;|#1BBpoMj
zfFW9n31$v@JzQ&q!;{GS7<zQsZx6_i4Jf!tV6>K#u*`5!t10x+BPE=k>CXgxlE(Wb
z8HjLJ80sy@V9mzJg~_-cvYY^bKl_wNY9(oaY>;x^2icwtklu6$lL>SPOghhuc;Vfm
zrL|uVb0t=wrx`_xg{AQ0qaKtnOfL_Vmkho%m9u3s7ndy$B?)(zQ}~oW0<{keSr0s_
z204zSkSiGR)4y!Jkbx}ImHjQF6!&qMb`Lh4S_Ym-C^|HED`y1JJ`urWR01y_pIN;)
zw?ah6-<R>$lW*P#MZctl8irmGiy$qK3ebh|E_M8|Dd)6r#1?*7>87tef^JSyK{oyJ
zH1)Q{bO;Qoc19W0NBKyFjtv=Os$;FkcE-u<nOwi;UGC3&(wT46uc#Ve@mX89L*)TX
zPZ8v0r0oAdDXt8uGJA(5@<R}1UqlPqSX+0aK|qX^GHTo*vdX7W<I|#Osu^kiF)in@
zQhBMA%|Jul*fBJALhg=4)(e;Ywu5WX-BRm-O_tY4I{;yIw0+T_l6--eQVwW~zgUn>
zd#H6&lVr-pp~umQC9L;wa1PFY3gZV88IbTlMt!?wBmSQfl_-70(J~O8^)T#oD>AaH
zJCwR6FljOIuWR)sGXe5EM6j%t7_nUFv+PD!>f*1Fco$;$Obwyqq1)R(K2Xu)Dn5Me
zj(Y~)S<1XcD*SxmQxfXCiW&&?gdV$~Yu5F4A0ClAtSkZ=mDSbRKu`M>iDGoA{)Z{u
zMO!o+V`R!*C&Idt?AtN^*Yl(9?4chMO+!9^K5o|k_W3@!Wga|%XSaV0u&ikAGdvEC
z4%bpvwWqaWT^)Z*fn;|RbBQw8hIHjDyir)#R*c!SV^FcWGrw-#x^n0oG$-UVW;bFQ
z5_UH$Z7ssE>59r3(>Wzx;A%*LN5lnnQ7jS{oh+=pSPBeBX9v|v*f9>|fl80NrT)dI
z-JO|T-t@w8&$7?EJMvRdmJI>0C6-%+PAcxbp^n?gVev%lua+<N>Mj`4WkJlGeP97l
z6ApqPfw?$~Js7b{oymo6RXM<E_QkK+e?`|I7o8lYyANn)$EdM{{*Kp1=y&Hzw5-^$
z@8K(bta3M5%DijK;=?gq%RH*Mu0j9glzv~S(2ZYq6Rk-ZN`R62l0x-9$H@ytJ%;mH
zRPI!8^AB4YI<4StVkCbrDZs+g_Sr~6u@8nnrk7v<HP@Ku3JQu9+~gabIzTg7wc@}(
z2aDRFbhALsWBgZckM~it#AoF3M$JU<8R_EUfRxq^(1A+Pg?Rq<Yn~~FJeD?oCOB5B
z{KgH;MwAi6hw>y1-aC#n)qQ5TEm?fwu+UT0w^2@F?(9lOEtGc(pD*YdED%R0<{+Z$
zrl=wIPcC|YdnKlUcMf%mg4VDP!!aD3Q3CcIR>#l^EUGidRJZQca@@H&byJ{TJ(vl`
z**c1JJXGi0*7Wyn$U|8S!fsFJrspm3_g+x_gCfVU|ED~Z7j;uZw|==7huUa1TDzqX
zIbrAQaa-tj({%Qp1Fv8cbXV><=AdscOp~pSIrb<g>ZZ_|PnijRhB4JK5NgW#QVUmg
zEiTOHqsJo%$^$vt+|SL;9VlINYjre7Cd-n3=A@HWYunZ1YHoao5H_w%^GbheGCa4M
z1SRf;3j*knmK)|~TAXT#gaSx{KV`@sq^rhv`k3Wf!=n(9WbcGMPxsxcoOjJLC$(tr
zQXgu+1^!UQhEA^HkTcz1*sA6J3R3H-E6iRb`agUlmAgNxTKft`&NardclJYOB_%5b
zslqa;fLl9ZKT|8!N8Q0ok!ZIy6(kpntp}-hdjy<pj9RdSt0C>;MeRb@xrWHH6zu{%
zP%o!OB5PYHE^gw;iY^bV{ZFytKTZ?k1K@6OxvX#=G=~bHK9}Mr70YYMx}XpFfD4x*
zuf2NL?n{kIysYM>ccbALEE<EHL(BYke_t&AcxT3M$DK0rI6C9M-<c=ibMf@gbviHq
zmV)Pl-kqT?Y~f1*$|ryM<(CW}akv0P%@fClpO?oK{M<4Sx~Tm{cI@{BUHi$-1VhG+
znx8*OduHaxPcr=}>bZTs7jB}%n*8GWGN|aB?w<pFD0A)?kLf{h8wA1sKv8_dEG3V7
z0av7m4wUQ2_;^~KknUvL;~zgoZBzY5wR)#L%p<|`J_Lsqf38ehVr9l+cDp=y?1FTF
zmDmYGoyFicu)&A&PCNSSXGXKVsoe>u33qky(T*H}d^UL{k*AD%Mx6v78FP`<6WZi)
zuLKd%&-Twn>z<|8kG){-b8^K|=8pWFcd5Pyb7?gg>z9E~iRN}8e6oX%YFE@wv)rR!
zufY5S$|U#id`2&je<Ray2D!xoBmcvP59;9VVvZ@QP9H_qN)*gluj+@G2XXH%R!ITL
z;5JGaDvsuT`G^lNs73)4zEiF%!A3UywF}co<t@e2S>MfQV4S0<9q2?)qAtVKW2g|f
zrasuc{1Q@O;mEhB2eCI-0o8W0?Z?}@Q?a$D{wg#Vu!KUNV2dk9d;9y3X=rFr!yTGQ
z>#5I;@xra>Y_CDuW;!`rGG9>Q0b2{iA8?w;aT3k;*Vj1o!1b0-aR8ZhY{lX~E7Gv^
zOgA+|Re7mwew6F{kTc=HeP=Q0Rqpoo_a6n~?c0$&=;2P!YUUYce)h)gosucrrq3~x
zC7&;;dFudtU}Y=!fqd+3=`wWIYeY>+X*CA#e7F{8%WW;5At9x&qnu==Yq=0oEn{%U
zECXFA3K<$@is%&Sc(Ms@q4o?hq)rSf1YxVl;NU8&s+M6ifr}ij55VDS3~RX4Zlq&9
zWt5l@K);&jJ{Z`7=N?@li)-gzKHUfQHMma8VHT%s=2FjG(nq72hCX~ZCE1uEWb3Dl
zxxNoB)E>K?U;iloYLQ3!{c*}}tY8Y*T=4A1%B>WJdE*6~Xfpz1)cfm?AClRPLGzrl
zI626SP8X8<CUgBduD<B`7eLpF&!Fa|0~F_WY31&r_}BPgQGc}VNz|^<qTd$(TE6AR
z6&b6j+%+8!8xz#Uj#!ziD0et!x}WsWylZKdSb9iC@?zFMgZFOCnzF7#oX_5Q4rHCP
zVvk~$YRa^3{(aLQhOeXxYtGAZhaBe)G3cAfOUX+)(C)Gr5yDeBlONfkAZmDtbPIZJ
z9ozWeUe$OD7+d-XkE|m-MHEqGbydo9Js&II5yb8?{X7tF(3F^343Uq?8(j<2Ik#lD
z2}!v^1Kx@Va>z*TBcy$|vql=^F%P9v;@jz#c<uU%c>pJsjXkT`<eJ!M5wbtcUmwg$
z)-Di52D|~s!5zfm10aHDnPvsaZX>T{GFUG|9;^=%vWh9?pBo+C+}cMh4$3rC<!xUe
zX7-Sa>XYUBcS(U2k?B45ih^Ol?l66=vUByuo`^wh%{yA?L&22M9u)iSFx+7yUv&_B
zB|i1!-HGD{o7oMtC&>=#6&&zh!j+(TevwdeB$S?%UwQ8z$pzEq7}M_#NXii~scFuG
zcv6Iu5TPP7mM(Ibn(2X-p9X$nQcK`6VrS93a2lAzrI>)_#3r&Q6Ua39T$Lia>du+s
zPKGyzI2|+Oa73{xqp}E~_YGiYPr<WDY;=<<$)Qhjwz`*1t=Q)=b3LVYo-mkOG)4g4
z@jaAF?1%={Vo2ZSdzcI*A@bFm@3|9;cZ>*ydCbc2$_5FPutW!{wO!vRxXhXfF3}G$
z{~R>m^I6;Y;}>>(|7GOp;}^bt{2vm8^p(4tzKqp>{Dw8(d{*=M`1dQn{};~!JdU5g
zaQHud$I{<F4?}+Z`@vuS{U-jGS&8opk1R)?!{h>>Hvadc`S@i1KUs-Cj^<G&bY;#Q
zlc`?cEjy6-HOr1&;QHS{)~lTRj4U!~Xoeqd*Qaf9=X!uF-=FUp&h$vX&S_U+yPkz5
z%L0^Tn9liGJE8e>AE|zp`bw6AP0==^%d?G`qv1_LNZc>rLQSa#vZF#*HD&}#{&I=9
z+<o&U(jQY`{#OzTAY|EtU#gu_T@ivNAN3=wLB2ScOae`XGq0R@b7n@nq%AN_)5iBb
z%U$>SJo_bQC*J=0X;GN}`uC0Sc{To*UWqG(B4BdvK*P(06^=Cd+e}}e-Rdv3XsHXL
zTAi)QBT?ds3w1#;D@U6$mX?^WL}XR@E~c};myP|(^!FO&mj>MNx^I5jaUfmuynj=L
zn$WU8KMhTe*J0$B_|=9XQy68xL9HV3`0odEi1CV;zk$+h|5FmY(1=dGfP9n&I5e{_
z!c)@D^YC)<Z3JMD$6*o(G*Q!*pmvk!XFY~DnoH4D&TwGAXo%#M3Fs1={4I7Ud8$C#
zgdJ$(Nusc!-^zhaYe_b{EiwOckX7BPTe+9Tz`y33R+WaHvn$8X2_TeO%cyRRtlYb1
zDKl`zw?J?pY;tcn6&Eerzp{M<?YM4~jRF}9Y9(_jdm%B~i@@*2H}ivaS20GKS#dH?
zCSKFF?-ViIA0Zd)MvP2Q=e>|x9AH`<1N!W~vDNzq2<<V2%d?AqI1_>xU+~vZMnbBx
z8{7D(+v0w(ShI=1N?tZ~k~1A-Wl2<d;=EJYK`z6TBw(sf2C^-SH;^TlW6Fn%T#0|;
zl-P^kS}&0UyoJ81Ys3Hyt`8#g`XU8orY$|t520yH@#SL4doX6?g{9f8nxsx;Od7-|
zE<@&kGQ*oh7=8gxF4Kh+1HO3zoE2MDYE4&;vx_i2Z!yBBa@m|lg95X+Hk{@x%e`V=
zDc304bGC`2%iVX0nZKaGrWOMc>wBW(nt#Y0e*H0?%MKs~l4rj#59MKN?N!PND2;%k
zs2aIo0ejM`K^SWX8pfMrp4%5{rUx>xhj87#t*BJ<iGf`Rr&KBfwSrU)QObBDGb`>0
z`z;WX5#Eq~eY*Mj2ES<~U6j`9BXeOE`Mf;~yk@xSUX-{k?!!zU@!qT6=F+ku;besg
z=7Yye6JwueoVXV$J*6aXl=Z!XR}B<0Bm_{2yo9u;YMZ~w-H}@%@jeFJ){kdZ(-S-`
zWh$<U@N<>t8nBLjHRY3ZG^No<$KPbfHy_g^Up4AsLV?R9WQx$FOwz3y_TiG2!XW4A
zVY|CO6B0;J_ACZwc6R2VNoR#3JxD_uJvYyVJ-?@83#=o`eg_cb<3lreP*nFYpE#Wz
zEEqW6QXZczonOMB^yCbXDsyU9j{wbT6X2=<#ky(B8xf%fe0H*;&S;L1`}&}0h1_X-
z1EFHp%`2L>>1!R--kf;mdjs~RzkSTV-gtqn_Y{e4b*J4xOpsBGkDs64Y^|j$9y8uJ
z=7LP5e-Uh&89?Qqp`rLV?_GK|p&>EV7!h<Q@+eLY2U(KdNA*T2t_+Yay%ueIFF=^0
z3$x|Y%z_PX-D1FsC?#Kg{as^T+|0kGj+q<K>QB<_<rvub;zPq#ds&pP*L-8Ez|B0U
z@`7dGHh!DV{^ya{&0Qf(zQe&+G*3Vnhha11$M|tK+|F_(Q0p@0Uox6EzNz|Z$%dc7
zKRIdEMRy8~ATT>Y0CI_!O?KEHFk9eB-O<aKyU?eiBXIf!ddV;0j9Q%(DPfzyvqzQW
zBZ(foOWOjM5H*3UzzdY2OW*)uYeX!e0{>R#0=BUS4XlbN?)L36<x7;QFjF2OJ8j@-
z=||gco3O+$5BWhzsiiDu%bUlc=RALm=Q9vldQ<LjF}Z{w8z{5_@E26D0Dz;)RQl$T
zEi`YHlzp7hj2;hJe0_MsBTWR5a#XlIrR#J(0SlJdwVMY8N}*a<ho6WdCp^9nhkbPy
z=UM`%+wplW!*jWZrZqr--@IJ>jqe2@Q*`M<#AgkJ{cH;n{AuJMz}y+`0~qV`JTXtg
z(cu_-GHfPH4C-&n17q9HVvvxK29r+UGEra+t<Mq$u1l801J-~lJ^*&EI^lKj{Db7A
zwVPAVq|vT|L$BKr{LgywVl;|z3Hk2Ipl?V8YtPx%^>k%q)UvHR9aLfZM$!CDR&wRu
zH~X2F@i{(7QXkc_DrL2L^wp9h00i}MOZi~dVm~1Qm$6L~I$gKix<S&7blpR&?k4(f
zM^T}5lf5i00HIX7!etoQ&@EuP&SQYohj3}UToi8d7eGc|0_y1jK<|wa`5j<w>L$CI
zc4sVL2!L6W_JtT9#he!NmxIM51G;(-T+W!5>{K+GHgTTGXgPy70-X^F8e`c9L{%T?
zg*QI$QRHRaSH50;gQju>A@gV&d03u?yXb!Jwb+k;X1wac)2%Odh*(Bw9)X>}yIIkr
z@&-D31UoGXXFLrO$4?19&FOM=qMJmdei0j@C%fIkgNk9oi;kfE$UJj6fBbl4Yp@gv
ziUri&BN&%r<x7kx=%KendVWWMmPA3RgCO#!SXT}bIJo_6M*&?WBrMv48Byk(7)`>=
zzVLBfL&CKt+9YGBkLb{A(GEXxnl(D<@D6t{v;7G(I<i~WJTFbb9~XGyUoM0(SN-UH
z19y5<#}olR+g9DoZ2Ox7#tO1?uf%D=Fn+Q=M@d<E4KOl=xvwW^P7wpPk%1CJt2C0W
z(=~eyvc{Y2R6XK53%L`UhLf_B0Z#G*^a{xwxi0<TylSdU$TcmY;y@=MrCx*B@k%)%
zC4u!_Rj+1M|2PmVUzRJxld6&;Iv$#IJKtP`-z8mhDns*+LRC(oWe)s=W%rU{^{SO;
z1`>fh#M#z`2=hFMKqEofu?D!0pSK3w+Vp^v--P}~N&N%~^Q>mBFc(@$r^voO;UNP7
zayy;c%(>Z01y@DK4tP%oG|vahZd*m8kBX5P4dl&uMrXFUzhr6%Dw7HY*Obw{l-^F$
zwyJJG9Q0}D-NY3fFx*sTd<Q@K?7)N^4gcJ{fB$}_sWch}t1&MbpV`@#>zzG~VP{Dz
z_`}B8E(gAKsT)~VO*|vij&75)q)FOfp^rOmz4i>)H3|@_iRs9O{A}=eE})DU?FO2D
zJc7|@AyNy_xrss7m?l?@Br9cSAUmj^iGk7p>vSIM(YH&XTbcQ6h$LJ#;JP)5Y*@bk
zoiEMm>4DH?GIK^PdM*UcmQUD+?Gf8=l5noGLOGadQwuaKpI$=zwWd0ZP%AAQU_d18
zR#RxAK9G|@og8#UYfZe+X);!nq+OELh74!IuZ;s1&wtLh^d@adRPjnubzr1xnzmM$
z-aOJ-ZI<39mO2;EMDR1675_opSBBldc26+LZO^O_rq)j>=?c~6&@|Y1$1zxd4~-|h
z1Ke_Ih#dkBhf+?~UiPe>y<M|SI8ec8?pn=c$?g!|{hpRC6DRDQ`EJEaJ3PEUBerD~
z=y#R2H^n)75{5KnbP_`^_K9f7t9tl0?etGfx-DB_wzBD&+0^!iZL3(j3icyb*l2IG
z^<AWpb?|Zf!}Ev3QFM&~UsBB<r5l`v&&phqkau`XPQN@6DU=O@0keNby-vc0dRzxH
zKo2Ve3>`7u5J7`{7~tMDv3Jb9@u_hUI#6%%NTH-IIn1-0G2^yqzrpJ1$~1I{Gp83-
z<#moxdv<cLiI2w1GN&WOZHz#g*t&qw7u{hD01zxD<+*&A&NrAABmm?q2`pGZl2&(O
zei?fJ=ZSf|#r4z-Rc-5P22RF_KD`Id#8brlz!zyC!TUscOYDTDrp_?W@7nNRw?ecj
zM?BuyfxFkis`1P*8C?E<&Ibkr&^YeuxNz$e80^_UEP;XTo}3bf`39RBLEgwCC(|)i
zErdL=$T|nYDBLt=)U*IR{yzFJyg6b>zLV{swZMTepqtmO=n|>a43ZNY#1)SZjAIx6
zjFEn$thSEn&=6q0S#60Y4YH~#O#>4xHTdt_O=!v+XWeM4HA~k_pYv;?e0rVMufI@k
zZNfnOkHY5D7(N+$FMEAg0*66tRizxOf5<{m&5<JvPG+1cWm7YHlU3DpNyooQ(Z8vF
zYhui83q1Xn^y9IcBbkZG!N;^!$#b0Y2Fhi-A?JG&r>v9~55K|-*qjVeA7>)5^9c7|
zKV?-b(()9ad)+ejy*Dsi$!xIBVrGH}v}9dg4$$sSV>zYuV3la>FVCmKpD^j$0_a`r
z{q85WpJ|S8*V*h{1a?wm8Mn<2LRuHcxE61{kgv9EnlvWI0UAtY&VdpFN}WtV*JIFe
zeTs=iJzN;OJ7)k8_*4rN?24K*8<fb0)-0xHXJrOzz0p7@mszL12E+cBsIo%cLDPFo
zP6i}JbsRgOoZL+@7HEa>rHvxJs7H3&GY}6#1E&#gI52X>pRpiT@?Jhfx-2P$@&n?6
zqO+7_%>jWTZ|b+vH1EzFnh;|Xv(Rdrr{<v*XgskU5whF5>1bdSGrLNp2-}uY=s1_C
z3|}}t3H4^9)p`uqPR6T#*dQocw4;mNV8PQgGmOr7kd^$OgGid>G)I8w?!_W$ns3Bu
zd68??Q{PKtu{d5D$)ZC`HIntxH60DI+?y^T-V`C;#HhApv{CN~S%q4CK24fZiOfEr
z5PMG)K}Ir!S5io+)F9SCX!VqXGvA?&HQ(uLh1_b{smkFXugVvhbXy9q8fb`IH<`O)
z@0;X~4~KUl=nUplv6<{<qXS9s28sTf5PxMPRxUdwjnHT?tcya9$-404INIUchuxYr
zj14=EA+=tmAUUO6rB7Y?BPet*m<~l2rIAI9(?(zFy1HB<zJ6k%nFu1vWW1p`QZVe`
zhNo{b<9`$BFA`ng_fIJ+IN_s#5>pR-pqsEli0z0Il^Vj<;Gx0&r-cp^JChx-LvB*R
z{(#|LmnE7K1#Ej0(nF7W25gr4pQ0MmdF{}FRDhyWy*@3B^$dxbDh{h@S;_o4tXdA$
z2Bn2?l9sKvFm4acan%O!e-p#Aw+qa13mLWz;s{n{1D#hn&xxbrqA7i~!3A<C+YNP5
z@zt5?QZAnao!O3J4F85^xw@h*ieu#UwIP(Mjflf8<6+rcI7rNYMv}RC`YSA@u+nrZ
zw)ImaN(~l6E8J!e=M86?HtqZf((AAi3ujV-xJz||^U$h_P_BmKR(kyi0xBcg4njh$
z0rSlib_|cIo-A2FoLP0Vt=g=$%Cxtsa4Iny$%N|rL;<5LWGhK>caZIv)z4Et6@4HM
ztumtyW0#7(UC^SJt+L?)t2m`PF`bB+$!2!w8|4Xz^au*{oE7p%#KDbY_G)gsu*$;D
zH#Gt7W63q;Ije5Y-C9xrA?DY%p~-?V2~|^UYTD0sj%ym3Zb_{)P1pP*Y3&*Nbj>ZB
zQ_bhAj_M??2;sgJKl`ea4&2aq{M8S}JF&->>VW2R07=#hEM>qvk$H%v6@STrVx#o?
z0ZJm#QWx)?W>#zeT$c_I2nq<;#^0ZTCmc691To~Ng5?{O28e=x6BWYKGsids?W9G<
zG~D|4cip694Z0tiOwoFm1CaS56rtYM4W0A)qy>)FVsBhEYpJV$!!qJuGc6?xA|%kD
zto{V0IXe)co!YMG{^)^eFc5LbivS5Fi|R>}d#rh|k%<EIW7xa`NC(pE#H7hFl*Xl~
zR+TJ9F)J<?yk|4EoCg?Yhj*zz+(fk=(iWL2eQFR>hGs2KZDQ;P9_062U|HSZOUd2R
zTTGMOh`50oCHVf_sphjv8%gEzG6rq+klzAXT}9F5CoZW2y-G(T5xj{6-^Zi03B#zj
ztlOWtPG`JCQl8$Xi~Jvh&{#pb8rpI!Bki0}knq|w(6OSvZpdvlUkoybW`5`#brs_q
zwu7XUi33s<4U-%JUuHbvNeoQQPS-qA+b^P#>}r&Sq>FgWe$CJZ-!uZ+$^~eec}?*(
zoDTD}?2V`ikP3O=jJ<OteNKuBd!ws@DK(X551Ni5KvnrQ3GpPV@KY5kA``n=_fve@
z|N4*Blt>1S(Af=iEy8VZ8Co*+Bl@`febDdAuSbj@3flW}4pkI^4)r5(#chtybI{om
zX;v_IZVqe(V)Zd5SSS4UYTRI60><sk;y9+7R+<pFWKE0Q*VhQcK5~SD>_*7xW_l;`
z!lceBZwQt)Yfv9lXi7tf$O|c9kVXkeR>Ztlog@6KYIGj8Ss9|?21+?FzAp~3sa5D{
zyHgoU=rz;@KD;glUQVc@R*WO<Fcc3y2w&2K1=9B}&ZRE8D=A2d^a_$)Ty+u$#@ngN
z?HwsWI$H{cN)$R|V|PZKdEh9x!RcMvF)gH$W0=i9f>V=M-f3OUuOboWoCB@=%kR?4
zt2#*_XSONTHE>D_F4fJm5E#-KVP0LAdlC+u?Rwj(nlNhb2QgwHle|KXRFyOoXxsQx
z)zh}{4fo;35Y*t8L<XA`*tEq|3g>ugcjER{H*??K_o(_%#}593rrSO>lqFe}v#KMN
zas#(Q6#YXCxEIu<m>^W^IZr;}am`ZGt@znWgRF#GO$oPw`DBG|Y4A;DcG&LzeFu@j
zWO#-#>LzU>?*Gd@)GX2vU{@Y+r6dFX5DzaoP<DZ4YHX*1OBUjNfGY!S&!?Ef*4fHv
zMBU+?94VRu)v8%U^acq5$X$bJ;UDqybVKCA-eid_TkJZl%J;o{_~b+opvh99c`e%}
z6?BPD&<N3Dpxu1P3ENr{gzvrUs<I429s{9_X57bB2exPIjBp%$5YSGcUPz(lLo=|f
zb)@)GrN4+I)KClEdu{6a?zTTG)1^SjXNWDtZw;y#qrguC#00?OnB6xJ?O)Y7jgh9O
z*mfc5orgp~&LI3QuO0Iit8uf}nYw}9Lo}YW4s-O(1))m&61SX?X~umHpFGz&1D7<E
z_0hc-8CH+7hbb5T@{pu6G>;7TX0_Frt`rGtF~~BQ+MMXMUO2Xf%NzPsl88f^JucOI
z8v#@0@d#JSHEqW~KBPLcG1JAz%yiM6c^dsHO;vaj4I}>hQ)2wRL|WFY<c?h%FP_p@
z8C$*gZ5eifjLoe~vW|&!x#(gG#hKf#3={w~+|&bZ!bx_mXV=<hlM?F}kTLr;Tj6%i
zE*^M}l_lt*!9seA-*@$px{LDkuFW>LjCW1MRiZIU{}eQ+@U=E0STU<@0iXvnpQ##V
z?YtX*aUOdl(n^v>pD3Dkm_M+HJ`%u)(16SfsZ#Wgv;!m(5?hv&H9Ykp89c1GLI>4&
z!jWn0DX*xIXT56qsOwAtCBBLlT;kRS;UH73&pG>IW~5nR#r7<7%kM#W66G7Vm|w_2
zApX46!~U7xNU9TSlcvoJL~dcN$D4YSrYS8YEr3iy$UPv!wFhTZtB3Jr&0X;6g_KND
z@V6Vr{^h%yi5H5p^;CHls;0G7AlDbtNFFrEszInMRP{GtA6MyZxQ=3f(=!s>OJ-zq
zZaOhF`nnpo_Iz{^Qr(fT%t8c%OrPXbns1tYy~2%+tWA#`A^#|M@J@wsrL89Z0u??K
z#EyWbm^ih>QiP#F3QA*Li;S)@^fpR~qJk>@2GONSE$V<kC&xBCtOcCPAlD*99-)2|
z_d6MLXYp;a1=*pvS*}^Ph+&T&ohtyopigoXKp(dyb4j~86ZBR*-#M!Kpwm5~BMbA<
zs*cOlF0DR?U`OVb0b<*#Y>?!|b!3B06xX4O&S2&T#LAcs@<~Fe7hosf4volP0)1+a
z;r=&Hf|pf?Iq)E@OW4qS3@SN^Qc4(19`RKPmQgV+!CU3cF&v-I@g<R3lRN+<c!uqD
zphpA;+S$mfCI%;QC#tBKqr?iW#IpBUhcr>Fr6V0OWC=K9y-SJeb-BalOn}C}Mp;{^
zLCk=A-0=UDE1$#*`6S(TrOL|G58&u{<O3uzJ1(;>`=A_M_)CW};Uv?=``<l0Tt7}0
zi&ELerU?EA(3K#3J$j19&jEz&WukGwmG#juy+H{M80N*+OJw<!lp?2)lJ*2Bow|vo
z+D+FG6H&(Mr9t`dgr&}rXji}1>~t4SbEk&$NU@KnqF4ziUUl=xK;nT`M|K01n7NCI
z)_0<#@XQ3AXa5lz0>_mN<SYHac6k0cRGANSTp{z@Dsuw7p<-vG{0tq9Q0ueYV%LFG
z-ab<j68~0`Z!e`G03%~MB9h~#p;9bW2l2N6w+nF6TSLkb)b7x9iLR$?Jm;)zHX~#V
z$eeLZr+MVL&=IaBjkf8Ur3OeE@=f<A$~4L${0jwDCyAo-8!Tq2Q=d=LKt#mXO#J0@
zxpLenK_Fy+LrCd-b=+98<@|A(jdDK7A1cgP6TMLIZ1-<UIHp>C@qTEbs*M4A%6FTi
zSy(v4NEcu{kl8di{iBdW)fsNLr+x~^cesePLez4kL#iWJCV>ESL!0}o)}<FRge_l&
z>aX=_lXSqB`%$YuoW*I<>=lJvtY5P$&P07)o?q>PsUnKEYHeMAF;yB-qS#WEM0E>F
z-B<Bxu4WC{3g35Vaw`NT9v^tcvg@10(M3jydj2JoymtJ?wLIa!9?F{Efws`<Q~Nf<
zOfStkN6fTw2kMRpwtc^z%bGhrHTn8RW7p`m_G#W)5P}@D+IR8t-UNduKma;HhQlV$
z^RNL=sV0Ue-PRO9I<`RRm|m~;43jwsF>GftQ{ZjiKb+a%H|>I4acA-`=uA||1^S2V
z<PnbH7H;|}<bi7?*7bXbJsh4^hRDpdTxxJE<`_-eG?I-PuM<2DsGE-ob>CeJGFe>F
zD8;bM4xq+K&FMJKN!dZ>kWe%|fMn#1mevM1S2%Y-?8ypCdD|hs6*_!9nGL5r$o5V=
zVXq#tLn=MCN!~Q21{!r`Xq=Kv6Pz(7L2ZxdxCaoLUQAGQ!;%9_IXgFnBN3sjADgNr
z?Ujf~*_)z^z1t>sBcU54rQI0<w`89}AxbS~dZ%OPSqTND#J4u+9J_Pccch)LRZ)|d
zY7%Y{dp1aN57a<XvUpae<Rp#hmVM(DTd7B}%iamp`fg$$cO&AY*y$2nn#bQXv%k_?
zVlX}$q0LpCxMRtp#<ih=lL57e5!?)c%G+87tW2O5m5ym%mFdb>LT~z(N+x3II6`<`
zm{-3kT_mx81zLW*{2N;YQ?v|O9r#_QwtuRTZVNah_o3$gyxImr=NNRU^4miRse_Td
zkTV@}pverAbS9%=Cy^C-pm3I8HXqSt+NE23fYy<Ks{tvrSa3n=zAgz&J&pzUz;g-|
zAk8`^W16(IFvjRfEe2ddeo~1yZUFpIc~Eju_ZW3+2U?&zn)J_2MO$r%%v5R9&AN1n
zed17jl(#!>sJxmve5Dk{wUC7++Gp~4<|v5*Up<X~`@<!Afx3vlRbM!8BAIT=cbEq5
zeQ)iv=n+RG35pAzZy_X$85r><Q=kD|&?0nv3Z7qe=PNAeWt@R3hHJb#so-^FqZ5ZN
zpK}hi6|e<#@(a--2~f~F)dFDB|2Ab;OJ`CQG-On(!;@5R<1D1$EZ9wi;+{7=!QAd4
z{vlO<l$sWA7G_OxX7+zGHn-?&P4OpIn1v(`0r*JVZ;)l#*hJgCX<NuOXG10_rSW%y
zn2Scy81u_?uGoA%pg<SPb)rNEJm0qA60|o=M$ZNz3)usImi3hP)yp@a2Wl|6VAOy#
zMR(4Y`itnunJoKRZA!Re+V&F2Zns^FP^1v!GjK5B$K?n9c@)r2OkVckc`9+gTUNC^
zL#3DZ60ro{q2s}uqY!Z*@2?(cLp}~@_ZII9_8h9bO|`}}n)uc^wU8ZxA@qc2TRU>f
z51310_BuIwd;cVtrVde*Af}y#ju@R~Hu*^evtyX*4t%~+>Vou2Wd{z<XX9PbC73kj
zKT0tvx)+$P>ie4yQHshs&Y7bAz3;ZTz`9W>GTl8%)g>K8!YwVMtQ+roMW|M2Tl6!j
z5ED|GqMn9~L9Atk+@E}1%r@f1O7H?*s(}eiHDys_gcE@Hpy`+3Da7n7nmz`0*pQ#*
z5Bv=Dew%@D9BwXxf2up}hungS0T|OQ8Z&H(QVsr_p+`%*UQY8GsT7F^ng?lFAFm&$
zT+?47+qFs^4F#&mO*m&Nw)!YD_cnFq3H%{?gvx`I{By{e+|j-YDBy4MKbp}?Eie=v
zd(4K-8JSPam!XvJosvSQajE4wgNd67D$Gl*pcuRW0gM@_qtO`iu^Wk8h`15d2qaiL
zVBVrFlNYFPmCG<nVihT${I>+L;tt8{Fh_TWZ$emjHSHKey?d38fPB9edUe+5=SiuI
zL>-|1i#>w?9&5|Bip2!C4)}1AQh6Efv@<c;L_dk<NsYkZ8Id<tF3Y)8?M&CSt2AZx
zZ}KBVu)^Xa0$la+sn{`*V!ukcF5i%GVPTm!{mhONf8K_k1WWV@s#EfPk<sQKR!yAO
z;6J33$Sbt$eN_`pjb#0eXJG%KYjNI<uX7Y=-f_o-XWy>V`k8Q2`!47h@bFV2ADAr-
zM!aW4ZYHu79)9ETr+J~7Q+J(?9KqFUt#}Oi$;lEOJiz*8xyqzgxl`JfzrBUbv%8`5
zw?i9!EJqr$Qb|Y_ke$lRK*zS_au?yy?~6joX+7qm8D<E9i~0obv(noIV`c)93H_ju
zGX~blzTSp@%09Uze-T^<;qKr`$&4&OhDVX&=mV=9BZ)a0+QOSjLysvKu$o-xUS5Ix
zxA$Kf25FP>!Gpjs#i`>m3d2PZvrG6OYz>PLtFraRPZ7MUI$t4q3r2?m!0){i!)kAU
zdh^>eQ1OIK6_0D+g3!azz;UF#Bl(v<jFk(U_7lsfel!L9qa4-GPRSH0Lqv=MAh)QD
zO}8e>80@C{nkSJHJqF@0J8F~SL3<eP_4;CBBgBXxFd&u`7{t0)%2gV0hwygw_WWNz
zKpEBlqrGbjYQjvTaq6_&X-iEN9kqy1)EX<UP`nn<inf4=6v*PeMTG?f1VyC_;>w5$
z7#0aosES2VBoM2Vi;?20K)6*Z5~6^jTtq-Yuv`oxdrm;R5Bsn)ec6Z21470G^54Gi
zJLi1om(~70fm$(a>NNBfN<chJZ|=od6oKJzRMY|c?{A(HlHfl?JPk+B<;rG6-$F!e
z#|}$0<+FRrG%c_;q-3Mx8&?@wI5~{aHhr^UC%j{`nd(-zXASP(a?$kA?k{eg&$;bG
zhz-vD%D`5fcP!EV3U`RsZj3E)4|`dhm7$Q5c(P2*7SUb+R;}&phsm9sv{j2*Py4H{
zfN?OZ@wMWiKLw5+6*lWoTdiYh(cC+NIJuI%CPB>Q6E@lB%ja*65(-H?f9g}WU{;T_
zGf8k%IChtU*%}1nP!^u25b17W8i<H#IdAmL8!_WXJ|-^=H_d#~li%ZoE!sRKlhulH
z+kkS^xgcbBU9@j(_VGd;LkPjy;w;H~cH2s8lK^B;zgW8Cc5iIk^u}yt&f|G=%ca^?
z$c)mC*(}9zUPhgBN*Pt&h<`Hx&pAA3nJl~j^dc9ZxG^f^#pX&4d<6=y$<}z(Gemiw
zEAecV*$<OL#;Ems3Fo5dH)gJoFw<-#!-y%D+7{(poUg^$*2#-<+ct}%)nCb?5Z_`{
z*~}I<&XOqt?Ow4!*w)}R{V+iWt8hzrKhBmZB;i>HLT{l<BuXog;}PW{=q4Q!)cLo1
z+dB_=EK@zy+6@f5;m7Mb5ol#7!ZSjCG_Wd4c0SU&gK?e!B`d6&0D>YEGgLOSD}UMh
z=HmdM96^ch-aFmtJ|t|C2xG;^yI_r~9S%0w;mSBvoi0KEq~Wm0SCUIZFNHlT!87(C
za~`9}&__Y*TkTm=kl=nI;kR)Lh77+->lxYxOr$cA>~<#Ub3SCy>lmiW51B|2Fi+VF
zCwXG71fcLFkUO7jFh-J%aBo8M0&<e&#~07fhpdaOco5NxC=u9$5KbcE0ybt50U@&9
zPzs2SJI))V@&)8Nu-fepu1Vj>yR#ED!3rSCnU&SEbLu`|rYx>e04-+oVX4hJRS$X$
z=e35-SD^w(mD-P@96x4nn0f+O%Q=1Sr2rMiv+rofYKOu3ehmeuzt7_!Q(gmBUjO>R
z_qDAwTyONn0hOgh>Y)$!ltbV;=1+k=Vr3&PA089Ze_g);@nw~)UL$qZ9zLznDp6c}
z`uJ#)N6|XSe2h$;SGw6ml{?KPYU{YSOv&=#$LUS3k+OW`s~N66bD6K!WRtb@x-;$8
zKh*3rUo9a)BoxNxgD>5&wgqg?(Ka+Z*{3Mk9n<<9c3p}}jg0ZA<TB32%8o|lk=EMv
z)NX#{Pyxhb@f_t8Gl)HB^aA|)sg`%MkEI@=l5Uw!x0R3weJ`%<s}BiT+qpTA+JN4L
z+Xg`oI5wbow7cU2*cZ{dB=sQ+t1hs#DLwa8)d5my6R5LHM#AH)`OwgiazYbtbal@2
zNFvCdu{MoW_RO4oEX=b3FS5;i2NjSZ&NAN%D%0cVO|9!AFp0J=9=lu0y4@>*fD}Lz
zIZ%LM7d<|UuB3^<@_G#vH^Am0V38`lv8`wZO?Wo8A_ED>VVPVZqA9R+kVtcL7a8J|
z&X6-A#jdpONic3s35E{g)0-%u{iATru0sRvQI}UQB$y#<*+~jf)d`$rLq`{K?hRQt
zX4FE@wyt$y1y-qgYg`Q)L+IvuzPlyo?ylNtY)gq8$dEBRj!chBS~EL3budqS7aD-7
z<75l>stVKN<nI{gPTfiy`(IKsG)F!1%>cehTP&~|*T?^5D_Qlm>L=(S^55WDgzl9b
zP(Z~i6Tk!wO*7SXb2G_zem?7NSB6#c%ciUN`<?dO7#1K{`_C<Mj?WCwi%UoZ#+8Ws
zyKKd_8QGa6f>H)-W}H{WuZ;Q9DX6pK*Jbbj5)fFm^rfjT)>74eMoRAC1;|%Lb?4E2
z%+os;=pJDVkF3Q=k0^+mRd*hY<6Ovu`vF!ZIouTr={*EPPPd@-c8g)i+{mm8gE9MT
z0R4Mt`HQnazOx=i;Ds=@b?6vdR_WlVh8Hv`<Ma6%&tu!Kz3rjC1?G{W1-d%lgw*_}
zD48`N{#m;Rr86BpQ-+`nEC5(>G&VbNzV8&CNNjJ>S@h^+q~r<m7INmKzE@!W#cd^V
z2vjUQ>GI=OK3BPA=Qe1rTw&)+;H-emJ}6Y@G=saH5ekUa{JMJ7o8m7&!TEU@THDNB
z@09Kb=TpQECzJNc*v|0e_s|iQ>QmZC$tXH|Xh80vug2oc#hKMeb;*F`SKb-<EKmow
zWL0g+x=6*JHr?1-;yGjs#-(KC9W5FVJIz$3XX4n~q)jVaVmN!q^Yh|@0{{a2zT*vU
z;X<a~T5kGcmxDTb4HWoM-9}+4EwiY^28KV50h?p#N3T?f72_Q)lefqNf#rw7Pf0ls
zlxg$H&EEt-M!dVBDaKyHWOo<jiy{>+vVO)oN$XaQgk(>9Mot4xw<rz;^xLwkhuQb=
z)|6A)ZX#nwn^O22G~zuXse?n<V2d3yM%>%$i`_SER<QeZx@`kDHFbpz2bF~1S%&OQ
z2jq(!W&#D?jS}3=ii|M5YQ1)CKpr)Y$2|Xdx_Z%H*w!)v0f_~)D=Z)mgU$X}f(1(#
z-1hP#5x#?vlS-&|280(=eFbM(7m;EnOQ!~@6JeJECn~er4ODtE`1CGv1Yj?;$_G>j
zMpV)OoI3)>nIFZKZ5WhEGhk1$tk_0PBl?$3F%}ajcc28B1{FmZgHwP%NfsfMi9nrb
zu4DP)>4LV@Qw@;YMqxRgR=p(W?sFMQ?!|yf3!h=dDN$7>_zyt8?sXyYvDSJCLYb}A
zN&clXS!Oh!Onz~T**9)WAW=zrgk)n7v=2PmC>l-%SEg+r>x0Sg0-!>ToYc;smc^>J
zOzl!cwFq5Xp+6+R7n2~vUX$?|Y7N94J2#soh8}A-jX2-^OpsR5;<d~>%1?DR-SB*h
zI0==#cy}>#WiW;+M4!%ZV4a5Q<%~gn>`(SEH$R6@4)k@yJG!bl=(E&nEK|xbQQ>;K
zZW|obb<n>gNqdONh=dR$eoY94EvTxZL&J+%F-r&Dnn19bdZo8?qa}n{Htu<Fz@#1b
z<V^fRFThDhWW!5{Vl55_KoaJx_+jNQ&lbmR+pY)Q*9xnb?pYeCM}7bnrIu&#&%quQ
zp_AOHOn0ji)As?kddAKx^p6Ts>@j$DS)zR@G(0cyp~XqvU!z!QMCEULDHyjz(JWuR
zap{(V&NXReW=>`oC)Jns<%yF_?5qm=I&^Sr?jB5%*CoR?#hcoG9W=b3+jM2`1!VNG
zYb;Tky5Oo}9g*P#fOcXxoQB>LAZaXKYIeF2J_Mq^u$8>Dvgy5vg0vOlj|{J5CJt+{
zt(tbbmAZFNUGg$U)i1VaF-r*he56+{Fa-y17gCq(at#wkEU@FD;i=y24Z_KEu3SUe
zBc!|1(C{yL+jrRpOJtgB<)il{8p)zN*VvsIHFYwOb#v2Z)4(xPhe|xHnnM0vsJ<Jl
zby6WVo|D@k#Xx|UZ!BG2bp|0k_W7}>&$^3wwSz>IY{KS9*y>;+!4(&Nb5JiXG+dBs
zjh2tHmh|bj;EhK7xTRcL(X@2mH3M061(M(&Ru8uYF1hCcP#z4S-qvbdXZ>-9qFw%>
zla+TOO3_Mx6p6HUASZDuJhdDrEAdO~b`Z%z81)P#oceEm)2F2YRBj4ZkCj+x9w8UG
zWE}e8?cRrlw2()UaKiyJfn?<f)Kv)Ng4GdQvaZEeXve6QpkYPA%%qY~!GYyyZh3x8
z4MqHcVMY9sje$@c`O}YfW-&})fbBx8GcWMHwkxle0)H_dC+7`+V6*!}FOr7F9=1yU
zEeniqlftOYKrIwhy?bSI$(#7*`Hjv+7q1447MLlu>uvNBgJIqG@Y~?)|649!4aI*P
ziI*V*y*{V@JZ8l=JI4!;3UBWF$p6v)qjd)sFY~_^W+dNNTzl11|9<_W6ETv|$&w<?
zt2g7nQ}>T}&5b;HXzGY><jBv4he$HLe(d{`E0r^^AOGVo-zI|>rr_zlBEM~y@5;Y@
zQ}EAZpPX99{80jzP>;m48B#@k=aX-L7FP%aE_IDv{k=hn)BB7B_h<LdQTiU{C-0Ac
z_I$lLZg-3|j<@ZB&EVdXodeZ<!9DK^_NgX@{qWxP)#mV*I^pAJ{(4aO&DU@L!GWr6
z;|@-DR0llSxyTuRAU4#m54NYCc<MedcyEAK8%uH<yIjowuB-X3B~Y&QCFlRrKAw7h
zuYLWip9KQjF~hbWubvSWj9l(M{KS&cbD0UhevDZ<VxT$ljMX1}`R(j#3)vU1-=5#(
a)l5(%NVb1+?qCCdv|hf-O8)uR`~L-L;2Wm^

diff --git a/vignettes/functional_api/unnamed-chunk-25-1.png b/vignettes/functional_api/unnamed-chunk-25-1.png
index 2b7ae2386ddc011e7e5b3072ebabf50abaf17121..2d9d87e20c415f5571ae425e1478802eb94b6768 100644
GIT binary patch
literal 400451
zcmeFZ2RPS%`#${X+t$vikO~=v3dxp=kdb6%6ot&}JsJ{1$;c*22npFVNcNr?Wn^UU
z?K$7N@85C!kLNj_aU8#A9sk>X9QRird_M2@>-D;>>paivyxt!2vJ$(<4v~>aq+ODi
zFDj5oTfdM<<ow(I!7Kaa9^fCOE!xr&7fEZx|Dy8aUXVyfNRk)NE82z)ci5;?G|!Ds
zXoPS-Pkwcge8&UHlUEgu^xtQ`y-LoL8l<my(qqto`!iMU!Z*zaOsPz4MJ*m=o*Ely
zTIBa#wKrW_PA#l3j0_LHLrt}J%5IIPJ>TN@POrCmqY;UU@ruM%t$!UUew|1D?|=O4
zyL6E5-~Ukcuifyk@1Om*zpUFAj`y~;I;r~McBvmzU-)Jc$(5tTmp(<S(AiFJ1L^H$
zmY9tsQjDaGjMMVG8SxS+N?KD>(=LugdV5oUh*(A9%k^a0z~Apkr0=hu?*7;J|9Sae
z3-SN<nxsw25?5k7m^agOLsnL{dqTqZ@FfX}ASdB9)zQ&WtD%N{zHxDJT3Pp^ho`X1
zpIpfASmG~4Z-|PD`W_Z4>M0N6oSU2DTykXNHZ?UhG&D^5Tv1UmK0bb|nvPq)=G7&y
zq8{ZGP2={gRI^DO8j`2j`Y{$|ZOgGrba!`e`k8bi^ANYbJjViYEstOE@)Q3hzCNrU
zv-NL}c%Av*ewkeym*x6F$JkgSO;?7QIqvoA*SLw;xHy{AS+TLPp`oFPiH7FplRYKA
z^>uYm4=gw88XEo?9yT&H)rwbO+Ofmy-Me?`>FNId{&8}Vfi?y?Z|kisE#E1|*G=_T
zlkeZ}_u_@ObAVI;TVv9V=FU!G9-f)1aDmp=R`J43SP^NT^(&KC!1|fy&ji{oHznOL
zGczM|`_k5CKR^EK!sDG26BGM1E32#b?AWo?7o=r9+!C!2-}@jmB;=#<4<*6F^z<%0
z-*&l)-MV$_wB<m}(F>2~7Z*cbE_e0ITxYwkudjde=Fi2MKT}gveGfu|gEQ{diwj0P
zd$#xBK~_7djvOn)rgyA|MI2_|$=@_?c^~cN<+a$-)}|)d+Sb<A+^j4Z5fWlQSjUuY
z<X!yvb2OG`X6<^4rqGQWH%fz9Sy`KtH3TevSKzuEyNuaQOiX6}G*kO(<yy0!K7G~Z
zP-Q4@W@>8fp^ER{&$+DT`J`Quky)A^tj~9t!{vK&uKfJ`Fh;_6v?G`I+&PuGMz7^;
z1r`d89Lr#SD}xdry2{GRMus3xEt9tN_|fm!&XgmR*t6#qAv}qBYj-yuDEa(wE0fFj
zAod%_`3m)_BmAatiAnM-%gf8!IyzR%t$fzQDt%=EeSWDK83BSPPMwmEmghEZo5mIM
zKhVmz&oJxxX8w6(WQ3`@E#D!towYoW{e@tx^+?;7u#Q}tKbIxl)AXthG!4H!J?K}T
zte$;r>G$C0=H|=;*>pno>9XPcYSDs%f@C{(6u;=Nj!Y5QDj`wQ-mX63@uw;2g}|kF
zMISnWv&WB%M;d+o`jv@Zahv6ux~iI*$&ZgR%LdyEr#?v^J9bPgQbdA8+BV=#<T?LW
z6FntsD|20b<(Zr51gr%u2cE^p$47r&SzgBHF=Zc88=w7Q8YyU3_v6Qpnx8e3qq$?6
zxz<V2K^!3uGfg`9&3m8NPLBWj>PFe_hs|f^yH)>2#&3#_j&5$Ot5CNxgtyc7LZneB
zuc`BJOKP?7(v`?kHN@WP@>Dez)Jsc&EueVYHWEp`_Nw?+T#{5dHfn9IaBaJ{lI8fs
zMDSVTkn^AO^YT{bd-jwDvZ^I0#+&27s%01q(yKFd-Tqxzfg_C5v5V$x1Fd?hwwtYP
zW5RWZxzQwshEMl57C+21?b^L-7o{(5A(GGH4fbx!9YbyH5BGk3*}Q}5fMuXq^tNr=
z2HLaJqKP7r_vLRy$Hrc8-@Paoaa!l`=uEj<2>0D%qwm<`5MzIuQ<_EQU6P&>M;42}
zgPJ2u*lDr5*lWmV){jj&Iq%*tvC8K^fBqa8(4VMC)hQ|4Y-Vj8xZL_o-4iFbz-Fv-
z%dSJ0AD5Mt<v1=(W?S@Y=Gh+Q`!!W9LeH)LO#QgFObB=F#6|Z;RHmPW9)E+E>ad{g
zz1iVb6uEMldsbE>Z5f6W4@!!Q!-X7X@x7g$ow)^$<KpySjNYd;v4blMQ~jZ5jfa|3
zw35^^uyA!z=O6yA49ztD@liER7gvI>NT*p<aq-9VMF<VUQI<r7c%>9g_K2qVYs{6-
z6CsOpW2U;gZ+KKxRObJr6nZlI((uZD_>DW~H)%h&b*uXunb40$t#yr!GY|I&UzU}n
zs#eDV$<NQfeOw#WHY76giMO|q&1gF!l#7dN<|#ILd?<Ro&iW^xP)@dWaKP^6UQbj)
zGT*p$cSTv5Usx;h#f66w7cZt8|7fbLOu|pS5Am2=TSxGkMyT&|ta~kK|CpVH<rFu$
z%=72ZkFNRUwIW6CWewucU@NtnnO<I)k`rDi>gaIl{r-GucEn+RJla-PUA-N<)ORS&
z@=ue@RJly(S>#~D%BySvLqfX6j6v@7!j3A-{>5cwLy0K``{k-CDu!#KzhZGM%*`7Z
z+}zv>S7vYEGm3l_hKGljS5}y~YH$7cXuRtXx9vzqv)V_a!NI}1hK4^-Q$~Jd(j3B>
z{z(~?{||vST=#Y4+0{2U^D;4AT8Wj4#oGpjhW_|)r@ZI|4UfTX`Ei`u4AU-QWV8do
zEh~t<YYGZf>}re=Ylys@>}(;s$=Bb%<2tt7d3C4ho$8kfEUS>9VES=E$N3s03&tQ8
zk0->}HYeS~BF=*R{5d8aIch!F@<yW(KR>?@R)3ndlONZYq2V*{9d1oSzUJAgp`r03
z!;rZk@U}z{PO@T(riNev4sof(?q9!t-M@dILHudWp~!P)-C=wd(`(N@D{DDX?%f;3
z_evrD3$o4EuYdQI?G_JF0p&f9%Y-mC)NC$XTg~mgt*os4@PW%jF?9%!<!3)8X_Kx3
zb~{gBUti(p(kkL9Pr|~(aSI`5joUsN-_Y8Hs?}ZE{be!XdSX$1f^rH+C-pRP6fX}?
z<Dn~3Qj3U^;(a-mgU5nLa;*OJn)D)Lbf4GL>;L9WQy%`RuO(GSjRye0b?uxI0Rc&|
zOvjJ^9;l7!p2*6|dir$d#Fpsm35vBMw&LR7=v|hNF8JjXmSlE0pKy-9<VBThGsbad
z-X)qhM=W}Lb{1hVmpfMQBkKIdm0b4|6l7bsZcTbl!)rn-WZzI-EobOh@$$5q=?knJ
zGU&DuRc`E{rr^tG&*Tf8SCA8>e=B7eTrV}#(&_>ZkuG?B=ktTji&t=A^J7te!ypi4
zGCK%0y!ei#w6yffL!zwl@+yxuBDGxh$<59lBwr5TjB#*qSevewGj2;~?Hu&unWH?R
z^rU=1BiGsqXYbMMpT<PxpuD%YC8*shmioH8&AW@Xe?6nb(mN~A=ob-z41D^+Lws&t
z+h6qSPnBK8M&>4FKA3twLf9#xr*VGXw$yCm%+MMV#g=e>tArLFOH_Q_<EckjSy%?J
z7HSWVA3q)=?!CW&_S8*R50BZkFQ_Vv^bU%sSPHRH#j*-sX=!P~^~qX=tMo>L__!#V
zbvPE4{v`L+4l*(_sRQaI3YKn;!J5^Z$E0EvMuBg3I33GN&&$jEXwso}$mXhwN|ybM
z!NmOc=L~MT14s%KaxHmwX}3im3Wv)x3PnZ<I?Pr)p$6u+<*8}Fe@HCs3kwV8oB@uX
zQ^LbFPHW0YN!2lDWoKu%OZ)ivEMBy|TNx>$C9!J-6%UxOu3afz^2PC-!p4xZChhY$
zUWOTe;*k_l)OH^hQ~^wzpi048;_Um?B5V}d0(>M6ZP}OdAjm%=!o<Ivs1#CvD+Q&W
zySv0`%|^DE@)^<0j~@fQmYZ&v4nFUYE~~0aMy>SYMEDsaTu+}qt=1DA9gTEc$K?#b
zmr}UO@4RfAG=U;Pr1wv*8@fYm@7;T^zBK}=+~1$VjD8lV4|#GR*zYGY#^VSgmGG={
z7E!n9Z#1S2NjG@;`PC+`PiO`iAQ^^e>FVlA$tEJ@_u66;v8+2Kr}`#wO_Cg!M9h1?
z*F;}nk~6ci;u+YFH>L=x#hKT{ref1Q`pb<u5~%j=J39I+!q2b!&mTj1C1QQI+potG
zW#iL>T#O}u8sh8Y<mgL-zI^$@U06|GUS3wVU(R7^Bs1yVyP}>h=cS}*NvYX;nHd>8
z+&)*o3mrap@26|8Eov72RHugN)vH%N^diYT`UVF4dR4T5<w+AjHCs1t9(-awKYnZC
zr(sHdQ<j;|!w>s8?;?x3uKsJW8xh|G^x_#99?mmtPDVlY=x1hRRL`@$YB;{SveX^i
z)YR1NxsZOh{_!pvU46g1ckgz89%@W%etY$6jXV=`W?CW;ckg_j-IS+U01-r6CA<$Y
zjSlQTaNuQ7P#i-8-i4`znp5+gUW$dO=@=?GYV;7-K`eyd-`G)9l!`S6s(rX+*CI|2
zy`b$)W8)5F{nI+%&hha{R)L1*XX!gkbYGH_lbh(}EL@(f<n3xNaB@&kP{1;5-E+7d
zWlx)NVYroRBFe*~^!~=J&I^4(OLJp}9Lhl9<)x*pXWk?wC9P`69AjVr%5OmF7YPjG
zGHQ9>R~{sSM9uZpZtC~0gJ&*QC+hRx`}qlQSJP!>=9u2l=)fy^mSoL*`=6ig<Nxn^
z7cPzFVxJESSPz3UFv(foyLSMaQbO<YuCu{%W~gzokHCc6*IB!(Czya-GI`FM>1oSS
zH3Q;A-WbRGc<jh}oMYZs<}jAeeAu+^b!iBXVfvPijt+Vud&S!o{8mF@=gh>aH*MK@
z;Jw3~$;<>U^o5@vEiG;OZ<Li4oM4f)<tc!7t}|zFG9T}u*Bq`?L<VqKUFb*p9gMIB
z3vl|lvWyZ;WO<XpP9QkI(Upl3dc;m@YAPzk?%lhqs;Yqbj9Ne3!O^X#m>)ZgQ(`tg
z#)i9y5-Y2$3}|P+Rz)1wX6-TH{-Z}f7Z+cu#_sU>(dqQk@|oWTnp&*pyuNw!X1vj}
zVb-zk?lXo>CnvI^yYw>f<`uzQCEzF}VW_fy$k~n^(=K|l4=JHL;?&2FAAt`89LR`p
z)QMgvgz`s6I&zgwbgra+`t&Kouvx0W|HTU~<2G)g9$J1&{jZOA*~=>J*tM%suhJ@b
z0k?x=6V=~~4+;};DL9}kDJ4a=b7#ZpqrAK}z%2ISKMI_ddP`qe*<3{ghz|ZmO^TOs
zy&JdeB(j!QST^eUudnWjifq~2ab+MC^~|!7!cLx7E9itBZIzS)qxAy(Ksb@SMF{i|
zj9WdxYoeaYnQq=EMVB(#Rk&8{MZG9-17!+u@l{NW*K-f4*gB&n;yhgVi_6&(!otFk
zAG{9>sU=6wzhIUFg5}0ReDL6(j_5bo9uP*Nd?@Q^Yd4s2XKrll!q(sTXms%KVa=x&
z(iapJ6<-{eu^w$#RZ(H#`_-B60MdC-$lmgK7cU)Zh$Z*E=Cl~EH{;{?<m9UNh%BEv
zdGbo+0;gskPU)vd+sO?Y-kvt?6uiJm@PrNd>*A6zuq|>7rEhuCW@p6O<;zcO1vxlm
zNB0A_bZ`E$ef##uuNPd-;zIA%zX8{>8mx;2ElEyJE-fv^QE~zOK}ne5+>~-76I97K
zT)<`&`LQyD2dQxx011J*Y4c{c$B)VCu{2DkX9WaqNQ@)#Wa!tv!kZQp6cCg;CPoMB
zPv%N`&`fV{ufFp^8k!S!><=$HD@OvJ|8l2yo@rDTvY*!P%*zJ%m5L<*khk})-MfWt
zM#D}`P4<<e+^EI<S<ukPHoxP(o35w_3?2VH(|deX&e?etgdQ+@5V*6ar)OkDBT0=1
zSVTY|7h&}D=~HYm;A3%7kveb(P(Hu#19^G*BMbTYW<BDk_oEQoX0Tc<=R&#rbOcN@
zf=f>ps#u<Zk+CsEhoT<bypCyUP*{BYAks*CM@PmtY7X^tAl>GY+g-(?0V?%d-rxEw
zuKi0tixrDkY-?$GQsPxxOO)ZhCx0ane^2B#etv#~U60+|YR}<EQQCh`FX@|}j?UEA
zJt9v^Ah~GFmg29X7+bU{@L$r?C_pBV66g}(M)ma;z*abBH6c^rn7}y!64v-F7U92&
z-OWuPn?Pb=qkLK}D^~cs)?VBuqA`94@oxTZQBhH-ya6^;_}8fe>oA1GVg;HHHt7r1
z!j74U-iNj85W+T|l7iyVqeo9lOxv@r&Xz_U09XMi<#t+}ZfIzjtRil(`24yuWE+3y
z&YgwD#XuV>mNo!=x2=cS0@zNR_|uxEhq}>@j0Lr!jF*?Vj7@((_|yX?=|Jd}uk+II
zs_R58&Hpn}>HnQF?0?=HTS+4QpV7DfrC87K2?)rAol9Lm1Sua-C|<sNIa&1&Y2094
ze3lCL@+wE#DHp&Q9%n;gi~&htVdAy|;#~g{rTTlGh=1Jvo1KYo|MN1j3i$fFCZe00
z()8qF72*}ykRsGOvzMtWgSjLY)^cqpwC$KTM_uXb?{7|2mYbRrMO}idRxZGa3{?7V
zy`90r4kUg4{CP-&Uf$kLv%^r*LSDRhk(dZQSOP2o)Dyg}ak6UP+fNVxktyHu)d64G
zblSz`P2S%JVF8e68r2)>hL?|ztdtZ|+6TafjEp4Jv^RK{so#|jGea|re^UM&JjKA!
z1Gy6Fu8*f@j@^{5n3$M>L4sKHTZLFpFE8nyZ{9c3^-t^998p$ouFGTBn4FrraN&Y4
zPz<2k+qZ9#bzaBD;wq7jL33?wZS$ZXKv*<wdKat^-@ttH{Q1vrJE-x#D7Jv#5rTGp
zAtAUY9Z=_=KWpCK`ijChGuG8rUY?k4D&zkB`}e}t#Xk^C_U+pz7bz?wDS4WQCnGi0
z?B2b%$|(_Xae8s6`<4Upkf`a-L)9z5C*Jz{csqk~vbrj&5Rkg`8bAcg2?CQb1oIdM
z_7?IA3JH~;d-4ALd*s%ERb?oSgIQ)h-QC?~=Xeh&gPb*(>I2c02?VW$L1oroN#%#4
z+kTswnYmQNgG;Zf!OR1o$3rgo%pED#_FbT63Gd&Jqe6ggr@epA((U30RvSIg5dXr5
zS0h$|iJ2K0y|%i#y1hN`yZ-Orzv-Q(uR1z926v>WUA$Nn7wJ6o^Ji2wyMwLmGBg+d
zSAD;K6Krh!=Vxnc>%~n*KR%AcN(Yq<>o$TXFZTF~6ij@hS-d?kFmPb+-e&;;u?#nz
zoeP?{D5<G~Ak&}8WynCa{$n9Q=0+^`Q;UrxMa%UFPVs;6;DHA@^GAcmA?)|1Ds@mz
z941JD8fUXvDLeuHcSPEOJvaqxG(Q16POc(!)V}Z?W!DrH8=&!WOF$?2@lKWNLeD#u
z)NY`cZ{<6f_8&ah3X+1#jr$woS_5?k%WTUsi)nup%%$f;%Qp!L1RCKbaj><enPx2^
zA))FpJ{{3IrY68lC~QnjOoLnt(5pde4k2dFpTFPx2;gOUxYYu1bf|Ck67ao$X678E
znVsLWijK>M%{MA*b>urJN0xeccxX8<#HOT}iitgFr@eNqzVrpNp${V?Bd+q*YY$|&
zx^=xnlP)+o*lB5YWuZUP?Ku>Z>TrR%FK#=~fB-ZHCIicf<i6_{6BA=E28jk*L?EY@
zrY!45%0|df>Gt&4`Pc(yA3l8O75@H!?5K!{3vdF?;u)n+MMXu3O+bZ8l?~E_a0gOp
z`c5Cv3PIkMe)mP{KY@=e`m1))@hg@*x*8*X9Y@vT_C$9vSYm7#HHmcR5dmRrtw0=p
zLq8tU;o?ejb#(>$F)dj59d6SJ;+*mR{ac6y0B1_?RLl^^dxYmjg>d$2y1EKaHox9T
zi8Yx6yNKb#xhB#=_oW}8o8C0MPoGnlN!+^i3vd7Qz$tciG@Z=Y0?MA8_BI)9&ju(J
z-#IWgwl+O@jbRIS&R6$6YuGCD6s2T!JBR}z43d(P0Prj;c?hc=MEU)6LK`sXN2aMV
zJ=!iK?b%aaoCVMReEZYg{XwV12U_1Y2AywD?Z$teS7jtqbUdN-u4YL8^zH9^okFyv
z*;rWSC(Aj#pHn~mR$l%m)U;sx?%mEH;Ls4Pt!G!3<^W|zM@HEBL<9x%zH3Dz=MhTX
za|Q`yJ*cAHJv}xfZH?)7ua%YpRPNfoopsRoz$=5s1oGoo=0`HSa91FJenCOh=pYT$
zM1w>~R#lglwidb+j=n^Effg-+%3!<GK70T<mJK^sTj=6aT8dT#ShKXG>DRqr^iYA+
zk$0hSMT)pY1O)|!g*jq{(Nln8EH&wmJs;e5&T;-;((`~5$_7qGn@O%@p6h!_iOeVy
z&VPM=5^UE$@tQum2iK9=8%wa>$hn+J%dX0oO-x21$0YE~uP%)RG5(HpSv`2@kec7N
zQ)J}iFMvzKMO=i>oH57!AYLtOY%)K5fTpg1C=*^D|5{TcpF~;m0Q*qx2C1i&;P%?u
z&WnT0A|iz#$!gHFa&ijRR+q3!PGk9VjMe*vH5A)Icuk3Z0MWVrOV*LvGSbo@m0LjK
zuZt~5BI!Hif{&Eu(1ivb$f2=_Kn5Y*bKq1I@^4^ZAQllUgb;Dk^`G~u-m$)0GgI<!
zFF-kuVUv1}<y}X|Wz)iyv-Z;ikUD?h#w#k`K|5hrOIOLdckA-y%dt`cMa)B4cihBk
zA?iafG4yH2DK7FL7q2=@a-9?<kZ&Ft@<WiH-z#Agkbg)l(|pqLipi>J;%_O2e`InV
zKmI0Nf%`Fv0j|(ys9_MNH#0r`ol=q-IwjQXYOi&EWO^QBWgYB~baBQq3^l}Gi<P3Z
zu&_X1l();#0?i%poQkTdCJ{x%3A#i}Gk=tk4^NypfsPkMUYFTcJ+zmaxxOGN6Ez2`
z&<Jg(uCA`B=_5!U(bf}jSp`Ymna>lGXNY`5RI+DBFQE2>hlQOyc@nyECxmyb$j9Ly
z(VMnzEwr5wvnXn8%pelsPNLPgU%bXYG&C1_yo<|PAp4CbWIZS!0c_V3l#<#IDST*y
zVsR;Yk{b3?Qu5eM_F`pov8E{U=t>_1>x8T~?RyVL1~LsnJ&~IrM^aHyad2>;I|2sW
z3SDztAn7DB^GIu&M2S-Bn>TNEP@R0{C~58`%gdXIbwmdmO8OFh(7hem9r0f?6rc1a
zSKII+#2)3Y0w<>WYoUU6<~WN*J${)+ckaw!-MezE%DcL>CJQLYiji*AGELYIbEe(U
z{Ad(hSXc;Qsr?anqBO_X=KOu=`wZ8|`8M6e*M#3D($BtEphl<mXFL6aIw8GO2RVa^
z1D&I@535$jzwU9fSXo&qCdX-HyK)&?>prQpQ_zK^DuZyqqN@ul#i6JP$0EUk7o9v=
z(b1tXc~U|`qQgc5yFF9EdhN}?s>zkGb617xOG_ouw_2E5>`jqpA)2XLVn$HN#2%aT
zW?2r_5k1aqwXWEgFAroN-iPCkq~PVlZEkMfl%h2@I2g}BE-hV-(9Ma866JWZuZ!n1
z)HcK=Pqsh0z>xL=A|jIISYq~?a-Fjs|9Pw$e_A);eSHa?gp4XIOSo;wD1A?Rdpr7p
z`s^N*NaP1jYZoE;T0PpI@@-~(Ts~e=RaF&f^to4iq08F4cPBf#yumKfn;EF8>Fh+x
zC`3+UopiYI4Ny|%3S*NF9iRE@mea)<%5Gx6Mn;T_(xu-*G3wq)97XM|4a=dIuIcK&
z*;miu!DsRNNxfl5PR3|jy0wh_8e@Nde~=8p_nTHqr1tLHcl9jm29oB%cSMVnvJ<aA
za{D$rI}5zSH85QpBeA1I!iV;>wzjzC!sH#~o0b-pQe|>-a`&AFCvlTOW=i6zwiDer
zjYFEYp)X&a!|^;u!_<}U&|Ff|Ul%Lwm!hku2XRqTKZ^4s(h=VCk7j##c=!sK`cB$&
z@{uNmOfJonlaoLs!ceT~X!cN1P(W5ehk6cuMn7_NaRS!Z7PdS_Gi|JJt?AJ8%nZ9G
zcap&{nnkgV<{k3*OQYuG4v1<%dzRG|m6aVR8>q_<9zNWab5TuA4LO<V&ys<|?C`;J
z_o96M$VLceL+kCY3a@vHhO9DHuvCT5yYs_(68cehq|3Ey*Rq*a7VlYDknbtW=;xsJ
zi@a{onP-=}QL6KULcG6GtO5sAo8pI;I?Z#fNBpH?e_zJ#tEjv@&9#vfB`E)QMdPo<
zJ<_44r$_VUp{yx3;ysTx^eS!bSEpZaJ%;jy`hJ9gAqL@(VxJ2^MM<d%D(W6O{)T)9
zn_F#wkZRuqCIi&3T&ZZ*a!C?SR*d&6EL?-u`Qq8LXQ)<zfmC#K`6$S~BC9Hq25L|%
zfRIr=HPqFqIMfeNQU)MRg@i0ti>yr#HA=rcd$*3ULuNa26(CYvcE6^!b~gwD7y<{k
zXbS`!q%X^U9e*1FK|p7_N-x0A@4Pb842=|646P49DJX)d8488=zrMKn3NK!cRT!*F
z=mjP%8qxPuS7Gl5mTg)-gR+L~TwHM_+7?U)Sg$%lC~=Jy+S@=!Zsz=F*D#<~h~m_$
z<bn|6078p;u+1^P!fM&(v8x9m@}TQgpnfRtvfqzAAMq`+T=DJ`tvV%Nzs9s|@HRx=
z#P&BFvb3<kC*4TZk?J(hx1V8DUJS}dlWQ3PwBA}G8jVh`($c1AO!K07Y~kF}Ilh%7
zn$6PRI8L-u6*Ofugco|rf2pPER<en2s=VayzK6cazL7;nqX7~p^A#1u^XR+u#ZOte
zxj5j;g6v0*h`zj*X`)eD#mr<nbnL)cqnGw#SQbn7x{~&6i^ke3SFQkb9OZ?Tq_(R|
zgpu*$vR{);JaTq*q)7Agk7Cg(83qBp*9p}KTYzT6_gJ%~S>tG@XGuxM1_lPt7yNrQ
z>Wx_1Q8=+dp8dUjecrTu3NJr`$xYB!ny^92Gq;J0gVf8i>H=)F$bI;^(UAFArw|Pd
zP14v)T?vYn&>RXXDyg%fP3{Q1-XDs7R-aLw_iiN?$BQ}%Vfi>u`diTR=Vn+!lm%%H
z@nof>kkQ$5IbfU8vwOR8ZBq7`ewK$mj=cwy2x?botf)A7?pzk*1XFI&l?cIBl?Cbx
znwp(BKVX8m+~W)53f0HYb_?ks>QviG6A3X)-d@vlYX>+R(*(K`zf|(J$<h-E_TI9&
zN%LKJCjd1n1vhIsPPo-{3i0#Xk9FqX303nz>tDv{XHQQ|%;_f&AAUt*L0dUaAO!n~
z?gnFd2IOh-CckngFPIe4K*aU*^s?tcV1K0P`Q~Vyx3|yfHP?}0vTKI)2DD8xV+P>!
z@WBIATs9vaUEPnmm4|Mfo(EgF9-DV$1t3L@fw(X?w@VYRP`N8n?<^S4o^6UMZhHVk
z7|v%QF4*}_H4V8M0x2v4=)w9P;*k<$=Ha<M${NK8pB1t`y62|^1eQ>Q&|KYD%?C;L
zcgk@$v3=CkaSQVofA;ldfHYuhQGPp70DvPYTwd$0t*&e_)-<5#a-8z6+5P?7w{Ks+
z>HxrW8}El-WdHvC$1=CN3R7`vh6M(~Ly@SI6d~k5C4eJ=Zb{Z1q#GNXIb^c2whU>*
zki^740G+CSu>;IFj?n6xh_=xDM2{C;nl1Rg+l{km3|LticNMIljSl|99S1Re8K)9i
zsh;_bt-A+A>jUgMBvRRSLL`#^b@$_~+Wc#=v}hgwnup9U{=KwCc>d=@P|DK$oSUEj
ziQeOCn>t>8Qb<U{=}1GCv^I9eFGCkyUsF@ghCB|f0`*~O_?U=Zx0Cv9<`us@L+H$u
z{Ht3xZE9(4EiFFH#f8?=SVPIxt5+eK<fO43t&~%I+lW<X+<*jVkt9LAVZ#O>&d1*X
z%L;63P;-F#piN}rD?Qz)J!=9v&doIF4b)nXT<gncczAY^kuAa`gwE25>j@2bFO*Gb
zbBTH{YEA=ZRt_093Qu(RA_QzqYNF0#=XrQ}Q?4gA1!*~FyR11Ujn<L5bwhNoVb1^1
zfVLcHtU3&Wfbqq<iG8s%L|>M&%{X5WZG}|rVo#up%BCxip|#MQ)_yea(fnHSg{7&P
zI>$+NcHG1itURT<!hZPeXr4)^IzS~laa%~D!Rx9Wn_%Ugr^Sii3?c$TLJS^-J(Wpx
zOQ*82v56M>1okbJC8?vj|7<JNJO`w-?J-W{pSSXBQmV}{GBE|bu1|kIt&(_|P07z@
z1igo$jS%&5D*hDIz<F>YoxZl6O;jeWTn60B(w)t3-k9yy6#)T(XpsUn1qHv}1Y<g|
zOvXr`Wh_{CKnKx%tqGHmS5dJAk#kDsF=${(l9Pqo&VXm?YQ2H86S&|+7%YvRE`~q?
zl6iFtICyq;{I3&%Elg@(U2<8TkVs{C<mQIhJi*BcmqVF9D>ys5s&y=(I1oa1uC+1j
zgcHY#N=lX%7KCiaMGf=N!1MEcP^JOu2=46y_Op_iL%Al)a`L4A@a8=v(ar15N=4g{
zOK;J$(7w30=^{K8bmh+FUJCI>jR^sxe8R%QLk~e9AZpdgm7UA69!U_sEGaqIlG+2E
z0&(C)T?WXgPSe;sG{i<5^qQ8n2<-O0&u%s*$L2={w$RGFjG%cbFx}#@w6shgEoEXf
zDYymq<(HQrr8V|rVC@qFwG1l^M6*(Gps$a~=yL)HkrA4u-DYMlc1ea*4s7;Ia^QBU
zfZ}66^;;%sJ65W;{pI-=<7r1(SXlOoFtFNy%8j6X2Jjad9&S94{rU4}XJ_Z`$hWxA
zRX8=orE2EUUqh#M8oHpweBP5)kU|28uTpA&?}1*^<cL@YOUlY>7C7303C}n-FfIt3
zpBhsok=TwAeU7Lv^}}u&Pc`?d>GARNA3Dk$gJh`oyx3{ao;|<Lad0I2bDasn=h-_r
zB>JNAcBE*DKqExHO;*bof|A_LlfVx}a3CT|wRPWI(_R#r$B!Oei4^Xru1>*Q0Z0;t
zfn8q$Kf+kVPj-P=#Vrhkp0X4)Sa^l|ZQ6b|``C_v%X!-}TfA#|7`BC0Cfw1}qiJ}j
z{^2)=a&(}BWaO88{Bdu#pTKPc>r`02DCM4bcw8;p%im*|gGTW4gMma0?mR5m5_RpY
zX=gifX+_1AqzNQM(4%@ZSZrGZJfS%PTpXGf9YyWHL0C6TnC4A<!<plWoXQW|Pfm{P
zLx-MX@0HSgJFDfU&1m+n>q>~h#rQ0<G+}OPN-9-vc85i7$7Z9k2!DUYkii2lATxqX
zqWjJA6hTLr=U^fOVfv7niR!fkP%?8j;e=||bB0XMWiV`_sh5&6hD3`inuWRuGX|ZE
zdC`u$n^B+Y>g(MMGNfYJ=9PaHDjrJLrW)ILaFN_4(z|Zb_WQG=Y9~rEjapx=^|M8Y
zxNsb9n9x%-XHa$eNUdaZirSZY;pAr)2Wsg~-N9fO+7|V$tP~9Z@8i~1kj<G2Hv)sz
z7v`3uAW1wuaCCK`)OEu~s2$fPw{jK#?(b(_E-kC5aGV~f1*R89jxF`}LXSIc?P~=3
zPK%3+1|$55@+{~KqN^Uc-{h41e(doV_Ws^p2y$TvawuLQdloi9aYMMAIP<1gf*jTH
z=`#*A@!Sd3<n4H;T|eEgd0fKdh=|Zohm}REnN%prPiOu1nx18O6O;R0x~jOC@LO%&
zO*_z-7^LKJmWyi&_4v?PBNaGL02-wPULenalQ%{Mp2G=&MW<JTOaEX&#A)wK?caZ}
zxH&+hZ#+P7w>9=%<xkbOhJH{LO6p0KiHA4z?Af!>L(pIw0|6JiaDhqWZYpy4u?z8v
z380Xd3(|n7JF+eG_?l`mR8&<V1H|AJ@aHF<o(Ax9+g1VNOt*SzgIYk2-&^PeNHq`b
zp@_yjkQ(_I2m}rfpgKZwMSHWG58`$9y<ZPT&voxP$nrGZm;2vs&9~z?G|aT_j~^e(
znfigW{-`e<B;GG^_l%;HgTqjLym(rs>%)f^Qd6#L{&<1yvH(Cp-D8%E*a?SeYg*iI
zsS)rQBvnHo?ARbt&;`_Yw;yAW?bcT33ZaSv&c2C@>+kE^zT+%f2Z?f@P;n8XVyd$M
z>q-Z;=SDlWw150n#8ilWd(6erwv5F0{RwS==V%Db@HN*o_W70PFg-lrbyfKHq@N%=
zoY$>TYKnT0t{R=Rp>v@yW@#6bO-KVo14w&&lu9qIT|?HmQy%D10#XINJbI}e^2%;_
zOO~U(fvo98obwo6di>ceET$mBNjDXEImQ%+K4%M%y0x|i+3Ic3bR;V<<6<JLUvTfd
z5k}9WEz>mee4J^^`}5~Up~u@k_&s#Fqx=!L>8_mg2;^3@AQBv^a7G^WKr_6X&(<#X
zygW>LkZSt&-Mg*GYn%o1-Oyoy#6{q+CfWm>*nCLnR5~;lgpxFJmGnDu(T9gZg4#O4
ze&eGhw7Jw&z!GxZp;oRq#<38E_;+wa*;bW{<^{bu?{I(UBalzhbNudr#jD$LPXP+c
zD+?_L{UXjky>!MFsc)yf_%YQ|c4`22iaiC;hD!Rmpf928g9bt2U2oXpA$sXzkH1yc
z3^1CiMo-=KmWmAll_!?ET15Qmfe9S5Ku3p8g*!Sr*AkW3Cv2b7c@`CtLwhoj?_wvB
zG{x2h+{By6L|+h)UNc$>PIIFgx4hE;{2;s)TX6rN9_TEbz`?WEY(BECrCgb#z4ag?
zBLk$eDe((CMfTcl)()}LFGE6L!I*#<5=yC}FRC)(cJ}A6ne3IEIqyZqPV^FV-b8f0
z+FxJ@oOPX^07U6FyLi6ij*@r?E)tV2q0B-7YC+f{cqA<S@9?0>?YWhKfdSQu?J!y>
zT`;X;m16B@hPrj@j~+b=)qrEN9ia||Xv?Ne{;~$eN3%EOf-0Mrm%^z4MbrhDC&}!@
zi9}R@r*r~M^Q?zL+Y>CZHTI&K<Gd4A(+O!Ur<q2mdpz`2`C(_cxZLu6cqp?czT#>!
z#M7;XQ3=sWK{rWrGa{Ultr#6I^#aH22xb_`G*x~scRGw=`axMxBF)QfbpH?ll$DmO
z#jV`)BiBHSmFZH1Ax&XXhkB6fy%M>;un$y;u(3(0uhtxE7YojCIS9e#-n;CB@65Jq
zY%AJxlpCyyP~Yvpoh;35$}m)(K|_$xmR|K!mWu*%XNVJMWI46Aiy-w+Cttx<cKgo|
z&ZcLsU}d;~avpW6hleEm!aMVm&)d%q$Mg7wh6+G>r=zQe*bN=M0j)0#21uT*F0HOs
z_RL%O9a>y|4^|8iVU`b5Gw{#^om4$`I7mJFojePq3@1m&BKqC;EG-4$C4tXpXJr~}
z%-y*hovVpph=l%+Oi;WOZeTyr?FOCCRvpt?k#2SV&!5*ODyJaQ47?!$z0=HJL>f57
zU5&PexMO+o=+6f1;M&TVOU{iE2vBknf~w{|?KJFrKho@b;$lj$iFLUo<sZOPP>CH{
z&hYdSNb$LzOz@dLbex8IAdW5r`1sWmet9Em*6b(O*@?!%?+%FaR6e`ZwDT<bW$EbH
z3ZU#qzZC*Bo!E4%s{uXQi{j!#Pp;iQ{)HX&c`#cc>k;?d(8>J6+mw`)h91I(Gw|}P
z!>rNDQYXmv5}`r5(<9N>Oe6=SW_H?3w_jKWoB!sig`*1nin^x=sY6WA_IE`v`em!w
zNQ{x7-obW4l!Cb~7tRPi+wq&F_HXyCcLz?B!~J1vXNL+gP!@0k4Ivn<;2=W#6TT?a
zw3l#e<1QJF9h2fYeI61J<Wr0`tipsjzqB;glFGowHiVTyx9Kkh2OflGco(qt_Mf%6
zP6s(bTMQl0(?>$9H}Ec;Z8HQ6nw^+n4G^fk0kIoh70|D1!1*{(wxGV~Ucn{=WwErp
zT>A3ma|Dp=a_S!(WNqE{7L~DEWH%BcKDj6Qf_wTmc6N3&V|Z?Tc~qBkF#poUi&tKr
zg6S<5xB_%X$gH~v?pGM^z^>6~L7n@6FojoJ-&&Ad<;D%zi&Q1XQ6)UOOp=WE00TVR
zY48o2W7JoI%&q1B8XQzTyM5;&?#o$kv$GdK!+C5TH~2q$b_!>@k`Y<MhVHrBZW^iB
zNl1WHa!;N<Z5VROgDzYu|3&s8x(Q8A0%AArWuO7mQRt#2Xbi6LB@9{hcW>{Lep&(R
zD=+J#FKnF%gRc}d*-*0zNdHVC>dg)d4vn1GOH`f^+~$zDZ|mum%X}{_otT;;PqtkL
z=wEE&p*QEGWayOaK3audTe>D4Ov>X_gscC|7XX<I=Z`u%I(?I@3%MU5;@$GWhC=(Z
zY$pf{5C&K%v$Ledp08P1=jhH)bK`U*jeP|r^AeOcKT!<XccKD6f@Cvs`?EjmwIV%A
zX!*Iv4fXV>$;lafSVw;k|N3>*&ISSvx%;kd+iF6T{C5BJvtiHvaUSU2%TZvuaEA~M
zhmhmEJZ+lG+G?>+{Ii8I1N=cT7c8EhX64ugR!MsRK=KlGIKd>fBS2DUr!aj%Ro#v(
zblOBUR>6aptSJ)FJ^HiTy%M2SRawcdneezyqq~{nf7FXpd{D!Pc?YMkprARR!9wI1
z5lC66?qI^^78a=Njwli?E-ria_@Xkz>xI00dGf}`D{o&(g$UYxSQL#_m>I0sL$O0M
z5T7I?BeQGIp2*WW;=m7wh3pwQI6b!q6rk>G+O!G!c;R<r5gJU(LEuE#awEo%9Qlc)
zi~EK$38{JzW?a2Ur#BIA6xrb5OKIg5$0BcyuKtSLiCRZZOB;qSTrYq)(BHp|@91->
zUB5}UaDu`U0gkaG=mgi9+C=5#h?quvLP5d%@S3#S_LWiI8vFw`OW3=s1jY!#MJW4P
zoNV`K-M2IFEX>@OiXCcA>j(V8_yW;VeDQ*CnZb~QgOJcH@MrbvKL2G6LJg?lT0}3K
zK%w<<)qv?h+Cjm=@tSN73zL0VUB~HK2}lsQ0+rU2YSh%?Kq+d&ksDHhXpmVNYtsCs
z`T%7Nnv;zOJkZ9$6v+<=oVNx%ipM)_y6y^(INV_gC`13exT7OKkDT=y;xtIzb&bpr
z;_h>yMYn3(?e=q{#wI4329g^{@=9fyo`_FProMtha5(h!^&L#V5$^5nJ!&=Z;8-~m
z9lvEwJG;CIPtIsn8_X2@bq$i!0ApT_&SaChX=6Z&e^LY@*X<;wn<gfQDJU3?KCopH
zZv|JQm;~*_KgM*^S`-WLrbs7S@iZA&21v!q`ThpH(NAt^Og+@WGNuL@5xRFmf9B<*
zX-aJaiM414)6&spaL^q(l(2MiML8x>F`=PB(JW=>{oWj-3<#2cRu>w&2L{lEIiIaB
zuH~J)I!dJ8wCh)&qa7f!GF?w(pP2leiM3F&PG+;*7zt4KPF7?i4(*pOVx=!rh76{0
z<RIn^_^89+h&E3WIdpi%JVCGfn3r;``)UB?Yg`KW?*@)1V7jN#*iG2XfsP>J2vJib
zU0x`uK*HPtw|VllL`+g$|KO2kf;uXMfi|VC1NQ;qZ{CcO-w@P0GD3&dNp?=Sj!%M+
zid4GDn1<AQEb}E-@`jCDAd&>ksx@QyLqB7&_V%4S$leKPRLHKmqGJLb1FidPz3Rgq
zG?5AQ81resL70|HX@RgWyg0y+^jUk%nk6227$Ty~Y1wxWv5=CWJaZ=Z-AzR&3T;FL
zxaabw$7WFbU?A*8P@vy?Nfz-C0Y?T(IzEGS`XQTNeEeMxg@(}Z@D2!U=%{NwzSMgB
z_?-!t8$j(KP8#|4R-oHkgm$p*x-%<|xWbs;8ZhG*rA~9$hS`mp-zh=Zsr;r+A}Ol>
z^{ly0EX>ctdGQ<R@qe&HL9Si5L>a;`0+|M5jf68~h#q#*Nt9O+=jD0CQbWK$U=+~8
zQ1?LN6X(V;q*=qPi0NE(F2RzCmO7RdgRqpX&6lF()8UOlwjhFJc4VCzp(X-z;8ycc
zL>VM}8sdsTf{}8k(A|Ow7Vjg#$JdJPEgG!!j^m$^1&oc2XCyVF<r`A8)^OpF4-N0#
zy9c`u`ZnmUUoRdSGy@SUUxPd`3RMxx0*0-i_R-POqD>5(1s6;&3f|M7k0G_fn7!Ql
zLJsMK02FA~gP|ejlW^PUVs~N*F+{b?TXoV#CR%<PBZ81(0BwoJAiPs6Pz-@r-zu_U
zWDW_A_io*5g1P2coiOjq-6s<vXon6-h(pDeEn7HL-hZm7&|fQ@$LGQM;tXJZL5`c3
z7kK_Rwsp^8!2?L#C_Hc*VKyWvh+we8!|mX{06MdS^(B~QRcSqxfKQ+kn7BlZnY+Kw
z1+a$Zh9OUROzW<<x3ta8KcG<rd7hZIz~13*2tjb}Yc}q{YwNDXO%LDfV-N?WIo_>>
ztR@Bf39`SWWNB5^N306WVAj?c^(lZE3~m^(J+0YFesn-%UcE}+g5-a4?x}nx>bo%N
zT~;4p&v9nv#4&y$p$=4d=;fWA1?b+x<ExLHM9^Kuc<rZ4r=HGxC&Q187rf|3uzN!{
z3atJFE@qlYNlC$orx2eG0t6!+b^vD&ShFrhV$X_A7aEE$n!ZE2Eqm7vwJ~0SYD`2B
zoK*X-;n~Nwy}tgM`i$svfboMA6tCr34)GXRV(BMkx7sg^<r$)eHMB|JgCazzohgp1
zp=j|=qiP!Z<U!^BV*O9bYaAd1j-e0y9O9QQNm}o+=jrY5#{rCe^M;H8*aw>ped$EW
zyCP6Vw3P|5XrF__64rwv`*Ft;j5T0}^_G*a_{EEahdv!+>v>XtE!CfHB5Bnb#H3;{
zbW-ZKh_s=A1Px7S_Y7}}Xpw*lCOQ@=lcP}q;ghdCgc-G$r*3^ISGoY+M)(5W5kES}
zrR#rZ7ahq};qP-7Tt5T6&clOl-~)t<L7CLN$%Q=-QTdz}(|>;lFoJ)2X0j~KNIxlq
zNC-HzTAHQ)yOM4sBOT%A<tGmE0Spt~AzHMu2<M$$-$s(_zovHOX(8~#qZ7h_E(v?e
z?bU}>VSFzF0+<|IG1Au|&{Xgb4h!@fH%^QO6TWYs1-$#2MVyT3uCiAZo3?B*7`c7Q
zv@<X0lpN;oz1MBzuHE-+YzP+n`#bdM(g)LLVpKIUK743<f9s(_{K*o>vQJ2r=;uD-
zBu2MuWNRs}!Y~F8wi_qf{*A;BD<l*^+Tgw7mVl=Xq-kKXY%8k7x|@Y0dfI8O!zN*^
zOJFSTTBqcSvVLGKVYL9e4bVAr=fYH=!}I_#?rwhNjS<!keZ6>J!-lsL^>QKsrow*k
z7eKm+S#bCX|Gt_7Gcm{<9AHDZ%Hb<_b9WE2A-iLY<|}%fC&x~fDMGv<eCNcVvC8@k
zHt~zDg6o5dd9R7lvrg7eH}eef>Zio@`CsA}{^x!&h3i^Lr&Ec<b*ANw5u(p@2QKk*
zusp~jd&}a<&FV?U7R3^nk+OLykt6wM@}j3xapU#Wn;SH)?n~W}d~6$#=jE$|QhV6z
z#Dcr%7xJe{t17~;i1~Re3c7sG?vHR7Wem=q%I}Z3aOCvghc|56zdq<j68-OZA!=ID
z27{g>Kj5x~{Yar>2kB>~5gu3)Kl%@;P*vNK6!jhb`w3_lGFF_}=cprB{`F*%`$Z7{
z@wNBz`j9Az^uK<g?pntNyk86o7+gtGR-CR~4=~l}jpqQ^lnTwYU&}V{Q=Hj4&(7ET
z44=K3<b0>@%ddM0OL*IHt-}t(sW(jXd2C9FqTQSlnXrnN9ds37g2dN0M#39`!zqfZ
zB)<Oh!hhD||Jrfzr`qj)tpls}bM8*DCneaJto$u7T^L=mr3pPHBI2-OA$o=o3!I~8
zKxoXj=k3Ip4a+guGs?bye~CuLWbQll>@%O6$~kg}T?O#NiC|(}=kk5RkD_!9Dh=Tz
zyp;h7`s#Y@y#9pqjIyrC(j_5#?$;w|!*sR+2&AA|M3g8Qp7`?f^$S@x|2Jf2H}QJW
zdcenDWtp&nLl0&{o1Y{2y}d*}cy@?@0oUGCY&cs7clCwce6QsfF=J%r0q3o217QM&
zy0`oNKCT?3{stmsMayV!Zqvjw5z?xKbKmp-*Dw3?Ee)rgP<$)@vx9J2z-=iuj644O
z>8vlH&H<zRGiD-=@bkQX0A9<>vHwW?+IrygMv|!C`a84Dpq&TlWOK_zm_-$<R)J&m
z99H+$6XIeN`3IsNf+K3U(p3)pLlT|XOXM+DC0EfrH2YzRWc94V)lsb9mNxc+09Xaq
zgW%ux_V)W`@J8n3fByuwR*;n6hSckKW%OS;W4RL&tuk}iJu$xkfQ>9b6u8}wjqE3p
zdQTl9t~eh6^ssfyElH(<pAUB2#&S^55FaRi43L<!G(^{O11aj<`UbM>gYBsWt$D#j
z9@9=o48<4M(4)}Z;$HuKZqv@?ff(Pav^I3UJfpd=e@BUgB;_fnXDBHxje<Sp8Hbmj
zeSM9dm8fJYC=}#l=6OH!FWZ@~gq}@0my2iLh!eNObP3`MtSCWI4~j2sAh9?TDJ^P~
z<X(U5UJg1x#jTj_-L}4|`(yEMJjKC`7ffm4*@fM#zmXSRi49>nZ_xD4ZYVzPTgPo;
zn-s)|)a-xKtY;EttO%#s*5WR_&y%(HjG{%LAb_u-Lz*N~Ap?{Tye0B<rA7<<?3kYA
zy{_>O$vF~uq!zSEaqdBb^bYs0@uM5%Ux9*a12V*q5(8(VW;BAf<9}<1LN#%#`1;?e
z3IE4c<A2WIU=eD>4wCj0M`tvvhfo6u_mfNr_@M}_eV8Lgr=knz7iMkfy1aT-VPLDo
z_-SMK3MHY#z(9!LgVh~=`_QJhSJ4%TfkFpkDC~C&gK?3FGae|kXvvO33vR?S$)Z$Z
z-M;=(oZG-stCp^xibu{U&SBeq!EYTpteRH_<y;aVN}ZV@KqriYD|nYqOub*78_A5u
zB_V8~l_{fHh=)4_RdAp&;D8VWkqAQ))$2p+06Bqc^bMQ@W5B)mMQ|*_*-YGmNjnju
z(D|w&gqH9?6AU1gbz$K0(oIaZz<Vrd#mTba3u!yzPq5Kcku;ur4S5SR#}veI^Zalp
zQXZ&J<Pg`qRo38}P%`WL=@1dPs?IS~&z5-8D8HWto)v(sEX|SK8})EE+vDa9q-|#)
z?>uk{nWE7#L71_q?iN`dN0UJpMjf;>I6sYoZy911BYMuW9Ralf^H79E72l->XXM{p
z%fU$uB&?wgPOV{ibsrsU{ES3kMs2!_F&xayU=Hf}kk2#s>FMcaO@|~b!6kctaP_>&
zvJ+suN$})9KfI0Px`{aG@(Q$ez0ZkxAMnd-9sB$g3LF<C4%@vxu-o!?upYQ4AsH0k
zI|+`s{wWvF!5sw$&B&&Mi)bNr+4gv2o+1m09$XW;vk^EuoRkA6lCy|ma5M(ksybJX
z%YfsXx-8v5>yRitkbKZV$TVy=K}#9wetTQryR@nVJRsxr^vHJxUs_166SO->uKMv2
zVRS5mCr^`+ihF7$=w-$|1R<Mz-7RqPGqFy&y?)ZzO6uM2kd#KX|IDVSRqc<#W!Qnw
zh4L6uDJDU&MJC_1SUWxPgIB*Kib*`uw%a3&3>78;4iNg)ad((3Iw8)Z2`J@w>#T0Y
z%P%gJB-fjP2bkQ6n)u8f<u>QnW10sa3&y|ffT*{K`Riv<Zg9|fA2{9pQpYD7KlP&x
zj);ip#CgJ47OdN5Xj!5Re6Ok^4jc>{XlATJvw&!mxmWKcD;~0f<sb~hkhCF$5j96W
zdlFr2JPqQexVZQ&Iwmdf!o#*d?_by>G>y)xR-Ubyd`3#a;!w_?<Q;AH1JUkOSsQV-
zyTIV^2_o1W43OZ!=7B{*ej!WfLQ@MWP~j(c7<ye5!h1F{h2y5=ckbVLX|3CfQwk|Q
z%RNy(`il&2M>?Edf+?Gb=PWQs*uKJHw)?n%{U|byMpQ@358&8EVlcYqYE!bt54>~B
z*--AgAE48>EW*59nI_WZvgS1G0||ihQ#mD>8{R!1f0-37cS^`UBBr({0R){|%W>bd
z8V<|58W(z!>k;vo7hm-e3YSxH05D-EGWrHNU(<1-7%KmXqOTdy!0>bsqUVD;HT4cp
zu_}f<JOTHYz+Ah<4#%Q$&O*Y0B|-0GDengTh|tF1{`O+i!s8F2q*L8k`?7=c<aM@M
z8=~3uEqo&pxN^zmwZvv%P(3U3Lt5I7Cf#S)2OMiojU4XLCX~xa1fHA2MI=U|_KSMr
zv2DaKzZq{nJ^{hXS)>zgKNL^s1N$UspQq%b!N-lFRk+;8x%pi8b!816$NBLycgh%(
zm5{3d#R@RSa<$Lw;TSkHv?SK^hIm%N6Lw;h1Su#)*=6JPxAI3#?0&7fnj0Ytz;bo|
z#zPYVKV2ju_L(+zM!V(>5!-wBG+^JM^@H9Q+|0$=uWWw#atQ}T?t?DQH=>^6`aJRn
z%q`!Pd}#dpFU!e^02yeP6Cuo~=k0qPYZ=9)8FwXE;r&_irJoP?K)xflsgI9~2(dkd
zuCiNs21@zsmgf?O4`vF|?v_JJc<<vE7<d+5KLj!TWd`&SwQD{th?#xfOpQmeaM+Z>
z520n(7tj&{25C~W;);JaV<PWk-Y2e{M`F=l962nb`qN0roGS05K2;uvY?zhRcktjr
zr4+yqSpt4sPL1L#BOZG3k8gxeBOY5B)mn-(O+3nn_-!PA%$5w}=|dRZ=x#a*u=n|m
zHSx5|&D84hIQbA9h&VUX6Wrq_27%ZFRY~y|Iv6^Jy0>pnBQKjUm~BlU5-BY9WqK)V
zg%Hw2#y@Q|<6RjsbvdCDT6($W7b+k{huH?wlVd<vt`s{C9o2kt(4K`Nr!VkEci<Y@
ztdN(Tkt*StdmbK+hY5Ux2AhJ+fL!Z+W3B8&iXeh=$_n-tJ<Mm>H@v-yq|kz1GIBN4
zlYQJelEC~iQdB2Z_Sz^qw`U2WZsJZ;`(Sl<JV#rloN1w$q&5Q;70;|O7=DBO_C3h;
zhGcvk)zF!&P4RU#fZX0o1lVF#LWRoeL(O%wdIF)H4kXr$SHrBB5})D4){g3TCrINX
z^!b)Wm{~?(PMUP@#H&D&2#mm7!sgqm7wLTi&8fvyH?w}7L1@kJfxRN@k)8d33JSfG
zv8(?H(m1-ON6)rmdApzz;hBh<u4i9R+bGP_;+Zig>n?00JqsmZp~6X+F-`28w>mM`
zKyu3RQgDW-1atJIb`2OApzBv=h4Q1@vK&qH-Llf$t3!D?O;)mWrigZXiS8n>Aqu!J
z+f#vyPIyjUY9`qvVGMxP;Ec6jusjPC@dV^>@eUdmOAww*Z8*1*1Z@%ZET)^LBXa}e
zQBrX~H(hg~CvfYX0gucr8C1<^?;pq%)PHN$Zq@|f+W|cA#K=xjUlY585HD6b@u~Y2
zGRiceDWgQU$cVs5hKx<@bl2IbL<I2pP?-5O8zi=pj;v!{Rw|0`n7R=e^>5!SkG@hC
zQPAbyE7vqdHFbxGxHz{@BFlec-*e~z9WR;4Y>STXx$Y<5T{yFTK4>{V2mKIorI>=#
ziU|58Vcd5W+80WYpWD!k!NU=H6F)!LuI6x-f-FfhKj#2=faq?5J-k+F=aUnQmdAt$
zFc?|i-t&nsj~{NJP3L)=k@bK|ssoeWlir9s96^_<O5W5y0y^||ndw>D{c@w89qWo*
zzK_QG6dt)!B2)rjf4JjhY4M9_cd_W*bo`r)**;=H_pUE!8M;91`q$dkvq%qqdrRgi
z+*!Xj^fqkTCmk>S;OpLfxe!ItV#|B9E+#=yeA;rL^#^I%vo2gvN1i00PWvS-W<X1@
zC<JkDT={0qe^?`kzIEV%b|1lbDdl+oML3T1=~jhhLwL^$#U<8b+c%07Av5;`GMfnM
z8_FAAGR7n9QaASgt9Ml6vPar?9-oWZ!Z^ksg}?<cLQ~nWH^Bsv2#T?g<8sfnivQ0$
z^j^Moa08wF)BXr_=hNTpiegF+gwig9Lr0iQG;*dl9M;BDbj-@-`1<@eIczrGWsvk1
zmpRh4wrXRvaEV&OKQOQfuJ;xhNe+&Xh=^6x8khiBMw>A09gZA!{_BvJ7%)Tx5>iwW
zDwlZb%qEiSGa@>q9xz8b!d#SZ#v5*4#)^4OBR@$z=`%JZcQglWhqDRGV9H(s&O{yF
zSyqOk>kQ{%amRGz+A?!+6@?pId@#D?0M~*#y`990Is*LIeN>GWbAuaV?n60;R|f?U
z5-<a7uT=d5V+dZP&Eo@CT@k8;a9`X6VkSo9U6$?WsjN0!liO^ZzdTuvga4Hv0{R5p
zMx}ZQkJ2?cA$0pP-X;Ov!O^ALPKQnoo_jg+_Dfi8t65MhQOwX}*gyUHGLFj#>a-Wz
zJB8RX+z8E$g<@*0ojH8_uHr4w&W<qqb}%)RFawq{1re!o^Qf^iF+_;_!IUB2;OiF;
zhT84=!|fs{P>rOCEXVW(>K*ZrlugDXO-Vl>AERkb9A$!ABiD0WdvgUH24DW6x6+*X
zmY6s|L9hogAYO^0B9y(XK6g6n7T7kDvww%+fCR!WiGaq;%_876o;yI9xeC}684wWA
zm3HDHX5D<*(TcdGD0z<78K1PDw-K5pGSFlZx!lu^X&{g=yWhcpW(N0{f}eu-;}&QE
z&<M1M@!djz3~WXws{i?!E{8E8LOXaZdf-Sj;A`BXHyO#9b9KH%#H95@-zSj*0?oF3
zL!%JY*us?^E!h|Xb_lEyV61~lfA7jsciR1#gpV+AetqB)1GSUGLul7eIes3)if^>%
zgO+)Ymb*fPf(&Ml(N8~oUm^oxX(t6)Nl|-u9Tv>mD^j^IbB2&6q&zX}ISXr2*M~dO
zSGf=Gc1Iu-qE2kQ+lvL!!WjN7tN5F+WKF{^(5uQ`T71cc9aotPen43}26@TjFAWLC
z6Jv1}&=*pq@a9y{dXN4dCRR_jj1qJ62(4e%@i!-*e&;o68HcV*DW~W88D$9aLnbyr
z(FVhnHd5=mT#k*!Q<r$qUqC&jyg7&eXIK5esQ(8xHK4)y<!{qqe1Bj(*tkbf-gTA2
z4JC6K!zWkdZ-IP3RVZ$Xe3b^}1Be69WYENKOIvA?NX~`p<fnTW&s)Ro2oAsh`gk`T
z(W7OYTSOMbU|=1#w@Bw397bI@|2Iss?|^RDeawd7g@6U14@4Q3{XkQEgm!m{hCcVP
zi6%c&_*vH$s$Dqpr|a-Xzxh<rT)V{nms)o(p<IvVO;r)tkVqd;Kv0wsLu#nC=sZ5$
z&94t)iEAVp^XMvJu1^4D2WNq!G|&d-MPQXZG-uxdSpc<)ATl5lsvwL2!*D`$fN;>3
zT!SeAj#@#uoWOCZ2k%l<;BYSEnX&m-g3qLe9+!*A!?0ubm_os$ufGt^5Y6<!Q$d<A
zY&%870?;I)ov0Rwna7bEN&d8Vah4jLNK)z8*OTg5ryBDKp@qeSFrJ;ceL2D5{Ls2=
zg%|%J&H=uTtmn7?yzu{{^@xqj+4w2tI!YCUhhRdRi%-Q#M@cCP3_?N!%`ff(W!spF
zr^@3=ThQhFkdY~BteQ1=Z-0MAs5h6r@w_zx{ipV!xTwFsd42*a>`&43|B~&Pf+RJF
z1M>g56SDT`Y~a7W0RKJ5m{?HKatPsDtrncRQdJ09$OgkraUVqrc`@ymJN%&>Y5H|9
zn0NucS7>31n(qLdMg{X&&IBFDoQ0Ab86|$ea@Yo%800lTJ%l=HZp6^&@+eWh*HAgk
zg9L#m!A4}SySR3_q6EY0poh<R?uRKBAs}nk;r*NT#cX`G2h|Xs0EMbTgnejR)$;BR
z`#(v<h*NJNk{q6VdwVG7!YwlQugF7?`-l!VwAx%$185A&cwPzcHp-_8JiY?3OT;V0
z_R<j40?#Mla#>x*z##0i_DE!S4U@hK(3X4!C%^0X3VlQhhd{!S!zJ|%qa#R?7<S1c
z=ot_eY$`Z0#8ca&-4C~+HiMi)eizZ&#eoMvBTu1Atd1-NLINpL6U>!h+pvldAmc}7
zb+e%g1qcO_-!dMoK}@5<eF~9I59_xZQ+?ofyx@#kJw8i4^lLBm7;rK95Ed<hcGB@r
zqZYA$&|Kq3QiMUk5$F-igw*T=e=RYQ(3n7s?A6N(5RzK_=n=P@oq2YA%}290QhO3c
zk}bPw=h61QB;^D73x65=UYg<+KEVlM3fle}2n5h@ycxQVY~T=BTdeNNgs+PrGI?G7
zRQM6NlP$hH+76#vA3FcIO=a9I(NqDBgeCX#!mgKMh0|qtD&H#HpK>ox{Q|ncxKzhq
z^PC3A2`&WEZ7MSTB{PoF6ZeBOfn!l34k1auz6~S2UbvMKevj<l1NqlQ1EQq?4xh@s
z@fdPEGy;e6nHnB=-3s^+n%fKN`VPujdl_=PPy6_HLP>+jvx^`0{WV8mAaTk=6Uq(5
z9zx^5q9kb*3ZKtF8GDPz{}IS@J9F5xs5Sr`40Q-WR)b4W@6H{jx2m`R2!!1(tNkvX
zYpH%nF+!LjgfLMBmp^&4(f`8Ud&hIx#_z-KyP+fv$w)}qp^R)&G9#jdq7XvL$_gzR
z*<{l&OC=fE8upBoktkav+1u|p>;68^U(bKfAJ6OceEo3`<8xitd7bBb9Pi_Oyg#PY
zKS#KZIQKp;ojBbE&`;sjiEOMp-U}KwR6gkArIHs4Ftz7|4%T1+_lr)bU09fI=%d_<
zA)wulr|CBMrXL;O`cK1fxOh?a-8ycFk2CUm{0Kf`{J`@V0of3$0jv;s%c_MVL%EsB
z0T(BfEA#GeVkXdK$rql|<aTJ3nO(aqPqqVIPi8+__Z{K$Z`1sKOUWvYP+7|h0CzqV
z3?{pcLeU1?1u)<WLJ<Uk!+15@TM!1T!rkM_EeJ47^?_#hg5tQ0n;-4W7rthtAqC_)
zf2vvS-s<K2tAfSs*!Pn_4UvdB0R74|U<DkgdY|{4+s_=VvD2-fIrfw91dDA7hEuRc
z(zCG@c%9;a=hwR5XxXr4AAyMSezD*s3j8>V@L|7Q9<~$R^mo-19Bqd=IW#z_IXb<|
zFa2WORV4B0>%6;f!7V{_mEy49=6&EcmT$Xihfov^$FUEKJ+O9d=g1y^EU^csYYsZ3
z4Zq62pa*64=NJ0dIV3Ox)?K4(g%=QmV9+*U0_5C;Zoy-CgsQqa|1bJD2(Ss0pY&1o
zBp2wB?kD-JQ^kF}(#B|EF8Nqe^K22&HFUX{8Lt%x;7rzL5Y%kHtX(95#EuTVej!L^
zrVV^@dV_A&4sYjw2+uS`dk$gN$*=KOf%V&agv-YIwIif?JeE%c1BTvnGt(drlo{(j
z51<-Wge6<I)9F6Y#7v6@Uxdpx02)JoJ&>r0;9o=_HP}9iVpL>b>ZC7MLLMLyWATqZ
z?~14VhzbtL-A(n8kaZz3y*FrTzr-t0Y*Re_IjuN!UY*Z@=eXS5sjDWG4;2vHwkEBH
z&7GDjz08Ug#P8cjLdIa^!_h7ESER6tfl_1KMIN>ffo(-~pG!dg{l%t`e=LPmS|C7{
zo&(G1@T+T?lqs%=QQ?HB`+A#H5;Kmo)uXZRKENkr?$>j25uv8PDENV8XlOO0B%NKp
z-gD(RHxD32_7cJHLn=glWVhN!7Ga~o9P^*2_sk(j04a~0P}~6-nPPmG*lg&y%vW<g
zC|dZD4yd(JOKCUV6FKB%2$hc&$M1~qe&am{z;$oxRwreiQd+>&9}sMY7McRLT*ts&
zfRLUb$>R&<H*0=5RsjeUEE<1@Qz0l1ttb~j;<73pk|$7(L{y72RjVsTqww9(wf_vD
zZrnN~al(&V5#$9P2xcRu+h#{wMo+U9K3^+rGmLgIK;qign-9|vc7D_ShJ)zOn~!r(
z^|=MAsIuV9t{L0*WD02-5cbzE^&!Q~bm06O+{vc&w~r5?0ikN4!mB4G)<$(=#O5gq
z7}Eo>S7<t707;b7>Glb<!;8Unot|3em1n`rLw#)Tl4rUA$k!%-_-VN`$mkUm##`Z+
zz<BM%JrQ^nNy-+f(U)<SC`AGFSph7o-D*2`za6%{bY(iDlLOASaIWXxrq3Nmx5$J?
zN9!>xvtJ|$Dc(=n*X;9eiKqnS<;@+*Sf&uM$xxoR$*o8V8pV74QTF2M)<=^?w0T0b
zvuboW-+)>McKS8YA)Ts0FJ%>d#E}%2-}KMJBosnEgLG#f*@Ddq+q}+Z9Q!Q_qZ>Bf
zQVR!f)B&YjxXTP24crrsMG4YXsP(vE*hc34_ut2{?ObuXxNNdu=izk7TnAEeg8~X<
zo5?P+r4%PvDrrXpFfsT}BDzlpkpes)+y+zwE}=9?QVmqxP&Xv?Jlo3Sq<V69myL2K
zF$Y-Ygsj7-L7pQrP-s;s-)%)kiJ&c&(3`#Ng{n!-1i2+PTq`m#$QuA0bH)FnMhH0>
z)0P8`%&;3Pa7jS?cD~L*+BYQVIj=SrVhk)N51AB^q|)Ki484xb2KlJp(kx!K##Q=9
z*zfSjgjm}kK84qn0Ne&*dUd}{rN$RU;Na5A4-xvnx7RUv&9cCWQRcF^;1F=+BZ~B}
z&e=$W;r8ixb=nZ4;Yoo>FwVZZfZKb3<IHFtx9wC3JU1L4`^?)sgxH-|-*3jsw4)%3
z?9B)UJ&g)+pUYVOAk~4dH8;7Rsb!#)L<9`9yx_)l&)OctDX0zS5k}=8HNlnl$88T7
z0(kFUkXxyK`%mb=6Rb?7%DKZz=c^IC34(n4oa{9BpOj3d9_Kr6Gu47A40?UXoB^$C
zf8A}Ln5x(k^%1V^qq-PJ7l(z-ln*{{?=?aDMEX+16*I6>2x~US*iWJ+;I~MN=)%Oh
zKTgmD1qQr<R+wfMg9|u>gK_p6o$F7?Ot&17eN6#hMM5T6eijj#G3XBHAxvyzfg+OR
z%qhR);Y36WH4J`W!?|xCUUY<n%r`_!bSsP%QDRpNVV#MgB$>zamoOrov_XA4X9pRY
z5Iv?X;37IraSF^lal70BorN^@Khm^-{#*5E4q^$prSDrpifsyi<~|-XS6SRDc!Qvi
zc2vN!jy^3k!4gz04@Q_k6wlXbW6UlYCB&YHYPS_e<~Q9l!mdClhbcKrc?Refazs(1
zY<UkFJ@@6VfLU&0g#jS$ps3}iFHb&L^?!aEI4`ss=&My$Uc)m0g1zkO8u$B&mgin<
zD2zKfYn`)JZ71DhquM`**M#0E9%1RtI5&^zUGZi9ij*F)Xi=WT0-T3C#9>KCngv(@
zyDi??eLN4RMi?<2f~O4N|JVQs4KG6|2r&KfE{2iU@Jcw8^{8RvRW#JpJk@+ez{Ws{
zfhcUtB?Opty4pBg9)s~tC|(hHa-iv~c2#>)<Lqkm%;QfXBHd>|kUba)fDVQDMEno*
z$cqsvNNkF^FK8rNnwpEH#i?|MiajD@UhGsX6@eXb8n5#g=i#=1=V{>lV$0+HK~PG=
z!Nkr$=5V@BZkdGJu#{R}$W3J5n2{i>L1GYnLIV_Ee6g!45l&5`jvxRi0^6F1+rb*8
zgJO%oJM4rk(ekGBcm_<JXhe}}7YqRy(ROVYyyFpI;j(IWvscBX@$Rr%q$L(IaY#~d
zCe!q?28hZu*eNXbbq9GE6IS8QI937*#tKGA@fYJz_C1vzlK+OJL+*DOEe?5LVa^dR
zD)8wM7rBf%2v*`YLdT~6VTAiYyWE3^M_`Na=P!WN@;$wd^aYk4JJz2U{lULs(pz9j
zAFf{=^jtwWCXu<OgcMd0unf-b&Y*zneg*k1*s=Aeuucv3jWSL>zFZYHs!L^9)Z)Ye
z3j$dZ82DeO>DC^&OW)3!^mBzXSc>R6LnKWe2-^1sO)QNVt&-_m8&?kthM&j|y#%rl
z8e|?|mjNgTaHNH!2P@AE+!OEMIjnq14C#-D8WQlWQe<<emI04@3EzM`!zac}tP1rM
znDTqbGf<;rLq`B&0*IC&AH;+LZS2xb<4rGd|JV_vCm%66E~1K38n&%_r)_0_dI*Xx
z?n}QbNr4Ap9U3}`svHazQpp5B-ClF;@&2f%9{YMR16vj$JWWzPPSWahhUaeRi`{hm
zS!xYl2@1Q2>48Aj!(^aLOR`sc+b^(9O9DxZq>NVS&zJ?k3a<>blSEt=Fs>zBPG}+4
zk@zr3-5H`ZNdEq>Dj_}$wwlZcqUhG9Du=6!{dV~!K7%MqqIh6D_%3kq_Y{EqvYlWZ
zi_h}c!9=3=7zmFAo7Qj+aeD=}?9TTedU5KR4jL<ruI(Wv5U0j263$=&j;%_Dji+*x
zVA0m_4tjF_b#c;A9sZ3_U{K&XKF*ly^__4P$G|P&4jk7-7T`%FmmXm6dcXe;KpKz<
zd+z&-HR1U`bH8BJ$}WezwiFoq0POq9L#Kf6qdp}fuT3z5q$kiPSUZ@9L#kYl$ck$=
z1st#}By8}%0nD?gRQ}r9i^N)WNvMJ#j>(aROirB&fhI5%-Kb=Ax|=8^$|1-@501~}
zB-q@Hq6yWW1XDA!E+iTlxwP?ziiS8T_CMT7ZlY6g8mqfH5m{QM-i7MR6wVjt{6Z&w
zB!9D&Bt*!~;7K_@>OAs^r+ta-vC(7Mu1|tVX2+El%eRV&8PI1%b0|UmX_-lL1?qwX
zZQY(diP*bx<t`LbQomm%V=>@}0t&-P;)q_J=sa&^xz0Gt+m})xWs5%<xV}xbQ%4m7
z8sfR{;Z%K}=i|9enMg-WZ^?#oVXFO%WR_DWOGQZ{#N4<hXT^aW<ZVWGbcsa+TQ`n9
z{jAs$p!yRpTLDcNJL$ns{+^QXN^Cj-nq<{?7BFS5|0E5ORV7%~fak$axBNb`q(EBv
zG^F6+TzYhS5kLnq<5(o;i%r8@;P>+1SAx<ZJF|moTuB6W02iuU?%(Kj1pZU<XZA;c
z0CY~>X-z`_$&0`u>hUmT#g&B3>XbXfNHo*vxnlpJNJT<v9%<nA?BL1UfkiE-A&cW(
z^>Y3K@?n&1p(T{jh(xbJda~}J0N&sbglE-L;@{cz28bz6f1mnzqR~F1EV0Sn;{#&T
z%PofMb_F0bm;%3n>$hvF(r(8VpKW{LYQi4Iqu*U-Mob!Ko|y&`xTJwPvz`cRl2mW<
zZ3;rqZrjS;zKLIu(vl`;my(w}O{h_}(R2ey()t8)$S${EB-Fxf`dpr5U7b_1jx~?7
zRm_aj><TTNvV_OAuE(j9$~EhJ@`;@2xz<;0;lXMdZBhr}yd;AsYXlRURF3=8^gX&$
zIh3FJ!VQ}PI6}Phod%Tnw>spdrA#n)kh<H$vY){0gLKrSC#w+w-)cBGaqM8Gbm<-?
zJv4~6I0v)YzV}LFSO$Ec-dRP8P%UBLd9<a@^hg48=#k}T`1Ak|$-zJyvbS-V21tc%
z)gJG@_GH=<(g(+MZ(S8%LX|+F#8qZtDA9LIq(6`bp&vul*WJ}Q6-lkX<-YAd##2aE
zIjSF+aPNSfGrX1Iv$|Nix~s4jCJF6o51dVn;V(&5@}b!vC`(haJJS&zqqDnivY+3~
zmU8PkyK0VgHU})&8y?DK_t4Ki8DBHcLgzvup<3<R1hyK(Yi}2`9^@H89sS7W1R?}d
z_bVDfxvlRE<!*-$u(6XlizMJ`FUCD((w&?JS>b_2h*q#m7iIl&wsy<yNtKe<gJR9`
zs4@#RMxz!)%w%TaU|<tmRRUWeiuqt+Anv99Q4QM?F78|(5)m;215YU$S}2FtsVv<m
z0cPKrO8*QnOHSdor9b=g)DTK?w2piW6y*&=p>+&_(k4@yMI+Hm6+<Y)xES6Wz@{i%
z`?@PJLaZ(md!7WDZ+jKMDMGDL{pv0cFg>f~9=g=>zYc#RwO||8LxQ%a9mrCQ)8SP2
z+r(-eZL-(FUa304DWPKMdKqZN*yXG|`#gvY$&Eb>108Q?B~O3yowKgx4MMC`UZ7NH
zezWO|WQZ>V^?B(Uq!kBYmpV}oM|??&?l_qx08`-icH_XoV1{X%qnu2(DsUZIF;N8C
zarbKouNC{8=zH$nYl2(_S)=cMHz8195R%CtTsY=>(rb(q${;1%P>R$DCU_7j5AUE#
zF{*D!+GerwVo@#S-a+Pg^`s=ATo~8?h=ggA*j+w*1G#lybi(o&B1(wcdF0Ut5VBej
z?&c-lz{n_qRZw}=L<)A}TeKSDgpdVKM}(B&yt$4?0aa%0d*(Qad}ihQ7Eho=l-p+q
zVUrtt*MMAU+jNYcjO4(f1kcSCzK1<g&L;2sU1Xz#nrmwig_}c@8bZyG$!_poz?XZC
zE0BwT8iKVWpgzt%SOK{9^w?J-5E`_*WflKYT5rT`Tei)2G0q=~$_n@69GX?uS^Wst
z)|kgGVP8E5?+!UcCZ|00&=m-|bR*iSZe0ZlOUm&FAx!mgCF=pu#UqDHiQXeUq6EXh
z%sij3B5;7ntpJV+SHE8#lw19e=-4Z360#i;S7z1%xDAL|>GJ9k0#xiI4Wdz8R2#O_
zcoQ_UZQ_Vrza}?`H;Mng*2_cY?ODE)#i`TQMRvfmS%vG^^UtCvvl$!m=SKA#+7Z8z
zxh0kh`uL3~;{p?+@*)%j?2)|5Pbr^odOp5?91&_hSPpxCRRf8V1aloovmp#R39o>e
zBIE#E`5Uk!$Q7(kU%z&Oo&PG%=q@s;Wq;uY3)0h4>^4{zJvo}yNTxkL(&WZ?j8HX!
zXc@O_ono)Sj_W&X)`Bz^*A2P29?=ta^8uKqxO>r8#Sv18A-t2esfhW4dO9iSAwGda
zM8{4(_FEUCBfK$)BZbg%VHO8`kqO#J<$p=z1{}szqBSIklMv#u;H6QbBNUr}ym5kf
z6}gg11?Y#J?cQW4QkA1}kKUtDyd1wagHbQX-4obT8Zja~W6Fc3e+XE@4Gg43zVLXw
z<TMy9@u+Bqny6v}BSs=<27EV1nlOkKm8iTU^a=Vpg%GRj+Lz1Kv*VeSETr-XpL5Rc
z^UVjK%AwkjNL72EkAAUEp|kyak1PkXP6>7#;{H@ik652R!D2hJLqUX5pxP-f0m6?G
zv#mm(!O1a2zY8AuWvUIyx`;!ODeZRbCr!3}m1T#UR4@5#S)y9<Q%oGZu2cZP!V|H0
znkAxp|FizM0inheyH`!vXZ_Mu(O<A82uZ!-Eb-|QA4?T@!f#p=Ob1-T2)7SImtTLK
z(evNOy0w9DmX)n=2x(lIuV97oti@O~6S004KHZe#(ZEm%=4#n|A34{c#?=={UGkKW
z;He`<O@KEbJNEfFKoo_NzHlVRW6=X+CmK%t1VH2CDq@GEB0u$I8ury_kSH?^iY4>F
z3<Kgi;F`7-K5l>q7C@Y?tnInF*lqHqdfAN-peto^`pTF8jdKq5X0}yIBy>pRPW9(P
zLqam~h*gI3OaUw+k0nk3Rg43l)A}}ki%V|xighe-5Op(=v(G>`xPK?}Jn}}2#Io*;
zJ(wDQt>b1k!Tl?8T>u`?pKjb_4I>G=DULA4xDZaSIL|X*#ZT%TxyZ%=f8Il&KN0N4
zgf<LxWJI)m`9#Z{HOB<<SALJad~XR*Y?PUsiZHO@IlL)<P$<wnMEkGT@)BZNMonxE
z=L3B7KxHoi7R>B`7m1zQnlA%bFXS@|TAA}4al}?Z4F)yS48@a7gT0BR_AeITgJlum
zrAgvAlH%81N+t|b>Rf~1;s-V<$|Tkw{Hc(;`%%YTfo<q)^iuY;^DssH93ayPZV;ge
zDSLtwz%+Xm{7%Qdy}!`y;45|B9!Y6PvLJd_jT>c_y6Gv%hqyiKTVCv5rCEkxW40;h
z)_aTduxu@9vjwo716_PZiA@<e3T-vjm~iCMsqN#Iy03a97~Z2?wn-Gn`d0FH&*c%L
zu$66ej{y3v5^SMx!nD)-bfVds#@{e|5^2kMN3*~kDSj-SD&05R%r`rDOabZb?GCwK
ze_`Vhl4u=s(Qx5L)hC4GI6H=u8WH!`G_;RXk#b_3&#&p|BTb}4xAhB$9^YgCWSna3
zfB7k%SY9o9&~PRR;YcPORsSyMz|L9<f=O(QQLO-F(mn=-Hl2iSw_cC8KKcw?Lk9<u
zrzt5u#4DE>`*afJ8B*MsB?-1qXSPr5eTC^k(2h1){y-AsqqF#>du6A=p4rEe8O1aD
zno^*a1f8`w4Z`?Lg8KK0q0IDMh~h~FAm5~uVHu9XZ}+9pMOmeZg9|{DPr~Q7)xf$I
zz~ll|`vXWLH_dE)E1>~SR@<T8iXjR3VyN~9BQ3)aF3GXTY#2l!9*Uw2A_L1{m%GA&
zqc43GQ6S+%+O}?FzyL5=&(oBa1VXADKLL{ncEj3$;GeL~J@YZNoAgJ7Y>ZE}kV6}%
zjk%oexD(LcE<@Y>rgS581a{BfoC4vRt|x!rCD4pQSf>ye>mp0ZON|Q&+2&;S>5MJw
zV~jU;)Ow&S?s#wk9*<Sx^xG93`_ceX71lENpZ&f)anN}<Ig*`@R9MQM_g?~@C7}+N
z&2C!{t5WXq?(V8vZX66G$=hw62p8NmaDW8xHu3OT{m}z`3IAFEp7UL#rSiIU+N&MF
z49x4UCp(=yva0;r5$-L341B8~8?a85js{|&d(SoV4!Vwp-bRbLvRP%`!39JrCafFl
z-Fx@UGsujEv%xW8_6n@FC*p58#J{A-G<2_ygnM7iZc)+zPLu$2K*~PwMAG<;-p>R~
zwtt-qNIxxelLUem&w9~TQk(-ULF{g?z-N8Xp$QZ+Y3}&Yj-{r~7F|iG<?b6}FNV!U
zrCYairv{{q+ZKQ&^!Z5rSi-08f2Hkz?OUWFDSTP?0|${9@<Et*$(opoJnlvz{p!*w
z6vAR3&q$)av8gj`X?tQ=6vOVS_Zi%Zf@oAj>5{ocf6V-ik@I)2GIeY_U{;(%27QuJ
z{QN)B*d5fNZ)`GE^*-vZrOW?t0ZOrQyk|D$m-d0vDD-x=u@uDAkeKSS5_v#{Q6H<2
zR>$PY;kQ&@-s;DJXaDGcHaa%TBp1|s``CpVKLtSLmYLCDhZL7JHaq}wxx|$zn2=CC
zN+E7VC=*|4se6t<$9L&QXiyU?N-=FQfPCSa<#p;GUR{z~oj(yO3RI2h{>F+1dRD*u
z)M+@4vdjE4EJO+orJ&vTqw=Djk8tZf*ThMn#})6;jMA1Z+nO{)kc7V2R&iTVX=jeK
z`Fabe-k~7cn{ui8Uck*GZ{IuL>i~an`vW^fG$byRrf&%FS={R0$(2lX?y2!q=j`CE
z^-NsXK+{vucLrJWac;yNhULSuP476w6@xmxWk;T6QGGr#*0umhOieQjo+KnP?R;g;
zwrK0`Cp;Ildsg;UHN5*i4tShhl$3$eo)embb-n$kPEvb4DKS(0_^QcNo5Vuz3+}Ca
z+=`&@tF)Ng0IgwX&8w~)B9#Yg0wyUv4=)6lGjAjZWNyGV98c>XG86+g%U6maVu5Li
zfqu_-I-cM&_o(8A?SinbM~m(1%<Izf&qc6(p)T>8v_|*e(jt2kcssN`WYWCy1ZzRP
zpx2Qb)Bks41Ut6_rN{97#T2tL#8pfVWhjD3=6b-(`Hk-A<(`i;l*qa}aP5OUWw}Q)
z-L`|0zWv21RBtlMFQvxsbo!ta*c&J%AT;0r4JY4NOO)RGzUZC}v3ht_JT60mQJXJx
z$CWR+{IhEX*V(sUjO1`WIZdd9OHz&W(l;UD(zj%SEE#WHRWl0(8HC}RJ7OSi4LrJU
z$qhM(R_ZDs;LCXU!EP7Xy@d3sw4S{`p+~q9-{0-hvJAJX(y5aG#DwwGNyHm{gA%@{
zaPHFd24<1zo4=svn-yl<Cf0fN=xA#PvLE5G?N>wSHcg12cPDLIVP>KZqBx`^cB^7z
zNfBtkav#0UiKeT3NOL7pMe-Z$6lvx^1^ot~u%XWPv&sN&h@F3Pqs^!T5Y?bYjcPbN
zM>q(jhf{or{?F*{7(_%t3<n>>^HF{-WY;2bB^@7U*2IOtJl5-(S9X_jguEWn<PKkY
z=bqo+Wd;A{^&$1$HMR#99D03{WF{rPDBC&!6epaw!gsI~A*W7IvDxk3J{xh$#&*ps
zALK`SLaQnbB^709x1jM8&9VD`-)w-*sxnX*yTC3n1~xqcTY%lPV)s+q#KP9nf2p2(
z&$_a8`Ut;a(VQ2(VU_M)(CBT2yFFJXAB1dNuw!=kATL-vb#ipqF_hhK=E5XS0}dr@
zjn0oFk)WN9<roXK?g4}-H&L}Wk!p3JwG@Mlc%O~~yR8=P<2OU(P+L{yvq)_e2(+&4
z-CEt}wqpJH;@Y+~lg%TqooXI!zdB5}$Qg!WF5s|#XDO*d7<SM7RK9Lij9nfg!ENWc
zCTN6Y%x53LTZdTP1LKr;ST^O%SY#TCgs9B&Jo{`cYTb5V;+M41_`WYH%$0-l+O_Uh
z`+lVE)bod$aQW`^T^z6WHv@jzmGcBdo!zUA@Z1UJ+P>ui4d}GBy>Xc|oWZBmpPN8B
ziS{@tqhh~vtUNp7*vj}%OpJ^Wx8l+u(}S$l$Nhu&j~u@*9-0H&r*%96kzj~Kv+Rro
zJ&e?J5Wm5Shi%##is*O}n0r8ce=cBk_yz(*?{UluCS9`-7O{p!LIuO{#@E$Z&sDJr
z1vMMOYnnLAf8Z42u^IHH5c(<3qH2mqXa2?8pQjftSH)Y5b`=*=@b`6%9JW!Xi(fOZ
zz*o(#O9M|Wi3o;PkTt#ynK84-YsJ7Lr8L!(*N#odk>RBZpWPa_<<m&~NK4$_DPOJi
zykphmm+YFyc{AQYjVITJq;UHg|DtpI<;(9Y`uU8p+VN+t3?|hePJi8MawW5yf$#H~
z+rI1gu4U5iI2PQc&><^0voaW6sC_2l6?b<d9~lhEc&%MxF;=)#Tlwr8O<5YOb%8jI
z*74GaJ?#f+qCuY4+Dl4|AD#!I*r>!}44PA(anB_&F|qkL*SCGPRag#zpm()LzrR~u
zc@AmhKC#}fot-Vs_0{Hkh=Qe@XID7OKon$wk@KPPU=?mq`S|g0T}}UOsYyv`qG@?N
z!ePf=oh>+;G5agr>+FH4n%ir36&=;NvhUwFQe0mT1Ust)-n;kl#gnKS^_ueX@{b=U
zK|H<1Chfk^(juU*52_2|SWu`h$Z!4k-+wWV<|FcG1PMs!$>{VRNqx@dmD_5RhdVsS
zv$I7oT?1l4h@ZryrKJxYIxTthC!*P&#*W@p2P4r->k@U<muxUn6S##_=Glx^zLl%1
zEAH<zFpso@vVPzEYt@{j0LE7+eFLBOEyegZC^kf|oGnM>0GXr!s!epfNpn|Uux*$S
z&tr?tP6qaSXyOQl-@NZ_Yd(^A#i7S)U$S_>IZc;)T^R6!<&nlcVgEpA*pNCgiBCxA
z6+BaO(m(6*pmG?y?7&~riSgZO*pcAWoxU!|t*WL5ST4?ZpDW5Xi0*5eY0Qc-It#;#
z@lT|=rwUW@ZrstqjI^04KHcDki^9L$a;84MFY@xqGBA5#k>3tFjCp;`MW2Mlxj78C
zO+y5Sl!xG+FjsW62h(}b>oIe5S4z-Ph!vrnl>zFmobxyvrh;NXE7K1u8k$2(!Etd*
z1NFOd-a;4Cpx{BZVn$3}U;!JWHw0n)zB|veDJ<z--%@(7H<<mHAKGy)0C+G~xq5XR
zG^&rwd9*wO4<Y01+hIT^vGyz+evAi<vD;rh5@;-WX;@srHYptXvNPFMnwr5y<EVZ6
zHBRbv0t|p&DQbtNQ<S=A{dIctY0DQM4#=2<e{H#I`0F>1!PTMrl$3{@y|*e-j<0ZQ
zKD=>lo$ZP<l!vaVYqN-9Sds9ZCi}t3Z<b*9Eg?7a_5B7W`1z=EGpLLsI!1^DJD!Ee
zIYlISnKDv#W7-U*Cn|;TH6APT=Cp%}m%P=Q-9IH|ZpV%}P9Fvx+(!~s?J~kgR{EY*
zP}6@zQ{V&=g+Jr&LrJv@DuA}nx$<AWJmB)E;xb*iI#eMcxo<hb<csp(!!rU84)X>G
z?c?I&D-mg~q-joej%mmA;sw+^?>@i?E*viwz!+#U0VYsoGzFzQ=uO5%F;tEp8-ia`
z?8nd!++yw0-qq+#hvAd&`7WN(7{?|e9XinQUFvtkkI_iQB75DUW^*Yt*wAiKliIcE
z)&W!`35{D#@=5|?Z$@KN*J4?S!-|W~fgJ~V=wQyKon&qh_DHDQs9^G3OR31}n?<FC
z3j%k=g%t{P$EXv{2mB&pF&+~NM5VC(m}7E_4N{=HNtxyC?OSZ(;^G)YkPWB+aKxd0
zW7CDS6Pm2vkmW(8+XQ};Q^O07aC&irknIrn%6|7x)c0QO@}R+J$4$M#fbAe8aZ+;0
z;N|(7*G@C@lyAyb%w;TBZ_e2)!X0>Y#{=UK{Bu{3pI=RVZ!pS3zP-VKpcE7o2=x=S
z4nK)8jM00A2su%u((n8E@@y`qSUaJVxcNt|QcgAR3jfFWY)D*l<6rHB=d>eHvkLji
zI|R<o0jmNv#jwEPiG#oiOUoamZ_22o2xb6)KBU`I3s;VFLN5rQS|y|Fj&qAR4jD1o
zx0=!;#@0|-HWzb*sZ1QW>EP;NRu?Pw3o4||E8xJppiY@GYPAqnv>rpJB;z#}rklUP
z9u41Z`ozN0nvk8=_TInddsVnbDw!{&^&N@DEH(r}=o;PQ$Z>yRXC;R*(yd#&)?u`*
z>1yK%49@<El$f(ixK`HhpB^a3L(rShVHEpu#~5I~h?kM`#oyN#jdQVc<!4MUUoHlm
z2~xEA<5IncaN~+k+}6p@A$}oVEyHuUcBp%ws3<DRo?kJKQ1Q0kgN?(A4BeNS9jN>^
zreqIKyI#@|F7Kn>85zx|FVG6!AX@PofjA=7ZH{h%1S=Vc0elUmbnu<DF)bb=qOE^c
z1FVJd4u4c8#MB#1y?M9Ss{1FATQ<Y4N|k%C6HsH{CF7cXX8Z2lDEC_*yp)jZk5`=7
z#hKvKBfjq|7gtxgtul(q{k_3}Gn-mlU*jE3tn(`>E)IA%<R|u~<)nUUN(!&6nigR1
zu);XKXja#YL*bte_4n~?`t|Mu-xeC{ahg5o=Ab%0FT3}}`Y;=2&D+v1qt=juGf)p$
zR5FhlOOC++_=29mRz9M@g>V-pnEAl@!!wSAerZho6~oIAlE8ECM+6$eer)vT&+=bK
z@D~R%cVV6jG^9SKv7LKUIdqEFlf-a{+gLv~vKN{w?%>yNN!yp6!U+biJ<@6{-A=Wo
z<z-;%HZ>)F_G2QPWyh41_)0#hN3VUF`zB=Oof2j@K)tfIZ^5yBckBTz7;Z@3AzQez
zqoiqM8iaw@*BHoz0Gz~$j1pvS63|A}-_CN+-oXJy>qLl;O4DX$WMs@Z<zv-=1*w#b
z-!3j*#zVA(HtWaTzyArG2SoP87}}Dk4#mQbMcxLNa8HM#6G?hBOG`^a1&crj%WCC)
zw{!G4i5`!1r~h4dwn?ary*C)GF@iw#;h(T-h;tfI{AhR}CupdEq!Lrahet*hK>|yS
z?X0lIlmM*QHj^QY^#%o#fQk4-Dmq?>#Bw#flSO|G3eU}#4;($J^xe})l%Agcx7j2q
zE!*}ed_0k)9KxQWd*~?g7{HG_{MAowt5EIC$9$SgX7_<0Mmdi$?)K*9tQ7mLvZMUB
zVnMo~S+$8kzA1Av9v1MDZ!Ockf7h%DFs?y35tgwtJ;IUqzyVE<TgZ-qQ8ZWw3D-|O
z(j^&6VX5mh@4<sw2bV*7jE-i8Fh``{HFUXp{<hPG3l0>TTNxPA&dF!A-czCOUvqJI
z2c3(t0#B_#L=PlB`};3e?Ic?XMOUob7}CUBY)L68gmeMBSe@cUNo)$npS*9xE+Sg7
zpz+XI#gt54gqr0w5rqPg;HGp=k4>95ZL-4C-fhk@<C|yX;N-mP?QLG&ng!Sv!)4A%
z?y-UXNXlrMc(D6KpRay%+l?C(E|1dFt#SpP^o@<#gkKVY!Ecs6G<@>xw<>lNn9yc@
zY@U8sxz5wah;%_&P?Alo#6U?{5*8K~z0TVG7ZDS|@%krBfT;_jCR=0R$(QHf;<rs_
zX%9uuB8IJcKZFNp8JT9<vujvgGsJgeOqUjLT}L}RAq9RiM;+*fGQFE}<kZZg9%vG<
zCq(nbut}`@xw2{JZtuR@$i|n{DQp{nuV6=h#qP%isd=2<y?;+h&2Sa(#-aEA03I@6
z&IKB<AMoa$T;xAs%e<YLIW9ar$^D$R+b3&l>$6aD)D8j~#`l25N`7MY(ahk>VZ?N#
zS`Yi@$8q<pXUb?gLS1HS3+tNCX>rEAh)*$$=EZP;>FgTDl#j~82{y;)!A`-wOk_{{
zt?D63ujU8~RA+X@$kFzurlz4MpE30U=q;MRyDhq*=QscH=8>@tiJS5-4RNrx?cmkn
zp1ikfjouNfQ#&cDvdUV%TaI-V^U*~0ov2U|oNSI(vg3SV6TSviUMW<V;9vZLzJeoL
zQMQBCOx$58_&a611}W*~Fch$DfKEHA+=x+!vsMviMRkxNj3`BLvK*vD`}HY9i=Lji
zVUo=@S;}=>(ba-=!1SE}DnzB9ZnY2}P7jK4S+-4#O38~SduqBn)}m~wv$SHu<63Me
zn|Suab%q|~Tq8(Vr0;eiZUs52e}bXlX3rg^0PPY)6u>$9jQUg~_Pe0%uaosCy^cnE
zMTMf2Kev%(V?{*@Qf0mNth3OCG{X#yo(#mXqVC;oZEc8weM@-<j2kxpu@Gykd6E{N
zB&b#F`P;j&i1Ii@dAS;n6?-ffl$LrP+O5UxjaGddFe{j%J=K}L5f|;^WVhIAB#Mbo
z==o3;gL_B6Mc>o?G848{Ake(}vF9%0`d#NN@$1|)dyuT8MLgyjtlWdFHcY7*VM#Yc
zU$iUsjE#+5xNyO;_I;mpj?z|tfj_6dvD7ypKn8+}*Nxsk-k=B+!q|p2E6Vq%AcXG$
z9-c8&U%C@=Qo0|T!hBYkBc+z(sNnpYiZX{2hiTY|2Z4`!6d$NraNY0v4uRd&@XjX^
z;}NS_%qmE72~e{9Y#DZFz9cjux-rX7oc=wOET#CodJgC`AzR7e^T~+G$#DcLF*z2{
z1lP*5_`BlX@CVu}$J3g6KSlpWT_QH~MTkaf*h;wqiW<ROYcSWk9X_D{KK$*2rd~5;
zlt=r#;B`M5DhXgH99$!)Wn`k7u@^j9Rq0raOg!?w=m|YJJ20ZOs5hs{>=#0d9lU?7
z`6D#s3?ejXXlP*k(T9FZPPu++{!mULu>#^;k^zzO^wZjsqYWv33xqEHC;n;0;LgTI
z!Nl$o_^bZZC6xEg(zb<Q@(MOkbHKgwvNCRs_bq2c8V|4BM9cZ+a`WnLE<`Q#89@kN
zpey>)R|>V9Q`a8=k$i|FJt6sj83r$m{lbupfNOy)GEdzfGP8J#OGvn5wn0yZ!~vR1
zXaT#P@Qj<EzkbRoawdHzE2|*QCCp4=W@b(@C{MRMd$tob{o~TcG{N0kfk8oH?tZrq
z#cPOpu3Q(sh7O>DCA;9zP<Ho8A8?n_#gDQZRPI3x*5v_}S+oadJCOk`SP$?-9g!^0
z<Gkz4_p7)X96XAVhbWGQjh#$)E%v}$@Qd{~T!)>BjG;Qcn)MnrHMP#!dv*v6Cf=Qu
z1tv2IDr&&k#^j_VG%X8G#P}5szI_xKe*JuyTS_j)Efxw99*w(nYcC+YZ+EvCUV4Qp
zI`8KYjn;|Zzng3{kFUkV9oC;IYvrS$=&dgWO?%^tW+YWb?u(z(ZcVuqgX{G)hnAUn
ze}FkpLZSmp8C+)ei34hwo$DL452CYe7U>Sm%#4grngYK+vc3&^9@AwyZ+{FjgpIQd
zKu@%;<Qii%moP1FTl*c4Z3$^-N7ZMalhx>5*8RSz3fXjMj0b4fUYLkmLn=-MZlv4g
zRvk7^zbKYtZw3#X97ng><pvMnf*8Tgc6xGhQag@rWOS67g~g^Nr6;Tb`$k(@em<(J
zp+Q4`4|2)hAwO`S$Iu`()O~3Gerq&Fz{Tw3+Q*xaW0;{Er#HDUZB<?~EEIPPWm8a9
zJfRLg?_CX(jVT?E#2MRbYJMQiQJf|Tc*23L_<=^Rr}}MU9b*7!1MzmK0<M!!gM<t~
zF;KU!9PPUP1X~M#lP*)R2s*e@*8pPO9jU6w+D<N-sMdo5-v1n>xSBZH&m*9;@@=>U
zt$E0JA*H5P(NDr82LKFD4~>ii(fD9vwDPSSv0~kbr6Qw-Iz;bzC<A59>KK0ez<Wad
z=7duu$O`xV(FP7+WMous8ByS4{+;eJR@>MMelab_#36A3sn&!f4&ydH493t~-0|v;
zh$~9eiiyvA%1-2*xD+zwTL2Ok!)3ZF-$Cv_nivAEjoPZh#EEmm3OY~|hSz)|It|W<
zpJ*{c;M@||X;tcHfI$yoPR8P>-EoZeUdru6viR|<*1^S2#F>4fA2*u^VZlmTLC@O3
z2jvdMrM?p?DhMd=8HWbro@&476#c`Bot44!fS+%}5g3^y$Cd;hkjQFLAqil*j=3$c
z3~04uGyF~0C@CSK*L5D=`jp`v2Zn}s4Gk|4`;i(iO3VQj8=*lAnT1Gw#Yh0FHD91z
z@7v?viTztqUe5Rf`>=KB67_A^1fCh;*)7F8N(EnAeo}pg4J1%+w9n4TDahA1#7Sri
zO|b%Ij!#4YRgaH8BwTi@X-g4p6;xSkkfIL)EU2ygh>;OwFR{8T;?(uKpm9V7^XMD^
z>Bw(Ix%@TWbM40#TzDG5@K<wPh(K@J!F@hTHDX>u4~RzwW-QFmUL!As-VB|8Cpm-C
zQqUrBmlZ(?E=;5=T;h=nUV}pa?@hO3a262g<l;9K`!PPEImb5XEqxP+xoAk*E58+E
z8=xez1l-Kk#pMz+A7(GYI}sfR2+d@7R#qy-<#gcaqpPK2YK0GQR#?*%?IjA1NkR(&
z>0%U_Cu2uCSgnYX5h6X%ng<;dU5RHvCBwL3qEgK-JdXZC-WOXUyr_0?9{`IEKmpTz
zXfS1gtbtiXa0*v#ZGD4+wrt)kdR?XA^NNKyG@cNmgJfZ9Wo3YDNmo}FlE+9N(E`K!
z!q4o(6KyV+&n*Z{xf_--_pcOcYB=QCZm+=Yh|}vB!IP8zM+~LWNS&OT!lY2>$z$|d
zJ@B?9r*Hn4=UGet)D=}olyP<#l>mNlpwDHJc78eYPEU_RTKYPi9|&Aq*FHi(4~-2+
zTU%@u7tF0f^sE1`RlEilXU9oQt6&RiL9%-=y&yi8|7_kk=)QMFn6TIgqT2!60T65X
zL0E4fO7&d-fvSGmvQl5@J!b%4h`XVCEfhCO-pI<F$^4mqv~S?TYOi-SyG0=6Y%rzi
zMmO9O&`3Kjb)#iu%kM$!-QEl-I;=9TA_oo-PVs~!c#@1rO@}WlDk))*_cv6kCQ4_(
z>nV%<4EKp%)qE{4SYS9VS9oFIm6!*Ka0nZz$&j)i+YZ!hQtb4v!TJi&>>@=~BqK2%
zRCaewEflvH#z{I7iHYb-r!7~O=1~}K96IHUgcPIuLV^szb|kG!$lKe$d?B!$wDj4M
z42@LnSi^WPh{2MA-F(~#4jc)4Agm6dI~p1qxDB8>e#>=?$m?W)EnZ-H1)2HO)YSC!
z^!xWu>FBh;HTR!j^ds|KTrPks4>!&k^sgY4i?B)vg0XAFIzJ0N4VLp<fw#W6V14|)
z*PkE>nNqGr)4QQ0oW2L3jiY{E|M~MFjBiaaKe6==JcelHNsm&#qk?jBa*0<wF&qpn
z6pyaO9&X4qrDlCo$abYcF+(fyF53nu<AdE4(WO;-ecldRsN52#<X%vau5^Th21t|r
zm@Ld=qpG1Hd-dC4SQ=P&1mXpQ(4*XRbRFhY;|f3RTg;;THG$?qcl~U!t0>G;fhe{a
zGN&F>ZUciaSX8SxT{E*poDA)_{_o$uLBU#j_$>9&I~@n^pJ8hGm=Iv&{0TrY>D5h2
z+IQ3oLyTdrmEUg#UV3U|5^@&hmbh7>{eSrI4Xya#;H{9!2du`VSUe%2+C2LctW}W}
z1<p`8m!}_xMFt8PqhxoT%V<TFPTx>3=}@nrv~*|DQZ4u%$5m89#G-Ptnl|rX6*zQE
zqd`&ZC;>MnA5E<;&IsS03R4Z=t^K&p9@7_GZb#SICOPxh2DT3kM2p)#zScY8&kRfe
zV9S7`Xu1Z!xmn`K^a9iml;ygfKYK<xuOM4Ge5VGO*xcM4gUshia5Q;6sC@U^ZENh{
zi*bl>z@AKM+=!43a#fvI1+`tN{IZ%b)uim<%g9JwV6YWDN$Co{+aVQ>JpBxQ9U2q<
zxkMbEDJuO;=Hn&Cm}fBo0>!;&KrI3oqSCz`T@GSpt#?*;dt1~g&Q48<>{GF7d--U}
zD-c$k=EYihGMm@f8xzHnUH+B9tRXnAyNn1)pgVUK+)?z{VbnR9CUHecl0viau2~i2
zB)nEnfrIBTE98Etz%b(%)Q^i^Dcc(68fKJY_M75Kx6K$v5oUZ_Pl_zI(y?QVE0OiV
zCRt7OACpvQ6Mt=M)=(`Kh*n;?C?8(yZMc3oWh6v<>4;@8rMKZ@SPX_wL^`Wp%69U=
zuH78~1eKM#s1p%8{K930mg11-BA^53waucXe{^)e=RWQ5s3?o6)XDL23zSk3IhCnU
zLT<15Q(8+y)6f?+cuSbejvuT0&%8Xm_$=99Q45xsbe7}SfpkGy_I&s*2PEzjJ74ko
zzi1r^NZEswI0N1W>C<yikOS&HStHQbQU~RWa><ShQ*f)ko(AUTc{@KqX6f&YfI;NU
zpCLK<>C-9KaNMJ}m)DbN2@Z~92}aHJ^<>@wYJ%ulIgP@mf9}9x2CPpl?Z23Yrsx()
z?bU=K3Lr+aXIy!>?WyzBP{^JL(*i*4Rjiz%Re=XwF|hP{vr$3k*RSY&&v5SS96v`o
z=n*1mTaUYMPW<xa3t1(<%_Wr=reNl*yj`bVmjNAK&Td8cR5y~!xSvI_bz~YV7%q3@
zdr!|v)brtXpsXvOV0a!8_md}2I_mZN*b{Pevr$85JjMX`MtaupH1`^7VWSNtrwtC0
zC0Ya+{y8%@7o##QiJ8zAzcw1RQLx36(S0btpn!B+5V(LzR7ijQn}&ul02_Z+Hz6vD
zn`YByOgq`!D>H05kLDT$*Hpc|{$rRFjAo!rGpr8Lf&mW>4Pe7svWug_0ZaNgwcA^n
z=zy0zMWlcCpU@^dP6=8?jPi_Z%s3$yaPscEAtdS%tVfb|l7~b&<B%)2W~@f)N98Y3
z{TCdH4Ouac#Hg6<oS~tipVL=66^3F`AZodO5)fg}ZUycAhDB!*bC5G^S4r(nA~V4(
zXZK9G*bQ-nQnF$LnRfZcdLNL?&Mf1m;I4Wgw+i1*Eg=&c8cO&-a21FEeCSlexo!KZ
zw$1DAhUnW@P~Xd1IBv16rKEg-021qni``<=d8#uEIBNLCZmtdQfb=9=JzTeb{lfhG
z@}y_BuH46ns3>PpkFd)y5sOSby53TJ_~yTWzE{Dwskdq0fg<LQ&it%7tCJ@&!668k
zAZl=kan{OX<?jMqEG#Va-F*7VNum`eay;W62AkXU`vH5LdRvD?dJ&T@rP(p+19|@f
z_N2<l9dB<V*cB9CaJZZy<A25wfjb<lnUmS}H7-}XD=WD+eR-Xmd+68j06N9afn|g}
zqj=%pV?YnF2c)jgT~&<CtD}a2M<TESXu)sGXzhJ&En}k5$RZ=;_V$JKeRwa#SOYL6
z9gN;xT$1tN?~kpltaRAZbqnJJErT?S7&<v!ss>&5)wF#54X<rxevgljPfo@nz?-_w
zfl3(WA|Yymr3yFOmv;Jj2Mj+v^V3XJR>}o?qDvpcVP1lWMYGBWWMKHyLifeoJ=<Fm
zX~B(HlvYwhM44MaG$ba*6{!k}tv;y_^Qa#9K7RZdNw*_1OT<*lS@xhQ5bY$h;})VA
zF8&p<R<eo-rUp~@7DG;gK^5~b@=+gj=;nWnmRxx2vT-hVI7WcZHo!#B8Df<1cZB<=
z>+-EN(YAwzB@*e!FwMc1pgtr-9h_9ld-|!4itJ@!DL_|5B0<qhPl|>IF(-Uj9qE4y
zM1moDFpTu$p#$-uGsM8cxwU%mve3GC+xPYZG;yJR0AJ15tVC;vV-bW{7W~ZPZWN;!
z8$=Gbc4~^0Ed0w7;x34dL?(Z&Q%%`Y46~0H&i%mNqzrU?xM(z%6pCNwex3R4!w7nz
z6@$J@N=flGZ2ot@*8xciZrIs8P+L)t^Jl5p4>17L7O?l4cYh^C3+!;1RAE(XeVk*2
zl9-!w>ax7VUW6QQ`bLntfEgDlag8=c>Vp0K6JQN!9{uWi1_ls<oCHr0MXQ_Ai4GAU
zCVfE<2R;>IMix`Ma}avN9zmJ+)wKnkeB?(!DXlOwMA1qOIfoO<r|1b5<=~bBM+LFn
z3K9=^!uuyoz%U)f04ii(sEf;@pb=25tv7^yCKhwY8S!j@^j$NQsg!`<)VDRgeR}~e
z1C$AbFJ@rG_jO0c#l-=59S6ih{u*v0`uY|!uU;0rx&!eN$iFFbDNmu%iV_UE!WbAB
z=6-puerambdd@TaQ{*<1G;<5yxE7?_V2rl|aRh(+se3Huz&tsVX^gWj1_A>>MZ~Xq
zQU*2=o2aOScrha8(66uT8{R=aB?}CFyg1l?Bc^Y{Cm$>L!Z$d(xgi8SkGvfZFZ}E3
zK*)nvY#RXG31{ShcL~v0I}li5=C?<tMi2>g(mV+-0TtH8#YHEGjDC%P)wBtx&qLP7
z7(+4VK@1z>dW<$Jf|3vMKXR{uW}+oC8?OtM7j!IhaUQbzeb0Fy!vfwGUN2&asu>|S
z5^co1Mz+Jp^07M%A&`Qo8EJURtWgslm-KF<EUTRP1kZ`Y(h~0-C$#pa)2r_sD+r0o
z`V}_=EQS#(Kn{TZFf$SJDkS%axCv)&I1rl%(HfEY1yJ{~_{s%y^VhgZ^!{KDP(GeO
zam~K>9|`MrVtN|ZGmtUJAX<r~zcc%i_GcDjG_(wG_~O&1aEdoyX@cvjzT&n6bt2e$
zSZ5RfN<`YV0%4+~@#-}-KOmL{MTK_W>kUdwD9;k2Hv$K&tI+v36gjbCevzpTtrn-W
z@j6gEBt%HaLWW#tT(1RxP+U|5J>GN_9b{Z4OmPHL$~>|1a!5sRH3+S%v)ssi{hBn#
zLm<EjC1MAoa3_nQX+-f5Dp`f^UIw+r|8sTrt5=xG`x~RA$WZJH@7Dedx+N_>B^2!W
zl6u#nx-vDoU||Q>TK=?k-{Q$&FBeNQ`K>9~pXf%hmYG5xSmwPMXD4lLgQ2zcGkAgT
zgK)AI3fybu^H8k;I>EL9;Os=mTV%QA?fsY_-TX<5eeYfz7`SXmF8tokp&GS~@;aUi
zyZ71i=VX$3lJd4pX+-zO^&ejgBz_c^#4I&B-G8RX#}ByjwkrSzK<a}p47dm6j0_E_
z%Y7x*KCTJO5r5x^RcmEX<l^9fqhs_yKodF>UK|w_1=|W53(0B`^)@#(!-C?!&(Gx}
z@ltHM1IwZwC*Cw}Z)%#v!1G~b8p#zVY+-|dD=Z>48ikqD03%sgaPvc>qo^DHc1V4o
z*8Li=N#oQ0CP<y37x?<6YCta{69-6B1P~oXFBqrhAy2${P__C=S%2c0kR?1kxKMO>
z2v$~%iz32RFtFNYQdGCb(SiN`))k^-q)q*2PJ?9miLGk^HVucL*bI^Z<Y@!_{8ZJ|
z9qsM$0-}74zGA4NML10tLiz6QQuI5ejK0A5Nq*cr{2IQ%jY<7TCYJA96k6a+kzbe*
zK8hn9@b2<0w($LW%|j!dI8Al2hqzIW!zeEk@e!0m{#^&eLG#V~_eXm*?uG!noo9+o
z;nPm>zdl_B=?TzES3@GuxT71x#x9JJT}VDRo5y3N;g8AR3Y<N&xR9I%3IAq)=38uV
zCKpi;)oVVX4XK|oU~R#XWKdJ#jM)qZS-Arnz$~n@XX7xr6`@9Z(@w31vt?Po{32pv
z7T_;IwtzSZ7=3E1rLYA2FWd=-)FF`$<OCczALO#qVw~LPV$>16Gr&k-b$9Q+eB}yU
z#9J_#liZ5)TqaX49ISB2P{-e3Ow!#4z$ntTNpD~K9p!6$h_Z%;0N@uWZ6wcD-M)Rh
z+<QG*r;Qz6&k_DF$~bHr5bGg8rme-aHPQ<6;}ak53f&>mE74~A1O&$LL3h_OL6!>B
z;gV*2H>^sLL?`AExZ*v|b>Cp9nC>m0=6^cjW)e0oQ`X`Yh#KzP$tg$@DhyC%+kmM~
z*nHUDV3i;cg5Qr=X|yZ^dZ?gqN9zJCZofTeQlhpNx<<<4f;)~JIl|7)jxYnJ4j~4{
zUS6otcE8n}YJ<8F+QUP}3g7L(vAJ7od;IwEvuA-w&g0d7`|(2`Pf*u*6Kl5x$>!$G
zIayhJ`q^nD(zDOBU;Ut_jsOsEMOj(d((-9xp&OjJdMSS-GOIu3P6SB$uB~Mi6$5Z2
ze)qG2)+Mq!i=8=?N`0eK@RuoNt!9j4-}aP2OV3s>jOson2954`n=VsR{&^uQ)IDB0
zq`rz=x~G6_N@V+D39ElXmB7g)pLU5eS&c9;{w007Pjz^_<!J4W+Dy%=bmc!TK5@M9
z)5;6d4i>QT5s=80VU2!?W-qF>@*jpB{O-jRonUb74|x=6n9-eYpW5V0I${P3L!2J7
z9@s*p;G8LU@SHRSbG<nftQ$mbXP(Nrgf-QK;h;i6=xMTKn6SS_K4N3=EV7i%5}47|
zuRhGX;TGF|gWNj*Cq3<#v8;;1GUjzr7ka{2DT1}&vxI8Q^DyhwC3mwaYq8<4kr6)r
zd9<)XFzfQ8vA(USD_<D8I1X_{WzTx_)gc4g|B3PNX4IJ}x;6s}7jhMG18V0Qn$;hU
zt`5||{7O68+CX;SGugN_U<cMk@Cu+R=EQyPez7KMqx$FuAJ6}ZQ55w{*n!{(QgZ<m
z6!BxrjiU2-QO!o!qsc*<Y=ywj)MGL-2#`6v(j`Y-e5dH(6gEA)vazJ*ZA623ksT+O
z|7iiqisIYXZ#_LIBw1e~qz*6<_J4X%RMFinCnR)o$+4%*7a?`dc~85;;LboSXYUG;
zPeZI@@KR3V=)!akh%|~#Kh7HC6v15eWJt!_p!V%@;VBcvCei{51V5Pr7XOg`5JP(<
zcLCa7^e1;-boT~lXYWzLKW;`~ne-1HKc+Nh+WZtMtZZTok`5Md@5#a+0z$?2E-IIn
zfVB6?B5&2O0Pe=YA^W>IFix0UtNoj2J#dg^!A6-I0XfZnu(S=01Tb-YzjZP55Accp
zOF4=OAH=gX04DS&x{T=UwhW7-qhk*uj@ReS?baX7C^~>8+9N9Jh%!!d>uVsRa5Xze
zw?Dj#n3dueH7xR_QEG`N$Tf`h+fcgu(%!C@D<?r%H$CrLP>Vro`O5n5Iw6^$mNPIg
zfd2ctSIb)3h-B29{yA)@1KTt!%jon%6s0_ZRd~G6%}BT?6()>b?DZ&;Z3B$@^_Sny
z$rJrvHh1J8kUsEBb_+=a;2t40Xl%}k0eB@<|4`UHLK0?But_4|5&ex$a+6ECh~CR%
zSJ9&V(%Ss|WAtGAWC>yF8~+z5@YYq({X8cEs2Ih;)89DOP_S(wrO(y=zP_@3WlkD+
zXK=`uh9S%ttarAl>liA`*$&WWKTPAYO3(L4R1yDpV}0kZ1SkGm^BB3yzh9^R_uGFK
z;{TmBxo9#>L*cATQfuTK5o8?AOMjFqcq3AilMf!jm43wmOwljcRP_l6*MX1N9dKk9
z$lJTo$ssFS0zHjCe?}2}U-`eWX7~~BJIMV1{YqmeLb!jv{vW@+OU;Smunop-0w=Tn
z^Vvll;1we8Yz{@fNLU}RA`}!ZVSu5N3vZm2Pi-CHCu65Dd5yKJtBaQNcyI*bq_|(1
z_}0q}p-?9N`}NR|9IL-yPw@!<{pbIF`+wJjg5v*w8=|J7`Hv|L9z=*{4D^7B4r!*T
zj*!#XSuQTFoCP9DiJH@^6nkutH>=HHZmP`^E#$*p$S!(@hZ(nT4?VRR8Zu<WwB9Es
z{D|$?-+_V>^&t2F+0bh{51cXUr6~Ec|IXj0ApioWMf-<`5wF)iv)qiI9XRsGuHRTk
zIKPnl)N9tk-#vdeXNhBaWAxn_TKv+DbwdsX&uYp-y=cFX`;drWAVagxGXe`5pO~mI
z-HSZ8#BJ6lVv5)U%`$7mJ`@xYS^yrAIhwxOmhu=M$5yzF%od|Ke=xdn<nMLc1OCq1
z+CKmg8uiWtiv}u~xPrkc;W?PxMM2>YU?2awQ4X5X3nbNMzZ!JNhxbwZHE;W?M>X?M
z5`%R8t6a<}qqxCL0xw*gOq>E2Lin0B770YfS4DE<$J_qr?=S)!(T>@uGMO<((MSE?
z_v`#`<UN0W8CcD$#6O?+KY#EHtBO(pV@EOwR|o;xTQEoM|6W`BYuCm|iSyc?#7+3V
z9rb^=5&7t)^UZ(Pk>-`r5o-K|?XM@j(T~Ar?P!%Ks{w)twrAbv75tQf{lEKY#LQlR
zdz_~^|9;-~zn_<X%lGYX^u3^pCr~IwOsUUbi&{k|i&jMdD=7w~AqH`5?ae<Qxc;xV
z?xRA^h>Gc@f=}=>v1{VVi&XgUXFG=n9C;v)3!;O4%AcqH=fBGx1DOoUNr_tD(SVcn
zs>gslg58can)nw*zL!So$j8(m#dWlid&rJ>6|9lbYEXcIdt~$_DLtGk%<V}5(AbSu
z4(~;BzkgroVF7S0Pobf;T>bblY>FGlNPLQTq>!YYnQ?zS#{PHYsGMg+B;)Jn_sW5c
z?%Y9Gbg9A(a%8q2a9-d{dQWxn117L)e}OeQY{*s^ylZz=B0#vS!0Xf?CQZJ+xxYkJ
z1z?Ur0ezLgK1cHW<m)iq-$y0?t@C$-{_nT{zq2Oy1ClX4k_5;=gkWX_z`aETg8U~w
z{v2pNNPE$!0%Qo_lg%(q2^T>&kUfK!449$~d3INqEVe!`FE2Ou5OyMx?}#xreB%Ba
ze>PdLTYg?186}RoKbRZ)(T1X*<KyE4936r%WNJI0Dv9>@vxi|s5D5_>TvXc;f<cf6
zEfTq?0KhPzo)gtP0c;m#6pnIrRL5{>d?9TM@bmNY_ov>p$!%fklCrXo#2~JI!#bHV
zm@DT>cULnj4AWsXkSR!WFY3~d)UKn#kE85IFK9D7R=Nw=qa#u-B#+d~At=G7s>hmH
zS;g9rQP*{vwP*qXI)UB?AQO-U$k#}%y(x!HU$q?uq<+Ye_19LrKT%hH1KLUmC}`6u
zj%<ng=Mb_%2>he)WTsu&MdTnaVSJoo;wKc#(TLp(OfmWQ^&9P2(+|i5$(#*j8b}DC
zuy+bj!>c83JWAEpKQ`;IfH5jWNRhCLc?AXf1%e+4(2ElM#gRL|;c<Y3P5_G~e65m1
z{2kl5Kj!B~=aJNp7|&}9^f_W|!|@5CK=Y<-6}T0y=89<<?e1*&^|TAVTI8=@M<HFo
zgJh48In&4jf%pL8fJr{)x8n&gDWo^{Dc-+|5jCS*YXzPKm<yf=DK7w-97X;6L=zGm
z+S)P0_Kprr=mZt)5*E1K9Runps{R@QJ1=CR;MSeGY^AT?ftmrRbM{b0o&OBP34@>9
zpvR0PthPb6<TW6{k=|Y+YKjg#F7VC^s^7miK~4+PuHBi5#12KC$=TV)58ny(1|>iF
zt?1kYUPA>r7Qho&mM?7<?(&l}Gag7vfdB|Qkb5y@`D2{kOaMbgPVgeaLW-C7DKKn6
zo;qKaA*+XXekSkg1(ZBdql$WT6qVgw16lOEI#uQ69C&WPM<OCqz}C<j$0)U#mZ7Hs
z^=^<%05;NW*gybX4E%{;tx~`?2QULKU<)kD*z4Q3Z|K2nsjN)J!SJJl9}jOjumCb1
z0Y`L(zJ2=umJZr|TA(j(D+mfS4gnx46fez(u99Wk_h(n!=qCbZAi?;>La_n>S1dMV
z@)sX~H~=SrEvZL`*<87T4!Sp})rwktKmk}IGY6RWD5^bpm3W5XLv6rZ@uxMlk^A+G
zeKd(?E;bCme)fR^gRMJv=Ay=gy$W{NYsE!C%_k3X&0&`|#!IDseE{uKD57+v5hDRa
zo!(OhH31Zgnu`sBF@Fp~MPT;wbq#WpR=P1xRb}Nlka!8B8@({dfA|`XJ2^VWV<!QW
zeAVZJXFG78<!}P#4!Nm$F5xJzHnz6v{(K5O1tatM?{}#gCx9K&dT;ji^+jUj_vqRa
zK+rs6hL4V{)75Rofw;t~TWarty-cB^Ag|MpqYD5+@p4B;$K5-3s`z#ZJ+`>&=%|hs
z%wGUD&9YW40BQQk%+Ac5{?ZSN<X4y!+f#(CZt0m89v+Uf7301W9~b-VT;XZA?AuN;
z_mjKN<3<k~>Z0!g56f%giW(XUZdv1J@D9nKKa`nJu(IqCz|jDm2$hCs*X>^O`0e1a
zgPqFp=Y9;l8Kb@4ryY=xAa8D>h7f^fn8yKd^HeMYD~&buJoV(Fdv#!o82F;#qdn>2
zw6d})XIvl~K=yBRvB|3aRdP;FPEQIFSJMk!SFu+hF%8@#Ci4oKH6Wd52&}|^EDtM#
zZ5DNs$PFS26y^o9B)%vGe_NE&f)n58g%Xey$lp9NS~}?IFQ0(fj{TZ)7jX>_Fn!yw
z4%bw2WzPSu3@9q0BL$ZH;`f`VaCCtNnB_#|0zo)#`7D1bai{xfk!z!85<>wuqAaRv
zsNHLJ7%&WaV)h6&*6B>)U}9p0sv!$2YewdjmD!!UcgLW~Ypabz3l|f<n-X!)jlV(L
z4KFC18fKzjbq1Ndp5qR*Pl5H9e%T$i1qcpG$AutD1@s@sVp+}T3;|zyj)ESlBskHh
zVvV~olhox0L;1K9U`Lq0ls07`YW<ljAmP>RQ+xoWW#f5ZTs^(L^){!!K)Uq#nHL|2
z$&d~8Klc&$uICq?pTaIwKjs$>@cz0XA;%v&ITw$3g1Clqc1lu`4+o(n7C5E@)GwOA
z)&mTPHVrrzaL*yNd}Kcq?%$(4JqbOqj85$NH*XGKs>X`pV6iuH3FD*!&Z;sP>MkuU
zx(cKcmZU-R^)Xb^estud7fFL0j2iwx&Luy@>}i)z#xWrvykY(BPgd~e9ZD@bYN7p>
z^<yjQ`G{5hVxgBJ0&TWS)aSO|!6tZeQT>TtR!WD+RZ%EAgj9T3Coc|D59dwCS4;bZ
zi<1*R_mrWw?PnMAAi5a>8P?YnBgs4g7>o(KVnaEgs~<jSPR59ORrDkSX>WHri*;cj
zJ!A;>Pr_xU$B>z@AoBG0!+XHnCH4s5g~)iW6!oNFQrm~Aqm=8{qiL|V7PD{!%NP98
zqj=Z91#s35xH0iadHJyi(k_p$iC;$DOs~|V`2PKOvngmZM%2*le@Gq0mc(2Vd>CvY
zi5AaaYv{$XZy$Gr`FEjEkSKNbLs!ZfHMR~zOSQ94b5bl&)>G8}x*0`;lr!D9f}!(Y
zW@b%oKuR4zMrh;f8j=nlv(B(6@DKpx?JyZa415kgKyg<cce$LGJX%n%v68Imf3~I9
zojcqSZqCkCVG&W4U%#HqT*BFBazmTo(aa*!mc^=nS+!7quB$@P<sYu{jl;ms!y9T>
zP)`J_&i@CtAJ+42Qi<vUO4jSk%g_;p{>Urfy)s&UkoMcMl|kXyvHrs(fIoluPeS?f
zCAKtA1gb#Qv|h1F{A|0nZF`idd*;lUyZYLu*@kJ{JUrD?ifiRj&y)j0!k-F+2~4GB
zUwInU)DNB}h@vU1RVc4xgMk}-x>FKm(VTJDh(BkVGpCdn7kjnluX}ulu7*#CNKRz1
z<$3&%eQE>m((5HpsHxRzN}=R};}#Zq&*dC;9)yMkGJ+qKA43EDb$55Xc9Z!ORK$=M
zfCAS&^;zf%vO>Rp*Wwj$1|&X{SNAtF**}W(&hlVCh&;Rl0}b3cF8v5OnB1OlnCn`p
zhv+BCCSGum=V0r~G*Nm#ItMa@6v}d!H&anbqNUIGC+HgspkN^S3B3XzGlZE@fM+A#
z3dF)78T&b{TPwd|!v-IJ|GLZsT=Z{ndj0}}8*n2}cIqJ9!pua|>ODJZQ7aaiIr(J)
zZ?Vp)ni`)R%4PpLGxOknKCIqFm`;SBDI$iTMWJaO$EH@y%gZYV-O^ZC#$T`IVG9M+
z7X|uzd&9zHNFwxKlmUs4C%I(2C{BCsuzG*v1EU13WrSFM_c425u&*!avko%j9krkI
zYyUwJC5#>flMi3NE@OzA2rt$f^B#+N>Y-Ng0~{`tPb06m!j~Y2B=uOuEA@w9JRmip
zYO9T+lP?#sozMOv0WXW-?kU%<<+iB-sQ_jZBfIX~?EPyYP;$qFw%Dc=ynHm#)jHW;
zy$XZIy2AiBVE`ruomjlME&JC7(=Y@jC1}n{!$!Re`V3hWUOCXsc+_l(XeM?g2+JW?
zNM!=G83e;uu6YZqrhWt|zy;J<OF9oUt{}SVJL8FUVf*{;?QUa>AfQN3-iA<MdV2cn
z%L_-5f9~LOc6EjS`akTwcTiRPwl#=)%o#zFf+We3ksyd5NY1E;pyVhhAUPNi1wnES
zN)8GLn20J0l7oOm0VN7Z4wApIJoojh>aObUs;>U)d(^Fa@8JM@@3nqwhB3z&a|7WL
zR<f{&2`jnxx3YQovy7Pk!{Ns^nbHAw2;Oi>mp5^yEDvuUJ48-S99EZQ=r@KQq;a0Y
z5vc-Op`)|2n)oD17$2gd<4b;Odh_N@N*G0>^BHc4x4Lm7`!WkVJ7EJO(n8=5kNiih
z@INFkA7QQp<)#-XE`GXTlI{SL3FH`1+fdQ*3J3_$(<h=Y!rUIEC<la`MZd8j!wk|F
zznPx|`wou87{%hTLn%TG+T|GeJyslARN2K)R5)T!!mo>ackdPf8`&`)WgCcrS8r)T
zyMs~7Nu{Wu$-DYx?uDor9G=Syf)fS{fqN%L9YmpbxVL>TO6k|2T+HrXK<M?vi{f>t
zL{YMk@CE+ch3UHEx<miU%NdoK_P2f0;z^!?`~#&z=Jt<J1Ynz5hz%nIu87it1EV^l
zIAS;$*%+EQz6OWvBnyjGd-ESfOvP2A`6|LT{;aN=3HTFM(Ll&e%qh+t0+o#BC<RX4
zf5Fp*GN|ogvVbfUulyS>wf3kmOs6(bW1+~&G)btI_uCtw@M~ekVZ1~-BsBCr48w4)
zb&DAB-d)K4J~t;QK!7Jkf~fo;f{q^w0Hqh@O%}6w`|#9MA08(3lbloGR3HYTTuV{^
z0YhF-m|;9wV9|vp989mD9!135WxC-%8A=r783^fT0pp-n3r#$3WN1i`udF+sUHpOD
z0}ohYYn!=whVKp3#*0wWLTIIP{rYtu(NK@_@^Vxl>FOWDDrD^6<1`aqU+C@%!MaUd
zMQa$8{+}i%IVdOy2RjnHJG5V5@hZy6HD!QtS&A|ppTX7DRRs%z@R7R$kmn&U{0bR_
zFlxoFS2m^kUZ`on+Xkx(1}|kUtBcvoETQ`awhm!J59UgBb%SxDxA#AYLa2bD+Q|n^
z7NFqQ1Sg^-IQD^*=mRAu$*HIc{~GvQysWEBFlKN7(53t<p2j<R<m<cpz=1MEW<1jT
zU3P|sKS7Sd=QRQc0K#t!=>Z=8pF19?2h&c2vX%Glov=&c71}7sByi+wYB4xBsQCdQ
zAfeV9sf|=EBev6?+_CQA;Q?Z;QfB54Fk38RiBJF<C5H{1JQVR}R8%}y7jI~3X&D<2
zqVfdhNDU;mnt<6()c&&jwDlZ`oj!CoU{jeABNXDR<RH8T3AQ6}sLlufqMgL||KH4H
z{)GbnzyI{~W=C=3JJy^fuRqe^2NaGRI(In5ob%4EB+tC_dCBUn9aaK)pN}fGQW_~w
zG@n0mzw$EUo=4B}9+Pn#)*?Hcx>s)Jxvt)W!yENh8}_EAk>67(C8jR$xD8K4->@H!
zvCko3y}w^42zB*;f1Ug9VGq8!>)$~CpMNd+FKF;zKYZ`sK>MG6Jmme~6C<3H6aSX;
z|M{2y{QHRf=b!%PCI5H1uo|`#J4hs3_|OR;e3s?u5{1j=aY&|xZ}yWEmLduQE7;#h
zOPmZDXJ_ZI@Nhv#j%SI)dD%ROjsv-<(uTmP*PNnaVrFe@o0GP?5{2?~v%0p7YvAy%
zssCe8`9H@F*X94OhijqI27g_k4C5>mC||umF0KCjIptP&OHb>!Z{R}7F_D@oqP7F0
zfhJF$WSTArc{3rQp%+tw)XdDz%fGjn%RyG?;^IOk1B1iD!a_ku`<Qqb4uLsb@K#t<
zl)z>4y0(#QZfR?4!vw*=Ht*FdhnqLeUo1-U|BWwd0K2AVW)e3gV6%WKf%>g@eLWmX
zctMa^j6m1x-MncuJ39+>PghF|77%IApLb)gExyRf$z{e~WM$O=;e~KAGm{r?Td-1Y
zPI{0}{o)<gLx4t%EF2R<4Xxs~ZTFG+LAGMmFawJoJTpKp^76i@((%So(Gb5gHa3PL
z6CYr9)(#(=kU(;9;Fk+I!sM9XroMgqc4lT~7#R8Z_#8cY6p5K{G9ftl<Jy}ReAE@V
zI97d6*o*>(grW#c4_q)x8k*SXX#B3AW3r?LYA~S4GJZ)A&ClWq5+W`b*lfqndd4kZ
zQdET7_@#gjS~bz_jcO&9mX;ctnvWkp#)rlATbQ2*&|X(xAKqqKF*Hnk9i^bqBs0N>
zLM8uaIvGy*&=tTnq^s)+-n+E4G=L!Owv&;fNm*I!xHc>-=_MsSO-(N$q}Z`z$Nruj
zsHD*Z_n^G#G_<h+`Lbi0>#pfuq!4p+ylGUJDVw0Yz9smqizEU$pur*OzkKWPyQ{6S
zG4Sm)iUG(MtAArMfXSScs=B_uCw40}$oe-}k$@G4-U+BDc>nIJi(Dh>xT+T}d_ht3
zgo?isZSbO{qJjdXe@NnldJS*Wv^Xov5(H@&Uh|^(#p;5gW75ZQvmZe3fHE4TrHDwh
z_?LT0ta$S%v6iqK6Z%EmLR>Jui|Ej4C@G!6_QGT;DT!fG8bB-Rr6@iV1yoAmj+D36
zwY6ccPN*#bhw`H+&(9B8&%lHLE24{wSE28~8WP4@lU~T~b5Pr$xxxP{14jue!MHug
zj~&DK4j9-xFap0zZEtO_VPb${kw+ItOaL{Wsl*8ZZtf{GtMj@S_4OM(tI+Af+6g38
zI$*dHPgP0!ANl!--un5ETP+G+PA)E&y%%I<^BYa}fg%cVc4B%}O^xf!&+TSLVDDj!
zrna<r??srik0IVkFDfX^VB(^pszd@b$V`ll0RX>6XX=j)7(W`X;bNio!|0DNmjc^;
zc5Y5&JFsp536KwA0~UZy9Y$;19h9Z}_U{jP{FsA{?Z=s{)KoA}CdS9bfX))C;vYO@
z)1Xwo-`o&8b7l?gG+@9wNBntk!N5^PagU3C;Ykb*C{~1gSS1{ZA2gJM_Bq=9mV3??
zZo<Iayw+weh){z7wVOmhk$?%OY|F8;d~7unP|BAx#LCOD)n74bc+6=wS@`)2@?sea
zy*4H0(&El~mF2~@l)Sj|+HX$q3@B%6KY)-;d)h<jR$E*9VA0U#YTj+k?>!IfVj}e&
z1XPF@$eg%e5B~f7?2w|4r34yJbl%Q4@XQkT6JJxGjNLGt1xxVN>(?e{jPY2Y#H`h@
zS5oqC)s?_80!GaZA36LWE{sm8>+6Z+LH~dNTzW2zAGrFc{Fzo|-USA#LgnSm2<>xu
zrxlKzf(MmxzUR7Yuk=kjJIgB5u*gV^JK6w)jafn2?+&PIhmCIN7-bdG<Xsyt)Uify
zWf+3xu(Mj>eXMYw;2Cai?nMd(Jw0(QF6E4uXq~6pr!n_E;^K#9HlW7@)We0F0w;o7
z8>ti+f`K7d0{E^Jz3t$DBDeec_gJ)M%GepYC~pfW{yj8At(Le{6r7h&3=E#Zz4P+g
z?CI&*A*`?A>bipaU`$2P2Ll}{vD?<Lt{}^LnUZqe_zPOEyuDVImT2T<?`{hZZ896c
zw8rXBrR05(<(@@LQI_5!EJ)<z%@51ka*BwE#C7I^O{cA`Ej9F@e3VRv@YgZt_HL$0
zs1jrT+|ls|9qZ3FoZG95d$8o7DJa-s(YDwh0T(sC(~KpsA8t3=D4aVJGqclmGYU*0
zP+u0&^2w+J_&$>L5*YIBY-LqdZZIWF;^S_uP*YRSpTFzVN>)CK5#R0Gw}%8kI{kor
z(9+V<(D0<rmsq?<glh|Z&2h>q1At`?t7y!ri@8X5myZUpaB}+3X~*yw+>>)j<Pnuo
zN9TE@Yp=olj*E0=8PJQ=@D{nFQ{LZF8ZG8zqt`8~-il}b{1_hC>=t2<7|cJB1hyUH
z(}j5&Cif<`gU~;WX`MAI-}BI!avfk(+$^9MjftB>cX?drS_i*!p=1)EBajL%{CM<f
z+k^>5j}AZ+8ka9;q&S{qtg5Z0IqP}G%u}En4c3@|$pVyTN?Ok?vNAJae!BsX!F!MQ
zg9ib}#Uh;Y&k`HL*PvZV=8q84fu;73fQ_p3SFc*z+fPHvm{NyLTooH{5mWK@YAio|
zVGt4NY-tJAQPP!>F~OV*kpgKA0rxRz?+OZ-(C*r|&$v;k52HI+D&S$D&TJ#0OGTQv
z`G!zrmSw$+_;f*eZ69v6Afyd^L-AUeS-~%@xIFtgH|818?Z-7G<B@{>K&DkZP?Q(~
zHvgR!i{aG1W5O>1DI|=+If3}}J;kfb%W!=6DIEe5^V5T##N;m<^Vu!bj?m6NDwR=I
zz5)KjY}-Zf{P#&4(LE;8DDEW?1HdBWIv#jPe*r}=IS4^6wmydU;33Dh;1NWuIa5s$
z&$V}I4;GXma|4+d&jRueQ{fF0lOYV(KCh}782rFsHS78F>W7Ga2$Ie_c}_Ah5gvMz
zrW{hr%GoGu0onkOMZ6%F|5bo_5Y;%YuoD8@{JeGdn{2gc0I0B!yOI)uS*N`OE!ucI
zi1Ge6{v6DB@#2`6BPV}PcX>GjAKz@?*^TMNMbk#@_HW-HdY=cB04HsVlgw7vulK{Y
zu6_E(jz4lTGBUR!jBYtNpo$60eS`D@!5>Gm-ZS?-35l>&82I_qnT}5g6$IfPKzc>k
z1H+(_@Ap^IGBYhenny)8ILH(-W{iMQlFWeMN<3!0dJYZ_!~{KC_jID57NI{!u~yqY
zQG|ORBYewHMrQj|GHM%KL4028_MCVK>pCs1tZ>ZgbjH$Dw6wKFE-9fNu~T~dPtdl%
z{c*qmaQ{L7Sz;W)@OTmiCESdR(mx#Ue#MY>71$2yh`QQZ$kLIKRNy;>OAa{W!O08P
z-rc<0!x}6H(5gcJoA2<z01|U7Kmue+_(3ZroQdvnLez|-I=D~+BpSFXY6(twp#YKa
z_lFQjACLgxd$n23Ug<PeOCW3T`i%=fD1uoUKq34$zJXaLQ?3!llkK=GY@VEae7w9k
zeZ8o+_U_$FxVj^CjgAuREHE3O3iMvuGlv`VbGHE||E~d=z?l`rBw~WBj10VKKEnuO
zs_*GrcYuQ`Do@mIpq>K8WoVn3k<s7tY1LL2(mx12N8i`i*8}XJCvDYKS0`4eU`L4R
zaA`LpH<C7WQX&c6dtB+8Z>q#4?hqcs)tkjv1O-l+#3v$FD%rGt{fau_yp7EyijSw_
zZV?Fy#9a%mu(P+1zP7OAS!SjPwj%&TLfwifK}JRfSW+gYrgkUYc;V;lV!H`nzIdm5
z-!aXyv$MnC3ib?Pv9TNiC3lv{HH%R6f>aQqcL6aC5GNo!GgDIu=jl88@*BW|fqX-p
zm92Tk19t|&V$Yk<xSK8j&VohaNop7xU)>_PBdZ0&(3qGjSFc)IT6O_z(fM}fFydKM
zM8rjPbps2FOu32t-8Fdhs;V$_z*C630x@z#F?+?eX=GKr-H3<?YAUMe*x1L$AHtB?
z;2!`(9t(4Gq-oj7tUY`8LSG1v93a13oSbA9Aq)^P3hT-S(q_wHel!Fyh=O_~X_xPL
zsrQCR_$a~wcv^yN&~%7i1gADNJ^ei_fN&pS#0BGnCknV4cr5np*@I_7Nm&`b2=)Ff
zH=Kd%tw6O4S(}0cIu0S@szZ<up>YK(w;>q~!{8}I9+ZD<X33zv-Mq;Y8j%OI7{S+N
zV@<>@bDLihmfJyEC2l{gw0uxj2o)&APN>iNOT#dqvobdy1-e;H&W+E5CAJQ?Z+Ft5
z&Ihbt!<Lzq<vQ6TVp^z<!VD2bR9xHxb?by7sjo3;fru-b^Fb+aaD;Ki6ylb1`qs-=
zu1t31CF3msFl44vR#qNL;Y^Q?i2<#GSYR{w_8AI6a2II#jH{q2_~erD=FK8z5+!*%
zy1)#8wyL_<4th7@X$ars##3Vus79>!FsYsv!vr4J4DNMmWZJsA#fBAo8c4d+GK-2t
z(av*l;FeZ(b;ZoRUg?1R7i1jboZ?|X3^{r7<ZPR!t}Z5O-7pkJ+XdezQRdB5?Ms({
z7>aUptLo`}hYxQ}jlv@HF=s2yn&Jq|ph81|02(4pV8b}yT;>KD85t-I_Co9(Z(T4!
zM}TEfj!BVYE~%-hA@1Sa9NILjp0O}AWLSFWcL_d<h-S*l0qE@9^=5Fuk;{fhMk3LU
z;fUd+Mn|*dJ-D6P!>{G)D!NQ&)lMVeMUGzkFBc#OJGHjX<@)-{OG-+j<HEp0n3=}3
z|4~}nFmQB;ITUgxF?TAn?UO*Y3Pdzm41TA+S5H1%3IY^Tm5_`WTnLyJp*8e-l#j?D
z(IvTtkjdb+H=wOCe)|RzZfL0SMJVu|Jc&r%!1%jhPNMm}mfgNT2r96hF>ne*aJql*
z-uVcw?P^bC>+24X1f=<L^Ws5sAK2tk41FqMpN7(B;u`{oB_z;j*%(wWT+~)q&%?J-
zO65I=Vig7KZFTivIc<1FC}}f1mPBcVOP<DLc8<Vfky>7UqC$C4e9n1etx#@V3ge2r
z>mRq~r`uRtPmGLEK#2e9#K9@j<k`~xSGasg?Tt~qpDHVTS|~wcf|3EN1^p%xVq$7s
zGO^zX`PszArXh!LSj9ol&)tj&4xT}Ke6!cpFDWiA4zO|P)2G}<;tCkc@15Ptx{)tu
zeEFS*vhr`RJ>p6S^$lN@l&oP?hKI(yEsNaD38xyfd^DekagQzmwUF{GkG_2Zx>p!n
zJ1j0l;qD;Q5WbeLT}XXX^7A2BOhyDg$<4Wc-#++X&0we|G{_fwJOAbnsaX#~2d0R@
zQ3B^OMrT3HB8eI?>FH+cbOB1<`49Ns&-sJS63!_E;b?vXxMQ=632)vw0?@vHER+oD
z$C8I@u;mOT_z)bgJ5fLlWZy)g%HS3r5fQRf3JKBkro1*8##6k6gl6~`T65#Ryo`)T
z$2GA<^{7r-p6HJGQ=sp@L$6cHj7*M_W<pF9RV}eM9)<fK1Hdp%+NJKx8YRX!+j_HL
z$xQrKx?lx{*t}+JR8-LK_t34$$UOL?IgN2z7o_I_{{F&rv_KtD|M<E4S2N3^oy72~
zrpyW5`_JKF$Js%7ohu;;31v`J!C<x_o*z#kh(B;NK-Tn|$ocFP@EgTC6lDGTcB*S?
zf)Ry$@a)eQjbG48SGmNBqI$i}otaDpN!(fYKOZ_3Cz)kYkbwTw0S~m<9s?5-Vi6a5
zPleyQFz<dmH)AT{2N0m+z1P_h+01|dVr3wQXW~axyWLQQgJV(c#tskxe1P!0z?7>B
zXQcy<iG*-guPW9ac)^d{7G{6FoG`Olc$k-!<>$on4R^b`98sD<(Wc|&T~2w=a8^G^
z?(*yoD1{Xli!Ha}+NvoGAu0ixn1~5HCQ`n-)sLH3LAStkzZo~&#Zq5s%3d~n<pNav
zS+wO(0|ynYw<#&!g-02vsHt&TcZ(lS&#J3a>YYgxfVbi3sCi;mhu+hB_qGNb0f-?N
zl?~ozfK6~|uP8#svC-mdYb#mq8^qeWirA{>DOGz<1<S!m`D>!|&Y1{vD=S=p%NaWs
zMg355Bb72lUdDt)N$H~8jT^twD7#$n=fe&b7li*rYNYOHj#L!ZkIu$(!tbckg4i9`
zTAJe1%;rZeV#j30y<9$upg|#TQLlFC62VNV3a7WNbQCqs!=VV?G63-Ii0uyy%h9f+
zzG(uRS%uzr$0wT0Ou5CV>z#L!!=3Qai-MOggPeGrQAZIc8yD?FcUo$yOz#*apYbjo
zJ0zcyHkft=aMsbJ!{n!}tt8YLO5>rN(4~9!=%atSVS`+30kOn;$VOUE@FHbtQBhY@
zQ;<s$l5)Gn65<2Tswb)z<GYn(lsC72Mv@@ci6t!A8Jvzpw_mVan}K#OMUk3@rr>Sp
zz|jYU72nzIFr#_EIj{$LwYqzMe~AKes1*G{za-2$PmtbHQB;f-un1}0{sHb2GM?7q
ze{!0E;Zz3FaVIE#M2P`5Qicx-KOPSnJ{7K=d@k+4GEyg*Yg0a;ZMd}<!swp>Q^PQQ
z|4$7KrvwDFQrKrJnLIw7;N=ZImh;gZ*%sZs+-u^l<rzNUw4cSWAJzh#3$JfD=RZvZ
z{q$VK8Mj4FosiRd{u-_W3LnMD>%M=F=cGD%R5oG_)5xk21Xm_SjDcU41wJ^f-2i`x
zFzvHuiLKx%%Ksz5pYPoaes*Q2A|BF+<Kf#4=7)C^i{3MzXJug%?PpZUY<?P-E)ku^
z?5riMs5GBtWSB!rm;Yi8Am${@fq^4xkfj1=L-+FI$HxPo00`~F5}?Js<!J*5zliM5
zU%W`!)YjC*2^vR$MX-Ve-G}q+0|Ns{FQ1=ltZnZDs;Z$uTFi`22lC#D?T=8e!Mv+l
ztUqi3bT!1J>e||5)|w9=o}sKje=wnLfl3b9o`QN-p03$@_Yoivy88Mmw(5Ape`_m5
zv?Kh7i=RE~2ZX_RB)bXY@u@j4VJfCU`(Iy~8;&+~cy-=7$_R^msIEf_bjW1XwOm}5
z(L@urT8=#z<N)#k5B@PW#xC%Wd=>I9@gka<j9UHp;1sT6>>M0uP0m0%fd@5H6`)To
zVpnMA6&gJyrG16LYm`hUPlnj*;oRz7y;}8s|Ni~pQCzUGts+Zea6SlAu-KB=+36qB
z`6*4xYP~$7w)^)_flFYl;pXA7jt;U`kHxlOXQnS(Rp3SY%kQ{_gd%i4zkpE%*0gPo
zy%G~!unbiW^n`K~QsS@Af5QW`>(i%X1hNUrKG1sM@5HdYt3XjTR_Hfe73<!kGCg(b
z6j*n5w{PEQN!wo_E-ZZ6JW;9=1JAQqKo}B2>X(G@;)aS0CT)wC2<sVuKiN%_z|far
zY7xb}KUzEaESwG&kFVf_!&U;Yp`y4zMc->=DqGtrygIK5>Pq5gfGEM4C#?gZR%uDe
zXb!gp;8LYN*%UhFWIw+>vsOtj%7}%`#x4djSHT}d5y`hPiT7s-ClpMIvsyHnn3xFb
z3YXdQ8IriM$jc8d+9)Y1&dtttq3`EhjtAI?gVXzuSEMl_H_hoMJ?|h!QE8keE+y@6
zsB2(q&)TG960#lywp-O<b~F}0jjh-?InTKN;Yq&?+<@??ysVg7>@s(<FjziQ3z-LZ
z6I~e?BdKX=A>V0kAsDTKtrX=20MKRU_u2+P8KG*=k?iqo`nB#pp~v`eX3Mi+_w&ik
zzy3VBd-o$!W~cpbB&Uwh27+_*n%&izsN|5l8yos>flZ9-lNJReM@NTBCtPDkd^Vg6
zQ9a=yP}+j(rpy~S+T+$n^@|rP{Z9f}>MnNTPIrN91lTr8O+-?R!4G?dlevFnCY48l
z4y2w~Jd0NJ>sO6MX;fWM3tsoRW?^MD_k`IXP&U2qHHO{Tb8wG}9ViMpC1q8aA1xd_
z)Z9AwS5PN!+q$(Qo1KjfCVi**yWV1O4=)aGo!zVUxZ76~M?wH$lZyy!;6p?fF;1D!
zBL*ZCRNyR^UW%7Z2C#tncLXQ5HfUF<$Ex<YTu5FTYg3F6MVVZ^Y7%B!nh!XwdT~R*
z8xL?O7wq~@T>FIOXpcE9ZrtF~;<tUKc4pwhC!o|ny=BZ%^~8mgR7<n@fkjMiTMuvj
z<2q4J61TV2V&;Hl##L{?qLi8xMK$mYx#Js~#7&LzNc9Wa+OOZeyF3}uDsL1ZYu2|g
zzqFJ)J}D?DIFV<HBQs29{^H3N{+vP#Bph~rXlo0Pw@O8PL&q?2q#oT*O*c;kc3DHx
zXhR?BtgNXSYVC!%JTAa>l8-OMq6D{1P3@4`I|uCXL6|uA-ljzJNz|okbR-Rp{McWF
zRKvR+kqnO3EW<(xMMXtTat5M9DyZ7<pIZ4@H60xn01P;wJlW(iY8juMvN1ZsiBrl0
z1^il}-2?F$2`;Vxm-noG`ETEjtmhXzygQfhaeNRbi|m~Hv|#7H9i4A~%wHcDpeEwQ
zPfgMXyW+*Gn@7>omb}fA2TSiH2PZo_DOKgE^VhCDRT^r4H0J6PBjF*c(@{}jf@qAB
zgixzf<iJLia#2%rpnm=9WgO@Wv=!hLJMX0vvGeA%HZ%mZyaIp|y*|RvjOka~s3$#;
zJIG4hMtT672_>fR{aoDQUn3(+fMlvP>}6$ltj6YGN;Y^h#7&^+JOjO(j#RF@#|Lv{
z8Uu$rrelnIl3Q3TdaF#;)erN3(7ABo;pyyu(%BJrF(3)4Kz!{K`KN4sQ8(XTBPYTF
zckSX&Um+$Y?@VPTF5mMEzK<Ud)H7v#rnt%wJke7aTIW#bxIAlOI`IMMXZ6+z0$0VA
zFrJoT2sSo0?&#{`tH{wWRfsUiH5>i?n=7o17a{f6?&Uv!g0_dz$c`MRIH{Om*vsSU
z>@3_lzW=}h)5R=JN)OyFV}s#<X&7b#?~0M?eedO!FNJ_D4MKN)z4G8!-V-P8X#U`S
zzs8{T4i^aN5plM}LcrUmyY)F{Vx~1~060)}6o5h58(f9P@b2Ba#`{Z-xVf<GZE-%5
zIp^6t)<~c>(q$m0LDT}?5TN;*)jYU5?k*G)ne?x8Q<Y%|lKYNoDj{%oGdM1;uA~`d
z8o~4eRN;If1J6BSCuw(<uyCppMe~vJty+oHQ1jv&%}i<z|AvIK8;D^w>e67ZL{Wp5
z_rc{Ant`}8Jg{*ovG{^Z-0;sTjFEv(RPp&eY^JK~_dts~lV$wTxD`3~aN~|6Xo@K)
zF4&BeC;(snsq9?udBo)*wH#AOMh5HbceCdi8I-~|Pw4pqmceweb&+*8CvRs1HVw$J
z2A_75<gnc^8S-ha-nfrU?p)@aPvwuYR)t3$_y4<G=ciGunw4Vf2WbtX%AY@eoXgoF
z4FC&UQH6BU|FDlECQfx0q^re4*JJzueQe1CL^6!`kYYN-^@T9yLw7D(qPmcoW0nET
zuJ!ggIdFS=KXV5#i;KsfKAR{v_6BvBeEp${w4|hbIF$K_99z74G$t+Wq^)&)Jse)$
zUDA;(F==|~r1RR6d@h5*z`<_j2L?v$e;BF`5XI8pBRsGtnOGW=A>=3-a9W}vQ-R6N
zT3?^GCGfPMAhZ2>p-vvr79PAeol}?O6x=_WyuxOaB2sbjQp|Z9lH+)as*dE1(9W7p
z8Zn#zXi?Ih<PZ8guXeCbdN_@&S)+#8qtk&W$z)ZPlm_RkP)0<(%F2J(jM1g@Tq0Bo
zi@<c0c!MXB2J80)^Q@o{j{ab7ZOv(W+7&Fpi7s3TSXC$=9dAzFxvHlqFHePnu(gX?
zQd>K`=|=%J!OZDrC^Chh00<p&v0ZEe94Zr$UZ5&Osbfk5&zD{MU!{a9+yb_@OZ@oM
z9S^_`u3{3jgQAz1l$GlVWdo@+H%55Jf3qq^%#Dp%&T3p6<$U1~KrU;>ZQ!)_=az`&
z^)GMGi<G;(V)JfxG0WrWuorZ`L}8w9;4oZDgZJYu`0`pRAcNgQ8#ofChA$YLSW{gw
zwY49y#su`|Z}mYv4o*&E)5ZCzzH21|Up}+CxV@>_p9~J;!@2F))mOkNHtRzG?48SW
z<`V~^CJg2QafUFn`4Ri(qjGN`ra*AwJmq0$`ls2ud=$9Y#A{h(9jji$xN|EwC4;3c
z81uEFh)KJwrA0<Hlwus)&_OaNAq{qQB9OwLIvZ?$d^|i+SV7Pl{+ot!YM%-c_)IBq
z&Gq&53foFGGLX$Bur_UmIWz2WX^dR_E1U0Ej72UsgvZCm(yaCd5~oNasfj8};)H+i
z8T!+rBm%cFY3M(GRo;lj_ejlJRa0v#thiE}k^vzS?|t+=P<DcD`6WV#`Xs*|vTrD@
zo(;~DOkJEv=tlCZS6U^sn6UR&?68<Wb8BS^u@eYbJ{@4fUwBr2y@zUsL4flSF(Ez~
zNMHC*P<S{wKbPK&wLc(>AJ)y^z8$qq5Qf`?$}~uo1W7jAN*py3!(1Y!_&}(`#Z5pT
zegoDfuqWgF6RZ1wLR9j`qUEIN@odB?q@GC5%4O#`36Ge%T3a(SCsf#u_pFN^7_<Mz
z(8$!pq`ITrCowhkHIPCW4%uy&qYmwiRyGzl=zZwSVR0Tre|aUPI5DSFQ^}jv(u>p6
zj{*X)-s`Qq?{-$dtehP1bNR$N6zp%_3}Swio0cZqdrtvYthu~&dtLSR$GA6dS=9Gq
z3CuYNyNw^Eg>uM5R7a==OMkQ)^LoU>O}N5J7oAj-?IVyfj`YK5o<b*2I<ByA8C2oX
z_bx55FWF{u2n~-?#59UXznyf^=hlJZ4R)PHJ!-fjq!F1J={M0mE&IjzG*lH0>?bfN
zP*GBPVI>Y=lfUp5Ec;PB7=7ykZh`iYbjpdy(|n80z(sJ8DhIopTU%kFz0=l`Tjc@x
zk->*trZMjSq@(a18`A)a4sK1sbS@~>WkDgBKMW#AI3~5)+(camz_o!X&~gvtN;9au
z?1f<BD`i<@3se&qe{Vg{bs({*FlKwjG7n-@ZkH=L4d@f%r>1e$o3KHPTBmAiYZJ%V
zpC!Gpr;0ARq~uI$oFIueVGW}E9j>3}a}J?dkEVUuI@$rX&F5nCiPtu;P#^zVSNVA!
zc|5CwabPeR$<{kLnem7s>9w^Wr!5S<TTxOWK^mR?tZ(e&bnJAjkI#-!hNtl%;lMqe
z<~DgKQtcSGeJqjA8#BfXB~D$7z{{^xD(ss2o{WKZS9iC)kfpi#^huh+TLU`^cE#=j
zeZ+2Um#tkjp1QnE=19|2R!hWzK@AlkfD^C&5I`VW1^FBL(pec9gf}gmk(;u`dfkjl
zbgaxvLLZ|TJmIu={L@pfjb9TJ#H^-wAv4UbYF?5>hC1^q{w_?mbn#rfw`U-EfD6vh
z{zynb01g&ssI-5`q(3yZvUCa$d$L{~DLNTQA=b!UpmIBu!84+@t<A(n2Fi`g5vhA)
zccs}vk;xb~DlZF(X`DCvF{Kx8l9KM(pDj~~^48P1bo=6<p3#bsLz-ZwvSFoSh(~OH
z|MfK6<f>A4>dv?HUiM?>5sCQ|vmu#`DE6@iCd=I34NwJ6LPvC*!OES`-b0DqJb|2z
zV9@d{=8KcbV9<m4G;l)?ntV^p)q7K$iN?~?8C&4IzmZu`+C5Fg{Q}=P7RC5v*Xzd~
z?V;@`4A8@j_xFD&-e>x$yTI%qe_;f%hU%xqB;o95Wwi>N6s`nPYiwZufNb1^c}lPD
zIfV-cNhsK^11Z9=V|292E%OhWBZL-vp@_&x)i~kc@44M*7hH7cPxsPAZ@TyGpuTim
zgKrCofFGWK#}wvRFX~axC@Az%pFT7a=Q<%l-@<}TMm;Yt@3w`h(HJc3(Cv+?+u^^9
zP&dP`6OKPDeIFQ@%lcmdfdhpEy{YI<)&f0L2?ykZ+V!k8`UD;Cc63~JEKh<j*vdhK
zc)$>GamQA!rBTJ9Q5#>i^ghN+zeJ)f3p@&iQe7FRlBz1k%XGF1hv-Y30y{Jhbzu2!
zMs{|+F9$gV#aWM)u+k3dT=dQ%n(26MF`f~g+1l_0HMHLg<x7_$mty>!1QgDn|0h~5
z4*<l}d#F=3VAoL9Ls7ynD>EEFEnDZ(zG!_FZJZvM-QdUL5`57KR$_bjA~%=&wvY2R
z4sLE@#0`K2^J(@01&3E4PJxD9aLM)AuhCHq9MCk^UOu=iHusb+ucBuu)mXyp*2{E3
zR$4PHQPF||XDTNrC)W~V0MqX7XKce>?xx0+H{J^TrLh*(v+<Ayfs>65J3|>Y=mXbE
zjrtyOojMiaYE+C0M^NxJviYl5VICt<zf1>*hp{qay7xyinP+=cSwLzkGyg>VL<f)8
z`1p8>^!c-uW%W2C0J=JpSJ5xPoJxFg&-0(oVnP@oAEItO0}D-rry7mjzDtw4rh?oT
z8=En=s7kmAfE22L>e#SACrVqa>i|q+Y?1P{+QTO+J9}Yi3BA>s<sb~qva^39mg4XV
zuh)T;f!VkKE32Gwj^(z?umrdLT8^^@Y3&HZ-?Qu_`~-|E=+{tR0yY{sRZBpToR^7!
z2Z`djkEV|Y^k^eYt*x=OK;CrwqtBnuV<(1PL?zxmAe`E=t%M!}SX7i3a$S9_o0?6@
zS-#EXvi$<5z%_oB$+$DxjwjtR^Apz%YwOQ@iK}JNG7KcDbyMs}i7(NJ<F^S{UN2x+
z_yXJk)*HMp-4@1%&{(V*JEoTOBMt*MM0ZsP-%IDPxR409b7RGoFI*K4Pf_zzCKeUN
zlq$t(Er{Dct&7x4f9>N90p3r*W)lrd$gif230!Xbm3G}x9wn7St@ZKa56qw9es^|u
zYCL@uf9C;l9*@Y5)a8!cpch-u(~6Fc&ZoE-2Vf=CDFdD4yLW#uN0!1VEn7tbazv6f
zupL~V^>bf!PCro<KcOK3Em`u}uZZA)ssR>Az0Uw;4rt8?(}$rOH*mgg-xi!YoyvC1
z4mWf3^cB#4Ff!oac&u;pm@_&q?hKS+xOIM4bMrwu+(*SX2q{(EZ!qP+GO9~gJUaNC
z#b|N^0o3`5n%apA`zP|Mm3jPa3=HTOA_}W?^4Wmn#E96RER{2rk*Pr2i&SD|YWl6%
z$rg(e@Xy8ANFaPwa`DN2=9Tf7kPt{^g3*Oy8m{o$9xTMY;<*e{G3OZ!RPv#DIaITw
za-8vZH;54d1-+FpL$x2M+)w2+@;mg=XGa&hkW?cW>9+$Fw-d81yY1NL^LBP{(74xN
zLW=sG?I*RteG3?ph4Rj0q5)|Jj~D_Rnhk8>6r3ouA~FSwswXC`Q+5&y2-$G-?F2?l
zPuGRD4S)K>5smWiJb8j$w{AeSWWBC2k;;QPf?28XHDbRK8cm{tZQge18?^DzJx*`D
zJ1%_lj&GtS*-(G~Au=*!TieFvZx|OM1U`(`a|Ota`6o}{6&;<KQx?y5{a1XY0Q;e-
zb$7r}@ckyf6V`Q6$}*-4LM0F&{u&!Qb@}B9M#lTS8W%4*q9w%B_!p(+h5QOFsz$_J
z8o&4bc$Q<KnxfJpXOQ@E73Y%HwAQ74u*~p<)v2%|hLN}ILsV!WF)#b1-<b96ndG9J
ztGxVf+xMDG$Es<6RR=LEF**z}Ca}Xu`TWepW*o1?0uD?7zeX91pP_=qkOaDA<ERB_
zL#Qw-6DuzlI~Ag2z`YUlQ|JDew^HyXfCi!j2UIl3<$dSQ9fZV!f-iesFJk&%(&-&r
zP7hH|sLt$N+y)9#(tHdT_%2Q0*@M~aD~!@I+rSw(NG*WLzHh;IO!xt%N2Ffutcqje
zGrhm;oY%64g^8(3eVZHQ@!~fplT?n=j2Vr09}{)JB;FKBLFp_#Om}#Qub?DWY3WZ0
z!-V}zs6=dv^A}l$V4QCQ{9HaumFwn4sT!qTuv2zMhTP=DZ#s*<%Gz2h97oi}brFDC
z2yNOsG22ToB}9dYUQY8_%J7|wHTBs$CcB%+M5NokeGAzwi2~x&r%$N!s))*49~(i=
zo!gRr)6;X~<Ncjjz@l66<XHm{2EwMnj^06ZcO#JJ&rX1z#Fp2;BX58>*P)FmAfDIY
zOE*yHX0{VRhpepZpa+P<rT7T;La-PlN`&P{D4ff^n?+iLfeRi++b+R2HD!tX#s9UZ
zwpMZU8>4Oh>%STM?}_rSBE4f&Qx$ps^<lii^2!Rv={U@bzzio0oN)dD$KeY}PXgN2
z(kJPPN=ox!6rC-``~&d^1v4g<IN4EF{+L;<v>jFb#Im4gMA2VjRTo|Wm(<ol9_t11
zC*d!iIw{kN)Iqa`gE(PQhV4#BI2^TQQbtDOF-p=A@;y?0+%ufV_@4BVw~uxP;Kjj*
zMx!<{HnxJjmv*$^M*T*Y*CFlcJjdyq;xQP%`LX!3j<jWa&UduMkm8ID4PiWkz3lZi
ziAWeA4n!NDMf5uD-z#viv!lZqEfR24B(f7M!+=DtSW=1$34LpCk1Toi`sK@Au#Ar(
zyLW!0L0~#+-3Rt)6`WL3Ik~KuavwfIkC|H-cd}e*mgX0qm5vo5lw7DvVMaWO@hv&h
z*^wikHWpaf+4t<)C5*U*q6zzmuH?$c!L{L~*QV#<Q2?%c_Y$NsxFX)8I`Gl;2;qtN
zmkSUb9$vwDhwbZIG`~iDq%U8+dZFoOkhw%drsd@ki}nj*^!$N9h^(1ZpOrYf9Z*g;
zlJ<DzK?4FexbbyObh2PNbnrS<Qe1OCSK0HayqsFtKOr&_g&ohy=cFjI8L{@`Oyph}
zby8pNtJ^OiokXkF4UdoGdF;j-t;I@ATT0<{tZ(!g@<gV-7Cd&-2UZ;I>=-_cG7Ov%
zvgxTR&LQPq)cYrKl8$+raqj6<x8wk!^ocpFQ+bl#q>A@^{<TCliR%|7;4^`09Q+|K
zj3#k$z~_cL1;EeGgCE~H{30#YGF{iZfho;X-^7CubznS?#_Q+rdZLGpkk7|){JD*!
z)0SgI&nP1Pa83O8%do`|(FIZkpm0dhtdXETAVN3>mY@PxdRwoE=lR2jJ4$qMAYC|&
zKX4+wwO*eGN-U3nz@YIK;`S9)QqxSujmyqdP+Y^Kepg8o>q3kiDdSJuU4mmRGzU60
z{z)0=8AZ9scNR?R4UF}ct`aR{aEf(YH3vQvfWATh2;`&Lv~8}lgS)5f*}h#K;^E`F
z^6n<Rh?Me}dpu4B-hFWULtMoYhhHI>ARG&zfB$mg^WjbTi7*OKlE49b-V*Z=*;k3D
zwZ1-JU#g(%muE!08kp#gijL+23J7&O27+jeQJ8|mH4IP8U-V-j=ioVj8O9&1bQsUv
zg6Xc(zV9DCT!%dg&NhIK!J;pSvg7!bl#3Zv#c8>@jmCPTZO_}$eHn{*`xxA~QD4oT
z^FoOcgIMqglKIz~z$pYl;Z$u|$u*Z?AHkbnWHanQh)1yO<gn;33;1cqYPU~}*_{k4
zadGbB1gl-FGR}gb1PT*F@nuj`(Q6nvb$btvjD!|efKlT;%mIPxt@2Tf!+<*kdz>Q^
zc~@8n4t8vT`smTaQ;R#wbQC=e1O)^*IXS8G_c!$ry=na}+l&X<(C`5W--w;qnt$Q(
zBUQ~I<-VREP?+Pu*@=%wdd7qid)3gsxr>?-apC35mnh=DeVK#j!PR2NfDJb|hTtY0
zWruq_#{kneOzs}#I?>V5HLlgWv`2XXWx<H@K}4vfGvpa@T%^8Cc*UMm4+&oZbl?*v
zUR4M34gn?M`)C4uXi7H-@{3=6En#+KQrIF(2NrrS+9qsbVHhBd!Hh=PEKeCh%gD%R
zcI2MDo+V_1KYz~7|I$>+!(a($l${VU1V0uc+y2b|B*$H7H41}E$YAqva$7y1*1_4A
z`1O)x|M8gVl#OXxJCF%6AffU}j+gLo1ufOMH6Qa-0Qj>*&oRtOp=dI-!Y0^b+8*cs
zo#7E9Cch*+9f8NiDplN=$;BRXOc~Y&^zwrH;0C#76OVXIl}?D^@q4XnGy%PDEx_0$
zYJ4mz0EY-n7X*6fF+*q*vBGW<*}1*NxEOm%SFn>R0Q(Fd?<*ju@~KYiFE%wYk`8n+
zi?rBNRdCjp<(veV_}C`=&+0!cMtwd`A3Q;DKhN-^^x@~?5SP*&ndn!s^tZBlp&8C0
zonO2JZuZHtnX&eE+evYJ!HvCXI4ULB8j2i-tQdM&*WP{(KG9UR-R-}E9?)h@B@Z8u
zVwfLjERKKMsOc5O;pBtILTzM^B-`}N);p%&zSDkW*DK>)oSkpdnjFkLTAd-OPUcZ@
zDoH<`);=RMGk{h+M%?tFN|ZjGp^;Jad(c$aHD-}IjVpKPYv|T`>WWNJhsFjbWnf#7
zPC|~swBx%@DrfRU;Ke4jN0%3m^IRt>oWVG<B(w#Er<@H80z#e5CC?2yKSj7$T#=Ci
z_KLiD8m{}j7og?v2t8pW>CI%Mv4J^4Z=Tj5`K~GNK0KL<l>xBNJYhC)l#DFsq0%sI
zaIpWwtGc?iG=G9-g0^kjRxNi8>Y?b#PVmbS8w%7Jf`cA%UPwB41$DHGcH?6wwljbs
z5^GCv>#)K7ZiyDHkaY*fGJ_4vy<uM_+Mg3d2rt4uKA&YL(gF&v_FwNYErhIGF~VZc
z@ok-rA&@u?1{@ha^toGAFt042tZV}x6}d@fa}`xRM9i+PIcS)gz4lS^NS@K&ZpQg$
zmrSo<n(f@$vVEb~tMP3jx~u@$<9?o`!^}YETj&XR@<g8520AT#kM45DJ$>qh{YEIg
zr{r{?{=k^|W9Og#kK6cMcCo>b7lUkKytf==r|Pl~EL7+F*#vz~Ja4vFy+4~CvzjxO
z+bQC379EkSt0-NO;*my-&jtVx3OO4BS;^p5b}|`qGP3E0D~y3x3dK8!R^I**RMgVF
zq@i+5NWctM=KY@AJKtIyI8I}3OiE%NLF~Kk?enJMH*H<gEQn0N$)>ZVa%7nTYvU(5
zpZq)wCMupMv5w)gF=*#$%E5!gj@IPb@%5GQT!w|0Q-gD~jAEh0?7rfl8XMwal*oN3
z8V1i~y`YE_?-ODM9X<p~X&z>R*5ES#h0e>EmI!pLtcSAF!TX8c2XViirsgM4sCNS8
zPrzVFO<SAU(|o^gUFu5!t(k5CRQx`FzHMw;>=cIaSyhNNRQ&Wyq<$gjE_^g+LE<Ut
zJcH!`U)ZW?S}-qXn%aV<;g$lB!h3pm_{rjt9*OL`r_>x%9@L%8>monIEq%}DV|8?X
z7nC--87j9B&VjTMMl~iTscO;i#!qx#hbH?6SwmgjV7&qr2F9n3qOcJYB%Vy;ZG>Dq
zf>@V1BsPfdgP1^p3Wla$$^_>hWA6sHs==K~48hY+o`G^rU#z5}a$suWA;dDiFM^h&
zy5MtCCcS}OcwKL8H!+;@Tl<DWnHF8q>8rgkZ|~KEFGck#X+4H#0LjYIz6wXj{{`av
z*G#|BCAMlHYb~s!tZth)SPkKmN$2x-@1n#v5W!MQDyW|ueMr#whQ*x0ArnWxiu4Ib
zB?P9XVpDRqnmC?%!!gVNhq>lZ05oj}6}O^0mviF80go?{0WRU*p37876vkuna&rA=
zY6cBrB|M7cm~qY~>a2-=`M3JCo>6WNTW`vaN}B!t{{CS_jJ6Z6OG@H2U!Y|g^S+NZ
zFL)tTQBZ@d4J{!^-UW}fFJ7Eg5e1KyMfHa9upc10S=!@n8hedEmzm&0XV8Cxe(G23
zQOE#IYRHPHN-kvgXwoYCnq?4yBMVRV2e#<Q$S>V=loIa3I$TrdU-k4{Ej+l4p0VV+
zo|DrOc#cKiD#fx5p!@<pIAMs3GuR{MmxOm%)vi-0et9?OMD=}rPwnq@00`810V?T)
zA=)($lMk^iUop*R3yO?nX}vOAw$=9F%Y<Fo6zko#?*%W@Jo`bPxtBzM8&d{YF`TmX
zVigvSD&+gso=YdP`xVT>lq71<$>Fe1A1D!Zoo6-EvkRFRXg5|@Q<F3m8$#wWT4}jm
zvrm_0!W5HBN!Ia6;O(%RJ4iJSnC%Ld!Qh#7rUPKX#oup;4m<89D683J@LoV|tw9zZ
z74-)=eZkbvU2Gg2Ev>DC-i-Wwd{MmDFzZFR!}sP|6?)H3r{k=I^PbjA6_m4YkO}0@
zN7^zMLT0zJlF!|20d<?-qeocPYKM9k9BA}?4)eeuCGLXI1lJRC%eMj;M`1beJ`93w
z0YS#W`wc(}{3M2c{xmc<zkKmxL}=(7fFN{#m0!NJFWgqcW7~$h3Zq^lbMp>_FZ_b2
zurS!aI4|I5mVteYx>rCT%*iyTuu#Nh@8|MzL*(5-lSPPLFn)sz+6kmrv=M7q$v$W@
zhv$jry54!;^Q0izVmEAJA_1@g_b-0;4kAyqO_7n2X2!<E92iGHDl-l=HA$Nud~8&z
z!Zd(XR`Ge@sscyUvyfIG#SiXcgIuxqv1BT&4S>=CXQ+h%55}#4EGbY7p?1cQ8geb9
zU_jQUrRNUxN+$?f9UU>d`wamE&${oyH-db8gzJ;smRX84ug#5mgP#R4F<q$hJ0VS6
zVHpPx*3{H=(DJr504AVzb;gIme@`swR!%g!i82jAQVIC6JK(C2_{gVDk{)29kM^xW
zPl}^7b4q@FW8H1>nI}EE+j%+^(SEs$^oo5x>r!o2eLf8|k|)KDfEd;#i0dx8@AWIO
zLc!vA`*tdgJzVxcv*i0o;k&eeQ*YcrjiLOWG7|(iV&GPnkE1a^JFAg$#V0tJ*f|pc
z_pbKictGin9<A(*N=WDhcFSG%ryOmvqT)EvKlm*w$;q8?f)@0aNA6`|RIx-9o2yV&
zRh?>1{sQC<9Bgwlv-IR-aEzbE$9sTWilf2KYw*NB%*_XwoS{UeHRd0X1H8d%6<Eze
z3j}Q@Du}AsmkCP=5ABYNyE+ied6H$^KWIP`>fU^rfs;&NfeEk%<7<;}^TSCc*3Mu?
zhYyc76QrFec^+#_j!n4vuBswny1@HXMIjpOZyXGzrxM@0x{97XJHf;>2qM^E87{8L
z(|_P3Nel%u8#~gJY3xcnu^t^T=FIVHW$yAPBZ8S##qUT_i})WT2}p5yBAfouer06^
zLmhiR2x7;_#=cgdP{TsA`qit*8Ag_?Xh5pBB8bH8deOxD#;X0vss7>NlYASBM=@Zb
z=D98j!Gam*HjFI8jt4MY(=aLR2WFr&l8I*>rOn_Z9-Bnl@BGPDVBWM;T+G$4r}Dxy
ztpEuyEea3Il-dJgB$T`i$GS$C_KUs(B^zZ9AzndSKX>R~mQry01zEHo1MaIRGE$%H
z3-=)CCx_qX9How5B5D<Git3MdNKFD|BjuPNIj|dfZ64J;I;dXtWJwTuYY-6ZQu2iO
z_!!RC2@E}v#rUOhNW$38bDt4J0Crl=7;pA=g-KBBO-e(e%+1BcJRY`sUs;`SPp+U*
zIwi+3Z}9wJfG~A@<uW}rxU5rzVv-QUh}5f-rDkWZqst7Ap-DCR;BS?t8jLn_u&_8I
zqYA=im#<(QOkb>MU?E&S!E0y3yjCW+wdUBpCi8hPMEJBZkUp2Po)4PnItI1?HXi8v
z0mU4?q^_qIe~c>YMf2!{<Ra#Ts&80MpN{lAwOGu_N3A#G6i{YX-Ac@D6KOHdxdDX%
zejkAbF?*x_u&7S!?r0|yAYK|vA-9N2#>UOfLSbPxa6FRhFy(cQy`Wa0YtNU$HNlm`
zB@Pwi()i>rdufHr@3hww|1K2s7ca_w%j{_$Eh)9>3?@fS$=6~$4oSps7}hb5Yps|&
zpya?nPUH%@Al7+lrYCQ^uWD#8+`cediwn<uVcVcj^Jx7M+2n1K{Idx381M6A|7LUZ
zEpR#k)ikxY!)t0UBK`jTCVpI(U*0J*Wa7G3b_N)p{&sUf+JrYDY?SbH0aFH=4F;R7
z=)h+b;}6or`F8B1lzh61evJq68Vm+dstzYXVRg@gidzMD38(`wQvhF}#9-cyXaN~O
zI3$WHDh&uPAcD*1ax187p_HuJl2Bs^McK&6g`JO3;{P5U6}I@w#Lb<H=L!m3fRC7I
zKk@X%Y}g*^d%Mvs7I5DNtn$6Y1+(mFqz?42Sm_Gg!wi-<@jf@lL=))89jC0^yu4eG
zn!SB{y7Wyi_&k_1fTq|CWE@j@@aEB;A<skT0kZOpDg6r;#8<rQiMT`v_~3IZ^;BHM
zy6e?16tR56&&;rmcLua5#E9&1c3BwNMT@&R#ybWFg2*N(M`TP);jXoNjJruj-vd}k
zKTkb2Itq|!alAb+e)$~05PhtRDNW0|1Y^eU=QX<NL>-_c%~m-g=m_38vX+TJb{P^9
zYV(-FTiras5QO<hQRYG^^*!i@`PKO=VIY%ot=mGMg(W<JfsF{607F2GMaDtYp&}#u
zEO8ir0+r$qZ00!(Ku2r`5Ef*1KqG)N5O9AXsbDSxh;$Q!DWqo}X<__@2R3V1JkP7?
zP8f|sfq)9c*~JBDxT%?0%l$NnouC~U{q@TQnIU*__0OM-tgNq$rx8PoAk{@SMIPn<
zcn-0g?+@!BVlo05zQFDUng&l+v2_)Qhp?4F=KT3J2$bp!cSDwD#8<+GdpQTUuhT2<
zaKfBm!DZ2s25Vn^5bivEL2E(1mzJJRY_y;8JpiJtAkGIs{icz}$gq%=;I4dWY-|K7
zj51DIIvI_wxcEAHQC#MvhhJ-ZpKE!)e}9&R#0g;*)16+6XLe1nT0n9q#Oy@_U?stv
zY7^DH=x-CKW~r&Wc%D09N91oXV^pk8@W(j!$r9_)puzuKoc8z~CJ)CzEuVCDV9bB>
z22+;PVg;!aNbB5XLib5M7TPX>*L*l;Pu+9Wg}{6BH_JJ;i`0I`7Akrgl6ozpFJQXf
z5@>859cg?T>FV^oukYHK;y(B$p3wgUB17kR(XQhIn3bF_q8=o<aX3G4+iMUQk9(zm
zrZ~gM2>Fg|2^B8E+VqcY{PU9T0<M6+K_@(^pV;)c&KW3Lbn1mi2??|!FZF&UH#3t~
zLt`tVT54QPI-@rxK&{6hx~Aan(2J3F^sQhGfTusv*&NMMzz<tsgr40pL};Q>eKm*`
zDS!P(HhFw>)EX06%X(<NBY6{vsiU)Omn>ZXlqsm{$a<TNi4tYPyPN@Dn4S>D24}^T
zn>%1TA{2MK+jP8=n*p4DU^{Es%DQ~usziVUb-d%x&$~M{y?|nsVrm{?NmFL@fz}LE
z9}Zgjw)XZwKR;3nFH_SrHBupB4L2TGf?B?EEwOo&)Z*;+zgK26=m3q7|I%AM9ehr{
zk_w~(T=0pf+k=}&6<pexNUJe^N<Arl9bR)Gf~<ct8GZZnUAuQ5?vq)>AM%AzFb$+@
zsfTRDg$dRIAC*n+_pe{_Dh=21^P*g^6Hxi(JDPZ~L|btE?{8po1>4cv)IPS*_6Uq9
zeH}!;C06F7>;A&&k)orNU-y?yCy2F`b+%wvGkDNs4$=f{pc<Ps89e^-nimgssmsdL
zi1a@s*LC6FisBM#)9Qy9m=i{O$j^_>;-Pk_M^EhX#REjLHB<*B=cGd(Kdwc6O!g~;
z51?Smld7iXW?-B0@rKE>uxLT4Mn=6qD>=CU+uP7%)FgOm!dwz@5-q^jZ{JSRp_cy*
zB<lUj7{`nPJd~kheg#$ZOtU5WKukmA6%|8^e8M$?cqh@*#@2nvcm%Mc_Vj2QBw-k!
z!t{|bde=Q=MMW$07LZ~oIj4RU784^xjRdKgTFJ>C?1T1ePy(VRJSA}A$&#h!<PSXz
zFY>+0fepNe5K*-DO=~&eT>rqpzBe~wt4!ViF2QaV6k`=F&Fo7-NUtVzx$P({acRC(
zR^n}HjoC)#<k&!hiDEK7F3!6q<N0%J_#DO^{@Jn^?pB3bxg?!bA7C!!8D~O?2h{9X
zwF{XOY!n+J0Qm~K^HYDv9337BufdX^vA>JU2@Ng67t4U&^UB|pxY95L@CUqgnhc|8
z4`N)uHJxcC2j-0H&U8dJgE(R8N?FP^&t(~%0=QxxbB?@d3Mmd0QrKq~l)po4FDwD;
z;ZIq~<TV1vj(|sx`jt7j&!d1y%76<7_M{JD&D){ERk*u!0aXQ|8#u1+;xOC2wd2Ho
zCeras+S-VV1;@ld?EyEJUsSZKu~GGma0fCO`a=RdWC_YVi16)>+u+pa!6Ffg_WK5i
z|Cp!Y87AMez`hiANfj0D;#40U6_p6)7QKRBJArLRa_M64Ei2!M{~1e6#&^&%aJ3;6
zVyZw4&F4eQaj2qs45R?zR~e_RxFv%r>A1P!0k}7K^%(G7v>uQB{jL9$)k8TUyt@Zw
zFOBsF2>~-=h0gx^nwl)|H|qLIOy(CBqWLs-nm`b7f6@xuO>4j2!6Cw3gHC@Cj01Fa
z=&+HW&|V-vX;2g-)nllIIuRcOq6rtktU#y01&EAnCm7RCtzfpM&@#%XLtrR}LT3Md
zV(N{i0Hn?+0qWZ^ZpBC-=!s$b!hks{4hyCOwT+GN1wnTyi1qdb1<02)0_H(-!rF!0
z*nrW7UIi!20Yek?Yvtuq;Jc!00Th5V!q1p~MqK<?vx~;Ti+Xy5m>Qx00-TMHkwhKQ
z)wKfB$Z`BN{}~0$57BIMVeuu76jbU!YZ_{6LFvVW6eFmKuc?pW$s?!O@l>~>ogx+U
zJ{LDPEM$Aw84?kJVY7B(H3OLpJoQW5mYx)-GmR39S4Cg%T~-5D_c=B_H}^e!070)`
zK{<$Xg^B`*00%oeo^2bf>Od<_a43d`m<0~qC$Qn+R3nT(W{#Rlknam#x}>QIf*;Xe
zqhmFJLjmexsL`QSg<%M6Qczj_h86@hl8%o+$n(wp$Ar};W&A+;zJk&izySgYh(H)h
zP~fhSlZS_ef!%?s0AH8~!Z-?YGi4^xI;f)T;oy!w2@=^dJZhL|;M`!ojgAHI5ej{1
zNsN~@f*x8!j|;3xRM2MzYEG=-I0$?Q`O^hZCw}GDEnbgSFf72%!xs(;q6UE#W9gn=
zdlek7!SPDQ$%Vup2?C;G*bTz_8>bxOPPkq&)aIUcsRFwUpDkNckrJl{t6$FJxVgKp
zA;m(i2mg%G<eZF*CWt88+5~xdVUP+Jm?TYOx$FX@FVusxuP+#Tpmv<fpSuRG0k)yf
zip}+zUcWB+RA36zc5ID=PI_g)7(v<8bQq=i_&CZ;-F8cww?@}dQ&3Vy#l?Y8Qw=dU
z>Wv7A)j&A;V~1qK;#-hoFmJ_9IAkuko+Gy+kRj4gLRSK=pk8HfZ-T@>!6bl`Jz+F~
z`8_@?F3>E_*wR!V|J#S-WOYR_ufZDywh8*OM@^cSBHj@+iSl0An9?$m9z-$2AK$1v
zwhScF-^Iy$@;9&&t8(w%=tJ-@*Hu-aqO*kR9&{pH(kR=f4<2mS3~arxy`~587<L)L
zCbJVeXfZPe!z?s9T3$7sa3@EffJzI-N*s2d0@q`c`&W|u5t3~&5A+<eK!w8%1;uj5
z^v1kzNdXE&)VHp7c6`f0txH81z~JR;>+4HlFM;F-9Vs|Jp!A^thm0ElqwCTnSFKAS
zB3fLJla!`t$J0HT4}f0^3hE&u;}eX!-v=N=Jl~w0tr$-7@>;;w16~k7by|fpfrV?8
z#wW7+&Cbu4fb)PLg8+}jhYgO%UT`?Tu|nw&E*<>3@IgRIVe55aMeEkW4;j9SrK<KW
z+gx2XOiK?Q7d^l8*(Tw8+4vONYPf{KacyjToQ9ejAXU>-*-U<wu<rXqqnL6cYeK*y
zB`pmF6OJWeR)ACx5P*gS@#^>3Sk|8_L&hUdlUOs4S70dAaCF&@VN0uy(ZPp1?^HHF
zlGmfY9RD?3^2kS(F%7U_I_fhoOr)$4M&kmG$uaRD^fgdoUQmiKBj8v(Mj_ImZf^L;
zs}ea=_~BaHrE4%Qquv^Qb_@mo{vMx`A+1Aar57ETn;Zn5I^94Y6=z%2I_UPyp?*5d
z3E`lD-U(C<pznFz+JP}<2QU6LDe^gDhKNPcSbK#os5R9^irAENN0#4Vv<ZW*{I~yX
z-(HH99>Fm!O6VLP^D{dl1ZilldA0gIe%$$%5{g$b#~1fRBx5C1M4f?|$y5$`a3bb_
zG6_REw6GUWxsi}idCb+Hj-!8t(V4nUgJ@6|Rl=<B!xxE(SZ(elSHFKj-O5M&au-VN
zU*V-xm~!apnTeKfe(CktDXTr?v5n4Eo;C5b%%czN49axj+s_?U?^rKC_g`E9ElJ7k
z>gwkxbWFIc`fgup@7?S8kNT}(@(S_@26k;S^*nmLmrYi3tiEhhbS{{6m=62QC%Q{`
z1=}}>FFCOo@GUpQ+p`|y6(#xn6*vFsYAx!l^aNM-ixtPBB@nWDu7U8h@ba1v!DvD(
z6uAbX+-j%U(*OhU)ay9m#LAD!h)en|sW-)T9=np@Hd#pUNo@+3dfnDDcZbEn*`>#!
zMY&KHsLeV`y)Rh_j7`DI9NZH_JplI;mJInG-esI7w`L`SLjpCwsnTQwnq;U&SuZZ!
zOgW5hpObJxK^I-9tMB?N6$!8se?x8ZtQ2-5&em|`4(3IF{Xsu5m1CCRa?<?u%BM<P
z+qhNzs^>73J_vWyedBD$0q1}}=r9?9oVI>%@2N-H;vWwGeA8{Sw>xPKZ@apt#!p-P
zLJZ^drCtDRcHI6DiI2Zu|82|rBHiKG-Q4TaiA|cS;^c3XHt{F)=9#<bpB@-deMj<v
z{s(SmqyAQjbi%y|jdXy4{Dmi=*m>cJ61mokB4%gLvk9J{)u|lZXkdE6J{4h4dLnyL
z<U_yEx^d#4Vo+IuOE2W4csc8U8fn(W?3FtHMsM*S#8z4S$vpO7kL>>0s97CN4z$yk
zi{!X8i3fozN6-G?l?#}l<GvBj97}IX!HGOO;LmS(ne+lRabqu|WJC3He4PD>8b~2^
zE?m8MSsI&n^luiz?jgjV#u`yf{u0$PATR*1Y50PJeLu<<k-kx}$#=w(j=b!2hpRU~
zZl4t2X_$l?VL<mLx%JrH_p_i&hS*mF7!k!HD<fT1HsP(;KKe){LOBEHY>{u;H_soz
zj+qNNpmQKQvwYJI%L0Z%ch|Ptn=5%Y$;0Qw<%Y=;$Z}NU><Vp%|Li$>V27bYKhGs_
zU*b-WjYF>6ZCI6_{6v&}r1&Ce2ihdln=ofO3EM-mJ;!&`{Wxp=CuhrtL=~Ye$A)3A
zG8C{P-2l`AkSpalJ4yfbc^KAL7X7oQ&S^i%f!mG+gw-d>9ZAD~J!y9_{oOG2Ki?QK
z(j&Qf(*^s-&#uq6T)wo0<d_63B&lC}Af4iYa%ZF1Mxu5efz0>z?$389kN+WJ13=cC
zzdySHt}@7}M{&tX`af{~XXE7E&;PEQJR}Fo2J!WO{YH(kiAhD#juShGX$-1s;;$Y0
zj?g|;y42gl1<sfykjEEeTS%&OaV7vB8R~6kPIZ5=m&7oc@Dn+;h2(@9-s7h$MA3w^
z9YwZ{<ip{Hq*H%i@&D$f|DWI7|NI1T4gYt&gf(A^@ag*>g~r8|Af%nyv4*+60T;6L
zcUXemuOv!d9NG%aV=xc4H6g}0dT_nIY1~<8Z-%u2a8P(`Rc_xPGJ-jc?XCJ9SdORD
ziul8N1ItCg$>6NqjWBiDzhDcttP%m~#Ew&#Xa0=IQV3@U8G|FI6W{YEc4s`gi|QJ)
zM2cE1w1qjKU*sv_F9j1PRB|8I-@Qg9S;42x+uMOB?cZ-(lPRw0A^ZasAoqX)@-7A<
z))@}GC}_ySXyq^q6=pA3Sm2k6fB8s(ctk(TVn^Dkcm4MmR19TM5h=v{eeq}ppZ@*S
z0;2#jdE4L=-|UMWz;~iBNms-OXOG7pEQA5RcgcSN?}iC!hvX1HBwgSzUr#*vHVL~A
zQ=ciJ27W=TC3tcb5W|o>OT5;jZJ?(h(rd$!z>s@4R@B%-`bKg{5xvlv$#(;LZuJ8R
zHo->>+xqv98d2yusM9}Tud(ETUA@;)Vqc%3HE?-ABdksIh)24>jCQx-E)*6LJ0c(D
z83l^J0S>WU^z(2M{tl57Ne;Pf_p7Ld7UJFQ>m(%44*d;RpL?-H=ZVD1D01oAK8(|9
zf!`_=F=2Diunlo<h&4p5XsF!}y!rcy<cSM*jBJ+b2@ZnihdW!C^{@wV@DQ<8ZRdl%
z+$Asn7klp>)^i%QeP@ha83qv|`xKH*q)23+T?r{diYO{6NhO0sC?Y8-%BB=0LP^n-
zHYkyj5~Y+<X)o>G&o%e+9Pj(@`^WpwdpvX8$9>N|W9s+&ey?k-bDis4=fb_d$FLmg
z8j`5DmIfbr<^TLQ)O8u^mEm3cP^{}H60aoG#EX3Py!dcs>{_%F1i~A7{4`UMKP7sc
z`k$9yHuWlR8R52a^g%w*f2pgF9EEumd0eh*UCdvK>_QwPzM}QtX=4;3Ko!iAXx|^q
z%m43{9Zd(k@$RcajJkI39l{PW#aw=YID214uDPb=N|R|}Rh0OCQ%8x@(r2>n*E-Ba
zZGq6)7QA$wmJKoz!}{D@b>-me8M#TZk`iBj>hg(sY6VJExbzOstkaNqR#Zmc;*4Fd
zVbhYz=S}@UiHS<a#HgEu^(@9s`Rk+|{#L_I{=Y~S|0f!AdnEo25xI{bGN0sYi7wN$
z6@n4X?cf|pz6061OBVIn%!m<3M*DjF(Z#;gwu_|@!~v*N?&xGi#a(p5uz{hQe%fvq
zZ(5!={pBh@>M{wvM<Z^DcvdfJ@xP5q1-#v-;`OUn{;F1+|EZk)tEBxz=|g(@93mhX
zZ4ks3Vb`T*5BA^xaTlW`CwD`^hHY^u^^Saj0dQNXvdhU~1&lRtY<-qG-_rv$q8GNM
zX$Lz!-rrh_%|vKY|9wHf!wfAeJd`7yhWgH)La75LJy(Uok>UdpJX(`>@ymsnnzCPv
zEb0Ftp_3?1#xPqsr=QRFWooE~s4|T#yytSl)xOMOeF%?08=m+`cR2CI-;TQl#f`WZ
zBKCIhN%6WqEvs7nQ|d%o{&yUck_YMwjtgYr;CcGgsY;X+)HyYA^%;EL&Gw({1U%|e
zCW?KE?YXCxX>0!m|3jxh_80ThCK*0RZLi@jzxXIV)wcADO0Bmv+Z8p(fSr7o96xl7
zk7CFh==`0Vz{RnFxFUR=uv>H?;Id`xcPg4WTaU|q<2SN9Z5L93Q;#kYdLCGP?nsnU
zb}Biy1XrtH6d|9eEV|wKOXBK6akpnR^d7FRqS+kFOm&pByIQt43%TGSK0DRFrX{q^
zNU8_IkjJpXO-%EatjYo^C3G}KaIv2-5%}_ABTL?`b$$l@DRFqgY5<k;(}uTkDR|X1
zf!S><)Or&w1mQ`ww>?~{XsbT*5Q=u3ST2+sfXn2Z8?rp-{tn^WjXOA6;&5B!OahVH
z^~$?Xs&DvgPIo-K(Lm}t>x9J*PIQ&h58u$Q(r}xQw(yA0?EZ{4<hzVsIxmKnN+ils
zSP;>|uaLWz$8p)$O2N24f6q9hHP9J|^WLGMvHbWT^m=ADiL*UK)N{K&VAZT^Fk-NQ
zCqK%{$dtXP&{23mc~g+@hC32AR4<l{aJGH(i#2U?ewPQ2i#1J|U+pTh`9P7Y%r>J%
zp@!(G8BA%(;*U|fbMo@eaB{^-sfj5IxcdJ5<hY-ST1n+lfS=$Y6{IK7pw!aRBKN@6
z1~!jEg6}`oWHAjyW*0WTdDlfkTIRpccAGk<dvZCpMWHO8nHmtrf)ZgGOZ?{iPh?FH
z4rov)_)K!?XbsAIJ?WmyTe5sybUg9}=z8wYXI%-~Cl03aQAD>G%IaIaG(-9kf(4}d
z$qu=P5`y$8EMOn^&!mTy%~Rr)u$nGCH5JWJ3hXqz-nvQWef)R|D*QKWMdEDcv1o3;
z$EKbv<l7DTh+-S!5mgy3`)e=Uo;Gb7hs8yM$4>g*Zd$`7!QZ5oe(I|$vsKN#3K1Og
zF7FsUc{p!;d;aH+rb5S>Qp;VScP<b#@4*}cGtQ)Zg)X4Uxhc<#e`a`fAJ6B%jOBKl
zh$Uy2t(!nI%rw_7*GNApjS;@j(|oAJ7Z1Pa=l2^Tb<*azUw0lA`b{@mmGL^xc$ND1
z>rmUNP58~s_0);xB$d8g;*i{da1RkkNsW5Y2Au1<$!wa!PE$PP7Bz_(b*?M*^1;IM
z57<iecDZsD>0a@c{~G>i0k%5^u+Z%^$Ui&QJmp&r3`*qr8G%~hf<#0pdHaFkZuG77
zn!mmreN=dGA`S1q>&r$>omv|b?BTw9%Y&V|A7Ur-UMs;!C8@9h%B+!<mK2wg<jx^b
zYcALa;eKiQR>GW!1px;uLfaDZ-hdZ0J8C`40vs)?-P9cepz6&M%EWaA%gn6%z}TK8
zjSAy4o$WO)R!^!PcD-9z&*aNoJ__aAmh3`ApE7jUA8^_+wG-E-n(ck`hu}?z*nRNU
z->inoAAO^NwIqJm3*B@~QCyvFJU>Lj&6_Z{xc!VUk&L|9L+-%Hx}TbDMSYF`g;2Kt
z@u8nu5v{wU<_by5NgQ0(c#f&q%_Pa*>;-cX>s_5~?e>@KpvJSpFhH(<!wHSXY`yt1
z2VPB0`O_$s*Eh^|t%5XcY$H$B2hKb&Ma(aBGaM~HYLqdjj&qonS%OOZF)^urL*ndw
zhq5a&jX%Tg#!j&L$paW<XKkuAoQ{oM37e@FT+<<6$gt#eNv#OXMB>1}9wi)*bdvy5
z)b@noHvj&mx^TF4p+i||bC#z(Llx!O4o+sY@q*-R7(6N4dzS+=NA5w0MNZdwx+S%;
zgAzaN>qQWJF)<>pe~t~w$nBHla7ovWwlYn(=s08}^b-}Ir)yo)QLz1SFD1iohKO<Q
zKk)5Nq@6xGKXcaC)2WN}`2&2qVEHe6AFS(^u79Rc(=Cb>6mY0Zs%~=#-{?Z_7$F>^
zad7l)=M<tyYo<%bbS`DVBa<F#r5E&AuxRtiE_dA>!x@?)F#aSowL=4u)Gr0r-uRZS
zx0)YNyX)4Kiz%0aJ?G-p8lzj|lqvSC6OMPXq|4zn5$6T(zoX74K{_g@YYUm~NUVF^
zxf54=amBD~d=c`L<adyEw0rYu35gbgEx(+die+M?-95p=Je-y`HD%ZUhgLzNYsWqA
zEDP*xE8Fg4G$?N_Lrmb}#Sb-!PlTk5Pix3)w8NH6{^R_)b9>Xq`INZ4uj?vlxb`}Q
zjc&fx)5aV!$DfVgiefA8I+k6rPrvxOGQv~PM3+?~XmfSD5(l$@IrpnEzm>Nqw0^1)
zb80bzR$hltj6?D{|0#FtGI^bwP`~f3pwV{TYaK1NH|pdfT0GTm%Xf+8T^&8BOvK)%
zIYwgMp%WrcnhK!70x07%@J_GVOQUl)%MQxrciR#J7tOI_`4@?Y#~y_MM2vZG>Csot
zKPtjq>VG=|R2i8}Es3Qlw2G@0C6kR|HTgP3by{bAU}x`-Yoh<2us-|)!p#sP-SfZ6
zKN<~h7P!v+#9?mc-F0uAX;mn?CM_{08n;b6)OPUq-V@eM*lm{_O10ARxv<Hf!m&1K
zK=A%gHZPS}{h8kB0P6AYeBNR&R@Uw5WYrzfS*O35Q>nN9k^QS#hPQE<e?N&YhyQ~^
z+JcM);uR=?wo-dt;W((1RwEZkI!`=kZeBt8b2ql)_E(SD2x+Q6n0Qt(wa|3ak95p(
zt;am$+;D>d4qN&5V_x=j)e=3gP1f?#l4w}OXnRISbK3a8w&sLR4VzaGM8h7uGbv(>
zRdZ+6h13JY)!RIGEjg)}aQZ5OG6l@Y)l4op@b%36=JlHOucm`v`>0!kKg0kSODlQ&
z15|`&fnHY6iZk9&nYaTf)I52}cxqiLD+MarzK-Zj{Td?P1V8f3@2~o&!B>u7&!{f0
za^GYe`WS@AteB>`KC?N?%db;A{au-y_s_roC@t=yuz9}h4!M1=re4$8$()RFm591g
zedEaO|Hh(K8iHk=q$|yRIW=Rsji(FA0bEqgt)Uc{rwV>yxM}h6!LIpkEVY<x<HW84
z?%aCvDhYF8lVxF?T1{z!RBZB2@)~3oK@e>)l<|z+ZF?Sl8l-XnpqDb-ep5y&C>tBR
z^Yf41TH)K@;1!9dzse!bGX(aIRZd%2H7gRp{Cs`TjbeptSM=&TA9Ara+h#3N%MXMm
zJH$uQU2=#ddTx6Ghid+Go=A$gCpL?LOcg}hSHWIEyQdx``Y*Y%kC*5WQKk1-X>t9&
zmL8cE8WO|0*Zt}yY25Z$B6;5FKJ#kRXD=2L%^Mt1nvnXnuDv03$mTRM8)BAe@Mf?^
zCvvUfYW1LrBNzYh^z_7tY<tNl6vu}MkYn<h0wZ=CJYKRJgP~70YP{cw$Hnf9T2EI}
zQmXjHv!FCwv|diS3r@Yk^l8sE7Qm#;uU`^B;wPYY;qVylNl1;UE!qTrTE!@1sZz6v
z?Ss)OJS8ql$qLQ%)#v9PwJd4hOE2C})8$afTUDOEKVU#-s;ftq_>B~zslAxR@B8?%
zc0ZE_uuV6kRUZ~pCP-t(Hge4cpE%0nDVI@LQ8$s-z2Aw#mr)irLqFm(#w?*tIhXXF
zY*$ETp!mgFoxe&wra_<&Q+tfio9rm0dZzl?+UbzZBK1T@$H8Xs*O(Lm3Pw2noL^)v
z2g5SoJ|K>E$iEj~i)DP-iPNS}XO+x3WK4N>qT>Us$g^@h_8qCPtX15(C#5j|rw!AF
z+ZXg}?9#^b7YKg2@>7+lO8$*g<()@C8SDK54iujhlz(__WrP^ppx#mX_hZ=F(EhA`
zK>+wpfaC(r!WbjRs%YqZ<1_rP+|7<vh*`FjBz7k7g$0Ju{VW+<bX?Yb@0pGeY-IXn
zuRAf^vWtX!#{X0=kXiv?wa6Fz@y|4JjL%RePAo3YB1}A~eIbi9SG+bbD`>zOqntpe
zy0jmgg1*I&92xfC`H>N$0=__tBGI&)u8+3zDD*przTfak9C~!7?<Pb-T!-NOn4y=2
zKNxV6FcXs?^ogXo;cB}nf2jHP>s%3V?U20ZNzM`dS~YVap$U5EwrAPFbj4c>hc}_j
zapA1%KEVz9H{ehD`l~o`zC=eOySHdPdA5Dl|8Zn}RbN*vF;BfLaJa-sFX*kq3o<Pt
zo?dLN{-mRjO{GMS#_g|(4c~%ji2&3io;H1?01D)J(;1veX1+Q-JKSKpXCbnUCTg(P
zj>m~qTbyc>9d>^-h~xKv6h;?MULL*Mj%z@s$wPEWOZN1=th7^XSUN<iU*tbWUUbs<
z*Go9s7W&yuk$qMEG)ksUu0ufn*~4jq{@nszy7Yr5-^X~)se6>rey7|_?JDQBo1JTW
zN)HKz;exg4qk%#i{ohN+SDa?>_qxX|t7jD*qjLXi$h5ILo$K39#O^#`2^~p^V5DP(
z3~ZqOCUIy6#UB2{fj5S}KZ;^2^)B&sp@u0x%wfggY}uI`i{!-`mTbE!I=w-6@9Fs=
zARm(Ljcg@w$dS1zJgVb!JkneHz4WAfS?#Ct-P5JbB)jlQb!;V#wTZ=wb&Ab$w_{N6
z-VsY8&0*36|MhItcStBM@PzckBiG$t>&vKxU=cSO9uh^=H8(lQigqlnpqLEb&LE)2
zH|A^voQUb}c<_CQ$&n-9Qy0b7zmt9V(5^^6v$O36R%41G69Y#~>bX%xb4j71Q@`ey
zE0-?4G{a(jw$i-&GS3r0&F<4&{_)@YVu)_OPy%N@8Z9B=?n7RXm^BuM&AOv~=gz7O
zzaP|uc|bzbM+?`>4oN5afh>0G)L_;k{u>IS^I57arLL$b>>-rCSKm(u<dJJ|yO4`n
z<M2;7{V`&5%tmHD&-2H<yr{hzIgcC<8Az}UN3W~xa-5HV6kX#==dzwpik@1ZN4PW-
zS#gvG2KGCPF02_~m-&9TvE!rey<(5WjW{1C(=UA^j|dEuMFEm}mmVu^L<Es(t^sjZ
zS;k~tus3B<)$bqoGqvyZmN;DS9|E{dpc*J0>?x~y_l<?Z7Ltc<J}MWdcPmw$>(a&>
ztHubxW?!&($^6VUP_U|n2ZDL04yPR|LT?F<fx|t}wr&`|S%e5l=L4cgvuID>boKk)
z?!4?LPKXu#_Uy*}L=N4@o+F2?4VrvsfTSy%($A&n&hPuH%W(1%w%xNTLV`y>A;;z+
z3xgz)w)~cuX0rMd)zF0<<DfqU4fyPuISo`;o0GdU*)>%iQzogghh#chR_L_8`I-p>
zi)lCRfSX5U*yQhd)a_}Hv%!EciWe3#nUDIWyqLnxIlD<zEaX!u^$Fz_+(*CGmqe#U
zMgcb%PZ7PWe&aP4b~9SUD?KXr6G$pL%zu|ri@&3!3*J#KN>GZ{<!TE?psC@)28&+T
zat?2j`+pH*9VsIFaT`YlBj!a#-@^I%z`)V@Wv?&G*uAb~LbCgvul{b(NS~;5T823A
zf(Di<{j;qW(DVaf6ZMowO(b*3Ua*S$TAwjn**w>6#Gy;AFCqA`V4PBs3f-k)?g%av
zl}DLb-GXcwADPDc(ws01@~p#}P2Avt|MLaNUT~Zv_+aWTIi6uX*f58M<Wa(lnXO;%
z5FqKiQ9FevPe0KH*QtRgdjvW3HoS+#NVosK(g(PKG3TR+cmoe6UR^m-c80lhv@^Sg
zMD6-U5z3nJA)R=qU%%af=7Jm<ld@%R@<5->2Y3CmN;eKQo+j@u#Ar1&AN7S?&Id|X
zK*a32Kj@u{=Ae?HbdDHDUvi}_>g8mSKK>mF@6kY8rH9718<C0@itKJ6dG-z3FNlWm
zS8Ug8Fk$pZFndP=N^ghVJxGNcT%D@UkPV1kG-SG6x&LO>?5YPGr8)oQvm{AZEeftw
z6V1&MhrM>}5mC|@!nrR6vaeZy@4G012=%GV_C$m(sZ|qK&uuAz%@7DGMi#)|rnRsM
z>4163#p~dHAEW(w<Q8#b9^teV-V9r`0e+SIBYE&{u?<arc8-mYdddl;bQk3|12hjQ
zsmK~a=?4LVg6))grYL`7x|U%q?gS1L3vX=N2CSYl6;s$=1LUDk#EN4W)gjQeI>+Zi
zTtY<MJgmR}>_O_(9=!+uofmNSY@~HjBDGZ}H42gxPQq-61kvrbyYXv|$h4y9%f1WK
zE9fGQ?d&gO#d*$N=Ur!=*lWD0)m1~7JO1+=N__%9z&y=SQOs813kPQ+UA{>|B3V)D
zYO4Gy|G|~xr%Uvi2q-`ETej?yiHBl?zu2SXZ%t-CH%9OzA!e&ru5>26DaVv}ks=W;
zlfQR4bRPO4sFD)Z0_z3Y&($pA7&Vv*7PNT=)}KJ&f3xIA=&gVW4Pyp<gQCc2J=j-F
zY3SBC_i@z%sqE=Do`DYWC9~5qXV=7qxNWj~zwX@85sp9Zg&g(NX!;O&4h_EU5*!KU
zmQNJ{wqDIaLHc*YyVX+7I>%1N#;$6b(7C9A+NPV)F61RyWtn>^rMq3}Dd8S8*k$A1
z5BEc$(%Y0_B-dm&kw6hUJE74OLPClv(YE;{XXo}Rgc*TfDEO8<WIN<9As6Xr7I!z-
z$bONBVAJJ7w<`|jwdv0MxnkKe4%h1J89AJKu37U7Su5#-bx;NLONPyS6l5lg*cOlG
zj+C;_vY&r@ojW&`0va$eGsjgzVu`vS3)_?hxpQ=VrALD>U;CLT)iQ*o8I7>F(3n%h
zosow5Y_eHaePflNT_fL{L>a+N(=&W&W}IxvXdgv&A$73S-rZ}~yv4W=zuf#eFT@k<
zY{_(yoa%s_k3XVD<klMII0Z%{Z$|E>^DG<;d=4#j;&-3dqpCRfRqw-mdJmVFHCWWn
zBMa!q;i+Z)%oq??4Kp>@XE*rON5I~EYT*}=i!$%)9}(;BOTGw27jtBqw<h+Y$eaEB
z`gzHN`M-qPj!Dj|8q@b-s(*#jPuMa2Q+>OTpeedjC;pn*w-+enNyY83(fa56-iINY
zo-u0-Dc-yWFu4AY8{XjnPJ-oQaQJ-3{68cvOcPJ-nDE~kV&=aW?Ep!9$#AI42I!Rg
zfJ%<h%gkfQ^hVH>G;Jj=YO*snwMkGhMlrKtpO6f~q-JI=Tb{@%W^FfH65P0$!=?l>
zS!K+57lDG(b828cYVofiU`NP!lA3RGDz)rpmN^kCHRI0bz_JQUMp2@EyzKAa_N;4P
z8Xmj&Qrx=J>N_AU#u;qgi&dpB|A287?rb^HURVm>yW3OZOQ(BYr4>!PI#O^lzOxUP
zeCnqXD2fsa6xSKw_I}+*{rS1Gqj@}q0MtV($K^9{6(nZGCE_;`<l2d(HXfqg$3+}t
zmKrnrF=&L5h(jfp&`*aLI6Op!F7B{yv!Vt+oGmsNo{TH+w~_rUZ`u~!<H<9;xP?P=
z%ikuQoo=lVOtJh7I-#Pa$BZ)Prj7{{<|kaJKwK|5fixd|+Mmza$^$MCr17>D`&Vdr
z!}lR_{lGsKX~CtVEpp?pdO`Gue7*|hy?EwbgZuIoUsBh7VsJOeW7EUk20G(}XsN6O
zo8PKOups7?ALt`Jxc%JDDzG}gPt+W=llFON05gK*;yD$mFQ_|){giL+YIuLgN*%Fi
zA=q;KWsbM6Z;NOn(M#ej<p0;*nXO9HCBySBCc)B<854YBT^ET9$ELMxvgkXqFHbok
zxX`$V#Jult$=i{nBo1x<k$3*=S!;#6m48Uwm^WFw<G;rKmp3^OVX3_}q$OYbPG54o
zkA#k`K<S5%>cYU+O%DgGWX^5foqZRMYoNs<uDgObxX+%K6I7rCR#R<C+ZDC$MlZ`*
zo(lmq<_Rp%6!_Zy;e8P!7@yMp@8{9e0<ViE8eovhwHdV=^llzV70!GMo4;6NM8Pbo
zx9u+%C(N!8uaD=oJZ6W);eSQpL(gwLg!g^XioTzb-Z=`@v>Xb&H19tptQX!Bq4Ak<
z=^52RAuJ9LlaQF@B!YlMp8>o@CJ}yO4@z5r&Jyk3fB)0l^azH_QBzb51$wp&X`?73
z^}IwFzk1cG$=?%c{;T{iorHC2tDue`(uCs2dayd(nALM?#WY?n@R*m&858vvvH>Km
z1Tk3c1XwA-$ib@KyDKfxXR3{EUFIj+Y!W>p{+EkT0B0mz3QTWfAOryX^@r$=@z=<+
z;yd%#|M>_1A3P6lPrEM9iXue!i4}?T_Ma7rso4L}JPi+L8m8I_a;OZN2}*2pNsg85
zW4z7wYEt9(cVz%O_y7KruY}CM+_Naws@Ev@keKyBgB!5T);;n6?`p0KCnz+-Uv=IW
zf;yjlNxp3N^zE!VIX>Woo$j+yHSQDdrnV>bmyv0va(@x4pfC>Tlc(7tz*V?~T7N1S
zjpq8*Fn4qvWh;XY$I;cE{A0D37&GFJ$JO0K52vsLKr6s(^+f^=0|TSiH|m^*cfBvj
z8L2^HN&$4oaE9;?rnz!v=!}Y-JNd*By;W-4X(fO+Uv~P_AtMU<5r7EfxPsKtaRfQC
zb4%DAaBE?sL~Ea>I7^@mq1XC9iJHgn7G*HGC;?s`PfjYJzsMr1SZIkxi6$yAVKogs
zL@2{R<K6s(O`(7*gMeLR5po}k4q*gdBqvjF-Jvxn2vw41rdpXlaNFmNUFl~UZWNgv
zlt3N$j|gA`ivdkrksHA|Y$YSXQj<e|-&p?&(A4<tvK7r%GnW|7?nYV@TRMO1hYqy1
z)?U3@%$wrzDSX)o*z^sr;}QuG%YQ5+N@f20I)1uoqQ07uvpsQ?U_J@j8J|RL+G1t*
zVJZ1tW!GDDMB)pOmC!Bl(cj~fK=Vxu6u-NbDR{1YNeEA>h9>?WB|LG!afDa}3|9C%
zlo|z;goJMU@;9R%4*bTCZQ(`nCF4oZd@3FdV%TwG$5NIQUq8R9;?om*9~Nv~RJ!44
zuKeUj6n(s{)F5g~E-f#&@N3r*_TMT>xEl<bcW$Nstf`)5zF;{YTONGxR5dV%v;Ny#
zV>AV{^tVbUrI;9HoD;o&<IcZE@DHy(kLheJ*$++dZLfPl#PNL(AO`-tdmM5u9Js+E
z*oUN&mR?a4*M?Z<ZTHy&Ij%W9fg+bQsQ$$tW{r0*`U4m^EZegCA;XnqGIn{*3Y@>6
zj~)*v-}Wj_AivBx(>3?44_#AwaigVpll<|s0I}<oBn~nLG@5uBN}J;z|9##EaZ80i
z8kDrJ&nRV%o#})Rakut%&E~CyBs1mkK;>oXt5?)Ws-9}!F2yCRB;yM~1-0uIzi8bb
zs2HLWu89l6or(gU3PjM{8OlNb!Q4~D7Xs7da+hMp<LhU=9OtvCze6gcL(9MU{YrrF
zsiB)iUG=S2+K|tM`h_mM()va56uRVd2hGgp@~p2QPtSF(d15lR*3+eAj$SvGWu`IJ
z978?+$4XA9SW&&?%1>pNmgRCifLJqkw=L+avG#$KQXE~CML8qFw#|}m+1thapTX^0
zf5Mb~o70oq;BKe3t>MzLl`H>xVmux;$#d8gg~gDaXjfLf(cRXtW~S4xpK}6s1XH=@
zaNFxHap>;<UD==O;KI$gVV#EGl)}zEpkS7QXLq@E2gS!fH#Jb=k(Ko&ef$M*+7I1d
zmLPf<GCMoCjgc!RJc=`E0B#*SBf2`L_Q$yHosv`z-6&vZN5V=#@fMmtq2Bh9Pp2px
zLuRxi*M+8FQtZX>NmCF05ao?==VfTayH}q(B$acP5FGVZ*EzlUH!SJp&dZOHR&-J$
zUbJyVAn_CwP;X$2H;G`z*W8_wLh+P?1kvf@P)XN5zq}Mz$hk}_86M0v*Zg6!4{R<j
zPuAo|XOqgM&qZf-`4Tw|E=@oc`<Jkxpo}wGE9@wlu%wdKGrI7Q+ueX=5#RT(p+5_#
z;%H-cyCT@$r!>sm*BdfUu$m<(rq$(BUQy0cEqLKOe03L0-hQMc-zJ%?ktD8C2nMEl
zdP@hxFbfLFb+*_r!T-GeG>z0Kn6M>AZeA1>_c+fn%A)Hj+LiD%kpOr!qTxhF;tGLX
z2u(6Z5kdlxVx87LzF-ZVN-?(a<OaNf%kOuAl2hMPOVtz52I(0|zlU<6#HnRapmbjw
zidB~UHBn7o*tsr7Eac*{b=FieP_$)#-ga1S!;|X=z6Veek>iPW{ry~CU79KmPmu=)
z;jO3+pKJ#+3j%3xcDEaX%XRaFHb}j*kf#6o$9bYO<++ZQS@dcw2CO6I&cQ?>YGUj3
zY}d<8P%n9r#tX8vC(aj1>rLP}dwTMso;#r?wv})Tg5Ix_pdT)L?9yIqvgQ+>LuUK&
zR5Ndp92Ex0*?<EHU*7dG{^T!q@j76rle=l(>%A$%^--U2vgPIF`wX74m12%uvnZqL
zB2827ZnLLm$KWR)i~g?9*=XguXS);Z;-K=o`32(C5q*k7P6fp_yj@PuQ}5~mij=9i
z<{XJ1zImsZ-Gjo@zq)|NO+?H$KbVz^I?HrLvx9J{5Hs8!LeeP7**hYxE*-&5iKPR5
zx#d-@J(Gp(-OUqbinpk4znnr$kp1Pogp|n3QFm1l^>M?^%a($Sp~uM)iX}0K;;4Lj
z7TtwBQMcubVv{4Y=pl1VdpJ~-vdn}EvX|jBSvC3T)9W=mTUPgr2%3kcK$PQR5g6yD
zkzfJ?Eyh0@ii$}J0zaRtqGKSEuc;$2n!V>gK@?<rdzf65EEd_$kK|VeY)T4Cy+s+@
z!B6toZACu@_NomAp1rPPl$7>RhKWRg%I6U$i%1|Y(~p~xT0HxCA5wzoY}zbb^sk_C
zR%yTGZlV+Nr`Df;lutjKu90Yrw!Y-o;?o;qT~q$7pF-VHJ^f947hN+G(k5CXTdp5b
z=p#InVDU{`Y10d&LETEmi-Vs(xAA5tY0x-umBJ|g3+M7{Zll?RRns9DR-(0<ngji|
zbIQ+%?k?}w>3eIr=0*8jIq9@7goKHbIO_@F95(I;4i{hufDc>HcY_QhADD>tF7chG
zx8}lVlmrJ~3^6OQ%p;Wp5!PK$8zm3LZvE>3EK#DN5KITHcuY#u3U8g?OAV%@5FU`1
zI)_(>H_1+JFk_>Y5Q5LU$vAO5<|`J`fFCUzT-f;@w1vl<P}p)l5SwYD`Y5Dz1s2w@
zF%9i@liK(Y^dtO2F0`T%S~b!M$B_BegFGti8!kBZJzfyBxsJOj9W!OQ=p}PAnCu~y
z=#x9`X$|uNod|J@<u<FcGtghO0Tw;%3t7h}6E7a79g=s#+7KNc0S7{epG2dAaV$F2
zUV$y73kFZUT8Cdaa#uR=4!H;bwCckvhe@~gT@s7nIY<HRHPASip|l)sZCD?+eh(!o
zz~^;`4Kv7sW=mV&%gvRuK5e(=ak{H);d`R6Pqf;3bCV5?q}OG;zP!sEFd<#6q@}wn
z^aE6y{|P{3n*f)T?fP#At^&8Hsi}o?NIf_t%+Bh;l(TdN`@pGudw-*0HuwmOX)UGi
zlJ_ADOnAe6E$Gm6@EYMZKY+t{c59g<gS!Qp#4=7M)cyGMHi9-z;zW|!1{i9FG6|^P
zUbK1UF|BI|yQhSYtIHfT>(|Mwf5v=do&5|FD^E$3tNb7_rVkZ3YS00CT)(MySS4@(
z`X3%g7{Aa*d{kib5|p|;=*Qa&7cVv;rsy+d+I7$^DD}3^H7!tglx9c=-ClObN6))I
zu(nGlS1$mX1Mw9ugFM(fPzjuixm6F|S;A(tMTA`n-VkpAGN;`#n9M|^=G|^UOaOWj
zM)j**e(PnoZeE^EGB-m_lXEHMFRzw{)G?k|U_3o>zGNnhj0*}24(w7Ujr;tG?&lqE
zljJGpmmR`OudD`aI&jfekr2^y15jgVsen#}2%tXUVI%Qv3GMN$MKYrE1GyGKg9N5*
z<Aa@h3Aqqx=Uq4J#ws9aRe{qGY`d7oYjd0?Ex9g4$C#9(+q}|{ri$rTzUvs-)`sr`
znsRjsP6(9(GygrMFZv-uI*Td(aaqd~2aT@jl#}v0V8BM?1#~XQdaC_iikxiD;KtXt
z);uoJQBWbC3#9S_e2WEwdU@Ym;<QNc2@GZ&`sTtnvT;!_a_J~@k!+XF;#@>t{Nemw
zQZ@hDxaL+>XkMd?J>ML{h{Dy29*OsG)T#a?FHD1G1p|}j00_~r<uMD!mrymk_n`kq
z?=+eV&bRL{MJg=re{O^em{|~os3sNPc<x0$Ald;0mtjOxzL`0Tbm5alY;`J(@eK%=
z%bz5=XNuoKL`PP1>Fn7z^obd9i|4$$_LMwT`s5)XShPqH4Y4psU<Tm62wP|_HF)L(
z-HAJwEoy*^J{IjyVsY(L7X{9$I=4hlF}?S>0l>^Ap**JCzCWS`oiG!36_gUO1lrFB
zEANax6sbIMVx?$dFtWh$V?@!f-K&uw_^LB~6!L8F%TqA=A#%?MG&aPs2;d_RMzkkk
z-JeqP-DtB#<SN&=zG0(f%((CyU|pek+3=MDQ$$R!3DupxGxr#pBIqf+bpO*Lv@EyJ
zP^Q|Lp0u>~BJn8Iy&EVLAFw>xGE4KQ`I<SvC1rZUQC_DyFF(Xt6Y{b|Z+s&^N7@;Q
z6Gs3VXgTwQ?_f(S$ze0O`Il1aUom2~Joa$Gh^kKRUyw&y`x2|B!Yw#C^|>utHhiP`
zx1=r$z7fdfp}rcpDa4j@Zl)VdxyGY&*kV#GbILe(u#AlT)$y%53epGp)OB+gUB{W>
z(ax;Vo@)0W9l-i3zsralL36%{n@O0W&iY^xF$K~N$nE&1ck(Zzms3&B5Y)<a*YN6@
zhw%<{i`}rW@Ld)HvaWIiA52?pNpIHVgRa`dKg#w$8}Sw7G+uM!D+;o&6b@-70dRN|
z>q3YJ>{KFp+~%?A-r9rEko#DF*BXWCd^I}{`T;_iPgt1s#bo2}pu5lOG&}1x&y5M5
zi=?Qv>f9K?q^6!jZ?H6N%gXqLO3Kob%$BI`E+NT}al3*-_1wPWpIR<eanKZ9U!Hu`
zfZ{|f0o>;6M1YUcOhKBN07P3pIc99#Hh}`32^x4%i+`ICWh@VE$r8ob=ihMa7t0+)
z&&^+oBK-%Ez`xT3-q(*6x_TAl9vL<H^~$|%`ov&T&BQtKBO`{DLYA=3RNd3cwz_jJ
zHU}B4{aIvoMZ<>qo0$Bgb9eR^3eTFBF|6lA2tTc6e9Pzd5lz)|rrZp}ePe8t;gO|U
zA=mF^ui2<OD!{P+vA->L^<1R1PA;fnzLL>y-RLmG0Utl!&iQcigT)qeiw8R&EM0nJ
zca-)amsd40pB=tUp6)xNxc#>UONe6veP{cs)yI@Mt*ZLS8$9RqI48^&j)Xpv_RKW|
zUf#FDLNDwk`0BU;DJ~TMkZNe6%C_70<(E=H;Do^5n@vlR;9*FXo9EFBo#(oV$H{z(
znk7V7$%xyA2xQ-X>b}w|&@J5xDs=>Vqt{e#mmyIX5j+6k7c7?8ltdl0J5PvwS-PO=
z#}6Rioi_70{91V{ctWK2%L&mn;Vh@G^nC;2ujKLk-8(&-GKy|@do}@u#T>m5?u-A)
zp4tGf1!gb@mjeT>sz;8~R|lG$UN&^~vSnFAcS2k2NPF3p50QK8IiJo#!x~#H`CjW>
zyv9tQG_339U8fiK_^F<rkpWDk2O(R!AAjJp!`6>Py)9heVrsJ;92`X1MoY^5*NN%N
z@KF`%$MoUavzIToQqH`*aqrK?@n2{cunG~s>u&B{lTf_T9a^5Y!gtVl|7hydM;5uO
zrhSs1b*SxeV4zr;Gp@#&M(m!Bmt&K=BwO@na`j{6%?IWS02a;g;i^+6d28QTvwg=7
zu_6Mz&i}lR*@rU&p)5P0x0KxlE+(C3L(5mWyrGTJ=Hm?h-lQ~bS`?)a=yKiMMm{*z
z*&d2@baGMy1G+J&We}%ZSN0Uk#sey!flSdC9n&*2x!6yB716AiHhP7gms2`hFN)2|
zii$=}-YU44JEecFo#0(ESoO|O`ES<nc6+Db3FDP{FA_2xUxgW!d+{87z#r|dy5X8u
zw^2PNOsPx{S#<21k1iN3No{U7+$srFgE}q@J8d8>qz@k=>-Nsx_P@0N#x@JC(IEd`
zQ{#dIo79r+ahSxB6Yy_Y+p=Kp%$W~qh^dbkYU42e&gQAdv17-0FiSH|>s|&eKyP^W
zum`L(fV48-n!gB)2C4xY=<;kau3L$E6z#PzA!+I6%p#R*CVEfy?{Q9Czn^yw$X{Y%
zLhk9kQ#;J09*0qS>FIACcO(GRpcTp?KyuZrMyjca#lcODjY>b#&*SyrL}#DrHYEqE
z&6^7$RfNGt^G|*+wm?{)r9FtgNh!xCYQ_!t;j5^>yy^|;I8BVXU($q7_|h!-e(gsu
z6kPK13Tm64J{B%Xp>SuLq=Gpy?c|P57cM56ynm+t<?eDRuFt3L($lYJb&2zM;W6Bi
zeL$?$rCsSymBP!VPX*^TSk=;h%Di1rkmFbRzsbu+*?p+9-G1Z7e6(<ZET>`qNlB@S
zNftkoE5JV$EL5k@9WfoZ#u3v@PyC#H8$t*7lZtNUsOaS89EgNqr2DhN^6!g-^qzhJ
zCtl)va|MCjQTOjNF3RR@K$w&9iIux||3F`5`@*EnpcPR&1M;CA!!50>G-;SH+s^8?
zS>wl#S5*9t`cUi2F?u~5zl}TkHvoqpJS;buZ196zo!vHl_Uu+XnWls*dTV=gRp28|
zoaos;qK;fZUAp4(%$1#P7=gA^3+WYtb31&oNGvhp$;$F2I;NMik()J!-VwnY3b`Rk
zvg5*LK8cCJDBaNFD>r4fT?Kvpd;DBT{VSR-@qB6SboNK3v16}u=5D&@E%jO*gEGzQ
z<uO)n7+EZ*ql4WG83m+y!PnOTq)K1g-A6hMx_C#^yA@*xjN|Te-$0u<2yfvPa7+nn
zn*hjU69G1Vo0;<J!eaIz-8o|ryn1C@FPhLMrk585esjSW&&7DeUEeM$S5NqND<IK=
zD<^mhK@<io!~}a}aP)#j`l0%|-7HwMK7D5Hd8#b6f$E3;Z}uY%4GlCvj?_}fMB?J&
z_==l3=HNK>Tk)Y{3n7zed`Jdke+jY~VGE1KcoPuBl0H5R#M^qLVUsnSt?2^i@^INY
zK;t?oB?WJtX2e4}ZrC6Hr0KnDko-l{HUme@AbA>G|LYy~3^5q8>Xc{v?(JK+2Q~(v
zE*jtPn^pG-E#G9XM0nMD$Z6b;US!np;_yxKW{_Q}^61e)k<0e1Wc*07cW-RSayv%>
zDlDiYekC6E=Q^>jm-nMkGZTBeG@E@02`sRcdb}@0FvJiVZ(LlCE<3d3<-l;C0=W?*
zFjHJGz^yT%q3@c>Hc}<OlQrA~e;BG)h9Z*=z-=6DV-43Lexgw@n~IaqTj_8^$O!!w
zT;-2GwEezjOz=Q^S+r2=dP}+%HEw%-?WctMDgC=4r?hX_*xUQ~7u=4~`bb=eSgnVD
zCluen8qCDU^!~%n+M2}{18YpYV)Bbq&nS-#@|C}Y*>-^?r5|Th?CSe9dBA_^Lhu_{
zwrm>Ax(PoMAsf0qUCt!qBjQZgs8Im{0gQ>9J#7v^Ok7>dB+yCwH%_02YMxtflbp(?
zxGQ}xOsa^YRYgg-)AC+)G=!bs&k(JFb}jh4CeN<iMS))2ecY+oatM^~Xgm!6dXbc*
zkXh^VmEB&JuML@y8E1Bz^25lc^T&_fdmSvUFaq_Il_^r1Gr9p82lfb^a;1gwvaIn-
zZiZ68Nd{`lGk~Q~n?*r3N-bj3>ztg2C<rAR!?USbuC;`K`xYGdJakj`4Sd7rEX_fO
z{Kviz@59$+i_NHs_r@kzUeIvq*mPf>`Ra=~eR3)W?|lEyE;8!lY3zAg&%1YZ#>!>E
z-r4es6{G*I=!JFX1l1d!IP=(9z#*(!y7X+m5-9<?(JT7XgUvUfF;3buqnkBplQqFT
zE!c_c#<PBzk#SMVg!7UZ?nf?IUTt#w{(2o9VV7zg2!*<=d(*F4lWSI6&CJFh(%ZFO
zkIA-+1$yrSXWht0i+07+Yf?pVW#to#+{t&&7;3C;KNZWO^+>qO;wz0=A-ZE~^IxQv
zDiCZH`n)|{>0}O~S9oPG?O4e;Yj5pw21zR{|0p0@z8YEr2QzMNIrt*kG2Lh#i^6Ui
zjJYsO`7>n^Lw4GmyXvq%%<G4d1!g_(hHYc@==imv({6^GD)C>t{NmAxS3{&uzsT=)
zxL2IbE?ZgI%Y=fL(ef8w?YFgkm6oP!seR)^E|ZxzBJ)r58<R9!I#1;{b#V5|y!`xA
zE?(dNJ*ND4E^iy`8)SKlBa>b`*dgb4bB%J`EJ|YC6XtY269W)bc;-B1amw=bg>P<M
zk1`!9jGv3uJ^iAuHLeZ@nSAzI{trSZkz+Wnz@~Uo1VXT}7JmBJt*NT&hplu~`Het?
zk~F~ETj&%aGB?<;F`>iYeso)#Q`*pG@X!`y)Kua4G2t{w#EdY=-Y|uAq(?I@eP5Qb
zn&lzue?0=4Y+>2y&xGJ`9y)bpzra2d-A|tESC)h=A+nJhZ3=h77TU131U?rhy`8?d
zf-|+cdBr_t-A}uH%nkzfb>C|4#`_f=L(ebv@n3$ZEMxfBEt1X~CZ4(5vJLaq&#8lw
z0^cqgb!PJo?SV02+sUPgc31k3(-x-g7H&cQi&;{r|7ge0vWMTovI)Kn&udrTZ0UA;
z%Aon<Uk5pO<u<Og%YirRJL<de=XX6z>imKNxv-vECo49$WLE>^@6Pt4a91nJf=k|(
zcN^-w>N!e|V?8PY7yDc`dqe5RafalkS15<0spXIWvfiEU1K*Yx78Zg{glR9+kkJ-C
zrux^pNQQrz8)*!*tuAQ6jRv6+#Tul_<&p<unF*2V-S4c^=;Gu9lt`*E|Mc<dx^2bM
zy7r&cjpla8I0sn1jUCg5(8wgGXQM&)c^QmamH0VRF>?KH#njVKQYx@mG1Mq(Yc^Ex
zil2t%XH?WrFVvNhhzJ&d6f|312w8jE%GAt^I=Lo&C5h#v)R>;^`krt>k9zdz`mCa$
zNg5hP&i@t`Zl2b|P(j}5dFcfh1MI&m^3}s$58amo`{k<sjnk~ZdCET2!4>+-s;c&6
zvDkoPx=Hm8q)>!+rq)Msuq7<!7KnHUnwNQa1;a!4&n*KdtQ98uz|0B?;4fdkeEm8d
zS4TE@*M9INE-P*cd8cCX;xvIHCrm#Rc%G7CMn*(K*sn%@g`qy9<AIN7E~n0cpQ}X%
z`uo?f5Bd2Z2G`@qNvk<n!&zj~Eb7^*yxx{;LJ#-kBjrtxb@r`WMi`nYMtg|sOt*y;
z93tHd@yp)gg{;0bF`03W{DmM%=0rt3na^-vT;v9r4Jgo$So@VW0TQ0>%nCw+Ga;hO
zPAx*!^dJ9#{%i2nD@p1E1VfsbU)+0Xit!O+Z`b6p%FmYO3{sehhnY&Q{@JjSuyAgD
zVpWiVOevMsg%v6xYbT5!e~S_*#3}*u8Od4X)=e+^&lRwFkCO#==$vc~I9oYebL!2e
z-0^`<qmNC9Ietx^yd$*y_4Vr}ct3smwC;#1_q^X^-A-E4#jCfz6M^aDXkW+ofd{f9
zH-)NpL!jzqS3yX71jP$l@@D>wS!p2psgozq{g80OTiqd?bnq1?B?c>4LBLb|0fIMm
zse?Q)H0qZ&`8jA1WzyH0ni^V}<>enB-syg=@}0KG;t_f1>r4Ccu`vq2!L024;*NRZ
z<PRc&r=zz#-*W0vHhbJ(tgv=xwrx?O>(#4QDb}GBxJkp*Y_5~SY_>Rk^LX>c@~HL=
zKj1k1iE%Z48jcUY#g*{Z+&_;!zSq`Px;fO{{!y3YzPoHW;B3lqb)1oV^<B%;!iX)J
zuGht)t0^{WjLe33>#iRezO%VFS?X)<p{YeucjpbRJAdKA^*ArDsb)&*>V_NkuP|#_
zs;{35Tnw{1CAGq0Rmj>yCmTjh{`-4%_1!7KdLxSKWV*L~EO=rcao{TiT+$f(KmPcG
z;&gnWL8kpy%CMyOCS$j3xD%SB6}o999XT1F&bw=Fq{A@ZjGj7U#{0uzQa9}w&SIaz
z-g_^pHA<s|M{rYfGwSwII#f5EPH^_^Q1pogpi@iKy&}Ivt_|CY#&2$6FP#3SyiIy=
zc?4>@dWta~AvYTu8pw(3A~riyb-y?sHo~>h%8s{wnX`AGl@~B0V$#fAZx?PFS-90p
zrh*u^^>)*#t8~J4%G>sS`NDjz_T8Yn1J!2DQcxb9d*5tzkz)CM_Z3;A8up+#4BgjF
zcRQghvH8g*y!WiVKV9-T6CC^%6dx9U$F9qUQ|HX8I1ClC+;sI^l|eZhS7~dzpE>hB
z4v>9ND`j!z<gWC$Pj=2fJ!HhyHKA>7Z4aUzF~~4+_mXD5MzMJYY-In4P!lV0@@;Cv
zLwVgNWG-A(v9I2Q8}@6Zm3}IWl^)NrWdDL+*e5psZD>2b*WO;_m|uVMttW1YkIQ)h
zEUO$4FctCmNVPS>;DJkRgApZ8rBk*ZE|?HM!8<**bK4c+PZ-nx6}1&yaS>P_>`OmH
zQF@-L{jnFevTG;2r(=?Ie>p{HCoxLOCHC=S$o#^{bT_Z0&cx&=`G<?$RMZdW>hik>
zR=*#%=k~t8=L1~4?SIzS53t<0^~w;vUOjvMy*Lu@aw~|D7pHGc;mn!ME-qT`cUsLt
zYi2=4#;U0PCkBl&2rsJ{cNFXAU3<=jpIf6qAJk+0(BGVQ0?xYxSGa;@cR!Ewo$Ql;
zx2$3VC7E<{-MCwM=1*TIO}rGkc0$axbibOvN(W8a0Uf|C9+1$5X6#g|JJ%j^&_2-S
zu96*8$HM!;-F@0vop&*g&6a#zk%*(85(20l@h-OM>%{sInwvQ2-RV@$N5LvSJ{_gw
zZegmffx`R#F&G=L#|eb1j~<k+*GlAlkhAnL50FyYRdc!bglKl)x#sn99oI5SPg>sW
z(rRACjMaLJ7a#9(W@r3DE$7og!4P@rx*+RbTU;V^QWoD>!<pB0{)#?`0I$aM>GPF}
z_Pf^(b?!gLEm*Qj!G1SS4dfDQUlZ_qKcCU|`1O`#^PKoIzZ_R?caL9r;K1V?w<_iU
zvX9m_j%|5LalJb2x-}H#F^f@t4d3%&nwJi0eVg}4SHJr0$7>KQUY|X&R4&#s0~pL~
zUTUtatbE7)l<T;Rj>{LZ@Fag8n{mGP*6N-EWB(G!hu_Lo+-GAW96@l|uPyawiTND^
zDda{hLGyFI&d^l{eQy=xSH&*0LHEq7T`+X`v&y9k9*WEA^O-f&i&0EVHdE<UI)Cki
zbxuxka#;g=Q7@3^bU!0bLJSb@=C*{wjE!=3OhhDl1WbP~b%=TZ%UK?ABc&RBn<JRl
zpXvU3TAy?y7n2OXDZhoII}(=to$HDdc6TlxE;@bLt(P6Y#Xq)o=MUGurb(w4c7;N`
zJn+bcVO#5<(^H?0Q@JJIlT=>%t=`-`E6DXga7l0bJM&Eio;)yu=OL@;Ww*-p&ts)j
znpk;SA})IKQS2q5_L|rhuf23>dEWh7vw964I`sD4N<VNTG9u`Wo$b}>lrH&?B4U)K
z6dx1FDBcdN`~+K^oRoATthy%&i-yaX_`C0DkpU>@I{#LIXP7+M5<{5c-Py<s?sJaL
zCq>M=tF@FT!emYkz7cG1k&*EsU%>I4>^#YjnUT_$#6e<NU4Pz~T%xYoF1cu$+4XwI
zdT+ME1sixJN1i+E=V`+#;rScnZ)9Fx`0L`@&aao$MxV!{`N-M#>TF1(Lk+lMvZVty
zBSdEHn*_yDk2BiAR>5r}xWK1)0N|%=Z1Uf|yN_RtR_FYc3Kly|Qx=bN$x*r0_=@Li
zma=&ItsR_6Q>BGpTW2Dg&6?0tjGFx=JG1=r5!26$qq+uL`^+6<eOYDK=oTltrafb4
z&tzYs*Kii&$nH(M3e2u`Qa&%sO(VhdKJ>WTTuq#mVY<xjL_0N_aTX;{VN%P)T@rTv
zg@rf+R}>W$VU%&>z#{^nMbo{M(?f#up3?yeH2vp*$qC0k|98AP*O44=*KzZ$fy}!4
z=7&?e{g`CMxxh46tGf~pUtRqmO;iXGQa*;R5cPLT{;1vdBH;_g0uq~gR4N(>nE5@z
z>mLqnpikuAOFg#sAG|L|#bs)I_Sw!YiMBiL%=}CCL5ODUsoh@LiwVLH234>!?V;&1
z9bv~*d`2>_N80FYg1PXRa+s8MFjU>e%529D_kC{&8Y)U#`A~v&-15F0_5vE_LSQRW
zg^7pOrGQ$n>9sS`XA=lt9XZy9IU9aJ%gRscy+#3)$j9p^(*>83YM7o27T+MNnqe6Y
zWm$uR$3QJ72SO@8Ki|>O(aqI$IGEk3+ct)@e{af|dc0uCJDOX!?WX?w`t?d1qgk)B
zqQSIJDE5Dv(YU8r+2J>@bGgl(DJI>kpY~VLeBL`G_ne>KgtkiDn|DrDHC$M(_p^Oq
zRX(d9kf$!4E;UR><`JJ18sI5Xa5_pX*SMO|zr<Yr`?5nYk#Oc5e~%L<X6^g70%am$
zvY=z}{@rWq%vRnvVU16OOUmNy-{Q}d@HpdNgYfS(E}H=f*seci+Op4!11CQy@bSxt
zKNHZtZ~MjX`A}@m)Z#YFZEvrg0EI}A-Aiwo$`FI1C97e71vRlzY5g5KRdn$}4?Kw(
zbD=fGDkt1`))yE(cMC&q;@as}azlnF7Bn?A39@OW^!Ltw&BNTCL0Y{*T2tH)9P%Hu
zPdVOnF8<Q!Gh_LJ%=AZh*P{_ohXNjX*U#EoV%Dl_I}(R};pJHr#<fs)G`O6wS^C0%
zhHJso=X?B)Jzf|QdZ=I(+@M9lJHFZ;r;5CivEvU`PWraft3Js8(-b$bG(c&u(gAO6
z(ft;rPurQNc9rYzDDlg^jwZ6JXiR{2oJmTbePtJtMRy!iyxOp;%1TGpBczp1o^diT
z3J?i{^efsNp$3fAgPEu!F|)XLc9e_<?z|=Y)9U(p$<DAyHGldDu&JolQ`e^V4r}Yk
z$fJ=VQI|@}4lf&IH7Q*sY{X!lw<WaH6GKB$ENJ~T+9IoEVCX^b)|oVE0Z2yiPB9S?
z5joSb(1!CHL75`R0{+kC*<QR0L1%Vm75vW&e*MMcjMioOQ7y*Xyy8~^gN~{Bn2RRD
zk{555+6>&iVnc}GKgqvm(hU{^zG5m2u@JoZ<Jj1bd3m(4pXcV<lb{8?2m_XcElWWG
zf3w54CoOeRn9(eC#>IgT$1;T=3Isfw8QUPYi`zgDUDh#zHOqc8rZi?RtKFX_r$qx?
zK+D3(`p&FdXoZy=U@tpXIZaUVAB>snlbm{rpc0>dc-wImb&0`?%8psSUiSEfbn1Pl
z7ut6?=rm+ktym$P4B;~9GdzupJLX8kjj57-X|*ypy?lIRh7O%@`UD;yj|gM5ndov|
zf&7HbGCq63^e@Y0oDt)at3E#Z^NlY#>Uhp_yL91#sn^6|?1C_?G)Mf4VZ+kEltKId
zyjT)u3S${>aE~%YCH~EA*Xi07Q4?4&)9-U5Vg{50Ys%h6!POOSKM=Mq?55g!-K|zu
z%A2=M)X?~pcw;vMsN}B+a&CjArKJZ92+llPNc23Vn!fU1pG_@aXlx!_Xg+JNo#Bi$
zuM{uY*H8U?FCIJgXYJxExzjh>q2}msnUa(w-Bje)?c9vZ(+>NpcmKWiw!@9qnZOGi
zCT2m`GgUz6S&Y>%pt5W8s@eQA5ZG(OSkd?8dG{+GGRyi4#RN)bzq4nrZho=8d`_y;
zxS-@q_iwhe05J?g;$J7AB5)4eSA@U)i7+39NwW+Tn3;Whuo7pO<(WaQ^5659&fGRZ
z{Zyat|9UxJh77@CZ1%Kib<k*ViMZtxfZ@&m>f4v!^Bb;Mtf#-@R=%k`wS=LL7}VLc
zIB?pcRzt{8fb{?epklR6tyVOD69x=0SYWevZ@&oDFt06k!8@D&pJ5>*`_8w+^0lw#
z7MWenxHj*%bo7)FgdoW9GrVG;0cTt2q3MXbp8SV56azBw?c@CGr}4GP&yn81dprWK
zwzRg=Hv7b8#(MjRj{5X7Y8IfR3Ry;x71oT2O0GACbsKwrQ+GNiLhdl*k<gR>o}*(k
zw~oY7a%his{5e(*yjK6IJ?uKm3M?(}b~qg4T>2@`QNo^pTCM->Qc3L#3imjrw2ibg
z7fagdq}^LJNcNVX8j%E29iMc0E2f_brw2FpR9yRPy+gX!%Qs-|v{xP_i(TLT{5H}c
zKlg#fxq<VkeiHM)5(E(=W_R8rxg{PMLL&g-L9ugq;nX)}eh+>s8N@p|+-Vu+=Q5}^
z{!CaHAk>Jr;ml3sd4m#qo)Y6^IC~d5xec1U2#seGg#hr6B@KJvX85kSpU+Tst;l;c
z^;>wa;j-^Ln_S%gs4Y7=cU*EK_K0eCh8XdtsUqK(-Qx(L-Ab)e2KC<6+A%imh1gUa
zVlkEsuDa!zYZQw_YF~hBHq7m>Mo1W>ocER%hV#vngO^2>jkKSiTA|?<9zLm7+JD>U
zJ@cl5W+hIKzWX4o6lx%G&)y|V21{Svw>~0f$^1duXhwvU&4Jo&#xD_hyt18S{3YP%
z(SJz?1K)=JPFVQDKN@$6)y`uyHO1njgK}Aa3{}}VvXFI47G5vlIqs~hUXZ!BTF4&#
z=fz^ZoYp>0fA4OZ26I!U3H~MeA;SNPmWv(`6_a%L$n&7^IEp?X5q|%EJvagQMvdha
z+}ioOQaNqh4_0%JP1(C~$F3y0Dg>t2S7z%w<x;0G{hsVPw{&UuN%p-KP8+qY`qn1U
zJ-6@FBX-4|D+r016V>g6rm;2(aVFH3zu4-4%%rcR<eJZiAlrbf4K6<*3S3W)MF0FB
zCuP#;@f;B9ol*iC?bSI_a=5Jb-CIA2v9;<wa@J5|a_|wH8V>&G@a0*T=}JzJ5vS<I
zhgo?Jo*G1Iket2zP@8u3j-Li$Oz2M8CM-#R56jq=<SeUax0|YzYa{e9F^e*Rn2&_+
z%Ip1mwv&x9Xip1wmQ8y>lP^4T0CS<>+C&(r6(cUrXM^A5fG!3|_mNe;nNWLzQWM$y
z`Z#Aq9fM{agu6<X9p~5+E)tk|D`#<!9zB-txHqoT#apxMKSBS6`0d0fNj)zqNl8{T
zMZ_E3Trp&CqlEFG%XOY<@H(8PiA#diB2%Z`zs`KH!q{kh9A4Q<8_%&Lr(Mv0^D%1s
zsenDuhJSSJ>UwP)I=3qUOCB6Xnsve~=TzCVp}T+mEb^*fY*)d)r~_ZYYsodfcqT2q
zndN%g+jr@XDE2i-cQ0kH?l9J-DJUqgIDT^9)(DsO<(qWgUivtfoLx)leaoJm73`x!
zfPB+MFC!zPc*tjh9^BHP8Z`d>Yps{^^=cdZVE5evqf8gsXuvrA*`tTpOl+7BIn9-u
zNjTPumN#&40xT4Fxi28orqlBav}|o3{8E(FE8UuUc2w~h;$-&FlYwwZY9R(&t6c^u
z75}+~W|Uoy^n~szm77ElJX7+!z_{W6G*KpNZyx;cV9H3wMO#{JAwb|gEApu;a$<8o
zefk8~>by}ci^YuluRQO$`uVism&u;-aI_Wq6w1`T(^7O*UV16yItk*mahTJd5(BB;
zz1Mzze6aC$#<lJj$HuQRN%S~#{``MP>RH#>w&f$^(z-FeL;E%B<0VuN2v$$rTJt_c
zcY)2>e?r;l<Q{g+h={D&+2W}cN>)1tOZq<CKp=hZP4hpNY;k~$=`AI7%G1*)Uo0hM
zxkPbZAlhl~Ng-E!Jf<A{_kU{ve8zj|Y`V0{g-cnTbLVp6+M{&;sFX^{J%;(B;yE2R
z)#4afW?<j(XgJ$=2M=aWCXH`OxPc7i;-r<6t7k&to?o02cH@`*zAn$gt3M~Ou4P(`
zBNx^+tePsDlxjG8XVYF6He|)@$8zEJzK@T0v@C7c51)2?$K9k~6>n{hz0VJv<1jk!
zmOf*;o*gFtl0Uen6jeH(`R@mf{}sM;&aZMJ>rL~5!oou^a^2@xv!A`#{*a9grRo+I
zTmZ#czK5LdzyXl}lD}RT-ALL5U61Ch2CXD(KoS3fM#c4_nw^#eU6}L=-8szD^puhc
zgOlSOC)-VVUB+W$X-2V430NQSNzFQO_tD~UD(fb^=QH%c^Omv|bq!isMsQ9ekQ49Y
zKqe8*&v94M_3L9@TwH1o(}%xBp(nO*y6(}bQv~S8g6p_(2Rhmu&@v#mVMCc09PO!n
z`x)xLH*Z`HQNZ1WoaFSFj0(}gomjPed47V63(bin?{h*7hh0KX7q&7k4&2qsBQBn@
zzot;>$KNcHm#KnBiGFyZY`^q!QCV$%#E;<L8qG;sy_}6S0OYfS&}@emjNZNIn#mOB
z-?lNX!`New{#&ZC3jEQV1vnNxsM<%3qxmp7CU!D7GH|+?N5;})x3>CvoT+e;QuhtD
z8e=OW9AdT7{VX$iZ+K~;?^LrrypzNMOg~&dYyC*Qc1b!H*|y`!cPk_o-A$)}b;l|@
z2p29Dm9_1|{Q?3Gv0Kl@rL(+zXKet$Xxi9<ME2`O9r(%=il3@gJlO)wgS>dUs;byb
zg!F^SD`UO}%aPFa4s?&ZyoruIgBD~A$fkHmmk%F4pxfu=@V=RoVEOn}OmB)OE`t{J
znizC>0e6B$q+ESUb%F)%tjV<Szp+DT{CEzc4FH`gSbcBtv7BI9UOYle-=0-fRWRd8
z`Gl%L{rc6Cl!&CHZT1-$)CFaMBxZDG-^!0<_0QhfqI0tU?4-nC`RbX!T3SxJ$V)8p
zE4*<}V&28a+7UO&O_ygMhHmZU_T*4|Y{^_vx<_X6w4@E+4I6L|TD9~hQXqI+&O_1@
zc%~J<x_tZc<tMH1Q$mxVgemC3K=foC)EULjNc&~(BnEqB>MD%9yC#&ew5vb=Y^)V-
zI{s?Y6&05mk0=`Nt9=9ctSgPjgU@Hwf1=Xx-d)7!HnjNGhBJbsXjNiXra1QR;lqX{
zWg6+RV~?gOIbcBl{tv3(I(^1<Y}vJb8bE06#snG;e_io>+j~KzOXs1qvbPU*>D6W&
z!75u@F7Z;}DaLzGJ)Un|;D7ViuLEeuG#|2<K6?Lbo01dHOLI#DqD|9TJfxe_a(0x1
zU&DCIpOlj`Zkm<>MUwWo_b3iE&A2+3cwwG2eLX8~*M>c(17RY0__(X<$)gwI4yw`}
z|2H7YuTKB=5kT7z>seQJ*{X>(gbA+j5hn9`A3rzobSran24BiW<9~oH2s1b+sANBf
z#P6G%8H3U=gTZg2)~R<-oc|st2Np+#Mz7w#|Dp9LgBxp{5pQ^{+_mE|QxB60ix6)b
ze$V2iEt~nCrcz@cT4Jg6gbzukR~J4F+k9@8rsfvQK$iaYbWH@tO)Ato;5}CJZxv^X
zy~JhNu@m|Czb2t6e+%YK>I+q#`}+g1KK6j>fPfjV`wwA!;eylGUSco<RSJ{=q~?N^
z7VmH0cypv9>1AJr;1{=y^76KJa&{hP@4>dfn($?Ht<^B&zbEZc66UeRjruAtdr9KB
zX_9tnoaUqC5eY9g<z8^gk?%}DKe6@}j==p1X<imOb+MmF8hfmdNscKUc+Aah<D^7N
zcYpt3X>UE$sW2xv$)#ucF4i9zu6ZrfV!M2?>#21mi*yI0<C6%sWh9=)n}R|)omc4J
zO?bmG7Xi^o;->%`zd}n0aiP+Z`zq_*Jgu{#g^3-w0M8RA4m6aXJLvkGXV4_<Faqb2
zEh#nhJg!q}X}KL@jzw@kbtNaCpTCO6h>vSX*AR6Ix0pge`{@K{OUr6{nV8iM95!qY
zL98)@I#Y<ajC;Hd1)*zgb|gvghoK6;@S9k_g9fO%Dqp65|F_U4biCSACdoSFAWf0&
z*Y8^VmGkGP<AH<1W73xRHnIfpiu3^3dcxtNfrX@~6)k!9KmP>LM9w=j=uRke87sq9
zHSptw`b#g`%EW~E!Ow$#J|ihF8?-8G7c^C3+rj&F&COe)Oiy;}v(}b5Os&*9X`#M2
zm62}pd*9x@S(~!1=@7koUv9NJ44E_7+8nEYskEY|Zv%!8zteb6I0-`f8%*myaf05>
zSJe2fTCg>>WadVGa<ut6c5q>0f^P@&3DIx_Z{ED&D@b{pzx1AnnU9>SWZn9j!Kksa
zo<A?5`9tL*R<iO4-ydCobLbn0TH{YW+5X1d+0uto=&TlPtB2gjTE@Dgt%cr`nBaZ!
z0{5YxOxw+6aD{MPc(78bbnCIundDN&(0=o8q`%!zWHS|x&4FkiA3d_)u;BwwPc#&v
z3q&vJ!2|9F=S|FfYUKz&;T?u@Z&>iT^e7whFszsjKMF8jv3hl8YO3(@pe05{^@Jrj
zbo)@JIDb^>2xot99Ou0vpA$r--b8nvViot-KxDJNzC%WgaDh#MXvCT}>)pG3_|9BZ
z+@n0X-HB1z0S`wE8#c+nX@LPXGJoKj-HhcUfWGeT?r&Yz^)+F*VqMgT)(W~&h`ye7
z+i`q|2~sxPq#l&?BN6b(bUsw7)3{}%YRja^p0~en2a79I7xx*^CH&b(2PcW-=aqAY
zYD}KI%B(N8nDLw?8#l*}wiwgo%_{}G^UqUi9_BrJPjQ!zl4Db1mfGy!Ki9VWY+aaX
ziBvpWyRI0saBBYaE#Y&9e)=WUG?C;&t%1>hdwS1o+1YU9$RB%0)sgHcYA)=|DmS+o
zGjCnp*C-G_i-DO?+uFY6F4<hL+?J<Lr!5H%6ZXrI-9{fha)ievpJ&)_#F@=3Z)6et
zGf5*73@U9`)1!-=RAN6uxJu2gf0mL`M$^daj<1QzK~g5!0HI?;@64Q>CLtXBwV}J`
zVwvN;q_{Mu)`Po8R`0ZpMucoJ=xtUS|1IT~`sHZtlzIF78!;d8_S{SvQtI2IcwCog
z1hQ&8o58*cszciuM3)@#(yQx(Qt8|MmkJQ})m3VR%Fw6mA6HY$!{kJ*J1R_0t5oN;
z^x&+OHaXd08!9^|oJ`TaY@!l-pz0rnHqb{*oO4|#IE!a+sJ&6&^_HUA<c(F-V2KUg
zoU<dgbb_<IR{pOq!`3)v?ixIGZmgl+U?=pvx`7Y*0;;O2F;@Hkv0D-h<qY!qoY+s2
zX_Qf$A1p=;VmH>TpyJ;zS>?s<egresN!?AR3?06Bo1oQhFgix>&s%nov6-1Fo-xq1
z%SKrUS9{RF%XK?!t*udY1_aCq6{brdMCT?IZ=Rh$fQJ~fLn9rZEurJv{Ce3GUcS~}
zRZ}~{L)YF00-%prKk8OE3%+Fs4?e&NZh!Ld#hVFxL)L_zHu$3ZmhOwmvo{0H%*wFI
zqC}~<?{PLEp%Y47EYL`sI&Fftm>)rc^c{qrQM(;jjZ^@@Pj}pk=-a5d@SLIZJ+gxs
z=mE$!Y$^}O|BTWqkGs6Xc#E^gww;G}pjbRX<^Nsn{ZmQ>Xk!|Qei_Gn<i7?+ojLRR
z|6=dW!+Kob_utH9CQ0TL4GJkyGL&dSr8$+%YAH%8BnufcHL8SUXq1pfQxuJq%2bJ^
zq@<z@A^Y`YeZI#&_CEHXzdaoL(;sV5dcU9d^W67!U)On_*LjgSqcT{ALa9bGT-6Dw
z-m0V%pLYBC^-D<Z^4G?VGKYH%cNsHg44bwCu+@F-O(N7+f@Vd!yMdRUo@$&&KR|(7
zc)O<=WV82=96Y$8Dl<*1FQkW{srX?>4kAeG+;6Ri9<HqXg=?G2t9FHz5Tfj>=ZG(@
z9`wyfPaiYObwOtD^Jk&35Ut3Yb}jvOt@icn!J|iy9y8`+ZSBP!@uW|JV02%?NfS#F
z^I`jP<0!UA&-Q%0cRy+oUx#J!gbg<|SafXH{j`oZip&xmt&vVgOrL5f*tGM~*2uw%
zT!see3&sP2e+(3c0M&PNYdrEjuL1K;8pYn#)Lh662lasQB<Yx7*IjpP&H|waiwsrv
z_dm!<P@56`g{ik>rqiosC%%^x7zjactm+u(>crvNL1(9uUFQ1lp`vfjNaK-kZBxHp
zEt98(Vq2BM(~4k3A_*M~73a(L=IXy0)s1$O!cy>tKInAkAj3E90e&D&g#Hs)QmhE6
znb2_&7z8#Ap0RGkK`xYlCa5@g2EcH0c6^&UVZyyD=}LUYcI;Wa37o|7`#Cv?oi)AN
zTM0#I&*@emx23??K>f>qSA{NkhajBBI{~)?A;7OdjTb5`OyF|C;GoCjNkkn+64F`f
zWRKu?%{1@O^fw2gBUB6h@HBaFkwPC8(r1nnDw;~=K3kHBWWw|~Zg^HOEqwNLPhSZ;
z=nqa&3XD(Fi)8aN5;h)S0~H@^6S)LRi|tIug|-1%2OA>2YD0|KqeqX>sV}l(vBRcS
zQU#Gbb3SR)Sbl5AEWF!E{+Q{*0??w<L}<bE`PE=gP=t9CZDcy=;YpX_-j{wiHW#{B
z*siCs=|Gc55eLcOjsGnf5=A9Br!e+`R%2BTSw^I??(#yq-0vn*=9CmReTTLR-<~+J
z776VO=og+j(MKpYI5*J{v*LtNsW_mcEoRER8I_Q0$C96cWb8aWXaAX$|Dy;IU_}D!
zhYKF<5@5(sFI+0Wu<Z80{+71o()2;ZCifN$_*I>JZ_govV9dQBkK<_|0BZu%;WtUA
zbMy1vJUoPEJx+WwL?krV^DGsvZF{f+fD?ua-86ax4qu=4))?4uq6tj#69hfpCVz2c
zl19A4TGod$7{F>0LqT}-JQ87Iow~=zc%K*1#7)u>40=(of8*V;jB>WL4dFRrup6T>
zgS+V6Tz!{pvFz1+9>AyO=4ie*Z{9p=YCo$5^L>B|xb2m!X#~|m!_ocwD8n*Xr0fI@
z1m{$0zUeDs*8a4j+6CMw;Vh$hWX#Z^ZX`Hpv-x!-H4rt}BgTOx6YaO*7B<=IprB&c
zx-cOby%gtAc=I3<6qe_VvyVQ!eod|R2^AP}s|KDD@FR(siN{G`EzQ}p$<u|7WjebJ
z?E06$dQ68A13Bcl{lLLQFRIY#z-plh21WM<HMJD_ZR7Ar>FDTC56C8IrxRrq%NhNc
zEBOr1RB#><J|iPTSXmhv8QgLr0k8Os1p@?C|K#l31$(h{dew6g=L5GG(Bd6fgExpQ
zpYkR&J#tW11k&u;vu9(bG+07H7_-a=w^q=a&$n^e$>8vy|HlboDzAXlPY*{9xF(=4
z;r1a|AUSCGzB0^w`SJ=<YL%4KjEsiKi+hUcl#dxO@a)Vh+l+_zeR6wwpF>BVzKcJ2
z$?)ii={Yav-`olSkmxSXI|n|cm9nO`)-j>gY1JyXpWik+Jy>JMuCkuA4X?`b0eM5{
zV7btCU(x6CVJ+j3Hw%8d%E+qQ1zZb`h)76&91T6@S+n(|P4Vx%a1D>k=6i`K<pb({
z#D4WI6b4SqO1X?VlzvB7XZ_k~A|g|(h>gh$?_)$V196viOXlhOZQ1~c=FiH`*eQZ7
zJlz%!@nLf6FU2KP3`<YVj7+;glF7oqYCc<>gTO3dQWZhxjrZj)mM73@l)o5B`atMO
zF`u1y$>NQ|(zvf=g>t!CRpzFx6IAv?kR@b~TyAR%b`U>o;EH}Fcv6Wepk=q7Fo#9}
z5X)5wo_0%#O<_KaQ)xi@n_@mSII=B}5BIJhbXCHN$%3#i%y%*SAA$cc^Ubm|gaSW#
z4NR)}!R~jQR)TL@JDF0sR9yKqXR0B$14;HA#%k9F)3+k#xM8T9Li^ybTp_&9-5H;@
z1``D9WfFf|^ne`}ITVg^(kbEXEP`G^fuU3O=k(ly0>ProcP{HVJ80{nu?Z^e1Z7VC
zja-j*gB<s!1tYg4^Qq!)5bZ><4bFmmr07t($clbowuSXO<b&!-?P$}!!p(_1_d7_K
zN?v&DBhpu)cvr}oU@qJu;`YA%`<<MrXdHOBwlv>}$T|yD57b1VBHuRjxp9(?zP0Pt
z_4VQSOJ#bNx!;4XG({Di)>Qa8YxTq9<aRdG#^gUPki$LGV)@r=mSK!qv9mkfEC|c8
zgY3H8c`KX)bea_^l1O-w?IyqRsucT`IJ>Yn9L)qNDPmx}#4*d_V~<A(Bk75x>zzZ%
z*eEEH-*oX?y{-|9uwC!zGE6^v_Rb6A%!?{M6W!TeKEFH6`{gL(HhDTcJnojzjL3Om
zqc*0m?&8HaH@)87euI(0XpUJi?%G%VI8({die4Ctc#bAeS!(QkTB^_zbm{qzFbpne
zLr9SDJAZC$<Sd4r(l;@`{AkyyBZP$`auJPP#nX*DoEeYXu=P2lxK64olhvku(ce)M
zG1sT+#rQg-)(PFu4%M+049$eE<Js%Ch{#^2zliK%&b6&#vZIP5nW%<5Ibn2H(IeM-
zitLh^22=F!mzD4U{(K$R`R}(o@5KMod$P5BfXKWS=8Lyg-|gnKh?{@t80U+=^KEh_
zOa#_Ex5RHxVVs7>%NOIT%hoP4GFs6nA(AOfS^adk*H}vp(LayN`^0eH4%vSvY_e1*
zcdnVvpFNa1!c4l(um8($YdG~1IXiQGX>>6ZV8ZAgB6qZa17Wgb(gX~f#EvM(R6Gd=
zu!qwNRvSzM9yOX%b~ll*zmQ7Dk7i<=h@A}S0Fs3#6|Q6<=sPUs7=vguiULhz)~pTA
zLemV#(8dH$k$M07W%}Q>_UHNhFI{W@-P@fv`TzZaZB6HCBVu$N;7w>DRZuW+-Y;mt
zH`4cbh8(t{gEWx_ugr+inoETbl_<Tlfd5td%+wGp>9<2Sx6hTHGso_A73X_^HZvPV
zn(NOA=A9zf+Zta!-J9UMckjt7s{CE#%GAzS(dkl34jFb{Zs5RyU=#r=J%k^d&+j}4
zUO=iqvN-!+gl&ti>&j26gBEuFXo#WNQ{O{42@(L@J6GAb@#Dsw<)~gb<4t2@CBl>d
zsvN$`wf_79=?tUpH?E!liIkF*G)<4#A|kRv@y{<8eoI#}-69-BM*WThIg9MN4($b_
zBk{ADCZAGzO!l@2zp%^f&yQ88pE2^jQ@uaj@q~%+lPgM{pXvCTP4X2F#X4VIZ1kU3
z&#%s!Of{AxwpVzEUFY_I;4lMJV#c_LKcC@H|IXhv8X_s_xN6n1=>5;0KNox|)*cdW
zqsU~xKfmC)efy90cJcc0B%VkGRhIqv3=W;|a#q_x;kr?=a0BOU{_~;LVQhJ&_i5Jx
z0+x|C6-o{iUe0oekcWxXRlj~+OopQA!H`h{a_#@ak1zT}h|8x7zaYC-R~Y{}FNNxM
zmWD<{MPYXK1u8f31Ai`0PyvXDY<V($?&WniI!|`H@qeyqxOZ$a7hySWi10dBr2jnj
z!$ry3)GI%K{ZbVFb45M>Tv2vzZVaW9sK%cQx0=jASbnWDjTLzu;%dw5-4ecaM*)RB
z|8FfLKNfrrs%ijA8ycKmdv^Y~>z@m@m(dSIB%1Qp_RmvK+Vtm+b?Ll1qu#DTHY}~l
zqZs+hZP`~MX_S<d1kiBZQkHLc%Ag9TbIZ|rvzy!D0|!VfMMV1k09;0yv$2tXa%I#7
z6~-?LV<jhx3&*0JG1YJD$S&1w6gj%Ayk_gj&MhFkfymX)tLpsYxWk>-|NozFR~O9L
z@gFU~{|9cwhj?KfC-#F>Cmssc(Nt!X{Rqv_5njJ`&44e4#~wZ`n5hAr3>Y=D`EBk{
zr(Qo|uO=rm8{#xV0l*-^JQxO(+)5a^2jU6U1JZk&zCh>#3Y*O&I|#d-9te?AM7Bd%
zBWG;@<$x(FK`tSi4>iF*Q2BTI401Wqrm=7vAU_KiEMSz@cv7{Jq+4OseauiPA=&4W
z0B&?Y{!=S_nPYwf&lzhPO9R4o3KTrki}LfIJl_dN4<;zxl$W0mtuA~q?*c}Qm!@*t
z0ir}%f)D6xsxd0hoY|ex`%c_ZVe#sNkYzFifWrLcvuAbR2ie%z%oV<F^ZVcsn8LkW
zd*q7e0YASFR3MRK-qh5nQuKA{){`*`y>c=i0t)bj<^UiMnAZ9$wue9)P7rxis6EFD
z5nrTPKT4S2d6ljPmx|P4Bo@Ei>ca3Js{8oS`BU`S7!VF_l{fQ<3QBkKanNabAQ<8^
zw4+>oFYIJuZf6KBS?DK`#GQw7u+bP)SN;Ayh@-iN#^<7>&uwi2un3gYKjUKO8+*E+
z;3Sx*1tcNdHyfK<viAMaa50KwZ`v`Fa|;e)$?=QyvUbVy&ZFj#*j4_&KF<Dj2iS&K
z&(U+?3bDMwx93YsiyW0YR|ob<Uy(4-4#W@s(_o>^Pt<(^Y=uJ@ODzT`L@GCM7SJ-_
z?WpExAbIPFncQcD>qW{BI<HQwPmfO+%DFUBs9jM(Mu}ETVaHqqGtu-8iYlT@*}=31
z4BEgNf8$d%S^T@gMqZM?qSqfy>rZ%ic!e_x)Z-dc5G20oq16M;k^iy!s+$?xdcn6s
z(D5*!F$GIL;+kmF-ym{;FJT>K9L2+uVB%}0=-&~cOVzsRIxtbii}_vd{C+?MDAKV=
zSN9=pU<zIqP&q&$FfBiRes}W~$1QQlm@#%J>ex0c&w4i_`cs5y5j=M|gYT#gn41L(
zOYaHmMX*MP&Q;q`a{u_G2~nK!G{o?^1TYIj3jkZzh0qblhX`$!^>yWe0Vh9=HFE-M
z<GNil5I0LnNojpj;?#z`cv3Bb_Kyx_guz&@3!U#=L8qcKzfab^d$fA=n=?vSH`Qk6
zj;~zDYOhx1Ny?<T<UqED{@ertQHc=@6@QVzE1(>F6jF-X6hSN$zc;f+;!Ne?keZvD
zL1faRupXbnXBAibfn~@ckD|Q1lC<tl>(@uW{Ts(Q?|XJh@Td7bUR^L4p0?=ys3S@-
zs1bp*{WU5;hgUyaCR_-2^<=lG;x1xq1WvsAH+qyyT(ZAMc217qwK#FM`N1JF7nh#d
z2}ZzL36V9{de<R1SPKv~F>1P-;sY2Aa6=F#xW?lz-vR&7<fLSE^?dmk3a-N*G@fHw
znDS{p4n9+=nB$y^Do0Z*c+=<F#c3Iac0nDi2$WI}>A}xb1L!2KUDT8kZq?b|LeY2U
zx3NSKV<6wq`?PE{|4d0<-Vtd2#tq*eO5jk0F;}Qh(e3xygXp4FoXQQ~ou-`xx46eI
zgh@iro>{gO;s3yFynFYeV4dGI0$@D3R&nGc&`5>t=FOY;CC5~QGIF<rGJ&3os;9LH
zSFp>Xw~$73ET!i_I_2HFcS5|_+Y8~c71X1{F>m;C6dCus<pi+__`FLWMF=9Bf8uf5
zoKSU*3|CYpi$vCmKK&}C=Ygj&1|6N-h-6LBz4UiBzl*f<ZB80Js&_{%>dr$*@;KAQ
zY?Ra3J6|6U9&8kL=$X!hL$W&aq{R;?>c=PjeKSYmgUpQuQ3rddP7AJf5*xOi$zQVP
z&RHHj(6y`l&9B`){BUTh{7{+WDmv)%Z|~ijh>`1y=4gCo<hn?yTyG)k`k4U5p7|1x
z`gBOhrPwH@D{tY3=9NAM=yjKNytKj5k>=(rGf=~w8M!z(cQyZhSnPhaD1-Iu>yWAs
zbAIVJV!CGvDzB5$dcJYP=tya&_OIH1Dk7q6^8(@4iHJlm8zkht-@=oVRhz!Bq~q!b
zNlHq>&+T~}1xN~Uq%va$J*Fwi$+juZqdj6|@_1P>-9|0{w>UF%@|;UdblI@s{kwM-
zrl$98uA;SwSEQcL>{IBqs;2vVgi<qS&fJvp`Jz?GdTg>dmEJf{k(cMg1Z0(&%c-aY
zdSt}JY=GM@^;rJzRrURIRRiJCA~RnSXkyj#AjQV3)OzFiL?GzXw=YK*6W_f!9|8WG
zYHG4G?q4{4TJSH3-M+=%{uNu6(0(1EoT{n-hQT9NE9PdD%DZ&0KxI%|d?F)4ZR57B
zTfYDx#yze3eKjRzu4nVOCT&EVKbv3gSy`rk;?$|htwx;6W&fv31GJ3XFLYSD2QuVs
zx5p0-U{c$(m#i(up<5=z?(NQc>)G?_jMFnWOUcOi+6We%98U2^ut;Ep;XCo#g#~Mr
zpF<L)%Ai~F&3Ss5>!hvssiC2vXTwkS>9FW!>A<Ct`}_6lSGsNezqiNz&l1`HSlPe(
zd{}3Dvqi5b8bluGz8mB&H@xj-Rn=*=_{pp3U#mDGt#`|_Wp_&Ei4%iUnu!o;GWc6M
zm7AQ?GB_rGBQ+ZHaa~<q9!u&#RG;3+NkmJn6qQ2_EBF6-yItyi9oS&Ie#<H<)+|1w
zHE01;fSJ2kRkiZ`L6AA(FKVF8(I3<}KYjdokbZ4Z5Vn$x=B?Vlcc1m=qt~h!t*xWe
z!zl{Mgwy)jE?KSEA@sOfa12o+WNxO1GH9nNdV&)_#>Qriyn9JZTs-y%C!FwQl<{XR
z2;QomDIWEudNOSd_4S`}E*_?dy`*HpRzqFg>PlxvNBNe=TG~Z1QJ<aPjovMs1tdJK
zDbqiv*VWZwHox0u;osRhEb*}t6440>h2F9=XL<sBY8|mqJn?dvV`1*U;FWuqEaAO<
z4j(;QTUo_9L@6k1f1B#D>5Kd#mA=EOYi$QhUDc0Wt`uU~$5(AyKAP<Jor}Ao<p;+R
zc}dVtu~0){3KK_0%finB@7ZsJiK(gdO|3PY(ap{Xvn(8{$c=@s=<t6(BTS@c3P%Js
zk=Eba-Q3)~QY_L8FBo-+6iwUvsccYpV=F7Uq#=16E0|q8dJ&zNI2SIj-sR(=sP6Me
z(YLm(W9Y_>jp&#V*)-3oZ)%zgXH&oM_ptKwdP9NCtM<cv8cpB$=Z$w+@107Pb>|hD
zn{lm=;gBC#>n!VnwNz&0YJ`hi!HLl%E_XAS4X2$uQSR)yb1Ug6u4puEt$6m#6Uk-8
zq6_P`#E|@8;edWA-}``cE`KaShUHOcrfQYCX}pX1h~49aj+<45|Gw>u-GAQp9J1_l
z=RC(cVwy>7ZQ$GmJ9f-&`T|)c+K50hfCSPl#lqZt;>3yaucUi$NQG$Z>+E{EdR3By
z?DA!$hDZbky%jg*tz5fyYeTsr4XPitUDI};5=G2v5?^>|v%5R|qj-~cO-&8|Ha<7!
z&sEGznAbUm#OJT7Osl#`5`4oj6}^?31P2H;_pMvkZe%oZPnYkstcjYb(ur5~vl`g(
z`Sa&PUPcKMW!T5-HU@Tv-qC8#qf&ly;pa`Un~7^~-c;33p0&V{HWQnft_fx*Z{GaH
z;f?8=r{HlaTRyxK^$_`{_U~m7F8Uk^FMs@4m~F%9eSi2D6;)MXx-#LM6fS>pPpQD3
z=Sky06)GBe4NGorlTvcH@m@8(d~!)>Gpmt@4;?zc_qL!Uy?5{B^$qvDWo2Zt@7>!o
zGSy;iw{Ck_7sxRlWo7M7nWv|hOBL-HuleVO7#$8QY@wcQJ{S`pk7C4i!GgW`Wx8}B
zyC<NcL9>IXR##Wo)QphMrGZD4<P1yzpcRV!?ON4HTb%vj`=K7+&bhuOUz}`+G#JIf
zb*%Xb!dQ5np6fHmWg`*@vg?ejrOUkd?Zr!%s^Ku@{8xGWh?KXanz748lQ~GJgJ!6x
zyq(iZa!;c~ADMf`!!Y3_f5WYeUQJR}>N7}gqw_xjlu9e5|9p^W1zmwOPOfBa?|SjN
zp}~f2;~E?&aKIIL$SGoX6heMxD&lK1$K^X8OjPqZIY-OKERCtBgh5FrVfFR()-`<v
zgPCmm!@q~<g(GA-yW^M2iQz-Z%U+F}In&a2lDP;&l=?IT;4)S6!dXn@j?cIlp`bek
z@bD;f`u>bV7xmF79#WfdXbbrSw=XUBMbuQeM^fCy*|RfyN-kR^%nw7!^TPShau6vU
zwO8OqzBQuMf&4aGJHfbsLrv^-ThQ_2Q;`~;J112(L$3?h4i|gFbA*hGM5)}cVZ#KY
zZajdg>YOqZ|9zrvoq=W0mfC_56btZM@;A0K{A=4b_0H{pa=eDrE>5HR7Xmo&^zFYp
zZmwKxGOI<_i;k!jD^?s)U6?=;O_52CuIIm+v4?+N+EJ`CO0Fakr%C8zU7du4gzRRK
znR@;xLlDWJsPQs1^fO9A?*?iGUU+8aW-a}xn##(|8#}>0Cn1i_8ESM5`X1Q6ucpvv
zjNm=&uVZ<`#Kn8|=po>T488?eOs?$A%ZZddtdlFbfL%Vj9zN7Xh=)oCi0dUaY*Eo=
zQaSV(m#M-pUR=v2L94%V<r_*QHk?qBA?|&zN@Gpdix)4bHP|ksk)@^AIhRpRu&>E9
zfP#<#p<6<7;OKqy5QY0JZaUt4o@4v>?j8O5i|pB}S51^c@@qcfast4}7qUZ8L(W1p
zEX`Ccu4$P!UiSWTqmuO|Do-Oz$!AR@=9d5YrpUa}!$*y(;2HAVcR1=B7__yvp0q)V
z+)h1MX1-lS<mB<ve{Z$$EAs|*-tK>YTV!|VbN$cX26aa9|NL#&xXuOipTBk=j}83)
ze2wcZoaO)XHNP{H_|IQ^YyCT;sPh~DeM^!5-dK_UCzj6!xf|i=PWJWGYGx(=mL|7U
zR8aWGSxr?{b^7$U?_c=tJX>=ss}VzoE_<24-=c?#^^3wt{}oXod9i@ECkqn_@a~=R
z@e9C|pn%T_7w{T)Yx+Kuk~VKl0Z=)+<jw)8?f?1u?{@gVXVaI&Huk(THx~i|2xZZr
z1!QVSz8o`V{Z>c&C)&UNh2LdQo?z~2DqyKedLrXKqSzAT8P0-1C>7_%<X&fM+vprl
zO+7j80Li`&WqxwyDW4E99)xY`6DI7bDj|6njK)?5h@92O+3y?T@sl>_w@`Ul*fa{u
zL^P|;s6WW;rL0t11d=O*eIkS$5gjuaQ`IMaL26%zTy9>TXy3klVXJ<3c;j?<@!~~a
zUmY4Cp)DGbd0sI4G50B1Ge;kX2UrEZbNE&PbBT+K19Q^j1w544*t0{CaT&m>q>-Gj
z0B_QIVJA;Azo`7~-AFP!)$iFGXkIm(*tbug$duVMgj1mimfA@(O4slPoJ*8=pj&zd
z1`1QAkTqSqa;1*Wx{#2Ng9r6Cstqz7H+gdWl`F@`9bnW{=~kN#cNz!2a_f{+3Yai|
zL$X`_aYhG%cW`VAlc(0LL+)#7X?YzHD<2PmC3(~C=+Bkm9L86!ToKe<3@kwC1}`EA
z9UGpe0Kl<@K#jm-lI17M%BBJeZrRd`iW9{Leb~KsPenO1eR@O5g~Om5NFrn7@1u|x
zj`+Tpc`V951a8Pk!EvtA1f(_N!I`_yml!c30ynQ4;oXe#Ye{(hBB}JisrII<-m&8+
z>PoM%FYs|7f!jMEiJB2YCTRv#kTCA4%w>w{J?=5G`AwMQT)nFNF-m|j4j#-!X+YVR
zc<GYG=fVlSr6x^!@aolXK!bGyHmoiwE*?Q9O@c|6-ST@ISCQ5;>eSuRQsVQEtko-4
zU@&={oS7Cxi<D17Ps6Ppp&6AqIcrHfw6-%jtR0Z((r?layJW{<PJz#F8*!~@0+(Vq
zDj*{BmMy!+ljU07U0q{e-{`VKlG}%*w6KEpxz5GKr7DX=x9Y~mksjMYL2#~%jEb7D
zB@jI@1HZ1#u)}3^Nu|v%X@TEF_(;|9&U*`vIKakMAntS}IHfvVPp`O{TUaQJG^C%3
zlx?qizd+>iOW9jMgFSopl*%Sw0m>CdkWsOI`s8}TpX?kFsnDvpMmf}I1hXfB4v&8S
z)xUp#P1@b*$kuV}$%9Is;M=2DufB~z0Ra=PM(IDE(1vdE{Kr*gR^0NhAfC<55_5KV
zZcOo&)^nvQEDc}D7JIqzyT|LDBLq)P({DTktydLH#yT)+4PwU%TkLof_D$;UBlKee
z1qPmJpy&asI9F@q`Sqi%larI2qKb;jY}dECg(=7VPoM6f4?=&#q#7>1EcQ2_O;Yk_
z-$O!&%no}>+qGU)zVv}*pNSo^*ZDm3OKi>yBV*$vnnh1UkIDyU0k3h2_o6fy^E>F&
zspYF4m_<@R7&jv7-Pn2U{zo<G2T!^2@>7%0v|gUF0a#hfmyrn;kxggmHhE|qHe+L6
zrHqD#P5-66JE9mn2XivSM~e*wvTpPFNPK=}IGLcihaJ{OWo*2?@<8)zk{u`X&Z?}`
zl#$Une}au4pV61llo%atbg#j<#|f*4T$SppG2h<e_j_f|22jp~Mux-x>=i7L@PutQ
z3|LN5Qh$x~eji2Wrk{dg^4Zg{9mv|^&gq<$oXf;&?17@=@0uMtc8o{j`SV*+x=pqi
z%}8uZq^fehpFe-b2<RGVs(Z>Ig*kJo3JV+A93>vPB+up!G`6&Kagu^$I{7=V<`Z6s
z#!0aq?-83WJUZ$-d1>*Up*-%nkXtA{e?gj{73kNy_cwZ$@c-3naz%JLaNse9<b-W-
z;MeMW^nN;*s}dDWh3@;r7tVc4!&+Onl*bP@FvurLTmKzPiN8qZ;;!_wnVEFWL?SUg
zZrQ<hgq0!cnmEN-)_Z=v7#*#<UMleT+qbKc3`BlB-gR9xp$Rmr@8VC~*V0nK%&|a`
zW1e{rM(zW<vu5SVis#vS0D=+elvAicEYyrA9MPE3-!Y$(K6r|~U4kt!ie@@PnatQO
z@Lr$4e5s8%ta$$FRaYVvHV;Pjaj68dipolSaT9K7W2B=wYnIh(!S$z{c~Z5(KrK{u
z@|Bl;B>bfp`AN)Q=U84^dWucOy9zS^zztknnxMZJ81<e0!^I3w;#fg-H!o-h&j`u~
zptUi1boJO;etf~C<O!1A3y-=w<S~t$C2v})WF$;XW8a!4*rLv4YSP)$waw8#Do19&
zt*yQ4d`LkKv%<_Q;DxI<s>3NfI#IJ|AjDr*HhWFZ7;FYVa6Iy?hpL={0*!BDJTfj^
zs6k%KYi@xqp=KXBVnmW?CSjbmW7U#{D<OX>oUCnaD}~WXRPX|Mmg2ne$;JM0^K^8w
z0cT=|?XVC)8Ucx_bjEFqPequa$o<Z-073exrDd9$TB7qW_J#&t`O1o=Wo5Kks<<5n
zI-3*R9Pk>C_}5uI>RoVS1ixlvMA_rV$8AiswO6sLh_E!}M%_R-jb*=@>QnPwK38d6
zq*70sA^*gy`Ri8(0~lX6KHHTII=rbw5^o~jHywHc5P?Y>4;Rnh%Etqqd^)f1@4=NL
z!}jfCl7l8H4`EaS0HJWTP*|sCYz#A6`<rvZV%BB*i2?47FWjlwu)U(}orQHC=!&mA
zT>~rV_ct=5vjvP}wX>k`G@2oxc+q{N`q&>Id9$3ZMw2O8f`3bfqV&Ovy5#ZGrkzL@
zIYC8AJHSIFskCA_0UueMaG;`2RZdp+zz;{-FTPP77OX`6|5j;(Oh7JntZ^U3Z~6Q_
zo$)Pb-@Y6U-Oy|@8_#zOKfFv?TB&#*$EaP~w=7<~Sbk&gl+5Kyu#VU1@@G@YU+AE<
z%{0x(E-0;7&!0y9RgG&L9E?dq8Yjh{vMxB%QCm2&R`(fgZgSf-F;29x39o7Z0H}AK
zG+n_Mg=+f1O9SXVy7WRo581VPQfk{a&IBB9(7@<zc#G`<*QjOmkKWCg;2QPxJVm&h
zeWRcZFY?G-mqQrG4lp*u*ZuZ(ZU492RK(m@Hy=p4<Mbabz&|-`nor^ow%X(dp0|6G
z*i?Kf{jrh*m!zj^=pqyvEE_KcG5h<w!@FLtwJ6YX-V*Kb5VpxZH2ASz%NYeyDHulV
z=XCY;OQ0xj-1ySk`T^w<_u4(~T<~%q_CC#GloBjm5V6L(y6B^8pGx}+6~$<AakS8J
z0#&yhn4kn=k6eg`K>2P9C+pv;c1`;Ul}+q|lqQ`6ARz0*B9;sYUAwqdmfb_(TExNO
z5N1I^++hbNCk<McSX`!Y{v7FcX|6ILmztV4Gczj+3gT(N;`qLnl9K4WJ#az7?uQj`
zK7L&CYPCrL;%<5y5(}bLF8lfU5j*f2AT~^+k0Cz3fb|bZ!HGcsDHot$Zf|EN(0$}_
zBo-tYWN)Y-SCK}=FBl<(?w(%Zkz>XPPEFu$D8Rw$=nY9vPp5ClW}9ZL*0J0**4Nh$
zo-U&@e~>Wg4Nhl<OI1OIuK!ztDnmvum5bY@Z#mfZd_wOx!jQ5jmj=2Wj*NUw%t&?G
zF7#EYPWRmbQdV&UpQ-(iCoLzU)6VwtrA^-37FH_Q_cXVIh9{>AjRdA=Bj^kv9VY2*
z4d#3x8)8W>{rg47K?J5+kBpl?eF~#TE$}A2@WU0Y7x=s&k(>AGal$oISIO8tG}OS3
zkDcu~2%|sr&oMDf^Sg&h4ZReCav}X@JWqk%A3Ws*-95LA6ph@uB*T)_k_~oN5*~3C
zh?b)xh}TmAI>S&led%JQkTV({N|Pr~W^Od@MmP;|srKG&K_Ee``gwUm<MnnvC@;y;
z1tUdF0@vi;K2e+D*iY?HePd(7>IlSHthQy<HNysD+RWOb_kw4MQBJ)!YXeKs3eZm&
zUj-{4a491rBkTc8O#Cljb{1S_THjA!aYv-KaMorxZlM=~qm(5_o46qp_NLP!XT{-$
z8^9KK_n^NV+<}EpWN?p<#II99Rlr9fi#?DwZO$Cw@I<fTxY41qscKO7V&RFU&ODJK
zFx||Fp*VRx=l<+tPe`qBMZJP^9-5Xa(Le3bxo~>$cvr@(yt#OCRaF(j`W4Qzunnd2
zEqP}Ifo|TK1zI26eN@^}sL)=)7oF>d{;M}{ZV~8%mooLtULoM>40sSLuc_$`gqb;Q
zIY=$NGzU#pH&xGzjg{Z8PWB!o53shP_6t8SeyKQla`c}1x;oEhDx0BmyFZ1LWB!6%
zYmuacgfJ&!{ra9xD(T8o02l=~a!l8Czm+iQg-lyYwYA~tvF|2}badR`m-yNoGn^Q%
zyynpTzl}-zhfD6%8nhQ%CBW>!ZKl~pMMul_fqUxg2$5eDHMDy%*y@`%E6a}_I<k<)
zaS8P;vL5YIk-0^nvzQW+uT$UB+Pd01lQ^<ela9&p<z>UOK^gLPy+O($*jjGrQ>Dt1
zJ~h=rtFYg$b1efzcec06rAWIB8AA^1p0bH9@rr+(U*0%kqda%+nXkS_4j*3TF#($T
zuI9%O`<*ed?N(1Ov$q)?R53w+#E~Wb(oP9ecaI|tFe@`X=;)-fjd)7jsk^7MaH0OI
zl(T1dg?_7H1y(2%Io6%NuY+>#@m%HjW3ui`9IC!xNBZWp7waYOh0gaUMt}V^+1AI>
z^5nN=SWkB~6GTY`jm-3jyz%7S8Ymv;A7q6S?B-fnSv_a9)S0#|hqe;PK)~&$kMYHI
z)SFsgwcx=_*F2~t@n&U6wIl+}9&UY7rzHN%1UrtsADxA<TQ2MZlZ$>#mg5sImURf^
z^kR?{$98P|A~SNSR1Z}ssByO}f})E1{Oxc>)NdhK@XW0rEQ352<6`dviF0Bb=W_0^
zz34zfrE@8lKz673Y+Vn>DxMhZN~aq(?k8$Tj2xM0b!{RYO?k1Vz8<3u=)bSZ;=Fk$
zj`YUZV?x8uuM5u|IwV?E)JXY}=1XOIzVvyV#S~{>Hu;<FRyH=*w#Fp&@e%V|xL<N#
zah-@!$ggYn>I$Nr9ULM(*TH8Ry-km>GTXK-{adw_w)Q;={yt_plYYdxe)>e*@j5fX
zNmFld*-d$yjI9ZQj?_(Fi6hOv?WP_%+{;c*Ike!_EBojnwLxB((SDoRPTp|H{uI^v
zqm$F3@+NL@auQrKwSt(3dh*us{DK0pS*OG@-=B`l91=Wa;iPX_EoJ?qLV|)??9TUr
z$ihHSEcVFAp~R-~dY>o)h9+*MmqW`vI3$GjIjy`&N5=K`_Wp$z&%xU_<nM3qcJS+f
z;K!~J$N-F-`cgUPX|6Xi8sy?oWt0tSQuad2U)n2R&HjD+V$XkqcT(0ihW{cJK78az
z%^2gA5}M=2?MjXU?dWfqv@SMbYC8@o?^l_HF5i7PtpE?F{L~&$n_8OIYl^0=4g0~4
zn?Cl<T8Zjm5)uz9b#k8&=EQ8{^TI>6#8!Z_PfPh+R8Z|G7<B!b3`8<yod^Sf&!qIM
zjbkF7Le<N2LKBIhbXSxGCUT0?X3P+8AH8C062;<8T##B~N#7L5=q)uc_-FLhyp<a^
zd>4pjNh;gaug8Km3;VO!x82>)B&)9bEAROFeZ^2kajN44a-m}6??9g4=6t9nO%@Hv
zhLk%zk8;!gB%^#={lDy3yMk22-p0)9%ZCvuT};;<JnW&EhmIyD;{@5v1mlGZd%Ao^
zp)~ArfZC*`K@P1%)ugQ>aHGlEXe0udbPr^HSNex1diBcK0n2~>z^ExfODy=YfBFQr
z^r7lz_#vIzckdAOF4WO6n|0#Jg9rO1U7k@ia)Xj-3dY+KmW0Za30>!7g~{P`B|7a4
zzO2kyWN<7V%8_*OYisLYWwAkj|2?0~jzb?_8DneCmYqB2<m%PlIrp|NZAHthtn>!$
z<qSk%#@|RxkNEjf2gAHRIyGzPh!J{&y6gEvmLM=;KI@bP!Z|q|eIZ?#qen-OPo|SB
zadeXdag6pB6_r;~!XTKEfomXQ4w4L-O)5zGZtUZ<VFNC-0!3L~Ue58wI?CAex~uA9
zGo=tjkHrNAe<1<MoZ`B6?Zmc}PX?12X>%l5$KL)n34y3-8DrTvV{2uE>0Sb6)O_kO
zHw{@7hHGZ+O7uh`CZa5$pEl^et>7$&kgr9UW&E8>N7!QZl9|60Uf4H{IQW1o9m7mj
zM@L~y%zfKHV{7Z!f`z-sF_b~}m%4qIGRJ;RmjtCiyt}H|M>>4t6@`3lTWfTDJenT@
zo!YBRFClMOzkWSu_}jO%{Mz#la;o%NW0)2^>-#n;D?Tid$~ZNdn;)-;4-Cz@|144e
z-a2*JJ-1~dW*ewA$dY=seFbkZEuUOwoVy{~`9<pH2@#9VQQMQZaI7a^y;@RU4l9OH
zdjGzCIo%iH=z!SF(+nXA$Pkk__T0MFL6KNV2*ZC1Ll0ay$U;pLO+eq!QgG)BZMxk|
zwI_GpPAeK>Ap$%B0W6ObtXGd7*bfcuej*^?GlQ9LdbViIR?zkI{6@+SWJf~X(z1E$
zR!CPveSLLW*bo{-D^s5G{I%AdL%4`{LHozBq@)&@M%tw(D=Lybpm=r1WT}3P9xm^g
zCINb+&Ykb;>x*$z>8reV^Co;#A4prN#kiF5^9P|R60$V6e6Ck;8{hPj3X+Kd{(*tg
z-XXA<6V%n)NtA|6j}8n9@<!wqzxEpE$&!;($E&M{WjtPEZ5<XD*MNPh%na~j5AHfS
zL`_Fl-16k2{LGi|!+hwpN7lae#DA%eQa8KLpQ)}c9}sz1Eqr-UNPzH|$_4-&F7!X&
zHD(Gp@H{K2@n3u%uiqkX^paUHAFHD(DwgAZoL&QXH7M~=dTLo&SxqqGSscF9WAWn^
zXR=de8}?e9NTf144~|laxwH%pn^))ryJ@-+jjh%-D==f8NF~lERX<0{@3@N&WYnlp
z<XDpe>n6y`>e}0%i9OFk%Z!Sy>*8FRQO>HPCcTnJvN`p$C+^xuziui}SdAW;l5SHz
z$Nb5#u#zTSb93>Oe$$=OjGRZk$F?QAy<TrbYo3N=B_!snS7(wp0}mFAj#HJi&3ou=
z$JVZnXwSPjd+p_gvPCanhTBB0EZ)QwpBU`O8Bum$Bbr#KnYs8EXAwujwu7rKB2#uV
zxV7_FSc<Qf_A;{z<QyX<>M<Iivfk@lw%*ZEn1P|+b!>U{lx+v1m7|JL6Td&B`hsl~
zv~5^y%#DRnOpHD<ryR65HbZ0B$P0Dv-*0}@TYkimovcpo8%E71l{Ph3&;J;zhCZrF
z$L!MhAisr64Gqs*>t2$5|1xDt%C9p0#SlGui_Xzlte_HAhtCC-5YX9_rl{7q1@2?W
zSaK5HalE(oF*Pta{9S4~c2sUy(rQ`#L-tNJw&1ouiSj?e4o4SWs=^h0%Eu2$l>Q|F
zf2{`eycmJi=b_6R-`qb~am{AUni{IU_x7<s9-9;#PAB^=6t~D{SVK<xZb87EchhrG
z6eZ`$4%g!7py}&N7DdriCzqMiuW4IaW>oP7k1=$;=fz&3mtN>;P<d2R;-9?tqB>)@
z+y*AZ#2`w%qe#G%sJi&6HrpgJ{ngDF6D_UXXU>dO(@nT=!5zhcZ(KrjboIq`<9n~9
z$>xoQ>!N;xx(~*Mc9TJD->BlqeGX_-OV^yym53PE8-K#c)x*ghTYD5x4#&kG8g~FW
zMzgqroZRM3o6a7tyyYEkv+(;=^&<mT;Sp_`DKA&o#fYL~$P~RJ3*VW%?u*3GVvT&a
z+W6H~SNghj1&qCAq*RKXt$Ly}aVK^7NQm)P!=Io<y*0i4rLmXPo|}yS#6g>R$qrjS
z&fYBd<fd;{9jcD_!751H3Nq)F+^f_2yor2BFh8TbVY~&cK9RddE1a34vpUd<(`_Ia
z1|SQReOPGd=DzDUY)GmqXV^>Ce09EOm$Wh)<_<_Clj??t;+Bsf-(>cY4E^S!wb)gN
z#aXUdbK$Xwn!AUr*SkGzkW(6IcCv-L5r$dYd?QS59vFM+(u0>06T!8CN)MHe5)->W
z`4!`!)YUH<N<>KOIluXUoLlaPmg%C6zQv(&ms_eX#Sd}&wi-EEgK^N_VmX#-R9}k8
zLr*WpJ~^(RRmbR@?2%sa;<c@ZIr9z<7tb3|up<Z~Ii<jlt2pqsB(XR@2cPiYAGV~+
z`KYLX)2EBJ1qP{MX>$W!Cgn8Iu_T+Wq(a3_HF?9*?=V+lLD+Tbu=LfQo--Sa%k~wc
z;Yp1t5Q_DhqgoEFSl5;JH*gGHux-(Mhb>i4MCT78?x6EN!%Fl73E(gWg=iMXlzd{G
zR@GE>9RmX=ZaV1(@}YCAttPSKY|Lx&VqshHIc7o&=0rPy`j}2|TFs?T)5eiy{`obT
z24YY;0cOgmz1C0b*0N@QpvBh=waFB>Tg+uJ?2D9}yl>w=Ue>gB9>;8LLa@j30YjT4
zzDPa{U!*ND+ED_Tf>DUKx`SZWaky0QMvS#&t;NR2LefZ?P0RR@IXv2ZhToE7yDM-2
zU8x$~+#@6!vA3xvLXfv}@-mf$5kxxm4t941m?3^qsN6qs%SQx_sk|3%XU=kSet1$%
zZt7tpo?q^rJM+rFP_h?m=nstgY&Ss=aFjPFAXF@@TCslp!(aUei$+~@O~uYKOs&6K
z!)fuREi1^)&3j*jIQ*KXdvK=gdz{Gs@xBE}uB51VArGSsw{_BwV>8EIJWnic9`lrd
zS=6)~B7y#hNJA^|l2B_Ki7{i8THl!Jft`g7Eo{S4dF|B+gw(aq405Nce;s*OL*2~V
zQNgch)Yz~SCnD$dFBp7OKq(uF(+4fIs8umtcdri2=h8@FY+1Iv!!3Qz=i%}g9T>hU
zF`ua8X>BbzdV1w;pScaj<`UpW4N_*s0bS9H0>N7yV?Ts!D=jH8dE0Zgp1*6_sh!JG
z<Emq4_;q}$G59!qdER7&$f@dj{w%h&8`lj`J0Y<U(e1fQt<%~->!PC%SSin%73#6j
zn7D7{F6vv{*4AeAL?-R~n?AcAmz6=cN9ChxX;zw{u_~uu?Z$5Y(zv*!uDT2bpEh<#
z(8xJEW~TUp;mmcf7qgu>etaa4H#4!{_}4p6`g82=Tf@%QQjEl46d}~2)$U{EPzrhV
z+TD9`z#50k*A{>0*#>P`nmo>LfNDQ9Dbe}T37KDy-D!L?y*O$c?}cvYe%u?|4!i`n
zd#K;ANx1(w`RIUtItv!a+J}qUZyL-!iRuuH-gKvh8s;qDb<_zox>ZNj?VCgmGWpW6
z<cbB)duG;E%p`;T)%?0!gW#J|GW4BSF9kgduorgb3)x%BHbH`LF)s`By92GLqnZDi
zGmi>0^atIiT0=*kTY*kdU`?Dwiy%k1{|2*yFI{>EVj60L#Z7+3v(eL!Y#>2oA6oTj
zZEJ)3%a%NK;nF2R66D<dHx!*o*rk3DELFR*03BH;)LMw9);l}1*l+hOJ6oV;aEQ0t
z=H<0Hh4G#j2}_{aRH>w1B<%tDHOVDefR(;qgYs(x5w+b2KmoxHAk7;GXp{1Rfd_sJ
z!(!dktj68k*m&^t?YCBL@Cw~7_qjaop#P;v^}}8Udf$t*5$}dt8CfzRsP4Ta7=g(*
z*|lrewn|C8iP#Sr^TgwqSMg}7LMz)_z)iqHCO){CF4^I3h4ildpqtxPEa&|{&LViv
zJ3mIw)5j$6!)!p5eL=?^K^DqvMJA%@&RDw&zye!Gs~~zU0@W%P!KkL#-*4d4u2Y<*
zj=7ujfz0dI1Ji=HlUx&S<NOdvvaGGT{Y;3D2b{~1w=y=qn`v-qBANI<&aljYGlC@?
zmxbMZDPjDCs&uKh<K-Wf;q}TG-AtW&Jk?G!k&gzv=Q{T$-9-f|E2TOjBG9<+Qm-4@
zCkpb^e#gNXYvt8XXw-{s^E@%A{T!C>f{B{-NT;!y9-oli2|4K38HrnBz3*Nqy%&D-
z?{fTIs{Y!e;^y`lhc08E=?}^WcRvvl!W4;R^)66e^5ORlXSIKLKD|7O<nqk#kk${U
zGCePCa$OW<df0H{%PS7u+zg;isF!2S^wzJJRr~7&P$QMi=Dzq;tS4QmXdH5{jyPE3
ze16eN<!Sbp!5%)=+~jB`e5PuIg@?N}ML?`9T3@4JHcdr^aqmN$w2_z+MUMITofoed
zdahh1OBU43VZek7xivzuiPZC$ol!_gx{alsol@&@!->$`6_oHic*#M7?r6BpR#0#f
zV8kQL<zCS<4WZv{+Mdi*Pmb&D9Dn3%c*nuwva&T3txP`TXIB`!(ch!07y87=wwxAH
zlI;2>{1S~Zcf4GPdTvLgLA$Ey#dq^`Kh8FN{;HueL+w}bnww5Q<E8rV7pQ6s1h<GE
zaPsfJyHvWZxU3W<;l5=Ie5$~QDW_mCe_-!?ua?taVdM(JwWBE03x@TOmDVdsQCh&p
z#Aq%c*?3|=OHsDNt>u}Dtt~CDBQ}@a_VICE-bTS-He;zCVhcg&?=9qsl#85?XXl1y
zvPe1>ZE=Z}nKNf6>mTBLu;ltFO5?}ht*ngL?sNr4iE`6F^-I(3Yo7r*je506i~gK+
z_4;+Ou@mlm*ue~d1A*8G4xa9-FolM<_9l{}$m9?8VI=RE_#CRYNnD;*^6Hfj7(C!B
zYryRC;78}OhU4%gOp4^Az(Sx>RFsunu`h>dRl2{%hvbu~q{V$UKG}3YXrps1-(g~6
zf>OS6BbH<_(a~o+1P^OMc=YY}x?LIuuH9iidtbD4)!msUpXMAU4%zb<h4RB?7NiLP
zQ(Ro!eM!#%Ym^$jy`_1U-MV(Q92xq3tM9|r%2m(0yKRv^G9swjw&fFtz!{&8SxtYZ
zKApi-Pf%g4_ft%>pFUNaF=H!2&_j(%ftu#N_>~GaPt@h+9;M@TDiy`jN;-I2Cx_-x
z7|n#5Uwd&?9)g3Di9V}Wza_N2_Zg|OUP$T>hNKSJyle<k&e$s*>3YkSJ$d|i>Xa$R
z+9@GJzJBqWxpsL_=$yEtMx2EqxvQ^|!7uwqCT4IV#C>~xTp}5gFmtz3ZMWKwUPCLr
z1~ttZ)SV`lZS;$=7|>f1i_yZatvi17^x;Keu{)l2;_dN3jlFl%(y(IA{eF)H5AQ8Y
zHbxd{bDrAHK`)aRqUGRNKO5{|HD_PJ!Ey9lz)$%bb#+#%D%xYT^{?IIsBBX+bl=aW
zjk@Kuw^+mM53g?j#vmuN*Bhm~Qi1&zC*uwlaq^^W&C#xjl6A97e|<dC6~Y^&H-HVB
zP&|p3=J|~Eo~6z3+ctoN3y*Amd2AIro9oXmmG-l%=`ritC51T)we|P!^QyG$bt$+n
z58*k~1h8UE<b4V`oUDYA$er+w;1Y^aW;0cLi{08Q(QkV!UA!1nq>U~?=_OSKg@waJ
zqb$DK=I7={{-&SvT*`x-oKJSO2UYa8y{>;_JHTM|w}dTo@i}(Nd<hn{p!WCGO_SCP
zJvTJRYplQWl$Vq#9$~Ya=9$)t=vC!m#%no$(B{mg%W8;yFDaE>ZhfDK3_RM9{ey6p
z%gKLHq_M%erYBa@Kr+OX!`-J)^#ITn7G@@X6ep>m$Sc%OkBy8>!nogAf+(v58#*Sn
zoz|A=>#R3@7Lz-s5uh!5?dPP62r;%u{9@adEuLm<Zl3cq;qo4bEo+u&yH`MduUxsZ
zqv>}CtA}Y&oyd&K#NOv942bcd(I8+Zb8<^G21?l`6ZR$}abYdKshoA_+Xm}584G=j
zt$u0N_A*)jeB%l8m??gjFBa_q#6W;1kWDVRf0*`@arp=0l8UsE%y_0KlzW|BWT87h
zVOVZPR+{Mhx;mskUi8uOp@w}IfA-=9rgg`Dyp9ho(uQeyEYKi}4ae(UWykX=C>H0!
zo6ygY(hNsIXB|qPDHm#)N5~&OCLSjo@UUasCNw8D&SqJZs-Dp`89vP4|G*xrg_WUW
zO`U)&(TQ=^ue3E_T3*D=%`3@pIQ1!j%_hD&ynvt5Nekoce_F)&t&~xyM?&6s7pe2h
z?<x4HRg9a{-&~5Ey?lrnAO!vXr|wsJ%^f4<(xu5aE^v~RN}ro~<4?(d7kT<@v@%j^
z{t^D2<@64zp1Vy%1Xc6;)vnr<>Uq#&bDkTskdGb}I^rm63)2O@6Zhcj;Ju$OFNLcX
zEgZG-!u<<)&$7D2D;f*#rxq+ddgI2{dGqeP8<qJ;W#YtrN(}UzP_p*pB>TTNKmWa;
zV7Vb8>$1vUlnPr0-F@-GdiFr~gl~H_2BLJ~U2rw(>QS6U2N@nP9F>?Ng{7&)wX+#t
z6&JUlNEq_y_M6xLINo%MPx)Z~%<JUzbpsUC!fxN5>v1ySxR3U`$jr>;(vI<3=)DDg
z3qXuS#WH0(Ni6x7Fz%MDb@Qb@!ZgvGoQplJb$<>`R|^CUGr80RFxQ_XhlVJihwZF6
za*|Y!1@iLpsT*3ncpB%>gj7gu0|24~yBEIJ=>F{+ou|js9a2%nX1e-cL&lga>TNhU
zNH?^5lxPFQ7?BdBR<MzWfX3k)@LGM9Hkl)?Sqx>^NyDYg^bLnFzZuC)gPTVKSfn(*
z6@$ZFgp;Au!NMYBjDb6ox+A+EHq6U(D!7fgSnd1waxGSay5C(~5u*i!3DX<ju+^;p
zi4H~#U{rn2Z^~qw8-7JE6^?p*JN@v$+$ymYlYI`NN8ZJ{o(j>=@r+e1xXTv1DzuyD
zhtNcM+voQj>_BMy07EG#@RPq|n59i7&`64SB-w~&f~Ue)3j8L}14Ha423T2IKK_0-
z+Dmb=)U`IVtdf$eVB;0z21MkAW<+RAm@g|nF(hAGTU+MF`Lh#NxssBN6VY2R-}LM*
zrO9I=tB-DddwJVeJtlJB=DD|d2z-f2)pA11D)S452BlkSf4$Wh2(rO5S;6QWAOd`#
z5xNN?-*g!$DpJFAQh0(@>C=69ltLjs@p=Z&T?}f=`3ge~5?@zeUlIvuRIX?Bca(LL
zWP{?Wj_n_wO>2K=EgH1c&Ed^2J;K3y7z%&YhWGCi=pz*Bir{nS&Z#<;>sE}8=oRGj
z@<t=cI>!v?=i<eM`T0yTVpv`e_o!)8rZmw35O0f(iRU|oXgy*>@8ZXgzkPY>LOXx1
zZKJuqdISoP9gMZ97`TTzg6kZXn$`aF&*>n!xA?g*&62LCIr9#t%}`HywD{Q8$i&vY
zrElI$g`6SW@zs4k<r%Z(f1uq%rc$bVicie5qtvloyyz8^4qaW9+rE-y3Uh-a|57xS
zd@c`Ed^u0KH3Qw}gMz6KGh54^J{_)q#`ZHn7#*Mrx&3~n%f%2?Y2a|k(SU^0E3`=*
z;_^TfeNpu6CT+*YU!TqA2%`y^xShB_|Kh5k=_)FJy}G68w93%1pNGj-0i9nHqyu$;
zWx<t;1++;3Z(q@`Z)i}sZEhFDonP2%-3G0Hc6g;cZGOACw%S9qN#gs`fuAq6QMIA}
ziY`C!iO2ld=d)6%QNubrj)42Wj|L><N9+gcTzqos-`20EDkwA%;Q8RVA)Xski6bkx
zGw@g0mR3l}WmArlbOSu1h}+vWie4WR`S?$^6ZHJ&G+u4rUQnb>%A2O$=G5244{LDF
zTC4AQ^t*u}{iJ^0mJj7GGcn}b{;gx+e#{bHzf4xP$f>9}wf*-uYPkyc#pfsQj~f6Z
zbT5w7u3~CCV95PtU*j#4cB?%vri7q>{?a_<V7SL^9|kwtmv|B3qn(bI`l2^Ls$-hj
zPjkHOvSk4pyF$jSyfexp#~@EjPjw#h6WVt;T$#>rKID)(ba{})C(#Vg7OA#HNTX@-
z85k*{&e6Pivx&SP?(An$7yG|OSrC2O%1s9o)_L}yb;s_oo}TJvOQT*~Y@S7eb#~Yw
z>zYZykYrhv-{=8$9)8BP<EPibplH!1*BBJ972QW?Qg6dF+21QR8!xS=WO-$m#<%O~
zd%VgViGCI9OL17$x%DEp1k25j3t<3Qe<2~CJKmamtW5S#jhnLL%ky=r&v(6egGO)N
zx<$Wi!8tX&=RbrbdjI6|i4(=(afOP@{_{L9Mh@~3h1SrTdqXuI+6z<hHXvtenWZl0
zmqo{&TCQ|^Yolaa%CTdQ5E|-aSq{+{$W0NOf^n<Dwk**GTU;y8|GvB|Vn6w{+dK8>
z*w1H4G{DhngVDAg9I?vt(T_~c?z_ifvq)2}wD~#t%!vJP7;>7u{;+@avX#HJbQ#R^
z7fL>-sNmpo8s(iXXfFSK;A8)$QAU)!kz>zrYjE*s&UwsfCv=ZmYJEmh@logb?el-)
z{W!m(f4D(X$B=uM7_Y!Z;bY$WR*?%QPfDEFh9Wx6P0_0E^#;}HI;G|1TahZ33KM8z
zCBGHu9NV_l&F!DsjpxnG&2j!Th#^OAVrfhD;D}iKV6^4a6`WqJ&rYHCaZJ#PaR$|h
zS0uZzFslg;4u-fBMNPYE(Z{R0^fw|0cqcvw$>D+p-90j(nY-=}j)n|wJ$K?dKXbD<
zxjKy)z+S<j-D~bo+Qupr-r>S2Ur_|O0783Wb$n1zP+#%M)kkj#Hpq;icRcIceZJ5n
zIlHZqSjCRZtE4blUiArv==BVf*TsS#Jc9jI8{sb|dN^jo5ulg18j0k}>C;Y9mU8xo
z<;XPq&r8b?oW8$p5pOwu+67lHukU2l?|m3Wm!mQI^qkKOx^jE9{vu{P2gaV5HFbG=
zu0`mRK`pnhN!`~NDAYTyX%8ho&L;VPoYoZjPoSBCs%p>}>->z4E{!ljlzYcb-hoPG
zhulR`An{9$W#a03&SKUa&*ro33&)s-7*1?>k*wepC8{wH7088^2ZID7{Es=pqn+(}
zJi{CR+KPZER5;c?Q@w1yk?<UTwH^3{Od&Zr83IqqW3(<DqUN^ynhkmdiG>Py8nHyB
zdT_EcFg>p^muCIIuSH3DHkJ=E)QBk+l&C>{P2bkkv@9tYW<5M5HT7Fu;w{F8PjSj1
zP=N|?x{KLu5B~8hD0j8{__qEu<=!4FXDGcp^k2nwz`BvDj?$)Up%P6;u)tbgAmy&h
zS@wLkZ>&`BtcMS2^VErs9lP_}-%4q;UOIQ#c6NVousK|rktQAO=NkrZJIV+UMYFTC
zQodPgWOlNf>=4rYW3lx;T*_jHwLK<$>c~6vcQnw~UpZ{WfQwr<Z>AAcEH;)9w<Q;N
z^4gd+moLjzotMm63~jssUiF(@riI|eP~i_%X_hmHe_2?$nGzw>7hlnUh=|2Oa#N-Z
z?APzDZI>ACV1;^`o9D*A{_;FijlB47f}m$FPzIBD8(dskisy_23G=npASI;G07v&v
zS!Ml*_nLBybK&9d8M^a)2cr~%(`jY95nS6NQMW-YL{9EPVo;r6RHF!7fYCpR3-<6R
zy%>zz`h^?@UcuhpzDj%vJ-*ReReO<OzZXq89vU4&Mw5d5M0x>c)1FU0b>Qt%5X))r
zZ04)><_XA8`W=FpZ&+UcYmV#JA2>C!^luMMwcX>GByr)RHrkA7QQP4(QIN&ECQ>Ht
ztIE#G(yHqvz)?zD>r3S#U({t(2eQE2@`Ywlr9q6*teR~Wu}|0k{4xtFe5+^y?F65H
z($;U}48IghIOUkGFfT(~ZpXHQ^s2AK+#>Z@Nn!~AP<BIPuXEOxmWZ@!pS--~ouMmJ
zux6)FYHbTt?dna1qSdx~&6?s%x>UO$#O8kkQWUrSg&gN+OxbxH#FK7SLkUK^CK7dn
zsx2||PZ6kx6zRT|BUZeLn0r1hE(1c49*S5)Y+T@Jy&8UkQ(6sEwpe_2@tHXhU;BJg
z8stq?v35jljYuK-Ym8pv*O@L~zJo-WR_dw4GitrdTPKTNB)RW<dzjhT_Gh$~a1wKJ
z73^oL(D38=#2b&bV_!veM=QqvQ#mqr<a20cqS5I&-eLMBS%Kl<7>v(1?@)h1YLMm*
zPWh>5?c?I&J`7!pkD$XiI4QQVG$z2`-Mnv;R*%5y>guq+`usH5<dVgtt?M^N_7G_v
z+dJx7+&SScZoj#}e>K>f=BFg83G@sV%7mqh4TJ0V2yZw`)TVBh@btHYQWzYZyJ$0_
z89{)}`Js~s6<zEW+`nJHIs?hDSi4QzMl8j%!THMRA>!gWlJBRCSvaXTCN;w{&8FEe
z1mdVtE;tog>M7MSNrFQ`5W^6HRvi2N=9j0b0iA0e6*Z)Z*<bznMLkzAh>9%3%q}&o
zoXES7)#?bd-=m&*Y+|b%IvGIK=91&RI*^*y`jF+!KHtn8bay9HQ*@Dh64o#3Nx%O6
z>6bzdj(@*`|9~~1Hd@1C<@}adTu4ZO0^Dw4VL{7uRkX?leoHuchsDnJI~^RHvt3OW
z!87J2P~3<QQX(ic)>KIxKM)?|JMIARlHT(WjaB!Gi<8<PS;{3#aJ|YiW(0o^F9%KM
z!0tN?w-=mndw%YJ)3xT1(Q7O>+T$HvTpn%j=|cfIy;`UD-eOv{mJWY*+voPT=_ao?
ztjGRi%ZSKVUbP2yYR&XR0N@ilkJ*>mFYPZ74QPcA3JEzJs0S7N=qbg0MOXPgf()v7
zQBq3EyRF8ZDCeyOX8VeX$xws^35%3M(CeR6x`EY<b10p@?VsaeQ;|`&teX^)Kad4a
zQ%U(34({}KQ+NLKY|%eo9nsyqX0!&Jn!j{Rw)=wSo`K913%#a!5+sTIU7(gDRxH>6
z@RzbAXn*k;&FvEdgLJgDqqs$28OXqypI&XN6EpE%oeF{sN)N`uGt+9x>O=3?&%W}e
z%OoPqpQfu-IG<<1B_t%!Btk<JsodT<7rM_EV7c%bft^+qOjO4I?YSpy^d<Zg<iAu5
z+cQ<YbSk~PXO*-iB5XO8hp8Go&CQjYJ^PdrGc{7cRMxKWfFaHllrC#~4*Ess#?5+s
zrNGjh93AskUxl$fF!up>^l_UX{pWdiG{>9PJ$U?B7;FaOxw3y90^@!I2d=xUp|U1V
z%vmOO+g)ar%%A(>oKD|znQhcr?EaV!f43SeIIyqyec3pUfk&fG2An)OlkU9qp0wN$
zDOzUC<ZgVwjBLIQ-7BDUfimJQoZYQ%@6>d^_V-uyBUZu*l5H;|mm0T{l62XQ5J%r)
z%BSq%a&P*}21t9+T9>VCpJ<yAoT3E>s^`Ds)7w5SpL)qT%J?by0+l{2oa(;6_yW|r
zSbMOssjzWy4hoP;>So7=;hZ(Cw#?gGoE9^C@2vs#%QA$=`Y1x<LBD_zl9J3dqA=cP
zHBnJ<Bl&ef&=9CJ6h7HbYpVtzrCPmY5moogq*cA#oA%F=I<4ksXKK2j*F(CyA+B@=
z#AZDi<QAielm^-S+VdPg4+zklyi^5C;$*}Y+9$2-VOvI3cv5?Q+~M|{$a#=i(Z}Zy
zA<l^kQ&W_=&Wmv7EW=t3kj^dmf)(T459d!vf9KW6fd_YwjET{^J^GEqme`Y*FMnkD
zowQ9xYg)VhIfqRj_d7x<Z&)e}?tmO9j}Q=0(n7+!aY}2d_Oj4*8zYC>CnGH0<l;`x
zyXlTZ?lDR)9jzmK=gkY1*2@k~d|F<9i}7M9Gg#1i{%<GN9KURyRcCTw?2<}{Epm=m
zx8HktBv4m&=i*=AWDU<JbFNoxpTQwz{2@)DrvnmvTE^$u?whK<)U06E4^mK$e<1Lc
z+24er%gP=<$Sezl4tt;}=&RPijl0L~WXhGm2hV*#X1;caBc&Fmmbhh<>=}N3UAO9e
zJ0+nd0OQaJvxlylM_s*fqv+woK|VqE2TtnP9YRmkN;mSlCZV+YSQ)xmpiUZ5(sfu!
z`w1X9`T1j+nAci>!>iVe=ajsp8;$`6h<j)aXb+}3?<ju|aB-fOuRFUotFGey37|Bo
zGn!Y!9JYM&2^hbrkFU@!F<`)qY10Nj57C(zaM19Kg_l4x3wOJEnOskJ2@S0q<@f3+
zz~4k&a?<p3UbX68Ug*6GM+LGKd8PlYE8(|p?F_o6zE^hC!r$LoCMc9o3CP`h@t^3W
z3BnD}bbB|zW5yl?mE_UHnD~SQtJkv-aa2`LHlT!VsIsHr!d-}rM>nyafpLPgDqX}y
zBI4((A0c63VMH?aSTFBd&|i65ljKX|&G(%H*bq*QSqzI@#l0t0zBKei-_D00`e0*e
z^oeWNrdM_#S2q2=vc&Rs3J4pMz!(?9OYKicP;{zpK0AMp=hCIIqETRm*Q!<m*c1C_
z&73)P=_`iG^c_8Q*uW^kO=`(5R+M&|w&h_ToWp6x>m{|ZD%okc)AKvHpBqm=3ES8i
zK8D}!nN|0vmi%UKcB`pOG%CPv)V|EU!go8E23MBeLVWlI@84`L+`%@J<^eN?N22X|
z_wIp4ND?=igK=o$75<=FanpFF-^X|FCaI_{@^)yY<h?4I;#j7iH0d%yGIF0_bKpKA
ztL~@SMj(*YfraQQ?oXN-Af2|hP%^SoRYe62y|(}AAR#PF!SCj3tcKdv$$QmW@5Pas
zYVBz<{kHo-b~d2wlJg@_u)~+H{IVtcULBDvbGWV;EMesHZidXE%Is^n*&A@a!t42Y
z$*elu9zeR{qYg{kPmCM8Ffl%U_0jfqgN02Vdi?mE@cKX|Oq-*F85E;GT6lMCiv$_#
zob^Y~#l*~_#;-~VI`?UuJ0}28x95`w*J#P%z`UI|{z854s9xQ>_m<IdfZ?>Hqr~aF
z{1nyBl}elRgq2bV%D`!2nvKN<L?O%xNVY(PGR%2jR=t+6bU6(G#`M0gM9yw6`>YU!
zyv)b+SJ$W9{&^KmSK6Ueqe6-uault$snfT`d{H5M+z$^6Ejc^CTfN2gys7HBAqF8i
zM<yRWF5%*zyBae3pTY~;u}XeR{NI90(#SThv>%Li{qk|s{FYqMxpD2<(AX>R*YVSu
zGKEoXpl>yC=MxfWHS%;`w~$-#Q7R4~i+*OQ3{y3~)3|IY%Pp1`UXWZ|@Q|0cpSF7(
zo|NAy_9M5IiU~P~y!;^zp92<D44ho@u1ig8=GUW>L%wNXl~|IUEkgpHCu&xerCxQ@
zc$^<g@+m<1!P1ypQ`J4J>zQ<W_Syf&-kXPGy|?Y+S`Er7DnywYh!AN|8B<Znm?3kK
zL?R+HEhQOJ<|Ii{+?j{WtcqmLJd>%+Q|9q^e%AB6zhm#?J@&EppL=-w>v<OL`}_S2
z*Kl6vbzbzEAgKJCqYKu-Ckg~FM|yj|<emLqTT6(x_#;jU58QVMH2t*S1sDrRSB;SJ
zA+m2UsmBeAHg`@lfZc?|f)r?AYS2|t>a@ko(C{c|vxH})hQi-V&QiQPKCZU>iE4UA
zhUeDZ)|lx-@tr;*l0il#qfA&K|0m)Nq$NK`R(V17Ak)MutR~4x1LF^8h7dGw8o*GO
zuIbGMgXhZcKLtK`;2VQ`lSf4QN#E7aioy2mV|LD3KtIu@=>TwncU&wEfu92mB`Oan
z)?_y=gSCS%L`jNCspCul+Ns6nvVa}*IH?~xaG;kcsG-AWPP7y8H!3;Qn>KluNYr-#
z#I42cB<fkB=>|$l<Cu5RoMZdpTWSS=JLg$I5$djDoV<d9l5k0A$jK6Xhn`p1XtZb~
z_U*%oJf`{R?c=phiY(if=*vKAlkir6U95Yh(8+`c*q@p45E?@0noE6mB|A5_ZujB0
z4`C+BykemFa{>qVvC6#v$m6hw5*s^TxVBV?kmks(o6X5lP+B!K+^;Sp#~k7K4hdLa
zr&p2&v-qJygMgYMP|?fJ7ayP3y>)9Cvr{GnUdUa(MlJ+6l=%P22ajXKG@7FcqM^3?
zpAZmU8q-=7fBh8|*@Vhav1ZUUeCa8)*B~>_9(G5Ek-dWh#wA&v&Huv|Px&<`DV=b@
zf|4xo$+A%yXS^uHw!yEUY5N*tth_(N(k;8t4>E{GvYy~rL2L_#%%H@F+Gf_=GTfdR
zBx6wD;2q}}Z_9k@)Q)*TkH^(x{#D$~L0XsBH3{E&>{n&I+@Hhda|aI{GDUufb2d9M
z@k39?qZWOr7+MGHM|0scXTvU;YOqul76H~XYMbe}Y$u%!klU&^J{rL3QRRbC81FRW
z3}MEic0T@44b&tqiW)8>FPdqKI~f`aZno5R_*%O=cecj?06PrQ?P6PLsh6nsZFE$D
z6UDjtGdgduaMSm)bxD^QplCQC7_HwRm?k3CMmrzc?AzP(fSy-exn^!Yf&y4^<bfxg
zrO2%?UYUWB5rCH}iykVipDr9iqy*(p`~z%+9P4?uhBSC#CQw&GSeUOXkeQxtUi)^-
z@3W*ph=Rjfb*BsUMfDDs^^wLtd-D%$-;n|(aMcZ4U%xrCp_nG5`Bh#>5wpDSO=qE$
zfY^N+TV<YbHD1wIxx~MjSC7J=az<LDJayW*ZK>pE*rgjkY13?X%{#9eR*TQIcWq4*
zwJW$gmd&;1=aLMbNej?59)=to-UF7wz?7?Sgl~7|W#iR6*&T1X4=#J55Jps^pdg{8
zr3JYLv$}Vc`o9z(8DvH&$Kr)!@_9}rB9c5uCvJni>kh$ckqI#a!n$ztk8?b<c^D|*
zF!y=?okGkXpq=Jfgpvp6eRmjxfrxvHO)agiw!{YDqJuN>qOSlT{g%RHc~<D^lIE|`
z-3ye>qGag3`1JRDb})9YqL7bg!Wl|HLlgdv_H^Gf&Ljc1p}6<^n@-VskZ?&S`KnY;
zsNIyDYY7zJ(IG=ZF_g(v6T&WUWPEfs3P%A7v=(>_{4X+g3m+5^rp!8vk!q|Ceg6K3
zk=YnHPf8QBO!>SMyl-$x(D(HPZy+c4L_0OE0uceD!=Gh4QFb8}B=>wM9qkD=;J}~5
zMjsvx80)tiNhyKW1LsFFgWRsu!Gw{C9$hV(Q)}uNL;@fm!8-S`CqjS0tU^%A9GbbP
z&|&4hpjeC!Ba|KGR#6C;%V)m{019-zPHfz~AQ_4N<!%Gt_-!6h(Irm~SIqmeL;eL~
z*X{+?uoJg5N~KWn-Q!pP0MtJ+g9>>lmz6qp1RLF_ml24DI07{U1bSj$5R`-9D-tGo
z8T?mpzOL~wpqpi8#|2jZ*S`SdA;_So@xVz2e<9+v3EJXdgR!E|0fFPkc!>)V((-6s
zKzkKW=L6_nkdle~>I$0ZWjc9}adVphfkbuE0v*_R%z&^_mds@M5rDCAeR_l!<|JO|
z?c7A`*vt2zxo#eTFxPyeGisYD2K4NX15qk<Z2J)cVIEkE)k$b*om}wvn3|SGXjo_f
zK{=Tyx51TsIzjDEUIi1PVoA^L*ZfYQWy)h4Op@ik8xNq<0wU9R92{8aUSPdVr{14L
zF>(@4fVt%;I$e*(Te5Qku0|gccn3t1#37Auw5z%-G5JMgCmLNuPo4w^JONlYUMLaE
ziStId7RxP`JPf2T2k5?B_Z+yuy&okP`U3qkTX85g^&R>zD~YTBTpz3eCQW^4-8DF&
zOH-+87Xbz6*Z*pVPwv2kpEwHjk3!DJ&6h=4^1Znvt6vNqSlD8*l%r#j%{63q2d*$q
zVCX)*t=?_)qO^;?icy|PI&$w?py+tsu?A^%kBN*6*ZhyY_@J8c;z<F=sXLvIW?53w
zYUYeoyqxhD^`p*lf2c57gEXCfwo&NAnTEp8LvD_jW#1$QV0e*lo`-F)5K9E_4Pg=G
z2=y7UQ?IvLWcP=}${QYdFRr=~J*%iH^nMEg)GKmxR#{GMjpkgGa{V9iC2V%O?u;+z
zSiseCK;7hk4q_y?ZV#q+v8+GNST*U{Nv~ef)m{E2xprE;nDzNC$TK1q{}ytA!zEp<
z?j@hjOwygmQ~7aQF^*QMUV&Bs0ebTZ?sfMmoHCv+XJCphJlVt8l8RzHW_|Ig&#>Od
z!DDjB9*gp17eTiULnHl%Ma?T}YM8S1cH$}YTC2r=UkYo|ZS`DUsCE<3az8Ng)NR?Q
zLGafftTz=!N7L&RMGvR;hnc+6f>s7)=rk3XT<kAN%OSB0$V^FH$Vz>U=EK^c0+*Ne
z=QnuTL%L4esCc@K#70;w@+QB+0!V)hihtx!ail%25_t?@0LNTn;UTpZzGd|80M7a>
zxeCE*#XSAB)qMcahFbk`v^_$1tdCWLlYOEa?F^`6uEWlNX882?^93i+AUpvQsnRte
zt5F@sN0k0S1>@%!PKYsPJ6o(#T`kEUK0&1%9FV_2_e`6b&E`8bmJ5o}(!yUm?|aDY
zTsSlaiRPy-x-{8FbQt0KU150+@xLP!;6jYQ^FLg|?qp8!UPpTzlx~nRZ{Rd$6`P(5
zh9!<XYAiQTnjvHo&c3q7$@pGy$MN!R%o1aCSZBb)Am`j<O!eCAam}?zGHyYB|GTnr
z*SXw@91$30`P=_27tlx79xv{Y<y_D$@PNu@!*P26?w#Hqa%UqY0h#c-oQLbxKo7u)
zy*xDzVq>79$f1mxBFZY`qgju-@UlBausl{pJK_@mpn7~zhPpWyMlmMmVSel?Bm3P7
zp%T`+mna{o?o$Zi-)!}S8E!zj;LA=bDr#Dx#X_ux3#75{!q~w>P;Twh!c7B`z18C^
zB5_6A!&Izy^0tlM^^;f}m{8q<vh}6gm~!^F#ChLUOJx0Wwp)U<XY%G5?yrBC8Repg
zcl)q5tpjDPJrw_c4+(m`xRWRvqseBk5Cj?VZ^M80A6Ohq{(oJ7|MbKETmQpr-_%i}
z8VMq+n)xBBGck~D^l2+V9trl0x|I;snG>1N3s4KCqH2)dZ8VKtRX1m&IzthINqH){
zYL86?r`4?N%|GK%sJ37Hx60pNMgJ1c{m=h_=fJ0rh&!ece3r?rqoPFNo8TwJmyNKI
zz!aW+cBHx=bPEv%MV{(DZ!nL{c*BP#RJK0EGhjc5I&$aWg+EK8q+tqb(xG@TIH)Mk
z|2+h3@3Vhv2maypuhCxr^&iMA2`Z}pS`VVLM)kjN9F%1!5KmP#MJE<GuZrKh!4-eH
zIEgu?3f~Jsj)G=r;`1JS*6Tgnqu2kv4q9q=F~h!ej`Sb)tb#MbZ>NGED?hmc=}g+2
z(B36%H0dXehz*gYXx85(KV^Gdv+l?JkXNr<KW$`EtG~&ldq)cGdeTpb&{e}<IL5R#
zpsd8lMhWCieD>Z$CcIF#z;v+=0Fgk%dLhK^CES@UaF*p>{P0Z1eu$`_1y|E`ff}vA
z*AY7`l;ExJn@QtVJ`~6qcr!&u0;VZCGR2|c3BF@9nb=L}-n*j^0h4lEtRsH*dJ(Ti
z6{VbQd0hLixkFF@{r88Q?bm4$gG|m@4NXF3HzW7`;n}F~aOkKbkEeofq7?asRy6)>
zIr-+Hv-tV4P8idTls8yZswE-3z=S1vMpV&E=b$ZgYOeh*Uo>H)#?r!Q{9TARXFBMz
z;-OJ}EG4vfwT$bD-Tr4u=ARzruQrrs!vCMOG_;}mA&F72`iM{amLQ}5yW&`Nd>(oJ
zz5V2)+%O<<&i7c=#ey}Y0bny|IuNb?Gi9(1_t#V@(%SK4u+!hJO5`a+Aqoa0ac|fw
z1hHTbD;qbF?~UWas;-tx!7(C6;;#FRx$Pg2bp1C!eZ!Tl=EJCmUp}=RyL$gWw!B9I
zb(sO^K?Vo?^GTfeq&H~CFu3z+qt_DPjQtR0F&iQa#Xx+K)rjq<U3@wCoICf)`dh(B
z{{D90>DJa(IAIiWG>yga!)LsB^SuS21W4Yj0}lu|9wn6-Ke)(<)!p7D&7{YWZVE2H
zx)zx&^p=>!$4M~b2&VmRj{{QxLQE+jOMcPIcS9SbPg1X!2JrH2q>8m55+H7?%dua<
zJpwLqm<ub`Acf=wEAniJm+JKpr#I%HM52uNgZxD^_kZgORBzzcpdcY<%Sfec%l~iJ
z`#*LD|8sZo|LAWVRmE`{F*3tUa<JY(`DC~Lwhw{h9T}Og_<yJcWBPU*6afLOz6CSj
zJhEZGQ$3Hr&`IVYUPXcnHS6liN(>4qM`T5|6cEe%oHN*TS_Yh-i*msCF*2Yha=DKi
z@lyG>&CctIa5fT}Ibcg@6fZ#vt*+JW`Wi7Arf56D(0~pL`M=+<El{}yLgb9up#~29
z!H7$_k4@dq><B@1uuUJlb<Dg#vwLJ@A^rm81^grVJF;5jVa9GmAbUi~`+>sCQ$h_5
z>&^_>jx<juWOnlM^Iw{HxBC_f*Z|L5K*3klTP^5ry_KYwVkjN=svE@<<e;yd#<EL&
z3vr@rX$w%XLiA{dNk<phY}Je<ovE5LFTFgKl5QEFteFzZ;<x7nZ-ex@<?xJ}b$gY;
z0Rrh;5`FT=s>RY|p48@>VmTo%l|o|1x2?0s-bdWi#?r{=YJ%*b?=+f$RZl)DhdCmH
z&-R@g%b)8+qwWyKi-H<^$C#Cv#vx)qCQm><D;kqzDlJ{nZ24+P`(hs^j4v0i&8=w+
z?ZdtLjoI7zzpF%1@rbWqk6W+24Tfm19};0i+i@_&aRLwDF3)BblDyfNqaUUVTw+Bi
zh4C(tjZ@>Z5Tc=~0h6oCzV``QfVc)OtEo*FGa2xs0TbyK3iJ3RQ>5(Z$RIR2W81MV
zC@P5oKZ5mxu+&KmStXm8YezTqGA!&gMD!w8r{4$bse>@VjX#EPBlO!eW>fHeu?JyA
zw~yj9jN`MuH=<%MKL6vvQ}Nkm5QW|dHvp!E?+Dd*9E=IoZ$;%sPT=kuqO*6TVE~T>
z{W%M;!~5PL5nCCxbQ=bO+0IsdsuzWAU!WMOY-7Qi`dSzV&BFj!066s;-U5Z27xS;)
zkTmJx7J5HquNXD|QH_iqy2=4-e1FZl`I;xt1Kyj%5UwFJagoy(cW5JzL=6tuSm(XA
z+b28ijohtzfPpEX#6{2~JT3GG&;)_qC-Zys0XeC?)^$bNiGvNDn9+Tntf@%yaKscb
zy&Uijqf?OT0(t+O;Y22V`)erKT8TPO7Bhj06AAR<nkU<?+zcE+oV|LKsO4?H<B1k<
zl)cRWRSYs}>RNT!Hf5MK@Q&#NzHr7u0g(aRs7v{R6Ob$jyZ^V_!c3Bz9_g<N2Oxzc
zybHZ$`LwB6TVNVXEhdp*8@1p11t3cw1V!A%#iy>*(OGN+H!&^?Lb#4&>Eu*-8|bA)
z?&!znSy%>SyDY%{P?iPQk~8u7PXaHsM1k~0j6{Ek4#~xHm$*W8YvEmDzv4^~Jw>l!
zQ>+B`b5ENDv^N1#KXhZofE>i@5N@P@`Vrd;Y<>;f>64p*;63I^O)o?wdLeAd?lh8u
zym#vJpMt(^?|~$e#eia?-YR)n<qk$_3{y3vd{}><7dw^!vr*!rQ&C<HeOCEm_OWN@
zXcyep#@D#)1^uSLczI>t<K<Nn5A)ORVqo9xq0Dm}C^c-!2G4KR)fk0!ek7ULiS||0
zAmwzSWn*MG2V%TP&*KASvu_Ey6g!X2**MBL-R0P*Wx^!^Fud2Db`Ardn_{~yMyXgI
zL!a;Y51T*Ah?1&GQaUi5m>iz_ZogaTs8)>tlH)M=<&yL5Evk@Pp(wG?hZWt%CUSg1
z5828QObchR$ci$*kT=_UYi;tW8>fCcrd0Ie^h2#oNo@&kIx;krscjALg5nXxWUX-7
zMs2uj(?BjcRjgTYAZc5JA|muY3ejhc<qV^Jhu*G^B-B%-+?w+{uynB&xQP=S2qibS
zP9q{v`|%gl0EHmf4=!@>y%L;jfWQMxh4a#;cF>;df+2bl67sR>6$Yx?1805nQwyg)
z`?|t&9t_?3R#CBa50Tlm(xmLhthJq7B~0nl3Mwi!XeKQeGk=DPKB>Rl?ijv)4~AfN
zJv_$zQ{3X>+tpQ7v@Zg5HzBmK2f6_b6!o7w(777Eck<~6_?^=016+Qme?CV?kj#m%
zl?7mZK(O$~Ypb3rb^NU>BoqVVIU51t?9L=(UV!Ny{++x95^ehi-lf8(C-@$~O;=7J
z@v)@g_!})l*v?|7+)R_Hv8-M{u<nXF)RdJ8=z`HL_YitR9mY{v;3cG4-%G9>TDT%s
z3j^2XIGlh&voF3M(nL}nhME`G`T~`D5{q`GC(7eLRiFO#vU&E`D<^@m#UrJ$i0r?=
z5GIqrTE1TT3PMNkM~qt9hX`%T525QB+r6+nZXa+yzVq?{)=i9!PRE=%x7A;6Jr>%j
z*JS}vDxK~R!fEJ6t{^_1_nQ{Vzv$`~wZM-=iN0No_##Sf_>KXT3&{gE>E5k`-(s}6
zFo{RLkio#jBso)mIYk^1nl@_EeQmw80lsKw0?&g3plV9^yGvm^?JC2(ciJ32iz)R0
ziTTE}h;N`KTa>xsLp&k^wR)8*0Qm9o@y)X^CD;ng5<itH1mwB>ZuJUIVh^n?(vF~^
z?`W^tji?u;Hx3Jw7)p#o^+M37-?s4F$j>XnWCil&U+;mn^3&m01e54zSc{!)^JoaW
zApVsTLJKSXf^J?;uTfNx7tBxpLy+mQ)8p3)UY0FLSN?WA1Hb5``PYuoxc9I2WUh33
za`=h<%76X(2TbbI-11WHNL`Ap-WVe4=JE9a>$lHUy#yf=@wbneE?ysa6Mft1dwQt*
zve4~0&hycv#X<lo4GS<ZMG?igS9ZA7E;xq9%)$;^H1wdubt8s-4}5YEJq&9hfL^|v
z1Z?IV;vvO}wpUjb0x&xDDm;A3D0(+&wJXFmKrZA<%G+soxo>&;VyR&1R4>O`7PwXZ
zhe=X~L`5wVtJtLPbU4oX04RXJfLG<ipc<VV#pFD)qF5z4(Ho}4F1&KUn{!j2V%EVK
zGkEtOU7%EA5EHZAWf(l(zU`16gqQWYEy&NWn|@Qp{j>W=GV`|doFIx0H|1>GeI*!Z
zP?;(cUi}@p=KG`r(LKrBzJ+AXvAMwWz{7>9Wv`JG1eUfGG#8_jpj=0w3XGOho`(Cz
z^mA?rML^OiuW>G@R7nKAmV%a=tWiSMXqY-AwMW>TIroP>4+Gl9b=o(UeZeCzWQt4W
zvGl{?{MnigG8HHF7LT*2E@KFsjk(ix3qqcWgh5Vmk;{D}4<G$bF`|`lRu!OtvsS9)
zy3=N@frcsbkunc%XEN*(B@UMBzS<7r;Wl(sq`^%AU`)vC9kU&snev>Mx?BYy8u3Bg
zBS!?1Ms}tMdrsZ*6J;bt1H>cpt2`y(=OUP3w`+Joc8mGQ4_G!i*Yqj0yda=7Xjed#
zhOjaI#ZkQLKMBxs3A#<(h4t(@Z2LJRp`K$gq>XSWi!b)!#=}+Y$K}N}+}{?Y3oW9s
zf(dvl&9m|X*>C$etc<2r6{N}aBQ*dx#}W%|M4gCwu^JJ}H~VdH)-s^tJob+FN(fMu
zf1NQm1e;c%om!pLgNq`>>|gVwmGuF^N#U_KjjH4Py-tww$=3Xs6*dnhI|4Wq>q!bP
zbGsFH4p%3r?+227XW+^RkXdpEQQSB+Swv6p-`ZWeH&D+5(}cEl5#<>TEiSz3TW@4_
zt46Ras21wBE8sW4rLYDqA&?^hX4sR}`l4DdtWEDdQp$VqbvOEz+Am>(uINpnxiVq6
z4*Zt%2}Fm>K-GOj8CT$O6_Y8K-VzH2_f-KFX2*23+iDsd1u8E*=RO(F0`;SIaUl&_
z8$t!~7|C4(194lxjd>0)XIBTP_cyw|5Sch8JwS8(@W(a_9RzqIwVfb^Uu)NMT_`vB
zh<NBD@seY?<9@m2*yOw>=>|SoDXC8+ux=wG1xb>eGFlox5P&gWo|4PmfEc`Uw*%#u
zBxQcY4xt6jM=3g)W4OE4{lS`>Y_kvq$>*|@IU`H^kqR$|f=k_r>_AEfq||f_1{593
ziUtoQMF+PEnCJB0>*?a5osNVe4oA~Gl7P^N@lr^QaqY(gI>cfeEzJlVars9d(knjJ
zsld0ThEiV+d*wJ3*SviVw(AyZQfj^MoF)~|{q57?iwhC#7Ou`D^;)ca;aI(c=mA1&
zq1mmf7)L{b!$tJZk_#*4f+EO{IUk}EJuB%oc4HG2^*batzv?e5;RP+*$xg1UjJvu0
zim-#mtJ7ppG5i@nnokkP!1m)9?Y0s8o{VTp90wOqd3)p?)C&Z!Bwq!K0~&;+1q+u|
z5yIS{MHRD|tS&ED=1t0&a%gYl)vE1-OYFQ=ef`}VDB(&td)D&cQ$bVHbca>I4PQ)h
zTe+3gvHuCP&>H|mQOZ0Ju|a|f8#dUQF$~`oY@dXP3>}3pys_~A02gP{5!=JVfq1#l
zZ1i;75V64MR6t!LdTVc&ZIj-wmv<F$!HrJB(!I1UJM{jgwOr2*RNXDJx*w?}4D1vJ
z-68+$03an-6r^TF>Jbj@NFR)i_LDn!j?UvOvEimh(^>N0IX@yyMvjchUa9G#h#Zoi
zref-G5qf>!0c`_@k#KaA0lFgr0Rb*Uu;M!N!9d+JT&Dn+3q3bpOL-`&qd={GS&8uq
zybh%DB-WOHt1;oj&NRS$opHvZvmUwn_X9eKE4y#mK>^@5ax-Kr=S=M?s;dhTX>+(t
z*U$*YZG}+WE9`8MOMq65vB6}O=r}7<Y2?b6Z1-W8)ehlYZUD-ji!TR&KEH+jV9>pd
z!N_oCAF)^v={t83XXM@>Uc=+s1#sL2c8FWOR2R9bMv>!H?+J27eutuaT&f{&H6)>g
z5nUn3MTg+Qh75d#azi}SiJgRy@)PLbR?`?hRHE1SSPt?k4AQE1K*&Z$bLv>;&%h;v
zb9Ip20iF{14rvZWY%GZ^5%uws(T_NP!~x*6;V&%8n$A6iAbk%qM5qFZjy1YC@{-Yo
zsMAY8_?^V!c(ynJ0u2U{g;RyR`D$+0%9vf@k{;4|H1ani-#!Nh0B)C}l2R8OkcGt<
z!jHnMpWNYdgYWghj!l_JE@|;Fh>RqE5rM;lK}V8|vx3+L9Y7>`!vHVWsXxiMv0vn~
z&6hw~bH5jM<l1MH#tqw$<Grld;r2Vc>m}}~)k$&Kz*67sG#;;97YD(?I#OCCh2y;^
zeGo}G8r8jT3xY|_%Uf1W{(*m`sYw1+;lnf%b;BU3Uqx`mX34i)&+BK=gaDRz7Vif*
zI4VjuS_NnkoiwN?7nYWn6VH6T95#bs1!@iJx0km5k7^;rHIsk;1(5x-4F9=U^uNmj
z@kE>FqTwVTm<yv04V7{n8l^v?&!TzuR#H9^zodjkRIP^0fs>L{l%Su7gM?%xe?B)N
zFic7e6;<%i>j$*nZJmHeAQ-LoPi+kv(&0lpsb1$K!(x$t|9wgF|M34%cF}g-h&)&W
z&m$t(XgpA?yZd<y_uHJ8>#IOUDO?)M{T70PBAVfksn1Y3Wx!zL>K{$*htIeT+@CV^
z7iA8y1Pm1TKd?s}q{1xl>;5sA1pJ$LiVS}QxE9Uuy4j&{^gO|u?bjiU7j!95upv7Q
zZRD;286>ughheBa`!~Q#RlTaexC8|)(ZYsaM7m5?yLPKprddh&TjC}V_+cm;0f}JM
zhGyeT5f4BX6q8jt;YtD0Kw}Mhp`&U{GX^9kM(P7xjCAzFyTE|mMw_<7rgm64w0+_8
zUVuD=@RJ-EL_OfKg}FdU8#VU=ka_Nu12h8yapjfctR$Q{krT&n($LU=Th)q(g~vJ7
zAgyu@obNm!Q_4IxD>F?<dxabNS+5YV2w@?d?VWNsRc@KCH8j0K(OC$*O*{>Ox)7T@
z+0!umq8F4A`jGJOTzmnZ5;O!U1k@7K-_Hd)mn5A*Ld$;S>Y3~1Ku6cYyO!U1ms3%_
zQR`wRV=iG4;c`K*!ZZw7+tbesLG4n${)nB%gQRX*?`OasdkzLuyA(3@K#W(ld5qqx
zcuX~<AiDMI_uVV|71Q_F;Rs27{mFeJ@`~<RV1Q^G$7PuI=1YXfk)q*)fQSpw3XDFE
z0PsYpJJvvdBGAw1*kT;YhLAaq*Andq$aDZi((U5Ek_-<On*cCS1mFN9fUQA>MN7(!
zq-aJaCTty<di@G?3?Nb27%db_i^y)gF`*5~|J@pCrUo<;e^SPOE9`#60yZ?8g#+dk
z6~&tx{f90<m&;rS!N_GLvsM~4p~zl_2oQlLPJ*{{i|5%WBfrx(ptnOp8srL>)!9d=
zNrWMNV*tl9Z>CYFcUBih0nU-FjT{^#`rN0AU<ME>yx9Yb5$><zB56Ewk^~*d^OXQS
zLcz*^6mKRD#tSN35-8dsn@|g8YGqQilp{{>3GhuI^(2A8KFU}S-X0Jv0%`zaPBMRA
z1n8|FFs)&m+jKMsP%MEl0mKn32xH(*KOaWft(u^=C`uk>Kw$E`=jxVs<+oc>xjy+Z
zzCo_WRbmFWZ_M$b73fAO{sx4<;{cM?np6Q_vR1jkGrt_trsc<(M;LC=H+~f_1I@sm
zdxUoS-Exe8P(OH>^lVf{1aLcV^ot!_#a+2gm{t0c{lccTeSiQi$$H>8kRw&kcSPgH
zjh}$Uh}I&nz&L(Nx{7;R_xt4id~plh8200>@ZNU-BH?O@l&%}Dk0TW_jOslYn|$5z
zYM(yhW~Ds&ev(VJQy)dlN#QpF+9SEtQXR0x4ksk_PWNX@{A~n0fvZgbW<E@PXEeq&
z;hI6;?x_ZX!7zCmtdU@ba_e;RGD*W1J#ruh^gu*%>e~S;H|ksUu_*{wHe4`8C^e#w
z*fDtG{`^yLUn6a5Wt~>&TM;2CNi1-3F>0*Uh3~}sk)Q_Vsi28F4WvJCtMp8(@4_}R
zr&IMRoa_j?T7r;vFJ(<#Q%}91L|evRvtBw3U@4kl*LIyS;AWs9fcHOz4CZ|8-050*
z`Bx9|NijJgJ%&P#L+#e(hgAw6Y>}|$d|ZQj|LVtxfv^4yf)|H+l?xkJ+=~Bsg93>-
zP6XBLkT9EKguO}gt_z$9m!2XhJIqB18$o_~M{h<c4<OSLggC5npP=icPpqCkvfDBT
ziP<+IMFYUpeImZ2;+lr_1F<w8M^)w44zzpbBZ5`e=tOaaUSb+csApV~aHd;cfIt}J
z;PA^RMIeun!FZ6En8ScPV(%*Z!jNEJ0)FrqPE$?-b4H?q2bPvqQgTBsM9KPx?q-O~
zOd$9vTQ~z%m$Y1zG2o6jZVNsogV@cl6;sIWqHxsjRAjtQ=DoL9&0j<DjtMWQi^_mH
zPS;#GpaWOzR>~C`bS)MaUJ?d`QH`lTI3hwE#s|ijtFRpCu;Kcavddt-A>N*ju<Mj{
z4TU8lIZ6uCye1!r?s3iUCy~G9Bj~5MA$aW@ZP&G>E}R<8tJ{{mVH(DfRM8rA>cesY
zMHNISkBf(NCmcg;06GvEH{@@3cI5$BK%zgd<hFQui1Ns#8cYL8K$hP^fY=e*lri%V
zu6Q_R;BIV?c7UMl@0WM0yz~nAFM^B(q;3|rNX<;$p8~?07ZO;gUiXc=tvU6x^_(nR
zP;hTK?8G?p5<IdlB*FaxG*mpILLzE*IGJQ>`T-p<^|b&?r9PR3F_lSJxJYToh#S~A
zb#o`MaY(`jgeOaL9ZpAYO4Fcb0N~Owy!?~f)!)7ipwt4-)>E}168R0*N^7uq<4xQM
z5agCX*nx51VE4!TQr8-LMu7P<pqTLCb2xKwr&3MEFr6S09{+UM?WxF6Z0v~wD9C77
zm;78`248_E%f_z-aKNPGqBspC@b^IUQ@;_!1!*4=bIgzllPfK!4Su6#&fvRdzV^TN
z5+kBkj-rdo&zknO<QF^`3D*}O|3>jod9XZtis+TZW}A(SECo<1vZ*}L6t9z+C}-AA
z4V&72=gYQ#01ASn5W*c!H9&))<)Ss7H4p;X;p?p#nD?OQP`_mjPLabd%=RB<U6eK-
z505=)!*BMuE!IVbIQ$Lcg`$sMKc6>EGM=44%pX_$M8}$8^fBV(21ePz3*#+#KUS0b
zZcY7D4(#GPAY>MOBrT!<K*@1I2!LRKR6&}TzKKu#y&iJGi6b;$XhjBI->o|R3mc(_
ze3hg#V9!FZ-W;ofyHD7B?h;f=fYc|GQK@qK8D@f|N1GA_y&4e01Ss~){{p+0_COxg
zn`#X3f*Ybem<DP1e!U-dQ3MQw8pW4mJlMr!xQ`pkzoQoS$qmR&V)oA$0UYv)PbLoU
z<ut<AsThbN0vnG{tK$8AA7vsGCQp{>G4;<h6|SDR7h(r9Yk(puDXHN^A(utW^)&zV
zT_jgMAKvlG%}C}Sngkq7Ft-#Up5O*F7m|Ju#F{3`vR9J#g8U1Y4+iNiU=x8HZlt28
zcTelIps3GE2BBITtWJea`EH>KDp}o4bdZTy4!Quj-b?iFrT}dQrWp<)WfYmvRH~xF
zr~v`;ja@kDI7>(ZbFfDQPI8EylI+G$hzJOdFD~Z_H0DBnvl4N2d^PzD1g_;`YFHrL
zW7KJm!5&QcURES01SgZe1@n8#5z_{nRG<!KP0Y~;z5-M!;&rp9aMhn<OYs<d&IJ|N
zUs4u8&<F4)`1V6wUjk&~1H<r3_!!jq`96N`%ZKBE|4?~Ct{yJ;(gRwN*uD}ZrV;=M
zxnJ2Es%0X+O;r0cA*hl-rAuO@QjnB4C5I42&NW^o`YR5KOP#_=X<M9BW(*iK*1XcA
zS-9AMw;?Z!fI(3AiWdMeU6+{#slb9lXy3jTsumfJM==#r1%s*O0zRxH_AQ(*7R0=k
z?V?SB3^e)pwJk$p!|`cvW*2cgNjSVAcf<tlJ(!F(Th4{X1FGd_*!kI<CX^jXaW=pW
zJpQr2sM%wScJe8D<ZV+|!l)r#C<Nnr+^BlF0RIoA!yt!Pbkjh}=1t$qe&;7=<47cc
z14rUPNScAoQ?4uPg<3j4^ct}C{kzRzcQ7diRlu+K9ydIUT0+M(pm~{$t-FtMF8@>&
z&g{6S1G2n>%;r>7@dCO+aZ(OL4+%jD<0*Tv<Po!z!v^dGaeeAGY(ogCE@4yf_KP^D
zMops$M+wquP70n?=+u~a>dAJm8DI*dlhFi0;9%^7b-KXA1q0u^x8oiU1C1Wm6%WoP
zY<{KBc>$dnHfT{M5DW+xMo+~*Hi;&N&PT2-MB<)2aue%&cCJig2z6h!hN;ZLl?3y1
z>!=v_e##kaSlbm1QIzL|FD)C0{ZOx+Ld4!>Q>WlRyp^c4-TAylFn?OL=Swl|38YDY
zmXA-#<p7fey9sxg+RJkhEsye#T89%#Q6tc({SCy`$Gj`Y_Ll4tk}ZAf0mONiyiE{H
zdeQCpfa`-&-5^Lz^&5jxn}LmL2aHG9C$C+MEQPdxa_1Gn0!_e!0C+q(<Xwz4M1^{A
zZN5ZuXP3oI)c91CX0SfCt)}nc^*uZ(28FBeUK*6PqI^JaCK0w8^;*_AB(Qd~037)i
ztayuHfudcP5_5(qMY)`5n|n#s)Q<o6VHqa=D--1gtE4)f(qahrCF9FHiIy%*J37dI
z9jK>@;5dwIkU3B`)T2w@jHT5h4(?i=%|r)))8VA7EPG7hKG|VYw}t!VGazE>*HCT)
z=>3SO3x-+oDo0AV7)Fl5lM8+{X>U;lp@}j=Lm;~5-cH0>DdN5QxdczdomIc*FH2Gn
zLq5V-XKy33Md3%f^{K@W&Kpq!MVHId9C2I3X(^k3R>td*$qbu#YyOTK3Os>bR`~sH
zHk2|75koagKqAH0k)AFU{$YY63|8ra1+RmWmu_B<2b2C|mtT6B7JBsp{cbFza$~=Y
z(Ep(8W_VxzK?0e}d_+1h$2E#bPVn?JDZ1&b!IIHG<p7*jg{oBt;?mf{-{2!GqEs!0
zC}e}%6YQc0q7Fc~H;j!vO5#xL1JeaJG3=5M+)yr37$YBoT!UCf$R?c4C!M_+T`36s
zRuFsJK()*|7B;ILtP9zKV#LpA0KJRrD!DrOG`cYWie$y`$tmC&bC$$y2dgM0bESRQ
zOUqry_fc_vzC(+J5FGL}iUIo8rChj8SLzE;CxaIrfqP_>*tR(!Je*~TW(|6ravw2{
zU<?=0627GG8)B=4VS}}trn><<2?UEpa3ysW_2*$>4z;q}9as?ciNAPiVQP(-_~ez8
zo;uCVH`9MZCXJ=iyyxGrHVqN(Q)B%%iNy{O^v(BvA^RP%OIB0l=*+HY)N+6fJYD(2
zJgIOGWBFm3I76^-)J&G{`LYKOeT{I~A&g1#D4274I>O`_g&x|xB)Rr$4JK^EHg&k?
z0E_OYK2omO?GhJXSP>c#n;@8cE{XF&kUcLT&7mbYB{=|udw(`D&Oc1j&KM$OfeFsg
zfws(XTnhJ1Udz`D;E=Ko-SGmqzj$SY$r!ef`{kXpJpss$_na&`Gzoj*A^8opz|on<
zUj67@PP{OQ6+76;7sf0Oy#d%Y715A=GHY09f+sY6@i-erKnuhPpOjNLb`p9$7*?=(
z(lIRlq<USnW&Xv9TgX7!cbgUm`}<opsB4ZkryQ8PtMeBaC>+HhJLcr$fe;`6Xo7ab
z1%s<z#raQ0|0&|6p#g7fob*Ac#xWj5E2e8of5+WwabtGU5@Tg>(aYd$icjA(^5KdT
zK?)SZ>2GrxxEb8@e0!Qu+~??Y!SAK8Rf+ws2uSF*ba+507Tr-KY05q{W-XpVfc9p+
zvISv6`?*t{@9IFvasjVMDt>*50|w~&kR_Z#&!iD)6Hy6uEFv_nKLzu)+s(x?laOrl
zV)#qcm*Bh(#`YnqBMgqbrV@JuhZ`c^OBXPZFadcy=P5x%PYv);vy00lske&mO9p^J
ztcujJ^&~T0TCs9C-=URW1d(+5@!(6oo$wybh1SN8Ye4p3zyR7tEW#arc2X}&$qlLy
zwnw=Sg*E2mBWPczU(1I4i(}tOH!1Wo9|2Haq-4HP8)Bd4+sKt@ud>pd7=|4-X3|<w
zvP9U6;h(#l28z1==CXu6Bjmuo124e1YOi~A-EbNkNxC%5O4=-P6VA#h1TaVNqBDb%
zwql|ZN^B3}35!zO*RU1in5RJ&e8GPn+210VJ>eC~h%~xL#22Fbz8x3=k~Tb&(9zx{
z*xo55N9m`enTA4IaIYW;yWN3452Sv0HNqz*jYacLK5*Y+2wI);Eizv!&7)E=i?aPg
z!A+Jx<WMk0aCx`gszkc#(i(j9A{eIn@+bd_TwDHyl*q8OkT?~DQUvtrrW*oq0$1>%
zN6~2urmD#yI>&WjvoM9{2ND2A^zPHagUCTn555g_mlFXMn0h&hyj#Xd5dr$0USHq&
z@QR9oJ<L<++P!PIdamGw?R4(iWeT-Ph?B9MFjI@8O+he-%TFqhQfwOWS0)1^mpKQf
zhR|kehgIeblyiO!zzwH56(m^okKbh4<;#2ZtvE8Rkf6dh$Ihqj=QJQw^ZMdzlluyH
z#H-=iU!YuJ?$mBBE^GC>5xk6VN>!z-RtQr2*UqIV%_Cv_yhiG@fq6U<{MN<`_&QBB
zb;b|NmggGkpjAdVbKs9X)Iq;s3oyR`mW4+KzfqQQMyx+8(h9Fk8n%L$6(p7?z9@-J
z76Dg#m_!dnmepcGSO2RnZ^?ec7k6{WTO5KQ_c+iQktxGyM3DPGY3{j(hui~3W2u7*
zAefXAcFaSr{?`8#u#yXk^W3YKg|Yv**rL{K&k4aBKOh;W+YWtXE~Fm{L7&wHNCOzp
zS44@;No?_P`2&Wtuu1T2kN2X08!#APfrp)pQA1?NLBy(viY4e{Ty>yOm4bb_R3f>m
zLOGX*co!I`d~(GPXi9#W(|bIFUQ&Xq_#0SkM0u$^3Gf<XX?v`93?<n-z|DZxOi<RF
zFDrgc<6k%9z?yH+DP($Ix=6pGDN3FQzThfMk48%Ehh;Rdsgd@dyH7}Sb0VIHQknxj
zIji&6-S-aNYuM#^BYJJ%f}|NDEL5OSRXvcjnRM~5rxV-0dL2e^$8oC@Vrr<<%fAam
zx&E+qLNL&1RH*6vw~PZp=%H0#Z5e=F9|iOe2k^Pc*aUNFeYihe8odJ8O(|<0`8}-h
zF*Cz@WCK`>F}P3A=(Ffc`6jlwp<WD1J#OS0-u!~z5V8DdMdi+828x|CRmRs#yHhds
zb6T(<Is?#z`cRY*2S7EHwttC858S^ZHX?%WpB5sJJCYi!<&C|R=nuO5nGLbh4!f7%
zW^Mz7V98XP8pC(T8ssB89Jfu%N~=LSLim)iUt<})<xa4;;DoBkPjVi*Npol9eH;Qt
z;M&4rLIiniUI#5PgVtC91{!ku)Q@rJ_nqV^!(laY+SIY-SlFNZWe2RI8j`baS#3x#
zVzC>2&HE9cO_tyYYQaG!&0TA&b16oz5V3mAd51seE|81<9AhTz*jHbOL-{5>*ON&d
zxHIXk9aY|!Rzr>XWmwRE&aLgvr7D)}98Kx`ZR+35rwO|I_VYw!-Ho-#d{H{z!6D2n
zGf55bh(90O>i$^!!RE7DYAG2Z&oq{9*J#VB>y(`v?9`U~Fr%h(Tl{85y7y}{g9puK
zrop@^51&=;+#k5S`uvw&WLs`!<@EsHzLiD3(|5<h0>diW?@G?StlmPaG3y3y;NoaG
znaSe9a_ok$xrQkUnFQN<vZsG|n1zikq1An!7u^i?kTtzTgr$L%DaMm&D}z&^NVaaq
zXfO*IK5Q{3=oErWotEh#WF7FPYL2@?5?Qv(8uc{VRd#fcT|nq<7y)*itldozBK;a2
zx<Lco`9VGY3b7)!m)DR7Bj|pQywKZL&C{<o?W65(jUCY0ke+`JnfUMvaaA%LC(MFQ
z?aH92#ns|Kzop(~CY<AcdRz_-g_QDeHIrz}R8XC%N+g8PpxZ;1yAt{lRGgUo$>22<
z|A(ZcWJ4K{cjZw^)rhO-oz?CqL*?^;3CSS`t05qQOX1*Uz{H5Q4=!ak?GJy8)6F#8
zprLXjn$K?TM0ll#UcEX)h0KL5M#3UgCK@uI2Q&a%#1Gjvkzq+9=(iEBRYQWOR_6!0
zqE?;==7Hq9uH>XD)_fV*Z?&p-^Jc8;Wtqo;{HVet1_oj1;v7mO$}U`N^X~8oWZ0NU
ze*WEGe)fZ7Jv~HT)R?X1p>hJFAK?&Rc@L9Cu_Zx<L=zyf_u3_fDPp+dhL@26UT(B9
zXU<1doB9+K_Lr5>>7NH6$r2pnf4k<5$Fzp`<dH?7C~tqyyv2P7&$<>#lbf4c1I@ue
zQ{Yr(!ZVo?X^NxuW|wzspc7f;oleFSrsKgy0m^x684}cZo25jxvN0h?Sa(>Cj3X*;
z{Q^u=YT+$*PouFuXXhB`Wht}DnAWRz$rn>m7okQ9W<~)Z(K0{vI8`!I<esF8t80PT
zk=BbL%*JkVdk<x+nR?U!Ejug`S~=Y>hc^hmu|ebDANMXYcOdwdHryD#WOi!kc;B-l
zIo}*bb+<o6GB|M3_%cR8+uvG7X7N4FVP;4@VQ^!r<{*me8VLu*`%OpBrD6JEgPj6q
zb2|u8Sa$7t7r!@Ldu(V3Oag7=49q4Q%>M<+Bc5Sop6ZFZE!t6X!Msr9R~}WxsLE5v
zUaQ8)l{pX7(na$HhB3E1k~;DC3L;Wh@EcM;qla6?@io>I#xO`U=yB4e>({R%-!K-P
zdkj^+pxcy3HPR@`BggUc^{wA?6?&FxjX!{>2A7s3AmG-9<Hxw<1a?a2M=Sn>26+_K
z!0#_pjTLdW-TRB96Rh|v3vV&(fpU`SyuQ9ZNa@+GWPW(CUpht-fD?JsC6?#fG<dpv
zPsvB^Fre<;93A~|hYa<PiZVrJ@%{VH@&)5BUolVl`1+=|1SKUUDN;?Nn8S1nd>kFE
zPo`f63`<5j4Zi&)n6(2l_j)oiZMS9lPM75&pA%W~N1LK-d|btr0*Bu>CRJ8amS6;m
z7HSY1scGCXXBJ1t!)y()AO9RTpigBDJ+<N`zMeHGJn0!LNNpz?J^KAnU^<M~IWgy9
z#vnV)dlG+_(=-4#7^|)DaqJq-SeXcPu1H34D(<^sit|i^W*IwV;27p~6aILz*A_Y7
z^)+E2JGqT79kup;=?+SZ+#q!%yfT$96PTeH*E3Rc#o4<73O=Rd8KI}Wt-Ct<dMoc1
zcPOSCU8iUxYBQ!blRoPsivw3o|7`DO!z)qBZ!a4Lo8F7q=c^x7_pI(wO$=L_rDd(V
zTY(&zcuu)D)hC<Mis{YfmZ57_Mwa-H)C<PfY$CXGaI@_d>7b(M=YSZE_hWP+V-X^s
z;)xXNfq~{Di_rDi-~k8MET+uo`zkz~FG5*xv%9CK=k&3gV!pWUcE$=AdqP50r1p&>
zAo<&a2^Q3_AtQV@hCcyco(3<g(G0<gy6YT`8Q(H;N+IJ6pm*R4sbX{lUKY+nODh4j
z6cMLCPJ5d@BWfq9?|0E(It0cF<%-k5@)wpto5;^Ej$CusNJF<kYTiVx;O{pNYamk*
zG#81i=>T=!uCLo+*ADp7{v8$^(LMnD&AV}@BB}yHt4nQ`g)J3};?DDTOa;utK>2yh
z^l%Mu!N`#LAeii#t9KAhyvh9eH<PY>Ft0jr_iHN-KnE9tZ9nqUZ;*0K_~6_G(4Kyn
z4nMBVu}_<tnAmRVYTlkTiB(9Y!Eo@N&khf6d1oN)MAwC)Mr@&1m;#7ikNCC4A*t7I
za4pb`c@j|00&<<AI0sxTC@xbX{c#@U?oJQK0-%K5MsjJq(g2r(vp{Kjmb&RzW%&De
z01Pk5DpmB|5)MbCyofZrjH1OK^5&OohAAqyWuFo5oB)*&7z2T%VPV(M(}0q`44<(a
zGQ*VFBf`QOr&r)ZB>%-;18_q=1~6z-p2OXQ4XMQ)Up~`p+eU`}kpcp!dCeW)K%i7x
z&jV1jM!Re8@PqjC3qpcU{JVUP=f@v^Yw>cL=hr1319I$kUZ8b9u!-(Q+CoN?ew|}F
zIBM_(X{+WFtWiTX7dRL%SS>!FhOisRI%eWS<KvQ;{WDw*m;o8sGYM@_R<C~gW4Ia)
zqM_?n6#l&DUON=2QZMA^<rVoqurEk!LwiI5KY!VePnDGlA%<j9XW6#E+afND>1K^f
z$Q{v~(A~p&2ZPNJSSp=h5XIz$LCLp%n9vLwtIF~R3|>SC_3q^j*Y3|7P4|x@EKAhS
z*VZVhtNRV;cC7CG(^KWmZEZ`y^fcdTfDsTsG-^_!$|eROlPHw!35gtBXy_JCB4--n
zY!Mv0oc5c}(k9?%+i~bEPd*SlhFURnCnuW#Asu20ZiSYZw1crW`zAWHCq^CJDRVOH
zQmXTTqdoNrfL#aLRWkP+xq81U8dk1Hqzfd}i*wJctWoDpS)G8Llb@qWdthX6hs|*R
z3v|8?sr{arfp$u`;zrXq4lV65n9pi8o_m4a;ry+H#PmbMbSQ?F4RqszeOz_Z-hFW^
zWc%}=^&GG4_9;$*PHf`7J<H!iK+f?hT2XsWyDn#0b0|U;)O5;41s)uCFM)pz2g6sa
z3Ee-}Wsh7s7-*yVegH?Rr<#m>)7N|G;o-GzM;s>F^3)YFkLatMmzMUj^Li^_P#^|i
z6k}%)VCDf&5#iu~BvH@E>mL}7zGor_kRgyM(*FcM4r;<HNC$lI=o!0q4TH>6U?gCc
z>j%>ca3?T0{Qx)j1r3RfO&1mag9~8Zd?cZAy>SpnH>wzIO<_xqZTBS2e{kKzeYjT8
zq+dsQ$lH34^$nQXozb4yTw&&-`kgy=?5hgW<f3Kwp3mMM`fY%Y9h$0d+}?%IZZ;k#
z^xvW=HC!3<d$O!zD2`p2J=#c7muEc<;NhfjcrCQ{DE$0AKdxb9EqX4Vk#V!<Z3n|?
zTU<7Oc>MS=VmspnfPBHdDZaj22PGy%B<6pt*ZQ6ZyaJ<#%3joddSDi`hEh7<3bl|A
zV|Ds}Jn$q@hixLn`KQ1~hC5#bvDXdka*@kdyqD@nVDVP;&WVeOA<XJTlw+m;xB}T&
za9gY^_$PRICO6E6<tuw>7c+I*9#sD=r!{@QSfKV-SGYakomWLwRb$w(o0TK+Yc?3J
z(Wu_N%cmc$5SO3q6o~>WQ|G-MK>PwQevlKgp|f;#W5O=#f*%_AH6#!y!Qp$+7(`!Q
ze5*b^{kTu@n_$&6fc1+;7J;RhV*KYY!x60!Qf2Is4$92TJat!L>*6<8u)^t|VqnMF
z8g4&Rm54c$uEbZWgrS9CV_(GJF~FQah#Z}qSj?qPP?6bPEdtlyAZS0x#kJe?yzI=Y
zkfC^yLFpRY?O;YsbtqcmB7<jB2c2L7vmxqw@#15D#nw?kMV58%o(xq;9q_h~{Q4Mt
z2~a1Dfq%@h>Hi_l9}kQGburLy2BGz{o!fH{P<P4Vk`lhAu-TyodEFYrz^MZx&-$el
z6Tu_H*mdH-3?O|pI^Xro_15X`LJbwgb5&rMCgIdJ`pP1{!#DJ!2!bFthVg1`6Xv;w
zE19fFNmg-{m6np41wtjrwI6<>;In2<5i<5Svsa_Q40LvOVmTQY##`YmiX#v2@~Qmv
z;MBTVWPKp@6wgWOvOb^=ES*N>TGmdW5~?B~9z<2qQhvD%{`{f}5;Z<G|EPKs7RuZ&
z7k3+qbP4=)+qQ{Y_dSmP{_{r^;#2Jh(6c|;;z-I}19J%j7z>Q#G8dZGlaYDJazB>p
zN<B}C9J}!fT?;4!X%}{Z00Z-HQrWXVAuu>N1KT!uHl%QM@fLb|dXkWoz^SNLn%&I&
z_ZbWJKayj83>)RRt$(XQ9A?;rXP<hzWnIwk9WNWPFoaSDY=p+0%=3huS40~q5F|K9
zoI%KS;W~)?SsTMT;~ojq2R)oc_!>mliE9O$-JeB7B+TT=SZsPc-irHA*73&dv#Ot0
zvhGBgjh#JQ{Nl=)6HRh8CIcHbY{)|rj$~z*Jj}NUT4->Pcbj*koK|+WK~`UcE>3uV
z@95~^@)ENcN8|kw_A#yV_~D-!t)vK9%qEkh8dC;4P9QLVYAY`g<ERNZrz{BZiwyha
zqCWq&GBh#Wk;h47(9j>u9~^-=S3f*c2NGbIVd0zK|GwHwyaCE0!M~RfF`&c4x|idU
zv4UtSZ^u)F$3=8E5Z|+@>>R9@=H433)PKCM58%)_p_R|L1E8-TLU1x{UJ37n0Rg>R
z&iT@@PkU~_1QV2*^<j}9_cp`*ZP>&!c_9%AEGlkfB&NGz9ygnqFY9U5LbfnGTF(-P
zz`IZ95UpU~ag5a{v=CAM9yCewPc@o&y)~L;x7f|wP1m}9NWJhYJ=jn*P1}U8C#qE-
zO$i4E1~taD;9zKb72~=UEKkW>mH9z>6I!%@?(O<xqznz;9QS?o?Acxg<=<$i{ib2b
zpqI^tSw%n0%P^`aLALDP_tr6)vZf}@<!qka@^WZIXu{F%?d^?8nUX{m20M+Co7(yF
z=Mm7Mfogxm?Sz;CL3@;Ok?fFH1yf}O7_6|-fcxHozUo6|xf(&rS=RcL^>H#+Sf_ju
zCDPB`&v=1}h?8QHl%(g1sYVU5J~CsH8#isDbT+BHJDCw}75?_md><u1S(qD%uX%GK
zBMk`<DOesg_MY0e6cWGyDX}Kd8_>kYdBY>X%GQ?c==H}ohtp!OJMth^^LWm@56Xdv
zC163x6vZb~pga%s-3$psRTHK<NN5^U2XG^InK5lJ77keA3ei4aJI*?&7_dpNz*)=J
z(-9Kt*O>FS(y@R3@&#;*0})xbQ?5XEa}9tJ9mb;Ls^U5TsXSuMF1Yw(ADY7O6IioK
zf-}B6ZFH>E*Y#)q+5B6nnenW&?63;rEwWlCQoKTTvWP(wJ1!*j?;qW`xiS))rE!MP
zCKm-Sl)%CX$49tgDD=iV$6i;gtP;zs86snar$mQPThwR_>R^ACm>vs54lcKb2WTbd
zS`;y@`-+rghwguD>Nj&6>Z7SLs@f$5tVuD&kedVW3c@=ax+NfVx9I_u!L!A<68XU4
zkud6H-RVXflurgX%zGHzO8VQg9yHMYgUmbTa`mWx0x_%jT8z0ujG9X4iNVLG#l&=n
zhve0FoLsKw-j?+^{i6|Y5OzfoX*y&R3`>C<qr-aJ1J4fvj~X2QR1-Y(g#Sp>;++=V
zW7m#-%HP2i;+zeujqv&_&Q=ae$*_5E@$db6%~)k(4B-e!?J+MuUww`Sb9!!<PAhMJ
z_#U^;i~AL%-!dUGi#ox<z&K>gIF*k0Wl;DM+vC)?Z)h-{;VP}GJ0>8Ih*)KEauTUv
z#7GVNN!!aEFC!x*&FXm?R*H}qilcBYbUR(D>}j&g((O(=R0a8d77=}Ov!A+b5IbaY
zgHsMpi|Jr0Tl^6P3<@|!8>$#ELr`MlY;*3x%3)@~b3{HM;aeU%tgq6J$HSpbwfTi}
zmWUJxNiTA)c?Zxa-_CR<SACP>>B~LnRQ~n>Q6^~JS|fK_xb1=Okf!&!O?1F{Ar<i*
zIBih?Rs@!fSuAgPE4$p!Iz3dsfM*~FU4o+`7&N{Bns9vT6gA}XAcz!y{raNvMn7#1
z;?EknPrvRcisr=`6c(6e5B-sWLKyF#;LalJQmZ}>gEQQoo%ZP!zd?cSOOGQ^L`|ra
z_plZt;YiD=4v(90$ByP-)V`2C@%^ek!>8sM9jncBbZ$UOv78;iNr9$eo5<Kn*t|w9
zJv<V#n0<0#p_{q6xsdHuL?(t?JDQs8plJc->y|(Nyu9fxD5n#EyxPo@lgD4!uzTCo
zy~3(8-ZQ;78K)d|_UhHw=8Eajjekm_0)YrHMDfYOT;Uo47Y>wA#?;=`2{m}q*yg@T
zR%4+<1f>=bvftvH#ud0|0tr;^3=a)8seI2*dd6V0f-W|#mKPVtkVkFH)O4{ygjM!A
zVC(eok+4g+5-N{K4hlaP_1n@7D*_c{GHS5Qv(s_-aBpBsOb`dPX-M+Xczi%l(<*9t
z@023-W52|3tZQ)ltSb3?SO0NNhIeac2e+|<&S}1lF6Qmq5wLLZS;&jFS06x$%>=rc
zKwZ@5ByAppq<F?R%G1LGT+l<v=w@7J!C(&JaEKxHV)E5AYYN<WL?0*DrJ4FW5c#!Q
zYJ6jlNS2IRS)qF==8$S5&%!o0_#+B6P>(WxgU=@yFo+C6fo#LoX?O<Ap1khC<)(k{
zb2sVjL-aqz9zL`}*Kii5hC1<~@SAp8NSE9Zepu6wS!J&f9=q+BM~3YsNhp1(7m7>y
zg+ssZQHoZ%#H*Ru4xkfgredUoI2V^ODh+UMyEg2Tlw3oe$X9S=5w)#&$*k;bXfEiP
zOug{mdd%I>Du#Db53RnBY)1H8sAI<CkHrFW<@a?35ayXa($h1A0`%N)lM;Wn-T13d
zw<U)gKD_70fwq*A`pb;-qHVQ6!`okeZ(6?~VsZuRLWUMp$}HRAkV$b(pEd-eQM#s`
zP*G7G^raw90S`$=WB&U&x&}O|+F~>VD{H)V;<0l$>}kkKP(}$2&dtej1-BW6s*EJ=
zve%D(=gAdQ5Fjz~swE3dok9~37B78@i4XoqUSyrd3D(O)CY^6yK6>b;_`r7c(}NRL
zWJamfcDMrMz5_TBTil<8gb1Tc{y|szz3rO$kjt@KTV0{t`WVl8!v0%T)h@|2?5X+0
z53y_0KnB7pZ)oMPT-e4Arj0n#Opv)R@~bGQN6F39BgY%VA^Fg#$rc#2{NoyE4ft`a
z{}<E)N4vXKKG&$T5pXK?d)y|vm=k%EGc()h=p>Q;#p}*|P|rBR!xO7KC*i|$>{wk+
zO^Vv--T(IR7B(e|$U|rNpr=r}cn-oM*4qyVqy}0@#*{g>#6@#@-b9c4JP05dJ9Tws
z2KKBtG+X_pLL;!}m5o1ew`KM~03+z=DX!p<kb_?Zfg2un;o;;wuUkw+xx>9Q<O7aY
z*z^=_xp1Nr;tQz7G+SN0bjjB`wE^Y|8;}rBn#<wVc*=8a1`SItvoO%pM_<x}FtXuF
z{?B-f;p}*)z|tW27t&$lRC&`xug^I+8LplBQrltHPMFqmYDEF)M=zV)YftxtZHpMX
zP+GbhdP`Fej#;#yYFGrUqI$2kAR=|FGym4_jp?9(`Iu#E*ss7;gA%Cx^*ewFoB)$t
zqk&yM(z~tA(9iFZ&|v_q^lehOO{h^gA%kOl(r`5*-qi6se3KR7xE)>ELJ<x=)_ZoY
zW{VlPP29mKnDEH509LW{GBO7b9`s(xJnf#UgQ6eey#z@*7U4_Wk1SHIVBEly^h312
zY<M4E)-WE|Mif&70CmCSfRg4>UX6In6FJCd^W#T1*n^?|FQ?=vzgBRi<6Lhb*C0K+
zWi$#01I<z`FQ`OFOd-N(6H5k}2yGg~kmjEV$2fAa;X`X8z3>1tX^dOkCxCkJF<nPi
zhHeCXuJ7jn;cjIYyP}nanv-!iLb_lNF6SPN{BBLff6xq;&jY+cEUEG^jpN`XHAbR_
zNr~UPsgaKL)azpvVCX!kJe7M&JmI>=o$yNsO+_Rh1ZI!y74UD2oI$p<fOKh(DS0Sl
zb4_u{H*5A=)mXt1U4QD$#0a7Ec{INo_3f6I0akULi(n69cW-6_+o|@8#z;-IwcpUU
zoPpbvmq)%hhYXKzyaeGH!!72Hwl<vw0fTJo7`%#RqOOWK6iO{olLH!!-O*MN_9?EF
zm9iZ_^Z-VTFonSl2L2tZ(TubXXK$(oY{OWK`N0Wx8u#x&36NnT^){e91xp|YSu>pW
zvtZ9hk0|=lH10=kR4^NJKWXn)hF0;GZQIx+UGw-gQ!C&#)J;SY67q!7V}Zdhz921q
zsOcS|IA{l-77(cNOVnP$=9GDTEqm&qb;5PoUdU}4MI>n4@Ey3ocG|iR{*uv6{oP{5
z&gwiQ<|1`Q=#RNABNjvd<w>NokNqARTH0Nje-BVE7WeCl$k|V9-@bkCCL3?dcX`Lx
z`K<o9hGcA0p(-qJg6-2vIg!0V5%U!E!f$FMfdHUIW{0l<QOg=uKMZqUzkUsL^q<>z
z>CBn?@e_Cf6tf?W_>du~C<|ZyNT@l63K*Vvj@oz$yeW>87B3sWM&ZeTv01@wl4cPM
zrfd*_RB>5aCjg((F%wLcMDPJ#7IlN@_gy~5kcm0>Xse3;12T`qR+7$h?)!r0NhR@_
zAnT651Gwai#Kq##rmdby8*qDkhtd71pCNlUBLHhky2_<*@9pg3l5sR|J=J|fX$(sx
zokKY`ab{b2AT;|=?pmv`M;cKbEKS7T&7c`T%OT&aq91Pq;pog3dLcVAXuVljBwD*C
zSfO5noI5kDQ)N$!U_twL=#e*SIf?u(=HFEgl9{h36=6k!yd0Z2s{kqL;e^b1$)&}`
zuO%hrym=tp1LgL&?WsU90J=OZ5h?L#(mfZp+l^3P)D;kX`b>pvF{6)$x+yKY81wGk
zT0u9YVN>i;cf{}LG_S|xvv-$F9J)Vyfcz(h8Z}b5nd#|e+{iVwA4iHgs5;D}wC(L=
z@)i_g$Sla*+m?-A!bzDOZW0s`$wAPI`UCJ(T%2Y&T_CsV0<+%gW@bak<p>8Dz70H?
z0x*sW3JS1a$j`4p4`9g5_x0<&2LgLpxxn<r#vr685i{-tGQ#DR6^u?>MdFE|oP4i?
zLCN)_pi^pcrIl#IYN2olb|&dC#_A%bohHmu^a8h-S5;Ijqx*+YIS`)5vM-^p?9j<Y
zQH+a5coSS+AucXXL^!a09~3Xqqf)CD@>`p*NK*qaU8Of$VayH~vJ;{gLWv(vLOBET
zwXK0t+R`pyX>x6c55C$=LYIVm*%+%YpoOXqvv-4YX#-loimP1LCqQK^N{uQ&s9Ve}
zkl)z~km=OuK3DtVf%kuoMvcl>kKNqCvJ*DGaYKWmR{C@L-lUheZQtHi=O2HsYq<X1
zeec_M?~oane+_2?c~@06V5l<yCMlKn*9Iwj5SxN`OJeM=xq^@l9t|O&WlO5GTe+ds
zcIcU-jCv?s&JE9FrGFcIWFOaChlPb_VTS!fIJvmA_;+3CXK?`IfRq;!+G@3V)nX^m
zR`9rs>J5_O6Z`kW(L*(U$1Wc(WTt^i5-JKhl2EKz61{yNQ$tDrBgkrHWmwW&<U|1X
zGGR!;ycZM{JbZW*NH6vhn)bEc$xI&rFHa-<z)%0_w&gv3y#C9VUL-=etDx>at29j>
z0+1tXvDVx7B^(GQ__cR?{Ma;bM=*?>s2bYXr03=?!61^dON(Eb`(Cv;Rjz&eq_p2i
z9!XwAY#w~l6Vsqi0S*r@;iaYnm?JPk`wLlt5KX!l6P+9|FA-{yho_9V_U=9N!S8u|
zGJ2a96J+ZsXcSX@f9Tdn8fd2x#9o6N!)IS4Cn3zDRCDQ_UINJ#jmP>L-I!0?$W~kA
z;Gpf+3-I8uaE9ZI*-D{G7hL3SW@bz!sz+7vRs$OP9xyMcy2!H4xXW8g!Z;kj_-c)P
zhYwdS>h{QUZa$0bH~t0lA87@*kUfha9-5*Ht!g46v$`mMW&a)qhIH$Gxdh2kFs-oZ
zq=^wEy7$g%B6UM}3hgJ%a|j$uC0t-h>m|?I)@q@al)*bpR}{>#;R8rNiobZ+w(U9o
zt*lH{IMz1@8MMJnoO+WFTK7a#&EQEgl04nN8YGB}5^)<x%SQsD`*m-mk&bZfZ6bM|
zs7c<tWao$*XS>JO&rgFLf($n<ba|)twPWw#cguuhPP>gJW7=N?EdYk|8iX2gZ@F8#
z1`YBJuSf^}hF9WvZO2<J#l^-T^&!(zPRK^R#q@kGYTT3)34j9O_HQH%j*YDXlPTkc
z*)GZ9k#f{t00%!vLH5Bx(3%5Q4f_7B`wsHe{zg)tgt0jA;aO17^$x|$%k(%#?^}aT
z$-+I=70`v=gf#<YN*8q<wEL*@p$!HKQRhY$aKM6Xi!^0tR1R6$E62z&_DA=W4~&gL
zN^X;}2QInJxVw*_3i!y^ZQpwBjUU%WzZ>e=@vX3SMHb<g-XY-laPNwv!FvE17yMXj
z7VF@|3(ygUM-bHccxm?8bbNws4_^?V>;ABgob#%lLz$6{bJ_f>u|Xv68G^04*L6w_
zvSyW$=<eIxAi-Do*G8%4G?S<4Rmi`5tXa;@Bx&effkGS4O6ECF8{gZ?NG`1z^qj@>
z{y+BKJete4Z5wtqOQR?nh>$TwGS6c_$vj3f&kdrYP)QOphm4U-KQjqMh>V%1CS@oy
ziOhW4S=aMC@Av1o*0<hwt#{r2xbL`r!+9R(aqPpkZ~M0M-WS9Y5-^I%633Pbu=_3E
zwWU5Xa)>o014TF{It8LuUgFMcZMDCOpxP8XMp;KW_@NQ|JnULB#n-Y3|6^U~P{#p7
zAtVz-Hk7?^Q<spu!$QYIM=zsgz-Q^1ZkpC<t~tW-0(8dy<doh;m|{YXhf#uP(nHk<
z(M`@-XV8~*KzvV3NSM(rxQ?Rtr=zyOg@DL_Ib0MF5Fo_MjrRKa&?e+#jRZ&EkDh4P
zadL7(KS#P&YF~f<RqR}x8bF+@s5b8AS_89}ubX^`;CW&gnZa5n=*!2>y%+<tIStMP
zpE}D*kn%Hbiai1#>FmoA31$N<B`+9L2yhmVo0^)Lq3lLHAHXzdOVtxB*t3BU9j>!c
z(*c`B$aN;g*}DVXNeC?Ndwb7eM+46N=bwLI@c=fU#ISiM!>3m+c>>Cpawe{_?iaco
zZ_k5>q7R0CAm#TFW@ZeqG;izln@4FCe0JX9HNV{`8UPn=skIrE0t|5W^l5!C<4ks*
zj6B2IR#^dBh#`QoBLQL>9}czr0w<yrr!na>)d6fZLP`NE96k_Yz`b|yhXv}ZtUcZu
zR(z4*BkpEoyd*Af!2KA}I>rZOgP;OWzJ-$V%Ju82Dk`Wk6La2)!9on(=pTOjRxZZh
z1A24k-4`rIPqu=%La2DXdxx?h4mM^A!RZpxL#(Vx0L|dRa!-z>6S^oEiP+>C@}1C%
z!@}YJ4kE03Ki~k>3)G;^6=_#ah6V;sVfD|g5d&~gibAPDUQRCJ%yVFK*jMje1;v?|
zzCrV^uJPwH6unFvR~+FcFu(-jAmGLsM_)F2DzmX;dD`GCq(ab;3YCyzY(<1l5GGVd
z1ely`MbHt=-VveOiAz~{<eksA>^Ol_VfV#+^c<1iBgaOyx^0mf*H&I!ej0!d;=-(A
z9&aKf5(1?SR15*@<_p-5_~hxIKamsa7ube5eH_3()U%RgM<?Pl?B>qGeq1e~Z33DN
z!ZxCJKUE_M`X&In5xHR)5`nUq>csV7mST6&RV#h_lP6B(U^hK-eTds{jO71;4dW$}
zhQ&k{^XTEj)?OtkVwN>An&Rn>oB%j4z%U7kiQ_L-Mh6FzJ_%RTx0CgW5u|l$<X`Mz
z-~0;;=Rg9ZIdI_S_y>W0MWO@f+GK!f9BLb+ru%mf1FgV7g+i7k2n4Ju9({XW!YWY^
zDJBn(ipRR=0Rc_OMx0|O5HS<=*0WTITQTf;1_$1L)j#d3m7IvCviVkL@hQORDX040
z>{~|5P5UH<fMZuD%$)anZ$E=2M<ROeXa(~YnHx6<#Mx=lK1UBSI0F5md8uD$ug}{P
zWr^D2fsv84Xu+e}DLq4-$<0Odi2?^8@pV{GP`Bu}zM_^^sN<FUHQ(}DsJ3oJKzrcQ
zEA^E658d6}n9$tu!+G931rO_E1I1yezd(2<+}*S4gi-yxL_WUq6rXrnyHz5Q<-9j5
z;4JMKs(W3J${Oy^>K)^Gnk3N!pX*F>2+C;n)$`cZqYbj*r_vjQj}5|{!@2=+IM1yf
z<9|N>C+iUd)OhvUa*~Nc`f%_5Njzh8p&@W2m)SNF<7(6!dF_*;<1=K}k^-lH|L%P9
zcnCLzYVjyEz1#&M#_IBWO#jN*Gsy!h&@9T2WU+#X07lb^xIxN9kYSZSs<BSD_V2xt
z+txiJNNOHRPc0Mo(a;cs)<ry0M_$>dx3{;;j0a4kg8zMBAm!BHRL_tOXeH>>6Z}HV
z(Aj-N!P<ZEphTDGi4!gJ7YY!5U^BK}YX0`e^y^?~!M?)I<{uLi5&&-ow;Fhk#rh*%
zbm~zDkL*R(&TF98k?8VT{t@EOO-tRfZQCczAVEX4_U2T0bs}BCV>wF&h#Y30=vvSw
zdM6Zdr|DOHxMl&>i<nnSdu^rF5^M_r#hTz8B%g`Yn6YXX1X~xj*p_uS!r3YW`CALZ
z@M-OEXha|QEsT;B2lR>0MOxTtQ*M8%3@AN*ZT0v9h+P7Wq$u$doo@cun=TUjb8JzJ
zm$#{gr9?*D>xN>o8cY<?AqfkPTN<*u@AKe+9=bR21!%iAxg<Dlaw-ZJgQklhoVejj
z+rEyDio)z+`PN|h`_8-YrRiRgKRrD$Iq88GXA}bO-@7-Ym9Clmo@ntv`~lQ@L~w}6
ztO@x5jHy(VlN)NM+_EJC`Z4HS(-QOwQNPxTnN_87BLLV^;B3mR@76N-O}GCS7r^|X
zpI<bf4&vo=H9QyV0^8tpN24|O`9=Zh0n7~fBVi0%joJC!*B6h_2$DHKk2F@M^1H_n
z?4X@bxMM>zijtVWkz<+rVi;UYH{|$KOA|CJCuhL@CGICpIYvxO3=aUngh&VX7rxlo
zSa_BzHLkvxI2<+nzR~h3!hYz{So2pK$sq?rRiz&>j&|?<yq-^RrWk~JB=JxSd@eyg
zF+6{GiwqJ0qMmJDOT7}>hkTO&Tyo8QR_DxN+u@miVSCferJK(JOs@a(5*5e2th*n#
zLKzACbDMW}0N4{!6{zUjT}=$feId;tvPXs0@k1$&gLO!IdJ+7AMY+oH9SIrH_lSi8
zcphKyE2C1i5#x8^xm1*t)GkgkfgTG3N$`eno3roes#+?SOmCRcGBCIUddHW1?Cmv9
zM^(@cs0%0y$S3`K-&_tQNaQ-xi1@=VGeU8w#)bDQW&%P3b;D<A4R`?}_n}kD+Kt(C
zd-j;E^wcar0KkBX#4>;sG;|W$oSK@N7y}6zfi|GGp2|<L5(F&)A=ptiHt<6@fx!w1
z5pz$_1!mhEf2ZIq#2m=4TNe(v-MR&bjQ&_3xIe_vL=1&p+Udl8_b^?6xEE%fiQha0
zm9r_hJLr7?7)R3Y>PnDEQ5?VP>50$*s5{wX_fYq~WjuZVfB-;7gGiLYO}v4pifsZm
zhbK<;-pXA#uBM`=*1FO52bXV+(@GRKQA!6;;pyo~lvK(**5Epa8M&Z3NIG-o6{HSO
zr<=pnVNA#j3i<|~!Q>=C*TrAoy?r~hOy#pyc=s^P6%X|m_)7MV8=b%x75Pa*^7d1H
z&+ue0ao&-&rDO^g(o11s8WCL}eZ7Xz8z&Uc*~;7;z;_lN`kgy>;5s3^@Vaw@UMl(z
zYOaNL7_>62WM%8?@47cNHAS>2gZB>i3#Xjpz6;3T>({R_z__xa0v-|Pz~<MA#(P)(
z$TN7hnXBrHSfpNvfe_8VxL;t*;^;y^IFGeCuxr<amFgjIk%G2g^3C!_sSg)=5Qr5{
zJYIDLY%Ew>a2t_Ta{yjOMTO9gKoYg>qH|bC$W=_V)c*!AfCo7?bk5p(2wN02cB1VF
zkCQ;#kSYSG=;`UfL3aXq3|R^Q0AvPe(P;+Ol0oN9RvGjNzhY3h375o0l~__#)N_3)
zt${EZ0v~H};(&LAJ$V9nqyrBf`;SvQ4dyHXR<O<?de-H}4MIg5$xSOBHsF~X@c9@U
zju=4#ViSG-<M5?Fc{(X&;I+oVS6G8}#<-hXX#c`T`1A&*G3ll!#NCG&?FI3E5KcH)
zSqG6TK7IPMbL?DS;T3c9XR}sU42p?1Ml4S*EFMl90s7(YI?YG#9@g7KKG2RR6)`YI
zsG-7u=FUq*Sige-c^S?eg0U9#o8p|0Zmx$W-X#?~v#V3P_U^TGyPhv&0E<MlJc6!8
z$W6c{BJF`2ga0L>F7TH6;yPgs>_in-y9r(%9v|@GFPIrVE06`=B>)02A<U*Eo9$1*
zjtNOftipZ~R`F3M*(wrihTsyfE4cXCVDzSjYiFl%y>&Fe2PDEr&n*ymLkvAdTCcmv
z9%mJ{Hb4#qFIL*is4PxJ0VNfJq&0G5^dlY-7G6NX53bfSucx<H66YF=(J7}Y;=koC
zyh!?3Ugv4e<*dMh_>LPy)d<cJj}5i}Gw}%Be7qC&f!~;?yLBsitiADJ`xqE5@@J-r
z3FsnFe9*H1$Z9X|`CF*h5zR-vWotw&zorJ|I%EXp<>f?hi7h*Fs*1If0JtB(xC3*W
z0pv1<oENze@Uxe$vuRbPKY#tgUPWi57HR{2Wf2Bi?=nyQ(89dR({rDwH!z8bz0!IP
zjk?=wt&yqci%%PoB7QotUS_E7^68xn<$s}dT~RIv2Qqlfh)&$q$>#b9V|1A}L3f<Z
zM(W5K*ZpjOV-L3&(v76nJyuF18jnNc+SDSMlt(l^eCANV4hZ_pzLk=)2<q52bLpSd
zk6j=%TuCLGYhRbeomRi8oN$4(fVsy!q4Y8R`EV)~iSD!!g%E#mF~!_JgZFO`iVCSS
z&%{Dlce<erc1U1Po%%!xLvc}&8&J7HrW~KK^`qavp8{OcI<oEd^kE^P8xWmn4TfRi
z+9MaQXJKWfJo#zM-DX%eAO&GM9-+90ik0K9ICcPQ#00aT$2f{o9&OxEzWaosz;k!7
zO+iVJCYT7pFaWJ1s=`RRuVCh}O)d6aX2|Y@0b!4w8>Q!Q!jsBSJyjP7a-z?}UO-H0
zQwwuV6;?`5F-G8EHGC?KvEn89BAODQU&%i?!F%}dSD=c=s#e+YA(Ot}RIWsTb6|gV
z0-^2GGHpzHV83{;<<3$ih+pg~>5!xPjOLwWnkCII(5}<}6e6|j9s(C<SWud!#DQh;
zS7;AAdBPQeDCvTrnbmOwa>H40%DL1O>^s0=$h{N=|3TDslRDxYm9G7|0cX3ov@Klh
z$K+FxKOf41^^U!h#7bbg;MOarO9Jgm8{<vdTz4PdLIsLKIAf-@NS|Tf9L(ztX<}wX
z^02vaFu1GUr#P-M5QE*~<KHai`TEwQCVH@7dltVaFf>#&oy>C3Y$9^_a6C=PhWKSV
zm{|C^ETa;(#~2-v1o$ASJIPVr)z%#>26UDhK8#U!_LyxJ_7<!n1l_UcgC5r;j9{*a
z)-Hw369*bD@W_MjX3FA(CM$r!Ngo+me@uQuo>Ik7T~);)_?{;O1w`VBv`%ci?O4~*
zF-xrG`1w#x7-N)~GBlDN9^L{F5j+<5aR#v8c6u$D+|WD-k*O=+W7kuBy@$PU+kwcs
zY_6AY!l(bm95g??#$~X~FD_!TT~hwdr|%YSvVYoHc>vqTeV4*o7DABTS1y{n?g8w1
ze9icR7LGCy=nY>4D5Pw@o^<L3A7i)Gi%pw0J-%o~XtsS)+>*GquSDQW04h|vA27Ee
z)HwS7gec71(I#r*6`~Q!CBy)bKR-ge6@TRb@tZaCzXE_b#BBfh^Plzk|9?$%mv-2F
zE-vUfvxR};_EIeou-Bt?tqdZz)Dzo4C2+)~_@vL6<Phx@V={yAxiV)3*OD{{`wln2
zHGcljm3@5{q5fB|@&Ehd(v5is?MuuxU7#L?H1_R~D4aZ$E}pTX3NZEi_pyvsY(P%r
zwxkkB0fw@_K3<G;92^i5nG4^3_C4rrI@J63r*RWvu@s@KEY5`ZB=IA)$HeLHzbGh9
zqIKopfB)yt|1QLT*5vO({NKDGiXLAh_I<QG&_Wz=Cr(3<A$1(*r%aH^WGo^#Knd&7
z@Im-Rw+ZBr5DaK>Jqe6TCm6Hrqoc`f7Z6{f*m$T-n4;N&6BVj)34EZC7czIuIYHH?
zSA9EFUai6Meq_5H?9<Z6fq{^BqTYP!27DEKsc^P5MYD_shEv^MLxI}>;p4$>M8oS3
zbwiQ<jT?`I#Fsl6@(K#nvYXXR;hrccmIKyjO@OP5X8nU5#EVukVR3*xKeBDik#?bI
zR)oqo#aE9_+vvU`cME)u0>cj=27%9UZFLlaf`zEJi9X}V>k+%~hQj;gMhXQuSPF`H
zMh}lQFt5E&R&C?JOX(H22cqMYX8-;Xy!FkNKfY$|YhZc_g7m7lZ)slRYi{R0L&ys<
z4ZEx8nAG}sq74oo<qgvKO&1s1oFFg-(NHGz4ABLJ#r+<81`NsCeE`7oWLH+4?BaTg
zO@H69nU^;U;vM7%{(&v9s#4l5-lvi9Bb3=Oyp4iRj<_eO%m=#H5%ytNO;Vx=mO}0E
zNq||1Y1w&>8~-C!i%u#)EnP@`QHw3v$q1fot-0d`58}py6U<K`7aM6w^j<Hz{@;5J
zyow9?SeW(MJMRZA?&eWKgYXc_2`z;=|J<(l<7wk7CaNmrjgw*Y)n3rDu4kY(q6ppk
zXE8Xz5Rj3T9Ty5z7R6UYjUC_vwHM>zMPBh!T=;8YX;BqJHsd8;XU^Ai1)_;kF8ml6
z20?)e27+6K$I&;mlVa%4rN_U2bXo%_EisbdiW7#q#dD+K3LQlpW9r;X>(vbYY=9KB
z6>b9$u${^91ME4SR`(rE7=0wezN!2a@d22aO>z4<k|z4Bv5^s^UcG&zxOyr5RtQ(_
z+_|%V7BuHZ%XRt`q|Lp=hJCVC;54Deh*YDhi=KF>Z_z@4KU9zAjqQ};_`43N_(art
z`1K-*yS*)QLgE&%4QWExuK-```R8jHJaw-FEFlzDuU!9g1H(Oe$!N9+B07PNO}wYL
zz2nc?xkA0ql!Yq!r>_*QK#xA|fm+HTJBl~o{@mhF>ff_quhZX(NBU-mg3c-qK_`%w
z4S4W>*P;8uuaWn-P-OV5ZGSJjhHc#3=^wR`KOfQ!*-DD**RRKH1y25JYKlDv{*oe;
z9&i_kY#pf=PDO$3YHYgl4~6zUV)CQ?)ZAQvzd!!V<pUacDf{rwq`xP?z%=H+U({Aq
zO!fDn)V?P^kM-{Zd)}GYpiA*}+uyJ51Tq9G(%}t-i-EpAjS74T1v~#4wl?5lT8xg^
zcGzZ=bDJqb|GojenzP^<q)NOc4&QC&*8OTwky%(;3cqYcA?5RDahcNoYx`T&{#{%`
z%Hqb2m$}gkTPdRNAr_C%U%I^n_H+f}O63xmmC$I!2gqoamv7lTMI|IF>ynK#M!dy0
zfSRFj?b?g{x8{6a_wJzr(>g+Z+i?)B(alB9KMaD6r1ep?y#~wx@iZcvF+{n5ix5x~
zcP))d5b#M({Kn1;q$}wcJF>Wcf13Xs*T280{8`#R|NhUP|6QN|6Kj&0^!ZQlcf9W7
z#}df()J+4oK|o3KdIe}xQE^>XFr=M=qN5FGs`0|haewoSHz|1Rn>TE*7I#G{3@Rk1
z<5FOYqUk?@o0w(=0O#Qsf+P%xiiPXU*|R$SpJAEfqYwW{Q{L<Xr~?!VyTwL30%UNI
zZjFMCfIv`?D2o1!TcrbJ6)^O+d80HP5K13sSPcsH6MuINohd(w0b0(qwAy?)fq2ul
zzj5meoD5W5D=h`so)KR+#`O2=Ud+$mz>$yY0GnfIcvycOaqj--UG&kG=K+=iW8Hh*
z3jg7D@ZX(!dpoNB=yf}(SM<QesKN58wY4>vb<ibPw|>2EYiQb=H*YX3iJgy+5A+ip
z%x~lW>$h&osF_DuBh(8klMeFxp(b?j5o)jC!b54!1o$Ap+_TEcB@n^~)P(BQB}GpS
z_-v>jg7=|?2~UvHLV|o9fH)Qz6jCVJRLa{U_@J;b{O3YAkfq1JC5iDFq`N)#z=iKR
zO5Ffj38rZ3BU)ojQ$r~PX*ftaP@Yd0n1?KK3|N1l5Ftso^+JF%k}nBZ3piU$`8did
z2q8cyZD`Pva7Dnu{Q53F`?4#<#M<eP@g0D>rMMk?%&7R@{rjBtkl952;|1ax4n|I`
z^I%nnUxGL<sL|-SGH^!uh0sLAoD(!<_Vb(*2gVGnysc?x8@vz5G#><g7>GMz`1|x+
zwaogkBYNGJ956=*jFhkd%)5Dsn(?NO#HK=9jek2aJOL!<ZKv!yz<Nf(1$4|Pfy&BC
zq*r!|5&!GwJps24Xka3DFI+#s?G4g;;MR@Y*>v<75NimYP*pFrf&AAW2HY50a2Hta
zU+h2!F)Vcdp9lHxZk5smm6=N`C7#>(g?;3#H_Ahhbu-X60XaqFw1FIImmQQ>q5bCq
z;1;zA2tUMrJGRyo%ApC+xqtzJ->44a=l%QlFA}fdjzby$Ze(kX7%dQFOV?I!{_KLR
zgpU5v<Hx6<r`dAp6`T>!4t2L7utFyRVuCY~kdOcnj+%yM2E!oGo=rDdgeD+NmQPw#
z!|VjK1s*~znNS+!IR{h$xcNyguKFAO_(pbPrx%PHod{W_{52QsezjMir^Up@WgXl~
z&_)e103@XysYa=Eq;L%dgj5D>O8E}`0g%eu_s1Z^)YNgo4W|j9>Ce_OVra-3t(2}c
zh~%~|)ZqiaO5~v|k@3`0TKXT<3-O&PH!*cgn?Lgby%8Bzc^81>76s_1Mc3O<elMhJ
zJEj36p{-6^f3<uwBT-6m6+hKH4|o*&Ba<7m6S^=7Z~H0u`rD3PgtMcdpk)ggMhB|a
z<6b;b)RsbN>(-6xV7~!OF&GDz5e;op`dC!Vi^z0+P-VIRjj|aLo)+|;b~t8;tIskL
z!#G?Nh|rPZ_CuVmE;PyNgG^{J7#oeH<5E*L%nXZxwgq7Hw6o4@!050IpE`&BT;@dO
z#LMB%01w`)Cja*%wsthiZ}j&8CnydZtuXf$YWN7eG#uVFHK`Ve2?%H>>*Bz-xU^d?
z0UiL>XyA?Zp5Bt{O&5r|JG-Rezo1U^Sya;yd}MJqVXOyeFz6cz+X==Yw}^;=z655~
zK<`>!e`0tzb+LW%&KS_-5iw2_sLyv-Z<MZSY;?YQm46=5BnA7?Kk)-a;3yiw@ss<h
zqr(|y$jcUYU;?NNyl<V@$jDR!<5?GZ;AWfR1Vu$lP<qLgK`e;w%Zv-fzP`TU;o;@K
ziQVEJhzZ14uV@{GeWLg7-Gf*SXFb?a+Avzi#>oOJioV_MdU_G%gqKju{F6FROzlMJ
z0O$PL1<y3;1++^*6_BGIdKphp-|f&sps-D17$ugiAMgJBf4`4l>h_G)aCsqPAecis
z5tslP8v4v&4YFbyJm_~BMZ{iyRQvO^)~<UXo1yr@UXLm(j-NyJpCiWg_lPZfuIDe-
zb8&UW#3w2J%rHw3NlZ}i&8_(Q^((r$(k~uNBc=ZYMCXpC<y<$p&!0X)7Y^-!xZM;}
zUv>N+#abJ-0%A7ydPw3E@Nbm&()|3%OJ5B2S219R$MB_^4gmoZT9zpCs&&Rmdbp!D
zR*K(Y9U!ff18)cJ)33QX!QYBo?jAlUaf_(RL6JyZ;v{5}aAJw?M?D@*^+?Os<~-Jt
z7j@xIDDEEKvhlVNA(ZlA?G=8%{0detea9LKZ&m|SFxkLxH4D6uHgb$Yk@p=_VjIjl
z=K%~m50~8!<1VFctb_Ln6{6ZcvTe)hKeKedmVNsCd5!Vr+@W1nEo(cMhxFK)jW*tt
z_TT81EvLS#YMreTH+y+E?M>A&+MAWsRNZ#m>N#f8QugXv$M?!<g??F5pO|~+7Pm5<
zmse6~*xx)inq_wB+Qo|jQBezHp|B=763)=;U=oI*LJA!@n+g(WnAl4ctdRMAW2Nz}
zs@P+UhC8raD%!#^XCW+%QWeqy6coLb|3-&j*KXXnk$o(B8h2b%qjV=STEzL`>`QFu
z$B*~+Ux1eY$p(<>y#b7gJq{}AovJ`sNDVMM-@HW>gTC){H6pyHm=c%zb6P=j+1d3y
z0j}e*63aS%czs9~9dr5g<x45H&cIJl^C2TBdgO?w<j0vAqEH7Hy_NgE78+ntd)LyU
z@yNvwn9GQMwAil_pFMx}Z&X7;!5pMyZe!B}Rzzp=hRh*mZHPmobEz}=WyK6=Zcdjk
z3%N|Q*kzZJ$zy2HM~Z-*yk}E4mEC`F0nTqD+1jRsgtWnmgT|u22f@+Xo*w<LeUpD~
z<4|kH7!fAnC@KuWE6B-ha1p~GrG%UIu=t&+n%-itg_((omBp(!P~b)d;zvW|FO;WI
z^#4#@{X9Q^21gc+0{{(ApFR7St-ijVYWsFpRgbk*aaEVMra$3$L4GF2x1#pfm8$vf
z&6~q$c|LY*0$rU@=}yo9w+)KzAy*br5j8KhAg*Tqz71tu6s1%bRu=hy)-?hv`1gl~
zQ1Sx{$Wh!^{=bs}{5PS$(b>PRdPDg?*dzXQ_OAi^_di~i`S<Dl`Th2u|3SF%hv>i7
z=HHioz4zBq{QK+Cf49=#U;p!!|5;dy{}bCM<f6qsic$fDU%zH%$YTVK0086qH&8@T
z+C3VOXhxiBGGN0h(<|~#*zTcZKxha@(*OI#OniL7T;NS3Pp1gPv^{P>dcMWP2mSl=
zpHuR`&lbdQ04b|wF$<kXCAq%8#t#TPK6}g!yUO=Vzx0-HKN(D8#qEQeDT>f4i1{S<
zRtp@EqJZLO5w6Ta?E`l1(8jzMFOGpuGA8p6MI10zbdf9yUWb?)UlU15DX?b;08w4p
zg?~7SA+t$qBS;HV#&T~3G!Iau5dAkb)|}BO80S)Hn-qr;mNn%d0uQJVh_5_IPEJlt
zG_wfX96A9VEMRyLD+h8)i2vnux3;#{)ukaa0h*0rdd+QZ>7S_4lDdE_6tBZw;B_dg
zVqhs0mk6fo)D#4SvwVC`4?b2)L$b)nH;wKS1S(+QaBy%yiJ6U!?b@|#j~+b&sc8i4
zR?r<fH0g|tJ|k;ISgxUn?1g9;MPJw~&Fy-GC@5RYO7cP3sK98iS^@YW$csNA5Cag6
zWU98Nrn#jhCpR|@4YLSEy}cDtKjkX%(~fTqfgm@~wjz8J$R_CX#g81bIxZC6pXld;
zlo!w^8@qIyd`*;%LCXPqL+VfuG73QU;De;buDf-Uq^wNva3P~Oax^hNzrCWOqOPtE
zUkY9m;J}mv|IplFdsoQA%R7mMAy`Ap^H*$bZLur}@R5*X`e`?uy{f7zg6BFOlQ3(V
z(>7WmmG9nJ;fW!>OMKLTl0V$v{rmSvIBq*CQ4F*K1MFN&Av#<alMoB1gi~$VqBdK9
zXBYHCo=b#c(xUFQE2VB2t{kZU8l}bDgh(;><p{1TFdq8G7I;cxyg8S8C8k3Wh14&+
zkj@6H9Q`^mYm&rh9X#qV?Fx|f2p+p`_~f-xAD>#pim)pdv>MDoq2b}+{ODPly+%DB
zJ4#7ODaCpfwSvJqenG+TQ8G3MLfz&)=k)ZZp6Ra7(Q9B2E-G?CT4&c6&TH)WdK7r;
z^sB9Q9o~$G1ccB71-`+2*+x9~hmRg*lBsRola7W(MS-E%ipKM)eIVNuW65g)%37}?
zt+i;xgAMvYMkmgR;${>hmU(9uep#&+9;Wue=GSbQuFs$rWpjf5G!me641Zb+J0Tzd
zR-c=?dXPx{CX%9CP)n{h|4HFe6QC|gEKUgs45rFBJ2`;>6Z-hEj=op>vO-WODe53L
zGHL@RjhDgS6g#?U<Hi%Mpd0Aq2y#juIdY`4FdHiwn}?}sQeiZVjH)X4dH}y$?Ax3c
z3nF8T(3QLH&j5LY4v}Il0lMqj^q}u_-9}P22yE!<AvDvLYw<sVASXe;P>F?lmc|bx
z{#GcUfn*J?+%!x5`}eT|c8~M&t^!#7P^S<c9E=n<oS2sYw0D^UefJo0%4yEJp%~E<
z_<DMe?hX$PoryZ-@6IA1;EZkWXr9CkQXC#!JYysRT%v*ENfQPYwy*7^6hoj0@*6h!
zmoks0I6f8@f`N`yLRYnX=ww-Ih%1U$0IiS_$Hv9s1;~KV=w_}PVk{-~2AKRHyB-b+
zG@6DK0VwT-%#@-|Zyq>Y;J*aWonV%bh?TSeGX(T~5FnATCp}K^ky{fDlR)3TLsq&R
z2nykkJ&;v}8SPCuW9@h=9X%Kj#=$hzIx?9%)*RgdNLIm<U~LkorcI^xXXelr6s`_n
z<Tc*)=HthY7@O#Bj{FEV$I)C{p9tvY_vYL81Ye2D2S-C}7#r`s-kG9=-BPJIrC7bX
zE6gc1SY+HtC1dRKk3a(eU27Sz+<))@;tT=p?OJz+@r2Q^(9U=o`29HAoltDh3iSXF
zrLge&>C-;9k^|`Z@1MABin|Z%*hK4Fg`NGJu?0ihpe}GwUmUHyh_OkW``FXS<c5zQ
zqpmcCDcabif*6k$c`XCZO^sPI&j9k^TY`gw@!4n2G+>nmBw`FdMXtSjjh*kO;|(pQ
zPg$$mANcwb<LRNQ;*5~q1@;8dG(^py%!KO`NAn?wuqifrFw-&@*C;3sJvrE$1~?WA
zF7CFdbzu<#E-=?(9;=2)NeSuyE+U{v$QN_<79suzR`ZMz!;X4iI7IQ)C9x4P^sVa>
zx8~o>59VcOPtDAP@)lwd(0Ef;$G(r96Fqsb!nERCfSsWKlg7T2Lsax8U>?i~p%od0
zfkCgUhI)|R;P>yvB_(L}NS$DX-I)E}c%VpCkdKcL-WyWn<L3v2Mtdw=3^-3@V4##_
zD^8J}o!x@tIf5$Bt)oyRrH-DQ^r08uaIA5i;I(BueE2qgAuJZOG#I7ukr!oL$BzB}
z{X5gTgVdF%xU-sO7Do?P!eN*{DTDqX0H&!`=QA=gn&9TJe1HyKShrNPx97tVz#_wi
zh=4>3Ii(bVt6<w=M{6yfTR-J59nybJ0EQ}3_7y$AT4<@jp=fKdY-+qjb2_!ScK3k;
z>0e9hkiJJDv0Ndv0^5HJbM8P!a(nVUAfo!Zx*<g<9y85V+hH#gPM@!C<HINXRy%}6
z3y|!B!P?l+fW9QqWr>;CE`1o!b8qKP-M3(V)b2~eXkb94rMDKC_xeiSn$L#Bl4Iix
zDl@pl)Tj3}K++<sEC99z0Q%bH7u0AAQ&NV|D1r}$6*7fu;|yJcse=wr;sqvji&qIz
z@7nby!XkkIA7OEjsh-X7)~#inUmDI`?6S7DoTDFBz-mIV8Xt*s1>Px(U<)TF=YGtP
zF{snq$SEGAg!$UXp(V+uk(iMo4j2L<f!peG5%l&k*9WH>Ohas!%pu-s^gpcjRZ(rM
zKi&IMFU{6}N4gxYSu6DIhYyP|6JUh)kv`!(;+PXcrHfX8JF%SyVk0+9y~g@S5tFNs
zF9yF`Szbnnf)f)R9PBvM%%rTWtZ)~`_Jrlp6eW02AFbX2b{?KXn2&V+d@Qq9eqrHc
zPu|fLC8V<mWoT<PGeTdS61`=WU1dF(khS?;**|w(FZikp91(oYw9ok)X3CwJ*teg7
z;l0P+uDyw&um{uJ_9xQLl1JC?r>6%eHz;TtEHq|K45L@*VCClG!r3@%{Dgc`FMI*_
zQkbU9PP>0UskAsVlaOXCI?KVy`2hxoc%%by5fMsdJ0QwV9@mXLcONoLAG^DczX!Q7
zwYx;S8=H^L0{wJsYn?89UAZ6zG`)Y%rux?1ctq~fMfHP+4xxi}s1^;>W8>qsp>vbH
zy3UjWFw)TH7hV`axFuHh3O%fN65^M?-VR&_;v~)3VR7NYo5k|l2JdJ+A|8v5K6vL+
zh++GnXpd=VDW~s?7uP_}4a|e@9+KQ(JJbTM>TOp@ihN039Mrm#W{bV6-eVD@gf+8}
z!fC)Elt1q<9W@Jod+5+11wOFXGq+oYDg8PiuaS1q5u<PdWBKhuJUx8XQ^A7df4>Nt
zuUDChMfu)D3O*2?EI?lXGzl+p^5jWqxPL*1+#^QGWAD0O!&qY{7r+S-tS6vza6Z7|
z^(W(XcuaHVK`<nw*<cQEWY7+E=FFLAd1`=s&$nDTxApuKnxsI#_S-K2&<#`kEEg9S
z;YUGrk3~-fOg{_lr(FunK|tcnP<(=tA<@2wMFDZx8&9eF?OQGO5PY!;t^qmJsZ*z5
z8oz)4PJ99epdsM#s?+p=YHM(8>{&@u05)j9+L#xEQ2MU$aY#b)7!~JZL&Boq3Lb-#
zjPM5qZI~0(Pm|}fOJJ07XV~Yph!>livti~DNrNH0tMttT7?!#j21IoNH8SNjc^+<V
zcuhP20yvz_8gxUvIsFyD)%e&L-Ddr+rq8gI@?|=-vjeqza`!`al2X|AQpG(=MAUws
zIJTihm!dR4kMt5lEyS`~#@z|`gMDTx9wf~igmrr8eh7X2r*~xTmQhx=L)pg96JS=@
z5ia>aU$d7bC3!E>yl~3SK{|k0`1pgWKE#-4U`>QH6U?oGf&xCTv$GRRxF3W}l)*~f
zSJJ*Jzd|E2H1APz@<iZ|fcAiZ?ZD_L*R8Gc0I@G$p2C9!GX;tSZV+sP12}i?BWxw6
zP7VUL_v4VWePfV>3D+Sf^xxLhSOax7Cv|>IsY`;*N1)_eHK3;G=C%SH7FNqK#=^#?
z$bH3?NtPS9G7Ox&#}Ra=1ZAaw{*3+$;T2+39iHz%{rFN28f-^?qQH`3@_iYF446#I
zktUR*Qr1MzGa)AdLj>IV+=2pulP3uS0{S#OcIlmsClDhMd_6eT4r~Z8`LVrl>hS6(
z=n}$K=h9Lhz~mFy0oY!oG_A(;R#rby3EXn9NZ-J~6?P24AA-u-)>aK;V@O;Zpilz=
zmLViARA68%c7{%;y5WR?DT^1Q*C;3_=L+(n@$qrgqw-Ln!Krz;_q4QjLTwNn4hlLb
zfUclTr<KbX!8E*?^H6hnQ`6)?ZOGQl7feA)5H^Bn73%tVNV}A877!)cQpX3+1%#2r
z=&$D(NFc^gV%nM6K}>*wgiV>2_RF<a1i?zPDvze7E-wb&!VxbLw4fKdeAvt=99D7G
zX(m*vqs47r$7bHz$?@_GC_zw3*RNlWTU$OVEF3vHNT(kb8agGXAto;FlAEBY2S@Mn
z<%oDi#f<<|!S+ox?)9hVSEzPo!Kr~X&Ap?l!>c^l9DG^9a(AeX1u0>fXK`;o1$`hW
zc7T?)Ds260Wej$W3=N$@T_yK>VtTqL!YnLhWEUDDMR6DX9Knc~J}Atw^X+#-Q0aFS
z(XB-+>(!IRN-V{fzpd-%UQkGk6Te9mANh<kaa75&Updg1$kc>Zl~5%Xlu!}YS~b4F
zQX^*m%KHsSKYEepeXVS5i=i#h*Jm2prK1nsGnKt&2%pSJIeO%bMn}dKggl%)-_T}L
zyYG<1Eg?J8&dD)kPud1DK62AP`xU4A^6{;&UKw&M@UtDkd{~V9c?oMb5EUl8--tOV
z@Z(a%lP6nKn?qA@?|yuczzScZiIlBGeu^!S)T<*ReE+^rRUa71L6yJhXJ%)KIdxz@
z|1_6h<wTkD408~MkWSm{4-XD%>*>{mIG}F?w^~!fD~mu?HCm|L!z3&nL5KMXK>xDO
z&0h!EquY{vY5P8eqb$^vqcIt{E)K3WC^)oTjTO4WTp%MD6WB6D?PCstncp7FSvj1|
z()+`D22ecbuK0|MMq}^ep<!V-65*U9y&h{S&!0|&y|8J!FDd9sGmO=nNluY&TSiVP
z6U*#CAlH~JWJ|Uxu<s$9vSAW2HziEt`|&TRWRNsnU0hlM+Rn|2Apv|FCz*;tS@`0C
zuR6t8%gviN=f%E%Ndg>#4lC0}gb+_Wp2$Gn19B-SP`i*!9UR<HJBU3KI-9DFs199x
z4%XH++AL;f{V1f=HYoa>d{qlp0?={u%=)vbMkT)yS-X7L1U<q5ee_}nDzRX0!@xuD
zk&%(QXp5vsl}A|OAdBpr921pSX!GF90YZ{SmAc^1?DH_{Ah_UucYXu?Hy<CPqW8zn
zh2l!15TtVaNgQdH>HK%RLH9XI(T)AZKUaa;;o(w#YJ1bwHHWs00oUKKc!;)a&pBxN
zAVERtTd(ANC~^WgAWHg+sb4oTirtt(v2<QC2r`*)7+&7q_1s`aoW=Gl(tv}*G^+RS
z-g&QOE+OVXlf3-owm|U6kVea#JrxRhU++$jpPzix1$^WjR~DVC%aIGzgdaG<3CHZg
zE_CB*cJ^horXWt1c_9FFWEicc`LFe!>PCXCgx;hgg}dl315hB&#pOGzpQ9&+>Jv;B
z!qLDrZ4@=22i%IVyTc*){TqW{fo8?>%t5iS40e0s+$GP@-U|;LeTm43+6T6%hMu2@
zP5mMQRh^t>kcXy7rfa8P13Z!``bdTVBh<5z_;r2>e<-J@i1QtJMXi+6OFa(7Aa2MS
zd~sZf!}D;S{DN?E0aUhS4s@wtc@QnX6Z!B!5j6~7U#h0ViNJSu(DU6J?Lq%JNFzi>
z@|$TGt@x<RNSIV?D9Cuqy-Zc_dR$aAs!aW<5Ke#`^X`QxE^439j%`3^dQm${l{}Yk
zEzS6qdov%6JJM;+{L%2m(_JmW5?E?M`u_f>85t`uQ|cbJW8_Y-1s-8ifE~LN)RT}X
zG$;gZ3RYsl!osdBJ9MGqxf8|2?^;!$MQo!xei){5wg(D@zyW@Zah0ON1+v@J&I#yI
z?uQL2;uGSkk}SXtz>i@2{v0!k4hiAE`P+W@E6;QfD(PvHl@Y_tjN30c5{ikiFL0~D
z!j}$<i^Gi!Vvo>%cHv`VBP5Oy&y$<5TUPPZ(I(p`Vq_L|P&>gh9ve9b8yTV2;{4Ae
zv&mUmlZY{BYXj{aF*I5`Xj{i76Q}v#@N*=Qhvm?O6cQ3*WtHdN>#;aP%$z`=)q0W_
z3W!h|Y9JHsV|bcDO6bXUg=i)s9&Br(EDxc)55~vX*epRa;k+rxseb;v1T>@4f(Th+
zh+nH0FJTg>I>=_Q8mK*03RK*c*5CIEM+qW)#Xtm5DF21VCn~D`zCOaBG(K)@Y@Gf5
z`*D#|H3TJBSLTcz9L6!fT2Vp45=-{|y9DqriRA(4MvH~S<GB;Z-sZ?TXsh7-5eZgp
zt!a)BG-B}Y_Fx(ePe4ARv9?X|8|aU<3bz~XbQnTzT6ZxHg)DNV|3*$;UeG|5AcHZb
zi4YD2-<J?z`<9GCttl|4Wg7{P)rMjPe>rB*3QUal2Ph-FhEsCgd&A_AU1UgzECV+O
zhlxX9dhvG7SN9Cxbu^PK$D>ioJc|4os|ml4F5ySOY~cC<Y+&L{KA^9LhBHjU;K|&)
zd9$q~5rs081l6A9BRCxxiU7RC6>Q~&sLK0_oXC2kM`UCRsZSqh;8SSFO<EzSC?qN>
zd1Fdyo&&HJ;!<+VE*+$!*qMBj=M15XfE60}2}A*=0Xs?e1|_oO(f<A;Vy~I+d6*QH
zRz1;7($YyfYMQQrIHC*PIap0@M9|>yv?d)uBsKo^>v1&cA-vqW?Houb*8s4z$00_-
zvV;h44QA|N0HZnO;ckrrsIMh5l)QYYHRI&EL9M2F*X&K$M?68eI@nEUWgwtJazBHO
zfS_2j%ovL9h?D&`t{3P9{^{xInH1n)Zx0=rHB>_}um<T1=vL4SUPgN&A_r>6Ju8Sa
zkPUG2@DL4SptN9(lR<I@c@Z&?TA3YETe7lvGeQcLXrZ65bAVlexb5|=f>}W(f&tYj
z`E>Anps+yZw696=r)mb;lNuYxp|BIQ+Jg(eefu_~Bw)0$$8oxmas`dh0V|{&R^&Px
z5_J$bH=>Imm!{YWq~M_O4EX<Z3?Gj_BG&w-`U1?CMK)U+m5Yw-U!;}JoLL4;80<S+
zcCt*JMZ+SbAE0xqP=ppX0$2MHOZij6P}EJYmN7|VWZT*MoHIYgSBa&yrN!hkKM|?4
z&j<oLsou1ktnVX-XS6q{7LX!bsaz7})~(EK>=x?#I0i4CAWg2vhP}%bsmJ{VfiXO9
zO>*dsa#reEWZuw|uG99Lo_2KY1`wNMzQ1}#711`@;X-QBx~1YQgjf$ix5{&Pwe1Cq
z-v0iLu<82B1?Z1Ui{UX!3ca)H!4(slyrpQLGm~d~adENCUIr7R_CY*MzsL$iGl{C|
z<mQB9fCfY(_2Th70_k&)jW7p3eQJWF;z%XTU`=?`e#`wNmSCLc5MFVh|9q^vXPzNu
z<K$s_y&!LcHq*%5moFpilu}8`>`zrVlZQq}3F#QS);J}WG>ni$LE7S25+9n`7WJ_k
zl4FB!s_*lWlgZ@fy|sH&RnSU0ta!bcuk-x^6wX0%pxcb%i`Ak}g_Ry~rC;QAwN2gK
zLYVB`pc&Mmi#V`m+u)Ys)2SxTM||X3^KCSzlYqUSeLk=?o@`J+N@8taCIvW886X}-
zM9}boDID#^W-IE6dwi>a1jok#@!2%2`H*`A1^<*QOf+{w$Yo9PMh0*)uRB3_;dFJ?
z(b?;!Lg$z)mxNboKjl%N&kzNz#&L65LIN`i2AqwU221*$7%FeXUiaCwy`w|>NmNFL
zXxMYU|Kb8TXlPLD=3+krB#4_m8ha%S38f+%HurO0!(E3CH7;_ZJ_zLVsH&%!kWj?p
zcE@YiEL9T8swzkefozicF}l$p8?A_nii$S-jf{+@;~uV_<272ZBEA3Pz&ti-U<a>0
z67|Q+h3c0LzM;ThD`V>z;%Vpc1g;>^2bQe<NrV-ZN>VdYiIFD<VGd{9BQrCP6_CQ1
zeT%c-OYF=25=vcLm_k)CbT~PUR>AnyjT5>TXF3Y-P>%9*#a`BPaynU1B?seQvtj=M
z1@N|#PjaEuweY)0r*GrF2P3*r3oRs*{eE&E@q7R4SD95417{B`S?z{)QXjh95g_$*
zw8BE9h64Aq6ia6^aT@+gQw*$si*QhZvmS!v?u>IR)EV$iBFZ=DRsgG$nK<q2MqzT@
zRu&5o<ziF_HZ0lhiJ*OV;47}mJqm~qmvI_=A0iv*!zAK2Prk!?%zQ%fTPIT2R9^lK
z*%yvCJd1>cJJMgWuM+8pTRP2X2OtHuwF4dtgZnxQ2|<jo$Cyd-c-FaAd)CEr8|0zV
z^8+2P-$RWL!RyQ1*BFTBCKnYJrcEMkoh37yI8Co=f{hh+nd#mFtQuXG1cW;yt**@4
zu-e=85doQq`i;2XxEwO0CR%nejv^ZU<9d4YV?F$?&u$ZOVnYa*ACX-lk2vkUgo71f
z$F-yS`Qe!Yp`pdON$9j?z@nn)26vk`F>h{ff9A^%O_9#@&$wU(3#bEk#Y?a6oa2M=
zbJ%0GUtHo$cGHfLf|nRn-}~{Frl)fi45i&QadNEx2@>j7;h#&}65BlZ;EGc&>@aUo
z0tbW0fop3`RCM(8lo}tvew|iQy1>#raJ(4xz>|WV|7gfGXCas4?_mr|;0<lyH$2Ve
zsu=7%6p$-XfLKC9x)35uQ#2e*OywwlK5+y@;Aa)apwQLLwz*5p6ImG6;y4Dq0F6ZA
zVA$yS=8crXeK8ZKn>UvTup6D*n8Ek*@}1JMbza-#tldLAXZ;DtA!rO?pJ5bz_M=Em
zW}v*ODYfdTU}T&VbnSzM)okLVR<JTUPAXn0`2hBgt+rRi5Q-{&lqHCf)i@A?RmX)r
zFk__L>K1as@t3Us_C(FXGbj(d@7OSq{Fdt2F1I>r;apD!g~GhN{{6p^Fd&Wq;h9av
z#@zfZ-R7CuQQeWw2XS$X9zjr5NIguokZOY|T9uWaS4^>^y`pPq6#xxd+CR`r*WnrI
zJ$HjHMuI;^qvV={;~8rur7fd}W>G=n({S`^0e!GmNf><#eDsOpC#>Ed<Ws@Puhshj
z?tKl%n{`=v%tLFLP5`V#P)-G!f%((Gc=XTR(IBs{4D&rW%0uk~ufBWNt|_3pNK2qC
zN+@QR?oLIeFfISy@;U#X2aaE$u%ks26REdPLs@wg`V2E09^PxIJ-6e-hYz+NKRzCj
z)EsiKJ0|94;3#4XXPUCdYtBah`aK8)Q57Y2A$_X}#Y@fQ`~>_64ieqnCs9%L#fcpK
zvR2zzA5YKDvi4HB6l$=HiEuq&2ulQBfvy#9Zf+2@h>Qh|^>%!_-@Uoa+PX{(=Dcah
z%kvf$VZFTV+qWmJ?ZBwPD26SX$Ng48XySj5a4$V@Pg&)0q?^JkM~^^a@6nl|5Pg`S
z@*6tXpv&ik7(}o6k&1w1Rp&i9sE}N(09u7k>A@jgAFItR@Q$`CA;^;J&z(Pwpp_`I
zFCd5VEE9Wp%N8-vV~M2X<aZ(mBQBy$()t#0n%^NBVuLoYDH0Gnn;_~Vrq94-!%1Y;
z1YJ2I2po;`n60e9AK|Qd%)!Ay;ef`ksDlK+f5`=K8lbG8%xA73!^R3&>TkZ!74%*3
zp;r3<=3TOi>mw%~{&kdwG7%gBN8>>_GtO@YTki8H{L>8%W9|;)Fo^XEcRVyTQuNCN
zen)O|VV_A#pe5I|*LHU5ETW<y_0#_lq5u~mVPQmxrx`W5=5nAx^qsQS#?DT&zY&X0
zAN2r2HflGw)*2pwwxypzVNB`SN9_cj_9WaY+(7$CxIefFIxMmP_7rY%K}or{w=7|S
z;PHtOk@B=V3`BZip<txm*RqR*X>Tb(aC1bHkBv@)%DNTMNyJDiA;43!a33+K=IqvC
z7)|6atN~?+FHm86{`@&$N2H)n=VIv0C*qTYI=#k_7eK7j*3K>+`u8AEBHH>DDwpI1
zR6yH2>JwyYI>-O1esoCL_!u!B+mK&lP|Dp46t<%0?>KeA<SSZDysJ$Dvos|;sGR`w
z7Ar#|B5I0{%m8Dosbxm37#WO2{{<HpLbUhZt4=g?(=A1H09o@`TQw&wdIv{GpILQ;
zvCUo;!Q8h9L#uvZ3o;LrA7Io(Dj?~Bp8zBY*%BZiz-!Rew6Dn^E-h07wbuR*83yrq
zYQZOdBL8J%WL!L311D<sn)WE;G*@(BU_7~@>;x(@GjZE;#Sd>-;m)LW<_an;l~`kQ
zlw6tFGrKoUtyfWQbd^v{-Kw(@0Ik-(+n9>ific~~X7Y;?ST(JCkIE(!zb2zaLtuD4
z*!3fvSBB9!HzSRf0nqz~Rbk5>jw~II4fd=)C4BMAtJ9g^?HwFcE|y?F7ea)=F-67R
zp1bRo%6-@a{)skG5s^Qem;CBgbm>QjFl)z#D8T^d><R$7QP6rvgAis?_V-s7_+}~1
zN)A2gW!I1o=|>a~7n;0!^3*8_sNXTmKB7yr4p~nmnmJ$GlB}`e1k24g6`3E)Uj(S=
z{ct?ZWDlhUrd1?81RDApa?J95ZlqUV#h~7ovRxT1Xf7Ic%k4;IWW+6R*=n6M)YT1l
zwm-6-_7`n8okkfZ1y=8ly~(gn1%<}l_db-m-Sl!_hn-SaoW9fUW%TIFH{w@OYpCsY
z@2tH(v(^O3qTKb(5G=!u<dtq+i6fzu>-VZC!3nMZ%m>;kg2pYA4xl9vks`BjN`9c&
zxD2vSLiq@>CVW2!2XtuMuv^UuV*QCnh(o+fS1s}G(2_m-Ms<aFc~yOQ%e}~|a&;*h
zAaKz$z6BsMBrO|RV^kx@9OUUKi$uMqAMrDUf|&xS$QlIh$PlWZq_jDkpZW&Q^)X5m
zc%(8Um0cERQqv|Ia8dzT8wc@lb*Lp--$;6=nFKLwJM<y6j@RK#qFX_#S_jZC;^;;p
zA=C_jrl@<MZg~wyi^R>pK&zwbra`L>4#Vnh_{jPo%9ZJ0PX{g|#03k=7S`5E1%c_<
zX)M%5B_+xPI7P^-JhV@$@15K2BL{?`b@SHFk6IGTZjAC01@ID(Q#j4I_Cs<lNS|eK
zGKN$lshwW?6G?y0CNE^N4WOtig;aKTM-))5IGhwl@ImLBs}q{$(JWA={%nR#5p-S5
z9T{p_111(Et4gCTA9Zj`Vr5}~!NMb|*>TiK22^V$X^dE?het;>e{JW=*S<cOo0IeO
zal`xCT8u5ZpOlmo74;FBD>5|omZvA`iV0D61p7b0bHy+UGaDQI3<H6THTlnvzONaD
zFQlsb7C(5PI*!Dqbu-=#RJdz}H)gGMmDeT+7zevbQ^8NSB-EWP51Pby>oFnG_3Js0
znlln&+Sv7W4MYjwa7~$9J_`JXssBN(wy`|(?$KMv7Mp`nh7O@bJt3>=PU&U}uaU<o
zXW^1i)F<({=&QG9aCEe^wu)e|(WjEKk>TN5!B47(hr6O@Q;`>u&%Z7Pgp2SN?wZgD
zLP@^X95Ly$OA*MTqfpXP_t;NIcTmzjx}cX?u+#lXLmYq7;n6)od}*PW`f^<72b6m@
zHr+O*whOCpZu1@GV47P;;P*{oStQDB+Xm7!hLlH1Vd3GLWol{t0|QRbYp;IcT#@>e
z)G~zxC*3?1=I{|ce>Ko|Rn;sM>QFWzmqo#RCamQXirt8S+v5OPAy`d!tUsf}F_Nm9
z#>~URv$&$Qps=pBNuyHtd4l<i1o`~?PDuHLxpmD@t_gT(3%1T{_$SE4(jm@9?FF%_
z{QW~N?CF!581oEN0gT{l!>yRp@f?H#t$nf&Km3+0FAaFGq?x#uOXOjez1|F(nl^Kr
zRdyhM>6}|?JSmKZqB;>9?ZX1?ar?ipYq*A(!KKdJH2d=9i&*+VZSuoY!n+i_FEMt-
zUSNsb&K6tjBU6wJD!|6lOg2bFj{`kO{TPg6&B9_1eJUjIC;0inApekUvVL@;iKUA7
z#`*JK0106GBEN7#FB9Mhs3zx5NHbN5h}~iSrI~pB4qe>-*TQ-Vnybd`hVlxr)BZ{t
z>!;h)N#0hFIQBD%Z9V(oO&i3mYCOf<s2%AAP-#akJ1B_hSApL3?4X{l`{qIVn04SZ
z0#JfxqB@bGSPj(C7In7MZ2VEZRSUH^u;Jj8?pvj3AuK^>os4_4EjyE)|I_4TK{SU|
zOIul4bu!7%>~=ts%FWImzY6G%INo7et!tGAf}DJM3PPP9&U0wjA%mvBWP+YevXQk8
zDOKa%t<elJIi<+ZsaGSc@7uTXyh~}Jh|@<`_l~4$oGfbXx4pBBlg`9Dj6AaD;|<6a
z&78GYQ`<GGr>aVHBRS6fC<@em=sFU@`S5&BP0o46vuAg7IAOmB-wd$pNQGt5N-CB_
z`3%jW@rrj~*r6RJh!_+og@L~Q!s24^PdOdZu>C}LPUZXe6Od`eo(1}IhC~XU5WtMA
zfA_+i8_0fMb8d2+6RYvH<sZCuyPZ>cgcF1ox;u8hXM>cUfxCeG$(EBc#`Vd;C!g3g
zY(gr(d~ruCMdBP@)BR&+)@xA+FL-G^EvI?{VMGku&Op3}b7%31gNKXjCZvLZ?xQI`
z8$NZb%GD5ATsk6Lj6@)jNE>1+j9DV%_BScAz1mjsi)=WcbUx{%OF@1<B2r~Z4ILeY
zW73(FddVJ~S9}B+mfaXw`A(dGWZzd{dYD)84ZXg}XAjBBR-#G11A~V!unPPva9E5G
zH`eR{B+;%(xE)kph~!oD+8E}?Tn31V1wn<JoRNWcW)wNU!{ngzBv9p=Kxp6-b-Au*
zr}uQmOOA?NTrBqYOS=HUJ@R^w8Wk$|9gG(bJIul1lm2GrxVU_h+3y8oc@Aw70N@m1
z4GmZtE%{$Y(uQrdWjUNl0GUAi1kMEzXRP_D=gfBimpPuEl|Zg!Wn~FWZSiXA3r=}6
zE^$#&(N2Xeb{!_BrbyWKObh9Tfq&VGv@%!1q;UYK+EZ%u#zZGKn1?0AE<SshkmY{(
zT`(Kk|8~wszOd~e6<rMr?K?2B^gu2Fh~6tyeGwx9=Jvae?1Nydf&PM8V42G=@H6PK
zgOCY(1$@`Zo;=dG+CDI`G&PV)1De@Ul^wu<!E%TJ(nv@MLKhuCU}ok;*kLRxF=7XP
zAAo+a=Nv%h!Psye_5B|SyWu$jpFnu^5Hk=7%==*E)G67{M%QS5=ln46_0zST6iPCd
zb{ZjLH2N!~An^&BpHQwH!}CP)#(VzNOGJwZ5Fj^~!YYKY6%TF>sUxua)XZ%#;EC6y
zbI@xsJ4?u;!T;M@SZKpFp-MjT<0<NeGA~*{DoW6yHkXgvS33_qN@?sk+lcz-n~+wU
zjJ<l7r>*RT;ynmIA*B*u5eaX6l%6*9a$Dy=yTLw5i{d(ZwK#K^?SvIbsF!iJ4fBrZ
z3#tVd1>JUxjEI;=tJ%Urjs$kL`gHt$($LV*Jh&d!(wTRDk>7P>YjPc;H|c6^H64>m
zVum;YSO9*pVYz9cnE6iBEG6U!yksDAJvme(DcdrETRVp|{GouWX<r~&<BW>Rp@Rpt
z3{%#g-U*8Z!VN1{D_)YO6PECHMY&8Ifp>DJ$4j6u5T@aD5^<9r02;)lK}|b*h_iie
zw|4nNz@0ngz<3DddTdk%hE;TTLYoa#R<+7#gd@!$Sl}1--0kbj@T#%ivN+Brz?w)Y
z5O!N2ufk|0;P)9T9%hV+AXlJT5lPQEo!SyL0oD>h<$81dsjanRO-7x7FrrIByx-V;
z!E3e$X9)g;e@9vo)sfa3h22jM+IqGOQP93ArecoK<9>cOV<7*~Vkp<@xfPkt=W2&=
zE>V6ZbcIp(!?^)_fEb>gsxOJ%aMbKfU2sf{3J>oWWEcSJ5Gj3WYs<q~MGTI<;Ag$2
zv_hA(z9%a%npNOAt7*x|tQ!q%;`EMYO@2F`UF*MR=OGc60l7~7hmdtJbWrPDlR9wV
zK*#oX7kHjVMMXVh4lTW#@d+Tp8Wf?cr2!2mWeFxcz#@liMz@;QN-=pf|FXmGa@Xj!
z(wNpDV-vjn+nxqmxr-;CkvXf{+e1$zvWknZzzxW}w^*s*n>gH>qr+jCmmGFi;1^;d
z-4qf%MrV$Y^q4@kxP?xj{&;+I;O?Lib)T>=f@(?Z*Clf<`2iG~^>FX6-rkuiZwRxT
zs*v@dfAgae(VBxt$O)yM4LbP`Is-E?)zQq2@|zo@y2W)5R3pG1K)=wQ{rjK9#=0ZD
zCQ8I{Bvn<Sgg<iri7n%khzK$0u3`(cR#7L(K1iV}R&sG;+&!19P_{kmzR5i8T{iDq
zr8#WxU|dc!V1E`aDpw>?ACN}~s;;FaceXs0T4e`XeK-#@GvAdg$Jxf<M19|u$@!m6
zA8!8a^D^A1<pZV$W;4Y|tXl!l$?2MbR!hx9NsuZk`Jb_Z%KCK^4WDEI8qG#d^lQRH
zM9oMeO+-B7^(y-Q5&_gHf=d}sUWtbS(>x%Ewty{$>uNeXA3GZm;;?#cd5_)>qcJhC
zD;SK<?QUEKsm;x5OozUsx(Xw-xU`i1+<H1fz;17k;}cL!U7vex-AzN|w6^M2Ese5J
znrgZT$~prEcVKYM9PA`qrtgU<t){26BdMcTq&q;HLZhD)I2iO1`;VUL$`-PQo}DL3
z7r^*J-htv%OTnefR#)+40X2EASLFy|un!S#t1MInD<gp;ym|k?wRK@B)LgHT&iaVe
z7&x0-S%rAkz*Kd7N@7`WHA2=%iv}+SiJ|J^OETFWhr6Ol9F1!N{drLPSifo8;Wjn5
z&5`wPDJhdv(}T*HyZp4&)pygL(0Vo*5*-a*AM>RXVPu4L2+L5gY~L~iI##N6e+g74
zz)U*L2ogik+URe*m&pZZmY|O8*8)S(nTa>~ly*w(m5EQnogE#>bo6ySWI|-)QOF6d
z3<gt-m>Nss*qQ41=h>Gpu4MfC;IVl&n>)*1hxZuCK`hg+zeq#7T0Mcn$h@))qd<Ht
zN8n6j35dzsGn_Ln;>*8cF-V5zcoVG4lp!<f2^oh>0Nr(qy+$t8?$(c7YRfHo6@}lI
zk!OW<Z^F#E?euqP5@7fT$`OAAuGyof&J@7liXJV~QlV+i4FrE;b{nBgzyM`;xB?In
zzCO9O2%<}HeFI0XKViV2dVh&>E;t;7vcxVe4KxNADtB?bzDGd$#m-!+*JqH+po+^J
zkaw}`2(=GFeYwX|=s%dIZ30btnGF6Qq7u5zaLcu&YQX$xq9cse&6A%vvp)=V%%UB&
z3*vVew6wN<36iPs+57^6h{xmd2?6I=C&7CAM5lJ{TnTQ{@m0PQP?;Ilu%z>)w}b%%
zXYyt#u^`|vc<t(pYk+5YQ(RzgW5cnqK=|zmN4BF!F%l`zO3R2us)J}r&GsbF2UC}t
zIeDHc^>c}%vf*go1>6z^F8a+_AJW<9Q2#Cw>7vA;<1ae6K{ae0bI^m*KxJsMzoBjA
z=i?iav_;fm{yJ55_RxG{15_!Ifb@CDFD&jb&50<hB`5-~UWpBx<xZNJm`IOTk}m)u
zxk)uLyo9Z<mMHtyG$$~OWl-u8<Mf={L~j}I<*9<Zar<X~?W3gyYbaR6#xf)kQNn;_
zKZ<rJ+M6Fl^%gvO>IJ(Fun0T9$<-w!uqgs`T{tI<8Rtdy3+TGX3W=XM5yVvvAQ03j
z#k0E@(TY8~bMSgMdzyjDj$Eq5<Ye#TB5wq(afODeL<&dDQj&n4^~R!DLbWM>^+1Tl
zp(VCicn&ZbruS9W)I4zIz9qU}zS)Y#ar|<E&|+tUm|YNx@Tcc*5F=!PD}{{qK&qAV
zff&X)RFk$4ITWS|&RoIu_3X@Y1T#e?vvh9io2<oPFIqm3NY!2GL#|L`!45A0!U#1T
zT|{W87V1ZgjFb2Q4EQM4x;y(@Ad~C>cm8oKH%s9TOx<<9d>JbIvMAD`YZUslXl?<Y
z13U~sC~&bjs~K<EN_Qbt1`YTkJ^<B7B*bCR1Htf^kdR-Xy`Xp_A|w>}?E=UGlapDY
zVSBukNF-vwDJqH(0zn`*hng)a5jg!IOQD*JXuGnjD?FO%1iF(ugIv>&gABoA`ehIZ
zGHUU3`}XPL<AZ~90Tp&GWJpLz%&ZLneW8rxtS{bi0XSdSZ9~>GoY*R;>RrhOQ4dYZ
zTMrupsSH_jzv=~1#=5Q;z3dDBxEyC@CKPfBpyd5~P;`hkCnPY4%B7#r{QmtSFE6Cu
z*=q@Bw8j&--URfUVM)-*uZy~phQ>6^AeI4&omCK-c>f;HR$h!NlQTw+xd=cgusy_?
zFTjFAs*GSA6>P-o+1c3$qo6`H1`OLT4}a0yXQ8osdlvpRgSm-k85o5wU%(HJ8UPGK
z!%yfXDc@givP^JDbS>Q~7NOPP^o)AylPE@@{L}vc6;%ME8(w!zY9lKlnCNbXIy%rg
zI2`GG0jZ_8_h#C?AbhOvQaR*0<(oW7jLy318vV)58>Z^A!+@Tl;VVdt4V&S(8XFrc
zplx7%+Kw`?>AH%K;mP~OjOeRhk`{MkY+0M=NKKdscvGs1YxC0iqC21o2ypLgHS)|e
z8$WTySe_CFAp6dbUY4O3sWLzv21$2AbpMBkBeCLv3<w4)^Re3b$`_@{g(Rk=SYw$G
z-6Ee6L0d6=88B${BS9$Z^5yJ+096!PYei23>NLEWz#aoUJoMwhu2wHJQZq(25rw}E
zWZx;JAeBc@9!YU9B6)2oI!VEjIhzb1En_77`E_qtM7_3V#k=T-i<OkMT}L#AFXRQA
zZudpB=H-4sn4SH>Y(dIWAVOyZX-D@URzKd!+wU5nHP_VE7P7Xz?j$;*f$jn&`uXhX
zhjYw9Ltu89i=~sP820Ru9qb<&!St3hv8$^qH>)rBW8A~>R{QSkJFyUf`3l~NIWbu%
z>BCzfNEk-O8wvJqb#rqdNw)NQF{*&rg0o<UARqflv$A##^j@*ZEz-{2>>#bu+{gr!
z-v0@8wmOut^O`FvWcVzwIxy}vZ$k;rQ|Ucs{~EBSeH#LukK34<GDij%P-5;QqXL0S
z)*8uVUT-WGRYC8#Uz2ClL9U0$9DrzEwMdzvJ{xiTX#y+!z+h^8%})vDM;?^l^t>zU
ziISlEoL3;d!Veh*ZEW7RrYCe=wGFv)rw|$ako0mkC=LnPW7h|iXxjK%Cu0oiC3m7c
zgZ%vc(Ke_0{q&15Y>a9H%g(c2Wo42_BAErn^m+yC9+2j<*85ng(e0d$Nz1$!oTfJH
zQ)TMNvyKVD!Hv3Fc}u#O%|$=>7c^_UoX^<FQ}Gs(#v+thfV>}ETn&Ls0_HM6T842!
zdZ^~(@8<`?my6@`4_W}rPy=RbBg8`dXlv0Yc$zJ-Fb7d<!`dUd?ZW>Ldv6}r<NEde
zx6K(cWhPT8MUg2}nMwmnqtc+rtTH5$+8gZ<5)CL)2~nXkg(8JgD3YP16b+Il4H|y0
z)AzoA&%gJd&+#00$9LGfKErjL*Lkk>Uhnl@YtiK=HqDIZbaWalCJUyt$+WIpF2`&B
zIOZ1F8UKq|Cdz#I%0IcJZ-L7Wl3Zv0W))%L9Ge>qAKn=1v~uN2RXJz>J0EsQUnyB_
zZ=bsB3bgppxf{1^IZJWNxpRSm-_Xw|E#mo%IXS-CC|t!i>*gl2w;pqJA>hfIas3%;
z%i5#AHQ$X=$lSb5Y{Hx2vWrjz<Gu$E$u=7v;Z#2|%{o+DmhS&+WQ~YOogWyH>TL7b
z-!<3n9EAA3zN0MuTMMuQe|lH7MSYm&Mu%<BdWP4I$}KOxYpS>)>CqB0sg$ry8am>9
z2avYrm@Kvg0W~^ve{{wqw;7Q^cuN`Hney8hU?+ej1^#6n4=qyVII@2p?EP&zVc@so
z;yJE$>(R*|azG_Ns^eK%#d%`%s)t8A9h6S5Y;e+9vcxiez3<JN&ahTgtqqltIp^mm
zJJ6fkqPk^UVRC2shitrm&eJoMBUrL$^l8NJ91r=uZXjw_c_g$Vfirb`SRF_T3LUo4
z44%{~EiQBmI(m80t&|cdIDH5>0fq?q?%%)vN77MO2452P)l-Q6LAFtv-Fx=*Q%CCq
zh{Ir_d2-?ZFg0k9>EDPw2jk6ULJ<C&v8TF@#}&S~UhWh=TKSKR(Qm~nqN$kF8I94_
z4=fT3p9pLal@>FGZqUo~H9YLM?%b)qwW!m~MvLtr?r+@3wg&y;IOZp~o?6&#?%W5E
zH7m`{BQI-5JySn&>5}}+yDK7|IS+=P4ozvHU|_|%<QjND(z5r5E6ph{Ev19S6?oqP
zbC*kDZ){uhBeU0LoO(3HQNRD?B)s=)xSti8N;lAKiaowDCOAz|bFRiUle50QVs0}U
zeA(71O!MHSjKCsJAmBuiq&#6l3u{Hzn_xJLw8+K?3nQb~WX#GJy$wcAFSnP&mq)`L
z>=kbbpqolW1T5`dxG3@xeZ5mve@nqMhPO!)tZm1i`T0qoz8a1psi>&1VU!Mgh=^QB
zn|3&r0y^j+s9zbd#@6-*>if6Dg*fHtVt8w254;2ccIT%@4-`fZ*kOcxNR*yZTZ=Ne
z#WS>7o!q35y0^8=tgz_6fC4b)6|gB%A?_#wrFsIP<`$kL71IPRfG1KbA8&42teHuR
zD&so)H`&=?$>;G<P(8F&@uA9<JmAZhqVSk+W7oYnH+K`8puAjBRv-N(@v_mTaK;M3
zO}TM<AcCb!mXNwdtN<*8{^E4f`SJFR8|D<Pxdz-K%j*>#UCEjbafv%{{no9-UDaW3
z6}F~tZs!)i%FSi?twP0F$IQ?+t>q>g{v2}4)_;2%tyic|X)OF$yZ4xu1`j?%Ya`&^
z$k1??^|G((xlCuipr29f>ca(;q#YQ3K68Klxan762J`Q1UHH|c$YW+(51AoDbZ*|d
z%jhJP(WCzv>hn-z=v;Mm>$Tyw$S!eFbL`c3{XCyxnB}|h+_ZI5UAJz%ar*R-FLfC1
z+7LOnsR@h5Ak7#Zcr5hik@t)&_?*x=Rd?~?bxeJ>ba~;gKGe#!o|+N!P_2K@o%=-M
z_s$j1^btpYR90FU8>fWWmK@V^Bx-0MtZ-<f)(Sn@6NHjtj@_!6OP4N<xH{$e;>4^{
zq=&_JDp6A$6|USdEe@V23Tp2z^6TEK*GX;JV)wWmb2Mfl7=qYSRi87I3B>Him^?qe
ze2lN`xN*PAbM$seA~L>vw@G9?t~r%R-RB69$m#F#FW30MP;qV~^4OT1)7qT)RcA-b
zAIJ;ODhv0I=-I26n9h;-qhZcaE5(^JiSk}_(6IB>enT13sZ)bbzo*yG`;Q-M*WDD_
zykgn6Lj8U~<BS;9pvYOMr_80%CLcIHyjgd#gXA+T!7vJdk1i=5di30h6UlK7Diso2
zlm}QyrsU_J9j-EAdG`7phZa#YM5R;5HdRz=boV~FvAxzu%Pjp;09j-A$lb>u3^O^R
zm0LXSDKfB5$Er_k%uf(yZ!6*qQ+a!A&~X3RLm_QiK*cRX+(RgYy*X`WvCPqB$_XDM
zjpUos&v!eb)z!d@*$LGtf77kiiDS@~Go6Teex{D>(cv>|1da1!?Y(>VuG-2?o3(iH
z)9cqKAuaPy)CD4fkQO-{n-5i2ft586Ulv8sOic0E`H}^^HyU;(TiE8mKU7pv@%!_m
zjxo!IY9=V<i(GiSLu3f8pnQCKe>p#9PtOw|0)kK3r-#@yj47D4{p*qA`I?BD;PP&>
zs?MKg=H}+I2-}5PC8nV#@(3aex7x2u=gphvIj}IoQPW^KRnEB#150jipV03RMKat^
zCA@pu?G9(h$XVX7ONAo5V=+;h;vx7#XodqfUii;bz4rGKSV>&lC#DxiMvm84B_5$0
zEKtAKa5aNRW>E@d8KawhG3;jX`(+IqEhw4!=SIjnj0KQ=K-$SrwuEZcytvrgqRS40
zgQ7ELnKKU~H+KeQn$GuD=iJ~#LtU~H4y5b@$=t#)0|#Ex>#Xte3{7<#6l84K(|W5n
zZP?Ijh-xwP{8=iwhhGc(>~a39nX4h-jfNbIarmOL5!)$R?NG#3-+P$s$__~_HQv~j
zGv3RcDI3ySw1}QhE7rdG_~y+H(rIJ~iwX=4(^FDpr%v_Tl(rUaCJ!NoyqC5l@o#xi
z(HL2|eCT(m@#oXdq@YK`d$!z6A%#U|<N37WTkH|o6;+ip+Oo0Nud7j3T7GN&m?1-l
zB4_@IW<ha;%I39ecLK?`_1D(LbJE6ad1!%4s>U=uz2{(b4-b!XzLQB`m|~#%H|k#=
zg@TpfY^<W<MuyQ8TkH0OhGm{l8j?@@7Y_}Tk*G6Bc^6_}RhNsjWBov?8*JuTm#SUh
z7Os6oar#XBz@@LM&z6p;T`Z9(N7ft9eJLcQfXJqb*~VB?6p%b^cIpvY6s$dEBHNE^
z7BiZ-LDX(1y3EIkljY52?snO}(a;5(^apnVgb<jGWDB7SqpF8~Xa0^>f$(a}s5oPY
z9mM<|C?pU*WcJ#%7jNCV1vQP6;VqNM#UP5lEN}fDq`|gAMFHp0gcb}K!s~f4gil>{
zwQu-fYT@t(q#c-dMUGW-!Ge9H6j(n{740(CZ!=dmMYtlW)tFaBMIsExFruDw=d6v5
zLDic#ZTd(lEPOC+p0!VQEcL_bb?(&Z)r@igRw;rezjr#P=u`o%452vc>0Ll6KRjtv
zP?A)y<3XQ2d-n3>jOo*1Q%w;W@g{WA8clEa8g+*$<lEprVAUudYzG#QpyNKsCUXXH
zniwC!x&#Fc^SBoWb`5^<wX$;QvO#@f_!xXEQv?}yyMzdr(4XWD5hPpH(^He1<bd-a
zl$?g`F!6hpoxM$nmQAC&NlA&b0^VT&oZs2ARwOe-v76FWZj<NMKoN@c9W}|BFD~;k
znmTL?JRU}-66}RY1m9l5h9qyIt)r8ckr73%8NW>uf%~G#ZN$0G^R(YweCr23shJUI
zvLhN$G=JjkdH$J?A1{2gxeVv8VwE7fz4b3)fT3>zPebtWLx)n*OS<FK>C@(hhIa55
z4i2AKdq@h*#499GhdK|W#pWg^@xWu=BaOz|j2Syth%8AJojhq2w|43F?X|pz%wJAw
zs^;34N1ZeBaf#C7t^eeBs@f@jW^ess`~9ai3NVFpaVtK4Gr1#xFzdDQn)Te#cO%0N
zoUtieE-DhSacp(b;EbJ2E{!fJBxJI~X7Mw7wwY<M#Teh=vvwTZ>O|76Tt&zKVnjsA
zmoHP`!TBNiyJ*oOF-!5n1sYoDdVHo9;5RM1DQ8OZ?zx|K4{SXGMnDcVg|Df?g)BuC
z$!L?KBQ>lWGrtNzK~9rvSLadlYB*HJ_<Fv%%0kfixLfz`3G{d*4+#f{zMcq`67W8b
zj*g>8-{R8SDd>R0;sFK3HyGnupwGorMflI;$6D|~?)Bv+1_lO9JHo-ia2K!4jo{hi
z@v{^4*;mvF(KRjprg^wlFYFC(znhRCih0(p@#)7rAHmp)LrcC3hRx8($l2K$BhQDa
z6IjDErya>SA>ICFhqE(!7uvhUW6|@Ot;vR>#Dk|ttWDVphIbPZ9sWbl#EX6pL*|{y
zE505dPc+a#EXD*&<_rBIm<RCc{C>kbP`3a|K+TZt<@bjzx|BW6D>1Q+EmEQ!M70Ih
z3SNz30Blqpza_F~r1D18C7c)w3+m#Y&}%R=bLr=6#SQNj!q#s)IhIjF_{F-hu)hle
zroELeA3C9T!O*3rW=F~lu!!k2wbS0^f#gEcm5<eokG<?IJyWiTN)JWFiB7xgO8b5>
z;)06{zMExEO*#Orh3~8D#}_jC%XH^m<7FQTLzA|G=f8j7e9p+&CHk<pH^u0%NyQ(x
z43NHoC3-Rl!u)1zk9rM0flV0sj3T`uj(Mr65esInS{~?l=x2%UyzOrCAw$m$9~uTO
zKhnIDev|%IWbvAoN@c2q9-d$xyJ%tCZn~jJJZw-JH%_a<Lv!)s+6&AiAHVvr=wM)M
zskLg1*#NHm(wT<r;!90(ZX_7V{kJ}Y0HgTd+Pq|W5FrMHZL=>nW>Q#j&nx##g|5aV
zmmqOx)u-rPFAPQ>QQ1uNydpO$YLQ9Ifiam^>x>U<?r5;*c^s{eR>jYv^3BA2o`i&N
zH!jV<(2!Z#bDMp+f-Acpn4i94d2(S#z02J>h(|Jqv8<44e6w#y(HO2E9Q@B<t6+Cr
zYaJP*$fz2(5O!m!2SU+SG1u8-uqodP;e-W^j!M3~)ZF|%MyzZxTGTbu=b5-hb0{Cb
z8@&aST$6dTYm@Y#L0(>7uYH7i^y@9>@(WFJ(k7m|`U7x&7^X!UML&MLxEa|~VqVFD
zli}HPCC~1<y|lbs6p>gSSw44aQd($}oUt!ffibDCeT2HsL-zQ^`Si8>JiDZ->g9@4
zjs)!fCa0<lNW_EPA~h(Fe>U;CQsCD>gGDL#xxS|>%8P?`u;5RgbVPu5qUi}&O5&#M
z>IMo$sL|4BZ!t1&+P?%?KveT41fW4~`f5teaW;iXShkTTzT)+5y{w!Zf{oR30bM2j
z?s{gp|FNtKFlX)@cg3rdqW<e*J!4~4K?u#hv;#7Z4lNxcTInUOB)$sL@KA4khRku&
zD61_T6Lw>y2AXLwx5=TB(Hq<)`Bsv+^IB(88b&f-cY=>A<)x@v4!euyG%Q1jhF-5+
za)?(k0d-?}4igm=41Qdd7~~pqiwrNPu>Ac1OEtD7K<mm+lI_Hl**;6oOx5}Pqe4Q$
zzr)gF>6-2HU7Vao1t!GoI8#vks-@qRWNaeOq--&AM%p#6U{m;r-fI#Pir*pOk1V_9
zMxA5WKb7aIIz6a<gZJbOrgpX~xM(se?9dB7%<zlDW2aB_aV2eje@{b9*t*DX7QXkS
z_kDPkbRBOfirM7`{%ew>7`oQf^$PmJu~)YBiNWjiJG5v($zKwE#MKFcoB?oUtH~gr
zKi~fJ)7~!H%S-lj{9&uLN>*uk-xR0GtLF8URnYZ2ziL!Bg|GKFJ^Vh@Y)w^|-HN!x
z(y*e~aeX=u(45z|Z>mH$g?)c_?>SQ{V^XX8sRvseUCi{S4B2~5-w73Ld)<MUwwy?#
z_G?r&+pjddpI_k@?q<58A6<gQXzpuKR}~Z|&yw)_C#k))s`}+R6|-~?Pr~8AFBhI|
zs=qgb&NH#fp%NuY$u;eZdrPF7@`!tEt-k5!-)?lQ#%A>V34dCKN=W>Zq2<+qk#osC
z6Z2odP6%ut=*)<O>p|}c1Iz;-D%V<dROq}`Or)U8?kK)h{B-K~zrPUcasEiG^4|qP
zVATTWTq}DkTI&Avx=y|5flBpL5)zVRryum$b0};4+W(;0o#tk=(`^ZDTGc^9dhCDo
z`uiWO>iKvlHOsftuCH_z4`_n9;ad}ld6BM65-r$M^`JBWIUe)%?nagWCGojo9gG57
z`e2JFW)HO7rZePiM~^;M2|L6XfmfzH!g*JIBvi!Qt4f@mU=JF^gs5C~rRQg>80d;C
z7~2k(t3b{1{JPAjHOs|elR!Q5r}sLWCgnd<s`_aE?=Qr{{O@{*Rgm}}YzKw#C?;4b
zs2~=7P1Ghkg8H37ueZHfzS!h)A2pRwctw1I^0^{DPGPJa&6?WP=#T%n@F$rAqXPT1
z6LLDYthZ-Q(~PEP=HIbv*_wa1YwPLF{{219ds95{2{bt{cRla-8^A_?=qOyB<w9ps
zNB_Vzb=pw=Vz0@*{TZ~78S@9n9#aSiJ-pRU40f>>`}wEMVYxPrQVp{-t?NYNX@%{7
zE<_bMHSGHsDN5>nDBYCSn=8I$qWBh%4TKG_;ofQo2Z*+f?lZ&@xp8=~XoCc3{gG<I
zWEKf~J-(n1{qNj~?mR!W9+~oj#{3hWaP%Y;7K(3K0dy{dW)F3_9mYyMtU3go81>JB
z#f6$jK4Uo)9K{DqO1s(y;6Df1&mYWN(}#&8-NDm+$>sli`#SNWLj%V2fwitit!C3N
zm6sWq{`s}YTq6d#7yq1WKiN&)%c8}8`T_ApH$c~jNXqWTd;jN?Jj7f6**Sio#`o)w
zkEs<MJfXzghQC!u)6cEj8z^Hsn?$_ZEymDW{JBG6Bl;-7qf_u1S+|6uBN9tGxp<Fx
zW&QCPvL}lYlu0l<;BR#X{hwcNXEx!b6Fz(HD2TiE7Z-Dy)$TC1hG$AV=C|z6YO9Jn
z{j0m%&Qj&Mhv<v`cfL0-Jukt~hSI)^hx=aG|DWrADz3Y;UOztyt(pjXR5I<v8ZQ-V
zypK3?5{l=-i|_}Gk`MQ*U}ew0IS3f){+54dQ_AP9(ZuTgT`)s@p>Dq?CD=$&zf|_;
znj_P+W4DlY)}LE%9O>Ud!r<H%rpPhZ9d0}8D4k{QGuZn16<G;|{(;u6Wuv`@|1sEQ
zIs?O1%wvl~y<`+`(lK`DY#|Jv;b<+qLc|py6W7;f@c%n!aTEOf4-VG<TH-$|^^a<L
zf`r7oEyg_c3n!Ot{CKM}Yw^f@LO;yF_q*TDaqf_Aq78qHFd?z&YmiD}qaDN;*jur}
z&c%#g@9S}ryoL~T=d1Z$M>swFi;|!yVqK;)i;g|iwxb)(;O`){k#0Zd!}FNu{yD56
z?8WXoGbTT6Dbx2TXD%#ryoV&+<h>{dKOiX8pH+6jK!MbVU{SvDSDN38ikA(!341<l
zy7qvcD}v2JO-h}|T-y-RQ!s<S&+nI-RpMu~AoLZK%gEiuz6`un1B5@UCQWJ?q3eGj
zXd4N5T99qqMqzWVIA3Twl+37k@eu<?7;V~?V`fuQTr6Fp$Dv#0BCpl;?=`;|3&B8n
z*amsdL#h?i%$Zy)7AVi9U@y_8&Gl9D51BUp{@LDE&|U$5r(BuM0YMhi9=!bp1xhx{
zt&3rm`5|Nz1RwR7Yb%J>DCwHb7<Ss14ruPfv-$ou>Zp{&Kg;IUzafUKLpfgGbFKyC
zlxSq~-K1~j<?v$&=xW+qUbg$aYGjD`5axpLMH;LW3+Vq!^zB!8$&onz&OXrU*ViD&
zw>0yL>Jch|oMu;jj@cAd-Cr{Ic<o-MR5avm*Ds8T<UQAagH`8sK&s%1jZrfGsdaY`
zhHR|zz3;In)OC5PmA4Dpu6eXxy)bqUuTLLPCLxhBi&vb(63U)DCZ0fPxX&-_Cbz?t
zG0j&O#^pcs%5eyi%6B)yr(SJseUCweT)6h&*9UlZX3w4nzr)W}W`&F2pZq=z(lUGe
zHFQXj7j9+4*)@M}8$En8;sJZB5fXm4x(`sK*mX+U?=N^KV^~7jvD8vG1gvzRF<E!7
z$G|eBo9Ql_#8eiw;npB`a($3fBi>ZTK3DG1oJgaqRE<$n?Jz8o4dsLc?p{0`a*xBe
zJsZso`CqGj=H6n}Wi>h$IyIiz6f?dJ+q1Jj!Zz(TElKHAM3FMB!*bbpu`%7c1@X|@
zBOr*Lj&QL|UW$1eWVIt}TCQarw3O^viaItTH|okU2=`GxbMoZL^!em7cRcDm2qt(N
zyP2vRI4k<3{-Qdbf#!KL3f#5}vAoQl&9;fBBnV7?*M_vW@U`?wr5nZ`$REO&rl%Th
zS?oo-IqAwxxMJnI#V7bz^`8$razLcYe-)q9SLY#1&LDSInbh9qsXU65iIUPL%rXu-
zu_j$fTuHujE^~sr){mUNc?J5n?t9gE#5?76#_VLINn^+6B5##_pN7CuTYUFIY@SN-
zAV;ak{kpka1b#=l(d{|gY+PPz>L8n*o(s_43`ayq=)V70O_fKx7?M~0xQ$ylQWXq|
z`g!aM6l14;&RciZ*|gC5#_D4;$6aoOYwS{K@_BRJfn#%Oo*`P6E1mktk%Kr&LTAx@
zV<#N4VqTzx#J$;ZS)q$gAC)RO{=W3TAj+xL$8_#&;I5byyl%1#h4!r&j;8N%FUhK&
zgrQ#5{^i(ghYYU#>9L8E?n+UZa2DX7?^c`1)%Gmk{?>dQ{meGJoZE0@g9k&Ii)J%{
zBa%&+)_WFvwDAdzkJ-5w4YqZFkL&fp`1{ptkJ`)`v{V!Xet(Bbnysn-{%&-CF#|@=
z59d5RbUd+SooLS|3Vti*VO9kf%Ww=O<99b3ikM(i|6C<WSP_@{gBZVNTA|S~*nHJX
z&Sz`=HIJhkf-y7DzF%gCY=1FR=U*|HjtejPh^cCl4!n2B42JpqQu6lW^G@lwVXA!8
zo7X2V70BtF5mO}|IxGTfByE5ZN&bbL04@04!M*Dku0|802Ts3#uqn7G?~qC7EFpc!
z1_juCyLv5gmixK-xR>g~3gVzGbOO3*03G2=i$3<3@(;ru%WoFkHflkeQrnYvQ2y-y
zde?=18?wi`0b=fs!EhF&L@gLxjoBdma^udiZBK@*v)q;@r#YmMhc*6?=ZCQ*Jz4I&
zQ%?q2?;5SX6ux>G`mk4Ol4d&?xSXEs*`tTmN6c8j^;SgeQeKZeWu_m(U?bI-Qg~6=
zyOh2H?wMileg*;8<Tkf6c-_0A4<70co@3vwV={A_Q~K+p5`_O;i>dfmYQ&kdXU|^$
z#^;(IjIc^B{MlcB&epoD#p~9t6>bX;!(wepK1W{U$Ody1p~;;y`>!ntHMq=vAs>6t
zQU6Yc>){00Q3M5zzwjfU@LHZSx8@mLT%z;N8afFhADFjQTw8B(Z6n1NvvApiVM=xO
zBXiItb9WNKC3#N4?*fGoBfms2sN&ItEwJjRUKC5mAT!IYsWCEp5r~U_f_LD;AMksD
z<6t=S-c2bCH|;;}GtOx4k2_y@ILbkZ__SQ>JzH`Qw`qWyJeaQ1-yxy!-(jK<Ah}o2
zVNyR`w}@GVX(I%7tC`3CXkU)CJ}-QlVodlBs%ve3Lu;a_=v#nfCZ5DoRkp!=RA+Ok
zWIFZ4%x14M6l(R^1{MpKcIpWxe<Fq{al4z{Ir_d61T>!GxOP8d6Owo1smyOW?=o&k
zsqmDKM%%8@PnP5Z4E9Yo;aeB=W)$eW)1cmm9-ZcDsY-XJUgYMTWBrm==qAd}7Z1nR
zDW^(oxg-Aj=U~~Yj}|)ZVhc$A&V_$UtYwD%paN0FJlUsTFVa+nGRmTBsW=E#4;qdY
zkZ`)-t(<VV&*~p10Lv%T#r)rr08mKVYLc>&f%=l33-(mrG}}5t&i2mm?w%Ce4UOku
zw!}hFH2%Wdbei~8#*uWe7>1p^biPq^{}*vG2ztn7uO13*)_=-0^}>B+v=vuKQLOFQ
zRmzRtcKxl>#EiEMm(B(hH!t3*bL#WAkFoZ8IwME0?6Gp<gtP#<B|rM$Xzhk1;V&iQ
z$~bP^zhc@k&h&rwl63D8Q%mNwmr8!@zwA*P2jP?Z@5;tsrOsB0rcv+6YBj2R?z&To
zi{k{`#AT#7o|$C3fDFU_!z$J~%4UA0XTEdGeqF)MjhW;|`rQ)nNbniWjiU43JL}I!
zzMhw=u88TRQ{(pr!Hjg*v_L1@fY-}rPps*OrI_p9)`nwg!zM+z`>~2_39`bilCN9H
zN;H(`h_<GYy;n#BhxFf_gU>xu@fFK$pnfRpug^0uksFf5LI&^s^<HzN2W*Pu#~yA=
z><zrm|MjJFv|0b|o`?kvhszcS?v4_Jr&Bqu=XQTR-PC@}qiDT?v@6@!$fkXb&)i>w
znXn2A7vfZb1e4KrcdnLsgjQm$IGoydHV`&P81dcziByw@P`UEz;-;9p<GzkR<9VDT
zG)b4@|0pc!i4hpE_5+ppN@Y78&+FKASkWi9zg_w!3f#G~mbY)rc0^j-EC-n2u?8oN
zA1+nk^te~9m5WvKx|o3D!yN96F1(@|YETNvH1Q@)Y$Ozv{(GJ2-*?vOHF3>0D5ycL
zeGn8bd*@zsNWM#D>`V3E!*_lDr@9ifz>%Nms^j@HZ~M#Cm$n*Z7fy-}`#f1)f5p_z
z&E;3G?UQjD8V=wkVMNmjy_SnSBQiz#!tPwX>qIf=_S6ApvUz?%LBBbZ{%a?T2QEoi
z))LrNX4k0ouQP!Z0FDc4%YY6eT}wkTpQ-U4KOc3K{Ee$EnZMAA|CP+`N_EOFl3#{V
zmv&7wk6l9qI{7HFGB)*arq#N&<L_qq+xAxb{5JwkdCU3sDEdKAbcG4*v3yJDG{;Xu
zz;%s|V#(3%&ACZ)Y=}TV`}{F8_9WTo*Is0(oVeGyji2|Yjr?woy$o_l*ubzqVOU!1
zM$LczH$%S)xsjuej^uiK&B#0M6c}r@T()q`L}r1NNXiHId3^`WXj~*L15UaX^nbV@
z9waI91zj8DF0pokspWDlU>6RI0P@pp^q1jPM)LlxYW!xJ^<SS|#xdg{xra*d<keba
zA^PvU^XtPVA@&0BEDKiKifn}RGUf(D@(Ems1br`$50E?7-Q?s!;keP55NGUaxd^<w
zPB|eXyi)jHkq7PAd(KY)hzuRqlCy3(7Kk`wQ1kN%3BU+a22j`X30t`&?<&#-r#g)E
zq}k!q8J~8AgBPoV5QOhaAAhOS#D9$}Y2Lk&fQyCVZbO`sCF!&Fkx1x7T@oR*UV<!7
zJKfvqx@7-xe$LJk2GhIrmI)3F3@o-%)x_s`k8FWNQD}a~a0TrZyD?<_e+kF1Kk=P>
zh?{DmvZVaKi-$3qOEH2R-^GW6S_p`gTQ*gDoR|=lh~O~HXQXd&YNx+7_VEqUUxq35
z8N2qyQX$7k7&B@!^pf^x&q3CWuS43T+odz@s%gY3Z`oJQS!}E=xARRc>!mFeZBj8k
zB4O1%)$fc)`|pBpXI=ip^pGBX2|nyN+V4)UJ7@lRqpYXc&eF{>*W~m9f_Hw{z+s*E
z=|p#1_6XJ0OFvyYIH+g3KPhctkt&I>gpNO#Su+3msPk<0I9xt4hDG5{kP2=5O_P_>
zGwsqj*K(fay{r8{iAvP(H=eeU81qz$d?%-$CKRPOI76k2i!$jGj6ZbA+qzh7xN?Ps
zIM*heYwfjE!GT)eZhbmJl=Vu>|Lyg-aLAOkmOHXE5wchl3mV+!!i15=kh>Cc8A-lo
zE=CTYjy)CqYVZB*>~~%4b%4hFl_H&hx$#>&H`M_>3a}h8zNo_@*Khu|gl{(J@}G#k
z7x8oj2j@Xuw%%+`2x-CaW3#T{d<pK}ai9=KGQ%n!JrC*(F5or}D;0|x$GLd3_y;Nm
zoHFr+Te775aOXG6iC;dkjo4nWIkyo45VE>sOh)dl%K|d#)@_;)i~?>DAEWQlY?K_0
z)!IBh*QgstYv!LHw1Yew?xVt#o@`=vqW8ePG$!E5+mmZH&lPacTvu<~>Kcf0G2HJ=
z4tFF`l1GWevY91*u`wOoZ-WM=bCRBUu){?wd)26}-OT??-jk3x(dU1YC@k3jApP-w
zUD_#KA<uzB6m2lSQ*ViTC)pZs0AmyB3-$p9cYLcW@N<HSisPewokdc}uZQQ6ED>l3
z-q>+GgPv!>rzZaXaj)^0(zW|FG53Bb)^CFg*aCjw#7wAAlsIoWDO`+&m5}HF(h&bY
z>Nv<w{QST03j;0lK@Wcv!)yqCc44UC+UoEr0nRf$)ChIIq1PGH9t{ys>+3`!o&Z~&
zdD}$8QO3N6uY5sb9JNWH$y|eGjwtrxmVv?!B3&!z920x17b7W-bU^F_|2|^FOhyk2
zT&OxhQsEULV6%@$%h-exI0yCe%i-x)WF$s~3_ff1lZfu?(>qkNs8m?AIVA#5NUqIb
zdTe$Ly%Ws;ZQB#e1ICXm#Vy0B5f5!ap#_We3yef%Lp<$o#)f}yx1^H;{n$k-3(<E*
zaEagFEytR`yoljBz#>Fo6DvA+5VxIa__W7S5Pzo8bR2`^Azq#ZfS|)|7J0qwmeiTV
z258RJn@hvBpW^^tlS~w*{>?^(mpFe7elMaA0vGWanVHde5-#c(UI)DkeP*D^E(~V;
z%oeOLfR2}A_}iRhJX)%Ws3s<q0K}B9+9^l^%}+xXac>}9aixrVMzUP>cSosG3@KNe
zo{(U$-bL-fp+*u56ni())kt)XhRE?(H2J`S5orr%j_}_2vVrUEt(d<iP{*;hht9Z+
z#UXm&*RbJd@1&H%7s9A62o*Jt5QZz?mdqP+k5vz!QU}Hn1cu}uSfNJznEsuzL-Oa;
z>Rv6}FEtql0ld|?Sp8{F6*IW<V{)G>uRxbo(Dv&#eESB(jMXcvpnJJ4RGy5%7)@-g
z+*$*JD(e1@46X*SEY{y}#1y$I@Olp<7LG=t<b$0X@v;l75oJ(e9yVbjkYf*ld-g)9
zxj6q!YOQpnHsOe)<A?Y%c#Z+_Am)7MY#n*cwtFlXK55F@fcGO`k+G_}^W~U$Cmv67
zHm&VkfPUhV8MH7yw@XrQ$lV8|VZ&AmR*2++{P&h8!yM;8*j85(!+5Y<<Z%b@@PmkA
z;!f&R#xW#XTCHrg5`2mm!8HNkft8N;tDnFSuV;1wg4>BD$z#&n%a$hMs@3VxfY4LW
zxCSBIdGDc9OG*5G%&86Luc9R|KEFDMzqkN$<@2r*XXRRMUs_jH#qqP=Tb~p6V%Da>
zV^R{*aepvar^=gx_?tb@``MBQVhdjIF>&BJqWEG0B#nxwxb%boC$-@}2r=4bl)&mL
zveQ-+c?^N^Y&riz%z}R9=m2g#G2EWP2G#et_+fHCJ68t@_)TLMe~nz&DyE4oD>G|k
zF!0okZxsBy(~SC8y3#d2ch4q-X@W*JF3)ISTm8sx9O>R@`)Wx%%;xl_L_uTxNFr9G
z)7PD!HXHkPveLeXW<oFeUx8g*E24D_?+&Kq`5kRSBmHOCCLj3KzK&&x*qLYd2HK;g
zSwQ007`fq6URRAHPqathXG=a$2+#-s%a8why9_|st-OUp)TLGG7}J!}mW$)xXvG0B
z>tNHT-S_O-y(Ru;7aQf{If%oiZ&qj>^ISX;aN7t-v+;?iI4blO%~C$<0aWaAbUIs9
zb~ea4)zV6>`1nG4k(tM<+@5qI&vOnJn0>8lo0?tQEN2i_lv0OenaLJ0nN&@s^N=ns
zNwe_{)`*(@<IdQNC4$gRQ3IjVKf`sa@uKQGRf)d^ciLjIzehYEhNmRr896JHcxZ%D
zSX66t0g7kEfKa#%g2aFwCe;H~tn)Cbw?DZm)r;G@t#?+?S?7oR=$t?t5#Anz5GFUN
z?WWVh${$!D9{!$!y$zb<<cYm*lHbG}H4bB6!dVHoXNFaVK<xj%gdm?@wXNhC^)4mD
z9W80n7~kd)Fav_&*|l~Q2}%_f3R4J8Cc8WkPXqG9bMQuXHVvsJTue~7LhaC$7iwRi
zUi5CXlh~nf{|-jEU$mlDlg5Lz28KhQ+mWq9z`OKDJ9oSP$9diLXupJ!4b|<D1DQ|i
z;*i5>4UaZGl?TRmvLybt8?EJ4^MkBT3zjB)C04_^*mv<jhxo?38~pq+xV5)FI(&?<
zB2d}>VZ)^om0c#?$qif*NA~`p+1l3DpJ0dQz!;yns`v?pOkb_=0Ai|uywS=a+5;|Y
zE|q3Xd6@?w4O__(tH-4cyc?6hgmVINT@&$@B&YC8aG1N<Im#%_`D|#oRzf&okr>SW
z3#S#1|7e=aU!QYl^c}qWeWO=?hM?Y)WcQ{ZzcfnfA^U*~Xm4vs8f0W8okPIdj%w(8
zwlcPR77T>$#o$hV%mcz)B8Eu0h8MATG%AFN_u^iSwFFmBp6K5Huk)J^Wrf}!#?YQ6
zdats-MK)qGT{b*XmRo~)6)b@p+U$Cm^Jbw+B3QefzjGyq+-{aw#Tj`{3nSkQPT==n
zG+vutF(xC{{F*H>84xU3J`vTH30|Uy@|{(I$qi!?KJV+{4!&;ug->W}dg&np83M9?
zdURL=+Z4|GE@5$gPBrBYO@et>7$80t$Z*Vo${3x!%7s-=piP)T7Sdit(u0D|Irs}=
zLhmZ`&;_0XXwlqEI7j#AsN7=`%CBT!9k6rV`}gnl)tlBVS-pIj428J3gu`J2G$%K*
zmHMe$gcBw5=VGCjA((~1(QucH<!Dr}mz6!cyLi72GEj)e^g~;aVM}qG_4U+hNXuhf
zz<suzNC#2Kr9ZneenHeH_Z$<mR+dt8d<`%0HqW+&@SA_Nu%kr8=(K$*Z?#mmqr`pT
z(O1|MVTEo@cQt3s{RUL<b&fWqt9ZO8zn@sMrOr)+=gAQRqK{gifx#TnzHtT`TzvPj
zAAe$w*yO&@8}|~Xt{W{L$z+W^Fuf$H$%7pJaf`gwfd(zoG_e<yG8$fbv`uL%rG%^L
z`M)U*FVoU?e>pnC;@0LS%-luQHjakD=lwO^Er-n=E~Otded5|{Y(-{>{iqpaIo1Bo
z^k!bsXKsIyQCw<&4Tf<WG4H|e)7)DcPsB`dH3=jBO1#k9tl!WX4#76<5C7?iMRj`j
zq13*RqjRKbga9!otNz?ppw&+E<PqYo#OY}?yT=ooT;(EJFeGT|h7*B-X4wM*+Fu8`
z4YG8a?{*f%(Lnik|6iD$R*r&5?m~x5_pfdzf|hId>^>lVx@%&i+-KePL*{R_ykhOP
z>`?Gh?e4a}ppsDVD>+!ke!ZZ^l3^(1jNU-88ZDbif0eyvUSdcR)#9zfI1u{-Z3Iz^
zqGNB~)~Ei9-bX;2>C=G;q(^^N8~Sw-8)5^+*h)YUH0q$igHsiXdYofG0HFc#G)hrN
zm?{z(1gQ?72%KCVuWgjRHcyBzxEY9YY9~Li#C%~CB#_9sD)*}4G$#=>^cXmS(%1h0
zy%&OngG~`NaAp<b!&)wl(z+cCk>cyU#B&PRiws1lLR@wc1(Bj}#W{8*sW|cRqyYpr
z<ewH37Q;Nf0`!bI77`RRo5Y5254dCq=XGF7k%@*(Jmsl+Vg>6g$Q4?xk*nE_Pois%
zkf8yb3!qZL!57q*Z9SBcc62A3NJ~X>M0ZcA+25}}C*}H2SDAwjJ-r?acY@AEFFqNp
zFaLF&igfq0SR;{{a1d1}{OwCfWfNA@@5RJoFpQ6*8>y>(?#bk8L~|*W$;J*cFEcYo
zz}?l25R}U&&Zf=5SyY=D0kr|xnih!SnqcTdSRv2BTU_cs@9(YkV|i(I#B-eZsdQ`d
z2ua|Nu^N>gzdsg?8MNbhwB7Duie5Os={R0ch@iG`;o8jL!+cB4t$PKq4c=9B=$>o9
z%~g-~LE72VK|u6&;~NPL_Wb&Ox_SItjF>tgzQvBriw!Ay7I5VDjLB>jf$d%D^`57t
ziAGo~IqC8M(jPDo{L5FOp=$tz^8HAbzL4Z>;+g}Lt<JzaW0?Tw;0DqEz_-V+h;izg
zj5TuU`&8t|uHE~Qu;BiEoffW`>-Y^gG6iSxcw0tZ5rGG(N;<~~oLkh$Ee)pfg;IQ6
zizy(N+}$;B1E1!Ts1}$96)Peak*3Cg0$}My!~BC6P%kc|*TqN+SbfbXC$k5<v-+rb
zLQRKX%={LNQEK#7Nj&Iad4E^6aL{xi3Hx(kv67h9M!lsh)LHTeLa^f$NJwl+%qDb-
zir(=PY!X|@WP7*V#!v}ncG^bc(YoAZ1HqY0e$CsJjx+^#4{V$Dmqh9~ofPY~b>aBu
zOu#3KqgI~kU@0%IZfEAST-iC{bHUQ!&RSV7?;(mxN)_y<C8r9{aXdJ!`l`m{eQLZE
zovw&<yI(<V{Oe`j4gf26_dV?Sn%)+y_^(4`xfPp67h%RXflMWvU$9;!C-jt?IF9=h
z<69h>;<CJJ_Q96n^Q|EUb&8(DUX^3Om#QX@zZBfB)sUQp5D^?3iPEAn%v)J^tKFNG
zjt>Ux=)mE?cuYa&VEkmf^5PNB#!Gq$4)J7_AqivFOh9HTH`hK*&eRIw^bT1U+}Go`
zQ~X%;SlsV3e4pwv8ZNAMzyFf;Z&|AHpw2Mk(ls|snYR6cYMwFUfsXt3-IX(m<T2?r
zrJIrlP5qX2Z^NQ}*)z`az$L18<g5yDXfhsoJV3R?Zye-)iPzZ!t91r7J?c6#Wn{w}
zBmIdn)vML-iS@$<9n`Q>jLUjl=#&wcOk@T8@V@#++COJ=9me7Z;g|1Y?3_~lUQ{Z>
zAN+g{%{;K+8&JyIy}y1DruuM;@1dr?nI8Iw62IqSe;Zvs;V)Iz28Jc|+DrR2-|9<C
zg92Lg)W9t@dp<!-W>w<Eb=~(A@4bt|t*&Qm%c!QtYNBfq6h1SouNH0-mXM^seBRu(
z)0_F&eKDS5qd)8)gb}|Es^a;G{7>3$i(+Y|8p9;C0=}=K*Ml&=!E7H-pE#D?<Uw|O
z|G*<x3__EsK;?{@epXAhC(Nxjn?|ASX}4<~)`fd@xcF$G-SlScJl!*lKAW86vgl+x
zGffR>bF^L3<MTgVNyMUtad=Swwp3$wUy>H9YHVwdj+p)HK>uahhSr&92Dlcy`fn}3
zGywY#A}VP$<E=W47fJx5a^L3=A2hyz1aZ(AWZtIln7q~TM^lZ*w!;oQclonB4C|hO
zzCKQQIj6rzit*ZTwjwYy_4|sAGg?Zf3>%St_<gB?ROP41t~;*WLu6Sje~(kGS96yF
zI+FjzC)bJi3YkThUIWa(Uwar7$TFuNuvwi^D|_-?j=ueP^$CewiKNDifnRzsrRPBL
zqfLn<%TNz>>ea6`%?&L=+P5r@eqowMFaut+UJjRxB?_Cv;hVo-5Z%7Lx8cilw%sw7
zwfu4Sp?-2?N@?h#tt~N6N~w_7ZX7*P8-fB$H*~6n+c<_eymnG^6Q_9OFMycTP-Yo7
z5Kicq$WN5+PVr76RB@WkM?1ft9SoC|R?PRQXkq{A#wQaVjBPs)*}UxCOk3mXn&&Ek
z>pcsd)C%q16D^Goc%;!(e6m#G3d{@rnP0;sjPyv6UR}rboW52F8vT<e{>n;M)bdGu
z@Ze#pT#=^6+Dh!s(W+N?z?SIFF*64~$k?0lps8A^r!3*=<wmQ+jnk)M__c3$P_foh
zH#$ptFRN0rD&wFF&wt!V4Izgzyx&m_2bElW(A}T5k0C!}*QeNejR=ta=RrxpO2fzt
zT+n=vu3`E&KJEH)bg|5C*(L+7X=jb>ULF=fnDX@tMCu~FA-PzvFJG>W9F$57BfY)u
zd~L$|KkJ!$W9Kw2Az)BmdN8rE$oq>*%eUXSlrO(+T+1S``fAlF{ZAe-$Ksb(*iNn)
z>2lg`H|#zNm7O)Yx9(s{XH<7C4w;au*Ek`)y(y%9eA%DiRc*Kf;ZMeVLgH9(xZMyp
zIFqcgv@uncm%pSixtT*^w^OuJyy16d%;Z@@e(Sd|DMZss`OfEkE&l6p*q`97po=tk
zl0-YW`%5)(&!ACzK}0}+b^i2hH%IX6%NgJ4*+ahMF6s+M$P&I-<^IoPV9ObeH`r5w
zg96*E5;UuUPK(CbPzJ^&MRVf14tJVIY)z@i9XfQ7#?wDF32V@coz?y6#IJqA52`e>
zh74$ib%NWcsxl7?v9}>#3=ZT~XRuX4j)^nbss;+uQeFbQ;~V0kYIF{pgM7qg5fL07
zp!AD|&LC61nMPunB*`XG3W}xnTo4L^q)2xQVdwm~6Byq~1P(18>(;EP1H4n@OPG}Y
z>{%lRPDCpnK^v)EV?d4#0vM2uyqhl#8&0-Yg=yGTvtB+MD6pJEP|<zz8>FKk0g^A}
z+Rp*H$~@x*1^sEt|B<y7#<SjvMARp`I_a~yq3yJ{268`5yifPjHxAMdfCwz1-3Qj-
z#jJ`bqT4B{tGVngurZ*R`*=B{ya)xdde(kV3>50Pu1jq*lltt8Jas@rOxW3-v3;6%
zJ_xYiYePN>e8KF|y;w_e`h|ssAp&YZU|qx10C<E%OmQq)<gI)~R)d2ETaXoK<{jOw
z#5b@0gVV_4rWEX``Q87)RJWFX^E#J9A<6C+VW)R=RFTmH$Au-YiKE6=XbUALgcDsx
zmhx;Hko@3q+;JEJ_5hbqaDl8&xrsrObla>YtY!u2IN8pAu>=#b;YEotAM@P9pTwif
z>!mZGl#C=7(^7a#n}*cP;^P8(A?JzY$kKBH3OA9kn@eJh${gwLfRHwhRR}ayjmdnL
zyp8YucR?ulEEqpR1aJiu%hvkDi4qVu=_UF>HtDdsW$`Q!08=X_Ml`7h&2YBqVc$kf
zPustCXs%+LvF11IGzkL)3EikMAMF&nA>Fnl@5C+^3R!87IsP^WZ}x&W7qlpWMX6w?
zHqZsbGp3Rd5I0oMw;L9euwj`2@`?R<yr{b(rdEv`U7ddH1}g|J28*z}x_5%*&5b5(
ze`Je(cDva*Mm!dQ=^K-NraA<E3haN2gnR?`<6!$0@8vKp;w(@pvU=!j=W@T(r|<Hp
z>DRgkC`14F0>0XMPh}1}l<h$s|LC+zwJfoKf~Kk|AnFk2TeBsofQ2|Ew_eDMsBk$L
zq4L+?VY>ZpRB6R!)$acQ5jWetCJQW0C^{sv1r6&`b>p@}%K+3xdqm&==BOBsy`-07
z^X5A7RL_Aci_?7na6%{hOD6m?E0l_WuRf+iO?J;5b5#S3p;mQNVe_!m#(e<51hO{>
z#d=tjt|p035o*)^sdOJMoN(n?r`s)67W|Di0R3ODBbk#`v5W9#YGGlR#%S?b?D>OL
zwP=jtm1ftBAdkJpEu_@s`Tn<0bn*w7S@x~H8kl||_ds0MN=6IVr5rp3!FzwQ>r<;U
zN6)2av>p%v1=szg_u%$wwdM^6(&~TY&vt*998oAnJ*>SvApF3f!hXxjcSASz4`@zm
z|D7afVO!kyX_iAm*~P|l^tV~E!uHjhH$|~nrN&j4LHIyP$pmAhf$l^Fmty~>H_5eU
z(pg~{T7*kSTg<(8VR;YNDSi)&J#*BZM6uSfX&6ka{J9##GKK7}(dB^M>7(lGD!NV<
z?0CJ@S4!xlTV_{b_Qu?k&i!a`BSVWC(hcsOzH++kd1;$=<%vrfojgb<tT%Db`Pqj>
zU&oaWcOx*7lDD<2BU(;+35M1^SOn)aUCz)Xwsp+nd+Z6yo%_dc%o7+1GWeO&Od(2C
z&teB8ZK7)FP|}|7fd6B8Yfds+H9q?VW>aQJC5(oThYKBI_Q2JR@rZ;79IT;TcW>vb
zj3TRT?`*wNTlQU2Et^)jxlZu5HO~k5wih2c_-0+WBU=4eFHh+--p}K~7-{U@RT}DL
z*CmgQ^5<C9l;>U>S=%l1#B~VM#)XWP6w_)GYpIv$7jVYOD$33n4A##*`9aymAc1qw
zr^#XMl8}?vUK%2IgQIXt$(9J$vh5;^?h`V(Pj^o`7aP18G{$~y_(Jo^AhOT`og*^c
zJ%3@m&ZjE#*FH@W+rjqWJ_YG!(oLtYucYpugx8pSA=@grR65ypk?fNXNHb3M_CNfr
z(`BWYJ7adgJpVey&g~KQZpCNAE?F*e@<*(~93Q^S$mnU&!<m8^UuS15u}r(|Y9e*t
z+0`PhmtB`e@0YE`dXQSbi)KqZ9+B|~*`;`UPuJ8^Jh;#9Csv4*>JsKfJ}sT!P-uG9
zwDR9!4xo}5j8&S%Ha`DQdF|W+H=VVoaetK|?JeJ6>KNc?jR$L#g4&o|@Rp)D`QQt$
z!37Lr+)Fe|+Qbbv(HrC1s7@rMlXzs$4eebs7Gg^55^y4&mpa`83W4DcH0<Y&)Y`r_
zpSK@~HpMDzfICW~>aHXNw<X(!<y>CAnorcxqJ(3|j`ogHYNJjZ8z86E+v{<4k99Ue
zjJbF8PEGUFQd)=j3|A_2PVb<pAHtjMGv3-A%BrR9ifVQtw2HpE#nuPC2Iw7;>3ir#
zzi~eIWSf6^6J^O(1^?`~p)tuOg7z0(CpJ{0PS8x4OD0kB_k~y~=QWb&IShib-+a`3
zhKSg)(K>1`APEMRy(MqgD8<@3TzITG$ja}+g;Gr!oP<o=&0VYOF!xVX><6aRoalBq
z>sw~fpzV+NSRbtEx;0PVtgbuYs6>+PVaMJr1?{Z`!m8pr1-pt@>9_MYsh`zdmJAO6
z_WaqiJ6KU06n?x&sH2vujS&9QXUL?|Tl?{AK7%LQp)IF*Zu@hawi8bICk_N|Icx?+
zl{_;GJ<p;gFWI<VGtVt+dlwp})P8A(_+`s{qnW!!8bTzfWygxu%xVn}ku&fcJh!la
zmY!|?{cisU;0<oYVJ%U$fEBL2$2oQHEQK+=C%aHHBv!uVZwrUjb<ZX|6B;6?vC?09
zoC_Q4Ge^}U_z#b|sZPE&sij0k=gFbAPQMAxQ#K#9R!w{;AFj2wLDtgx7OU0xOsz2N
z;pBaGbxvw?gh=xjjPa=0)#a02lV55RDIM8uU4ZNCAM?q81!_zhJ%6%m#?T+aGT!Ow
zk})m%!0nE=84x>e-+Lq-$=BrkQ$i26Ra33q;#x&E4C>%p149FiT1H6r?QGp!DU1pQ
z**OlVh{smUuFAVxi@}ky7+-Pyv6NC)cf@Nlbw$$a;WwWp$<O)(&XS%xaw7B3X8lsM
zJ0jCQ%-3n>#<cNQ;aA*C!FxR>FR1JJv~g7UoRm^c$fetpn?)o%{yRd<33;Qku4N}t
z#e3cXkLzujt(@Mk5L9(|%|#&Zf{eGP2JaQ6QvV*#lRgy?z+A%QQ_*0m^a7@Iw-&!P
zrs`u%fnVdL&jUHcNfw$4(z+$eSI-ZIr?uHt=sCGE>-AE@BUun)L+?hYh^Msr?FAa8
z>`tD2co~YWQ<L}G)d{8RN4YP#-n~5+h%D}n)r&9aV||jnGOc&Mx6cafMGT8iFZyPQ
zHGJBBT>2h~L8ZiG;mBbTeAP)DZ$LGbY&Om>x&HbXMYa;>)#g=bk2in-l)3W53DR#&
zo*Q1Za-!}L87Z&ukhrr|lwE9^e1BkMU#ueko#~f-vSDYB>PwATW8z$4rfy+giWgcD
zy6D^z*=pa-E`bC8Ro?EqO5}7jM@A2}`6(;JHZ30HEe7c%Pz|7|@wf7)&VwxbG#QMT
zyLUnX$(D2NKBDVx!It9&YqkXA?^}rqC$D5NhTYw&UbrUW4Sc~mkzpO@p}O)Pl$V3!
zy4UtlxjzZZZ#8rcxDp8|@Q#Q!3Q<zlL{SGYbK9Dhe!(JWn7YAG@o70b1wCJ%{B1}N
z<=x&{)u~<-xJ_-V-4ZWWm`w28wcAWl<B`D_67pHl4v~Il(ju%+o>zzl1bXfl9>@b~
znYqmk{A2H-p$bXnk)jAvbb&2sqwEC`pD3!vxF{kR6_R_w!NIDLPW+4Kry1Yc<6OS|
zwcMoj%hRgshq@mevv1yoFxf-B75DA?8qnjSxkb#~3%43h(TYIMNN<YLbJzH%Z(J?)
zUdv}1XXH6OlFQtyl4msE^e@l9zBW9aQ}D6Uq37eCQ_6qW=Q`o)WY%{1aW`bok8gvI
z0v}DRhtPGdCoOR-)Hh_y>b>n9f4oMYDTL2&ZOq$H^6i^m+_#m10cHtiyVYH1kN>ve
z*#;fQ@%|n^4QL>;VW9L7nhRHNJqr20Vy`rf3AXm_+xloBZAW+I+W4!DntFOk>}VTG
zN5F^~*Wdy#CnZGD#k|p5Y5Ptpjm(D66#XqOjnKQ?rZ@iEXIpBCVh<P#vui%+Zym-?
zD8I5Z1u7yLOCQ=>O3HL;!;MY3$|ZNSb4Xc@U1=9jdNIAv?)mfo*)g$!dB3b`dim5X
zAL<}5k1ze%TT-$Km|;JSnZXe*W#89;57WgcXi7<$_#_BbvM5kOugw=Gf4a2x+sIFT
z-+FH27f4f-o)htmexSN->sG|3hu&@<V7hBkR|*%@LVwF1`9X`Eq(v*&ufN=A@$Fc`
zX|)I1-n+8=>0@%@gq``w5&CXBYw%72bYq7$IUC*D`*`Ap#nbleSC})$(sfY;Yz(07
zU5mx6+spbKU<W{WoUswS>>Fhrqhm}-TXO3rL~CW6Ps#h4CXwpNk-<j|$8})_UvCY+
z0NP&=8iy`9N=KUUk1x_-ALiH0l*U&T^a6{`OP<vH%y+E6XY9wCwb{3o#*f$8y_-Rg
z-$>nQ-(8KJ^=nurI)B`(@sHlMXO9P*@RcAkyAN1V@>gdD<-NAcl}pZ0CzL3gYUrD_
zeh%eeBsEhzJDa?}=fyxCi=due9+L>7L{mi6aHm()XEIVnRIEPlPTAVR4ab#e!UC~>
z!^Eree1j(+jN7xTreL5>j5Ck@;iH!j=&97p^g>PWzrp6zckwvv-RIJ6v3AMst1HFm
zh_(fRMSE79^WDFy`g7=;YHPBUDc54Y8_{+!aCgLx&?Ec3;DMiaS;ayPSS*#DV;{BJ
z=g~qZr=Oto(G%$h9JXw*v48pQByvDko<CHF63-scIb+GVle;Pw&_CoUJcgpu)WJQk
zye!f#P|lG-Q?a7{kZ7T#c=Fq3faaSY{zjITdAPf431v_Zq$*L(^hr8><@?@^JKGA6
zzWDs;`SVTKMnSSOXqH^3aqWG2^(~zql(I&3E(BWky+vSprdB=Z^}A)26!p4J3rZ|z
z<n8X%rHkUzSi|f`V!ZaUr8FDe(A~_09-KuLD6%(3+ke|kVo)TnPCb1qh(#RFFXA7N
zy_kD;<?*5s)7>7^+CxZ_*xW1zb`bhbckykcr!$f!4Lu4&vOMO+x^b<G;umRYoe4RU
zw5>#!?u5JKJm74KitK7DFw)ExN@=k;BK^!-`%7H&<dkiOXOFDbDc9Od`zsu*Xyvn_
zhV&Ed#-BcYy7Jw=jfUO~@r=H<$jz;Vau|EhY`1$BJ-h7RxOsu|UkV$H7dWRrXG`DO
zl)J2cUit<fE`!#PS-Kv!N;J2e$pmuB7NbAr{~H$^4+pm%vkGOeSP`{w%AN*;Z@PO(
zt7dn7!V~{MQ{4?0SI1$pwF}zWoP_?GEm=>Wu7g@?UY-<dcu|(NC2_XDU!>*a<b3=%
zn`>-Yy>sWS^2#B5v+rD)x;NmY!pjZMP7mq%uTWpz|9)lnlTCb^DAxq(iJ-ZDzd>f(
zXgMjFy2VoaM&+RghIE+8I<u`g7H303RC<j#oVC4G+bA-GvztTr8utB|Zpq96i3qyz
zSF@-%%7~1fwo7~I(g78?*}!Jg6`9NSdtqV_d)Ul6>H^6M-|<@aVf2MryI!Rqo7ioE
zrDgE8L3&+!)6o0Mx{(!Kdh^_orA3$j<EM9hz1+gMXk-QDSmc1b<pw5y?e8i-^3(cW
zGRV5<y|irWfV=|-4sZ)}C(;dZnYt3v0R*2ilAOu*vME|L$I9N}JH7h9_FRL`;0JGd
zY%9|?^Emde+#$6Kxkr~ZuD=^V3tA5F1AxHS$P@x-f)>;*zN5NvpBrffOQlOz+&cYe
z&GLxNHUEnKf+$x&3^#rjX43yCdXlTT#!aT{RX;s)r00zms=}@W?w(~8dw-jGq_S7U
zl>JY(;UoMqU53rx8TFzzNj2RowsU^hCnht@Yb<v%Go*ZV&6WPkBl^pl3`^q7PUse<
zU!NO)7hQv=msdN*dCa}OWqOuW(;Nb>*x;ajqb{pY-Lf5*32&y4PSAPZAOwRn=`maT
z>9$Mnn=uC%1hYoR%*xV|+<$-j!E@#0N?69ClpNTc#myHyRrQxQjSne)l|I2Eq`J1Y
z|Kn0ljLN3Nhv@ZO*1l1`XL#VZsl~;>l<jO2D7Ysrt7)+h;{(qt*KGj?>%*-rf}3S)
zzdzv^Te83F+0YN7Ov7g+gL2lk76zYg+wdV-KXbTN;L(N)Hrmnlx1C+>lO`pc?b4gR
zj=*~Q*GH|`hUFltdsgFQ>)A_&;Psk?L;mOa!B)1C+m$!x40y5S+2hBnfeSDhK}_g)
zVKaZtnl+eFKbseahs;r{{i(I-rzl%EG<k<msEUp}Gh1zD4bkj#fX#R4n&5nC1soLe
ze&oAH)StZgs;*ORw=h7q==j?kE?>KTbOn1ccRcU5qs{x9i#Bd72lgUQ8`!U(DguC-
z;ypKBQC1TkChF_f?e4JXtoOqjw~Z0YcE7ewn|43P_;IghpGOu7^ryPDC<iJU$<eyP
zA6IO8pY@qnpPX~TW>LlHI>Sp&ABuZe+PXZW2EdJqJRa{fM26GDHRjB5!=jN3F*-X)
z<=(rv5K&A3@knBp&d0~|)4O-;wyP{65dB<Cf_n6WJ(uf?k57<`F$YQW#_$P&<DU4t
zZ_mrmAEa#gZrS|$8I+39Pk!2_E80z&x9zNdJ<vYmT}Fn>KEIXxyj`EsQ>Pkh^uD<;
z@9F-owC;nNc0~=7uGxcGH1uFkhQ{^s^wbGldozRs{xl_J^N0Af<l1ZQjdzxWocT_r
zao66|EF3S<LLT)p<x&^D`cXY)4j(?eBkala;vYZMA)wjwf{l&+_#h@G1}rQD2uzl_
z;I`yC@<^XKHQJr)44jGzkG}mpzg|*1r|S^=O=ET1XEJN7o>Yjbl#HKFai{srA&pp3
zi1H<;thVFqT~toK&y$bK+Vbo~;Wf30Wv_NEdb6NJXVS)-FyPTEZj7OZ7GNxxaZ;+-
zhLj_{5XB2}mn0Za%=E>QGqb2P$6<_)`BGd_@ezDQXXj2`yHd6mPFlFg@(OX;dbt&T
ziZ3sX+dJ1bxl`G~GZBX4e#D2eGpIh**_rQfb$m8;!;j#jBdR^AzBsshPkrRFg4d_3
zFP$lRdt$>Wi`u)oXTRB>oOy0%ukFu_uUYS<QXdk5!~6@v7j3_yS<LW${ferp9AicW
zHOO{b0N4u)4_AuvqF*6y0_~?Zy|)NIs2}LnWg!zPPW4W2vGhLj;&t=%zJmt~y4SwX
z;w2?f6U@|o&kyJ^W$01d%)no=GiI1YMU4zKc(Kzs#?{5f6`G7pzN7ou;NYkvzuVbB
znTVzSm*jWOs{Pck-wVD~Fd^vL;EVIQI>;n3jpSD`zrtvL1j^wQzUO-3fM{iAneo^Z
zOUpXbs*=`~&5e!5EKAQEq{aCs2B@XiMW(5%mibV{^`NopL%-=B8JfEFHtBE&wC75H
zAIqR*YAHBm;{q5@jRYkpIe*BUcb1ryuP@Fjbz8tGBuXGWBu3+_M${gC0!oA@Sb9ru
z^X91X+U`4xf*%iyp7fyn^Zc9F&!2y|e(IGMUG??!;@o~Zfni~d&e$+q+u+jI|JpWv
zv;yzZn9kNVw5^j|*@x2f$+y)$YSQ7i=wYfY115xeMN}pAIY6=pIoWnT1*!A2mnmX$
z8qY4Nv8c!F5n}vD(dDgV1yIaorN6aXHg@F|UgvMIxaAEE&m)F(i3m(YvHMmx#HzYy
zmxZ9@328?f6a9_2JL_GAoj2V7LKWxNZ*TP?W;tm<K%4H(S-NvX>(%LPkP&nj$gR<u
zkhWv<_D`QarB_@gHKI5&-Qj&~#6nZ^x}QHU+ax7^_uU<yb;mBXyLl#0D9|JNz9F6Q
zydRnSzMW%}Q*(IteJk0sEe;oFEpBTY+nU=9hRQfQ`}UG4>9k)OlW2N&Qt8hrj=PsF
zOfLAAHmmk?AsH4|<ZGGZrP!F$i5qv6Wy9>Ax$TaO7wPE_3TMwAQ)d`SZP4mTgWFoG
zJd$8<EdI$t_>pm<i(cy+0V_ZFW9s*NaVnH9mCruB@kkp_0(Yw5@r4b28a)u4Z0ZRN
z&WGgUehsCcZ6WaO&m>kJUuO>hjyUowX2p#KcViC8Jum#Q->8=(edhPNxm{?OmTBBo
z@`@*;HwK(Ginw!UMerV3*Lht;Kr-w^!m}Yqp8&VYbNi0ec5!w#S@wB={8pu>>r?|K
z_!<AjL*WMI0J<pYF|kzK2(K7nowibxD|{WqQ(k#})jy^^t@<3G#kk&3Mn7F|+y8Og
zNzuyYYi_KL0`2RHrumfAE>`@v#@wt#b+Ks>(~`db^U5&k##(yArzfNJQ1EJ8OX}gI
z=h1hE%eO@WWQ^5UUscr5D4?m_zVNFSOUAuMRPT3E|M{%bxiN^Fx7$w@^}e$_KHNIN
zB;E<eF+<q-%a{9>Cs~>w>#)ZtCh;e#3)3}Shcgapx}suw%PbCdeRb!F^wl1uFS6jM
zmO^5WJ$`w=WM(f(4=NGbm5RT9-Pjrzy$S&ZPQ8Hi1A^EI0}qdXv*pG5scmovUvoLe
zC_Mt5#|>JkwyTkZWW5r7kG0%ZY5s~#O-VU;?d-;%23VJpIH_Ah=YWtX&|hd_NIO~E
zCxdfp6U)2J?{k0(B=~yscL$>{OUa;Xo-kylL0y3Se`^8aH|CV%Z6;5iJYZ(-%EhFi
z2nkh`c5L5HN!y);&Edrl3y(fpRf=#k*3$XsD61taWR)WgC)5AcZ;*ZOXP+q>S+{Vo
zRVpT6$1S?D9$NnS@@0jR^s=H(5gdy0co_AIZf^~>g3p;w_Bn_tk{UYoSqtq<KbrOr
z@lP>-Nu+1^b&Iq=BLT*Zv&ddWq7vWgtoS|euF;X%&te|oGgI%*B(0?ITG8*<Td<vZ
zTJ@KwTXJ@-TJ;%s5cTc`;*#{6S@<@l@m-F(Peup>KK{9h%BszBW5z7l?MZLGaV6bP
zs**u3!)jnajgwmF1evb%4gX`251F08Az30nWE0J1#TJN*-oJl8Dm<&d=1O<6B|qF8
z>4;{g9S3i1uD-2UJEs}%YOpiA{QXUc_+a34loJG7`eLO-lFIb5G`9s1Db={^(9v{4
z=#}%o*10FSlKf81MIISDDPr65<yj)QYirEbQ;8fmy#LGl)B%{dzcK_H$F=p1=>K@4
zMpMAIsj@Atk=<G!J;pieS}Diw>)M;GF>rf<-na1|-(rDNBX3@*7<mvy-;SabNxcu`
zo4wmq2mdkb7VMQLPY(AOpuF-{hWnJ>>FKK`F=S?m>H5cy9+QCFU}YD<8x?vmMktX?
zyL}U8&tLv2CgEfK$cr0$Q%D!-QI)xVDzOA+hY;I0-FZ~!`D!~#kJm&skh@)(|FXEa
z7_(7ojAoEG25yt3_m)b|?5ESKLTh_F8av8gZX10yrrF?P&*LSRT7FL(+;hO`_t=Gl
z!FEp$#0jA^8RW?NJRt{Ev<_ojl!5tZ&e9ymADnL_rWR$YLMW7zb<0mpFTrl$gnpeu
zdCCKvq__S!<bHiMvkU`XhMbwc#R7`(=GTilmx>iMZW189ciVK~^5s%8zZ0gtv#|9u
zyTTgNk3L7Y)l$sRJv+nk+<>OH6l_c;8(>}tg=U3|C(r$IPVbmIcN(F>_8^${yT_hM
zfS7w*|Gq{h5z%pMG7ea@NVu76eNtU=CH+Q;Z(&C$=3k-lYL)6~KO?E<C!-V3pI^lv
zfLg=B3Q(#{HA=asORr6KOAvJz06G8JFF3o-i9wlr6uKp&90u&X!NZ~VCBV@QUge;A
zny;G0$fL4F##!*4@n3Zqn7#|qS8TFj`;d**_4Okrt@?X&ht0I1=C{l$8a*zA4vV!2
zmi_KM|2mCduADaP+8dCAU;9FAc2aY>QthUa2PVF1L(Gedio#~KS0@TZJZY7ns*#YD
zWHwYLPhS45QWepv->H11QKSAp_TD_2%eH+VRZoL7Cz47e$t*)=QZgp<SduBxD1=O<
zRECf-Q^-6ea|lsXGEbopC1c2(4EwnBzVF(<z1DB7z4zLCegAnsYdt+>_<ZjBy07aT
zj`KK=^E`}9O3T~iQ3%cg%NlLZPPEL06GBJ7%<{v=h)rgqWXnP%hK82*f@B}dp+gAk
z9;7^i(kOIivYVd@n~t9^r6>%wX-!%B+5;|C?u!}(97r$Zj~45r&?|zKn!E(CPR|?3
zvMr<re`u|Oz&fFtqQm!1`L~YGsPiFCJZj-F5vAgQ_igfYJ&UVVk(Wo)ML^mDSZgm)
zwFM?}t=9iB=%#2sXrUVgQ2U&J_Sph9DbW$KdhV@<DipvI1Uev5Fh>4x1@RMdY8Zd+
zfY2TlqI!#?{QMdYmfR}6FHv`f*1Hr=0WJyxB!L2tfMH`MK`K~e2@yLHn9`)n%>3Q;
zU?qNm>VNyB5zrVq^rWK`)Trcywx;G3_{f}RU#b=qaVkXfy(#h^vQz=pb^93g`n78<
zex@$`!Sou=y48CSNZ&eiMU?su<`&r8%-L{ceSB0@ly)R`V(75U$&-Wmgp1}8Da7L$
z-x8>Ym>Qb<fi0MV^)}QOZWsmLDpAB^>20;-&O4%TNaZuu<haNUwoD+To^q9Qh89Xc
z$hnZ`25tSt3st8J1#t4%omo_Qx877J>_u!>jEvWEBl=7I-3gR-AeH%}0w3VmmV~%#
zCMYFn4$E8L59_BMTz10`=@;50XT2f=u$JNNV^3kCN6;j<PP%@4X^17~OJL(EK@f0F
zPvgwJP($B$4xzvP8)qn(W7FqcZdbpQ%zd}a53@!tYlKowK9*xp`#Fgt5m-ach(1y<
zvEHE&Mc|20$cAt<*ruw=CMol0Qt#iwHjP`h8t|*BoSvCM!PyEI2jOnp#+a$GZo^dR
z@Y#A(J_YOY!lELCoxGBgCCKnnyedLTxgKsiRl3Di4t;QGqed8uG~|8bZrL^j7W?<_
z-?8IaN8Q)2Uol7|i{W%=5E55RnNXV1lFZyrO|3E0{qk&J^0VqgDo@X@l7l^aHRY-g
z+#V=hf|iJsl;N%YwdZ*WM_xZD11tEHP$5!ot&;+E73jiLzNE{-%d8G4?p06$Uuj)_
zblILBHHnb7qWa7I@%-~^rR_mZ)=Wi%PRrRCY=HSio)#prgscs1{76wKb<@^;LKLX2
zy{$VQIsVZ8)7AZCq}OLnUERYBB$usX@(dMu&V>gOpykJM!^+MMv$z7X_Qf`|#ma5X
z>_Ge+WQ%2Ea*4vciIT;cGmG5_N4O13oQR>BK<H)lofk{BD9af$zfBiwv~-T>6a+dF
zj&4d&jo0u1^aeFYZO6Y#4>mEVBl_`r^yra~kGAA_(;xl)iI+7B=Uxnu-0>$7A4yDm
z1^s7>NLXSEGc$x~$snshO&$X%kT`4VsHrceSD4DE>+hlr_VdYHD@{pBN#xV;Blh+$
zYp>h~_M}Lj>-bZ;uFF6)r=Od{e1l8V3b~MUVKOrzO*mqxZ5kdR$JDUz^<a8_SJtzw
z+D)NGeovn0K9*R;Akf?7askfM%Ep)C31#fZaH<U(j?_s%eDp~9^l4fSI5#JN&yBA~
zVNErCZToBYSx{I()~``*8Yv98!OA*1I(N$$Prq-O-v>9rp2rV<0V04!LyZR_Q{R66
zl)%g+lHMY)X-tRZ1Vr0t`+(#bX3RKc*wY+War}Hnc96a$&=>L?)ctL0vrf|)MDUR}
z!^&rtBgrHpDw%8T>}IiG$v1^rSWeC-<6bWVn?x!d+{(tuNl<Xi?5q~rqnXhZ_vmb?
zi@~x%*1a2A)##enat+~gZ-AAU+Qp{6`EdBD1OQTJLVn)@t0rZ0$5C(MX!eRc3tK4L
zv1LQ-nFHM1r~5BFDSk5$jKK!Y^Bnxy4s@NSe2cTC%Ur`dPvX?V<LOzYP26h;N*2#q
z+B@<t)5s>^1G-g2EuK_={_F_!1E)QxRY6%fQp&Z&#xuU`JjR&@yL@c?+=GWq<{Z6b
zV`Vi*I`_j-s?ljziqX`<g42;>v@|rf(2I|}+c7S1(7-YS>OCmA2FiN6qB|d6bEBIo
zT3nB$2WziXFpV`}@wvFUTV023o-B9Yzjx0YP!o(|R#Pu2J*4Zi6q^^27>D`jhS=~^
zhb;043tw@)1=0vo1IGbdBYoZ0mPTsJB5z@VJ~k36>fHNM_KDl87MflJ9-Fi7Ja~FR
zvb}`FiHNm`+7+E?q^E#EqQgmP4ADY>-~owCmm=q?;*bG3q&ni!%sU3LdAp!%JN@>{
zP|IWOrhIz&PQ^t<q@yt?y<MEFI*Oc^IBs41;m2FZGPTuw4F6+yu{Yrh&B>}8q};r@
zrQlEMn<3s?r%(G#Jgvdqrw`n1p^W{<=Q!QI>&NdkXodM-u-LR|(`C3Vpm1hKN>=@|
z4o>&tzN<*38mj<T0KtX5M0{Y}Lt0vZ^$aijpiI(uuo<Z^`a~>Xl-}wo#n*D}U^^j&
zKmi(>AK+?n;P~6`dDYj~19})NRhoa$<oX6hBMz?`7;oe-Ng$*&zEptktPtPec9qT<
zQ6h@VZn{n42BeyFwB~#F@>2fkDCoDDeNaDZV1U7h-+=6!nBXsF-FG}U#Bxx};gJnE
z&7Lybgr}1L|HMA*l~X=&D1hzAkxVSO=$|UOP4|HpTI61s<SqPQicyIE%96P<pjc^U
zJl@M7>k4#W+FV6U+<OUh2TXgHE|qq?*Pwi;3BW3sVOu)3P`qSIOvMl{3MHS+GF|SL
zMNbR+EDEnY15$2s3S(kqj7Fdb?-COelQkZJsX3O{wSR!$o#p5VZ}ZUC_gZt(<KHD|
zjFhmq=Q54mf#Z2!o&U3_q@(i*6RBaBz0A81nR^LdaioVH&6Augj4Xsv11vJ<THV>{
zj5g_B!60qAMMNHr-{yb+&b`=hd}J8}AVp{XlUd987&{5L=L+HwV6T{xWJdh}0paso
zx3aWWeksfMB~V_#=lGbArzL1^Vm@*{Wxp&D@mdD;;EAE5z5SDSOj#9DPEJl+_6kQg
z^#elpFyze!G><w9hO?#sDF!<Ec(F(47&+B!P5Xo93qsh)ew^Bs+Dt{B1ISj<r-USn
zU%f)FM|ww~MESsv{NuJ+1wMcXL9$ew){}kkpaG<X=iSk7-VAKn-Sy;D0%8#Keb-F5
zU|4O+xs&?kSaFy(Ge=m=x<X0aN4%n64?`TDRff)>I2qip<L1FND2GQc-_HJE+E^uZ
zn9|M5>myc2`H)8PibVk~)ZP~w$d{FNQgA9Z1*@YUFm`$FeqvnQy$275ZN8y1tfsCm
zWQ*wZn*jtcDZAt^U|55`%2s)U;#*p_ci7^QGO$H`Jlf9WeN})yXu~j~xr}Tjy*C4}
z$)cfEwq&ynqOBSwrQla41a-hDveOzflw={Ao15XEhgHJ(t!-_8V$}1{JIf1`sV_02
z6l|XN{>PIr#meB7y;ZV?;!mr{0*pIKk2|c6_zC1+J2QSQt@f3bCE?k_Cq4u~7I$bI
ziq+jGyG+dKXTsanwtHh)yG>Ro^c2w*z|LWDeS!qOh3Y296-?q`ZJ$Td78)MTVQ}(n
zDjWjOrF7o|lx_mCAy-sRU|2}jPHS~Uem|C<#os6U9!eoO9cr?{wn<B+T6}by(7EAn
z$-OV&C%IJ%7z+HbekY8cVAMv=O|f7Z{Xj>~PU>Nfwd>b^{P8LkFp&;07$xTr<l*Z#
z)5q<tZe#X-e%7?*(<hwwBGIBiL2nmgrXMugth_0#n*Ol)d(vry5ybjG+g9!xh-8|R
zi+oT>$S0J(ldM&me@GAE`bqPy{zqPUB;#O@r8cj-()GsC*q%8IzM}QG_JO-)^#C?e
zwMq3gwxu)ptvdwUWo|2!b^>;9XZgTw>a1_aC1y#V070*j83&(ZdRb7wa_}ICe>`X+
zIXvCFj4)YQQ4y{vz(&&R!2>x|3vuz6e$VTW0u71^ux9C}Rg7r&?tO*kCxlNB^O2q2
zkpr_+Cq7w?ZsV7z>$huiIwqaTA4RDF5kJm@ghc_&IGXf{k^!D{5S#f!hr(Vg8c3@L
zi*u4d(S^7eKm+b`5yKdW(}iWhE&`E_J95{xAA^It_v{e^p%kF=WmI2{Y1aXJp}v;?
zP;-}UPFUIbdj0e1nxH&iH#Y7CHJAtspiB3JNGa#x!%1k{Lx77!BjU9Edl$^RfgsXn
z*nAN9X3@T^;gbBSxEL2AfZ8t5PcVwl^_#$prDTz|SnXL1%K3(hX$f#dq{_d+>*83h
zTwGX~pB*d3*YGpo&A)6ObVNqONQnllW&AYm8s79;U$cvfN*$nsYV;jpd7fZMq2)7&
zP`Jkrir~1POL#}0L-qhX>lsW<Lgz@V6aR_h$4Q&*ze)|X_pe=6`fYxx;@kJxM+mL~
zInE;ng;1A|hlg6_T?Qy>kf}dNNU>9cGi-=xC{_xnyUfhP7Lo+Qpu-U8?P%S;hTx-o
z%V#*6vOf1oZPt2z)&b9!L}!3o;<ym6nczR<VlNg-2T<8XJkXM$yeCrZS1<-w&Y{;_
z5F=#Jl$26iw|Vns(q9iE?i2{YLAI+iXwc4h3ng?k<DHy2b~htq3~do;2x97+_%u&u
z_kfi^tk{g;#?71eNx3Yb`I*xr9^)?O0jS_9t>7XCe)u2qLI;$N5wK%QmIsT_p^q83
zNyHF=!H{48{9b$i0d`8>Ev**Vt4NT0Q1=0NscG;&4xI+*e4POD04q`zcBc9em_cCY
zIDz+0*gx#-iXtN!Ao)OYq6kGIK#wRF8JdgGwrFI(RP>`kUvv!iX(XC4P$7l;s4D%1
z(Zl#%qCP?5(kU65)|C@TN^bb&WdQD^Mn0n(UUvSWx3?2Gny5Q0f{52)nx=jTz0H7T
zVsqk<{RW?HfOw49kte4PrC&iXu|p@>AF{H1tlRR5G}@Cy28@PGtv$DMO%Fujke<3R
z+9QB~#BgFR%qxhzp@#1&E7{Fcp_}Inv*X5yG#8bB8BQVkM}T&UtiIicCk0&3B7Gnf
z$rzF)A_oSL6;I#)bb$n-Ns(u9lVYBLauGt{SO!H!8Dh>8x3`y<Mr}{-TaX3N;fWVi
z^FX0zugOh$bN|}QW}q1OfC~cj7;}8r%`NUz56Z1lsM7t;od}tpm_P-edW{hxo5pxP
z5E_ui`9FTFhDTdpubO(75`jO4zJIK)Ch)&nY$f3QvqL|re>nm#c5-^9Y)pmwMMME3
zZR~;lgZLvixCd_<z!h_d(hc*w1ejkm!lHV4dIt7^VUaQ{l!6Z2ViamGTu5lRh@ma%
zaM;PzPBMVXcb2wYfKsr>qeoh;fi4^}a8}rPT+@dU&;hnU7FPY?!!O(+eg%e=%c;-_
z0$`1M-?=f~b9cvXJGDG~MP#zu{?@HqU?*XX0K`_^D!VYoK;?EFdjWC?K&e5@t?BMA
zMW_*c>M-phI$aR~`32)KVyYS$+m$yIqldf=Q?&5HOt6B~tUzV$BI|x1n+=Nm*RG8r
zO^z(aerFC}hbf46Nts<R-to5ty0H;6Xa}o+P>-!h^JTXhuH*M4J3ITt?Ce%b$``y{
z>8K(g$>~Et1wTbPKc>D!1_$4q_^vQHj0s-qTLZ(wZsGQjZ2gilm65q8tkda|7(>Jk
zbTUvt6A~5{K73eyUW;^6)~uhNoefdeQ0Jq6gwa}%fcO#m-i;Ab-g<>8d3}f`0gzYU
zeSmx#*Ns_sN=wSbCn1jCcF)hx4<X}lvp9}#l<N{3r5ngOOU0G}A8fF$n56}^?C9v|
z+qWqO&)9%Q0dS1UvUCJAEfZto>I}5oFdMK#d}I<)5c-rhW6~+kxM^O7<vVm4xM1Mk
zq+TUfy7^boR9d=iZ?@YFx+qi}CgTPPVyE)?-u?Sk31af-h(|Azsx1*Y<5I~90tX*|
ztFAtU8Zpuxv|B6nhyIiNuR@$Q-OinH^RHs(()}QU7s~Rks;*8&Ss7D;s_W{I8yA+A
zmiD~~hc*L}{ImLv*zg<^;g>M|^QFz8I<{TedETgzs`=%nE{RvH_k}Kil!jK1Q_9Ln
zw4Q^c2Jph&utGiA#{jyBOcv!T@gyTJY|dNp7Z_z0qg-fOKqwe%b*QPSxtus@Zqopg
zsZdL0q(sgQR&2mOfBrePZB4)Oka`)Xiq7hOqu<uOvxU&}t6ExzDPCU(PD@-jt!ts*
zw62AwZtwfwj!CC3L*jO)oL$O+!%=LH53Ng<KkeV#SMuv?#Qo7dS#jJxqdv|b4os}O
zE1eg?!+v{2nj3Ip`1d2(O#Lf}^oYz%pf}=rn+?NMC)9VnR+ArEEzIO!NHKexd@O@U
z&}m1RwY;IH0MP~nqfwCq`@mY&PE*)*>so0)mh+1ckB$i_`(5%Ftx{nAS}4fKyhZT7
zw|O7MR!>5-nUDQoIuE>OUrXrvCX_1(RgVkZ-adU>e``_nkM&3#d6oG9UiW0)`!ays
zK_aZHD2|;x;JxL`Kd@_HXz*}}ATU5Ev9Bd+eLX@Fj8b^J9|MN~$q&9>IxOP46{CmH
zMa8`|@Z-l!)bc`Ngh%+;2JFrPk+ly7;H#g{*>F=Z2p~%d?(<!O0)sV%9m-AjugWp|
zNIU2YlRVOt=o>2<TlL#<VByUkG13YiCw0rwQH*yMgDiNz@iDv<LT;-cEY%RP`HkA5
zL;d^b{CFs;wOX!MYJWT0LpTMS@7FuXRVYarE-x~PiJd{4BIX~S2>!fwoFWW#5COU0
zxP-KGntQeVo`&IM5+zpA7yK!s=?qk-MX(NFM#3b?E1`a)o1m4kisH6}ZsFWkig%Ea
zZX{X{m?&*w1pCl>JoW|Gpcd)uH=FzuQC8gY1_>7mmSQjXX(~~r(94$9pFl-UG=(2`
z4)6}F&iA#ov`TbGHR(`~1t;m(I|LLTv;4#&CTCS&?z%5Dvu*EQPUqVy+tw^<h2SYv
zFyblj4}mSGC?gk&z~+e#tEgy4W~5r*f&}tFb8w=5+hVR85Q4p*qde6}cOH{F2Ss8F
zNZxXGJ*%|No?QSRBGeEQq$NQJ^7-lO|L+e|MGXL^M63`J!pD4gF=XjE(r2I1<7^S7
zL~#jGNKC9F*$m@TQ#KQylWMz9^qiyXg7oJ4^+r?6p~yB+p;{w7j@jxqhUb$pmmaCK
z^fnN__IB=~c>T|Vrh18f0H@nESdi$6=J69rw^62h4a<qJ*gm7HtExrzORl95XB6Jt
zX^j{$(siDT{0!^&-hnOTFXZ+072t)xf2D}_z;*uX753jRudKxXH*4~trmz#<@mpV?
z+2`kY-y(L*G~)tC0$G9ST06`_5ia=h<*b|?sk9oI(UHm6c|W1UdOeUtf?)XD*zl}e
z&vo^Kwks-r_}4LNBb${k;tzrC7^U*}^}qXaKzQv@jJXA=PE%J0`;1x?9}dFjTr9Lu
zAbYvDZ9n$IbuhxPc_=7o6nVTcWA`H=X#f3rPha2eBD;(^6U6Djc=nN>$BUg#!Qk&F
zQ&7~){7YL9Ur;7v;r_n<`{jQv#Q(^eNZ4l@{+kQ%-$z28#Q(&8{QhIb>dGksAyias
zOuqhm!mQmFN2w{f8wfSBl8yxHm2D!ticG5%uLl-`D8RyD5$X2er!l}}1!8N-Pu!+v
zO2*8OtvP^}*RNaGk&=tSnn)wtQv_GuAYv6U3=9{+_ahuPtxNb5GgR@Hzwf^?4O4so
z2*9<}1W4?kPfb|)#`PWK$lBK<*m*`&sIIYM?B$_D2@5}0zWr5m#VN!9k?6Y?hg^8w
z-I)LNTRrGv<Zxw%=fj_WKWh)Umt%0>4?R8ERW_nu&jPq=`uvL1xjhLdj4^;QUp7AK
zxE_gFMtDU~k*jB<?E856Dlgq-e@rwXFMx>n3nem}`}Zrs{m#Ee67J??lcv#??|<ba
zS9g0kQ}oEj-G2?z>=E+eUtJNYe;XY=*IJC8i>J;rdX7_nZKe4!@|Fv}rKJS(!1qDp
z)I_!6|ML9`&nZq*n-yF3rF-Q=N{J8%P>JAAvr|(}XoDU3^w$>dC9a-g&MQCP4$u=<
zMmI`sC@lV&{`-Lsn91!fVF-f$`vKTk3W}FwD~8*35;GX)=I8a*o8G-+WnwDpT3q?O
z*O`AmFB#Kf5Xo~Uw*!?yyNl!O-#2`RP?Cb;JLjpE<Fbe}P)ojhn_Ph(8GMVd-6itx
z&d`##Bj2UCesJY+)M6qmbZ+Kt5u#vnu>O&MecGLs2TT)x=rU6*HW=1!6&XZIk%SF#
z5~R7Dy!-C%pe1RVx{~iw5Nr-l*8roAbB<1rb$6fV)igE#KYy!})c<lx!H=e$dgU8?
zACe1H-G!))h9;i#8rss}eAr8*R=z8IFT6GnS|X5K&{^?1*%r_M04QeWj<N=*I=1ro
z=-d3*Pvg_mw!|wY3EwiBB0C)6dwC&Nb#wzFZUmn$g9GRArVkWw;^gJ8|NYtj*Y5NG
znX~=xPp>@A|2uKX|7NKI!ecQ_@3ll_Q&TpeGXQ|;z{o*2#~~)I5KzYiAqXW>h`5Lr
z-ktRz^<WTNkgp<y>FVl&^#g1L0H739E$kqmzrzDtR!$wmPGaa@$OAh>=FkQ>4giLb
zP!06}`YfRO0gzM*jg;x&+Z}MU-+vJ#1BfW}ej~4cR<F)Up`oGKCw%=Y2u*tE1x66X
z!i7GlGR!2`Ls<nJHE-G2hKraG3(*I>PzDgv0C(udjkIOum4%t54#<1^Rz_|X$M>C=
zxA$3a{Ll?ktDwQg!eR#e3B|gbHnU@!i9$O-i_k73fDkaOXuWCmpTWGGj~({l=7Lm+
zEOI{s!__gv{S>zym5hyz^;FU*ILYXbLBq4S?7f}L6o+oWHb#E{6b!_Q6pw*s_w}mh
z1zf-qpmMu!-@Yq}C7tGgna={_${MAiqkH+ccGMoLw6wHDRQu6<dmTLp0EX>AP+*W-
zu~5QN!~`S(EBg%UGFr5M4h;<rlpG^E6+oyBiWr^&iUiKI0EiJ)l#yGEWC3nd6U%km
z6KDp3#PM`}MkzTbpaF^T=52ot<3pAKKB9K`)A{c&Xc&25cGq{56^W!0x~Yeof$xX_
zo&^evtohU-4ah_R(}Tu?9$+r2;r>M~)YL>oej~jGEQehI3M5|oK^W4FrUxne@$-(e
z?BF}0nt>v<;_1_Qvzg@TC{DW(BXxrS_Sgokt*9Q<VSk}|jn(nLf@VAqPfz+4N#GM;
zEdTsc^QQ^B2(m+k5R3tc9ugQRF9aqqCfn=gef3X*<w9c!U^_F>F93(YbYiA-3lAql
zlYzy9)0R@2a1omuJGdLY9Y8#T2X)Dn+;-ea+~E?M_eIgE@u}k$*z2rM?R)VqDQAEU
zpyq<n@I7F@lN3I2pB5=301XGefSPU3=qP&O@UjA18i(b>>Z(38X3_5750Sx6EKGl2
zpFydMGuR0jot`J57)tckpcZ3@RRlxZ(jE0GU{j+2j|q9<#wA=a%DG^QYbD$TN7xuv
zH|w~o>oQR5M^Ld97S^@R008bv`s{!GIuww6>cNVy|KdfQ%ec_B;^|=UD2EaN5)uM|
zm{?J>EauOn){+M|)7#st&LJl&3xYxx${850>1XK<pbOYSE#Nl9OYxOgB*ev`AP~(d
zg=S#fE+ggfW5-^T2_&27zK)IYq=T222Mz$f%Pi+}*ls_x{m>4YHV)tw42>8USET)7
zm}zh3fQk#@<LpPE<p4TMY(r9a2OcCSK3E3hmUN7pHrpJV3eYzk?c89Mf>4)=qI*)S
zMSE5s016`4FtHz}7|wF!RL2f#>V2KCTkM*k1mKivsdxcRL)4S`vxig=aKn+-nbqh&
zz>Q9Qp2AF*PCGP^xiKQ70CK3&zfCqa6YjF=0@{0%^-GZn&M$_FSSQ@bKs5mV`pj$s
zkZ`5cN$Bp?cPI^;IFow#RseS5>2E;3=&EpU=~zXv^%=}dWY3!0H~iy|j5Nl3#X7Ee
z{Wce2it8XAJKjP|h?Fc-nsU;K&Hxrl9D-KCV`MTPkQ$g=J-fFk?_w(2VunlOO+cS0
zMdpL4s7-1~Zf>j176B;^<y_#ULnN8OiHHNd;^nK-AzJ7Qx_CuVMJ4~a@;(q~=WvoW
zg<4>ov6+ra9UzPKx_uim)~`GO>OXZ%p@IA3{q>%Y9+ACBFzUisl-xiRod=$kohF6=
z`6d!*1f0!bD4G+7NKJO_`}e&4_a0+@U=(0JZ4K;1jT+7D^JS|y7N(u%m{tbM7oc4F
zRc)_^ixD7-x3;RYX0L3<(SL{om{YQ`$+2{}b*r#7(3AsRkVK^shBvR!9r(|%%z~Et
zy?eSlP6%B>lL6W+Ox;s5uSrRjB38G|jTO~Fv%WKI4(*K0U~=M9buM(anuEI0QZoc`
z%i`8&S=dQ=@3ybl$<yw{qPVI(e`WzRPg29;lA}JEUWyHt5)Ow!n2tc|03HHy)#u<I
z85){Lty_<QHlz^71CI<_R0RD1;0<e}hVh-4N7IKUf9AzL+h2%5L2C)@Lob|K%~lf+
z7;cNc9D+mIUeP~*rrX?Cc%1NuGh*B;YrlJC?Oj2z%E?ND>#5O}Lp9u?$9<Qk5G0>P
z+1ZECJ-Z)B^GmC5kjXSa><Oh0wcHz;nrNUYuDUgaS{2mBnSn0uMI{0hXL<&P=ef`9
zPo45kL%%5(sE>M+&`C<uP7mwrOL?SQex8?8_XQ&wo(e2rZBj(cFP|LNTeE_wGyWl?
zws3+rt=I@DQ9W&ls9v!Rgfn-rw?F(bJ3AXCP8+amER3LCOSINemTO7CV7W61Cze4_
z#`P0+oD%9iPa!UagPiu|h0%#V@nLWg**F|;-I^KgJkw4@_yXOr(2o~=Xi+LhtE!HK
zD&7(W+{<$>Zx1o$93A9ory^JLaw}P;Xq%&(8emD(au-n3R*jrIr>QAm)hp9tAZjz9
zn(&GnYH(^@ThVT*Wf0l~GV8l{a@zK&tzh#i=;?)is;H~uVPcZjYbvZb3C9CDf&@{o
z(&*@~aEK)!UDDL(pJp8{giSOcqrMPT3lo>RfN^JbIY%K;@<v;ZRDifXeNZ4!5Cmq<
z=#7GA7gTNVVTH1-ZX00ZQW)6BrU!b@B!oSbl&~<vqodrZL=n`oh25)=qo=Pg7d9HU
z>#z`LyU@|j2G@l9Qn*?Cy!I3DVLx@rtt1t~v?~q=jEjR@T*|f@*RQ`&-eSZ{huT6i
zCl5Ebq{Bqq>32DK^MC%Fu^&L7en~l5`%%M2x352bBycX^Sp*r`-7O=XfC@^QpY|C*
zCBns}<Yecf6nM@=6aq&nFdVfJ<Yi<4B%k2LeJv_AoE#jft8qri3NWs$Kp#rZs$4|6
zEwnN5VlpZjd_6B2DM5nMR#xRkxDjkwaN1ty2UJG&!$ElDx-7Hp9;_s~BxUuuAB96|
z7T0@PFEMSm0ocd>0RgeGu|Z_WDdwys<SwPuB+sCI0$gnzs%+!=zn{I5(=434bVLP5
z9i`%qv6t~b&hfZzGrUw)z(@%eiJ9dPB0sg!9ox4vRfcILKHD#DKaM{4;QjF^#9?!(
z)##%oq7?=a7C^sC-vN{+9PpH@1ad<mt-@x-3ayZ1DCUR?ckWR*&u0riF=H=9)&QM0
zn*2!*bNsU})u`(<dNhZCegTI9kvq1J;Afb+#L&y~G-*!328?222qWXM0>q6hXmdw%
zVQ{|oX~>D#sH*x6B~ZzKTyakS+Egrve|UIQY@1NWz<pP>mwYf)-|Peg8(kZgKl{KC
z&(-1xEh<i~eEVl5IqZN49LC<hKT71;;4o;9c7dw_2pj$S975=eJj~5Kied=Qr3dLh
zdU}rF4%5@oc9{I)>B@TU{cHF>)YTcH5<YYYY(lh9Bh9XvXa$%Jr^v-2e*Noh%RU@-
zi`*s*eQJ4f12$Qk&(0CloJEvuP2E?mT7~1F9qCzuUaQD7K6T`b`Xx>UsL$PK0-5Ct
zY;tq6CWq+dTp*_E!!tygC}whQCEWj7LbBM~rG)$dTBN2BNC4MSXl7z!ia)9jr78qZ
zr0XWM!x5CNl$BWmWo2bmH(;E=s7)Tt0{>AxAmy_Hx;d~Mfqn2}a9kN#U;Mqfp@jeu
z(s^{?YEhT*7!*#y>Q&d&a5<W%fB*TDj3We-<dPQTdbd`EZCyEH*Lg_v`??$CH6wKP
z>YqV;Oj=LDNQ0J}MnMbuH;{#Z1ylf72RsHq9Vuy~;}>1QW)3(;B*A_7kkY~k<SdBg
zx1eFBr8ezsK~|t)<x@<?O4`&50OE6Zx1qNE!rU(qK!@{xjE=(I^pYZu?J(%z0*HW+
zrPyBkcn~o#avUTAKT#^uj>M0rcFC_izAvs*+Wh}ed@U0IdmY;iWCwlQ)cpKeFm!~>
zssh2qc`9nl&c#(@%7>~q(X#-_?Gl%20~{$XWDa^lY(ghxWcG^P<YZ$TL|Opa7Wj#%
z3nD&LyV8%Z5S0c!SHV4ei2|h8D_M{$)UyPB+{#@&6a<yd0&6)|>4L)|A|SWb#A^qj
zIoS{)E{L4g){}oL=4it~SUNphwW81`r2+T;ANZmGJ00!7hgeuhL0aDyeh-NSUg;Mh
zjL}VonXX>FIxrlWB~gcgbjsOMcf@CKnj-A6UKgR*2BH=6$D+&aS?b{_==*C;7$3i`
zWIW);h>=sNX+}0JI|b$u#YTZeOrjH8IkI}MB|PKcag#8KO&g?nc%VzZl_(M-8}&tb
zU$?V-`H$;z-EP|av)LDvvFBYPL7jsQ&_3nL{u)Um@{UsTItDpAGLn%()rz>sXl^d9
zbl@G>4q0-eK5)b~m_H~en42a>|ELmgT^Gv#?uS9vacSu?SjOpLUH^u@bH7ljLV`R3
zr3_?R<lUl=6X@xf27U~cCS*5xu*F#yR3TT<ix`mTZ)+K$Hj3btjLzk}HVKjo4#FZj
zjjNieiyD3Xqr~ipbSzEl?mz7qG=L-JI5qGKx`CpFE9HvYp0ZbwyeL|sH->0^b8=n=
zhlMbUI&zXzQ&YA3AQg#Hm=QuZXxy-Y*L&?ZVqAo_AfXUt99Rf>`G=Od$N-T2V@0lk
zFtVlY0k<`d3rO2@bM`Iivs0QN15BZ*&=!|~(xW8UQn5uVHugFlaV5In|FW0xOB(pd
zy}Nho{Y`&#lGQ6(TBc!vffXQtkc6j)&93U0K!|=0L~QA8_#J8*8Z9h0l2wq-fHzz)
zHH`rOA3HagH5hd42$T=A7SLm-Uz7#~MkJ?I6JWC``OJZ+X@}Uswh3l~(*n*0zUcyt
z46Sx67(HpZgLk(|j*pM427`O(@4o}bQGgy=Q2~%Y2n}^Wwu8-t(WfEVEAZIuKX;|g
z)jz{2fvKebxOgS9q&UqDqYlilp_pRy3{pK-m8Uyh5%PabSgz}eT>m|AY5AL!onwbz
zm49Z)j%4{gn-z^ms&{U_u&zDJ*O@o3?7Vm9+w<PeoSWG<wX@UcbGq+Pj4#f)Haq4!
zxfs#y5cRX9Xsl>-Pu)91+tJ~({e_>bBR>^9|3pKd@^*%T;?5T=@PA&bNzC-0*Vh%}
z|DS*Ikf{4VFMB2eUjF;_dLs!t{_{Gw0!9A&pW`4v|NHOAiFc7_`k&XU|9<3u7xsT*
z`zW2r>m2Pa&A_XTR8WrHr3V%R`_uNBBx`6(ue^o^NQ$H^9Bo!sR<nGBBga_Y)5(s3
zafFj<98Ths&5CEvgj*MRcqqUo(HD@(q|v=d2?INc)z`X1T*!Z3|K02q|37VZfxu3R
zxhLW-3&K1+zKoR4&Wi|0*@{G>BEeb&yi%VpEFv;EFmO0kf`U=)>eW$DSO?CerKh9Q
zG^mB)b|oqKbadz=B&ex@cTXGn;dB#m2B>~XuC_KdaAARM3>1C{V?-n*B1=GqK8bM%
zdOwLt+)+NDx)pw*=tduv^tQabJkrg6wYEP#D+@!w`B_-pI~q|-dbKg?DH0qMNYFfk
zk{&^fG&Hn|EK^ca5cyQ2!#yR1Pf$>hpC3{c8aHEK1_cE{l&=5#_lFg|h#vvvJ^y70
zYF7<{8>E5iXP|-t@r%ayHL6xxV6U&d@CJtt9fF^O?^nASd&~&*5nSp9w<owS1c8B_
z5p_!H>i)5@Cd$g2b6FS}DU7C6RL-3{2Te#2c)-Yq`||Mc2n(~U5z$^8-5igGaWOG5
zwBMaM^SP~!jCR=OA@k_VebD4kIEpGEQ*w#=-}eh-tKS{E-+*uqA;JbXsM6X%O5mn7
zMc`!swNrGW(%X<<^o+EI1P52XeS2NIcW5XjIr)R#Ly%;^_UAv^(bh)BTP*!_pLYOW
z363CwuE&0UsyaGyS3cF(BWxrT9E5)8jniPsdx7p@Bu)U!aqLo(lA@)W(I1Lz85;Zr
zKq0Fv89EUaV2B*ujvW)d6&vSoa&d8GC4;XBsWral2;6W)M0M9yG^>C8`ntgD@#6-_
z50Nh9D;qX%COH7%GF^Pl`s4RxKNyi)wrm0WImUHlgWEty?l8ayqygxad6xELmD6gJ
z3*>xbh6e^X4GIIZzaveD(&C4%=ZT4*5IWsC0lp8$welS;EGV$Jm^7+Y$;vD)F3!t4
zj<!jlxKlr&1cGWE!j(WUP74bQv%=(@Hg)b%*V0N)yoTV9kwQU6TC)Y$g8p}q%^>eF
zuyr>*Jpxom^!_632uN32K)44|$_&xXyE)YCVZ942X@u79?yhzQtH+C}ap+dB!ps4L
zrgTT7?)2P1Xun#F{a$9~5eH5alW+GWgR4;Q-n5atydFMOy?N%Xapn`xKW!Zy>_uEr
zA$Wffap&#otEy6-KljNV2gLomqyTqJD_n)PhPEQ4NHIoFsTZUO%nhmUgXj7H6ZayF
zRkScNFz7_&wma(BQEX`7JkE3D(rk|o0(S=>xG_I~k@B&;0CZ-d0nwZ~0V9mdsHNJs
zckh~;2Oq3&{CGxBucfuMYWA!dNZA#?cM{tBlbZMM-xH#{Kv^WyiTkhzFOPf<4h`Ks
z<BDN9^po|TJ9h6@$Xf<NSasD)my=c_Y8i@lE4KsF&sU-fNTi?R6*>uO$D+f+^udyQ
ztp;G`O;;+oqpO~7TUds&0stD0U#vNR_F(kY6*p>?I-W>LO$ExOiEcKQk|%g<NP4G#
z_Tvhk@T~d<b+ffMU%4!b#=Zid)2U8Nc51inCZw>67CT;>lTj{H*T_iYtTSsUj9bD+
z*KYzhd#D?^`qCnXUnP#@-R8r{z&;cwrnrzs1sL5Q1cZXx(#}uZexE=h$GRZsmpQbB
z)-ODKcx+6{q;gZH?ib-J%v##om{4boJjh#x@r<qQG`PCJ;MZ{H(>EjWhGj!Vjw>@x
zmvajhmGSx~_n<Tub>x+SA^?j;2u%*00WJ;>Fxv^OsQZx2n8kc(US7YYOd+H#xw)9j
zl3!3DBC$&vSkv9RZxDLZr#pZ~y|`S5tw8_tCDr$D-%jL_vFKG-Hy&<+5XHc<r_3{~
zduzUbVWDWrs*z|;`~t*AVNil}CxH7Bgc2%#%sHN?r;CK*hK4L5Y`1@Z6i%fA=h<`T
zN?;0+M=BY*nwj;Xc(B`gF7A6zPXiK}`CtG-0Y_G75A?ecY^HEy3M<U}6c=C;(rmp+
z3(Wv7$d0HicfIGpH^;x0O?y@&8X$6Zn?D^~m#Hg`8D%gY3Z+V#n$6g-?bHi<I-vH8
zj+r;|o1PuRu0$_ipu;Tq#FfFY%<L?t(KY}c{#myvIw5~}co<NtC~{_`<8TMy@k1m5
zb`=YF?bGvpdI_-kb#-Itz0f$9x=K<K_mA$sDJ~czQYno|`TF&|#i;7>Zx4NZ82B%o
zLVYo?@6x49$PdV%jdfROB8Tj*$phDcB#cG<=N#DI?N1J}Z~rEk{QNngZ+^eJ6KDJV
z7DVP+cF5eaviRm{4Gj$q3=G;War1ka`<V}aL~8R?%C)xZISB6q0|PLCA{*_IEl=;0
z<;q64$1pmCMt;)M)5ER7KWNTy6{2w*H1~wEJ#yB#5+FP`BoFTJ0Q~|fo$0k}Jl3)e
z@Mb9NRW%#~a_9E=+aKs(efO-13c}<2JwFeP@PK95KS&RN4SN4^N*A1TcQ-ke%^$gG
zy&%B#{%Ow^3MYW=wQUo~@;{#4(|S~MJ6s!{A6M)dc}ocS09Ve<6?C*}z-hxwEm#3T
z#XVAPqnj;IE(12OzyAaVrnR=VqUeCMRY*V}FntO8<|lMZeAexN{rBFo3^uWFQyQ|(
zW8Xk1ITqj2+zfiigG@Yr4?dkapg$;Ye)x1)^)~zy*2Alq3|$B7|B<KXEfk#~5m;I(
zRpgC73RYX+g)1ZZcH3w0L}8GNP=tm>g2oGS{>)F0NtqR*EA)l#aR5SZlqh1JbvxVw
z%8RHko0^g-R|pFX(E@~W%;$~{MZ;2E&Zp=gS<p!bXkvW%@#N5aEW(Cw!^29CqTyRN
zqZL|6sGp-7bC!q#2Yk`@((hj=zpP#3SBcM}%a7c&c{A!!Y}W4BVVEuuQe?oG0T3)x
z&S7|P5RE3#PvAdrpm(q-5k=XNicr985F@Xr<kE!m2h)52`+wvooGJ_i$p5$Ir!4I4
z?UCIBeE=AO!f8jL?WMtCNXuiUQ&vP%ZJXg|IAdJSiKiISL~)EE>H-}U%(zj~WoP#s
znSZaL1*1>oDY{Fe0s}Sm^;42)Q{2MB!mx+XHUNeQDqs%k7d|e#eQIugg}N-9*H-ZA
z7w2;$oCnYbfCKoY)KyAQQ0qqu8XHhVh7S17efzjnByNxd8<JYiO5hzptm}L7B<`Rw
z?slLw4j~HfSY=#EfWQAv$#aKbSdJc@K^p+Rm(`+r?%WbU4D9ifva$r>PfC){<O=LU
zL8fiDJZ88M{e&x5u1pxB2L%sRL`;m|6CEH4=RCAcsTmk*T_T}y<pnnq9v0@mjQs#L
zFK09$cpGE)lG8sx>kkY3&SgKY3^*0Wf;0dCoS%XD##v(pAdBj-_x5@xpc&G>vP51a
z?<Oj${zs0F1DrrHJga5qh^X_tjg5`K?|einxHhz&u#Dz{Iz5ypr>)(J57_g<!!y>e
z$<GoquYhXnp{$2rz%El(epHQMQ44#_Q`}J1dgoAlPR<n09!Na^^(I)ecsMzo7Urx_
z4HDNvJ76g^HSvOYl7xhTUO_mnrMwO0^KC=l4}_zvs8nsu6^@X0;tw1J7WIG?o6oK7
z&};||h1tVfq6IliOI=r&r~<>r;Q7LIe+(YdXF^@PqKgPe;uxwg<G3%Ow{yh33}Oa<
z@WAh1>3&yV%tU#d6XHv+o)iPc9K==P@g@3(=xAub0{nrcQ`gYIh!19~3fo3ZLq+i^
z+0|@R92NDj%s#+8Sz3g6N5%MJ;i9}eW%oJ6IhzY7l65%~Zmgx-vu83$0beFT&PFc}
z52k!b%iYBDniBu#z~dwsEP}{L?k!4bYilF2&*qKTL0fk7ea9{-*FxzfBy9+bP|jlb
z%3eAcVhIe+j#gGj3OYZ2^b)s$3}*RD)ZHleyQL>2{J`uI(3@CHj>i`k-a=C}ViwRs
zE|oY#qg0VLFE^JZ|NVkn0I~+C`2P6uqw+~+5>(#$`fR~ofUCUI34G~|eKJ-<7%C6*
z9pP|kNM(ZO&ebnMO9h2s(<%Sy>1nhty8P@A92{1ogvqbWuLGEDzstg(cYc08%PxY~
zXm!zd&}9xD<zr$Z(<fowM;zY58sTxF2;^#c>5@A~f%BiE-@fgmp*bn3si#)}EVEKn
ziJ*O;%|Kx5?GKNNUR0a$K2TUuGBAj2C`UIJ!Uri45xuZmJKwjo1T9*wU%x(E>4r}*
zy}14L6ZzLMp?c&Sg7GH?{??;60nDMnYER+gf`WpPHF60F3CLlnlBFaipEB4twzLq0
zw_T;RtV}wrqwU%}f@_+_(}UmzWqwxK_7+ryJ@fU58=8AumKW^??Ytj8jJ_!VQ9y3)
z3n2~*b4R`%tYsAw3!i@s%N8Rg^HgKt1IH333eAKHad3ELdMzZzK(tbI8aWoFlskw-
zS>H)!_!;$ZrK>{x{LEW__VkopU4|ETAMV#wRV9;cVAuIw`jNq*$dZpp;*OySs>xa7
zU>JPhf{g`P1&&8kLR3@~&e_AqCph5<VgQ5_W<r(K)t4d1j}P717BXb4#$$W(WCg;o
z%a<<?Xd#TaiZR6MPmSLiTj}YwBU{5ROH58y$Vdd2pxbrX8Hroq^kErTAVi3C-Bup^
zvJo#>EeY3y({#I-`H$`z>76@wQtn5^m>3fH%PUB{cK`8|Ie_p1T{bCuUMD6cJzU%X
zDL0Yd7hx?B)(i~3k~yG;WB0+9$NkcwPLz(JlrY>FzcJPwRvCt_6Ui3y*=-HTNsmaL
zUn8BAn(8_?e#hWR=beL;#z+7+cgnYnbs<wl&v5ouiSw^1*re!o9Fg6jtEXpGD|H6B
zz}pl=tZhQC54s5MR60L6hnB6%%Nt(zC;Tq~aIdeJckRQW8$Pe1!K&2O(NWTCL2}e?
zL3J1(lbW2|lZwen4>@HGxF;ZnN84}h?cvcQCID|AHqJ!XRpLwpU3)}xb8^6!f4jR>
z`|HN1!NCwWVI4n!tmw;{a-%7Z$&?iK+QYmGIyxbf9lkxVe=vwUt~NrtN%*dC9AbO5
zT{m#EJA4nP6&8j?`u&1+E_YO*mfht+^bCwg4xgY~gbxE-ZO2x*Z89i~qD6~SmFCZZ
zvc^VrPFl#&x^3@_9&XEwy}m}AnVEFAPWjQYqIVUP8~jMmC+$1XfBkrMUB7Y~V&yC)
zOACt+%d4m>3%q}#ov4jvcxXtt^dpiC%X(VR3gep-hHi|olF$8cXg=<aDuz!+=dQoA
z4BMN5k%J-#bf_#+rUGeX6Qe!>RA1wAQl$O!XN@8djk7jFLCei4u<+x@endE$THsJ6
zNXbNM+(TcFcd=JQVoY0`Zc5zS@Q`M~78Df*KD#bX>sWMjqo~(2-;TyYDa$QX`(fh2
z$qTo(lcgh(Q12gQS`QhL^IcRxP=J!(LCJTB#>S_c2KHAF0Nm7+a>@%S6!q@iY`>;a
zpg3TDIP_UeOeQp>OCGO6<#XiEyNUB@oQk)Wmx{;FHQsB$cBqk5tJ)^Jb>l{pd+X-V
z%&1{xmG}O={0#T^i+523fI^yghW@o{*LwKjCuRrjmgb<@ZiDQroF{E{d|cd%&-EqF
zj<w|xyBEmJq5D_k5H7zp{sjx%{(!dRAJ$@6F6bH6N2y)D%-PT0hmWl=w!*pJEk-)+
zn3RJ6>rhmIOLm)T-p0T1Z1F8yv=@IDS4FfWCnx9V=s3^v3w=!xTw+t!Lfi5ae)BCN
zDr5>0{1g$VcsEBlDM*Msg9Oxc&;@ITOZD~gip|5dHrLkH*3_gU&}eq{0s;hxLNOwr
zqkp3dqHnUoVVfW=VZCS19!ORow5fj=+*M>pruS6EbDW3Z2S{+N%P1hn1_WpT{7Bk}
z4%izw-?c_Ofk1i?d7`j=yTcgq&eyN?5Wt4I8%)sj&(9$l3Ly3jEW4uS#d7qFu8ho@
zKgSwnwzbg-i6g~d_~`C_|2{Gyjjt|y44_wxBLS>HJL3fO{}8rII85Bjc#U-V{+4}J
zLd8giKZ8kf&_um_J;EIkl!H@KFU7L7e~r+u75Vnm+62J{^61|1lg!Q=PafFrtjTtA
zUd)XTb$nVAuduKV<ZmBJlkaWCj>Muozt#r^U4ko_0WI4%82qTHNX!ITSm1)x&i!lW
z#RP9nU1Vde&=?5q0}VPoNt-QrA(5$N=l=c6Xa#R5mBvcmM5qJKd{)-8XD5N!!2ky6
zenwfHaMmV-3c#hq2Z22q`(p!q7|#BK`}Ze)E`cY9a^-{he5AhcDmYF+h^9Ws;`gav
z#!IvD@XXH49Qtz-77=L$o^nj0l*{aBBr^AL6=JQSw60wDPZM??5Mw}QL$daWb78|l
zdt6^jk3c2)>0R+lQL1e`(XcIV_b^4!%}$`%8L=Ld<3|wWA0j@e0M!%0c;ilR5b}^x
zBU2u-eF)0kCWyu|)|^LC?e*Aim_%HcQfWa2RKc!euei9lV5O%mV=l)@@l0M}Wv-yQ
z+PbX1IQj^o&a$^3(9YyPxp1`b@ra_Jaoc$IZoQU{4%n}EO4f}axMcA$+EF<wF5f=8
z4k1$oHD#u&Ol$*^#m~=9ZggV=P&L5s2GawJq-daLaq;5#=CkoBd_qF34+6x5h3l=;
zchb-RsBmZK=|pgDt~KE+tD>lQ!;ZWXf0$jeuqd!j*6DQv1wr)@`1jsEJ{LB5!z*Bv
zi8DfLO1H0j<)6RaMewUN(ruqVlX}{bmg1C@A(TU>1SwV)QZXq>LeSU-wQ(fb&#0jK
zL{u>ohZ>rjK^Fg-#x5urBxy+|p?Rx$B5R-%zRsxi`rW$+*$lT_!qE*`%hw2Yg2cB}
z^Q%|6t$1o12JkSU6ijKt%zQ$vST<Jf<jKqc3waZO77;XA83~v_)ERSPy~PkJ?aW2x
zb8~a^_lGFm5M9cB^HyLjmEMF!9r2IQ@ZIqA!!(tQN{(VrbwydoyI_7hKjY9RPk6Uj
znwjD8*nWS%9b#0KbiY9k`2FV(>!Cx68rBFvonK!Dg#4lVjDu{fwQRM!xy9j#EZNwm
z2TO#kP*S9!(DGsSvB5E>*oO668Ae7&PpgC>V}CC>*_iFfHY5<|*DwGVbl{BqVg&YQ
z)@Mi@*!RruhHh(e-r6;5cx8jMwX~4b)P@OugisJ-v+Q+4T0_tWlt=n5gGlVw<d4l-
zUty!B26;fK?Nv$#Au}T(;V5{zWd8Pjon5<jjeXAdb)ikD+Y?=H()+zAAwE7AcY^Rg
z<rMM}XZK3P^uHo^A;OcVxde40klzLqm%uB=p}sl#C6xV5yKP)l?gtCq8x!N>j7uR9
z%zIlbEEfxouGf&4Kc=>W<1;AMdP$&Az@)2PKu}O5QOL?icC00JSAV2I9%}bDBM~qA
z$GC&N^?0=A@y7lnH^AdowC8dzZa3FhD?7_^57u7Tuo8|4LWft{Ufi(?-svHOC>Lhy
zg44)v-gU|FR(?8XCcI_MGe%Pr612M&3?^P*w4pF7!yhzL9`_m)@cuooM(4qROyc9v
zPV-AhfwQqr*avC$L$<b-7cV{u33<oO>`E7T{88zpri4e=wHl+>>0#S$JQ$yxk}{8;
zb`e8`wIyG1-e9A5T)cAf<XYQ7#AZt%zunwjxyGqvw+{#;aaAZzTLoKLTgUp?BEPQX
z1GI)nG$ExBL3ok3@tXk>+dEAI%x4*`*d+@g#q)8GZLeRC?pAy(>3vRLUvYni9708l
zOP4;w*jG0C?0hV511Yqdk>yHJk57pG9x{E<g1Di5P+I|!DeN2YoGBqMpRv40#?u20
zcbzZR7W6pjlmkAqzMFR)S;5F%l6M;Tk14(c61lnKcx})HSZ&$T4et=C(?YsV+*};t
zHG>E?P_z3paU+{d-OE?6*0$wu_xEpN<M8eRV{AxwM^i{PV8}-P)!EdmI1YNB7Im8A
z)jHN&Nbf1SnGOGDy7qv2%@JsX_oN&|du8<1N8mxoMVsEdd9-_cb!Do})PP#8@stKb
zx2OO-?N_Hyhhh8UpB2*5(vGb5j_CtH1wU-G3l)@S-Hzu9Zm<-c!tEhm{ID9O5Cpd0
zE~~q^NKVHh9(h}v|9qTRR4mj(;3Mz#buXdjoCUUjWhF56Cc8jMZxFHmnW187=4b*o
z14mz3C9xBXlbewdMmC<v+yEvbKCJmv_C&y9M{N1Mduwh^SgGR#AwMO;swK2Hy7JUG
zOWKtX_K%c1=~0OkWl+V<cZS3BWNsZA9Q@$$&+TW4iFQDcdcrV_?&O#SJbkI-tdEoY
zQ{G!QZtyvFzj(nbVz_<~h4yID9+ecW!s>k$dR)6`Q74Ew?t<ViST_j3;V}vuFj+Mf
zYB#-N-y@qye`s?4IF(abJX+sR8aM`6zurBd^8*HMyRL5zo+FgkPmcq8QTL^q?c`u*
zPx{q>jM{vc>Sel4z;(Y`Hd0dhz1(F;XTV9DZSK|dX>Wj}H*^>e-BfU6#8N67^tCoO
zgA(M;|B2Upr^gukC;VDT>j0`IB3?&+NPu?#$UyeQ2@@E!=IHW{jt=bI7iW^5KkuqH
z$$cv>4F>lae#xWoKu#t@yYFmrH}Q2jKzSU<gjJumx_WuYu~x(Z(q|6Wcsoi08(Ym0
zy#+(Q&OKQ&;WnpR1dq~m<m0XlR~W0PVl`M9G0EY@@#6YxmtQoEq#UO+c-rJwb-!V7
zbHJ4=>c5st6tN4waRfhkB2Pnx`<Y#tSpL#h<T%UyO-+{0;OjBK6U`!>Z{ObQ?AOb+
zq|x#nSKo&6m*(7)PFqckNM~lgdtL;jl(+kSmk+>%Ppz6>BYpAGrKZxmjSq-1uI5fo
zPj?2yij*R&8mTwXifV4z>z&oxC}W}4FMx_6C&-*NOFvm&UYI-LvY-Hc9Y|@53tlIP
zA8<0}B3+oYHLYvZch)lIVus?kdQCUBJk%W4U5SA$8+hY9glRS>omyyj?&O^)Z*r~f
z+Yk$A?8TTAGIFI=8SXY<QmE2`q@h`#7_f)**EneigVLHjH-^Ft%Z0_KN3YOJ>^k}e
zkZ;n)a)g*?5-Qo{3=H=Dmb0)(UdUw|U)yhT;ld1}sK>5o7FK!r;<dlo8tM8B+xD>z
z)Zsoz!8p%xcgf?4HrM!~>Hts1PxZ2}P@TFzK0>2f4oQ#0<o!%F`|bQ51mt?CO+;80
zIS(59Dl=2|VpL{Z?xkIV0xBr})YbV;iroPF>6VYnDg<1`P>zX>W&WK3@+ATe;S8qT
zG(bH942VDxd2q?6N&xf$1)T@dvAY5gMPJ<_`(t0<t&S#8a4s%h1h?k+i4*w92@s|b
z_x*}L#m3H#+zApYSr3*!wX`@mI@W~?@c`3=UqY0elb_$e(Ok%8<JyTXMMA}z|Gonj
zcLoGEkNJEYa-5Uk-^VtMWbppEv^0q49G@5x|DZWuu3Jfvy6U4fy}F)e-5ph#eraXp
zE6ClQeBljq5odqzR)5MYym_B!;84W|I|dBO22NTvdlA(eQ1z6TA_D@TjjuHyYn*A1
zD$n<B6#a}@c@l;@ayfGM+FifS`%Mj~LQ4O!$T8CyfJ8I04x(G(*}ZpKBXk^D8^D5|
zmmk%)Tv5FM(NWFhHE<u4x~q44c4nJzUHpgbb$k17RGyl%j~+P!rHX@&Pe7L!m3VAt
zY019$E-EfIb`qR_A9_Jgdb6~Q474lI-n5(s6SMYnoo7Y6!%GNJ1q3vaF_A#2z~xiM
z)EveSE+?YGQ1N0`={Ewt+Po{Oth+l(QsFANRwD}WS!y~u=7ymzTOtXSXvL2r=4bAv
z{Ks~G#<OR~zxs#{LhF{+UQ<QIH>`ip{{1S3CuA$2a_6z|4e=CTTAo~D+pe<|K@Wc*
zx>{WU^cvE3YpCj}Yxpa<Jt~YXe-@#(H(t{cuXA!q?cBLqABJ^n*LKcJIqjB@gOn=p
zh7WCr#d|cLLGQ_}aPu)m1%=;e3LDb;09dwt_&OYSeZZIo+Em)%DQ9L-6Hu^DLkaST
z|M`b92j-(sD<UkMwlg+9@MnlrFsL5ZJ7bjT$P|zIQ56QH4$TG!3Q^-gs)r5b3UPm&
zg<W&=ygOVuB1%dm&0nWJd)5o+YQM;hIOwSX7y>$2rgP#%zN&o>&trMiD!|q)uZ=wp
z9G48OrP~AG6pTyx#njkXYlwvRc%PUqU!nq?kUix_(8(>|;+KMcPRA_^BiG=F)}vyn
zvyczMRNXtC3@G0jR52df4+mlyC(fEO$=E;Z7IuL7iv!4M+axub#O`_9{Y0j!sbh%(
z@s4mATLuWc?B+3n$w_+p2Gl#2Om-uHY9o(s++-j+m%!}igJCtSPdih;ez{dT3qrn=
zlP1>)S-*WT6a6FeqT{PE-c?7o{uNdM+0cmpRP8_Jfv%{Vz%gC8hB1M~+eNp6N7J)+
zcyhew1TYWuC6k%=D5tMJp-VsCoHpONr>FKV!m{FGXTao1R!o}+O2!jQCQ%{bQC-&8
z)qT(ZBl73uh&Z-WQuO_w2?Aj-EgKCf8=7QjckP<&lR31+lbt=@1GlVPOO*jr4B*c^
zP<1WV)n{)P;Rk|1igf7{*!^>E!M>{vud83Vg0T7;=pD#CU{#WSHVa|`_vtHFxC|AQ
zk-7Cq=F1<vas4{ugWE@pyCJ*Pd@>MER`A9ns(Id}_a8pQ%qXj<gt&gC+h1$tnw37J
zkdGn-swxd~^Jzfj&>aG(RlW7~Zi*W=Hl}IEWp_gj99gVD&<m@srry0;a;1koA3i+4
z>jjX4(GXss;=K3AfLxefxB%rx#W2~b5zrcNO%crN<C*%M=`E^Zxi~pTfa+!atUO0&
z^=~dfS3TFxmxFy@zg|a=1^f0g&`<#m)G?GV;Gi{zY<ZrNLNDbaE+JEd6K**u?|$CZ
zorX_BzCzHVQASA#F_Y&)J~Wi>*R5OBN!>2FFuk(7h-@uUEz2-ur6e6|z{kR1G9gz!
zxB66hA^Hz~UH`tn+=^S#!QnR+Q_5*hRa;xuS6O5K(b^|`?)>74kSLGO$bh)l@fTgW
zD1K<Terj!191&Oj=R7DWK-~9?@_*SDx&E<y4n(x(tKtEwVgsaqYQHo2O2yFW$#YmX
zm}3?xdvC>qGet<l(4tt?)dfor_GqQ7N91}mYL(R9O-MLoBH4{HpZn)SQW6ryIO%8`
zQQto?Z+Wj<g=Vaa(uM)rHFL-l14^%<6oWfCnEkT;BNEoN>(({0X1zk0gOTy}Jn9OQ
z(721x+iW)Da1(>pC^YL?cgULaB)mPQl2OLcqx307i^HQp7<1P*4qcnEUlSw<$tipw
z>%oKD);takEkKJEl+A7qHiDj-i<&m1U2#HIQ#TIooOXkU{DFQ0^tF!Rcig*oxiNZr
zdHwd`sHveCGDrp2hJz!P&#_+48cj_%esF(piuPD5Y@PmI>iE01Ht>5Zkcvd-fmn(G
zaaX{QQEwezT6CO|t(88vdaIRod2~ib-TQZny|4IUw!Z2G+35G=bl8<h6pHGNZ20-J
zVXJ)U1&3e|%2~n%EXtqG9r$(Wu+)u;PC!+PPJ@As$4kC20>MTdHahVzU?up<U1EQh
z0jvy^t^<81My?8Z$l$I@s<I`jkA7N%Ce09_9rDssRb@O8cu&$sZS9)@jO5cK1D%tm
zHUbPoE(0W&Rqhh$87wMNisF@-uf2Ocgc0pZ;NBvNMNnJ*&v*Ci6Y@@<WDcMp;GnL)
zqikrbL2N55!OrIon9$PEBl%`?vjg(^=)^=;iRWNFIAI8U&*h^wd^QYiMm83rTb(}I
zzHfDiPBig<$bY1`94c;OC@N7OI_38yLPjrgJ@&%5Ed%iOrG-fq%lAG|B!GB1h*@nR
z>rlU+;w_Ad+AkqpgHkPBw$S(4_P=|ehSk5>{hD9arn%sv7=8hPxTRoL69r~4wU7}7
zXJqh5XfL6@g-92q@{g6339tnKD&u_felRa@L{%FCV<vz<kf;LlQ}B+VUc=-kCHK@$
z`KhY?$&>EYe-z*DHd1@P9n*m2(pM>=q?$VA;38~z<c{gPh`Y4$1<TC<VQT+~j!Y%+
zX>xCOerETEjLHB%0f+PV{hRc=ccc5x60Ig94@|=q)eRB#fZ#2LTq}siQyTF&3J+5S
z)q(e~GIe_$1wdcPXShb%iFh_TMWB<&I;MG?4?jbp3dQ8BnJ*Hxx2^fr^Jaa@G4$FU
z3vX|imkicTi5MsqnF=}>k4XRu6(^9VYVmq6A;bZ`gsSS`Kl!LLKvE4_y=3YWNd5>N
z$GS@m!zN!YL_R>t6U9DJ5fO++vO*r)*nf9Vwd3@yPjuJX&D$K>UODi8i>nT8yQdOs
z%Yf#8;e4Uv{V{QIGGRP!3)pqY!_ag)chwJ|1kor2-{QPfvvA~kt<kP=L3FD4F7BCz
znrNmH-_@=4=K4AC?l^)YwQtZa);=8<AFsA{jr#ZPQC9t=GTq{IAhv-<K1{@aijS%~
z^fM*6^f%$d&;YT7>>c|O&1q8}o@m}{2M&tm1m}tLjjp?*fWB2YmVq;X8qc6i3$zwx
znV}R$e*S<VZN49xjUOhTH#k<~=sW2cMtnazu0bd9^?1)2pX75;(#X0sZ3(#6QVE@o
z*lTfLBCYw-UY~jGpukqM#=3+j<gEzx4(cyTu?_w_LKcmtfGffAAY(FN<sg9%@D|px
zrltl;@PJAaaOw-RpLlsGearD<B!XK+gd4cS_xSTr{gXe4x~R2{&0Z+hg0ABjukFSt
zoY=yl3dnpg`S@YDEVEs$_4ODe!|y07)&1xVI5)Mnjlvi$;HKDq&B8*#$Dr>mj(c9$
z54m?074Gzds0F+U7122clGt=U=!DQc3o1+ObQpktwy_~$&FKhX=vi&9?4>5PdjzWR
z#mL1uj&MB{M_n2op-$e8D6RruMhcIRUc=%GROmi9zF94?{@K$p`fpsZb`0qGIC%f!
z;Lh-$ckWEGo@9z#4-XAC5R8iY8T5zXsj&8qMyd42Se0}l{KhDnHoPN|?L%k{Rv~b!
zJ-c_4A}aD<P;B7wM`&!o*LoY`rqZLC=ZKT}p&&AQ@;s<yT?ip{H8kEh93p_2gkHo8
z5{blagM906S{@e3Z?&L{+)hy|*r*mCaKV0i@g~RKm_P?L%E9N|^lrJi({N5lF_cbM
zmkzD@3=F6kPQ$9-(ENnd5Doui_#u>8*IB;CgG{C?jEU)iyO4WCZ)Rj9Z9t(hg9#R3
zZ+XzP&Kkt}o`~V;;D^nfKV5TG7)%$;fjaN)q227%gKlUeAiz^GN1%f+fT=D5e{MY2
zJXFtH`(Vhn=vj#{NB)Z!Njm<tBD+jec|eH;nZ^2oM|yHH(<EC{?v>BM5?>3Y(aJGP
zLZR-G2w#NJnL<aWf?fdje)MRw;|A%HZR`4HWPL7Vm2R2hG>UowL|R?y4|**ylp<<4
zS8S2v^=~H4`B0swP21F>W~i%%#W(F0_S>`-Lzsa8FXQN=gdShZ%wt%Bf)qh#&n~Mn
zk;rlA&s8I%Omuof5~cP}JLovzRtD=D6?X~By862KK+b-r{OQK`!YZ3FOul&{eFBZS
zi@?GMqF$p3tv)1Y`}pIV414$O8xGPdd3<dGN6-={H==9+rF!T+VxZ-#RVwObqmMw8
zls{<PZ4yut5Do0@b!dI#++{Js?oL@}WRvjf)hkS%?D2Yys7b>WOY!vAVZkRSdw1#&
zHeD3(!{fwABxw9g6J$(ZpU}Vv*1t^#Su9Nc9&T@!xpAY3t$_LwRZwm^7jWp!rn0cJ
zA3Sb-LVZtZ>`!C2^t@Oq6*y9FDPv2W|7bG_f*uQAs}%zB=3pE(wva`1eQb362KcqY
z_=x$zht@URwQ>EP<?ZEoZp%<kZXAIZLD&>;Xq_8_-o1wsO-?8&Dq3txL`I2D&CtwD
zxE7T2E)Sf*0m*(d;3SnE@fj3nnr=>)k!V@wSBfdGp>u(nXT<kkNVZV%D_N+xjE6O~
z>P18kU3`=SMqq7U7TcsfEw<G(;u5ZKjWn1~&=3x<HVX_0$VRI(o2!M{v6@@N`QYgH
zM{xN+6Wt0%bhBM>L$`?ygC?tY7r=4^S+a$DPTDev-W(Yi8!H8=(sHAD3f1X|#2+B|
zpz2~B+&>Z^;L>k|j&``qpbN7gsUb3#^?6R!mSwilF;>x*A+V+Rh=4#&UY^VR6q|`y
zzY0gcEQ*k~I{yS}7#T4Y)F0_IZ*;)3R27VpziLx4kDT`2MqK#ag*OD$@rXIp-VhSi
zdobd7NK*f}fCzT}4DKDLl6lz~0Nfk1tMTRM7#JAJsb7mUZObMWZ(qc-kBISmcab8G
z)c?WWn}=iBzHOr_l_-@+84I^DR3tJLArvx;D5;Q$M26_8Bx7co=aOVhWGqAGAq`X_
zB1A|c^S7U#_x-l@&)T+kz3W?R{jomV@7MD@;l8i?x{mW0_G3Tx1Dc&e*4qa`^Pupa
zx_o?e?^_V781-iHDj@M#biVoWB~G^U2aYp9UUTM_*4AZkC^|obW@#b$EDTyuAL-U&
zPv`OLPo=b>ff^dRZSek0!#x-HN>T@-)3fpXdrQlxj}X}c49xlmz(1jIvbwXKA!Pdy
zH%)SBsXGdaQPCE(IwTt*z?;V^@LSCY71~9+(yN2?Zr_m9!P_;%@U7myF5U9<>C^Mx
zv7<kKvOZm<W?^F^gJm#3%;p4Z8PJ`9X6u86&u9SrV%11)DMjH|S#Y3j<qZF-c>CK+
zdNcSXL4TjxG_@$A8Hv;Lcv^k*NPQ*{Npw~g$i|4(`=)>b!J(;P<TGv?|KUwW=Swf?
za`4^S1&?cHMc*Nb7-xR=?5oc7mn~K3z44sTaFJ;F?V+W$WzcyVOb)mN3mc&FvGwIl
zURVFtfmT5o<KONFTNg!X(9$o3fyB{#j`@$O*{Dt=jkg}z0z*HqSqDkWDwH8CNnRN}
zWxb;^n5vE``-P}88W7;#_r0`tCdzRoao>bq3jD~_@893t!~uHm-u=&l8xHd5YMc_D
zH55dpudE5`TdGkqQC@$1+-nsecH5YTVH13q@8N;|{t4kqZa!<@XA|km@C&9D9;c&e
zYs`OvEw(m~noZ9ALA2xOz(CD)$Lf1qEkwS-#Q|4~xoVrKEqUnB{z)`Uc@h2vtux=^
z17m=Eq$Kj6e_mmZz4P~#i_OvRF%~dKRfg^eHZup_kbnL1W#W;lAykf_<NN0<AwWV<
z+@3T9G81v0NYMZnOl^J`Y1dTIhR5iXxBDv$ghfVTX>%cp1H}FOw~AY-31WkEo}h)#
zk?s3leWmGDh`ay;w(Zz%0B}FH_dCdTn4TqV;@mRCtk{lp>Lb+D)DY_zlfXCYB9sv5
zYZ@ASb1lxJ4-98MydfGFOK26{gy7_%nT{vIaNm%iFjYTBM$iNd@}z6Z_W-xVEio4(
z+ts%{SimN{f|38p2r#Ws`lvvmV~>3{nrL{_lgoKJiy`T)7criT3~g4s2Sg2Z|Lg7R
z@2Y!FhW{3~z!ID6Y|X%fjPz4FZ(mT0I}E;pgYWK^+q1{tvMzXXih56FoHt_hv8!*x
zcHP5=f}Kvc@>0N3o{M=ia0y2W;6h!@B5{)?PQpEhngUwiya~=#xV@V1Gn#XIu&Nk^
z)SoXn$X_};AMBee%FG;v9!{C#FvFr-WN`34P`jNrp%T5^mMVf<E4FXn+2iN2^#kk4
zy76eGLfueE64#3Zvs#D`wQvx7Vgk0Ph_7<hKv{u|;1q_HVYQvAvq+GUk}d+1%2WIa
z0sJ$MG;Qu}&+wSJfR-g07Y58|2kQ7=5H*Gs4Uy?dMbew^8Q*=vEV}&DwwkW4Sl9B=
z5;Mjhm7(UAmUYh2BqLdec|C6gAmm)P#MbHR=`n5Da(GeQEH$`qXb7_p3R^IHpbTa1
z!oq_k$Bm3WqU_;$e#h5*aB$E-PcL_Ur&>0Mm04LD=B_HP78Sh?5YPDF&sM;<*EI9_
zbBqx@Zuh1lg6U|8IkZflP*?w)b9T*aA(CX&KPfL!XhK{_i9Qzl;>sa1dDtvMTTAO@
zxdhGT&5vHcF9sL(cf?kcH7V#qamu=4f0~_3M%vU!haC<y+Bc8K14(<%`wfaJWYS0n
z#xPZpk`fvg=J@sHsusFNB)<oy8jc|j|AsT(*%5?8iKOl~=qk__LN0tjTpScf@aHgq
zf+x{f9(#!z{tfro5JVZAJHl2#PfAg-3V9>3hZhcBS0(NO?{6}AzMo%IlnfQ4oFDG%
zYp?M1WMz%P(*@>+)Pu0kkojCN@nWFf&ESL9Si7ZEErwRrc_R~Jpw|xmE=q(tvlv1^
zXD<Oy2>h2+yN|i&ysN7ye0=$QOU9>8InW!PJ5vY~Hk|9?<t%~f5XMeDumYyP0*#7G
zpPmilh^%!|Ra1i;MzRJqbd36qW5PTR_K{I}lt&kH!MH<)Mq3!_u`G!pqB#q86x-lX
zv$3(&<pv=JHPqMlZe*6jHaZl548{twYkizYD1v+TfEyPgSsHgp>-IETdIh8pUOHDv
z3ki*2Y&xv`6K*e5j_k99cQP^_bsD6(wPWuYs(HT(jP38Q(Zc8e>MiWVhGXY)2kKvp
zp^VpBb6_%An_LDi-(OLHeF?K*QM@UHD!jl12U+WmX4(AG;=Jq6v?otW5a50=-<$-K
z?IQ+>=I1d^^zOq4WhEty5k!hzF58$p?+VhW5grz-Y3FB8JPrWQC?#5Oh^nhof%AY`
z0mX4VM^Noz)srNi!I7ZKz&VDh0rDP<{VN+AON)w*BXPp#Kc`v7Mbn3-kxkQpr*5g+
zTU1-1SLg^i1)&DPiO0O(Fs6{w(*A&T00*&8UzeArvZ={`Uzb`m{YyCej|aWWQS%Xw
zPFLg?f<NpUWZ#OM!}5x`$Ge&}#!y>&8AyB9#^yLjBOFNf#_-Ix4<Gc4BwSCu2b1}q
zjJjYS#2+9kXTs2uNah>;Ju&fBX8VorWGo7-qelK9l+#&3%A0ogx&<PQOUo#B91fYd
z$~cV_6Am}?RDqzBR8)2oLe=9s=0Ww0bmmx(c`|*CGSGiD+8zD5Ln$^OLhq<S92y!w
z1$Z}J?nZ-0-_$P_<OhQ`du1b^`1sVL%{d!?3Qt-~s}00xoU;~>u(>V<f?d@c_)GzO
zef4+xGWKO$XKp!i5h+lw(4MZX7d^Kg17gM8seTwn59qX$DK8}ph8XMpaP#I=B6aI;
zj)xWg)Qh?`|NC*7c5LuVI;a@5ajOVMMF56O=%r5(9^1~l(!xf0M-vs$KbjW7l~0P3
z)r@~G^BmcNnl7T(F!pyBu$|P8&VVfZ*^!rvDzf&UC38O3nUE6M8fc1Xu!@4rtSm?k
z+ya~ON(5np%O+n&A!ih#tD}Rbd?_y#y(E}WEx*>ulSdw&d6=NjA@9+Y<;x*%1Ulk`
znD$pIOSlQ$C^Qtn(%nl7vkh*v($+kSygXKqNM=~9LJ=obVFIlm%x#g1wXyNr9t@)u
zf(V{5fX{nAGRc(EqB5L5_75hF=1=}O|E|5g7-(Y2VG|QJ7v@-3G8|FOuc%3Vr2JBM
zcQ>RK{LL}^qnoe9^#;lqlB~xovd>+-$jr&vI`a;v9u*)8ZfJ|EzE+&dm+0}yHicK~
zyy)cMn@Z*S`YLVDxD@wNoj#mv_mj-b<KJ1mV=E$JXWuR^E=qlFT3X9X!-a!x`Y<m~
zJO1;I*w*N)<^xbByVFCzYD(^T7{cAFF{Ohhbn?Dq&B)<^?-}$sF;LrsoU)U?_M>z7
zb>H3CVI$_WfEZ<AdAo^v(jB{>#4o5*$sP0`c!h>c*cUFUE$74!>3|snSe6}o-XU2=
z5PlBXVuB`|Kaa-(xjLExjpE>7SkJv(G*F4b8W`%eYqwJW0xJp`pJt3iGiE6@HOEo-
z^lQz)nV`hM&?A5}8u5oH*WVpF3Ex7Cbpwc7d-o>3_L7p8R+sM=J9rS$hRj6%e3Rl0
z${BzK$mU}z2ZCRSJ{XO|o(*NDCCp=_X6mmKyU>JGkS77??Guf786w?|Dvo8xjv&!D
z7*?q}zyUu{|Kcb^6=v7K$c@1KBeo1cHE?*?%G^98GSV5nJEYv9exI<*1d}ox>y&UO
zPbw&^eo=|$K?}+O<lRJM46qOjD=V;W@Dt##<Z+@9JS-x**PR~poy4ZQz26<ytzR#8
zx2>THWhy|ceKsdpz<<Iv8>U4u?oB9gK%=GWf2Nm+;`~usS{y=ZxirS6UR#~Lfk9R_
zc@Ouxpxq6&_kt+u5#;OzxVJ&i<Kd%6>P;>uPz}f{zfQfYZ*1JSR1OCzet|VigNFgu
z*3!`tDWim}6n!IE%pF4hiAZuF?Ldzb@gD$P3nswg<28O-+SzeJcs*w#ySR7`A`Hjq
zxu0faz(6O_YMh=%TM@p@6#{mc*a9L291IE^Dz2T8KQSUYgmTuU!h#IMQ7Ay^_DV=R
zRGvHq>h@K|dQTQ4HJ`zyn$1LcaoNTO5_#6=&T%vNfXKEDxk~qmncu(t0s_v92C5A9
z_8$4}@i;yGinDWq#O3I;G|GXmr%=R-RmdcuIE6FEv;U0n|K?3DGGg3O_*&5XF||g>
z?CGVn7^-yl<KnNzA*d^@?>Ck~IgjX2c&Q`M-3OvwY9d$KF*^p+YdJ5q<9llYiu7sF
zlVJxM(c<+8NiSdS!q_N3|Gjb>hr@@dq~`wliD(FlVE^Z>tgOiBu-@Hs^Jd65sdICu
z;1)r{51th>u3aV0_Zxn^ZG)}syYlt0vW>W;<bwuTN(HQL2^te<F8%q0#bh%;fzVs5
zq}!15qNs=s+dE84>Xg;hM-TvjTI{!2*YWo4xoox+T7t*3ur$!uWYDKM+F|ws_1e(z
zFxfh_XbO-pDf{;u8U_Z2p^J*Qc7Z^3o>dmH3L_9(%$H?UQZQG9?4D-xOYyyXQHXDm
zG(r3Pu&c?+@H0*h3KPmNDTlzgB$|!FXVW|WyhOy56gv5pt{f}AdL=N!LA`$c+i^V_
ze`TT~f{OO=9j{$9moAZgU3AZ0AX!8!ygj>VMaYjr%NekQbtTH7;^N|+-@Xt1IKUDo
zbLVCAj`BA{=D9^^KgrG*-OSGyb8}dy)SE(hxBvn0o<FV`VM1vZ&BK;&g6?~Dfn!#0
zq+-ON21sDNw5;Yu5mO^=1Z#`#J-KNg3kWMf`mT!Tzp?@lwEA1LSG^2NT$vN5nN>jt
zwP_BDa!^SL1<XhmxZh|I!fb?9cQ)p}om%V<L74@73sm;n#-OA5xp-Lc!+2BF1gFIv
z0b=$V6KqCQRP@nZ-d_RA_!s87x6K@h!iWE+CBE(G7-Wq|Pk%E6#s!Yyc3WkfX2SO@
z51p`_LLPYRcnDTY_aIB7c}Dh)&P^mlN3+c9Y5%nZY3evVU1EM=A#-<{Df#@rU>el%
z0)s_1+6X#jA5^0DpGTKX>Y^U*r3M&MX9j7vB?34cA8+j#(fMmQj&Z?Z=_dd&fR$A;
z36^?sD$U%Tni|#-FktCuZwCy;>bCA_m<2~bi>bpoXq4s^9J#O~0MsJ#$E_f}07cBI
zjCGeugI?RBkpX#2VE!1eIlKu?1=`J<IgtturU0eE>|}YE`boGY&E4GeIP=4o<R^3U
z^GlE+qY*J6pPU1D4!x^M;j>^KUx5N$%d<)M*tj?{yYfOI@fgn>W`VKa7{Jt8Yb_le
zR1u0;7CK>%ZWLj{;s}#fYIEP~N8v}#m?VJba!jl)%hWq@Z0MaI+7n3r&}sl=N{QWx
z(@y4Ukh5Ds?lZGF3<C?Fa3OHq;6n<Q?!Afo1XKj#syNLT?Ce^N?*10ki02sxc^|7*
zK=rVK^P~0!1;Gl*02H#X+1qc;(&`oFq3HvtfaxhTu(tA_v#}wo$FS+MMX7g3X8@vn
zh<tD{>UpGO1l2Ch`6njkGYS)o2MDr_+0cvzy~RM1?$sf4MU3oy#smc<E8#WPmOnrh
zjfaKfB5&V!M|?aIv&P8AKp&IZcG%e17<1aeQJIuzY*9f&;>Y}0ohJ*+-ZL)|?j#=_
z0>_WASa^8ui1{779T?bx_XGuB<i@wz2hq`L$DOl4+JsQ-N~6iXQB_fqOdumw8F0S@
z1H_0RL~$r5?7SJaY`KCgqb2PTZhamm9TDE>9x>W|$oz-~CRvKMAZ>Nw0+@aiNXPGb
zgdZcCy`}c}fFWn^6x_r<uwvl5`$sJ=TyR1K4VI{?$<TQK{V0pceC+pg*@sc&%L2{9
zO)4uZ^IG!(D^hDPxdTmI<T<T{lTRx(n~@cv&jJSXyG?=;!N;)dnt_GI6}wP+<&+hl
zO{5HU6Z&!jR+S*rM=1Rr{-&)BBGMQd??oA#TJcX3KfLV)8|>at<}O5Mhl)SLZPhsR
zv2}Hi(l~{nU2Tds^s!*ms(0_gPAQ#^{<Rh049WvHTsKe?tk8{u+=8WLC(a%)3N<OF
zPhDLLAVs5xg4zQvmLK)gfdd#H&Bul-q`>w2VLQ8xDHCj)Hf}@@0z+u|kPyJYSpWC$
z5#iy%!&)@Ag5;*(0YS!Xhk!0f`S?P1j^KbVBs-w^z%)>1LXMNxj9PGdZNmQZz8J<u
zl8zd-2lzQqW~_36$=?SVlUDJW^?Tn8mA=q|{Cvd6fP44;@*8tPW#nI703wS}>Jple
zU}tUp2}BIsF0e$7TsQO0&%b)*$`z<3!t8;|KsEp>ehB$k1!s%VN2EH?9GuviELTW<
zr0zvI|9MQ&`s*D=61Qhd_Pe)lQ%uEuCj?#4e8zsw!B{(yeK-61`yKI9v9YmWN`YDo
zrX>I;T?2!Im?Q(A2E-18(E(N}b&`*~_RVwGIW&|&>*KZIR?hS1OUNxf(4$IEPse1(
zq@R|F30O!yXma8ULl02!P<5~9OzA$#^#*tF&J4o^YwI~u$93lH#=I(uN(A_N$Yx|_
zic3iFdhP>|pK3Jmy!TZ4)2GDbFc^s%cob<Sc$k0zkca>Q!DELK7FouQ+YqLH8Y7Oi
zwY6drS&%M$`SNhQvNpy%Nec-N^3I(*JlK2-5vPcVkQ1QWA_a+*XAZ?I>-;yqeYf4u
zpQm8i;^W5xpMO5lR7Sk0x8Z|8T5oSJ-Vd{E0PXO8xG_+T0H!y3KGqAT7bI#s%ot)t
zHck?Ti2+;KbYy*jlten~XZ-v-Y5pyDPJwWNJ;6k{f|(grbP3Gm2Xly6zgoz$IM%?+
zPx^a8?E#Mvv$hC=K_3g#&AtJo3-uIx&Ho0H>Ia}H2nPm`1bCba!15@YxLq{+UloVl
z-OL$}NFNcgot1SS9tn;pG|n3V2~({g-vQdC=a4a!H2i?gi~FAjREs8hd99rtoA7#u
z$qL+>TsT6Ybx>j6#m5IUB?Z0%N&h0+6ptTa(naY5cXQ~UU%x=WZUQ3e1|IU{0q|T4
zuricr)GknCR3v~?z8gC#arkh@EbIbg8bN{qv<K*PdibsPxO0fdhwbr={Icxqamew+
zE-bOjxf`CLJ`6CeQ{Hy<nro<^@cCvV78%Rw(jGpXMu2rkl*O<g@HO~`vmlQ`(ge;G
zsXM~W{M4`_uPa0Pon16251e9VHKN}=x^;4m*#ME;fYLBn(a^BQXM~;L>EIcH)r&4P
z68PTl-!Fh3v@fec?m!F*<*2w;L>_P#Q0c(iVFt|wTOsgnXthGqYegIZ+=)JuhedBr
zrGx3@`~3Md4l>B0?g%r4^!?}&EDKeF1p0fpfk^%UEru@D^jDz-lYEGwTi^2U+mCEP
zvfkcW28<6Q89N?lb)|{~9G`QfBpyml;9JT)cfV8Z*fA}q6gg=SZqJp{A%)R*N?G#$
zQ8)oV>%aXI%+SL;^mVO9&ICH+0%??tpmzg0le4jR<#zg~iZs}2tikM+>^J3;2!y~y
z!oOs-(z~r50gjb2s=T=Qy&d78veZGY`Q`OKM3u`67j`|SyO~G+pjFUjzImU@?MNvO
z4SyOXK}>nF4c<z~nkOvT-BB2x4_mh6<-MzU{Y5hoyitFJz28CgmkqJ11%0J*3L55d
zyH>O4Vs3V^Wz&%@uxeJZ3;ubZsvgb>^T^#DJD~{1l@!!{Zm7fWdNDUcdg<ofyL%74
zR5^oSgS(buhi_lk*N@2i^m``~r0Y{dTW>THoPNK8xZHcY-AA?%Nr1&3#5D)rkfrJ_
z<d>)bri95%p}`!KYj*CzRBA3wGp;@S{exB=^lxR)#e?~VqZi)EA@Sa>SLIEa6%9NJ
zYLP@csL;htXcN`fwl;8Y|4QOm+>Y1B)X^mN7o0SAlRF5zNd3UGi^d$-DOnI{?$)VW
z;e`>Lsp=v0^Do_b@2~d?u-@_ILKvFAF!ZF70K)m3#UM-|X@p!+!tU42-%z&EW{Y+G
z#V@1dVMPP%2B%8;hB{zLIIAa_kaDA<8h)Fgwt2!TCqEw(x&uhi9t)GGB_tGjq245j
zbMesIt}b^_Hq~r4^5Me(RI4`MH(Q9>4mX0SU}Lr*xCj{Bp7N?GpC6M3(}E(gfl7`O
z&^nq-8sJ10P@n=xdX6^j7ri{JIiMs1=MmnyLxa<RPA}i>?H2UhZ7*F?@2R|(K`?1;
z><F27^O8oNf)Tf5t|S;JxHG!8EB68JQI`%m!AH75j}Sm4j@YqBWVYmVzy$!rdq2<k
zR}@BhZPBs<vH~@~A3p%V7&!Soz>TQJXt)&(@{yWOvU-K-4m6r?S^)FXcG||HwYYs9
zIvLc|{3T5IX27eqw!M4y9RK)qOY*-KMIM{G+!p0!&;Q5_wj4z+5dLL*kUl}<8XGs1
zV&O{SM#)V#9FfRi>j@g~TDIr#YmlbozO{F5`SyDjd1b&j4YsgG_wp+&b3=xwPI0v|
zto%&f;S(RpoBZdO*6IKMisCKFPmhW1%@DjyHL>N`ORZGrM%wf<tJZF}xq0lQfG<Dg
z=FlhB>AWZDMbpK7r?#?YB_?D&F*m9yaMsm5qqHh*hQ<2$tVjRn*$WrHYw*^mQnmgT
zlA7td+7qQZ7CYa4wUwlMt5qSdiLb-Etp1fEq~QtpSE2Bq)PIEwcLj<6t58fG@~_O{
z&BN{gN)c{3#`3R1;eYEzpwD0o%)Pxm_3NrLD*M3{DjL8}?yp#ut=001iYnX$d99K+
z{zYlS+aaVSs+ZZQ4l5ZUyu&>U9U^+w>8bwwgfC6?U%4h-;aATw5xU2uD5pzd6;*ma
zRC;h=OGNW5FtrI9L`M2`D1?dwk4sjhy6F(kA7Bc}UD@f*lz)F7-e{X%JLUj774Vi$
zyuAOeCHt?u_y1cr^?%8e`QI<{|B%W5pWfI+YbKne!{>7W1|X18(sQ7?n;wERkT68k
zJz(*hWhP#}qh3$-Qll4P(g?d#ZqcIR4_LT{uuo)QZ_SZKQ)t>+-&(U;7y2{lLp!jP
z!3&Xl3`?_to<Fv}0^M133N*-8tWoN*vYBqw1u|Zn$|ox69>qGS(XvvTN4;B1^>ta?
zu6v=5`~LB>w}D-pIJ@%m7+j8yh{*Fd0e%b(hscZ%H+PYbv}}iR^09RWDvw%rXOoOm
zYpBi)w&9#W)*>yVV4H^IZVZ3xJzaH#xRj+H<m=c{tlj<&L~V2~<3k=&QFTV8UKt`1
zZyC|68G_(w{(2G$D!3tP(Q3W#(KM0{)zOfmP`(M1ZLs*Y_QExcs&GJGD^htDzkuO^
zd?T_Px6w%sw`&9~m*LM+Snq(YsCo~*77zT}&1Hu^A3h9$=Z6JsWIu$GFhjs+J=M)P
zO|`qW58kY82!N}5>G>GT;-M9;_VFHBs&(w^ooyQFSm<JPs`K1^*+9v@d%87iczEF^
zw>%9~oq!ZIAP@mPxU8Jt^-WD|l2^`HovMSn8x}`r^sQnD#izOFG@4s00ra%G<o=F*
zl@x?HpKynvyS1E2GQ<NPZLF)498L{YE1d_;UuVZgeBLxr79y?>y#{#!v3364>oq*P
z$eXJ8sLZfA)a}MoP`|tv57HftTTQvy{*a&25Po~N?msS01;vjsfc9JLvZv{&4u`09
zU7PyL9VSL@ge3Z1cQ@|T?kd9~Qtex1cmN03uf02o0joV@rUctTerHs@hB}yBd^Yv%
z+kuCEq?b*9da~!%Dm-13KuBV8`o;!+sqlLeBlrcHpbNI&2tzl@3NS)MpVKaY+Zs{>
zVf$&<(QuI~PdB`M8)>XlyDDG!>6I&|mz`=e7uM=Q?oHanr)3VCkQe*M_kd(NxOK?2
zL`u%!;sls3*|Xo5SMVaHLGuiR4o=(wRedMbHB^T;K%#-GJ{k}nmc*Dxty@z~!Mg<=
z3#tEwX%m=VQs!-P(WHZkF<XjGK03#+F42+eHx0?#`H&Y7emW%b7JWUkCxreNd6=sI
zKA_}dctAYuurTiXGI@2@HK+U`5ypiXrDt_XtbmrrP^QmTuQ^KY6~97oiU;D8GoE*Z
z=E`RTfc*y5Wc4)-EEK{8P5P|AM{bAK9kkCdSILvU7aULgPNXr0HAdZi*^199GpDaa
zS*3JbD$1U1x>v|26^L_a5ZIu6fy5{PNCP`JZy`q@(Q*UB0A?9SP{q2eTSN7-<q^&z
z8ju@4xi$zqLctVyQReW<2}q8+-sgi61^*H&_ySn9@Zfda-&vj=#l62&(pPSW(vyD=
zK%cDxc~iC6E0r4%9({#-{=@z!@2{sJR((vZ%h3h350X&brT?S-+qmE)W3^9%Jjbp~
zHB^L3wJ+(d0H1<n=AncwN&}A)6VCwbhF1#qBF!j3U`k%p@NO5t&PQG6BWTlAl<V^*
zDf7#<j?)`B4rJ-aC*_D?((ru<MO_flKm806VSF*ML#IeOAIxKYN1B}9pFtD7_4I#-
z=KoIm_HP)XI-Cc`_#Sv(=OoZ5;9l8NKRI3cgWqQ;nNgsu3W1Z3G&?q`o9tmEH4MtU
zO@D~AFpn5lEKYxhMnRYr6%5{Eyllq2ah&W&%qmKr`icEX*U%ZhFSr1><J<SKRaBiy
z<Y-Z;KGuKw3Axy3Kpcs*WdF|7!tPV?yRj-*YVSev4q?E&<c`qGGSZSZX%}43O$Tlq
zAu67<x;?W<;@mpVdWVzare#ZF%*|x7c?VrkC4I47Sc4FEpIoTcGZbtWEi4#~E`vFP
zD8laz+fVFTxd<h7hX*bwE<p-5`~X`%?6Q*X`nCbrAw+4T7h%7PL4zlYA^{1bOtixR
zvWoV7a<=L&Srdjjh<IBVXjtK996=S59P=w^U0A@X1YhA%`~rNc`jxOX2<2E}Y?NzW
zg9!_=>lSeyJOyizS)AcP0qz;xTE~6BvhC}`7+Y*QN<fGq-259@fB-pwZm?xAXy6KU
zvWC~ViS<l}eSm;JN?yLcii&?TKuP>i;Lj~BP&LKE=)lhgrFaj&Yxm%KyMgn7IB+Dc
z73d>CPEc5ZbMChF!tIwnGLKk)y1-iRR?AOy(K532MvMO$Qj*T6h^C_NrD>=RryV?7
zd58LoQVO}tM@cG9^5IFso1b{K9z#e0Uvx8FPG{&poof<nHF5h{0OhqK*ASrqde7m<
z4B&ERBk1CsyIn{oA|v0V$`?0<In_v;EwVVsvZM~#6<j3s3&5{{&cpt}!3tz2xyG9h
z4J9gWmKCp`CME?rV2c$!v3R#J)x3`i0HR!R0Z$OkESfP(S_Tb9%CWa=sB+#CPervF
z_Iwc(#@qqys2K+Kt&r4)J&nxHwLI%n3F-yrLVYb+Xx1-u(590-AXQ#()1fxS%}{N;
zF!>X&<VS)J)oS8{OMX5#wwVR*h0gQuk0;sRTjE%acMAy#z2Xr*3oP05miP=XD>+o-
z@OVp@#8p4kzw8WGcEGkv(sBeq4H1dNEpbKmMzO?MI_*tNbPNAj;w13c<=KSg4a<kH
zY@+iJ4wMWOakd@wcz03-J1phK9%8nXaF%g2<wtX>J^`u3(fMp!f==A7**dsSmrquY
zYQ8+f6UIny87vfM59NyvggNkg<S+VJ*HfM82stx*@>n8zH%yDQ$Pqq%!md5En9r5m
z!p<^%ivk$`>!16U7RNm`yO!Ynw>#IOZ(hx=<k5VfgZhsKDWK9DfcM#7`VK;8+|KX!
zxC84zaV%+sN&3yUVCCTLsJ&+uv<&hS6Nyd))vj_Z;yW<rFVws9H0|Pi&+-!LF!Lu+
zM6Q4j+3-moE0(@RAfx>Q>WjZ{#Yk2O8p{d%eLO6lV=o5f7AC4XCz0-0fZD4nft4a6
zO5~*7OHm6Kd=&3$Z8M9DHpQ}#^vd*V*CcceaO&3=n4zl$j+*>)88Do@m`(4MsGFL0
z17S<xAEF_1%f=kfc&+<RAObh6xDZK_@$@W?Ip+?L=QJ7}3wss0chKIYZL~j0D;*h<
zy>~GL7l>kwWp@9dffX`Kd*i(>SR-n`QdZknpEeO^+&ke8&>CiEW#a;*d$Z)%n?*m7
z!HN>mvc_#k3(RYLsg3<?ps47H+~Y_I1|g@3$>iNYGRyv65OzV8*pAdvW*|(*yb+Uo
z0LZ!^A=!BX7?VujJ#O<xJk=ZT^$cnv$P&b{q-}h8FpBsymr^l%-Y3C9>-8_cFH{El
z>LoNQXY3PreON`qZfuHHB@p0z!xI5)Qj_;R7tR54*PR8^<lBe)Ze2Mr9Uhl1<YEAg
z4`$@x(33M=KLvk+Pb^kH4JUN%0m=rdd)viR)Hgt7DicR4cK~rcz}OQ(LA~u%-_tZ3
zPk6Tzc^JTS5+G^UOHZpe$u;|G*+MJVed8S<dJ-_q&Pia(m{QuEj_pUT6qhzI-kA%%
zR_ryB<V{oykr48R=bBXO#R}Oj(2Y`Pv5b)jNEZDe%fpKXP}bXFh9y+SvWudrnAEnG
z*+dJ#K_^yJ_yb_rTTdKk+KS^}rg(`JGiIDjd5yZnyz%x!Ba9#Sg8;#)*9kp0X4C7t
z$10QT*B<Cb`Uq~Y@>A~U_;@+YXoVKzz(!J(;R%;#V8{^Do4GIC+S0Q0J;u(rs|;?)
zo5R)8X4dLaoCMrvdL`NJ_QBZh%jLo^<|Fc0k3&|;pjiojggGNZWcgrp5ur=|1@l%~
zF9mb~{sIKX-?Nlhfhvtd94%?zhx!Fsdpp{)@(aD2?ai<O`Hn)8)~c{$;Oh#`-Djq{
z`OXc_@SD>z=4KL_#vmyP?BDmZ0{%1~=|LnB7>?`4eaj6<KyD{sMYM6#ZWsd;>B5r1
zphQJWZ7e7Ix{T~}Bdf%ov4?QPJZ~#89At`fSrnro`{!L}B~GR-`dIZU56}~FA22X=
z8Nk6hR_+1w!Pn%U{~6(zFTT9F%AQZd2&phUOoRnD&|CMeNjyir$J&&PE9f&I_CP=H
z-2Rp2SMbKFcTWzL{%(npx_TibT(oSe9Su>YC*(-1T``{J4W~?|@P7RK{KxWw4|ZSY
z#yNcMj^&|0FaU2e3;%<@MLr%vbu;vGzoU3t?DG>1*_hG>A_&lEDTuySzX#<+1S6^I
zIj$!f2K-MakK36J=Sj-ht`n*)L)w-|OUYi+t`kSB<}e_ijJ6GQZhT@gL2PpyPKzX!
z3|1oX&+e@x%ZX5Rv)Yx~12(o7I-Vehmg1PI#uD*X>8F$kWTi_Bbb6^jRMWf>*t~Jq
z0SPFybgl$bjHTfv4vZ~)L)UNzG{Bh*OzbD?`8XZhjv&i#Y;K0X8b+Irw1hz#VAMyC
z{4H{jEhY1)QW&c7IK*tbirnBXQM*baL=eIi>>o4d(3xC@_$R1VPDS6z+8^Za>0bt$
zqICMz*HN9})2GZ{6x*$hiWBk?KYarM(!%>ZgIQd`&Y*Rd50Ew7`1)Ur29igLum5|H
z;{Se;{|ms?O0q8<qew4hfpPVfDk?|)`t-+(n51Pk9;>9Lx;OItvn}R%1jbC;hf+It
zd$T!w3pl9ni0b#HNG}!=(y67C>|I4={soJ%D92J_NC?2!e+v%&|MuCLH!~~$Cl}zq
zYTLWkKGaMw9gtu8b>CChQBt}B^ap)`2AoSjn~NBvI_XB#F9x>AYnh{Uh1R~$LZNJb
zkRtLcp+i|({EQG0yJ&F+ZVf~Qb*1%ddG>>igM6iKxeElIqPjBlJWVX*`{B8Sz5ga9
zW1i~Cgyv$1y2#@pLnozTwwx*#EMiYCrbW<mP+G5Nvyv+azJ?bWgX^pd0Xu?pos;`^
z58ztmkOIO*S;A){LW4hNg{uTflwNu?XVsc@pb~RK0>GTx1YQ~!@b`@hXERp1z~!LK
z23~@_<@Nup=a4+}^5(ZdX<O8C*_||W1K+=Uz-qAMiMSJ@IgL^r$RDZodXQohN&w(G
zlo8k>1xN%b4lDi+tj{V60^D^hA8W;C#>ox4>d2dRyNxSFnB4gDeqj2!)(70Hp6lNN
zNLhqL_H>>F5k`Z@`h)@q>~1WOy^6>}qO4MF(VEyPTHIzU{Eu7>QYfqz0j0^{u0j%$
zmI7pKfEmE4l>CvN(%#-+^-Kr;8FZ*WaKm@iIb4U5VAS(B*U<AP07b+@BeeqD_|~O{
zy5{D~=-}aQVJweaGw0Q&eJuf6EX8cnME9@^$p<D&dkxHPkWK>H(%5R359MR*oNTqP
zWj}c3>r2_0u6%RuuHA=}yae?kiLNR3(v2VzvWdJvg?3+htcdKd(4=~Nl8Vv}UQUtc
zNzcDkoDFkrC>am`y+Zyp;v{-z?+to455SFQ4P9~afGQ>b<kS+@Tr(C1XP3_jraPb<
z>f;zdWukHA9%hxQ#~gmuFh@?D4}FHWD+O-sv#fyIBnFG{^}Pb!1@MM5^@ZR3T);zH
z#4~8FuYa7r3zFt|0tqhPqqw9;@VScnL)3{;c1_zMt4ovs(e3s$h!~4s#yyygULc%r
z_z`&<rDb{$i-b)P!c9>)SzVoIu!xk1S5mD@CqXpMfvnL$grh8pV*a)R2S!M+vca=m
zC=3hEt2P}Zn74g8pvY2`7ZP^ecsp`$iZ;QL_PsJid;RJ=$`ml|n7@a?zEfLZK)6Q0
z@?-V%(fjJ9nhK6k#o|vpcqQjP=ZCOG6q|#n3L3;Dv_%%~XI?aJag=g0M@5e5LK~Rs
zwpaxil`DSgDQNP^w=x&vk`>N>8Zg9)I=n-l#C4S3e4W3rh5B9w84!;g^NJua__uPf
zo_TpX8&gGc6p3zK)#gf4tTD0EOUL7QgJ~m`*PoseNQy7}&pV+m2>42ynOWdUtLzLL
zKov5M+-FTb6Y^Ej(uqoxy^|aE*x=~^WAMbuLcYIE7i7Px^O2`NAdw&di?L0xewK*~
zZf%0=kH=_vA2+!Qbis4B(JUlt!|ryZ^J2JL4AlB~YWvQ;J{my)1}HyLECuz5pa*dO
zqY<&VFQk&S|NP>6^reR7%^}p7Q;H4`2R4#U`Vd)CI8RVmmqFWbV4Gnoc}P<=%w|Y8
zF2Hv{<w?WjDx@_cO<k=<(=M2kCIwj#f9oZ<?a(s5Sr<Y~97`Pqj_8!+fy?GR^5HTv
zSyaE5`l2yiQSl?=36AtZ<YNvSgc;|?RfY+WVq1G63Ehd3y-UQYI_}ZsIUN}Hp44Fg
zdKPC}Juy>w<CZlv$Uv@`Shb!l=qrcn&X441H5516<E};S*$@;S{vLj!E~I3vSP1$V
z)F<>9tsHO0y~59LxfV0^a?R%BTAsvHhYBzsTw`bnZ*Z{disA;RRDW!8l||O2)Hmgr
zip**vGrvXfZBQ~+9rD3}7y={-`6G|BjXWyCllKOusl3^pI>DMYRsM<XtrQFg!T$(-
z9J?56ul$b67x+p^@?0XYm%=UwbqiGx%1~mM-Fiqmki=L8eEr&qJTb-wiwC0*W7iT&
z`gr)C{Q-h%1M<U6(TA>CgD#0tD!c_$A4f8tWC{Kn%MF$0TXMi#p4ZJzL6%98o6|T0
z4-e6>zwpOwBE#Kf^KcC^xWn*Dmq8TA#lq3{!J;bfpC!8EQJ>`xuiRibOYr*sfDgoD
z<O#`fEG7Jl>byZb<Cyf(N9kHd&Qip9+SNtFZF+!MCm_WPrXZNS03xrm)j$dmc;^6o
zndr$L_W+gyEa-*QpK?QxA`rqOE)g|i3oj-cg#1c6pccb-B3gI*uj)^@6cnR4q57F+
zWMZgMq$JZ{!50s09>EVJevgmreR+){F=7`re4H_#FfH&Xmcb|wM;k8^aVRU~0jemV
zY250lT03B#Aq5h!e|RF@Y9*^LaKJUEC1m6=33j!05r|p?e=2|Ya;kwPNS&y`L`nw6
zArKuO9<JA=b14;F;-j+31eXAYmhvv?Cj)q5j$QV(UVLsi5W_e;3+T#_PY+5QFe6(q
z)!;y|S4sx+5y0ZS2b)e~!UJ8ThN+TuPWurhp1_i~OI1rZz})lm^6+-1%Da^wuXvg-
z1K2WUE6|}Qw>b3{7#4%kWJkqPL6=-@JJ74n0e?<;IWVX5EO2xvrq_i5VURLiurK*p
zR#BgC=OdjNEY2pU5Y7yhN(_!X?y0$RDI9hPCx<BujL~iM9~7Acj{<~)aZmD?fIj)f
zi7mpfT}QZX++C@_Y;xsDSC%_!k_2=Lu+Z&46mzf$hZJFe-i(0@nini1xX59(Xxhff
zXKyw91YN%V1LcYR0-VGLC99dAQ|~5-AMmqq!Cl&9RNa1XZ2V*XAeLeR8jHYk4|pFw
zr>#$M3*!)B1{l84UumcFQovSBBKl<WkXKmvGKMugm*;z!K_kJ>0fQjg(3)tZz+Qak
z0~TtLenFf_^n7eE9W(Q%TxD;)L^_e8U9l9T0mJx?Cpv!X^f0usi2vw8tD({i%ey7f
zykd}#qAp>9Uul<Lo^2%;^L*hJoycFrc3>yvNF7kF7o!S1$praE_CJmaD-I?pZ`=Vy
zRk9oi9ye`+1IoffxDnjy`k99NfY`M^Gs@DII4E7RL)~9Lvu^e)gXm2yxjp+%wU{%P
z`NX}W)PX8>u7hq7pJdxO)3a*JPTUiQ5+DhUI$g7?fSQzHIx9VzFV0chrrFr8tRRyc
zYtAkqVHiz1r^#afXm$*MtdWZ05LRj4lXZoU=Rod}cp4{Y8asDQ9qK9d&nnPEuA4~$
zEM`BStt7HDIw7G4IZ1jMSfTo*2FBKSjr+by2*f^-EkX_=ji@_q!4Y**8MzVusBTQ(
z5NH(%c6HWZ-F02cE;GLbfAr`)`&r_TSPt5njJ^Q@I^=7NGkgUp|8cmZj?W`EaOzLb
zf`6!8{OW7NH~lK_i__=Jur<~4rcTw<PQesq?u+MD)zwLI6!P&BP~>1U>VYdtzeyst
zc?PIl<e%OKg}=IDXlVKdrX{{)NMSk>dG6u?bjmuNrKjab5M~}Dr0Es_m02geH%1}6
zLPs|}vxE)(gUc3;_9bjSkvoK^uo+zC=e7P|72f6jIyDR_T6bXlT}UZWeWOIShm)Fy
zi`fQ~B8`eyon<78OZF4_fj%s<C~N@#i2DUZp|<RH@i`>TvjpjZN8>;>4qpQq4kV{v
z=n@pzF<WF`dpUe^Qx9+>1Rf3PpU(jHlidgCO~Q8Nh7xbLNDao~5}zhMpsU2jfl_o#
zfPBEVF9U@mjY9x8W-bHaAW#bGa|4C_pu~LIv2M3FLDi7O5;T~+3^E4fvrIzqv{SEH
z?eJWA^HO1}h(d+}F=ZI0QvawBhHQW6E~_oFEWqwhNsBQ)p0p3hgdXx{=PrAdVec<b
zq5Oz}8Z2nMg!~qv{|yc(3IZ8)&+9dF<`M9_&^I7s+;A_w36@~RkeH2U1Hd(7VQ|s|
zb)h*ONMyx>@ovgYujAxraEe67D)j8J$3=EpydC$j^Q7Bg?9~={RSvzsEz;wRFKms~
zx{%uEdd+mZ*}XgJ5uAZK0TL;-U-qGC_zJMu_E|X-*Wt9>+m}L4rm2M!DnZ5sT%GU{
zowTjeQm>H&vKAV|mEr^TxD3XnS?z|pV+;JVdp#x6@#m*&IGh=PYG7f&zFgs>Fn>IW
z!iKcC1vc+PP=u>wbDwyH0PTh+0F#1J`cOY-5;A@Qnuo)#tGB9*A{!wHIv`14Rht3!
z07jgE=4XKGopiQp*)(#QCpY_HH`qK%FTbZdXC$nx^_D_S`57@*?8E-!!OA%NJMKV9
zF3W5*0Lb;$Z_LFO<mQzRs$h#}9pY6?#0ub}gDD6r1Rgid`{2Zm)`7v_L4sV$G|A)w
zAnihO+iNzCtDaou6pg?nLl2G^J}C+C#Hg`@8g?g(cq0Ak&JJ8N&aeqQ_o?SLgq#Z;
z6f*v}Lw|sY5#ZAAS_Bcn0LJe$^+tuXSiE}ngMLBq_Yk%h7?PTA^<26NV2->AP7bKP
zK$J)k0k9UC2BApS+^`aueHXFE$bqm%G1Z1Unt__@!iDeX>b-|sCBQ5mM$M+jVNUl%
zYDo@(FoW7|5q?X=eURi5wH$7aKi&<)7bB{x30(XPi0=U+Fbo%FT7edN<&#fnP>Ih*
zDGC}#avwZ+(3Xqh)czb;2I=LT^g?2}oPZum9_^UCpywlg-RaYT7{rrwIOYKhRxjmm
z+`t#R<v?~L9Iv(<v#WsChmSX$^uyi7T}QtTNspI`VNS*t6zs)ugG|G%jz#GY3FV$T
zDhw4huchGrDw29oXYboz#qfuPaAB%Fore6yTKM~J`2T`=q1NO*65};eNXBmKoF6AL
z+Fgf5W6}YibgKGvGFe;nU>|q7SDkt^(lTMQazeN9|EvLDo>e~g1N%0(dJ=Fh#)A~;
zRkvt^%h7n*QPWB}1>+3+t$Wb_;+he35O?KgbR7scNS+q_XgP%h;4z$u5=R!u0ruW6
zy|K06pi?%YEfXUnQu8$cg6(5uzusI!cBxlyg=Xal4uUG1Dej70@$f@b9L%%b4;wg_
zCcHcMa@?4D(;1DF)ym3#EX5NH@$)RbVy!aU+}hAmseQ^e<2j8B6n;zd@9L{9^WO<>
zk3I(Fr;4mU<VEK3f`yiExTb5iOR<!pIpyS%0DsMURY5%oygF{dyLi@neu*g2Z$jY5
z)p;#H4OZJLzK4%*Fi?8z9{Y1n3ODyR?3p*i+W1=!%`3u*1)xMi(#30-=KvNnj>(BO
z8}xF}-QkRx4QR(3pwGzXaZcw_!i{O4P3NH6k$d1<_c4Na>-;g>mO7a*g&~f<?vE^3
z5ts3a6AX5XMtCmU>vzYF)%z<$=(cDm*)Q2?6ua$0f=8oM-e3V3rv^3pzZ5y|qZpnb
z>KKymJQu!k=3c9uQ9~)2P*hO2ryb?UmKOptmWOgt``3(VcV3Lyd$!WH<mYzh>^1a~
zX8^PxIf?sH*c%OXQt~KhFJWdtGEXMYAmK2aG(#chCCs+N=bs$?`r#Py1t=ZAjdA=D
zfqyUt$IIvQF(e+_<R9h$Xh09K98HrATqNDW)D(kwNQS1c4SoPDG2#erDc}IM0D91|
zkQUIbFJS|wEQ!A4?e6Iu;zzDjZQ3xm4mZkZ)D-R+t4xvU(cPL%LdGUEDe$D><A5RV
zz0F3zKz_!v#UojMPh|jR<k9C%I{VT#6gB4J{nP)MCTO@@ZbFmg6dn&40sPjrq^V3v
z;%&8bq;W)Z>h!hudPfd+la4p(?7VpR3Lp?679>6eYC(#d#>>x}8IeofhbSrgOuNu1
zNn3}KyhC={8Z{hrNB@>I{`xd6khZIUuSD=`i@oq{GWez2)>!@2jZ`1uNzJ`u5z}&G
zhztEShcgMU6q&&N8=X0AX3KLOg3G~QHI5qz@$>&3lwaPd*NJ9YNWu2-@V{>pmUJ6l
z(ZQ%>e>aIir8>4L?*i^7oSX4k&s;;57cdNw;-Tck;9YQ|kmUYM`ee4iOVGtX4q`bf
zE5B+LBm5iSfF8VXo-vQCT9=sS=z}pW{;>FXjzT|97x~ORJe+5>T#Qt9(0U;_ZlAAi
zLR@1%V4iGop(WM@?bSP;q)&wY_KQU6z<Qz%r+WsD?Y`R|J!SmJ3lI^~D?{S6lUw+Z
zGQarsnpK^pV;K|rXHj_&AA)mF$%+aMC0P{*=u-A7tdyX*ILI8gQs4GBMZf|9iTd(s
zjGLkfcNGXxZP{Iaae$CG-U(M*qy{9s9VVA^?CH=E4=U0*>57txnbym^;;LejRz~qE
zs6!H9q^bgWr-Z}Js~f5xRxba^_WXJHiAQ)vrOPu~Pm-wG1d9HHaT_>!*@a7KH!}Vp
zYD<q~=HI%jfI9XG<V2L5d<BzWHW)mRhnw=5&mued0Eqax>;6In9`31FhJ09`E#$gf
z(Bxvs6!Jhb{fpz<6F%#mp@`%T24b$*?OMG#O(->;CvgksnIZFZDRLsD1Ep6c^hZBn
zu<360jpTSofhe`^f$NA-(vCwDC{dVZyIx{o5K*}UjhntK87ageus%Fd?%Nl_td@o~
zJxOkJ%PITW=|J|r)k!dZh4WnllH;Ynp^~y?=^Efu_+H@?w{uSil5VQ_rU+ImM<)-`
z>FVXbi@*Wy*)`yiaHs%m$Rs)jFz80m&)!TDvEUPEs_6g~GT~=D^!>p(1k8eek<>q=
z4*A_8T{Y5(LwE@v_58GT!%+{9rJWaJw~q0|Iku~k1JwIt8`4N)$YH@Nc+@Z%PzU7P
z3&}WGiU8Ab+W8BROfrB3)&NF|t^p$;0s->EF>!HmD17_h?L<RtcBW!EE+WDS2nxZT
zSS4)g(9cAz$K{=H9AMTf#kN+N-%SL&=&aY!)Px|-6sb_q!$z@<<5mGD)hi%?s^S_d
zuZZPsf%T+?XTz6R_s{zyq;XhhhDm8A=OX(MV2j=xckaj1$bM(|)Dk2uC^=#lL=NJ~
zl5<*p4InS0b6_-V1K?s!9E)fj`iBGsn~eHO88d7|R{eTA83(s7`FsA$`V(@MiIWaP
zl-e@M;m?KRsA$juMIKU%D3pxs87Ul2*$u+P!|;7J_8Uh<kVbiUIFpN%E3#sYwr7dI
z0k?6llVelG(w}Tpb9+E!GSyY>-8MwA%)sPCZ_+(4z#xWQLQTiOQ5Xfl0^lcpK?Fv#
z6NpEbUr-U+ia-NQ`fvD=dIb7JxdU(^BX3e_;n5S^@7zcD1HcoKZN^f+T(xeI*&tp=
zO(||l;g^Ph+i0!|9;7FZ7Lu<q*oSCraT3T8031<LynT2)D#YSbAF+z5cJ-=?3l<g&
zKkYn)92hOka~J;sa5{Q#XS>61<RM<3V?ZJ8o$ohM@rsG3E;Qo^An!nT6E;E0XbiYw
zXWny*`jCt9F1!ni1voEA{g0m5VK^&`@(jtF@WKmZj08f3H!gv*M{a^TdZpVfGkFX-
zWylL3Iq>%ZaNwwLAekmf4tZsi(TGvZX72kI%An-W?wsL$3^kKH;KT&arCI#n1&kCN
z2n@Km!C#_+IJE;4WUtWi`~`TM*Y?p=!Y8zy5f6wvWnL-<BrFFP0pHuu(n4DD$m<Ap
z;Vrt6WRezHDeBq#ou%Y}6CnTC?$RpPPKG;SF`cYyj^L#*o7aPL#hc~yv!Fv+Bze1j
zR0&&J#bHCZL4wV)eeiq&P)5thNs9R1N*TkX8`B>j85&gE*Ep@bF8jIUoIFn6))uSv
z(eEgXw0fjj8@&NQq~hOSg>dG|!dtc$KGx!QgpaDx$APp*V6VIukL&9^BoOs3P0KoR
zM?mSiqTz=zw=$rYCIcz}K7-llL&ox6$hYC9xFOxTU0DZ>i{mlf2-3ONk~x-}>gulJ
zKAd~9IAXR;Yj%Y^d7ucn*D$oXEq*Zt&&lk@AHA-04%qw^`I0LU?fk)>g<hKn!j*ma
zuKAqMCiJT)AC*3hA3%R9j>-y@A_F@rmu#4+dV1+^t^8|!jdrwQc6>h5rFuF<7oil7
zTKT~Bw<ivUXBIeIZ_i0g{EDhQP<D!hD`PQ!zIJqX_6O|En`e5Chx7Ua@;<k=p!WR%
zIlKCjCUa0};Ha&aokHY*F@OM7#Pvw~1)?PP3{lXyu7fKBUHx91k=(564mHu~xeSC~
zr7@<N4I_|wDg|12(hG^YLw{hs{Mk4xzB#kKt{SzNnCH?$r-ZuLyfx7eWZiS5{i&8z
zlPy^{NG^n94n+(Kv9QB02&5=ZS`2>D*MH_z(Mux%`equ${0S)UH;>0ksL+OyW%Zkq
zI;;0#`|%%8;M&5$%n;G9_@U%`stAtACe9JmfI?$qBnHQB<8?}_4#Ur-8>;&&{~03*
zOUbPs1j2RnY&){90w8$(JTyf0bftDl^$^0VPA3_r1bWM%`x9x)C_Z^t_TO~pVIa%C
zPoLvh)%nmNtnZxR2bbX+fSm`4igNH*6V6&l{=a=teg6o`HCJBmym)8C<jSkH_G^rD
z&6VIU=|*(%P^a$;PL`Bkp?{{Gfd~bVgGX%sya~~}T+A03IOm1N)HR!+@E46+pUE}D
z&dr555q0c^_zV@qD5#OrM~m4AD&K#NaBM5{b;B=cM;2)&@QL;4(uUh;Gg~>TS=Oqh
z+-H}TNH3`SSU`waTrp}v^u)bi_wm6;f32VX%xVg7c}~t_^f5!8UX80fD5Kgt;)`<!
zewfnb0JRjpzjL0bFst5^q_mT^H>w&8eL@_i-hCQnNpCb7F20{mskRS!-PYkEg;W&#
zw`qFhG%N+DocS47BB2{4L79{UU5t82&iIZ$$bon`ij7vV8zPKteWV}FqNljc&4Bj^
z%OcEm3CSQyQeZL2Yxr#A;IQTL3J~mHvkw64ECIn~k6F14>JI2|kwp8O2A9#_F6SuU
zdD{-G@|({76{e%%Gamw_IFpB4ksBNwvp;w7z2-#qOEA|P#p@W%X#SubiOV_CT#gqA
z=Kh3vj6A@I6Cc`B-<%tlFMw+6TDzkT^vFijQxdLyFoKago-p2Gr?CJ)o~!iSWC!Wb
zxp&<M`BsC#r|@t8p&X}tftCyDyx{ZZL~&Qt)9lV&CndYaAwxrZ%9q>vWKi7;NiO)-
zaWwG0pi;)CC9Q=bAvrJ~`KIN}8ftD7;^Nt^<zXk~+i{PXlA`Q^2BGXdZ)t><RYjwu
zTyC2nLEmsHaIRJ!`A21FR>b}Qj*|Yr5LOHQXy0)>MaAR%9>bi(FyZR&{(_1uWnc#*
z_x3lwjL_E0mkiQpF|-uZX~U%gx)IzeKlITDsVVoDyd&OsBv>JM8-;I2mTqbtU44<F
zr}m%6KqoYNQ$K@v#c<~teie8N;|Cd}RLRdD$^@8Nm9QOY-HG*W3vfx0xc>bguu0<B
zD>M&MSdIrN_W+iJE@CP4!I9Tgz*hv12;HgS<r4Ia<Q}7YfdHA3LEx?<emkq##UfW=
zLgmwe;8>;v9umx(1XNdPtRAX-U;3<W@8tM^GRxr9?%Dt30>n8KtVw$_0^n@d&;1DR
zSDF#N<?VHnZeHW5<YM0GcX}bG{o$Lc!sD25xm|qICv9SrrglHKdM>v#*V`e)kBe<N
zGss@>elYRuoC}EIDX&)vpDq@6ztj(i;}}~za=J>@h)&I?A*pZHZ``q^;2R=$TvOl?
zt0|y;M}_YZChkOn^Q6^O9pBq33|w!K=5++}Vn3cvdQc}1+-@K8T>71mVogMlpP$n8
z?DG}My7%<z%T?>EzaJt*o=%mUq3#Ll+>@;bv=NL5KN7}*_`HZB&F?ea9!BRg*R5*;
z|Im9WU_8JbNEsHd%%RvO!|CM3zm*3mfZvO1f1Bum2ty?2-{KVvo}OZQQGFg%fB39`
zI@&bVl-h2GW9sTcg$0+ZI$F;L-x%4oA<OC<rwZhzRIb<dXIp*CqPc;_Wo_2bHN>*o
z569Kj{hII9oJhfo9k8ac#xQvo(}@(hsr003FRTV+pFd?ju75YgEkRvizYz0dAUr5n
zKN2$H<xf}%`3#4#_UU%o9XsL`P@v4<sIOx`w3Fo2T^9(+u5SDlD&3>Jm1<GS^1w7x
zre(pI4Ebo!d^`mrj<9M-36@g_9F=RD1Bjq<`RPWhscLn}B_vgJ9i;;i#>vn1zl*zG
zx2cANqiNoDuAVe%p?eVA^d6Mo3GhJmn{X!*Z-8C2pJx%7qb1|p-L2tbJbE9(4;EGZ
z!+jc%u2(qH=^X{P%p2Ygh37ZsA{rL?b(uJQJb&^4bS03UG8yM1FI#z@$m0BetgAg8
zCItPYjLk-~C2T9LnMfI|zD8Syu_g5o+UV3*n_jb%4-WJjW-X6SM(}$wqf|G@rxr?(
zjC-dIW&sACc;yx>uI{hkj)7`_Y4?IQz>J(@N;PpRd$i#Izsa?1+b*T&D4=OE@EEOM
zFL`ti(x&L7Q(@4Fc<!KlddYiV9P<OqS?GwG2(C(|TLzRn>#c79M0<5G3Od*pp%(TY
zynH{F?@=45c{}8O^BSf-&#cj4B|W)zaYK-B)0aNLs{s*t-uL`BY^!!VWFsV|J(dU&
z6nVI@RK5Hc5=;WS*(GcYZJN{9Z>-v8sq6~+S!Ja7S@@E`8{J5X<@b#2Y>$ENJihTd
z0?ICSzz1;q(86~Gg2J60m(2L$mi}-}T%89(FF{oB4+I;6szFSj_g?Dp+f6MkBF+0>
zz}*Iv$h?fY7{62M^miccj8jLgH&idt!9Lc@a>Seh119&ke8gm!15DTZt+H3Q=kj8o
z$G(+;Dc?2#!djzMTqq6*7ixu$N=}_c;5EeSqfhPhapnNW)_m|^&_bJN>wNQjILM+N
zdi+K}+{JGN9j5BMrR`({j2=K;=OjtDk*ItI0su2XU6oCJ^+v=Wfiz%OxAdQAwBt8e
zW42*BXb~^OWuOobl69A{J!RXxC<?$n-2>&JD1_Df4U}y>ckK*zTicv&{13+N7(}_8
zF@NP{R!Q&pbDM9`!Yo>7CmVELq26RFr@p5Lm>u9G<C01Ar3ApU-eQhE**DZ?ACsWy
zCC>)hORuv)9pHVlzg74?DX!{HfIfNV?<YQMoejF5L}~$`Zj+87q3~Sf7d1^yLIFAF
z{iz8-bI7HcAD`{wyBl+O?1*iLGEQO6x8J%@`e@`gw6uMt4bOt70EEm82s7cU2e2?b
z^R-JKt!NWMnzE;u4ZA)Eh}*9C@K|RMS*UPxPzt}aWuziYT%1`~N%mQcK=j6(QR02!
z8*BLo1?>&VLSuTrW2p@=v2FB3xD&=NdZJv^%kKJW%s_1tH6&zw-gKk>M^5DVdIrwD
ztNb{DJ72^8c?Tu-_lZ0)-KD(z{1i2YuHVPn2|GCa4hu^7-15~(K3Sag5{iZ=6DN5(
zns&;%I*vMfc6pC;Ez6O7&uvv3h+B#nT^KvbwVX~n0HS=iiRue<u%w=g#g%q?Ty5Bd
zEt<eG)2xz?-#*?Z3YRz!Bnm^ArKqX_?qk<z5)kLy_qL_KUZ;nsaNZDiSZYRdX>Wm$
zea#|sH$;+XH*e*cc4ttl5f}h(&u?y57FZp1sb_=D4R!-vMc}yvl8lWD(mRa)F&M}9
z3k+=b{HthlHuzCk<*Lz>OeA%6R!b*OeK29uJSaO2b1f#PPJP1ojcGJRsIfOpyI#|E
zgN6Vgd^bFJx=0}?nT;LjFxeJkT;JSWMtmi0E1o~JR_}VQKxms-!YUt}TlQ)kbkE~%
zfIR5JJ<!)pZYgv1ba%db&NP;In_i7g54laHY(M+ZCq716c2vM#^6uya5$UjI?P6FU
zC_ZQ)(s}X^xJY+t^j%7ThlN`r+<@Lz0@V9$6quaPIxgzzlqph+nG(i>0z|B$C?z~}
z*a68spDnZN;E=;8ih`4&dD{()Jd+HTG)^{@u)%Y1>@z=VF}<AR&Ed=Ls?GAB#&rFA
z$*$hzZUQ!EMP26y*p3z2H1c+anQ6sB`lvxGbuK`T{ud~DgiD7(J-ZhJ5u{8mQaq>q
z=Pr<`1x%8Vb~Zc`a-~uXY3R+`^?rWiQ_<nK@UAkn_>Q?;!CJcQj7vc=F%U>G`YE5W
zr1Hs=@$S?+9ZjYWn@{~Mue=P+!3>_UtM_U`pb7csSqJ>1ZBg&te&c&FR|)r{tw+17
z{}0-BAv;yrq(?6CGy2ZU81%P=AVsUue{j&wG@L5<64f0qQS}B5v^B{b&_q~d4O966
zJtvUK_al@rWG{i(ZkpHYp(?m=@O}x3Q`=%ZP{2S5h+=3L#iC8Wj3UweiLhbLE{9hj
z2Yx5vtJe(vX$jPdnlf8~+@cOEC6BnYqocDtVZJ(QKr&J;3=+P?wvGIOOfz8&j&Fr5
zb@tD!7TdqSZ_GGPrvb^_Z4H57ossPgt*x6q6~p6D_sO6}JK6YwlFg<e5WbMg?EJW2
z0VvlGU|w%<`f)|Tm`GpQzT%)6u(G=B=YZU?YmKrng3Nb5vSNW!u&*{=yMY)U8OH5j
zd>0Zx?ceo;ZRx!4lR8sJ{$y@K((gR`BK!m3X{HwI#or0bqn`>!NP0}TfS`&HrnJqc
zegEj5o<~R=Kr<Bhml&ZPA?G9@xx3Fv{?btu0odrPgj4+o8$q8j=`YZ<(0B@Kbd&;N
zaL3G_G4a!Z0FL>`6rZRHdjGyi8D~r@ZQO@oUgvfW12h{h)C)ziKWp3_(9$Amc_PP9
z$>{+@Y)ob4;xJ+w#igs^D`|b8Y_2M#AX@;;-~@O99iBo27Ibl504wUtT4H-LW%n1~
z1VWw<z%m}r7RdZ8Za>M@oHzL@*n9KcVVZrcWV%}EhIR0lSEV$w%UeD4_hmGP1RN7w
zpPQheCURZbf5+AXP5BdCJoR7DOky{F@HLvpiKPGtc0xi0ZW`BpCg-&=TyDLpOTYNG
z&zyf~g2+oYycFFYtD5ef_xxBgPW3n=-+m!KN8LwcSJ^STMD2}&Jp6|I$<(Pn$;s5W
zDT3R@l>>I%Hf-yxIjAh)d&XyzfW%S8$L6cnGBO1{+*MYU_i|~eb4I$t#rb<n%Lk*W
z-?Nuyy4+fCT(~gPA_b^4c3rUgfjdkSA3LeTv0MY$Sve~IzK!9BFhs<O_SMkl;vHQ6
zE;YzVK>{0bV2rnay>p*8qf(%y8{joak8NA}LpZH~i|7Q}mzjQlG7NgP-^b6TKr~Ff
z0ibF4Hwv0O%lAHd1wPgi5%5{)AQLtVdkj*|+9_!uT-kp((d>jMUv1Kpd|gC<GMmi&
zZuM*x=hBwN2-4Y^ljuhE%O#x)balgWY(&I#C7o!0i99FIT!^+zM`)F9zZZq`H`$WT
z-Q|eop7k><9ZGzfHxk;;zF&N7klhXE#hw2WNFy5tQ_><$KcNx=QFP;uy}J%2?9fSa
zlt=(C4e4VXFf}wB;^C5!h~;-g`@XokYF;kstEO<?_*Kk2R^vao#+ED{sz)pDco>}s
z1nh7kR&t-EPm316F8lx{HMth}Ra5({9K!mQPX<d!0iszR_gpRs;#yp0e|Uo4%CYYD
zkhfX#fYpRTeQ<R29APJ$6ion5YPEkfg^_*;TvfnrfB;Gm_QBzOh3=hzgx~-n$`7sN
zsIR+!DK`Ak@%tT2;ZW?m#%nNjtXsiq@?mOf7m})A3c4=-@wZBO?#=$axLf1bc9jJ=
z&D|M>xqU!uP|5Y6D9S|M1Z+eA^5+<$V#S&fpnAX=2^R>cuI<!6s0#$*Y1G37R}uik
zE7Cb40#exIgH<*W5yXAxw;UQFqnMz1qMk{K8LioJ_ODO!G16s=nz}#B=G<QzASJRE
z6tO_52|m|1xBl(mbe{!ue`J-)!{I7l$0m=3e8vV!8sPoji>Y~}<;@>GzLs%Ha#nOS
zq#cPbKv%4f9sy5-kWHtXZ__dfg3d9D>N^(n14MJi-y?aQFrMjT9JaCt0YMKU1JUr1
z!2XG1nW2!7s3<}^k(+a6$PWQRAX8d}J{L3J_aSw)UnZLLI>Gsz{pZv|$+s4wo&)Q?
zlAtHZSn;)7lb_N0OYt9Qq!OYin&|!4(-4P9Y=xD$V>;X&sg@FZJ()3qKcUD)m)yLO
zpf&L*V!T{4_7I&HZ*6vJJ!-f6gq46^%q(1F7d{lYf{(EO`LT8y=+A|4#Dt`cna`t}
zyR*TfC{Ivte8TXVUvP?7B^5cAUWTqfalJjd5=X}ukiC057lHcOSMB~lLTYv7mo>?h
z$!8b-?k>62<e?XIV2rlUxnlkckOxA+Cd~~rafvh?!J24W(BDDC_cQ}2L{bySeO(TX
z!o7N0;E%|Gb&3JQKM&3YK1R!cjMzm8vo5)&BgN-x90HIhEE9U>cyO!I_rLEdt8O9L
zG1%Ce$8;OmL%*SmXp6S63+`4Va^Ryan<Yt!RjW5}%|y9r)m8|kDa~Y>WZfU(O@1&R
zMQEGLApcX#vhTdQD$RtMf!_q<IBY9RZ*H60vLioqJiG5`C(=KH@2H$@dpY}Hpe9HA
zQtHq~H>^+$WJ>ov=0(ed@z|?VnY1Gz4oLV())cJRPt-Yv^wygThN-JHsnIs}URSk3
zp%&};5)U0oI#2`s>LZJsl!;`6Bn2-(BP!V+uU$W_O>4i{ZWd3vZ8FPMn1cEr?7eqb
z&;9@R-{~~7sqCzxj1mnEGD?&*%(O_zR-vKcL`G<$B%vatk{t~qNlR9C+L_Uyl=|JC
zo!`%2zw0`V?{QqmbzJ{^{qgzwI_v#@y`JMS?vMNZ@eo{ioguQC!)x$nHUYDJ&s7J+
zq?%Ms_|kHCyyZ_~?py4H=s63(LF$)_$Ad&tygZM;Qw|ZmH6$;0o9Q9=e*)I|VdYxv
zBluHH)}<gdcELGQe%t&^i_4c%++#D0l8U9vr%%21@N5$!23ndcbw@Xh@eLbgkUC_W
zv7b1L7%Gt-=fv@q7U1K4>2}}JcN^MA%?^%wNo!TNUe40@XVbN+=V;5Qp>Otkl#OYL
zYtgGXtDwcJYZdSAD}q~5!r`d#7?iiH8_u7E#yCPNbHNhGm9h~w&8P#dP)dn14wTNp
z0udOpbpcJjMr=wP)$rCK9(Ub8c<{-klcmKp_RyJAdukV$WJ#Aey)5WDTBN0cf3KK&
z<|hB`l)6Js+8w@3`rYP(p}iT&t~pj&(zD|m4;l{Cg2IJGT_PUvnOZb^x(;=`0<l0!
zpK2e3$H+VE-xpf`qxy~Xtt#(gK0>}_hudHJ%Bl;NgdB#QSMD0Ql9d*M+Yy-$n)l{9
zc2Y6VF*1Do^r_E?Rpo*k6``aE*JjME&b4I#6qGyN{L+Oeed37Au5v#je|5bXE%nXa
zE^Pu8M#<I_DDBlMqV%VWJZ@v%o_47p(mnTDWp?goRy4eD5n<18Po%`<p|&G_U?5(-
zJIP)CRaH<**d&;v;F-4UF#l)uCgE#DB`#8X>rKxkNY+K85klzfh;b_0JDW}3D?28c
z&uEx`Ai2A3w|n1f^@2xwdo4HiYh~99d{&>>y6A^NAG!U9BJag!C4x6q8Eo!xU=(8K
zGQox^t^UrT-1BB%!Q#Y++Fn0SzHXw{SR=hG5F3^OE3da^J5QR~>u-tY`tLUI7JM<s
zJy*xkr{r!21*bpHg!H;aGc#rziszNB%u6{<xkoqf>5<LYk7}pG!q)g)lYNE{Yodm#
zd|B?U;tnButxdcPUq-Q_Ma7`xDyr&0=W-Kl_lS_lTy<Qvo>mTi$OCMFrjItNr^_>q
zHQq_RbGijn-nTOGu^v4pf$KFjB@wgq|MpbogkTNlOE<7Vnd{dHNlbxHf%h2wwWupJ
z4V}g<FpZ8>RxJ2Y_J)QQAuC4swO3YDN$5!5=|+5KRCdq7lY5<5*8M{CrTZ6~seR2W
z>mXRGpxJIXy{@_!Jzg=z3;!CG9Tn1dcBGt5oj}m5(s308fd)NCj3v-haZ!xwyE?;d
zj^-E9B!uhxN_QxuYRGI1n_t2S(QLlInyfk>8e)~FI2JKY3X!Q|y7f-;39n$L_RH%p
z6J1>k2a^T|!_$__To!Fl02-hT^zfhN&BGjUD84;q`_CMx<4u*Ak1;K2Z}@;d=dTPq
zxs(_@eU~o~ON!pS#p`=Sjmi`X8s}|7e3%jbXHM|Q=>p-GJ-UF?r2FTi$(;S8mpTp7
z-@7<?WJPQB@`AE}S$lOv%2HQ8>IsiAw51Vd)x$)!n%+km4!sjfPcON^sE)psy&pcz
zxQ!0NLT~kckND9o&L#=THHJI4rh@u9zK?_>_5E#cfGbTI6M>W7`*lbJO6~KP1nK|U
z++*-$(|sErgOMw1bvU+<>{`Dc<I>0d_WvqawZ~@l{I%l1>@BU^hzy7<9K(~^QHK$X
z)d)1Rg5|tZR$G+pqhgMYeK>I{Lg8Xk`eV{M<>izrhq|44#G@shI{#U6O8V)Ud#A!P
zb-WdYi*ANCIK*1kIVpk?ArezPg9ZS)a^+-#Iz8_fL`W!l9-2+&6jr4^?rkvG>`tfJ
z+A+804{=s7yq@lvU>7m=_nbE#M1*fegzEn{+w7U%Wwc&M&w;86w=B1_M+VB>mJra9
z#ApwVnP)0acFcl25gG`$HN7?Yw5bC@jIb{3m?+wY8fD8eq2qV$1`KLyX)_iYdgOrB
z3*J#i7qr!?4rdtHD4p8VI=c5B!)E;iMCS!7ggRdrGd|ddn)Z^qd(Ws#yF2Ffy@fff
z`Z%n_%Dl4=$bXXPv2%C8lN;+_^=Uk_4Z`Sy<nw*MY2VME$h2lLS)gS=ZK2Cc+QB!E
z`h}_nqkH6!^Om9*EN1c6(5^3kb<t1WUGlMM0W@nJEqk_Qe~#R-o`GNQHF%!HI>=3K
zxV7Wt(uU8XM{JVctTFTCuVgtQsMc3jPi1={yOb$F9!!QD8<iFZe|fO;WV7Xpk)tvR
zXCl+a!WDh(kx2@(Nfca3XS%)?Dn~|(udDmJMDVF{lponHkE;5<)=AH0IW3f;3yS+y
zA?WV@x{jnzhr{toSAf#)Q?FsWRLWb+?N|7DW{a{DWu+-z{W*-qC;Qs^ogFsfq(Zl)
z#nB%|gsk$L)US~+r0>w`D<Dp7tsh=%kr;07(}*k0Q|UWH)PsM0JMpJhJZMw#-V>r9
ziu=zLjB)g(PePZ6!HDqxTX6WQ)+nHvdJ958&_)-OsSL1$b#iZS&uadmu4;c#%g-Bd
zd2?4k-Y-Z)Q6W7}(*2oeSR4Xy8!;IOGeeYU1(GNZ5Fr(7fPuQy>v+R{K?fs{_HKGp
zB-mDnWA1XO!x`X9P%bib=sm)%Z_%18CE0!G)MV3D^XA?sA=w0W#b*8~SN3)jLQ*1K
z90G>qB8~<41S>|n^l5c{`|xiEHO&(gnjsSz?|1)J16_FVMEi$?#7G!mBTc}zh34N7
z<&GL<H;n#%oiT%)W$)m0_qgnfkRSI(h^Q|v35>b~DaUR5-44Ax<AHG=q!on=0nbBQ
zml6SzDTVV@&~Jb81?zbh$cBO~1Ve3a_dzl#tASV9nGpZlQQyeE{gpgmlF!2fl39Nv
z)hJ>X6SX+#4ytT1ar{-biCp70`pA-;n@g)03?XQ?0In1=D=6!TVkE(m6Ovo%Rr%qX
zwLFG4rUAlJ>0kr88f1hYKELnq_X9TIZrwHB_OgDD;brH9HD)Ve+6z)#PPch}2|~@O
zQ22RHj`tm~5FeL)&ea}7ZkmSpa~G&fqwCDQ<LD0J4Yg}^s|Nrh(0I9fcjz07v1Rwx
zh~aOY-aXqt*DK?rwz!~$nWZ)FT>68%Kc3+SX<b0IKl2HPW!V_NA&&^9Q}>vNjL{~9
zP`Z4{`Kf2W8!1E(PPWEBx6pfcvVGH3447M(TiURbuZ7F>DlW8gHlAD@v4!g>+T$SZ
z8F#ChDve;uK@@okqW=xd1*=>2GO<(~@m~jD*(qA>)el|)VhK$}kLZW|ww}cU8QOwE
zjL=X-|1chZ|El&TjnONba{(vyI84DOzD&bc@<vwR?0&0gIaN?EPEq&jUe0|Y))yH2
zaWQ1xBp>b&^w9F}f}Is7g+|Hjq}F9^Rru_oO?rOqMZc0}$$LFN*ivTy5&KAH()b!K
z+^qMrWTE)wIQvx3Bv(zwbQWmSvqCPjdFAv3mA8Lj_ZT$y3ysAKG;%&(fyst@rhkny
zRj51J+B`b*MY{Vvnq2gWk0BRfaSQf>59(wtJ0w;lb1EdoWo13^lA>H06)R|v=^IMW
zHG$P>6g2To(N;-L#h#J2+oL(uo(Kh?GX3OYIj{F^zWpcLwxv%@klt?X^~*V0Cf{Fz
z3zbj)2{xHke=hWZ5B>2p-17mGhB)5uU7)@99Y#*Y#4KP|8R6?d>*YCjqp^7h-cMT9
zx0!{`OMHU_qISsg$FY7%TDjyG&W$=d%U8?_WD6Lxp>QsJhW~eE??xk=kz#avc-IAW
zt|v|JoYX6^>#;X~A!$uy>#x3n>jqAly=9U7lqsHA?CDMYGPX+(X(o**8QP~dv5t)N
z{#!R%x%?>l8EUHq0myK(<0S{_F$<VH=C<WT!T2U9>FTR;RJ5frVfl%c(>J=__js-{
zf5ColZM_4N6vp{gEnd2%_o^X1`lSFD8|3Q;>=U25f70yuHFc9C3qAZ=#&qlBJa?VV
zuKXKv+vaj7k%sk7YO%*DayzB2)<TYW4p#g<5Bp5IcwoNsP2+`dU#NuYRo4=cR)hw6
z;1vsQ>t9b4jMat&1%sLK;LiodsSsSgF9~LoV|4#0u38v4ARe4<iu+!1KW_G?CHLr6
z3ybANYob1t$CY|Bpa!MIXA-m0<^yrox7BsbTt!j%SGP2y=^K(SZO)FFb!ecbI@ApX
z9rvqvQM~)1sX~!ON4NbR2Z$1+QY?pitqq)S&zNSVilH_y0sOuZy$e3F=A)5mpLeAy
z)n2eyr@8Zw%+Mic^h66cuBGOfgW-;<<Ic7Ae>b+9&kXul<tAF5*3h)ebJ3zXVZ%^^
z!PA4?9+;6dfr)fj*zkG1l(cj3$U;^3!3EA+AGjs>&Mxa-F&nk;LFdB-RRNu#|KZ*_
zj#D6XzNh2`7_idbp7{I=H(ao=?)^DRCe?BLIzfnC-6}+x7&HPRFOE6axQJ35QxZg$
ziZw#pr><2nV0`$nP^5GWr~u(P_W8+_>HIerVkNY}wH!__+Uv10*Er(rIROM;<j=@A
zxOm%LlWnjvN;g03Y(Al|m;_b($Nj!nyju_$%+vZPL^81Lbw-X{0W!@B{CQK<@&e=>
zx`i0P6u9ogeeSV8y9zC&!{Q_w=KMP&zy?veh#yJ3n4nZocvQp<f;P9RA?XtO<kFj^
zeg_6Tfzy1&`#sIaK$L-pV?|FQB(gBXb)il38iMPBs6>o=pm}-~=-$unLFul#zIuLp
zTot6k7<o9nqmYH1%F-ygjg_g@c<YJ|PP4+E+8qYD;LnfI*}&jG&|&KLOjXi^3ssA(
zrsB4Dkgm{(+lP(pO$|&0!<1Eg6{15Un<lVR`@C-6oUNaKZ!*x?jEiZQU6qr&4=Kyb
zuO!843@uKCL?dt&;o2+e?|8VvkhAlu9{d6-<KW7ql%P`$%I<tH%9j~7%65E;8%p#8
zjU|IU=O{)b6yCE&)=Kf`DcAdpZDE{cAq&aoPX*E7uDpYg$Nq?FAfjn2nwJQLo9o%v
zB*6l=_gIz_CNPFCZSVkEGB-H2IEeYUSPr;IDfk3Z6c;Q>($67&a45eIEyi5_B7+kC
zvGA2Ry5<lhNfU2iN&yVtLJ$m{GFh~W5#`nNSb5N>ub_VWE$#t&^_KN_0=T066dt^h
zfcF(6L4YU(dQ1i^9u_kp2)QMl`hII#|30~qW&gt$uE<jh+{ejJVL$2N4u_xxfBhX$
z&x4aK_8T<>ho#@RB@^H$i-0g(4XDzz_n%4lkKIpC!6eh=Y%)qL6#^!Q?<-VN!QF))
zMPd6VMM;joC{FKxX7SzR>-m@n$^1j=JS6eOx3G2bNQhiGQ8&C>TX3TB0G~i)f)Ghf
zl2Ec9k!Y0b+3U*3jQt4UC?XEee0NcgJLHVTZ9Lv{Qi7M8u^%o*2uv$ZR0l;hkcb%t
zm%>mx81ZM6ma*9kX%oMbwh^|K4s~o=+3VQ|j2a29O2U;2x5mSt?$x+#)#G&@JW0ud
zK2xS?`-N&Gt{6k%i-Ya#8B%gmsb$bqIOUtyY0n1z#EcD+_3!1aDyH5zZ&0<`QlL-I
zy?uRaL3cyeEeMR<s}mAlP=$YlrJ3j00Pa^Z^hkq49qAm)tY<Q$ugvNkI^b{Lxp8{N
z0`?o0&FHE;x6}T!^1XhE0;!JLM1u2p!ueC#GZ<a;tBvX6`D5(dZw_@rL)Oy&+{MTN
z%b`7A<3NdiWOpZzU3lT}6m930vQ4VL;h{&3-oC%F%y@}e^5TO0^=D#YGChtej5C>e
z5BCvSc>Iu5rO|4_YGCH9@_PqN0hOSTwST^-Dyjf=udr4?2cu+{D@a{H*GCH7r*_D^
zZ+BuL!XwR|?Bz3O=%I+i2W+lq9O&06H%}@K5{1gi03*FrgD3;PMR_83#DmDYzPGJ)
zs9R6gSc~R4=!3yvNgtr~n;aOcSa=eaBypvS)$AVzJlq>>=Qk@K0r|<?0R-=^`41gm
z5Sngjz7i7?W6;y{kW`A?vL5g`ZIYZ;y4u@4b`boc`^Ro*R7iXs<`&i^S~#>)pDU7U
zRNWUXdcpA*gcaRA9FSyWX~jRin+vLZYR><mDOEx1Ge^B&8-49m-REp{@(_30@5iHU
z_!}V^Rf!$rTTo@;^R8?*OO6#pRJr%Z*ygS3zL%vfgGUxc6-14GHw7|IZNHh*0sJzS
zQ~z*unkE#2!Zqe!7uU?^<eY=sQ1T_iJ=p&?_Z#<8B_~?uJU19f{R@+{10%{&H$>>T
zIn|Z)7gHUDL_Fu>{J8+J0{YqYZ2>6td*ETJ#|M%w&wq&~n-}Z%Y);o~n<E=Bo%_8v
zNvH9ObtDr8%csBSfADBTgcrW1tn?1(vHyTc6@s2&zcmn3CLHR_J>3-Zv@4PN5(yY!
znPi-wHw}jU=1x|F3%nuH_os5nfkwT{o)cTS19oP%ZiiXLRb{l~Z*+VCX#=Vx`xHae
z+lb>kwa91*&u-v1xRGLOd1&RB=}D=7CtWZ-oz7wiQ-q1kcoj!?Lt+4nOUtGsmg(yH
zAo-D(lfva_sNrs3hxtf0+gyWCk?Af9Kpv3RCCyj&5N(~~T~4haIBucSh5+%c60&&q
z0n9~$qrF1O#cB&GS52%rnY&h?)pzo@CZWdT4pXw2>+#HT`qU!e=6TrY!t;qtAYvSB
z6AABI^cRhjjs9wFXVjvvYzc_jPuWJyh-)}4%hI5Ns3k&O)O@dFlw~`C_maFE&^&l=
zF)0$N4{?kt!|LPdWd@VsJD^FXt@%Vteh7|wZ`GB~jPDXnS;YMg><qKXTQmR(oYF?7
zLWMB*u|p6+fKZsWEG$9idXq!CYtLI)B{au;m6W;$o;wLA*{Mgt_yQ?(u1@UTF%C<h
zUq9|wjp&(`(%F6qb_r%uSWNY=o!)CH2%IE<B=#shcra@bVIaZEs0M>2)8vz~3h+SW
z$Tlz~3)U%?n}SB;@M>{ZS@m5GrY}rCJ{U{vMe^lree5jFa71N<=%UZZ1Nu!F8orl@
z_!;|&eyE`<9^NQ%^=j`mhjC}E)Kt0=3&1~WB#<dY&P!&D$%dQX7Qn0J4wUo}C@tQF
zc4D4fmREBExDGg3e&+hCzI$i^cS>!!6aB(m$<E;gH)V}JDXsvAA*U!=Nys)h$G4G|
zLpK;l8U;TG;m|7vYEZ-(v*6rg+%|@b%<wi=F(WuZ6uQ#@$cTr8tOMANfoPWQ?(Upb
zp~@$C(;d28xD(tCUTO$2hG>8Tbi?=(0_q4ju7YnCadhliilC_kGe}>;iR?tKjAn#H
zg`qwPGkhd5OAM@7DSF`5e_9ySw`w_v$sYzkkLdRatxPi~#B|?@-CU>NaF1FEE^8du
zvl0mEYWW8SVr~F)5Re$F%^Oga*bK-ms3b8bwk^f&)M`{)o+iTtr5HjQRwLRw9Gg^$
zzkScIkQU-IxWr3=CFnF>4mFj@YtGHvD%t+Xp{sG$wsocL_G=Sr3V}AJGa11OcF5}H
z=3=rig2IJBnu4yyrtmEy8t-e>F-2hnD$w7gx1#F+USjSEt$GNpOGs)7X2=I@i7<%j
z3Fg1yu}g9DVIRum3Q7-xQN1Pk_AuzryjXR#@2H4U2{-6ODKt5v1vmK|lHy40gNSrd
zhp-^?b^7km>#DqVh+qh0Y*Ol1uw4k}zhC{k_@$WJacAvFa2XF&DM@*{y<qZv1;a`K
z3yPV0Y$B*5IJ1Z)$Og$7;CAA{){+*P@bS3a7sSQA{+?&WXmiSeZdCL<h*yMcp5<zb
zmn?65BauhXO4FH7aIT*68`#c@*UkqtE53nsj-TzymE-f@2pd6|BiMTc$^38L!-wbK
zYG5HmnGzITRM;-^G$XD_>+hCtjqJCGJBJ*ghG#xOH$<-A6^I#Z2f@F#_A_?y1V#R=
zM^>v>t~^8y$DkNA>ws?1y`@y#h#Zws?~&bzu#H57Mlyd0XF!7k{&WySPI=aCRF+%D
zOw?prls8x+o<oRJ1owjMX5l&0(}Hnvds_-@<GyLHteCn?&7l>8=t00Nqh!Su))g}W
z13!Nv!YsfbP*s@=8=QQA8;6KooZ%u$)_jnBklk%AmxxrwodQmqzODK^#bjDuFfn-<
zC`TWYLG?h-oKyga910KO42bk{lT8SDd4g5a#M{+ZHW#P=dx_+4UlWpb!5W7CBSh>3
zqv3&5e9lqs$4HBp;hS1Tje<p<Yivv!`V%P6rP!z6&ri?C7UrzKiXl9I{=BeW*yUE?
z2e@oR$ovAds&$y=F^l6pNW&sPx5P;or(?#PX8Ch+8k*E$TVOA}BQz%|6O~)6e<~5q
zm#IPl2V3YpZgW!JWby;q3^xr0bC?E?4l6n%Wv<Y^&UNg8HROax@T+v$6(|v&%Zn33
z4OV1T09l;K+axR;*7j-g-X{)4UJ80!wuC@T)YC1@xl?*))sz%{*Im*{Q<X1%l~ocL
z^U+XY_=wS4h0VKsPtl{pv1js@bt8L&OwQ`9Zxq1vxXK18o?g~FvB+*w%}C(li>9Y%
z*ze|RS^C-Atn`&`-|O3$PPBLxBRP4m)Vu9R0s{j}yPPJnYmU2aIK3GL@S`6+-@xIw
z?^N%7xVbalF<~F;Hzl)cHTEZB?S%8$%5zGd9x-)lHk_(v?AIrf9A7j31~{+LHq%yv
ztW6ZIIv;pQmt>0>Ev|Dvwfp}08WG>&)^*$x$DggJR*#kTt|v@S-nT+7rpls2Yo&Kf
zOlaYgg9&8@lSySnsC?AQIaf`<@s~>!)miHI;C}X5oYU};MiDa$QWg^iWT|`>asp!D
zh2O)nf(N%=G8pWrm!mm0xa!h@2@7hJ-1X&db?;?kIrS{3tHNO`^06p|zuR~AQPniu
z!}<ta{pB+HOe$0~u5#Y`Qccrv*>3&e2<S6Qx)Z{u7VjApByBR4W~<Eu)fg**e%5y!
z0|y^)CL%7ndYupvRO3317a(l1eTTkJm-iF=3K6Q<cbLB(5={e@T)m@0=2F<vAux4u
zi1juLKj|kCP#T)2tTSn$PiYQ?=*wl$obuV-k_9y*y9q!jQVh~S4Q_ChlOpe6Heg5D
zld{>jI5V?)RLnLu9s;1ZGgw8&Kd7iop|Q72ps3>TTz=!=@mZGmw;!{HX24!&#Nf9T
zcBw+pY;m_)lCkS9)c%TX1t}-R-p%PqbLW|6nh8Py7y7BJ-x?UKnliFCfr*_wi}ioo
z98sJ<%1Iq0@G4hOltz6P>L!JOJ(CaLON*d*2`SIePN%@oYxmv^{R9Qk-ytyjO!V2a
zQ|lhjoO7_#upY*K(b)yjA`}+FJ=SUde;t!_(iL(@S&JuxllRssxHFgM!{S&<9yH%t
zXP*VVuKR3dbvzaNvCgf(MT7x|Y5xzsv5md4Q*G6LTjkqdE{mG7N<@hx(1ZxO+P3v3
zV0h`!h25$bvO`+rdu`1f=;RZ#Ie$jL!>JpfSa6H2bz*mpjxRG<Qlspyznp(#CZ_+m
zBx)^vtvrVeQ&iOq8e7vf1;bk~-+Hw5e;)m&V<~}wj75!T6f}0RlEsfQleL%smKc^V
zTGvKBJ+3Pdk1p$fH5=-wN8mF=Iv}SK8p=X(As=!8TaqxA@bNY~l`6{OEn7GQweAak
zkhsN4nXezwx*RP31^rm$I`(~9EWJD{xk}mxh|Nd2$9`>4R?;=!5wbh<9Gv8`1z3Tt
z`EnBCkQyZz(gHb+Xa{POQTDTzNGX9%h{hAI%%nu?d%3%UT2NL0%II*=(B*^}<&J&I
zQhsL~*Y)<gJoBM`mj9(P+d7G1Nd#8`{>KwCE|zI%^5E@N{?kNt;=mzJ1sJ1|4L!7l
zD-?J0ivq28CtSfI*zcVd(Ma>Ke+vGmJF4i8&Y%B(+gSy*SATssx<Xhh!gOuMLVZ{)
z&Yw#5*0~ZpLZQ6ZII>wj>;H{=XKxejOB2HUx=5Jr!?3n}>VIQcf&wkw#=cV7yhJ2<
zB<IJ(HZ`ma!;W#ZMi_~Po0O+Ujg^&<@OlNW>>|?T_8X3fmhpz+2INBt-ha4^<WYWf
z<+0CTOh>}8Io568|9@-7*8kh4s{gCq`ac>O_CK%l-{|jq#-8D-+8wzDg)51ahyxB@
zxksJ>>3*<<IWZxr*{MIki|Vwhn{l8gqiou%&-_b5!gn@KS7<RY`1gA<XKibE9Ld|b
z&8GBaJ}yq`4mr^lMKbon;lqc)FfLtM`?ti7!EKjF?k)#g$axk11PbJqB65cWj=mih
zju}cjvE0|OXr>p&-2dk*pJ&|m&wexJcmq}5BH3;G>78vI*|Qu0Ok&QQk-ij!JtnKW
zVp>9v_0YDq{AK3|@JZGsm+TkQnH1KFLow}%)PycJ<gz%kev23b(koclivU$Z@0Q$6
zUcuG(Zg2f8xQa6l@=%d&-TCJP(+rDD&12}sS#&%1{ip3$<~Gjr={KPR%hIK707%ez
zau0z-X8waW7cKj{PT`6_oZj{VX9aFV#5}Z9SV0Gg9VNrthLTuJ?Nh?ie+B_HYtMia
z&9>K?^McfqoWAm703gEIvu9<@nD^5*z(->CWpW-t-*wZve~_ufQpXRA2^X1IhvwIw
zqLhpNd8c+^?BY$t&w__Ika>yrliD646|_PXrs&-hbRrtPw}-^EL@|m;!f-d=GwNnc
z+d9o_8;N7Mn>x!`oN&^&7Zh4W&7;3q+x4@=AQt%<|3U)CK5q=Ma+;$eo_%@S`pyab
zynZ}iD<o(?#6Q1k+x5dzgz_7aU4_qk1lphj%v;i}i-g3%wy*aklfQ=OHm`?FzYY?o
z8roJ}es;K?8<Z$;e9Le#XQj;|D3}2$9cPfP^L(;%!5+wA1k(<&Bc?lv`@TXfJ(!dw
zmPXE}?O!k7pBoZ_`U*8*#!$v(-S=A3XIN58%da#51?5*}5)xU7z~UTs&aj@0z%}Tl
zV7O>DhvfuS9PuWXqdqJi^RHc(C|dxDlFCk-2az}xP3rZ5Qcm!Y4r%89Gfn8f;UNF_
zmi=(p7tZkxS95c7FY->(-c$fwDpR6uA_hk!a@3+k4TZZ#nN?s!WUcTVj(=I!bA=c&
z{O0;snpBE`3?#flC`zYl;5@J`!F1tP34k2(s+fTXHk#E#Mq-EEk)R_-j(q7a;Us`m
zVp}#-Le2v4s#EuYb)-;9n3DZpTA`GuFk+^avjcf&!L!6a6CuW42hbuw|H8>3h3=zi
z&%xx&o5b(&MOH2hZZIS{x7$ciY)S)A{>+;7fTsa}yhW~y(TIoGLf8RVvHyB6iLwb4
z0Kw8mYfE~?U6wG6K*M5hJLIon$KX|L2LON&IeMtA=;A7#C-{$KkDf5k^K30wk{F8S
zO!4(5V6fixL%{{m|3YL@5S+i{o)Vjv2&|wbb;;=uIl>0XXM?-c!p5b)N;EKJ<K$$P
zi*6}Ob?5diAUCF%6Hbh;zrdTzsJ-=59r7Js`XYBme~XNSM4z^imGZN})qu+dB5B`g
zh7wq1rvK0)ns1JnVg!UEDs`xC1ST@uq`kzn!NOXYnv3ay<gS|Z*$x`gxr-u5^30x7
z7IYi^i2@gJ0Q^hqZk6pMhJ}f{e7Lg+j|93gdaoD67?@U#Cf>iFf<tcy!aKrafhh97
zf|tWC+@`Qsa`0!rHE>;6n9tA@@LbufH}D(lPixK*$g!&eU}OqWLw~z<)T}VZ2P&;3
zB+3H*J&rVBr<v7lQA9iyCvr_*Dir|vVY&dOfUpEB0xchaf~X}!>PP}j(I5In`4Qd}
zs9qBe67p#E%c7QyzoVAy34n5T-id+EOOJpCaN^R)f84Z61zigO?q7`sQKBrP_JEQ&
zDjP_uN}9Wi1%p+K(0p-3UI&S3cL#_~%vuaM#WN~7e8v<3jfRcf8*vY#&hX;x@cpc4
zSJB09u8Fbn*JtOSvJ5fC$A2P%5sfVYy7+WjL}b9=6`e<5Ql|A<9jXWsAC?9APd%Km
zjVvK(F;-ku<8cSSIg{yHMiJCjY9SP?dwHo?@Wlpe*3A~RZ`Nq%b<rKC=r1TBqG2-a
z#9Tke-9S+&ln%gy%-R^`(8^!Pq<tEv;{6*BcQCDyi&gF}kyQg8-N|#jsG^XmDZPFA
zr$Qrb0Q^)S9j8C$02GDLJwyk>%B7VY8L2CKxbJ&soEDQ#FbmzZvO!w8l>@9-1I!D8
zb4n}FaKJhpin!Kw|1kP0FY}GGq>o{UQVQoJ^ls)cSx=xPm7Z>ZRVCrEAf<E!a(ukK
zUNlVqUG%raU>PBNtozehKUWMWYV*a`lqFR(hTsImNhHDU`r}ndr{(|Q0&D@C6Qfx~
zX$CJ>eybLmrX#(}1xtrc`;Aje>DDa<l#|_d{A$XxioOz9KFvqqy`mk(tZG5a3K=7v
zpp4f&4iQB!MD+Gc3~<)stbpyW*!UVYWXq>+m$5ehUU5qjM^wB($}TRh&vphDl!)0@
zJLJU3LawykMt6`sD#jN3Dz$*_=~?If)-kjE<2m#u0;?9pv8j2kfNXwrk^uK@*~l))
zYC73t;hYX<&h?<xMGPfv$F;X<930sdg&xZvVwwDpSMBkl`{R_~_EdPo8^v=_PQ^{v
z5CPZ-apeQ-MV;yKSs3~KO7=rx_zlX7Bx|bo-Xf5U#?a8FP2NqGJGU=7XS!T)86-u|
z4qmJ3#EMamt}uk&2Ch{>+oYwPR~{H>i=jq6HudzG0T1Gl?TH^kea!wJO!_Dy0p0#X
zC3d8r`O3VBIWIWhq2(y(C_*S+yb=vFizJ=$-?-WGLUJ{5vaJGVMikN?2Epn8-I?m1
zPvM~byPlI_Y{o8(m9dVf9$<~aluNV1Z<g()Wp~-F6V+==k7MI=YPb^uqbiv5m&86X
z_fukyjH*GS;1yICJS`w{8KkP+C@QUVZ4=@TFeefI(c2#IZJ={LRm93oYeUfSQi)Cw
z=OsHtSIlzCb7eBqm_-+bLS(QHhs5{&Bpe$WAeFKEHBCJ!hzO8l#k1DWu)&13&2A&b
z-z6=d=IVv-7^Jp#!IsJn62m%)c}}x0pXy!W6FQSMqkY!=oEl-%qJ!swk;UNWDA=@M
za!`9EVs@bX`X;ps%?}mKijU7I2nlk8=n039_y%5#VW33RB`mRw&v;~3BdT+Qj|`n;
zjd)5h602$eyH>4SS+4kHy=cV;$mGw!tSV!{q7^)7f33b(g>b@H|My(ne_ISZEeK<-
zzN!(+*!8m%1{+VgHAl26Y!npIsPD|*BwQ06gz8$jz35&&OJb(tHswRCE>lu@m=4`X
zpusA?bO|1!x~|B}bb{i$Fu`Vu!Sb9e8|a~Uh~us~Vz#9x`m?A>O)#AqttT+Ush{Rr
z<;*F*A{cP`?rTYXL`*s@=xb4$?kYE%PM(<sfC!-n%#|!ImtnO1+#cfm1Xljw?K7hH
z`zC*=8a)V~BFJDWzx*}TTPcC;&-+M?*#?WDDY>GD*-6zk=5=v9tr+09j51_Fa1Nz8
zm}Fv3duGZ%GeUxb%%ki7mI!DYxjp-G=&n@S=@HhmBvdgDTF4%r>7S-{?s}-;Q-zjK
zpBcBp?}}J+ibnmdBSx;V!4(L7QuR^)41Ed;nyd4|SNLg;a(~L1e90T|?&a6=l7;B?
z@ma9Se)X^Zl4u**xkIk2;X~2bQzfP>Z!gF0!8lvk0O3(PuXlVd^OP3qD~>XCS7Vji
z<6d7v&ws}giatTbwxkyQrc`-xt~pm;GZI^Lq3a6ba86?;JyNN-tQHM)Or$GJ9>2bF
z+O&$JKV}ab|M0(VjB_3wxx+z0Pw>z)H1!{kIs2oNTBi-&FN{BX%Eugk-8Vr_N)lo5
z2VQfZ664^!8!AI5-8v{c<u|Tv|4A_oN-*13q&^U34bF8&usH3a!~Wy5b_Ymf+R|gW
zclhF6>KQp*D`!4VR{62`Lbulg7g)}PQKFJvd-!#pjPtbLL`LCLc8K~2aYasIJtZRK
zzUxb3Zel0fHr%RiD;ipAhp)+9Yx+2~oQaO!QvH5P#aRMX%BNRd)A0svQPLdIeLW7U
zb;a4*JnaO^@f~zCuo!knGZS<i)|qb5|H}6rlpO{^_0Q9J+T-iI_nz6C!q1Wt8((U{
zoUf*{yGo%Q6tl%V#}}P<x(cTVUzo%)qT(#UWO?Z*h-i)fuH`le&u!Y3v!pj1OFFgO
z+7Sc8VyN9wp_g;?lN4`k_J;p9HX*r>wV<PsvkU#P`}~~;4v>8M@#z?1l^aA$B`8w6
z4P0a^Mfp?cIXi4<8#Ote6mv#fK-UMrxvya9Bz>eT>P-{yzB$n(?(U?1g5B#^1O7Up
zqM@AI(y9CrR7a<~9s1pWt4euRZ>OT9%F`0~LGM{5Ya-^sq7~kDcoNQWwHj%Sxsd&Z
z15Qz-F@uolU}|i`qta-Z@f$9F$0?Gc(JAq&lqH${Zh-=vk`r|vaE1X_>lq6mK8w24
z;FvvBjzsfw%$X}mGXZ$`iqH5fR`g0}C!F2|L9LX$Vz4^XY85?x@eczWM<|*Tx-na3
zj$b;#foPdrpbhI|R8@uE60Rjd)i&f0dYfooFc{JV0-uxB4z&c9=2UQ(!iMKM?);hb
z_4*qpPx4MFC1H*gdJ9C5yE4_`v&HHYVx~ee1w^4F!T4hHLy@3j@v$;3V9G-nxdDQL
zw{;;l>h%0O)zkaq1hXH1<wf&UB17fz<Kx5wZvP2ueH&TMif_73We-VOE9#$1N1}Gr
z@ArZqGnfN;hk&tkF=jD^{(mQ#SXpo!WBeREgbb=&J5LWkDPm*F5u%bwVV>>OVn`i%
zSNpHJjwO1NZjozGmzkCYOcL=Wb|h%GP{rdsq??b|iSE4dIxq*Rk9N4|#>2q5^c^6=
ze5u-iQ`74(cn6bO8-A3QrWs-A-{06|8-E(0Uae508FIAZ<2QjMXGf$}b9AL7VgH@^
zP``J)P4Z1lR#K{hM8g@u%O38PI3ogf5cy1o;OG(_itpDK?w)=w1q9+<&9jKKM9P&e
zspt&^R70(@?3N{`TvkbRqQ*N&7Zibq{b;lGT_$2IF+p6Js|sUUIbfpd#JW^ee@7uA
zVh5i=<}GDmBSaT>yxM`$p>`9Hf$}>o9R92x?E#n#1@K~cRFGCFO^I$vJW7cFy<Zqb
zvaM5bgmg`4zyk_Z|A1kMZtX&7H1vm*kC?!OUw7QrAK_TbpQZ!jBO<R56Yy6OKxe#H
zmLC(?3ys>oBAg;4HP{X0wPM1;iv-(~AIwOY2N8X9`bzu<*#6sIa3j(W&In;M<Ok6S
z5%^ied|mdKb%U~Ir=&Vi#(Q0s3#Nnl6t@35`n;`|nSkrNR8(BR>MS|PG@RMSb-ICE
z0o9cV3Lqhf0zVwG5XJdSLYTsHFX0;h*YH{*M!Y5XA5UK^=FRq3iR7UL>4J|F@>;Gc
zIge;WDIOgCwXV0IB2hB`L!Kq`F+1m^&{N=9=g|u4y()nmdDZoXnU8`2!Km=?hrmr_
zON0yL;M{AWAE58*cTlJPUd+;VivAb&+!0Q#<r4SVo+;C$mQsWhgUR9q)bl7>iGKoe
z``<$qZRVUG`9H!t{^vvfZ%=?T4;~^hO;*f9!(3!BqwwJ_!q$yQ{%<8iVF6ki)8Z<z
zqt8a6#c`>gw@vE1HyG^RU}5g)9`x_85GcJzi&;?4&;N#mazIyBqM`NW?fg3m7p@nV
zURm>OcTm7%?e;UeobE8AZ}hGm?RGv<RZ#EMJ7s#nS%rP7T}Lbolj^$U++X_chqgo;
zbneyeV@iJ0@{`-&C&jGPxPAOkO{?eYaNv%wdEfG0t;-AFH6or$>GI{v^ZeR14E;~>
z32017c-epdRJDybZTtKG{PTa-#L0OZe=S>q{Qxyk(|u2Z0Bm()P=4vC5hF&ZsxFYy
zKW~rN=k)2*^SX7}kr*BxP823DCpTN3>GK46PIj}W4JQ-*g*6`4c?Ncbgz)2+&9d59
z6(Q;@lPA4|-JNu$<gb8);~JX?ke1riAiyJRcx3TILcbSoeVSRn0R!AV-1)1vG(3E8
zWyjk7VjwGc!tg)H*ne;cwUh_Hv;YSr6YhzGgGlG)%a<1~j<dD3)zwYIiEIOJQLwt!
zNB;+KA4tqC*B^H4*RN+jFSUg~&CP=I#q2l8e%ji1T1})fzJLFYJ=LEzi%Fmph7Z3B
zVw#^%W4Cks>tBEU<s-BI_;E{9)5n<9GeMZUI&w@5ppVQ<vr)b!m}jyGjoLs4WBlY#
zeSQ6}U)5M;=wc^^4Eyo{4roQ*u4eZ>k_ITw$lTn1eX~9<H(|mA1%-$EhySI&h_?X%
z=}8+#qGX5$XWKl%S$q~KNEN|0jLoj&gNPhL=Z<o_(TbXbN2~w&Q|W5|9V7gJ_?ii)
zR?QypMid{&Me`1I+8mF&+w|tTu}uwB`p@!8<wJ)iefaPJMWf!PeozpFpNI^&yUS<T
z@ZrnJwnSgsxpNP$>d>>NzbM2&zkr!GrtD>ole07Mq-X|<UvOA7;n-XPbSpTrY`dhU
zx(|^K*qoqyg1x$Lg=nWrWuKfgNpF9jMQW+Qt^D`3EyUv1?r{QW;1><&&OK}>dPo2K
zV$m-o)7?CiiAmrUVo(F!(fJsf&O@0cWC)#mHvM4thQbDj$Yq*aruSmhV*IavJ{6^<
zH6J&GBbFe>r@7x2%tMl*3ksq^1piN*f4|1Z9@paf$jDR_7cWn6eWt1%yyzIhXAaF*
zX@4KUIZXm+B6`K~<BhKn(HK$q;-QL|5WLQ04x9WgW&4d=Z;Oh&i74DHnbuefN*>U1
z(BGfb>;dn%=6U=--jjMu_{uV<CO6?CxX5JpLQcuyZ7#*>9So}D9RSZ4$F$x;KXBz<
z@A4tP8X8<z@2zzu={VFpsj8&JiM%!CQH{TUc$WQdP{jJ{x4iz`au2?9XIiGt@aB_<
zfuNeb3rdZQiK!$3=&GBJ&4ojk71Y_<%8EvIC+L2e-t)<Nz}l%(rykusrtJd8t<}-e
zdYzwdl5nl8%zoDfh+^l?wE|zh{%O@!4CV^^fztERgnKV9uRM^Dkl<r*0Z0IBWc<~a
zk&%(nld47dw;^qt>gv3~&dm^LoVJaNRPbfVRE8SPQ}Fr1e*0=4gW?^0{CMnQ+fAp|
z5zfz_KfnI^(nLES_h9a*c7^7YZ>qK%VxwbXv{h&1=0z_gPA-4kS(*z;*UQ$Mw{Kha
ze-G)yeqj@U%I)_HYl*D-Z-Ufo<mTz(@@wC|eLHm*4k9(Qu2l?OeY|~mNuunbb`p<8
zFPy(yk3>X#`?C3iDwPTyujZS{(cmy0ZtBmJ$^bWirL6Sw6A1v9sFpB(+<EL)Gq9qV
zwc+Pys5EBG1)}pX4UKa&jmxFCEs$+qcDD1DWoxlHKcFbr&#H^f$jr2S?&aX%uw~2D
zddDt1hb5z7j*pM838K))cutv%*Sv3WcDoPJS}<_;w~C6NoYcINy55*mpi~QzQds6p
zavJem*Yo$5*|YbRXg_v@I1xQ*(}D#HioZNd<@VV573qxBloc#*qd9W|L&TSKxY@Hp
zgs}aTuiIKpGYe!cHdqZ`wQ}`p%bWWbp4i*a&>-fX-JGw=!0g78uU@@^D7yS^Si2U_
zwxZ~UPi}H-_54d`f5LD?IQMFB?=K(NSRvAV^ysojFlR^LL$Cn-4j3wFX#9d6d&{%w
z`a;of^TPL#$Kep00QZL!w$RoXhPc^>TtP65GU(}+MZzHURCUTILF!?>_qSS)NrHLK
zA$ms{is&8z<tqYajGG5VM-WrWQ?>-wInV7@tbktV)Y9x$*`Y`CJ-Jb%gn+Fkh_Dx!
z_VThgxXa4w>Mvv+|I+tS@D=oGFsv^(twqCXUyCCbC$bbEf7wO^?X5jiQoFt!A`-(>
z!*z6;0W@lYAl3^1aHsnt=OvpXAqJmYeR@1#AbAIV;yQK-Z3qPK#b(19aigmG>_XZ0
zf9P(hrK4kIEpic|v>=**H)Krq_DmYcVO!~g?wOq$yyv0OWiPiTVKhhr36%v?_|m0I
zLK=EpJq`o?B5&<)0ZNmYsCN`f^D&09_X<vM4FhbyhBCDTCM<=TIEzMt>~1}KhE!ak
zm*Y^aE$7}}ye1(fI56Si<CM`6OvCzgeX-Qet^`8jdjFdr=BB2m_Bhf1os!^FPu5;~
zY1uy&hs)vZ4y1z>Il5R$U@63Y+bC5~-@A1EkI1EvRNs(KhZnARRGzs5Hka-9kKfR)
zb>KJ^Xg#V5vx2+qoZ$L>Sw{Qbi@Nme*-FMh-NDY%>QWZogxX!S!WKuzSJ+vgfi*2x
z2(}0aNXZs=OHY9SNOnzQlMV)*l9t_Vx3!o<&m;o*KaId2piKC)=QUa=KtbAUuB<u~
zHBzcQ3-1ClLx+0bnmc#y;NLs8Z+}jskp~sLg_YSdMz2xa2!+RnlZ#@(_jyS}W8))+
z)Q6nlYVwtDUe@UPG|rx*kta?W`SOF~uI}#e?rw8Y$fNqSFQ)JUk!a7M?9YGscd38k
z0|bBV%w3vzqsw}qo}u$$7)I??ns=JKnix7rV-f|)uIjP=)N;<-E#EC1!Ijnik(Y2`
zKF&*-d9kO$cs(BisIlL!+s#IGKN#cO_@eZm0K1~Y!E4zVO<A`rq70T`Wp>ZMJ`@T5
z!%G9WpkRLfEC(uU?jD&NcPbI&3bJV8_4B)T?*>Mu?c&sl6Q8_$mjX2$*1woeQZO_#
zFoEzubS)tk!G{~on@4IG?eA{_icfg(3`KNo*yTSRbV>1uzJY;%VZ*+BS-)vhHJRe^
z5QDL1k-M*%<SH4AS5a{U{pIt#zJHi*WfdM1vzoP~*7}%ay`6vAn9zd<yVcJ=%Cv%)
zFTY>_7*jhSyKw%a#e^|Tg=7Qz^zPleUq6Ajq6p+FHZ?Vg;xo&Bz4|lIgwPi1>FL?o
zMe>NIrUL$dzUx~3zDOFVW{BlWc2!YVSD!jn$n%gSpJNpW65!6P`r3Krps#O}=6jml
zfjTQ+{y0}>$`k=fBlk2k6A_W$yeNSn@LIm<qW(f@(Kns1-th9$bd_mv6w_J>J0-to
ztc%0L&q+CDL1s(H7=6QV8?l*aR17MUAM_Jaz7tmZV?G!Mko-ps?u!Q^+N8!NWUn>p
zpvLE7*8jc%QW?8NOuwDVI9a3XW9FaD;j%?HZ4!f?2n1zneqBN#bc<dDXGxdz^z}t;
z6;K9a>+t--kt?N_cOggE>*Hg6bMx>~qZq)Lo1HCkEPhNB!il&pt{tPUE;<pW5Vyj+
z-QPzZzO>-c@Hk6D2QigFlsa*(h}#Db8uXcw5T&K1)zw7=Sp<0PJWjqTKSwmn1!xk=
zoQChqot!S;xZy$;CNu$wD-5(FHjj5i>XXrL1;Bm1OqZ+F*DEM5B0X4V>CifI)Tj#1
z4V7qrUtiw?2N;?P89hA7;PvU5oxHuh2ffK>P9MHfEDyDe%8H8Ljg8mXdYrGX-1&b1
z6og!b!$iE|=;+8;KSGN*nBJTMt|WS>H7j<%5D*_{t|m@KM&@kZX>Pn2@#`|zs@0a)
z7DZJL;e5H(ORQ}ij2eVo;nScdFvgV=+MVKEL``zWj2WqiD$B}fV{Vl?@h-e6(@TjR
zh7TW(jeN3p#Li)Q1_mJRH*iZG9J)6wcC7e)R^rr5<xZ4tB@L?GDTqJ$ret;Vl*MXX
z?}?VNnwl0;8B~y(8XC_1O+;^VW8;at-=vJ0BSG$~KWVRJ)U58#Ek^m=iQKOkw$}}s
zE$9uN%wz0}eCuL?;B5UQ*ZSu%%RO%l4&=dY)9p-^;n6-+oZa4dov?u?KT&bLMrFs3
zA3q|yE8kS`O-xM0Fvic1@w1Vu&@r;HiLmLqO=oRQ;cCdvocE{Q&SPQ)N#4ZdMzrE>
z)BK1p<>g0$gI6RUX)cYbq_4(;^=I<*C3p7f-`^oQfqn5Gs+-F1bZlY}%yA%1ZuxX=
zid^x(hsX{ZG}gBSq{{ND!|9DLo<Far(317=;RL@4$W{0(Ro9p<JH=iJR(ezV>jE0V
zpY2gfyOQ~Z3*H`4HuG5X@<he5uSDESuUxM9YiWx6wV5MTAq(uXOB)sZ5#zsYjRzK{
z#v|O{KOg_0_SC1sIUteRiQ@GOW>I^rORl~}YoS^+kNdTyF8<6%i}usOu2L7iq5YS|
z>z_Cov&9YEWa&SDP3h95i+L`$;8Cj@ygKewhfbZ0no!-P{@yN#3ckkClUm^5gLI!e
zcJJ%s)Av9{yS4lLG{!&aJl)~a8=fyoc1=uLMKo4cPI9Ue($v^^%;i&Y@oB#_PAkqs
z^}Ww)P_*YK)%Lr!JM;bf=9HLK@vEkuCUzD2tgy;g96d_8k0uA<38=%Z>$v7IZVSY?
z|J7xgQKOPJ7w`P-9uv)vktE)T9<qIau2=GiJ<#SVZy&jF<Hq~>0X^zaK}cItY@T85
z%+@|=((-C<VC>rpor$-Iq4?z7ygYS>7`QsVH-4ee%4z+LK_L0MxbZqCu+i(0o>?+C
z4PNX)<=0b0>7b`<&W-cwg=89AuQD}i^ERv2BGTXIy=hgGy!mHDXz00bH|f`+qNes-
z=_>(>@p~t7<7cIP+E1rtk&k@bq5di}5{refyiZJdol?5bo;@;7ayxdbYVC+Tw=bqZ
z+TtBXf9<TX7T4Dwd;hJ>I3pvYyR6R3n~GZ{&7Y)sdpn=-sP(Tgd*EA1UnIp+iJXgn
z7H+G|8<Jw-Agx*}waw%1iWMuyd-$Oc%zE_b$@=?#2D!qmkTK<G2=sTVd)@3SJ{(CS
z-Le<zv7`UP1rX_V={t)O?9HMt?}BX8k46`e$4h0MZ5LfY__n&(QL6oEfcuVV>e5{&
zgz4++qoKB_zOcsvXVU#iQ9mUshnJU5`}E$1Mggl^6jG=Zfiv4G0Y&7=MJp;3TgwHs
zKz?#J9aX>Ys&aNLb}U)-gVVizBO9s;M%H_gVXO~%z)MIEsD+n<k!TlFpjLSx&7<+M
zb>y{xE3_skD0q^pqoZ@txK!v=pHyImK1tRr+>pE2t^Q%rB;N&Lhrg<gN}Hu*(o5I9
z?xE}Vi{x~~Q+4fvH@<tQ(b@pG9=gtb`}XW?n#`@7(~S9=HEUq?4ABwj=-a^+B_)xK
z?aLF)*RH)VUF&`Gm;qKD-&|W4B0t}pG?{AKM>V_5s8!e9>(?4}Jf6l0bzdEe2ud-2
z=$qL~BNmTHYmORSq<ZVr^Dc?%hZxIrWK$<|B{{h=LIwA@^V>VOw6sJji18W^a%riV
z{o#zLdW}mv{)xO%DJT1TV^Gm_c}Eu)4c7r(fnoNn^4^Zs^yzzt0@bpwmg!CqcLIwR
z&Zkg-0b2Jnas1pU^i$@Er<p5rcfv@6VTb%i?1SujXOuviOiJItd?LvgWqqOo(Ycyf
zhC`6MmBHq5&@!L6nqC;z85e$D^X`ydy(Z8_X2%W*!VWG3>8lXMRcPVbV*wy&%J}tR
zrM)+HKs|rL*v>O>sG=9Qloag44%2(t+1c@;$oai$vyU7;Z2$J~A!XM^i;lYaedr@^
zDCV)EHy4CdAayF;gcQ{TLx}B0Dl41x+KrcH{-}15PuzOlwO~l!$b7GQSz{4I0|&l@
zG!8U`k!<ux&C?0K$If=_*8e+E`?*f_w!{|H-zzd}22JR?p8fO<A{Catcpk`1R>#%!
z_arSXci3$1nrcQu?;ZA8>r)bHLgGYJf1R>NpLqo}3D#8D@#E`ES7Ta1)Sg-WCAxSJ
z?BDQM^cl#1%H+ungkE94mb_fNYSmdEpPj=TloZ@i+-0QC|G9GE?y&Nsqw6KVP`~<=
z&dKPqb2E<uq7y#ZafFtZmWD=QW@ZdQ3~#lzC^hBo*U3&ovglGBQCCT^({5wZ9WwsX
z%F59u3SD>hl95@}cqGDGsZlHI?%lf;wSMIYBv(v4re`MAUSt4Ng~_sRIPX&*r0Epx
zmMml(lnj=z%M@3|QqhrQ%oxL7y8(a*anM%BV?OhGypY7?gJ=Y&&fc|ir<l}^jF4KR
zI?5+4t>0J<P!JYm8=lgm&)>#mA{81lc^AQa<<m*^@ux~upPL;Clw~OZ(kE86?^8au
zBK4nygiScNr8hQO6M+aB%IMMbeQd;?P5d=Y&n%oCMSQuM^5aLZyq?i?YwXALJ;OS~
zZN#IP2r*HJ(vTrTJc%BtrMzhV#Ke(jPoD~*vY7@&!@!p>m(;F!vpC>I&Z}3kwW4AI
zD8$&+^ULB0B<kRcj~ka$V+;!Vnivx5g~jAvdcAg|rK-HWr<O}W&_6t_N$)bk3nV(3
zHJveI{d@QRK|xcd=F4Vx)qVl4-;yf9=odBIby%9_mExXv46KP3b>}v>o9tA!@77}v
zUrIsE<lp(U$>`g6J`P~$(32vK{rDs!eU5T)P!Nkcs`{p=){&oGzj!;o<@+oH1Hg-^
z6DC}`Lvq+LFR%3P9ecbVJ8W@raap{$5NA5W(UE!S(9qMt*=jdD3SWMVIn^cVPu{+@
zYD{-$;6T{OJzpjc`~yTo3!0Gpo;Rrs{29ai_6hHO-Xc7fzox$Ijzr3_IowAL;#mZq
zwttkhQ$LuQnUQrhqD}16x3Ah*(IxLmwfy46i*s*Z`0Dzsq_h+^mi&}!94f-jF`pP2
zdMu>>#3qPu&%Zt_y5T<Z!1$PB`&tPkYu}kVtNsT5Id}8+tEW%7UQ1s;x=>K{sa$2e
zsLgUbE)04$ciWQ5Q7PbGzmNKG^DQP`ninVF-3b@hGZ6ffwMm%j9P3&ZHp~(2X7=gc
zUVulRPIT<lY2Bh1gHKsU>UG&WOTjm?Lhgd7k#opx;*&|s3H0+_`iIzk4@pZ)+mhh&
zk3rOdNQK-V<K5V`^`6ofi!T;b)!dEY_-y-YX@K(X%|z-M-d|ZY0A8_b-bsSO@M%&)
zvxnTFb?RwF<cdEb)hYZ84G0*PCh;`RQX}`ydeP_Nu)Uw@Z>{<tKg=?rBD8o^rU!{e
zpV1#M9<dgqm6bPf?Ld=X0t)<2dlGA(cX0dm?a*AOnT=LaNhFkh;`dBBN{r%WfN>-7
zUzKrPBD0~^eDEM|P)*BW?P(&Ycp`?<`Z>+#?}5|8G&XIF=c+%;w{9+4E_N=g>9S}b
z8q=`j_*`1BM0;pF(^*5RyC87LqVUca2Obit>I-8nBRLIWCex=3d{=XAGZymH#_wFb
zF-z1vVBH=aGBop+bEV(!y44ep)p^a8P9;r-m3Pke4jH@D(r%x!M}xkRooL)#d=`j<
zwaULy4r-izBj{}Z%PUpg=gsr`?KNzdrPmOoT?#_=py1mweGS-g-Np~OIXQ<r>hb#P
zR+&f{52)(`hA?`e=*aL)`b*F(jo_HXc-yp%Z<nS<Ukm`=No{n=EgD1of{+af`@t1A
zfoD}id{({A@G~$@effbaRQH<sieijW(o(7nba0T8{&IX)5um}Rd!yebyG<(UuJ*Kw
zdM&GaEYj5L#Jw6-cSA#8j@_-xBv=Rw)nS#Aqfj}BowR2a26@4aH4|1{f=)o&?F{ZP
zpjw<p-74~a-;VVUlA*eZNzVeGqR<O~8hvcNb&}JWBbkrZ;*#fA-@uH1`fTYwb~(Y-
zi_^Vn;$faLKRAnwKwjTCSL1>O{$b`W`u^>6k2!j9{<p$}lhe&V!sV;6m_J2Fr!08t
zo1j^X^Cy|6d3kCl_mXj5G|-ds_LEe}pN&CzaXW{>!_)s^dpo)ymm*QLWaH0mQnh!g
z#!E6pff46M=ff)M7F~}ikbko97Fb+=jg?&#$4d3;wajXghQ?ygrBbcG7CO1O=(^QJ
zS5)wJM^?>nrKyYRXE}NK^6lN{PVMkmIjz}lu>T>0@p5wan{AbqmFr!i%7C3QdTNE}
z(Qn_{>G`zk)0J7UBgVZlopcB;MSN=Q9IxWLHLABN*68<YZ(@?qk#wO9C6@oHL*V=F
zTA1*~tnOD!pgYp;8~AwWg$rtX^alvc&{n4$;H*_oNFaJyQmg9GwJR>=&?>_M5+x9a
zMY3a?m^-{v>pDEg$=dO))v!nw87x8K;NjixW#H8vyPLn4Z;uIMYe{jsg)Qu+a^Cv(
z^=D5n$sbbkwZlD0y^%>zc5{j@Fl&#P#8m1JGp2|^)f-mU6LA@;sa*-EIm^NC_F~3d
zL&Km^9_Vty+;f})jiZcy_1^C0CVjC3=2FHTYm|G4Qf>BRSxdjAr21EPFCaEJH1n2w
z{RP#zbEasdutc=MHS?&OJ1SV-ZB9zt1+nY11bhBb!Xjc;;oQG&CnbG-nK;wTnYz!X
zu${x234MIS+}G3r|2#AZgPfqT=82xQvE$Y<K!s=PjXe+9^DUnO*Rk>wZJy?AqczF1
z?Ws?$79=GlVfzY?J82XhK6;2jovO-58=9K3Lqh3h6tL>a-HQ@q2Q1aEb~zzWf_&EP
z@Rci9%A@Q$JP%L5&QWq+p5^B1dM<yYU+G2FTg{PS@29O;Q8Q>VB&-V!o6KfM0WrNl
zRY7PK9fpUVzgXe&nk^qonAJBwJInSA?Q8Fysu1{O{Ho=z7yGOJNr&&|FxY>*!B+yM
zC*HNzYA)PS^Of~d&wm<6#fJD6SeZ>TRzAcKBjb$yPl}3cf45xKITWAdI}0|R^w?Fj
zKU*7L`03-vV*vrD9Of$@if?$I3R%2j%t+tZkpy823KhX`@JP~4V$Aw8wKJzP^Md%{
z#y4@Z<91!pEU|Pxc%Zbu@~su?>Esc%f%>Xt>ae)jSkcn(OzZ2=S&O|#M0nK95!Y-a
z5wb$-5pxNNjx%P??EdZU`Zw3(>KcF&9~hHy-9%^L-dbN*Cwk21`V$S6PdX#zXkMhG
zsF?Nh($|KagQ^e@H%<q66A8burQd5Ht-gKw<oMU7yu*0?ezL=%asBC4oWh`QVKn19
z;5ffgSlr0_Z{EDAxqalhm9@3>m7LY9SCfQ1f4vPlxxv<Az_w>;0i#VKE;epo1FnAb
z%Q^arK*0{I45UI4?{<)anqupGuS13tfiyAkijsoU4hC**M&xo@!y(Ux>R@!^@iMHv
zAWwn@Q4L8*sHTlM(|^<~11f&F^a{?3PSSgWAtXKf1h7Sr2MW5i>hay=RXau*bQ&;p
zsO^Uh*uhx0W2;%uny8{PM*U((WvQJ_Eu;V4nTQcTiwu7INw>dS<I_}DX8FqL?T||F
z4V8vXd}Q?7_=SCEYM*%KHrcu2;Mb?y0ApW&NHeb-4&r)rQ_u6Yy<}xKT0K1)HR{Ae
zk~ucvkc;Z(57X8-E@(9WYZY9hE6>~%Y*Uk2Hggx*Zc5vhMiR;U_WOisfK_+-LLddV
zdMD+b!=S6{g)Pj_xp%LW%CDKX8DS>+sr(>MT&{7Vogt+V)xuuaqiZiQ$Pk0AC<z@-
zMoKDLfi4iS?-g%v_5jNVKl4pTh1Ngevt=lyhtpoIA!8OArry~RMMu+*<Lm796Wd;B
zYISpSqd8K*$&WblQLSqBBkK*8{88y7s{yO;&Zq}QqhYLByN|0barpH5TC8J-HCGQ0
z>r6%B(y&`2rnmGF{qlefey%mnz0mJ%`nH2le_S7|w#LkKdM~D|gl<|uJ-nh5Rh-4m
zvC!d6nnNmP+r(138Dm(tu~*_-P>iV1-cHoLAb};08M8*(LNmpF_o^hHwB5{^I@2;d
zu0?I^Ih(P$KkJ`pgz3gnVbL%eaW#V?Ig0QzO8KChl+48CmabHl9@lTp5j6Hj?l;C-
zVZ9kwfUCRV*wwGc#L=S-s%;^`QHZ{aa}*8lG4qS3&N)y<7P!#rq&CPWx#1aIj8PhG
zo3!Hx#thi=JGiBK<j~)XFE#skjJ538N6W7~<}`JTIoY25JfD07ba%qV`Kmd^ncv#7
zrc~|6m-lWDT^{ckZ(#gM2NB@?4>M9NFdxV=9K0@m)1jx#yiYnWof6;X^ggi{+*fhn
zF|~C^7lvrj@#NGz6+w#!7yb0;{rmTyRO)wToY)I=8{4(^s}5f7e8MT>kk7fjV`Q}g
z!ou8}8yt;BY}pMf3nOb~9iOys-xUWX@81v@UwPF(bm};z)AU|6oK1tU`ERYM=OFdD
zSN=W~p|J*qOwC*O*|Q&ld~XVI$bEKVZ|3PriT>HE2ab~1p^|(1i+oL<0WpW;eEiJy
zWZok>GKyW~2ol|f4C{<=qKBlUWRD(RP}&9h3V0h^7_YJ_p`VVjay1zvmp|F5+^~&K
zaoOV%t%`2dy}X&`kZl(1wd(e*TcQkTo}}O#)-zgu=%Fj&*XOrRyJr8S(u1m40vOwq
zlaUJZ!c?<r3Ia*K9_8jPlsc+xYolU)P%lht(L9wFBNGDF=RP&tSILt&*66B*ls_A6
zIld}czkZZ;B6QAyj2|zjjg#U_iF}VM>es1f+$V(q13{>m(r*2?l9G}t)mt{usS9CQ
z?yg((u`r>CGz2sv!S#G_aIg!&x842DwI}!Q-%rQWkhB!udt<Zp>hZZ29nfEWy?G#V
zPQ(bsm;T;v=GT_RcG5l!aCcbldDP_YbFv-!IdZ`o$To6c)lx*nX}0OfNS;WA@JDA}
z<%~DO8LiQ~J}O(gC*2%HKT1K6uuTJo3k?byVR8tF>ex{G)vIx<pX)cT*I8>`SximV
zT+8O2bAYT-b+NZyC9#C$zs+6FF6bjIeGPPk66TVZJ1e4P_~Pf=jVKi<9CepzUrYeo
z`0b(n#4K&iAP?=HVegKMJZy5eBr@a+U6pN;t1g}{A~{)5G`OO=TG@6<dJ?9Z_IK*~
zr;i_hL$Sd+bU8J%TQ_f(m6xB#ZafU22sP(?Ey&DOvSiPN73@B3Sa67wTHSuvt!Xs4
zmz}+i_g&<vQDOAX&2_MQj~yz|td{l7oqnQEqiMU`jlou_e~&2B{cxZnuH2n4tB}Xx
z5FBgOVPo8^uE)l%!vTQ$6WkNPj5QLatNffCQK6@)<Few%>ZLgcxOoHp1my9&zrBHf
zdo$&2_BuNs%pgGi^6vI8_Wn$ez-Z`lMX$1>>t*})jZ#vbt+uuWevkI>;nQn1;;f(}
zJ@*Ow+pN6H0g7R{Ge@R<9cCM5^7I>6g)_-IgqOTwY9FZFqtb(DZeylo{Hxj&$e;m2
z24i)Wdz?N)UIk=oe((#8+meT5I<8~%X_}}yxiIuOE&>aLIZ{~rttidbFiP3_RvkO~
zB#i-^aY;El#;aEUDm>oOQlzI+?@YS4X<_@PRY$v>Y2Ld{mM9O-TnLq|Pb4urhrzTI
z!VGYXox=`~vzRC+7foUcZT+JB>RzQ0yOvp04hJg2^WlZBY*yGpfz4*%yQ5L7Y!u%$
zPo{X|Sfw);JO>I_%~UZz<KXIyz^6K~19Q})8kHMgcXSs?oLiAe>WNOd4~PtWmsL(T
zmdXH_j?ES|6@!!e#^2anJfptS9cx(g##Hj3EPG@+3%t7-k9x`nN+}l-4#+6mjI*6|
zZ#{KANGhtjVq|&wXqS2&{p<5xWM%h8{JrcTP~WL@4}Wc2>ej(>RN+_;2rASZTi9aK
zrP0dyxw%uQz4DPjF+&Cq_9%PU+rXgDph2R&k7(p_v`BVi^T@|XqmEX0mvpa5ky|lx
z>x1c2b&6z2(ia|V4bySE+?*NxSH{gvlXevry?bX}J33rGU%z&0okynzhglABGmM;=
z3|vEfl30Z$IWVFS^hyJc5~16<m@WO<PoFY2c2~pi-%w~P0V0N~Is}aHv;ZL>RuZjf
zARB)g&|h-+jFiCWu$jGK|907W+T&D8$q8pVHjTdOaCRy8Uw+|)B@d_9ZaBGV;}6?u
zQ%~%yxM=!7&+M>?-dRn}UCtI0duQt2s1-#akWX|%W<R<I96ib?l~hoFh6syMi_Zs#
zud~H>cInow8OG%Jlk@JMs$G_CHa~ybh9qx?%F`roJWX(jpi4j+mf5&*L*?{IpRGSc
z?E0SYSw}$(S3}z7E133k3iqE-QdeK(DR0@Uea5(-J~OY>nlg%M>kQlKijFCHyKdiZ
zBC32orBkloTf(g~lIG?%!M{(mPOcj)u>y_xj>r+3ny*|#SDTv$)lDuczS?nI&eGQ*
zd-rCt`4}C<!ATT$hKd4pK~zoZ2hb}@U!tcn2U%O`;N!=pYj%?2{Sx>Rm?HL@GOq$O
z^tvlZRkzFRS&r*yxd?QNc5YqcmCbdtd+F-x>hDjPrKz^&i9J;_&Umn4`n`J$Hn7XO
z9(M0nw+Amk@z<t4A3wn4!ZGiB7po)K(Y<^39x^J{)XE)R`R&_Oq@#GYl`B`iOL86!
ziNt<lV`C#DLGM{juMLU$@w@%DLQFEo<3MD898KT8!GVFrU%y&Yw5+|9$(G03CeGMr
zsysg-Y2rVtPTyUq`tiXKqrns+-PeDA|8(x%Sly4UQ)?`GfZwhCT0!n}IQ1T~47==U
zx}8qAjh}O&B}%&|XPsePtrST`y&tF$m~V)#m(}lz{Os%|GAyE*Uo@s4S)}<wa-5>#
zdFtO!{y+BKJgVpZZTpSO5Sb+|L{da#sEidwG@wC>q70dmCPk%W%1|Vc6b(w0%wq#(
zN<tDPG@waBG)WnHUhn+wy`Q!A+JEdno@YI4?Ynia-+f)K)c5=OzTf9?9LIT_d5z$K
z-+7*}gdkze_Ikloo-}zyz<V|@G=bB=B9eWXw&Mr;^9ZeiJD)W^vAsN4>UMVaSL&$R
zUYp$pA^Euif86wCKDxeO-|V6{Pj^(5{b*l3x!+-{JonVjSL)<WH|_FaveZ7(?yWUK
z{^-<Z&cK+pecs-cmX>zKBvAK6ao8+#w?S0qqtm>HC@SinIXh83IyAZNTk8p-ri_Y9
zL@HEc<l6M6vP$cQ+8bTXJJH)W^xe$?=1VPGMz&{9s<{BjIFQ|WJ$4or7NVSUtuB0>
zdZ?}8yhyFzfK5jSQw}ua+ZuSPuh~VfkA_wOVf8yE3>$U|mB9D#tY<P+s}Hl8aln9e
zAvT`E3J)N-Y{e(Fd~eeipA%=#oLQMQZF5uYO1EW5Wn`OQJk8Iy7^m`FBkM-_=rzT?
zs@Dt}G>D3O!JL(c!ozQQ)UDFZqPrJIvd4^{(}o?2=l)i`x2!cB-z%%l;_rE&7g@DW
zp3G@jd^@bvM{2dhz`dSFDEl54q&C~+<lpqzfZT7x%UVtlVD`nTFGGBGtaVUoDW0I!
zG^O&%fIfYoR#yJH_I*sHhP%7_(T?*vZx0L!*+k5SI`Tj!S21D}3Me7JsCe>k$GP&C
zyqa7-`fIAG)$=@ZkDW`u`*g?X*;!dx=RjF1PGsN|7&b9wl7DjD(hCFH7Zw$P6=^aU
zAAes%J5sU0ZsQ=e&5|8jT-H<>U#X#0Kp~P*I}3e~UD}>eeFhDJn%|;(A|xaq0>Ki%
zm_?gEP=@A}3*XPq)B0)m0Gf)gUi7g@!AxyxcbTX6C4cz~A0_ENO##ynzYRP;NNR-Q
z%(9_J)fQbgHhmwrAmJqykJzw4bsB2OiY){}Wl=>`ME8E(wjB?XqCaN0isW7Jl~MMh
zo0{7(GC0_c8ds`up(JCLaVl4xoS%BplKkF3u{bcK>}P;_0{pk<(s>RJ=dFW;dTK#m
z0bX4mE2VeqeEj6c7bef%t;2keiuKL7kjT+q$jd^#?mArQIds;-lu;K;vA#6*R^E}8
z<}>7E&OiVBqwbiM%5+4sv1NDT^MR*r7mT^ww?_}-qirVnQ}_T3DE)3?bZlYjdKZ^k
zs^HjZ>K~RLvGE)GXXJ)wp%bT0E%1_kv(>e~i<{f}=O_1;q_reI-2o<uf|VunJJ7)~
zgZyN%JAn|oxZ@X-Dh(K-&CvB7-folGIOuO_hYn8H?#Ljw4@2bT?^;hBCl;Cb)y4Rd
zee3=ZGaH6AD*t*q19BlRv>yH!YTN3$$G3)t{d@Q3EPj!mqqbXHAW_!V*1m)HoubT$
z%b3rNd9x@#5+~Mpl~zU4Kx5ZITch`)GLr5CBzVi&7pnj1uiJTf3pC}-`oHl)cU-z4
zmobtRH-|QTOV&v=%%x_WT^RRa;nk3b5B2JHKqepoDTq2%e{8bN^?^dmqJKbrb+*$H
z#;4<E=Z>t-y`B#9(qO(uZRz;XvI|FTInh-fx;S}WvCZt|R!%Esm12`XKA5p&(w1ND
zr5{~Ur+mmdX2I(v2-unTo9J-7IpoUdp)*8{ceAhrfc^*G<MEF{bT2IOp~S3`JZCv*
z?lMx$gVHHt$17UAgtRD>xfdqO=cV>cPD%olHdvyPog4<40t=|UWPV-N=SfxBjyXw7
zOV>bw2hOwO=a!<e8&&AUe7fnMSU^JotL7if4my##z{UXt;X#V>x6u3_YKOzZMuz9K
z{pTNWviLYT&Gw}~L{<aU>pxpw)Aw$u&0SW=!QMQN^nK>BCjWf+a*yswH*ab@e!GMY
z`D4LQAWP!cMW+M4(|?}}3JbGqN^G|R9Rx_!h}*k(xu$PRj~-%qSDhYI*6^%lUAukJ
zK+o;=6MYOPdVKLW+AV}?p5*uPE1UI^UFWUVY+9*1;5QR-RW~BZaMshb4>3IILYvPC
zG!3y*mo7@}Il(ShY0{*HCiE7UapBQ@k0LK`=<g*<mOxoAzPP0C^u>z{$<uv&4;X!D
z^XwjDI>u%gfY=d1!q!gG?!l~UJsjoH!HpkZ%+|cMcIncnCT$Me!iBHsrO&;7N|-B`
zI<+>)NO{@g<>p7~zJI^&ZTitme?~Nb_)L@Bl%GF=KEl?HnaiOeG_wZjuEtgeWL8Wo
zTY^T9F1#=UT#$pUJbvH%_wONSuN(eEWWzA;bC!&pEpkGrSfyV^vjRu)mH%Qm;}XOu
z?=AJh<_G_!1yE|R%76U$vAd#TfmrFx=ju1k0sLaki&I$u7eEIi()_0(prYdwK1;lo
zC;pT44+SYRCNk@mgyY7I8@H7%CZVEsg0$6uO$NHr-=VB)((100?w|K@sdf)ApN3gO
zbZ5?toBuql&e^JxpGUJO7E2H`UftTfsfpdF2M<#ChoM+$_FOi7`c~_J+toc&`K<yH
z9O#SUsp;p>12`-lH=3iT$2i(qPd^FxH_oJiPbv;1NA~R-py7VeGx|MjiotH94W0Sb
zJki{HW#xx<wWQ>+i~$}i<iQE*hi15AzlQU9xJIO4();MwtMqi;F+Rh?J=$x(qJOW@
zo1xtU9{C%Sko~@1y1lxXhsY`!6UXO!aa1;ll+IG0#tF~*C-UCCdz|i*6TQEK!-aj_
z6i_eS`Sb@lNgb%r&dbJ_ez@bl?=I1BblUpIM`9-Mj2JICdg19jgsqgT?Gwj;k>o=j
zMIBdicO&@?uK4rwEkd!m>FXdE;V-+myT_=^qn)_Uur|MdMW1+IA^Fo+u6;V|xfn1>
z?by)+^cqtuHW8tZ6~ab}BE!bPM|tv$X%1%gx|5WYYQN88{GyxBuD<ZPs=f9?AbihZ
zaYTIR_SdzuU;z&Pge8PWKBaWl;zC5Rk;2@K2fm4vV!k%$Cel38?V<C5Fh%R>W~y!p
zDfyOY=K3yGnEi1HOVG^2hnBjbBBN?uLa&C=S}H0kRRt>BG6EzebPp8o6doK-_2)_S
zN?)p}u)^HR8pl4!$QW>S*x;{8>_aF$)lM5D>?(2Z^63A@4A#5_aqkvRP+c2nMw1KE
zlKfPNrF-{Wqcz*3(Scas!RT2p&d(ID9~Yfvo7|5O2N%}ZY`*ml(~E=kZ_zJWU-6~0
zWzK#2pI&sO#d3oN90>E$38H9^#3%UO>%Gx1fn(_x5)&usHo>+}{hbe=xT<iEB76Ik
ze_f#9E>A%oHuN<oBzcvkWmrx3a|{+_Ey0c5#<91LPfK86k6UxD%1L63yDeEe%TNv;
zuPUr+s#SY!iqPHBVTV2ED_h@m=Mx2G5k@uq`h+oMm743a)Kxi0hqG&*4Yx6QgjkiC
zAQUf51bKQa)b2q@KpPsJA60Qt>y}ZdVp;Dsh*W@b2?ICm@*XW!IfWFNqksOHytJ;f
zm`FHy7*z*H1iUE!*iX@XkGWl~)_1?&ym?PI-J{PRd7ZeZdf|@MntJO5h&P$L#h7T6
zHC00}VPjnw^@k!YR(EK7?W0rN&S;mzW*tjF1@cJ4*H;H}WO?)n!Cx;KJ!Z@W)Q1(F
z`M2mA)hc9SH}42Rc0@gUxa5i~0Ydr=7?5!OigW3#_;6T-85;zLbm+icB!Mr_4oiC#
z<zadC5i#GL`Hj*kqQ?F1USDaF?h|XF|21KOeB|6}!nx@$i%08u0abjSFYVtWO!u=l
zv$Z{Gt#N&@dXv)|V%&rK_xnw|1I;9C&8uo@I`ydxWEUBFE9!rpt9Ux8)&OiKc~z?F
zI~0A^w-zA&u@2!9ixz=0E63`0y6t$pRHhfLBC<)V|Ni}3J+cV4I5!Ii?%5_^L}$6#
zR2V*>lml1zr)^uc(p;~XVe!_6%`Qjw`1>D9y-;=`xVd!26Gnq3XT`e{IV%b!>uPK3
zRgQyiK){_8q^+Y1${Pj(7N$(iN?k`18ui+|$Qd{?xFP%1FwcqGcm7YhvC5h`F-%)6
z?6H$>lwUnrS^4qeAycPrrl+gj`)yY@*)un$HjNQw_>#0OOPVcAO`kJk@MQA%roh!%
zADG}Cw$g`?1+us9+rT<?d~vjUY~Q|Jti9ni0H8f?88BeLRG&XyA2Uv;q@*;jzH-a@
z8{t<^_60wPv^8<W2>oV9l=C#jSu^gH-3!?FqVoD+Vb7$OC-`)fb#z#ZOooG~@c9Dn
zjrCE>)wgc?xsgQNVa!3_+}`8E=A1%e*UxdKNB<?#x@+b34ZC%AHYjN7Jw5e?&#zSh
zql==7UQ|Xa$+$mBng6`<NF3{wfm6EncU{xJs8@A$_1i5!*DpgSu$)k^JTOX8)NMcd
zUp3waec9xZ3^ln`!v1}sTC$QK0Wrj9cuHcTux0@BCfl2a<B<aDH^Q_S8A<N1s=$ZY
zlH5j&2h!><eE$3yNyRlf-n%Uk9uUpZN(MJ1534M(43|D~XkMdWOApNCDPQwCbn>J^
zQ__MSJ2{$`SK{mIT|JgNu$`v14m52OnY6O1%5Hq<N!1wZ{=ezT554(!-13<mxI6P7
zT5ybEhBnt%a?1QTuOt6#CMwX*>7`H!NBlAld#$`*B~tz@002ZwP@1Cr(4oWUhLu%U
ztNe2Oa7S*x!%vgP4<81emK6TZ08Htj${@=?30!;f_vwU$gjGQnTPoF!yk1%+yT*S@
z8`ZSWDq>b&-5Tth=35+DGwd*#n(_HNGT)DCZoa;1$3~-=;R*^zQ!qjLv9=!hX2Jlx
zEz+)inWVD+{N$Ydk&!m&)Lsk~HS(Rf_L`oKx+~xr^YVvumf0~{FD6EkiO?t|PmLFq
zbmxa65<@~Ps91z9_GpKbJkt&<0?u3CI`iY8F>G;QU?4Fy)aPBN?daq;HTAFY%y!1-
z!BiGy4Dkp!$GZPz+tmHJIu&ZkCjo6y)bYV}?bHcf+4<#RIS|dcKNeiBY43xXRar{i
zw{K!~vYwtE6@=P%vE;V)=fwT3hm;Y>1~qAz0}3hHf+ZZtJmr67bi^TS_`DlpV`B0B
zGk@9KcWKA9UzSorR$M(?4jyx2;`f(5*Rs@QRMy@r`<t4Zc@LJVdLK^Zi0F{<>e1wh
z`>Q{G6w^970nbJ;_jo4u!xSWrW8>Z~6o=BS#Xc?Tq2)XyWGn71f4m!>bYViFlcXeG
zabiw9Z=KFQ>NE6WiLaRyNLOBrn44y;L;LpntA+E)qmaY7hB3;ffxxJeojM(SdXvX`
z$nH_^`?ZT60zmc~II!@fesA4?uVinF)uD<4>9N?jX;XKQ1HL~yz2;8G!FmZZKE>v2
zaB(U7`ka=%UHkTIayRkpj9mBXyXn6Nrw>RK4RVUd_8mL=4E6VCoTBpC5I3ps1~IE`
zZIx167b5`%(Gr6(8;<-8Q2eX*qIUlxmDyhyyR3Tb4(#Db086rnIr=fy>3Jne?5Y$e
zlEJE*!3);bgS~z%svKH*I!D)}5QQz;!1@Vhs-bh3`FOc5t9q@gi8K)^!Ma(=X4|%H
zVHQOLLuO?^eE4j!ZNZO{$YB|=IUR+>Ws&#2ge&qIK7Vc&AD<UijCcF?FLAfm*2l#u
zr8}xYdTyE;P#QREcF&ppWv83eyPwLTb7Ek@cC5w;QX<;Y2Q~CrFO~+S02Pk}j!C^3
zT)g2X#I1JYyzoU6qV!@uVLz5j<WsY{SSl59@L&xt@&E430j}d+WKIo#u+}wYpt^}P
z{!;614e%?+ENj}zN74=J-v#JFg;#ziW!DTdkL9h3`l}%#-udNFb#)&3rK-V~?8qP@
zU26}!+jFh!6_WITCu4_NP&K0dd?q*8uF=Msi_BI6yVPw23p?QTBMNq3B2H?-{44%*
z3-B1FOKn!|RCyhJ_8Z^~u_Ey0I8~#&#~kOqoPOtdWz|$e;PB4JAC0*f9musGhZ;TR
zfEFr(M~&zD^yu*o8#~KapTQKmtnz_U`2){W-n)h)f{H-B?|AhAn7WwFKPV><=5po3
zBZc9_gMX@YO^!p$6<1K&L3{Vo9ft^3$q!N}M5Q1hUiEPYwYT5wI0)0+zeTMQhlfK?
zUSIot&|D$*Gc-Ijd=8f~S~I|sab2Eg7i%YdT=A%fVjS+Yo&hq%*;gqQUgJHN>D50c
z1_un7IPYs{|C8cjLdrGBj5Us95dgG6*D#lvT;g;Bw~*l1$|_1S8S#u_rD2|OfMLu!
ztr)(mZ`6p6P=v%D>#nbuO8zl>W@`qSIdkWonZBY5WWe{_?ad>c`Y-Wbtxx&gdEl&V
z2mh}V^y(D9SqR0M?dvn*GqJ%3E?K6+hD^YnR7&4kN$1DyNnd;?NIVYuA<>;zG%w)<
z;IgJkWyyk>>v*G8W&R=bGs8F@v>w6Lr)x_y`WD*r1o$Y->o4xuu>&tY<+`j0ad<Y1
zGX~ezceG_L)O^F9hs1+!G;B0vo{bNu%I}Y$Fm&&ml$_ju;IN!#@yjV>OdRDG1~&B2
z9XE0$6C#V4A2DtTL+QYvX$G%jSmce_hf2kht)S^IN){{PLJllEjJQPLtJF136$bpp
z_*d6O@8z&^l&WeS1zJDRW6fE_%#lc4gk-PNH3#EnzePl|D0ut}CZ0}+IX#|=--T|J
zH;DJxc~kUmz*#5x!?dL<XfyE#G5MTM$we1IeQuI^sPBh_rE6Uq?s@_?TpSnFR!N>K
z0l^1zdlN#rYK)bQ&0A(O1u61!OKMP1kl^GaC(i1z6R<Hf|85`MoYtofvPWQ)l9V)#
zROu?-9aOd$`aS&iAhSI$udFu8S~*#b@-B)yTq~sml!=CZG8|gcc100<s078?dwuXq
zRqsab2g4E9U_--x2NE!moPGrI8?Sf$%{U=Fva_oq(Uu!VNN#^uT`i1@82(IOoN-1i
zdCh?s2HLsc!;B(OGoP%?JL;(DC3Y^g{N8ly{GOp)&rG$pZQJgfh<N<ZxA!w{*>onN
zTmeakQa%s+T{TCH0E5acGE8kV@bIjOh6V=Bz>Iu0pGr$(UOl5p0EcCx&r@*FwF8%N
zmOaz1`?P#0FJD`1u%^jv_T<^Fj<6=#KNLbIIWN!IYwUtsK6_Afy|;{^A0*O>JD0S5
zdwJ*Mvn(33;oVkS$&cvw9HHNw=szV-CEkq!yFfBgRm*IFxc}XX8t~ye`8Jt}R?`cS
zLnU==ZOe#WSz>r;jLn`oLvTi-*{R4Nnpli+63b>06XgGzq@`v3$kH#mA4*@)vxyUF
zTh1&RKW`Cpi;Xav<Hn8SSBbj>zfmvtH8Vh#`D5}UO~eO)^KAHFw6Q{nY%SfCqY-}$
zCjjxKrB%%Q1hZe9wW^uBw#(W$3FSDm&bhf}Km>F%4{df+ovjUD&~V7D(Bx2rE~CPf
zoYXQDC_Lo4a2=CBsa+){!3jsdI^Ty6ZIsP;5O#<S!LbL8RUm7q?`Z*&qX;(+AwF1Y
zChUtqUBGd<pPr5@pr?lipxk++I2~esl`+9MxNqOnN@QP??kY#plofyw98tt8PBqZe
z6U`07RzR~1lYwwHT$bo8e(;=*U#qJ*l;@wml3D2LT91_hPPE*a`DpG34<0OKqp6Qb
zjCnNWF-?$K=fd9>PUkxqZbOgHpcjLGOop{LtkzX}BB_WAGI40n-o4HnHYih?lSj^W
zU62ilfFi$-FF|4Kq=jHfJoQ9G>1HJr6&rsyySZ84&S77C{N&;!`V=UAm9nJ@e4B_2
z2y4+jz~UpfT(CiFjo-d~EbH9{|IF*+x61`1buxV6EBKGMCth-NAZ9i(pw4JO(n%i9
z^IUo8*rH1-cFReU;;ET}zlsqNUg)>l+B4R-l7HnoiUlR`!EnxZ^7C(y18X;Z<veA5
zRYYf()icZ`S_$3P)L~-mkM=Wr@hhsO?cM_!1cwPIEh$0pjE>40#R%nI#ZPrNW?7Mu
z-`U>#5!h2~<1z}QuGLk&+r6i50la+t^r<kw<t&sB_m7K<qXrKfoDg}oG$1Ur>2bzQ
z(>0VT1m0D@X13E7f^fA}i|X!Mj1`ggxC*W0pn(JJ_4^4E@G~z8gZSNaqH(Qf5?Q=d
z=^TCiy#WCMOm97XijMy5fddDAsjmAB6!v^T!C6K&!{%Rp__j|FNRjgS<d5QefyVXp
zoFM#aYU0h}gM0@^bG!s3Rx9i)j-x7nyUF(wW<zYRQ0_c;m~f-sZ{f3G!{WvYrWT;$
z{O(c#kC?T4(m0)KID0^lfV3cXdQ?!5f<&srSD(jgzc10$)TBUPb}^41FMTMO5AG}J
zaAT*;k)XDcQd09ncJJDCp855_9c}&`n1ASpL{!$^m>j9l^3t;=t{Pgizd%s>C5++_
zT`cW4aG{R3%@-JxaM|1zW1|7=yji_tuWM(UoCD9a%NhrTX3kdYdrDrt|8JYywazb<
zrVNUVxy~(1AMX9?;<BgRsxA$+Dc9@K?og(JjYN`?QghL<=)os;=rs<~Kbze1bwEE9
zeQ_5r48uXs$4|_vG9g?WMC@T;FmF+fMg+6@$}LxjpSTS>n~x2X?s<gji4XQ=fWd}1
z@v}x*Ue)^p$thN;AMev;YsO6K0_k;Qd@$^*Em;0;4QT`HtTqT6uLe9SoJ=RZ`Qu!4
zOupZVp$ZBLpf<Ct#rhHu?PC4+61v@5N9EVCrqXZ3@+ZALrzl5c%(b>IMkyFOt*Ee&
zVetWmGKZDt5V;SvG*JsGpX$CbnU^wjAhRlAWsg3CGhTIiZ6$Fr9ld(F&nmj{uVp)8
z!ZkOWYh_v^lkKyie)5u>uf@{8C1uMP95iHZJH!ifJKt&HQpUD5xV=gw6DR7}Zk1_>
z@bn6bl-22ViOimJ)<XV>J)g9h)%V?VL45=+^ZPYrIiX<6xIx#%V^OLwsi~`5p)_Ud
z-JZRAjRzp`$b>!xgZp~ceb7Wd{|vCZi%ye|=*+KAnW#HWXSUv)Ijmj3bn_&a`1wvV
ziFVr0mc(|&C%wncXV0FInKNfc;kTHxw?@OfENv$?DRC3;T<smM;W2*PxSU^!ln>l9
z6}uTn{I={22t0Od>WT+lng*QWY?2Bv$GN?t9EjNcV{Ay7;(>dc9D5G=sv(@|JbTZX
zYVx0_0S00<&U~1h-0ekoBy@X=&tj<v$TWga14_m<z;_s`K6!}*g8Th+|Lk+5iRVkQ
z=>!QVn;bOk#5-|va~o~<;$2(!vw&s+jzd->JbQ@+&+09OSbTrz5EEay$x*z>ccrEG
zGGxZi5{73{d+hKOq@+DA>w6IVL#8IBH{T;6@ydjb{kW`-mfx1fQ<bwNZ;wDVzf~2K
z{UpUyg|HLe=rHzTkJ()%-pypZT&zZI_f)$@X`s4<#Gj9?-we(`hPA_e>E-*lHWJ6<
z5H>vR*2cV)fB9yVF>>&K{Dvg{+l1-wllWijKs9kSOA5>rX#n`}j^oWPaq%w1i`rw%
zKqsHWk?tDB#_HV?N(^hcZtNlv1?Cxi{P<_Fd4;d(CO!N2PbNXGp1mzY`Ve-7f4`#q
zg{k=8ujE#fk=B3yfBWOdS$Lyv-NnhE-k8$5KSm9R8STiy5W0D^-jl>mmv^rC7dUT}
zboctttKo-!o5r#?{GVoe=u<4fB!u^2Z-m64)&Fhn)%wW)*B+n$_DpMi$p1TH)eoP~
zMXRSvTNHe&Su{mnUT*tVlkF4z_S}k^Y4%zDNIyfJQTm@R4a<48|B}LkHTEW#%3m&A
ze)dxP^x}*DaXr;G9+tng=XN#Dcp;x}sb2Z~T&AJr=gHMiCvR9k;9c02e2f~mj{3hp
z&4>S!1K$6~KUQ5qMB%QUtAlGIs8m%t)s*O-JA&=T>o;#6HEeQn`ay;7KWvitE}s!*
zBZDuk5P##_Mv7|2h7auBJK{_y;mSt7j1D93N<5os!gd%ONSVv}%o0zfyPc>xtae@z
z{*qyT<IC9XkRpBZ%)rr6pzEp+;6FvR?kSOw7`g_ifY{(~dj`#9*wJsN#ZUE<N?Rv(
zPmOKst7BpoywJ~S*;M{j?tx4U|4(AyPz{Zn^vS-*Pm=OQwSVxwu;|oVuHG8kN&#v%
zlFK?z?qrA`u7q*ne_lw~-F&y!B{W-0X8nELvmmKjSfO(6wokB{{i|;!>HfL{aQ5W|
z=C!&G?yFz{#(HpsZM&~XlY%7_TlcO=NURv?G_vsiec@-XJAf4{5^}968nHH69v$pi
ziGL_^-&a(am$9rxLc;Ptui+GVoB8U?692?k@*cMaweHH1*m|&!+z<x`2jOI0B8efZ
z#4{a{^7#+m-|loYH#hIl^&@iSn3xsg*p4M3vF7hVxc!jQIUie)9lMNKvcv1!^*6wm
zsw$0kSg}HU&?Qo>KM!P5{Ja`BtZtt^otOO-&OkKfmmY|pSJwJqByRtp!z}h`@z$I=
z|L@NeniT$;sdjc*;*K3Nz>oc>_Tx|Fx{Jz3LLyd2)E?pCbzfelY;u&6^6#}roHOI0
z!kJbg%51EvDq>NK)S~9Q-C`~NOAEjuX?sWfb&tQ#MQL>HxlbikRs9k8jBou9cEZaT
zlTsNlcxscpoD<~2VZ(m);ni;Kl|Cj){A~j%U!9!OMuTN#>kD^D^9M>gor4WlvetyJ
zcP)z<tOXeg+Kmg#AH<;{PAnCYkVyFZljTWTICvb?xNj=nNx%00c|(TicNGg-7xLUz
zda{qFcJl}`a>&c~i#+T&%+b-&S109<Ey57ERp;u$$t!f*OWZzoPEcth7mVL1J$;1k
zdRGZ0Nn8ytUL3ok)oxW{dd@<Y;nFGbV~h<<a>d8h`nBlqTNVEtsN4D}{r4aLAM{LC
zJrF1PT|+5|-j^?5_}|LPH7i%fn#Lpd;8fr$!5|mBRU}S&E%xAh&Tejbc68uPA5Ta6
zsa1I`T)1#fgvO#mfR<G7wxE6Y<<JnIXbw&E#>9uAcE!QnB_?-RVheuW-1v1&^-VB9
zupMbbuh<>?_s<3D01Ge&XtA=gQdUl2`X0oIi4Nw;6lrZ<wP`Ey?MQ1#i0aFK2R7OS
z-~d_&ik6#G#{~rKndDiI5nd%TEKI|?W@|7hfO>z8I&+X$%U)DsPt)T@3!G}K#(pgv
z>70K8de$4o(7OqfahKV`QVpnLyYKEBa{=Oc2d9m)fx#L6=L-cGZuyFeim{0+RHy7v
z8Ba?U6rT6GZ28`7cf=zMAqlA%ea`1;ElmdkBB&x?(-8xlv*+bZ_7hZ2jkUEw8i$V^
zI~Lyhg$@&E_M=Dp4IbPAasfcQU4vLw?Am%^YE3{}cr&0xZ;FepBmdw;B-SB4i@HN2
z9UK^Vk*1ud2H%yCSUU60tJkl817{LTT@b4eMBKk5Xvq5pggUC3WfvSq23Ns3)3OTB
zpootqH~(ee^b!;N0DvQ<J9{@X<?7AO0{Xmg2yFUhSPM7Se5d&!M()~qgr??i9to`}
z5C1D8T_jNLdu=4#=bH>5hD;9u%)$@k>CtgBvIi#1xOKoNCrNiuKR*NBzH8YtBpVFz
zCM9|7P2X`NQ0D}^*8@o0pS*EUK}$=^{!`@-2a@r?J@m2%dJZx)e2iXX0^WBI9-MkO
z<MO3THI0oe3~7@a-GEww(iSaTDBZM3Lc;w^px8PVxN7-w;Yk2YJZ0+C0aIn9V;B*w
ztQH8~jF1g&+pk!)>O0WDPG8^2Q>OHvDsy-vNUNBHJ~6_52U;P(pfPZn0(z(kXQx|!
z+t!%bsk5&<))m5eXrr8xHD~ou;R&<Hh@SkX$-HYpbO_6PlC1ZjTf5jL9856G#f7VD
zzJJ#u2BTxLEF9ieB1$c00=Z`EP6SJ3$Bto@G>s!d(?gIwVKx%S2EL$DzkK|-XuNjx
zx7sCs=S7<hD04p}qwgk-t(H(4TnbjG+pg%iTn&wv;EQggdHnRZ!?$J%&@cmly?ggw
zv3Bjp($W_wg=JSRz><`S&^ILkj|vODU~8{mzYd(HQoW&#gz=QtWpU4B@GfDGoUE*{
z4&+547}0n5N*s8_@B`OJSQX;#lP)u?v*8J2I-`#-UKADz`UXwni0Z&^r~LNrUBoVl
zVxPHl=dynN83!Lw?P1q{r)~L#ya7p&xqO==Q+UaoEwzo(io=za1#Jw5K@)W!x>&tC
zit@E!w;ZpKfj@9&=cvC!G1DPjWz?w8h*P%9aYH!euL}y8Ov7C#|H&K+zjr)ePhfar
zLf0u&YmOaS%gL5|_wU>C(}oRue(kGus@G(q0YpI6(f7>4a)3cdZ|vWn1>P`h-tmPR
z2mgrWKm-&wS65djr?C?!8bNzA0P5%0xqEkuk-@r~&*Q`aog=2^kQnIe>w{8i@lgo7
zn?sXOj(B=<>MY`n@4S5ZlF<3&$tb{70tT`pdHE84q3)ohg|JL$@uOFN;A9@YtCkG%
z%NG<CU#9Vxw^D{qMl=Ha+%qB_V)gCYw=Z5?;ps#D5`%C(3;N9D7G8xPDLDyD@7^tC
zm|*SNP)IG9oUpKFt`x?>{HN`{@2je?Ett>fIM<20`<sf88Obqk8uy+%w;qKAF0BL$
z`}OPjyUeFyt_pK`-ZZ|mY8~?diid<m8exubwM*a48)DRv;cYwbiizXK#V<%W5)u;9
zoCVZ$EHF@%msm>?Fw({zOb9O1J@tdDk-^+9lP~dR57N_peSL9%lUW)85(DjzRu;3+
z*?i5M$=;Wg+a|Q($fv=v(n%UYvUskr)IeB8^2yA-JOeFim$@+*fLQ<9H3~cupqX7l
z!UG`Fq_gjGt_PL_j$evoC2kk&6NTOF7BO$kLoGW&Acs+8@<=B;9znR+A4sqfGOZ<B
zTgW~yne%0QQEWuX>;};PH~W-t`=?{)&~}(O@y-0kkI2r<Gb_OGg{M7n0+{{TxSOyU
z;8rlfd<i?c>J71seeBxh00v9B6=e9wf2XZ(2x$U8u{{YA$Y|0)9q=eD83Igv<ZS<6
z(0s(Ld7HZMgL<8=1qVrM`LkIz1}6(KGJ=&Cfnd*pk-?8hoWB~=T9&Y-6A~c6<3CB&
z;VlLl$*#KqV!5@v+}Na94PdH+5O=O>LFgD~^IC8>McW_s{`cP`o+z)SGO`#?XyK*_
zcXoByfq;NNVhsnX-r<UhIBv39lHIrGE^!c>Rz-77r=e)(D6=do1B9k<h?dqLuI98~
z-|lg{>h~=&!s~{qO)mf8vK$_G*#IwEDt2fOqV!Gm^2qE)OM&796GLZr_rw$LgTLv{
znxE=8Z|>j1qEE(mbiJLnFI~G<ELjtiWr$iVy%At)qVeD;eEc_$1CD~$kV5s`cTPGi
zq2i=TlaPC)r?2FRmt5kc^E-0vG~l{W@SvwW@$;rZtP9$2enN1eaV`SY>>q9jenf~v
zP=<|8!sHDVdgtc|Q8&eO3(lv~ng5dD!s8?)c99eIUb^sB!q0Hcnz+K%KEn9U`2o{X
zd$SwLGJDI;;L%=1So!;c_*6L$d9t(6Z7OEqOt0T#C&`LS`f-vjGRJDWT7qEiw09b)
zja@#;=WStzJ#!mr<;8FJ_4fAET<QPJ(9!Yo@>2M~4Dq%yGm0|!yplgvVV$NNW``Ku
zSIM@P-(P*ocQG3l!G*|P{gqk)hT&Py>9A=E9k*aexLpncc?hO~US^92F3NUS<GLZN
zwwj&3TAjHfJh?^$8_3Sr^r*d$7juuD2(W5$cuyaXzRk^bMf4c-wcWT_Q+6J%O0vK9
z1~1+blfP30OqqpAq_)f9>M8H=)8Sn-&pa=N#R#Dvy;55i)b{;%bn@HdmadC+y}i6>
zQDUe5-~j5KI;}o=(xl2CAEP!hrwN_Sz?s3I-$KpdRa^d0vqwWwTYZ7*4-9hLTSs4=
zsrdw<5s<r%mW)K~3k0riHkDg%EYdR6rDiQ2Id9%PoMQqt#Eh4z>5HLb$A07G@e$|w
z?13A?6^>8e;-aeX#o@&GSMNJb(1*fO$X_b6Ys$JpE`D^ld5@jQ2-c2Hu)x{*F(rXw
zUs)`cGlC3*seuqCSU7K{3v|8Ei;24%OtmNCWVP_JsM+q_y<49a{HeM+m7Iy>vHY04
zwAeUEHd0KRXuS3Hef%a_3z)n*dr;<Svn-Mlzn<^S^PUA|3zdEor_Wsv-a)bT?>}Wj
z#hlDd+mGs;pjNV>tcI!Rsu|)C^wy<EJMGJ_{NFIvLILpc)2D=%MJS@YsI!R-UYn2n
z(mi!lZT*{iz>kvjQp;3OeGu-CmDp}NF-7VXIR!l}svuDq4UF4Tb|$qf<>)O5(W|2p
z&YZzoprE&DZ6w}hwKn?iY<LVIhE+mJJ$(4-ouxh36(-gIVz?^gmmNKJtOXi?k>~NQ
zOF<XV^ClP>`RWMi9sXxCj7}U&N@{|-V-(1Lqh+R_3OifS`%AZ5hMJ%x29lZAedzdl
z9!1Hl`SV}!Nq^k)Ubsh1EKb@m`||yat!@4hhM()#pIf0b&R0kH85dbq_>uaBas2pn
zo2wXjjwMef#Xu_Mk;dyoGaft;4m2fxDpOv)d9!Qh&WA-4UfjX^hQpz|VAhPRXH$C%
zznW-W|0h{lT{TU}X)JBYk24u6>uvBye*PSC#P_dnFZ?X{$Ab~f6DxchXc3SKpU&10
zqe!sorycin2vR<`fmBm63GBo><lw;}9(oRG^Er-7b43;UBP2k+q9p9PVm8r;+qP?%
zsXE(jexPxIQGDUEC$Z#WH?kFVJ1WT|hN*GH%aN=q=NlL?(3?9qD>L)bwQJ{0WBo9u
zE7O^`dGk*`%Rb9It68LUaPfN<WwvSPa1_Z!_(aEu#CsH?q#-4T1N-(-8=m|jBLq9V
z%d+Kb7^O~J;xZx`pQplJ4Y0t71QB&@*ic7Q)dz;Fc*1=b`dpU|mxS)}a#eCz^>N2J
zbDnSjMAJt}5fS0AY#D>&>0F!%D86e4<+001s8ukUv%7}oVFNgqXNOUe+d**tgd_UL
zw!P4_n%In;4VeV8n@Sj*O+R4=&LS9?W5V+|I{DhLzWVm|XA_0XlZ)d>?}b)YUrE^9
zMiK5}W+gq8?yZ?-8?}mTKmmtO^enG66no6d<ob-<aXz#;fo+{gwT3z{^fu<y>ZubZ
zVC)zkwCN{en~(3_$=#STNMh^6UZRfQ>f_<*X>gr=Z8A+Sx3?D+Z<6{&L=dEUJg)I_
zW{lXL`5D{E-8hs)NBT;ml|NQ(WgMR*Cb;_0@hkDzWdXpb*I7*6z*mtI`wnJ3)Gu5J
z+3z6RzkhRWxuJ7M;wyRbg5@<_ap!${MwQx3>v8hbss8=1pNxu<%Ut}8w3FL{!Pp%d
zp!?S%$T`2McM~RF&FwXC;HY^qtb7)O<4h;@YHVS>&i!#WJsS)zk8K9k8<kX#2<#l1
zU|7Q^>(=I-$wQWP&E*j%i$P*)PR~=`>KiIaMudECpirc6_!Q$#sSue==`Bz{PA3=U
z(D4^pS?l>rC6-@~9617?ABjHQ{3oHL{K!4*YX>skNgN#+e1<*(<T0pF%49cWr?5gT
z{6p18k;IVum<#FLKrU5D*FFe2amK^_-0_)*V*7UOsO6{E2D<k(CrYr_ENM&@Pl792
zn6fFGPJ*8+;<(o;vZxv!s_1t2bpN-JD`_?|4;dOc`K;MUhUR$7%CE(sM~PZ8;-r9r
z$>>=4Y?>)_P@w$!{kviDCF`-|*TeQqta^KTdX5XP2b2n0>*}lXUc}8^1t4CX_F09=
zN?6agAJ_S2e8AvoDoyd+p2GtuyT{M(E(G((w6jOgDk>@}pU`xZ$t6OS`qZgYrXUru
zs+Je4ww(L9q_o8I7)_K{z9dfEzdtj{{eB%Eno&mE)pMCN2zp^4V>aTIObjhm&+c>#
zIg58rMUayj$}1i@`E+pb0v(<0iD_kLHfr|(g1aAnV|~G}>eq;OGxZP&8fEU=NZwKS
z2}EI>aJo<>z>bfr_uGCKZ`|nnR(mZ9yRsFz%N-m9_~&oric|Et2<*q8--G-O)_oBu
zF$I0uAm1{X7@Do#jZZQ&O&k}TNxy%;4B=hLANrs^r>}G^TCO&Btg(xa<uy*$;)E82
zX-fGXL@Uc1M%<$jO*<3J&QSCE^_@H56CaAzzB7E(cHX)(GB#$EI1Z|>F0wjb$0@ja
zek~;kt!>CBo1>#=VYoSB@@kxZs%xjVFJ5jo*L&(rQ$|j<4eqwp)PMA>HnPB@)3s^d
zi_OHG@~>P`+dJ&xlhw-KPZ#|#D{6IrZ?b!kYJ#DWUWZ{Jmd(_r>A&4h<}kgqYPtNS
z)AMLa%665doQ~{9Srl@7zRS2(ZJt+Bf&~wZzSiX?Nr^jN;!twjF@hpw&rjS<^^Sk0
z%}KTGg5IR3Q3_6t&PK7OBO1#)Y4>;<p{YDb{`{|~q5B8P%8qVI9eI+L(LWosCidF4
zF9;PBVwg}3^jay5$exFA5@|4OjWWHL=am&#l{~L<SzAw!wXWFWCuoyckjTJt=2JTk
zvGkZOcwwKm{XlPJsA7Wqs0-?!vOp$=*0@c0NyqfOl2}>t^|0jlMiaA<3uqUwc~U?v
zxcq_oS|M$;fP*vtoY6TcguC=z$`Qu#W*g~_8ei$0%6aSYK{jc(*!?X1`Mrcv7LRv|
zu|Q(qHOPf7?bB`<a9m09^X+mphqRRCl<p-nPz`nV)rsAd=?mI0GT6X*GkveEwv%=b
z;{C>--`C4-m@;XSJ5qm7&o{K+=zgrMKCG-#od12hob&>73zrQh+A-1omiG3jA!nz5
zI?JK0Dlb=TYVS*P>TDMep~CfM<}z_rjBWT%G@A5AcIshAkGAX3!EY+2<C(!mJ=?u!
z7`5c{GNk0^&YhzHDxbAB3@5X)pCfHpunoEh<OTo@62w=|oP*RzHX8Tu-n~m+sdEKL
z4?5JbW5>6*Hd_>K!S?}djAu44QnLK}-rzzD^Di9vpfB{ZFRz^Uy>g`v_p|zsp{rRi
zlN~uN-yZ+r;IkjcNZ#e!Hf-3C{zVM5V7rNx@%`O2(O2-~J|~lzY50yCz|uKSfN|%|
zAdbLM_rEZCV>#NzF`ZzwdgaQ!1>EEb02}gWtD}oGrq`YkC<ik_GI@IySSI;UcKn*z
zHk1q!tx(?|#T(;JO}b%PTV{6<8TEXJhe&`}R<eHm`fG1Mg^&~}zBss?X-rhF2dB0c
zlWLySAY5u?hYKKSq?b2vAy`;dZrC`tTeohzy7jzr_3CG4hY~gE(Xbb@+z<G7WRwMi
zJdu0Wc}Q({qWxzcWsglf=_pGrIX-?Zy=wYB^7+ttQ<2~4YdIMj8YWq0ZkN-0`R4l7
ztHRTkGg3CIeDD7KMa;>i7O8Pp8>kaN54lUY!0gjr(O%oh$?4ggT!U2wlmaNz3cFBF
z?~YjOnsnhXR_xX;<(DltZoHnTacgnFwQs2Zr?AH5+ymjM!4>YiY2KVUxtB79@vq{G
z%94p^E~$)9pY0N!A}0wL8f2gUCio7u1gJvcXka8hYM0@^sUH1C?$=_fG;@g&BUae;
zZf`H;S2k+Oo*VHGhbE;7Ze#3UYzVaKp#3Ii=W*4K419HlELy06W|H#DCFZa}xv2ji
zYfha!`D5t&=eKU0DKv4u3ZetDR!6~KzH#=Gkg_eicA0+dIg+N-JH`EH7OC(YM#77e
zhbSqn;iv?C-7OBjN-dmWKUOSTzl^8A;pjfE0;Cz5WD~pieJLG!AGrm?wJw|l<NcKM
zZ@eYfhRquBF%RdYUbSBAyw^@WsdMReIcT0G*&8x7V?9^BSO3`L)qkYCyjZgb3S{}k
z<zT`ltKeg!*$Z-c$tV2@CB%@2d8I)GXBV!UZJyNlp7TyQrMSs7X?UR=*9a%vlHXl&
zD>(Bc+Pjg8Hd7{C4J=&yFJ;ov`o61RrHbddEEo^*cKX^a>rPj0+z5H1MbU(a>Qg_U
z>M^g+e{hK0E{AkW`d!@1wpHL_Ojd0{8r=Tr390RJN#UnX)Ss=4K63oH7nQ2>Pf?#)
zSgjy;mwaB0?oZ9IpP{+yYxKR)f%m^Or<pcC+i-o)9?9Pygl~?3Tn^3XJ9EtWvc_pB
z9bFF}6&YpJ<ss%fjn?L*P9c0%R0w~67%El(yl8Pvo_3NJqonbqoFpiDSi-~o?gvK<
zqcos$3)=0fa7R7UyZcAN_>@_hnnRi~sbH2YrQD?wL@CHB+PX|3EkF7@mn3K}f)LZ~
zt?Xz^GVa-hC#d=jRWd=l0b8HjhccVIEGWH0=B904@8@SH8x5J5_A(AE=#x^2$@0aE
z7lR><abO1l;QU<Par$nBO+9IRaM(bC+vmpV*jQl`1q_#O*S&LRQ^(|mET{lG-_sG>
z)(-=G-tYBy)C)4y@*@wNLmIdE%gI26w{A_T1%`iq%6n=>HkrnofumCWvrsjB@vWXk
z_uMuqJ#mX_$1R!ZkeMVW8FWmu(x`ER>BGhoF)^j+<`fOPvkVA0#_Y1S`M926lg^wn
zjd%4i>BNWLN_1}ZK}E(7HgKqzfJHE&#IJ6sL9A)Q%m(70pT>US4<)kAby60mXiAmi
z!@S=4KDSO%@dw1XB&e&`T@-iM@n+};F<d#uCV)=cs>Y6jib7lSp6bKdgS{1+o*xYj
z6$<vKc&5xIV|JCPn<|c^ny09V%;I_s9{d(eobJF**1NWBZ<*G99h0;6c_&vtnBQ_H
z1H0#g%GVEZvH)NhK&2Tk_jY0qkJuLR^(r-1iD!S%5>|HrkOs}5Caf>4yC@lH`djrO
zJq6fVuD!{a^G5ULZQ&<?z15F6IHYe<b@zCF)c*bZ6Ms|mhwWW8)Y{f>Z-DD{5ww2l
zCvSSJ^kn=ms?%Q#J@Xxu{FqdHyVu<<*LI8x4Q){4uXfsrZl8|)nMi%<uG&gS{IDP;
zVxpj!|Ca123=$PW+Su`!;bosbeQNoIt{d(vNV(to<;zW(WXzlqq&QRoXjJB1+^P=e
z;9`Ojf>$hGPOMNYPRo7sCi&>|N|!!MI^TZqpn~86*tO%wGtM=5qBRCFS+i7Cuh69<
ziJfyf`Z4ed;zUIwX*Mdn#>^uWyrEzJ@8qbive8!5l8OZu=PcC)8TP2)%^S-|HBY)R
z+eKC2vtYIQ-Eutdx*4gw)*?Lb7<pG}dlYK;?8r~AdYK!!9Q@hbT=nYa32OJCFP44V
z@Wc#86J)@9DTP(~UTCp0VRzS8mc1~Zk-i&yWR@z)H&%=p1{D{WqE7dFK4@P%iLK9D
zMF+QAeqf_tmij<!T!EpG$^7oz-U*#CorYh57@rS$Z%4^%o3oc%@!^7PeVM$fZEWm2
zJS9)faLYPm8wSN`iWcUbx3j>gUFO<WoZY@dJmY`WR&96nAG>s`ZT}@bc0$pOT9Ry*
zV6r4XmMYQm3&0@FxXbFjetztJlfJNJzrVk-m*w@qIdkVK8b;5aHH*cjI^#3rKKO4g
zf7A!vxN}OtoRW@%y`r+85T6sBP5^3_(0|?dlx7rfGE-M(SX{<`X#sSh0QNRMymsS;
znB=kz9-`fYqd(&H7Iv^X`<tkYk2Ux5`Xxlij3?y@Q3Ie9?nId;&1rqqudT^hPMd}|
zK*<c6{5518^k!{NjlLKa6$PrOKM1S3Jm<`|67S}ww+cvF=w4*$8G|&;h%DH&>6t%M
zY2#vefcOKuJIrCCsw#8(Uln$<95pwp>K3kA;kGq?#JL-5a&%lC63puUjJ7D5uCj83
z#jL16Kc^ANmW<P4*fP=sH-ovPe9QfmRuSwi3Jv|PbA&l<*<%ev7=u%@F;bOFvbLIs
zLRczED()z}hHd+kIK1gqJOkW(bA5e6_7j&I>t_WFT!a^$a|(DhWoSd_q8>Z>Iho&$
zSOb{-Z#!*cC20gT&5dJv5ea!qhcv{px6JLjT-rfcS(o*wQB2>{H`LUy`rIUhu9*qT
zVH$}&t1B0AU%00}`W4S04_-n>1wA>TEbRx1A{^At-y_#kv)C+0yJBqc_o@!&w}5B`
z8A=3?vC&dcuqAlA?Cq6#cKJap3`8q_{|lc2qHi%|Z4XKGiUJ6iImOewr(@cu9SG|n
zC}*x!T&41?(o{_s<pF5s&6Y=fuAE<V<y2w~h|zCENL@P{4JL@f2#<L)xld<ZW8=WG
z6ENPmg3Kt?A}BhqNIm8+%Kk!_crggzedZfC)s5zwmTmO(jYbnQpMTr_L|Iu9r{21<
zsx|BQ)QqgpdCle3(KywA`}WMfzNQ8mMbu<ks@UNJ1@!DW#9^?y-C+h#Q|}>a(gL^V
zqdCjMk8D+$A(<0*$Vr0e!aStBe1WZSmiR>SO31C;zZ^R?s*Bv9Uvg60pFs)LEV#sx
zYQON5CN2zs7fW0wmJ(pvwQI*9Vd1mMQXIeWVpkm>7!Q2-ruEF#u`_ddCVHmn@j9&<
zrS*KW!fuLDTm_HZzltNqxI8c~;~misl6ZIq0N0J2Su8j7#t$%c7o|C-$F_rB78Fcn
zry7j|&45os6?R*zdF162w%i0%q=4+IxtWF&NQi00dJ;Uv3br0IXWw<kGAXPjIN=K@
z;xF1Ux<aK3v1Dx$YoGkHN<xCAf4+zfRCoAP!@fE(Ad|Uz%k8agC|ZsT6bMa-Ev5;*
z>wq4rs;_YCI3s?C-q8S00(OT^Qb?O7_%{v#cNMUwDy{b^Z6-x{S9n22=T(3C0`x9c
zOO_V}h>cBng1UWHODJbL8_1=SanK*eRQbRi-e9=Auo2RJ;M*N(n(}{qVr*26jsQ&(
zkPU7c&IGZNx#pusjS?yWpgEpEyv6;uH$7fN9g^8gbiy)s!DS5-;t4vZpIi{QR^XRH
zW<of!F4<3rJQ0W$^41jk&-r+DbE89MvdY~S>pomuv<<h=G7Ol~0^E2_<){Y~CtTem
z_J(^6XTE|W^P2XDhPt}>7QUuR(p(izVsUXX5cZ4TCgeP3Z?=5@9IE|+P?a@Z=zyl*
zWe`V1mC*FZ%$UeqxF#Nb5}_&F#7i_K+H0e&TE}}zZNHi<mLY)r<x;dZ&7&$)nXWNc
z---iQyxB<SVfpv&4P`EmMu42As;&+wa|xfNHf>+*UaT9!?R~T$_8AjB&$%Pcsmv?R
zKkxL5Mn;S!h?XhBgL@qG_`G_&*%=i4yLRpJQ(a6uBZv=o_aD?Xh+r0s3?`OcxgxM%
z#oj);)pP@btGkgMs``QMLksYp`j5CLLZC&#?5D^Z3Rp#$dCU}84RQ}b#D+FIWOnk@
zckkRO>{%0Tfxm+wM)Dt#Q~=gM{?PUP?fD>6ZRTux*d^LD9{8<M1T+<y>({R3ogUlB
ziRDl$dpt}e0xGep4z|W+-IIo!Bdd%_=}9wC#{Xv7TYYF5nBuNeIm|~wSN7hqOLF^I
z%#SAYZ7*^2TB|xu&ln|#wl*DAbKk`(ZD<U>zC2d|n?x5bHQ^o&T)CZc-%h~>4d{JV
z+1jp|z&6@%LT^@>n{)TB&f-p*Ec49EyCxaq;^b6`i(%OiO+zE4;77!>6}a7%%n(aQ
z6C!Q`W)crfGOmscwgXUWZs$~6mPvg;hBvQWBQ!(o9L(0xbs+0w&%D>xAito%Dxu4s
zr4Ri#bFEQMJc*oUldT6wY;xUx(Qj3jpf(~<xLS7EVd~TruI$IjMwwDwyUL;mV%E6x
z1PuxzE)Jz%&}_pN-BV>G08US_K#;GrfKDqKW~RNwMh+*<6CWBn3$XZ_5YU~(y{QZG
z1R!3XjPG$0i-oXR35sJO*E0%aV2~LdX^MzNTVVQjbz^Qz8$wn~bHT_wtP!}8L$t)|
zOXv2T^!X9Gsp)H2@HA@FsQX_K5m0PA-_R>*=T8B)g*b`2Qc-#Chsl(soR5YsOF8+%
zLhR+}&M|BYJhgbyfJ@}us#iCR?b*a4UfJo@z?+=-^x%cAdiwfw@3+bduGL$y%mu`=
z?)RV_3#or$1BE47j8UeItwEY_wBoHa*E8tEw2YH;#{9b^nmppUOQpot8v|PvFH0ua
zzcPlF;N;^lF0=FpsS5LFb3u-~+-tu9M2_0B!phj{1?@6QP(a=2w#TcPQ-U$_qHeEt
zU|@>Z{~DWoc@LoE)5M<;<)DZUKtv!;Jb@9@k575Hz;)c3zW2dv^@=+w@rR-)-HqPG
zf0J|W;jb0ZmcsN4%#Bj`Up0ds?P5(mnB2y^W=6-V6|fHE9q&|K4NL;ZV}re8mHEhE
zkoJ$@m}fWF5(5Eh`2H6`6AfLxT#%v2^y~Kn9A5gJ`66m_5c;mN3wE;ARk}wH@Q|**
zcYns9fmlXn%et|;lXzr!*%E5YQXc)`2T_vHWCYEY`aa`+BC|ZxxC&GoRLlQPl|j7Q
zDm$321&%EIZD>U3rr=z>LjHlC0`T_2O_c&vj1Q$zg|$8VDfbJBjt{%8p<xL*Euh^d
zrqAiwfHkW=ecDXWXd79L91HLZqPv_=B7HWqhr%M4N;IkVjvY`3$Mg>0qd5Nv<^ya+
z4Pb}N!4J=F-MTeE-A=*dVU$|3K`falrEEx&+N~&6lZe1T<E`uaky=<>^b9c;y50|)
z9Yq)=7K)Wdj3C0+F~HkZGdu#1wOmF5xi9|R$;E5}RT?|i>B-S?2C>hdg$5gx6D4qN
zK|eHP5U8mpIrVQ3fkaDh%V;?jm3fUK%Q9|A{gKN%cJrx`hmL@o6ZRw1r%#8UU{2J&
z^4<tbdJG}@K#9imSwiW&>5OiYdp_mH3&Ah)M9tx2-jkt10t1C-DYg~D(6&Svy7a>m
z1kjY8>GPK_imRXfHXo(Q6eFPSPhvD1Jn<4n!DHHZ!;Uu1(708*HU^l0VJrrkd;!78
zSP2UfDypj;)~#E&Y}rA|T}ZH?pjtkDLJ7+c??#1^qeh>X3FnW$-expb;O0Gcl0JoE
z3wLDF6a^kXmI9245&}7v&rSSbIO4qpc}e-u!Z=@vP{QCgbv#1Gt5>cV&Y2?`TX<{o
zK*MBR8k~u5y;LcP_yB(V_<_p7cGW5^pFbKSM^+KnJ&Fz+#M&wVO&v%a2Oz_8h(bQu
zs6V?A7?#=d%Q+6Y!#``#TO6KaPGcMv7RLYjb(aN}V!kaX;p9+x`TDa*hO%}?(e1q?
z^N}29<Z*8b3ed<*0ER`yq@n`vLAzb_xY2DdJc*Y}XD8L6YT_#wjs%TTUS1xX82LUv
zS_r4PkW!yE5+<PUYrb)rJ|6wW<0Ck@ZrHGnv0W5HwC<ErLd!|9cXLx>p~HFl;-(If
zIL~?sL(32=OG_xEG-L(*i^cF67!9z$1hB-FmjY{7_H)~m&EL<K&2Q%Hd)-qyN{xZd
z*t7fh@8{<WGd7R~q?7Sx!`RW0k%3E{;TETQvmoRN9d1qTddqLt%$d^f!Y`P{i{^(g
zNCu*-O~%luRhPsP&l#!$K6A@_ALOW;WVBP;wNUmo6*tFFq5Fv7M=-{GAu6YO=eH&%
zCJNRfFRP=(yV0lJS6nylxHYKk!-tFzt2Po!nrJeG0L#CCWHZEKoupDX^8Vq%cPB`U
zLCUY&MAiR4CDgY+QK!#w__>WJ(yLnsM^)4itsOCGXHGX+M2h`eL$*Zh!E4vtFly1#
z6f3zh4f><YXmYTZDtpXhW5v*02T)8P#__6$l)?oAxO|Tk0`!4@r!NeK4j6EZW^r+{
ztNyKw3$~G*)fR*qV8n+sQ1j%Lbm`&f&(RR;JE3R^90&=6yn$XQ7~Uc>tvP`Y5VdTd
z4f_zryI)Edqgq>^Cy0r@{%Q+JN`x-1wT!H6I7@-nu3dzU_+NW%M1^|YnCx!i;TTS`
zycs`d&K&Nw_)v}?Kh7{JAH+~V7Ik&pk}oqyK#wA1Qv;@ghnEzrAn+dU^7nmM{{v6J
zm*>WAYzYRS@mC*oGo0Jz<t@G%KBq0a*WpO?KN8<MM8)y`$@h;ab@&~T$B)~8p8oB+
z6+|~By@MDvo8D8dRp<Y%>p{%S*n)#UncbS(iFgNLI3|{9alTh%FzMhYp}T~iCUJNc
zd5>d9Df7LiW(1?j0TCpBbZI<#(F#!Q8IIDqsVHhst^@KNb2d_AiatBLva*s-;4D5I
zL`B%v`||B1B+~c)oltnLG<4_++<YDcv=v+-GHOTh3b&5?Z>7C>?h+FJ2TAJxHw`?j
zT=f}dZ!Wz_Z;TCFPs6e>mYqdJs5nHbuj5?ZGSFgv(Jz~+9Mape&q`bKI}t6EmJb^?
z3<qe=5`!TR?$2yaNQO85W!q8W<}z;FKH0Y4!cMeakJd+f`R^2q_-DEQX~OdV-+$aX
zIwRq@hsu;Xud}c$v8MA{vAgbN63S?La&;@TBIJgBeY$V;5}gWAaM)-XpDRqwm}^Y$
z#MJpMIELo8VgLfCZ0AxBDP*1M@YP?_5b_h|X6@_fc%6@g%4^;A#N=dXxqq2S-6e)0
zrsq^rcHzG#CkZZ&qVm$EdmbsaAV%v8zLT##MiYg^HlUs}+8QKL@MBtl8zU1$j*d2z
zTS3`2)99UO{li96n?(Z6blPq52<a-dDqofh7n+HpDBX{e;X${~x}^Xr9)q`0njye1
zO02oLUPLi^Q$!+V*N|5AJ(=^?%pkTFL>h{gLYW)^(9M_Pz#XBXubK5nY!Q)P^7-??
zfB#)XA=WFP-Sj&ka9i_@JUyHFdwDGzsd{Q^#yMzg&0%MUQP^a5qY3*AW2O6=>eI9=
z5qY>NoCaKzWuHRc!;(A8<(uq32`-%)cFbM(`^bic2<YyU9xU`)zI<o1AHr#9;YPeE
z7-dl}03snBP+&vWyd}}~U0Oj6_v44VGx%<>U|V-zrJ>=$fKEMLM`!*NK^o0{=N`8Q
zQ~)-Swl$pI-KKO=FF^4wdTTU1M0eqU`k-@n_kZvoGgsa1uyQ3<@d}v{;KTd<{CeFe
zc=bvR3@c*<Pcj5SL7tcy2@O5yH)06252(yk^gzN-Wc+wFA&8GBq2ejNj3y58b=UC8
zcut|G585pj+}YWI$I=7Atv~!JBwM)W2;<_)*KPXk?kur1k)rY85)1GI|DT_!jUP^3
zLP+rY=^FczSBpMRY<H*bp>qi~h=o=+NmJq=joUcNF@%S~EKb&e6R6Wo;}3s6+f<Wt
z+5Yu-+0{)#i8O;}w?N~-$MwU-qX$<_mt6+wyG&lX-4HtN#2vXVz}s2ViprIw{eTLl
zvA-xP)3Gpdh{cSxc=}E0FbKo?gcgursmAG2^FA2_4WBeAQQ=-9UgJxbE=|2~7zrUN
zvlD-`wD%@q<la0SW!P%N+MBv@X1{BrJgh<nD=B^kntR?n2O;yE^L0EI#c8{UY!p6I
zp5nbSY*6sgqYJxqmwdzHg#C!g*7-^;i5TzIvnE6cV<s*`ydSU)<v>_CXM)nvcuz-5
zCow71F%1jAn0Ei%lzFvWTH;%B6&1o?3juXdXj1-)W_F=brK?*#2#{4yl3sFymj}Do
zXUXE}BZQ`s;W)B+S;72RQ#!oa$Cp0G$|r%yfDUBT^rgT{H4P1U_KT=HDU77X<{ewc
z;iJW8+3*@`e|m?Ml+7U0I6Keh4&MAup-I_EgYki0NKvMknm2En2UYo!<-<o?{;HL@
zMwh+rkHraNlnoD_>C+k%w%ypz!qxR^WaL|*HjHg*zx?8m@Ja`$wSY*nf_d@9S6txJ
zO>gOF@O|6V{#fVfom!_GtydsNT*y+uRsP}OOF&WEw2pZsIxXv&-u|#WYWROKG&7AK
zb*c$-#<<eWgPH)-4p;afnia0vl~LA)+#EI&BzoSk)9%5(Dm3p6ylk$c4YPt&+u$->
zK`=+p;>O76m>AJRMrf?P8MiGz8(;`+hnf}eI&01Mx1Hk#o9ta}9IZesy+I?z=bWzA
zLZAhdw|@5?;1zq9eEZkH1^l6uY~QWRXA7M1<^z6j?1GP%QG`(gw?DQXd<ijyZqL5^
z^v}z!jW=EHn<E4pLOd{|voC(yFS4^Q0H3y3nLVgiugefpI_;ELjB4>oH3iev%<-Qb
zuM2RC%jeM|3mgUtvb)Q_C1qtNU!SA#2RWyHH7}mEhe-n5YwytL7^p3N=O)DMYp->I
z3A}yf(m0p+py1#JoCF`PjQ#TYvt9oAA|fOYQQCgR`dHI{rKJxaKfd^#B;Ax&Q(5mU
z81;ckceN!9`h34eF20aJVxTixr*8Lq<?=wL=J;Uyjvv0~E-u(FPy5|4hz(l2Q=sd8
zXBfA{U>4LoQ%UkJN#VYaJ-dc3E%muIpb9(&9JS94*6hs<9Un`iqIvb$&*WE&V5!l(
z`Wy_T_k@-|V|M0lI}x>M+8keBR=iNoZwK{eB*|sjDeqevm^<=!vVV%DX>4FQ8`A>!
zUk7k0G7{?T?Q&R&v~S;@N0unjaCYkru{Qnc#+Al4zMVCe;pt6SglSk*`OLT-!AqfO
z{#jlR+?XRwXYu;a(8>{$H>`q#;yUL0{2`1*pj8ai?$&SY^iSOyIZ<Mz5g@;`%PNU<
zKT7ADZuD<R^bav-hmaNuSK7VkyqqLuI!bDNEP;TkSsFcY;+VUW_A3t`{(&sP<wNKM
zfK8VHmui3K)W~1G+21I1G>GJBl^Dj*bFhd_llRC&qCm!xLC${YP-$72?OBUywBl@Z
zy^@}nr|#&x>W*+7<Z<J9@9<vd5RDuxcVuATQVh@Dy)%XdOZweKYC02_%F`BFS|+EY
zAT*R&8bsFu>kj;cQzk<)7(sz?M*BUJ<LV(qHApQ2^%y=chIP|o&?LQgej1u^K;{o#
zDd`y*X>4dM`QFPadD7<Z_RhWDvV%jwV$>Ptf^?(Dpf6`ja2w5OIRa^{kgp^<_OYMq
z3PFHVE1MSKsvH;e7h{3s@+u#4$ikau=4VN@q$)rX^VZ^nkq^jwv1F>IgQaxwJDWko
zTY_oe@#AMxpTb+^JvqD8$C=>B6;)DHOs63N^^kaLvlr!DK)^ghLuK|B0j=NJqEZG)
z#Ims(`XpT9rG9PJ!<Du<q@$FS6buX>Q%vCxZuky!L*OBDqLk_fK3t)lm2XiY<e9cC
zo@Zci`TF&Z<RfS25w>xbb$$PY?507dJBOJ@21>O8-EgDSwG+l+2i+Il*t<8ANuc|E
zTBz1o_5&LCiL9%g9e$4x7}s=dNP|ts_OBLY04Zc-e1Obijf%cLV`MKctsn81zrJ}1
z^f~tYg{r@0pmq<ior_4M$+AMVTa*Dj0%S%SLbWn}%^6LA3ygR<P#*q0xcpq5aU5H>
zj{4|&NcW~KDMYOjg@99V;b-Rm!>9ulyN6Jv3HC!y(xc!tI!=tG7q3gC^GC_z`ee)q
zm;hRKJTyN+BB|ea3>(7gUB;F&(8yCyM1QbvzvXV8P(RXOlX#$q=3j}s#KWgSWsQMQ
z&C1^MuUhcpGL#nlxGz$@>Z(Od3NUGa=PXy*Jln(XK9bj~mcktPVOpjQMJpIVTgL&h
zy0NCR;L0@bd_D~3ZL1i<6aVt12-Jj|C5}*r0#OmHH^<J^&8?xfwpY6$!-lOwfD9<%
z{pXiQ@e_Vl6RztWx($poswgKY&{30P5}7=+Hd&?4?V<kIKX5XGT=Ng?wt6lmXq-!l
z`>2*IYVYW=FGOsRHDAHAXjq(St;fgs1A+F{{`5guvso8~ycJ}Ds+j!ZbRTN8adiBO
z=nj2Zvcw7MqMYZ?$20kKK@nj-fh&1he@Dvag;@!&sTUibg!0+AfuDIa4v-nOP%d;9
zg|sWHeWPkhr_VDp6U)QluTYgWHtJ~KcKI(YK=wMLza>8(Uw$apy2i*FndOA_VIifP
zH*fLKGcd43atP9Tb&ECi*F;93s2E_JUbG}%g-jOF{zw5wyD~2!AGDn-;&uRL3Z?C9
z(j>D<<|b13hhujykY*%jK?5Qhb);Ha;-5Zx1T1NN&29nPH?40F2))?SM2V7Hr@eC3
zDpYu$6wUk}-fd6Up^p0q+)C)CA#wAr7}1NX^=oMAx3|k!$N7{%OTIIG=#h&=GKq~#
zytZ&H?<r!bSs+t>U_Ky2e2cutu?a@bEwxTyfgC3BJtIdNTmkSXB*pTRK%14Dj@2-Q
zkk>pH14?R`PsQpPNID}2+(5%49Os)|&WJ%3m~3hL9nB=4b*FMr7G7P8Jyq1z`E4Y+
zoGvvQWo1=RqLorH&w+R!^y$crmRDpU^yAR~WzYT5#lzA}dCG!lGL}|z<<O+vj_cRo
zf?QAX;LV5SZ$M3%Ja|>HO`7EaXDF8DUt8<y>vJ{JUJ)p(UY<+XYzfy2p8=U!wdIJJ
z@8-L#nCac4N8r~L6glpE{Dlg-4QI-ktuyCsx$Vv+)gC@9o%x<Co$&{+go|EAAy{JD
z*9`ua2FG4)(rmvo&RB_z8m(?z+bQ?=1TA)1uIe>=H<Sc{e6~Xbk7)1iE-1GsBj2XI
z1<rI3#_Xq9&$nmnlqt!?@TOvSsL}ybT;r<a4I+|{_{#7)LCII_{_aC0R;cr)(pL(V
zP_R%Fb33}@n+TWZnu<2?Z9EHu^A%!YuvTwT4X%4fH1$_7`<>~ntw2S-Fgu}X!)V;V
zqAt;(gtgxeM|t17cKo8M!A4W|4V-LR^dcw6<jQZR0D`jjy&_fD&AO$SrXiM%(>Nxd
zc5`t-5qZe{Y(?)B?l(h&vNjtCH{|)V0ebLEfnVo3uUq#&*n9J+9@qE%yKN(6rVtv~
z(MTvm$Pl5TL31imDw!!!$~JC8hBOEj8Z<~sg$N;3XctAL5TVJ~K&apA;`2OzJZt^d
zdVXs?&syI<KCQL)wkz-V{l4$(I<NCQ&f`3eI&M|ogOk$6WUB|B?_+t03hBd_3pAlC
z5}$y9GDW5O?Rap)K)+4@mf4tVj2NNcqP(Tn5^u-7{w4;x&BXShZ`nSVOAd5dS5QuS
z7P4ja8<lI0a;Qf-Qu_D~S9iSjgG=?CmNJ{8e$J0eC&I%K>EG+@m(-z^kblbHjnV=p
za5iSZz)3QNOxW68edhmF5Znyh|A%+hd63Pk(uBY9oU6eGN?mSV`EaeADO<wQ)zuLs
zNv2V4@;)2o20qG~*4Az$lNva?uVVV@J=TE=eTO}^?4R^B!N4%rV$rjG<sr=(AsMz7
z6+x$yG9DM1eK%X~xXsqgPGh#7**5iAvqHZ89CN+kTZB(WSgz4n#e%zzM_R5H_*8~9
zoY)lpWK-5%c12F$DR5<t*a@QbKuo6y9&l?7Y!;FjDH(O;zx_`;S1^-Hl1B*bZO2$4
z_#sKPxa3c&Pe`8Wj@X609k%pUzczxTfuV(Yvcpt<Y5e#MS`$TeJFkDOk_wO85!8@e
zMyi7(WRCCj8NJ!zQ7nh3s2H_=94Nw@0`a&>HwjXcf5_YAh3*qZqrM7;3#KXOE^*b5
zDr`PBSR_H549=VMZN<2Zv~&|Gu}xENIus~8TcIIpxtnbIASOC*XTa)PwRc|4G?N)K
z*pN~zr1cGD+KJ1oPn&OFGWzPrZ3|AjsC*|&39fUdQ}8^b+f_@`Wo0^(ChZAZ1&RVZ
zJc&9Te*@v5oT&e>KxneQjY<4|jpS&Vv0m-MW$#v3UM&szLte4SYB_SdmMiP<q3m(h
zVByjHpu0^kUa}u(Fr_v2GiRKFu%C|gyq1>vl;R-rL74XMdIeKnv5zR-K!F(KW7R;o
zOVSJp4lZVvW6VbD8i+sijZW8OaL$rlkWvm}w%~!y!v_yU`9Mwdpv{A4#?QtL723nQ
z2%>Z57(RMQ|DzIR@1G0Ri3APsxoi4L9wDE^00}v!nN#{Wn3!l%wR3nS9$4J6U9!5+
ziesR(pJIaZ+s}~5n?kiT)ySI^6`hE|A!oh#3m=QE95MO~pS1o?!km;(FJ58=v<Ih!
zh-XECVPT>N%~8E?fJ;ak`D}GNViO87<|uwlqBO+_fig7u$uyfKbjlp@aQphUpYar@
zS5}rrM*Gi37gKwQnMu1l9Zjl+47w$`bxuiCaDW&h-=4i#b0CX#3j+};VqiGhYN7!7
z&!0E(kHe(uQ2@<?pGXWZ*)8HyKP`xZ?BU>7PnAxspd&;C;;YK#6%FK1-8f^4<Q<oh
zQB^%N+Lw{rH;Rxk0VdeO*`Ysv_*QrE^1X=|LimXj<W|DN%x0H4tQ0^6MGc2<JdyaG
zOOmY~xM^{O*&M6*Lc$^7tfHQTRN<ZaRFHP_W<-A|CpS#cJ!HX5pyKz&ojZ0MtuR2$
zn+)G3=cc*WbZ3k4nM&>+20*A_N@fRQkC~Jpb13yB?iu@T-F?O`WunIVP_^XGUC<io
z-xjYv!$Wf*SNrx|{Ll7oQc@O=W(rFnv*X*xMvlQ}jfeT%_X`nbrL)p;A*6Ejs5fPi
zp?*lXD|o<>8mUJiD;n7@`QtaNTUV5s8M9;JUxTyGpa1vn-8sJMQjQ(s0bLxQ`Ldw%
zdS<vrK`?okwz@~RZeu+WwB5F@V6eQrKU-pL;^#FV6*OGNlk%_w=YH6Nts&2zTVRi7
zJ}nF7388mR&acTLl{hxebpct`(~(RR+je;$6Y5ZO%!q?7>*`hRPzEUgV8px+2J_~H
z;ji;Go@TFAR!0LJ`A0(AWhdM!EDz_@ujegDrBduz?CH%?)9|c*N}qWtyPL17EbI`)
z_ZofH(HG2{QSbNhUlcZ8`oo`nT5p+frHR62Uu2gf<a*CDJq<HYRhPbFF_R=M^ctT?
z^<6a5(!8%^FQIaAr#n4!mwk@%$9<WH&YR;1l0tpNFDXsVci?|0Y#H=J6Zzlzdvs|-
z<$PfN?~z=Q^H6_+YyeLnc%@MJJ08(N?QlR(>wTJ7nB9$%)}LUL>bz(91b2sM^;L1&
zK_xqzi=^L*qNsjxsN>CHDf*(9l{_s->9)=**xEY5W&mDaDfQ!aU2pih17#{BnKL2Z
zki@<n)l3&Gt~upN;xa8+Uc#gHMSVSEOWlffp%n-dn2C?cd0d9pw;5W86_qF1O}Jp^
zzO#^>;SxNZ2poOJOubhWb~ol8hf}yK*imxqn|U;8`1|`;CwmlgL9VWEDX7y{l$v=w
z7~^8C!SFG$EFqjkj6hG#la&7ODaA;VU%bH$@}{8mK5xQYhfGawe~bN=`9mR*BZf@F
zyHZFgkNo%2+E+9!-LpbF&RDZ<o$O?<6%Gy&5AU)51O2Q)XbD@ZGV?<;Gmp+zOGl5k
zJhEH<q`&sJkx=gG^}3-U$)arLZY>|=)D23rn>91%8ijB`XO`}n3Vo5<yz1i5Y3<*P
z!y_O!m<RO9H@i&t4oc0bdZMb*vLGLCZ@1!4M_G}Spbxd6d8jVC4o&xFy)QWJ_*Kv1
z>f!V2)m(ndsi^ur`plVqZ+GL*!hsN_69azxkD~##f@PbB*EDYvnV(0I;4e)wzx7TG
z`Ub5n7bYwd(x&V`cO(~7Kb<;UBm~1&P|lh6bZgnZo4}aV>2EFy0=zKf@br!f{^I1A
z@I<Sz?uoJg&{xFhn$VG|s;YtETQ)VMKe_F*vTGE#&^p7(*Y|3G@ivPtQGb>U+uu_n
z=hMBNg=FyY#SMUSOD+PZh7DCEyu6*`M`3+CBCoc{r8)GxX<rxb4_PgHCu|!UxyDdf
z%MEcw$fWg4O0~t2&Tf_PD8f~y;u1h{6dNJ3;0lix$D6U*snKAR9$p+tnT2lV4jOH@
zdQJOhu&tcFKW^GVQ$<a*IJ?7`S2Zj%f|aFLzsFU5bgu6}lYTAd?ZS(k!@@nJbWT`w
zJFutfmCKhdUBzE`i5ZQ}mrxXj@7vaO8_V|W!aG?T(DJae4YrGR)V>tNF=adH7K73k
zACFj^E8Su%$g&7Y>7W*VhhsyrLH%d{r+xKDD=V8s__vatHhkG&eMix0LyejRbg3o#
z6)KX1J4Wd#^&N%s`P(IHZi_jF{e7IB>fpG!HwBgKFzrF<Q+0b!><qLpg5=ki`ffO2
zMuR{jVkIKvCJ{b9%>U=CH6NYUS4aT=s8<`X`uvA%J<#8ii*q+i+d4Ip1?7#Jl3<;3
zy76>Jsax-$h*>1<b^v$eiT_hUfsl~7|GYvMaHr4!cb-3akm?l+sz46GC4RZI0-&%)
z*IVM`Rctt^A@QkU4m<+>o~?(qO`pr;^K<(>Wd&SsDzQuLk$%m`u#3}%8@7y$Z@nAI
zR(@Ff1z;e7y1XVMZR{nP@ZfQsW*ll;-pY}b6c)Q(T3R}66~M*Qi9uT;*JyyG^xhZv
z0Z!-j3mVB~6xT)*P#!re{Y^Wb4>52X)t}?%FJEF>;E3(DdVo5-f&a~Ag97ewvgXM)
z4Hz@d17cL_iL>mFyf&!|f+J?z<i|7|9Nzw4PH`vM;7#O93roKc%LY6TXob}T;yJwS
z2`nfiKemspuB8wJz=g2ESXo{nWmOcbch5_AL<7KY4f3}h>n|fyX8pa`i*xUCn(D=Q
zY)|q8y*E+FWmJ-s=E}lS@=SP87QUdu?t(zHvM$Y19W%!C*jqt`_6gp+(aS4jRNCUx
zKaNiGi<UPYQ@K*Q#uA__<<<#<8%sBRRQP!=<L8T06tKd8DD$G=kFK=qDR+BA1qV{i
zLB!s7BAI1T=YW6jr-YO66i7M@K4{AZEcw<%sjO_Jdx%Q=@Icv0N|kzt?{ZUsxMr8S
z<!3Xlo_O~5A_s@c!K<%2HB1DSvMSNzWE0$OVW5}+!Eq_n#e}*nJBRt_yd`@9QaYf_
zclIDQo&ZS^I1YU3t3**Qj^xW1uF&yIT^Cv*?Q_@F_M=`?;}_b3XN<yzwNdR}CUpO7
z^vYLx?!#@(_?h2vdafcN;^1}CqcJT8!ACQLLqcMwnbW=ldr5dS!*-%Y`;wy6SU{`T
z!N@VCc%@GrG+Fm$tC6R9rx_C|Cl^exbI^_|R6e)m3E60phi8l)E9uRfJih>82>PV`
z8)K?XcgLuW8@KTCcykW=6@xBtLRPs;9HA`V!b-C2F-%J7=#JgH)soAlJHDhyu+)tq
zN*Bc}@&w(MOqE{;7}x`G^X8yY8#a-uxVWW~%O$v&K~zXqZ%+kXH;lEK2`YHf_J?`<
z5M!0&SR|#ch1Z-T6)1DC*5LZE?b*F$Wo2>A>f85pnG;-h%GN{Lqc~}4j8sL!BKb=1
zX+>nh+;4p{J{f*KtngFyQNp>C#;w|cn6U4jR9gOO^G6eQ;F!%5X*@_3Y+YVcnRVr1
z*e*`5J=$g~zAX~nw?(T=+h3fbZ%cnwXDAmqx-ukdIYmj|UNL?9_PzhmV2bxd!f3sF
z+K24zwH=bi%T2+PZW?D*nds~k8$RC<CpV?LyzU~StG!9qu?n^?-2`+3v>h_V@>5Sa
z^MxIE0e)ya8*OHPCaZr>xyFsY*REZQtzK!7c<b$(H>yVl!)cc6A^Vs8Fi#-cya6#!
z&7b>JJKdF>lDX*q#y=#cRME;m{d|Uczj*&fl`b=JqgY{~Gl+X6E$w13A~hy@dJP2O
zNEHooh_>aGSHE#23&2di$CD*@M?U)t@QEK}vI9W*Vy2DGxwW}Ca_2BRhjbF7z0*#!
z!EdK+j4F)Vtar>1)lBX3es|O2Vr&07_4g7V%vcz*Il)nzVMG>68VUVS=D<l=iNIxj
z48DE+3XU(-=JICT*Y!lX#^X_JKq@CrEv&2{ZG-3TP-RV4;q5yrV)S8*nGVH#z<xpn
zep-gPRlHbm!6P$~Q9%E!*Y5jsYM`Z>#MX;jc~tY1)|pf5i_~BulJYN6+{Rw*DXTR%
zEa!*tTng>kMl_-?tJYRiTbs##aVeFzQ=WFST2aL~^IF`_$(;t(ScUjKdt@s2WV<9r
zC!i(AuHE9jo}!5XRMo_^dh0RUhk@I7!>r1#YMH06uWxHxMhKs8IQE|g*h8_8G+HHl
zCA>6^=hSECaud7mwVJ<q(gG)^Q>EJ_$LyI17)_(|b>$tUwI8W!h?Po7;r8a{jVRPX
z6e|2Y<E)PNJG^J{eycxWL*W6k)ns+Ocm0-+y;<O%A79kgs$Pj$El%+D7s5*pFf>@h
z!Hn9%*?3<Qjz<7>fMC@+A+1B8oad2>HTx$}zp6+vBEi0FnCMj8;;`xIr#}vky7^Ax
z<JX0wV*eUG!bSO!kBfR&wS9?;Ne5m)CQVS<pMR3io@AYZZ#ttkNP!4Be(FF7(SAH)
zAk#tcNo8bWkBk`Ml00&Xib{>ZHJHSso^M+JEP70TB?5`N{(bOJ3n_UImBO+P3F(%N
zx3oe&wzNc&@;BC2kSSn+y*nX&xJ3o*xze~}uazoex<@7U95&oxpF)&u2?XfJ^1hs!
zmi8;dj~%N$B$5(ASxwynL0!ggieWZcP)r}SBe1Y4@0`{vDa!LrHaT^ZY4qr(2V|JY
z^hc^Dw&Qqe*t$}a4+s#yw?%Bikge{MtLHo|OE5S%%>DLxZ|SZ^r!1=7T<$LQ{iw-_
zL9uyUSC;mpKU@VUlCS|t-yn-JoTkzKqP+UWskpdzysraesw6{AQLnc>knGiYyW(Ge
zy#|lF*<LcDZ(GZYm%&yJh4OO{zL~7lGJe#l?Yl)c9XX6lj0fBwiNBdxIyc@yGM>j=
zU;pgnj6Yo-Uuo<m7hQeyzL|#6YDe42*We4nJlp5r+S)4giMXoJUH9fwa{aZdkRF7p
z%=q83W0pfZ+s$msvXb-iy%8k}kST3J`!b}DmwfgxBUtxc4Mkg5_4*EY3me(S6uUgI
zeTwIQ(M#97Jihu`pN@iID9DjmDAmfEm}N|RNq>vd?K`(py*~MDKMj}s#ASxcKdAk3
zDofOhTY~N7*3yO%r*`ArN}h<uX!oSn{WU?KP3GGM<qS7>&+7G{q%O^Q&4uUzn?qwo
zHvzt2>K)o!ME*(6x&~~yuMF)t2Xx*(!!tL%Z%lBn@2`&-=aqDk?kMDLNi>37Zd6xS
z+e2NUI*gqblN$&*8c-`{Y**87wOzIy3Je_9d=@9$b!!(NPF2W_N|j8Ry=KjMfP{vP
zU``R?;Tk_@X6(@F3p%4cxu8H!?$!3)0Ihu3YuM{lRh^)QBDbD|$4LASbH>=vB5M=2
z;7L!He$T!Iw}a{{djGlQDX+4(H!eD0GOCvxH_OHL^}-dD0&P)?{?aQZG*}EjbFeg9
z_<W-Aw$~~%Z>H^J-L)&75AWNT57ZcJI0zj8gR2@>W-X={7rSEcpyEfi#0}|j55(|D
z{=F`;E|=tjS5G_XD{}tx+tMbURN7+DXkGdqE{dSq()2*^(hK%7{js7*J<=Eenl~dO
zeN!2O`~$tfXIi{B!bb{;*0rx*Erc~h!_MNYwdvX0j1<}|Yz-XSWRVwQHNQ1cS6%&n
zy-LUUz7Mi2dW@U5&}APb_N(jdKQWv<{dC=j%BjUV(Fx|Gl}k2tHTU0Qv+nQ9CA;%i
zm}Qi9XbD+k{<FH5?$_z1O!sv3(P(zL0s?0K?U*in9388_f(}6l#TaMCw@MGPrql~q
zs}TEx3MW(tZXCg72bcI1USu3I<@9FW?b64SfuDyjI#&F8XbegaR1EO)WjbcNX|-_<
zV48D&dB>XN-5?BkMMO@WQYqO>kRagIv!~!FWGubgv$U!ohy*$BJG_#qTU|0<?9u?#
zeNNA#rc)30r>3T+vhqDFoB*AxDqr$MVqMD*NcKt%HQM~K_u#=E#a3xs3}@%Y`Nce~
zet{t462Kji!#)j!jOwJjqr~r9$`o(qLgS5W3yX4#jlQ%s^YoLlXtmMJx%k8FJWsQ#
zAuUa-x<lEBgb!Jt&%gRp&VC2OB9`a>%9>j?Cq#OQMR6xXzw5y}Sb^wzQLI(w-@h+Z
z=!n7o1Z<ke3vGglSDa|~em6jrwEh<NdjF{j=v}5iaj4_WSUz_%mq%TuIr*eU^05sK
zQD7c+KJH#Zs5}{ADD{zL4o{>Ycl%P%7F{jtBqal5>1N<c3#)xaVMlB2EyK^952<}Y
z&U8Lu?!Ex2RB6GdK;3Gz2V;SQV1L&b>sw%5D>ybQtn61LMO(y~C?X?pdB;i1E_&%%
zK~Z6O2e(_ld-aLdwrjU2scrPvI(yuD7mHA&NnX#8559(O1$J^h<&eI0)43QF7`To+
z0Tc)eMvG?$#aJwmC2QHXP^&!abDr9g?pb_8bo0p;mQOiHQWj#~e!kY?h(+|__roz%
zux@?ZCtW!?r%y*kZEcH1-z#K*-oFx~qCecA`(vG2j}4~>xo?ot2cXUL+VY^o+jp}d
zY%FXln_DNfru8gVf1W+bH~DRCZEak-tAzD70_KfwMTSGue^5!q4mO_hX-9oG7xxkS
zN)6pC@@IvN0Jf(F*ichcO<ubI@M^5f-grwrQ`15+D~EMXjl$A;Llor&ais8arsGH#
zeKwgeKFQN{nFWgF;?zjTHGkCb$&)9Gt~<Q@C5C=}*J?jI8UV*zjhCCaW--GO2|q1%
zOF!{b@0u`9xEU@#eKD}m4NaaD-v<_x&B9-*C10&NlpBW>VRF#lXTZpco1PiWpjsNc
z|6EgXLYdsafo|>Y?)UEB*GjTbwjg6QJHGW#j#5LTqIUKYlf3px-6RY|*jbuRc8j~G
zMRT8x*AA3@2}#_(dt`)n6$U(r%S9@5;lqe|Dxk(rGtgHMTxpFTvFRnCif}`mGU*xk
zhcB!LS(CARtNA^Fv5zy+(MMMq9+c^rldV_URY(ez)2GX*+@IA`PQa2!kBVotS0lmm
zoH@qT?F_&j(dM%&u&heI!CFdKe-5Y7oFnPSYO`-~FuA{Mt%PWjOvO61&M0B;fdeug
z&huy$FWh8S^3rGOjNZ4655*CVt1Qk)<SFPzN=T^T)a0<~#>ik;u5Eqsbqc9ad$2cf
z)U=q^>8)yxT`gnJl)mFbTBkbCBUdo;Cv2Eoc7E)@;3dqf6`4${b_pN2nMCgMfKXM^
zVWc`f89WtYn8_P#^Mdl7RYLcwW-wLukB5<s@V&{qLvt!&R8zB94+tNYxM>cIpx|9H
zC2c2(LvWu=ni>!5M`CbB0g7d?@cYZV0tV-$qTGIE$rWe-FfY_Sn0@&Ec3Qjp%P#s+
zrjztiPNXyd{!XJ8>2jqy{+(CkspYv}AqFT97HmdF&26F%9yE5>PqA3)1&A0k{twB`
zAU3xrz9J_k`S5tTOF!xpZsNV$hrqmAzAWz2m*VGNx@}wB?Q5-aWyHm2a#Mi()`2`%
z{jj(`rJ;(su&M?GKA-DM75Yi@{BYM^vdH1g_JPSkS<6Yf$&g%gZv{i1?^XS|yVn$N
zZ%p>o<5iNQH-G#Hq>vy}X@ewX>KIH4UZ3+smQZ`k?KFw4x>1Fq!7k+C(zoW4fCNw0
zm3lNQh#rweq(Dd11_$T+bcwn4$n^c!yj+J~-W$MEIR*>e77@A1S6@#me#!ln@sFSu
zvG7{Gc=1WJWmTS=v|nrmG+gO5cFAVjnnl?wH3t%Jl=V;2`%jk{o4Qjg*rNUIzrdav
zdwjZ$zrLIk8Worm3SC&9g8M`hdP(l~kZo+&+`a0IfMo!aJ6v7>M{DoFgBQILp*n(-
z+MIh%o#g2Kawj|1aJ?$wMT`UvQ5-eu+^ppM=FgwOI_g-%F#+T?@6aiMR1KgQbi9B6
z{%294!&&LyJ%7MHTAlY?bc{1zPsmVWz*?``<^-WPVS%-UQ*FpJRDTpC=FOU6?sO{E
zQXhVTM(&aH+tfeG5r6iBYXKYK(ak>fy0;+f;`$_rV}nQAXKJmkoZ|j>EbTUigH$Jw
z`zAJa@HN!!31kj+H&mP1vVSH8>eTI14PAxuL7?TS^1J)97av6tkSlX*{gQOCvX;S(
z<rNNLP(${;R5Ugg&UFc1i3`shP*D>z2dql9148ZW!rJDlhE4epHW;JDkdQogXW;#y
zi$7jCL7kI1unC1;23l)_1_t&l#znKnp7K8cCXazZ-;Pc~$uxGpTH!4B#5BCrH<%Ak
zeU;+{W(3*b2lBEW4ZDbt-2XsrkpZl}`ibGyanS*yd6E??v=mBo<19dy2K4T&Ahiaj
znG2aHzi*bbQxN88-OwR4&KMkMi&il8kV^p6Ou^9n^kx!gEs+205+nikgV7lrEEk*(
z%Ean7qK)S*ST!N_)UTbRfg;u94%9BPkM{jQE|ph5yu%s=$|#9CrjWny9N=K0p>vo7
zRIZy<f;mWX#1T)h?!9~Wj{W&Y&}2ZKmtXqG`-@rC(y)K#+*;<}d^{KrM00cJjH^+N
ze;dS&PfN=bOsn)IB~X-F5K~c?FA1?)2?tipDYE3|Uxb32)UQiF7iZ^tsxkP7{<A5a
z8Z@JQfl23;C{eLdOOWy&JZHQ+OPWsuy6skMiTbm^YDXwg9ZrQ?36+vP3A0>9#hqqb
zLa^(;PhZMyNhzD@&Nxq6Mb>eS<P4ug%02!K5XT;3A_tGlHg=6cXp-6%3J+U48)pj%
z?Al-Jlv#h4<MC{a*IMWpOwW6n>L^8R8e!zuVx>NLGC@hH8WT@5neYh%yQiM8EWSN8
zjE;(manBxb@|toFn?5`Rk*NA@S+k<vE|_q6jXRHki~&2sBagQ<?G3zcuvuG3SS5H`
zqA6`j4JjiM08dh3PS=Iaf=_?Ft!eayFa2AtQpRoi`tekCj;mb0sNnCj^^gDDzpt11
zx<rF}J>}A5q@SX(W<Bv?Us4TY0|PUAyLA9`u@iz7x*#qQ{W(hqC1H;9uG<bpMovRe
zpQXIn+xy+LMV$HE2<Im+gKuZw;`CAzk_D-+CVdqllWE8c(XCg_xvDFkbt80RPE$Nk
z>b0;^7@*Xr#3x;467}@eh?7Kb-u<F`<(o^FFK0o=!ix1!x=@v6SaBi1Hai)~<!;@&
z@eJ7C4BNjaIV_GbF*N7d@LpnmcqK{-zO!cC1_?l>;RpR0(krp{2Yr8!67_~LMWk=0
zQ?A5gr1zTGlzN9B9#5LM5%P6)e0rdoY}XsQjAPZQBu|vt<l__C7-cwa02o(VCSq>K
zTKA)m=6?g?#KHnJz<!=G3fsd`CkCp|7BHR|?Y^B1!DJ}t(xxSmd{iumr<W+IsN|-d
zhy@sJYyT=LDIGo{Lj!f~2U{V%nUJ2*5%>bl!gh=3%e*Pm@+dJX&2})p^i|6;&O~@c
z<_Ru>W~bDgGjRZyI$u(LS{!*Z6`{zp0Ooa$KBz9R|H1#h9`ygA=M~zxg&|<Qq&{bh
zAT5+iu3_!99*-LlPLJHfv*civP)fR7%x((q<f@MPfkO4xgr-Mye=Xkrz-?J=vOL`G
z5}|pSbNtlFPmXFwu6LX<5c+akzLR3{LMd)*`tn1fYpg)=k&)khethg$;uwU#^<TU1
z3q6gV?r51;Ddzp`h3v9j>Pj^7R_HC9vArnTB0vPin2+zkZ;M7^u(=Vg0%=oT1T{8`
zAXNWngn!$n@WV0MYNWhEm?T#xIwmHpWQ6oaHOS`)dh?9kgfL0_w9WOs<j8{uq6N$d
zEfFozOIcM_5X2^~uYUP=+VUf{R*6AhXNkS#WkvsD-~%>rv*;;5^93c1-FV46IG1(e
z`DQ7UCA*7sy6$DJth{RDyS7K?TLeV&k?Qm71OEAEB1*9(*OB*Y;53+kIS_N`hbmm?
zY+&tL9Z|92)UKZHIqM7k+rG#b3rF0`sTa&)XRJ#khmRlC)uU_iYQrn?Q!r*LT_>y8
zbUlK!IrS}kx<<#ITLeCL$2OKQ1_eXq&aJR(o17UT_3+ZGmb(B<m~(WihWWu{p+904
z(fMN+Mdol9?*w@9qX8<!qnJ)7X9iM*5l$!PC#eBRx)vJWk2Szy6Nj)uwj*~K4yUN=
zw)3C|h|MCS2u3})`!0GMt6+6BUSK9a8(>9@xjb~HYIV-^{SL$divaL}4%(@s<Er1k
zKes+ZuZFwDsU?sw{sTAW5j_<)1-w95)<CA{ySQ+1OI+PGY_A9_Z<TEB^KOEen8=`c
ziBhKf=Dn==<q4*h1bU*b5Gh+GnfBA-II}hD)}^}dET;Lu{=$M{D#bLr6^7twYW4f~
zh3q}&u^-SY46#!?2ITe^Pua4|60WD!#VM(;oSd%r$Hg{f$83oGV82)e6ef2%i@FmT
z=90wmDjC29v}f5vWo136iWlUz1)>p|C>ux`u+DvfUSzFOv+#tTVs-H~ww>?TW6L}o
z4B7JwEXTx0D~%j^yjg0PmDQ8LbEYOG=J94iz?cv`PI_@JGxtVDCt&KMvcvQVM8aa4
zX0b6bVt$>Pxk&XEAf<hp+u?`zgb2J(ivzNF3SA(2qCAm{A_6!F7!^YxLZ6-99SRG$
zQfZDDflkG>!a)iA^{Q?G{2MH7g|l-sqbCDz1?<^l8nV%Qe=j*P!%Z|;N+CsEvy$CO
z_Jq(ZEo^dov`w0X2H@<lt65owqZ}EP{#xo9sl6pe49gRqIkAu=Ll!>w!2@#adU%|i
zrK(+Sa%E(6aYhjqs7>AkV1NhrjDSwkmO)lfUOrh#$%O+1$%WomXa|;T=&A1xTNS_}
zHuM!4H4|NR46;y;$&O!PP2vuFjeW<xYUQ=rVHeIXxzYgnwg@My()i~rGQvY?oudB!
zfuArK1^y~bNHUW21mHNI@$6r5)D_FF^|GRGMOng9j2$Qy$>6%o@HG1=i1@G?<#;Db
zn4QNbPM%EpbTiP*){9>pVUy3KlEL~85_@kIuJx7h{wuhAMI*qkka%ua%>kF-AbB_C
z8Q6-sq4Hk9Bo;_HM&du)$4r=Db>?yG@@*M1iGIC|5NWHw5Z%6})!5t5uWh06)R<|z
zfnWt$#%O3!f&N9@c&)94r<$EZL>0irkpPpvrSAm%Gd~x3{jdpVw0j)39G_XExS{f#
zPVRQ{pNq9RR<;hP7%=Zze9B<h;0iY_*it~Z6$z2)9K(WtW-fp6b!wr-(su$H@;VWi
zFUYlNpQB8<v&_JZ9ea!YVN<<iD2@Rse+4YCNV!lLY;hrS)0?oW2iY<Q9_UYiff-WW
z)W~Q$;m1AqJW7DJJ9UjXSZQ+#pP)yPwx#d%14a)rMSu0_my~JV=dV7z6dL*jZg=qT
zr)uKDL=%tFv-J@u{(9IAN2er423%YAK=#7yb)CBpWFY3`xVJp#4a~ybbOq;__f4I=
zVY`aPbmY9v{QD)Q76$0QVvyalZDH}%6-Nj;G~CK>x}F83`lan*qO{@Tp~Htyot2Kg
zm6L<j(HJ@;O=}IgF~U<)W>gC-f?`&?^!LB_h6bQ&y5-YgvkkPmml#KR-_w^oYvGX4
zqk}}f5(H|tD)Gds(nWJ180R%|CZ?6c`-n=FSBPJxCZuRdq)D$VHEyfm_axYTutKbL
zH)#-2c&D-D8d#N}^B=}kMZ>{3F3N?~%|2Tguo_mK5`zs&N;YGi4mTtpJ96X^xvwbS
zZOyG-zkFFp$C7B61ppFaocxrAs^a%8kj)m)+jY&Nm*Sn^vwi#bPXi~WIy|0K-1c~(
zu@J)orcQgka`hYt?$p{s5b2*j!y5;5j$^;DeGU$pMDs9%=u@nlp!i~mW8^3}74(@8
z1~u5KB<dKr?Yh|>+&57BfUI`jP5F<f1CKMaCLGWiJczk|eg>g}<L$h2Zr=;i_6@6O
zE2f4J4uU@q$?M|p-_G^0w-dQ=o!K!{rcQO-Toc1Eq+rm8bRRrqh;X?9f;DTZ1V|X7
zv*F&&*$U7h_bpkZ{Lxx&3IevrG&bVH(s+Md0j-<+^<CEuCWl=KYdf!#?R*k*2cL#?
zk_ubTJ@?7u#BhG`vG1IWq}C3Z8{+-ut))pNMi)8X07}l-q2A6fwsUp_`h^&^HaDpG
zykx&x?(gm~loV$*J^C@@R}`Mkp>aIz>x(~K!qwpYU*z<|#eKks-Z?T%3c%)img(t1
zKHqpjjc4)cd3#$62Nj>~iS4(H*IvHlJVK`fRBywA>$LRn@PaaVxlt_;@{TU-*fLzd
zA2E=`2`%9v#?0lWG6Pd}^V5=&U^7&mYi7D%sMbboYO(CU1qB6xT_Kh;kMIM5C{nyD
z9P{jG001!_nd~@CYMdMA%|`fo1V9LO!>JrNi@}>FNdMd0|0zgC(FVJ>{wp5G74pgG
zWu%i7GiEh>+_*>Pqb+)OEPmEogI~W6myh-sg)n*}5LD9Hglthda(B@=012iJqU30u
z)&Zs!V@(gZ48aabJkMYkhBrc{gZK_4yRY3AuB7`A!#108N9`HCAvn-~*}zLWLSf{1
zC#9p!Ev;()@yizvaDBA*An%mOE-7kB@-vIJ7-4i&!LCyy*-6ReOdFjZrq<R+AL<Ed
zWoV-bPmb3~nvgvZEi@P3*h(`U1Bn*u3hD<-aw%vZD6ui_i0Ke%#J26*Jv-B@VNfe?
zl1GkS&c>Z9Iwob~F(GWq59+om(~}d(cCVSJ=Tc?@{tKxuvKHtiM+2@ZU#J9;STXwi
zdbT;|TYlwB1}AJ!4I7#_NVfYf5b1*0Z(V1uTJ@gg=2CDud$F>9)3?5idQ(G+<v<I9
z0LZX}Hw6e6PH83CvE!`Svl%BUci5+b?Zvm|af46HH!uj4`9P6I=JL7m+g&ph3@H!r
z3{PiZm9n%F!G@_*vrs~Ts>wAlZAZ@48eeLFGCgViL|D08HV}SfSlFO}1FgxQ7%<zL
zSPO3Wys=yDjueh@|JIt3j~4u30Fq^#>EC`W{{B9NE;<)B|L7^l>H-3c0CeWFKId#<
ziZOOO6t50sYyh3784xs{+7$FK^eTD8*TZ859~SzHDl+lu1qlvWt(0o)sSJ)r{_<2-
z7$)unTlx(`G}(KzI)kKBr`q@K=x(uD16a>!?p$lGf(+WIU=yJd5;P&)rgUbNt$4yc
zWE-EYFQ5gqoSZebWxxAq1_OG_$yKu0pyDST(Wg)R;?0|8k--3W6y2dYZ#(kOlYEZ6
z(HBWg-I^$%i3ugP;}*5+X!0q<0zaxLRI%9yB4t5B32QrCHBj+qOBZqn0f{fTE;np^
zaf;Oiij8F18*XGzRDh6rIed66M+V_#>(;IOGMw-6qAz5zKX0fb*V!Qb7$f>xDm8Z5
zE21n}!VAdKj0T15+qaLk$p-=@fs*0jRc&|ScCVnIQfZI*F)`6W`4>AV?;CQWM=yzk
zBksz0s%!}fRhv<2>gS;ClDUyoH)^ECdpex=9Ciw~hr%PR^2u4v$5!m(!>?_OZktmY
z>C|-)V}SMYj^GV?%gki<W=>g<SFq|`ij&3RusLOfOlKu+_C%6xf7y&HJ}u`ijX1xD
zBO2Nsyu`j61wMKriz&_4ro#qZkl*+fqa>!3@kvK`giYRfj#pg;Y;CxG^fm=pd7~8-
zvrWI4DkWArptU-A#OCtZT*W-VZ%5*AS`4Q`Yw^WDZs@aSxy-{E6lp4nhNosVAo_3l
z*ahy>GQ%I!I*VD%($8mBXElYCcJz{P`NIi!iH~kRK(hq6kt=OzCa|a);=Tpx5t@Qz
zvV?P8{e^-d6XjzV!zVV)vt!3bFRIA@JFl&~;XMkUwinrdy62;jvuA{avO8M{`H;gI
z9I$sykEldK0T}gjoUZP*DQEW{4h~Ky-*-1;<O{;bySQ%>=ezy7gnP&N&}nCvJ2+I*
zy05$OEoefj<4u81wVC}L^MS%s+GB#$Hcj*zS^GL`bPZhmcPTNJTu0^{l7S-KIRe3f
zW4*S1zQwd4xmZXB5q1eFz=31Nlm43O&48^V8Gfm4?ZLm{(JpgwnP;cVAZ&Q2m_LG4
z)W>wg0(%n~Dkr%Z(tRo@yi8h`Ut8B-UGb;VGuF3g4zC%T5IqjTmoU)SKSxF-)EAup
zOPKHRMn^L4ovxKIS?92+I+!)j8D_Yml()K2pj&rCC0L$ReIn*kU*7^5u5f&lS+Az9
zK4opyjt$yRnBj*SIB|`E#Mzkxnfsm6(|MAMvNlz?ASZR}*0*DJX{k__JP>hsFSX0w
z>$Z-Z&Gb+wdHuc>r}+4<BmP{|pJC2l%l3lu8DiAv_f4dZ499(V`Yos@bC~D6CFI;i
z=TE}KmmpjQ-W28PwCik7_d({(u3aRyX2Fl)6v;2V0Qk}yTbQE*cu545yUg5l>DlL$
z?8MFzYZScn4i3S@YRrsFT*HBIe0E)W@F2HtB1Q9R?7vvsL0e+$pDBYki#q90M<)r1
zwDUtWe@SxiuVe5>|MTaQd+2ig=j*@U{(U3<zrQDohD^l{>=`6X;UKvg)u9{Y#FG*l
z@TGY}-Dt3BECN7r;3sp8|M0#o9{XfyOhTe6%9XWJKO8A74UGq4QAlhZE;6eBe9iy0
zmVW)uzq7>B{r&a7`*!E3!guc`N!*C~HHx}v(}4(H>0lCwogl%K(b4WdWJo=6LE__6
zWaSvKT-T_+b!=WvVLyqjbIfS1M4jQ{U)^U~pu`OPm5JotM#4t9ddP1XBk}(KZ?5G3
z?7IK^+rOUce{bxs+aMwFzgyq`bJoeI0}<j{x2}6jdP|?}NzEE4RyO%!q71pnq&8;W
zIJ7PlycWOJ?xHCha3DfN<*lEUAeR9`jUHX#@F$aQTPUKNWrhFyq0_&G$BN$6)s@pc
zo7{8a;u{i^4~YFYw~w$8lXxLfFtz1G)~<CMH~KTT(x5HNHx=}LMTV_^Vz*NcFg;4I
zi4F{M>!b7tE)oX6@7lzzyLZnao30-lG_+s2_<@pWX-P;7xi#wLYPb$$(xtk;%*%KF
zJ~ZwE114e+7O<V_9>Jf&QD4MgGyl(}3}mcHz~gIst7h`;lhObAzUKJO*|gozH1nDK
zKtwW=Ilr&(7H1zrQvH2|BM^Rb<!g2M>DIm?MUjx`i9~`H#n6(~3ua^2DiPFSd!gHg
z#v$ToL5?)5?BHnaO>qKs9KGlg%*uLQC4Mx0?}}f4{Xl<+@UpVD`+Gp1oz-20%9yRc
zKZj%;)}v!GQe$K}NF^3e+AA(sV$FHbGBF#NFdx*C$+k~@kM-jF3bTb9uY`moSh0h-
zd632N^Ow+5lQ0k`Fkc5;Ev%m{*6g3$Q@ol<_x_&Zro8Y5N<zR=SzSFXdi|nB&qRxH
zvjIQZ+O_oeZN~CzIpvOwSL|t_FK5v2_v^WA+nK$V8~{0ozzFe@3W5g4{WOA(f&7K^
zeXlyobBaNg`(g=b3fi~$@yDhnGKs{}Rb=(*J9GJ-#IWB_G_TtQPkXRNG+^y@j)==M
z`kfLetMRRd>gt$Brvx0QYN49WSs)?N;Xjvx5aJxrB2OLbOWjrY{^B{m*G`b4&fn{P
zFOlFk&f;(U{p;3l9a|lLCIYQ}t1LA|#Z#bHF;nFE>F*oAb?j|)E|~mo8ue?1+)(__
zRoI!|suoZPwPk#PSRuWCouFGik&cFLTJsL;W|LGVaM5;%(LmA>b<EhPH;99){Ih@m
zBHS%FHm7&uqP=g^WXA`kLL8kZ)1VUw1K(fAo{}=!j)e;)P98WtN?ft{a{GO2#UHnh
z_<gVb=Y9LM)GeaHZ!UniI==qrUhMDd`~Tqr*R5?t<WNH5=?nlRgo9EABHX_}nwfmj
zZd4$O@&yOGk-VCm{`Bv~b#OaaCzi{$PbX2nz^)PPlf%R3Fh3F+6lBiV=|>_C%h~@Y
zS&}#lm((muzsH<{4iYABgz?;G&CL6>K7@8K01{pJ(v6m+Me8uTr3XREBoTM=SCBBN
zB%TRS2RwwDe8ak`2?*y&e4I9MmI&GkUADcbsrd#w9g!Cbr-H)uVl(FT{=M|tBDpB~
zk79)GK!o@ZCYI-Ki&*n<;$)ER_^Qhk+sVnva4m2GXd3}(4+BL*^oc?H3|$WJo72Pe
ze+5_X6Z7eFP%egKbd;1tv&aij3hYvQVD|(&;XO~8M4#eldat)Y<sW{1)zuvbXaXzY
zi<16+DYK{D5&Q;;p$!4Tg&IfB;JXzSQ`t$JD~}!tp$;1x8ww}Bm`Wc1@ZbD=VUP`_
z3?)Iq5Mio_QP6;;+=m;czaL4pAtZX>4hR;~g9d$r$}U|0(#s7;XEh?Z42Ae^;24Hk
zg_9!_YKWK6(lGdtm!CfwGzeXV&#=)2Y25z=7M0XRHcKgFxfB=@X>Ok=4_lb5v%N7D
zW&jZpVWEdSz55G_xkegFSOu>D>-8v3h{=pz4zm%Y5W<%)cq9DHWV?%_h7Z3Lco-;!
z%CrR?ylT;iECy5IwA@UOZP{)<%=FSF9VMoif&Px3^MM1xMGykaHx4P7@@4-+ONJmA
zZ1}gV2Ic7T`d3WcyV4b8V0GC>Cn^B}ov7a%g*>)t^{&_#m(HAd3%0Uv@7@h}KESza
zTR}tD|IPhD7ZwH!vF^6>owqwN00VH7>YT>d^zos+B}{bbY1>^n(!oT>dP#pCz4CQZ
zk0(%OVV)NIAUpa58Tar9w3U1~%-%-$Jj}`Y;du@~89YD-5p44Lv*E6b7ry|~qX!vl
zvGa~rTQ9h-4A*&hdNbBZkCH)~<qc7FU%q@{-B&PM4}WjzJfR8p9_$$+esj23+6Fy;
z9T_JVegFR9I=z&5&*TMuEr6#yWQ%w+r9`=Nbf>7Lg}+7hafkq*NJ(H&%X~Qb1>;o=
zSyoE-@83w@TbQHoy=;5q?p;^lah((&2&fCVX>EYa^^*tjenG$Y%=ArIk(@>!rT9u(
z-a$dlWYnrq_9K`kBzCP}W7EpPy?dXh7sDi$)S0&*J_rR`smYR(lK4EBm+G#2J(pny
z7as^646FCVIOW(!p0tLq!i?Swm2)aqVEbC^Bj_~<5(WiDYrg@p#Zzzbw_9x9|9ef_
zvjjP6EIvT{F&mF#IqGrN>4)+`8^LAS(b9tl(>{A8tG!%M7fMC{gk`L8$DEfIrVNAx
zXaz~S2~MPKSJVcJ-T`(HEIlj~kBwLaTY~=ux)Vmga57D&_E5lu2E+U9sZ-&Tju}%3
z$!6@>R2miFQX~QNip`q^T4J&NyI3D~L+nq`Js%n}u#-fbYGvQ3(*Qm89#8rW9H?A3
z1qOF<1eBG9C8NL>*?y*84|xGgM^pPHk1--$%!6lH$42ntG{4OnQ0#v9@}=V7Nic9r
z8X%SE1eeU`gKDErRNUbcjiS6U0s2dMa<phu{n_qJHpkVsm)NbYDfmOe#E;l!vP_Ws
z*iGD-T_yHVzfwbBnYdi!+N~Tm<;2Z>K3+B;_WYHzSrBDJAJ2DJ^3~6IU^?6+*H+Ny
z<K&!3D_S%&;BMCWTF%!i_Z~cWP@fk^KRJv(H_PVS)kN|D(B-^lM(f&Hw|UJ%d;p$R
zUNfAl;{oIn5;szQho|232%<bg&;GNCIZGO~arGHc)QdBfaPXnC2*{k*j3$uH#4Mt&
zSh=ZpK&}uTn~{Ef>~;s6lL*$Vrq2zIjRLXE0=)#_vV;cOSC7kPd1fHinBv@I&WE0O
zDD1V!{3NRXPVKgLuI0`}9)*Rh<)kYeXYd&;#^1SncX0R4E~hbK>?CDgPMUkN?40@l
zT&)&?OWKH`9}q+{l2Sy+nZ7;&hzJl!2tXR}9ewP^GyUb|lUEoS8?VNm038JL30?6L
zTM2sglaZ1pd2#Q1{st1ZzIglgjOP<lrWk(|VeQk)p(fDub1vSF*<9(}yCVsUD+)M6
z;xjy|udDUDFE$6Ev-ly8m?MR~xr{3^(!jzB(JKZZ9Z^@p=%Ch;7Ujb6XsXxaS(tvk
zr;D#h3SGMWh7h`No6S}-L@&nbGvE^B960Ea>&9`yd6Te_u@9TC(*<x3%&lI(h*^9N
z1rGHOZ1RMiFS4?-;J>@QbX!*1qC)bxwc>Z(AuP>jr2nL=nrh<tNp+70DvH|WBM}~{
zJBJyClzbdLZiWRC_U){VTXgv|(i+U9se6q6IZ_<T|G6v5ACPd?)6=u~AUfb@XS<Fm
zTC!%{*!Cmri0O6bPM;o@FR!Mpt&PmOV|(ipVu<>4gJ0rA`L*Y00$H0CRm=>T(v??S
zqp<uJ4p&_Piy_|g2|LC`gP3a8Kr3@zl0mi3hOrL}XVOETBfmw9L<5*obRM82s;d(-
zHn&RKVtn@TrqpP9?)EEGm}9_-81%R4eqLVc)eZZVCrv6Xo6lRB{0G~sE+Yby;wNpk
z;t(FY->>$uO7TNswIi6aZ*zmYM-_S#cmSHi+SotnJ_+U@?K#3O(@sIhp$h1itYhob
z63COKselryV~2%={H?>+eo_$c<;U}@1s5*KY!RA%xw(h6G5P&K*DUX$ZHl)B*~F+(
zF2j<}FfqkvCMA!Uym1vLFm2#-_5J$y|H}hOs^h74ypB*SEsj!KSjHnTDG|UJ=}{Cz
z_4J_%j!!5kF1AQ#8oQQ>i8Ct-^79u4D74fT=ii*Q0+K4dKtbWq8h(#B60!kAjf;&;
zwYRzEOOco`9I|HlP20B|dqQX^%~BLsd9h;;{}j*r<PI1?#|=WRT)KqR1FKL8xdYq?
zJm#Aofr1AmkxbCIsgmJ55`*%^nsW!klLE#UPT{P6w;p1KmtUiWcY}cq>-PDdvJiQN
zTIGN)B*y?6P*KGX$dA~bdClk>+2nJ@Ec)lT&Di+Bgc7VjQq$$qhg+!So?mZ2q^L*x
zsMCzEP}7pW8;iSLOOX27{`5-0R^@XUFJJ8|>OZT`?D{c#D;5tJv%G$&omR%vD=Dut
z?zpL&rsT*4WSzeAG+Q=jl-vHN(zBzsZQD9*_TMpC^3B)YhP$4r?;w@<sr|dwiK*+~
z)YqrFwj3MxgnQAxEbMsw6ofuVht#HSMlXY)&{AKzF!@<_wi7EuK<m(F`7T3on7^Q$
zVi2TkiqRmato!ny6JqN4w`&0{)+q`K3d+jLKpZ+cI={r(i{U^#c+dcsTf-(bdc$*C
z&4yC2Vn_<ruu{nS42_s`y^0nlL8NEbuJeqH=zOX(N<@uHTMH(S?x7XqxgVN+pI7o9
zi68#d+aM=)>C1yRr^CpBKPYC|igqU;J6ZVS#iaW=IfA2q=ZW#$xgVRGLqagvb;YHL
zCgZ=yk0uwo;R7M8tcjlLo+6);@4W8Ue*KyZy_#<C!2PE@w1l|=Svt;2Z>A^TUq$|d
zs(t7dH=1O0g65G;z8u9ZR2ca?2%SVM$L{>%8C<sTX&%0@tCZAwS|iLH*xK5TI4}@d
zqnC5USU-^{$<JO;qJ<C%l7u75a4LsgU$Ei>7jsDO_u&+IjtiL9{$oAqJU96m>W)~q
z?DBfDJRJt%Mo&+BVlIOq*{$%sk;GVZ&36+~918FL(j0U2Q)T<~aq|5MXPgG6z;KRU
z{>F|^V!>1BLy`(`|G{2^7XS%G9WgxM>ir+2xq|yQvEitAwnak1dHUSBa}j;HUm+!N
zL-H-*g8+6o<P2XiF;K(A*I$3lLWy$D91sMd4l2Y5{oDQEA<ok|#3_YX$xW0U0SFk(
z;0qf@3U#U*DtxS){16}_e|<c>iK8<K9U2GGP$yP8*G*Vi)(8VQFl02eG-w}1pRs^o
z9+)q3(vrf$G0)qdmX_kBM=KgfzRsOGvH1mJ5_VkOk;x@&5%~X1@_Lrz#ECE=uKTo%
zLPTTZLvbw9UGw|r4~b|F7g*(s-#bVgnn6XSJN7s7@xqY~%E#<s@xKmkZi~d;-w#Io
z@gzcP;=khSzu*3SAO3q!e&2`x(*>dapaO?kmEt(lTaZznWl2c*&ZJpCgj`(%QJ@u&
zE7x6{#&`RZps<5ivQjl=D8jy4#hXUmAq{*WwkI3$uURVlA0EO!GmxXVufC-2u)lsm
zl;XOk_xV+}{kp{dzeCo4zy1HbCoiJJxb(jENk~0hzVan^5|quH_H>A1>o<05Uq+S+
z@?CXt5%~*%<^77$3UAX8!D)E|UXXeD`)8Hq_}F|MelI6Sm@>V2-~Z4bJix;5pS{r3
z<t1+HepvKO0MuXqiHVIR<+EH=RaG^X@w4?!lBMwIa-I{@QkEjL(4FY@&p(9Udw2s9
zf+@Ff<oZHFEbORVb(XrjPh#XBu0vNPPH7H_TM$my$NnjQiHDCH=4(zk@EkhtK+oju
zNO(e85@wy$bbI@&EP1cnK|Td21NmhAMPmu<L%vJQ$`Aku>|HT!#`(`#1@>?^C&vbV
z_kzglNdH50Sq^EQV-OYj2Neqg^rO)1od!>c<PvHZ=0SXE$p(-yFqsJ~gZbURT;8I~
zuSXvA``&PoQVtoR!zN>a-Z^c36^wh_kW?Ki@zIn{{N=net+2%|T=^5aTK}HR4CmGB
z4%si#*si1VD0c5xgd69RT{gA#7yx0y{4xheYWro1P}SYN+wb%PaB#VG;qU#(82~(R
z92@*JOwY>tN!>Y9Pj6hAP&RNLeAXf{B+r2d%G^hvC)8VOI5=e@+}`vUGSQ@21>)J*
z-9A+xKD3Jgx9euG4FtWGubwSEVvEn&PuA`JU}j=*?Ut|oRn6b=uv3FMQ<54ZNwUWT
zM^V_@WlB9?lUhKQffyQ){NriUJiGN*)}(x5s-tN<<eZg89-E<YtTNh-*C7)$PRvtQ
z+V1JA6f(*lP1DJx`rMvqD;Rm-2I+3Jfn`6`{bC(sqPHp}7XKAUGvkmGDW>#$i3;S8
zFJJldLgDgsQicncZdX?N;S4iq<wymE#XQH@x&cNbh7A)ZF_&2zs;MnIAA%N24vnkk
zYr27Al9cApy1VY}!>P;e4_rHs2``4(#uh%lE|PF+i`pm2{-3I5Z`Wqv$PRT<7S0=^
z=UVN1O{1VgX10YYo8El3<?l%hRD6o`Z|f7ff8dy5$Y~E+>53>WkRq3m8VwW(>U4`0
z91#D5?22(ys_eJma-|hxY^dYcdU}SFW3z-GdtkcgN8w(y=JEpeSL3P$wrOsrV8XdN
z-Cj8E76<V35xkx0=*+U~{_RwCuBYRfA;LlVSe&C~z>60zSSFCwoELrv6bRRHH;dgU
zujOWfobGKUT0m`D{_Xt5Yur~lJ~9S*p`%3x;|TAL3!<=P`v7k;xHroB{K<1SxhT5&
z#|2k4KL5s?d5OhjuU3ne;OKzukCB!q>Um(4{hFkf39(}~8fE?4k84o`W6jCR$wf59
ziWIJw{HRfzezt1A>L4NgE7uqc3t)a`&^rqEu}^H_dQwANc&2hxm`CzL=x7AaZy2eh
z#MV@U5d28vb$%%5<*KSc{DF+W*lRi8Lq1%-A_K!}*idjAd2dUA19&OusP<JdfNtpw
z%@wUNCeL=CTLvu4ept6|fzHR2awLGDJYI^X;3sJh^H_-rlx<Q;|DW47iAv~0#MOE)
z5^gM}j;)zmj_Ls0;2%TazF_9$6Po|{k^ifFyVspQkubTR$8~Z;BWgOkqMC5iao4@m
z@TjKx{jj!WmtF%M1m60kynP#x77T7R@_(d;gO;~5ae`yErOf0~Mvb!qK?<2Ba7igW
zm=F9zp&VbHk(Tz1wTE{B!rabk_@#NXBCB#QP>%_fp#>ZIfBj0vYm_{zRmJ%#IsV?&
z!?Gn$gX8d=(Ii+-&5K|qbSnIBPHjjLN<x#Gi$%bOTomN6IhS{(@PoxH-?wid8u5b%
zf2TenY$D5J<F4L22rve~q$-fZ7#>*pZF|X&t3~Q4p)fAG>jA+NKCAWhbG4!Sjt<M#
zvU&N4b<$J;Afk7kKli5;VGTh+yqJ=r0!Rdu0;GarkC-N=^-y%Q8T&TlPaZ3mE<JMD
z4{m%Y<8#uX%x%8rv=8%WSPC1d<!8%tDgTE|ez_|;rm(TJSC*My{K=*Q?{T$i-|BxD
z7<ICyrpclut5+k7Sa&v!^XwS~!PzFM5s%2ap#bunxHIYeLluWL%P2f&St5V_`6mFO
zT7{CL;!DnvJ6fWUUmEhE4kDzP6nh5~+`k>14iB3R7pff~m>0f0YfoRYdlY&Sk!E4$
z0BHevrk2o3^zrctK2THfB<Aq<ZITi?V$n}}jk6105N8=RrOA_&v@$nPX|iW+8c&I#
zlB-~STnCp2k=hxlLR@p<AQRliP0kEIl4jcg>Hy$vM~x3;W@}?p&GCdx2#eSr`Fx6q
zUAFmET@MBXRM3tl1d1jE^%+$suf}Q07zJdPkx}{E7*u16eHK^^Q?j5;QBkiLq8qce
z>NZ<`S=?djEio~Q02a?7orgww$}xyv^(OF=?oA`>$jm5w&RIPr{-}??f;b1&zH3UK
z6jnM&kr%TeR6v>aGE7#DMysFgL>0W5zpZ}mW605?mpH1R|FN>l?%au;dw3Y7VKv8n
z<+Uvy&Qh#~0{To`j1Zw2HhZtNo!tOslN%VU@^LvxitX-=x+@&aZ7-3lWc`(}ga;wC
zR4Pmhm<kfcCc0lZcJmia!_<Uf={T!PAVK%jo5$EPwMrGa>I$Du2p#S(&A)kb38(hW
z;(Z{?_bNAieCf^YqvR6d7l#i-$Z1+bZ)IFaIqKC-$#*O`MsdYRAZHg%rn5;&`}(U?
zy?>8%^;7tL%ijH|3dspiVXl>>U%=*LAdZKt%ROd`%9h@w6y*<<=rPeB!Fe9DLBVt2
z@gAGz-v-1uuJtj;mpDr+D)vdME%WfO0}>@4LR&&x&)o=yGK#LBL9ax0n4PVNh%dK(
z6?qdSA)1(D%VN|69Tu&iNQ_PHPz!}wW6>o|HH~*1nzV2RB7}pz3Qy`wb-wG;r5wp-
zJY*(zuwNVBZ*)uqeWHLt<K8ekO-^<_D{Bjlp6^X{uFlFfQH2ayVcZ7$JO%VsOmg51
z91t)c43i-+AuWrj?^AWD7;)TjJ-k*DA#z10ZYAg4JzculLE-k_M<-dTWo^vn^K-X&
zB?F&HlP8m&nEZ^`L%vvB?Rj}aP2i`<!4kR}cNX>PDB(M?v1HuiakR_o-``-7?7WqP
zmu1F7*y>7;Q97QEhZ5>cU^gz`>7Q2+vs8vINRDS$1#5~*-e6(XM@uBO^Vav59MKrD
zl+(}rZgD#-@<V@rQC`c~sFjsf@OaNeZ(4(crWn@POnk|4PJ|q+xp-@9uU7<n;fff)
zlg+Wm*b{eTrbSfB4`Aj<{{)Mf?h%n}bVTv=Vho}1Qt}_}<65Q_JuY}JGjwZ4FMX3M
z&iTzg3_H!W1=64Y)0)-;vp)ZRcTu+B7P33y(_U3qD^-x}+4Ju7olKAUHoSZ1U3I%_
zlN{}s7GkFWlMGsLfS_sfk@&m)ZOIPU7$It*ZHta@nSd5y(LedEeReb`F4Y?faDAyQ
z2T)|?fll4}YrgT%+W~)8`TJ*F1;<|L^^<Y2uDh{LdHC-&4u}5Dm<XRp+jd#Mhtq(H
z>z8536iu^bB^O7gpRf8v;$B>Ehg@lcHaOYs>Z(t(?U@N}Rcs`)50IN``?KFkq5z_A
zsaNX4V;Dkwa@J;HMjuq_F+820HIg<hBQAjk#w%8?oZ)sH(#dZwfItgD5o7e(Ter^$
zm`=bjD-&qHay!O(`yn~OtF89%NPMg$DY={Ak+XR6h`+YTqph)EOQ;wcJz)_}JV8Uc
zrs3(8jjP?~S4<e9WSiv0)o-5gZZ7$KuYsB`2v4f>=Uwk0gkKoULm-+g(m+F0MX)(y
zN->mS|AvxOXYJ<PeAFXV{uoWj(nc4cr@$mg)9!1vhr33p9nP9Lg^c!Wno1l}O%C}S
z8RO^WcNS{Dtyg&zq2uOA9x}ZT?CX5xSMxz@o_lLCX#+ghRj-%m@)@ZVo(IXH!@Gu>
zW!jz6O`P?#0C%oXo{x<P4jRjszR6tnX$MCS-NgsB$@f@XbaCkV&)y!sCm>*DQS=_O
z2SGtWNa6VMYPIW=IcFK-T%YBW+e8cHA^Jz)x~T`DyysQ<o1eQq@B<HVvl1`#aM8le
zeag8Y*QpD?u&+qFUGH!CQ)Ey;-b^$1W9OdItA~ffx+haiJ*Yot&c=8DW<P)K=-xuM
zK$YMF)Kxor3+}z7FSMbhaRxZ`-X(`qB6`9~B>W%0(C)hQtmE}IN}ijM6f1$F%#SOI
z^0MwJPtD@tkL$VQpF6s4XL;hIZMe%z&<4)`%&sixKXBkvK1+;$8t5Md$M>c8M}<d3
zAPp|~@Cp7MrclLlj8f#(CNV^Hn@>wZoO+Nk2JnSNb5D*=Cv~Tr4oS+fx;NtRoV_G8
zSH^7I0GPdsM~^5&>OpqbMUpnMb5nQ5JI*p7L1%tnp6YGp9a(r!?6)3HGGWCyy^!kF
zYr@l&o;;Gn*Lqz?wZ=#*W<qTifFM2hu!bl8j2B%c8RP)yi8N$#e6g@9UWBG9pK{r|
z-ju4`IeW(J%yx3=c*@z=6w<u`-(}3wa?8y9`cYkb_QY=VEfr<pkt5?C->P5c<fM39
z(Y|Ky;RuR8THLko*QTYW(m^>y0m?bc_;g%bC+oM2O%%;M;v4no-j<z5{2u%%(PN&y
zghXirt->b{<~K8k+tkDy^uNo~fqKz}p3Ua)^kmjJ3oYF3^H-Lp&TbiZiRKreDRnhB
z2e~C7YNR>&4Nc0buDj%Sas`C#uG50AY^YiH^`ldT>O5{N)2}pg!!(^{k}~age_%fH
zDg`#{dNPxT)c73jjNnHgn78FS)<YZZcak3qJ%XD4L3BXhknC{s%i`w%BLpi{Xl5e7
zN9?%d@nlqRv<y9#!PqN|dQ36GxXg8)%5^Me5{va`&mM{&%-op1g9eqd>`q)R_zCS|
zvywr}jKPoT@O+A-&=CRzZ;ln&;J}5@^VpyXi-Cw4xz%bLa0Pw{Y^~y@b_5F)s8SWr
z@Xnqz84bpai09EzY9u($8FPxGeV!DcymY^Q7l?A4!`MAN!!s~D`z2M?2dqP=D6s|z
zaR0`Tbk(d%jJGN)EA=}o1c@4&-+7|e_~lE!!EyP0v{1X0!(UMS;^m?ZWyE@ma2w`C
zALH`Y2(^KDfC=^%X~A<~YfQKCPCEJhnFmY*lLiJI0G>|_`X1I5$RDy?@k)?_4NT;7
zs!%_%A$TPQLQ7D>C}9gNJ!kKpUM3puo0}5I-FWw_RE(re%qI&k+f#ayr^-l2i`$Jk
zNO`iU;4VZe?k*_!b;1ux3-^3LUp4f13tn`SflNb}pwLZ=C@Ch#X?!3(;7E?bSti0s
z?d>~vyy#o1?0mxe2sR6=u&zzGQSf6v+hF3$p+jP(okUs6u`5u3^Zf~NXOE1^eDUsG
zoZo=(Kxl<9E@m%EWk*7|SeCh(RViZEzr`iwpKzh_^q0ASG?670zmPVdJ|kb8mk%<-
z-c+lZASq3G`5nQPfaJ(=&dFxvwb@6vfI`x+^W%i4yd-?`Qm`oGbvC$tew`}@->^Wj
z%dkJV!A%IU(wo3PA(8;G3Bio9PawTmfDaH^Sk&X&>;f9qo{3byacZ|4XKr3S^70Rf
zK~059p4=Ny6>`Q4X=M~hJCC1a4v!t(V|+}Z8;fd0l8mgZs>4@-SQY7vxIEQy&WU8o
z<|ftogY<@UJ;t1fnRb%p_un!A=B<ai8bco1OrlPn6#F=UjYx|}>h7^IOLwImm>9oL
zKPn|70~*LbX~!ZXBYRIz%g7K_Oxgm*cDW2s&%H7lLRToaS@@mLOq(6*X?yVNmdVVc
z$GE!wRjQQ-7|?lR(EQqUG%dLIdkg`mSVbC6$=#Jau}JWa9XX%(w=Fxl!fe2+dCw92
z@~d?p5=2_x3gj_p)2XrZMqi66bR$7#^7im(N8OlfrRC)X$FvQn2Jf{@OhS3ZJDKyO
zMlR35ER?YUce6GsI@cz3=N*>=Gr?45P7JAcJv^gIOMxEtW1q*1qINmR9Su}6)ZO%X
zmhJo>*D+(8L=u^Fq$zsTHkN6P>cAs6@=g*#YiwzsTYGz7GE0Xt!)VWf=4pl>)QjU6
z&KX}5_V3C7{R>#%ltvA~tO*_w@=^vB-sq|v(!Ez%sXKnWrR|)74xUs2pW1%<Wa$vr
zX!y<yC*bDEts)@I3@5VFj+s}-J~9|Mkwna8{V~=<*M_+lVx5DuqRU~}Z1>3>;Tdn0
zTuJM$u^Yv*$EO2#MtaUOG+a0@An9Rt_KfC8HMOqKT^_CN;vwt$n2)yKZ%=A?eYK9E
zVg5h=bSZ7!JkP#+>cxxXjL%(;2WCA=<70|yXM|I~AXSUyY@Rz)(iv)>yK%yZZ)PQj
zA9wnCNfSAVLk+?)=1PV!y|<lv!Y~WG<E+SP{=xc<sb-N*qw^1j^4>U0^S$JI28J_x
zV9<#0yRyZjYJGW<zwFKzyl;{Y6*EX1bBAJz8EcKR%K=mLw{N^8j*Tg6&Hjas<zyRa
znHU^iB|7DROb$_n);2an5gr@eO=Tdt_4%}u7cUCjWpiliJF<hKnZsR|&J#?V2TDeU
zhR135jW}k&A2>5IUw8J$V2<T1*{KsZSfCixzNwdO$mYXOecQ^O*Ts1T9~X?K$P!#$
z^&}q1w1-7R=&~%`WhDV|l?+3ua%XWS;Ma|r1?Bo?Pmvzy{1~YpxNI879wEy#*QcMn
z{51YV8FkI4WkLKy4xrof#GwCp0Nb-ImQjUs*X!i#cWm1N(7R)0O_1+W;28a<ZJPbI
z5$-&w7OVWaVMsUZ*%s;F<I*P@Y(u*I`J61r;(faCwT0W5EwvhsGpvR_a{@YnHuQTF
zs1|;j&C5XkC6iM#4I1$AfnQ*}apzcFT?ens6l*l5T7(*SA;I$v@p3?AA}0vLssAQ^
zAw09k>N1Er*TgaoUT`8j9O~EW7cWj0Vo_qY?!^UDosTY?A3t)_V&QJy7hBKhExBlz
zTP<Q+hq89Y5hppO_n=xecW#&N-TOP9=*>GlJUZsq*QLPjLd7L>a(BT5EcJ9G$PH_A
zD{|9fGX-Vo#SvPmU=QrsLl2u&JGOqpMbv;~H(gxlWoT?1`XK8ONfo8o@z=c}3{kO!
zWX+yFUhVMp>+M{G#kHNf0KG&rkp*Py5g$x#zwX%|EirZnP^W^=BpscEjvaHlTJO(&
zda|!Pl8<>{EE~$fseVmAXSy%oWq&Q$vqoEs#x!)b8jI0vg}zv0x}{7IixQ-!kbF<-
z`Ylk7roO4WckUE}j2EYZ&~pumL)<_lGB1RlIDtyajNuWp-s~9}QfqB%n>V7nwiz`(
zn_qF1uzVSjszte*x4@$nJ_vijp6>H+XoZ5AR<uoKPdj?RpF&R(DCrHv18J6*mJ$qD
ztE&Kuik8oG=nziUXYkw`yG5D{m<N2w4^oNB4=v*vZrv&YdYN^Mkkz}*F|~kfc-+M(
zk`V%|vwsmy5eMIt+qt>!oM0p7Wt=N?qnJEr^%)kB@yk^k*2>B_m3i*!uWT~?sb$`I
zCkJXvlT6>ZLN{T~cK<dVjHL@a^E&SEcUHU;F3s6(wKoS$<s|86X2aS2n<>gykR+@s
zUp(R`D7A9M)g>MtYWF+Ooj)G{A<<zcY9%$gdVr9Rt@KoApNpaSaw@gmjU5u)yVmvY
z5<Ot@xLL>c-8s7E%wxdmn4RP^qJ$w=`@2IYN<8q7Yv^Z|?r>)jcRmtahp5_j8r_sq
zs<0(M4}f8i`WkWNbbRHFk&`DjLyZ9z;Hbrq6NE<)QV1Qq>71<P<l<H3HTUoRFZSLu
zsLQO|_Qmd2+ni7dDs3PLDp?d2On`s^5F{xkk~x5=L~Uau7(hURi6lWpMT`U+MFByA
zfP#XEf`ADL5)?SUN8f$!saticZk-ReZk^hG*t=iDpZ~L-wdR_0j4{W2^29LY#bfY~
zW4A8XliQtIB6cQVQ$KPVAyai7OXqI=A+`FTIU;}2PLO3I#Bh_rvE*j`pCu3UwBEIQ
zHyYU#gWfxi*ORq4M26O)9#F0jq;E-S;ivj7yX_50)NkgYR?le-^^|hkgZ7zn7(k69
zULHSo?Afs$Vffx=rEO7K;K+$Nl=7*1_DoHjD9>(JJeMP%tn2qwY6_zebGBqYvcNge
z#z=icuR{dzMFRi;CoLU%d3nm79c7ZUeBApAs3|q<mw->;);DTN_%glvmz~TMwj9~d
zh9OZJ4jfVY-Gl4k`CHY8lyr1<A~{-;D2O{Ls}p9Vq<MC^ZumIsS!B?5Si2T#D!=yb
zeYra8Gu;Hi<%MO@(YIXwKF38feM8a0TOL3_ms55*ol$CJY+*rF|L{K9p6!1gES`f3
zywCGyqw$O9tdMHu4o~l>l?>>QGH+Pro}w@ELP5VpFt(auXmD^=Dq5m(c-yzd%y_b7
z#k7`<Y9V~XdhJD{UbZwg4*k6O*-S#VZhCLrh_h|RTBf=tJ6{uK-3`dz`v;N^u6MOs
z3zU3C)+QyDl{SiZP9AT6+eSrNq_LP>)-K&!5QkZO)T?^?cKzbE>m}rkQR6+C;6@yj
z%zXvw+k?m&ql-7(WMsVHTwHm;4-f=_2|cxWlJz6I7O$=>?ZBmZ(3N%a9@-9V3BDp)
zx&lC3W@smIRyP7PJV@#LgQ^qO2kt0x))XGPMCq4kThQV(;x3}VMDPHIn%!x_w!m>=
zz<@vlhnRoF-Aj-I9o&*?zU1g(Ybz_v6&|cfJebnAA>P#PtlFI;IZ*>9!gw!G9PZ4O
zL8)gKGpv3wD$-}@49jv590f%iP{|)sc1R?>F1n$_^&BK4Rp1ty{p~8n>2<wf=^Dk>
zO5+1shEW=Y&@Rs*-etTGAo19Hl>w2rQcsXA&3;Lb{l5lp4&<Kl_+E4WhDXwfzW<sP
z)yE7yFTKjsXNWgLwa6(;KK$7uLm*8DV*R!|N9|zoT9jq%pmYSG(V@(9^?Rh7`LHe7
z)~;$qIXpvfPVpB&<k<8e4)UR4OSToe0en-|I=>_%RDU78<L3FzW7}<)TiJP3H8(XC
zea||J8`i&zcdWZUaXycXuqm=8FfdSf(m=^OO_x4S7<#>5(8Fy>U}cM53tGDCh7(@C
zcu|)S5MJmg7+VLlZ7F5h4R39{a*oy5U0<YQ_ZW4(FlY_;V)Au%OB`?u`F%iu8I>mB
z#w}`ZNU1XA=PzGcCoZ&eba8fm1zWharUQZu2i-+qS{ToCDphvw;!e+Q(Zj|f;yQ2{
zCsPN?EEY|>DKlwO%yZN8-UL{7{2_g1%$gtVD)>RNEdN6bltqBV7-2?>7WSiVR0uvx
zPpXylL(kL$fP)EjJ)`13ehiacHttvtD2n;rkcRZn1JgEEzl(ECEC(KTmZT!%B`5zl
z|EkCTgO~HF^6PfTu6Vl0Iusb<BI;=C^xVFfI<KW%*r}>&%a*&;p58M|N5Fd#1WA+~
z9T!z_&RO@W&Bjyot=8L6lUzSJG1O(nQDwUk6|<Iny9lM7e|DCTj6K)s)n~}KDWO_*
z{9g6JI4+jaJ4b9#3*qK?GVS-1Bd=#)80Z<^J>aO*T8FJy!uKe?DXKbx?M>#I?W=1h
zwk#k>Oj=v|35_a9G#8dCmnS!S<-OgC3p(_Dq*~SDdPWN#oKMe6YJ0i^O;H-Jx-hqw
zp`qbdS--21Tdb@?4BYdoVlE9!-FP?q&ip${4g7o*%wl6&A3h&)eOHzLaaW1u1BFKN
zBEvsDDKR$uX{K{ncTUt1A8{e_*tG@ZRZjF=VyBnNeVTLJ70V=mj-vRmz(5(#_3t79
z>L{|GL@zwAvF+Sd8Q<YJK5+9Jj_x<K1#@7FB)yM>8rMyC;v+w4qVR)yBwb3ld^sm@
z>IZ<<&@YvUs=7SK)-URuJTO8;#rtOxB7Xkk%c^|)^(L?T`!CQwIWpKsjid$QzyC7_
zI2nO`T@qT#o`jPrjW-v3X>_M~1+|VCM@|VIC}bgG%EDF!iuU&L0m?AlsApH9<2x)d
zWRF&2WxUMb{*m61T_?8RyK55~92&H7cD?>|{c#QS^Q>It9FK_9);;a*!GnL4op+a4
zmMN2N_1SxAa@0e!6LAP&>PGqGdB?=WOu4z{<j2s`oE(FYnslmTGLTa%MlDOIKn{9)
z80P^Q{fO3<blaXeW)nvDNL07E)Lu9JXIr##iTqxf(-)1GH5(6dXB%nai8rJF4>&O5
zijZo7x$1a}*3-M8c1})V!6t9HLGG+P$F7{|2}ni!VEkb$W;2XZx-MF!(XPg9#0gjS
z!=n?fA)Cjw+b+AV-zN<asouk42?2ciMnyxHg;~dq6QO-Fd=ISalPG=0_EYs3vGz#Z
zcJPI{$wb}Moy2uAc*yf1GBb1Vp|N|?`uU1c=dC#GlUuyMQ-MSVJz#Wigy-YK&Xc<S
z=Rs~PRI;C$_R(gYLMVwGqpLrXilG=U4ILI75+cYwz+VF<FSi#|26xM5j==FI4df0C
zQASo)^3E~WAP%d%V(?cFrL-mfC!gM#ZuVmxm%Z!qmlTzDwskXQU;i*@Yl?s6lAvrn
zq(t%MnS-07Aq~vS_WpZ)!|8)=>S{f0mPiGkQ)y~;8GXCCFTJPRIz||C9WeMw+JeQ`
zcfRerG_32d5&xvHiwL{BUb3N!R`$1VQykZ-CLtt30`AGr;|tf%d6t!x1&)#o98Bp4
z)>2mCFZ^!?ATF4^!?aJ@<&=%V<62G2)%um?kMJMpPPlw-DAok>B_O(~cF%HN#W={%
zRBEi>i#a*P#}LL+ns42(;SPprWs~0pFGpNi_G7whkd*W81pWBz!T$n{2aPq$T#$77
z|F*id=RC&$d(_!i=56q%fSc$Y*KwgxJOeff)h*R%X8ont<omQgooT#u>@w@!Z_`?M
z9hyPDBHCeFc_AH`KmYGN@?rl*<vR<(F-*xdG34a6i^w(W#ZOf<-;6xA0JtgmU4K0;
z|JKryfddAps`On1a$)DNG<NYJ|2f9j{aY7L;G(z8j|%;L-h#62aVA>_l{H;f3_dfS
zo8MKMMFm{Q%ho!h1LS(}6X}$s_5m}2kjii_Cx1A${GqMsbYBT^iB>hKH}|2zR0$67
z#GENy_`2AAfYZa5ePM_eE~)cw+II>4vuwvhmtLbLA!}~`{$1GqK%_HgT8decdN?dT
zV7a&C;kFENYdgm(q|#q}Jf4cbud4bh&nA4_X|+nq$CgTAh_Jd!#425fp!bb2my`>)
zOq=PR{E=4&ReSnMsWA7A?@(X!CR?2Rmpjyds(GB^R8y&tsEM)t%zx|Km?*IDjfDbe
z;o2JzB=Uswjva+=razvn^vHM)V5d?t?^zZ*n_W+#cLxp>1%f~jgXm;b#8PbBCRv<5
zR+H1qglHCt;*MOk;Mp?))I_8JM_wH}SJ!@sQhVG#4iXKi8|RNvQd3j=kUe+tqPvk6
zn8&sBwW$4psrO`)$fLJy$(eLND}-*6hKawq6VVMdkI&|Zrjfqyv;FnP#HcI^bD15Q
zRlC91yFM%MSddz7ucC)(!fP^$K6b*Ctm4onUoy|4fBYLBiu(%{Si+ib#uI1HrWe_`
zW}@jt*ZzGj%?#aLrGQF>{xWJ$D6Y27)=d}&G1IwjoWK}iN|^MefG;zmXf5WS9bn-!
z#sD=dQfe^wWQO{|Slv0kb!JY^&Z6mg56#neu95IOyy*E3b(yH!$|#pE1dV@35M<qy
z?OzY{b(Qed{U&KT_yqr9PeCVsSGS1x|MA1cbFA5(J$uLo^%7Gk<UzqWoa&$6;M`-^
z>{stGtjM;xAneQ;xmBy?dVE?Za$e}j-FM;<bZxQ154ytUN6zisVm(_`$3vRzf2&>G
zfwFM2Edjnl_HX1@Xmsl9wAE=-a`VEt-f|v3LUuCI+(xB6uH*an0WVz>tw8X0NfLCK
z?RrW1$P;l&r9(FwxdCN1yh0mA_^KZM_Wd0%#@!TUkPqjq=mt%SYF-9WHO)=oMl6b%
zg^81}2g<s2gR1lC;=h^vd^+091NQP`eiJ;T{(#JQty+<KxBR@ksj{=y=vE(W6=*I`
zgI4lDP`gVnp9UyAhYyV0m8-Q;m^Y|B-mc%srrJE~J3H-C^$uNHv%M>Oh(lG<ty^Ch
z-p%75%)Uewk)M{_jkRU`%NQu)PFbH*m+JVqhSA#G2Zv@PR>q))kY71qCLT9Dn{v_q
zeA9??x)I`YT%8e7WJ>D}cE27`NoGkbm^^ivzJ3Qqy0GVLC5c0~l$T+!X^)_tAEn`O
zOOz8U44NRoxz-?ugJ3oRVe;7~xNkm6Spa7?(R$igU10AC2{N<(c>M6;B8PrR5aJJJ
z+TH7{2cqy1`urOU|75Du+1Uy&=<&E(L4lpBm)T|?*CSTR%v0B851k(GtkrENVE60L
z$lO#N9Jog-S#8Y3n-2NoCcb=?yeTMWvca-tudl5?vUL{i9mEn*ZJ|+MZTELF3ZL2Q
z2FsaW32Tu!;M}kxdJs*0-)ij+kmAZtoxQzk--#0^Y~B#iTK{g<+){oc`{54X=N#+}
zZ(#JyYF9BG`}~X@;cP@w=;a~aT3gE1He?7ejQtgVZuC>%RA<iIOHgq=MB?Tj{f@c3
zBV;ABd(HoQ+y0N2Q@x_VEBl|S7H%&0Oyde6<V`+_8J1tyiSI*oeYv#aXVXIIN;7AL
z3yFASkk`RW7b<&T`o#zC>9=gWgKjD?048$J?`{FNCf(euc~Yq{_Jk|A4e~qH=y?e%
z;<XkE(-FYCZcXjFQpD3&a9wK*v^nxrzi8RW5yxE#Kmaj0509hgibr;O!yRV5Xv#9A
zF-gu-tD%+py`McxbgGqUh}=iE4c*D6`Xz!>?C#Gi@o^=xWG^jcsKu${uBe9c49o?6
zn#Q(ismf7YuC!rrvfK|_L7lu1R50+b0r`Q#5s6MfLlZJ1{^D_D1J0TqT%|y7NJ-$z
zH*^OgAc=ja88kRz#flXakUxDZ+!xC%J9oLI0C5%3`|`^x(AzW$*c_rc@=s;Uvie1u
zK$s(=<>ZGs{HZ=8Gk?;~n$DgDE~QC^3d0;qM-{zY1Y_t9?x%aRgWDVG$*f+O8s9HH
zQRsFGIvh?DHearE%Xu&d_uz*gBb%Q9My!i_L^2EM=>N}H^rwH>&L2wbdkK_D5K6Y%
z0!}K(DV`?^<E_lGYkb8z<t{$qTZ*=`{!0<W7Jz3rQGBL=$~{j+$Kbp?o9Wd`tP8k5
zxkh=GNMNjwl(x;?kTQ(qTR3EZ@RPu}sKy9D*EPz$pq;CCH&JN}s=TV`w*M+3r1`xA
zl+d6(BuKRMk4oz9Nf<L`%tI+XWI=-N_}^3}b1J;5^5YhT$k(}9eq^a>h2MW^0hBh<
zXlyOLAd~jpzy3%T=EWua_3#T6m=~T80WJf-E!@+e>!K|6Me^}>FQvYHv;vHmMHJ?S
z&3x9LqCpw7!pc(>*Fah;ing%T@8MhTI<mX@e9xcD+-W_Yf4|~m<av2MqQPHMC6Yf4
zKK9$PU=<y!Jpdzg509JWA~@mk5nxXwl0e8eZ`h;Ij2veDj5<8-#&)6*N<XB}d+B?|
zCu`@<CQhaxlBRJNLoNZOr!z}lzjDb|>LcD?piIH0@*DLqLXrI<d`^XhrGUkfkZ1tI
z@o7*{C%nC>Q~$xH8u$^}1Ltb$_q8fWA2(cNsy3%Qss_qZ0QOj2ej@8D!2NFe_8<WA
ztH@Xi6uve#wu|nJ=#_PKV9mSVM9!RT(KWn=sMvmY--pf0Fx-=^^=8j$2B&Tq)c@8N
zVF*{ll6ux|xz#5Uk?O^w1_PY>>Cb0k8B(9|3ENro)G2`m`NkHiInSOg@9HO^wl?D+
zr5y62*y3THyJ#d5#2sIZm!I9f<75<hKbTxmg_O4EWeq!_SJQPD!mmkJ)~sKTMdOYQ
z%SlK^PM8QcN?I6HdIys`C`Ti`&R3a$S3hlW&ElE-n0n;lmzQfF9FA36H>uvrN!sAF
z(rt5=MSVD-#NT5YDuBQhqYy)9kE>>FirN!TrG5WAYr_FLdy7I4`S{Jb>pW)~Xke-)
zu=S{|30ZPc4qfbF1BD){?2U?t$4}Jt$yzvM?3gi;+q4f1JX~nD=)pbmk!Kr|#l0Q{
zYalKvODZJm3#-2`4166^oRhPW<kX;F8~zN!`}zEi&0%Ua{YlKl-PZbSA9@+&6J)w0
zrxM6N$_n%E4rwk4Q9D%%{qyX}Y_xupUMhU;UC4op*Y3RwO&wTX0(9M;vW;^f(dUOZ
zl7{JT>hGXmzkc3T$nT5qi}pv8BV-u5(_2|Ksy6?8Vp}DhWQFnT9UNq@6d-)0T1?es
z*)x*Y0x&^(mpy3x^wK=$8AI6WuRR2)^Mu0W%NR-xSG15IA`~4x6?wi9J&w_99nEa>
zh{ED{^JO=iD5;|YGHuh}U<0V{uv9D?ZX5Ai(e$XV&-#y)&-y0tlSTHaDJd;+b~Q!O
zACm%CeY!x&*r0gC`|{;$!sgs?Ab4CI+V;yWy{?pOv`cCbH%+ve6R<24wy$3FkNRV0
zme_u2FzS@*wZ=uz$GEFuL8~s2hw}5+fhr^yM{NdrLwJY+UY|cE4oLX-J_Ao~S&I|)
zU0;L@1UJS&HS>k75<rT-f$rn-EpQZ$j*c9}fiAl3d9`ofBC2X_sb79z!0pL2tRYhl
zVR9E?9`eV9sLMu6SP>c)_5*&KY+x6J^_lr*x3{m|aX;$@!(~x!45|=bBdYd0pm<~y
z=#*1EzL%nN#bA}TXlngU1sC@3kyaZCro~&}3LdMx?jcz?y$>jt1l%740gLUGXWG=b
zx`Co|T=Xsa7{!;MQ%NCY3ZKrzXwU)!0|WX4kw~+#xER|6KW`A%(G4n@9Y)VE(@y=-
zkA1?~(J=c?xietYl&puxfoB2-xl6VPZTObgCykh8Zwjj{P_V!E%?U>y6tg<}<aJtJ
zlN?Ssuf4HZde!v%u;4j4^0MF8xIkElHVuR5U&hc2(oiDSJ*otH-B-kWut-Gr$G(zS
zbN2>VSe3^OEa;DqhUch!i7`gK#!av7_AM7-+mj~>^hYWye+8i>Gj+gI<e-}#!DRa8
z7k@pKoal&RN$EiwubL`Llg@6yDLMw}O6G#H1TqnnT`;86rB83}VNW0qK&ZhP4!twv
zC4@Jn1u}sf1X(JZ3SvDeKFz%Z4py4p3^u0V!Td^DmH)SR^;oqS+6R8_7ZL@uEI-&L
zcNN71$pcv!V=A|%C3RZsaE9P^K7g-cor?-api0?R_=diGd5@Yr!(=Gx;T4@(tCH+}
zPsY5XoWv7E7_i5C&B(~OVI>R~v9Nf--wIS*UCEI|=SeJ^1A!PXoAGDSyJ7+EUi%%*
z-_U6*#q%>Vs)@nD7J~#E7E-S7_?w(79ILvPYWn)>ofqgxso~*^ewb>KZB$lPiu<<H
zH1k6k8rm^Xm_<Wk*V5na{TvaIq{3cR6PqXJ>uQrF{%fQIsK$>q9V{DqaGp?$lfx*r
zk11Kcc#*7Qd_ky0tkAY_-Kl2ro(#Gq50$+(FbJF?jus);VEK99ZsD|OkHva5*v~J#
z=smAm>|?+g=kl}J4iT3O?R~)c!7^jNtoyOq`7FmsevV#izcZ9BCw*flGg2UQ+5{Q9
zrm{Lsa;_EGCi3x|2i~mTmZ`JsBS@@Buia66>$)V)-TxbvRPL+&Dsv=pI5BfgO#PTA
zRhIKHsgTXdzF_-oe0L1F1iHZD_ZJ83^ei5lpTu!WZ;hU5h;-H<sz*?5^{8FFeq9sg
z!_};~qRm7H=hohRN@SNDvy&UKh_uM=Qgq11>4+y>xjP2AtS95I-uqZSsgVE4O<~nq
z$`-~^BeCc`*_!^d==hUkpD)&uT;+OT4b2p`+YSu1brk-6CuD%xx#%2?Fs>AP$6mR>
zfXuR#PqycQk!^w<v-365j^Ok_VaROaQ*$2l<doGx$iw-a`}a!a9jC|blT-yITvy^~
z|JrfRyLacm+K~H)eZ=olqxnA~S~5iYu^8GEMUZp1H8+pg8WzNDH!Yx_9Su>!R7d5n
zkAWSYM8D0|Lcx1?`<wZo)7M+&$Vq@Digd#Mz)$BkLzmU8;wF*a36AiDsE-lg@sMx$
zm)+~kPoPj?`bf;f)4QAp96CFDuZolYY*m902i0ertmK(^HLtdI&GB(_H2?bR0%5h!
zZebuP@0&R!6Q5Ru{mu|!DHM8_mPsAx*-8k8fJ9!ELDx6MjQx-CDjs@w?&DN}g#;HP
z@;;;3K&}P4Ec*26BQe59b!tVPLbV?rHd7)_g?RNv{J>LU51zfo^$4{!)VA7;885Q3
zGi%4H3mRjNNJEY9Fq&Oq_ZJ-9_XK3{Ge55Xs3~&a9bp6bvN-ozuh&ds%W*~}mp_{X
zdm+$=bdN;E#ztQkmhnRw>QYO(t7jyK%p>Tp!gZ^t<Tc6I_l}M`_wL2tV1B7P{sK5K
z;pQAG6%TL;X_5PCr-Zl}qn1-!XCs~_KU$lypqQrjszVBv)V66nJq6Gp&t6w=-u(PL
z@wHIbRPJAEMbinVe<h<KltSm}pa~!$Y5}jn&!e*aShb~8>*7VbUtc_R%Yhm5!t55(
z9ATc-zd1Bda4$Ti#4bLcNRQXSej`{v`|T=6TAqY%-svJz;HXyi-4{ElXV-7s_yHe5
zkwWHv7h?0s!Gn`zWJW2h$fYTYFxcn8WeML$rKK)JR1ga`S&7aC`k^r5(TB`fl`)h~
zw;L*F#_`zTD;Qh1vLr&I23<O}O?QQXUntK83>1C|^f@I4M;yVMQ%z1yRh1F;4iu}z
zvkrzM<)qy!&%^_vn>vzp5c>59!SrV3QGju0z?e7lb?w6?Sskd`8N3tjJANuwcN;fs
zKzNnSW}_Z4a?xKbc)e4<kX}5X<A^$m2u3Qhi&2{g4lLl~=sy!|!36<ezY8DGNeLU3
z%w>Kyhz=hbaOdIK>-5i8a%Yf@+pl{~fd;`(Pc?k14I38)BDzA<4LBJH;(fya`4UJ2
z;C?Rd*$$QFCe+R^(E^bvaFbv(mG~t6E6;>*8R!==8w{Ng*w}k!-Xi@YM~vGwz~4UA
zpZHBTgd{f8__KJpyYqHNB4Zmnb_;zP@7|e0?oo754X8U5{kEvz2l)ZIC&qQn(=pk3
z=E#x9>?`z{9@H(s?z~c=uzLJ>7B$$cS={f}AmjLfc}CS^lTJwRtV8<u?*!-(-U5OV
z(wm9`VO1yr;#SWrWlC0uAO%RY0p=-0Lc{ZvmX+P%O6U02(fsnY?a*n()q*CAt67`h
zW3|X1K{(Ez`<gP1HE`=}iQQU!4?&QlV;vg+fVP=lK7R2$9R2AeGvvnd&z+aZ@tP>J
zAw(Xsj_40tSUiOJEEdO}&X}d(FgXh~m>XUM`oL1tM9@yU!H4c}ak)ZfM0VxY+F-YM
zO>t4tP8#jd?a(@ve*XMxs&le%s6fgGvLFj4hfvYpBvYhnnAxGI%}8oq>M=uMwh02q
z5#us&`hxVanDu-QFNN3K%~b+8z>XmJQz(xbI~HO4$_sP%bMA3%QHNV5(Ex~0U3<jW
zVeMLFs4$E%!qCc-!!XB4I3DC{n>SykU4m~wi(R5*NJ-*#$9I5@y3O=O*e!dk`XDcl
zw89RaEERVkzC5?YfxrJIX&`Sw7|Z<z0BJ<^$R=W5dkpVXJV%GYXiSrfV@D!RtfR^y
zoU(Pu5uTtAp%OuIT>>$}mZCf8^_t}zbH3W;+mVE#NAK!?Yus=>>tWkCk$vL0xlxoF
z)EBZ7`u`KzGnP2A;?mNw72o&`6%D<9GdGh|%rC0MxxF<wMJk!){gC(jScYM7H|CX`
zU$z|N*WZ6Rk7~4T?b;n&1H_u(YvwcmVq747vw%GF^$~uMxYkxxoqRAy{FhIEWDO~V
z+NZ@O%NC;PVzvcrFUOU)v-b1n`*PpDeY+b!nN`N_A^6Tc|5XqqMEYq@*V)?Hjif>C
z!KV3ZpxL;$h6f=&_(+3EZA1R!O*qPDbS}{3K;s4BdVE(E3gwLcqy7ft;aYoVAtfSq
zB61qN$iZsN%SQa*E$oCT$KOxryE4_)>`NG5rD?U)g6EEELIV7_fP9nqx>M`$;Hqb9
zw`@tcHA;snn?04w(+5^o7a`-hZdrbD+T_Xl+oTV0^;3kDuI%d7_-3MDTzOKTqSm#a
z6L$mY?wTvz?jbF;uLGv7a)o)SZCHF<f5{SBw+{V$J3rd{o~)q9$drbo9ohU4pig0y
zjcDB?`^=Okf<JdNUNaDko8$VAA^t@VOB_Yye|j|7Rw9jJ5r4LMqUBOD_tKi@y0(P?
zAbI{n$J=Dqll)t6r3zR7vN!+yuB1qQ3o`bPA*mrf${J3s??;eEOMiXtL-dJDR2^uE
z?FhAVwV#Sz&UST)eVRY5{DOmMj7znxBeQw+=G=Nj%Cb+TG@(Yg!_HjOo;eX`ItsCB
zKQe222VMxvp<mBE!25bAJcJj6_U?2x#|~RP`>DTdK9Mjl*|c463wwdxT<yoh(hU_7
zuPvh0Kd|;Q@9X~K1Q^ZBWm~6#-WV=jTDnEZeCfGS-sZJa7bIS4j;k#m0qxGX&;{Qc
zaDLhjF%PyQt?Tw5b05d^69qVKx1xx<?a$S!T3<q1aAmECh}KxvQhoPmk*X0oCV~~~
z&tIa-XeS%zF%V_?@+I1u)RAYD*8Y9Rt(S<D^y<~Cop$#UA0;KV?K7Kc-^a{frmVt0
z{BN6qU+ee(ne~EY$UWT6miHex5V1tG@PgV8GL0rDkO?ej#^X#S4J}#w`L}%ouQK64
zV>jMh;zNfIH;FTy?9?XMSBk7z2QxO{&p(ye9a(?<Z(E=Iw6vJ}h*Y#2Q@YyEn01wX
zAZHassX;$59fWJpF!vG>$?C<S6E<8{(tmv;9Cdq`)SFELm`VHTcmdfG{;Ejh<ca^=
zU+DiR=a#CTYrEg~GBz6|_1HmVp3xsVy)_~Wa;NUh?rrpH`-9%<Ls5mRY}p%WlBD9M
zw)}~7P|hmDyp>K*0$v$8X-qL+X>{@7Alo31O6^*Y&$GU~a{D}RpwgPx`Kz~oVW4U7
z{+X)jZl7lko$%9=OV|{1kHYT!=hygkbfx|Jb?~S#?cnFH|NhG&|A($d!qQ7gE|S7a
z{*grA{Vj4*`QCJCj2J$A;4SKi^0Kmjax^v;*9kqJ*S`rHqR&C8pYG!q&J-tDKQAlm
z0F+VcImuX>fK(9WtGA8$OZ!x0ep=J4=?zsT9~TEigsXsYiQq6Qw)WTEvtsDz(dIxS
z6FRht$d5@#t!xCJ?h8i?fBI@)zChbLrJ9b3*|5Qz1p~QvzqX`>-l>PL;mH631D^|)
zzek7u^!K`Y{N%}#a5t6~`bc&VD0)YN4~guX@M}|-8UvY7+F4ulQ^i|D<ihP?zy6uT
zfCbxWs{{}jX||p}5Yc(~>kCA~j$o~FNNY9@eRO00`BM04S^xFVNcidho-g&(N653}
zdj02L#TFRfJ~V7LzqLl@KR=o7U%^Rj{+Ji|9-!5quuXW-H3xqlynX8Qh|qyEx_Yj?
zGb=nd`>&s8bB*1hKOe?3xr!RN+EMVh`>%fxASkW@=6Y*RQd28_`t-_~a^Z`K{W{ZY
z?0_SPti3f4iqrolVzlmt=Gmv^Tli0r#$~(V#sLw}SiZpZm({RJxM>k87&6x>O9)|L
zU#n0+vljNFJU=Br&nn;>8Lj)Z@TdE;<o1EKx<75)C;fWGM)Ks7$+U*Z9HU@pq-d5d
z7{>?AJ@M-`8XO|S>TGX6Gi5r7=<uHp^4IvEPiSOyCM-b*7<kN#tF{utBh)GXyaV<X
zO_)Bt6sdEKp#h-v(sQ#d{gwzfnAlH~(F-@|>Y$m=Z1Vl;aDG>A@2{md7#48hRcYyT
zHMJX)B!!Q2{LlB6=X;aBdf%hIU~Zjp#wTGX|K|@UfKFj6cC32Z&*k4|_G>rEtrC5^
z^VhZra9__~7pV!A?j^F~8rd1g#PS|qY8IA^<vM$I)z+<7-aO*77oXKEvEqwfXc42J
z6sB;0S|QRi#p>)*hfkQ?bw5Y`&5$}fepmQqkO+*)q=1U@1d(pbGym6gkpB+D|3AN!
zK3=X6USO4!17H;bHSc>(&CO2eh?#?R{kor@9{^1gDGrT_jKqR0JgeoQsji-U&|0vY
z`|;pc2#GrC>+6e=03Z%UHj-hVZ?!cwV`uKPMGyx^mjCqWO@jf(euOfJl$w|5Of^wi
z`U%nFzV--zA&QCE9mZUPQh_mpN5CTni>o{=&l9$H_pqNEeBX6~p(dyVGnE_&LYFH>
z@rPC7dOxQPoHk<%V$-k?*+fFcP~lH3b$|WIT7awLM~`mbwJX_W0e`U1VU#kaJR(L`
zQ!K8|jW5}lY(;B33Zy-5ZuHp*6)CK;yq@~jSM3<R$i+?<PHfMNpJ#J!EQ9ZQFbu!~
z6y=IQwSZGVDfu2dHX?wo%5BWilXjrr`UBOVAcy8xaPVG6Zcu{+G;w-?VwgBActioZ
zQ3Rbv*2LVK5hI+4hENZO4<4jXq0CQ4oaV71g9lgAvH^a?oA5h#j)2w*O+<ry0PPPK
zk{O15TuiwdEFI$6n_dj?*W6s66DMf4X#)O%@p%r35x#LI&v34{rpVk_V4OcU?H+*r
znm|lu9$!OUE3XGmBD8%K|MM!%<7H{-im&%!hvmkZK>GRWE=S4%g`U&ar*B_{Qt*tH
zO1CeVkw3Rh2aBE#HchD*>OQ~NT8t|~Qc^)vbM<Du|Iz|H;S0m+EYYq<ixsf}VuovS
zr<2o?>D|CDM0l%&6{#6$qcR`ReX*)=JV4^fpmeAA$JYkK)P-)o7>>t!QSK$|?_HQf
zcx!vbAN%I7SNr)Arz5AuP#!u$?la?p8W0%F`_R(VB>1_oVcy*n9r2E$J51Zl6NrnK
zA)^2SJ|FuBNB|5Nj0z6xYcZ5lASu~j;lH8a{XymV*hf6V=0rk*fMN(bECkzyt$h{<
zUeA1c#1+N|*Ea}vitev7V5hrooxou0xS@40y-+@8?ATvq*eca0NZ@ODQbBXM|EUQs
zX+SB=HkIwIX|N5mTR>^GnUhw=mrO%7S66T`{5N)q&>BpKmG6>&(|WbiHBTE^B08A;
z$LZwKJbUHB5#Hy8;$Jsqjvrfb_3iERc9?Ut`iqVLG<XjTH}aEu|3ui~+jdVzVSU;B
zK|>AdVSa99_ZP38m$`mc3G~#ak6bmp9Y`1nyk$Ys2fUI$y}I)(J3Dr+-p?nSZTRc>
zsn`=DykD|e`9w$0IpJmbHGn*{>eCRLo49f-*AMeWu15CRg@+qZ-pOs~kbAJJ*`S{X
zPzI;&x_|J`3v*B4udB&!G32Bp7DKz|ZF-WQznV@NxFR&l4eT(+mgCIhMHdq&sY!4_
zff49`=7TRZ!||=%_U@fIV}>cl7)(nBN%oc=J9g{<4d#|b=s|i}5+DG)(2K1>>W{$A
zTN*HPOUpoPh1^-3@DQXL0J3h#f5~ImAT+5kOL6(NcXXEbS4*ZRHm&23@P5lJ_x`h7
zWZ!;<lv>=g#_0VZG|FU`KLABE-Zu(U?27|@Xj!xI1V9OIf<B_tMtScefL?`h99J5Z
zc1L6gZ`9?Z(G-S8S^{+cmdrGs<fh<_OcMzX42-t$B<dWNM&CI;t^?pV=$2q}6&xJw
zT{E(8BuSORD9^XP8Vi@z5+L5ye8CXh;>Cjp3$Aq|wm#a>S7dhcx6c&{4HY7yBXoae
zgLcde{@c+Unxem{8M~cuQ~U&#ZWIk+-uEi^ycHy)Vq2hlVY$(vTu(_Me6#$y+bW@B
z5rb^$$odRxk%y|}>(Gc!6yDJmAK?MumHoGAa{^fQkwb==Mg7Wi^N0UT5-<2DTm48X
zAooh9C&fQgaPhglv!267M_k;VMf2xp03MCpP3o8ZW0Wvv9Kg!^o?vn%W(Ew6qwl=V
zWPYyq5e!uqISVk_htu}cx54%STQ9lMo-VqHy+2o6=T~@FhP)3K%vNfjCxK==6;(1>
zVCjx~atq8~E5t1+gg(%@M~BmP+7milSjq>777r+FXFE(T-Hd-ueSK`{%HMwm_y73_
z#{HbwXAW%g<hj143=G(D&rx$*YifA8s4yIG=%pBwDDQ4Qt8b>|R&dKsmSukN$_tN3
z)~bT%{pX9(niIat`$3Ej=y2}j+Un}?+p9d@-E&Ngxj8=!E2afI?qMm_-50gneA9T7
z3v+!(UrSC-P8rVQou2e-HK))G#m%<jRRHrsAs3qChcB76b>WU{M!vB%i*?*|l1EmH
zlG}T~jCyQ#J1|~UDg7&ZEqoQ!q)O7yPa6K~;<7A)>|~?MeP-0czUsKp31r4U+OJi+
zC@LxjL@eZ^CPT$-9sdp0RD)L*R{LYjKiAb+q`K}32${Mq1&qGhMOIdr6d7tM_=g+x
z`}G}jV;IE=Wv1iCxMHI*vgj&qR)6F2i60|f&MTmGG~H3qGSgNg`}$l!z(`=YCTDry
zL`E-QFXrw>bp#;&!`VAvQFD@HJzc-PgI;IwBXY#F?h849-2$Qu-Tu=g<UHUKz|}`S
z-V1+odB#?t_($PN<v{b;K6u|lwq@HL+tZrpWkzf2j^~owQZAm+6T};b3%?<(9XZ77
zL&B|vvL5{38&(PKZt!1}Y(CSyqxWvW7PI;rY)j$Zk+{51-pZF#G@8TDzxv0ol%c%r
z#S3m5Vl9N|GSk|K6IwaQYL6$$%E;h7wsgmHR3Y@+_)e$d6daCrs6VCmk{Qs=%RXT$
z=U6ylZ%hTo^vkq57JfZRGoe^F<6%8B*CEk#rBl4-C1k$r;xiZa99?IIJj8s8FnLm#
zmEriduax^xF|im6n@RZa?YLLH$enK{I^x$$C}yJooicrECCnhN+LHjoI<ansrk*h5
znQ9ToD)9|SxGhW`rIA%i)!ygQB?pKrAZTVaf9Ea`^hVcy2vot!=cb6(P72~{68n-8
zdi!Wl>|`V-ri+o)@)tmD;epj&T?eQ|HFs=WF?xv~U0qxdNU`AM1dR-R@Enr8Z&(9C
z(0Pk%I_pTxCN&9yE+37T&josIei?q|v9ZYiHkGcQ-bf=$ZtmiaZ}4fUOqW1Ab0ax<
zs^eK6SWamy7awHQ!Q^|rMds?iNZw>V%teZeZ|AOEW-}|m6h()qc`Zy@5udvK&EJPU
z<z{8Uyz1_^vM?e&3o?fU^Lnxgnf|%DY{BQfrWNCDJ<}e_)7;DN5DVI;U7+ya0zVVD
zdyYu|gFoxPx#BKp%Ai|hq@_oGuyod}fL2`Cyf1VEq~76UUZd_{Fq9em7$D#-Q=Sbs
z3PNq!B_Wf_@6qOMD6dg3Aca;K<cci*&#fItCW!2Sj;zE#p6=p&(3#UJXw7E1qk1N9
zR;)874krXl@6xQ8^(Dm#6<=m5W+30d?+3WLq>>}BP>exXV7i7=!5@SMCx0qS(o@U!
z$wGK(bfHODu${jGK-maM$yvNo$inEz$kQ+MX(lOv=|dwvvMvS#4NwPKh()V?xV(o>
zmEO4M{6$WvuIXRUwrm2l={g+L{Su^&SKtkP@PP7m)C)%jNy=DlDlUMmMNP1}AYctn
zU{^7@`)w?l^S<!^1Zr`14EGDxM}(ACPt2nHw%sSO6`UHFKty|N$ds^i=LG(pFvUNQ
zDsbRs+H*~t{r!=Ao0S!2n53RdNlAfBW2IebiK#tU&#oCTXwWBaT$tqS3~rH@wl-}N
zsF=BU8j+NtOkz^N;!>G55+65cC^D#_kNI-<ce<2rWsR7C7xAqAXWjF^6h2U|EHMf3
zET$>2($X`3tTbkRH&m(Mut})Fo#H8DlW4m@JqHEO$Zh;42u!n`YI9K)c_O1mG{E)F
zyVhN}kft^Ts(R!+Z0RWsIw!UZJ_1N_P>F<Um*lbw;S<r);cdvsocC|vE=L?fb@HOD
z40ohSsIs^}InT3DkQm5ZsY5~n9@N0rhA^RN`@`qYSGEZq5Vu%D5<f%xK<Y2g_h)zG
zFVxQ1dS)-(#QTSDD_}xbgj%V0xZT3vMNW?+9!>ddm&2YpgeF-PSN)D}BzUxICA|YP
zn!Dj9`v8K@`rgR<+5>*oZNlpPweaVaC;aEWp+6&~@Sh^{e{Q^=|M};yELr&P|JQ%{
z>UHk!O#!b`_C70LBETTOfDYj^hEDu%*!O=~jsG)2@Sotsz74;ud~UcXuq}NsR1zFo
z(<%_tq3Zj&D}zs-KQDz412n3AHCxW8PkToP(<)vAM~Pfe!r?<@SA%JX3mhyr)@gS)
zK4=8s@Tyg-5-R$O^c)lHwSN8jZ-(Iii6O{(TF%fvksL4!-a$f72dH3UTidOzZrvNG
z<ryh8PTO<Jmc8{+LBtuyOqpK2TM_T`n=3n$(P04TxKL%R^WW!38;KN7rY1k-%G*GU
zO1D1MekwR|O|e*Wy9RHOt(7TvlFNS=E#^u|W3c9fsa#2oU%y^k;U;prj}TQ(n@~1D
z6Ly{-^kC&_ACikM6jz6>@!m`*5fi<GB^3H&`u?#d6=XJZ=A`l^NRZe!Q@L@dWq)Ew
zk9wk>)ng9pMe>N38O$H|JE9y!rD*zdU%yT#b&!9I)=ki~Z-2rStxoB(ZtKnsm-vUM
zB#=;bEoWkVLv8H}?lr^!XmTE)Q7Ip^aM<lVU&GUeEHN;+?qc1&Lq$b}rit$)VfxmX
z-~bTWXUbgsvjXECqAi@Zkjs*cnk--b>C-35YV>kt;W`7nyn4uutnc~u$byzY{c|0?
zd2ocIP*m@XKpB|HlbaA%ipZU-W-2Ldr`6{0w!i45qzi8HKf1rmB?!R+^u-Cjo`3nl
zYw(|c{$=~G_DVntx*7-GN@*!Fgvj?kRI~nb3%AV;ZVv2l3oW^f#zEBI*ujdSmYuTw
z*xJMBTtK{p_B_O`2u9?k(c8>lu%L(Tj2PKC($*mZ1}H2Pgc>UotknH8v9(EK?ic?)
zt7qvX@HnAkLC;sBrmjxzHS^HuF75juv5QPssB(e9mLIj({+(sOnHh`A>%1249Kjr#
zya>0~;*H7DWz(yIzmhSD+t*3>)-pVBMArhcR}H#3(*#u%e9iE)LI*b=I=#HT0WljL
zoxQgy<_g1uXw^i5B?w5O&H$P*a*k<LV}=gBVDkos=new6zo$=>lXJ{C*su2VY;3bx
zB(B`R^XC&SF0he!{IS}eDb?!Z5J8ck<~KfDgBLm5)yAW><nNo}_mdX%6_Hcd^%1++
zZc~9GvRbQ^N;c}*AX47j%gi@o;I70Rn!o<4;vijhv35oP6JT?iyV>&I%$_O8<kUJf
z1V$pJQD|D|yr7PLEg-3wblx+IJ>vbQ*F22Zj7Z;q^H_QXQ74zZX#<ADF8k$AGsrf0
zb0dtF1(=Ge!etutV$KPw?3hM!;r`77t*W$pI!>%h0?V=R%X@*~pZqh=-2m;r_}H<l
z)@?a<;>3{lL_mmu7kiHtusvHp30;T@6{F5By)>YI|DyEMh*V3rKGA+cK=7H?hu!kN
zrskg<wAYinhfuAn&Y82yxtLm_<_MFFM(3~pTbRacpSHIF7dz$kkC;XtQ`|v9GCF2y
zDgw}u&7b%r+NHL5p4IMdkBB+Pzm*KcJZ7|6_nBh{a=O$g+NWj*iWJm4Pf_kKa=&LL
zaUBa~3X;yblCxh58&yp*ni;Wg`ivDsRMOH>N6UL3!<bo{2Kl*1Xl}#|B@~&he6&m~
zNY5iSaNhYm{hEutd-qOdlGS1BG-^fMw2zgW_W?^*m@;Lt#KEl&4$n!<KtwkqUL?;V
z1MXsCWytg`@&Q7knV9$md7eS^ULc!OJyD+V3uBq|sP6H-Rn_mU6ri;w7}Q~FPXDI$
z%d6hKA%O=!QoQT(?*!yFaz=a4F1<xhKK1Db=ez4R`el&~cK!JAF6gA$(M};r7vS*i
zg0hx$&%yY+$Z5Nuxm8a-n}Z3NTW{A0QwoDabutkKa%#*Le)@ZtFqn^HtB|b<H08wE
z(J{ks-Gd`G@YcK9S|K^rG4l#m0x-iS!eGNvjtf;jb%7GPE`|(I?l6IH4+lm-6oHWF
zL`zf*Jb6+OBZ8AYa~;$-=L|0r6iRICWK01$$pg$^Qs5$o^A2Bi^*Dze+tRp6`R&v{
zlyib<W{mt8^^K<sM`=Cp)nohXIb>B;pUM4(4z?(#*C8c@gV^|3m&%Ez!oy+z<k42K
zTj=!TFatDzQY2WY$ja)e$?ETz0i@}0-b3c~*7fV3v4MC`6bkCUdfPZC-`d(V9J)WF
z_j^&+l`egXQF2}Rmb<@_&mRtwuJr2Z5|{%NfZm5l_T2%)p%v@A_|y$hy_k^ctBJ~e
z2cJ3>60-5OG+JgD)a}XiS>&Hlj*u%q6mGk18xranyk#&~@(wR*Zlp`2E4DXs1A0fx
zC=P~DZU>{UdCA6^!*lQ=Ka;Tw$|ywp_+`#~=APP=qrhrSf3*{#HBnB~@>s9alZA~g
z(&$C!k3&Sw7>;laH;m}r3?CmK-T}44i|Xi$jyD)iML9NP&>*2@>Bcp#Xre!;z2CB9
zOp-!Y<sT4$x_8ZW+ImA#N<Oz$F6;~k?8T3s(O2r_pWnW1ezr^-1b2~V7R3s#m?^+%
z0ne~Y{Ns;5FxxXK_X`YU0O#G^m3trCE`OrQwde02(X)%bH5t8Xl%voef(uS(?6xa5
zsh#J0e(0s08832&8c$-&<X5L1{l27qpGHGCZ2^7;My6;YTLGR0y|-h+DXt~tkwT}3
zu0}IHBWwVYgd(b|uuj2`YQ~+wa8QM#F1Qb_&<Z*RG>>8BBm9J4)YH8N?)$#B_A~b+
zl}m#IwgaKS=Whb};n<q%bjgY23y|8rC=wW8D$u|L08qgYidu+!+qEXZyuWb?b>;4l
zFNV>oa^_6k=g*al90d7COC7#-r#_n5dQL8L(=f?o=;3C}y~!d3UbwT92^m<0-kh5(
z@Ln7wDy1d*`mu;#2{&|e=6MA&It@9zGuk%-83umBl&E|6zIS#C-%#{ZY*uzQ<koxY
zMi4_4RaJIWJ(JcEH2*ee&XcE4yOCOR9tEfWy26QWZr@o!;d<n43;vFd{!?x+STtfP
z7x}2tJ9MB8(Q2O?GC5->?^!s&d@#?;tbjA8Prn1%19@Sjm$r5~xferEt@n6yl&Ku`
zbae2M0bl1Lz62DD7*(xvR2*FPHx4lul?Rg$I9jM7s7;9uLPwLfwzkuR`U%_tl(v^y
zG~@k15REv=fn;cup9>5`m)lijs2Dt32OyVjSEIU7h@!hFXLTh*)i&&8BuU|>20Fb2
zAOv`zK^^Q!@)7sDJy=%HK`{?rTG8?ba5Q=)oR06@vBMH4=QV3CbU!BmEryFLn7GkU
zck$v}idJHXxRlhFhK7bup8%v*-cGxHgUiwY<uw;Ejg4WaP8oxTv!{eHD%c+ExM)8T
zvs``%J>(`#@E{ffkgZuGfGc?)kHU+xR|PkGQZ8sGDsnSQZ}txjo%ppU&%AY5Y_gyE
z)M?WMv3EKb2^iP%<;!728%h%-X|sTFXnuX|*h99NZ{@#XVPSV&K5f)LYiVi8a$=;L
zY`5gQ7df*#ZWWO8Obi29w0CfzJ?0KC8yajUoi0=EY`+o|8hVdcL&Xk;prNRk&k#DY
zRWcX44V04(gSZhaX@dR8jR7V>x=i*<WD+hTmds2`j1$Xg^Wa7c|9p?1M%QvHkVnlu
z^5JdvK9yYIWS;XiOuo9l=PK4|69v@l%{%Ga7pZ)_spOb3n?)mfkj~LA7nQPH#AwRp
zl!uc=?%zK;Y@+Dvc%dif)ul92VCTDx<2dTQ>r3UFUpWt~P+>i$7dr&bdCO;91z0Jw
zr;es(z_Cvh)2AfrX~%zIO%}HD^(I!OlYB*YD}gjcD|XOKC%*MH=1`vZS}KHjw7pak
zQ@@3{cI_-HD+2^dr0F7N;uRDpq%Hiac}$wVLY_@N{p`}+bgyQ<iD3syvdjn`Tn)Jq
zDo~adwz(sdXl!g`?v3#v?oCzKI3h*W=g-p;b8ztE*fxABe3FyHe1`pX(N@F=?rU1S
zf<42JWpVA+tj)nmvG_!1ZRoaziHYyn#sN|>Vj{6WlDXBfI^t|Icv-JzC2WWn888R?
zib+2IqsHqE@3N%cnFj!2{fDNh6nn~57A4eOf`C*=yLZ4S!UYtiIzu&W&2;2u&H4^l
zD056lS6BWfz1;I{gh7D1nVP#lZ`Pk!lh0K@htx!RR|6cHVfO|0<d(g8wL|l~pTQ_s
zV`?x5vsu3J(*4Ij0l7PW<k(3oUftVlU;kQ`My{whRwDKt`)AN@kCs9`_>99&{e>*4
z$t#}@VezrUhYnR@VDZ_dZR{Ifv0U1%Tik2}VfjJih$<hCndRa&y81f%?C{H|EAtu#
zGt{u=(~v*^H0m+EP)Bf(-zL);e4HJp=<3+Hj+9dK>}MY9OKRJ>b-$69zGWB3-Fhb+
z=+xFggMkfi-zEW=2S|7~n)$E9sF*k#E>n_VjoWE*ZSitbDnQ@z0w!4c3!L^p<Nj`$
z-$T=puHMwuX)A41X3t*PwGv|3B-nFdB3%ZQh)=V#jbmk%lnUzd-7pjHTtx(khnh&S
z8a;Zne~%sCh^S9SZkq2v*xF9BxWC!R1UP$tJIJ2~ikNMUvgh5yExJj<G~_ol`KR=c
zFwArnz&+<fhu^g=a0lAgF-g*wSG_6_*tX>&bLRt@MJ=j7QK(=Vb3t*$xN!jX>X~2q
z`3VLEfBb#e>?}SI#Bg~tga{TbtzeJ_^3nMrW{dnKCFba=ekpDH_H9Mt>qSU+>QoM#
zA~IzS(eN$=_6yg#ym(H{Hd>NrPKDE`=U_)!958lUJh)*Dp02KbwDT-*vZ`TNvVlH6
zO7C8;nR*RVqKG6yLTJjkHA{W)QKR79;=E?w`WD`Rw~TY3)AsEGPrT5KCNmSyEZ#WV
zlq7~>jq5I4rWPrniA>en0$i3U=&kQ>z9W1o+$_z#`nD0(*RL4?Vik<gCdXoD<0T?=
zf-}>i{pqYPmsj3V(bOz|7wc<30$2HK>y78!U>`}E!I^}I$E`e@ysm`{d(nGD5sHpJ
zo7xXoefXfDNUAtHc3)?2r3a&*%NeL9g%&s@>K<utZ=cPO29L(eNlDGbTAe{o80?Hz
z4c0bjn?|#__*RXN62&81Lh2IJyW_;N#c`!HcHeh6-^A7RZfEiwEIk80m~ly)4Y6s3
z<Oc9v7$Ev4KKd3=tW%%R^wv`AaGT5QY7?mtEWJH7owBywRZ00Vah;0pbGgi~y?qOz
z_7Gn<G%XZ7%}^B@Z)y6VOhX^Vi?F#}lZm)yd*x}jn$Tq!(gFg0M%Cux!O7igL!~0<
zxDm$C59B%;IPl}MOZl8wTmqPtc`7i_zNwf*>dKE|@DH7)+t|cyHg%$3`c(1e;mVpu
zaY|^<E-2AWOK22AE?f&f5kydujZ2Y{3w3q5O`14l6wajAZ0ytdej6rDp1gsPKh!Do
za?mFLiyh021Yb*(gDp=?Y~T_=4mM*~!|6i#y-}~=e>r>!J!RS1lhN;C`AL3HU^T>8
z79fLcgJk7QXlN)Ms5T`K*8&E>od@ob_N{yqS0{`Oru#yh5}PJYQAj{3L+Qhw=dDiH
z+eS5sgcTp&zkg2>M6M@*;4fa#H`RjxfWoz4nS;<m%d^3+wPsb}71Y^r2&Lh4`g@34
zY-XbK7Or9#u}&4z5D=D5>W7h=78w{oT=;+PcG<OS?2PSez|H5(o{i**isvR~r(NoP
z<e!9;Rzg(MP$f!PiCLRFR=r@!fN#{x1*PZTPwz(H-SVBHLt<<2j2+c)9vm4^ag)^j
zkKVoQ92`O|wG<RSoZVf?^-GvN9(;sClvLtnOj*-Lt}>UJoRJ;oU4R=x^XYNd70wVm
zWMhzKs$W{CiHFa+mlw9Wjw*g=rW|CN;LA_#sq^^Uj2-jSjJ+(Yh1NrY46hQ~v6DeT
z=A>i48fP%e3GKO3_x`<u+96j`9_wwXcC~Np3VA7!=c%7o73_`w`|rPDK$l*^XqZ9-
z*$}So?!z<+*C7M&J$}5@QH?zy&h!*Mn5J27FK)_&beYp3A?5s3p+<lAKI^(LrYgO=
zRk6!K1ls~iKWIV}lm#Uv>Z+>EY#tKE3b<P|D$UK!&x(qQH%Y?Yb)UG%ACP%BH8sJZ
z!qsKa?AaPWjKOGWco!QR8+-rjY+~Z)>HCnb6b9VhBm*LyhPWV^AtNed?fFv7@LPJx
zjR4D*c#&YO(fZAzI}VI5YrBaItBJG~kp`LzZz}DhK0ZvBIon-k&CNk<5N^E0cNnl5
zE?QL3v)S|ggq^&Djxf_ho^Bcq3z#uH0!nA-P@~)BfSk0j(et;qd%Wz@Fh^ueM3%gC
z*RDc6seiPZ7jZtz`SF<}6Ijhz_*pkww{st`{e-sgX%9Eg&(ym%M7<}^0Z$R{mkMTa
zY5+^PcE>FC>4tk1-x9m>&h_+kF9vSYMhCm(jMRcFm}s<ncN<T>i!!16Uz3OV`Covm
z)Lu0Kl{PrvHEDhKyIj8&$G`Rx>6267=-kzo3q6;0q=D<vSi*u$x4Wm?qcyHY+=xs-
zan)U0QyD_-cH3vvz%{L+*^VuoD<ZO#mb=AW*;M-5j@XfCMA@WY{PDxRUU;FcI_pEw
z?(xpw(g2vXTYT`>XWDOYkMTe8xPvM;a>!Q<GKM8zA}=7Iq~HfqUCAd(Jhi#IF0+5@
zx4PBIoUNS4iGq`J$PoO!y+7xF!k^IeCE|%1jR72UVSteL)sgRWlz!YR8S+QcN%yB=
z1d`y{20PR`Io*Jln{MpjFdyz4qAk(I^>FCi-%il7P5npeMZA3ZGSu=t$J<w)@#89C
za~`;3WlVj)d!9{Hb<~WEYe<dZAe%Cp3hQJJ9X{+Gz0_o7g)qEVm{=jF;Pv^=dl0<n
z7t`->q9n)St2XG^xhYAOo7C>Usd`2+bz!Kw;II+*hB8Rk?U=1v=XtO0y^POzAl`ve
zP4UaVfesxfAE|%nQNk>T(7w+35ijt8^#b&7BPH!S99zg5JO)I4yv<&2`X(&+j!u<R
zz}Nb|LLxo#$-DROk)GV*cRFW3IodL%ewLAjhDOKjVoEc%fWEAswy!IpY#aNJZKgBA
zgSjj=5;InFrZbf@blT(2GAPsa9x?`_6myyB({BnDxe5}y9gDBLQEF^&-;?PSybWIo
z-JUAO=Y<8&71fKtb1Fmbf>6V}G43tltHp4|y=$G+kaK(;pQRTO`g%Xgzb6V%p>A(}
zD`Ve>wsoF9$k$MqZlb1t%3bZQCL}Sif&7X9tUaY#s=j<l$D^?D&``9MR7h;RCt*iq
z^6S(kd}p==TOKd0n;ZKp1rqlzWXvOfK_~WKuA``RctERJ22gy}jrhh#MjmP<@=qFa
z>ewXRpha)(y3uyTn%fnjDT_KxSkzCwS{<YoF{@@|`tJHc;wlT}WzK7x=D(X=%>5j&
z!btGP&i)2^Jc%fNEsi&4DNh<MyLUezjbU%3P=f+i2r4J}y-aC}NY#;yaIm$_dR)bD
z(V}<z!^OJzR=TSjY+UQ3Qc`Bu&uw1cLP3#Zh=l?-7w5^Ds7&)AP7;ypUNlr!Q(-%r
zUR>vMg_Em6^R2K+i2FqAv~_3s#>#O<YJWhj+jG8OVkp4D&|<VnX7C|0oRy71@xqza
zz}c-b1EzR#pyEOyA46L)viTt(T+V<{)gO0wxLR8?LD3+^{GK_6=ham$6m$XJ{hvnd
z*;O#<!LhY`C$)0xNy>N6y?O0d9i_NHxA{KtS@)WSy}dnzLDV{L|H;<lA`BxJUS@AK
zm<2BgnST+&qd^)kU{P6jCHFv{UDm2*f~KaGYXt61>Fa`pymxf`z)K<~i#tien;Bg4
zg>l_Q)G<0sc-m8HFIIf@w=H~B*L3LkhvL}vPxpH(cT$(t?0zn8vqwo=SKF!9CnZuh
zg680JN^3JH%n+~5qF@I}ksqaMwM1{3UBE{YTE476|C4~ONcsH{vcfvW>4I)5gjwE?
zL2qg7g$}7krE9eEatY{*ZN^Hc6cFB~QJozf|KyBT)1@zzHUjUS)zB*eqo-f`eb<g1
zWw?f9ABmok)6LnxUe}>AL`a*^B+i{|a&uh^8|#?&9FI=ZkPj=>)+etuJMY;UYw&X5
z0n@#=jW(`t;ZY0+eFttEV0@s^(K%q)y~86$pBeGpdr|WPM>UC#+q)vfy6^>&H`9MG
zJQ<Bq!V_X<U7PEkJu!(-g8cm_rpg^&{`-fKTYNk_n(R8f&gaao8gA@!<yk$m`ZnCC
zP|{RZ)g7%P(zLma*y#QGQ_bXka=N>noz*06$eqXUv5?vIY!O2{K%$YOL&u`zd1Lqc
z*`?GgvgvmO`L4&j;^7ld?B{NT$#N{M%E%OGl7w58e$S)DPMw(y+~F~AHsx^lkG&oH
z124a~*V>d>hZktkorS;qmWW5s7CZvWttHkudwF>U_#N)4_wKTa^({oAcrh{0QD=Fx
zxR&9Ul;-0G*@U4F$AZOq{-AI|{*}schBnF6!-j54u7unJTZz5;<I3%D-@*m+=lgX3
zJJMYA<cQ2VGCmcFYE(48*ZV&GA-3wPrZW{BHZTK^+1c4mzVf``<HsrKhZc!ONapqS
zLPwd|de`^L@j8cW)oOeN)0Z^kk{lE?#V>j76tWAgj(x5;%mmm(4^_U)*4nvK#r2c3
z=Gp4Yj`lbJg;kH8svEg>jvpnTW#{GK{bFO~%oZn}{!L`xN~CfFWowj7@CbqYvFvFE
zoh^<p+L(;5rpaU_<kA~vUw$K3Yw~n^|Kmc-goQQKo9QTL4=}^XOIhzmTnPu)E!w_0
z^Wl=}PoJjN4ECn^KcQ#drDx)Kx*?1ys4E!Xm5+jY#p+e1PoFOBH1y9e>w4vsum7yK
za3G_`IU#?^(TGco*i;1Q<{Oyx9eSa^X77YIb)NbFi0_skWeg9VUHzIe4v$_`?9o|&
zN!QS4u!P#8ai<*Y?CNDjOON^ZsI7crECd!Igi2HiBVut45(~(E^5i8Nj^L?Nv|hV6
z`@YecHHWZF{iQHCT%(90n+(VeB9?hsPe@k;3WQu%p^qHU8zPd<;8TOpo<we+%4xzZ
zHa_L6u4vDhkv$*Jiyy8(&@rG*TxE8S-{_e3McZzAlif_TyVpBJth^yoOHE!rr)k5z
z8gsQl`o~t0j@#MVigr#R7ku7xp>qGmDQ+c^ZB7Ey*f=1YZs<sjwGR|Ny8>IC>@;$!
ztSQcZy?DyUxk3bukd-HEY4p@{ij0oVF`vU^9^~Q`PWf{jZf90j3Vt#{>BjsS&V={F
zoO)wVvg=>aO!JisUN{aLmN{-aPgqnnBpda9DAenX9n|?y!?icAwj;G}0<Hmp(S1pd
zvz|Tge?%~5Aj;A^FYsd|(FCfr_UZxk@dr-<zvF$7>lh*ZC7;hA=Qq`lZ++*LpYPJQ
z|J-3{U@j<GK~toC4#zB~yRZLwgVlT9^p3q6PA;p#2-KdWH$%UeTp_g*%HW<KZ;nVc
z=;e&QJH1usF1bUbQ?Pr68H0}Ibxugw<92zW#L{W_mdb9xRWu#~LAlFn3qBxWH5ri|
zIon>{J{26ip}1;rkKK&)EpOy1uPI2moBl;LATi73@|!@24Y8*kn+9hpHVS(>OwUaD
zEdH5AT!2f?WmQ(%q*Y-OWtLJ>(O7S?`3)q_$k87@Fv`HaDU4kkvf1VuzTeS>3Q@Nh
zG1gl08vy8RW?}-K;Pg_Dp9o{UY^ko00Fs1XMW_zXvns^%Un3ivb8qB~Q(wJWS!juF
z3-k4RV3Yu+bZ5`|TK$Y7vQ?qqF}ZQlRwk`wYi6udW0cKMb+K_#D!vCs2D@tnPTQ;H
z+9D?-6WwlYsupRZNd6Bg=Xvp|JA87>*RRHt*R+p+=%-~Gr<eMBiN&`#{Q;X;;Tv?&
zHMF95&O|2-Gws#A;l^`Bn+*%{&X-#+?6zOGwr^aaBgT)LO|v<31Ds<gs;HQZQ)b#>
z63^9?fv~tqxm9AGl<eqn<EFiqJCUK@({o2~p?~d!S^TImV=nI*)lt=nQPGnln=Gm+
z42fXz?oqFWW&yv8uJokAc%ppp1TUxl)*^d`Rl9%xKCWoo>`bbp6JYVz508Mie7nue
zG>sBg%-Ykq#$#?eNpt41Vk$@B@^d?B;8B-UvHv%t$P}w#A|ef<vJpy?JW5Y{#Z3>N
zG2&W^k7rXL@&(e9)<4ouGa?7Cl4ptHX95FD5f@N?q#+SNV!DwGuX<{(ymw<i2Cp?h
znvmoSXW&*yNZk1d%W|Pq@0r(iZ-2IVUS?(#^*k6Y7Y89;chE`Ca~Q9wUAkF3FhYvS
zlyU*f@GR&s6%DD@aB`9<HEp93;(>CvhZR#<==q3&uS=_LBEB4UkEqKrV0D<*STiva
z5=r{d<O!gC-WPqU8=?;fsfn78v<CmKwta<6!LwHtwP3>*1ugP*h%7+`>HZv7%*cPH
zh%%4iEwIh-K{JU)Y;oMsmhrfQ`hNM4-8b0#uwgm~B?8z$n?oqG4<0;7Ei5^xa0wGp
z&~Cpm=%*|Cl8Hmp<a-7xjag%TJLQ2=?Bu%dF8#JUIjMGdZE1J$V1jw!Mz^TElj9dq
z=SMHg7c_IgK_qIzxL;l>0)i<8k413Y>4=)O;BfsJSE-Owr-UI6LWu;Wkqu%D<@;up
zyL(Fa5l|1lkwk^}q-(&IodqO)20Kzxeh=^r>^@ABYVop-<M&E!+}qa*RD_ae2}YLX
zycca+Ymf>*%#{$GRY3Uh?w;iC;Xz<pd!ulBvg9zFSZAo*x9uwd9~BhjVEUJ$;xgy_
z*8qOfuLU1WUSRNM+a2vxXyjHh=_?93vT#MQiJ`#6D_dXw1=@(ToEi6fB0fPAk${e&
z;=iF&;~60+O<|k@J;!Rf?_oU853O__J8N|H^hsY9oc}af^k+xv+_hp)H(y!Q`D%*q
ztWnDEwbngooWSHeQG0Q3$T@UbsbFSZi43#@;wGKHq8ky*W}G{B?)L55!T(0;eUzB&
zC+kiy##&T4Y}v{~uJ3GX<D*X$`hLElK1N9)q}4jgEOT^(!j-64m9*@|@EB6KK?%KC
zTlPYfa9)M5>@+w(JNwj?En+5jEYm3Un#3x$Z?r$fHlMuJe9Xa3Vx@8`Jed>g?{K<q
z^gOekv{hrSCnX&jD1MznrQ=FsT4zq9*>c;bje@ZaI7vfdjJP1>6`CSfpOx#~jv;{W
z-)GEi_SPgnxA{)M_A%fTSV-%dr+j!F)^`5-6UT=4^NZK)F`fL;wSW2kz^8jRh{DDQ
zBNo6*N>}UWJ$+h9a=|ttF#|nKOxy;$#L@-@1!c$OewnLYt<?fz!a{I~`vxz#$Y0V?
zzAK$*1<G5zj~a7yoz}VPaWfUZESSn)m&oIALDg>FGb@8-D0awH#*%7%hi>a&)3m~{
z!NtVBugA}4?7AS>d2Z=jYvg`Lx(e6;jw$8G)+by)TV7oK*S2Tf)1C}K*Gmkv>-iwq
zYv%q_n3|e8NY&X_m+6AED2>_cGY;BQQZjz$fy*gFJZ{c167d##J=fUPzS7%l5xrfb
z*7}gIZypr`z?t;<(M=o!4CU&Wy>shU8x2gWUro)<&qsHj%T1q~Tc94D)-|fmOw~1e
z@$n$@wM}t8(xMT8926^?=JcMfFLa`cZ4E}K9enXz;q2+d1A^v$HM}=-$U^7e)3W!U
z57!M*q&8MpFUc_IFOSxP6WTPe0&grJ0U1^2W_?vfNfc7L2Nvl3ZaPwLhA_x5%~-3p
z;$-#fec1=EEmjR%`Uby;-g8!xso4xN2@Cs|T4}IpL?MPf0^G?-R=v9qoOK8KFf<ab
zZf-{oAC@2dbgC8j6!8oCUfD}?Sb$cspLw2NTlTt7bJY#D$><dAM=`L2-lN0TC=E;W
z21%NHJa_KeVAJ)!3bqgmfV*RuNHTi?5I{SW{M7R@iwMO#!L3ei)42QNuZFq4fo_GA
zVJ0cRnHOlalzpDtOKu%KZ})R@JC-`mkTC!H%|g{uW!7Fny239QTjl^Y3NLZ#*}F`L
zEcW9@pld0N#w)>Z`}tw)ET~pZJ6i+HZN9d(dA&I_nqCDOrv&fRYkSa^)%t9ZFNn8q
z@e}X9OQsA+5!^hq#b?wX3RX>r)YxvM^=yl&m+#PCNGhzpa=^5S6W2qD){gn`@#9K{
z`gMrK;0;ySaM|1Gdx+}E6>4Iy-u?A;f_Rkmt6?WjfvUn4Ezq(h*=_<q7`?>6dPcbA
z>i@ysdxvxV|No=frIJV_(y(V_C97--g+xXvMUq)I4J9ktDSNLHQnnH$L{=yxD<fGU
z+qpgZe7@(Ob6w~BbFS;0bN!C%bA3Law>MtT=VRQ*?S8x8?tZ1gHS~{~g3n4HE;zl+
z7~1<psZwU~>VqEMO#s_ChkC&859<1e2QvM%s3ABbU5}wlF#e$AKufd_;1D(mOlwJi
z)582BGBOxd@R?v}U`ZdPt)NqfA}4B{i49+1J%Ba1K;jFsOFAflqnv5J3khh^U~R;u
zoD*tBB1ofPSUvdA5|@^r{7U(b^Y;_h4h~|4+Q!;)iHM*-V<q?2W#*ur;|NPpujAFw
zJM;FjKT*Hm1BEo;QUgWkDLfSX%aA_AgymZE9i)C_u(Y)7@$9T-E3v1xnpzY8-BZHd
zJ=cza_*7Wnl~hj5v}*7&e}7SW>Jyv)IUBxMT!L)`*?f7C(Vrbw6vP(zQBZao#(_MT
z!<b-AjhRwV#R~knBMnq=fD$^pc}1RD>Q(e-BsUEw?!2pLymNcmCy2buNo#lO?>dr0
zqImd$FkK+6S!Co44trx5xP(ugj#&ePhm%H;AYQ!e6au*3lLM6trW)B(sIo&1>g|%M
zyzT~{_aLhv-n#`RV8U<w=E^|0akK6;-}-s)X}>c>nw5Jus-;01f|=r(1n=VhtjTm0
zN=6K(y4W1_iOi&#UcY`F?5(3UjA|%LA6m|IjGbrJ#P)(!h+61Kg7o$j$DmEO8s#`Y
zGpFL@x$imYuZR|07CJhjj)Vws=uj?vJd7q-3v^AHn!q)J3v+Sq$t>mgcKvDJ{3n+`
zWuS--Ir?-K=g82|Bu8DJysE`7@r9>japy>51G;u!vGMR&K#l7sQ%0MVVS%lJwvmnQ
zCx=*sl3bb$V8Z>5w+P&VMl(p*LjRzw;T9^uNUw^)D^ze-td!Pv*Gnb1om0#<?;G#9
z%vhd=hjSPY&Hd-<LM-BT=<*I1hkXn>PJg$$uDwXRd3^}q?J92HoXzXkRwntR{_t6=
z|J>(W<?nCb3Jrhxe4WX=d)}g@w8dvZ-_>^){Vr<iHI;^I^g>RZUFXU6*mR7g0F8i0
z?AU{;pX`sRH~w^_-l@*&O3=|G-A;9OB-rf#<>mRlCqQ~3b#!p9{b|G126lrJ5+=Y=
zC&zC$#Ulho?SeETcdyL8QNDC}%MABl<WWj>vu;1`YzmV1glK{43WFjVrT05|;oTJ2
zXDTEu00;Xn8N?;Y{LRBIf6!<#z!;LyZDbC`lK@Z;E4iX4qODaPdo2Gl-f~RinJAxo
z8~6xrh5HnYjEr3S@Jo->-l2a+8d)w}5Szg>cVmoLBj4(k3B%;LD_|_8T^&N=y8vzT
z2-SMn-Vb6RW(ogvodEd8F3IPHHjYXEhVT)8<(yOt&W5B}fFYp3hFlI_XL#kxPquIa
zn)#au57Mm8Yid$2(?nVG{cPSLn`RLIk_NrfK-`-8-4157H)rboNlkd<Qy^kXU+p&8
zt>h3F5A8OaTEUMUPJ>avLBLPY)$1cTu~UrZ0Qlr*X63Mkt5e}+>rIE^G|k{|(zfM?
z*X#E)NC3vVmZz$LdwEG8KH#(thhQWaQ-a3A$#b{Zde+#Z`gz=HGi+64?|z*m#FCRL
z3(z$~eL?H!1IX@O2R3^=Tr|xV@SP+!vG2$<YyctX(p7%}6g8~f73Kj_WP=v<K`Jn8
zo`j(Tz0yI=kpjdM^|`ydb61RBdFtN%jS`rq$B&PyepIqg5x@Qh`lH&|gMKRC;`&Fc
zth7>gmr*`b`4Ay$sb8R5<WpPoT_)wDk$r<gp!Ja+FHi2%x1#EWwzNL<dszEOpcNIk
zAc2ZKC0+Klxg@Ko3XP2htH?G*c1#RK5)kIR%}=_C%-dpg(m`i$@8&4<I8XYML(r#%
z8gBd(k;^qq2;3Hh6V=c4AaxIybl~CU2CNzm*w6+l9qh#m1IDWht#efu1j?M9K#|Z8
z@9~_|$_45)iG!&-bZ8O*dEL>$NLqFAi(oaX6rPF{7hE}WB~E^aK6bs{W983Wbi>6P
z5ui{Z?~yJYRN;sUNer5k`3XR=qTX&buwbn4*Fk|%1bpyYT3Q_w;A=cuikkk_D$wJ5
zb=tnFe^abjm*b)Nv3QQxx<eGx)jV0O1vk-i6SylMHLFRBL-J!MrWV9Nl0?<2KXi0(
zof%3YRs{oLRauKY3`AIsQDm)*bU0djS6cJLa4@=E$%&aUUJdQt_xIl1A9Ez}fxWt0
z2db;8@=?y^&b|nUxUaqgZA6b&`n(;n9MW???#<1j7IfC6l?y-jkYyD?@58dgH~rAX
zyS&g^MmpE=M9dx*i0_ux2rpPvszSuCl3XN?Gms!0!e~9<(wFZmJ`=YwK_I$1_fdxZ
zXp<?P=D5g57CC|ZA9v>teIqISvS@xSxHrYHt4}_luE-JLzt4l>Gm5MEX2&M_*cCVJ
zUmgRCn)Od8R`5ktz06{Jy~zzrlxr^!kFdsEz~nHkd5Yb7XHCisf4{^dE;`dU1XwP8
zeDDzR$}VFC9F=zAyWReW3m}N{njPwPn>sP+SO(ZuPRx#Dqc5KXNPt1c1WHwVJq&>J
zX!gf59mDR_EOxrTw|%TmqJYhMv}y*T$%|lh33G7r0*T{_w)wDUi^i^e%!J~}?rzSr
z>Rx%(r0#28i-;4dod{|7<u+fxc(Dhmsa4@{GO{Tmi`zpZY>I0tctpK?L#fs@ng}qD
z@1Xl+69}WVPen3pIVynxxgoIwSmJ@E{Y$&`;T#5Twi-IT+d$9SaoG9J-G^8<N%ypv
zlxXJ>4UYw_{;es<gAG7kp5)oV<UAC|5$$Labf$0h3D4h>x;sg5CS@pN93r$P(Urv%
zM-aQF@PtOQc2Ly5yJ83TzGwdEr!{5>RR)YyC7eOpX1=|GaXlOn4f7apq+~SPw<9(#
zwP*r4H{1N6Gwd~udn?^<B`1pQ3e0lBOtNxuUx-<6t~dJj?CS6Gtpa}Qf4%8)wqT_H
z?%e>ybdEZgG9Zp4()z?e@2IEJTk=G$26m_VC$!I;nZiPFEV1lC-*kVmms|f+ix=Kj
zOfv4S<(C%F;u`HX+YYOPPKaS|j_A!#eW|Iu@H_+2s!u9lC*k<$R~I5M_a1CB%ID#9
z*|0y)ZWN{=<RJMIOi6tG3r>FimDaMA7CeimuGbh~^`$gz1Qmpgdx?=8j<0%G6+F^b
zzn-^ZTMAZo)O_@CRc)*x;bWJ9<yvvg`)g7~<FjNLETNgO=-k>6%u5VI?2SOS%`Zx3
zzxA(Etn+|1cAAQ-?r{F~d`ppn7d_enRNDr9cwRi5b^>}!Gdus=%42UyQS72)HCIZ=
zP4V7`hl0A_*_QmNJ4W;x0DlBY1MTv2Wq9NfmLT&REJLpOv4g&v0gNKtKa+>ja_~Z?
z<)621-HP7Zjv+5IC_1Q&Y`K4e)Z#(4?)Bp4=9Yl7In;Rr24YSciLP%Z>*I)s`2dfl
zL-QrX1R@d#ZI!l(M8%g@Ki;P|dgA(6n>F$Q$OsSm&$AhD^*-2pL%FS>!$SAx;NVBA
z8J5*amgCEyByOZ)YD0Dzd-KjeTe${zd~f$3`m#BN@~5;L-y!MtANvE!+=f$*St#wm
zih*&{2I?8G0XBo3;yoqm;^z+!c@!HUus!ufPWZyx+m~wo0Dk_kjc}Uh;-LSF9@TMF
zp4g3rX7c+UZB764`uh#Fe3e_bW|7I^2GYJtoI)@3)BdsXw(LXFu`!qnJ+QCVhq6!$
zXT5WWivTaLUtJL%m)JDc<yD+yrBy)=*PXsYNWk>o000Z;ySOtUHg-Sz1xB1R*Eb-r
zR9&15QsY-~4=^dslbEca1>J<X+Y&1kc<u;-HAIn<ZYz@k9`YZyNXU|%ao&9k-jnv0
zue6Z|DC3{TPY`4O0;MUAaoUo*6~KsD>@50Ek3gMv%Eky6v`XBxl_Fqw#4-TOWgs4+
zJ)1$B_K7%x>CEY|&a*3ei_7C4EPJkk8AhTZ_f6rLIRN0f@(uCDf#La{$I?^hbZ~1^
z<JpQo{6uo!>k?M9*1g;z1u(W5l8o{d!uJNr5sP*A#h-}4KtZG}bAl+2(Kh9x+`eVJ
z(|BbneC{HcL0wO7pjXrAOVl+mA7Obw@9BVmisw!C;?^x?e@9@L3pA7-$67rNkesTr
z-4VNmXK7f&gYk$q@I9z4JP8kP1;`VtW)~iT%&+_)IE;_26`jCjM@)i*^6r>|x_TX;
znBb1(xeeX!e~Y?x7T#q(_6#rKJAxa%thJ6BG99UkRCNp`pcb3T7&!LaZ-|Ki<-!SR
zlTteR5x#<QA7-73JV>b!(a|^nCWta?^nV4-2h$KFcTvKI^V*tNeR1<MqVCK7xQ_>V
ztL|dKoQpq5qxi2EMK>UMfxNZh4HSDz!F_ModW_tB!9&_QenticOeH_2kWuVoUJr<P
z5JU=r^QFZg*n?)k^(#zLQNZOZ+&6ku^WmY9PWzQ!HiJ_GSu*hDk5(fWzJ>P_IQCLl
zX~Z5w()cI<4|AWb+_h%22P^4~{r0dfjEfG@UcZmt=L(NDcf=t<RyuJE%jgyIF)BG=
zR4$9A?pN_CYT?L*KLfyEk+~51aj3_7F3V%tf{mLum&Ihd#2{u6x50PM;fsFJJ_v0w
zyO1}v4hfx-vCMu2#09?df4z5>g;)}{M9M}T&Jsaqai$dn9r5&!a3wz=Wk>3%rLWI8
zoC~LhsJv3XYcSGIeacFX#qq}-<w0b0N?ykR7lRSYf4+dn`CY0f`6G`anc#fy&1s73
zvp%+juV`n~gPRKIx(q@S_zD;7sb>0vHQ1T1Ve1>-b--!SFW2V-5+j}gJA~f&hbODH
z`!sYT(ybDwH*&oA#n;0JjnHU4bAq>99T4#FUiBb6>uIqeoG>4E?iy^+KDTOZWJD&V
z`=Q#>vR$l0bCy|WI&y$jaG^`a0<(e}GLk?|p8`tu$SF;Lj}N)(Fz8)UNMynPVwaRx
zl$qHQu#0Y{f>y(G#UpJgAZ1^I*ZUq+bt@bmZFY7bfi$b{H;&V{3NIIP2M}H(jgySQ
z=|v939Sg|^=sxaqxxRFJQqbAwZkKVW6G-F1yrm(U$c;%=i1iXCvhb~J#o4*4i_?v+
zWNjW3``g$0fA}?kE&Q#9zlb;IZ;oCDAjwA&*3)<l)^@{Mxo`~WfpLTakWz0S5C)t$
zz8q?&FK$a{A&dV@cxaGIsJi|X)km56l1ID$yh{Uys>Ea?k4iOxPs{d^L4$pVZm0V+
zyziMneTrV|B#b`9hCoZLuQH#5HE_pGKiDnI8qlx``^5x$ZQyp;SZNsFCj&}oR>y#t
zn8cr8l?q*P9zEd2)QG4DJPM)@0E;Tf;yEklw~xvTK={I!)vS)ek`w<9M5cMLX1VN9
zCMslx#D=TqTr36(0z*f?Md>@epmV$2PGMaMW$fgOqOn5Hof82<ibx{flhR**(hYRo
zpU{1E3~2|va<pr$@$!hS;Q7#F0|k^xQi?mg-s6D!8%Ge50FE2v&@aHD^tC4BkECZ*
zFAL?pe0lh1wGV<HCj2WXC>$E>cm-zZLQv3gLFbd}ognUo4gm!!S==U%#H;qQOeL!B
zi1LvA1W1geKJgnT5tvMc{L-Mu%2oS8?7+-u`w>W#<Xq6_+*Js-OeC0cV(yttBf=j_
zkR`nvf5)UCY*1y$`L{OOOmI^OfX8rVX8g@drK8X%uLCsNtCX6~P=d#^_2k1qoqL!(
zw^NB}r#TbwH!Te*x_J>5KO?sKZ?H|hWmXorKkx^jrs1D=7i?6CSf)+DfjqiG{Stjz
z{pWQ!6+xDB3Q9N`<d6a&49G6kx5Gvd2k;l`Vi4B<XC@p0q~JPgl#~mhh2rGg^X~y7
zimM1X&fiEAD4sYm(t0lqm8ODm9>Oho^86cS{uFp+L$F1YG;F2^-V;^qhE@GC^)5~`
z6-ATdv^fWFjdkoP6Sk8-d-f~AajCp&{K+vRalWuR5U-%{oqRo;QXG+z=^EH5lpiU|
zI7~Lqq4TD0(C|^9DpsZQSH!ssl=me-zRVmKrr{TB1+Xg>>1R2$SgJ?FudtAjJ%%v+
zlW7&D4;#wndVQaikc<aQ_=%m)zX1;(by?p))6eJ%s?rU*BHMzXbB1#UoqiFsNVi?&
zHiNq;Ub#6R2Ul=Vu_E=S-~xI=tPvG$d`?H51<py?T%`aa@i5gqFj;Vxf7L9j_db@R
zD!@k9_&d`^Uzvo|n9f*WrsoJX`HSSb?iJ%lk+6d!14KqxM0J}sICv$FIziMlmaq`=
z0>}34N}#8sV=JJvpSG|)DZ4b35Ws%)<3LUgznXNMG!s>~%}3_r&j5(^JvuhF+3Wql
z2V&OYWde+kaUNU|k@TLcIl<nHM~f}%|IbARLUt(QFDoJK;UN@R{szMpq&t44+`5&d
zlB_!%<q><uAkCTEnJ!6|TySf}{S$GXzdo%G-MRDMda&v)DzA)9gYD;Ue2YB}c0Ihr
zYFmzg$+We&fwuOvm(0?E(AHE<&Sn?`0)ZdTT5XY0T>V?N>IQ3^B5g;qGce)L`m`YX
zJtFk|kT(kUVT+Zw#F_SqT+^<70IcEa{e2}R2Nb=oL3u;dpXrsLvtErwR>`eK6%GZB
zjaJR64J1v*F<=iZf>QZZ?U8ySOyTSa-3<er6U4i`zk*eXeKAIqj7+G)L-@I-9|?&d
z$zY7biilJF$$bdS%!ZdO4S*zjPciiygA@~0J(#W)4|fV4;m3xq;(ac{PMEr#3BVz>
zM)rj$k<pIP7Wa!JVXSLz_(5fm<ODEQk5)ayqN1e0-UTaiy<2fN>=m3`nR*}ctH?wT
zg-ci_*Z14SDy+`1qZ~oy1$B=3W-X2Fh&bF$3{gAVxQ)RYut9jy;mK0ph8lDqtRj0S
zUcgno86C3dGbO-}AP;lp;(WcwzFbSp5V#+dZ>F;^A)Je!pV;LaB{N@RmijriXgxr$
z9g{-`3UdA#@~U~Ln4@uJo`6MrauGfwnP3#VVNhUkFx&}e+C*^(USGkteQyh^qChTU
z1|=4^Y53nqG45!l$@S$g90KyrCHBsb<U^q3ia2*Z!Cs@w3fev|uUiIdlY;>A<QM|g
zGMw(ot+E{6<a%R;1qz4ahsFnia)CQM3kr0KBmI@SNGGzZEYIQLrGQ$2>=$}dnT?!2
ze0BqrlJ`b<vphesT8xWQIuLeVU23M;vvuclbjBknTZH}=$CmT|u@Hz>^MPez-&hP4
zv6gk>eSQSJF?Hj{;mzpP+v@cmN27T1!(KcgGp!OJTN1F4-6Qxx$g45me-iG38fU@o
z39>rVz~F|UXUpo_6I{i|wtKxN5i@D-LNm$7P5Jl;2{4GxIBjTC5_FEhlAMBKJ4)q~
zHLk6QF~JN+N!=~TA%JA)JIlHeH1f}%Rnc~W&4Jkh0b_Iny}=i7fIYRU0Q~{EqA$VI
zsi@Z}r{?F+A&nA89M!w34=RQmvh|a2=Aky=WzyKaOod-S*1@A)+eMuqmlw2iC59(p
z-Q!!hUv57muJ(Q>S~X>vmM!1V+{Vf(HTv@XK^r?i)a61$LOf5z0bN6^DfsC&N<ex=
zw=6k#rNc23eA5fZ0!iSZU6h6{zw;2KD<`6&9EEXZ3kZi0{d|R4(N5N@Y(rlG=paZP
zkr=_;(k%;`V`hozu~W7zhX7Gx2vsRil$?kvn168b(a%EnoMRUyD3);41HgjdPFZ|>
zeMx-)j>=YppP!R+4z{k^9ci*sxi7l{Zh!|q9De{$j;pBFXa@o<a*j?6vcTq?$H)FD
zFhg?_EDqKE6L8&rNBx&cKplBP+PpDwNX--(XBK*RL9_?418kR1SrofjnW#K~-5@0*
z!+LNNr~uvAUOl9^+PTlR>OiRYUWqsYqS<D*3*lp)fqS62idRSRL`={hd+oo)ARgR&
z@<}Hho6v`?CAkN2OBPN&@;dxCj;ImF4?wJ8XUsPCSNIj=79WJc0_X1Wt3lvH<^>D-
z19wz0{m_AYh9o*MG0=^ygFwN1s$;yN?bD`UBjqCA9SS3%)D)eeg2HN^_6dZpLB=Al
zYT=Du0)Pe+9ytOaYW(68-MDJPEXR7{-9jQF+R+1w`u!X=hc|-LlYkow2OuM~SnocY
zTaw9rlW?cX=j*Wf%Lox?&+idLtw|J~5~({H8weEzOHZf&=Rh2m52@S>6sPe`?-eZp
zZf+lg-vx*t21@hkcI_f13*;$4f4D5X-A|#qJO|-{6LFf@GMV}Q;Gq)SII@dq7$i%W
z*&bR5UxY~zZ_v1Jj{AuHh+nNG>O@#;oKl>sN@Gg_MI;>P-N6gJ!tx-q&iR0J{_N1W
zX#C5U)>v;60^&p?Onnv=)rlOopFU#q{S#o#%%eIf7pI9GhDO-4#6j=~4tjWzu*Ti;
zJ@B08=#+xg28si?&ps=?JuqFy#f~=Z^^*>_x}_{ktBu&IU+j8{KMUx#-+^@(<>!ba
z970!*lCD}c+YRV2c#De=fBqjdU=<Tz^N>O4vT~4a9RXmF&QJaxC_Uy+s5D@6fhG>e
zPPAFS$n7|_JVBI56;@u>tS{N(y^`_J3_*D9n;XZ!W*M*IWDyl9L>XTDoXt=d(puth
z25&AQ>+mkh`xwPS^25+Y-JPsGO(F4!89!#F(Pic5J0fMr`6KhT5hlaJLxJeeuEC!i
zr_Y?p!KQ))f$N=s5ZH_P);XXBXS$ctQlb}-D(&eM0<H@L%vm5qwG#^p3_o%r83sPt
z@$Lpq9W=YQC^KG~UKZd~O!=ltldw96Mtnu%LeN?-p`h_eV<hDoYQUUVJ`Dp8TtV|l
zT1U9>rd*^9uc6i|u|nXIg7CuU`YlH=DQ?4t9PO){>T^Yrw<C%bDIB=FmH(j;RrK-&
zXkaw(ID^Sd>Qf(?9Sjz@B{Z3DHlJ>&S;h-m0QzreRcLD^!RF}iW(JDeEHTIkcotf%
z#g|fKmbnZonIgMzw1$jPr(Fg4q2>!u4}6Ai#0f_nxbgaJ%;Kq$w$)sMx>b(A1!~ky
zFRUVz0+@ZnzDB@fUTu#vTk-Pjcfs)5g8m@2*{ctzw8cmd_PPIbH<;PiK(}KNMB0>$
z=zCuXDCkiFsk<QFAi2l5C!~Vaz=un|SIXJGaAXbDMv~G_XCV*XAvqyi=yP;435pt8
zFPWjbLFx$~oT&`{cGzi5XJL=XM>lD2)#M}ECp3=zflBeMoZVPRU@Xt@lVA+6f!E--
zMBw_27lqvZGfesWocaTYGJ?TUlmxWSKC^C78z#LrFD=G-BrQkn=S!9f*B#4<dDw>H
zGg%3Y1IaY`HyyS$d|L#Zw=qecDZEWb@%CLZ{2~+{x-pV4k*zYrA?I&A%REeq^q<-z
zE*_zE_nOR8qHY1&R&;+}&NI1|RS;T?BquGbteJDCWlk`*{lq4CwKXsvrUyBn(wqK)
zfuszE*r%W=Q?P!8v9=Glek=jF@?eZ1Zv9B_i_ovg+3dq9nEb$N;rkH9kI3cYDX~uN
z**}6GnS#4as0_fn!_mD|m_d<FPjtFk)R23@!Um3UQ~Y>)1lcEkf1-~SP`?!VS@h4G
zp`pAF=NUDoudIR+{6ShzlwtzAOem=IAuchFfSHBzK4x}3#t(#E-A+Mej}30@<B*wG
zDO8I!f!N#7ddu6dj*X-yG#aj@`>D6?_PVGHB3uXw&;9?E3R_R}PqOrX*>6;zbXY?X
zVuef^`VOYpoZHoeaaul$aPnB;8rjEiBPm7ef2=}8jfn#AC(*gE*Ih%QV~)z>BM8_r
zQS*)vm1+u#JOA`Fppd5gUrDzA<2Qtwpm(*Zf%pa)a+Bc(io|YY<7MDW4ut_-1XOg)
zaE3y5rUBd{9EVL&e%9iVV~qIYD>9^|-^ye0`%oBs85H&<@RJnQhjUYj9ykTX|LQyW
zzxYA^zvK1)T?UG`$01@P0=AJx<3)tTf_$(xR&c%d7SI>dPtxI*T^OaYTJ}F@^Y%S8
z)M^lppjFjaXWdFKeVHSiL&h~>=muUnjY$s|Jka7uYN2#gn9V)J%>+HOcyIWI+rkV-
zuAJ1_-U9oKTvr?OxhSaR$n{A)4&#qlG5vBX!s*u3o&@~PqR-!_FkXz19nx-5<<$|-
zW~QL%EvVqc?67??ZnJs_*8bW)M#~4U{;O0~h=kwvBomTh77>|~@|6*l2{(-I;!`p=
zL-|>WaVwF?d|UeOG51@N*|Xpus(Nq_e}6NM0cGj0S@9AVMe#e2MPUn|tL>2_CPv3h
zaeG<z-*ajuB}_ob!jDlT{hBX6&=FRcVt*76_TY#JJ@4%SQnPJvB-~kejDJHgZrc-W
zF@PYt-rJuKt)+Oo@=uw&F_Y$wBT4;`TDhG@*LDSQHx%Zg06nleXU&4O)Q=`1vt3NZ
zY>bB}3FEbR(d8PVrJnYM{Ve8wTqye|^*F2>cdi(rHQEZh9V)S2!o=7=c83y<8K_V`
zQT394YDuEh<cudJonRD+#)xd|k#z1lcD`W)-c5ykS9>ehgFb-%M~4%GXKxB(YyLfH
z`h8jr$!w&%1GkRB<-^^Uu#lw|B5xk65BXwAq<+=EMX}Y(ns^g=5(eBJiSM{+h@e;m
zc;T`Zz8&)KX<JCe9K5aY#P`q&Fiv=&fe#wh?Ya%4+5&-v4^YhtvR9rIN`xZ#eWCIm
z>p~(zJy%A+1Y~Ax#c4>X;7h;N!yevlheDmKNF&f*(8@T9KWXEc8vnhcwg61GdK*6$
zZa?2YLrjOCiUvi8vr~uSt#mWu6F<MU84gn1K_e2VSz?AdYu=WAMF%(t42eaQ``1{&
z4O~E2i+O0@i5sAhrK|b;dHKh^9gT(?9^Y1gdB@lO$N!y&w3Zz5{}+CP+ju_Zf4Be?
z|8IRrinXT6MOeI8$_}Ki*qujKz*fL<LKKQGR60OW#}{H`0K)Uwky(!@xWu>QT)ho=
z2RPCINHV2m4p1bp#cHFa76E}^T<a%kZk#_V``wr@bo*ZOMI;j=X(?T(7l3jEj?*kx
zN@UQO4?bGrpIwauin_emy&5g)WQ#nq(GV_ITYwz*AFGVW6af(t&7`pt_zlK?r+QTb
znH}h~(V!Rq`*xwVH>MiR3AzxKMm~LHK|qrrR6w$LM3l&(IokVyzB)k}Y;k&+jCLJM
z;a-EkNBi$>+QG9l2Bz`k1JdhJX!(?YZ2|0p=0<2#)G`$NG4|vB`<~{%B+&)=o(TX2
z3Mhp341<OY3n{%6V06Sax+~HZ!+_-wZ6Mb&)YmV>`rtZa@OfhVJ>&u;f(F9jh{O}B
zW5(ys<s;cd(o6mWJdPB32)|g@h~O&I!*)Y)UZQ|2(6>z92B8ZQ{$7hM6t~xfkb6KS
zfmEy*xye44(QLxgK%GIO?J5)x!VPS}oRicVGlO^=C~SpbTA@~BU3I{#!htAtA?F5@
zeHBDW4wex@9Vq)SjK|oAm(%3b2HO_N!w8V&V^dX!E-koX?MOZV)S&wL?88Hk@}YH5
zFYzlAzrsFVfGn2OYzWT=5I30or^l+$=}mq}1JE35HLRmVI~f3N4APw+$Tq8)w^Q7{
z#=7n88dIn%3`^TnXfgvik?OhA2juo-X-}|*^d9>!$oZ|VHFFNyszMZK3EBP`Ngr5{
zEMRpiii_Zk^u5uCmOJMzl#mwz!iBmn0iP7c^wm#uVx9I)FjZQaqP?i^tF7!uyKFzp
z>YNg2McfGlrU3_o@9&)rKlX<_dv?Cz-maG(j=wMkYf~PIrOseYZm}9^%kJC45lQ{(
zA{+>YFHQh>4LSDwSUhk%fDTfylsx+V=!=BCBZ`Rqoi6~JL!Vu6=qnP>U(qgOX#i0a
z{ze<zxG{zXcS+2M*%dqqkd9O*SC_k2qe4S<@<JiU9zcxwkJz)?+V<Zq-)!i7dRb^O
z2BZR>tm2&bB8#?{Ju<8F^{m6L#i-4pnIZ{?E6u+2_v`PE!0?!x#r5j4Z#tDWF07!0
zoEUg~)B^pMop4q}<PfX}y#dTf(j3SnbDB045f~CVGXZVN%s~a(A~xSS?(TqM2S>3v
z;uL@s=K%rY@VGBD#X!xAEm;oVCO|oEL8g8lcjy#Jl2qS7iMF|**N3N<_783Zdu^r+
zO5UiVhx487f{y{#;(!<ifKWulfXCTu5N|<|hfIK(d#0+UcD44&Br5G=?YXE)^QT1G
zssf-eS%`G(^~RM@6LQ0h*pY*s8Pyb;f?tut#i?fb8~r4j*X5smeSCVG2Jr!ep;2n5
zqVkD=SVI&NhYt0D-J_CkHb;uWb3!yP{P&&K*Iz(-xBSs#>7(2Zsc9&-Zn~-Fc=QYM
zb>7lJ^uW+cGore644@S16f3B|c-?DN^uQ&g**$@u{%Ej%21gD;`4?U5QK-YUWOKVh
zA^z~(UgmT@L5a6(Fr7#Cgv*`;V-y^YdqvObjS!@?Ji2VLbDW0C9qf&d4ruinLo?dR
zdX$xM5!|POW8@tm2>bn>1k0_qi-L!&z|BprN3W2U(`5t`A=c4E2o>!IbY;K534<XJ
zF7<$0h7u`L_bSkEwWMtnR2xm~Q(ynvT2JyX?6qp+`%x~#P-5@#*I?kl?1=YlMozSf
zJ8_8-$6-D<W|_Bg3R^(p6JT%(Vgwoio^4*f0&5mv?p?NW6?)n*M%A4&P+O$om$sX2
zzuF2#h|*=kmYKWo;{2O2;6y%FEGX2J+fY!XN|R(w7U+fLaj{CzVw5rwmFc;@AsVF6
z4KO1Lp5mp;Kx52(20(PQ0Zcpj8<@Fz`3eeof_$bKF-oX{jY0Lms_6G?Pz4D}Bz;xm
z1~^fZ{%})7GVBXG!8xoj`uhgr+55M)Q@nNmNt1w&Cc^l~#prhLR$P4wp_Z5Dk>lS$
z+fVFqniqCKxamN*Vke}&q1YYsQp)Uk?p)Ck)LiC#qc8+W{u&zb4B*IMOlCmUfEw=N
zy&cjI3;HH2KZ@<%wQK$J>$H^j@3NQeU=$w04CPZf0*HgqwMs$ZNhLM?CH5JC;+HMT
z(?Fsf%FL~UmCWUS)r3F{sHw1fv|}+$;Qos)Pzfua9jb{8o3CG8s)t~z_$PiSs=1(!
zK&l+3We{juEL&YHi@{9*AXDK`2iD@;Lx`|+-}-K;G8;r_;XH+yTCo^R!lT<?2qN;b
z1YjSjfAXoU{i~{M`U*Jk($bQr7&P=bDw$8kNH3#=pj`42{9D4|1w)MxZ~I#Lp{V`>
zr(h(3O*d}`7X^imI(ey0MQ97X6d)zqbBZML+`ccg^^vpG{H7Hi03!yleR63QoaXc=
zj+kv;@6)URu8=9=Xav?V3-(bRIF4}0pPUKnm&X7U-3$8-Lv`z94S>^h>sB1mfmszA
zS6xn06|T>6pUD7DO%SEyP2Xt|{x`$8CY*<`O~%Cc``Y_34X?548~JOR7>au<XrLJL
z2zE8myb77$Dp33q)hwSeH(X-6&dUOP2Km-NhH&uau`*3yK%m?Vy%9oHBwiRM{qAc<
zG&cZ#JXH4JJ#p(qkv91|>KV}}>-dCedw+zV`cKflDG&C$7>0g}6(-!YgK~L^5hEvJ
zUj9Kkvof6`qhn*Uc+%W_82<>>XKM3LwY8(jj3CKM_(+u+J5Yyp2U8m9!b5Y&*LDF3
z)r&idc?Q}Z{|T71v9iR#!i;XbnGe{omzI>99bpJ3qflKhr~FN{h46kEU)UG5ZEPkH
z5ZI!4C<)T}<eLx18>T^s_8I>QN8O};jBh3im@(1Mqf4j7YiRu2&?wl;Zp_L-L?(@E
zcXy%onV3LNk_akt@Ze0{?jVUT{S+%xF&#$>@DlP;65haLr5H?gF@ih^+jC~P0C~{d
z2dre}E0EO8!SkJzVgb-NS1V<NGCr}xXk;LC(KKz0v|!XD6ZG6`aWp!?<kI9MMF<2m
zFvPFGfCJ#KUaA9;7%bM^-5fA#F8n3ajdKP4Fw|dA5c+33HFWE%s)ibqY9g*iJE1NH
zIoiCOzsXblXqCeSnnPR~3F1G-O5{3D_Ih@rMsl;=G`#W+x{!pDP2s)Jci4|cx4lh}
z&?Y)NMA?{72bL;NgP($7@3yF@MS^wPa!p8@f<on=Oa6Er$sG<9N>121P>~fhz2NJ+
z5UpWje9MO6YUW$A+(lqTV48#=(1SuM4ijN7Ax<Xs)CSe`v*hy;4iQXy7bO|G4!9aF
z?{j^<04FCWq^eNYL7ojNBSYC?XU<&<SbFAeaa3JU;e~Cgc!lDe4G3Z|$WPJ!jB+CC
z^aypy8o{c=MdT@NA6&nVg6RUeaBo*hBLX2P4N1iBBp5O*A?mg9bI(xpCfOp8fzfAc
zGZgU}cK^P#hKup9Jb$%86L=C-V^mzu;!O5gSt1<33(daPq@YM&^JhPqE@6RIFjT6z
z#N-6*u*^uuFD(`(zQTluRt0)PZB5O`6JIrCmQnhSBayB93(EhwN}gAsw+U`)BrX|3
z;!`%kC{>Y&6NtP<&XU?=oGb8|Ebx+<Lk||)(MkYHLq4n^>+m0_ilMUq3y9r$pd$(&
zjUg0^aF*K`GnE(b6HEAGe;%T4l9)Ud`Bk>^s|?l5GX=-d7A<L79fCF*gIn#U=mA&H
zp>Y1s9U{ZX@z;x?fwp*5&I9#4Qhr2<1F<WHy|ayz`ZXYHjY#oHB+~+1TwD$l|D{8(
zE_$qv6NW!2k)Xsv!~fFg3J`u0L!d~HSb71FvhG|oV!UqC3gct@ZjW3ikr!&i4s;aS
z#~^oXhpp@T4mJV$Yte*v`ZUB5$=E`C!P~xp&=daqtmSr?OPnbZlw;fR1-4}_io3A!
z_%wHzHuNS!P*`JX=ZpMaAEeWj!r+Yk#syMuD+BKlQM8}cDhSGXAo<6Z2;*X-ZUMKu
zeN_pMjGKr-N3-B7Lxk564R%Nc2Rb`tLEt9&3NAiScw7|JaS7EQg-JXas=4U5jB7l!
zZlW*j*|SCL&%ccRzX1`HeiK;1+o&N{8(;q?AnyM)%=`cU#Q&2^$a$o51BGTN0n%V&
zj{w|}HC^i2;W3APraj%#C^;bsy|#+Ezl(L?r4X(KpqxZB>r5mNMGn}glp;!sR6iuS
z9M|EQN)z@1h`UUXOnhxdiH7{|ZCL^U;Oqa(-w25qgAPlHj;1E*W(ZyG07eF9CtNba
zbe&{9pP$lEc^%?Ip-LRLx#=8@wlYi+>_x(PKzS|jP4+G87MqC70c;rvH30MAvyzJ<
z@71AMfe8N(g+ZPagPz3<XC@NfAoKxg`Cm<~<v-PqO=N5h(*#<nU7<W2zAOnS0ieJU
z@(9sRKO^AnB?|`t+|VVgR$;~^5%eTjM2U+72t=5;hS(nr>%w2Ev9Dec-8Nzq|HH6=
z{sWp2{`5b)eC0|P5Ld8}AHIS$afQu59e?rl*LCuUgow|8-j=2&DTIO{fD*TOr^0E|
z8=y&XX(gkUk66$hkoGYD;Ex=bSkn9g#~2s}O4YYqS#BcD4IhS-AA_z55eT5E%4BRY
z{#MoZpF$o>eJob_JrCTp*e4RubQ)`dR0GDlJ7l>7#vADH^w)L6c>r0?-zZ=!OafGM
zHR|Ts@;Zcs12}|z-vRU>5Z?957aj+nB9PpW`N@qBJP@!@0IZBbZI{plf&En41LJK#
zAZdq=3<K+T8(<Vwr3tqZsFQXBBVsb>7e6y0Ln;xvs3G5tLZS@(DZ%O*m@OK}^ni|7
zfeVCuStWm-8KumB#Zto3H+1NndAn|h_N$*P!FUe%8Jp@5E-Wh<1vY&O-(tZ^LW8&t
z5HrHl<ct&IU!;~M-Z$()q)>?Tk@i6x3ZdvYJYIerX(Pr+K94mdzztBm0CW41t`%+I
zO*kcZzRFjqq06&4oDoLILI~3E<fI|wh7|yGQpD|Y8br03cYZS%$)iYpW#H$PYjB0#
z0pYwM>ZO&ap&h2!6!PSapCAd<+zj7BK<-e{UK#hL(k{apOp$sqym|?KZJ3{fY5AQk
zr25G&+cupNt6bWv4$4XDAGEdvPsqH%6QLkM`Hg$3Qs)&;#MrZA@`PtfAPRyUxE>>e
z$-vSJ4+&x2LmEF$*}O6X!laz^1Cb-)PKHMG!5zDzAb$759$_<0`#~gJ9w_P4xzEkk
zC50P3eL00pa;#nmq1ntwnU(AgD<zOn;I9!D6vtR(oZS?wN=8Z5j1Ue`lcAb86IINu
zZ|0`;$*nERq=6AD;<w^8f<Fu|U0w$siPM*|9?vbbL8JdD3f&dYlH5j+R(fDtNw*Go
z70fmjJ4LRsma-e;^z#Uh;*`J*8;;dXp~r}DGXcK=k$luXg6@_|qmI7nMTD}w<0l~0
zqVKb&dH^Y{=0z;Ru@#Sys4aEx%n_?K81sUI2_Rs->3|ABVy-YXn&IRVDnV^|n#y%x
zTcGXiLj?$sh%_H>96~ZGhQmL?&fjPUv#B5op~)g4Mhtx={Ii8tAhR6^(R4I>?~u?1
zu~8{FXILCYbLbA@m46`(gHABDRKhC2F-FYW6=^j9y}!_-dT}Sy?G<$3g|!f|!Q`Ud
zjl*bbL`eJa#=P<BFAp7)>`Z)?1sWN2bm?6zz^o<gx)`KPAX8m}fRWDBR6ZEU;Gm-=
zHtf3{;$YCifUliu90Z455{35o^Tc8yqZqXeeH#!3^fNYSADYmf1dNp}v}6e#;mI}Y
z0tuoB(Fk|CuP`BRc>H<RJvtbLqLr|Kz&L0b89@ys9e;9k4sA^awMnb))2K^HTmN`l
zrSR;CkPts89ms$O1_udZM&dW1FbKf_1+*YkmVb>Vo+ZF`;bG&ae^8JZ6P!-@H>4#I
zt!SAQnz(8NE=o`3Pfq|;vkx~MK#NNDDy*tHQV#-ssKgLn#zHG1UYGHbH&bkDH*JEQ
zmNiAH+6NA+oxK{43L!>T`(rlC_@DE6p@D%7urpX6Qp$$GPyLQlH#-<Qi2>zOC}YJw
zDFWyL+`!(*s4GxuAGmwJwKrzsd$laqa4W}eMpxio)uEZI%CiFf9CZ;#29ZROR=DZ-
z^#t@n0!ffKEVuk_Me_Pqt1`bkoecbkS7j!SS$;DQS51uHz}0v{u3F;ZmC^TAqG181
z6WQApO+JW*U6fb+y75qsKzn06HRb8ow^SF?C8D}n{`T3ej*XW@HJK02Ed8>N`BhqX
zbIeYn_2~GbTNxx<7(Hq!__SqZDK61@;dW=FrdE9=4oEzF-yMPX@P!C9+rDq#whe~-
z<f&8b4s`VN8c`B?g@sSE>VbX=c3F{L9(wCd6tQyu3PZBHF&zkc94S_l6BDU5@kOWr
zgh@A1GJYgMHWCr&fhHy;VLFiva5yL`AaA{P?b_%E5fKseIfUd~g|@nbif1dmAqpRe
z_uIdN!LawA63t&^{8d#|m>G`7Pw)oeU~H+L-GG_)<HPS>_AD&DfT58AzP^X7po<S(
z$?xs$5n0)hkwjHEfa&OomRL|A`jDvYy%bBqXK?f83^s*fuc`EoQ`GV{HrcY3+-CaK
z)zx3Wes!g5WP@XB#wluZ+X(4h_*pzO8`}WPkA#GTpD5G>VDEqYI;*8+hx~M!kB*Mc
zZ+bHY#YR1T382`I^?-BZ-A<nl#DwU#@p6qwGL_ZTu*!#pgwoYUbM`=@7M-wYSv;$w
zgK3x|f`Wks8HFqpXb#2ag@uH`YkzqA_VzN5g0o4sIJAc`q!YObl%IdTproXfc3(ng
zR!UqP$yN%trSZ4<Dx&|ike@j$JREa&L=PW+$sc7+<6GIklZTU26TkZPtEjMWVynZ8
zg!J_E>FH_g2dRr73Bf-L*#|*E0ZtSniSUn@Q($JMJTNcnjf0E{8LzR~e!r0jkvW&o
zpFj8V+Emnq`K!7~qus5o4mia^I|M{TLZdjkkQ$@w?v{=x=xQ#G<O_<K^RQuSJN5PT
zzqPeVdaOi#JEN&-4fGM0xJt4)>>wN4eRpMyD14ro$vP&2-d*fAIN>zwPitxAqI`iF
z6C0UKr7&{<NL*EQ^|hvCe~4LC{NdqK#$KXO1zVAko(}S;1%@`hc)@>O7w4*<OBh>&
zBZQM(efSA-tbw0Dxj8w9Mn*7S!wAXV>vz=aPoF&-hJAwI(S-{au3zUpuZtZ)C<SUN
z{*SY>v(R5HCFEfBwY7;>f<i(gIGlzH+cs}L2RIs`#)zme6~=EMm4o6@XlN+1OZ*<b
zo_|~+iYsT%$R7-Z7UFf|`RbaQPkJk0a^W0d2v}q0;|^<(M3IYgpxc?}-xo?~$R5)&
zGSKc_{pCBNnD)KEFXd&%e+>-W#6?0DefTMKZ(RUlxOTf^YaqCCvfCEn4bE)!VP%Rr
z)%^r-ngRRp2{*5Wg@xKm8yReaGd~pJ{O#K}VAFb_VL{|%+RzQS7`6v;{L@Ll;b;y+
ziEVY|ZyAP=xf&*6#F6{*)8gMhf3j=7Elf}677_|7w!C=pct8UOD=QiMGdFSP+Z3ED
z5QD?cn1mK3JW2l`F)@*pA^`8fVNd_Y>%zA^144verlcLb)wkZNV1J+s4&eesv+N0%
zJ(H3K@Grx8^%@aFmv%zg8F8*$7yFJGs6dj?f%Nz+%>`cjIW#1W7XT)pwx<J|(g7EX
zxhz8hQpc$gOJeT)G^Vce%K6D%!99db^M&h|SKTw7L%JD(Ljj_ro1<RZr%&5}(+DIE
zb2LGQYZ%8q*;-pOfIf=I^JvzCWN7~(><`Os`1VcT*f?g+IK358wP~KFVEREp8G&g*
zQIQ(cMI9X-Xiq^g9W%gg-4b2qKXvLekgX4pria7zXVMr)(E*AeUS3|Zdu^}d#F$*y
zICaXhDsI^WM)HbouMyOltW7UM&CGkvUEH{a_e1NCqh)>6VsirOEB6YQy1##S!Tz2Z
z%g2>Pd>fcVf$=Zz-@hL=ZBjWUvL>*IdVSoB7Z|XtOkDwPJ}?3TS4~aX=J><irKAkT
zEBK<fmvch+(7}UMeV49Z$KG7M69!?i{`(?86(D5OeCyhYl<0CGtL=Rpo6FnV3tkV-
z%O7k2S%ru>LzdRZhw@qxboc!I{h`4GpW22|Yg{fkyWjEGpo`%&z4`9r;9w%|53mM`
z^&B>Tfnwu{aQ0^Lhnj8JH=n=)VZbt&XI))gYqy%C9XMcVJelgInb{DIHS`A$?DqX6
z3Uh&fk6@a{oo|SYoPwK1_S2|4qphpU@*kQnCYvUWrWj$Z{IUm7cMH8}bGop)+Jb6v
z|Nn3Ss$r%eNF2`}^@WX<b^XST_z1BB2kzK3z5?Ka|IEOi&(7{%p~BRJx;9-Moje$d
z{@XA|f6-9$__1u~cf_yQdZKppV=LbP{veRTd?Za^6zw^37ps9TS*$1CD#<Bl#sZdY
zrUgHhg{5cMD<Uim(Cu^E9HLEqJw23Vd`bmz`gwUFr0JXc2k{Lj4-cf-BEAJd4R`Ne
zwNhDd@G2-MzQk|-XogJg>z8%09KUeopXF|sDt3gL32I;naVP!tU^T${9`9i=gx|;s
zmg54UC*aYepT9=1JF)uU+HaLqa&|iP!ETr1LHp9u(jY3<CYWZnuL7@gb1`o4PDO%t
z_&WaGzW46lWnJObJ$?G9<HUaB9bkaRkzemFa?sd+v7^0x9xq$IDJ41i17r{LFJD8r
z00q`LXJrkIvSk^lHPX0?n=c{$^jmZ8{CTww3n(d-ZwlN05D==+tg5>D6|rAkB}EgL
z$Vg!f3hI5&#mR{lisvt1KF(sP^ZAWALnhg6_~ZNEmSFHu&rUdB70CoSYR~)kH|&wT
zc4Mxw5GSqOzrB?oBRO@>p2cuQ1xWXP_`uG=0hJ}Sd8ov<j0P*hv_katriT%7<;&BV
zsC_NBjhG>}jZ|u>smVCb@&<7X><)r^{om@whPQevyhk_U8l5hB9K(PJj^^PHjAhEj
zPgz}Pp47vHmWhVggJv*e)4xZx0$f3P_hh+y08>qV6s>`gF-P)^9Ubl5qcn_jz{tqx
zy6ndyQ<|RcW5o&!h{l7U{4cLi0fT`s=4h^5gE~cOW+rmYgT^xG=!f;C38;rPstV>@
zOG=XQ@dr!yVK!sx(P_8~8kW=Nbad#BDY}j{bM=%AWnYm-Q!Ox9Kjx!5et`F|VIL10
zn<-#r8kR@<H{03$!JneEuarJOAk^Ky!9hXSEiLB@e`B~3BION*pKEJ(N$-(^8-5~U
zv3oGHd9F>jq^hO{+Xl2@bWj8`wRInWI+}=K5cvz)RooNPSH=alNq+U3?0(JEkRj8$
z^k@x5q89-dzq~}yfa!!bs8!$H4?rXH`Ev}M918D1H#UIWkZ7ZiDaFOEc))&2cUT+5
zS%5gE4;z3Cg#<HkHLlAQbHR8)*T~5UTGB!=-O*7@q%uIpfGQDP+DT<)am=5w*1ak7
z18{hJ4*RD-fYH9$ceR|H3NSGbwwhHsFI>QweBQ~Srsn2XZ{9e-FgGP{Z`lv6*Z1%5
z1@e4$RKz)6MSU?f5>4FqN%)kX-}dd>8Qlal11!$Xy}TEih<e!UY7DND8q;g}(QnL+
ztj2aG^0X)9<l4P;^UfF<LDKvQ(j?W~E`7uwDDD3K@nfR5(pO)q8pTtO`2qS{CxlJm
z^IQd#Oic%%5eu_lou3+`WotWy?v$fh!#DIJZyrXkcw(k`uFKyW5R}B>hO89K)A@7f
z%*@RZP<)P&i3!|Rx0{8fKD9=Jx&k$tRIO7uP_V~--rh)qwoBEkGeXMR@8AxE9!MM0
zbvXFy@QP>y`t%*c4~PuI^IIvcyUkt?G!PIyW_8J{mQzev*d7B^^YW5D&drW@lkSFx
ztKs7vXF@_lXU4l@N(-&9${4SMX;Vmm6r(B-o8kXuyj9vgj6fJ-b(c$D-??)fxHTF~
zR93I<prYE%`El3Ioi^xW8p+vcX=!=>{CQp7Rscy2_3M@7<uNJ+;h2i+EyCY$UlMmB
zW1_IZ*)Lzl-c1=87(jWLa`R^Dp#zxeR#`dF^WLj@x0{>WrhgNnFm?M{A9{AsH&P>(
z2%~bwF7>S|UlOLn%&yDM>Uq4e&(rH1?+WT)Jd0JE@9uvH13Zh-orZqTn&3CctwI*t
zJaT}BmKJX%-j|XHhC3CY1$V;@oyZNqML!<Bh1xw9Q@(EBrof{kOxE{*IUoB%C8_V2
zC1xlInC{p$8!pCW?2I5w#D)Qdk85d_rU}8_X1M?Ob4Hq)vU%PC0T@R^-!h7_Wk?%c
z#Os8_#8&{*AgG#t>iYFeU*D}QwH+OiN5+$lg4W>3)}8xX-`E&ba`*=-gc}yFa`N#d
z<t#O~w4kt5$$&t$0Yk>1SB5pxeXh^u8ZzSew$daUJ6>40fG8em5xRti91w<*j;rci
zt<$G_YDbf=CIn|ajf?9+a-X{EHSQ=X_n7yl^%)4<N3m$E<{4OmTHF|U7SvEDAt005
z_y&0p>NOV~PIz?JJr(gC&ra0V(WzAT-?435NdBR#q6?@<?m3T~c*v!Ndv8f{@=x^Z
zU3wE<)!YoZUEj_FS4AaCH1FL;)NQ+Rr26%v4mkcK7dOT&i}w$0E^lZE%oXh3vq2i(
z9Wgc<AON-II2?7F>PHgifmm4a9UJU7Vk~IeR#&TX#h_=PlOak#FGoJ?lD5mKP=9rq
zB2mOR-^Tc9yoY*<;jjf6g<ICk59mR)JDAr0-aR>a`SmTa|7k~>-a%Zw4OQk?`w*Jo
zSe48!T#yyrZy}q#<?^-M`Nc)8dw-{={rvq+6C=x@(TTd-Wd-^=dD{`rh$l}Tn+_pz
z!m_D}a0&>Z2ZQIP>iOfx?+9o;$;|D5sX)+)`HfBE>6w|u7$t|qHmzaGFkeyOciQy<
zq&S$O8g}h*P|%50jnmrNegOf4qa!$O(>%Ipion1hS*wYO9y#YVJf#Is+*S4ro;=M1
zym$YK0HiX#U2Lu-VC3+27`5y`n>AX`a~hy(?kV8b*Ec?1-+tiO-p;D3R8*oGrwcG)
z@s<~iCSAV%T@5VRQ%MfP*VpaqejS?Uf7s7UlLW6~9d$}sd2@mr;;Qv8YEvDm6gQda
zQ3L)ryQpu#5Vijy*)>H`n2)seR}tFQ<<R)OfJFzEQ*q<0o}S7oa~_B#x1KqBn>@zd
zR}z67(rsDYMsbf7$(CobdNkSRm{?dm#yXD9&N@r}Fv)fV{IquRSoS$rjHKe$y9pZc
z=*H>8633|lagW9Zo#RDPclgMWgr}%^Wq*fG$-Qeo(8eSqDEk-cqa(Veh&<c3kBIrH
zYq%r=1W@h>JmTk$aW^b1EFV66`J#C(OaFcYDzxuzoGhV+F~o)p6m9E(!N*x-h>FJv
z#Z)b8(*?wF49qLQ{#^ZGZ{eXJ7$~mxc`E9)ISuY_B-`lxL9V>*3AF1;UJb&5&Fs%#
zziMu*f^M=!fX?j76?V;8pBn&F{oF7i&s6$sUZeI!`-MI<P%1pwH&pNnvn4NdeAlL>
zPM|u8$3}40_8A6HVZ)-ZhzJV3?Up0^6jL-mEyB6bkEt>uDrR;0g<7TT%gme{TkuK_
z20nD7@8bkrBr12$vO>RP65np$g!uTjENQeGZqT`J0QZAjyYs<L%YGigY;1b^_P&0e
z%Fmzuoih<>+rHlEEf33@R9J|KVMV8wA2eV%vci%vhI0ny(;G8LNlQlynu?2yf6+)%
zW>JKk#sH59y6zzS<A70oFx+(;5nezkc~ts-#!|2@!^BKG0)0g**q><ub$b8-uiv~G
zZ%Wx=yhm!!$o8)S?d3y}Y(e54TOU^T!k*bmBkJE)x{~%wU>Jkq`2I9y@_9C_(cis`
zuv0g_!6-rJ1v&*St>D6KuQM}wF>cO8mqZ&W%sbb|rKUP!DB4K49PlZf{Ymo)XZv7)
zOjlngq67+Q@WcwLUyg6-`Sj(J`z9x+rZ7Qojngr4@h9W)m#Wc-DDE+XYAre+PkUf^
zoNp%r)_~%;jEudEjH&`gCtFbuLXI8?YhAjG1J-k(W#`pp+Tm8GOW!dau+OU(nzL-X
zSuQ|hCXVCE8Q@<ixj<%TCmr1kd^`qH+nseQLDpKt|N1skZiqq3AZ9*rpa)RebRPAv
zTjMHN?29;O?Vq70!3+i9pdgw(d**>g!BlO>9IJ-p;^IhW8Q}uJ9PI4uLZa8q%!oi3
zf<8zQA<LiEapLN~7Pbw9+ZXBSG(!{UkwiRk%w;yS!s|*|#0H-ou6t5LS-0GMoP1ek
z@A5$kXRFU%IO(wU;`f~K(<xWHkbi)LC^q(G#Pg|K+3JeRofgJjbaY6v585bLK(rsn
z&sy}EjED40SHAnQ#NZfiPq}XsHxCbslyit5hmZV0Z?A8kK6ytKoze&W0!RAO_nkX;
zj(U$nZJ=4c4CNr8wb8jTz${4WQ5}%X-t^#owA?SDJDBe%Shs|9d<gN6kbNGaoVqiY
z5MN^Ws5|N&ct&h!n(kwD-rZ^jZzIr28kAW?$IJ|3W~E;9F`z=uF2T@JU0hjqQ9OcZ
z4aFFYW&HrUw2~60p={c)AzO)I*Dgd+tjn99-OI@lN|&c$zPUX<<a^pv!|Rw!GFtC1
z;9k9sBH{o+M7C!j^Z<<CSL6Uyfo99PZN?H%`UT(cPiMyS=V04NPNDWz2Snl8(MEO*
z+m!1{h1+zQF59;wpMrbO>%>I)d=@1orMe?5tWyhaQ?o!GfX5B}m4^9~l#DDcahx<M
zwdW&q7t+*P;caU3-@y537f>L)yW^htZe?c%%&?^7MTEvcl2WXkR+b931!kf**2{1Q
zt~eVn?{yG~n|EtzHNm*m(G3onFP4Ky(K|k_(XwW|Az{;?H}!|_eOT4e!X;?Oqqvq7
zF3hG>#qZ9@!~{uFwt>xi>u|Ed_zkPORuF}3&P-P0fb&AkijW#tfAeV2)2G$*aUc2p
z_>=I4I5n!0^%c^G2u)94pZoj0*PEx`iggc5T%0Z$F_J7iP#bwP*l6psi%4~}J22t#
zlD|l*v+U=%nW^1yGJCphyP<EQ-Gvcvceh=YP(co+ooZ;Z_8oXR2VeU2qjHw|eN@*H
z@nZjWKECIu7HpC9!H(VB{tm2%HjaJ+L&LPA+P$_z<rOA2s;ZQi4h`y^8wMF(ux`h5
zqZ0kOR|h%job`~y0yRngY9AOL9wTf{Gq|cOFYg)Bor(O;4u=aqDcFOXiwjrW-|&?0
zh0k%3e@3D3osB81*CcMVq_?06dSzwNO7TdbGOKlo{+|l^m7gxOQ<n6b6KZN|a%C9W
zt-x!iDgT$)JfLFcp4j>Y%0>&O2L&ZWt$)b&D6>g5VR}TN==<xcSpI&0;9?6ReK~0#
z*YU@KSEN1pB|l&MPMGVk)#%29ft;eEVea&9I&yWkoc2h5w7r+WC;U)?_{U-8Dp0&~
zLzN1u^=f-0>0aLYt*<!Nh5(!G%EA%rXsM~08K~e~xUQ|O-CsM397pu(Px^Zh;zv-<
zuy;7nnevd;(-L8x<J9i%?$aYBtN^E+^<0CLS<`h&UF9A=e8?c}DxUX)&lGMqGVkor
zw(Z*o|3eE#(1<BeLqkKr*W5uS!c#`MX-Hr?!lAu^`#1N5ZXp!`Oks3!ajCm~!pUhK
zEOM3}8(X7*Mke(O!b}S%GRqYh=Ag*)wD7aLi}0rMm=A!xo4M*?w8BC|MXF;*)`)na
z?;HSj5x*g@CPB*L!<Y{CSPVHY!=RhReQx<fl6UY+Wb{bvnvyS)eHH;v&T=>t3;37A
zShwFF7}jjWYvSdpp4S|`bay|lTidUFTaDLo1g52N-DJEyji4E}q%}Pl&{b}05BN_%
zia%HoXkLes(>0vyF!D;zQps({so~a|^ETK|9~B1T&fJyv>Q!Qp@(?DJxXj+aeTFJ)
z{)ZZWPf@1lxbv=_q}T0MWdbgx>uwkRNIe1$on`>ATF<@E4hUP~<7Q$wxcP4B3HB63
z3h6*=Lqe$6H+(Ub8VLl@_xiPHL`gm<IzX;b*V@i-SoXDN7Q&%KFhZdW(A#U06VIP#
zsG}3QjV}9AwLOioi1R%Nvb#7ro%TSzA*MJkHg;mNPl;Rl#P)kO_ZEVe)$-l5?e>|T
z5A>`1@&zQK`_nPMpdXCO@UzzW=X<BwhwfX}Mqxe(P`X21y3l`FnzyTqm$gSYd&&54
zSMsU7i02fi{U=4i-ZuG*<nTiQFT>Rv!>|->p+$Q?!#bcfp<cQ~5Mx-mCxg-L)s94q
zBnx__#jrBaY>PPNz+XfgsB91I`wx3Ils-I%ND}8n)MLd}C<KZLGC;DBe;#?NATQqm
zVPnLjt_cIKG9K;iL(TIFCr0WOH#)z*#N6k%{Z8g6Z$~ozF3t;JK~@?7bk>Snp2ea5
z|ERVNA=+=SI6>;f6DaX>>YqozSf0ha>}ibBN(F5yu9k=huE)#7L{IkzK~L+jz48Zl
z8jsd@bw%eM7ADGc@7~=hrtqE28$`Fi_WHAihEb|6gU@7#b%lU!Yk4=$Jo*!yIA%8l
zpAV>5c~y+`*T6!|rMtegQIEyaB&%y5_JfKiJG<gj^)~T(bDQM;CYeX#9Gb{)>$ZgK
z-wfdPL!}*pgiqDgCcJW;k&}GM)$~S#Tf;4`U89twT>SmlwBCV269#4-h3tD1jtC*m
zcP2gBJMZpiUEvOcu%)Oddn^65=p^%_0RADiWe4}z9|PJWQ1wV4@wvHnK-gd*k+>!b
z2x$szDu%1a&jD6Il=s3qB$B#VJM3|<PQqC!wsTpGq1&|$4Nn=DYQ#Tyk@~h&+d#bW
zKoV^3hV|>eXw?%Y`=t-thFyNtR=j%s`sa0C0Rby>^ZDsvtq-^AOggqrujRGJt@JxA
zkubJqaC1`HoJJ3?!FO>$1u_utzs2T-<UB^=rs|v&(TY+Y^+1r1+jli)eIabrA-}r6
zcMe0c&$$Li12`Vi{U?Ura}YgRd(pmb@Sxm%&)uH;BtxHHp^_3S*z=$P)J2ZT=FGP#
zHg>Nsyt|7q7Wg^s`oPf8JW%N2pn-|x7Z%QrwA3K*1|t9CAfLoAhvo@uD$GUSLP;rs
zV+W*+&HW7*H+Qd#fw}rd%`AQNRO#vM(P_UIt?mNNeXwekM|~P&?KlTHR5}MW5o%yf
zz()4@hWb|tTn0S9YWupuj*nt@raIX^uTMwGNpEYko9)xNGCR(zgHW}{7Y$1Y<f+>4
zmJhiGGk2HKb&I%`cgO$l=0cXxs?5%<{pHrO+gP2FIn?S*c*gU+-QWhFIB{agTN!ye
zY99bt&}o$)9Zfni+Ga|Rwk+Q+!+dMx9jI^g@Lu}N?)tl|Lf1d~V1;AL<rOyL@A{m1
z3YA<Ve|I8V89qR9cR!elC=Rg>jDegVDYLZRU8?bjHCLl&J)jHmH+ISTS`c*kWwab)
zck{C%5hsB)%Dyy&3%X;=mb6RR7lPn|2M@kS!0g8tDZq4`8q*>!UOx>Jda{e&NX9>x
zeaoWdDqEZYPqmTcS;lEQx*WbU-(s(c+@aySM#Uy@b&rg@q}18y$VjGLyFh9&KvW4B
z;e}Dj(~=wO&v$ZYW}hoYXE!9aKnB87s`$mx-nBC`G|DUVEZ=KSHL%2*7On)s_<=rS
zD<HCst@AySCffz?0EMxQZrvFzn>1<=pZrwL^bqEf&(lmSANylTUux|sHQ?6%hnb>z
zLk_xkVk;lBQz+#HKF_zLotmB9Cv>?A(ori=gMfW(Y&wAAAV2A-n;L+&JI1vaAOFJy
zRFr@i0sI^sDgS0?#3EkK%sja~>-*>tdw^(3<JM<b{jS@S>8E#Yhwb#Pyf}3E{DGJJ
z%Rrl3)#{<w&|YB2vCKV+jFZd?eCjq&>9)CeedK|C=KL=P7guT2=R?j#>N5Hp-z-c{
zzJkI4?#ZQYaNUb>wWsLJL)R3A@V%p(d6|>b+d*b!j{4(onOg0l;<7D`*}Kr4ZeEa|
zuc@YmVGOaai|8S4N$PRW6BGM?{LqZKm$4)7sjPRHs2A?pEc)qyc}atofZ{`6U*9mT
zma3{xS=lOF6cv;BVIU?OnLh!bgc`0~H(r*M+K}!6Y~dajLG_V~%e$EGu-DfCorD;g
zmoN;+{(ee(wAG!mk2mYIfE_|%!odN9PQem9+#Z7sp0tD8T3p#R<16dpZeoVqM<eg>
ztXoH~PI+Od>cmo|`{IB1=wC2pq8zNRkayAeAm4yUr>DJ)!QK<ta7<*tv|IV?bCFM;
zFwoIq#&8MV;!J@y90OZ;b;wFV^H)@VQW*+S#zI@7p`qbX4|N74b9R%qm?vOu!oF;3
zT70cJUc)sn_nD8k_m)d)Ao1X>I<<lzuLPHyN-4AcIASPer&EP^@8o3mqD(tv^WACm
zb!)lud|o&E;M-wgyAvn@)C3_dnU1nH+S+Z6I0emT*0ZB}vxX}xE1+s+c6;?@eympa
zwX?VH?}$1gacEhQa?u}IYunusimF&-mCe6Vk@i@aPCIL{&#1umKIH>c4sztoPf4$7
zxwe)~|0mNmOUq@Hs{lz_Ay37;Xp~~14D-<28|!vpb38gpP&-7W7KQ2U!naXsIxT3A
zvKb5^qA>*MIrXoQZjFQZ9MpEv62XJJ@bY#^aLO>kLU}aZ1{yl3atqZ_ebvq5$H-{>
zFh(@_e50YJe*5>%r?>M00*7BJM$67iym|F1v&GAfS=Y|em5PdL^zZ`A=+V7)MrI1n
zhibf*&ouHFtfBbwFMr~DU`@C7?ta9uL{9hnXrj9F?tqDb_l?GDJ7f_|0`|ujyg6Vw
zfaTC3mHUv1Dkah&Nl0}*aO^;;1}rw{&HvWknMd`QzkR%#3?o^xRE)KVp(0VCk~O5U
zCsJ8LiBeH$v7|*BQX{E}eA`hfkyNskXiqA}QqiVJn?yXXoB7Q-&%e((&w2iNzJJWo
zIJ)oeeP5qzdB5M+by19d#>uxNUy6zGYpJTj=T$s?+FM;;U;q5BsixB!3Q(S$7^&N$
z+*;b1o0liQo}z%6rU!2J8D|)hLcU>lf7S{or<8!t>(;JiJ-1iF6o|0avFHQzmoBxi
zi4WX%JNwQZVT*Dbm`WX1f9H|ga&A4`+E|qwlI4gwI$!w%)O=&_-<<K25fk!fH(wf+
z?OF@sMz)&ZzZOfV(EX~VTiDj#-or!B<>U$`{VY4N=wsf=AK*6_-7klsa=3*$IM(o&
zf0;U+wO{^n+_{6y!_al^WFgP^!UeBN&Yaio;oYur*10c=eh5Q8W%O>IJ$+i(L<}bJ
zL+!xa3=b*M$Qce^X_%B8tHQ2CUQ{2vwbz7<j{aAT^6uSp4?Se))}ub&+TMKEuKZpi
zd&324Y-@grz(wE(lyOumR6bj7+<5zi6PtPy*5>Z(>)l$Ns`q=F-5<8py-<68TOAYD
zno#F;pS_sp9vm#l4CIG>GDT*i-?6~R$mq~lsouT8l;wso32jyI6p2iu*Gs-mr^VPz
z?zcM=3lAQf2zEQ=T}0HX)(z5&$0$sY7^)&4!hR1|Xmt|VsB6%6(+W2Wkw3K?X`a;8
z<32tY<Kj+7{_ZV`xOM5c(N3YtZNl1?wA%U}nVRICeoSPI6@|xp=1->1>c}Fiw;v+i
zt5+{`o2$vmgZo}B9We9YSQ&3Hfr-1bu|dXC^#M(3eF3cUe50S*vinNtC(LfSHp->4
zSDIGHh!bH0B@-b1nYKkMfnkcVfk7_0V8yeF!E<=2T4ryF5Q!Rgra7ZS{>twAVo`3*
zXNgIsX=hT?EglW2_c;RMi<esedgq=!d-gB=WwHI4gW=!QpOlnrUg#GcHENdYg`m7J
zFdWLL$RRQ4AvWIq)8)Gssn4&P%SL$^_Mc~9a9n*jY-K`$Wzn10ua}Ls%aVMG?~Q#}
z0<fPG@MY`abp1UpOV3N{9Uq~+&b)hz_`7%S8Y9LD(m0cp%v~XYfgizH60h64s+Gw)
zIy!dD7X(;?P3!*x&ncxy%kL-7n+~0A5xui?wnx8_`5Bru{O-2+OW_=&7l5h>%Xv94
zof6%TPG|NNGZPRw;-(X*65gT5(85;n^t3d&kHtqM?&!8a2fHtvQ!vpw<YE~aS4?u$
zP?e$+t~*m~C;jOfeYBgoToO}!@RhG#nYUcS=Zd_NqN0>m=WpU+t@!!futOOcIx}mp
zk0|LDW=`L7bCcg29`$PZ2=E}}&y%OWxMA-()omx%qBbk8O><!irLUS%KnbpsK4w;n
zFVxl5%}DNK>wEdRqf;X%MF1?$m^m}aY^An#I?acRj4y|d8R7#8ow~QFi}_X{8v#j6
z9UR)|L)84;Nf>|D=$o-|ZElH}n0(5jd3vidG8%n>*3G}%)OPM{CR{e21|O4^oBY`u
z)PDW>?!epG*`BF`6n^Y(DQ;YCVq$P2xBwQ!zVdd>Wbt3zc0o6jn)+X_7&u<<KyQe+
zLE~#(K(@Vi?`@c%o-g^El2UDDrCzsj_CRZiXsuOF64WskJwI!D->=HdoPFqLCgoh%
znk5V|eK%aoPH~@twstpgO7W~s)#h%YEUHhGwLM+I+g7=23X>f2{tI<kvFs$m0`V6x
zXN%kQqQb)Mg_BcLQ)hf${T21^13!gN#~N1~8F{+9JMG`kU$nTic+DHrAt^kVC#`iB
z!Z;GRh%46_sp)C@^HN0_=B9b37k1H#<nHrg*|`IQxX+9p`qa5X!`Ql0!hF1Y_iR-X
z2Mry%AGEzin9g1vn1S+fgRyatiA)2!76_%+23|9rJFnb(>lRTkF<^nKtIo_KkI@<$
zfn7TbZTBeG-*xJ29&Kj%K$zf3KDMI6m{cB7mvFdWe8v`*4e+G{Xmad=?K^hRg>`@w
zNm36WLl`B+Ymph5aP{h{MT@*=yq(S33VlPvaW=!`<&D>^Tdd`(Lx4mA7;aOU7O-|J
zQROB(&YD{;TO+d=s<x$9`-$1?rn-fD04?JSQu&ga?&pwpSpK4{tUO_^ry6P%IR?4l
zr(HHf)(S(3c*;yad6<>80NLKvtLigneuiZyye`t$k8au>Eh5ITh3U$pAxtfL`ts#1
z(#Dh(ZY*-u{3(VKxG!;~cW*(x3%kve-^wo^;`0)Kot}G=2-`@X8CqJ)y{|e5N+uA!
z^v-^I{ZSHhbadbluO}qLT)K4J%d7mRqw35RTGr*0^Y7~~A|y51?~yj@Df)!N*JT;i
z&6_hn#qZ|ZxVUr!j!HG6Vj}hJZ?(G!!iHVmp-C3_t2=k^LjB&BKM6$@BA~rW+vd2C
zXXO-}hW;blvI{i;^A#zzw}ySy1IKGfiM>ZTNBOE4_f}tbne+_LQ#sX*RqF=%>3^>r
z3r4cudk<A=NW>icFx#Yao>8Y#S8P)srL?rAv6^5AcdwB5clM{h6j6<J$sk%;k$rS_
z`8g$c1jGx;>z0_zrrM?#n<n>&r|)n9PoV&(PLGU=;*Gf=bU`Z$+aN49X2sv8O=Avx
z#4M+`6$c2@eerMd_Ek`6RLQjW%MrH-$Kb@|4%9!S0whjFvc5LzZK@jvbp*+%4G2vr
zf_P1k5cbNQ7(O#gKb74xuGH3GNZWSp3T5c<t4#3@&#1$P?{Hu4o)5pa)Kpe}u#lbd
z)4>@7DBpd?nCRGnHA-uX?f=yt`wZDx7kC)6U6V5Rx9ZJ^l<sWslYQ1#UQJO|^+Zu4
z7ow8hGF40KMsu@mes&iUnV1)JhZgLztt%Y8K=Sl^Eup2$`c+-s+LbF6zWjQWRtrn2
z)v&NI`0@~Gf!d7`HO^gU)^pa>ss6_r9U99&;IbQ^Bgvh=W?y8SOGkUV#nrGVE&seV
zaVEx8{x%9S-qA0=N$SbT$-S<rNgOFG#Aw_8l_UY}=!1`sO?mK~J&qjvey6;<aLb0Y
zeypu?_#hdXJDu;5-^qB-yK`W7%_e`Jfr%*89UHR(Xd|+J*WU7GNswEX{~I$o&55~^
zJ$u^g?B~eJD?>-i@8tJh7Ks#|_4*Xrjuy1s7yKvh^Wt8oYOH+^;sU15Xlrftzl>ZZ
z0x@gGZovG6CH?ioV`ICj?n@ntD#$m6F-o^yxZ-Do|JA=<3k)=>?sM>i2PqL%d)V4Z
z8T5T<Z(crs{u?BK651L$lJk_q^)gmN66b{~(3c5O9@BHUrtPS0Ms111AOXm!kN=FK
z^Y!A(k5?8J6;(DiAT?ZN1X<T#?$$hx6%G@*_`dG0X=cIX3d?si*tvQvTj>3MpvRmI
zI`jL_!}O|6yXrm39Fok~!Tf7WLS4Ka38_fF-NnPXlvN3xz-=7^qgeW2U7o3B-sNy9
zDoU1z`SCGD1lh3mtXYB%EF<u1SYGHxi$c`oRp01hx!9}{66ow#8>}@<YT=wu8I$rt
zwbz@giZkPW!lX%C93Pk#R6KlmtTjw4Ubf<@<v>vdh}f24qi?rfs*wL%QV%?}SGwG(
z#}-2w!YXz1E^I?iVJV$TihWDV%g-dp);flH+%;b9@!nTjY-wL%no!{;hDizg4FL2w
zd5^%m;HfT$h51ne;q5<E&5x}hW6m*av8E=~wy)_E)E9NeS#o#7Gm~%U=3csRp^9Qe
z0PbXUzo2YG&aX1-3<%39T_*wYrT3IxYGk&dT^s)<)u~tQHQ47qs0SY$1<s!h<c)vd
zNV<0IT1bcqyFZA;M~)ncd9<ggsp;s&DUatP#7av~&k_4Tg2udmHOpc}>h8^(t8q}!
zz^uh>>7N&()k>=X@P*80VR!&F)iw0#?8QZ}2R`>GCx=MDFp<rx##{~w5oG+y$;lNr
z-`3UP2w%XNnus-=bNi}=Z>Prxafu)xj*MJmViLj-8ieQQOET=_r8lBo;=Tct9-js|
zCk|zN4DEy8BVw(8wqiPcTa+~KL2|6tD{lnUr-LZER3&kWh6ade8mW=l(IMi}f7O*;
zLT7~_an3016820Z2Exw$Gvp4ZrG4R^<xBnL%5%;@=>M7cO*IqfYyOxb<h41v6j9cw
zg78;B7{O=GAQ2*$R8&$rM<dSHLn4vL9!z%cmXmRoN~4fLC@6@CG)~lu`}X;w4M*us
zNlb_$GCj=ARSZT)^r6%hHP6I76LLtqni?CK09{{I#rN+3dR8k<zk2mCYLJQw;}rF!
zaDR;Vop0(og;riNYe0mRbLSqJae`HixqWq7Ko3h$hM<&wiD{#rA=ShCAdC$-eHzrF
zi*M03wrBa^xP~I3IT=6OvX6PM<1{s$(5f<fq`0tfRKFtKZ(rudeb5AEtdttAKTyd~
z82p6^us>0YE;git<bMD+q=3T#a^O*|YfxXk{Ps04$&;Oe_sGsxCjkZ1@ea5lWP02Q
zJcrl33$*KUe=H6sc=?1B4R!D{c!+Jt-&Y46nv^zv$z`NqhG*65NvM*a@5Z}R&c88<
zoO2MAj&L(<TaQLOZ&XA?LcyCQJo~aUI~>Y{9qD7q+scLxepOxF%8ZrX*U<oRr%^U2
zKCJ5+_KFoIPTTXVAQ!6Lk-tC}$rPB3du=jjL@hB?UeWY9J236`SFXya?PNp2p^={&
zC3lM7*)g(+Y{|A}{*!8Pp##4#>`3!uM6Z~X!JFYy+eEc*-hArp?B)2t@#F+7akEi<
zr3lE<&^rF3wF3o#ilIPAR&KER(^qAPfzQDFv(C-0#Q%){%`Y*r1-s8FTb+(dC{))?
zB~5bVTo@lTlPC=GE7ngfSn!RnVs$g}LQIS$0%moQwo8}#zC&K~JUl#jjI!_0gdKnA
z9>KcvdDj#s92r-zK<O_myQ_HQAAkHoP(2m-EbH9Cx_wWm74OtKqJ#xyk<2<56qIAs
z=(EMz`U?g_c6-XL#X{sWcBwO8TBq^O00|%)&6zRre1$=(!Wcem-$2)B^hw_gz+p}#
zrP!gqKn*=qis^9}<F}o(s@P3(_qgUK2d3{A$U}Xsym5zgJzv~V$363wcV=J&rp#dB
zMf*~E9k$)OcdJ>5Ud;EWG)cyr3tH`&b_#@vN?dc~_4!*w{%hh*m01<BR0+zFqFN?9
z=Z_?Rp8)l7TSPxyoizRv6ETzpG!3ssr%btHOl&Mtk9N}WJ1?)G(@I9>kRJl&@SMCq
z|NQ7+*4fg!vL$Y*UKw%<tI<n!U|vyb#;vwqH63mP5U6q5rRa@pim$?{6_?Hb)8AjJ
zZ(q}dO+B_Xt%DC1h8EDHn8M=RVD_;>oHo?g|DyhuL=&AbiJ0cVXKzVsP}N!Hw>(i}
zj*#-q$O+P@WO3D~RF1kNt8L(c(s49k<jCKTrVsT^PTqqI#@zh;J@ZtNs2YxuBvN3G
zZ>UJ<^$%Hou*N@C2ZKc~!OWRB%1)Uh#za*nU>Shcs2&@inVqflV~B!;xCm}x5<6>j
zbWAsSw@S&X6w2;I^o5vwjnb@)J|u|3E>)7Eqq_E77VK}puwfJ5*hwv&DPG^ufEL__
zk}$+(j)ul{5(HTOLnFW(W##3y%rnLE4=}BMFQ|r9(3*{I-R(x{;m)zq(Px5!7Fi?D
zbF0phZWwdKBMC$Wtlu{7kJ<g*j~uxRi({3%AORMIeiS@lxg#<EQ&$(yyy5LzskXw{
z)f3&=LHIzwGMhJzA{?X4hjoe=8Jdg!gOY9W%M}|obW@c*c|Z5T^khf|(iClErO%$N
zPf<I5`EnEDBJync<g{Vypu{j9A$m~Q^iu6}J|;$3ElZ&SyOa*r6|(c}C&|SKP6=qW
z_4N}b`@?L!CA|WSDh}DIxvxcW^CO((s<P!0nX=NtXod58S?KZp-8;kQHP)iAFui2I
zGlgY5Ep(BGM*+M=$Y+w=!SdSE(9zu7%&sAU2vktu5;itA-cox`p}3#QObdb|yfBq3
zfrvPQ9d&kgp4>g2*ydf}H;OR0fKIIqX--qdjjJKM2DD!#FI_(|wZCNL`}gl_Yd_-2
zfq0P0wUAZP^#Vzcxfi4V)6gKS|H^Qx*_z3&{RHV3ZeXaatRra#i&YOFmXMZ~CNfcj
z5tGg?_Oai)cU;oAu{y@2?ZWhf_17Ay%U`!Aaee_+m@vf1wo__j;^X2n;8TTA&0PUk
z;h53eUyhD0<g#?-pe+v$4$joCd;csoUUjjZgM)*;y^ojI`D=U`lx5uJhbEdPTI)-c
zvKAaVe=96(wDHrUE`j&nzQZx0uSlf3mPUoYskzvUi~U`>vW#Z}q4D@JTFg>pC!5A3
z37m7?c<1HA8aC(+aB=}(W`{Z->P|v%byXFShkK{z2Ok&~wvq2oRxz<zSAV1Gq#E9J
zYwVdeN1r~L3dqT}l5B$aH*q-ejHVgRL6k0Jyen3Wfs}wI5{XI(ru-S(DBHVx>fE_^
zX%U4-XvK5Vz$7euZmahvu3fvf7c!8zXkucb7<y5GCS9+Dh&9GRW-`>hEqAr1Pw$|}
zA?PD4MkN3w-PP4ak}0pY#Y<h>#3S;rgA`3PTC^hVdG+cQNE|kDGAS20g}Vxuzm<^i
z0q=trjbIrP5CG49K0aR1lwfzu-#!6|J^8!dC>h$)c4KO?s-`9|F4WD)xP|J{aPDHW
zB4I@-mFJ^^0ynBhAmHMX61t@L(3@ZS_%V>25VU8_La^Rr21i!mt;eD-CZ%O%QKrK)
zzQvI9gJ<a|DpFIbVMwvD53%RmTwzkoD)lrT<|<v8bC=T7yKyz#YCaNbP%4_`pm?@z
zqmFsG$ZhL$`WZ;ApwE#>9>Xu=4+;u+?8Q%>TuV%3p<C3=n>RC~+%I07n4GzJc$sVj
z8M9YH?>bJoq?Hd9MS^b7(OWa(*pSREg(Md#x*7|eVF1`D+i3;D{>TYYW#CX?`W*J`
zA%}UvPM%5$<8U&XDOQ(0Y$(=qGt##<=<lAK+|H7yZ!FM2F;G{xy=5fmuRc@<c@%CP
z>nkc+UiJ2V=xlO6W`$`~rn{F~3>~baR78C&bicBj7I7NAEU;TNmQcW{O`a@_>=9b7
zzODVk+dEA46Yh}oH`7Q{4DRsty>E_xZiA_;;R=!&D}E22-b6a;s}<_efgm+jy_t81
zbg=YlwYa5F?sJA;H+}iyoROBMnu+j9E}|j(xctgnpBzRl2cJRopnc8|FUdcXc{EUb
zF67gV8(QhbsYj2_C#w#SEh>|OivXnwPJjOB5m9T0@ATuUS=?ccz6Hrv%$xD&gH{?D
zDXI3R>Ebp#;@SP0($PG(wpa&AJu*8QJExU8Wf+p#ZZYXO6&`KI5RDxJB7Z-Z-&y29
zixgWd!u&ix2({Qg9nKATtwxJJW01ywA89>GwDGP7Yp*;r-oTzS`qt(`rE>%WzaFKG
z1DtYO*k{AfoX1U<)}#{4vyO-TlxZaB_>U`_byyy$sV641>7to#ICJd97)}UFEJHGQ
z6e`84Yb5`{RK>)$)FRdTwo*}Mbnn6kR10-(sW)$muEq4BI+fqIj{y&e(00Ur5X+Rc
z5q5!gwTo=DT)%VzA)mFHWfn#@f`F*&G<ieC5DQCVfDjbjdM~SLYduS{#6}Ge=rl30
z#9@=o4_?2n;j#p91Y=w2TRU-HKM{SVv^gy{$r@M=u%K6}Gp<RfvsOP97qhaUskW^k
zsNkILk{g98k9|m@ivky%o3hy>fXQ#$@snUF;WqVU@#jcjSNir8Ti1Wml-^uynG(q%
z$2KZn!=7x5KD23-Co@K1ZO*J0Z5K8<D5}WxvW9EnSho$Xq<Mgc|I&S~#KHJOoJWEN
z2Y<=VMT-~X4gBp^Z;4ju+P9s04hx=MX+AU$|DWpk-tgreuh^f&n)SZ-6`wbxv=-t>
z<cRx5$hy3He;qn(n5s$270c|BIXRS4+6S)1EEE?r6PdYnC&zd;w*zv8*A$}l-)?0!
zy~$~)=ZO>j><YwZPZ#7DjHz29><Hy+s<)5l$8{sJPRY?M)@LK5Pqsf%S67c#mnUb7
z9NtGi2exH@>7YT)7C(udw=A!yz}xqiJhW+{vOb$35R3hiX2q|_>*HD>AvQZ{tA)kU
zRV72VfdNs!?QB3r=G=5;-R30@Jg~?SO#c<L-9GjPbL0N=$t>$FT=w6O>5p6d=jY}*
z-ygw0KOL0)@z($OF-;I{|ND7$bTR{x|NW>6*#5sCr|;z5e}2rA`u|J*?>yKgVKKrh
zX`euJLLMksrPwGR^v(HAr1tbY`Od=94*uLu^(>gfV+rAnngtNO`9zKYcAxF?8=c+;
z2G)Y{Fi5OS>Yw?`Ki=N|{7nXitev#0VG37JrVa-6{eC&&w?3W<EC7Bi{14mN|J~w_
krT)0?_y7K1m}8gQdzpa>7j(QY2)D3kzTv#MIp&A|3!HufMF0Q*

literal 270993
zcmeFac|4X~+ctbQDixKfGBqhBQ)SB3gd#&BGa+QooVlx#k|7cmA(F~e#>^o@h-g6O
znL?(_!@DhA&+qrW|2^OLJl}t}`|7@Qa-PR=thKLw-?nYv*1GO0DadT1-a$>FP&S=C
zbNT{>LRCbe&>dU160dAkyn(+^)*8skoTe<1|3&4+J)lr_QqG<}p>i#Ju*Lb3%C`l{
z`SrKne5CQYz9CxaJ?q;;)YR*I<W?|fzkWxtlB(7Gnr$7mrj|NU$>@NOdSPtmjTFv<
zQBjOrR4QIaC2ZYmv3}RgLSWo@mCzBV!;R~9CD-c-^+mY$+S#7+{Vif=b-~fmaUJf0
zLb(&GeU@qYI|>yK|FPxoR&R}2MZP({^MCMyEYHA7%JrLNced=2_T6^APL)FWaKv?T
z@P&L>h&=-(c%OkiC92$_Eq@^{(&OO3K?<d_F~`aJY$)aU0Yl}#x3q4rG!?m#>&NlS
z<m*2#{J$EHa?3J**2cFJ{k1XA<+!w(G37L?$xudZ^J5b~W=yYJ;IR<Ivk6hT`@cWZ
zA31a|6~XJdHPhaWRdD6l57>U#BJ=f?@_qpUhfhxg*xg17r$@Wgl67^;3!5@5U3(s?
zWti6|ow&JH!2EmjJDaYGhg@xL<DWbie(5m_Sse<Rr9GT(UT?MI+4&ghKR>_4>*P4b
z%04`JG2zkU$64-klWA#bc&FRKpZX0<JmG?tKm1umWwU=SDK5<ZA(!ucY?5L2?X^`)
zHa;dv{cUFmx0axFYwnfW=*LG)6e1<uD?_-wJfAUgYfk*C|DBLpZC3aE?8Aeq_^62w
zbc((#A}J$#R#AcvH&Q54-Gx-_hXy1qqpl1#yxr4F>m0x)ZZKKUeq)2d<nQ`qJx==9
zdIgKWk_*@-XiaJ&#A(Gwj)xm`(k_Oz%^MFlXPvld-Inj({{EkyiZveV{ya8`xa-_8
zTE@<v>WR@YoQ&ktKY2M*a=wH11V_)S3$e0m^0(Zrv{n^$9NdOK`rfRJ;wo_pPYrZ!
ze<fy?Chy|lFv2j_GAV=!a`y+}*%Th}?b@?vk4L(w(=dymg?dZ2LszcLd}O>21DC|y
zVCvlDP^H5pE_`;7c_URChg`_sk>6~TMNdJ86_o1+U$Y%>_lw!M`4g-d<>`@PT1mWZ
z&9+>-zG@7yMSYTvk~c<F9Iw*tSMYuC{K1{qs`0u`Ee>*_JmOQ$c6Lvnh`UY(amYpU
z4zv|`&hyvc)g1z6r!xcCB_*=?4;&b%7X7WAcI7<Af=lJ)X3y?}s;`@!`<~(hEb-cW
z1ixy6`e<TitIy2Za~CdLSh03Bg>z`0AJJG6s~UeX)%eq~iO;9T2Su*7zu&-{%cpm9
zX)?3mDdudKH%pL74OepR*i2t!O?`V~W8=(EJ!ynsUwXZ+`M1}qNs|aADJO}=>8`Ha
z;w{u(E>pv-E$@sfLo(YuMAZzlY<u=8Mu=H7yy@<Cv#50Lf972sAwCx*$8&Z*h{bcR
zsV&=a=t)~V7K{N|Fq}L>t=A0G&Ax3NobP_1*lRxGu`sKeq%BfdU9aP$6V5PN`=S5D
zYn4}HHJ*z{>Z1pl^xXIb1O&3}!q*HS(6alZ??pH1dCc_q`ub}772n?U>r0&Cir4aI
z&(8EoE;)DJ-M`P{?9^}z=4Cc2^&%3m(SvQQxcs_LVM5klUdWI7Ej8QKNLh)p>21GT
zn`zZDTPMfcV*f>+^_abI=bb%fUtb1>hihZq!~{rWyG{)ominGmR1C;}jfc(>G-<=q
zz|xC$og8iv?aknRrqrK1UVU$dQEtMYr|+p#E2~rU7Q6c&AMAJP|NKnaZ^uI}HHn$-
z2QmR{tnOWSvcpCd!5>23_Wc@Sq+CCI(w7GF#Kp~RInrKO#$tC47k(mWIXhLbB>vK*
zulnh@==2i-^Q64VAGbF#UrbPct6zM(Ue|R9FB(jW4ZbPnDVQHy?ft2uqQaUZwlddw
z+-0KQ2J@in@J%&FCSVc6{5?@C6U!g}47KLXd9!<5`<*vsl{;Fh@bs8e@+R7n9|1kq
zo^x6gLWxKh1xpLR4H-AvIm|>$-NRC+Zwv|!o@g)(C}DZ#xwNRJZBw#8(k%s_hR7w!
zM2HNZPz!xas_B;(CyYLfc6hVv+J5BNwBzVYB#~Mf_LeNWzPJb%vtQp6XIQBy?K^iO
zO<p%hvuw(k9BQIv7PLUR8W<R;4CO883DLR{G@-00E}(alS0m*)LYQvfxoJe`k;8{Q
z7H9g_)9nj){9)SWF=tt|=qxdD7{$+h=JnoCgeu2E%_Dw8gGrJ3k>Upijj1Mz`ntzX
z3^ipgVL+eu9qs!RHuYW0S}HTrx%Y9F=EY=P*1R}p&URN_jQa+jcRRDiT@JWl*bNs|
z&V>t)dNJwVv0^`BTvhe>NKDi3>|aU!rNWEzeUeT;zYbuHXcc(4OG}s1O3w44CX4>g
z9e?J{R#7hus6s`GK7CBiaQ3e->->-3gl)UaA3bFjFbhsDB+=Rbtq?O;d40tygGp;-
z9z^%n?K4q&ei83_Dnn(21hJ?bhkj)3h;*N9?2p+0)V^*YE;1s1bbh=hSu@=%sWi`Z
zDj%PawIj#rH(Tm!4Di91?PnkClM-6&6)F&!?T@+1bC*pYVYuFLEroN#hkgo0Njt-;
zWk+7VFg-Q5R>q;;i@Q$DT`8ORrl!ZO{$^+ID@+uRPWI7kaS@R$hhN|CZq${Yl&1AM
z85eF_fy{KowC=gr2BsvfOdhOC<WA8dyh=A-)W6OBdx8ePB-Q4HB<YC=zGq5Tex#Xd
zWmtUtw#_aw5{r*lH&^(IG-AusFk}BaBPB=2N%yIi&@+mZ>nW#}SAkJ))nhC;KdYMR
zY6W?DKdS)UXDnraH9PtS5!VJ8LRQVffrYfQ+w$CI#>gT-F3q3)8Nch)Z3HDl+*3zQ
zWCU~E$g`7Pr=?#abnLrco{x!U=FJ&QDUZ8*F;VjmhA&{4ahx@DpZt^4Lf_-nctsnU
zv+N8Rn{kb(tkX{<Jv}~8&Qd5wAtZ75R3bUwP{6_wCE57EXrbqHhqtiF&7jZy0$+`k
z5V&!QkrGzT?^d^|+<9eC;#2E~RNtIosg+bpay@RWgO@jVqzKiz?`JBO@4%NhnS#^#
ztr~ZeQEKnnYrVUwe?D(F<^74j<G|t4HTLFF{MT_K1fos#@Q-vZW7qDK{*|11`EL3=
zlE=?IJ{*hmU&23D{Z!PD=U_l{eJt*wmQ~kfAhxQiO4|=*{!r16pI=`+IFpHl#n^Fg
zC546MZvfPN`EpZ%VR3uJX#D{yaoMBX+}weM4<0-)%mCC753HoUjKIQ0%}4ztkq!I9
zTz|fb11>m7PmKW`LiY5_H!HfezSggHL(_&mg`do}&aYoV;Z$D!_=BttU(c@*W;Xf9
z?RmeBjO|g}3HeMlJ$HwyNi6sl&!q)KlPm_V@*!8~gk*r^;<fxzJrsqU-}%fftp%P-
zN}LlpZ3{`L=~)WPak)qT%a<=*p1o1R^oyI=fQJLf+{nfrsTST--H2sUVxgDk;?Ny>
z;#bej45i<gP=jx*?~VZqv^<K4@F?AGA~xAzh;$|^CwI@S$jontz}Vc@8i_dz`z{7G
z?en;LYo4o2q)!&TxJ=UAL&Jf(*t0WCb_`+^%H7Q<BgbFalVPxGIoA6M*JJgZAFUmF
z3|z&wa|S4&l%?Yqox&Zf1dUW<q@!`IzI^vNgN(T6&%2#o7KwJ>KSd+beZPMaDXC1G
zmW|p=L2_YYKEJ|u`;q*);T&y06x*E9Qsx|oUm@}CKtm7qE7I&R!-}Z&BbDoO)CrIH
zm6WLW%Y&@x$10kEV9wfsA}wHC_2|(86OZ<zU$rqvrFR-rk)|2AF8W%WRZw6Pb+pKH
z`7@@~4X|gB!R|Wvc82x!NVJg3$1Q10ZYT*GHph6MB#==`h}8L7Yqs2(-Y}cV<DRGz
z(ae}6CxDW@>Su9_RJ3gK?^n_8ZD6Ndzp`A=6BUFzZt)h(rvR*x%Hl=MKqEOn5}%wu
zdyhB&4u+GST6}4KJWG?T)JNiqVxu3X+B^#o3CMU@Am7(8X^$YYO#-k|q}C3Vl8P1K
zF*|S~;3W4xtxT&=RaL%L)<A9(R%4_>hC6hQ%7+W{yTp1XIqkFc1x~6KMv7L?y=G%S
z$DP@H_{!&~X#oV@AkZSh;?Dy`34%;|RS#1r{}?1^0d~uJmHIK4HEM{~;omW*I)Lpg
z8dGAi9%~l>9)*E=i`2HIX;hfTJ)K1D=`+jy`8uAgFR?PWE<d5Io`1@P7*bMF7!Rp$
z&a;^ZlsatJTQ&E)U<p@iL@L=L!(ua9!qkQ&S9@=PF3qsaAC;@wvB`2Y2lbc-*g+qF
zYpT`tcbe%J+w}b$9O4TTU)*OWoIAXjdOw9_BFs=>zhaDVTgcWHUtWCDFnNk;h`+%=
zOF7Q7JOOl8`7^!vy>2tC{BU8LP5_YeZpm|vCXq(Q#%qV3q%7s3O7KS(xD386W4t!x
zd@e@XFZ<7Ttp`CtofD6RZ4546yf|;u>M=KDf?9*}=qK?F^b!D({<SPP%_Tx3IUkn#
zvof>ZquP!vOhEhi*O885Wo2bLmOft38Mzzc#TgcjW0OtReN(SH$B|`uQC|wHU%h&@
zG+SRlJ$aJrV#1xQROjcbCC^@+8_Fy|IIQ7XMY&!?x4gV2aKk$GwNfW&M|_W>%&b_o
zcJ)&o5G#jz9VWZ7kuw?^8aEMt$Lz1<Z`kCumX<C2%+om;R9_V7ch2LzVy-V<yohie
z&ccN_N6ZQ4wy8!O#Q{|)UJg%4NT}q=8T1z&E-x?VyvmQ_{Af9e#BV=h9FE`Hz<6lg
z&-)=En4wG^_dnmTH0~X#M(uX4m10`J1D#+6{;EYfMGj|3s794`8v604jNS84Lywfz
z%vevNX8J6G7lV4lq#A+4*ykE{hgM>GXLZHY>d@%E^6KULAxnVYuyhc4y)Zt*vT5>X
ze6lVeE$Dz^XYv|MELFu?FX#p<+@@NbEPtelbom}3Fb@GCfY=781l`jH_QuW4RB$C*
zG{55yv_T-{b@_`XVIlTKdM@7GAE6F-i=aRq2gkfJ*8LG!xeelg{?Bkp&qbRuR;Pj3
zP{-e`3ts`Z03{Jv7jH1XNY=^8IOR?ARMd&Ui>6HLwx|;~0Sql$bDhBy5$2B%8_7;|
zDQ?&)D(d?6<@r4_{=O1~GOir_l7eN#Y@Iubi5Nf@&T$+fnQr#auX=pC%b%YpEkJ`N
zW@bOJ-qn({XFdtFRpC+}gE*Xr^Jpl?BZ5*=Q{}SP0ws=qK1C}IB*&q%hQ)&CUHc8-
zOR3DM-)&2n@`j~9_j$1vkRfzXOYG7W!-e1C**Pw3eRKIEp?81v5yI8LJ4{Up8o6s#
zQ}53M)_sVev*R|V-(G*?FJpNSZ-b9ucb_<aHW;b?#GP8a$EOlYm#IbAW#USX`ASOg
zZnE62@4kp}$8b4xez+^{GEweeh9WJPPDSC=36jf3X1RXz19SrGcOAE<-EY&f=VSC2
zlxe|-i_FOT7cMuZ7<tZA>d9t9*cy`|$mfmZ^A|6`CtSf7Idq^d%<LgD0;eM|?hsad
z`nA!{AfDBf>u#w+xNVltWUW4b{yZq@#*GFPYb;hWQ?Hc4c8@vyZq4J+*+QYL*|q#3
zJHWV*uv*oLbfomb;wMqhkOxU2!Pn+9<PPxlpBI*g;eU5LmP4YwT@4FS%PK4OIU&Nv
zym=FK0E2P}02&1ZRa663!fy}<BHvxV**HHlUeZsYrwb$D^gomj|2HJa|KC1C$!8P0
zgxG*5a^S_fo?sDVmb0@3%QC0bin%6%dTqKH*C^;G;Eh}Z(Ud4oj!=87%n~D@GjOeH
zpsnvoIyp!SkhYFF{A6+Ie+F%c@Gz(*xy3Z13D3=MT_6N8v&B*CvK-Qvb@srEpx55J
zNkK-Q_L!)ZxyQt-`_@1TIcc~hCtl^1ZTT+mqkz7^a;Vh#BSjmKZ^PTK<NxRTAqxP`
z{LY^v86VXRi3PMk!0anB^NIM2iNKr<sU|hZmkhshJ2`JE(=fCo02VBQnV}%}x@93z
z7T%;H<gY|>6{SHhWd+5j{3fl$w(;M7sQbVBWP58OJpoxiEOEUA(-CUET8sobWoJpB
zh|=Tr+tkd=>};rVh8V$BiyE7l1rHh51n!Y3oi|ZO4MOQIalQ+=Kw7Uynkf>64>c$X
z;GWPIPp@9#zHmE69_?)=??DqbvL2gw{y;(YavBr`2_#$4@Hxx;MkCvb#uC0?gM)*B
za>wdb2@73(te}n*MfX^NlU_6B@<)%PP1ozHU+{pr7?B}vwv_`+L%RPW=IX!wNG~)4
z<fEU9uej-{<)O9BKb`_C9n7rfK1o$b&$)H`_U!=@vsW%x0zK6M_PcEg3JD1@-z@vR
zE>;eeV|MmevdQJkd+R()eYXQJvZCrKc>@f!*wu*Zc`i8Yj=`%#&>1aSvi|_c(Ivk7
zT+tqU3OBCAX%CRmXO^|JIJZG}H-+UA65jhlL)<^W()?W85}TFjkF)_y=WGd+9x0*X
z;_gE~(lK{z!dEZhJG*0p%~`1+RzzTxdwH2N5fEsuD?m~@L)-3?0H(bBw=m=(jdU~B
z+HWO`A&(z7;wcm*W}d%Xs|_@Ybc|p)ud<3U>#%T-(#ljK1QB_%@i5*d)>gOndL0pf
zSj1+heoLazhj(>%cRxGj&7+ah4O#$j{|#y3Nl;K7SZpfT3NX^b;@odSXTu{1AX>ZW
zN2Gb(v$V>J3LlT){y+74de@&FK74o>Tq>H=XX}H94;`RNF?L)hwPeL2zt`V8@n7WN
zbDJLNOjt*mWJfK$dViawhugHN@yCxJp$As9IJN13L_d1^>s!2#^H?{KhDbA+BX*Bz
zeNTGo*fs)MEeGtzK0*u{`?!XA=%JeK3{*R1dWZ9u0XJ-beDgqUB*7`LMkfawp+r6j
zg?2abMKPe!v+%|m=S6BuOdLiu|5D)%gl%kG9Hfn(c*;b*0*|U{0E^~yb2a_AKz5_6
z70;q0{$Q%K5#r?`So1UQ*8i1RuJWA$t80GfP=0?G1Vft$HPE})sD>hCh{w4u_9b=g
zKZS)oKn>;JAWVhQUrR$^Lz;=d%5eTDa&djUS~7GENFPWC?Bd8D{tyH412t%?FJt_(
zu~?xU5xi8II|+r%`tuuHX%v7t1Gg@@G*g32k%ToSD=XVo>PJPxG99@zABQzx8#)MP
z2~9#u@a-k?6b4~L>&b>@(-`#dp@7Hy^zI}!itG)5QQ>W5NkbemuO2-Bk_?2l*H-ny
zg*u2@M8+TQ0knE++f&hb@Ga+M6BEyoLMjjj)7IP>ETId?K6C(=ik`Dyuz=*U(~4x}
z<rfg@tq=kscL1uzob;l0`1w@?laid(`$SFG*$8BFX2%d#d3MKbX2RwmKdXS~zw7eP
zS)Bbv*A@+!o31FiV4tj;_vEPQ1&lwkn*j7FtUf19Gr0Z3wpR=vbt<c?BP85s-~Y2p
zQKTmrYIrX65C9-?kYhqqkZ*f4pP&2zdNeWc#ZF)nNQuv|`VoIj`Kk|CH>71Sh+BWK
zPaauc6#)o|oirDUHt-_I_9IVgmyZhl?KU6)d*2k(AYWeqq2YnTGTK#UVr;xmxQO6i
zBp+OP2>(h%9|Y3^B1pm#T@<4G&x=}mn$<>^m?!5!OpIIO0C8m=)NT8?3bL!cAxaK_
zohZhtZE-S_KbiC@kad{Gzor;fZsp8+^KYJc6Cv)}YxWq!f?nVXpVik_$q7=9Z*j)V
z7%t}3;z>I_NRl}=Z!5A?7V)l<w?NEo2!J|Tbbi>Wpi;!O?(F^&cC(7$10ByWL}BI1
zK;iP;)~Q7LcKi57zqn`!Wd_7v7&O2@_hERr8?<u5?HDdzP(Jfna&a_ZI^abD*4P(C
z$)|yVTih*^e%ImVlizjvMA;-Eo`l<C5%7d}LpM5m_N<N1z_<8%(DtoIPi|>Ve**z?
zBv@JDuJQ$Rx<YyNBFZzf&z$C!vTk^zKQAZk&$2ykkp=S(mAS$JHE2@|AzP)MSy17)
z16hr;?FSA!4L1vS*@J7$z6RRwt9pqb;rF&|Oz8pA?^2saMP6-`pieqb-GMStp}J_(
zmyC>z9&#}mihqX?LM?U)mx!if5xaJ^_M$FXERAcP{sEGY_Vn@qm!6C3MJ6d-GGWL7
z!e3C{N}<{mU%TS^1#6UI$bKJ1l2Eb{rDG1xAdW?4Pz5-PihMppWJdBFjQaBxiVcY~
zRW+S-aD@eJ^$tJ~xA)gqj54*Nu>1jkjYO+!{u%C%v|>gWXKIo4T_JF<dD`+u{}hm%
zadr4(R;P|`u;5mxJbuhVrM&x8PCE@{^+o(b^v)+1_f<Xqf?MUypWfZu_*a((1b<}G
z=<EVf#Wvj@Eip=0w`@Pb?WDY{;pL*I_BshAcep;e4Oxyf-NZy|!bMU+#ZKMcc-ONm
zW)q7rdN}`_Xj13Z$xg%T$Zq$X1^4Xf0%<2AizNO`C#PUpQn6iYM}O`I?l`}K%L?hO
z8dLX8J>P8x<R$&ms(l<BCCxm!$R%r--;_TrJm5`z>=%l4nnmMbHNy$sO%zTCRLtYA
zrjX!o1vDWIqsQi^L!Nl#!I^@E$tGkxMk5=kZd`l;^#<8e=+gV^Gvj?rsM~-=$fvmB
zOPCwG*>e$M&dAOjS|%Qe+F4`Zs-HwihY<^bh)$5Wez)WZp$$^v!vN`vXhn3sA!|47
zn84*hb(&p%C64#iD9X#Xfbc+)9Y%T@aT<U$vozfm&=v3nAhM&_+Y@{rOJ$E%1<QYV
z0nR8Y0-VfsmfVYU`T1&3fY{-<Vypyo+m^(B9hlb%gJuE}8vY<v?WNK)L<M2V?`#aA
zEFkn=C`Pii${LOLegfNTNHLO&P-SS9VRsLX=cw`e#{XyF%l_Qh3;rl{)*xPMcn)!K
z#XJ3O1-O)-4?>Tm1*{xb^Xv>28!O*~MCH?Ff^^F7<O-ELADYrujWBjW&&7Gjb0Otx
zdHK<-_zm)W*XlR4Lj2|iu|a^p80Ihm^NZ~(?kQ~#w|#jv)ql~8bR2faZ=}e|u1~&A
zMWNZvk^V}qd$yB`-R%u!&*`MZx%Qj1wokc(SkK#@mN4b6(GlLANwwbVBw!#4oPvEO
zF5s=)8yp-Acq);tcvgOwBEe(TN<9@1=g`c&nT_)709XrPeplk*uN)Putw%Xjr$gLP
zL_fGa1oA{voM@#di~zD10J}E?(51HT&NWA;c!M>wEwpl8?RR{}s!+gBUfbYho|LsO
z`_$ztuEm>#R)69-!5_&^o?}|AWTQW6uaxg!)wd1xr`v1h$5#&HpP^A;wbHcP?36=S
ztG-q-MOw~UOY#S`;14M#o#4nEl?g{Qi=jF|T}y*Jo1Kf{A~CUsO{|wnKIp*%)C-kc
z%H_#GK7+fT{5eu1_TX0$a*~X7qY4Z7@bS^NJ6I{4e~92EA<_HDkecr#YIt!ObXT4m
zPkahdRBI7+{gZ%TqTw2sFJC_0$3&GfV#tPN;q+a@WQ{{SgiK`L7J#?u7!D2&BDo=%
z`A9f?5~#m5#ThARsaeHQ<IU>WfD&Gd&RWzc-Xn-+;MF&C0iN@FJj9<BwWK4mj+!*x
zr)D^qi;y}&yDAqYxVYx)Q%GW{g<}b?`SS5ff9XozXL-;L?JVytmG22aI*Fm2B8Be5
zoh`LP$iJTGl#xFZ6BDQ>;#gz+Q%Bj6Svv#PZt^<0YVF49-Y{!|%1~*Ut={d*E=DD@
zwcMJHjx3%+dddRT5xwCQq<b`5302sBc<<~U0qeX!7Xu}42q;3F(z_-a1zkrFTI&<@
zBn>n&lB`IpC|tyj0H_;zvS{-_#G!BT0CYmLCHayeqdE!UZN@FAjoRFhmmvMC2C8nP
zaGU@Z4t5D_dh@~ag3*2Ok=p^DGd-cZU%q<D>f9Z;ii&3Av8H6bf~)=_5D4*P3#w}R
zUaP)9TYo+h<ABA2>UP4UxF$$mQE>?=iF$G?9zG9M)y{vs`D_87lO-4+@U3k1C>n2_
zl6>PptajDIoRFp5$3F>;KRNv@K{xN1nw+hzZIw<0-FaYmkTsbM3vh*3aS;dKD-k3#
zq38i-1F{AmLjiD(7zanJO;@Ss;&@~mGF3m-{v^%xgQHIyp()iu>tjK?0MrL8V-%2w
z7gAQg6_Bt|cPLyvkZ_!mBp~q4Lu8zr7_e*5_f~zW)Io43+(xbB1HF%uG`Sva_jxq$
z-G|ca^QSxSLHJ?3b|16lfpwO(1H;QU&h5QbZEYM&#ZrQy$DO^iYE+b9A{0k`m6Ma(
zz|M5Y&o9>u(x#M>^7))jA1<tqo7Ev)YQt!dx>wt1YHH4xoo7?parDZ4J3gePNRQc9
z_hWP%8;xwLrL9_&A06MH`<nPbfa?vzBfPTAp>=NL^(U#UGgJkv4a=yTMMmP_&NBAJ
z5$_R`^Q86d`BF@(?FIBdskpkk%tx;%O_%K~LnUIE+45j@9@u)VAIXy4Ci1rS!_6~3
zTuFC^!`qcNV+pe5xt^<v%mbto8LK3=gY23Qcj;n0&7ay;iv%`N^-N>2(}uIw*|{O!
zflauoM1{Z{HcGPYI&|QG>HPGl46B1m$7?P|0>9^Lj`hx&l`NdtPT{m#9+-O@pCN^4
zT;H-=DDb6jp39c6DqnwOTC*3N!qXI$)v}jLUU~{$S%iupa|8qt)Z8o+0_8qu>=&UC
zpZGjtqu*IxZUc}{7iXeBtSP)nQ9(hXCs=tb;UEiRvijTIHPv2QW|CC>AGto?n*?B#
z#TE~#nX&b#<Ox!=J(uR%`X?XRI4ikM#3&p7#HyHH|5m?N-@Sin@sVZ2n+|jqYQ8-K
zp9uUteA;Kev7Bg~{ZnUTo54XG0gm)QY>-Jm{!S-g4t|gwEfQdOcpe5Rhmz=x3L?QZ
z8fEiOmEL@MLTP`$<=wHhl&G|qU`f;0fFssy+cowYO+Tg+@>1M20d+D~3VFx`20U@)
z6N8Q8;B0ztE}h_{pZ)E*)Pz7o2igI7A0ULZUBNsKs=T~EePuM0!T+3N^wjq|6GNG8
zBvnCHp9W<m3mGEKBiC2nV?JFQ?CMF#hTnudU0#uzntGGQmVPSfbS*PFX3yzutAMRM
zkjh9E$pkd)1uj#ZKTpNalr3oXm9t-FxA8S{dnm9{GTm7)5(s9$#VytPK~w1Hf)*|9
zq@HDKG||g#!}2Cau3++qIp{L_?k+bLriPlFD!IwW?wkF3f7huZ*9G=+0_WG)95V@c
zrQNyoea-b%k>3dVVR{_QV3F!yE>OZTb}Lh?C#%o3dvX|fyhK|e@Bk=7lXXE0O4L<3
zs7JvEl-JU*GysVpcR^QohejM|bD4!dHSCeCRE<8DS`y)R(4n1|7N!)@VF5ORe!Z#U
z^QpHX#xe)1bncxio{}n`jU18S&MH0BKy`R4(}KLh$nb~rKyK)BpUxNrPIBqGCyZ{(
z)|PF3ZMG=UG0dqcDZAOzHa`c}to3uw4^`S*!{aaVsdF6sr)S<5Fng%stsyQ)tOw}!
zvyOjCgfP6PP+k%>zz)e-nyLcC0O{yXE1!wyL84m~Cm1HFC!^6&8@z$Sae6sBqo`4{
zGhLE=fSQ2r!@fRYzzN&5RWcEB7jN2jtRnq3@O%NN8N_U)b~*=g7YbD}!d3ps(YsmW
z4z)zrX-YF4M}~%52i>ReROAtxl)04w8hfLK#(09-^FgIMhFa|Af9Wj&3no)b+6Las
z(wB-bZ@71QKGV`i5mqNq!~pTh2GlL&-KB6LYu&0($-37%yu`eJp#Ki3rwuDrdkC5a
z>Uj<z>%AA9C%OJgav-$~E$Vw~0@q&OO8k+sc{HrYZUunlGvy7Oc7ua!6K|X)Ds?v_
z{|q4uEA^Q5BEr6%9_ArLdu4&rsQj*cct+Z>)^NEdQo8N?KNA;;ZR{k9HKC;7M~Tza
z<yBSCRJe-c-7)tf_$8t&0{zzOd7@I$CNKJDQ8}H%7<5Dy6M!lz>El{xi&SQp4q(Hv
z*tCAV3+V5jf+}D}-~zf_!QP)^!W-TjdV~fA1^HMj#_gVYRJmfzs*$vv?C63(cMdIj
z5h)J#we5Muu8<rS;l0p<a3!UxAGkaS2H#1$*vfbD+X_nTJ?Iadf8afG0lMmCiF@%v
zns@e|HIPL|7ZUCM>{HirZi_Lk+n=)MbYb`Q7U^Lw(w);bp}lorZ!}u<tfGztv&6}V
zWxy(hCwvP0Argmfj&)W~$V#y<yeCMPHrw8`dK5a<68@0pDC#J25u*RHCv)(OTTOUT
zb2_Q0eU}i}{GpB^!Tlzt#>U2_=RS?aBi*#*I0aZd<~NkS!Msq`e8mXAb>gf`mBTRU
zK96!A4AAHh=?j05Jar}l9v_2)yak#W7Ta5<LA*aWpXxy#wXRQU0fVqjuTRu+LiA@@
zEv$Hn{HPM8;k`0K^(rJFCRJ66((Yx&@DdnTM;EBY{`}`Ih`0`cYQW<h*p#B@Zb^QN
z=qqN1ED_%RtG6Egt}drX@Q+NtgY=M9&9)AxbF8P5n1zrHAuc;9R6?d4pO>I-ARNN(
zeo)XDnr|g{mFbok8G4hR;+93Ikf|_IXGR#NZUTDpXOH5hd4mcguXWzF?OCi@dbamn
zbslV>kcYJG=ej>WFq|amHHNw7*sp}sy78At2floNMeifjX26x;zI0UlxiZN`e#Bnq
zfyA98i26%`{Ccw8^}(wU-V5+yJc`c)ib3cv3|V_>O>i)#6SZ*B<msMSrXnKJK0o^)
zI$<MaQj|!BeB1ag9Tz^ZI9AC^TD<$Nb%5h;(`|2t+GH3YE|{%29n%lZygAKO`Lhxb
z7CeWV^P%^&zlJ9dDZ@9&66XncD94)@Y)3>rzySd@?^XKLYaCk~>b(N?u&>ohEBZyG
zk8Mk}QE(KP-7sYAO~Tl6=cdm>%-ZT!ve1qU=hrI0>5^wz7=E>AJ+ZHCaI~Dhxgn4Z
zps_#lF+#D#xAvZ^!w9Sf_v1Ohm(eu!P3@Y&jFbhidtzy|T`k^(TlF2j4~2lf2Z(}K
z-3u~QHVQ`pkk)hyNin|L_;+V70GIhRE|CcZ*`L_2qX0nx{b!`Sqk2*yVA8T56N-xh
znAeN``KsKt|2O4o`?%g=O8X@x<VFONcL7-^#JMY$oIRA$RP@Wu{Oik65(k5fGoNb@
zD8E?kCkVTk7*oi__>0_b;eACTXE3G3M2d<^MK0TryqJT8VkiFci2bE=&))B5|GIsA
zz+WmFd|reKg~cb;p$etx>GY2gtS7|W)L`S0zH4aR!klz&L~810ziJa8mGk*ci)d*`
zFH!b)qBVXee*_AxDHaH-g%A`zP}zT*1kv|+d~~}#s`$%pNlq!9vw{LJ5}?rlqz_e%
zAHNE-U~FtG=&LozvRufa?GThfyaD#EVmcYV<<L9-0E@b{o=!8<${nAx5g-#yX(CKR
zi3)<322NU(ZW5deb5q3fQRTkmI?=zSMHnh6(!8|wIha%9&?q|n`*?aJQ7wE-z6VAN
zDH7l9y#yBu=Yg3YzO<B`{$Rm;y=}}87TP4&Ky2JUAI)%3Wx7&tF$K@`NL<Le)eTah
zG~*1a%i2vjKoTG_z4oH$gQKU!++qSnF0eA~)U(e*ZEkce+_`g(ob)^R4NlJwY(&Gw
z_B~9(utbYzlf+L?6RLdS0=e;3Ce+S9;j$!>Fd7Z39UwpaMx0DDpanFi&APil4(P!~
zm&-PE2;iQlzOg)a4sX0CQiNOHy!iQ)KN0>E#JzIO!DbSnZoaDlXXvWp;o>6JHsre|
zs5(i?1mf1)fk%JXzW2n^beZJIB{k>^oqAg}PBrK_wJ@aGRmJU)rp~UtD;ae32f#@*
z)gY{(_BEh~GE>_5ZqD{K(7bwnPNL7w@UE!IDu;X4!BAo-3i*4~TIN1v)cp57s>!+w
zFgf_LrSj2Q4ufbO2o2HjMtwgpmRG}q<iV@{YUUf}ZS{Wig#I70+|4osQ4RWV;a!+_
z8a6Qk)Y%VTE{9&6js*h*0d_&eQ3Qe#x&x4?gTJH#Z`6p5SH*_v?l@`zUse)y>hLb8
zY#F9sj8TpVUWTHc3AYJwCF}$c%~}EbU`&;or-n?zsGxp(<8~~MWRSK4fr;mNU>gQo
z#Ba92sY;?=Q<He#m+cK!If)qO4x<3j**0sB@vv2P)co$SwGx9XjMD^~2Ou>733M<7
z!b`%AE))O&Q31cqgxgi$ZPm$sIq~xwSGc((JZ8iLjC@^yk;PM6L7K<s<_)G)Rzpt2
z>e!!eoORzR5v!%6ug&=K<z_$>BtMtotiF&#8e;&yD0UE6d9*SH2~azF<@3J|S`Z{j
zQPG^wZN6eeI+jyV@VC%_`=LractbA~HNS#nh5MK(F_*63)q0~aVIO$!HS-64t<(8U
zPM^=n%I<)M1z*znI<G-V!i7E<i$mJcF7P5`9r8uqBe_-AJJsq$Bq&!$I7h$r!=Qw`
z&5AtV>A68QQB$D$*aT7id$O1ZAwMu@>B&~bHsiP8XjcnAv0F_@Cdphk)mXmjxj<b{
zs9wb5$IizgD|N^>jemV-)AdMh=RGq47i4`RRhw*wVBuGFuPa%acYe8f089Thd6(VM
z^|DMeW)qOC`~AwjH*Vh==b3*o#>B*g77|1?OILG0-l*d`lz!JUry`Vh2_*msw}d6J
zT#fuzTwEMUDz%jkT~PFj2n|JZNh-i`=b;Np7ZZO#f!T7j_e9nXcS*%8m8xThX2l24
zFUz^)z<3>~pPrqF$~R?gZO71e>8*a+1Sxk~98)g|eG@ePgN8TTMJzSsq1ECw+}Kwz
zJ<<Urs=k{+V!Hh%5RL3}lcbfTn+oN*Oe`!sT<qFNJJ$7Ev~z8wIU5hqmm<E8fnRs(
zPK%&(jIsdrao5XJfK-I=Yj1el#W}CD!RsW#aA7dDW~}Nq=~F>-MKkx((qL{glm~5Y
z&^PX)QL0TlzCuY|sSVE4g9<MH^w=LXbt_Wsz*dQH01;A1@0sJS*yqnt!wiV6P^(HN
zUMq8$ury#ybdQM@wlMV05}oR;*J{ZazszOg`YGj}76hCA>*0UUQfR{`m4~3p=y)#t
zDbZl14DP?XC<&9ofjSHl+`}GFQU34NJa8e|QGgt?@X6Z44slK_N9f$hxmCsZA43yH
z3Oacw8ZncIck~NlArlbBj-^etaocCq`PF5wAR2?PwW4UE0Ptw0^}`qZ14g3VVbNcc
z^@!<i2688)`W%>LfDkzL?`OC9umy3F2M0DFpmV$8t{(v03z17`*nT6HMf@86o(}Aa
z7tG*pB0Xo+3J6mxH|kP6=eyX!s72tfBzPTp2^}H42XtH?-vLr-T|Z3iS%31`=d=5t
zPLRGVN+E2|#7I;vWQN8Xj2*CU7UIWWAwjc=yA0OIFzLeO1}dZhe)_HleK*bAYoo-i
zwYQ7-p_UEg`{j1)xDitC1`DI77#1l}oj|L=7KwcdXqhyJ;eUXdnTjPUrE0mt9)4fM
z1G3puG~w_N`F}-vtEP-Z()L3i4iYq~Y7-PA0_p4NHrPTNvyNt78u6B-Pk2q{!1qYC
z`j{ki);R!lc)!RFo#DpRILtY;G{m13c2Pi<@tABhaT;vkuWnK(c}~!?i?KMmWti>L
z8R0<wWo)h%g+I{`lfB?LY8|nT61;^L`4rHer3FJhZe6h8xKywF1O1nIbS#ytZP0ha
z6Ut?<Z9VTe{G)AYaSEOpj+7qKUjPl%yznPj9-N37BIac|Oh9n}oiHcC0J1d+=&Da+
zsAMu=)Q<y8m`l7KPTRA$&JMn^_4$G8DI&|pr*0Vb7i=w4j{r{~7l*2d8+}1o?Jh`w
zcu<&;$0J2_fY{65;S{cT_{?db&dAfapQ4mN+)~H$v3$fv-mePeMt6e<+GoEhuSL(A
zG#vuD(Rn=$XeReMVj=3^@i6#sN1K~S>UULEQc8%tNumFMZ3AJy&PEh_)3!nQ!$*9F
z|4nEt#~s_RdVd)AwJSruJIIl$i?s}FzekH${~)RbtB6ySx|eU{L@=+OM7?qUuli(9
ztS8ym5h?p@%?FVn0$4=_4;*;4`d1_EzfVFgkBTx~3AgQh53lEeLKkcp{A;8+Zu;%5
zVHwhFP*Sw7&p9=A0A^w#Tz#)m^53^(@5Liz<+ExlC?(5QF{Rl<>TjZ+%Qdu--!O9c
zPPrQO_SJ@EtJ3n0N0z8z^Z^E!USZLk4Lx<TS%5+jo?f=pp(S&lcrNj^@-Fe%{pSTG
zTe5GIWgh$N=8+;FVo&G$LJKfV4xN=0DsJ#Br9on>MLQ-GKZAvD^RlDyzqgxnd5_`M
zk$%8FqzpLD+Ar~$REKA_x^SiR{6oct+$w$c+bWiGKOtn-aZr>h_8OU(gxX{C7-fm<
zIb8&<l+HkJIfHaDRpDbadMvL)Chk8|`JYH8VetQbAuqDKI<am0k_Da^<a=m$$eF=S
zJIvpy6iZn!>`F}BDr$M)HFs=nYY1cP_iUBl+@aaeuf%`-A*ZF47P3`k=<U~#ttof*
z2HhDPpEJyN^04a*H!(ik^!U-#ujJA2KC!;>s=MKRZe!tn2-WTvwO4}uNy52qJ~Lh;
zSzKJ~r*aLb4UX~&xf?fbd^Jmd>2c;vIY5jBBRhUCN=A6&8uEOiuk;G?BO}{@ZK~RO
zfajz$;B71Y8AX2Rd}!HDdwd15d8Vqem6a6=RH~{uesp~WBXJ&6qON0iDIiMQzNtOf
zV|XVZATyt&z~hXm%NdZf9c?ADQ%C+uO2-FB@)Mt<aQR}THSAp&>geFVT)z2N6Zxpq
zXW*1a1}JSWyotS_myGswbaVi^q@2Yr70UHPb<39C;4Ry>HN?o!axv>mk^SEd%%(6N
z!_WNo@cQAN9tH*mqxehY1&!x_Z(!fSgRkM!Vf@{SolGFXKls@2)Bn1QlIfWlbh*nq
zTQq9PfY2OAvlLegI;7yG_4o$aNL_XM-<JTMx_<w*F7&Uht(B0F5DXwc<@>KUh>42>
z&s{3?Pv|Ee<g%T9#Cd#NpK<SpRjXEk=j#Vt!V8pD|K8UjF|k}2nvH%K!8u@-{zGAu
z{B-u;L3ostGly}gOxwUke6sKLn~(1BtAnkdLb>X&Um+aZ-fljAc=N`MxCdJA5dzES
z0&JzDYiVvC+_4qE6;)!fJT4XHP$^KjKP8A6d{}2o#_;O4ze5>bg!VvH!bcw-^5WjV
z6WUJCqnFRJb*n*&CWH<+F}kl`!;dN4#>=yFl^U7@HuR-`G=tGMXw?lebM}94`YPi-
zHFFy~JLY5dFAbM}kf#2X4C?jNckXkJaP#>#KZiuOOG2VGa2I)F74dLW+7BT_i;9YT
zRoZ)^@%O7@|K8u7z|&dY-etH;mhBtygQ(rhWBUFTdbOt$FIHMn)6yoZ#)sVOA|Lke
zP3->XpMSP&*^&~bzC58mf9K(SR{4v{yxiRA=x7#p@>x5V-t49HAJR<Y<lvAR;KWUC
zIB@8YO8mEyl9E|3Me#ph0ZkKB&S2~Xi1{QiFi=)zMNgLL(H*gdl=iPDPoBg;CD_yB
zv+(7QCUF8B2VAUcg3F@GfG&Tn3tWCe^8N8Uf8(BffAVh(kZ(8teJ1kl_04}nnS6ii
z@6(WPcm9n(@-1c8-;p8T{^z6rM+0(SbAY^`vaPGPm$<BnT6Og3QDNbC=qY-5c<Act
zVn=@k_G&69<X`P5!rYS1+Ml1Qu{XKPAn9yoWc&`QfbA-<%5v`9`M##6rm0DE|Ni}4
zTwK`LPV@}maInhpv9VwpMqG=g=Wn9EuDoIW`YgAZl)CeXejlXABQ~8Mh;D>EFuh-w
zU>PO*W0AEcChT3yCNGbSjs2{RDZ`et1|%^wU!J7OL1h5n6Gzx#+t8zg@UXC7zkYG$
zyn6TU-AnStDJdy=Wcf=#;LqP)K2deCfn7^QMJ1ye_apiU>L)xB5B*mv9A8C|7O-yB
z9?(;vP~Nfd@+L#W>gnmxxH^QV7==+Uc=*K3W;!~<M#=$h?nG$n`<iM#eM(PQXl`zX
zRp#{R)57g?_*ChtCjylz@gfm?3Cqh>nCwLjmvDDM{zTej92_1RYQeUeFqt*(SgYVR
zcl`Y`l;`sE^NC#m#w9o%p|e`Y#o*K4pZmL3_Sb=8c9jKq-@3)gy6&7({5MF<^2jNJ
zF2N{U?BXu=)!`y2QI6u9@iyPT%U+ml@rFY+2Lwo#YGvy5^z<Q(ln<>kqa!1aAaKCn
z8Fw*$$o&xBQvP=(3VWe}4H4+m8JfIz@2-Ik5=RC81w#u4L=~C?@FvmBV77oT=kGgt
z*B&E;PqT+d3T@lg)>fE#)^FH=cG%;<O!vv(Z3SpX(A*)r;6Xj@-v5Kk?L0M5(zrrc
zRF|Kca^27kBuhG}9IV^C`Q6*Vmb|<?8NVG49Yt&Wq>mVuZuIujNix;rfAa7l*)D^)
z^LdS~-6DjQXyqf<MlRL*N2~eLY~wqwbt3U1)!Ma4M-~78q;mnUpC4_Nfq{X=5;v~P
zj|={r%q!wy`$W^@w3eZvA@PV{2i6|$?!G>VFN^4`!S@;T_{bMj893Ba5u0eZH0QgE
z3JD1zxI{!mcI_gadrxSKr%s)!Y^IArRwV9zbcXS94D|G9cVF3m9a8(}464J*3JTRX
z)@;}zXfX{vPbbesRb>cX2T1nxmg)PZJ3Bk)rhe-|H$b_<iyNt_T_87(jgL$H!T!&e
zU!71(+UtU=yH;IwB$ea+3g2^5l5W#)yuH0GY3srKQJa{Un8H}rhl@GeVz7I2+sMw%
z6mDN;A%$4dQ8mX1N5|k1J<$iN6TErzCX6Gof0(#6FCUJ8>El#(SIIpl?&)lmgAcvE
zvH$M2#fWr>jF1Sk4w@B-5Lx%nCZX|fUP;RGq4yT{SNNF;Vo)L#8+!(ggcz8Z3LwwI
zCY5n41mz1o1O;s9#3b3JjKjs#1bF~f8g_GCw;xD6=toION!>ko&Dq&~ezXkwZPtky
z=)u@B$7H$y<tjUeIVaS#EkAE?hcK)Og1nbc;j<y81h%NCt6$Lw+uc_a8Q|k{kdyO%
z%Q55)_#7K1@8F*J{Fh5k)SJ(rKg0Y`X+`?R1_sfIG|w-_!_bP5J=Z02<rQKUno2e_
zT}Bq<zzH~<k<<7z?9dTIkRRa;mUkoC9OhK|2{ws_KKEL)X3h6=J@7b2TV!ogykC8n
zZa?gV`*O}EUZmY|w7gelHR{=xzf0H40ow%924J#(+wsrhpJdNJ!!1oIq;>Mfu213o
z!Nxo2)fXduR#nt})5M;F(e}bsrt*!avqCfzu_N+jnj2&t<D?zDJ;-n_C6H_u%{0PF
zxiwNwAFatlFMU#YHy-z207-3L8|`J0B^@K;IQS^he=94ixX1i8bIogzbTCrZ@z*XR
z(d9x$ndQJ^=(+!SnAVF&c6W2T%zW3^cjLy5!690?(f4FyUC?(lTDo-=iXwC;jpl?c
z=x*#+j0kRl;N|Mdf_-R|-Tyu$KI&r#G>`hu#xHA|TUua}-SZ<GawX;%o0-3rmBpd)
zfQACTLKA07nXT0`(x7rzV+Ae2fa2=vDr8!B84l6tClIpvI3Y0jr9Fv^jO>?snnL#t
zYtjR#FXh(!64BkT!5}5f*Dd}#-5Qq7o0YI?=2!_HUi%j~P{&`0^|8oGwCz9m@ZjOY
z7moHVK)K(}Z}0&T#U|nQ2%mfZJ|8w}pxk}T@2qQTa>cH9BXJiP7Lei?d3D{8ssPMj
zaKL`#^TB65F@Bq5v58=>D3_j~&TwDfROdZj*-DImV(pcR9a*xmuP-J(>E8t5)r#!n
z;J)(Z984F{^3f#|0BDzJ{~al3XXlS0ecjy+KQ{warSy=!La_Pl<Ca8M5JuL@ufbcO
z^yOXcxN+iHKf@Dn`v;5_No94Aa0kNVjF?;x96YF5;E|Wr@pJvUb*LBxM^lq8yxO{T
zD-=@m6Jfm%bm(;}>+9>2yw@TOC6+<c^`g_(K?y3af)Z(V=^T)%EblxB$4c<i{FWE;
z%#E)acO#eMSs#YL<Dr_-m0~xJCffW+$Elkq(>2B2rmx;S*?8b5*v>h5c>$r$B{L|W
zwJ|bCe6eA$Xu?(-38^V1Lt`H)LY9PO8UL*)oadhG{601{mHBNKav!Ib$MlGPEe~n!
z*mT`aL#XJaJ<3Ty-HwjlZHl=4SXC7qQdyl^U0FE?XHaa<+Oy+W)2J>wnxAfj-orNi
zS@fPDvIbh=3>PbW5&&uqE`1S^!DNXeQUdb=2>W3DuP0yM2-W$aX=~=4(9lo|t65Z7
zh<wtH=gyrA=B8b@j&G?SZGL)7m-*?pX-Y@eXbCd3BA>$QGO#ED-r(d^>>7^Nl-<(e
zf^o^L9x;zYX1!d`Uf?`#4Au7X()&wx(u>+Qzgxwru=YV1{=GH{7_H&nUPeYnWA{0T
ztPi<`L}D~lPE07MXh`b@9~70-=_b0SVq{1N35D#M6t3(KrKQ*`^~4jp|JtF(RGEwb
zlz@J0STZCL{TdOL(u1wCnH||PxE9~te-qvG^tIZ3jk#Af?lSuE&peSmzvleS6l!Yf
z(_t0CEf?4Cdmg|dazXT;Jgh|IGr<PU3m10h9L_1%g6$KHtdIE<e?C7urIT%svjuqM
z%0uE2hsZrfLQln<>-Y)%I)3lpOW`LFOv|lEcMmWm!2%^w&rvrkAV3k2(07pYn7xZj
z_QPEoz35awbbO8^i_bm3@H@@^DVloN5dXMw5v72nO&vz**bC?la2!6Ia?VroO&$6W
zYR^utX_QCGbx_}i>ni<a67*`Cnn)_rjq+ycF38PnY-|_VIsr&hde97D7`lM9+unZ3
z)Y;T@01|MbZ`on=uRz)UjxC|^-n5}shsS%O&C-tv3OWLmrD%oXBrJ;tzCB-lTJjt^
zWPwDG66SO7UZ!PJ?;9|n{~}((H)|x9a^JmEcT|S^NG(y*$ar5%aj_xFeci^ve^yal
z{#5|QJ^BS28S;&p?hzt(`;Hv>5`Jm$gdn#B`asy5j&ocpqQpc*X3*g{+-Qt+M`U*6
zs%-=N{G`XRLk3>@tb|`R>M>|)0R>J?Oa!;gVs1t~0G26KdzLXJCoez04g6lwiLjzC
z;Cuyu@XOEF0JVnV6az#P+F4UmQ$h^~HP`MH6LUkiyN(<uQUI@3N)BgE^cgodx05Gt
zoX`g*z=Lok&ZD2_vnc!w44<8esUE)Fffpc)&;{LLpkS2Ix(Iwqj(qCu>H_9IptVa<
z(i3%o<{qK2nr5-fhv7wCc%?%6BREWK2hJ?ZNk?+ThLcw8sv!q~zz+bIU~aCMr{@yx
z_}1;)#NF+=v<Ms=U5a%acK`n8XsHbYFK_}1)}JGkQnUiN4jtM^LxZe6jXoaN+>=j8
z%V^Z8H_L>nqp^(KX{@x3p8m2@bnO)bLqk6yAg?sEjj?>s3ib5%(x4)@dL990#`4>=
zZrv52*-##x+cc~P7wGQf!%?Y}CIRu2GI-|yVgWv=Tj)G$c+N!d8{SSE?eD+xODm)|
z*<p5ah%2WO9naG_=o?9CwskX~SXfvXA0L;$oo(GF@j!1Hk)5z0ee?;I$=x4>C*-?7
zg`=+YasV>2B|0IMx7F-|_5LbM!uL=Y@%W-#T$OBoKx{2lQ<gs944M;L6=W*P%Nz3D
zb3U3~bo}0kZk`cb`QYLJgVBGBokBO#p}xL8Fwt9?E@=Aj9A_HM?c28x&|8w1w<cqk
zM(Jgr`Nf4f_!c}_=zQ}Y1_zUU`6#IS>YXuz*Pq^BX6oUUIQRhWy;N*anV)yt;D~Bn
zfHXb$0`L?|7zi^NQ+UpD5t8+@XU{Oo={wn6h0y?o@4!z*9Y-eV>pv^*d>cu~ikBB0
z(lR^zqzBY7yxL651poa2@F)C{i2w)W%$Bof`R$U?>_A6Nl#}xc&~NlHAUuF=vg-hA
z`_tJMk?@g0CxUULro5a_$Yt)!6&wMQz<;4|`eNS{927u|(x*>j=gla(Dze^)#`d3I
zWl3yOj_YEN9rly`+p%@RzA@}Uw8VZnICkde=b>DIqQc3BV{&kg(KUo5+Oiyb_daSI
zL(2_kQLxa{f5pyGq;j$c0bIAOO)v2xcB7%guZ^~1dEg$9(4%9E=Zr`T4IcnE%6+S2
zsOP9itn3#a2g42wZH|EfSWNgYE#=_o!Lz|YH4Rz{uZfO|%5!^P-0!rcl^FB7_wVlC
zzmH`MMTI;B<$4T)#vFz=BE!6W`L`v5*Z-rXU=aQzE|C!VUy;O~UxYl}qk8=K@k^JY
zh(JRzx@=^GhPzy4OG^vPXFsQjAew;1>sM5CGB)P=IAGrUlVGgj;;tEg-+|V3FFK(Q
zLgr>?b+he1j*j9j>U;L??S^3z0o+&eAC2(8t!zeZgmtKMQV+&aS??}e7@4~#3qp8w
zUfU~CSS(QtW(%m?+}&~DPcRMj)zziDDhyawzGvIk5S6RMMYuVi+}X1qfx;5?h_@Mr
zH|1B?E}hEC%CfPsLEaLBQ3c5xS&3cJL;I~k2?QXdtz>M%23<#D)<fIllLa4iG)VH8
z=6CS9z!7%%$PqhRTVzkNNf)>aEE^OWWZv1;wJBY%H0x5b<+mQ}3btx^^9-nIt|m7(
z7qZ|IZYAB5HeB=M$s&RZV8I2gcpe+u)7?$>pg~du_HoBKNyP1qq<4^YkCHz>jq{Dh
zd7lJxN7M=AT-a;e;L1mf2gfd601QC2$5SL@M?Y>pQ#qjO?k6^#i&^T)mI4z7mu}jg
z`nR@EW|kp52i^OzgR!4R=xY0MS9Y3SoG<lu`1xgH<AbdJ<{Eh}>4l!fP(tn<#6d^r
zCOwlPBlR>jp9IEyP1I7$X6pN)fTPABh%&LV4h;;LR=-JvyAh_zu)wu+bacS#4cQLn
zcAD|n?+zVZTS=+?&7v6@kX9qWJeb$E4LZzE+mzYM$jr)}twfbA4E#`522rAMI`d~o
z$0$tnkWxj31sfQ<Hm||Mk2Q3SJ;!eGG5AKd)i7b}x($pLGcz-j!!IP}95lLQ^D@dl
zS8iV`xI-$Hos$XBfbXB9*(nfr<L9=5guk#CZq7gWA}VSpC+C@^rAs&y5hq-o{v(x{
zeDEPLX-m_57>eIBuoQCSfo7t=pC5xj1FB|<U<0~F!NK${I{DcNwO61r8orUk9I$@h
z!)<aV27;5~Bfi3U-tYDGPT0{wv!SMur<3t&1`i+2dFrnF`-8-J6nFUTgJ>ou)?Bq#
zE8XnjlP5nPEt(B}H}bpp=Lg#<UXyJNSI)!~(|-pBiDJ3q=a=U`m*MddWgVeqa6%bT
zo9#UQ!msm23{jWh+LiHVd8>L8at19eEkReRHz(U~<Ow$Rm_#_gKnb4ZfR;u~=tJyJ
z=ss3f-t+O%d~4v91hsD;6eoCD$VTuJCkg?AJg~6{FVD=)nRAKZJRpFgB^(0DK)sza
zd7Jm~%I=;6yu7~=jpUfhPZlFM!36{cI&Hq+$~1AiwY!oUe;wu*Sq`#)2s^4yN=s9j
zrA~(XLi4+(zJ3QxC<!#wKG2O9n%WkPi`SyBbku2Bn_2J*El-2_1D%k?t?DV4?~3oS
zJqD8wk;T#?znT3h^4dTSR)k7{wFSU@&BcZ6vI8lCOg4k_3*eV|k*J}mxr2@_)#2C0
zo|RM;+e>$SHkltgiZ`Y7>f%BZL~3CTWqq8xJI6S=vZDL*y?-`^vlA#5-hwM7=@lgD
zja9*FG+uRtQRtoNmwf{_A1EoIUyA!CM{I$H^g#;m>)XF?A90S6EzL&1BE_7?z=w{u
z8e=sQ)$=d92R0}EA&e<{mqCfS0IUG~<~ir3@uW?T&RFeeFU)e_O}0-(A|N?@%X2%I
zc+ncxD*Qu$QzFZ@Z2;HO3q5)`22$XGqyi;vo444gg@ZG6fUhvF@^BNu8~Fnu6I#{`
zc+<PAtnDhm$#cjs*6uP@CdFJS=98Lbw@>jhu{T7yDY>X_r&Rl=*FG$H^8E9aqkwn<
z)3i7WPOK|uo5lkbxwIcU&4Z3zfP`yrpB5M@j7URnRG-{9+?1)Psh6YD1CdOM*GN`D
zfrf@A#PTYbiQunC=kZu!P8ueNnzesSsvhZkvz04fwS>LVV&U7~#l@+Duv9fu(^rc{
z=``Ckg`1?BS5|*tQ@YFN63f{0l$jn=3`pG*6w@OwC$Ss)Onka!)A!Pr2ZjM$DH|q#
zc%0%jq7vNkUE2EqG~bcc$8aJ95dcFA=CObeVkm)paSUL>nho)$N1x!SHnAQ{L+^r|
zzX$$;r30}!@7Ar=uD6EH!2XW&(}=VG>eZ|G4Q%TowsmlDf7sVRi;z!@%sy4{&VwxC
zP#6t+d;6<b-#{$O%yfb`8O;h|c!MGKgr=JdikR>4Ge#yL2Rh)rC>H22u3WitaV)d|
zzM@?`JV^<=GzLIBV6@}sWZ}DX{o$wKwt^)L<VC^g-+;}?+dg;htfQfUb5j$i3*y~4
zgrW{&A1t&_B|Td!Cco6xk&{(W_P!=)d}(URfno>dua)CyX=C&DF*i=dGe9PUlm^`a
zG7}gxB+QNL*KZfG?MAN%yzz?BEyi{0*Ed4`g_<G7ASNb;zR2F314oW1Yc^mFU{3=|
zS2{*7#g(Y!;GHR9NxFGv(&d84D;dtA4j{uDWuX#8)m}NYF2#+Mh~Xs*;GtIMK32O)
zTRS_F*&r`XMq-o@zPIyEreU6q4Ok4+d`}^Ox82u`Kr_N&gXKUbHeb5kEJFK5h>!1c
z)CjD7jJfMm6D&NW96#wcS@-v(tYQvkt}q%|5lpKf!Q67`LLKfAhbQft^%H_b+}UYR
z+p+NpxD(D{OSKS%8iF<&)M?lX(}|DJGB*J2ms$Z=v``l0!QFf3caWX?e(WcI!hDxG
zs&*0yk7$}+US9hOILz~}`bnpP&S5qN==^7+V`8wMc&Az6Z4>!EvTNYW3%|ho7|!1E
z?aqXGt&75l?VCC*vf*6p1+Uw;Ay@4fyi$0G7>FN_Uv@fu`UB)azcjk-+YhvGiHK-4
zJ%y*fLa+(Dqj7)RmjpfLM`da`_2SSQf(7ybCnw)b%Q1B<m7JWaTyIc4t>#O}A<sAE
z!&{%>*SrHOtui}2B7$du0Z)6N>_bAk#RN_f=%<OxD->)SNtT)g+WG4R=qPrWEB)}{
z3>T-v7891}V&PjhcUF|{qCUJdmWI<#P1AAr9|c7ryqNYt0Qfw-z<=s-Qh8FN`4Q%{
zOZ~Bf@1=_^>>m}uMWE=9oQ9;Mn|a~2iiO1>gnUy4WyngX0!*I_5AyKnrd_GU4kYah
z7Yy_hDl}k}pC0Y<Q@r-Ey!;VPgoaY+D`eEe6cH3ePDNs3isL_4a0#6l;w6Tf%H~@k
zZDNU-L*Kr%{+F12zmmFo1E}ZaO!U~BhYBplI-_tg=HP9)C9Q62Z*~DhEkFxN<va9H
zHueJ1_i<u8c2JACP9}V;0IY);rt{{~Kch!@1hAPJ)$K0hg>z@l^rH=O@nXo}ci0VZ
zi1pOKmp*i$N&|3I%^z|!4h|{AhCd_f8gg8Qs-~uyQa`W=49c3{qr$?%25&q>2Qx%q
zxoTBrYa?}F9c`nR^)waJ$1)0e^yuY+Hs-9qzyI(Mj?EQr>BBrW@$oSmuoMY@IKx`H
z>zMB;miP0OUvT-@oO{N{D?^qb@-miOF*M311?rQ2<L62NDnU*eAed{mA3l6w5wZR7
z<%=E4%joDRGFSk+WImdJ*m}^_(eW%X71Az5y)X+i{077(M<*xXSm;~xcohJZR4fd@
zz>^<Xl<2<=Hn5B0yHZ$z(<70v&>5t`DIrjHvBbl}!+{~Kv6CAI9`PSZz8ue+3yBRI
zUS<F$unPoYDwKknNXaFrZfn=9G39~+)B~<_WP89L6LQeb6F&VLlZzSf<ibS$1BX-t
zXh58Cahb&)7yAG{98KYgBS~-)*S?sPR%}1tBzsQb%GT{@^jHb3O2m1U^b8Cd-!-9N
z!ubI`>r)$q*2@qop@<yy3`N)K5TfE5kjwD7r^i$^lWMQvfJq~}eIEu0g#7hgK9DNj
z2nrAW69ZIrV`znYdV1aqlUGm(VKfWPvR{jJWS)I!!D|Fwsis4R<LtZ3x3u)VdBbOI
z{rm7xEB+e=En#PX9D)M_^()&guyRBa`DygY47OI-JI^jmwZ#hHtn0l^wy0aNUMbwd
z>Ke(*Yez$H!nW|AO%-$;aYd^*X)9`@dC-92<5QY^%)7vJ8Ha|Khra-2z>awt({!L1
z4JW6*fdS{93I<cwFtKyoe$BH~*HBsGEh}1D7La(hs8*Jj-?@7i8!PYIKiE_-f)g!p
zdg$K0d+!vHBR4@4?1#$NS`?=`!kp03l?`UXn0LzrYDxC~P1v2zofzerI)T)8&&6b4
zqk}7ulgR_oiq=&4=p7)8uJwY-Qr}iP@BPY1rpfq6wnKc{iAL{NpSp*JaNdso=XP{3
z51A)IrZf_f%OGMCFc=QWM!-APjUapdANJlnn(Ov$7uFm~lOaP%Wk`n1B2=Obm3hio
zlvx?0q=Cv<GG)q?GS72GsYIr}<{=>|5g}BBeO$WlXFq$r|LnEbde>g-eSg-op5NWX
z=X-sw>%7kMIF9o;UFK%&{8=Mn`%-n%aa-ATtfDPJs^2t7H0G<Nb+Qcg%T8RE;VDe<
zU-)ynR&m#?)5t@?lbXXj8zhh1y0CU&;|d&#y1d$uZ|q`#gE=m(rC3iZ8%6M!|2+{C
z?^Iq~3%sddWJ-gyjVm-sv<#;)xUkz<Mcc#OJ<`xlGg*V*s7$LRbYb&F_;cPYxZeDs
zeE7rj50I1B8szjd*$2IUZ`jjAY+VE2oEg-=1RAs86m(`F_n3F76ECl_D3l*U9oF65
z?SRfz7JtHwrTtS-Vs|@a7&jQK(8d-V5|Y$Ssq1b>O313p(~QcX4PF~?chOZ-I}gDW
z{9$UdIA_7<F)=VO7g&9oL76jwdZ-#WBC-;sE&`{%Aapf5z;ydk-VbK^!-t9d`QAMZ
zJ-z$)@0XaqSl&6uK6+*$=wWn%l_4%zat}|aquxR*kbs@e%wog}vb-M#(Sf7v2FZ8X
zGwDQifk0W5ay_|i*(MGav`xvavAYAYyr`bm4{Qund@1>gF5~F|Aj8#dO!F_kK{Jsz
z4I_~ab2FNErghERzZLG!VD{J}nM1jYTno5tUkwHC8P_32r)TeLoO{G#+o!4V@$n>_
z76pM)|GAI&_8%GG<_?XHPM=px)!-aCmEX@K@Nr~B+N^Tr=(&)+9R^Nqe{D)0&L2Ra
z9uQ2M1$J8KW{*l4ar{~82|WS?uI98ZG;obQBtkl}aKhRJl~h#bu#mM~4TK$>uR&=%
zi;@EEe!9qgPM&PU&nA8Lhy$VJIYZF6d=|VVg?9*)8=b{XwYBMR_dt_@0wi$Bt?Rb?
zf`IvojCuVq?kkWMTfhNlf@VeD1%>b;vFCbUsyQo=gtr*Y{NQDQL+=&jODOG89{Rox
z<TETG4TM5G8R*wQZIbEVU~HJvZZVo?`&MZ-eLyL(Vx9klgtnhnrh)d)-w18uw`<zA
z-nxH->~X7s0Qqlkehn%WYz+KnA{`Ja-FRJ`S7hnVb<z|08#gwAHG9AYyg?*Sdy(2c
zB%&XaaZ*~XDLQU$Zao1BF)?fd+1DI-Q3g=1Ky6uZ>1S40X6HDq#9@hOJ{cJ!(Hru$
zQTdEaOuB}K8S|ji1X(xK^!8TZqjojUs!UlwgEw-SoQ@X~rK)IQC6B1(D^?`C-LQao
zFlkQd9of1#!*5%v0Zs7m<LIBo)2yd~fDz>JS}O|?6ULUO?oMAdyYu8wi1jotenI}Q
zBl8E^Ta=t6&=uI?Wot5mb9TZ-B}pR{+K1eW6Q^y2xlkO3oH#@hMZ7*_u>WoOWTBX<
zd30!h1U_HSx!Szk*-KDrckkdjb%yOUT@?p!7006gh1*8V6$ikLA1gAPiaY!>h-O8Z
zT+#j&AxpbgNr&#4;d$U3-l9@hxo-`H4H-bh@Xk|wKkq2?rxP@|gUT9DOuD)_5u1Xp
z4I9{H*KOF)n|tQg*n|4z;(sdZZMlVnD94W5G(?RFXl3dYTQygiYn4S8;6{wjuw6fI
z8yx;Ezs5oX5D|6_Gn+!Oapr!|xIY1L9ILf(&{P;~(haXcp&B|?Z!6UlZQcrh3tV2f
zB45WJMj>%UM<H&Z@urTFlCP*lLJ9Wb2AD{}gNW1}JwGA*dT_u!ui!HJIv_GFy8U;5
z4d;FPzVdMyNAdpo-Jv`_W-$vtH9`2ZzB0E7kgt0odE~X^CodtoT!~!~+>{MBhZypZ
z)8-9|yKB9^8DJ7@l;2*ZFZa*>>Nlxf7XO}l-8xZhFb_V@S4~%Ij&@A-?h}}!ar}at
zce=e|VHz9)Z`d$)0Hj%@GJ&`Yj33SdD+>#6)J~3C9g|;#qQU_|(m|XbE{^Zu<~}Mq
zC3=O`wqAb@H)|m~D8%Vq?Utg68w_TQ*B_jHo`>GG>3Z}gy24xo)L3g6=}f^Lg#Mp`
z{ZDEe)}-oYKNvcCc;=B1Ml?Kpc*MiTIv)8)(_yfyRtXESHMsR3GGHD7ho@!NZ4}y3
zCY`}x%5)42c%K+&X2351Qj}rA+Vp%l7rK3LfAfga>5n?HPM}tn!+ug&Hw*E_=;wYg
zP1hNxeiB;#kZMtt<SVuSd948XseN=8wszKrSo-o)(liqtYr}{0Bq$7>Vwd<&rzb?v
z(%Gs7?PJ~cJVd^rBPR64w%egK2>}avX-Lb?Zr8m4Iu@PG@)s|bWbFnzNOnBe@nNu2
zD5`K+=5a=Rxs<o&EYN~fQj7z^h3;Y`|0G>~;>y2km%Q`$w0qH;lN+P*A9*nuBtD+1
zaBe)xT42?A<k%0T3UiUhAo4?CJMvBtDhS~pf%3>NaPAE(cgJqpxN!$RA3uMqNuS<B
zREbT8Lw=5fC70;TQ_AZph{OG3<yD1?#L3Bt*kGgFgGHi(cq-gt=|tFz>KhwT^W;p0
zSJl_s;SDvdY28-0>XO|Lf!YT85YOzME%J-SffoKQLyddHANXn03~Y2VIK!5Bnl16x
zh1+4cUWl}#e6RKtt@6~Mv~Ys${L;$R5)6W^Bc6=|<gKvk2?n9a+inKnJYgi+0qZqf
zVe9MzCnqP-Iwu2XS`gTs4BU1ja2$Z9{1+0<y58i1f|hC>9L?ZMXZ^bisK1IlC@M1x
z3JPk=%t#~LJrqn`T!Gzpzd6NneF^IHo^bY^2PX1+c9wXm@bX$C;=ukd#K`$HQX{Tt
z7{=k|ZpUU^_ZCnS;`)AzxAfgdY-y1mo^shQD0tFf$o9^yZXHX@l(GC(Ba!rsjK(GB
z56yMf*B`GC`|^U6RFPY6b+8U>{2_R9w(-;u<Zjj5ImN`Z4d7HOHg6HiRux51n69gK
zw~PfOa}Mo43@y3&{maJ;7gIkVbJD*YkDAE;XQ3jlDBN6;Ff@g34Q8FVpLM-%YQYBt
zt3#w16;wD*fc;XVEt*k6;;+ESwGVD!h+}*_Jn!nu?itN#go~X!UVP-Bn%dn*eBWf$
zkIkmdB6%4op7?yKz%qD2PK>1<m}Xt4x~on^u&-`j?>!_^^KBYAHB%bgH5$NPbY_G)
z5SLhcClJ|iVFqVq06Bb35J6PnMcUp3Xe=tJ&HFjrWU}q#J!FByvh32Xs#5_^t5NZ~
zG+QC1nToi*XYZWRE9P7KkT{=SIk534D$m_>x=NJA9rK)?0mXe6|JFOxotK9OucO-E
z1$ofHn>DAn=7sJ~S$r(xmO0+QFZ{(!V{T$*W~?m}tZXwhh+!Cb0mukFfR+*fd~n(g
z#&wuQMtdvdI56(T9X-@YpOH2hT)cP@C=bfYY4$L&MST~W1_MAeh3tqC=I0LbD9qBO
z6BuD#MBXsULTERqRlpZJG_X9$+aL*jhVPiE1$RU%;@*{zQuAF9%ASjK+(LTMztJ1(
z7g|yd8~lP&ZI^EV0-(2#kI~bMl=uTXcJ!fE7MJ|qP=lj$C_?+X*mX$VOw&_h)DR3%
zN*Ba|YDjMG4e$>PZ1p7FhXvO~440Pb!sY;ZzJaKLyLkO7=|n`8XZKcCr6tQq+8<c!
zkEJ4G+og5?b<<H*!@)gef(DggSz}Yeov~|@A>Ta-H_(}sPrG5mo$>K;VrmQSxa(d^
zi&jYQjo0W^Pnqxg`t>)_O_T--E$+?xoB_Kdg}H=+0%xM|*-lx8xXj5914X~gu0^>1
zzwcQ*x6ibJJ|6(2sQr6AgK{E5nMEUVX2Cd|Aw~8c@yAo1nhh95wLkXZnT=yF)|s@I
zgJvi=X&DaEB(LBdfd0PsOJy2dL+~M24?;Q&&^hJX{DUoyFN`A!YYm!A)&iQNp`$CZ
zFmE`A>b<NH^{0@R@-*0Y7pEA|a3EoLG6&W%U06yD8m?WqZDHudwwZx})5v5F;kHRG
zkRF+?z<YSssEzC_Wd6Hy{L<^689JHXn7j^{y1Jv}LQ?Q0NAM9aNVJrj`~90tRanj7
zydgFjE$Ye1zFi2h*cuTN$ELo>@Bdsg0<*Ir*^vtG`Xc8a_lMguZT=)n;Sv(hN77(m
zZvN0-sTul#X!j#)EKw{>S$Td9*x)~pzC?GF8+;w=5@u?>NU0vmi6HddedbyCv!xG^
zElVh3OAy=srfTIRRQJZfh6x;vnypsASl`g_8y5xMjI}%8LxznCn}9z9tso-uoxf^W
zk*Nh1Ba6tX0iQH6KO#6b(q!Ug+a_t0(>|D3D^+Gu?(zU|=0N06?&rW}Tp~Xb2qg|C
zzLC=9n9Nc}-2l@(^3VlAOMB5$h)$htV55I&pX6?ouU#WeO-%$RiXTTaa<J@|O)M;^
z6m4!C=Kh}dR{K(B-pi|+ZwPE5`0LP0DrOX!Uyw%#Qrggptnu^cBxHvyPj{$HSL}Tc
zp_4*O10*r34N4O~el#ORdtApgWpbBuPKPu4-7{Mme*Sv2^5Q1F@!KgcW4{gdNPwST
z)0H59|2Y_1)E!7!aqOH*tmK8to1=}f(z>%EwoQOTcJ+)P=SH5Fo0|(5hsbJueX;$%
z^;T3kpY23berB|TCAa1yRm%MRuWVAzJiB+BPkek0>Z-n`2LFOcaX@+@)se<Egslz&
zpz3DlvVrY#SDpe9Liiz5OoW7lApd^q{Iv<}8=^9FFPO>mgIElJnn89GYUJPsh@+_z
zwYZ{!f)dlpwHr3bkpg~j%^Rl|#@$J^vk(2laozQtrAVivvlX+zj{Ix{$mj|0EUup(
zz-(OFLjhqEVR&9^Og|}=kpYXCvYlpbXS<~uEq**|%n-^uG&w?-p@s4jdQD(vKo#FK
zc_zD`i_ULm=tKx=Pl&KV+F~DSbwPkiAO%T4UWE+ELW}S6Ba0j!h^2Whkb7pNheSqN
zJrtW?GlGKM>gk>^WpEya@C&yp#K#=A9|mk8Xj0CX84?>iIx>=;sR{5Jwb!u`P#g+g
zHzSmj%OKe-h_dfL5vE*-t+{KP=uw1K5U2pq1=G~j)HZI~lpy^P#LYV1j$zvNipCfS
zi)?8D1&UlW`GbB2yWQLOHi8(vUR+$9)qoChvxjc*UBPL&9S{IDsZa;8a5Xvr2B0ih
zv2F@}ZKtLyaFL&L6rA7Z{OCk1?<xHmpTo1}jfIilmxUBs)G}J#05;wf1)r>@D7@N`
zHdu3rrifpfvGw0|84I2KCd|m91b92w$y*uVPEWBJ<I;c%iQld{DuUA$fRgwLrf-Em
z-1NEm`H#zeNnv1TchFBM8Yb%8Gt1w3Z7c`cMPeOjHgg-yc=GyG9Xxvo_89st85$%?
zu~cyc$gkXb%QDQrwi($qzvX#ski5(=e}N~a*`q>zEgG(-N$R3=*Bq~P4pHq$!cK&u
z7!^LT9Ki4oAP*tY(R0Lm1AXr}>1A?HF%2#Wg-BD+{GUJ0f}F6A9ff)k2rd4ZAI?j2
zGtpF;BlgSRBmritQ02wlpq5+LYhtiZ4t85XfvA66nk+I8Y$I&>35*N$gy<P3qP6}=
z3HCekB4!pA7GSoS?}6ogO$#c5qB`?Q91O`mzypnV=pgL@bVvwC6pyGDFX8aR^{zf{
zulb-pT3sm5+_G|V#^Il|cKv!Iz=b$tKpgc94P)S-0eLUt`kPm;-~|#7yb5`8tEH%p
z7c<vWSQ7yhsMx({kCN-?VgKWCvOT|cjpl)D6ZE{@TKAH`Y1g;a4#Q-{@xdUMU3Go^
zD4Jw0VwP1X-KzxD=AWJS3kWpy6~BZ15~*xj8xr>WcXV`sT0b;bnvyaEq`?6u&cH><
z1W;UlK!$;{Y(H>u9~ajnED`*f-nBWqxh)cdDTw&h^Bc|+M12X+rLW>p7P{hY0P{eg
z1aDhRK?A{1iT0HDSwXI)%lgv!3BZBNEb@O*#l+b(D4vAk_Ss8{1hJ<l90n+8RRK$B
z&spdBqnQ@xt|6ttA!e~UA|>>8ZO`dcE;x~^xei|g@CVnkI`gf-oSQ06ipbK|&E}(>
z2*}CfXxg)MU)eaL8<#^B*0ky%6Q7>x<p|+=@=H_-`<jkZN+Bk`G~KFOzgPzH2T%!g
zq`cix^-DhxISE{uj-4yqst!|f(msuC-W4UHN{8l)ZWPEV_1rGvWj8L3{xNC`{(K!K
zXlM0f7}YL|mnkOUiey1_{B4qvot+)*J|yKwQkfyt?ObHf$Ba*gV*1(wt059l4$qXt
z_b#2S9(EO-LvQJ9MeF|3dQ{A4n8CI)IpRdqVm(b1@j#m|UA(x}svkQ#QTGLyUyKbf
z2ha6+r3mPk@$vIc)?II*#~2tG7?qjc!NqkJMRSvG_Xv7D?2!Sk-Vz^>vyhRhl?Dyx
zHasEf&8j<tV`JNynM<Kje8czT008<M_n=ton%Rh$4oqdwh%o?uc$X4HDp~O8J#-kN
z!(<#-d9AaE6(>L^l)Y82Uq?Ef*|KrtJ{a5rBZHX@A84#P`K^n-c?pQ-CnJRik(qNb
z)w)t4aBOk?jl>wcs!i|$z<H3BGD%KSpsY|8V#BYer4=!5hK?3UA!4P3@CKu!6aJC|
z8>={~?z~<K`)~IMo<vTByXI8+<<g3ZiePIzCp%DNY!++yCOtY9lT*v9J6dQblymKe
zh?^=ERn{XE?kE^Kp=Cu(0Hy_n3|1E>fv8)M$^ZqcG^d4VL=OBUg6}g>SD-dgq(ezr
z3biBR2yVX1*m89NfY#69c#yOKYXjDkfIoP8{h0&P=pJtP(~eV&^6_qP@B~!zfYMRl
zpwZ$01jXQn6Xhzq-vjq#Wnch5k_N7Z%=pE!EL*k=-W{W8Z$k*;Tl*5bS|b%6TF($z
zpgF{-z7f3TE<s*7xeB;+;3Zjsh{b7Jy>4q?^-o43@syPO1yr!?9ueWod6fhID}veK
zokpbWwW6t-=WxdnatI=RL&NvDI-&ry;xU0_<7MfBBE~r<*kI3-;`K@V>ByP!<?Zba
zFpCXjCJe4XTj}fTgPdQBFB%$B(&H+@vRUP5<Sd&%YVHhQ8~)lv51cXFiI*oTko}TA
zi`dwCKw*%Qva_?BnDkz~dKKMpqTuYw@|C%W+#AbfXZq&i{&f<oin@GS1w=<P+T&_J
z^0gxt5nj5#(ucfid<X<;prW<)^_eK|LH80xB&7fNR!>y-)4mq`2ps+DQdO!~v+4~8
zkOIikh!Yr6jc1S=KxPZ;L#Q?(8A?VDfJ%4-P&P_T)f3hNM<Exo&bGTv0#@?lmuR!f
z`k@!{`wsO>_mT$vW-sQbHe6j{bZ-a$z-&=Mk``E;niS6yWHb>G0<fHH-O&J?albIt
zIlY!arH@j&1>C+PF=AC54e?J6AJ|r}dXjOW)z~J+)HLZr^0zg7FNq2+bOj;*FFqyX
zNkA$pzaEIfJI!7^X<}kh8y?MPf-sI87FQCwFmkRa9XaAXwXlWC_Byg59;OKHZpRk_
z>4t^v>ul8oh=@0yZG38qOtP6GHpe&{Nm>U4QsU$Lu)|vg!K9x=7e<QKfuR2D&jbXG
zAkRxZR0F7i!FT1=y{2r5C6kWsc^pjNmeN`xoFc^n)gVc{t;bhqVwN6YY`xTrhV|IJ
z(-ou0zEsg5^Wq}g%T<?*1j;Vs<3N)z9I=jFmUI~@zh3d20a_rP(vkz6;k2&$gK5vs
z{!@cc4c<V97~IrO&fmju6l_LbhCAuL{{GFLGZPcOfq^bgPMZ6+zd4T;_7^~^TMQl9
z&qff*%I)f_sJel?WH?@M&*T65xb@Fh5Be-Mwq7j-RlU@0TB-20x0Uo)fbyeRJ+M!p
zW5kg%h$Dx%cLI&iC2U}YdH>gSG+q)z?;FM2R=yQaLKv+pGgaRo!|};a*OxZQwaM!W
zMj)G)zLt8J7r!`o%|(B%faoI4S#lrkx%E`z=6!xT+b^`CrF+}o-XHSU|I_hIR^k6`
z+49ZeJ)!?<0m#q8x@$lTd3ACaaujN6>LmevC~4Q}oNQ{+P+W^HkN40Zs7bAq6axvt
zSAh0oG#T&^`Sp1L^g>(1yJ|_@N>%jg|5UxY)O9WS1bYuRjQ&#Fy$xDzPy!cSQ37Uz
z5+lqHqV*DDroCs|@|f$<ib?A5a>my$cJ$!EQq^r*1p@PzOFg^vFaA%T<o|oU_I-=F
zZgI=hx6QqL!=$%NL2D!5u^dxTzucAU{c;sKU6<u@QBo*Jl3sjL3>3?~B<7Lr!GDNP
zXof{>#-9Gtq1+>{Z)&Vsxksh@%c{jEjUQyb@O3i{&bb#hK)`yI$qlV9`$foWpgQ*7
z91C_YHF^E#xBq{*CZ&ba<h{qga$orK8-~LLS|!e3?<v**-inEhO&)$yRtAc=K{S>$
zMlD1HYNQsplBP&Zrvyvz;K2hxmXEB@!v-CSynTO7Wu*RkkeFIImOe%L$7P#?rdKXo
z;@6~h^sfsG!_@cQUajNq9v*0e?9$3a8xR%C-!`C}<z7o|Rreo~j^@s;$yO3KJ|E7Y
z#{WF0v9+~TE!^x+&0R#E=&&vN?(y4FJ9q7Zl*8P77p_F)yID%J<d1Kk7RRMeMKxjx
zJ3QiCV(XncgkSHI_`BmXs(!f`E&bT%@BcA5X+3iEC?U!E95_*9Kf_{JJt3WtmZpEd
zReCDK0<dwm9RI4SKw6Gg{cSliJ-+&JrJbD}>HY5_whaH?4SJf$w+9_dmvO@z^br7A
zuJ5EtD}G61OGRZvbl;%l)=LI6kgM2~`R^U87rBgj0<M364<eJuk-fV5pJ&d%nu0O$
zh;2f`j^G}Z3U5QIt5W~|F=$HAYi%R+I%E#v)z|+#pQlIi>)4nWBot@Mm!5C<pY<+B
z+6lpt@te1(5O_}&W$4e5wM+T;-{EHwIyhJOuZPXOzyFMf(OESJy8|jd%KKRI{<Qx6
z11X@jfCqM_`*?eIba$VMC(n2AAJY@RRi%d8+1L!t$qWzu_@z%q<RJKZtq?p?2Dl&H
zNnOHA-;(BUyQG>S;G#ST>J50;ZQb|0E!=_haOUtVS6ljLK7oP9LA(@dy}Q7wf!}%V
zZ5yhbTsTMmkZSP@Y6>KBj51n)*-*V1l#s`v`fF@Nzl*3XCnh$@ywW1ipttIuM^QmC
zeWX^2xaB<O{nR#wJye1=cWdjk!b|wDI}(w(i;G8%oGeE#BJi|;+qZLcvhf`X#%E@-
z(4UOI=458)9PdrDcC92b-lczrKOYxgOG^th;{MWKP$$jaOJBSHo5#%NzYjrv|3AO|
zcX<x}UtW`%%pS7Gsk(qVk(NEw*ifT>`0xQCu%V#=QfkzeXvPL+4MgoJ_BKpCASgOF
z+ED%ikX--w@4XN(f`%Z$rM|u%H4=q_agH)#Vq#oepHac04WJj;9E^1frAc3$f1WyT
zSlieNs=X*y|7!~s!1H7)BoP2rF%VwcXc$OyFZPpOk*&Ul#=X+H8CTF{#Kjn8Z6|&l
z%Ji>}$}kfKlmW<R2CYjw_hNT<)zm0Wslgxz-Ng99WH`*9!xton&dyW7ict|cgOq^a
zYJN42K?W}%0EE%Du~l`7)bU4%9av%S6-1cbYu4N%>BZTzEaQ9P!opy3AZK%@uYMy7
zi?tq47@A>Gr0neWnx8E+8^urd0JsQ7IOll?KBekTKEAq~M$f04^p+erc#QFloq#4n
zo)jiv)OEFa+qP|3G`mUTL$NUD08t5`tFEOyB%~SRaX_~?F8pyPE;&Y|!%*sPT0_4J
zl5#946xij3(${xZ4^9%15<<w?XMQkB0|^L_!5)SwZM1l=2Wgjr=fMm80|QQd&zGSa
z#Kvu<)E=}Q08*^;WPqhHur-2^3q(ZE8Q=$PR9L#nco^S?^FIOc&cA;OeL(<E2xo#z
z$OZD%^XCCTfQ_Mm@IkT+S&VJXhw`w7I5|1!-vP@)#Lz>b5X0Stg$1x<l)Ub9$kCx^
zH@#Ml=m}55w&hGPBD=cQMzDeIxm3q#daA1*KLgICOm=$08V1^H-P*4R(YW16bBRVO
zOjGh+m(45geScb1Ow0w>I-Vvh?K6xX;QyQ&!L@t$k>Mk${EC+#G}JpJ5b@;66Vgx(
zh3g^)4TXe;K9upMUB6yzKA*gvYy676_Qa<JJ0KhBJlNxKDVFWqKL9}kR%~O%f%)2i
zK!|={#sjodGsLK{w$1}Cf!8L34+>}syB#M(U0odkcmaH<r>E!ic3&O*NgT|Jz=ES;
zH9?IB-MKW}SKA?wFJNj?ht?F(4-VUJ7_z>9;&w8v3x5;ukkh5*egk?rIJvl>zH<Xw
z3yms-2eTN~0a<JEaATy%8-qqTQ6BU?J9<EtYGmp1e(e{YI)v9=l%Fp$I*2I}xC`6x
z(srJ?iRrpGy}gI>DlBkG967ZW@mZA%<{NL(K#-huFE*C6+L%S-yyHHuoP|ViZ=M`5
zwBqqsL<K(XTw^YbJ@V<3Ii<lyfBUoQ&Y?u$6Td#brcM1qGz&M^W1HP)$fx|*V%D{_
z6@eZ|yoCoR(|Q3eX%c^Fdmu+p8bat#;gs%0j~5Ql0&HWeY%h1=SAa!Vu3oLIO%6F4
zdEobup}qiY3JB5K5y+6<wdr)bG2=W?#LfeFavsThgyvu@yWjS=H8nyR$M=9nfoKDR
z{H82%7=a_JUQ)o@wfvi!<p=}cf*{gwdl%?ONC=L*aFiOB1Gm8euxva;i`m`tU}^w*
ziu5HS211QNI_HALfFeR_=@xjH^2!b1AVS^a<Wxw4wDeNccmg8d8a+~FrZ{dXKgf%Z
zNBV?@@QpuU)O(1@Cj5MZzxhT+1}GejTUQq6-L@DH5ljU=?(3wB@!Ap4f|WUKCLq7r
zFD*Tyr(a@Fi*R6EetA#Eo+uu{-4Srw7~RW9#9%?`RkW}+QnH(v1z^P7_Q(|w`rW86
z4a1l5vltbZ4sVVy;5>?BvW7ccbTo}P(YB2vRaF4NF*~up(wg6d6Z#A|dwK*9y&q`u
z<9Z_S76WCN7`Srf%91BfXlvISQxv#DFnRb~q8tZlqb7W~M7jc$+aO33Hb4^=IpBvM
zdY7n1khTa53JOvt5u8{LTr#+knK=riFgeJ+s~{EI5!fUT;x0}z&QDluEWg1IAXdV*
zRuM3V2lj(yuSLXw<seZdrTxanunC3yN?8v#ioSKy35xG9`h6r_hC^kI78X&7<IVAV
z9IMT@K9EZHq_h;^RKw+Vgvx*N$*W;u$o&f5YwdTARAq$pNbC#I<g@NGeU6xpEC&y<
zje$W7^L6n<eeExN9U78!{B+DS`$1Gx7wBIg(sB^<C8)gzVTNvWjmRmue;-Wxgmp;F
zpHyfb#Qlf_l0F=L^q?a>FA0Ry7#bQrbk*wtVu0jxy`|<+0QVZV9x^tQjC%h+JE88y
zi$ewY1U*_2xB)lv%fVg^ni=mwxWh&QAT~a3B?MP9F!$E<%618mqVM6OH@cGyvx3T4
z6$PeCHTCsI#rCmTJ=(B`bA>$I?5qh_2nc0dKP^|qUAW75E_0wiG2fU4FXo?oScj`H
z#2oen$6cuVmG}4*@MYy){_RNl+6};WQ*0DJ=l=cf09-340sZ>Y89Li9Ahg46CRv>a
z`|s!XY;_(f1y8`7M9eYss{)t*g_&>qfI<5c@GdX`P7J(EulO`H@n_m*5EkNhE!{p>
z!>G5s@-Z|rI_}~NMA$E!2+M{;al!$4tLfR>*z7I5|0D$g`19w_G__&M7Rt(gIfl*;
zTrja4Nv_Wzi;@$xBDhzm)0uNhEfEwEzH}d-*w&wb?N)5lK8ALM*`m~Jf;?`6A;Z3(
zn5sh+|L?|3_09lmVEm&zSyEE+_UA_!*aSD8f|y^wCeBbC(fZhveFbw(Qsn`nR~}B?
z(~+1u4ihty0;hSww(NPE&JXDZjK;*RoTzECn2sz75dUT)uLDw2Whkc-<@Vi=36I;0
zXm56n4r?C&7r=S`)*d{q0q$cMcQ{<0!ztLvFX*V4ueB;wMqR<GlSkV*yeyRDPO$aS
zTY&LF>URIEq0QU38>^|s&-r-3_H4ZG5Jd>o(hu;6d8%|qTp@*4&58vFpa%6G`HT!E
z{@CBrYOnw8+jpgQM=`ZUv?K~C*lX_5K9qJy9T9Q(u5qVb<SXwf5DkncOks<D2#Y8h
zgv--Et4?4LJu}!D{zH$liJBN^(bhIY)<`@C9%HzK6gCj;kgz8{*FV+H)i9&-%b+e#
zOsNs%z9q%Xd=6J(;*Ylg!Hl4@=I2&6l7eWqT7da10O!hUq;nX=N8Vy=c$OmwkNfwx
zZtIRivjl2$@7l_0^E|BSs+B8;rsu8&6WKIML|?dT4c&cRhF3|&yi__oABNBydi32M
z#{?p>PZi??HdL!XpmVI?DT0i~V^moeFI}RorKO?4Fl1|UuXQ2e;YH}ZH2Y0_NY0#T
z#$oQ~*e1agRyzm-S4f^V%ambq1`=DVNF4`9<SUr%K`#>;5mAin$A6mC$in|*Q90;q
zLANAD9#<C}!jpp{opCvWz@rw)!F91ElkkEnV$=Bl{$n%lU<}ZhYb14sF41ZN7y6o&
zD`Czp%`%9T0S8JjuO8h;-*BIR0<V-;PM(BNmrdksK>0}H)+D5fU}_4Ai<xA~pc0s2
zLTJm`_xC_ela?YVAT^HOqT5!6%NF0m#_Gk37yU%!fpY>%$jHb@EJ&Gu1u-VXg-25>
zH#?LBt<|9^`jIT2&7kNR7?RcXz^Ec}((J|SF&KVB&7iB>j)O=<1s8cFB;3gS(Ju{R
zCB?HKD`vql<KL-6+)JUs>xNP#SI-OUqjt=Gl<az{yEW!{p#1OzMDxfX(}iE)Q`llq
zJn`m%&>H=lC<M~eZO|Txu6XiLNUw0=sE85&Fm%J<>grEGK7`lo3=bVRa8%dS5RDx=
z6VW^>SXTe&a004=$QNfseX*tZ4GNAWy{oNNNjj3Mb@SFO8;7^GwIkqVG5htw_&cbh
z5zy?$de?n&bcd84c95in5t#rMhQ@2Ty2OQ#M^dNV@${yO%2b@BRX$^W2FFSi+%~%E
zTYqhsMR=iQ-9SSlTy9~uV33T|DDg1{(WnlGA_1WpVJh8~f{%n!BiM*|aRs0J1xh2(
zQ!(@{npac`7iXt!2D){BqoVOAu53Bs=;qmSWh|-0Y^hL=FM)i7@OUW0_p`w5_;Jvj
zopPzibXp5nOn7Ka%u~FCo$1W`J3gKTq=Fiz$W;&ztjzOwV+uLqo(H;whlxs>AQozp
z=bz^|q0fqpgQ}}jZFa5Mr|_s6z3-%_qSm?UJ*abZ!%{F~VCP?lnaIDFlV1eh7uJVx
z*?7gPV;(F~gz^-5z+!+IaAeOOI&{t04sExM?d?BtORk=J!*=4qf!`1t@-vWf0EBw(
z!u>}45nOk$oM~FTnLPjmuHO6WPNnt@<I~ZV-cNL~7#s(oMbJ4Jc+|jj2}vW`j6<|y
zk=!9Mga5HzZ@E!~`7DHM7|+>^&x2&t=#s+~<zz<$Dg$0iIfFkI=H?hTlL|TuyP@Rb
zL}K9+yg;?4f2-S%=prLt`gqmgX|c0N?-gpu8K_TPU5ko~8?i_jV~^H$gb}3tXgOoU
z`8b2uXF^9!JpqSPaH^nUp`QPQYif9yH=-8(W@sIOeI{{Q!tG$)Dc4+vhJI-@cO!gJ
zZ-_;F!?_Rte$N<j3#|qC!rcSaEWS_^VCPRE#U(NGnTME!M0=Mf!p-x=-2M*-<IVpA
zu>4=fM*f3bk(f?(@ZS5bZB%McSFTz`Bu-t=@Out-;8BOExvF}`I26c&^~wqQST+9R
z2l@eXTm?<7kdyqmgIUTjwbgrq)89ARKQ<;{FxJ<1umv~z2g#x$zVGUrC(>|Im%euV
zwaHuh?|;Bp^56gY?WGl=qFP!LbCi2Izq6s0wsCv>YCp=M`bC41^9hC0tNH>ZAOM?|
zVC3e0jP%OJjV5Q~Y1WaQMaxX-UWw-jEKwo{eZ%A8h|_Xf1d7{1H2f9mc$pdiPm+Z(
zBks9fyLJ(o4D=ykVcL&WBT)%?z}Es9543!^0J8P-AC=cbfOa04A4+ai14MRdQhph#
zL^7cWg%7kAh2Uq;vIGlwo~c%(+b=7-0J0wCF|g(a05XW2kpGR2jv~H(M;bOV(v5uk
z`SXKEj~*QzRgDz30F;Ic$7c~UYT1wi5UiVJ^b`j#?=TXH0|4hCzCjL;?*e{C4{{M;
z9>lE<-yiu*kt_Q(iLJTyr|Pug)P_+=mx5b>{Duq?_}cs!6%{&S04`w@PmhnMfW`sx
zfinDegY@_N4u%Gy&pB+6>%E3v7e(hq5FL=)JbXw1L6jKcmhWJvE%?`rn{i}a?)nVX
ze}KcGOG~cxo?lVCr3o;xvDuE*9%6#7MfVvG6ufO?Cu%k`GCrE<JrDd!BP@bXf9I?m
zP$YO2AF9<xQo?n>zrE`+ZVPe)!8FgoAzS4@8xlBN5?`AfuUuKc0~9<Cs14mCg3cV+
zSU7Fa+$Jx6r5eiY2O{TAAg_0}A|r?3hnbJ*NFQpjV1@&)@C?yYR_|)irf}q_dbY6)
z?a$picYZ`o(X{ZstFto@;}yll+bTV8SZ(Uo3m(f|f2}^ck}Xb6Iaydhz#c@HW9f+>
z(y$~(yT1eNx*|1dw!C-1Q84FJla-Zqbe>G<M!yQ4eaiX|04}6MiR-Z&bZdQI?=HIF
z>dH8J8BL^v(kuMYX%njUkdkSGf-}wi<pzg;wE*1x;Q!mk=uE^i6%*88I-I)5&IhLi
zBt}}EsSeS?k+yJDG7xxGe4JL(E2%lyPyiM)?H|4&&4W@X`^*g@#>SY3x{G1!I_c#}
zA8`44;vR`&MTHw}MUO5hBUP^-H>XdKd`N+_oFbZ5=A1@8eIkR5{avz4R;^ymV=up>
zD3|fsY1ZY@5<dH76F|cO(w?>^1M(nRGpi2IbsSV)vt~`t6OSLCkN2>8!N4^9$yC8(
ziocSrJj*ll({|upGPn*L_=62M%t>%(lk%s_j$@RSu#{BSk@@Qkp9Tk))T|9L!D$$M
z*@?}G^&IIi#gKd8EogZ>-g2n<Om&riKmhUD!)Hw9MSDXDD$Zd}JjQPvU)9!Y5(<E*
zO-=i7RJAyvK*NC6Dx^!09x~9p#yJUMm&4(<L~37212i9fRr44)ZkH^u!hu)wwfIVL
z2dU9GH5>{;hq-Bwz(Ei_PJwbc9|!DMCb3~)Yyl9nu4MrlIOHtm9i@mTJ4x=TcNEFs
z`6rp^s$gj|fL$<-dEgj*2KsRO*cX4@YM~&cj6`37W6%s7<Sm>l>TYweeg}ktA*=5+
z4*kl?Lr3uz*S^4U=qs^VipO&e1>(?3kAQ<IrQb$IPNAHH`!KYPC@c7lpZ>z%0`Ryu
zLmz@|7M3#1ZWwW4yZG|{z9NnscUsHe-WWI*X{O(Tj{!KGwOvreL;5B$y1|Qx&YmZ;
z-htFs6&$o(VMAKJuDW2X^G_W0VNPUzSl-+h>xnk{V#uzHbkv2&?4X%mMEF2s41~i-
ztDgd~Lc4)l0ny{bX=f*21+Z?I1N#i(MF;u$p?%9QD2TwDKK{BSEI|WTwc&`{y3Rh1
zbRfT-Ks*JEpWeTFH(regW*z`T43m`^C~pCl6N6oAUP(#4fpRjtF+2q}JZ%~3cCuU%
zwcub5m{od{;#dbx3cW~`XD0>8H2dIermKe^1=_8(_2Uk?e0UVC!g{EZ$fklb=f1sr
zNAM;PZrz-mw&--DYb03yovts-BznmeyuQ1+y27RA8;p0iY~GAg4<XYX1tuxO;JTkZ
z1IWhuPr#bfta^ckx?#&qw`By{8)dN5Bx`iSBr$ODD&APj6~A!KBeY7H>#8>=G7>*U
zIILU9)c}dUdv_y+3A)Rdm9wZ)E29wIn^VMLv1C?=xF2=#;_0n<cgy7k3~!3zb6Y>o
z=;+XMaB!fk#^ltb<aBJF))YDSNpsIPHtY6g1wIH5e^egzih!s_4KNm0GYRco-miRt
zcHKI)_cssPo?)1+BHNlK#Otu{7#rSpg^iXp+ITFZanXs-?^x8OW^+s+fKoP$qM5dA
z+2!&U6bCxPpcMDBJpixVGct^n!S=%JK;!D+-1Rm|5vs)Cr>X>00Y)=@07eeNNm^RE
z0<HO|JH$*ki2`p0%!;y-xIkgQ2?z)v8vW_N#g6Y<N3m0ZAB;ZeEY#XhpFJa`1vv7*
z_lYUm>TqQM&S*;)+Z8uXz)cMQ;Nhw6?(Rk%251xvJAMoEz$&4So!%CuWGsdsh<^v2
zAFR0_0=5BL2gVEX<OO*5Brn(bp`@yc*4KDNGyKQ^+Zp^WVxLIfCKrM<q6ZD|0iMLj
z&=6ilaAGCa+BjW}(Gr5f!cEmP*b8I=BcLbHc+yf*NCcqN^hD6c-7Hmvxf4FXQdh5B
ziQcO2*49?g2t7Sy#sK831Tn@k1bj;{L-%?Gx@7UDm9=y7^YJ+!3?%>=3A~3Zwbg@u
zm>;0DO42x96WYkzd7h#dYgx3}us;XwS26QQRfr_>5yk+e^)xn~oXSAo3GxlYsdqRX
z1<qpV|4A~(j)K)n?H-3fl-8|t2LbG0u&cCDLmudvUIRNIZ0JN3)nz(<twZtXEAg`y
zK^;|}t_22kRPEqe^xnMirLeGYJq=A0&mT~XSP|#_J)e~O?pg?UA}F9sJVAYEbkq|Z
z8l^XjM$Gb^DI|)Zc?9$3al#(JDQ3EnG$8dA-@eDz{X8lo;>8-%<FBDXuUeUijt+QY
z9D8;L_^c>8H-{Zrj$@DuF<(s_1%-!WXJeIyckV-cJ05xlpvc!0OV70^MO%e{LjwYT
z`xWX0%uCdsn>n79l|?4xp(SWiU-&b%E=B>+Z5zhKo%6}7Tn-&Ngmf&_`h|CP)LV2P
z0_#XjR)Hy0?x%01KufHg<FSzx9%z4h$RDZt8Q~Q;<Aua!Aw|fk5^9Y)KwwJ3rY77K
zUQgcVURPBy(mh5(mRLXg>lXpq=JF@dK&b4QtDcJSo~XzfXcU4QGY4_lJG%T-ErFrm
zR;(wOQu-21*@^CswR_PbLisj8Stj2mY~I64Kkq67aXbGx_{dWChTws1_1M>$zU~|X
z0jy?}HBMVJHVedAN<vgrOYeojwPm@1G`aQB+q?CIbJRC?-;XqoQv;RSoO0I2COSBn
zBSRl{)PYe*=g=f9v3S*QW@f!RINi#ASSB>@4m5#fwx>cNE=XCV$bcU!_YMoQ_TXUW
zQ&hMxJBVkDp5`@o+~q`rRxKXf*|=ESl2dVmAVw(F?+2oUUHbtLBC=-(Q<KbdFt~?*
zO}p%V0-K|$bGTyBUwD%Gj=w)GNNt1~jS<7EkgFg93(LwHUpi;b<<S}rujK^!HKVmQ
z*f4K<dlwNx5|5(#xBg+m)%koJ5^5Gs7B|V^6T6h$kWS;O+S-BF#1#UcV}ueuD9OQg
zE9j8x9QzMDCUc&Fi3KnM#Nx?}&bWE3!C4|(8lu1}2YG-qpg4FzbY<U*iX!f8=+#Ak
zz}Hq8+KW>|>bSfW6Jld~F#R}Q9SN^O|5uHWc!gTH%b2UN9}nPQ+Pp0<2-2h@OlM$V
zQV^cctCKM_Fp#9qbD7Cw4$KIdhT?MvMkPC9rXI(7i1#DLaDBT^y{UU+w`J;&n^odm
z8*I&Ai8Cg&b8DgRK};F>p?CT%_Wh1WegzQo>O?v9Qy;n|J|T!&g`y^T{cFy`LReo>
zu?oupz(gC8X1-tphNDcJ;zie782pR3Q@z`XIS_!@W@cu9wfNq?eGC?wZ{HH$6T~s5
z(CX@O^YMMd?o<#L!w6AeS{+?oNU0AX<wf{Ju7g#_x)r3QBAH_NQs1jUfID<^`rO&Z
z#&%d<9!8b1(BY&1_`!n(Nc{2FJi&l>h4Aq{@~A|&NG>|~D;8$=BC$+RgFU8#q9U2b
z4&i|h13FCSkk2r7vJaR*OE71LhAW?51(h&fM2>i*2vu+1V8p;~PR?xyE~+uGaKQ^2
zBsPXeRvR;aM-Iipsc=H}UK4sF0<~9Rq(jE`hr`M6Ho;S&Tq=vZ7Z)dom=A<(=*nAs
zOnjQOXAKCFRd?70`4AG|4i_o#qg`v~&Yi&1&Sb7KJ_3h8pnrENCsAkV>gonoqAz%@
z4e+W{1%rfc*W>BB?jvN*K;ak>5Es|c(<|{#XRLI)Pui{D-;#KZP_K0dh#qtc(^ABv
zKF!?;_#q|`iCcV8_3luNXuECqA)}+JU@5F#cV6@+#I?6m7M5uQGezQH8bRap5lI<-
zGW~}mr%~LrkDRu5XucbV)DuI}11nKLK#CG0?PiZ9!%bH1@9+QYnJ2nSz-yz^i}#O-
zASYfv{zzwD1Pzk702N(J%&Ww6gl$gvtTB~wc$X+399f6C1yc)r)AbpkP%>3Cu~Q&m
zc%0LP{i@yj>f7ij%GbRaXYR4a`N&8tob4vYPU03eHkUrWqE59&Hy-S>aUE{nF5JIm
zJ1o~R-GN`a4(~a&@%6^h9BPw7z{;uinHV3P)^heLP!NqC1{iIdnys1^6vwEbyDYIm
zN7_3GR3FUkcclVtNPmDKiqVjFbtQ4^r>y_kdFKM8uVyzAC1gjcA)A2@Tha@psjO`V
zkV{Otht;9*7yyJwgd%MjkvX@}iR3;zl{(HxdEX0;_=q4!VTuhJ5+m1e2XTx@ns4)M
zC<C>jKsRHuI2%!=k0`PwE5wwm&7eiW^U4I0`E{0u=F6A$SLL+Pth;Teg;zG!QMLS9
z;JuD2jj4HWDU2m)$=j-d6~dJ7)BP8$Cc@Y2&wCnM0G=PdWKT<)W-zkjQY|ic4Y}9N
z&6?8{&-)aNLfeAu^jeb5Ax*scv7xGp@IIi?E9J4>b@=z90yj1~n+jX=VwAF#lL)M-
zvNvG>=FML7I+EAt6NE%C08F(;)`aT~VXSbbE#NZQq)%<y1+ey#($W}9e|&oI3u%@3
zz#e%ay9*t#G!B{k+G#DQ8<QD*w!)^%y0;Xu63%4woqNxqjYnQj?~}O(KN?#e=6aBr
zfO0?6K^jcvDF7!qPm%keDHl$jYCQDVp=cZ3Z1mF!|FFy<jx(g@<_30x4t)Fe4Yp~p
zyIH#z=(P>s!;P%0(&rl3%<2)5p^O3>Ccw@81=|a!knyt_GS@Scc5j`9khZoO{lF<f
zaKdTNIbRb@CQcz3W8?{ir?IAHQ;(LiWP;MA(iWpnpP%2IM#v3S9)!YRH{S+o>P=($
zX@^R7+ZY%a!1>Zl#$6NUS}EJxL6~1<L~~WDN6h{Eme*w?m*uX{4IJY9Fw{|{wR8eC
z*8TFfM%{=s1vCB%ze~k8{RI-pNWuG8qTY2&m{TmSSH65%cydD4YvCybL^yvBc`@E1
zP4FJ>)+Z1(QN#wPwS~{H>JfYR*G4!KVC8nWXcf*!98Lh96EyP(8NMQ%yj+}d<)-O*
zAr}MZxY%HDYgk1Aykek3z4yrydcabkCKzP}91Hc&7#BithLoF07Hvo+Gsly;PQ>V4
zTyw$Rehe9?Ka00I3eAV=J0ofpB6M2RA@+Y0$okm$-rc+7u<{$~ZXkrY*Kf1{96;9*
zbVz9Iuy$}T!P`UD1J{D1B}dL+y~ffb8Zby4#{k`!(3YgY`Wh)WZVIl9N4cVJa5n;x
z7aaylX^7o`wpyev0yt$zFVNKV261U^MIy;h-;C*N^t;_b+TNb3dGw;z@D;m2)(`%C
zf;vAvMa*B~HX>$yEI#DK)mK<<j4j=BS<s_4Met>!=VRqP!*Wf?YnxVj3X7d)HQLzy
z^g?c6>T*`*+`#0mGt|h;G9efuhH(NBHmM*gQ6JyOHWtNjs+FE%!Nm^Mz>jXm-VX~C
zgOS#>Pm31z2<DHp$Q>jrGx$dRS!i9y>Nd!?4qiZ?o*e(t*6&3Ele?VI$!qYD;nb;a
z?5KkA3mE$`!yoO(07kuX4d^_W<w@j;ZISIaNI8222^o}4S}Mo!f<U#Tacv|5kgg}6
zhKIXQ22{SO<QWH9tlm=tfRASMID^tJ;nF8hSW7Rc!G~m6FyS2<>Jw$RXfiED+w(Bb
z8U7(eOGZakCbL>dIegOu^whttEz#=|IdO|$P%e49zywgj>_IY$gv!hqVz!^5k?sSu
zkoa(>t|W7=EK6G!X<IcKSk^_qI>)uHlL|?JIzK}u#0GffQ4sDhxtCcA*S^o!%oKp(
z%ydVNZN{uVl#5(LA*q|Xhpu-h&a6f@LcNrYn00V#!Cn|5IMP-OJ0P35$!slNQicOS
zBQ37^%H`D0!sNw2i_7{PCSyR&bTNkO;X-X98cAQMY^4tSuhr&ud2F44D3ZFJiTMx`
zK8;JOVCmO-roaBz>oz2n8o&CGbD+@fMcHjE5of)xe@oUGX#jR@8qN20k531x3q+6-
zuG%=YTn$86eD&KZwZ?Dso1fj$6GRgyMbA?bpQ*4Y1DL_Zgz_JE{DyRuQysTSuwcH+
zU27&WhJGf4&c%gDR29%F=_Zi|b&u6YVrP356Cl8%L<;M;0def>>%(P2%Ony!+kJQT
zL|913E(7~}%&}ar?oLl)749&LDC(6Xv4IV>k-<h!9sgFr^`GMxBz8>y2yh)f3F>!H
z^&Pk7t|H2++`w<^1MC9+W_M&gN$L6Fnb)uWH9eZez*n(@A#dR^Y^nD9(8fW#cvYKJ
z%|L%2CnqDHg@T|I(yB@+Agn~i?xAFE{s|{Yd(qj^5oVA{E#KCx_@7()vF9`0>BbZN
zqjy>So`xBFE;~vwYWIKq{P}Z~T{+_uy<Usk=;@I^Cc}G$4+2N{K+iUq>}%a}1>z~>
zE;C4x5xWpEK&lt3CzaoeoypFqdTM?0!GS%O??y!ELdW&nt8+eL)Qe%p8W6V6biL=7
z^6}570Y42yqcBRT+W8RoQH!VNK}snr6tWs#n-fjOu?OL?;Qv$y_m;O$BI|U@xnoFZ
zxX=dNFV9L1{D`qluz{9qtRq+c*D7OcPMXx$%n#qLKSim0ztu4qRBXlU@Lo!*F3IN&
zew7?cROzMvUHbb`ItG{x=bD9lYiV?JaS=NdryAIy$n;7)PeP<fNH`~W6DlF0I5kK#
znmb%IxI)}7Ud(_C9f()B$D@l1JT-$rHko(#?w#6zlzmC@rD~09nxNjn*8&+rqI?dB
zJA`q-?ES~2)2_4|VQ?elr!BqAunsk^YjQozDT1sURG8X(C4P^+54Cs;b$Nu76U$eo
zy|8(y4|<rWom%80Vf+XQN8omuG6>xubQx+&uEm^rE-^FB8xtpgp;|)1Fg!3If(?zr
z9grGk^y$}1823+cy5Kbo_ip{sO8+Nf*!XOq`1<IA6zTqujDhr|So{Q_$*n5jPkkg}
zl$Df~H#1r*$-q4cLkwwXT9|f)e<TtqzR9^FrJ0{wwms6Xhxy2OV+KG-+<mZ^BDg=Q
zRskvIa{vNZMy@~yw6>T&M5*GalFVXw-5wEh&s%()fOC*hB7n0Dii?ZGnu9)Td;GIK
zW@doGTpaq3KU|<Y+i`l^u#=zv`!}t;x9yy*@Iq$JR*5UF78gFv>LmPm3$Kunddc^6
ze}5HN>_W_pbe$uj7N#%&mynF(suZ=(g6}-?m8uPgOdRI#CVz;+5R;4v`MkiudZKX1
zaEB}K^c$#3)4Wg?jNrl(RX@3SvL?XJ)Cja%=ne&YUt{x~<sFW0ZgQyGai&52XHAG<
zp3x6fO;FiQ&de|xZEGq3oJm38#;F^*qKtAcJPAv^I(xU}SjEbIgDsC{9a(|W8LTRN
z2Re^fIo_)}er#&HD!<#ksJk;QH}yL^Iwbe+hs}`mzJ2FWTcU>BVzlctsCO35N_QpU
zNIoJ1;sp&<c^LBI>DJYJH#tUR%@bN$T5#N`&t=sNP9pr>*(R-%9pw9AgYo;eXF~fS
z8*4qYqD=&eb}43b^`8NSFx}?2;WX=M)@{!cQ0#B+s8-gqJ@bhh!|dj~iv`zhlhV2`
z>w4kBCV{v_EneLEsb1NE8k=wO6{bBID$`5!ft^a4n_S?}7_yr`m?q@XsU{$qNasxv
z*Jk961Wg0|!PU(Vt69rhzdwhNS?Jf{viHBN=<u<+mz-dSfstY|UNbPw=}uK>_wb#F
z>hiY~;scpQ&>FCnot?;*NaHmiW8i8UJoIVI8i^xXe!`!%HG3Mq4-O``+$!asURxY?
zQdzlaA?%bJ@<+|Z$Ivjtd{%(lui<%<u#)nWKCmgF-m*1Zku(OEg_TJm`t0{l#4Q%i
z#2V?;#UlWXw=gn>OtH)BY@*YD2?_@$)|`f8MF3<-a)PpSdenF5&c@20iN&e_1hns<
z375NY;Z7)$(6mR*aN9KT^xf(iDY$P(5l4~x$QjlfD@e}azIBm0dDcfGw#9Kg@e<V$
zMMcIQ_2xYs^5I>Ihn){!wKVIO5#kyPFSy6t{k#kfFg%C+{&*xk@ZV(rC$nsCezp5S
zvU0NE(10yP0+hraL<2w4$2RCHh?>a}n#tAQ>>npVPn9y(gR#p)K+>SXcvtRu3H|rz
zmQU?%OPli?#Ev97p37u>>va9?7V4kT09-<%tkSc=JG#O8=^2W_U1b@!_Iijos;UOG
z6vHyY6#qu7_tjp?cANn9>;sh*F!T{|X{3VnKxHMsRqM+u;NP=F=Lu}0ETa=O?W4tP
z&Z@0Beg~)qoqi8qEs9wTi%wR@{5MyOdk9?YliF;wqXd%N)uPj@kp%=<Cz?Qvc%CuA
z73VnZ4a{@sb$IB2N3G92E@W2TQPuZcQPK9yYOdejO*XSnhk97UW3HMw-h|7tHi#)R
zcTA-S@&RJ!So(fSCPeMuJeso_RdAv1?c2B2WEdZrYRQ!{l^zZ8`Eu0K?A*C_Y<NK4
z$kHF@?3R@+um1AF6c_;CWm5--MFa!O-)|G<G|lVM+dvpru7PqJTo7f#kMUiOnY>v)
zKG^Zr1#k{_c8gzj8Ate%M*HvOb6Gi;aZ@BtgW}-fgCr~SgpuemoMRswPg0HG)=+p5
z4RRRR7rH9n9551j2LQ|42cR^Ipb5s!ocA3Cy-FRj7TLYC=M6k1OS@)>S+$S1ybdrJ
zOBj`aBoTWHw!$6CPzw82XO7Fa_Mr0$7p__6vqLBnGCiJx@acK1g?lIcvt)NyR}hPs
zvfg9Km8)MNghqdSk&zo{wp6~fs7_OJN<D#iY9uj?Fn4EYKdUv}1S`6mk3hshK8gO5
zktclv?MxMxq9QtUn<nS=Q?-Z^-{%f44j+lmRQrwo7<zZ?*j;OG2-=XLbX`i6b*m9}
zQ`&$z-P3xWFGZU*>i465h?*JAN84W7nP}`t7(b+SvY}LM#lot=60|>dx>8AC^(FMB
z5Xy7K39aK}ni6nE>F*_mR!1tVS;SB35F7qx9z-=}$@B5y%u-O4VmYkfSPj<T2onWH
zCtCzsI^qwlxFKB%|5w5CUx%l<eacYaQ%gkO^W|6l3Y7%gMI@)lg~VtOZuiE}eFZ@d
z(}lpp%|THdDiqIZA24X`fD-!)dO?lHNxNks*4~`f;c}J!4KQy<*juk?H252a?J37O
zt5}?O*mm*{ys5Od#E|v~d>pgbj~lmc?55vYlq;2ACBAwMxxlu(78ee=a@hCET3k`e
zlBJTIoED(L%L>$%-{KQ_nR07AD;FPZonos(;<b3aFay@?xYOb}irQ(^hnu{Nsg&<x
zqr!`CF<7e1W7@iPLl>xkmX12*nToxm_Nd;|VgcvT@ArDu*{MHWQvKV6E8&Vm{1DwS
zu7s|wGK2*mZAHG;?70vslx+Z}0hKE)JH|;&7583go=9Ze5!3mwyztos$|H_HtO89y
zuDB13fbN8H6K53NDad{P;lhETcyD<DVk9|=)shR<h4y@8D)xFBC<+8q(z1gubkv+a
z{Ska4cm@&6cRNrR0kgh8uIslGX=)~%D4~3S@SkS=`X1mX$m1b)h{PxXDC3Zr;A+J}
zM1Y&wXtcc;HZGNw)LXdLRxqH+#TEKVKx8^R^mr5k0D+6=JD_{tLwx^D+j|_)QgDI2
zC{dQjM?M)V#c3EMFId^y0ebG7>J0}!ygGipoP*Pi_=aRmbaj6D{LHD=FIvz=Duk@c
z`?%@v8{9)z$I`u(v7`F7o$9NCI5iwwMn*1f?zy=+h-3lgiDN<vmIw^F110e_5>jW|
z7X?e~hCa$2Z~yzR-@L&%#Ti^EfJTAQf$u>t#hW|q?J1x8`r;RybU{hD=-;y>*ttFU
zQeEu?z$hBe_;YcsDrIvPQ#}<!5ZOKZU>|BFCliDDqXQt3DN<Nx-B!$5r8Gs&u3sS4
z^M_DXE~oS7ftAJQbm79)HhE6qxc8;IGC~afG9v%g0)&Tu#>64L$5;hmHCx&+A0m3z
zmM>D(6^y%?SnE!r1^@+RusX#bJzW|2q|(x-(ykv)08({h*bisDHN`ZI+i+gGACRiT
zDbj0J6(ma7lqU9FCWR22A_{58q9m|nHo}R+F$U`aVV)=*ncmwP2e`-5>c~;xm+9ZU
z(A3N59_}HbMKD$aedjasE{WNNW8JRQ$a{}>m>xWMHRmjZ(>b59C49tIEMG2~20ONs
ztACDTq!NR%uCszEl82opbjT>pR{zX_mH>DzgPm*LeWBH&9Y7c<ZK~4NWM{DE->^V9
zE0_|;ktlg*4ObDfGZN2<c;kr?D(G70gPy^0pn>Oq#4Xm<*E3{B!rXr%3^Mjije3K<
zxbDG|Jmt>~T(z8)Gq>K0c}w^GKU;?3jwe*|0{9dB2ltsf1sJ4iDIyTdGEE7CkXa$1
zYtCZi*LsF^n*B~ITzd60Hsl;2^vEGF43Gf8gk}=rW|NZh95ONlSAtK%Z_Edfb{>k4
zve}1T3P|QpUMvGkZp}UeOa(0_7=K-HabhP__82=4SuPr<aIu|)RseNI!Ow><u1l-7
z1^Nqy<;1aL`+m_4frVekBASAa7l#2isO`bpW<_mshYkmsB!-p&8!E_gBsz{_OCz|6
z4nOGy{sNT$GG0AUdx(62|KT<K9BKcK&kR?ZW>4cg!Py2-4rpI`g;7?QVTA*D5*xAt
zz#N!LC{LlvM7~&zU;iEGa~U}nxHS#lFhx0Lg&|mBR?$IrFO+D*!)gnzpvysy38)QU
zfBkDI<1FmA)iEjANE;aAEV7T+0s;kW_5JCOpdkuwd!VTr*{kt>o&Y(~bgnntnrFc#
z<DoTRHwqx|Hw+T_JJ*cB(Mh*{02VsiB<u}Iw@l*$Pe4)lG~jd4>;SQWu&5|HiObL%
z$&+DaV>6fSHNO|VZ)u%N0Ua?wS=IK8p6RY#MSH^GoW~dD)6($a!xaE=iP7g7ogCiY
z#f^kQUfDB35if)wSM@EN>X3BMjhvlWrl(R96SzP+BOG=OgN2KWh~Rc9h2Si8m7O*L
zQLmgT7z*G|*4PhZn?G<Zw~$qWK}W$cEChY*a;r-2%alX~1=l2Ws|@$17BqGUFKBD>
zhT($KuJ3O0IWHSrbIun1GX1Axk)*a=@iPu70+fYFTDVSgxUiwF?hghxJA0J@aXs4*
zvq$QvqSa02Q($)!)B;Emk-Y^4?Cj@g2k5`NGZ{iUhmfJ5wru*~2yAPNHTL(fUu(Z?
zkFsi6*ST9>yw}#*Z0AJ`jlH@4{XU(zz$`6Zuu8W!uxHI_4uAUe&|-U;pZ4+<r&(QI
ziA(W5$PKh0CnivpYwuaje7FoEFPiQV&qvA)%|4S|=1RIwwj`ZSC@;?EyY0cnbWBA>
z<lEuZjf%w{wl8g`?)kzYPx&sWy*SAWhpp|C3<cPcav_Tu{r0V(;5=-+QteJ5;=u<k
ze&G8!ax$CV@@E2K$1&3qlB~-2nmaJInfxi|FnaAe;AJJswb#n*PdZmf4d$shqc1eo
z<c^lDogHEfNG;n>EU%<CkM-w+OK3I*HP-U-rF?I`a)baH--mHu(Mnj+6CfDpj{A6&
zd5zT}&><*qiL<EywT^^DF(M!7Sc6B|*w~nq@!>g>P%8kT_Fy&v&ulajJY|IagG&q%
zKx4G`1-|z@@FHA<*hp_ZreB>-I$Ir%i4+&WkWX(0*a$_CK?$o4cSJ2>1W5-0iH(Y3
z34o`APg}wN<DGWKcHCd!2`k!8G5&LZnF~i+oMAPB?z;#01R5h)Y88Fa`$$VihkUch
zSj9WyIY+Ik=rFEuBMjNDcH4@vx_DHgwYhn`0II+B84B$uJ*X~96M-W(H(oUz?sSmw
zzP`b=gE){U0Aiz+6SHzP&1n%)kO<7;{b07k1)Meb1>!T2203JZfGsews3n&hZK1Ll
z7)s&;Y{d~qa2U^-n{Rdf{sU!A*1Rs7ixEsf*rR7^5LCg(j~WQnQ03#f5nmdwR;?dL
z1HhWqtKnWY2_^`7gu3eL^G;5{e!I~6aR9l;in71288UNcgfd)J=(YZd+so&q7W8yj
zZ1gqa4_iQGqlZoZv&z{71Vi+^zz0rE+YyhCDaM#Q5H-9RPrLvR-KnUnshioz*JVUr
zfVvBa;7E|bNHkPe^XO*90INXOiqf(*S%qoKjE)Jm41jP?Ev*)mxTq-Ike<N*2^Bp;
zGSVxkKj1H%NC*vS*Omn%g?UEiFfu;3FUS+d{TMM!$z1gF^@VvNrqpyHxZ(|BG08Xi
z{CQ@v<H=9S=#f7g`p#`#J#b>}UJW}1CiTOc6Yi<-L^yXh1?r7D*cAJBy)!C2WtXXK
zbBbe+@}5DBds@z|3Z)OLW6i4U-hcl2=78(=e5Zz=_va#OfBbI$by=>r@#nnExw_wV
zHs$zF?=2@Jzdj5BpVfiA_|4?ICZJjTm$6H$HgKsKRt)06N|3>3_;{#;(oYqS`Eoz#
z062vuWt-TX_D9C%aZCmN&)9u14B4E8eJMO2??{i8+^65-gj?ST9(8yV0HDE&Jw@RQ
z-IzNNCM;!|=Hmu@H;hPRLN^WnQLj6`zh6Cn|7hd=Zgb5;jdqj5xzv%}pUmtv527*r
z2W-;hW;zJ!{|nw=L|P4k43HZ52Mhrt=<-o4>z>Ckj&1tqj6A+3Q@n&DDmCdC#`WVa
zPu5^l6FX93)(!_%<gx;usaa&7;N}Fb5R23dGNGBp>-X=}+1-xRGojn?N^u>N%O3O2
zisXpIq^YMik?e-y_Rufo3Z%8Te+oXD_mmcp*_8}`-t)=h87lpI9!q2XP=3rWfaWx%
zHxqMGRFpJ4WI0vtHT}qn?51CpJiaTN{(+-`mCQ&%qEK;eipp1Ww71-j#3njd%HW9*
zjJhZ{?XhzdbgG}dKze_+?;QD<O#i@T&w)WTo+(h~SJ0Rdc?x4P*cNKJ4_*-}-b16@
zf7F=#xs{BK6YJ??0d!PCn&p>&Um7drc2ZGRe(TgI!<6o!OaLc9!TX$Q<u}N*z>`3)
zRZ~*AUmvS<;wO0Tb5_h4N~g-ZOqtp#`u<)YcK77*Q2FHP4EfLEuR^zOqdNW7tzyn7
z{IO9Wb~WlmJZXV`fu;Iq-6yEVAx{W2lkrLV_hB)nDh>3zxjhQKKJ@KvOZT{{QTd*$
z=zo1&6+CXP7It36{HR0ZDTMQJ%xkJjd(d<ivOs1N42qmUyEAkn{{Q{Eiz43wZ+iAm
zK2BV}f6DSt(oWGGy#Mo|>+No5ZaskIt<E6A2^chFC5FEJ!&(iAW{}*%udR%lmVZAy
z9jG{rxm$!M-kUO-cv0bE-vo_c!MZdj_iGt6cD=vZu^Db4$?a&qVHw$cI)3Q`T2LuJ
z{-1w|<#g>oMphfK{GsQ`;GE`#bX+dE0=H}@vX&j(7<cb=u-kenKWU*QV;ZWJ<eTE_
z;Ew<N23d!des^4!9Yho5>=Y!Uco9_bpE}^0{b~j<RxD1ikKsD1gWEB6N4kx3<BHSH
zeO|bGQL=8!&00>y^BbOn#1Mo0Ln=vO|1$DD|92KQz62sBy>_Zs&jF<YPh=h8rJ~x&
zic%DGh)6MhFgF5=bH?y8s@qwJW57^zv#wv?%s#UGYDWrUUWN)L7i%Gr;NwTW?*OJo
z)o&S9d}#hZGp_&hqW`nc|F?GH|058@|I#BaL80gq-h+sBCX^1vQEgKml(V?9R!!fz
zb4PBppn3AKB;55(E-b}3G!F3?p*JN+c6JEy{twIlVDHVtxnAGCU)5?=8m&l5R92-D
zi$-NgB~41BWN0lat(6ERvsM!!8k9tc(wsCI5+zd+%@Gj_l?qA7@VqW-{r0|(z5jXs
z*w1n7{p`ErKGyGU@bz(huIs$s=lgu0@AC>ZE(*X8Vjn{rOzb=tdoDrXvU4A1bpf$Q
z1V(p}n7P3cHU99Qf9@xLZ0+oo7gxuTtTj)jJAT~=VF6+p32y_4%nLqA;9V+uQ{^~l
zY5N{`k=QZ#EI<+l;_Lde?#WlMI{4?03$lZoZV>`y&9^Qs$>QG)@BAK9R1@CpL}K~Q
z{s6!aQ<Ap}wQ*HB$OR+evman7E^in4bYE2o3F%29I1ZTT(&z5S=NA|UQ}u571pXkT
z^B;%6@lB<3;pMIVgMW);Rhn1#myp;!Py|+WdT&E0Ayv6<7w%O?J<Bn`w>0N(j9f*<
z9#ztVi~yF=JkC-dQ-YLW8Vf@bg;9ggY0yRZ)+h553+m9T1)uE3m#3rUpXtv5SCe&n
zw|??q35jf-&c#%}$s;rPN5|k+7`eC@SJmY?-{#-XimM2iaTzI?=Btc*`duWHCv~nt
zql2_xRz}7<+hO8^x_5q1Ts)we7eg6UdBzC<rWj}?y>&iK!+Foecdx--CzXq~u*OE8
zeIa>REQ0bcSLFh-va*=v93Sb&BKHyBAaY(p&DER-!D?RVw!G*|q?#k1@lfwOe~g1h
za9X{24L+|i;Wg@+!H54W7TRcqyuEoaEolu<3R{R`JGORy+SzJy0IGh`TS@V)L4v(6
zDG>BJUw%|H7cfbl&yx|Mfg`$0>~Q~8Y#84I%6h^YTrH_ODoq6=w^{!@S7L2i?@iCo
z%h)dmKO>)(OyrXWh&QG`5Afd7_-Uy9Vd017)2s7|uFvt$`S;b!xQbWbc}>i(CA%%z
z`qZ#9_G{wA<3=)&JAov3dgU=I%a#2ljOqp7^<tEfjnCu1-LFtNMjbr)#Yzc@QHAKj
z0FN>v(kt(0R9fd<@wZqTeu*3WzyCl&;(u~IX3iJ3d)Mbtp$*H*8#NnI8w!_tac5@w
z%h3Z_g!YI9XKKl@pdi>j-3g3ggOd3OyM<6)`s0Ew5|*QNUeLZja~1Dp!J<WdUbQgI
zlWakCMg^f|J;@g!pznuTy3fo=7!t?E`Rr@=U4IjWA;}lI_ck;)CZygJb&v{N*r^&A
zHG0H|GLnWb3<+Z7cri*KACL1yF>Y#65pc}Y6p)W}lwl;!w8BHafAXm?Xx>Xhz#HK=
zEkErFNgl~AzX%k<vY9i0tfrZ$fZ@)8GmhOPlqJNz_6<Qc9B!W}C+1p9xmkF8d%jeo
z;mKI2Fm?4f&b9E?W@@pr^cse%S_NuPL20}?cy3u@b7g3RbNN2`Ps>P&NWL23<6dow
z^<Zj%qenFaA-|AgVFiAJ0*1*=A)F;Ci=d#$&+6#WlLKchdgOg0D3%P1TY^LOg!tlR
zJu{@YGfTX3q1x?9<UG*<)5pEIVf1X+wr5L8p{n5*yX=oRcFY-K9?zsI0vL?hwl28E
z4nHFfg0wF4L<GEP(7~0a`IDWO>9lV_5*+kn<R*CMS(^9MH^=G*zkUX>RCLu|z~zJh
zXcBs%goK2y+5;qJI{9!tD0|{+wI$s^XG4|mHjNovSSp`pIjvy+JNT`;M-b_u@bV(Z
ztE(HN)Lp`I5$A7a+*Vc~c!KGajho<qZedVy>|0PD9_2lu=;8&R(^E^`C#oK*3U^a|
zaatylQD@GSQ^`xTIR9cSGz-Q%(fRgl`O@%4qE^08K+W15Z0=Mj-utX+JfC#uyE{$L
z)$ipFkGUqsKhgFqH($uLHxP<q=75@4a}JcoE~n;uitR?nWPX_qi7IN_A`zxxgMQBM
zz|Iabvt`Sdvzz{LZK>sI!e7RqI35n0r@sGQYGTTW9BxpUmBWjtzh?`1Rfbseh=bP0
z>h1bjbQhk<>MD_KU>nd)A~HenTF}7vim)W``hKhQoqa9VFyRO0qa6&K{^2aYS#{QE
z1bdv41X&{x0n)OW3f55-lX^y`W1BHPu*Zar$cAaDft$(^yHZGjxO>YyULhsrB(Pxe
z_Sac}Z6sUVE#x?^tQ@LIIhgpMAk?>w>^>EatoSL(%p8pW`n#XS#>LFE=CS#*=BcU8
zhI|d6gZj`{1kH<aF5rV6JT~Gk8$4~x@0@BO!#Z$)P_S3fNgPq$Y7uqr#l08KpYQVK
z?;e=}!$N6!IbDPvOap2Mk>cFqS7fPS3RCZqAj1yxGaRKzBi#9K7TMln?2V%Ho4>iP
z=G1h&+0{P<5w)fVFsI&d$jopP1DPGIgpkX%fuR7Q+DyT;fAM^3ICcGd=H8|FDE`5w
z=y!g^nTWkSpDWUJHt<mU>jnxgzm{7d1>9=Xh0?tyisdF%dT0gP?qm^1wy;n&@z;&5
zkQ_?YX7C2m*Nu}npSKy-D3y~_E*wU<d>tBsvfUMhO}y0hz=UMa#u)}*L}L>aV~>?7
zJb*~gL)GTdhXbrR5h<FjwiRuTKb_i>-PQBU6}+Zw6qKX*p^0zW<tL+60>l06vW=9t
zG9H7W_d3;!c0F;I$i5NTe^}3|xyFP5s*h86RlhB)AO!qUGBG)BKzn&8r~K26mtu@C
zQ#1Z(dcmX}M{1U3pJwmv^B!XqXE(wcL2z(NlN+;s;rgKY>n&<&D%v@vdz<=jX-OI_
zoH$;BNLe3jK4|^smOdi5P-<^w$kEHQODlX%^xj&Vwv{dIN!@@Oq2CAI_m)zR@7h?l
z8_0j~pP1Lg_8IAU_|(Ehi`?N)EEv-Oc6xWU_$<TGuhSps6G|2rk($-Qhl<dC{P+-i
zaZS)46_IN2lSKAtfeA7)IhaRIzw;e`IlDgW9=uR0^~VrXh~C!RiTTpb_7q~Ip{rhC
zpe2l|R6fz9%$&yGo4O96{5f#+qMl3s?r4lpoxJ(gB==U`_PRg;+fst?kvSS^sCRBF
zaSdJ+(Ll)4ZB>lU6b!^v*vADCU56$A#&A*xXe>gwre!`m`djs5r3KytmuT5@hzgsP
z?w$^|Xc0w+9xE&70F_7H6m`tfz;E}*MC|UEx<pn{(TjOfl|Q7qNL-!W8SQ99@fY~S
zabc52HAg(>>PsGg%cQES+UHcR$%eGvYP(}62@T>}w@)|9M6YYasCTJv4=o(oNFU4G
zApPGR#x2MjdflS26gq=w{_gV@h}UrZNQ{P?KH4Q46Kb0@*8Ta3;RJy#k78Ju6u?+v
z>Z5j6(kG$7LjhYo1!OZ)Q$6q(YW7SpQl11k82Pa{+Uy93Dzfup7|LS0gP^ZX;;l`p
zXnAK9pgFp=z9iDm^qLh!AqD$#26DN7PONILoSJ%$qk*#|%P<6zAL5P9P<^u<QM@v#
zmpeS!nIi{H{Qb@jZrm6zDtEULh<ktMZw=FJv|lDM%7vO<jA7Y+h=Ihiy+i9A(kJFk
zt3bdIm~d`p548sb(^J>-b!L>uP?Qea_QGV3*^-3uvB?_crDJ#$KOyWG9+1F?aHc1f
z$B^dBrnB+^Dlh!5bsP1!fq&NnsUvkFe)H(QPh69&E+-)h?OOuMmejXWx|@B{h(I~g
zbGN=rdNrbIGMa7aGD(zsht?IIA<fuVz~s6z9-;Ra0nZ!O&u5mzrP@p6R)Gq`D;xF&
zaQg=~O?yD-09LFs88e0xHFh}=HifU9Fkv$@IDzl6tQ7t`MD{7v?m0qpYeZLm?Q>(v
zt+hrcbuA~EJ2=#Dj(`1vs7Sj#cXm7TliZB*Z+%W0pX`rU5`%3uHU?8*<6aSYiKtKj
zYl6luk6Ljk*uY=(c~>__Jysla{FZ_LhQSjzEPC{wX}6PKp6MoGGN)N$kBa*(iRPL9
z{bsiN{u1py>m(s#L{3b@q-5+0_>pt2zGoXrz8g0;KZO<D6DY=NMYd-R+ipfvY?CUB
z{aM7GxV(-ZsU0~P<BtwHx(5G&I7nA2pLW=crni%*(w*Yj)Fd)PqYhKM+mn;M*g5w)
zUL90zJzm#!n3lBPU~Q_QLCY84f1;~A-;0=XF9*To!bfk!$2fQ3R5#<S*JY~)INUj{
zhnK*jvrnSFgk}wH6Hhq(L}`0teBz8@5<O)b?7Ni-ls{8_V2bJj*yknF%64;kJx9*q
z_yeajg5T;#-+0QA$#ML-iSPX@EFS!~da@{S;GJHYR>VZ^Z?L3Qmv7Puvo5^t3!s84
z34E}2obhG(>p27fsS@YA-DT(0w3fJi(XUCR{}UWMsa&;ewoLUK^|~{<ZGZA5CPQ`&
zlKl&!$XU==N=}Thcn{fhuShheM(#0k5dKhrHq8qEiD^?k2;48tx0~xIj4$T3?wrCY
z9aMd?i^Ry+PcMD?j}}0%iAK3X*2}iYEB-}Rb}E->t+!wM$F&`=aA#EL|B}tirDa83
z77%L^Z)=Xp1S#@F>!;g`?G4h0?Xqm(#G^hHs47|8>frZ2T#2{DW&Pwyy@O=oR5XRO
zmS`up0jvWHbaU=4o#;j=P&)dW&sGV1sPgD;?@|GM79AT<(vErO9CwEe=~X*1z#d)x
zh5<-5nB(3n;^U%{%`~5b8<D~iPBq-0auJEhK>?(7`l>l)ORBp|_+0vj0MTrGwtUHY
zm)|M~D`wP&awo8Z#JDa;4<DkKs8=AN6ZPiLbA7_A#67@GBPtC+I-t5CME;=hj$dCA
zlyEoe@Cn^%vAe%c)9{Vg$>(E{nkG~P<UFx}FaQW4@R!MZv&W85P72V|Zpp%haOme@
zQI0nH1RE>5)x=5l2&_Lor!YwA%n7GgC>KPVs<&MWu$-H_i3nYl^LNx)GL}a%$sK?6
zkr*y(Mi>3ex{1IEBJu{?`-e_)v>4PxCcH9nXevxOSfri#R!2Ve(Uz8^&N-)$&{x9S
zVVjLv5CK@IVmWiSb<wORCW$lyhM<3G5RK#N9}u*C0p3LWhs1?>3=(-$*j=k~!b`7L
z*F^-$^f9*R@%lk6(4>3c6i7@T+4EJ|ppJNy>^^cWKw|_72R?Qz*L(s!9%;z3sO=^#
zzA7?vqgHWu((d#4$4?XVPYfd(i_xx^x97)*<9_+pXUSJB;9s4FN3*+|&sLUhp*&3)
z_8GxGvH+FS2dSb!XY(ya-nlu)q&WvtbxWk)kF+?gurytFLQPTp7eo(ww?v1a|NCoS
z7-4oo+s!Fmb_++PD$H?l**cUJXCd*WiU5JY%w^Li!%K}8vtm|rsh9;B31c{SMoTzO
zbpXCF=kIt?lEZLxjvJ+p0GJ+(JtMp;O8X~41AvjlKNU$TE-xi)T<#MqLhw>2q4-X3
zRMM$YKRdg!3xMX>JVV32uY%exWZ>8$Q-T}8fD~8Hz2rCKWE&h?x5O7a<UBjKT&ZN1
z=yqi1!~c$WxvOd(^-t=B&>ZbLb7drk|LD9r7a~b;W}45=n>y%T5<OL_RvMej#CeD)
zzy?MZz<$8EWZ?I;$eg`C_iF&wZZIj+zQ#63k*@i`aly4{-l%=Y_<!?C1^{9EhDnKd
z0}A#BxPbJN8&FXA0#Ei`wIUr?x`~GK#+IMeQbwwN5uJLV@JRCj&B|lSBiyzEJX`>j
zBR}h~J0XQ4f9sfL4-VZE`#9(`x{I;bHSqTnoSKnY_X$NTe?3s-t!sf#?**q~V#3Lz
z`3R8`<5{C&-H>Ri|2E1`wI!7cpxVRpYiKF9Vk2aCRo#jlm6)&ny*SbwIEOF8)7nOA
z#~pm(o+-}iGS}S~1PzrYEG%VG71k_Y_56Bklf99K`Z#!aq-^3IV-Vg*+v%j|0J}9C
zkNjs7B&@+WkJjTsCG9-guQZw2=8CKDvYBSG2TdKKzck%EZ~DiJVG5^7#ygtxy+-Po
zYWN`>Xs7a%|GueNL<^}RO0<=c=C=ITXG{~U!Mem}Nnnf^R3h?mZAl+5`xUT&vOziq
zOX;KF9xxMFEEH<-GcknNR@D}Fx=;jKU_J_4o$?o4#;z&K5(CC}=9+P}y`K>M#CNT7
zy(3Z=8(7@Hf)W^#u*~ZB7jB4Iu?vMcDs$-kYTE;~>gPeCIBxZEm%V;$F4?0k6X0tb
zS@tTDpqxsh?$e#sCwsNG`h0>45jha|2*pldPF3{)G$*1}l)A>fsl%wlksqptjVX=+
zf>WaEzzMFLPOrH&g6R;wXKS=2BuqN9_Dl<CSgNe+Y(vO+5o<VVzhQqPb{JezLgFh>
z-@=NxEQJ|d7-0(K+<5yFY;!Ngm%O4jUi-0{T7L&;c_r>7XL>v^5UPSl8BJi8Cg2NC
z9|o#L&k;sWLCxKhBdMfwQDgGtZz0y<V{`BNokLd=3!W*D0Wk^{mkoyln>v0W{KEP~
z3zQ_kzm$L_E)i}R_ou8P^t3v5vBl3$Y({cVEYzt>A4nu`A~hgGQ{$v;Opl-ZnyoE&
zy{W&%%s=LJ%f7~2%53qLlNj}Yk4V$CC6hh;u?xhbY@L#jUX`7I+K`U>%CUzsMYOn!
z6`8vERRh5oBGWA>KZ`dq#=RDjMi98kUpK|xK3ehz|5D1pFZ`u}dj^|CHSz5sv($|y
zB=%z-+~4pN#IUq!7w1{qvV#=2>a=6PUc-12q1tvXUcUktNX<mgu^OgHTmNi&!!a8h
zdGKj>kJ*)Zur7nQ4wuOGb8o#KmF_s1w5T-@si1+4E;~`UQi01zdpQL<t+n^4b7w!!
zCL*|?KFA5WKfPCJg1yz;#_9A_;iI|d%M1{>a<k2dWAp|wU-qj;8l`Q82IU;`onyk*
zvJRm{W|D6KOaE%}OS6G+r#qf37EEb}n0qItrcar{(B<`CFJB})lD#W%EtZ9P!d+~B
zoiw>3f^2hNf@cO|GvV!GGicQS6ei*_tFkR+n>N%GI#NSWmrI_85dINX&nMc(#0UV-
z_EULuNK!}>&_}-(H~2@DE)vq6$zW@s5V4Vf<sAAb#ocWVJ*c;7wY5Evl-B;q$!$D%
z-qo9nfX*_;X7;S8iZ<CiKFQBS46oYKT`(x{T7cnPPO#(k3NB8;HHds{)h^lPY-F^T
zj!uUfK)mK0cyh?HP=^-RK~ju$vj-J_VS;jJ6|t+taBig?zwr%(*qh$T=mrypOCn&o
znpiUzo&kC!rM6MroBZVc?K;ph@IwtNDPc_n^D)CfWw^jBnhxvO0&9^~)V%FfhS0Id
z@^G{iiZ~m{WRn3HfW%%Ou8=B2hfHWY^ogkFy*<1ATaEr*A_%q=Vp0fiZ1vNhC6uR^
zTfwVKNb5t25QiQ4j8?L)oBWobNJA3k)*O^E%ohEPyI_e<^RJ<^!N`vEDAv`PRF#m>
zC^F=eMh2zN!ZZL)>2<Ki!XGRkVjsn}ebL>;+<7=Qp_&EN3UAe?i)UN{&+>|~ouc34
z6@SLc4^lO2B8f9_kzBv%XSIM;7~*qSaeq~7GZi45?ksN0)v%&M4QUuEG8!Z06$HTA
z_20?KoIYPrUW9T<XG`aFz{bz7p^x}=ooIn5@=t<=plSiCP_+X-4k6X{b@Y=zATE9f
zbdpv~4Vd}<IT(dpb5A@Q_@3y%@13e3q1^H|YF$L<d^F$N_>#=AVJe~L2;;646+~>5
z@QtVUYfkcH;qa#DUaMF3lG`!upTCNNJ$o3Tn<|JOS?&Bs+yyT!+%9pb%R<9}ae`1=
z{kMd~={+Q#;%%^-VBz3GD}Zq9xo@QBqJ`HCBty3d8os|?UVhE1VclSd)mppUgx}z#
zVQH~v?G%Zb_mxD!Mfr&spU?&f<s%A9-{B>MKB9yQ`rVb?DSCBIxrv;E&~qquM|{5g
z$5|Mp_B0l?mmqo2P|g<)D?v9aX#h^7f|rmm`h|?LK7!%#Ky|D<t;~DltUt2z`}v9i
zJ_JI+l9H!sz6)9&pxsyimJ|~ifwWZ0!KEc&D)J2QOa~{C_Dpgp-NjgmZhKBC=76YS
zMf$AaXGLEIIvdvcRFeY-Qx3oob*8vA`PH+tsPx}|@V{m~>f%&UP;hfn!LUy6lwLS<
zQ5auZBU2&(GB|0Fr?$jQsqf%XY=(7TtTgi9_xSIwFG*Xtxc*UVtWI{eH}Y=}PHjE?
zD5r~rZxW!~%zv@PKl7C&B)W=vA-{%qzI6Vd|4sg2M^|-@Z$bxS6AUiqjNL~39Pm9=
z5NWGryGy&hG;=hpR8-+Kh&z4TQlojfn=P}`fA&>+`tyQgof+=iYh*WV0*MwQiM-AI
zqq*wafj|Ox10m822@0pDji4!#pJmIa>#l`x=fDV3oG*ndx@>fER%4bR2!w2p%CS3Y
z7!I}2fARW6MMRWWzkL)OE?C)#?<2subg_%R@$~%Z_x?G4!es=$-v%_y(-MP;zkmJn
z8weV=C-cY1tWhUd3Bv)I7yq;)kuxh&VDhIut7abjMoN_jgT<F{t-MZlafI>VS%reH
z2u6m?fP%2!2blJcLrx}W#Eq>J>G}&(J2<j0B8?${amiUwvH+N3BpFHVn%j9AT*!)J
zhB1AOc07@&x8d`adgl2n>ZU~Y76WiH%cvZiKJr$YAz+99)KSz!Uc3uZ#hN9*zO<HV
zV&4y~un79C5vEX)d@}>h1*pf<zSmpG9^h`$YI?JabbgJ~hV7z2!39k@CvaK~r9##-
zv#H6(+*p`FbK}PiSUhF+d##VTfJDa(8RPVf41Kk$+{UpfEguAenAF6Xp(Z`l{?61A
zXQc-L#^&2r9&Hz3p=9x-sidbv;+nvRHxG<AA#(u=4{XzB{#FeWwiDo!XI<Xky54r+
z_|a2F7184a<Qmd;W+RPGsm+x`W#(GH+A;_Kg<i3`t1^t_o1MDU|2_D&GQMZCGX1WQ
z_^l4EFn)M!zM<h}DErEeU0}cxwx&@>Wa*UICuxDZ7bTx<7rN(td}`a~AjNA^CtN?A
zqocrIqms+xHGeO^--P#06X&##Y3ek{LN?aCOk}Us^iQF|4UPLxeP!CdmqFN!T-X5&
zc-y?y@uG5iwcyz`*3D}yji!7CT9R4gqQBYlKyUp|(ro&)^q8M$Bh9A)_gb^kBpn@3
z(C4^5mah&Bv)o#r{i<5RC`yaWPi))Dqnj=#ilJ$Vga3khKp~rVCP~?b%T3G<7S((E
zO>8Z8u|ty9dbmeW_%-)kGGj+oUZf;fBFMYRlW*+ER0K^z{080f;J*z})+97|@iEBl
znBq2>G3<B727jNJuXXsbLt3vf{wQ04Fs%6kLqj1GpxczP$|@u@4gI*zgvq*K76%{c
zlW}OD;rwj?N-k4(Fj8TS7hXFs&-Hs`EtDUf*BIX-v*EYRJJL|l>^M}xg<Q6zIO%iJ
zM~A7e1hE4K)RNWD6=VbnP(Z?a5}*68V`o#%>#tQiv*g3L)uY~AjBu$8!9ZW8xQT-x
z#KT`TW;HE2g7U3${hjh^DkWYO`=>rh4G0YMLK?~z2?b)_qLKl^9gF^)YJ#Apn462g
zraN92@3lcqT9)aMLN246_RO}ov|PSZ8~hV{yKzl`lP2@B4X3^qJmMgu`6kkBl21dX
z%seR--EHQ5N^W+7V|L8pE=TQhh9oxF!gXxx;zgDpNwAW!K4q1ka=w^TOt*9kC69{~
z-P7|x3C{>bX+_N8Q7BnWF!EM#-3UqpD78^u!H=mB4$ymhD8&qgB_m`P<ePZ0G(t*w
zuZTf24j(VYGeF6UR$+i>DUpuJ2qMY~^539B5D8OXZ(Xk72cu+nv?!|7xnlpaZTqWs
z_Zp}XeqfNIqR2+uNc~kI2Ga8kD22m|Yp8WN0Gil!`FoZsOFMGro+JNJaIHRB+ek7;
z*V;yKH{nEJZ7*A4$03^0)A|sK&RX6+-^b{e)kAM9U-49il!S=OuUFhcyVB$B`>yWZ
zm=yW0_k<B^jy*B%^@XQ;G&=+aen3O*1-9Vc@3F)T=X{|`UKkHfpDXVOH{I-Y=#K?e
zCP?koTOWZ=E#PS%sX_tVqNteC)>x#}Ps{;OnYzlr%wTfM$TROn3(ERfL5N$QIX+e0
zY^XN#Z@%vQMNkJ$jni@M;>2h)j@$iy!P3ydZ)Yu?(cLjL&DwJLGJW?@tUz4=(7CTc
zcY#nswc)6!4~kd**5}6K`%1vW)31cgnN1e%@L()2OiV|jm%!`wwwLcLPv0c*p~W!X
ztD}9)iLJ}1IWX<Q9!9Cesc5rcUi5;@k8y8NE)1b?dWxvPX9#kJYvK4*JI{xcUnVy^
zS;+O0^LV3+`(4)HNBY1e0`>E633S-#%_3C8?=x_wqZxiR|NXglIk%1;JvxQsOTa{H
zUrr)La}<h&Ml|3W>#`3-{(AY1s2P*Jn+#DS^VaHz9Z?vC2$W~Jq#r``AsmtUF677P
z{LQem5TyY^fELF=#2SUHQp^F+OK*kde#z^=Hy8r3hH>ZYE1_w5C`BdwF!ejLEqA+%
zItWsCGDfdwOAUl-l?QJ@TYQ_askQ1@Z&5Ynau9NP2pWSLBLWz=@s#oyCQpn=R&Uj{
z!`qoQgu|!n*pjiFd?zQhp6JU(&79JQ+CnS%SJw5DPxIm=G6n(MNgV5?vb?HKdV@U<
z)a75xn?C-f?nsQiPMjXqmiSlV&`Zv841&+h|C_|6wXDBacAf=!M@U4+6j%keq0EYX
z&3B7Ohx+CfXQq!1t@p4Ts1biZ5sG?zavitxex|V(Qvg3Aut{<i6hZIvS3AbF$n0G-
zG_p2B^=^*^0hacYU61zm5eAs3N_sqZXM^D9^jq~7lfBrLRET@do{=(5x^3^5EZr*-
z1=oDC!wD^#i$kYOuqsf=U%3Q7&L{nUP`^Ds&EB$M_U<3Na=Er%i@BoR%lD0;KJ1ll
zJXTH3nSu|6?5CU?=mzBswu?2HxSrD6id#p^xBxc6rUo4%_!}KF7PO<Ii6mg+uGIJG
z<@%=Z$plEU$Godns;aMCDKA}fRC3~q6*SNKFnG}6?y<b+QxS_r0k>Zf={W&!Y|908
zWEPR>Q)Z)WGuS`m$noQ@=`U{{7^dSs&9cdjQjPa^9hxexZXLfoCYK!io*S91_QHHJ
zb8e!uZ<Z*VE+{ZDNbW0oO`0{Dh+!JB)0Ytrxu*LKnWWt6Q<!+s2Jls;dl)0M<7%fj
z81#vqYsV#+S@vZ4j+&v6@tG<3r}g|Y;KIJY6Aky4Ys9O^^;&jpVCW>r(BkQIN!-h!
zfUB71Hn;VMAvbwi>D?!*{y6^U?7YMwC$t<UPd6HJIwr;fszmO>@x5+TnUyS74N%q8
zOr#*C_fnp>R?F9?)IS*Cf7tY(oHbU~<fl`Mx6E$vTaI|blTf@NW!{J=t&}$ahvYXq
zdt-wdI_TTZJ)By`4cW3sAN%!s_w-gHl^710W=s8XjM+Cr;!?y@vi0EZsAwPmWmCVs
z6hcZpXDnQJwCJ~Mv5P`FLPm({^5c#*PII^#c$6(%?P8ZJIfX+7LACt2)2;s;udI8#
zA2F9uN;U%tzc>>79TWKid`!6?X!ogwC2N9j+UFTWh7IANT?Ok0qH|nfp+`S~U&1P9
z8W8Ud6(U8I38D>Z?R59NFwL^q0<Lh3xPRTt%SwfmFj>QP4gS7R4FJVk?Lm_nQPcnE
znQ+%}8|R1Bh-wcI;p(Kl9n#}mM8T6POG7NMMZt6G?5!b6fuO0kR(~{S7snJl+`Pyp
zK)H0xlo5`h$wa7y&Q|HV_WCxEOk%8Sed@(UzMgwVJpEaDtNuIV;XBeMaarRuj60?$
z{U?m9TMUf?7@?r`)myjqBw_Bo2Tiv^F)D9=77R+d@Eu-O6wKkS#dH*?5gN;x$)aL)
zmbMJ7+B@C4JL`1+onXBE^)};{mw<gX6)?Cz*jr2>;w#vbi&5L0zi`=i@QOV-v9ies
z;aK4|nlY*7A3J{hcvx5~kX^LqAPQUX<qAXx{s`?p-`=*F|6r`(L_%@HS_7m{+c{-i
zrEcsEae)KxX=$8T8H;;4(Rm?sf@}Ku`p@W$@}D@Q`vvY~RIqN&6K`&p?Uf;01LW~R
zqoo?!%Dr*BCLE#p_Sci?IIH=hf#wW6GTk)-3aS155-&+nw!6jlz2?}qW-t8&+p8aq
z7~MO0P8_#IZDxZ^=CvmDTmwAsexs(?nynOwEzop^CUOH%G3lN9h|yp5nyzal_4a9g
z<U?VFr!2JMJktA&vN~Ew4Y&=(y#c5~9{^IF%v-AM_@;*$cpczrL+jV8IjXlRk8i49
zy`^c%F_@Jq``joy8nWfq(o24mt3>nFZ7`;s>~}F29e|LZUtYg2(a>q`_+^em2Zky=
zDBWk6T|7E}Wc&K*T<W3wPa18c1+m%esffMDr_Ah=Ucbya@|3cZU()VbCxv{W+;Clp
z;)ZSLOEy@ag{00uK6!j{5h!+|=fLr6Q(F0>dq3Xh?9VSdQq|^JMAOrm{$h-QxLeK^
z!SEaS4;HnoTEph1okX3wLkm^+)ejD%Fq<nIGFN4qV(ATWS5>@tMY`t7(@0%#<y30e
znkVzKrp;K_o>2)*>@)qAjc-EcXTEb_f>nBHY<t5Q-PZ^sOoj;Lm^))bM@f~?P$gSW
zxy_p@>ozyJPBWB5^$I+fN;h0D#%+)OhUev&FU#)wI)7S+D2Mw_sWF0yIO#J?qgnY@
zYHz)plcUDZw?5Ha<-Ygh<y)#>9mn1~0hkomcf&#J++zn_f?+i`w5nx)dtE3h$88_i
zUaJwz;;aO3ZS#wRw%vF4sN>2A^^-zb;_ok2!s}2mwVJH{Tl{?^v&t!>r|3!kVR?9i
z?P2RNVOwT3oy5aIzT_(*GGKL;l;RZA^6`^43mNwTgUfS!EQsbtf{Q0Pr>+fIemL&x
zTk?^ECoZfIxDsZ5PrSf)R+q1UnrdBUFFjhPd<sAtj>28<wJ=-*0EIyoJ~qgy=nVpV
zO{r;)OP|HKd{TNjT9?BiXwI+xw!6OZGKNyq+g>D`+D1KWxWxQ9dMjQbUZ6;13c&Jp
z_7Q7Rk?4rdYz-eIb>V>@dFr@i_44kCg<BC8tWPx9Mxscn^Wgn7cDk^lyH`Z%V=5!|
zD3JdP<U0j7pR364!`rf~_slvu)HdM{f)}_=2nk7D2bcUoq&69}jF^0HSAU(sC0|K8
zJlTm;$RYVjMVd|6DGNS9jGb7jxcul0@j4Jtq!p<FZ#%cIDy?`eV0jVyJ=T{9$*0NH
zb@L$z>z7%#kNd-VSly}?0I5#PjWr^p7A!f!#>jqL2HD96<^i8@N*rvj*BmCfc9Ra~
z!`##^iiyg&bUS9glrqlKjv|3w=RVB-@s23^4gU?tsiQ?WZry@{ZSPc3GcvIV0nT?&
zCG(j!OK<JuwGF)F9cg343nm<hOKsz(%!*Wa0%RLMs%H}n#I$ieQ?lSN_16)T{#=Hf
zE{~Z>{%gk}5efVLxW2A?|3k-)O%pLOHZb9pjYY`~ZZUy>->)s*a<A`W#c|uXq+&|-
z+=;*t$DihnIXFhxsT@5Da`M?U*Lpdx)t2_m+zI#rc-F*aIZV$J(v$WRYvEl*d5bJm
z$glCh2%Yq&10N@ryS2R#z-8_A?FmOdkcM1<Fd?7WhRu|B9z!A3+LD8rRC?*OH&w9u
z51*WY83$R;hb?E7O;KJ@^s5Y*@%b2<GUrBD?ilN9$z8v3Wr#_DCPVoCL+awI`(!D&
z)c}vhUCl!)zb0tX)T~3X39@*y;$IJSN7Y=!49Ov0ShsSS2D!u??89DQ0>U@WfyNr~
z11fj1449MbFtOEoX|KaBa!0F|obPkUW&O!qJHy9+*d89H|Cg<m<>6t?lRNuFy?s7z
zS2MYK3ZVwuqK-5TAgWE%DIEIG9v7vz&NKfnFHo@hg~yw1KPq`%W;npRr8|!ypE9%0
zka@rM%&Ajzy#88#a2b29L5=HEm|a)eUw3!E8=aj{zlpUF3`J15i_1#^d+eni95J?#
zyL}4YE1badey#VCm3E2mj{4uaw^P;XeoAfp`?Jvj4=M&q5nB(|ojCBBvT2U{s0~io
zW{yv`$Ga&cHfzy#aF}=O{9z9opLz>XTkVXCN}gXXq4fN!%SHdNR!0=B(;SaNxa`)A
zuIGsPG)ulGv0jn6GgXd7#ye_Du6ou_pSFoMRjVw{T@x&WpQni#a9LE|fZ>FdF>2%Q
zSH>RN{z5O`Cntb1A!O8)<UE>jM>@8=pwwViTB)dN8NTw7?Ac5+XqL>Ei3_bJev!Ez
zWor{sw9eq*@uFh|>z1c1vYAymrTHW#9c29d?HEe|lHZI4w#e@hRvDOBXc-tcRLy!E
z`{;TX$%w(jl=ttnU1ca#(EhQ3({|Z#vo`eOGJM#S6d5t`aG+{(ZCZmGhpUj1EK%?6
z`jN*1)tskf2}NM%l52SWNg+wtv=Gah_WVW0X;2d;8*=!>R;e#ac?anPevK;IB;Zio
zO&h0+V)-ieY0POoqH$I`bjj=_@~DXlr;d(jm3{T7!!vX<z?zd{GJO(aP{*!^Oe(+Z
z-%16}*I9GP$wWvRcCJE+4h%1)QC5~?iKf0m)ZW{_{7(K!(`-X`S+fJ4U%YINWMAvo
z@pIl51SiP^V3-^Z5)tfAaL>ITFeeT$ULmHUN>B1Ku_6Li7_*r?=als&`mG}l-;(*o
z_h9abXFyCx^;wm{XDOpJ8!h*o7=J%t<YZ9({g~t>oojvyy9Lr3vndhPLzkwgO#DUv
zzb^5MT*KhzZaS?_*(xJ-+(%!_-89V>Vam|wGQBwHIv+1ydy*>r02mktt8{|mi#Du{
z2&aqHa&u#4TVWH1A`axSMM4a?Y12hQHhAT>?|rpKblI6b&adlEtw`l%QeC9XjOGmZ
zY5kGsNpBW=NN<+;xJLGw!LwB@)eB_~Bt@$Hbyu?IiUB72PIG)$^S`?;_S&8M!yr6!
z(!}i=h27kBy!uv^xPAVTw>vk)E-%a!Akw6ARW)9=O=**-#SNBUwL*`*+iOhNfhN0N
z9jiW+YMgh4i_mCPtFim+?Uk6oLNtF7K0<{rMhv5DTH`l(ZrJZ)N@1kpJ|D$efo3^j
z^7+p}oJgU*ONpegvO<6uccGY0@8L|gbnm;`<(0z%=|mujGcc2PCzOYbFPPg<U=_Ij
zme&e;#0b39YHt%b5w760yqUsx*@Z`f2P3~+(k_`FY}Isr*Q%c9mPTiSZD%k#n(>je
zkAPdS{q&Zy8-8@#{yhL|_&@8mhqW+L@0yC#j5S_3WslDK{xJIW{4pnjny1-K>~c+E
zB?zVK*>jO{Md?v<_HXWEV;P+Pw_#YE;a!TG&)MR}G<hcvv-l}hccJs1y?u=u(yQ;(
zDVLi%&!Irr(ogfU-|=m|o@uH5Zrnf>QNY%V6_>zqm?af&dua}vK}dHX>FsEFnYv-v
zRfB^aV+36YUHzPf?KWUBi@To69N@sWe&1C0pO>#{c6ol)bEDQF{g%}Ao+mU-BV0nS
z55849Ml;iROcF#f0DJvrIK(YY3x?=%YNSdAk8(YWwAI<ll`&ZvKD)qdW0KXg0*kvl
zc%}LTn-oyefU3~uE!lh$)3Q>UESt99xkjPbt?Bk-2p~q_>;yo?@hKb_t})^HK$ksb
zgwW0Tkk%eNV{__mt!X!7r&k+4?K`#jz^7<u03$$7Hl+&2<*JSa0i;rwP7<e_3GvcK
zr#nz9j7nD59r!&!(!^`myUlR{7ae`04dOB~3Pv9-S@5hN^Ui3eX8D42xgzPa-1wf5
zuOSR}79#?{|GVR>tWUNU9C-^N6j@DetA11~rlD^Krw6a*5hFl5@QXr=_UUl0P&Y3=
z;R@3zWaQ-^4s_o{<|zwj1d%5HV+5vBOK(<It<3rQ%l$oVhX<q-tk?5M`NUOSIqR$(
z+M3&#RScR})m^z<RzKSP^$iLAj%$X06d>#*R;7(uHD>ha0Dxe*b>c<wqH9e&VV)%O
z10YU+NpV`BEG=$p$F<u$hkZ33cKwuF`VFO#Q$%CI*Z8$JOEozRf27a%DNK2E;v^U_
zWX8G@x8Kat_t;9lt=vD#1kQD$V$~nohbcC33dU8~Bk6uovY?+;Xtb@3)$fPL8|2&>
z{kdsykq}^$Z3=;?cOSVYth|3wgvf^}%flvp3<Ek)<*hq&os2olrE}YM{W81jOck!{
z6Wa<OY%xjwJ2awuy=8N`^&dj<OVY&#fw#KOF5v8mAp;?6=b<25&gFs9^bsY6S>uMI
zqlVlLnP{1Z%#`&GM<WUJyFl9(Utbv%2A0ig%DC5>{8z}@Dr~=OE5AoB+zIDwy{VA;
z2)k3qF7rqkparej=1N^seT0ctn!lCzxu|^j)X2pizJY;i$#xlovR6DBxj2ZyKkkV6
zV7aH8{vbfnlC_&Vuc~172H)V7j}D!a3VvYDDA+zn2A!%XJHPAbnt%j%r}Rw!bJ5-F
zQ*{OjROf8f^d-tFcT}+1p=M{KYUj?Ki_2dS;O7Unv#9ic1zBNjK5}%tq}}ho+c-WG
zQNUq*ez(wRR-tM3IA_c?Dc5ZcxJaH3#I_$H`%Q7d@11MPES~IOpZLKK{-UibB=BBp
zVQ{nh*sOCi<J{}6)iPJ%#^xd?cx3eJ{P0=OTbGkQocA{*!2~`^rA8YMN8xd>x%)k|
z7?)PZS@Lg5jyDl!*|^s{FHLUza-^zi<qXRH8VX{z=&!{a)t5K5#h`N<8L;Z>8}i2c
zgO=khu5|R<79~$P7xI3+h$%{W-Nn~P+Lf<NmyAs)_cgL@tqb&0IaQzL!SO(N(rd}#
z3o9=)270wB9i6198oWGr<YEVv@?K3@_g6-cbQ2aO?MksD_217PGF+sRWTGJ@+f#?m
zGT)0kEJY1_o#(C*<jbuer5?jdjQd)L^hYXX_WZ$v2N%h2vl`khwD`f-JEQXoVzM7%
zeoo9Q93q}YK6Urfb-ov7-TPVD)pF|&_YmGf!!SXrHYY$B7T(;D*E_pzilM&l_lfhE
z)Jt3xr%k}8==I=2O`ZNOf|K^|0r}#>`-yUY>(e&&LH-3Nf`_%%yd_H_C~x>%t8^90
zR_6+0m$^}!?vI*lg_Y;$tkoPdo8z^Zol)B;`Xq`wAgp@XiTuE7oca<WvZ(zrAM;6T
zPNb%S8~0jj>pQi4t*?v9f(8k<Zh@bWk?WYc5fZB&hRXif?yZ!!!<2DV=_lx?8)Vgg
z&)bBg4{+SOr>dG<{WUYQR<UxHO#9GEa!wmw@6s1WDio<M=Vo!s1$r#_Cg?e!yMLDG
z;ed{7(i!y!yK&(5>tZ|B54Y>wj!`$zdicaQ<+ZpaILp|HO_+}8wONCWsT>i(VM6DB
zq%Hy0qwGSDb)J5kSuf66?+;Z}#zY&1#rzkE3-!NFcrW@?zS_SS8Y6Qt_fG@ZmjKTm
zmuL&gDzF9<faf<Xp3saSP`*OWZzk{VUaE<rj@#6H&u<sj_LZOZTmK##7Jtt9K!;w>
zj*t)<<<bL>tLIJ5y9RJxw@ENZ^b|!c(q`1n8{U2%>@fN%@jPWSP$BP$9bJejuJu{7
zXhO9s)p4({Bq+>4V>#G((t`o^>JNEs+U`e<(3A-w&BFeB?C;xWm97*hy=*`H*u%y?
z&@rNS!aYB&J^zql({xSpFpGm8Q<Db^5+r1P9ab*QGz5&z>?X{1a(m@*P&WZO{$1X~
zWy(W}1{b{53-2+-K66U<hNHrEjiTu2^3qEag*z#ck#uS0%|_6LH9>J&?4v7Pe2+yK
z?{IHMX4d}Qt7GcMkHW>G<AV<RzS?)oua%hE;4-7v=1;p}R<jbKuWa|Qiq_b!2fz9G
zN`2X}Vz|bak`(Z`^4i#TN-CBSvyMa<UjiVC(w4IpMX=QNZ?Wp(*L&{Hx1lz+v`b?L
zS$^eTz25W@of%C_;b?81#ymupDS(kcN{~pDHWLBf_r~mOd+F6?Tv>c)mZ;I>DqdT}
zNB+sbpsXWC!~stg0CBJif;&Z$2e~NnTK2>_3EkYPjnvk7l>Qw_9vnW#3J**<J8LVP
zfuvpWjZMa6A2bq*>L+C>^svmOieuS`#QO_o|2?B$Y!cO;vfdsxuamwBp*a!@GSKCu
z@DR+PZjtV1-!z5SxHfkdmCZj@))tAzU1vFV=>1|b{y^`!@ZC4fp%L!rt=>A*F-x!~
z-3sZD2Um)+a>de4#`4DsQ8^er#D0_&r{zDd)3D^{&e!>?w4`CURx?6fdH|l2QCw%n
zI;_I_WLe=SMYJVwwLqR<9kvo79f|SEJ#u}9MxM<`(ik-JieT#=d<hZCvMar_TY^8q
ztIjR1qK=whJch`(z|KT=u5j>+-Yz36YYYbF=ic*0#HQT-zZ0BHa5!z!;kTkhoT&p?
z5o6fWzT;Fp>qX&eibE{AWa$LQgAZY|M2+%6_sl46c6PVn{Q13Jji9;?Q5Rw@IsmVI
zGBR4?{339VXe~=u5EU_-_=hgE@J2s7oNv6Dnivk3AwODEJHB^BnVGw~ws7O%Jk>-?
z!ule?Z|Az&-uiS?`9yj(Ew>Acy|9N`9AL11g3{A&{j~<zR{?>hx>sEsnerm8^yc_t
z=!LWy6eX;h!r_m5l{ymc%G|)4>P5C_jGe>BCo^~MUj9&f{WqdOoE5Az&-3%_0<c?c
zKT7luw6+)U2n`JlX(?|j8qaeCi*mp3F-wHX`MM=ZUblY_>@-cA&c{-=$e!KT<tg%0
zpZt4N##c=0E(E+^PYM69h>qq6_P9Z@BU`pGT#PsAvCY)1TQ{rFM~|o6RKL<|*-za<
z55zr8eOEO>Ei<F~8g)?Xozu1Yjqf8j<?XT;sd*d5l@FPAK4gEt18TfxW>ftIB>#e*
z99i=x_By1RpeUFHE{x=J=LZRaAO}j$)0uWL+k~>c{#l@z+f-rbU-<_0%)}69dJ1Ur
zPHm)_{C49jaN9T7bk|Yrst5jI@S%-_2ztXN<d^QSh_lQP#|+ITxCt2PN<z#rX5cr?
zxhBvc1Q7s<4oB7hgt#ZuOif}LO%S5k5sr3*GC6aPJ-FW~p_K59VA^7|GDsmD$ROkL
zKi;pp@B-=~*DW1!bT!IXc$sEI{GvI8UKY&>6DB~^5dhKaq|$w~m^N{ts-Q+jDpR^-
zZokWkmB(%3>1X*<bPDyZmF}<IVAK#BVbFl30BPW~%%_08@%?*LzJQiKK3QspLNZ5M
zN_XDdvm`OR@dfu2W9ie9F~$|I4Xt}+<BUAI?z|{fOP)LBgOg&&r<bqJQPClm%C7I5
zK^=Q($XX*zQAO`_DOUoy4MV=R@}>dZV%!Hwp3wfX?6=uJa&AY*t^5FPu|gNRD``S6
zu?rHSHw%Dl?J~48kvI(fQuTKa1?A7<qQ<s})M1{Coo|2np#fEe!q(Hj>F-~~OwH4$
zPj^2dbo)}qH4i_W^j3J#a--U3;>}qnjT-b!n|}JFtRXw);Zk;eC;6DaOedmNC^51$
zSl|QStX4J@RyTV+XTGgqo;A0PA5`{rpQQT}oOUJGUUcN1dEEd0^HY~-_@aiU94yzc
zMv?vih<pFG_Y3CDb7oo6#&m2@u9c%&s!VAFtHHzv^SN}|LQ7jWaC4X4KGIr9YtBY0
z8i}{9D}lhZa0&)BpUxiJ@43Qk+g>2d5moc)nUX>og|zJLr#UX+s|G3jL{)K;-Q>51
zartxm@0rD}Nr5e~au0#SLfodfUcoiSQVy)PZrcifFvd>lhj&J=vl&Pv+wCYi^4jS0
znDeD}+Qr`v|1sM?ckY;oPP_~CBMOV6%{l!NnW7M#Qm_u8ZRfN%1-PGH3xYvI;2Qg@
zcdP=nneBCJb)$+hw~ob?-|zDB*PONHg}MpSXZ=QJ0vLS;nnmuXmFqjsGdsi_aV@@(
zeID3rz{Q^@7z%>a_N9GZXzkFxhTJy4FR_Q0d4|Lx-ax=|RJsGlECwL|88Cy&=x2hF
z8-;pbhX;=2))ww{Vr<yn*z_S7O8<pMp1!=-IAP;dp-?e4s$-7KZL(fGmK*76d~oyn
ztT~XjET!<OcDSMAJpb0)gy^dCY{^^Pp7LJzo?!?NEEGM9^;2?m@~N2Xq6E4)L7=xu
z8;yW>zx+zF2+@H^54gQA#d_APH2OALD22@l@eV&%(1STJ2Cqf_t6M+i%D#y=_xbtd
zl$w?x=4k~JnWJ=gyxlJ_P(&AUu+R!gEvaEeG8;5(c0@^SDwmjf0ItymX@w9Q0t8mU
zGjRX>W2tb1ScFjLDGb92tNt8XuxYc96@W2`sLTL~M#RRS(`5FQ$65oEGkE-1<IBa&
zvJ%Y<^cDyu(?2?({G#m<Sj5HE-~rND#LfBn)s2so2dI#qN2g+9ed0M46R10pn7UMJ
z^%nwsE_y-qpx+Y}<Ww&&QVK-YLbf2((=N(N9kFX5e28eMm3X{*DyF&N+A}KlF`@Jv
z*$^oG_>|4A|9BT(+Nfi_Apv+ZbkasjAA56;@^kPb*M~XgjXk?OXuR=f&PD$7pK=!o
zeP!XbDk1qGn2J&+<wk(aD;|Qpifmmp;8C|0bHKZ=-fa6%?a>L*K?qodVw%u_fD`aR
zyM?}9$Z?4MKv$t~8Gkupq-)R6Z{syWI<=i^%3M|u)De-~rqdgx`dvN%Rqr=i32UHG
z5aMwy0|KX!kqPvCf}ELR&UwhzpvwVA7Y595y`01;<|bJShQMZ%E?8P^V*KB!z{@NH
zBTyJ%TVLk?;ySv=NpJ-0MHR?Yq?4kVz+==XJy01KH6AikSZ`1#Zvks?6AK>(%rMRo
zKZP&`7F|r^t=3&wDE;Hkor0L2(lufe=GRXYoA5q6kbXJ3J&idJPB5vQL^B?jq#Fb`
z(bG%F0XUMNLEpOkHB_lWKupTg-?uuyS#4C~mn0omkosideS0VX6Fak`WuiyT@7CFh
zypGoVj&*J_s-VDf)$?11c?Zi+w}_oCEots>e^jBW&;;n4&7Qyd*}K&at;q`&CI&<V
z=iZ6gt=w$qGEOCKLE4oOyY$Vo$6(}iNBtd*-FZ`#NT1aOzPjuhuml{N-I7n~Rh*cN
zm5Uxv>!+sq+4=Sd>@2)cJD8Z#G^N<yOD=Ne;DVq5qdF!m*)!_!AKkKk!SFzSIpV^v
zfbk8R3n}_jjIc2KChtw%NwmpboIqQx4flZGh47JE<SiVE>s+4$-vo<x^C-%otE<O#
z3Fz_jlu>&3y>gcSv2T6BowzxO`?HP<jbzd5%fvcQa{lu3Rn8N+moMfw()lJ#a4avL
z=_$`rY)`iM#;b&j2)=Qe3cO~CXW!wveQbQqJ3sOMx`I3Eb2~o~{!fT|On_vC^+c#a
z#@bn-#J3GPi0+En**MDj#r>bpNsq}o#<o@J3YNL{h(#C4qkkL5jqhAl-rclJ>3b7g
zyX;7N!DgZPqj;tKnX8_QToD3^9NAJOX5)XJkDo<|6Q#Gy+_BHP6XSfOHC{{R=%T%S
z<*h{PW%oaEyvQ&|`}WVm|0NMHyK|LiN?rLcK7DUr*%3G&rLN@kCghA5Z82G*XSsR(
z*@{u!c8nAar~LZwo`nBvOUAwobBQfAp;|8Wgj(NQk`iwUgwDd}6RJfws>yyg2T8n9
zK5+c_n&MGgw)%17XP9@}@(hnfcVMIV0#<ePef@nWI(u&Zr#b|TEbV0_<a>+Uo$LN&
z!UeiW!1EipZcR)-lUC7yEN!gKafBHB1mMbk<v@6+fju$dLznDvI?|Fd*l|k88`4SW
zjs|hv=4Oy2m7OP+<9bL;8+0lrW;9JUn_Upq3IVX$H;HxYh0zkf{$JjW5_e+w`TuAE
z{(HavuW6b2f3hzrXmv5&PDWSd8GcF5ht^nc1}*o)O&3ymM>r*W-3#zWAv)~X#<#uL
zvq#*Ib&+_ZD}V)}YtiXprMr&9w)AERqGt`x=bP%zy?T{K{<U2iUyzJ-tzoxE`iLH!
z?8`uC;On>pm(iL$1_O6}&l!wdEjrk;RT;7ir0D)+CHMHHYI@Y<pNkMW$H?eo((1=N
z>9kkkQ+(b)eqn60L1)G}<Jb~Pphm;LaX6%9g-MST7?Ym784)M?z{oM9XC?-!63VJ#
z-A3kgB&n|2%I^cZbpC!5nIAXUSNA3~qEAV*x?UvI{_{LODts{oW*430NZVyRrcttZ
znx7sxN{8q7#5^$atuQGVpq$_P8p+{k-QH<DM3B7r;HdjWySEDaM~SRH;-Mb*8T^}Y
z1u*s&Pe*OwK<7hQcK%y_2n+Cx!?tWtAjP7Gv(Af$_-DoRY>hX@MfH}D*rI|EUgCh5
z?K67x=*$F1Y6~EC$s0V$jwsQkHoONy5MjS)S)BOqIzMye5quBP>$W2-BW1Cl@HgCH
z<=^>x6i(Nu!fAe9MHWsbEu%*ZwuHX8PiN*?=4M$fDiRAdQaq#OFFAirfOMI#bHV@d
zfydrDe|dxn3Z0mjvpavBA)Wv9P*4!5K)mvd223(B3;3^pC*xrryn2-fpF8b?c;`M$
zP5%VMl(S#{kCbg?dgq($A{H%<Y_J*{BHa-48Y=H<<()sx4)OG6J=Y|j2E^%`u`Opl
z_HpWe7VV!a`G@{0UX{Tf<cy3O9kXd62pg2Qk|=7FE{Gl)F0K)@cSH@GFqd9rXt$=U
zk>jzCjD7~e>p!$hfFhbr>F91-%;Nsb1<jBBKV$0qf3s5k|LGqrQ2E?d;>}qpMa5E7
z9fD<~fX1BHvyf|WNAxJ&^6F?ODG@?>atYD151R|7H!jXl$PhFPS;r){2KPCo@{9ji
z6d~v5Y~MoLfE-k0O#tyQu7B7g-&?DBMFAe{!E6(<Bzh&>|7{&8Go>w>+$jUkwCiZ&
zqTh+eQkuuw9*mEL%8TAOUV9BrV~v7TM@lF%2)G`DVlm)`R9X~xME4j8oEx-~@M<81
zze%Db|D%*a;E+~IwhI<6oB?5j1Qt6AdK&n}@oYi1=1DS-%~u8}94%5<>?ZnN-)xkX
z6jqGom-|Whny>~D&{=HrIWAd!9Kgh!@<HEa<cA3*la|&!xDI#^7@+%c9UQ}Vm&AVs
z^8=5hwsQ=gQiE97heboj&pF5%eH!x#L9WW@D2evK8;`pTe<Wue7Dv)X<I7o&yvsB#
zLM;p1-9pBMX$ox!dLY^iMtmc?ZAKxGPjLcnMHF05U1L>^4iR^UU*#Q4FiW^XT9#qn
zQ*$0>YRIw0tD2>~kN`%e^_@}<E*i2Q_VHE9e-bYT+$G)C<|h@7o7+dEpW^#NgQ(S9
z!-7GpK3j=i2n)-oTRD7zya8Wu;5FrwX{6-Ja<1qvSo0qk2_=zA(vdM1eE)9-%%U>5
z&<Pem-{h0}pFHjot-5r$ab)keQ9eM)kS}gX_mNuqxzPnenl_8Xo_~EL`vkHt{b>yM
zOXTj~f+Z!c{y>rYh#VpE66w-l5)fJvuT8P#t#Gx`Fd{$suN6dFAfqSb#uT2d)=xX8
zRkTUKaFQ0Kpsc+3i4rQ5?P};8v9+Kqu$okFEufOT7I_!uZ`se~NTb?GK{GQlFjHE^
z6pT}c7vp(Ys4r{&jEtY0bZqrmG-4_ySKkUg?{Xg3-kqfnY_>`?3CiLYqRO2&k*J<$
z_Z4b1*zjD`9R#;mg0ezAZxQw7&v3<)$O3tM53XoJENrUyY0!xY9s_<lmgAV1a_Dzl
zfkd%{<^$1o6%!NF4i=TWmMX6A9oF}cz)zuXya+Ft`s4MNtLv*xL|+mcRL}J5axdNw
zx?M3to;}f5K`ViE5VaVDTaVgBZBSjOv+@6&(G^5hgSKyzFnT7$|B>4>$sJWjj5zuG
z65&|_cg^HkeHVkgEBJy?7pZdg%Asa9N2`^TmZJ0wczw}{8o8V|t3}~c>egl|hE&d*
zMhG!dz{@6~_-?&}{-LVuy)Sb!=FKahER)`Xd!aBr^zq)xtcM4Q!v6!gk$UxeS9Z`>
zr(P0Uu7^(F_Sd5yN^eb?`s}5D^-6+Y*Ct(NE8S6J6AABQA7QlOiF#L2)8SNP-VsJ!
zv(Z)7To;y|ai_lGRw9pYyJ*kTDs8&Xx)X4!`8gi-<Nt4Cd({$JyzO2=(jXTfQX;Bi
z%T7o29Ru~YjrI4A7x}q*ko;ZvVIk7;38i5$C|+k5(ygn6JNA}TJZl}#8SL&hm&39^
z+$^n8yhkQaEJ!y*W=pS&mEH+ub#;51Vnq{HL!aqp91w&IYanil!-W|d_~JZ=DZowa
z7ubxxc7cQ4`A0v{jQc$)f|<ED6jI~s#3MhL)8SKKu|v%$25=~RLdyk{hO9xRNlsuY
zalUK)h}k_hQ(26cteqpg*NR|Ll83py`tuBRCQao9{e-T`7|YPi6H2Pj>j_LX#fcM>
zwJnXpM*t5%SY@AhcwhVHO<Q0k0;k>9R#p8hnypMD9`~hTaAUX|thZ!^KR<PlGV#4B
zgP1Zb4!^30_#H&%JwaD0yxerP!_YAc15Lf>13zE9mX}0j;_76vtG~s1^(_@G!vu7{
zf+f@_M3t{3yh=t0>v57w^6{EQKW`z0kRw&L(FI7h$}{|wY@v=;{t+exZ2CnXWR!Xq
zgqq%n=&;JAaqAS&vET$akwd3`Mt5+I=!O9C_@qDUMRlO`ud(-pZrS^^IoFmOXR$*R
zA#o)jAlGsE{Pf9%OU!$!c;#jL;lp|+EwxVH_d(Qi5v*9MHtQl<vW7I5nyT-z>Z{t>
zH{yY~%xd$%gfr*k`mtl}3Ie00o8J2Be;gh99RKP`R23adB%M$-v_e1N=qq?Om9eYl
zBxkJW38?RfTv$%M;0KhpJC%nH)`;dD3_b-Zqc^_=n=aXsCj3zJ!;UDbtHv6gOcMqy
zCGAvpkYSsx3_r>F(A&QH#5DVIQGfK-u>uIkU}Wj{o;z+!F8KJ4n&?N4^WYnG-{uR#
znrqX+(yYbfr?!!uq63qZ_+eZ&wYDq#Xg2%xzUN=$E=zCp`4~sRd_-&IX(?GZ(Hlz5
z&1&3tU8EPAF?g1`*GBD-VPwkvJu+3xqiW?ucly~gIDSVKa9zET)!TlE8)0DY57JVy
z3n7yIuBJSo^fSc1#fd{Oa`D4}jqH2c(0bXerPne2Lcin3IK@q(E@Wzj+YjZMtCQ!3
zMd>g-4Mdx{!hVo2UZea%`=CO~dw%AG59cpHtR61XaU4auVDv*E_g^usx}%oXt-KG<
zI^pbcM!o1-{t!M<TeB$<icag=Mta)1PpDL|rIIZAf9xP9P;Hid(mQ++-w<Iqy*u8z
z_d|-b+X~kQ@A-1nJJb(I#E7Y8>oEJdO)>1jwN8cWF*-H3{gHb~Tpb&qa(VauQNuxz
zRukc2r1F7FEN0EulxStH&UEJ%98*9NehzOS)s${xm@y%fx$n29Z+uEu{k@^lxv2qz
z_r%lINPd^wYE2%v@W=z5+$Pv-A<J>RWMamvgF@by(B1^L;lql2yna{Svb5byG@lI8
zBQi)xD~a!3&T9J>D1ZT>XIj6n<z)DL(#+e~@70Kn?~iO*`uWfh`=hSXT(ZJv$oU5W
zDY$uBxvwhJ@q#5w0^8=06u@?`N#DfO*5}-jD<;XgLXY!J@fKq=m&D75V!_v8LU>1w
zyLZ6#v6iY~rh0yvZ=FLQL=M|Iy-zS+uGhyJKX%&dGw?}|K7$Xad*8-+2>kf)<~L^7
zO37OHucxq7Ngsxv^V649Ys>&^tk!+Q`!2#!L|CpAU*}?qI>*np!;)NU$69eg6Uu!P
zMD`~7_jP_OnLmHuwAf@q0C|{?!|#vI(Q9!q|4@CU(9)zHpyX)IC);;uPSAeJvBt$o
z+NwjeCsTGw3*N&P4W$T9Y_jV3q1WE_d3)-3AYZy6B)HmD%+UVHVw?0SVw1Rxw8gL-
zLuec1Ksrn6gzc;45A<BWY@G*0qT({3ggNN#R1mAKg(nLmYMLkS6~z+k(8hD4@3Hf(
ztBj&r#d6af;yu!Kn7ZX0a_{Kei)vda&?M4-)(LD;^DjHN62LUmo+R{+moV?fOl+iW
z675rS1LWnb4uu~$@Ego+?N3e=H>AF&GJRlYQWF0iSyJl=_F!+vNiQEfGP9g_`T@@w
zN?yk=dGPLq_&2~Kd`ns{Q6~VpNT-aImei&ntfQ^8<AnF9jrD?GLtErY-D;$*2eGYq
ztuz?gSwz1Mfk#DC3hfVlhcr;{5R(}e^ylIgzGI-CcBY<Edzf7GB69sqaQ#&DAGM9b
z4?;{8N}uh!UH@SB-h=oiqfM>2A8_B5OZf^Q-qk+s;5mX4UHdhlxTY9@t0>b5-aGul
z!%F~qVdiU&&k{NVrOyYjs6)CcZ6XqB>FD%w21zUkqa&h$KDO!ToRw@pG1ZSpKs;DW
zbFK&6Fnf817+-yUC1qbR5L`SM3^wXMSEDLX{~%b2`!3}@>^N3^8KQhmp^9qOWEx8e
z*R*i0O)s+Ho)e9eE-9H^>ySpIqTMPSLV+gq(kN+3p+<8xMv-!Yl9Z_7r~JvA=3R`d
zNtJ_?5S22(2tUUyoUf2)4{RTE;G!p9mpv}!6U_f1+6$2dh(W5NUqgsYJ|8%CtUv&y
z@jI%NnL8=$F4im;sU7B6fu|#ctblTvi*lqy8`q1Nf1wo9Du%#M%t~(-?Y+lNGLGT|
zM)Z`t{7#4Ck#rTX5myuPo}?uEdAnQ*?Ul8uimd3Ci&sg==Lm;X3NKtJw|<*0+6}Fe
zOE(;#@b&ZS-Pe?n3VsxgeD?9~7B=%IqKu?nNU*eoiE1c^;a6#gi~sX!B@3kLpC=6K
zfToq73P8fbryRpJ<$^=?qqBV6!OfzW=^Jw_%?V;bR5So9nwP)gPxMbE<8R8iqCr>K
zmZItH?@b2>m7o#jF^GU*w}c5i0^7t7Uy%j~H<LKEszZNrUvR%01`QLy9lg!S3ukIE
z&%Q!P5=}J9(&Ij_5r=&lP{&75!TcLSO$%+L-%+h(I6o8Zm$(_x7QP+vrQQ!pir88!
z2z55L9ag=b2CJAjU~nvcAb_9mjg~Z7$vqZqUF8{bO4rnXDoyp`3xSguG=2l;_}!A{
zJNo`@>fd!{o=Cj8c}p>KCXoC^=PXD-csG(hi-~6xvRS190t$g!WLs=TQBPs-s5d5~
zCIa~KXx}k+PGB!lgImMrP%nhHV)X47_CK5(`3r8FjkF<4ejD;tIFLAFHv?}qEagvN
z3;T9_%OQtRY$}mclQ?6Cj@t&xH({qiwA%zRAku11CVy*IJel}p6jmPXuI@t3n^N4k
z2Y>NH?fAtO&lVgf*4rE3jT=3A@?_yypTq8&#;Ynn-(&dr&M7GWP4E1Fn5v3QPagJy
z41U?^Ikgg7CKa+-QB9>4A?IC1jEjoxA)$Q#$<wE7OXK2|IEfv|8tg%zzTlz@mPzcG
zB^-gNc<+tJX0nMpM6KR|o%4x$K5@~(-nfgzsGV}}*iQfZj%j(D#zCv-*o^ci!#vL(
z#|q854sqweBkkBI)Ls`mtl{ZWf%xUBlD+{4XV+;s?1(;ix0^)JB2kJPa2SK8Vmcp8
zOvRK2dTgT;KAimh^qDjC#bt4g9oO}0hb6+&mK55EKOhwRTf;}B8ZK!iNC?r$#Q}za
zC<R-RF^C~M_&-gCcj0vsGqh+U^1XPBc?oaHuCR2=X{v$&{EQ%p{O?S*q;eS`<g2(~
z`CnnM#|Mc7j8A#aE+PFTJP;L@6SDkn&7HPJPE-Etf_78KUa_L*pZh+HZv*uwOl{~}
z!Ou?M6WY`)`NDsPd=lSZ0PHjSLGw+g-p;ip)B%LHf4Z+GAl&4R1V23IR_6&sTOhH=
zRrhBs?Pu-TX?SA#<m$5!+y<Jzb&%H|ak-3`a?poFpVwr+5zR>$%F1rn%nC+qIibkw
zp6ER-52sb$r>Ru*aChQ(UzX5Y@!cNPuF#SjXuXK)xGcufP-9p*_%$G(P(4^tQStUO
zVC<sflg4UmYa5@)51p#U2oFndA%AvwLp1$p^Fs}Fsh8~W?Rs9boayfLO*jN9GTZuC
z|MA*$)YpYl8MvlzHne+o_Q}B+Iu1SMwB+}QW!rb@H{vwXoA<DYfV8OK;4-2=!Cx(E
zD<>-xkwpyKCaz%>Ie;!cAn+NF{Zuq}c`L8|^*HoPTu^%41g_AcC9jCgtBdu!v?}@~
z1#P6bKDRQy8OAlVT-Dpsk*iRcRc^|GBy^^o=I!~zP*fW}?zY@c(hBNWi>CnE8lRLZ
z#u#v67Dq++tf(KK+aTQGc=2|(zoA6)O{wDIac!(gWK$N*K+`Y0p^)Yr@=n7X^w6bu
zT&MSE>t^}0qx(Q?0h~w+T>{*YFI^>nLs_MA{8f3tzU?JeZ+^*6?JB34E~gbVZ&b&n
z#NEaTd@7l?FeQSy@Lk_QKOuvm@wt1fH)Cr}fv`z4WBixOZPYYB=*irKXD2*Lbj^m^
zf4m6;x&q2uXnnqRvzYE&;?#r9x;%@R3()(rYW+Jh7a*1*Ce;2v;LdO)*oFXQ!n5;N
z$2H4raOM4pVh0_TvpxuOs16#f>YZv}lSRUIae~l=sO#yh0wPC3FA0_6h8#gkK*A%=
z7Z#wrB5`zZU=1W?xWGP^n^9AmnhFN;YIz#rdv4vdaWc=cN94UZxqEE0-NR9-X<7_i
zzV+^?Db1;-M#G4lS2+;CsNS#czUk7*SOskb50y>*3Bi()LK2F>M<{~K2Iw!95Ehl_
zOMf%=K6U1d@YH~!*X@1ol+6OqJ#xafh8G@H+u{ZK7gG*e8wRv1I>EO$zpt&{#!dj0
zjH<v<s$3L!NHIim4&5b;64>M%TU8pph4dZZ_Zi@m8P1c4?zyFT^Sz_jbP13z@1wBb
z*rARWW^9*G^tpo0>t0X>hLvf%pdb$I6Q9Q?Plal%UvSgzU8gg8xSMT>s~h;628UGx
zwjEUCgN3Z-UQrQw2AUQqee_P-;Va7GOOSAhu}dqDy%&jbQ=B-&2=7E2bR3!_0R1PF
zXK1j52cjCo>dd|YZ%f5uG^<>Nyb`MUSF{TURcvvpf}__+O=Ogk$j#Yp*CF_&whQYk
zPMi=T<Q_G(O?(G|KHhg$nX>T}CGTmwynl&)n>snTRq`6<W_7GRfmu<2Xu`xiNMgQJ
zOlg4*GAQ;UJJC$(i%Whzrdt*m#<6uo(>Zkqv5;()fP}B)t_-Xi@a%}V1{gA@{9fQp
zn+d=86dwPN7Jw`Gfur3?qiP|CSgS9+Ly*J(_`)GV+{=6oBEG08i$)M)7llb&0D?CQ
zZz|Q*S7z>!6QVMT)W4Zpk&gqqe-K$LAU~jc)ku*^p8f}WZywe2{{H>4HyfEL87h>>
ztk_f{NoCklri2U)GGwR}8Oqm`1|>roN|c$6GG)w^29XG*gi3`B70v2AuI&B0@B5Fl
z)>-SEbFZ`Rb6el_`+k2-pZD;3U9ac$yq?$dnssrZHhG=oabQho*`wp%*R<M80!;Nl
z|I?#OoFWIj)kW?@^5FQ%1Cf+R;m1O09y!zZrjXn`U_G@T*#2B))piQx8|xugl%}s=
z(V};Hka6f8V;{?sH~S}eZh>bbvNN|g9(}qPu1IbG&w_+^F#(Q`nguK-#TI^&ztc8w
zyN<&x{=VqOc#aIusQw2#vutS(i-p$b3muA=ot7ScI!7$L>ld8LChaN!?sE3>+@Bw<
zMEVI0$LknL>Sxfe2e4i=@BhdQ<$&TxgyVh#eQ@<!Uz^0;_@uxz!hZivLzMs?Mm5mv
zDRX#9jozW|ZW;Th`^w67ccoJlDep-H)>5j&{fg=W2^Ey*2zDwdR({!2g}iCaw~QuR
zlTrk%qB||eyCJ~I6k@BMpOb}$?kfNCk#*v?3m+K$r%YKkTrUk-NjpX?Cpc8o+Dd=R
zEiOokI=n&X9dW>roQXY2%_Z5fjz}dl9%Li%IV3{Q$<!z|!`NXa>lfwCi7Q6!RDnS9
z^H~Hq*h?LmQ;)ffZ%HXC=u0TIp1(}2CEjgkuZ{cbujE&g>5UZwkaa2bZ47;?uk4&V
zpYdhXjKv<`X*+knmqUqL-xRR;>5)W;6BN>+c6DU_MewpZbsp@Hf=aKsfTatUZm=Qr
zvRnR`Q35mSP-{ljGJ`19<)I}_r3CHdnon`m6?Wvd{Nx*Q$GD%rio&${{wg^N^S@b?
zn6+~LqxDs*lD__aeoNSsycQ}fZ7&+5jg|Sy&Qr*FR@-s%(o{7JgK?08iNQxUpUgYb
zScbQT&F;x9)nLJzHLkB@SW;vPaG6(2($%MQHbADutIzfQ86tHjGw0lBeaeIK3BCu~
z?mtK=Z8+rHR`YDuF%K(ZrJJadiZXb2g&}%m(8<f7PxiP?q0VWy`bSPS<aFWjrsbTd
znpG{2otpl46&)>NJ|6Gb+i=oqEVM?hUqjf>&CQKed<Hp>I`)qr78}mZlu0-jA#*Ip
zOIk^p&CcUg^Dz~hqd_rdhW1jd`<U_sxJ3AMYGUfIN)<ISeA&|@bt&2Fn~a?n*7Emo
zhmf8o*S2kIbG|{(6Jsqb_A53XW2E2Nsk`Y$qg@?~nyG*6J3d?4u%+c0qlj%C`)%9s
z+xRI{N40HgqTcQ1rlRQEo?cEmXNS8wI#xG`eKS{5_b(R)H!n>dTySu3?}OF%iQtBa
zwL0dm#Ko}W3`w~|t*IEyrOjsz)F4&9AMi~Gy{f70Dc~I?IpzF~Q#;k*XI?KupBBO(
z1&xRdo+l7E!Rp%>zd;%r8XsG}6)5wc7BcsJfz9$GNPDFML5EMC(i_e*6|+pbj`O@$
z`E}^uXNN_u<l6}qG_OI7Tcn4aWe--k?3=bhj8O)z`$B)H#&E@#!_Vb)@2BOlLvM4y
zmgMe{3wL?nPf^k7(K>$lW@r61-xsD`e{pP5_`!mPQpMDJ%Vl@UJu_ykT6FfVkVoKO
zKb>~wZGz?|zkzy<Q&qg}rDAv?6=qNLsNW(^2eFfs-Ch_U0b>#Shc29+#_hsqrLy<G
z06{$bFID)1hgnQf1&^ugl!8xE^cxO7wQl4)V*8i0lg|)O>)1X%jy`A@o_|;2<<jqL
zRw$gdmbw=YyX)E?DUg#;&bB0ELEZ;ZZXc;js<Tbnryl!Ann=|iB$kbnUYfNm_);}#
zB-nxho@nIfx<2|7SE|QRrF;qzNL-oH&#gW}mvoCxIz1M?OW@;(Eq%0N4T)ef!iXq_
zqt%03N$3dR6l*zpFrC%2diS2Tn@sgCk<(8`3`II3(?%P2o!-Hx>wFe89LD8fqENoq
zQwJ?YeXPIf<RhWH*Pm$Qv&PIllx;x$E!Szg_N}ucA9D!(>VNG!?Nk~&V0PHSAJseY
zC{vy{ukqND9d;$%T(_;cczc)k#w3-|TXq=q@XT+s(;#nBj#<a!Ao+ag$BfRbehn+P
zKm-Mo)0CbP0=nd$X>Z`sV8a;Kn%${DOPN+lNdJf_!~t?E!Y2<#N;a>tp(hR5w_e;2
z#Q{t=jhnyBs%xh?o}$S@$_SzeN1p|?q%Vx91r_R!bij459l2#5iIHHV{9IvI*xDq6
zgyRX5CZTtDLvA4irr+Eeb$1=*=@ye@Vub;UF6n=Q?jx>%DT(SINayCDqEBu`J+$Ke
zK(Bpv29;jVWx#{f>q?)w;Nk@GsG!e$DG-=dEdUCLclXj&5Rnc16<<ivK2>voGx2h;
zN`a}#bL!+yOZ4A})y9O)SKR>B5Kb7&45*zV%>@Gb>4Pf{*P&)4$%1=0d&bP>{GS>c
zxiU$^ag9wna&&f-&5-hiW??xpsz$d9dyPji^S1WS898P#9Nzqv<B}q-iE0@kNycA<
z<0XY4WB{iYPiP=5uQD5;c9Vh(4m|TI9BJ1lsMi$7+d*}!4WY2t1$dmPKtX*s5P)Jy
z56J8=oY5bC3u`DuGDxJkPRwuW9tqA``|d^U(c7$`?=iL}xqA<fge&(02dP0tUuCPQ
zXA;uq067d=Unh-9cp!~sW!Al1N*Xk1*6YDvKkSxRh9Fbs>^o5&<M7(&C|IMog;-dU
z5ci7C<n(D%duRs#KOW0t%6jjpj@Hmfsr!7JX+-n#Ei}jPuE1Dr(>3s?q*$aw0R~ey
zOsXJRmkt^>QR6K#m|1lG2IgrSntz)%iw5(G{5N94&Iz);^K-7L%dHC~a{KN>K6#%N
zykZnPvnFCd!y59P`y=~)bc~g3!g^Tw!|$A53r1*Ll{=XIT-o-xw6(h2K5=D1{7C>2
zy%fnq-@-<e4i)Dpm(o7(Q|YHr^7tUevC)oCK8BJ7>)z2wu9~|rT`C6IK{}`4%@<;z
z8)bITz{BOm6xu3OYkz-O8Zn?#r%qz^6dfiKAoW{oZ^mG}r5~re{AH4UL+P+ealaUo
z?V;IvzER?)<nHeIJ1n0h8fWWx`>T&rYiH=1u)`p<LE8sn_s_eR(RD%_v^WFq1DY>A
z*d%bx1-d?xNZ{<kXA~tO$k(#%;MTBRM)BY-GE9v2f<-VId51|oM&9`Lp0%%((qq5O
zd(3=LxQ~gLUCbRo@QV9h^oahpV4_GY09Y1RU?2k3MqGpSh%$Rv(B2cM^%phu2OYe~
z1oq-nS#cplvXu1V%_Zfhq-3&hSe=aF6j1kSWueR?Tka=zT1cq!I!^PXYSklJJAPbW
zBM}xkm8al=y$(p=(Qz^B9RbZ4`?cJhSpSLx3eE;#AZd6@+cQq6^4UwjNzDP4o_Q(a
zStpQBeK?Zk1+xm~bIsC*X`h1N?kB>!baoB<gXc1gxSU`eMIzSwTmKmvCO<b1B^h=~
z5qpYSQd3ja!{>Y5()8SHxjaI0+9K<jV%kWr9&l)*K21!$Kebb5nM&m;@~q?jSwWgb
z97+Vdp>-O)Zc&}x+SSZPZ<rYxndXK8*A|Xa-(hgyT+O~!{OtTTePZ&u7Z3K{k$1RT
z56{gRS%i*v)AsOmSAj2OBL)r5p(Un<s4o_A9@od4nm&-do=`>=28Gm!va9u*%_qO2
zcO+H4DCAd0X7{e|qFb*2+*<}^7SKtZFra>tDh)+7i|lb|Aq-w^eR=g}!_wUh<f)5P
zdfQ%2<WDkwD?e97`QD?H)66G3CPu^i0x4c8uFX76x;DJ@j{@)yB>&*!bJgbx_u<qZ
z;>hylG=DQrC3z8N`V+nc)s_xQ<EYT6D<Lt`C*kDdE3q5z&)z2*e8Gf?4x%yH+0yD$
zp=`AtQ8@)h(F=*26L%-3bbPaO#%DJ0!>U{yy__KooYpho>IQS<B(=c%01HP0oY18|
zU;f~i*Ua2%9oK2jK@*$UY65^{uDAe@9#Xv?M17dU)U>vfeV?L~eLh{^@uEuHb0{%O
z1=ej}(0PA<f1&KsqcpO_DC~ijYX0~%y#q!m+0H#b>GX?Qb*jZ=(ABrxGXpJhb{OQ1
z&0mr1(CYBU39=)bj#D%DZ|d5&^mI8nfJxZGuE$CDD@bRPU&tmuF*noRRlQg120Io8
zgzKz<(z8=}v}W+?fJJ63c8=4H0l`|9IiU4N>u!*Tq#qcsxtkN)VdP#;Q9w_r-pqz0
z-)V0yIJT606gk)S)WL@5UI8-hy4s6cB)vQ1h2CQ-pR_LY?-u3Mk9)3*G<%|WV7AY?
zebPebmYS&Bl4ipi#~mM}2TOm#5ObMCWS<K^8sc_s;V~>4&v@0<28HEjsfZXasC<Lj
z&rGygU!!X)fKg^;9yr*kKODvJn{ch^sAC-Cie<-R=#mwOzM#+2WeddyERaSv#6j^h
zx(~4`Eh*_%tl{q*;v9fzX&KEa37yV?KpxLrvxFdYaQU}D#Y;P<rB9jC{yquX-N4f6
z+aIl&{>1lM+=P-H8vBg3)Lok@tt_#4yU+N<hMfW^NAL9BxTB(N{tL#+$QZhsbq!@u
z<hNGs=hMX{8wchnDEGsE+agU~PdFgWRqx{Udd83(`oBKRi+UIDbH1f{sDR7lCZLZ+
z81>WHRl4hV2ew9mE-9*?Jm;AegO?Id$yvhfroIbY2T*<-<?3nQ-@w6l&2iTn@hEGe
zwz^xU=G{-(4?eJE9$x!lJ0q?2O{34wU%pfIz5Q+E^7%nsuNoWH4#s{@rYA+aouE|u
zqVvS*)^$U5@N4Svaq9curbdkMk)F86s_P^B>~Ju(9b3~>qi*P3PZ%!tjwnq>P}WAH
zh9Iazh9lucoc&)#L$=6Pt+6BD7k28~<KRl=A8(dtcbKl9Z{L43#9Fu38usfW#FMOn
zw$hM9;{f|shwTROaL1JFsMu#bw!~soJHxvMA{azh^Nd3`LGys?cd?oV&-poZMRrK|
z&*}7TjJaBaZmBAU4&T#1B{@7YRR8)Sxo7;WvgiTYo!lP&xSvKS42Lh<7bqegrouuV
zcIIoIRMeDFMWpO3wpL1OUrVobU`Oo;LWg6R+jK(7ggt%s>^6Dvyn&t9iN`6Y%OEC(
zm)-&6PvQ~pt{Y!_u3g=pPS2YUys;xA_Plb3Z8v^irTeZ@rL?{&)RaQLo$ICBwrQ0E
zsh+f|+O@5!(v@QX+>%&46LQ4a-VV91dKWOHqUQ}jRYEJW<8cZrz<8!*12>xVB1@>H
z+#L+6T7KM$Ek7^UXZPE-fBfBYGIv3)XoGl?<4#yS{<0{ePE-dh2%8<ee?QHOb!X}d
z@x}@ZA0DIOaQ5Sli2=*U#d}k1+}gF^^U#fF9F^{L^->$F=F-Epsp<h%_nSoH9LHAO
z6OHeTt7%%gfM!0Uxiw?lDDc9TX@1;UD2(s&eLEuC>-`I~BNueEbg;L-|7AyR^xxK2
z3m=u{+%Orar&me!;qlA~<gPC%yW%?J0&iyp$fCAMKhNaLt(A;Q>NeA=aE!Vp1?F2E
zszP3Jpc5|#KWI{t&}_Q(d)$|*_LYt8U{=?6-4(aX+NDE}>v#rDatS?W-U1LW48p6O
z2M?-woL<xdI2`EYZ>b{FZS8ZR+b^8mqjH4c9<*I=dLsq<>N_W57yL&YXsMDu`_`_J
z<t0?egp)w0l$4oQiYMeX-BkNr&b*9~n=zsim5E=`RRaB)KyhiQE#_s^0@bZY1=X6R
zSvytki(b_tv1;y^>=&OB^#9h4Z<*n{C9iw@VOh<CU7Nli{tsf+vLIDThYr-KJ4uXM
zNu=dW63aBtV!yd#ZYs;PS?9A~cJkiX(RlwXSi)2$(a)*BRP`M()jqe`PJ;%cBJ@0y
z3L9&=JUry&((it+Gh`bnp?cDgdrf?G*FO<joTNgMFfh8AO>FeM{-|*X#%o|=a}D3K
z{<Q<d)9UkIHLD-=YZJ7@xkEzyS8<+EUi9cYH|Pm{Q`z<EZIhdUe?FM3)>zBE_-KPa
zozlO_AP_3+VvNF~ALQEP0-<9l{+eXCY|w*bA!&xGz~XROqFSBZ-fcUvoy@3tUHs0^
z;ia6uo!qMCUv`_PV?$7Q*Xo^36AjVe25oN~UErD{5xBxx?Zsg6f`LQbSE<Xv{%N6)
zafwB`fKRn8yUFFl@(y<txbkkV@1O%HUZE5Pb`gwduFN5xMGWXB{?_@rV%W!!1zi+s
zPM`TxBtNs$2DY}<4cgVD!Pp`H>C!hWq2AwW{-d$hU)w%Tsajv0wSQ(u+`(koQXe20
zmf;4J+(n2};aLe(oXxdU;SgJ0h29&ezC)*VGnjvzN)C+KXQZ~JjINA?WrvGyHmNg5
z=d@72K6wSxnx7kL!Mpjk^p(%si_gp-zK8}Xar)%Do;z=$^_@%zw`KhrJMa|*M~=a;
zk1A0eU1K61C!-nebK`(}bNge9+bZ9HDd=RkCp%&Na~MzT5F$sn`e@c}(A`6B!I7R{
zw!hk|w`fRR-Ju+5D@b^d;FTYLlnt?^s}#2Ibg1=7L^97kz3X&Ij3=&*`D>X{Hf+gZ
zi(O6j8HXwiWZ`Q3yKq%lxI1{=-*6b+HhZT7c1BT?&`0Fbk5fxB?XbiUXLZb7(;=j@
zOkI8<i)7P4^^ZNrTK~;hiW=RL2lu`;xvJvWC(d=r;&(u#zc|fk-z!7p0)|+R%#VaP
z(O`E_kTwU3K~Y5e<In-#%9COkYz$`Cfy3mbh};-Gh4u0q+v_fu(=RNXa{eY>%9yZX
z%h~Fy?#o<-G<*FkJoG*N0}-XK30?BT^rp=k^bkhj-jU|hhC_-rU%q%oT2mRW8q`Iy
zABEmGvY_Rv;4mO$(N)HK>^3<@HpkDRF;vR9Ri6aY2hOCQm2-QU*{F(|+oZ!NZN^t#
zpjf)Aay4?vAlL7<@L;nmoS`4ohxfSqfA&yac&IW1v<q;#ZSK$z63>eZQ`Mul&ViSd
zA-Jvo{QkrzOg6lwX`n6dDr$IgxS?T93^6|=U~7RvevDrl^n+_xo5N$x&6fcd0AdtV
zz9xfsF6UJk<jFigovynfK!8|*#fP<Ok81(bn1a5)@-0ySgcMlnYx367u-e)_mz}I&
z|E)gj%DR|7EY_O?P-chtJB1^=ORJS4n7tj<=bWq4vqR~}$Z#JG)y=lgMpzwCfIU#&
zauh*QvC=*KUg+(u49v?I340RJy635!?dM_6gSNxq`Cia2`Fg~{q&AoAU>&KIpUzg>
z-9IMTvXqM_oty*UCb7}!8lq)s3)l`?-|k(QRY{vZ$$3WkZH|wAq%zuL;y*ywcRG}i
z!WJyp)iP#4TpwCHJU=B|x@rhgLuai8<%<_=LEK{VZ+Q(HwDZ+jqX?6;t+;q3e20wb
z&d+7srwob6(BHf~I+NcdBir;~+@wifsJ3m*cDlJwOU=5!o=Cm5tK;jGzvpdnG(qRO
zC~8>68dO9%DuT(_)prNa6?RVd+>AydFZ9Uu%4-pFqY~S5`LX=ZS>Lkmoe=|z%3*`a
z08Vt>S~|3AFHfo|k6ySyE6~?)f5ye{NKgecA}T6BJ@@Bg@N1ANd8o?O$z>#^2MW63
zU+qG#z0903Qk7wEL@K(w^k<2K`#q)5_6Bz)%ZldvbCZYK4C6lK{TErMdwFMUwkmIR
z9o@ID)3|-Mcpe^FakPTWruq$PLfuI>s7~H(A1b5<eY#`H#%*3TC(-!W=<#HV&VAC7
zyYCMCCsllRf7=R@IGH_=Ht^*I87+JenOZ{Gq4n0HNds@2W%pm1?IE=Mz{h)zMd;s=
zn5&h&A+iGL!OyzdH+9YRt8o9{(yOub|GBumJTogCp91ZLp^WXw)4h3AJjvqq%qR72
zWBoPK|C-s9|1@1bAeLft$5mcZW-CtlhJ~AGtcdfDBM<Ux^inAnnD~v#%Q{Wy?vVCY
zD4eswy;@xtvE`&J3<X($<S<tVjiizMF&VJ~kLnkCylhu^O%DUNs2y3{!8^kX=gD}7
zElw+>1U-wo6f0Wl+c0nPfWY54v&>_=N*3|knLqQiG&Jn61i(Gg2%3+@esf`J1myld
zwm(7)6V=5AkzB=o_-V9YLijfwjCP}ki%%fimZE|vZGka!=^g=kzW}F7IGA;Xz3$lm
z51FTWuuB<vVw+A6gem#*1j_NyLJj4qT-VYsp6~3eF|1$&hElCFWgn3}1UUXghuJc=
zt;}?_jBV9S^MNqqQ@U%m*72@TTAbe|X~{pBdxh;7!tWhv-6qZ$Rwhnj59?FpkoG|`
zW0Yp*%wIzK4ZN>@IJdTh7vJjZ&mGS>2vjpFX!;>gj&NBoqtYMY%;BPRZnj?ioA0OX
z+c@a@t<BW*hg&sooIm}=^4I5Z4?46ga75XRm`rkD+ZKD>UOy1VwakK1AS4ms;5qj7
zlOPSdV&b!y8cWc6v?^kHojj7(jwCRSM)hB;nosso9HE>^rtZxL{@z((i>$|vvGWY{
zSAD7BD$#%b;YJ013CB%^FN|x#i^GQwemfIBQ)=cXG(?V&ak<etC=|Q<VvdC@SI+P9
zN=l@ibFU1Cl0h_qd90kOXFkJhH)~SL7VYz(X)!{`+74AwY{mGInoBy)7t$iq{g$k`
z;f_k*N5{u%E;@oiifOBKs&okl38TJ6R}l>)9HjQ1^rF9u)rksuOWSWOOKa@M5u#H4
zep~m);!8!@anm0L4K<Pz5@v`ySZvm8ksg<>xT^B7(1-xg`D*qj7~9e|#-5l44f=SY
zm_Zzh6B|%DZZ=o^%j!xE>vn;VFBL#LM32a%FKgzL^Km~O_w`6f=uLgZULi?EN1?k%
zf-&q~kNOm#(@yX9-R`1H>Dl_7!>JijIjGsY=q>OZD&1Ue{tMf8_t;kDHu@O1)4`!f
zXgBrPep06O9j>*ZZcvvIzWfyIkK325O2()EAT=&WPf0>AH9{G1Mgi<z`<7pL^fI8g
z<hul6VE4X232&(;%H#uL2_u~PuR<wSv~s%r<Cx{%gU#Wt+rJB^j<BXsT=|uMe3nrf
zD!S*%?*xfaAiRPw{ado1D5GHPDOZKz^SGcu=)~+(cd&?V>ZOma`#wgL+UTkP66h=N
z5dI8D$OV68J;~8u;i&e$R3yljtD9fv5?;GJ?9U$)S0gcgp^C^93;_UClGYLJ*)_t=
zLSl%+5EN^rr;f`CqP9ckZ_XeQ%a@Cj4?t7&DtldQWEx%m=wB^>3?fKa8=iZ^mR8Er
z6}HP>BR{Zi?jG+ARmgpM_4}e_+q<f%%mKhvEi3zV7LTW`hN<G@v6gsdYGl@RWCK+R
z2#xUX-qxSTh}H-<!Gi}qMMcJkk}*kl9<wgQX_N;olL7f(Jkc_0xr5gV`y*~|1lEmP
zC)S?<Ck6>JtWg*aGW_#jWZ1L5`U5nrVvDSiM!EMUn$Su!Y<H(mO3x@1#-hrjxA<ns
zAf7@O-e0Skx&w>2io`W9-P4*m@Mxl3lM)t9*WUcn=8+}Gb%gmPt=qlUQMb_Bnkp8k
z;Z?RNy&u+{4h{+`rWt3mV{n{CrvLG~d$%793Tk^bw2(@h-S?E|o1<aPXb^i)p=x*I
ziS)zaH{7P~*`Cl)nW_fNHdXz5Q2uU~<!j=5(1q2uS|moxQBSaXOarFhcy?(#qiIQS
z;?=%2@*gCYj4UYs!D%dJq}{`oD<_pYeNSljjWAx)!Prx=+UYx=$uUByKCaT1wjXkl
zxV_$L)BAnBvD5oZL{B2Dllq&N3K|{>OW9jf|J_@;BktaKWGYQVP&}I+^t+K!=(EFM
z?{@RHgS)7XwO<fv<_tR_k^|v~)(p|qd{57d1ShIqcK(ClhYtO{oCYHlJSTK}jULGW
z2XYp_<Bmojo}Ye6IB351R2<bAISC?*+^7alLIRYz@>EA0v$#;`MZ_bAP(?+Zn+DB8
z#(0wmSULOCAoHLDTGDdWShtox0Q}wK-5wZH%~^euFEg4TG0GK;254u8mFInQT{Ke4
zY>3WK-I(T$jtYWdaHp@0IpHoSNzkDLmbQzki?BUyeIXy1x}FUl_a&HCcvb{mZ3U~_
zub&#fYV&;3A)6bkodfDte@Z(!m<2Agf31}R2mD4IETQbtN}qO`%dUMm#?p2s^YcgC
z05ql|u<i9=%J-z9<jw)wtw)?me?!S)OPE1|N{&@J#KyUQi4E$L*&4}vfV+ZI<j1e|
zS#!8Pwe~6B^}y#Cg$p^VtG~t?;*m;0R4m$g{`6&ir85KFL^{cjikXbQlqm%CE>7A%
z<O8BaBGp<B-xR(>?W#2*5l%s8)5$c>k<5r%l4G<AK{+#eR#Pvz@p}JF!t=n|CA>QF
zR7s%FID#Rjz?;$7eHj3Opv=fssd<)60VV=US1%79nn;hn2#UA$o_xd^{+<Wq+_{-r
ziXzGYNyx>?1W$$;l=?6>QS5?&E41G|1?5Vg#KMt(IAYd43kd%5jCK2EGM1Tk$f^tB
zKhKPMFp$xgLi5j_*y_<o`o1}=m+;YpIr*fN$uU4F%7Cljo+a){wSd|sh-biwg{6l8
zw%nFj4v9x3<hx)9WbM3i(a>?}q)o1%R573;re<_decbbtjn4eo%16sf+Rx%bsFBLR
z9~zSS&p7{<D}Duj!<dNFMA|`^a?G+72F!W-G4tf6_o%fqF5segEb$N#s%D*^Z=656
z1BI!x(@wtt5N821@ITb4Z8xi-*nTtcDEt!dcmSdnt*Wu@1RaV5N8&z*EPq1gJ@bg<
z2P==eAs(8I=_B_z5^t`Yt`~l-vx(#-cOR>auKbsP4|vmbIhtT5>*bxUlg3uwtMukW
zlE5nP_+j+KSx%Tn&I6!ZCKPJL@>P9j>;xw3YA-mYoEtH~WMJ)qfN?6l9-j_&TXN~I
zE3e&WIX7{;TPgOIPUr4QKX;bV&uvZHPlrs3{+UBIHXwLxyZM$OxrsidLsw?$I?vJB
zXS{K2&9ISed!A`DPOaa^(QPlBXucz%-^btLH%Ow8xj78&tS5rAIHeSyMR+?}d-bCo
zJUjShKLho@{z5@;kO2}ef(6ri4*f%Gl#X|!po-<|_?}$rsHDObEwtbvs1=4^a$plk
z2at1|>u)suR})PyT$WCDxZ%EMSF8IBnm>ZJ!;G0~v@_ZIG}3iWP-$nB6I;wT#=+w@
z>>YV8it$*>5-xXr39~`Pe#Y$L(1nu4{m8gQ*A5b}+5W|2+Re9}iAEa3(df(HQGLX3
zf7zg9Qocn|ySnX@Hv5dD`0F#SlWIUqI6`T>J#fi^{7BkgxwMFV1gZ*WAYMG~S-W>n
z{q+aeQ2(J@LZ!Bt#&h4<xk`ryC8+q?)L)&pT%Gn&#Z?VP-Qa8w3ktH#vJ5>G?djWv
z!QJ!>S^M7^qj~rKcg`p(-_7Gx8?V^7Q%~{SeXrd1^KB)-*iOd%JA<P8<gD&_qc%6#
zVNrjr4H^{lM(Wk`jV7Dc1-_xRq*EOgl%qMVykk~BZ=hwGy){5@4n1z&mhZ+Fgn!+3
z=->guKYoKAKN>lB4f5X&+Kla_TK_+O>n$V1rE$Xnj}rbhJwl7F{QJ-UqJP9<e75;4
zgu`SB&4(&BIAp<5Z?aT$@@zO?-Hr{q-J!Q9HpUo<NKm{^m9$KrcW%%i?!DNH;;iXA
z9CwrkV=X98%X9j_@=SNP%bkJrfR@)e_J`3MEM!epmw#<tH-oB^2!!l-@lg;*lbyZs
zkLfeL(@&__nKZ=(ZTYQ1k3;Cr>v^5rNyvstER^_pI(6=|J>?Sqmwi|Nd;R`T-;P>G
zV+aVo(p<|iq#AjTl<vUic#5lWg>N{0kgr6{+sKb>Xmp53(d!~j0AizGT>b`zuc%fk
zt*b0plRHMq>g;uqjYS8^g+r#mx&@@P3X6yE-LZ{`L@P#5FgO1K!)Ubq1&SoPL^gKD
z=&C`x{}8uA3(4B8ER3-}jy<%Mb-tm|VfjFrn=0ap%C5&n9n8SiI$ilV3*yxV3l6AK
z!?B#kSh$oJN+j!{NM!3a1h<Hq)=!~9YFf*l@i8R;Hl<u~9^6pb=^c6il%vi@aS{!E
zjOCkKN^YhuZ1`o0?Dl3RzrJ!M+5;ikf6ibz!xtrpHU%b&Cl0*q@&aH?G-*KN&h;Ox
zzz4rIEV(hL-ZQF!Zh*wvx>TSI&n&;}45MVES;bWL@{Xo*`E@6do{Kkm<^^>978LL9
zZj-=W|3JQNGU%6LYa%Xo0n2%Mw_9IB(ZokqREPAQl23nZpBt<B=Am@?97E9_lWXRF
zeM8~`_UbUFK4uij=X6bcuHL`i#OrO$aUN>TSRfZNJH<&U{Nx<4%NEtv{3*qOatmhT
zLk7GiGg`eyY_|{>{deE&c3n4QalN@#gI>SMgTDUjeGhgICtnn4#@M@r$qhMjtk{Qy
zhGy}i)7LZb{Xl5w>><T`$S7?&ogxyP?{uiQ`&zNlLjHeIu1J&Z%8S2VbKj);n~Txn
z=2Cy)PF85~>oxo8bmXsS4d9WCTKW4QEt=}5@tPz28}U`&AXZWv8^9C&*X=Sqp&I`L
zuNUUZHQ5HW%S{Jv;dxLWmVJ(axF!VmOMH<|v6>!N<q3pkb(srC@RqQzr5}}4h^ht`
z?Y_SK6dSzV`wgLp2&g&88(28Wm>9Az<yDFeELQ&S;QW6JT+|1m4R)sNH|cUCRpCD(
zC<L6t`2xpBQoTgBhzEc$dIRR%r~q8zkl^6lWNiy^)&-_Nh9!rk9s-FCkp5KzPlgPk
zpm&p)CLokj6{V(1Dx!YFXRu}hyjV#)F7V^n@l*gZb$P!UfxqmpeBsfT5ziJpgs?h?
z>5d`agoz#aYHZ*%uIU!uapHYNS`rHVZ&{782pDZV6!J|P{l)x<6<5yndZH62n+wXV
z00|5PN~akMaXP8~-^7Wp`5^98B8rr>PFs0FoyhdqNt04ue#N2VcGLikbkQFV*KeS;
zfR9##p~5t~Yp&|x1XMzBXS1?m??laqwDDNIb+C@kKV<-wcv_devLEn&>tJ$(Ms76e
zw+1SGS-tzO44b}<=>UIb&V5Bc)U+(4hBe;x_E(4pf52wHA(>~FJi7M*+H(=9^nc7u
zrgL@S2nG>aC@iXh+#qCoks<t&pLuk*!GCGPm}3Nh)IS-!!}FF-`Y)prj8Lcujul<V
z$7!IWk*LN|PEb=<`nZKxF?I;AqKyePFiIygNq2KvZH}Q+!X1un%lbVW7^3xoRh~F5
z>Us;FNwYD8<mkQ(7o|32oAv~pUV{fzu5xU>;SCdg-b0bUgn1zBOAGivZp{(?Wib~S
zQN)MCc$N7NevDQk8ULv)A^{bH`{|!Au%I9?`fN<6X8QyDu6U>Z65HfYZaB(aC<tLK
z4@tqSEB;q}MD);QvTTn{l#XobeRTElR7xqM)WVYVubCIv7luYg3drO^($57;nS3iE
zWSE)wz^*VYU$Mv5F3IA&fQ6f%bKmFcHCrf7QE*sj?WWAg4e?1bE-7TT%Qz~TXsA1{
ze1WnW%qR_enpQ74@7{KF7>b)n0wN=6@i`R9BgLZI^>XJmLupczk~p;_adpsjJofVB
zlmhH{#n48?QPjxX5hCzooZb|bGejSE{msv_+}hRJi5nr8aK2ER8MERYn&FV(;L9M?
z<cv3x%iTVmm&9(v-@+Q6P`VZ~7UO?4J7xt11S7N3_I1CW_+{#~Og57?L6Y@VS<Yrm
zQn09#2DFah10;}7HT~s(wG#g;HF{z3zoI}Q?VUyMWp0j8xyW6$vx5e&hZKV&Omg;m
z$PO|c@e>0qhD>91BW`vZ2Lv(q8$E%U>SazADlpp(-lGA3RJ^JN7Fx+8NBoPtkM~tn
z(4^BIjmhE~?O-$6Gig)KQq$@9vMM9RvkdOH@dz7aoG#g(jT`8`sBcC6*|wyg%OCgt
zYwKj-?z9QHxa6K652FI8%TbQib;%oRWYZTairzcgnTpsy2G1*@l(0fuB0YPhH;NY#
z^eAaRrL=RX6}y}DRafIxSMC+$EsePtpqXAqQgCRD&x{Xas3c((mYh%~RD8gWNIoHf
zgFHe_P(=2F79zDG?1(|w2+XSANGL2);tZ6$oGZjzMp7Hg*3fo~pllku9R{3Gffp5c
z)RHGbXK$1wNx@O{2%nW$2l^BzAyV?vk@j{MX8`-PX?M#P9S;+?=Z=v!Yd>EK+Kzg&
za$nMVT82%Q4!o_b<%Gi=ypYctElo|4S2y-F;nlX=&KX_vR%%50->LfDY%uV<6#is=
zW<LhJeBnwqj!!hA6EBzw!2JxpkCXc|c`+NFhmRfYag2+lE<ffn6kG4d({eS~O9MWB
zFQe2$QL|uaayneU(vXy!5Sd$?>p#<1KJyS_49iR+BaqU{!ivt@c-W$n!FTaP{f0JE
zX{td`=b=ukAtD`?Fi>NyhzW;F!vrC4yS_O=Npy-#4pFjUSOX0p%0K#Umd%7p`Elua
zZG!iEnPI->1Y6>N_+(bOWhM;Y-hzy08M`TAvlNWXWFQ*b8OBzu6bTaR#Z%_W%%qjl
zr#pX92-N3mHmcOz;I3R>SUU>%Fd88fcoQCo3NJc}!_laVn`zjITLdbKlkuu0foW%L
zWZslz>$uf8U}^oA1J+q_z)}ZqZRlEG_seX4YO9*U767f{kb<9RISSvz*5ST#@fRh<
zZOeF0*hT+qaSp^Y5%e<=hL8jUy^fAqQQBhMK`Dj9Q}@J9&pVBk#|-s^M6+~!Z`MBq
z6SH?9RbMAB!k&(fj*F7VF+{{oKA*1=zZHJlqvK$6x{sUcD_-0?lI0QK5?EPbge*Ps
z^4H8O=MD`>rJQu)Gh)?uBHPV0B>P}!Z9gmH#+714%GtVLNsCv87dSU9Gb7K&KczsB
z@MuEhfHD+dwXCxTN?!uG!lguGOTTJZ>+9%8N(L=JXRi=`5;}Uznne<G!eo(jTi7cR
zmm1|X&4>ZP?G(1yo;fGRWM*NvWSE^<*nZXL@VMbb;Dj>Po$1fNH}%XHD>`C(;)}gI
zcRodq;<VlJ_zAdm9$<fGlTZ`(7>GNQ(;>l_F8!}hk3Iy+xDH=%mg#OvA1nCu9y;~+
zZT}zLDI7ZrAKMX6{a%fvul~>!kyR4T##~1HpuSMMd@Q}lj3+~%mXIKpefDdg^!Oy5
zImqUHoKmI_ImhinA8K5GKi@T#3ChU_hJD4{&DY=}TYw!~%c1o6^WLj6w(^iMG`d#N
zX)+;;8)OS_n;13DkfNpG*Wp_^J<a;d0k?v>jkfu08bH#@pl0JdK2Y!OQ}_hrq0z+-
z3#^$6u@GD-^ih5bq+M>HrTW_a(zUskr>+;j--I9{!vx)a7K3zvCi)+hTQnZ<;H8qd
zakUljUY{6j=(<i8PbL$Gk@N@E6-%0OX+(UdM@Fg}gXj1t+J&T@xR7Qijq+lImQY54
zaX^jkN8bUQanxfM)+n<JWCW4&k5$6M%dmxrlZU-WABM3t6Xs`D!?OoScj~FP4PAw4
zDYQttwqjZ1{+)N_4&B2C{o%hmIJm<~EvuuS|FZ0LM6_J4t*WIJ9Gu}~M%Q~gH|Mf$
zLQ0Ea%!vcH9v>W0O-tWvCLJyt{rEZ&*9O-{%hqx@nY5QIGfvV+&-1YNs2|Kmq(RF9
zgz{Z$;C@e(Cb3dxkpOn=^@oi#0#>%@y<9}H33Ky6Coxc*#)b!j)rKOPryhA<uNKuH
zEz<3;_WO%oEBYD;PC0e8-I!7CZMYk4)W^e|*GkrsD=8|=(z=84k$}_YhXpVU9JDty
zG&E%b$YPjf<}40~mE)-ce0+JXYlbIvHc`ri7w0`<na(A+|9Z?=pdttk^3KVcP4UTr
z1X?<t(jp1+AHbFV`+hV&i)dNY`t|7cpqOBpfJa6-TT0BSjD>XNROIdndu-DO*5f9b
zo70%K!oIOoFf6O3>%LXzkDSbIX)O%}k5Dp*Yya2$Z8JN3CAx8&GETthGwVsTi>@lA
z(6ak6w50~qk!c_rxgs|Id^xk^e4~j6Wk`l}5rWAh)Ob@?p9hlPulN-LWqb?ay3@+<
zNOR?n(PH1mZZm6MZdr_S_^y~XIhkz=ZYa`g)T?S4o{5?vARDi3YsMF!)zSgJvCSQp
zoM9da+Fp&}hCCFbx#pW8YaMt+T>o~1-3Jd!`XWlIl+=8&bP%7YlyJkk-BZkcWI)#s
zHhJIZ)x0h()828gwLL<qHpPM^Ipi0lMG}N(!yclVeZz?$=0WU83YzPQ7UB#}Y@x6q
zBej7Stvjy+in64H<j6ZrlXDITb{#2&*c_1|kXV&}dihn{ywa8qrF~KGfI(eA;~eqL
zLW+9%0f&P9lEj3Bj?!Fmn6k`Zswnk!`6rl05bdnf-*lqhCK(UQopM1fDzaDP`w8bp
zMPDOiZF*4v=6I_cNh***;rtw@3OGWK?(0Z*TIWCDn3|?^+jS-DO)6^(Cw>+aEQa75
zbUsK^303I)?i+q%?GGG5vB2^)Hds`%<9$mxwj*9VF+OYot1De?t8*&Rb_y<By=JR;
zXj=UoyuGX8$dYQ=c>z{%w;QzR-greZ4Z(o&gZkFe(CVdsNOfz|(0BdDQ??Sg=u#A%
zGIwYy;t??f8~8;_UWmS=C9e^Y5|ZeZ@V(nnYG{Zw7H2=UAtbVnZh_u6w(Py!rC*RJ
z1psf3sg0(jqYWxy<a!x1$l~3yc^bM0Av8uEqP{6dh?rM_eAUb+A)sOAZT+qhM>2+w
zJM&a#Z?8m$ywz4Q@Lw$e+3&^l5Z7ij_q~9fv0Ev|2S~-I+7~Mxq?I3HDu+CRgF)+y
zWVc$A<=Q7KUi?vHMu0|{HV=MuK6nYZAOk>%C^AUSA{+^T;6I`hnv0jYy^jt2<<`pK
zg+Zy~ja21Z6t)F)n0lw1wSPmkm?7p*!23j#3bsn`FDv8||G3<-@rKZ)&?x+eKe&pI
zV_B{T8c1HwF)evHwIQ<CSacMknP6wa#CQPHMNwUlKq}AOS@DR0v@5`2Vj;{bqeqAS
zi42Yq4q;lXe+ZI?d7=+?OV#lF3@NklyhR*Z@<f0P>1jmk-c}4s+mE#$dIu7bUq#3z
zT$UtO{I;~P19uHZjJO$N$5C`hS2Hd~yAmDBbzy#VI&)jy?tu7z$a(TbOE;U*lY4qk
z>AB<~&|f-ebA11MQ!K)*{QI9y=KlY;%JctH<)F96@TKq<?AC$l7UFqDdcxW>y)6Cb
za?GX1|K6ZSG(l@FegHxW_SkF?oGK`Y{cv@jal-}*)9Rrt4uChoUM4Cn8tiyW7x9#=
z?#OumYo5*j=L&Webw#f|=1qN`p<A0cy20;S2NbsL&Y9fy)5dLYH(GQsit4wqY10`m
z7e~!|6F%YjD=WnxN{9cxHEqmnzZlE)ag7@*jv1ry<?Yr(jd!mPN@y3H{59}mY1o2%
z^PY)4I_*p;IQI~hh{3*ngHqPJe9`-VsyZ3gPR^<au}fcHyA*xzq`Q=di^CrTcAcAH
z)X+!9s8+2SI&9b~ko6u_#i=U$jNk9nyInh$s_=*2*Q0-Lurr#@Ht7yc?T_z!p?Ewc
z`Ff}I(b8H?fa3J}3H>DAk#yiZPD;|x+hI)fkp6gJQuxZJ<JA9pi#H5rBrihsT*YNX
z<P99dqKSCB;djwCq|f(OMM;7R0$=r2d~cV^3L-4$qBV>Pr|2!RND*tlfB!x&Pi6Jn
z{}3NV#tHI)MxRAZ|9);3QZq=*jO(kGzcWNF?)=*ArDTQ<Rf|rGucI6x;LAC{dro@#
z+a#@}jx8D%GzT))th=|$<P7b*wd3<G-XE>BD)nqTU+6?bupHJ){g7xN06}v|EoXs<
z5x^Hjdd}yruF2X;9sChaKb@+Z&cw<ND_?-D+lS6@M`N}q?`l=~*STj?OiH9*HlbQ%
z4N0wzuh}P9aPB?Pw=GKn91Q$&j`!pS1PFiIo2S@eZSb&bzZvk%fT81SyxgEe0W0cv
z)fE&JWK;^|4G*M8t8*3{x3FEi=iQc`6BJMewH3L$I(<I0sa4h2y4tTbH8uB*#GNL=
zN2H?T=WcD+^we)z>{un0L>evyKi)sLV*2zK2>ir~Mj>Fk;!eG(5qX2X&*#lI+<kfO
zqiV~|TVzNb0=otF_B~qtp!xbBbeNHmk<TCWet1rwK0T`;Sm)r|``>AK&Kmoj*2rn&
z$4*^}!lh5ixa&-~>Wl8tE9%6AnKS^#tFJ1Rnb~_(TkTmu{w%!6G7{UzBG-s8Dk&-P
zZ87?&7=fep9?0YbVot9wdPsWOo9iCV>7{<uV>4J#%&*`R%UL?To_T?oslTnp$a14T
zXdxFCuLf<!u1+!^S;7<7!m9~6lu737cTrYNqW|1!qG6<?2S<#1I2TQEN<oCG=6t^`
zN8q}+`yEcTdvjyjTeCw&lXvQE?$LTt??<uwjjN3}x23E33Fqxkq?-LzTJ`#J6&)oD
z0z080m3QD#hti<~W3fklP~z_mj&yfenckY-YY*H#Qu|ByIdZfK`j0`du$)2HVzhSg
z!_Wq+*%sr<-cTo__Wqsg4_Pub^{L_A$?1HF$HVOQ8&g|Wz|KNzy=Q;5_$bbDVm5Nk
zfkzNGeaGojz+;A|l#9vbk@qJe13Hr;BdWr^FeE%YmW&nl*A+@cBt1pv`QYXQV%{Q%
zbk*F_qWb7m2WfBHKd$av{`bH^Ie@#H9Q-Avr4JX?pfIX}0`2|!st8+*lL~iyVMW>w
zYjhogf4sJ|6N&r8)J7%>V&n%Dvh^|Vt)&&i9uys@i^9om-viID<d<qc#n$%S^1COZ
zC?Cfq4{GlQV8O7ZVUd{_ZHd!`nxmtmz<aQe*j};g*RpF1hy(8F^U)X8in<!b5_S|P
z&N~_z-9Ze~$HiQ~=;h?S_3x_Z&)EIX4DXJX7G-fjD(O-O-CT(hd3tc$9>_44?N06f
znhCHs6KpDbh01dbwYZilXj`1bl$<SlLdL|9Q`=@Tvrv*7o-DCQ{Ab-iiCdoyaxV3L
z5mY#YP#47*7>ge&BBF_m56OhgK9BiXhcdMW8uKrM1+2p_49OJ*H1tPGO3GYdG`A9B
z+<L^2Wd($T<p{}*>WjY=R8#xIs-N3>VYtMrgs|$2j0|xwMY<_9(wH#9k^!*w^x12D
zp1NT5tyF!s;o+(EEsC$k*1yB^Pr#`i%3FWrGD&B?^8&9DK-f0RNsK^?w4_QYXTp$L
z=qQ;ntn){2nip4+t9<wZ+dAN>3v^G&nu{xRg^&p8uRi%LPcrf0!@f^tFfsNELN7|y
z1{W)073O)*C?2T4hFrWd@6K;+w~juXoSe*uiZ=lmK$2t<SWu@_dZ2rWVYA@~(9C~<
zzt1^%VBU*a5bu)D$b+IE;wzSDFVPi}slyIPE<q!W|MmH=0Kc3@szcs3Y@+f6xng1P
z#*G_+yF)&7z1Sz=;X}VUAKYlU7apeU%s*-578yMJ^YvFxt2xC+N;_YPWyaVpAL^cm
z*0$|X^=P^01$r>mw-X$oRZ^%&ydfANRuRD!QdSg7B}5T=W540Y(+Wmp@P&$p1$-`b
zHIuK&921WwH>!VzQyM=k?23|$@amj}s|shBv{aD72^TXN3$9k)Z;0r!1^3zo5CzRS
zYow%7=T-Nn&MIQN!_#Nnu-F`sIUR@0qIY>7RV0qFbjWL@%?3dBs9AQsraw?#hw_j1
zokETh4o8tL@?V5?x56i)8N=eC_4umDqj@&Xz3&kfI-(9om6{Hd!?8o7Vj(N@I;Qme
z%Cx6Dk`{8>@nc#~(+Tg<{!|^38~wQTbRTL(VyZOO&^9MMXOPTXlK#gNkr1Nn{8e6X
zJq{{f+zZx%FGW6-(A^uR<J0+jtz?J_osyLBvwN7$;G~C2$fn4y5uh8v)78_{Bim)^
z|Jc#pSdH@ny7RDq+u$_zSLaM!+6;<{iVA}vjO3vNe3k*MCB8*(2$L3CX?z*7CyY)G
zBh~2WXt6MeX-%(!=H)0hY-s^;uCf4vrfc~bHr#hC<+fGV_fmFC<4o_c2%>U)RG0nJ
zG-aqnO@w$$Q(sxC<46myv{6jnz&cg-eSQ9{%Q&^2-Ceb;-UsPzPU`M4woX=&)>VS0
zqJnrUh*xCL10Il+Hb||S>z@X`NZD0%->P+RGy8|HT9y|s6F5$KH56s|oV#}@=3F{-
z0akGT+xyYg(5)IN&nZ_PS|<F`kxbSC-u2iSzdf{9mJmgfo_cA#3W*9ax%Q6rjug$d
zR!I|Xy7)X#s<9JSw4nL?z_UVc7gyI^tIJ0VYm#$(+Yh*Y&Fp!nc<!v5^qXr}MV*^$
zJ?GmBU73%L)hf0(8b*V7kD#6@Pnz~#_flPYknwrYHX2J#N>$8SXQ12o`{r64(AGIq
zq${COu`&)_OF;-h?Q`eof#NiapK<g3_f_8~{^=7+8|89JYm}V<A1c<4X`-@XRMvbN
zUD7!}dOB~GW;%;;Lvp||_54+7Y5UfDMoWrrcdvrP({Am9{&xfrMtz;<<~L|u<Hn7h
z3jzqt7t($gTUGblLWSP3W*zze{JIP)!lQnN(*pGJ_r}EjCTPIJ@W&+2l3fYUl%_Y_
zxr#Sx`dF2Qc_;6rs{B0$V;V{8&y!l5vHA|Jf3y8OPF!od_1?Rf)A2Pwzj{qdd`nvH
zJLk+9su|J_%~G`ucZT1M3kYtNX#D+RmAbr0U;oXE)yrC!Voz=7(4_IVfzup^uJQD&
z6=unL<)1GQx92`yv+WX~mN?erk?$+cdCF@d02h{MchvNQq)Il(QhW*?<22>LQ%iN#
z%ITBa{1)s1P07i~mqS;Kj1wsI%zU^2NuN#4juzUirOHmD?*7$3QixvE+nsp~QLUw6
zqEp@UvL9`l$JPFMnt*e(<MFYbiV+Y<VPbsOdWS4z<KiE|6p%W$YM;xcU|*q3AS!!$
z#@lTd&+vpGqtoUfp6_%)wn+NQc3zM58^eS3sO<e!(@$b7!)sPhs^PG_Jm%e=S7^3v
zyW~zknKQH&#~={@g!MBl*|<H{!@cqnO*&tQj<!`^FmbY}scDwdQ;vtQw`ce*i328`
z+t*yZ=4(wI7G>6~)$5)ZH9-_{Zp?AB5CfTme)sldY!RHq9IJX^(k^eH%wd<YYESDe
zKazU>)M&9~v}7Sc!EF7DMXQ`Pk3P&(ZjJQ{-6TQ5^-z4~igBm%1+*PFHN_MYg*3P>
zJoH4}4X4CXcA$JApoCq`#VS-|i~E5HHo*OaNWf+COmec!aF^qOGYY^Uryfr$=rA$^
zeO`o1jYrZQ3M<P|>fg;J@-mU>qXfZUI%oi}v=?v|fZY2zYig-a8@QT&Sq`9bApRL#
zYYYv&*k3ZH8e(QGT@cFdQUws~WSiT&`g$2&n6sHUyv6DvchC_5E}0Cc!4;aorJ)7t
zaOa%m>E{olo@bBr8+t1G02=0?VA}%II5*fSMeE>V8J_gGWWuRKhvp%kr|M{4yIZPo
zE1ylglI1d?<am~Q=2BLTSI$UI8Q<&kmguV7Dva>kxo?hg6p8|Fc3Sd?SIz}y(186D
zA6RI?$VDi|e-G!4+ZTbn0qOorscbFBd3gMYBA$e87kI$J`;Kw7B~nj28Zf_32h5w`
z3?`z}Vp_zrNlXk&$h#$R6#Q_q#Q>8Yt>ap0x~9kKcw3y><^8?haYScAT~jzQlD<>w
z+u0Hi6B1T)ES*E2Letq<pa{o|B(^|l@g7TmbFv$F$~W7?0dF=Cx|ZGT=VIY-ao4<K
zRkfE+OrUK50y%4KR>Ue>Jz&AEPJMdyDv&DE=Q5s@=>SDn7Z>WEe(587UmZAnXi4;&
zRtET-+&!!_c(8CJhnA5ltMNK0?%H)7|E_)mQ@ia)&Y<N_1Yo4%U>)<U4O1%WM|uYx
z@Itv2%_EDklpf5W;X;L4=A{<;ofRjoLbC~TvNSgcZ^!exS$3T@6FIEB?@B4~htG(p
zj?NA4x$C<Cm{uLjUljF_t^gjH$sac;35*~13UF<;bJfY+0i4_E$2|e7wrl_B|9QO(
zTq<qaKArOrtxl!5Z7z8^VCZ=^%}NrOqiP0U)@5=Av-f(9EJl|*CFp`Mly@e)nffAP
zK<X6#3FWg>SGAVO#wKcipIJ_i>=+An<=9ZY%~Qe*b`NlEYK4nk?LOlfB^E8!{c8sf
z$zlAFkCUUSK!qyaNe{KkEy{DYjeHMY@LTs_{LK>+bX)Fh;&;vB4&t&S7f#`gMSX*x
zmV>6qm^Zeq-#R^7`s{nAW>Jnd8EQH(jkq_q&jr?e`o*4MCPSCV<Pfm}OZ%Gzz<H;O
zabx>Ha8u0=<5nJ=0_}>PP9de1i3~1F%j9ef=r#57s--mPY#COEM|>`do0hMlzSUo@
z-!POw+_-X=Y1ZpF@uRlKJzm|?<=MnxN>{zgaAB#Y20uYt3Oq0`T{Ar?X`2%*K%xKO
ze$`c3ZC5!w<*+(Crpr1B^Y8{AkZvHkqHV+KzPr>kgWS|OT3yBHsLTmmgkjQ6f<Fk}
zi+2U|-X5kT6}O}7n!lLg{O~$=rd5(8UbxU4{=-LmOj^fw@QYzXX*Ac!%g<LMmyjnG
zmgbIs1}Ben`yO2z%&Cu?u=eHk>;b}r`MPCJ&DQdK=?2xZ=}{fR8#fEcJsUdf3^2qr
zAa_9L9j$t__M!Fdo@a>O=E)%jN4&&{@W|*YvyQ78A57fUtJlGtpZN3P*Xl{OVWi59
zw7*$9W5I%aSiw%KdhOYH>hZtP7Gc_2@_|ii+l$PD!|CBD2=Ar#ks!+8H2Wu!n9DB8
zTWnADzp}DYv-M?{F=su<m#37aPOjM4B=OOsn;=((KtG!nSNeT@MKF<6?rttq84Zdr
zTS*SleEEuOxF9(C8Ck6@d$M&V1_iT{WBhb_MS|U5+N`o!`4E~&T3#o&_>;!o^hut8
zU<$LZWeg?KDdGkYOh7;#k30@<EUF-R%2uiXW%7A}NuporKWijtn}xK@>56bkRpF#r
zq!ErmBdORI(eNqGw8pLUj)0vxAY#}o(jAn3^MIs>J?{0r6f<GP0Ll)~rE)o<hy(b+
zSn<B%JzFU3DR)~?Wj15-7UkQB;{t|T4H$=AUjh00v?Ebib2;}NXmUU7LZ|N^r)j4?
zJl`Yxq)viS6IDgMY%Wp3cKn}-#!>dIs%`SyjJK)ZjGUb#)U%=;CVVg_L=^yOB3^3w
zHTAwnnKH0;*)`m|+SR;{8gt53Si)l;cc<byw)<^r;ek7T6NfV~Gc+tr_4`6zEp2hl
z1EJd5E$_2u`hK{g)wmRGB)dE7A9?g!<4N6ghApK;O^eT8PcHDT02k`6_A(^%es`=&
z-iig8!GN9P>hg!h`eYTh9TInJjP09a9DwmQD|N<6pkCKPLG&1@7sl-RjzRJLq1iAx
zsn1NB|3qfQ6vh=!1wDqD>F4jYv;EpIusWlfN>F57CD<{(%!5U5`*0M;g5R7*hn|qB
z+hkiggK?LjSaYhDE@s3-0#w~Tt^@+#Q8=u*Kg3GmL)ZD^5O95?J+t?OmY+y1{Q9+1
z4WjQ-oJk1O>CbYfM<=aaym-GPo+=?{J9k+x-dKAklvwnm$*!ks)1z24dhp0C|9Rv)
z<gaej9Phh_1uxywW$ycsgN><seU>BP=p0q8>2BiR^m=NT&YCATFK%hYx=L`4YW4-i
zQE-eg=b}dZhx6WZc1q*?-A$4#L5@-T8EsvXrg-b^xPVfiMbAR(;8p!yCLRb2ThHBh
z=rS#D=T}rG11`H-k5N9T5O?#Lo(CxWHhJC5yx!`cvU^IH3S{Ph`WKVs7HfAN_*16o
zMW@UfQ>3}7(9m!J*z6~m!*}IE4wjYMd(Qms7%3I7oX11QOS+GOO*c_>eeU?gu)EEU
zcx^uVjW~clQc~(b8@H$#GN4Hi<NeFnZ_41fME^GB<yTfV@B4XGgr5blU|?L+)Pdqe
zH@VG}Q{&WpHC>y2_U1V%B^p=NQyjkh-KegC+c?v|5Sd@YMzA}S2hLf?PKYb3B!@Zv
z*S>bI$wtF#YRh8-YX%P+R?ey6>*QzSUVp%^oZ^;L=%aje`(s1fuP^#Vq*wop4B0!I
z3S_&~qg%M2gZS5NiJ!av$23IO^7JY8Wdowwlyh_@JTRN^yt&|c>6s-$XIq-7J_pCa
zQ}}D8o^IIl)`<ZRua}W_UidbT04JGFS**j)S^2R_raw4amC6Qw-O|{0x!2>w#NN4k
zMavlU=j2}Ip)(>HcgTJ0rB`(R-o}yboX;83f_MQ^-ZuO+-GEWs;Sshk%_3vVTg?lI
zQpMxrVB5C>+CI;1=x0u_s1=sx(WCyaKS1@3U1|^o?s~e9?P&42=UI@Pm<&0GjOmI4
zw*B*5TcxufEyhI69XHLGh^G;M^gy$Y>!+{rya>L6{VYpK?mjg?ulrB27V7Ty)Zp%+
zEphTVU<JOzqLZ+Lw#1m3@Y-?Jo7BS0R~%A*lBpKHxz3W_?eIBf#(#U{&V&79?!H@*
z(<*L2X6Kuumz;e0^7LwJYip4euGEi#Dwp61Y=+j=Ej+YIq^F=-#j7WpPA*fX>+yX7
zsRq^)YsJ1F`6tL6fMveINNxC?J~OSeFBL9%wcb*pw!h02w;=5a!;b5CL=AM=`PKY4
zK1B}AWl&I)iu9G5Ic3FYU)DBqKEi~DsO_X7sY^7Q$aA;t;`evcBe-J47%oUcc)wOW
zhBce!aOymR0J6(`82w^t2~Tw!G34gA@{5NKNzr3A8H%K1(m4{=whXyJkE_>2kogBj
z9(Zu8f2OUQ;~>yRzgw|I)OLq`(O0_q)H3bjliQd{eFwM?dpW+lhnBn6KI4UT$tt5A
z%>T(@Cie-`Xi*i{l|gGk3ZA$iNz-e(_`>^-oWl^jyl?=}Rdm>s6+gsIEs(mz#OJ*u
zk6RQ48<{B1+BRtxr@C}ne*f?cRAiGkdI|9ZyMtHgM@9Iu>St1#I9Q3WA1&ygl~bT<
z+ZoIZUiJ5p!Nt2e4c4@Mne$iT`&DcOb-#-p){rsEITsxm>f-#>ZsLkz%M>&R+&c0^
zJ6>flxSa#Zk&G3>%;jCH1r(bYnf?Qnny8&wH}`*TW0}s`><qagums|rWhvx|9Jnz{
zHW+|=<62!}sY5M`sLJHkg$ozDdgKMgysYc1X^AAETZ3_9jHWh&;?uaSt+{sO@mPX@
zX~2Z2gNgi(a2|Jasnr*zKK0H!d_%hQb&`4$M+_5<H5Wj2->Ypi|H+)5&Q@W*nKD~3
zp{~D)XQU`zV#)_7S~oc+jWrso>a#k;|ASO+Z{Kt30=bmwz%!ltR;j&<KJz9d`R|kO
z-=SI*7X?_8OE{vDEuyEYBP94_4UA>^f^I|)BTVG+1~j`4M{}I8Im|9_g!l{y3k0Wt
zZ{}>Wse~hk*#Vj!bM0Bk35$Oxn6Azn`g!FOIV$^0%BH2Il{;rXQxPCszJP-A!r!eB
z$^O`qtQgn(?ULX54j<fCLnE?g*{g|(e?3<iuv$!cnoa+({k4_UhZb}E={5P$mDSr#
z2u<dFb-CW-A!lkx#eE@SQIX7vwjUQ%_IoScx74q&RInPSVU5LxlU>m{k3B1S_oe~z
zqNp<E_3DYLjK&eO1!<yaTjR5LXP;O>{#T{PzsH`r_^2BH3o}MO#fxiV=bYNO5zFk}
zsrKq!88+VZ*D<$uyP8?n#2pbbo;xTvy-RBiWtX_eQ^)7ONGqAbgCVJW%}5hSahk4j
zi91qYzUzJW-8bUDjc@;ookL(L6MhB~NE6wM>+ZE2My=#1#HpwVZxs)%%*=jRa#rR6
zeVfOZoE_Iq4Ng~$;)Qe>wr{{m82~c2?Fx!1<IX<sQtpmjfK?Nl*70GdUO<zU9O&F(
z*j-}XmI*nM`RHx2nk-Gwx7K+CJ>bhq!vmx5awrj&oSbsuXWJ!Dl~a<mzvC?9<%pA@
z->rQ+uhE+4_BNcycC_Q-Sl2WT4K$t5(wCil!rq3b!wYG|3qJY0qtzRn5?m(YFf^`D
zEFpO7KOy+=-D%S|i)a~G?NrU0-|6H;Q-ZsD+#{3I^^m%Y!*^ZGp2B19QR!2UXI1ts
zS!n-ibky~+BE{1TJUiRo!NH9}5<5~2ub%l2VNZS6tBzlaVfU6MAYGZXvC%IgZ*D~N
zjU_TT^c|eu9RSj<bq^H2K~fuBE+*WMO?(T@YZc@%#Amr@;KElIKQI8NQ|3qt6EQ!-
zkyE8VxbRrN@4kXQVJI*aGn(~h_-@$na;U<-Wm#-_s%gd&D=IfnxApMw5P-K-K!045
z_Ew}4jHv)Mq(9*D{+Osa3or@zxlFA^{WDlRlp5s8ufqa=zW*CGi#0-S>)ESHLz{M|
zyyH{P(bms1+7iES-#(cr=oG%P0*-HU(qNojdYqg+*xO>!zqEnxw5x5)5UgJou58@U
z2cG6h)Bc<U{Q6rfGG3e8X%te<He2phJ?ot_??;R-)m7DpWe^UQ0U(WJMLI$Q`e^tZ
zw3((o%(?EPRjjXy+1G;LnL*Z;(7+RJ+<A$k)PcOc4jwNdv)k^gbRlE!Rq3zvcAG}P
zwwuL<?rWSrsL%<(G2j37)u5~1sWOcxJtdhxo;#;S$TFSfC69?T>iuC0)$k48XflC-
zS_ul3K~t0(;&@WF!<ij3%suL%j7QN=Acp=dp(}!PYLot%xv|oS@smC7dr``jlkc2-
z+}3=b&U@Bsz*?U!pJA>TO1;i$?URou<VYqjv+GqOf1cQ|Y^R^eeI@JH!*dWEo;rX1
zZr|6kgib2PBK?x$#V!MxPwfBK3R)af+M21f&FrOmOAIHH#7^heL&rhq7gw{m{X5ih
zoii&qc$6zYw)Va!%g!l9p+d{_bUhWPH21c%l+SA3fEf>&N&HVZvH_d^`)9v&zWF1A
zxT`$mF_Fls=(`jKOauJF4hFYNEw_B|I<mK>`hkbkC`R6S{YLJjWbloolC#XE5rxAR
z87w9yk~`NuRcJH3d@(xxlbzR*?`zIHs4E`7+ebK)6;RX*MOX>XUS5<TfEs)m2Hcxd
zWN=A5Ac9W%y|?G23ktNpPx2%hFV2E>-&4(4&S+ROfrPZONa9V~hxBb(#a2yISkf|1
z!TTO=pEL@DNdW?!)Q?#?&@`OSC}^L-KFaVkNZ2YN8C|ARXHi@IwxQJ2#ATr@-Hwf2
zjf<K-G!~qpR8&z(2$3v20=h8j+cND29JQ_A8<oGI{Gzst+72fIyO4#22(X2bQqIu|
zN`Ib?JY#ln!}5VdMoyX8dEr=?-IIj5$U64t=j9r68%g~s%hyzE#DDif&;1X(h!+a+
z(e?UBnS=cp13*+1qTwhgko6%aJZ!JuO6iLIK6j`5fc^Ap@xONJNL&wG*^%*178jq(
zsDkf1-&T)we;Klk4Hmrc@a>(xwWrro8QgU3H7`AA83A+Lpn2b~ZEtPxZB)+T(T6SA
zVvfms+h$4(3;HBDT;yz*Q;3k1KAKmnxFmEZM)-a%WdoQ@>VQWdoXaAoD+xM>aDBwI
zt(oU-`|nusa}RG|Uo5;i-*}X!2PM>LJ?#z~`+eFcMtP{L+6=KDnQ(TbR8uytPO3QD
z-{#syo_Wwn@m1RxeirT&i3wF#v8GyC%@S2^mXG1k*g>O9#LMW3mp(YNU3fcH)BE`A
zcD+(6=WwRp%qLyKYJ!hMi62pV!?lEoOjqtKYJ_M4GcVA(V{$ESS?_%}M4}Uw(q+}1
z9B{<Y;>@5?JN;6XomrZHw4Cu;Jrt=)@4J1cBDq4L{~*9R_-0=$x^{Pa1(y?lw)`4-
z`O}|8N8OXP<L|eurP}}Z-UienHzw>7w#%!wz>Di|f29Bp-ZxEEy0s3?@B8Thc7p>3
zu$2{Y#=84{hrw=apZ^=DH$uII_dQ6XZ8agY)P+>(eei6$&YB%7$ExK#HqolrhuHU+
z+CIvwU+ll)5bZ7_0&o-=XlNJ<;}J`XA=8fbhJ=(e1ws_1vieBLJMTLxqdV$~91Z<O
z+~mX+uGU}<q#aqEK1QZ8=@&Mi#Vf^X^!#Gz3{}QxMqTF~BKbUWOq_FYfm%bvK6=Zj
znoq!kt^%^m0NDn@<UF>U?_ZgM<23>ep0adUs!Hx(V9+c3*HB!IO?WPxzmQ2(zt^)_
zL`lRRr$(udu#*5Uyjfz3L7&{~FW3bM9o*TFqDA7%)%cI$rJo0=6I};f(Jkr)ROzL;
zPa0gr^7aOF?z<H+$pcNbJ#padkP){~(2DY!0bvK0UF%(pF_sWT>1)F|i>W%qD)tIa
z>x$&PaHGxhLBs;G`&-eGV|&z>JxDlYMG{x~mW*?xx|`?aFWuziwo09h_4m@l$U|!$
zNOW_hV2h330uZ%Su6glvjQ*r3^@zz#Tp*9|J8ek{YZok7P)pqD%oLg@=RbUV)Yqfw
z9!H^h=Xf}_I{dU}>kr`~#MpBrM}uJd9&&8LwISm<NJOjTqR@MQ@O~kUG|Ltb?^5rQ
zzV@c#XOg8XlnGCsQONpp;q%wq<YueRevWCMB-vu9j{E*;zXiEj`8`KX6;nE7O2x)W
zdj}m(A4EA>ZT<6*)#lb2Z!SS(l3JTA2-b{D@IPJxsBGnJ7t)?6?zFL1bH7E0R^B8&
z2i`jV_03dFi#YYnYu_g&C0#@4)LcO!Ct$joYZ9UtHk#^izh3{L;j0ER$~x9_ebK@_
z!);E(+Q})E)A01H%|{Pxmj#F`VE2W|B?q>v&HC*Y<@WLAn?g*ErkHY|u{y5uig$U<
zcXf4bpKKg>Vz*%9i!r$N9^RJz^-{@LwekC@Hum|q+W1FXQ^w9+;$y~bUBh7m8T9wc
zNmD;&6n?1p!D=i%SdI%K<DDt<-TLujmP`@PG9x>Z5#0cKoxXkh-f4=Cq2;S8=H@z5
zycqY`A%>qs%(C;Zl&PHt&ASaSOj>&|A6Wz3T1Yon4x$OdLaIMI*-oqadIyCq7;@Ia
z+qiwr9#QP-uwX$M+A7)$v~ou$j`Mxp<O3Y-q<0knw#8tq-sTji%Gfd`a6A|S7st7O
z;dl`#45av;!fO}(DW$=M8&WGy-(qV=HDn>DCRK*_ghtVK^axik?$sxXT7z^p%LwH5
zt0>-dWEUd{oKrOMyYsZqO(z_I;6yUO9yynL)_2ZXNh9+HJ_F(jfzsGJsAu$r3y>(f
zl$sd=+Vji4J$u|pDMZfQEdQ|C)T&|F`PQd{B}tgJ?fF&a?csr`!L$)~w^0x0s5XnM
ztyKe<<7QkV{ThW~z4~mZ&#I%XIzN9z>!%D&km4@p)KZP348j|Lh9K7~mk}nC2SjOz
z5}lB;2Go6g;(;;(4U~g$v=G0A*pYFko`@keJB{Nkr_HI7^(cr}G(QxBuW)2OeZBuD
z2a;FCa^h+_z<}K=_GeP!Iybit?Z^k#G2MG@hFEF8hZcDA<8$UJ+Ga)j_{-qG3s0N$
zac)hHGK}s(la3BvFX`Rmuh#Jxy|`UY-*C!+@0yF2=l@{uO`x%E+jrr+r%8$w(jX+W
zQfL%WGL}qbrjQ1O%%PA-QYv$ZWR5Z<vj&k2$(YPTh|H2X_Ho^w=iTqO*Z$Vp-?zTK
z)_?8qz1OpzRl42x?|1#K>l}{rIFAz<>t4hu2+AY}(`yjZstX6b_do$M4-Srut*9Qu
zjo_wvB{4r;UBnsJzA#lR8F<m^V)|n9!UAk$I4b=39TGE{qkfH82xcr1@h9kWH$HB<
z08cA;x+?JK-*Q&S)etS@p%e?Xy#_)cGH=nr_QF9RDG~vLP_)Ul?bkS0)xr5}*w<?c
zd>!?`Z)9v8O3y!F3L_ByAdk(2b{I$p2s<#-olxNqci!2Kh`S0)!X>C)x3d8Ld<qPU
zGh-GY4$Xt=1WP}G$QS&rL||9h+1Y=yS_pFhz@j~PcGC(M5`JTQFJNJ+c9QQze5=a;
zKH3ZWtO5x?sxvpb12oa$m5=Hp+$?#g2m6bNKLZztjfX3yKiKuKt2k+#*Z>DV#KU<=
zLK^;who7?<9s@;$&<a3A8$oOw0>}o7wp#Q3bHJnsM}be1fqK|wybF`rOhi1XQDL`w
z{(2VMgli>jHYaA{4-e~%e>L?Wb?Z|4=Z(`Sr$lfL7wm#>DW(4@d1trT6ZBupIFA78
z*HGl@ONl>zTJ6T0+t2U){1sTjT77KB1)5W4F$!a6RapdrW%ONb&a1L~TdN22Aj2~t
zj@CoCvc#lo&*R`1WETLeX4+h~4q{&y8%Jumc-}tdeF&HbUJFan{_6uJh)fh6M68@S
zP(EGjG9_(u+`ym}i>{|VdP*9V9Wv*_rVPyl#KM=}<qf2%`HE(J?{t)~W7Rm9JYV}*
zT1<Xn*Wq=sJO3e9O6mcxk(u}o^rRWjNl-BS=Xtb}S0zuN?b{NBWlbaT)MLTnz(`3_
z+8c=Gh?H+o#40`6qtaKs;H~W5M-;L8polfG7Wg1ZVxn|OyD@zADGl+u&jffW#<>8C
z(+cDS?vISe-TN*fQ9z)6f9{v-Qu;gk5X6F@o`61e>8N)qak?P<yHp<`XtL64+A-Cg
z8b95PSz2}=s%jYRw)EO90n!UXw+Ss1xh1F6MLbIb=FdPnOqnn&mC>+LG>ej)YQr`r
zX$WFzP%o82DfxFnte}ON<OSM=CKPv0HFP{=8s1C^_*7r%T&xSN1b-iNa+247gs}c6
zT2ZA!EtkLm`BHw$ZKdlU((L*@V}TH-5y0ube@~%YLInRb4T4xUs-+?`Fw>744^Ui7
z(?H4rl&WFJgq3|5Th8~ttTE>Qq&*c_$Ym7LnK#f<q(VY>qsq97@tH2lH?L4|!e*nW
z@M9zEHJ~lRL-CX>>0l*+8&p?Z_mJiZ&YegFyv;~=jzH>vS4R6^zJ~wPb<X}bb->^-
zlGp!NEgJvjROn1J?Qna2M({mo@=>ld#4eN3AqrQ`&CO5e;m5DQ^_W5d>)RLPxi*O8
zk31M^&CiuDgPBMF<)39skujwqc1l=9&k=d=>H)n5VG1SM13B3nFbq&xN!WZ9ASGZr
zrMin0%Is@26##YO1`!-a8kyHQ+Lg;_j8s`Dt}XyJ`#`S^FIYk<<xD;+C@E)hQ36mu
zpo2WHw(=$|X{H=mh#vV#_&jitC#auER6ybIcb|@m$>3Z2Fc|jl-GV5RKtKWhYY3MT
z)PUG?d=yuKr-&*bn<QVNg1`~OmQR$ZP)djd)07L*%>m}%zsBS?6GduWg5pEyWH-YJ
z_^(<1a1FZJTo)etCguX+wL>*%;5&tKikqaAIH+ZZeYRe7r6sB_BkYKq{6a%@KOP}T
zby8B2f8kP!ipjb2=jp*)zjq5dF5_tO<Q}DP!Sz5h3uO=>(Yg0-X;m$!aJ)>5Q^1W8
zU{}#?iM%iFMM~8qF9%epI+QfSo}Z;qp8JA~nF@H%WYhcW%b#`<2`naCpg>rJl-fA3
z+ql5t^nl*fDQ*+(25}TXfqW<#?q0lAKG2xr5LLt+ZVA_fvX{R}#StJwWpD1&Nhy%e
zDpdU$>QUw#AwrR=(+Ff9N$UOU6w&u`JpgM(0!p>8-fDWvQ>~cekC7(Tl2u!|RAmSC
zE#N*O1=p&61;GPF%MBDMTT!IPSO?$mz0f7Y`~Bs_DPGj?(Y&jmMlfbnK}xr8q_{o@
z-p&JxsX(8$`Rq~wZ*ue35o8_$%~;Xq?(stuSFV6PmZnh40vxPy@;!x%tT_X|;7Eym
zMMcFBGWF@@SnmL=@mp}3OD0JTNmP_M&9$_{g&hKh+#y_b$RiLNX3e=Rp~Neq>$?p<
z=$g$PTpNFk9Qm8h6R#o&4_Lv@?$Sz%3TH{2vgB;Tmy;L<ZAG_&V+py@W4?JghtSqU
zQuRRl-4x2{5<;&+qtZD?NOWf9-C80qg$K;>2&g63E7}WVEH}}OuIS8p)XOeuiTpw8
zhkz<Um5FZp9J7@o#m{B^;|#XC59C2<X;yIDAqyaUH$R6&&!{--3&TO3;x4&TZ7##|
zw3_<djlt9dZof+G?>lDRu$Cef@aV~t7JwGZwvxxOW}}MXD5Wq4OQmuT0#TUlyz_`B
zSLQbCH;v9y&yh0X;3uY>rF4>3JU#=r#9W~E2w@U9Z;>S;qn1I~UnXJt=C6nJkV&B2
ztw-VjBayY1^ppq+Cb~goEFO<sWCcCz#>q&WJYbs|SH&r*)VBwPhLV|-S!*BTOI2i#
ztU65)8=!u0xrO9Q<rSL~33f?%+|2B2C{m}bfYFk|Qbx=5<s3@q+&^@b`rQ0e4W<dY
z4b&S{DN;qNJU<|9*~!Ou$2<uDENBA)f?PP|+js2(0nw~#CqNJ)&iv_9e9}W#({5mE
z3L|m319#VWJU_Qc^3g{SMhV!li@cK4q9~MjH&m9u-#LSCs4D`F%P4LN>qK7vX=$bW
zzg9VWH&0sjJ2nzkuZ8zZDBqOH?xiRYn~$Ahlvs|W%JP^=F%5;1KxFzTT-0%S<tJ>G
z3&X6@u107pa1RH9^3dNXu?tTRAVGXE%;D1lB9lic!dB47cQ8(p78jgD0UkCy&fVlh
zON|262rn?FP*T_-#z3_4{QQVrI_F5_q>%P?6Ai#qTsV}JryB_m0bk(Pr;U$+rwOo!
zoS@1hvXfnn3iY3x5GZM%nPQr?F9ey#z#7CrsdI1yV6dElLquPo9dr^1T|I{FmC=fo
z2-{XmC)l8_oF<uv?{P;giN74$w=E1N6}D(&(;L(#kQ3!af3YfDCs&S15q#6l$WEc<
zzS;?>=iZDDXz3EXj4t!%t}v6!W4t@Ae!GI}<#+@g5;Ur8&-epG6t%ZFqb5;NQPM}Z
zUP1=xH*wTL`Z>#N<DoA^8jm1$gtmd8i(O-EtQ6OuQXyD$uJ!hhfJ<f&6cXT-HVzY-
zH8r}bP$4JNffan(;Q9#q4HTnjf)%n~;1&3?X8c8NFJ$)?L^m|GcnfJ1$hbl3Z-V*Y
zkBg>0V6&}NMH(-Eeh$n#$T4s}7O{^+QH1xcze6@XJw@s_5xb-;+YFXN^Ta1BA<KH(
zRowYiy%=-H2&TvSZm|OtN}VTp_zY*zx=6V9k5uTG|KS2OfhZ#iXI=0ekS9^azB_0Q
zN;7Dff}>M#^pEl-j3v+~NoFI#y`W2`344yT455>c{rciH_@13rl_xW^0{(O?r#!t5
z&_d8$-~eLvUUSlt^xbX?DyE&I1e*6Z1A$v0V*wB6*32%0ye!U@ScJQz%~NIHMtaZB
zWRG>O)(MXJbW|wDpgsmB*0KqcUnb)u|L|x0L9v~%uRDbW&y(lLOtu||&n<efUTAZA
zn_&jfG}@`2=|X`kkqkS@xB$I5RDONi`Q0+MJ%Qp$CI2e6=bvzp;x(w-@_^fh{SnY)
z@V$TJc%oeTl?D;>StzUBI_UrsG~*gA!&~r*sA=M$SaM}QJ06M9MU2(LyP~i^rxyJa
z8N#ge$(T&l>yN6z1|}x1jf?wMXYsg*p?~SS(A_ZtNTHo3mB_@0L*e&|6V3sdulV!=
zct`im%D;S3)+=iT>(Nc_-|p>c{8Q~NQky^U4h$vjXDHiL!n7K10V;^+(<>GO@MV>8
z@}2*FLkkZdJ+eeoM2u<r-`jggniR`t2M9Bw@*<QPl=W672UCBe>kjnrTzXN}a?`O^
zKuAJ3l$Pu>;6`<3M1TcxB#Qf5JoO<h-q5^s=Zz!k2igILc4nMGt?Y>O|L_Ll-GFJu
z*_EU0lu39(gQUNwc^(^RGYZvLW>0JC*8!5@ib&TZPv3D>BL~g;ejqd^F4vY&<nyt^
zgXAVQW?DUY{CJwofge1hqOuzW3*2jW4|!!=uBW!!@$u1Yc0v7_`5^ZB>2gFt>(9L^
z?e(Wde*xpTi;qeRFCoNv?j$>~_vV^g0#&3(NinS^>w8dtI|&NB%6h3H&GX%)2LU6v
zjG9nSJOsszbF$ZfP601vYifl-AyuP{KbU4whK1BB*zI@vCD|J{sC#2U1hy9BnQs%M
zWfUoI$-NXXr;r989Fn%&zZqqfVH7>^F@fjYiIOBBG~WTe|M3&D28pcr*e!DYEt@>)
zw}u1+FgLvcJcme^SA788#Et0W47<+NEkTa9JN0POdK?8_Po&X}o6uYGhtQ}bD_~3m
zgMwI^TCps!gGCZ1sMr$z3V5%~d(h^7$O61`%!)#-oOfDY;4C>&w+2dKQ7H1OEm`ze
zc^oD0Kq<$M{Q`UhbmbQI15W$qDT}svbS7q9fZa<1hY~4HHo`@MwIRwy((n>hAZYmm
zH0%UEkPB1TWgJgEbz>qZPM8$Gb520P03R<);t0G{G|)zt15o`I+w1=%>i<=7hxUaF
z!*7;G@%>uiqwIK8KGJ=?k?U$VDF$Ia2LAPmJ}S{ct_4fn+%BeY@IGa0iE>!E(vx8w
ze<bZn#nq~gQV-egGTc*Jzh0i>=0l~U6Mu}))Z1P-@AR0i?aP_a3+Kb92e0%r{Fojr
z7sZB1lP5Np`r`23vxv<7ul=~JreVp`yi0&2Vf@1&uK{II-U6V)H8nLw-D^(ZuhFDR
z$-)wv%Y4b^n-+)-5OYz#K;4P$UJpY;1cGEQoP`8cAy`2cM>aoE&|ka~1W_`Z7vTu}
zKs$O-LLYzp&yW9JAhIUk-oc@?IKvc4T<}z%r`se__&FF2rCVNH03pFq_j80~+mg|x
zDP%OIjk2zb%K~c92M+Nad8)8P&u_U>DW28+32D>?27{-T#GRe`er$X{MKk)_wtL#*
zPq6&&4zfu)&!NE58;CONclZb!4e^v0=wgK(gt`>zxkEt&;VBy%8(3)s7d&y|1nhc*
z%e({Uo)%=@a3-X717<x!I<y-U7Z=AtQxP0>Qffkd2Z8L?4dtlcVk4>R=scf(m6CE5
z03?3NR*T|$QLS`;cSK6*1C`^dgI-T-d5SP2AO^KLdyEG~I{l@hLWqwq9<siuI}wZo
zE)Rwh{DPJS;lEzqotMpyo7-j6Ef*gXFC=%c&mw+_x6M$q_L!t8WF9A4f1}!vp4(p&
z_aNVXr8<aYa<h@3h8#Q;Hf8}iBH7SM#73K5o&%ZJF_iqXc{k?NDGqUSaBv(t)b!2(
zB^B=0yDsaUKvz9|?3jDU!b8+u!X;h8Y3`SP`c&DxV0fSrtAx5-c2ZJ#QYmzB2Rb^o
zX?GyPpt~{!Mo-@hm+@ZJ{?_={CoI%1D(m@-DQn?^Bg-d_nK}EVnbrw>usFitj>Ghr
z`R(4Y1G}sIFmyJ@_VNe0j~_l5qkx1Uh2DoQqiCd9dt-sWMd`v(O=tJbr2|%crZkKU
z3>?!N5IdhgD_Gr|!f?svhfb=I#{>t&6^j}=Dj(SF>96S(2oMk#*R%fvsnXV5du8(*
z*@xpfXy9Q*Q?}1MrPt8XI&iU)BWi(dGlZHD7HvDa6q;Q>zkfG2N_Sf&d8RMNWx*K)
zwC?93%_}89=xCimB$75Ws8f-|<1ta_!MEUk+Sy&_u9<Re&wl>ssv-;<DZaPKc;-j{
z@bFv9@3NjAap|eImitd^*f@V+xpc7AiRm-F<hiHn)638q?l>MLx?haqzY;L_*UN1{
zkm!;kf`g|}%k2wXgT6#QzE5uYr27*aZo?ItP^Y;m%j?&#tEu^V^+O4l#H!eZ>6cIC
z__Ig8Bm@~OoIpV7$=?{!Gl38y^-2>mD-dh!a=0;r1Ytp%j9y^MX}_aR&-h9u4FlLC
z1AUtq-_P-wKxox5pWfSE|1i^r4XA`&Ms(EZi#{FY(0zm86+qWjBjFUbCwL;mm{_6-
zCG5|%WKs!g+@O-|;`LaQ#dUjkcfi=&O6uq4vMO#V&~k+Z7Pmy%zwk@PCSfCYlh-hx
z%HiS-KcAjmO6re@5<WLZCnXJ_<UNH__aKum8@9@8sC|HH)tEew@*V*B#*od0KL%hP
zUL;f+?Yn|a#UwM$n>4n)R>j$glP^KRQie=6!VcV<EFmEQ{)H{7Q|OM8laaw8{@m1b
zXvdBuRCz%fe@iq(Ff{0Mh>3~Gs#U8HQCk_by9yf8sQd6Qtk$1iXUg}stmEe&x@zdx
z7B(Ptfi18-PMAVi&x%f32z~Irkz7q5h@=+Sp{2u#6Xl3LSK~!X>+9!Nq@2euAb_Zq
zYu(Qr5S7vAu}f$z4b4VaA3z-uPoC)O>0O6e0Yx3Md~ZQSK3}&$81CTeXsl>}%5%Q%
zI;{DSfB*9<P0^L%eC~@!sAD~BMl?@Rcam1>(OLIB5bOpniJ6sEWo5^$x*tA#7#+>k
z(s#xHrA?GK>(<@;OUP&235S?PlhLtWpI&p=;vw`Ac`Hzh`K^d0yR774`g#UIY<lbp
zEekMqfFe73sH&<W_Vn@bnfUc95d=@9;25B%36fQQ9q_i`kPt<89Yp9~f#r?U>ZV^L
zgig?tan*;pxC+b462Trt(!dUmBLv}Km9U$B44LO$IIW?9&fa8%w#vFjUz<I2pwM>5
z+q))F+y1APXQ~<5o(zlI^Y-y{UN_$mrg}9*^3HSD22nh(7=Uf7dI5ZliIFkN;Rf=0
z@I<~Cj4tV>kq#bHTs&kSxN}%vwzhHH_u>6Sw!s6o7R0kCHMzRF5)`@c&bkdJK16FI
zf{!B#oD?J+BEk-{i@w(ZRNWs22A+SAvXvju7H$j>@x42DXlQBCI{X}O0-GLA2%A^)
z)TwhCW*TfG*rvNM&J50L-MV#4mMnSm<_+oA0E~>|gB^cqNcUAqT&zYSY550l7GP;z
zl6DQe2i^nF3^n<3<T9V>O0>b?fR~BU{D$rc!9=B#Cr>IUq^7Oc$K0VgNCOe3^SX7a
z32NMc(9rpacKo+UqseIkHTFf4<e@`^D@9^#gZ!N2WD6}VE$lVu;Hc{8m>SU$Fa=YI
zuUxr806fi?1A~KG*RQVzMFokU3#tT@!)=)&N393HrAq4^z^X$vzQq^oh7?6TuLm{b
z0p^Bb*qo0jOM@}6GQZ%?oja)AD;~B`+z#1YctBM3MRXRBA>gx;|NKJkkv_EyTGT-X
zUOn2ffpzCMIql)!XQ8|-gy|PQ)sVU!f6ePA62cVp3c2)P9~m7CVn)6H3*?115epeO
zBje*XP97ImuW>mM9oC^dy=6A5ltK1snc?AXmq<a$qwizz$XnV{-hWUUJ@&IX`&E}q
zExIlS^ZViPF4+S?WiPQE4-cEi52vQCWxqO$O#t6(m0Oc207w-F$@lj)M@QU}=Zfc!
z$l>-}M&49A;sj<-REAEzC(rEcY^arKvd-($5urjPDXTv`xByMxF{HJ3y}j?VabKp-
z(s_Z$Xpcp1X>MM#tOkb&|CQFA{h+MWj|}i4GVq%ZVbblpOPST_U;jp1Us9(1=yuue
zJIUYUPbUUey;vjOV<0FX@H{E$goX5KHtV+zzs{?he*s_YJIoASE#sCgf~^9gqUyiw
z1Wv_sXp>eym+5Yn4|3l@_=Y~AN5_cUty=)oo~tF?W7E|IW<JkJ{~ag3b1%2tVNlFF
z9}Fvat4NQ1ErTA!N1gy2Q@zQP3uHn^zT@fZva3I`StncmsMu<$?$BAdDikUUt3#ON
zGI&ixn2s+#G1DMZ<=U@%`8%Ird}CS8<b@Ty^e2E}N3}fRS$#GKMXF1gO7ijy$?Zw0
zOFyvh)Z8@hKb!iXk@@y+(;U`7wRx7DvF)dFPX$S@lHT}Q=f(cadtb4+)?7iTyum7R
zk0WHSV4Jz3>d2h5Zo4?$I@oxOGL!yrnzUQJJoBu}IeMsR&yIr!8%BC?2FuG&Fm5NB
z_lW53z1b=#x%YmNH;dD?Ya8kKxwyK(aq!*SJJBKgfX!^AW_Lx%o}4Lz@7M}ax(X9V
ziuSv3b948ZUDRMtPf4jvDqiMr?Hbqz(l6CH>zCYs1m7CXS@cbo$i%B908hN&=#KXH
zY8!{GQfcfQ4n81r>xt-;ylH1-j{Jzh9Wj}0D877)9-f8B7HnVt*F2*l+Sj$4b_Rz{
z+`V_Nw&^NByRV+6Vp!PXbh5Dhi!p_%=4IttjXhY{!>caYw5%*%zH+6c<M^kf&$h`o
z0|q|vZ@XVHB6roJLphs$Ha<YYUA3k=#b?p+HrC~^o@6hG->DQOny|X`t-HJXt(9vX
zot(f3Y=X=_$kNDeh&~8*O5z|ZwATjS-CQc0=!iyVAZxutLx%z9Kqh#0a+rmQiBQzB
z{TrWS<_#1eE)`E2uk7pVL(*NHH!(V@ATJNzm=x7SDWGuyjuK+Z5a!q`DJiLL`l@b;
z1uP)^HdI6X0|H{y5<vMwDdG^K$B~h%pJ@@>lBawBqw$}(3Gj0tY!dCk<O?^BYYq-{
z05O3dz@rEW3BAD~MkteRFA8Bcw8Q-Z5Fs07^$b3}FJNOn*O3b78jbTD3k7>bLBY%9
zbwEG>dLbo+g{#Ju5cZ3s9R)aS@S75}^04gGId|;|O!n~#zj%#A(t4==(zFUA3eBL0
zVQhZiKdf3vQnIb^=n;mloEsU>=4h55$pSKsXyi97=)EW++M#m}Xa;5(M#CYrwzd+g
z9r#yAkGf5T`T9M4_%zfSqO)!A6G>~duFpH82SMZOr471!`7_Lj;Y(aRMnd)BG_)7h
z^vYm;UnfUJMcue@<8$L2NJ-TO&Gx`vnw#}>S6tW*vzMTx`XORWY(jN_YWMao+P$ED
ztE{UoAA}D&`%4Ph?9Jhz*|u$)ul=3g!^KYla>t(P9vmd>QZG%?qWijKhlRy3atg2d
z)ROLvou6L3eN0A4ggkF<X#wADWp*7{+>GnjLz!fk!MP7h!vg~9z~@R&-;=!?4-eeN
zXD{SL&Terw#cZ3et@}06$eF!b2*jV4`MWMbg|(BJ$!c#-S9vIK@cb$|ZYO=xGCO4a
zdJlYdbZ%r{kQHg}W=X#E4CFB{LH|6T569pxk#;#Uqx+~Xtq~kD!_qio?)=Nk_I$-m
zTZ3PKgZ%DUqt>f@&Hbn3{<k_{C08C^eNqh)C)iJyvwGm~dd=5w*bw`N{Z)^e4>;+1
z79ZZdlPWHEj><?^u#|Fo7ORjNLXD$(tp>ds@AAmx)$n$h+O!<JDmmT7xZ3m@pw`!W
z;kD!}5?+0zU*nRlXELQeRD4*GtsuV7N=u0vTU2)EuvTm>xOjE17{}+&pUsdEB3SB_
zf!nJ&xdl{0J9TF_VD%ZT#vXey5h2-(2F{`AzMJo*2<!D*xBfwkblD3Hs6-P9T}-gL
zO5z9Pt`-exMejs~{`%{$edf*agAHI@vE2T?E@_s){76{%XxxH=>XD3Lnu$=hH-xD!
zvJ1RMAb~V^<NZ>8@bR|`p^|;Gh#4XX3Kl0z5JE%9=v}iTntu<w%wM;3fMZ7HWQU4m
zfTAs~9ASri#R{PA%WBx4Z>K3h6o7+;s0MA>uOZ^?t;r`Kpm_4+*PSKh<zag-2VVP(
zl0NcaA$ZKJtOJ-hjz+Fd?<zo)=YM&!c;y4o>ny;X^=^HPnKKBt7Xp&rx7LU9N#ZcU
zFc*4wd3oJVegK`Rizq}xaBoNE(evkHhpu}J4Gp1j020r>dm~NnSTt9B`SKR}I=}aZ
z7Gx;hebLBxlc#8dze|CD<k6tZ%F8b7=NIBVbVzqr$Ers!3RQ;08&5Es_EzwKYsZ5!
zI+xP6Z*r;98*d7E?&PD!^*MS$A8cp0L|3WKV=i<yfL5#HufI&FCY)l~Ww1x@^tEeQ
zH&+#5ph&_Wp);?}-&c$gk<|K8Ik8bdBkAG|vwh<2*tvg@<V88Ws6Wo|sz<%7yc{z1
zPkMf1-PTA7_ke?1-OQ$?rj`c&q_IW8LGW%_^4B9whcD35GlYINWQ?Ec`aXFgVbLmX
z-LIW;_U4p7_Co2S?Q&$42~HEvCkXA%po^lY%Gh?Av^&D^gk~i}cniFcJt892h#LXU
zqDR^VOidnxLBzDxf2@)7z`9rK#(_PB2xJSuYm`;;^YTp5N{%OS10UEa7YncUp>Sje
z&vg7aGLYh;BB<7-$H(^~xBv(Nx)#gAmW$tVo#KeuL!<qxD<2R*Lp@Pza)ryz*E$ja
zY!TF$9>=|$JzHtMp(HV;i~i~7h75(&(@va~(cj$ub<ElakgfEd{9rVt5MYI%<aM&G
zq(=AIDPvk0JfZIY`zJ)ldZ6;1T8l3K{qXeB(NSdSR>xafT0GZmdgh?7qH-04KkEXU
z*vd+k&)#rXpakn^htjRhTGCi1hKKtL>C`5znh<3(BnXXXf{Bg)u<bO)wj+`ZsqY+)
zgi6(MC-#TU7Pa&o^Vg-19_9MsD5?+hJ#j|f7=}Ej{=(x<cu7VU5NKp31$N%Y(C)(G
zVsA~<TjKZ+nL^!X>do;ImqygW`{t*?$Glrl8sdwhl3~XBTCXKqw!d5AWFHR6ww=zW
zyg6u)g(GHw75zJ-qB~K3;p^+&8&sU#-8jI?|I%r8TA&W`X%S1vPOTIv`@${E%qF8R
zUK{xu%X_vk82{oa;u-Xk@-te!m>`uQXnqyS6r`@A^7Mw8_C5`J!R$@8PrN%G@H6e5
z;}0@sFQPYGmcyFyl$rqcTIti*4<i4Qo94CW4Ut8LlAe6uSt1Vr`fbfSK0aUH*u9Gy
zxzY`3%fy}{G}0Oy8KJ;xq$FtZVX3*11Pc5szNYu6*G$kL&4w=c`8@(=7JV>YvYK|Z
zEu-PLyIMTXCt#{HDhAIoSqAU<#b(*|2M$RePv<%mV%5!ac)JODaT|*TFL~u=7ep*b
zLgESdR&?f3U`<0=@<BtA0Uxm|xm~(4x7piILYg#mYR3+Rj1WdCpRw&s%Z|(qN7f;i
z17z3r;|JoRUL;uP|6xeB#$E@9sy8qU=^8nS^L;>K>;`oPMZ1%D9cXp!G}D$}X^qxu
z3#UQ$@obJ!V|rbVo)hPRTm{R6ISTQxS2`-UhEnO)tVus|BIDyoG1J}6^ZsS24@5k7
z|D{jAMU@(}+GGoP_D}4WkBXXGy=s+k%Bjf#=C=N5xv`S<IX1i(VH5Uj6O$pdbI_mT
z9sYnYjr@y@X&~CopsK^Je!gLv{epf9&|9o>1kvV0XrY;)R@Btgq~tlIr*nHY>g#n=
zKpKfXG8lE^LCY-e@?-SSWZR`@z<rYU^b7vV&^y9Vn8}LRPWQ|pCNzwn=he%xN$67#
zRhS=O-LQdY@-0`xAg<>6+*E4S4LoZ_J-%t1N1dHUgO^9Gzbrq3u^#q*!VY8SMrY`Q
z`(I$10z#rmC`4TO1)<8OeQd(ra(woh&iGXzEA>DkU0o?>&a}nH$HN0C9!{5=eADyn
z*)xRDEx*j6T`jwX`_;vvxUa_J{pA`1Rnb}vWTi8ru+km>eWjtRUiBtpKvAmY1?$yP
zEg0*6Hmk|vW{n!zWpiW{zwj?}tq>mJTWk8dgZbV`$M{2UBa2?C9}M^OEQU;3VpDKW
zcefdWRQ!k2MN3cb<Ly0*b7=ea0bR<xYg}9$v7j)CM8z0r4l1q)ap-I=6b142reQM=
zt~>lyb>h%q{#{2~9XlD{rFSH^GD>QE5Emz^X9>C(6DM-1nu~G&Tk9V_S(Wcxqmi?B
z7Nn(#wD2o-yL)(et-fzpXf{^6t0eQ1HGmD^JXsSNXGpWb7D?y4tmY7Y9;LVunr@wY
zsA=`tx*#`1&xgcl2&{C?jVFPDv40R_g(JVoxzG8>MMg%Z^XO%vKw)o*5GKJU;`d^i
zoC|16eJZ;f8Q(W3-7Tv!)7lZ5Bt02o^dUj(g5i|0qS^mtuZF6PvE<zL-Ttr5y7WAQ
z<W=bhRmbv-Jlx%Zc0;}hJrqNZE&c@s1xVjv-S%*D9<}<pzX=nVP9U&?*t6jb4^Xld
zfjYku9y{^h<+l|Jx<-3;tr?0Oi@E^fy1yOqDKDO|sTw0b=2Z?)SLoiy<uDt3GqnB6
zr;%ZaeYFUIk%VAi6lwkgCR8tefN%Hi6cl;UchCsFARY?WLMA|@Bklm#19t@FozYF(
z&C4s-#MifXIm~N7-iX{VkK%nCqsy9=hlh_RrKG6nUG!TS`?%nw(%r;#-oKRYzEalm
zkRzcT)#n_Ok-2n6cXxz>rR6Blr#;D7bDPWR^tUC2_ERn6ZE~b@SXX2(ZnLxp;S>%u
zVo9_d>?^dfNPcyGLQs9^XPPJcETDlG@Q;|dffNc>;;>Ec!wK^m#9iQ++(7mXRftq|
z3cv*1=hyk>7WCPGhL#7hGTaaSs#T-Ef7i{oIwR$092Ov3t~GHB8{5nLlByW5h&wkR
zc!$j-$X*pS3fdWd<>YLf&WrE^me2UNC>jJAwpXtzS<ML~WWYW-Oi75rdgQ<_BnTqA
z55`Vu%f5gA9${UYwTjELa+>bg{Q+TV8fhxl*?~$G{OV{x_gY<aA?L}rWQEi$iFFs%
z2BaVJCJ8N<{ntt%2Gy9I0k;C+@pz)|@*@YXPo+%UZA75gcXFMv(Cq!K8|Sl{N8<E+
zZulQ`U)O8(&GWCfmB0CRA0@emxy!A@swJZGhb52I!&<*iF3KFdI4dRV@?(GZOQBiZ
zI+CiEMG)B;cZz{pNaRd6B-H(c*r8&pt1%<|Vq-2T4_hLbgh2Gi!op+6jxAfU;%Dot
z^D|1s6vD9A=3!ZHcMZn~$;~l^R9CB&4DVizi{B?CbTX3*Ab`KN2)Bi59^RzKerk;N
z5h}`!!&YlbXppf*?<+=olX=`oc>u_A;o&beza^;iMg{FUc#m;Az^3yTFUDtylbo*r
zw$I5?Y)E8ziNeAD+Jx`H`Az?<9jgzIjL*NEE;Kqw65F4bmxt^l__c$P(X$C#z%GS?
ze1-$>MvA*{OLkigM3ry5Cl&Oequ1(_<v*@t>3Ooq9n5O#>aUZ}T<ch#AJfSG=icF4
z-AF%r<SwwF0wpXg+~z!MhT7L&ZY06`99eh7-Z+t+nK_A`<u&8HCnS$yVZb=3$^#`3
z<8Jp^K2V$>9p|fylr1v9;gfzlzuza`?IU~fVlTR8b_yvgD*EcVz$|qZpg_<N&9D;#
zFqQIGGOTh-aaov05Vg5w0{J&l=fI*2ForY`eUkO-3%jw)t*`I)4P!GzKvm>{*00^E
zDs#z3nw`HgA<zZX)2AU<Ka{DA8m6a)ITvj1Zl=*slvo$m5}n1_;u4Kh7~-t(8`~Q}
zf0Dx>pa#1Etx++>jfZy6t<y7`QWq*djBZyNn}e+0LY{^%>$<PD%&)j)^KK31aI6q3
zHYH<3>O$%PuXYw+y=<YHjbzKtE<;ZO$PKt)`&qq)JXZN+FSkM|tg${vSvqa!;NInM
zUm|nY!Z6AxSSxOCgwB;KNfY+p;q4x0$#D~6&6IN!%8X6V=fT2UgQ}w&-Rq#a-Via{
z=Ti|oxw&Q1mOi&<_WDea_T=cSEC}wrU(rK*9px6_Wuz<shYm)sZr;&LmlCe|WPkc3
z?|mx|$OOcmtq`YXGox1yak?EeOY+C!<M{^PmAvLmmwSjQC`>E}N!5mr9z<A;j9@fm
zVqUX0>WQXiKdK=?x$HGP#icKm?iP`#?^!M5#HxFhui*m9d6v6z+}pfYF_eEE9Bwr8
zx;)P6*kX|vng1>uo&@W}m>lI!fHyjp#Tdqj#4tX*J*2{sqd|48LJXWwSrKIPqZ@}k
zI|f~XF%!J60YrHJkgdIOwGuE__fL!ph^%7wT>95vh!qW*4H?bF?6)Z#ZY|_Ve{dco
z;C)wzyWoc}&-`A8?TAGh1kT&@8z?9~`7)#Z^K+g3TWumr&GM1HJl-oD{}Dm8VCzrV
z@~WyTB!lP%!t+CDiGupiswmH?v4?Nxx_31S-X1syhtVnL33BmUcnj!`K!DH&I1>Uy
zh(V$m>@Bt)DDJm!-==3j+<aUAdQ`3HZ{(tcW(c@phgM2EEDxXs*k)2##<RoU!Gr}}
zeM$ZcoLr1=IqI$hB!an3&jsu~oC^n_PAGy11pGu=h1j+pMVsWzkOvPQ4DB93E8)Iw
zjRvwvAE4lvYv%#Es!|!y)Kv3*5hjIB-hN?R&r_5(aqOu=>N;wzmprl~OW?IvtyvRx
z$o}O7QNf2_sq581GnP+=;1)2z9#=8=3)`KYit0GJi1PGMQJ(FK#ZPh=I_7Wld#+&U
z|2J&=a!bSupd#C|k^Eby_oLFgDPGC`>nrZ#fg4{cGAv_fk3jv;OB1IfBEud?(2dbf
znNcxxfz8j)huC7oAW$vaQ8Bo(+Wgw14DNv0yfkYnp7PQZfX_ykcSOCgsOW`Wc3OPA
zy!r9~#X1Z|9ArZYEjrWf1$~xzgatY?eW{?y@Z*h0-$fW61VrfQ=!7IEC)+(C^|lqQ
z`gp{Oii#*XXlMksK10-N`h62}&YOxSw#|Qp>}RT1lQYl`$EMpJ9-CSe_dCBpW1fmt
zH%b7Sd1*o=F6(R9d$wnws%q(xnwr|V7lg~A%Ldt2GI{55oLfzyFpyBR?hX3eJO3zF
zFq+QD@t^P9<KqZ)gj+LmY&#Acid=x*8+#geK)x~Mti9~)Q!AXymSdtjV`F1WUsyaa
zHaX(g={j*t$#3kC%##%5v+kz#m#F34oN}>>eLo6sdVK`Sn11N@*x1Yu$pvjSHO}nw
zCr|pcUPw3mup-;^$zJmO>|QS-v&`ZInVr&~o+s67(|RCCHEk)|TFzXsg}W$!r=fSB
zJ~$GYHEPvDLPMY<h{GOYcO!^^uwoqaTH=6FG!~~|vcWWneCsUO7#Mb=^Wt+qo@XWs
zCBnj^C_>@%fQ{LJvIBetaN9F_daUg176*JmKEKEweHiBk$I|4&1>NgT-%+%=O}|wK
zkuoBb>*&}8v;2+Iq&cz-Rn;oQ#-vpq5k6u=Qw``I`TdFg{IexDGs9v(q8y}Z0~iYd
z!+<XZ4<1tdq;vQTMybusU9WLT&m2>~TpRyrVu?yoQ4t|(ojO$uT@D;o_?JJ`)zvt1
zXnuh-E>Oxy<jDYep^BZ^n5rzgzYGTR+2qJW>_)x3YgM;%gTcY<2vy_8C@$&iWszRF
zxbfuWTSEHLGy8QbZ_rxQZwb2iDr$Ej|2iy%J(P%gS~U@=6+_IuiJwaW!D-!PGs4ql
zIdi@0k?^-%h14Vx2Zg!w$dsSP!Q7;D*IV(#iGk`E?;cwPg%UK2ac<v^E@5y7OAst{
zc6P!s4mD-&!WE;e0k%L@Qs2-J(834|HdB77!+j4C63xqfbNtEwWZWl@21%EnrdFh`
z9=pxnUFGMn*HKpX_BjjbdQ21pS)khv1}Ps~a?4A@>S%?WNpVF3pGb{Fy39|VtJCQ!
zD|9ZNLg)zv()}OgN=lc$fzW@HA>n#temOT~YSS;d_qEO+h%#MDF5B%fWwfMQ@BT}T
z#3J=6$aL3?@SN8##i!vmF11g6`s0Vf{`{bYi|C&ND2PDLEg#jTw0)sxit#3Z4k(Tk
zNrrD}@9stey9M@&_hsv$)9Vck48XG%)muD5I`G5z36E?OGFCkAGP`r5F|3_kWw5Dq
zNfvR&Cgy&f!Q=pO(zq3z{KmWc=+5p;UejNpO!}y1sA@(f*8M~!0k-_{)VkGf?Zd+s
z?P4}N<H-8;T6s~=%|$*tnyA>aZ8S$6sI(8A(&5E}^HO%@0Jt%|gpdm9Y_1?jN5_-F
zyA2`Ew}LH&jfp88l#T3_a@U-kzC1ZpUwGy(z`B0H!5h`TrU$^x#l*&vc6Ym$<42Dc
zpe8|=K1;@sWtfGw7SXKGT1Z&swj0P6ZmpEAIvsGhv9U4A!|GA8bc`RYX-B`yp&WRw
zNuP_0%l-E4kX#5#w!vFu!>s|J-K>9*ho=X%6x2pU5)WPE28@S#_;5?ErnqF37A_KF
z4qTwB2kUp+-90#K^@oMKx9h+|0#XJ2{Fsh91o$(P!(gV$L`}X_R0Akvd&Cqt*a+0*
zWR>SeMn;f=&qEj0um5gC30ll?>JxN35l0yn(pdMW&TBRNv$>ZYx$41~aUCq7k%?m|
z`hH`DW8>q12l{~Vp~gU$+}F_|j6x5>pUqkH`el8_D;E=^;(s=<8;AeoBF|_<$TM+r
z|B27iODX|#euR0#s$gBOp~{StijpZ=R%DRS__&5J@%~pL?FlLr3Z<slJ;(|3tbd$5
zd-f~>&+na`$SasPY&dVk<?QT?$RFt)(c!^yMNw#aq(i#Q{E9eCRze#**C3(<cyE-E
zCx8EbfpO<R*>U~ov8hBMaJ4|@JM^K~<1Vl=u}Es;;v^AQBRxmZ_Y*T`(7v*h=EID#
z5`#OQ5qjxnv6e8NZSaRZQq0WE%U7&G@PG$9j)V%sJmU%&dp844LP;C&NPQp?B|;z#
zRzT7VyN(|}CJ}9g*lGuAKoAj**Bb3Wu$0JcF%5D%I=`^KH*ek~a}v=ch4Lkg$CZsQ
zvW=)Om_JnZZBaRbq9AHWU?zP5Qv#RU(a`~mOr(~PK8*Lw)NXf{%I2T}!6I;PDgo+2
z9|K6;jL09sCP*VLP!|kPJdJ^o_-&NkcJ14jfvOG<Jq_qLPuyzpY^$(#%8<B#vITN3
zJgJN3+mf+_7cX9fpT-42(ZhUZt<yfNek}6y-arESqyD>MUusQin!TTq5&?7h`rL1b
zY_C<ba^lpgjh5OWCN=|v5{Von1cm{PptcbH<jJ4*_e(!Zc0t_j7v^d16%r!y2C?;h
z<0J%KnKC#Mm*;bX=`hJ2|JhevLY`%l`&sQ?i<fTVZa&nvF}5vvWjEhsL`Z;NPW-~$
zf$jGZS2mrjI$9yQFs+=gj6jdRg8rv6xCf0gZt0$g^M2-!!sBgeEp#qMJ!h4_ix5Mk
zuXbg0@A_i^BJ@lBtlAz4xrB#3xHo_MDBnWAFAZ~9>HKXv;e^!`@oYSg=J^4{IV-+G
z3@y5KxC`13pT2y_v<{x~>*_k!BKhaf9|!{VB%S5r2g3kgnvH$Ue&DK9W;9%Lr$GYS
z>{h`LCT^AaY$^A<Fn?)|zfJQEJpW*{-^>=p$hqX>oTqroZtuQ(6a>)lW2h5Cwj<S?
z6iZQ#f5QBQQ9eeEbZcau!*PO}w$%_f0!!@bBaT5s;Q5Z+XbBa~^U{C7taP`Ae?gDq
zDEBK69x269ZzGqLF63|>EfQF{$W?Nm!R(#b$u(ifv<{<StUW4z31M-PQ#wSGA<3h~
z`*C|LFnyL))jicUHK^3u97l+7Zv2tl#9<ViD`MuXZ(@Pcc$QPb+~#lN>#$$}3D?H~
zb}oI#So-0^cX;|Ba|K8c!>NZcfAKp{d{{kBG6H$UTz-S;b+kh3qdy~_evzJjYW4z<
z%feTJcN=E4s-h1BjeS|pd1^;l2ovQh?aY>wan$9B9jj`)T5+p=3FWI6C@BcIUnIxj
z?3*SOyS-mxb4VH8tTzx3J=u7V3xH&n{KI{daC>BqV~=^atFeq<=dn^rmOCn`J1(l-
zMgn*46T__gQi^P4+f|`%7yb2-BHH`hRX2mqMLF>Pf#iJe0K+{3xqB~5U|>xgQ<{$0
zweBy93O#jy7)qY_33QA4a@|qg>$;9K*Us1jWvQTF$<v`u=*2sv<Qlt29xZV-2AAsZ
zH%o1}gtn5!*SSak`7v3E|E+5h824(al!G-WC+L2tMgJgP*Erz<jt7eFMt6YkDS?3H
zx{qW%?~^7Zl>IAw$b)^yc>m@qbf#_Nv8A>U{6D|Uz)+nRc6Mv7DiD2h#B{Hs@Trl&
z4)ZU3VQcY4O?7H!MobXxa1?(@c(Rhx((3sJM16>R4hpliud2cU1xSz`LuG=W_)Ri+
zj2ExmnT-hM?AfQUI&o!E(~O%oJtokw=)Qe+NJ&rLTj}~PS_q#?eQlKb52c*C5dY!R
zFK&qce>W3!_YII`@b82>$8@8+H<`iJ99Klg_s=0ACw(GC6zuPRF0`GxSSN`)rnE1B
zimCNNV*!+mSPS1rj||UVLIL=-Fx4=G-KX|%%txS^8i)7F=}=EkIV7=Wr?-&Lc&ANm
zHJLL6c{=0ybHpYH!!l2sQ9u8}w)<#^8u0|L6w`zYc}_Qs6cF(TiYp`4HB2#vM!yBq
z$LaS^lN%D$Uy0jJ{S{smyUoqbH8iRb^F4ja5ehv6ysion_eF70LrjdmZEfvUqB-Q}
zQrSnyok&FlD2%*&_YRZ{T{Pt5_4hB#KoYyXukX@m55iP%d0!c#w-zs6PV#L}cOjYK
z;rZD4`7zWG@G51be$DzFCp~l@yHQ;mA05@Nt=gbmE&{ICWgd6x>ip}+l05#K^vUl&
z4Wr&P$7eV9YX~IKJ$UKCS@I8*3hHW0>A;BHa&s$h<T`w~71WuNkW9cU%#HNvxb8t^
z5y9u<*Z<tvv^4e30<|xVB4vP{aG~@2fB%IfRIEr}L}36(0Y>#_ssDCzaXA7PAC;1J
z4H!Cz(I<SU_tJiiYH`;HTN(sm6)|GBzIE|4Sszd@VjGZ|J=~HNA7sc)KFy=9(%Czq
zp`lew+4xLA{;Z5>UB=^oPksCw4aXZI&r5hQk)O5eEdHzt6)#}P8?|o!y)vxSXL~m6
z=CN!LazSWk-L!`KA&(Y+VgI#}v-|h&FOfmmLq|(%g+_kz@}w^4S-or5t|4OQViAs7
zO@8cevFJBA9}BHrv*w<+H-LHabBgEao)o|s$B_uwYVqL*il7aF(r2Kd5v#iq%zj9M
z9Yo7MCCmumHA3}!RM%y>CZ<YML<E9p)}9`4k?mGOdYBqlu$;6uxk?v({OF5tMqD+x
zmbypCH~q)vl2!OG+rSlGLE15cI-qreu1+#63lue!XbcSvQ9L<(_%Ldi=pDu&EpcE7
zNWGEKTThRfTv*v(fJljP>(&;Of2gszW-=1YMM66;aYAH3C7raJVh16!L?$nGhzx&6
zydp==>-VG5(x__BVsI-NjMy22bQmvuGe%~(#dZ;qK>+<|t{-mB{{Br~Kj6UwW%Q;P
zU{(U^f?uDXK5q4-o%%x5mTulM1A+GvOz}iU4^oL0^4_T5l3=y9wVkn21!@UQ3NPFr
zkSPeAfxeWCQ-m}EMFJ%0InmLD&nKZ~0F!8I>Hzs&8#~152Od9o@C8@^dN$~;8<&!?
zZ@_Mz{Grza6Uc439J~VNquTd>{e1Bcluc8T=8+*|O=6Mu(I23pP_pg<@z9wlqJ76i
zI(Q_$8yOov*}9Ih*BT9o9Erf2fO|k4);A&|VtMH7%*@wB%`ZKwM>OT-KY@Y=PLNsL
z4ART*-@p6$`GL08_)KQ?p+V{idwPx>@P}W~lnyisV<SH`+<Zp^78yKDeS&)NODAY9
z6ai`gSn#u_=Y0KEs!8_0O>EeDNLvAKsX$lS!9xteg<rl{fDbj%>J7dGX-i8!Gu&5F
zV?q|BeHGO(-(VKh0J8AvCadG6OMR$zlgWBW3{O2jRZ?8cyJrsp2mqC#!OFzk{N?tj
zok5s!Ph#qc!4>cHcmBf#Xo7ujs;}P<<$N(QM^IN+gikc!2HM-tt}G;N3YczNy5tZw
zyW(3`ueOBpbeMf%N)E%xx&k4gN&1#he9iGd`xjm)V=0|DG5-7aRx}x*hlYLArb`QF
z%L0IK?}Bpyltvh+J2ewdp(>9Ta9J{N0`c=TU9?63TLvqnDoKL$GogSFPO+n-2H1sI
z80fE%K?wl!@psP5I#9Hg!w!rB&<%aPi>!>Rcc&?K$$T{M8bDfpx*@<3Sy<wNr0dmB
zok~Y<3tEBD#$m~0@EE09h}(`}5sA74APc9F_9NJjkZ;w|X#^-urmUjs_o~e+4~`62
z_IKdZK)uuwPN6UolK2)^clkqpa16wZQ2+-B@CRaw<5s;nAbajZ-ku%vweSo|Dt>;}
zz_$T&0p7mt>G>8245%Vj{a--9_wuCrhIW;I+vAOZi@M%yTFZK%y`#ghsb#N4>vaHV
zsD-U~^JuR##z0`3Zo2Wt1Q)}#pFkiSG@vn0p##3lPW9BOap(^m*Vuw?6AcZX)-Ir6
zqBCr$tX&wr{o_kSX8YAuZusCwp`n7C-rnr1Vi&*0k*xqaF5N&scntX<el>Rh)e11N
z!2Z<C1PGWtEJyiz83muAO)Ois^bg#j#Am~5xJqW_=B@>)pPcAEw4hH+jT-9?mZ2LE
z<_P8l#H0xRhG{6kf<QyJfVtE=oJaa#+Q=LML=7b`+P{DQg&Dx#+ZN_Cuc<mZ&Y*(C
zA%sgx$<2)%ItHpKrd3_CoJNbm2^#GJ7zY%wxU%E3vc9E>b%QdN0SyCz07a`l*a2y2
zX|r58>PGmB#B1$q@jO96@YhR+%+|eoYjg)c)znx6t^=n?IPBI2=n%}J&j_+nCPQpd
z&z=DnLu<}?o6j(okapGA6M8)L5}jx$Fw)iT13+OJsTh;iqc%j#EFd4)G_tgs%q_&w
zDsC1BM@Pc8>T!bn9EGwNX3bYZTO(TrIA=zO51CzyqMUYy8S+|SU_3#Jfxj^6Z`rbC
z!C7ftrx3zi$#$B2nWZP9`m&?1@8H>Y6w7I}Y$}EAH7JMK92>Jc!hqoEoc5F3m=d5l
zBsK}sFO&+%7gA(M(D9Ca{QS0BwZu9(AJ)`516V5w(^|gEREjjzzYP<w6>i3ST+Fh{
zKN_I;8_a!gO-Ru{fU!F#feIapJDpbE7i`uo>i|>}Ft@6M(C&rwQnGZGnqzJn>Nko)
zrs!0Wn1C%sA3qa`JLk5cXESXMm%8DM#EasVmK%8bR>uKUHD;J?VPzfo@q=a4ri0KB
zc=n95WH09W(yflQ!~APh0SchB!h#SIL^|{_XqkV;4c=J|3n3)5hKA;$lS*8d$4@-s
zmbWf%OSUa2XlZ>z`ztrgrUTEv`mN4_lY^I<Xri8esZ}H_8}UT=U1Fu%UAg?oo%{C#
za(f7k{{Djp@ILK89lh_~1%Sw-9FAJTg%UZvnVA`%O`;!@^vQT}A}WezHR$0#Q%gXJ
zUxXav6Eolf?h?wo+Tqr9dfK;d3|&>41`aZMZ&1s7tz)LqwHe5<uT#NKfYlIVGClHy
zCkJ$vX3p%AajJ>$1|0%s37dVpwt<Su*+l)qj~^8S7a&5hdczKeAnf<6C<UKC&HV!%
zeQ@q!29eFbL_ZW1aFucYFz!R?c&a1D*DoK4aFSkjb@bJRB-}b+c;Avg>FMbpm20WQ
z^>Oif)Q^A&*#DCHttW|Rca{7kp^%1pdq$&2*ExPSj*Y=*OiFyYeS<9aU;cVxton)q
zka_^93_%c7(H_$;(+_@&(+YFl&x}$H^w{K+RmRr;M%2*q=0;QW9b4>)AHb4vwmN0d
zTsjPx+u8X|wrxlevV)husd<e5J>t|{7<ljOMYGzKM1AO=X{OELkXnT{=hzBLNa*W<
zI4-pRaqc&84vs4-7IvGy7TdK;Sx?CRk*%R&2R^l^+fm_HWhD_o@z#9R-a$GKB&@pR
z!B{KuhH5UM1(JTrX0gwJ)S0to5Fe#awo-4tg0$}A=DE*d6;He!YND{SvzLsI!?B@!
zOm`*t(W7nJ256RMw7)Sw=a4M}U%sAYbJo0$2s3EJzb=G3$j)B4emyTY*LnYUsF|aq
zZy#GC5&3XHHtGB80W`?SuybF@-QU7)MhJo!(CtbN1%-0*81=gmjUP#C#-J%NvBsI1
zTeog;m)$q0+L7tjTq|g_BHUNNAs}Z7VgbmrM`z99z}3A846EW67EVQn_V`vJR1DYb
zJ3NY=N_yB&(Cag}!?P&_V%+<;tF70`ce#ozS~_@AD{rn}1wFo@dp`CW3v^@Ty~>Fg
zwUV78g0Ej}b2u^sZ@ITVMNU>0Ws_4DeT1#cdm-o-(Y8PH)?h3$8IRlour*ly2e;`8
zA%1|-^!0|nkt;4FUq!jfNj>fzs2joq`g)@cRX6VBR>A&V0s_8yFPp!7A+@6w`}~`J
zBg#rzQzC-;2lLxUc$Ar!YGkK=ne^!(k|R_vuiu!ra09_c=Ox2sN4{lQ8iNqf_m*92
z7&=cQV`IL0>8NgddQhL`)eY3|jZ8|~!OyQ?^(Zdx2LcWWaq*<)A{<ptZ78<rW-B3t
z-OkBb*bV9<`ldnyMzwe_5EhQ7a}2n19a2gZcRFQw4(rv{)QAcTTcUpJdmQGCG(#=b
z{Mw{vV>2EsY|O<H5Ew|dm*FGoZHIYK_wpD@c)38wtyJMa&;ey?#M1&#hyK3KHCGfB
z?{>|gDKA#<H2HtmQc2Ws0)m1rDDIx%c)WoObG>q9fHatChChCcCI<Q_Ta9=XL+~|G
z5oFl~o9zPsNSI@*W1prHl?n`c-C|*3Q4#|$`eXVTxs(wu4r1ja&Ye5ALnmF{yblCo
zI3U$z1Bs=3r>3TWqIcnR!fQSU_oG!U`;rt2Mi3U?C1ioX2@SFw*<RKcm=u(hl#U;-
z5nMx0kNH>$*SfKaYD2}VUSL^K`piUeuTut80iYvK;9`raLdx1OV`Q>TSxM;y+Ecq|
z5x~HQT5A0bON`kNn1rta>P5`}KtRCJvATp}6H$GFFU4$>A^cN0R1(n8jerzc0T{?S
z2w-p?A3kKrE=5%xgxasz2&2zXfQ3}}byz9X!TE)ReD2)A?2AgoW4zPBN&$-KGdJ>X
zi&EMFiv|9&qX+76j@&RGXdAnF8h?Q$cSb{Dp*NfOwMwXsIW3$AnzD(NRS#c`^oHp8
zW0q%Ugp>`DGq?Kni~68TH<5kTD~M_}rtpFwgwgd}nj_=mta~m#>{`<<i|#Zsh0Ds?
z6ao@>g{pl)K?Pgo(h<tOakx`NQMw{!$(I2t%ZgTjC+k}tx!O%7)u;%T;_><~1oF;v
zuvY(C+n8qJt=)eaxryxDEOnPZ%`;MigdP`y8gL^~^mLktpTk%r;ZcWBtCQj4{F4kP
z<w`$){)0I0-pljdp{GNT!eAD1`67r;Vf7^YwAzRDAo`VV^i^{^ojQz3yCmVsFoQNT
zEFA6X)n;3KP42~h0uhUJ`5ffg=md#^QRE7MoX*;W+mn@$a-e1y+NfP#L3j>%ZVhxy
zzwg0QM;oE9H+T?OVp2*40ru6;xrrtVL{DLdpLl<TNq%XNZ+g~)C}gp)9NNJ)dk{TC
zI9b>LNg9bywo(J$cag+bpOpuB1=a*SpoMR_zbO3AMVGYGiRu+x{j}|`C3JjchALH8
zgYR(MK?nF>Zkpni@^{XrmCD}NR;gv?*4C%v96ct!eYF7JQw~lBhJ(jsj@^ISJb7^U
z?%CoiVo&;BhKD~6?_1&g_N^mg&EHD(dw%~ePSlJO$7D8~Ddnrl$ZUQRt!>x{Fe5kC
z{y^aw<RyDlfJriS>~0^9(odXSJXh@1kc;EoyZ#=OShH>Z>v$%T9dN=h6+9cy*JZY;
zO>HYl;rZLwMNw2za1v0+o3npAtB+3<v+2ExO$lHz8gvvGP|$8k%z+>d(gdVJ4Vun4
zVio;VBN?~tyC6&gfD@D(Ll_%pgpZkU`GCKOLKrG}2q6(`<ltwR7>8-+Xdss1o;kFk
z0mWQVOd&_IBV$2=8Jy|klF>&GBH1AvC8%zqzYxDn{(~n)HRLmjcQAulkZ-^nV#Q=}
z3u*}`!nXK9mLHi=X7xRiiOXA~q(?f%)VD5YeT@nCn-1=ZgN29BIBqB|_34=q-URVE
zsstVW@E>XI|B19Vk@3XKRqDU6Y_iD)vrL4tfjbkIt)<BM0=_`)oFgY6n-5!B=+8Ci
ze%pbpiz6N@MhpEgnH^C?Mgh*hgl;*7PZiw)7}M}2stMa|&Qo$Ovi1XJE;b@B26Zv~
zCm)4lXRs<cD3FmQ!hCpf{GX>z?XeqeP9h5_pcJ3-v@U-v;biyN<@cxU52D5RIKq)G
z?;8`<&9>^sn#PLp@kxky*7_cN_eh=y?(>V*EupMB2x1c^&u(r|gY_YrlnHk|yTovi
zK|YeesV4&k0I{?|ga=bESMi3a0O>|R`5V(n%0<608M+Z!1i(NDoXx;x0-FSgku|FO
zrZDTssR@GyI3kGjF&oNZdL$s^2<}`SI}k%B^p?LxFGS1+sSsUwFeu6G#di|DYal3y
zO!22ASOk1JSklByFrGZHtxmY$kz&u)+T#nytudmG$ALnj4cu}C8Q~Epv$i)96+T6o
zm{K~D3L1|3vtmw5Of*rM%Dd8(v9qp&RTmV+pkd=Fh=mpnkil&q=f0gcF##eZT(J!r
z#O(P&jQX^MCX6>0FJ$HbF(V{Wn(wplK`0=VikOrZuKk_#s)97ABOlO<e~&mM^wFbr
z*6+|sUf3{-My!f@w9-!(&B}NS#nWEJL^6LXE>NDJUhKSK;RL2`Hrv!F;g$ruzsYkx
zmZs$nKiH+n<|EbTa_Qs(%`c2=^0wYyg-A<%{d(5zI_VeJO?&kt%6-25FRPtDeZPi7
z&S0eFI_X|~{$ZO5hAr)eMg9~haQb=kweR;mHlM@tO|RJ2jm&00bW9Ro=9mKd9CfMy
z%7n=qyZ(5|WB&0s%mQzKXC@PLP-k%d0`~jwzM2aV+w)pWTI)f)9D~b}p@ANu;k3>W
z!Ys~v*0l$)Z4m7@VJaJx2RD4PMvMxv5{JRp^@_Y1m8zEA612|C@9yG%ERA{dfz=Du
zVe{*5V*-V1Atq3)JBxS#^DFhwm0{?XF3x;`519;wHYYI$OVu4feKFhU9o#nKKm@R@
zy3?8x(CFvhd^&pk!R`UVtnZ5rMUr*=;v5`~3&v8(G)>~JhH<Iq$_RR!#KHaoa^6iq
zM&QlFo2?+nim=5BE^N!bE1xPV2-b{nZ$CzX0&_T=`~jWWfC#9<pfcM&tmD_FUi$fe
zyY!=~vYs~fuu8ov-Si=(BeU@*vQ097QN)dN6&pr2K@XPMGzj4~ad}|Po2>HiSDujz
z{sFv<%#JO&Pb!K;3cG02wwa}eaC$q}T<gk#ybq=&`%a39ivB_`{sOs`oc@>C@j!Ls
zu$iH2fi9!(F3?PL0UsJ5!<T*sTnU&@fjwb12cB^YThl18RiBtS6H+5tAdVRpPgDG>
zwzVih<GoFP8sRdCG7$#pRR)+MqI`IoYZKNPN7fNTjGi^24`vVWepCX7u*Yz2svfkG
z{wuou%^U?OMEM4&y_t;@G+$A{LWf#D;zBaG63GP^Yu7PT13-*!{3t;l5m06aADQT@
z;l&`NG!smXFK!>Gz!ivH|KW_V0^{8Gz`!7SC>qd1)e_+i86n;gh%SXuWI#@rmeI;(
zb&(OV(v9-eLmn&#i^X(PL@YWN+OYy>%1navA&HA_y<tq^>XbRQo|H21S*7)4@*N>a
z!z2PSA(}imoX~v&p)%@v3qV0pbW^PBnXXHAevK&Q`hE1tOp~EW7;Ome0abxqWDnRg
zbJ`c4?8UA}ii(*#`1oA_m)ix~fv%IF?vv-`)A1^iG)LGlFA}YNU-plolSbb?(udsB
zUA;bANXa35<-eMYQAEUwIV6K#I3L<J7aQGqD@xvqI?mZFdLicEpY7$_Va4vY^TEp*
z<W0qfnu}rb4}iiz3{NWMkJBT&O-o(~ks(azARrO!ULe$Ji2unfQ_Rg+Fo<7S79EF>
zizw^{88}`7Je>$+5f~FfH#F;xL0stu_^}XhI1^uWclvfRj51oj@ze=m4OJDKdPJ>l
zSOVN6Nvu-d>Zn!dn41s5M1VkLg{eVd&|`-%69j3#Li^ngBcSmSQ7PM)i3=buJ$bbH
z_^zE5!DOV-LF@CDvUm0&$HI|~gE_&rc?q%$ToR&6TbxNU)e6ImFz`)VxR?x<UAk7h
z5`iNOeIy_yvhS;+im|fj;P-2%0hW~oRVS=)BU*iNaG}sU2JeOhX&PrS0#*_Izf7V9
z1&;77QCmFewgf(Yys5z#q94$@5y|<1uTT_7bCF-5k&diHNfcrff3Dxhb%jH<7s2YS
z2~&(rgW0PFMDtg%3(g0G1GmnPLNWN#Bgq>G^GoYR;Kcr7zi78-6A0;3x9qSk&2~*@
zUIV9NPdp|V{Avb5alg4X>G;kPY0h)>+|Rj0JXKb4{XLn@BM8=z3j*<vx>tnadn*l)
z0X%qRJW8fJT~9*Vv&`11H`249g~ZTE0U)cotcZ-&11`gLpcaP81={#Uth_K7n81EV
zw3a(a&$s>+p7*HBXEH!qd;m1U?fcQ#e(>fi7DsT>zq4~T!rX6KSfWn0H(&^O4VCuX
zQ(^nbXaZEzfC5@S7BT6hsuKY%UUvUs;j0Z3ut*V>A2(SZCB#{5%JmM3ew>a+wXs^?
zhsTHu^^#PEFM}@(vnUR3W5q>pShxyn?#Mam1-s&a?8Wk9A56=*0p3?Odmq=F75pIs
z)45<Ryx2S7`RlNGraDe18{B~WQbR+-sHfsH6oHDX43W-@<Xk9Ue}d%nU)obQ7=i~;
zzU3Wy3*PdWjDR;S`*J@HdH~L)DQfyva$zDSiZ`RoTJsLhMC8JyadX!eB9_=8Hm|~G
zL!9w=BpDz((lPd&;+5NZSFm7~bh+k8KYxjtzbQ|<j`T2ytrziZu2v!gz&{0uv*xes
zQDmd#x6-+B7-xCsXZd!m8kFHaDeJ+}s&8HQ&}qC}79x#rVB9}=zm!FP@AY!+wN=R6
z8n7dtYOA6vg2-Y+uGkix?wlV(#DGk&L3@8LTNH4})_qsBx>)Z9lY@EY?Hgo0Q2zRK
z|4>9Hn(q5CW9t<#)2a9FzBXUOaR!mo6`%QI4Q?n3n`j<a9gih;rd*QAMWRr=Y9X`^
zzJc?e-|&W7FdL`Poh0O_6{*6<1m%%7*!4;rPc#EDeRKGDZxd=(L$HsX5u+ES7m1R<
z28_*-CKf^3Zi{~#qVsmpcYn*U=%-_F<&+!&Mq#UG(O5&I;XG)QKOxd%YT5Wok@=M+
z3Nzp(C^~Dc!;otgzHvxvc3n=uO^)*FV|*&w@YjI7jsT(~5bEFQz;9&OQh=IFt97U(
zt(~fZaX3wYs@8twjij`!k}DS|j*j^7!gs#%V=8Ac?C?(Z-o6s)w0}QVMSg!DE7h$M
zCgZsOipQeFd*Z^AA^MGRyBTl`gffabP`bb?3?lH=H%8vCTu%SH9}HH4!bA87q#l$S
zGEvb*&b<>LE0{bTkvte`2ln*(xf_b_a9L>3s>rK{Dcp#(8@N{rf?(vfaG?EAJ`Xuy
z*$KQ=9||7FWRHK{(h8#rf;Jf(2NR2Z0<DN22#?tiK7$L?V~a3&mpSB-|N729uMKPY
z!#uk?Fwde4<BD>7LN4QQWZ?N@RLT`#$I#{J7<Yulg`MxirLMmBd<x4KkR0<aTU8jE
zAiLVq?vcRcT;Lfk^d+N5Wf+YdjBVWne!x&ec#$VlLeR>&HjrA}ISEIgRK8pnBNCHy
zem*}ITknDVDjC>ZjP(P%^4EKDh993b=3~;yx@zcaJfqpi-T|KiAydEPRB%e*<4Gh<
zuDIspp%<uS0(eM<$USkPh!lI*o_dI<0nL-34t4;6W5`T0f|tm-Tq6}wBOiz8V5qQ9
z*`f%BGJ#fn3mGjw()9*`#}(L%k~KmKTx*xpb6X!Ek2~$3nbAha7ihR)M5~G(Cpe$m
z`Exz32lPy~{$k7g#G9%IF>$Xor{N!<=OLKWgfC*5f2M`MLq(c1M_FyC%*Q8hDs$-r
z)`KDz@pu$u*6EQJARc(0l_(g?StxdHd4obI0u~#%#v_IhntX*#++k_7BCPNg8G}9Y
z`7MO>J5irAmPN0Fvd3^z9z+a(ncbhOw#DSd<$p}Qyt9{$kZ*TM4sQEHCWn6Vob7hN
zx^vp^sJ+p!Fv`BLuI1nn)HZd-U4<<^pH#t@5iH2!9jY!&X6cAbH}vdauEJD-%{O|{
ztt6cDgUl)rt-!}lZhWwNETS?6Lq`wK9V-n+9QNPXdk>(jv!>gZadcE1F<~Z163mK$
zAc%=*071uysHm6*K`>y#F@YojQ4EL*m?K8SfC&L*Kro?TAPATc5EYbge~mNeJKwEa
zr%s()b;DQ3dS7eCk>~k`y?d`-y?XT(Mgdi7$$Zz4E0CI9svG}N@Z&5mSB3-@X2-I-
z4!V6?88e<oW|SB=FUc5t@Y`B4`mgHZp?CKfC)&)0vFBy|eR5{&wlKf1vSgG_{dmlY
z<!--^)brBo=^sDz*rZ2|G9P60?|g0aE`6S<OfQNyaD>##-PA|WY_j%lCKh*jwIO3v
z=LZ-5)dGC>{E*cz;vU0cyBMFecFFY_2n?%bFExAj2WST8EWrw8mOTnPuf5e&mmSgB
zJu-6}=Qml^=7c?_wC<GUH*NdrL?AEjQ{gM#!Zke1T=p*gBXxuTrQD>vG1UD~nl-YR
zpu+MDbX~-T$*{OWu=@PMZ1Ai*z5urNGmo%@#pOKXBgv&Ii7S=iU^*6HSzp1?^_$pL
zKDgPJb*Ev|IGI;dm7y5)WsJcK<rg#gm#QaTlXy|E*6J_LffHq%G)FUK^->wa$7q6;
ze|%t~{I=RAPv*%4LeG3o-STUggMO%@SgYc(F*LGrM{TRSSGTMsB`R~7kC$-*skEG(
zc0OE|d2bDSup8o->ujsUrUPv~o2~-Uave*{EXR72Z~vYUba{#EUt@<<6_mc;>Hk~I
zuPGy{Il<iS%uzZ}pz{ypNC~cOWkdd@Ie6v0;`Y<rhVdaAE^gj8b_aKiz9C8d81aSZ
z-}L#)+szF&CIpV&*;ixm;47s!4_Q<WwgpW4nN4Z^s<ex=0>0TXwS0Q}xDKxFY54QL
zm!#tAQ{)<cby>$*<jcJb45i{o`J+)y>%?+e7z-9+6eZJPGAtNBUC8rMdnUu&XxfTv
zG9s{J|E}Nr$fO{dyvgz8#?%?C!)o!htbOrmtpP1}#o5w0{Qge+iK_T9qm-bWJ^8@!
zO<<a;bQ+a0lT>-}5lk3VT^3@g_6%13`Vh>QC!LuwOh^lv8uaEKGKONa+I%z0vnn+o
z4zcF@+${C$S=c&)L`GT|k%`h(lmO_WBa)=F-}m$`w$|S1qwAV*fc$Y@>`o2n5>B3D
zaf<}xG-DotN*kxTQV}vdczgI3BbWZgF9zQm+52|uI_vGj;G@*ms(P)@^eH$nRNqId
zw?`A5KJ~jv+skl}W*GGi^JNub8PvwqT^1Jn{kqX1RYqear>6e?90v3mP*nfZ7mklS
zv+><eKBA-NQzFn*qSdUR9d`^&f=YAF6Ry2U4cu3@D>{LWM9ajo@YwNFe(ISdp09vo
zSD#ktzm>75xt&edAIeg;shEF^vn|zJx6J^eUFp7Ik-P}QB|GM=f4TDg$zo~69=s%e
zUL+@?H+77Ry=_(-VhQkf)TdvzR$cl=4S8g($U!W=d2v=n-5-i)|CVVDyQFgx+GW1V
zLm2Ram|9F4c=!@n26%?OivY}sw!@nZ>~p#a$iQ%U$M$-I2mc_|^^YtwCK*Z2oJtt+
zU%kQm{Lbx+s@lD~D^S1fTF3JX78Oh%*5low9raJqZhT=&n`Py}6{#9wF$Eem^|@5X
zOtaz4rL>!EY1`4Xc=~e_gh)*jdiT9)%COM9<uZ1?a;df1Z@OE0UoE|LwLLwm#gC3Y
zB<&m*VKQ*b;uY)jSCci=Kg3N550U9a0(%=-w5`1-eI;$PE*ok#v-O(w@@)iBd~gY%
zDP_5DY5b+8-yuOX_eCnV`#fr~-KzC17RGaA-%9st=3jeH_4`w^a~#Q%w^tbtaLqA1
zYtg1#`C_efj>K7MgKVOBPSPmXbV!E5p_rU*DweQXHJ4g3F5|ESAnKVpU^c+!%C8?f
zz9<rO(Rb$X+4uQ588^Gh8-QGnjJd<wvvbN4tE=1nIrE^_*x9WXP*X7Ppiv%p4BE8y
zESo_oBmSDLbU>>wD^Tm76=)S&vDHqJ>t`7b7NU=fI2lr9S@B|b)t%LY#>OP;bft0-
zKGbvm0VEP;HE(S{PsdR-oSg9#FqkK*zUo6P)im9+vscH=zLqO)v-s|2<61*0QcUQ8
z0)5(OEE#yC(&?DL$CFxWLPgh-Q&NmL3iazW^40Ivj+j0k&FSdrb)nNIb@~3ah|HzE
zN2;7?uR0wCh4p&Wz`tKnf}CZ=wwi_C#M_wGs3SL=g)ENm2hR{+M*5fg6il+E<P7>m
z{a;F0RmHUj&lJyw=9mulpi0sHPZ?@H<Z{`_QJ<S^c=8+MN*eVsA{dGG1kf&wi|y_K
z-3p^`YfhY)1p|T-Ow-I0VM(`A^&P`graso#*PxHJr4R6Dvs3#&yUr+>ey}vjsbQ#3
zgA|qFmU9bRc4t}E)^kTD^VJS8vsmwJ^x*)t{<nR*3B%^oApFn3X$LMOzkk2|IukxH
z&+V_TO0#LUOfZLyV!e|D2^B-4GVOYww!{4DHuDGHqx?k}lPDZ2fQw8|26d;kFE4KT
zM~I)t6U~H&YJP5&>bI6pE$V!3o=5jK?Q6Q~u2`0r%n;{+3&zhDBS`_RcPz7DXg!@H
z@~mLqqB@hdsa*bJNif2`kp}H1wX(XG-`Kf!xAjxf6X{6}@8WwDC`bj%-?dNLluKJi
z6*bu~<ckbGx!xyNupe`P+r03gP8xC=H{n*b!T^Y6^)}_Vb=e8t-k#3xhd!td)2!Q~
z_k5*l$V0#UiOvq@<#w;_^nD(L>@wQux7Q@byRc=biS`6zOFf}NJ`{J0)*V-&wV}M*
znU?!5g6o8hhF(RFsr9P;NuH0e(7P9&nh$r8`lZLVg(%?a0shBJ;9S%r?iUG$ZI^Xj
ztLfa;ykbg(b=7x{{ZFNuC%t4j-~ij+yCBBjOMPL)758m=Ys+-D@K%Ma00+21cM7C<
z*(Q~WnCRue0DptT0KNU{M$$Lu5n!1`&b0GmzxF}C&*DJ;rDi>=hNRKv>t|Hc_tnW9
zFX+8iB`w#!n{l4YwJ_-VW>@;#)2!Xxr{<H=3s|shjr3KHa0ba4ay-OrYrJcOrj4FQ
z6IG*>5sDJja;<#px98;78@30{5td-ur2bP1Om8JlPUsUd)nU{vjy*Ai<jC^o4Ztz)
zAz%>9J6v~?!O*m<uAm`~_Ev}CzcBQW#H+8#%)=zb2#ns?dVhZiu6WGx0^5>nJwnp9
z((B225f@4q{7R3)@2?K+Re~1ZcX|#VO)Am5@Z2q?27y1bh+m~!k@moM+Pb~k>2W77
zHP_Fm|8el3&hm$@_H&QzSaS0{bDFoDvsql>@tBiPbT6%y-Z{n0+&Jug&!lOFgXcQG
zjLFFJ-rxE>NW1uM@z$zwZ<&(VJfW710*bfIzMU$Hn;0+CIdWV!Kb@>IOYwMtr0K24
zfKi)vznrQuKH#4t;J|D?Sh&yRREw<|g(HGBPr@Bk8-KrhyR`k~1}@!omMmP5aunNg
zC}u{H4Po_ibf!lbeP(yv%MRPKXD}MBotIj6O@u|h{mo1a(}9qxbTFf%k>RkA5Xqev
zw@LVT&CuoJoi?_qt76%*JNpl@*ZIC>5=nl)ZEtTGGwG4FruIx93b5hRcY$74#IAqC
z^%hAH$Ce4U$$4~UXRBqD&@c6^mD$~PD>e<}?@|dXS}0Qezdq}Y8goq3r0h<c_tZ{b
z2pQk7v!wE=shK}M73@6z>pss+Yf2}Id+{*fRi&=F#`>4ZV0Kp+^gg6Io(s=V&CCVR
z>tz54gZGEtR~epYC6oqavCZBBjbTYkq!wh+(R4HU=-p3!#eerbdd|VGo|K;7)L@_H
z1(JzdmJ>A@DTs(wW*|vL;q)BBL8I#}%m)#1ueK+~OJ<wDx|>aPukLA+e*nlLzvZzB
zVF7JReGk6YHLJ4r%x9^FgxyqU9^2~A^YUI*tNlS>a~Nx+6TgOm>HU64F(@OjXz%jr
znn~Sv%LKDm+a9pmXtYiM?=<)(4wK+LFO(wTqvvg6aMLMPp~WR;ZN^k=*wXDnD@gyr
z@$S(4O$S-j?F~Cp++s1%RbJfgeld0h`|iLew%KPdAYag=*J&thdFRDL-nzZnZnDuv
z-z(Ti`^ZS<f{*(*(aa?K$>|r#tZ$}PZJo)bx?g`;R9GlnkJo983ey;{v|9C+!LvUJ
z8}P2f@|nx#2J>B}wI#xFg0X?S6WMs~(~xwJQKw2z93e?a^1ksDXczi=XQI(U&fVte
zt0gCFZ8EFdJYaK0U>%hGeEZ*I1pFzgPEj8_N3QBHA8steyFWGefT}r$uRJftT;Y*@
zzCb!P1Ln)UU9T5F1sQ}6$?=L36U7eSlHhII+FZ-R-?Ep#N&##`@5=?DQUf@0n;Mr4
zRU*;u>9h*ub-QhZv&pT_u5lXi;yCuVfg@*@lz4(lCTDv4RBfF2O=&RM;;4_OO2cr=
zv5T`ut)gD{*4Ep`EeS5q$mC`f(f?0w!E>KyT)B-}d6*q+jg{S&xk?9S{Y!u*jjrFO
z^QNtHu0*q2YgYH!Y}nMg^6-R>x{V}6tz7yv*>c}CJPn8C%f$0WV;9pcshH5)ZRHgE
zkuNPfe(pHNy4BjaYVIIEr$!VjU5DI0&Xy!&C}9`9d{4Wj3EQx#+Ii)X>DQ)^1)Tj2
zDf8?DJ9L$<e7u+)wLAxzDx3rU3*-mi#a#vXho_G>z$v=uXbt;P$=0!_25Y@p-L6f9
z|0`s9loKDI$zjtAk4Lg5&l=fH1!M9YSkLcIUFp5$!sZ6d7DH20MU|B4y?}COe22~x
z{i%MEFi?nap@74Wi>AB$qd1U4B@Ex_1z2l)?M@Q!3EJ!`=L%802Phfj!i_y8=Z9ON
zz^ds#tbV5$d?4ZMYaGDgxN$JkN6qw}1IUUNz9mJ<U7M>l88Qu`QMJ~c@8fjp?si%V
zt*bSTP;F23v(Ph@XlkMp#su}&6Qg#GIC3Oq#CI3UFSq-(sRuAohEIPKAHKd(s;eOZ
z2a+}Go?b&-H|&Ml%HjiVLaQZq?7QT}U70CZYj?oz$nZ^ykW>C85*<pv-3HH5Pc6z=
z0~tCQ?*Hl#-3Yf^9JWzEuJVRMwk5L1`AgAPHiiGveADo)CgoNUPJLT(TzfN4mvl`1
zk))!8z+5fy1FACCe&2j+^^=uP9hWDn<N}HQ{x*IeN%D}fCW$5sA(4&{<Kxoz)m0Z6
zCcL1U!KwZXum}#fylBRH%AMgF|MrlfmUao#e1ESw4*0qAN=HB748*k7+@fsPE4|`9
zV)nP$Td8FoEz2<DS6=wQ_#&`k)T&ccE_9EV8D^iC{<y7PYgZtK3{Q&l@1kSoWG%C`
zRB77VVqdy20|Q?7>i}V0g1}R?R`-&f<cXRpr*161H4HoF4%c43=T!4+Il_Ga1a;vR
zccG`pD!{2WQ$J57s_u(Z-BPI1ou&qVJ`el!VlDFr)^8s3c>Bci;Hk2+lG}tC73>&V
zZnrU2<I?z0pNe#qB|0sVEH+*a^|{r=%*OaRRpe3qTN^&XVN|ocP5aMh8-?kg%Wy7+
z<H9A4M2#7APQ6K!xgicM%12w}c3_x&)U}uLh-<PFYNl3{ASe>h_wqf54fcA=U@Xk4
zOK$@!(bCEIqzb5LYHG|K?tsH^wG};<Q^*;wK84tUAqeMwClz!)FW&Bd4}2;j-+Mhy
zS5ucc%S1Z%1n}eoFoyLiSA=%Y6ySX|jnuN+<xO>@l&ZDMRzi7UQ~i;0#bnAr^5rND
zD9y7<>gZ__&2?V?t$tolaKlzxyYU4gJ6D8`AbT^;QmKOmsSak8hE!JI13*DyJgd9c
zGZPl1PMGS#!gfFY*1I7Ul~3I)IAUt34{jS+%U5=+E2->4qU}_1ExoB)+1dNOd)x}z
z^o0;M<m`;)o<aqyF3$;hr<d2QrU@ao0P4Tp*AkhICse17vi)db^}qGGyLD|v^uF4P
z!v{N#FlPSCYIvmb>2M~c+zN)@Z-AEVD8aG);e1zf-tizi;grd~(jC1s_~|*$Wf_Ml
zbn5<@59N?=vVFEh+G*A<N(Z94{=@rp)AVYh>7}r)qV?fBYg5xK>iB%;wN8GXiEStH
zyc?>z5&`L=^fS99_viI=S!&ZcfVwh|d&TqraNqdUvDX8(NHu#b(L0>t1Wx>iZuFI*
z+R2O8lof@wtvf-a8^WQGyKBytWP|*p+H<2h94-3j8bg;G)-)S<pr?Vsa>}9ZYm*Fh
zSE5`@s7&nB+WQKaqJ(4-{FtP2QoAq=ccB&xI9l1xR)uWshm>exh>A1p2Rj~TD+nyF
z{$=aUCbyyDClN0jj`;ZO96Z)w|7H6%ZT0+n{{p1DI<KFy<vka<1hknl^t@^02|K;<
z#2v%W4{qBp;glcgw~Js`zUBI<?IX+Wve-E1|J%kHx^ZTx&&ZvDN=Ii6GLF$;qYx*b
zqk-{GGb*~1vwPPrFHQ+F)JHhexm)Z+bMqj2@xAaqVmaKfNFVTOO8<pTjLX?+q!}Xm
z$k@7ee3WHJ76;%XfA%Tpt=rIopT`{bi>`F(rL5@a@%KQS1`}SdZ<eQioHq*~kkMb9
z$a`k2NNiQ362s2xwI!c%3FlE7nrF#G8rdWwA|#B%O^}YOBSDaTapcp%#PKlQGS;2y
zPDZmt3o5}<x~Cc}Og&ZZ!8b?Tsq^<cjld5%@m!R8`OAACxFw~Rduy;bYU|tLpuQ)(
z>uN{s=-I9Gb~ZoGYNpa$ySN}P<KPSq$*NkHc&?Vvwr2j_;2{jTrt%|v1sE~u)2@(^
ze$4izD%U;Dq~c7Nlb22loxSy^Oy+-g%8bxcHh!-4`!<NMO`*jGfzD%aw_>*RXKZAG
zm_LGwh&}jsYHBsl(UW<y)4)<ywG;auB6DHjZvp!jl>QVcDUr000ws-cN64SyBgJfI
zeF*((^AR<0_N_#qg0WvYf%y}jKJzZkrP-%qn|CF)!1$)&&i;h@@%R&yV)<-o2IFyn
z4b~TS@A_i#+RuA<_JRBbWCRFHUeZ)X22FM(r<)V+(7egd_eqZMPF@6Yjm|o$%zU%>
zh*u%ZJ0cvJAy0wBgNV2z?dfY1%3<i8%BU&b?!{c@>5zW5Ys<IA1UOMWGONyQv1zwT
zr=y}~|6WlASb!%h1Cj|?flqzGRS+>TSIeMEFDD+yrtKf+%^$#SkRzYuB!vDJUKI-Z
zB6moX4rf9b%Z5O{2oF|jeA)OUt@c+cYU33?I_vb>FURxlEMw}cqi2RWGw!N{i$TyN
z?VX@Vs<Zznvv{QOoQo*48uoBVn!9nP5Rsq~BHuRJo=B`bKrdd+CbRpad3gK_&Z!R4
z3DGB3CqwE$yr|?I{=MUmAf7p>Zv7pau7)5^gHA1t16Y=M@E$}~hK=3JeNPfNj2u_w
zZsI1*ayPbVt=|95#OWy`thWK$V*u@C^&5G3s86)%t-itrHgTVO?6BLo@XYYaF-b97
zor_MG51G;{U5cvX8!t_*I$ihu^lFz}LcHx#ySm{U=f7tM|NYIM<vmy@B<6NhZ_91Y
zMf#qP43~}Fm8vl&rf!q%#`!(Qgf{-S2oh}W__~Q3q<O)fGoMAelXJQMgff*q)GO7h
zoQY%%MhibSV^-^#E2pp`U`Sg&%Ou?@{&ZKtTHC{_&0Fu>HZxa-l3#SPxA4&#N(ZAz
z^(`-c;Bis)v6*0!w4a&H1i-#$-~82B&5+0I(AIW9t3xN*<JwO5G3j+^-PG+vuu;s0
ziiFZFeg8yj)ulOFtB~CG36<v8i`)a%Ojoj+kNOo&k{Qj(%YWchSVaxip@V$TVK6hT
zj{VNQ@#}F)uk_|Zg+yHstIPEtW-1?MDj%kyrIin7;4u3>I7rIS;}bnj?CR%iQMr`*
zJ8b|?kQvH%+l<-xc0O)aWW%}YTY*%?1E>FDvp!YhO=jIK-CGsfav+7RcM1B{{2}=p
z{$}GX-2?op+Zj-sk$iB|%{$t+Dk{LvGP?F?Py<7Y6czg(HGNiVj<zMix3Wdt+6Xmt
z?T&XX6}t>qu<9WXQA1&9UeAAAdXRnAy;_W7I>-A6D-V^NMp`F@ZUy?DciitGkBu7l
z{{x+D{cv{F@?B&yHNTqtKQ+m9mwEO_ajEB+Su)}m1)K38#ni84D}<A%unlN!b12I@
zs5VfD@z>mu!SS&t$53J2crQr@y{pT{*AG9PfA3$#wpB-3!1?t*7+kNPzvwtG3bxj!
zmokf20vF&kSN-_-qnWVZ%6ETNh?z$Gly40?wqi~N5KgRK9}VYn?0_-U_}kZ(oqss?
z2WOmphZcNan7)5}{TKf~{|Ns-EWiGLZCU--cqaEoFJ(`59|O!of8%bv8mLyaRt1hQ
zqL#WuF}cQ)^)xB28?Zk>Ng+wP!cFN2+42nz>r?)>Ptbd@8|ZdlH(=Jd_6mlFmZF{o
zhoy8n$?uZWpS6&m_Wb$N6bWD_bQTIhif~wa;Q$u2Rbw8d+X>Mwg${f~l_GSYGJhP?
zlt5wlgzlrlFo&_-Y7G&#q24^_Zzfv!7#VQi<q<hrUW+B*f3V4+%6$HV@LG)(hLdUZ
ztnGrhv`=jp{xsMX<@`E35)UYe2@$Qv!g0W1XxAH9S*ddYL-odgXdKwJ%;=!}^*`$W
z`jHhq_JntH=-Tc0ac)9RG(H(lfa)^9qt8%ZV_b56&IovZ9Tb#it-Cku5yQe2x;X05
zu~T<Cx!Ji%;ySY~dMES9OY5Ww5VU`*WINdV*_`n0Bc&%SSV_LFj>(fBDN-h!S+$e`
z0C-2ImF)YNU%A!S`~1$B__q}k8pC*5^L3O@Pt)eF^HThw!>28=PD$h<6}lbfJykm(
zZ-q^dSWyShFP#VlVz;#}%z-<oH9yd9er%L<7Fn9gM?2y0FQ#-`Doy|aH?^NWDN|f)
zhAyc8o+ua)DN)1}A-tMSVJWoUXU2XMVKBMR&4W`;6!ZRX#155zKTU}XSQhw7M)%0;
zKZ;46g2GXW^#$P}AsOhwD%aoux-TSToN@iPD9X3=Q;wVQ&)r$blj{fmlh@w#`Eyi`
zkhoU{<jW0=mzO(BtbrdK>TxPA3V!}zz0FLFAJ4aCHid`{gP4hyFhEQGy0W~nS)%Tt
z7!Z{$%ocAwY;FBRZ@tU?A~cZ9F6nc@jy38*R@Bomet?tn&j1`AD!m&rw;eotnBEON
z)3STTBsZVf6-RNAQE8>d`M@Dz@^~CizCT4;+V%VLFGJtP{_j!k|91}f{i)+={jm(7
zLm`4XtSlumNWso1+D@D}fu{gWx(gXIe>{~05gPL)d1if=O_o*`JMj^^zW!;3(SULE
z{70!W{?rA_q!hGsP-ts;dA!bj_@5#YbWrUnTf&Sjp(iGdOs%Z;cdE{#FcR>Rut8>W
ze9*iHubDoOwy+8jHRMI95KK&o_S7=9M?)>pJE0(SRPXrWC?Uo_ILN+wd|b*KW;9H=
z!bQ<xhzrqY=-hNXQWRCZ0R~)k(B*Erx<z1Ex`1-1)(WS?{|Ge*XueZWMX;93O){8v
z(ur5iS=?HVu4l-L4jCIOSjF-@a$sf1Kmt7ye?JR#g-!P1?*@kYzmsP-`Q!#|_YXtf
zAn6tPN5)-#|Ar5=@$!;sii=sy-42c2JLP<{9>Kosk&$cj$cA>m^IbP>Vf3#<_fr91
z%8*zDcP^k~zcZ^4cJ0CWrQ(S1c*2&HYG_K;8NmFH?Ks!=mHduLxo|!k#gTBRF_C9o
z7Rg(GaxV&dhmBh*2JhNc1<e(~!_iq8hUPMrXG2whN>;awU)jI)QM=1#E@w0zucWa@
zl{Rja44BHP&YCbz6SN&a2uo?H6`mb=B+kA>fr5!#A$h#%*yg4@Qd|n+Aw9ujQ{e7+
zP7*j-48mMUU?E&(27Em-fa^k|hNLSNRX0W_CkUIw0ZBFuZGXsO<4umzxJc7OHAdy#
zHq~EY*VKaz)>YSjHmf_wy$+j-6qRvwulaoQL8m08jbBQZDGFL1*bkaA!1)&`U+AcR
z7&ixofeOniwBwK&9iL$_%X|b${1JT#6HMG*P;s3n`abjEeOUrWw8ZFzf&}W<{{7GK
zL`3(MAJBg-5^4zH)8yyFR$-^?M%73twPs6YO$XV{#oIJt9(i@J#)$P7p(3zDFR5Bk
zTm?y4M|H%srs5$EW1+88D<JEq!%9f3#EYn$%lOFVKq4j^jqs0H9roZAu0`<=60Fbg
zE=6I6nw_edI#o^-{re1l{GfI*HgBBb9EeyRyD?c_UCpaHQ>W&pnT(QUZ>gNWU|5C*
z96{S#F|I>p%IYVJUy=RiB^U;?Bho{?@75&_s&)3;>${?k1}Q1nWTV(o@tji)eE{ng
zx+?Bmn(_PXWX>rmTRhCuKGWavQCb{It9DSNb5DJ-I4*ZHtQjvu)U(5UL<}K<-<`K%
zaVaF+&bT67u8nGgq|c=yNUBw>UY@ek3kwN}WqIr@OPZ{<hHct!`DLbv<*U4>1eihw
z+p=PrDO=h6M_Yy0Rt7}>{qMExR#O`Bt-gby@xaHeez`19-fVlXsb9%Oee3r);=duD
z#dc?dwnHkBBr-@)*tIR=i>O_FkW#icy)pjPuOG5JSAhhGA&T$pkcZV(r`J_OT74=W
z!y1-pn=}n7sZ6@K=&ENF3C-b%B{rTvI51l`G1`pIVGZ?#eg>+BZu{ps|27$k86_Rw
zkh3d~*PBlbT{V)`ktxlD)egK~TS~M$_eh<1aYD_Y$&K$Z50Ti&5>QR}P4y&|7dRcf
zkk`)KD!nO`eV!12ZbTxrdy692b5-iC<*@x?nD264#9s0JD0LV*T$a3f=SORkVKZ4L
zD`+BjcIC6LemHwSbZz}51lC^yfn@z4>6x`Qc_8;HxMbdsNaDjbU0bE@I*l(^74D%%
z+4_Tu$I%_c9B1}c#>alm-gv>+vAf#8pbmOkH*Ue|^GIZpB3`N~ZW;ow9~Ic~6@2A}
z!plx>Hyi1=*MFxVhH8iF6Pnyc`zD2y*{U)>1FEfiS`!>C3OP=~PppU%euzk1)#=>1
zbNdi<7N-&1Fl4e%btgnl;i8MbYMw4evU>`}@y@zUrI5_4DfXDF4GMyms>YJu|MF4Z
z&wu)2Dsj#i>Dc&Y9%;JQ7G}^gphK(kOR30;?Gmx3$An{#&_yBZTg>k3x9TO0Smu+S
z%Z|c;>4z8r$!r!X=9JP`*rkJG%;L-7So~a5sV}ayb4lIuPAr6pU2T-AoX{N2fV=8v
zkmxnhzmMK)LBfIzgV<n0jqQlf-snT7q<)?K_`!$+c6`l-!mF?71wq1xs5uQflztmQ
zn^zDdt#8)bx61cES0Eivn~W>qNk!}ILm`dgX$oy=9GUFs4c*`JFz0&aL)o?S_+xGA
z5--3h6^Yg_hS017VKUr2@-U%nnC+xBaGD&}<2Uwea~>{k|CnL^Z^4glkA7Yqdm!~I
ze6r5_mi!+0td<-E4!>*s8vJ0SHS)k_#U)9q%X`v#;^h3`bAQUZEzg-2uIwo)(Tdxn
zoO^#=LIH{^8Fav{PVQcat#fg5Q}BY<n>csvT^LpBg$v8HC?DFmySxFJzuvb#QhPFo
z%4}85!dcyjy=mXOvsXr@zRh-taJNW3`m7%UCb75;a&^K$EcKP~yjitufV05i)q(hY
zO`PU#e*NxFheu)*&zZhav)1fuiOG;-?-r?4HXJ^5(6b3cAGFU=@Nd#<y`s&qh7Cg;
z?Q+{D`i(kp(Z=ydEB`62o==?W@3*jR+e?0mO(wN#b<$U_%an%W?7o@@tLi71mB07g
zy=(9;?QOH0y!zy1zzQl%$<h9tqf5)fn~iP8?lkjW%U*F_eiC~(jwQBe@~53u*ny6V
z$>#|Nr^$Vq&1!bX2*Nq!Shyhtru>V{7%(Uf#vEFX_1KRb%uT1Hnm(c6OP#i5wwjU;
z71MUvY6DV4B*{G5)r(f0it^&n<DaXuTpO*3n@82)70Vo{gVD05;uiqHENzZes{Ipa
ztnGLnM%>Z7<{gnJtS8v3a_j!}i-;oI_I=wc^kM!Eaqc@u^`v2A%o2=USMl0HSHFej
z{q5s(7~@6G>#Lr+j}mivW3-1_PV@qu8NDlK{#_BIN!iUga0!h9fA8wATk<`hOke@z
zVUBm-m}%?1CrUjku_a%wpd-XiZ9dy*Yjz~53h?XJRwq6+?xR=QJf(Ln&sW%-XlQF4
zICzxlUlhb<bH>Kj&(fJaxWun_8{M&3c8smv1U=uMPQ18R`M@<&iio_U2YiMfkUGMj
zug=eRP<;9~&{jyuUvGFby{F5(i8=9ZYa8|Iw5`aw`v&8N%{vHJKkQ2Rgy(6kmJk09
z;$2dB1_PoKK!eL)#fZNo@58WD;6hQ*myB?t#NdH;6~NKkOr!lYp8r%T@5`UUil*mn
zkxZzyfbqIm94~<hNiQCQQ=~5~I3EWSgOaP|(Lu(?$T*uS{l4-o#CLoBnDt(~DCdL_
z@PKrw9;l1lq1KNx4&2<;sMhpb&S&X6?%k}7;-;VeP(=lG-gYYzkkE-darskJNB+IU
z0@S8H(poNC_7&`qS3=}pjm&`MI5jz3sulSck=6Dt4(3Rq+@W{T6omji+r4I+G&;uy
zqA{e~MVTm^o}3)V`Rlw~r^+oOH8oYbEeTKV--c8?7(tM->U(a;8=3|N(~=o?1)y9>
zO;&lTwyZ2=Sy_EJSL>D&0OGEoU^1ZdO}ociz}Pk7*{mLCU<drN>u4=(-sv5`2mZK`
zF0UCJp3=}kId4az3|4^b7Tx8<A6)HA>J}zRR9Nz0|CdqCzFVT`a~t@vn3lP_4SEi9
zm;o3N%^HVRUWpvXrcQ7hudRd*shdNPmqOr!mk`I-k;lStjO{y|HQ(pThqF?01yM<l
z;i+rG?yo~GG!q^=_M;FDQoxDx3H#uu$$`rGKl#*M@)KaA#>LNQjapTtZ}IIka6#%%
zHJIOvr6_7hEd;4G3L2C^H)#b;LMu_(C8>Cf&QSslUPA}$*Gmk&ZS3Y+vqG_?8T{$P
zlatxxQp(FeQLH*c0}`c>1jeUOn=vC9i{l6R9_k>6d5&D!Sn46txiX-VXT?~7CnWM&
zj-Xtr6ltM`AV`+{9CalPKhl5jt=}GS_77_3_OCzn(~*>#@oYs481p>r)DAa86ZgD3
zcSa24^kQexGrQo!DQMYPfVO#!*#_1DcP_aZY6UqtIV+v(8nymN#ijX_ecZ=7dkZ%9
zl5DE!)SakbuL6K1eI)zvqHJ~7@fQ<Ysz8$2JlOpl-be&U`uWmLA${GyG_X1#V1RCM
z|EX`b4nn$V(j!e2q*F(<*fgBXEt1Sazic(RB(cKgkjotw!;aInMzU+N>eP9lc2g!}
zE1OqI<_KblC;5!(AQo0*%DwgsBfC+qXqlELdK&3=On2Q>J@?_?6jcG__L&e13U;fB
z&Rwn}=ScvFppi5m_q*6J^3>aX$fDZutvJJ(0DQFh(HZ4ELl%RmrTMt5p~EBT1*Diu
zGFWsuZ7`p8a0{@f;gsvEy>M$OL5fBe*zR>Hzr#!Emz*O1#O1w41mO%md%9C9s%+I!
ztBt~a$$`9C%uS6=NEgJS^5P1=G-^Q7<%+(izlu`0r{o*TJ=kqlvq5hH{olrHlkXw!
z=z2YpnFd#0)to|tN?NY?&I7>$;4qq-C$CzfpICuT_bZtUV3Tud%%Ja4?Uj6lhHt3a
z`qLP1(AMq&^X<u;(5-%^SK?gEZQ|ShhAZyqDHO<{a=tR21uD6h5CG{Wa&$;;1pg24
zfa+)Fr*M=UAks-LI$~f}sEQW`Tf7?ay`tEdMua1`dCI~9n{b3nI?kR4@bJe5w7qbl
z?cRQyG{e^P%Ok+$eS+SVg`{tiM3ni2^<h!fFQ(Tw;NT>e=YL#>)ORUaR>SM_>OI+*
zp+T)!rJ-RVukRL+L<;$57|l&lm|$aJDSIn4sDX_*&6=m*r|;l1d&;J-T9?@{uJX%D
zs5klbKd4BsfaGud7c>r$+xoBKo?|!nrqU1cTJUGYb0|s8#|YQq5)v<>24TbJCaSjp
zs|jyK(tA86T4CLJyBqr70sSipYKI3YE9|%0-p6T(>ne7FP#&o_3FM-o)_dC$R+>h)
z`sejuc%_ujWP5k4llIt9r6OMvmxzEm68*SPB}FTm5;QUr&LN{EZ~N1_aic`bLc6oG
zmD~)aWWDlU&$kNeVpOPmgeO$b@w^{PAF{|S#+$_a0elf78BiE;xDMipXt0p9ZL3k?
zPDnlxP;wY@R2*uutpeu4md6j@1(AedUp_Gdg}itju#GPeeaJNN9GDW@$b*DM#GUYd
zsa04%zW}v<2822-`XxpNxkDLFmAVE0hrpxUO^55>=$9jw-0Oy%j9_uUUy_PLX1`mx
zHidTY4%S(Um5c_R2j%UQ1_>z<eR1)t$k#V+O+7o?y5Isx-7)60dAS<W<z%mCcwJcg
zc+3{T?vk2LLrQP_>b^?wl)Lwz6f32ztzjQ5^R8t@M-<O0<pdLmR`%*0*<f$S`0*RJ
zU~%^y@z{NaeW)w|T#Lyy1^(5|MP)%?y{jtf;igJgsqZZLQpf66{1(`5nqAx1r>>NA
zFSG)`xPg+jhd9X1(rg6HG%+<cWVpUh>#)^@Eq9oxY|gBU7H0_!FRPw614l~JdPwl=
zCL$6keieskD%^AA(5$H5^FK=VoAbA{IBoYd4BJWl2RvdNe^AtMPHUE4|Lqwl;M6cZ
zcj9+P8o}WXuk+vTn6gT|us%?(8!5gld6>wXz6o=v{dy^3)9u_lrWk%@p(o@Db<8T|
z)&Xb`EN#1-3DL4FTyPoeq}X@tQFj&rn%Pt2`e^&6N18$rKj7`l2(=Znnr-*9eQ96X
z%Wg$C8f-#+&WF{i|BZ9aMmlS(BK3_`enSjaNjss%CGjTYfJpmf$RNi_B<k<-qU_`>
zr&|v`TjyCFnKhqUB}N@F<IHIv){-uvg{LmhC3i3%P>a~{8^ZARAF_l4XES8<61m>F
z_0y)n=ly`AMsii#dcG0N9B1t{G6T4Z4^&prsE7q2or`vyr4dRTx$j^JMd8<4v7^$c
zqxa<E2W<O~i%)Jp$J1&#=Ur@zhMf`i{jymcSGbW#zx`H!p*O4JO46Td@Ha#*2MOoh
zcAVAa81Icn*fb^NK(v?})ItJQilcg8tvj{1c*Ym|{;GEU6#gWX_6ioc%$+M{E%X6M
zBboF|a!6AMfMyp}k0ASRRJ@>`yBtOf`W3TBKaa7-Do^Zp8g0~WF~4h#nb^swpt$Ig
z@UNZBQ^zV2c(DQ*cP3*xQXD|Hc3byxdhFeJ9>T2b!Jfr^S;v3|q%a;R&|<NZ;a{*m
z(7EL+SuS1zOBbyoqF4l}RH5@C(zt^`xx#u&CbsO8$u4geRzs<M75$3Fh7Wuie?Mv}
z0+S^T(@IvCr|4=eFX0YFh0ik=`XMKcL@RQ@{7m>stzH6~hzkxXf8Cnd`$Bz7)}a5k
zDKL!!0c5@)FNr*U5M10yt(S_@E1_lleU(>`br3an9gk)8(!;Pg`{KuOmZ}pF5%Q4J
z5Q4);T#g6REG8Y$OmzpJASlzF=nNxB$P$u<aFhTa0|{`Pbc9XLy<WyV0oMf)%8uT3
z)$AA8vD0yRe#JfuIe@M!45Rx!mel^T#aeE!S>H#w3n|uQ^tgt9@F9Yqm)aK$eU}$M
z&~RA={5N>v=`b#_@j<Hv|9Z(t7~45Q`uT=-!%GH(KRTq00G{IGMFJyb1GZqV8^f36
zx>X`kd&?rl$K?-hjFL-T0VFNmT-#3n<0Vt`rjPeq5G>C(h)-bGKc=!rFonJJ%H7nS
zZi_vqGZcS%$)E>@AB74T$BV)u#*PR`IA<C$eQV}bHBao9Jh_!yN&s?&Tn5LfV4WjQ
z0caD+aOA9p9W6A78O{;#kpC?iNO}ESUuQh4oGC|c>ZtJqqF=TwFF{0p93k>}`FYqw
zu`iNs0|DHlppvTBRTd*G_X+19qeIkVfsuggAJRTTHmhw}$)o&)^o>v`Ng03@fivsJ
zH8Um!_CPp786g_ns_-dim}<5m23sUf#p1CwjT0``RC*ty{64(;X>!3W$grm5$I=!?
zmVvK-pW-8&Ty%wxqL`NAQ7yD2tXoJQac^l`dFc}oCFOBBRjoHwL(gRvaCG`tM_s@c
zWn#kj3Hp|l`=txpe%%TV)biCc98+kK8&PN%<EQD-r0uYO8x6gQhvMm`JjpN%@Vg=C
z`Smrc*8kdUG5)3@Pv5a~>@G<wex~4QB2y=lHM>;S0b`G6gpR<Rw^q8e29L#^sn?~P
z&>2SvD~5+%V4>Q0?}0`R`uz$D?P5pXt4-dAO_3BA@APT)o`655dfidpZvZ+?_OI9V
zQJcJO-_Di6l(1zuAcu6>iKd4!mOAq`Do0-MmHG<ly;Vzl*6*X^2QCAz6|6@N8&Bh4
zNb1zRtz+BVV<CoD2E&Q!4XZYy1$x)sy?fu^h?}Uy)sL!Pf+LxYKE5&^vfQy)YkF6j
zia(5F-`%eY>i)-x4$HGJ%UH{X+@iJm{mDzn$oCpvCEa~N#N<mynT%Jgp(J3tsDmUO
zi_+tCc1CX<V*K@YwfU_!-_DIQSeHQQt6Rrv{?u>!;usda5W*~pSx)Yp%3eVip_H=B
zmeVjk+WJV!U^;YW<&)Nm;l{Xx3QY1^dR>m_KD4|yZ$3E|mkr;KcJ}*dvhe-sETf|D
z<5u=lpsmy5puw#(1seOJ4nHo^2YBQKGz=QRQ5aYo$UJ8cYs-ckN2M>S;4Bi=V~F`g
z=P{`CN>2Dp;sF~{NZBygTou_ktS(CM`v`=`-<UKijouIOK3Y+d3No+^Qt4>wqj_rh
z^;}XqV)AFm$lFbe9HjdKUiT|>RqMuWY%(UV!pRH6`JF4iq~rN1B7gF`*el{v65!pV
zdZn+?ESzOX5gP<-d;il}hV!H6il5FhH2o6d9~0?mt}J@yg-jMi-MRnV(RuyV-fyO@
zD0g#b5@prppTIb<N$ta4L5|nqQ7@DUfO}`<1aQ|{D%X*43tVja#1)M$lHX<Bjhk>R
zE>QXnNG9#o{|EzdC@YTHH^g+JAkL|k*5L>2<_Z7)TXSRFmSf%=@KKXk^jBjd*4$I$
z4V6E&zj}TO#W>cN$3+Pm)Pq*5syUVuUMv)U9fjXmWXgtAjgpsPx-I0uHV^)%Eme6e
z=n$~0&}ahg3B1sq28)@U^CZyVWgHQmZG{ls1yq^v{qgC0J9_PTSj08ai0K+R;sthm
zpx8bRhf+~c1KbEZW;ls_W_iW)JTylT8|Gu8WnNjM{1?=S&~rxF#6=wA?yv;tZ;*tS
zGVfY=1X0`q3WJQ_9hq%0>T&0nEOuns*Kx(-aWTI16|qqyKVnG(VPK?6f0w`X7N@UJ
z^se^}q#dScS}!EPuEat52}Icq*Fh793E<mmrOaa|`^HntDE)}yVq);+DdiEJ+fHPM
zd-S@$0^PBtOI6kOy?bAvOo34B>YWT3=SowRbk9_c5SKt+rlI>j%0PD<&kR&iVynGL
z8GYJc%ox~&{zYmZ;}$L~$%@rVxr;#%k>sG^p*pNEP6<-qJ@9fTJn1MUma(o$<EoL>
za+<8<ne$+?;TI@7->Dtms3Xrv3Y5v=>=0G`*;X3A2GzV_r!Q}2RG2#pW+CJw90Eqz
z8csgaoF0<0MxYP7!lC0dKP$Cz6}%)@;FK|5_g`}^u0otL){pAzP*XbPk&dz={jq~f
zOkFc<XJKbxt3Lw&Z^<xXUx<1ldDAQ|3sSjNwyw^%IY}1NE3^Ls9q#Qv9y6`lvaQ|?
zmkJ32xMD-Jxzc0P;z|G*VEQtO^ZkACn8<)uNj7x<b!_hBA8@u>26f^pM-_cs&Jm&Z
zlJ{=sz)iVe@IC512YS=*PeR)-Z~7Wy(x7ad`{^2S-YLdaW(DQs=oYC@*E|zG)p@Y)
zTd;bNYTuHR&#aHx9k|6^do*9c!)d?hJaXWIqMmJ~xkKegW_=kVofu-~8X+o+R%$aV
z_C4_BwOT}4b$p7E&lf7*I%-uIHu?{VL5dclex|{za8)IrdRG}^nXQ|MtIYXOp4du1
z_nXt-quvb&D3y+y=TrY9`~@3&G&yYSr8lXtW$zvJeY<B0a+}cQ9PHF?>0hBJ;;cp-
zbQ#1<?7JYSR|JYIM{a&g&(%hhQf7C}$l<mxs|7^nqZ`>8qYtIFb<D!d$ptgVun$dP
zrq;)cix_Bx;4{i(ipHx_5I`XpKB_b<lEa&+=|)Tb{7S0OHAr`S+_^z;eGz-S;iA5g
zt>S^)y<`*ukOAeBt_nBGr+8UR=pY5|Mfu$hN7$dqkflEDTe`e`8giX~*(`}BDa)+v
zt)^eI$bUcZBxNtMP}id5PIBBOowI3Af4xqqy~tTA)FHYtUgnmMXFzt--kV^KQUY2J
z{P+yUUj~{`&*A2rz*GdNSuTm!X{qn#-EiRv$tq$4vDZ)kk|q*ywCoFWTKTQan@X7a
z*7dt=$;g9W@YRzsNmTipzB@zqT&y#a<Lfi&sI=YoJy7V|Yvv~o3V7g6>P9tUfqGVT
z`I{V*y$HKQKF0mDlF1V*nJhNQqkpYrj{ajM)1U_~Tg`gM?Wf)_Y?YR+u5^FV#@D8W
z9G22VI!_>f+smEABZN{;c-@Rq76rQxbE0li@3m?1ieR*O&DDpPxC~BDbZh4Eg!TCi
zaZAA>X#*;AJ}uIph|UpWs6)x4=*t-8duUYl9@oM4F3WjqudI0ln5)#cqo;~0fi$_@
zj3e;iNwoz}PfY#{p^)(I*l{tNBk6(SFN5D+lOJd>Yw;&X^22if<IF2sd23@+B2EuX
z9=GU5R*KCT&ES3?zamnz^fm$-(x6fBP5>!|hch(92>lOWX3$9`%9O910={dOJrs3>
z_U#^L;zqfxC7@AkmCjBO@)Zz0QEe|_C)VsYBfsy){cE&ve3*JuDTgDIXB8FRaZ3}Q
z+tA1_X-SQF!&(_@XPg=y=P_u*%WL*sS9aWS$M)Fo_HgCN%PN}MsMD{RqM~#K_mlX4
zTHkY%e;Nf$dfP3!1^L{~m&JQ-ejYveHzk#wI3S)U-_Q?tTsArFD|oj}y_L+#sRl02
zq%XqQYgmrT?HoCDw(E097m}{*hNiraxoHs{cWHTG_@S$G6o4#_2x-esjsDKt*qT$^
zYCvLFr&+cR?2!E4#=ByO$s^bPzC?5(8RtfBtKexT4tnXHq?+;2+@qS}2R3pu78Pl$
z2YKRclKAphwQnw;*eP3U#UUu*BU1nLJ3Ib@r%}GvWM~joqjp~=Ok=Imh|?~62vtea
z36-Uf#&Jq=bU3eXJN)y&KBpDH-wl^{ESgUeFYSCE0-mx4EW=-d9Vfr;T<x5Aa(3G<
zj~)$(rGYeg+0A3I8J3Ms9+`SpyM<S%(Z*SLiXGeL6GqpZ^8e)$-@j_U2e4ZA>|U4k
ztB@$@mI!yBylfm+GMmIoCL&N01@N?Zc+D(-q<lTsI!LCZ)jvedr{`h90lmrnk0tB8
zA}ihQQgq9=x5*pdRfj#wfvhX&v3#>~OLF`r>+hU*(KyMp`W9ty#5BccU~@GS!}5qr
zzFDJ69(dR76s!(7PkwxWoHAqJ?s*wzosHl)l$T!|r?|bFLm`XSmc};sg9T%A8}eYX
z_67RFcho>EyaXq3uD64hfg#NrBB+*Twe~N{3GFYi{R!!Ve4>J|Y8km`@(Oov?`9Xz
zWX{x$Z$egdszE@@8Vjob0~_@_xVu2Z7VD0gumtuY|6rsTtMB#J9f~I~Gve>>ow3-n
zw$-Xjb9=k*3vAPC<(3r66*{dPr^KQ46Y15PMRk<Z0Amx;QF9=E0~mSZ#h22vqk69i
zi!15sT1zu5vbB74QL=-OB3gtMKUQetG`@Y|i8a*2jXJ{sk~7aLOxRv?NxyDKo5zUm
zwm(cb(!QPVwHl)yJ&p%E3kCGF(@{vF<fje%-HQ?)opv%Sw$v>7BHg_F*RZ06D5kq!
zj3pe!2Fii52yfJ`7jILO?cX4Wf@N%t|ICaO!jm`@NOqf*O99X7Pm|b8lUw^Q48wd{
zOudBMqi2g`lWtbA>&zcw+;ip9O+<TfX#O7V%Bkd4J2NFbVuI?Dl+Rj6J&$l-v#-qP
z|4a1N?wi$X8w8saSYtkELMzjOfBgp@S%0oKy8CN;r2fjB-?dJ;8Aim7h`fK%;`D<l
zufKfpbAnP8%i&^AH*zErw(9^(l4u7420gvfoO!u-_w~xP%|&TJk%!LQl~Aa}ggHp;
zDEo|Y8>Tk+X7?#G)237m{CGg`SzWu!*~U+IHrnH`>uQ#PQOdb(L-~@VfR1fmECG#5
zgLGA^L5p9iWbfQ>Xb-!Qbr%A&_(JFgdAb+-rTY!FQ8VlCEqnB3BA_<`F+`_H9wLGo
zdX=2Bo1J^|XL+X4t^KdQYmC@j5Z^@V)R;QYS|xts8QE0N{8xji(*?PjhtzsQ69@j;
zGd1kmZL5lXzUxmuZDo4W(0$;=n4b{t|MbMdY>NKli76%y(bo$%X;wD>IDCx2OYA}6
zj6EFZiAs-mN8oiORml0$!{I%f#q|A&b5f=*XhAt|?3qh_U3w+-FWc6!zs{_ud81R_
zEGHys`kyZpmZ*KNDdQ_c9D0|mInlJ_?wA9262q0nyq7QvB>#j--C1f8Mz(#ABOL;~
zbN-9P47Zk4(@F#|AT0```4^KM4K-c!LGzm4la|yNi#dh+VfDiJ?Sn13RxMRcTh)LI
zJcX>gVJ4rAYVhUIKkK;e%V?1o0VpCqqU<cLw~;v_HZuBtVk0voWkgdkQofzr@_<+o
zow?DthiVEbmfE#5b59hn+Su&eS||NWhuUqT1Fx84PiQ)zyJR&;Q4_d@kRNZzs<K(a
zRI#~|E4~d35o>a4X|`D=>j*`D%?B5=g;-y35cC*-q~)#w=g|gffAa5{I@$lkAZmjf
zlK*`Ec0U0A@)Xug#NNG*$u65V4L?BePXM@w^no*gH3>BtRH72w&?7D{E%jb<vR?e2
zWe)ld7j`zAPL{fC{e+dy`&emHE+BVOcDJk3X|DDA6c?kCThH7(dvpzKSKGe3HA&Fc
zKLct8_Sr7^<ulRyVb~U4eW$=qPvok}0nhpAk$qZib29&yf-_2{IHSZw;f%8BUuP6Q
zi%u5eLZq0wsKPLBNW*T=VU0UJ3%XX;cZo)L(ddJ-UB~n?#`;rw6}fCE$+J)b>|{b*
zFPDjiD%T;Y@?H+#*&6!TH>XbQ(xrEjHAa63lG1TtuIR;A$JAc@b}QlnM0jr(XC9ur
z%j&SOEZtco^nG3R_d4-MV*2iRKcHj(&}qjFN<L;1S)0eJ1r46?5f8E+@keL6YklBw
zG6aVvnYA0YqG}+=B<Ew{0{P>}&wBo2tj$;CxF<2w*5KkAYQWZmo*Wsc)&E>&?Ov7T
zOJ9qkC98qnFqxrA?!4q8>Le^IPrrFz3XZh6NFJt~t)V)E=3VpMH1CFv_oR8Z<v$h?
zBVxCiXuq`@OM~DGz38NVJmEX4B)Ki0)#3@Vd*R~Q&xc3ahI|Kfma994=)v$}y(6kT
zQiwO_@z>T!|90R87R!6jg<$dsV(J>FXH(m&d+OxDBqAPz&|WfFE~KDGyA5J>9UCa6
ze4F1AAorI19c{h(^Qw_Zcf`&&>nkjO;zu}<^rS(;AEaFU<EC(omm@c}!$XLcVpN$}
zZ*o+tp-_eyzcg?7{YzxlK^^*OUaWzk(5s!1*Zo-&#xJ+cmQlCbot?eM`i=k4XE8)$
zXKb`0`i1Q;R^P%=+STmfN?xPTQEAHvq*P|&vxdAaWl=S0J4ZV4?L+W9S^sY*e&VR|
z;Hz3Dv_77X3Q$V|<hP#o(&u)rq-sXlxZ@cSrahMW=FGlCL_pT^Tc8J#79w9eNPFe7
z31vBV)*zR$xEQ^!XvdV0$>T&?UA^wKNPSZJfz}YEtL$R%!R;2hRep*4sY1mX)Y9KP
zKxc#49tiro85Edm8nBEEFsU|`kuE(-65UR{UDRzky}oZ2Jfi@?UTxiTN<8+5B4iHe
zpB6`TiHUwKvvaAQ#QHmmr^k(9u32YtajX@8#`VV1y-e)*&V%>3KpVBb{Y|4=T33PI
zvpUN+PkXqO{VkTI+2@XUD|u`=QKIV3pqh5sj=4H~(##N>`mRANw!5YL>RsmVG_tGV
zKl-)UHJQ2F@Vbch4JE3D`VD~i)cn^zaaO&3Vx#<D4TF$e6rybKc<@qONc6kSAnL>(
zzhYtbeNlNq3MMj?=e)))=5)51JvjaC{BEt*sL#o98E|l9L2fx@@o2Yi-;iMLs{Z{n
zJ5q85;bxyjLW3cP*jC`*ZbR1PiQv4M$9_{cUFms!hi*_M4h3yjy+)HI0xA?YP|N~x
zc`*de#Z=K|;UXhk<Lq~{T6M&GrXZ>3EC^s2A1{o>#oz%*3UHL7C-jlXmw3u>nX7oy
zht50+KI~>#4Luftu{llFYG!Mz@tWlTUkP<1E;1A#h$d*5@4*y#sEUgW9l;|v8j1VG
zmH*@-!(RX(iz);iX*org!gqhr15!NefmhMq@YLo%*!_OZok<Zxln{nB@%zdfOMbir
zMO)FQ&O;)=e$uEG(#JBe(-%p{ydL$zf*`r^P1qaNOHr+Cw%XE(9i%DVu_-+=@#T8(
zZ*t<hrQXHNZhiWXti#twF2YD=SHOWAQFZTMa^UQDZDCS6G)BZ8t$*)u%{3p*Q|O8#
z$j`Yx$f(3+dvIAO_};fi4|+)6#IAud8ONK97(yl{gNbJSkeq(y*-2OwQ?FfgDOrC%
zY#>FYrq^%I_Wbx<WTt%6VmlvB8}6cCxu1lQQ%Xly-i|&!MaoQZT4-7ZUyghy336B8
z!@(}F)^iB*z7<vA)ZMSI+)qo39JufgDl*9R)nC*i<uhN8{pZ5g0l`=Cu^v$IN}SRv
zgNyrW3L_J@*Za$Hr^IDJ?L4Eld27!<!2z12lto&T70E01|9F{GGAoksllJ(djBxr(
ztD1UC#~<<T?R&biBJv)F3=Q?qC|W{Y^ta|K8d`Kv@|lR)L)ErDeJE^s{4%|FdXL;L
ztlj{}PfN+<8u=dRI~G1V?$cez<u4|f{33Qh<xzw#v8ka9`d3`Njm-NOYf*8n-`wk{
zyfMSmas+CD7upXiw_BI0F|B8Pbf1}WJEYajN#()nA(qxjyYZ=aw`gFw8;D|2cQ>&n
ztAXFHXT|Q|b7$5S>Fza-4IyZ`A@Purl_m=WKGH%>dnv5pK1wRFJ`<1H&p}aOlug@$
zAxKN=w`A(FaVR*zuYKb2V!8mZG2@c&gh<hzJU{iB8mB3}VrPhK4PNc9;hmyaj6;D*
zXnV$%jb*$Pn?k&|sPrHrfXC1!36$|w5ou%@-Q%&5A+%pMz6P^{4nXYLNN&{lZirA>
z2DDO2N{7jVw%O++!7e8`Q$j4W5dYUnArf3FlOo{jYKgPbcEqBUnlh$WfgSqgAsep5
z{7kGG+zh7`8)&tVc8XJnJIEkM*6z_xbr-W;N3CAII`~yx?YG!(Lx|P<<!(3rs#*%|
zZPEUNjSqQr4uwb^RddZm^;}2^LsXa5FJI(v>Y$H1>^t0VA*b2p+F2?*W9gCd>%ZBv
zF9C=Iq+k(G@-+;CD+72Sj;OQ2TpXgRFAlLFYBeZRXsr6GQ$<f`X>r+y(*Ldb9wjMK
z*p2JjLf@yL?NYof=lxS+vKy}5Vi*Gpf~U6*%i5attQhJ%Xw^04kKDEtksjQqDexM{
z))vb+vDW(ZJZSKNo+?VZY2BO;Pu6KUdHFzFG0H}Rzo*i%dqL+R_#K<v#zdxB)amc7
z)4TPkp4`mmHmb1kv*%9MU-I{g?tagTbKMM=8D(GztzEqDbkjMxZY!Y4n_f?9s4L3p
zx*rL3d#bZ1V3)x|(H%T^aLGLLeus%IYnVwLus3nrP)HgP{BoAw4f^Bk!wckd-vnW(
zl{N|JJByYecE%|EM{qy{w&)ps#=mYlCsyaR^fL9_d`quPE}@e4NC5h%^4Z{3=^fqa
zqc%=c9@IQ{0ouJ|_boIJF)1)$`*2QS{W+=Jt>H+-BvzZB72DaTTKEi@J!qv00iwbT
zHkk6{;F6aV5;<+ENW59Fi`KUiX1Jnw4>Xk^3#0H{&;;WaM(CS59HcJ|!Kxv9%D-ZQ
zML>ssgujNn-b1Z@;b<t>J$C!h-1I4W%AF_!!TQsz(O!G)NtysXOLD5H@fv;2a^Cj)
z>Ath}ZK!}<KI|qvTTDy6q|jImGDR9#=+9zo(`S>x)Rt#5twH?pub&##tQ!%RC@Ow)
zj(URxdP~6`Qd!h4Y@D&>I<4zQ>MTPD!}+^M_@O*kd~u*I=`Zw(G{{Jr5U&-09n=j5
zyML~JgHZ4${d3|1YuQK5@plrk?eEPV!t7#?Gtd2}i3|b|(GRe*x|Yy^^Ne)`RWY$?
z;Q`CAO5M{12kJANazA6Rz<__SK$A_w51QU;g5yVUpOoqIRptl5!~)8%5!6wcR}<86
zz1f3)6&;u~v<Opls!s9lW2zsNtIQ~w0OjssG(d(TE{|ktT-{RO3CFsvekf}11!x(h
zvyjJil~@A=btN}ybY<WVp;!uD$pb<#z@wGuz}bQG#COW;Pq-B-ERFx3F#7TyjK_c$
zyO>N&VM@lc0O_Su^7zD0|2T3qYiUvcLKiUAgi0!wFeW@8QpTdI<H!K{k_A9sj{ddc
zjz)=mMehwB#NlWbE-J}@^Nj8+W}W>=v0WP)TDFy1k(hE<VIy-Au9;LW5|bWkfZ|ET
z_cFP2DfdC1D!L?M!xH#f?tS2djCMnT>qxuKP^ut`T%&{ipYjP@?{qGnYe8kapNgd<
zE2JE!%zZqzTp!gn4~Wob^ddf*vo0H2*a?jj12ATQb6E0X&kgI!PdtP;hm%~67iP-6
zM<E?&jwDT{#iW+XgxJ56u#4_2ZFoB_m8k-*=uEhx2i!)%s72jmN!M}QWc1>zWU-c6
zh7J)t#tK4earHkb!3dX05$Y%5h`zux3wMqTY+LnComxo+7wG$F*N44n|AxKiS^>e7
z{i=dz99r45_dEOQ(1f=;E8GhrE)-iHe-p>mjWRx>m8M=>>NCwHm)GZI-S~^s$ekC-
z+QQ#=-p5#jEh9sf{NuyATBar09pxwWca7L>Ro7w&-Yo;H)o?4(NsQ_pyqSOS`HLxC
z-JfpUE2$WXnoisFkon@@<ivFPx759Q9G#ITCNiu&z+CiHDzzO}{rP(^s;>bdwd((R
zql~y$?A9Uq-o1F1e)Lp4_o{e2;p0s#_XB?L+gg>;i<96L=gK7evOBG3J{Nb)Q#)uz
z3%Am~5i+7MSShn^112}4&=kz4!9%G<eXM=&H4Ua;{D!O(Jz{pz9aC{-4J$=DA9Xs|
z_QXVn#B(hE$?<PE7ykZ?iOeQl4?Xv{bvOSteW9bm$A_AZWQTsPlR>gzJ$w5y6}RXG
zh;4sr0p<(u{A|d(Bhg?3L<Hm3%Vn<YgqJOPd@UzbGX0hXPFxq&<v)M;_y*}JjTH}Z
zm$`0nG0}#@&7BgZ>W{gbebe9Cgg1HkP~JRq8n$~X-)M#M?`-{zzO@z;RO^naT^5T(
zf0Xe?6W^8!nKs|OOSd2Z^=h}2OG9nUM8&#SVjNVOAbTa|wnjHBWpt909?VSYMnjOI
zXy)bDe_^F){r&wf+0e{s_u&i-YIwJ3W<L6CW2xNzvyLeU)vT4a9(@&t?gn7@u2<YD
zp9!etkU2C+NnwAC3_JlC>ZpneH>O+vd-X5le@2u&r)w50K+zhYXJR`=zJQiu2-y7S
zsnSdi7+_N<*lE&w;h2eaP^`^(i_=s|$wi$O9pJkzQ@$%)a1`hBMOu9K!7b)uCGo!n
z;XH!>r4=Bi75w_&Ry+zV_0GlcEUem7G8IaGy^`pc9-jqN05KbVK@7N>bI-nw3d`;m
zHP36Dq77A%Vzj9d1>z9EV8a?wQBycM+|{tlPzytq-fuh5{%}hSY&qs!TwD$>xWVAC
z(FEiPv_{w|26Q$kVYZx2?m8xw*n^WD47;?vFBVO8P8^=niK?t5jo!*ZzbOpu_JDuV
zpJV?=$Z<#$BDMX%PAx|g;a;1GuxWhr2~?&@IARuHNMqweUIifRJt5i=J94Z5Mn9f6
zv?VPq>*|Wh=P2FSY!)RNIgo=s?OX?~6%V_L)9-{4zcdy1l4@v$ft3EjLJ!!lSlUow
z==@9|Mp){lBV<mThRw{Bf_3XZFEq4#IhQMsuF{>E;A=1zUkm6}UK>H)6$Sz-%~$Xe
z7n909lx|N@lF?|4(9xTG-k~+_Fb8dI4!L46<0l=|H@u}SnS7#|k*`?ASiOeV0@y5>
zX(5nX+RMi${;GY5=Ck0b8vd`jN~8`Gj;;SiVSj~Ki9_=!_P<4z&|hEQzHN(OF|8Tu
zDg}J-O|#XNFly5>=@y-&ao9Dkrj_5e`u~pBkjK}(G?&E8c4l9??QFn6y-ObjaztGq
z;2jNYT;HS8C8$Aq$$SKqpd$Ouv<4ouC9R>W%i)Z3<K~^hp3hi!M;cY~AfshEoub?R
zD+|#^+yCWYdC6>eKXI?|ji%BACcpn_9N@f>p&~{Y50O4jCC#o@dkjR|HrDv4UfwV3
z<xSj~xGrUwC?BL@OOrKavyWs#M8XM{m$`v`as#Vn7=-Xut5x$te^XeuI%RQ#$q?Zr
zlI_Oz{rTpHf;hu=?Ro}mi)zN-$t<G4oBH(EcL;>-SWr>^jeW%tr;FY7u8o(Hc96=O
zx0hSAzpi#F|Er~5D>01vYb<9rX{`3MXF7OzH0lf9)LzBV)-OO`s|~;8$*Z=~PRUzA
zElcdIqm<;!4nq?Fm1!>Ht3(O*s7xVNo>c3O7+3Go_c}ZpKyidO-S5t?n9po0c+{K+
zW-V?xRx#)$q%$w_MfsP$_6iEVPuL75hh3$-2Lw%Q(<0bk9)ntW#24`^5n5u!&n3Ze
zfyT1+H_^><LH@dGU66V!tTU-#E*GOJhWUWN(FkA9!d_^9MT*C}&21g`H<aL5aDXB^
zJcZ4Jn{ua{#Hy)b|Nb0MFGvgX0vh|)FT+a>H<eQ}I0ev26eFL5wlI!;NT&B+<BxhF
zgm%CC>qvR=$y2d0q^PUB{3*l+%CxDeWwge9*WMYQ%j2CakJpQq1zG?3B}=S%4mQ)+
zzZ}j%e^OGYU;cSp8~f5ru3%-8v&0J}Fpz&j^-cL0G0H^g6#J15z*OorwJX;72Ecn{
z_H&yQk5K`%z9Y?eL(7`_+!irkT3V3n34#gxXH<7uOl@`n;$0DkPITt3dj=?(f5hF;
z%I&!OT=&hx=S#*xj&({}c3NMADP@Q3@FaMI&4u&o`L^Z$3MSA_Tz~Dq188!<D75^a
zu$%ke&0YW5R8VEClm&yN+qptrMPc3TFxIz1a2UZsnx`+`Q^_Dvt8crrw-uagbq|O(
z{m|&B9=_r#mBs2`H!GSjsi3Ws-o2{r#A(QX$C<EDhQ`)=EtY?MxUcp+eA<>D11S8+
z>TN?bR_@I1Q0TIlBJ9|i23U0kND!*qdTnrv;A6&Zd>otq-E~hBy)Ns2Kk`dIi;Odd
z?(-*W>b3GcvTl8xPWPs}tB!8zq&lttFsmr%^;T{p+-45QTi>(0=KB2Jdakb<D4uP0
zcEEr~-z;t49rt{E@>I#GlK96pCf*bFF{9zq@pqBNMJHFKH_|iz=K%CvP)iE1n5=(J
z5-0Qo>rULNUZU{Fz);Tx$j;}%y>D0L%E#->=gRjhTZb=@8IN=tJv<R{X#M*2q}9tg
zY^0fqcqqenDUCRx-z81%wS3QnhlUCYk0&eN6KdyKZD|7qySC7jm&xt#{E|z?Uq=g#
z&JMcFFLCzGc`=Hvjl%SO>H{iscpGqfehX#O#NLvE#mT$BC>;4s##s%R7GM#b#zs}J
z@5nGoc2w*8LRN}JXzW$E2{DA)wJd`@$4tEA`}cP_GMpx3aor!yb(xtZjP5mYS-yh8
zqlTNe@YcFpx1axvo0ix{J&5dEtlsaOLzq#^%<$cFe{gksNpU%uY)%^SN;z#~o-JxS
zKuKX}^V~)9R>cuRV9K3xr2XwqSID@~u0vK)<A=BCzxV6s&!5G$*fTRH5oT#ut$pG4
z=RBN&J_A^MEu~Uwec$>1Wo^f&I-y?!Erd|l8S~I(<Uvw{p?_-x%4!n}I5e%gJ7l0O
zEO_EpoIXBsHfhJEL9jV;$UXW<Pil{4i_=F+rGXP$`mb5el(TL38V~Fgcu0Nq&yNhv
z`x+fQoTKzyO>QHF{Y&sgGka_2hQmnNW3h~ndAwOkb$5IX<;zBagTN(wCV9h34*Med
zi(`w}I1A=V!72LYyaDg(AHn7Qn<m@zEU+(G>NwWuO<2qubmP_Rgs@3#(fwRQw0$)D
z#BO<u<6=0S)@grz<z0tsdDk7MX_6OlSs&VsY(<c8XE%rVLRPAuW5XAlJs~iM?TJ}&
ztb1oho>+v)n-6U=C1*9@V1B#bdYx1gbJdtP^eihrFA)7>#*FO)d2WH4_0KJ&(64tQ
zX2Pw;Eb~u$^gdn;=wthK3KW^~usy5!UF7l^b05q5{P{Cso>?<}bR3o6KqCmjH>(Q$
z4t-ubC%pvxId3%`G_MZ=htWOxHG1!ThFrS%xg0+PN?V{@zgJ@RpG%frlzHofuOB_e
zTr-4T7~OWF#s1^u_|IV&@y7&Q>)z?V-BGRi^ywsVd&7@Dfsy-ztq@}?lU?7wB=)y|
z^jyP7CDgdZ^nrnR{YH1t!Z)if4--7=kcGXE0~(|P^45VB>y=q9)8mg^P_~^gM;5tW
zCWqgR1gkYp4}KpM2l9@2BV4_mL$}Cd_x|H`)Hwlz#OCSQAUM8y1H>H(baYt17lSg2
z?bbaLIO-G?6*F_@|6D_@JvVf*I{EM34)%hMJ`s@NoSC{5JR+%Q%oiCV2d78((RGo3
zbRJ~FVf%>&Nupw-3w6-<f5=6S?~j;gY`@*xEe=P>;y89t%q!lOej;Zc<MNq))*f#Q
zhwUmB;z%p{Va&&g3*@JCk3KA}KND+h*&Aa)X?Mz_qN3P>haTR77`(!_+$jg*s1hEw
zyb&XDdBpmgcb&5NlO?=+ghahcGm&6TpFVw76_T?!uG4c0s=FkqhgKYoG3eB*dv{~*
z26(l5?nde-w6}c9JTdL4yb2F1g1ACm5rNNkib(e=I(s1k+O}hBhkj9Ti(B4MN)b;&
z<CGGvYdVbVTTA<A@}nc_e>B8OJCj9mTWN65*ww%j_VeuTA5Hi^X#?pgoF&c#q5lVa
zZvxD9|F@5-ZY^{xdx~V=ijqQ-R3enUBukW{L?UZwA(8A$icn-}k)^UliL#WCN)crV
zB_hf0Trb_f-}5^&|Cw{n%$YfJ&iOUZJo7wvS-zjo`~6z3*L7Vl`$6E0{2206OQTD>
zkH|X66Y?=o^2H`KG7{<Vot?15<zVpLQm?j0FsuTm(m)`?y_N8E#-6K8VIp#&jO=vQ
zfy9-(j~03`&itV&7!?%l?L~hBADBa!7!ZVnqy7LPVCX>F%uyL0yk2MX;J^woK*gl+
zc^`~(h}3W$arqQx>x{}r|IwY#7pA+1KWFHD;?Igu3U&DNS2$f@f|=-G;?H6YdJFh7
zdj~yBYMki~;m?1+$^V1TWb{4D>91cd9|(pPcK?GLcpXQ4L7AfFBcM}Mlwq?zIZYw6
zR2M|aZQyZvvF3R*Hu20Hi8NNs{COj{%dbHj0_Tw=UR&pX8x;S4_vIdZ4AkxClr66O
zdzRF(WoxXqncJcn*?mL_lP&SAHy8<Jtnf}0#1-W9X$0NmI69j5ERt+>DVyyj)nka%
z60?DYB*Ph^-W5!*6IRN$oHph7AfX-GaTpdI*#|J!D1i{9T?_5WFQ{~7_YH*ZMFPWY
zunWZ#S=4xek%^it!>n*vbsnEHxTf4393*9>&Drx`{{{*91qiZ0S%|fQQVYKbi5S7<
zBmYa3XAmNg+$@UG33l%&n(m~frH#D8u-Fs>Dq33;#TXhXo#MefJ!{$2`~zFwBy21F
zkkXM$0R*K;yo%2hFh!uCR7v66(kd)<UdJ-bh-lp-pL4D1`(V}+Q3KoWb5F?({ZL=R
zo*|&Vj>lnJk|}I(uMV-VgIJ-e<rgf34-_tg-o?5<*m(jPlb}Zu7zow4D7<Nf_*LtB
z3%0CWDF_4QAf~Ycwr2-)JHbzq>M;k{Uxdid;i(9kHv7Xtfpj}^0YfEY;IjpD)kom(
zuttAH#1}7~5;~a+nrJgdMMzkCZayQnBVuOq-!<S4m7k~l5HN=Q5b2F7CVD-t5@Qi@
z>{D~rPq1<^ANFBe<ry;wl0?cBH)1Xb$mTakqy#UrN3h`%%vkcJTf|~eVu9T-AQh#z
zHCC*AonW`dFTJI7+Hd*P{EL206|5N&+9*;ZrkE)cX5mJ43SYWr208nsNAvNTY#%&b
z6MXv!r_K#-E-vf?lP>l~i;C#ZW~gGbaB09c3{WDh)So39+2JEnJWuBju^Js6MZRLU
zWJ2i*^vL{5-S+yCQOOoaOWvVoO1!BbqS}g#$o7jC=Y95hI8oJ*(yensV`Nbyy_64A
z!Y0GgYD<;%^z`biW6Ywb#z{p5SKFehQs$=SB;L|tnafiL{Uo&4asBfmKM1I&-CW(N
zbz|Q$zQhd*u*%UI>PH2=Q_lyS8<4S$PE~?T2uh_%uo<>c+9K6-HiiZ@fgywMjfvQ`
zLzBqjnCkSSOHOM@Yx7enb%Uc{JM215jE#+Vd?%R|)jFdOCTUql{SLA67#M_-w*Tce
z)6TGTX*C}<$<HeLpYmI3UmlOpcwZQaVQ}P<laZe9-aG}ei2BPJtm`sO7En28M2$-^
z0a*+)Gc%|szkrQFi~QyE92W7r)WOk|k!V-ZTh7NFWHK=Gn`Ty?V14H6Ucvf6LR%4;
zyg`Ag$f8BB$RoM<E#!Z64fiOB%YItqGx8K@m(k6g5QqFG+gONm)00Cr=ll4sAr%rT
zbji#~nw*oVlLm9Z^IilJsrvo-*7vmP5_~mhUWf6@iOC(p+Gd*41z=|%etE@cLP14^
z7Ts9j96&|iiu*Qa(MHn1kVJ?)3efYMLODVZkK;>x4s7)p#z3sX{8Olwwq-`v51%A|
zl~SwE{3klb+)xcAK8pnf1uRr9i@_WGlilhnV|w|^;vgJs`?c3F3u(U<I){KxR1GM%
zKcGs&$QQOrSKXdp#XF`w8iGNzNNwu5&xE>_JB~C>o2lSLA%i#Eg3pRgv}?fh039JR
zeG+R^z7pev@f<I%d4LuxK3G)L&kWfn9K<bx6Z7mwgjPng|M}!Kv0&%)ve@EWgCmm%
zA`S+t5xFN2aA8`uykw#pGR+C}>$}wwL)-BIZ~gND71wlB{2Dkfw_)=oC+I+;#7Dti
zF%O%2qpq9QnT2ts<>B!=qvR4{g6r!rrlvzE`5`Psu>?H#H!%lLw6eW{tN=x{73(uo
z$8t`TVJr0-(P%=(r~_YC6y&yOh|!X=D117NDBcH`F;mPj5yxIZq79gie=iz=ew2%l
zd}-ZI);`MptQUzC@Tg4PiHZ6wBhehQz$szjQlH1?@n#h35Lz2jI->DL%`jQcQgMJN
z3taNZf;Sj$n(6B5PPpOYleL<p)TXrU|EmwZt^&$#FHoU~2Z7!k5~u-7e?Au|+oKQg
z+_aI~0ID|C%fOxp_XVSd&g?Xd+*`r4(B-k*lL8NbR;Vb*u93Py++OO|XEMe~w6Ay_
zjlig4O>TW+E(UN6PA!p^Fh~U}1-775P-&402>w^Y2$9m?N}|JDS1r8+j{#5gl_WL7
zXy6+bzk!(m3c^>|h9J<J@7-DzjZ=r>v~b&JSnKY6a)|IT4857wvpGPuOHQB2iBJhQ
zf@c6=9qYf^kVaMDe2a0lN)bl+oOV7qeJGv*;wiIi=uX5mf#|=WeO|_aLU6mXst82%
zhs`In&_&~$I6s~>kEkPHIlxDS2bz6Qx2EyHjaHXtdmz=aG1Ix8n3xy|2OctmE!<Yb
z_U@{{-h@(q%Pb<rBV0-9Q73`?v8iYBW*~fDe__52$6l@U?{tw+6<nAcg_=IEc^*;C
zH6uA%tJ6;1iPEulmf-2v5iE$z{*z;c6e-8nrF6v2u<Krt)DNq#H4on^Dk>ToFQI+$
zv=8p~;<ZnP#CCp;oiZ%R*SV{0DR`dO`kx$InJ%DiA0jV%+{xfbG7huo<Th+eDadwx
z1gDtBzVm4Qa(4wNrJ|$r0x+bi{)A$gp&>_ZA=^$6yM|j~(;M+1A|b&Cd(Dz<z8+Q?
z-|P4i0^be4rbjS*e|}|^<7?Wyc+<&71<aXr*RXCphn-%Y(Q{Z7_5)l%1d#q^lu^OF
z{L*#;iU1LJb#|uAKLo4^Sm(kc0_XZoUSO@RGs?;2`LGAuH3m)WYxi_1dsrNa3}17*
zShpge%aIpoN{W3{+xp*Y@M>vB67YAm<JEEM{hZtn;Y1*ZM_!Ld^CDtT`<dH(4ukfl
z^`RFy(I{3arFyR;eLw2gR2B!Y=lMYnu(Y&H&pdW>g7&~!EdjC5(duRL+5|8GMT@=G
zttz$JA@gV=VR|$nNNx(0q%w=1R%9}yhLr;sf+Jn7j$482r_n1(7)g-tm76D=C5+xD
zY&$8@$vIOkY<J2~3D|vGYe#brl?l%*nbOf1&G-UjO=m95R#>Bmr*@!Fs%Ca_Kk~w*
zV|1*`nBHGex;$0ZQ8kqM;g#3*{MCK|OHV}Z87ZkndXq~%&ue+PblZ(OwxiW$#1Ads
zUEdvPz>-o2OlmwLVM5q8cMS#L+P=1{V_=aWXvGzz<1WkFn6iCwJUCGj`vJwCfrI%I
z?o*iM*lr&@$3L+jm0-%066dLRfgka2Y38*q8MQ2_Gk5H}!2i??<%~q@rFtXODQ<TL
zkh(QLO7<*~8)(afdN`)>p7w!SRE&{5>s9f^#o(u$n=7I^rxqZp;K>|+-V++SzJ7jw
zge<6(A_{OVa=*2?-f?sNfewH{8cUDWXYIcfc&`z_hrU$_5QVCWMl!x2>Sg=BYq+;^
z?^*eFD1Co}Z_l=jxb~cj)_HYy>u20l3HYinDLRuog1(seu-(}l6Q1|_O6c~G631q;
z5I~@lJ+|D3#jgI0>5}^Hn<`<Zm!}NJ^L6B!4KAk%sF(ZBB>3%4$11{pNoeO5QU9Vl
zNlUR1b1<Zwq^)fLQxF!(`~^XOkJ(8hyW^`713<Y-X}F2yQWmQ$iV2X!28K%T!b$)f
zAdJ(>Xng8`b+iZ<0;4-@(+QgQ$YmDpKqHQrvOEjs8%N@}4SFa7ACeMCTdz<Ud4r=Y
zkL}>`Z)Ja-1+eIZDL>^;6ql%5fKQODj_NtQJE^-rz*nXrWy&*y0oU({zvQV5o#wKD
zZKUIZ%?BnoJ|GQHTE^?cl<B~pwWPem_^SDy<<@jP>pJ8;)9c9%25e`^YF~tFTOr~k
zezZaHTNE?$;3t8&pIlOji=2tVo7AwAI}s6GLA^=6n}bY{{7rcKjXG3wGVwVkRw#*2
zp<FY1nG&`mCYNlouJ{20v1yEv_#CpDEdVK!#*QLPC`cn>Vt%8Y0qNCRo$sU}2E|`N
zu|i{k#A**FW59F6P(;^fU8ZDz0#caK>Q80u<n(iU|5fYobZL>C$sOBb^ez7sf;GRy
z!WLu2qBVIvJjY|s!1Ik=PwTa;#D04Lt3_6;I(VGNYAj;c(WEO=hVSxqY*JUVZVhH)
zp62!#5qgm?pw1E^pCS5O=U2qm!#C6g!*C-+A8iqy+BBE^RA1X>p!h;Su(P^P2-f-+
z8BjZ;vWPSm6G}ZGyxn}@Vh^opY`ck1`3vY4qY=CcUn5CyuQU!FF`R4h1E(35lCgk`
zRdttO2B%EE8{dWhRyf!H0fjejq5WN)KUFbP5{PGg3r8;*Wh@|k_=cF7`%irKl_#Gj
z-ypLm(WwU}<>XN?^7m_%sCL*V9kY5VrwsGnfWJkRLxgrxH3#urz+5cu+Ee_NB(yb!
z2C5=8tm4BJLLtI$C%~o-m8;ojpPzDGeaYmzKKU;wjDkKgOQ)lFqs*GVj~sm3<!y0D
zd)Mv`Gf;$PTP3RT9>yO=Ib%C>i_A1Rhs5;EH2wUhPyY>5r*{e4q-s-y`*KC&E-f#a
z+lRYu9Lz;5Mq_4LeLi6K;0w!5l{NXYlj*}w39EH1ub=1T{L@@qcG?=;yUQ|AaT#BD
zb#*nlc%XZY9zp$a!>aljshi^y68oVTVtdFb?NTL?TDIXVhVf!9OqN`QuXdovC6F-j
z2>#_&c*>&^zFcshsrVDRWZJhJ+qRq=<%&0G1qXz#V!9!Ja${KFqoW|9|3uFq7^Z>B
zj;uRz0Hu4Ri9p@BGX?440Z=pj8&!eG`Z8HRE(isSyPgB+Z5FUj1kd(2YKNHn@!(RC
z32Rmn{tq;HQvg4i&~4l`b%%Y6sM+5?VE(>q`sRi(4cZd17qo3?8bO>SCsNVfsIJ~2
z(6np8A;<}rR_ra-XOmzGbk9Px>@*N2%JoK`?-=6LjJ&+a#s_{xfFD?)D+j7T7oK$R
z8Q)jWB{_a70=-?pHsXmID*&Bs=u%&~H#-^M1O(WH8TXJaM}9HEl=T<7TC@YV1(Ue6
zijLFbWRMPSyH)p8*u{jWPYGYws`UNAe-9V8Yb-+JL1AQE4vv3gd$LqlbOz)eS<h)2
zx>-Bw@sL>8O_eteee6Io%0h+M|1`c%_7!1;?SI=kXOs{ijFr~i?;9U^;i+J5e61iI
z_-Tr<ZZFByy<*^*cjw-{d%aH%soRy+;*AGBjDEtsQ3J3tS>m=m7umibKyvUdp_gce
z%R^+a{<*<5X!K*h_nuSkpO9k6>!^Mzpmzl2w^n3FR~JkWGMfJe5q1L4K3Il$dbH2*
ztqT$I+$!Yo7nmm$LNN$?9W(qbNTL&ct|gv8PT4av_JN^*>gURrV!G2f@+k1`L;VLP
zDJkY%@UO&nd*bc3O~-);i@H^-K@knH1$;iKqf9)<l7S8%jU)EAwQhocRlq~-d}PYk
z3p1O;TO=oOIl95fhgUQn1uZ6lSTO-IfdKw<R#D1D%s#$h8V5ld>ufrXsu&AKLrNa2
zh%UVp%&T|;tyfk!$x6$UA8ChQTg5pv`;h6O!bt63VytGGqQ3vWj@chkRgPF2>$j%M
zRv>}!d-6^{IPc%1vti4+$@FNagxXzHMaC|kY4GU|HuE`X+%TAo$5I@LYnMtQvm4I2
zpkPE-ruynFb~Iv7M|lVp*l6J+MPAq)0SGa04(wy{H|4Zjbxy{ODeLtHz)nBVv8SZ%
z4Rgh)!#&I5@+(-yHVE(?e>@AE1#$1V5oKE7T6T91)7Yf~CgI`|(8#q5-tHSiid;56
zpR*8orTx__&F6TWSB>3(R+S~B81;hs$cNFORVP*6@qFN1C=L`<C-Z+mqc1x33Qde?
zRt;o7mlejpLWwcq_pqF-39FWGBpv#|?p1iK;AsFOBcoveEdY$-tt^XW<CzUuX7Utx
z!YR#7TwGjmR@gAGEeHM1vsZ3t9e4L>9c34Z-@5xf;~JiRecAeiZO#&Sc7zsvG4)}q
zbcqRfLetFUX<BS&H?dn)>g%DH|Kn%PXcn-$_wk%XLCVs<z9D1n(lAQ`pp*67=(tpo
z_9$qkCKI19SCJAIcSA&$s;@r$gHb0Z9$`UH?PXxm(JWR=Z|^B`gFIrGRacJ|fA^8j
z+ZPi0Ikh}aTE<cjO087Dqq5y`+RqX3*nM7yUrwI$GI3C@KTH<YLgV)+D>nc9FrUd&
z2sy(z_>y`_Xz&JpUUwVd&#*%L*f}QTGZ*Rm8?uhpPjllP)<<@pUNQ5)guORfk(&=e
zQwSs|LQ$~E!vy!TaTmQjG?b?*QJGD{0a$wYFm5%?bQ_gBC8c`=Op9-YR|~Y8X_k*i
zT#y#o`lmUL`=%rKzEA(eu2XDBaECf=1g7rOM)tW@RdTNroPCds8LaBs^f|6{+d@TI
z;R%aw3RLHL^Oagcx2KMgKWE_K_K=pcqw$&=9)|<m2AFdZJVY;bZ?VK%ClSL%ejhwc
zOzg?~5)5$<R5daPmA742y=q{d0?t7T4!{c8h@zC*RfLElg@L2CDa%84cbJjM%|(6=
zPEJ(&-lNlo;|WJ54IB-Lx#l=|+&mhU7@LeD3{mXMuhMD@t=qd4Z0nPUV%BSo<v1aw
zp4+6g!`}-L_&r2j>EmBfuvuBw7hdc|!#k`nP~3%c3OwF@7Rf)##*FU0-G>npaHKjk
zVbKmh(RHoJ84zaj@m1Z^NgynIln05Z4-^JHuq&3N!WC1~<Mp+XZ&!f*uETJ9q}?8k
zT8fLNwf-sPlM7P3<E=H(jdJ!<H~L}$5Z}T1PIpxT4i;)vDx&-iQI>9vJw33`=!t*<
z5_ix1H$$l=HD0BPegR8P9Nbk=QY)17pU+*(<CG}prR@fL#81{e9r^Wl^`i|&Z*Jj0
zVcy2x8Vft7Q7}{a;iiS!m+VxbC{O{0JJ9-E!f=Ao0?t<8_yog)8^xw2vLS)k$ieyx
z53}p&T-XCx!FyYtd`+=WpxG_s>n{i6!webOTI1zf-!_0@<kfWSVR;N{-7fg+X*vDG
zOj4NZ=H*b@8lH6=o$r8F<XkFetSr;#3<Nd*lIkgmbtET9z*ek<z4GbckoBNC*It=f
z!4}cO?Cz_o{0!+xbD_`A!9iMoBm8#86M>&x9<o$uwI}(^Z4H}B07NFjwY%eBtZaIa
zbjn$(PQ&#CRQ3yf4*quvWpB-jbjS)GaSAj@waO_U5{rNPPaHlz6Nh;?zsL+(S>g-+
zIm-UB8bzPCFGvw>+j<Ll1ud;Y(k9<ZtIM=xg3xQ2!=rW4E*RQ(h?wyRu5JJ^>RS)u
z14~@HQ&G74t$|*>?b<@Nn8(!(Y2<^S)yaBM8Jka_7IWFY^i^VVOh`)JeCc!g(=8FI
zU?`bGH;nY+0R<<yIF-w@9KqcK950Q$i@h~PnE3BdX+rtMZHe0%5iRz}T*>w`j@^NH
z{eZilW)5DPPQXR+i}9FEWHDupdx-#1RJXF+ZhtvCSo5p?i4>K9ZvAYp$LBYe{`<1y
zidltJXI-mA>yDphGNB+NBU60;$Lh(-{2rbAapP!)7b-4P)UQId-WM!C+d+Aj@CSfx
z6IYxA+GH|sReaskWBSI3hxYtI&*FtPnZzz=&;S_>p%8d>cfA?f13bY2fEtT<WRt^Z
zi%Uu_#p1)>u09a?PawfjcmV7`#!dK0&%GU{38yH^NE7Tg={5O?d|fVm;)QL1YZ>)c
zC~qX)E?%<&s>YAwliZk4XETJ}X6NCj$z9gT3MU177O=XzyGOa+G2Omv@vS!ZFYwn3
zwQ_t>=E?KDWO@oHRC-6^l$rvp09xdF!td{3zNquAs$xx4FbYFaF45&ZQOSb^ktcpr
zcJG|uHEWrS@Bbv6T@XJtwqv<AN~qP7IpIQK6~Ni74mY?!j^Z6~IXj~$|6Gqw%}$?;
z&StSVgUc5L)H~;Y=Jr@15cJPMJHCEHq-y=Rf7v)5%7GAenWlhCXEZ0{RpM_INa}#F
zLE3cMb$}}g7zA5Irf^RK69Tv2fN;>gp82?}6>@6UEq|(;2_M7fUnMTyjphgAiMGtW
zTGQ9-kf{K>NqgTzMV(v{^ChmVuoeMSWRgY7pQscF2LmGL-gh33dXXF>PD?**7G4=z
z7=~ffo|c{^JA`_X&~sN(8)BsIX1xyfreNttiqn5iPn9u>UAdk16M?cM9F5`nx|gLC
zBvnej?lRCQ5buJUp2<4T%p&QO=9%8H(^*3G#J{2)jXyCK3gpKou2IN#R=)m%RpJ54
zefZv1eyd;bY5Sb8|Bs}T5htJ?8n^63M;~G9Rr|qiO_y3}UZgNgXqn7_U9WnHe*cp}
z6kRdO1)6&J-e6S`#0szAauJMIkELgB%`dk%pu|Eh`sHqdlQB4AA82!S+dSMWa^Jh+
zxiE*cIRSnFt`0H}$uqx9rZGLHS>i1M;~#1P$TB;=si61_@FU%gi9oa#(y&(ZX>qmy
zEE@TQ6*tl+$^HT_<`L4BXQ%IOT9>{|$;HlU6lAQ7O_!6imy<f4otXeGIc9bh$tcNz
zkz;7T014kZ)F1UBB{Gw)Y{u@liyHN<<x88t{z=h#6Z32VR`ew3KY}Y3aimo4Vn7I_
z{+hAaLZ^AJ6EyDr=h=HB+o7@Q;lBgi<xO55flnOemQt5~5l~&T!od3GEG<LYjDc>L
z@jjsCYbg9vf`zBbd^7U@Isu(n*7E_^tyQ-h6&Xn`__e8hBNT$R4^V9a<nYdFy#&%_
zpWN>yhtSn&7PYvbAc#FOopvL5DF(NS>x1G4*&QIX3gCZ$vRjPrbhN@n;dY7|Y$AY6
zmLnVl-k**0r3gPunw8Mf3Ou>1#z|>u;*OW{Q6)tyRgl^iK<RtDK6XP(fR)6P!$~r}
zUWCM-6<}bmX3V2%j>$=YAF;Lq*gLP+b)yZ5HO|Ra2^1-N`{oi!8_(T+{$k{)QM#FC
z{!9^cd!F6jL8~94@!~1uz~GU#CRKCL{tch0p9!y<f6!E=_pdJ6a)2fy;AH7)*VU5}
z(OX|998vte;5&z?RfXPxV!{yZ|5iNVS`;=%V2j%+s2*t<8{bfGD}=j^I!Dk|=$I~B
zGxbb$W|fo@*?*?1u+Tdg&TEKJk`<KclT1#2C|`M@@lNbJC|}w>R8oaK`BMn(7`)G^
znko;$)?yN+-pqf!T5Dsgz|`GlPFGZbF}9YqR)n3k?FInciva-EwNg)u7fL-X##<zr
zZtD@49(I`KQaX?pZi-Sp;310zJQk4v$PfhYu@GmKMu-T;SS>={=5XJJrRE_V1RmbD
z7|ErVsKLSr+zus@7{u759sZD;o|a~VRtaRaB8-zl6xuEjd{?#`-+=V+lBF}QsRM0e
z^d7-aqro|T5_;5DSV>_(^a#T^kqoc~WxSWVpju3IO7F#Jn*Hfzi0E&ZLTtd!@f{tU
zW+Vx`t;hYPZt53C4$gF8=l^tJ`M~zK)S?T6WeY53Hw|~8>l`QjR>2$6VKwTKH|UPF
zOT&&+j0x<l9ZD_ucZsF(2MDH@=vEX~qctbsKS$^%CtmHwIDdyp7fr`+*I;C0kjqu_
z^}WUP*t-TIrXkY4#1i<d55W3(mzlvph13fHCm+~<{D9Qewt>(Qpl)p6h`cqBYO+m`
zgyEvQFWyPy4%jBDeL1v^Ut1k@8daDmaTtxnSFR)@Yr7DXqIv;Lu^vy_pPz!f!>gVP
z3Sfh_YfESrALu4Ia@Pw*5`d_XH>2p7quWX-r%sIi#KN|R$f0aHA7NIW<wi;jj!Vfx
z1IxYvnbLM@q99d<_wZoV$y*~M2kye3AUP@%JJ!mo2Ex5H>PupEuG8|>tS2-9cx#$-
zXs6Xn5ZmOk4q?KhX;=sorYgSA^mNN6Ma{(IV5_BfrCK5SIIxv@>y}K-mlCS5kt;y7
zGIli+C4+cHmpxyX$W<VD46^Mu!Bo#tce2XuPE}osp{ftto0JKIG?vDtODyGF$0BSn
zaCWoyh+iuW9)K0O`q-t?iOQTF%ftZ^zLSq%i><#7{3@=7KZsYI>D507bQvMg&sB{2
zXNVCm;|0{!o-O0ynEqy0gxODOV-@7qbJ25vdq^7^$$NkvEzUoR<;i^xkAC6Y(hh;F
zrhBI7xK&1;EMk*KBpV>mgr5*!)PI+ubwHO#aT6slGBPs&8vqyGD67a)cSE>Pu!5kA
z&@Uj=$AG60`18B-5-b{8;B19J^%w*7iOGH+lG`H~U`|Bax`;bS`(==?aMJde=EHJ4
z>Y?3Rq>Tdy3V|ecVsE)829M;;h@XvDuYwUc2Glh=K4Y(u`)^L)@M3Je55-NHJp++Z
zFfZpu*c_)%g|?fR^0>1=J;-?h4{$wZBwKNwa9_OejjtPd%KwxdegDdPciR^MQ0vS=
z;GFB~h=Y_;EWR!s4`|&$_)|I8-w@>$@h*#gLym?c?k9ecMJeH?O?t`UQS-^&9nDy_
z`5tojCk!-vizcMX?jjynx3Va9fKTUW^i21?ndayiHKkh45&=}O`rgp0)I(XI2;4T1
zA8zSgC@K3O!9Diy<1nHbnLZ!_5e9?{w-G-R584Vg80G5-a)UWc&+VKVrX%bBCoO(!
z7R#L}$)cB>5Uz@vnc`&#KGivCxc1fdX>b+7gV(pk1D8}&Q&TN$LamF!z~JF=eB^SA
zpO=T6gKXrk6sYgosH)`GnBkWK#r;$A=NmK1_{6*kehT_eyK(O(E=&A+-_7as4*3U=
zCBKr45MO{8vu~H(5WYUow-uit+9N_XzKZqIb{a!-Wa`xVA4Kk_6?$WtPQa)fU<d-7
zO61t8_AW;ZrAkXnY1t!o>2hG`-~=!S?XGmz{oZRLdKe&6O>^?x=a74{XM<vq<dKQ%
zgIr+Xfz-IFHMq)4Qr&8>lL}?Ml5q1|*)9@BuW6u$k-@>4{3i-dL~gH{W>H!E+q9fI
z%<~bTsRNN~Z3D9s@r$D?>G?tHzWrnDGGf3Wz55JS;$dROBg~fGH-ia?PHYVh43d|v
zz{Tphvi{f@aQXP-r3-c>ynOYFOtSesx|2BMjcIoh_LH8q=hegUhWBFXsMaUKuXS}p
zcfHhBgvkeuH!1}D4ZoTLpi+VglVfs9joWMpy+MoukAPob$x*cI=*B=MT5u(~ya90)
z=K;XCEnt%?pRLu}zN^DPBH%@z6($_;%aQoD`|qW4Bjd=%ST@Xy);?voG(aA-UF5_h
z`}aJlG=Q#f{UfvHwy~f9lUy4|{4Jr_l&e7wc-^e_HN@%&W@sL;BLcH3tiZ(Fyam1s
zz*)_|-oLd23>%H|vq4B!PUQBgzi0x%7xq0pFwIb})x8}~_%WuYAu`yN{FcVTZ(Ss(
zOpIJ&^ot1x4%jA`fBf<8esIBB61K^{OKC6$8i=wUzl<Jky8gS{=<vyq2lTKMg+<Du
z->B$|L&$6Y*N{a+SpVb9NK?DQpErL~i-!CnMh;wwH?RW261%lPstGv(V-DY%=7;EA
zG0^wI6d?M;7MyqxJ3hir-3BR!k~JU@u#4)^dI3SOI8iBrY_W4~IG8^~6+-*zfdK%0
zT6BLN`0~!edjr%QcO{Tp0aR6BO5kr@5uzN>(*R(FUDioh5uoJD=e-p3loqX_03cs$
zIJ9H>-UvT>&|+A?f&<^@Bwlmp{2OikCc<orXryW0jggHjz87Ga69dd>-0mXdWB&66
za)C`i^=g0>go@Q(#FpNB=-lxwFGLlg#04xiMT`vX@7M7IX&*xDuIfM@Sz1}Mm5@sa
z=^o*r!tAG<#t{y!E%4m}j?d><#)^HkI<!^&iA{AzBK<)Xw&^mm5OFa8VUYz74hhl|
z$?wpEcntTt<IxObj9X@sAR^q7xREV@=q{WA#~!1no4e2$hDCTIF&G;lB4cC%Ae2ya
z0F-3Mh73#gJaS^bxY^-ngJ6YF4Rr;1B-3^<AQW)>JvkuKfjP`kV8Z}<n3<Sr6xxG^
zd>X40$w383V2BLQ>iwG$B=qg{cVrq_yGn5O5C*lH@7}t#rI%O#77IIZQbHIs9pa2c
zOPi7VT1|qA4*`&G?~J3Xx~LLL+0T3IbpmX5rh&I>mb|?E3N=*5=0m9}Z%+2H57I#5
zM!^gZ>hkJEiYL5_^sEn88@%-W{Gl=<Z0eVXuljwTcanK&T0<5NhYfpaWr2vARwy$0
zwB&$v9A*mdS-x8Ms+8IU|FxHgI&HS;5yT072YvU<C~1*}r9LY!tbTzyq3oh?n(f5`
zL$NpBE6gSv;vT+m6A~_+J}FUJ8^7X3>c6-E8K2yQ&cP@_*NKWF_vs^OOeDTqphFPx
z7E1s-1--76qV`ld&4Z=^*?q=BxExMxec+qnLaK+FfdkY0xGZJ}BNazOlP#T(cH_Ct
zpUzrQS}S?<56yJRNki`-hG)wxS*J3!SkD`b-dvpo!Gtb53c%mLOQyX}RieHE)xFrR
z++YqT7Z(Q2w+2LrMT0^in^WgHrbv}xE2N%+)v6iUjFvE6KE2_cW=#0W&njHomK?$;
zo+0Df!1g7+)8F)clZ^g;whWZ-bq&}SsN1$&0mja)5KcU3OSgr9fn)jzt>wrEYLv6&
zU*CJJJ<v@w*C8^BC=#?Db?RA{;FI(+{rMyoF|$KCpDo+V{pS<>T7t2}oCaOoD|A59
zycXHG0VN>Iz7I^j8H?!H>j>{Jh&S0DTx>(zBdmV*MIN|+`DoXMyP!LX$t$qBUgdqq
zS>iqC)B#oe%Ei)*E2M(tv!y32VgwhaeqGa2QFH0jth5Vg;LwWl-*r<W*9E!v$DX9k
z9ymu@Og@)J%e@Ol)Bil3wS72-xV_`m1xB1j_}y3j^Si6Z$byZJ8p?2&0z=B`mWsPS
ze9Qt@$5qMuEm#JMbj4X}!(O_~PXS9XjCX;oaMA8J`$k$VSZ8lux0pV#zB|sy3{M^$
z`y`I(0sd*$$y6k6m8Q?3T>VR-A3S%)xn>}GmLW!LGym|vg!=ylp*}L<J42d12}9sn
zBhjRXC$+8=+ZV#4uU=+g*nyKDeL%|AS+*BFIsTDrHaRrt-iXCt^89Pf-v5~g>t1Ig
z*R&=yVyeq$stbs4g$oBXEIQzTxnWS)u0;;&Sb`w8Zvfx^(J#;hksnlFX_6=+X)Oqo
ztrK752Snh=aRjyg15(wB=A>akGyeQnRPkS-$^T8dnZzO#``?#9pmU`Y4a!P!aSgrC
z$O27yjt5TdKzT6u^XIv*urabmA;WVK`p~w&ShhmrBe?@b6fm3#qCS>XQiIX<9&S6}
zKtn}aL;ee!x#G~J#2%59k?BXrS7~?@NBu0>jhtRdo6PVH*2?l@U;J~R@lq|MdhDDL
z7c!&a3ME~e&`#?gWiI%BGy0-~>f7sE5$>VfwJC3csBi^oiZ;8%oqW3&^-TJHArO^`
zI;H`4qtHlFAcgW_guD^vI8u!>J_u=cGhg!7aP<On6sjn%tR?Z`H8e?-RB;9WKMn6X
zTTY?!K_vH?+m0Wz?a$Z9**Cl4KX<k?2jasS&Y~#0;%ZeqQ|5jqEX%F`^E<vq$wYME
z6Lw@MjPmo~M*O9?>dX(g2{?i1KAnxVGv8w-Qm?443R~>(KD!e|stFG%ix<rNz-Uem
z4l=pZzDj!=AY=_k5IF7j=kKWMLE|X-T6DKFLPJ%D+KoE9=ZFmr&Q#!n`C5Ow8Mqov
zgk@wh^|#K3j(J8#o1KX+8+2gfD^ckDMg1XzjG9qdZm|RoZTmL&I)aQ*<;)8dUkOSZ
z7Okc1#@e2r1Y<HHTy0)+cqU5K!t@c9Y(c{jERRV$yC08ZJAKUg7xniXt*owA0+H>O
zC?2o)A<~JhSV=VndUv^U;S`ke7rABbR)s>X_@3UPG2sbl-z6mT@^ONUbY{L&7)QV$
z_-<GLlq-1-pZ>y(f>sjX1)4RTF+){VRYb@P1w7(V5}IvE#!L*b_O7u=vc0Hant4g4
z;Ww*6HU^n4F=G*SxrbY<To7tT7PXjyMtv7&INb|cx2w0eH^{-KK7)feNW9PYObZw)
z3c!6LGTi&%vr8pqmR>rfH1~8kdVp;77k28n2n@dEFeyvbjt&1Vo|g0<Nt*udze2eG
zYq~yh72InEhV+ikPU17s>;la<EWsL4oq}li4!&*(8^8<9KoA78nz}$ZVbad!_bluE
z`Ov##n708+HdM}NPEt{9;^O83fFsBl^E-49gyq6Tjk*(1ZYx~w;X0@P69_di*Q8RI
z^8UfSASy^>h)(-JFnH)`m<A?{mdgcV5>(<z*#~Yd$M<vnnw~lZxA0|f1O)Rku0|BU
zpjUU2zdfwp&@V-sj3Stj{JuUlFfb5OECS+yoe~cQ&=x7XVcbM+eI$GV@%$tF$lg!!
zFr@Wu$5lozY^SdJx5y<H-!NQ&b*A`;v~~&t*$fI{A6+oj1S<0c1TjgWdq)ldOdulI
z^YdV2!Pj?)#H|G;fpQ*40Q^joHWUvBKRN1v6GK{ba$Tk1QQYXwdAuldshI`}yC;~}
z1DFZm3iM>M%Q*+2<tN29@%+O-;In)N+5#;I?N9@VxToG6VKjoSr>_P8FEopIJ27_i
zwn!j@Q2C6DpR5179+wmAx$#h>DJ;3d5{HQ*+F~vU`a5Vh4tV|eDwCyhHmiZCRJCXk
z&V&yOUL^&=?B9$KBhGvY6VwY=6rBzQ@_In9gwI}sC5(XAu46O=+yK4iOORr{pj1@`
zp#!fnqsr!13SK-}29bCJkOC%|_=^t!4h#FnrW*@?GNfkYSbmyye&?_e<2G^0v&@p>
zXPE=%b24&o=DKfuTzy|{M#Fs>@vl-__TAxPbyX|B^J(^Mrt5-dr4o%9^>l~Tw(mc0
zJvh0>;b%%jv-Tm;wF?HvCpl?{v{>bfj>@O`UCjymAEF2Lq_M|Bt8cGUUqLW!K9_qo
z!;0Y5o4=%k+FCHZVfN;|)Y!3|QXn!ljkM0bhNp=v%ww@`8Y|rhqkK`?a^r6#2VNGM
zQ6dwQyRYs;aCM<5ex>SwO=4bWcacP<?b->fR_$<wgWeIiRbuc-i#cg=9gc`joDhAN
zUzxKRdIgZO4Bb>`Y5G%G4)w;0y-yI$usb_rqYO}fQgOo?I3nUF0aQ?!yfby$8NT*4
zp5g|?S6UN9wvp(TpxTOwi;DyQag}+|2&i9=;2#eU!+jlx!*$WJ)4FY(L0m$e{QP{p
zftmycx(^NJ;ybV(6g(`liiFaMc81O)XUG=dA7_K3g-398fF0Xo6CjQ`DTx3p0N(h3
zAD1{qFj#@P6vOF9CI#O{7#uHOVtL#m`owSuaLhw&u}FVpin@k&eB#@1jwMatko@X-
zEfzbuic~v!Lf~(Uz-Srwz!M+g+}-eq2q)r^1iuIJ00Bnh6xvgrq=5;*bivAwfKkfb
z^0y@E`Y~~0Vb2ozk-w^MaBo<qd>7<_LCpV{Jnj$|u}ztgq49O>rCm#7NiA(lKGndg
zo-0%4BP&Y)&%55-hri}(^on>gGr!c=31~>hLByQ^A}(3^ER~)xkvzcYn0pdNKZDZ~
zFyjU^FH>6#We2(zu?N%CbQhnx-Y<*3BBXF`4FqNp%=w9Ue;_mo_a%WI!r|~@1tCaq
zOuBmY>ayQ9z;U1kiu@oon_<T8i6MC}LMCx?^87LDOo8et_3M_sd-u{R4xCB$cZmmn
zVUG!@w6++Wu93#OZNjh@M8A8r*XURvbTP?2TXhNpgAw5MJ6AUVOI*@)`Rdhu;pxqY
zb~??&9L1`v2!VxYn%kFGRaJe(?^u;NpJDTCvhfK$gdxTJ4TPws{@4ObUloix;4gl4
z-O=^n-Y4pxn7zl;QEjmLe)-+EgFr#>vjrX;Zr<Wv4A)tJfzKYy*c#kj%1^5*RQDAX
z1M@3iwl)i;>b#du{U|PMO<7S&{`qNC7c~#Vo*ZF3`Ku)QzZM7~cZah=fODDPYomcO
zlwP-RV>cCu$jko*FCgeUc|qYnKI2Qqp#*+=;h4;DpmdTh?Vbo6w;)Al4_Fa<do9_G
zk+u+%IDYFQjcK(9qM0KbnbUGdfy*@yW5|)!n3KvZj=fEQ3(f}bic=Us)J3Rp!Oz92
zrXTcMh>3~WzaoDcH?6{hhYz<NfBH8`!ZHlgPXLdmRYj`GI>6*+uOj)b-9x<Bzm7HV
zxLzMab*ls|@O9sb>2<dC2QkcdTqNRVq$Mr>3A0q}m&zHs%gnHO)}O}#c79;fwTz`6
z;v57+)yDlM$_I*IK-eU!aXN8=F!1VC4`D~Xu8%7QhX46+$Pe`R1+G{EP<0|0;NHbt
zy*dufC6*Zrr-ge%?ZaU+3c6f-O;yM~=}f&F&j1wjHIaWZjFuWIwkGLfHgML(*bB=@
zXrqr+P;4^no96N{nP~IF1}l*qd2DIHYu<Oniu8hQC==Qqlz<>gBD7=hD6IjyyDvG3
zjr>eu=3$(;#)kPN0NAqi&&)znaOJX;VxF1xS_A2YS@5>(&)bpgrr=X^Kx!Pff1s@#
zOhD;pmn$z8ik~8pTh!N-Sqw8?RSdRVxe*7EVhm0OG)CzUiWr;HV#YuvuW>-T!Ak3t
zDv<VRk!s^Gs__kd>>*0>(>VJhcxKFUWb%dmCYuj5OMJy$GRL$sn{`?pC!+-;JzSz^
z%T2(0gZ_=SeqRsXD-6-B)SChLpM{fsQSx59GYaHwhaZoz+w)^IlsmU;6ijg7qDz}F
z0vk;%Z*IQ=lWDM!4>Vy<0g4u@ruLeg7JPdUFqbu%C)%`NLI~t~-GhB=)^i_2wxleB
z^Ha8b44gbNgaDRRoOXbVJn_!5V6V&l+)T6!zJGoRUDXF5L91Q)%}3ir9gn_7ZDE0$
z<}5qr49g!rLdjKW;f3g^>Q5$2O-tk$mFy*oaFPkO???c$jAQ=EH4Z<;n+PlAWBla#
zkLsxXu1YbWW;mynPcAwisr93k6O%>*p1;s2Y$9VNjU0gLvi6=h04NTP2M6#z`5r#1
z<$0;fcXoK&C8;Od(;W||&l^SHP(j(!IE+;=;)4o(fzq@wsk&NgY6~h6(iJs#(Mk+r
z8`&M{PSQ;Sd&04ahz+GDe`2}JjA2yrdw{R!Sjz7(a~suOU*l(Qz|$aL^EmH>6aFh_
zdI2BP<xn!~>(JRxf+dJN3_j6lwf*jk<sBFZl9Z?2eryA79{7Yb{83wwKSFjrxY&9A
zu#UNr9f0=c)SqhykjNO2dGl8MiZDFiyWsh&%E|zxOj`Jg@vZi8tJ)OU4_cIl26o}%
zT=Ej6EWJZ4o@1=vwucq8=Px)g5xmp1Bo2Y~K#bo4zb%;0&?OF=nojg31}Q+VXUcLI
zI>|E6VK^N+^HR+)C@KqJpNYgYFrkIVrCO(Oz8ss#7cbF(ercq^#$hBvlyynI101g)
zi=lKn^N^8|PC(uoq8!=r<vNlD9(Nv0wA+`H%QEKtG5bxts<5nJ0D8tiPugut9w)JC
zvl%we_?yW-X$2DnhooHD8T^V66pEFa9!RYaEVAu86OVeYd5UbVYjx>=H-Fqj#6Apf
zMA)@$FYM!jzmzLt0kz<>zoR$wz-BxqJ|}F+#NcEPiW?43ab6avcYLOxchPP5by%|v
zmzIhZvXuMXjT+R!M~~sOEV<G#^gKK@I!J4VySp*eZpGTub1WOgpUg)vq7~irv+=iy
zn8`5)&s{aA9xX8}Wy&8UV5&!R`euXr!s`_{J@C6WP0Yfa4fK<X-zV@R=T3HQqb)C3
z1mg&2m-6Kw@WpUkB%YF$`aVOPl5sdH$zcG6CP&`hAJq-U+4o-n23TRSyMWwNP<3=e
z2cm=}E9AzIr;f^I45bzg^QdydkD)RXy(+QbkR0)UVc=qg`g56_&0@I`01OntVSrS=
zd?{GC`1<AOp=fg^CcSXkCq)9<0dCi;FPod25h3n1zFH%KT0IP6EG`>6q^&d2#}zgL
zu`JuV4qILEALxGfCPB8J0$xJOq|aQO0~&uJzEVLBMHui-`Qehy*FVBTMD^QG;6ff`
zUAICA*$myoI9i}PC>)zv@Rf9$3TL||4Gs>{ips=w_vgX3asT7f2cNm4$D+xn)nB1o
z=y>AG_#RgA6NABTcN_^t^41Mu;|AgvT0v3vS0ITHDss?cJM_F#YF`uj5oH4=r|~SS
zP7^Y<!T==^H_lLNk-L-a!YL#vvHaFUq4@5?HtaRoxxoC9%JXaOMNVbM9s&oJ#Fhmx
zOlj02v%}4EM=@|iOv*!FRKH{br`!X)r;DzAFbl`^&&LaBIB&SoFS-IT<~JBAx9rYJ
z97E~19Gg_okT9Ed@iCm%zfOlQwv!Yc2oAK-8*kq#?pj1d4vM@ye;lMY*mp;(Y0*b%
z94H2z3Ez6>_z7uNB$6>u57?b)Qv_OPy+Z?$*J#TlQX_fbEAZ}eUCF@UDMOL8LP1ec
z5v(sFUDJ;NwMoqVEy{%Pn39PUz)z^K7k<qFW<+NX!+s!2R<h+o;QYp6_ag?2sQnIP
z*Tod0Y9bw3cj$=C7fAiSA&wN{cZKr=ys^YTer90k{cg902{`%@<TL!p60mYUfB=M7
zLjq^a>xff~rVi^&v8DL|pgR%>B7-F0C()0G1XEbu9856uH5lD{07)mmD#S2qRB@Mb
zva`|6(Mv)QgyFU#o`nhiefp;7R-QO0SQ@<-skLz{6)^Bvs@{|=2&AeBHxHQ-ze_Pb
zF@gzM#-sh(2U!;ANt8jrfxfA<oq=IWNAdUE**vrY9fe2Fj3rV92PREdR$Fc<z`6zN
ze|aLo6`T+thBN;)W&GcpGXB5ine6R>Q}=u_u0xlQs>Fo<2l?|M>YB!37#d@v{q<hp
z!?MkEw*U`0tIwKN<Kb(Jz&49Ei28Q8axiRWrR_JP1Sd8Q1B)WYr8Pfxj(o-q8OAQr
z6eXq+ng1VHeV*2q2NmVSbQT(0XaAgcu~>eTKfNAKea`b3IoimAW$ON&-FH5qa(T_i
z5y@A%)nmu<{W`+#r^~YIfhvdQ!)0ZXu&V-VwC06~->*3rrAI+F46(+GIK`hJI*g!j
zC3;%Y7C+?7hlm5Xw|$r}Knf*P+Cf+Y>O1rnUx^KeAc~#`n#_F2_yAbAp(g|DgBm<p
z<}b8X7X%voqp0au2uB~$Ws%r7mNB#vIdcNx;}tyDFtbK#ds$OM=rx2)SOeky3~B<X
z9Kcg`C^+D&e7R#*c%)wJR*gXhKc+bcuc3x2n~Ye8Zl}I@Xorwb6!OlDF95hNLNP$9
z_t;tlA8ZEeF-+bX@u-L8aK9}6_3V=K=%*ZXyMQLD^f}lgRFbUolO*7@VC8C0vk!zb
z;FJmP;y~!lhRB7L>j>zb<lBr=nN7V3$&Xg6QO-C=gKAm@rSu6TT(-CoXtC~Yo~FZm
zP$&1YJsxu20BI(Jgktd!Nj^Jrj23PKPNIHDHuzFQ_{L<tL9Caz3|TyWp6d;Or{9C~
z4@4dLvUti!coC{NxMjfUqMmJ$?^8>7g6Y<xb6%bw)eno3ovsSWFTtCFMJCTjeWYj@
zLFA^jqXOVw;y*+lqZp;o0!N~s_9;R{52uif-opac-oACqfheNvB$6){>NvhaI)a7E
zQFLvUi;B@Tfq-%xg+%#$)I64qCbV<eD(L68d&pz3Q-P5EGulI`+%8C?$RY$uJj>3C
zc~WwdyAPwznMjQtSatX5-5tZV{-GVSG|5tXaVwY)_^JNg(HQ6ELFiw;<Op&f$TV7d
zrjZpW_o*%czfO1@u);U={!~G0uFJ5A<+sem2cASR@#^KvRVf2-UK~WN!r9+U9|<S2
zZWVYBGPxHfZc*Zc1H*3s5J9=TXy@UxXnJq|6L4qHZS1*iJU|vePA}IX;JoaoutSt5
zxC3e#RNPC|G@ALnGjjjq_f|JIH}AFIPu*;b4e;)~&ggxbyc`n8@l%R0kx|<qf&pEi
z)ZG#AK+H4+I73HSw-bhdsgC%g0m{7qu?mER%BE_de=HE5)0adHU04BhsMKI12b>IW
z=A3~7&@Yej7rSH$MaMjAjHo4m=>W4<_s7P>kcB{YTm*|}0SERm&GU}ZRC1cXV|k(+
z)lQD^^G<neHVXNFgPuB{QSYn-%nt!3=9W3?2?8-J#DzN0gtIsSnuP-j1SKU;G>%#0
zE?+u22(=Ro8ph$AS;KLRPz6BL<S>fblQt{ka)OKmt=S7i35(<bAG}NgQG8X00pdS)
z5dx3%eUqr$Ch+gWtbMYxvq}986IML+HTGTT9khYQP~Vbk3Oo#X!E5hdOrFIJv|7r-
zpO9yyg`Qx+FxiXu{&@qjbaI{p2-HucVpxP{H7^Tw5h6Ti!adOy=H%os%b_QCLD>vY
zj_Mr*AY5Kq=m5#yY4bx`^cAnP$Cni|)?c_``1cO2Q>YA4ufM=E{(uXHUXD@|E*w52
z`9&g<kjYo^sn0xa=;`{se-{FgQ8$r9v<q7C8(kyLEBBI7TEOy5*E5aQ+*+H%tscQA
zbIU6aEb_zV#kgrVU~{Bxw=JJnBAvbyOYnGRQrn@fM<_@^R+AYXf|$hIcT{uT%$pp-
zn@oO2gBxTzeh|{g;Q_A#=NX%vK3Uxs<Yw`w0Su1q1N;^E=$J*2PMTy{*X>tEoIwRQ
zgd+MH1jlGV($EJW#v&@Z;(ZZfiF0eZYuB!@?NXnai{Ye!MuZk)(!T|Ej>88xxMq}{
zcylD^RP#C_;r_1l^AC)JGQk|j$kGEAFqzb-j{PDH0OlyP1WE1hbYDD0dHg=JFOiAX
z0{?h;?}PfLrJ#|E*FZ~?A<ZJ<98w5&TpI&31v@wy(@b*qLKM*}9Y9O#iH4c@%9H@`
z`w%e^#X6}G@RvJ+R|Bml*$&V`u<dnGx1#lIz-z=$r!7v9UW}jvGcBEe@hcdo7mo2l
z-i&1f&RU50chR{tUIU#<=1#?w&`+TkzX2){@Wr@vU#`sn;CX=13f;G1iZlVp$p+s&
zsh!4Tm^lSc{G<h|!!n;x>F-PIMDruztdnDLiB^ky*98NJ0Km6CHU~)<&oEEp*XO7`
z<`T{in})(S3hx-y&CnEb*Kq%`IDBK|6$Aymyf=(XX@G)ZLT(wm)Fakt7y8@HJQ`n0
zdKs#W&s?n#d=|oQ@gW8ZLd%X52=jqWT>u3y4ZTP2i4TmRxp?TX3p9x6J(&tj&m@5?
z?uE$tj89BT>Kk&q_UZ<l*Z9^P6y^a5?ch2JK7}vW>M>oMt?<!aH;mudECk=;6UVu^
zrlYVzR-x?*e#M$1?oWVz;U}Sz*GHoN>q#7r3KVPemkRjxo0I`fb%JG@*nBi0yC8Jh
zDJyeQv#dsxb}l6y{s=s7!L=>LxS(G;J2@5B^u_oS)R6ZUxhnF-&NiF|cbp~mAmXw&
z<{&9VO*?}$nvnN&IhbL1>}G9y+R=%V%Tn79C{mSy?W2+2V$oy>#MYaA*)zZdd`)ZZ
zCsXj%S>SSVSr+FpSCF7mg69;drgd7uhMDG*NFa}}S<WhxOdPhTW6Ob8uSF5^Ub1k5
z@JF1?tSA0klIS-WP60g5m2c|3V1;{~UleY_GZ&3_STQEBW$<3A*dps#MPHFcDO3dD
zZe`WsX|4a5G?|dAZk!Rjb>!XUjWcL&=3|`5D%phuc^<s3oEB7!*A4he!akQ3HH2+#
z=XHOwplSc&QU}H<|ImHOh>`{BDlf~()o)~ot@so07fso4O#Ym4Q2aK<vlSm5T@EiN
zCQeC?rTS=^$N=2cx7#6!M1!TSLK0aQmm`TL*5<T%^8`O)J_&I$<F1wUF`toHIBj#X
z@s-V>WvQmTNpvVhKs?!Ixd>2C;(RA3C%W7nLcJh5Pr3HXF8-Z)w3+`G#9YrWt+!Y6
z%nV%JI%^)U?xzAsA}ll|M=;l|#;Bi7d60A&EdCcKcPj9*(5RAe7aSE1R&L#CUQpFt
zS6G80%M+!c+6HA0mR?w?aWx6BS9OP@G!RHysI$CE)M>ufa?MuoG7p@)jdSGU;!?16
zT*Wrz-jJ;Sk4!mt-R)v?_BJpkZiCG>xvwEjOnzSEa~Xje-6q0$F?tEQd4nGhK{y8l
z$UbQ+{fiNRhgC5$K3GTn5;SJaeQL0;Q^jYBMshqy@N_UHlPvEtU;^?>Ktq=b4*4J+
z0o`~M2j83X*D5KE<M6~QIJQw5KyfR&)+EFH2Y!{H<;r6OlLKtDx9>we1qs$g1g%VT
z!J~Z9EBBs&8wuA^oyykxi0FezeILM+5a-B%hc}QaRMG`RJ4W6@`;BpX*wr;aHC!=i
z#D6HRP17XwlF@`Iei>&nJ|ttbbrFgWEbrP|R0lYR=CuphMTLE;kc<emB(=t<0k#1S
zfLkowmxJJ*$NT`pPC{>Edrq>VP5+KC{=WGJpX;%g00@NNFJbaK8~83ge3bS4mL!Nj
zvAyB?=5_!Bw8)J>G5ppW+AatY8vuJQkg_t7Ccz-NG#Sq%i+ZHH%>i(NG(I}pT;B2V
zanx5a$6qLyqP7<dLA_+$z}K-$;j?eWi(+DO1HVgOD6J;Es4_^@y)h@;zkzm<Z;f4|
z?BQ(JjtM1MZw@~xi)xv#^(Q61NG=RswD|dQxn;^6OxP>8qVI3}x~6+X*NsgA)-i-m
z;<<|_j-`+V2Qon}prgi360y2m%iJsb83`Dg(#7=62ji%-1+epT1W|{gxpHJF_bJ>g
z($j!Kb*{7#{06ihtUO}&k0nmMD~5YhMSKYnnQBjEkz@XeeVrJn@HPqDXRm6Lo<$y*
z+#XiVnLgwOF*M~p3LN=NH+w{a!7%o|DLyzHZjOZ8IIDe^xRyRL8B&Y!e$oPQIOJl4
z&DCipd2{yKs6k~6^Rko)Kgx4#fz_R-yYJ`eDXFDvkn$^TMmiFi1F2iBEsPfCbsHcf
zvW@acRr+R^fU(QC3v;JK?{9lY$e2OZ!Ci~iqv_(BQZ#MsJfGppFL_+fjFwZ*uY7v#
z+QvxLFN>o$Mi!%!Ecd#JI%$Wi&9vBe)@XCI9lt~u9l8;5=Ee4T;$av|&@YD2aY@-3
zu%7zvM{%;&nz9i#WRx_!@OqaLs9{3xFAxE(oUUkdy-!aEMus7TrtU-Zh(Ykr#PDs`
zMw0~=KX9LZBN^_vr=^1|j6o8|sQoak2gQQL46|L2f&PzaDYM@e9Kln-oB<hd|7Jtc
z9c*{;H!QYpo91WZLbL&14mpO|hIQ7wBH-f8EA5q#5eO?yYsb;Gf<0o^cN~RQjBI^?
zZy{Hpo%x8E=NKdM==1qy;jJH~IGD&L3o6DJm6fzh5=HOwCIK<A!>ESqRw3$AO>X50
zP2U(LdL&(Z!e|+bP1qRIxHpSTXGf)vWRVY3<1@l>-K^=~0ux``H$O#rVAW5dg??MI
zmPKMaC|%HNDZen#r?L|+hfH7Zp>FH;GpLS{nZ{W64N|9Y(wB>7o2<|a;eVt?D|qa|
z+6a5@81qra5EZ0Q2n|%uUe-L*p*pzuH=C#PU}(k*0bzG>cE_0gV;EkyqEydbBzhM)
zNOPu9w#E*f#BBxWzR~_A-U$0xiS~!&IWs7;+>`JN@LAi@ajSy3Ef^>d?SCya7%wLB
zQ2XW&#zp#|0g`?EwBwn6#Vk$u1ex6bibqlHa|)qUq~+&Vtc<EEdEyS(51wjAEOoAL
zBU$O}RREkRKV;3jy(hEfZ~5lMAG7g0P)tl~$9Y&kELH-rfYv)e-;#|EdJm?qIEj#H
zyru;3ZZ5qLyyb)rS}UrEDhhQ;5so72?`c}5NLarM$?9)RD+2z&gr2+G2>?9)J`+Kb
zYDUz%q`4!LF}$nbk9j>1Q)tT;SzT=Y!hZ}k0RZZ0oIT86@!mF|L7+L_@beWGw%qA!
zRZ4Zmi8$C6p9j1a+@%#BDPb3Y(StjJ@zFLD;tu`w=BZ$S(y_JOj1%hu%+HGq?21xF
zdBW$;JK2X#*IQ+U7kOf1<e#_XMQ?TWvJD(eMrpdA$x1_OW0tHyX?+FnmhgGvP8eH>
z)I;j_m--EdxiWHHfaf=(UahU84Vjn>jfSAlp;^gw-`HKaJykxdG1Z`9x9^P<IM=L@
zU2_bzIA&*C&t4O8e@FWpVp}1B;1*#?$7}_#{i&(;Ms2+Q%;07lE;%)M_bdkk6cBB9
zjpK_@;cWM2qmJ6Ej$JKcfa2-;7W5c(uho8%2Q@~G7rsxRG!}KD3aPw>VfpXF{U~4q
zr@D}Kxth9Yj=yS7JO=f3QVbnbziSo#bqWmda{DEmOn#_nTJTelJ4wk1M&yEMZkhVz
zBCCpt`CF4}F@?S8Wsek!iplD55fxMgL$n{&s<oY8%3kDFZtBfLXVGI-w6_@@xbcd~
zBv|DUGDzq~-{Hp{0wQ$cb;r<xM3iXz<l=k*TNGS1P*&wTUrT7Ft$KmBDpI5NQts&W
ze)g4ePn4k_Vsj-og$_ra-lP~wVl)2ap8)L#h?h4;U7GqGnxC1kbtvGP>eS{4_nUMu
zAKShAE%^l67>4zGlHov7yp~|Wk4TxmV>N2o0Jk`dc@?p9ID6=rEFMK__nipAHt+Vo
zkLaApxryx^+t<5ayf9hJH`dReLmFt9<81av@4t=P=DOOnT?ZTmTx;!a7)pS?fSxzJ
z<K`*Hl#cbE|B|IM$uZu}iiS{~_MZYJdyftf!Wrm~c$&IUpb*%0eWn5-X8zzS1R09n
zVJ}x{N%v(QABa|O%-P1BgadC9#CqK#1&GMa6*mndHVSVYT>N29gn#&9O2j?ztll~_
zcMdC~y{N{Gg#oKOXW~f@C<D5JAZKVCm)kIU;nBXF(|OCjsa-U;I03v5)-0^V(ITt2
z>4B2z$$m6cnSe4-aZRBO|AFu)nse|}uHmCkh}E6253fT8rrr;43yD#ZTf{@+aan7c
z*5F+)bOfV|5H;@Oi5{F(_nQ`<<{#BlhMszKQp?lwOrvK4(!AlIC01>l?!VLkm{O$V
zu~l=W4CZzV#=<U)Pdee3%6c7U1wp;QgEf6`pQ8}Du+corO3lt60Wq5SIN+ADVVQHB
zVnpe~Yu|&g@0N-<D}NscXJY=j=y|qa=^?X{)C!OzZTUhfq;S{#@#*zFdnzg`Npm&S
z{}IQCd9q1gc=Xl=4&Cn5`pnHp8G<1f<=Iz{-Ig7xlbkyf5R#PFx{F)RbbrN|1g`c;
znX-Y$yGiq%`SpTZ1+|aOw0>8_>8Lgj3En-f5SO<#srte8%vzJ)U%#r%F~&9b{C>aN
z`8ZHb%%cmiXC5MCU37k^0BOup*DB5?KFUl?CIt4M_nxv8f@%bI?I7qQcFlQfo^TH~
zzBj5vNLTYo`p75JuOj%+gnLPjo0DV&_qd?gXrJ(IpuYeD)E@SNw2Vu?Yj6lK@ALJd
zX+aH{7Ag<cz(U63PR9y7?0{n{b9qC<zt3ym^fK}raAHh*guX2*qntJ~$Bj7{HPK=1
z=BrvVLW90Dnn!`7wn;zQfEK>{IVj9%sZo99V|R)HN6&O6=Jdx_Rz^-Ms|_VwXj}D`
zRXJ6`QDVK>pK@{j_t?s<3kaJ_itOy@K!~4{^*weir?S(3KD>j0jjPTviRUg($RZXb
zqC22A+JyZ0;cdLYM<^R02)6?Y(OfQN@C~_!mYNZ&)S-i6me%oHTy{ON*CJ~VX_nnB
zkIm?BLFH(EFzoFqpb8VH<io?C7>M`nyuG0B+0LZ0s2C31wKu4D(U;S{bKqstKJ#*L
z!Kf-AZ)S}7aFE)jE}#Z5_r=6}fAer2vwGPvQ!sHm9JJ1~cB2_v*SD~&u`nO(9WcKH
z`2-7Do(0BV;MT0Gxx^#wk!`u=?OIwX1#Idke~i=~X(!P=+aoq=EDz817hUA3`DO73
zrEve*B#3Q%^OY7n1FgO(a?LoF-SWr`pjj_``~W*cRU_i<jXiIIIK9pt-@!G#^Gk6<
z{c)t#cnAS7%|yXL5E#ndX?!5Hz8G3YiQfn{TLaK7eWv|DVMxS$u(akMV{Gy?qECg4
zf^^OnyJiT-P1xPw;EtC26-VZv=fBi^BYpvZCDTM}WGa~Vh%{TvwIXD^#!;zI0muLw
zhzP|seSc;B<&@V}^Ma7Jo>kSM;jtOkrav}Nyk1$itT5dKaRI=OS(WN!%!{DaHu0T#
zM-M+g_98*<N_cDf!Q1e-d9Y?!e)3%xO3&bpQ$WH{`9vf{@iO@lKd8d_15gun8CC<s
zl;2qJ9J=w~g?g>!f$@sI!ZS_1!%PWDr|O1abGxj6t$EzW&9e{Pm~${NPickPII^Qq
z$bmhpE}Xh3<$Oh7_IFD3HFP1)P=vyC85rnZpjrUE$jWMBEB4bvcIw#AuN?q|sJgv3
zFJL@GMVMs&pphvWSG0h2t{+8xnmqBeNRc-E1o%<=uHPCljo<Gf4syyzNS_BykamI1
zAl*8L@>}wLYXX9k?XEuFoPLB_l%AK&7nglT<F;Hnkh+2t4s$&sutHDDxRjg*wutFP
z$I5LdMvMHB2Jr_}mF!qgh@I4}-kUjPW_)>QBdobL7Z7faclOe!2xSmCd}yS?uw7tq
z1oNMe@1YCOOU$d>dS3idM^fWV?!7{*#>OLPT|BN+N$<G;ssDU<sBdE1@X2(VX^(u^
zb3*dY+z7K}_OloGg@%elxj<hJQ3nsC1cUf6XBTK){fb%2*P}a`n1}pOFvH~)(wq{J
ztSoc^1ENy}ETpRt!H4fzIR~xTLA!R?hUl-GLKS1SUvg*L)z+V&+H~Xf2OmpNaI|Wk
zqT!QWJ2shz8(t`<(v!P!!72_W*m;bjKw3G}jH;06M?`?_@!fq8^J!{xkCh<GR&BbR
zSHJu-+6eBZy!}tk3NJh0W2kA=+wXw>5xAjV^qOR?p^!2}nF!eSph2NA(X=CmM;KPH
zHHoorau->I^m<7AO-GX3pNG~5?4a=(3aAHmZ?kr6zp0jY*qSBjk&^oioif%6@Wg`R
z7XPQ8QE6?Mc0}PvG|Jh0g078Zz30OE_yUTRU9qwvj>2^Yx2nE~_JPi>wp_ixE7~MX
zLJwoDT}6fOFv|(;xQwz8$R1bh`em-;H=Gr_6VF*WOjzJI&v>SS_MU8}tm2l$dM-Hz
zp=FlJS}*{E#&_mDc77141gQ*DE=ck1XPKon?%DJJMJIM1InQGzV*vu5oj*~I?mD{z
zW9C+>fTbFVK0w^G)%8a>yoShvyl;Jm++em>$$RehO=0aI6P~6K@Sr&^XDws#r%hLl
z*HR{L7;ip!rvCys4L4UhEY!JCEnT=$UjiXIWM&1PdmFY0vc`To1(<{vyHT>dWIspj
z84T4K40TDlvXNtEBBDQ(zUMJ$LqyA%eF^C+Wa^H#olMa2S5>@QJY5GFR3~OL91J9}
z>5OQfcr|jeLI4P*FPME?=kYBE+&sohPAEmVH#|y#ZE)Cgzg;?70o`g#+U#p1*}~#G
z5e2V)>g$J3*s9SUGd$lb@(x5R{-SaX%t3O<?NF=U+*@`F?M=$@2{`O<+4{Ae!oNS(
zOQv6_KHD)9qly@~3;>E&5QUH_g5pz$ueMIy%ynqz1908(e5Dm64**DsdrKk*6v8;F
zwDlf2ECfv`u%u<{j>v&O&M_sUmISOGtrTt8!;1<5yoCLCkeW5)ML^~c!T(wWA403p
zD>2p(7zKz?xby;W=(4+mS%^mlyA-Vf_}~WQ=jRg=kKT`f9qw-8jCOm}(k*(SJWg4_
zZV6&ufHN)h)y)O`XL%+X-PC;>9k&o8&__TNR-aPD4@82yUz<Y)MB6vbN5bTMc%8K>
zNXNW#>!!-BqW^O(Hni;;BBhg7zkF7j?CMeu!iVN0Yh~X$HY+lz+rg^t4<H^qW$_!u
zWBGUX_-JreB%avReOtc?mhb`oOnzHHFe7%N-9v;lwG+ia3_*k4ljViFnVQ>)P5A8n
z82ea_(<)?-Y-4c_6Cmt%4=a`?g1(9zu;AK$S+J$aNTQ>o<FAtyFJFF#CZi(|ZWI!=
z0!A`!LwnvrMIh2QA1*Sx&zO19^!TrV$RB3=V0VL{CFnH;o}8$g3(_5~Q!o6uF(_l^
zEkNuM8Zbp2ykPE*dbGE^V-6^$B!+HxuNEZDOI|#&@f&OE+KS&*_Dgad1+OP-j?1^V
zs~&WrhqChY22`1)Oa`inO%~W&Vx^A9I<;oFz>$09rrvu`nY@(F3g=(5n?<p#_A~~l
zwtw-k_`~aoYxLGWVfp6cTh~bM>6ZVz0<-r^W-{Ig%jFl68fS*qGHNJ=t3*$uzt}EK
zOj^=be}wqF`5K0ZQa|)dz}!=Pdh;7@9e|_>lW>383f|Ii|7r2hspe%={WnTf;+(z$
zB|3utV1LlPptb%#mS6jWi{Dy<bDF(y&UMiELnY9@#9qBhR%&oQIRX6w&Kx1jh{zN8
zNLlC=!_1Q5gcFT3DmOfYLY#Kdj)s71?F2>=EfPU^B<|nXFDc=sYyaW`d`pGS2opk@
zzoPi3mby&?BAmUrZWGVRr50Be$n?J6Wx1&8exf}ii(M!MXa@{kGxYksu!a{C>w@f^
z_S6BOQo<O0qJ~B{J*y@#3F%TRzCwT<@fCm>kQWB-GJs?9tw*6Vj>&Q%0H}^szNjB&
zw59PNb0Kad8APBKC6N>Qzab<|qSB+PC#M?(?9~8PaSy5a#(!k?e=6d=hRG3=5$?uv
z8_>UKGZ2%EBaVSsvC^iyL~0*y=@OQC2PT|VoUtMC)SRF6y(3K`VUq<nG=x{N!gb;l
z;7_$jh52}{WKJOA$MFuzAEt>qOabN)-!(f(;5+P5j)Q~Ip@LnFT>x>8BdUjBwF_bz
zxjv&m_r{~g&4TPD_kUyW&BLkw|8~*Us3MXf2~Cu_sDwx|PthVFLWrcul&ScJ%#o=K
zAw!vyF^Wo)DWW7KDO09oDhlU*uYS)t&w2KB_H~_oJ=fmXIr~|E{J#C9wbti7yoURA
zzwQ^F4xx_b<tBg?i_`bKM^TD@QxG?f*b1fTnfQOXy9ir=0;VIy^G|*0R2G(|o4izh
zaMzMt?d59rY@4^_S2{_|;Vr&iHVh_nmEliTn4Cu}O5(~d%tSMK@L?%{R4hgiJ@vj*
z?)t#ohenwQRtc07WSW733D8H8y`czumA(5kWNzp7em9cjBHwlU+qy{R16=}rEOpgA
z4kJdXH~{Ki>eu_D{h#$9+2M1Gm;XKu=S2F-cP=^-oCd3b7$f9>QQc1RTR;*?N^_(~
zEd{tbtl2bn%tA%uyD_mNWJ**JUBe#$4mE+$HwKO_R%=5Ti&&p8CYc{t>Q3c60c&s9
zNro4(Gwp*23;~|qFmW}E{H}+R0C1Hn?t{!{W_x2~6p@q*?1>5}tR>e<hd(-1yAsCx
zrKo&VOw2T4KAcIcl&Ti5=c@K}n=q4*-dJPj?4x-U_m-SV2p$J|?nUwM>ZbFw;+~zT
zDgFvwb9Cn}5fC3Xd8gEk9vE~ojzL^jHv_^`IJdGm1M1VHJ(L@0{FR1ju(0A}+#UZZ
z*IOO^#{}ps5+_7ROuphT0<l*_5*P4t2e=uL4EK0X{gn3rOswk}1gr-$aWgZmk6}C%
z@vZnVVRhROnjo3%0Q!*A^{N}OIvMIQz-{*6T75Wn%A)n2sW%8G<PsoU>;@dfyX*R|
zor3p?tP5n`4j>9(xd>hLkGR;2AnLMtuv{{RdCRnvSH>vq@8ISOP5msxr9qwHciZCy
zwyDME8J4hHgwW`$fq0C72C-t6foO06a6o9)xR072i_qGH-Dn4OGr^3<L$N~+0b69i
z^R~GdKBgD-0P6m=5#&6gnze$HB!jr4F<#@9ywwa%1j<_#JF-)jVr2#ZJ^>b=)MG&G
zC6rIxSwN$7Tn>j2Hef_MnVHHaUHuOo&o7ia(JP<}9(N0J%R=zukU5e<0-QXVA_}lS
z(vla@8L6g&w=f5?*JYRjlC>gw54m`DhFH08-oikIz+WI;<yamv!5bAOoGZ}9Q5da)
z+Y@~G`2`WWBJx*Qpac}<;nHxWjgYgT=OUSaeGp4i=HT;g134O{X%yH<K~^hHoV2jd
z?$QHw0MR+(jd%DfbeZ=ce-}vj{$v)WoOE~rUQspznK%lM{e|eywesO`&k;JT>ZHPU
zQu9U@iR*F^8&4F{aF9r`97Yq*{T}HcZVb^00slys<J?g)10H7+SOFaR5O9-m!1eh3
z3$P8OUWL*bOe6%z96(ketwJl<xkb0kvBw2sttxo$-$KJU6c+o*jfGYfO%xN;xPWp2
zaRg4oDPiaKMxKro`!M22wNN}}mN}Lh!Hj?_2@w&Az0oQax5pZ`UEEtF1Ycq=&NeQ|
zUv8Yr1YapDIrrXRYz)Sdybn}d!QZOlim0K3<uZ?kLsCk@fiQ+Qh1DRx`Db$;JlYf)
zG=T?&ZS3xSrTH6G=GzR&==C~k71YED(D6Lk0|*sSN4DrBeVnO`C2h(#cSX?>`t_<m
zo!xvSTcEeIQDfnnh4`yalle;Db7H@v_Nh?e+c8W=>e>Huu?IqO8W(jI`B0hozEwV!
zmQFrlNd}9sJ>dH4kcvO?BeI(n9`EpMdbq>AW_jFV*3N3p*sZm`N+vaj2*2{nLkwD2
z(=c;s<o+uJQIYv>Ge?Nn%z`yo1c=Cz^Ub~hHdF9ajSwLGm`dF8W)T$lgXYgOMBw&l
zM$9IAY-=!xq*FbFQQ?92F!P<}rZik#X9(JfcRY-66G!5tb3N9h3)!Ue3l<@K<x%pL
z5md9<Y(zEBXxow*p2&dW`ajyT%%{a`m+_K*rsbp#VrsT0AJ*;F`HzYMgjtYgI!Zc5
zkb<8H-KpQN!qK5{MY9}76pdkG6&c=(3;yRnuF-Lf44A5;y_+!JjdU(7CrrKWg_*L!
zSlm2Wgq7cdt?h}Os`-i#><Gqmf&laH-xTpZgzJPRnvnSHbd5Xi)3U`~^<Kl`I~Mbo
zV04`N);_a=@<A6kfdCid-2=gudQ)W&Pzof&na(im$ZrVWT^z7H&XKTJ!17BP<M*kI
z?%f(DzNoA1w^3t)5(EsJVwp)}gKuTs_IFv99#XelxSO=<AK7TazsG9}T2T&3&qL&X
z&7umxlDFK9YCrOqL?|D;I-~e$A9^Us``~rx|C_X8(&K~II$FC6{;>2v^>I>io#`k#
z%So{XCE>T|V%J#??-A08F3({L9NQ8Y3g-aIiWtNp->xl5NrBNiZEa2X&7}Q-0gxj)
zeZO71Qu*&UgnVPla_YYg^Zjq0!7Aeb#Yd796k&Gm;Jvn-gW;Y=ZlG93WbBYls2`W1
z9O@vIjjpW)KLtSpd&`Y|eB!+kJT}r|(p8AK1R!0y5ab+YPXBK}il(jVfA0SO*PZz9
zT&(DrF)X=HWg`(chIQRJxEr$tmx_=q6X?{HIntAZpi@=LfkGMYu0z!9Ohi>iOn_px
zf_Mi1o^JaDLOB==_s2S08I5x`v~yZ#gnB3D(CJE=ArhfVa#ARYy5zRBq`*dJab^RK
zA*4v!KUqpTLVWg;>Ky<Es<<BhN!^W~tt-*lx~&aOjx8y$%~?i4eEuxZ!`KolqKx04
z0yk$DXrj8+@<;>AK@y36j{9?g)=i%N3X|>#cy+^Joi5&;{~WoZhdQFYOg4}<AN2%c
zU44p4aT;TS^+xRZU|JL8yf51lX_%$pOW<Ss_oIy%+2HY!w^~XVR?mP^4tn79MyPHG
zJP|Y|beO?N#FAdYh$CpsYO1Rfd>I5XBw<3=5wV=V&*p54(i1QiYA8$IAX@v{b!hLO
z%dz4Iat*@P%wT4qP*`bwyy=EmI1=rp>ylCx0<0J+6nxfc!OK!qNf{yuWO^m(7eq5B
zdLwz9E%U*8?Dx<u0MP%K$5TA%-oHOaj{(Udcn>6lJxWTHt#0`3!?bsL43wLodTOqv
zMB~Ovjr_D`l^Xji7)UbY=Z5QUeC%z~w`1uCN!k$rdj76ZXwXKKU$+HTspjtAzeMZ}
z=!UI>NQ+yo!;>h|Rw^#_jf)8~eA2miLkn*fp-uY?)}4tChq2Y`ok<!$J5lul9dCDd
zQ}Dg-@S{!uav+Pds^caqp7*vR2<=^BICX!LNrVoC@?g(Qgp>rf`HH7`Leog{MYi6L
z@>qyxjUE=Hh4zRw!bsUeI`i=*4;fh!lPueUh_<k(h&X~j(u9j^Tz+L(oAme**?!2$
zSEMfWF)c$LKn4$>pGiErq<Km3SGe$Gi0F~dHj2-?2Y|2h0qLN{Wn&$Y2Eq&X?AGF|
zPnYQSURYq_Vp?}daoxJyD@E@sm2$p($~!jv$x}ZsxYQ`m=(kJCd_{-txjdH=K|z<!
ziZ>POj^}yatDwm1M95D3-u~D=zJKM7h=?uU+e~|EotiE=Ui^p&m*oSrkHP;$Lg?>b
zf8G*ABu4(e{_Dqo*WrKJn)omiu?k!{lH19+PXHPD@#DvtGiS20vUZ((<c4w*a`ydp
zA4m-obi)S^9}1l*K;pqE1|Xqq%KCo0YG|a&vy~HD(-uxl_B?h)k_I{RhiJ$q7*n|;
z-?nXE0Df=Wxbeb;3rv{Sg!T&XWsz8Z2X(;c04~hL_L35EI%<~5$jE>?D29#?*qC7b
zaWD)gCIk_|BS|Xei;Igx@=ip6$gM)1*O-C*<&Wv<OMiadudM9-6^S&iyV7(2eucwH
zGjr2tj~_qIw~vK|1>&V_n>LZQAT$pIT)ld-!nwAk=jk2E0|(F?GlU4|4{9`+4XZfp
zMth-|<_;2bG(HE0hc6=6!`+%jEN0hvLyB{cxOmH3^3tVJztCP7em&Ox*c&+%l0=h7
zZep^svMR}IDL(t@X*!Nlwj~~bPmv-{j*suQm8YW%KqmsQ@-6)QHUP#!tMm8YrpUOK
zWA~vQ0r}J{O!li8^W}R`L0)rnbNU2T`Tj{TW(avGvd~Y3Rv{${3Zfe!420x4%;<Sn
z*Ck}wK<>FvjK^$8t&=52dHOl6qN1W>V`Gr(+2^)jOzZ*@f1s^v*RI{?&PJg;QoH#M
z(ujws*s~+a2j>B1Y)nvS8yFlkNK;<BM|s60CzSk4*#P!C?&Roo!UA?8Awr%1l9yL5
zH@#hacz6?#I8eN7Z_k*Rn2_cxpnb5mK;pX(pM~RN+DepCw&bH@i>QbH{8{F@r62_k
z;#d|{sp<$SgJ3$x9{52KjL8Oc2H0m}YN{B&a|VkBBmqq6`-4V95J7@st>=v0Wv%6u
zUZ-^UM~b3ew%#|ibTh2F^cBll6WU5~%0f{}_u-Zkcy0*$7ET#)adDxE6bKl5NxY%?
z63Xlb+LkEf=jQ{G<)@|xYTg<jA4g%7008h{_zV1FKK1BK!=iq)FQG{B(rM}BatL;e
z`n_u(`c(O_IzV}8e+!@o9Go{DM-5(M!AGC}L1pqi<LfI}N$kT2Fx6-{phN#GMefbG
zYL!8$U0lQ4g9luOqNC2dqA?Xv40AMBLf`QYB8Eh{N<>icOlm`2ApD>#yKT!BoDibT
z4|2bnnwo%s06?-zw6g<YAr%N53se<I^$_2`e4|t-T6O7WH)}Y9V{KV36?k16Bn%}v
z?vbG(BcP2$a6wG$sO@z`GJU803SvQXB`p`of&nxqA~HhW3bCo^_;}9En>X|D5H$~?
zxAE##i0JU(;D<X}n=INg4{&mGb7OUlP=^6PpO4}VM)8Pi;Fj;&z5C4vWa{ImKPRC-
z*~eP|T@j!`W_%*xJthuVwVad&)`NgRUq|N+>g_;JY~FJ$9!1Tblz}1Oj-*2=nDxiR
z1U07>CgcA7`yiT0!D`^k0VLOU%F71`3NJ~&YSsAk^z`_6U~n+1B%eqv+pc{*8(Yli
z7c@+<O-1t`t$y|DRegO9b^)ww*dQ3U)vH%CZ9e!2D+5dic+n&*h0}^%ywb-ss8p(>
zW8v<~gD4N6R`|m4DMt<nogF6M@fsZ+?d$6s8R3FO;pV=CKD9S*-XMT__569SmlDW8
zm^w{*ZR%0?LAMvCI_)wjFau3=&c)HZiIkDCc?eM<rsd@7dMz;U0m4eyvcrcDpR%xU
zS`c$O=Q+g7$7h0NMe&^={q7T%n0@N+vUKTESVo9~2y{m)H7zUmioBmlSjP~4LMjM3
zEMaQ8s<H7$m}FSk2Z(fmF!I~KdK7R9?~a8=1OZEBnN748U?-?|Q$RE3e0cI!XlUpl
z?FDV7m2`C4Niyyblz}GSiZ!h3^5hU&0|eh#f?&r@!G=P43Gwr)lBRBouoa%4y7(LR
z7!M=yeF$b}OGQ5Vb?SE+V1toDp0>2K^sIwmh8>6-g_oogy?BvAVh%@?k&>6!AGGa_
zj24x@73<Vl`{?pK!Q@cGish3?2@lscGKzP2`uX!`)bivFlH%eFv-JeB-eXm6#KiC!
zgF;vp$)u1$?W($T;=yOch#5o2fia@Ak@|F4LsQfEfy3>xva;F}P3`To;HAB8XxQua
z6HQJMHYx|#Z{_pUzl-uX9V1u78QY28+Ky+xQTnW2gDtyZ-yPbiJB*{8D&h>1yX|%0
zxN3Q1uDJNH_;2?osbQO@yKP_Af#2C#b?KqrI4&iz-;rxieN#A^<m;<oXO}yeZFNUw
ze~*n`mL4!lH`JHj?JUXK%(~>}4bA-=6j-3$E6zB7*0Qp<-xYJ9rlBEM=k}HMHk-rj
zAD%un$C<3|z4Hhj($m8uo?Q*MeEs_Mg(fd>>wEWnQ(lpa6F3Q8Ox%TKwwO>V@BEa|
z70-3(P36H3)bEAYHt*Sny}aJN15dKSqzyG*znIaH5z=fqHRWt|;z(Mv4crz2UCZkC
z*umI34pPutIm#cdlKdyB6&;P&^oDdyRaFDi($emiYZTdQi?FP}fyaq3vH+z+al`#L
zWHnCkg%KIf%*;$X-$kJLePuyPOiKPR_k0yCDJfgBlr%Nca`duMhr64e9w?fSlr%Ox
zobaq^=@SWPR$A`fk6`eL+@JY*_%+cv!ni!&U3IJe$EvK>AuIv<bN6)K?p_H!2DhH-
z6>@TN7Hy<w3nrFTP>;Ve9){pViEYgNQR^Jl>)UW^CV&3iZ(o$Ur{ejG7u=E#pCMv^
zrC_c);PClS+)5?6hsnvNuwIt2L*0)b7wq1*@0jGLx1T;8l$68=_}i8+v+N=DFbvJW
z#~-f>d=H$st@PXpiyPwVrJkiQmbY);R&S}o--31!cJyph->ASYX=$S&k>+sw=E*~2
z$(F*ey1N%)-Qaj%m-geR<LmQJ=VN91*02ln8!e3YC=}frMj3pi)BbOnt+qvrNN3D`
z)1u@*X|Qip`*YC2+Q0oe;l!4prwB!^XaR^goX=7WjmABD_}~G5t_*5)a55a5jU@2C
zkVy1M>racoT_^#>OJ9$Xo_B72v;fJb`l66t<`ySETz-YAZ(gZEjv^&ST}ZB)cl$*N
z?=pU%Q|wrpn3(t;nl$kJeIdp6-Iq}Yp?;6Qpi(nph>!&*Nl4Au7P?2K&K?&8%FAVo
z_@V~bZi#oM4*c^`uGXLtMV8+F{yPPQ?^rcs53;g3y10lyeRS}|PVJ<<zlZb$TVaP<
zT3Qsl3#aVu;P+_iuaw11qmpc!a^<x9I!V?1us7TQSdSP%#-|a{Mgv=TCO8r&huWQu
z;0uuQkTwm(Qm}f4D3_NQnZVj&I12g3Fq{Znt22Q1$BzmxufudugJLiI^2%Kdf!EC2
z)U~w8R0RJ%n2M-5?o@o(!Gi~@9ebY4kWmE)(o(~g)6qfIxflT~;sB&uEk{Pq>)Zw5
zX6x3ibpa*`GbaMwe~w_l7Df@6uvF`8yHRFcO3dGM?08QlyDU?0P*kYD9-w+KVX7TN
zc!KLlU^n$M2;GsxxCgNY7J7Q`5LPR7pYl+k9J*}TC1pj~68@!r+4vo(X(SJN&!ag|
zq*%Fca<QYc6XENzp+|%cPFfbwHV1D;!gXYR@K!djdwVUL4L783YxY0~9^#=(moD8(
zO!T^R2~Ob8=cmiE>M&1;dA`8h+#L4wwy&@6ABokgR;g-fu}N&=;!2EKoB8$g=XK1?
zTaA11t>%ghK32GB80-`kJ<_jo1%4)Q<@SA*E)EU_1rzvC24(Bc_I9$2b#nb^*Hjdu
z_m+w1{>B;BPwoslAx}Pf5EZ41&<@3*a<Iz~ceJE%7kxMM^!$ymUtOsL=G4c>2jwm_
zuk&x&a>#kl(5jv3r5|VBiMvV+A;k?kl9AVJgE(7OMuwp@w6Ag5d+Az(^XHxOe*OA|
zsXCw$yazJ`p%A*HZs}?DOfm_3@CzPKRa~7F*WTxCy4M}Q3kR=$iR<FUi&1inl!O-_
z2Ynb2X_-N;@=1dF(310*MiptKn5Qo!Wv}|$BuD1h6q?~Q_F=+`qIo^&5Z65pB@P-K
zjgxr?pANMohM^4t)LvYz)M;%PxW~7O*w(M#W~c95<;c=9NZJ5QEb-WLtyO;}G@5PO
zr0k3l`_gm)V6UQsWQ2t^ACLiKJ95AE_s2#@^XAG})v@}<!(Fn6S;CZ2%@M$>8Z4p^
zwx%rUJvQC)VAdJdv38huRV74d!sbQ5S|hS|HJBKk_p}wzG_H8?KrcuC94cetP2Z6a
zWXgZVAv74~wP@b!VQg#+$GUb%!>R?w;Iz)$)PU5(Nl8gmxj%k>el@!ip6#8(wdI|@
zfL>Uej#o>}>S2S#X?elt&4`&&`w%aca|IECuVVe;udhtLf9}0Ub(vk%*hWv63ZJwi
zG5a?HKx&j6xS6{GqPAm7fQrH3z`)Li{)9uer`Uaij5TkzDe?VCX@+@Uv-2_gtr0$1
zS(BR$G4YumB#<x%FZif|0qVlI$f_dS?99<2u|`Pfu0xZ$?2T!pdPqN+W1AB8YunDO
z-kM{l9iv`i@rrTY{uJV3z%N9K&MK)}=ucbep!<{Hog3?CLH@uUJf*`wJ>IaMPn{#$
zU`j`?y0h~)TI<js%v^=x&->!6Id1w7PVy8Ro0y1H*itBVG(l${Z&leNT05Fz&Z3)}
z&j>BbYqFH878g9j%#5sK)Q>?UtuMSDN$wRreG;GJNK`w*65Tr~8z+GbOBMzM)VCyX
zSaDbp#dgqn4ylIuHo9|mJ$sgh2E*zJFe!1Hjy&!KLkjA4>)w;jVk|~RM>i4!)F&C%
zwAz+9oZQf?5<d2+@`U7u(!u=jYu6wRSE$DUp&NFQ8`8lH5dQ}H&YyMWiLs~t(9wL-
z)2^={Yx6{Wan(vkZ_oH0d-poyo!7M4#7_E@91)LY*|Z6pX=FZ&Q>WN8Q?24_G^Nv^
zk0cPof8dDy?PU@7Ly7~*@BX$5#ke2t)f7sv3oYqd*sAGc8)vo7nK4G>gnltv9lugT
zWRUM52dZ|&1&4nWFuwPv=vYR-J#*N5#51UT{e)PaLk}WR9pGeu>V^jf26}pW1_v{F
zZzk^SNASldU0y!;d^pvgk&zMiAj(!wKtRoz<Dq=Z$F{cP&e1u8W;2M3SJNZ)v>cOK
zOHa?T_8xvfy*^CO#B?M6s%mum&k@9WB1T&MHaFbs4~^Mtvlycf14$07a1vdv3!-L(
z?1+EN;cS9JLRezyRSZr{=|mp~u!lVX<3j$|dt|Sq3mg6*!AeOym{nlgwEn46sTpmI
z6dx^GHapIgmFt(Boa_*N>b#??D|@ByK`E(Ow5f{ItMe&TR#vj*8y(c$-dZ&tKzH%t
zPvi*q42zTnkKOfqmwTX2AhB8LEb@fZ?dP;^wq>9<24Qi_d^@D~=3m9x&VzM;ELfaB
zx?}pz(FfgYSN7KzJx_|6>lp$#Llu$n`laLST;QNo{#Z5)fk74WrPbEd!haq6a*yfk
zcPBt%uRh%K`D{IR@C2a*j(iYn-ed}p0@}QLnoV~#9O*BMKtA4g7<uOhMq2v0i14Ap
zro6K9(xF3#MCsC+o;_12mg)TALuHFz6x{MSI6a-W*>V5={l1YA5fMn%gp9NNqcM-j
zIru85JV6IVCves=sBc{?8X2^lvrpy0;b}Na+{6T$!IL44gM}W`w&pWhwrvT@r|q(;
zUWIQKvVXhYQ-5!ALS$rQT%15jpmG|sRw!pd-HhHC!~~G`o06{GJ#ECI{ou8bs!rbK
ztf-PjZ#bqC-3-PS);8q)6R<I$5G{GpaS$j1I6<j>&CSiF<2CCj6h{`4@B4WB1q5hR
z@+Ec}32op0Y^<BAtNWVg<~zGR?bmW*vbVIR+|P;dh|ZM1nt@dTE9%$-<@)QKN<`Q(
zt6c^{RknJ%MybnMAb)q4dl-ml2~t?jM+x}S(m|6**RAv|h5dDN`L!bqZJMes&l?#G
zj*JL%Ro}C)w%+MwnG<oW^lo%IBH+M4WNjE^>QWai076YZ0_}H2=Z>)-{QH37))xND
zUW#1gSDyubi+pyQ$G~ea?n%2PGqbtlS7iWRxVZU)C;Bgi3Q{QUOaDG#qMY|ZadGNY
z9xfuJo5gQ@ZH8%N=AQWW<qH)Y+qe~oAGaIvu&|^#w4(Yf(PQE(p8}=@Yu*fSGms0w
zMC<J8>iYEQ<WJq_h`d;@-c7u9>(+wXIM+~T**(oQE(oNm=Z~QeAAv~bf~>HKHLf!W
zu4C`sxzoSE^}OzGI)r1{Zk2UFOh++jJv9Uvh|!LsEQ^@YEPUz)-{g<5w1`lU)DX#L
zU?V^ok5;W97t}5CzE$i2RweXFUqs%V8WP~=2W4H+{oaW;W!wwj8IFNvTFwsncZn+m
zLw+I$fe*!G!7OSP7dJP~ioK&F??C#UFEN~=dDJXKKuXDXZi22f*u33;7+tsb8y-@f
zx;PYtr1u=MwY3HP1o^8cY_6Z*@_EA=5ixE>WCjht8Wim9zr!qRRBqb5`7tPt=CeS8
z0D~!hw_C^kB%iNh7#+x`tsj96tKB}JTX)z~-?nj0TzQP;0yD#hiJ843ZZUDRnB{{w
z0>HU<L=q(Du6DBM+!!)%Os78=bmCU}^5x4xJW`&@LmG}l*Js3w+qmlQZJa;@DN^Xb
zDiSo4JUCB(z){sGfN3+k2r!!BciW&xQOY7Crkz2Zhde>5g0K?lc|lPl)qchD<%kKi
z43e&2zr7e*_czc|H5f&J23NpxZWc+!F^(k1iMIteE9IMu-GE(omYm)BC~qs^TA02Y
zwt{>2>KPcI?ZVX5)atrXaNkVsyD4=bax5ZcMMY8Sv#Z-jX7s942bP?@WoZ|O)aDlO
zTq4o}1hBenFXw9nFNBma`6^cMI2;bbqXSY>K~>-)vM23YL#YY)o9)yCwSe5H==2!m
zav8ZK2_NetgPO+BjyYg&AW0lG_=F4!wq;XmbTpEEWW}7U9v&X>?$qK1#CTJiWsOn}
zEoru;%Bg}EQ!{eL`S|fR7}mJ5*TAgZJUy|IJUpVBe4cJ@f#-`6MybxbrBSo6#XE(I
z%gT}^&c#>j0Kg<&N@xb)&%GDXH!}h{g<?G@&JcW9WsZ2)`y><__(XE-+&l%%n<stP
zw8bUr5m3e6xN+RZ#s*f~&pzYsT_!-gKy{V(-DClnNC@a|Zok08;xpRH$7coZb2&SP
za7ZW)nX5xd@8xCrr4eNh%F79u(a#$Z7gr9e06yTcdndbIy;8~3-yo=0)6{eipdC6%
z8*6ITiB2GWY<59onY=R_R&Rgi8GC!oSv!^$f$r`P5c7kTwzeKX=#C6?Cmd<<X?68r
zc&rK4VmJXgz*~Uq4oFKA;YbYc!|q3A%UIT|^Q{4{?cWNx!B01bv~UBuiL^D8d=^^0
zRuXu1oVIl#fbm)`)qno^N1d}4wL>WUX&!)F0%Ga)CcAg<);TSY00|@`P|jt&7mU-I
zS9x!yUj7wS_%W%ZsOX{7z_oY%M4gs|0Yl_PhL)n$qt@;d;Q4`$nRg!Zb0s#sFar98
z$NCJXWX$rSLI;1GF0V6!fH>>>4F{8-VAa(#@O<wOWhoBt<mH8(I3jr2zfhJFK~q`T
zMI0h@QWN<B$adCvEe})PKIgl}Xa7q)c%SytLzl^%3r)=QaQoVtdGssrpVKn-y)t1y
z-vMXGPpU$V!B&s=WpHRHD<hQ=;YOYpj0KGA@7Y;&C?O#@4-1ZrR6HypQCZHhH_)a8
zyCxvTgKEmwiU&3)HGw<|@$o6So;`!x0Owt__R#6?a70X-BythDAo~HtiTo4qg#hn2
z?m4ER-7|fQ^BcqW6REmIQJ1rN;MjKhbJ`*ycbaKVqL0_29)Iu8^a3Z}DAP(LE3J7i
zvf>CMW}p4b&Nv^s>T-Gp1?YkcxKw~r=8<mn_w<;L8Z6vLFEJ9wphn)5H!!^EJm%J-
zU=zVji1B|cJ`**_G7IV^cb0WU{85DB2pmL(k1wP=LYt|Gm{@abYir8K?(WA(&#03=
zuU@{yyPHh?lR9U~Bbr?V02ZI*ih!`XjMO$joU}Vpu&#cz9mdNXhb;CcpQ+WePTkco
z^%Kw4cHA&T)Tsb`awQXw9VU2C3OM^oiFw^c(EgDf0+Ec**cYa35$g(;-gfDtqV>RS
zF|ICZA-C@|t=!pJMqf{WKd{pM!vOa{JO4%KQ4HhLnGy(uP-IY0(9xE|RB8*hO{rVF
zM?P&XH7yMjLo#@;B87p4IHvacqh+}){J0d3cVBqcM;u1@C02d?_MsWQ8ko0Da<#n&
zH8uyA?%K1*33`+$MMb*%p<ZP#L$lP8s2h+;;9y0nOFl9xN4$9nP?}Wo7k$JUX((!-
zD)hRpPTck>0+zYozvV8?z5B2adHguCJat>bBZTxos(nUA256FE($dnM%0O#LCm*t9
z&J7v4AchCWqFvF@eLHow8;bJ>BqdcvTu>;VZ6ajN;^Mu0ODv&DJ=Wz33}emk0At8l
zI~I_1+N#Zo^YU84D5uKBs?Oj2&`*0Tg4ldG2;U?z2EptlOO_B2v-~o{N&pd<Du=Xn
zCyW(#0ykq3fa|vH+hcb9LZE=ygAv7+rY2zh<1nE8YXq<^uxh}Eh|GUXuN6*lH(G0n
z_Vde^@dXti55hSwg2H}%Rs<*t_^7K9cr-LLL}z1dxV^m>AyHbBJB$;Ly+o><yZVAU
zAPR^}KPr$B5jh7IV_heSkPjceQt${GqR6kV8>y!?lcJ0G>$=B6cIS|I1D%A&%9vR4
zxI^JS(00Ttx>cfm!WGs8;Q}m;!ll|VRZY!AgcGibcIfx=W3!|tmzK-2A%Na;LtEVa
z3Ow1S=$kh=xVQj=3?sjd|Cz`@Im83_0Ksy*CjevUke~pG!ol!<!1{Pj3_>^(w4M+A
zY+z^@Ht?190!#{1hIP4Tith|jY;mc+!~@<7Ua+dD+q9yBgh1+DuHoCx&Q5fF=IOtB
z@d8UmDDOZmAxRbE&2VQu5^z{sNJWgG0)qGi;MNd|5nsNb)Zqf&1q|i+PmiPo1qC}@
zdE_qRCch=bEkMDrM^#l-fbs*cT|>b^5=tnlsx=7p967Qh-lYoxAOeDtG2x2B2cgG2
z_V(^M0l|w;kG&tiy8;6Z2JtoEzvFq>AO2)!r2&Nf7!lo!k3eF~wr<@xT3FbH^c&zd
zy1G#Ilob;bL&*`w4nF0fW1qT#!3Vf(5m|=!sh%@9xd?MGGNruq0+3v|D1^hvgU@wW
z$@22{0z^Rk4!~$;W}B~HA-90NZ>O#l%!T75Z`4kMIh_V85SYKwUdV8eR^xU3`t=Tl
zdDKObL#t|Qm$`-_hFW)EtYImI1!m~U`f`vy%s5B){NvvT`fV%r&Pk4^_XJlkWg;J+
zs2#tzw&N&8B3O;aUF>zoC1n?ZJy#V~x#D$A0h0yxucy@vDefmgxIr12PgwiN5!A-q
z#$Ne)h0uP3?eA}tW&(qT;raIFw(p?kTWJLN*vz|8QBmW02n5vnn;-Zb#wSyJR6@xp
zzyc+BPL$yA%rD^lV3}NSynuusL`IF92LR^Nr%wwD3&<CS+l%fRiA+@Pzy*RCD&H*e
zNrv`Tf9Vtn`gj*`IZ-}8Ji86+guno3L8-A`O@Jip5=u=O4b-7L*@QV$7+QWA+xwnz
z`7C$9HkdL7Wg0)j%}K^%$_5ELp5eahmYVP6U%Exn1~XB{cRavefp_%)&g<7Nx3jOV
zc!klPp6Opt&%a~GOGGMYRl}a6Dlmt9e6*+MC}3KEaEQsCKYxy&<B)fe@=tpaFaQL_
z?``(Jzs705g%P#TLu6<a#V3p~xnV;v1=N1@P(=3yevGte`VmNm{n1{$w1HQ&KOXwm
z_K^ScUq7ZT3FZH9YZBSfv4rv@;bm*<G(y_<M6;RWE<7JRd^N_W49MYv)X#%jRgg#s
z`3Co^ai`r^<6Q3E15gTdX(PgpHRJO1{Bgns`Dc0IGRmjj2)^DkE~&Bjdo^fFr$&Rh
zXn%}Qq#efp`tg6WCX)umHrKlrA9>BTboBJ@f_pCdU4}Ze!2y0sJBN&0!NTLboSb3g
zE4kdP5uSrm1{-eC^&S-UI+R44FV}kxTprMC-s}><6lQcR=dm%F)kBV|6igfZhd>cc
zO{m_scXg3YW>AR|fu$nYcieLm{wurHwksKNE80G5UsLQzTj|2@?!{7MPH?O+nJ}F2
ze5l?a?IumfsCDLYZ;1$KcCV3pFur8>283=AKLHOO+wXT9R_#3_T|pt<1=Jw5$LQKW
z2+zWthv#!VAO->r;3V<00!LH*90XP}8P%Ybowp((T0>2(c7S;eLF~vRjH9W~5j-NG
z6~(Y5Sg+#ZVqok@*x?B;B0vH~APb5_h?(|rh8+#VU|{t;{gBIKo|d}EF=VqKQw-fe
z`6*e->E(0nUn`|9%{fD21BaKizi^{#9|0p+S*jWuA)pU{!2oxR-tI3013%H>q@=8D
zbiiPXcIsrrW`~|-W+%x>c5lVgJ6E<Mk~BKNWuRhY)B%-<DGgL!_9Csd=u}fx-EBkX
z7WD^k6Rr;8Uxm^$SBNu0Hwtj=x=rEX;ec<wPoKWC`X8-C$>tPGJj{;b$(JZxa~*Ic
z{XW2e;?BKB9#aL;6`+1+$DZG*`+fY=BObX1|3-z&PyCsD+G!Ts{R4Ic_(i&@IE2NZ
zBa!gt9qRh{0Q)-+D4U38Dx#K*J9msS=K*^ibQ^zCY(<8O1Gt=zMEquzAb(kxkB`p<
z11(X6uZYBxz$6TyMu5?OsQ!g-mOJR!TLTA!icK-wdKMPA>}iNW7zkLB#TkG3wVLk9
zdICQ9JQ5TV()hI(+jceXy%E+1aEv1ttJw+CI|t@MeaT~#sz8qEO<&QoID!n#vBm7u
zKFEC`NVPG|X&nOUOJ-vuLxGn@x!1+fQSfLgfsv|^J|hI58S6T6VvmrJ5OU-FuhN&f
zbKVNr&$Hnmh5azGF1^R#>BSxhJx0ie<0@YMl!#0rj+VMcudk0>W9f=D1_#zim9zIC
zkVH>_u6a3{RS_PBb4g9$p>fKP;o^3XSz4e$?arPjL)OtND#cI;@B*(nK72-PK-vLd
zKM&8m-7$K$#HbSKInW{`8^QWBAjL2;^6WVsns15Dd$bE(JimNR?*(U`Pj<$?C<ctg
zy+<4ecR+h3pOn{NzH%C=t~j<VZ8!yjAm<8%kKpU=F!n<5qL?|2<Q{CPO{^#;lR_2>
zwa{Dj$1y@E38)078YnRb4-YBVVm9N4PV2>eTZM$qB1xQUsv~@otu?@sAXM@xr4N@f
z{*B(Jn23lJxfAD6K|mN^XhI~hp&j_H>N5TSR`#i^Uq1U+VP!XAWn&evvJ*EDF+((m
z$kQU(b#-zQf}|04TJvV8c3LwuFbWWg;K+*VWk%e+YYk=&6eR54J#)9y1(gQR1NB^E
z+)A8Lf|LkQ)3HG$IL25BpeaZ5`W$@a5CkIcKp9!&czz-@%>G%r0z5+Wb4jAagfGCs
z7G&rer<~S~c2b;3$a8@oc){A*8UQ0MLiol#D9^mhbaNDymi7QYYNbD-`9||1U`f(m
zzZ@7hi8(-m0g!eHl2x!NQMyAY4+Dq7Dahwv(NX4t72qme#8H5fC?ej^L~Tze<q+}_
zGQ<LLeP}3yzX)#qX|%DC2o_Q@xEjb*yu7?%lZfmW(k`4e!2o8X8e}|3_SUY^K!$Qx
zdvnh7?Ur4uc9~R6p&y!qH5RxBKwk{nq-p?5IEN?_cCEH_9v=>4p!stEjO()#a%RAT
z7}Nnr-5gcu`NRvjw`D%96)K#_Pt{{VXFh>MO*E(o=`qZ&q2P0_w1Jd}(J3xHXCBr)
zo&*0R`JREHE)eGX3f|F5R04LnoOE$=Dw^tm=1A%Z6C8u|Y!Re&$3;-57ZW4xg;aAu
zE5h5iE12sdEr7RRKL`BG8`e`?i!|MJ0JMJ}y?7HjXvJyrHjA-2IXPD}5{DpOdfn9j
z4T4+#MIR!o3WWSlMuyeH7{|h=`#4`C12-x=e-2naizfCE`5YjgjThq@(IbxJ|6H6K
z-r5zptbc;^fL~fI4gn4+Ky2~N^7Y-_azaA&X7bJUrf@40IY@yLB7Z-}<>Y2(-|_Yc
zzym)GB}GLWxud`hl@eeHhH%h&xBmeKY<a}Sh=+%V<kMPOuU#km?|``ui%FjbT5MK{
z!f$W=a0lGNYEld85)Mm7P0_pRIw0S!O3yNo+pG7DwHNK|e1GcHDZjf9Vb~M+^!gai
z-)}PnZ6@Qm$sNs&T|>ojtbJ47F{>FESoI6)Vs0e)9y$3<YSq`<lPOUNx9B_=rJnjX
z@e}_0%Ew5*KZY);IROjmBQRmiDJw2s2yn9Hl1Yqu2B+95zsNfa3%~7~pGQ=oTU7N@
zQDNV%XLFl%iTqmchNhmC;emlV9BM%5aL)RCYkNfBqQ&1I)Tz=RsP4m*9VuQhe=h}$
z#k%wMmfDhE<#FPQOw^|m3dSWqeg16byCQ6UM`Coe<-5r`5IoU;z4On?e7<Qb(zO<A
zUy8;Gs{EZhcdkKw#N7|(``zXFQf9b7_LWE;<02x=M&B$Xw{91P`q|Tt5ST8P(2aet
zueeoj?A>ukIKy7{+Femj%L0+)=@q~mweFe9_w_mF;k_$d+}+%GVc<$0EO2~)cU~i8
zARBh1Nj_$S2P3C6Q;Xx>$wN(B2Of-X^xrFOCcMLCiH}#)^1{x4;{t32asmU6E>_Sy
zn3$NtLqar;93k4g=udGrwTCNk0gh&uzzO?;Q#>*{ihOJ%56=Qj20{v8Civ-*Ha!Oi
z2OXV=6>{M#K0dr?54^LZ4S^590+iN}`yJ8N)>c(T(stg#VH~9s6Jz7(;?wzSKfwyZ
zRe|p<j@Q0@`wkvl0J1{}SjV!XORI^@cqqr#Zq>$%?O<BVK+CV8K{H0ZZ*UMGa0+T}
zDoRQ~t6n^RKGOZz^rFI9XJ==CqZX%5vD&34N)rx(TB$`^I`sVO&1~yYpH9DfH{pG}
zdIlK}*j_Y%=ZM*biXS4JDTH%^rj|(eP({0OZ}Qg<i{{lQaOF^OiDtO>b~O`|{rlXw
zQO$8O48gny7!f_2VD#^&fraP&;OkMUzAtZ%E@6~sfW+*>m&LoIMj$OEMN$FSO)N<D
zqJRMYGZ)tlhr<9CdcYF;MQ^y!$Y&Ch!9X_#Ptq-4&hh(KJCfILD^-E?%SuSRPkb7;
zOqPx>uU)T6UTHvn$)qdKpj0mx@!r;>G6yNWYVPOvR;Xu8{P@9KsjahkMuKq0TB4((
z-qKheuDvzPT^E&h?!1dgV9oGNE6upw6N%l7GS-b;=X7UgW}s^lpjYzY!-pqNp1>5Y
z?iT;LGCVYt$ezLRrB|Z_4|n|2WpvdtiLati+lU=&>;{Cb5w_tsnVeI02h&^)44yt(
zR;M_yC=1l0ls@?KJW;-w+B|l*1Oe>)iZD(c&`0)~Tlrs}pFVExn`%+VN<0LZit@fN
zEE0JoSUG`D8tl&f4I7)aIoa5({$7xGa{faM7+3aFQT-@liwB?{c-4b9{l+vM)xC;k
z!aIE@&{urccLfTEuf_dzq|SyZlj_uY6W9vz+y3ZO9yF448&7%qP``Yl#pzA0kSv|K
zj5Ckovo~V)u<{Cx7kc)vr!_CJ5F<`|#eoA4Q45&;{hPj<6E6DvmwH56{2LXE%$_dw
z4X8hh=EK8n+?pC1EUs(M#f_P30lLI_y^-Dj?OQn_Vqj`@WwFoJE@1Y@D9CR8YAT3;
zh{6AU8eOevUNTbcR=Ce<>gv=iJ^0ecV09oOQ96#UgWX+we!yk@>aAs&jc^<UwZ6I<
zfk6Dn&%j96?kKy3?*cSW%Vr3KY=kw;u^WW+?V$6AVO$zoT9w|53#gkaTJN|eZWTO1
z=ar?GjT+KPG(qt3^#S~XMO&}0s-`C8JQ#6VO;PbVoVNcy^i35a8mp<PK`e)KVdaVy
zt5sg0BG5S!Znp2ALB%!TZwUY-8l&4@H^jVV^SNwl79@K_<XT4Yy<oPw;ZXGa0|Qsh
z$5{gzL~r*7ZSAH+ziht5gWGVq@zGND?K7kt^K%Y9MdG$U8KZnsMMWiH>kg;1frtSQ
zMnP$&H8*o$xt<CW$GewgNV~{52*d`|??XvPV&mgIzP?;Ldj}~N-yWea;5WK-D-#zV
zA8+eb@c#DiMvK^4_J#*fGtjwk4Ta6$9iMZq{gXIS2$&C>bk-CXj=B(c98v-0Gh_Wz
zQ&Wg`1b?l&<2nBMANe4oJW>7Jo3(2ZYgwYUGRN1HEC)|}f3+5xxb@t63Hb`}9e^z)
zL4;xk|0)L@7I%)`TA!>w3WlutXMv7^5s(YhT~y?R-yu^X5KpbjA*8iAZ&A!om6B>t
zIYKrrQR*$As<KZW_07%qaee?9LkXsAeD*jf8oC@w{vRKEcbggCRL^j9aQKDHFn4`;
zgoq-?h7HyoR{%l+-+h~WT2Upf8Ev<Nz!NJX(L7RP$rQFt`n2w^<OSuH<f}>x=IDo$
zk#W8^eK8^|Wgx{*YnlW)(PGC*pkqj~wKK@pwwMlfJuGW@IJe)|y3=JCMR%jTSdX?(
zpQ`X|9c=)Dp-OQ>%-ksI-o1M>N+>IWtm5<`e&hZJ3|$|JZ{EC#&|mFXeXS_{+O@4=
z_ME%Te7t19g~#1}Q^kInjc^KFE<T%we#V5BmBxc`M1X##!eQ$<I5H0~6r&uNE2pk^
z8wDRN14*ZT&H0{G!gV0!|AGmmr_s5Y!2YT|pD2Y_f++EAu^h8p&2M2p4R5ZteS0Tt
zftO+g=TkPu@D2B)`ABIF(TGBvU!*Q`xO80)Y`oS<iGBM}yS!b^d=I>lR_#jM_5`o+
z#}PI=>RS|a&nW`rA1gVt+HgMO%h#{3*F_mNJLy#4p`Fm+xQzdBLRGsoax&y<50zf}
z2-97AeATK|Zz>OB@WquuyK0nB?q2!<e}IF5()6>;PK&iSFZc@C4S+#QZFMZHcQ&<Y
zwFMKbq-d?$&r9Ets<-PmuAa}0iXA-Hs4QSmkn)u4Iu^bn%btx01ud{-Ix9Y&QjZbx
z`rL{dmvfW*TjtAs`t!echlGaW_TQ2B#}a)Tu$lU{na2n)N9y)cM<O#hb$9i9F%Bjh
zC5iR7$3MKu0z3*{E6?W5$q2+jcY_IDJ^T^!tDq&!@u7IP07+06Z*%~i?<iLLX<47h
zGQUCM+^Gy16*;Dun>TF^)6Z?HgWY8ii(`Y6vdbPwNu5kFXgt6)#ns?XznP-39fh3P
zD@eR>q%KMJtW8*nZm%t-^4`I<FcvGC+ewS$Bd8>Fw1H0s>439P^Fppjc$GL=LUH0J
zlglN3!NS11cOqRxuFWbyeU5W~Bzz$>Sll$*sj!{|d!u;xP~>dci?4dTM;28!wE%V~
zo)(eK@frVQ4gZ(8{{ZZ(uM`zzy)*B4V^S>w)uZ{hx~K+TUJ(`;xbB)2I~q=J7wAlF
z3_|-^MH^f@yZ(cV-;-|nA%0jjeB#2FM)-4^fzH;p>|;XBtNrt;mA6I5Z%PMvPBLD&
z$*zwd*Rb+`{{H>3)4-ADSNFa(<wrzDwh(63;zLCAr#;V$1-t6gEo(m+`H2BtCe4)+
z=k{hSP;o_02iE2eO7WZP%F?mjqO<5NAm<D<x_`5Tfio?NsN>GAX+P4LEvJxE>%`iB
z5VUBZH$xY*zJB=v;0MWW8}JiEbO;k+n40EH1uc8Me7g~Pf=>sY^AMnlpCj#v;BI~R
zU=a7cw|a@bG(6Irsm-2g%lYsz{j<IjTAOsT5*_G;KJJ@9bbIB>l_?D{VF8rqKk`Rm
z%KWyb<MYnW69%BRWXK&aj0U@L6@Ag`RX6*7_@-{&vjH?CVbdyOgYyRzn$PV#oaOxo
z^_WfRhG+m#+<8vUsXugK7)aT-1rla}bkP9WQ(#%l?4Xj~yeLwtw#aJU;l(}oEV!@e
zzP@~uyxLibRwtci_ua~j|M$bS``UPx{K`VD=bgeh$1u#_QKUc|v(mH}j4mv~e#DEo
z4nd8kXHFs`B!kOw=<~g<=jIA$mErz{a%<jA*D6}q_~K0d_fJTqTeVuU2|pOg^UaJ4
zlj|!t81!E|y(dz;%AgUoG=G-H_XxS+mh^*EYFxJon^Y9rfH}yV>JN~vf+ggs_Y@j|
zlvK8Geq_AYdf!6DWJQ-Qoxh*LuVbe@9MeYMKGuEeC!hryI&%BeOGxRg4_rJw_{OQm
ziMI3G`(hfOC3HD`21wLGK%&s7(fR(lpQ{j3#y{G|S9ZtfFuVzwWQLd-92`_T9|M<~
z`<jR5nHFGH_-=-xKDZCSr@y@%(L*$k>kY?Vg-jRD2HZ0jiwBC<Ev$95qHneK>7cnK
zOiWm|A4S!s6y6;>ln-uC`H0LuWAUzXTAZ{?cD0q#tH4@@uj{*XP%ny2h<z%&{a9Y)
zB%*1fLd&3nkFHl!JN5jilfofK5=D$Mx%b-k7o-CZTtn}__{}~YLDA4jf;UQj_j^=O
zu)8a;1Kf#0ZDtE$dK33=QoZF#zQK*YKQzZ=-Z<^vAnHXC4ntMxiDdibCA65cWPHg~
z#c$;z?yU6Q_j>gL2p$pN-NEn(cmcQ_>^zZ&=^(Tp1n5fk&-R93O}>q7<%Xge$)%6n
z>OjbWqfq|v;c&NG&UoWb#07|C&zhp`RMEL6v}D!hfSBvox6Yk=z8oPSYJ!!QYmqXt
z9%&QucsK<tIlYQ0gz*@mm-k)08l+6x(bB}Zq1063UmbNHX->~mV>HJDKkmQ0JJn9G
z8gKu5Zf;FIs<CYB9`5d0t+taYJLH{RKZ9o6Yedo4Lj@Rw)ib6jydqkBmiYphP@ZQO
z-r0Obj_H4IX~3qRV5@%i%;g^D8d;C$pC42y)xKVV2>C~|p&yc2bX}kZnp5wuq@-l^
zkXzFg<@!;2m_h^z^<Dzh2SGUgp!KC*x*iilM56g}>bX(DUdIlE>8<f7y2qY;S}^V6
z^-qfel?|SAmoO6}BYWOsKzQcn2v1^i>-XZ#_Peb|W(BP018hW*iNu~glYnZ#2duI>
zT>ldAmA^kw<holAn_~(}Rsn1f6qfN`cm#szg;PhX7rOx!5EyK~ML{aKJ>O35GAdS`
z%^dpu^ZW2sDO9E*->7f!YUub$560|A?`d`*=nTOGPCo~%t3+U3j-8T{2K~d80J;jy
zeP>F5gh5e&L2&TAUTARe`o=A}QAtTv)=fOKs#7Y*`x-BO#OY|QeD9r<;QmcnX?AwD
zR_^mzU`gtn4Yi`MO}Eaw>NV5tG^&TJ-1bL57U9DCZ66~;M)JA;oMG3v!X+dw>^yr2
zA8uaUZg=AA5%ge~=|slH0(O*(bJsx86-dnqAWs04u?MS#(2%xjCj%xF#+RRBWMYy-
zE2cc@6I<}kc;v;=gLu~8AL0Fej{7;yHi(6UbI*MH`o`zHt34Sv$ibWb$qb|1-tmPH
z(+)8&<(?}jhF?D)O3ZmTm^0?wv@>JVcbj&f)Nr4AoA5_^(}k+>CGO0N$rQ@hKSWjx
zO$+ZH0_$#Gn+w&-<fmUt^%gHj7@^%t?*I-?y{lhHLPElrOX>D^4Tl#moJei55vSj<
zVZ&$$n$+q)QmPPl`EP%O6I2%~(g|Ya=rE|wBo3?PbP90z+={Ha;llzzKR>#H_1OOP
zqQ7t(K&wGzRz+F)&fU9mz!zbc@M!C=>5|bUBC>QffF)c4;u>Ma01OtQ2L^%54$wzV
zpH@~;AwV}GB`7RUOiy!5J7;cmTKjT6_uxDL-xFhU%?1)2=0r7D*Go}JX&;3C0B9m~
z0wxauL~EFT3{yCz^SZs=P;TQpps3(D{Ty!7=6H!4>&ymHqzKz6SPN^_k?WX$q-(7I
z*wwW`#3=o@6L=^n|1`K}VU)$7=yKq>?622k*?q3lXp$FVqQ%mhRz@V2E+?_HVcp%r
zrIbNPg_{mN8tU&q{jFp;+;|Xt2EBdriARi$6E%J3CM7N-WSG#((h~#`0q2T72a<Do
zKiQgVYrnyhS4*Q=G(SJTLmGXF?}60#yTHXyy$V5AhxB4q&bK>Dj%Dc~%)LJAqfWYy
zkPDz2kje%FkDxd2#8>1zYXxtf^&R#;drQpXstr8|7kk~rThI`*jZKsBVmoqt1S}~z
zljuRO@hS$g2TNK8$}4)zKj*;<e*3a<pKY|}_BHc}+_SxfP+BO}_!e+`5(S|k{Upe;
z6%-!K;?DbXJ^e6}-0GEt#hJSiy*#-XjKOXG8u0H%z2E4L(7NUT+D-LzaHF9y3JrEM
z0pg^GfU!{!U8<Uzp%cmz)_&mWx7~6KgLPzO(Z6v{LcR`o;dD`$5eoG?o=33EN$C^j
zZrv>ku42ZCthBVR-cTz9rM@{H?sw%0PuyYZYok80F;Q**g2Qf4Mlg9Ya*BGruNues
z(BSxkqv*6}<raH$nwP;5L}L~V{yZvy$jG81BF5qBk=i-oD}!+KURIS%cLPs;z35nf
zzl7Vk+2}A@4$(yq5sx_#1pw=Td4i74LBbkHl6~9K0u=(G&1xtSg|G^wUgBnDEk=@!
z+k;eN1|A9jRu7B`q{blU_u~s8r$h>l>>TWyMxeRqgiugWD2ah+phbBt3+!h0Z&%a=
zK7t;*2s8~fFEwp#z-$Y+(D3Z@AZ)#O^-74#J^Go6eug?**W@sw3=Cx34uTV+9)jM1
z6j4`G6BrT#xB4E{yJgFkftLLi5e*1YHhVdp>=>j-lh#v<6hsr4RJ3+a{wV-^6Bl1;
z#O=3F8lSQDKZ{<AhpXv1hcU$RjrNE32k-&(blreCH7~9#kWL6IK*f|$;%69t;3F_i
zqV4e3pfzavmpl_KY)BdkU%%$Cdy21vC%H5_j657UD+t3P{QQIP7uZRJ;NVZ8o5;q@
z%nWOY=J?D!uuMs1>gQmyLbf5z(q&E<5nya9o#ycW89CHc7<y4^(k3QtreK_dFPpZs
zA#_CzBC8dx>xIa%Cqy)p51K5#Jl8B{g|0COCPtl6lx$GHdVZ+Y4s9hf;5kj45D3`h
zh1yE{RWuEqn76+qp4RMCRgVCJ6f~dJMXzJLDE`|=9K_B<jlKN*s+XD5Rz{!M8;_>b
zxNw1@w5Z5P44hB(3XzhO#Pr4esPq-@D+(8QYqJ*?36LM_HNQx-ppb#HazI8t#P_}Q
zAb)V%UT&0k8HmKu(cXboXqQ8~JKI<v@-u92Lg}_By1v!APtY&2PvmN1!JLhAm@>Q@
zYL2S@2Q)IM<rNimp$cua^Fq61=`Oja4y4?k+#+&OnrSv~Af>1Jr+vA1j|IVn3FRhK
z_klwX!KL7%Qom=n4q+o8J-Z8CFybPlo-*`99yl?|iBw4*OZ0f5=%q(}YYdJlGY`+n
zLQA!fe7@bg9gw$)S1ux{1s|-_Ofn`q@w<48Iwn=6ASOWPm;PlFpp7uz2XJ>l&W(N^
zt58j+HV-BFeiJ1#vlIXdtLYK{V@5zf<T)VjeY!LT{d?{JqG6LaO}FkP)9cZ5xt^0#
zFp*!srnA!%r74tQAY~MX9JBgZ&1StdX!w*N;7S2*A+#H6-M`NAR)c;C97n8ku^Yiw
z9?Dp7%+z*EdNobWkLY*298_6}_QYq7dscSd*3UtPxRITmecie(rf<<c1a2g_r7;@=
z@-4csqyaK*Bm}A1_*NQuZZS#F4Apm~QYeG0Li!`{U$C*L>raC{l`P|)s-FV`-Hx_1
zM3TVO_jR(X-w-9u`$P*Ig&45L0)v7sT)c=1mos`z0JIyTiO~Q0g9i`LibvE!QO_R6
z6$6_u9$9xsS35k}+^{!@)Qd11z-|#UvhnLYdHc2q@h$L3d^$9G3Rj0Ed<K&MXSg2q
z8gN&=usP44E85%J%gK>?FHo7954#<!>h%PV!7eg!Njl(*0b}HXA9@&rswx~HxGIPf
z+R<o<v}U03qFEjeBH}gRgPs+U$dZh!yvl)m^c(qygw&&sb>`h?ol8J8VTn5dQ^JP9
zQq3Z`!8gUHgNOxe5U8dEv=%@JKF#pK0EY@DgOpmqxAKOwMHB>nFFMUyU<*1sA3_%c
z=EV-uRS3)(`ne@iyl;dyk|^`pIOvX;uDVq?8JOpNt-j7xB?uiHQOhMf07Sqr0+%Ox
zgIq=FQI0S*?M8i(d>XREI$gWdWb`O$mwDWa1lbLhNk_T$oy_ajtz&0*1#AQ)Y7}3c
z;QWYH@U>vf(aq8f2okcLp6Ds?9?mO^E;~az8pdXyNZ_Tw*qc5X;SIn@)XM`0#ND?f
z;MuZ6NtRt|CXUGy3D?ZGPc(_X_0ErfkM^}ZZoSJ>-HQutW-7(Dt#uvp^YqLexk#Bf
zy~gn+4f)oexcROx?7Fw8f_>zlqP9Vq53AtHD#yjL=xN0r^x%BJcU(IbZHh{um)E0`
zFSTXB8`Br>f&&B7p34@>zpYpcWqO&*e;!;`!_N56uAuDRF3luyg+7^m!@({0wh+$k
z#BQVq&@4Utn0M>e9>{(c>dVaP+S(4G`C(dvUD!Ye;?ZYcY*+M?IE!lkN<IMD_tkkR
z6HL-fr9Sk@jQ`g=m10IRhfF8|bgEU*;f&V4sr_yU1Ycg`KKsJWFYpa1RqR68{x}_$
z9!If2UxxNZ`mg-8L+Ce01khN2?g_rW&mfhXW{UQsAMkW1gp$1Y(Hru!<HZXrbhRKM
z)elt(Ty^d(f=!f@oW05Sl++fb7%aU-m-Dsy?w4x1-hKLRyb#3laYq4GGp*U+K=2=p
z#e{LNWRTa`zc5o|zI_e_R&s1}xmWmjAscnM(6cOJfc7O0T}uQfsAH?8@2Vw~0s(Lm
zdZjdn2!WEke0Tv>`;1(I{8#+qkMZ#WSIHWlhYS=mCF83T+h#Z^6mgQr|9$<hQd;`F
z_~GH?ugvVC3!hn5nRj1IHwX#dawVAk`Efl(9oFT{@+mPJHsrFeIQ!H@DA;J7(R6F-
zHvQn=`WNI4Sw*fK$WfP1xiYfOlKJq`+4<|;2a?_QIPHwGj9fiFzu@8o8S#To)-_S?
zl0VUKO!-gNqx_$xN2sptLZJl2<5)tKNw5`EPk-ngjNKe1g3`FC8d4k-{-Y?Q9dTLZ
z5OGwQPPGt0lNU0VEbswUuK}h6li?uwhY{-kFQg}<W07hNaZnf!!ngraQFmdbIMjGR
z01mJscl!kDY$$R_9{WLgvh^O~X_O5#Tv&CKGww=Le0<3aR4VsU3RfNaB4X}kpb`%B
zLQ4(Oaa{ev|2m}q5V=6Sh%hSoFXg|Z!}*`8CHg;W@BW`<MgQw@4*f@|)gx`fR{zEY
z_#cuE{gwr1L6ICG_$@rw#Y9FxCa>EH!OQ?LVC6I%EoQ5qbG!nI+oF&Ny1`mud?^7p
z&~kyy*4oO7@9YzlI!K3CE$3PaWgAA@QlL1A2!K=p*3*sPDy>(+KUbxRP_8k?u*zD_
zY{WOfyN|pFB_|@BCLy_WgNgG|<dc!&dgyAv;MAn(+jzG4zkdm8&^IJx)T`=Wp1V>g
zWDYTflCDgx>&8Ek28<E{VFt=_++_-dN``48Y7yz*0#-m4beE`KphkI<rnLDk6I#d^
z)<SKN&n0S<YBXg|>YbZP<A{C8)F`!JJi)2IKkkQ$whe-n>r8~W2!2!)2RiQls(DiH
zApS(AE;P}j@4SSP!kXR(%@S!em3|(dhSXb<W1WDKWC3zTcqVNP4Gj-f^d!~N6dPl(
zHf?}0w{c0DJ)!IaD^C3%+Jl}V)0CO72{zgi#YoAJ3aqw3DVop71R0%xM9jkiOzE=v
z1>DeenouWoWH2rKJmTQnj6WP;vDCw8Z{*=G7;-lwgY6tQ2litr?TxaCUnJ^|=-j*B
zNX9%su|cn8>k>4j6v3|eyu)ygS`3Z7Ze}fnPu7X2Et&Y%r(fLACW*LJT^jGUinca&
zekkLiuaxye9?lHhKEFR$(_3hpsk7r92!hy!6|TcrH-Y8&w&t|`H$3{eLKNwD#b9cP
z`cAxAO&ZzBXtsat{ugJGcY{KwE`n~0RkS;ArGkGz@|3#@SHichfSaZniMj|${|PYZ
zZUGa@*a+QCzrWwM@6lvq4nDiV<X^il@CqMK`}B=ez_g$eJ^yH>v;hZ&CX~s4;#6!s
zB;cJfH7v#hVq@!Q{)3ei3A7<*ucxwv!k2?W7I4gPJA$-8UNtV;4LhFbHBk%r?sRq<
z<&=&xon#G$s}TAtaU<%TSOOUNm?3^j%JO}nO|)*M+^1a+?CJl|ZT*ke|9|5*p=}YG
z6Cjw819vHQ5hL_0QGv0x{@(R){~q#kLirQm1(B&o$88mKcR=i+USp?F7`Ks?bO_Xd
zdDGSgea{gG4y}d}_M~i!L95<JfSG`sEU+54y^eMZpP1+2bHL7qUe8OgEzpXVJmf&3
zq%_cM4(1DFA_oHCQ#p|23ABdF4IQTFt>lBAEFMP!gmENa(GjevUG`@?^jy9F{!KHW
zGYI<cAhE_FLz>)dKp_$`TwT1NBXJwrKgg=McAGu12$v-k3P~08B@&`u20OaEqpi+B
zXFluN5}(5c&=^25g$j9Bt416?w0CiKeL}?$LsPa_R3g&>R?fZWv?d61rx;NY{e;vJ
zMR~#NMVO(Mi4r<Yf-=r5@SNkAwt&>`JV+0y574s<-^6$Ya73UZK+QbPnhwS$LZ%U9
zq`NNv2tlxY`t<31AZUo|kt{o2Y1W6q{^-<%p?FSE)p7xgxDA;nt}AX)H&9_RcLi}a
z{0QmlL?i^T`UC_4Pzu9*SeA|Dd+#ly1SFyI2kwR(mgj!&QU{$`%-*EllBV#7;AXMy
zxd2|6P~%bb@L`mTiB1cYN7VlrAj372F(?81@uJBT&(8yHIa28_;jTUY8(zx%sS~<~
zk;d3_kI|C0D@X2QCCJd3g<sJ1BV56ThS1ID4w9)30;Y5nMr|aAyY#LuRSzS&$)Cr;
z97@F&!TG)T*a<o=TJM%~C&b3OzSY_<Vo>lIsLMT_+lT+KK$)@+vxSZDSY!ryu?{gO
zfSkePK9H#ZFl=$^uAJ<j@n`bf@OQ9FQ<%q*f=y8h*?NasMeO21B4sR@OHz_>nF-Yf
z6X26%*l~jia#~M}1j0f>2`%YYInM8HRNPa<zS^|qcL9CD4|L3$Ku9O51^xK}SZ9#A
zL`Sl|3iS=+w5jtWmwD1dx<&z4gp}vNjEi;R{W=l+Zc9YV>DTYxi=GH&*SNgHjCoxw
z;qn7y2w8L3D-1n_Koq@8i(>Qyg6%DQ#P)u6C_~)`qv?eE^RaO%`;%ar$;6(91+f3T
zaHMPjJVBL^MO)^Fz4z}EAXHTygll)l3<>&ummcZ>0cr=n;uR*=7D&cGlOgrw*yHTt
z!NG*a954&Gf(w{<O6UgI+-u&c;+MgHSB=-()yR0L$L%>kMT~8fIn5kty}iy*7638J
zCJ>^CXyxgqM|A)6?}6Ca&07|YYneQr+a#eS<j;?AZ<(RYG)Nb@HHf|X)3o?UTc@e>
z{6mrM8Zr;axr|_T<vR#h8eoXr!jEt=m)65R=p;EGJYs9CV%u@k57_`1lYfZOkX*B0
zr7((}g#H&jNz(<!X${VxRTz|_Jr$QAjp!fjfMrK&K8HFddXHmbVnDK+z_9MI`>%LH
z+NGqcQ7cls7&4iSA(|K~0SXE!f@6i1zH7B&(Nm^q?(sHo?Uke1xLUp8NO{8BM{fu4
ztl=~V`to<9c6ra%D}JT-M|@L)z|Jd7o=F8)=1f&ANK?ueH_2W0B5DlbTdCRk@RS)e
zo$6IEKYSm^Jky}_WE3Jg7OpVP{(QW6>z9rg>AeI)zj+jwHe}Y}9YF5>vxXQ*>s%Hh
zT496a8*q{rE}Jq!IZT=ccvX%q4DySPq@6ssc!Y}Uvf-rP<a=p%*r878oD6q6pe%d1
z=n%}EX2AgX4KM_0olZ_pFy?92(E|zcn=$qhv)9RG_3%RoSpq3D<Ve>oL{aF$%GzVz
z7|uDzR0Y+M9EsJms741y@ro-}N%WBkoWbB4m+LTayPhTi&|BS60(^1s<zVeJ3KYpu
z6WunTuAy<Zvvd>ywHz)K;6Gqb2P(8_iz7fY(SD7Xhk0`p8RRwNV~=oQ)nv{j3XIR%
zb(V_Q9B(ij5<@&>gK%-RYvWri9vQ3W!Rh($-XY*2tVSXtM6rudA-BFR2QIv9&_EvS
z#ueG3!on-AP#oOyR31g|H!$UoF#~fpHcQ}5^1{?zAVP|z=|jl>BG%I!*@NF-H!pJN
z)n3B5k!{`$&{D8lTcO#G@&nMMguQ1bx-s>Pm|_@q7UQ}<WS&ZIINfh@hIC|3vBD#s
z#VvS*ag`Qhb+xtbm@YTDFCSuha5uA@Kq)!93{Io0{gpO+Ce9k(4j~!V{6~)I{Lt~o
zbH{*0kq<ylpt70Bp?Zk~JRx>Lp7sTQNq<X^BGZ1sWw_~y2Jsg)10nr!Xe4mXc!1&*
znZ7$_JTs4KD56^|y!Wl279JjRkWZfb`91j_sR0vjCbevS-9wng=jKNdzDaVCiA*rC
z&=)v~6Kvu70rmiviMeY5-NefK(PA(bd*Zy>pLeLXhF(5zZ5R<1HTj%H`}xOX@PRd(
znMTQR<fYkIVK9r^a*lt0@b%h53@YNmOi1FHAf5oS5W^5b2eBf86_-#y0*W4d{_$AF
z@KvVGJaSi|^7s+n0`)8pa2Gbg?tZwT!v*(q3A4A!Fv^sypXNmI-`V2Gkw#Qfw*MdO
zy=gqxd)xP`R;^~0Xw-<(AZbDwDx@@^c@V2vbJ0MFCW$0Unq+Lyq<NAkMVd%yE;58d
z#)>lR_u)Lx`@Z+<wIA&Jb?qm6@5_VhTIX8)|G(dNIEK&hIX=g1#TW9;t!GHKq__};
zqK&P6j&R@6*PQqL@v~=l3a@|QXa`&{q-mYn6?w*{Z>LYmC?bF_Mke0C0Roy0$JfcY
zAYxYn)5ym6)XnfKfAG|lN~q#i!ef^aERvwH&~!}8n0uOfxdE$y^x(tGAfCMYV2UE(
zkr=SOoP%Z6;9ucM3BF9EwkLW(Y|O{`%Iu%;4MMsjme5d{IS@ei1oxOL-#68oQpTQ@
z1`&8}1<xxH&6t}|ORnL@ykegJ3G<NC9cc?<IMk%CrUx9jelm$}{`X<$use6>juU<S
z#_pp+X@%NQ@V10G_fMZa)4Etj&jiQ7Eo+bNcld+K^TvfzcC%f5uOaWD^Q()|8*x2#
z8KK8DETAws!Qk}^uf8;bzx<#jpK@%#+);S4o)ZHrm`~TsDHZ=@xAE%<=5ZEqM0!t(
zKCu|rCqADT$rYZb-Y3OM_Gk+V_zIs}ZD4H(ANc`Qr-;qZN>nY0Tc~_h%g8D;uNN*}
z+|Id}WuD{sY{-p|4o`3aFNt;RYJa~W+Bic#!2hFVg@EH2guV^ZHZC*TfnY0xUd<d9
z#3(Db>U=xWvT~h=s*d;~8JL1GiKvMM|Gn;3Tf+;|pRpO^lI~LW3_njoca0Bh@X&aM
zm~S)RfW#I@75qB`_AkNRlACZ3A0Y=xNveuu=D|O5dH893Tv9K7{u@;Kf9d!jfp4}f
z5aTM0QW8x)?`S513qr7V6LCAU*{1vkZ5$~iABL;W;4mni@K#(sMxVd=&dgNK!`-<H
zs2Po1xVCo_g>A6lrgZ46*yDc#_56SBDN=-lzGvJYJ5o8zb$T91TIU?Ewu7Hxxz0*8
z6EPXpO24x(Ex~FgcUJcA;;DU2-AHl$3r(vAb-3B*ZgL+xt-RU&@(F^>*gv0&CaJX4
zSd1+c>q|UOG9wDGSOjt@?EVe>Dak|;|Cz{XoYo<d!uiB$$N)JT3riP2<pKt8p|$1m
za>{+Qe*ENybBnJK>0l`2GbACLbyZKOLs&YBU?+!-*4s7K#p5E7omeiJ8^`3o+eDM3
zSNs(y4cZ_w1z75~WlskIhR?xnPciS<R7G2RT=)p<|0GwLRnxi-Trg&Aql-j}#a|%|
zOnz}ntu^^ti`d?ntka6Q#ifwfbN(zw1mtZ<Qe9T!b_;is&N!ODE2?WAv`#U1zA)|?
z!3O8gKOmrOZP7P|@P5HU5)6(A{hQ9RmDsQOih3-LULyvR#Xft3ajJ~p6lykjL89-W
z5(F%+`Kv)xNOP-x;}-9B^zlJ$W5RZz`GXG2)ek;6ILfn^(afTMF2NsRVt6beX^ER#
z&grRVUrUSx0j6Sp7w2q0Pw#+a!U(12H!GZP^*f$IZ@H-l3R!<9DgHIamxM0lIrH=m
z7f%SG#b?Yqz2i^$Z-lL#>qhIIjf^^TF$;8bP`aLfXR;RQ$0hoAL!LOHsFHjhPrI0%
z@@yHy8nHF_yV%BC_q5mPVfTbk#CpdX#TKNz5-dy{UU*fkd&8fB%&wr%x|&4Pf8z2`
zlOD@d>uYa4KE@(*|BuH<aHgLeo@fz5k3U|9pDK#Vk7Nq*I!>!b=NaQ4J$}LwoFNX{
zY%rx)7u{X-A>G6cn6+O@Asx(P{3ILoXLB}u8p4XP5=tFw`3TbRsVP2&B@f!jH21^+
z$k56&OpVkf>H>>(WfoqX7SqfN;LgM{Ux?GRXyHPs-_ue>cy$)<ZtH>jfofHB!L53k
z*QlM_r(RYHKDXhbObsWlVArNT&YdvNHemGDI05N2y+8>}jjap4`bKKLw<qDbO(3$n
zv+|?zxAXUY0w2)Ot$}?~vDd9CTu;r^EsY3rdSJbllcvvE_X=^C((4Gq2<3%+`$!3w
zW3dcHsL`WG6Zw;A7D!r4qo}})w#j^1a}@ap=N`X*ufM0K<?G=M@R7PFT&9nFQFeK7
zQ)Mr((Gk|;j}FyJfBbi^AkfraK{K-J**Rsn6N5#)UE`-JUUh=*C*@D*-~B)a<tdyp
zzKchvUoRg|FS85?!IhU*@zohemPPhn<SR%czrWxKkF0p3Ne~=UMPuduhKAMd?QCyi
zk`c{rb;Xe=NMbUkq!0b*@M`Mx>RjX6-g`!5d_MT`L9hG+6HYO8nfyjZ4stVk)5%Ph
z&uoa*pc3=eyJq{DyOQ$1XmU=y^nbpM0rNI1XZ638y)U!*p`%BK1Lb6Tk5H<v`4x+J
zl4VQ+Pwqx-3&SRpfTgwt3QDXg=?5||hqz?W4|u`Zkv8|-{^^R;Ig=a<`Lp|4#J>GT
z`3m|D;t}JA{<h7IP{m1b=)7${u4l3|Z{E8&f`H{J*q8GMY-#QNt~FMZO74Yf++qj%
zF2jZl;&RJo(Ml!LL3xB=Kfg7p*5+^l3aHzoI^<m5on-5<Bw*WVSupZ@_)9Ql`yKld
z<+y8d-2L~VjJ1@|Qzj_kvS2{F5+TX>VY3--19WwNfj$-Sdf3*p08}7ZbgNZyTl=NX
zuYP<IW61G72gw3D9bBzN=%#%moo8F2waXxA?4nkxeSGw1VyD9c=EiZqCHECPj1}0W
z=A%bUDbss-kTVyxwR#uLTDekWuRL0GTK{5fuOWabVD%E0XY6PrLy}-mwjFwW-g`cU
zUn?8(lg~PIKk1N;+Pw?mO@0%Z%k8FlVlb*)f@#rLfwTh%zy}!N3<woMPQ0ocC$ObK
z@d#p5TEa!%nB?GRUdnv3aGXaIpJD(e(fPvn4F8=Mo{zYhpPzNCYqhXLRP~BU+$#AM
zuMTrl`q7I3MI!OdTONvl_ZRQgc$lPm5}1?3G*zu-l<T2Eizimpg|22b#pPtcc28)T
zYUc9oefKP8lQwr7Gm*ff)2e74zi{wQBJ1`w7j~1nRb!5|Pq*4XGIN~ogXv)wJxVkT
z^WD{Bdg!zD7>PevJxGp5i#z}%1gq28K*oPg-Q-WLMStzCIZKo^15Im;TWz_lmOlSj
zX@~!_J8E+!8NB5F=7EC-5%e1Vb9YO~p$j)Ny`U6@y!_H>DTb={hh43TV<2QSymEYp
zZ(+OpKh78CPQ`^1>N1KgZ{yn%=Ws%_G_UO(75W_ikQv<LJ*ZUc+9q>AnPl&A15KkX
z;%|Na_ZUyIU|ut`fg{sVQC%Hkze7D{*<2>L-W%KNm~yY<%u02{sd?IbJH9K(Z^MJu
z=E_c~i~?24)yQ4~B1aNa+PQsM-+81CfsS_=3h(*RKAl8{Y&oIBDRQV3b4G|66@4M4
z(#W|eL&FLAGwt@^%ZKOXZ=3;COKcE?L9c&dg~J@@F9%ZYNzOjDgM<QmZ5%{b058^_
zSv%z&7#Tk8vpswq&zd*CImqf7sSR3m7^t4};SHJMKBFH33l)DYZQ9jS^4++Hid~Hi
zqD;<Q-ua+i&yvmso)-pRv5kB%Jx-J~xk`Ckq;>d8Eya%x@3ii{%NxmVof^7Fn&5t~
zlb-RlNs7t>gZcA>=aBEvj$XauDnUo}_YJY*7DR~j3RuU;myaDB9J;V>=~X$pxJK=H
z=S~me;xLP2SWQ(pj$FK_!{5usa}nVR3E~DA<d`+_zB733M5HTlO$jY~`XAp=2<5Jb
z?#e2aDF8;2YvD_J<fLJb1{b=ixQ+Z~rCAOZkh4smuxIzsDKVS&$VB#bb}pB8%_Y^e
zvWMe2-PU;_;hNA*{>kOWuK$+01k+ZMUSiL2;t^&%_Y!!#kPDC<rs0QW+#o0g8K{VI
zR$W;R<CJXGAphJI02`}_Vy~}|ik`fA0py*TzM@rAce*?T2qS-7{X`B$*~H2QuNW|h
z0Z4w+zPi#t9lo0jhQs*-1<fejZ?kny3#=CbHjHiF&r{vVQ+r~to~QQLncDu*4-S@}
zYqzh$!VIC2rz$US_<@NT&Swvcv;#n;^Nq%mI^w>5GY2S|FtC10?OV|<dZ(;IE`rl)
zilP?+?0XqN6$!VB3gyB@i|FlNOdcw5EyuX*w*gQCX{hy9lQ}>#(5F@a+(yRy{>HD1
zfCnz1igSnMYun`;?T}q5EHsG;QYyv>r?3C8!ubQfL)FNoLVN>%!>bdD%8Va=xt3A#
zM?c^2UofZm<;39#>aHP0j%Dq9K|u+mXeyU6wHWy)`Nhyo2?nM%Oqg7?p%B-ISzT84
zxCN&2CR<6Q9Rnr`+Jb1~LYz&8wop61BiBPkp{e(o3L_A97p;a#+Ra3u!X<y6aKZc*
zz8<B|j+{7XJB%ReK}!+VSiuZmplx;6M6j$f>Z%Dvl3pL$otBf6Q^Q2Wu?xKBiP;qV
zZ_@4L%{(Bk;?Wx?E+#4uo56d*`U;!$3Y04Y+9fq*+$;;ex5l6WgMBJuYpaWj_wu;N
z@T}JP%crJ*u}ilP&F`XYge@#y4+_y2e;aWhZJ_ev74;+@!FKKIV6$(Gi>EhN)5cT#
z)=X{nk^^Bx!{)lz>1Owv@0h7;Q%0g+HzUB&5&c;q$l~%aj3f=R?!m}gNeu?p8`NeO
z%L{v^ggNW4ZO^fIK1r`lmn}W?^4hLz;y#Lx-Rrz;LH1uR5>=S)glOzU01<whCmmk3
za5$9#LayE4_xo@95U!?OQ=KZ^OP!!9I$#e+T0s-N-U5~)h-}L3eNOk7`t}@WEnI1L
zDg9urj149}-Lk(T63gGu<=0w5TXLDQL)H==WOQ<Xda3%Ud+F_baHj3HbGFUvA1>1H
z^;>#Zs%!*2_e&o0RQ6|dmqQhWg+MLi)+RT&Q~J1X%3lf5Hw;IDPQHii8&j_cgBiBx
zHSZf2J<hCMWv6Zhgrbd&{Rf^ThU+)$s%A~1pl{jT!I4|o1^3YYw(scCKMeTRNZ(2i
z%`RK@(Cp(&tq&$v#>-^T)}!^f7x&eo8MR*d^JJf8V_fDc+4;4xbGTx=(z=aZ?*(^k
zJTF{*5Us6lXt)`#t@9^$cIt@{07aW{9P<^YIymcFHhmdkHs|x_c|(UOX!O<6D&!#n
z@8*<85{Gz9S3EX!#jGA}{A6r8^+@v@!I;gA*e$=1_pg4ldS+YK_ft6~$;0x`g{;Ju
z2Z3!-Tgluz?P)VnWWWa-CS5s1NvQ_-&LcM6h?u`5tBFd{oK8$UN&EDYt0N3aax(uz
zr*NKq<My@<BqAiZZsvVPkBJQJKOOqK{M@0Hw+pYkG;5Gfpw4;jQ=rFb??Am^?It49
zAgbR)4GT#UY;KhTChWd-H6LN<Hg2P4X(~tI;Q$q*@zRh?S?s}4_lY6P&L6l$e*-as
zanxdh4O>EQi2?~6v3`iidIrs`$~n5KXZ6^%I`PqaU6FYy$vydu-Z<`z1b|jDK*l_>
z$0XYBAUQv<S$zGXFo!QJl&dp6vR_T&!jFVUaUWuTzq;TxAkNynvq~BfrlZolKJ}D>
zAG9k|ZnQ3Q+_cxt#=4x)AZ)}z_^d4A35d`4J9q)p5i$+fSQ236`Q^*S`$PBCcLy{X
z@krK(Fksuyox#NcsDaY}pU_og;_(vA-fsfz-5HxS0qKaoea{x(J2|>y(|nA%m+>qm
zZ6{xRxgg-q9~$=8v<cj>g74CfE*_QkleD;V?X9^0tjp3SWWIa%jw;6Jk=kK12;B}`
zV7!H4&bH7d6idFdwraBpX|aqgRn`a6!eZ)H3iAGkc+7@y0Jn&$vmS-4l{A{bwH%cG
z?J6#gB*IhMYB+T>3Q#DyMiHZ$cV3q&Bf*<#M-Pbl0Nb}iWfs&pT1xj{NXDzy5-o~E
z0rf*pi(YA>+NxV$0^~D)``Iz!-ELckiaDW7isDBu7hMc$(>peK0&tB<1`h7HQihuc
zew!QmAn<yJ>fVOsrv2|L)s8(dk}A4Nnvctd$BLfXt(DS89<11R;KwK>yCt`{Qc_)X
z*(3L^(zt<H!2bP%2=MYwEVt;PZ9Z)I%YUW4{^#hmc=)c9v&rVIZs!%-59r1k*{{Mq
z(Td@J;|_^Ph2@t%lWB?DcSDD`nn>!>p>U0oZ4X-qeUu4Y70p_x<jy%t{lUrS?JC!@
zqZQpn@f=9o`9n(jL039aJc{D~d`Pcp>lRVpEOvre#2N=%P3(9}H>FYocEE4{iZ&(8
zsWR71V@j^y<dsq@(}y6i>wDnmLDx8u9X7Yw{4_a@^rDBG3+02bKL+GqDJQOSn{*vi
z+zrIN_U@<LseH>#Q$%+qI@_3#PYIyMxK$*&eM=2%Qszy&Y_W0a(wP@|tK#)3XD<1f
zQ;1P(qN8#2!NJC!+R6ti_8t8(O2fBe-#%H!(@tcGaV-ZmhH!M0O8nNoH|WQZ*|bu8
zrizCzK@Tg2I~V=q^=uM<pLcGQ^C|<zG(9lh)@SY8lmCelM(V2NE+N@=x_wu>35Cjy
zRf@!L%~z4Z<~A{%-ZKz7rE@1y5+kaYy3MFhc7c|h^=(-+(+r!KoEg0=BWwh993tc?
z|4hj!db%cP(g9_^_}d0Lt5mf8X&wE@u<yzwC0irC)*R&BF6$T0czR}})uwHGv!ggk
z_gg-0wmWj>j+&x}q7Fy0GABln>}_v92Q#m#<=HrGx2=wky|jOya6h&>t_kwFNiYil
zEXo9%DR?(1N*L-ijdpy8>Pr~n=|igKV~m4A+KcaN?bgnDc)|XOnyO8Y-BuKj`o}f)
zn|>OnJ$6^`cj_^wV_ZVF4=7PFmi%Q>$tmONuk<<WUD=f`7Cv7!_bz~(F%E&!!bZxS
zk=WAcQ(LA%+u+``vsrylQhgGNLrrw;6k8k^VVjhDaYFcrxeaT2=_{z4RW)p*R=a#p
zOxBQKs@y|eJraiuBYzDZv3B7oDE<N#iRM(ORbF~+%`d*u1}jXmxb|D>SO3WqNh{`r
zK3?%C>azq1e3cZM!4Eb_LFSg@;^(G86;nOg1);Q*4khgFp_csYS&R3HlV9u}WSVY_
zs=syBTuGnc_iN*dKMrI(BA|!*%{<4aOD2m?{5|GDl_H7cZyMVkR>mC3z4~(E?v*m~
zcx@IrQrrhzg6DUM`pnbscQNmMZfeJ?24mo|O4x-@$*z7&0h6#YoCrE)*_D`*gidHu
zC4CTP3er41j0d6Ls9BtD7Mgwjo3G{8SnHPca4RiVwObK6FWWj}!}alg6-wPp+!d1^
zOn+Jz_`a29tJ~kBFlVESIudv~Evf0Ib1K6eor=5S{zxZW3JEz-6euDX?6Pv~N?V`x
zbw?Z0T*!j>s{4aay4W=iSGiffBD>EGGh4Gu<p)qoK=SMx62Lb#=zS$u>_SMo<kT@4
zn~EnE_{1;$ai(TXLu^%%_RY`Rq`c-c{QDuSGGH%(Putu@Z}U1V_46u=G;7=Nr@3Bn
zY}|%(xt*@qmrZKZZnIw-mkpVH(T{^h)MGZKpzu}UiVy(<@)Go3odC-;)9pgx*Wu3B
zZY=(umHjRr-{)_xm*8^vkxF@}gp4s3%a9lBT2t(i7bMw7_D}Mk`^EqZ`De~XT1+}#
zn~s|e?<85l$1DCL{kj%~X}xZ(e@btr<)d-^J+=K()MK9CJ6z#)SnS2v_y4LVg)hsa
zRSS%0uWkBC`^Z~4M-CNMevcgUKh}-JLC;=U&+*jVQ7(t4av<<~JmPt%&6P2y1-xc1
z^mo<_Pg{1V*_NJfMDK2GUNMOg0X6AwW`S1H?Y3^+pM>o>teld6xaV2dHN-?7?V_HQ
zG0xnTs`J<m2A=cMYHU7Fi8;Q@?v905$N}5={Z(6ESwjIM!>wJV(H`-Rv-;^2=@!x4
zb<?Ec^!5}1#qtu48{E3&En$???Z)|sC7_@O#`yzp<rjXSGk*VI$o9b{o%_wyUK`OX
z#zjN{37+J%*lof_y|grDJj1R9$0A5ph#$RHCjUxr`RLh`6vYo9hcf59(VqH6@AgJN
zLf8MjnUQq-?RHtd#)b$65?~Hlp)p6E$fh=~nYMS}#q0bUB2m!<LK^KxQ2E76-rpbO
zLqV2bas$;TS+Bs)GJGU%|L1xA%{oqAlXk7xtA*q}THgDDZma`F$X-TXGp_24MW!Up
ziCrN?$k&-Fdh0*xpJZ_J(DdF~TIaI^u8(=VJbjX8$G^{CQFSaVIJC!k|LFay(=KA^
zO?5}RS+q2yir#dU8#zkK?uid{X+SJq`(D^a{aa6-p>c2sNoJ9+rHfh8#_paM76^|~
z_+aJR3P?vPK`J{eUgE!R@(bt6Iui!>Wc72+;!ex6OOH_1e0)RKC2Pgj$#;})-MEo0
z19hJcN|s_2IK*|%H>YI|`LgqE2S-P_x`#Fd^^8X(Gw!UV0C<4i^2;QZ0=O|v(lqT>
zV8x4i@y>5x(*)kQ#9AI|<335_x8FPfFP+%!DkF(`nbz9n-9Vy7(*Oe>=h#XDwCzos
z>YVx`NL+T#>v->CNN8G9PUnGkjBDcF?)A+zSuih5ZxrD~?<HWLj-NW1rnYNdq+-<*
zyPC_X^~W5*6kq3Yoj*|TNNLk63kJm`r$2nop6^y|fk%3k9B8d^u)-@)?oX2nqDPTO
zL@LgukP!B$;imO58_rZ9{Mow9ZUbIR&8^?pXqW%^RXzKs$A;^=dT`3zZ-`HJ*>r#Z
z<fbmo-{72^cAeviR=o^}NZ*)$ew_B5!50>i95uA{=<x@wowA`c#>l{Sz4Sh9Voh0{
zB)u%}=sf`602XTZ1_f-x5}E3{;shRb6c|r)`uEx@zko#VSYBITfs{hwV!_BV{AIH?
zYv$r(2ds9Py<k}q8|8li{7Ay0&G>u?V8#t}{5#P5z9w3+VDx%XCH&tp7j=Fw42Z}}
zwk5Rf(KTs(W5Gw9Hj%=6TAoIsYN3;lqaX$Bah}n_?j%T@rsbL`52jxg+r3XH2*P>-
z*7m(n%|!bnRD%GY+jKpHDe-hM=z^5lp~94Q=S|B>GXh6%9R<Pib4UpW=3#F?>Mk{q
zT@pVC*^sa|@PbK2+EW5wq-}cJ2LdusD2v~#?yzyz*N=^jsU;P!&mL5=udl)R0&vb<
zwL7lDu<m_NXLng7PDU#_t{{A`(UH#`#@(YJ)A^3msPfeFJp{d#w%#3o-7nKslQKn0
z@L`Tb{JEa$o5N-rwX=1a8xuf2B6&hm+yz0d5uc!#phv}pry0E=5w?Tv(rCUT-m?tE
z(|ctVy23%L65zajH)(08{OQVZ?fFI-4P)+4gw^3P<h02v$pj*m5#2f{p1+MjF~x-@
zmRk#eGUVYto$9xsg{Su32L~NAj{N)hl`5qZ)5sX6yiSuR8k9V+<ijYf0Apj0R}z<q
zDc6fHhn3nt$MjE|iF(~#PFc_J%=vO(pS&jXSAEQy=w8>B8P~SC`U`WUSbM?<WW*)8
zZ?o!jj}(dMM`jt20W_T0`V4-bc(AwY*w&TNmk+hPW#L6jV(tn$`g$jIE~mkLTIS?z
zoFD0hHNj=lhak$TH0B=tJa*j3^6YD%8iPHvpoI<lJ;)|R;N#yB;1kpm*O2LPmJ*Ya
zr1Byt5{X6LEmz6n;$9k$OGIEMuIrVHd>-v}OCjb!OMMG^iXEne5zPgpyNzzlFVKku
zHCC!U*rT<;iMC(nbB7i~?gxX%!W<@d!%;!QWaHjdMomSd_i`8Ee``1-*I=c?@8i&C
z-84aVV0+jxV>j2B0AAttYVC@UsU*I@P3~@DuF{>d_SD}Sm9eu^T<in5jUBG!DD~+y
z(<JkWTFHTivPl7ABle9CGC<!dEhWG4y6+#I1r}*GTBFDC;nMdCsyhs(AOntS*?%xB
zjb6PL102XC2tY|DxVJC6zMVp3pC}zqXZ`nbCE|Lg9<fn4Zd1M|z>n7CNiXV|CcVj{
zs{GVQDz(2J!C)OUEBpOH?y^%=C@?3aAb_$ip9;|jXzuoqG_iu*CHxPo7*0m!yqU6s
zlo0cbph-Gp@3?KHc@>mAZ)L{1DJG%M$qcii@0zYfs|HT!9|DZ+!^l{WPS$Wq;Gp%l
z^8m_Ayig5Rz3FM|G&nIX$t#aQ#|?Dys*FT+?J5ns8Afj8RGHCHVy>v)5T?q8-^d43
zZt_rAnR?sR^SDKgtL+Faw2C7V2gDJKt6VADqnp<Llm4WqzlN3cx?;A8D-zzQfZ%&=
zvVZ}<XkB$)n@nc=C4W7pubqCJl;wX%>$KY}s9H0Z;hD(9F09ZU9qO&Bf4T9gz8Af_
zEMS%WUL_b1TJ#;&+^ntsGQCNJ594}mA=-7foJEz)n8=xoTxXNa#YjKl`t`)YZ96AS
zeeGM}pb;~#SIqIncRu#eh?$hQ@s?70ZJ&vmeUk<SNO?7{-tOp+E`*KST(=&yvYoi1
z<DSZ34IvUAyYiR&LV%w}dnhbYU}f+|^#-^D*u*GuJiv#|yq)gH)h0j<af9qv!U4$g
z8Q%YQF{LBnVy)W5VMa9L5XE0|OJYb!*`*YJ=D=sdxP2um|GRnnR>u3_rR|7ag_F&p
z3i+1uyx%GZJKDr0NJDXqYPBIquH2%ib<2oP*j7;s1KdS9VkP`{=VL<Z+c$fCCGnY}
z+nrb_wL&@&W-SRSD?@PY3X&xBEPCrCGc7qiJKp6Dl?WBr=AU^Ku|GI(DfB@2k|Gp&
z2-O<~ul{^7P8*!C3bgGDR4JK6O$$xiKl7A7OU>rjxibW4V6`{K8pH$5wsmvgdI)bD
zllSLg8-q{M#IxhCX8FXI-u*TP%oId}uI*-<EH|)n-vQv2n)W`4N6Yzhw>Gm~Y{icY
zZgY<49RG+mY7#FIjf#~_<^mAei1U<@9_YUUvi%~c7aj%4J`}llK>xChLM4<C0_N5B
zom$Pf;($jU6EgeG5)_FJ*F7EYkl2h&uCEmu-4Da+`@WBng8MqX<k2E_@>Dy}52o4W
zK&euRK1HJ1UT@<#c`4`IO;5)<Q)I%hcnum;Zg5R6ZrPi*d3eU(5biSVPRz3{2I^66
zKbpg-7CZD|u>bU`4fJCDqNvxz^4U1ka{o*F4jpWFr%OIrMP}DY5zg{?f~A!W+@6*%
zCFj`MKe26MZ6LGC9?~G*%JJ%P%f|pV3=~3Y*R(wQf*!TUVw{vK_aY_y!1Lwh1`;Yz
zB~@yE8I~+^otp1Gy$m`TMm3ePdONM1J;qfoFK1W{D=FohdmH*FH-(frB*~!sqQ5w+
zeMH?Zu>s?H<bP>vRB~YA>n8^)sJPuT?o_>_&BO|TzNXEGe0A2pwj_1=TD_Zx_9)&R
z8IB#2x*A8Ni{mOnb8s*4Y^B_mzku757(|xWdOTEpuEDs~J-@<pc!d`*-PluhCzmbU
zZC4d$nmo{@q8LjZU+L-qGDpOGbYvU%FqPUl&D=+UcU98FK{nm7)&_VC?BN43x!qMW
zgJykyFPVNJy41qKQio3<YEu3gfsX^{cq#uHtfU!)5XaoT2y$Hiu(EIJhClN_$lD_z
zOPMNs`ujadYThA<?l=2-0haAF*vqz4nx%!=mE|<WBh1{;^(QYflvLZTrYzVm-~8Y<
zXce*W+5NP5nA-rd7K3&U>|toQ83%z;;9kwI3X;tum!L<!_2MoYXFR?lp=iAEo9+Ss
z_A{Ew7Pgt=yrMs7W6ckZ$>mV)>0<bssLh@fV(YQ}M-QH`4zW_Io=jKEHva48s*;C7
zuHZ&sE3R@)DVo|<Wwyf#bso4N5ggF&b7OOOG$`_&WTYMQn{o^bBKTAO<pOxo0J7h}
zT6?x1%Y!d|dWbu)DjFOfqatz`IK4SnKgd+tiXXMJ2n&V4wBo~0k}Z;o(|jQX0DRDV
z@=2x2rGCn875BTM-($$y5O0YTA1Fa274g>Qs%w(;b8T?%W{txuia?l`Lei$>eS^*(
zf<BSuc5b&6*aY~{o>+p7k`fBR#DR&OyIWG{!&r&LipOjo`Ya_T=;J+So{9jxmJ(Bh
z{`I-A8DP#POZkwFc!|OQ5z>In$o`j6ci_OVP&CeUV8bluU3Hg}hn#|FKJcf(AmTuz
zR|q;V_0xfY_;!*xh(9k$HP%uk%9(vrKR9vrf;a&gd-Y8s{}TNHfzKFgr4N{b&C5Bz
zP`-+@7r6fY63S-Z$#pX|uB&h4B(rEL?;P$mY$w#v6J5u4NTH^}W5}=w5a}Q)QgXT=
zkdO@HAcmsk4yhT^d>QFd&_4@%S<2Cw-c&uN!GX1zeJ6?qdo<1^p4s#tu*BNHF(~=m
zqotmGyNSYrFpfF#tROY>^DwM9)y}Yysgs_DGT~`CprytTp^;@A-#E-b!F_+J*&K_R
znHyG2Uax9YG%Nec*HRna@3i8#i)93w`sue<vz5n(MNE1D>gsY~GZE2=uA^tpNd02v
zP*K&?(4Rj$@?d&f)g($?lY&NTeXHN7owP72ryQ7WdwA8gtwef~!A&hu4ML~6b!8o<
z$LddMIQa;5++vwxah(E~MpO?4UsEN!nXBLSjLF{jA6ilvrMcm?T{r#e5;FOiv7*b)
zHRHZ#Pg}5VX4uhwU+0<lKaScFw&z6NWv};BZ=YD-d%f=IPw$WKyZ-uH(8cak`gDIW
z^=tm+QQya})LHbjs^(D9uS&z=$&T>Iem^bVSiLUC(J0q3EqP&dZuI7)+>aaVoSxta
zg=S_N@7SF70!iiW&xWREJY>lWb?Xx4QyBPcdl*^cNA;-Z(^<(buJaSm3#-S5kFb><
zkL>gdu0PJIwj4~ahDb2BQAxhE2x3S^(N8lPC&89`4hWieg_FTWaKB+goq;1bVT!@s
zOD37)&dncMqSAkfa$+Nh?pvz*a^r-NTugcN`uIRS9qSU+WjV>gUEfC{;h+5oxwHLk
z+FM?dr-5gc_fVn_htoD&!n&1cyw*8d(e&Wprr9RnE&DU#Z6~OQbO8_Znm@&9$XuA-
zaF~|X-`;zoL=GfQ2lF(y%=dqAaKyvqnSE_uX!^Dk;+p?7F!?oMCJI|J)k^BAdr4yz
zKXqO08t(;>KUXXLMy4JnHZDsQDszILzeqgww}QE$aP|55Z|?H{MGL9_K9*K<yzoB-
zA2fyAtCNkX-Sa+%G2KG%#nh^$bXRcp4W!*Duf0Ozc9PYhOH(yd=s#4jN~A033V9TN
zX<FA}@qcrrdwaq|j;7Npf%GCUkA_!!1}dzEypW+KwIZMbC6$&6%G>Gj={2lDx2S9w
zBBl@d^kTc7;}5MSg4}sy`Nx~oDJ{oOF+lL($$twT{1>`==TE%5`#)R&g;ryvuY#Zd
z!Uz6;?n5;78KjdUS_TgT$X_EV{pk$D<1Hq}&d>tSv=}?K$+vW{NMZ2Ja-m(VXK}I>
z{UALAE2DDd0_a@Q^)hvy_M3!f|2c#eA6D0A-()pBrM9oFP&a;K<2YgN9y9^;>so!r
zY?1p0Hfa@U4naYPXxsMEv|)W%RD0C#QYc%CXIFk~<*s*$UF%WT;pb_s`K5>=!B1k7
z&sp>hwGu1XO@}~f4n^?#iZAHKp?08R!PRMXpsxSFJyoZ#1XlGLT?(TS$83Tef-#5c
zx4j1y^*MZvxTZW+SY0Q*(n*jvcrP|FEn|s2Ar9E?D)~=iRJXaJ!W6GEJ^Rn<I#Gb;
z((1Xz^znw?(OWcV`QNp_?hD6Ew^?IGtW&n;9|8rTu;j^SD2lls(SzEGh}Kr>J@@L5
z8*f?n0{2R-TU~%vOx=EW577b`S@)0^g)G1jRZ9<oF@}ep*d_7}tLr{!+W|`cM3v6N
z9@Kr%(f?sdRP!0M%bI@B8%{T~J>_+l->jpOg2IA9o#fr435fz}!(6leJN(m*bUDS3
zmhaPSP2|>Qv1Q%M1=iUe|Hc&5CV&e#77F)XsKM>!8I~^+`5%o@kb;t6U)bCx_gEL-
zO0);0nN~r=Eb{C%26~W0-f-F`e^A#)Vc`V^FIoXQ{U`l<*E5?pQBbJUZ?NzpnkMG!
zKe|!Nb`2FmD(bdeV~qDGXM%P*xh4nRV+HrgB3p(1Yeh7^MY2Yz#HNR`kb{xM3~4|p
zH4#WW?E7We@FsPQivInqpM0Ty52f3@q!CY;vMQiKx9)%hCm>X-4{NG0<|1kCspH3y
zG~J;awD!eVx2mkJZB&#L{8hOV)n6^$j^t^6e!K7A4x_@3+MO$P#x``B^TKbqGukA!
zc#MZ_>kdE9`rleH|23N{?o>g+M;{Dc#0gU8knGa_%n$ru1TOjc#6c&vm(xjrE-lxF
zhIbByBaSgYP)>?}Hv~lap^!yq@HYr)zVnNy!KIP7bAz7JQ{O$qx=JWK*pAWHsXDun
z1jl!6sc=^(F^oHEwR8%$h1`{EFdKP5pk$@aJeO)4x&I;MM4Hd2SkNkQ3{)d;2A%Pm
zr=+3m#3atG`JRqHKzB|YpK+$Of`V6_T9N-eX)_=PQj)Wy`mFAAj?2*m5#opRaH3wu
z4zhnRy&_=1x7U_U_YMbJLSeReUNf$;&k(T!R+*eu$r>r-gH_yy#~}J0STKb4G$Uj_
z7Tk|XzJrD$9ZT~hWwy64;Xx8m+%#VP_TpO)rn}QiJ(971mB3vP)GoT!)^Om3Lr5kI
zlo|_zRaz!|eI2CNhIq9ELgVsJQ9t0MK}3))WXBIHZouNyy<no^n9ZVZoT=d7p3=9)
z+F1%AdW^I&UfLrMRLwhF2!rqz8fo}ZJLsNwx9^4Mf#13d%7-`~2#fT<bk=uJe^^Id
zwZF@k7Xj0O(<TnNCTY50=|A)%E~!;<C8MA+giNLjbOSwz3m-oiViNIO&5d5l@GfcW
zNklc4q02Qlp99XWq_DXi=lVi|NGO}8V6{;weCbQ-Cw7(A=#aG#O|{_#`O&4bu81c8
zm*9fYJ~M$n3I9#COw7-B1E@oBIz{-ed?D-O*jc80ffJXKvCagxicSbP5AF=@Udpdl
z9-O%<2%aKOFn{`s-Np`VT*tbNTS|XfxwN-OG0*%9XhKr{7VzYBGF;Wgx&HqCd@G~Y
z`XrrB?=Br6Jhng?Z;A;_Wi~X(oY9*gm<VnNB5eQJ)-9VBP-nvi3R{%wx<@VcLu7z?
zsP`?hqFLrU3bn%+4j6LB?@9^;idgznhFG3ztPf)YtYXc_H4ndhjQ}YQ1)N^9z6*`J
z6ehyH_o+!d!0Pz)tb67obafZ<#XeDI0i~Tg_KC<-l{qcUOWb#=#8HnlN6@Nm-o9sr
zj}X~B$oGH!^T+V*2Q1wLTuSbIWN$q=I$&i{pszx#;!(iWDo6-HE{3Oujz2sQ=d3u)
zg*S$!yoQP*zEp@iGy@cK^sG0ft$jJ6az?j>xHzeeus}1|FNn%Dv|`VSb={mA#Qq3N
z_Wo=o6=z6VRmbWnXns|y+BI1rj_D$rDNE{Y$&?cx;TeKgE>U$wx}xEUeNqFzTuX}G
zLG4c!tnDX1yg|yGpJ&h}va7%?#jG8;v*yy;h?i0GO)EH`*q=e>wSc`a00%Cu1rm$U
z``1BSZ}i1~a#+lMS||3tuZ7^)%&BO<n41U>{FZiJD+QmG)OQpguN=`@NBpzJVL+#b
zg+d|?+9|Ej++v5q*HpG53WZr9tm$_(_ZoE-@)f&~i2lU=-1x=l6i<IXObc8|(e9yv
zwck-9snvr;lx54Lo~0SVmHQDzr!hX^rHqKP{t!&lPlyTa3WroyY%YefbqfExX0-He
zgt&i3CHgAD0*OC4i(pkAuH&a<Tbtuv{UuV`GU+6E&B)tO;#<|~p=hl*Yk`4>)RR|G
z@%Xh}@f`spHE6XDqQwi>wxa&{-)!ytf@hap$Bl8i>gi=e$nn`etj<u5_n!A1$T_R)
zVs2H*+OstOr(T=K9HyI;7TY)a`&AzIsdKV30<<i^Z5QvZv&I>2_5168yx&QF&-w$s
z(lWEI;^DiSYvvn#B7Ey|&G60PB2-1Do^6}#>}VKI9jN^^QGj{ZsfoA27{0EV9;%1&
zby5!&`>Aybnag@ZLoKx~>PbibP?`w4p8H$xWr*U)zjA&kXY5sK(Qe}${+0nmOE+P6
zvs4Lh_L)ckWwBp~gjmhpQOmx6yv5!i*buZ5U*$Dy_-(G1^y_&4C6&pDLT<WuZZs2D
z351g#reRSQuTmdS%A=XB5Qz1E{oFkO8qx`8y<y>k1=8tYwl+;lmDV3%`Dcchpt10r
zhO5M%(42{|9yq57n=v7cKQmj=a_{vBHDtGT4;y2@`KKx+`u1yWdDwmuEopWJfC_~B
zaR+w1^{<%n7vr2a#@Gk-n{lz*(m#y&R9&0M|HZMXzN-O;W&~V<a~g1lbZ}a4AY6+C
zKlCfaF=y<ZsP!#zdw}y2R&~LA?rS58FKg4f(JtZnzDxN84$pIm=|+s5j)#+`2M$go
zwXak~mI$iDl$3_8uG);}e=u{q^*dLHWOXf(FRey0NYVxat16U8>OGTEYcR6S+{*2(
z&0~0yh-0u}5}2j3;|8NTq{;~W-}=wPx~pAyD$v2g)}y{cH@7-Z`+!A1J?HjMOSj)I
zg)1fw7Pc_8X#enjoxUUOvHAHo4flC$A>1MFsAYIgLdrYzMKM@Pl(RUi%$dvqWc2B{
z0umApGk&KgOZ5qXj}8O=8G%K50@)WZm!>WmsA+vFi~wG$Oz|BmG3H^4t~rMn1DO7z
zoPPmjHZrY9@M6dSqVszz`n~<Aad&cyK?y4t8~>U4+;JL~%QU4cafyO*EG#m$uI4ao
ze}F*tF8?QvzJ)Yem_{zUKvN>*e%BK3K|se~+C~=^uR_d>Eo*0)EMOS*#OM<dYPC3>
z_1{0v1YS*?LzrF*i#)S~!S#^Zzx6;!$svw}mq<ZMN`Kpel}S7#=pWJhCq?|`s(KgO
zPb`N6n0qMi*mUBGo7B~t_6>?y2|1SV;fsC07m2i0i15v}=0Rr(vI$j`zxF0gs>{GE
zPKXpmhe0Q~u<_I8-}w?Sdr@-GvDo=MeL6O*c<PfZ9p}>Q#LG^;aQS}HWwJNw>=-p<
zYh9{ZPN;^XvhR%oFCa*9beDz-Bf8farWew?fsdg&_6ItVSK)Dvrg+f5S{nGmV>$KL
z;%7&=wr5@X<z5Eip;X$?aQ>i=6u0nWN>^`;QA;Po#QQr`Cwp_BD2(}Z((P|qZ$T%B
zyaubV7G|b(w_V64bWon>Y>4gweM&Y}^;G#K@NqzAXJ}kmy~mR8MbiUKKU%Ym6Bk8F
zW$j9Eb-#q&pHRHw1TjGFO;-Y*1U<8km1H%@m|uqy6Qam_eMOB>j%B`;^?BPnQWcdl
zae4+vz*olIk7&Zgzr&~=1qGq5eJ{1{hsCSm*vQ86BCv1#_|JLm_XlX(3DXcGFZEH-
z(iAu^Q~51G@{`<}lP5(0nQJRfT*?Vpgl)m!32m0(_bHq@k<80HXdE^RNLAqj7M@~y
z4WZLY1Ya`z2nPP7PPXg}`A~%{m=n&}X}Tol6PyZn9~(t6b@JVoD`BJwK8ma70X7l}
zOFabJ*SmLr;#w~ghc#&pnvPihzvNiTHwYCS_;bHPPJb3%W|ON*uq8vGE+AgfFU9)q
zrFe<g`T2h$>(F~m4&N7wl^}v6*48Dw|DDqPmlsze$^Xp{v9bvJOQG8a?CDBwR#-UD
zJt@yaj3?p6L2cUYUuN!Yq+sFBZAB<xC>vHw5FDZmEwx!r(y^o9u-%NofsLzZf%RU#
z>iyNPc)U4sd`^qCg-_6HSkupL|9{XV)C`4S6I|scb$wG0$EP!IuAW=!c9hPUu<hHT
zJJ2y%vEe3xA6c%&S}QBn?B@^mne;AHx9T=Ywn0I24I@V!pdxtOAbc&NZar~|fpR2{
zMXGn?M%DZ)>b}R09ErlT!v&sBxR1$z-m{dlG<gN$Vu$j^5X<c&EoE*eD%UweqlWuN
zP;`#qkKdK%Eb6KJYl?3D^4`+HDh(xSYHEL^A#@EI_R=Zm)%T~;QX(!%YT6OBQ0K;%
z(ZX_p_>^!fj+li?sg$|D5GPDn>uu%rDAZi1$<>?l8d)6gr(ogDvzb-j5aVjL`5VdH
zz_n3)tIChByB-QX?Y9-fN7{5wy7YyQN-9}aLC;CUbs`cFR+BUB^rVG6wocLu0*J(L
zm)B;CasAZehtmO>S*dmlFj8<NhG)#%OH`WWH2MiqjqR>MpV2q#k9GNJ59Xgn7!q*!
zaEEoigbPOIwRp?y$~%1zNpsNBY_UT*W0;;JLShN7#za2QiJw!nQg&Rf++Tn2*39Hk
zCHp1aw=U{i)1YJ4q9ee!?HmVep_ug9O{!0!!*$=Bo4G%4^%$8eB_pz<zrbpZrAS>i
zEQykwgSW{%I80xJe}gaUVwo!~{ZPAU@cn_I_ZoIFf|mT#<VDrNwN;C!O`BE_#N8``
zkq0Fp{Cj7Z6FvcKlFGzgCm!Q7rSV2|IjDm?y_QQ|U3#pTz;(@O-ne0Xl%87`D`_gb
z?=#VOzHfB>XX}^uwCW~a^>Zx2n=@R|3@yCtIX2N~ClH44dz$v{uk@J`K7H@^g>hUN
zEn`DS1X<jsMyW9VMa`l(0O$E-mW9z{;q5YRL!=ExX`MQ@{ym%!d|(Vm6%jn5?ijEI
zzUeIW_4OWnTRDsuk#pEKYUhhH7abxR%z8gI<Zg<&JXY-!m}D-^-;-;V>Ash#5FGdV
zjm&y%#(l>d^cKovdwR;cE4Hs}*o7FSyN_MzCs=XtNS~%n96PoHof0xyoWHxi^wl>w
zAJAK?_$#MJ8V3a=0AhIs?y&!pD0U)h8L!`~v&HREPeYaRtVh&u=rd3(XKlJP0QWOr
zUDx37^Ic_^VwxeUKyI3;wsSwn32$js@tF$}O+v(5@-M3SY)}x})FgCo-U_GHV5>ZI
zY~{06Jw6c74utGY&a;?v%wT~uLVCgmic!zmcbkWBE~kT?#m!!n6zE(qzAO+bTVCjD
z5H(3Xgf0L$7IBO|`$5D!vsd2INOeG;N_e;Ntu_5!-H*x^kF9UGUP;q;JBZxf+FuZ&
zv+Zv|J!m%YpJ~%xgB?%xZLKLICYZhUNUk^J8wu-y?t4Q5M*o|n{wMbj<a%(aFHu2%
z@{^!@+=NcaS)M|%4QVY1Pget{0p|0uUBYsqlx*YhNJBmZ^;2$9GDwxsQQ!wW(T)|=
z=<vX;Zg0D8__&6YyTvNlg7SI>7b31gnzyUSyiTHA$O1%~3u}v%KQXNaD$;kFYr?!*
z`t<+T%Z;u-VOW5YW~;FsI&Z0u_+LC{l6Lc%STO(~LE5Fw<6+oVAR7GBdaU44Mw^3W
zp#+-6V#xM)Dt{V0NqxwW9~=om%7~amx`D;TZX&cf58W+UaX`w_!L#b67juA(prWF}
zB|{|GBFw6!F>>kPhS?%QXlpks)MY8(3SiWM1`aa4$)qHtNuo*&7hR;eVE>eLoK#Pk
zHCST(j)aJ|cjv+$YL`1?K_I7TJ60%`7?;FB=6;d7EqxQY`Tdep(u*v-9nnas?;~A_
zv;|pmy$I!2!m~sr<^gogr3{`W24CZI5gLN%dg3G`BV$9bAa)@25)&7t3pVlrbh4h>
z(_o@OW~Su{50r&SRTMh8-kN!-bEN&+dgsqrBale)9bIz?Epg$Wc=HhH=oO(hNU#P8
ziQN6fl!5HNP4J!d%1+>9Ch0eTIdW-J*_9iCLNP7*2+MZ=^N@gYJ~@M#%DlP0CRyqp
zZBK_KhPnrh*q{B}#c#y@&T37Rcl8LW{r*>olb}Z5VqZZ;>HlrAFqA`I3Np1c*rUrv
z&D36OAhcP0K(oi%vg?V7jfc(g2^DG`#(Q*Ul0adEd+DT#Tx>xQT-<2|Zz@QscI>R?
z(eH1V8fl{+bi9(zi3dhz#|e`A{G@sE^|?xza&(dFz1eWO+bSon4MddT%v&QfS33ku
zfy#Ja`yHdIQb&n0B=6a4^cY;BknRjttqa{1IGRD4_igrW#B<Gy-m#q1iN5R8HUFgM
zN0B$Q&lL^W%v-}_*lLmW)0qf)u*Y6z437KxL?6xGFvU#?9Tp2g+OxWj!>t&VpeQp}
z0j@E@4x3l)wgX}#k?k?Ri)2AH98txhwz&(Ux(U9UeOkCZ+q@62fGFe_Tf^pSS0t9)
zq=P8jwGBzgwA4CTKa*2W`nx<RdcW*Ki_XKryAu^XEk-!l?^2!h-qC(d&*%O0ukIQf
zV#YmmZT!0@sx-Hq<|&M#gw)L#T?!!=K3{sp55PdEc@GDra+$R0-VV9-52sMhD{>EZ
zJfM2;rJhOfc_r=J^y0ZR7<;$B#DKQ9CMv}((qV$K0MJ2#RUE%Rl{B7ur080%A28Ov
z{PcuCy-TdrrGtiZh@ePfa4lWd6#<V1Rk^K<`g+ple$p#yLXKBO;=|44%gvD$fCJ7j
zQ%gHc1M8mj0{)6f4iapT8ZExHJ_ubpthoUc@VCKE$0?|S<@B@f@RG>a7OBXQ)7N64
zV-Q4?_B!yu1fwUSi35;Y&9=xB2gUXMn`)nRJE!v?6$T!Q7cH`7(ZTW}@Nz$!KSVz1
z8bGHkfdpy!8H7F0QYOLE+m_za&);8KBnKtwshQ8uZdapuh46(E#Nt)>51CHvJj@_k
za}X5sQO(b_dKBtl?`?Equt(dC^(X8z+3nt-sp<~^mc+fdN10tIG(8@cz$K75*W{<d
zk>m1dVjR~8OX~w1wB&UW$jcUzWlA;1x4Q|y9?e%_O$cx(N6ZZkB@P_QqV#YvH2a5Z
zN<d@zb3L^JL?@-MP54hR1Tm|m%!hFiU<F-sPngZUlOcIITP*EJ=gVy;{rV3VVD=3}
z3PACH5l=V~2~#!k|0E)YVhJKkl&dBSW-pDt1sz9X@t_e8AW+4&bvANw;{0M!o6M;M
zZwzNOiCUiyIkSzN`Ip1h>|wq8jp12L`i@#~n~)gUrZU}GTnsqu97@&N?%-jVx=keF
zDz)9gxLxrC^M46unR;L;lBce0PQsuI0~U}ZbAv5Y>uc*!+tRpPf}h`;`xeTG^j#43
zGIfhH&EcCxA(ka{elusoF-R|f4Nppn@2sX9gm#qqJ_*u4qpx2{hes4Z3oW@)tspiR
zoll%uCtfOBcj(SU?J`_so7#*MDS+&)-iw+(hP-TTMZ<PBtu`u2+C=^3dtCEK-^9=o
z@17n3k*c$9HW@rtQERl4=4$_OA5EPU9Qv5;j#Agqpp&@Qw`%VU%TMC#vpSpsP>?uX
z?e-lauh72PJirBf<YmjykG@?xINw{ftqWAEpH@t{awdM*9#zQQB^073%KPKgU8^c^
z3%I8k(vy$@(zJm2^+@pSxVpRRQgIA<0lB|_<V*YJ(kY@=oY((7mA=vI8^?Z@!BfxA
z?w*`_Z7*XbkJ+^J-uP(~tL1L|cEATNZ5P*ns(u}>1?t~j(BTw;+7O4k5xU18o(2u)
zE0c6SW98UK`bkSmd#l`vcsXnGq)D{SNp)${3&(ZrVJzUOwMWxh-98IA?4gV2jDAH3
zh12d4bnqwWT8<u(-x<Gg8p`gbTHln6Vk<bj$qyTI7FOEt5OJG_X{H+8^aq<NY8C2V
zZLee>pgPM&{dpKVm73Od29C*4wXnAd75g~6!O@sJ5z648pm!!<nZZ<fI=*uyp4?Hh
z+>lw6*})ezX6Ch?%ANT`<7=1H^PD+!>JqUss++FXkI5j$+Dx=I&tSx9z!zcl=7p7?
zJ{9EE*>ci7qiZ_z6BeYG+7N_@>L~|h_eL*KW(aDehhODZe(5-Mr0_rd_N<v0>h%rT
z&PSKd6tPtJ^WZmYWXQ|hxE+Z<u_J<g!aJQfZk#mbqPx_7o#AT|g@oD}S3sW_uH+)y
zsn}W8m<UR3xYRHo)^CRuAb7S}kW82$p$31>%gH!H=n|duJ3GOya<8el>GI_9<KIYb
zwR$Ho7j+Zm$GsW8g&taM;-+NM&9Jl}V-Ur+lhc^F_5!0KYQ%3Q>ylrZKjEakIY0Yg
z{?-$3VDg!KFs^)LS*kON&~wxW;d}2m<h4b$kH=_3v(rhT?gu0^$n3GT{=@p9=%wMp
zeWj=HeW#86W`hSqeQmQz$ia06yr_ll$5v5)A04)CDLOT2+b4KGk53auadAr1k`^r|
zVP7O}-ZcEp)Gf_QPg8mQ!UjD}>q6^JfyZXiFH4TOG8_{JKjYS&UG>?^n*HGvAT+-9
z30?!1`5I#Q{n?zx8fq-L>jGh=s+S(M!<Gv-duOMW5g@{#&?12>ymnShKfr&2g#~Lp
zr|RruAF7QIg~d&fee=#zpOVV#<qxR!D%E7sicdA>P<S#kF)k6tk!T5ECj$dI>DGKI
za@B=opP0MWkZ^I$&grR$*Zm?0&9Og6^qN}VJchbOdHx&i(uq~X8dWW*McyNa^U(fB
zX=>`DqkeXO!pBw&>~|@6N{8c|mC!=sag|Zy-6tsS@lc)h&B1=I#&dQ3t2f7l+`Tk{
zFuUqz&XHyF3R=WAGkRmJ^XEtY!F~I{ZmAh^r<u#g^UX81IBP0h`#%#9b%Wder@qg$
zU8|g$hjwtrV1Hz)xOY&)%2XkI9(nr%JbB>yF<=e2_!U3bXPYIF3g$oWyq;>!f`U*7
zl}a74XQf`lEJf@1ch3czq%&f1VnS9r^I{fcQRIrteY7Z_IGPx{$MA=h-Of?LH%qJR
z^5gamlG5U?AkQlAa#8)+d(KTT*o<*q7i^+jHZJa4B+1zfd#xN_!crVz_QtGAPDuK{
zh9czPrQTw9@vGd8VR$*ZX-x}bvBau<4Y0KFH`ffGPJb^IcdyOAEK@g5v$nLFa1I~p
ze!Xn^o)*Rew7R#NJa@lIR(Fq{2jWy`Y1SF-wis=126l8a-uAs)()ze2y>?u8tbS6n
zty%wFpF-J$&_dFL_||hgSHD|$aN#Nf8WD$(*5ja+O_M@fr5EcxoJpm%ZYf7iFC{Oe
zhihV_wR``y^9^R5vDz{H<7KluvBNEK{5Mt4jslv|zM8Yj0O+SmTJ8J_t+jt_f1l;z
z^O_#`QI<EFdG;HuUj6IMa#$D~;KwPU5G*ndoL-%gzGpeE^U=B<Gpt4D+X>Er*>&dF
zF$FnDM{zkFD!>zWcxfpYbHLuU)S-T3gEN%Ieq7gPXxe^7O9{O|`yH>E76!;bSeZr3
z?Wd#YCJu^vfZ0!Cr3u-GGQwPg%WjMrwWPJ}j$!C^8jWu3=7#$Atj0PjL-hxDaj>7%
z`)O!H%aE}S_LqD|Y@KiZxpwPcBbr~DuHU?N_Mp<CuT(|p@@?n1cCSAYMysYAXiQR4
z@h;i*C!0$uP%ZlUfaWmjQ(!Hz%@r8gWyB%agv_qzFQyekx%y2!BPvyB<4e7b7i{<}
z#&eE^X5y)zB;X5I31!JV=)eJ$tA6Jrrcc=w{Z%AS8+KjKBEVsn1~(HeZnd<x1oj<b
zek^}2pvd2wHNtG2R$*<h9y<L8UXl4RetY1nc`i4=d)P~Rm0xEN{7%kU&dr<H{8P>^
ziX#Bc>9Ju?=ix)IUS91Rai#Hnp{cO;oxwePe)2yq%(;D?T?-8enQAUf7Q&zl+=X|>
zIKZVCF>IxmP3b!R3D~8%m-?m^9;y!2y`rW_%4<K|%~9q4-lm%yXn3i1Y^gu^^xB-~
z^?S<wC+@$L)yT&PFy6PTb?p)FJ`bnb9JhIMwuRl1q3Pqgwv7e4VAAYa&S+==mPQ24
z2q6grUCGi2lzvlg^RTmsw<^|k!7Wmxh{vSqAQcI{kycumgr9W2|Fb7>THDBO!Luq3
zRQU}!T-$5yUS}VjGTp<2UFjSP44$~0bJLk&8cyRF2b6s$c_f9X4)5BBt54xO5);U)
zip|2d%a9_!J-1$uZ2INaZfm<mIyA*vaTC?XDG;QyGrOV<jq8$yh(i6I9)DMhLg4`O
zdb)&a(#%zG3sc*EuK2;rmna|WDE2OlbDLaxV}^M}U}oG#3)_r+CRrUOC{E~eO#kYZ
zx*Khwl$UaB#MTG`@i7-|PJPIpwM+XF=l%-KA-tk3IlF#;Z8$s+0SQ~S9}vPmSx+zQ
zXG5gyjOmkZ>x!1rH+ks>3sCT?SDEgj&|Nk=u3q#~1Z-%2e&W$Wy6a|vFN8L?-4&L0
zB{0A0_Ndmkv!|E}s0=JGxT5}h;-erE8vqyO*PcgcBLx{r{HgkCe@Mdx;gRO8udTs<
z>$p0~vm_>ra(12X$-zJSe7(`@LHam&Fx$g-pj4?E&3e?`<8kPkJ}30AUR8g7e11O<
zi;%JAo9Z_d)@HYR4tG+Fs*>_3AZPzE6$|m}Qo}X$hafJQMUv<woQFM~Ht=tDCJ$ux
z@XM*w=lmeORGFhd^bCil<+CZnvb+cJOVUCkO&;&A^wQGO8tka`?RTiPO7`StT34Q9
z;Lk;CetR?xr~WkMGA{q^3EzrpoUvl9PT#)y_&cB6yAgx3=8d=A{kuQitL;8*RCXU|
zcB7*2W{}EiIuw*|1r;EY^9zhT;pUawc@4DERo9LE9<k@0qxJR#$>0{u&wf*5?;vsz
z*NqXdOf>5?Pef(yhpm8rcK_h6M`y#vjc|xi-I$vD+@+V?3i~l3?|i;DXlg3jOxwO}
zeAgkS=OVx~hk_QLw2tOiopV~5s5MMDTlCvuFbqjv1ixTH3ld<7X)xf|(K#{t&YY`O
zPYKaTEZ%(py!ellg|JY;_K00GZghI+7V9?B=o-=2;#Jb?{H-9!)YC<A1`QIa%}%-8
zBtYW|bW9samgw}lx6j7?6+~Lw&L~*t=DngYo!9e$*`8t;anB}i&ZRj;E37z-7(wcg
z(u|~9h>4R%N(Yxo4R2sGB4R%iq_(IPtt@m<cNJWR^s3iMGQE9D6?x$i%Nj_t(a7q(
z^}%>&k=9(kF?sT2wO$hwh7#;O|H)bH_%2jky_^0NMXl~${XKGqZ|(Ug)NDlk7Wz?b
zH0xNxO|HWud)t}RQ}sO#V4csM&<+~6jE2iR=)_HH&B92(q|p?Nf-LUxY{L40{69=d
zy7SBS=V$^eLCf18+*T56Mnut{nO+5ZumL#zc;|$9tEq#I9vaCMiHixfFN15QoTQ=Z
zt4i|VJ3==B9`q{&#M=G7mS&Xc3$h6A%5IK&dXOddFm9p!0re1xUTpRnI%Fr3zl)+n
zSQD)%B=bdhQ7XbIX_(}H_i@#PMn-)*;}p}KdCcDLktF$n_|N4J<XTzFy;TR;EQkPL
zb*WH$?_&YL&};<`)TpNSE4*#9XKVVCgQNV?@!cs|FiGn=cn$W*(Uui+pTLR{Fx26k
z4Zv7uc}I-~hZ253L7^a>ND1i=+p#j@BkMyDy*+_x-oa=~6^x0?^;b#$luvxpbcKg;
zV&rP~oNYsvza8*%{xe%DGkOpEC7u$7)AP$0rBgd_algL>BFy$`<X3jSRJ^prbw-NK
zCqy;Zg{GHS^pr-tumFXoo7c}!6f5r&KxKoh!#H8-c|`NTaUW9=Fr-dG(r#9W2&|uK
zC`;6^$vU`N;(!9D%RCw(3(CO>8jW7XPZ{e011vtGVO73eEwSE3zB<(NMIAwLL9&!`
z^y?=*e!LMwr?x4fCdA~4i$~pi_nB-SOb)Z7PrMvQcJhs0%jmBskB*w$jcmmAaqm;6
zKYKT6GDh)PPmi89emQb?kIx^1C3~lC6|%bW8iq``O`?i8a+4Ixp9jCeHYJz%UOHau
zy6q$3xYFRH`Fjlm@}q-TZ34Ap2d8VsbsgCD$D=9kVfcr4oq(<PuKDl>GHhn>n`yq5
z=Yco{2PO02R3#gC%^kaYWVZJembz_bg*3)iDdq|$v&#TbcMcvAsjN~$dr%9A7{(km
z35X^>|MSSwRq4o;nWn<b`F+G4E}snaYl%1|><rP-f+hhRk_DrWZuiH;a*&b+KrJ0r
zW#{EYV15E>z2B2=Y{=Y<ICsDNIS&N1A?tt*3W6gk23&<pm{nTRfIVVGl#5(|{4I6n
ze!*oHh)9RZEvH0|mn%44GX4ypg}9d`A4_48#_MCMpJ?7~b^Ga6xE-omG9l{wg^T{8
zUJnjrVl3DwoYv$Y5g%5?8kQH$C6x|)ZL`l}bp3rMS006GjAU!7>XHrLF(LZ1d_c8K
z{Z47zP|%!|QFbN{*@9aUmYt0rg#n3K{}Bpas1Y!X&gf=#7aeF)z2sqGVR44hic~dw
z+?JYXXtOLXpBrGL-)2Ay+V8&<7-7>sfy^uj=Im;?1lSOh;_uLty;`w^f0pSd(RZEJ
zGTNkTZN)=C|82vLpEx0<jI{~GNMG`-W~_3mxb~Yb=(O%%*uh^1MMAKBiE{B|Y4HcQ
zG1)A6T>A_CsVLEHOKavkJdDpA$u%^C_|Sy*>Jw33nexJII2}ihhJ6gYeW|7NU7Aa<
zu6Eh+W<>D1s=bqUmxE86D!%2^BnFX=dlpDtGckw{Wb}~?(TP!uM$SAuX1aB|Wk$GC
zhf0HOD%q|5*vGx91JciakxDP(bW453&Qu=_=q2o*5@vCvcd>PRdY$$Um)OJ34kSb7
zdu|ba$F~_F7+No_>5(7pYDy1={Bk2&5MSN0lxnVG$D!78m-d<L1tV$RWN`i=3T5^|
z{@K(&Js-fQ>>t#AODm|oV=i)l8R4{!rNrynGJ=fc5L}=Z(G(F<#}qW3u;~^XB@oK!
zF18HB0X15*Y88bHYzGSFBAEzhn#hJU{{xH{CK>?6-z2HNq^1w9VbZP%m7>i!ZC)s-
zX!uKG$`kee`kVT56AafUG&BBSjksPiObl>?f}HdYQ1cstB6vZP47)5C(xF)xsZ8jy
zQmP~}JdN=H-*KKVUcQtmFldI)VVugrIqTn%u4DzTntIprJa;g3@h?Eh3daB7p<mJX
zFTK&&GO_|VJ)#;Q3}pr&3GWb1I9@{+UB<hsM)#)#VVCW6H<d4ug4rm8c&f#1HqG1;
zQKztUG$za76E&D)d;co?hH+|kVTzIock(0x8s(Z-pnHx#GPw7J8(te|u|Z;ISIS&w
z7wEfg7eVac>Vn12!^V9bf)}RRX0S;B4662k4DA6_;J^Xtt^wVXzH2$gyaN?JBhGk%
z%(<6|I3&282QMEx2KaxU>c*$p|GJswngB0N0^Xu4km2*}T^{+y-i-&^e<!4SCiRw2
zfa1@owZB!9zRQdoXU&g|JxGkhPQ<rYcW{{5Jhb6r$5*&a;Yzw5Y+A9?m`d8MohdXA
z7ccwtcF@<RHK$SYtSg;Yt}WRWm^(FZ&{ybWJ=$4I*k52$I{0vZROzvJI<||RFRlhc
z6w<L*ZDAj}O8><*>#*+Xs9^c#0|5a;DY7PZn62ZFe8QX9njg5w@V}j>OgxFE)fZCJ
zW6L6l0VG`woWDN*^$N0tOO3qGH58I`<)%($-7~JF01a~#q+C(kOA^HPWzS8N_HD_2
zzS_Y)&|;CIR#1Z@>k7VDH5r_wGtRaCkWQT~l<RH)=i&~o1sT4PsdvcGxqCt-m{)ae
z4k;yb`2IH3Jx#_aqGyKB)UyRs=a3p0HXFxt$Ur0{-BrHM1o#lm|Mu-vsOQ{=yti%9
zFVj91rvm)_2P01jYhf*Xyq~uy_`Sqd2%_oMcU8{RY_&7Ch)Tp6#&Zgp3g!DI90yj;
z5Iza{qy!duOa#*SUJ+Pzj_Rfzg2}WzG(gUi-;=D-XP|4MdJ}_@5#@Twoh88<={L5v
zb1!%ojxF7D%8_#ktaXLgAZq;bz7olxO1{CNy3n5Nn$4DukK%==)jGp*LG30(i$m4{
z9XW%!U58MP;WR@z34<~l|9vLY_i*O8(Vhtg<-R+5N$|L5b7XkGf5$>FNR{&3_K(==
zvP<{nUn5dZ(;P3fbTVzrAcIQ5Yfu@nuy{-#TPmjtj>$24<!|}Oxs(9_|EAMj4z`+t
z2a@!p@cp_(-|p}M<^}qp8ny(ZgQIeTL?~)|a!J5@D{uFzmI7~i?KxmjJT|dK8Z}dl
zy<iADFwg&^@Plqp{?<`z2%~=@jgOB-g<u8|_SV{mX~!Ckp8Bae%W^=!%a3Ae&o$mv
zpPA`9ntF9pF`GHXYWazq|6J?VU*%Ur<*VIZf3dc1eP-&%YxCyxv9fO6pxre09^>L4
zPxLixmi@~-XmQNT!?*T#KfIv6^PQR_f%?CmK03C;GUwT?-0$zFQr6|IWcnlqYi9S8
z*5w<>sLlf^Nvtd?3h4HOg5(BAfNY6S1hr)hWIkn0?5jo2eK+A}$aDg3Me}PeP!=dl
zB6YChvTY_XXDqO}qW#?EW6&m|KS<2^_+X=IOzmunA%rk8gM&jsVUgV|MXWKOO1{$K
zO(D`Q1bE2_b)`y0-do8@g?9NGYcK72oGwv%00w{0?+DaZa9vfas?T9x<Eo67@q<d@
zWUwSOm|V#&ajbAWX!6bA$jXl339!C04h#S|wjy3zm@jBm%81UAA8QC#xO75>kHzE)
z@q%NE65jlpwjBh16ex&q<^i$tz4m>)!=N$zK=V~p?;RQN=6W~1_?|Qi;^b_jol4Rv
zsQ}-eV`YVR1c8cw;=O%fdq^0{KUxfU_vm10{o^|W6ow2Q89&El_oNXoA5UsAb%csl
z$c*C~^$iqiEt~4D>pj=gZs~-t*6Y#U>-6}GRE5s;n%Y)+-@KgEq;qGA<+qSj7G>8U
zq22Ba9I@mrQU#ON*9<Z8_dm8_LJO@H%Fh>!b*^{qbTj`xL)X<)sL+;n$Omofi64#G
zyy;BN<VDCem}l6*o%%w&J@(ZCMO#tHV3<zaY|*hh)=WX&z2mHonjvFDx72U_KzJ`F
zM!7c?*=eUIcrbw#raU{foXPUPx#DrNMIf5<baJUtqYJP;Os{$)tEDlh-sAd127lp!
ztVe8ZJdc+DG0>ZedVC)>6f4Fa9bam+@Hx5ueai~8E!R{3KDpYc(P_%99DyOxU*NTV
z2hKf{^Ot|arS?mE*F8^)JddB^@PdnivPUl@T@;9z_xgK6u#sVT1v?uILmgnKX|AJh
zM@T^^o#j1&CS|_rP%~ef|M^KxV{~*BMs!Fe<&BPxo{}6^(A2UXhvpScF3J;gJ`ce{
z0R|u6*mlI>@SP`p>mN~Ws@ZA+EfXV7OBHsCWjzH&qo$f8bS|By(~+MBN+4tPbw910
zoK`FIKF^#o85O-u;lx(AJuOG44N=g0>q$QkKiz!W)3Cr#{r~<KvJBm<MxW`WFybEO
z^BSPrl;rvfO@&&b^8v3rMOUG*;61E&g&W)7Xe-#+4K!0;U191N=*59=e=}R`-pT*4
zkN=PM&NHs6b6fY=Vj&hRXb{DMQbe$!SP+#GtRQNzfd(vy1&IiXGz+4F5<51eXhiIa
zB8XrED+orcC`E-8EEEv|k#+uqoxRV#=gawWKi#|6Z~t}@g}LT@=lhQGjAuOKpX20Y
ztMF-K{XhRew`X=cNM$mXj*6F^sj7<gRSeK~_D+6Z@c~q~?ToclbWK4fw-@V;+P5=&
z@7}$|L%CY*Ubc7+)Utf(GFysuXM-{`dBK#wr`c)C>~|WpN3rGQ)9gQgAj|OY>yhy5
z@=O2Z0{myO{?E1fAM3Gedodl_D)TdhIz#YfFe!+!lt0IXbQyk$tv~uZ<%*Rgw>k^y
zZY<4x^trs;&Xj9JEV-MW&hESmi)Dv#9`U4%{+%Zh0QpOb>7v*bN20H4c)8o=MzoRj
z+Z^3MWt2&OT6TYLC<UDJ8(rDjP;hn_YpAGLw2`)Qy-cbk(c6aU>zRX;pPfBaYblIm
z!@4JpnPD&D=QD<%({}%LI6uE47}|a}V?>IC>(-;yBPs)6IlbnO<UH=o=3tI;jl<zn
z#J1GDWyeX`azIR?^}eW*_dPg45tpCKOj7H8q}#?A+swn`M4ysp*YO(1HLknzJKg4Q
zqrfT&YT(EmxyZPH`imSL^K(8hYG+!3e$$a|bFl=Y^c`UF067PmoaN+FIh|C4o<g^E
z|M(SBYT<4Knq+Rnj_mreekq&+o81zQJIHB48140cuY)ICB$1X4Xf>4Jkc&A?_$<-@
za^kzC()cFXrB-HlPY-%qZ*9Wn2Rt4gF+GVQzpxF;Ch`_;h~*D>{2mqzno`^67LnCX
zB{Z2{#0__^ztR`NbPvk=I?tZh-;=i7ldZmXr)^I`>H&08O@G3^sL8uII)>7T^d8L>
zDgCqf>t6k|kTTE_Ycwv9cTJqu@!3*%jJ^8(_itIQ{gy}%vP_lfHIP64`{#qnQngUd
zbRQ-Z&Ug2!+hgK3mp9er`)4wsmdZu9NkNU>%v-3ajNSF$w($(vOgLG0|HG~%R=o3Z
zt9apajpe?#hz=B1w*TSt?YpQ1?5f{VNAD9>J6{TBEvz?(Otn!7XteXtq3>k9N4?cm
zY(u1MSy6H4yN<ADVI;x7)tPpoVL$<Oz>9E=#5w<1Y1P!^4sll{9K8u&3^*><=RwUy
z-xe3)boSr8vsj{0-DDm4h`W}5{^t+=xAlmSSHrTdqi{js*sE{~!hX=_`2Z561GYbS
zKuP~5+cFujoKtE9Ls&=W*oK8fMoraZ!8Hre4Cd1cIv6b<aG#6{o)at!Lt0=@ny;n|
zFvXy}tG7?{XF{r#xi4@iT!8&309b$Fd__F2-|xini=-w%97$z}#C(b={#VyhPWm<F
zun%W%3Qbtxy`DEKLotmh^?&1^KpP<<emjb1@ZJ}7mu6b?8DL!?u6%?MOD=@c)TbTe
z4Odg9C3$IE6hi{ZFsP_oy|KkdXLvh0I>r{MTx%dn9UXxCxR`A*(qtDn;8`ERw|s$p
zqF<6ZjR0_k#)GlYD-2+%mVy!S@a<NAq4*#&rX)s-`Fdb5V1jAX=)hy?kXb}tL;k7E
zTi;s514iXe{+%5!19-WJGFSOknN4dKz#=qW5K1$sZ3pv~NQX9zHgYB@7XeO@azXJb
z267XOm>VPA8j7SmnbUMK6H+EDyo}+X!_>mc@a|`oYmve*AljPv#UNmMxr#{)<h|C%
z4ZEo9vLd?)nPEcNxLr{UAg~B@j<zf!m)0#^DYPKsWds<jzKB628InPbwOJX!IAK9^
z(AMUS@c^rER_h&|KsF?ys4YVVrXQ6RHc-`tt;<pC+Euz5?^R5jL4#?!5x1U+8@MNJ
zu98@zmQ#3vsx^7q<jiv@Q4nnZNjfXMQ=aCwe<Xo^E{*9Jw2$Pn`G0tW{R>>q^N;QV
zNn7fXIF0D~uH;JT<1W3{h_+VU98D+vd`1_Iy4y|~E%#Ll8MN}5bN|DrDoQVt(@4m{
z3AjLMNTd^{>?xN1H5O@oT1Yf&J8;1RX68ArHDwmn6Oe&08&kdq?ees>$sRtve7bUg
zn#)PpwGyc3VMfNb>c;}9U+{$`$iNQWTZ`dC=Zm~2X?ANxjI`(D<C}X*q8!~>Od#|x
zt<2!wQ5zm&x%BF{@A+jEk|<6FS5GCFjp{7;hN#XiP;#rDbd|`wKwhE^OaS3C8yOiz
z+29@Lpw8B4X0q=nv9Y9eR)f-VdQ#^5A~lfEp@m_r<U%I4a;8~)P=aJi0#Zoj%%=KB
z+WE=z?4RnE(dyb-cig>`eokL2bIP#E5s#*@xk|%%J*b&Zdk|hdq~X?noP@AL&Zk?p
zPE2)42ANjU4JO=qz!NI0Wx_@oZ7P#yH*FV{%~z**h3*++Deuu@SS_d>80!|lOVfGu
z`8#KYj5$7E{L1f@fPX)jW@CLyjYRTx=cHkJi7y4FAvv{*w_(UBTQ=(>u-k};pXii4
z#qMT%Qr=ie{R|{j1|$;o_gGM*_~}XPv_6Tm4wei9F455`qJJg{Ug7OfEv@fMVl!Cu
zUN(|V91($J#c?SF`9J57g}o9r^awQ}3M<km6IG&{I?lsy(IUQfcG4<$3+K%A$q>NW
zW}OG2P%MLfAgr*070vd>{%g~zEMm!;SMm@8qdUvQUeVVuuY+>fYvj}k9I-=8Qt|l4
zMt6R#`9*ZZU8+ItqOS;dM7q`~R!XZCQYNi61}Tp0oD>U3tjCmbcUKdb^vX#m?yx?=
z`^%i7=JjF;>5s=@;Fv<oS7R=W65*wFroijr=<dA|pH55E)u<6pO?=_h$)o@fT+2{f
z&H#s>{|)wez{bD;y{C_s29415$!B&QB;2sOQ6HRr#2x?WT@2H>g(-8toIJ)jxBxGY
zJzKP+*Z6heF96OwoGS`yM7k2u`0Y=}+1R)N?SN83_D#SlVX>^fj?J#W`T7uUCYHtd
z682Hg$G!~4fp(eY8-ul>X9+7JjyavS3;m1IwgcRVb%#C+k$~R$5XZFe`V}50aahDj
z0qOv&G(R85J_h<H@F?2JG6uNa2lTi*R6Hh5YNt7mbHE<;XNXOYOULyxLC9ay^jNxQ
z>LGc+F|r5dzmnomg!Wg_vQ`d&gD`5K!frl=5@ad_^(^uGdPPbLEK&1i^Zj6YZuDP$
z_AnP{wP}|05AS=z5nX&bp?AGM6<U|Y8@vM~iFR3+$%7SXc8lo?LLFuwjWIK4&g9rq
z01|ATnLSx-kiU4<9RKLne}So4-0!|z4rM*KaPKs2@ntN4bo4z~IET4RrQd=J4~+En
z_OSCoazt!i;)=iGaVTz5esD{^s~kXm1bf~IjMgGvcmeo&pC&ckJ-O_ro04GLs`F>>
zIEnk2%gG`+dp^U|(L?vsD<OmbJVX?Vsg+EuJt|jKeDmD%1F-^)YaS1d>D!;yta}&f
z%hj4mU>Y!>d(WQHpgl6~7js*S?w+_CaUN_KTP57|r7(xGKFZ?)%S63IFyIB8mRpC-
z={^!z78fVen)s>VnQoD~hDoJ20QA38duTb+TkAKVbIT!2&0EHKO?ubORQmVVO>{sd
z$hMN(>`)Lxwo^to!?x<Fj*hR{$s<|TO_d1H3cPqNpn=#^7)(ePzQ71_W-f5U;DcaO
zZ>X4ns(l+Xp|_s&|G;%+_u$&TNv9s}QM7dNvEtC#7PJE4<E7*v`17lkPc5~QDZPQ8
z2E}V0DIy&hQy&y7-(7s_*coj6U@K7^fi^x2LXP4X+7aMkw+r^q7$BNR6sw4SMnpva
z!|1H9L3QnSlLLtaYD$SIBoZJ40{*HEmnFfS%p7oGTmprH5ggb4hxMoMoe>Y}qhN`r
zIERdu{F92=?!9~M=N3u5kD09ySP{cFT$^w-lMycu0Yag`hdtbL2)kvwN79&x38<eu
zh1Dmub_7{74h%6WYeXvA6|fmgF$o(?deXeB$brJ{{sOco-bB?ncgF=hC(MOoSVaOu
z^MlKRYB}%MUK7VH-}I8XN|WT1!J{J9%}3-F$kjjzGY*?6X$;A@dsnX_kH}35Z^9;|
zH~j*khe*g$?XU%o=1dj6efxp>9~V$wglHmD{8$HzoD&!3iGio$KSKnj#Nt{8Lv<!R
zJ@Z0&{2#%Sz*5{f93CG6JN!JojA`J*?KG8hhsur`<K3g<l7_b4|4BmrKS-$Rlc9il
z#ZS-Vl6nYws2f7r{=GPxWhH1vPy`<DH&VschyKCU<nPBwQls!+LA!8SaZrc_xV*vd
zT0N*Ff{0m0Jn_5OBtHMoBExwRLxBA&lp;z!)#lg(8L?ec-}xS^)1>JFxU%_XH(|dD
z@p_7>o^Er1`s0zi`h97oJBXlpR6g!lW>l-X!79Hl(;ud$2fckk*eAjiH(L&op}r(f
zkYA-a38L@|ihWb4{-CrHPCo+Yb&6@a@(ZR}D$An4L9^#GeuxlIDqv7CCCbI*3KPiW
zw(wbiiz1CE^7)drbX4vde}(Ex9R-<UAZX~(FgvG^vr#mVsc*#aVUTY?-0Vz8n_NB~
zsTo7QoJ<8tSOfB9ymP+n32i^p|4k|?<uB<ZH(BT1w7p|jm8QF?&vnm8DK_8uqu!_H
z^var4{DvJi`n0#F8g5bbsveRf$HON*FQ;?WqE<t4k(kc}oa&#4h{vYqDCDy~;MTU0
z_trT0;I0PZH85bk%8NW`4wQKWm~ZM`j3Tg}7uQUNKdx&h%0uM8Z$dp8%OnI+YI>nV
z@3=82V^bY|!8C7${UGjX^U%*1GKQVluy6n&+MMsj_GJ%?ehum}nBblmY<FpnwW0B#
zz_w>1TPQ|rQ=$Ls(NQtUROCyq<bFSmS;kRptF}=zXnBQ<O!ARJ+wK^1S$60-o&94T
z*_M>|m0`|TXf-z<*Sk^sJA{)abBThQJG79dwPW7Hj!f|Mq_`S=a~3aJ)X2LkOrSMv
z>J;>tS5pLuE|FG3i(#-xK9Z7%2$<c)!7}xMK;WGe<|z48naT`iJT|iFg9<MQMcv{)
zq0GL6dR?bN5%8iUC2jZO9M+oZOgbndcloKS^9uKPTDOld&zsdvZ-<?EhfF&cm=qEd
zut)pG=L~>!LH;pfyJbNcu0x<YX+E!wWc;!(advI_-W2H!JGiX4i>-;2#>R+H;H@S-
z;C+rP*wnK{e;FYG<9zp#)!Q3Vc$N9^U~=QDR!T$wcZi7{9eiULqZ3KCo!_^9Lh|JD
z?hA0Uf<M%E-QQ-;5^E$63Dtr*DN_GM!i9iQnBIOt{AbHf{qle?#%ArxpS$~{Mr=UW
za=A(H%ZGF}U9aQ2g`5l2*8cJ6tBqRxOyq0ZNp?qa%NJ0A94}`v`J)73_Z~J|rq)n_
zAHQnyTS8AEb!)fk^u5stj=lD_IW;mUoRZFc|AX7N_h(Pyeu+Z6aIDXeHFIQHg5PLy
z9YNA8o#n)!eh=CDp#U%1PmkQZZn)WaY6H`^ulOaAmTu6coM=s$4erBYyhe4G@V<pJ
zk(Sy^HDUgU+E-*#>1vWnZ;yw;+QZoeGDwQRpHa_{hK1@c@DjvGCr{Co;L#lhwEM<W
zxY0wZs02lksnb-L<#n;;&?={6#f+{hGG=v?1;l$SrTXgu1dZn}JXlfw_O48?uX_XF
z3Ztxg&#YzALjNFx(*ed)jKn}Y*8~O=kYoGry=9S~w>L)?{YEk`N#y<$eX)dOzOSe+
zlX;7dgW}418S*2&>zQZgC%tYE*pt{<Hw5P2;HY=>is;@8SiO2RiKfB5X4>}L5<IEl
zVCK@<$#}-&B#yGYgO?7Yp_8cm532fEK~!F)_PzPX@VrymhF2>dMW~5Li;%CQ&Vd05
z_XQ%H4E7>N6zK*KDEg7g>4z}m^a}wN#@oAlP5c_ikY-{*Z_I+(B~4xy$<gayu|hYz
z?suW5kXuq6QH)_PAxo)yg9i|8hE!uSlQ`}8jrtKs#{mL_0z*jF&Xa~CIYHW(h@1%i
zvHh|*7+G_PXw|k##{kmsB8W+apQ5w$P7xJn-F^)+G5Y47ci7SWow@|HmD_CHWmpA<
z){GOyT0SANg9??@=1QBwckjN_!_o<*BzBWM91!6tK${KquXWCx__<XNtSsr11zZUJ
z#*^`PyI26sDDN58N~v52x6nxA)FvJ2WRON+`i}Tn0j7Zgp$S)^#U#R7;s6?bf3l<e
z^t*TOe1UQ}e?s_!I!uz2%C69agpWdVIuK}Uk8zf8<o`fOqFdBCxNer3l_UV;`fZq)
z$&z@}n928E<qeYG8nCt4Roeh-@NSX@aa{Eas&E26d25@JFHGmA<Zc)(&aXiN$VgA0
zxO{_18DK}=vjvEH<rpBV##@#Kk-z0g>5~dJxp@Qf#1xdlzN0nfrO52QAPC%2MzZ%m
zJd;^R?vp<Hqp)5I8)Kn`Gm*Bu1&Ak2yJYijuz44;GwxQHMK_E4#Qf9q(#N-MP~RL{
z^pRvQiIG)phw4L3gS&|rexNMJbTgo1zWc=fKRwQDd^|EW*2LB2k*dpnjhWM>c@naS
zoKJaJ{84rG8NX?CW!-3bkyD*%HLH727tiS)BPH3kO1Z}oOd!LGxPE^Lw##Vyw0<hP
zx~hG3+R~4SYz5(?s<fiT!aQHd@s1qHI|tBBe7j@(x<auVMTLbwT%CQ(ma>YH@9D7O
zAg5Y-CVwiXkL6*>b4puPxo=R9QP=RKTR7jQ*X3O@<7ktzZB{rYsw$q9J@G`u=noxo
z$eW>muld9K_hsq=iBebJrAsb@w+WKuvOzm}ad*)a6x9ZkDNJ~NPoU6tX!cmOt0AX3
z2Ti(uVVqiOwR-tQw_HCClv>lYAFW7d-XuL?)XDZc*=4V;cpYZw-$8fp$`3#JKIV0n
zryK&ryRCT%>q9=X&3|wm-Hc5GA9pAD?}XwXLNaTMF1`Jjo00-;D{F<|0M3=lR3ksN
z*L|;ND@IGJyJD{&(NFMz1<8b0KPziQcEM6>sG5WW)jO9nuB^k?@lq{{+<V`~uC13e
zPtaQ8WZ@AswL>@KK^84dpDlJ7tDZT+IIfe0vPQ?QDZaGRcV9@M#tVwVI=p3kjjG0o
zi(d?LL58A=a`10E4mSGT<V02jT}QHJu)j0rSJJJ(&mKwSXR+}GSvQ?SYH{iY3?Kr<
ztxb3HYv-H4@GprnWlC9HR3r{r+V9zxTg;}a*;GO4C+!<bA7|3#V0R&;iJg3QwtL;Y
zM>{G`o3gE0!_btEf05*KEGzJDX+A=evQe6qxeeVNA-cU$NAb9Uo%Nl9wn|4mDy4dF
z<B&dh^axtp3IxjkIbFvy9IOR=9V6RL*(DSO5Q5c>O+GeC2HF937d3;z3k<J<I&TKM
zh%*^nb>=+QQL+ip1MtD`4JMvxOu_ATPIqA^*t{)ezbX;wU5YEk(v;hqNb>}CLg=$F
zxF_f36G8rJF<zPx{Y)(YjWADmfG%11V*gDdkb;#e*FyXxnlEVTv8qbx>I0%Q_Tcok
z$0J=`4vuuMKX5b4|9S52(UxMASO!CA+w-9dz9Ct%{s*H21m-OW%u^TN+N6d*UWAJg
zh9aH;EFWOPV8;2kKnEt2V#>b>q$xD<wdOKt3yhfBNDhra#tYN@J0qyvO=Yab=&2b8
ziHBE%lzc9FsD*WOi4n9NmBLObhn>#ScP^}Pr8~|7t%x#EJZXVpOdY==z%;?Q40h`s
zmfq|A1}S#~vkM6Pgi*%3cV8qf0<VNJZw_^6o?j64Xo*RhWE@kJyukDLryD<gyRfn7
zCVV)5U!+$gYFfAuPg+gGbTi8!x-N<t%L6>1dT?5n$RVrsZDQC09i3|dncLb_Bh#^s
zO3IC`mjHr-4HISj&t|Rj1iIY#cY4Yu)c#J^Cu%lj-@pn67f!l}=N2uXwlnNINlcAN
z`J=d?!URXz@zpabs21DO@84eoUsNO^u}}vNh;)n~UlN?t5-5NmsfuNek5LYafFPuu
zvz~$dfMDTGYawC|<@?}^iJG#syg*799o@RkTYJ_7ttKAatmW6QhW%z9ilT`y=Nb)5
zBIb`AzFp|c?45~)Z+BR<oH%cgaYbHi<lVjzi(Fm)`b?45a)>a?7Zg$F*EShr{nJUW
zQQMC!@5St$<(W&1ckJ2o3${6|qxHtzv?2l(>v=J#-_#QNgnk-K8&sywTmGB*R0fGQ
zc81oW;j6QZ)=|CksIvv>{D#)aPmh<v7UXuN`KMya*b(^{;}YHhNYwg2&hbiN91b5e
z=|OXyGKX1sB*l&*@-Ltvv%-@?RLh*7Kuo>uQ=7WAE668C=Q^x&`s6!;FK^g7q&mS1
zN_75^!4JL?CIo4PRT=+9hvj0AhmD2zm(?gOshjC&>X0uW?%xWayGcK%UNCHJ()d)v
z!t+n?2c%kQAGFRLAHSANqz0z`ElvH}+?d~mbnS8GewfT;<%C)!Uo@jrN9{J5rm6Rd
z4F$i)PIS2Ow4+2>ft~V!>RW6rCHm-D@ceJ1+1I1x#AOZA`z&eMcWuM{$2O`TyRo#m
z?PoJ%=@AS&wBGYX>)89N^-g(JQ!j1WvEEivK9KMe%sBe61^}YVm50O<z0P|p;D;-K
zN<HeWX3cPu;lqd5xD`AysQB3*uIMDt%7|q|6h3s;T=Dp;t$$pPaOv@vHW2quqo!kO
zq+P!vvVs(9o7v&GsH=<FV;L!L6QKM`BUa}8B)sivg<pUK7b0|(cXEppy(U>(o1Y%N
z?Js<G=(B7u>UNY7M`XH-G$moQqEZ41tOIv2g>nr3QbP2wZ|B55Z_a&U(6{eL21?I_
zYe|elN@Pr~sb~diz4ck16ZAiT!>ym5^Lz63Akr9AHalVAgbJM^2%b8B#fSEZBi#dK
z)Jgn~vX4Q<*2g_QP#q-H$vO~EXGdz%2!GCM+0)GQL&M|$Zn0i#<V%8L8AsoqFMhm9
z*HxuJ3eJi$(07MTYQ)nT-bpO4_JiY{VlbQ!_L{^jb~)cdLX{fq#_gn`%gMN%DqFj@
ze$iFu)PyI)BXog*+MkgWng<yDaO1N3dl&*rS754Eu_z2-^fE69Wm7Tm8SzpWy!pJH
z6SrmtUF8AQedDylOS=J*Fv}1V-tn^P^2x6+;@o96YIW!tI?t+xd`m?2rTokXaM9Hq
zU)7r4Tjqg|tMvuT=pTP~k}du~p!w20Y~7Jq!=}SBh)^`Y@0>&QBPN5kL>GM3KXL|d
z`n*LjL~YhRL|fPg`ehFDrN)fDEm`4Inub^d86b?RO6OO73!I7}$kk^NHXL1%jGz_c
zF{mvYMfkXq(#UJ6Nt=T^^<VdApJlwl?{e6)q$nNM;O9ujFFHaz$+9m%LPvesHQb3P
zKlwD;8Rlg8I;8HL*{%(4H`^*F?H+yO%>e)u?O!L|{bOVAF)d9MuK#=mjh9-qJGawj
z?ML^)aeC|j<*s1BxQDk)X_0&_WripdafTxnA{Xq1Hc+A2>vPUO*Um!Fi#K=AUsPu#
zDznt=yIE(%w#+R_gh01*U1D@gUJ$ICt6pJ5RPL?AcB~5rJxpmbtXHjwU?i%Sh>nNZ
z)U{E$e-MdT`A3cHh8b)2UHk1CMP$$o`#Zf02xZoIyph^T@}$AJ()@;ah=tdr(u|o>
zjH}pP(%-+5F!ETbk(dd^NM4SQM_o1tuk=*wa$lgDqrZAR6;@bW#}(G}$dDST<|G^?
zZ3T;3MG`MM;_MH~^7MsEH3Eyd^|_2#P#ZqtNUFnh`v}k=$o98VYIGKaPs>rAI%P5L
zH1c-fc;}m;gjs9a7VcW&>eAzJW?YQM{+x#UfBDgPkY1BPkd1xGhdNv^RqXBI5xS+B
z`{_x#|61ifiWUVv>^fCb!&0Mfgju&sggln{MhIN}uGr;M&s|x!9(SLBe-VhNfkS)E
z?kw_kcRW?QV++~XnL5?H9I%9Q*JcEcRHP*gakkr_v#i?5G1qq#wIm;sp`Vk+xjf9Z
z5Dh*+os_7T&VG+uyV{gm?Goa<;*)3htkLE#{FPy+2zF4$R_j9aqaHpXP*wT-4(&q8
z4ejkZ#-knN1%i|Hk%*y$3d6xh)Zj2m50@|uiYQRB^Uoy>NqMR<iieT+k`u%M?d1Cl
z;#*RH*|aC@Re2W@IgH_Jz@x7po`2>y!`bRtzCwyo^uL)}yV@|Pck`yfrET<eFK;%g
zt(}TDXue|HVB<k*t}ZV|Pd5*C9QEhpOglYA)1C_dbr&vNaK3cA`eD|9=_LVC&1w%X
zRk7Yu^|Eo0vP{EwpX&zaZ9{we1(j|!7V(s8AI1d|7Q=>1IXJ}4k)o}{&14wagtM3%
z`o(180J)e}+~wYX%sW&n)pn5=N3o4a1FDPErXyPSEsNd|m9$vIf<9|8>3ZYC5s(HH
zqmv6>Pwrk5Rf5lM@B0h~CqpeLytKZ^__i`KM4RQu3d4er0pzr;=&AO3{a1EAOVLs1
zXu20=A*&(3ocN`0V*lEKn4rbc!%rgCUvuNLiQQ#q=?arFyi6`IGp4|jnQ>#Z)9D^R
zdT}Ih1Vf}b#$^tnJP8f*ed?~wU#vWWV1S5`N*tUb-PJ@{uYhnBrhk$oeuOL0G+n!J
zWm%KDcD}Y!_Exy(tA7j9>fo(9uERf8{Q3;@;OYNZ@t9{#J`U^fxa-0br|VbytVWlB
zf(~&Le~sH@<DS(b=geTA#V^K`beUZ)>~qxbBr=NJL$HfyT#CIEMP;l8#i=i(hfVvQ
z+iY9WPSldvF2^{8?1Umxdug-ezf5M0vM}nSYQM{?Z|%<7<bL0jHi`kCH}zMI4;$p3
zr8D%pmrnA{R>PN2Ys&8n&noLcP~+!x!g)9!w2q+ynY9%Vu63^VFl(eGqB^TS7`@}S
z+|;n#1<j=(D@EKxk;jf5ql>3|O?mAN(SUNk^s_;$mNge8GO!3ZfNHg5$zN-!#uZHH
zS=jfLP_s;eG(Vtn&?0`u&sF;i=Dk`lIN@=oh1C@EKU-yJ%)Hfb{~sIoj3ZSKNw#;L
zS!!lGTE{G-WU@qllR@B2T-smG`Quu*<j1j~p&~Fo@qln)Hihn$+8_;vGNp%HGd1Vp
zzyx)}*;3G?`Om1#9Oy`^=sEkYEZ)s9Cv2>m@5<%p#{{u|13izWe<iY#CX}&1jo9qu
z*w`Vj=E{*xT0=YRBwb>l_6$tOFBWU8tkURU^5W-o-UrG-A2?u>waq%uxuFSKmJiZk
zo2FK$vqii-ob5j4PIo1j9$9g@&9{BHYa_6Br8|8CRW3(^G-5aQR&+ZR*>rq^iG9`>
z$LZcU6W>gfRvu&k+iY6KiCR!Kc%0h3xmO!)fepa4*lVgEZ=MM`&0j>$aUo;~9C=As
z$WOfXh0rC~sAW7X1gD(RX$z#|1Q<sKcNZL>MEeQF^D%vPBM4*)G)nN|!&2djSCFB*
zyqo#pyScOey({W37UUBpJQ&_4_3E=&Pyw~Ru*J4%1!{jH1%`uQWK;s;IQH$eus&}J
z3><@3?|~T?>}Iy*h(T80>LVo*g3@cDe)aGY0&K}6Q=e-bt!lh9S98VqPhnfD7o&p^
zy3f`^`AlmZy_#*kYr%n<e*Ce`Jd)eUcX1rG7P&nNM|GIfz|K57(>!^jdSbvI$0LtU
zd!XF?V@=Bzx!F2(&AQJ&`?`T8?4-Pi$$7UkN>joq#I`A0qHulQ@F~O1M=vWoPM$%k
zA{uXm4@+scB25<Unz|`uL~^1Yytmm|0Q5C8(PL5F8?iP7lSpQcC#4XaO(;h9x3)`p
zouXSZ6G-i|mPpR_->Z6bT!rto-&U+x|HP_G40*P6{v^E<5N+P!g0PO8Ma*Ti-lzx0
z-sE=%$xYn*FO+<#(LJv^Tz$tsE1X23ji?x%s<v5<{0p7VR!LJVzfya+UE73E#;<BG
zzbhm(k9YuVooCZ228D!NJD{DK)3XZdPm5yzW+sJU&f7q&YHyRaW!ZPVH3Sqc^Nn;g
zdFi%nV%>6@`S{HRCn5t4C7)W94eFst15(fj6M5S9&~Nwj+|!Ky=~0_*1*^MVA^}O}
z8->|^^YH^zle`4dY~|ybT5$GeD}4G(66-ZgM|l?y))0B^v#pDm$GY_xY-yEtAK9X}
z-<B=qr%~fwRNa~brzDi^Rkfv81g9u#E2#VUCsmYcMkV~F{POe69``FKSjx;7MJ3KV
z?Sm3_jpYzQwwg~oXR9%INhjZxl^g}A>B{!2K48P;?>N!w4<9=nfFJ@8?N3IrsS-0l
z34eIu!f)L$(39Tkyd`@NdEV9cHU6~YtB<_i31|Bk&M)ttHHAIqT4g$^^;ll2s;Q{?
z7L1Q2@b575jcB0(w+&9TS)@D4RRZl{?H9M~JG;TeDdyu(@cmY~y{bs%L+f(43rV5#
z5b7J<GOFfZ05$I|`>{U^B$EUyN5|Jv1ggDFU_bn?BRhKx+i*Hl7_5Mo>(h{jLCA1s
z?Y<3X^!n^O@l={Vy*2%MUU}Q%RFlPEg|#pDU&Wv{-!@~H*C}ZlUizz?@5Xb^H6BDl
z?0tK^qRIV0(U_fs&{UvFzc|H!8~~Rn{{=E8c3?y6n)2d$<vEM-R24Y<1s|Ryav|o)
z9;S9d2C_Bb<07aV-n*pq;vP%cXBoAf=%?AjuO@D)?oWAVRP<FM3tpbw%^M3#N&X(J
zTqhC}E|DI0m;#{o>a!3EZ*34Z1{5_#cv-F@b8v2VCj8l13|Og1D2_cbW^9?TGB7RL
ztMW+eyUfhXfd~f9@$*+b*{SKrHvTdD+>^{P9Ww_R4?1o<Xa)tT&fgtMa#NImcfl?K
zH%FcMEOw71q9A0>;P+i@UFue<<VBPTV&`91<i9T5mq`O-svq;XJ(dmFQ<=M_XEM`N
zM!yl7Ws}qIkAs~n+}FM@-_fJCw{!oJ3A@`12d7SkgQg{X5b>0WJ`S^p4B1Y^6+@I|
z1mUt4jpxKQwm4(*4Ntbiq>h?bwEsp`4~oJTS{Fe}aF-JaN=OcIq>>y`Dx|pE@#?4&
zgaKQfdT!8LihqjR2J;9$*?$@+H^M3vz6~KFZi9m4JZd`<Q-T6)WaQE<<x%Tf#M|WB
zov*eWgHgh(b-0>q*@F>cjP-d&!Uhv%1tF3Mud<daR;=K%RZN%v!u)M=&;Irk9RZ-S
zwh|5nzaBA<{SL%mLnrH5;xP6v#vqpOr~DM6_yypus{fy4tvzQ-O=exn&Goa1u!Wjf
zQ;yZIPxeL>ONspZnpIRRZq?cZ$32=H@Ho>4^J06_IBt1;*nUcegeYdOp}S4$(g5lQ
zK*tDMfVR20t68_Lj(t?Sw*bVvBw+BPhTYyp#dmBr<Zr}0b}I$b+aCs^O<{YO;Kw97
zaLx|GN}Dt-eL75G?a_!k+9MtWz@*~(Wf~F4T^u|PV<zoklcm`376hYuRRp4-I;Yp(
zJ$podK(Q<`-G)#a<?A*@ohMjOO_D$smFOWc<~Tw2zWEs$0==o#-C_CphqI;R7}gU-
z5ap)J&iV8_v{^XP(&s5C*-Q{D+rv^?&hUx{h{Ky!fXwo}YBo@4og=G%L{cZCTXSzx
zt3T^z5kIvUq3@&7oeEmW^J*~5O`kLk3)dW~`uI4Cm~?SgBb?vB#3G{Es5LMdd&o&O
z?V0$BSR*hzpU(D<-xK?LxgvG2a;@9XnwANc^LWJQrUn(u64*`m?%v%z^e&Z<z`BXU
zyP;0F;$Vp{#r*tz&dG@Hw}0o;+t4R_`DK9x=JjslSM3iSQOQ*1CWn;lJxJ5KK0DY8
zu9;_#>tpu9HD<om*k93bzkQ6x%!W+{*|hv_W_dEQf9}b~oD@UvYbe!qbc&&&<ksPZ
zuJ0X^!<09({tZ(%agixY;Oe;08o)Ru)H`98VXd(O**wovjC5vV78SWSR2{aCOu{4s
zqh<IffnZbDo3wurIFKI?Mo21b@>O^p2o;uqFCzL;yDx6Z^8vfgZ)HpmNziLxlgKj@
z=Fp`(f3JgZEl@<Q056!ud~ZssIBVYT74{D8F`}?08X+Y)IOVDX8XYZx!N(lFTY_^F
zB?qty?5fs^-+sd~R<L2|MJoNfslVMEw`sS$7!CeQ<eIPZ7cP{?VdMFeDepCyQI2Pz
zrh^!)AsHQZ?z{5kmE5q{VzX9xLq-L==vv;>)&8ZWa_WfKUXxyC#^|0}aN1|b1x*`|
zmi8A05B|3P+>?VZT>Jde{=njqkqZ(R=<eDvZ*j)yIFDsnvz-iQg;)%Dl$g@=!|hJ1
zPVepGeeIjK>ADLmN~)J<<X!kyxFY#jY-};-bf1~O9}-Ch>Q@vy-5ZEP<nKCzJUu<T
z((l{8^<)Q!F8(tDK6?|x2IbZr%*zi)#!TDM#13S4Yl^L2qw?}{^sF^13^Jplq7oAm
zx4&MKjBXYwMe6g&goM?1+-ygWzD}4qFe*A^&X+G=P>HcsK4r@~kUn+s%9%WC)*3(%
zLCmq8L#dar_9tRvw*&_hmcFSo<ruQGBH<Dd-$k{hb?dE)NWDMJcP@@lGxazbd+O97
zEYc=NyV<jAS^m^qoz!n#eh*nb#Ye?TfA_I0=|6aFvw3}ALNCnK=%DwXJ{cMqgeY5O
zXJ@Z>vo|Zzb}&$1Lz(DJ(%sVh^Cmy2AG{jD!%#2pRW~E?PP=46Th%?29UVtjU07B-
zHZeHG_1yH6k&(a74L4M~S(Id$+V54V<<!NC7r#lFXK~Ymzr1v5a7O4)9;cL~p&<A=
zeu+fBWW_S1;Lz7sR=zt7)q`0K2KjhbHJh#}!{sUa;%|deUhW>YRYb444)ai~I|@N$
ze$|_&>kLppr=n1nyL^>;ErCxjFO8jTY6f{5s1HZ<9es)ddwtpwGBj%fY&CxGIxMW~
z{)q1W`9_wFwr))dat!}L<4^OR+wVL(U=^C`M04C`v+JULbkl8ureTiX(}#nGg&6&j
z)u7W5qz@|Cj7=s+qZuNJ1tnScf>^YJK2nQUB#cE$YBs4y?a=fFXo>vB!R@hT+aCB6
zAK*?zMC9%0dNelH75nCHmq2DFbtRV(o|jp*?bh>n8WHCusIGNJh{cp{2LRX!%l<0f
zu2C0qJ1s3O%hY1pv}t~Rb=Grs@7q@m7k2RA!D;>{e@Q`d<H?gJ#l^*HzTHx+xebx+
z#EtdP+RXpM+7Lw%DafXEcAHSzW~@=vZ<{M@LUFgo+Yb?NxVpL$WSSv#r246rYPe-|
zmTH}+3~f4n`V(ilbCSWA<yWs>#T|D_>IKk?O3TC}ANrtXX64}EFl*K<SgEz95R+JF
z8MSD#LeXf9+bHjtt{Tee_S0KWJiXScBGX!XK&7{fcH0EXX8R8u5QSkG<dLzAxLZne
z!ti;v{LAoe%{N!NsdQ511W8REZq}2hPvhg`y*xb^FdeIX`(1nYnt}RoOcL{kw^vuU
zip~P>6YKFkwCX+_SObHTF)=YGPYNHS-AQfwZ?N;g085uF39*XC6_<ScIPp_>1-8;p
z=}-P9^E$J#v%f$fc0@gf@B{X!Zr{Fw-L-4}*eVs1B>BkG?CgtiaULEX6_u5+m1VRd
zHLyA4A#LZ+uALTa(F3_z=B;MAco1-qN`Eq&%=zWcJ^lRr#*V#<-Vq$?bN4^zr{of_
z3&^@{+qSiRo3_$)adS&y+=<ouHnJW}IotGub5odQc-_=fr%t7(rvnZlQ{`fBZ|~;j
z#w8)#y1SQgcKhh-Tb+BcUG{mOKHspq0_O_!1v85m^y<|s&e5Ip^h4MSAYaB7c#*fZ
zp0RZGrt9%{ev}i|tIvp^xOYlt0-g`j5Se|%K3B1G2y(tuRB-3Lq2dr-1qKEJt9-?y
z=<DmZ)#?{&(E~%X#MQM1W_kS5mrtm%;$TuNthhn;E8mFdFd#yh^>0s~JzGVa$Mso4
zITqUZEwj=pdoZvkPb%>kqLNHU2gv4d=qb#v7U_4I`$e;Y)S0(xT{@@dciCHt`4K_A
z6?R`UgBOXt3;B}iM_Gk>r4JiWW*p+5_^|6woj9?WssW_7L-<+-L!O6o*0@CH&ys+7
z8%$fZzIJCra?Jhtd#X-73F&ftMfMM7wx>m5rK9I$y&`Zo&EGnR{H^EYH`zS%#)DEd
zot@Dz?M9TL{f@`gi+y^HJ#^xaplwa`UYeVicztIk^qXL&%Bbv!!&P=4-&o4Fe0hr#
z;g6-c@_H_;Fi8l0?J)2(XN|LaeHPLq{YJQ`bn4)(nEv~I)2S(qb_jjQ`Vj;(Ht+CN
z<85pnirn}wrg|}F&bWa}tdD1!UuR28%U`nxo)%>5&Yu1I_uJ|>uv^<zBXQTq$eJ#{
zYYHc`99GcCw|eua$(gMvoG|^%k^Pr0UAn$-4gGp7*pKJJ%T7c`JCj$gkC%z7;;c_}
z?bbmNnLjSQYiw?VvC?X#ZDeZtU1aZ)-f>To=*WdQ1#8d1Z+QRyz0@_@y#FZ$a{`h)
zq!e17<8Ytrvjeh@25szPye7u4sdnF<A82Y$iwEL1QV$Mvu-u}d*+6DBVxHD)N)1`u
zvi|d{eS;&Krh_bwKR3r7ERr_cv=e-#;fJ;=b7wy?Z(w!nUqj45b;2qs7u~9JV{`tx
z*L=)u<A8J5=DaDpGp$j;xv~FKRlxc`e;^FzwF7P#ssv<7p%C7W(Fc8%Za=~IFe#~I
zZ-AsY7CNcq?I?c#UiC`jxs9oZ+3)&P|2ccY^L(ogTp>}o09@kn%a@`*fA$3YfR+E6
z4)(t*W;k>RPQ7uuL4aD76qpF4<O(WtcR3IZ3XAkgi@XuiPQ<58{fue(m;W1YPvPHq
X!p)Rn8&4mXYmc>@WN~`T+;#s2i8N+R

diff --git a/vignettes/getting_started.Rmd b/vignettes/getting_started.Rmd
index a02ad56e8..bdbbf6710 100644
--- a/vignettes/getting_started.Rmd
+++ b/vignettes/getting_started.Rmd
@@ -196,15 +196,15 @@ model |> evaluate(x_test, y_test)
 ```
 
 ```
-## 313/313 - 1s - 3ms/step - accuracy: 0.9815 - loss: 0.0803
+## 313/313 - 1s - 2ms/step - accuracy: 0.9810 - loss: 0.0842
 ```
 
 ```
 ## $accuracy
-## [1] 0.9815
+## [1] 0.981
 ##
 ## $loss
-## [1] 0.08029716
+## [1] 0.08419506
 ```
 
 Generate predictions on new data:
@@ -226,7 +226,7 @@ max.col(probs) - 1L
 ##   [1] 7 2 1 0 4 1 4 9 5 9 0 6 9 0 1 5 9 7 3 4 9 6 6 5 4 0 7 4 0 1 3 1 3 4 7
 ##  [36] 2 7 1 2 1 1 7 4 2 3 5 1 2 4 4 6 3 5 5 6 0 4 1 9 5 7 8 9 3 7 4 6 4 3 0
 ##  [71] 7 0 2 9 1 7 3 2 9 7 7 6 2 7 8 4 7 3 6 1 3 6 9 3 1 4 1 7 6 9
-##  [ reached getOption("max.print") -- omitted 9900 entries ]
+##  [ reached 'max' / getOption("max.print") -- omitted 9900 entries ]
 ```
 
 Keras provides a vocabulary for building deep learning models that is simple, elegant, and intuitive. Building a question answering system, an image classification model, a neural Turing machine, or any other model is just as straightforward.
diff --git a/vignettes/getting_started/unnamed-chunk-12-1.png b/vignettes/getting_started/unnamed-chunk-12-1.png
index ad4a14d9fd6bf348607fbff4fc169aae5c1473a5..9a387be7004aac15400ca8eb4bdcee9d7ed524f1 100644
GIT binary patch
literal 32138
zcmdqIWmFtd+a=maa19z<LU2ifyA#~qA-D#22<{}o0*w>gA-G#`cWE@ZyEnJUJKxN#
zyXIT#{+M64p{o}~*QrxS9@%?85lRYD=qN-eAP@*$`jfZ{2n2Kc@<xIM&d|$P(g7dH
zj-RxhK_JxEFK?Jhh~Fa!L;;c(7g6&_JA}CVsH#6Bo}M&PbMZlhvvKK^bgr3Hg*#wS
z`X#YMun{8<I{Px=u$M6dHzcu?LQ%00>4@J@^|K@hF#X2mt#NiZ7HAg=%&U7f`QzL6
z^@E?l*u`M{a$1^?v=9&QN>-AAp$z`!^*;L;B#M;1)|DlNM48zAza2M4Az#mVNMbjJ
zn1)`9wiq}Vkp~(R%2;)@M{Rqq99$u^XeXJOX=B)e(Yo2tf<k(`iW@4?x_4cgI^3v9
z)F-ms8})jVU3p3SzR*-4Bb8||M&{(ueYMl5nByaf4QTH{M-XFU=wOp$OHF)BzlJxy
z=Q7h$8A##Lx?+SoyXS)NXh%;Nf(I{x6^z0$RLz7Sur`e@ic~~~AOIXm{+pv7p1wf@
zmMD#~Oug-bN&V8)<Rlt`SYb1cIomF_sPWo@MW#HzEBlHi$;Z2t#sf}CNy*SAsde}u
zvh00ET3U<lQ1r92Gx=2ZA{Xk1o0HXsFQzIgD(!wxg!yR;Yc&>Qqp2Jvm6eW{yA!S6
zH!Qv%^CG9tR-U2BHLlL$GE-rFE93;&B5#rT$*3a4e+a*(0U!BQ<k}>&gDNdU)JrsC
zh&YNTNy|fl-W2D5`0&A?(P6qUf8b)wf)>5my(<`bi!3QV-n3@n$B!RKlx7Z)Hs6QG
zfieDtwyO-Gx|Vi}n;J!a1##?U`!2S$5t+B&J@^rmSW@2LSQWmhvY0{Px%eeL9`N>c
zTv#~Sx3?QtupNvD`1b91AfxUIGja*IagOY|lWMU9b|U$#NG17i#hx4AmzS3}hT531
zY&ms}$f-W*Qiw}Q5#Zw^Y$o5++&sTb-Oz;(8wznWGgeMcPGn?cv7)j2$%<XA4xjrm
z6H?$wYJkJn(=Jq2xtwE3Wo*#+b2B-?$s;N#d<i`#ES&1Q_^F2mv!O<%w`R1$&TnzF
zrV?^mX^Mzs2iN8Zy2T5p82F#(D{u@uKc?`~zX|h@x$q92C2R^~6-wSZk7c0l?a`v3
z*4d0)?2NIp4AoXKMaA<6h{ORuhObZ`mYv5WNLR>9<|=-qgrE`-#Yo<|k4NR)0{zjO
zua2;rblIOSx;mI+T|&Z^IBkgNgCpajj%Y*e*@T(cB$l^0T+%eemnTBc`*xC;6D_%M
zrq(o;{2pxvNewGJE015%Gp2OH_Wo*)EW*_z>FMsa=#TG7@lqxBge7i_Fb7f2gp$F<
z#iFZJd)l+lj@#0hnZL+goZx-oaviIF(_g1X;O?r&@XC#ByS|S52-(FnLPD#Jrc1Q$
zs~Dg!VX>tKtzNt1**%P3-+oY_<tAPG;-KAlPT4zS|H&Wx*8c9Wds?^YYo8duy<q~*
z>efNbx=tmi6aj?v)iNXO2mM%e7Iq6svMW?izvd0-gPoiA$nCG`{HzJ37-3Iz%UZSE
zs;EigkLGMC7SDBy<Y1l>{XT^@1evxV>jO>_L7SwdqvNG|hq$j|gZ7!T?)~3j<1Bho
zInaqYeM~2RmAE!2r1J(+XqC-I;^D4`lSy52=diLI4#d-1jHGTka!baMK}s|$Wa>R3
zU9an-!k^KqBn*YUSrC8kez&cQE^DtZdxHq#RGzkUg;gqiL`_`ee@*^o06a`#Swx|y
zU@ekGB(l*UTA)gRT@)1#9}{Zrszx0kDW*N^zC)YgLB(-kiyq{6mQ6jUb+F;cEyL`I
z_J?t{-rFoB5a^2_ca9ViNH}Th8H4!Tb9H6n>^r8SZ=>7Mq8UMsk_-WhUW4D${olWT
zrQ_eFjF_)piT~)>`pi$}$QLnmRIPpF2F@@ej0<aktbM_$N98AzywoFeH^jkQll26#
zS+Qp&csQ668VVD5^#mT4!q?FEt@96BH(l*H^Ypwb%fS{d8KJKp^}HV7YIcXA&>b~P
zFHC2*A8v3WFZWYRn6YkM<m?43#>#IiRb%7zz&J0kTz<*b8-hB`FNBwcO>|65%k>V1
zfO!!W6<yG+sjf~Hmzw1OgN(FfHuam`^4R5IlcUZWI*+LAxxO@F9Ny+0Wy9fsxdxq=
zE>`^uG}3u{w&3fd9v_@TGnX=>s;R!Dv3&S}qJsr5G2SHvXNMIWC7}VPo0V61rz2qb
zsu&x-MuMv6CM6VczRsj~vDSu|%l`4S8#BtRPA@LHtcv!*q;1I#J2*2olj!~Wnam6t
z5~YCmb?LHi)YjJ4NXo}P>skzA&QjNdS+04aV)(SC&LZMOImqYFpXZORXUlY*^?16I
zP5WY7*_>{({)X-)pL@l=@s^4^?m40&tIwj#sSW?qjF;^E{O8KfX4dGdmc&Hi)CW8t
zJ0*i$OZJ`WY8+-skP|yAIT1(E+}l^bbr>v*B(rp2;E%v3t<PAr4=c)%b4|CoIVJWO
zWc=LD+j8|w0!AGH*wm7GN0o&(vAc|Ae;M88D~*W}F+zW8-JU*?)|kJep;0fFPirb-
zCz+t;@VM(GU`D)ot20dO`ey)#9H+&=k(V$zzCo{N>#e<JcUqcZ*s8N9_FBlt)2oz?
z2YlA41ErGulR)ip+_4DNSh%v-eHXP7z7*fDdc#(@c;63Pe0Bd4-pq>^PpOv0TGy(T
zz$LZ1>`j8Z5&T!`xj-Zh37Yc^RBf_1D*CRL!TeHw`ZdxCW<>@;B?7sZmzRJ^#USN9
z<u`0hWp`Ndfj-_PHak>J6)29S@l2W5yf_TSOu;QO>LDjLH#a%CXpItP?}Oeb0xs*T
z@f<u*2L#Ul`RPUmQJ|O|%?sDW)6?@}d!%ORh*~OEPFkA&<|kk`D-2qj{K``052oKq
zg!Guxxbc}XQep5_ErfRqGKx|_SSoG;INB5JI=l3=caMIU*@?r?>=$O4p-HA};+1Jk
zSQ_Y6R<6~xN#M(Nav6^KU4HkXFUp#kc%6x9^49A9a<8E6d1AcyD1lzJrPe0dD`9?q
zep52a?Z|DdAviZ|_UYuQTEO#mJ&aZ%ZC-G5K3mpv)6hG2dFNpxigk2=F`#gl?4-5c
zs55YFBtTVWEK`UTNwg6apFVGl|6+KyFC)0H-*Cm#M%Bk>DqmifG#(6o=%<dejSUaH
z*lACBJS?kU(b*5^@Bh^B4c!2ln+${8n(g$MLV&;po$!(_0axPX=7y1Rv9o(n@8{k6
z<R>LAmZW*x9(!OcY#6CHOO76So{Hjc2+@7+uQU~x#3xK+!wdDp_Ak{c3V2ujT~OPe
z>|-Sj$7?h@d(-kyS|64i+V5hy2+>%=;6NWfeiVX5|5XPM+8s9=n%Imunv&rdy?Fd>
zHD$E;yF{R=Z*zgt)A$dQ(a+&mh9umMSyFLid9e|3apdfl<6}b4%MG$O`J{aKgC9WL
zxV;Z!B-Vb)6C};3Wd5%~oKICp@iE|~n_Uc0i<BC!M3Pyus6<v@EGeJMuF^;<2N2{C
zhE@$l-gr$_D#-M{zED;uaPE!7EAW6hoShKxPST04FeV(ZhIq*;7OJCy)>a+_xqa)L
z%O+3k+RLDJ7q?;A3i?VdQ%U{~4s$E+zQKR<o=fL^?h-Zg{ofOOzH8eaq2m2@<lYD)
zOd+nh0qN~w>pki9>azUl<oDI}{I7Ia_xFSZwc|sQCWN%yKA!uv(TNkltqdhG4-5>{
zL)@0Os6VHCnU5G5oVN=}@wX_|TgkO2g`IvU!z(g0Ve=Ffj<S_wTDTw=^R9RI!H}AR
za22|1>g8CWjWzqRt1AR`C<F68f;QvU>((+n90c<4JI~AXztmT(U8D-@E>mizb4xvC
z<9}kZgx7*Xbm0)Et;ai1*Yc)W7(qswxU6{}u$pr~0jkR4GB8=6lIUNB1juXfla0kr
zS!5}Yf<R%^sXIRHuGI{YCnv*sbQuhH1e#_Nk2T{{2QKpvd#(EXo$Me~e5Ub(E9f%m
z&aN-p?)%*!yvv>gNUQxw`Ir!C!t(BqA+zX;^w);*T4g=P?V}E!T1iCFkprbaho?V7
zKMzdWl`W+^Oz=)I4=M^(nVzgrX`d<WtKgQDh6|HMIz65Yn+ZX8KhL{aSw$2T<>YYF
zyE0Cf(P2oO!KO93oRv<YD^E~(!iVgmAs9L$UN<7bOCcScdEla;t$$uHVJFf?0Qu;S
zD9Ky}>kSVM)`$l<7<ohWhC<O}(7F@n!va*iPS-jEQ)>fu)=z4Vh&d(?uHJ>RId1(7
zLd47{SH|(dIY7j?JXy^9%=H=6oHK7*2RJW=$k)Wgx<g#QpX*`4N5UHT;^Kir+N!xl
z%DJH@Uf;2H+S}Kzl=d~sahJ>Jjt6J_M=Ho?%Xxfm9sOf_vu-*Sg$vK=@}lsW2gk<z
zPdgFinY#6cjh$JmB|_v@Xx<6M&6h&xk5;PE)3@Z9EG!nfIm7xLLngkwHa_$Uk}YuG
zM=yvhm4ZU~;*4u2s?d4e5_nc*2$2g_vlS%Hl)N^eVd?Ks+2EN#vTZgAUvgU7D#xG6
zK-`$1<nzHseG5=J3`m&T1H`}8isKwi_yz$qj3$OeNEk{ROk-rC1!7}E%ca|UxXBzC
zDl3_Bvs&Whq=E&$LkVJ{d>0A>n%G)w_5}}Z?a#M~If;@Q4&f(VZIvW_Gm}joecd{q
zN31v8_|%+7?suDA@ddiN)C>4RTs*v5yQR9@q4^&r_U5H}*KTs?mUA#9*DI*oEc#8q
zr=14|2dx`T-ta$=fP@uBD+6T#pULbxY)DXZoW}VI0V5RdfwphXqx^XM90hf}<&jk(
zGM&?!%VA}?(zv@koe$SgJ3D*8A9|4)@^t)lo)8qES#W4rt#!X6uy#J3)ct;Kq~2)R
z0|d&zyuFpXauhd|m%4oO<_$2P*B<=TlF<zxic|voHa*uv2-^6)9*-M-Y8v|1dK@*%
zYmiji*6WnbcHN<FJ74XmxO!1{hkN<LR8%?XVHePEcoO&N@~jMe&|sk>Mz<!+p^g4-
z#x4xM0RHZYv9SpQV$t945w_m^gAj_1WQjcSF*+2Bdg#_E?bRu|pYH;ZDZ30yL6Y%r
zKieS=@9V?8QiP_PU|E6(mY>az>s?Nm&DT;i%D9G}TZxT(VGn)1a_PLa%Mwpp3~w&o
znxD@4$e;7Zy-;Nh{O-0i%5)NWwSN4%^>gn8Kf{Rx#0lM=wJE3bx-u+npKbQJA1`ay
z+e7RjO?M0ELh8GCd={f=1f=%6&U}9jnoL(L41KTE7fhii&rpYQ2(t3TY2S7Hs=w}|
zO|4Bgb}J@=uNH!@pI=+wCjCwBWvwSoQDG%rMd4Z=k8^NF{fCxN5`iB{lfgvt7+oHP
zCa<VqxsjbajaiMf>J=)K7?g{VbZEQh39ltHLSiXljfc^?Klb#q5yAFQAM<~K-RA;<
z8ooiI%u);TQwh6p{&ZlGb|axxu?=SkJ@Z!sQQ=4`^>Je#ljSg)?s+qB#hh-5gV4`&
z2Ev%rSz>TOOG0YuzC|)af>iuFc^zJtx8jm8k2X2&2^1p{iTYBk(vuwchW1N!M?si^
zT85v!ua5oV1g__2Fq)AMb}M9m{7N?P+KUo<<C{zUGS~>{gNCj^8bS#knJrc?ahm&*
zgc`Jh7$o=@2yMMt@%A0!WFrhEx+3>Eg&Be}`UoVLM^3Jiohcb)$W#AqN=w!|){`Ej
z{1vc%c!*Bp`aV`AN^`Oa!qlfikrDXl5!dV;#{N;vq*g~D0H&k=gKYb)iFAy|C+Szs
zPj#5)ECwwNYJACv8L9<AE1;8>#=AU%@}!A5Xs4Q_<e%n~wgVU56QCQ-$4x`ev?|&B
z_v>NgWI8ou$E4_p*AV~x+vqnR2L%rEU-=5Fg=F2}K#htg9b?<9u_ex~<_v|rE-mHc
zhJnyzV`D>x^jl`;^Zo46>Anr<6f!4!Z@-8;Rdj(}zw@)td6_yLtJgWlWy>;*7!**M
zb*I^8Q<xJkE|90AJv)EIan(eZG@~j((IYA>R+KUMF${AHr)#=2@9&lrIS;#SKNj<l
z&TpMo4U6%;-{z=W4T8S+hb8GCbLZj3RgJ^NEnQ<3r<swBT&XN|J`gDPi@Y*VXWzwR
zsJq2*8jmV&?>cqpgV1YM8X6k2<K@Q0=89(o43bau`t_GXKo%30uZVfwO^Hj#ol~Jh
zeUB-4Ym}NSnz^)oStKI#d<sAtcjc^P<mm&$w$FPcggiu`4r2?nKh=r>x$G-Pr3Y?r
zdDE-NG?~*`bSmO%#l?S)V4FwvTCzJXue5p#-LAm_$c*D!he!}Y;?Y2EoZw>K>16E1
zLQ&O(?#G%Wy{+W?gW0m7kr7gU563eS(Sa<(8uO7tR)Wq->`IH9xA-r+axn2#@q*vw
zRu{Ga5RIctU8>7^dGYeI!GN;-g0^;QeBV37C)!EOj(Fbjr60Gl+-xNvf`XJ`Dl2V2
zicHPh%L{mapy4X`-mS58^r$;tUa0P1HnPZP=xO#4OKL{mzvtq`PZnxio+PE;<U+3l
z9a;AC<;)o2H@GQo*Jj<p&7xU$Te4%kydME`(Q?Vs-yFeHy5Ja%15eHCsso*hW3z=P
znP1J-tVphAJ$7uGgpf7%AhI%3M%Y`BmbePb#K+=0Z(g=`g&y_nfj|sMDnoUOtXh|z
ze%P0BA?UI{P%nlABLUNSGSHWd*`(X&9OqwaWH<zbWOmU+9M)}g7?6f4Am5zI^r@a4
zE3}y_2U5ARO9-|o7e){?e{f(Rh0nblVoxsM<+5NWww&+|jcD>Hj`YEqWnQOlQM^CD
z1$I@wia$Aj%8h+@>A+>gHPB&%102un)BLUVCmd+n2zSZtn@(+RphnG?1FCKD<EG35
z2`xeUr8ZlxY9NXHBYxnC0kSVkcKqA_y;?*O$#d;1BALh9*8l<-N|$)`2_CQHA$6aW
zP3}-*ab>haF-H>GAt$3usG`u*--pKMpYKfv%e~{x8vUQ4#9`5(nm6X{?zU%|uKUw4
z0YM6Li=qbzC3ejt$1G>*BFyNk-M0JIoUWWisVVY5)HZZAq~8+H{UFs9Q5MFs99yz%
zW?y3wGO@Qm222s5FVwaUa<{mV2RKOP8Jvo_rFPGKL|-q-gIdfAgXcfzi7?AbY=j}N
zKxG<0Yy-Q(G<NaKU`qkNhc{GRnnTw~XUo=BiXv$YV2|2$IU&r=<dPg}5<!9_?a}&G
zCcVLib6wxEPv^=b$aVnSMV8d@{O>^QNAs>8X%Gm^rSPfv#CpXk9rB%7|Bj65hiUV^
zD<nF!9Kn}*eNZt*#;>;5pnnqjJB@xA5#$gQdMx-<xS&!eq)~E3B*ef2c&t8hf7|*c
zsTiV}M&}(Fp@ZtK8@0f%frfuz{od&Iu1(8#JYFw3upLvf5v?D?kP8TTTBpES&z2&G
zNMzA<u#F!j@t~o;k^pt!Od3=i7_+-@?dtBsU#AL9{JlBE<t(GAmUHz|tXRrR*w*Ab
zUlbyhlauoS(z~GVQb^gw2tk5I`?F=JeX+t|w{OYbnzk+jU<Hq!%(CmuGS2gjp=2T9
z_qOsG2WIS-oNt<mFW+6?6Q910le)c<f@&dv4r_Q;7yN&_$Mj@tVtNH*op)y4%pF6T
zSN)%TwUfdC)Cb6>s}bc=+4qJ&j+NeCA1pC#C+Ew<1(=e3G5dZ*Tf*FZL@!ucq|wm*
zxn;`MHhapz-#yX)Tbk|DBx_2SkI~Nx3=*Pw?!6NH7+N28J`RrB!{jC;O10Mrt%l3?
zH>A%uRC4~L@JFfjtDh7U>?eOo7nw#&Zl>`#CrL<=w|H1-`>j{wSYSNGpP&izwU)Q{
zPS4rOtW~I9Z1%-DthTiRTg%ESS<$DHTtLq6>3cX82ccr!o`)l3`yn+C)R8WIH90l)
z9>99%LPbuJhlk}Xr;*SkK>-Tsm?YdvoV#V!)sxn`87Zx0{rlidV-PRVr8_vrJw=|Q
zBKbrR$XT8r7YhE=S~hv@X?i1|`CYDug_MEW`RZZvoQ~k<0<yoiY<J`>+ctMIStgP4
zK$o8r4U~hhMDvWsA86jc$Brcr>w+EMTfZ5xTWNACR!o_c$JpU|B~em(y%CRCJ86Ze
zJq*X3ud}t)%rtbU&Pkx#<g(XSze;2=eq*EMe>+>&q=o8!4-XozBTa)mfQGC=2D*<|
z{^ld+9{i>T-ZjSEp&}l@kg5Z77bTgy@VGLpU1Qz|Ies8dS&N!;&iVbT(o*{C)uYfs
z2#$xntxp-0SH86*ZR#BC&~^v)cR*%ZYA}4JSItil==l6$i~HycLhBIYn59Y_pD(P|
zEaU(~en)pLv)39(-7@{|tsFDxL_dYF?LH-@rrzJ}WWoi=?Nw+zJU({GyM;Mmw|zCf
z2gguTF{h*5|15E^Hm}w3PvJ1VI`u~CU~GS`@_o3jMx7`3e*~g8|CW=KVMLu@?*KH4
zs{sKc-i8)_>n|x%_C8Dy^aFUr4M0K>(1{WtZyK~Rm{<o{^|X`{?#47a>@F84zGL5L
z$w_YCA5dSMpZ}m@YD?s#e?OtEt!)+u<k8RTF?&twjBgf7#Kjv!mrHIHoWEj;CWeS$
zf?n*PeSAh>-O-xkeODOm<<V08G)&lXhNAODGQk3n5_a^niBS)m=2EHD20q9A`1#YL
zpQQHwZ;EBQ+j?g=d7D7Z;rmp5e<bi6unI&4<~p*~qS{v^ksx$vUZ8+6!3VSqZ!+sy
z*ovE>`g{V!R#D?J2vpmk6Tf-NUEWmWBMXCxM{sXuCU;^N+we}Nf1Bk5khZ=zy}QU0
z;#FdJY%{2F<^y?v>s%my1n|4IN=Vt(7(rwJv}tzg>w7<j7X4|jK4dn|=|a&%93Y8W
z<wB)$s2$!DJw7L@YU4+lZS+KJEoad!a<rc!-oOSFm}*wpD4uTW%^*?!#eC@_5D>Ha
z9=DH|@f|KX-?ZNNInqDZv;>k=5P)3J-jEr4a`RSaJ3g_B>$A)O(Ev3ywfsBPS7ZiB
zwe6*AabyUh{1ES`BR=)}2#R1OZNf_sh*{@1ceTgFcI$&1CNb}fEUGgLj4<v{Y;2?l
zH!syOgAI)=kMI4Zf9Ud5t-@mlYVx19GYtTTdTb>4{uUBhu~1X+)mX27^~i4aJTwDY
z6e&@eInV&N7a&P^(nkT@R-0+-a2I@#D0Izmv3am4U;z3wej3*+4k0$4eW*&xpQp|A
zIiIkr=ZuUz?LW^N6S|?@jd;74#BBU=OeM+rhWya(=izfVz;b0ylV2uGA_CFt-1N2)
zyCBc!-H4NR#*{11>c{oGN6))gEFrMMa539O4ZD>1IEH8RPW4N6(bF)Xlhq}-@8Xaf
z*?DJw)mV3QB1+oXid|}|;BP`en-*gkPwEx{&4@rS4*_#I9Iv(ufp9&xQWgk=#uy#x
zP(W{OLJCZBj;nr_w~{_`L=$rzXCrS)<-e;C0x$S?<<1HI{Pe?qVeaQSQZ`ulNyLGj
zD0}TZMUzr!%+}siGeOY_Rd8UaegFl;j?W&yX>t#iAC4(=wt_TWH8c7T_V;_V*h3^^
zi1ar<QAIYL;8$ty(2X@CtTNKRT^Wa?4j$(y652I2F)=ut+`FP@4Nwr`XW9N#gC9pC
zkVrcO`u5fmibE{WZ_?Ho&WcY?zS`n>QDXOXO{~QLa$d1e>YOmw5E`9uU^GY_94Pw)
zY$Zj3tLZ1xK3|Jk-wpisMn`0j^~GaUtqlg}o1c+sALiT{)XFq;@5=kDulr6GhCe23
zzTW~OAE{cT`xv*K>E*y~drXjc)#22aiOn1;Cf;<q$jDU@A%~R~*Gxa#wY&LafmM%9
zbd`uAADP+`O()ton4u;_w9W)6BnSxu=DnEx*feS~=K?7x%i^b0E;V^v%((?zwF(Lf
zO5F(f;-V4;bh=n41mt~-WxDl|93n+-M+}10q=bJU?pgJnweQZbMd<?6h%b{{&WU-F
zlkj!SWVA&kC4W8>6Kw8SWehSfF_qf1J(|uTU?%~*SY#cAEu{Zs6+O7Y;eeK-Gya(}
z*q#gzq4m>VLF3)YmVtrV)Y4~<JW1R5NZXIv8YU(i-@|`8e2kT_u0_QlgZh}?pg_`B
zZ<5XXG(e}9yVM*GzSIIDNZFApy5kjfi-v8EXQgy7!Z>AYWs<gv@#}xUW`yPd2E6Kf
z1F>p;iZ5~ymryr~^EjSHKDsxA5w=~Gww*is!v}aAyqt_Ip<`^r;n4+4cUWNSLAhAG
z<ka8^8U!d@c_Ba;$Kb`vNvr#?d6Pw_;#6GVeIo3NKpFg4@FUBgk=EVKzD=B7s(=2|
zkiSijxZ2l|<&+#zNOhk0>%`l|j5avO9phl7Pla!HUxj|BJF*>b*JdmiweUqD+c?`z
zVx_@oevNlT$o`sJa2g<F)m0gg?<3xSk9pOW*+|vgfryv4XsE%~m|Uo^L-YZ~<zC4h
zhyr}oJtRD5eo_8H$>R^GG|N2+Q!a;!;mL{@HV0u9&#Jw?R7Y{YtTf-@1UasHHS5(I
z{QJi-+b&$RdHSZm4n~4Z*^+G8Vn`Do66LQXo_mJChO`xli*U^!IIkfB4ydl~SfWTE
zsmrh8MaRi_zbgZDsPy1MMewv%0aiGG+KXIfZ_d!a)ods!At3=7=N*!$8NQ`;-ScRM
zATrazo>1_<3(Uz|;CV1&cVR$DOxiy#u5xZ{B-oc9UII$S3D}1lE8{+VkkLt#R8czC
z8k9149K9n85i}2Ay{e=-QdYIyDnyw}mE}3^w45c!eP(UG81o8v%WZ^|!GUt<A^?|s
zU!+>#_i$7{$NhD+mcx45(#9Mgl*YpJI<>xZKHW_F!m97@e`x`3Hg9^i)N)Mt-7hQj
z5V1t*JOc49r_*n4Qf$FDhMI$;Q5Iu*l@F6zEZp970>JBOJR<5CC2g!CA(2rtR_lI5
z>WELC-{miWkQeJwEApT9+~`~cocO1uB^pi?5szGtBr9G({E=oX8RlSuu~0IyeBQ`L
zx%r)sI`~j`r_@}Lc^E6b#spDVlzZ*ebf3j6m=^2{o^Gt&^AK<Q-nj?mPoPz?SES%-
zE?iVX&xcsq)|JN(D63Yy#}xglRAK-@0DvG=ymtRQmX7CUj|ft%;7gcqkP+Y(MS7*l
z7MQW^U})OxI_cbs_QtT~H%qR+3Mz)MV6~ggVYudr!=E%AUSEC5XGF{Lx~cH@a5j)N
zi%5XG>Mn{OFXh4gdlOaE!Fk3N+W-N)zld^640vr%97|MPgDudY(W6thB<edzm9{g%
ze?p>DJfKNmt-u0An5*3O&Cb1&$D)qKW9M5R@r>!Rn1D4P#wbuo2f{3-p+CL)Z%97^
zP~m#RdGq~Y%~*!t<C$TS&x~vZ`1}vCCSNI2s4;h}DsJB7SiE6-DlVQ%y!1X_<=@R}
zCfyX(FvIwY`mc6Vb7KKqLx9480R$rH3z0B>6N)_9PLRh=7-AN53XWqqk2GWO3$#>6
z84O?91Q0Y3QlB$f7-9!}2dA~_Ciuj!cr^ynxSdc^A7<y*7c9-ZIBTZBl3MMdyEFii
zhhKZTSru~Lj_x%{5%Tx@;v9@68a|)~%`kfxZv|xHiMx3xn4yz0egavCD@7!NNWqTX
zCm(xf=oU&lu>zcR5?@CPfo%w`B~wF)=?-TAD0B|d4@ZfKshbTUh>~zbBcyAk|C54z
zAogXCt#K(Qitv#}!TZ5aKALW+w|YIXz634;-fy3+-&<{{r0&(iQ&bttA9yl`42hiB
z<osQz|EK-n1T$B*PFvUJ&}kgIcHrXg`s+x7dN9%jzQ0#eQu0xCarr%CU3Ph&S@NC`
z4I_-Y($2b8Cu1i_Sxfo{tijj9!a{2UEfFk0fAS6G;L5AQ-%<iT7m$e{1sc`rJhh_^
zRe%5f{p*n=H_D$KBF=1yjl1-)bf?jBxF-A}*l-D+SLI!HhCa@16U_S7X@X3@?OD?F
zq4(aztb;P^Z4l3^c+>8^3&2>-Fh<Y~a#(vdT_U+UV2SSX2EMNJ**~3dg{ECVZ)TUs
zcJU<smX^u-^BLQ&)?01oq}2E)JqXM*rrL0O_%sh!ShO{-DGkC=gF~l^#YZyI4;$O*
zhngNRI?(Na<w%ZvyR0XywopWzU#KMXtV$J;aopO2p+885MM|o#rKOc_K^Vy8s;Z`@
zmW~>w%Ej%fYPHU8_S!4KTJvd*Z=rs^T(!^}u=;+TSrq{r(!_H|?r<ziqDUdga*Y1(
zB|=h}iD85f7sVnv$vdfnyATjs{pn`Pq1d#*-Pe(VR#+di;cU0x7-`QPJ$;!(pu{y`
zt9e{=LIKYL;i4D;%{$KN))&9jnbj@``BMwn9ID)33i@IYf?%v-HnjT#qmN@wVVIzR
zW_QEE4SF|h(RIWis7^SO&D<kwf)}_p;eZ3_#rhV0VB4(-c{W9j&qU>IZI;SZx_Lzz
z%zzvOE&dtkGcsxiT05Qy+lOrzVG#wKx*tNT+D{AgQ+#6ejLLN`sv##I(ju#<|6DZI
zo>d@WlVJ;K7;0r5)^`5Q_6Td3w6*H@U<c4!(39$(bnd})UJSI&y(!E<M+Nd=tR|~R
zRIk?`ZI%C!lgAXg+8#e0sdh1*(&u3cT#ny)W=`U*Mw1CF+SeQ2u!tx5qVD7+*=@bH
zhyz4Wj(?v?THlqcKhv%}eqMRN**{Jf_~N*e@V9((WeLBh{8c}YCzJXg562(ruCDUC
z{#KrE__FDGlC#hlL*HM(o8ByY10ELxY$${O&hD6lE}L&xIx~@*hohfH*xrHS&fg5k
zx%G4oZQxR~Fu){**vFkXwJrHVaGGA*U_S#+@6_es)h#KLDBa`T4Jw7sZ~rO5woPi(
z{MDIvqlxj%QU}NdC(wCa)>n<*FjxRDQ*Z(?^@k1=PW<Jvd=$PpK${GSu?c@Wz+o*t
zq6<kQqY741x+A!Jo8L!zc7ovtRS9ad%Zc;naT?D_&~EqRXDhnNqHt$r%kYQZrG44d
zW6^B&x+?d53-$~ZN`eV)@Lr7oCsPJbcaAaDJo6ru+><}O&qc)4I~X|N;`ca*yprnv
zyQBpI6)NZcaaARw!{aQ40XbMZ)3lAs@8SW|3lANg;7hkv`wcz+AU6Bj%uV}9{p@oM
zSD?`E%vO)DPHD|HWB;Za_RpAp6O}#j)cqA=m<<{EBCp24@`sb$(yyy^YBLyyJ&edp
zoNxj2*{B@4q7YJGj0uILF5lX(Ei)TpaAYh$PK_;99|$~Mq!Q@b1TNKpqo19LtMHXu
zMXcj!zX5_VnOBlqqA{GleeA9AumKi*7h;;x&-Zt;3~#1!!16T*n84tlNO0_XA&McX
zd;7hsb+Vs0)in?#l~6F#U1Yx`<1HnB7-8pZ&~J|Tt;>le{lnDele}(d6DtxWu92St
zWS+B(o8$aJsNRR0$!@tNrg=_KzodVp<TD~6C+(8EHTGCp{dyfI66Mv|$6b1m7!uOJ
zz~qWAhP+cjQy5piZ0FJ92-Ra74Bgc`j--QGOjkp|*`Ge5c@5d!<oozPtCpZcbjba0
z7Um2yszwvi3PX#CuZN|9Un9aX=KBA11lMnLBEF&1(ALUJF8b5aRqx-pq+bOicaPA;
z_2+J7nUGa3P{8!>kk>c_`zt^ur*zvmC*N7;Ib?Jir2iY9=*Hj8`kCTp|J`~liHUPH
z{g5Z`CQBj`<;=OWkXMi+w3(;fKq+t;43<gqTy<=-LwpRu75&uvNHWo!mR}fcq08~~
zOJ4?E02L&6NvbO7Q)xw5k!x{%Y=Ux8q;aq}&NJ-`7#WIl=NlRQ+?BYeiX&>8a%}54
zj7?V4h__X`oUp(Yf3)X~!&2cA3j3?(v~YC&`XiV4b@0@7VxARr5yp*}BlTCERrWC|
z5I!Wgt8&r$8$zFqSwR7Z!;8(v;F3Aoc2@@?(xf1-4Yv#bCm4`PGFif=*uOQyB2-{c
z<)r|99Gq=k!8y6wYp2z1WTj^c%2=&UZ-7j{Uhj4OL>XNCRi-RtXM>nyd|*ZRW-Fh9
zCE0vT0rai;27!}lwGgmzqK|1!^vm3)=gDELQq8sk)t-h!m=?iXKZr*%btC8)zEAx2
zWIPSs0Rx*rJ-EHZ{)hb=_P1B($yLD|f^+%KHT|bc&r#fmk1OpUz-kBes&-KY7b6Nu
zZHYpD^WiMtg~)vVHToUc@_nyv_~c@nC*Ow}BT>>+6tVm5aBd_f%yTdPel=80(nl||
z<`4g;nhE)m<>%L8Vd*wMK$sb79H<NfTH|J6LK$KM<e8nf04QLDn}ps)Vvgt8`+++H
z4)m<Zzp?x@3JKb&*Ku5fl=roh@K`DjmKq9?X)#)IFx?Gayn2dh$B(EDU#cetC6M`Q
z#a`6D`6UDn;5v{U>!_1PLjGhdi>xw^BN})nMI7CQu#cx(r!5oYlw*!Zz)9%8w%$LM
zRsX-)jd8E@d-Se!Q+$%vhi8?*E3<rnTx&ZdCP{3`clr&U`#Q%s7wt~N35pBDqKY)>
zkh+x2RmPRj#yG7&gH`#e;v+&HZB8SN;_(X5baJpVf&OVlHGaHx$ZD?uKw+2w<DV!T
z#L<|Hz@6Ql2KGMZ=3FJOl4#2&ESS|InmJ;>{Z;r)aTOC}GDqg&h8?U_yRj24xKa1M
z%+TMlVf!q*%rtSB`9g4z=+<t?QOA|~a2cbTX|8aqEE#vdjUsqlerqQ@qX#{$d21;9
zPdwQF-c#4DwZPxt!sOtTCycWS{|f-WVC-Ov&ir`Jb>eK=BgIrC_Vk>Ip-<aTnQu(*
zuO`u)(rq^`d<eu6<pQ^_*73QwPc=&?&$)St2tv2|HSEamDEZ_#3urMOw$^=;kGa#N
zkkrNe5PT!lx&C*%Yx*EF5?j`Dkhp2>5l<BM&kVsb7`lp)xIsI}Kv;b1j%1W-@<l!~
zG3GmN8&?ePzpO|O{M$Amp{be>tmODrU|M@d>(^esXs2$Z#xiX=<t{hHM-TFF;}>%2
zy{*V}(9c7>dNkw<*Hp<d{q=|bKi5ZRN|&YSvOgvzD@uJ=T}a|YZw<pZ==1wnCi7hP
z(}O?N*B|C3s2~}o4le!(>^WPoMjFxGQ>LyUNfM|1${&bLX1V&`?GhKS^9MKVdT=W9
zN`m=++WwI#q6kAMD*Q<y>H65&@(16HK+Ai56#F@Q(D&9<2p>$xRQAUVNfb)8k-ImS
zNlVA_cTi}7260oZAoRGVqMhF~ot~#^AAzpkAM?w6(aM_L3hzIogLUB?3rro<M0U#5
zMBKW9^_~Ckw*q#fJ2!LX#eDZe)#h{LAd0$!PjJ5baDac|gYHm}Fa4qi32#Ej6pqM$
zL*YTm-$%8g3bPV*>d^n0=gZ>#!j?%oJfH46&Z8m?;(s=Ew}^SS5l1RzIVxc@?|)b-
zb!2N)uaVIs5Jjz}gzNo3msbq0{z|}GFn)v=5skX`i11~BDXC`{XSMUEDJrN_QT-?O
zVwlFcl=CBVKPLozI^qj?nP|0;LfT!9e~@(R+eXQ<|1^h{2rux3x1oCvv}P4obAd#8
z$;yE2H3CfDVWR2g-yVd21FX($jpRiY`3Qx8DDu{0dw|wb83I?f)U0RolaS~et^O@d
z^;n|oM-?XQ(d=A6#2ZbtQV%$z%IWy%h8jSFohU0H5(ij8J;F;M3wQeee(YR;{kj-i
zs`IGY)O^iM9qLv*@KB!Md{cVsbtKKRp0wW0JM(u?001eJvR;{qvywvX&lcQpGFqgX
z?`u|_0M-1s?7?=a+P&c2G{sbtXS@EdEm9yTraJ|46sPU4HLgv+t1S=Q>>0@gC}BYX
zH^c9xtaSv+lnP||?vz4XTTB6T;FSSezJ1|H@lSd9fS=3=y4f*-!T}M{Dx;as<4uPt
zsynC$i)C9RaI7t-qUPrL1D{g`yxqPyqjh76Qc+RCF3E-({}!1?jKMPD1k<CRTHuBn
zdkA)1=yozP0ht|M;dTg3Fi4n<!RSLuS59RxR;V$9nwEqz>{1vTUUU{M|Cm=RNVxS!
z^4u-7YT=h(_lDrz>ZQC_P3au_+7KUtWGu;^QzLh*GdD@^2y$QtTlCNtiY+S#QOw){
z`t3o7Ehk_4XaHBqj5WmG-DqjrlAWXv089h87y*&|&{sQr)S-)(96A`*(t*!Tje0N+
zb&<G9#)L7FIP><wxkm5;s?-MYzAMti2Q@Xz97}JIK_J%!j&w3lSC?N-CTAtY)?k|-
z)qTQ%h|pL5@AkIA9#tQ<=v!%^586q9QlHS_)}VrG&7YGf<{crP5#hT}A?O~j=<)#Z
zK3X??E-Yy6%BvNq7Fa@)1oGXp)~3goB<<^w*iRtK6)aXIA1YN$*87mwzt2{VT#~j_
zY2<M9eE_lj(N=8p^U3$&-^kSL$`Q}J?~owENR;p(VOT+~9WjV!3c;_!;8aQc!i%m5
ze&25R8nvIEqt@Yy;gdfm`w~=~#4D!uxMOP;Y`-35Yn0$gykDcf-`(Bq>go!>YzF+r
zz+P#<xOFZ67-DW^C*g#YgajTQp88G~N+4D=wy)ZYU>E&1T<||72k<2``_KAXNiosU
zXtrEk%9RQ<pMFoc9y~c~8`gXWfO(wn;}kiM|CSRn^aDO+uo{5Z|Nr<Ht0W@^33F=D
zE8iF*>3fex9ccp8gKTnZC&584Yjw){HjzDQ_{U^_;LdYaRbpv`m}zS$Cm7*s;7YtK
zeOoq^m(^K~$6TsVfC`K#Ho-nd!q~bat|3tBLDHAl@{S_#@_yF3mL*v>X{&?NhK&#i
zVE`OpqV%maOho_%(tj0)P~go1#V2(OK-*YAaR|dS&@9*M%kyl-KmK#Kn?<cMTq0#s
zoJTu+Zn7xxT-a$%K*V`YFG1^`Z^*P$rhfO(yi-s%ihj!W-lKRo9GEoGVEiNf+ih<l
zvY>ZxO}Byq<B$3m&o_I;K|ENQoZIcsN8o4ZN7UFu^Am5LyX3h=G8RwO=eP3V-i9oI
zq&RpS@3LzH5y5w3@22aw6xjA*x1<Iopy4iYB2)UN<s3R(pz0=gm+4S;l##+yC%q(X
zD1XwO+(h5b;I|V&4bZR*Uk&a(2aXL?3D9i{g6bUP4cdHoK&xl9M>a^z+}@9+VS}$c
z>lm#JM&*^?21D9W&|WT*B;_oiVM#&+QT(~}K!sbIf8O(rD;;;)U1909Wrx}j{Uzd3
zr%|snMWS@Mcz;%w+URHKuLlcK{F6kQjEm*Y@)+&8v4fY{`ZOs<n%hc4Yf=71H8Jp|
zLrtQ)f}UO~tA6@!E~V+oM=>(vu&bqtaZ{`@ewOY>T^AV5j<(OsBR8=h4?g0z^RwD{
zrGO<B58mi+y>pnxx@>H01h{f(1qB5;IkXo~X&fs}(&ngmf+IC5A{xIqN^Jozrgzwq
zPCEIK<;@QjV#25o*ZR@F*>|s>LKSE!irh9Qxl)St27^!n{hpyu4zc8O?aGtq?PFJ-
z_d<YP5yKVmKkEwN_VXWzKorGhK%o^OAdu4}27yde7yDpA;#|jAzKP)CjLOX1OrT6<
zxy_f%9`ZN@b@#b}xMv?@0C72wG5ICakW3^jNG5UlHzKeQSPl#9$$T#-!j4HNN({OH
z@r!AJ0_HzdV-WcvOHo0=pwe?FTP!T}9ZhdfPw)#0t+o~HI}R5xVJRmoi|Wb>xWz~n
zv2kj^g8{}<W5Yn*hnMoOJz4^o1-t*Vr^IzsYusn$-pfm5ZEKZ3-y95}gb^y@2e6$A
zCr!v_Gc#WREB0#j>j}p;uW3Al$<u&@+mMqtIZU1p#4MbXi4O8Pe^r!Yi$5@UL)Xyq
znle~sGtR6-D9ET80)4u)3L>U$H1k~ln+~uyxZQdgsM&#QWyxyS$3yLlEzb=%s^$4J
zDzrP{XGO`WcD~Mk;`-9CNy*cpwwUGzJF+GQo;~-Y(PGQJ?QXg@s~3aB)VV{A=PU0V
zo$hR=+5}0L>L@M9E*Eo|8B*q79#Ab6;C`Zl3{X!^k}=wb0K8{st2c}uh(Et<!9Qg^
zG_gW%dW5^D1UT6j*!+Xdo@DvJ3i^*tRtvtHNf}3+e|_$;m3}P?B^{DB+nsHOWgb7A
zU5kt*x^GQ|(XtkOG(B@pHaPM$bgLM2I179UMDhT@0}rvad%e=kVute#%$k~r@=+){
z&#{ve1NQeh>yYX)`}U7t5f`^sba#BAhq>B)*96z6xMN8+i`T8K?@HXK;}ZClRuVH>
z{UZeoDrkScK5pVze<1eY{m%nkw4L=PBX>MZ6_7vq<-KL}&Ga~jKAaFd*@w-Yf+oqQ
zXv-~mZgOc~yYX{;Rvgf6bb8&;lUO?NcVI}9LS%5PW~INr=%pzaQr7mz>dVIqwC+mV
zM#G9ENqFc+($>-v9Tp5oPL26#wVgH7o%bM4KV{=OUhL6VOTHe_{%P$w<L7L~Qhlaf
zk7m<@y6cSm0BhU-gjX@MXRIZZ)o^JlM<oAMZVgtG%$uVl1a+??*@AV-l6LL@WBKD(
zfF=b4LotA;@)7{1vtkZN*>uW)ay;UBa{z7?0jz1IP8oDH#B8Cz-)6Kt0sCjGeL{|^
zx8mSCgIL?ew@EOLW(DMJpRLlY2NyDN7N%X+OTVSQB^B(<XOA#j-oZ56SH;l}Q%r9A
zKy!Z^X~)j<UdDen<{4k%VS6H2h9$frN94q*F_P7pq>n9{?*7?ofR&_gpqk{gpkl^a
zfK0Iiyya=|1GeQ4J}RoQx>k}Io&>U_anSn>;P^iTKQntSc)V>gcGlYht%touLV70Q
z<6av}ioevoj?D1Ty+|gKUrWL%l1yxdlL5Nd-ZfTJY|&3!FU5cVr(>7kE-6M|PXx}1
ztc{S{;r#IMu#!7q35wGyG4B^@EaG6sFvX5Dv}w(OGOqt+HE0zx4haoWL9dz3-#-bz
za8#n|3iF7*Co8S1D^&+M-HmhtxzHo`&=&)Nmy-Yx!_T*8n_MfCA!7*9+S^s+w^m0*
zjP1URTHG$vT%#&sUhX4ae$x3@t5O;Y^A6UWCOzS1{LiZ+V6f}n-Q;}L%e_C@y_F^s
z@_S^EeaoG`X_~w0b1`zTVs`OJZoSM^iM%7wbY60Z5oB}pV6WaJP<QcdU4H`x6rd8<
zO%GC9S~pB89f`KvQ_jK%)sjD5frq5rRA!m)x5IKh?Z{;CQegsihVv`@mv`*PEk3vB
zo6oi39h82*^MBg<)O67Xm9J>#5S@?Pg)stT<baYmgUvWlG8g&9x?3wzKUp3MJx2(6
zS<n0>FzU)Zovv|~iiu(VqFRq*G1%ICJNJO<<}I>#)r<I02F@US4jQksQTa11jauWY
zJ35}DFYC`um67hfD#my>{cE{m+a4dLR7kGH>9c~~Lpi<c5@h1-1(-bc)18n4$;Y|x
z%Uvzd2xap0Q{J25%-ca7*@C+AM$GoBxvTmLqzCL5LXt!Oq2STMwF!Jj(|Yx7uUofX
zdIj3o)|&FFF+b-u*k&RSy<~4*vF|EEbsy61f)e1_(I<-&758@-8zG(MPYw*|OW#<E
zd)w*4QITx{s!+5}^2aob#TZX>qk?`_fn4*y`u^Q4|2(ms#U?^;vV^kFc8Y}?yDb`q
zg<+iY;%}>7uCrf{(m49{&GCY_o+&quC6uh`sRAyODc%`Ge%|eF{%KEh%E7l#RPo$(
z<bEmhRqsUVXC9&eHZ*ly6!D6=?;AE1h%a%ajO=s*_efZ%A#@^4x85Gu1y)v8J(;#M
zt4{7^&*0m>05jEvK(pET?GfwdXCH^Yg>cks#-&&B;Ti59Zd$7wCfK%`d-u-$69P;9
z$)!8Ij>Q%9LF2z5$_Hij?VpIr8?DQ0$NFl2GUIvk+Ro^CAG;Xb!KUuqu`P1ySIjj-
zj)9D0!ZNVR&WpzF=DJ|P$QE2awHH`LE`UF)IUjH`NBXPFiDc)@quhU2h%mHh*YA?1
zyMS=08mPRiP%RjpQ8B2PdwxKXPwzpJuW3G3h*j7f&seBg&sz3-6k~Q8wOV42nd;+G
zj4=yYWS+dgl+A6fPY)+NUF*0<E~>wlW&HAMyxiO7VgT+M;Yd|g=lAh!ec0XUwm_W*
zPC=~n1db?sZE8W!`(P~nbn}?p@(Vl0W2v}@iA_MkrL46aM|P82!Ol?%1cQ_>B|-o1
zy!D^kEHB(JEYTR1Cz?=oUs<)Nwu6D1kOTbg6(ky4fO+w5fUo^52F+r0)_W~PM4HAg
z9At8xt?FYPPuT7U2?0F+hJ9D~fAN{ORq>KEhkD7%PHHEkTIbukB?{}pKx=_J@1l1T
zqX=4Rn+TEFTwlxG5$NN}!Gi*Lul=8*YVAp{2eW}OV`X4?4WI3ObFA6a|AlGDx^}U{
zEo#S8H1OQ}y$BX8l#eNPGf?>}1zwWz&j8H;ER_+|g9kArNn;Ko^uAZzD1pjav75j|
zDAoAcX)2*F2WJ(K#SJ@ygt;>=xcNcvKv@$0%-ZDdJ;Ke_F*Pyd6&1%uM>z@6UaFeg
zewb;CB8^>`>9kR{nZuD?24lr;0$NM}NGufomWn>fM~BZzs7oPS%WCWDh!dFclR0`u
z5~dKc&n^k%lS(_s5<{f&QuB%q^?3b#)!@9@Nt^&ctuyJKc?y9Q$(`UPiQRvceOj~o
zsx*R1+`^2NreHw`{E^*N&BTZ#YSfG!Yu`m803N40dmQan;IWv4(T`Gt>B&ihRs0;K
zaHUtEm^R|9Q!{G7U_n3$(4H_~K59Tti@K$rTR)FJWDwO0urWpU1N1IL`Q{A!DIR2L
zAou<K_BDRPJ^(HRZsk83MG_LN-YIQv;JD8n+^Dts+eaMpx{Z(Q!UDEKx8W7=@!uAi
z|98db|JTmHBu9z`Q+S5ko|<>qSpTI3$T0r-V-7yJLR-el2>c?EEuoV!TKdb)@(B;g
zqlTofii!9X1xvImud)pVK3Dr?fdFpZBLma6M*UZ1@VAa@O1YQ9{ALKvGi%?MymDkD
zWn4I@Mef{z09o7%cmOOWBQ=Vb9JuqP!XK75okb4?fEth~jn%Z2abNO!FuJ7iE0ID=
zz$4EbT)i-_#W^B|Zds(K*<4RjMqp-p9>NiX+|JWt;RakzLj7J%TN;C^P7i~1q@Dl9
z3;%cC`Ty?u2Jjih#=;Ew$dYlM&qhbYO?}K;5QyDjXBkt|99uGC|EfP%z`BYnml}v+
z(ILUu0<|YP*A|ele7U=mvb)s^UV1oCfRlYU!`u;T+lQ^*<^cyARzq#io+A^*%$Bn@
zKoxes9E{!i!*?)so->0+E*=+><V<E+h)D6JmArYCB3z<iY}<-juE)Kwd;ObP3IImY
z&&Vv;M7~dE*~cG&%Nmu+$mdPZgit*fF4r{K4;fFq)2pdWw8uv!mK+h)!oz2{jYOAh
z?3$2A^0KZU(jf(5r7JitlHy*%hhx1d_;UQb-QBy?GmFEV{QUk;93Z;9J?ro9ueF&=
zvX;OKE)f(}OCWqB&AP1>n_ts<eA;e-5R~0D_6HLi8@tm|qhtmlcfkg)E)uz-3y35U
za~F&%wU|0B>lkn~k{n#>gh}m1$8mzbx;DcAN(!=ckrQx^`}U5Q_z*U0-EO1P;LcJh
z>va`YK%Q*!{V`CZsq5pPuF&@-ZN$9!kLhXIBES_h*+KN?f8%XEdZ_@x=L<T(@YFMi
z_m2c8JWWpSG;-u7`D&+HGIP%Ff7GQ>uKiy+0H9EIU++Is05%kxTAly37_s0xt!wZ7
z36jhJ-6GsljH3(A_HN3ebkcAx0iGkNP&vEh8d&GdoGHEP@E3tFwuHp^IoIvo;8H0K
z2VuliC+Q-201I{Sx5(uLY~Ke;tIQGa2T~*J)r(L1l7daQA+9m`%QL<gI~8r_Yj_Z5
zQ-ocI)Aw{kGIqR{b++S=jq=tjM1*>rkyxV1kc6$o)ZnE|kG<=SXE6W%(PvPAg+WP&
z&%J9z+2T20-3N(~<%rysKlCuRZzzMiQoSVG)Jo!Yej|WJGM7aEvF)M&;`@R`BQXsb
z^qY)5k^WC)xWGJWni~%UsFD8x{%Y!l47)JG`{Qqr))h!CoLT>ofNC^_%)46+L@QqC
z0V-q```CaWYMKCO(d`fg^!Qy&&e-oC#+zNxm6d$(cCZ#lSed=qE%+aW;KhLEb!(wC
zM!GC}2m^h?3nZbWt_ZF5BGiqN%v3|E5dZ%WjVAkOUZ>3MFl#r*u*o0)0qjaD4&z)3
z%2c`GZ;`#9QT_qs34qkDgUuvt6qL146aUHNpM-%jcO?t)l+#XH-rSMLlVlno3Ep!c
z3YaYn1>L7FIV#0r71vFZ03EKi6^nVd;dJ(J^PkQUF<^QPI+;8u0IfBq!{2|l0=QRv
zHL{nl+DoBJZd6_%{<}Q(Mew=R<3<W2Tw}_J0Sv*+w|hWO`)l^7XhoRB?RahFA7*Y}
zHwfPfV5KKgO3E1oOZhMXmCk(qPX7XWuEchc69GW3>TUQ8QLVh*!=K1cwWEFdAr`b)
z1&UvE0i|bXMS1+|{+rSE?q!=*51!oaA3ITg|4(6W9ah!%{P7;TL+M698bP|dyHn}z
z?vj)e=>`EQX^@aqy6Xr?gLF$b+{Nen{GR*Yz5jSP$Ia&Kwb!0CYi8cB@i$hQbs856
za2^PgqC56)G^w!+)J<jS7-2(y^>u?QB_AqxC~(qQP2g!?R5`DnYDO3V7Tkfj%Leoa
z6n{Nf)er2j8h=hRF{BYzSE{2$!Q+Tn2rFj$BN~@mmbJHL-DpB%ws-NPaud(={mt^=
zSa+)wZSzZegExcN-@k18bX?c5^2!@(@O?dFsku`zt+%~AR-0<6VOC%$;`~RU2v&sa
z18q@8RS=XDv&9c#>y$xW<9)Ll)U>9hD=zz!3EL3Fur*8PjAJ*t5`x53Mr__Uze>vE
z{oK&reXd8Kt(4Kg<#I0&7)V@--n0;~`3r@?L#fDS-s>fIrA4EICzk_^bP|0u%s6%E
zuhkRp4bdQ+pLO=&4{HA(>}CJr|KfT7#R#3g5PU^=F6&`K*T}cjOXVU&e_mNm&m6KH
zPS2Z+Vr^$GZi+Z=iBDbKskedzsovZ#GZulI%b+BwXW)3%|L*3?PMJ;nI816fPY5J<
z=BjgwThOY5emIJ(1$AEsO`ypS+5!oz0Qr5mX^Uy4x3v}|z4kC^NjFcIM*?io1rCsF
z4uhjh+c)jdsU0aw(H%+O&!0N*vBjiRy*F_*HL+h@;HkkdMN`SXz(&B`ZI?a%0FV6j
z%Jsq1$!q(L8k-vBbC`lF%Guq1^TDX_mGc!SAUzm4p>7~U6T_BNDmBh$&Go>Uooc=r
zCR>^*A`9G`q7yNyHml!t%D2gP^MZ)VU{OW3Hx|A^4k`7-V>*8tX_@m9+NwRu&0ovk
zs#t9C-At&rAerk9_C}K6zknALglE*=N$AkfvAl@Yp&jFUDOBNhQjyhYGR?bR>^sRx
z0mXs*NeZGaFZ$m}4Y6h`vR<0T%<xiT^W_XtBiRbzhfsdp(^?zk_^y&fFLE=)>Q*v7
z7j=vN>9RUi;oKHo+_ky`@-tUN32$TClM@77r%OY^$+w?w)SJ|djPgmo+izLTKdv0X
ziNttmJDE=_eX?c@=Z4kOCY{$=s0A?v%|g^q^;wOqB2&kaI5B=@K$S`Xwf;B3MPQh5
zC7VN5463K|b&RIRSL&nE!eia}=$jk6pGN>6kG$n(w!#NhZMsHUF)k@Iq_;-)XSzU5
zYopoo|G&XINZh{@S+H2{0WuWQrLnBe+N*xNRJx|`-xii`O#R1YGQO6(?<w6{CjKs!
zbBL6{hF|l=JD8+CZgz@)yi`x;S{n&Ruj)q<c)IUQePftW<vC!uJhoK{BkE(!^E=Xs
z0bH(Ei%~Tx!BINTW-Hp$?HU*L;ok21PK@-2-Z1=&0*#dhI=4-#oSeaT{H%@k8VhUS
ziktsDFmg?OCp^9XT4kXvAo@;Lbt%_a4|ZQD?9=;X87$YxOWnduGac3fC2V-L)}w{j
zI`B>V?~k_&c?CSQvPD8iEPJC0{T1WqF(WCFyHfTxY`fn?>LILD@<3|(F{zVk{~I9Z
z*Xlb?E{(QKH%L^IWrB5`{|(E3pGtf@zMzNH5R6aGWi*gs0EE);XYnVjGBLASIbP$6
z#isKA9h(tZG*MDqOiN4q_ii+K+wTG{x<JE`Eh^29mhE|7Xj9{?)!q!vXc@fW5AhRx
z+oJHZaxO-FY;cA9pxstm2E*Icn`C_N-PtveXEM#V4ku~QB7{9hSkc{Ptyui-kPrD8
zmC4pGk7~Fv=`>hSLQ0Wr3uKOCjXD{BbfZBd+=2pHA7-@c|1<~9u@)aCSt{-&*4y@;
z*e*`73J3^zU41%66uNcEJ6fugy1P7Vy*d$}HPkJXqNP^2%%egsFf9Eb&RUeKg1I|C
z6KaPIHI#3r1!%3tdZpWb?%B8GqnUSGR$1jvCop2z?$HKhZaLy>&b&tfg4&EXM*$E~
zmwPq~=5LM7$RU)rvKZ(~#`SVD``C~vKkE@GOLTd7P0_1<ILOzwr1HiE0p&{u@*g*=
zp;%K>Q{VLu?{Cf&$W_#E#Rs~(3vElT{{3T<YPh~WPC_hlO;U6s<W?-4#N{d1WE>7s
zoF%`W(vwx1B%PNK;_FZ^kwD&cZ@GIURMq2DS~2S>vn6M%y|TWniVCQ)lry#XF~YTn
z1=fmb+kHFJ_m6FRrNs2-A1!Ru;33TXQrc?ldxs<Zx6rgK#hkn>VfyPCf*2q*)n&%-
z?Yvt;4G$?Oy}Rr=UIT=RrOu$vY1{W?BnnV>7Y*j&=mPJiS(xb{<oW}qkDc^j)<#Nc
z-@ile{wH{R%JWsdY{O@{!2uMpCSuHTGsz_mbEr1G90oZ^9PkhG@SE{q>B6%9C@EwH
zT-4bJHG*l;{F3`n+>Bjr(z3A;XL$Dse6M9P0-ZN}g#0#+PZR1nontuy1S<o>cUFEP
z(FM^f8(1$TfvD{yF3}TPJcM%b;^J_3w84JZkXORt3HoS{Jc0NAk4*VAY6zv=RH&rE
zOsLftX0XL8<<Im0+mYul;4^@O1_1@7GNw7oCG^o%4vqXT(IB`|THo|qx{eobD`|bt
z4d37T`d^Ojy&G+a-8oUR;d2;2<$!TeSK;M5n&{(#Bs;bKs#R(@YMkX=;Yj`Ao547s
zk;^;eb)@~@Gpwnmq*yIUPG;m?C}*4Mwle@>!T3||bJdvbl$F+>86Bzgd0uACQHvZ>
zBR}f3KYDl@BUHCSG^coZT6OplOs8)J5%|e6*<PX2O-|)-RKFp{Y`;f-Pz2XIAaOs!
z<|OnTUPCXhIf57eoqeQ4{wqwL%4AlK{Hh~}bb>R-H|d9e?f*<Rs`}2c#OVpnycaCQ
zah%({x(}`O`Sd6Z=T~&@hYWs%pO>lx|Ia3Jf}^oGn4fkiA;zrP`A~IR0aEZF54{TS
z5bvQ~8!O!ke%s{0;d&8k20pfxLJg2Q2|zMyh+t$^C0*3`pTUM@IL8auV(d@}f~aQm
zOCvf1V=Ez>Ukz_npE@>R`FsX6qMxsj3#ur?J6x{=Lu=)g@V1yR4=WFK!B=Ap!@V8)
z#mGhBX9tOtSE)P%zfXA-k9#Zc#%=dHLgoW=`Qv<8w)lSr*@fckgid&2(7O|K(#@m8
z|CY+0fZsRb$L{<qe#h+~YTyd+G6)C=TyE+WGrWix2w{KFtp0>uXYw(CRdpI1`UFJY
ztdqV6*<4l@v$-HNSyg)$9oL=XFb@~(+MhRIUQv7Y5-dd8IjrZ~V>zMZ?GQ*o=A&!9
zbV<`$z^$Bc;DCva!vopCOfoRu<@`oF0v>zt3yrXlY@suEN!^r%XXR&_QQ!|Rr`NPf
zrkQI<Sa6xe8cBF`PgSR7DBx~rfm7qqDtarF@7?`cltW3|8p|<gkoE)t;q#Qiw>o2a
z_vh<UG5A{d9}<lq&xV{H>nNXns@&7_`95e#2q<b%WEMZ&tvm@EQ<HGp>rPcS(Ulxb
zQt}=hca|t+X`l3?lC@3@@}y_YP6yrF`dCsy=3G8gB44R*`<=g8+uTesWrqot!)9!E
z5ETE^{Nhh{aIBPhO)e!ek0h!P<R=g>7<FcFt&Qgt@;p8ZvxbC8%QYLDUx-2#r?_mM
zs`fsDvlR^O-A}|RT;JVQ8+CSk`SR14O;uA%_<rLHc84FuW=GD{r^@7C2B;1@r<Bjv
zjcC$DYFm>wpOa!j7+nDWt2YLN4V_vpktt8O&-xHG85>JKN6?1Y+T)EKQLu$x33Rs$
zF(30aJ_N%1Y29aeb95%ZXNr?$*wupuq$KoG-!IsE(4-dIWEL}J>88kU&a>@N@f=PX
zRL&$WFs4MCq)%NnOM0|*sCE8oS+sxm8LK>PA(t?cNnkO^i#V>BpB=cF9-M#|@LJB*
zp<cdd%CV`nZH_-#w@95K+LY_Hw);ih!gK<7Tc*j#G|j98{W<|hROrPrQkHII)Fb~o
zC85t@Mg2b$0QoF_>^d-RL?}p#ZTcohC!7C&rs1==5ysa6OaC--$d|>Yq#rJo&&%;&
z6CD-QfJXX`oQx!<7Vqo2p|EsvQ*({a7Snm18d6-}i)q1h{jn#PZ<L<&<Pc~Q+4I$B
z&PQ73NnZ;OnGX>o$^ZVRqlSZSoV*eh6?o!IF1}9m?{O;e6)B|>Y;VaJwVy~Cet7Ts
z@O=`E;0>urU|R^3oUObj-WPTMT+`&k-t4rj`-vt8m`|wR=)yZXm0LwBCJBDQ<TGnf
z%DdIO3eeE9jIwUL=h^k{izUi3eCk}M81=i|E%C{%9+0R}wxW3CwZwC`Y;oPEt>cFk
z!)_?Z#r1`J!4^E)J@JxE!h(B%)uoleT^SnLO;7*M+^Oolikg*>@G6i0vU7Q9`C#fT
zW;S?G%2wfCKZ*8FsEMlCi?^NrJVHVb1v3kEPbHtV_{)T((>P9o(iXz)4V!zhhCek&
zNwe3nvsxr^{?L16W^PQc9!v<N_k=nc+W-C~92m3S=_YZbSPARIrtPu;+T-gxZ|L(c
zGI4AtgzXz`;^~&VGuW%DV*AqE#<fQZ^ZT0CEB5#GY45F#HIjwx5JFUs=EmOf8-+uj
z?dFJ5ny51hmz2`QMefewkG3y6!);xpYS~FCcRq%{+;Srb`~^c9=jNXc&*QyMOz)|w
zJ7>2hzmtEw?nX`h_IN4S-_%ra)ah9m01-lI=IvY|tCEr$j(aB0Se`HQjWneaLG)DD
z;(a5^tW|Mw335sn3be&zKRrF2QPT$k2_p?a5-7|JWp)d)H+M9wmZt_@W{S=eT08SF
z&^YJiX4div37t*r2%r9zVZ`$!3d+fueY<ws^E(#GErSs*u6Dn=re4VQnZ^<7TX#Vz
zhRs#@NbpR#%i)JRr$BzY^u;MCFR33Zg2!-~l3u?M|Lw9l=aq{5Ck6SNZtK)<LATV!
zHg`2|U7hJ)!e3OE?^ML@2~&fCeqcQ(CnpE8@_mXqz1-Yf+>l?sOGlP3g2K6?p>_<x
zX54qzzhg0VoS4T(SxdJ(aM_0Er3GIB_cKzW;n7NcAGmjp7o8J+K0c%SBxp_}!9R`#
zV~T*jJnf!AzwowJ>|gBw?YJL_J7)?l@3zHugIu#P7fU_rIv?RWP-MjszWv*cCDL}4
zIG3(5*Dpwo0RIY{`N)brLIPW|!4*nYe0q8+??~Vz2uiOgzKTV5|2k`njFcG^^0F5+
zP1tM*@3TOVn!Pr-oBjSbdA$vt8pGG2mz&^St!cOgG~&tAzUj_#&obvK1Oe?Rz+Wz&
zoQZFtv}EJUsvqgQfIBlmW7A>aOK_8)oSw21$9Y(<zeuBN%>3W`48`a7NrRdvkrSMU
zySux<HWl2i6tecYyxV0oN=+@uy_g`2vTmy~88B}DB(@V08pJ}~x|y=C0cC$`oIAer
z@xK1ImPT$<`E^&!B(4tgpw&q>ef^*?H}M|s#j&cBu2e~p=HHv=J=33Vu#i`N`Ldc}
zm5V~k=R2}931h=BOah-1Wq%scsq!(TVdylI3i|nL@>mp)@HmTX>s{=K<!I)itUO{c
z(+$F2W4z`@n{H8EypKLAvUr<BLm*y~Bz`p=a@U1yJ`zN?&HJRW*@W$>FyJZY1n^W~
zMIpoYdDT~gANx8vQHl2QH7Jv5kk9Jf7U*b&JgF_D#MIO>oF<a!RF=01*zadq&Xl?q
zf4DJu+m4hSsKllabBPsbTrB8s{pBMN;R&H6_r16o`V&1oG7|Pd0s;{fd}O3&{E;VJ
z8m|qNDQCbfGKwqqkSgZz>{ly)D4l+^v{ZJgwESo&AX<I##Rj5EkNXT&YPy;NVLy3{
zucq226p!#tK5%YAY7Gwc67Kv9p3WB<G{}F~v^a!(?b|%9_fy|**M$-Dy)VKbVu$az
zg-xvLo}Y%cdNTM0(@ChDUoO6w`DZP;K6~Fr!k`PYerP#7ysI~gQzep--PkgbZdCg>
z5U8dAd2jg1&=rXzHekov>P#EBJ8tEFCxq*YLEfcGyz}m;bvgi(F8~^{`GbX@9*cS5
z1HVip{p4XM?^v~d4(>8-a{JjU?ySlU?%_IIDU^^H5gNjX1e!AMDK$WcqGKKZlGNYd
z<jCjpS2Ap3iF~tnVu^EayIshvffy@u%#|PQHJ_ZbBZoo980(@{E&fm)Qt<YB+P#rX
zuKmAQ3iViI*Qev15$HtX&~uTXLi)D_46c86N6=|mGPv=hl&Drc1yIAgcHK{AUZZzU
zaK0ytO0k&RtpB0Oh&O)r#wnGik+;@01^<pIH7~C-=qWI^wyDW$FegRn0BFQ@Ds&Sv
zGKlc-q?R@h%^LJ$qN9`LEqb$x=;e#lweGKPzad6AZj6{&2_d#k7PV9D%;zhm<>VCN
zL4rrOx&1Go6eSpD6x|RnmI|(7kRyJ^N3Q-t{?e{LZCII!v+6f<^v9IliA%5Qtm|#o
zgRVL|gj2I7e(!s&z7kKIzSNShClK^EQX{2d9HBvs@`O7<2J8bI(qrrUo_@?8MApq^
zvXP=Q+xkIacl0e@9buJ_W+m(}o@H+{3-Q0APC!*|6&V?+lb;Y47Uo#37`hx_d@&#I
zdy&p&B92GQ2k<rntep{bxd1vfK9>9N3yA`m(7=3-c~DxeZlF#owdr8F{nAwMYcaaZ
zDUcbl45#jrzx2|93nwSoFw5g+hHM^|bK%dk(SuE51CfwhW8mX#5!O<>U(=+uCSyvG
zoukXsM>mglGrdN2rs4kn{-D!Ax)4giyMuZFg|!}hc+XoEQ?dpZ43E^=FmRnV=sIGq
zjGsMxik7xX;EFpfSAJ9ky_nheCyMy7q-DbJZIMn(Q<Ky2MCWa4mrenNTJhrViU&Dg
zQ)X^&c61FgqSE3Xb=(U|p&r*)Vaf7E!&#LQQ7nPf&1`Vs>mLr$zWBEo6STQq=yL|>
zJZX%ab-ow2>6eV0_V6z8?I*W(8pVnXQK{>^S<D7BbBx1M1=a}`<<CDe%hBpJ$|g!B
zvLu=(Z{T3QSWZW%fTg19tX~L>R>2)|L?B+q?aG4vnL64TzgTMqAPaaX$K)$x6pUPF
zW}q6t>N=Md8biQpX=!Q8nK_k`Tx=8ed06CUOjf0FJ_u4mzoxBL@_m`M&lEWp_x6^Q
z>>=y_Zmt(#BDQ{g=0;7;5hC7*N92W<aVqK3Ias%1o{^PnV@$Xd+k|Au3*{8LU{tT5
z;4D1=!E@_LJ47wTVK$s`H1xvZQOnc-l=%1}s)BFSaJW*_j;2iECWL7`21;PGEEOUg
zE425&$|cQ|e7&6B-x|+b0jo13XL@q7yB%go)78v<mn&T3!XeXa#UMMeStRBKGV=Gq
zI;lidsYzP17#cmAUCCBVV>XbL3*I(5NOG=h_BWvUA&bj&+o^SHE0)sYo-)hvJ+<IB
zWY&>ZLKPbloYpjL`&#H$zU+f|d|O-Dz5k0YRk4oI`h-jJycz(sWa?NuWilIXSs(i2
z*Xn!0C@Cp^f^A;CjD$swj*Q&Ao{mGOjGlQ%etw8mJ{$OQ!EdnHrLOZBA=0d~*LTA9
zX~CB1WpBX*Vp%@MC9E}rZO=s(nbtQ<#;Q>LqC%&O^Q5GtEJR<JV8b$hn}s9Ryugs5
z;qbgvP<USZ-CDyRZ0wORNkOWowzd|F{ZSWAr_VGt2zZ7McULmf(jBKmO!}3LL!-Vv
zK6;fi`km~TtG_})I40)fM?xT9Uc#^Nlrv3AOpKK_1B57rzyXE;ivz2?fX0phPQ~0t
zsC@{aa{^`A(tH9<RIyEK?4d5V0(jUQ>g<y&3YiN(gsx3VeLRBq@lhL(4LBOoilND>
zwNEhz^U!>bW?0(rb5;XlFjgHJW$Tx0oG_W?j&Mq<)3BS|K~CbVQmy>lT>i`{)#QkR
z@3=)J*i6wRI+gr7y70vXz2i#OD{UDO;`-62^cU1K3VAGlRtb;;1OMIiEtl(Pv9)t5
z>DVfEc6|FFu7cp{+tc<p-i`mbn?RzakHL8@MS>xs$^mw7g7k4>Lbcfw&t6;A_21QG
z>T#RaWjY>{_RkT$)lGx|@&~)e*7d%)LhSEqE)JTDnntT{g>CAq+p}wgm=*|0O-<(7
zE%y!U@gUY=R4LJ(j8!?&c3iH%N8-1XY@nt=@+~rXfsG|}&IUnNFRQ;)W@+*`@!s~E
zL}Up1xMnS)vwrg1i=@!5&^@7?u5YOXxEJYry!6{3=UdG&fn)#6IjaV;Nw_(X<_k)#
z4(@+~+a5E_6!2PZ^U<=Sqo!WpnJfeeKCk_#&Zoz_n5q>>sFnTJQ!1;`8%A7AA^|br
z2WHXho8PcWNt}VL8UMrBWe#x9^epCHtuSt4##Yv4+pur`vHzmlwd4yY{5SyxN{R1c
zu%hDmwb^F!Df-%}9?a#13_?;i^zuq-V6eY1V4B$(8V3KO)~%jhu38Nw#-RR399>6%
z>+bJBJK~Q6p?sX8K?EsEqf5agqpTx9c$-vt0x($VgE=$y4Bw}^nc?Am$H*6(-=gwK
z7`+FE3$Z9V4tDr`{7LYPV@A!VjUl4%yvL7{>oCy^Tu7TP9()v1{of#Wt;dTjSLpHG
zNt8|7)};R$MjOhq{~}o`q=ezOuXMc{Bnqvu-HcHY%J{~UT{Gt@wObe3d9bQOK>SOD
zkmdUCcD8+Y+*g;k?>`9?S-McSe<D|X4EPu&Qq4HhNwhS&KA)2Da0h|-y$S@Y8(`|`
zHi$fZ<M+R21MtxrX7|1OGpEi}gRR$cK%<`4Yq~pIou^z<J|ER+KX<yA$oir9X@O;{
zvT5js&@)Qp@fyz18>Ox7i%XKza=O|tcBfQx80Qn!28q7yYNM=LQ-trw$gNixigI=f
zWl37LXKfh}2pQpXkAOZ)xjHlZB=aQZtei28oR(iTlaLw%T#4$}X;5qXry$3x<4#_l
z&TKPX1aJDIW)&qhTT~nRq0nM2<T8eh(K1clEr=bxzRW9@=bI)0<wtc%B<1lSZsW$m
z4r#CpJXkIQ{IAVyWQaKIakE6Bnn6sz{lmQ}LjMs?|IcBPz^8_B9iwFywJY+td>j5Z
zZ>s9-mau!tQ>8|SzWtt1lA@ub^YVT~u%1V+GJOAZ8w)t+OD;db9BuKx<x>nqfEbTg
zvBLTH#}dIoco4Q3vL0fWT>6f99k^r8JB0&Yh{7Q(=%?z__VMuGapRE&IV{YLe5vQ}
z<*YB7JY!9{TUJm}0VR97z<dJqN?=p)u|StI@ZTgNF{PxWQVdVeI^?mInDy#5gfc?@
zfa~<wiD-BRUCpfn2yrt5FRmc^zu7#U*=i14tm{;lc&X6?PruL?)N4U#9}0};yhvow
zS$J0e2D$^Utz5fO{)m~p#l`sGvjv1rp$?rAE}aGiV$Y6WbI@bl&E8kzxIaVY3kTL!
z!Ih2QC(ZfY%yrDB_Q^CTX7#Z?dSi@?Eqdng5o)##xFz4y9@I`?tm);WR8N{Cp|-W#
zn<TszNuROdhQDE}t0jLDw~bcUNVevFScxAt*#LY`4Z3am9jm_BH$pGf|1QU50=S9e
zP#uM6CHcbFljq$G_epKGszO{8QC#2Va2lZZFIOgMikYj`Gg5~yr_}Spiwz79a}c82
zHnbcV<tSTCtiEDf_n6pQ4Wh`tzJ~&a%#zE90&8d6TE@wjNm$6D>{NJUR*?X;IosuT
zy;BFwo6{n%h#oE;LUFymG>bOJ&Nx7&UV1v5n`olQ=4=0=QvGc37p;%q?_f~~3mzne
zWAosSW%~+HRcvf03E<-50-)^Y&!4A8Q*nL8u(w7s&K`QNUx>lDZH=6uAb&R@f=HtG
zwtg?!+CVqDM~Dr;hphe{EDIENXz7Clf2@_dwKADp3<&{5FOX9tBBWmo{53?FWljWP
z#vks3mgHO5L4p?-JyB<{Non9<Aw)HDi7-DMo9MPSf|apgt98ogQ-~qn%&m>sD_NNZ
zyr@yqg;5q_`S7s;FRR@F7kUz`*t}t!xKmKh$cgH5ox}Qh(IOw3SC92RXZ{}in6D@r
zzNwe%JZ_6T%}q1mw*?r_O>wNSusj55&IjmJC)3E`_6~h=Pnu?0LS;CAlUucYco-1>
z8XGinz-Q4nZ_v;4JNf~F0Ch<AD;*EJDdol+*1R5nwYYffW-2sG2#=lLCH*6+zxfv0
zIV6zn^&?kL>=i}kW%2GJiJabY#}`#dA4WVmIlnu7<OCVDIf#e?gmi=1;&-imusG4B
zR<a{PvY64>@nMvWI<hynv3ggE>)A=TtyS+u%bC+gia=cjrW$;hE~G&}psCDWTF$kf
zU`}YD4yl_mOcPg~vWMzxK0SnOf)iqmZHhgm`mu-}BV1Q<mtGPM*z=f4-t9CaN=JIg
z&kPQZ>FBo-5!N>HFr^z;mjGhkxSkif#j-lzkix1$&Sot7<5jkx&y+d4$Kia#*Dh_(
z3{wzDIU5%tiR)7<lnxIMUpi!~Q#|Po#=RwGImdGW&Li+Oz(^d__LDQ+6u&!d9Au&Q
zwp#1&&g53oDzE<5kcKWiw=A5YtmZuY)PJW=6$bz7RWv0(7vbH_^^OfOgfL9L5vJK?
zS3Nq2(pmobk?Fr|=ekIRomhE%x$eySay%FBppn~Q{`&fQ$5BNqAA1Ul3#kBfMyqvX
zWOR5r+a}kW5T9e1;E;m0Y1`JNF|`4G-9Ni=de5j(h(Ke|5~cO;QZ+4a>3Cqca|1(v
z4x^<81~dhJr$h#nM>I^#j;pCTuPwkU29ZFm1gX(QAb(K&k@H~MVQBcmu;ci6{>&k1
zV%PRHL>5NuKKPYSt}tIgI-T52;&;noiq^!WSKndMSzv6n03(L<wDRqasdl-?RSG<P
z2QDQ?AyY{#Di-UT3J_mM)|HgQKg;B@d-(zZ;l?SBe5pzg`nTVfK3WPH;_Bx1bT&Qv
z@0=GBY$jucaDA~0MS|d>hSNtEiHHeVw=|cJQ_>$1r&X;(01VZsR+kvWn~yYdAG)Wi
z7d*l~$~yDk*)O#@FKTR-6r0>g;PjFcan>xk;e7EyYxcZ&GW7Wow>LAbNUUs=A_-w*
z#8q|o6B1~x97+>l0aQ+K%lpcKF51w&Y4T;bxJ&$sGIH0ZZ2a<}@A+m+VNh~fnpM4i
z_gOt2k|%Wu4z{Z_x0~nDo0}WBpTH{@#^Bu0b-*~kbunnH7WDl4X@(hJVf{tnYUtAk
zaTqj_N0Z+lUr9tH3o-QZa&SOrXRjQguV@w1Ykr8Cu@SDukG}3KZ1ImQ1C`tWaYuTN
z7QB(QZ<MEZsdI$Ncq9`U7PfP7n<$__cL^K?L+2wwLP4a=IIu3Xi@R(mr=U^^D8lSc
zwh~PhRww!U`~U2hvHqiq9;11*IJDT@N81CE)V+Ns;w6rglau?6Pbcz`jWSd0f&wY^
zgjN7@iJUe0cq%E^7l-Z})2n_-sD>+YyHj|5?QLYV;<fApHiRVyZxzzsko^c~?KLI(
zw;wSz4Q+shgPW~1u;L{4;D4VI!>aj(LI{|}k}2q@rJJTE?hp6eQWcPZSun>64fvg;
zDV=$GETK51MY~I!<bH^U)m^KcQ`}GY@$ti%<rwo_>FVmLf2-MHjTk?|U6)bo0%$;S
znSR{))S=ha`vlf@VY=hPrj^I7ET<uI3F!G3vCwx<6+=MEM@3b@e;L~@;ry@N^XT<Q
z{%G9WTTx8Z+Ia7fu~CyJ;Q_;TaiW+Oh<lVRjUVAx>btK$jrr&$PfwWIcTNmep8j<u
zzD}gU$l`8a3_JpW%v8w;7*RVBiv15PT(|znA<hiyJNsCJvns^wiC%v)?4lIAUen$n
zS}+&HN*(mNL4t|2FVDNk5%rEgVB}5`A7)-{vFYb%CDzs=M{obDMfLY1S8#VfA964#
z_O$ln6oO(e-8j;L+SR5x`?a+-^NLW)MSO<y3ydwzu~dg_mIcbloF5sI7GK}$5z#Z@
zL@D$0khs<#^BQV6GT~lCi|_7CR88#0BL90Dn~m2$A=54st4z>hRJEhYA@{pue%JB0
zuTi93>s?~dKd+Sl8GKwC<SR+kfc9{{lzJO9>-i8Qh+M&$YiWVFIdg%cv^ldlH`D8X
z18b0F7z?Bh3nIP)MhM<81cJMfVnYsDB_lR=HiGbGCHZ+g%xlMACbbRUjHXh4oR>m<
z=>AX)qUM87MocOX9B8Vk2|wW=yizm@5J+NJHjuK?hd`sUVG}KC^cY$(=Io8@)%&QD
z^Awg0J1Q&lSii0W=XD@t_93zk^~<n(D;O^S8NQn+x+atPBrPSiWcJ<&gud~7)?Ywy
z!};(a5TulrZ+|P(O57WWNqlx<L+ZWUWr``W<X#Sq7oE8&Xpiq9{VPSqqs<-H{fzyB
zEbaoa;_9=9uU<(%F^i^mLWS#g@Jo2L7hflEw0N9O65)_A-|=LbnVFH<at*L+>Hf@!
zb5-UZ8kR!v`E<^S5l#=k%s55@QGU9cVu|jujYCmOQ`b0ypDb}n@@)4t$w8Xgt5J@3
z@_Ir;MWqGWij>?$v--{-=U|VykVc|tVv;{<Nqt6*QoyXf6ZNf-e$QyOOg;JFkQynH
zx0@=4t}O_3cL;?hwMHQ7!J9~Q><<14rB8a15&L}(-jig2y)In{VhS*b5W5%mH+LeH
zWxo@ZBQGMphxJrbiIX%sIeJ7V;Rh@S_72r0JbR2?^IEx#RW))O0E;fw_Yt{(^E8n1
z+ay3p181}G-z>`|VjRdM4yWnIkVa9lleksA7dYK4DADIc2v&?qxjenpsu{3_iT&kR
zc*s34cq9+E;{o|%l95tL^{?ctb_~w0LD%VXgMJ0n;Kk;jT945hF-XnE#$1ko7bYoh
zjXfI(g7wE!U}0g=?T=t%k$|hV{W`l37@9J%E^~U9=_WNLP21EFo3t|hA>tGbw)0~F
z_1ab1;o%>csmsIJv*_zlGGd9+*3rguqE0WrzUY0+nMk9l36Vm8F!U-hz%1{(*cC+%
z#(#HtC)~YRY|=oiNvu}{n<BrZ$?(I6iGVd41!XZi)yMO?l#Wi3naB9KjJWz#TBlc1
zt?g>+Er7946)6B-?D@q-z-OEnk_ceA+uJCxe4_@J6{;iY>Ux^^T0xN(3II7Ro`?EJ
zZvlLr4|wZ^(s79AM@kuS$%7x6EzP+#EOF{G6M9X@WM`qcS{X5oZ2cy$BYzX?<$q2g
zbzbbXn`I>zpf#N%n)+83H6ALD!HDT*TcMeHZ_OP27cidT;NZA2Ppd_B0gBoQjbH~4
zHU7=};iW_dK0cq@kNNGyVYy>N<d0hmcq=H0%(l|_T;=TQJ97Eeo8m@dSAXQOAvALH
z9S&Jg-QPULF`tU*%fTjp$OH*IV`Kv+3Z1@O*d|gwe!E)m+OE1q_=LCj4f_?{3AT-r
zmj~0OQVObN7AdH_4I?gJ)aN8^OvjJ1WJC}lXz@9Y_kWW?@2=&^-=GTPq0QKFcb~<C
z)a!KWIoL!9$8Zkv-b6$#ul7JD83}{@oxcGry#ZQfS(!_BggXW%e_E2j<>6dFtA`XO
z5-%_#%hr;Ztz06kfeSN^`M)5l#$Igv?YR~|O^{+?ASakT0=&W2q2!_~Z+>F2t5Z^M
zeyX>ox1F9a5Wy0hW*^G-PLL=7Uvh)3|3hL@$~)`<-y&EDeb}EYIjz7UfHwv$HbAEt
zRAl5*SLz;#kn<h6_zvOo9j1`&o1UJYXTSY@+f)>pfad}7#<@m00lCP<Xk$(<+|Iz7
zXdB#|!+?p=!^0JI@<7r$#5hRaca~h}^HRhCcaH$}xXvpYI`LTPm=t+c*E;d26!}%&
ziVbx)S1&R_^Q6sx_$;PKSctD9Q%K99F`V!3?*R?+CWtz+m)q8vmdWbpIT7e&!`$?s
zQKAfXH&b^nQl)&)vEX=E&Wntwn1|JIdv$i>+_9c(TXs%Ms<q3>Fe8UWM1aFXiCpDS
zpZQ$A@?5zhia`UEa6pfeHM+Id9htVEBn*(&_g4L%oS=V2<@I+;>57$eZpT;5i4M5U
zo6e-dBfSojD`s}ISuo@gcu5CLkRstoJ)L>6kAHd!{3X+hPY3&l5iW!8b)}|+=rtVG
zJCn`4s*GZTCPgn@7%Mh{*M)VPRbK?DRcqS6oCMz5(Iwhh08Um)iUoJpajjzxC9+Pn
zAqENxLpz75N1G{8gb88aTB+nEP*gR{H<<jEa?}U+;Suwz%+spSt*aJyEdv8XBo@V|
z*x0rO4+GWO9;!$(6xhFHs`n1TUX-~qg7NUnsd-)x6CW;N9wEl~@T;2o2r-NrOMuv=
zpFeF{^b%KogwY(lpjN07A6}j>3dY5&$R@F=(zA7k$cO9U*P|NiRcr_wec##ZyAWoa
z#(4VT>a{d9G6MX0SF4f-b;HN>rtO~IUc40*-*d<C*hw78*Pw+B@F5VfmNB<9(=}j{
z>-l)YPqIX+4SMQeh({F`JjbtHq|D7}AkKs_+)RxN!P^{8i@I04Ayu!$=J4&j`;}R!
zyPN_vk(-1$ey;t*6bNX=s!D#%Mx?;M@mfUg4)z*W)DiKWu#sanhm;)OR}sj@!r^IC
z%rp#4U7`eY3=uw4%1_3ZS4Zz>sM_))mAg&_I-UFv5<EXz)QZC(f2K4JhlN}oE!~aA
zKAG8koBlXi-TCM{R}W+;yU>y>m7G-)?pzKlWzeRCoU%cZPc}aH`2koye3=E1$Grqb
zxcK;>*-j(aboY8Kzuhmcsrk1tm}FvN0vZJ}G$mZE1R3!*^GH^=7o6y_Ipx9*%VGy_
zCuq<=%JQ}ZtxzNH<1#~N@Npw#q(A(m788ZVs`4{3);6it{f>(ND%qrwF(edYSy(q8
zk_y)FY*dZk2{?2C98mpRUxS>n-m*DB*@h#$ojF3TfURd%{e9xLEn?J02r)G^Emq7>
z|5cwv9`N?rs)4k>BTXYJlnGRC7;0+%ZgencY4a%pC<l`F>C<l~Rv3okbD~*=CUN9m
zBE!DC#@g;6Ny$5`R-;fD&%U1#AV!GsCeK!;;ZOJ(N~w#AmNcG&F}?4troQii_^G=9
zXUTBpJAYg4mFl=&y@K%5Q5^2AsrgO><z8h?O^x6E1sJMqP8;>s;i$ATwA{QWwOtXD
zOvDt_a!n3`z(x>HIw&YWsjWrLsj`2tgm4DbKk~ZL2nx1=j$*GV9Emd<Gn58)C1FME
zwnnxCFobO#H^YZj6a(k_p~2b8&Zx49FPxC%hU8RO7295J+0^jmaBIIfIE*t#$)@5(
z(A)~G^y;3tX3}BQ-H9^&Dg!3yR~aLomj+}6ictlCu~#~~fBg2T&Q_j|TSO#icd7^k
ziU1H(veol&Na#792ZZA2SXe<o#&>i*NWh2i`6~ksT?f(M?P@ipxsI#Om4}In-z{=i
zqQ2+7vS74xac^4^{zTpnJqt&xJ8p-}*`ND$ft(edKQ6<0@4y78z4Q<@HeLnn>vwaz
zDU!9(N5`NJRqVGK@SVVm+yB=K;9zxtf!**tSu{yAn#fnBmsE4V5d=Gn0`2l)N6M>U
z-cRMf_bTwzwqKU3mkqTiKZ$9|n7)>KDK)tn1qcj@G--5zsqjtnTgz!Qo>anmbF@fU
zLFrTsok&~Ff<c4rGyt#L3PeZmEwy^d%gZ~+sJ-gEKOL^ntN-uQ+3FQwFYN@PQRl!y
z4!UeSe@YD6KY)&P&&L`x+WTBg>wp9JqtKV0FU!qib?DE(fB#w?KCMT9ejA|A<nw{x
zQL(s6a5e}<!~kuF8px!=IUN@XTQ@&i|3E}1RMkP`AqGYS>=@=h6)3wx-}(Z;W0%%q
zrwnT*$5eE4Fg9z|JqFWKq-MJR>!ny2>}>-4t?kPb{O(WpznOn=(5P--RwLLew<fC7
zn25$&b(b1fw!YS_+bp)2yZDu#D-RDzWT9gr@Rzv%{pJy74zI>|M3%|;p7`3%rb&g_
zC_Phu;F_oTg4kofr-k8m$!n$Zknl*X<>n4a0h3^pe9>F;MAp4@Xvy$H(IAn5$lX!w
zdv;5U49&O#*l9<Sifk9}pcW%)3dEX|>sgAbb`#}mGX=}Ss?oN^e{rE`%Ev0IJm{;~
z+!Xhx&<9KV!J!eWkgUkPMaEy^U3+tAR|Y*BBD2{O(q8E7M|#c1c8w<^POS$lrd^9n
z+SyIVNqpswvc;XV4udfV_)=t^D?O$QKP-+}Vh9ecf|k;kn+@E?+m?>9bgwp-+j)0S
ztCBGCG6ddCwV6E@P~YSr-8G*Cm1pPmd-#|h)QC1ceppEGwcp>m$U;MGJeo@m4WSko
z2`C<moqqSYcKAMx0?qEwuo{Ws?qG(6g!{GdsbPF0x$M|*cw=`i5y@r*g$(6ua^3~+
z#e0{^bmq~6lx>H{3?$9&71EU#pI9=RSbvNM5x1&MH1BM{7_8_Rk?d^7-nt&GC5z=O
z@zu_%Y<mT|^i5!IC#b9lBnidx{q2bIFgv>F>ws&tw6v3uuoi8ZFy7V8MHfE0>x_2z
zF~@_&@^t-@)%Sw<$-mn8N|B`!!*=e+Mbsfk!EKvIeR$u1T>4q{z1SzRP(P2)^bJPP
zbeXY5`p2~7v>ey1w0LOEm)(EoF`?azZ?jr%)z%6vxqsI}8Y`GfI#7m$L&bgRVVs?j
z0|ZqNy}rrF&U&0`&ft7__c^76bEigtF4czX@k0AkMGVp<1*cT+&FTU3l~M(X+x{}$
zzqTYbkOxGkK56rNvRC9G`q%Ya$0*2d2B*cL95ZW)Q+I}glWN3B<K6h%aRp{L$VU^R
z+-*Ca?{%Nw&{I*V2Hi;!|2A7p)>%ovxI~MKuU^bJG(?fkE{UJXXuI0K>HM-*XrLyD
z)tg~Kbl}|VUHuuUcgJ6FuIp@H<o4XYwdD4i%1fCU0>NiN@S#QVE|+iz6PUH>&AIG$
zel5``p(wWjeY@WU_ptFpW&`>fHa&3!Qu?Xl(Um6y+O80N5)qtn%wX|_b1TK#BZg0S
zlS74s=v_FM`xaiyHS8o(cpP14-qZSMqsDSTZ4!a$x$m5LB8+)G(f-BJ<)O0d<Jr23
zr!)0(=7lcd_Dodo<wg>L>+dxoX+^>X_TAOj2g^|&29ED{#OB%>{vJe;4*Kp&TCUIJ
zJ3hQtQeNhLI&0@keCK|t8MsbwbTwxGmjAm)_xbt>H^W_zkJkRyn-24pUgzacD#DA|
zjK+3<%YZEX*HMb`$xz=)PXb537m7M8gzZIT`%C4o$X%m2Q5}`ki{<2NT@U3d4cYtw
zvKm;Zi-NVto1}UTHw6d^H&L@?KShfK=w{-6+zI+JZ2xKTh^$W;MG~17=?d7cg+Pq=
z8<zH)atBia7O=>mt+jWEF)Va}%aZZFdoS=HjkJyi?HBCFD-I*+)*@c_G%0_1?vq#S
z{JkjP#rN_rvL2NQ1ohE<{Z%#E1zr<o#XmJ0Lofa&z!&e9^5og<^Ik1vF7f2Mtk6X^
zoI022inY{QtkreKfdK#M?#bWp%hw-UY4IThxeLv<1_fF~i#FHkxfp!(t_wXaZ@pO0
zL7`X=oNIZt4fr8!KS5jt$Op#%-<RJCGtlzRg|tI$4g{2fna|=Xzh(Zo+D=ae=smA8
z!C9;yL<N(tle;?~ZoCk2=DapmwGBVF1<WC=Ty{;i^JGX1MPMARjzfYk=8xhQ#z-(6
z5Ydlus$d}Y%71>i^jiowE4+K>u6n|F_s4Z}-086;miq2#yH*>K{YAXDcLKT4nszpU
z2=MBcQ6jYyTK?`DCN*wE^E3tCFHWS*Y%=VS4Cn2g{M`~9$;+Pgnz*C$ejRks{vp02
zC>m*DaxVUpt~(0TX;4t@tQ(0d8$`CuZuxNocl|X*uF%d+gO54>+Y@YRA>XbXUeACI
zS<=hHkJe<OJ%6AoLG_zUM7pPef6wx^+||0=Mq}njp7;!f8%B-~*JJu_UkP?$6SP||
z)Wd04p#~kF_fum~_P{>@`YO61L^Dzcf1Er&?QYH!y)&Ek1C~!M4i?ktJl!I4<F@(2
z;Z(x=R%=M`rz9E9qBUlmNh*2J%*b}R#l^BKWkb=yJedF%va!S|-%iW7=KIYHN>2s}
zE*SmeH4j;}{}S4jRSE;CWtNC=_!vn7_Ir!SQxV~Jgavs)&pTo_Lz30w-`;+d*wscq
z_TV9_IIMJc5z=9Zx>dCJrjVZqQ96|{K>YJ~GwbvR)DA+8+S{%{ce`ME^%WFY2yj%L
zS?$vXHOf8|=JJTl{yLo{2$BxC<79yuGvqk)iKhePVAi=Ao8^4%P+#Gu)!h<9CgLSF
zn?r&d)(Y}JY(s86i8_5_$|MycC!_3ZqX8b4!)_s^FU%`Rx|aSiL~=XwrIvl%6R}i9
z!?Jn$ooe*`>CnRV$|fdg)`YqMjtFA^qs4E%(M>v@8%|o7I3EV`&V9E!PJnd*)>Jz7
z;AKz9#rfWJ@gJ1-dxn~9{f0hLfqWgv&8@s(1@NC1v)rD~*^_Pf6jt0w{pursjJqrH
zmGn&m#bU!NB2$`=NWjwqll_8pck=(&TYADO!geE@$vS`n1qeh=N?Ec>+&JX_0Ww{)
A7ytkO

literal 32193
zcmc$`byQW~yDq+HBsYkFfP_d$s~`=6go>1;fEz)&J2ojD0u~_ME#2LMbV|8N>4r_$
zo%p%uJHF?4&hL(K{<z~ZWGh?tT64`c=leeI^E~egR#K29#G}H4Ac*k!GpSb)gm!%W
zjf)Q6x&Ok175sI>=Gkj|2*M}5{ze-@dY(fNJ@i~kLe(XH1L^EeHady3v;1;mB9<t{
z?;~^rP12e{5)=O!<|-~6O9D;0eg~1S15*tljl7#o&pLif`TiLCGYU!N^cXo(1S>;a
zMSXA2S;t$emy+XtG53UB8<NgkXL>g_CQ6Cpbbma0v<y=wLu2s!h)^y^qWQrjyIPWE
zaTy|8|LygiSuMt`(y2|R2-8?*zhieX*BPHwR%J3naf;rO@BPbP>R-oZx+X?f%7>(~
zhP1Tk>#6BzhNL<OCDb8tr5t%z5)_PozDrQ(;CZ>qSu$W&P8%T$!*q=sT|aQDeREh?
z)%;c|6`f4_{^0kq^~id3p@f1&il_yFr2Z@^y*osK8(vXgSvfjvX*$$$z?mMbpaojN
zB%d9S38pT)dJd*aJJpH`2_ahiZVbu9^&bAxwicubud1qg{)IX6L^;@s-VcWEAHFg?
zJp4{y->5&)@^PHKon3lZ9KW@Kf&#qCMo&Vb*&m;x>pWYEA0~<WUQ!k{F)<<Pc8KZS
z*4DPMv5}q43RO96YDLhBLiA!{V(jdBJrTWe{5ak>Srh~k6aw_LDlA?P#X3BUGJNeS
z{Ct?VyQCyy5CdxEA!FLtRSHS{n(Zywi|)N~i{9_8-VKHl4G5wPVDRgl$|Sw@@$_|u
zQ6#Gd{i}N`U~KS090<%<=|p;1zE=6^Vze#+c58#j*(qvOWl7dFiM!!=ekenZ`HSi5
zPzF=JGFe93k#;cmT?i(LLey`J@hFioVP)4@l1dU_`)>5I+1lcgt9R)$u_X_$qNf(U
zB2vvMdO`CO>#nrV^jjVQ8Y0L>D1qsUEhi_3`mx3FT3#V3Dd`89ca5GGFRoZOA9-E6
zufCtK;4+o3S^HsD-Zk~yzo6y@F>$3P#|s^rSErv<UYnF_MHxo2ekavH`uY2h?zgZx
zjLK_jej7DYo6n`5z-NEs_^w#<-0Oz?Fjbh$v=!!E>C0MXE-o%cMhxiT!-w<p^XXwR
zF)??99KJoaj73^SL_~1!r;#F`;Hq))^E+Rh9<V6HXK#kn8{Q{HJ%9e3f*bx)PA)Vk
zC@3VPtEWc?%>MpTFTd6As<+KP+1c4ED=QDh*#-&yl$b8h_xru3#Ro2TV;%cnz9}~w
z-7G8BZx2cGx;!yj5PuxES${O`wh+c&Qc{w8!IjFoHBoBZ8bHXD*4*ipZl$0?Gw_sO
zT3Y(R)lZ^2B@PVrRA)Sw2<cKk4ORdJ1qCq`Z?=Rl_M)@P$zpWYdtojPOo&#qakccR
zY}a<FVRuw{sj=zpns)CI4PB4&bHOQJ9O8oWMlUgB`NI2d4g6YEyr4ZZ<jqQTd(x`(
zW&|~4bh$n;hp0$t-r3z%QGv%GdG29CO`GT>PiBKV77h?zzw2W9`uDN0j=)$YOmHEn
z-*=3?9)Hm(N$f&5q1*NN#n#$87P0XPMh=diq`L#Wcc`e=jN4Vehfm)Qh2_R~Ed1>1
zGMg+j4f-0l_%cz%HF=CSZA`q_;-{OqmF0MmUd){`t^8gy3@C}-<((D;?IN0&nP`b^
zYo!YVorU@fJg+()JiSjZ%&k*%K2Z}j*c;S=flkatG3PqJU(wlhih6%05ZmxbsB6vk
z_RO1SI$Sv(3%b^2*^VDjZ{ah-Z~G7T-Yh~g(wI-Bc%Y{CQk!C*h~cPr@87-i^`#6s
zKVFEqI_?n9RWCdx^rcN{vOn2dG$kfjn@ArF7TNbLUHa{Hb&-+8M5}li$!^&*Mk}|9
z1%-VG2&Lwia@>3&Lc3EH+<k+Vdlvx@wL}P}3Czsh{aW=PCyjZte9s!SEsS_Ze9C8y
zt-;+gT5|ih&&5mzA9EP?<%^Rc3wYss!y8XNb;}3(t@u8LAOTJ#se50u2WiAT>j;_F
zUXw!=21)Bz5rJq=(oK%nYXtbEJsMULScEEZpuCP&Tb+Y7b!)u=M%87oLO8W6$Zp*t
zq!kgkD-vI&)3%w4nOHG(Kk$xD!3JmC7$NOIXEKG<?#|^KgdA<vo3>5n>q&k>76uh&
zp+0M^4+@Zm?9E0eb=Bcx(1%Sy(7A`yvk?<-J?SV>uT3ohX;w*vrNv75hNpC-cT1>}
zO+PJo8%e3ZJ91Iaf1Sq3vph-jH0w$?F#L{mJ`se{(*pq)!4Y<~LYX;%JnD-(iBJ1|
z+;6CIyr-e-%8;`T)?);3SpB2Lt%t%e$i9Dif!`h3X>x3>*~;gO)2*_C!01@9{XpBX
za;2de(XrrEE6+T<S+;7d7u+o!Ph=x>Kl)}oQP*PK?)ajJ<FSnj9svg<M49JS83?^S
za?yb&`5MOksDKpM&3Kz^?`C1CGn(<H%3#F*qNKGtLb!&I3+8v0G!^OREEtBimslZd
zQd3h0(_|~Hkf%o5YMPpqWMrzREhl?3nu*Ko)g_)Y4IWb!mgb|mqu(C9vvTs4laP4d
zy=eBks`#qXd5*Fk)0ft_fflQN!h(p9u>Lh|zINr-@}itCFO_9w;&$D>LizUK@F4Gk
zt?*vK%`6NWqOMga%hygQXjp=VXx<KihtF46n42G}#c8AUe%MD{dt>+9iN5~z$JLj5
zpNgMmB|1;vHxp29eXqb`Ch%$DNoz2{0!Vxc#*53G8PsH|?sr77MP~;Hb6NKG_mkZB
zerDI1Q*i$B;?y3@(<po>Jid9j$sT1%%ZCX~iyfbFH=f&V*B?bq1TLQ#n&R6lB!!3L
zCJ8Jx!=f^1oNUJLXWfi<x@%a|s^BgL4b0R%nrx&Ef|7?rg%YwygM~y_9DLN5Gh!5{
zy%eb;E`=C}cbayDaR|uDwG{F=tJeasf~uXB5C}w`dOqpRoAvyzj0p;XUr=|Ol4TFt
zLT<6cEvW|6+)`WDsml$XuO_otQB)zf;^V(A$)iD1Rl=DuHe6q(7?ijus^X<ilZI~)
zoxP+=w#0|L?-&Nwbr<2Ma*0TetV-2OjI3S1tV<_s`Sy!l1t#7YHbZ+m&L1Z(=u&lV
zj+|Hy%tvs4TkvZOm%||@VW-+P8Oa_TNO_v4OQM?ec7Tyj!0m7&RXWt-@l=)V!s)>}
zjsni2gU_%;HFDNYiObo#u=I_u4Kz^N6F2zj)G;r};c#g4VaQ#G5-o9a8VmA%lGTZ=
z6i?i^sIeDL!SO5GX%ovA4TY_ob<+VcKutq-ur`t-cY1{7J=%CN=+g4%JqAvctDf?o
z0^O_gcQ!hq>rOarEUM$jZNWD?KHLWPw6(R>Fq<g%h-XEalbws}5;d%7`FJW@IZdm~
z#CDxxN2GWvsP~27Mw=0nlkm9ZuPY1q1n=jJyBm#{`lpt3UxoKk-Z=(!n55Kxtq10Y
zPO#hRf2#B&I_?KCgi)7DG$)VV(Bd?4M9+%ER%@84$#+@z%5++FIbJKz;L?0xq!S~9
zXJ?OW2`j}_mdHF$P2EF!%?Gn2x@}bx^II%w|LBi-v}}|#WL2r1|Es8|C@#T8#Bpt8
z3t6jOZsrKqRF~M%y;rvh5b58y8?QXa3f_RNkSJiQf6|>KtsKv5Iwaoufdql<PLt)D
zK2Q1)MvVcYesHhk#$)~anHG=r%a28@EG5K>$!^x>yN4fzo81-Vv5lqffAzJ~6a*Is
zq4&cfT5sOB+8(ZrCI@`toMJ-ncy7#B&X7VR0TbjL;DvY~W{5Vq#lFFpE}`i67G<56
zmxBZ9cMSK}8HR@shiH)Vt?#m^Rvg(QXmwA12rnKDnw57)apu*Tgi>%PR*K*8@!dYE
z(a6_Q^VBLpp-|1%Tz+$DZ_oO!gzXG|e7GIS?;?zckH6_u%RAxXn#x+BRc^LDUEdqa
z=XQCr%r3I)wlmw@9!i1lGf-(g!;<7)f<Ov5tPFf+l7r|41y6Fa5_92<2jp-^UkxOy
z%V=-qjOW?%+ELQxpy9HAeElb7>$=^5GU7Rpx0N^J%4g^Obv5!Shco#gor;N=ogi$)
zo#NNQM@GvO)-Xq*!MCJ@*n=i}W>wt33Ouskn1^Q$6J2_aHNu~*45^Knb=UpriFtH(
zxETttj_4EJnxf^t1nn}DPork_4=z<WafoQX3EF3lt>57@8+p_(q7p`K@mPDKb>Zz6
zI^-QTlW#h6_#TJWyHWA|4uCj?cLXS@WYD2Hl>2m){H@24pFZi8YY97VxtHRT9|5?}
zE?RHvfGW~!rOfncZQwH>t1rHSzWHo1&;PEr@2)I$-CxqU5+QsV*c`KQcDS7&>|9Xx
zCTsY4XziZJjC~8H%cbG0z_A0ei-Fi-bF#ce(E|WdX8G7>UaQ^7(#88LgF9V$O{R18
zhBILvMb6vPw>(ozxZCoqAPB_Z%g6ZZcGWsu8mqNCaz2U2b^D2KRx*A++f1Ri%l>lr
zJ<dq;BFDeCEq0vt`2`_--{N)EMCDbIN#n$D(DqYD+g)84#hUO&3W~cP@64U-FAFF1
z1=4vQ&jr$P&5X~r2CC@B;rj+Q(a9QDSn*j-@|wCm*Ta$ss<WmSxq66+)7pY{wp0(F
zGhArAJY6q%QwwmM%m+$|;WAvMo_&Zu<Shwi;n>uAamQ?RwD-bnX#X^&V4a$~2qb_x
zh<_KV!Wz7f3koyM%hS)}SjsgO%9zF9D0H~S#?FYJWc5MNSfTS0uY6kEYuERwtgGv5
zFuzcZ!X-0-fz>BugeOZu%<`9q5qE>EltFf|l7K?=T>$tVmmbt$@tVlnczcRIcm`0<
zSt&O*$XN}E*cTw{MRUDj!)^J<@HBv!4n8NQ>%Q9#_VZ0r(#)4JujbScbnZLNAHw1!
zvfQ$5|9loKjX{58*!M2K{i8KA)=8UZ-fL_}MDyFKue!%)tHW3d(V2RGvV$Xn5xyq8
zH?V!PhcTqln!w8XLat|#JT_s9Kt)-~RaiOSxC@c(ia;(>Hn`^<bY68io#-;}&Qg7X
z6%}roP(kjgCTHt2&I|q85gjbvaAN3VgO`q3ZW;2aGEoEt6tA#c(epCM5)D#PCCk(E
zX<<zF2yum<G8I?p_|tM1QNdi_{?RSRM#dtEOszLuFGe3*P9B#E|GjQ8?D~!cQy~+c
zy2o@5LYe|zV;`cY%FQ*5l0atMi}976Z7#M!Rm%1oWn^R^T_j-~M77;!z5Ts!N)1({
z&J-BxI{rAhY5tOVL=;=hL!ijY-n#STH^CcL(#rm25@~rCzla6zOz9Bx5bO|iNhxKK
z!~{b5Sab#@o!<iYs!%Rv(l99!5qKeE5xL(@DSC)YBORV)_!=6xnVP~14ZwJob;M)`
zVWjA^)@h{^q6id^l3xZZSQpK_gitCgyuk&HCR{yV3)HGG{dCp<JxmE0ri7w0RY;Sr
zT=i`enf)XCVnm43)c4jUPVeU`5HxhuNH+0y-20OE+_Eu3?9`-s8Z)8RyZ4v#=pM1y
zaq_vWtJP##*c>5!Dw@UVWuvaA@+BL+1Z>)UzZVpregOe}@q(oWvl%G<!>&m7?S?Ze
zTie~I#G<NmAG_HL>-ON4_e=&!rAXlJT-$-1hE6*n^AmX*W8YsdboJBm=}Y~Lb*S`{
z;BHPXl$B7_QP3|J`rdBL4cD<>31Ti5g?Xo@vMQwAh_JcYC8?~SjDqQ>tVqV?H@!s-
zeGMG2O3W&oyc^=A^40gVt4Ea?ex@mUGJ_w+?&k>th7<^jniq(&97;xm2B05}my$}f
zN6ork^<B1aONm!Uid{>M_^IZ2ne69cxpk*YJ!<=x*T;+B&A{sg)?_%amHBBckLN?V
zC0kPDwQYplyim>a?V)J&-x_ww8Nhr~aOr4PSj4pJyRM`JZq@Dg?I9DzJWh>U=`dH*
z7kdN8y7Sg}+aDSC>6mhP?*ui(+%_L8;A})sYMldB=H!uv{r2NE<G%QqZ{IW<ND`L;
z$kfu&5qaC>&BkOD*-#AP@%jT-Je0!w4=SNJ#5EH=UtNqX%Jl~FcvkS;&t0R}j}JtI
z<zqI(MY0&jmb;l$(R#_vcxmtU4?bWI5sD7Qet+yzI6*^AOG|<5d7{ifMM#2)!N407
z*LZOl!G)BdIy_km<?Fx7d&wx$uLv)K0wU1!i5kw`2ajcAFK#vOXnf^<UFTvybY~!$
zAxOV#%G&*OPF5&)W7Mx-^ys%x1hLQ4tsa4s-3~fAIr3<)LUxS;UW@S}RJ*tu+xJm3
zkEIynOBGFb7P<aJ(OQqQ!&_VGCqE-V*wj0(oN0YZ+)C#q;&k!i=Z!V(+#j!eSQHaE
zI5<GAFlOpr9sP7477`i?C=V)X>Zv-{{kTnE30+GUSHoFBo7tw??S}RJlK?{M$4(pL
z`?TBlquIMR1vd{3szr@|(@xD?B1mQll&sV>82jzv{L8xmL#i10zo5G;Arka!tRakT
z_d<TPq^i_9)%f^+7%Ql-o)N1y*I>*WTyqc#f3$Xh$XA!^1ch<9^(p-6uzsjI0MEf+
znKKG&b|><^Xf6mpyR{yy``U4dz*L(J^emYbG(qlcopS*x0jDfy7CTi&9e;dt0P<co
z4NP<I!f4XAuID?foZ%qvzEf)Sn`6Wzmg_<zo$4j?@MO6;ds6&*)GHMg@s#X4;+DY?
z5p+imSdTEevs-*oHTMMss;p-k%8~e^EsUo%j#AqPh~?fmXOI8^1EB$T%`nhQ{=C2#
z-95TN7tz$C;jK5<;;Qn}!r$@1Uc(ZRn|+pOr=B$l6{#Mt40d1uCQjhmK??63NfiPH
zz<O9!*2>2{z6IAf+9>VsdpIyC!7M?Bb8)x~_yRS(!2Tq0>vFB@DEJUdb2ZtyT8F!C
zr-j0}KPw=x;dCwU!Gi~***^iUsTDgEaM&P-xw<(1$<j4rju2kkSbVv7zf!>EjX>%V
z4(*@@HSORSCs-vqv}AZ!i5SNHF2=oxtBa$I&DRQJeJ%42f%{{3@#_!^uMC5N%h8pf
zVvgZoM$fB05C3{aX<f-{KN-I2(_l?rd%xs4s)c{%BFE?TYdB(+>7sb1#c{IwyC<^7
zanepBp_F5N&lU16L@6cjZTvDTmjk(MZ9A(_6Cl2f-nastYEoKSS{fR6w=f1(I=z80
zPC)Q@2Y*3>w90<Z?lIpYyHh0WRW5ATO=Ghw29q9da2!n@X^!UhqY;*Lp@sYgcoh5p
z<aE!RiSQF|c|i>`G+KkUiwEmt39F4<jaP6e=y>7J;pP+$A(g68(tVXo5m7UjnH3}1
z0yy<mS6-L9j+bfqBq%WNKcEwpJV3+tkbe}{tB|#C<lAm}hB`SMdA{-qla^cLsAXTn
zE@P9IVY5u<Gkk0~Ike96qBNr$fPcBCIL;FXjKx}L-qz!taU)aI(GVGv&16DxnBwMT
z&KuvI>Po6eW=uAk2J^wxb{4mztCQs<;`Sf&BU-pvWkNzS+(;2lJ<5Po)`$+qZEYTc
z*8RE-I^Dd{8st_SpRm{Ed37T#<U1$fYjv<TIwU9>K~GyWQE^MV;?`>gp?Idk^c$<`
zV+S`Lf<M!@sd6dg5)_K2I*F-&<}5%^GlTvbtk`ktGt1Ey@@z{*QDpkQZO4gRSU?1b
zlZcbVmq}u8oKZI46lx<Lc>?n4VV@S9-9P_vlwo~ktW`=b+3AlVUQS#eQh&91gGf^U
z4Aa=m)m(CN<3E8y)TmsM+i^)rNjq(--!7=2rs$-?y1c?#xv_lmb*{KHl;cC&_$7t3
zln9Avi-Ev%M@QLJ744pwm&k~?rd3F?E}^DcWsAJ~jq&H0m}&qQK}5dC!ZPzcFH14W
zNhl#WIQUlLhy03~x)xPemxttvKYrl|I54sVwtk9?yhTEyJm=g!XY=G|b=%piBwQkL
z`+^tZF(z#7Z>zeqw~cvuhku_>BP3=IH%DkDiBmP(iV{asYJT0C!KJ0U{?c`|7Ixam
zmZD|wTcaXCOc)p&AH|-MIjo`I_VPSMji6y*fAu^}@2-EH_DJ;MrOfvLzaWTHP~PiH
zO5@LO>0=KnV_(G!Y}9bh3{mu<JyDmK+pRl~O!7RK3rD&z;7MlNQKBwu5C>~(qj{T7
zXRXBII&&VZpz;zsU&k_`!V=q-weE@Z5o=w%Y#6T@)E8Yigdjgzn6ky3o!EHUF2<Z5
zXLa0kq2{mZM*#;(B!O>|SGgmuPLyhY+3KYz6gqc=-vK1$m(8bl{YG$or@6(FwpX)~
zrXL;Z@nGk>?q-Oq=UV`RFUF$|&i(A@mr5BstBEQ6lE3{IL9{j^iXpvHLY%czfTwf3
zXz0N%K>IK3Wx}cWUI=ODom}vg?;FD~8F)o1t*<y|!regi0#$9<RN&&hn%@T}!~`c?
z?oXPzIYCJ}i4LI}x0~K$Fv^B=l=g#qSv*M;M<(q3z`#Yn$eDS;ZOBCOGhq#bL}qhy
zCXe`5+)RmK_jJ3p=y3jU`1V+Te;2~=2Z7a;*Z55@5HZxuHP)0EJIZgL%#I)EV%x?u
zrXb&qzM;clQ1yB{aF(H*CS!9cvLiCU8!t4G+CRf(NP;%B+1Jl_*KvQ^Atbdd<BNju
z!+R#0MLDS5vy`WSQ6TahAY?yKK0u&W0Q(2-1xNjUxDw|Wvxr5DSH+sPjB+$L8iRfp
zvTv*28wPy?n`7>HS0StP=S+0m6WkO$y@9iK6JF2r9+d?+f(Vyy`SSgn8i!96%lqdP
z{_S)WWC{Dfs|>1ccZ3l=2*Z8Y9!c#5!nCPMmir@%kKu-yb(I_s9u!+nRTv(cPL{3y
z{2l5M%g1R~J#>1M#-&U8I*i>BcdVNFYmFu+FE1H4)l=M)8@|)J7&t^d?0gBo&}!{1
zB?ieuM;Sw)CX&xo1~k;9B6Y518CCP^D?rK+;tk?4(}Y&&s6F=n{M?hc0^EdRl9)$F
z14jSB+HFY{eVZWrV@k19z3K@T63I?oN)-Q}#l?%mDQiFigolOY8}LW7da4+l&^PY%
zH};kMu#dIXV;~N|e9q6W#v8arI5;@ivELJ&UqP?1zqz^*ZS2I}sAyxelXVqafULoC
zNwkKo((HaT$$e|4>j&GT!xn{_=)BHtx4R!D_b2SkP2NhA`9Q^&doHF?*T9D*okUK4
z>(~VvP%`-VlUIyGU_z;xisIIg3Gcqx)!HJHT)e{m@riYX$zZCFhbOXqa|UB5XH241
znw)}J8W+h$fPlP-#m`~zQ+_|$revJQJznK62a@cnqvP-!xGNCSBoNiZp5vlf*;Yxm
z#up`$ySuyH(cFc{g32RSSD!8xa3Pl(2j?Sd=zX~on`CxDOEM1hmJ&O?G^b_e$1mrJ
zjqYCmS_}R@AEy!1EWbda8xKv{ZYSr^CXQWJed-ms;a7F`dvVn@uxFFeZ|%YLa99#u
z=D%bmz2|x}*ROIjF_tZ8CvCLmUg2!^FcyS4YxQL=C#Y!JGh#^f#+>VB6yeJyulK1s
zN9hP*C*}D!Fv$aQFlSHP-Oo?98$=z~RD10Ew~To28}}tSRik+`8Auhaf&*aY>qgg;
zk&)2|Ib;KQOIcajvw?5d#erHZ%bp8ezIKal$@0yi5RD(KMjqT`lY>+I-5bMud!<NP
zXn=IjMzd38E=4)Ahv~!_fIaukL0jUs^5dPM#mjy%arfgL?FK&=ks52UG;Uqe+rRi6
z%D5{QEj_UPV7brgaS;(6K+Yf!Y<*xdko@uE$A|rUsmGSZ2MLXLr=KZD_y;Q7AhvA!
zjrid+^y~ZUQnOiPug}Xb!j)dL>xI_0QCD)V>I*c2Q4wbFjLN6D4B|QBE7HpD8^s--
zW@Bg{eq2%W37y3qFzug)K-wG?`~`6t@2ME2MQLgG3|urdZ!y9pmA{sSSBRTe!wNNy
zHYtXR>%gjh>FW2<ZiEg6s)6I@)_xMifU-}hb>8-zeGl%bTd|`;U*tOl3W$=`Y{*io
z&H(HX6<Vrc@rn~bcGJcP)_xWZ?nu+Vy6?dxYrxm{iFj9eaMV);%plVNPW?{<AD?q-
zak4)+SdM~g4)so))=5Y-h=w5|!2hK+jMkQ#gxeHZb0AxLs8i=+>tbLaCDqpJ#8{_z
zgpB^Shnqr`c9~2}Oq?`v7G7Nn$FBFZvDnDsoPV59O^?pw_kob`Qp2|>a}ViZ_Autv
zt5;SPMjVhgy$Qc%GIt`y?H_`69GN1L+1T0QD`;;)Ja7FGru)j^WN39y(*pWPWq=T^
zF(r^b*QGfJ#UEgGOQtp>ME0Hq65YuiCRQG~tn>dVM##HK^$@RPkwlX3=>`9Uo3yQ(
z=bUKuQpBc01ijOk`PJ}U#2be%E>~P((a+dlUbLEdN-%Q7^FCl<$)XaSbFzG=c!}m$
z^(ydGt4%Ep3oHg~>F^cTL|p*5fm9bBp2(x;Q#NQ+T7U3Mn<$Wxi_2;#{W;*@UY9Nb
z=9c-(D#%RCm0DD_2+xYiC7tds1#uq+i%J=(3GfQ~3_gNPvOQnoXgUIeXw9NOCMLo?
z{d2l(&z#M}(=iwd+^`WhC)Lu3aT!!aLudxGva&kb+IkX2rz{>v)0|sP*9i&=y3V!u
zV_{)|HSkPPk@&`qm<R7<0OGW>w>R8=8a4Ik^gJVwTc^*-&Byd~REE&g@|Je}R~&D^
zawc_kAK{@xG*#~9hTXZyiMDu=%!x)E=(oqItFRi6@B>NQ`?8-HPk{=YDC&m(L|<Ru
z$*F>uuBqI74CE6u2!xi`uU~_cJ&bt?P%a4amT59!=@g)9&nYTW${QsD`d~U=l;Qo5
zp;ZUZhT2F9a-Hf!xK*MCnW-eRlwWU%84cVj^Lfr|uL9R=>6f2rb2Fu1{38<%_J4Q*
z<ZiAN+*f2~VbLflRnOxTMJaN{XH;I`NGWd&`EI!tHqtF@QwHodqI>3=>0MYyWtv_o
zdL9SEUS$@QD5Vw=v<~p*OO+C|vMOtaFKAY?v=uP=tC?ie&$5!D9>>WSj$wF%bT?r!
z^aGF=7kk~@*uHu|S^*Sa@zmy6p>7h7Ud!2b<4m%IZ}=Ty^e0wUR&aPCyf8~HKTTel
zjKS~louMVktekywH-(*DJseuOX@JOcRSk^mkzy`iH6Fld0v~;Gb&}&ZGDj`0)9+s>
zhL%J;fAj5*O-;=s;<MN9b1QO`O1<*D-M^Zs2jd-Ih2@)-Ccm^B<b9KbuT2NWiS64w
zX9IZ;ukun74E-1x83DDY`@yKjjT<*)(v%CwSXfy2Eyj76n3&kv<?v2HcziA#*Sqxj
zmxIuqRvsoV9xLY7MbBNcEwtL3P;ALYybNIgcH9#(t3%0r_4QpB197o=fxbDLXVDs@
zwY@tNfo+TPjMTy^fgwb)<#BwJ1c$H8e{@Oajh1k@b+@Rey)GPc8^(W`Czm$(l}ud*
zO%+V7yI7K>|G|m--s3zeJ;Cq=fr1wRHEZRKc1?vnN$;fe(9zM6k->%x3=G=ZkyUl~
z9z8038b~ypDG&S=U?thw+JeOd*&J?6d`^io)Bb4QaMt6G;gl6{cbM9bE-3Vl7IMy|
z8kVG3IfwsB=N@*Tfb54`6D8*UjDAYJO7~X2eE!S@ulp7iMNUFOa_d$kz}6NMB}=_=
z*1xN4n|-iA#sIk-DCmIPhy$BMLPBEk>Gb7*QX6}SGbO6A%;a*r6E|T4-&gg{(Xf%w
zQEsQ{oz{*J?01&VH~@i>T;b%HzdA{w&KuSGc}c7M+o&8#r|O^P;vZnu(%Kpt60&mP
zRwdd9G4$yXpKAS>@ht5M<57i=ukq!z4FFGoCb&e15ZMXWy#7Q>3kSIz_)(LC{7P4H
z*ulgk``R<YR;~Bq>9fdJaAj2Wqu&h)(CijcG3u)<Zpu{3=j1|RseQ-ZAB7Sy_gzby
zRSiF-H%i<~H8~I3)^|NFNo=c+13nd;1){@Sf!2+&69iZI-G4<=xa<Spra4XS>~zP$
zlf+dc#=ao*_xA?`d@bU@tX%u;l|YM8JS6$sS0<?0kb*-BH)YZXWRs7dK0P>@u2tan
zaQhXcr(I!@BZKcS?P#^E4kd4|^;eoOIpdH>W}otaSfE~@bJE=8mg&5GKJoEP>;g5M
z+Ys*}GAX|v+~2`(t}x?~L@EjnY^L&bz^#7Tez~t&r@*X|K1Bau`4lF-89|-n<39M-
zfru7uG@%!p!H<k|C&vNUJOCMd<ieOgF`##Lv|R{i`~tE!s`YTO@s69pu;ROp@cbe3
ziFb_;Vs1ZC?vrTUCDQa$x<j@Tt<i9{ReMJ3g&y1CJypK3$)Cp`Ct2&XIXzQz0hqSJ
zw)W>=*v^94c<hBM(Y&)X=){CFsjS0bk}vR&lPf1I=#lf^=d6sye1W3md2y<?$tO~H
zp~txEdLkqLThvYCET2U3nKBw33ZSu_Q~u@FCJ7tXdatWXzLQ_S8{VKsyYJ@N5>gQf
z$UX)r3-35t5jbwVmN$WVQ_(LnT5Z2f<)$mNoVwJB2H7AVDd7_46(33CrsRTO#qo~f
z)K+&S8Jm*QQ9yb%U`Xs#Sgp%#XnwgKpxUh+4de*kW_VI?h9)mo^>%$@#hDLGveH{Z
z*ASGMOCz@QZLTwKmA4inz!;q<QEE?rb8S}8K}}OBkLXm=7NoBe3V6G9^EgJ6Gxg^m
zh3c%ohwubkDEjI7>%$RS$cB)TV}1zaap=9oHut>Ky5}@O>WbyePm0iemAiiE*x2=r
zRmiF9F~t1#Fh^BzTthEMln}zVS!wT^9%V@C3+qo<t)R3IN?jS7j^<M8@WO#8MI`BH
z;U^7334Jq07e7kAlitd{XgLjoz3Ru8yLJ!Z-8yv2B+mKxPtk2U5tmQAW->6fdHA_Q
zC?Q<$+mN<F%BYzrZG`TOwZwq+G?cm$X1!oWMHqP~l{X3m=)`5ib>J_fB*123YXZmb
zMbyJ5*qdy1M7e>LN-Rk+rR5Yjx#HAPi)etgN+ZL_H1^jV-++L_0&(vg2I@dZIn8#)
zKQx5i4;^e3PMBQrQWGt-gR0i%jWMeg29sw<iK<LSUNY_?z-kalsQKlj3zN(i;#pyD
zH8SEce)vB1_r-PxA3BL78eZ#qX3!Xz>D!P<Lhb&iF!R4hkiW$C0C^<!BHNdD^(<)g
zE|q-Zw3?199;nJmy?h(AE3H}siD$}8zl^P`l|9811Y9E!YR?-3KCrPF5NBa{*KUiZ
z%A^^L0YBL5OSr5T2%MW!m0qW-S;JY1bV3d*$GZ!{P6Xeu_!B)2CxV8vd<1(*!}I`4
zMbm-YWBMJQbYpuVPRjKrsd;iBE<-wvSHy~GEG^I5roF{(U6(n3KJ#BFa74fy1K}ty
zHQ`J9Vs~_oxTp5OQFt`Mk?y5C;1Q}8<yP#Czdee15_Gn8=KYas<sN_y9W%*SK$!zr
zEcV+sm-FLY;1^n6UM_^!BS0_?aZ+%)3soE1?AJMUC_=NF<B`@Q>C83$VDs%cuNZ*>
zKSK0y8~`W}F~C?Sv*v<ufltn%RAygdGFf(ib#kFS;mk*H%e$}au%MQ0M%Vsc0$V3V
zxe$M%uQYCsPou<4p>E^PFD%A}hB3WMfa<CT72iTf1f3|N3|xqumewTrU%*a`SRqYf
z3k($XD@})v(7kJ&b#<6w4CzFd*caCfERCQ(FR*PbfM0eySmX0Ncae5uE^L!(?8~t`
zb0Uohv>q6kBH6w^sbAweR4VIZi*Y3c4GBq#^1`h^5GxP=Jb-LwT)C%lLSe56!3<vu
zS0J2((Fndu8~m=@=-K&^A;c+9-KMtkJ^is(+59`(QET8yJf=ngTh|$=Q!6c}miAAW
z6%$2_dMKr9)%L&HPk`L;M7e5>U-FsZIUzy<258)|5<@v?$3^7+{rhf5TWTfm_~c+<
zd!m6oA3Bh%XL$r->YqQ>1Bx*Z4FM+}&8@pNnAQQJE5;VyX;(2z_$${jY4rsKg`CX>
zdcTi7=c0>Mz`>=Gr68XmbdcaYeNBBwvTKlSs>|Pn6CbTxK|9O`<b*#aw}>QhfA-8I
zM=u4Mjujw*A_;UlU@7~$p%$i3yvgDeM@hD^8l^hrA%V+qaw6>T$e{BhU?;dhMj1$!
z09ikbI?uq<Y2LvEpkD8@@P@l8ve7TDQ{J3)87I8%7_4MK^|GsfAFp>ia@(%g_}M4Y
z&#hI~X;mpJE35d8M`F^XbYdE@&KcNBLP5cAI+PCbt8^%NF~_w{0VsLrzf=ZTpO$*@
z@4EOE;aX1}mJb2T=c`!Vr~L53BXD(}mY5WtKab%vSAY5PCDYlzL+#~1f1XkETTjQw
z$Jg1+wfOk>;8So}fPK{DjaF*Z%f`VWV!za5>Gg4qzFm5wcm}P>V;SiZNGAC#|NI4i
zl;P3#44^Yz#i@Jy`l2#vp1pk81OD+CUh5=ZBlPGWPbJ73$XaJ^F0P8H&AGX`(bY4{
zg~P3TBK`bUYv#9g+3x@@hH7?uAnd2X8o1ej)s$!cQf3I;f&5bwMy0C<2xwr{0pQum
z@m#e$UC)#Dw>%qDm4$!{Nl8husubqr4AY}DHHN;4dwD7-gb$j;^IOB8o0CbCNc#Bt
z=I!pXNh~g&UWyPwJxw@1PbL4DMZxqR#z-lD&65D$#Sg!;U|-odq)$rA<oIISTkw47
zt>d64xDTkp%s9VqB(Xm^Ipcny*W&ki@KF09B>_<&29QIun%{nWoSB(vHC{BASJ?34
z*)txhFr!jI5fS&T-_>B0i_N|`Dz9Eavjw$VB@-4@VLuk_BFi!dO;|(N>ZQqJlPbJf
z@$%<A<WrwC{J0%pTZ=5YlC+~)cB);p{9*_kv4-3n-6bOZ-SGEKY#gmTB@|J5loBw5
zLT_NKwFPd|-1u*Kx#xtZpE2yc+S=O<4Giwxy9XfX*O!LzF)<Th6@neW4X+ar6Y~V5
z;?JKyLBs<D7L&35m(Nx@vnG$@<h0UQUz{=>HDXjs={BSbUa-LCYtrbr&AVmN1{ZdB
zM7ZC@?P&D;oMLA1gQZH9!V5JH@hAxb5O7Us=fwR5*{}73)5iFcB;e$Ql*rUFjM$T+
z>MnQqR~k>2s8=NbECpme2rrI6jhL}mPESt<`wob_(lRnK($a=TMj(DU&bI{vfpg1w
z#sl~+Del}61$&4|E~<#b{9z2O>mQQaVlUKe+?)Z6D1Uc`rzBv8iowr}GUo@SIvfrM
z`yKqrlM1|-5wYTUT@$G1sTKTK#a~A0dZCjox^LXqeU`1$`{ExQ`&D2A6w`YZIiqH;
zqauron1GX3ijJII9YFXb{BGNLQd!`q(<2X+K4e5N;dKMa*OolA{VMv}ZOOL3?Lwd1
z$L9~0G1>*4)bC3qL|?Ye;WyehDh5*Lmt32N6@PYfRI;|J7KjW_(*%}&-xgZFJ5ct~
zF0}m}04#UF{rbcx8zqv+Yx>I0uH@l`GEJoy1>|y5Sn?UGi%9uzYOr+^8zq1f@~Bpa
z3&dxq+6*}s3=(_hFEw_zO(wsL;BZ$<;*Q?D3?KA5oG>mfE(XOpKYyj;`mgn0`6^}F
zAu#FQOHulKF4BAZy&Paz#wXM$u!{xl{sax0d<YH}blv*{GQii3$h+UZUk!~F1o19x
z-OgsA1bLnc#9U=E>c@4(Kj2bQQnN@4=8)KnS`cb;@=Wk@H>c@AMY8Ms;j$TrZey%)
zrt=dcA5||%XMqH_Q%x!jIt-FB_X|P^7MfdK{ggx(tOxqz__wF58<7>3asA72BCdPD
z(nq!#wE>_30ByzWT;xHpT<ct(dL1*fHQ*J?$g?stPx!c|*Jto|Xq&6F*&6<L{glc_
zL-VZCTU~Vnx#o54IY4Q2k#d=51yb}1yWo@IL|PYt2v}~!h}S4#bOQxcw<$DRzK@%7
z`g~CXIY7}rEJ*P_RYY`0pBjuSV=vFD3d0R*8q2JQQK?^t@07_o>j7XQDtIGo>S!I&
z0cx%i=WRHXFf+#59@FW(Yci{I;dD8Y0x%_fAcuavuL5AnwTG2f6hR}3aQK`3<;Z`h
zTtKs7axh8TX)ps9i`zfPfnPLU$k7aV(||Xz4?rCtl!3{zqKmXP>N7z1rm*&x#vQf3
z#1`$ag}?x60X$*>d)d(N{sjANo7J-?WFEzwC!Ih3IlKHN5t88cLxYup6kw#H<~31L
zS65&7;V2{;obgT20}m1rwRNCnP@Mty9eAw7KxzkZ8TbI5wx=(q-L~&f=1;ZVnIsnB
z<~8l&H;4bddaWhvft!*-P8+xoNVDMUh0_fv88yFf9Y%ADXsVGv;sx-aY54ufI?xah
zn;H47EHE{<^bcY&R~_w2wYtcYKTis4c6$b|8dF^_5;)^I$3C^Nq0|Tj)V_^;XD;T@
z-qXYjAJS%%z3>b=cGs7}$dw1|K24UX7MZj%nRR91Vi5;Bu1}q%dJej0s;O3r5!#bO
zJeVY9z^99&$3=K@xS#6L->Yd#K8uVc{Ju&NsUqt;4T7pj)Md^K>l{qmTEfOVn2<(<
zu)CG>qOh9??ZkfPTb?SURpRoZC16j)*_G5@@U9;q3=6&Cg@w!hivoSi`}dWvJ+J#G
ze=OC5TkL3Ll~p5SGfXGn@^oHT_Z3$KdJ$I&I$R8_<Iw{Je@h8RpBkUUWxhr_bjbP$
zp4?-?EQyOnqlP8=#rCsih;)A6RD+7;5f*^EcxqVnOV5NL7a_-66W%e>w;BAtW+o;Z
z_a&t4+eybBI;|@IoqZuGQHesqGCIf|tlGU9uxVpo_$2mqXK*LrNQ%7qVphuy-mA>M
zw=&_|=)($1aC{QN|9F#O1>E=E9tcuAg8#EZ?AKNjiq+BvT@VK;1@O7FRP$PP@cHFz
zh7~Y6q};CSFPJ3jlN`MjfSi<=cve7YTKu2BlEt9EUp@gyWMcCWIP-7ZWPU~s<WnHX
zFWLwIBNlLxsm_70=x31CN#X(QsbCQT7ecR*JxJ1Jwg3Wxlmy@g2P<op&77qF8R+Qv
zoN!knDIp=@+c#VYXoa9(W9SC}>E#v^q|p4rLTO0}+ix7u*Rnx&yQX-;0#Q#30q0e(
z(U->+W}|D-x}Fax2|(??4$1`ovk!NKAG5Qw^YS_X_RetfkAOmO0UHA^uWGQ7qfi2A
zufAS!5d#Cm3*I~>J;u9t&u3zmU71c{He@$InKx@A05z>TP!V98z5{^?SR$@fc|SkQ
zv_Z`|G2rrDa!rJNPD;`pidD}uq|E#i@m>eG>pK=mFeRpqT3cFx%fj#qahm|U7RhBD
z*L4e|$}TvN&y+&jNm>03UjzBPYwDk(Ex1K1Up=Xx#k%wb8R(D+lTpcN8RE?!NGfhh
zeKyeXZ@K*gt*F-IxBpLehi45(&Sc%&1v4A-SX%jO<(GNgKb95gi{2(Ti;6<ZZ?pDF
zKDkAndmN^gs`LKYrw6AX4}-vv3Gy(A@qhC$0Ic7ESCE?DN~h?pgsQ5lgM&ki$JZ5X
zNn8eq-f^EBzRHNWZua@GamDa`U_jo*#RafV9<+K&CM(QEKibg2U!W~PzZkP?fbd^A
zGk+i_p~;8;CAei>z&-!}<n=*0@_h)Xh5x;7_k&5m{#cSjpaR<H1ye1Bg`Qmm8nM_w
zNnOGrr~H+4Ml;ILG&ZVIevwTn!{`RX-XX)2(ZS?p*M>jE^tgT>L8;!%L(sqf;(U^I
z@H;*33)LrN(i3rj77WG<O3juc`D*ed)6^d?Ps&;%=I+R+mRDp%TF}Tu^3jPqQ$&XN
zLPlk$j~*rHhEf{nzPv>ttsIQUs;bGv20>y;+)8*L5dQCx0k4begd1QU!Eih+^H8L&
zXRzd*{yUPmIW*X!|2JRNHhTGIbV>}Kp8)r*2Y|JFGf8IGJ!C1lw$(aG9*3_jk5cij
z<Hm}YyV%UIqWGPgH-}4&`$0L1;msNXc40(VnC0)P-9JCS06Vgf;~FruBq7#v_7q2*
zj@Z>@$36gBNNX4PJwdz5LqMAXMhZBZfCWzgWG#@^)YYR{LqH?|h8j>Kfj5sq1s`!o
zSbZtySHAYxAxk!NJqH*4eF?%HVbocJCU+Sb0h>k)-4S-uX>bS53@=ZRc!4{nN~hp+
z0=sLjg7UrpDs#-)eAlT4Hfzvq<O1+0Xqggl*`bHVEh<2E|GE(XlPnJcZE!UYGVjpP
z5I~>|_1OE#k_8+-Mm^sE*Bd!q{W)umO!~)vZAb&Z8wM@_xPII9L4yvJpq+u%-!A~y
zwrjV=m}=ew;%gL`mlefM+SU)q$;m;-6A9$BIT;xnD`Yx!FG1l5)Oju<w9u^UrIF&O
z1Q1M&fwf^{;wA(P|D~m+X{o6YpyIK?xj4Zt?C%F^$*04fo}Ll2@-S-tC{GdFdB5RR
z1IkQ*znSs}dF65SUaL<l4Do_qKYHA~$M_II>sK0*PXAO^&{C#U3eme^;WB7yN(OD*
zf}j9BTn2!drLfXH)pI7yzM6xsJlP0Z`Df37X4t$kn1(j{NtV_XIM=}~CZnJ@>ST-%
zIol`&oFpC|UM!FqK;Z8dDF=p-+=h3AfYJWX9dUp?rt?4zw)o@oQC)YKmDw)*2#XNE
zboq(j2)u$ofQfmTnIRV)Ws%OBySIG=3_-+nV!taar^d%MuY_~jC`ce&2Ec_Ke%U6I
zevZaw9p?x0ZKiiCO=azZqj7xy+79+-)-aZoUItC}E4@71dmTL2h^Wir9H>}b762P}
z_9o`~{(I<;I?uZnBEVYS{*^xk5&*a%j=-U(nXg_s1zNAP_F8~rS&q2J$o{9y-_ST}
zmcLUEl$46u-5WrGJ^!QMSATW%JBWnwd?Q^w@4!W*gl;e}Fi@L=d{uniz<`Hx!>P8;
z((Jm=&RTtZ%JiQc0rmhWu>Q&pr2gev`GD4&2Aza|dnI50L#da^q<yNS{u|3dOXP<S
z1W;#ZXHaP9m&bb|+!?v7Aw!jGz~>HpmkX8+W5BJVku@!xVSg4uS7i+f>&gi#5H}(~
zuKoJeVRyd0tgH-J-av;PSTSIMgi$`wzZNfydb|4i8rBPHp;>_Q6NDUVdXG{~C)sf!
z`jJuXTcBKU`c&h}9)0zcAJ}^?kDk8U3O9}NJXsqTEIji<i?I7IEq0oV0ahI3p&@}*
zFNlf8w9&j;RlzP8+nRO*dsLZITSiK1#K{>PxbFs_ilMYY);GPu;71WYOt8y;W}rX^
z(Au#LdLJCT%N8gR@!!2ZWdDmOT6z;NW_^i)n;ggMaaTt8G!G!*%2QkNtN{GpI~dai
zwd;d-KQOVdB4<*1)cAoFxf|4GaQ9tlu+QL!C&*kp?n1!pmyGui9Iw-l8<l$OpPDny
zYuC7VBKHF`6_el(V+uNL8@@#?>Ry5vX7rbmYiW^8>OcBHZT*3oUm<H4l)Bd~3!qK_
z`O>J=6|8Ko=+Mknuq{1znsFBS$tfs+9W!>>RkHK^WPiJAA$-k)h5+FQW0R|0jr*KD
zhJJn)PCK-~1r;q$eMb5P1vK2ag?RxiIT191ps8~YNO*vKJOxZrVdEtNRKv1iRBDK4
zSs<TZ*VP|xKLSO##@r7O5zqINu7JzH#LjMS)BT^5ghp8CkfNMwk%)sADql#TK5}w?
z%h3=$t9w~6yU<_<#L*TZ1lm$B&rcBbVjx7L$|UO%2k72AJ2r)5+H*fZlm`S}3~UIv
zPNq%-C{X}uMlF20;tZ0402y$)YW*1D%5V^43U<<war2W^R2DDwCFpg{Qb>sX{P`iU
zwho=C3OMN^-dR{|0L~&UEe%YF5D2Vho|JWK0HRu%0_y?Yrtr>5bL-!uT!FO;u8j+5
zf~U5ibpXIf8bP~sc;OYe<L6#8UKi`S*1(VJjfMe0S(%0I!N_2q)p=58)uc_Vetae+
zdTx<`ylwAG6`ic5nhD^-^z`%uCk2?b%<}^A<)>=B6JX{+JHqFVq0a*)+=YGD$-czP
z9f*ofwT9MFh7&_+E+tJZ09k>`LIU;`^vTMK0Q{B->~wirnh6RT85irS18yRTVD}jt
z8Hu3=o`SrV?Ii5z@Sy=!?Cl^`*rpdvI%5qE=^%u$(xeh>6e*CpECBkX)l(P|IYnp=
zQo7d+eu3G**nPk*fjv?=Lli#vf5K5%?$>%kRw82UQzG&@sDj`*`mZgeTjQ`QVB;lj
zw<D}f0yu%3%`H&SC*0hT@LKALNoNj2OLn#Sxcq1LFY`Gn=!goqe}%N~7q#}RIE*Ps
zs7S1Z3BP@Nts16Frn1tCd+HR90f34$QUMwELr@SaD=VO}P0c{(fGwbQwDJ!DQNU4e
z3>x@G&A=PxtD<=<v%a1^-yTqF9=3zNhwT;q{XyBk%0c-qq8>#3W8&tNSsRgQdUqgg
zR?aM?KmTG?-47OqTvSetJdmXz9{n~DRAok-xM7lOB(8g?86#O&&j8ew2F7<W*BS$R
zpQx!Rw#@VHvxD_Yz19E_QGv1y_Cr0$CQRLO&xN0bzI0p}m@Dp}#mB?56G{N4`#R7H
zzUrTyZkP)y@LwPu|M~N$_+3YhP60Wm*7rBHhiMUFZ1u26-rIIe=$;?0J%+J>WV-|n
z9_{Vz(34@*dSPP&$XtM*eaIeg&3pJ$z>hD_wlhrc1I`|f=nuZh5>ELbRV_~)Dl+)_
z`5xBl{C{`>z~=xx#|@?r)b!8*aAt$;<ODDmYE6|tEr1ckgU@3r$K*^AqO|>`{bF&`
zY_uEUs+9}nV{mrY5T^8&Xpj=m!5JZMV5UP_xZZ~Z?3X^25qw8^vCn`hri(lT<SY>p
zQGtzPZd~*A5!v#p12|7~;|F&(_Y8HNw@X}S>&+>8xGz+1R?NR!H0q6ATwcDWq5vTg
z9Tk1L(@G3_nUswxfuG=c$3h@6T|gswDqFqr&+j9Eh{7SJ#q>s@99AdrG_q<yR~k^w
zu%ZmXO^E*bRm#c^S9cn8WwVFq+t}EEyU_dX!S(2CL4rw7+k5C11<y1C=4<m`-&dz9
ztwHzxY_o6P<dsm(4sHRDnheD0=g0nX4o}6K{pIKb7pea?_Ok_!f$cGHHL^RLJL)S7
zVo|yspRH8?{yAtn+;DOOMrFSnH@ig4%e6A5yhwnalfwL89wP8Uyd0=951nDx6{&{&
ze*?DCH><0wfgOnm507o)k*FxrsMM4<7GTh;!cc;z%48k3dxs%<l&p?{#UHL~j3d3l
z>mq}Uje|kzn$iH(M%RVgPdFO%_I1UJqv5ls!FWX++N0LrulP%HwdFDEdyR(I4?qv-
zs>S2DGG&+_P(j4@IvGKM#xY?JR3u=J0se^CLRm>*)BxgO*A#n5s=<{V2E^dEATV*|
zIl#+#ra_L6|6W~_=@lI5pC9LZIaU{o`=4vu1^yZL;YP2khiUtdjDCMZ2_|PfDbmV6
zI8pFe%zTo)?5}2phZ)zKIe#B!C8*)%p>*>d3ZS*yw{Krd=*lIq=+?P$;QR#PTrpl?
zcOgRj*Awuxg>28<V8-h*^S{!)2LNzkd#{fbUdwqHIP+SC;2wPi<BH7$a9Vxuo%{7y
z9jXfK+kUY#3qlllki{BjogR=6Rf&ATw)=nZ8lJOBr21i@>T?G{E8w{ip&ZxGdB6oq
zK`E3xXku{f&0%7LJ@IGq!&K6XGyxBc08fWF{sC&q;j%hQHjVdee1N{HBu2{(#}@71
zXNtg*p!fSaeHTAK&m~b~eoQ9V2ulJyF9YD!X%6|}zttpDNU~HMuRJ@b3K95SWGUff
zQK>cq=*fN9_5I82t{fM)xq*;C&<dX2;qL<nPuyS)0iR20VN*uCX6N6((gbqBKO=)6
zAS6s{GvR^Zpv<*1qC2MoA%CweAh7<Ad7u9&PxSxy``@Tsh!dA7IJFEar$EYQdHIdr
z5;Usll$nrpmxNLAA<lO@ZZPl-+ysPcpxJzon5d{IGqVvOU4hKP!Xi6a1t@Re=<bQ-
z%PT6{0d1F+S6pqIH5W4Asyxi@u7JAe=FOXE^gv$|7S^5|2g@8Z+5w3HB*>keohj3#
z60-GQUYWYJq?i%`TX+3F){L|bg#D+T_kfpoWYZ}tBjW)9qbNQEP%-fDy91A`ES-ny
zo*iIE?(y=XGOd%x&g=@khi`wjfA6p{Zw%Hns7A=z*F3Uc#1i7+eXf*;l38D*4R%f~
z<>PB3ubEB22>APBwF8Er)G?(9o%S5<NiSx7^IIDM8fex*h!ESiY|!E{C>Qz@gpazp
z8%11p-oAK&2fd7Y+`H*C%xhV$B?s@brI8<gh0ovWw9%Emy}$K)Dcln4lJLl!mkJO{
zXGin`0^?Sd-P7L{vh0Nt_9|8t_3|WATClOO1Zjw_X;Hv=n(<Oa8G;HQJXJ(hvldzK
z3yiG2<omT{hmYic?{;9v0((A9HX`aVP`nrf1*<`s&J~^a=cPp`IcGYKQEBy`MUL)Q
zWQ+SJY_9kXD$WaY7$XPf^Ai>guHVJ$^(ke7b*ca2YF%mx7+c{JJ{|&lV7MxL{V&P`
zOvK;S4onW)`I5Ij{)_A|)BhlL|Nbgy6Eih61r8_xqyZmr&3!*K=mZ7+WfyxPa6U7E
zMuR2oAm%SXCxT3fwxiR!L1>``;a_nWRc1Q;!t(=o0!c&!jyD!QdFL_L7OgW+R`{dk
zzJW3OlYmbTD?Y#^(d0$>U*4D_0hk_27X17VH-62|5<`i!KcS}cqkF|<V}C6X1xjiI
z`-2Z8zhpM;$IcLs{_9N)?<{jtsHnq}kt>FKQxdpLGAfdfepHf3wgh#goh(#NZs7w3
z=Y|<RBAHe3=pHmc^PB~4{a3~|OIq`xZ5nGRHQ{|f+?;rv_Lp!jSM-<Q9xBT+C;n)a
z_-gjCq#q2AgAEvzcbwL^%pBWZ^%%r|zt*;uI8}@XLaD;CR5YMjqDwqG#qL0u<a_Ca
z>+xGXzuWx^%ZPd1;V(ID_QlMX)&XIC)(5V%s<5IK_-x9b*Gmx3u1@2d#5Oqji*dKU
z^2cQ$OCF&?=_?ggTH<(LS#OJBq*m2nLrJE>rC|j@sou|uLkZ<L#`Wd(3-2?>N~y>{
z5TzP^C#Ibn99$U8{OevJvr}_GeS1Ji$TMY3g9*6NlBf7D82pKjP4t@`kw%-Cdi%qw
z_<FbHzWS!Dtvx3oS^eR2tRBGdt~V!6Hf83I`&l!;sR@(p6kc;liN3&STgpfCo~i7Z
zDh6V{^T@6wF;rnC>YW*rcl+g`nso=b4!F`0emxEBPmuWy=p1chgwbzHdh|Xjy#ncD
zusw;v8uGKM?T>-0Fgb`VrUm4u0P4j&GA5P8CZ^rk-fjfev50_0xSdT9V}&5ZN;Dgf
z@EZ{A7?l(U-~{Np+vCMgGhQWtX9>kwH|&~dS}m_k-ee8wt?Jb2Ke)aC)ifPW6!4(|
zO@A6dIbUKi@c5NhxoEQq&S$UcPxW8Yz*7HLRqq{-)gS(kA7n&E5sE}1aU0pPWkfa^
z$=-X9?5v6?J0vTTk?g&bkiGZL9%b+Eb-zEq-*@<<KYDQP`<(L{*Y#Yl$CE4QyULbU
zVI!&#Jzma0-hce)4d4@m7Hy6c-sRyLbT-sRUD#dvRRmjtTlE>+_-DHXmFeHO5)1dc
z#XLykOc`XIPWEubNuWp=PXab$cGE|FB6j3or;(9xq*jQbiT1vv5ln4V$ENKN=^InZ
z5W^2Sj2d6ntN7lIy>|Uu0q{Lq7+O$wY~|p<A*F(r!^kv)=#@c~B~~uZ^ogjdf>k)l
z_(ws%IGbr38sXkYTlV54(8Dimx^53zl~QqT*IyE!;qM~td`L}HZtlNy<Rw%n)E$<-
zoc1F$eR;|ZzR0Jg7oQ4LL8)qxhC579TVH<x)A6Gyw5my9k_uiB-=~p%ek|~(oq>^s
z)M&2e7RW9JgS?D}6fO&Jj9i_*h8zmFO+^5sV)d(*b1f^V){7lR#=WriY}4Ur(CK+@
z#%l91EcnwdMaBK@vi+Z54ag0y1C0mRZ<j$h0IyaE(kDP1?8S|GSLD8S`N|_xeN(nC
zUg%Nqk6)AE3`#`kkBTh&vl0_r=T(aQ{4<^kBFgKEKW~{{m2lO&ZZ-*MIjer%^SDK*
zZJ=57;6^DGyRNk*3rj8O&UYm{a@#L%IKWU1e7@0J4+%uzQ(btWm(g3ZKV}X67L1io
z(%jF!8dJb!WpUh<<fZZGQs%7Qox?IF^5WlTxlBxdupTnd(alb}Pw$x+(8z4ADJW8B
zg}F|&8Ak*uP7wkBA(V;VK^8wj|MbaL9CILEK!!6PGIOwzHQ`hKPy+9WNYiUKno(gD
zU@O4pF4F<TLIpWDGR-+Rj9R}<6gJ0*7x26N8d?;we-%g~K{IQWmLlxs1|n$m{sJl<
zTnIDTH&N#a`kt3>O>DTPKhapb&wE)ibKgx8W~Z+^+b(PcY&(yorGG-@hqfIjytj(U
za1EH-#;D3xcC<*zclZ16;+yVE5@_ejuoyKsKH)t#9hXiNK7u=e$6JmC8z^73bJ`XZ
z{DePdr6MZGpJ)7(SBx_nc3Z>B_-`OoYP<5{7I@{=;v!OxOWkvuvv_G$xCFw8|HSu5
z=K}}BC5`70cBbBHkXGL^m3!Ne@*&@=>i2*CTQZWr_e|e8!Q~%w$3a@YZIH2r`jg~5
z`rdQVF_np{!n|U1%YFTQxiP4B${Qsv*ze%&+tp%-IT{d~y6x2cgnMN@kW3Q=!W=4F
z5y!b7AokMq{akD7CBPI^^z3M|;wwrbJHo;Ba!@s$Y#-n^3XE#-=t0u7?%iX<%l^Eg
z$FBV3<U}cnH+<;@aWXUtwHJpO^+Ftv^JhLCIw%6q0=$^*J2Oh_>6bY?9hLp}sBOoe
z7={axU582VA#u6)HopTcC#MrYARvYZN~{YPbW=lF3Q5|<X2BsLkQUYIY~KC*w>)0{
z?ayQ$AvehTcL5!Wz0EGgd);{MC5i6|<43DEpqjNxm+=F*@5c{G!~()~0E~zH6AZ{c
zs{`E@x{&3MaHe@N(}0oDo2|?Q*wQ&+B(Nag+~P$U8UJ`pg(qp`hRmPK*9+}TDt2XQ
z#bG0=D);XN($L<B|FBlszGJ>vwvp>89V|iD;k1`tdn?3n;yiPSjZ{arm`tit7>i75
zF#U(;=0(TG4FrjprwnnlPx|x=X?1++(>fH+Gq|OvK<1gzyF5`S`Bfd%X-k1<VX?!)
zugJ5LgZiB=!7kkB=x8YH04Ij9lP8}P`?;l50%+B@)&pq0Yn(cFl|)_nsiQ>?D?qZ(
z&^fi=$%^+`=9=tid!|k17^8ox%b5$A)L$X}*`?M=1InKp_M_GEHVmCrwN?ltdBdZp
zb$0iLHVSg%a^6T3XykU9aUq~$E<K*k^p8hrgJKTCLu@nuX)yTgbn+#Of@vp)I!m2d
zI}ViOgG8zDMyCBPYIm0A=8Tqz185-#X#8wZdE;SD1kLBYma<O2u_&Ah-)}QB1^`c2
zdNI~L%jZp)5W4!9<Wj`^CSzf__0X(Cg)kN<xDPyxUVNgE>Lq={b2B`hsj6ICi9QY%
z(xpAm+BGFZO6fG^3RuOn8GkNA0$yKVA41E@n!g}nVP(bK%mI0F>BYrc)_5IbUI(~H
z%im^~M#BGQ<e(;La2=pRngx1b5VU*z__2q_A?!JIfPcbrnG6z*{BjVS<(KboPU>J8
zs>j~_7EE>(JBr@sjG&<XPLNeWB4w2!<sZNrn2yYC{`(>Pc(AE`Laf4i7T4A`r<meU
zqFAv*3Jc-E1Sy~V(<Ew26Qy+`<n-HyM*+;3xF!id+z7UVwnn;`mOJYcoB!>E4`H%m
z7*bka(21$6C)6kWzVhD}ei{;d-u3zCRim&Wgx2OHENViCfDjfzLW0^Sd9qlqRb?qA
z!4$mz-^R<MwX7=<!>qG)4qmmf)&<=iH-6ZUvEjcLEd~|eTHzG^wm)W{Ay)^*2Nr?P
zb*-)2(wAPg)u-{#gt>bF9UlC>M-!vh%2xOU1MxE;CYqrQvLZn3KApFSrj?csMC~LE
z%_Bq-*%}QJ`ch=3)qYH~Wb(c=eoTYoy|h6+$%bhKoTJ^-L)Ettru{9$Z;Kv+Xy;a7
zmGUz1*D#A=A4s?<n2}+N``{rtz3kx*U%OpFAviFVc?OOp{2<O|zyDLhr=;PN{z}n{
zQ)j~D*z`QCS65>bhpzv<uVc0R8E%NRw5|ueu5KSE^^}X&93SiZojo;PVqh+H-3}B}
znP{7zI4b6G(}Myh=b8R;tQ-TS0)n`C5!zouN%J_DSP~<rW%W!MNeAKJ+H(K87koZy
zo)Y{brE$)FW`hq_7mZqJ6YX<QX@Z9voyZbrqrZu-lY-%-jAvTN#FyWyn@u87_jQQk
z7`TUzqI~P0&eeY=K82zvPwy^_uf?eH(<$CErLIMo6_w|bO-@5;%`Xi<u*4r?dF8mf
z!KRxH-wZ=6O_|$%>Ul=5dUY_&awW_0(6np}m8*RF8n!-i_ql%$<-(Ixb$_7A_EZo|
zYyaNElibn+TNJNu9VTLqkI%CqsIj-QPxK~Y<d4|ve+8ni>qgH0(gpSCHb8C20e+WS
z!>h#KbT7WDctd7VPH%v1PoR@BhbgZtiyx-8n^zD#P+PLUZBJg~G}lLKk>O)2^lRh}
z)_$4cSB#<`^^r<wukJcq(sAHl_~i5Blh5Q}9%=UuZ&GKitrInH;#>EIWHWj{befg(
zt)=}aLRlATO-$w8yOs;C9`Yx;3MeTRcdl$5^Y2qcsYa?__nJ(!vzX+Z4)Bdo9aTr&
zjGt*%1eFV4VAbBkUgdjlG&ecK>5uAwg^}*ZMknW$`8S@Dcz-3Y3H35IbI{1;-Al*r
zrf-zF_N1TR0%N8)SdZ*tGBI|ZKQZfx=o~q!16_7`(w8YJPgStp68vqewe(@Tf?Bm(
zmz7jPC=!F1K_iM@Y2dC3+2mB-t7|MBMdL}Pls`#fNtTV8pvk{?t%Kw3PIsA{lB4~V
zQZnC(${Sopd1OPci$-JW&+y5M%bQ`pMD5zOI_;~FzSej`v8L{K9zkJtRYFfw2ER7;
zeYj`fyXP8=R^cy3nC5E2tdv60e=@nY`Z(>P@cVLaZe?vB*U*}eQRj_H6=R)xaU_*L
zrv20T`1mYs(59v9=bj|fCN=fM&hsCG6OHj)2b8s<{F@g2Q@wRu2vhla+w;zCnAT}t
znz!!n77P%Jkd5#5rXmsPL`nH~znKtLd%Ko9KYhYS7qkx2Q&4|=>n18=BqT{{c}ai&
zLaTFO+QFCuHutKBs>>LubgjNq94>zH0~Ls9TsqMQ*1H8J#I@e8@k#Ax5TxV0ej(C!
z;`kS~)K0VEx(7Wat|PB?q8*0vQ?GcIOG3BtRMj27RxK#t<c?6@=ZV|gI?}prG#*E*
zX0?Cq*am?WI?{dYm8-L4xiS9s>hi*Z2}R0dtu9V))L-J%D>#c=xlFj5f}n!4bj|M(
zlwz)6Z)zs`A*q13yTr7bJ^F8Z@T1<!jr!|rEmo{Ysme^h)0>ZqN9mZ!@P|{L4BXt=
zb_}BUP-ETXH|&pV=K)GKxq(iJEwZ#3a$YM3CPphchg2~YOWqST?cr<eCaB7<T$+xa
zUKSP>4yqTHNzv0I2~}z<fY7REe(Bg-)uEwQpbY7(pg#?cUITMsy9@;n5Nb?E+F#X+
zP&%Fo#a6#4GVSs0&zG{gXxk8yADn7y!SY1A#96uDPYle=K;mix@vRJA<T$vvYc5sz
zTWaR)pES?i34axh<m8*qU@6Cj6GQ9Jw$NqWfzaY%wmP~ejNm81#%))xCd$N=T~RTz
z(K(d<<Dbyh8wL_kB|`-2cs$FtB<zSwFF?B0RLT}&h>3+YNhNKo9`Ciw04B>%$^5gs
z&fPVhhYfi+drkcNSxi2jtOXmDVcn+sg0`-%<O-m6K}V0(a^;Vpx|}XJQknL5r#<do
zm&pEkxSVcgXTq8A)0>w<<oJpfuj$mpt&K5}(VRVd&*t)ADRb0N?87s;<htr-JPb?*
zsVO0Q+xv4yxn;az3DcDdjt=az*Y|&%2zfpXA_?$HlSzq=#!|?5Iq#pIstd$61sOLF
zTf>%EN5ym!?~T;K9^@<I&O+V<5d~X!2Yb!z=GJ%95kc2d*GgjKaI!7*5nPGwzbo~~
zNJ)u%J~YvL*(5Pi#|oD71x;a|)$3Qo_gkeEF5_<-FF${E=%^>&6x^X!?Bq_RFj(Q{
z2zv~m9P4AHx&NBig^x#)OMQpk&Kst3QlNjx@lO!;Dl0e67F!XF`nc9RzA^amWf#Zq
z3O(UWSL-<?nv2sGin)3F4cxisyvZcjesP?YB;Id)8P3l(zWIwZ)0y|gD%m$*(LQc5
zt43mDYziBg{^RqOrQ@);WIsE8M>Wl3(oCWIcS57+8JC!yc>;|n@7KsKhO?02ziOVH
zd*TKV-<4){!NOQ>^zy_051LJI_3fKIDD3)C<&!~4%m54<v+Z-K{u0%%F7}kAz39GA
zNSp*P{I-*E1CEby1JJ0&4<x48tqxI8*#sfT6BQqxRmYd-SS40cnRgGng?4_QD%Tri
zUPD5tg{z$w+Xa6%e6e{-Er1|pdS~m&)lZ#FSQDuOitQYR2xs!BZmS)<vU~0bq)XOa
z8R9czpRoIC_M7rY4~5oUymw^`(>k&B=uN^J`H!p=b9dtP>fPqQk9t-U|Md6wN7Y_M
zuDDf)os5k{;fx5<V<Pp|)puOKRO>-68>?|(K3QP|r8mcps|5*8`eFaQE$B!2{SC8l
zaODY!_zerc^{x1^3{8_|-o_8}WkOl~F1(9%Ds$<OXX#?;d<2bhC76GDBjol^uJF9n
zzTWxoK$*w^Rx=aP#yoU>dE`1NFP|pr11aJj6im}lNg>dIiNsJv*nc4#yB~aTb6`A&
z@7sM|8$QZn>bZc^ChVi&2Q@|?Qrte&m&G<<iCI`IgZO^KO%Q7FL~HxC!FPCaz6V4t
z&@2vG-8w{Oq~o)tpI5n9LSzZZpC_xH@;be?s;=^Tu<s_v_7hVK?6JOydu&>$@5jT#
zKO-z*2}lUM5gK~^aQ1xcej`S2O_ThB%_}Whxs%Q~Q$AN~m*cf`rXCl{4qiOM((#R_
zwg2dX`RJIhCpo@p>o&&!muco`D&n-ANT>JK)yoLT*|C#SK3-|5f{94^6an3hXZl)N
z71qOS5|4pSY76vuF!W{MM~jwh)SvDYd(MPJZc9$++xqIOypNn<;vZezNUW#S(bDqR
zUG4?Z#Ov31nx8(g%lWMk`mKh4Wpr+nuoS<-m8a)mFW&p;{-Ndoi>UkN1hjPEr2#pM
zs&6Cp{`&pJKYO&Ys?81^um64?peiBP<@ym;Sh{GmZe38jZ(J`<LXLxMAF~|!*)jRO
zWeP>}HO$O6IZdF<etd-0l_Kgl?aVHwdYj+tPxO><ipSjLrTI)!prfE1*Hu^N24)i=
z3QbCR@bLhj1i(0`nf$8Re|S778?!Q6{sWmDk&%(`v=gCtcXM;|@Zg7x=)V4b3X~Ry
z{ge;|yRN4?0MtNzKNqc|=@23(+tyNUP*54O;jnw~!R@93AA{xS9#sdT6>3#Lc5KNb
zd-ExkxC2|ZaryZ4Cw|}29t>Zhiy!0m+<Ij-P`YJ~lj)R{;RSs6WsOeCfA%-fbl{Qz
z?XOyP(zI*9nmZIDO-&86k5!*eZ1}8x{W8QC{xUUH^QRL3Nk=F|rNj!%d`a;QpfD4Y
zq`%@bvp1}F|NgF`nRXc|=f=@edHEV*7%u9uw<jukLeI}Vpl-!$UPzw|D0{xXZ!nGU
z?D|ooKqWeb<S1WW?4sM8s-tNB=0NLZ`axw<;bXL%iIMSVYpeZOiR@Q(=vb%_$by1L
z2)SQXAS)FoX&1`;<F)G^6BYGuqkI}LJ|gNv;B4bX8>D~PWxR-gbjnrF@ns->NGxvU
zo9g~R_Fa5}{aTK*{?8|fL1(4Q*;zBhP^#74V!Q}(nq-d=SBjDAcx0Tcwr;J`O6s*M
z(wEWQ-VS-8k&0iEK-LdhjP~2_BU3KsI@q@ez7huqrzj@(rTcg#p^{AJ5Th8A$oEvz
z^Iz8Z#I{5bq*7q=Qk_bF7H`69q)Sd0B`Z)_d1OV%PqE2MiHV7cG|+ZLnd3hx->Rb-
z%Gv|f`Ewx}uU$nRmF0R6qykCCW7%f{zF+AKM-YTF)O#+0<p;XAD2a<dN%l~DKp+yJ
ze6iGz(5<rF1;H1~u{~y(AehJdf$x?uJDO@LUrqT4($S@z8Unf?6G8nZEbQ*pNS#m@
zozPIVJsj(=#Y`?$X4*FhAwNQi`lS8^!FY+?(=6=V!Q;Ls?4!h4b>rolcO$SPGGyLn
zjM1y#e@5|wtuTnByuLtAU(GQC5XUe5pguu8d{`AlSGT@y6WvBeMWwT`w}s5Wsw16I
z2;+<)oZs(zN0Ca@w>83on!TZ+JG!qL{AIq#f2Fu;823gY&Nba4_Kj5BUV4RZf0lJ&
z=gXzp)mV+f2Pk&3CsE-fVf0iCZL_sh7W|=oF~`@ox;a`}s_rCW1Y`Oi>elNWC$SOE
zbQcA^Uub?sS9PQ3igzW)(WoUzG8`747ojvQ4_;U0FoxKHVs2+Hr^QGmcQBS(091%!
zElou>D~jaZJKPGO+&YlXRA!P?aryi16t?xAk{b@xh*<SL81O*r$m;wXbP6YR`#8I6
z#M2ksi&&3RK60<LA)+jFxNoy`ar)T;|JsKQ(vUdt($3nfUGi{Hk7|ta_-BK=)$NKP
zx0KLl1jxj4s${IDs#@N-iHLyNP~=FU8F#-ny+7sYHmf>qfgYQgY0&|3sm9{Xn6MiL
z+Czi3S9AOF=w<CLPPcb<Ao0rXqJ;EPqg82@$2OE64@LL5T{ei!yS)YNenaqX`|X`J
z2ugKY=f8wQ*td2LqM&Udec8ZEsajDk>7S$Nlj*NNaVms=`r-x#YO(pmb1zT2b5tql
zWT4heM`_!nwaa2n4CXQr2eO&$7K|k}jUG>lS308H;UcW<v{U<{i5W~h3IyB?Jta};
z9*G+LB|DAlESl$aE|;}<;{slu9e<^iPQ4$O&JBTo=RlZ?QAgCUXi7YKM1_OB!;OtR
z3(@!+&}pXDN2b*82RvZd^ab9PGcEaD0MnP9oX1@rPQpwm>qjk2A1<B=9JENr3ed8%
zvv0neHuyt@AjxQLR3F6a%U#n6$g)K3Dsu&kFfuZ-va*6H4PcU{#?WhMfyKG9=W+;|
ztex=1PBEl};>2hckC^yX*vk7C*d@N-jjg{F)S+&7gA*DlsK4Au*K=_`E$V5%K2nG@
zATAV#o0u17wW(_HYzNx4<RQW!jqC9AxAko1t;Ikr{LO08lT}}c(@FmP$tvLYEjeX>
z{#Lc$*|sq3tLt@>3(}wXm&NbkArOyI2a8p7+wO?*yQhb-6MI*BxHr;%JS3=v@H-G-
z0FX#RNy!iN0}e!=RclMUzPCw`-?#7{R8T}_HJIT^{-}au1C5n5lY6y#7i}*eaaa=}
zHe<zW#S1ay$%2Aw9)<EQQJ+aOlZaYZe#_tLD3XiysJ=`W*-LVR5P5;T$P)CqMnP6$
zOTCHeIp@w#y?<RH7pc>&#qr$3i3nb}#kI9YK8^bME(_kLYD{R>b+XJ%%eeGAYr*Pn
z<~H*jYA3wpVVO-Vg%?)CWxuhku+5$9KHl#jz3pKz@t!W7>7S6k;XSc@jf>;80&pDe
z9vryW`L$ESU5)gcj3m2W4|rs-_NQyBZe#8uk^T9_4mK=t)h;Xl;HU)H2>H0VxrKz#
zQx5FIVEQ$;o*CF|a-8U8nDzmcrdjb(r2Eu2hd=%jX~@zrk!;SkpqI&+=ld#(5D|n7
z`-$J7%bbU?<bq+?-s@3SS~&`veF9#=-DP2H1P7DsdWg&NUq4&Y@k@LbDT&wI{?@F<
z-)3!Y9pxV5`x&myS)nhc)eys{M{w~%%EhJRmA=oO&jA4Rmx3C`N$GY%54!}>xZf*9
zlhS8^OImYK(s(YXaV&91S)pxLjRwn8wbcoRd7OnL>GAZ%@+e8(_J6_4lsy(02otl<
z%=@L{f&>#)R0obW=i?&1s4#|J?L`d4bN5&jsYujmv|N3KDYnwS+6u2`qnqRWN=D4*
zjaNkCNo5GKvO=ovJ^fAh)bgP+NlXx4kj0JT?afKjxKlh*#>O}EN{NK%Y{;W%p&nt!
z^NDwObU@uE_tgbNN>h9tTy3v^T^(TfL2zgyHFaE3IF-o?R96s68?5++>SQ4X-1<|d
zLI*dWul=-rgaI6Ia*8+pSV`cHC5w5@2&km~wdO(k`bx_*R%@M${sG^VN5;Zm)GLn(
zBLTazZvrov*-=-ri>7S@CB0Z`*`{pp39me?J^j+1(HzJcC(5ojL`b=Mx~hDW&IA$>
zpp6K9_YTf*Sh7FR^!&FV8`cu_{=Gx>RvE0X)3sji&!10#lW^<U!FeUD5R`}dJKH8b
z#~4V>xz%~f8jH6a^q+bK_M=6nN9=owa2Ci^y`P`D`T&T2nBTeqxuzrN`f`RutK4&~
zd>W*mYxJ^r4Bj(P(P>ER5;;N+kM_e`1-&ktqhYFFZHn^ZO>h;Gq*U(7YDhQ{;9vQO
zsWl|bmzU5Pes+o}Fm#b}ho*r8<4W<y!8B2+B9f1T7;Ju7@;emqH*B|Y*@xiAZ+$po
z2DCQgQFUgeJNV`aVTB;ko-8z$0ni_ea9U-@!o1Jd{^<D!PbC}Ik%uwBBF4|Jl{=_W
z5DbsfSoT1V<OX&+<qHSD;+I(cdHWyn>odplyzLa@v$J{kB$~8&-_A>DF}^ijb6iin
z&SLgaguwV-L*ZTG6afMIshWS?zJFUNG!epQ^;1h-D4jC-OjTGa85m4|^Mx56KIZq`
z9eke(+@}WRHgFq&RzYPcM{{3y8C=vN^S=csMXxGS-}d>lMJ#gst1{NOkp3!COWz?D
z_wCu4<`m&Jv4!OWLcc@9&@4jENQ^5{^cW+TD`S$K`R`HOxzk{PM|x{7BOPhD)4@9@
zBoDHft-^9gq#2Jp^D~oZh)OTny|#$CIpzGUI*r`)`>YBl58lRQm0sCs<?!Xs&s+)f
zOL@O{40^n{Ze|1`PfYs?at$UV4^IxsbZ0>m|NNX%;uowAVPRpQG6VJhN8-73Gp*%Z
zz=o69!76T^#{zckG0_Lui7gU8GfNBV|4;`E%c|n^_)%Qft?+&@X(gQdw?2(Ykgk;b
zS(Hg)U8z}D%dXzW*ikPTHkpiC7ehQNc}F4^Zknwi;-9POg>#I{jqfg^iOJE<iF~Yf
z5Mk$D*O)G&6-_sMw1mM)hjTRK1qZ4;le)?jilzKjebMX8?9RC9R!!sSjgU04C;j-L
zsxZJ7pB(K;*t5DgF3<D_;{4;@kWqUkZa_`s7vlIqzMq^e=+$j3<ndG;2eZ+S9y*EA
ztj-BXPg6O$b`ZUz%~kRr8UzZqs}TZW@?)!Ji7lOA8ibli27I;P%vm`%XiHJ}yBX7P
zR&6VEQLaDJDp%it6R!T?L1lW(y-+^tIR$r34C24W7B`t#ACbzo*SXzV9mYkNP&VvL
zP0cs0ZX~8L5h@guWv0$-L&~iBO^b<3zq2%|-Mc)Tm=Qd(7Bw=$nA?xA#M(m$Z=bsh
z_v{`r&hAWKq^>LDHGr^QPbe8@{$A5c^!)zT8KnJwd>8*s27vFQx%Q`cc+2jrOA(}I
z%(HiQmUV-I&(Xtw3EwAJ?%W2Cu0ZsBge%`jr~MkDb(1&lRrvd})8tp9m+zBgQYYK3
zzP%wpcoS#UmGqN$4gdfTWigS<O!J}z?!YFu5=ze;F%?@b&op98<cm*5B{+lPdXV^u
zq8c|P;cfX3Bp2NSW+CY%P<9+09YJb14x=zW;^1(RM7Vme(vN%YK9%F_hKu0sUJNYu
zeS(EZUw{S^)9Vv|6^y>9$k1!B(PYN!J&R8spH|`mKKUQ1PBJwS)D77VB|j3%(JB0B
z2YV8)4@7UCot<^i521xCp0#yANV#+AU=au#q#ZoaTDI0s!$pcmYwQMdc&kn~i;|;E
zo9?d*M^SG)jPkSUhn{JSg{#(2J$g;LpDW?T#Zk9kJKu-sXfF57=e2sG0BVENU~03*
zvv#Vxd&{lf2izA!GLkR-Xitx(qpU-o)^f#>IQ_@vjesRb?qRQoIf~p9wL2KnK{hPH
z9J@N(7LC#1bobr#uKu5&t|BkO=6XcDzI=-)yM@;Z)@-*c1layKH%nvOvb+hR=pmZe
zIYovK--hh8sknbizqG9BM5r8|3P>dn4-dx)8Nh4q@gt5E8j+r=(pYaA^@#Lrn7SHU
zm3VZTGnnF~AJjz$PPfT`o00sQ{uW2XG}rSlrtkE(tFpDCkA**E@}UG>d7T;>{wdXf
z0)wi!&i2ex;j|?B8P7DV;z!1iK=Ivu%ACE|ZLlmH2Ge~%1o@5<1iLusbTz8m?*>^R
z`^(k>iDmAqtB;p|o%zT+dWf3#u&BBd;x6;J^n43Qa(XASCBj>I8VxGlK76||2H8dq
zuVN>f!)Yn*WL{V?Q@xDdt6JQgx#$@q+&(|~<o5uN?j2A)6eMxQ?vG2#*M21*={T<P
z`>^u;GP?QgkGOmC>0v45&|@NNdY&us0RiG)F%fZ+Psi&!v(r8@x^i24KL+qym!UOu
zI$dq4iww8_@Y}_EG#RuXrFeR8^L;3cw_Y<3Ia2hE?3gF4Rc_Z`)CZ}(_eVzA{j&$(
zK6}LSlsA+SMIc0s1R3IaS)-=zbnK9$I8v0jf`aQgjHT6|Q>$#7?$<K2vFX)$SN2*^
zbD7Dj599}DJ0uv@P!a$#(rWOoj_A%g#=fHNMsk&^8L2bttFBJ7f~CG;=li*_P9tAW
zsa(#(qXO;uI;F6;4d;)0DnyWmp}p_d;RvIrl9_kpV)J_2#Tz^P^m-pP;#2R8Ek_dY
zox1Xc{iY2P!X@z0l~%U`IxF4SenZdd0oSonj+PA7ZcN*+htws_Y&XMW=mR=&kQh{>
zs9>@x5$-Qek_6^bAx_f=YUEr9|2?NV8Xs%N^1)`@rHitttsYtYuQELNxPuqJgNN7l
zKt^syZZ3m2Nr>7#xRxoWnnSy*QJ-|}QsO-7+gpyu*A~C4@lt(wThb!0jf(m9dSOXd
z(21Y9ffy;l4<54d7jEln)((lp3S3Afi4M>|AC_L*vo<FPr8%;(JL!5u<7Eqh98D7M
zo;Pp=sazAke2$S>V7VSOx{~5eIs0WJZou3|m7VbdCpkrCElfOWIl#M$cKY^A!s`%{
zUJ}G$nB2*uAaC#PGttrRciE*yL+>m~CG#1ciK?z_l7J{0t=xkEK~MnMl{Ya+EW3U0
z$ylXAdyXRSGRPWH{ygsv#O#nedhy;buu9v*Vdp1XfW+VKBHWnKA0HInyXSkWX{c~>
zJXelplw4!CE|JWQ<$0FA7kzf>PA#;1;<Y6*p0@ExxD8BnM_qzjF@ByJ8bc5>ZOWOz
zsQ7Ut^9hAgtwE60J=E^9U92v7i0R_{V^sqVWZAZFp<FLzDsAIiu7GzrO+Vw?FNB_Q
z`tmn+OM1nU-YP$r-1Q8}!j8M{FQ#F3w2w>5>GIewB>f>Tp04Yrko#&y-ru{=SbLAJ
z>LG}B96DGB`$72HD~qi9U5VNhzrTx}S|m9yw3Z`r0X6Kn=&*O&oVvy5aR?<GJ4Q?b
z)rZ*Fy|aO~os<1b{%50=Mr%P`o%5ED9+-0XNei3ysA;fhI$e(sDC{&g+zE;LLK=71
z*H<ItEYC6{j;;~emLBdJcHew`l`K_*G&51%9a|0phETh|iUCuAD!h99j2BvlkNQ0k
zEZU87v&}xQL{p9;Y6oZ-5wTHF%dEGg_>75ZeqO8u`2%(@nQf1nG}hy53=gxDf|>n;
zdOW-4$N92x`3joZpV$kh7`{%Wmy7+Z{ISaA$ZX0r<JmaFS0ynmN<6tcc<N%b3jTO<
zO*{9i-F(jWhM)ggWQ(TpfkHO-z!R{`DvRDbDr68_4M(zk#(xJsKsf~kum!F_;z2Mj
z?Sh_1v-+(X!*(f}s)8>$eDc??Ux(BH`GVnhRD7FmPe`!*pXHZ5%g;>8{LMIK0mM_g
zQP)GC<B7t%7#dH&uG@_GwcBK&%(Caa#!;biahV9)ip!Xp<4Z!SYvNKdm{-*$m9NG2
zb$#Eun67y%TuoQk0+zfdn^#%rLgz+)L|DsQ8FZL%)^U$G-`)9_39&}=&MAri9<Py~
z6gg7~@jHPqpDpASpZqML+F(d94MA-@L2X!$w>gR~W>7-lUf{Fe^{3xWK1wUI@FOEh
z$>V!<Ck;kvHX7|_nR!DwzVyE?{iBZP`NgFD6*E(nJSJ7WP0loJw8*aULU{TC2EwcN
z``7ZPUGNRYyx0UOufFE5tLMJRc0=MG3(+mdzn`_13wt!Fh$d4XhNiz(m5M3fr&FPK
zOWBzm27Td3Y-}+rN6E&rP9U>Y>6itliUx}oYeD@skmrvAunjva37T&wQGZrf(cscT
zl?-B%bJ>=@_QaBbBMiub?(_fZWaH{DzUp5Lf$}qEDo*0$dmG+&m@c!Eh|6AjiP!!L
zbD*T<F+#5$VK*AZhN2BXYQ*Ghw`~ixGz#?c`&Z!1kzT-6@b&c-6}zTty^M{GAtz`e
zNkIL;;2ofKt|c=1T~Hd-L8co(&kzuT-;}T5{^TAnJG(uw0&sW;WNV`RAJa_`=o5L~
zJQx44T-5RIg}uT|G_g`TYo^F$l-c(WrWq={H;!_qapWm+JYH^NUJX*C=nE7#yU)xc
z$Y;6kq3~E(J8g#hW~8a9qWGo7mEEx0qFlD#9Th&Km4vr%E*>8$y%@=;r#Q4ogM<DU
zM2pe-^hvvdg+nU?ZD57V>%IRkF`+|6xq>`OUAPeJ?o*J#1eSg9`T17c5@^R|nvQ;B
zhXH5AtENHlAc-jqJCFRkR)Z0vS+Q<`+G9@J;($(k_k%Q#Nele&H<_1SGI=O(rz-oa
zUdFHcWis|L%BkOCb*fx{TrrsmTgijQ1B0Ku!p~kQx282?cnvTg8HmgTA1ss>`0unq
zmIsOiBV%J|ya%I%Z$7y<WK@{<tTZPz)d1l9>voRwcS0a?{p+eAh#NId-nxdoBlleI
zjejY8{IJH8VbQ^VP;L|z(<wb_lde*KI{z-K)E*1*TUx|zX?cZ>qv}Sr@jY-5pO?zC
zc*^I!B#HkbUI(XDxMY8jsVTOSBei3fnvs@)cGJ$ZZ~f=^hU3epeWd1}4wvV9zJ~-Y
zUGC%;WyNa-lMwIDm)p~;Q!n)0Hm|)o1E?c;BVKVG=mXA>$eUIJx3@%VX6pTTkEh=}
zRRnz~q==yp?1Qd$!>Fsfn=h#oY=LfPn-=T6f`)0W67QfczE*8DnB(d!I>=Npvneck
zvd*>S>5$3(bL!V`@1@d&WK*w*#>1JIdxB+Fm*M-DtDq06ikUR)UHL{MFp;Cjfqb|{
zG=t61p}cv%kAr9=Ffpax$459_S`_By#2c|5^~+^4k%rErh*~3Ab)xQ2VKr1XB}rZF
z-4JxIE4|6v4F}3d>DETp@6Ho%5!idoN&5~o&=C{fR&m0*q_QoBfq4Jj8(vwvu*g<P
z>A_*sttv6u&UeyU2JM|O+{0Gbn^<~TvX;K;%=iS?yP;4GQ&90(eIB&x5Y;7XaeR~P
zOC_M*;#jagBD`EnjF#;C7io)=-G3EWq<vBCW*b-%EqgWG{u#<T?CrY<DqJF5hBq;t
za&PWjmHMvLbz69XlRhY2%)B&b;ATXc@Lx+62GL|W(pTY>kch^DYn<lZq|JwSbZ5!K
zg#q_F%Bx!3dC6Y??0CG8fSg3)ID&dD?;j%Of#Kocwi(54`uj2wrax__)4cw8YxhOE
zx8Rq7m+al1GHpeXSU6oTP6(-_Fc5|}GY<aE;%!AHyk`|AvW<;r?aLozTEhu1<6~A6
znlxb*>g3ZtWRk6V9~b9(N(rz{8X;^-GP1C_Ajn5joy)zt^;rx>0fu<~ZJWy01F_y8
zv#&p|D;Kq~y0zYGj9xTK?-tsn$Xq~|``PEBGimbS7qg{r3j-jxubPYudY@x=PnoeW
ziXKiy9S<*ur2G2qnVu<yo432GAMC~s!;+-(kEPuTn-<|XnRLI)$!<wD(OyguxBmDt
z9-afk74!&uN0a6?0H?~B6pwxyExyNDbtI!Cs@oRnUT-PCsILr>Nt&N&_@UJRwhT?>
zoL@g$DmEN4cLsD7v~Yf_6Gv`4$vqneX)?=GJP_jt4cQAL8lR_^1C(bO_cC4Kpst*q
z<w(-aS$`4<S88%_gyqbChad>$Td#uA4my4m+Ai8^m<FK&54&XSKQP13(ddh4ysgbn
z(d+oZOrdu5hV2&IIqJ;A6uM54w0Ii1RB=Vj`9looneJO`!Q*kKa8Z-pfgoa@Ue?LW
z2-3q;zvi)SX{nIuK41L#CTk@Mn`g`N$5Oslp%KWhrXrf2LR2d7ZF_EHY}#;iD2TXM
zHkh9Cz!@>1TY4dgAAbt6C(uOalxH{$!bwd)+ik%w6am|EgUifAF!b4F^kKEx5xLmz
zKY0O^ZGv>ze%Q)E)c}7Y3lO11TE&iIogs+1fFpyCWB}8`k9#tbEZmSaH^y1f3$8|(
zD(Gvq$+SX*0$wEn(IhDNZ!N7o;)|m1Oc9+n;Y4#g5N$n0^o0|&pXhT0!M%-P0DKh0
z<nFpjmb-$l7?{iuLn|rIC+T~OSAaza#D8qV&Cl-w-bNv6t>X<VX?~-@_UI4t>(}A3
zDf1P8A<HLlO2wJ#E9P~h56|7$pkyxgyYP)Og#{j9^L@CcK22ul@87#SJCG>*6WB!W
z{d7z^dSPK7P_%<i%UpY?wO@zvnSR$cux$`{k_C2Nu!JcqzmJb!LVqZ0Glxjh2zd1Y
zCG>^;4I&lMaRFeO_Yo)*=@@=x2cfo*>BjQ^i&1NDDVqNK_Yd-&(Ozs2CwdCTjTb%3
z2T>yPXid}q_i+wj4+C-!5zFXxpkA6s-yQosHwdf)gF3W8yb1D+jg2Cov{Dw=f~Q_w
zTt%2-!FaR0;lP}s`V|&iF4$FojDj?Z_VHPD`LF5_h1hq*aLZey))f5&!On_}gF^!`
z4HxHU@Fn*D{)l@s6i-S;N%<HrLC~E|N`*PLy_IyW*xTo=UcN?Bry25H3ne0ATL`Ui
z*KMejoWU#&@f2QsG@1kHy8+$i9&J$!(iwk&f7D=(txFl{>pgZd0_Q2@(}^C=1VZ*8
zGzmQiq-#muwKBI!Nu%T9a1b!NfOQG|?jicZVLt@&G2k)R+=_T&rQ+FG&h;@dv7O_0
z7<=9Fy8W?-QHLbFYF&6TEB#+E8mvmei!=XbPrJ_dK^YiAK?F5kY_0^u$7^?a79NHg
zoCJ(!Pw(C1y252OkX=1(*6$258qfXpwj!JS$M&)%bsuM8JA#00H^5dPtVgKZePe7e
zo>Yf%46uh|8zWE;gDSSj>%a~yk8f{~($H*nz(;yClYw;Eu};JV)aA=;|1-IX2*ANV
z=%@ulJv*XV;@xU0SaN|Z^Af}l^2Yw(D?<tw^x&}91Et%Y$U}gTv+D`L!U6$9kWhdI
zMIbG@0df*Sb6))61DvJzhd*TxW-FwJ!%v=8AtodOO+xt!M%4#y>ssYo#i7a@b8Am%
zQ@s=bK7&AYD976!Cm=i#9Swktz;Flr6W#=z%~t`gTu?ItmVXA3B}flsPJmcCx~&A>
zELzIHwB)`%;s|pN0<u<CR)B~HeHo<RZVnful<VJ+&<hNvIf{#ki2>v};szHE0^sBR
z*7P(~eCKB;d=@<jLIa-)E(;cBW-F*%y2cN{Mf)i#N)^tivKn9tmgt2q20^-DbBBk`
z!_172K;Egcme%qlMeuufc$RfinW_qKBrw2Gm;fNj<7wA<3n)zAG=_j9GAV?qQUQo8
z?KmkF)aXT7YH9<x<wPOc&EI(U*YNJzusy)+;8)hjb)9K=*>ydy%x*F<C51h7PD%7^
z04%Lgz7~N0*{U?L^AEI9jF7y|z>u$9EKSw|rcJp0-pU3#iC>q!c7K(-z{3DjI?Se$
zQ5(oL)k4RcLSV|Y&w(%uYls+El9TfUpXZBDqHq&11~Ci(gAc?Zyc6`1-!UD=ADq;@
z(=AF&M6|!ZCid&qOuE^xVaUbg7Zf~!F7d#P7H?34A3cs-srQ#73;I`F(wRbnf;<FC
rf~;8nXY-itA^d+v*8kaQ&Tk8(n)yY)d+7pavLmumijv<X3<CZaLvLfi

diff --git a/vignettes/getting_started/unnamed-chunk-9-1.png b/vignettes/getting_started/unnamed-chunk-9-1.png
index d31eac9bcf48f4e6b7267ad09353d096b0c8bcdf..df21a321026e7b4ead2ee93be92580e3b0df555c 100644
GIT binary patch
literal 24497
zcmdqJc{tW>zb~wLlqQieLrM|~nN?({Pzo8kl%dQr&s2&cWC$gcA!RJ{tO!ZSoGwGi
zJkN3&-p}t|`+eX2{;~GE_CD5H$6ot*?&Eo$y1B03aGu}qXZoH`E-TAZZ)ey}MMXue
zs35CKMYWEPifSX>*3I~v)Vi?}{Lf~+i}JEmYs7zE7bOH!Q5~XEls$XR@zrRD)Aeg}
zJ<^j4{ATOVZ8+w;mEpwWd81smcTc!N8ID?~wb)g=es3C49o9EAz7b~e;ze^!s`_p9
z6MYw6>TSCouJt7Mb?c*zXZLH;?xycwUL8!5u76=ud;B=7Q%TgF)tj|<#5oEj2W)Ep
zK9vjN+$JllCu*S4#c_x|ePfyL<A+}++1c3y3Lh&d?BrwLyZ2xDD+ArG4Y?WtERuKo
zzCB-J*}SL2@aKnH*NdF(+-PFGHq((bQgzPS+uOU*cs?EvijLOplF6k}VSMsG{?vcl
zcmHSnR9ou;mF#{N7XNqej<%iL?72=&P0fK+US59m_;J~@XV;xPb;|1RwBa=kjn?|d
zPgo>&^6hx&?XBNk`ba(NikjM^Q+H|)9X`Bi+qP|w0|MmFpMT^hr!f1w*H~ZD>h9f7
z903v!2BJM!y4)$esiYzw!|KpCPBZ;;l7j4P{Dll#X=rG^M@zHbsf#LKoa!BM5s{^4
zDNDLvKV*A7$AVev@1K1?$ur4OU-x@yY!GsqHZW_?-J_L~UF?0X>CjZkyw=^hVYOsZ
z(Nb-lg@whBA=|Axcgm)bihlR^zfFt`<dE0<JKYy9>0)nA{aoSt&NQuO5fQ%UeGeS$
zc-w4JpP<6}JxY?XzP|p4kzl%c%}b#jxQ(AhE;(xY(=#(NMn*?p3Yl(JQBe^Q742S{
z8~IvYEn{MGY;}1?CFz$&b!R6_cXzi@<6CN1H@7QVT0T4)sfR>G54NV>$jBedHYc--
zh|o((N!_)v@yy6LWzkvqBq(U_emc6?xVVjZd3lR7b<$S771R-<!JeKQ&dJE^p{L)<
z$HzBlD;5$O%Ff3}6UL`o(Ou^4bU5z-f8p?TJ3FDEprB_VArC)$Zt1P&E4fx!nrqU0
zjFXe|bz<VT<l+Yp9`M`rU)p)_<i@jS&!!vl>Ezx^(as^BLZZx<`uaU-h0{J&W5&kD
zKlAN)s6LU&jC>`_C!?JD55`h>6!*?gbOqq`un7sNsjcPAjgjfDzP<GM#~T^k=bL;I
ziNvBDE#-GLLHX9NOyk>g#+H@=S7POE^?Z4<b;k}*Utiy<^uE465y#0xQj5Pge5tIA
zXr`%p#t<ILt9^lrVb-+vm1sKU*~^!S`6<7D-+ce>ox+sO!P6G6hJ__1CF9ddR<#mS
zQ-vurB~CMMXMdTdDPm*!tmNk9Rb^-0D%-s6o}k{Bt?xg4xU9B@W!f-UAFDQ&fUnIo
zZsf{XsHeACUFq%ViMtqfLbIt34=!FgQmmOW@gXlQH&-OWwN^N!bTa+7VfEztL**3}
z$J|dyuPtXJW;@Rf8&0vdbl8vmI-FNfP(2^dRrppp^4P(hn=_VDQ&U6P+1U*!*%=wZ
zd-v^2X)9-Fa?RegYgdCetGkh%?%~6SRc&o;i`H096`Carc+pV};lk3*8Fp@dkF}O=
zE;0R;Lse1nNzE$n-o1M{Q`AkZ91RT(Uoh+B#3v+Fv#cmZOLfH9D_xnfoc%ixO-`0B
zBJokv^S|!&2s^)vuhvOucalb0#jh+=C5u)gL2R1J`7yGe#lrOTIagPSm%`>I>C~Q|
z?|u0q%Ovg2^!)kr!5?onM7d5YV%;9(U0du8(60(Q94_rq>?A(pvAnXvo4AGQ^XJd>
zf=0XN$2&}IDSxVhm0D6#Q!B8rtcIJo_a8V=isiz>!m@Si)-wspk-K;AzB%vMUmNMW
zZ6DXy(5l+niq#cLU~upTY~h_uBD<NzZ27g+lbZD!-zsgTrarH(9@^H+c2Ax}{pTys
z(W44e8Jf9w-&8G46jM^oI`k|36>$Lz%{zAP3|U?5r3s#!buMa3y&>-FUpTH#BAqZN
zXDd06q~yc}NUvW1R$Hr{JcDP$IM&+O*k}+r(wfFo%PnNtsly=YoH;v`5Pf0OrcFVd
zRjf`<E-n}2<Kwv~$#`-7H4#VJ$mG^{3H3&C2T$GpS}W;3RZcgzN=fc--W^a#PdZR&
z!cj=2fP$Ak=DswPAUE=NY3Tx1G1rLU?b{cy=D8?yQ&WK`2udxwITn%i&L(9elig*i
zXU?4Q%g5?y81E=B9Ep^0YA`B^6n7BTyfgMI>$Y3Jva+(OnOU;gi|0HV-~4UH+w;=r
z==exx>8Yf&FznRf6$h=M=G|q!4nI#H`uq29I+@jZkS%tivZA82t*y<V=+O?kZ`qNk
zSSnI~zx<iUPOHl1J96a678Qx9ji}3}g}Nv@#QW%6-NVFWHoI+uxOihpc7W8<Q6?s)
zYXNC7F+H_oOch3g3=%7|ja)gpg^qi8)RUi$^^WCq2|GoHsV9|(3+Ojdhnn1U6ha+8
zoK9BS^tk<(o^ugP#A!?QQE_f=#q6%h$-CvkPc<`*_H5s=!_T6?evCo+mB_91RQ04s
z1qB85&Tk)jd9kAnoVxw}jI8WNtxTg^tFI>-vP?DFgoQ0TlXFjKD=KdDXBNM0P5E?x
z122oNe#JJ*X4#?5)!I202i;0E(nwq*?YUMPYzAu8M#*(`D)uh;AiR+y<@V$1Xf#R@
z!d0PF-@nTm8NI6-!^&KE)HNpl2fuEfsk|`WQD5Y;ATCWuOUouH$viYPq;T<~jJ5Si
ze^%*To}XXm(pX3HQrQmv`244Fzr)P*bkLYgQnR(yg;N5B49{-YysWa|J9aF8Lefu;
zLt7=GQM2eSJJsUS(#so1BqeW58L1BKVZO>^nyg#Gf-;q#XeBZGLuRm};I#9?xZ6Mx
z)?;r{T5+j7tJ_1|!HeW1+dsmnRZr;!&NYi%va+>(7#0@R93U?wEMabT=vR(q+Ty4A
z_={oVF8H$Ms&D%)d^UOdajG*ZJ-s2Q`z8vuh)w^~^LF=3=I``>e<9*D%}K0f?bH)Z
zYSJ`&(GKfNJUq%zqLgPz@80)kob}ZC8qYct<&H~Pl14ucaUMCMeQixxUS58g(zR9*
z!hJQ_q*5!?wb+jt{nfGi5uHH}<Ic@-adGVn?ze8;iv0fkXo~x-r<{sDhR3P-sB&HZ
zo?Kd(2nY|SA8pI1OnlB5dR$X)_Ui)L;DH?DT^FUgD*SCo)Sup6iTy#m=9Apq+;Ri5
zoQkD$Bdy(IW9FG8d;9FsNQz0Mv`6Q#u!$$9;?F#rBlhF%Pq>s`zE+8r8f=K)aqir?
zAK3LM;G5mu-F>=$4-Gx}`c+}{S5{I5<-*KmUAMXCN>UNkVSL&ha{6VHiFer8*sf>X
z_R|y1J1Ota^!ciaq@k^LFqhI~e}tu$-9rwa>1k^;#6xUsZz?F*o~XrJ@O~fRPJ-tO
z>_+T$9hpU~WwYyI<pX+F7P@j=e;5cv3Ae<u$#|iAYg!femOk3Ch)=s@(UoHvU1c5>
z6-6&*`>tHy*3K@)ihTd^g^L#h!^0o{{CT}CmQz2tztW#shD3_05<^+bGBJ~MnU}={
zC7Z;4)N^bWNKyAC?x;O)-_+Dp`odbfuztnPHu3r#b268BSB7=pg^aY@Xgm6yg-&0;
ze!c8^C{GF<T-JWw8U3A`J<WDM4-4yBh~iXCKmFH!q=lNVcriw^m+D?;k&8aoy@=}~
zzs`*tHK{5R=E<~x8$uR3xX?!l2?=ecKR0}$Oq8VMZnU+x7gkRr`Mo%GhoNJ2vW%t|
zU43ytdhH3e&C6E`4h|0TCGIPe0Upa61Ox=Ammc42PE<3hkJ*%M(c$ApneVP-_4v}9
zq+vMn^x&z@STS5VU6iej=IFKMU!Lq!a%VwhKe2wphL4U@dcocA>W%OPTXwQaG4)pX
z$>0eYR$ll<Ps#%3AakAevgpXC<tv&EN=&~Pdi<=84wF{4nX#=4Iotf>ojMi+L&I+T
z`qyvYN&#P9HxB(s);wi9+_a5Z+&;w3;=>!`Uzyw$0jwSaE+=fPt?$l{Y8n_Ayauc~
zZQ1z{MJX+f4;$0m+UX_v(#jFEDbI3WI-&`3jnsw<Y~U-LT)&H1?3z9|3-Hz=g@U!f
z!WFG_xEi4JosenE&Ecl^iI$!Kd}X)}tI6MI+e9<fQ?%ZwC8=jK-1h7l7}z2$EiEf2
z_wt4VZZ7w3_k*f2Jd5+*yO{Yc+C_p5QGtJ?=}OOzwAPKTR=M^ppC&7yt?Df=PK8T2
zokY9Sv<qOBCUCc!Lw4g@SYY6NKxNX$_Bh4h$$EL#=Fg86sIk$n=;%Dn>2%&PM$W0k
z?&Gx>Xw~&N50HsQrW?-eUhzuQX7d&rmdyM}FO#lfH$btgulP2sI{$E8njL!V?=L4O
z=hf_n`lC;oA3Ms!laZfN@i@V6PByKz_xp=eBDRAB3h<pMS?m0hxTpWvv16qVx9$6I
z^Yi*kkz%xf$6L2;yJAPTfVUqQ6!ak4WA&iR{ODO{=X|9K15^A_6adO@K<?Vri<*pA
zFJHdbQQ**<Q72V)f5VpV<1V<;#Z}70RMS<JTMcoFgN+FRiGPn5kph~Ut}Rbj^C?`q
z^rhI1La(F#L%}4f`<=|{>gw0p+SG|#U$@iHgx!$Yu{e@mNJ{f2rO6u_9=>ql!exBN
zg9qykZr^6-;Ym^sJ2N#oc?J~icW>{8NaZb9%h%M@-ehL(fAr`PcGM>{((hVrpI%Lv
z3liileIYPKD{DIdL%R9Twl=myhp4d1yu7`)08071Rf;IBsi8*qKFY^eSrc(OFf!5)
zEa!Jmxvy}Mu|{nDAx_SX056~$_~)**^*Li>Zj1KZ`+)zym;cV-8$T?bon1OWN9WC}
zo!zv+KW_iH`@F0VgPx;R?Z26L{%4TC|6lg|?|BjbFzMF%Sot@tBO@;K25<ZrPwS!P
zEfke_thr}eb``V8$!7oV=}`rRR`J}yr>onw#Ij!VGZ5b(1%=HB6^S{(i^V{~ob%tn
z-2XkV{r|YnB3{O8FVG@-)P$>Q^EqAar<PXHQmmF{jmwv`QVx9l`0=)10K*YKPDNEW
zng<NoiHXMmh}fm2OIj#CK0c-RVbdbky`5}P!a?j?wru%QUf!&zi%Oar?5^OtlTWn|
zusn56>Q!W<(>H}gfkK7n*yg2OT_x%i^wXmf5)vj^G|qSV3l(U^#S65g<CA=2#hvfo
zEseTHL1QKWot81Vqk|b-*$9;*B~yilN%?26b4R|NjDmtfN+vTSWA$(L4F1Be$3504
zU8eCjqD}_oIsQ_(E$Xo<0m5{rQ5Z1R|J5rd=m=59s(uN622o(<zg#;_X9w$do;b-^
z=KKBj$W;ma;$h+!tE;Nco;gF6n3R-a!NbqL`{2QYPzQwR4+^nk+06i_zrjisHXeC#
zC+gPS!1MGZgFV`j!&4!F$IBb#(iLvqVk6XmD_3q=F#@WH*bRqJ%b)k#s=`=xQY>;8
z{)Kv$!0qq!jDB7W9A0#q-(Oyp@{=p;v;2B&Ctv#YOP6-Gx3?cpWIS-dOW6F^Pe<9N
zaNM5Wi4!MK$+O>`I5J$Ng&*`e%of*dVmfwE(L@55@Ep9Yth%~7S!~0WT@Q0|9C-a6
zGwi^}mj}&_iC$63{xr#+^ulKNtKD#wI(}ZrrX^=UY|elOAN5jnggkSMqis7Y4SNK*
zqV$gx?bB^1brPb?9tUg*GzYXlmTFmt-wefXp1QDLftxjZf+u;(dW38f>74#iCefXs
z#(C?h-uDa+KK}OY8>N$vheu|lC1u3@4CkxaSzAH#U)MgqN}SBM8#$1kk&(i(Ux);n
zc9fSlW#ZN|Dgd2C>sK@X7DeuVi@iwfpAT8!MY8A5SA`1D?|he-m_Kvu#0fbpfbHA2
zKYQ`w2|#OXety2c+$Gsi$3tbN|3l5ppy*ba*Bh6GajqOb-C`!-nB=lO%^MUS0Zl4U
zJ#?&jaZ0}C#EW=nCV)yWnhKp}oV%op7uO$3Ar)!OJG4|*CUjOB4JF|RYN8Fd4-{UT
zoSgiGeq;r7^psV)xY%1p_V>dgO<OVMmK1Gokjjno^EB-TPX}HG4JI&&utodU6m{(`
z88(wlhK}wsH2jZf?c&k}d3k3O8riTNik#;jL_|a++ROPfJ?X0s)1M!0%M!ZA9#lVA
z%5(ZOo$mqOlle+X%`w3D5Z;7iVq@Rb8yS!xq#c0x@xf)mA|jx%x_ZZ+J$trp-~PbY
zcSq`t0xw7xuf!c{lkEW@>p>?Rot#ki_d#Sidi3ZS$dI*>;@+8=g7iB#M+9_1LZg$F
zm3<8`Mj%uaqi4^bKl<~>00gq^_KJwOcq|?y+6AqIq$H%P`|0T?_wV2T8wEkcdG;u%
zym<?2bCDb!$r}4S5PDi%9Oq6J33^@kzlUF*wmhMI2dxGb=8T%!^Qw~s@H$l|p>IvW
zRZIZ~Q&B<w@`RRo>(AG~g_@+IwN*;EkZEX-yX|=UL12Y1zdw7{C#vzZrGru{C@WVC
zHz$3>tt$k@{aiKo_xE?~d#-WGLgnJcGQ5g)R1i--fj;`|;gF-3%rC+gu^;_;|L@E|
zZ=LkoHA$}mQE_qo(VrjbPg^iK&knAG2oP*|Sn-^*vskuemp`6EH@@NaoRHzS{fmo>
z^d0Kz>Ru@6+S%@R@AAt9u$Wt;%##COzH}ZKh>~>C`}&N{!9BD<X+0IlC5KDW7l0eI
zWka&jWRp&?>obez3Q9_99qM=Qen_m&27(VZze6CWxg$z1IoCala_rgmJ~#KLK^~!P
z%=eCr@Qk(RU9~EF`%1z|b-1G-XSlAirba2dh?-GYfhvys>RSa`TH3^lyEFYt#*hnd
zuN6pBMl)0`EK<)?+(ue>jmv##gMx#DhgVgvURBC=SXr_wUn=va31e;q7BvZpmU7EK
zPr2=rkzrG`mh#ehN_`t681hao<yY^nTTK6{QeSA!vD7u|DrO!ZALpw68d<T6o}cRd
z`}dB+?-P!*$Uwt_aB9%#R2Zh~evs8;S%8U|S?#%lon1MWg#aF$T3>{q@dL0|0=s(V
z<m8;!x3#t|1^u0Z)Ov`Y|3{4QIM)lS*$biij@*QR>P{I$p(Hfm`j|)MZqRMbeq}1o
z3(70kMM>^qWwmUkx3GZ<?nBGN@3v%{MD}_5^h-nh#hO&eQIpYY%Uf2K=B^&kabIzM
z$}HZ}z}7;PUXa%_5Fna&l_gp57hcPy495vD_-)459LSNj?36$gJ{Ee){!d?TuP6A>
zYsFx$0}Gblzdh&KbMT<C8Y6i0f!1at!Nt*xI-n`N^zY!ad+6xySg}@QnzrtO7N62n
z&LFxoE&ua27h2S9+G^XOhE3!g%ba&BpvujQf1w8|C@P+Hcb7sl+p~ZF!w(<$V?8dX
z70th_deTdtb(X}-3JeYP^6}ZeanmNYW5+Jc?^jD|?kQU6EQ<}0G6|NhC@np=45FE9
zJG56vM`t%08dl!7mTlX&m*N5YLZuokHV4)xbV<DuC|=lbpRk^OkB|F5e!SIo?3Y@0
z?Y(sy_+1z8nn4Zx=%!Ckvh(bY$L-KvR#j!ERI)&ow3CLY_J#PwKY#vEvWINNaFeMO
zF2pAb8V8w0Knr^8>svO|m|$8s(bxA#MS5jFC@$BCJ#HE(XK}n>;`>iuiuunE>mXjr
zC@L!YEntCInYE>FqLG|^R+W#xKE*&H#Pc5G`H{2|f9(AZ2Q^+UuFZ_XX3vTeCj3r!
zeh?`%>1_$-koW(MKjqHMe7IFcywBjiF1LjnZApHmy2cwnCHWoG#RZ3$FBD7^pVHJA
zGKmGiQdahFN;kZiQ6WfZ1H!&OJ_h8>oSd+KRv4k6nWYyuG|MRl#i73|Ko#dAKlbra
zHH(GfAT$hpkDHHACA(I9qL72q#(iTfm{W24wrv8My5lDLP-1&ZBJ(sDC)}n!n~}_~
zjdjL)xuBwGH?LmgD|$FOYL+o90TDKxtdXL1nEaNMR*7PCj9h_-pZs!7^TW;iqk&(6
z-JY@-caC-b{Q{M1w)pB;0hH$Mn$uk&<~FD<i*v2IhSf*iir*?gh}gH2?>PXTK~7h`
zU3E`(8frYaQ1}fM<J!S_>KJSE0)q)NLczX#`B|I(hI@sD5(F9a=|kmXE07cv%&e%p
zKmp{-;dYa{uoA8srgB{HHa_*wYQmO#F){4EWaUej1YLHWi>?3a|2{vT0XtmPtnw!R
z(a@{T-#&%xJEHtEuDyuz-uB|xsoqMpxroqEu7Zw+mACnyCsp4-((s`%)6KW7>X}!G
zk}$Ssj#NHUkXQ~<V!*1IPevUu%;DLT!ws4!%sntPWZdG|G_S2qQ6iS$xwX|5+Ulb#
zd?yk<eE89m|2F*8gkH_7P~M8yO(l6Y3faT0s`mEZJk+fh(gMYH^3jqUyEnCs!Cr$*
zb|A;H%2o5$tv91#4|59&l1Thhog5J^+{|ynd;0p~6JIqjk1co=H2ll=rSxApsx3iX
z|5KoF|5FB`3sY|^LPGZI6xgTu%gxd$1{tj7W@_<@IgBU&^pA{;)XT$@%&6Q#6(BMF
zSm|3wLG*=ayRZ{CuFWM~Q?@AS<LYZp-j<MgY$Lb6V-<+~XLRQM)$h5lGWs)#vKL4S
z3#)(hJL|dmjd9~!uAG5^0YVMC9YDQkS}M-Ye(s}=t!?J$Jh^&P$J@-V3l!Ukh5?S&
zfcEA~8X6jHv10h|B$n4(9K|<(nmlQYHiPb`zRUrAWPS19{s=J!H6_ZzL|BIu)_JB`
zTf-lF7Z;b~q{8dje$nnrRf*Ce5NfJxd8uJ31W3+@n=vj#1+j|TU$SRAi>_a|x-^n*
zz6+`n6sUB=KT^8!N!8dAf_dikG3V1&Oy9kkRde<0*4t1vasnhKFffo19ea9v)oLQY
zhYQHesYz->WMVwV!=q%e0NLQ&x%Fc9qmEs|u=%jz&ri)7Vz;aNee_scUF`AYd%AiA
z9N-Cbfa(_fh`h72v)=-CZ1U+lwM@safA}<5sB3F0SiC%av2E#y@~hm5z{G+`Nf-6(
zRqRGjSVXxMF1V5%2abP&P-8HX_{O$z7NSzhc<dWf+dAQ;#l@!Uo0sO<it`}Xd8C+W
zF<v~Agst@`%d}MwIuuv6?dZ=v5PB6ZTqwm#zM`r5<n`-Az$@3MJl1+E1CsLV)vA7e
z)E9uf%|LT|VYvC?)YmQd>IWFs-5hF&4;QhD8J+Q4TWT$tT%M_Gb>uGqIP<Xyr`lhV
zdwr-$2(S(p{t4g)kLx}f0Lu9{u8~LU&fLEJc9c-5dp=gOlvM>ExdL3FPX=U3$rtOq
z`OfL_;92MFdi7$;xVU0cb0S0Vujkizo&GK^UN3y`jtMT-c{#an^C|spa;C4OJxJD8
zRswBnG3R|v;=5R@vRUN*?<|1F@|6Z7K_7}ckcQTfZcs?H`A%x;tlTjH{fZ_^*ZX$g
zI^mWWLYHfLcO^aNSkv?Nb{$!H`Pjn3_P1lTFmAW(V)~Hi2cG9Uav|v9K$OdPajaKC
z(NqPadYYqE8Z`NCOAks!k3<*V<VTARF<(m6w{PjWC#d4yzSXkRpY^=gZRmxT1*FUp
zd_=|K@<kseQR|ImUep;?V+?Q`9U-?(b$hdFb+ka&+RwTyj8MU)2w`gU3!lCUf950+
zkN6gdX#Ute?~7$TK0nxW;L#8CJbs69os5x=nw$JLfpXNd*u=cTw!?68guV5n*v$o|
z0>HU`6B^S)fMGodW(RTz7fxdSr@rGv(Sn(9DqeyfKxaDLfpG%cq33slsi~_K7Zwr%
zYne@*<Q*$VjceC_qXWyx%0ASTKZUPAlQ_xqNosjg=F68aM@hWbdOc~RJ^>k;+KHE{
z+E};!>@Q4AOpHUZR9IV?@Q8<JuC=1XlO;M}INF?NbH^C2xGK>>iXmrRgn-?EuC{-`
zY25Ao%KRZJRcN38__oJ*d4;sH3k!9A<yyry<=dJ1E|A;WG+#<MrO(3V=5f~!{Cx8n
z++$u|&GawBS!pOcTu|Y@{VB$?RV;9r2wj>Vb4l>e8gc4yHx^_t{uaXBu(rA&-8Mfy
z|N6U=F{k2Rt|(Ub#dRPqajG9Tvq?xWfla3y?%ccqcm)4oPhU}0wI1-je(Kbzz~|2m
z8~+}+DbA^H(aN##J8>ibwPwcc$=TV5MMaXZu_j!HMzNcCNu}jLs`IyN(u$Y1>sLH|
zKNXdf!~^AYgNMf&gRWRaNXRuFT`@*Ljp>Vbc-@xeQ|qMMmR`f4g#34uoBOP~`T>Z^
zwp|@d{qB~r3~>L5=s;mnQGKTIRS{$Wcw2v@2_K-P<vwwC)9)V}9R8GRbQFCTokn@d
z;v<=J#%y8uSJEf{a9yGN!&YDIad)P!n^uY#rv09AN`cxZr>Cb^{4$cY0~MF$<n*H3
zv+W6F#&dd(djg-)lm09Jm0%F3;)O@>2p$y{7Ut8_Y_q@`4+kf9o0#Y-VFe*e^WGC^
z1ScEhGQ@{DU86A^2@9<(6T_8vgZGJUJ?qBG%DTEMP-!YN@ZXyC`QJiW_}}m%g52C_
z?lBaDON^3pRMgtQAZi9FpxEUA)wCm~Sz2EzDqadnnY<!i+??cAlb<Lmzeh%VmzS5d
zCU}k=+m0VjnR8cg^oGXf18ZbA9i7GzH$N81eB~v!DMA}eh<49!q1?5%@8{-tUopw9
zhI+6ATTp9EA8`g~8Sk2<20wqszZe}IHCYKsJ>fr&KW}qU9b4$~kcSgy8gCBi|4)B^
z49a6xgOuc5ky|RmcCgGK#AP+uo0%Q=^!;8s$)2%{QkVAJ9<pSAk1@xKKBy!{t*Mbp
zWV6mfQEO}Ko0fGhZxw|gY&!kQHaDGdN`q{DD)+0|PQH^7DJoNr-xNZB;}>~h#;;qq
z?p<nXc7DpIPiJ64or$ggdDEcj@nmM;cc3S4z~*0ZDR-irygr{laLY09yqR5M<yE=#
z{SXA%;ZID?&7DVTs@(kH!-v0{-d`1FqWwo{tOlEewxroy0fXL5kiq0B#5Vy`<-dXq
z)L|uTNrWf}7K3b2#ygAa*Xhw5^n3SQ(Z}ui=H=!d5)=CslmF`75mSM}ug5|kK2uLh
z$-2Y&e(I)UpxR~{mYwjuGKPuZMrUVdmLR*yaeO@OMW}+Zl5OkDe3Nh2zUOFte?%LX
z-27d$D{7=EI0fHn(w}Tum-N(t$5rC7Sr4)qXL2k$465%!1!R+Ak<H%2z>t_*EJ^&P
zo?~Fm(T;+0-0WdI$(pYz|A>ecyM%JP;#ee|e}l2kN$yO478}cf;x^$@lAOE~MocPY
zU7*rVM&X?Zy}$%&`tiGT1V8#Ome4c5=b@o7nVFfz-e(VdtE)?!U~Bib`L9##Mn_zz
z*kNnQU%M7!b`YT*Y!vY{>?~M=F|S`!y?F70u-+j{jTn8ArD5?3+f-Us_VI4_xujo^
z7)#;bK7$AM@Zm<Jt72nfO652>ohPR<KSrU^{0@|7`HXHGU|F(q0Evq7bi?W&5DJ%A
z7y@_Glc*8g+tObb9guOSj<6x;B$;A12Xr@o!ct4t&e;|1I=yXl%t>9vU^_AbXx;z4
zAoy%i*M&FNdF~ioWVeAkr%#{8Z*APXIR@W@MR>Kj_G=)=N7ujBh}0EidtKl>?xn%|
zwWg*VaPO%_qe&KgKbqThQGSDiM&PS$*@P#L+ZgREA}0z!R%a6w+z+tfnUHYg)vH%`
zAprg9>q8Qa)?@W=g4$S;A;M<RTZyDVad9zp<8o*P&tAQH3bwD=VGmG`49rdxeZ<8+
zx3wKaw~$5V#K+ecpx^=U;xh6>xmLYiO7m5pKd*c9<_*GvNM+DL4<!u!urO~3w!KL9
z6b9++fYOi*P6=#Bpl--{Q+;8&$4IwyvP@{fT`#Mt8LI0dV;3%J!{xQ}-~#~iwsg7^
zI-dZJiKo!aMhM{=B&YH}^Uy8LV0Tmf9vkx`#HTumnWyzD&pEe&3030t5P`VFH8Sd*
zFNCt2|5EKP-@p!~Nl=Ndcz-p4-~5+g#u5a9O2jSrt$Ji1K6>;S@XnwOVFxA&$K9;1
zlj}foYKFZH6sRv%hw>h#qFxdLLuqsSe}uei*RI7a<~?ORt!x*8NJ9f$WL`B5jfj=M
zU4>3poyW3TImN}r%iZBW7?6K`xaGO=41<8)!)w>Cze`+$`C|;jByqm6y!<?BO>i5<
z=&9)1s+E-$1g)^+!vjPl3KSfvrl0K^Iy$eEr0y_YxaFuhnQrK~?wOvxKCN^Zykbo?
zZD`2pi!>udj3}tRqr;G*x{riR;d6KZnr<MF7oiQOy8%8bK6&z_WmJ#aL`oa}(oh+V
zbY^}UB9z=>V)>mO0S8IKE%aqO?%B=`UTG7i-t5^FbACsgFpJZlJ@4b<zT93rfH-H6
zdQz51A)wOtU8&@T#zw<{)we3q7FP=1zWaN634F-BB^r6G%{%EYPn1-%d(#aRGCYRx
ztq&N-m18Q%K%`wEZ^-4iI61LNNwFY@B-G~h`Q{ltJ$8hP{J(v>0NzEQQckSfr=cFp
zQ<WbLs`et++Ep0&R~kChf#wbOkayh4Ah;EMP$=IGdL`G0x`swC78V?${eYw0gM&mG
zcH_p4ukkF9y5o0|0Bb3S`e{&YaOaLb6o;=>Rrj%lH}7J448Tdxo78Ob5y?j)d&R?3
zJvz1f6Gz%+>p!&n_s1=^9oNbIVX)%R<))E6{!za&?)UHCl;KUAH($U7dRI9+7m-ZR
z^3KbX9C-Nn!r+N<70tD1H$dJ#Zg}SoPs?9C8^8GxK&kXvh#|BCA%rrHj;*yc+o%!`
z|6MWYPe37HZhS+7YOot_=i*F%P4MX0m}yg^kzfOIA_fC2B_PqxYi{Z@CXeMw2oIC8
z_K>LJQ1II9rB`OJ#?C=tjK?Yto<pMTFxGtV3K+2h3If+ins!dDkz`i4-Qwge?n8%O
z%iX~0Fh-|OpHqfpli3Bz8j+k~Kc<x-Wdlu|8{$PaA%q%Mih^^sE&0;%32SPkwO+(d
zom&-8nKgObT`oS#_-uLEwQY>h_Uae(_4T_Ufu(ms=P)GKqGBD1cGa_{V5b?7!yrj}
z9*J^XkGd?>$%mptybsH@t-E#!FFm{;89Ag&h8ngXMnz(!#+55p1NOW0(V?T<myCnS
ze5At1v^?rx=cne$mJx+@7ZHzy2r<0*q7ZT>sN;Lv#@?=%wJ;Yz8UnnOXFf6{gp+LP
z=$Na=SaDn{b1MLXT+WRGdukALeoc*}=F*Oi4kO8{2n!QY3?v#4v9qt|)ym}e_~9^w
z(3BY?{7!Y9uoPrrz|BtOx&8ih|3E;RK<`7h4KcQ!eQ2;xi@TCE_=4?-%tD^cz_IzU
zfq^3hlK3FJkzhS>ha@x_#2~QlkgyOGZ2DtQLlZzLotS5y#;*%&C14c~b{0`0qajp2
zi;D7xzZVxD&xW*VWdN)9=bM+?x8&6wpc1nkyg6;`Q{d#}R5kym*(6US{sr2QLdNYH
zgB&FFIu@ejVgpkW8i$feB73N?5^7R|qoPtB1>es;_J99es3lD|0A=b%vj_{Kk+b6+
zjPQhuXF9PSr<yf%yP-6f;tnpgsEJuK3g~T=drJStgz2q_m{=*y(BaBh2F561;oaa1
z8m>1fC{znY__ASxzHRY0g(`kJPN->W8du%@fPxKerXnW)lP={H0@QFZ_7ZEESA%Xi
zF?!23w8O!!k`lca03N?sA8)f>2zU|G|M`wkg2mfkEJ)Ep4d+ethyapV1+>%$cU8~Q
z=lv%n*WG-kaI$nqI+=R&231wnahH-0A3iAeUkFhz;fI1yeliq^VVz>vgDaE;dDy=&
z)byLbQ6m;5;QYr0ZP1r2=zTcaQom9(Z8M>N3=*&T1vJ1|8}=YhB9}fnKQFAs2SInf
z)-9_xTH4Z%G3#s)ClNUlcNizKXg`p3Cp3NlVE|ob6z~m*FO}Z~9hNmP;6(6Qt=BEy
z=^Z%mH?HU!7<(xeu^(?aHfKD0wmVrfL+;}%ofm|%{XQY;WPS?pwc||9=@hZjJ?EXA
zM9~nRJ$rV!mnhz75*zS-&t1Cov}y@PeF0w~AuN&yfS{)k1Jaf<N~c$09I1cCCWG#s
zEYfAwUAoB?pT?ClLC&fBLkMv@nZ=rV1mK;v<c^(@g-wi%kSwyaqgR4HD$YRjX5-_l
zGl)c4(-O*t!a@Lk^{o2q0)@lj>PhE`J3`EGanZ4j;$aaXpzk#rIS7Cd966zxXC2?B
zk%6`&04oJ}{ixJ49aeO3_5!Nc0-HV03P1i<T|MEV62#dS;flDR7oDzaeX<qRnTgIK
z!op*ca-~PA@OJAFWCpjhvPLtUK7k7@MMz~-Xms6g1Q#@P){BXYGZL9?#H;))JQlmX
ztp<OvEzd-+eL-qMNjVOo*;w5Y_r*%-)%)S$;hWuLKOo1SX;i-+dDe1phDoS9J@E9>
z?JAxg^o08oOrCIiIFwE3K=9AFqytdd>F(dZf4$K0W7XXzBS9kb?M1`tkBNdectYv=
zhYd0C0L)<?%qRnp?h{6GR#a;{d@=OUi4(gZ#tY^bLBsf&afb%o@h+~9NZkRxK-Dm6
zNj@lMH~bKR-EUH-fBt}ZzaK(l$&6%}sHmuj%lz?^Cr`qHd4OVb4n?W8l@1$II8UO+
z{OxKm+SF(Gi9{0&w!8B^{5~DYaUF7Ti68d|0B?b{V=d4G*S!rgL^WuF62xyee62a>
zECAmv+y>?cUXL<qC8?+UfQPD0opE+*X2vr*nw6b{qb9XLzjBvO?%mIS2w^ft#(Nh`
z(N|R`QGo|r(^#<^?%1yxNLj5$m~*173b-#jUW$@9=v>s9oR8u};O<0m^3Yl}a$$hV
zY1b`~ELj|k4H&*D8@Cx7{pg7k721|XkKmW;!=5{kGXujX1|XAAgTL>R*2fTor{@+L
ztqx_GEXyt_Kr!yxuyr1iMo@)pAxd&^aFh)fqyBt!qu7T$f3De~uK%U6!a1gO#H#=M
zK_)R<0tJ+Vv|3s>Z?%ChdJB2G>mP2e!_0vX;4zrsyL?{?-ab6I&v1aBA_H(pQnG_f
zKRpY7WCDQ7x8X>9AD`^Uee~!j<j<$t?nHMZSbck;$h7r@>)Of!VSDl$H*n4lHOe=;
zh9WzM$rbm%eRP~$TzW_e?mWR`#3@vFw~5We!vpxDb%EdDk+*ks>YGi=cABKLR++Tc
z*!1+hFtGod1&{s=QB5u3;%i4o$JYo<PEJkTS5ELOC=dsUxr`_}<WuaZ--Kb1Vs^83
zLHOlKv!DEJl&?rmJ_2B969#q1B*8tbva+(W4S-VTA+~q-_isk`4hakEMaCICuL3OR
z?o`kDh|bKUdZXQY_dWo60^Ptrm}nq^eWxwj39In#>ar8Q@%r*1XE%(2lwr<f*oEFf
zoX}9tQwtP|4E&#=hyAapdFg2r>)J<<e*Q5sN6_pa6@HAUd4-hDt`ZMvB+zwfIM}uC
zT3NjZg--bVC@d|_(v9YRG7{MTBA#WMG$;DY%g~&R65jh?T~PdAd4T_(Ix6upE|-B8
zJ%#W1{uqU9S#r?L11&|wZ7EaF4H0pK$ulbC{el4r-WP8<ORcS@Mhid`h>5}xHz-86
z@WWZ1G}|V2t1v1CVqUDLr)M~h(4|@xr3pRw{WM^Ct21kDZLNeMXEQx1?gF-X_r!#5
zk{kLdM*LW^er=Hy!N2f{jJ%N&-OyZK&nzhdjVy5mjTwN>B_(qV<~z8upm9T-`mcm7
zyN--17D0z)<K)Dj{r&t5?*?i@DGDutTaeXoLgiqU4q|Z`aCcZhm@<qb0AuBqm3Ki$
zs1tl&(I|IP1&!BSYC_0))@U=`3%~C3w=?FrS@4T{iC^sd^XCxqx6RGX>N&LZ^pBrD
z-3fJo@##~BJ<yE)LO=Y0l^T&b_x)5If9_Xy<kcRhD9;fB9F#lFxDOw8Lk)QM?%ju6
zKR}gNVs_uWVq@eUBAs?HQrR2f>ZD)vr*1#>lauk=Q|5d8b`5v5Tx|FBX#-kF=*MB;
zV{q_5Gxd%gG7-Y&W)kcsmk6DW2pGxBYhMf7H)p{6PmFGNgGpA_{8LULNN4Haz}vd}
z`V7@0VFi_v$pxWuvVNPOn(0+7%|mq4hRW9bUDWQ+e;2VMf^x)n*1Vb<b9x?V_6XTJ
zC2&R7!IqQ=vxBKuRfNA(dVUoKw5M(TC!BW?-*hnEB?mW~{RP>=gL!jfdCHP)KiSku
z2x_de?%p*_mdF|!_PI@jJ%9cNQm38tp}oq81tH0tcwJe|5eRGtq7~{iJ`8D~X_zZ)
zS>RWo0?<(tt#$e@V-yhk=cksVqex6lj6pS0U<i13IMHI15(@wh{nzi`zW}sexpCu$
zJO>y1MTuwKS^q=LEWMLRXB4?O1d}mj=?@&!(*1b&eq1TgBq!CmLd2qCZ-;$Xy>^Wg
zQ0bBha>_*$th%o2o<aZOrHGI8_FkRP&zx0*sM23|@{tYc7`o!k)#bkgI>KmzZdVx_
z``yyc;|P5s23ZDDx!#4cW&CM`!cov*0+TSFw!o}H%EX)9LRx;~lSG6IbH#I!(H-sW
zn0eTZ?LaDW{tS0YO&r+!%XDA00D||w7>@0!GP3P{{%@OQt*IPB=mmiq<aD4JZGyWf
z1L*?P`Hs3c03_%H4_p?M{cR9g5R-ci8FHXDQbtP)6O5iUm6njuc3=XM0AmC~MKQu)
z#yQMZfk58Ma7VruQC$bLi|vq=Ak?4H)I5j?<lV7fLQIm*`$6S}M*@aPu{-!gvCpBW
zol{la3k~`yW>5E`UWBD&{J%r21}^}-dJ(P_$aBM05P4K8D)V2NpP^;)TmSh=>2L1z
zSabXSQs|<UG*S;wffe*7e}6{W@3~nUC^aUZ^yh|~F~)luHoe(DGf;)o-*~uk65``;
z;Z;!GgLv{A^s)T$9*(Ch9i7-B;`XCNRB{Rf7>O)p9k?X{`_b#r%63UQ&)$OwP?a9U
zNeq@DBvqNHmPE`QXDq!uX}I0h)%B?b4-ZcnN;<E$G~7NUpoQ~L=EtDC70uWn3MOFi
z^_jSIbqCivDx3cAfkkS1Q$6Kq%jxEVjwY#!QVKa2;6f)_@Q8?9Z#E<Y!B=IUB9PM(
zs9O+~0_Oc5UenPrOK&H%4a(Hq+;i}o_adaLkU_ZT#yhm<273GZxyYSf)Qp@|4Vd<d
zLoiECKNoYff{+U{);gly<~6xbpKgs=DYhF4&D1O3UR714Xy<{192dF3eI;LS1p~9y
z{>H$fLf;S&=!kJURo>Q?*SV&yROFcSkeV1Xflqwx_-PDcCsBsUIeZ8o?t<kKZ1_i6
zf6Gq#dzk2um}|OvAm9x&EWGISA&E|8=bpkRphxO4@moQDz8?BS_(?;b#A&6Mf}erc
zfn7O1e%e7r#i~r43lRMo6GAZpVxpqAK!q0}yJ904U14+x{x0HtxEd)KLQE}LnU64g
zLFmCo^`$*h&~TBK=5%?<k$_opOrqzQ?934r5g`osa6|{2&5=aVLm1+AE2OE#I*(OP
z_mw5?Q0)?TNm%!$L%U5;h8aO*vGI~kyt~y-)02+1yV*6+`^kCYYkbMkr!85~kZG$A
zice2ZO??J8`VptN&zExgHckdW<H}5=2B-ENF%e{Noo~AXWF??^8<}5(&oXjp^pP_3
zhAyu@rRT@wrO|;T?sn90AH#gxp$8Bsw42;Pn@yHor~fEoD({fAG%Fe#BC_YMUk|UB
ze%DHJsrmC|${d54cyYvV<N5QOFh#nXmiDlVPhzvlRHw^CH^zpts#x`rL3Us4-lCIl
zyA8o{3<Sww6&jA90(65YltZzy?Ggs=heoOUu?y-KGKYj6a~D%!n7YijA3LUf=hMA)
z^ir-OPzK-rEEXWrf|IJaogJ{e16TBEwr>599}R)62cLM}+IrfJjF*6$)aaGOCtrFg
zsF+;?T?kVF-QXZXoydBX!sGKH=J4zaaMM6Mt_vLww@1rYgw1~5L*D2OZoGnm#Uhez
z*7m@g#rdm22mbmXbsB{9^HYKpM9eBt2gVB$5A56LiEQcD&c=lbM$79=q%5nEDufp%
z@u&8`PRObfZMABe7&jhmmg`OvT>raS?s*z6_**um!=Z~9(agFM#L0c+h`dQ#dc3Co
zOXvh<5vXqI=D*0~i~p=JLQNCa)cq%<*MMG6DCEKB@a4F(>!Cg{*a<TDfwvsd9*A`4
z3|tuvNu5^=;)Jgep3LWvCBVaTajL7fJ;`sOc=4uinsB21#FerlRF@Q!wfHd7IZQ}u
z+OZ&N*fs{X8)SYJ6Im!s&&~Dm@YQS9kS{tjuV4TiCl??%VL##81I5d^>-~HOBKE|<
z<F{;78vEs?k+gKf5GmcOoDe0pd4*j7*h?o{cH|pckx}mnUi3XRLGX?(W%CtLh~nuE
z*@T*X^yt^okL%80te+r;<<_Wz&pKEb7_@rt#IFGUkwc_d<J~X4uRd~s;S@gAdaZH8
znJZTU%gv8)alMq>LL|E@uidyooCv_3t<O)tznWXp@aDHC`;M5Yj(4tH%Rl@F$2ue~
zDP6o6++i`~ezoJK<9j0VQobdd??%3X73GLZlwg6sazgNp<{l7~^nl|XW2gz~<mV5_
zA)t&AAv3eHf~FECl$4bJmfS-y%f~Qel`ib+_3K7fH?Dqv;qdXca`gfI%fnsmqMX`u
zrrhG<ebPL6lvV|1ByBM4_h8f>RvRQUUviE?aWW1Nn7?=q8ju=axZ^}%O;L#d<x5l3
z)2eF9L;mwFWjBof<@?lUj)w3?c^Qq;leWwGF~mDw)xU6IJ0d#j@*GRP7|8U{L9z=o
z-|A+SwYAD0bua|l*1O@JJd4v8uD<aOmH4`R%mgMbsU16#(`6kzZPN1asut~o5P?Di
zko=plyL|eZLf%tjkTtM_g^@AoXT|HYo<9&92`10X%n-_0M&;&Zp>F0qd){0>Att6}
zPZ{fbU~%=AWxQLy($Gxci2S(8+oYtV(JPVoZ#9}98w!~>+;n^)Xz+D6Iw8;M9%v=&
zx?E?JnEevpLSn48h!)V_OiWu^7a$08fRA*XT!G|~K{M+&h&F};p@h9q$>MS;+s{R%
zDu7v>tKivdAOMobid~zqHxwqQQEex!qthCi`>-Wmwr3mFv$rJ(HC<jhR$OuS<Na<s
zy8qA2Yhu6h?eF1lmJ*{<AUIJY+QyaR;^>G8-F3(>7g|QZ^23IYaa>A=n~_xUv&6&0
z6=^HEwS6e7U!VyXj$oRVLsGJ+!Ua3FCF0bbT?dX|ziHdaCnm;#(0;tx3OZB!f<kjg
zZImRB_8NBMXV^tLy^^?+O*;<UBu<{NYrdRs><ocPeWCt`Q2rf8H&<7a;~U&4ySv(j
zM(>xX3#a%Ue%*|%RZ2v{p-BZBA{+P^{vGGx!`)cP?~;>cFJImb?2^1#vPPszI|dUC
zh+w{$t&!iB441j#c*j2*J`LLVQGKu-QoZy_<jVZt3Xhc$UB}hG1Evz;9S~y6MK4kL
z=;|cf4l@Aa;lghM=wNE_3>pB+xnb~JDb=A<r>@!)iuKe_{eW)7u`#!q^wAvfMg0b|
zC4ZBI?Cc>)vJnrW#gP~UQu9+ByO0=K2sCn>boQD*94&}th7|P~vsfW0!hDDXgpYb4
zBbn7OfdYFJ39E9q602UhlvzH-Z*{_6US3-5^N1P@uS{GUFTcZRT2JV5CQV1%v}uPg
zSXi9EB*zsJ$)eHz4vbqwF0VJsA+=m?cUZ9$;WLM>D`g#>j=j`L8il#7(Bvm`N>-Gt
zq9*Y6aFk1A`3lrUO&vNa#QRh71(wbt9x8)21*9B+VBzU^htESMet;R6wm<fMa2pFD
zw=E(*o?<ZqUCW2YosmOvFH8w9jB1oyL*P_<RiZiq5v>wC=33XA5{#?*o-;z93m11t
zZJQgv%}@UUC@1OHZsxL!k6<_z!rn6KDwc$)0B|lajAnEW#qlEqGhZ^{!m+rF3KTAM
zx(GA2;=v2Rb8XwP!-2qd#6&KbTGBCH5{VF4yYmL3GDHdyY?(yYWn*VAO^$tmuRvpX
zL3<M+H8wpxz5Qti5~n_IVf_tcn3d0b>%C~6nVAiVIJg7_G7j0ck_LF@DXVXv|9UlZ
za(=iyFC+7I4Ffm2hLtYltau36#^_`|GZG8MP3|tcsovwh|M6`Rf?(j)CXUHy&%I*O
zk)Ksg85<ovt`sGqxdfY=wj|~2O@0Nq$Dzd7{@gDOl$v0uaE8`h(CXqapnDMrwdiW)
zuem~t5;QSE3_uKFhA!I7#lhurf1zMdoa??i8tLVGATFx!uY2zP?~Vup2Ejkr$AAKP
zGv||Gj)4)R54nblc9D%t;StFpaZJo?J32d;;+P!RSPsRu^VmTL__}+5R9!OqHFVt`
z5%EsPy5?qDx24(pB_$qye342QE+jtZi=>7D`Zqo!RBDx@^$vzVeq@_#L;ed12|2{W
zvjbvs{)B~_O-arbKX?~JbO13HBMgNjTJ$v3bWnp4&eZ;}*X!>N-WQ+lNg5~hGdWF_
zhy9{U6W&|D0V>45o|9wf2of<6$K44PKUgJE|CLkN%L?T{T~kw2-&`;zA5{EbKQ8Kj
zHodi$R^&qCpD3{`y{aipcZkhu-yu4g^?gC~!y%iG?mod0ueXEY!b@%=Zhn8y3%o|!
zC$k$5By4E0{}7w0Y}FK#ovkNswC+~Wk5FGFpG@ob;Ub%2w{5$LP_w`pVbLz>CC})P
zj*M1=1-6OSORIP5r0Sf~znpeau4Q>lN5!#+k4lc6MD<j*kc$3^<2wGmQ|oVX_H6p3
z@MXunz<>WAsJf*Y+Bc$z>4$4quU^$n-s0sl*Q}wYrRChwivVZ~!Wrxz*u%N7{&#to
zwKgW|Ex#~>eTfu;>|jzUHQ(XG<-NTa=UCp!mMmhn@jvof{}Ud^f6!Ny%Q~-n>~3yu
zevFTAVI*&Dc|clHilnao7L~1Tz%ilmJ(R<HE#9ld>!2Mve<<J1&W@XdLk7fYY|gDF
zQd|Rv6BQM+(x>^9$t~^&D#4*sQf;k32NVNoe%7f6y5mTEle#KVm_-d9c0HzBJ_*wl
zzcn{3g?v#O260FoTP&Eq>~KHGfAdTlYr|Z5fp|NlMw$||!ZkQ9!On)`sN|7SZd#t3
zsfVBcwJPg=D7kzZWVLT#VBuRk(k%~ntYaf>{n}cxI!AgDCL^~I*RzoTlnmnbX4eOp
zgw4Vk7#V5zl?gRYGJ5-Mg|)6<>fUJza*bc*AZ@*#OEV>6Aj>aeCi@9L`5d?#!(q<`
zzExFSF*i?P+9gk~?&bavpLzoA7IS+$t92IIZD_jBU4BgWU~j67<gSBzj_+dJ6VlS6
zZfIbD14}NRt%D?+hC2MnN^MXcpL|#6L>SIRVr*c7WxK7><Ok~>zKy&gGE&}UpT{hB
z@9lZ?ninu0T<@%3q?lbIfFx}{?Hg&_+Aqk65RmpaT~}`6xb|fS+cvIq$_W{*QXEfK
zJssIN9ontD(JwbJ;L@1jg3dMg{LtOa_mDH3Lbi!v={{4qJN?Wq<~$Y2(npSL2KU>$
zq9{!qI*;kj9r(I3_LrBV6e1(r@lK}!W~Emm_4A_b2`slDGVW%9WF|FXS?4yqNBPT+
zTfHN{cyXu7Ov3DDqvQYl^=-8)3&DaQ`ncp{4b^9!VbR#)RGJZ%wLifsf)HYgb$6Hc
zVB)9=O_{yusj2LZLY%f04<Mnyx8YVxef=9Se*E9FVlMM$abb6vYn}eXE6o4=meOM{
zqS{A$;D8B^XUeWaZMlMj?wZrGvxT#=vlHXv&tr3lIZemq7Zgy8%0KyeJ8pUd`T#qS
z+Sore^+a=1)4TL^RhW3cKI(fTWWN6`?fTPFyBIk*5k|qW9ZBOQYwjyUDjp3Gs%9Ib
zSFd5Dz<0KRieq=$(LfmBAmi6CHBbzWc3`NF>Pmp)`GaBA@vxn*)Jaj2_4Gs%w-mCG
zu9!f|a7MlCvlZhg+JETSl!gl*aYhhy58Jttef#$HS?7fW2gf2pc2q#1KFz6kJ}b=_
zU*`NLP~O?}{k@NmHTVd-fq*#y<X6fG1Skhp0cbgIEy9#-BZ+an#CPw0KpK?Mx{BEk
zh<JjwgO}su;*Ls5k!oscOiWA~pc2ht48ltgc?dl{A{~Z<sfhZ)%*+hWI0W)t0zu*b
z<W)XBhye`=oN>k@AaDhcu?d5FO`V;o#M8~m8G$UY1cYY?(e#6>3MTQ69X)y(m=a4K
z_tJzF^!6<m&ec-IaL4b{RYmt4=RS^)Ta0xSXuv>iW_D<Q)3rLEBjr(s&LM}5^bW!O
z*F>mT3*LJRl68VG9moSwB39>Mt2aX&TlQB#G|hd|lhv`q9kDU<j(oKsh~jT2fPL~Y
zHX`9(5^6|9j4;s9j9Xp#wML8G*R+8~;x*=Rc|=9qCajc{l$<e^fw3F`w-N2G?MBf^
z;^NG!!11C5l2W9Y8tmcurFKxlv91!`lI1?WEEJ4;^BWk|L-H^M#jAK_Wv*z#t!~~0
zr}up5vcYj<n&1F4fAf#u$Zwhuy{f93+y#KB4sq>M+8Q2PGV98qd<zEhT*r^#?b@ey
zt&BT_a{e<@ZfcsOOPLrRP6wGAZCwy`oV;m@2;e%8fT)er1G=iJdTz7z@@K~eP_J81
zJ~>uTF%4=U+IXEbz@NxTI_vuA6e%bf_%%AM8rFATs}7F++qUG9tU@k!D|n8v*#bU~
zo-CYHlmx2w+6`Bw-I$<~)3HEI2*R9xIL%8ci*6Mb7H(N;PIIT7E{1}V0<He>8q>Zs
z{*i#GAOmIz#~=2!QBf?Wi<8~iHtjgvO&zmeKUa_Az?tI-jR$QDhet*>>}sWjNSr;f
zip0;G*Zp%9(bBQdh`r4{EON+8J#;1XQ_+UOz2%~5M~Up4<l6EKDMy@X7xwTl#x5sw
zC2@Emgp41^mlW34H#WxN`UZis^IiX@Tie*oJx_Bzk2w~W=(Tu0H$@!W)e238;pgd@
zH87M-LdrqY)6=$;v7Dsj<nO)nh;KPE;|z)vQ0EcMSeclct6>Ep??ZGGFhCwCLciyT
zhz!!BIc~}_GM@GmohcO+a)*x`Ar3tJR#P*#sDri(AK~5W*JtMD=GLWc**j4f^Cz+u
z_N-g6FsjL%*lN|0=uBo8<S>xOo5;69(4qwsWW9E;$Ya^iC>fi?(^N)AM>7c`KJw&9
z)CYI;Qf+E#YN<-Z*4ujBv1$tajV&zZos<dM(1M1Yg6PF(6Jz6#$E#N(HMF!|HrLm8
z+eITprfzKfo_60pn;H@CvRlL;&U||>ra|{lMc?r#j+u(2o7&!!bh_tom5HfoM&!aH
zI?|-b2_o4rIyx(c08H%#C8fFiE+~8>n2;t_5;;GX%D;?h1T}=?)cRCOLE!^rYGd(n
z$Zd1EU3KpB_&Eu@uysZKcmwKIRv*V~qMV1h$#$ul85Q=+I2)qS->#se1&;-bbZmr&
zp|_lN{>mSGLpIqme~2p}0M48#Fd?Zj<wik^{WzB42n5Kvd^d30>>0ALnORd<$+98V
z(2W^Kd3o<pK3xl!zw%IKDplxr(*ItmoJS0&5P8%UQWp{=sifNOCuQ9~aE)Vjlr5~T
zuXUGu=&~t`f9pKP=fwE%;t-ix%(<@P>3BYHe%_uu*12BK!BQZHoI+@_1I5JTK|vR?
z579P`(_riGv$C3dvrz)C6BMt=-%czNXBJJ1)J8zhQd3ult*-VGnl!>XEQ>MHh7jDd
zZ3^dqnMzGmx1c&A(&E%1b*J{#>(<t2t5PYS#r#jxc$X}!(U8qs1qSM(Tfxm;$Hh;b
zYP6@Iip~KdH4jKcw=_4$W0n1sN=QvrMvEuqAXcGESHh$5HGoO9p*I{cDHywo8f%pH
zJR2OfscvVN<<BCS2rf8;EiybbH1${VXHj-gNJPZQT;u-e(3dYw3TP4<t7^D6$QH(Y
z#oVHKPBcPg`ISt&yvIbss%fYF_8ms~=lCsvXD-#Hw6yftKS$iO5YxO%Qvn{Mtzs5<
zafo^-A?i`EbR7c=B&R-F$(iq`mY~D;+9jtsSnQxX(FAFEMa%m%62(|r)PH-gtWbOV
zUD3mt4pPz|R-kO(SO~A~8x9s%S=^p<B&vl{asmfse7DZT$)IS%8->)DCKfWB=S+$y
z;{_<+S#BiV!K9UTg~u8pxntR6SE&K4tD=(Ke50wXGZoP^QWTrPe)DnA^BZt$c9<;F
z;=7mH>fGOMVR1YZ=X8$t`v|#fPukI`-<~wTEG7UBV71x><YZ^hSLN;1btswql72k8
zZksSE$QH6aLg?mN7?uVjrEQ30gw8NA6=`W>WAnpiGkbVzPD$@8js4frCb%wz-=w@k
za5!K<_IR}_o=8JOgM`1Sk<o`D>n>fylxUp%J8x=+652Z?Gvgnx&l2ji)Iwezx3sh_
znn^RdZpe!lL#K9Op!lVvOE#<+_nO|~TA3nvUALUf;S3#|W?G@1^7(+Ut0Z8pRP>#N
zSHz(Z4jp!3mF9rm4Jz)_!OuFo!(%cst`g@nynC11KZQ^}p(B9%e}KsL+in@hyQTTI
z;xU;Q)S<Ht<5-_v?)6(4y#4kFT6Rjnyr>)a>kl1kkj7)tvvmyJjFbb=V1&kgHcfzy
zt@~y5u1iEeM=5@XV<YBo1SJIbgT$lL2e%4zE%%?6(IV3oPi~@S5Ih<uy2R7dTuA_w
zx_=ZyV+2{VTi0|x!X+g0)a#=Dk!K`TB<HSuDyhNNQbj34m2z`GB)S)J(;b%$XH+nz
zYbMjw*5=fqA8Ob}`Xdfv;Wi-Nwath*2x&G!;<yUHipw~E@Kf*STl#4VDk>%YrRBJJ
zT=_;<8W%-poCfOFxjz5Ro3n@pG{Bvev{?IoM=BQTlpBTI1wr6N>%JlMQS8Rh;we8%
zBaD1rS@FM#VKoLR*Ib;OmxyHZN*|v`J;0w?<#Ek)SxDt``70jMrMBg;Hd9?Dif|g%
z6fpHM7!i>7W*FqX%ibV@;Wn3~+YG{CSIwLAGnKY@Pa5f2Xy}2^9YmNfal(C}z+Do@
zWXv7o=B~#O{-_lN^an>DoM>)sMS%DJ>g8-)nhK*ZZbk?}!xe&-xgabu3M-Q`o%WrY
z3XP0zH&H4yr<<*LXQFKw)nFJzps><DaLDGI87|s%B}FD$p)MWGj}4V<Tbrm&+tSwa
z;y+LX{Q(=}-uLW%&U4Q5OT#^5ESq_P?&KC$?{HJOYR(B-mqv2iB6eStLv1o%8w{zU
zPXrKXwNgZpp#((<bv@nPu%OKyWA$HF3#F-X@$qn&N~u@ZO_o-SDFzTBMHEHD7recP
z<bqa9Y^Tqzd>%V;JWPe;?fz!kFfiS_NvTxENq<#91KH*ytr)g~-YYNJu}6co;uths
zh@r5`z5#=gU$^hdYu7wh&uO5ZD5F?ynL?tha7cR=k2bsW=W+XcAOHEog5u&qjDN`R
z(lF|gbwz>Wh#h&atA|;X>h*dfI4qT1jC?$1@LJ3v+pCgM`vrWyd0kfGWB7IsOcPn<
z_0q)&jxsi{DyQ6`l1)sdW@J=@e9W0m?AIW{O5V`5mnZ$xAs+~k6?LJ8V4bX#Fbq)`
zeK5q<&eBV49RG2q{WOn94g~BX2j0T>a*I3#1A$D40Sl_(FdoGz=xZdE7F;nNC`r;S
z2u<D^{7I;sPUlnYwh;nIg=(iwp@YNC+9V{s`ys)-_ic;Eq&R26V++vKUQBzOx^~Y%
z<-%wq_P&3Va(mqj0nz#a_|%cg>gae-a<enfHy3e-i@n3;R_skSKm+ciHu#Oyn42M=
z!5%@OP<SRiT|rS4z}}61zn^Pbl_nBB+mcnU#g%TLcVh2P-PDv0QBNcfx@7=g*?o0u
zf_;(=Cd1sv1H~<yhmrNgR}@W!$L7iab=oVMJV8_sV*l%L#t2!2!)bvGKL4WRAcTrW
zt?tJS8#<(5HnMI5!vnJA*4o}^v^%gmOU2@^n8A}C0=WvE)oOi|ERVX3AL5_j1i~Gc
zDyXoTU^_3aOL6u8nO1+_QvRE9w(D+AXCMI^up<IVL+uJX55tB?M>PFz*{ummk!k+O
c`f3QQ<aS-CPPsuF!Qc-{J(_t$cPO{&H>?fm1^@s6

literal 19900
zcmdtK2UL@5w<nGr4=7?oK!bo3r9_G#-GU&}6qGI~y$6wAf(n8PNCy!Ck)|M3Ae2B9
zP)aD$O9&C^y@w97U(fyK|NYmRxp!u*J8QnV=d5!)hmhng&)(0k?f2zvHN``JvHV3t
zLvu*!ro09X%`P+z&0Z$j{qU3Q=4k@_vENu(QJ!W8`TKo&dISy4SsEqzt9RYrO_4pF
zjxTTR%&%~>I&$v1R($U4?xXCC$wztFPuL!c;S}F>Y}n<($E+96KR8N6qEFLls0Y5d
zal_UoiNlfac-Z432VdMcdy4)zZRGC#d44JE{C?+bySWxOk`;Yu%S685rm*3ogYVX=
zwhTr@*urBLW8-6WL{;1N(zNHx@pp65&|r=j(nQL;?Yez*VBZxbKPDP3#{ba|Ob=0M
z>^m~G(IpN8*Kz}P{o=kJa5!q17<0h$l-v==0q23reG)#~UVmOXapJ@`RHdDc*zW(}
zQ~%@d{Xg;>$jRKjckdooUS3|Q^E<Wl%cl=Nx_sIgNBVC5<z=IJM1H>T+WKV9@X~mj
z#_ii@0~pv&22|Zq=RCl&zmo6CXRS<~EUlNXUq`R5x>=D+3yB^b9UTiTifnbSPVk*W
zqjM4z4}N@UW<vWSI5;ZLq^AD$8L4w}az-92GiRpQcd>`e{UucHxe^Wk)NigWemBnB
z>v{P~EeHPEZ8x{#hIlFM&MbW%Ej~WauO}+`{DwzH>g%KUd}Zj*o?|rJCG0qKGf#)#
zI8Wys2gfx%oV~!OPoIQcrgU_RY-8`<z58auqObUouw8EfO~}dHAvfAY1-gBUbvLNf
z!|-L_Q2y-6`dkXHPoIwy$GGUDjveD{mfiTgNyQPOK7QmSd-qJcC@CvP{QhkmQqxmp
zr$0_15SFG%+QP!Z=Y)jr96fsUoUpKlpdp@jczD=xb?#vukr<&ayCx4m8k?7GxCL*1
zH26blXM1b-wbV>#w1ECAxTaq$n(NaOu^0HXKF>{@x*`AJ!w1u^Pj(4E_)3F(o#6e_
zn_gacQM=x|_I<_dQBhGEy1M7Jvke&I8o5WC;*sN#a9>C{e&R&kC(U$x>3y!q5%Ew*
zM@Q2t|3lwUKN_ilCyjdxZPH)fti{01Ev*qUuBO%8N=Z+*;o{=*U(C1exDfKgwOMAx
zl21FkJ==(2?=h2^nOS%v@TAqtmoL+(LQZ41$=eC;Bh0Gu1x0ow>o_qx;XVS{zH6j0
zZnWi*m{qHM+$Ec6^DjLrZ0USx!lq+wBZk*bLPA2e5koV@iO2VCo-pM2uslQ7|3aEB
z-I6$cSU^IeCl?*~{ex<v?ES}=PaE#AMn*+h$!*Vv99l8s#(OMJq{7*{7wn`e$Gm-u
zD=xll*Oh(WkZrWF!C%|E2|hpRl%c`mU%L(*J{>Af@zTV-?y!z}_wF=V^5MgW`R0u*
zx&@YD>J*J4d$OL(bgx;l949j~bN=H|HJtcpsfbzqab_{wHxJ9<^Xt`p)-^E!qj|UE
zC<^HAdH2j8ZX;YAdT^ozy}f#5S)4sNgH-eTd*IfZdeS{@ZJY7&al=ISwco4rgQYdS
zS^DCHndXLAZSQ6{u6OR=vXD|)AflZo$H?wGwVWZff#`~lBy8i`ORYIZa#o$6W91bU
ziCZfaG%*Iv)Y#}MCL|=3|FFvceV<MA+qV~DV`J@!ib1Caiar+<+)ePDkto@iDdXbi
zR)8`)ckbNHTen`(vx=#hHzx>?itHjLCal{Att=NwXZK3UVmfr1@$&MX!%^Dwqdc;*
z+eI!X18;0hWac=Lyj|z}I;c}$nkS0lFFt&xZea*lxi^Seq(CiR(kd)0Ob<SpN+n@f
zi{eBqUXz4!3`$z<%CKAOv2SNzyr8xF`TlC}0XDDqaNo#w<qwb>o#Zxh!_B>99|xXW
z^{we;x%*`<<yrS(p*xc{mvekB^{qgq_Z7)K`t3&)Gp{6?f3aHW_sk(u6?PRp8dSgy
zlJm_1NiOkHuGC3_bMA<EWYKMUbaXI#bwhAw_KX0lv|Fxy-1k5xR%)5YvNg_(T<RPF
zAIU-;sQ;|{WoWD=S&$f1kzz@yC@1A-YUh*+cy29srK-k>jqJC)#@mcBQE+>H1jAO{
z)YRmnuVrW`sQCJ{(9|GhXB*eG;V{~os+@29{V&`wk9tDnWLHj^00!<5DZedUy-dJ!
zsYNk0B}IOFt=Xp@D#`5Ivwgz$eV4~uk^`!%tKYD0s$O%(vPxwi_eYAFrSp+<*jjek
zjH5o`UZ$2k`jjWfrAr-yV(#wlrLI{vZ@I<d8mD!``;XBhH_m-=q(p#$jxJ(p$w^6F
zonV&;hZfc>w__lN>E|zVofUhRcsri*<F!c6k=CT^2dX5YYUNgEYin;-iI=WlH#YuI
zV+Ys5TxS%H&7Qc_B)_=0Xa=nT6;fxMSL?k(=Fv=7i-LQCGzin7+A#acfQ3yl7e8%U
z_pzh^CnEfXDeG**yDwfjJnm`U^cTi?yp^j;vI4*DvAH_$IMIGShZMLu``NlJP1U~n
z_ceDo+e$w9b38oC%?UEjlf|VBQwzhyr!1`u^78U-eQ8BSow;iF@5hT*F^27so1?<Q
z*nL*3*rZD>h*6Zy5pls;KDyyGyD8e7BSI!W=&RU%O_OjGs;-?~Rx_da0^H;bl{hig
z@Q4Vr-_<V{1_IdXd%jqdmmXejRYsGkx+aN~+duYg90`qz;)>u=@5~x>k~ACq5iCf2
z6&RT6xiSO)@#<Ay^rdsKxbq!{>t~JnPeDhAixAZmFQp7u2l=f(P4?C;M^a4{gWS}@
z;}LODM#HPruX`A3(^TV9lp?rGmto6sC4FC?9U2}k+1htYRW)j8$P{PCn(_86i-gly
z=xIr(v5jUZ%FlX*w-iE9)A_2`u05;%@uS}*?fa%bcXaU8qpYlau##f7Hda{sZ(HF=
zW;1;y_C%Z=X{3mspC75=>E%-vP47N=DnGjg`|jp^h);Xg;DqeOKS#9u)w=K&+-IlB
zPSqbl%$&q%fr!VWO?4k`1j3fnn!A<bvt>gT51Q%DyS}1OH=5|vpRBFqI^BEkq@V%s
z_wU~q$|ze+gWA3A0eyW2Yb44xf>JI?+m!2upk(@@CF0d$)FniJtlis1{6Vv}LD`v!
z!D{6g3MfbOw`3(-y={mUQIy?YHHjI9HWJOR%b_kk@3#r%5beD|t;12Ud@g18^$IN6
zVY_gIjpU}edD`X>k6c(#5IrGCCX|Ddb2zUC6;XroRI#?UHibo#Z_$jBfPU~!+I?Yz
zbj#YthD%UTt<+`uLP<%<LZ;DHJrqI+%6V<t!sMONi6e&Wwy>WGKdjuV=?u;If^v*t
z6fEb`tccjyPaAKU?|rYXu6=*i*A&{5>~iwT=0lSN%I)v@LVL-@kFKBO*Ex$E^W}K8
zwrNARfql9JtT&Bfr^6X37nOdGNddBC1ICgae@z98tuQq+tAkpQuF(FJovjA9Q~+JE
z`e9~I^J>zE4;;+GrY{Cm3;+K0E6&b^(0~-yS7y=P>q|mj$r&isHq*LruJD8e!P$z9
z?DqC{w~l52gl1do%fjXjr*a6Ybk49U7hrc#hU@fB%C+e(>xj6MLnGgliHV8o0f!e=
z1EDru*?_o?!6u)YnyO)7zz@qK?Af#3=lS`U#+l``)e@x7M{ui#Ls=-hxD+AB9D-fq
zv?3K{6<D@h`#m`LeZb)+9A39=(QM7*rLj9bWv)4=rCb`)6<v2v!+MW|v$|I5G|qAU
z{Q2<kaE|+>PGPXvl?@E~R!FWsxV+L*Jwc-i{&cnY@WGl8BGe7iRR9f^_xfB<G&(V-
zXtC?mU*F{<iTR9A!fDYow>a=Lyf5c=zdu(=_h0V^`TIR-a2I%akEFLTPd)sq40k9h
zR>Y#8>TYP5&<Aby>M6!ec5ZI&?(4;>wa>LRG&FJpXxLz{F5bl3k%}B^&bYQOTK){p
zv)%h&^9ucE-Tpe3sa<E6?w2^;klk9+)O2P*Nslco!VW>4AFMe<3J3@we!P<`&BDgQ
z6PRYDmwVvp&k={q=gyy3x^Usb?uwkJdk)l7r%8J%*myB=M1@nYUc3l>_wJoXPZt1%
zNi(_IYy<l=p^+Y&^OnDHBhS>&%8yB9oP^6Mz))o0Z=8Y(7q#i+f4_01>Z;qf$lToA
zC+P8s!KuTC4lqqU?A12%=92NG(%+=a&(CY<>0M}PY1uW-cIYs?XsDpzpOTuoFJ<;l
zvI3*Q9PPlq5b5pJ!Cm7qT%6M1^BGYR-s=S^DJiRk&01M{N=iyEb+Y<I4z=lsRf<3b
z&Tly9v?MDYGnhLC?_BeISrg9K|DcC9Bq}oUX29VyLqCF9WmlX><6Un(dIW6eq+Aev
zFfCERO;AEYcbz&t8B_7&^%-S{Z+w1C_}@9HC}jn=Kr>Ff@s}Rt`aqj4pWze`P=(!b
z^YR5z4t~~w_>%&9wO^m^@t!b&%^u5gh=sF~&T!GJ&LU68ZR?QNHMegzA+qZqfr7MW
z76C48{iMkw7eX6+aA02!(skzss>-d2&plRO%{^1cA55B?n|mU|`SOP^{O|rlEawFD
zi<UmBLX#@6YU94bdCdeaDo;P2l0dgScc79^cF}aa)^lP2?Z@ObQnxYft}e6reC}DZ
zv^znDB3x<YCQXaIqoczM*uB)*4PJav<NJ$E#$)NC7eQwBQ-B<!;T&tSCr(GWG7+vK
zW~m?33?FB)p}KW$y1zV5oaSpJz|z{~$u7(WJ}pf_5;)<`o@b6~?=G`Hdi2PDS><^Q
z**<1*w0T3CSJV(VtH%BNXQm!r+2!o)oHp^O@n5x>+2kHLLJO1*@I|J6&A8L6s(~W`
zRqU`A+ls|?+lrm%<BP~4Z0o5d$_nh;w-4c@K<Af59cj|6xNS*As>c}3iZU`Thy<A3
zx_2+C2%vSA&(8K$IYF_a*&^7pCb#>5?um_=EV-0<t>woWHG1>29s+Dr-<@Y-R`dEy
zcw!<N*!p2=C=G2FUb?l+6@zRu4T*9_{3n%j1NMko`^5xQR#hR0C*PhdnWx7(#C843
zLBE(7$H$M&hU=rgq4xi@yi`BOBH@?<>lxPcQAS4gEPY&bd;48xDVNXu`b8>G*DA{*
zfVpGE?DU=}(tV9mU^WzgEE%r>P=05<i(m@GYGHG+SsAI3?@UI=%gFZK+a20QQNU`}
zoK>u>tQJ7IP=5F@$pHGw>B9l<<a`X%RAP-wLk^#jtkqPP;Q|I83b?4Q<nd@qM#ilv
zmokH5hpSL2ri&wu3jjQopty}{X5n&r4`t!{!z(MVofi-|QAQS|cr6A`^%g3psl{Ia
zfnlDAy*A&?NMl|RVQY!I6?96d@~#{(MyU6f1Dgm?#3d$O511(O8Hu$JWo!X_R0u58
zTHL<RL3cJ-y#M0q$AFOvVaeL4rMEfsiEjW}C<O4<Xq@*Z!Qx9z2QW^%+5U2WlaHT2
zx5IgSc>esk`|_+|V$JpsR?HZ+r`*$(Ol^!4@3JER{$*?58Akhzw+J!6eD&(D_1RBl
z!TMaAziPuc5!vKZR#r!rxW|&ElA0RT$_Pkcz8UdktgzWDzvm^}7A+h8BGS?ZBh86u
zglofL9AscvvD_*5SPq9nBr39bpA*n4nERNOFg0a+o|iZD^Jo65QmCzbr*RGUovjse
z(^I}Ak11OIN+NWc14E7Cl=du>LOm`Xo|^P6{1%lY?C|SKPOn~;Ru51?kb1h3jXbpd
z53zE?d9`NWmr7h`T%%-hvH5J~A&n^61<-zk+}5~y+Jyf^p3U2l$80|p-D2l@ZF1`1
zn4BlwSfx2Qd=D^0gSgV0n@}c_12|QGKLcuNc^lJ_XQGIF5zx2Q_GXc0X-??jhK1e}
zpZ*Xe|K)4T-3N2Na^h8Zp@u-v(7?S$yY|PeaXDMGf4s90XtY^5Nu605Yq<^(du#<O
zGv6dNd-LwyNYcV>S(i$_Ia+jt?9S$6@<MgB{L7G#6?eHrZ))t&(9lgs$9#DOg<DWP
zTvAf{C((xdF@pEkuaa|fFUHFGR6I<y>HHi9bXy5l9han}Ua9M>ARUvS=%$!+Yl;$9
z_gcJ+rxv%`JKpKu!U$l2R$bZgD(^1eX-bgEBxM8TMNnppSrT>%+Ls~jxstiyq<ANW
z71RHRoT~!|a_;FI(=VBANRS~V3&||ha{7B$%vI9alU6{_LIj#=COE62^4LL1kEM^%
zp`pL>y@7NT0{kUYK`^7MGZ9}e$^#5uy5o|7CAm%h@a^#;S-y>B)ebhSO14;@=vV-?
z!K2+A;CSgUa_aoL`PFu72{N9hEgx>kPFK~osmpnjm<L^<w&`QMVJ$hjZ=G)si5K~E
z-TxZ$?S7k9x3fjraqkm^3x)H78|LqgUu=>p?Xp0*{o0j-VdLfJ7sYE8yUrS(zi`1(
z-@Z^URXfKJ=Z%`E$?igL`pyY|G9yO&J7;8OmQQ$sAmQKTV<1%eWruBNHQ*r|Flon;
z24>ta2u29E4yobhSb*;FCIbgZfb6JU!#q)==C%(tU=|+B_B%ZMXwJGIFYjEjs;Ozp
zjPNoL?gfw#y6ze&e2abaW~}aK#R@F#RkNLg2M@}vVotn1^Ar~4($DJ*<IyoOF+@;`
z%O^HS7{YR~D%@UYqKpKcXa?!Cu1K4=yNEwfO7cQ+aWUH_?ynmA>b7(m2dPS;9}v#*
zo-gV4XOeBCJn#q0L8cpv_FNc#dgp$!_nPGXWY%k#c(xO-$z%61+r?RFd9ZHZ2$}c;
z7a!~FsqilA+VWg%j0C+-#^wdcvw-sBCNfQGUY8Ctp$(TzEh%MiRkt)|J28CsF1sZ2
z&;bvdxJOvsp+)b})zuYtUoa(0s-<g++4t$FbFS|I$#y1gEKgc=E5d5-bMCrtPn?{X
z80sr|j0#beANqEHP4r&#SdtxFz$p`P51pOPNhh4}deQM8wCFPo(sNI2dkd7_iducp
z$AFY$X!LpIJJ1VfVj}2VRkDcMMJ8Q=zI+a(lEJ)}o3lc!+Bk_mem*{BD2JJ%zHz70
z&1;<9HTDJX@E|3Tqc36l4HCQ-l&a>(OCAmS+n2O7+g`uoR&wN@xOc_bW_uhF4sz1U
zRL->EuJ`Id!0>Y>LkIosU%m|Wpn-Yy286-KHBXmoKtl_4rFd)Oysut;N~Bx=CQEH?
z)gTwi?W{F(2@5Y2HD_sMBH!JbagUEoRmjN5a3YpDjsMo%xpna|8}>R@y4JeT8W=H1
z(c}V<IV_=s+F-Rer~qnR4|(r}i}kijOxM@ZdE+wGEmS3xZCI}7Irr^AA!wZfpd~9S
zE4Kl2ec#dEK2`)}saNJQjhD@y^#X0H7o_D+uzsx=85#97G{UK061P;O9*+b)B)vSw
z<3E%~@!8q(-{>#*#C4{tvk6?daLcCU!*el^as}k%D7o=z%yK9M<&voY4SEsoC;;4L
zj1ZDwZ_dVG9XT+JX7ieR`e}ae1&|V1FInIH`WgeoR`yIx4oKO~lvU;i6P#X_9>`i>
zshH!26%AV;hWc(xW#Hy*O`h0w=U(Mjd2?jyA)=+I5Q>|RK`R%wXgW<;nS5Cn!82L0
z#r~#d#z}G*RM3@;ipKipW+vcP1zQ2TD~Z=(FAUvBce3`Fx-_G(oHVFB(LkcsvIvM^
z4v_N@26btn-*Yy0os&Cz6E?RnXno9`bR&5LYOjr<PwYJC82y<ploGf354Pmej&#|(
zif-;_!EeaV&o}MIyR$D(y=@zh1kJUMF9-j7kwgzl@dbdHj@z5=i8WDxZK?to#PP*m
znNYWJ*t!Q7L$)fP(=D`Xs=F`A0Y8?Em3W+1gE{iZY@{I;)okZ$Z2qX{O^!QYGAlOh
zYEpE3{PeQhZuiE?&Mb@OgkIRs`nLCRxc>FmU%RPYz8adEg?kPhKJ_{&KVKtG(m6xj
zN9JS2>#dRLFhx{LQ=%NEI1!{f@14g(lN<O7()!YPviD-6IO+`MddWzK<&L2@Xl#F*
zG>+4~m-2dS0q3WHUJL~~45AHU(!fRk?Qr>-j6vIBlr%4hD@dbbzWCq?;$6&?&B`jN
zs6_ht?akB6f;SI=5{l@3AS;AS^_T0zdrsvF#b17OO;|)^80tC_7V=Fkt!PC1`(oah
z^7*q$(>n=H@Jyl$ZMtsh>c)0$9PG@)0J;PMJXFV%7|DL!|E7mW={auhS8%y^Hb><U
zB&z$xJlJcYjt9G=+e`2v042I<Z=X$%lGgE98e<0#6$VvMr5G7$!TtNL)S_`_VwP+(
zZx2iJ4_q(}i|D_rFnoGSc@}u3x769@%FZeD`v8&ux3!6U%eQIF_wq^8fOZYp2A}@!
z%)ZYjV$p=3Uc}?25nsUfBFD!x(Bi=S3d)$qy&dO=dY>sCJWNr1z8}rb{%v>^if7n@
z77G;gzn(PnpCvr}(XN11J^AT6Tz_J1sZCwAL&k0HJ+T?^fM87so473-3MH7V5WHaN
zW%-W(ZJO0L)ZS}xpx33Oocr|gBkgL%{==tvNjMC{D)v2oi=Xu&FPM?w#!Z+RB`bzx
z$825k2C;!z#9Za;3gd|r-@_Tx7!5C`Vki}@72X?9W=v{B*;_$Ud@{4=&;gJsDl02@
zSEOWye3248ZIByaW#qjawKUZua@M=1rbgMy>J!*V2U2`QFT;!O(7VE;qfdrbC<e3K
zaQMdQ$0WV5Iv8>w#V9&x?l`CfL*U$k*7v(hR=?01b`LJ&LdRxN6_Xy?W~cGi@3URy
zAa|-dd^3S}j?8i)>VdBrr?i%a2FDpmr%+aL`_;a#c-Q{(AS%OJ+q@|JM?xq;5DMrg
zM(Zq6F8ts2<s31bdqs%$;Ab8Xhz_1ZY)<6*WMwUvh7=Tt0#;iX{P^=luC1Rdq8dXF
zT>2CO+RlcRsAUi>+U?G*b)$b|0UBet62BV{EQ`+}dV|dRSaNHDm1a(_xw-i*kR~^!
zIYhZZ>o^JOK6LAq3<0Kz0X|R_=DGcVckZDEonv8u|78}n3@xzjX-yXE>FH^O9f>BG
zHhAB_zC^c0tfydm6&o#jwIL#v69feU9DOX$EN)b0d5#IuCx*7DUPM}-f%&tmH-JEz
znwpMv4#10?D28V~zyYe}wy5}Y)V93aH%!nB*|?6@$GC@<(Bk<=pGcf$C`n8)Ecd8;
zPRI02hTiQH_zI4o29soSK79%U>!x$($uFk=X>LF+GdnvPgy9DadkiyRiBXT(S6%v7
zZD^mrp`HK|$~Y$`riFXmHhvuRh{FNbflGrP0sE7NmKL{>&sO0(S#Leq1&3fGhuvwy
zYUoPU6}Nq%9o8o@I}knXC>t9ZGyoUf98jTU*T*<@^NgQ^1=F#}yX#V(E&VQtC?MW`
zPo7D@(CDZ#P+?!0b7zMb4cXINXZ!Pk5g=qrhd=rO!vR`du)6+YV2A-`Utj7xS#51*
z#!!6C?RhRxS=|yxGe~FnovY-#g1G<SaPXfD1n7JdHf^p;m+pa{#)mjAz?(I<f>_1X
zVDX&%)v|jc58CoR&m-bC;5veuAlqGa(bOrk8aTBXV0(L_iLvoek40h>Wt)mSdg4TC
z85)!V5F$X$pPIi6-X9+rLNQSFw?J2b-SHOqM!*c~!RJiH;|&h8$?^lXf-N$!7Jekt
zAD|0w{$o5s@}Tcg&6(DGGiA7@j?^h#q*s}Je|5s}O!5nahSHz9cx6BRsk877*wLAQ
zfyTgL*9WtTrvTQtz{dy53m>xI0~$u$1;qCRb|>LH5vh&K^lRPRuc4;)+O|8Fy}PQo
z2itN@1+b?m<H5BoP)1fRD+HZ-n2`vMWmZZ`wSyrjnZqrsGbN)aJNVpQ;NPmv-rzkH
zS+;Np2nn^c%laQalR=^Y(N2E(q7%XscHkuXZ{SGNy@=By*#y+o65vED2v_5SPQ$}7
zMu$t*4y6hp`oP+uC*GhJ#Sto%<qH7N+JL}?lMARy0`-8EYL=%{Fy5AKb2y-C(#IEU
z_vsqB?Yx1C20vhe<puDA5Gny=2+hTlOiby(nTeBw_}HGgjArEXk5>iS7_$EYQDR{v
z4l~gSC<si)WPKj6<vKs>at59hSQyu@2gXFqzF$VaVsVa>^SaAS-#B%|z!hW{4lt!e
zKI7o?*T&U-CZ^}4Pg*aM#`Yl|{3*nPw`ofY*3#F{B#pvnIB)3S-SXsi&KH1PFCZgB
zo)keFotDEItmu?EPmXkL#Y!w@%))U6i=d0|A*F&{OSL~!E;Y;MKE&ug|DDO1l!Mq!
z%ac4HctwnjS(fZ<Z<wtu41XwDYXbbPAZFWrB+|1Hpe2`xNSoc-3r5bq=&0!E`p3;O
zJiupgAbGrioCF{~5TC;k@H^G>#SB&>h@N&H(ueG@tbPw_MzdHZV`E-e=HM4z4Qy7|
z)KEH`jPop!-N(LcRxqKgH^G~#OD;Obtui;>kcr?|Ae^0`4F4f6N=hyL*bTcXm&YRw
zkEBMT__Ryyx*+$Uxp;<=G3gMC=*<maT*!SZ%VPTc=SDQ+Y0+rh6~8X9UFk>D!Ck=>
zPQpUBk0Wd^=Kzy|N+(fy)ZRJt{7`Z0u@##J@bBEy&-)cv>6i;%zsU9xDI082>=y2D
z=ca$_$yWnl|M|TsdDPKSZU#zFr`;E~dE+W9tc&TQ&-ys~R#1VeWDQ~=x^h~^qungF
zCFVG|46gbv%M<r-W+0uGf@J-pk4BXi%<2X(tA9`XcEkQ*Ppl^rhoR8@y*GhnP(9dY
z3`(7pv_9($6m1UGML>6V#tp+AdaUoVzGQ_nL-g~=$jDkCkLtQ*E?N9KxmQ5R=K&Qv
z^v#=7k9`SBADb{HHBpsEkfN@V%oY@;AhdcCXWdahz0X8FB`wXPj~k6vU&(e>ZL2o_
zQX-?=>iO&;+&E!Ozw{po1MwU|WyDP{!l5m!jV2NypJTQ%L-t=hma$0cel&988o1RD
zbkb8&h7RnUVP>{kRnYdisiLw}+z5Fr`bmQ_mqxqT{#Z+}L8tmk+6IMm`A?RBk7+v5
z{wd$4OAu)gL8m2J?Y3?wKqg`CFLnl*rBHk5Q)_lz9&g1ls9JXK<_6t?EC?={>eR$^
z8HKRP8G;Bq?Ccq~6MqkYeciOK0TRhjbbJ2dze^k1D^KrNcuCy8zWx+KCLyHu&Q6xj
zPnnsE=9f7Z2pc7kvKnIWS@l(1u`OAMqip^d_BKnBCIp`uIKg4w@aEy_9z5V*l58e`
zj5Q{0Gq}l7R&8lYpidy;nC#}<bqP=EXYH;eLyz%PlB~_tB5Qr|1Ge1=8C)2u`;&LG
z?JI6E(*<M2X||tirhDg37%AT$X_#8s25VbHR_N#{urOOo9$Sp}m3)EUx{owjC?xw0
zHxLu<W1GP7ujH#m<f)zls|Zm@M&bHl%ds!h*@*|Rjw4)wR9g7EcT+`n*Y9a+a)MqR
z4qLzXPAn89#6{}BqPkzCbn|9tcXv0bso)LeXLY{NO=A-iGoXL@K0DYd)(g`vWr##-
zw+8BbUo<d0q=w5;gFyM5F-o^z0?k5lRHe=a<O4sDBg#ZUG7lko_T<t!kcLb_OLLs*
zlRy$uu;s86?LbT?Ys0?<{P3}U-ZfCtJ(n(wL(8iN!kb!Ltb<)CqquM0idzqji{taz
zAW=-mekS=_WQk8g3Wg+{-Zs8)KH$e?!i5Jtr_TE9Y*4mZn<qLl%|J0Ibrw~Pka#>&
zTf^ovmEQ-MlyENPp_e=fp@^1}4@%n_#Erh&<CLM>B2hYIBdKd>h)jeGL3mi$Kuq2N
z4G`R<sof?E-6kPQs;a?ozKYbT9!}^7p&$)lee{(eJ5v4qyAou?xWvV^vCEl8CDV&Y
zGKjEc8?>0Y8xn*6rb#@=n1N<&Hr|>l>^yN7=M2==*+?Wg11b@LiS#U@wdvTU1UcJ=
zMb-=*G5_~~duyNVqeG4v++V9wG94O!7GgpSziL^sK}W08$CRh)2do_u@Y8-#*W4Tf
z`gg22CZT5K0=P-YEgc3`vEBO@vm68w&t3PY7s7vv{`2d!9|7H6vgy1D;$*6NqO3pW
zZU-bTKrNMTZEG7QyHcV+70g*L0srN*5y22_%fV%wRZ=#rN<-|L8FHPV>BU-CZTcV1
zcHejVe@PLLl^GnZ2Mxvy6lGygs%>3UQ#2g&T1yD_91z7Qumq9RAxH}lfjAl0NZoRK
zjgq+lk(nqs-L<7LwfBg4hG?ZXLA^f5#dQPNehLV}tw6e81_o|mH)dwVw6wK}U?UB;
zf6{c^Tz%A2@;Dv3iYA~z?G4Nr+>`Awewyn^tBMx7|EihT|J0yu=L3dcS|GRmc9MkD
zs+$#{5n<RNNwUol6)l^skfP;$CI=hzUQb%w!IHUKw{HDKJ@U&J_Uf2KNDFBJqkrvz
zI>aIz9sIu%PX13l_Wx-F`d{)}bbV!L{3<{wuLb+ZI%DRW?}7VRfZ8q+*v;aH8*_6G
z%$KYq6$h^mnzo3l(oxSI=><;^*0Iu!8#lhLTmm-=gmcfYD|?2gEqZ9LgUSlHP!qku
zBIDsyK6PM%6aSzd8tLwe5C{iH2VD%Fdv+c+peV>49B^WT3J)`He7nzy&GIrW`pwak
zCr{3C-S>V+P8#_2OhyGsVy1u~xNBTWR@M;IRPcEu&u%4u8+azuK&-6|Ehs2>vLs{O
z&ijlG#g6|K2Axy->{h<1UphG4oayR`(UFnI3Dh2to*})%m>K1n_K*emeHv6CmN0V(
ztUP!pC;pd^4|*PqaaQp8^T56s|C!_qA?T{niW&T(+Id02lY!rV2ZSXtJ_H2-;IuNO
z!h5T=3*6m<f^J~KYi|~|o7B-_@@)OI0h%M%2ik0xH$)tiV0UCwe6HXJ>4LToH|Urz
zJG@u-4q}t*S5Llb0#w+hf`5(wBwY^SsX9>lX^4SG0R{2s5EBv6lysgL8f8opf^#cG
zx~{BOv9?{wZ3ZLo@?VOxA4j23xfbVI+<=MKL8vPQOdM1p(xk-fdiPq&Tk->2uZ~+G
zJx)SxUvc|pk-&fEpJN06OyczwR_Q_xd_(_5N5=)&9d{UBWSe(8vI%kw%KvYc&r?Z<
zHF26fsjRSijHs|0-2Zn9!ItnTF@;>_y11X_4K&JQCE&{0C3dy9-vsk+(&x(UCXm|C
z2?{plCVZL$dG3#HeBk+yOyI2GI6;Z?zR%8Xx+|0I{vXW|{9h^}B=NsH2`C-l3+G8l
z$3Z>-_;|HFF7WU0nVFf{!g7}p5s-T7`}bE6PI@>XIYHs@SJ6g4*eR?yJ~U|EyJt3%
ztMVx$qZLAfS<AB&xv`q>Kd9s)uQf+<A|O=Ke(#`u`xvy5*kWc$h*|Q}?rt5#+Mj~J
z;=<x$DlqnMsG~FkZ(|f;7C->mX4V!*1HT&|q&hrN2%cN`|K<N}eDS8VyieyAG(a%^
zY)~THT}8u^1%Uug#n;CJzjFYC=iuN-ah~i9haffVS5dHhpst^P!XlU(m?zzw<*wpZ
z-l|D+FtrdUZ?v->gXH)Ue74<@8o{sr4uqiFArfJIA`{8G5MG1RuWIPC<qQG5Na!m_
zMh_Ar#K`lSDl07@sk<ZIF)(n)v7hV`t~2jokm4nzO2E0j36)~okJkm@$yHZdJ2$yX
zhRYd(U+&b`fLsgY`{1jF<1iIrV9nOSUWw$Eu|4X%D5>3F8(hN=O>4bm<WzVw!6PGc
zAL?D`T_OO*ajFOs%cnL=F9vaQ=m9<|0v0+AXDB#XKL3N&-(&_n!y@dKN8`8?%j*qu
zgiA)p#bpAbZ(Fpb76U3xn<kfaInW~ka5#?waG-Ox$&-uS6t}@cWLd}_XOk)*pIkZ(
z^9F0a>jbb_WF)yc@7%d_bw|b-_bAhk?X-lWJR+ME78aH+BU#D1x;oiuK5(y!Y<oB~
z(^SH8FLv$ldr%zwA+`mB7Q?``!vRAM3=E*A2>{dc;Skrxl0#Bp-T+nD5fLAcgjK?$
zqRe!}Arcxbc>lsIVW)y5;lOn_<jm6>ZqR9dJ;ASy#FaWj>=rEb)q=!7-u~g!WB2qy
zKLRQk22zj5lJabQ?!&5-+}zvX)Q)P(aqN^84h@FLMI()<9p)(>Eu_nB)!ilEmT^&S
z^P1VSPK!Qf=sA4}D!tVH48~Vc-gNHQw*?SaSUVrPjw}OCCI0&6Ti(Bu1t}t4<cpwv
zC9b<ej^mu9WN&7Xy3Fz)f|P6c5hXQn!p$${AW}mhXiRj@4>x&=5K?A>ST5bd&IcN~
zk^AFOG9I-6vA6njc>3K(&p<lEQ6E+RZ>DoAM0vCI6W+Yxg`}D6?C6(f*$5bLfRsiw
z82{tldC8_07DSjz7zU9??j!5TliaW-#it(sW6Yo*Ujq8DGEhLXfy(Fplr*)sqIGEH
zkBEm>{qk5UEV!f0%;%9L0kkqCW9>{Tv}#j8P-Z@Vdq_F7ugRi|bq!`)Ki2IA8%cl2
z`6u8&f+QAizLj`*ho@+oBFeowx0%V%?VH^OY_isfK!Z4C<RbCM2GC|ZU^vF02-B0F
zs&)5n3JD9l47>Z30!7Say!FNkrKt_l31qj6aTr!rk;j7>EjDgTpTP??^j^Lf(hG3>
zCT<fsbuQ)TVgU*`Q~Y`bwdvlc+}Fp`F;XU*YEat)`8lwb1ast`ujf4;L9V;fQ?DcX
zZK7+0oejAVl9`3YWHeg8s3AARwclNFWxZh!gY>WE={>-n{JZ{wE5rvozdaF5-KB|+
z$)e`#Tiwg+&wY*<au<SZoT~<MmbUjSg+Q5G58+dArS1%}?OdEnhPeRnzXpq>K`;Vt
z{cFBkUcO@WH=X7(Ac1uKnTgBrL4QX{gcq+HXlVStM}g&y!o3U%YQLXmSmvU-4HClJ
z2_o<1W5<rkF!!!htc|b|rsUY!2M+9HRpxObSrU}T7(01kVZpHkqVd|^J{!|G6wV56
zuYC!R#Zv`m7{4(uMPq*4*3?YaM=>$2`0i*8bO*bwAE&X01Omh#`603hnHwiN#Srl%
z#NHq~s#i!~%&;jy8`V5;x;~b*m|U61T0YsxhU*Y@fpsfl-Vh30oGb};#!9IVrXm*h
zxKfKlAkQKwU(?%ps{0OZ6P8~B{j?`s)pVl?^159rL>HD!uL2J%pJ3Wy8!$Tyu06MU
z!e=C+RuBDY(n@~E2uQZQI{~{zzi|^*bJ<KO(as&H3zCKg!~cc!>M!5431*ps>50p}
z-ct$;lN68m87&1$bmggm)WN3*P@&KyK}c71ngpG=X)<uRq8sRi{Tk?87Tq=Q?{rVr
zS_oNzRN^iz7Pcc}J~$y35C_Gm1UH);Ed`->Bg?%T-NDfj;Ns_y@<A*?2DYp&%gTWH
z{8K=DJ=Ttv7R~g}+?#0##=v5wT<v9orvieaGZ^GHbMZ)Yh6rZ))mYHL)Hg07wsc9k
zb_O)G1#nm=iqO2g4(?awx@!#e^fE})QMsKGVBP)H65#ys-RgRU)~Tfan~=e?cLzG<
zHDPqne5MLh8*)H5OA7?lJ0L~Ji9xK#4tTpBtoB$dK&?BBk+Wi7(9?@fbaZrW$Oy>g
zA7E75FvzaHN3_lx3JTVs|MmiNEkKebE>M<Gx$#{!%kfAo&=aKPLhxk-z@bvUb*mK%
z83HSanG4xUJ^Ec8Dox#(B~TnFunlw~&NGyU6@(&wB2w<~=qPJZo9JMTFQD91QUySR
z4d9L`5C`46wbIhoI5l`RSQx)K?8_i~@lwK-=O875YxY<7ya2?2(gzprfpay(D(CAy
zrImS)4~=H%-VUxL#+0zhVZDC(C@za7*wN=6Vil|PJ9rE{6GTfHXIgRp0IeLX_CW|-
z2_c~roZQ~U?s&CyJgA^p3Ql)zZ4Kd8MLV~zxVewPTFZa%ZMXaHmps=-0xUq~$%ix}
zVc8V~a|Ch@G$t#+t&-97sjiMfHn)5Ts8)G=lcshVdNxj?MX;b_cy`Q4K}sePR@r)c
zzyvLt0r4Ba0dI&CN2es*hY+H$ZTCfYn6>i4#(e`}0fCud-p?);1GX_+og46Hiop~@
zxUvn+Oh1xm1fjwNkBRdk+dROJdBDeAT0O*(-SAGLZLq}db_2g4>*A#xh-D%n{r3DX
zAnd50&2cA}E>HDnl5!vVk_cCZ)UIB=`gPmmwzYMJ%dfAys)X)pX<0*7V0k>}Ku1>|
z1&^e<dI3ZoT6ql8-K@`RFYm-ISAB>s7GNsYek2dt@pj=^rYV^@nVHA_HfccZfzc#Y
z!1<BU(MY`boy21WFlmte^YmfG+yI+ipizjp0*&tMp8fQp(Ch1w`639pgolQng<A=4
zMwC#{`?x%*Q(*v}iSV91Vrj>C)R3tVq^^;CT3vm8IB0*!92`vj)j^IP30da5OzS~t
zb<soan<N#`r4o#yNv{n>2%FX(f!U}S$kIKMj9*CH*;GW3C=!%|0n6%HIpvL&*%+Wi
zb&ZWt@O{cK){+AK;V2W6#p;Xjg$DU|))t1g%U?Q1nYr`_dsd4WulOqbyTiSg=JXvH
zZ-PjCXpC)e-;%?~zc&T^e+j++Z+z`Z)i%-OThIUi?M|9V-|hoIgoIXb-U%@=h%exJ
zL}70S<lha!{JK-Pr6m#%^<H<zwQpeK<Ii7Q{}^J}r2m%lFFU{dzi~bOkE$66t2(lt
zZBW9Ih+Y5msq_E?%4uy4@W3zxr2MC=APx;7QdBTQaKztJ$X6_305&`_@*GTv@1k@8
zLN<ex<t_?CBreg6ll}&nk!osc<x@6bHZSM1eGCDy!Q@BK!Btt976y6`X_z!5&eMy-
zu>JI=XBL)!d&%bHPw)2wXYmhi>FDTujd5Cj<PzZTOGmvpotaY+4mr&IDYG!A4YVd|
zY;5ewQpo-Qco7Y~tF^Utck2pFSkMZ(`OQ63MEPvaKUpG%GbVtm9fCNwp!IF`T4ZX0
z`<EjBS!*WN-wFhVSdjczw@Q&A3s4V&=APATmRBR{0r>nipk<|i&h0q(<51#8DZ~vC
zQ5n`6n}R<gAxOsnLIM3`S=k5(8&~nK6EK4?-=z`4<#g(1H9SE;7kMrKCm!M}hG2Oj
z*9sVi;YwI^G&d|%?CBFMjF2=?0|nU>+&tN3A?&7`T=4#2MhFe8<qv();m2?9mqqjj
zNGsocD`q!wkJQkhGT#0vB2m%r2b^b@v9Yl!bZf)H;cGHpH;@?xKj58iUlYRQ2a%Bl
z-~ETeVZPI&og3n)y#-ds-ZLiUoyGq;hJlF#7=PFUX#)3rr=fp}4Irrin4hpnfN2K|
z^8i{JwU+>y)kv6&BTK@I*y_QnbI`0jPE}A6R@$o)LDvXwDc6VKggRVNagV?Ci7#91
zD-rGg-%|$Am%vwmL2WE)^wOnEU}M#+vmMzq21P#KYi;2H!{C8yf_OoL;<Ka8i44?>
zzbyXyOv6{@e|rc)@xN+A3%}I8=QafBqF~Hmq7&LZqDR0?9NC@;hT&j;s;{pfn#>_Y
z!$hSrL^Dszdg;LKBolnv@)T_jO+LifAXg5#X&8Y^1r0!O62`UvMAVUdKOT?wKykf?
zsZ92>5K%z9eJPiz(9KP+aoDSJ1NQn!SMpUa!JIl37y*zeroit<rv_-)1Lyv_3Y1mY
zb;dv!XCJSV^?7QSHaLR7sQEw+pr>AZ5-RVuuNIs)NXESQZX5>C8F>f-7<YfBB>=f;
z+_?j;IXnZP4s6BU(t1~9pe+|XVEAngt|Kyk0uPPw53-Ls1)~rk81&e=tEdD?Z%lnz
zfRRltD0Qb5GaQKv8Q%#EltEd5vXTn(8m%~t5$Fn`usb5Y5imm}@)!wm<AJIGF0c$`
z*M^QEKdRZd$OmsmL=Yr-AT?8bU9ZUYJX}-Aj`JbC>a$K>Wv0n-MxdeLyY}p_eX?uM
zId}&$B?t@@3Em*#TEH5&%*;}eK?>OK0kW$hc|-6$HNf0}OwPiDOVsVRlq>ahbzy+v
z_IjQYC!zITAVGC*4t4_E+;LP57}F}t5eW&|FxhmfiI{OOQ<{Y$v(lHGWl*94<L0H~
zB{|vIo!~^Iz|{NO-kP6D*PS;!{~nm=Ewal*oJ5Bm*wP?-n4nnG0|K2LSoUul%|*Q5
zW#cy&i0tBoq=I442Sf?Gazo}892d3SAbtQtDvkEIN$RM><pRJURTa>N#Jty?8V4NQ
zwlKG~w9>X#YvjcGVq)jO!HUt+(i&;m23T%w-T(Xd?*ib=?%wajMOY1&CE)X)8@oMf
z<lB4tvcr{ruSHd|B%mP3dpS+|fh`AZ_)py3Te$}qQ%|8y%)oXK<cDe@?M&`}H`xTT
zi)nZ6L;qz&=0L_7%FiU!Tmuij{Er7eh<NaDqAf7E<-E~X>YQoM7RZDK2Ju|${{5`f
z^mI?}k>NlX&zwZrK+JfvR8GXga9IY9>(R(Q98xMoG@(5!*>!Y_GenvL8DzEJK<p-<
zUMUd4JbLDgzzW5Oz;9GxASzDj0rR02QmW4!b94b(nwp!J3a|+o`;qb?`|J+B+z{}Y
zXplURoCF{~5T6n78!O{kgv=9x=&2f#p45o(Ju<b5zh@$|c$9HEO4|?=tJT$2QZGnP
z$nMi>w?h8@Sepc#K9l59>32{z-W}2o)<8Iaf-;<f5EuC5UdYMYb=w2Iwy(iE4&TB}
zL+-)Cx`LB~qu{FVek&xNjNCWe^0Q0GjrbkQG%!4T^2Hnk%fkmYou+%2i(Fu#ciAm=
z5w<%VC{XDRq;1eW1W9iWqT?FV$ey$1f+W^`-B1$b*RGKAu_rmt^!4QIbbfnFx!rqB
z2oLt(5fSn1(o&o)C_$xmFt1_TJqzoC91YDy6;vS0?f^cBu7rc0T*@!99{vHgF?%9p
zxhIOIKst>9$@&fJ`-6yC%{*}88s7Vel#~?7PS@SN1jMuI+z`YqHJq)A2kjntP)c8P
zqO`jPboZGeJlr9BBFr027fm7hIVkvRKpxKni;F_C(Z!DjQ$SuhdhD2q{hav04w$Gf
z);M(yirPx2rsKxQA6mTy>70`C2U9O$o+^X{8LILA!fdOzkjKJVD0!$~L~b1N9FIQ|
z26&EKGBR65-YC%ZkO7t+y!#gj0+SrDf1YvBgujN2A<|Dmnql)7fs>Q-5M^wF0SEvJ
zC*d;n=3zV>TZurM)UigeLF1)dJ*2rol3gLYCaT0s-h;<*a3S-=VVsJkS#0%?Nr1rC
zP*MP@RRE1f6Q=dlfd%Z5NSTe5;KO&oNEH3#D$H#`G?0iq+{7NeK=_gYz`jH#@E^|)
z2>XY$0ps$y@GOl3);HyjFftB2ih~DtR7t`NOh3J%{I;u{Hl(aJ2N-<j#u5ayv5Rq(
zVtfvN<R}iNC9YLN_MhJ838Xx+Bnto*vOB*srr;*S(_%yswG9!+0@$#;;MpusNVY4Z
z%biFT!QR;G%uG|)uK*cTNs!h>@@|okuqx%}6BLY|A(xf$^C43youSY$ktfv<eB4c-
zqr<|yC+RxF4_62IEh5rn&!4ZeV?)FQVsk9zCu}4=AWxlz$B%?VGHR?7cnH~!X_62T
z6EjL>;)PTixKZRHyN@tUKLm{tMCYVgxi5NI$irNiWjqRqVlaUMbAg_u*prK_Z*|1>
zE7{t9R)I&YP-iI^x9h1OO@l;QDnKc+izDSDvqw<FT_!;J(q;q>UO;FT<f!i8UT1E6
zg$#ncTgM(yR^dLk!sHpmSA&qTJm_LPfVje8%c-IwF&$^h#Zr@+!(>*jL{QR047tdK
zHrl$y4oSzEn25936{!#-D@`_-{2RvjY{b!@k!+v)ylf6B7zh0kC&~M=!#_`rYVYG`
z-7_@nvr}fYu3PR=h*;sc*J)PB=)fAJS?@>d^PkksH+?BRUwv!=%0U;9H4g>5k<~rz
zVg)k}5T&b!jf4|MRTm->H{~HZvMEfx@+v4O5_tdt%5zExG8kc?ht&cM9)f=02-3hT
zh7SU&Q4l&G?=8$6f@cfmV3+w|ln9bR!??uYN7#E;ATjuFnuJOR%&koB*vxL<(^IOV
zIZU9_-*?=$n(n5a@=L1|8g9EL&gdNM?RoKQAFc7=A>JWW>v_Jtwn5Kc&}z9H^ZOY3
z#<nnEi?#VqHQpleKx?ClY{l;b5e?!-Q?s)*>+9=2D9tP_#R_xoOpQ;+mpx=Ame&Ka
zMXa|5cHFTwQKTP2OY7?_Wo2b~c>L)t>`qxTsk{4yLh}4e&JFiM()8uamruC|t(`&t
zzGMzhJo%uN`JGCoGRS3$o}xE4Gvg8y7grQ?8_5wwqfM-=#v22y8)w}ZbA*0Qy9}<b
zUIy{V1dH8n6wi1qeEP5n27@-Yuo#@1yO@xW(C-X#%dubuPS7qWTbSNrPcN^a`1tsb
zTA7c0e0J`SzM;$yi07!oH+|@uyphutI66AY`-<b$hZiunU0hZcNFtFe3-uns7bIQ6
z?u@LjOLlg48r#_Lfk}!SDP%Gm8XBshTdAn2DZmHV<5+Lp@nUWM{{6<ggam~$^7hZx
z_+}R;r{g?4Ja+_Rk~Kcbr)s=(EIHlQ)&@_}nV(*)sR$<9i%Z6<_4V~tcXZqnjcc?N
zi<<`lHG6iaXH<MoT+?n0OO2`pS#0<DmrSAIZoB>;`~#jOB=dw|Q>fQ6yF&0d^))f_
zR%#BaQf?FOT7Z!Z?Gv`8IFiCyuKC>>N74V|KT;{XZt!VdbXj4Ug2%DZDBVz#&%XBH
G*?$A0OrCB4

diff --git a/vignettes/intro_to_keras_for_engineers.Rmd b/vignettes/intro_to_keras_for_engineers.Rmd
index 4e25a30c4..57f1416b4 100644
--- a/vignettes/intro_to_keras_for_engineers.Rmd
+++ b/vignettes/intro_to_keras_for_engineers.Rmd
@@ -178,21 +178,23 @@ model |> fit(
 
 ```
 ## Epoch 1/10
-## 399/399 - 8s - 19ms/step - acc: 0.7477 - loss: 0.7456 - val_acc: 0.9654 - val_loss: 0.1189
+## 399/399 - 7s - 17ms/step - acc: 0.7464 - loss: 0.7476 - val_acc: 0.9627 - val_loss: 0.1272
 ## Epoch 2/10
-## 399/399 - 3s - 7ms/step - acc: 0.9377 - loss: 0.2072 - val_acc: 0.9764 - val_loss: 0.0786
+## 399/399 - 2s - 5ms/step - acc: 0.9376 - loss: 0.2072 - val_acc: 0.9771 - val_loss: 0.0776
 ## Epoch 3/10
-## 399/399 - 3s - 7ms/step - acc: 0.9567 - loss: 0.1474 - val_acc: 0.9820 - val_loss: 0.0623
+## 399/399 - 2s - 5ms/step - acc: 0.9571 - loss: 0.1475 - val_acc: 0.9816 - val_loss: 0.0633
 ## Epoch 4/10
-## 399/399 - 3s - 7ms/step - acc: 0.9648 - loss: 0.1182 - val_acc: 0.9860 - val_loss: 0.0490
+## 399/399 - 2s - 5ms/step - acc: 0.9652 - loss: 0.1175 - val_acc: 0.9863 - val_loss: 0.0488
 ## Epoch 5/10
-## 399/399 - 3s - 7ms/step - acc: 0.9707 - loss: 0.1007 - val_acc: 0.9872 - val_loss: 0.0475
+## 399/399 - 2s - 5ms/step - acc: 0.9713 - loss: 0.1004 - val_acc: 0.9870 - val_loss: 0.0469
 ## Epoch 6/10
-## 399/399 - 3s - 7ms/step - acc: 0.9753 - loss: 0.0868 - val_acc: 0.9886 - val_loss: 0.0403
+## 399/399 - 2s - 5ms/step - acc: 0.9754 - loss: 0.0854 - val_acc: 0.9879 - val_loss: 0.0407
 ## Epoch 7/10
-## 399/399 - 3s - 7ms/step - acc: 0.9761 - loss: 0.0790 - val_acc: 0.9890 - val_loss: 0.0408
+## 399/399 - 2s - 5ms/step - acc: 0.9767 - loss: 0.0788 - val_acc: 0.9899 - val_loss: 0.0378
 ## Epoch 8/10
-## 399/399 - 3s - 7ms/step - acc: 0.9794 - loss: 0.0685 - val_acc: 0.9878 - val_loss: 0.0443
+## 399/399 - 2s - 5ms/step - acc: 0.9795 - loss: 0.0688 - val_acc: 0.9867 - val_loss: 0.0438
+## Epoch 9/10
+## 399/399 - 2s - 5ms/step - acc: 0.9799 - loss: 0.0666 - val_acc: 0.9893 - val_loss: 0.0396
 ```
 
 ``` r
@@ -358,7 +360,7 @@ model |> fit(
 ```
 
 ```
-## 399/399 - 7s - 17ms/step - acc: 0.7350 - loss: 0.7737 - val_acc: 0.9276 - val_loss: 0.2389
+## 399/399 - 6s - 15ms/step - acc: 0.7355 - loss: 0.7721 - val_acc: 0.9287 - val_loss: 0.2384
 ```
 
 ## Training models on arbitrary data sources
@@ -435,7 +437,7 @@ model |> fit(train_dataset, epochs = 1, validation_data = test_dataset)
 ```
 
 ```
-## 469/469 - 8s - 17ms/step - acc: 0.7492 - loss: 0.7477 - val_acc: 0.9125 - val_loss: 0.2952
+## 469/469 - 7s - 15ms/step - acc: 0.7502 - loss: 0.7455 - val_acc: 0.9107 - val_loss: 0.2982
 ```
 
 ## Further reading
diff --git a/vignettes/making_new_layers_and_models_via_subclassing.Rmd b/vignettes/making_new_layers_and_models_via_subclassing.Rmd
index efab4a2b6..3464f3bcd 100644
--- a/vignettes/making_new_layers_and_models_via_subclassing.Rmd
+++ b/vignettes/making_new_layers_and_models_via_subclassing.Rmd
@@ -481,7 +481,7 @@ model |> fit(random_normal(c(2, 3)), random_normal(c(2, 3)), epochs = 1)
 ```
 
 ```
-## 1/1 - 0s - 161ms/step - loss: 1.9081
+## 1/1 - 0s - 131ms/step - loss: 1.9081
 ```
 
 ``` r
@@ -493,7 +493,7 @@ model |> fit(random_normal(c(2, 3)), random_normal(c(2, 3)), epochs = 1)
 ```
 
 ```
-## 1/1 - 0s - 139ms/step - loss: 1.6613
+## 1/1 - 0s - 106ms/step - loss: 1.6613
 ```
 
 ## You can optionally enable serialization on your layers
@@ -842,7 +842,7 @@ vae |> fit(x_train, x_train, epochs = 2, batch_size = 64)
 
 ```
 ## Epoch 1/2
-## 938/938 - 4s - 5ms/step - loss: 0.0748
+## 938/938 - 4s - 4ms/step - loss: 0.0748
 ## Epoch 2/2
 ## 938/938 - 1s - 1ms/step - loss: 0.0676
 ```
diff --git a/vignettes/training_with_built_in_methods.Rmd b/vignettes/training_with_built_in_methods.Rmd
index 89bd7164f..dd47d320d 100644
--- a/vignettes/training_with_built_in_methods.Rmd
+++ b/vignettes/training_with_built_in_methods.Rmd
@@ -148,7 +148,7 @@ history <- model |> fit(
 
 ```
 ## Epoch 1/2
-## 782/782 - 3s - 4ms/step - loss: 0.3410 - sparse_categorical_accuracy: 0.9034 - val_loss: 0.1855 - val_sparse_categorical_accuracy: 0.9460
+## 782/782 - 3s - 3ms/step - loss: 0.3410 - sparse_categorical_accuracy: 0.9034 - val_loss: 0.1855 - val_sparse_categorical_accuracy: 0.9460
 ## Epoch 2/2
 ## 782/782 - 1s - 1ms/step - loss: 0.1590 - sparse_categorical_accuracy: 0.9538 - val_loss: 0.1323 - val_sparse_categorical_accuracy: 0.9621
 ```
@@ -198,7 +198,7 @@ predictions <- model |> predict(x_test[1:2,])
 ```
 
 ```
-## 1/1 - 0s - 191ms/step
+## 1/1 - 0s - 172ms/step
 ```
 
 ``` r
@@ -434,7 +434,7 @@ history <- model |> fit(x_train, y_train, batch_size = 64, epochs = 3)
 ## Epoch 2/3
 ## 782/782 - 1s - 1ms/step - categorical_true_positives: 284268.0000 - loss: 0.1658
 ## Epoch 3/3
-## 782/782 - 1s - 1ms/step - categorical_true_positives: 284173.0000 - loss: 0.1207
+## 782/782 - 1s - 973us/step - categorical_true_positives: 284173.0000 - loss: 0.1207
 ```
 
 ### Handling losses and metrics that don't fit the standard signature
@@ -529,7 +529,7 @@ model |> fit(data, epochs = 1)
 ```
 
 ```
-## 1/1 - 0s - 460ms/step - loss: 0.9638
+## 1/1 - 0s - 439ms/step - loss: 0.9638
 ```
 
 For more information about training multi-input models, see the section **Passing data
@@ -610,7 +610,7 @@ model |> fit(train_dataset, epochs = 3)
 
 ```
 ## Epoch 1/3
-## 782/782 - 2s - 3ms/step - loss: 0.3365 - sparse_categorical_accuracy: 0.9041
+## 782/782 - 2s - 2ms/step - loss: 0.3365 - sparse_categorical_accuracy: 0.9041
 ## Epoch 2/3
 ## 782/782 - 1s - 1ms/step - loss: 0.1606 - sparse_categorical_accuracy: 0.9521
 ## Epoch 3/3
@@ -623,7 +623,7 @@ result <- model |> evaluate(test_dataset)
 ```
 
 ```
-## 157/157 - 1s - 4ms/step - loss: 0.1213 - sparse_categorical_accuracy: 0.9596
+## 157/157 - 1s - 3ms/step - loss: 0.1213 - sparse_categorical_accuracy: 0.9596
 ```
 
 ``` r
@@ -661,11 +661,11 @@ model |> fit(train_dataset, epochs = 3, steps_per_epoch = 100)
 
 ```
 ## Epoch 1/3
-## 100/100 - 1s - 11ms/step - loss: 0.8017 - sparse_categorical_accuracy: 0.7806
+## 100/100 - 1s - 8ms/step - loss: 0.8017 - sparse_categorical_accuracy: 0.7806
 ## Epoch 2/3
 ## 100/100 - 0s - 1ms/step - loss: 0.3661 - sparse_categorical_accuracy: 0.9006
 ## Epoch 3/3
-## 100/100 - 0s - 1ms/step - loss: 0.3009 - sparse_categorical_accuracy: 0.9106
+## 100/100 - 0s - 960us/step - loss: 0.3009 - sparse_categorical_accuracy: 0.9106
 ```
 
 You can also pass a `Dataset` instance as the `validation_data` argument in `fit()`:
@@ -688,7 +688,7 @@ model |> fit(train_dataset, epochs = 1, validation_data = val_dataset)
 ```
 
 ```
-## 782/782 - 3s - 3ms/step - loss: 0.3428 - sparse_categorical_accuracy: 0.9022 - val_loss: 0.2337 - val_sparse_categorical_accuracy: 0.9291
+## 782/782 - 2s - 3ms/step - loss: 0.3428 - sparse_categorical_accuracy: 0.9022 - val_loss: 0.2337 - val_sparse_categorical_accuracy: 0.9291
 ```
 
 At the end of each epoch, the model will iterate over the validation dataset and
@@ -987,7 +987,7 @@ model |> fit(train_dataset, epochs = 1)
 ```
 
 ```
-## 782/782 - 2s - 3ms/step - loss: 0.3654 - sparse_categorical_accuracy: 0.9057
+## 782/782 - 2s - 2ms/step - loss: 0.3654 - sparse_categorical_accuracy: 0.9057
 ```
 
 ## Passing data to multi-input, multi-output models
@@ -1171,7 +1171,7 @@ model |> fit(
 ```
 
 ```
-## 4/4 - 2s - 559ms/step - class_output_categorical_accuracy: 0.2300 - class_output_loss: 1.8162 - loss: 3.8764 - score_output_loss: 1.6820 - score_output_mean_absolute_error: 0.8826 - score_output_mean_absolute_percentage_error: 1826.5676
+## 4/4 - 2s - 540ms/step - class_output_categorical_accuracy: 0.2300 - class_output_loss: 1.8162 - loss: 3.8764 - score_output_loss: 1.6820 - score_output_mean_absolute_error: 0.8826 - score_output_mean_absolute_percentage_error: 1826.5676
 ```
 
 ``` r
@@ -1197,7 +1197,7 @@ model |> fit(
 ```
 
 ```
-## 4/4 - 2s - 418ms/step - class_output_loss: 4.1489 - loss: 4.5443 - score_output_loss: -4.1125e-01
+## 4/4 - 2s - 392ms/step - class_output_loss: 4.1489 - loss: 4.5443 - score_output_loss: -4.1125e-01
 ```
 
 Here's the `Dataset` use case: similarly as what we did for R arrays, the `Dataset`
@@ -1217,7 +1217,7 @@ model |> fit(train_dataset, epochs = 1)
 ```
 
 ```
-## 2/2 - 2s - 757ms/step - class_output_loss: 3.8547 - loss: 3.5725 - score_output_loss: -1.5209e-01
+## 2/2 - 1s - 706ms/step - class_output_loss: 3.8547 - loss: 3.5725 - score_output_loss: -1.5209e-01
 ```
 
 ## Using callbacks
@@ -1267,11 +1267,11 @@ model |> fit(
 ## Epoch 1/20
 ## 625/625 - 2s - 3ms/step - loss: 0.3695 - sparse_categorical_accuracy: 0.8961 - val_loss: 0.1873 - val_sparse_categorical_accuracy: 0.9469
 ## Epoch 2/20
-## 625/625 - 1s - 2ms/step - loss: 0.1752 - sparse_categorical_accuracy: 0.9488 - val_loss: 0.1405 - val_sparse_categorical_accuracy: 0.9576
+## 625/625 - 1s - 1ms/step - loss: 0.1752 - sparse_categorical_accuracy: 0.9488 - val_loss: 0.1405 - val_sparse_categorical_accuracy: 0.9576
 ## Epoch 3/20
 ## 625/625 - 1s - 1ms/step - loss: 0.1277 - sparse_categorical_accuracy: 0.9626 - val_loss: 0.1217 - val_sparse_categorical_accuracy: 0.9644
 ## Epoch 4/20
-## 625/625 - 1s - 2ms/step - loss: 0.1009 - sparse_categorical_accuracy: 0.9702 - val_loss: 0.1138 - val_sparse_categorical_accuracy: 0.9667
+## 625/625 - 1s - 1ms/step - loss: 0.1009 - sparse_categorical_accuracy: 0.9702 - val_loss: 0.1138 - val_sparse_categorical_accuracy: 0.9667
 ## Epoch 5/20
 ## 625/625 - 1s - 1ms/step - loss: 0.0823 - sparse_categorical_accuracy: 0.9758 - val_loss: 0.1086 - val_sparse_categorical_accuracy: 0.9679
 ## Epoch 6/20
@@ -1366,12 +1366,12 @@ model |> fit(
 ```
 ## Epoch 1/2
 ##
-## Epoch 1: val_loss improved from inf to 0.19488, saving model to mymodel_1.keras
+## Epoch 1: val_loss improved from None to 0.19488, saving model to mymodel_1.keras
 ## 625/625 - 2s - 3ms/step - loss: 0.3793 - sparse_categorical_accuracy: 0.8936 - val_loss: 0.1949 - val_sparse_categorical_accuracy: 0.9447
 ## Epoch 2/2
 ##
 ## Epoch 2: val_loss improved from 0.19488 to 0.14489, saving model to mymodel_2.keras
-## 625/625 - 1s - 2ms/step - loss: 0.1777 - sparse_categorical_accuracy: 0.9475 - val_loss: 0.1449 - val_sparse_categorical_accuracy: 0.9590
+## 625/625 - 1s - 1ms/step - loss: 0.1777 - sparse_categorical_accuracy: 0.9475 - val_loss: 0.1449 - val_sparse_categorical_accuracy: 0.9590
 ```
 
 The `ModelCheckpoint` callback can be used to implement fault-tolerance:
diff --git a/vignettes/training_with_built_in_methods/unnamed-chunk-26-1.png b/vignettes/training_with_built_in_methods/unnamed-chunk-26-1.png
index 7fea86b75827cd61acc186745d8a2ae0acfde73b..00956151abe8775f548fa50b0c12311e2402e3e0 100644
GIT binary patch
literal 206413
zcmeFZWmuJ6+ck=}frW*rNLeT#AzcCnAyOhJBBdZ5inNq1jYx@<5)vXIh#;v_f~2$x
z3ex2Q>DpuRy!X4`{b&FDzI`0~<2jB8Z(yzKy5>A%jB$>+JkKf1(QMqik&23nM*fWS
zc`B++hE!DR1lO;_Pxh=1mBqJp`U-N=RLkUl&kGU)si^i*$xEM9cZ?YR;Skg~*(9=*
z$60WOTIL+P*EZ=jo?bgntd%M5C`~nKd@AkQ=rU3!=lkPIsY!ab!RE5NI;z`(k`FxJ
zsk-VxJokkw4t>*dZa1ZkjEo(2g^b^d<nI%Sp67OW_+#zZcUSkDJ3OTi;12%%V)Na7
zG{}M6+P_~~{7P;A{u|#l>w;+h@Be(f_J7GQ6EAgapt2a-B02N(sgUJmr-W5hRNr1p
zOtl{Da9>^=S6)p;m3>Q&+3A|%(_=64zvBKH9D9N_VhqatdLq{q3(2bqoe&KdzOMTv
zQv4nH{q2$FE#kxHI@qbGiiP&`@Te3p;4N;LH>a?V?WD@qq^lIiCI0=o{W|QW#lQc7
z7jv3a{*M)^xch&8N%jAmML1PDu1rOB#k}#Yc}qogwG#dEcomPqua6JHc@I`r={k?l
zO%AoF&`0wdD}Q;)Yrt=`h6K*zNTcd!GF2g5YGoQJm$rA8`F``ESJgFZN~VhxvpXwp
z^t&QJ_1wAiFus4NZgYhqpl+Xc?DiFok@eln-d)aUqnN)iUWH5W)hDXPntaQ)tegpA
zS3AJNlO}E6VgJptK0zfqh%!G`B(?N6K`PipiCp#exNU>{k<v(qjxfHZ*}iDGM_ZVL
z3}mmF`AbZB99DW3=i|RukfS+OC!BKYwbnbot&%hQMxMU68!BsvSK!yBMM_+F_GT4T
zs}-ZmpAVFUiP{98=$n=7yw@roS3liOE<qI(S;d{l$({4_ZV)BXzMZ|dI{Y0yKb~Sl
z?g3e;|NgpS{*oO@_V3reP2`>b`|Hi~|K~61sz?Iz1qKGv`Z5Uny;KU9RaI5xyp+Dr
zs5YvgZ7Nl-$Za@(qDD3MX3tY`2dgjP!aQDHUP4t%J9&hKg$G;GSZMYNcnmhZlIy7q
zK6vEFkq{m|W^90Wet!O?w`Lcd5;S}9+jmQ<>gqc;?dIkTZE25_4{UYni#&)oz*U%R
z`f7fp%%iqY6=y#`AAh!<{N6MnQXlU~>Gqc@`}km!a8*ZcZf-`mpSWz`VI}?QXM&VZ
zTcs?r3qC6bb6kiyd2gL+nr=bj<+o<)N#|KNZrrHf_Qo*7Wp;dqQqB_l{Q2|zubgT{
zF4b?%n!Mr_Ld)*1-K1Y1BNIN!|7;^2hqU)Du9&Nz9%fkmc#mL`<<xm^n>n(TD*Y{~
z!soV#*$wW*E6?uL$~02)VV1bHIN>_|L%2qKEMMf;ecBk^f?Etei@zSRbV)7Gs(imS
z^qSYWCc<aR3O6BO-ppyZ@Kr_1x9OFJoV0YF{YV!U)H_gFjNSgbbvaELmv+uE-CG0k
zdqi$rQH(k+8e5?C){IUmTu6zYMcm#ZMLU<zFkipa%ew9)b-Yh4cA#)>@YQ6kd1K=C
zNO1?<<$TL;*`^cbh*sON4hh2rZ=OPrQt#afItDiVh%npH@1j$!2L6-&WP2ZMOuBww
z4V%GlT;t<Od!($2r-<?Nqb<JCZgcECGuf8kWMgH0pOm3t`Rw66(0%uM{Mq2H;?-Lf
z>7x+<iqVo1u~>^_y2bfv|9e{{Mc&(w>_zM-KkMv^c4r<doVPl@`1dQNbhqBMEPLtC
zg;FF)LiOC7{*HCJ$BN1^*{SD`S01Gw+eD)B+mr9#zxTJlHLsjmn4K^fY)slt={$YX
zPi!zzq}zu-MkQL3#2!tVGFB$ac{Dc%SKyQMd2vW^=F=8&&YI&BUql8IV`FYj4l?*e
zqHE><`MAk?d2zZtM#h`3-MQ0s+V`2D`C(sV9=GJuOn24<gMi7lk@mL=^vi><()2s>
zora>vuq7Guqj_oW9f-cHA%}qmR?EowSbxd(16!_<<r7Z4GnP@#T%UgRQ<hD0iYBev
zOGVENgYvS+`%ml7{P`7d=FoY+v4;36S-tC0TG<ww=#l%oT^NKc_7u%lUq4@VL?w!G
z(=N^mlh=#Wol?pO9P-%1nYEIt8C_+sB@{w9JxOJFNW-wJ_V|Qiie~2FZf%9}#d>U8
z&S3JZuBQfmo=o<wSMO!m4K{i%EG+m;ucd0`K19W)VcDJ~FMa>fqw<SWmooHa+Q&w4
z>tZ7d*4<_C5#k6&Y3V~|jfnw{3zTwN85x<<B~H!Un}%KYX_YJ3rt?PMA04^;`tsYe
z>M56at;?Al4k3lSemG;5JhiNHSyf-0dZ0gDh^22xZA#V2%bFN!ObWDL`uX9W@SaZm
zK!1YA6A9@O8-ddNHp^x-Z>*4Sd5%HS#j!cfaG)XHQ-P;IAOLm7)_>LNbwM))XD0hw
z()8NjDvOU~>9-_mW*T)(eY>uL)XS1i%S=pUqjY)m9KM`oqH5I1vOPfL_v6z$oSdB4
z%ju^1i$8rbsJfM9`t_x>losmmZ@jYT=JIsGqepUW$5V5<ys|pSqFtvn+h+yMn-#^E
z!;GWkrKJ%NyF686WexxG_u~E(vMj#|#V+tCeGs;2J;rO%n(i}IQ&pu9;n?L#KeE_n
z938K1jE+-UBQ@6_`?dFGv!=1-v85*p+@JdeuO%Y~7*dL`B)h~Dlq1Cqmjo~EdZ`i}
zCGxlKwAv`;yG?C;L>~S~`p0`a{+em5h;r`5t5>ha;89Cc>tp5oOFXt8Ji8#)_a!WF
z&#@~f%5sgy7q@TsB!xc1exyCfK)5tI4}03{bZyRohl+K|J;}deHP!R)0%nbDvVM&8
zeGxXwzkmPUDPD-^Cw)V%v(PoOBkyc52d`~^-AwfIY&4oDm%R(g;b-Qlo(WO6I7Z~F
zEi}zqQkkYZ-Q71c2yt>;NX}ASz#4h7Dul#ExTA*|=eNioi45Y)wrG1ZGhMVCX}?5z
zE*h%$tJL=#cy%PDaQm`^Vc$%*zduQX7mt`P$K)n4-s*ornHx$=kO*cC9K+^NQME>m
z6fH0D+7Gvi{Jg!IZ*LJEu;dYo+tVp!hp=|TP#yih{q^;E4r4xLWFsDKrr)t!*YRqz
zbLMQ8SrbPGT1C$2`=KXg)LR&z-n2sTiE&?Eq8}km>qx|nQ<XE$V+Ff2jcb+ZYpbh$
zNxXL8Cs_`glnnbK9sJ~2Tm|r^Z|Y6Q;_ZIewUqvF=#2b3aW6XQ-fAif#g@%f#k*c&
zqbP_X(z?$Ikk+=2X0LkJ%O*YNnO~0-n!dZXn<lKhu&KjS;NA9XUqu%FeCiI6V{Yr8
zOTO@`3vEV4bgbf_zYmkc_l(xbR~KLJU7VX#pl9K`b|yu)K+G^-EkT+0*57^;#6yO2
zyc+tI<3zRa*WM?d04=4Pc=EShNLK%B#`IrnmfIIa=?%XwTX1XgtUww%ocvH2t16y;
z1+W-ag|I#g#R_bd_<N4g^>2NAhZaq~%XgQFYU{aygzL7hSX5rC?>Rve>o)IsTt7$6
zc;iI7-maNGspT-f5WG-DYSC8Prdm)_STq{=h0?5bJ>D{|Mj9d1{W9FD(u@Y+BA7rV
zE=4?v!C1jGchh%*JxI+>nHod&O6psp_`CabE{;0HN2Tr4xrNd~L3t}j5$!_j_EV9Z
zJw&=y7EP^`IKnjs&7MKW{m%#LSfuzsgcse>G9;SLaC?FCv>%Gwe#7HByzi0O6x3#*
zGQoz?j2pi`&iwtVV({_)1_~N!JY#5hI0L!;VHD)j-pEQ^F}j{wMtMC!I=36quGI%u
z%le6Qd6?j7lBNlYL9%D7+t}Q^*@aE>oSPg9H7nffXb)7z@NXxq2smf}Fi9DC*C~rc
z&9FgokH#gONf77C)5)`^kNh1qY85a$J4=!)h!TFiP*hp8Cn#%ZHyvFmV4{9wqACT!
z70=|1R$bb;yfpoHLxL*ms^i~!IsRnpS1SRhCWL^qnjCDR;K@txZ=lsjM3nl~X0y+S
zHow-|t>;3KaT))7gdm)q;*!8rqsX80V+s3o_|TfYUZoX^i8+iiA{4xI9DWG4zq`&t
z&18Fy@ka3~(koi8K4OZI^Jk_k&5tGM3}GER`rF@ix-p~`&Y4w;A|-S)WLjudj`JE0
z52Y2Y?2+LD9@U#{?#M!vgfoEv`sktlC~PwK*3on!AkMZa>H@sXluIoI2=&=cwVS8W
zL?MX93IUn-@Ea)t8`icQBhef55v8_^U>;nOBbH*QNnKl8>rd$}FQ1`wAC!_RYNZgk
zX`WV)t?rz}67mHdx~oY;&v&9moToj@lvdEJ(c|Ll%QEGDOtI!IsTrMa^D+~?)v*9*
z)|KoDC11k$)l!?#i$$it+X%mx1X8X*=@uPB5mX5`P*zrEbYGybcb|T^S>EYs1*u&v
z+BQm+xuD^rcvO?`Io1ok!bP5|w;n$RT>q~1deLGRppc^I*vCyg1mne3iwC2u$^dT)
z<zzcjx}F}pdP*@ugg1m+=W?>$Dyr<=|3S}Bj&gGPFkH{)B~6cXUukJlpx13>W<*dc
zi>3ixsUnLB98Q{6KD2{Uv^-yAjeRJ4V_4O!C)NM_-bzq*;*LzG$$#pDr?7Si2Y%%|
zyMlrNEmPe)2qpRYQmK-HN}{CYqLc4v9oDOHnSeK`g<o5FO&j9$%XjVES@P%Cr*8Cr
zFBEtw(2NjUZS2zlVPS_@se5-;Q<tgAF$LQy1N3%1KE1`$q0?2^G#}8p{@wM?pxk(I
zR%i51PXm7mAww4cvU&h!-;KpbHBwiea(rT<8$e86>>W0y9#6OjO}=u{djjEFf<0tf
zo@*%*67kIkDz-+38vRw+OoAf4Kr83QRc#kaDWySy=XILWD4-8%^0I`l)rkKoUY*(D
z{P_v1^+0@R8OmY$(up8+vQjjnovW$0_#mz0BAln%jLYA)=rZ7TIP0SfbyoX(fS8P~
zv-?AIijk|k(OCUJ8ejS(i2c#G$(wA}>K=1ncK#BLsLQAd<?-Kt`e6we982I46+cw@
zLrg8`J-*6ih;GZxdj;4GV;9LK-Ipe!)r5u}M%wLS5BKVw)u1lw3DIg`b{U`VT$W@;
z$@NY3r{_QHHl5eGlEv?S*ei%lKV-1-ATX|{J&YuUwo^|quYQRKO(8e1%=Uz6w_7`R
z?!5L^5@hGhN+lSK@#gkIdAGd5#lyqnC$%*5RMPcU<pI418|n1Z3Z}eJ>9QJ~_3YaW
ztpR2&ZCw!@^+2;E9*5slakKXUx(~#gRdNsMj{!BtI8TobMMtBAtWZftcXXL@pwvo%
zM@laQ5&}rI!sdp&Vb4S3sef_$(H<$cg>@~RXqr!tJKe05T>wOjv#1>XnxN8{qRDTX
zlAeAn3fBml)xGgyZRxJquaOli5E~#rV*Fk>iN1KN$|ba`7_E0!g#7sGd++9{89Zb~
zw!T%Dbii9v!}JKOwskE>lq2?(v$(sEZ0h+wHS$dk37e!6#m&a%K|p<(A5-3kNxPxu
zQvTZGwERY2JOU2N$y!-iHOyjX{SNvc=d&yj4O6QBa~zd;kC<)$T%Z5(@iW@EJN9xG
z=iv<A>mB)gwXrGj;wQy|K}8*BdqTT`@fG$0qARE;p1T?=^#~Y5B=!AZa-w=tgRD)m
zMjCTMr2XaNz;E&`iVp88lGGAjE~FUYhhF58J>P$QdUV9HVyW#3t3nsq=OfWBe5Uz$
z;*1tuXG7Z@1wQQ8Wk!eZ*76K9MYFx(IRh`l&%WixUZ<XNIs$@QMi_|i-Z$>q0341Z
zZ(AB_uA^%tO-Etj+h2A9q0(t><{DvV)vDaUsPzi*zf$+mepYJWJF6cn+sh(8nq$yD
zOCbBL#XHo@XGHrFWBt6q9S-}(AC)aa)SIdVj)HPlwgM*$iOjNw#l*zy5wf^c>FCca
z5p(K6!{S(-rg3C&*8UZ_pyIp4^!H5F(^IvQ^TU}${;SMwO1_}rv`_jlGgg$`^gy7B
zr00iw)CoGV0(Lh9{|q-FI#4+15IGLE0}Y1v_Bu|F{)ol1vG<{~iknKzW05`Fm*+d#
z`yv4><<m40&+Yfa-3LuHL_Ud3UB|qgUFzUme3#@=F)X~d&cU#?n|74|jD}4~+fh+H
zu{lrLCEJey>h0SRk17mlSc{}-H-J=c37y~Zp}Jn~3*9WLK)7)f!d|f^`m^IbtnGPO
z3C}Fj2Af|q0KRxjy3R=hr|Xj}b-JnWa)@;Tn4;YB-H(vJ_L0FyHT*Jb#Ey<4Mrcx7
z5>*e*{T&FrnP$skLQ<?g-)SO<0ytoGkG4)tesYqY(%JHUy2Cj^!tzKMC`BoO@&eQ=
z!G7*Q09*EeE#3g|#5vBV1^lQ)us>P?>sspPG@9Gj1#+N7--(uUHa~5z<7dI^9moA7
z=lThzjK$-zj~#fHMD!t0Vl=lKKZ`{M;sT%44Ai|i4L0u-E4ze77=L_oJV{Ka-moC2
zxc9GmYH!}W$(MW&3nav$ks2Y;jN+VwoAlP{ZXsKE2@S)uZxt2MfUbB8xXk>?qM)|F
z^i2@o5F;|6Og-BMsm7o(h&=;AaLBjs_wN#HRU*4HZ4z{ug&rRF6>@^@y<jwS*!;bp
z$0d-+<CGFw>o&tsR^MpsFoU4s^vE}1WK)#wLRYDGcd&-8d{{ZOk_e$%0xj@=U;RFp
zSmlje?0Px8ywE45`Cn2%9$`l4S<&`iUMAloWeVWPi%A$LHjC0AK>g~DajJaaVQaK&
zndPNL1=p5O*UmS6K*j7Pwa|9BmDkjz`1UFWSGOksYS#;AjTS&p*t<)cKJd?$-su|7
z{$>^bEMdzQs&lstr+GA}s47dy6Hg7dnU<q+8U>C>p>X(=v5L!gg`SUmuV_zr()EJL
zi{Ed2qA_fL%(PuKLw~Td4LTjQTE}IZ=^qZrrdt*_xSBi<fDc_g{G;Ut0F&3}EQ=tc
zX!Un$Ye`J-34@^Pms=!@Il8&Ip%}hzwI(7N(Oe9*X<WXVB1l|-+pWv!je|^2+e^=h
z%@p5RtyZQiIs2Q&Y#id3HN>KL#^qY~C4Nih@YpT7&m^G4p??;(uEFHMf<?d(bmnRZ
zm)6DMryoClRLWikLg(MBp?xx6AR!1mG0r#P{o2l{-yf)1IH$57hJO_9bfHZvTDny^
z1J$x<smL6%&I)9)|MT;M>2zgp-1HN+>PLscyTkWErbHz#{N2~LTii$uEBgA5TwK~c
z4TMIC^KOmR{^zLE0NFM(i3vOtvV7mRlA;D5?yi+L015X|Qx-K`oWNR1e|kt0=%)i>
zpATYSI^E7__otz;IFv^(YAO_B=NX>KG^&57&T$h7f+=vYr0Y#OnnKX$9I(?M$1#+|
zc;A*+8fhiwRtjSF!+S{4&gj0FrprWtJObM4&j%Sm>R8wLsrU}}pT&3T)AWksG6<j{
zS`={4N<B9`>C_M>&+E1@lQqw&o1fDt8X|3~d`UAyzw~g$ikzIy(*xw1wJr*Di3kbw
zdhcybJ*T-*wO(vlA}|2O@4_&-u@Zz2G&(xSK+SrHAtzEc;@h2xQVH#?ba~!=nP(+B
zc}?n`d$2881(XUdU-uKs*(*X-+?#-uyiJp3^+T-8hd!`A$M%V!;qsef&lEuuzLre`
zYwJ(6heSe|Fzv=}tUx57)rh}qLIJZ%wXPRj7+mZGQ+_IDXNH1r6nhD}XxA71S|4a;
zM=YaCf4t#6pbs^}v>PBo$7O!xU1_RX#8}a?<d7iQ+1sxb2(v~K@H$ToXLXaZj;Ler
zPTF!<UW1xv<(YN0nOufjGf1|VqJc;JEhr@TipPCcA!*vZ|HY%O%A;vMUUb4%2?LEu
zY9u2IXS%#adY>AcPgUJz_h+lrLi)@#Z`CJizgAJPF|4TGCjd^n1`?u`#X(+#i255u
zfw)CbootK7>&wCRKgxU=5`0;leg~L#e|ry367S1m^M#ieidGN>B+)SP6L9=Cz^*F|
ztq5fnf2k!$Mkr<S-9yc<y^W(?vs)#g%`gM8DmjJM;4Y3468j}^Wxh1?fUF_tYLxvK
z$QaLrB%Q{8$F+QI`rrUAm587w@Wu2vnAqIZNVt8cr4D$BTG=jc?L89CQ$D!aAcHrr
z@6|?%3qRb#w3mny4Kue_t?udx)$;?@JmO0<6<+JE68-5}%rUzAWBrxsrO*!ypd^N|
z(xAh|{+{MoMRg_I9txctsLofvT=1%|Wm2=h1LKaU#~lrll^zC*V>ci|=G1+kuAL4X
z((koYtu{0WzUI-*7qLXyDB1sr@rm6Q6^TjpN~v;mbbi8i5>dSyX26!nAfr_R4@7rR
z&QFnk>_?tM<--w0)P`#=-WZu;;__`GL28kGK)yoi6{K_kJNGkrW&QQBfx0Rh)U}^K
zd)7pXV?2STlK~wk?#DY3EgL3+^;J;_f++ryb8I8?TF?ve8o6et&5R<)42K)lRD6N<
z#SL9hSm;NH(t`c8YgDCw9e5U#nK#e2Qk@te<Gw4aFTA?w8>)MYeXo$kQvo-qKc%95
z5rOu;C|3Lfljxa3Bl#$LBA|@f^^+L7h+@s3T{vhZT{7N-1^Y_(>D`(K4<4`^04pjx
z4^;salqDrZJ+s?#^kT{x=b=)YS_xr8&Oj}prD4*X=*?TxuO?YZg%Ld|%28kOnE*S8
z--WJOvLMCTFVK-m(#H9=(H*_$L*yh-RNPTR^9S^p-nh4;q#IkVt7&Z3rdB~E^vH7f
z<KrPh^bP<z%|1c^W}x&wDMRZF9;XUZ9NPs1@p1`JC(cLDb+U<FDnUL0xksrxy|ol*
zN~0^6is~9Fq1!)H=|9>L!Cj#JUceKaXb6FXN)EptO11vD;fOS8s5LtHNoC-^GJlq6
zdn>};7iK8dkZj6IJZWQ*%ho^dP?sS!^bQ|k*zIt30_qY2Aw$R%5fIw<A51&ZM-5yS
zCtFxMGNHc^X1Al(+hYbD)Bx3VD4K*fp!Q>13m|cA>ry&v5R2<O%|S`?m*$2vh2Jg@
z1B!YOrI$m!V!`3~#Q3;A)|TGi9XzlLLWCd0F<#RJf}N!gYzGwkwr)(7kLz-l$I{|_
zIlXbjb}p^#FCr^A>dH4e!u3{BGj3BkPLU>)5#XB`lnCn@(Y~GE9!o$>3sTBMbI-FI
z+Jzak?<3}0Y@jrBWF&RaPvihmXLdActFfnUaYNOU$uz1a9Nz~aX1OgKm6DLc5@$1x
z?xKIa0L->^C(!`BcGQOdXWHSqIDz~ow7&9dK1vcJHH*uQ{N<&2hEf9A_)N`~+t8`I
zApr8!M}P-#S;kKC(1C?RMq&oHc4%;fY-)fJfi*-OqKq}9Ucg8L>eRt5Kv4fe*Lhit
zx(vpDe`c*@qPqR&1!)Fx04BRNbnGs8_DMi~&H%iQ%LwB)E^E<qXF>^37A?ZW%6jPa
zyAa1tMa3{aRx)6GBPt5*lDlYOJO|Z7*h>P~$Qn5m>{yH088-pm*wmRZXCkUi)^=j7
z-w#w>NfG{Gt2PXoiE#7M2vM6D2+;M=R7Eh$i6t$!a(@`PI&_Wzp-?^N-Y{b?%c!Mz
zEM+N;@b_)Vdy@u7(75lmUzK>T6yJ6Pizh)C5E{B0q6>K%BKi0+iI+i_Vu4iRTsHp?
z$*^z!D|ITT+iyp0-s03ZwXzOO(AXfVU^EnDK;$h#?mVZfrIU>)(U6ANmZshnvHC`W
zl2Xv(f*fl#^jzPT*c>>(3}BY?`-y^3&J<9NbdlTQ6F(vf3wJmkT$-(QF9S~32PUaU
zyA~PE>5g$2>%gN0_eD8#lQ$(|7hjaxlROz&X+i|8KeL8PqZ$X6VY<dZdqmlXs3k;R
zCM2CxukbiBnvA0TfxA{73==u!wu&w2>G3N)J<j|xcMasKZltS%xKZ8nlk6&PY&ED}
zZ_K!L-&yInKwvJ7nktCDg>Waz>sc-B5mK1DY=qpii-dswcL*9Lh)GC9LM4zP90nMd
zy_}3?2@VKxv?Qx|Wgh6^eIf!C)jY^-3536ErMi(3IZ^oy4Xmxvq=;kz2`3i&8?PIC
zoWz!CXI^FS(ekyswYTlozHzd_QM2iTJHRNB`nF?b?PR<AYNJ`M=Tl@or~ALXyY7M>
zrHUJgGq`DE(+zay4S<#+ZN8f7oaCFLX&6_&1rha=sI|e4x`4b9P$M#K_EctdlO1}>
zXL#1LWf?u6DEX{W)8^>NH=50ZLT4xX$|qa&^pbNSvs%Mj5~9hCl5bDy61^{zl$%S&
zI}ODiBAGx+d9QA>GTU||%1O8EH@$EH)AK}OD)r0j@;a&y@WiUSjM<Ry(bSw6fuz^(
zP}d!ig_)OQj~Ae3nTB_KorPpr^_TxPDIOa)ZQ|cMkC2i+dGe&^G`Y{eLoGbJI4_+e
z1=+%hVC?ieq{)Ha(PCux*|CO@9I^(Owb6;&1Mpd6dCKubk12D&YW0Asj^GLWQGu1<
zZw3{Q_iLu9>5djmcaT};Spnmvg<0}PM0*MA0f6j66|FF_kiXtGqku+6I6(=RIvwgY
z&!kupYKByB|Ii9+Z_NfPU_vd{kjC7;%!Y7xGAh%jbYT(5I5dKEYD9)l3kz0{5#?=V
zHr*F@6Y4|R-+`~0x(WFaw+54VZ~m^>QB9ADGnwI#8pu~gm=Ohl#aJL)z`=W=4GK9y
z)nrx>e9Kqw=x1Z-y+KUGoMiDED!l&|Cb1IK|DdUzq`*Uc^%SyfFTqwlqs&3I^hKvB
zA;MZ#Hxb?mOv$?5DJko-Cx9|nI?!3<zA%~Q&h@oxxc4sx6{5-#lZ`WwRU7;uvgjO#
z)=QXa31;+_$NeEhltVym@I&8JI=;Lxp*c4BJ5Z&JJ8yW$@`ThfEy_IyWq}|DB3_Ar
zrp6+JD9B#J$Q3q|KKULq-&nkNDYMWU#6}LKjKQF2aq4KjX&F!9j6UPdiv1LH%><@*
zsJ}Ugm{^5SZeEAc?>;Rm#!wkD&<M(*RX(Q3*J?e1iXaO**p=qKXrA%N4pgbUTDX%D
z_`&zVrrjY<+(ZcRX5t5-ox!SRwi|;eSR?=CK>xb+mgh7)_s$($nzx&)mt!eK9<X4T
zs_}QQ>F6sLDk`=}ubxN%l4cVO4#`v@h|;yuy}R6BrEOb;ys9vAKoSB%5QW8kA){5$
zYk~}7(3V(8-6ho@ba1!QGXYN`sv<nMLD}{K(n@NS=|UiKlXgb*WyHi0Qglfdy?giW
zQ`8R^lqL#!s(Wg3pk)n(qbtIG!KO`{x-snYhY^CGj9Xe!9SvG^9DH#pIS`O1HM972
zpl&qsk|>F>t_mDF^^;gpe~dYJvEp>-;vO)w$YIoM0P&XycTYvE^@$IMA!2>Owa-to
z#`LU}I*PIpk$a9i-q40sg1vmjLJ-Lq=t<lhn1awb{IJu=vS{ObZ#%GqNzB#*uv?Ch
zK;VH4T!%F96<bS91U(THH_#rw=E6~d!u+*RE<U%1jnq&JOZZVDvz3Jm%6zgW@Ni#i
zqJa`JL>0q@ECX8x$;=h?-H#wj-f~7-0$j+<gxCQv*zzSp(9yK~^Mk8NNN;BnxA!6Q
z1z6zHO`w@HzSa^(o7jc6>bL!%Yzc-hgpue-ktwsO-ZVre7#5JZD?ovS@gpYtON2!2
z6}DO#_h9E(Bl+Ez7JO5=(0eQ;+s-&Vb3m=qc3+&du9cWxncl&y;RD-|G%iW__lk!G
zGTa_y!qmF4PPD#;-tg#Xy%<I^*ako2i9u4cmtV7aSE0usdh?nOqbMJdLyHziu6N&F
zy~Tr!nJ@{Ug(U0iu`KkOM=P7N5TQ@ZGA@%%8id;TFyv+~`fSAfjhG!{@&hsWeM=_n
zA8Ai$<_@Libs^+c=z+t2cylFws<p?zR%FJR?Q)FU(Ok-mzN~-#n@oRTtzaJ$Ku>%1
z4i^dSw=$y&ugtOTjzI~LHy|qwrfAjc)B22PiYwSW1N^*%cnLnNVyDu2Pj&*x@mz-~
zm4&x%%DE1DY_C?DZNmNKKcBV|{&BtRKUTx%7V)p{EzkJ{WBWTvd8dlE$E-KrM*=uo
z_0pvr`plIcbNfwj>XxMvS`~{jMd%<6ro@|g)(~s%gZw|XqLDL$p;kxyYuG}WmcYy+
zNB|`;%Ms%qAsbm-<TFQz18`4q$k@B0MN9QGT?xQieeARcm{qyVZk?S#IA_F`pso9W
zp1biIZy>Qr#i{|&Nc=p~($KH>gC4DrPblR4JrRWD@9-??J-Z!5{x&g6KEJe5U7;w%
zW+2FX%~NxYXpL874JyM}@HGkm(zWADWCBREcv?8<iklNxWNIo4t<-|4_8I8;mE!-g
zp|aVmUAxvAWk3$m=9KUc7^^JC7gXnr$efEL--=rl_i31l$u2isIC~^++5YD*6ZKcb
z?_NQh-Y+v)f4K9&`ROySY7vVTDZ1S#%D&PER67P%p7mP>nH6F<{G<%MH$4o0$qUnH
zPJfJd7b+^}m^!xL6xo8-n&YT>6ufDu!(IzRs3)|5Yg)_sxQUCf0Fqn-5iKWT$@u9l
z<lS#q58U0v%{I2PwK4^zId-)rO4<P_DJi2bPq%-1wACLhTO3HBiv2X4dFzNqaa%tU
zUAQN51+P+B?7(cpZ}dkVE{P58;Rqrd`rlvwUvTpN?{E+QpKJcFvTZ+D0RjW4#$4|m
z#&l8SkdD_0@`S9n5|k(Ql8HtUBzqXM5u>yuwgH_ikfp`TDr!Q-yHYVGw1)RJgamPV
zX;p1)3$dJQ<-iV<=(L*Z1icgZy`bT0i<AE+z)5odZ?WW)q_UVaYfhmT6%~blj@Etd
zYjmYJlTa)io^oKT&C9dP3nJqc2dl&dZ@=y!j=Z_qXZS<d0=UK+ao<<Mn=#m*TxRLO
zsCQiXQS~q(nU@3U!3UfY)}>dA4W(oX&E;xAHX>jTKP$fYl9cS?=2y?3?}NZf2hH9a
zGuSdf6ljP%D44`ATaTXH51p6Mx@5a`#r{XqAY4`AhF2JDK-Y->Qs})KBFt*(T5Bqq
z#1M}i=?X+)Bk<u8<n*1pb`hJ;N&_S~5Q1&DL(#m>-_al95Dhev3()I#vAFy>Nv6dw
zl_U28)_W022BRm8P7e}X0@cU|vT!MU63j%OArzQ$l-n7Cp9dlA54W-~6I%`3JqJTL
zHFtn*`QgpwkTS7E>w>)1|2@}+aJ^?ZzK~qd3%*NQ0cP(4<eJb({E5mBg)tU1Rux7V
z6ATY*2OHV(+Fisst`0=*qi5YVKijDfJE^h@k$ep+gDzpxL4<;2IKqgn1v6S%JS5>!
zST?y5=(FLJmU56`=64;p%we!~y*IStmmU4#{L0ZEJ9yR%+7NNol|y(SPB%Z|t}b+O
zYE09U!kg}d9^}_i;4C-RQ4k9n+7rp0?i(iIO#H=63_4PXj#xmap2ypZLkuybAilCP
z=!tt*)+Aa=Y8Th#on*2OoxL314*GuwrufdUI}PQ6Xy`)N^=3VIlP?fc12H&SdWHT|
zoXxT7M6MZl!+`_mC9QIPJselIJ_sL}CwWQ6H><q(>cvoh0HH(BTl`9}qUs$3nKa3&
zv1M~<Z>3-$>xtYF{uyCJi+KaXV*UUYJHv*U{ici|Rlg;+Px_t=Sc4L@jX@FG3Kg^W
z!C!ZFGxKo~>*7_^hsfM>N&MJ>9iSX%H}5&NZ-t>v?S=oM3-dCY80;}?`;h}X0NBqg
zbS^J;9(EHxn?OvaK!Qj7;2e>L0;jOh4v#|E9#CLeXIGR`Pux9M1}3PXj^F_>GrS*o
zSpwxoh$4PoOwVoBunT5TdBFO2C26lwjXmL8+}~Xgg)vRs(08}_k?=osAsv8h3JX!N
zpF+Cqi4-H%7*{QVze)#BF|N#?1b}j6i&+12pJgeybcC%xtlfRc4I;w0a-`)s#6z?D
z>k-HWWS#0^j1T#1tbAB70}Ll&G1so?f`(zxCt~QhNfZ6SLJP=WeYn$ss5P?Y@pJ9I
z9wm?+gega_+7Mf^KAcZ0TVstJJ)$f&5GVZz#}#J2Ottu*-XePFl>~SXEfL%@q=www
z2tKv0;(8a9DD}&Szvj&QW|$tHTo?^+hZ3mT^=t;Fq)Kti+m=ns7uHj;-2u-n8-pcA
zWF+fr7oM`(kvxAl84D5C1ectU&fWN9j(>jASe+IC&l*=Y&BQ?WM|05<G~iGNvBv?#
zO9lKT{N4?7Cc}a$BwaTlpUp$*mQGlf?v%_LFRYmy2JQ9VvF@eMPu?Skkci(g%b>41
zeD?}-0jA_?vZO$w4__hw(3<3n<SM`Z{u~$!07`)^NE|Z6{BnjvhpMW!9^*21!Fskc
z8T<C8Q(a8ow`=Dnp>w<!)j7BGUi@q9%@)FDPb=uzT71`plPxjv`Kg`T8gCzHUA3v<
z)nB0^shi(FrHf}~R-Jc^w&{(Cgx_~sx$REDSapHE|JKc$r7^I!9{(*L@J2{T=u(lJ
z1V-V9vm+xS7;|!SU%=&lMoEdz*~QYp(C`5Y&p~l<t>FxeEz6J>&}vU9PnY}9^WQge
z`<7|Ewa{&mBir?#J16#STfYY1Oia|()vtg!_IDOZF-y9zHH&QBx^)jd{V%Nj+x{0b
zhYlZZ(C;YsXGyyrI>@+V`iOvRS6A21?(Tzik|uTgP;2wb=C(2)_lS#&+dH)1MK1*c
z^RT1H3zCa0_%*|pEn8aO+a2ZB&bi&+ZvqQbqgTP*yZ_*bz&|xLHK&gDZ;VY(-<F!1
z3Vo2Lp`k&<*V@{8dwZdEw`@(}j+MLIWk^fD^HV64^sKC%>lL*$HGg6Jd`erJ@l9qX
z&gI;O8cS{&v~Vv<xQW=hEsP#ucnPULj%-`6<k8xCX#t|fKO-Ra_ujsJt8U@w=(tZ(
zQkUcX`uo@8UcUU;+#F(4mwwpM+1UWn%3G_1RCXSoa-8<S-09xscasaVHkY)thy$_N
zvPlpo$x@ImnM!sJj*o3^n$IQp`Dr17J6q<+{@2ptn;p!c`wuHect|ZT1U7dhsU5NS
zmQAA)b^N568OPbPXa7Ji`JU&n70S?qmoE?NWj}vTg=))M6tI5HDPJ4`*~D@5=-PG+
zxVh;TQ5bL7*aUiUf1Lh=LjH4PM5ohgTcH8s9p+#YypKu@r8Zis961{h6~%o%+qgC=
z{mq*lCJ3UPLp>5?*Ihav%e0Y=|0(U}CJD*O7Y%;=+uy%`e+~_K|NMDHL|pt+YwMos
zaN$Q@UK=;l(7+pbXl7>S=jXs9$N2fX{i|wf9=?9fr8#_N-?l?WFv*_c=f^Na&rIm7
zg2J!CK@LIJH6Mk~pFf`tlqi;AH}}_c&6+ihp0D4>;a;F(Ef1Dl)YjGxtd1xG$H!)Q
z1Yuyp?RY%u#5`<Ls(2|w>h_g=Co6d>GX2Vje`e<AC#HLT|Nc2T>bILmx45Z^uD7?h
zuD<>bLa9C5(i;G!3mfGKT?k#jj739J(-Si{fgB$^?hDG}$B%DA1l^QnZf^dmrbgM!
z9)0l7c#lHTz(E%li|;wAF_n+j+!eTCXZPCrNk~qP$cxW%_df`<<RGy2&XRPwKSuqA
zEu8!-0(KZ)1j(n)makt=-?(w?u;SBwH!YywZDVBIj3Vf!rK`(~+j#iq&6~@V&=dOv
zbP(7#aGK_hJzA28h{&0yW(d~X>F8K5MBXp+_wm`Bl9G}RLG!rF%r<s*b~Y|9nUg11
z=^GeW|14gGvHZge<$!IX-vWb!OM#tk{I1wuQdWkVzZ*KQho4^=nhOpq6^pq+7K*TJ
zVm*BLFdn898%d#b%K<CJ<mBw}^z;<LAPTeSDlthuz8x%*E<Y{OP(xY^UBwR^JQxrX
z^6<%%9VO-E-B`@JuU`$Aocrtc<JC8_FKF}+4tC+BP8llopQ#b8-@UytDJh$il9F-&
zWPo)l#R}u%)<CmAFtTUQo-6%zFa8<9Z$;df@=KOhGI*uzoO10N7?Aho{rvg!FW8hW
zy|ZHa-P5C4ZicNXFQ-PX@V|D2`lP>jkrm5+o$<KiHaa@Ge@0MA4n)g$ol{l4gSm-<
z63egoeZ0J?&y&~OIk)?`)A$<dt<2YkF}2u&b;GH#`!h3+DSAb^CU>_}<=pJqn5tXA
z(a?)A+H)jZ-RRhu%9EOvK%>2;Dnv@6^!U#Yt0N>`pB3>OJh)m<PY;%<u*@xjf`Zyb
zZcOd%?F-qcgr7crI-QKlbpK^7=34iSjNZO^a|)iEcr))9Hs}Hulf^dp?!Iv0LdvC#
zf3U(Q5Fr;+wLM{tJfWendwhJHJb|a@2B-{oP!b{21*mH1>&H}lIZ#<y33STZ)!qGT
zY|I}E{jsEE&EVkRxr-NxKAdje!o6ng+LQ9~o~pGd_5qQRToh{sT6%i=)^FdsA$@rG
z`j(<y9z#Xw9~wHXsYws3&#`02ilNPJ-?i%@(l~SQ)!fHDw9s%r!tuI|p57CHWUcGm
z-#gjaLa5KEAJjNZvlC;1V*&!#8p%q+=j!E}|5gT&;zmy;-62-iUzq<EqbXmxc5RKF
zon6J}&nM;NsL`Uxc2|tIp#$|^3mmx2DCMSYvAluQ9ForeyeOQVn>>mFU}7n5KU|F6
zC~2^n`EwP;bz&MO`*a-KNSZ-$DP%!GvF`ms|H2jX<=Dm%C@wTG*Pm}uFlo2wEM!6^
zO^&R4ap!qz>J}Rt8^B*CoFFdCwdwowcR=lXk-J{9z5VES_P-D4sAd3?e+~?6prxf{
zV`oRP*wUI|K>hv45Bf>V&O)6yB?K6{r0nwo6sf4FsPB<z8*4r~Y|;!;BDWnd+?M$$
zyXe~I9haNUQ%y`wuc9kwWy4&2Z0WJt7c{@(A3vCO?b?Mi8iGMVL07I^xs9_eUFgAa
zCA)9e)t%$i%HE9Mladt^a{yhpF56OzG(n8SQME4kT0r#ghV=oaR5pm<)TsCU0|Or_
zD>uSbbJESNuvM(8s>*tP%JSJ5@U%&N%vyMuJYY5LnxS;Fad0FDj8Sc)rM(ATI|ik{
zuA$*>axw@0X1$!8+?sXkn$I}c+6rJYdhEu}6ItdhTM%q@b#=D`0=AP{1q+lwtKyX6
z2UG5_dDi{Mj-}V^GBCP!?I%DF95vj~3u@fm5|fip>FKdx#9xN#l{E%Xmp`{FsP2T^
zf6TI700~3S!NEap4u8L8_wL=wo31<v2oLuL9QgG4^9H!tPeBrY9uxBqG6W+WW-!r@
zxc;{7+xPbze8LTlwiHftvN{28z|O8!i=0;737md^isE=2)9SSl06${#T?d!wMJ@7b
zj3DXfk8*JQSz2@skZf6t(wE~h%Xe#fl<WLOd|eB<o?Y6D06%GB!iG80eYeHAjKTV^
zUq9)&FP3tAvZLJ8*|u%lZJcR5g+GrQ#%u^1Sln7`_hAWZ`@m?=1{xX}ef<MRnEpGC
z@6??WfrSiX%-MN7l>TDqmIJ;2dI2n}*Z7y;d-RBUCmr3aWnp0<O|L_azzepGnE!l0
zOkB0F@WFV2(k%-<DQjV2;hJ}8?3%H0cdgWN39O4(Oib3((9qyu!f6GCO$dB4SiiBf
z;Iwo1ZUu{2#3)X{Jt`<5D)AaaLqiiYGqUBE3U0kEDXPcu3q7<n3<`WskhPAByN`>#
z0Buaq$oK%*Lg}g6?YX!_&<Wb$G5wPDFhj2bAVJ0d1ou(8aV>LPLPE=#X%x7hKYvmg
z7#JYtq2I0otGfH(!CE1!?;e09%`<z<?jh+8@bjxZ*FpZ{XlEK-xLvSM9B>+*8^e>e
zoSd9kO)6B;B^M=hCtjOgMWC-UJ<PPSmoM){if%y9L~}VMFTVk2Z5{w#9G8$#)z#$=
z=<x7ZzX1A*8^41%6%Z7B0gPm6Y5DZ&)8Y>wR_W^MW)%thhM|EDcNFB+SO}KpJ{cds
zN$NG0whRwbSLo`T`0CYZRn_gFTMxs+=)vRv0ghWI<<5-ahtrim!Qg?dKA}clH8OgL
z#gea<m6ct$u(<eO>T_i!A9uPgOl@4M+85xqzWem)o`7sP(b=+oeEY_*Q^u1~Wd$EF
zdXW-0>6MYeyY{v<O42KhWmIx@X8-bOIa%3<b2F6Kk~ySDA}Vx)9|bys&p4)VeN~hF
z_4hI(p$#`!l$4aLuRxZ)iE_UB4a>jB4iZ2_44W4*EXJp#d`T{EYz(?J{@b&n+r!H%
z#&=?Jvd3kMqLR{VHwfIJx*}{|hJR;W-Q!Hx!^FGnyNCB~rV;}K1BZ}M$c2RTp&L;D
z&T43cUU(N3#Z1S{T=RMC?_XO&#MRNpL}y<oB*;Zdx$DLe+`mb?`^_7^N?_bL8gX%P
z4ne`-z~xR4fBysk+lqlbZZp5A;gr_WA4s!=kWk(~hIkMe5{2F8(sW^NuIL3#O?Gt+
z#NnQDLt|r`@gsPLIn427B*l}hE?>SJgd<rO)YT82cXxCYiH$<fJHJy#6@BwLj-M5q
zsVOV_wWC(|O+*9-%jD+doHH{si|{uGx=_D%?T|48P1xhd+se9w0s>xmd3(njy_)Ov
z9{dK}gMV@-rKH4NyLN4l|IuT|&LwB#AjJn0b7(i`GcUBjZyH3&5wd74)>MHz{{CQV
zp2KynTRhCr%k(8Wi!95^$|gs|{^d$_hj8pPx?;Jgsj+eJn^H2;Ngz7EpujxtIwtx8
zu277w1WDZZd0Uzii{)(3L6g}~;lm_;ak%sJL|<*fKq9j+7|%-p+(cA3FCT{2I(Z=+
zJUqLdIB6tCc{YGl#0v-rh_roeY>eB&C~_h=B!qA9)h(`~$k^$zPBkMVqm=i-K|zVI
z;N2L{MOurwE#y=<B`3ef@y8QbM2|+UFUx=a+&ByD@;f)+_q-!;;#n0HmEXJ-m6hko
zUBJOnnyniCB|=m!u?X|bZ3_M&At9Hhm!RU@9~4E9bw9msb1Vi3G#*Y58FE%EQwj~I
z5lJFid%C@~_1NH9W#yM_gGISiXqy-FUcP*(e(KaeN5Zjhuw4KAJ3acXIp4{~3}Tf!
za@v@|PDWAD=hCzfD8=&ki{&3a+_7eCnXP^)rg2_X_2eRKJ?hZKL|8(R<2~KopW0+k
zznIH6ZnA_2On+ef&z~!esB)!Z_CVb}2JW*xy92WM`S?uC4<%b^_s$GGyM3D~qp*r9
zF+Dwf&YpnuXB_T<iYrX#3M<ZrOl9}Mf$$<NjVD7AU<scx654$6t4|S)Cm;~9^&H4N
zR9;s0vAzBM2jd69LIVQ>m>8#DYFWS;lN&d0?x?E4`W&<RE&{IjrZPS*?h4M*M7)W9
z|Ngx`9<!pN;^*(vXU}egSJV$j&U^!1E-*4NS!4MFZUPafXJ_|DNF3ti+>Ee!fZI4S
z3vi0y*o8WBB^aaIwT{P`-G>_DHyv4gf@ZWYs{ryue``jpx<y6B`X%oLU>fb58z(vD
zUePi#`alHeLbKR;^wTXPqr*6CdlyYjJ|g%RLq<l%Ldr*rpwQ56Xn`NlDWwNp{`POj
zDEk5M%Fm&ev<TN+_hm^!ks*CpGk-njG(E~>poaq(rkYu1XYo9m^TENvB!T45oH=1=
z7+)u;sj2xiEKI(77aAMNhJm|*#3QOp{8^I&Ums;MA3b{X#fW09(Z47U%7RC}>wHLv
zQjXI^Z~D7;X#q8`tb1m62I+SIxDk*8xuJqXLs!=i{p~6S;WUzY30ZA<4m=SN5hBK)
zsw*q+dV6m=b?TJJH<X)4@$sxnXlxgAySntjQo9k32kdlnZ?2VE{7c(6TwhvRN?6QG
z#juZi(ykuh;MfR&T#irOxYPq0hAwuSu<*bRR5Q%(sga3i<mI=s>j8_yoplA~nga%l
zK$-`+MqSZql_fnl$*AVI&S%#Z;@sDnYQOZHX`Z5m^G7>d{8^;*Aq!l}v*&se7%1w0
zl#A;@Lc$TU!YF>tt^R;u`Z&6NfS(;fMgG($nC8{oJD3beJxG}?fO=!mp0xosmtVHg
zLObvO;0KLsMb5l#HBYu-1vSEfn?3kM3_o1!=2Q<{U9gRb!8hD`<><F&p@4t@AJ>><
z>o=1)fORR(l8cM$>9c1WM-EbMkS*}I%UOAtpzWiOW}c5|X=$}wGM?P`B=L%s!7p&9
zOW79mRK<uZG`|uQc~xy~pSZX~pgwy6o?K@0^N+W-w{PT<Z&B9-J=?Kk$ILvc&!6$}
zk9fdymo9l=6_E*DAnZ3B9E=w-s><f4$4F;K+{J?Vi{OJVB;6J~z$Sqtx*^R76yPjv
zDWrXop_azR`<a>iL=uAhKuBzSeEb<@Wr)ba$8ZGqkK=Kw<+W<dI6j`SjwZRgr>Au5
z)2$3bo0pcB&L}95R*ph&4~r5Dm#oVy)E%IIowyG0b~q58>@vyD&VK0czjX&KEqS!>
z-?gQsrLRE0UMP)0$-0E&g!g8nIzSOjoZ<9~Ga?@!acXxJviYpUpUFSJPPewUdeA*h
ztwG7WYG}xM^VB_0Se^0HeOz3=6HjRvgz5^bnw%D9c%jN;NFaIUx0<ObI|<!_h`Mv1
zzkDgd;hjD6uT+|wcdEuq!2ij4!P)MljLh2RG`-qZDeEq2ptiG3hrc;w@J5}c60z;y
zW{_Tlq{o~?z*tAIw*ApomYA$8^#jZQEr$H_A~lJ5B0DSV{*x!?eQmewIYtFKxv+@m
zzGi6HbjJAfRj5Xev*WC2dOv^vzK>o2n%;$+cf_s2uDcO=T$qog!J&Fpf94jhvCg7$
zJi^R8cp^b-m_4vy$6~?k$$o+OLF~KNmTBz0*yVsH93B}7K%@bJ0dO%us)OB98gd7D
z9b%@Tj8Mn6wzi1BN7>klJ35$Qcqe+*<A8uW2pZhxF9Z#2$aGLF%0i1=X5U=;d<NK)
z&<C*36G}=nMN4x>bqifQL6pdI0A1if<Yvn+I6WuAby+Xx#(pxg=|mWn;MTW+)~2Lz
z28D!lfn<m{PjSND`yteM{PzYRIT?ks>U#N(l)JnXFM;k3+{rEzZ;)i@lkrBYsi4bT
zfpCh}E1z-@8lJtqy^nWOZo1vjMMg=Nr<n^VAnEoam&@F|VvgYhSs2A^&*(>g{PYP?
zQ98TUZ>|i%)qng4xPw{#Os<%cUFsomaRwyz$s0G)r6pG_v#tpjG!H51X5fd;VmOSp
zrO#9qQsJp|{le3z52(4_0#O7vcJfX>OI*>H4pe=_JVTIed9<@g&ma<TB<rRWA0MB(
zwROh1wsuZ~85*?oXT$JQihkREaicX{UFuh_zBo6F(pXtj<9ArfdCEd|5d?lm@HC(!
zpP*omKOBxbUZvc*vr4ZS{Wp`w8k8|6k_WgjIVDB$cMjfdLg69r{$VlrCT?e_8(CR#
zH^F!GYyr6RlBGpQe%KsNI*0^^=D973#V00KB;DVoM$e<6p;5*gE#>|aVxj3c3d*%0
z9#CvWMMV?CCNgdE-_OVQ<l-Qx=3|P>MV{*|xi-cZe+b@5%I^X11N(=<RMs~N&dUzI
z_`(%ozpJk4r4oazjzy1g;FBjM+($lVPY$;QQuby9{|>a(%CmocP7BIX#MuWo+n^mK
zVf#GoEQh>-FF1sSLlMd`@E?A<rH4V!Q#E!y>sDx~u>}yT6`C$uo<Dc)M1kAlGo(%o
zpxY-$#ap*-MN~b#$$IF}{rV`>W&p3l7o6Y9SX=X=za6Q>5%g!ro%n`$f8;x<asH^U
zulJeTd-ih~{?s!{`2PhbHkG45)h`5G+}!cyi_gnxkgrocfOf^AZtB7RqD0=x!-o#l
zUGs;*c~%7Xeegy#5+RX6*y@;HuEQ8JP`_fJJCw~>m(%|9-k-o8W`vL!BGoUDPO1Q&
z&(S6K7~=5j(Hl&7qSJ=gt{wBUwzZA>yvf_}Vh@N$BR*;(5f==QEOMH7Xx3RMq33{c
zXd)_o#zIR=O9F@reR)n^-q+v0K0b*J4d-QNkt+?qF-<!IxO1#BG&ool2ANpjtKNYs
z@B$tl0#IYqqE(I%`LHMN7axl%UACjmsZ)3Ba(BAAx+W(kdc5p|%HneiZLG$m=*^oa
zdyieYEuBGsHT?a1u^|bN%7cPhRBWuQtdO}C5?_O!ovrqZjEub0@>VEGvPjJib&JJr
zq+Reo63}gKYHDf(#Y8x=?0yVeWqbQ&GyAErPM)Q;snkz<C03oow3${R^vRPa($fH=
zji}z5wMRl7mtkN#Y@3415;oGD(I~WP)hhi?HONVmBSiWWxUh6M><v+=iqB5IyW<XL
z<=&yjy1M6v*$`$e_}XH=WtrZCg3NF9{+zmc30R6rUp?5Lyu7?!kAA^+78ZXf(MwBM
z>iszFbMrgkW@$}`Icf{gdL3Vux{Qtv6BhINTb<mS4=yBt_249`F&n^XmX!VQrQz?n
zfkiAkcH9NCxB0UmH9S0=4t>0A2_CSY$c?OtcOZ7Mvyf@_adZ3Og8;^gKGoL;G(7^x
zE3T^Y#`Iy)S-a3B?B4WSIeU8{gNLh4jnjXmUtJAOv8Eji2eWUNq}Ca@Lp7saQ7ZA7
zQfmP0ABToMDlR^O^EQuFCB?*^h0w`rZ{EC_l>IMX6cLH!^7uqEuGswO=;*pBm=lkM
zTnT~@4V?ta>;sf(Xjcc>*w%wdoa(8nto&27yr{BpFX<pKW(AKPws#Zc4A@aOfs`A$
zq^g7q(cn3aT_ojC-MS@a-W&>XTcJ86%i7K^KKKrlA(_6KNZ5@T%XZu)lF^wn>;Fzp
z#`jXosHPYSqsVMV14?QOJPb`D%O(-BGeWMlt?d^?sZ-_s{r%*g^$kC-yA8eP*p!T%
zoOitUe*{Rv=LWsoJ)Bmb!1&T+UNP)C!$Gv?ofX%#gP{yI@Er*a3GoV+&*h!K&}Ye5
zK|$fp*%u#)riIMC4fC)~_SZLt3YbYio$*%fh3hNUFl!gx1hbUe@u|@t-mq%obA&Ps
z<j8jDvGPetu}DfvG8RxHX^DJ#XH5PAnws3JjkEA7lo=!@Cf<Ja>Zp4s5E%&6?bKB6
zST~e?s3diK(W4MHIx-DW(-lpeamymrayMjDRJJ1Tx0x)17WlrA(&)j1LozZP#>p&l
z?Dfv@um!}Y{eZzp0^^|Cs;Yb8;aa{fP|{8*DsBc<CNYaR=BwPbd-p-m<L45hqV%Jq
zqdwiJ-9&7Hs`(r;4u%FqEH$882dCTj0%r|uEH%ROevN|mdt3PwjtmQdrramk**8Y&
z%;NT(xM6Q}MQByszzr~lY>8cfK|@sIgvmy9X$&HsW}pi7rgY0p;-s+0-wyzACe8p6
zC_*xRE6VHizsjpDo7aw+lLssUVh*V<?{8G!Wtf+nds0cs_kt``@a&)2IGir~;ZE79
z$p<MH?tN-7Lf||OxVkJ{-roLf*K=PluP4b{ad0z+$>7|%ZG9OQn^KCdnwfb)CLW*G
z1IWNZ=ofDQUi;ifao`(%hLX?za6*vD>cS$DD*S6s``y=T7`09)?#Rl@N;j(BduJ*u
zO91URt!xhM{6<&FhO#BNrn*2$$Pg3aE+z~Q0du;ci0qMc5$P7mE<VNpZA%6k2b$hG
zB_$<ndT~w77L1b(87Jl@&F_raUA(w!NMR@NcscBwexP_n5lc=^CS2g|y?gXWQoxzW
zSW_;h+w6>Dq!``y?c4uQDBEWpPBB(iRjnk9X9q4yW|-$QHKO55*$SPm44+t0hw)?5
zpuK}bIYhb#Q%SF0DJ=XjGr??W9p)w_SOLu=U&Bz`u0<poI?Mi#dClnqM~`}EtLS$O
z6V({QN;>vwSXAqG9Kp0S7EvOkW4=w@*T2c=i!+?VDVzN6ERTLj=DDIE-fL-OB<roT
zQ7gZ)1Xrqq!V+*3R6nM4Doj67BUSq(R8{v*Xy+Jm9GI;7{P`+ms1-nq@s<~(M)<K(
z^*E-9b+_TtUGF=B24!>kIh$AVXR~AjQ&Uge`4w1mOgi#Kp}N;)8mnO55Mi&B+&wbl
z3+i<CcP+Z*&tJdp0F9plZ-BwO7<MGW<jZ)m#dId$irl3D817ewKYMoIz=3}-fPH`j
z1gy9U5g%mf6T1KMfzx)`*E_lEd7G$l`#ahTp^fYvvI1A5nisUUwJpIPUbVEO!6<~_
zDlniwfBzD6df)GJRaId7rk%cbkQ2>r_yo1Rlim$m7j~-{-!1vH(sI`Q*OWmPvs{3-
zljeaJ&t<%iLT^s$f31QFOh}RPo2jX(EStKWyBNJ<KohbSu@g;ROsiL~z63%R)J<{h
zZ_7O5uGcOV1&(97>Dhy|<M+|wJ~(Ck!uY~;{^MvYp1A(aHE|J<W9wEuP_?V?xJSM9
zex>b5JKath$~(lObu-Tv3;LePzklPbu7fcBGtEVXNLSex`y>Z51j0v5qQU<`ANFUd
zAZ1}6-`y?f0s#SGcO+X1gh3YOlr`<u)zw!a(Ag@2KAl%nGqt>P=Z;UHC779yCNowy
z!+F|Db{P^Y{ruidkcwMz#()s~qybJ0-eEPrG105^Vi5`<BK((n@N-^_2=3t2Z2G!l
z)?XM>`P{K$_NN9C#<1~BlaXscP|$sZHkqO9Q1RiHxGv?sq)prMKgd(f3`S3u(Ixq9
zVY~+G5?b#<Ha4n>_}sZXH~Fe@_gZ1;<g%?8@0q{m=tZ=Od=3i>qevM;wY+S5c*}*%
zuQz?2R^QFAIE-F|PjWi9<Yr=$SbUt0>ic^;?`2Iic+iIkS9+wYWd^g__j%CXtQ_GM
z^L^Zk(~6cQ`Atnt56V8kvS1z&v`<ttY*Ks~164jFzm6R1gE%?Cla&T-A47ue-&g>T
zR)&`>-`*~a_jsT9h6tnY!3I{(I9vnb_X!>(oGZy)%6t=Aln;xP?sEwdkv%{*f(<{^
z`_5v~>7xTMwM_YJN%VS@w)G?C8*Ny%lP<ez?ul6v#tA#)k-qbEp(CP-UBAd^<I27N
z5)KYRrhW_w=&k46s*>PADchWD$4vTd;o_gLCKEJ-%EBTlRZEB#%|oQvM~KozMX;0-
zwK}Bw)ZQDPvne!8OiW9C?d#nxG}TUuP(#kubcbWW^yJ#B)2frVZ{pvgL)86-8ebb4
z)*OC4xI^e0E_?o60uV)(Ko<~#Y1=649rlAfe>xQyEUy1ypr^0jwV-iCMC1~uUSm@e
zF(;I{Wj~2&DiXFR7c{Kgc~J^upXxn_48DH51PQYgWx4L%b-iXEUTt<x&Js*>Ftc)c
zv54#B+?wPtc%G9n-d}$ZzQPSe`!OFlm5L9pTMHfBV7L=n6zHzGMZ*4or%yd_VDSS|
zIHu@>VyzbUNsIrXb*)#|J1&R%=s}x?77;h%zSVjR2MH)reZ9TV6!#*$*?4$DpV2_4
zDO>xhDQR!uQg5I@Qt~o1s^@HHfA3-Te*_bgO}6T!BnKyF94phTY*UT{4(MXuQLI^n
z1+lD;KeS0WeuLWgJR$uKsy`F8Qkx2*w=f^S6f5)4f>#@(O?+_My$n4F=7MgVR$mxq
zcxbIBMQg5m&`v1J%U@NLHZU<c<GlUUo~CH;PPCsZF8iPF?;0KD{_B#e^==Ckza2xC
zl{T!3wQ^AU%XaU)y&R<qo5iVl<YlL9=F_5=9=Fzo1P5nK@pDZruL}(wZ*|%CAH^yC
z<bRoTwln)5Id@ZaZ%@_zfqDdmSI69abNlY=>RdffgM(Se%de_oR2n(eJO>~}KB(+D
z92S={jcB;_3bnfx4<KG63|}ohHR6!ai2V5R6d=Y1qVREYan&bZIEoKN-3%)Hlj4{)
zf=_o{u-3kJsOvj9T9h^X8nW3xd}MMk;*Y^9kj{eA#fu?DpDQZ<yt`h0P9oFVHAKg-
zYLG*@PKOajO_ljnH>CS&_=<K68E&%j-q_{!WV`SUjAUzWEj2ba?wL=PdIx-6!X^&x
ze`oAWlM?1QydD<=ZGO~|1`a2k^`D<7J?*X;8DXQJ@t)3x$N`V3NtpMuy`=MEs(gqU
zBI7EQ5yGKWjUb3%#vflIv{wvT?f=Ezn}=iFc5TBNHC#;^WNM;8gEG%iks->M6e*G^
zgb<mUBV&<FDIr6M$V`KwQWTksG8Q3Zp1ytQ>V2N){qy_d{oZZ+?sL1g>%OnzJb&l!
zcO2_jYhU}?cOb*7#BY=dEPFZ7v*mBEL3-UMVFbQZXQuwsa-ru92SkFOoHtZAXew||
zm068Jr;F(QojZ4K7gjuAiq)V7FlC<iT%=&oz>wb9i6=R4{`?mx`F1%CpGA1Qk8H~q
z%-$Q{Xz)nL+bmku>bTlh>$T3~!+ujEcoL!d4VujjntB>ZWp{<2vkxUE+)YPM_d_Hn
zaMiuH{m(|NFU*<oByZljgu6qOD;P9LtXTrqmS;1Y`saE|;5NR)tmmux`eXXJi+<$A
z#%JCFu6iTK5^r<GJK^!;6?ccs;*5=rpA6oNo6Q}k=kMYsx@xiSc@g^7HS2HwEx%6a
zp_=uwZEI~^kGvzlthsmg%x_-Hc%&NQ)SO>ju(Nw6@WsRzSkudrl7+~Q#)Uj0sN;__
zj*<2-(Sa~5Of#H&`A^)q)Wyihro-n*_X(q7uh7oM%+Y*Eo^o#7h|!ce8;Wh-@6x?J
zCSN0|A$ea1=Wjg<mXTwXYMx5++iZNfp6J%sImYfmYuMN!2aorG;hM}_ets`%iVbf?
zK~B-r8SMi7k#e-Gt6Z!uO*PfEp+SP8FKXaD_V-9*5IT||I<4EZX)a2aD`{zhojC2P
zH*DM}dnN?y6;rvhsQ;a+ha&{>8~NVdnyk*tXio>YF3wkO$#ouEg=FakzM}Sle?~_~
zn^0%!*Kg<HDZ<zJ_U+sH)bQeF;v{&sDzmZTLmx@zqp8|E(r<)}of#7_Cnwf!TfctK
z&o@HR|Foj>um58`0mRcQ?#gHcLW<joC<&`o<wB<l+_hYMcrAexqK&G%Zk2ynm=`c0
zLbQn{2%fVyuR+mQbW8P%`8MK6jA(8^9rhjRGt@FK(HRV+Wq7!k6Wwvq!sd*E+$c%=
zTjAk{j-8QHUGeCya7AVeEi)%)Gk!mx%a<>Icvp#`;9JJW=Ncd3wRDs0{<0JyZcNnd
z27<Ay@jquV2fu;&Ip0U$?vSEgS~CUm#`deBywT?OxH-D^VaL(gC(n-57aUHw{l8^Y
zhvNd}Ztc%}0Ti$Nk{%ijh7+olM`D6!1TWNCHA7eyQ+r-$t-S)DOlthsWtAWVqm$r5
zLtu9s`ZsJc+=)<HIDv1@Wc2>)(LCk4Ays9*C-VA9LMOLvJREY-=E8*wNWbQ#0+pNr
zV!CBQZNfI0?TVad|8fBW+zht%9tl)gQTg@23pA7W?%OwuI{GLUtcJREi)8bkyS~T)
z*5YBvi3NfZ`uxOb+6Nb$o#3#n#j4Z)VZH^BIGVyQQ6OSd6@c^s)Uzm9;Rg;h{aTQ3
zUI1_S*4eo&72$s|EwY<8Zys@EW5SchWagsB2?@#I4iyv@J^)X^vkw#({oLOG^0%ah
z9mtgX??>6E{qM3-_3`rZjyofkpLmqK;b2bR1uq9s=jgfq8XCHukdW{}_8D3h^XAQS
zd-#wO&>u=BFUMd0rNC98uCm3Q`P<CTO$}Ua?-#2pMF~wu3o5dWrw+F!aX()*WHAp9
z`T}40HKKz&L1*ORAW-zL=xIG>^ocxRaCh4b)<$Rb<ySOQCa%Rq3Ip!*bVl@Y^Y-?Z
zA5%&`>-O)XUi@$Jd>G?oO7)b2c-hWnWPDoZJ9ofORr51nU?nqay79-o!QBF{AEz{a
zsH`N06p=-^xVUH)lZFvQE9{TD-HF&jJrE!^{=+*im;eIVcWfAJZ(`7dN?uq6;15ls
zJ>X3se=yb-#%#>`X9>Zd<z99_=wMN6Zf>5Fa{0znKvtSR4Q{Q%^xnyz1`8AAFXMsF
zPgPxU6?wZtWKwi3TMgg(CqEL+P5EbOI3Mzszl7k7z4NkW3du-m76RZduxl~mCTpDk
z;K6!`GKmD7JEKH=TF4ik`pMkRzG>selh6bJANt}47bN^T*>=jtETu)B-rkErAQvv0
zME#KCG;-_d?&uT3_58vA1%I$IRW^S_WgzG+dF|S@@|3#ZyLaVt#ZEaFUMR%gFXOdj
zSdF<+9Ah{%E`t?{vdaMM@0mfsc_;rtI|;x><~@ly=%6gYxmG92A!^Nq8Rwmds?A_D
zye=z~1NCgjZZsZMlg$+_bVuR@N-gxmGEY=}M|rXS+S>vrJz_uEy}N)}4D<tIq_e}r
z*W#NNR3VjWs#tB%fJWr<QBb{{#R1jgkGt7=O+1IZBA?`b9c+8gR+X&u#$4t{!h-sj
zmP|_sHkirwFO%=cbPn$K6EnBI{{CFad2in;p2tLe9j0qvY|%?w;V)9VL?(bMZjsEy
zaQwl_wU@E9mL469b3C?Bit!e4MxZRZ5K5qoD8sLzH->IUh>*!L78dO!%UUhnchL_X
zTm(4<Z`ylc{0GP@_`ZC6zg-i!xXZXpoi&ag^}u@mHH1oP_9@7y$krPz+$(SnGAPe$
zhQR$inkISv*NJJ)ndGirOOd+hcN%OX?Xr#5634DtsBRoIT`pAr;K2jIVD{ME4wXlC
z2?`qiY(Dz(7x+%#!<>A~dcwlO8}P+yR#wFpgMHV#*wLf2e*2onzK<L|3Pgx`?--c-
z?x;rRPuvrpQ&YPB`NVOY8mAwx{)n$=^<~3<odJ#yQ281|w9HjXTs$N>f=LU=?w<N~
z*8aH}#ZH~u8nzwGNpbwCdL)O*SjWm$K}AjNQb52;;8={VuCCOlL2$qX#CE-F5nG^&
zwHk@TKvOf=_8<@zG!CI0?y67!_VXt(WQm)oskss<TB_Bpf*f@<wSuFMxjS|@^`6K6
zH1xMTv#T)IOuTK;yhHnEGT>}rI#xZjx_=F?zKoZC!x{G#tXskGU=CIW?^-=#0YhW6
z^HX%zS-%J`-=q}tiZfOZnOSko5YmFG<TC*(1oOYyU2k*ue?K3nX|d2ByZH$Mhx-6%
zFKeK7(9&ZMLKKOw_@lki?t1#@Y%Wh^HwW5IEV|+D{*C<zZesMD<#%1iH6>l##zeuv
zmj#z;bL}-MexEoFn1ZJH@lrny?O0}4l=WCRZyN3G?1;rF;+1eT#eW+fZKxqWOerLF
zsonI<{4-au`;whax1ck~m*+rW^3n8soA!ejuMRP!Glf;2EV<_d{IKe$63!@tMVpcl
zLarDd9=VpDo?bJG0-O=YQBmP*uNlw&DPUz~-B3IQwD&Gr7tN9<4Gd^2AB-Ag-FBv3
z1qjslp_Hx29iSl2&^6@~M;tI*)8iQpMcy$=H9Fn0ZWG!i1D2fsm6Zz2Mhj7~9)(d-
zf%aX5I};Wk^ctR=@73S$pkR<Uu&jS>!A-4e*RNL$W)|eKIeyk-1@VMRAQW@Opo<O@
zBUnF7&qOUP!UU~X5v|3yTUb?AY&>iEL{;-J+M{Sd&*Vcc3Yz?~&fTWQa3;_-z5ss7
zj#;keA23uv^I$W(%Do-XTW<ou=sD5a+M02`H$o=MECgde4?rQ@JBAQiaPX8Sy8B>w
zy!`S-m$kccArb9Si%E+IZToh54CG_|@OYGT02}54ACG%r<aLusfMAsw>m50=+MqlV
z{p+g0*+g!Bvs+yvhXf;p6c*DRiJbM98#<10kcVAsAssLv$Fm%1(^)I42Wg-xeR%s$
z^HIl4rNo6^0L8!d^pw>mbp$DJEy@GJ+ysoTjjI#1d%*IOdC1vSL;-5o`6wsOIAq1V
zOG}b*QPk{HC;`DqD{4HXr1Y3sDlTx|id77&_8&QNS0-pXH1V#Wfo^?E(s@+CT(8FY
zQ`7<Gy81e(sG59Qs1vlw`ff_2;%#VnoNRj+r1nwJx<$T7S`9N#U_Gr%O+t?s$f*AI
zLq?}gT|vsMSSY8knpgkIhsRS4z|3BVc4`1B25vB0U0r>6bTs8o^TGuSX8IH_&)T7#
zA`{WPv<NlE*N%?28BerUBc}jcevLssdW@atF^x11eH70bbs(|5Y;0T$)>ZS*+q(c$
zvXq?!4bUtyduu~O15ryss(b061ZN5Ka>{CUcCa4@eGf^j(Ywe>KzdiMUZvO+>a4F1
zNj}eZzXZ={5pdKEK*9Od-@nhnLG%$Xpc-ml8%X&SKQz2yaQTNJCwM0RGWMtQT!F3V
z&Kobh3=Xpg<coAS=z;^1dZri=gE+7codzOjr7i0ZIQ)){!O6)<#Ou+P1xS`!d*Wkb
zKQ?7+nkZShn(#w#v(Cl2Fc%?fI6Si`r@y~{8R$BGxoFIsy@fT!kFIT5<65-3q@P(Z
ztj3Jplns`(H@oNtKfOgWqVL+{y27|PYm*JGf3VrkgRU1TTe6H{uHjwAb34cd$;-<x
zt=kL=IvZOE>zLN^ySEOYBl!}1N3`Tu#zFvn9&R?FeISHv*)^mQkmXUW=z+BgBG87F
zx@eC@0<G8uVnrP@7{M5Zd|G<JiRvA6dF4=mCZm^B)v>@<P)jY)*vyRIWeO!vFGk(x
zb=7CuY&I}tw6W=D-Q38BmUz(iYm_=P3b!Q9riCxVck1<0%rN{K0_S$u$(OTB7s_3n
zx?~GvFZHV_0uy5VJF&#kC5N90x8_@+yPMpgq@}L@u&bKu;Y*wsnAdr}75DEF#@c#^
zK<xVV{reV;!>X!&t4?Ivbp^k+_D1ugCe!B8>f;q<d-lEE@}jwSI2`=8jzUKKc-<sg
z#4_I8+qNw~AGC)ZgIL?t+FDBjfw;Expr8E;-OHGtI3-8I`T6<h9QQU>+GVKked9*V
z@-f`q%MlU!UIS44V4}{*&qfHAu(PxO8tyfqh6#E#r~$5`Mnm1Yx)@I<u_&){Hh`P$
zIvFa)dK96QgRgL_M?WtsTY~l<jcOnv-0d{%hI97C`}d1b0Lh`R5Av7a(X!c`ixKf^
zki!7qBgHO#U`xoM<(!-(P#&pbP6draOvlbe<Kh)$F=lV7jeDhW`Ux1)5XGX=spc9y
zz-A<>o-;bH$pgejSA9AVLc4(0JQnLxX6EZ?))DyzV;)%HFL693V{iu6DQj^vJ;iDj
z1<~c!2Z8{?#)>*ib8|1^O{%?J<cHh{&@M@CQErSH(+a{N5-#JbapA!F_8_w<uc+t+
zD~32r=ypFglmM9LhBlXV3MH%$x=^*KISBok*1}quy;t`EFm#n=)|j^f&K7rsn!w)Z
zGEU}QyEYe`x?-_tK!~Xsn*;CK10~+UmcAvDrGN6*w^`5~TjFPp5+-uTI@_~nPmCwj
z)HF?&M|Pw!vhM#Jasazi$g<(xa8ve{(r~Fm@oU}|;H)=U_#F)eG&LXX=xzj(j0up9
zeF=F#P?H**iRNi)K-ZEcNV!((p6UGoK0fri=b_Yw$%QPMER^ICKTUA4m_gA$rSs>Q
zE(90hVrGQpl`B`CV+Mmb_+Zateh$5Y1w?5=c7+9Z1=Y{dbPF$}x0MQ)g0h}}{HVHh
zVL?cT&U}WVIj<O=xK6pCK173dp=D#9EFK;zvR~j|%k1CZ1ogKkX=$=(C?JM<U<ga>
zcg{&jE{PS*`G7O%?74F?m~aC(hQ^lw7^LJ0Jh#)0d$nSTQ9SDby3eq2q7-g#uKt-b
zN;6J?-2b_wUe+UG3uvW2z{w&;ssNgSzi_Cf2KkFRc>B>_n%IlT6^h>^Vm`n%6c36>
z4*N$?<EsHko7v90Y11_;r88%?E8785*f24=U$ZgK97Em_(~Xiqe}5HJp^<M8wn8e4
zz}hnDgKpfyhi6cG3IZ3O5socVU`o1lPTJTxap)TSesR-1JzY2r6NyM`v_e}hU%Iq~
zn9*j?ZShReOnQZ3P0_Jb{C!>SMo+9sMLD_aJ0F3I3_4zrsuAYZ1OfM6Gy^$)i^e2o
zUH0aWfsJ-L6~MoYErw_|xnfLmKapT&j8J2<;ZdanSQBa325+mYwTxYHm?_(1n~DAi
zxFT%Xuxt0uObPWoictF{p>$%%k!{_oSk-P<xdR!n>LE2XCh*b<wF6@ka1Ll_YX_+u
zL%!tMjHvtCF$d#Fo(6yq=d|E-G-+fpIdi6+ch@dotIx>1wSAP?Yf3M3g@A94BxL=k
z!vj6gb9pwk8P9)paZqruVi!A_rAl$dk%@^135WZ5Pat29VGF2~Rn2V3$(n1x*N6>$
zg%e(JqTz(&vxTodebR_oY;SKL%N{O(qTlbk#A-IS>h#Wkah%zEZ{NoWv0-ZJu)2D@
zrSJ$^gaK_Kr~}i`Y$;k@B8+`}wP?X4aEgO=jEh(c*1l@)jlr3~{}$U?4GqebPi%2J
z4zmG<b<ze3e9%R8g=|;tQ@Q7NfI*{t-BHB*;>8PpYy!RGd-hy(1lR0eS_)!U43244
z<oz2A4=5_epaEJjuI~{DV#}SK6%XZ>lulxj@WXF@-Rp5SGbPseWZASYj&DGXwqRm6
zzixWmq-#Bh2QdW&GW*#odn7qIIpaWbR-FXKB`|`6Lmh0!6O!JR=q%%d!Xg09vlzw|
zk2vGB{&D9D<eBD4=Hnb*abU?Y2~YFgXt8L2cM5#sWxhBX^plj-RR5!&6?JvP9H%BF
zA4=k=C`v}*<=I@Le+H9Fdy4_kYMT~s?2xF^%()X$WAKseOE@~A#x-)jR^Rsz?c{;Y
zGr`)LQUl%h0Z#jBbc~g0D2%`E=(96H8fW0$bg{n|%jRQSY^F}nDA&4m(m(q#d3Cta
z0xi|O>aCWyxc^KMs;H~)la^*c1u|let9bJCDJD||z1K*3ATqAU*^LPxWCCtr3*&q`
zfx!d>#b3pc{ANco{pg4OQ`j@HPr6{16eAYYg$7<Q2j?Tm$j?z~K|){!&CB3o@Mz|?
z8+3lo<VSmAc#H<GP=aFGZ_{CgS7=nwWCs%9y(mt;f^1nfJ0vvpA4^M1Fn(WDR|iV%
zKX!}_ps!J7Hue&ZgpcFkw5`IxAY_hiU|r-m4lDq-#Vt0L8`;kij6Oi$!2M~DCjWM$
z)2H=ILq*R+!rzQ#yA_%+zz7(jwKWrGGqqX}gb6A{;GwZN9Ab#LCorr}G|2pO%s$1h
zu$`jXZi)Z`I~qOg@4te0B;YAv)J+ya0l!G6AE?a~rvbTN!*m}SubZApV*N5q8FH`y
zoUmxfaYwfgP2#Vhjn?4Ihx)?)5)C>BK?UIJK-2yT@P`DhcwC;RrzbH0k$Cn#{RCq5
zN&;!=L^Ui~bvGo0q}(w4b62B>QQ=2TGABlzcv5@7?XqnG0g3JYg9oYxk~D0Rl$0dR
zt*L;qn#-WhXuMUbS^-#(hEsrB4|FMBmIWAyfW+C_Wy*kWiq7=?q$It@L;=&#urTeX
zURoQux#xpqxp$0aJCxJ^BDR>LD;2S7nvHVuedEq;+g^c8`610r3lht-fI{xXK34?-
z0o-CXBclQOxDct}$FRakGYa)zn65vG*c|$k2kIA7K@nTfJY0iNT14bt{2BrV9Z%4l
zTZ?X*7lw*&uI!)tomUTqKA<2@3|*euSkwW&%>yuRFpB>Iv|j4YV(1TYNNdrgEJQTt
zMf2fk)K*l`kifgDYGmLsc9RM=oNHGAQlaze4qn0rP>v3-{D}oC;?T$S?)`g{AYD1n
z)faI)54>l#x+`k4X^LM~l>jN+6RM+q<Xp#xmmwqrfSe=MxxM;A&Q|#D?LkO@e+eVH
zV6(71fT9Y5<>D6*a20t6+6B}Q80SUT0^8NdD-7??j09_ZVy9l-b4=L*wkoDgfuj~}
zCl}3Rh&+UfSj_}W?h=kjLW7m;%%f`*K_hX!7b%4KvD=DQmC=I6umuL$@`|#+#<TkN
zat>srmIzjDTm<IU6(F@$>1>M@05e0z5mV$COql_iTJ8fq>o2m=P6YUG8u<`KtB7^W
zJR^tb_N2VL-5Ak5#;He-`xl2wlG#h*34Ac-Gj!#|_)JG*etv=FeY@`;E_!%*tt>f>
zbJq|9YqdpKeK9!3%j;_RPOGetr2&)M-H3=EiZshU{6tAtdkAw_rIzKB5RwQTcxLe>
zSW!tS((vhu0B&(Gn;{nyB>zRJ_FksS?SEpCEE`nnh6JT>dqkiTh=oSu*+DHt{RRL0
z6PYVmCaC@l!O7t8KPN-5^HHJ{{Wwp6#*fa<15O-w`7VAkj6UPRbT5ICflAL(F;m38
zh%KflKtfAdj$LcF1i~Fo%EkrNQP5KS1iqiP`6t^!MvsUXfNLGlp7R~f$otZWdeLAY
zDcSICH@2Hu^cIf|K`FB4GzpX<jyaEaY^eHlbai8qHk3Y_LI(LD3o$X$`b_zyQ-7H7
zRJn8n7Tv)GOpfXPY-!=@wcHM(St(0OJRcP?LJ)&i1Q+pRos8kzwqRbGZA`VQG_VB!
zNh6R)S7c1gdGuUdPL57XuW(C7z6cls`y<ApG3N7l<#SVUbs-QA0mm2iIUvKI*~`Hb
zlyc$s_dTf1{cSd#X845TP?Q5*6_zxNk-F#SpGp3i_0C^cO>M7<sp*0N)ljAATc_Vf
z$4v3nU{XM>MYNfZ@%G)juYdH+sECe@RnSE<CNd1aqO3ZlyBX<e(a%gBArzCQh?bB0
z#k@AsC1?P*c+}j+RA&BXj$o!HQ^6OUO$uF&`Qd4f((0x7{S^ZPom!?Cd0_e>Bq$h#
ziKGItyttJc<17HV#=kL%7nF~t{<g7-@1G%>rArXEu3Wp;genBn#9t*6k?;~@17Z*-
zK;gNg41OO1_kE;MkF61sF-x`6Wnw@p`S&b0P#w{KC|sk(0cIq-q@(|x+M}y!NE2uQ
zo|4E9-8?5zn$+!k59-(jP5}Xp$7Zv}1Cz(_X*d*?ndz8UAdL+bw_l8o<3PuCM0ot<
zLYy;xDd>h}AddhCpC>sNX~D^FFRyNBMN)bl<=CEM$3kusA*vaDi6c)=Ts*?Uh`@gh
z=_|PD_xkVFox<b^NKY$_^Dz>GMc!F(9l8Pck?8<0L7^)nC-<thmIYZ`Xt4?AenC&V
zapT6)ZTv2}a6Qt)UmGAvNPR35fu1zv7A_&ly(i^bTDE5Ae9WCN`2+}2*O3Tam|5H(
z!18c?*gmj?GBPsgBno{hF1&(O@$fl3M;hTqc&ve5FouEJld<SPG{T8%k|{Tx>`w!6
z1DC)$mXMHG3e>lRX9p4bMMOov;k}p)mkfd`BVYlr2O0s_YFK3CyjTm%EOeHG0(y38
z(%Fw?D?}8C!iIU$Yw8jYcnc?@-XpzQwr0)oo|Vs0Ng8Wz2*R{^5msxmNyR32a{wzC
z240HX&@jLZ9LUmpk|mke(YOeNgDarK-V_{)eEirGv9n@pNY!RM{JS8$f7l&;*E3aB
zUHx{QjwcYIH4^9Dkm>C`bjYV)1YK7VF)>7bRtz8oc<d~PYVHgq7;=cMXoGC1szGh}
zvaoOt>6n0%DN@#{@o0Ua*^K+H@H4wvDO8!sBq=G078>K4HEZ;><mh+<bNVv%@(K%O
zo6a;X>dNz@vDfvWr2;OXq~H-Ro+2^9fj7<6)zvM1M!5sPn-4skoR@L^HdXs@pRe=;
zUQ>N~nJ^Ep1>j`>^E(Or`26s%^^`@Rb_C@OV<3wFJASCDx`x96rxx$&cdH?m{0|Ox
zrh0*0yQEPqBJUy2a5K;*S&Wl{utJnF^iH3@d82Sj+GjqJHB{cz%*}xKXvD*Z#9;y)
zkbHoJ=g1QVAS!ewq7Oso8cr7)u}4My3P9>E)Tzm?BCu>(1zy-N*dWwRbQ=9VANe=6
z{NS$YflYc6DFM^elY1go8?aZ;qC&zp{R%D-P3&W!KxQ<K-@aQ1HR?KWMZrD;zD@EZ
zSi9i=Tt(+PCN_2tM!q4p`W$)W2_vJ;hrdP%!BJ)w;g9tFK*TtYX$vGB;67hRqD~1C
zNwGi}L&YE*OOEs^s3$<WDHy|gmES<=5k|Tw2|(84k$?k|4S#zT!6H#7^AZ*VRoF;n
zr9m=7b~6{)DDl*2el9GG1M{wC=OoJaLr?)l|8+=(C*%Uo1fZDOmy~uDU3!27O51$2
zmxCtp3(DswahPSmaF*5~y1~d@XJJZ(rc$sn@G~DEh;>9;`$cJK;)`5)RSq1JluUty
zUx2BkUjRQSCBnCTU0O<m(-<Y4L6_9iz=71W0Afa{OJCRS$Drc}JFM?m-L(ufAP%L2
zUrmJzF@Z7@c^vf#@O=RM&`dmvD4M57BZNRe<bfZN!ePefJ_-P}e%ZQRVfz@4q?~g9
z^b-S%%=nm=@K-<yzXG=cZqXWEj8Yf%B6U(QDNiOvCoou!{qhp|3|eMt{c^z?hYxSy
z;W=pBh*pybqq{6Pc~?M;EH6Dt0F<@$a62zDHf#hHkgmb+KWS`y1Dt8vXoOuNa1=ZX
z0s9H9gNt_f!pS8@|D76ISYFPArU4es3*4q*-Wo7tK?bL>bulq9h>k9VM%z^^`5S4^
zABh3V7zS_M3TDHhVzvv=Ad|t&Bo;O%GnatGhPl4}<+30=-Y|a!PUzoOH!{mBS7tjQ
z566bl|3Q=c;4PkQFXf;FjWln7ZD8p!f&`BFFD87A<>KBocbg`aB;c~Kw%^;sV=ulg
z!aJpdK>UfqhRKXwLPE3%0q-vW-7dB!ksA?$CWwG}U!mMoKwt$yeA~9s%p(DKQ2#vv
zP7h=pVv}>h*lNa3$~gONCbG%vG|L8v6--v>X3!Y_E`Y`Evo#c&{&E3U#n;u>6N`sV
zEQ`iGFGQy4HTCY@Y*|%enA2xKV)X?&Cg@uvehk%|ZjNK@XdXHQse@>s>LVH2aC=bx
zGQPk!m>$VZ(`CRiHNZiG*24>2(lQJZVD^h^^JeHlvEi-}Ll#+pJGSfzbd6U_x#WDh
z_2AQ{s;Vk5kRU;@9bJN}XeNRp0NTw$V1}3rkV7W(5jqQmBI1r0K?7f;Y$1x{!$(mE
zylQOJ+getMDc+aOy)<nGs>C`l5D|psz;njbpl?{%`|$9wRm%Z~Gav^L-8?k2!y++q
zHy%SF$~qXvoS8j)_H(oc@D$t2gw2eNNh1Z>{0p35C(O*Y9WK9bNgOb!y3g#wX>k>!
zj{=LYQgl{kUxR$dS$q%ve<ME<3L1Pf$VUXh-q*2TysUssX~s_sqZa8!%~SbQ?n==#
z6&hWn$-JZ_%ZDWl$JWx;#OWzF;A#l+syFJ2%G+93Mo(b-V|X=MdJF+A8T1CnUopLe
zo1<H_|GqOOljkj15Xb1lx)$4MKbq8Y>TJ)QLn*ph^=tb2Z*6VYu&EcM3K#TfTUhXb
zG{Z9}f<;zzb>J^rGxMT26hV+poCv@<xetgRhl9-0Kwxy!rOE^-{O&GB%knHKnL!|8
zV8CD-6l1<S4#rp!^-Cw`yz2c40Ntc}EqPuhW{G^xoptk1CCWG=lzj_#om^(T1Vk`s
z`E<VE{0NrnLM*CPW=hBrM~8a!4y;)=@5<Cu`}S{|cZ?>V_iUGb$uw`BvCLJEh?N@u
z+_{uJPgZ)*QtgBUu^m;(hd(8URUNez`)&Fn$|8+JLq<00aPMBLJv)xBw0e<sF2?qO
z(!vc}Qsn2)F75FdNmJk09(rYnYqNWJnXRPjuif=wR~*B3kM!j@CB8Rj?@^h{;3oDR
zx7{nr-WhuGraA8e+Ds6m1=o0TKt*2EL9(ZvoOSXlo}!K_NOBp7m(ZvXsQfMIc>_{`
zY{5?JvH@ROK%*_}o;?8TQk*Li8YENvK|&XnhpI<1Y_2rJ|LFZb!Iqo-M?2p3C|b9}
zW-_7of&<)71nQ)P7`%K`AVsWP4!fN-fLp{R^I@0Wr?P0|7kXDczZYQGxG4kXjpW9H
zIeU}XxHu2;tQD7>dUMEy31{N#eX`2x%EF%zUN@2GyoAQ>WA+UWMlk+yeAGTQXgReC
z+s`3p9xm*@1El&c>%)y!TvCs0tRPkH5Od>NTh1NbZ}uBAZ+F{!Y8`hloTW8WT747r
zJEd=*gGylmY;2xZ^dmtj263WaW!a*H=yg9g2I?^3FvPc0u7E=ho9kG8<L-f<WadXU
zHHTITO41)S!LMcP(^GQKfZK-eBM)n%vB550P*cM04SoOkY6lef85jmm0EHzn%dmS3
zXS?hcNix(L)>W{8QG_^&{)gQ=1W)ojh{Co%*uW6H^#Np}gCD~hVFLt6e#aWcYHWq*
zc#4Fk%)Zf|dzzW^hYN?{|8w_|X>fFz3q^JV<fQ2P;PZp;8QpW(gfs~ki>KoSVt!j}
z#Agzn%z=l8xFot4j=*VJ@JWzdZZjt*%L6=_ryjK<uG<_NY<KHFYIn6Uprr^d6)d1s
zhdWl`%7=d(at(gu!C)MB1-i?|DQGlOK=f0L+FeQ35l(^oCF`G@+?NB{@~L4P&|f}V
zm+bh6877ItmA{<^pA33zOT2+A*m<Z@KskJ3&EmDYB{m(p7y8K5y3<T=tnT2HvvQOg
z88JHrLK&Cg!<mU+*QFl!Dgj&MYm8DY3$0eua?!daPVO`Zr@l<UQh-b?M|c*wZHg9Y
zBtN%Lm_O5o2;7cFLjeaefv}&i+;GTsR9)B*f6Kt&E}9gTIG#A0fpx7gDT@Dnaeaj{
zxBLkdA0s>$H<2qbzE__vT?T`+&bg1WRv%@VTeJTO&cG^Z&&Dor%Q;Ib+kaouGxttE
z=zsrv{=EP3?(0au_wUQsjrPRf*MENT|FWA?IKso=_Ra-cyU54o_d};Zh|GH(2a=I@
z0M9H138i%#o~CFY5FT0ylkO@)(wrc`MUb8>L^9~*RsiUSm9%32{K4zYmrl&nfAi;G
zAN=S4zg-|J+Orwl%CfOzxyV!k*C)*Z7XVulY(FH~iCR1q@o|KEo||qJ&@rb#eB%3;
zpx~_qbLOo*3j^2hua}n;A>!IO?gq9fp)pH3(H0qF<3<L7V{;H{0(uY`?#Hr-oIOG2
z9pR2b<QfEb%s!1j3i@33->xzjX=>@FRoL`0F)+L~wm6x$?Z16J<F^ze{@06XDM|AG
zeaTC2nRZC|^WW=cP_X;^a-031U;NL_`MVkz82+<z{=eQG9{sv_90IGzh72MgxrYtF
zxy!70O71QY$UYW7!mN8vHPoh5)HZI(^n<gb$^+Ccg*4VN$bMH1=R9j~BqNLQ4z2t+
zTXUT3Ix8a8Vm0z^B2X56g&UhU(t@NtgD9Z1(R_;EGdPpp>1%%iCd!d|titST;!rM(
zX`hKSd-8WUuVr8`KGn9WEgL^)yq)Z;QM?W09fFb$l1_RpiiTzf2KJBjj3L)4;C>XU
z)-#5)+<=RY;;B=}5eM>$VZqnw{O78NjTTu=O2Q5h#c2GHGXfwl6QY)zE5Pe37&(kf
z9{{hiIkn<oeEPt2Y`ec{*4%$?py2z|xn|L_ar|tNB1N`I1!TJHuuvR$HVL;N4ggC3
z^(#OwNDJBVSUz$~QeAgEhA+b^PLFxm6HWK(`Z|mG)9=4ducO}++0aLlY_;r!?!SSS
zuVR$i2oJtB>un&Yg)m@RgAC~Xk1U!y0VRn{$gHz?rf<Qg!^rfSV6gUjB+Ex~@Cb==
zw|CBDkY3JPO}<dSSOBo`o_g&Ynj{WV!NosexG!$EMM*dEU^2@V)x)6`%tRL#v&WfT
zvAW%C{h#kHgnroyd>K()8}M2n0ux#x2>;?(Zuuy+SSM!ukk@<6BtHaR4VW`J^l>$w
zBlPoZQ!#YExpVqWo+m=EKpL=>h*75`4w_!r7>gR=@1&OtOkbhOWt4`+uzdG_Islks
zqwjPUCH>I1L|B|W>OTTg4n^~WK{<d@sz|?*eLr*(DGy{|P(aUZH~LdHDOJE<A3`mE
zvl-rbI_+mLAun8{*8b<i0OPFYJ0-t(>-3UyYay4SVh($qgH^0h;Xhafi;?Jq^jOPr
z!pl+>(kn*3w&ZBUxN`b?NL#ctE%G6Cdy>Xyti&>_*i6ferKL$ljM;4ZdZc;0@Vk)Q
zT%Pc4oL0$nWpwK7?~ihFXbsyrU~*NK9X2(hkExV_zcz>pH<urhunKHvhG!wlb?YNI
zeg(`a(ewpJ{$#|yR<J4e15;eQ?a$SJmbb1L0P6)+=o-<-ICF!A=-IvA@a93Pmhkk?
zd^iHVx($Fq@(<XAxz1vE{)jRW=~dW3AF<8E$~pm&lSAu(%D;!b7I+%qro{E<pjZ<b
z_@v%Rm4k5>oz(YN!Pio+T@+YO;&sL8>t1JX*ILxM{GiBsAf!i^Oy7--bT`tTY@1=0
zWHD`Yb3eI0z%8xc5@s<$^yohA1=j_-iz!vj_}T_gSWGJR`15v|<c<#E_k?2bVh9K9
zNT!bup8ohd9oGHRyEqwX2wtzlvO--koJK;g9m%lI`1^A}N|)jWYZtG?TnCies~G-$
zcGiD4vFRDznV+#M95zf}tOoibavZiARaRsnfWgd!8Gn}uz&s*zX&2+36}oS}S3~b5
zUBOAeXR8vzCjFUd`c_M;tA{-80yZau#>OlSlpD&2_G}GMY|X|KHr|T0%}i#CXTD^m
z6-@#WD*)M(jFcAXuuOj$E+DVr=V8-OJ}5O<W5A11gt@dhf;}@WC^U&-=Fw;UzH5=h
zjg9#?@F}ZB*QaFvyNG8UrVl+}Wj+`9<#wv&M`?QmUjc@e>I^jC=0cu~`!nLP$cs8d
z8pltwpM5@KgvTII0`^nAuQ`2Mvj4%Qs-SOps9p<q1m0j)({DOLs{-Xxd}YlgWQ=cx
zzWd3oamRAywK_R{@lGy~xmiaBuswWsZzz){-^>0#B^86iCzPdtc=nlJE8X$9M}PWP
z{`@PLpO)<3*MAbi|Nmvn7n`V7l76+7o7>$0*5Kq{Cj!e)ifYL)+Lgg(1PxOJL)WXZ
zGJQZzEjCNaT?O?xYY8}NkbhQ7_PG=?DkUBs1w?R8$uxENC3D_VGNB{sMFdVvfbZT7
zM&x;TN;F3W9w+ZgF>;B3B8OaF$+X?Ad?XH!#TZ``!6rnE;v4*8--uznV^C}0Av5l9
zr#=Oo<m<tWUrDCz<eollExvYYxY#KW)>Z<uC}@W-C)YTt&aB{LCW~)T;k^x87)Q`|
zTqpa5pUKOXLDkqB`sY><kmT<MK7@MH^nTo%X#6pm6|aT~T<F=C^Hni;YvqKN1`!+R
zn_<GoSVjht*c)8n;#izkeVh%@jcuJPHt9!5Vt5+F#m5_9%O|7Lh$h{E$1tm=5QL!b
zJ#C$ep{Db_h0M)>@VxLdl_svV3Ee<R6oI{CbEn>vBg}%=m{8c01<Zn&xPxwstAs1O
zZktNQdpyY>0(9Z}r5m53`Qj0Qn}w+{c0k+SQ0JBb+Fju-1-Hj$1dH0Jzg{k7A~dr*
zB{BuRqzBm^(`At}8a-G<NH%o4yvV=<4?N?5E%L{61g{9#U@FTB9(TofSUBKM1Xj@#
zhxBI=o_MnqoF<#^8hBgT$a9iUMgb=HmM4&BNHN;p&E)S3(7AQo=omPl=;?va!wMwv
zBNAKIVrN*9lHp+nOw<j~@D@#&KIx5J=y}7L(FDvtK^*4x4q+JGmLgl<;fA3+X1%G=
z>SknRr6&WpK&>wm`F3SF=@XLis2sFq`gF^s$X5rYcm&e5j~J9An;x%bh+j;b`jM|E
zA%fr~o6XW%z#zSE%R75~_NVaM2gSG>a2k+*hn=YJgf?d0ynux5*lhMOW7-ZAUfzX}
z`V0@8fSA&9G<Q6abxW+;d7K@L>9PYH^f;xP$+)i=UWiPM@PS=ECuK?-9hl!gYxy^u
z`LThY(MvAl<fj6H*tv%bPF2%B0k>ugrQg&01h=ltUIEuM3z6FGwCP)a?p`<>%Am!b
zci_SNTI|`T>ikLD70m5JQDXRPA2OUt%xLlmowz)GGTwK<psOooetE-%JKM5X#5{$o
z-8Ku@AfEsAf)V0)+Z$Vv<?N$v^Zo-8xMuhpd%oPeb|@yB=Ec#V?bF>4K6xQFEi0Di
zp@w};f4<+@4|^C3;VYpdzxTF%xx!{L6=%ju8R413;KuG(4(#$Y+Re+!g3HV{`xjip
zSHKuk_aDHK$hMaOxt+E~g<<P*$_>avknqjvC!%qX2gkJ_nGlg7FWC|yDEJsd6%ZV(
zjpSy8=3!E$m{F+Jeo3vP7qD%;$g8ID5B}&-g(pttvZgCvgKFW8)h(VkTA8#TqT3Om
zh0*e&RSnCgf?A_+4suk2r)|uPt=EG&;1Ek;MRs}}EdT`dHA@utKN4Lfw0T*;L)0!l
z4qwvVhifOcP*e*WC8IsET0*P07<kP}G=AKlmWpjdM~)S2fkdBjh3|-FUbJ!1%wLIJ
z_a)X9qa@_pR}A$4b~rw|hgf8BH#<^>h|Mt!^oXg__AU3L0$6B*l}399Jy)2;dw_ml
zgsshj$*!A_5H!>^LK!%1TrF6wb30}YJ)!$QvXBUn&5L&Nv<WtVrK0f3-;Z#$2%y**
z9^}<c-w3ObiX}1^Bs=H<RUy$-+|cuIAJ)gFOomf7#%W}n+Ql2+9N2;1>XI7dK?nf<
zT@H(p{ruO9yfA5oxS>+QWF8raTfxw#kaEcEQw$6pK_N(CgZ)r-6{AqSXY~#!eiPi_
zy?gMMf^$>M`f!qvu8uwy=1mx7^5E$mZm%1Je~n20JN{<4?Rrrm3_nDD#97c6?~nm}
z4t#BBM-*ITKE4{a&Gxvf7JBP~d2sD4zI|vn3+&&IgPXaL06a?Yn#igZflEl9Ltz|d
z*xr^-&1{s8gAlOa>*CL{(ZZcB%fK^xClZar2z<zU54TQff#!l{*(P!>g@2<vwI%4%
z?!p<B-o0}im6t3yr`AWOO?5vA;JfW6({V!%Vf+Hd;E&+VgXw4K;lpRym`<ncb2fJ9
zVHiO8$LhUcbpf~!Ysoi@P7K_DRTP?Gh_&q=tfMpf-flBMfJ1F`Q>;f-3yYA~WqddT
zLB;@^f@(iX^Qh5bg<V#N)O*#{t6oxkw-lqA!t5K|LoD+|QKz$EHF{&cN#qUX_7ozI
zTn5B-gok&-<dFSNoRl}|h+2cj)$rs{{#wR0=TDFo>SVj?<mS$C@;?JJjP1S}9;#;C
zDmYX6ddyR<5WqhPe(8W=MLUhM=`>z+R2@TcYB&Te;~+*Vv{rKJB-mZCICZ0u)|=*P
z;e_N*bU+0dxnV;knirq&_|<SN+MSrBvWUaMXfi~7l#0?XI7C5L%c&ZGTpzc=jz{+#
zO-o&P=!~yQa)yG8c6Dv4iF%PBN>|Nz$HlvczkWh5@oAL+h8=RIdX%QGPdVQC*pm}`
zs>;rLRT>8B-Wlo%k2VR$KZCC3XoPD^y^@5{dn}5ex$!3OG*x;Jr?j+_udazXSLr0{
z%bG%2H5MJbPbT>|m$<uvT}MjvCV9INqz@$=&lNM{!0D;XfxO;iNE~aHzmyJOGu;a)
zI@Q2TKa^dC5yt!jQkTQWTQ;+pp%ea{XUYi;|8T1=Sd&H{?j$d@0lc4)&b9!>C@iqt
zxf5gY(-#?CnTWXwo=&68sh)<Ssj@XQ7-9ncI_r8G(8>(dwbQ=*NgWh0EgdUaxQ{O6
zg}4zRDi!1Oz`Nm}br;rdn5}7loJa@SJs4vTciZQ)p=fz0Hi=`6Oq$7>&Ftq?Ws*yP
zodu$VKKjRLiRR$hnv8*CM&_dN%SLu({pA7-9Id5&s9j6}+i1+NAI-2r;6S$q4&TO(
zmqRr;OLH145SR<1D=%Q4=^Uv!UT?w_L(#c=Oj&UP(tGgQc6`F*u5MUcH}WbJbe`v7
z?@l2}G(HwIRu|mzBEVQ{5MJMiO@7$7#EbIC#UfwbSjT$MAfzo&H-bNNOfVxS^rDc=
zsm}b#_$o6&!L=^KCp}pOw>H3j*RvU0dkykji~DhaT$W59e^^PJDc+fbb!HE-WE;%5
zAx{XfVHZ#C@A85qDTyDEx5-c5U2^C%zm+a9onG?V4$S)<CHB4($1Him2@Xo)*(YPC
zr`niZ*zyHIi&B&F!A86K>5TAbf+ijQ+QrZpk}w*pEW)w3l#azBUc`5+j@U;+DzeiD
ztbqzBmPl3Qig@9ju6H@i#09QHKCtWF2~;86IBeEpk#(>DxYc_mg9u>ycsRW~r}}o<
z2B+gBEJVaI%{kJFFz7C7(-s*fZu%+aC}a6%<zWA3J!Gv5U;DJfodTL`oqwx3p$$_i
z##1k~8wZEga#X#>n;={7o9v2=kphaK{0`yIXZL{_w%w?cx)^j+r>1LGS3NoT%5J2}
z55Y^SYtj24i(jZBI3rL(Xjg~tI=L4!4v)IZ26n8c9&b%Pw55kD!YBPJB+U$v9cWvA
zUjtNVmgaPiP+)FtX+Gx$a><(;!$mV*y=$j>*NPs48j(O-X;{eMvyM6la=u>eRN8Vt
zZ&VF1QFKwBlI2aP>LgUYFTKugEsvygv$<=uWbkeBvo{+azq0(Gyi+YOgOTRj><rTU
zR!Ce7e?sr8nm_OKF7hkqXdEsQw$AH5d7Wbw+40%rS(qa`DAdXytz;cT$PN&a*HKYX
ziNE{{zB7-yj#1AW8Ee|M)H~4WzLh+cnBDDx@|?aE*xAYO?DjMOGoeu$unb;ad!F_8
zXlz}HxT2F=o!ox!LfiOEuNybWES{{p3E<PqepV{c|Bf{rIc`ZZvYga@)Bb0#mD(F2
zEa5>3%!P~Z3oz@;KbVeCuvp^P^SN<ACv^TiEjo}qC#WpSt%e@PY6=2XR&-oF^nn^C
zRGD#Pd;~FMxYx&(9n<u0L@bd|(zR?#<PT6RR2KfSf=(tV8x|4XNqHCMNJN`+_FSG@
zIqZZUG-c|k{d^oEaxyfVk8!oYXHwPaS7=p$+Y$K86~TcQ1|^~0+RT{9eOuFrnFIj6
zX7^3$Hn=&J*iB9#Wfg?c`a{R0W=5&uJ@~xS?ESFdE`s105$+(56kmf(R-Y~pR+FCn
zL?Vwr_Q;GX1oFCuXTm<x?L?Ge0+X#wq#gB#!kAF8(w^UB^WL244A>4MEbGmukacl@
zbMo%UF`HXm<Ki3iVY+SZrue^hy)=p?X#a6F-@PW{JNZ^C@76%kE%ft>fLS*O8QpUS
z3#l0Km0XuS*N%T>BqSHbqM~>@ek*Sv!E$3_KYRgiSUPte>xC`YZp^Mu+iN1IZ^|1?
zOEL_#uQ_==s;gv0Xg4Ot*x@u`GRFUfD+Np?N>TZ$G?{60jIx&9+?glRxpQ7eh2CUe
z;9~<ydFRGV19`^((56q$UtgM!!z6rA=tut8drOz)Ak%q*i6%vn83?r(IKKtBQg-sM
zM2fMwc^uzmQzF(u5aC+qQEw5?mv}(P+8CR#)ovczJ;K$M1F_juVSz@6_Ary3f*5b{
z1m+p7JA@(ivek8^15sRFVZagH+LQs(-e(~<_bT2+Ruu(Wd3^z%6J)im<2+DlG8Rs!
zN^P0T9b=y}_|)a7&JvS*s7}d2vSQMXLy5O*$Cic6Jf27oc6NO}_o5Rcs-BEh_{mbo
z%jeBCLE8mmW(`6t%TPAFv29Ws6GxFKKlMhMqdZGvy0)QU@tmz`e2l&V9#C}G)c1!z
z(`hA8H;fx2$!#2O6Xx$G`1<jT-%50%4p+NQRI`p@ijxY6GLd}TG{N2AAvlAYDBP8!
z#O!m*@pcQr97Jn4qHfa+6#2fx^d_52a|X6bNL@N}y|F6{e<mr{48cp5_WrYc%ucZE
zE374pRdc<q0Up8)=Q*3l&oMTk*NZee$gTm&lbO>%Gugd^aw9Lkv%+kBC)KUJ^^KE4
z1eDvo0|~Vm`D5uD7X=UqTZjXJn7r(b-Q0%NrTOi+qd1^TMA`s|-k@M*G6o2s2^*j`
zYIMT}vLE#fbjpE<2}`@tf-6Q@AG0vVLWR!H-#G}Q2zR*$%!IT`^O2GZPZyT=!9N0z
zh0hx^KnA(he%w0j%X%pf(KUc4D2v`VkZM=s-r0wl-Mm3%NfT&Br6QZ^cdQJuh?y?k
zk;V!Rc0H@N5NzlR`*Ft?b&!!HSCW^)?M;RImPq>dGB2kU5{gVVZNYGLy3&i@{F2|n
z%G2M?K^wQfXfoP*lA|}@(o-qr(559)<7bN|x!N1av3VR69Qk^^3hg043lC9_3wk#q
zCnRfEmQEAzP@wAqmVb`$@Tu{3&k01J*y;)Rv<CD(AGCBGJBG80G3}_?NM9?<7}R@o
z0?Q}Y*YBuVHg(KsQ;<b68R<#@_%a!j8e2e{y$Cm=!n>?+$9ljaC*R$Z6bDV#&+Z+{
zU0_4YwDSkAsfBzX#v0rGX+Zow?`ylsFCfavSRQ%ob*8<%7Csx|7FuQb=}6=)a|aF=
z@haeK<IAswXqUU{k>MZn6r&F);80^*>pE^#G!e|%-iR7J1Sq{Uzu%8*9ye}WUb0(+
z@OASkoD-*hCY8KjTw8S7@v8nNI<l*Hu@38_(x!MS9{hf@Iea`e`~}Y9@`kS|*xVV2
z>voym*jTg|R{`NJJMX(2fCm9v2<r}kLvPd0B(>{62m&5U$rt0MftS7=^&|F}qJxO?
zljCNu?Ra$S`G1XyO+m=OxlUMgQn$Rn2Mk2XuKHCNn%4&d6BMMZbh$}g9-fUR9w}k2
zaxz8f?lT!aWKg?9U^#S<oQfu)hI1FWKyFuf*MMJ;#mB_Xnll%o)HO8&Y^fHsGwI5|
z0IBL}k5eyas%@Z1<T^g`RM=?hX?Ay(S)iV34Jx0zAXgDfh^1H;u?DJ%kp+ysnbbl+
zOe+yNjS^##22OfFSF2BnPkFy~+GfMk((z9YcepEQVOyc$@u&-0l`pUp0c~a!S-Flk
zucbJzpm9K51GN3kVdvC}c-e%`+(3~jNV!-`@boj`b)6i4Y>?L<))Of?TiF!H2Ftzz
zI#ZG39ZjeN+tG6Z6<fjqlSQW34pzYG5Rxt|&BkC2CdhuJHrox{@x>!6!Wm9&Ta#im
zXEVU!BGU%YpzKV-N38)2%HibX^L9YPiY*4673z>u8r%c`tW2a0S^9g&XQ49BY<|>r
zfAKHFnhMuN?J0)oZH{F@`})GGkEI}@w-70@h!<g2iTxfU!#Y>$;{>})VKY&3;b}lw
zGqnmj6RuGlHkd*WHOv+|!yYxL0V@W3FrfrkL$$q*wH!tx6>EamARGONHBMckR7^V+
z@h0VsniBcXw!t1`v74yOwqoqzmfbsiO?^}hlnaPq4iNI$s~bnF()qk&0mdMOO}hA}
z$-{b?=GigC!RJlsHfXMdg5`elz1iP$<Hkd8v8^9>j|@_tE#8x(ZlH%H-|Vk`(4FXQ
zNW6DeikwC?B-8h5BX=(~;Iz%bW{Gdax{<+HxFP9?**P5;UL#XX@gAXK2dup|o`goQ
z%QM=7%_6`L7aXKgjio&t&2{Y|k@irV2o(S3Q)tc_psB1~{pJxNqzKGrKVb%O13}F&
zXJboX{<vY-&^RI8l-ChP)pZx-@gnTPCM1M$P(x0@nb3=d&nC3ogRDNIbXX4Et20z}
zP*1+8U24ifLrcqu-3E`l7zcOGffqkfgAx*4ah%<+Yy1K_=pS*ho(6Llv95rIxMDrE
zw21>XsNcT@i5oqfL8#1#HjfMuQUsB_Sf6s%1W>t)@!!4!cYaGvjg-;a7gkaQoVaj0
z>Yu_?hqmPr!f~K%LM)?|m|aY&%p!z<SWgX(oe3gn7n3)Mp!Wq_Ab+9_@YUiu%sRf$
zrqH$Wq<p2BwRX^WJ72B~`C%(xV0(*pbSHaoH_{Xdd<(+NQw<tirzS+&?i!R|E9{Sg
zTn5qn3JLE(G=DE*hs|V#B`%~gjowKcq<~|Ufg|jxw>X^Hn*bV>o<BZXMyMS+pxy)p
zp_pEawe#5Ov-%;)l{m$fvk#(HndWC=R>5G3a#-O;C!3<9$IpMe3d3RNgpSfj$1HON
z&81xkDz4HtiV6$)@iQlrvK>qh7m^{k9J`fFRWEU4pH99)Rl=GcJcYmvZY_aU<hzWs
zO%70QQ8H!7p;kpKIEU;DIDI6TT!&w+)d#4D%}o78tmI{cY*GSFh$TGd-eyhVzgR8v
zMozN>CNnA4fqV5wOb2EBhz@CjPmD8F(bV{$c&UR6XBi@{F5!k|99}P=uR$L`rs7?n
z(t*T9h1mn-s`~N48L^liu^dzwpecmtqmMM%!L`c};~J1eS}z8}R4g1X+EkRR4+gS_
z5A;k;_Bc#UM;W*A5uCHDAy?x=yQ?_^9nS*SsUg>5j3+fexW2}N?B7ukFU54NH*Us$
z%ym)c!5h6Gk&%|$W)j5{k11Q;k^Y$paHfLzX^<*<k;@bT+i4<8dujurTTQ}_J_mN@
z3JhY&cwV4;pxtTw4=s}O-ydAphv10#01B|E39&}x9gl^*g*8zkLF#LaTCs$CP_K&O
zP*r}sD<lwdG7wZcGm+MJ=>BiGlzS?FsR(Bp@XR`=PxTw+);OH1%SCyrYdLlthX_l4
zH-gpaP**&}QxgpnP0abPGf2qBY{G_)&cp3Vu&%vmJE}c}=n>z)CQT4Q?eUmpU21H@
z;FM9?NbMhl#>6XW$B`%Z_w3$Tb-jcqdwWl#)E{Kbyxk7z$0GuY9mm**FRvY6*VS;L
zJJZM%)zUyp4|)W*wr<(-dN^%LYOq^e%)=2X3b*&JE0F%aPTA`-!&VN?+4I)z2|o8>
z+p48ARxZsvKWFx`CHF5a*!orT#opyA(s2(Idz9Dh-y1l4{S^&GvGB&VlPQg7QuxI~
z%oWb(mj>C~3L5dR4GVKVJs4ERE^00F)Cto)mDSxqzAL+4;q&!}ouL#N|Gs`0-K7=Z
zC8eB4Q{S7j^rA)e<5{-LgyW;s5G@&*D+}h$J3D4{6O*LVuQz=yJw28Jsbg={u|rgT
zq7yB>-<zEqfQ_v~J+LF|x6O=DP1po!qO<37OHXxn!lsoQr=yLUrAH!o{n|JgzIS6H
zVEVOX_Q87++yHm{P+A41J3}H2n2HwraZ5F|TNF$p{H@P_DMfy>AD_>g;{pv;AiGOS
zJ5`nqpa&z50PrP63cL1suRJgehAq)Zu(siO`N8xkFnt2>?}78@GhTI|aes>hxw=1R
zHa`*5{e~Etf};9Wr#NR&?iP4M-+?gBYyL6`PBh@UyU|*tm)l-_%tw?X_Y}nqDyg6B
zEnaJgbZ&9{PC-FUl*8?J%#-0+@yrQ`%{8EXF1h5^eekF1$EV<Em;nDxT8Q2d?_M=<
zF{F8)L0^r|c|)Yxi{tr)gbsuGajqKTnzw*q9YyoNbjbJ%TCag<EF#&|`WaWO>$&MD
z#6(Jv4TXLViD*U(jxHz&XqO)=$G`MXH?wgNQOh%|41#t$XGKOuiAI(p_ZNkwZRG&o
zYZC%N+65JVU~2OhFTR5_y-w37+X=_$HP=}uRBlb#jPw~gfP5N>je^1u;>!5?&oBPD
zscriD`jI%;%Hi{+jdSSY{*c3pZvd;e03P#$VZ{Lr4gX3$jHUG<2rI%r2<_6BZ7n~c
zM=3uwIT3|{)?2>5Cv3tO-bB4B56j@n0rVs`Lcf^jyXrSvsK;q%or^`(+Jn4%J^C=#
z_C&9x*9v`*Joq4ckGj^I+_FG$IDFp`huvlz4R>%vyurD>C_o$1#Kf-PgWz$dZRZ{3
zuEK(X*9iX`ar^`zI<Jf@x%v|4x*>Mq;cBdDKb+lJ_vNluBXMel7shRTqqjg#w__09
zZu;}T&Tg2#5ez!bs6_gJS-izrABmfZ8KDjE((*@mQ~`Xk2`AL5ojZ4CV(tdBOC|W2
zR<s|jp&~hggKMD^?-<^o9HV-%n(=oa&!B)TCJJD)8jxmgy}TS45*)y7fw1Rm(1Z-3
z!jxiAyjaw_<;G>b8AI?6yNy$*1n*&69U|qLchj-O5t}XmsuOVR@`t>Nf|8Oy^hZdB
zil%FzFtighT6Yiy6;7Xy&d$a%q}P26yt(_KUU-B2iW7n;OsuS1m4gNRFtL6FfZ=*L
zmBGw{6H@LN>0FTs2_!g>y#_RJ3b(Bfdu~O>7t~x*eQ1b=B)USkxLGYqaV33+;gJyq
zMMV#M2l;d7Qb~LtWPpuG3w8=s?`?k{vKgJMa@}^I7lD`%Vjg&%=C2mr*$55dTadcM
zbunQE31`eL-2b;gl3OwU<LQCy{_K)pDs%76$wL_U@j3J9w$W+u5Z`9YCKy^JY%*SP
zn3e)Y%3#Pa{WbBg48Q#R{EuuuT?WtD8U6jkzyEt*gp(UwvH$t{DNVcdf4*uiM*cU_
z%%8t|GD`v7)BpK8!FUVl_Wyi6U(AJ)=zqT4IHtGM|NKL7Ut&1@&zIf0|6Cx$f9@bH
zivO%Zx(okjHi@O2CO|l@d-qz-W9!~OD~=Ns$}`41Nk@_+G~zVgT$Lm+(jywcx*|@!
z?Cftsut;KBz53s*#x5JifS>c6kJ&g#OS@^rAAbMmH*_|0b1Nhs&+YKdV!z|>4DEgH
zW6vHxjH3V%8Ud-EJ{W44@<6XAL+PGGtu4p&$HX9wnU5UIUikASI2~d5`^y&!70)ez
zuKAPdlZ^v}3tIuv*q-#kUocb$;duCmg|A?}EpMtI{mFlS{bwoB-S|JVl>Toz5H6=@
zi!)@r1PEq`d~CypJ|xj4fTg~C=FBKZ#kCnAsX&NUoCdcxKvaJo-8$vmnQjq1_z^yI
z5uEy=xyH3K_xCL}$Jx&{QCV(>#Nvhx<q#5Kf$nO^h1pt^#G;9F7`&2`Liqs8`_OaD
zh}B|Hkw7-M7oCCImXlcHJV=|)?ek$+VO!8O#M5NQa8@!gaMrrXymY_7nG6L5qVjKx
zB)=hfivEf%^X7f-#ckxX%m*i-%nL(_wc|k9FF$8K$h>y#PHe(EIQieA>BI$XCm6>>
z0qU|w$49}7Rgi1T7I_+9Uc@T!Bt8L9-5bCvEdVI}0Z}L-+xiSsL@0+uqI+wNO8NkL
z4N)MEsXg4kh|&B(Kd`<xpbxg9F~o!_2Fa>5oPD@8<NPVpiBb%p4`3BAUblYz`lc@3
zUmx1Qud^ms4Np&h0I&x@-iPAGjT`bHY(ybTDvwd+r1Tq9%Xg|F-_I#>i#q;}fv}7Q
z{F_>kN8bt!-HvJMruO#Ds1R<WKOhePhzq(hXpI?UbVCoS_t!69z#w<uOF8sKf+?uH
zRU_u>*My6&rNVFG&tF3jlHSRBV$_zNh+@E2g@CP^k?5n8gL%oVl^B|x?V*1L38po&
z?q$nNKE?QjT$rUb+E=oZ%94>bA6CJ)i*jG&w&CHYKm}41T?)viYojcHL+G(_bK}(T
zzg&Rv@kpg0zCN@Fpf2T`YVoD+B(EjHB3PSyC<TR22;H*^dkYY*BEhI+TDtVB?=zhj
z0qt8vF%8NC+|?IPQ1Q%}#}LLnKsO4Ptd?$}>oEG`Q51@V#U>>yLmj-CR;}6ubJ92V
zUC2^ZNAxWic3sMtXpsH9m|+x>I@_S)<_lVbJPKVHa4DiD(}-1*rPLhEk^ofvitbDs
z|EYd6U_I27CAwf%Q_lrv2W|!rCTsc7q$3**F;ct_p2)H-;VrSS?8?ZLWB6=NpER=(
z?yYYzC2+1p7l}1MxybBNF#M!qwUX{t?p6vBQbq$%->Mc%mR8T<PU(++c#Z&K{edV`
znbg<idepQ!aR-3m0KiV}kW_oXM;A<g5A>`RYf?a?sG>q0xcMnC;E|M@#(2hqHTeb#
zBBx;Uor38Ota9%M){vbm&v$hV+I8~Mw@7b%%4?KAtq=+cz}qW;z52i!3*Hx>^XC4Z
z%LUMRRAJGAS6KeLa49`l`0j)TpRh(zaj_hI7>e1xg5`%+Z{>IBJB2a!gxVZb5e|Tu
zwtM$ozmyGiYFG>@;5ir$*nDek-3Z7q5S7?l^y|7K*38_}jY{Ygkl&FuVdD?dm>7yZ
zHV9sjAHI@~jYANh;VFo?p`tKITMNkkHrT~COXw<4Fo#!)shp!?1-D=`z=K<;PG7YK
z3z=^P5yTI$?S2VKNplN(K(bW-VOM?9O_FhEDu=Z42-=in^DdO$ydY6R?#+%38}@?L
z!wFol(%c3u8zy%4?Zh8`ASq>6DXNHt)e5xro&Do4kmsd<L_KET?uU0vE-c!H4F)XA
zS$J@sTlTW_SC!kn76ctb8H;lBi>V~$vi!iZJ)%>je)eV(N|#L1?3R^%0~nQS>sCct
z=nxxPRlOF@>%_yEW3dzJ?s>ya3#l4!<5iStL2JjC&+zT*&ZK-E5J&}_4i*epS2t&g
zy1jJES&_liE4;viorzg(E<^^mE&CXT&E~1ms<Cpb#;$=!d#^6&?mn2@2Us)?Dn2V&
zS>tf^o!#9vj}g_Pogbo*1ZJTl(^(@3_l0=FL!bx_WK=`Ba^D10(%nvTfMakoH^hB!
z3pRBeV&vf=j@(%Ar#etHrPmz><KU-h@vB$+azU})%FkahC&|+sy`-OY&#?OA00bT}
zC(5B+8s~;M3`lh##PFQgJi~vcW><E47iOVUQ2k%40~D>D*SAs8Uj=e6=en=0QbGJW
zmp%b1l8rs{`%^|n@n}-Z!9#w<_gp6s6hA?n@JlYsY1DCG`N5UU!`<D=K2WC{9U5Nq
zDeNi8sf0j7BWN%gOFA=K{Hz~1Wk)4SfjvDptwphXO!7Do(h?P^I1ICC%OS4Q7X>3@
z<J*dgip0B{@N^4{i`gvtL0^mm6oEC<*%^<S4YtjBtaF;*+W3FQxcp<N44PY*0YinN
zbcsUJ{uh}Y?6TZZKxhz2qWUUT5tLr_W^N-k&zDJvl-S!peU28h<QH=g3h!lS-|b<g
z$*|8~qV%t3aL6vR64yyUD>iG)^ZNDcVvt>|bc6;~yf$Ng*GegzCUH<&5Yp=p<ug>g
zSTAAI<{1ag=J&l$Ae2-hYW$O~{wXoA{d-OF!|u*s2+!HsH&16+9G6(;`f$sJopTcR
zk9%EOcfBABVo5(NQ0(S^jB`nGj#>Bl_QCMoJ8+`=1KEEItry(F=e}rSXopU<g;<J3
zP9ng)2DZws#_Vw&Keu`fukrx|PUxh)1#)3q4WH?>c#bH4Cy>DVASiKD?QLQfdlyh-
zy8!ss1=4PC*1wjU^x(CW&w4X(y1qEu3CR6!M}NQ<491cVS>Rfg3ez?KUq;_vs@oPQ
z4#lkzoE2Q0oUa6?FywffHhJE7-+>V{m`_21lFxP0i4uy>K!#7-l3+eUVTZ4CgDOy;
z(;hnjOK>Ce=<dE%!Xa|@dFsUea65(!bsV&X+0_UkThTs#ysr)tPQf^NMIw1tKt^Q#
zH96a=6u1%m_!N8l!R83ai5#F5274zhGYr@X3#NR~^UYrO<s7Tly@xH}wR%z#{!881
zdJK2$6r>L4Z0%G^6q8%FEEa`8q2CnUVh{$B<JU5rk^Af&aV>-6s-K!77?+*h7WaK&
z-Q^2ikBXnckN81|Q$Qc96+|VWz+CHA58yv6lghrGt)<1bAi@l@$u4u?-g4$f{>k~G
zT#2%3LT>s$2#*44-W~#t@9hZaXOD=_<~w^0^}x*;9{eDa>EQl}2d1`~q9+&E#<`e-
zn~$&DuqQCyd9c>V8ltN4olr-XdmFej+;S4&VGHIzOTX7i?&01@ff(Roa2vS1Z$R2@
zfymn(I*Y)9*aV|O<bPSO0lpZ5Gtb1qu>)ihKLoG4Z;@aq!3L-mc}{oD3Wmbx(A;jt
zt{Q2}|6P9I+8W(y&8bre#EWAWzV3qfS6}|rL|KO;UE|NXw35VUVA?OL#o(r{?TY>T
z%Tz)5-U<3`YJ36~UJ3Z<U2zYe@Hu*uRQt_-QFv%PUT-0HG7b8rwa?5$va>H<&@eTP
zqr?Ydz&$got&m<&s<241Eetob(AI|G>U`YI$2ao7ZEx{iy^rpu3BuB<_9d(WuOD>b
zlI3vy?H|NcCC)oKDq_{NfMR$DOvtxTp*#hObss>$*PtA=;ucrz$cL`Ag1r2$&-Z+M
zeDLVrptaZnxb_xs#W(1jZbm0~)uBTZNn=3C>Lh1Ac={<xPY;}F!60GFdnCF;;`X2|
z1p;<cKu_pg$#GcTfLeMgW5<#uOVHGI2X$!=2GzI_hi}o1z@1CT(*q86=k;s(SLT^&
zX{MF53j2WamVgv}3JtUrfXO3Zz>q^tmSKQeit)z{@#g;9wbdpkdR*H>!h?dik#X%f
z+t_1^?CTafH6fEjgM)XV7p?#jK7a4fQ1$lqA3wOU&H^z_Z~&>7Qvbt<2y$sW*v0SG
z#JyvOa^7giwwopF9C~HJf(Nh*+vvt20e%A~)g6FU1O8273~Bp<YVqUa!|RoI?UA!&
zmB7$9Mtp%4W@k@<$IP^54G*5RKS;*XHeYaN)k=z`@r#VHlh1b_rY5v(9I!MEfzi{C
zB8v}$<ePD#I27aDZ`^p2r)u%l!^R)kLJ690w=7+NOeQOhGBSwFwGv;DjbmX|)+3>&
zhFwTyMD4pn3k@3@8u~iolimVKHw2xHvWWLK4<X6LB53MgIhb+0TSrHjQRS|0;H8(I
z!)!-iueX|ED2;{z)3PtzC_dyh4{AQxJ4fA>lI-s872Bu;`jH0uZx)P&p5kr<k2ZpW
zYH!bY&<+TC>$h)yJtCknMNt(L<Bi*+E!o|w<Rj5{1OE0^kInE5gAH9rTAAg{g%(`I
z-^T(D)pJIdwF5UHtCI{1J1`7X@g*I+sR!D-(I+~mG>LLe6AtBVS%@R*E-o%v=8(32
z^t`^y7@=IlGN+g5BOi2k;(4LvY?D<lh<ot9506BBLj+{_K4C#s;;a;`&P5}H&KZki
z`0yRSgv5`6x2*v_!6)R%=X?$+atlJ7`Ae56a=`~ez?jXGolrnPjqMJnP{~PL<fH8O
zNDpz|#Q=kUg!<{-sNny@)|<y;o$v4CpEGl2tEoo2R4P=6LP)}yX%Uj_I};&$mYeJ|
z(^Rr%OOZ_3vS;6!$`V2)?yUD@joT8kZ{O?bd_KRwet(?DIiF9(eZSwY*K@h9>v=s9
zwxiinNO1zHPyW=E0fXWCOR((}zD)GA!uXW7K?{bLa+g7lB5T^*Tl6qPX^oQQq0K~3
z)%M^>bZeppHMp7^YM2x&><l5CV;+Hj+tu-bCqKbsC*KO96+L*$6?-k_j};@0Yns)C
zzwMNSP~5rqK-mXuY<fH2iYv+t+(dgsW@n`T5NuM#-lWI6!oZ|i9`|En(jR%v=r)J~
z@OOP>{m=IZ=(V<Jt+v<Cj&?9PqWg}4=hX^k!z_Jl%=~v$Egr=kTA4;NuC5|1@iYv`
z6qYDet#gp}+uY+)sDxzy<G3WQVsqLcK6_OEoNeYDLFPjSLO^R6W1YHXP&p4{J%t8l
z=&}dO$^0t1ok04@nVHr!d!+jLI%)v%h!t~Q;|Qna>xche)*qZjT6A&UQDo=6Y2V5@
z)nAO(b>5ug`sU}2cNjq9o4f3Me0<_Gm-~dC9hQ|ff;RX7_9|1v#;zE(R4t(&)~^r4
zxX7Kz$jB~RgTiY+V+)P|G!tPcSQG(I{rlT*6~TO_cQdXXy`)%+^L&LG!-X%FL??h;
zpnmZb6kt?MLt&{U@0eZ9InDs1fM~9St>XLOYPK1jIBptI89%IB7mBJ}0kt!BfPZLl
z<p@DwsYesj2QM>lK>IXPG(s>Lqyh3QN!l@z<Iu&2cM(Vx&@{rqjrtd$Yzo7z>4IPj
zLQ!#abop3dZZE5<qzsA|AymZIu6tm=)om&OKe|d=6~Qr-f;xQ~48KI+dQ<j=kW>M!
zM*^zGcbNIYCv#L>1i6<DVvbgrEP<B^jfoMW4}|tc0axINtoI9FFp+Io6GZd{{1^_P
zd<cE?NEpBNE}BV2yzB(fKi_1qwte;W*Zq))sAXSKNM&oKYAI7iY5Dfa&4H8d!Oy2=
zXCu*c*J98lEuV><RS%?`5}}*D;1q+>XgCOaZsCd^{@ZW=lgf@j%%(($V;71O@35zg
z3FmEfP^E_<Vt=fWuBTa$j*7b;aV!DI3O=^21N|7i(VSGzsRqegMyEMKV0>B+FM#9n
zBVv;$&T)i5)~1Mtekz@ti%Sst-w>Px&Xn&~VN9I}Nb~**1V3@`F8DCuzDv^Xh<(L!
zKAPEv<Kqx4gyJZijE}t;DAxO=XCJsu)aU)kh>mcHR=9YPs4ilB-luA-K<L7c(XOmT
zgG1<r@JkP3#32%UHR<g9`<rhRFb}cL^HCUDVgxU-X&K<4h)D4!ReoqFw!n__2CDp;
z3~&FXEj4fj)p8l{7Q()|P?)@<IjMjjfEq;PW^*5`6tKf|q^m?5k?J7)((j>}L>V1I
z7{~v@3u+}Uwj~A6IkD*As)IL<cFenObUgjg!vvr1R_75mDBMS&5uRNg*gxL=`=5Vl
zd#aY;2e)x0R08p!6E9lI!G51^3jDSsKYhB&ubUnJK2290V>t>~sOTSlSoO(&ri}oJ
zpg!UumJRWa*gG8g^T?s`5Rkuv`5=yt$kvkLddObvuji)V(bN$DhyH$hB$m%9d3u(%
zUDc6-){hq%GFvADznpr!foQCEcbkVGQ@lbQ15)~&GVZyL%ca9o8&zRd%8=i+iez~R
zl%qs6rEetp2!u-|EM6@atyoz|HGsy(Ds5~Xg*Vy&L9FaR^jaOGsF=-`{^3n;fR6Cd
z9v835(N>NE{d1icJB_wd4FcD6!+)6d+ll_eaA_UEUoah^i;LG7im@F<Xe#-Dd*QS4
zfj4x}^Fgu5i%K{Isr?mZR(SXA`w#SOJjCkbD{p^#F(I5NL@Zh3Lz50DfDcfeVx+JT
zVqU1vGTZpw<I~>}40YQ3w)4Gp+XT*Exe|}a<cC<^M#{Fz%cCSTfQi8oVRlzZ0RlY_
z6jsVnufT`)1Bl4i83O<wfvJa2`SsI|gZ~t}J^#sv?WAd>s2Wsp_LWjreBofiR_0vE
ziN1snFCr$e&n*ApDK|AbY6xKT^$_@bu!a)FS0ug~>OJ3FtX`ZQ+CH;yrS;s#JN}#s
zUBVeWX2D#N7jNFWMSZw9_TVLitDzqm2DG4oCi#8Y;wNO<TMJi`193Q8VAtsHdZGUU
z_;xK=4oJK|hhZ0GTl5+ompBu`7a*0n(0?RQ))^7Ec~U?#<^I9|ppqzi%VO{4;*YXd
ziPUz7m?VA76<Oa3&8e2WNA_{G{5Y0%LikDajPCFt)WADQpD9B{Wd#WH*6=7URQW8H
z6ZEDa`bAB9T)+Ml_IwHd;h|Bp#^Fyo2cF>CCL?10>a^U5iH_6$%P}IRw?3Y<DdJH*
zqk(=}6IFIq@ot8E4_Z+Ta10R;Av~6qfkfR3wH;vumzI$dw#zjiE&cq5WemRlQ>tZh
z8gNS5op-AZ$9m0UM5lRA@;(dlbcJ$V1EHZHIq-!5aKzbVg95u4^ER-|(d~-L%9_Yo
zQMP56XPdDL>Od=MWyeJDJYL_nuuC9811cxy;5%m{x<A<?U5FbrVUFh4tRReL)w;zT
z{q2RUnhOdf*y-o$#$sia+CK|W+b3X4d&98w_0N9|EI^X-YDdK`q}Yu)<H5kq`t%PS
zO04A^m9ub1`6X^z4=6%E?WL~6TNFwS?W{77f{)B!Aq7fw3_GY?SKtW@DY3uE^bMl0
zwcRR)0xb%?gg7}ltzbb|%=2pCf|SayVCtLGo8S3zBZ8Sb6eJ`AkdlZwx&%+N6x_sO
z5s$|A*w}e`@KRBelVe2<@`;%Ac&&7Vu%Vt8+87B|QOu)98M3QOPMQ!YcddB1yQcs(
zH##dX-=R^F(^cqzwP9|$BU8Y#e`K20`+6K$;CX)1q_Paa)HG`HWYou8W9rEAE$GZP
zw5)=&zv`WVRn+PU@8x0}Xh?u1A~(Br=gzBz4ZDHZsN#4^<SGML97e&Wjg=k}d8#_L
zta3wF4=AlHvmSPT-vlYgdmAilP|}1%G$&1l#Trw4;}i}#$JKjo7+-hp7->n3G;gE(
z${9W6{Edu!Edr|EcJ>_lv>2uvmx!9n+ui&fY8OU7=>!#axm1o0_;%Qy?p*QN;!whF
z6IF}#eTplrlUAe3GmT*#V*+}I(SnJ+kHyt^3e<5}Hc+$x2kq*ZBo)Gn%VH;;P`)~w
zvQcy-qs8Lt&H$c4Kjta0^3xsDD4bK_?z**(<5L=8hT;YFts&VYr*0_L!j41TF`)ZV
zNRxsVp9%_|IF_A~zs$`a%^on2bah@z5AU5)^K)EE-|EJnFA(I$UnR97e9?{nrqD_`
zlpYHR(VDm`8>W*05z*}&hOX$=@Szo#U=N|&XZUM^!o5@unGGtx$0FqXkL&kAU2*G=
zrBYjzI{oVtr`taOmu-=#yC-BVU|P>}38KM_MfW|jo*xo#TyMNGUiAuJy1g?I*v{C`
z4co>UXiYv|X?s`;q>eBzt1h=2vnWS6%kO`IJagy9e{!<3-$T`C)hvN5^~*B8z7`ws
zcuCt1Q>2^6vTeios!?8f{XIG9&og~puG!jujt&WVjFw&fuco+Ha)<cOeYHM4EFXec
z#5U*_6F*A82HY!k6lioZ+Fj;AQ=so*V`F3529RpI)dfx7o32LmswZ=0%En`#JPc2F
zRc(1&yAIt)^K#lYP4_$J^Z)%4(-#Fg@`(BGQGLf0{7XV$M4=6@Op+Ze6wKL54gbs7
zdAL5^Z_Vlda}j4=!nzlMY`(R2PxH<gbjR_!x5av%VBJN>l3@%v9f(Bag#Q>fdn2@U
z?J#>OVR2NN(dWG5RCyP4myW-Lc|OCwvreW3n&da%v_n+V(K(KqD-UQ;m;2PCL1>we
z4g<A1)>#dt!v5tB4X_d!5FuS|mY}+84{Gy*ZbXawRjoz#Pn>cI3LQzxb9z>4qdfN0
z+m-#VyZeGr?ZE!V+{wo=5hBW~Mstl>AWr_<f8jc%AY$EW7u3zZQqJ9e<W6v~HXgyz
zHxg)y^!LblcUp1Xj4?8~h7MJ{erWkwuJe#umjo1RPjqKcJGg${e$x2oCe<&EJ8o~_
zO6(W_{&2j0X1JyP2mK+>pp)`$EAxXwtr&Xo8Dw;y<2BpS5O1Sh=~y|Mzu&GYxUfs2
za&-Psj|($PoIf75(suj%&S%tKp>$pEysXd{%VmViW0n_)5sXVYB(+V!28*s`xmgX2
z3T)f5#Wb&h`q;9>Wbr(EmJuS*%QtLoof885(n#d4caN9#)uE;ut`<fu|E%3xwaDmE
z+_OVwPF<azJ;g;{69sCGwMzVT69ow?JMv1mYFD$WM2;pD{`bF*Rk2V(X>F4Ea_RS(
z2Lv{Q=FqdfG1nn8mbZIz!tUtsQK4gQ00>DG1<DbdBC#AT8q;uJheky3RjjU*cp_S$
z+~3B*LHeswKv9lMa}LqB=6s;8=R;L`<_zZH_K_@*um|{83&{uV4nbUag}O=E7IU}?
z*cq#OmHYPKRdkl(9>=jv=e9FW5jc`QNMSe2$;qKY2)TcMXB(!m0MaWtI%b1EZ0<4r
zIrl_w*+oUgP&@=b-}8tid`FI4*oz^LWZ(qhLi$>8-{(h}L4XY@x_o^PCtN`hqlU)P
z4FjM*uGPdCj^sg&g=UJ|4G3NW*c>%v0GyUcFj%B<<3k;OX2tDK1;qA#$N+mr(^o@$
z0HPBbCAvI5Ix2)Aqq~UuN<f(^vCy728)Md{3^H6J{7a2KFhcP}MB*Vf6ytdcqJB<X
zb&}Jo3t1zv;4Phb-11%Jep@Q-0e$zO12dn38Odo*(JQ{Ex%WTBGY26PhGe;;FbJ`S
zO7ANml_xQsyX|QZNy|V3n_ovN@iBhnKFklsGrw>N<sY%dx^OD{eIYC=N&r~EU0BBd
zGI9R%A0Qnvk4IubROiRwAC)d%Y&Fo*sIZ{w8vX3{NNSEBc`sPr&ez^;{BmyY`aEDT
z@koqYvaT6|eBC~XjjDB+)Fefn#O%0H_?Ec7Yn9%eaHS^<ZfF&t<_ANI%m2JCd|Eh%
zG(ccg;wokjh0vZxQl|t2A|Bilm4e9A5xAiBXft?`@L&@6V}S!aS?Ji2yp!rg!kJ-H
zb1_9D`2dh2LYL=}<CVuxqj(M2Bdt}wvS8iizGd^~3qH(reQk7a7{omug6QjyXAG{h
z9+k)y+Rwga3(tYoku4xfLUHk42EV3xKe*JWpE-lH$9uGV?g_dRf8+q{?5GC-Lu8+L
zdIh3HvO!UH%LjPEE5J&<4^42Q8?yN1Yq6&Dm???J$OiSUs65xuT|~dLXNVjoCU_oS
zBmo4_b~cEB{E^P0<}W^=a39+WeV;zJfLXJj(Bs|Cg~JV38y;i%Jd}hHApfM}19uZ4
zhIkB>pY!i*jF>;dhJiXft#x0%pTMpvf@6j3a`+(Z6TSA&e~1l7NvLNqh@O8<OAAd+
z!!ifQJjgHrGiP9Wo;L4g^npLYIap9nMA_ZF7zG)3RR1tDGqb}EkYGa)M-gGq1vZy8
zP;{Vsv$`iXxyy>*FY_@^lPbRa72rto1NLp%XT#NQE;!sjeDdM8`u=mBc6%I3AT(I;
z?U>@RJG%SGs3a)q$)Tj!I;b87-{gaxIDqO-;MzRKG|r>QB)Sp=W7AhH6L`$Ea8cmL
zzqr*J929g9C0CT!cYS&AQpe<6+|Zm&D$9FO`Hven=sE)bsQRi!Rf9h0)NC)ZOtM~H
zin%TzUQ~e@Lz0F{4B|)&yfPxCw=)|8c~<fpx4H43m01z?Lrt-NMK}c&hWJYPjwv+*
zP9OWOVoh>-`)4T=`}8;=3?c*vgFazjwfwv^y1Vc0iQhurk9u5=AKA>Oc)!lFYyZ86
z57Q{{<cu0+ncVy+;kftrOMoXD91*+b$L?-gzrNY_r7n+IQhfYVyvtPj4`z7ka_#V~
zm8XQ1TlYSAn&h*(lIVg&VK(is1*-wHjD42#Qnb>e#!R0WOc)(CEY<+%cmXS#T8CW)
z^4I;J7Jw%eJ#&humzOdFT~5d6JrPois>o@Y(2-Y716I}mqVsw|&bIw{mn@MK=4?`r
z4rAINiVZB{<%mcF%E|X$x%y*GNxmyxgEQ9=*oF4S*|<6Xp|;@4`|Wq%rD9@8*I<S$
zf>Y}sNpH3-JFQ_4;_s27g2G!HCF^(_ZtmEx@|d8A8}c#wIC|F>y@#mttm$PZ_{fO6
zf4SPgu?tjR3xpE8+Xq15IvXp+1*Z!f!9{GrXJK#;Nn>Bq?feIT3+Lyqai|)5O-G#W
zaLUJ9$iw^41P|C?n=R3!Wdu0MoMm5>gemNH1$~ne>O#<gH7t<&yM!G8&3bAMO2eqE
z+hX!dSUVKiUhY-03sWp@#VD9s1B3v0m}}k6={;Z(H0J;Wx!I-PyVC^%=;H;fNAi<h
zWggnl6g0dCcEd=MmFdOxw*r8Wg6vSt?)T<)>|sORZPn@nD~X)zZ#p`pSsu|$@HmMC
zu)zsDt;7hZi93|u4c!IQNH{H*@{IM5$m5v4_l&9s<?(uN#Nfw|m!~;m5Q9XmEihOT
zmGuD;um!GkmuCtm4#j$^<&2ePSv7_!KYR?W+V<cGQ1n^Jd&@Xx!FXA_jMU1I)_#6i
z;FGThzyno)E5esLEiGY0`J1`EJCFC)<K-)8caP*ppuK?+*bldD)G<=r0v^i7-M(uS
z`C%BXa|*zWs18kJiH0WE?q2V1eu(4?meEz5$*}=76O&bY@$$Ho;+VC%8x3i(^S!X4
zrEB)DV+KygUvLf{M*40^!>D2%-UrvYI^hK#GikwlJa&?3IYfJ0I({m!R;e6qOp;V%
zAr=b-{WLs29(%Mk-y*&orO!#5*omnrbt)?8Q%`Smu6S!Tj#6pf+gR`udFvgRI?Rv{
z3{Ridf8?8WA_MJRR6jKHwt1fat$TvQ8yHR*(b57dT6J!?+)^uz9jj07NA}bKeds3;
zLO;1DI#O+C&p0GPsfZG{M<of&weg?pV2t%s$s?^Mq0m;8{}}m5i<^gI+cq<o=jnjr
zCGz$ku);`uvDv$K?@*mJysrBsBV4bGzYNO?r9=i`UAw{!`YlgYU%qcdbhRsdm{4=1
zZ@_Q%Bp+W@;^pZn*^8IPD@De6{Bcjy4_-1Vc2<~QT#N<=u6_<_y}qXysFite4B;^?
zObz}7Bw^E;U0%Q-qS(a({Q7K>uzcC5YQ4|Vc#nHAQtnxgAEgPwx31wg8@xARaCQ1u
z!l$&CQCbU}I%Tlds66=?uMl>;qgk=8ak0M`yHfKY=AK5BPeq`y%wUg>+B(EadoLA^
z_4WW~O@wKOVII~!t^GA*d$#vSIpEM*_J3ZBgj0O9%z<RkUgmPymSvfU+|JjH;fR>h
zW)9ruWrtHZy4Cw-w^U9|{E3Zk$E9cbVz79MmYnVBj>Pz_>a16DiAYrpAf?wvFR@No
z6VqY<g893=CX$mLiGDG>s|{z)BdwC~Z#`sV_<ZY3{0Kf))U^D#Nx0DDK_yGR2U1f#
zjzM5fyKTCVFLv40v^j)`9Tgtb0tVS!?7rjah{IT?RF3UDvdxc?EhQmJ$xGv>5fIa5
z^=PYyxZo5Na-T9{A3n4xtWbO)XZIeW9GHUW8XUEEyzTpxVDc^~Nefnptc|b)%|1{J
zhEP<07IPOOxf|S$9zg_e>T%B~E`(*>o(A~^FI;qq7BitD9<aeZvmQ6T-Jk)m3-3mG
z3~0)&u?aaa6UrE)u!TemF<eU-upfT@#e&yBfX%an%KYo4!U_rsc1hUW*8}|2{_OG%
z__gE?)ofI1hJVoItn^a3D}6D*jMD&gCIA_3L9A&(AFT--N}O3R(;FZSFVq)Xn;{6(
z6SwR9Fnp7Rc6=KrXR_`fxK~EKe#zAZkgQklrDGf-&iJbmsovdusI>}~=#Lj6P~q#E
z+#zgm?MwZyJ3z@}roy^)9#G%O%WNBY00}B}+Z`?*k?*wI?I7(?Qs&!nr=v^xC_gXG
zn-`MC_DK2wX^;Lmc$3(E%?tzGoHPJf6pj{f>RJ#q-=mpLLVfSd3I-Q@H9skGYp58Z
z8_JJP>87J%PKQcp>ej7?mYszT+dGCqxY-pUOr^rY;i4_1x64pQ4C76_?ZO6_SmS4s
zZB@G?!etzK&NR3KLo~}<LRBDMUxA6=TML)Ag4H>Ume)X%DZgaPWr0}GKi8qQ>aS<y
zCm5FPuWqF15Ln^wWaL-UYo8E444`>z@dL`QX+R;V=pd63?oyxM_>Ns349d0Tb-}(B
zms5`CkMQv8-zZaP1Bom7z7g&(mICT%+@jl$IsTEPv2yf4*>5=~_{;{&#Y~{V+?dJ)
zjPUdEt8f0+SVcE*oLSGcgx=@vrDeB@`?+IBqE0eW+T2^mQeKY&yc3I9|KpGE4Vp9+
z+Vm7&z^f`qd0J<^cZ8Z1_}uN0gJp+R-~6L})Hhi*?p76dBJ?&bfS}SBzj@3k&A%5e
zgDl}`hPhKyiG>T@weL3Yxy62a*MpWdR&A+;@}i6RvK5AJW|(=^s*rO^L`Y7B>6G4{
zACmXMl<=wBi;qJ?mW;}MuI<@o{vVldunrljhMKXGULg#Oye<3TdAlKNEd0R(%Q<oE
zq1He*U<I5`qU&mSbgRgD`@X~4f_Ug|<Dbev6P+Gz@Dfo!^B#j}mJVf4TpZfoI;c9}
z@<6niV@v$y%i+dF*(JPKt@`-H#3;jSdDA`d;v|_X`Aloc{n@&TW6$m#yU#Pb<xh$!
zC?JkxFo9q<yCYGHd;Wd^$fyl#C!}Ci(*_KOdH62AIi=NpQsB?LzKsdtyXX)4-Ry0I
z&pXyYpY!^hM%EV@23k7LevDbYKfOsNUiTvVw!8I?NNj}?k}cvNZzw5BqRtq#b<TA2
zyLYe8{ZuDSO0D@&({MEb?5kqU{?gb7+PTQSL?0n=j2$)WgjsCo2JYV-`;-BIxOsT!
zkA`*odgACX-X}`Kz(f^LJcFGzv-K~ejp2K82);bHiA4*$`DQmo7ah_MReg9TexxTP
zd^CUnyAwSwJuV&`1zQSAqVwYK!b))4)~(_8y3kd=M@Uwl-g=z>%+Zin=sOWxo|{Nn
z*Y44)J=0nzgOR;q1j<HjVOKQ+ItIL$+iSM9Antdu0Q+hiIKE!5E*i2|p3f{8*lo^U
zYnA9ripl8aEbom&Pw-fF>rHmQ)H)9}31-7Ihfhbf?pkWqO8GRHP$<e@QrP#unKdS|
zXKb#c{eQv$_de6H!{lMHn}Xhr3!8nU_e$~JlM?0cQ2!N@=3h^<^RB*q@x>L13q@be
zH7<07V!7Q2aQmUQIksHkHf8?OcB>issl<v+yHEU+>w@B!>us|7p=K5mzPLs;KsXG)
zn_TUJ-cc7NrSd9fL~DE80egxrcctlyL3Z-s@xs|NFb1+;SOgC>iJn;Bz#bVf+iTIK
zSpziBOWU2;KY9=|yW1nXpnBS9Zx4;YJB*omPVG3NL;R{six)&LIMx0GlK@BRpe4L-
zUk-M}6QX4`Si3j4130j9@*%K-P^|KK_-ciqSh2b%jb5v56;=E_RAhPvOxz!UvV5S<
zYYWm)nNFe&RDx7a=$;`4){SQU{|%%ACD0ZXl^SpHaaM&fl!k(e{QmVI!;S)0aJ(8l
zriM%j@d&)arGA0*ipKV)=4g-%Jt6~t-UB28&n+reoF_M^08E|mwV2sZ0JG;&ing6f
zH!V(|Y{iDqPj}tBYB_NOL#6Y#I~aKy;&}`KMQ)i!QNTx<fMp*rBe!Sb2(tiL*t~kK
zmyJx~23MO6>Cuax7x~R5P6>gogU705^D0EArJV@tw}zo}k?;_51+e8%h8ev#NAisd
zuRVUeV4L7N;&RS&8JINsa6Bo3_FZjqTcBm2-J1X#fW)J@y^=MGTDu+t<#D>>i2CBD
z|CWJpGaRX+FM4&{XH_4qze&~wG^++hj#W0@_yK$p60JQ8c+mIIw7)R%)P<0I9<fB|
z9u1fx=(M#T-0F@eLkI%)KNhpQpg?zefLGb#@}(6`+S`Da{$Irn9(K=s>^seT+dc_)
zq^tkt1D-}z-^`CT8NU6#9n*9LJNye{3d^2NO-_bxlAYVfS^`;5lHjiCOoJjdr#3KS
zC})L{2f<f?H&54<n6<aar%j@mNfBZ#;ftx~T<B)0ojEhRya$rIA9o)6yC<7UeIuV`
zhJU@VQR!jyA@9-(3LR$VA8}Mp1wCgtb^PR`Kj>E){$#Jujw)bzRjexhH7sw|-N;Y6
zcN}?@YRMzQ*3gesmB#I^WZm2%#rw@^wV74Cm|5MW%{lQnf#$1+`1m;OlF-N8Vm5gF
zE`n8mAZLFqv{Bu*d9%sFA_=T8l6ziHFY0v1wH6Os>t&yn`J~69FoLa^oK{i-&ejj1
zjM%adX8S$ako(b;;)T`Qw{Go9k(MXaV1+dDRnve6;sK#0*1oFRsLzK=LKx({>va#j
zfk}8^^hd`+20*#ZkTKp_q>g2KLc%4mE=r){-b_&A=|23hW&@+U@gtzqEQLM#LgbG}
z_Is(gjCCfvT`tF|S3|~aC>GGpNOLpE2RxvlqQX3)zqaQL&R0Bcup*%}xG~uYgb%?l
zG@)A6Pe1oZU;ihNT+f5}jYo4!IQVT(A8d|Opx5U1uRd7cMyBcIDMIKnVwZXhn<ejJ
zgyKAmP2-VqzFGH2p7-DHrO)-QLi#0YJb6DZ$li^eHCUPa?3%NIMkgS2d3-RNg;i3y
z8GFxHef;4M41V4UzEQ+FDqqv;#+KY=+~FA13Xwui)6<0ItE2$|5!}2wQ^^DvmN(}$
zFgR$jU7y-@iM-3<sl9Bv8}KLyv0uS4>w~~0H`T6eXCb<;?MFZKlj~FhTGs86{0Xl^
z-2y&?rH6MxLystdA?Vi4+e*jpbi0(HG*WwV=7)xUWWv@9`CX9snHN3jf%b3`zO?57
zNWqNow3AGkQ#0~S>gcERUht#h{H4dl+=EO;m%7{8S8a>LGPn!<{{ODXGI!<ZO?{A;
zd?!wvU@eid4j|Zi1ABnCG#Bqwl=nRQ%e@_a1>q=V74U9^gj0dl3^;svH>i{8Wqm_s
zXhWiU_zaha)S_q8_dRYk{}dP5`c|YnodvfXT;XrBR<*@Eh*UG^wVzabxWmW))3_w^
z!rQWg89QK(bA~u^o8k2n1ydoX)s6-2QOy266}ZsHw+d5ZhxKah+%%tF`Wf(wRiO$Q
z(zdPA1T1{nUR|4i@`-c{c=vVABF>0^G}=t^I*1U7phi2|ks*Ps1k*Kn(nOG!hDKWm
znl5GtvT7Fz-uA&JdqrY%OI|zHml*3zK{FUtekm?4j!#;;V?>E$8`(KI%Bv*4@LpYZ
zZC))px}v`zZNF~|H@Cg|>TU&l-YG<2se}f+Qw0LakXOy8m}$B^_#e$nlLZ4Lf2t?i
zgu!jsmYslE16B_}(Z;s-JMhZ8!k(1I&G+4#?VAQc7RAO=*I(N6ZBsjTq*s1G8CZ1f
z;~#eWj5IK~UW>>Dg>_X%*i&lMpn-Y%pp%uZFA4TUA>`ubw_?zm(F`;0z}D+uCTMsB
z>WSAuEk)ma`)wV1n1S$+)%7<BnMSYFhZ0<(>L?Cr{)tr}<3YoESpyhj01;=-la98+
zAHx`3jKQ+I+7SAcu$?94YtB{$0{sWeQhhvX72Ie%T5m$P73j+0=Ryo+Gg}U2JL4sL
zASg4iy7j;TV1&Rpw-eS$)5}1|`;idL6}BE2*byH1=v7BHcLbOKny>~ld|<LkO3%pc
zsFlhC(X%AQY+h|ePGDw6M<5pz42iq&x>hm)cZ4BB9b`Q{15hapRYqISr)}pxRWXZv
z{(SrH$=US{Cv5^VF+y_~!LZq2?CG34jf#L~=qkTj^#`!?;(k9UiguFlnK%o)x=A5s
zM^diIQ0Eun7jTFacFl3YMRKyVt-N3@cV%lNGwZLx!<Ecy5^D?+aML>-qi4a<`gW`@
z;1o0OTTF+57!gnC^OhS&5wK^twCzYV0lu(n(!JuN&MujzT}p$yr7*iu#bU8cvSSj?
z{HJXBoc((_j%8C<)eifS$F4_2MYAh-8fSe6_uOC)<nRA45<0846q6$h3{>#HDMii5
zPprqDV{s;Wm$=n=6d%X8cwpV8y}I3369v{Gr%iS%w5`Sp{mY}<pWE&0T|)T>ZTmOO
z`M->FOd}l@sn<n4=dLve$4bSmBQ6xr-B1Byr4VRVD1bUwTo-Z}@Yn-^6b+C=0zCY!
zi5|v)(hr(-lyKu5K4?&a^jF_|W<<B^=q2k;w;lf7y}t!^P8eMRd{b~el149u;5*BI
zDMl~YmqTRc6;~O<#_PfbfLjdiu4LDm>)jk)U)sx|yR70qAMnnjq0<Z*D^b-mT+I<p
zdlB@4*eR=s#`h40Cf*;BwtUA+WA4Y%I1ji$%E%8pv3mo)io_70d>;z>TTSOUjA=*u
zM8ODQ{)UE7P?y9nKBaVD2Y00-L7k=zuxyD(_>QW{#mk}cvBTRY=O(m(ZPxcgLiU5U
zAURhXbPg~CQ=Qs5Pk=X}M$K`5<c3E2y;#xG(Rh04`AE+s;>f6Xr|!9P=gP;IdanJq
zZ~6k4X437E`9-1|y7jv%a?2XB5E^jITU;rJLS_g(v-8M@w*yVqKlfVm45|{a7QiiM
z1jyfJu2k^$+rr#^i>enLZznkAmd1*Ea4P}A9?2i)!tqC)B<j6%-Ho8j0Ty%*un5~v
zqSZ@fH~Gt9s(70bDJ@XmxEyyY9S`9Qzvd^qf1;rj9m<J8LMOcv#9LCHbI(YFJ>?V&
zDufC|Y-1N@NfmT&8dBt^G(0wcx4ZY*dQ;IX^h8jm5pgzE_}A<%PF<r9M)KORaX%Tq
zY<tLA-@nhU0DaKO9v9on(dyRycCHl<Su?aOK4+fvbqE!8u&VjhDaL&<Afze2`y{4<
zLc_zaG>_l?rptbEObq{baSe5SyT;UwZcY?zwRkHX`OIaniS3@aL4H(NFBwj~$S#7f
z|BD<3mTs*P9CPE(@T>4QHS*hTQOUxOFlW`cadz3~hGrY)u{zlkl2W{l+Rq;IsP4;`
zwG$s-INQ(h{md@Vb-U%f4Pbu7>Hl2~VhicU`yeR2hs*Seps<z05B@lTuJ5?bhf7d-
zr+&qE#ee7eH@nNY+Z>Y;IP$+;4vKd*0arJi>jDcPVT8U0NgrfWNGRLCYDq3o;cENt
z%U5dp#_O$vKV<b^w6yTnTHhNexEfrx7?hoIa>xIyY}hk&p8G%lQsvmXH9<G*dy``a
zU5;xVx4LJW>}GZRTZND_fp_kdxtbmu&be%IHZbPH-%-DQ{qtRceWHK6`8>N{V`dj!
z<Z6EZ<E%bAP26d?942?;1+(*w-M9+^+<S+lS@RbWs1II*=`L*W1$&S%RSX=ewsCU3
zh8A=WnW82HMe-`p9xVrO4i*<)$--{d1lg2f{{3fsk%S+`@_^`62g4X{4kpOkGD{~F
zPhlI6F4s4-en63X3K|suw{h|oKNf$7r3;Wd{nC0lP~srM1OMPPh*TVaj@Dq;=@xP^
z^u)o;YQ89K*0}<cb9^}SKCjXXERm{DStjb_Oc}nKv*CGECI&pD6SA9;uhV`oXWX1S
zS%A^PN-|}ITKxP7tQ9SpB=7GbgMKn-B5(bIPDiB;k)ma=gA&53*vRW3%2KzQ4s^`v
zzG8%r?b>X-V+@<Lloq8IYvfi$dcWUZ<eMpIfmr0$^;d?a^jM*(+l8sJaj&D^b1%88
zfpzYJkDEU2#2{CvfDUgk<_?f|Yu%4c{V2t}cBk#$j`lb;#?y?Giup{vS5h_@r6AvG
z_KbLXBKvTU_3+mY9(Fx5ot2{x?sTli7JBv0mavwEN1yH=jBz!pmKe2JS7r{d@S%8e
zS$;;e@~Dv_qgIPmuc=a-cd>k0X2Nsgs$^5_oU|N6{Ahx3C5vNHE#*dOczbXK&#=)E
z&9DjAGhJnN5_#6=HY!XvtGjGSv3dFGmdZVinc5PAl*xWQb@bQVBqTHiD79ukMVkp|
zX$dWpFw>!+SvBrGTS8@NFVNV=c{6=OsHHbUujb<kX+i%Ksu8ZmksKA;$AI+{aX9O5
zVPX}YG_cp}%iF?{?0Nx4nu)PI*9y9j!5S1Pf{cLaX~3^_9oti2YDRxMcU#r4_v9zo
z#E_5lvtRfI(EJ6X>@I(|Snu++WNeab^M;D7i)OJljV<-(%-ladZA?nB&gs$I2F$@1
z$eB!{O6*>QVXt-u2=AfNS#2$??GSKzu6bWFrO>62OLBxTyLtpq*^Vrz9V$8G;|*8v
zy&&QdBlB#x(eaej3_tpLz0=1X`(Dk#rq;;1LbByCF2&kP2}sx9NmTMGY$m_W?!jZt
z&8Y}W!cQMYhr)|YW-s<-O|BYyPE?(M3--G9L6~=GeB4>3#g&yXchU#2-W|AgSAtL4
zOE#ah&2j>8e+@|0NVPFU*;DAHchF5Tz}|#Bt!{iLnIQBJ;U9m*8Lt0;zi0}agS-^4
zf_FeVJDwmiR8X;OdkO?+#7V9FOx0rRt1sPX74Q?{o=;XUKnASYZ@WHXv}6$)CqKyX
zHR>Ir#jUpk?D2*0AmKF$XQ5+#thl(<Slh_ZKNE4869Z}1YbW7+>oj@3J3y|M_?iuE
zD;;FAnRbZ=!NhBA&(-Fx$$9GfwMIw#3+s|e9DSnq^H`DlquP8=2my6HxZ8JbD+xMK
zo+$fM!Jta#qX&gDVh)R|ph=hVK?R?{lAK;}lPVA7fWYK#QXNqHB%X79q9?6leIBC=
z9*P1{$1Ou0RdRKP+r`y=Z5Y{*s#PK|)532~CjU7vl~cy2+8d@0tuz`N)+;3oPPqPA
z8a{mBD4$%<_^S5ETnT!<x0pr_kl{XRRC;3Y>x*2y44##Xe7EDTG1wXB_DG@_A%D|u
zm2le+nm&m~A5F&&>&VU3B}m+!dZi8BT(7*`g6h&Ic{xAdJe5SU*>XT<UbV-yQ*|Z2
z4h6uP6*R3&$cS(R7%(k@0kFx{8oW_j@_h4hh~yUpC5;6gOcGfpv>LCT1fr{Jliy~}
zUoUBM6>=B15%{@B2s5kyf}9>h!4F*SM^Ce2hu;8+B7by&&VOSx)_wligNTsP+k-fA
z4Mkr0467-5p^hxPi@p;h&t9gKj$_bJ!>yzJ3q>r`7cap_hA=;ep-=O_vGL?#nBn@w
z8+HT6UW@&3$`mRr<%%2k8u!qRY(h-S{1wx<M_-@J9j53``>==me-eNgBmklZ^9#SG
zymtMB;^19lX5DA`=#&jUw*gbG6k?V3y5YyBFa2g7d5k@9=qlc-zg=3`@9A9Smul~g
z0;z!i=nUOE#3zhZTq}4`)p&RzqMmIGrr{0{M}8;+o`L^^>0w0dB4&BXd9nv{*01NI
z=Mb2)-qobMW`y%9)VZ1zpKQN1L5#WBj7wn__FRbKY17i7ygiFB`CT*c1eE(Fe4aK%
zMtJ*6R?RKY{%PC%w#2%{41bx^;+VpPi+!eI&2icYM3&`-GniYf#&l}T#QnD6X@(HL
zM@-6lRazQEH|m|;6#m>?#>h+Nc3YpmJjrqtzxyBd7Gw2D-wr{QPNNVy*&g0(-ww&W
z6)V%3z4%$_cV9WfwNcC6xUF#%QE4~Jb4t0@gVC1p0PW}2Zoi=a(*m>(`-XwBR@A?3
ziNkPn|0lsXVIwikyatJmwPi&!jm5ak7}L}n!3GP<f*3g;@76_+`Bu%`$Q{8hQ%iR$
z1T!=G5^U1e&Of6sp|4<k_~21KO_APl-#iDa#GQ0#0>Og@bih^RBrW_1sJEUf=I}`h
zJ*~D=p-ucL5CgKb{DAEblX^bO`(26Bnn+Fr{6u?f4m{~;wrGQu)oQIrSKwLDsV2zf
zB52W0AV`?P17FCe0a@B(3CAq${oQR<lSs$+RzZx9(<1kG8&{t_u5<3MXao?#0noQD
zHR&m@f(gX${j1|JsovH)1mZyP1})Wr%i!GEg!;U!Pxybe|0Ag$Y*=-i_Ibd*+yu2b
z|CHcVk%7hB=Z~C`gzQhCP&40fZZFDQfzgs;uvrIsNedsZxGfX*5l3JRu@&pK4*zu}
z?@&i`Hn7q<-8+x4?SZNx%pT_D)}#B~1+lM1`{RINe*=}>gT?>K#Y3ln-UAEky23_U
zghM-*5+41Xt`y44AI$eI`L%McGwV4A2llxb;`1h>jyw-@;A8Y8eRZVcJJJif_4>aP
zG(zSHFU$rnMO!BV*qN-CFRS<d_nw@1tyw<!E3{3VVTlsv$dSR&I+4XP#l~RMAgM&p
zHYnx|1Z(#|R0=CrmGD&`*y0U#v6VG@*cU1u50P~^zvSdQLwT;&S%`GhEW)J8ND?@n
za)<TLj2b3oOaOk>KyqV6eW=3OlvPxYA($E%>_@n?fnMBQfxR$db|nbF*O9ez#!B)7
z@E~dXS8hEpUX7u-VjPl;qJgLrfVS8M<CW|y{u{)qj#FV!9X|pF_PF;<PHckV3=EfP
zVa+>X8gc)*Z25g6uG0*Zv1`t)c2m$cRAa=cx%++Kd0Tab0#@<UHz(^(ikZKDbv;hs
zQ`zu<-O-1IhMkWBAsTn^=pvic6aC@pJxH4>4~!!`_;JT;PfRudS?dL^xfbAhXSmp9
z;l6oH<>K$SnS4>&Gr#+TYN^4DpSr!Yy+RU(fyg;IpzJ?nu+JHm%^VMLI}IwM-oyEu
z(f+Equ2Vie<1cIj-1x=!_|z-aX2|v@yl2iyhbcpXd4IySvVC?2<*ic52vm-B?ZPW$
zy$ftO%taRtb#XqcE1Cl8BEA*swfp%mjHRjiS)p=vOD}+BN_aF$`d?q4vRYPTq29U6
zT!G2k4p?JUOzBA;n69k}OweCO1g}(;-=n#*iaC<h<>Z_Ttv)?AxLgf_`#nH}aM~}J
z=U`ZW4)^c@vzUx^75NSszG+5|3s3Tng;40>>Gk%nMa3V{uA3I`5%lhra0k#0_nL;~
zW^c18uAU#(E$-{h2NeD**%Na{!o(!ypGi&d`pZwy16xMYMx+V&t;&php;BD+1Cm<>
zLOk1hhWZ3t8142UT*lNd*01_dN(E7cJ$mfygCn}xyzH>|q>U(ErkF?W)YZ{yw9N`R
zh5M|+D_JZK6Ma-;BJ)uext9!{#Wos-UJT{c6}-U_GOh5MKpb*k4)~!BJ%a-Z&Kkrd
zt!ux`pxh6i*nRPVXu@A!nf72oIb*e6_d!Y1)PDHh=<W=A;5n$sjpGgP4J|(S(g>_5
zd=(AI3o^z|n9YI0Z*EA%CrWgjlFP~5?uD#=AZ&uihp&Vt7pyo-*rcUddFqZ79zGy&
zJh6|<SdOZz;*X^w)l3|2XGPPRTQV^%;V70x&8|PgbO%r7C_4D&fZC8Ys>Em&RB3D1
zGm(Q`jN@2@a1}_@{3?Z&1oxn}`?Lc1qV#CEi>`7dQ*Q40CO-+A5q!unhc@^NhKc-q
zz{Hkkwz&J61ZijVmILc&ifOCKpjU4uZXT_ufj`B6$W)i}_TB&;>9os<_Xqq0GoI3W
zK#`I5<yi{3skg@aT@YyiM*b*kwiI2UFj=|g!gh5eJD}zyw7_B;XZ5++hw<8Mt68m7
z@x#kB#-$C*Uv>aUbZ9q&HuhEfZ^p*#!|eJmhWe0oT6l0+H1paW6<+ILRu9=3D|H%q
zIe2IqUVHuxf9&%VEK021C1ykh<eC1cl#gT-52pllv<~1>jxZZIEgo_}!QNyo2Z4A1
zSt#eXTcs9afk94&{uU7zW$d9as5Zl4H*CK`V7=L>rbLG<GpVd}E;o8me#y{)OUX!4
zxrQ~VmYsUkjelBvB_psjT$yhr=#_U@r*pG1=ay^rb#*;d#T1UZ!i-$c>r<{okG&M{
zwB*5i;f60A4WF?MyyyG*e_w(k9>ogAb?&Xop!_Vg+iK`Ol%UIHEs^W3y#N&-qq{@1
z0PmsIh2wz9=!OEuWcqK|3uLS5tvDuGF|S-FA~rvr=@T>QlgPfheCRMXDq#FsKk|yz
zQjC<7dHbWyexqol`P;)BXj5;s8C4_?(Wmg2#ik(%Q(#SF?Z7%aee_XJ5BI00*krPw
zz1--J?)#AKH#WHMcsTZz+$*1NV@1Xu6EvSnZo5u|A#H*-vP`P`@+_<#DXa&JFeth2
z!~>-HQ%<fws9$sKVrWxQZXI_v|MEK)QWC8xtNtw?UZ;=Hn5CUg*Tl~+4VgHlH89H?
zO~(M{)j#3Y`-13{qeZQANaF9jS+rsn4bmPcg#u{TO0O)S?C7i)dA~mYgSr#3gJTdv
z^Nt}nKGC!y<|AP6nW)0vqRwkCH)P?dRbhKjPpFdc9oqVk5Mu5{o4jb>!M0u8OeShl
z+O!nS%wqbi^M2`d#pA#OKS<@G9ZxgRrBco;74%>~#O@4wBH^pFI3=LWD<j9-rU+a_
z>8>tvU%!sw&M9H$t(V8QZ8k;?O4<h4NF*(-dhUIq57KD~(YC3m`&Nyvyl0Hn%ULnj
z+yeQOM+6l>5ntFBs^pkafICd<#Oha}vNUb~HVT@*YKUNrS5}edVy6#|R|~6q#tcbU
zd=OY{tGkk#p6^F+hk^Ht9}ZP+knFs>nR&#p_&BVBOVc(aCOvdiG}_d!6^RC_1|jsF
zuzumS%e3V$11M51cn|E#Tw67T!gwzg+sOXrdnB$cFPr;WidH3RuJ!0>q2u5w>(D4Q
zNClGUOs5D^D}7=U;E7GwdG;M}^6UbicmkQNS(v^oIcU?;yND)Q;Y|0<^oavl(^hvC
zuk<)Nh@6q+=j^L*SH#zUYnav>9+I0^sJ6W*GyOg-HgiEozU@-tcsr8Pl?w~E10xE{
z3fFK8^*bHR%>~-Vokp6Zs5<dZ(ag-hEJ-<fI}7<tkhh%tEojB1#gZL#xs2+QzbH$g
zMG+0UzfdQbSxFA?NT7-xGt$@^@c%7Z!3qAavAZ!>%AK7hiKB>jWd5C7Z{GB14;2}T
zf3%<!?Rgp(BrqQ6b{aKvU3>K80;=HHwJO-(T0NkGr!V1(y78piu=Ua&)hRdrXtgC7
zD!P~~Wa=$+g5jfb&};>|`|afSpOf!n;n8IEi)|&#@%*sEg@=8Y;n1XT<_b;Ggg0fR
zw4nGmGJ1p>?9|`_nt6U>AGEh_+z6~=3P2CAMbD^4N0L5nEh^w>hiFi-ytuuB#uae?
zsfCn5EW-z{DQRRC?)<fSE10X;X9?kGokUS16YTh6_%tTYn$y_Y$X=cy29C9SwI~3l
z^n5^7)CNh9=@=#A!|+18ccH|NQNhHwxWHcn`Z<(Q0sO!-yUjdWvrqzqa8tSL0;~vs
zQB12jT1{={9*q{WGjy5Q&3v_Pf%ZYFVQQQay}`Jdx-f@%st89QUg}rRw=VvPego84
z5i=fZrOP-_sqWf2FDU!oq;;$ydDhLx6xq3BH;(9-1+Ev(D`(rWAXmJC?^BC6t5*JX
zvoaEQEgjRnnV9OfEa)(gw#_c@=CUi?-zx&%P3JPnE)KSt_zO8cipJ<6A2>h-m=A<}
z8SRjS5wuxEqbcOV)uRBbVyE*o&m~U1p$PzhAhp;X5JijAhtOJ(^5bD<1=?nE!p5%y
zrK(#%+r-pmor)64+MLbqK%XhL*BY(WS*hDJ%+|h)icGW(e=7{489ttK0p|fAmV~ov
zZJM#jGiu^G*`T-DVyH+h*@<?An0tAVoqe^b=D;F<Bo6AjMSJMnvRkzD`EbvjvhmOP
zpdM5NG8WC0+Kp9@-8GG0cDVv|d{S0!LVZjt^AuW`c43oHUTe86Tq}vAC|qmNP&?X{
zz|g-BwPeQUvPtybJcId=Ke#ETQNLpzXo!K5m+{aiY-{YLI>8|tzyhCH7ZqI=MXkwq
z3O}KqNqcYc1!Z-{r9^7>*%EMkx1TQt@>^GIr-=)#UW+<74D_f#=CmBZyYQV1G;o|u
z!&R?W(_Go*VHZy(dzOHPOER>;spIee0ONkK>iM;Nxvl#K@S}LcSpXGEi!wADXwm6I
z{s8Dal5Cuwf3xF?fe$!){#X{qB={g+aBXEi`#3R%!$+LFc6&dH`}6b5Qo~Q<Cc9nD
z%Bb+HI0_|K^sG15IaOTgY{DswW^`~bf0B1wHa~kN8<99*(G3Dcr|Qe7AQfnN1|>9$
z(kvK7Se)tx&KB-af=|xRW-p`feCW`D4lVh`0*Fl)_y@+Nnc3o4uDvEG{!|DAzEUq#
z0FyRLwS=iTJhm&H!rC5Z%;Kdt;mne)K`6QeKIAx$Y!uLkXaNJeotEQfqlgcvp(Tm$
zkbKe>KUk%vhhoNIAkCxSJ(z91hfg8&3ZAI$EWvbFuxn2N6ZJV6lE41_+e6m!bXHL4
zxb;%(>U`{z(sC18P|?Ud<u=*ZE^HnV$So(!wg&59Y`Y`t7cQ-g6UP-&B=0G?1~9@3
zWn>&m3d?pP@-nv-Xg~DhEI6bPO2J*z?`te-dNj57RWU!atc=P8so5-m2D~`Yuh=V<
zncmZ>0X0!4989W5G5GW-t^Wekg^NQ0RcUjE5BR>pBi3l->v|XIJ0nmlDJm+aIPbMS
zP;JDzHM^0iT;?&`s1t;Z1)`%8m?LSt{*V?R_e$N-SzNAoxlx)I2+BpcMn0PM9CQz6
zYHPRGGix<tsgkqZA)>H(bl`z+#gul88$Uy8rTf`Z3uBL_1!}lz)-;4L<<%_ln8@tp
zDi-wyPGfN>Ofb1|9nM#)X<0gFcxlgs*xppx8boFZ;kHTEfdiK4UCT=rqT@$S$^Dz1
zYNh4^{Y4n~JYJ3v7`gJgp#e~DH3a?bh?lK-z7-pNt#j@TVl*ZB+Ud_2_DTE9Lu*I-
za9J)A#(}Dw7gx7#(UWco|DqgDXm<DOG(U!9F@?h2KzD0V55*Gpy{(l}D|wqAq#Xr{
z4~R<%Sj-rkT>yxzsB2HBtK%;E=h#Dt8+I_6@8vz9)_ucwDKz^#tPGr#r3F}KT~r^>
z*Q$7{ef0kV%)4<x&*Jrc)sH8!k~=086_w;X`WnHs3e+Wi$UGMgB5E4_P`jPiws0I1
z*A_x5`!`CDJj-RN%PkM7Om)lee?Csh$EK+Np4;*zv-RzvS8V&GwZFbp-fZtZ6`fX-
zk{mz6s}PDLC}zr+AkbX>Fx;?(pQ$`!aA#jcj3eJxrumE0nn}O}EYd42QqwdTb%_#R
z52sv&2)ll+BfJp<2hC%Ln6oO%se%e^3Zc9-hIq3&5C{<;Fy-^uEP7)dyD3ImvIwJe
zVeo59uu;cq0$Nb@YO#<@R_6rJ^*$J8DC$~3MOTZZDiOK%EZI9cT#{)O%K6p(;Kq|W
zHFlr4_AB9frFB|c9Nq%Fj>=Ys=syg#YSu4b4??szeg#esN(WjGE~c-}*S8dU!i+z0
zcQ3sH7|_HprCu?MS&0y$9YU@*pTw(93}j<nPY7;)@-Mybk<}hJuv%hXIlbC{TVpP!
z{n<)5uQlrigEa`H#D$`RlgeW{1<E8OV6Ds)Y#Uak)qxb{6KtA+Xg!>MyvAAZy#D>W
zz4u}2$u~OQ4Aeg66e4S_u0@RI%4f~Q>i7+&Ed~>M`{^@fg^U3dZ19k~$?IB*;Oxd1
zaPEqV;`vbC`Kj1dgK*X>kN}68PEB2??tOoy*D;j07X7Y9I5VQjEv+Gb<d;MwCEh_1
zc$|EEct^A-5s2nN3oc#Ma(JtrpthvKI-Hb+^m5M=qV=#QC72ODErf>LbaN}l|1X6z
zwSNrFhPK)1M+XmxQyn}+d^?&|dS<?K*d9^XkM~Kfmr*8MkIDd>90I8$ROg0~Ugy^D
z%&bKD@mu=P4G6c3=VbAj?U<6KO-$K)*;rB{G`ExX%_bNb8(JPUPRiglJ9Nio%DIv?
zpW0X&UcB4KUT3BFexw@MsnGYCrQr*^e3(~^>Jqp6WMXl!6wZymqmjp#Pa)#B)Bg9e
zhh~d?_z;#2-_QG<Sh~mG;R2OAtmm#~gWXF~_C19)nxUAi>O0bm>tl4egPtA`mJ4fD
z1eRQywNd!%qk?PE6KaT*7vJS#2!g7q`?NrwUnN>Ik$H34toUsUL`RcG&s=c#a#hS|
z3HdxmEJepkidk&TSn+n)<S5^C%tOsEghi&JuxnixLCbeDUj!S<Dzs5i<%r@x-lia9
z@J}FqCr6tlnFdbEMwzpy=Pm$!)VREeHsT=mD=?t3_zY-Ze=YO?F*EL6I-%Lzh%mWK
zH3YJ$g;RVAm!I3?<x_RW4A#u}M^d5H`#bEk%&3#|rMe1FBXa75_=E5mUTgVp&d>gN
z-7JaPHFonjL*(N}M7Y;&(xLGqBhAzSKa`nNs%w;;#+P!V_DFUN;v>O;N{*}<>BG$L
zH;bnW^O+s0ZGU9YD*lh%3Cw~^aUQ>V3BtO&B!{rH_lC0k1Y3E*nXY&4rf<OFNy2pS
z3<c@Q+piC_&b*#sVrQjUmrY>3f?8Hq3Ti3^AW5OLM3*n0AmazwhRRiEq0mWvW(%M!
z%+RHc4->NI>+E$%wu3r$?a9uU8U)slmoM7goU}#nSqlptb)!|R0}Ro2YgU-;c#l3Y
z9spFoip-{+$M)G~ehamOacg06F;^rwnXX*suh3wILiK-$in5!9)95%AEW_(1i+W+|
za?u4{^yDgSzTNvW2~!Ekcw~4sFzud+ED_#RS8SQnAR*}E=Z%J4F(1rZF6%MZX-9dy
zjwT?RxvZZfie%R-jN#Hsp~szCz>d4l5MBk^OO_-9b{(F&7x0_P(`ud$RrmAidIaT6
zKZ$Ij?ah7kVxo@gF+;OK!f7S%JxaS~XPly&a`B-0HJRy|jBo}*L|LiFG1fI;Rhncb
zImmd3K1`Q|<|5if4*6KP&RTH0LR)C|l*}z+9Z(zSRE{Qx6_(lY-lX!5>RPAr0nf3~
zWgvLU^Ff()xyUwsBT{*ZVZXq};lXaF*COAPW%%aZ2|+e+YMpEuTuTjD=#o>nD2G*l
zf-PazhE}-BbU5y~IE@Zy23MqtfpM;34b>SlC)KP@u4LH%7lTO#1|zZphaWdb^ab0U
zjTXNQg~1$2i0v@!v?O%<O78)F=^Mp&>A=Pwt4k{QZOg?I`^UMA5OR<=UKKTg%cuW^
z$C|96<`SB=;qYXUi0*@={jJ(9T!w^@G#)x*-Bb1md&sk~S$0r!fVThX;#c0><W#QE
zhPLH3`&%;ZbDYU7e~dAMgTzlKux{hU3icigR8Y%NXajuCY5zZiyR6(|e!0OSFSt)X
z$aICPe2(T-0NEYOixMZ}%i=o(_)_;jl|rLb>Mn+Y^rMjLHv3n$ofWD&Q6R`^%+8Yx
z{*RaGm=N$#8gz(Xdig!VNZ(PPc_pL3xX|oo70B9D8c(v*(gGES44fLoEoUQbpFVC^
z!wCqxGCDqm(m-I@s|W{vz^1RE4>2kA*dI#8L$vla8$&%;03P#CIqjLny7c<gLk$~l
zKTb_9yo#}*ZSJxPl6mNxmJ0GkaqB{P!7Y0?E?ZddYe*2NSZlv9lzgA@sew27!I?+I
zfmq_SH;PIkc%2$}s_3}3`wWcrj#_B`oc5-o^s;T@X(MjhP5fHA90rk2P^~W7y2+~8
z!40M@mi6$I?4XHXx6!yaeL;B01LA*NkgeT=-%oif8|PSaK`a};jTg)yJ`eNUS{@^e
zDU12LjXJ2vFxv0!CjLiDMEnS-zP{hS-gq7~zR{2=P5x|i&#sdoS6(UbE|V7v=}t_0
zp_+;q<L@U5KcSmo30>pF?t6&t#@qVyw)w}~D74M*Jx%<wVW-iNo|)Cy%2CFlwS<M{
z+M|*AXuPZp=oTOCMz3Dn^B|G*pB0N;<CBVir3}U2WvoCZt+i8Ve{3?I!y2lWKl^q=
z%k0{<K-A7tyQ2%i(nUg06}xg&LW^6B$}bu!)m=o1%iiKMn0azaZP;-sCS!>qbTlEc
zcYFcq&bW6x>XlNPM(;RsC-z8*(gx9lMq6;6p{Bty5`$ldD5zER0-ofGG&yJ{`Mp&q
zEv82>(%TIk#apa13pWE<Mfg$;Eu*I$)6gFs8(*f5DF84-B>_V)tL3jpOS74hFc~^b
z)Elp4EB5b28%h8_tv5Xb+&?8Z*amWqUSys?YK>`uF6jxX86A1nJHFfFd}#?A&+O?8
zsKOrW%cKShm$DxC75mbvSXz&~t{%1hZh}qLcAnukYhE80jDwuO0Z>6TZskrwSny`U
z`w8?1eRZ%K8H;!{fxK~a&LqO=#T?ij%%{nMVT*7n1A<p|cmHV<jOvQuA6*}KU_pCC
zHNBOq&``w@@$*g?QJ}*r{zqg$5$p(kGHUNHl<y1jwg#_f8v(2Um^Qxs>*s~c&$t*>
zz?XJmTEPz@qhn;(lIyrtA9FO62ZEq7s>D65|A3q!<}HCDe;a*h;>RGQ-$jZdBW5R<
z0Wo1NaVw38`try02E>^LTL{i0w1jpJKJ1|474-%OVFUD_X+t}$hxFG<G=YnsR{K2V
zAvhmo9IiPmBoqksqYMzdhjYLoxq<b)E~2{D6ljtI;!h3jz9aPgq5E+1TYT{+;mP+!
z4gGQOf{;_Ujh-N_3+-**DAjkDAcaOJ`fHzHfw0^C(_;s)kX0lySzEM+WS{xoLCmY-
z5$J-5_Dw?=oT(<k6pC7Z=(=lgmeoGa`ovHjNvpNzkj1wJ%gpFtw&5?u8k!6-#kF^M
zEmCs^Italdlwyh44y4WDP%o<vEyD*sZpoFl1!xMGo|&#Xi<lQoe@xi%PSg=Mm-F$A
z=*JU%9pPh5e0nwDA`|>3U({^cmfURt*w0A^x)Kr51dHa<F3IZ{qj~NHBoYKqpK1U&
zk&LdWY=p{cJi|RG7##8J%U?|5B3C2Q?S#@X5dC`QU#5)Fc5$Lx=TKEYoDn0**c?Ht
zR966YEmd3Wwk_x~fj*2Dq@Ex$#JeIN+s%MmY$L^nCxmabGE*M)Sn3ydzjV-g1ZHaA
zuK=({8~AVxvF*SC#tl1ceTWRBWJe+(QnZ-jT)rwC1E9o3%Tm{l1qO&d00Q&V3Nqos
zLn%e1n?T`FO&hvVvkq-y_y3<3;Pr=$H^?I=P;B)A>JC7!SA&tp*y&SgVM675v7VNW
zV@~5qL5f8uaD`2A@q*Cx9~qRur^-UAuLK>wmw+xZh9>k~bKwKvk6;WH2oZ^N0vTQ`
zUxlzX%1_sO10{PPh)`ZMjA%@}BtHbG8hsK!Eruk%4D+mC&X*XWw4y!AcZlYNQ%SO7
znelW{P&Vk|Io9ejYuywmbiE;$y&rUZJH)oE34US_kl|D2*Wv&1@2@u<(Odk4r?4G2
z&$tJpmv?^Ha<CHjaW80b<q|stC=(EvcLXyE4}cWQAuL!dzrglK>T~B1$hj~N=Cd0D
zF3xe**)&`3Qz*g6*ys*~w8y<C{EbO0;j}mIpvg$P3r=k%wsfs6diNanU@0wjko0Az
zgpG~?(%uBfv1%~-hRGf?9PRbielAn5As20#E)9<gsOV3)qRowfZoOpKff`3@CCGP;
zeFg;*$j&Gc1~Fd|;9!c{)f5$h1M*W1KwK6{X0pA%W9CQsR`lO}Zwy3;w@8cp(ww?~
zqg8My0~4=)lj=wEJ5^}aIIQJf5kuNK%|eX44_U{VGjMPebwS814HAPjz;(dJ^{d>d
zm8+v~P_;`-+c-ExuKn{ju^sC7BEif6Lo@;Q_7;AYpKxA`xa$mPgEtxaAJNM_OzH=`
z!l!ROeZ~M!gm%W1lrb9$24XJfa12dQQDh+oPnzM7G=2D6RhQT=m<aCMenjUdk`DoU
zK7+9VfjuLwl^$h}Nri>q`2?I~vGdSoWRTQ1=ZVe5q1{xs+T=V2zdTO*hrW5n7kn0)
zW7Y6^%1;qIjhU*SNk|5nlkWR<b=t;W+Ykc%AtpbEaQy^LXI9mAeT|fVnJhP1oIw`g
zNtsT34hTIC-AER2((rkx8Upb(<x%Zr+f@LTyAD@BNDc5gMj+>Oi|D<%!^oEiL<3oX
zZ>4qtNjWH|=Aa!gg&zPtc={1=l01WqkeXXT3~7H(X(Js@0MWIDUALGM3vhRgVJfqy
z(K`yzrgma2;b%pMEDnQMZ*fZl?tE(N#Pyj#RTKYW2@q)|PDaeEMX><D1`f<N1mRy)
z0?Ic<*+lsq5#%he^v8AMUT3PBG#&#l<|5p<WQ4T1xRC`fmJk(d063cBbsyD}lAv9}
zuyl2pcf**ICDCk%y&Uunkmodu9(z(KgY(>zd#<P)>I%~!xU#%43|EVMy)A;!Uu>b*
zIq$XvKq(GLIlUY7DFHx7b}U6t0RNXqw<hp<9BG7vB$^8&T1iN@!qkfd&?F=G4nt?v
zhHX#%%F*=Rmv*(IW$-|Kw!sx60En@6@?*(Jh~8_6izKZf)cG~r^=(x5o5w(8gS`-C
zORMPl64cBd(7y&V0u!NolO1hfcRpdDbbd8&+m41<IXUcs=M>aR^F^KT2|whf^&$Xg
z!K0M9NnPPKUV$@*0-OU!5;{Pw%3>=$%Ak0n-$crpjLtrXcY9qg8!!JX;>i=x0TJa{
zGw7pgpcLUTvF+*7(zlIZx}c=LOw}*i;OB2Wxm3b<?_l)!1l$03P{D;Kdlt~NVCi?n
z?hx?D5i;Ld23RW#EPD9rD#BT1u=kg^a0bj%cM@8UEg1)J0k<OtpDhtVM*-hTX%dQ0
z>LigOGo{yNXX=0-({I0e$=^turXZQ53Sv<uctXr$8A%=b=#6wPJ-lSvPp`{tL!L@m
zy4H%r-%C%G?g-8b-4P7dCJBhI69&3N_T@>2cvuA=!0<x9v%)BS9OKfS{H~w+BYIWz
z5y89%L%)5NPP~Ern)H&EFdy^5$e06kBpQ~n^Yp9TY|0SPb)ayAqvruSX@J0m6g?DA
zf^WP70Hq4=%M@xQs(|R-MQ{ZW_#KH)SB;??&h^`m_yUeSgU4q|HORYJYd}<KPF)OU
z6?!a#%d{4h8#J)#&x^wO*wdJ#d`5=_^3fC~U*Z~1dWt_i@i!rHN8~JaMkkCuld##j
zqi@UY)-y*A*=3$|Sk&v8N#Q8a4k|1=k)4+_hO*C1rt}2x!nr1&l`sc4v$0XztDrDM
zFby*45lH0|eRGHg1!JyqaZDiWSK*|-69$|&N}Gr|3%qE9IIeE`G1|q1PRv82_z0ab
z`ocI;Ca|6+&OsjIW-0HBv}`LsRFyWp1^Yxbq+nSWM)6JI#7;uAYMit<{M5Ig3H~_A
zOtBvAX9IwgoxrCP>zzcYTt&d%4Zf6JqXRwTDTlR;6b*wQ-}W|A1P2-Yi;<!itPNr!
z53H>CD1_4bZ<Cd!`7FGatKhJx=0wyDoe~SVh`-0vch|-S7Tvl@IP+PM26?)O^G^<`
zc+l}G;WtLs>0emvK9eTrtYlU^b9-8oaX(tQ3q@$y%;NAZF1x3rPgHNb+c0lHOJt}1
zAA4`&j^+BjjX%wzBoQHH42d)tA|*qHWK1%ap^_={P=*FFgfd4mB~*yaQ-ctSh{{mP
z5M@Y&%)j%}-k<N@@An^gfA8^mj$`lLmgjly`?{}dt#h60Tx-4BooAT4kL&FXPIBh0
znKJ5KaRQp|RQ$NuC+W7<zbIRhpz-qZwbvTjb;(d~O9M0_L0XIRgkK<^cJZk<H4VV>
zO;yLXTpW?H#%T(-iS9#jjZU5VyYYy9%3r=;*ldnG9;1;y;eLR{o1Gg47V{y_-4}gT
zKkEK78)BEtW0R_m2fm>ojJwNOAgjFk!`}4Fyr`_*kRCqF{sG8%uEfubWBwkq!Qo?E
z!am&Vy|mnrE|4a{(mU5eEX&mJcUxBDT8V=<NnwLUyh7WE|IzPR+)7>0E@kdq0L-j!
zOBP8-qaA<hF~Sm&!QPgMdz`e29X{ak6OoG8cMg%UY}oHY5D?p6{Q&$XSY)s9R?7X?
zS@7u8W1-vxyd-T@mbG>TC?h_Z@vAYgXiHm#y+=~8CK~_~86ZToaRRRnptK@~_U38|
zWTWc~v>3-Fsxp<@>XwM^AJwGu(qcq5T1=_+U|AkEUNtt)81)#Mld%)o){bz$7@}F#
zVa`v1U;GKhdU>R>_9VPAFA)=jyAdmf#C}Rzg+_}Dq>yGlZb&WL8OfI03}!+AJev6G
zoeOroO}K15Mcs+Wa3S(K7VU)u#tTq{mWguBxsD>Qm#ku=1E4><BmAC?=3#yVkE~>Z
zpBTSl#)5zl&T5SC8j4eB;buF@h-00mgE4o@*iOFX5?=9B>=?(jyPoZPjCdZ^0A)T-
z!gLob2joFu%*YLd3;7ihWvFzzvOB@=upswd&5#V)2ZX`r(&d~QaFk3Tb>Sa#hWQXj
zdd4zcu98T^bJg6CtJ_B+I~ty)3B|~Pcm)8#r@SH~NQmk`Pn+{x6STsp;_Cv)$xLZ#
zR{hTB^Ps^AiRpVUa3J+h%@er`>|4gv(k1F={M*H>vBi*X`mk5&g2dB%Dxj!UXI0H>
zIMNqbdW}{j43p-yQjlQ&#KzIlodEN!02FdAQM);O?UhbsYlW<ZMa=TQNDN|opmYll
zMN5n*RLSWD^$7JnVcg1Gmp+0cBMpJ5#|mJE^!V0N_9kK?GNWfZ=OKf#)qId_5LjaV
zDSrM}YCl>K?ubbH^8hlF!y_Yg`V)6@k{~%xorUs!m)*=gm5!swn){ebcrAoieQ>N%
zX?65}l1>ffQ~X`4giix-&%dym`GGP!$n5l5bs41YGApqttJuyTB!SWF0`Lo|owUrB
zb5cX#uSvgM6q;q?R`8?Bw=no0fGW<FFjJZzfu7FJ4?MVocnYc6k&500P)j?V5hLme
zYJ9vKtXVj1aJUsYQWFSd?hlzz)Ybgj#Z<eT?<f6deO0l(O0%lRUSfSQ{8il+xtE9e
zf$DwM2==o9LEbrwe9wh~(UpE9C2b55^*xRZs%8!9)ZnF*8@CGM?$enrqi6^Hj=g_F
zR&znAHuE%^4qT6)8$7dqeIqJN^@P?swGh_=^NpW?J;hHb2Zz=(X{3>`xA)r^#4XJx
z78s1a%}&yXzF--pM{QPBq%fG0AkJ-m{4t(zIc{_R$+I*R@4OfC4Crg5<GzDJNju%;
zOwD7nit;l4cdW_z%}Wm2?MWLVV<`WAHJkryrXlseM${%y{`;4d|LYgwO`5KxJKBaM
z6KP!AW(q|+5Xu74fl&erpe(aY+k?UGE;|g1bFpn703Pc*t-@aN!S9<JmIFX#3sWfL
z?_hZ~ay=MPR)BD^vW<~KX^sJ++eai|Z9<qz{^({bsBJ>L<OgCfa?3DKc?c9wPZs}|
zwQOac;osW||Mf0D^3U1J2bbXR{Lf!A{A)1t|M`~*8q(YIFdh#L!1O+v<GV!HLs=38
zmE7^@PSl&6#R&Hu;Zzjw?{C1Zv<V%GaKa$?5^)MeKE;p$_jU)tU{9G1LR+Q1kt;Cu
z^)e(E{9wT}PR)~Csd&!~*>j0<i`UblQo@IOwG3lvc3%BIR|tdkubI#*{x#8Ao$}wm
zr2KaguwDMU2><_E1aE~!Q<dPU3!MxQkaTTrt$n5>g2&vk9}t5@?{hk5R6SEE?71E!
z#6#^(i*w)ul!&2tbd*>XYbJQnv@PXOc~rx9KbR8-|0i1t)ZOCAMvREJL2e*^WxYtc
zjCB<LKkWc56ml#Q^~kPR_P|o%RdkjKwu$o>85Pga2;O`%CT#daUUIx$1R^IB0LN^w
zbvo@Li@!d)Xxpgv6FXYj3?KgicK-f2GlW#<yBDwU!`VNd$hl~1<&S-ob$ZvVp0N1t
zsU{)xbwz<%XtOg1u5a2D(I?@D2-(#*`4Jp_G!)_JkX_V218a{<78pvx{no_ZO|^mv
zO(V%0g2VC#!MRL2|Mivqi(mQY3y&MhTCgN-AaDMlYk`~dR@MKX)iNeD4-pTeFp|+a
z9@^7DkaTijWCQZm$a}m^8p$=r<|?_@;tB#X5$2M#5Q9@ZDMaXSCBH-w)oYVqLOJ=C
zm!B2axUrI5^DSK*`HQDjN67=SDptk=I#)tRp&Xpc#S^Ip8LU0xiQYS#5r6v=`Xo`C
zyn)hkc<~LPyx$f-OO!OhsP&{p>>|ca{@#0^%;GP^EZ$zrS3j;`zFN;Ujf-xgneoqc
zXye;7+kNos-?IQXMLuDCkAQ6_$d737CVLP8`lT#Ft~BYzH-zn0QftMv<q719ML{oG
z{F8hYrJ4WF$8jxQiVkRDSFrHo_pfp6BhS{mSYh!}-2Yt4eeh@5V0m>c+KFZxB7{H#
zPVzC{-xqg@=zap-yw4qPF)}0@4-KXH?SUZix*zYXf#;L^kF797m>WVTK`fb)e}ROB
z2!30%1<3Dve@`YpRg^z$aieONfRlUy+RZTC8d2z~zgKF}i(2U|Uf$lM99buX)A=#V
zi+8TRn;eFx_lOASZaxj}r`d}zub}0sfq?BZQMy47>7^e;d^yuJi@Q%<b@4q5*)E-S
z$09*y8DA^YKDcDXTE>(kB9q?A8qIsSb6@Q`Jj@uJYah79h1TE}@21QSZn<*-yF1eS
z^eg<A$>rCUdwKnSYg=1Xz3Du~xS_gxy22r2vS?kO!ly|m<4IQ2?Jl*vnAFUfWOB;d
zItc8Vu7*CwVH&w^B|jyvB+U>vjm1BRtzt6YfDSC{HL^9P(~hcc`KE*^IDcN+nj*sJ
zNiQe&>$Ljn%UhZEQ7yjNeRoe!PXbz>DPqEswW*28WwbgMmylpIHT&~E@^ASoe}BPq
zo#AlR#lNVmdi81znYUww?no6h>;CiG+NXE^{Y9UXtQNmSK*v7ha}U=(4AlGcN_%+9
zD4hH;ZE&}UBSlVG>cYet5fKqz+AHLzzn(sSzLb`h*7aAvM&iERyLY2$+zL=`yAPU?
zueocB1mUHOjEuwzzA<(wS8w#p+O%&C`Kr*TrF`W6zx53ct~q}Exa>6<o@iyd<<H*8
z(A8jsE+>MwGyOZsmn^;%mm#oyj^k(6zq7U!p|c5eg@0WI`GvG2NL?~uDD9ffD<cf_
zgVt<bLe}?xy)-PU<Xiu<1poTOzQ0#taVh@!ug&K*ME}<(kssWoE#!;+bAkWIzZF**
z{I&l7`VO)v%Y^=Hlz)E1;-8zA{Ix6o*LQvM4$S$_Gx^syIX?aG?fm!Y{J*x@cztQT
zvCS2=&{HVQutXFop%qL_OyS7#ZQQu=WlfFD>C?QJ=R&t-O9=Xtc43UMUW4iB)4oWt
z`n|)v2TqK29{3rkY!Rf-gdD=<$B!Sk6{AJFph4lIXf95HvW|}*p>bH(IE_)wV4M}K
z4N>8RcC!3!E|BWN#=euN5|hEaHgz;a#C*36`T6;L$1~Zk2L;`XjPylYvRTBd;x5B%
zXcN6=v%sGt%eY%#U%xp+p9xio8`(B*K4oFC+SJq(y7%R1&4W*^!kksRxoM}kc^knD
zuf8W1MTZRSCFHcD^ZV#LnaB~R`eTIzc-Yw3MnJC<!rm1zIh;yHnz4ey%A<Fr{E-sj
z*tLruIZ92b)LXYU-9xfc?+HW`^hJe5xm6i~fIEp`-p>sXO|V;H3Vb|fYevO&LqS4j
zGU)tt`qs1p*z1l}NolDRY^m%6iCUJF___hy&yARv00@H&o_KCX>oP1i|7EH=Ar}{y
z8V5xT3hGH3M1K#z2}AV7_Ak}dr^jf!1f&Q@e*L<dn!4@w?b~KE=r{WXXclzYUbJ(%
zjs}!LeQ`TaEq(RsmE6DznS(|~?Dh^0H)HRf@dSe1N>P%N(T)uSNr+aBTX}d^pwT#}
z?v<EmxA#ueABz{Rm97o>&R)_(3OU6kU}QC$vl0?0m@x!V*Q!UDs)GT5Od(qi=t$XG
z%t;zY8Jx1Xt0GkW_U#5_8jtU3K<4}^Nf%;Ji0|5;NYC({V`XEz4o@1?$7gvvE=~km
zte_dRuHCwS|7HxORy=a#1#+)WZG~S|Q-&nuqJD5P-qTE1$DO-gxKQ!#oldTywl)JN
z7gt`4Y>mXDgoMxLHhaXxn8<sTlw|DfcL4=o!58%O^;zycfNc_?`o}Zy=mEBhVB{xB
zw}l2)y$IiaoEB+Ld5BL?>?TN7AA7qBO6>c`mfotpgXu?~AtKUgo&fOo3kYaBdy$5Q
z1`<eWWCp{nW8>mp%cJ8NZ+u|*?b~eV8b%G#c<j4C9sdAQH?-4vP#5wd+3s>)US44)
zyM}0yW$2>y0Tnz7rewU|&!D3w_>QaCa2x^x)JTt31JgUQg#20lJG(_i6`)dq97zuK
zy87Mo<ZvE)un~vT2g}d7ZQCl0eA9)xXkz%I{_^!&7Tq{cMSj1Y<bG5q5eUjk!jkf#
zs`{E_C1o6*@yp@n*ym@<kda%4<mGd$^Y|`a4UKf23>FNIHQIRi@ZnZzv<Ymn(8$m1
z%<JedKr_`9cz%@ts#wWO7-|+*#SLK9^!nwSC3N^vIsrxyHg%!ezW;4%hz?CoOF9e)
zD(ZqTI#yY@yD^prGYGq-p%75d-5ML<i{pdlnRM6&xMPxa_oemkXT*Mae@<p%0L?M4
zBco=5QfqfV<=wzKxQh)?5BT8OaAKK|6Be8h|FcB3^k`>zTik9b8cpl%c&Sqj<b^DK
z1+tanyV%Q4prB{zh7GSLE<V7rIV@kPlcLOs(k=!h?w$eTs^foUugPh@#LhpDaSzBg
zuEMuqP{n?R9WRAeA5m4s5YS7|`<h~pC@LzF3EN*je?Hm&2H*ek;2$ia%qA%*i2<xM
zK*4X?O{_RQ`rGohV{h;Mp8WcdjR4EU#Dp#an0J|mG+bQs{Bq`RQAiPpBmwiDl^E+X
z*ip0r)v~EoeMkNvl%>TkKY~FtC%~Zi()#rCz!F1WAfGG^wtTI&z_Io6axv3LDxLq?
zB~Mz<y-B1id*0c(g>+3#tauZMsf{O5JywXWT*%#<_MazLODUwkg#Ritz+z&Y>N;s@
zX?%DZ-G&VtP}d@7H-o$n1#KMop8F}XqO9vV>i(<%2m@Y0O{JcCbhGTWC73e?+r1L!
zXtyKZADop^h2S>`jqaW!r;ZC@OsZMQASx>Q8)|?+A?BulxVX)j4Q>k~{T_cEVd3oT
zOtpITGbA+kde&0N`LvmRt%eWd@y*T_fto8AN_K(aup2kLF<h)0<svZSzC}e6gF{1K
zdU~!xy$}^s9joz;m=4&{WfBq+Io3rg&NW2WY1LW28A1^eM?Ujf%oMqC>lVKrTKXGY
zZw$A=>@RWmUwp98*C0B8E!hZR(6QXGo9XHNRUunKu*sQZ?>51lLjAG?sybS1{k@Lj
ze{PgV4|`xe8{0$IGhb$BFW#2+V1ZP;64&_i*Eg*Xt*y^KePSOT9v&GP`Fv<+(*v@c
zu=nf;l(x0!qYqS?dg37wIr$iL_pCsL|57}9*=w5jFdlR*k|G;24mf;0BBCJ6QwbT7
zI;QtJ{`j<cYHG@t_8J~rz>)>(Z8&)JrVufr>_2GHjX-R>1hNr1E31cQ?Evr<)zzzT
z&JN;qpEP^1%IAg+ZsrZzb>bIga^gfo#<+XYUj7m)8=s8E;2;_p7<?WY3W7zc35Lc)
zN~reTV}$u5(cic7^RI@K9%h^MJft)Xc%4HR2vcML6`aMmuoI%AqqhnRtEEO0e*=Lu
z#dSWHf}@}fWeO54S4&Dtl0+Upd<Zjf37NAmIF$P;WfvDyTc@?Xy&H0?YjAf~?NE_1
zK}J`F-#~+Hj<))yCVD)YN{A%`unXBR5aHdKm%n_ck$T&S8AkF$Y5mY0Fd^g>UdVuV
zfDQpu-nT$W6S_YNW8{-SIirN|K)}9cu`1Y9I!sqY7(<$YcA@Pkv5PWa)o`An@i3Zi
zuc}&!U3uQgNgi5{Q>LcLXK%xCF<?esHG~v_0@^s20y2M{_T>nkj$k}l{zt6hmmfc}
zTn}GDn8hJ3&Wt|3=bfFa$mz!5xqg<+dILj4NJf{DC4m*#xp%KT927cQ5=io*s_MSs
z_2b%DqR+j(K6nPt8XMPRekuC)3Sm}^pxdd#tG8|2R#;T@tfpo)#%>)dvhO`<V?&E0
zYj*lHCp;#$bm93gOEDTOVBEuUI}EL}#spy$E^gy=aB#p|6Y;s{&!6`$E88z9D2R6l
z1Ag31P37X^T87%%LsnLTIJg0O?sPfM{<dpeUVuXg?mj?1aHEpZ#Wdq5(m-x@&^!k9
z6+C)$5E?4=486S{+S`*4`B1w~|Ki8pU=W)hoY)trK3PH@T)la7J@(=%z!&^l1x^Ka
zF@a(@Dr5!+=CnOl{3?W*)<-^n`{s}Sha}2&9qn07B34MmxG>1B?`Kd@5a8izF)=ai
zOoL5u?}c6q(q!{uK;8E*n`h&(=%--bbTLtFP-qsJYX$6AJ4nB@vy2RL(R6Y%xpGwX
z#F@0`T5Ldxc4B`0vu8`t=2`aOK?=4$cXc+-A`A-oA{jczfIlH$nqUluGAE_Zqqu1~
z>ajH(Hw<HYp8q{^icR_=GerDVXgzWr$ZmUiO@10#3+V37J33y)B3oHm-2h(g=`q7p
zF)}gH6{jb=($viCD%h)JI5KPz1!rf`;m#)^Ag#W@>g)?LBs)0MkQGx;2nIfBX-R{}
z7k%s27dUD3m_Ded$AZR60od0ZnAWQH<9<Ksr;6r=Te#!>Fs1ei5#d>|0U?lWjC3U2
zlsA_3S#2%#>C>mx)3wQ`xV!ITm%iwS0f`+Lb*b3RZv$3M2Hl9;_F<h#-_Z$wB6EA|
zu944?Wfz|7;Fm~<(X$<eB4~<CjY5AZ_g_>HsXWKskJ}Gom5~QkDJ(2}+R?$bj$P_0
zJi&RGD9#zHAF#9d(x_gTaMI>sO+McY{uuh6;}a%WLX*paA-XwgMa<w40AK>q@ouBc
zUDyZh&_4lY1QoKqo3U%T+dE*IVG-mpR943U*S-ZSM}$I*hM(UQK;(U@=jBakw5|*a
zm1EAC_{HBlp^v*nKCyLaKF%X6+^HNcyZKXNd_;uLqenZhqZ9kj;d}6ckT(lsr(ni>
zFhI=q;6B{ihXUIzXjD#27f?{8N1zrg1DXnI0p%%Mm%$sMq(qzBLHI;YPj9x9lT*vt
z$w{b`Pr;JIyWKR*ZO6<j5WOn7(wv-2(PTK(R99E`B%oJP;9DAMYB~mnyJOE6-uL$P
znW7vWzF{e8sM<OKjkabfUpr_X_~1im?@7KjIyN>ipK<vTH19{(3du*9^_kJ2pH+=H
zit41p<)gg32&)S}KJ5oSSp`{9s7I2wG{!@+-??)qX&mFAH7-1>2R1?t4BZeDD{COS
z!BQG*e=#5M#PjT5vTD_}L)Hiq%*?pU7v?X#c=6)wEIaJdPKW`2{pd6v`Vjz^H`w`P
z3v8lS(RNpS2hn=^2Tvg=3!K0G8rB#$$rASi%KV+EM<Z?`#}G8NlTrANUPOM7^$DRR
z#I}b9Mf7k{H>^6*F&TP9CIBv>oM^Z>yp*{xi5sFHwfv75U_(H8##cT*zNgT?`5#!A
z-2h<t?%Wzf@qPQcP;V=P2da+oUp1AMzyN;1!Bx=5(b3b7*4jLsHnX%0ga~g5R#Ef9
z<Ls?e0PxQdAldc5;gC0dm!`1{TjJuaCl2=Znrv!_j*^m-voR-(90JE@8Hp$K^pc<m
zB(dAs*!5fr3^g6_g|Z;UJ`WGm;<VJvb_`%h8CYQ2t5>gnMu-q<4Zj<JgumZA47cIJ
zSZ-0<A2^!ubH=&S`}VN_>@;1n<PZ`94eR(6iiwu54Mt?!Ds;1qvK&^Dx>j2M@uT6c
z0~^`dg^Iu2;B{+3e;5<&Z6bVaX1@k?<*Qh+b!1rCcZGET5Cs3=E)&}Z)?xl)HKqxQ
zK-3%6mk#LYZ`54^lZkFLOf?dK*+JN!U$DbjsxA3ok;NVR>2Mua@y|&d`Or?W&r^Mc
zz$x?>h63>{e#vbWSV1PVQ%)kvIK`h_+0>-@^3|*GS;HGqQQP()<D%Q0lbwC@x&<t1
zKvg>Cw$rh*XTJUM4C2f0`-kBIsajiG15EHH1~UKb>^z1=WONRARaR9|LqDXs=M3h3
z^pmOap&neEoZQ`o_PqgZ?}?CZ-|2X1kLC$<?q|Ya6!|`6c|tw1rf<Z?Zin2{aOdvb
z8{$$Ce;-1#iGF4=+fNJtl-o89KupEmjlGunF1`+bHVYwBkbH24YwVy>OzS`MkjVPs
zWO%CS!!SG({Dk^WF|qgG4VtkZIWXSrLD32sIyn5R8289`rvW26>`Tcroj_N-a<fhp
zb9{6BMCJkA2n|i@_vrijl^UHneH5yTp>KK)^T!Y*Om+E@NhS3&3=)8G3pbMlTFQ{V
zklxkb&&P@m=kbH$_D{U#=H?iybMbqv(U@{8#Jp+Kw>j5G<Q3kLgL;A1&UJT7Fcp<~
z_VH=*1RBeaU=-4X!4`IQqd_~18JBslK#&|ez2+|xFQGRVjON&MXlRk8(V=YbV`OCX
z9D)>vN_BPht+SYo958<e6^VH|KgOKUr6pjDrMSl))?|C@aX~S6j5KNN#tp3HIB>w&
z`U%vWD)sNNA~loln8s5xuZ$_{)JKmV4Rx@_R7d;kXs@A=^W?#U2cRwl+74jmWg1Fg
z+sesQO^<cpZgLK=qD@copg3tqzB#oJ-o)3zulz}%a(w~*`jD<}bk`(1BV$Vcn1{yo
z>({TMD|TW+!okMh*bc%^9%~*LfI-pom=LZ73C+E!8i7YHyZ$g6^M(z{t`E`k>hP#|
z>6;Q4+qmFW2+L^#zXWzT$+&l0Ym{fO<D_Y<M>Q1J!;B1`cB@aNo3^Rl5xnR33r%E1
z{rvr9pdQU*Pe$H^90LS=83lzT1c0I01a}$Gh9HRxh>na7&|hPZ(L!))c3>cw+HZds
zL=ooD0DiToKXVl#`g6v{k58|~po;O0vH3tX#A;$Fz%#g6>sSDXE);BT*qjVE1(|jB
z*)lYh?+F$MYhM8)7qKeC2Rs=Z-0_^Cj?AgZ$VgoXVo?~C(O9prU=F|r>)DMmEI9U-
zv>(Xz5g-Bv=<HrG%0+1Z$?YH!m&=~ajp_qtL{LH{D!*w!IzSAv452ndo^5m5dQc|h
z`BsXoWtU<`Bi6>V)$SfQ-Q`ATz~2b@D$c7NJm}qTb3mA+Cz}@p>9L0E*M9@b`p69e
z>Bg(K;o6y*nJBTR0=2+36D}tb92`tVOIro21Kuq?st{uHKy10^t*s|CE<`B%;C!A!
zaeq?L9*+l5FQd>`w`Xd%1Yq(xmrbNkBxcpMOl-m|%{GG?q!<`}h#NISdwLn|-;)4d
zlB_Xszz+pse(x3l=UpLKZ+V9i@23!e!ed@;bC;nTLGcQgw6xLVr+Xbq|8!CNDGgbp
zZoc$BI$7Y}r*Ip~Fhr-)0}Js8GXdnSgRWmE<O3E%Y535*v5=6EALayftesaSQ;X%*
z%J}Z5KE&N?)I3|H`pwQOvqu#{n3=Rrt^pb`UvP%4ckj{PyLV5w@H__!kyzuZM<*uU
zeHTd;h@i&NsYV7c&0&r5t(X`Su$(v<=dxc<6$6gkS1pf?jeVec2gTm>uoXhTinmfF
z(Uh+3h2MXUEiN|zrxt*dBb!?eOZ6L3K6ROmt!)U#;x+bID=Sl%&yK2q?6Y9c^$q8}
zQ8^6{u@O$LNktH!R|PbA>dcwcG!GzLBxg<n+l_VLIl-VG1U7MEucO1&5dDzFtXp|u
z^wxs2ZT0oEgn#I2`1*t_KB&gC9cstzL;QBY^`cTgkUypm+#!n3v|F7qs-co@?c(0h
z_(sCi*4S(e_95)dMlHGNQ@B)Yfg}{}xo^rsfQ&SK(9t4x3PCv7nwc$KufPjYBNljX
z;i;+$TkiPN!A?<G`8gQ^_b$CX`gv`wvZF>9B#GlJv6<*}xr0m{`E^EM$BwlKxegya
z8r<ftqN0K^%Iw(WK4d7pB#Ic69fx5<fFI8?6|QugD`zGyb^kTyTHtOC40jHiaRP~n
zkt-=yIFK{6;B;GDFIrk!l8+ydA2U}ydzKF<fTtZ1<o8}ld3+S87m_!;`K*Sg1%Zxv
z7&DJ5K+XD%8}CiVF+r(xcdWiq*~MGqaEj(BpzcRGIhq}a*;b+d{!@G}3QFurq`Lb|
z{R_+tquaRA_v+Q<HS-8J(M;|3JCT9=;8W3?DXgs+{*pntIB+Pm3{IG{$TboS2<9Po
zgI|$XSEmC!yoBPt^`N@M-A1`b$HsVORxMxtw6an@w;amOLJST*JBPDN%|mzmp)QBS
zX(+9Ds+*Z$%36IMU`R)j6N0o#w57WDKKk7I%ph=|HM|R8H3G42W%bq;7XARn87{M;
zckwajJf8KSjfD&l4|ltuQK@dKL$Z)ZUdphq-@-lsY%s;x7z<b(juP4ZE)5)u<kj?n
z?>56;A3?nYaL~_bq-%rjWW|?dyvO9<0bfAtfjzzm?fQo+m5Dn`-WId^vN!tKcOlS!
z0Us^`cScYRWd2VPnUTn|W*)~1GT2@HhcvS*n*rI0`q{`W1EA)dt-G44Y9-*sIJ-U}
z>k%L#;5R`50VEN;sH?dGvUV%Dd$l`KHwQa02kav1>XOC@Brp2Ls~U3>RnS?b?+~TW
zq6TckkOPG6CUq{`r%o`|g)G7{>1)gZ26yxWnm6oN@)><@E1|B=Wq)E)89<Oju@|bN
zDyeJSBa~ODsWq4cz&`k`NgT0`#r~12eD(UZYR){+A>H!C=%mIFtUxmCOeh(IOAI+!
z08U9sRvQ}|DC73mMhOeIRe$)vEF&Xx_QRt@qhKW*dFLOWg20tlAeKY>je}SU?imu=
z!$2g_L@fCX?@vrUBHT;Z)af2CdT9^{qQC{6ehm2CW$psV5~z;^PI#HfetqN!P$0VS
ztgUT2;yo}cwU?KdReSzg3L!)r<x8hKR^wo;L^v(0tIG^>fxEr9u#mKng-|%K%^jgB
zSk&A%l~<TFwX{&>V*DB2P@`lZ@XD2CsM{I0zjPLn3(uA!G<!86^Kq&k3Pb90!{hEC
z3OHNJjq=VNlV>_im^c`VE)cAIDBpsGMl+>lHPWcjy@UbV{X#u>@K}kQ>@#L&Cqae#
z1qNOLS%*o~A^4i~x!Xo*nkm%0N20wkrm(iWAXH%cb{g>ARlwF7m>~0eYN|#kU>@Cc
zMgYPeWIWJH4Nm&_26_2xjD^Cq>!)B`2=n~S#7{^7qormFkIN3f@<NP&E*)CA<p2mf
zMMWdaEUx0A`uX*%ZjEAH;f-Qm23hXuF%)|6x9M_<if+`<&`?xSktMR6Ohd`mZC_Di
zzz#nL&Z4~D?rTraNgT#P)UtW^(QLqpqe1GvQ5r}idY5^C*}Mwfa>4?`D?dXml@HiS
zm8wQeSVI)G5~eD%%^f^<u66rH&|}n}BIeQmaV5CPV<mngh=bhB`%JS9Z=f~!aVa!m
z7yc>bIzmAzf)&F+?M8c)YNU$t^7aDi1p<FJib9I-eyj7?Ha<S8PoF-a#O86^G-2?O
z_jnHF3()>0B+UZn*Oa}?`}*~3Kz*t(xGO^Lo5h0GXpEg+>lT!pds}u*FK8YTDPv!e
z2S+Bbk1Acl7WD=&>{kH8edMO0c3f1{H58$KMwt=a@+J7_YnW$hCN!<;1Z3GNN;p#N
zT#aAP5N+R-8SzrYu`yG}lLz-7Y8}p9jdte#@N1TPTmY_aa_<KdL82_h{Gy9fyGSKx
z)anH$6vX*q?T*>6YDO*XI*9Ho!Sp!=I3PG-M$f0jyAMJ_*5%uLq=r3gEJp;uWEj2~
z_86`Z!&j6v*jB7q@uAdhHxg!#5C`U2gLrrYZsW=j^)HicQP`tQ_i3o8TyN!~T9OK>
z0h+ubB+SUvu_)cPf<?C{V1t1i`w4N25v0|m-X|Y;)0(>)gFVD~kKXn}$<F<Z$+0n6
zRKz_(p-oL?ABd|eY|bokXKMm&aFEHKuxf{^d*I=LI;9-{eyrq4jQ)LDS9hUz9K{yP
z(aP%@tmxTJ5KsO>wZrc+Vh4pp6B83jO>|<QJ2s8bAz4}Z1C|ubzTAnf`uYOGQn#ht
zNHJvvfQV7KDAH3_AB#zi9mCLjkPMCcctVcEp@LxPvSr3aCr=}>gYRXS8QzCyLGnc?
zgy(7diY`r5YuE0mN;7i$`f5{+d@D=^D6Pnk8h?Zt{i+E){yNI@zCYe)#rJdud`F2W
z6<AG{ag9UgmI#i_DvZH9a7szdM!Z)ASU&Z9dCg}19fE>8F<z!IwT0gagqb63&l}%W
zu!<yI)}uNJ*m2&@?is8Fb;}$$hcx{HAtB@PhvJc#D)W1Kx(gm4N$gvJ#oMZrfjO7@
zSdXUZhkr~}+k!;f5>P@T6B9bIZv_Bo5CM<Vlln)y!4D#k`0$-u=&9L&3hT8bhfLzs
zd{Q)j%G{i%=<F=|?<*WRa=nf97q0b@&BuLck3dIAGpwV0HQMz&>E~TPjp@&-AP&}p
zqYbs%7RkEz*WTc!^RyyYFJE4|eED*QX=wsIkV^^`3rE66dnm~YM2F*0+cEo3fVWSP
zyTOL>>K#Xn@C8bygJ81vxH6d=8yK*`&+^zFee-K_FF;qsuB96MXj`{p35CQwDw_n~
zy?dAQWEK1zmt-&ki>fEa;H|i;a}k&iVw1V&%|SZN(JGADL|DM}aqkh7f+~q*Glhyv
zuX+d7Ws>Z8_VT5o<8>UEx?l~YddlVnKeV+e*E2CQ^R*rS?pj%0ty0elPcLB$kh6kv
zh}?ZUD0gELh=aCG$e+h<_mJoWRWDavQ(G&L^P>&%w_M-r`udGHQrt5Sv;|Hf>Pj*N
z`Amaoj3Gl!MP)-rM+b)LOcs&hI8<xaTuj066D=@k>ygFfYMt~!HqyWRJ#3xQ1WG;c
zn9X7%<cvK*f2(sdzP`S-<t|9~Hx?nk$BWNZ2^K+-NaK8Yd3mS@x}j}tb3tWWsE1|C
z{giPxGGpR07|XVK0OCV{k!L%>g*Tc`jC_%;aq}WPNy9Yy8$<?uMAR?bX&Rx-gKV#|
zeI|0`QGGjjcy{mBvX_Fh4C+SSyI@e{=g7#LIuR0_pW>e~2XB${gh|l_&}HZEGp{lK
z^LditEnM}W+o$Kx4>{jXOXCaj@_IQpm-f1I3WhK$?)dNzvvjG*tyFkumlHJ8<!0)7
zb;3eJm#<p2s&O9i_90A<>tRN&zh$Q35H@1}MGU=R7)n(UIs@1m>hWZL+H#3EV7>rY
zQDbYuR8e`m9^H=P)q4?@t~YLHfytL2-5*xhf`%@7&TGILafpc>cf39~<1MgN>TfN;
zBn|9e(l|DS7Mm?te(r9@t9|cPjT%D00jVQGVgO(<o1wY9m82qvW_?SmG@2SXG~crZ
z?8*|Lx;%mwQI>OeDl@Z4HZi9Y^SI<4iYZq??^6-z#m2R7pI+*ZA0(KUC=a#tGm-xj
z%eNFlPH=<G1%cAO%^RRL*|@PHk_2aW`o0Af_((45_P&v+oMv-?(9Oe1qlZA(!J6_F
ztE?Egc!mLwY3i&3KQ8)^jLfC8b1*LiSrKA1$r^*t7=QdWv9(=?d;|eaNLxJ0&-YD9
z>D?aE9OhoiP62z(7xN&^pwx9e+JULhe0YI5@N819>O<XZV4(s4CYVkbz|0}`UJ0gH
z11L#XU%#TQt@uqQGl(qu%@rV5iMu)wb$M$7;5*T&AOlR(5`|gb_F`Q~*KcfdQ)B}A
zISk|C#3F0;ei*4W<zdb-WDg1Q3}d?hN`N>D6vcA1RaIA~zB#(9s;Vj{<m+t+IRN<O
z2Y{*qG3H}6scgKhZ!F2c7zLk>?U|OL0_6mPHF>XkI0lXKi&KZ@K6Z3)yKvXl)m0#c
z%iSF#<+cm<ZClZXAnE9s7*1~P<zQmzti$5s)?!n4pv+#;^d`Etv@hg5dc^0exQKcD
z(h6&O6&9xYw!kkd=C59}<^={q$}c!OBH?lE+A0%M)4Tu?fC*&H=`%=D4Urp9{PVBK
z?En>?XuPPSqqC?fLdb!9TluTbvhwZuWo2dk%)a$OKo|hhjOSdoA~PY5BlEehFEu}_
zIc(f9MKk?#SJzVV+=`3KrSh>ekjD^iZ+Un^%bty=dJ?1v<lX`(>Q2u$5=dp*u;B`V
z*O&E){@_GqQs=Rr8$3eeNhSl_6Gun62P?9r3>m0H*`g54#;<3KP<g@$C!)p;-L-v?
zhdiD;kFX6BTX;YU?U1OCRS-H*+X3aIM(ZCPqnSEzXqYuuq@X}L$N=6GTQF%?lEmG>
zW|x5M(=)Qbb-=E1hn-rKcK|KE%!jcKGYE%Ln~RjmCygUjPYbGU9hgDVNS~#8YF~OF
zeWf`t_PMYD@>u3N-e>5iPF6pSiH=Ty0sIXL?@M?0GUQ<5s&fkpqK@2Mb`aWI@b?7u
z`%5byi-Yww!3i$*O*;<o`5S(+<z0=(HRFp1dawHWDht)>h^c{)Ib~_Ny$uszSCH+#
zr=U5>s%TQTsRvI2do$FUm|nLr86Rp!>K(O_=hjI*d5Q7Be674l38P8Jp~Z;s+jC@1
z0XRY(;6p#7+EG?pTl;2ZqBh=U)^28Q&M3=`4TBdCu?6Ps$dG>lh^nLu8vb>YRTV)S
zL}X-ubB$ZBp*nv)-S`tceQqZ6x?TpP`bADYr2xJs0-oitYQE%OUDQbK4uzLKHb<}}
zstH~2l?L;5FT;10RX62Z>ywli$?f*6omA%EUk>Vxqy+cYE;BbX`#jcg;IOlEe*Twl
zhOm~TBbc7;OG~ztFKi}Mh(8t3f%)tlkP-JxVDCUry*UD|C=w!D8-&lX6wUHEB;#Jp
zV{N?V7J0EWaFzDQ=_v?6lttu|C1He$Dr`JGrOBjT$)hktR+2E{LEX!|kuw=ciTf~<
zo16OriT}&kgCr5;e&WVj3fR&2b>hWT>((8Q?a0l^IRv=cGI4@Te>TOO)@QiqN2g!2
zU?skp%ZMh-{$}6jNQL!0E@gwvEh$K~)|Kx9eyk>ka$xi3%{(*lk}gLZrR<eFoH|bx
z6=6hU*4b+pt-lJ@05A!bl=yWVa@!F0e3H9v^EFJPocg@gtJLXBf4?6JHkFMAF&w!H
zP7oZTy4}s`{l71l+v@y@?FGnSKeP7md$VaAV*<+9FI@-#a)wBf;MSF!)9o(ER1gsv
z;3f?YRUDJqG9E6m81NWK+}zLD&>npfxh}l}LmF{-V>hlCenk$E(Rj{QR&~#d$Uvbi
zbOi3W(sbRXUCUA9`d)NT8Mq!<9A!Jep1g72@<qSnYY#rQ#@*UJ+1qn8S@}Sin7@H%
zcoEdzNjX7_+TEA19bMc5Db6uFECvS$xn1b$kpZ=_$xO;dj*5FALeO1nn}oy(;i2Xs
zb`x`R9~`1n=#<1Vbob#4uPtI0BuS!MezB-)0>^|-)y6UzT<C`)hiy39zeK;%#ISXu
zs%?8)#uQ@u-)I4$@+xy3fG!#*Z9(1Lb|@zN^226q_~&RX#x}ou-+^QZNrhyH0W*C}
zxyM}$#>OfZ(y^fGhzs-;pu`x1xF%0n*1dkc76+d6Gn8*fqQ47D1_fjfZ@x%ao9=q4
zFXT}fly@TXe`zUlV4jyG;G9&_`Fk?&%1V+$vKZdNyuBAE6W5kBZX)4>MhIjeTfiO8
z6#vj0tSjyuP*P6^@%#{DW^Enbr)#ak63YVS|8;AisYnRC>&fp!=g;pXgJ=y8y?p?<
zL4%g#5eA<v`z_gc=GaX5z_*g8;OMyrVi4!#HJ_)PyhC$0lqMFHlvGqzDdYr&UKAtH
z)#$tJHDq}>1r!kSL>jmbAwvy!q%v%EzPoBI0QgYEdlK5gEDY>8>hZ?x_>Cmx-0eX!
zHl9x1q{oN0A$OAkE(Y-l02fEQPFNs=J<|AKbJu8TX_W<~X8>;xekx-{TpKzx{He@x
zq{qa@Mu?>LaENA6^q3>*%a@q-LtUfZP<xCuAhbcX#%HH0|Nc+Y9Y4SM;5-qG8Y)TD
zYE)?deh3MwPkrwrl~Y2?#N8pcl4pAQM<noRIk^k#w=g>-xe{s(#y$w??&C%Spkr=!
zbY#)Uem%D5mT@feK9}L^z-4DyG}G<7UaTfUK}3ZS)~xjONl^xD6lPXdF8E>`^>=ly
z#R0*=3MM8H?159LWVl%fxdVbFHrR7<r|;qQ-JA9uu;@o#{-sA$?&~1s&1$OB14%DK
zB6V=aBCWN;)r=s}<E-djtT5PBh{*dS<`{)qFI~EnXlELy5pZnXxpO4qJ!!Eu7?^_=
zx;NtEg^R=FkSaTcs1KmHHM@C8ZqB%E_>~?gUiqOO=Nqa@N^ZTKNZn9TQQg=)iA=M&
z$Mha#K#f~9Dqg;%0%xzo5X-UObC&tZ!n`B_8m+?+Kf$!UZ!d)t4EMNx8atDgUCOVp
zP#pWZJe11%?Y*_=`?;*_MH5^j$?WBSL`$<Pq}~9Lu=}maCx7XcnB4b+CL=0O6_pAi
z>V=Xd$mQFOf1EEs2_U<T84WcxjqUQ;T?yRjS+9Mry>Bqh3X@z{p70~i2{K1I`0JwB
zDUjvR$*dxxV)r=x<0QUB5NoAHFQoV<^AW*-EEy5kz*KE(t7>RihwMkF*b`W&;y0Ec
zIrt{zTYX<W;Da}%K@id!>+|~6tCI*D^|FLg6Sxc9FisX$;ecOAi107RBG0+$L{RTJ
zrp#>sNaik2H!hc%P*|!;l(oobgjt0l$&Uzuz))C`tcz7>ASs7$`?XQ+{P{;oE4m>5
zs07ay!0kmkHyt`43&j4HHUy7hJ<t!))mD+<Gt`q-grR6QEgz3pOAA_%g#7yEW(EX{
zp=zjD1L-Xv`=OK{LLY$>#_H<c(6nsS+5-n?M)gN5yAlmyg&?DrvtM!w*YF6X#<;L-
zX0Erlw^D(vr~la644M+rpe&-29&zxEp-d*;eUf+2Ra8{S3J)2n={bLIR<dX6gC!x9
zD-uV3gtxEL%;sZty>lc=KfDa=c-ipsfgeAFPoNRCGMc6+?2$P~a!K;7;L<7p#ZH|*
zeJ^bek188B4<w~(U(vz<=B4`Bym%9ZShi*wbdOr2xcyMG$F<(C6C)$uIIODA;X91G
zi#p}(tM!MaW-m9zujV0{6Qt@DJ*$t#O9x?ic1@LW2VzsAS7>$in*DiLMD8z}>Q26C
z)K(qOl}NR-0+j;;LQ87_<ISoYd9#wPf`$C#Ib%%*!{-<x{le|i=CNPwdsi1{(FI4K
z03xA+P_(i1D9ox#WfI{%O%aPiCB$Go27ld+teJbPmXMe`Xb8d>A!<0L--%)LWgcFL
z41CdN?TAJ@0433sK<O=qfYs=$*C{mpP*hcI8Ho1&0$lscLGc6uXT?}kPa!G0uB$!F
z)31Ueok^wyLVWIaw*Bt8FM6F|wG^LRU$dDgE*l9Tb9o2$x*Dl^GZ$*1)eyP!pXki9
zJl-mT3)@yC`yK9RJrh%^ofYNt=g$<l^Hyi*bwbCXorN7h`C`7=3{j;Ey<ACwB(61D
zIRI#)CeG+_P7W;c!-6SD_jz3Qk~v|xAKoHUpYq@z1pnX^STt`#c6IN$<~Mb-l`_MU
zf|GL>Wrj;KZSH{CQZYILUDlO?0+*Yw=6K!bq$Y~N_SSg5BN8~`q39bR%ga7hv9e}G
zD=@nl$^#8KH%ub)7F9;I0T<o>d=mj_>l<Ua#l?R|Y^zO2n?5yC{vOTQbVu4g8D@-#
zD36XQ0&oXfv8p;GL%JC{<TW!SMne8kg&EALR1wV#-HK!m=FSG1#iHx}cc-K29S*>D
zJ(sNxdKEPirV-Kpn>01}YNDSkdu42}2JGmb+p@LvIDc7LS%b{VwwincpFAcF$tUN>
zg^J>wX&2iqmN-RF0dmU%mzu?*A0iWMK9A>58#A-JW^-sw1_;BcXA6^zE;fq6<OH<F
z*{6^lX`dQ}^7&e!T^Hwmo?E1yykHP(+<BN}mFd%IXov}BKj8VWpa@v{lo|KD@g!iV
ziu9c~ZEY+F0p3akV>5#oFQ?03nz{85**VhVz9`IH$sYV$O+y2c4TQgCs{lC#Yj?vc
zdD!44{6xt~@P4<FLe{PDhP3b$d<#p%9B(mT9|AZ&JzHdvNc3fYoq(fH5=PgRQ=Ws;
z@`FiNz7va><{F#=7nqdPsb52tw9`m&MOmYMS1@Bs>S7t5oVE@&?>eHTw<;l+A6U7P
z?&d^~JuO=PERml7ZCWFG>Cz<#CFnBL{zScb0qZDm-aJLSA0h<-SI1rb#>FNVpeW6t
zD90=|l(dxveX6AoE%h~!!{3~w)4O9sLph7Kvc6FpZbdMIAYkKMYk2MZ5N463=xyg;
zFT%h?Qps!kZsf?F<JwFCk6hT3`5W8zDX{DKW}Wo6h+m)y)OEfR0-pv7FgrR4-j`GM
z%cA=gSdZzO{=fru`C}s^Pq73$&(yPNV~|r$MsreS(l6);6{Q0nfs^=9;v$S$g<5I#
z5KIDS$`cNz4}|pNqat9fSs=I3h~0`ba}|y(_Y6j>Zh{)jvv+U0Va%nz!0(mKsF}e4
zyfU+eP~JUz7|>i&OP`D(DXZrxplE&nu#||`k)F-cS_7+Hfu3>l1JDeG;3VthmZPp0
zC9jc3amzAvW>1z8eH}=HlO`sbY4z}Zh@-*=F-*s;vAV|wqcU=AAps-#^t||Q$fgED
zG_1($g{mK<MtLQwW6cmag^t69km0;EADIcR09Snqjmh`Dwmo+)0M6|*-WX;z33N+Z
z_vA>XG9q)j4Z=KHBCIAcNc%h#5?}-Qr=R}eB9s%6so(@%pA!U?Dcr*~TxcTFnmJ-m
zSve1|V03)UnCmH=M&vad5*G<(w^6g}Y!aF|3s^YEwns?_*glk95F<lU(=xGX@IOdL
z><|`S2ZA=M^qz`MGxS2ce(2DXZHN%}+yLHr?`>alaP!X$#P-ls%MBn14gg;|z*Yf5
zS>9!E<MJgW=Kv5UHv@;sg7(@tV_hfzgqWI<Bn$SHb!c$#5gcot+Vh7XutDh~oPUtC
zCi}%MQBmK50x@K4V74xSCaIkuP0;;oe-I@?BVXxdQlItmAh?dzTefT=ee^<ypfw%A
zV8zdn8$889IY0f&f}4kj1Ik_q@SzEmhhrK+xi6zuPgXHR#h0&KS>#7WQ6(ya@R7*B
zK(KPa#DQst5N#RBKT6CaOCh)6Q%rx$1156X2%<*k4&R6fMvUSRDk%}B5XkFeq&1oL
zY$0a%8QS122ZAMChD%4?!9HYLyyXRZd~Rap8bK^7qDlpLFk`p<ARLgfZ5c2CgHnM6
zjJ!)7@=dfPeTO6RB=ksrPR?_ryv$rCZki$>gUhUd=FHjyDNqP~3A#aZGIk^^wA*c9
z1V2E-`pKo1pX7!Qf$N62S*Yzaev=s~527dqRhoc~j-;B4sIXgqk{*+K1x^DxXg|7|
zot-_&8r~@Y&D*}9pR@wV_CFOLOKfdz^Jpp;EP+9i#28Bb@%()uT@4&ow}y@vgeu3@
ztrQTJvfy9XHg9e$_qb3drQ?ccjx~!iizP}s+*0TPvF~ud3@Auw<dGun0ENBoA&&rg
zZy@9?iKi-jb3#tfXN89^)+BZC{K=pw$`M*;aySUlmdYr<^3yTa{DtWOVG;`BG!;7Z
z(ZaN^fxrfZ9z;8hwrxevoQ-JLtO+muQihrxVD4%>L^?)BAC#4$XgLwVZBC#LTr-39
zFh?rr)k5UXAPKTVOp~T90K>BidG*k#trVoH4<~j67}(@`EzEx?_mIRcpR6Sv4=aHT
zXnL7duAp;*myhT$CE@IQ7uCwjN+tGW$SPJKW^y(EnXjC32qAsT*POjA1-8aDv#*i(
zBPvO5UfwPwkIvcpJt`=uMs6SdL@gT)t5m3EjsF<gF03)Wv7~3iu7wNbrAU`*3ow#o
zFI>~huSpT3<Kv8mFO5+kSW}J%CXGO_<sBL}f6MhhibrL<sU6M@xypYi`CT89z!x!{
zUiBfk-4wrr+SCSx=Vd4;q<;Q<bY(b}Z#9#!kziz}Fq9Rc*&jZ9P@I6AqU>xZ)Ce@|
zwn#T$c;)8ib{R>_qjm$|VPYAfjHP}(@cp|0LVU#ulHowE%?CYuRug(LDJhAefpya+
z{^>i($;t9FBzeM~1<_EA8$>dh@8f@>Si-L!Bky?HNIE5_nzY|~4#~w;SYc9hu>n<K
z-0ginfQiPA!<b~Pn)3tZa6Ts^EDUuqH@3D-Kodr4EC=mEXAB@i(eCL>1`%9Yt+oN%
za)Y)&L{A5}D@JhkzoaK|1ee3I))aYoKoNowkKSiHK^3b^WE^`C`Kh5mxE2N{H-^=l
zP)#-E#bll1^uxvy?tN{D7rxK|X<MMxJkuR1H%?bKH^wQKNFTgX%pzg&7^i-)&|m$h
zOi_(Y(a4^gzfpGqBE@p$6#5BBh`B}7ZtaqvU{>y>c8Z9L+W`SvTAY}W(9!$_u!24S
zSJX(Wt%m{hK~W+KF?xIb%Z*X12?HN7I0Kc}z<8g&kk?y?67%M8I(A&Rj;O~6Y{Y61
zeH+mLRtTjus`(8x@SOBPMZovZ5tGWzRK0jXg-p9}JBUSqSN;pP(G7!>^>Idq_02nh
z1g&w$%ihftW9H17?rcyA(Duy3djlHe+mI)FH`S4lf#;=IzukqlISN_<8c_1`({bjD
zNuTmv=n-KuH7AquF7kN%lOrU8I*GWElubgYeY#)WaTIhG=sRK$+&8<RAW;1LXK$nG
z=Brz&`a7Q-xFhYMmo|(u7Fu$g;*F9VBEiCOF+sP=k<n526D|w$9%a?~k?5C80V5S^
z4U}*N>F<+J+4r|T-UgZKubiXp-9&JH5won4#$kOtdL6*TAH+G7Z$Extdo`=5mG@1J
znguO0G?*ht%&5B5b8a8>(!4XSuL$#nuXB%^!9|WBJB?(NQS}ZQLNC6R@RJM)gtF8d
z?kSi}!lYmQFIVh?9k~hU?JzjCMA^oetyI6!hVncrTgrl=_0b{Ge3t3Cb5IM4d(D@&
z72|-^OxU*EXC#t2v~*hsA3auQSTr^|I?~_BRaT`JNtBv!HIY``Acv60F+mv{rf>5>
zK{SV(EDcFX;mNmZT?!IBZye=6hyzyH+uI|X?J?~LQ>*Rx3d-ka)zya|xCVp~;=k8H
zQOOateL?YWCXJzVMCnC7JC}eyA_^bqPwI?1L2f4HB==-pMc;@dv3i8Uq$CPEIsq@m
zudcJdMNo}nz-{{EJj4{pHFJ{slx!&CexiHZ_g~R@OpuAB9J_pkZaMngv4J9|w?-`>
zbH>JP_fk?45w^2TUI;_@tTAx)Czd?d-72l5MACD9>P+J_pdOq_!kYp3&lj)*@-s-&
zu>g5Dir?wtrCN%NFxlhZ%u5lpCF8)Ml18W56Jg75bcr6fUoR>mvL5-)MRLeALLSXR
zu&lc=Pj$~QD+>#0>bg-zJEap3%OB5^)Ci;OZ}AbPX=Bj%Ud-y2#lgvV`BmI?WGnPM
zfmXASd64oqFkE1y5Rn~58qCbrF)HlVt@QDwA5q<P66h&u9HbN&+-6q;8M>X|F*xT+
zF}irr3V{{{CX;1yzt_AJ5-sU1rKmxW0UD7TKwxqKQS-Nm_n;_$Vk#`h{{4p5TTqTf
z^nA#8>Q!prCWrFw9ogFZ*fJd4+;ZKXNFhQJbO5!|zqV-eM}>x7Ml->aP@U+-mEdO5
zes%o;+&58OhjE^|=-omEB;~FY4tW`=rG!a`4pV^WYOg7#455?(wT`5)Abp$}_MeO|
zAT=3L(b(gcQHVw~rFQXH0n$z-l*aK=UrD7is2V#vyMXF)xL_QlE0Fc^4UgT6!EA4E
z-VYilPob7QRMIdo`_ZGP01YQmOO`Kv;J_B#I(=Co$&m&Qvn;2;E-hvzf*%wDIIn@K
zL1AHjF0cIejkvfVi1A=$Yli#Bit$I1#SE?1tS7o4gwZDWJ-v{hyl6+e24vQG=(a08
zNUaO@as}v+oVaoq27zzzL6~QF7<T6;KNXA;UV8hGS|rXCL3{sD`f+R&-E5N;V7FG;
z+Wz1)3|}kdm2(H><Y>yBFn+lV3^bv`37|2WFFy(n9;o^!RC;{FTvSWR=0$_L!}h6f
zK^GE!=fmSOAzj(=`!D$7$!rxD*R_imJ4Omhz`773$u1slC<(-;-rPekK%qr9G>6NU
zXJYOwcA`O!n{j9W1)d}^9+0~FuZoE9V_><6q1-5J^zip|1=b3U<Mid^<)5L(J@;d*
zbI04JCK6T^mXw$xvi(ypxhA1!K{y^e<I`C<e_8yaQQ~a9eC0|`q@jOul`ofh3V4}~
z00!kI>+0+4q3>Pc34yUmnI&;Zu!2{R+yr7~+~IpvO|6=&KLS^fTg$-cytO<6k=8Yw
zV#umy7Cw}WJ-UB??q~Owl^O@-)ek;%(^0>Do^maB&$XN*w(kYsKK!Mk6LYITQ)?{d
zqw?;I+f9R~V<OXePD%9mY1OmPN?fKEc<j9^qjSr=cvOYN%$nO3{WIfIGj1n;<h|FM
zSu^u6Z{kdr^)AX!tXB@4^lB=qN8)~Scmhqpo>Q=;JV<5i{)i~R3@5Yc{;};#b23I?
zAykkuHV5F2vB23e1KH5@;JEN6%+E_aeR}~Yo`;W5(P8G=mdLzh>A$rAo@aAt-9m<&
zr!yt??W+YZvF+2RPde{2xCM>N4G0~IqPUxQ5a0KQHZ!M$<5|>TTG>|e;Ls0(VS0VA
z8`G(o$sfWHUq7z-ag2~Ub&#>{fR=9*GeOofTPA`KRy=t!0Zpv+cT*kMw(T&2TmcMS
zzX=Iy4G<MN`GieHPFAn0uCA^_pl1$QW)sw#-yn5@mngdX0R6Wp6mukAC(V$t)P!sW
zBcJ>*`MB-A^l9jxQLbc+GV~qLs%!wDUyTebOeoW-jB)eRr)v>dreJ~`#=sqs@R-hR
z^(LLaTMKOCOEH^Q6NFO9DnazMP)1hUY~=GZduY{OPxGaiA+9DXNXCH*vP2zKQo0FZ
zw+8bG7HWVr_rB%Xwyh3mip-aGzM$NG<ugM3a6;7p-yNx!x&UFqCNOfOrK$bQg!O|u
z3y=mFYcA(718~uc&hk5_PCy?S@z9`1(_w}yM2eGeZ7ojQSMv~J30Cm|!qZI$JWIhE
z8WQ=Lo8cF*?W3p#RK|*8xZOb{MP9VD9NXPFDlHlVu!XoM6bP#pjg#-90&%on@(08B
z4)h|cUG?^@8+tm0(?hU)11!@^aIoa9fYvGKX`u<MZRg;y0q7C&MDX2v_mm+>A*?4X
z4EO&1WpA*Sr)%-1Jq7XNcF*t60y^=4<je8%Bk~|f(C@<QboKV~;EaWWs6?L$d(a3H
zp0lorOD!_eT@)02F$v``Q82+G-#|<X!aM}<P!U-Wxx<I+aO!yW?rr(Lj~h#i<Nz1x
zqJb9wN}zdO)YYj#eaMZvo%;9#wj&>&VQs?3MV;20!V-7&^__8yeRrhZ4c?xIXxzU#
zUua<dHPsid0@=oOo5#P_JA<eQ15{Rm>EJ;VliAtXyI=_H2j1;~yw6uoK|vYU@ktu#
zV6j^Zkb*dQU2t<#LdEsh89Rvm$f(lS@5T^7jp8(^0DbR$yL`opO&$*)6iu4LTA!{f
zD1aoj7TOhb&*srg*W~$zk}uT5ABW)e1;}!u@Z#W*0ae5g?2t2)U@0ee6&BkM;RDgN
z!@UTgnx~62cPJ_f1^y=x2~I6~wmjMav|m`Kyq1^-NpIOB=kR(6Ypn3Tz58MbB60tq
zpuFP2%lM97NS7lJa4VlZo3;m~7;qn83*<@d5zeX@Un-&x-myL7Cy?}XOJujU@<(F*
zu?0!bgzawSOW}rU2sVU!`}V~p2M7ZG=E2INfv#C&&|<eR@8_709tylt1Du1_OE_Uo
zrEU)R)?kS0l|jPs;Ozh>+|Z)|HSvJ(bno8pqw&GBTm)5I;e^ZwW3m6|=x8YT^Ph-B
zKt0=`l?sAnB}K&>ko!qE_TNz5{XlP*PFi><)G$h<u8_O|H4;C^#@+^<GD0ALpu@K1
z?aDEvscS)YnPctg>FBlsvUCj%?NH|T4-5>$hb4>6#j<ei+^b_%r+XB<YmG<9X32X+
z89QGEb?T0~#~<)Ay5li%f>;k+uM#MAN7FV`t=ppm*A;(7Wc$yKilYD`Wzq|gn>j2|
z%r`t9n^E!6>u5@jU701pm4D^R747@lH7891VG3g(X8AD;yx8p)Fu-z6BxQ)r<cQMs
z2Vcnn`<QuVsP*w-cK3dcpOcd)1u6D}5E|v<8a5q;9OR4wa1WFfz;}t=is$L>`C7l$
z$2`-@q3__$nZ8b7FWAmVIL41>c3?}P#zp`__l;`2FqZ6VXlrB6#4qFNf!m;GX4Y+L
zAr_RfpHV>j2B>?~U5OO$M??l%*ZimHfh80^TTq#Ip)%(&46?Vyt&O4pKhkk8d6(2l
zBQo^?h<{JP6PQdgOxNV$dbDx1L|`(C{?+(Jb@#cGvyn&0?^8~`LlS52vK;y027B3q
zqKiMHc#l@0SP0{E7{e@(W1cj*kL=XJZwJ*E|Nir(_W&_yPZD!%2D-~PgK;BRzJe6J
zKX0a}-$i|W3T80q{Fp>FsuFmhMKg2=iiDFY<nC(OE~A$oUh7H`KpUBun3y36J_>&s
zi0JvW9@yYb683l5_BkrvM{Is;7z9NQXqlump})T4gAe)`K%A{LKRc%827yUsXJ@C`
zG^}V1SZzMr(x1LHj)3dgXZGTYYL^c^c(wRWse@38nS-UmJU3<~6_qfYS4C~@aIhe?
z;1w%z?jeoiMxQtl$p<J0$tN9Q!Zj<kFZ}hze5D8qi9In#D-?T_u}e4)Mx#fKA=C?Z
zA(_%dF_%SWc^NQG0Jc|6Dap~JvyE{;`d55J$|@OjD#VKc^{UuMjyLfyhI0!b4(d=c
zG-$AW@oy;<K6hevp@(5O*NJO<i5WE3eJsQ$tPSQdTwFX#Z6IhQH*|CFfgV=q+x@@Z
zdN8{!+w>Zl7!!&X8%@XX%Y5sX|LGNcaup(*2twSTOQ3S`Vmm^-P2CC_k{(R_x&hMw
zD{7^591V>VUWi9IF8Xr{p9L|<sAg9LBEJjY+66P<yd@?4HdHOxj>Lx2Gcq1qv-i*M
zQz$bi5{-g)Q^xrtt!B_c#;Va1oe}IIY&}tApF%=Sd-!e6On$Pt4*$Km3LuX?0j9^#
z4_)-+F}8#s6qqJT6CmrPgWz66SqaXLcpW5*d$2RG@|w^0;(E!;`?(6f7t4uaS*-t_
zvu_Tvh+6T11Kfc3rlT4XfK)|KF9si=LR=Y862fbF{$tPc@DY55RHRWRAXhwkTS*`@
zuXOzHCd&a4n#x2*M0lyi#ST~c&!Zr3`Rm13_t!`K{R8q}Ryxv>^xrSht$gUW`R|w2
zUuW!}KmFjHfNr7xemz?9*FO5kk54Z9>s0;YXHQrD_2B>UGw-$ky+Cwf{qG85PyFA%
z2hR(Aqoz#c*mWydfTjC$8h2X2g6FvWFnT?twtqaFq+Kd?xya!s-kLPrz58Ivc2XS&
z23Vz-1vkofM(`W!n~kte=9K_JgmB|ATpO;j+&sDCY;)R8)B}Y7aiL-G5UMs==)N@s
zkEYHGLlNc3crpd=YI?UEU-I!plUTwIwAHN*07Vz!$f)is5&4&${Kqj%5ajyr`KJ6o
zZh#hd=hG{n!Qatu*&@uziBLZh!BY77Aobe-;=RzstQ<fNTL_s|!AmSl4o-ng+Ch3@
z>Hg^J+Xmug2Y$*YgjggJnnSd1Gc3tR98h@G@@f|}{XA$c6!2^s(UuB2&?qM;cn=ZC
zL-zS6(jA7(7G2#!MG4qqnzl}~N`1*^>FJeEuWSd92R-_5=D^OK>L9~4F7G{LSNr;P
z1VB!prn0REv`s0X@p!biKNTf$3zD^)9@@6%r2PK<JN#gfcb5$OF~b#7&=2xoUXB}t
z1gLYJ=lC3V3|{L&miVpyafD4p=f51n6|9Q^%q?;szYi#-jOY=fekEjofXj?=paSo{
zx_*5P0ryDAr|V`XAup~~T2@wBQ<JV(yPX!jbjkWbVMQV_1Qtw${bHpfaSK*Y*}&j7
z1kHXBZh+|}0ZZWB(k`^U6b9x8$`J%Z*#$s_#wv&|w?Xr-(}#|_g5Y~^ATimWF*!aS
zft;!maKMnZ12Y&?eiGM^j|Rv?2Z2C*G%h*9v1Nn;sc8J&loWqI(t*wl!<TprxEP`U
zXqCX)tZi#THu+w>k4ac$BrmQzG(4OejZVndgI$O~Hco|Tv?I>Ax)u_$-8K$5gaoW$
zyN1*)DSUIxvL25_e5n%zlWf(cF*xj7)qX=z1E9wd{87){SsJ<T!o)+O^PF4&h0u!z
zje^fu62*6b=P~k}GQTnoOa<doS@r3?{dUwxBK!$O6OyuVVvms0x^Rq@&k0U`kkosI
zBBRAQ>9uGI3l%8JdI+Z=d8s`)2b#XJp`oFte5!%J3I@;*$+Xse;Pz$f@9ej9WZkV2
zsRlMb2U)_^>()hKAqiAQzLVvy!};@cC{9>SOB)F$#bD(+b}KEq0m!D!@oYvRud4$D
zTbo}}TwF`I5;f_myHXhJ2<@+Ce@;LJ^m1m%J_Cbplz=!2l@iRVePAPqSQ0|hI-uPh
z$iZ!QyoDVx`dX~>B^XXYN$0_x;;YbRhAgEySkTeuG{WK2l75hzbE2Rx6t^yM@%JMX
z@3N?(0|+nTu_!Q6b&!Sdh>L&d0_czCKtA*h(VJ`oO&pG3;lsV@iZ_6O<~S0g0MfM>
zc!Qw&{Z>+MmV#{SF4vKhh~4=w=`H1J0HJKTXXhEDvw2tffa2^y_{2L#yLN3}|58h=
z&Wp;*O&_b)DD-Lpq=GF_#4bQzKQ$7O?zFu5a*nl%;87gIsBL##58?5i`Kjs;MMe}(
z<2S<ub>$ke>A+C<6!$L)*CBFNK=yC*>u70(a#($rgs;7xm?YjP=DnkzHQX5KiSWf(
z8&#&H>Cf7HNeKxh5KCs@$9JI*kp=XctDET$x<%lWfZ!I@?YyVTidpYBYCLErvS~&U
z&!XHqIztv`79`cve$hK-+MbN#c?Q7CZkQ2DT&4Q}PKzLf7`veVj}2ME7r~&uk~CfN
z_p9jz$--l(dJGeR3_TF888x-=9|sTBjI#Jd%<5dnxM|agrk3nB?<LC|<H4gRgl5tP
zXecWBJ)5dUm??kwaO_911&PWJm_APz`2CUNH_g{lNJO6*;7DRnQdK=hYHgAH+&fcZ
zVU-S@C*%%yT6#~LnWcfjxQ|rv2?HHu7qzk6>r1Kg^`VHQUb9B9*-{t9#*AnR06<{s
zRD_Ie2I#)XN^>`4IkDX75TJ9qGFm;xV3ZaVCH8*)te*D%Ct}uPsQW7@EJZ)2D^g(X
znz6A1NBC&6Ac)}&@q#(!F_d2hE#q;*sUg#q^Sunb=97I8wJ06$B;B?gt?Nj;h?LaS
zD94C^kL4aSj@1hwTO%Bu9)2AjK6>Hn5=Rh?$56C%#X%LS9c|?6YCMoov6#3$4k}Dv
zI|Mxebasb;zisbU*HFF<&LaqlnWEz2jrUba!gU<QeJ2b)!S@7xeT<A0r<8I#xT9Mh
zo0Tm~o|Ixzcd&eXUDJr?8g?n&tfMK}Wx6^M<IY0x6hc+%8n^pd6L$8qMjzLiSqV9_
z4zeLR-{+t?yG#r`n6Lun;^~#O(|eutQK`|NcMW43jhc?-kOT62hugt2RLCeah!J&D
z>*8J#f`CN>)v~9@B#>D=&Hk&QHomJgGB!Fo%|&&a*9_mbZK-}Cxg$r$*Mp`vZfcSJ
zmj6zJlCb7CLP0XntO?1|sp<ps8ci)rNT(AH6z{AYSoj&@r3fC-x_$w+W{+g@Dt1AL
ztRm6M@t}kAMhA2vGAHB)jDc>c8bUkLstSr6erPxIT%Qnu9gb5kM0I>ejs-Tuacw7x
zljd?_+u*z~__hR2ORuyO(ke43=j<VT51dpw`u{QY9#B!9YZs_FMobhmu^@t?v7ktk
z-a&(+QUpXmKtO~6>C!uxke~vhQl%?Mkt$t=ra?d$igaNpiu5X7`rWVpy=&d9b^dk!
zoH&{JzW05~-uv0l)i7A+(a>Q6Jx)H`#v?#nDhQq@Z~8F%J@8)+45oxZGNT)^f!_J-
zjc}xuP0aA`)R376mt_D2BoBB-;qU3o-@jj0Regd1LPh+N$|kjP|7FA)1po<;8eBAk
zVAhy!p$p}aI#2-3W{opWWVrC~CUiFnREc{RWmZOx<0=T>kOe|pT(?gI06D1v>X)Cd
z5j7gtQ2q(Tbw|E8Vi55L1I=ZkwgSeQ)ibgV$daO-`|fk27bFi*emBCrBVPT{6^22a
zBLA15@^d^_ok|xle|Zgb3<>I>xO`4&nouY;e{b9QBljPX)sNN?89!g1#LmW+*1^7O
z_inz?Ge4OC$5#Y5b?bE*FlG`eqvjq528u+=vRhWV@)SZ<D*tsi!htd3`M+n5oSr1j
z=ko_B;44N}u$s8unzA--u$iN)e=ZiD5FbFc)+b@w<wqVs#j!&p@S1|Y=YocR?HLQO
zfXCodl6ujM7zdH1*)M21rEd8FEd22wSHm>{mXV+b=s`5f2{fcUUu!0wOeR44eQoZy
z$dDJI35qb{VNGn%D2Q!#n3pDn2_8cYD5jHOq6n%-{MvA*Mw?B6q}w-2pF)^5yK?A%
z_w0^=gTzi-d!Gepl>|J#-z^S-B=;0_64{FQ5vbyR&=Ay7^bk7od<8Mm?wVkK)c3V*
z1*F37>(~6C>?gnfdHOU{kVf_(J{=gqlH#q}h+9Zq3K(u6V~+Uk9H2FUs+=UyI$(?O
zX3(X2u%YFJ%<arWxh<+!fzelw)DtVPew1I~0)$zC(7~x@=Le(u{F@mYJxB@$BX93M
z7b4xry<@*Y0fWJrvp4zx027sMPwg_Kei%;p&wBaAnysB(5E+1kv>k1fHi$`^CZ!+<
zJpbwx!VlyZ|8xTpiy-A#SloPTv3m{vOng8@1$6bVkX!;zg&^73BKjws8*Vwpp(cuu
z&ICS6BlbH*J&5=?6DM&C_ClXr?*KEw7)-mgTgE5^g@=a&H~aDD&06FMV&oJDW||)V
z)y}~o1W;k`El+qfI&acH13R9r6Kou*-_i<s^g}<tAP8F(0W~r%;t)I~1Bc)Se{dew
zrPma&LPQxuPKw#BhtDlzli=+HpW;TTxqzC!%NF(EIRGM43fELOFw&@iw0{!pHb}-w
zI-Nct12-?MX^*=RQr@m^@uuk4EvoF~B?grJRQJM0)oC461B2)`NN;NJ6(~?_$oCOy
zj0XB;0PYPzsNKiglV84s$WkudZ3oAe0RlW?LoiFaGkizgdeUB_Q=s7TJJEIzs;lYK
z;I#1yl6aPMSq!YI2%jsuK)|t!WS&ou+8Y~_EJX$top=AK0>0z~H@6aD>G0a^FpDMb
z^6mMLf<upb0@bMi2Kj^Eb&`0A>~C%jLBHRSlKin_1R^D+9-5UCq{u)ubQu`mqa<Y&
zmBR6fON+!s=%AF#Ych&NX0|vupcxQ9hSudNQNWR*Gpto6z+CXMCA6pZQMjjWqbF<X
z>xHSMh~{?CcaZ9iC~*<hOzb#5A_11RN^9n$qL~FIS_`?;N!%tPb0?n*k%t2Ikz#(8
zjcKBO8hd54Q@Ara{|iSjYdYNZF&wQEd26T_N7`Y124V6)7au=%vbQIX^D@lKtYfYz
z#9}Qk@16g>)<I=Y)P3@WoyDFpFhs*o%(I9+y-JuHRG-0!+=>bcegpzUl{knh(GJAD
zICf?ugEShmhZUHyV_`5Lf6*J8IDHDn#>Owkc5gXnR>^WkL2!lPtZ&)*PUnR&jnbME
z7cQV=IYlzWMNv=askJ~`lkN;|X2_%@t*kX{dEJGD;Rf(n6}pkNz;??Ya+VM9#hHja
zRa@44mxqkSYj&`mo}8GA-n!-DF;F{8E+=qJ{$Hl1D?flVSx%FY<jtX$;XuZ1a-Y~U
z<qz)waAAsWZd9UU23K$rdg#}>8y6S!;b}kv5PJd-IyxpMkr02=(>6_vS?yu!A#n+4
z$ckL&k_d^25KVCO_~KdYO2V*H_nW4Xg5+4Y+Nr_AZ~`EXPOei$!xOQEoEZ5b4~uX=
z;5BCE=R<&-e_ALuRDeneT1PnQ+P%lTK4eiw;T}Ns2_=OyLl`FV#X(&oZySZL`Co%h
z-WWf8wpo!c;*iw*+*~l07(IYjqf(57hF}lKy+i~N`9k=<Ua&#Lf&18LVu!BV0WTH;
zt-plyGbl3B{Iv>-?KP-5n&F6Czp^i!q;Hb;VUgdgq+Vo4SuWpPmBp-d2=2RCQZc&K
zp#r>}T3p;m)^iA7sN7IbYB-EzK%KNQBsmab>zxcjF`@WfaiD~0+`NxvR#KE8okIdi
zh^VTmxq`7_K8&UXVMra>L@=_U)#8}}P8qQ8x1WWT6%>M?&-nlYdT!Y*A26akfa6Bp
zl$Me*Mkv(W+yEtI0OcnE42~V!1I0Aax~~n(Zf09}Dh6XD@Zq6^dw&0)7U1E-rx@S?
z4@de<dg-J_I|XkUW$H-~fS?24x3s8}N8hoihe;hchh9bpO}IW3c@qLrcBqobO9FKM
z9@dr}Wd()tYlp~35KJ99n;_gYyrRSn`xPV*8fs9c_Ek)~)?cOY!RL|`Mv5@roWw~W
z)G;bqoP^)NM}zP~BY&iH@tm5P!syCN;>k~1VjQi>Ai^r6LArt?(p~JVf@Xmi8may-
zUku4Pz{8^8GX(&;2a5_7AOn!u2eh<p9%nA#kDw)wnB<a#_90kB#%x~xAAB}Tj2gq+
z^-DvbEi@qgN(2GqwCSm-P=nTwWO9;&I_8XB|6@sAwquc7KT@UYZ^|ePaT98SV_+kG
z@HHQw{{=N}XS9tryIekl=hhVlS2R~gNP~#Bx@Qxeuh5%V!>d3KNS6qf?ljK-v12)C
z96C{bbv|Fq;fDMj6ODtAf&hZhI$?C1qK73BMEnT#RwewV0K*s-j)%LV8##_PDHPP{
z%;I7gIG$%7IvI)9ePujoDZ<gxUB-dM<>{F6jE$etB2ro})70y)j^S^gSy(SXQt295
ziNCWO{!1U=dNqJ$Q#tOvNv9JXrorL{A&L1Rl@f@BETkV8G^J1wa{@?lfsY{cZ~xGc
zNq7m?$8#cE5&-~|=7fL+N|mSRUYrv8KEL}^oX0h&aTz(dkr$$HhHeOP?o-tSlM(V9
zxkLfK?m_qNsWB7-54w+(Y=Cqm^BmO4W5)o%5Cx1YHi1(hU>6A`t}euLgCrBaluS%k
zG{Ym3w3h=<FR^E*1dvGu#B1miGWF>F_o%4a$EV(>p&7x5{FoKo#6l6S{XeX4iIoCr
z6pUALXNaDtc%WFQmv}3Zat)j1MjRXFTGO;tRu(6@;HcDG%m(SPN7fv94XDA6gQyJ0
z=dk^(Oa%=DYY>vJ1>jI0>~qQ0@2mawj}SW#()h|l&@I+~4phdGCRGv?gf|g5A(u!T
zAD%!Z^8|1Z0JHY=h|c0)c3R-B5ut&<zrW(sfw*C`WU~;_)gbi`;=jwAeqTUvLC~o|
z4PcDIlzI2h*iAsY<Avu=oxXuPN~Yv<Fwf_s9gVHMr|>1iG3!K|n5f)_YJ`-<7@OhA
z!LU)!QJr^R-SFH7AWL}yQSyyYZ`)T|S+N>|SHY6E8><#*esn}Txdu0}wr(3ZsmXV&
zuCCG$e2i!ME9DFM03;Cs!Cqlrc0!VhS>Z_}LLWc|VH4>2`f^WKtJ6+$bJvYrB<m+4
z&KmwrmW0bM!3#_K)i1TYVl59i5tzfVo;ohl`FxZc(@r-ONskz)z+!@)TXK=2YI>7M
zr6Ju&Z>UDuN1R@rpWeYY5Wu>R`kfDn{sl?P|AAL=MKl39N;jbcU^q|+XEh!`Y9#oU
zE}&W^Jb1!#0cuKEQ!>r|)?~Aa5?zgG*ih8@_oj=HN*r%7LdrD<W3jkGABcvKpr0T6
z{THC$XkPbqb$3^UnP%@I^aTJd|9<d**n1YrfLqV?oP=&1;X5bA1Hg^l06ObXVx|hY
zLY7w9hxJ80{7FVuq09O%U$)^{>%XWag%)goRS+#tp+uxEka;g$tr~P&1Z_%{A`YIU
zJ4QW2Lxy4M;zG2W{c!y8hkt(E`x2s2=0WY&xZ_6T6Ch>|f!^g{_fZYxsZq%LO0Yi~
zmt^|SL4yz&#+f;+qfWP0^RCjtzl}UWZOD;&<{DHuPeM{Ti?l~W>>0$|3xgS!=8&Wa
zq*63(C6;We8~E^h0-dO6Fz$2AZgaK#*1P}x@G~f<zk7Ecf}nx?4*g)@ayp*2(-=w0
z0SePuWS>mbyGyQ(d|+_VtQ;q@Z`=cIoH+cV$CYNp+5iQp;I}^qyo#%WK1*i2B?iyu
z(Gse1A4tE~e8QVee!xq%4hH~g7Yca(9F*VU&QsC7GRTM_7<#C|<KU+9E2ag&n>7M{
zz7Vu^f9^NJYa2j%55uO}n)o?mGn-*%W}}qN3YJty7CD}7mN85no%rLAcOwOmnDk_$
zYQ94HvSmDqg%xH5>tGDi)gnNq?-JtWGfHWHZgZ~h=x$G+*)OQi8H^Yqf91S>&I0BR
zM8`HeRXjb`?g{T!4xb#wIpKvh*9U7hO0%d9$<rY18G*cPHqjQ=pcLoYT<Iw1rt-qz
z<T&OX3#T?qL<+?Tr~qcFXA_>g{J;FtXmNi~)(a|-qZu8~(GNPA*%Qi~e9z9;hCWbI
zv00MYWuh7*mQatS7jL}q5%&K6W&1Mb0F8lkx{a5Xln}L>0MyiJRo~$AM*#p#s*z3v
zX5{h8qsmC0gW1*W!FVK*(H~B$*v_Zzpn9!S>rhqzz-N+4IJUGZ{evyTUU{duElryE
zXhDfzgFmeQ`r7Y5h~ox;vxZ=5Vc(SvS4`=15xk5P_GnCu^47@^Eu?Va<b(q_PDC(c
z9`r=*i!F8WO;(7#tvx4P5I=r|dH9S^KcvFMwHe!`$gOOfY=e>y%I17x<&BR#4n$Gt
zvO3xa*O3?ZjnNzlU>(Law&!|#I_x|>JYE7gaT|8PlS@GLXI90sg%b3?+I!$BIDh9a
z!iz|Knf8jPL()`GThu2V!TbcTCoq<43a@akQXs&8Cf#cq>ncxy>q%+=S`98#!nqV!
z-0PsyiQ*v>(S?mbTkDb<1tCrj&guZNRoCA1Y0u%qR<?%&irkkn0oFd#pvYkP`x(Ru
z!Nz`<Ea<=500@{ixknWev#_2L*A~m(${F$6;OOXdj2Nfl`f*G6;*6zXay)l=3fiId
znd&3dK^!IKjv5^tI6<8bEq&L@RFBy>b#864%PC=*TaS9({)l3Eu2$yFCMjNZMP#s6
z7^-leNr19Me(nyk4SpSJ-&dF_$)$)`eo|e>VP_6?h2%{b{>Fny>ut7p9*TM;E=}<Q
z3c2VP21iCxT%(b%gsFW1IE0rVq)>%qEs{clsD0nPd-vicT938iP1PHrG7W3;L<qbN
zctm(f664_}u4pYe8lQ$mcPK_7e3-GTfL7pXFhyA)KL`6VHGuRNU4CvS5uwJu4t`7u
zd0yM-w=lbV=H$uBBq@5nMRNEsQkZZHd6HBI*U4RqME~JNiI#G0KKck^p0;udIdCTI
z&@-ze!i~YkQ8?-`EDLpT@jAa^M@&ku7ZDnL7lzlt$4OCOO+J%9Fd#3Ql_O4K8<1^(
znghNN6f;80?$^F~sw%uf?-XezCm|-50uG@&-VF;7w6NT|C$)nM2`;|CM$J2Lb`@aU
zb;KKP8DC(G-4o6mNZoE$<M3&?p>fIr#yPLhwXkN4qn71qAo*RhG#ph<WO;Q`FBe`J
z=G5=^I=45I6=P$FpC)sCP_3G<lSLH@ocrOxj*g-}tTW3%HZ&~Eq{&tfP!1$MANn_<
zADqU)6ow$lpLV*QAm;U7)33h@s`UJdvVX9@|F%#ZNh1#)|0jVrVbb>%0_vd5p{-=-
zoY`|&p<<OE&!SY~phe-J)@vpVo~MAoNy9wN?L`;eeDg$@LDjd{+&GAAQgFvI2{((Q
zdfQycqElu*i(G#C2E~eT0VuxPq?Doz{ohKKu}x-+D%B&N^J?`TI3q7mz%n)DRxzU!
z#V@f|ix{+G3@v-N$-KgF@qXD~Gl?(h@!RMgAT<g(%aPI9N@M|`cPJS|X1zL4XN2QE
z2BL%c(dhE(Y8ndq2<bp<a=C6(4r#o9;3o2)+bYkE*Kdz1Z#X?mBrV$=wdEua6cmwt
zYDS8X9JS(R3e5>F%AifvaqvL8KO-Q}9Pb8Gr$2?QyFc~2foe)cg%nX=D7K>Eie8`%
z*jOlz&&6u&GkHbCh`GxcIs%T(vz)<`zKPySe!L5qP}*$v%gJzW*Zzm60*36Ng01&V
zAFc30qgM<3CsjgRQZfx$>J`Q>a+jgm))kCH>BE&6LiUm1rU>XxVSTUZj4h^6|HLei
zw&P92el?Uz*>3`x>-0P9Q0SNo!7KX7v2@Qb&&G~pFthAs+Wls$jLF8=TW*Pxp=#pK
zCTbA@8qxu_n>c2scMIpC>duI)5p2ZS8^EkpN6){y@H8BSnG$d^)^6oetekM&SRP(5
zQFHkX<Fw7U+PC5*L5KbX+y&90y)zPo!FP(dVk-Z|8U1NQBGeME+-)vS^&MnGAA>>K
zW(@rKMEpf!GHMdimiUkJUUP^`z{JuO6o!N706P??!EX|J4yo2hdk;?DaSZ0!p{x&f
zmBpPV(0xz5aMXhgV(@bJz8$-J1LN(9z&XfQwb%7^h=)}(?k5ZnUc2>SAnyaJ8D79x
zfy$290TBd%h_$q$=sB9pS`bByCZD-aqkn?|-vB(gn~Npe_uycKLv>caL`{c1{F9~#
zw+k206~&9wO*cCC?7&|F;_lZGl!Oc6#9B4B6#o*u#Rs&i!i5%|(~a2p1cMHKYCRl(
zZaLA{q@<)?LQEY=&o?RKtS6OTQGKMnDZtDC8GQ+i!lSRyPoYXb*wJukwpCrFbIqk2
zhA|`S2JMY;%a1^Y6F{VAlT<N;p9cjs-*6xN2hkSRCvOYUg2N?wE<Oc7JlkHPyrZ7l
zHQ*fp2KC3`mN#u|7J$n|1kO<=6UqTO(rSQ+9jU+YA|y!5z*7UffTc(y;S|X3fl@~p
zq7X|h^xRL;>K?4K%Lakry4Y#Tlj$5Azj6pzf+C*TyoteqBH7~?K`$gCV_w6dDYYVg
z32^YNH!hw<C`NE2fNK28K6wzrCInWt4i^ZB`5JO<Hp_dR#=Cq#$bHe$6@5m4OwYS$
zP#uJ=paAa|X(Rseq)#VER+PxGV}GJg<WAKE2WyPWuKDP=F`$J0N`E5VwioHh`R{sE
z5>69yW5Cz3u~hLnL&FuyIoG|15A*Dno_|7^tqW6e)jog!{Sx06eP~AbB0?;w=<Ca}
zen698a2#{lVF#xHJ9{othyddnP3%FCC8GU{BJJ08_<hawwGYX~h*(AHOJ=34GzYvG
z%k5wp!&iR^=8idkzo38sDWrg@Qv`Z^M0^$c-U_WSgeBOhO6)(h#4ToGmVqlr!vUOc
z4^6a><;2@;M#$Eq6NAEL1PgNa-dWb+S_G3N*!ER@v*x6u!8X9s<OMV>SCF({B1JbB
z*_)Cfg5ITd{1m8H8T@G<pk_cxssNbkL3g=VD?SfzlHa8tXju+eo4#8GcZ~qyXNX$3
z>X$+`Kwch$AT#On366y(H|NdNv5Os9Wbp>6ZLsk&Ra7nu_!>dn@z)9HJ*M5%fZmaS
zQU)K&U{}K1Wl2jpPu;ZG)p<t8sP$qGd+!_D1c(y4G6JMrCu!h`YH_*KyF{_Zh|Fw~
z2Mm;#mgH8l*kwTusedLP(OH?u?j`L9SP!Dmh48f}l}g?pX}IG&7b7%QL**B#VQHa@
z*hRU!`^}EpqCI{L+tX){)`jR{7(@g5*ptV`(DbWe4!l({>gm%LGRp&dON)yDPS!Ut
zM3o01_>E7!wL<jj!t|1U7Xt4BNZEtk%5Tsh2UGU7>6Kz+R1HkND)|q(Drz2TP3l4w
zv#!Z6NJCk@*`nsHm#sd%em9a{3nxoexR7X}zq@wq_Ux`g6me+bGA>*I_IkBGPPT6(
z<`XG70642*7{;yIZ0hRg{_lq{MUe^`W^A$rjZ{9w3O?G_z<3s9WM?<01%!kX4BkoJ
zwdsCim?QIVKCG+j=Ry*5MJ?LIH)p{Tvgyf!5BBk}j2Ilm!(i?+w6t!D<QIF~uG9(7
zmQ=(Z&G4mn+mGA{137QmlujQCL)Dr>o_mgHUI@I2sODG=GL(&Ua>c$6ZO|3_a^MqD
z+@hyWRS(3}=;Ydm%Y)?<7H1=dHo{AK(A5rk0TkxfU6k7`eSKtF$7~Kp;CI*+(GF$E
z7+82_p1K-+<HDs|rOh(x<Q*-H<syVc&2fuWcmW&~YGEE^{2Mu1gFIP)U4q}s%=3A6
zHb{hcGyk_%ei8_s;Ip45FojKIMQ#+06xH)FVzz(IiTT+jO<?|8-}&aUw=*hup~v13
zA25g8)PcpE4@|}v9RuhoZ~9KgekfoC-~_pB$rs)aDBWM~yAQDM^!DxBLBc=B^i~@1
zx%hAq)Z9#fUb5IhbX>#j;ux1fa3vKj10&0GvK1*oFduY43I-}!^XPT;Qlz6!48bU4
z&^sklj57(X^n8aNr45b@>ETWQqas}oL~HX-*Kl0?(zm%$*kn&ZFw3Z|ZnbnF2Lw!&
zD7F5EY-3pGH^OVqio!V)m){%@ICAh=BvTfcS<U+Xkiuu?$w0-s<F%FQd2o4nb94JI
z%FCNKjPV{D1JRLxrX2rQZ*HASx%NLxm)(Bb&5|ZSkDd8l4Irx~*)C99SJ#?YUth05
zCbDK{W}4{3Fb>a74JhdwMj?)@uQ~dN2`ua@U^B09Ep*rDuda{U4D0ytU!PmxOEXH@
zRvzanSl`KJl+xv2WP4a$kaFMGS4U4DqoG2yGg(F{^Gq9sA4N~I&NJnO2U+^V^gM+`
z&bI2H<jc?IaIHoM#c#j=r9LllInwRj!N|FWO<bihv=v&2O>Pv}2&?}4&p#uREwBO$
z?xV*?b7CH^PZP)EW4`!Y82S@)_D!;pTy41u5bbDb95is6!0$xelz=>#qcb%J;#CW<
z^^W8Tm3K>jWKEN{PZ;qC#m38U;wEwilZiHQ2lQRuS*}M3TO@(|O9#%72x2Rhs5hQ)
zpco+6VEPpw48J%`19=0zpvQ)xyS_~vipw+tihdSDrP@~FIyWDCQy27=Bg&KdlARke
z!_pf6^#T?bqCYn9TnN>0jfmvbO_S~hbx7_frzDt`^{+=ZUoS5iT)2)Dr>xH=B&_aa
z*i`{Z8AqcYT89$7Kw5jUM(7E5eVX#+%OC4#&B@#MA26x)Woq9O@<`yz`Cn=qbO{3%
z{O+f1l3Yp9k5Y(;s|P5PeH<d{_%<^_&cJ!VB8e|`m$4Vj_dg~A87zFmXrYWj1Fb-;
zA?Y=WQX<em&GQF{Q<%VQgUIBznkRa$Gz{2h(CA<r^h({@Qdj7!_Cf%HO4q8PmJv}k
z0T|}C3T9nS(8_ijr7=phaYVefr>E!6x2K#*yr)6or9#3F+R@%#P^{oTeO0)&39kMU
z3&RQ+XC7^XZO-7}AVnYpsRySm+{9JqeCM`s5W=Y_aH4n`ufOe+-CR})UX_rLNG1I(
zszHr%FVA2zDCshte|jTC(170K1%MpdkpJ6YjG-*kC5AEpN*IeF-LrSNHfP&)Z6Y$X
z5L<+oHi5ZJEbsZmjj@<%VX}j81=Epj0E8GB{AT$O*~j1*6S<6fQ{BD~)nD!ctAs<!
z&;}9SNE`ew3LWoQsV{16HPjax(EF+4Hn&LaLPv#(ZHY9RI|!?Qp$;^+t4D7B5xV$%
z0zYT0t-AH=J|47z(i+eRm3qLA?NJg>?X6HLu*UiNI_i)2^L+TUJcr>d;i5*2Q9cjv
z{dCVXHlDGzfsgN-_JvjeKo+(K+iN$aZ8|bzE<PTH&D?c#quDhD&$E}_-F)2!mf}@-
zZ@1A_R4o*d%iN9V8bUEQ#5;{U6RBkIBT$WX<)`ry0|}j6lSei42yy1H>vW@Fwx#7x
zS3hGd@AtAn9u!r%XO7tabpXLz6*3J5I;`z&J_hnxey#-2X||XhL0TeQB9hUtuCT_q
z3$fKg|E<t|m#5xf^{>IZ{2I_z6Z%)317WCx>Ix#bG*Vxpno5HcqUGY#{6!a!yK3TS
z_%oG>A}Sw}>jex9rT(;);h6{bD&9#tFYZ~HKx8;@?wT?eTdv8!4(txv&UspLdhO=f
zvnR^`9_~ipkw(Lut~%=I=I&t}BZlE}gh%rvzRAg(7H56UUDs({Eao_%g1)IS?cFS?
zubxk=`D3~euR1qXw=lDFmP`uerO!%`0`2>seQzb^nHWPC%iiG$E{tIgaDSS&eLyD+
z6W$g5sc2S~q<@y>#=r)PCkm?}!q+p1<`8mD&{K_pGF*gQpx*@sROh;Hz85B>OYKNE
z=)Wila=HsQZOF-J+6R1N&C!088l`>i(AAQRoWqVfPNUGeCnu$Wv}nDb%3fKhRyJj!
zYxLLeA}!0$X$KrIa#qUb>$W)YDx%=c-#KJZ%{5aBb`l6PJ8H4xM-}_3_^DO^Hr$p!
z?Fq?fE93jFO?GyeH^u*3)2Hye+HTpk*CyIYf9K5c0W?WPhGFh8{*I21jmH8cyU7wB
zVCIfre^tX?D|z}8SO20YR)uS}2DCP674~untsaTSP=_!6Bj5S$`OlQ?=ni9)Pm}}w
zDg8;t{x98awAb^zi2!KECq)r=5$%h%nq{mVxP-dRqR_+tYiWTdh9~vUen8jElUZIW
zdkIj3TP_1Bu9<iL!^kG0R2T<+=+L;%&GVfg>P6fFkw-$?Tq9VY)P49s3M^4-#;xJk
zuMg}a<^{pSJ*0uLd3ypSr6*DbhmWW7w%rkIBShVxe`+3B*ceew6+O<5aD-ObGABV4
ziO3PFWlBB4grZV>B(#*h7&zE3fI2F-JZ>4P%>(H6#e#q8RN(X}o|@xJ;++^IRrb&Z
zVctu3_3`f5HFe|vYU!}{zsZ`S`gLU{hCzN|-DNUEl~qwl)~lx(R6rpWS+d1U;ay*B
zNo7!hwk_x6`Oam!)3MukuX3la<>ozjK=oM8X>v%EzLEHO`X-uqZu3GC&~3_5SFk8+
z_2PYb$C4jgJgEE4lLZ6Bot}_7%`avT$RD#H*eTy8s!@w1p0P%?cRT+MTa)2(l>K&k
zbFTtO>^nkwwhR<eY>FN!ng#m)e>YsgtkLb-S(aC?|2jaZ&+aThf26~3`|Au~Hgk|U
z9NvO?yiPiMT!<O-e(YVN4<i(F#8tfzmMnU8P8Fe#l*^;_A_yx9AOK%<+T69XJClFl
zbb5?cGSaGvsZ<+L_11gvlp4Mi3t4>YzrDt1Ux*&G*Y#WeTbtd!7kJs7-hFq{8Gq+L
zcrPgVv$Mu{q57-lzJ`1Tqm;0=a_+zv%v!1<Mk)UukrO@;qK7+4*vZmc>$s!i^>*5K
z`9%t}W2;q1*;shznbI%yy;U`vX4`kj-lF;CL%mrZvEJ)Q96!nqYWA&Lid7BJ11xy?
zE9VQ=s;fF}7-7IUSMgr?wW_}RL_^e*b?bA=Nza28WXo^3Gtfy5u*pphxQ=5G@bwn8
zwK6?gCK`b(#thrG{vEYsTzjafx!I`Ji1+j=(uh3$Z`>cZ;FNtLrzN3703l>C1>Q+7
zAK|XllT@5Uk2@<W$0gPm^n3y%_F7r>w&Z?a(b&P;D7_SYUL3?P$0^ov2s_>lkDv<X
zpu1Mpax(d?>SSt4wsbpY^vkP?SiXj?F@%*^-r&dvo57om2y&^L(06ho#xsZWyUpKu
zrJbA9=>V*9ADpv$HbDX9y$Q(SCaMo-9UYygq3VaG^Ti2ggz5azjXpF(m8)H=_IY-L
zr;li0)a2-vD;LgRSHMkeS|@v|Z6VuA;l(=9kL#Y8lqSN3CI^D>Ap7;)o?75b$1P>k
zn}+;jDTlBPcpMDlP}gvP3&<7UM3r;Xo=!SV*e5<AWnw$NYz+asXFbSn9y09M=14|A
zi4&)U(?p8yFn9qd!=3`fvzRhXPbUVMA=l(vuyytZ&fXvBblk>^9Y&8KNB@q%ny?CL
zusW#XbQ?;M6F^Htp~QU%oUSGh6=Ecvd)Ycd!RX9%a|0#^NVF!>bIjgED$y+N+{6ro
zF@P#NSiZn=K=;M$M~Oao@*fzrad&-+)4mkNee8#xpk(I|Jeh8Vdnk4R+2k=VEWr%3
z+h$8y!4VLeu1h`6^+^^mED$nlzOdwN*mWWa_9*AF9$6}{cJ;<3puD)YRCxdXeOQv!
z#JiDEU(%GUdV_<7pq}!CYy{1+iYb3&qN2b_W;B|tE;g0L0?ENY+jMI^gi}4BY*#xm
zd#3>Uilmprd-@<E#C7ybg_$rAosHXEyF{jS>(|S^_wOv7jz~n;atsJ}hx_5u(ozaw
zd^g7re9&FtW-tMBk~PRB8?djFs8;Hf2Oa{t$3M}oTRUKx`XNHRB;LPfF1@)nXNUQ7
z_y#DwSQK42*s-F&Ww#&ZXRn}nt50GRyZd~~2BVvB2Iv>Svf87h=ZWH-E<8&FG*A$0
z-@K|_$&yN`?PoCu@2M}@2)l>)wTwy@^QM3oe#Ff|C)Rc7psmAN-^tH_W)&d48j=Zk
z!5R;S?QJZ?6T7y_<_5S%4&ihVql8WeeSQ7ev9zqQjdI8Fur(BU7q#=R{js?$LI&62
zWc4x{`Y9UJ^iY5HE&ZOo`DPdgF=cfXl_&GC&9qk=vKLwALESv1^z0};j=7D3QVH|-
zuU3h&S~!xNm@MgTEY0zsEez9i|J2sj)^C(^9oKLmF<f8ZhBvJEz<$|SHf-X(gyZTT
z>=^>_i~kY|UTZWX3u%_{Q+O!PBc>sJ-JuZ&My-}Kuw93cxcED-@-If{qgMMdWYCA6
zhzLNi1WMD!v)48si*81|;3pOaJLemAEQy*>^Cjx}%>MJ-@_7grAAV$$5;?iKr5{-#
zgLT9k1}!`a!&6h49p_}c*4|)OiY^{~8$PwC57YnQ_{Zp_0sU@SJWW_&)zfz=@KCmk
zth;w0+1a;BZ){Y-k-2TFhbrq7s=lW6a!`*6qd<I5V_4!N&*D_wK$H2ijuf?uDGG57
z0W-nWjs!siCS_H-N=DB2?lp3V0)SK(DXeJUM?@bI>#~bek}`hgo15Pyf4y=!VcY-y
z;cfokgX@3fZCBWT;iunUf4+Eu`@qHDer8ktS+c)W<od+T{lEYA)BePRe)GppYq<UT
z<bQwUoh_IBp514eBN)L`!|`RoLu%H=I5`%JUe==pO=X6YqLhYlP#dkW7B@Ne8iXUc
z%DaF0MHSLkr4CY=O|D~)-bhAauRxC`Qj1dKI$EeG_N>2e<6<BWK4b^yVtH5~H^=;D
ztw%+*61cz767-$JZ@j#)*yoQPDDaRP+yg}y#C$UW0$xSgqKPU@+WJjCwU>F`H?ZU`
zglCqbdU`$o%wGUC9ixvjaMieY1D6<!6<cfCA@M2DKwot@3#43xZe^{OKy?SCz^{Pe
zPtAD}tqC4{cE07%cL~wVUnAHIC#yQjWF9kCe1x^cp)71KYPakwk6yIZfHn_|-4m+_
z1Xl6Nc7m2oP8J`B@MzwdN3?$&$wbmoq1+tumK7`(F6k<0n})b~!2$B$$c7_WOgL)#
z3=K0%e5zh7imNutc7FsCPF~&yhtl+!1x%(Zv(aElI_h8~l2$xLRYc-+6mJZckn{Sn
z6%A+#0L29?!Q^MU#cRJ^n@d~LkUH0sO<xRk?8JZ;VJQnbcv9}j<tHC4coE1Z+Q8XB
z>vH{1>yBJ~j7Tf~4N+b;=`xC4s<{c)=X&0Tr?3s|)}|l0Gndy6!4G}uTCV<5*1t}T
zbu-^!yt07HHdO0nUAmyDyZZ|k3`<D=JTz`0haN3vk3{FAhn{*_wTE^qYO(Lj<wExY
zF;ZZM$;@5YyF~YYsoO2O5~MA4=<wkfu2nWclYkbV=Ml-+qtK9MrCTWiZq7j=BuE=c
z4`!9Os8eV-9WgOiSbMF+#TNK{`=FHbSHraXBlw>`x7Nt!qFZG;EYUhxLVhZ=;7N3z
zU+m`ed2N|?1RvYr-Exl$OqrFIWzNQ%<fNpo9SzU`w?Uarr<kK=En8pIy`uHE52)@N
zB72w_U~%Il<>(EoM&Jx;1jr^^ei6l1jgPp_Xd8qA+C*gKf}01>${OZ8lc{}#o5R1H
z7U*?tD*6(LExWZNY;5BpwjB1=uhV1RXO;!}`86ZbovfLDL^bLB%dS>x;uYPxSQ-6W
zZCIiy3|UsdoJck3+-bW-oOhBN&Setu0D_K_JKZGx%%rCFu^cB}9Q;DFlSF^t1sMkn
zY0Rk7xphZ_^5E_bIbWlL*ZC*0v6<Myu1huogZF=lb6IqgcTTw^eZqL_>oK95m@0_g
z8Tt62?+YU704|wEmb&ojQQzDQG`d3?8q%_n#0b6hoBYhSg)pT~@txS3f$5$WS8THx
z3%(zbV`AHwvHj)jHDn)+mk&zK9g$O5MXe`3y#11zU}4w_CD^5~hj>)uRctn`dsU=f
zU|gna>T*ml$}zgWh`mQli`?Z*L`MtI021&80=*eEX<|M!9jIORG&HM`IIxKeAG-uV
z+b=6?BThJtNcrt~_Bvfn&!iXaH;{#T35E!dbM$f{SkFTAH^6KOyH&>-hK9)c(`kE7
zDx?5oTY=m}sc^EvbRT9MNQ2Oj1qB+A$pJh4_UU?vB$Nu(v9vUK`_DJAg)M+qcMhCt
zJ7hE0;m$%^m@7tz8GCpt_Tzv5?j#eRh3rp`CjC^`xjyOKmT95|EKFM83D02+_2cBi
z&SPDwW^xP6Rw-H^eIv`&M4wxKR%~Tj<Fmi7QIqTRht>|GsN?~CDfAwP0WFT9Hw(&q
z<kYBZU33wZke#vo5mXe!JPp5g9CvN8^FjSofxEt^SX1zvBov8A7)>}$j=<0?`6<Zm
zUter3o{TNa)S?95_ut1E4?&N3<(PQByGg85Mr8%c>K5eSTxlT}SSyvTFVb20Z~s$G
zU*sqtqJPjlH)TFTYnuwhvL(Y!1Nm6bNhUmVZ5KpN#U5MQ+nlS;6cCaxz$)D<4zhFW
z<=^zrKV@68duXBh>?0@Jhf{U2@pdOIezOrhZi3YYJ*iu+XWr}_%};yrVMi3qWSI_$
z5$gHrOD+{xWhj6Vu{0oF#4Jq)2Gz)0BA4;f{?zV`^rcC9GG+@epigfD4l=4))r!9k
z(T2z8KAdHQMzC(i4Y7uOxBanY?;B@xU+kQ8%A0u)ecwdRmwA92?T8uDT_!_~C%Dok
zSe(MLy9&5)FsK_V<I2NP8tEq}Y?<xmeGsQ-*HL4LV%)vN+L{!=$LWcNEM!g-qWF0H
z{it2CXY|CGfgQG>TD>W<OP6WfbU4yz;eCFVFXMbDvVjSW#ciyW6KSl`E(OG3Bbk=d
zfz<_k6l1J5@*NoAksTfQ1z2_n9x<GePF2?~V8B-+Lq~W4^w8o)dS+qoE<X2+)7m*N
zTNVU3IcrXd?#(et<6;y&-933dTAUYPPBa<clt{NX1TU*H0%b!x`ooiYH4bOZTTcT>
zSAaBJ6@_dYDkZc&nvgaZ+&5^fYx&+~9bMwKVD1S@S?A5&7jWyfkWABPkFqoI@MNdd
zbMOOez{qlahNtmR<M*yHYK-oTBj!~w84$r^(GRM10b#Yc=dDL5U%y3e;DM>C4(<DG
zb+uYE3#g&6ou0d}HmnmI#++WYf8B01DC$RWswKcB22P$NrtM&R%_X>tx!w!i6E7}b
zk1cqiFtskz;;X=x%9Zy*CxJ<Ln9}`d=ES(zW0C7i<jUOomwNtPa6ZqEr$Oflu!;?G
zpSg<|1Mw>%Qxu!?f8UMJFos&9*>kC9_?ZZZ&3X{|(9IDD2J--^xa<UOXYVVlsds8b
zn`D7DV+B_#57ICC;f%y?v!y)stlI7=SELwH=PfSgQklPg!QCfdd1KgqH}vML7_6+O
z=d&IbO#Z}yk!i#LA<tj9MQ^id#pL8x2?s5#QgNwz&S;q|^RqYb{ol11kP?x&C-qd_
zgvsD_ZSAN+yHZpc`yziiDFcMO*1lUFHb%LTi8;t0B@EKWKD7PY*<H>N_bgp{V@6}w
zSw>_$QYba!z~sY2rO=w=vfaUZCWGU1r=h~3JQ?S-x!*Aprd^%d&8-FTF72&1Tw8Qr
z-}>tkJume^mzmgyNZX@=&TK{s;k;Iz=V@o={0di{?xz_(TC>8mZQz&uFkZb>{G<`k
zI0{Nit&qxCM*gZB1XjaDVqdaG!K?I6VZCe#%_3p$@MIF=g_JshL+0BLUrkI(GDop>
zb$@F*bbt$(-FXu|JjtJ@!%=>?0^oD9(HrdRwd$d20Ortv^Ci2>=bvcP-w9~L<^Ekm
zvqFw}u_3Dr()jUtH!#uqHgiCiuQKJM!Tq<NVcU#jL#9apX15Irrpdhg0%mqbAc<2)
z2?f|)SQ_)k1(`k(gPI-~21$iwnL0$`9b_P5vbU6;Vdy*gzCCEV@wh>s>NislE61LO
zdOMU0LiBvqAbs~qVOo_7*D_}rFXH~7N56$V>1&QBX_t8vp{#q3HvSI8A*^tKqa=qS
z=1Q`wX8Sy2I2+#ArIuC6q7u&}eCo#mUf5w!<r3r=eJNo0;21-+s7VUjJ}F4|oz0`r
zcRW$aq1ez1>Jbnv<j)T%bK7UUuR!mb1-?JjO2`kEKk)s(jGhK|cs<vXlnw$hYvZ4d
zk+kFD=!G5-a~MS7J&JqplqOdYA~_lBG^SBzxm1GF3NVS{{z-{C`=l1bj#6Di_@X?L
z&CIQzhR_RG_e42d@(Qq+rIp_7XWhqPT+5L8#}osTcKB{4rUi=aYzA~^q`hRtTo7qm
z>tiIfa2jjd#o~g3;_+GI0KJ>AB0&g<-D82(W#Vl&51LJsJ`V@MP(A~F<gIHM5NlC4
zCAkUg>Y=O*fnj}eN-ZexTh~Nzx@&PHqAf$W<Ex|RyY$<(j(0wG3qak5*TqM#kAN0u
z6z|pE-Ca{|xKn(4y0uA><ixeB9qi6!&QpxPj_{`WHmwoDXe$GwDoRNR;RRd1$RiY3
zNJC!LnMAgb0JE?yy4pSwt&NkZBhub|w}L5beE0xlvl@V|TS4EWOj03q0Bz6<a?J3E
zyBWUGv{4p?cDaiEAx11LEGa1V#u2ctp%PNg(g2vz24zJX7TD<0(P0b~a`!K5;HI_w
z7EBfe;C+>J4T0E;))z%2)Dn~#*xzsAt`Nrk3PX(w1Yg45TtOaeb_AW$!0bMS!Ev0c
zRv;L{h`bA<E!8R~%l6R7N?+gvjUrhW9_|5E7LCFr3Bda{hK;^C{2?3X?q`L2brp+_
z7*tnQ4t`>>sx*H#*B!QCqGVEZ&;YQ81y~~T5?kgm<ZuV;Y}u`Xzjt|V<1EWWZX**^
zI8_=b$CKhZ&|6zcY93GeX^j@sH*6tXmB7@-YL(J{N3Qh0aQ;|h&-Fj5<`^BfHn2IX
ziGC>eJHVHTnWAc++rEY{Tz1r_Y=&nB=s(rQgV3VW_(~p9GDBT%1Ytwo3zU*0#7F`u
zARlHImfD~A^h;78)wDpUAhjitc$af<cf!yLm27B6G!QGlhY!`zk&(e}{O07ozrZ(2
z$P9Pi`4ONG?ZwYagG{lYFfOz_D{jN@J`u8^4bvrCM5B{bP*cEo%hugpe0}3U>nGng
zLw+S?;aGQb^0VT$O<u7r^lnePv*WWS6<=Mt9>nUZge@c31%Nh60){dQ-*iG6V3sKl
z>MI0uo4~98`WlXeYU!)tV57X84$r|(x9(AsKeKOF!t(a?!bp=G!luL6>7Y(X#gZ?v
zyS~rxTwam#^YbH@;rJ(OIByXD<-*0Maj>Mjf@;1&M7r?)3o`k&v5@60?)#|J96-{_
zERC<HS06$l<>5|OK24fS3+D%2&DB6W6wudFy^CNCre~-d*WxeyN=`c~+&5{<CUO^v
zOwJ)noKn_>jgq4!s7G{6rW+o^u5rHGzA?h7whTN8B6$+3EHULm317rv325ArMMvzs
zfVCR2oYO*ZF-{m*rV%zBmd+6-EEs?ap7bl}T)^rWu|6jINCWV0U7VTEfJWhZ47Z8o
zn*5Z_YEpc(GOV}iVy3`3mzMJ<%d0z3bzt>n4u?u?I#?Idc`ATa1@IOR1X(gi6%4ix
z5ZFS@V^LO%r!C0buloH7Tpx4HY>=rF%Bev#3XvNe&t?`}Lxcm}C$g*OQRGweAFmyQ
zC>@lKQ-t)1!mr~H-4L$fBV?45LPRp#xL9mmozG&>k*?j6)~BCl6lQ?9bD30c_;(K5
zm1%oS`n8$HI$k!|Qs#`Jjjp}ROfHeBZGV?)Mw4&koWQbhRWUPN^%dJExt_NN9evga
zM}KwcEK03m$7IVc>0CZ<6x0YOZOA+cpdl~eJ|J&lb=SYZ>@rd1N%Sn;K$%5u0_GRL
zPoA#?kKX!9&Iw9XA<Qx7W5zZ|&KU3_G=cnui-Y5C8bf_PX~%__uVdQ4PMk=Ks2F8{
z@W+l)4nbZdCN)5*$|R2h(QSFgZ`9qZkNz0(!9-&AE$S|Nk6-?ftr7tfct*eEP0KH@
zwLtEt;9IiW6d$dsxYpZEJDGXj@Xut+&#VMxJq-Onnu6T?HB?B^`k*q~2JIr^_Sl<>
z?N_<)G<P3YP}rvTCP}53ly?NeRtlLA&PmAev?{Kwl6b&_D}yye94tvpEZSlE(fa(n
z&!G-`=Q}Q(&J79Byolx08|#%lC*}^cQ2<|!3!abJM6ZIY*C<Hl0zf^qSdcC-NjCNP
z8r~3LMq~P)L#!+p+2^4S+=uo>AL#jn?VvO)Nd<z9yqJM%N$~OTD5C>ua8l2~4}>Lz
zgF`IM5=k1K%Wp=V_U_rEw4|jfc=^U*xtfKRvXcrSs)YKN%*G7MHPYRS+kkjYQc*ML
zu|B^S;U0q{EuY@T)NR2`%3bkNoXjNAPhZl()9l!}(>~giV7CYptx)nkE(i|`O9BKd
zraTUMVsv&JXF;CZ$M>OiwpWn?Z|;(f5Z)^1CCd<dMlJpyf}|rAnq#aJrSLssZAohI
zpqD5_$0JN*XLOS%h*inVCT@2YLdmC<jHMH0i{#k%@-oikr(NwVI&83g+cvlX>R-b^
zL3EM<oXzo29FQhw7H|OOvCWol=b7fYB3M@7W&mih17z;lq6kxkD#ii9xRr87_T8xT
zEqh>gVw5_bX`iOFrJJnfz<q3N&&hkSaimLibNBy6{=Y?9|Ne{mmvk#rPJ;E5O@cs>
zDfO=lFs(?s63pS+y=;r_?&@yKkTV(z051ge%8pk0Oyy}I-AcqHJM|mZZ1A7H{G(9E
zc=0~$hXT#63c(##mjtpSWM8^J1VOJE$ppYRW5eNi>A8a6e>iKJI?k*6D+k#$jzKhP
zsjNXqnt%#c?Y2V5b{!*1g|yHk&nhbrm7#D-n)Cb;>ps<lf+!Qp8wus56YOD`@3&xv
z!d~14z+>3(TO3mE@hwHJhwLyA_{cKhkIPy?rl@W@65bR%|NmW!te-!#=0b!<r<v^f
zKnvkCTG{$+`C$$Cz)!s!VE+x>wY_(ljVi!}A=Yh-^AWpr7yl$Oh7PJkgW6zme*;7S
zs~)8%0|IOh4+AH(E?YLK-t=By2nR$ns#Y|+HFuW<lyO3yf3|92dht1tqo1Oi(+e_M
z&Spd$h9#!#+Bct3D(J{&q)p{vxHt8V&$_0%%=BY<P6W*rLI_-owC=86Sm`)|>9D}6
zrX(7RF9|Yl)}SDMq`R66*QAM@V27(P@V1o5Lh|L6%*GYhMAPh06llnj6!CBR(3NWd
zT$G22j&5E|A0q`0(#}x(^*((^JFHo)Dq!Q$xU$d9#K2?b-%E3ZJ6q|&xyl;lpC4ez
zJ?JcW=dxDjQy{tmF>&hEOLcXu{yT3o*m^(21_uP;X`xAo{1;r@vr1<DT5N8iC3S33
z7LDMI6EBd0+<$WY`|*Nqr2fH2*>%9FV_3uL1JAcxo^bj)j<5NquIm;@d1VJ8VRLev
zS1Mo{v7`(P7=lM~)el2bnd~X<m~n&nsDM=gk_RQWxQy>BQfCaxO3-mnx=3R+*yG#(
zto_GbhQe7lbGjH)>*gSZR0=5jw>E;Nnsr;kBv?fIQ)J2LF?H7rputI^SpYxm3o{j)
zL)nB8_X^YPIpDfdB~+HK=(eZjW|>FlEJGOIZEOf;p#`jO@u+Ai;H%M(4)6PV)8GYD
zsK|A}ycf;BksA8>FWmg7=NnSk#BhgX0n%tADF${x<na;wkyKZhA&VQrKunS}rp6Fr
z1oe`S_NX_qejUn7H)ILyD6?MT{xEYF0TAkgh{@h(lD6@D<YkR6AQoF?5GzfTIyIB)
z%T)c)1!rr(K-S@$<*%uxq4ddE)KZuc2x=@a-B0)U_HO&tj{E7o-kY8+u0Ysx*PmEe
zpH*ZD28sz6u<8se;#7r+LseuJp73JkDdNCm%=XvNnW@3}voF*|?mmiu@-}0TOfXpW
zmUPZ^rO(9-tmQT1_VW0orV-Ia;hCt2xafKnxqTF-_iGC;t%si1Sl^;zs&O5K4f|wc
z$8ZzWWmEHA;i30ROxkMddbl6NVkKROiU8Baix;)MrjNK!@UKJB*ZCvDp~OEsd2avy
z{evBBJAxdBW1J%!#(RnjGa9a=F=@sr%OVv6wq1lBj)6l8#!Dr`z452MjYz*CHsc^$
z<g!W^BQw(fM7V(7rww7~7F7u$OKBKb)JKP2C|_=B)+Q-M+?L=r!hY=ni}NN#(jgr<
zB{C8PlhAwQqM|kw?@7te!E!I!gfN=YB<(){2-~=QC666nLbG5LWnDY!T??vgTH_!Z
z2LWkc;t!+OsE5F$iHc_)Ka4(hwj2*+VWP9ePCLz>pDWkW&_s0a7gU`U`wktNa4E&?
zo+pMQtZhsID3T(oB@a!Jc4n}6fzp*L{8rI56KEN%h2NUg*Y=xvn6da0k%BG_3l@7`
zeW3zT;|=$J_|Ud?3uP%uZYbT`X`uSM1V;P{yY2j2gbYN4x<d2MH(PtGV<gOFJ}QK5
zSZ%1U6`KNdWIhJ>Md+0y=u8RN$7PkS&C+B#uR?Uc|M1~SyI%NcIl|GSsO~pMSgMox
z7KdVJR~E;A8w9MDH2eht+5uvaVX?ePlxGAG0#$P!kb2z=75$M!%d;WCG_&7JjTZo3
zr2vvy!5EySXCcyd8s5xDhZFm1N}qAq>q!XSW<6?C7cJh>USSA;-IjHK5~M+?D41!K
zo!6qrEq-*{Z3nGV@?H#?u%(Kou-!<o{^kwJ!T?8~75e3tJbWK>gk)=*@zSMUdvmSF
z4<39(&&u7wd3mW72GXHMe@3R|R7shXKhi!TkxQQ%y;|(x_2@o+@px-*>=C3fFC|OS
z%=+4ng;aQHoR*U6o>_sKV=g7v#p^e~G@x~gqr(p*^b<oBfRS@G1v!BnL7hMebEMze
zwV@E_zSKO@nVZ}}ne%O1Wd+e;)T+XTHv2KQoo4`*H$#Ia&as``xec7?q~j-lqdz}u
z+q3iMb1(8O_g=BO^Z#i9It4s2M@(IvDpn3B#(V_|WS_HgI2lxin)T?YLIQ!N+b}rY
zGDsdNPHmCK6yR^M4OuL*h4AY0jOsN_7z13-F}^~lnP&WqZn&%CVooAFegh}moLJ=$
zO5v+}^f6aF5*YBtbkzBhwHv@&QgE8baQk!zb5_>SAw~cF9)5y1#qVQ-a4Pa|FXetK
zd<&p}T=j^H2tBR#C-Cxi-~}bU?I>1-yf-|;=dg-41w-*8aL(66&*;a@lS)uLNQ=uT
zU)z45)4#@bu)7ZYUJ7kGFIzVG13*W}U>s=+$WY+7Z+FoRDL}K^f}4j%ElBF1Lfc;8
zxIfSd=!}Hn3@qB?5DdibIZ9HJK8a}fc*Rh(xcwoy-pEn70FIVm+f6D)Sa{j$@km<X
znL~Oi!ag=PH<S4Y^0<@eY(RZrH9A?|c@udxU9`_b9n-@}sBB+Cr9fy6QbcQ@CELcp
za0WXe$h;`T4a0|W-FuwqR(=m2+!fElQZzRVA<YvAHQKvF&v_K$@astQxb^91UGz4W
z8@-jdN>G{-8nef}qKNq55R`Hr7d3@c0H_nJOJ?NTF#AvHS#X-z3waQUPz$AG2c)sw
z-j8d;cr1zy$p((3M-e6}c@_C6*GC`|Zw1&k`E2o{-0oKZ7N}nXB`vMKLRgC}4l&U(
zeJCf=K$VA$Q>+|_!M=UT4D9yqA}@5F2_HD5tDtq8TBmK-J=0~8HHz7o+(XB{qyCB%
zvyK1ih*nrBD;o<UQRHSP{56RgdR0s*3_fZ@PBp{)ymhGL9RJ<nsz6EN&W>*jDSz!5
z+4mjVUu!^l>I4oQ)vIwDRbB&BG=j6%ip)HDXMhAkNPg|{i(4}6A}Tu|(7D<<1_7#l
z#=F*ul^Bp*0CET-P?Xb`M*KP;+#5q#HE%8n-=rU5z)?6C9Rk`9Hv)<Pu}lU@mjRUn
z8Bn4O1%cjj(cm>;0S8O^yIB)5LF0(EP6+7w=IGrC)O_rs-)XpSpm@6uw+hnY!8}4X
zq&w1(1)Z%b{vM$dG4nw>O*D%d0M4Nn3&9tJ;Li-9fh37Qxn-To4sFbm{Ws`9A{^B1
zcc60!+by8jjO;Ky|63>8r<;LdC-;|#b?mT8S={$tq?Pb7qtSu>{`Oo^X?mo-(D@B$
z%t6PkV+`BQql#NVT9qiHVI;{MgB4-i+VOblpSr3*07hB0{!4g3RNMJCv!3^;AA!Uc
ztWr!yg1rgB$x2R*DsUL(e*FY46t+8KC}@K+PaJW<;4XKrE>SV)VDMm4z&-L2U2m$Y
z0_I!})WWZgE@HWObCFJrWkkWYf+Hi{6_+&*)4R}Sf*^?Yy2v$${{Bl=fe-aKnl&`A
zR}T+n5frrL&Q4f+S{J_RB5<1LeAA`Qq*EtO4AgGqxWV&<)QO#L=iVzI0g_;rjFe1}
zGEBdd{PPO5xB}MkBGeHHYot>uaHTgG&BeHKNgLc@h&BLy#?=B>xc94)$Ae2W7AQGI
zIzSE3$E4VS6$`;i$g@I%a`E)A_mQ0*jEw1nd4JOJGe2z$`f1-5TnOP4+8Y3_+xjYn
zFdcGu5an%y?J`c8M3@b%MJCsY+1Y3*`j?(|8NYp;w=WAIsF{1fS)Vst1qmAfz?Zyr
zkn6f(WT^HkjkMXHut$+bd5qm&c>q+N7{fb`E~8JR2Q4xLkE9JsJN7ts93&MG@dqGv
zr(Gv}G*}QhVp?Xf^7{>qV!xUT1^??7J6KGL@_6IMB}hfcZoN_kNW_lY+dAeFRQ^{7
zIj_BZz>gWGuaL=wjqBl~ZUAm00>cEDy(iHq8(s<mV7hJ>>Dfpj0`g|5<3ZV5NNUrf
zm-(cndmKtJtD1oPE3Pkk@}P>;%s-Ta>@ZK#rDxA+!Bp-zC<q4wR7JcUY4BjA986Y@
zGNCr2Q!Kh^3TF;^@HLISQEj#*5Ri|$;d<nI*i5~a*cH2m3atW*l>!E3FftxJDbZhJ
z0H4&rj&4HGX2v~xHWF-v?Tc6A*ed`=7y%>;TUAn2v|r(wnI{0LxPAv|s9P4G;nj^h
zBv)}Z0D|}L^I~48c?*`T7Nt`_@;3Iy@d0INnw6IQ_IB$AZ@eLkmX{HPYyH;?Ku;K9
zl%<M*<sAU<mI6q^%+SRe7TQ84NgKa|mAJuo;>7zKd3T=DD3~=gU0dzj6nQyeTaT})
zw3KbrCzDKp&0_4_Ps^Kd%s(`)_sAb@n#mf>9(lw2K+>z>^ODy~zxVrxt-7F+jSedZ
zYN>4W0}v~7j(B_?jXn1gc`MZf(OBrr7s`dMg%y>Q2r;HG0v#*PyNM_+Ub$7%)_$n*
zk7O5)+&Wu(;HAppw6zF;JuGDMWHbi?B_Yx;P70;~b;+0=Zg-uBK1+hf0F;6e-bmOu
zOI7G}8?s>l_~uF%Y`oIr%|(o`?3?a(s9+L(Q=(sn1YA(~G`}ul)0#S(bf|-gkuh(f
zn4)<}fU!ah<Y*J1khpbutXb|!S|veZ)D2Z+fk!#Wn8%}3f=?7Rfbsv5yA;}DyGH2u
z4g%wG_eDvj2l`~mW~Hb1H>276J|VsGCx&eUr#-QFVS{%__zR$bLPy@A`;7?*!dEFd
zessp;t2R0?z{yxYByRWKtQ1gQ$D&N;h76(qtMHU^c;cQuHH3n+$t(s*pA~^hsOOT`
zWBZ=-Z029A<my+}qxp?|2c1gKv!#xgM|Y`8I*u7rGd1W>G(4Ix9uYpiuS(>4#dGT?
z%a4Dc9}}FH9epsyOeXZNQn*0N&S3OQw@L))X%%V}vNtT*ak^-jU3ay<aj6x6<JY|2
zwi{#!y9(g$&CFa>^T{Ha@QpIUTMAn(8qX|_UKE7SV+#ewY}PP>Gxo0}Yt3n%=xOS+
zEy4ZR1!zgz;vlkuOcsSEB-rSX6Y_ilqESJnmA35mG>coB5%gV9czv~t06i_aUI3=1
zo~#2K6)6YOeEiIr+R^AI#*Aoj24`K>vLv9nQiJzMvLn%iACh$9zJJ`9g*5Mk4oJW<
zh9uh-l5^W_5=BvN4bJMo62le)Diw=PIaS}LZaTwbz-qjZD|w|8v<YP&pMJM#d8p<R
z!ztxRW2*kug#)?n&qxRe$4c{Xif2Vnd34~{{JVS|+?jT#WNNuI*ZD|{p6>s{d(~H|
z^R4hpxJ>vxdejCbUMn=LL}!BlN3c(a#R+XR7FUT&7R2KdH<D5Pi0`_D-sW)fsrM+Y
zj<kX0A*vjxC-?vU`wOCU5b(i-64-kkLi<jB@kM8v3`(o|n-iRB%6%CR#zP&=NuJEF
zT~NfhL|ZK&*G0;BQj=8TG6ABcLTG}yx2LG=20jS5W0eIlWhTZ2Ew5Q0OU*v$^YwMd
z$gB|Pn(7(XFaI78GOA)+hrwmDV~1Wc>>#R5zm;#UtDs8p!;U`mfT_d>BExgA-F4rQ
zBUz-?=3nc@S)GTU5}~ZGfx&V>x_f2WnX?1N))*Bf1utqE3$OrzjG|MFOJPwkXZb3i
zJ4hr@kbcrm)q1|b?le%a(S%3MDqH|hVS$2w1<|Io>naRngn)j)L91?DVEIq{E9Oz8
zlCo;PW^pa1sH%g^IpT&znYfgtA_b5x%mECQi<Hz2R4T&goM}qGY*PdFKz69(S@+o$
zq3Q-6qJr`_H;`R=tf>s3A{q0dSJdr4v0Lify}{S%+7XnbF&IT;`8g<KPW>b9Xfi%u
zyq}1&N@uFrz{W?SH%bDWlY)vEQx8q>k|xtAm>5z-b>+~dNk#+ED?k%7^rjCt>=6t9
zf3}axfP_O-OxBaU!u2`>q82-ikUb?lQ8Ex<75Wt+)c%+n>uq;W!wz!1%P${X_%TF<
zk#TwQ#qqoM@7p(UqX2fMN!r$!9%5gIOtA~F77C`9ARNee2Kj8b7QVupS0<AOptKK<
z!?yjC5L>$jVQa}8ShX#%vNBRlV$_oqmpkx#@d0_b@#}EE5MY!e|0vkwTjE02Cb-HS
z#5?E;3UhUo0SlxK$639aP2{zMAN$3Ln<9}fL1!`U%Vte99FDNiO2x?_OcGFBEKN=X
zv0&maTIVZVw-atM&yIg>YO{Zh(@$jbn3wPW&Pb3L@}dBaYXad$lg(3)KKjQYoX4cQ
z8FFZ#Av6Yi+c3V4Rl=_jSEzV5X!0}gn&5{EkWr6FS5QZZ0?}0nq6|%yw8BI=yh4~L
zl@W3p7uHc9tB{{aXw9sGy<qH!kPQ6w-Bv@X{YQ>W3j_z2m*9iR%ONpkp#VtRta9l)
zEf}w)kjt<*Yr?NgFj-UUt)F9HMyLSDw_TC1^d(#ZN6=-c!`Y6ISh&ICPz--WGkTat
zq@yv;y=t$nB`OtaKk4s|hQLa&OVwVU&D;rafOjjCoqW$*R`512fPG@iOa#(+eEH^v
zZgXf|HF8utP@h&}YT)wq5usSo89xWidj~O^_)h~$nas-|Ix_sC7a8l!FWtw8%XzQM
zS0-3loI~A9AWp*CBc3RLF!+c}O72{;BZ)I;3w6V6-aq(#1$0fw)2DxoDo8D&LmtEb
zs|om)++y?>24oyA-)U0kN;9Y)G!CbArwyphQ?9~EL?c0jm0*Q0VOP}M+5l95u8+SS
zt2m2U@MA<I0Bl<w0lXDHiF$yT5wt?^q#V}1+Y*=*!*bY0S$$|62k>ll(+PV;0QSTM
zrdrqy!-fcT1yiL<`e@eEklYkhR60<hf_vab2*c4FMGHl2Il8+HtXtD?ZOBcB{zL&-
z%+nLtMZMVqwQ*aiyE6zu9fa&_aO=cl%RnV+fwG<OU_@Cr^#_gcv1(`#4_RevCBA%l
z8ht?ZbTA*8^g!ENTD|0&MeXA4#kA(*ORQ9Llw#IBY<0@uB|1>Ak_MqVC8YN><IG*n
zscu^RhYx2n1Q#ah9pRP#TWjp=`R^ym4<VCvElQmLq|O3J&ZXQCT=n<!6TpCiVVV-E
zvWCSA@tp>EUXKvq50Y!6TTPn^9x;hfnD{pT@HO&~wn@(i|Hc)x$$5Ic@J(ig)VOic
zI6Q`%ugI1Cpsyi;qOf;$e#9P;_uo;cIRW(oLl${psd{0yfB*{$znFy`sCYALYzs+>
zos@}_fTI4JyYF1k!Ah?vWqR?NsH~WVg-7IMMM}FDs8Jov9{1@2hYWy|x6V|;b4qR5
zXi`)F5ccByX|icsSN>-4ssxbo49|Tpl{#?>Ue~3J)gt*0e#jYXFu?nK1h_?a_4H*G
zEzsg@1ZdY?MG8$<*9@%lMh!UMxjOgn-$9(Kskql)(JP67OO#rXS_PQTCFj)csYBVU
zT|2O6&ux$m)nvJCOww?OZPS$F9GX#Q9X<=+A!LzHfqEcuEGZLb87s)#nzA=r?gs8a
z!(~`8kF{rvKMPY&nm~)t;&O5>PNQHr+QnWk24v-6fxV>qBBa8PAlaew+Tff70N;>k
z)1EM?ZK$6+jq<`}D8f{#!CDj}GvW05aivZGcJ|Bcq?3xw;7k@U-Jo%@^|lFoTE5y<
zbBoO5C4iD+_m*V`s(omDt7{BX<Jrt;8w|$Wt&JX-G620AfDS`DEbAzxV&^ZvoFzCd
zDubf@X#~)s?p~np9*Y|uiTIznCO(A3#E3@ZKsP|?b{&o>D4!Omj2EAgtuC5+&XoVH
z;KdLn$+F@(lkoWP?PMllK0%6s#<MO}CLikgA=xp8MOD$)$w%9nu11R)b_45S_e=up
zP=%rAo?cV0(&VW#a%4}T)se^8(E6tCp9w$7nNRRg3P4qvY$jgWi`bj*7$ZVkw7j3(
z@h_6l4U?pXNu^F-Ll*oH0fhA1#dRQt&H^`?zP|l+6sKh2n1-+IpnaxVg72j`Irw{+
ziA<H$g;=rp@<VACPB{cut4S-DU{3nkiAO893%v?pCYEfuFtvvEo|V8%cHNisO=`==
zFcKsLi0pIIYEWiC)wf!%tO1t7{fq8JNU)AY8StSadwR(!+An}-+4u8~x59}eqJr=O
zDPa~7nny5PNS0gSSg%V*xl{sR&;ibny1X5@(Kz1<xX8|F-9ht+eW{>eth0s#0fnSq
z{(aTwmk?-%$Yy(CV0Zryd+!}p<+*(elc+H<c11-gDmIEBAWFwX6qRD3Nml_8kS@KM
znAkwDP(+%7f^?-<F`|GV5V~{)w=@Ci9lp6X=H7G8?~d=!Z;bDb!yk#jX7BfX-sf3m
zuDRyYDQ~)Sk5uttK2&b3-p(Ij1~7?`77$4sa}t{*e72D|A7$;#<ddy@0fx~-c81Al
zgpu!^`IH9a={;kZ-qR^*4$)ukaHWDwy*kR|WW3{1`z7gvGY@mWK7!At;13yGB#ONg
zkNIEN1H3kabKWRvAJ@yD54DJ_-n7QO7<4;>$O^tbfH8;#*&EVP5_Q(#+&QT-E7px0
zseRfEJ<P;mNyzS#NUzNUxR!zzJc7t;Q=cJmrnrafQTDs@g+j<A+bdQjq2^PCxUu78
zZ6ERjaDY@1-_}6puLcwN(rXiuNn`GIICLvC5itUEK!8az0?_<%0?=u^+ozh$)5yM&
z_Bp2XQ3MLQqA9{aV)bhG%pwRmrl3%=FBjP)n)H8@XZ*zqh>l7M)M0{Bg*Nu8c)}XY
zV21HGgJXg!Vd6!jqOz=V^1r}n?TbGLsKE!*Bw$=wcB<P^XC&65P*h7lC##{Aa^Vq@
z#x8`Hdv$l9#sJzr;bR#3chP(ds7o!j*u}tT>Jdj^MWKF^H!f#4O_^rguV~$bD~Q3|
zHtik^8$3ZGHYnz1Hu3l>9lE);0$x86gEU?ZvP0Ha75Q-r_^auyX6G~QIDxgAw!OXq
z@a-AL9L!%22CP;CZg*XG+&LBqQ5Lf6WTiZ$?~Q?Pb9wR&e560$GxtLglu~49?#x9K
zkEpH(L?nuHi22$-a$EybSVusa!S9ra?-}3v^HJb`_h~$2JcRE6#U<mE8fLocz1fS+
zKO)Abk2r*s&P_o5o`g;)sc5-0$FjXoS^}~>;3sL`6)a<m7B0MiLe)lax)~#JxWGw}
zVn(QJUu%zohS{I{R9(=A!7OO#`DqA+3Z5W@@5<c4=FJ}253C(P=Y-tCt#~sBxLOj(
zK;qm9`hc}d*rKk`Nz^qc8!4F*!z2%y@i2^hTdN>FP<rZ1@+3aTY(jdsZdM8vi=ege
zw%3MqtXo}&Zo#`Z>(wP7fXEnegYthJN>M2=85B9z6&uN;br^>lR%~Yp1IVi=EH%lZ
zbG8n>jd%9D#IV&vs7N7kBliGaC65fY5&#r`vUSJyWXEx2m3D?X<>nAeduu&xn2oAr
zI2OAL5~^zO$>#AYDHl%Sq#OY)ZdB8YVTLj(nUrB{zbw1*=RlF1kqoB*V?idcwJ;<8
zS!;=CY=`vthbx&4SHBxyV7&LkL11{Qko(@|B6}Ullf%;Jwz6k`Ln56p^Y&E9@A57g
z{g^`>TdER3ygdb&6475Mih>ijE97u++$-mVZ$}WQnd1Ks5Z>E=dH)Ed<~L9l18%PY
zj)EEvT;U+m7;{!g7V;tb!;=poW)%_Jz&fBz8x-xHwYQzC#QTA~R$-Jw3gl%_y8VSd
z{O!G+Xg=aWDD)Je1TR1Z;NubC#y7GK=S;>yj{g_pCPMB06iyPVPLCm*6$2qiIwyS@
z?aQOQIf+o(@ItasAia#c>)4@Ds%#>J1kLW$3=j$l%*i3y_53|l=K;Z74Kh|Fo@H*=
z{cd{3_mlN)lRDtc$Z5n1JKSsNoIx2(yTi3s*!X-$PBvLR08yBWT6l@gp{x+NI5>`)
zL<>e}k^oL#H|V*?jvcFnbEnFOpMb6wp5X~BfEtqY#zz0g2>&wJ%bsGVqoyG3)2ykH
zWh3CjCUO`}q;Ur0F{lhcZ%O5tvNBm`u#)3ZOpAqaxas5L5L=mS!#EaD1<pLNO&c5E
z9!2LA#hW2gPN&dqT7ekb4qXfsQKH)yS|r4cM+&$WXaMm0Sj6VuJ%Q<dz3^h4S60YS
z(VYsgW_Yv|RT1YJh$0yT^mv^_5u_T>ak_*In#6*Z^NeEP$Sz<7u}S(QBGKEDs}1K>
zL&CG9cK|Dk3%6%41jBL{<_UmoPE;SNP62j^g>_4_JQfJ_wF2uTiaH8;9@xD~^l|{`
z5Q28uF7$o^1*Z*)Kj|2?d-p}>3o2Xy$Umsni)aIDgyVaRQ+@%fN=n@vc@cD<Xpupn
zzm|<H8pDFsFh&!hm^y4v^9u{ZeNYh%sB#m1Es(Gm%|MI*itK?K_}#w!VLVhHL!^z0
zeTiU6I61cAQLsjaO+A3yID<<<7fRZ`M0G-EB=yPB`$MWYwyM+<3n@0k`93q@79JwO
zAtYvn(l*L+2jOUJH^!7J2NNbpJDpEG?r((esW`P%tHVi*(a(DUiId)w1}=~iV}IU{
z+EWbtMJHYs1B}+7=+nCuHo#V@2eJ^6i(BL&Tq6P^LG*Bd-Pn1AkHFjR%-^$3H*=qJ
z45WJ~cSrRu1rUNcnmYj@U{gh5Ry7Y$ejcco(MLl+*&>s%j>&aOL&z16VD)I<8xi7d
z2JaLX%~`)&OBG_t6)MN285<HrQ0amHISI=LE=fF^?Z6C%fJ)(WHwPbG!0MHL%B{!{
z>hLsaPEQx;TU77Db*x>zIvoC<2}1!&f&xd7K`u82aCA+b-}!m4mUKfvB@$``7!cLC
zkyFSlPy~V-$!b?)2f}NtLhU&d?k2n*!c_K?U<{<Vj9Cx{g$+=J2jYtcLBuo&qiI2b
z$p%3|bQ6fzEwyP+_(Is;wO9g1*<Dxa%9-6@p^Tt?7)8cTln<*Byrkkp6{c>+X+gE#
z`!sBlKh-+5t#4^6s(@DoEvBYXdz(RkLETg@BrpR6F`QE+lv-_H*JAS(Lkt4$2M{+k
zQ6b_X<bzsvdV?x7-5^!xeKuHRsxEJqF=EMgTc0LwgYaUNQ^sQv3)RaiySII}ve2m)
zHG*cmW?thvTMr>X6Er~$jAD-zr=jy4&?KV8M;!@_FL`Rz?9I~r<)&@v-&eI98&4#=
z%l67F?c`}V(QJ7z_v6(~D7}jZ48<kLBbJC1YFW0w>2|yO&kCDj<eE2&FhlfR)1-t#
zdkBF!^+don(PSJRJZGe%#5tzy7*dJon5pfN6M{o^hOx(f5hNBopcvbf;{^o;ycn~9
z`0;C}t1F6DHR+c6bX))|WCIe50agW5aumestf4YV4#v;Rm)mEpt2eRj0XVaqUF;#u
zf(e`rZCQdCE^#FR0kQ~~iovrr1r2}>c`C%@2pb?j0+fo8VXuE{jC`hY$Z>g2i+{KP
z67@IRBf`UjTVk7AS05#;zg<R6InNkrB5?qyM}(X-iJ9EJcMlYrCy-yJ7!&#7rDJK8
zh}XAcmY+@kdxm=MvAcxj!LO4X7tvS{N3q-8L&+V}{LTTRRl^Xv>KHXXZAJo4YL)Tr
zD$6J7bWyneE_e@z2kwNHYNZsZuYMXQ<a*v-A}oMULz${}__bx0m*jxeH`<)^#HV`=
z5U8Wb9;jFk{1!wGP67o(6#j&K05&H1fF3<d)YX;R?*(@Y29~$K`x5MdhqVz6?$FnG
ziqeQFeoZhHl71+<#39GUxG?~KHnbtpQr0^v{rf9d!7-q`0+L3MAM0o=8^5eE9UQai
z8Hp^5FTeE+8{A{Pi2mTEBKgfCW!&XbBh@Mbi5h)NbN~3=5>#Y?@+)W}X`L0&`fUKE
zmrN;8K=-{1ILOA32Y$Z~Mc-tOeAwZ3c!mLoft*nUyc>`>)VJuP8SCPT!21E<e_mj#
zlVKK~w(?_+nV1Zp<iXPm60A6#A`xc_BnDCF<4|$?8sh1XgHrb)G?3@{ITdB{Y<B1L
zBX_nIpx!vGD;2P$Z6)K+8U!d7W{#RNE9#P#Cx`g@e(K-)h$Vbk4VY)I_FOplr!hz{
zg4N5Gq`|z2xv3M%lIQuy_t%~MjF|=Q2ld$39_yDmJw+r%_@{1Y2a%>6DikRS-W$n=
z5^<jAwdvkXE(XS!Fr6+vGhC&7e5F2OYD-4F)xocMz(u#Cu}Cd;QYv5J?C0(cKP+C{
zgQ)HNF76;S5_Q`Y#Zuye^DuDmhAT<v3nUkKp>4NaaWBi|X=BXv5{Nn;D7@WA;d>7Z
zzp6n_n4|>x-gF3XGp?<ks0Ez?L0Z|xs>Mq;e5kmM(c)_trabyJd>LzB1dhQj{w(`m
zlBqHdk^xnNNMla=o5v5G%?s#>Stg5qW5JxSzjGa1@JU7j<FRimP=o<Oj8V`Fh;0)4
zbH(;w^)Mo)64b_Ir8#6v=F3#B@}V-|3vl{mMDW$sS4A}npTYi-9{==-jjh<H&|Y!#
zHFpNEN~d4e_eYl7+QgE9MKKJTMabO^{cQJLs|Fk3b(5E^;sU5jYP9aWcPw+!cvv5a
zSwdaNCdUq%u&@+uN<jyzs^jlj-?2R06DVeVF3kia*=-RSSJ!?<8t%_uK};eg4;157
z0l(!4uDrdt6Vs*-2sooKuLfA{Viokk;1H)zZ2j~Ei1lvm!(l+AyZp5+22^e=38mI0
z<TA=t0BT}8LS4#^Z|A+Ar2Z-Y)hmjyc`?D{`1ZjNG`ymbY<uG{oV~(tzo|N~JbZ>a
z!m+fTDFn78KP~I`qtn}((i`xuygPOjtavgI4~uGwGDUr)52PyKl?D&qz&%N(Jko2#
zHr)9X6R?-ugXo<mY-Z6uqUP5jMWriE#kq6<A-X=&k6*Krap~c+GdT_{EH)?6#Y(c5
zxC*|;92_+~h{Nu?SO<^Q7OPNj4))1J^xj2P2p+mG4edr3URBg_>PY5NP<w{DmLj~n
z-Gfo`VTjnrRBpKL<mNsIER4e6kjVYWGEw0eQi*0^$&AK!a7GOvsrg|Gh^UC#V^QRg
z^;_z+#mv>M(LB8in8m1yiE9tG^4oP)IJ%tx(MDDN=UoP&k46_#E%ltobl_j8EF<f;
z#OcwV))FYhr7z7dTGvP#Bp?^K^RnKantz5p6%p4;c50s-*9B4-{X5CVP|?O`JFg=t
z5L{Y{3UxU2#3rB1Eh-0vJlOjz)<6`^1G4@_4$$=<#*$8DZ(zABb@+!{W-6wpF|eOe
zD1IKxTD_P=QQQUm`g2foo)!3>g~k2QrtVbFeX1Fr%<Tr+)&u1AU;59zT(M**w+SJF
z8IsIdfgf&n(DmWW&}2G~^^aa|LZ1nM?Y91NYcC-DjZFF(Pc{d@v9$q4B|ltAhcX<8
zl4Z%7KF0Tdt_Vr`?bifUY9vtE-^P0L(e4ji!SeXE2I@6hV5!`_?)gEy3`3m<17P&U
zjh^tH`a^rC6_@!_yxT*RJ8W)V0|wPwSo!c_xZWa`m#ov@fB$C_y4t377q0z&{=N7i
zD&2bMz?PRz3t0}#R-!yXJyfl44tyuSi5CWfq!17Jt2^FuX2mBjoXq~0`OWkHJaM~%
z=YLFT&p8nFKPIw&{XQfKcJSZ6UjBRu!|wj$OAt-a|M4T1|Gpt}MgF^n%w_mLZ=1CP
z4&Slp)38b~4QF(h;-xPuLCM?QQI5JOm=y0KD)BSNK5(FLmYdMQf98;!+&IoYcDmu0
z>SfmteaCX(mpdT-_iug35}~qh$t{F><Jmvr1()j(14ef)K=7#XY54MH{OGne{Q-;B
z?jMKo$^Y^7a5Mh?KYqsI_?vS5fBx*lBJ!>O{&jgh(|G**5A}XT?cl4!mdn99$U)H@
zI^ZD0d_~=TQfx#Fs9Z%NdvHVJ(Q+&&o*`R5Oe`1&p(VMLW1|_q^$iP)549tqq+|b}
z7VvT|sIcz5tC(+84+1$68)3c%6D>W*vlm|G_?AXYGuM;lrRa`5d!{xUAXl$G`5nsy
zCcZsJ+)Jzd_4J7oC(=xSoV{*lzT3kCZ{3aO-}ql{<jki3ngJ}w|9K1lUuXF~%sqGL
zs89>&n0?m2@Rii|t6R?A{^d^i<ukt&6mLGX*6-Y*mBHTXo>7iksSST|Mm5xOrUn~m
zcW6CdP^%uBV(MA_=*OQ=9{!nk>-DWme*EL))!Sn}v!CpWTG(Dl{`AT&o}s1G*O3_}
zSs#>;d-r%K{4MkA_GKCe`OnvZ?|=Q*L-*5<|9afEZ6+`M&)0wd@V|@m|FUzc6S|mo
zKBfvT!~lo}LksFQE#<C#07tJQz|GW|YD_U%10X%_Z&!GRJ>W*{4}bdfsg4<5HxpP+
z(w<F;*+2jMa|Gsm|2jl(XQ-u)sQ0MS3i{@RZbIhg2N{5+Vi<8~$Cd%nUv)PU$d|u;
z0;A=+Sz>OQR@o$|lbm3XB>E@Q);;BT8nbIzSi~OZ3>NcMAiCt0jY7`#9MM}*Sy{p$
zUiINXGRQ&%6#9i#A-X00WVpS#<x29C)laX@sIO$c1H0`zp$Zh3d1Z}}TYQg`!=B_V
zHCWV%s=8kojRE?NLc0%_KkUDP|G(TqX{efe0UFhG8OHd*Z|{02iv4int{TWir>YL(
z8y3|l;6U;yOGm`G^8;EAI1RAOZ}<n6%jKScUJoO&z?{vDW0N3@e0&bP8<PNH?#|Mm
zvVMIQM1kYwgUX<4gnP`*s@3L6SS(z~a(RI+P=a8qS3jxMdI0*Ni)DHF-NE^_*alcN
zq1gFnoc5P_Y78@j#B6$(1xMA7s^uORWWMwwa!{Zq8^Fsez0ivC@@uS*NW&Ty7MVn3
zqiYu}TJ#)59<Nbg*Dm~8DZ3)1=^?Fle6Pg(tS!-iql7S)$!p9`n~Q~IdJc0{3_x08
z>+iAKk5^hfT*VA<mum(tx53PZLKK?<D?(XVzB&bT)iq-lmUXSBu~EuSy!OB;rk#4Z
zg|r7w=DE*!jUZ<@Li8+y87$}-s|J&(Lty#{*Ct{WDnKpq8llLcXyGH~Be@Mgx?POn
zt0G)muKg6ZZHyj{&C00Nd5sKZF*nG{pP0t)T8iQ#3E&B9^e3R#R*6hx`MyBsW>7nN
z$GbK19^BEx+p+|1W7<acN@WZnz-%SWHW?6nymE4KmMSh}VObo8VIOh;<X4Ze?Ynpy
z^NxiVtfA#G^k<r1t0P4IKWvdPx~?0+k`{7C1Kin-DaU3qx$IB>n18H#m1pDuy&FI+
zpX~v%f#u3a_0f8(A&BOlkVnYySy+S>=AY>&k{bbHs@63_S00Y!lPwt0*>V|xYd67|
z&#9b}zqOH-Aeaww^p}C$^Y<`_qri%R5a*wax%_O0cu!khT^(3TTIE!3AEhA64K<Ow
zZx@xuov|EH3Ig{X11IucO$`nWwWBt4D>yiq_dYVU<pNOQxemxy%~2m^x;C^$c>?~>
z_)C>O?}qf3NO~BTuanRUyY=f;tSt_pXDl&Xz;ZwwEo&lxT(myA83$WmJwT?$>cAT2
z4jK309a8UZ`IUo(h4*t@a#_q*&Tku5_g~C|)n-SApGG(mbcS9lDrO*|s~H+kWh3w3
zm57R#Jg9Lcxi5hJJxB3|y2DtN7vWQG7v;MfP#F7ljt)5M5+lss)7y+a!P|*>dtTVy
z0WLF|#v$8pGoQ`Z=~^-e9}IX0(>mubs&M{TpSkxh$C`Z?x*#}!K>8%P8~$^wMV0W0
ziv#}mu@>vi!lRJ_2cHcazGg2Q-ub{2sc#GAV*;ie<_RahvjcO%UQ`iVdT|sIbFq(q
ztSQ#n57Pa-#j*W`xd}d~6c3GdqP-{f>QP+b2gdwC+w+!}rzh8u_TO?Lyqgc|Snna6
z7CrLqSjRdcp%u%3=r9i^;he*VKB<NE=A=scXFwTP3dY&}%;VuoBd&?i&eOxltXRRq
zazqfqtk8qKlhnujw69KW^pwS3>wq<84c2G%@CKH}IjA$e2GnpY9xxLxHk?{1&cn;U
zjPX#OLk!}SnV!RfFk=)t9BFV-WFbB?raV8@%o)JKLua&visO)}R2i_iCSX^&h)Uz<
z?87;IKgg7$j-yt^jX$X2&yHM7SNrX2Txx#*eEasu0p=oo`)>1hOv~`?yUPoiC+^#K
zEUQ=kYwrEuKm7mE;+WPAGoRwiM?A8!ve?eyfGbbJdcQ~gWi2b~8xT`Ap->*(H8B6Q
zi<RcL#Zw!cKv0}slAAaO#bs;s5*&t4Ii^d!-Q<;X(6LIoXs=PnV})Oi{d|KC6$?x8
z@o(=T3~JadPzv54<%~cHh|vHY?KOxGd_Zq0T(n+iC9dSA?DWSK)&KzCpp+8pcy75W
z>Tx#^&G>3Hq9jFtT8-2y2Kw(jJ5jqWo`m@@K+8;x7&B!&Q1fkf#=JhdGi0qpvz-P+
zeFRR10n05A?+9Mg5p1i>4uR8G>DQhDPAxbo&!J`t%7&gND!!^ia=wKqskVcn;#NF~
zJ|zQGLI2lV7k!K4E?-fwx3`}jj*`6vSyy4}%~MryiuWqq=NB#T1JjhVY&J+rPBy@A
zgwDSD;{hkxQ=Hd^Z=o{VP;%Hz+^@fVLC{nH1pXYw6sJzD#-Otb#90^6UC*lg6Z3Cw
zYtCmP2M&Whu|pvHOOj2z8=ONMrdT_H9Vr6TPUa-qf7DSm;Z}wL6hnVqIAoL5ganxH
zaG>DZr`{l{s|6Z%gzL~*C0YgvmGPmlxnLV5P(Lu!QucFs7?<o+eR;Tkp1U=gQC>rI
z@xEgj%u~VHYaHvGoJbCO4M1Hl#7PAJ5W>-nXFYclV@Phmbne=<>pJXuAs&qZoU@t@
zQOdXl1qBh{M8NqyN8sZKs+R$3Kx|!*`Hd(lFZY4ZIfQ2zfuh@U)E)lWjpzspxMLfm
zqNmp~I|rK7`c3;TUdJYS13Wgy&bz6piQsi=nXOIK=YnjQJ&3V8xpVQ}o3M;`0s_{f
z<mX9`9*&9v@8|ey;%$9t&fd>$tQO!C>WMod0V!DRgK_$KG8hb|Qx(D7xr>G7t>EV8
zmagras7hw9bn@iMrJPb?7`c1;Q43o5Zb15_5FNI>?5E4H0-*gCpzHP=9$p1%xTosK
zIOFbELA<yqg)$VYOcd&NLU_8qBdE^Ff&sT~)hdeTP*W((><HZy6>M`1W%Q0UW^<4y
zdeiSdmVF<^fbg%DxYu^{E_kEzEe~o_1frENO%-ez0P@b$y_O!H*EpSbff=BCz7=aj
zM>QO)IW;E<m4gU?`6gp1M!rFDq#bet*RlKi=G-6KpqeTVzcuF31X_x(*`~c0R)VIj
ztl5f=O4<><PODboS$m^UN{u%LbZ^+8<#DJ<-vCbZ2DLek6*A*z_|XdZ?gbYyOQ74l
z&&UnjTzZ_>uhV04?KRW){AtY&InX_1o?E~|p!w4)1G3QucpXE=12JBhV-pHG(>MC9
zP2jj#0Yl%yetCmQvu(%|l%^IhMOlJ^j{dm}#MdTMXe1ZyHAqK`%6pJ18aUxVZ{l*K
zDe~N~_reL(r!%YVpcSkxg)-UzLKavksXJ*DH?tf#WaL-<dITZ;fm;)`=`UEpLWu4X
zV!3<;6SM81#VvtCW{7<=qT{wl*EnPRNnn_DZbW~M6*-$zpsia22odcd{rGTAe^_jE
z^%vPnZ0MBz%5$UXdAG1$hf%z5-=8<3#Eery5!ZU72vY!)RU+*o_nHn=ot}%ij@4g_
zjDC0`i_PL>jlSnXk#ebfHl=7Vuc|FELl=_u+wB3b(Is;%er~|W!%L)0jC(+Q)!cGP
z=ydKk%EhC#E)}hNjoSan1sJT%DSuzz*I~26Q5mT0w}XSgZ#%O)y=k?Ej`gA9Nz4di
z14rb#rTb8jH5y<J!Sa+y&7r<{%hF@2krBE-8V#h}kgHk?RrE{IBL?2$JHQ0FJ09$V
zS%|_AV2M6uhCMrwn&`GW;l*TkC!`VooGgQkTtZfNj3c-RB73$YkAN0rwoc;IcUaPb
z{CZJl62$U#Vjmod@Rf34i*acO?2(Jn!?Es(PgfVa&9RD7m<%uz4A75ejiJHD7ZY%l
z3Q&_Q=6H7j@s|{O+)OOhXH$^L6ks<SV8$@(azlB!;l5@`ft;_emh{JZA;hPhPDxdb
zjUF$oiznuM@zB*SS(L;*%dun@8Oug&d{68Mp#gr{I#7PxYyk$A@Pc}-5D!Gm1z5Y_
z@h^iCZP&TafkHTNg=_lZ?RPSB|8M~s%J`KanNF>n)@Tn^!wn4dcT9rQUBYpJ8E@_K
zoClR|A94CMwPr+_1Eu=wkaH-+<6Scbj*G4o6bZ<;zRpFf#A=KZR{g|as9n&9!awn%
z8Cou|_4sg+@Mq7S33Jz8hc|qU$e11V&`Y<1J({cw-=GJg?P-MfvFe?$we*0*P)uJl
zGQkk^U?Z3J=a{Y&!;fud!S90^;BD}Vn(ZKs2sAhT2@w$0a2d>Q9H<JLP0>2oOK0Qd
zU-PLG%MqIJXVQbo-(Gn$y_RY+@>t4t$OL-AAREUSgJIN$!t5=O<+gR(;!@Pc*M0U0
z#*9%y15(Uzt!v3R7#i0#`M9QEtX~aL5Um_MUm?)i96^&ZK!Rt~2g0#6dDQr;KgmYJ
zvI)WMUw~0=up^@n9$x{d1nszM#Y`yaX$?TZ^DyFLK53HG<x3Tml$4}rd#o2VKL^NI
zZG|esN)Y*OuGe=J!XL1r+s3Gm)aoJ-^Y}@mC!y)F4M*VyF}-xtVW&<%JM+uC5Xt?R
z;C>~=ED<|dtS2s6SbPU9h}Dm)=fmFBJbZshShnP#<L|g%xscP~r%nh12KIojxC-Mp
zz35Wmaf)@q!sb1xuuz*jF$?Lf^&o3t6E6$ze5#f{pKJ$S>>pc)|DMB!7ebSVBf=5h
znfHJPw)BDDvH{};*a2@;zzkxzjRNg*T2=k&w)GJZe*~Agy|H!H{$lG3X`@7toW${7
z>e1r7X2rfd)JK*yi^}aA2-ifwT30|}L=05Mtq1@UAph_lrSxbQaHKaPbFe)88TX=(
zdC7Y}mxXKstZ$USh3%1PaSok@8j4{u(gInX;p4B?zgS6@nA=-yD>wtJltqJpAOrr~
zN(?|2g0w2gD~F2p&<y4CvtZ7%U-dOk1?bou#q7jwMs9G#WYOd-EG&@F@S!bo6CDbq
zI&VNKQ0dD?Z6N~U5Nblg%%#8nRJV!%3l^3zOX^M3_Mql0*Q@9$e50nJ>t|nzl(y|A
zUjoyo<qe4O@$n%tS?ADXEl;U5&sZorvj*M>x(k6T`$W#x0!uqPBs(Vp6Dix9cOwHa
z>;PB@^6*Wk`Zn!3bD+2ZG^}8IOF22jMRW=WJoTry!ds8t(8Ap^Ltykm7HZ^8*4Vf$
zLqdO3A7jf+EWKxo5c{-)0&oMx`$8B3m4r>=HirC~NrrQUG}#<0?=e?MQ7zK{fGh{>
z;sMmmBfsQ=-hbS+3HrtK$csm?OrEGidF*(djkb%7zN{WVg1g8*of=!daqq4lR&cUY
zTv0r0G?C9WyONv|#DI5f-HI%6eVvZ2k}K#l$1`{03vG;R$g&iR%1&1Jw9U=-&$Xd|
zEogfc3tB)9mTPXj(3sQHhc31Wtn4Y-76{Lit(}s!K)CJt@n`)bPG#`?m9y5b(F`^=
z*N+=09D6aA>uctK-4E#=x@V8JApmyFScFn{I)lJK%D<RH#@BTysMQO7xB^BV^f)hH
zN&=Dfn5i=$hKG}`J=(1|QH7$U_~kF*s9__V@=h^BGgfT|0Ls`)%oqfi>HVY^X!_p(
zl#6WVlUmGTW+ik5=lsbRvcqr46`nE!4d^D-%tNfP$Q5}Wb5I5RXlz=@>2eVj?9II`
z*d>QeWXs%O>f0c2sGL29E<L49*JqEwa~mMfG06)!%CG-I%4mE>&|^X&7?zc140a~g
zy85KNpOy=bYB<6aMrb=Y@Lmc6nb_VKgc|naIu9)b5jmip@Z^65WAjL`bpc!0#PmOI
zxS`%152p~#aL3%@Af#EbLE8CJac=8GnL``M`^t@Z!k|ZlW^{&o9*alVb_iakxG^7u
zll5qG?N=G)ly!Gi4B8j3W9wVf164UsP?KcLee`F#HLU+yv>xdaB3d-8g;)1WBh-N_
zmL#|dC;`BLj&Fzsi=TS7eZA~C{&Mr;qhO#(na}*TnDYo*k@s~+5Wm|Vm?hjm6rNsb
z?8``NwBQqkf5v3oSB{{`MZjb}sRd-BZ~8SG7TaV&{i;;~y7rox-{T*UX*r*R;EtiC
z{a|`}TI|cSe1hG;QOyUuRJ*ggEA;0s3*Pb+ENcmm8HKP72kc+x9h`;6ft+HfRAx2t
z>bJtK8=VjbRXFW6(ngi5zt4TWCi{4qQ9ADI*ABPGh<z$Rg_`lvh9{CanF#2xI2TuQ
z3zj6HXSDY*!7pj<z*>Ck+vyP&pRk8x4fNQ|jPY;ZocGQrOyJ@8ZI!TFV_dxj-MSP^
zfZ<FyA)2xVKV42)Svh(HX~$s@9dlrwSVI$$_w=DnGQ|vRx=>+X`2DXvQ3vaB3XV1o
zp}+Vh7=?M$SJX*xvn{1c*%P;a2%0~urE6g$*@F4ee%Npbokpi>Tb+C2H%K{J1-gML
z6qMv26toGnou^hAV?G|!1|%WI(*s7@@$5Ws))*2|=;t793$d=p_uB3ocvIrFq;$dr
zZ(FmvSO)mlsand2z|&KnvfQ;b9JroI4}$u2^Dd0{(6@{{g>p|AwQywRJauTI%qi|b
z26sf051~CGZ@+p29OdV*OJG*X7mNBWGlGPpFiDu$mtujdY8A5MGmul6j4|^MuwIiE
z$2}zE3=6Ok4C8mF%e#N<m-n5K-X!Dd*rr{<y%iJo^gmklW6obtnqFW>Q6U6p3vi$9
z_rbve)~#_N?NUt7bA|PsxE?1dC=A>vzW_#P8z-3zKo_>McFbIQ`B5m59n%JJ&A~6U
z6;XOh`Z_2)?}jr_PtZetprl=1UY_{(MJ^|7*Aks%{AkJ8ZnIDjeN;1yHOnM=;>M9e
z2UZ-gmy?6ov+cDbMqK&bVo`A&M~y<cxx_!;H#mn3!Wt*>{NCFfC#8LKHRsAio@9jr
zw5K3Retsd37{WlB^q^PlI(>Nkgi-VI925s^#(2%lDY|a%X-AG;tfLaRDZb<Z0eh@{
zfy#*e2p%8;yw_;FvXQjD@b>-t(_gt1JBO-{){^jFAsX2?QuU{LcEMl2O=mDSt(=@{
zc@2SjFq709U|lBO@mSXugVhUHxPQIEEAD}Oi?9U5=uc{qtLyaX;YcWH&mn5!-P<S+
z{dw)r58zbPGAuQR3@2+_!=^v&WGpdEg!^cLw;P8ulpUCWNYi1sHem^sLm+rR*D8yC
zXI^b2eVliQ{l@kCdreM}cQr~t?MfvM`vUjA4p{h<Y%3yjiqxN!mH8&kh8Q1<v|Pmu
zhL>}ELo&9a74$TsMGljEFLItR`>GPv!DYzpBH-kJhp3z80uF#SL==S*fw|ri-A_|S
z5Ot`U7Q`gMIi!prEja%r4tWZKeD2DG38cWAD4#0s2VkBY0R)3BC|bDKc`2PQg1p;o
zY3pXlM*_j$XYNMn76Q;hL?~Zx>=~7JACBO9<B&Ji9J4V|r+CsqPgQj-LOMm)GGr2K
z#%RG=mMi^p2jIlJO9Al`La)28*>cG*&oUs;+K!VRJ1mZH`XrjBkQWsio1#1P7~~-K
zFj`RgcufW`WyWQ=a}~AdX%hR9nWD|TtT_V?m`M-$QH0a4!6#c22D$F^lyt-L`$F71
zl!Nc5$y4E^{0<577@#eKC(j?mLIF6R9MjMkQ@|a3M&<xsgj~nJqKD#G=H~;BO<ODH
zBn5eSZvr^b=s&=`4X$>#ar(#Q*AUA5EL=y9Z2)Qt5{OE9n>^t82EYw}vNS*jshP=`
z2l;J;m65tbA^om9&E2r;20&qq<Vtyv`}Nh&&rvn>9f6zApwP{}cd)g*0tfPxEI@>k
zlqrZjgr*^!nCb*Jkl8Vwf+HIh2Po=klN-g&u)4=Ixe$0Xe09>qI9aksel1}pz-CP5
zN=X<U&IMmWhDEoJg@yH5dp><{Fb0rL+6WET_5?=yXKR?8YJNpk2$mC^-=5Jq3O)de
zJbh<}vZda;2cmq4oK-(JeEIT)(jF=Bp(zc4NStwTs?nm~;i!x5OW2tfpBZJ<&{>6Q
zfkI<MoQUZwoP;(^Z#QcYMN{0lX{1#*P-;l&fqz&7*x&{_5pt`Y5i2&#fMjaYLntPa
z_axXx-&DMk1iLd`NaKvhjDli<byy$lZfiaJg)k$KzS~OlXKnr*)9`!j)95i>=ply)
zgwM@At8n+$a1?<VeSWy@>lqAx?X8m8ZF!@eAYan<vqmhe?nre>mY4Fy9?~c^0npgk
zI$p4!qSOi?cPTv-{(&9s3W!FGvcuEpfhM#I!B^X}{N<i-AXIk(O>#4u@bwasI{*Ws
z?cR{`eZ~V^;$d{Iw<FqBqjIKwe+viP?M>(<6hersY80|-W9=<2{k!Y<YkJX<=mi8r
z#C(Gt4klG)9f2{mBbxDq@W4(&cJN>+fO>;JHo)lUlQ87tIKA4YKHyD14r$LBD)XG1
zX`Q=;q@;gMSb15QH`nafYXqd;^sweiMw5zfDjwo}88>H&&Z%Nu-c%-q?CCuUNV7f=
zh6x=nb9k`^XN)(s75MHZSf&AtA{N8^iF@`5@a5Dh+>U-jrAuoZ)gUHt6Px%+&Ydi*
zw$P={zsCz@X9ofs;Q3Up4wvvXx361e4c(CfbU!iiGdxDj@_aknK=D34pmA;hp!7K`
zwRLxNlz)R6veyB;hmZgi{Rv)du*H`FG-Ju190r2Ny|rz9-uNM>Jbrp+m&?CF*TU!M
z0K>o26|6$TAdjsZj(P+f3Y#Q?#yR(3$F>`QVei=}A$;Jq+ti1q1upca9ud~_n?UHx
zO1Yel>%be6I@pi>2-Y<|-}(?uF)x*;5hpyU{Tan9h=&be+X(~QS&hWc2UPb0@T0k~
zS!1gMxzZ8oF~d@4^jx*0zm_+(7B-$J0HYo`h|Mz~Q>2nmUR->Wf={~7abO5*p*3z3
zhr!+0U0VT1Km>vfjrlhT#X-Uy_~qx=AIhbM6pJ_8;_0+O8^sHulI$$HqYL2jBJdax
zD;qza;l4Jgg^l{9C0CO0^}%Vrgp92T8ErSqtHBO_L9wySKxV&Nzl+eh*Vlb#UJUvr
zzDAzgUhXk#`wjsw#C>|nm2smCNA*~Q_};h*6o{q4VaVIc^rpgv$Jyy0m%W_e4B8*x
zQ@#lW1}VHgcI3;VpVz$x^;OtCVg#lOy~TY+7s?fZz?V?iKk!7*ExNdDmw<pb?4Do(
zXcWF?I6$0CFXWtMWubZg-ThTaNOYf!`hUhnd*dwx%<t#IE?Cza1sHvK_rM0fezMd^
zUJf6yQy}xx-HcCn6^_O4bZWv*U4y_?XDOFV&Z#NcUMS(;_*o!hGjO|BIn<$}&Cm|J
z7A$iwV2$TLDKdA_;E@@3fBF}T-yR$XF_@fo<g#(2$^BTf@s_a{>k-qH9LYl##Q$P6
zHPPs?h8ni8Mlj{b=t$7$5@+U!h~Q$XTj;(J>1lK!Ee+*BJ7mJ}Y^4seXhLAlr*!<*
zNoJAoFTaLt#N@9%d~bKjnvOU3yRWBYfDrizH|9d${bzso0af!tW?P7eo@zHYrcg2v
z<b*yI@m`8`rs$X|OLRSUgjwNpYe9LaJtj8cCSuS+{F|>iN`i28u%cI!kqFrpmxZ`E
zC6%xbQp8#^LhYufT3+d_0u{y`BcDaA!+fBG%>>x#M;B51;FgBqbAC!|{sS8TmW9Zi
zc~gujB!E<80k-+-kQwY3%8gf#jHiU5lYl@vn;DN^ozM*SnpvD+@*Q42I{Gc1UJa0G
z0Ob@Z0zxo!{j3xkfS<#jEE#bGuG_v-cH$_4>FwBnx_y6tB1jLXXyBHmYs9aQN}*Yp
zLHHE?NOo>Wmi58NfoI5bGoa%_!V84e`bn<T*U@2z#K?8j*+e1oA8g+XP$&XV2SbEX
zge|w}Lu*R${+=y_)L>@{jMd#KM)P+#-ia;mPs;DE!vw!c2b^~rFa2{Y5hjup9X3eX
zU;Na;#4JQ4^mh=RL`w`T*dXw`VY)gXhg8~lj!ecYSk}7jHC@`8#W}Kl*kN9RfCvcs
zxepG+pJN`#S_s4=C^rJzeE<-}9)o1&PPi5thkm`g4>(b%ek#*CS(|<iSe2;s4=h&K
z(LiKE1v)%zAGn5tYuB!o1L8vMPE@@?P;(QUphEJPX!-Dlw2%mQhdp`@2@^xsbRF4d
zQAvp>wsawU-#Pg3=U^~Vpw9-3;Hj!(27<?dPxp>T#!*y7`f4fC0_cAi>rJUTRf{2*
zD-jYz|E+`QxdP6lvo+6y$PZLbLW`d#055{6+9|~&NExHs*pLM$p6I`Rox(PrsNmJe
ztGw|Vn)RH6pl|^49)*CJS0QDdc|D0>L7Ff+(?ldp8^D3dEbqfTQDF%`rWzXsPeo+{
zGn{wfNRxln*GGy#6efyvm~}Xq@Cwu#-eZ~yCdTfjY>yy)!tGJP)9(w)hc?(+;}ARb
zVrNYsYl5gz1OhL?DfEyWA-}-RSkZWN9iIOwa%Kt!Y7=xfk-3>BjhWZW2518Z+m?y*
zOeuv*iNi;n`kXmm3ZjGdV*HI@0YX4dO15#?{#m1B3I3cFBWMxAA6x`+;i(-0%!g^P
z+sK-clfmoXAhb0=JH>P8$I`*bPdvhMIYt<jXNp3Ue6UGW!>;`ap3jGs^F$n$6RA{Z
z|8RTAs;llQ)4a|9=Y+s~toQ9l|7B_Zm%=&^6*D))|1~sSkbQ7NA6{O<9E!szTA4X7
z#kZS<J!YpV1}=7{I!t!!jD@K3LY5Owc@w3UXc1y1lgeUsndx|s+2L;sP^-~pVt|Y9
zz=Q2K-<6G6+8TPNYOM!YSbhRt&BQd`3;use7n;RfTxpq($+`A%+vHNO(z7BZ*Iy{#
z*;@QO_{FP$l{e+K{Ce~b`;sk2%17_mY*=zqPW&<ZuV)|0E$3kWyurZhto;jjk4wp|
z3B8j#Q##oRy$y~{mgznALrxwoVPjJx6FOn#31fvLSuIT1X5LgNZ~mX{Lcs@4{R@qE
z{&ix%JqBg#-@mNBKT15#fBX8d@GkKZ|LyDLPuu>LG|v3p<uynCl|MEA`klA`l}<kY
zv;XTZ@W6{9v_+TX=O5(c<b?QlWOn{9B>1P;;1W{9w`*6?$I&2BGymuiG^rzyV!Ui&
z$)va@u&f7MpdM_^zU8H7;m?zua_7)t*tdLHQ5r<~8l)1TcEmU#iRg6S3|rU7kB<-!
zfBu{EyCU<Z70vY<xIWr{iWy+?gEBUKdvYJVQ)nz+ebRQ738V~+j*gD2*R7(&NXa5U
zzXly`$ec959}0k`?KUnhE>=N39W*_w0IjM{)R*0@nIs39Kl`UM2vkb&6~m7yz)0L|
z85W;1kei#KDjitd2q8>FBH3HEHu0blXarR<nzBTLDM{KIxDx(UftqEYWTTqvY)8Tn
zxWy^}4u;Q9E2htZ@eu=6gTUfzP`86riALWN8yc%Z+8mf;bP9Aqn$)udp$+0d(%$H<
zeBud$3h3jz9Ih?-u<!Nwk4|r(YeKS<u>i$L$-g*2P!Mh!EEyi?10ly`PH{oz#Um;z
zS~9wUi?(ux?)<jgTn?(3jp+|MEPL+y`WlM-A_|NcV`bB`djTmW2j3b-W+B2Vk0f<j
z<2as7a;jxfkVQ*QG)+|}Z4%7lLS7t!7Ee<10JuPN5)lDjVidE&avcL3qzWbMYSeCN
zY!WV#W{N3}kp71>5a7UK0Ov7)dNQ5sY~F3M%sFdYJBfui&NQDC`Xjy$_@CmNJrDuR
z--j^p(^r`-mXZ*=re^gIAI@Mx*J+eoy#ZQLaNULIgW8EfYPt0B)gqv_N6_U}4Xlpz
zQ%JIflqaLzm#p3OkOC?dCv+>JD7-+Uf1S=JTvsiri&D$x_}8xH->UxM0?;6dD@cS^
zKy~lvjT<*e6=k%)H4*9XDdfL2kc<WyL%bqe@)Aq>sseedbOeqxMFwJ=Cy~mN+!cgh
z<RPD14S2zB=(7?vSCf(ja1d#<0?T0wRzbg_SLV)5=8nGHxM+SyTP^Meiq#Mh36&9R
z2eAVQG%gr}L?Q=K0YCZY5>}h68NiQXd(C()JkV{0S4WXtvdn%w2l`-aBn47|iZ(}K
z@nUk5;wxo6D-4@}a?xIu3A$w<X+>S5$eIp@NW0W_*;~1o02QXu#hA!)Eqw+cKs7YO
zNkRwQAkA72@D?z1En1O5`W^_S+55P}cVcWa0?!kG#Z%od1v4!OntXH?fXaV5^M|)2
z#23_7GeGxf;y9I{jWJ#|LSiU}WFP!K2|Kwj4Rk(|dZ0ssC#Y!Y0DazNoKmN;uIbu7
z-xlS&Z^wLdE8Ctkm-*Lu*6>M4G*qYIv5?w|h@Xp#OAkgm*&KDnSgbl^6FQaatQ4S=
zKynpG?Ts|kag?B`Zyh;^kCE_I?Gzx>GjV^2CWLSAZOw5~0fEW-sP+M85Wy+{CMCyd
zfI<*-{Cozaue$0tgC-o>m3*6PVWv8CypR(<L{VGQ4+GCNq19N2eL)4??c26>;)zZo
zmfi~yISg;u8sCSHq@_GRvGBuQ?2{-g9*s*7pIrE(hL;aHh?mX(_z?$R)OWgI9${{v
z!Vl{iFd~9}W{6B=#YrEyOS1n^J_?Fy1g|-W0i`oLqay<AA!38kZ@el_0G(?RFui3Y
z8G|ioUYvm0iVP#PXJkRO#D4_&&h`~6d=wer>p)=aSLHu`SWK9&R`PkJhNOUi0HhF?
zAP0nE>oHs=DGEVGoCau;<^`#E&Q1><Cm92iMkty>4(%5wi-L<R6N}rqGpYc`5bi<t
z-cf)4&OSMO^O~0>i?J}~{$rTf@?<7h%zA6py9bvc5(N3(I6z#-kxta1;_Qdv->7m>
zw&)DR0J?Cd>9@dY59*G^%?^%@m8-shs1}E;yOj1o$JG_G(<iV$1B<imdUjY^S_X{V
z!;np!^n9b<kE17JRC!3bjw#Ihsc4lh82d=$P`;@oNnNl--Z#r^+42j>TxCgN05?g?
zEU*}YAn0UBx>&fP4m^_IWPn+x7y<l3;Zl+NuvgKb<w>kF4P#V1{sF9;^_Sou*l(<V
z*?VhXRnvC(4p=wQ>A#-i_DIx#be?eY5aZB*6b`LmDb!zKtE@WNja;eOU~?Y#w+N<5
z$bDu+gC-^h7Srr$Bwd*~dAM2)h#U!6uAeym@ztNy)4>#QO1G#%H&7%>-#D7521iHz
z!BjyBH4?lS6V%!8)4ON2Asw?v^yh5+ib!zLQjo4e_H1S9VRH;>J5Xvcv~5Nj*flsy
z;}v_GGOHN9TRog1W=>i+un+1qYntu^yfr8t96$<6D$5`>z-tM+0UfY!gsaF?9GO7P
zy`Db|;;;Dw`F6k{BjHuF{gm$S`31hg=4l9II%)Hpl;Gr4peU@fn)|XObN68DjKCl_
z=xE@nk?0CNwQA^=`k{vv*qv`)ULNiz9L!S_SWfDRhY)tm)<ZY%ci)>g71Lhx@m+du
zb+szVw}Oq&94hkk#1<N)0zkSN8_TZk^-ob#*F8O}5Q~hwH}L6#THc++#6+A3noFk?
zD0~R|gqPgq2Qy&}Nm7}_`e}d>P~qanMs4W-#Xu3BT5WM&d1Pc{8r?osyW~3b>tURk
z7HOlR0h!U$6N}3ug=U(6ftt@#npfRu-Fo+4$4Qa{#6Wq3?9o8HF?r9|$Hxc65COmk
z#{rnU;O0~@_kH&_^kP`~PyfxBfhr3HA_A6mPZM!csI$iT!*7dK0jo|-PR0P><wesO
zDeNtSL;@a8$(=%FZT@wDl8(Z8r(p3You}Z3bf)1M;y_7~)b8B@sbUbfHbSn16opZa
z;=@D6lB+_Mg5>LHf{|9Pb82b*f}xvO|1Qu?G$3n{6t`*ZT0@7czk0q2hugj1osMeu
zzLEfb&8?nq&c)nn8pc2kgfI$$#aFIeQ3Dx}y~}p|>qqfQi^H(qr1S_E{}@&&8ger@
z+!~-wu|oO@4mlPO`2)BU1sRo#n9}n*tscFQ_W8(&8a}-m^T<d>7^BSUO&w68bOxAR
z4W_GF-071T>~b6HNpoYdy3y#Cp<&$#db!<g(l+sMYq6wIed6H(e>;6A1}lzenUxzR
zm`2OM5D7AEBtUqpw;NdNMWERmJ2V0bsA?Rx&sLHMdFje(HW<LqLYC&fLfbn9_$3-~
z44XoN3B0EsA88Of12YC!aXb6qCCR>PNd+^v&Aog!aAhjs*nO-DgK14cCy*$lg-h%q
zfg_4Au<Cf!xp|YI&+O~Vh*%2s%dI>mB$Cz45wm^Io;|xmU%q_F-;-`qwIkjgrWzmp
z#s+F=3V`6h^u=N$GjN=RSntKlAS&R;4oz*8Cn-<}p$OH;qrP30<^rfj$=u-iX?{5H
zXD&HFbPT$4%f}@ki`KQl3m&M@G%ra)9hK1++l7QGp5-<<CSn>SKN8ktRgB(HZF=$I
zg+C~;7$$Hk>7wNRvWQ_Yb=N2yIda5*3Jd$40?JIgfdM&W_yn(p9T%znZov<RvC<I+
zsO_*<nX`*oy?kT_&Um>Eq^E++fe4mbWn;eu-rNbvxIq*xH4(5IN<>1es|3Vse$2Lb
zmo#(z`j7ecC@vV|;{73Xt%Xf{A^kWEbO|Upid-Gg;47g4WtjlHmHIW1JMTsmAlfXi
z2Pk-7{><?3aLVh#!XVR~h$GrQemt8TLOMLx#$H*Yn3>Mj1E8+`gk)FY%JX`p>{x0K
z2Ax^Pt@`p#=*^a(hQLqBM3*BTAU~qPq+kiVVpT4<hrJhjP=Cin%hNzEzl^#-)KC+>
z@k%2mxK>cvlcD5QH|$R8B!e*V+7N_P9J><$=UQ#0k9K~OtPqq>TF-BSuPas8IV^+%
zq*1K2MzVwM(WLRIsccC(5}=MhS6nvEmkaZ%mAW6kAU}&9lUg&3m5Q4VMi822;~ETO
z?zt~D9j!s65W5eP+3xbZKwPCwL!q*J9i5ysudd*-II65%W}^=tM_5{O{PQF>P*c*o
zhfvE4>|QIN^Zuj3Gw5Vj32pf3zN=?ElS{Z@yt8$Edqfj>+UN2!aCH3GdmC%oBZ5h~
zHou+pARVp|KVY+i+gn9uW^JKe=)<J}+&2a0F?$4~25Ry=JV<SJ`PcyPQ)-~3ZP)Bf
zTI=BNFE!qf6{3U@!~n`grKhUk{zW0wfQ+W}5|{XZ9*rIaASooRt%cNsr1fbmAcjdq
zre$uADvJP{2c!WNWHl5~B*)R*ZKNv{azi-AzUPNUi{9qvbN3oWx;aQ<dhbUe3kbL=
zOnx1HH?6-Lg<le`RcI(D!9dDw5j;MDwA6!^KA18Z7eLIOUbZ+(ahz?^(CvNyJl`T0
z3YWQHNs-EC(#0Wkk&;|!EFw&#A|b*YCCG7VP@AEU8bL^8dWk30&P+4<fa33wPIftS
z<{^qoGC@56fWzY7+uVF}j+9$1Z9`%cPQD$rqY+l0ixhi^*GSs_5E%q1sBGAtEGdcQ
zfS2lqb`0f0)HV)aD&L+G&A}DWy#zdLB)-N4Hb)PZF4d&!FnV)@ISOI(hCvjMMuC2!
z&6b;Q^t>dfsn?X!tH8exVmS3~%vGToA*SoffImee+fq)gM=~ADpdh+NPY+|}V*$(4
zY&iQK(z^#M&>8iYq2YVGZN48ns9EAOb`X{4$2Bv5gFB%3=F_eM`8l6~lYJr*W80#E
zfr~fnZ)cKUCRpc#d<zBKP51lf2*JWhbQ5T`CUhz&=7!Q8rQ9TVsoqEvIff%XZ|6bu
zM!wCWDPxO=hes|1<4E-M60$A0%O2#iDtL(TX;cp6n6#13gHgYK*qc2t4f;aAkX)+l
z?0AF#25Mu|AZk<kHlLidYokv*`gQrpovQv+I1-vZ3p%hVO1(F|z11;5g(j<_)@4?7
zdka~6tr?n64zOT9hlg^gJW`@64BUK>rUO||rrfYYUo|!|#kYhgpj;aRzMEDqEYBg5
z@`MrZnkiZVsp50VnD5g@p_{_wQwYqLE_{#O8+9*Cq~V2IHY%`PC$=6u4A-4ad(r57
zuRAD3P#i*2j=Ripccf^UKSHvt*w<+m%@5-5Z!>|kGCiE++}vzRYKaS^6DE`?$OpTd
zsyPdikXW^{GMxu3#|7Jv3K_#f+j8lEXsBr(oCI{CQ+sV26qbI$(<eRR;NE>iw?_S`
zUg`SVVjSdn1a}V-JTTsvG!U(Fv4O!1izQPe840V+oUZ-*IXx2C&F8x13$arwfh(pP
zU2;c7OBav6{^pAMF?4-s?uK1E01tyLhltF{#U*MtmyGLaf@xw(bJ|cot$_aHmgC2d
zL&RPKS|nM{BQX?jLapV(hnK&RYuB0~Wi51tTX=~%Sj+l~_Fem8;Off;^JOrc--s<4
z16NHWMo@miyyXRp7B4O-ER?6{4x&!qqC2Ch!lZA>jCNme6!lOa=tF1C?dZZNjE(xc
z`H_~i)_%C<lF2mALYi5j4JqJd>0MBzQk%h&NHobqDReiC;Qpo^{QU1?+~JXWWo}B$
z7{r&#OxB%yei}N&XC@t=wDv<vwaQJ*6bYNjjH9zN9(Mg*#y;G28ifH#<h~>_2lUW#
zTIavz;MLQ3RQHYEfwGZaW6+0u;oOZ|x9oPgYlO+OafwW$N9Hs#M$6N((0v(1Jy_eE
z^1}Q{HZC`}o`Ld&b#Ti=merTPTflD7gSMR1dKw$TZrT9{aMeQ+N5>2n&GKS`3-tFz
zQi*_&pN5pgXhS{#3Jy%q%GDsLOoU+jrf5K)rfj+%L<#mcQO2Da7lwRFe7GeG`fxWH
z2<rTytX2vE)V29Tv$YeZ!+LCikPa`UffoF*-{L76-hevN*`kn&=KI*DJGjXmKCHE5
zy}-qx{;xn$;u``b9TLt;j><%+sHp4)gX~g$Ci}{jXU0T`7%7l@*v-vtH!W9v9X4cd
zk~;vqj=!h6m7yl-k8`ID(wMmuI2}vJ-U6)~q=9eMqextrHcp`g$?jeAG#)(sYmLP}
zjVACNT?d<zUNDA4MC%Rev>O6aXr5K9J@n7cB)7m-TdD~{#7%Vy=|W0_1sXf9+F{sF
zd^ehj;O7+~%va^|xxL*$6N%TyeSll8Jb_hFA3;iaJbx1A<9Mug#NlOC&mm9R#G3$?
z<%U{a7<08|wAtV9D_qsIBG5-RuXzROcgB$aS=J+QWnfi1x5L}<%)7IQsohF2=)-Ve
zc=(--2S}_IB}mfP`&$2(Pd@E{p~-z;v29sCB?tyG^URm#z-W)4G<p(V=WGZxRJBXH
zDcQRQCpcUUP|Xh2!wt1;kU&SX`hzrA3=Ft7^)}HRIeCf)&gRSNc0Xbez(bKjqj2+a
z<(xX)PaV=Y%!ctbnuCyCC7d7dh80IE;q*vV`Cdi#tQ6sOIB=EzsB%7q@u86ZYOf3U
zVF*IilTT8G^7y;u9lStu(lzy-9Aga_tpcb)1AYUGHR4V`LX?*M*ark8c|eK6I@HgA
z{6%5EMbil+*!wJ+E{H^GQ%{Si=R=w;8U;H*X-JlJ`&CMDkx>>ro5g-K_otI&Z)|A+
zNpQ3a#v!R9aZ-G;26!^ddVC<Y_-N<mPm%JRPTP8|#}+MFOY<T}K$tM4;2u*VO3l%c
zV1jJ3*0kWH2n>g5vedqP>NxHWVxz(5F5PN@6&)NHu-23%-A#0|MGj&oN{VKmbw;$c
zKUnj`U8XELx*<r8)@djbRG0lDea(%}+;KbDRPCyY2!(2(22SITfHx8Jd)2PJy&WTe
z3KSXOpsJvy@uNNR5fb*eG&}7C#E4{CY1sWo{(Rh5EsEG2t${kvh+kn=4d@=`l`=*g
zktlek#-<%md!%^xV6ve?CA!beGAw{0EKgn3VdOfQ?=d^X6_65S9c;XJ$lQG|7z2tT
zY*BUKi^c}t!x$Wn2S>$a+ue)U`>uHj$((cjBmiu9zs`CHNh9HBV$t?sfj&W-ZDpTg
zd)giuFa}s|>)&CPOFpED8Wye0*3UY23TljTtBq+5u1z|km6B%s7-?D}xELIsp3{B?
zXLxP$ZDN9cUg&85@J}oW{+s&k)M3e=;0~j9H%=ziUXLx0Ak+m5U{V5_H--uoCAlF(
zpr#>{5=0cWfZdat1oDn}C{`v1NCoS{A0}z`b*`^SlarTkoY^!;13CfNY5=zYlVT?<
z0f}bO5v~z`a3@#pdl&E=BH-2t4kOiSg#0u)*$q4DYU4BjdUX#j3nw6y^S%Hjc|!Dg
z=2PqH1uLW;BE_b$mHX1kYHO~(A!*lDc?iK-CH5+#zrX+MC={iy_8@}@`Kk1Y*SWVh
z%r1d_w(s6M!s&pwM>U3FaOlS@WYL|7Hcr7{Bl#5~H~=3-Gk9uRKX{?sAC82=h%Xc*
zKQkOM!Y-)DfEfYuG#rf6W7<039KchZq7mrERj{<?wRzy(P`@=@6GVjXrl+UVQ3AGh
z5W7iPp}M4~#V$<$56rqm>_AiHzwC(I$aDJH!?TqJdRd42*ODF963|L{s?e|-+W$34
zxm+5Ioxrlv0p{N~$_2Ol^vp^NU}+dXQ1tfi!G(R--hFg0+R)BjAm7*4&6N)IbxmY=
z`*g#;0X#Rq0X*@z+XmHKAP7%pB%uHfEJU=K!WH8fJ(A@H(<g-i^ao;`Z!XDBr~$~R
zn&nuv>a5sQY_d&z?mq4r#n)*s;zZlJhdL1@Doi<GxFSx1v4sB|rW)`9_-hzMWNDbJ
zhY_Uw>l7OJFmK`#SDzKyRNoX7H0T7Zc}_73Y#ty<FhJt<^ID?QV1u3_Net=%mbCZm
ztA#;5Ik~wb0Ae?}4|u?cLM*@fqa(OJagf92Q*nabg8b;ApvK3Mk-W%yWBBo(uUS0M
zv!`r2e*=v-Mro;02Of1}e(#_s5@Zvc1*jierlqAjc!!DG?#Moy8C+p3r>ql!q0Rg5
zO<<@-(#(QCwc&=0_g9#sSaMZ_8`*{y&Ut#K002OA;#rA>0xImjwZG+AW?4%INHx><
zqA%itu8zoJ7FGTq`Jk1(vl{J0pjE*LE6nT#0gNVa&@8>)Z7LB%Qi8e>4mUx?*W8`W
zduN}4<l*+a(<l^#f?uQIKO!Y94XOFD6b4?ADa9iu*ydgoJnbk;)w<ZJEy?}AMQq>_
zlCfE@v8~N`ug#|;9LP2<xfRHhuNvf@ufaC?W0b_Qyp)n%Vq))2>%Xkmr-~y)(7W!_
zV2RQGj0`aj8Q1s2X%kqP_{v$gk#02reTw+z6m-r@CW~&3HN*|~hDkY9R!1b*wn-fe
z6FULG`x>ViT<tHh3md;|>-v0%ZB1O|(9lrH9T{l4sy;mMn<4{X@1BwS#@rCf7t*PN
zWI^n<yIxe^pz$bHGhmN$+}w`y3=<)&W7)rIm@P36D8L3i5v-z!V&&~wX-nL@xotDJ
z@Xeure~81@APIYT^XHQJ4_CQ#WW=+TG{5PTD2+B7MhzorVuHBX5`LX$`!~iT2nDbD
zY(@*}ipcf$>GXGs^3F(><izJ4x@%C^iR_jMnZ{|OwiYWVt6k~g{v_b;-E;T9`iB8y
z{}uG}lem;+VeiiPR5*nOBveQqD)^KMfIZ`JvseF<KuwpSD?pDQBm@ogzeJr<heJ6u
ze`e*_!LoC%ug8mYy}XY*V>ST|MTVxt6I?z`d!Tih-eQH-J(zr1WZsT*{bR=|ZL6{q
zaer$t{J!1`$AJA<<dL}yi>}tda0>^Z-Y2kJl{jTnmNcvIufP84#_SNPJVE<e6Pa;+
zA9qx14-i~rdP;N$fSlA4bi{$0C<1vN{QC8oGYtS$Zw#CW-?J0AlaJ*r1Z6)l7=EJ^
z_cxSORaJG{naNh}yoXfh?WK!Ofw_SS97ijy<lrGOudQ5>LHQF5<K15S-t7ksN1k}7
z+p#O9P`=gN-qGK0nIX`Av6Zx`<2G?}hIToZjDyT}uhblIryWpO{?Tn7M-a%Me)$MR
zh`%58e%am~H1n=x+2J}9W!(rs!W+@m6*-79FNFSgH1>XeVy1exS#H0EZUpGyC4(C|
zL`{7^jv`J-PhLDx0)JE;rzvu64acp7svJZOJlOZ~{<@+n7Ygi;`R@97*S<L(C5p%%
z8`;-0w|6(Ks4pDf)g$5MV{;f0Npf5=hPohV(8hwI5TK4K;o5cU&X4K9`QGZ<St#@$
zkI7K*yWM=}&kx;w=k|-A9{F5<SPk|D8Cfjh<!{bP)35OL>kB}fNG`k^#5e*15VBD^
zDD3t&&@b3EaX&=h5ee&4;hD@XwCC}FkC8<l*&d!-A82Vk;&f?d(k@o8Z}&P+poNus
zi1kephkY9NNvunFz){)!C1j^}sI526jt1iA8;x+Y!hX%0UQ7wzr(?W>XD1j1R?XOm
z95<gJsJ5?@yvyHqq1F4D!@DLUzz-&Mdf1E!I+@{!pfAi$PXx}MKhH!+WIb9g=_B~R
z+H)}^tnb{Vkk6k)8<55XiHLwB8&&Tfeh9v5`&1oI-P7lYAGF;L7rc2x6}W~Rr(_%r
ze;(`oxyYxlt*P_+_F#ndq9FY*V6wk`X}9;b%FjmLUPhT*lO9bgW@MQu=CELae5PQm
zSu)}s61~2dK=~p>vo0oNyG|r^v30UBi}FQ2^L4+=7lobKV^<}=ZAlM8`O*{{9u>J;
zC|{ZlDKKE*;9E#A{{$W?X@TvRGwnrYtpXF5H1i~Rxc`flCKmbAJARk?F2THA7*zZ*
zD_Ak*$7RcE_!4vOMo)vCOMN)rBSG3p&KI)+<@Gv#5c%nsKG?bMSTVE#H3vMfTzYuu
zE=1r7U=5<i7qe#Y>G^h}3AdWguT@aiiy-V9y<L%m+R1lYmw(ZX(uw#CvHT+lR7gD^
z4LpyL|52cGb!0vN=`+sjj=UY0C{?KI3<(ERpk4xEpmkO$84@w8&F1yM;Z$TN{J0J3
z8;x|jc47GtLdq1#i(eVQKqsU#*>xQUS4bOXB!G}in8?RT)LB)8zD1AUtPV^4r_P9X
zA~FRpfSe>OEqwHv%fS(P1F1L@x|ueI0RO8YKt`N1pePbTV=Q4?FgeGK0n_C6&I+hb
zytfJ3afack`7rIGR`YPsKU@F+Zj`_Hfi5AM+yH6=dtXUS2hHM(dm-lz{=aZ3OGgn6
zL|NXiqPG^===cWHWc?qFiOO(MuOD$#0>CQ{i}HHSwlAd_=M#H_{qKof&5{h<A&G|R
zSS;FxggKD)mC}tY)q8;f$iFH+dHDIKVbKOue}E37Xpd|q72xW%#`cYNI_^1#QBV=Z
zl#=`r)CCFhQcO3-=6+%}tc>J}JC*@|aek_6T{;*E`KIcP2#@PS%%vrp>;yj)o%WyF
z(s%75LM^QH+&Nd@DJ0;k2t%@ak{eo%;3!3Jn5T{W?;R@PF`IMnCqLv%0W~%d3NLl;
z;_~Y$;@3x%w@<e&#MRA>hM{1#(65wNB50y~3M~qcYk+-kMCJZtgsCkpAHccN6b547
zPM8fE?4*{cCrtxn!2c;57Hjz;^?P$o#~lo#f~d98={aR}%xnx|fB^rt!rhaV@4W3{
znyhAIwocB)r(qT;d{UD;Y7O<Kl2P?NQxIyi;|=@_t(&4Z($*oU0hC*XaQhrXsI*CU
zeV-K)+xPtCT`<O2cf;gqpiOoe@sa}@-9X3<&N|$?eSe41C?MBqyN$@U@JaPL0Zh+{
zs-qpk`KaAw6yZpeQ`s=1Ds>taM7CNb55diWd0bX?xi2v(z&uM}?BE>YosDA$M}=f2
z*SB+z(RAo#bw+m04cc7n?7>E@-%uru`4ahoeVY91AVz#YA?!uMd>QwAvfgC<A6_pm
zCozy`@7LWg$OGGqgB(d50-$~-42ak?Ky+RE9Ok*1HRK@yh=m(}&>$e$|BG_4ypPP}
zg@!zj3C8+sWf6__NIK3WPXoMrH}C@EHlHF_JkPs-=*XDYm(tK2<p0^pPMmP?IHYgY
z8^ufYT2V%{9Y0NlbMT?65W9C68ul4j`ZuEz`oN;FlU)$T29?ndRyyKuIyc_(@=AGd
zcHd4Z(D9uKGxE!x+$as09ON0x3>k58O{Bc#c<WxYpqv6ka4`XVp&**{bixINe9W8o
zMsB0CDpcE0bWK`)PjySj3E&ItH-AbXSUZ-1Cc^VB>pmk1)-HAY($y7>qSA)ia<Sy6
z_13}JeT&%H5{=o1Hg2D?mK_$vCeN<wyT*(%qtGU_F@aE?rFC>aD7QS%EGsN5R6Qm7
z?;!MtKOn9U7(-+fB9)nEf+HVzo;7>>jKS$r_h>f|X|AnSs4ABU2;dx|!4KNIk%qn6
zZT;9oF5%>$>tPhV>QJCct_edP3b+h83Yap;w2r2=Wi@ZuKFcoUxTkPxbYz6&8g(kw
zVR=6F_99LM#7u#3P!EVa!+@PO%KVTGpUVV;&%N_T&Y*xtpO)DNfR5_5(ejRAvL1S5
zVla5@F<LL*hn%%R4o6Yagfmd@<Xl0vWo*K7!1XGzc8ZoXOPf&2t34%te8ySv@R;xU
zX-b}ybno2Q84C{Xx8oa?P!{szI*#6hzrH5duL6Fe4xDdU-HHu+UVbfHgVhHg>2axD
zXlN*vwyOrj+hIA=HA4Y!qoUn~OiDA&8Yny!Xc3L8qV|gb(&+Slaq;@E)3GQbr;Tw(
z5R`$nrrI(XbxP*8Xs7@vUu2bw(_eED<Tc%hmX?-n9GORfM8vPY0`sFme1m$^Sd<Uc
z!wxYE^XhRIPRq{ZUN)R{KZ8RWX;3V-DP3f20ag7{35G@jRpmc3GegDfAiFBAEASzJ
zvbU6sf-XWs+SaU^EAcCQ>6gKk)UTYwMZxc`h^CEPM^7OgUuv9+_&pud{w!>pLF8Cg
zsERFl3PX32z6AYyVrun@3LVm>W0X}*hUcMQ<f>iO&h4{U9jXLJk8*tM?++X8zkmP!
z#nMZDYow&4;{0c(r<+#GKl=4gWG{oCj;b_zp{KRcN6}%#sT$Z;tf4>P`alUE3kMMH
zQ#Co)tT7Pt4B#8-tjOQ4e|PbsMfJcdz$%M|pKHMMa-!=I8x4k*oHC5CVMa5YQLuNZ
zCYt#pKT?kkjU1wyxHr3aII3H`JUmB%O8TH5FrGj?a*!~k@hDPsVfivQWV*rD5aCJ<
z<``3)<cYNfTSTLB={zB(BYbTZaXr=cOwkM!`8dui)UoZSsu{d<`M#1{2|qbxBJc-5
z2+9f2dYha%*!XD3(g>K8w2aWON9oDV2rW#bL8(3lb6I;^^Yl^cs{op7p3#iLtbxSa
z{ZVMIc;77|oQsQ$1_tmz*ytwUz9{5cJUD0oW2;duPsK*Z5*@_mtp$)1#wUASh>ucV
zOtC=B2;vzJB-3mk08s`eLaoZ6Z$TmbDE&R&4p5OLkq2nb(Agkh&HoR3?*Wu$mvxE0
ze&(DN6jaQjC`lwJD54kuB?}@b2q=hxfC>`KBGEvSEIA`d$(R5o2LS~M5(NY#XPC9Y
z{_pL&b*E~kYO3bm>94xF+fTytp68r>_FjAKwRrb?^Kd5}MK0%qAsj7T?^=&LcNl0?
zmWE!zbmTB7FswSwNGnP)rB~vNLmdicQ$YtGM00#fUB6ZTe*G2$aFsfI9pV1ztDmyC
z+X?ys-e?6OFFU3hqBs%Dj)BOeb<09~^qKQZ4Nf#l9&7m<H*)E3L1vV^ymKm)Z3ELO
z6~Yyd+CWx_0v?9Fd<d13AOQJlvmVK0kz;(fh5u&!e*rdz6v`DMCjPdNK49D`7pC-+
zDp^dqQ1r|xMnGA({^XVDGrjm2x)?rQlyMF_GS0GQIEH5%68odx=KU{!+A6jHS=0ga
zrJ#8<wz)B`*TJe9?nv_7b{v4vvbo=9u0!_$xBFK}?v*%4p(O*F3sb(jlQewz$mE)!
z?AKD@xt7|hv}l*)gu&+Hu0-sq>TaYlO}8cxQwX&@S+NEtk5v4ZgEiAkOjjUm4V|4)
zNYg9LCupz-cU8h>K5e7l--@pK{=qyM^4LA76_p?h0mxMvG)?kr!3duu_6Dhz?mqgP
z32YVW5+O61<nDsjLuJjuyZ@k%XN6GfiwLGEkaH1q3W;rUYALs*kvBfhGj7p-mOGz7
z4Ya!5+pgEIaq@@mSult<Qg2~TNK(7+^v7`he#Ko()hYm1>K}uIys!;!+X+|Rm}>(s
z(*XtMQCoqBfFCZF7Vt{sfIXo)Si!;)PL&ky=M-8fWUgCQZ@TdVpiMlw@YLwS+DRk%
zLbBCZU+)J#EFBjrG@V68p`l{cBK%gbQY&_U6o45w4E-FUR^M5v6DMSV5$^?^beFcA
z1NxVs_x1s2P<ASSFBgGA$$H3Vd(klR=dc7dgy^pq#we2qaJVS1_cxXul}R9nkoz|Q
z@Kwyw2YA$m-xa8+YO+P|JFd~I&((16M2xQ@Sy?1&T;b`!c^+H(yNz2M%T8=i0J!(%
z-y6d5?Pv(wI{eTBl&)a8yqJn3;ueD={RMqE)D@=*)?RK;FA-Rld#lU8huhDA7<^C!
z(pv^nq&yla<O&4+5gC*()SpGGlKRECDKAy<QR#w#=)f$^Db)JA5hSd9(H(C=a4E$P
zAzwHUBmK9WOT)PZW_%hBdlFe*T{eY)Dv1pM53jwW-MfT{3XHu6Na_rF5#>0!<j_ZJ
zjZPSNe&o-?YdLNPE@q|9sv6*E5(qa2O1tcT;U=j_M;%O<YhP!L8i=I|NS4X(7lUsc
zswro1_D;HYwS@$ET~CPe5Oe_RDF7X+K#mDSF|b&p!PbZ(+8H;u49E(Xc`V1lC4ot5
zIdxmfH4jbfp&NVxx5+Xq_J?ZBi=8lJQ$Q{lT+&WmSE?ItFLEWO6%I^tLv%yrYrLx}
zEM~`|O$F?LY(aW7G%<A>xkdsk-$#>GAqz^@eWgDI^0p6sxg?^(0l!A*klaXOMTZAR
zg*~U?BoFbd_oNMql0CA=)tr^prfnQyV)7Tu!1ySM{q7AeA;9-HiTSGB8?KAPrHCA4
z12{k=z6@ZSaOtfQu08+_TXE-*PbS)o@c>t;Q;!ua!X7Ms(}NZju3<@sl_lFaL2$#`
ztAd?Sio81lg(c8g9qeyYz}eI~pjzG(pd?Gh+unicRGVN=!D#yD9|w18ec+NI^L0Em
z6{rD-6D`QyhJoMr!Ag%DcCBq}PF8B6Nqj@34M_yyM{$j~E{LcH_BGMlY3uy@b%~Az
zp;^S0#FIR>KZF!F+@h8bNHDQSSMOHQ-~Rr<&L4|4?M7aOa|vhLf(_Bt-%nO?zsI+K
zvqkZTR^a}U<<qWQ_#fqgNeslI#};NjLFM2;7YLb?t<zT56EWnwE+>IpCT4JO&{e0T
zva&J;Ktc|6eg##5=<F>WM;B0XeU2@Ni<0Zzd!x`C$hwX#`SFHfeJUD3sfW27*6z#a
z1{+S698-yAlasS2-6LY);Bzzx*dOTyr3-ISBFve6t2ndgVKB=+E8Ipc5&hi@<jpN?
zZ13gPqU_+@uZ7>3cYOpks+-_Ac4Kx-1eWIv*pHD{(zB3$q&MOe$5e{LB=KiIxD>I;
zdPEF*YiDJ}`2#X8hR(PIv~Q<xR)QeIF2tSKku-2Xiu6Z*duiCOn=P)2g^wAznd2Pf
zWty#ZmJrK$^Q)|WM7i1pa^UOlHF!|Tw=cS`$qkpI;n99|?b@}vbLY*=?jEB###q~W
zbdLf>uIZ%$eY>CprBUZG(1MF!FT8E(?Z6_So^TGu(ZaVNyuGG;1VplD?dCoG{clJ<
z*BN3XO#Lr_86=EhjM=5pfGelkK)<08bTeiEe_QtMEj6kkUTxkRl=d#|V_epIcJ10#
zsh1>^*(H<df(1)H97i)6Ex6?06xe{PdNyse`{d&hb4)c$3;=AC&B5`A$i`gyg|}mX
zDBN`rO{x_tL14|$8N=YrF2u2@7O14Q>r4{8CP1B1-=P@k(Q<>>W=b$1m;a8o;_rTl
zjOfei?SUXf+rC`Jz~Q%@W2XUpgtvm@@l{UZ`}@Jq!x@9yZ)S=!!&$NrtN1)(jI4AI
z4{!-!A0LP68|ROk;*rt>-oJhLx6#b-IE|DWf-};%d!o-q0iID#xfM-(htMiP<5tqr
zpx#-PDRxlt+Zu~7fa>n{L(ZVYNFbFryp`<H4^A;TkRs$6>#`uHm&aC!30DNEH_t;r
zmJ~tXjAV!OU{d%bn)HQ$bpnvKy|fQPtGyLgctlt*bsr_+PZ?A(JWgsL`XE@CM}9hu
z{&@*cuvP#PY$CzYPzT&lUsmtt<UEa^>`Wm6ZLj3bs}wQVQ3ODc7dCm1wXMQs0<1x*
z^F`pf43)6&UXm%uAih`Z-u46Udl(I3BX~JJ)>@j5q?jmOAwNn=WD(hJ2u6DF9|b;4
zY#hDFv(f4gb4Fq_YHHL_qFq0J7M~iB#F#%`A3Iu3y*Kcst>BU%BW*5x!Fk;VY$UL*
z@E<;+ZY|OS2iG6zE{^+nF~-n=zkj)s@zf<!8d)MC>@*aCTud^LVPaEcG6qN91or*>
zpg){MAS**v%zZ#F)9as^^MXObfv4}KH(HBTlZ-+_+9972iIXZYYEq6)QQUVsPwBWz
z(x{1#ARCT5TYw)yr~se<iF4_w<4VD=4ENe_aq<yiKZg!|)~kdqmmvpBb;!rcMqV_B
zt3n12+d7P6H|Yup2q&R-rI7=0mfsnPV0vnDdI_wkj8`K}e_KJZDP1bcw68vA5eN;b
z!;@-eT@UGwLd!^uW{{9U5kDJ>E~1n^eN+pjZY_ZJ5{Lmz(vU%sEW;&FNfsCqt!Nlf
z79{U~?AetJWLlcQBM%~FpWKAF*VqLGm2tc7zCEKG@&;s28<0w7$Jl=@U`P|)2w9kX
z$}NLQV`vOUfr|ihL3x+n_Hb;W%V2n!LdIz?(g5%S4`cWZ`R8-=L*!6D9e~;kT}pVG
z-$s@K#_NvD;8{z-b|=?g?9EhEn~}(SiQ(`(LtPp_jp{CP>!zioYcqrrPauvo4M45}
zvO}m!K#HzZdBV19mjvB(pxu<{fz?Eol4aNfeNbzlMsYFLFckL{9JxPZS0Wys#;6Bn
zS*TpXz=MNghT(g!1{PfHly;H_BzR<@2-4ee5`lh`*(=Bi))-MBzgDE&*b;Kk=z+id
zff94#_G%?!r<T@tc<#lOK~p_sDvXV44EzLglsiC@0OnkZThs{zO@y~FqOikTol<bb
zL^3cnMXeuO+KCs24nm(qBe?9+09qyEpPkLxo{HS_2u}QI81~5zAz>hYMH7?8QOnDa
zFR`IB2z%!iqqBez)(D#=5aS)_39O>^z*Q;)*0N&xa=OMSSBG>h2zigznfFRlK7pd~
z1{NS3Kz0Z^C{(*cAWjIw5wgS{K+0^4{yTVSM5LjXHf>DGB_XHE<GEkQX#99FF2~=Y
zRj2X1NZUi=bbBEaOC=ihDTru-1prdyf>i>;DV%B$?3*NnY-07q>y)0rkupxGA?r!x
zo-eaO-wuI)Mdr-gvlKCUa~E<;uc+UF7)s3wc;QCi*7=3(1=9p_c%dDs1FH0MpgX02
znZ%eXZ*aru4k7$!xVPjKO>KDsKO$NSwavkSqns3r<lNi2*nkCcZh@!}e<2_tl}#qI
zY^t%rCVO=QI~hMg<T4@v<(@cRkw~5?2;;rU6PAvezYU9K7+i(E6^*07m-QJ07k9Q-
zB`bo7OBr_J^+u|eu*vr2h^cNugQFN*kj5JcP^XYgev;I%ylA`CK%`;LWP|0?;4f#O
zGvAjnF$IHRM+FtpW8+}gOZ`ctS%pab<q)jkiz$PTqX4i1D!Qix_8<LAuTSrJozUuj
zgQCjN2fP@7>Akz!=x+PDC$Dn`92ydE6>U3Cn}R}blhxDp2J2obVipP$F?_N^NS$~w
zEOsUSB``1YCVg4aheQrpl;;T_(eYwfN)sLTWi~i(dYB5SShzHC@-c80wb<d@;SR-@
zi=hi|viczt(M6LD<fcfK5Uh8@ne@+R>2g7RsL4J-G-5>iM7E9PYk0J5NeM@yF<({f
zD39^uhYu4kYF*Su0#nc-;W9oxjuIs=!>X3#Q;DH;IR~5~6#A}fH;Gz&8av(3DSQvu
z)*eFbFt2dlunNDNCNos*CoS&Qats*Vq>&{2V~^U>p#cd?A7)0m<0G(9i8>py9~cQZ
z0nuTHkF2Av2+xH?921j`my48utHdT+;%WAIK5fs};00&S#8hiMDhCkA2Jw+2)iFd^
z(gf09h9s_PiJdMmmrH<Ou)7_%zrxjBDz_HAvix49OOEg%`8>h2{?ODLW0<i19R#ek
zF9G?X3T%&;*`V)gWMpMwD7WDaxH#w1EviCuu?N9)cL8I?-S@gDu&(dB){9uAwDA2f
zu_cbRF*qEuLx_~UxHOwC)qGxyd?pSQD~U!Nwz#1Ip>#;RJ^=@r+iCGyJ_(R2u3EvO
zwj7Yk6jc0q<i95NSTInGgbTCB_*A`6KuW-~@SSz&Ux$!Gr0jiS^~!V+fz-p}1eJ4n
z6nH#U_Q?)Zhyc>yb`pAi#(q>6Uqj?C`R_EJ2~;~_$HsDiI{2zHf!)Z)A1iH`#<;;H
zoD@5S?A%Fl_&d(7FHV|DM9l~>tNZi>^@`zPvCops%k5z#5C{|49Lb^tivRF;z9Qe$
z+#G~*S?DW00!}^b>i<YIg7~#C_E7aSZZAt5LK&21fF^&ec-97yuwy7F^W=nq8i84q
z#I})f(Bl2zul&qG+A@r8;lI=J2fd5|nWKv#8-=^&apeRM2JU!6kkQ;@3x4)51zjAt
zi$BM0Q&S!EXo(Z2Ar}co>IaGDK02>BTP1o8P8vi@|DIBfP~#b9uwg(`LeLZ^Nj^ko
z%f2J^%J>;^CaO@WEqZkW8OqgRo?`XrO90>m;=~zuejqmFXZ-8Q_S~G*D#TMZ1Olsr
zZdjs_?S4Cr88<6|=q4Y=UwoFGYIPl~&YGH6pwINLe`;rFO-=aE4hO-nimjH9Y*OL8
z{0kAP9J{BZgY#*DfymT}xvH4^f57bjzt)ES_dn_hv<m<G%RBtJ4tps4(;bLpCInJF
z9g9d><(=L|G)`}nZ$l6U9GbWsN+fq)pd@JOAnzwe1KU(@^7Um<$T}90kia|Etz2Bv
zxK0gkl4&aaQ_FekzmD0|zoGvgV*VSh|F3_=_@5Bg_+$CI%H*kh<4ci0%{T7r&9~zy
z{Jin(+${+{zK3R%&w4U%4sS-&veiakMW1hf!7XR|B3Fz@Dl=Hxw%*oiWEtz%L(<bf
zxfZ?Oa>+M%hUljI%Z~h>C8jhoKJKa(Zt(l|ZGkHD+e2od)t33m9*@r|+$|WZHk^9k
zV?O61;fls^TTtDE8I=u;<6?*8%76*@p~g<!J>BQKn~0?av{JiwmV+a`2JP59=_6pY
z3X#gaJG7{D%TdMt=7;j^_0vla<`hP$rK`B$C-O$i33^I3Crw_sP3jm>=t62@-dKT@
zLLHZbKq&s^x;FBk(nnFG^u~`ih2d!uy*L=#8<wJ?IQ4{+|6TZe%FpfZzxw};O3r_z
zpeFTCr(WxSOiBMwFeUXbZZQ%pM+iZP2!tUB88@5_i-*RrS{<POiR^e3BH{?xy73c0
zacp%X69e=2{=+yZdpxLQQtA+;ldeiKA`xJM#c4NE(GM;HF?-DbByqCBqSDqto1rfh
zKJE;khlBFTQ-P%Lo7926uY;X168HwumehXy1|bjt&$iv3#yDu`d?KI@3>UkzR}%xO
zwavh)>OwQSJfQ>X$jd*IZ(_xR6qVKV)P>(56RHmGo}Gm^Iu>M~v}Q~gA@M74DYph;
zj6Ga?K0;sF;5MmkR=Ab-$ghi-aCmki;O|#LM|w4i+ACjglUg8j`~0furt(RgQk<-K
z09-5+6;OHbz@^Zjh4}P`UHSBs{k%j_4nrgH>et)!3wT#g29Sg9ry=|dj7~4EXE9`*
z$!EaN6Lx+JBcd(k_)+s6JwQTIJ_1uFofWp33p{U~fN~T^9^k??82h;B=Wz}{12>+o
z<D}%F+onk`DW5zNStM5?>HMp|71LqvsBtNL{c2|?rz03S@bjSaTgWmMo%J$t6D&t@
zapjL8@Pl=L#xMOboW}5p&Vz2DOo*JH2i-M=&Qtqb?}WyK!!x89j7uTthb_k}(3L_S
z3{k1039-<WOiT_MZktRs<8}lD1QY?tn<YddK!H-873Z7r+GKTqt!D=Pa4%XM<v!Xp
zYQcN+?~*)saaj!&m&>nvPMkbRQnU+iK#!d_9_$e@6mV;Q@4xn(uKJ3Ne?REu&rhx1
z!M}{-9O@C<HG4x3<+tKp+A?wZ9?GX5ed+k5bKV`k=7f5C8vx2%g@XVKfnfEOu@X;n
zb?;Y)g>)Gn3y@iOeHA_b$&FLbe;nZbNwZ8S_K^GQv``of{cwXE`EP*1PYSxWl2fEl
z#FSx6g+C+=w{<c9vyH)sTLlb>3-4S^o5U8OZdrT~zY2e`>+i>enV1hhW`^2Q)7eCK
zVsSk-j}=`wJo<+&QlEgoKQests~3=|-o9(sm3KP0=HKQ+rXKq!a$p>V<KFXo@x=^|
zDdV3)t-o(8bpZQLSLXETC7PO=eHB8epqSt^GF_4CfWqs4U!nrs0w89))VNv^0`5R}
zbskvLZ)hE<KP;k5Us@qJwZf8Nr*2&Yzqkk>%h6z*sq0~Jal9BnR0=u(T_-N!M@K0s
zVEAbMJqB^8?~meTwZ&!0GNv&uMRa#}lklc-g{-pj8Z;%Z`Qvvn7|OH${syuE0GFfG
z)Eq!eJ>)g6eVTk2=y_KMx6#5W|7%}|<-CZ}hSqI}T<Fp;dLIDKZeil~wSq2`hJC}s
z!=LS(eCy+X-#X3aP?;JQZ#-8Y=iuaCJg{=aTAU9kweGt<xrTY(Q%AmujLSfK(oxxW
z>rX8m$mMcai$r56-+$HdT+cy2dUWyRW8Vor=h}c#T)lfIuy6x9J@-GMD5rC`Tz+z?
zQ>KoW*e47fDAnth>yFGrqI+qf&hSrl#11f^`J|_QyZhvCKPdJ$s$>q4+Zz%tx7Qm{
z14f{ADuiR0IaHCNM={MO;*v|-KMXf_8vG(xh&K4w++hh056|US#Cr$ysJUEWdf|1<
zWp3XMkr@>i={*=OCVQi%qy=}h!ca34mEWS>6wv^Cil+;t=_Vg`e2{Yb2J$qE%wb$o
zkU`Y<Y}?3idjTkmtdR`Q(GwiB;X-5Q7Wk#t&mB#L9?cf2>;L{$3>qo?m^;Fs4;ps=
zCnfw(+WX(jmkW27lgkDRRGGUl)>fKu6`c<?djw=m8?oa=Z4?Gylgt#u(Z|PbUx%Ko
z&OO9pX!=G0h0Z{Dj*Scn4IPeI4T*<*90OaZC=Eh+R~_hcG`eIY_e^8iPr$d>E8U3i
zJbi%9uuldVe#C(uv2kNR8XLcW_s6$kMw7__PCAozOn~J0G@sx*7`#N1C*MhQb;*bh
z5=#LU#<-#Ez-3{>*QMS1_3AMh_e!elz}Jg4G$!GH?mQ}kV9sF2&Gfb89J#-@@GxD8
zDlQZw4sQAaTS|n!475ko(f>1mD@J3|X8cvfhRG)^&4$J>aN8_2`ZN&A5_c15bmu5}
z11A8EuLl*w8h2@95(|CPlBsXn38ion7f&0h>TF6{fxlt6y#vNp4~A*~EDAbqwiI&;
zxBAP3kehqm^OL*37}H}jc<+8(B(%#5{T<vi`IG-HZW_J8gEso_rp*>(ce0jaUkNe`
zeGDeBS5z#98Ye9eaY5)9=VKg49efEp6%P7o$A4VU;ZW<pho%h~`52&Ql7#?U2O^gP
z7>}i`9`xH_qE?uYf#7;jEH1~GUafIF5+gEOyz(P~s~Rs5m$wjEjduC1AKO0ObZvZV
zTQyS7EV65ZBb_ynx35FhWQ|f>T+ELa7TqJ{_)4DBM4)5exl;r8c{C_1WJ@q9nT`KH
zJkF&n2aomTTP3V;sb&CfulKpX$T~DM^c%X^<XDd%cIuOc!+}-u^76*Q$$<VyewzHB
z)zMugr?biZ@T-swSN>8AWIaJsgY^69_jSR0O&n6ouo@5|kAgGkpZJ%w2-XaM|6xbs
z+N3~Et~^NB1VGMU!EfmTRQm~8_N2r;EWI{6?>-0NkKe&P2e3AGoC88baX2PV2dOEe
zC2OQ&n75Z6E1~^OB^fy=4V;2FH*U~%7y;5x?Wpk>UI}5nOQf_+qW`~e4`T~&!H=P1
zLW=BJ;8e0UHZ~dy?{hjqMs6CArGxW|_$Ri<EXzQMum}2b6!bMnLAU2v6vEMmECixK
zNjn2cNx%)<B#S<XyQI~urs=CLlH+5$7QRzdF)BLJV#T@OBj*CYxmoB!ovp7aLb@L>
zsEho%g<jpQYEWU?<TQ}P4X+>nxxN^*&&a;?op=q!<l!<5`-reUb5a+uZAKsRXU?S6
z#xUX0(T_+z@c>dP<(i3i(-%wBz^WmBctNZHtVvsYhsc$!g_xEt>i~ZoD}X&mD&kAS
zX!->_(Z@?|lfLic?N3iSIXMYTj7~l$hdpA8=9m!mZX!zsh(DE%#@*g=Pc>CluY4H(
z$l6I3MPV#RrL2c635>-2Nr-QJz;a?>O=mi`Z!g>$&IXBS<FX$<WjQSCIqor04ilKu
zLAUwjqw;7yY%{5&04w*|cNBgVrQqdWI)Z`gSEOxj*G4Xv%H9IFYDqzz-f$7FaOKXE
z0WxF=I2qQubM{2_rCq`KnzRNZv0c*qt{?WZ7i3q3Hv`?}cDl2nT279LD%ZbkdFJh1
zv!t;s7>{TGQcxDUlYVh=aZ<N$!(;Dl)s3^Gh*mb>3i8Q3#1kDfd;_8pxh4T;=4(rW
ztoadqq=@`0SI4Z$vbO87$S51~X}S6e#Ku8bhq9zYwOhxyUu3LABm+IAb*P#ol(i!E
z2B|`$;^5)Q{8rf$W!n~2c$<amE;`U#Na>axd3g$9x`zdDcwb}?$<YH|&jMxnOefFA
z@n)MKbQ>n^3k2IKZKOZx7A7H~C$HT4HjA|~h`{wAy4q1bUL=4GYE!Oed0>%1!_!%z
zzz|UoklD;2yi?EBCqV$@YeHaR-%fkDFi~8Hu<)2jr-#UNSnq9|M>Z8TBh>@X9Tm0n
zLduc+6c$ouc}JlIAuP%ONw-E!J{>hOxcQu;>pEeKw-#0hf;`f6EMnQn{P*aUksCEn
zz8Wl8<4xUe9NvM8giM{$_x-V<8C2j+cz!074GXW6|5Rbjq>(K4p4K^NEdB~N_iFkU
zl>k48*3^kdYp;IhcR2wfiHPp8yoi5$C~`kRk6iN{d1PvOpf`K=2P8O@(l?vJd`=zH
zfkg;T-(Y1wsq+ppQ5Jel@PI?uaO*XGJyCgN^*Kj_-IZ<E)Ns~Z_nTZNBS`Y_#u}4s
zJgg4IU<k)Su8wEW_J_yu+p_pk7$lgZ=-OC63LVf%D+)4UglVK7SejfI6&=jGX7R$H
zfZ{7$3qy>l1U=XmgVCw*9(-=4>E#g-YFw0dBcqQugd3VlFNS(667g{|jLXlhhA5gH
zZaNx5r;+g@3_xz@;IISDC<|VFOaw*foI9!eJ0ACkndF)~QDodvyz~OoqYF$&m)j$N
zzsXGZv$3&hGwZM%Z?@DrU<eqKd7#>;L{e#hB3BDKUOEG4a`L8P<JsQEkl53%H1H8j
z?Yxfw7A0FnDHmtwK(N3jjEHiMj#dk*oABGz)HlMDg{F35kwU9t(P*Zxh!F0?GY0@6
z8SRB`eOoX2TpOH&Ge|K;N=-5a;UQUqPZn|G8s<lbaaj~3+=RSC3z}>QP?skIlr9CR
z_~CsLhH^?uFMA3KD0EcRok0(|EZq;R`6fYc0ql}B6CEP35)G|#53`soW5NYIrGmgn
zavO&qhf4<>2I!<5fWJg9u4$bg1LQMXF`|lH$vmKflNdjq3mUdt8vGVY$WFD)II%UJ
z84LKR=#~a}9(%{n&hW7_`TAn=2<YN)DK*1lO1hm)>oPT6&N&;@fOlU9eYCv6i@!i#
zxM)7sMuzCnBFk1j%STS>9ycm3N26%?20c7-6#xzY3Hiv92^{l4J2EL|?^&gYbUINJ
zeWMlM(1S;vb-<wFa?qqtwQU<(x|D%1H^cHOn|!<C4RLFd$(A^_QhX29rJ(sLmlALh
zw|j5jy^Y)h_uUNBePd#{(I;k}v>s1M*?*(s2Q%mn_UJrhU|7q`xD$dxVX0mrt5--T
z7j#7_gr8YT_S;l~)w?xS)0K8<`(!p07LTYH-h+%IxZm{h<;%b6xd*CX0Tzu6AzWJo
zHr&k3fH+nMdEQVi8++;K;*tfoC+YSd*daP*lAVWUv&yyiNxW|Z?>4Jutg1$HOeQ9h
zYJ6KmuzRgB9=4}#964(S(gY`iaTqcO-_)aHlOr#?XZuD&kzj#=on%LQ3#P`Qp2?UB
zi0z2b#b4EteGj>&#qW7ND=%>}L_D2}TRWXjl<JZrdQAQvO!8;)AVj$I<@mdGJo`eq
z8(O{0%{=CZ1{A^fwu~<?4_Csmb%1Hvw-ggY;txije}B(a=;l=kSO!s8B@4Fpc@q<4
zD=f}xj6#(%Vl>WW>jf-_OT4-Bi9)dm(#C?Nl<UWt^|jC3+iZ~mw$B>yQ*LI`c)Vu&
zq0*q$kfwb%2wxT^<RN69kx33ESzBCGmTmLU7p0uLKV!o3U!SP8zg+;!F<+*{MU2J`
z`b{O(o??hSrh4yKDI!f<d?$+UFM?_{LHGBlrc9c>*p85sEhUGzn8tUOLxP5s<s53p
z``?l|YZl(4dmngJ0|mn~TO4=DZuFGWKH$fPNX-JB5%1;Rg!#p>#Lr!zSiV>%f6RKz
zN#{>UG<ouaN?M?XZG}m;*A+t&sXlN@6#zA94Q7tBZ`}Zqf!yxZk}-zTnRN{o*yLcX
z0nJpU^&6HHrQD`hyYS1@(F()diBdBZJwED0p)ViLm3`dI0V$idd7X^dCxqB%#=<45
zZPQ^9rrT#-GZBV!O>X35l~nI}v-%0W0sv3X4a{ZD-bk8E0*OPxw6D!4w+%5h)|6jt
zo2}V86U-qVi~d;e(nK$-36s`=w`qM5w~(o809fS0`cRdW*_Rjpf4^x$5VQ<(xw0;v
zd(rYm?4@d3#naZF&~mV_K4Ck@Mt<pU@OR7Ky>n;IVDFy_K{%k1d6i9xlu$&)P~hN|
zO|1h%gr;y*BG*-^cWWI!l$WU11?pQ_LSWjMk2gGXDe?ZQ+o`q^mOAKZub=2zKk<yq
zwMu1bsU*n`=aN%Cf4yz^sb5d5h#DJ~nuf@mxxHnlPH2Q6%r?UfBoP|Sv!7U5kwr6>
zWYYTD*AQ#(jp2^j&_@2HYbe?E{pi=u9@?JDk*IGl{Z){ph3m5mvrFdfriLAgyjmic
z06$ABD6Arb^UQtOrDZT(&!b!|-tga7=Q5zzy^$sWNJ3M`@d0^_1UNDg3pCJBo2-E_
zYg&h``2^gV(jgB_)D6NJzAwH_Zzz_YVNHROw&zM9G`Gt3619r&-4ExHl>lQvEnbbg
z$qs5MSvakA<epSs@|NNt@JqGiYG)BO6t;So2_eGIsi+xmugOGEi;YB5d~@>aZ$u*U
zwN>MnDokh=BN){|U&CY+G$9AP{+Yjh4&A4X%60zT9e&VkGys|xK?1;5+tJyX{L#~H
zTov?;?4O@xWrYx!ftXgLZ%I8_pVBsb#vf)qd3z&oW1=)HHBxP9G}wi7Q<0(D#*GvY
z0We$PdK1HVtp-$@4T_I%(_EN#v#pF{5jK2O-;dr36gh7`Or3;A#C|(+stY%$lH6D2
zXh5PYN(1(Qdh?CrjZi1plRXX%>P)7d<Wq&yt2mNflMW=fpghtX!=95lYQ5K+!SA#z
zedR!&aqDcO&#6?Nin%Yco7HLV-I;b3w=cc)3{XewK=!bsyJ4v~%TkoP`AJbp7EdXM
zEmW8UQX@n$nO;Y0Tn9Fp4f2O@T@a<n0q8S-I^A)Yk;@!Hdqc1yK5WG1F>pKyAm8<}
zFO}W!?lVxoxEWZakM)Q34<irf>0m~>PPVkFeUx`y#osVtgeJA0f)0l`as&wvel84B
z<uSp>(5@{VCcRn6GP(2H6u;)P-u$r?h(sR0+BuP9QOK|^?VJqys|fK1A7veCN+XpU
z-!;3yo;r%E6x{pR%v97Z_BOkdwzs+HgMVlPlSiKVQ8;+80${iXoU$yGHjC?l+c#36
z{!F$0Me{t}$RGM27aCORz0KH{)@5fK)f$nHea4dylVR~vRXlhl3Jg|UiAdfIvUdc1
zZ5<DhW2b7@m73GA03|mhAS%J7<ZtDpi46?pVHQMQ2lDm96mAl^1FIuK7U7a!Ga!N}
zvUk?aO67E4urtXyG6RNCEL2HW--MQdmiY!z<1wE%-{_jI2uzyz0qA4HY?Qcb8a&9<
zO%=ZAYs$wOwuwAJ9mWaf<mvWN^ncW$bTs8qhdP_`0bIIAkiAiXO!pS0q-0Tuo<v(;
zoOd8J(_LHz47JGVQ5aJEXE1?6_RU~(uLQUC3Gk<HZ}iS0HQqOQDWX$DC<zzF0Bp69
zuDO@0)<8Ni*L~j}2-e89kK#sH55~R?nWa(vHwbT}_297?04n$eHW~|dH<<-aJ_sZ`
zFq&Vv!JB}RLNl{wp_(V1X3f1<_<itkC%FJZPYRT%WJF@OH&R!3j367#8KCKP^b>6o
z+BWqFUXqh9ly*BrX_wH;N2a>JA3y-P&0}S@!svDV#r=jj6og*rG*>83?ko3E?|Jy;
zTk(7p|H^{8N3UUGK-vSi(d6b*RqfluOr4lfc%SQi=5~Jwm>>hAF&gL^J;^u0^}>@s
z29ssm<x>1?FuY69X_m#|09%dpV5!l-@M%}ZhoGkbc%vJi7SX>Gb*=``B`&zui#1PT
zI0!*_s_T<aB@{Z5&WhNf;Ey~HLPjk6s*Gek`^gW}0Oik7LC>rH*~-Yn%PsAwD4Z%*
z5bGU?CObEbVnc;B#xmGp=e!h}@)sfoekp#f!4%Cp#dF5;WqVC=Ngvdm>~F;>6H)^d
zCxg0iV7G>sGgIBGj$}O=RKi)2k$uCZQ5GvHv(Qz6Byu2`aKRnXr^{yYH$A{yL>9vG
zY6727mX(>LTO6sPxo+|dCQQXnw~u0BnZ)E(XH3xK+aK3YvfTJ~!2lQ7w(;lMh=vA;
zwONZT2S1v6Bd$f#ZX{bmDUq>Wu|8f!{3|>3XhlHFUqv8NsTZNEyHoyrm^3FvG2CY%
zQw(ZuJ!Z8WAO%YvV$xAOSd2T)09M%n^sHK~qJM~dJ3G=R#>JXj5NvSGlNYDmCWrmp
zz^9bk%%>KobPCb^T86NsPWuKCi7=r_ESD;8^nZPE1JPlj#0$CK-a*~dc-WG`FQ+V)
z?fapB*yDQk4zBABpUGCx57d8(3$t1D4~mJAT^!|9m?0(;Px8lGJ<+E&v69e@i2K=z
zoH&5w7pDQGO}fIE`1Q}Tz;@1nz)NZAA{ReoF45G7#xtoCB8Y*E$&?dFj6GPM@zvfV
zn3Prr{jdnM7+NDLY{?k-7kH8^KtZW-je>7aT1qDw_ML%ZQ}~(ajjHe~Brn0WgUjh?
zUgz+XA|pXE(4}Aa7nvqlk!&l0>;jjrwnDomK=o8c<Z;~&MQm#}#OZO(^}nJHyt=<g
z@Yl{%A1oZG%Of$(m<3C^g(>HHs4-b1c@~6>g-X=O+y2QLi7%jdp{E?<(ob@V1IM($
znEI0pj!Zxlb-)F3jG6oJxtu3q-KoFI#+8#A{ZDO}g_BDg&hvl$qs%?o<NEvOJ2NJm
zY5)4$Wq%D8r~Z8G;FK%j)Sos?`_DI0==%H5e-`I|=bdx+M&fRC+gp&f;9}ZVuE}8J
z{lLwtfq4$4a+d3#*`B#Hzl~uW?0JiLPW@G5W&vDj-XO0ZMt_vymPipC-@%i=di82_
zCk{}{_RcVZrZvJ^Bbu$N;B)3<PnZM5T~wyN3FsT&WYwYz)oZ73z~|h9>D$~XrSKLJ
zDc5YsO&`rk|AcCYn!9(jI^1hUwe~})fVnv*JG&04k<JHe2BTFQNdsb|2fj#|#o;0)
ziT@wldE@JE53sJx=o?==n#z0V$g`^-?-iX{x>j25%6Y}#sc)Hd)qS}#`=dWa7JcLD
zubyx1v(fj=yYs8BZ2uNqER@^1*z&i_ch8re-$i>&JY$<V+pC3=Kjy~{R42MD5uL;K
zV2P;JtYBF_`ahF@`hWh{r8|O}%m3g0@;~4HzjEo``MjQgklsl9+=UA>;9by=gAukP
zsJiDzi}`QR)YjHOgU#y6lM4~oz6LKv_H*V~z(z41ShWADNf&`yzfv13D{C}55~A!Y
z+5CN24w$Y0ugqzWT7v#GTy-3umOZ<N0_G9!>w6cN%7!uft`0KsLpgR)-k#w;UE8($
zq%M!)g5(53ExLUvoBw7YTe~q*ypRzrAGQi}2J;f}D4xRRYj8~877t_Q@D)ZIAD-SQ
z%6k)LE$EwxoM<D(Kz@Gy6{1h(u+5}7q+=&fPG|IX)~p9o&C!6Th=aKG;LuQX8S~n;
z^$;T7j*b$Wll79kQM4EmOnSEJo`Y|QJ_JjWBImOC&+wJ{jy-#9E~achg$AGg?Q!zN
z8-OddXd|3x0|Q2SP5@gC_oEBY2Q!C3$xO(MTUlF+nwjmFkd(Zjp)p-tTpW6AYrxr+
zed&Dn?lcxG6qFjuOhLu!Z7rQog2+WmOKNqqqOs^dF*Qctk61El7#ACRKP+q~zJu+>
zmwCy-{tqydc0Pj->;CY}8W~WH+#VwTik<u{2!#)2H;Bf)tFB(h{ZgW(!lL)rQh?FZ
z=FhkJu!D)s|IF>t!RY;3GG;zbj#r?>0wYPm>U7Hj<<8n#hPZeZUfqxa2ID_eR<4CW
zAU%bV5iirDLzwmWgGz)XdJ%B%2iuEBMn<|qexvwi1&RX4y6wK_T3TA*vnvQynI{F_
zBAeLR=iwn=$Hd&Tvs;LvlX-CDmJ2@p>b1>@6P%Womj3?!07cNV`y9YFvjqcxX!-28
zZfK~~hNv#Zs(3uQskPM_J7-4({EfGu5v4L`%f<&wd{2Je;l4!F`V|DY@vD!f{jRGM
zeJ{vX16~+^fK=GKa_N|=s;Zd|jccWMoWkIkKDGuEM$&c>k7o%hYhI$=(S004%a$*P
zmc_hVw{HDWojp_vJ6*sPpI{AgUS0h|at(T**vzJzH*bDj2Gii_i`VVH1Idw0&21Cg
z_U$7>?T_FAEM2y2V?rl}+7_KXdlm)LOq9{{<!zUX#$}e5uf_*aP!_>ZzZ~4l{+x+i
z>eOv`{rU%dpS&}8MT=2`Fg5e7c^KD#YL>*8m#kUCZkcL1p&D|CzW_BtAu7}3CV*?^
z0C&`g80l`DkKWX0fsLElOje}V+~}-^!jR4ib!_#AdYA}4!5T9Si11AI^<EgBdt2Q3
zdlb#U2atdd-gQ4HXd7hBCqMldpGvv~D~9jxrKgKr%XJLGqoWJGk4&ZB&z=pdKnIYF
zt_%BHV(oBr)Ycx^HuT`(!wdM?{Kl;-WMpM$4wROci=#vF8e#%*XdT@-acs`eDimCY
zTU2ErnnQ*3y{sK?JPp4m8SkW_(U=7Y1NR*q7B};`y-K(^G&;mXvzAVA7?@pz=1X-u
zysLp=dlrAx#A0NEB_WMT*$Bt`IDJJ2J!5bfnNA!(eqrwiC1vGR>X%P=VcXOrzh<Nf
z#|vJoP7&R`lr0n@6Z6#4%%qh?V0!Firw>DbCBGA9n6W;*jS8PV$6%yi4@|tfhHLGv
zD#u`yZwUqx9eG73kWN6TWx<t-xVd}O(i*Cw;K&5TXFvM-Im#1*T^g{tn1PB#KKaLk
zcGPA}T<S9*@3DE<75iNSqe^Ix?8HDW6m(60y?#(^-Zo$iF=><|qG@KBU6#W{=aQ3R
zR{W$BQ!IGkz=2VOEOrObOGD0IF|5yA2#VsF#$V6S!boW;-0r`n0F%g^7NfTwJbrxt
z0#i}eTw|^J`ntM%IQ4AkTE<%d)fogFD~<Z}%jM_S?$-^Rt^Rda%y0$Iv%2kDx6VTR
zmQ}`pcU1Q@b4CJMhENzZd7ExFM1Iu@Lx0+_MQ>?rT~ifz(dNmM_u$ePU=FJ&3x~57
z(S#qS=8IVP-;FHU%%I&3DfWY=rh>X-pn8C8p=Nqa?IeyY7lxMsBF%dF@};jmq)qPP
z9mDj76PONa#MEO8GY<_Ihc4cf@dA`m057lNED%M4&qVvyMqBGOtAN04gLe?TD+}+^
zq{Nrm`lNGjuaQ%^-5F+PX7+)fZYw!CI2HnZeUIK&oXjHR2uG%mDQ`hoUI9pEHi!^^
zdU`(MyXPnKKL}odLoVR!6%i3p=JiMQ%@t;x_4wzjc9?*(WRHUvc8={!-@&`SAoCnn
zQ&YP<hT7}dix++hX%E^lQG5w1?=p{nV@#L!`?qvU5s1Xtf;vRBAUyc=HlCruu1!Cx
zs-gi--P@hyb8Rebe5glzXlP_;h>e3I;<b&K)d}cBeF9U%u?#im*`1p{a4)zoVAyoY
zW6=&Jb#-;f5D6V;joJIY@6I8>*yOdnVQ9JNx9wURGm0=q|J1`HJkA0DjdDx)a96!J
z%F#phM#sM33Z12`t&MM884|iUCN2)l(M^tJj~_g!MVCdV`Q-R1ydfxOTj3gs6};#I
zzyPK?9{?Qa!*T;|5zrZ$j;m1KEs=6(eFK9-@QggP+Iw|Xd5Ey>ic<hgYgP7Wz1zTr
znOd7WcNh`=N2~-<3fluQ-WgGodV60roDipi@y^L$JcrHS6-I>1o9A2xfAt;)zZJTX
zE4(K;0Ia*xI2!c-FkjBd&tHTUQBa<!%b4#WY_8lq<YI)W$zK)mJKKKRK<J?szv9;5
zA2vjLTfC@`d3g)LSNMd@r4PMkh$4L`EtR*Bkdj&iNNK_1#SS0%7vg4qW+@gsO4eWT
z`?2rnOCtirt>jQU+=p#?@i1`AdYo>`6y~sH92XV+r+IpGW@aXUTmQ0dB^Zp=UaDxQ
z#GLnisG!wCK~}Ip7ySW`wh%8F!v!CKT}<OQ>Gn7COiL362jniqi|Rf<+k~DxH?{%#
z`1_E)EJCMHR7`9JWT@6V?o^brpmzUkR3rJbkIZ2cvqdh#2EB;OsV`nEN7(j{H~YnA
zF5`NUn;8U$s5oIEA$P-BEcAq{jkC8>n!F>kxTM5HJb)GRyb3*}cV?D-Sd>u<BqPO7
zk%0?o9cGHGfi%@2kNMNzpNXuB6Q_~V4P*c^7G~wVc0U4MLS7ggfEnP5bQ@zh{yi3?
zD9!x#dyV$?_Mh$TcHnfsE<>M-6(CW`ag(`hTPpD-(4dQoO9R#z{`g?r1vtaZ<YVV%
zvxbPvLtH7_BMP#!KcEXT|CJaX)_OTq05)&gvLoW=&6|%A3d*{VA3OFAA_(!$>vk|_
z+Oyg$UN8LTYLraCJaD1x@$*0C<je;p>?3FooRBj-bN1}X>@a|caDz=N9)=}qEw0Mv
zIa1>VK*S7S+&UD4><J+N`Wk%izJ7Ha;71Upzy#vgBQS={%r9Nq4sv^F3HNf*6OpJ0
z4I|5M2$68hUbn`(OC+H?`p>Lcvre?(Oxi%>hVsk@OU|7h58OuU&6bggk=LcHtPyc(
zRAxd<Orht5q+|vHL^O%F2cEPJ^6+2*2=xz=8;?p>78ZSw&Ax^yI_$H<L#QL*l67hD
zezfB6!op;`-2zK=Q}&y5&wx1563c(Yu3|=R&f>)`d$2hM5^i;E1RZ;mpdC68Y~0+-
zV8X~?_y+`B0EOSf4|$H&lC;Np5JF+VvLvF(7%3);(2WH+(ZxDR{<nDe_>SVfg9ZyL
zHo$Cb1R}c`40N1}XeKkP{6*iOXLSZzDrHe7_<UQaLKY}TGIOTQM6WspoSWP`o}Qj%
zC0jEJ(5f*Jifw6Y(*Tf3*UAM5qs5tn>i9W7KNDlie*n}z$u~qHfVKhN5b{={l+?{X
z)u5VgVF#K(Oc^5LLUhGedL9MMK!2!PlV3pK%lh7$<ePi&DehqB>c*!$c>0`U<HmoG
zPn~G%>+35@x3r9C-Us~$qt0qpQ&ZkYTYi@LA72iVlR`1q(0aLL>bkn~agYL-M<D=d
zfV(tcnm($P6{n`ckLcmU1~BeSu^3x}*n=eVA51#m$2IJO-K6}b9v_<jcK<F!zNM>I
z&xV2zNdFA<z78TKkAE(1;Oy-Dh%#^BsJD8W7GliXCn{>Op`js+*j8hGMFGd}z3Paz
zu^@!qx)8cIZ{7MKJ9{49+Qa1J<O=V@w{G!+bomI)9kbEFrMLt+d3f$4^M&r<L(op1
zjI*(C-no;u`+ekGFx-3U<z<hP^a|j_uaY2#qkarcEv-7dy#&)fc6w4))`c*MIisXB
znRO#XBANzv3H<tinHm0a{MNS3A~<qj0qhwB=hf)Bg@u5Cz%8Wwqhn)TQ2O|kk-;V`
zyl(5(t(^PzJ@)iu_3-p80@34r(OKSSPoLgF2D*u#p9$s4cKOg1uq3g@(eg#n=CM#h
z9VtgW&i0%I3&b$il4dsa&r69}NYY^Hw-WFDacnFPo<5+Fr`QWucP%a(k4h-*GmzOk
zd7s3(qDOHe;OGWngjry<^L3R=_e0d_9W*+`0uGNCK+rtea>8OLMq=yz$B$=cJ9F&Z
z!TM6-5sqp-E&<qXPG`)RIrAs(V^BK%6mnEnz>KdRR7m=SSikd^E?Hqke1Ycb8yi1-
z_DuT1>gYmnPSB&8KWEXRObpOGb##0HocLpL@iGjKKLX(>Q`SN2;>B7VmN|3g9K=gQ
z`f@>6*9IYo8d=-b(o&RLX3d`cv8HA{!WDgXjD4KHaABH|kPs@2cNrNov7l?5SPR>Y
z9kGZz6(ei0PJ>8h>+0&x>*!eH6Z&Fb=o=cAw9c`vc#ijswua%=68S4U@T=~HP)hsl
zq!lQ?5^fLd29Ep?l?=Tr0Rd(#<Kx6c0ouLT34W+<1$RPw@1{N`=8=EH8D<*xs=kqt
z7`4BlnG|1qk!jW`ul;wisl4`^-23+J46@dlXo?{ou7q$Q@w8qcmZ3Yb{TPb2IrHb=
z$AOpmxU0rEBm}_!`-+M+=&yzM;pF`3>N*4^8KdWW*zR<6_zg+t`a>wRpcXd--q$81
zlz8gxijgdkim*Zw`o6T3JTc3*JK*=SAd|+E9KsDKAt7;ITYI*mq9WRMM<S8}!^3xh
zz&afShWGe+OV+NPKYjZ2Gs?=ocsfuT4tRKY#C4}mH07;j5jJB&BliOm)AL$dvtZhs
zkesZma*cWE(rHMhN5S`^vJ{gtkij|~p}aRCDajW>XXdg^$BdEL=hxKJhj({(FYC$q
z{P`U+)o2(BMg-+Ska8BRAt@*O1Bj7UoQG#Q)B~SRcu~q%Hha_OQ>(dTj~<-`IZ51o
z4v<~hw0rj=5Mk*sV?9xZ|7>klZfMKKUdc<=!)az0=iST6c_%qp2#Ekr)du!cxc~B{
zzC&(E9LdPDQ9W5ik++L*BX(<|1}2>2x}d$iJ({?D$jY{=y@jUBvdSIq6^ReH7Tkx3
z$vbR5yy`)ed)R!Cr<#e|V7io)6gIk(N@wt~)p~Ao=gvi-2u-oujUx`#th)gLTQCrH
z4-0xZLQ>47ya>nJ0Pz<|(?&tTHB+_t2P~fhw7T9E6-n=G_R+(Mto8U3j_vyT_3J`Y
z3#0%U2!ip#^Y(dTt>{4pI%~uwBo<(EH|V|N<Ku&$gD7q}2%8`AY={cG5sBOaG^+W?
z<DjoH9bfnK*)t#x^S@P8_)>5~!kn9z7jJenNfDm}UNKVIEr?z_jk9N^36CC~EbAVe
zM>Ky8JzM8#??!MWcmCeUjT}qMDFmzlb_Z~c=A-WUfCI>hBugDULPP*pTqg6qt^w+G
z1*u%Ru`V0o%oPUZT-S!VU)nOGbn9u%S?$z<dPl}9&BkVaeP`#MP!%xtkjy);e+%W;
zfOaUHJ-fC?KGZSAtgNgoe1A_m{DRp9Ers#Ia?zx5MgB9_ViCg?xd;z<n5$oSdF8vS
zo<F~#%@sK%4}Ka}>LTJ)miwDmuh@DZ_rQ(S_RS%m-#`Euwh~*H@45+P9amhpoNyUn
zFIhNu?$1$M$isd3_U+rBCTXx;wNT>w=6G-0wCVm3=w8gay1MTT3rk8g;J=wKCMI_5
zgjzZ31}5YI`P{fpPZ<efDP&B>8Z9Bl<Gn8e&)2hVThoNc1*WZ_<3fs#Ll)(;bR!oZ
zpB9qi&+rQrl^ez@ZX8Qj#ovD9REz3@GRplO=aP7@rS<ps+d}v!cz=(_f!oe#k=h0K
zaP#nJAQ#UC400^n&*zWlV~MRAU=|xWcjC6=?;-Dz{BF5&9633QeD$=o#x2-kib<Wi
z>Yylw_aZyZPQE6c9@Tv6V@Afq<Csm1iH&_pzZ8NqKgT4XJdz13Qt7WI=)j-y8pl9_
znD+z{Hzm#6JxzAS!X^ev;B1{~)GhLz*?Ak0h$Cr%8QO5JD8PA;%CSdHV4Rf)h46BT
zYq)ki2A;#vhZ9jU<?`7xXE*`WEI-$Q@T7%JWSg3rnti?U+c!>}vWwVf@_}cUluKJ6
z{^bHoI#Yh2JPb^cT-+cpF<JM;<X16jM1h>M@D(|3fE;IRa|R!v|HayZgb*CaeE<HC
zkqHh=7I2UIJK&NP+>IB?iCjVx2SF@gqVn30mNm)$;sT_1jjLs4W^!ZCKuAgg9Z!><
zujS?10BJT4o;@tBqr*-UyjH{M)82gzmACm0P2mRD)a^@#t9CVxXK13<^JoFx=7iA0
za^JesYQA*%oXY=T`R66Oo~XHoGqdW4cbo7h&WcnAnOuJ_h-q03>jeh|Wf`?sI3-$6
zq@6b%N)PJpkOs(*3y^DO{XHbtJTNo#%|RVJUkkisTNF*%F+%+OoKWpw<<vo<O})%O
zF?mo?Ef2q`QxdokCchh#Mord<Smc5mEZZm`7p7qMJ>4?Bq1RXsqXZs}?XW-fH$s#a
z%Q|UlYMP5P$6xY2(QsR*0kX~|=e$4(<3y^c38P&xrw|acbFq>yaaBmStX9bFMbV4)
zsemocVQ>#_GFC!Jidt6t0~_jgP6n4rFt0K{-=3!iWV|+BHRFWD#JLukV7hR8?FJEL
zN2!U9jEjqhN3N@zTTF2>F5AxWBV%b&?JBJTCf#r*I=axQ#sIPN0-X-C;ojq%sg$x;
zos6pr{KC1w5c)rvRaNIS56oi(BH;VhFFL0tZVju%0^Td%PB#j~>6BU{>9xm4&U^p<
zZfpwXp_X|*Qm;%5RfO=rNm~XAmCV~Js>xyONX>EqeY+x8u-PpOn&y$W0xPwI6!rBv
zd4Q8h_oWXH59j_W@_UI}=OoS=u6qOLWPBRIyaAktlujiQB<*nyP56p_%CV(}w$v=K
zx>r~DIan&_O2$0fr%#szLf6&TVn$h*;wiZ;chuJ*DEkVn`x>RDrg`Kzq`s6z%s?TA
z2-%)Niyb!xm-fWGeY@II#3eUZ!fE9(Gm-2eedBClgSr_SD5Ti+qF3qz=yp`;z~|uZ
ztVv%j;Sx5AqqD*@3#RO1H@k5aKQ~+V=FJ=SwW2jCCLcpb0g62c?hdX87~#>#yLYeC
z*Y8Jkg09?2S3g%G^&Q(x0wBs3m)V2EjCWWI`6hzns(En&J|W0(;wE3|8RR-^2q%`-
zT*p%p(qyiXFm^IoLE5R+-9Xq>gG|y*`>gkMViUCRUJJ249P}IHxtMH2%);6g!yHn3
zug}^wYq+67<}tFWZzM^54OIGa0RW2yWtD&7Nykph8E`3n-vf{_JS9Q!VxvROr!?Mn
z{blRcMV8$hL@+feX=rsk01UC{GTDx5CTeJCl$xf)MD5c>Ev<Qh=MXtyv3)SJtW2R;
z0GglA;X|`Z;wCVo@NZ4H4}5brmjX}b&)E@w|Jhfm@3=g>?X8C{`^l1(<kzIMJ5h@@
zty`XwD0DS1w1YnZU=)*^&cg^qRpndv89CR^s>7esK#qF%>C-1G(+TT&=Wb;8|M^p!
zx;tb4sPt3o!{$$ttWR$8!;hb7us*=`xFB?<-Y;rX*@Mcs33X4(W%e7{Qpb-k9Xt3M
z$y;##EhKo$s+9OTx4ih`ij5T}aW;AsGHWMTTOVZa>8;*Aj0k?D{6%9bAeQj%@iR}k
z$j~86NoohH3)4|$yi1X-C2_Jhw6p^KJGW*iV$-`qT-<4B#nD_N;KhMRKy^IJ_oR=?
zG_2Ed7c^$SW&-JYBewx0?*cs#Q1p@Nkr$WlbTkKDUBLuBfN%ph>-0ttOcqz0+YFm~
zRwVkLg|uXYnSA_(6=Lp9!1%VfRAQ5U9q^2d^4pK=H5ZKG?JX<|7A|~=p`|&2XP>sK
z0O_se2y=Y@*#AcMtU--!l0ur|M~`wN>wTW{Vt~}bv+u5$=jXW-NXtu9unjdc7)<aa
zY{F(`@7aFI%rgcBy(_>4NUGajK<DH=VhR_EZFLZ#KK!nTw?TKTl)n=u3i?f{nmMaD
zJYM#%a2Z|(s7BqM`OXs2&-=C^jfV-(JG{IL>gxA^Q2Qasy^fAPq^`aZti;za%R_-b
zZBWpu71Yz$(AFz5L22WMOy`)C)T3w5mI4*GMIJ|)Ic!^~9$d6ucvnF^?jnN9Kxmyl
z{r<K3{y6y70MN^J!F^+gae7L(f&s5!QBdbTM)3ZC1P-U@$c-C(z(9{Qch$b14m|oF
z071cWA<K~ISVMOGU2g6I_(Nm@^Vf%z@6b44eNm)UobXzFZ-m5)!Qg(mAU{7Jb|x#=
z@LidM%J^4H%TJ^~gek)s;2zA>8ShZ$U~1$%dTTo*XXkI*uz`U@aXJGf>;<s;kT~88
z34uiGMyh7$(odTJ#&<za&kid-Cvffh^^1XQmdQnFX=!0pY6)y@=c26wbaKPy%`=dv
z9YU~V)B@#2ZS@i2o2Oy=nI9)BBU20KMYXa<xlo9pdbC*%{+t~jxW1_vygyVOhcAy|
zoon0L!kqSZQ-6-Q+mfC;jv&#;YrTi;?L1<VPQbI5FQ0%Yb=`|c%?p=RCMuO%K?C-v
zHn7G2phX#<O+-+BA)$}xk6}4<0$%AM!lqPWcyT6!vR|MB1irwIcH_2fvyp^Q=Tb#Q
zr4WM?@oAkv+t3fZ51*(2J*Z6vpM4wIz?&|QRA|oJxzF|>V(IKJHOAFJ<~;jv{b9T-
zznq6k&jtrs*YT?KYrJq45epW^X~<_12!({_K0rW-EP|{*ZNJ&z>9h%m@SlbT8fX+j
zudP*TksUq8NBr!9O4atz{aNpeik2!UC@|7gZ=%3`5EZot`5fb*%#~OGTtHk_F$yv3
ziFx%on5;QqYiHMmPmbvyaj)GBa8#EgyLynCdf=KH6*RXX4LSq=<`X6x?*Qt66~MgD
zsJg&d{8?W=9lpyAO-=tW&=Il5EQ&hR3w?8<!rH(zor7M_F-b|MsMxn}sWDgk>z5Tw
zy*6-hEk$bd?d`So!WN@*YTFzWJ32b3uh@B${7{vmCFSHCfAu(7?|D<yH0H4L%@bSO
zPXG)+8WxRcoX(&o0STN6kE8yf(h?fp(@>Gc7bocd{KsOnOT3|>!Kz{c0shR}tE`~j
z@2fUP&vGtGv!QDOAxLXyZQHhux(axGIswwp|D3BqX-udC3k%D<1Tp{3B^l;&OEV-s
zpz96xBWKyoU_ZZ2fa>q#4*GzzgiffHt*sbJ1k5QD(2qsf2vZv)VO-}AQK>V0fU@AO
zpB)N<>+j!J=v8CMvH)3NPWSk65ZzYXyMKRX&~At+x?~+fVS>^RQ;-}`!qjxc<TbwN
zR{w8i<VTy*?U^3EN7iVI-%e<VZ;lASRFr-HFo38(Ku7)+_9tu|4ndHhAGjwffG;^U
z0Vh{uzbwIKTP^_sY>pEpcL8yb&c`4+chVp(Vi$iXC|HaX($P{15Bw0PS0Pg6i~Eg3
z84}o)8l~EeL2a(&`+@GueV{94-Cy$aVMTcWd3wZ+H8>lW%EFn^!necQ4|RI9Tb(S4
z>Hs6khwW#)hO{`coS*2UKmBr1C&*6{^HFswYE&fxzeW<80KqIm)eyo6$qEb%q&SLY
zC=8L^RxKzm4@AoV8$X~A2md@%;O7+aj6>C9Y^MNY@0E(hXZe<B7-TktL}N{m(u&hU
zE;{%XAAkuo<{+YBob4f+>rl#lg1nEm9&#M<g9jO?kRL$s9ccF53rx}`zd|K>qCDp_
z(k8}lV?=!&JnZhq$L|Ho0VKl)goDL+ahVWMo@SQ1mG%LkYch-Q&5p`xyMzLmA9nI)
z-`~AspzC%N!Ii)f03a#12Npqu^c|p8hFL@uQ2qr_%_u}SA>YVGRLJGYlesAAeVz8B
zduoH+2RXp&vIzjQ=)C`vo0}W=eDR<7-9acZt*oq03)unLsKX2%T^PXTwhY}m7iZYm
z)I{Tj{Dw`o4c&}`Q2jtI_XrV)dXL!NPoY0)g}^_rqX+)1wUARej1iO4?2q5SGlNAX
zuZ*B&2T(lBX)D`<ryKpB-hQ}T^w=|5mZKKMM*hE3iq50ApDkql8Npx4FMx@Oi4#}J
zDo^oh$0OKV8eaQNJ|tQKNG|@o8@sGtQq5)49Ds{FaJ@jexd%zrrkc0Nce?wm_ecf~
zX6NN;0$ti((2U_9SpnZ1070%uvbj+O)>Vlc__w4K;9@S4xPG0xnG0*FE;+Ljg0xPJ
zH;s%gqQUzSt16|R4n!c}*&&cd#9Tx=1ySm~IJGGhD#kDN_Bl&zl@(Av2icV~L#X;5
z9hZ=3_<`f^@;$*|lOLX`M(t7sd&Wcoj?WzDBg+wF>qot?gPr}29vsaE-nmG(_xb3a
z$dJHN)*V7HIOye#R98hg(~x)a(u`?9ZR5KuUtgH6t%~+#E<&S^lVMDJyr!_(;D(mi
zQYW+l|2X%c`Id_wlDyP8MN}eMs9<&M7+>~fe?n?)r2nRMXX_?SETxycGm_&=TdFW;
z5?c(a%8q6WR56kKNvEs#9?;Hb$#01UtH2dat5|!a3qMmWxooeBKZeWBta@B}y%VN5
zA=Lv1<-@l%zzQ_n3_z8<{jcsnKM!l6ua%48$1lkab<F@&KmTfL;{mErH#+_q?*R#5
zuAmmc$a{^0&=-<L?9swIlz>A^Z5p~4>vehc0Y|(K(Vbr>I1IE(*ZTb7!-usni%2cH
z=5ZvP^bwUzFgX8UpLvF~60V$Ge-?%8VQ04<Nbbf#d|te4&5jX-01rRimu_y{2({rx
zyv#dBkSr=EPE1z|x@qyJcG@*nQ9t3(<(23W_1OPIK;aiZo$gd9|Ee_{wQTUqc?Mh|
z)<isaCC>?iL+p$CwzRVE-XwBaQ0C3<4!;#V$1w%EN<kP6>?J=$@TwCrAXftRZ)R53
z<D2QDHL@nghgPjxm9j}-E2<Ev)Sq{1goxigO#!n(PKbl^P!8(h9Z{DylwJwgdF!C3
zcRB>manlCe$|1jP-nC9)VPWTA0+h;1+EY<cG0%Pu)4Fx}yqAl`($WC|u`tK_s0ZQ(
zFt8B+a*?|_i9=X)pcnmB1@jBlIbXgU-@z(*G}~D)DX1_EsPMhcLx(aYFgp3uY~rg)
zp0|ivadYTSG#0{#i$<e>4in}fD`5B>ElD0EYE8)HpZBCAAG?-R!4x6mZasYLJAge+
zr)E6*nxLWavD*r6K;u(tMy9IZVu<xT?MoaP9-i0X{rvd{^y)oB(3C&FUwVn28yttU
z(6M<pW7e#B9WF0legTi&y)SX$dSR76OsIo_VPJ&+e0vRSZknLhkurh=)-yUT&av~_
z<;!0HuNrhvQVuft^5&al7`-K66R*r0G(TfOtjX6ydZlpUWYx16c%>4!5b<MT8Yi|+
zeO}mx%0xhX7UM(O5gsoPD1xOE=g+G-_FYC5-C(&ePi+vwJzC&lEj?R1G_6iX4Xv#s
zxm>M|l5^PRSs&iy7w8#tH%eq&UJ|D7<-5zcP_sHZK7R7#$@y(7A6;-or;Zz!Nw)^F
zfUHWBGcHc@H1{a^b+M{s3JOc(-GnEZho7#$ryitLs~fak9#aVK1)Bgm-5UvOeeI}*
zKw5$%&7?xqb37rt1`%j0Jf%1tuxv5*ps`NxOhf*y)h-13hIDfmuDodb`E%yf{)qBn
z4+DS3Bxp`$ysnFXoSbpC1#WU}zmsU4I|`bGRPIQ|WqYrNY^yX;?B=VanZ3kJ@phi8
zKltBErK3mC9C3O@Xn6K*W<*=qDXGH%`wI#IZj3g^p0~i{7DA8pN<Zs8SXM1yHn{+Z
zTrn1b*6`3ajjLDllhZ@9tM73`626*y4Fq@1EPGH;w{A)Nc?%)b<1%Et&x&h@d-+!H
zxQ*VIU}D;eoe8+raud}6<{3&3u?PM%f)Z7@21(3!b%f<S`n(^rOIq#VD4i@CItIfN
zwt~aR|75hcyy|2`7F@t%+-uRic##u0WA_-xu3a<MGrr{JUc_1)aT&me3ZMM}gE^~@
zH<tGs`}0?IprDZp^mcd80dgg80ZdBZ+5*+nthOsO;zsv`c=A0E)Kp6{dh$KZT&S5Z
z!~StA>Rc%)t?-Bl6Ngg5O@Y0zK!haGY>2xD5LT#18M-`gQikRlb7tR$9R8i}pe)ok
zHfD?hZ?%O}X6?ADs;Y-cVsdg=O4J1X%eQa$8G;9>H9FXJkIS)ey8DqQyM;7?`SM=<
z;mXqvkvW}Ly%n-7ttY?AtE^$(+wK@)GqOEHqIdz0iI+H9pknJ?UcC5tJ4%bGv2zE4
zqf<%h<42ETEZPRp2hyK_{mb#irR(<JdMdy5m6WAQO)LSH5+Vl<ctT8-xy=<j_fJO+
z{5^*(K=9J=5;hO*X7Tm&Q!25ZZ(gucF4v!Fq%jrEJ9Yt%S5n^z)x-V4|8~a>XT`U&
z{YW@wq~Da3lza(&aXSI?VO}!Nqdg%3sgaouO0UmPyJj(SB~L(L;C0~{t^?d$To))y
zZbICqkB?Kj?!8nBz^Nm?CDC>F6Nc1e+U|iNRsfV(L1r9%h?EvgULO<Jj-c&QSytvO
zzwVWkhYZqGFQ>@5!%-&IQziWuOYc$8sIelBj++xwH&pNj#2Ioe_z3&JLtbxC_bz8;
z<wzQ>NV5la>I&#Rdq599M%e?77nqum!i5iBLjmD2vhrfkZ!}J>TPWJYMpQ;HR8UhJ
z`wB6rIb{#LU|tu?;FBXr&q$v@#q|;$DsHrzbCE^3q94qKmd}#?O{f_>Ojceb1nzb1
zM=hF<Zz}MnPvzVNku*#OUu(h^Xs${J5r}<#_pa>~zZ@5unIGghaNwLwH3eqLP9!~o
zf*LK>C(Y-g8_ku~3l43AJ6Q!fm2fR&6<mSb%xTYTFb=Xc(D!E#io=^)LqNySb-)2s
z*6;AH+s?=59It|IUQw2to11lYd12ua3{xs3*5C?K$_W*2Ll5G~?)%ZvT&UJJtBqqf
zFIl&4_wXPJXCJiK?v0?U=(2ZVB$Ag-xkvz{&J%j*YT`<;Vb6kULbltio4_1>?Ms}{
zY#fSvZ;j^&Elr`s5;1LO1Wf0bE?T!g*kBkC9kCKR`>|B-k^X}f`M}+MEf_^$Ei49S
z;*V67Ga8@x)xq-c7+kKBYPu3Jg3?AJ!eE|e5fbml16$Es2F?&}j3j6fO_5s;Lr{cV
zx?;rxq?^8gEMRga1m4p|{>E<L047oG_j=K;sn$dJ^t_nlcK}R3Ms-w__2*nYAUL!w
zHe|2pJW}BnD9z<LDtmX8HjczcBxZ6Q@aW=3KQY?{Eyf{96_=r?AZ(xynVCm>qh-|H
z0ot7hD1C2)iJ3XN<1Er7^g-*ZYOv7g=KAK4{fsRP$<lxVUoOhh84)}oB}G(<#vH<O
zUSIeq7r5ue7yR$6p%pcN5})TEUagvHnvtlw^+m{m3;}PnSPjOdC>Vhp<!3Mq?V&?x
z9yjwL&-W0qTt~fc9L(CLrbi$dzN@S}%Xt;ej#+RADP?y+`w&H!2<A~&B8jIlA^qlj
zR-nvJ2VAk3J0j*UTC~a|upjn&L}b86nU3Zvk(!ZpJ^-uE*4CEaLFvtvm;OvX{&hBb
z=vAD1w}owsC$b;w?OMy6MYNvFto9EIdW=fPsv5k`Czq;I5^IHVKOTixMng_$>pjqw
z8>8<tf#P@scTq!*jJh3`Km%Qedn1nQ^7E+m7Ewv@HVtKzD9lvMOi*kGD@J)`U&rdJ
z;}6}Mpc~o+`j*d1b3I;Ze278((EQtj7*>wEK9R4c795P)<Q6zy=snLwVUF!P6AXe+
zz;g;?wNF*o!v)8I+_tj)B3v(HV-139*9@a>EpQP&;d!`lYoX~&B9{pW?Q{Vdtb6g|
z#eH0Ip~g}b{&`70_<mX(%>OPm|8MQRc{tWx`!0SrXw;xcNXk%#s7x73lX=KIlL(n*
zxRtplDG?PnA~F{;rHo0VGA488k|9%yO6Gm8p5OPk_xt^O|Fie89mo3~@2leetj}84
zx~}s)uk%uc2Kb`Cv{o&5sKt>DEvwh2cnFz~ycM76+*$EK?(Th>htMbcdAl=bVDML4
z%jnRgkvcwVgFF{#T-5#vjPhvq9vK@W*hy9N`ls>hd5b^3t*l%Fq(Rdt0m>1SZ+3Bh
zNC@ym6;Gcb!@D^D=b&}{y-=iaNu=i*Njr#=8`!CnJNGi~U)!K3%}m*6QoNb|_4%D7
zyF|75$$x2%19xyBCM}T0Oh_2yvC0xR*REgxn&8D!xyd7lf`1@V=V9k^57l-!3BeKv
zk&4}NzJ==%|0u=vC<-cEP6p*WK+s5Sf}&#W);&QT@WFpH8B9;XisgUx2^OfRPQlLk
z%AJ7nIvJz^|FnHindGR*tx3a>&x3=jjyzjhFAbNH{UZ{f;4wK-v$-6ZM)(rElFFVv
zi~LQrUD!MQHGo-|O=r%XIrC!nc?mZ3ke`@c^Cb|8^Jx!*>5{_6Eo9s>2#45i?uh+?
zC2r;&N0v0{WZ&_G@0=V91(fhc^eD4UJi;g`t%rw6l7gY;51;r|-r8LjPcI@YjACGQ
zs}$t&z$9G1O#mx`o+E#8|D;Za79E6ChlX50ut;EPN~ZP)$)u|iR0OIu9<thRE`*+g
ztOl522fyBRW)6;eBZxW6^cr<l6*}8g%vlbcp^%9{u1ArD(z8qHl<#0*_y^oQ>~Q2x
zEPf;|=I4eEdyE-q#{UCAvdXB{!=VY-$xJW&8k5=X7WlGb+ClzeG(^1$FtjW8c;kBh
z{H^^{72;@Ow5|}x-MBga8l%sn(7-@C=WC~_L|}z`OnLNie*OUnLPDFEgB>CEe%s0W
z3l~lba`v)!q|^ay3LN8rsHkQ3A#w3-Wv7c-{jXTuf~DuA%;pb|&MM4fvPxVh1|dsG
zi}>3Av0&rd^mq%W{P8XYzP~4a-Nucrf2KuAY`WgDf9V`M9<8}qCp(JEzm~gDXw8|n
zA8Te{K54-Iykz7u<@b?hY-?;^CetXOMvwfS&dtriZQ<j=2%(Y7KuM}*49Q%|uIsRS
zS_{ObtP{W@j9%?{vkwUize45X>Hi#Qu+qrayAKjA_Tat%^_1Y&i8}A|*RPc&<Ivs2
z#iy<dFZAy`XJus$%K>HT))TzejV2h*)A;P#NWdRWfHX5_*-Q)Em}KwX-wvqKrZM%p
zkD-G@O2{a$zXLfgqV}z-I<u4m81+2Ui^kAFpIhmP9VHM;LN!%aqQ?BF4x!qBYVJw!
zE}TGHT46#02k3K&qDb_W1F9KvIQtMTX|RcVc{c};+=Mr`tns)B2&3<llYL#iz%{WO
z`+Q!!{P=M@GTT9RgJGmZnWvI8@Jt+}ssXrg9%|`}o}T8A6K;n_vwGPZ{J?)Oov*=j
zt?3uWP7rlJ6o|s1pAE3%N)*eQa~<vNRZCG?$Bkg5*!t)M@>N5wu&!s``tC9nEnr$!
zgZZDC1mfy~yQpVxXv@IZ*gm`GIHZrNxdt9@OM&1Kg0ATOA&xd!(1@>x2|r|KGFTc0
z;g=zob4hV=&?ydzJ!2Xw5ZAEfF!5y}vWrE#hUCE1!*8~*{!*w8O0mE=CKf=d`bJB?
z$BjXI%cVRCWUawu#a$n5kEJL52I><o{gsu4#T=Y)Tqks@NEd3nv--|)x}S#Gu#Vl=
znp}ul<Jb$AeFEQA_Ku11AeRN^w6o*8fw+~3fEanF+V|Z-$j(B(HS^NKO1Uh2Ll^~Q
zajKt$=U7PjFy&T+(!9g_R><|9a7fwiMglhkiXS{>d>FrIIKC4P9u9H*+Vq>+4|i&#
zK|2SA@S}yE$XAABkl&oE5T7|K@7OoD*eYkT!YAIg0+qDw0+JfQwOJme$WPa-Ti4f}
zlNk~7!GkKb9UT+FMG<sn*LGc+o7!A$raSZ;8+Xb{VBI4%^DiNeyg9o&^fpgx2FF|(
zSfU48aR8J?&o$2$T|lDm-|sP5Cu|+wIDEQ$lvheB)~z*U{-8FhR=Fb*x88Mi?En`#
zIp8d44cC8}a~%i)t?65zWA8cd{h}P%p(R)OUs?eEB*1HB?~Du#h`P9}bG?W)CRz6T
zeyV--YB@0P%LAXhFYZT8c`ly4Vv9C2cSO|zOy47YRiF;OfB(L@Y#K%OBWp38K#M(f
z7-XoNkrVk}2WaWtO~_<nhoiXjh*{)in<hq;_>c0PZJNkBf%x&7wWh|e{}kq^Y$qM9
zy5}`CYqZv({{B39KK*MN^YD1suajkP`+_&>;2&uGadA4k{i@hU+V&2IA>wc6d#kyw
zPVM}YbJ(jrt5*R(&nTF?c^{GV62PrbfSE2%_OC@x^bN!w@}O`|5CJe8idYR@eAzKD
zDi@moW-yGM$qni#fq>jnMB~|dK#bw{k?bZ<*DVF?TGBDdgi`aERt%JdyC&v0BYr`B
z+Yq_0;x;}sRN)7Je^JbOk(10NLyra_<t7{k3IKfg4BODukNdQuH{%^Hi|-IbBDq@&
z;9OB%U4P`vy*qb!myE}5D)F^xA{=XiOPUeQCu!zBdq2V=pnmMHnW?G$Rn#Z?6B%Q8
z&soPUc7KG$#aor`JzrWx|Nh4$(dsJXnFnw`R5dLOc+O1nOo;2ouflAy<v`PW9=ru8
zQfGHqw*pF$JQqEiVA+c5L9D>{Z-^<U!Ar^K?emFmNI2}tdC}y!hS=p(kO2`2sdWw~
zW<T2eDG-khX1#%}_vHr<9#r#XE?X0(cnhP@CNECvE?%>ujO+g76TfKVwO@cgwJU;g
zdZ2V~q$o~?0up`~6nm5xsNq1btih^4s<9Qn;3Tec356@SPmKBMKhiRGaEQ2jmni+a
zo?h4l;aN~<s2qmSHBRdxsj)X?eiV9#L|#CnxuUML%#VAPp7AfyUI+dI4o25x1qClt
z%dQMUMEG-kX1CRj#I>+%@7`-}cZrZ>I5?2?LUT+hnVeMCnfgZT(s>6Fw?WmHu4@)H
zZ43N(Kkr9{IAhyp$EIJtjEQ%YBH@g21`6cKu#dogucB?KtE=k+%v+u~c5Xw@UrYL_
zS`?YnyUpZA{8|Aq61<zB*obzV5J+NIDaVT+^_}Hjl7`|I6*2#p<ItliJ^xaY!&Sle
zY3kiy_JTS)7!p0Xe`qOmD!W1z_sktJ`)#PN-WV-G9!UBv#y6^-pdN<;LdSDMkzLC=
zNU7sGk@fSG9PU7Puy~2h7%8&6+<)obg-)b2E6?dRh#gGfBNKR)MN)w+uL54xAFoD=
zX;?wTqzh}*rb^-#fFgyzr6rnATmH)v6${_Z1#TI?F$a$5QYT-YhX?B=yg)KRgh*Q<
zwI$MD&7E`&@0nOw)`H<b?ERY_`G=yi@(E~$>~C>gV%PqfEX3>VpESxVDoUf~N}_(U
zKeJModOVnF63EiYC@-xK=Kn*vulFD1E3FAfytjd$x%^A`gsAP%^76MweQXs&9oJ>X
zkD?eOU07!f|E$VK^W<xkqAR6?n6Ds6^g*@HukdnJxgWD>*+m9?NPSk#!#8QOvUWq9
zIJ==llTvv|QJKzo@zg)Tk(*91N_%A_8yrXGWyZ-req|n=AZyTJit{62HVqh;Fpjt_
z=)O~B9ii|pe)<$mA8O|<O@&J5v()3I0>OHH2J{_Xy)q~_OKV7~dX!ctb|krvw=DKl
z<h@(BG;HGYr_{|KDArbrjX<aQ8ffQ14Bu>>y>n+fjEdKx(u|C=Mvu)yhGVzy@Qr}H
zf78ilJo3Z|%91TzLZYuiEWTuG^u*rS$p&@!RdZQ2=^TMf*1pC0DfRck&yaJU1&8zT
z@wrVyE&wBwrL5MgwF~_6o;1FHF`=&4{KVGaVH$0&gj`~h(O3AqD+qlFHts6aAv2{o
z=+z3p8rs^ffh9zAx&)nl3X1FSsm*MiynYIq!H&Yh!g(%V<cZ!ryMdp|&>2(`Hu+8*
zDJ^TLgvm#)yzdp_rNQ=>a=Aj_&9njW!!mpN$B&mlHZ=aWeod<Bt`D+T+}5!dJ01$d
z%c19juOJ_EQPb0l`XrVtVfU2r#6)}1<^6?8#aC_a=KK}+a>FOqRRS9Y7V>=l-=!gk
z{r~AG$2ez@{U)ZS2a#k|O><F4FH>3)sFs%8I3ibYlrzMXT|op?pK6b7<>NoC06I7;
z|LvPM8nxUFh|GCllhY(TK>_ifTlSgDeTmta`Hg~|>E^^|ca(T1ke*wxi)3W3hX6@`
z1S8y_5QBHe2ImK3BzPP?Y|_k|&nRMx$#rxOWqu<~NnV_|0^JRN%d_Y!v%Bnuw6cM%
zz}klm96sj~#{%dF_}p>juD}f(wY2yRn)=k<CzqT$OMtFj0``NZR%5N&^<N9)1&&sX
zD*o^v2k3a2{jx4!?Z^8EI!Y*KfF{JH_#@|L!)MSDs7q$-WYm#v12Ze;l~<T|mhISV
z)<*oJci`3s0&@7gW`FX5cA(20KivEtEj~PVi$8B6P1N9|Z-{n0PggZeCDJ!IPhnk)
zFQIsK`!)rD6iugzxD@k1eIxf+Ureh8LG=SBc*F$Dj#|0ujsATbS9TZElUnlfn@`&D
z#EY-hmE{lX$xs81{v|HV0RWyI^{b9pWYQh`W9f~uBi>1N%7^*5?%H_Az9sq?goZ|?
z4vP%d0#%!8*`(WB|Hw0IaQ`f^J8PNX9N5QL3zpqTX&QU+@(iQC);FbI3q*2I-o<Eu
zMtgUq<bX9e9B2db>{|HI;JF;&D%2i&yJ<8*R3o78Az&^vLdR+M8*z&qF2}dxfh=v5
z`QF~r8}^hl@bir0ItUbPLV1Ix#hUJ=3ZKQRSFc{)blO&=kZ+yKhVPE<uXF+@+|r7#
zGAb~NWVC`uIP5#ih~c1pW9gB=!5~;T1A?2Z5^xVXg?d3=G$o%ufBpq<+-7W=ppcj+
zrVrA`d+U2GYITsF!Lh;2))%|-D?=EQEvty_W1%&;QRoo`f&311>*27xT&*gay(#&Q
z602$RYJfXIEhV+@-=DG-zuJdH<V_5yRZ#B)WOzR@1Qsz)!eu#XA~_^j2hgDs?`IZV
zfU8^7xn~E^V!Q6%^~v$;HOj-%<5k?yyr2LpwaC_92%;el*Zy}8UGor1{C<{=S^$mM
z1-91*k`)pjcnCo=)NYgN0O~?&cfre(E~zFN>eUaEPhAVX+=u}yq!k9O8RT+9TJHil
z2!(N+?l+7tgid-J?uH61!@BG;urM#64>26Y*!nPRu@u;>gC_q6>@kp*iV+RrGcPDr
z>PPJVZm8f<G<lFl4kgci!aEk+Y@wk|Wug$#@?&S`D7P5^gng&Ww&R(tvep0Xed2j=
z&dzQf(CePPuOR%?(fcHDDl{gc`L?1?sEnYMyPav+Wv!2)FQ96_+=;f!mX1AzCk$=e
zS0ycNc-0Cy3GTb<=k`ba7jFQ53@z_;b#<j#@BUEnvpUelOdY8W#*?J`J|8G1zzVkb
zYlv9BgP_D!#R(Y^E-K{><0!HqG9afs09O>V*7e;tb0@d6vL*wtLIW}#E=lrW+u7J2
zvuc2AP1K7)bv|IG4>b&$6Es|7CK-RP+2?~Dx>M`{^ro=zlEBKnGXg>Gi^N*T)$0V;
znd>aR{JVD_|L$*c_faUV_-%7DqrAMla?=j`ljiTl<d&H^oOId17Y0!69r{XOar6>B
zZVa0@QvgqNmHV;*fi&;lj|1%j=D_5s#|8*G6`&zomsSP>3dZ)#qszoaL_}aR&Vs~L
z9#O~K$tjG~ck+Q4TK@(x5%)maxKssx?#Vq7Rk_!IBYg%lA3;;_i*4qzC4EIhyVjP5
z#Xz_k1R(Yka>cA)fE+bP7n((_8?FC_*$nPBx^{^0p+qPP5MI+V-S0TWrl#FAyDK;*
zmFT;DJXTTA2BYEKJa-tBm|peC&UFuk6oZ{53!%H0A-XQ;i%F~aA*Zev049mULF+^%
z#9l;126XKuARrx&oh^~*dm<f>xUnXw_%I5nol)GcAUy!MV@Q9|=<#m|*q_0;g-<qu
z;VGVnE&dC%b879JaoLcip+SlVEP{*7i^S27)=0^TW<hIc7Gm)gct%>>Y0Y3qoU$V2
zXd%I3Q-{4?|4YuXrr|nHgi^;*+#HKV0Lq6pu~zK8Fr^=ruJutnB$zak-`IJC6}qlf
z->K__0aA|jK2H!!Blc|oysn~x*mjWxkPN{qXY$cafzY!8dbG%QnN9zktfsy>KyTmA
zn-jde;_72$LX%-yr<>POV||tuN%x9-)HHlt^E}3l5W{E%e#pHMpYsa~2iX~hpvg!F
z)-dr1%Y%s=bQ8Z{WfgC(Cd?XXG3>Bq{;eni3EI}KduW}&`Rj4?dE?RZ>FH@SPOQ)7
zd@j+5UVCWiNIE}Lip$p=O-^<XhQVEBsuHvJ3HR$2<K>Hx8a1%DPQ4SO_zzGQrxVi?
zo=ZWr^y$;5OTgThQvelx8XjH?T!u)0UbIc!lW^e#hra)hZ>|6?ge->h7{2=tApi64
z*WETFg5eq4IjL%nPDoD@0ZfI<Kjl;Y(-V#E>^E=Ujyz$oL849?9vhF?K_(I2w54<j
zS2lW}Hvm^Jn?s&@<HB8}2K@|$PpUdPV%<{pKpC`7FlDTh&f!}J%qoe2XDweQuXQ*%
z`lwAJ>Un9XMxgiw6|)T~_vHJVsJ_TqMO@~gCVz;6;4JUkG6bj!5T0BY)mJo!#6UiK
z1p?8A>BG8hx6vgEc}zdo5Acm^-ha|*{;}Xc>?KM?1lwm?$_S|0I=db+HE82_9_(cl
zSzB?^;x}I=^*EtU(X70`peYA~+dJObsUR2`#(Pd_i*e@N|8*QMuuOza6Hs#Bg--zD
z@;s-kL;GASu9*Z{Hwc`XrrHgKzlQ=0_51k?7gm4&b!Y)coH$Tmx0%U|I6;DieDXo-
z5m`1u<@irn3A4aMEf`E<=8eaIzV}P+cZA4%+yv`aAzpa9$};cq<JT}}i!2<LDx5*1
zA29?-JDvw75LO^-s)l_-9@Rk%8w7khBi_UO$ob3+1lgD4794$yv3$q^anUT&hR%_0
z`6s(1Bn*Bwo6^7jgqwfTs+_@qNxFkiM({mktA^tlU)M(fNQVBD9@H&76Qc{pp#l`7
z3Dd1ynRh?J=$ThR`o?zX6_bzV&ZA38ZUQrLU$N>2j!M}9KelhLuUB`Cf#)JQAWw*L
zFn%x_%?Is4wD|b!R+{SxZQqPo{%?o^Wu8&!k+zk3e;J()0zHn%S|_hwc(fl!?y%R-
zHMmO8g&x-1j_tehK{n}^>Ha-G<JRNo;o*@yIh2`7cL`wmb*X#v2>`hBqK+IsEE>x?
z2{DVgpn9Cd?F+qq9%j+5>RNHL`l=3G12P#BQl>7L@3Dc9EXX6^>`><8ivkxUoC&Yh
z<j<e9CBcZQ^B6lGiKONeK7if8gK<*$MnD(}v^-rmCpi(JxUAEiz4K#xyE!^{l}B5y
zn}KBXR9IVERi)YP30I4$vd$|O_b<O-?(C<EMzeN;#vzSynlpKAMVBntw8BvYvOu0K
z>e#`!1b67otgOM<i8Vz>eE!_h_1C*K`A(Ji>5-Ak9+j)GxzPpIzv=t++e=ai1VcO8
z5BS%*10#rxR|pX<zr%+y?n7MZFlx9pRi1sSmYcB+g}nNJb9pkWNDDlpi>4X@M1AV-
z=a#faNRmf8*Tjl?x(rlC_^2b7Ne=rwqGAzxp)%!z5u%X4Tml{LgWb;J{d-JPn%I)&
zmxo{%=KiE8_*D#(CW=@zAEVEXjUb*VbV$d0xg$uxaH%wV6_i_Oe(SFRO!x+FLb6Bx
z+&|@zGCrD)^~<`(8|)^{{DPzNYpKv03LH&>NZUnrn}-7zV-dxSqxk#uw7!6&=~lZT
zuesmZ{l)j8)Pq=54zdrqO^qo6z$jotLhJv6E9%wLx>v6*PsR4V(r30jb4DYb>mxj*
z8aDnFgUIfM>=_O2f_A+i()0Ku{`i^jFG{T6uz@LZ5!I8d{zb~z?RCM`%}V&BmQ*3#
z9ZAKByfW-P&<<<9%aC^M!QV27e-VhT>l+2NRQ3X7!YBcv_j5qUuED?X<oKQl;zAVH
zS$zLdOCj{bc8;vocy&`w(04F9SIA;12(j$#0A_!E!BqYSPuOB>xt~0NKOn-`|Bsuj
zgqc`3A!*B)OgpmT&{X&WH_F8#0U5>cz0%Tbq*d?8sWH(IpEf*ucn+?+tgGGV&O6(X
zEN7QSo$OZv5?vLormFfK1J{3iGwb>`3K2T<Gv`HbA`(Jme_3L-s3`)+pPRVCHHYDn
zV(4zjsBngV-J?*5;=GC<@V*h(&D1%L%Setr`uW*#MzHaW&ku>&D(-i{CEOoxm9)#^
zE~?i@eX>-jm|o7c!<@IDD}WEv-QOX!dETU=@wM0|c5|>JaY_K+pgBxhQUMa3Kr8hp
z2t$e|kltN9d$;d^a@=;1sec(h@nA<q7ukamzmY3ifJ4oq%I&sx_3*fQ@7~D$w5~1_
z^yX;=Jqy#)(lT&~2?$s*l%6^>Q3td~!lmzu(E$ItKi@V0zPCFn!3%{Sapx%r*F&A;
zd9E6&RK3#QQ_@H%KD`bKWDz0yt9c0934YjM`wV^K1keYQaRp$<<ToPOVA~%GccDqi
zH7s!Hajx}4Gf>ZMIiRX<=eozge&x;7-q!1krs%7nZZZ$w%fpg|6d*<SnHi(kvV>n2
zK=Ujru2ROhsvHS;Ys9NO1&nzK=vHMkl4y4sR^e#qFk5SL&@hC4lxxt}D85(kY`cvv
zLVM9!*gAbdr*(LZt`5q7I!$IvF#3pZh61*0TVf9T%^T#IS|Sf9qsl?iz}sB<4<Luh
zjrXBw2?Q7wd<qGJF_iKrkvG#VVWT{+XX~6_nC=a)|3@7mfUxTru<!|gD>+QkwLj>S
z<a*Q9!GCgoU`Kv`YfL;HHvM~w&cRfa-|k8j@Z`KGXkFiznBgdzJ3FI2tSJxRp+L!_
zT}xXB0j`FJwWJ~XZ}VWF$*3P>x(_nnpxmd#Xa?Qw76c8LCN4(y?s~}7-*c2hSAw}M
zaXykzQRaP7(R|m&SDGmQzESUF7n(9W*~~CP=+nr^I@q+d&Z5f%0(|51jD^TvZPR>4
zGT~edW0LnJ;|NVQFl#MzXeTLa+Bkmg_g}v%fCZVKJ9mxP1VJCkKShu3XMpiT0#Z^&
zU)-vl;qd`}t-q@fT^M9(vaanUqHeFBHa)g~_6Z6$KxZ5|L!@G1<*C`L>B@Of&MPm~
zU};|?qvB4R!kv^RUyMuAG|#-CxVsH>znmClc5dzq)^~D%Pz&!JZuwhgVh3w(5GAHC
zNI!I#|14?8`4y+X5r^t9`9W%1+WaFmaA4^l#Y9DQ@^;Gw73QV@(S9Vg3rx^zh%FP0
ztOOj;@OXaKf4tk%MAvHZ#gF~VFR!cQ<P6!x&l}9fS=kq^EtH{cm@=qUk~-tYHz(<o
z)=D!sjIg?YKO-$IY^v#P!t))E8@d$K92twcwlH^n)Y;sAua~3zQ^mkNx03VwO%m(p
zeo5srPiOVINlqWG_q{jad&>REV19LaC-h}ed`fQ_wW+rtTJnUEcc@<+71B~q%=)Ar
zO>uH^n#8zK1D^6pZ1Y>+DsK|uI09h}qQVBLqV;L1sg*b@mXH(lp$`{{Ra4fANKnfM
zyGuheI5KtjdbM3ZgisArq|?!Dj>&q+|Ag-P_STqMwR>|T4|A)Jy!s5^c=J}9L`?56
z+tAlwuf0RYr}%jOZ{SUt<2SuIxwxniqWmfEn^P1vJ>)rZH&etHgD8Y;8@U5!Px$$j
z<L}8&FYV*p!olHu;8mF*#wI9TXh*lFTH!e;Yn7mG=YwJj@|MkpWfxsCRw(nPxsR$?
zd3yaL3Dv9>_~_Ug)UN&Ti$9n-w1^o6dzRYn%j*c8nwZ_VX;Yiv(L`t1W2{=?2dPC3
zc5vNdl2sCjv>NpO1)rY#^QRJH3{;)8yQ`_QC3wPS$Lgkf9!_Cgv#uOX9c>M;U<z4$
zUsEG8n<Pg~ZoBUfGn*2?E`c2W5aQ~wJSsu(-IKse_9)SN95;RoOx6nA|DLU+N#D73
z6Q@9UhN}<CFJbD4P94W8%rdORukVM!c{m^gWgnlim%P8caI)M-#k~R~3@rdvU`h6)
z4L0AhYD=9P`gFs|hlWI1?ORQkS7?x5_f<b3N;-=o=G^Pt0jgaI?m{(Z>g8&V%IAx}
za+BK-$4hTgw=&GU=2+DDISt5Gg}t4Khpqf>c>mBe;<sCwjKsYXL8^9V+6=;TKFaRo
zwxpyarY-um!H2d5A670C9YASM_R*aq9RX1{Wo9cauQ_4LQ`yf(6<2)BWKVQvHK_fJ
zm7-?8RAgVmtQ$eV$70g=U}HQ3)8q|{92Iz#RiE@bnPLtDH{__B0gX`TwLB*Z5QpdN
z@B_h~o9I=jVz>n*z&;>Ax4?W{4-Jh(c(Dc5T(xQx&LP{o_B-$Z2DtF#8&bbYv<X(!
z)Tp4*hFe@*5B<m7NTf+0CHePwBUMQA`6cb92s4eY)r^v?#KdD5s3i8|_iyrhhv8uX
z%(L=PuTRR;=O_s6!)){#k%Nmx<RQhaRUWze2uA$d#rikxkkG(Bv~B$jmOA?HC%J$&
zYP(4{)AZ8FQPR1dn#$chfQxSuw(BglA}u!Ez>5Ope_iPDR1k@I^Z{Tey`PC?Sa?dk
zu1>c#pg8NBF#OL4#2nz3y^ea9Mx$vdTqyBdC?&Ivw($fQ%}Me1m`6uP@5N?{hL7@5
z>vs{iZ||NOtZVX>s=~Xb9}x%Pb-=sVDowwr9ltM-{c&;IX>T>OwC(FrcXxks7;|Gv
zYk+?6Aw6q^&&c(V5b#W~iVhB0Rbgqc7zzS`Rq@C$85GK%=e_X~E+_7Jjqm|#C`UdZ
zkHDJ-B!|R?o}L~8S`lL=w8kWENf9@z^78pH3>dJqb2xPH2yEIOhdrDYB}3?0Z*Z>B
z+1niR^BJbJxS~RS+_rzlW-L?V97H5{A)amTpM{1h5@iBy?jpo!!sywFSQ<hb7#sa^
z2(s?a$?gTG5^xc1AKd=)UpO%Es$%Sp(CIQ8*H4$KGeuf5v=wm8<dkIPkElX>SB~W0
zuq**JTWprMph>X?Qr%R1bmW>`kVOcCrtR*R1pgN!eIcj9#gJcESPS~O^rt>p&n-+$
zJg90Lq2EA9dI&mQj0n4l;yM!MP1QJ{waPt@1F@%u_vEspAr$5|X%HuGVP@WoO%{d4
zQ^G|tIJ>yGXuT;jQU9E9BxSWPptpR;sslVG@mwnCZE#QM)9S)iV~G~HRV&i)k*gp=
z7DN&!TH;6*Xd51CL~ryyC}X0L-NzRAIXUGVf2sn>YStusSr&33%MTl$fs57Z^n|hg
z*%>f5HNL9$W(wFMhyIw)p=24r91(#gn-MH;3%h%w@LUO$o9+s2xi?BfA%-u5y}n+1
z8yY$6MYa5skT_IB+4crFJOr#fSQIFU)D}Ozs;POX%-w~_UB>}8i$)I&4b`BE;=u_6
z^2vj%8b6gA*CIJ@fJ9L&(3Dwby0q{Eqrfkd#%58wW<_xI<Zu#{fR|ShB-)wDsufs`
zXdJV<RCLzt>*lD}ELwJn^~5p?J%$KqT@U)!49%JGwRCienQv~wY$4+jaR3<tMoXAm
zCN^gPkU_BH{kwPX#?GQ?g?Rxb+Ii}aKDhvDM0j{JC_9JqJF)a5h!1;iy1)u5z!4!Y
z1x9l@xF_garGbs$V<&83Xn~jGz{G^}?>W?V`vT6EGiyiwmlgmHkBWHeyIuFQAZ)!a
zD~obFoiyDf_{}Qe3F>X0K?hj?^&t}@qvhr%AcUgR-l$Xqe!;KI(nA#$8*hgbLVoOM
z|G+>76ss&d$b?Bf=&J6%^g<zL2SHmV_NcO~4o1mDK6r4hK^!7KVS&Y0qm__M)Wu8d
z1B7MCBIM;m<WK>QXza;^0fgGUQK(WLjE#-a7qT5Va|y!PXmsg_87(1c9f9|`76hiH
z>wqJ2Cc`T6VaY^zLOsBZJ2xt}VER)vAc3jV1=yR9ryrw9uiLu^r(M;_8=a;^D>zwh
zC=ofo#DsCViTAv}Nk0Cq-O&ulI1&bU-dVV5zB(cB7(@(7qfNfs5#cehqz>bJY$K<H
zhkSkD+wNY3W}AR2H-Vej5SeJ4HS{@!TfB9C%r@rW*{g#xJ2AcCVM0P0fEJ^X5(J@6
zDb+NZ7DOf*aO&r{(T9nr@0!U&f+{+b=Xd=0KOJyhV1+we!-6QPYm}dLSzq)9Tz%p4
zxT=#WN(nrm9R$>?4h{|uX5v_8q5=+Z1aM+9;sQv~KCceO{Yl7v&gH$w3v5A=6kOd1
zye7aATP^A75_Y358J|W)mG~g%^fMzn6D!?|{$p@g^}?i$W?a~!N83=(jG!$`2eh|f
za22|=({l{c`<GreAIU53@3#yXhp;ZIqz93RKIb!T)=Y$r3>;_HtmxaflgJ+;G|__6
z0dlg9A(d#ct}xSVYe2irhC&xt*S~KLT~4&YS;BY`KM~0J^wBV_EbCtL9_HAsXL`M;
zyX-Ln{mJ07M(`xv#V)VM7HBt0{yafe648m6*_@UC5G%B28b=}-H{?DcuJtulxR1Nf
z;y0>XFxIA$zQkdtHYggp$4_7UO!qJ@T5{qn9B#iA0(cfLaWss~dS@(!;Z$u|UqB+-
zI4uF`GWs>2l_d2fnx>DeJ7cM-@f$~EQvX7o4*RhT^fnZP9gnXPh2a#oQ7*usG;(&p
zvn7`FK!DPMP($ChY3*7~Iy$=8QD%RC<)=m4ih$4toF(3aZqL9hvt+fwXRH9<Oqm=;
zN@9nehH7}@T;JXn1Gi2syLnI#nGnO_*r4yU1O65-aTD~2WvT}U8Rc7EuX3w<|6aUw
z2z5**$%A4Ctgw5e{1?lNQ1y(!gcStmq4u}hG1U90K<s-S0E4Z5!T~FzZtv4o6PI|7
zz)!?F@>*KKNe*xnxN6?<TRGcWowI6<fu3zc$2-Az)n_*$q$usg(1DJAZ>`gq!?l!x
zQ3+NJlB7{%{Ntn7SznL?w0+uX>Vuu@92l3v15Vd=vI||)wxgRv9D2$OaUrq285$YM
z#Jnj-DvT0RemQrSreR#(dZrya9P&b83xyd&H)~~_bcJv|2%CLgXaGKs8$%U3e9i7M
z&nBB^moG0bfNxg|l->b%py<hdIuZP>X}&38(Q>m2^r|;Tb|@d2<zS3L;`4h|n=I(O
z+TiEOIVnJ(SR)4zzU@$x$1a2-CXL`2YeJwU=X=3xVRvMEJq{&twktCWV>|SI%=gah
zDGqjaN7qhru8~RH(e;aLoj?YcU}W-*`Sc1TX3a-VSxvhwzynRrF{t$pRtjuwTelM$
zHultK?*mtKwqWYVuU{<-P7kbrx5v)8P^r3*rRzy6*ryAP^H4M@=0j5*D7;Bh?!tQ4
zQpYfMLN|2hk&a#OQMWuq>c)hwY}yMRi_(Z(iR`7@aF#-Zj{wFgtjOx&?yduVA_cLa
zpnakZi13ZRsmVz#oYYJ-@yInn@?Go4p~A0PEhK~LSA}l%>W3)$nb0r>kGD`(SvgsV
zDAbF#<|IetM{41v>c&1=jp&ci@eo8FA2V}wvK3esJ)CNBn**YU_~tsDpX-9o6#J-;
zGE^5y@rK{}^HT6?y=6G`)Uql(`P8dEZ3HLXix-PA3|t#)o(Y61HlAL0H=i+I8VK`b
z&s_eRU$#nGg<8rG#8qd>UvJz(0nFHpR_uLtf6%rY12vkcXI!;YmexpvcUGIH<u^ze
zQ_&OXAnaL^k5B8>yelPI)$;w%k7sXYAX&ZZTL$2{Zo%hQi$0U9UsU=jrFP{bc6T_$
z9qw4ip<iBZ;380<;Bao`Q^W;HOsqilU3TVxSJlcC#QE)_ho6>|WFSViAU7KL_RYzC
zBeYL?B%}N_4_5wR%$3bswj{&CRu>Pxq52`fwoF7umMmkV6M9u)8=zEX+P3XJ63Goc
z0T>yR^wa;&#_YKibOsXXWAOFsnqU%0jEuD!L54w{HfE`^-2d!nxTY#uVn8_B<S!uG
zs($%0$;J@rYcq%;`qo2-_?}`9Bc_R2yjfFMpN3S~uDARdSzhugX8$q$tiu$JH%M|l
zk@EfJe_oCG8idp0qoW@{Hf%Vu&>4572rXFKyK<{VNLU0lap&+UtG!l`-KC?hZAhNl
z;z`rVgC}*wkX}<nJ|yj86Cr@R<&XMgugg5?t>md2ed_JVEt(p_UUnh<g$vWk9zo~}
zuH!J1mXyqZ560JTMc7Ap-VGh0>-Lhp;}ILa)(a<2WcE>nq*6Nw7Hv?L2S4Tj3g<^^
zRe;Q!;Jg(F(L>l>nBPbIdFNjD;XXqwdXPMu(0busCLRQ_z4PsN^l`*K<4_6DhltNc
zPx_kV_y_EBT_m&vLT=+@W6h98sNhT@e4c)mi8`s;_jJY4twz@`2RhuQWz)2JeWpTR
zPN(LpQgQ!lOSb?e5t~LQX45+9qbfQNQXSH0<nDcWiNAs*<t@-FF)=Zvfol+3Jp|*s
zW_DS5^?&12LeBzJHTnhcgQ3D^PEJ<Vrga(c7q!-90;TFx&{D;=ok2X&frwGp(lXWV
zV_wqBe9Y2(k+@LPl{yw&7WR9h*O~F$=U+QnSu>#yCNX{;96qv6>PDuHLHK2JS_Kz?
zb<2+@7D6(SQRD&8TqFLFUgSaVuV>RekH5ZCff8b5!hh*$)!C2938+O)OUHw)Viy0r
zI~u#AIjxk(YBe2Q1^~IJ;g86?n!4Ix*qi0y*Q#aOp+ZRv-p$8GQO?{Lf1`zmPkQg(
zcm2#yUsbiYmozsx1Y^L2FsssWdyvW6gxyun5Ej$pOKXO2lS=pVj8)N%Hau3*#0)+I
zO(p3FlOx}XgvmA>!QjDz50Afm{rV*Kry5^cbtLVXZPOm=!Ymbc##61EBp1yOI8lQ`
zRI6OdCnG5(m5JJW)Pq~k<E*ZQMY44qSS)^N>CP-)gIJm`aEQn9V#0@(eL|M3jS9_b
z_Bnj^(=qWaqdAo)&7sKv;TuaWpg)<9`g$<vr_sN!9MGdaG{}6~eV&L)EoO_8(ZZo0
zk`G<ZynEL;X3J9n6{<?L5DFnFkQI-GO&>s4F#<7+xS{-TM&>}N-j*hONOx)e5Or|S
ztvgs0>tRvjEZUgR+6?DV;)zO9<A9kosw~v!OrhT$>>5?}iC>eXmWOMjplL?-rAij2
z3P(d10*$(0N$|u_ur;KAY4~?1(7b-8h-g=d8ATvJjFoh|{|)0tBJx$j-WW>!8g6>`
z_cUiWr`%pr_U^4!yI50OtAkllAI9cI#Yn>-f=W!wj6wTa`9J`w=YGm^xa~n|^6S0y
zgiTnNu9lj5bt~KO6R3XljvTVj*f`EJpld3{^6;dVI2Cp};rS|evBz707pna|!;xOJ
zE5r{(uNK2f;;OL`?3RQN@fl|*q|nDsS>{0CeAg#gRtfLk4$VBVN1G3c1zx>MV%PAu
zqWn=+9)&;q;(Gj-d>>Dn|Bdr%aT1T3r|Y^rz7k3HFZ`z3gqr~5w)v{>3l$tuMT%jE
z`x2>rcI{AW(#h2Y$f>kC)f^&ec<%nJz7=+NL|~c{CF!E%yt0MNyJT{=j>>TQ8|8aZ
zs;0K9%kGN<6s2u9lldSsYV@aDOnvi0Lo@oL%|<E*A{!pe0A#&O^)H0#3m3mKpYAuD
z!_qLU{*0LJ+hH~2%KG3*n!3*6T8DQ5KY-sl31sRvtup)6s)cxY4|r9Fv|2hnAH@xc
zhs~~@_aDYUKPf0mGv<nPkGA2?XhscVHaPjJr{^?MO8x0rft>ts(r4U)@I*~eH0kL7
z`ZXPv#P&bmR5af4TIXupJLiy8`XriijA^RKXgTE=or}9W4V79Y%7VS>hksl_0nM@x
zy$0oF7&qk#Mwdm}BddKb8w5Vdt}vc<H_-Y+9l6*)m>OsZFd&VniGzLN*OQO^Nj{~7
zlvI>G$9;lggZAPD=R^YpuZXBjZQDRkPs0EzIdEE>VCKY1oUA;j>JC}q=4-?@cY%kG
z5lSa*_#G9J1ePRt4Uf=2HUOM)5=oIV?)!#G1a4CG#7R+=5z(rjKx%gpc4w(X(29QU
zT{v8^qn;NpD&pr3;3!7`O(L~D&#qmGCPQZ(v3yZbW^~#3d3lxM$x5^=K*vQRfiQIR
zN55k<p!OP=s-{7z*7x~yQpvrPloV)L3;}!bLFRS~gjq0>q?W$CLX;gfD82`PBMO7H
zwuPz`8YBCh(a})>vaK<w&jg0P0~p_RD={%q>!u5$Ox+Bi<VJX6h4dXt1KJ@BPDK(m
z3e0D;k)EfNoW^qT2nn5nZM*;&2Vs=c8M-y|x!h*Hzaj_caj_}k&{s1_-+<+^;;94~
zL2USJ_+QIT_rUYTbQtwG6ix;x6?TzPWhK?=BRV;#@l_D@R1;)OQ8V3$LL+HqIFE`Y
z7|lkM?vJNCfhEE!f`+?V6JT&?)CF8Rgs}%@B-B>WgPL`w-hYvcP(Eg6y^xrMp@auy
zco}{68mNnIEZz(p8y`;t)C0WN0Seu?`8NLr{lGWc+aRvXhnBz)1iP%9oNZ4u5Rw7x
z;%Gwn0MQVXIxT0az{v?bB9a_mg2)Uvq+GK;q9)6|Yu7PQHrybLh{;bSNRNlI?Ao&K
zM^{775Wu~6uf8%p+>Y~%N(}M%hQ;TT=7E$>;y>7Drf*RFyP2Y$aJc1Rv=4B7%sJsE
z=={PyT{>K!;mBGl0rn8|d(TleKn=IB#8oJ^`XNx7#9(w}K`I&VVAc{lIFIDETgrl(
ztY74bG*Cct%Zt)<kzr$$N@bV?H;FmIv4`d3rKqV$8z#}pRs$>)je9~&Tht9#u3k-W
z@_sZNL?SS;3c}KZh7o|uXe8JNncFRtss`?5sCKJyQw;<%pFYqL4WgPHCp5|vf03pI
zr)<&zg1PeACFeh1Ca)8k+eE`9_zXoOJ%uZ18G7MPl5K%3Em>~JQq9XM@N~Nt1LzhW
zl)ifHfFf==Afx{(LPke+|6UGd7YpcKa)dA<qs`S{J&S(h#*JhJYoK=edyG&dDq*!m
z+o<@^&<$dQOfb=uf#T&ye~3Uqw`{JUqU=7WXhAh32?MbInXEh1TZtq1(EG^Bl`AW8
zN(ON9Xehx|5ZG#EAUnGXICAMg>_mUR&V3oWJA(j6&HQH4U_V_AP&O65DW$z~V0L$t
z1OvjLD%^~@?bgm=BU(94^=b9Jy=Lf2<jF0c%|si0i)ClowQJYNu*S)o{_`h^U=}c<
z-pvPDh}vAB528r~f{RsX6f9b6P(s7Fi%Zyd);%OlSF`PQWy;d)qf0aDl0aqA_@zAk
zv;@f}b}WI2xzQ_!Dc*cYp$BC4+LJaoumJW#53@>1Z3%ln14n;?$ALcreHcQ|CSGf~
zbP8|HigYI;hcZUR!-EZ$XSAbr5Uf2px5Ow6y_A9g^&7IMpO%(Z16kRP%fl7sw&cec
zw#8+X61YY4+eekj@855s`I8VB_)Rs)CsMJwN*8u>aw1QIu5vnACI>Votwg2_oF9R}
z8|D#$b*@kNQYfQ&+RA*!tCvx9tH8Zm0yr<N0W-sg!tvpopc|kt$;JygjY2~OC5W06
z>5i|euV=ok5Sj?V$e~e#=ZmxU4a@-m$?neQ_xJU2Lj1%znSPdxWm#487(x!f4B?Rf
z6G<Cl^+VB>gc!pv;*pj%2HsgwUtgd8Xcirha*txB{FtJ_u7WkBL2?BeJL*;PV<22x
zP!_w)b8-UOfWNd5njpxJ#1z@Skr94mtX9h6#TO>-!A~&tD;3v*(@B3NkiFqRofsHw
zkq5VjenHmniuN@nb#-PO{vT6QP4lmaRgCAK?@2+2l_gs2bV#ph8upl6?jgqxZuO%J
zPhLu)p8}vT6F$o}9E%^*(?v%4!CJ~BjN?Xjg)9Y35QDSxgs|A=sCgeGL;5HKvbH4L
zBO|*T5kymF2**b5nbHN6UHynxu{9*Ymo`=@g+nQECfd_>{pL;P)vK>*P3wkhDWfRa
z4J<4ilmzMq28fevo6~PQ<;U;=jEMv%Vka~;H5GxzyKYC=TL}WqYBt%%z@Ryp`W?><
z0xb%jdrA<)LwX%lR8(})+dJC!Xdv+(Y(G<gqFH$&o&0Wq<tK4pbK!LN4G;6-aFkxR
zNy96ShQcwWf8XlWt4Z93smq&nn#R#R=N=d1#el-_<B`*mQRY*Er5O+i6(IEJQ4WTE
z7aEunRh51XX7UfrVu~~`S%r|(4*<2gZ{bht2gu-xOV0&|1u^!nL!OvEh)S~Q$;ZG=
zxCMwWt@t8bGI5JrHjLu!M`i>}VqIM!Nq~a}4JKpBGw241y{<iQJ33llCs#>DB?4}2
z*KrLarbZ9<_~MNr9G+;fcdH3U*GfAQiiS}v7hb!_=pC|%I3`N4bR_*&*oC1zy`XO^
zL7a!29({J9urzGlK!3lYt7|?+!zg-q6v7zsdO|{i^|_RL2tCzEeL_y7M<%njs?SOc
z@xl|7b?9I{Onni_^+ta4oE6GJC1M@FL;@y?@6sAm6BBM>;r!+W8*o{-q0dposzKsU
zvIp{-a&w(y!mHP@AN4bVNr+f922+jf>54833xP_O2BIWNl8lKd@trFK+p2<Pg8)Ve
zpmud+9u+$qG2&n>dWCrbf;B8K7<dwl5u-k9EN{TQ;WjRVlo?P}u5dhA$9Gmxm%T%!
zT7Zk|qzFrTF?OdiEY1Y@I30N=oYPvC@u%!adv6&77#=CdZKztRDl0cbtM&r{2hbEP
zLzT!-usA`A?bR^$bE-o19DG_qPEH9$PhH1vyxqSqtgbE+@H|;N&6o%fFz8Md^kUz-
zHDx9~58G(ce`(sEle00$R2tN^8GY2y;9xa5CHVJ!gM)iu&tSb!{LUn>*9^vovV4EJ
zB%~gXd)G5`Hpjp=A<Hm|sLI42*0~Zj*4?`gOR`XwT1U!3ehdzky?emQ_L|Tz5;NDS
zK{c{hpiOYOa3LB=1np^EDt}818v6K1P(f?~yU_?GFEBJg{kv7T@dxm+#AAEWhAWAj
zE+*%!3L$K;VEx#W;W06M0JN9^?1#v#$DWGBg;HU&NWv*3y-`<;%~4hGki6vEB;mUN
z0UBOsh^)t@7qXZ@A{gR6NHiwl11EI(0Ym23B3cA@1b~Sk*&uSU*TQ?v_2m<+U6lD?
z;}j0fwJ|8vH{cma;<mWDVs5qSzgxDu#7zP2bApIjTIja@fo+3h{CG6F@aqr;N!@<&
z;zhDw^$FCuI20y!-pLsgKgp3}U@+I)0a8n>Sb?S!<<}Mt(NRC5o19}Bj6}#saNQuT
z6FFNlIXOvQOn6k3UUQL=K@p#N=Ue<~|6S+ugRw`^IBJMr8#ArH6N^a}y97ZQW`NrY
zx^27f){G#$MiT(5&J1}Uxek+k^{<RHo?cGTtz8wKzbx@eY*&>*gR;FPMfvP0E%7IE
z3JNut|4u)MvabpP6jkx^#K8|LW1b)|`puSR^&@xMHtq1o`yWvEpN(jk&+J;OuN|K6
zpR25LV0rCJs*Vis_9zH~lP9(Xf^F-E_+dcVME=)f5dSPt;xx<`^?0e^NSqA_$|5sx
z@-e9{Wk$DP>#0=hqUaA*27WYZ-qptN{3J(Qfh3`0t>ch#PsdgJb``_RZQwxmArn1x
z2Zv)Xn44QDG&}e33cxfv6ln`<tsC+q+?z)8j48ode?QyTq03_D*(2+L0`&)Mk~t8I
zaqYq|QGK$3^RggX0IET-&<}cgd)MGhRmX2O(}gD86xMhaKVMfOh@m>riFBAseYTQP
zZFRDR&6N(5H?4l7^xj^ltz5M?!N(%N=FED;<wLQs5<Fs9?&DdYUb&57=FgX#p@RfN
zNA718U_+rWRe_zvdA1Z?!d54YIh*WC;}4mgXg<_Kg<~6FlO+HH7$j9zqDMguU8QHy
znpw7tl9>htvZAW$jfVN>3**l6>M<wkQfb7Ce0!Zg2436Ap7@P&D0EA{oGiSgG_AJg
zFQW_`A_|SYj<=f^=Uh-yrd)AgxQ&U>Ssi;VLxKn^IHe@!{Op%rq-y0V3XMal9S<gU
z43%3N!iM}56!?6Io#H`jBFB+AipF9}{Dcy6cgM}4)bXBYou7UT2jXeCq3y<BqG&WA
z7~h|{GldlCbCR=|8MunEu9BuR_mD+HN8Xru)4ecKU1sKu4fjQc+h9?Q=eL_sUQE>B
z=CTppf9Q}tq>Ro1HX7~3s)<D2mt{y|NU(QgyiEln<}E*krm=zSqMpooIBGCK%Ax`t
zpt<N0NT*3;_U*ekb;Ps%a?aiJlm{tpLlzW4S7vG64Ixm98jd@sa~CJ&m|Sus+%OZb
z8swe%2SC&ITJP_l{y`G?k(W~^YJT3Q9X#aAG7SAg%g*Ic&iov79&fLk`tie@=0Hh*
zNyuWbPGYhCP#D4KDPjZX2`eec+>i*c)Xq&MEFl$Q5D%{OLie4h!VKYR;G)yEiUBY4
zG3=*Vu&Ep7QJYjFpgC@4;E&MjgRG%ORDmVii6p9!4B~_x>bZClMYzDxU6f;K7f~Ig
zVMwC<RIk^lpkL~ev_`(u7gWhHl_u<h26@K_rStdjf;Tndy63T;a-d&hmLIP*eIuBm
ze$)U+a?B6(;(*IDB<0iQpF>Ab14W(P8^sXNslre!ro3@L>gD3*_E=Mga;cikP9k^R
zh`T+KqEI|0U;A2S0O2!S*(1DoXH(bcpsUJ1E=su+dj^U@VQ7a^X2g=zDEA|?L@*QS
z#FUa(%kAgSpEpEk%kEs*yo{pZhV-xs8+HK4K)wd;wTk}w8={bl5+33XMkicmZQ6)F
zN3&tryuU01_oxi;G?a5WMTZ-ai$bxxg|zyk)XI;?K+hjnhy-E|^N5ZEoTHCWiIJcn
zMNhf30pjnP^vBDeZ>m2$bpu@njL$G2DF#D>+VMASKO`=rvQaMS0<X|0oImw!;diGn
z1VG7cI9E^1MO$0SJl$-8QDncSQ2zcRq&xVzI{DZiGrdp;CAWcw0$Rb$CU0p9Z^6Qi
zc(&_A#YCZGmf#|8xzj)fzyqV_1K#_Dx*h*V*9fGFBRjr-NmoKzdg|m!%CXSnTH=%!
zcA${?@fJ8aw`|^g=8KI6(F4Ca^Q0HrAvXCHls`u>TIklxuS_U5#BZ!>VQOQbB;HTS
z3J}A_CDJe1EB0$BXN-Tky<{y8!>{t|C7;FW`-3qGWo4_9>~S3fH8rm*n;l-av_zU6
zUqQ*dg8M}Uz_BCys!YioF3K@A1;f?$WME15NebndEP^NXJUbSAzPu%$iXZu(UmEeZ
zX&e6YTNKK+T>zc`+t<Ehc)<VmWpD}1*nj)Vqx_9R|MA^3%l^Ld|M=dE<=j>){^L7X
zPP%_r{y)CG{?`BdfRz7>R&a*{JB6Zi2z9S3H0j3Iur)J_kr->WAwb{22p63bqSydY
zC_Ia(Ys4dIG+FP>6bi2{s*kDx2zQ(Lu&)b9ubJ^R3Z-QW_M3)0K;Z}+vcEsh?h<mb
zPN|jnDR;~P9gss!X{Hd9M`Q{eh%5LyMwekVO8&=xd?iY8{qNrTtBqG9WT=7Y=9ZG`
zXf^-$_vx^qA}idO)P>a4W<3cpctkGvU6E(f4U5-OZkJl~?_d3&|DAqcyg8prp6frp
q{tr0A5<!)61UUI`UoFkqORIHdGoRPenCr=(DaoCZ%{g)A^8Wx#+N2i%

literal 118726
zcmeFa2UL}3w>3)Q7fm!`ELf=a0t!JuM4CnvK><+^=@CSv2N3~jCPqQ3iik+F(VHl}
zCXtSU(mRn&QQDT?*>|pOzV_cS?!D*!?T&Fj&Kc)?dSvhSeV+BKHP@VTJ?GX*RmJ5?
zxtDTqa4c6krf`abW0?sD$6|@E7vpc%FA6$}|FziYxS|3Fi~j3TUUUEl$3_k%g+m&S
zp#yCW4>ZQ7<%dkoyf^CqO?P1ZbA>}|-g@4MY5Zkd{n;mvE`F(Qd~(gxca2}(->9_J
z>$_D4jpsd&KJnyU^Hsvm^<RB){N8r&%lG#d$d6o4UMBtWrSpYlIg_sXz1a_(8l;@h
z{+YI-w=42gtOg$9?>{Hk`q#w`@csRd_}`Xq)|f2%=byJe+R5|JKW+TtNYg+6#PQ9}
za(c19evTbf-}m?b%)xQ+ORpDS{PT~iPTpJo&p+M1_Qdj^f4cU!fB(vp(Jv%9PS)_b
zj=dW$+YzBLfscJuY+e(ka^k`Fk6&cqV|DXRt4eKacj^r*Z^t9t{#~p}X7yl8M)H6a
zhsJSyvtU!M(`aV_Yu?7h2CrkDQ0&X2HL(^Sq;vN9k3Z+)rQ$LrEb2AzcX5e2fqDP(
ztk)8Q|F0hJf5VFW=PUmwh1!2{VGnNDZI~~EBh>s<(Vsss?}(tM1v_wJ{LV*McIj|%
zTpOS>@u=AG&!3m<U0jl=n~^lI=Szz9gTZCLf5Gu+<bTwe>C>1%<nPqZxiZ+9J6h>&
z-IiS>oHMX$k4~!RFTecqK-$4N_tPKCL|vvnJrFZh%}Qv?vg_T?%pHAlJV-c&;U|3*
z2g-P1ZhH7*bVPBDnmnJs+|0$-5%L9k^-+#*LL~k7Sv9Ybx`IP?vFU}0xJ|oEUqiCM
z^z?KJzUplDWwYJ7>AQO>!+4klEVmP32PBI9_;@OUM3M*m)i2=1{pg8SxKF+M@b>Xx
z%Z#=w?{FkbtE&8uc&{z`<Hj;Z-ekM?Fr%a-)0ScM?#Yqv^q?z)%}1)jWf`Sg{C#96
zTgAA=FMhXd?QTzLX0F&!0c+us?mAk^xYZl>hs2jjWq!-c@uSfFFCU#svufGLwUX1r
zSV;Q5h++8f<Q0E@*}Z=9AD;#6^<y@@G!@)@3Xh~JJKYtq%}qo*@2Y*DWWlt`8Rx1i
zgG_hUY;1#}dtPB$kg)#4zM5#YxHE}5+mc*6TqjLJEehr*D)#4_)kJM5@n0|c<QU6v
zu5;bCp#pKNN5rVb<K2p<@Y-t2dG_tDi_?rxwcthO+R*!<CBvqEHc>0tz`X44HnU!C
z=_{6JMh9}3`LQdv9DTNQ*|L`l2|{NQjxFci|9x89m6q<YNkum|H}g+_{vJEt->|NQ
znLj%eKgK-G>9J+@sZ~q5w!0ysty+5Q?O~yL<|O{wLGkXu%cu)~uDtX~k59?U7faIm
z{wIxEQM2l22?bRgCk0tCTpS!4N_^5+d?q{G<yB?JYgf(o^!Aq6xX-q%mvQ__wO+30
zfw56vZS1?q<3X!t_nFu3T)cvxcZ;H*k51a$_$7<C&(FQeoa*#8o*DaSJ2zaiuEq3^
zU%wfC?;#Ks_7tzDo$P?WRSej!Qk}j+Ku(4+m7ANJ+4In%lXiHSNbdK2^@$<l&C9vP
zJuQ;5wnwQ&@O$uGIa|-zTZV8j4R_j(PY#%PSXEI`QHDdr!EsW*?6+NNx~DmX7cJq4
zk(=|z$ywTmILaNXKAti^(Gp~?7a`}`*%hFsnmX~=gP+^D#Lw3%qb*&7QHf}A{Pg>h
z)rn!sk*j>8FYBaUT&X5AtZ*5D*f_E2@Xcp@IUR1ZCgY2|Jv~u4#H?DL9P#dKGL2Aa
zF^E5txL)kSiEptdSdxIct3^$2vxCNXVcX&jruUrud>5ul<cm{E_(L4|=}}{1&CM6{
zZ(5RkQB!tF(9erFG(@QK#XV}zb5?wM)VExGA}MD`IrRzx+c?B3eOFDRw@Ci}_?(B&
zw)iAD(xMbBj3*bh*L7k`bl0p|^NcU#js6dG*rG!(DK}5Nd#b2-?%cWNiJuoODb_X(
z)9gIZ6oYl28&7ukUBtQn(QCWVpZ@s9wrpF_y~Zm2ALywz<~@A)vtsG*LWeage>eG)
zgQLzHHSL;t6vF6T?CEG_T~?L*f?aL0VZjOG8}A}tIu3VE&33T(gACoAcx?+;rc?;$
zC}2yUsD>SQ_vp~|Ml5dUud7u(A0PH8ekfs4UHKB>c@)Pi-O;_fJaG8as}<3F4n4u9
z7hHIa6%;0i+Qb%q+E%lLC&z)wuVv_JUp!snvthL&udRZ(S@qWRrVAnaKmE3d@$(Y?
zW6padF23E!?>vyQx5IYp#}#_?3fIo3_k1h!1E+yboz&-rm1i&PbCZhuM+)5Ljr?UN
zJZHxH6+`w}b*j0~oW8~Ilj?a;Xi*g=E$KG#GNMRTdT<Tps~T15qAL6i<-}s!M4dD*
zyGp6(6BgEu&mW&)m_Jks*nR{rVl+FL;Xhg>FVBzc{b0YXk;`;<P|D?x)n^ObToZKC
z1m$K2UOFyJ_da!Cw%d-qQ#8%pVxO%jv@q=Neqz?%?CqmNb7Rr+Z!#NBI7~S*^Cl_&
zh0^zkh={mN=5&~Z=3vuohq8K$3vaD9{#YHEa&^4Vku^Wb7^x8PuMjDa8O%h^y_d<f
z%$*ylpm;0vlO9@6uh3$hQzTl#yV{|{)nT_m?*8F9TB4I}b*Z#O;fa0W*DHQQ)a@XE
zrdfM;+wsoluQ<IaVnU_tjNUwy5Vh-hJvCkBZeX;tE#Gxk!NPs&y~jK3=6X>h1()$U
zE%mc!*BiQyYx=Ag-OMRB_2&ao<D<pHO){0q`}-S`MQSoH-~4vNDkmqW5I_9Zg@J-s
zoYHGP7EIGygbpPqa7qt6pDJL@7Mu!8#)=l<WCf(pd&tk7OJP<8i|q(^ooGsB>~&4#
zK~~=pcKKBE#0iHoN<MoHN+{cS@L%0&HH$-TJU2CL){>kz(VS}4l({g|&-n1v-)~2_
z<K}$)Vdw7MK2>tFySJZs@LhwRol2|A=$l(TODJOclD4yRp{3>|uij!#j+3&xP>o}r
zB;li`{=B}z%PNPpFe~QQ2H?0l(G>^gSzg8Ok{vFqgAA^k6%SLET)dq7-c@~vuXh=W
znO2@?Nk{M;8SQ!FZN-CR*=#F6_4|s>+lM)Kr4JUaR`a>JoM&xII_lPX*{heEoz(Wf
z|25TW<x1M7;t5Yr^`}Q8t!5V{3j}fo({*pzJqww{j_#~EOS|`{ySWi%4iOvO<DU~<
zI00deL*DB|DjnCo_m^MT&)72cP~7ZPPg(~a`N3Wj<zl(yf`zLmBIIPPoWl;-KSAD$
zG>{Hd4!n9o#o_ItB}Z5b^TxF?>Z102wNr?zr;i?OM5);lsLUlbGs%pwO}cz;?6^qY
zz2VC5Ke`>fGo*fd<w{DPsLGL6Rbx}{ZY*0@oN886oo+J*^nKn^q+*Ne=~#_#*MI-0
zmO;;bjTQa5<h+RSW7ptES5E=mpZv;KKwhpsNk1fAHc<J<p+kq7CoVgfloneg=iS#V
zKuDNaqM)`N&rARQfOMKggI~n_fT^hS#DHUOm_z48i;cfsm8>L%UZYub*b@|7^$hE_
zQ&lYq+AmYB4&VG%spZN<lPM)tuiOR{rTxQ&t84tq#uf#6dJ<f)(yh3^C*$j8Z)pSf
z`EehMRG<Q5XWMp1hA$oc*b+8YW2vdU`E5(rpG|Q?|Mv8*I1k$<wA0Kro6}Li_9oAx
zBAC=AKRYSh*0%jvlr5uJ)Vej(5&x`$cP2U@A|fK)s;l%)%K5UpMlIa?EZ6sjJ6qqG
zRF#|VYP=9)<uxd8)1JFt(ps<Bb?fwF<@9tM;#2;?2kgzI2L471X})&D{$P#2Le-TW
zs=uriG0-!LIN`8T$*9PSOM6^5>yk<F#2e=NJIk<SmD8cRAtn6IyR+^4&%EzXI-9bY
zp)b47{W1e}AuyfU!CII+n_;crJQ1Z5a^LN`$y-#mQ~uMP-Xi7acF=d8d|CQiO~7?6
z*5?EA*SFRuXrDHU0D`$!(cIJ?_W02C#T#0J_UNAJOY27!Ov%H`M~1&VAT|10jL~T9
zpO`xC@$H7Y6WJCW`L3v|n>H^@zVc-F%gYVQY&q&vnyH#CeWy!Do8L<Nem^&T@6Nc-
zF>|-}*MIruN1pl)Z4x`?Uu4yk^;FCD*gQp&h_1Cevtym!Plt*pwjIA;=quU2ehJ^@
zlj7E``_1d(xY!v8OS$LW%uCZMKlg>v2(uRF@jkAVE9H@jpSsP@6io!<sL&x{9I&WQ
zD4VgC;7ZJO`;NWp8b0N(Y5%vye)D<%XDE6PuJjOab38Mjefi_v6xYg7DcxMH>Nd59
zk~ZNf^AC*ub~zc>ooz_cuSrkTez~1fe(uv$8`Etavb+J|VDtoc<WwGtE9H9petk)_
zmGoq*-P3cgZrY+CMh%{>)_)->I5g9A5kTk24-+r=gNKaRU4e#LcQDoIfVf@%5{w5|
zqL=me<CDHHPy<lbXj3p>ot~_lAsnR|mR&Rp@M2Z$1USb^wNeS$=SvWaF?Bb=ut3h(
zIq&NDsyBw==>U|CS6ZxU($&y37fPQLxKz67*5oiEp;%GCRl`yj$LI+avR2y?u9kcx
zal$=k`h9SSt#n(`t&B_f9A2BVG0runU=_R0D5#WwMiT+lA*X+D+4%e-VkgB41;(4I
z0<S#s&e^O>-D{auA(A4DOoP0uN8H>N)-oGnPw${?xkoqs_5S>=K+IK*`3(ln+a4Ck
zL^^-lc@)1`*67xF&ZB#)kDurIz2}cL(}~$*6~h{<Q7b`7d!fUa?u}4uIRlXPp`*b4
z%&27&*7zMj%+wnjx6@I>6luTJC+F_{^tF`ZDv4d6af2G+hQ>b@W|;z5$t!w}?{B1_
zcqz1H+ZzLjdM*>r+}M7#KhekG>sy`#9lc#yVGiwO`-iSA;`DT%?>Bsh-YLegz)jqu
zez)p@kIx(j(;7NcYgC0sdz&r}cNT4EVP^Ly?eDsM_z8~At%mZnBJXu2IHc9-4nwU{
zSK90wj9Mmea1xj&!+h^-QH;uKIc_U|qVeL?r}x^)>|J5!T-IE#yuT1$pnn>vE+TU<
zy~PN@XxpIgP=pK_G0v_WkFM@Y{`dVj-NQQiHOE2UwArDY63>-elx3~a+L>LiS|LC8
z{n|Y`Yg()_^qYtEbFTOz5eHsvv%EKU_SN+ja{ITdh3i?iU|j{!26Y_I>p86^tJx>I
zlGW1haGiyeP<U_GY3dn^COQyLRGV!XzLIPV^Zt(;gcA$VLGt4X^_z3@?-h-7R~Q$1
zt{ko~T-e($YUnm?Qd}ZxEgYy^v|O_FJCvF|>9T-CLjXtn`x9Pv0=;_?URY=7vNB#P
z+3&)ehei0Jb$An-%YC)G{A5N-sfDb!VDnOShEC_H->HuI+yH(qJ_l9xZHu@FiWs%j
zC+e;vJQ1IqpZ$x&xV%mN^o0a8;)RO5XEf2ctMVLGoMWa0WaX|twjE{V7U0lP0|SFI
z^NN1F8+i6v?yTbmz_9$5UuOTz#9&CeTC{4|lhe_vkq_DJ=c9zZCiDc)%ujtT+ivEX
z)QiXHAinR5o~-ObYvi{h!ey^<1fqc|VQ>#?p6PBFWy^#1d?z6x1v`zXB?2zr%PmYX
z_WJJGxziJ;Qry%zSi<6r3~O#wds3u=&vf(Vr0l*pG#m@+Q7IJM93eu?zjnUjpSYgG
z%W?FLB;(#`)mz;aK|$uNG3rk{tB;H9N$Ege%ZHV<#R3TTXGvN%em6fe?kzo(`BcMM
zP!%v*dA=#m>yJPF*pJrO7#kYNt#$I`cO?SuF0UU-2u06s0PHz4Z{R$b_CU(+LU4WK
z^K*wo@dCDhjMSC{_pqO<F(y~+^P$`GFI`aMALw<~d+=U5b|QFhWMMw)v$x^GNRaLC
zU#%+yc7OShh1l^UGB1CoFMic*CBD;7cA|0p{-1t4V;6M3=XJ2{KuSq*Z-jezz_eiV
z>tL~`hdnq&Z95ASW(x2twm7b)<7vAS*BLl&ikH5BZ(O*E2P|jiWV<tijxY5q<$*#K
z;#CBS1z1yxp+Nud3fx%=&g>MNZqOaZPQfu|f7OGS(@o$9ADPbpcGAAvqU4uI^@gt2
z&%S(%&!JTmkZjdxZ<W8A{JeO$+ssGDt~-iGIB${REF2_%M7Fdr$QQyV>}S9!7N)4A
z!;p&y6TR^I0Re*q{hWQ#5re33p7c9zBLNYG-|Ptej<P=(h$EDuRm0*~T3sUfQ=9$v
zcY5vzT<*jR+(Q1^eFdqTveIt7%<l;qJ7P;1=uExK0(Jzo$e|{EyJ7EEB2H3W&m`#u
zzf=@)L#O@@9T}xhUc1+#)bq2`mVJNyJK(H~t(ms;RA&;-uKen&uS88NL#L2^8ORi#
z2qjs$xlgYs0#qm1YiHS3F1^VA_Vpd=tL=W;uphZsL?lo-QqXxdhXk9Mdg?D12ToR^
zRB1=p8k~KVmUf(R#}AJ^1b<Sre8VkazJ?YBwLP<O0(r6c#yW$uzvq0pGc4VzrPybK
zn4mMZ;&~lV$my)L#b2C1yjV4<?DwDhE>u5LqO1)5_n`gQ*~Id8fV=18I_Z|$yK%&<
zobM{`V?W|SlYr}sIBxHu#^_+MJ+KPpVWSpA*6QTk5#=K5uh{{7=@=&`$H8<q6>YNb
z$KS4{TCI@zbnw}!*rlzXsq3!$yu2LOED5J^D4U>_zcn$~Qc`+n3z30E6WS?8M`S?F
zaOH}zKk%3;eV}%2>;V0`X2+rxGT#O1U(v3gU$UHg4KTiN2_GAf9vmtKGkB~#3{~n{
z=*$$upD|g$a%B_O_|5Vg^tGLA-tbZs;C!Xb=xdZ%6BC=SK0i78iw;ncyB11a8?RH>
z9Z{?1)be&xA~-m@Lg^Dud}a(|&Qfmc>h9K@*cY=HQN~3KTSkYQG{5`Yup{Ebc(6Gw
zFmDh2=A*>{KL;GY^b=nAFtPh5pAf6TY|j&-ZNbk;;jTY9IJRoWgR=kU&vS{>|Ig|E
z{~`zbk8k<!P#FI6mH*%1)BoS?B|DBw6D%5%Y>{`*u;rn0-Io_sTG1m1=Haw$ybd5h
zoiIC0Pwt^gcGQ<!v#EvSPge9!o=0BNBc)pyJs~zpNKPCF11ASFIs|d9uL%mv%NzFK
zBc%4C-9XOJJciWrkLM(GdE$SN%=Moj{J#ytYl+RE9yg1K*wj#$UEaQm>qDKE;hvH4
z4#!rI1%Y`I*Am;0tkX{|nrs00Dh7M0-J}bgc+Fg8NdIM7-WP)5GQ8LHVyUA4*uT$p
zX3P-;TxVpEfftq3wY$%LN0<Z%xe)ML00f`fWoEl<HaJW0gmD9dU)GfZr!ayF=FrlB
zjyND(FgK#yxKg5wa4x{`aJf)&A+Wm<`XF^6C-f#>;E)Q@yYPUCwi|d^CHF4HxP*<b
zCfgiDYIF!XXc;(d1N7z<a2->B9V#WZbm>wf$Smr~26-1z+6FQienifFXil>Tz1kP6
z4y4BG-q~u$86d>81EC>V42)-l4ko0XD1x#61`&|4`uqalAZu5b<^Ej0b#ig3<@885
zkrwNrJ)8*Hx2LBnLZHKCRD~E=oiuYU;tLoc(fwAd$%uj{zcZdY@)O6UZU3htuH52Q
zqas=KMzTZCmZ!Wr#3p|OvDS>V27`(?cNC1Lrd8p#0HKQg^@$Zu4z*jaFJA6t$eL*g
zF<7@|jRLeLAzMX_d&KeKY^D{>E?UCrLuvq{5sbV7$o%SbNk|WJ^J6unH{F_T#A#-b
zf}3KMRMNK%{lnBSgXm&U(+Jmn;IxiQEGWv)sJDWjClnaeIEkh97Bwy|+~O~@id3oC
zdb8$KGd=q{O<QzeN6;E8OQh#Yd6a2(N%?qYLLM^)fuoztLMA9iAEFbNCiDkNp>7m2
z2SmB=2C)nGUio1`FMI_4I1w%*WgLM#QdAvzT<D@=Q-IUyz|uSm;Ven6GKs_5Tx-`B
z144BMsL9;|cb8c@jOAJ*9Yw0v((vOJuWl|M1_@b=Ht6;hU#2<uVm<Iwwr?aF6x=-9
z9UK1c1#ojaV_|9KG<B3w?CU+$mJ?+D`k7L|6T9BZ?$j;UxCc|9S1;m77X~fsAAydM
za8SalkG_nXoP68>>RK{rD!MaYRr7$_y;JfXn)CP<85-|H7p3pJapQ(<e`1F9@B#u<
z*YZbw9mK!T6*u$rk2TRH0FOLsa?|Hymz~?P*}!GE=tFCkoeVJFwh7sxOvV#;)<UH4
zc0bRvFU~8qUanE0z<wa_?iO1Tw4?X?EH4?4l`FOM?KM-F9f;9L1A@_lcb&A<D!<U(
zk|n9>YjN{3S}Gpu?I^5iTZ%rte`So2*XBAN1j<XOlQwm1s7<<G0-$xnqQ&2O9_EzV
zilchFZoSifjw~e3YYB;ZS$j#$nyQId(CC^6Z(f*~(avp?H+}{hraTVgYD1UdBXA6C
z0_U<CdS*wh13uoiUfZ-tVBwR8SG(Lv-*2JW$)VfY(_Zl5iT4w<p;Fq$Xf}zJ41N6-
z4>Utzge%wT9jd!0)4-{%MVEFNB5-bNL)@{-bC)8eJeKe~{&;z?+1`5+Us!aCO9<6M
z<H%l{M}l{pPjh8mh_MvXO7g)rmi+q7>T+>6M5Jx4x*|!S#|MN~EZMjYnDcfTh?1hl
z6l2M}By=1A_IxC;6<_;e(G&DLBzY+!e2u`A`aShKAV>-w3v7NT(o4IzHA2g_J5cLP
zR*naUt_L;k;6qPawrmU1yZh1}XT|upYhOr(tk~BK?0(Ci`%NH1Tcbicc)v$kU=MH9
zZ2N(WZy)JdQ}Xx&$s_KEAr;N;hmT(@U;q8aVHF+pD=YJyCoAKUkxie`^9H_u<x%qC
z-P6y{ixzzg`-mYSiTPRK`~tdfPs;4{+RU8ns}myS<WiH!NmpM!>GZ}DUs70XA(8i7
zPE_P~+S(3UIJ8n}{B4qc&ie0vJUBNR!Sbn+XUR|i5GQ~DFUFTKK)jdjcPqcQD{@Bc
zIob(Z&>BVHCyajo<*O7B2M=VOuTa~rlVKfZJ1{@lA%L}ieKpamlXtt8dhaK+GX6Ng
zR>KQe>b7kV4I0hqbrtZsJ$T*R@p@fR$4`Hz4zAO`tWE?X0c4^X7v^Ub!4agmM!1ZW
zInI2HKCX9~pYz};pzfXTUYJyL#>FQIj<f?u@JC0uU5)M800N7I;1B~5?H60!#^VjU
z&-)*IGyL}O3a;6O4W^-YM$^!t>EXOeMFb_DM!6wI$s6Ka@z}>2M!l|$F9i*0`Yo0(
zRl1<rLP+H94YgA}@lbphHH+s%nMgLfY~Mh>mu-g*LJ9%p6grobMQh`=M6Q5z{{T%p
z7l$qd6m?XXF=Xmt&;x-pqxYBK*p~12AQ^JC>i%=ylSC1`@>n4<YKrD0W;fzE4sjfq
zb)iuY;xtj}ZgUI)MH1l92vQ>sL(L&^%f|0b`w4)hfPXb=xjHeh0iWVexetVcF$$l5
zi%kI~5x>+?aFbTsl6E2XcL8GE`|$e`0Uu)KYWlBp9@v=#D&SOAKv2*+M85Cb+?)y%
z2mLhp6nf4_XetQ?dQr1qd7wYxB6vzmNGsI3?*H@JqRxhb1$%5;-iwU0A5l`-y5QJ{
zcUvX^7pSh*ChAI(9t^3^Q19})Bg3#76!z7{SL7C>Sg)r3Yykc8JLKDoP*DcZmXoep
z2o2**lvDv!a5k|(8^ty#pnru_?;m!hdDj7lAzmWEV8)O(X&*h>?X(;`1hx6t&K;Zw
zK0E@q$+uZvUY=;pMt9ac!%Ha16Ap+S3*UC11NB0c8cR3!L0#A)A790F#zzY4oB1&+
z<cV&EHErI}mlxhVfw3Y5OL&BYPb%IyL)k*eO=pDrMb19oID`lIfO3^;UMEN*GJ%j#
z1T8s1NE-<54VQ7cG&kf>;0+gu@I<pk^2PRCCkhpG&l^z7z3~PjM}5|-cU6VNMS(K}
zkD*qGP0};!03Im;XCIiNMgmVdIO^j5<N|)wJ4fEDA0BV$HA4!RLuRUM?1dgs*=V31
z`l?HCgx9ShHac8qV3yPy4Z*-VUZhFQ8?ST5&bM!i_k1n=!%&!+b(X_KS9_Tqv+uT3
z(9t6XVk?tE;@sNi-mh@?M8Ya0mx@E1y_OylEs-*cs9kpH#n3p61{za}((97+rLBG*
zPsV9WopnT+>jZDH)7A=A<JL2d7!A~Pl_LtxjHnRnxPd)Uh{BA`&fp+4o=jJQ-`PB>
z@6;V=HcJYBdo6Q`{M4a(3mMP_N9svEO#e^=(aEPbKa)X9{T->roV;hB<)I^&fY4Km
zbRBLRe!#x(8Oe>cAb*wTc5BY_-YX);O9NW^3!P-A(VkdmW}c+x*|TQ{^om@UgJ`*P
z0Aj)M(Am4kh0FRZZYxKg0&FTcg>^jM-Q8`H#)Jg9XB=4K-hP`SdWSp4A?rm`JSE5-
zh9YqqI_2)Vx$Z##7{M#a(u1j|V|%ioa~`*Ea{%3u28!IMw;7UJwV%+=ou%gTWQOSb
zZPAL|b>Vo*`&Z}38$iV2WFD8UA|K4hP;todjE7XE&A~ovCuY3^+73DZcH`xO9wY~q
zwo4UzV&R{=XC4-J5!ELaXfkp>0hroIwZR<Kw>b^4Ip}4Y;9a{1dP$Y<u}9Sl8>W!e
z0!O1u+rh;tkLO)xyaUZs-hKcTb6L}Q4qt8wutnv$1$L#m@diC%%4K=o?JVi8_F&C3
zL;7KRnQ>R%JUf*XlK4J*mr}O8H(EgbI47W<hr`ZJ+ll6EPQeSf-m<XS1Nejr;u7mt
zrJ!JrlOLF8b*-;YVCdfRgGaGqJ}4>HSK9#G_YbDlta<~FP}SySs^AKvn%h8k`=ZRo
z73QcDaXLigK43zP+@;uTn37KFUHurT&04`EmS0m#pDQ_k#{A5D13(86r#}3$*@gL7
z6>o_8xrJ^dAJb7Q_7{j4XSunJl<i=UIzK(yYg*_2bR_rY0%(6lbQhg%4jpgeSYWQX
ziEGSgzba&0K=|0epgoFg2ik88*V1dHFdYV)B}^@9<I=hf??4b5bc7)RXW=OIAK|jj
z#l*>zIu%>&2Stat!%TpP)vV+>IoBD&+eUD0Y?xI+jidh~A-Y?3YuBzl3Pq1W(l$s$
z)4U*3;B4GRL-wP>8zW6<`^S3`bw)IKdLq*MJFw@WiGj%0WUy+<f3vTp%QDfci+6ia
zyj>%<y%A!#x7G2<ybYhhV36PIob%mbPKu>@Vt1TP6^FtM(z6^?DsqKJ9^U$?&C=Q;
z&-#)XYP?yC^3}WNqb5)h?`Nh1cx)x$@ebw*@7y6kaF|!*--VKs2wB~srk^rIXOZ{n
zH`<zGgDsMBpl@R|<Im7u4Hxs+6av`uP_u{6JBs&zf`<>8@dr+^hQmv_)wXDcn9iF#
z){SEakfoDyN6M!_#PKa#C+q|5wHQ^Nnmi(JGynW$nX>-h9Dun<2zd(d_AbMQFt6kj
z<O2c?0&Y|1^jyd5j^r<1Avg8o9LzP9R~IV7q)X7!cA{_cMlL9Vlg0>65Mt{xaoiZr
z9kcCG`GoAAz$_Bk?Czw0n6wO`q)emJxAhSs?5+hYt*HCL99bdAO$P^^mkg&ZN(j6{
z{%2D!s>jMHIw6;7dAYIxINmgIn}*q+xuaNVEe>Y$Y|a&{%K9cE@|}8uMIqwtt)jA>
z`TizPCsu0rC~rpVrQ<{%_b(8NCXljW7E~x50tR$6k<olo)u}rTbXPDuK%EE>QnZyk
z3iVUGn3Vn~%7SW3I>^SZ%ZXlpA{_ed9t=4T0t74Pfh@dMzIjxbFWBJ%*BuIl8iaz{
zP?)w7Lg96;P#=ZD`4?&Wl38PD%%iPZP=WXQwIO#_h%+6^UVX9iWqHnVNsd?z!8+s+
zAAjF0lu1E*cQP<#4Me-nP8{iReSc$}*PvFozi2B7ZAWJA_oUc|$p0d;`P6t{-EmZK
zJF^mM5z-K~Yd*5jOzp5@#kp}R6Z~pRJql2EN~B=!(k)}?@}b!y<wB?FsHY4fbE8-W
z55H3nac>u^+`omv-}5FUs1NIJj(%g?kOOu#7Wi#luzXE#f~Um?6hq;W2Wfm)8aGKm
z>hy_VP`(~Ho8&E=z0<1Wjew0qlSvRvLJ8CYr*;VKIPu_QT-ILMyG&GQ6@4_ZMd{0d
zD!sGHXO87#wS#8Vje`A<aAXE`g856d?vwN#i<A_t&%Cc|UDl*S{;}Bn^guH-KCxsc
zkuWOs)F|=et8t5h9uZBgPiWG7Uwm?uelm<+v0p7)+jTi_gwS$$?<d#~g{~yQIUVh4
zX?8>x86r5p0SGYtMFuAvWq(k|+}TmUg8~}se9y^r`>1f(NNMirU0DwM8t$~_bji9;
z-gXXhIbWU^Z_%UAl@zQ|ZvPT{y3GvGF(u+LPGIE2Xgsx9RS9IrK-&T2<GtZkmLm&^
zn^tsufS4jOE+GLkV|n|&gZASq9Lg7Oe2Rx$<CF&%MA!Lt<jcI$epps|&bP<?3G1H^
zQ0ykFS#?D|E?~}%s0K8=M+~$~K|r*O^$Y7~7v53i*Y12kDkH!h5Bbr~Sbz>MLVrQh
zlG!ZbLP8fdS7omZH2SdO3y?*IYfq~kf!9`9uK=c=VMqvMFH8NH2VzJ>d(ic%p5bAn
z>;9;cggRi%1)SjfQ)iQp3ukR5Jgbtd=*!JTQb)SLW4&BdXS0?|u1i{YtTQ)3O?pu@
zODB%p#lv%S;)vZcYXRuvqvS#MtiQK6tw1eg$V)Xwfn5i-MQAi!IB%?)5Q-6^np)S)
zffp6k6Z&wuHkW*ILwQF#>F|uN1??4EszM$ow-*QnUjpTWe6it31dbmggOn%F36`Si
z^pg1_6}TY<$F~st+P#Tk2l8O#LUAL-x64uh2hTcxx+ir|e!5KU2|RgHx*|cJp&rg-
zdgCO#wA?7F!sDdO2wrC+y^op-J$#!ch5cG{#=6=~D)KB&_2gxZ<X_mVCs{X~U3LRP
z<XL)b@~>RrH9R#-K%w<IlK%_1G!7mOkvB@obaxCY>S2911J>r!UK^9y_R0KpVD@&U
z7{X)e{i_*4Ou{C;^>M#47mEXJTV*AXjCd6JQ+dDvSkH219I3Yx@>F>fKY4ew^lXcw
z8a{eyc?a}cd_eB<Lt}crrXN-*aZZ^}za>mu!aCYq6>me%W1zTJn(H|Q!&sfdb~aGM
zS(8Sc1-hM<&UwUWA;K@rae>g@Aaj}+Mj~-Qld-)A)XY;;gm03g#Yd%@D)<A>FU^mp
zaqnO+^?#&YqO`9g2Oz3L2}ww7=bY$Ae&&@b*~voqjlHZ|P+^Tny2^;#ID-1mZV>vD
za*l!lX9Syu--4aVsDzrXNY+Ws*EqJ^1OchpeTQ`pt#n0r6auU0S?Ypv5?frLyBclb
zv))fR@{VnK+01u@)qP**Ts{vZ#@@m9(AFHw&pUXjN58${#IOBP2Mi>$KA{1=VA(O{
znwv1V?Tx^u6-yosu&XwTr&n&JW=!U^48YZ!vfRKy#JDfu^1a4n)XQP;IG1bs5l2OE
zNb82{#;+oIQ>DNSHI+TcQQ5E`EW*pa1^|xD$#(Ppz0CGI=V2TlMw&2{5N>FA3xd$Q
zb@`ujqg7Sr0xMUNi5*$Z>Z&ZdD5Kf=&UFQz8iHvb$(;Yf;Hh}Ez6jhXCwOwV5qYvU
z>w;u9RyiZ{=_xaIX=t&|9WLH8l#u$4te52OYEEHe%E@YkI7z4}R>5xaDDqlG`7AIo
zQG8;Tx=K^s{pLO>%6n1vZ<&C*Tsr<*f?x=mE^-z3%6AbsC!Dd9?G3%TLVzC~oP}yb
z!fr51N>%5g1panrLO&9V{$ks|)=0@6lfu$j4j=8955t{;dL>rB&u-GQ&$7O&1L?7N
z;_kwUSY(Xi3Cp1|VJv(Cd*ORn8+U&cIM~!YCvd=*N3xPu9^kVg+Xe^JD;Eumc~Zvd
zZeECgA#U^7zT$U^0(@$+;~GyCJ*|?Sm8`Z-1#n#F6xRQp@#u7KWmrgECX!vp@_P_v
zS5sD~A5~oAI?#t6Tv=jY!#>u}O-F<I)q@m30e7O+61*aS%xDqZONEBmQKgpfT?)Hn
z=L5A2c8$A{QXYdQGsWuTAC;b+?p3d09D=EfwWh@#A>P{Q@k2<<+{5;ZI9X=y`(H>n
zk=KEk9CqC`o_sOh(y_3Q?bL7ur)rB<XPZH^pe-kqOHMJ@f;V*xWu&0;kpDJ-2fErL
zEAmX*yB*#rWLj6payF=(+UB}K5AdjX0-4BXuxTz@2F;_1=7_uESC5WbAlmLU9KXN2
ztSmRAss{56p|5qDK<EY7cQECC5$SaIZ;2lwT9Pz%BVgE-D`|wO5Y5+GvPjWL#7afp
zm2iIiM0WH-*6*RJz@#txGN$F9)yV!}X(x8!wI_K;qD};_VhK1;@?S8~6ycnce*Kff
z;;$8socekVSB-h+#;pb?zx(c6@xK-AsFt~~BKXI3vL)rmZ{9DBH8!~rW5Tbt+06XO
z*E>=&?N_O*ixwH*+Pu?2L3k09>F>5`(M!{baEGda;P<0W;o~v`lP~HWlC^((?U3)B
zNB9`2vk+;)eD1w`OP51_!`qI9iHtV)q?nk-Ue_H;rx+3P8XpvsE?jldhj7wo`C7zn
z*qi*OjyOevd;rre#ODsD_`B}zm5ZMpb3vPCq4%n8TR-u6(hftfABZO?gF)IGF0^OQ
z8^@%$U+bYL1P{-^y<+OpWw4L@n>rmWZ(lzU-2#xnRWLtjV`0w%dS*NE!QPnWUZ9Lu
z#HCfcuW)e{Z<JPuRKn05|E2|bh~6hJUTV*}Gn!=J+LgPdD{pc;Gt4j7VZzBqis>HX
zEwTDsTkTO}9Z9oPbdb%X!3%@0CtOtK6-zx9opioZzW9-jGwXRsvSQm{p7Rgs-xS>x
z4iWfMt`><$Ik_QGH;DNYcW7s4C$(J0fByD!pdZ_thp`6Yk4TDX&ae@n{?Bl}PfNn>
zu9HpO{58F>xVtINI{7Y6gu2OuF*T7<z~?VBa*SP8@t)kPYgPt3<TUDH(c8c24wH5`
zV-C#G+&1B|##nPC&Elk$$xL%k_R&k~dV0J8JJi~Hr@+8;;#?(ZMo_m`Y}YiW+lUUw
zr<c|ZulHtGILiJ7nIMZ!7Il{&XY@2AZEjq8wOn#wLHZ^BMPsg$9X-J?9V0pP9$K^B
z>Eun~2cZ*M(wc2=fsXcov8b)Ox;hQ|Ft)7a;kJ8V;97akUfP00vFJ=3u%$;Tr6n)5
z*TwsXI{XToOBX=>WThN4ub?|uEcRbl-Lh#f_=3P;7nKg<+Of2%9@e($pE%5qLs_j8
z6cI@tJ0RfC)0*$<OcP8whU>Vv2E|m311olfSSa7zr%8-Ej6gg1;d3DOdf+!D$m50h
zD-z_W1)3ta)wN@+k6FFp-h>Egwm@ex1#g>>R58h2zTGywYPj!Il;W*A`|3HQZa-o_
zBNzqKZ)}WU=(9g?#WiMV*dl9Sqk6-O`LEO-ui@f40=AKKE>ZK^n5k|N_YDy8k|)%Q
z3Z{xgE^!+pKM{9K2Euq5!&9eLIkY>O5>*@R#<LX5rC;^_{vO@E;T1i${>bgie-3r9
z?#i{OUvJmGa-==WM=Pi+um7lVsk7u*=w$oDnTNNYTXR3miwJ!ljZcbhcGkk6;;^0v
zCm%^c5`H2AoCmGoMtP(|m>{x&o@UrqULsi1#voE{N&xnrS(&|i6A?x|X$6+uBQX_u
z0{smD6}iQyVx&szA-piiss;v=;5V0KyQA^90&*@9`rniP2S#3E-PUZa^z!l|9y3xc
z9Ah2QR26BYEEm*a3P56Gx|MDupX35at9tWXm*mH?ENiqm=K_shr*X~HOhWU!7+BH)
zAe0cmN*l-4xJMW(-7N%h5@=^wH`sdn&CoY>u6+2Gr85^>vs=H2dt1O8>x-F=8am+T
zWeR?J|3|_s9GV$M&@qN#hf66RMaz7vV}8hC_=bqfriSl%#IeM3z0b219HQKxA{@E7
zrI@XyPNvBz>>SXY$2!!WbKz2ZV7J{Hg|m`r@76b{SZPjvk@0Xr{wgw;c^xyn>(v)A
z<P_5Arx7%GtSl${;fykA)!(a&4kCW+DhkYo<hD`ulUV~UxBD3}Bin+0?P{0*mfz*m
z@0tb;jXfK&OKD!Ha#Cd7Be=~TgOalADFhVDntixc$q3U#b#mL|l4RhyBk$*s+kBCT
zn`pT6P*!GNT)Yu-3PtfwA)%Ju6q7Xg0+A3ldGNlNt-`=q%*5%o<@(r(co_qnNXCxN
z8dy>Th>oA-bqBd*3}#1aT(gbA#5SrMKS{~v+rcT-b^Dj0`jc(W7{53fXXrQg$wNM&
z#Nn<y={k<qs7~7RMkde<$QnoEeSy8w_I+zuqw=iN-P~47U!S`={os|S`y5Z85{~?(
z<;y_lZ0yRl+tM4Bb2EI@@Ajniv93JU<HtRD-MK;)y%fF^LH4J0(Tcxu9Ky@TPPF%h
z|5_cb7LiaK1b{+jtCLUBP9#+K?>Ze-dsWek=*Q}z6;vy~S-t&yZhk_7<zxhiwSe*c
z1icbFk6RE6Wtj8h@4qM~i_LR1V#yPcjL9)51ImX6Y=-3W;S>+;ku(g4$k^s?jk?_m
z#!`Emt+_O7+Tbum#m4zwfQ}+;ly-b3_b<fHF|q(Z>!zn9+b*9Lw@7DZRAmjF0nro$
zer(I$@B$1x#pSUI7WxbJhD&$fJAKJ(TL-PHkAFZ2javdwOO0pXY|@+!Kb<DIKZHC%
zNJufVs;jF78*0PJa1b&5+O!j}&^fS|W~0Ck#_yAAY7BlgR$cktWAD7K^KV})QNTcw
zIT!TPgPRb;YL(J?H}C&9Ydjg4OqJ|>{H}zM=Jja8F{7DVZih5bHFfA3daxpzZ1kD;
z1H~0T-2GQ**M0!c9D8r8E1c<UsWovX8$)$MgLB^lyM6!<THHVBGFFAog4#(5KS)V_
z6vi+7%$Q;P)iPfWgBPgiTiB{ZYRmXgLu6#+aIBhZT)72L3vX8q+C`0z<2*LG>P*pr
ztkp%-Za`G1=*ypF78IOebUQB(!)%kfhDPe_xWD@xL#L?jyRm^LF~$}(u|c;GHFz{j
z7{4qNE<JAOJmmE<gIXkpP6RSeno{x2=M_xkt#}y};|WSimd&ZL#c~)Lxt@uZr*n10
zl!D90<Az@7*-diOpFG)&v=V0tw!kv6qojQZJwGvM3=#L~<Mop==qL!e`1Haj{mGo^
zj*3puvMO^~?GCx&TcmB1n@+n#gUjwB=gaXwmWgOM&7F>z(Wt-C_E!?0=*k_9-^5Nb
zJgc0THzOvk<D{6)<?Is>S97mk#fld#$A(^8;Gkh9Yk{d5o&&OKuUymP+*}zy`Gu*!
z&cmF@7YBXI%Y*K&4*$Cspqx--pdZmcBpr6wR&@s%8klALg7RLpC!_V$pe#g!Km#?1
zUd6W7dE@n=a_97PbaX!23rpJ#43bhY*R?}Fb~qbC(97y*ixOX+btsn!BhAJ8mQ1Ib
zo|4%71eE?b|6z~`4lE~Zy0)z92)cm8@k(ACmbM%Re{ddyZnB^jrs|U2G~*k&e>ow|
z1EXb(K`{$TQphmVT$toupLuiv+c9V)fOP4Nvb6&g%1WEu;iIkK3#mahX|?J!7MML7
z8Fpo<3Gwm^?E_{*ty#9OMN0}Ha_|CP-@@}00qijVSbTB*ilI1hF-rXXiNB%v5-GS`
zS`Ucw9SvHMye2o1;$H+NmVwyx#xU0|G5~OKait)8v%S$|i*=uGaQ6k<R0QTkY(fg*
z?$b-uKAoLn#bi!&XP9RW85aF*`5mbf$@#rM*?{{^7@2}{^>EY~HD%{kDds?#Q?J(H
zQjnI;lHw!n%cYrZ8g}hwT^3YTW^1xzedCj|1&zNL!^88@L(}7+UtG(YbZ4!D@5e+B
z8rFu{3XM{IE`mLKzIgP=4{Y_;&3n_~h>3c;X=oofX5|#fN;1>GuW-8!aqDUGEUDz@
z!TiqVX3$A07L~MaeE>S{%A_oHVI!bJF66p_eN+N0pV-h3D{ZM`6%5#CKQ;pq)~|2y
zAjZbE*c@4Y1U6IC++MtyGWwU7vojzlwi4A@+|M(^+eO8s3}>X`Jc2^PfZQ-$I2Xxl
z8qmm3WLYzle}{DR!pj7uCU6J}Y)f43&<^=&rO5vL@kATV7AE50?u!>MTzgn3Yq29U
zGiMKzI#i9kqTIk5FqMy=aCNze6H$f%#k<tXK|FSd%k2U2k+_Bxls^`kFq7m<{ua#z
z-6jz5p)*a=J+dw68np-Wz}Fl}V_BQ#@F%0pCd`&ak&?kUYZ_{Mw0M@>&cpebI+b9!
ze*PQM<Z|;fXv&gtR{F|7siPrmtYYm!@$kZLNoK|-B<!9BM;~sR3q;*xpNqoK#QX47
zXzx#Ia3H8s78x*w&A`do-z?FvApeXVjexE@aG8VS;wm(sx{S#VWWpPqc9K#|SHwt^
z@TGR8&CIYj2kuU`2jx!6>*kNPtBCf7ZC!5sX6r5cn#OvSyBY;o%xh>@e2iUT5qMLv
z9&G>XnR$u(E>j*CDno-uU1%i#?YTwbnv*TNV0^*R*aO8Ydv=_Ve@MsN1zuoVQKylv
zv$>A8Y>^+Cku82=bG>e@-f<M3vd(sAW}B-a)Z<Q?$RvfTSoHYu<BVm7Q-2IKkZ43O
z8ExNM65<hXTksosj-9u5z;ssSOOtu&HXZ!%;k(r{u^<LyF=$KP0Sg;zzJtYwzp9+R
zbCYBS%a-&#H0eXKr!gvT3MToe$AE2F%oZcHC1B4C%o-|fT>Ev&2_q3K6;p{XSVxD2
z1d^}T>3v7odAIh9XVpXD#JH2QK;6KAAB1AND--l5$s<nQ+jQXtyAs)^Yz^&V{ejCD
zWgu0jU~&%j4n<lEHc^W|7?<AOl!75P9rrH_gzMiDpF@oB{MFfsL0j;iM=@1K%qorh
zBJ>YyXlN87z&jCaw{Y0xx2uM(hal@sQfxR@<=EY4XPCPp<Yt?yjw&gY$(3jUFw^nL
zEBpy`km>yXsy-K{b4SHgW15~m7^2~<Dpz3<*X~yi`J-VpY4!Nk#pejBPE+|=ZI`j9
z0(&m+^t~^2LZm?|6egBH)(ZMdU-#)C_c8}ZRGRT0Dha*TS@=jXzjm^5LX0`XSLJLt
zts}Owet8}7OD;WP5lVH6wBPix^$issi*75Q^Zw(DizlQ_c9hA>nK;AEP&>0a1_%0$
zdB$y7gtD9z3|=q!8R#NaF`IAj@vOb{g{?|=^&lW=kLOJ$t#D>ee1a~tRguqLqdp}Z
zd!uyZ$dRYv*Y!yAPt%b^iKxm9Fx6+Szq?gwM1VCcAXOxdUZ@Mjs?xl@&tmyDrMs=j
zUx8OR%hF7mH$9Sb_*UueR+=Y__pw1IuRVF^!lVbjm1IW=Q;9@s>9)e>aW1i7dr?tQ
z8M)>Ln$zN&ZTp;~2ewX^RLVnAFV74(xGxccQ<+x&NZ@(Yt_>w-meOm2?%&@|%(*Y4
zsk@1H->*`wHQ@K-2l`>tDQ}srSR#!{+k@vM(!*W^C{*fJ=DhFD4Nz#;ZJ&)$X_tOE
zQ|T<37V2<UJX9iJYrP7n0;N+0SV5(`fE&A{fc?ztpvLBj$#HE8D89RbbsOe=?lt((
z+aY)F-X*1##8QIN2}~hjVL#FVXoe=?%Ce@@Xz7ip;(?Jq<4hb1)p%Pp<XyGy3WX94
zdaGfMd_A3?homPAtD1c6lvx`-cB)6RU;{KOwKeP3nbz4D2VZyL!PxQcv(Fzr%6Y6B
zx?hw?Be0&y!_9QZ0KePRdqIb^(xdu`k3Gp~QUcGJF|L=;E~!C!-p}w6BSDirLgQ+v
zpAE}NyvzcO0kyi{!tv&VYWr*`8Wq$QnYozAXHvISC#g`<*h?lzgQqbuJMDK**F)%i
z04nh{2y9iXPw3DfIVREk8X$@?fs7gW+yu>Un-<z)R+yNW&<zzdMP~#9<_*pr7CqnR
zMSbr<rcJnE?IIY|0T8b={q0Kh8SQ1{A(Uou|4=FY19r*TaHPF1V6<H;zy8Fp_NN1*
zvaQY8lQw{_C4?!bN)_eq*2&T}BtV72$V3H&hnDc&8$Ubpp5vKjPQ*iwM&>xPLmsN#
zI+5H@Kg?lB+n<^=vbK`t49lnprR*u-Jln2!d(&Ymic<5>UuL{oAun@(xArR93!<nX
z4)PNFaxXBj2n{tqQ9rcf*;fX2;ulE8AyXv{hYVvC3TY4tW<@W$kfT>|1GSFS(~3fq
z;v(3DBr8V^__l4^hTr+SiIdrn{Agr~1eZ|5dwv8txd?98oSZz&{g5<7nl6OgP2}q3
zJ8;<(%EkeLVaCNix9C5M7)Ot%CA;o{`;7}ixHIu5xGII8R-C5nIAzD)k`uxR42sjO
zjHLRL3WdWZ0s<ryAP;XPuEnu0Hu{P?U%GW0H+GOcIs~IBLo8zYgb%OOi{H0K%3mf4
zg6>j)@i@>=1UD6gyDv=dhwZCR<|ZV^5(wWsL3)>z3}&R^WMhvq!7q2<LRB~)+Wm-5
znuQ{<9Z6S!zJxr`b@5u5KHrF%az=zHCIH*bdU6lL?;%PvD+JQn_ksWt(^U<0RSk57
zHRKE+GL8l|*Q4o8z4&f({DL8-KjGmS2~6(lw<-iiT91nbjCR9r;h+l|oz`%lb?bWg
z8$1y`M*)B6>J_>(gDep2n+U*_NYIpbg4^$!UsK@9dC;hkB~E+<1J{{cpl(LM22%u&
z{_M&X3mC0T5bY%%lG=mffJtjXDW*&5B)dWOK!c3*CVK9VU#DOK5g=;{St4oK&SDmw
zhOwYHts{iTYhSBQE-vD{E)O({_S%B4I2ONC>F%?)hd2hR==uvs6j-tF0=j(#vT<U;
zwbC;9BRhFIrizCe@G{<*Sz7n~W^Nvfm@95r@M7rk6H*I6^%8C<I9wYyFYN}9^dry5
zEU!F8#Kb_;SDrD$q8=Y5#-)g|Ao&Z-Bdo!UX922$2Ni;?aFfOvmOn{XWGt*Zq<~>8
zcd7TkuBMxP>I_cn=y3B$S|>z%mIk_PoBoln&M#MTrHXvRv9{;dBhjMWujI!7NJ}v0
zX_sw^NjjS8J>d2Aj|Kbkv>$JQTR=}A)*MZ__EC5sM~>&szXm?bs!9Q_F2+Qj`n8MP
zJHY>(4u1v1ibn7?($IM<zogOgN1}Hpw?D!CJF(LBDsG^7Z?B+Hsf+A;VLGNO9VVpc
zavyMbZtB-Ocm7I^bHu_}w5wkSIh;X{e5p`7_2AXm!x>(vNonN_Vx9V>R^hX+MX9uh
zM1@;Cvpy?1y#;)n9Q*nw^=xV}@j@NFQ?&ZzLOZBXn}IqtIGV-}hijn!)H8Dm9T^Wy
zLi+=$*hDapWeq}clgn40G?MnnEMT&|K%!f3i+xFU*$Q+DBS&cT#P^h~tt|;m&An3?
zY-GSqC`2EQ)Teb77^VNgJ}Vutm0UZY6qoQuHQa-Y*D-GmM5A$_9smy)Sh$^h<cV+K
zc(~7rIj?)VQsR_DLde{0_%RKYm(p!;we?Ufc$Mz1Iq>b$^GpTCc!Qw;Ddj&uF$)C|
zq~gR0E@X%tE6eyW4af1D+mUu>R^2^2-&(%@?eYlA(sk^CyQgEl@~o`4o5Zj7d@pOq
zp0w+_uD2!aTpP#1SJeq`(Nq27NBjczW#f%x3|!yYS~gUV4*v%ag)PR*+5c$y4R+w*
zKhFx$NB`qz*T3JWb?D#5bC%Lp|KsP>zu!Xl%fDUowgzm}KmH}hvwt7Y`RYGk`M+dg
z@2!CsFm4iNiy~AC-4Ea69Q+R2UH9=}?2Y+z*pH=bI5<iJaC1Q7+f5wLmtbR-w;~l8
z15w@J{JevwTA&?Jm;L!KHZmV_;bZ^tbFKK_XSn_=K|<%_w}l}L{rEs9!};L2bQ!!f
z&~dK%N7!rII$u->3!ON~QL6*=u#68gsN&y{Cn-TM>&4r19Ji|h9(VyN+IpvmzPWah
zqZGfD&2`{!*S7yXm2(_4t|1l{#_C9t704;LZs%t93mjuw7|!ZIhOv)elpn|P=+uO8
z9bk{TetyCKp11qY5%|wRKlZ=mp#LXn|5vjP63H6eqNt0T6t=}%kmc9aPZs<&DMblG
zWEtlt#>LKS1v?Lz(`61cQIL?NkJ;G3l9zAXSD<mD35*2eMSw0({sN912M>P&RwQGB
zx`;8_?%f!9(iAaf7tZJI@i8%;P>_p3-1WI6z?>ruAG2XE1JcAIvA@vSP4YE?pE2lV
z+zSd4DkV&SuI=8vdxdDb`&~le@F~K;VxJ3a{}(+e&HTns7xf(wV8E@DJSxh5<{#C@
z@38pgTpZCz)lKi`;CTBTP1|3S4F%-@$y?8KUJ^I3RKgW<pkILfD{<98iXSkQ9_FVm
zaH9jia`N-9fXLal_}ld-P3I#nS1Y|kPw|gmIi;bo27OWN%Y_yoYcy-|FBfpZ;A8gC
zy7LCiN`{SF1%1c<{H2k?cC|z@VVokaa%Ks~wVO%*T(=j<-1{Lt7=uP~8YFy|a;dxg
z0|+OBS@v-LHA`1a%lL!C?#b)WetDq?)YVA42=;PS%^=OlJAVY`rI97(Jm-eJ3Wosj
z@bi`Cx*QX_IflF>xs#CEd~Xq0iTOc)Yh?dO%5C<)y6p=_>j)%3gIQiYG*-JSO%3a7
z=>-;MDIab-5*OD^x4cC|HT!JZLxUDT^{yta54Q$x6q#pl@Ifoczande^&Vs_kUF&^
z^rs)%XAmDpFJ{w?1do=I&<ZiQE3p6<Gzk8+w}(o51NcZZp=&PL|Grq6kk|R!RnIQV
z&9dv!HF3HU6kV*-$A0KJo==`U$sNP8__FWUE&UOk9I=SC^K^q0n%qziHY&yCg~S39
z=^m7iPY4NeLolVVcV`1Ncku6p*3O4QEk$$TEu+A`)`?^Kj{goBeuq|D;#GP!0s3E@
zqd+?-F>>c#rQ+vsxvfU?Q>m~lXbOxVT0XJw271Iz3Oa`P(&{H++})-M8d2ux=a2mu
zh^8KUWZ~+Lm8P*a|CL{G9Are%#mI3qc%-4!O}7Smd3%qp2U&riF2#poNwes_i5D{#
zM^I+6gOdB1lCtyVE$n}@ybdzfQ%I0$*;)(uC_(o7N$eC6@rRqd+l9Q03=*37{(8TU
zX&C4^>c<;A>Po}__{1d6&0lg{D<<0n$N6*MY^6HvaIs45p{?w1xGkU7G)$g$e*{1b
z=M!8H@z=k7yzi@e{x&umKX_Ox<V%ZbV5-I-im=ij@V!qVOZE5THcq;UHR09YcZF0|
z@J;m%8K0j-+k?Fl&uArH+^eV*Aa<DV=w~c49dLZX8K!IXGiT0tz(9DA@%i6f#)Yy+
z(82gN06(#h@>O=bgNse!A~&Sz>v#*hU4P=(&$DB`^c0M0r29)4I6&N1gk;da=`Spl
z4E6~TqRX%+N(o`a{w!_*LhHA$TVCNvInI7d4qughS%bLuWe1F!`YYIvbNb#^rBfd^
zQfSdw6fanq6foz4RgwrXGAiLVz=YmX_EK&PBT|XTVV_=8urzSC#g$-L9=&^D9Zjot
z2(AnxFl!sCR{Zl6`<}A@=PADGgQUnEbk|at?kJ(5WqUrHYw@M%)HYNNf^gri@=(K{
z9_`h5q=@S-zXN?SKX|AM5d%X5y=*MoR|ghkZ+JgcX%_}hM@-2@Jj5p=eMoLT)36pi
zYwhKCFq`wk%dEn;p8fc<{&1=mgTY$LcQldA{4}e{1?);cOJh<~#X)&IcoSI@QfCju
z{f{4&f4>O(KONG>|6jsGc2maDr-l4SS}XCpl&1J)uPWd$K1KM@;2<%HmaUnhbaaSO
zCDxD_O}b>~@!=iINIt>TLJ3sDvJW6Dh$JJL4x-sRV2j37v(t2g9TA^ZGNUJU<5D9c
zup3{P9DjV6uJ`0Pc7zh;&aHI1XK#@f4DzZIlDrQMh_iD5vdvoLRA0<E4ahCPpCCbs
z|I|k{cWxqRU4K2XmYy{_?23*Hu88%e>wk&*G*7m{+*7Jeha3o$wZsVOxlMPIVw4Hr
zZ|6gcq+3W=UE&Mii-`{erKPUE5;~$MvP2<>O=G+s`(kWxcDdkWhY3+Z>Sq;$?KMIx
zDCNt>WgA&}tep@9Mn*=6ll+!jd?)$%aV=P><2=d?1Egzm-C3$lUAxQvc)isiw(IX+
zfMBrz&$r?#HTZAta~2SB$;7u66~3%Tw}9IfzOynQk7Wz_sqU%J{d&6lVK#aJFn1n#
zSjx8m@FbW0Kf9pqsi?}&kNr{=A7%Cs|9wAVAMS_EbuBpCk(AZ*@RzEH&wqc$HGtJF
zuY;NpKu-8z+%80zCp13_7pKLJXO48xZ%3KwC~U<(Ho3JpxW7b3Dmy^qV(`(iZwUkK
zI9hE=#&n=Wr;pw@yTO9{ki)PNgW<pUbb!TUPZoi6^1&<zji4=C#oqF_pUT{s0+ip_
zDj*a(Te<=iaBCUgsl-zEMH~YHROH3gk-B~9>QYF1r$3y$sGzK@%$^3mjBkHrO9Np<
z_0x19abx0ZDWD#myTghS4PY(ixR}lUltW<q{E#b;BAqYgTk3lfzibR~<_6~$ly?RU
zl2_&+!rX!w^MePQYY^uc=@9>lwo4^|IRHxS0bwi1_jfdZ$cFobCfywYNQeie>**+*
ziMyql0-DMuVjH8jD+Q)Y75g?_9bwM_<Ic*dR{{&)(N(W>@i2ufJ=rMca%t9s=5^Ct
zL+$IeY5I;hZ`mVZO+ea&L+Ng{H3!zP-|t#ifth(%!5174*V3U;d2QdIzYe1Zal_f<
zF{UeYv95h`y>w{lngWs@NCYMi8{MQq4xjy)fbHLGd5Z21U|&q&Z8*1?WF6N^8%Vtb
z8tDot8g!uZj!L{?mP}d{+t3rk1(V<&DLTADcFjtXG}x3r;?xBL>^^kw2(I~BO|>>9
z0>_O#EIWmP-v@&1t0x}~-(_7umPv6O?xg8{5}3#_Pd73g+fC~}kf{9_Td~odF6E@j
z7xP?e;9MGO6I@-b*#!IQ9ONG#HJ8rSwivN00-bK>M;>@@8oeR|h$CW(L5?g?_);Wq
z&{DAF1<0FI%?<f$;jxjLG~_AawL3wwt~X8HVRVKM)D+w00{pz5Vhivk&!F>|F=QN)
z<f$=X*Xb7<ooyiWdBN0H>!_dK;ST+jt<_sLy~qTYN|r!d%!kA7sJ9Y=QR46d7^M_-
z7ZuvaYjXjtO8wKNKkralMRYL@?clZ-K9E3O$aIAT^DPBMbk8#$=LV<fX4>&7*4^E7
zvCOTn*$G2D@#hR(>#WFP<{(LT&oVPJ+l^*$U3FM1CecaG-Pdt-Xaa1l)COQYv`Em(
zFu|XkP7ad&w>-NrU!>|b+tN|L{uT0F0Mn&cy3Vf&&F6V+&xYA42xt*CF7vpQ{7qv6
z@X6?e{Vri)Ra^+HJ2)GGiwS(^vAkK$a)e<uWiWE0H-gbVE!I~`ce~&b*)`Q6_W4ow
z%u+JqE-0~x`L|>=QI%fMTd<{1eo6cKkq&a{r7>;MM5Haul)GDePDCs3+`E6jek!k(
z<xoKF%R^7ZWD(p0T7wce_rq}J$z2(0)$SdJh+0CtHf;-fXhmRB)yYZJH@Yy7eYq=i
zv{GF?Ocvcj4G^Xaq!?*vkA2vRB7Biz3<=r|dCaG8sPgBL6n-dCG1M}IBUy}I`p)ni
zl!-f<@=c9dcBZ%_Hlf)97K-qUVe=96yd`Ld&0Fiv1GbZrcGZi0SKGy}y9(_3r46M%
z6+dKIU!;wk5Fifwp(iQ^O%=rDUBH2%f!125-P8!+;`NdY7~~UHzX+zgyGe{B%WYp~
zc}0aW#<%Hun9!jdoD59=t=v$h6|`FpQX~T;>l^b>y8M)pMWP3~>Hs_~{Xbes>+LsA
z956%9qRV2=IIAcrDY*>UMZ`w~^dcT_9$b)?w!gE;CSjU+^C+tFk1XQLAKm%@2=Ci4
z2Qnhb<Je~Qg@%}!AE`GaW)#wBQi^BhcDihvcB!wO42m17WYIyBhUwEAe|(>N%(G`(
z@rD7zb7;1V(343_1SFzDoXHVVQYa*4Cf7G8k^G{WhbRC1{d)f{w0g+pn{KlzRrR}m
z<1t>yd3=|(_)k`fL&tkfja#e|`AwU|4=cuP{AHub5e0?YPmXCnf43%O_t(F0t@~-+
z$)D8CW)}U<rFG=s?)aFEnNO;@6b`LCq`dQ#g4e;F-~8=|rDbyqpEel^XNfrsg=Ga!
z_B-`B<vLubl$z}NPJXiM^|#j=OQ?0awjOhNiz?=Cp1X((HV)a)KsUy1IT|Y9J+hFv
zGH|w`A}t=xT{h-ybB8L*%d3$yO|M$R%Bz`hR<vRScp_#gm*9gThuL?&kQA?VZ)E`;
zq?uH#)#^`WpkdT`dBL|^6yo?YbqjiH{3v|cQ*gb?l2e8oXM<}L8IdE!pRzrN=o1v2
z`?Fwuz8jNPStF=%47lu<Dh8H15=S^5?Tp}=ga;UZi$NPMC?-}jk_+vzFtB8QRCKfj
z=2Nmey@WI{PFxpzS`o!4yK2-rt9xyYD+-rnThP6G&j7;C-nX)syaSWDIi^hIT1>Uj
zr_x=`6~+T|yw{5wO!wfr9Oe7FH$17Q$HmXrl060Uj0s*x@i*>@2YL|W97a6eB-9uw
z#GZ=WJ?YW;&K(tD%&FuqhTlH`t<*v3vhOVX#^k~UK`AM1n4~mNqCpdxTqSewIOZ{>
z%>NJe-a8!YzW*QB)m2?BErqPo9+ZR-8lp%fWQ4Mk8M4QzX%uHN5(&x7xMYuLpZ3Vg
zN@Wwu-sku6I_tha_w_rzfBcT$AK&Bn9pBp@_o$rX{eHcl<MDhv9}j&^%@B~&WfX$<
zAH^0*DClSD`ZQ(<G*6N{qM;|RS>uXoDK&rH-MVKs%E}KQm2swK4(Hwf5?hdY+H*i_
z(~!i9Jz!W8%#BGbCv;9PC)cui&dp5B9F76Bv)LFDIpoLv<g3{%*Vn-dsX<di#;<0v
z(R{>ApLF6!?(I~5=Gi$vIyxGW6$$ce&`jY}7l*==!;RPI_!>f&-ux+RJ9z+#DL__=
z`wMWC)t_i3X$qCjsZrIxS4G}Lj#o~ZN#q_1_V<qh#j)XqPF@L|RWRFB=jT^Eb4Crz
zGZ*6bDWR?$aXjs{cH6NkGL9rae(qD2GdF~maQBY1Ro@_uz5AyRzmBJ0SeSVZ9rWL&
zP8?<2<_^ALv@HdQBtiD?ja#x)JGGxAaZ0&ez!yK*TIf1sSL!xuD-s0ig7l6vkALLu
z+r0TGYKP(6%07PHaLRlRGmoU1MhX|E8#x8^d1vhDz8eij9o7p9UmbWnTn*jQ0PWQK
z8KoQ((`t=idghK`n=gKc>0AoID7%<E&G!;HP0uw({*m1Qd|JXi=5ZQnF`e^Zq19hz
z$8_HVQ+>nSLZ!%K2|6!x+syr+AzJR<v4eK4@DiJxf`U-%Q2!tcJ3IR{0-?f$bEr$Q
za<n%nbw)KdiGz}Q=dle+q)>2#X`{j54fJ5VU5{pEFk3Pfx1)O`7^m4IV%x9F$Hco}
zq;3398|K5x-Gxyre?3k<zaFPI4SG#L0Y!7)-r0A;n<e$}-x4-m_mq{D%>pwnS8|um
zVSQ;L_R18nw(ld@kf;bAOtV?C{Q0R4WQ;S=#R{Z3L;W>n+4^DbQe<6wK8+XeE{CaE
z<o0e)Qcpb=xaYjEgb|_vLa<LK&%$HH6+B!xywug{zpyB#BU>-dH$S;pd0n<?*8ZR#
zX@@ptrV{*RlME_}_tumP7ao%>)bwd=S(@jxg^P<%iq-e6@m7Nh#NY_q^E}T_SoL=T
zwV~n~9h9CgKb!iheXw+<1oZZmd2e-;2WlDPMqP-@C}SIJi_y`{tr=HpK3LO%{slaF
z!u^lOcwi|W<E?m%_x$5ACXza?-1<Tb`T|mDvYM?Y-nHqw)){W!gS`L&i_?E-weock
z7I|UwNx-jlac8%r&^o)iOrR{WAnU8wjwe-M8YS|c@DAoq(|f-R5k+pvD()58+*zve
z$|s<*6V@q!B6;aQg5bc?_xLPj|NbmIze378)O$qswNxL<I+26`x^wKZk&))-uTko$
zdW8FK9@zQZ2O6$pu~#-vAW4OcpMVlcZG!5aQ;UW+S{GoM*BGm`3+kkd%GeYh9e6*Q
zdjG*e=W#RJ@1Lrx|D^0EdQ~cb+xe&(2$r6jjLrb}WbJ-Bm*^V_I#>Cx!|wX~u!C%f
z?c<<Mfu&2AI$;ZVJr560fbcA;p`aNQe*<avcB>}D0W*)885tf<w;$3eY~COTu4OH%
zcD-CWMDUbwqY>Q=o$l4O%9L&0y44stbEAOXtxl_RF{i7M%k}${cSY7dJ+T~e%KAYx
zunN_}-KMX2YW_Y;zdxN2ooj&c>9P;~%{WMR$S#<e^LS#X08U9gXj}_Jf_<sf27pQy
zVpiE$-YOj-vW4NcWZz<nDl%$>86_UYF^x0X8Vb87da92jde<60)b6sSLk=H=NZlz_
zU9^LQ(L6TL+%U@acK;}l$$oTv*JeHCvV<&J)G-uQ)zf?0X~jdCO=TSWj9>b-Oh=ww
zoENvg<uoteBtQpfe)Z=?Mx<Q<7=a2}olrOpAhsghO9D+AlFgsD;N<i`rlJAKa}14m
zJf005>d2V-OVkm>Unyn=c_5xx2Lu^PmqyK~rRXGebQq+}$UWpw0@Wwv$qa7B%xwFc
zsgfYR=ZH9cpjik?Y~Qg%7L8avN@C-@7NwnMpWY(wFUH^<Ca9-A?M!|2=#dk0#_`$%
zzr%^&t*N1+H6_RqA4raP^?t`@hMcf_9^?}#j7SNN=@mz^)|C!Zhw>*9T965d1}`rv
zAb`ctz#tAEo}ALUXQ$R>m^CT`;r7_A689x-5KYJ~BtvL-?Ux{W$hqicMo}0NL7tL|
z3Y?z|qUuD%vvwmT_2y`3Xb`a<*UT--5XsWU0^xXW3Qmzuo5+gehyj0V%i^14fG0)v
z4Pt1I#cbsyoJRkex#f23*bQ=RAbR;6T^gcbASAx7Ow0N%mjH)Tl4s5DTeZnf0B8Of
z{^M7Gci&q4_a)G#?`#$R{gb(~5sC;VTd?_grgnJ|^ph3(2T_hx^@SaFe9_rm05bXB
zN(v;RUB+8+zfM<T(YjDq$^*>vjkvyUt{<RrF{4U@)bZ@@HXiKsjcL$v({8BIr{+Q7
zzBAXhp5ymVD82c{qUFaPu-?1`Ij>7YQvd!GH=5sz(fLl)b2{woDjY~}^nwjc%?q8b
z!p2X(f3~ut>a~k%-EkXFjC*jJRlAHWI)^J)u53<m2HPM+*gT}aGa0Hk*p$2avgM+f
z+iW%#2>KgDU-#r0iy%OZp`B<c|9uozF9@dlt|+7A-Tal6-~7tT1lGqTKZf3IxnnC#
zje^lpf6;WrvPCr-h{${QRy35XSWnshh{`GYl<3GGHB-1&TwY#>hQBWDvSOHovv5l}
zW0->+yZP6fEb;yIChvZ|NhXdKdB*m=9DQ&=P*8i9pr9b(p}siW^}R7I7kpJA#5U69
zCJf<lD}h>j?aQkLvYod|amzDoy3c6Q5U}%bteCCuJ_UziHUb+1Jy7_3xZ>?!_BOx2
zqcmGr;alp|(5wCAIbEHnPG(%I+wqn{E|XV@ycX1W9+K5(_majOn^^RHqN*u6m{67v
zJGxj{Btsit14!FxG=M*yt({Q*=i+T`(Jl61yMIB3=vuZrk&vov7ONk$O6>z4Ng^pi
zd@SG{QxwKuEq+%*?tgulwRaGpoAWS|IIG94t9SsJm?Bdtbhni}k{5t7-or1)Ks&Mr
zK|&SprHU#HJ=HD&fwQP6#2km8F>@b|0~M5rcWUN4q(1QXUzh;tr3|13c`8t<$gphF
zAu2JbGE}hI8wk%71a$<%I`C~XP<=wElGwvF;3H}eUjIRs91c2-8n5TS@M$y{8Q%nu
z&1N&z9!QYvLUa)o6%`{z#kq5VAt4FC@O#15tik*h(Nc<NtCbWNFSWP%eYd6Aak)LF
zEE&o72j$xej>=;i*;gY8uIfGh`<^iAy~8swO2E*T^1q6}w2vh)&%e(l)2*|ph0i=W
ze5Jdp@%D|@Uo(S${WAXB(~yz3{^L*ZKbJ6W?LYsFnUO*M`DdJrQ2oz8JM@a2%76Y`
zb!QY=|M<f%OpFx%k3W00V8eg%v>1QqzVN>v5LxlRR}iP+zh@8^;s4A-a{r6kQEUgO
zMjlK2i6-ud0q7JDU!<ieOg!ADtFwN$%3qi6;={Q2fD;ls(46Iw$9~B_I5-Y){_Q+=
z@;Ca<6N7b#=J*YMZB0Z&=3l>2bxK~o29=}suF{MXjaZP3k_+73sb+=WIidvJH(gZ8
zLalVPM*Z0EU#@+Vyx2|x5CrmTiy_K|;ytdQefesE561LG9gT8^!%w}MIYiJJ;7Z3n
zkqPpqYrv<IF_g+v-jJ0^{^dV@{u|<OIsWgA!K}A2K-<mF&riXU+;V&W)yJ@S|6_70
zyctt^mcf_>M{4^48(eQo*(MiIg%k9)^&r#HW0?inen|^9ITY11wMbhOX8ScS*mSsP
z^woQx6>dLyDGOiGH?Nal`qhGo>Cba0cIJVZvpN(iN6R^3m+);B?}fc!m`9`CFv0#r
zQ?^wUpz8SAT9q(S8yOV64bXFbirPCKBRCW6Jwv$At}anSFd`RFOd`;SezeWCsHI8a
z5gik=TS7wH%gYPRWI%#S{8N1NIIIu$VT;Ne!Y#u)?08&zF6^BC0jH@tjN>A(UDs9k
z(9bUtTNiz(@vhBmIjU@J{i0xYSRTOlX_PAG<m9|R*3_sbUl7$sDVG3WB@`Rah<F^<
z?!H>^yR#k3lsR<!0LK&EZmb&#QRZFDTe8o3tm3H$$f%N_vI|(}SptTTPu~hrhu!Cs
zAIpaybm%$WeWTUGz?8|IqiBHXB42RZxuVgpZ}UanuWqI+^b{_$@o>NN7NyLJQ$bpQ
z`of;xFambRJ<y9Oi+c0Tq60wIs6YKeLN*R=-z<$WR@az)Jx0WPD?<+$dUp=x16kg(
zZQFY_4*2_;Xfo>yWTNU~hZT!3roKj7w{M@GrF7rc!q@act6Z$6jT&44Y&Xk;CbdOF
zPH2G4i8qrBC@uG$-%&;*Z6d$FwShWLp(9^E{Nw)m7<(-S0B?}%en^B|6{&uX)fEn0
zx%9cTdsM0j=Gm9p0a<^w5Yt=HIQ?zCai;K!VWB_P`CiaxyW<xcYMMXQ6)FS!G_$&M
z*yhy7a*>RYk<n_W*X+|(IZ6|k(hd7DZVz|;8}bKNMFcOR?0hk6hNf+~{kli29=oh(
z-rsHURnl+Sv>MHx4DOu(TJCn3YaPUz(a|Whj6rRizqPk>)v+fG?Wzvytpow>rXht4
zNLv&}y&Y|&*B+RB?=|bgc#$%j8Kd!;*;$H~u&eS(_m%LE_NDQ4f||xYxP5`7;pNMw
zkXK&xxv=Abd~G3xB5N2M8|%!+!J3738mP-r($w48Wwv!g1zQ@0bOUBr10$cXyV{2>
zl6tUNW|MdD@hRYoDC1_@PX4%5Sh;FD`L^R<MQ<)yH9kF$MW=rIxt%*jM9xFRFdLqH
z$rzbcq>R80F9CJ6DTW+J%_8YVL8A+vD(+9qkS@4m|FuX6LKAAZmt!C?5lUWWMR8G4
z75YtE^uT*?4u<R&ufqcnH*?=?r8M&a@uzEYxcoe=7Fs@W89oI^+9O~R5SEsZP1kAE
z_$A}^e^G7^J1=`Ac`#Uns(?3JImN;Fh2U;xK9r@N8fc=ca2l8Ax5X;*1nyQXv6_OM
zdmW^}xE?NiJcM4WuMHvyn6{4Qf^!px{uVaF=Z>cA3Bd869e9fca2deDyMD}rH!vHr
z;>1xjBIJpMB$BLJw>FX%(<$krDz?-K$ZthqaAk^aM9gLKZ-RD%zu<~5xAD-`2;df>
zl<&x`q(z)VTg+@bvf$nQi<eEkTztO#*wJ7BXaS#Xdb&SlVS*`i-*R#GY_X2#UO`pC
z(VidO+_;hH%b&O=-c~oSE*qO*IpF$3z>;~B^8La!5VrnnaaYa&AaD0zlHZ}7@7Rjz
zkJS0TqKb-oTxRWE%Ma=5=sY0^$*pI}cnscks6HUCJRaFNRF43gTleg)GLSU{iP_l(
z^i3P3-132*CIV_qzzY)CBM=1xLS8RHZJ~>sfJs#^y2fHDc*rRdlZk9`3X>;uWKsRv
zsg1AoMH;Y*_VMR!0`ynI2jW}Yy_3>BV%>@b+u;lH$59K}t#~*dzxgR~`!vsmve1Q|
ze|A_(ANa8H#%)InXY@3iwwH%Xwe`2(!{~9NwoSQEa@`2D5Uv2?N)Gh+1S1YtA5Kwa
zf;Ejpf(6|<W)dbKND%y)oGh&;g+}EY=Kw!t_L~xM3SX}0=xBFt+jatvWYbIMEFHM)
zswSuu54QmN-??SWTRZK@V?1K4C6eDizsv)-U|GHt!e;WDl~q{N8v`QQRe0v)$trjc
zJcI21bEx42b*sfG9YxptQJ*T&_0tU(ac<gv0h4mi3v&8%E_R22@!ALAO{kR|#cexw
z^y9V2yJP%&1si9M638t%(~v&;?)vxYr>nx2tN!~EJ;Np9d^1Ka(P46lSa6BNKCCR;
zU&-ZyPHVM4%BI1wIS_!dHK~`_FhB}^ZpiC|iti+;uV9Pz43EDLw}d7p)E|KZ#vnJu
zpY-ERI2&~bcvCdQIeM|k^<wW3hUQ}$mw|bYA>SCIe|Lf|NiDcQs`!4$T?v>5VBs$g
z_#Ox(nlXkLbIYKx8<`ndyburXp=qTD-mLqQ!5C$Az?o14?7pHW?Lrxnl#s)Zi_{6C
zUyA2%MzsPGvN3qza4}I`Vu`O)j$^@v!^Z-Xwhap%zI~t}9}BQ0<>h0!L(nYwh|8sl
zo4uY-Qw6uL78<(;-DVwXhH*blhmJ31CvQG1ghyocU)JKhT<wdy_W)!A@Er?Y^Eq_x
zb*MAbtvltSpFN9{_F9`<w=Q{Tm%;Rxi=R1<Y~4)R@=?Izs$^Hr@w+D)w<BlAB7vww
zaY2{lNPrN%*fCMToQLR0i1N4XU%{u7uMP<gpblf#u*DsRX;m(u+A$b;pDQjd=6CV#
z2?Az-E2@oc#<Pf6<dpz4s6(XT9KqRf5U`;qTT)hbO+|6PTRnnfO-Hdm5iG^TQXX*E
zZb*6{IVKP=1*9P&YZ+xH|C&n{LrnnOB(QA69v|i=&v83<W3oq9YkQ!sQG<~`7v;{y
zL)IOhm>~4n>Y}8-kYZW<VB2Qxdq`<=z|`dstBBKqO3Z0?#s@2XcpgXO8U)??;mPw6
z_v)rH1tkW&M^3a}eY=vNs;E^GNCW^P+kl&5i+x1*k~#3C2xJ8a{wz+q8rV=IQeb)-
z2ZfJZ<cCr4%X=UR!cXCyAT0wjdOXgYSUBwijYi(>#o_HjLUq@4k#@#l6+yBHl*=C(
ziXv`*1pDHWV;Pnu`6fJ;kt%8C$1$XXvFUjfN}Jy9H+#5{c*}4X4w37DDdnX089?$R
zBLRHwC?q6OeA&`KJDIh^8J3Yn4F;woFN*8OZy(U5nPdhPHGB^43;}z>>G1gE8|(7<
zo_WmU<B#25rZbVGou`KJ%L_BbomCzcxA*A!C#_*F^O9X`ohM|SXWwCG?VNtG<wsV4
z%{<URQ#=jiF-EN17Y`O2)IFkmCvG0OHG@p`*|jWmhPq;qqhC4EukhvFu~|O9XkH2V
z#uM{KWslz;w)bQg#{T%L`vy6LzqM<(G0?D%>L=uZIeZzU6z=Y4F)<sFNrgeA(+LmH
zg5Tf=!0rnZ+QC*MPa<Eb;>nZyk;sjn`{Lva((}QTXzbhfcK<#6JLlRQELZ+P3)p$~
z&1KyS*)O4enb~}?3;WLp(9;<{FCefkkZ57jIF>GdU)i*O1yPNV0XlgvJNuB9`A~gQ
zs1+zv;hlH%IKzgppe=xFZq_ddvEUEL>mItDOPT=$zRU!anvhuv6AW+N_5$P&xOzTm
zctMIiCmj=$gaP<J^~Qd{f_?k>Jk$-BU0}YRSuzB5l`Eo{FO>~FgnqbIeIv&B?SS3V
z=5k2o8wK8erqnqII{QG6;B`?+$qQ}SXCPMy;<On|YdqYm@y$frup~eriOkHC>(<L5
z$ytnlMRHc4^JHUyxyfxA9Z-Db@I)DCA8%TviRJr*aShJWm$-aXjE(^!kle4)-y)&)
zAZ1~@5kj$l(l+{wj7!ow1-^A==4>7PN36?HD)7;@HG7-0QhIX;OLDcO;nix7ik!fd
zsBK!Y(a}@05^4XN(s1qx>*X_ttp)Qg-EVfy2ye0QZJDQV<VcLNiA1uCq#iVK5tyM^
zxwz!(cH-E`r5jA!h+XTsC>le|dIq;2uyp(UA>Xm1Jd{J`+U2XmSrk1X3ddrNzLpKe
z8oNJ!SgF5#n09&-f2N}a?mF(b#{5112M@k)7+ZSxMWuK*IVp(VW^{d0{j}N6A~{O3
z{t;G)!nO3WgIC56=S$&@t%>ug9^WpzVm{;E<XWeCo*grh;pd7s1bizTkyw2uCmjkU
zkKbOl{vN5$)614Ge2t`G_MGo}AM$2{9Z@XFEd6_#fDygs5m4cV7mx1*BSA;u%r5ZR
z*IRii0WC<#okSa#MUTMod}9{|!Ny0Gl|kSU5N*~km38@!BeU-7iLkxllug>mlHC;b
zxzC=rPQ+c0SQTcIw+egVJsF9(vrBr{P44^1YDh=@#67S}J2%~`tBhUNkRSfWQDtxd
z3+CqNZG7M|AeA*D)F_tU*tGvxXi95Q{hc+UQ(|gq8JF&-tlaYBg1*pph?MOr71O26
z+1A)#n}X*am7Nq|7G3?uMfH12eCJOz;^~($5?NuLmoo0(oaEOmA!6-pEMRd%+(OT)
zZ1P+E)9)uIpR8T^X#Vj@-GK7^obF$@%XyDr$bmys%~GjBg)49FHOxGr7vvpa^)}e+
ziNHF~yXH@t_qS*VXrC-1r;i}j$hb}**=IP7nV^5{BPBz=W7?e+>`x~rCuiGZ6+&ak
zWIjcVl*s0-xvIDuAE6hWj$a5nWOb~!w-=9C_`$1Rj!?x2&jiDwoN@Lotw693f_^mH
zfWbMDuO5ZcY#)k5z}aV=oN`H@i^r@Na#f-5fa<cBcs;Ly7<vIbzgjRLVv$dG?b#zw
zW}rwC!~)|v3W+J21uQ+l&X{;U_j;o&D85gi^eBK)vfP7ZkZwXLZi=+UD38<9Y~ifJ
zcOko~-EWVqFXvh0r)&6=LP^=bA-vtvKZEMSCVCd;2~LMA`W};I7tjm>#VU$ROKXsX
zlOTrp=LH$`?I3JM5)-m72Xq|1E+XunOln(}WjG0fxyaNKKVB^i0psnJx#ml!46=0d
zlgcFE9~$}`S_e(nC?L$KVhP}U|6zw85xoUyK65qClUTnfYE`*g15xaD3$zh{v0F5)
zA7hp{<LalrC6Ha@P|d#jb+&=jwiqvP1+wodYN$Wvmr`8+*)M4T6&ykZpI^$h%Q;<u
z_=P1>aE^n<2v;*^Cb7gvnw+X`(-%U=u8aOGI$|;5(_-P%GU(O9Ey^ZRKdB{m^~4)g
z`7fP=MJ86Uw`}ZJcG*JFx*JWBMxwM<5HCebzwFc7!&%F}?`<tm9}M7|bF>YtOR9!V
zY_F9I7~!+YTLc8ECvkfeDyQ&d_@MA}<SfM6ZcY15#}nT~S=|ONN^CSh)5v{(2#!S}
zl9i$un+f3j>4QHML%W3tfN6ac_zR?ctrQMF?D!FWHvY0xYT-gG?ZzltxT)50Sf{Pr
zvl5AE@Tcc>kRTl-_yf*w)zpq30jaN<pVn9M0@{@9q=U>I!Y=V1r6B)3rG)$Fo>DNJ
zS^h|nmd^kDdFdaGs4q;wducJ%CiE!NF~V`pq$Gc}n;yMF2(`o+NdKUDQ|+iAsoFo~
z*{3|WO;JyMlU{J^)}I7WsGGiVTk;zJC)Y{TYz}-W@G6u{v0fQ?vhdtE#5XnujFY7c
z?oNK4{7?N>eoy#{8|IhFAjW%$XmQB&3}-+@SSz%$?O0%nzK|Bj;;E)9$-7XA_2xhv
zFwN{OPCVa?;737T6e9aaY!4ck4hC?|*_Q+buM=69P-el_VV5_sBg<LVBI|R}z-wCe
z($Cpmbpq{owd_lLZ=ub<-&^T5DsCufY2mYCkf%QSld1SrbDnx>+nu99FVapZ2YKHj
zK_OW{pY~p|(wwbjm$)&ueMZrGkUlJihV&|n2Cj@EfySgTWvk=Q_pEiGHP*~wY??+6
z*la^%V<UHpej`5Y2t5x#LoPJ#T%Lf9zz*({2Dk*7+N9`QzelYWQ5L1GYxvlXZQHJ(
zo=X5+JMRqhwxIHQdhvi2nUG=heR6ul<Q)jga)j|=$bo5oEGn&z!g+8}(=o-%<Jr7Z
z*zD8A+{*I2Prhsd$D&bE{Fq<9FFyjEZH_VxR7_%ZSTjdTI+P2ez{jAfDW7LQRFBF*
z6&Op64b-#}fP5`(iCM64u`#Ma)W@|bfX2|!g>r{JdK3@d6^jEMzzyk7dmYh>G>~45
z#3XTXEi$M@cW^qz4%kp7M)U{hQm6ky8%l&4rnJD$Qo()^M!~XR-1|4!kb)5xp$CiN
zHc4D}F7*Jy7Hp;9JCgD*xvrameoS0Kou6Mo8SXeW@cCu)&@jKuu3fvhQ&O_OEjwS5
zgtmy*4;oyyvc=X-x}nQY;54yHXO8RsJz|?^gdTt3V1vQs%X>iTORqp_u9b;12k5AD
zE>ecHnl1{NGCTM0;ASrx7{A~Aa!Zzf^A@i5JB87=jz4oz++36A;b=qmdtIK&|DJm|
z%RV)raN7%&y+*r+Mf5)~cjeqf)6z|lBa~6yBiX26I_dUmYC#}Ro8<YT2{MI-sU+1I
zQ~bm8m*?Oeh09)QCsExut2<BhU#j|+qSSl1cAFe&9!e?Z!e_gEn%jGDk%{g0u<k0h
z{YM%u*!U!665X&VYC%zPaW#UjQGfo$DI|~lg7EwIBXGZS)D+QO9K=ldkD=IS&w3y^
zWf$!W`dR3F-abH!=o+KJhdqhuf*R;0DDmT=(IP_#Y#?II_q(F`B&N#?O)4=^hc}0c
zkl80Cq^YyW2V;O8eo;SY&0AUkpgu!?=^#-6U-xNrAOu?>a3&TH)X?5AC@t6pJdaMG
zVFZmLN7ZHTt$3Ll#Qg-cDG9(ylY=DCN4g^P=}ZLf)%bWoU1r^Mm8i({k6Ev<<&Wlk
z@5C<Wq4o^O+ic>WJ^lt2%Y|JcE&q!1Pm1cVFSYv}=gU8p3l(s0?in~so&Qell4bem
zQ6XB6PUF@eHsAcNw+wRRv~uQzwQ%>ciOlkC<K*omErA4D0*@6IKDN6G{0ht8E6)cN
zDW?WU7q#7E&l0E{klJ7v4!-kXs%;CxCKg%3Ut1xu?%urlb+;8s_vn?v>0*GEv9Fh<
zt(cww{Cb`O(J{>>%$$G8NkmR+G+WKDr-D&chXiaM);)emGxu6NZM`hvv?ke|R7SUE
zOZ2V^b>LlkT_1{SOWE(BKLG-jSvv8sF*sf+@<uPcWg$?wO;iEr3Sv(1#bn@!3*>EL
zW<U~X{NubuxJ@Zv7GZ}9N!Wv8;+7SSbLfQn<K<x5^aOd<SsVzPVZX#YmVgB3z7bgn
zsC|NSi6TZgfv&Xg6qzky4dTo6!y<q&#vn-ROaL9Z0wx*(E1uilS?=2Nn8VdrsB!h+
zo`$FsC{}2uS=;~Ix3?Rub8-NOMTe;=qrFpnT~V*xGi<vvErX@Sh3k2vCa-R2YN+X%
za7i%n8iqUohigGO##iX;5g<|rNn5w<S6gS<VP`|yx`T|iPU&s{fC8t9{g#6~2eO6g
zqbDoAbXeLjp26DNdfVjv3fToM^qw?R_LfLeQ)qgRZ9uAU${+9Msf2!#@ysXG3=a9i
zc|tCGr}}m2A@nv7IF&#!A^a=o3jf&7BGrGSqz&w;q>&bG*=~<|WHRtWKG{t@=++NI
zHP+jZW0Ch``IM*{_4QvWgt`TD))aH>f2K#hz~F7w9I|`NOwtYSlsV8FCh{b}i&8@e
z(co%yU2gZJWfzM6rcCqD27Oat3tyIZ-{$A6?C$953UQnA-i}p`n=si;285(8saBbc
zYDEV`(`UW&fCVvBu<-5o+KOuB%uH3^g!#dJaZ96@7gkM8FTA16I!*zmp#A@NCYG!H
zekNQ~{Y_nbE<dNem!n7q)yeH2%4AtaUWs5ek*^**7;MU#k+BewwU~Gr-oov6k7GV)
zt%=_-z^);9_{<$&ob@3$Kti%|GbQwHAgO%H-W@~bBWTp};cynkMYsOi8ge?5+W9$m
z8f7I<d<fUd4ySvr>aC<XzRN|O+lMRD0Ak73Q-$->&r#gPfKN;u4oq5eGYA4fh_c=5
zOVn=6rG0*3co&o*0uT}}LovJW!2qsY-Ck#LB)4+QViUo6o)RwMocXF-c*WapO8B}#
zS+PeFyr{ASc8(F!8<;vD0aHkP4i~dcz!G&LtfF`$tdo-eMBf`yDPXq#FT!DIX}M5y
z@tXqb355xOOOFQgu6{JZo>dHO5Gci{AGaN_NAWK<TgpkO!v1f0n7By9h>eZS7-FL-
zK<XIUJr0pGSo>Ovb#N-mohr$6r^G0!N4<K$vqh;5AVwtzIA0Qlh_Y>T%hfHqT8ymZ
z%&R57?%wa`q1iCd)J0<OQ^OW#*8%%}tCFuI8=XY*+QJgNj=G}5uB~XGt7w42n<Xpx
zVI@U=Y4pjdC&yy!SB-SdZ@#IeukAgw;>^4->z*ksUin45Aum{_LIu{XD6?sP8}_17
zUx+EuVfg~1*{S2o=+`B5Vq;>OBt<me1MwlWCR}2P&x#=D0EibtkAmmn0;X+haJ3ji
z9GS{jZVbIRmk5GSw3g*dRWR$SFno9p`4v-4kO{x5^2?q_x@YY-XVI!}M+%9EH0WbZ
zy0oo=(sEbCLOYOiXIi-VruKKf=vCpik^E`~gdWf`rC)X8sMlT#58G#nIUUIJfXtzH
zIaBupn8l;Pk?zv>XgFpQ8hF%PH}FH9pqM`K0RIMc%oV(Kh`4<+nKNm)d4)sX{Rnlw
z2t^8HaSd~W4`Is&r{I9=E%|4Wz;SJPG_}`jV4<L}Bv)N~()n`|tybHidtsjcbw6Jh
zb3T@-5VHX<vw_q5hkus}GReCDU6MN3=}8Lx{N(hMY|{-rGK1;*XR`p<V28DHhdu$H
zl)V4<QR)Gy6gEBG#fVaBn2hjmF!TtPI)7$TFqKOl%*6;CkSoymBCNBQi$l<$^Qr%R
zgoF<ugQG~_cOs*opsb64Inq(!?5eK}F{W!2s~E5{*b$tkp}YT)1pnUP*ZBw&WR+rB
zJ{_sp6v*?@=OO`#+}plZs+znum{@h_>snUUSd<uXXoJWM1q!;_WRNo~V2t7I#B*O1
zL}W;Xr(z@*OCm>+7BcY+eLtV$cLOJ!#WmQ^?^5o*CLRf%-3lQGe>x0&i25TfEe(X|
znArxDcZ>SG)H+Pto(dqC4>Yiz{#sP8{Jop>?H}I<rtE`D$P_b|m_eg2R;fY<_n;WQ
zPJ1h}7MJ!S>BICEJ!<7<pn9U#&R6d@-@IwH&aGP5=&>iN_J3yp%j)35=qf|tE~3Ns
zKki4ju=cm#5okvl)=GQ&ef3DUfEcUffoH&jlQXyBNY)J#DbuW8Qi)(r3YouTbc_8&
zDC}xrD~8p!1aAdQzw(?s-6g*Sx}4osw(=OJ5QLiydg0Mgi$IM1icTxlHf(-uRC2YT
zMvd@H5Y8J_cslR)%(8H7?K-*AQ8Z=lm|RG7-_|=+?q30F-QkGg#AN0)o7+Kkm76g{
zH=}-D)>YT}b$)ogC!xY4hzPXOf!%!K?|*fDA)bIhgpgia3i8ZYA$?8}J6Aum9<A|_
zyjD>z8IZUmt=vua?Kv!CO6a8E{&lw>a0S(KNFuok>K|^`X%b@Vlg&;QWbV^Mln*iE
z%Lyfolzd13@<k!0iSu9j`t3_^Q{B?FES_P~VIxRV@1COX^P5S}gj7IZmm}e{-@-+|
zEPKn^g4p}Lt#9*lC`Dh!Psq;)9SeP3+}5SHjUsOle?(dzxnSl)@GehBi_7W(Tzd1j
z!j>0)CyimhRd<_3Tln+R*_WM^Y!=K7FvzPW^F#DZToO(y0849tdxlW%U1D>Z{Ym4q
zwC6bux$Cpm*4Fd=#0Cu@{YFWbY~;=7=!oLcM=s*_S=|#wL6S!OT0tKI)}f^nxwaQ&
zZ130{+Lr13zE&@?&4WqnqR+Z747?ETd+Z(6sGmN8H(R-1-}ZzS1^{pEP6XF=Bjmq~
z+NVo)vj39ua)y`X#a!T@`XDY~u%>`D<SIb_F9-Jl*)N!%Hcin#xB<9CPPIq(_qM{e
zYi)Qu2Nl!!HBm;L1$sB4iwGO?`DJS|z!8Uh3%q=9YIP84aPN+y$MEI+T2lWYj{$f|
z7mRPee*wy)f;Bd7^HY~grxORu>M-glwb`;BMV1!gl8^dPqg5+%TGx@nbxyHkyytcH
zPalDHU%*@c&H^+0K8|FuH2qH=(B7-zJv+dm|3SPfXCZkcW^g#ndGY{CQ5%eOe^eBz
z&Z3Ssdsqk#Z!=Bi0itsTHp<|%c6$)C*4c?wJd)qf({rzYb}>Z{!yo}mEsgc`oqx=+
zXG{3vUIfzHz}nY@2i}nZ??0g|ozYu>E4C<O5PPC$Q5Mf;Y%Hp~k8@B>_}sY^5V(W7
zDX@`?nIZcfh0X0jn3{x2@>>Ar5WD)tUAGf3RUgj7M-W?Ejdj$`T5~at02Q-Az_tLy
zfqt^%hXVTLU<$fDvSCwNOf!~6>6pH%^744zI5v66lxNnW`Mn9+E$$Rld2Bygx7rPT
z9<b`ZgJ!Z9fl)+4vgyUPsivHhr40uCMQbYXcs%^|c$ixLQ7nYQL=_m>%O`0dBG-8J
zLI}H*R*$B<M%Hu-=HvgqFZN5&N?6=sq=0LeTGapNVR@gQ<7#lVo0pFwA6d>QRY?*E
z&`!ONuA-hz@UuWA*CAwqPf$xSa~0UX;y9r(kLaTN?*;RA%<217mX#D{*x03U2@1#3
z6EbeI@y~Gn{0IEX+yXUOmb%mi^w~@0#VZP%k`OQmrzL3wBAov2eKyXVWDG>Q?G#-A
zE$&V0my$hz>66>i#J}Xisb%UemO9-UO2p|2&C6dxoH3@|6+KAM(d{kv(=S0dM>iQ3
zmIS7;A`nc;(nFx&LVY`)Oyfay{+nz%IiY5_yq=Y{+tw61u&=_~kMX$txK3CVnAg;g
zi=rE9fWzDAiH5kdZ--|gc76;4Ai<Y!W}zNXjaaD?dxopoeE5#!vsH7@#NM7VL_Vw=
z-(bNf*sQ<OzcGR;xE|qe$i2h~h6{%Wo2sJaSz&72JUVGy8q{HGf8rP_C0XCAMy9u|
zL(l$;3sBtpySfVBS9RfB<grCHZe@Thp>5D^+eTIuoMIyh_Xlg(Iuq3%bPvgNnM|3)
zO=)>lG6?|wN3!~lVdV%mkpcmu$ICJ5w=5bR)cf2bc_2HFck{n*D?tGQ0@X{DJ1pDu
z9LYS`DRnBI_Y>Bsk|ESCg^nMoMIw1E*=bgJc5+P$O&G;)`xvgBYmk&-2}@z}x07;m
z!wv(Lfz;R6+>HI7VdUZw&4q`els<%w2QBLJDOjTXQ7~OOV5I`(Dq9%c(#HT{kP2;4
zZRW{~c4M{Rfv}C{qQf+wwye`7`Gl_@Va%lZ>22)ZgTBN}$kYUx!NWz^<H0)Rc37@<
zxKPZ&OfbPo5t{;lFl1O5vj5Tzh&`}I$Hwjf#RSlBHDVwkeT)sZYT=1Ki#qxEaP)^{
z^!`0z_MguZq6%Kj7-lepeV%F7_!N77z4(#@Is)i4iott4g+6bBg=aW_oG$(TG5T(A
z0h9$2u&xk8wKHkXOoGzIrD@MpLQi1rlN{~B{P)N7e7Q*5_w%Bxd7yuoe0%4`Jw&ML
z;jZJ4bUSD9&`cC@0Ob0Je%Q?MOuC``a0c!V(%BaILunZuj9C!?;MNi$6dWR9v(BKK
z*XG$<nm4{6?1O0lC&a2<0!UR%=c73*DMCAW69xd;bG?DxgkN(xW($7J#RGo#tn^9T
zy-Fcq&|OEHnW8U)-nVA?ZVTV^FHCwbM^FRSa|9B;+bMuSA{*QPi$P*jLO%c=?(Ua>
zvNQYd8?fwBey0w;dao#YND<8OpHN(Y(ZIDqt)^WOjC70*dFhjw2y6rpD=6WHuq30W
zr+3|stWJ(T2Frk|ze6F!z$tMZHc*90fwu*aS9dC5KVgD+xCwDyIEA)&20CE$Sz3EG
zs%*`mJa9pyNOpBwjH6qSG`>Wi6BG#!pq68Y$Sp{p<?JThFr;ijG-bw|V%Oqc%>yBD
zNS^~JnyG3L@!6@LFQqUR54KU>z2(g=^C(&Om)dx&VE;B^AJdw82n!F*hK-GoHa6Eh
zj~ZT~5|Zc<y^h%O-7k@X1EW*v2(y|Ca~H4$-KNm|(BEI*Ne5Nf-kCOHpJ6j5Nbn(o
zH(zl@2Diy5z;L$8V>GhO)gcm($`*e93R4Y;Roe~Nh+&IDqO?+i!8;Qm;!zS!YS~vW
zwbSPc>6be?I@F)ocNDE5{uq0HG$L-U3%8YV%G|0uvx5zJm+Lb|jFIIH76ztoJsHhf
z)Bd6cJ7Mvhu41fcMJJO*-}>V`A3f}(<uk#b7@5jph490m(PA?!)&zwo0ebmr@w9E^
zTbY-0=AN|&$f)t<=Bycy7K;ME=H(U^UEAJXZ28ls<y~K}z>0pWbw$FU>(%5Q8<Xff
zR{h`?53RF%2m`R}&|s?L$c>^e^Ov;cUL(zqEAH*Ie7i}$v6Xk5{E0W&m&&9Yf-gO3
z=GL8q=$S!QoMQGlJI_|&_9Z1H%A}bCgxClCM+ImNVN?R_(;yd_7{pWa?0kXE5(eUx
zN1#0Pp%s|#$K%ouOU5d2HW(l=q6fjRpf7d}3wr&7c|&TTv5^^g^8i+Ul99_F(w*SO
zLD0&82M@mWZP6z^3-Wb8Q9uSrmVyzoR^jaMU79L8?s*lMZ%GKl#%N5Zx}gN11MakK
z65o6e=G(No`{?=3#-0fyD%AsRg}P)B1XK#BbL?rw%EmNsoCJ)?7rO*1?*DKWY$PH4
z0t0j>xuO=!VdSAA(>e&jN1`y?EhMChi`jsc!KZ*Ldohz~fZ`~=(T3vv_uM7@L&^H{
zwyj%P91yAInxK<W)z5<s!U5SuQwOlEUL$KTX$n9Lrp6w73`Mc7HYX4M$qIWxUNr!m
zpIt^2wCx%Vzgo!E68g}yRT!xCjnNm9nQAhWOGHBLtFfbX2mH_gx)?4fqOvL~T;NQH
zFbz0EW^lNsP?L{9OKgf@O4y{t$!B&1ll5n?N;v@y*N2s2QGA$+C!Nqo^7BvH*YQ74
zPpB!T4YA;ARG-zx)vN<Ip!YZ7%aULULH8S7w-C4|OU7ctw}pEOV=gwq!M_?9TsGiq
z9PBNc2fHkHrL#YtHRlT&&XNww4ZdHPBw*yC|D<`lxO`U5xy&6q>=%(uLo3Xr8w*}U
zz<PpEAY3zdVTTP__0E8`>^-(ikS1wvmw3Ks?aH^m=n!UU;#y#*nAEu^zluW%*SVez
z*eJqUV~VhSPu1&Bxwg-f3YOO$I<Vrc0|NEo<TIo8BX;YR<Awsv+3pbO1t;@1j26to
zhB`_PV6={DDwq}G`T`)05udq*^N-%qXefN|2|ylVZ~xLU?W5oU-UD?$RA-Ih<vka&
z|Ar%mA{#~rF=NurT(w}A-GdI$R3bI+aE*swl?N-?WXcZ`L9}X+jN*9Od#3Tu&+bWf
zh*ZClszmerDYu3;#Vd+CFb5b%A(PoMQ}|w0w`nXU+PpgmnFSoyjK(g@<Wq`@BB*s^
z-1|(bBChv_IYD*rSHS2x)7Y}B`3#o<fp-d;70Kk0K|pXCti8;3k3+Y8`v%&*OSi`H
z=4(QRcCmT~ou-^Q*t)BEf3V2q&xeP*%2#g56~rW)0u+=?vNKL-{(F**fusZ`R^Ax$
z$hEnQ_*6c(TDNvBCIU)?ZQ@O(W~mM$h15d7rnG(Sx^+TU-=uxm1wW+N$$!M{tA)I9
zVde5nn6Sztm@?0_DlJ*L@fiUk!L*a4x?+ei3XfUmm(~@&7k##z>TAu*f-=<-KnJhw
zoXJrIoz(=rD@fK421ai5f<87pJrAbZF=aYt-73*?>(7@EsfEvOGv=5)!V6kxxW9#v
zKM@q_JlAZ27)iIBbISvC9+<aXL5LvJvstk$m#N{6haLV#xJX9!i9=1lAmjC5;a#Pj
zFh?P9AJSd|W~E;(e9eJ!rhfFmNVFa#>1Fx4aY~zk5LPB_*9lmQ>w(FJD+n*7Mj99a
z1z70fE2e>#oSOPD$qH5lLRBa@NoYzh#%{NiG^B=r7B_bLk8&Z5YrKN^wG$w@tN`dH
z!}bguRz@-EBFs||wDckrf~0YyDZE-+>hCw+yxB6ymGg3wwX?zIEY^3~sUL!^WaA3a
zRNtgJ+L#(#oj7p}P4z=G)djzX51-$e{2DE@ygQsl|08tn#uhTKLx?3i8|A?Q$Dtix
z3)~_Q6Cc>corW-r5^AzK@N^1GlJEeM@{QbUqIQ@8aYxc^6CrrA#*HZ@%IhNlp_j+C
zqRgy8Wi<-M`xWrLrqRxlJ;yLf*QN~F_mB)aiZZ^k4INYxspwEVidlWTMHpxV!$poG
zR}Y)7@gN`MC}(jH>IMo_RYLLwkqLgH{CF)*#Iy5wLtD%z;@}gKoGz-~e)7RaA(Ll_
zzP;!zt7oT1GH`Ea!%r9)#Xub+o@{)hK+i=GuLC6MxwN3045Cq91OdJ3{5cOGIUVgQ
zF_Sr(REW#&gyMmrFH%BiqXbO{e5Ho7P)7(8h}2|bQp|NaoxD8x_+)DUP>Y!TAm?ah
z?;RXJ!lsowbEY1X39^Ptsze%<hk!<eJwcXz@jXcq3nE|=Q06@t?yF6Kn|7de0%0Y4
zOkfnF*aVykBp#djOw42n3xu#V2z-RmDItD>{79IBz>uTxa)ilC=xt`Mj?7B{gx!r?
z)m`=|3RiHw10Bi{WDhgU?chmz_)GCsSqX`(OXel6vaQ<m#|?L06ZY+T$KGw~Q+-|g
zckT1Vded>&ZYrf+AJi9#e#}cN*(E4cqx_ZE#MbIaDT~0-_p5pCN*`S9#b&V9;8UU0
z^rhXGS9K>!gjaSuewJz(%B!Cqv`@*XEVZX-@4*mwBj~i=Xgp~{ZWDP9z-JheP>^q>
zd*U&YMOhJr;=LNr*P*Up{cq6Gl~<`%Ib|x}-#@zB(be2sTT<?2rf(5_I#;uyeocMD
zlzIb^D<vvh#Dt1Cg}_yz?H%#7Qpd6=k@i9YeTV`;iwNsM7>ON4%d7`?QmBvUL61gt
zt#dw!je=<wl|Hu4M)d>r13W5;W6F|;EUnugpGvNxpPAZZI{nR0M6q(?;g@nc6j>I_
zl@@bh8bZm1TN0x~-e{Igi;L7<t0=O0c!YnV<_YB6>YM!uobYpEK3UJl_3%=c`y+8y
z#M0BLFr`9@KCA2@vxe;Hjy+jI?0({pKWFROp960eVyE$Ph}aqLwKW!Rgnor!vMFkB
zmA=)*uW~(~JwEQEZ(h!<a7p`|F@D1&{H>g(7T#*R>1PLM6GStoRipX|`1x`GQ;lfm
z)V&&(eyb{?KBQV)_At@i=SQh{G@I~DXpR}9^=lt8hVY1Llb3GtV!ZTRIdi@8L6vid
zF7Gy1a6xJ`y09t>eD?-zg)1AFR863xG$tw8<10Q${6m5^@ydP0V7a+BBR4mf_|M3%
zRxmu^QLH+{SwZ}zCuvfHqtgKA+93qC2YwYDp&3HrDl@q;Y-N|M>kfO=*E4Os)LOw@
zm#_15|6{=^eT7w3*VHL#H#1mPst$CYKBv2GPs!p0HFOB==vP1Yj-dc~4=(P=T50K2
za|nqMu1S~<=quLWF@D_U88;mKwj<;XC-DIqD{#ZeZUamtmc4HZ@r+4L-Ua{C3mKJ>
z8A42_)U(!|o$KE=Tq)|WTR+OA;4vn$24DUHzC3JU)z1nm)m4b~Cm!mFxM}A^T(y1i
z#Ap4v=Im)4;ig_2BJV_ulDBy?-ZR0KnIH{B$<WTp71SGIQ@d9B&m)mIU<3Ujdi!x*
zMmROKV2x2fpfcj*_2z0t2USSoNR9rS&#&)$Pbl@NSwsmLAiCj^7#f{Mxci1jng6IX
zOAUBpQlo9?$-J5^^7u?lOypS3b_y`Awhp$0hzsAUm-PrZ*~2A!<XP4*-PSW#$Vf5u
ze$Bc4e7TLW`N|`b>RIQ?`?oE7E9x&rzB6@9gxv3paKZX_YcislnIfu)z}v}a<!I)c
zYbSe-bpihJBC@R8ykibLdfS0-N%?tK_Z@K!QubJI#7|vI#JsK`dDmiGSRuxRoziRl
zoUZzx-_KyKHC;o#TB**hLY>^t2yZB7*hDi&e#0x4ZDvy9OEi5T=u!OWIFox>;zQ$|
z-7BDEOMau`QC~x1?e9Ho#@e9DffJ33#*-{v_bA8L&XD{O&Fr<T<M1C$ldS!`53HFN
zFtG%UZJ7FkKPd{!Sa(X)1FCgIcktgoTmR`gvEUyo(M$iSnSO*wVEpUn$RGbOx>);9
zqYL-{T(@@r^S^KTf8@d{g{E-51VO(VGhvqjW+?6IVbH`rL#*N;n7C6J!q7@N^`NHr
zt(G&vU(F9-y;YUi&wZH3Kv*3r+DaLK8Itk`_E?^11Z-8&kPM|W15#vXqdt@)<||5F
znD(<kc8SjxoBK*v;VMNTntsv545LIRP5HO_662^S75{&}`4Peok3!bv9(u|3VjJ)=
zywpMA0(tymY=abbd}PtQ%B^0*#B{p~Si7=El`(t?J@TRA?zn9q)IkDclau+oQ-$JH
zcB{Tp$EUG70DKOneLZJV6rpKPWMvVQ$j4MtAAXAetcwxeWBXgJhVkhG|LNQD-!p&z
zzfbmmLnhPzpA&<LJn@@S;zEPV4HAPlrq2dkc~B;F(xt<cY&|jkK0}*<$1n%_0#l`X
zj6g88l|+>9O8#=B9Dv|_95XQf3#hQ8zR-HHg-HD17D*xIBBBbDU~)wIw7FW}Uyupt
z6MzobQ_jnPEl-I4=fAi`JRzAJVGfY{2_tcRtVAvnmr+nqfY0$xr-?!Bzx5e!-XW(0
zb=|-0iE!O;JOZx(oP=+1>tpxB&4KYRO77(EgzU|lNFu0J^!%7qLYg}D4~8K`Q6TxN
z|M=kCM3<VV-h=%RIN{n*Ur!XHpMu-q*v8PSIig>D_YbBc{vvfm&(uAixc5GW1?i&3
z36!sMKN#;Az~zW>5ThixZ-3h<AV&5Yt_l-VC*!k*WQM5h4%}@EON%46jUd5|_SNk{
zOCOTu3c{usP>F$0*wP!$gnI<8kc%JdA}06k<_W(p2;A4x4#Dyg!hete`@D*wV|}FY
z0r^KU3Zx2AT0Q0V?B8#BhFqSI<o|ffCN!pcm@#@y)L=B@fDus?_Rt+Y8JbZMS36F?
zX5v&x-x-Kj9(jFlqtNbsGoqO6{r+w$f0FNJ9Ky(4i)@;%#C94=aaS{h7*&@2`a2Wl
zc$I|Oqxkf7{pd)20dWaA%^>l9z~Cq<GgFv)mQkv)_>#+iQ1o2uF5H@(anI~+*o902
zco<s4?%urQSSe6RxfBsW6n4SgoDb*z*PCr*yqVLl50Zg$`om@q&W~~C6X;cln)j0}
zjElZ=?|=TghEI99@Y@wbRXofIH6(n;uZ%ldqx?U|*MdyXunoqYie7rLxd;|wiLu7@
zr@vm1lktMeF}df+)xL`LJ})eSiWrAtd^Poz@y+5GIkPIAQ6zdljimN*!q)maFXJzF
zGBaL4>_0A(C6L%^z_6s;s81CV5=zCAK@SWozAj`Gq&rtLe&xXL_hhlPwPkqN#-;+s
z9ed08H%dPlx2~5_ZL%mNQ~S(%wn#AEv4nBX53XdS`@T&$RWV=D1>FU&ng&ecqi=jt
zk`EvnX^5nHvk@^6*s&Hd4*LFI2mR@l-v^zYwt`2p<@d3xyU#dw438oul}7H@y=%f4
zznsB1w0-Bb`j|GcV)|%N1|QVgOYnhMf^X^DYS+w%fm`Z1T{LV_kR1^~@v&BVG1HL(
z(vK1!5@nCGV?lgc8=wckPf96+O*jooQ-~I8&^?`xe;aYH?itQm{7rW=NOZ0KSILD>
zzL50x#D0;y2C<qZR;?O+3@?R<1nsM@7ZJyWK(Q4i4YfwmzT^Y{{)u=%oI(EmbMn7c
z{r|PSOT^Rfm>3DT%}z4>a<N9`i(Os`2xxPV96_|niEsdpr!PVeKOwGU_dLC;*Res7
zvf^J{fFeJlufCYa6pQ?Ra?u|%aj-Xh;o5#a-fBlROPe5z+=Kq&jNl8<DQ{1&IdhkI
z)nl8Rd!)OPiqP{OI~MxjSrMsv2dp8r6Ay8oxgJ>o3>FDPHU2G5Ys6pPk2rkdS}5J;
zO<H{FYCrE|dC-Emda5o6K6~zFn3%E=FRJ#(HWgk(cb%PP2aj2zN=BX}tlIsUYsfNh
z-T;+1FHq^;6Q_9~GV^D9{*GS_1eOZWCb9jAW7i`hdA%{84(ySJv}is3v)@ma+e_b9
zH0CuePA12VpC<!rK4f~L2f+0whFmma1_=5#gOIjK?qo8S%^429tOY-APquVRvH}_{
zJ5+%FoILd;<gMTk-dix9kwT>5+riNF1_u8g@H5A?*$c@T2p=C17xWM~K--8>2Lmoq
z=cO2N&n4R7klwZwzyee-c$tuDj6O<3!?j5;=j<{CxAeos)Q>6zG=UY>YRCiv>X&&3
zBbKMQ)>mje=#Yqhw@9D5brjQ)%_Mf0sF|Ca6U}%}tg-CJl?*o4g>D8FBOMxM=k4Bm
zl0C89K!c>m<Jgl|mJ#3*q%bO0QzMTyhvAa>6mYCrdH(!)R)lOUDpE-zAxV{&7guE+
zb}AHF02giZT`g$_vtJUJwl&FO)smq>4~yRq$YxyRTaB=^7S`6+%isfHcV;^}wbdXC
zCE2%N&+QvlI_`-NW^W-T^$m!YOd%U{&X)Mh8XMl{iC*jaU7zPJQWOne_#5;$PS4u!
z3-6Fw;C{zMtq=dWuLdhP4B35RNB`M>ADCyJNM@b3;RjdZX6oGh&`?|Z4x2Gr5yr-z
z*B37Nfb7AsN@7&$J9v!I%uAN(!3^y#^et$oc7&-sXm6;idw(T1Yy$)o)eplK4I5<<
z58fdcDfM>9>^eUj6)5i(U|k0k9fX5REm%FX`gm$bpz2Y(<3&DyFo`p+{TU2G+wrD%
zu?Z9)#xi*yi2MnRHLba&u)LT^4UEqA#l`@`&;k28e6~B4R%c{nSYbpgiziBcBMsvo
zhNm_b^rcZco+2)4aa*mU0tBb3OLj;tyS+nX8IwXNDvXFkg89NS_pH51#9lJBz`{uA
z$mE^RqIbR#Xzy^0e*!UVB{tjgTrvbo=FjbUN~kqYGcz-5zv4OuNZ_J}%^KU@=ehmH
zpJS1GOsTFLQ-oP9t_EJ7xL1q%fIn_G3-YiJ@qED8wp<-s7iMh*5@Zs9`H?R!{QJw2
zBaaS!|NHH?h?~s2uZ|vE4|Xz@L<gXDK~^mYHXNkH_Ivs?lK55VPQGC!XJD`GPSsAx
z&nJmfI$nb-%|Vr%HfES<W76>+6ND)8_`O=Ngc!l;V_uHc?z{J97+%K&O#sfpva4^D
zH2)5Wa^En(*=LJT!el6K5g@l&N|UQYc0|!4U>Y~+NKHFR#3PNwYJdk6(PFVd;H@^C
zR`1pj@26~tm6L1#bE1$gK|Gu>Wbno)<paUniB~PLnxiVZ_MIabel`?q^e|^T;jlv`
z4TGKcBs?Ux6R^HKM8kr$1S-UAD2sSZ$YTVW>E!2$0#>a57Ufbc@qC`}WU+=H$Iv@a
zeYl8605;R!f&cYw6=m``k?~1I4XO@TZsJ%$fb~NJltmB2KGA?nKs=nkWlK6F5w%tL
zgl~l5HRZ8g(-o^41rZW^Y?>-DJpTsYw3D2uxQ}E!;D}Z(q6RIQI-5|3a7g@4iNypP
z?DB{k@NSL0B!=i`qeN8NfbhK@Y%;fcE8<N37)Z3}HWsNK#94V_%ASmC#RkhbgywK#
zvp6r(4-i)k;v3-7cP`u;avB8B!8sF#HZB*Ix4ntE3Pdy3Q(#T<4dc*#4yVDp@**xF
zSoI*aD+#A$u91(0SccI#x+qo<dnz3%5P9$h-FxV`Ii62XQjC#gKngRoJ?}C9AXkRr
zpM#Y=(!7y~I$3V%whq}~-xu*S;3`zy#Wj*Lhq{F}u|8QOYIiQU?L$*GWRk+Aq`E4@
z32Ejo))&OcP$j(`A+0G+eCfksF@D<6edWoH9a8L=l#|MthW#dXa`Hj5q=A6e4uuyV
z63w8pGbS@cM@*(7bkb1XuSK@FM?NI*Zerph0F;|J-IL+Xz9B=f&aFTNW1IpEk1MW(
zO3n2u5<Bl~%+(M0He)A!WX-o3hO3BLq9-_K?;vBdkyyu{etbxP7?HmtrXRQ#71?vE
z+<x37gZ*``1l+AVX_21pxN@rc&K^~_^j&4<Yu^`9OMHSnOiHLR%JB{Z84Y@f=>kZj
z4wzR-1?(uN`Ro`zYQe%(q_J%GE3m{KdlUbUB=2UZ9{7(9y~a>l8JyAGny0ZmSjJWf
z(wxobhFE;)S~yiJqwO@WYI{ZM3)_-ivCl};!fhWe>8gy$LdnLrNRa19c!n&Cktml`
zJO*R27I_1i*l=w19@a>LfzyCpsEFm6<^~LB(q}kDEH1okHiMv%{$mg+uQrwqfgbzF
zN=l^V>hXGYkfk2{e5{LroPcV2<@^Pd*L}CvohfpfCE2gvg%h^(hK^k4-`Jm&%qq9d
z!y2}PEr!^aUEA<<%MnxXl~1~n8pMgplR+!7OE`y}aTkgX+E#zN?<fX1=5j8ZAcZQD
z@2&qB++ExHxsG<tyPnuVom0lvsakzb9`4q%13bjaXetw2W`ntzBi1BW7E76u3-~jM
z(<W-oR4H|Xc<jg8;iywmv2rEU?W((Ss?EaXcdO3M#<lbmvha#D-8T=_bGtU&QOvH7
zhCqMgOUs<I<P`)4@GKWihbj6CzaP25<o*Bx<{G3iY!8MuB2hCY@bC%m-BJ*^;wC$i
zv#_o5K6gfR8~)&P!~nMI*swt0MU0c|khSyQ!w+7T;rjyUL4yVeMUvsTL8_^zhb>+J
zZzTpx#73CYrc7Xwl+Je4Fn430I3;{80H;V~z^->1%%+0;cY)H$BQs$`E<3<h3(KTG
z*PS!rgjCFx8RC!$63+Q18#jD9D^eaHg<7vFxQ-!dn;};X&TXQ;oFmu6e*2TVh=>9i
z_#6W1B+b_x6`2cISJd#?jvb<iZ}j<}2*H;Rr`X3me|~pQO8$YE(~s5?@2e@csdUVu
z9Uv7h=<A3?;LTEA$Q|zb&1;9zoH${%A$`dDk#=||#X72Y{ehItQX!!xrBvbQ7Gr6*
z;VA2<&f{+BF2%^@Jj7{mXAW|>G<G`$eILOPeHQd3K4s~tvszk=D&oSm>Jwi6V;Hx(
z<bw}f4Z_qrEA0F1BokU`Fb6lLx03jafwY3l>;(uvixOX?g^O40Y!c|VL22ADLzE7<
zk;Gbh$D(}2b~7LP;vS|0UV?MWwHgISYD(&h@*xu1FmdB?I{jry9V!}P2_yVy4nyTb
z<2cOeG<tnj!FrH|_1-SDbVqbB;PU5V*DdMKzEF5v?h<Sde`5cNm>zPZw?aIam3-oQ
zE0S=)CKaBfJ@QwS6or)D_+oQ^MO?!2Pld2b`u68A9dU>vHxrg~9QE5{m*L_Q-hZKV
zpDp$=dN`DSXV&<wiavr$hs>4--LjoBr5>}!Q>Y-A>_|5s(1IfMg>MeB=H1n&+*%#l
zeXKx7=rpHbPsbN&VyEdE9buHvKA<EXG-*cW7mv-`-yx3{p&eQAUPwsDQ?8LH(prF#
z(AqnOivCc4VjAX^h~&@hrAR496b&4y`@GCXX25lShq^P&Ziehh%^WXTy5sSEt2|QQ
ztrTgg`r*#8^WmSixjbriJ3DPJB)4sj-VbkEV_fa1%sFBgl}hZ*tPq=or!7R01f1Y^
z!|&51?1eX;L~rrLstg-1+Ru#V2QV<LaPk<uFqjDxqe`X7r)zsSBg)3t&l+cR|HlO}
zypIdQOYk1Uh=M#L7bT!*r!Hx%C3c~+4V1a_pC_7etg~l_OGgUY?qnnFM`6~fK_j7t
zG-zy6ZZ4d5pDv{mu!ukJ`&v|s)u^aQ*TG4=O2H5@aCz3eU<oI~c~!FQ*iGyK2*JOi
z%V`G1$dAapYu~Yldu2$>Z-9B}O4Y<E*Du$X9QYuKXJVCvo;^y>N(Wj5)FCg=p5+ZE
z=1R6@sJm*$1Bb<uyN2%THDeg04^fpXGlkQ2DIZ9o$14-SqdWca;emP2Q4Fs?;89e5
z=Q(eP^@pD=HkF2)f~cN(PvG(>&3DMyoaZN(hk_)eVsTL96nWU{qBKXMI!_|gcrwn*
zb(Bzs-*t1ZT|r_@I9a`wpoXNzdf}Ug@BWS0ZGj~vgQ)!!B1Kl%<E7z&I$$A!+n9gY
z$P)A^t9p&gC%{}KB_nwt$$8PkpR;G7BW?CrTh239CRq6nCxi4MKKopI<$u<M&mRN{
zCte+!8Hw6?!nn2@2>L^g7D|(QY)C*PiLeiGUq$c|9yjlLU+FfU=Hx$tvu(+V4s3`%
zAxpXS=_?WwYCZ?veW?L@Ze*2$xEMAOAFiKsHpHrm3KDFsxCLS;>4w$8O=liD_=wm&
zL1s$Ow-F4=rU_L%y2=H3bWs`n9I&K_!ww+a82^Vf_o5Qx<8e{G2@_+)m8t<fXJRv%
zeKCu6+$U$52>6lzoei514o*q|jN`aTd11imUn;7y6C1JyALj8Xq*2>=vj9vy;^@!8
zr>%3`?Vx?<1D-&Ra_t95M8;DOc=!+|VVn^0M?MZ`Ji|9IA)RRTv(uSLfu%8lP}uyX
za3Ao5itI|zqK#VI2#LbN9`9--F;1-C(rNHUJbn&0tt5L6T&d4J+2`U_Ot8Ec4_a+q
zo-0T(=P=7G%AUivf&awFMUZ^1P_sa-zTu*_#!nLy!lnU92j3Qit}$OntizmJ<3}8(
zOh9NJ^JT}~+0Cnr6y*Hjkp1`PtNDs&Wu*J`d1lm2P!L|3_z8AxOa*50_lAZ-+h&a_
zei}{xt4R2C1)r<w86=8ALk?b1#&)eOdzNCfay&8Qx<Vx}w59O)0c}N}R^f+3NhtMV
ziR+$IRw2%YpoKUJIj1ZR)Zz8*XHXLl%*<iR5NbOEyR5D;&CjsPJ4dDuAN286Ga?!^
zGtpCICyopQ(cn!Pc?#RZK+90Vs3}~&)xKP)Az(JGo>((Bln9Aztj?x%KNL&xlW&Pv
zvF()g@37_#k#S4^XdTrk`A{X{@2Rsn)uYS&ILgPIpT~E}f-&{jicak1q9*u667bqW
zF$6Z$DD1!TF@{moTO+2Fd@96rg7?G&p76Qvmj%b^=!B@KT=>oDq%;SoS@o&D0eo(|
zpS^rrfi5CVa^qF329b~>E#OiMo>$E{mNAI&W<W!`kV-C11s!O0D(uAHBi;4+{KS5P
z7nqRGtG*frUUQ?PZG3+oMQZvSK`(gdle8-{6lDJQh)JOmRh$apk`kahcr<~CtB-T>
z^W~81K2(VhZFwFgzg5<u1^(btUPhm5qjLdbO3B*S+-^Amw<afDsc)WJ@c2VYG#kgU
zW*n{;of<r{j}!^IP2+anJ3TQJZdV2><2S%ayEIRaX`pSb7eG69!V^<rlF8iEQ-?PW
zN3A?xJXLK{>LbG<EVWxi^GS#7o}0j<2$^DM&S@(1J<rDvJUX|w>VL5->O5-=$r2j?
zx#@GLq^^Dk|8=D=AMdx@tU5&@TSJ9R)iBfEWKKcWB;`zhDl;g_*{BiP&lgUsg6Xm&
zXs^av)cU@7NG)S;{lqT@Tj@s6S6@O)_7h5MgQHGp{Z9w)-;7L{@068?&PN|``y#LS
zPb1Ei5)<zTO(=bCOd{kUuO61!t)O{mbU5~|If1P(zK9vvha@+MB=<K6dmx&Le;lV{
zjj4Dt%K+@^_=zzzn+Asmv8ccaa7VE@TrS#r`tWchxP}5KE7?HaZ4w}a_ja4`nG1-E
znL7od&*A4|J7T)Z+MMdJ^<t=?nhtdaA!;aWKZnSkK3hapW6n9t6p=}IElAVi>@)IV
zW)^XzzSgIco1Zx20P58@-2A&#U_4{v#E+dco46hpS6!g|c<szD-0q9}!jI!9<UarG
z@K?kYLG{>vPq4z1crNx|I4CW;^%V9>r`aHZbS8z$BP9vjqSTsxumec!&i|m)H$}Eo
zNaCtSEAjXP@tVX)Cak7)vv((AkuM?ou+sw2-c6vr3l_C80{phkoU%L68_zm%U+WuC
zwD0zVSj3KYE1O6C7d~vYtvb5oJvjbPF#)(j+pC>>(D%?cPgFM|%eWPV(GgaN#VFC!
zFa<gi?Z?ro5jle6<#WDo);PX9bE9?Cn#K`QX~xEP%kJ^Q5&J>_iDJV&oBBZrZ=&BV
zY$0A#?fNuqqO=l3)32rghJ24x3sD8FTsq+Tzu0>ZsH)HHTa+|SG-`~Z(!?mDU<Cvb
z5bOv7QUsJ5DI!$?X-d%~Dk4pa2vUqxsi8{|j7mok=}m+UB0Yfg_U77{bI$*a_wKmw
zj`4rvje8H{jFF?lX7BI!ed}9mt~uw*TGwwV?(8J>Ey%+^!4J6oDK^Jw&|0tJ`R(<3
zR9NQXFT|hew5?jXdiBDD%ffmZT=vdWcjYK9tf9j7`I7|=sCg9_NE$nWT|V;PLb>uv
zNt(-ld^l}&<TKA5?cxz@b#R*E_32%tIrR)G_zc`0ATYVlay{P7u!VrkT#?JHv;?|r
zY|B$Hes(R?NA^1TmwY~A?dk-mC|XUuBg(eNb{OD>NgI|8<lc2VE8!Co?s2+h985~5
zs${nS&?4|mb`Te6Nkbd%eULgX-g_()h5V3hhGWDf--x5Rfm`3o<0nnMEEIa;&qSaf
zCFCN!Blc^!S>3^C%B|hQc!y3pUb?$SXW*msYGJl=!GH^;PwoS7=vHqo9R%j-PIv=&
zu>r!C6ySfvP5YC|>baGHgKK>iqPQJ!BdkRZT9)mFj_2@FFCGPGUF>o12sx_}#tpF8
zU;eP|u~$&>geJRy4@{`TVyg45^909!5t)B{)_|;pSNN=h6st#S&WwipQ@*p99z*Mu
z>dA&lnT7MHfY2p=tzeEVKDA6}TjR$4L41GKzDaKM^xsD;+Y9{*7hz$ga@qrCP}xcX
zsy_5_qdPxfdP==L3^_+_TdacGnwpy2y)s+cF&ua2wAU?gp4|zNM{f2bm3^07W|}Pl
z4M-D4iObnWVTMxS?+_)t%j_KV2Mbex<E1&_nsB@DI%UB(D=LZ?-ecA8^^L4S6)wbH
z2pCq7rzC<Cb4`<RT5`Gblc~QIhN+6EyYO9$Sp^CK^Ogo4%K8js?{V*n&f3+Dl^9uk
z>TY=IJb3EIn3(Bvl-PYvo$X6Ow&ov4_PF_YjXh-y_DgT!Y0!vh>n3%eqqwdxC(G>Q
z#i3?6v5p81hTB~bMNAO%)I};q1sU_>A+>o7bb)*N-bbMMm3)cH*dXU-hZg29)maj3
zfY6=Gl^OhDVyV>R;qNy}i1s~&Uatu9l)iYcLuNEnHu)VlM_+I6V}Mdj?2!)5)85!P
zJ&dJY(PYP$#E_j_`GgliOQBg^MP&`LK<PRs%Cis5iB|o97<*0&ND$kEIQHpezmb%t
zt<)xYz%>I~?%LTyZ}X33O3?j@LPUw*W+@E?Yt5BeRjx0(P)<F(Z}r+sjUUl;(-E}D
z6_*gffk2-^0fM@5FQjh%Q5+HQYT^9LmEt@5B<VR&O4>9}Z~&Anjqe*aZwAzUzxt_}
z-SFhcnetC3Gk{|hVWMUSikrJAcnVOA+sYQJzh!)xdeGJyDG-%FAI2VZ3FhMtHFHEY
zTkbGRa2Ki#X&p~(xyW44vN`dHAVTTbW~{RwD(w!FPBbw!WdK+&oBp<?FDYM=hMj&A
z4>c45t~%MPo*!gcPL(XS(?6=n!T2|{KRVq)xOW9)`E$&z&#LNGvcTTp#op*woL{h7
z$`bS&)&9i)bl0ayJ4utcQ3D-!i?!T{>3;5WfSD_^z2HwEjXSexo~G_HxV{(c!O$nV
zei<a@q=C@Y<Cjnu;es5UcgY0Jo^lM^Sv+gv(iPj6S2}(WVw))|!NFtzojl#_f`ur=
zca7TSS_QVDs(eMbh<o=1C&`7W$|6d>Ulr|%LW<Hg&!sG%a&YrB@{E!OT@@8y;i}L$
z9Na=c=nP!w3gW!}vIV&3HxQ;8zEN7^B&7~5D6Fag3%WB1M4^k(Z7nNq4pfb;4swh8
zf>lULW4HB={nMAIW(r0dLJl!xaub-Hht8b^*x3S?+jB)u@cs3~>`flYsgl#{eu^zD
z8UL6J`)a|w`LU$v0d5hqvxF8_38Z8}t*nRMLjt5-+IAikNf_hQM}yr+-a;F}h@z6s
zUdL~9TAe&T_wk2&^A>ymzEn5HaelBXVu*lWqkL(Bx;TiaayfGd+-yKLPOd#FKt<zk
zL*6E&8ZzkTMu>LATT^IWAaDq-mVTn1NQVoM2I-^2Xk0{BZUr2=C`5NKx>NwpI`(O{
zx#2pb8(7e@6#{=c+sEC|qkLpl>@ly&%-Kp<cSvchq?I{OE2D*#k;Tfy5c%SfnHGS(
zWT<cA8J-d&#Sf%hoF$+O8}?6f+)6oogL!5vVbgfL)Xj50y6WIQUWMSZKH^3A_ATdz
z$}NG77A855#cjaSQCS5IKP&A;2Gn^TpqTF?QNL|7ohSMIt$y1Jzgi`+H`HZ*N)$B`
z5d^4FLGzn~EeqKMkfplY-MHgQomnuQfrppc?++u$A8c+_y!=eR@7U^uE^?#t0E*N9
zX&Ot!KyTo6UvlYrXYl0xh6R47fX|Rx7?rAi_eC^<I~i-8fU<Q>k@ELAmvnP$NMPxQ
z(NWpOct8&qwQ5)a;j;vp0n6ARXmFC|>v}>6v(xf>n8djuS{4AVdFy!&<rXJSl_hq@
zR!JQu*nkEI%^4@4k!Yj#@Ev2s#wC{8R}B0`&$MsLMO^bi{8?2UAA~!leRrrir*VEz
zv$$e}$MQQT9?KCtmVbkjze~eaLo@1B{1ywy4Q2516ta-A?QckH4!!bL=9{}SNUTq-
zwNnQ7$2JxLK%C1CKfso$3A7jGv(gh7ZANKim<2YW8>!L;cN-O2)}E)xly?*xrPF2B
zY$z_x-elY6G~sS3Is4a3<IG)Mvl~&)KLOGe**cCvi1sf1C|m?L0%RC0F-{b8+Kt_A
z^WwX0YiDO-{OT<axZF0#4rR|V_t77`NJHeC24L9hW1E9f+y)Z1%IJ{2-Y^07lUj;?
zw^c${8BjUxzf9Gt_sQ?56aj+pcs9Yfm*!CWxYQ9mJTkt}kTWFur1YB0dtEWt>&w)3
zDxEpAZ=Kv)4=%-IE&uKJwUQrRJX3RxRWo<>aSu0vpkw{}mk8+P=q+{UTFxGM_Qr<8
zJ=$XZQ*-TJ9%U8f6@%sd<%4-`N%grp<$RS66+)GR&J{1MKlr_ExOnV!QvKZniTC86
zS*w~b!m~W%YHki!-4YSesO|RQxyj8fEE1&gQ{joxVQWq$yCUhhiR8gMe{%t17fAxa
z&J`mC4KZ*WQlGmDz&=TB-dWM;;D?7PUtITr(6fP3KD1Es8&pe5fF_>nX;vO-hdYZq
zwf}~?)QSjq1nLo&HjCza$<sDm6dYgyd0^;XA)qFG=+rH-ucDpgwzIOvIqh${7(NNT
zDm%29c~>4tI$foHP1g|6{vXjYlF@tzoaOBK8D+KdFvj$(ctlMLa+6^k9o8K7F^rji
z*#T<84qIWgKx9LctIA~!wO&qM91s%Gm~r<I;VhJ1Ue>UkWuF8ufVr*R(6oC}j-TWu
zDz{a}=a>5~?;N@d&_D3b9D#f`P81SdeN<TvJ5sYZ+P}YH4})8k#2i#_3^?Q@DP<vV
zyG#A<tTPGkcAS_Ifzugf<_$n`t1Ya7@)qGz*|8IM_Nbubts!dgfuI=KG{$#JZ=SQh
z2pJenfTS{oNB|z=3-saXTCz^)y^X~!f<x}((_xZq+(HO~&-2+h<P**m3cGu)yPaIs
zu~(#N?*{H&-|L67yty1m;A;fxsC~5S2~bpV?mM}g_(pVrQ{v{Z%g&q{a)dcUXFE;K
zicjLnkCys~5J1XTjAeJ-*X#h^nK155O*$X~+Xkko_k=ou6|J0DHRh+L%!A_QfS52(
z!NoSR4uiNY&~^^aQcO@fvTOPJ-R`=x9qWk^;x|JoG4i20jyDji9o<AvP<f6^nHvwl
zs0!Gy$wJXfdVo~wA&H1=<*lzwOuUM_STD?J{7&2KhyC9ycsYY5GF{kk@%Yrn`n{4G
z6B{^ot<(=+aeKIGR#0Ra%Zi6GR8}E4oclD?CWY;mLo1;_{i#QIw<)0vKoE`-_DduV
z>dgYg968%uL%m9B+E6A(xVieZzl=f6vP3v|g%%$d9rjZpsGIIKAO-D!Sg5#H%7*y!
z>$R^a1p`MkUg}d%_O&GS0!(ztKs5lm$6d&4PkR-XDVt{iTpZ+}-BO*nfXbE<u~Rwo
zv<J~|h|q3ByhVd_UUA?m>rnw0&r`#8C!NMnY!p%s1`H*}su@^qkqb(@0GF~71j&cC
z%|XsD<+Rg)+f(I4WKg7aAMC`DORRMFSZ6`IztzAjO>}YhTkD&xWjPyeoV)}0tUnIn
z2`ULtGP#3!wu9ObSe}D=uz>OaLLfo@!^9w=RyY74eGoB+@%OL52KX@yHqMTyK<MzP
z9{5@3BsECE@|4VIu_~jM3!#?DK$uP-E!X>Xg@`LaTX;*4>Jip~DjZ9cyPCo(it4CM
zV>JKvQ(yfaj0wLliW#m9unF8g1D1~g>k=gU0AQWO+^1^7_8<%sO4)s_)Q>M2@=a#H
zm9X`G<WtFw7akr*)X~SSK}}a?w-ao|VI*7YI%j)Rn=ZDk5NX&%1Q^&-=V?I_c|6#v
zlrI^F)4hxGcVfRZ(Cv~a1t__L0A6I1buppo2@j--BFDue@`8}Jd;%awS;a_L5<%1;
z_2^;E1dQKCP^i$y@1$<Dt82r>aF559j87Bd>JuEnvP$rnh}X&zBvKZtdk;#-X=SiW
zhXL>iLG88{?h?+9s6LZ1J6unFJu)C20TGQ2h0=Yz`r!roW@+cK{R5MMxn#3%4fQ|F
zwHKY^2E^OD+CX=&FC%CLfRR?y7=cze4EgI!zh;T-LN4lx!yDgF8#95dY;AzfRpaD1
zT;%)q7D|6{@gmxw>+$5SCC44^VZT4m45gdGku3`_rlG5kyl=uJmMZ4Bdwt1r#z2|6
z%{9-lvK!G^lF4yKcpD}gLp=}t0}G4E?O`RB>!or;ReW^V6elkiLM}@bLMNa|*XE@|
zF%0U4G(sstCL$NL%STk@>4&f93P~<w@s^UmtBp6vvONlMq2TaRuC+_Kt|*oV9}t>H
zZMTPbDU;X{9Z*DjK!qud$M6`JOcm68)GZ0V{NlT+G~?-0y+kZQKs6cONw3ZlTRFL%
z<<R^Cnl#F?&~}2ud@W+3Z7?E|@a}N98Ie@(0bsq2rc9!x;#sT^i<NJywKRU<I{#Lg
zAHkhEFM%n^1&K^X@}UGlyy1OU@`xkMGV&e;ktbs5L9#j4Q)kk+AE~i%f5Wnu`3+Wr
zmL0E1T&D-Dijr16b*ALzQQZc7&*XkznrwgZ4LYk_#CRbXEg8w%D{2Qdo2AOfXTG~i
z8X=<BfqX6oyoeK*hb!$TG_u%{BkEB4hFdM<q)$XcG_4|wqB(txk)jAQ{DD0eS(Zp1
zkRy8-GvC!l6@rhZVd%J)Ws?;9`W)uNroY3NO(25nKfewveu&S1v1uvYssH%ue9nCN
z&wswoxwy!m|LFY{^OODk%~n4%chToR9$LxVEPwxF@s0odQmn=Qxy$>AZmX#gy7sDY
zW9g}5mkEji5MjUpO*Us9AopEdGl>-56on8#wEKY_7bGsJ-3zfQ1#xCX><u(aTw#E4
znFs~Mwhoww?Xne)370F4dIyhq6-vt*YO(d*X>yI_9JF)W>c3>Md4Vp1oAx)c#%Q<t
z34g&-ob&g~z|i(ea@--Vx~C1yL-TpV#+3bM`_fYUe;6Io--B!6fxE#!A6zcoetxC9
zq?BG3wh&YgR;8P^@O7~v)CF;$X3|kk3ILhYQ(v)t(^Pi5Pr{QV19ld+q~Co<S@OD@
zSlD+vJU#Dw8-L5e{y3@tqRAQ~>e!(95c76KUZjx+$bc^E?<e-HrL6|9c0ml~swGMs
z!7P9i*VB!Ts%2XpQsD;NO+=*}_6?im@f7QgX~4#^miz#3wGA*1@&fOT5*a->N>Ctg
zKtDkra(P$E&<XXT)^2D;+U2zK&TF%|l2A)h6;F+F$40Q4nP48kpNJo#Yu?MFZG==Y
zYXj69aP-^?ek~i(jEMnFcN?+RP@74R7@B_3U2W}08}B9X@dhqoStecaRqv1WOU8T!
zPCt6kdEQMX<HnaKAFU)7!jG%jbb+<)G2trm;86cv61n<r6Q}KP1hP7Ha~@LybM@5k
zXjtL0iJA=|;|hb8A+)c`C&iPX)hPj*ks5mxE>_j|!}rwfF*GzRA$A#Y7-Vqi$IUip
zjmH6QEdfVQcYu)yf;5pweDSu8Y^eO(mNh@!=L^h(^%1L8_=?f+jOJlL;_~OHN2c;T
zH9iY?zy*l~p$gIP230v@kY-A#Q6Dc=ghYx5*Q<pfvOqockgF&2EFPEyCHNl8i@IPC
z6pMLn`K7k@?KjuEX}@05PAF!q?z)-Ly|-j}1$AC-?N)mDiOu9i`r+E%-t-#+r)BR*
zv|KKEStHUIWzyrkd&_1Lj3viO8()jJWU2Xrh$%Ov&tV@VH>m!?pKvyLGsxSD##2bu
zIViJGFx&${m|^2Svo*YOu6WmlD4N+YA2eoso|>fSClmlo@*a~Us{Y@=;{zch1+|%-
z=TdkQQIs-qP}s@E+a6pl!|_4dJd<WL23t*mrl&HXuhQ6C8^@%p(u)BLK9?{DvlmVj
zLgo@ktG|gqKUQ+T>3VjwccU0Q0fhsX#F!dL^lm=jr5s-A?PmJxbhM<W!-?WwSIo^m
zd5E2GIOE0(L8L3oI$@3J1>8CN?MStP_%4D`t*0O@Cdd44t(CKHpw8(4=IOcroqc$z
zIk1u+sWn9Tp0F4VO<<<<fShVCE+|LhIf@ND#oH<Xg9U2@x_7MbBLzdan%+STB7(kx
zl<HYZOsEFdLoG%X0Fgv#Z&O0`^!+c(sj%4IW(%qZlbMTtr%BHAp@YgF&A5KjRoNi8
zg2jNhTgmMc+p<5?5BKc-@SWd}4iB<rGj4oBl)nBY3gEj?B}Z!6p^y5ZYXcsID>#;7
zdkp_v<?rub;4)Y5qMFqTFfOF_xEe3jJhPZuR{q3YIby}=mS$V`lM6Wj_Ab3P@@Eew
zDigS{I`erT`&*(jSO^;x2AM37oQ#ej6;o7LtrYwCh-$qbP}!a%NnY0se%*193S~!{
z^{>BG=b(oZbs3@Qq<lsAnI56FcrtH)E?={qb!4wih}l<9sNDZ%ujAxjeqOJd0BRDc
zy{y@idh*QfVdhAT6iC6*!~rK*o1UB)^`D8zhq8;Gz%bfOL6!>uH1+^3c8S!Mf(IhK
zF>l*BqQuiSrQ#2M^-TEy)#WYYF!FFGvTV+pxZ_QGrIxjNntmE=6-VWxVmt+9Ve6dk
z(N&U@KNFzZHH3x^J5jr-9uI=&p=BFrS$nDWLTRsavx^RfTG+Pdt7_FUJiU`-PDg!l
zeb#q$yW#ZrJx@hqn%y0~j6Ak6YHhKIw+m9tLtl^vfiHk$Z<T$PK0J32O(c>GmF>tK
z40ZWad$2so!$RQa=hsC6{psJ`mEj5=F0%h!dT|LEZoUd)e572YGE?aO1ddOHwjGhD
zadcD?#i`&UjY5cR6CZ#CHu!nlP(y*b%V}LSTdpz0;D+(b*_hY1TKXY2b$7sEQq|B=
zQ(H%$+M_b^cXXtc2#99AzXCr<3eW_rK|}#jK+$Dwl^`AOe)j2YZq4e8q9!{urHkv&
z@Ki&^R^s3%=yZ$Q%3M)u!|2zt9ujiN#UhGlc@2-=p&5}3w8%VQoVJy~Yecj=0Q7eg
zgig?`^=>H{AA69a!T}d0;~*4L)Ik9PRZx+X=K@M4PXL#l?K^ujs6L>^9l6mwhJ6mR
zhCs1QXtynj1<`@D>JB0{Bj|Qr2|nUjM{PuuR|jc*DP1B}2dxd2B?vUf2}Z!(QvT%K
z>YD@7GCr^9f?Kb3`EWn#yJQ#+P;E>_8MA&NRFFu-Q=H4r<JS%YgNf3zD;Fs0oU^Am
z2<=4Ivz9CGa7jHyuZfANBlSCr!kk8`kxr=r&yN8@lp8|a0mvUTy&07#2nnce0C8F$
zr_U48n?m@EY$P`tY62vK62b1z?1f{}4>wWP&A>oT{&NY(G2BG@m@C12=uAlgth@9u
z8p1&`e6|rs0Xbu+E7(M-rRER!8UIK%b)h+K`v9vcg7_qwE)7^!%dgA>A%$K?F3W-<
z7HB{2V^3D#keI}A)!Tv|6{RI<Q0#ruhbK;O@*VP2&Gv~_DxQh8n_gxNbesB;w14vY
zNkIiYQCDPW<gvWOAjgnEvk(n)qrPvaH1g*62QHBf6R7F%9ggMgj#O*24s}=0W=j&&
z;`q4V6z7K9-9ar(LQB};smaJ?laW8PROC`ttUdds&2t1LU)SPEOr&vC>J!CVTsGE&
zGTDCyz|Ah!@jw8PDFBo@fl(!nIAT^@xdG_7!eZ3Z@+I~VTiGR?bt~*OBOCXHNw0<N
zZPZbE9sA6OM+Utiwe|;4K*y!YgM1ffb~jy@_}g_hCkbemp=+adSft_SUpqmJI}K&!
zy^P$Z$#`EL8RGv&pnoDmvx#G8s8^5VA-dGcG@N_xO&cgn2E<wH<s_i06URPYZR#e0
zosscmvY)z}_@nV^`9t^}$`~n=5)V)&suc-}!@2q%JiwD=Zf;T=hu=|xQ|+%hMSGyh
z3doe&yWKeV3WfT%J#V*8bxSP-jyaH_O9?Vb|FHuoHLjL`*A=i+H&3EZ-!B$kc@gl`
zGp@9O&mrFghlKs*!mtv#3bduuX6SbMMzrk3+gnpg)wWvA*df*PD>#KN$la*id<|`<
z^2uu-fOBrQW%VVA9Bc1BY>MDg@tz1@em~?C`DxAfuV(lqnV!h2(6Vb3d**1+$8#^K
z-npdyX4M-(c^{q<*nq{&y6Yl&(Nj-7{1l-eWjb)jc=!45$7bdXa%+063QwuK9N6=0
zI%w1&cStufcvv=}lIM=I%KB+}n=Fx&bNsgJc71qY<l|?%(eJkDiH(Wb^COq9j{KFc
zYbD=f>e~1L$e#SQJk`0^(SKemk1*@~?Uo?lHDFXFIC}wO+{3u<a+-=J!6@+WM-_IC
zn-Srw5Gy!am$^J9y?JM|dX%>kN7xu*g~^#AF58Bm0s-i!H%0s~chQ5i4%plBu{JAV
z)_;BkxrNbAyMep+h+Q5xU2_a=f0epX?7@m`ce(YadtMAeE247WzpsF7*nB`5Og%BT
zW3vyXc+}f(x|<^$n|*y6*I3AyKxTv9UH(dQuu@CDS+AwY80|$9dg`XOHYR99xQA(G
ze{o-Kt{n}%*Y@S~R*WYG0Qvo-kopLjc1;?WN+V*F%6%bRAzy|b9`l=y=m%-$@ewE7
z_0J4YD<1GafL7EpL)tC_8Ihs8y83#3lZ5KEGb(m+H;T!0_ND8N*qq;&Cg;mzqu+7-
zXq7jwA;(ewVXLWlkKaojR^Kdf5Y*{a=ev-r+#@9OxUDAASw%Z>g>t@I;PJZ$KLqkb
zhAi!B@AbVMmzXaR94MhRytUph<JJRnxl-PxUcxIM84y}ZA@U;jw-)vRxf4(qfw)N+
zJ<d%b1v9K79A^*^_me)wgp2WF`5t%b5nz-QXLqo*0>MMnBA`M6{x4FykyR}#A~2b*
zyi46@dkKsrdf?Oy(C%YMQ92Mjy{K~tmLSdB6aa`tKR_TjxmBBRN52WPK?|2C@E(Yi
zLd4~Un#YTTK$Hg}(Y}lFA;qz|QgUV|DGf|&^A+cqBBNltcMjAsaCN<itAU1#W1=fD
z3{OCVy_ujeqP*Xf9xo3P-a*3Y2hgacSA;848Rl0ufx<r54}*)tmU$i1WF}v*zej3k
zijf2bL;xn28;>&5Vrtpcs#=16S@YDLn2>Zq$7hCN`kodjP{#G!WPV(j3R_^q8#40f
z!(g$G2s>ZN&Q~4nAr&Vnq0;&wN@_qXTcMmd1Nz<D!Au;)3lf6}Gy%5SQC6q!iIjV3
zH3|4%$v4$SR4}Pe`28yGCjQ(u<UOS2Jm8-N4D<wvWZGN+8dJkKxO6A$h|Pr&n_7du
z@?{|jTRQv9itAN<jUSC~IJ!%cZI8H>&V@0)?h^Z++I08srbi~z^0itVe!TjjwOdvh
z@vbtG-dG{GG4W&~pLWWMnnOz6za72LUb^3~Xdz#GPxTfyc#r?NteH}%q;UagL_#GH
z>WnAw2GxC#DXBKh<7}VGar$Y<L|cDgbazd-AyqQ%iGSw9WL$Mn2yZe1^Ak$Y3%PIm
z5aAmO9?A)+NXc*Ct%;2vMYq5g`9SWGslbYKk?AV1A9UJR3qG{SHukrReTNK=b)&fz
z($mP+jEq5(*H>=82V2R<=0M=!#b=-$`yzi3?aC%Psq(q3ma40uGS^HdR-Q?Mia}x^
zS+KQM!&e=ttS>mXQ4)0&nll+BJUQo*Sdc;27dmvOH=^kU@z#_~lXa7Uue=ZCz~%?w
z6e5>11lp^A@=COUUN?lIo+_U(?)JySa!l9frw^%ezZN_W{-MOAU6J$hjYNRwuj$6_
zdVMO3NIy5Nh<57N>`!qzPCBn5aWKWkWU938N$>2-0LPHlQ+bK;zZ(Vd@XcL%X?l|V
z?S4b)4M*jr3x#e4xdp3F25)23Cb+9FbK;fx^f0^bQfx=lqfbdI-2Qp=@7+#K!qE?s
z*jeu<)>ydT(05In--^fxWJkl0ED=2lh&k4fNlPD@0G^OE&Skd0^@<9q@&=}VT`9KR
z8C1@Id|fm+ZcdCf;_#I%Jj0=Th5q;1E=fucTM*5ug<|AHi5{9VmJ(w%D#<AER2^nP
z!+M3L9GR4Nm4W0Hy3=&wnd3w_I##a`hBtT2<eY;=@>B!DwoqL=NcQfhc4tM*&Vf~+
z2S#^e=#j&Y5SSU)lMd!mQ5L9Fk3qrrzpo2z&@ejYU0qCR5>;W^`6av$Um+>t0IArV
zZd%^cp_~-l<5Mfdj3=@n%pk3{aANg=?%*}ZG4A336;N@4%ve5WfP{9qj{WbSekihw
znXb<ulT~20i;%p~n8;z&as`+iQthn&0(Zz7q#CjFB{jcL)d!c&GxJ0Szk>IwAU=>r
zm4wP*Dl*0*Z1i=SX357($EZfcuZ=wT0Z>jrvVv+t7)U0CW9jm!#(!!O2!T*XplgI;
z9o|`-_Qb7=;2yEQyOjD`#DUKaT%W|hoDtd__%hcP*Y}jvH{MH7iA$h@wG98?_|d+_
zGk(oV$Ao#tZKN0puhVZh`>v>8V&f%Hi94a%_jc;GRnfew%7i0NWZalVeJ*>F^bg2v
zR4wdCp1g34BpJ{dU2t+otQDld&Ywxe4gYV?h<<|hd#a>@_SnleS%`e=`bE=+&#6ZO
zAG3Vt5O($>6hE6-WR4v^)9s&snsr}wiJeps)KC&=iFXZEV*QCplM-8o*4VlE;t&Mc
zLUcMpYMpP6Pz$STgxY<)56yfs8B|uFWsxFVhGrNW>Qqe~PfJ+2M9*bG(BVY1RT6L%
zL_M@%rwSR#mRKTS1A7$Z=$h56)`upJh%E^x#>icmGGr=KU^{Z$?AcfyiKvUfF9@^^
zLjm^aZ3;G10o+?uEU2eykAG#>L}<!NBg`^OK<ZbMNx(%U+@*6CPup*?hp6y1QfrqI
zeWHpYI}NkViQSms*Sq>`?K#~5?$*VcU#|Q1ufLj}nh+0it8;#JA@<MgptCc#emP+M
zp);~Y4@rP;QXH>lw7MqM@b{BOPe<K&A#h8>QBcRWGxt-kN`>!muWq5W`J=YZhs>fr
z^=fB(d1rfn@yni)msaxHyv>Jd<w|e;Xhj;f^j6HJRcd##ncp4)DHa-8)%=;JkVB)h
zp1<8CAB(`O_s1-IV_%Cj$IrH*b?JF2Zz^h2(G!nKF%GF*Q(Ktd598K0O4rSssRG|-
zM@!FS7%sPSwOckbd$TxnIkR&({P(PKh-!3?M`ntw%PXcg)ZR<1G{XWJ@8LIzMa*ZN
z?q{cG7%4t)+JT60DrXR5DxO@KSYRjS80p1~C!Bz(#cBu8QJ{O$uTOAJ!U>Pu|NcQJ
z?2R)Q%5>*_Z2yWNnkAZ?(OMJKGn~5f2kt@(szX#)9@(jrE6SLTb>6=ZOmW7vqf@h-
zz{Qud#Ti`HCHu~*_Fs~W*@2YgOuJLw694+$M(9rCOf_`&m3Q{Vj;5Q1=$KzXt#oz8
zl$ewO`*u$Kn1XE#wxh3Aol22<9Ut4u$P$?hb@t^>Tp{}_j@7-}LcX`V+O_`6Gf=M8
zSFNM=Eq`uO1@{2hCj0xCu06|!st@`+mOKFAd+EY6OwOj-o=h)Fd#biSsVOwl_8rLw
z_7Xo{W+ye>e*83~EBaKvWY4;~c`bP;YN?k#{yH;#sZCw+r^%2Y4Wken>PpC;yN%%5
zd1f3<SPqPtIo>oDxZ68cV*xmlT8<<*&>qOhBD4YqkL-RtQdv$!L%OW9%%QsGU!f6i
zJ}{qK2qdFgC%1kph$jk!%{b4Gh7}<^DqE<i>L&JDzCjc#ewydqwq20}8LvULXODqN
zTejr+XBqLWw_~&KZPGN&bU|BjRM{fAq&~2uURO2CYjH`1yGrXvY$QweDQ+lKynDKE
zhCgnOHufoOY~{FBY?bMMPJDVeYiX|NSSS(4%tYqtTiQQ$w#U{ZV##ORD38tC`dy4L
zrT%Ko33E&NTek=)KdA`c;ymg4z;%dwsY)9cZds;C3uVYNS63oxkaO-`>|-fd7H=fk
zdq4IcwR!pTpX~EJ;!;Nd9Wm31lu&>Ki)j-A+(-35)t=_%X}GvR_0{cK&`CE@jT}|*
zi70hRQtB`ICGy(7(?9^0Tjk<T5Tdp?E;WanU4(8@CXxY00lWXho|UhF-sP4`2KO8>
zKnYBFYTEi^Pp<UU?Hb6&T^XUBIo??dyX}Zp+qV%i#@t)|Jb;eY+qWr?f);rUeSdm;
z{&lKjC-Y-Bo>H%!c04HfM*}?qlcFni63)B*dkvy)#p5xvz9XDyPgkOL#S#uP?*Oa2
z@_7|<d5L1WWpzEZwuF@-FFfLHyeZ%P_jO_JM&Apok8MRxSgRf&7;5^}T+|O*9Bb*m
z_~*uM^EK_2Y+Rx|D&MGhL~yRXgh$m?E?R6&I+ZochPr`gjOE#=qAu%z_RQngRgzC%
zj1XEx#2ko_nFuwmz4b&M%%APrqw1`0%ztw_9xTPR(wvsZ&_>%zj5lKF)5J~6X8HGo
zZhqT#FDnDfpT6fRm^Rw&s%mO#^V98&UzAJgLk5KX2U8Pdq)JZU7wbWE<bOX2?1;qd
z-dOekakwp|^`)?}K&gDR$p-BMdQFpBBVb;qa}veYM-wTj&N6O`%0t#NY!2P|Q565{
zs5?XNfJ7vPIF><{n?B+{vQx#sQyqiUv3gLww#_Y8az^=~Vm#pJ9|j&MMCBxFVs9i|
z-G9b(A+)kSd}aOoxVza>zf&QAUmUe<^6n>8#|_YB4b(2Mc%LHszc0QU+a9j^CDZb}
zi7R?H{+*DQz0C5toxiQ#Z)M|>k+KN0i$BL&7pd9Ihd-8V1fZyoYe)cCu4r#exbl(?
zxLdsf3yplDJcu)k<VyXwAhl1J$y9`$%|+rek1}F!ndD$m&2|rtIOHDDm>D)QT|4q(
zkWL8&^tU6#HW~(nXkGu)biV2>X0g_GW<ME<vncHZ<=zb?aS;*ni0;i~$<s~D#vO_5
zd?HdSZ@jfvY9B})?qgJaHQ7+I7%Bb8|3~u&+*4*nX8<Pq*lK6DRh_B95nMJjI{XyW
zT(wtEEg7Y4`%xXgwe01e$~HkIlVWqVptx@JNV;PPSKyM~{cK4lf;zPb8lAZ>yoX1U
z@+Jbmo#W-!;;31{3>k?*gj|Upba=YAGicV^e3Kn8(63ZNqy-Y@8UiT-B-xxp1*cP1
zmPuS^y2^>_n~Q+U1m2Mioa@>BnrMc^TRSZ=eC(dd^c=&1u)Sq=wKmwNlpr#DW$4>c
z35*Ig>9YDqn#vW<=*r=q`|I9$@MsNhim6Sg_KPpaB#sA6v@bR9P{JMHj#+P0cw`<!
zn-1H#YV9YDEo?zO)|63HZERC3zO;S&c)>M!3Q?BubcxVj06G<+6>6-Q9NV_TRO-ha
z%~q$V;t0yn=MUsfkDlR@KTbjI<^&O6o1}XTkt@E7{cFW4q@pU3(ziauo^4dls|a-r
z2|^@JPO<sTF=W?N_GYZ;Dz-h>1a+=j>0JFB0@e$Kn*Q4VaX)L8$o7LaHswN(MI973
zR)k6fGygYxFe_@%+x*d0<QYq%#BMkY$ml*aOOYuZaGreN)!PzswpH3=?J1lR#`UjO
zy23%f2*AER@(2>JqA|IpG*iZhhulfry&AVikU0@5ztXlaJc(iKQ>NC0{PeJYkK-QM
zanL6+%FC<NPO1sy$g+ilXu`VN75ZmWvez{Nf|n|0GiNn`7$1%)8UFXZHVAAcTy9`Q
z@L-b>N(oh+V}{wjw!rCkHHE|kR#4CIfOakk8IuZxO2WVui_qSW4AXD@Jb6VmInh86
z_oA{-18bzfjTao$j3sa<f=eY3VyPw@sp%@OSj~hb#>h*&2#%PY0|~2xuJ9-2co6w7
zNKqcqnvrODYASoWv(L?-?~|K9vl=g{_s9JwRcv3ojiCzsRbAIA)V+Q$;w~+^KNfvg
z!NyVd7)>mCqNKx9LE6ld*R14QH#Z*|r$zsO4_<n;8K{%GeVE1L`c>RY#1)G#<3=mF
zuK7LzTd&ZD;6$}%Mii#5GNbYbN(tkyqv1o5KHEl_GI?glUDv3t+|)LA5la3Y3wF3-
z+&2h3J~77s4SDNUCnDdfTz0h|r8gh@{>0hnXS{aR7P!!o;iZCx4`dqdGg9c(7}s-w
z;#!+{I?oIsdKKzwzdmZCPXa=ZReeWoMc}u0kRu9k*Y0p55NYww>sO%wLP=VeBy;C3
zAIB5}BgcK3aQ}#nF*ha{)Yn#SYwJ)G>h%3kke&(87CuJpftwQmR!u$f?-?M)XU&@n
z2c~K{Kh+)Gnn$ZHX1C%XU3aWP)Ff&2PTBzs#Myh;s{W0T&Tqj1GERh|+4deqI;|)?
z7bRG~Br^Jee&}kO9ji)TW>GpP*V#T)B5ZTl?5jp0J`Q#gg<P<)4CIkv`Z~C@VUH2#
zQQ2bVWsx*6d$OfEXE?P(J<%O@FGfm?p#-P>7oJE&eFR_|^Vd8IMSj_X&i0uGfSkio
z%AOk_Rch@=`>8=`f|K{*fXosNRyF*p|1T(#L<aP;fXC^+(H{ck6Gr2+n=f_E{Z%oL
zj&HIto|(b)20@)ek=dgeB3x+7@dYFS>7qb&7Ht53L&Eb2u%UvhgIYjP?IFKFAzX!@
z^h<*7Hmk5ZIfIHi@|8fFwB%|S?9I8!$ttZcv&kgRuMB1M%d~;BxEaCXGc2_sn4l@A
zhx{slvY3|IH2u|P(sXO}SSptaiK|Pyn9`<Szc>bM%K_FUB~my|xLL$a&eV|{W?9>;
zB|0{*aDhCuoyoor0Cd*GbxI>cA?X1XJ{9*J=Pn7ZsNojWaTb2%zW%3fL*+0Wk^mh*
zo<UaT-XYUXmk{u|3phWYlZrhKqCyT&H#5&k`^X<=U)_w}cJ!-J$g_5k>%Ll6E1Khq
z?Sm2IY=RcaJ*#FkR@LmKFnxo+EFY@VIGc&H=r>ZX-kN&z)5|y9tCK<ETN}BNY!wzJ
zh>LKwvkLjLCL>3BwACvECfTCk@0GQAfyn{mx%`?Ki^xr_+7GD^`P1tl0N^yKxLO;E
z;JZvg(?}-t1T&hrA&u~&su!n`6&-JVrt8Vj<JJog&G}na=zNcR#PViXAoo@B+KCRa
zbt6rNB+6%csstcH$j=I1g%`Q9K#I=s&^+TDNk-DGl!6ne>k4hkT8dnH_?J86w1o_a
z0mi@`l$x&qwCcg;Qoyj(&mISc#73Uf@<(5w`o066OoVnl3?|anA17J@;ixq7koxDd
z(9*e6)6{vUxh-nfo>rdDCRDbdRalx+y-?ak^vd+g$e&CRxLc6!vECm;^OmL(nD|>)
z9bd0nOqRNLDf%!ZZFXsv)=$;`Ko!?HfmFJbR!nI`-IzXESeA6_!SLqJ+}N6cMYt<q
zdHoXAxTLwl^uu#X<pPzi+5P~zXML#h+5w?;&=FtrN9%YLm%2mNd&kv13F@x(>7LY0
zQVfHlMm?(%fm-=|7;;iWn|QR@RU3u?H(1vdxw_JXloDj?4{XEmOoY#&peoIz=FbRD
zMO3RZ>O&cHq{=ee0Gj(UlXPb=bcuwrqn0Xa<B7n%=a1Yp#|$4o+h#2S(WcrUKsUfT
zt6Jn>=n8eGSbN8i8%MTeya^j4M&`0+8mz)r=7eO^7m%Ua^!KCzi0{>Pb>0U~ZUn$0
z){K0BNJ61S>_mUTsrA9L-AC<^QgiaBS~eyH05+9BhnFxmzx-$skn!Wo&gSGvIcfuD
zWSe*fLyXwjkU(=?OjC8KGc+7!hdilCp9=`5@dWm~M$=p#fYY=Y`WwCp0M}DtxuNyP
z*K7M_I3t?L<><nWPB6p8rUmts!@+30RbE|}Ih#U#9W)Od4%7)0arzF6Htn4dk@J~G
zPoD_3)Q~K-BJp&lb-5}KCZW(#Hltvre*XJ?H~=@Z;lj3as&&XYfcVS|XXN6t6uD%(
zaj3jh9m<s}?~mAw71#S?duSh=Dz1~+teZ6F1B9g7zzBSN)zBGax{CcKBHt3Ipyrg^
z5O!PO4BMW^&KGjejO{rl<ul5^<0iI`;ROFp_W+*ADkGm``A<&K9<uFyn|@fJP{KDN
zZS;|Z*1?fKB!X9sT%HclQW^FilGxd7uuVmrbSn+D;b|mW`TfQLl5_wWs{*NOl5C>s
zoRdv6`qVjyaH~2V-`(v-UbBEVODD3BP!b{Qq~-;zF;iUPB831;J!S^+VTe6TG;WxV
zc!EmJry$z?8~(Qj(`#5Q0p}ZO50N7plv-tNtI$y)eg+B7b-2YTotFmwr62Ag;zmLV
zrt=vDW@p+h{@9a0D@c-)kLYEBWR~JcI>ciz%@|b+=oLkZ&)=eS4PHV;$nH1pdG<+J
zE<{Y2`9?5`257sa0@PsS%17?W!g)D=py@_Ii1M1*In1=`2*{r`yn+dt^XLK;p(m`5
zZZ~zvsfFZ-h{9A#P`gK3YFdfYs`w|e#XRoABP<r}IPHbD??qI=kzJq<cpw)@dta0f
zY&XTZ<%}x9X=z>`o+OeAh>(Zw$at+e_}bo)bedk9!@e~R5H=IE1EhM-GKV}GvHTCK
zlgU;P)rXyeq2Fdy8I2hXcJKt`rtE$u4hK;drV@32y$o2EKDW-iGrT}7VcAxI)SC0F
zr7PXhY2Hh5l|WixV7$~6`iO=qA-cD|XH(A8MJ!>!=8vg%Q{9-#%rb;frKH$gD5yY*
zVj}_o$$&lPnn7;4L2iGsLtt$*cAqe3;)>X)HiHjMkM^zAM>=iMb(pRzQSp(>7tml|
z%<R|;9{#Yh5!E$o*T}9tG=V$?Tzdm2>1tR_`VnofR1BmusYa8ygp}PAP}9j$zx@2U
zs2mM6z;|QjUwo;8ozVpm*ms@{7tg2r$=%Mmw$O;sGxfL%G>_#ZA%Tc*O3av6LLV4`
z<V^cuU2opo*159}?b3N=e8|?2uWvu>JL1PHyFBZ7WPGs17ix7HKNX;~<ROnE4=L2z
zK|G|Jr&jr}Xu5;CsgI+J(KP5V!Q%?$W;$@NKIII!WE;98a@T^h$IZDwI*}pI9L#_N
z?$`;vd3;rJ_LUx=afMIZZ*OH~p7*I)u$A-SIRI)6Iw-h+azk__Ya%a|D)<b|z+EI*
zg;ALTkola8W;uIgFtttZCB@B0nKfpcWL=;pcj}xO_<1@<qe>LCr$ut7EA|_zU-({l
z^i*NlpyBtb6=fMxrS?6#k$%HI5m%<=d_L~CiEMi+Qmg<EpY4Uz&$d868@SW`+oQ)f
z&QF|5d~&VLPrEJE;}SZjVCJ(2VJubw(FyqT{ZommMLP)HDozY)FNtj-)egbS?#*Ps
zY5{=xDF}#hsza`>z=uW1)Wc**{URDsj?_Kv6VG2330#zh?ok^rwq+dc3OA!r2i~rb
zfX3}-q}*J|NR?bM$eSp>uNoCHc{(!4i-IVneRIot@H4Vm=#+3L`7<Wa*%bl7Ki3aQ
zUL2yp(hmHC7t+OajTuS>Nnuf>s;<q7GQRYb(?}%lLrV5G9I!Irq%6KE6I>fhF#3xS
z3pP>*lOSB-*rORz>(hUS1d{o0Fdt$((Dy94nooH-4Ur<lHnTteTyk!p8ozlGE<`jU
z@Q%m+cTb~KJ-;=7;3eW7!#(g64K|`p^6$EYdovg{-&DNbuhqb&;mUD27mPSclMJ5e
z5|Tt#zL@ECVa{B@v?kJwPq$n5Xjy%VO|9GColREMQkf<8Q}u5IDXaIG_5)?1v{%3I
z?qzfD6?0LWk=8%iKxzFpr1ewid~92#&bumLWmc46BPJ=O<_8}N$)M&DX^Id!tw>-k
z0D&UeV(f#@SO)k2AE7L!{HhAMm5F~&t(s;5g6&!&r!bYXB((xlBOI%fjDI|yr1OmV
z-U;$MF=l~Bfx>DtWkL2d;tP%pZWLn3q$9Ivg%kd67(<1^XNBRdNu15{5b^zlE1=!z
zPLPEyY|t90>-(9Y^`Wa_>M~|I6~d~CgOEl<#RB+%l5rJLoIavkM{$?bfw?tVigVjH
zlYHBQt{KWB31cI`qbrOgGmq(Q1Ews13o$$HvP;yB-D=&QR!H(Sa?u3cLiOxCeM;)V
z!(qhouvZ&BWl9x9K?WF6y8zZjIILip`wM*n20ptxkp(|ggm4fim89+1d)&k&23WKQ
z0I2F;qvf}O6pGTpcz^^d7f|gmx+g^>69e~|smZ~?VA8*!E<6EhfF44)7u{$`zEb&J
zDEX?NgEqOM-k4d#NX;}>>cL~UD~TW+aPa9GO9sk*7Ro*{N&!os3&Z?eA)k|^g(chD
z4nP&fCM>-ER62w5$PK?j6Y}UIXpI=!RNm1iq2v`?S`_wbw?$F<5|J$SugRye7hkqC
z?|nR<$$*e;BO!>9U4&{9OKeIseaR)zHSeHc-2uy@^(3w$b7V3iM1*ulBC)$E^Ab0Q
zd72XL-IbPjaoR&x*~X7{Q!RUNN=*D8w~S43J&L3Cz-$C{#yfM%d-Z{y%Zf9)eg%S1
z8{o`ndgL58KpC<xGGrj<{Ut90(V2M7o*sDHkR#@sg84*?Vj`HvIqrOc|J>Dn?Q^DD
zy5;$*9%<oGJY9jVp*W;ge>A9uuOok|Q;n<0AWTm^q2dh+^0sD6JcH8?&SVmk9H%Bj
z$=!#hMG`g7S6QxFJx7JNXr|thQ_<*c+EUt=$^;vv%cdoiM<)Uu-*}mvsK7)^m%{M9
zz1Q+B%@08*UA%NI=0fLrpNRSHAf6}J7AZszjiD)1@9$M=+Gxtja_BH|y7{;dk0IG2
zRpDL5EpF$TUE6d>m<9n{Zmx;4Aj}}Fd6*6v?x<&PM<^vj!lnBg`xoAqQiM&5_pvDE
z7dsrQ8FH6yuj$5uu{>(`zI`XAEsOU^#_#O{L|KkNTpWh`SJc}P(te?Pqy8JaQZpHF
z<LF+9?4l}|T8PJElv;^eo|nmhVnBH}H`<K5>c75z<X*xZZ)286l0Ks`Lu_6zVup}+
zesSpKPpeegtmS_*JNi-|+NHH`sP3jRI<kxK3{>oC-wHn#!3%5pOLo&iA*5_`ZOmX5
z^K)6EIY7q4vSp{mdLk&9#44~XJDza@<kxAL3e$#O>%#EMz1CP)mN$PDzg(f|R3W@N
zWyK0#sG$@O;i9?&CT95C5nNvdVIXpBu5oSJhiM|qhJatQ632ybh?Byx*9i0Kw&lbJ
zqWt4U3Nh-*fs^&g36`~PW*%~1v5ff!*Z)!5{Vxl7?>0xqbg=1Ws34*iRS<SdIWHgL
zkx}k9QT*Ce_YgrBgdnj#5wY<Fh&kQjNp+?Kt~E$+SkjMWSTumW9FyOfdn-=TNMc#E
z<OE0k5|$OdpwQQLYpwaI`ORM9Fe`2%m0EI~VmBDIZ#-EJIc=(j=t4?xqww!-EG%O;
zwZ0lV&CpZ-EU%&eJlYK;r#EY1I%&a;N2XL47cXAm7*_=M7c7pacW+jo{(;4Bqf1!H
z(z$P0V&XbCj2xEw^7;_{wtsy6M}o#&ivJ0*T|-S2+0rw)0RLFN|9&~z^9B8(ieKa7
zD<oL*Tleh~zmhui^^HxN)*Q0^f|d1`7u)Ztc9ip2zVL}yBJFW%(~WyievdhJipy_x
znbnuSZ}#CUc;)(P!;>3-ZusWr565pla{caP>ULZG>Vqx@9j{Kd>EvCTNs&0Mx^Osb
z^jY^aC&^eziv>}LY+0^<{Tqoc|Ni>%hyOZbgU0gw{k7?f-$_CF`Rn@EBPp#c5oO4!
zwn#}yHJd|%Eep6n{=htGe?%_5(>m3PjjaWMHbe7K5ZSXx6VbI5+ESyy*xBaisoPmD
zN73;249)l7^ZLC$M+&!6+vS)<(f}-S?kG3pX)*m$-(sjwl>NGISl(-SKOA66Q$N3n
z3WFp6s?hxI+kcg0@@)Dq2jIFQNq0Yg9s2S>7rq)TzT{WY&|sgL7^<|Tj~c8GFFkH(
zk>{iH_Jd1EX6UIktDgr(0<l*@b!e7pSx3yb$1yS6@tJEH2%`e6XRc;~C>&2gEW-EW
zl^oTO3qD3XZnXHdx;`k^%#KC%UO>QlWG1Dbf0dQB)XY!%NNGndoncYf0b3eRT(m|x
zE3UK0sYF*8<<(>2OeGeP<vCkdYW}%@8SwU~^ndxW{pSb&2YdQIi}Rnw`5YqfpZ_<}
z@PB^i+{iWCPrpO)A?k8gboCP6w!D4&w$gD3FuND|eyC<MXRe@CV={)j$mmz#?OQxK
z4=6`w!EJZ4e8SDmEq1o06^6PO!HP_uCL#LC%wSBmA2Hic;apKOZ!~1+6%6}L4HF0+
zL$P~qhv^qA&wC8OA{z#RnW<JsrnAKTO;1^epE>kR-D3Rg_9I_4@W{P~ZJ3j2#{2KI
zw6wS<LC_!l5G^3G9u~?|AcT-iP;(`?e}G#Hi$N`WJ}U*O3O?%O1Hm_R3_-&7oxxWu
zmn(YF`uiTX;fy{A+Kz9ceI~kV^7SVC_i_b6!-e(KcQ35RrBof*ZGpdkujQknqoWf#
zz>7bym{bE1C%2J%Pcn~_-GjK@VEIYhJEAJUh^6SA!W_2z<QzJquaG`(*|u%lmN8n5
z*z+Dt0seJs1t5z0lYh0d#p!Zv`|?XH2RbUY#3XmG%?B;;Yq|`tSKngZ6_*u=0a$3V
z9Jk8zFoTiC25Wk7m&Ho>Dr>qwkySu&;Yv=?AFy0rc!FtlTtEZJ=PLoq>GyJ~XU?pp
z(#yxkC#`M(sGJ#$Iq9FUv(7S@egW@;YfzDHfKsBVPb>k`E9P|%A3p5PB`z{xbI^_+
zYcvP*)_ouBKe$QS!oor>RN}zm%Lfd04+aMZQ^~JkN^gc|EAzq&PDMSq77pzvHJ^c;
z%7bY%-WgZ%zC|8@gtZT<G4nARVG%ad(*0JAX4GxsV(#AS)^ulez#c_T^C0){nD`e~
z3nI*9bN*a&TJ2~HrIW#)pTAWj`%5lZ`P?;4%;MdZj%Q`&o7g>=o104trfOwe;w+m&
zK7Vf$>a}oOg?IWK(GGNf@#4k4E7&JSVUPo#K#O8cJz9Knq_JATMXM7L@Ok~?fHP?*
zJRM(+NjQqhcyV$>=@1LeVLm1Q#qX*(#okXW1fsVa_aZ&0P>>1B0SeBIwuntoHofA>
zJ)c+64gkanw0TS&=O7?mgUeS|AMO^H$JqcSktn{fUKg)Z1#ixMwzQYMC7CJ4^8-of
z?9&(#cHN;ZaD9Dbb>P4O>PF$uS<KBG(&Y!|D5D%2w|CvF8DritjV-sSKHaQQknBZc
zFfw9E_LnTzQ_Ike;)IP{%(x8{!yY6@WZ!qu`v%n4*czIPTjRPb8M49R=2=s{e5k}Z
zIx(>^U|&>WAdrdfMC2YPnZ;9-@EtSlnGFka^vm{AT#Ct!)bUSG8eV*4gI1+wWo4<Y
z^l_f|{0wZ~^3QM9pK)*ogH8`e`QF36g=b%3pe{D+4!9bf>zX399nSD^WGjM9x&NWf
zrn7~ZipcVQ0ll?Mjo<rWIKLbR`;hS1$Eq!$ke%t8qRI5dl%FYFb2bH`mgTSYv7^Z!
zC68corAgu9;mIiB+p}j+rR^I0GfVVch^tQ^goUSM%6&f4@sU`D_x&mOq@=tHmT(a4
z%Khh&j=nEO$0FvQumUqxom@`X49rSZ#V$cwp8f{AG&qD5gW!cGUU`QW=r_zmE%Xa>
zyFAx@Qd$%a!gRtqF${@J4OomcW{5P6M#IcAt^Y4)TE%k}*tZ#Gjml;4WO?2giM)?q
zeCyV&(VRnzXFB0N(>COvZ)pcPi-J>oMpVat%^Y?^&9Y_7^k5mz>UnLD)~AAp-u!jz
za^L|BkRHmB5k@~lb5vSMsfw?tDZ`ute32(zp0pC90JI8#k0~7Zki{YDdl0U7B=t1Q
z#Kx`(6?SMGr1j9gy^7ltkIWG1qfiqn?0^9T_4o1J^fUf`eg-*+-jD8gU7)k`_t$ly
zK<1)+{_~+P|CiJ0|NJ0D$IoBh|5j`K|9No^>0~UXd=)DwpaE2W3OrLEtVHnQyN?%v
z^Z_9CURW46xAWwA&VyI?jSsbPlaCcmU%*UmZUWRYG?_*|EGK0aE5_m&a$$hqErWZ$
zotqR8tQDr{pAj~VR?miCm#moXoPj*$DnJ^3nkBh()y8;_`-^)scI*D;#pN`Lo}Hb2
zicSJFMDJWlk!!d~+@G8T=;V->81jqAtWy<hX?hdlulgj(Sj;ku`5Kr>)_bk?MPr^#
z1Bh&F3uC$qtcW=Zny&%UNn_|qHS-3<Wu?u$tXQt6vBPnKBq2(Qirxcii;vXgxW(Vg
zM7~d}nNbes_`qTiVtEVReql25IsR`_PMtc%0K|Qlq@y&XbP^*>HWYbrdl3&4Zg6^U
z9v!AZxP%?AUAM08$i(9B5Z<;}uSOKmX9H=$F>;hi^b@zU=x<8z6VgYS>V*k*MPzhL
z=u=ISu2hsl2s={(1IEJ|a|(?}1b7!;iKh<GKXbrlPcBU&p-y3BWTa}Wrj{0`m&C|P
zYGRYb3|;!#!31E}%+?SA;mdU=kECJX*0SZxw;*04;p{-An~dy>EVW5K#l#t#OwW*E
zF#}SvyJWLTNf^2IK7RbTA=T(~Jf<8m4FaECv+J;d-efI_^l&XokRG1WCgcL7=Ucme
zeGyu<i~9=W8=(P6R$0<@I`WS{_TsrsJz@HdVeua0FM4scPfblJp>Rj`>;bJ#0d|BQ
zZuMc<35tQ{xt`ncufrYn0pFoXFxTnKCVKhjG~B}UVSrVv+$gaDBZ6Gv09#1bu28=^
za7x(k)qfT-Bbj_DJ@pbT;`1!NXk;7`&GPs|x)I(awQy0Lv%KFqIydAZMSMy>YZ{+M
z9}4$WirUp|a=*W?X>kW<hAz)o$5wJ%K8tA|WSgZ_n|V43G~ctC7TPx%Or;N~PZaLK
zJY=R)hR={KPGVq6J(CPIGk=OC=a-!NG>6lZoOa<J!AnAi&1jn+Xw!pshfnv~+f090
zoV=VwJBE|2%rMl!MYJoJX0MTLDz8TO9^s2`Ho4usyMuEGEQS`Cw2-+3B_#eP8>O5j
z=m5fq;)!zc{^ki>{Jy<|94v?AX&Mz4=>7nZlj}{Xy!)GRk>krK1LXj@RhAyWcCw@q
zKeC&0kRQ=x+=djxQ%F*?G;Z{}x{@*lm^QnY)VzKAm#<Zy<lu?O{+h9RH69>2fRP(m
z?6V5o(hLY*%;~Mw4!V$e9ZU(yPE&orCN7*LVB~0v4%cclw97tf_?MKF+#!x1jSF*`
zuyaxG`uWnaV{k-eKvv_1MFYgX<?Y+k=B?X=&jVP+7ysYKLJ+NDDx69WU_X1H#cYrp
zfJaS6NqXRD+<W-2f1ncP0k2N&dA1f^my<2^kWew;jeVSEUTf4l!7SPx-*CoZp5$9J
zMuVbct+~Js@U&Lbt5OHra1QHq7Q9`ew4pXL(TYYZm<Asd!gldtOjuL)X`dA4Zes~s
zOx{^|l>NOrbuhnS*OecKVfQvfV@}g*`;c@{*RBc)`7s+DWL3taPoQkX?z;pY!?4FC
z$M6%T1}gOV!3Dm9x<Ets8q!%Z|4FNf1;v!yDaZkmy*|cchDW=f8TFq>FXoOccc81-
z)@ASD;6M+Z;Y73V=7+f#1O{xDVjw4aVG7q;W5sCB0Ep>)-Ao<rZmqMYxaRWFdSz;3
zKeJ^iYa@0fKJa7YWu#XiTOHE3$;!&|$72VEkxPeZ@~A6KCLjWHRB$TNzYP$i(N7{7
z{((6*5vn~`yo^f4CF}^K<h)fquZ30pXF4nUo1+b_nUD2)>g~lG#``3SI~C8N=FB}S
zxy1acHd<L*lU~=qV#fMxD$0*4v^uCWO+p(S1{EEQWQZr4eTSqMm-ssJ_aS8ig%|7z
zvXoc7N~5imJ=MGXIy*W#ApM{8qB%L3z$F5L=z8L$q9|orWYd|ZARwEh59x~~=5tA-
zv|&K_V+F^=uaa>=(D2m`ctu)*{K<6iBVq<!H3E1eG~0s#G?le37?NE7I)&SKh;;R-
zRAjZ^;?P5KORph%_0vU)mbm5gXu%;DSe8UH6^Op1<Pw;Dx?l6c1oQS3&HVe}GRdm;
zQC~H#UX5s(@afec8pp%Fe|*3&o5xR|iqg4io2szGT*$<e4x2B}O^@k=`*-K61khRP
z0%HU^JIW4E1gC~{Q+2d@Gv7WRfKd%UXAY$c#oj;hA<&K0ha+tth9><W^^KX9FWfG(
zmYfHycx(q-*yz%NU624wJJ*phD_jNq#%M?;#RvbREvsFYeJzDu&guzz;)se%9-54s
zw{7e8O#;I<I=&6X`svpTAm)Gn&WlZmx2y6KSwxf0k=1h#_LqoRgJSz60?B~ivnZb@
zGdwTy46POq|L5EN*vROpiYGY7I~fP5%<h^bawkn7<mYU|%|<7+irs$-&aGTXa(elS
zs6*!mFN=oc7s<GI1-sHI)5eo17(!b~_F7Ah-6dLkFjo>}yX#)zJ@Rw*VOQa9zhKas
zFM{*y$7lonVn+h;jg}Qrp@WPl!9sKOy{s9Y?`EhZU{TEaj0>>@H!1{+-m}WZ?_e7`
zk0vf((GKzRqMJL2fJL(xt*opR4!i@|Wc&8*bk&+n*kK#f?1mkjE}&n+Wq6GeTCpmg
z(PAF6@Mt}qHI4=rE1qDYM!4`B{3PvSB1uwGL>L8}ZjJ+;e~j>k#{b!~XLoFii_X@i
zN+>CJL^e{G^z^dz_OJ<wtySk8^3A`{#;+5ez4+dL_~~=PaRpZAevEWWx4~}7i7Y<`
zDK{d8srn_)7Lsd5b5_27?E#Kh{dT`+2oIBp^E)_^VtC<$T&#|I%~!;MFwmZkyX{og
zI2?3h>kNbuQnrzUFNq!^rDl5Z!B1)?oFPZ7WkqCI4lw^ty(q#{oRd7>-M2Gptip}V
z4DkB5i*1Wdztc0|D}t=yEMrYn+-76~h8xDH<4VMF)!1~S%Fty;aYR61h_`iFy&9n+
z=}HBYpmz3nv|V?1w>0n({@g)y%RNzbs=fPJm=ao=J57TSH2O}=_KX>57*3qZuIbb%
zsd4)8d|Pa6luHpat+h%eC?8nEg&FWn9$HBNp8^gOnmmRvkfBzyVC2MGC6s;;K`<yy
zpMQ&zk3Yn>_qR^BGviGl4`>Vo`NM`l{=_fO(j!yC&SfJ%d)!oS&?|Z8k4vhmzD~^K
z&-HT$=f@@h5Ruk2bU1kor^&}ES1h_uO(ZiVonIMWJ;6A9CiMq#z)Ococ#v~}pImQ9
zDViW`s%a>ddgTccg!CkQ4!yS6MBaaU)XZ^cRA*XM;4<UDhJIZ<T~JWKclxOXGF-YS
zWz+flhB1n{)_4~Kp}zw>3|7zj1kL?80sa|d^g#U>behP;6qY*SNak_*a-a|ibKHd}
zZT_t0jEsy-y25EGj>UZbdik2g4%Lr*@Untqbab#Ux+;{+(jM7zu>ajN@vccwPWcn6
zfRBoT;mgmt2j7dVUo9t7?i#wM%ATizqM4g|Y&@9#A{jA_rh|82Gr3br2vF*AZtkp#
zHMuus==hwp$;aVmsNCobaz>9Pah<M`$@oBd%VK>|zWMXkw*^7sn6a^O=>eQ0H^i&@
z?GLnpuslHej`XJ6kAcw|dz5vg5V2+J$;F|wv-6DTJdP)-ex*nolV+L>X61-XlPEi{
z`Ut9nZ-rL-BnaFb*taLo(U2cuodpQ3`HK4R>AN&AXE)|@Kh5h!#igEA*He+l+-)^0
zKX3bF+-0wb3DA>EF8{-+VRJj7C)A0|MheCphzcSv1AN50$i-DEuEE7yr7nwneegzD
zJ#}f?2w9fLD>bq=j?fgjE_+~<v!GFWA?;Svh(k`34F2cGx`bw<LfoEVBTdMr{1~t{
zI=Oam4&5@`pktwKuJ)KK$In@Y^AB!C0`|Bvma2o&<!_d+tNA!B_s1oFajwso*XZYF
zo-xs2=+cQCC24I#mbDR0@pU5$3z`H1+TLB<Cp0gQDcQ?kn927dT^6cEF3QXpG#U8Z
zC@2ewpVHd6XWuWSFiem9@#o4GTf@_3D!}CqxU<sE&dxXM@B8_^0?CNq$-MC|ND!DF
z=~fB|sd~86X&wv>GNb{9N+9<TF%%|zTi_+Onr=0`Tp=c4)fT^=>(?xT9ki?i`!w=4
zeu;ZII8mgc?0xo!N_<!;5;(HLrKC>Qlt#`C1AILrDhb8kWVFk4H2=i-RHcg~`P|a&
z3o_#|{Qfvc^Qgv5w~&g(6B+EF%zcCT%oV9zze}V&8qYv}BBfc#=mb!)Qns#9w;SgG
zOZ^JTq)G)%WW(hW-ZV!2&W6lOy7J9sPa%vH4uvDIpEYG+MDs;%QGd|`(SjIodH~O(
zIf(@a5yDS2N|>hafQotyTfO6y?zA!y=utW=hTAN}r@Wb?Ss)lobD*mM1l%sn)9%{l
zvK!H!?lDwqcBNA|?KFm=0Oyx(a+(}(5AtU?@)DFg<QYO<cjUN27Gna)-GlP~_&)WL
z(TByIZnW|9PB$kXB1$!9;icz+BB{Q&iSz_acQZ(De@d{?h2;{7I^JpswK`*5M<GU6
zknI(lb~n*}Z??a4bw%r7o8-By&B!d4#QgwpZIKJGum-gbI=G>k>5$!)P~p;{**2|q
z)0%2o<iPS}R~ro$lMTV4)Ud;Z1G9KPM{K1<8_sE)ORTGh^@ajve8zE^xNZoL=S2Nw
zE-ciAf486<Ioy;vzys3g+K)}R6RE7zVx$9x+;AELi9C{|{E^F2RTWeh6uo6}4|qE)
zVoZmc23oR&sE19yf=F4@YIg44?MbFIUBPHk5avJ=29>~pz4(l#oG-zwTp+@y&P0pg
zB<;WeyQEi8Xn5b7Mytt{iXc=nu{PQWQSKc3cas6nvA#Aa){ULaNt_C#_$izQD04NA
zyOy22Sj_}}#xe|!Sz21sjzIIrizZLz0_+x!9i<_r6L}NQJ^~_wHYEG)6k0d#xCV+~
zq|pKNB+gcJ2FzS{$QEAVr5VVlgnJJ>Brt)7bl?z6xq>QvrmqRNHO-U`5uD9K-L(Ow
zzFEuDWLaSs_?9v!(a><p^s7%U4$##$KG^#3X<e9qKtK^?L`Zb+QmCFh^9gn$MO3Qd
z)tNdo^SQ{oPU7j+PK#BBkn}%*G7D0T(1#*Z2YSPLIQ<04enc&BrN2P{Ll%Fe-UtBQ
zB7AUtTy4pKW*aihH3<<xE~yl51Fd@j&4MASHW8tT%R2kIZC3toq~iYpb|Le)scnig
z_#_+)82rU_z$OC+5-cJQCn_c7h1mbnSbPFDXrdCrqZ=x*LL|>Rk<+AGBQhU~iZ%3s
zp=UgXT@a3*oh6{ZhI|*7Sqw9BgWTa2z=5~ZO(1jt{QKkzc{h_ik9oUQ*xDDt?CR{Q
zo!vD=G-3jnAZuveCYAc<uS=zY$b>(7v<Y1_cB%r%-W>x#)>0^=9Z6*MNh)z0FbTM$
zx>4wLUP$2Qn8x<co`vEye3bP>lh0_TBgG<WicR-_3rsNonI*VnG1*`I(|=PJ_ut4c
z|M`uJC^B=+|D7rHpT+s_SRCSCp_;fFge!$8!bHJhF}g=|X<V%8?e;J&=zv@?t^jQ$
z6AMU3h9Pae005Ka{XS;F7W@`N?unk=Xo1>6^e#D$8EzLxVT8~;RI788h2=Z5ce&>X
z)1N%{Hv+zRtVDf3gA0p){2!C8{#zR$3uW&@Zp)~K-1%xgh%32K;9IxNC$?PF+O&GF
z#n+aHSKJjBeY}(1W7&nmttP>%zY^PQ^2zAPTAP-=gIfl#yDcAd`*OFE_qU~a%Nt%R
z=Sz(=9Ie)Ft<|p8ewI5lS~L3T;KV!YiiB)k>y=|;S<KJsz2v_*^c#CG766Ou@V~)7
z9!eL2NoH|L37>reWSh;}kAHy!(bG3Z+D)cFocQVwBA?XjqeSZ)oz-V@p~GF9l|(L~
zn1EAmTVE;{th~mM{{%OL8G>^h4nT#}->`n+|JpIeR!H(OZ>)^<2DBn=eVOZ4BZlno
z4rGi^PdebfV4}e5t{yWV0U4%<M)l&BcZcEZapcjP_wV0Nup2eaNz+2mjVQ8a21z0v
z@p}wo$u`E502cmckmt0J?gi$$FWaV2vQ9u&2EJ5lpl@8@LQQ{#?p<Qluiy}Irx&Ba
z;zXb)xnMHJGMm}4#@ulCis2`v9Pt1dekuxgX!aSg-sz`s{hdI<&G>DNf-6-c@byde
z>lKnj6Z!vQ@6F@6%=f=>bLKR2PMvC6v?w$Rp)`?-LW`2EWJ#f%lw@rv6_HLeHIXG-
zlC1?<!cd47XJ#zXqGZWdXDk^+gY3oqd|sOG_uTh?_v7*V=l8u&kMlU^Bz!)f>v~`B
z_v`iCUiK5YM%IO`K(n_Yr6~$?zG}Lr8-9mgLY;8|v*i%E<}HbKQn9-pjmrIkuq`;D
z#Xtk-B}pPXO0V4G=JB1k7ESlL1T>J@ghV+&{sgi>XQ8|E$G@-lgruDm6R}5{`WPJ}
z*k@2A1R$wzA<`>naI%W76L4TTrgciS0K@QwLq_&R_>R2=a$!1tMwh`FteJuV)|?*D
z#DR%|=K>#xlVX}%TZ>U9Hxl*W;63IS9)yoiNG^mi#_^bV;(EWphuIj_8_NN`!uWO+
z#KQ1601$KOEZ}cJn?`)So3qJLvWb(q7mR?+ijfN1VxHa&>7=3Y7Pk0_XNbIQu)2H_
zP%18-$z@Q196}J(!<Z!^Cr3Y`8&b$VX`cg!uom3+LYNi4O(|9r$_!5B;k+z4aZPB$
ztsfE3Q_6{++!$Iavsvc=t~GTKT7Xs>!7eU-;m-JHgrF*ueVX`oZ%xFc<5Lr3_|X_F
zl7Z5-n2v1g(|!G~V7OteiO&8;L6ct~2$IX|gs~omoeYJU?xPh{jg@23o;yQ!s~_{J
zwhW<E<iO6kZLg6X!3mpllO~<KjcQ)7w!m~1FLC0ZpMA-Rt9#-+G3v)gSN`!V7hN}#
zn&oO3?qEH(bkSj8oL_MrP!Df7gv0_-24FGEl4;jy_t-Bp7+J`^B(pRBF3P&qXgZn@
zdz1RM>vI*q?rAkgkOWx(_w79Vh-lSjx7c5>gJ}tuUc^3#KnPtH4XYTKkQD;{m!X(N
z?cYf{f%hH7Q=}fhFQnlG>6rcDN?W6hqASnNNj0FdC`5Lf;>$E}4n$bVFY-s}y<|W)
z80Pf|wo~LxoT6Ar`3C`BG;xgC>O;7lWJcw4tXCR(&5;Z;;3O|ggffJcUlf)`gN}gy
z*fV8l_Q>(!&!?<jz;s-{>^=lW7vs4OeXf=?m%d8ArodCe(^Ej258>?R*kNph&74_2
zdTYlZ-1X<Gjn7s_Tlw(Glch!)kod^o|I@YA(sc!kr%eOVxs4+K33V*!@heudR!@X3
z$)`0Nf+b3=pszc56?5QzxN+o=h-2B{ImaPP_|g0D-7U~R<TN-~p0k!7mX!&}1fJ5_
zbi$+43{O&TT8P~+@6&E5Mn_sk3r9N@)#)ypPD0k^-k#(eU{(u&w>RRMI8Q%37IzaY
z$NuPIR2Xi`4GER^(yf{w-&04&Tr%mZ0XnoQRvx4A2PRtIP>u$r=i1duZ_|t$`@frG
zV}Xl)9t4}H{5*)Tu+#jg-^YwZI)+boawgk5%*Xb7KP_=Lo`j39;jJMLVLc#DV;-#G
zg;*y|2x#=<3|ll**V(aS{Rp2Nju4U?tKr4JzBRhsc!uz(v~T63VyM85zy)uo2(LGr
zKn^Y#Guqis<B>eRuyc5ZkJdeaQ7(Xvim>s_z_Q;*Vp>S}^8QflU4R-!>08mtItb}F
zr|cXS`zC-9eA1}cJ_tOUQ?iyLa+nZ#6K>*`lxW|<o3mk7LmzVZZ*#s`K;k+#H<t+$
z+}jA}@J0_}RlR6Grd9Y#+eb@^#_>C3cwq$fgYr=op7c5Rq=Ge!ju&e2YO4*y(339V
zH(L+TjlH3kjE+7eM;qD|pP~Vi;N&&Nl>I@MQ;UGxun$rS-x*t1e@#d6AcCCa<32JN
z0d6`;BvuD<DktCJOAKLb<f(({7qV2^r+K_aQ-22n#zqW=3g8~bv~p+w&+(>E^oXa9
zWN;BcO(oAw5igK;LR;T;08;T1_^$@M<UHNA2XM9Ivg*JPSU&=Z{?nF4)8#6+N7#Er
z`EFmHcfWns^{hgC@?F23R0pZm6s)HxcUYam5jg)3f}~qt!x0Zq5<J<WnKHfiYI(%e
zu&17h$BVR!v}<~B@=?5C^!5SB;zA^rN<ss$EGq_^^eX1_6AsYr6oaotq|zM6Xrd`}
zs}bkU8Cd~{dTOs}3*IwP6KK<tz9=wQ8P8OKx9PH{hR@M9)Z;#37*wpemQE!3x>Kw^
zCKn6?C=I`ZEyu{dX;uD^!B6f?9nyVgh7*VJj%RA4F!L&CAG-G7^49#$4$v&ya;KJj
z07|b%<}4oi7RLzA)tdpxRHq}wuS_jy%Od9VGPZdJKE7O^7aC$8z>cBjhivM#>pT!G
zI~_AO|E^p8Ef$?DvkH_dt%4|&bJXNm**AH(U1`=G_E>Ha?HpEHI&&jXbOY#WvR>w)
z$cTM^53Ymw`WwuE<#guIBr9rH?%|!tReEC>oouQt2+j%b9!**E*3dd~cis^lmO*k>
z)j*6Rnb2qaAOYR8Y#Qd>X+cU`>SSVyn{7a>ye(@hCQHP7z#SIz{@6=g6sw=iwjWhd
z*;<%5#~?w<dd03-Ce5b}D|G@JU(&5n^5n^rqqAq=VLEYDLmj`}uJRb*;@QcEK>O$f
zrnVpctoLfqa!L97V{bN=%$5B3`q5ybD|V(OcKNYj*C#sU<UQGonk3}C*JbQc*>|BD
zGD*|E$x5(dCk`SZVg(ztNqYv5P0l*Caoz`lil@cJyvCRKFimIYIiF*09aVX6jwatC
z7YDN6XwOGGG&#$0)(hGMCzAr(`>0QZsg_XF$)y*(N1;*h^$+u>10B<88<aF{_IeLk
zQW_Hr#3jAnZ`UnGGo(&Vj75`AaRc_HppX#b-_-#57$s{)5?8f;1Pkr&ep%b=n=UaV
zW^m;7SWL#s>@2(wN;X-nq3YR{a2UIg7{U~GV&aWVXJwHhyqVlS_JO6BmSA*~5LkeN
zMOq>p2P`Kml9Pk(Dmrw8FtH#Vq9c>5xurMx8UuZycrbta0JT&RM#F^~sZ(qSedNr$
z%XaDSH#RtLyxjcKoTbTmg$JE}=iEUgq)f^N#U#PO#VJq0xrx@sU_&zRGnIa`F<x0x
zK6=Qo#4m6(uzP!p2dAPfJ%md9OlrBrUQcL7O%4S^KzT~c0XLaOS#5~7gwzN+Ly?{=
zN(2}naj|r`<#a&eOIMGQN&y~}s_(>s#ylh_hGAB1H%gE~>_L>S2&sj@hj%9W1Bhql
zwbD_4(*~->0D~`i>5TnE7&wFifIe6%6DaOs&VZm#=zKl8lmy9}qw>SIx=`)?fl%(e
zcig49rKJ!!w-6vhp$KHqhTp!WJse2iGUG9Yc8xsH%Z;c$mTlR|6P-rQZUlx4Nxx`<
zX`3W?cy*acILfF`SYddZT5=wg#aS<V=mOs7B;<et!Ix@b$-y_duTx8l>NqBY5aP}4
zpeV)?MLT5y;dcWF(K{o*Dm2gvRlMcV8XKJf+37w)FN2O<;aDu$WDfI4d4K(%dSD_h
zg8EPh?!s6@Z{mY3!@&(rG=4{n?09Z(6y^K8x!~a5O1aGr-_bt$>?mZtRL$5`x<CB-
z2cF84vQZek0RC=c?zThp%|>ZuNnL;8zE4!vNESduFatC1zT=PyC=Y}h0;@d(*~FYA
zat%_R=MWu$<+0#RX}cdeQK(#MOJB(Yh*-ChUyU!m@iaz#P0B}ZsaHo?O;jAt_WrZ0
z)*_H3#*Nnv-l6!|P#lYf$j13Pey0x{;w#%!`B|u;I}-p7w=6`$Dk!jH^m%{v?(KoL
z@J%0CPeH5>999bPF}E-xvH&7ABA44d^ldz=C{JNLJQKx?M(&}VD*%RUkvJNhmZj8n
zol*AB_FY2L(^`Pq%dF9@9svMnB&MxJO%PY?JoZ}$`m&sYyhjc=CUqpg-aQK^n8EN!
z7oHzQ<K5Eg{Ue-S3dA(|r57buL3!owZwExpEQ5L$W*EDp)!L_1C*St`b&I<ShT;lC
zEbH>1gjcS5$`QT^Wh*10m5gPo=XCTDhY)5vxT)O19uHb0qV?pA8z?`?a8m)F+_tdo
z*42c<q&cG3lS+^H$#m2)$J)|t6V)Jk_m)A|a;r9<d*hkm!>Ho-$Gt-Ib)w;Y=AvX2
zoA%|N8ZxZ8G&TN}ckk84xsr1wZ65XK8|QdU7FS!rvMx032AFss&<SymJI%sX^>GqX
z|I^l22FG|^1Hd5p>V-R4YTMt$S|ICx4dA*?;9Sns1~4rDG6v9fgxNpx(6uyyQPImA
zV+n}Nr})1EK+1o1)PQqyW`3y0$<6#cUG*DSMr0C;_x>P$(RGbz_pUKymHUu{K)r5h
zKZ5#;(bXLDJ*kwlbK_r*c$d8QVif)9;%^|f)<hy(E?1zBm%s<PmPk*i;}Du9mUq6X
z)JrVaKn_RTNI&{uzic-g!3?PI4Yk4s(hWSpHqM{y7wXlJE^cvTLb}ldY{JR-IC9l8
z&R{CUQF`Hh6qH{pSXp&x_(T3eTpN4-Mt;*%R9A7uqbLli{+DXuBmq#yGV1AAGu`Y$
z;JTsAkmMLtb~<^d+mIA34pET8NTJDTgv2C2TlkK63D)AVzs*%z#J1hKOcezc2lsM_
z?8T&1aBw$X_7Jj1N?&RHfUi+VqlusI^UOrW<>nT&TtDkaqZ$<CK}EyC69{dZo0KE}
zjil~V&&6(UHWn&tC~1^l#0Xv4hE*6@5NdW4r&&%D=-2=h6!-Ky=LvOI81=DNirVoo
zp(MRq(ZmkobZl==Pa}#5y<C5YsS3DDx4UrO`1><$*!Je&td`3$-t9C}*om5Z=iW{e
zBmiIj6hvV4<L>(Ktdjl18XWAh7Tef_I?lOaw-*0^^2Ei*9s2;EFnmzA?Ot*hiekm=
zdw_Yk#rS2T2u-0L>U`)j=y5pY#_u+lj_gPAa{LqbroW-85JIu>#FBFv{rv~c0$yhn
z@?)>pBWK}!J}{P+_A+!z9FLTN&3-dJ!F4F0x-r8;B|vj@Fs>Cf?w6ZNIX?o&*pmK1
zIt`#InL#KB<+i2|YV%^}yPdPlEL}En7%%IAx9;A<FJepG3`|P`b^gX$fpe5?G`#En
zoYX0O-rq4%$3=c~;lW)d;d4@#BBp59!Mk_<T-ZaY>(x$FiFjlkaqvbA8QMSoA$a^w
zq1LmxnFdIx2s$+mo!B-sf4nyg9<Ps(XJoEhEPmwK`ZL*&dRm?$m}(k|2Zg>nY!0Eb
zn3N51#cUQbrJW{bbaM<2!BXoBwkLZ-Z4$~>(Ymy?ISBS*$f1$ivKIr-Q3*~;opltj
zO+rQBxpTk^HbWIXDLcf=ZWxrLB7lsaIw0+GD0SMuG4ljR-09^1Yg;k)-$@N)nz%KQ
zzBm85OM<bD<z95$1XZr~eorAT;ikA&4Q%K(U60ti@e*c?==wgvtw`HaL%GBIK_F|(
z?37+Gwv@Sw86yl~+zFjaxkt)(#HpyWb(NhE7W!jIXQgD0DIbGbGMn{w?6$t0P2}rp
zr@nN#vsD>Rdu5|T+2}YPdp`7Byj$owGEqkPRl%4uo8v!9x7`r%xnu<apYu1Zz3$@;
z>|u>)t>UBERt)qmEe6PW>sddZ$?%@s%1Xq98UNmklI0oz4W{cR^d0}Tk;^@1xV0=c
zw%k49L14DQ{6zdSvuNb1C8wQv@HYaHrAM4iv%F4M2I_juJgUa@jXQ5HJNR}MV@bY9
z9zh*_p0R~G1L2`R8rF`osCfKQ6_NFnVHI}iUEDcs`-)Mr|BPqOC_aiRcFXkV{n>&3
zl1??5@%jpxrlzFwSaNPfvI05{NxZ1efV+&urdT-@>8*tay9*AwR!VPnINur4$|ESq
zBWG+0u*S<l9=8C=?tinf*I2F0bFpgE_40vvlBRz9-<4$}&Lsah8gpwbB0;$bps@+X
zv%bd%yh~ATxw@bB57Rd9qf^k#8)N|viEqjtyCz$+&-T3WMeT}1BbQpvd8-@cCUl@~
z;Uqaf0F0}oq@cI*CYk|QaTe`Ntb?ziFQHbHl~`_F(r;zMc$)oXV~$HvK1Ps8PzT`3
z7?~Z{RVn!zi&t}OsO)s?D(utEc>oHMG{+7?K_EAfDzDdn=`DM$osonR5UW?roZCcC
z&PhXqzzEDz(?dEbr#o@5#}{qX@%fA!IU4u4_Y*z@;=~C$Vg<Whxffc#GueO&{^_YI
z`G*$Z!h6V2T4lU$ACpHzV%Bvy<uZFTYLOYeTiUw9@#Qyd4>CVQnmxDIMRk@r<9g*n
z+yj=&wNEQ_uhvz+Kvm~g5$Pn9VoH*_;oZ;k=SZ49z*!S^$D7$@To2FkZP63@V>gm#
zlfp77qg9=fQIsx52suQToUEWg1DMJV9jkwZ#@$(F9U|AMm+XWzr+d%(v0WNz<m4se
z>SD5$GLQcaEmWxhD_g;iged6>1V3@q*&RHXzMdr!zP`gZ0UFE$Y3<Jqkzg%Ay#TyC
zhj>SnFFY2KYM*^y|Fw3V>=ejh;-5JHznGBjB$>NAN<Vc<gi}dZP*44#B)%>Oo^KiA
zcF(f@*nI>KZ_8U2_I(r%a@mF)v<xguBZhCmL9nR7?Bf0))VpCx4VjDE)E`~MX_&U4
zpO^Nv_n4Ouam_9cp2YmHOPBf>cg@izt`^d%wRK3EQtG=aRT;Wk^tz4g<`iwes#$f#
zZK#Q(vF@#wudZBFn+!YxtNzW#HYJVW*-h6=&CD&Q%vFkesrgMQMCDWo>Xe;;RM*J<
z^&nhqiD`>?dE>4wNs$PXuDL4%fr<gpoTZ%BAshAWh72ot=##%@Z@8KdFKlqscptcW
z;|x=ho}*{yvsAjf7$4YJP4UIRcKN?fk8n&^dx@D}=a;7cw&*-E@U^h-o)0B25Qqlz
z#|?F9OQh9z<Sj|5G>tQvaXM2PdO<1+Py*R<xZr53_oys2N7#}LJe~pw40HVmIoOuG
zWxE{rf44eyj`a#J>)S`$urFTiq{*;Tt+%(AQ{pGo0khato0RtW2VGkY6Sdx}&3=LM
zXS=tzabno?fn&ItmLoU|E|$Oh;@{4o3uiOeKJ^^YUX=p`?z$IB4!zACk8Xn*(4gLz
zQg<daF^P)92G8F-k`0YN^8p#?Rc5qptnXY0ihfb^h-8e76%wmpvGqh(l!0~s@X;fM
z+nKeLW4dOAw~Zv?NSvTaZL6W=qv6v2cC0*T`mC{ltYTa<x43gS;igg?QFJl4B&Wxj
zHX7|pN{uU9AuD_1W#v=(jhoWuT<+X|xrargCCWve0kmCpa@AP?2I5A73pcN<`1iHK
zZvAQ}+^7Ug%-tRUX*sha3!0k4SG&z+_B$y4T_coNOE|R07mty(n^x#xv7L11h)dXT
z`JDeQs!pv4ZP9BLnvbKXHqWa&*J-z0v^kTWsgbtIf^HZ<k;nPg=_?gRePUH18r9-k
zROII-Dc-qKP+Ykqjv5$%oD;J_feAti{s<prL1_;y#KV35k4FB|kEaop%i*8Oz{4$Z
zIg?b&F*b!4u?^%FEIXn6D?*yiL;IArWhhk6*c!l*O<H=Kw$n1pJ3m&$oekcRU59%3
zR<@F*PPJRbcJ25u{ll1UYtk6De=DLigS#w129^4(#&6fpI|MJxGKU;_eFGok<?@>s
z+8@A^hJ)J-biviXpFja89F?Z65bsQKlEm)=KIBY%C=+#0sc_tQFV{%+<U79YP_5WD
zioO4$gB4hr>!<_cvkwW_gBh_0k@DYBuOxWB1@~ooaD=T>MzPb~YtNADXp}Yv#8GGH
zWpW&j23{48QE3EMC%4!Nf}@C6ui-4MG@Epis?>Ufam#s?SiCA*faBxp4-)3T_hM!2
z1J}~hZf4K}*VWp#2T(2F`?(J4)DB?Q*TyCr$5i4RbePS_X(3&)@F4rNYJwEd84ySG
zpiS4dgtth$BR@3z?jcn0-7uzIOnN|N61y};amnO`iy}>HSGgCuoWfVeg*rzA-=S~f
zh;P@vR4HK<CQZ9Ob|bIgppz?<cC%_F0Bz)&?TtY|<yFzZj*6rijF2vGN~yE_r{`iX
zZeh~Pg)Qvu9K;=!d$ADB?Uz20pKbf^7w;PcnS@SXfyabKgb8(c2QJUne<^z6ZFb<^
z-MB_<CMt$kIn~a@CcpJ+@mS_Z-vg#C+h4O}y4nfS#5P{CE^T!c0|&5u=(ehJ)Tuq|
zN2aY|Gnr7wwm5g#G=u^0g&MVMrD#tOeKksGA8Bt-cY4*O#of72=)>+U!SI6b&o=J(
zIYl2Qr)NeB#i*c$SkEQ7LBX}0ECmOU2@%kYaa2EwqF$)J9Y8299dL7e$FZeoQtlQs
zf8U9qQEhr<;zo{P4-4%;#kM0ui1tm=xgORhv8m!F4uuk*=hw=8w#yd*Fq|m;g*U2y
z1{q#VNkea){>;9ZC=c=Tw(d$NKJB}3kyHG8ziwLCbiLE`&7dbhDV>3jWn&+<Wo%mv
zc-#dE?b7PYD$5>?9~O^yf=4U=okM{}UyqG=$4eZ;vohPf#TM(2N>A%RE~cxUIDm20
z4<Tz!DfOSBk&zY$?=DezW*X;|zdG<7YDMwP@<0Z!eYgkeC{$wJC}`8r6BDxWn9~!6
zjQZJ*+mM2au{95{gk*HGPuJpUGM4KEqp4<BA6^MSF8FpD+6^thLHpne$iS}m5;)A`
zSWQhpyYo<8S(lr3GP`y|it>S9z%5KHe4Wktb>wE5eNO(C3=tPIDi+>U7bu+z5H~D}
z&v71M**A=WO^k?s-oX<fxTG-nQ$nyf;!*JgLQHw^OX$(j2~HL0NtTunMWXZw<e^cn
zHqzG-?mEI+Hlqb3^ZtnR_-n!`ZhbqU=%5qF0MW8=HIqTkjO0|RJPfFEwwNU4@&#I1
z=h|1z{zkHNN$=I*%BrCY-9g>kv)@A3>Vln2^zNaOU}-$6+QOI#ejEstnSRS@2Hj9#
z&pxpgt)cPOo=*i8y8{usJX&&~Le$Um##8NrBxq5`RdCWNJ{W9_E0zRd@f`Vur#VWo
z7@o}#i7{MAOm4H>o}h7PEV0^mSP5B%4_vTZ2{9xUzX1sjUTbIKlA3mRQKvd>t#oh~
zgdM_;f8eH!C1RHq(Kd1r1QeN`)19Cdi@_Efwd>(X2)8<^${B!h3P%V{pV`qo4ot6!
zB-?NOh(L#~9SvDzw)e-Sksaiq_#<N!x?<^OV3`xZ<<ZdngXV{?K_>(5pgNYy?V!;x
zGr%v@0O;j;BZJU|fsL?X&7tN3tyCfAgv|tuU=@WM&kq)MY0)lGk`8=~K*B^#0}J7U
zJyn}jAQpuX?FcIcAvfvKSHMT@=qQTsCpFD2F=K=u$#wU}P04aX*KFYbymmiU2RXl4
zj+nhFM<A6d(|y<hq+^&&&D_lo&_*kooJ6HoglAw+V{WAjpV+7&!}dR#>I4W_XAVc-
zOXd7VEXsNO81%YuT%-5cfOklh8<x?H_ba9eWsaIa!(KQ$W82#0FrEnmc({dk1JN!3
z<lYFDBoN(?umpYsCwFz(G#WYwYsyYuy#17XFA%by`XTJT28fNLA_wu_tWhy$;9Vcb
zYY8$ECc49k)P#)vExE6W13|%+{Pz>kV+#-xrg<q3|JDUt)N16p-+YOQycCKHA;ORx
zOiEP)!EaRr>`I*N0S=K_Q75p{3jrlXXFLckL~}JURTpG(weYb~gjf?J0~<IKC+vBM
zmlZkjjT4dC<F)L$nj+P89k!fe9@yEa6Ps|PorjsJ$O<v>v@6hm2S{u_n+N<pDH3)R
zt#ahCV7w(Eh7GYLFCqCi5^WVYjJFK&!8rMAq@57MALza831f}p0!tNB=%!-V>Q@n#
z#+vU;@V^!>L6JL(V9X-KdR{K?Mk&2?WdJ|MV9PS^h_<fyR2cxHsnPZxepM?$5iJS*
zTQ2Hxf<^nV8vQARMO8J29}+^4K{T>W=wAF`bSuKO5ohQ6lRL0vAI8&fMMk_)-A$n`
z367W=Uxdz}NE^sm@_cTwSUg(MYHjhN<ZAJT@<IbtgoY~%w1{B}0q<$k;vYcafH`na
zFRP>&dP7dVvt5<>_*W03*$XqWKwBw)<P|SsXdUfKjkcO2yL8{#X)PkqR)GFduk!^|
zO*3#1Sd{ir+T+8c!LSAwT-Q-}6qCBJ?lj{EGFF2x-8Lka?d1l@s~gIOMX5PT8XZ88
zbM?Kt0O1LkUEBSXn<I-sQBpyp1&>8(pQ@G?E55?26F3}2)w&==GDy<2fiL)q^++X1
zX4lGU>{Zqs6akdFIeK^2J>U#GqcTZ+LHC`P{6HT>6w`s$AI;T#jYmO-7o-3Wx}IYz
zP+P@b^<dn542iTU?+rj+Yg{WlS<>8mC?y-nsswyX5xgRm5OJO4N!o^5?ic$W#shqD
zL@nE^<;@$J7|0fsvb>NaKHH)iOFX802xqrb3h%n;^?03Co6m>b^v343^jLI+N$p5<
zWMpKg;;c-p0UCZwy{s?-b`j@Ah9*5pUW+3^eqQg_y;t>7DCLKP;0i1QP+$$!GJ(DD
zjy&du3w{Xa6saO@(IgOnI_X}}Mb~sm;3nrvWKgQGP7LM5EIQ$sP%A9~tmi)0oREo*
zi;-aR4gfk=;L`3pz#R(Eh#Kv2I%pv<{kXf2TTO4HPE{qZ6KZzp=3>g)usg=TU)&5v
zt`Jzq?zinWNr{Ogi=`SFs7y^3sZv*OpguFOB92f8L3}i{EM*Gr#SX_Ipcp&sy!a5Z
zgkI>zk?>u39FOAwX5{R$7rC5V1uAYS)9G*{O15|_#iIjbr9`y?v$Ji~*n<-mAw5CK
z=y~iDp(#_7igw!2+(M-|v{NZe7B6j0#gWBcSE_}N5!B`h4e7offdUJ4t>Cw@q~GR9
zD^sRNITdx@;8P=d-0>3r9Ge4FBdmO4O<nR<b_0(=L#3@O`vLT0;b7u^QB@5!jIgwZ
zQqRUWp2#!KK4GI?2<>0|;c@jyn&;RW;2!J#fr~*w`*<Kce5Jz3hdbZy#m=+rC^ijR
z1$a==Po-3O7p|j1I?d^$BwnL|dW3BkO?+&8DUl&b$)W9`DX$sG$KH!7M9jV2NtfpX
zbdNg`n^h!H^O0%>+!vGIF;R98wX@8lM>XQZ0|t-1HBL7lW>o;?7K4~>+veJ*-xFSV
z4tWKg7z7cR&kK!6AWux5yg;ViNJP$UJG`kud)hQS_})wyrc;Y{0ruAuVhe!~e8sbj
z%0+w}Ng|hEz|u<>N$#%&<1`0+t1mFBS@L73xdalFS}W04KcJ*6_6Byq3Rr1JD~#<7
z`{;(>gbo}bXM-v!4m-vP&cX{e!%c6olemBeDgt6*CRcqqjgn0Ip<ojhI8%T;jRfq%
zFd_supH713crefnZlst7tjdp^_C*>|u2K_MDQ<+RtTydRjlfC*tUk7z(aU_g8t276
zNP!rpK@^xJ<8wSsG#bEDs0L0govG4|q?=%wTpRy=Vkh}G5<Iiz;Ql_u4g{i=UGJV}
z+gnq&VjiVmusG}n3^!Cj7hzu&#=T+BC}h+!Z1VhJZ5(XE;1ULR2tovW*4Z==l{V~g
zyFBUARB2nG`wu3i_HpsNrd~--dF8PuxVhmlV;l+NJNDssQkcn0$48S!H9m<D_gsDK
zeRG^wKc)&zQBdbrl$v3-2(WTNw;k$H5A^v_pN*}-DyVP;$NA|_;2@ir9Q|@(^1^}b
z?O#lln9qYOsE6Pjzu&YKn#gHv4}*HnE9Na`io0v2PTvvS&!YJ7J?3#AxUe`hVUl|3
zO<p&`>{2awKmrljza(BQvroduSgPQlyR?NW2$c*MR2boRU+L!pfMcz$Gtm0xO%Xa5
zg$(V`GcHRa{-q%=*|bo~FEII<q*>Lnr&s~kkjD{L4?Ww0^Af2co!CG{L_#ZpUBJYB
z<Do*BRwva5BdZ62VQEj_xm1}9nvIa|`|h6dr28c|Bb04tMLvE8Ojt=;d4Grc^cz~|
zo|m(4EgZG@%ep7~J+;245^$ZVECP?I5TZ?uDEj%(bc-BesP&Tf=NthgoyRGBeajHV
zQ#i~YZwOR!K*Dz6MRrFmYMDd$xCM(vW*+by|Am#GexV!nt(?nsJ~R;WT;zLPIwP^Y
z<`t4Dq+<IB8N6NiYf2=Zts)fvzM-)Iz6kM!c9is=ShUq4$MLndsGRZgiR(J|d79jv
zOFYQLOMo#>nco0L!9sbKoE4Zch!DL?HUuU*vdp>NfK^Vl0BoWAle8n0v1ZsaUCrbu
z`O%U5vAg<x`$up^vMjRS<@<&v&6#ke9}r3dH${=;2OdHYl*l?IM&a;^i5d-n50LQD
zy0)P`p}kJy>aTcq{8@hOYBxC0fU1*<u!P;)OpgZ^*matUcIEp9C(U7q9R%O2KTi3~
zFm;Y(<(?u0q=}fuj_2suM-#%Y+nv~RPkVpJ$W)Ol->`B%m7Kmcdkwztk3WXdf>d+y
zcdW7lwWPNHZV&P+`>vKsGfJo{YDxjpGwS(^*BM9)ck<H@E|@G_sy}{B`GWuaqsafi
zFG62t10I1Z_Ibb_ui?z9VFa4tchr7Lca`BerHk!m)Kft0c3t(kHC;H`4bKQdWM8*p
z@nie%B1~}IQc92+-xSvAEG&-DUD3$Y{qH|l#(TFvzVLr##7GS91cfCM#=Jq&?+5PA
z6EUq~cZE||&ZG(l8TB`r6<FyHV>s#>rMG3=2efbJ&>|hUsx4Ca3;Mj6PDTEjC9<f!
zGaG~;z~n}@LL`pQl=h?Q9t7HOYy9^R<;VD5miE}uqs~a+GKrYz1Xh(Co%+Ye!2c#I
z3uwe2U-=*BH~Hg({a@X&|7oV1|6BwivS$1;{6Di2&PJgCe0WrrA6vd0lU=7%BB5KX
z2)QS-Y!YvxjOK=vaAXuBO2J1FF#=99Ogd;w0csI3q6>X5iDm-nI3M~BT80elvll&1
zLSV^Ya{A!IQ_}egl?B(a7lX<aIpuql?h6*bK&D#Myz$@5!4rknA)Yju7z}p2Ssd+<
zD`b}m1VK}PTKnR^5*S^VYA8D%1(tXa^pg*4&t}Kh5WjejV)W-as6~%Nqbv$5e+uAg
zhg>ILeSf4-KwR08=oc4)PC*?hHuConD7K}VV?W8QDBSi(Ut0w1O8z#N!>QyA<CYo)
ziu<UanCNspbH}_ti=5J@pGD#JC>V*&sEif<kMD1O^JX)2)e1L9Ss??9@12+e_FgIV
z+F0W6xJD_@hPl}$PZze<b*bBfbYypwd;uRu8@N8=#-c{YK>yG8{?ILI#H5dPri;c4
zP8ASzwvQSL{?G!<12TT|LY~6#^LIG6rpdPlPGVvflR_>C*|W(~pI$`BpF#^=)X|A6
z#x7-U`fLQM-a2+;yHQy6&!>)U+^GC;@q%kVEl$**E;>U@-}=u{Dif#t<+reNMPEOk
zWWT83s~J9<(|vvNEE~T!`)>6j`$mx|U-=geC7)XTk88o_7jIC0`0R<{>#3T#E036(
z9Wy=VW~NzYZo9PoUhBPgX`j8kf1+B|@C5^xCQO*1qO^>kYvVQHa~(9azox473(l-Q
zQ$(ZD;_$dViYOSI+_r98c{}p98UqUvd3J?4&K!f4HAL$P%_>!IK1W2{#m;4OV<7F^
zzH{fLN>yIrb~OWG#e7!|G<!9ooo=ms4SCZwXytZA;*_(mW3$RZRJER6i?}vdXk8Za
zTECEuG>{U*>n6Sv5+gw&m0A+)6q40=&Yl|98h=0kA|1}d)vr}fUfvgw#DTA+a8jo_
zdewXj)iRSD!*cNxBJDFiy`2>pk>l4Ybfie6+jR~liK_#eRrG7;{aRgQ=i-~MTrH>Y
zN@T?rLTJaQOvZECd>*J?A;d~y&*w8F4D0WJ99|1=X4CkaS-?D8rf-{!d~jTmiYvT3
zxGj9jt?oiryQhYA$G`3GHswc@kfA|I*sSY#V(Qhvsc{%nrG_`!tzCY2OCncJJ|BMz
zJE@;LefqS0EQ+C_IHb@L0ZKi5{``5SD(wnj-0qdpXP>_N({<W0bT2@BO%bGF4hw_}
z(tdL2S@<m6m{Vu}xR<6k+zGbhmhJrfIo_=^TqC@>M)8OLxVe)zF*?{#^Pph961yF;
zRZR{`1bm0(gs((Sc~*Y9w<$-Se!90RGdTlc-N!5AYwOvkJGgaZkRGKk{x-+(yJ?@{
zwW0y{sd%jA_iB8)*vCzCd-{M)T)T4R$``_gkw=0BDY?{r=Y4%iVyMg)etW}>@!$SC
z12_=Hwyoo3@``T}N$llTt|J@oR!>bCE@<kn2k`<T?HZr%eKj;OGzBQArvVb+xGD2<
zxO}b-=)vv8TlLQKyRF7A{8w!%AyKZSa#>&D$!OlAX0^o;-1dC%*;(Tsu^~Zt6APpQ
z5U{6v=x$8c()!Ot7D$@(7Z#$~W!vr`Cx%~mD$hW;^Xs%`ii&c;d~m?**|Qlz>abk6
zH-M-&BUq=r`$*SHJ_yeYXTDLxkG%n)Cm1rT_u{LnKSKQ+w{O#?P0N7~nEQQkWt0jY
zLPoh$uA9>x2;A+V+Z++MDv%cQ;dH2R_Qtzt0$de{_WI2Nd2C(Pq!svsX}^dh8czf_
zELFW;q(Yk0&lt@jyp~ACJWlv!?jV0R#Y?!HPw>0(r>~Td573!1zE;WS{;ooHH~$HL
z)_ILzm+`-eEcje_s6YKpL}`3o{O2G4xFEPA|Ib{ZGr3dnu@^`+g39#+37D|<0O>J+
z;f$7LtPrxouEoVI0uFPF2E2XXA-f@`ECS6j7b@MG@eg^lB+6$k!2k64Cs4rrYVndK
zOqS<<N}LdqSd(8}n8Zo%E@Lo{gNSxe)y0p9WUgO|2maJ>-1am+X%DmUHi{T~3#8VU
z-e)A}#CBZZw<dmVzaIZnF@X#C)V(S|w}XJvL>c%tbLPxuwfX!m`wQj_N6!_`(uXIA
z9AgAv5rm7g&z!*r{u2LqN-S;`_R)2U!Zs@AM~;!ha4mn}r&+huaMGXfIab<nlWjkP
zLyWrtUP23oZEDImOp#DOH7q0(Z@-xfZz(Lc*=MgK+goJ8jPZB6s;d$&Q?Z7LFMuF9
zEU%G;AOhQqnYT%e8iO9sL2(toc5JHm;6_^A&I7VGb?+nv$PAt@H3!$T>Vj=Xm~X%&
zBCrFT!7PBNsNlpo29ENkyluCPh;G7gQplfWEYe$uTece+O{W@pE!>~H1NQ0|TZ~sN
zqy<Yrtf5{;7MTP+ltj4%deJOBfUZ+)(!wI(6lY*%3=UfEJ^F#RcgRCBN;|6Jt7qe5
zC(d)-HOV*8qR;no@|Q^ya4q+*|I;Z~l51L8FCAA#J~@!m(uuXEKji%$h;De*TF4p=
zmh8%mD<A8H;IT`-uo`${0c?#Cixm{~)}-&ed(^2B3vMgKOkby}kj)D1jfTb0=iC;?
z26RI?up@7jd|3X_lS9ex*HTZ*Km@+y&emN)<c!_A`m$>H`a2HSp;W)Tz3YM=1`|IX
z>ae|quUl-aZ}(Wb_}boz?<!6UZ}XPvF=ldY9{PNpL|*=Oe75Ew!i<2>uiw7CqQwPP
z30IJ3@#RAZ#u6-8q*ut}5*@G2pf6T2@3(8D87W1!iw!S5mOs+K>5P99SG!%#1GEue
z*=>n5N4qBtxW<<-oR`id5=;6U4fJ-O@WLfLH7s41b>&>1S$blUhp@KS9TgU4&r@8<
z`AjY@-G?}3W*=;E*GGM1O@#z>R@!^_?#*gy1C{VNM*VJipZp6*94<kNa|AS&E~vg&
zg2~Usqn^MllBT|O`z0kauYVYeU!>1C+kX8txAZ;RL;4o*=|1&Vs0y0nIcB1?wA4uA
zFi41m`T1Z$ILBj`&kZ2%-LH}Az;oQb%%#7^mOy^8TNmbqhPmDYBF*t0_D^>2cI)ip
zbJCvfSaDjaeb4=C3GM5JN2W$7Bp&KfjX;Z$TwojY%H<gg16E5Z-uc(*&{o<4Pa@8g
zDRUg^L^A;~p!on-e{Qls<}X3l&1Qc}P6Oey8M#a4-DtxU+g^gnZvqA70|Taz4UFx8
zxr8raIN~f)A0}+!ASDTJR~&LEoi+Idi-@YzQa$N=f|4bKJ>!Dc`0mpOsw83lJ1hun
zaUOR+qKfZ^`%t4L7vonCkdLN8hKd>?oegmpRB(|2(<!Ti@`|!Jop*m-M6sn1^A1`|
zcpw=gz!dTL*2zGA+jasY4-S}&oNi}GAo<AAW6P%#a}OBXd>mpcn)Lk8vno*<M<pZa
zEc#sIdMo?JAI;cg3p|=5eZG#>SmgL*ncG+cUC0@Q%GJ!{F$$D>4G+*@aMWS!`zSEV
z3h-KF$2BpE5_5ai+6Lj)QSeuYAg@Fn3g|;OMqTLY>2ZQH;|y<>TNT86Y$Y|WcJXP)
zoN0n$9a#|kxISDtZ7yy@pYDA-u5A0!``2Xt_j_*vO=<f3OE?FdbKZek(syU7rA6C4
z98G#^RnRF7!jQDPZNBJSJm&BLrT?5k(M(G&#~_wxASi2*cqI6_x9Be0CR{q3_hEcR
zo4Gk)Mc>`Cea0Uv`UPwB=2RB*(cy2i)22<MSOE|E4mD<=TXm`g?}E~jTus{`K>FE;
zS<@-5yKNV8#1SW<PBFDU%u)lDDNBZ*Ls3Z4_Cp{|!pY+&?~S{?#vkL<JU+%2|NR*E
zi=wE+gLOF*>vM<4W8A!)I@m^X;);(!Q-DHnv&$we_VJ~yL@jwO<C;(&;!qCWQ?j@w
z?242WRma@<f&*5;?S_V!O7RBgC)}-|$D0Ra^INLT!yah3^^ITn9O1${|Jy^?s*BEW
z-K%!IgZN@@$MsD(WI~@jsu7}3V|!s0HT8H5nm!~TKF_>@VceWI?5fXYAN|f#R<PVd
znX+W(j|<QPTz)_J$5pp*<356`Zv1~-br|Iqw+vFq29B|%KMVD&%*I;zXoXcy4zLb{
zNrbeQ>V!dVWXxE=N)64%6a`6IopsW0n}l0g;nHDH=~nF9&eaO`nV%ld-Q$mE1!!hU
z7L6o&@oYHe9D*HLMbt<Y?QK#Iqte49ZsA03|9-}_X^=m2LQ<iihRNwMsAih*VnxMQ
z&fdFEj^Dc!;oj{|Pb~T0x8QP~;=0~*d!d7O%t^=5&}6lt(w59KjA)yo)oqm39MVBQ
ztlkZI<|e72#2(Ap$Rjv9A~xK?$vhQ)yc=Ri?{8oowBW?V!@yxg)@4tIkA(iV&|=Cn
zsKeQ@IvY!hJo=e^y{%<<z_#s3-y@D)+?MXSY17y+bD$@r2>s`<Fonn&p6!&x>a@;T
zanSJ0$o&WW^($@6C;1wt3w6%JC{}F?(%(Ynz~gi?o-<<!P)h@a&`!zEXAEh`s8S*u
zG+foV);Ek9E#$moBv%(zd0`(F)KD-PTK=kvXFrh7?lF}`<2&fmn#2j463ra-_yMo9
zi?s-g-ti#H7r~N@Ql<z8XK2q$*wdLX7!pA5W7|$1WiW|U26W?F0yvGUQY}BN2BOJ&
zgdWNT+C9Hkcy#2R@PQICQF$1qZ~1}6-#U9?5qR9dgaQ^#49lb5!C+OxOaw;iSR(BX
z>31e=Lx=`Vrx-`4I2J-i=EIhElXIpV{&o|2xh9YVfvCp{IrAD!zV1kUtt&P(L5n4{
zuJ<!5R&@fV(kT`sUE#}KJ9}^R6xL09N}IW%)-Ne$i9#`CwO6Qdw=nl4G|U|~j9EzA
z++DhlvUw=tsg=Q?8biac_+`#!;0`QgePS?k26|XV<+Bo^0R;iS-Pb{WQ#=#)(!mI^
zki^Jb;B#QSPa{ArL0m=;`*E;~tQUFOEMB_QpU4R{SX7;UfO)~TT(uR+e#n19hx+Nw
zCnr7m_ctFa)b>9y(}*F9cH$$D8VPwA3|%SThZuL}gz7!76zA9pGTBrb_`QjMt)M*P
z?D!^}bzL62;8}}LOD|xv(1Bb@aKfYs2Drcw9tdJ8-s=@8Sf*^J|Mak@8e(%%3b{)6
zjo)p9MCYZ8H{8FLC!Ay}=m~M}0=Mo9K&t4_$=*oFpG?(`VKaAue2MeZzQWF}GIQ@}
z1nV(C2AEiJxo7qG0(FW&5mKR$ypap^Ukc+J$SsH-m|Z|qC_oY+dONF?F)$FbYyzN3
zp?U^5Wcw#HL37&VKIl~#zF}W%9L~6|lOuJnS$_ank+q9%5;k9@FkIB{JvNQM^HPP!
zZg?C;+G=6nKq<YPBP=LEK@jp{IXNFZZj423gw69DPFo@Qo1NK?>XW#kf?9(`3qU17
zICB4;*D~hqPz($aOEL5%3~WH{pQ#dqQmHT>lRuAx^R~v7+2t=6Nn!W#(V<r2Rv8c?
z7NKUD%XkxHf^0(RmhFr?xDhEHstndf#q3AZE(~enH1kr(iNZTM>*Rxc;UTPS$(%Ja
z=yl)3@p@5Vi%DyYfrvYYU!&ST)~xn7eAZJQP0uz?^{1PxdaubN_7>RrbEmNTyRs(&
zRrh=(i{kHB6y2|EPg}_A^XX4wnQ*KB`0M_40*dqLuj{_}&p(R%=OSP={ExZ_tOwS4
z(V_Sosxv@@BJFqRL`MS}+;a&f=Nb&H-MvfXO4kUIuJ)nK$(^#$Ht%Ss7|4*vzysHb
z58z;?t}-Mb2g5Sx)He6~i^vrZZ1&T!jqD-rcnlAZ$#YiqbWM)G3=qnqY1;FHdM;7y
z&d>-wAjKxH;rV8|6M#)ss)m@2aajBd5yN8O27Bo>k`b3g_zIJ&vgS+^S=YzGQkbk0
z+3}4?yLdPz)&19>ho^X|{PE}1-^O9>Pk;Z<KjK3Ce|iy4t#w66h1UA@KUFyiE5*nB
zJ^I*-C&uz$!9p3;@od3D^jELC4mV)I4PqurbjPz(Vlp{)FPm33FLa-5IHOsAa%7Z`
z=OWwFXAPWRd|Tr1(9Ct-sk3BU)=3D)5|k6IM@KLbbUr)YDN@(Iwm^qej}yQJTtFdB
zNK91(^u7u9k9~0D1nvt^l{q;!^12vJXh@YO^M-&Qx{lxwn+>zDi^2D0W2=BT4W?uA
zy5LZT>Q_it2FFBN5zx^*mJaxNWnz_<KTK@?I(PDr?w-%rtt^R1RH!)J^Ml7%7ZiHL
zdLrt-xUv_Ng5gvs^DJ5(Y;ynkud>Z*ixw1ffEQFIHO^y<N+*rs-bIhduVG(4c*TH9
z9GiLy0d#V-txY=+G{%ezvY-1<uL*j0WD<UbfWZUkp%wunVQ0g-Ul_c^PX(!FAz`57
zFgq=TG9@|(rd--{I{S%h?17u`vZwCkayD3+gFj7WA}sa6$5BEw0kH8wLbf)>c+<(l
zgcX(@P3$m)7(W0m#Q-@Di>qDGZeCfD&7gSZOh9@TfUFC70&%OMM!ODIijL;zb)!m4
z#e2k(QJ5#5i}JBFo>VH?zJ6UA?&+CB-XnE#nrqsenaO9IwlL?jgFCTB{Bspsjh2(A
z%pvVf+jv?xEG(=L+Pmz!d=z{I5w<ZA-5=2XBz=2wFoCju=6?*r$`{oWZeQoYfhZU<
z5(ri6jkVxRTtvHKA+rl>T$GEeU0imu{}%0YYmrlc8ImZlsOB5Q+#aLvrtOFxtG@q5
z#WvS@uD`;8E)4qt`qG5ZC>u8D40)r1xT-q1dR@ap+iM!Z%0UkH>xb5>FV=LQoOq~N
zWyPPlR5x;|svpCJDaIz6*)R&~MH^sJW@Zt5KaCJ%J}$Lvj%Dl!nHiMJsTnH$fNfX*
zcICV5y0Hr5ZKiFrH=7_<$Za`_@jJ*}DD8X=Oov)_LA?u6>@4({^mx&(!FWz0O{v9D
zD6M2HBQq=P*LW@{d+7A=SdSP`@TXdeNfTPW>|1ap?LzgluO2#yh)p3$*7p+4lr^Rl
zi3a#1yUL*>6jz*s%x*r>9_33n!XT6qmKE%kke>VIviUB}*l71R5o{$Ky)1X`oZ}6w
ztdY?ZR%64}oI3T%s=9)zZ03?RAxJ8aN<vyNQd^N`AkD0?*0c+Mu8?${CpOW2vPi@Q
z<sdy=%Z=)PNt}cWxri5Xy>KD>*|Z$*_=&-Byh=CI+;FHMRsLR3rE1daS9snz;2}65
z2f@fR(Q%LgY@KWQd<^xp2y_CbXrOtxquU{l4(2&(!p?q8aIk6L2K@tunj-YUxB6TF
zXEG7C8|y)d32{C3WK<#cZ$f{)G-vGbzOIqY=wwB7wAPy_m-)f6b7ynogr3Oyvu8J8
zBams<Z}v4e95tN!r*?d($v)f&NYfeyf(msK?b3uhI`hE}xVw;2VC%J~J)J`B3}6Ej
zi-lqI2%G27-zge{|N0F<K=d~-f%eP!->zfc73Ig5AV7~oz#1%LEt;1J1Cx#H>_3Fd
ziv8tscdy+!_%>%CtWzQ99kG8{wjb#B)+5cIuFKRwGf39~5$xSz{b#XRxs!8^kkd|n
z8*Tkc+THFz_uj<jju(+}czu1%DkA$gJk?4|9A80-LDh!U99e?CGAx@0r_|h*l_HbB
zOK?2(MHj%{z*|q%6tG*wV#<#J1FOl{bM!Hvyt!quTNX8<X7GdIO`CQ_Sh3Hc9b(rb
zcnaxRoK$x0qykzL=6Nx%fUa(i4W=!g!H^6w+8ih$4zA!35zy;y_Ko-VAPmeB7NX_&
zitqz?X}URh9x(b*m;(V9LCp60X(}VC=(eb}=inG?c*W0S160NM!)w^SG7xeiM5b=j
z{CK3O?5fNX<UBI!?N=K4%@#UnFqi>5kyykD3z#G=nLo^+2H|BP1Lyn1Jh^iOgK2uZ
z)1I7S4b7uU<|u>7Xp!P<FM8~lhRHpj-$tVb0GTB7Uc##a;>#|V;|6XYdac2HeF`<W
z&x45<2H7x7W@xTWHMG}_l4FC~XZB&^I1gic0bD&sZEF$X>_g!pEPY{BdH8Ip-;Uu*
zA_yu@<;$^FKyOP2cA@9`$eAjMJ!=8EG45!wynJomJFpqPG{GsjR7tx=mX<il%jP6q
zs^9D!qQ^a^k#ji#2(M0qGj267#3o?WC#fkfstT>Xpx3@WSJpN;Dk{m^RYujA>t|};
zmE@2*ldaAnm+vK555-h#J#=b!bq-O$6aU46&(^5|3igna3U($PowiY*_^YL094>8Q
zs;_HtcxdP%w2P4>$Vywqruo1Kwg}LOiJiV^Dji1!s4G5(_RgL)O9o*-u~lLSn|WK+
zOQA;GM-PTswp2wSN#tlT^3^iStF7xgA?Vz?=Inmw+?I=4*s9CT(}8mOK#iry@fk-R
z)E-ZVmx!6eTiCEuWwPTfY@~V^<8iOt-J$95dxNpTV$JFFOPZJcdgZ$*x_~OL==N6M
z<^Y_ZcOFB!tNfG0zf4|b(st`Lwn+|KkdJZzxv1l@2a}J?yFXx<ngR1mO?~Y!-N6T&
zu$A2tZ)l;Uqgz43H4F>Sd&^}eLlPKIRpFb0*HMqEn9kM;5IY43484dm#o&a0?GczX
z7x{mL=IUmKTRu3wxUv3wa8;;eJG*flO!<I2sx(#q?z-2bM{|z@FS_{L6Y7)2t6pRG
z36Zd<9<vIsGYPCu=oxHTSyJLOdMKo>Vrj{!ffLTw&AnAO?#TotO7;b}n<LdDKK55E
zy5<!{$yKMP|Fv&{YDkpaT-6t5&H7_GB@yM<XNAT9ms-(sg2hFvnpWSHLyqn1T_YV^
zSN2>^E~&=vx=`nNzp|xX29*m<I-S-}Z@)5*7!Iwg_`&0^>%@=bV}o(3_`yo)KyB!K
z-s!0)fB){b_f9nc&<g%`bt){F=zxB)hhRzAEU}(c!>mq+i(UOVRpZrd4enJ##BZ0?
z$Ou8YK}<ai>>Cfi&*C;i?}@zn57_i&IIkkK*V$r_5k2L;H`JIQHL(f9?e>Sg(J{RV
z1J(93BbZgQ=U2;!nJL(t!nwU=X;@9O64haxY}}dK>83jphhbce>A(DXxGG29YY)zB
zADYi=%N&p)YBA}cTS?A?zSN0LjGc&ow>|;Pn(?o-V1bjLyQL(jt+D~`H_~ElhfE1;
zSajw-QN@<ydKd#i*0g1~eH?SkHK@ch8EA#asXpO@Oit1o`((K_L5Z5~22$HK+odZ_
zB~O-I0&kA{VJ!15D@y0-%&%>SJsc8ML(C05+C7Z3Jq(8o)VDYrSEgpxZ{B09GIX(e
z(RA0y{T90)J5BfTjm(@hapH7lWL-P<?v)SrX=$*5LO3sduRc6BTq;e?q+h{E;a8Qq
zEZ4v*F<hi#{lKb;B4Sa1sf&P43qkHkfz{+0M(7D?xy6LtXk@`ra=fFP8pIDM<!+XX
z5%B}2yoSDLAtV^ovo*Uso^03!>{;=&R1rV3bSV_wbt1s60sP`un9f>KRZPJQ@YO*K
zd!0+Zg*jU|1+2lSa6Xb?w4rt&tU9S}zymkJh56Wws3lU5(K%$*9!IZP#VA#D#T`|2
z`!z7KF=eK=fmZ<2TvsD5njU4q;;2toe{!-3`IoGp;r`{(;+mrk%yZX#gj?e)Rwixk
z6`|no5`O-OT<@}4G3HYq8!1Wu2>d+&@AH6DG~M?)ve7dcvXIrvj)%}K?Zu!xv56F-
zW4P4cjU2=PR1vb`+W;9dF?sT;dc)33yu7R<+bfp7Vwx1bE$!R47dB4u0SQH<SIAgE
z;G8TPcHyQ_1;9Xj7%dHkhYYkc%Lc&+rKsC(e2+sr?8I75P{noMd{_e5EI({&Ac7)0
znHRtLaFwWQWJUAJwxGH=nH;%o&M&@s=#Xe2Z2P)_QF1D(#_p~OwO40_YF2!gj8g-v
z)V1ewMUPy%&xN`;qiWYk6BH4QZYSf<g#mQ7u&3pbpaj8q$ED0`LsfR{1Gd)y=<#<r
zOwOIWF6NQlw$<yzO(fGkgtmocHjR1DP^;`bgbK*KYmnBfQU_Z7y5Z87R<}ixfe*#8
zYerE(8hCgb>{0)hL!HLE%X$9Q_&}vz8%KC@#<$Yi$F)-cYAf>nC2aekST55nI31Ko
z`ieJ@E64^Q<j}SEhDTafx2gwf(b9%w2e92~?~_F^5yv98ZVRT(skLQfxlQ;7&(5}d
zW7MQ+<(5fl3%)_KMaQW8{JSu@=R~IDnSDinU{?NatLOa;sBJS(N|P{*rj4B;!zOpC
zxT9yeorLkptj#M`r*8D_@DnHO-##$d#rH)47@q-+Wopkby-^BL>0nP(PTC|SCw7>X
zYM@6hEp&yo|A5UaIN$Q??{T!$nc~X6vFetUrVDd)G*9Ll6gIC+T9s?yT-h9!*`c>5
zWJ^V$hmKq24+@$#O}aJH+SivnjA+&u4N@+NFmjr_)YfQ{uGQ{*BkUdau93;O`Lplx
zG#}qNj%4Qr*LkL1Q?N0!sYAHMbhO}BVb%Y=Go*!1e)g5<%$WtKyJuiq)AM)?eHu%L
z(W+w3_Y2Ek>WN6lBFwjqIPw9U5w_p%Knuxmw1X$2%3!3t+gFBcXg|r1ziEfXyg=I{
zBw{1wU~g?lMZt;t$ayQoX)_aAmE<EINk_1CMjiO5%8T)vm!Eq}*Tr&@Zy~r0DNui3
zuEp8G=;b&+4&r30f}_`3Xr2X_qRTjO8lGnzwQLlewr;6<in)wAz11)Z1L+dhYJRO&
zs6JV~w;e~I76bVGImHcg0)^~zc<i(x;SP>rm_;$p+GOz;+fYNXpl1O~hbr|t!4h{t
z#BhEddO`RMizJbE7}v`5KQh3aFU=MWZq%7nke4gG_`n<~EEm&{B^9Rt>C1@UMG606
zRA!8AJ_iU8mp3)f2g4@XWrfX+Z^-m=%GV=kbRVg|!WN#X^bR%9o~~}-GMP$;b}z`$
zL5ZdztPI!4v$}gWIvGtZRaZ40#xe-OGH~6rbg_5hm!{G6GOD=-9?dGEk5CP4a5jF8
z6MYDz);=7OP!n!5X<Liu^(3h9XCmSrzg=aKtQ$CQxphI6X86U6MTBJf#i(NC$ynaR
zC_z`vo!g)=0P+7~&y5GeZ>A2CC-6@-p7+kp1}({4a0y*i#iAeA#p-8kYV$~E=Vl=}
z#&K}$6jOI%0HT%C{rfd&2?<ba?YMV26a41H0n~+?af5E<RiUq#1Nf)==!hozk~^r3
zlw-!Qx9bfLoCKfXhc4-BzMYrABsj1#0|AD`oUF(kXbFEY)2OGQ5`JJdhN(gf*Ag0h
zLMp=OD@144!%67E@NRX2R^HiltC;=-Pw|;zh(Xs7j^R8A3JJKezI4iIAS2n00#+Tm
zANjTKa3RyH4wq(wu(|mp3!UN|gw7t#*=o=dS@_OLdgSg;{n_{Z9XD-iYd?r0A-5Rf
zhO*LoxiUFA2F@W8xdyJT^KuO~N~d34ne=5#Oq86uYeI9gzH5Tzb^(py24LW9tm<yy
zIut6@;ug*!*n*vW8KZzkxm-ir`SS~}fz7N~M^Tb_F}cVV3A2_jPf^B-ra8WtLAVU=
zNDZvU0S(r6GoC`T-FZi9hFfugmyZs3kL7HxO7b626MubA?I(O~8i`W#R-2K-?}00a
zk8B=-An)>y`O~J6;Kew*)WDFB<^STqc1%V|t%Cw5s$a}B=&kY#loWA=k@}Rw)wlLL
z!1CUC)XXd?8D;*GwwEAybMBUwR?|5P=e@p17c@7?^Yz$h;@XM^#1D_$#?Y6}esu_g
zbprEI<p<A7nyT7-f#J3C^GEQSyW+2HHQ0rsc1NoM=uCPTIu(A8mw)(XxLu;J1l2^T
z`qQdqTeG8W@nG(XSXGS;w8W`-Wkhh@LnkqL_sLm0>Wc?>q4i6!bLL+A_Y<%ObykR_
zMvKZc*GQ+zzWM_eT;6~1v1EyP4iUegUzs#0fGFdbd_?PTn)Zu5+y3j8gf;x!?f}|Y
ze&CM0=6rD~!6et<#C8otbav(Kvefh`NV84#-jii-N-4hto<Kk6^`=2M+ESFqu#1FN
zxz~)&_qz2@6M#8vUut``L-Qvz!P=a1#5vo^V(plDa$S@Sc*0qXh}kix7)v!-{T;~r
zMw}?y)MApqGOZaz#HVq~aY3Pm=HSe}l8FbV)s02(!hUrR&yA==9h4`XzQ5Gf_V<3s
z0vu=7Cci;Ny2jzX0S)uqeVK#Kr-MC8ca!AU>b7nqkon=si_bclT;`tbtn%_JM+D+R
zw726;L%;e#bh<m<JW<Itz%dk9uC~oNBnr)nORmAm;A+Fjitk*vJ$=93ytO1EI7Hc0
zL(w*Qt^hJ6+6%W!Bl-P&F8p9F{NSWVxlgK5fX`qd>Iv3T7@`0&LNNCepFmwMxW=ys
zsL4wpP4*I)=hCRf4czxWzwl0eOjkn0{RRptJMo;(sh^iP8<+&K|L@$F@4hDOT)Ol9
zuy+)&zaDut&=_X6$LHKR4s!v=Ig6AkrYjXdqWKGTCD&k2<d$HcZv5h19~n`)z^Pd4
z#y(xd0tv7|*q&b(HEh_4Ef7QZx~nu1pksHpHG0b;`H#KTu3jUB`*YSg;A{S=s?hNW
znF2VzTKczP8UUd(48PG1Yt`>_dI$U=tGw*n<kMy7zr_$CzqKQLv(-Fijb-5UlonUO
zp8?>Fg*t&8^BVjrbE&xG$-%u@Wlt`aIy=+?Mq?V=HT2CE9EGur^Bb9vLv7&Cy~5uR
z?j7IZ?RCD;*Q8}E`K&Yc*d+t2EzIuM9*sG4a!R{0?<2OWgyxl2L3JzE_K4Ysco=r^
z0g4D#_S*E+WJJK(4J#iy839Cyse9-U{3u*FcEod^Iz719uKTY#F*Iq=jm%?JkD<GS
z$JrZTsiYiH<WY1u9x&Bn)8JU$O`1Q-8t)3~j|6*U(dB|JytcK<jFMhF7%Lbm>g<g#
z(qZiaT8mLTB$I`CHoRgDQ-5wtOd1#_f6hfg*#u#D&Q6+zks}H_Zf@zg7*g`!<AYR3
z?||&3-#~|uTb~VOP-@+(Lr-=DIqqH`mlu3Lr*bGNE(Q84bTej-Io?3;TuP+FO&p%0
z^O0)XsdlCc2^ovT>%ik<Z@8x0{TAnxPI0o)Uo6pIgjJ>bnxyY?N_zY54Js*x4~2RV
z6kDR+N4zBpj~C|N04sEgkzNr~HQ#aU*b<AJ-unKD%iBiF<I+M{y9d?n*LCg{%{BOC
z;U8y(&eKpip?}ZXHh$Lo*UR8hYsvM$CzI=0;vm($vNEWnJ|T}b`p~AfJ^$GAX<ss2
znB4Q^W%38tWw8uwhA2doW!?P4kC%(ik|gV*eBkjp)M!g-K4aj<6se_$kT0>_(sDir
z#G~=SCp9m|@esOo+S165$~dP(%0pR~6|Zr|1WMkVI&Uh70Oswv?ofT!Bxh1esp((r
zjq>sf(sbxgEA2MP3DH*p37L6&)Qe$r<r(|ESH0YWK{438i<X)`!Ul^r&_Z(d%$bfk
zTfL-nlCXQYx<)oPuT<ISY&-y>VwGz`X-R~<tgY_WrzXnH`eEZF298a9uZ#+IkDkky
z@fTHMZup$i;$z;O4JZ6eT86&-{mY$(T87D{iW7})FTS=y%%bqz=pA`^)1(~h4V!#d
z*7{~|pYEq(ebrRUUuC9+p@YKYD_QoN{=8&ocllyPDfzzX$qqr0X_h969w)uO%J#df
zyylkj-CJu$h8D^G8l@WE`i<>?r@MREiOkN2ighBPE2A?%OPrm1uLlv^1ML8C);q8c
zo0ONoVGVWQz=_V|OyR%;R;lD3?6(z|nV1yNZ%7?qkbFa*{B4}-32X0rqHHM!Y`zcb
zrEW%IabG}%q38amG1TeDq65r+>o{0urhF7*AmWQGZ&+*o)BxgGESv$U9b?TmQv^lT
zMQNSs7=TpfZWR$TT9)vu)Qf><I7+VFm${{(eOm75Mmw!HWBTXMF+C(&qfY&_wi7hi
z*KfzhkJMbrL-O5?$1Kk5tma0fqP9)&+g}a^w|#oy6SIu2j!c98{gLm5|Loo$`9?eK
zYQ;nSnvjTc;Xg-%BE({{uasX^U%X2A&dVh#ZDG1;J{O$Yuf42Ec)5SxsqzG{41H4J
zXMKR9#y>s=>2gtDzh|_O;<nSR!@4)DYbeqJdL$$Z151}c?&>)Twm?Sh%Gw&d$OWCs
z0nhmDUgi|fTIOcjkAgDctl9sXo{IL5pD)YIs<KbT>&$}yrr`Ir0&Di$_Vucx(2pg$
z451NM9^LrhD>4CiT2rz;Axos7Dyl!MP1IggR5WYl4jZwoD`@sMooLj2;Z6Z>`q^?+
z(6>G-i#xy^KaiDiA3cG(yA8jc_j+(MT7^oF4|t*5v1HG)oXZp`Sm!Jt(_WTf1Rb9^
z#Nv6}5+R8=h+4PETCd&NCDD$?(T*W8j<?P`zWDG^B~&az?wFFXo074I(hEf;<6mCP
zGEuwzWy@xMXNYc^PP&P!!S{|YX5KQ`eIJGVT<>EX8nvraJNvW6PHHxv8!o$uPP#9T
zQ#MLJ{8cu}u*u^@Kq57#HwVU$R$+7Bc^?zs1%eKZ0~;`trB*oVw|eKFxs9VISu+O&
z0EcPZ6#G;Au=vbjH8rR=lSe-w_KWc@?IvdXzC|8V!pFBDq>=}kz)0l>TwI!>U1F)d
z>^D%C%s{f!jD<$Ms8bT9(Cyf03@THtG}^+82UNXsu!eOFNjf`kCS0+vU}o!vP_wVl
z<nO61W7V$ab+PN++m7<xO(<;+K|-sX;`QSrwsWCF$Pub=uZYG$yUgbMNT9q$IO-#g
zeoQ!O74Fm{Tla8f#KV<$AFg~7Xt^;dD56JJxkNwWq5j>6`h{1QCLTFBwKT>tB>zo}
zQ}gBby3#03R@T@%*1Q3a{&dJ3mAQFQnaw;AsUxLdT#0#nSw$gb^Swnnjd*S2N>9cL
z!HtpP(x7ftNr@Tvz0{G{8T+(ZVC^!w_ozD7<Is}y@8;n!F8zS`p7Tb<G20T3VPqSV
zz;9+pf>0cPxIcL$lZOP0LvqS;DluL>82ud#Px?TUBzAKiQZx5kod~$l?Z)|9wH;pA
zFXqlmwij!;@^v(vYAT9gbxu&P4o$gbhXym^nLXXO=Jn84rrR?8$FJiZW?6HBKvM-k
zRkdhsHI7plsX^0j*z|CzPH+K;hM1#w9rJGi6REiKfgdtsTG)FqiIF{$>lxR@)Iqn}
zYEJ4<+EAd{8a0DT_1ZpRJ<Bbrkx{OZ33I%qclgoD$q^2|I>KIYMq1fMMmat>0(-^n
z3ES6nVyczX#1z#7^KxbG8<na|%QTSjSE=JP2x_r~Xo&Wxk5HJ|%>Q)+s(u~!d7ET1
zC$F>XjhuhVqZ@>tELCsj)sb*plxn(Ss?}V?@0j^_NT5WM>jj9A<TeI6^w`#IZJ5@q
zVWSbgePa8In8{0!0EI$$yU@s0#zLqaUYXJo1Fla^Gs)_=jzstyI6-a46U6v_GYUAY
zDk-~YLBoq~4g>jazBf`o?Y#FKb^!~X&`^f@?mCJ{kRyJKEV3>R5NCbOB9NU%=eY?4
zrHwW>EIv!!e2eXYZ5$*HqNx~d=qx0<WkAx$3=4)saT-uD8td5*%?gSLlo|8lK2tKI
z!Dd9hoRT3Jz-Mh!?=zvE4@`uRp%w*`0*63W+<AgFEzIrn-!T7kXH(A@LT(<V9Z#m%
zGy&CJrXvD-*y_Bw9h;K=K&|E2M_`QM-I}`>UVerB=)dmY!n#nU+_Odtl|?T;4vlym
zs(T<fqCIkalRuGZuvewnabflT9rC#{8?x<uYh<jeH!frH5kD6Lu&}>ppd^5`bmLpB
zIo2Ws#F$gb0c0+V`hWx$?~*jbzUA^)5jk*-(OO!yBxDK|ms72;YXFMm;hj-jrHh%P
z)Javk03Qv>Yyhj4sj*pW5$|^o35SLQD40u79T{hx#FHCL4^QVX{+}+;SF@~9Ff&n{
zN?}(W3*SyTKp}g#RJ4AbE1`wd-^IhouKGza9i$jAwqOW4W&7R7a0-{ed~4K>fVmud
z%bW-K9k|M}HuvdLXgbN1qBziel@Go{-TO@nrnab&pUK3IbmQh2bNQOw-}YnAlm;H9
zOEo<YdzG}q!0B#Oc%xpl&3c!|gja_T>TVb+)O<TOz4~735k!C5)}uE|e=@j6xvs+n
zw`|ut*+@GSZgznxxRtl-09(p*9%Z;|Z6|)SdDBXIP1cunz+?4v&<n)i!p>|I*GGz+
zIym4^Z|Hth*%2rOY}_6Lx0HsoH_=6shr~~1%O!!Yr}W)J=W%}bf#5|t%t}tEs6@@!
z40O!^IBk&2UGy5V=<{zL-9wPI?~?z=2a0vgo3n$QCa%xc(ZqUq@hn&7g`07B8z&_9
zT{4iXUZnHl;IgC?R72+ZFR;o*vx+S1P6un9cvjsJ9FZInu|MHRe((N-v7iVeyGh9n
zaJnhQ3}DHh-)$}0xxTVwWp36t0i7ZuQ(jOD5ag6>i~V9%e5gaPR(8oz`z19As3s<?
zxq($z?J-uLDrXa!@NCbYp8e&?O2i!WF=Q(PjRa?iec<E~8WHKa0a@S!YFIiAOegj9
zo*>ap>*|qx&YmNo!Im4mw&w8m$`4pU9Mb@B<3ItXHj=T^7OX&+MMF}>^4p&#a1{L@
zX0TjxYoKWlLs_oapQMO{Kr@!U*|X}qORXNPMT(v1m*PhZ1*QQz)Z%5!0?C@7Bx%C)
zdX52wOb2Fu^Aco;+t+dCJ2K%W?$A@DvnMoY61h`gWy&1;3aR)sjH)?Dl7~YAIqob9
z%Q6G4yqad=ZprHHn5KuFX|8E6QrrR+((Xu+VPivm4_VyZ#69`q_+i#%g<vH10ab7T
z7h1@t7_SkuNGJMf(uD^|(Whc5s|5uT?TRb6Or|S(xbkEK#~102OuTJraOs<`!4yj3
zs2a*cYsh)2OODR%OQ2;8l+Zn^_?6H`+-bRJk!gMmJdw)i2V#}o4ew5_N>{slix(Mz
zx7$_n(cKej7KBp>+vcz`8m6$5ORybYf;bO$*8i)$tBq>%Ji{-QSp{jWMJrL59~=Tl
zr)>?++M=Kahz5j4uu#wq0s;X&Vxd}{<yb!`--T!kD1s;;#xNpa1*L0Om=tXN2!<HJ
zRpO}y#HmM_c(A*!cOCV|{_M~8XYVgaa*{Vsp67n<=en=!x;flzSrKBXM-MM*KNn%+
z(|E<Inrhk?7Pc)kf0FciK~jR7MLQ4S%$2)Pj9=_G%Gm0{4&$qJzuG_Lj!%7o_!=qV
z*-9~<>Dfgxcgb*EwkbIP$!l!sR_Cv5B$XAf-dCxndTa+4)ke(x;gxbgY?}?p6vH*u
zI-Rod{ktU(cS@C?<xxo#6IO9)4Azq#zUR6}hc>=<I1w#d>jPJtAHmRI8{8zX44$6Z
z+)s6QWzoI~PWkLZ7&g-FX-v^#8R7+%S+05g+Ch&8d#C%%P~=+?;&rvnfIwr;<?isE
z=%4Re2&q!&ha2i6bE&ms1>)-CH{+qGhqmIhD;vAD_R6fcZ4?HLZ60@^JF2^8B`~5u
zatYKpPdR;evdqu#S`vb*0!+UC_co$jwwivG4{NChk_;_CHRXdhzi3QYIXgz$rtjZ?
z_>$5bXsOqH+L`XZJ)vVL#q9Q5Lw2Id-EiTj3glEpfA%;35&*<;bMknX<DOq=CyyPz
z;kqE+J|VH_vcFkz#LItsUSh!=ZBoN0n&h?k2^8vLr7F68;hw1;;@;l_Qq92}i`EdJ
zikoi}w>wMg;62hsDh10)?91w8V)P=BfPfJjl!iW`4?F2Ay&up~1?i&N>so7x$V~yj
zkldqhJ_OHA(P|O&i*Vy&PAID6uwqS(gQUmeEZ08V_Sc#dHT~|hqKLFp<$^F}&{qOn
zAZ?5q=yGU@!#1rwjD@D^y${LO!gm}1Eo9=9U_E!3mYSFj(O{)eQF*YzncO(!AYzA4
z<Z(Uqj&EbLisGKh<_;)wCSQBRj;Q>-01HiK9<$>?@eSP;XY<oUV~hNiNJnH2Z;ZT!
zVE!ALi-YHPgP>4Y(kVQ|Vn_WglsL0kHtCO6odSoDWn9?K$S5><n93f9EE=;eApVFe
z@gr9JO+4fEsf%07UsD$kcX&F;^<`_?lzXHLj!74UNEdjz94kQvxBrDZK3H9QOxF$B
z5vrztW^e%#O@rAhoJ5F&4;pBfNyRqRt$xlaYpz<_777=ZcZYv1%&_N^q$J)0DE$|@
z9zYrhC2a=O!kJ_*?495=iA|WxbC)cx97(;~<MmB@doSKNi!2x<`%`#41XjZqAJND%
zSWR4H&0F5LhWoatV<e!9dVl%Zv!X3&S7VCi%XR$?G>wL;_Cy6w4istt)7vass^bAa
zq;I|K*M|~73)rKdU0Hg&&qB8wu@DW|vKGaSt=Pe%u~argMB@u74^<m9d);M>SZ@km
zv4XM^hUazbT!^wDSbg5tM6EX|PZVF>Ml|NM-TdtvKgf!FtGh`iBsq-yy_m-+4v7PT
zf>A?{h6M*I)cF;64Q!%11(vE?v1N(HTU3u*aK7keGjS|y-+eDQ(~6vZ8x~P?hrw}3
zE_p1$q$4+rj;hQaNfASc-t;v#o9U?c(OOg_mEy=sg0K9irz6oY$i9SZNta8z`MSfA
zDV-sE7G>8Fe?y%%ms%EZerl;EhYoAaxk;z^$@qecemq(b#gaen@`vl`UhfZHC(=}Z
zi>g`*L-s7*9}9NQP7;Y&;uec*Jnj=i#L$ZF_^g+6znl(V?9Etb%+UIbacACDH6ptI
zi0CdxZgp{YMxjQsD`L5ah({Qt4D)Turg-}!L3Pg`lUj~SEowh~o6d2Icz8AhkOGGd
zHH)E_GYcRCIk}X`iz*$`SjitFUJUSkDFV9-8`F>g1RVK5=W^%uQi~vaxy9wgw9;CS
zTc&$gD;t|vPIzQpTwRPbv9k5X+NDp--(^Hf90rg@oxB&$=H|7l=A;8`F>T$6Lx4wW
zyeExrPu8TI(COx;79Rb+WW~(SbNYN6W-6>jtyNmfs<Sz#2e0?L+nR#A&nsO&?l4Yu
z05K}Mf^FMtSGkTEa=m{`Qa$a=-fpZ}`S$VNhdMUGX-5{1UH|yKkPlfCCedx&<lwpX
zQ|4AR=z!{z@76r7(=il1=NO1O54?z)1ptxtyEBBU`38rVk{tx2Z|tmPvCoyt&U+5H
zS%e@voP>qAn}E4rriQEj7SQXQrD#1TuS$>VVgUnGjm`ri7J&Yej%7!_Dgw(cyImT1
zu<K9zg`gYiX;z6)b*N)^hpqSVJWYCDlHl$P%tFwnk)_KvElF&oauXbp`OD<8I!vU{
zC;HX2ss|#UxhD8~mN;GmHz%PTfbC7XA7uu;Bs>ggi0{FBpU>$qo^4YK+DT`|n#ZQS
zT}Ec>c<u%~w00h-LDr1v9J)IrE{;7{3fZy{Md;Z)5+BX5+rWkn*vz#tJRx33-Y>5n
zmp1HXk!`|&sQqFSdem;h%8w&HN*v>T+<i_ebp4K^7qKGWv1vkvwLd$A$8*9T^6H8|
z4x)rdYIkPjSiwQr5Z}yuEa@=r;5|ytRBS822u$V(c#o$AM##V<L9kn<x8W$+;%o3r
zTrzY~WB;#97DJS0;un8!nLkUBUmZ_R>Qsfy2MQ~R{gA8Omnj(BEj8s6Bm4x}ltG2H
z-4o_sfuJCdUfV8;20K6hKjTH-!02W5seD$-@z!n8|LEtZ^PUIzIr>uJeklfc^zwG$
z3>m%7fAs$k#bErmhF6@80T+<FP(0K!Z;TLU?EzIR+_a5KL4f#4X$kOlQ6Y&Rn3U0n
zsAh7p6^sc8woXfHbULkvkmmM#5V>0B*_0|SdrX_*N1oP_KMZ4P3#GPXL*pqSZ7$&?
z{^f4NY6ED|v-xpQ1YZ6(Q+0Gc2>;Q{3JM)~PLLGC`k{5GdFWi!r<`^d)zo~p%q#mA
zN6I>CaRqCfxB$ir*@2m^iu|Zoo|&E(S6qk>8?gK34uqoD{}=OjpCNr~O_%(c`;Z0q
aKmGOc)27gO=FD2h#})Fx75Zhr+w~7y>bmd%

diff --git a/vignettes/transfer_learning.Rmd b/vignettes/transfer_learning.Rmd
index a0768d301..a1c6dd6ec 100644
--- a/vignettes/transfer_learning.Rmd
+++ b/vignettes/transfer_learning.Rmd
@@ -50,7 +50,7 @@ ImageNet dataset, and retraining it on the Kaggle "cats vs dogs" classification
  dataset.
 
 This is adapted from
-[Deep Learning with Python](https://www.manning.com/books/deep-learning-with-python)
+[Deep Learning with R](https://www.manning.com/books/deep-learning-with-r-third-edition)
 and the 2016 blog post
 ["building powerful image classification models using very little data"](https://blog.keras.io/building-powerful-image-classification-models-using-very-little-data.html).
 
@@ -194,7 +194,7 @@ model |> fit(random_normal(c(2, 3)), random_normal(c(2, 3)), epochs = 1)
 ```
 
 ```
-## 1/1 - 0s - 487ms/step - loss: 2.8362
+## 1/1 - 0s - 436ms/step - loss: 2.8362
 ```
 
 ``` r
@@ -648,7 +648,7 @@ model |> fit(train_ds, epochs = epochs, validation_data = validation_ds)
 ```
 
 ```
-## 146/146 - 53s - 360ms/step - binary_accuracy: 0.9183 - loss: 0.1887 - val_binary_accuracy: 0.9669 - val_loss: 0.0926
+## 146/146 - 57s - 389ms/step - binary_accuracy: 0.9183 - loss: 0.1887 - val_binary_accuracy: 0.9669 - val_loss: 0.0926
 ```
 
 ## Do a round of fine-tuning of the entire model
@@ -709,7 +709,7 @@ model |> fit(train_ds, epochs = epochs, validation_data = validation_ds)
 ```
 
 ```
-## 146/146 - 97s - 664ms/step - binary_accuracy: 0.8660 - loss: 0.3213 - val_binary_accuracy: 0.9652 - val_loss: 0.1022
+## 146/146 - 86s - 592ms/step - binary_accuracy: 0.8660 - loss: 0.3212 - val_binary_accuracy: 0.9652 - val_loss: 0.1022
 ```
 
 After 10 epochs, fine-tuning gains us a nice improvement here.
@@ -721,7 +721,7 @@ model |> evaluate(test_ds)
 ```
 
 ```
-## 37/37 - 2s - 46ms/step - binary_accuracy: 0.9540 - loss: 0.1103
+## 37/37 - 2s - 43ms/step - binary_accuracy: 0.9540 - loss: 0.1102
 ```
 
 ```
@@ -729,5 +729,5 @@ model |> evaluate(test_ds)
 ## [1] 0.9539983
 ##
 ## $loss
-## [1] 0.1102593
+## [1] 0.1102476
 ```
diff --git a/vignettes/transfer_learning/unnamed-chunk-13-1.png b/vignettes/transfer_learning/unnamed-chunk-13-1.png
index 2679c6251cf955e18919829ead7cdad24d6be2d1..c72625c5f53df9c985b9ecb4e37d42b54a46c867 100644
GIT binary patch
delta 370688
zcmV)lK%c+r<`#<f7LY@Kyrf5UCVWnXH+Am5ee<0=Ni!O0Mx%^CLJ~<x#*tub)@#hJ
zy=F1q&tR|@jBUI)5X_Pc4q$(_e>Um!nqZ7ESrVW;iW4+>=FZ&Q`OOv1{?RjH1}j=Z
zV14)b<KBMz?K)LmPt~c@=bWk+i3rxPhBd5V4bNP#A9MrOu!c2%tYHn`Up(^^JsFiN
z`-5)98lG;f@&5l$tnvQ1@Kh*z8XUgkTlbtu|IPdD2fuUk{fqhu7ko0%x4-55>b37X
z^zWiRTRnKsT@RcXm{ZrPd-%Tlk9O6=-+e5kOMCCU^MR8g0RY6ucinvBjW^zS<Bd1p
zJ(mE02~X_4?@(KRJ>IsRCGCd-&m7(X?EJSs@mGKSFZ*JEe?v{)K6Kv$$NG<#_U~ot
zdx<-HP(4v;KLmKj^8P(UKe3*FtO@>up6~SJp>BQiIsSA0_@E&o8=idNj@xz}YUM=P
z7hluxuK(z<`91gUU(HT`I&UAo=g#|2hW6}&4~KZ~2d|%heDR;n+sA4_ytwzy+wVEr
zRYYd#)=zxwv-kFFe0<kU-yGt`I}Q)**;W5r*p+|&m2JhKF<$dWuKN93!hHW<zI3)U
zGBZ&wZ~wI~x0cVod%95fz^ymlynESDGPmodo9;Ruow{L;ef($F2j%gJsyBZ9hxe-N
ztFN7OykfC`SS;4AdGCXRJKuFpGiZ!fgLN<a>%-=WN_$L7o)$bacqbyV3m<>^7)ST{
zA315`v#8OlU;4ERDy5NTsl4-ld}-;lgvZNN_C3U%J*e9DiAh`I{qzT)dfv_1MC*s|
z`A1}m9({txKk)hflq!02^siq!<=1A`H~r#8Z@Z;``^DEZ#Kgu;Bc(ui=lr+NFI&6%
zr8ivIXl^=psxtn<_uf7DF7fG!dv3aE_sT<*=sh3$@y&Gy!F|D>K05Jc{?q??L!r9v
zoT*av(%-tNeM+V}{IQ?8aQpV{=ZyI%&VSoo`Po$`-&><y{@C^Y^e_HP-z<FW`a<<T
z{oUVxyrRDDzy8CX`)>Qj*S~h_{sSL9>+WYxQxe_uhAqzIYd&}1M_yDaJm)>9IWfmR
zaCLCrTkgoRo8Pd}yW-vZj{eE>XaDP`cJI39a3?3*{^Tp_V?XzgoqW$9T;`wm);rY`
zleS+!1*|=JJR^7~BHQ_wU%uoeul|Wi|M?$(Ing?cnz#JZ`>qbQzx9qJyZ;@R1egBd
zp3^=5_?UXvH{W;G<((ftb5I@V*e4`yjd!B&7w>aNK5!<{fAF>UJ$%nU0#jt>9&Lhu
z{qKL^^ZhAR^!;zV^XGr%uXd~CreE9WUi6N;zwnv{Z~Wxno>V9Q<|iA$^Zw-U-giHL
z%ir?)FC08_*Lz=3sJ#5|zj~T@bM*bsb-Y*nf9XR)^WM_QBcFctg!fp)ho5?R;9dPk
z_qJO%{_ofQ<|p=k=S-_1{Pu5bcdIY>(EjYfv#b7O`+I8)58v{YTMmb0PyYFJg^{2C
z@b6sYl$tZ^){O?m?Z5u@g|qIy`k&%|)5Yx4r(QW8Y}hpIZhhm;(b-Ll4}9fM{>z1v
zSH1JD{?}eN<rbRL(>1qo)gRq<``dT=&wBTMP3aeYp&q>Kli?GSmeLOeo*BGT^qt?l
z;>91@{ngjc_|N~yv3t&<=B0o5rftFXA6u~$e)45O@n`-|a=O$LVCtdEJ9fQ)<E%l|
zqko6AA3UCsyx;zoe@pbk_533+l|9-7f9bp5`U9WuPiZi=c=3y#i~soQEt4<$z!77#
zCIE543!gVGCSLU73+ce#gGUY>Q?0MP^A}(9>;GgfyR2!W(+(&$^$VZeyXQkUI1kB^
z<<ZHhks`;(s?@vaMbE3<^41G~C%3=hrozshQ)hgm=$3c9_JhZt{ktE2?KyDld#Vm6
zzNf~*hAUop#dQ0h-~7_w+WoxW|BcJhK<P(6a^I2Ne|i1<NB`_wqi5dzpV#@Pi{a?Y
ze)m_;J8^h${cru|mBCppqvYg$x9wX=+DA`xXJ7WNzy0#vN007$_l~=N{`kW;^<t1x
zBB0|)N|KD8khIPZkG~cG{$ulQ9{kha{LJi+ZC!ZaWM;bucON~AnzVA^rR8sb<gM@j
zz?<Lq)l|oEcDjzo&(wD>=i2(LLG|V6iA(!o!ZV2XeZfWlhUkaw`A1+1*FV|>?+<?9
z^Zlt*^ybiCy#7}|6<+&)TYvSDU)m~$?@_zeA+ViRhsu@8$VkK4{@*|TwQqd=bASE*
zcfavP&4*|emC218#tM%~^TS}~XaD}*z2Exy@BLVC+xvd$fBv8T>G$JX|KPPBIC|Od
zeE3(k3Sf=1tG;g1`Q93-ee*kBeEDm>edS+$=O2H0+r(%?j&4|gUu2c3sj|sZb>`i3
zzdJy|&fWOU0|ukJzHwXkAwp>R){p=BXHQ)6hQIyt2d<m@o4@|o4mQ?JH8>yNFyqB7
zTdhNTwQV79w<EvV{BA2Am$Yw$|Bi0@{^36~?{=w`!Cmiv-D}_T^|spg$&c>#&!T4b
zrSJXJ|GK>O_3y}kE!+I+_}D`W{NrQlbKf}g@^AN*&mL6ziAh`I9pL-K`=8%>*?Z0;
z+P&|Qdj8>=3bRL>;Oi!S;Pd?{wdmt0CZ|}&qu;#wh_RLcKtJ&3Z+hGRe%l*A^1$qm
z{OAQ2{K)mwcmBoO|Lkx7;8j;W_w}D$eW)b0zVgmr`fne9x$}`D_KyU6;$uHCb>1)k
z3vGGMCF}e`y?#mu>wDk(y7%p_UUU7r17H07m+tLsJG<(Ye^X=l<+r@*_ik-o_mXvo
zKl70ff3bJX4Lc5e^f!L%oxlFJkGC#+!OmOHy!(X*PY2Iw^>e@dj+-~U@sEBEKk?Rg
z-~LcS0YCnK`QQG9U-;E`{qfuW<ePfy&JFjy?`Lm#<s07fuK)4-Ux4TS*h_!xhAl__
z{&#=p58m+JuXr!G;R@#oNxSg8@2Z<03j9arU0(f%w>@~^z=1t~_n#-+E8g}eZ~p1C
zs5$?e?|jRrRyJOHak~G&=uf@kIZuqKo!geqyu9dtIoF>xsGePVLekcFhwmHjAA0$|
zcb`G@%dc2?_?~|xrqbIUWrF|6kKgbEpYQzbZ-4ufl|h5UckW}Cz5FHDUcN;fy8X@*
zjThf=PPoF(y>h0#Fn;x=*?oJ(^WXN7-@9SLo4M+R8<ubU{MYBEf98)r^lO*m(L49C
z%U*te!}+D~o`3j%zfM=Y>e}`E48gqvw;u?u{ORkq29LP(WA(!4UO;=l{>87|o9+0;
z_x!I{jV|A{k6rfi>%4ovc)(Qh!*|?z>+O5QWk2)EpT6#lsxPlZXV++s-~72-y9IA>
zWZ%Ah`wsGpUi;=>yu|Lk>DI;RSHA20uN%F8_X}s<-T8{=Z<GLFkALQapIE%|H{Sc(
zKRJH*rkfY1Z+PA&9|17fcI{QQxo>~tuFm?G|NaNxeEl^qc$Rzc#v30n7yk15-}g&9
zH$Cgxt?~Ug?F!HT<qy2))mu(|y~io-O;<O>ljZEGz%zpPhwAJey#0XxoL5|b&9%>e
zJA<0-=Wk!%+x3;34!Bpo;eEgR6Ep4;V(QJ;UG>5<iMwXY70<ump+WVlmsR)^k+#M=
zeE)d=u^TQwljtoMKJVdr{t=lf?YQPqCiuA<F8@K#ccir=5Z17UHLT$ojc4A@*fp$S
z4Qp7#Sy)r_Ygoe?*6<9+4~5fj4gYa}k@pr>@Azz0db@Y+x_$BJlI6u_t=sMP2l>co
zT?iF+Qe!zWTdo(gOy@aEQc{@$P{NT(qRPdxHK6l>l<27Vx+^aE);+sqsp5b&jCTe)
z85mMzIVw=TBLYVNGnO($;RV<juP>ZvN#@K<)R&hQOHT2kbEaBH4{n%jSk)?jNRvhs
ziR!HfRyvAv&A{Oi2Lc(!I`Zx8RTuEdF{L9A08AnR8H0oj83Ra_lE@h{0t*1lB9I}5
znt(9`ARvy`G9cu#sN!**9KUz(q#tbCu=QKtzITwS`e=R7?sfVq*N(MBWE@v$Wr3{c
zGvD_IgLI{xNY5)e20^FFc!8IHrmEQ}7hTOn%3SxAKl6Jt)7zc~Hj^~EfA4o1^``6i
zT2mYk27@*-GB!-IZW#AgT1yIZE+inInwr&0b$gv4@U%5)o*HYo;7aFU4KfxKibDFX
z=j&K=#sFEn-A>{t@LVEfj5*RHBIyVzgeA*4?{|Ba$sni)qL3%K&aec3u1-v?-W;1<
zKQ=Z|EEXC24`ZHYS=JB3K^R6+oJHxrz55Ow*pG}6fFWzFHDs+d00th4N)Q7;0JNul
z>ed(`CByH=EfA1#L<B?vB%7NzzVVHXM&nUhvZm<Q@T8E3y=s5{j*tC=vbj6vEcN0&
z_bc?gC`}*OLsIfmMJ}Iz=$1+zaWIHrrRxR@3%x-nFO~|92T>d`?%Fiv4$RaX4Cfv=
zc2brK<)RlxX|HFyD^aCbW-iNZ7I@gG7W_bP5clspj*>S=ipJQW<Q+U@4;|{wpG+#g
zE6iZ~Mz(3(xnSoxM~@z#ThPJChTTUFr6B|96&uA9tDQxQ&wBNL3+CViUs;N>)LI}w
zO9%vvdB!M5FlANBka5n9<`&R0t1<vU0?D|}Qftu7RmmXdEJ@>g<K7hH-NU`Tx8J<n
z>8BP)+Y>WWGxy$g%W79y%Nrv#Lk0jG3C1RAV&ag6X<8d8)=IosV7bjECT5QwJj~2Q
zg4Tj^%RxBkJUwiGT5ICGCt}7^xwP2UD%-eWow3>C^3w6SgNKgo9mJin=_WBaxwvw4
z;qW=<Y+C89sMXB#WSqre91>b&;7F`h8=W=?$1fBrQM=pib^%zHWt?+YI6yEnGuuo1
zi;MHV=OJU06BCs}ZEoSP23{&OrFFBuHYsfwTQ@y9yMFV3Iko15>$*=*KRhKMBFAwm
z)k?8kj>72BgNIJco!Ilh?p!J1N@cV*RufsP0W=_3YXOFeo&hjG1_0JtL}ZL{&Jj<C
zR4p-%oFjk*w5TD=vPW&pnxbFBlY`97$#34yI$a1r^bNIFt(2b6DitSOjF!7eZESpI
zQ<Ev^^}<enyXS&K5ZWYlWl`zK6TU^+>GaEf;DIOy<>$TNM{oPHzu<8eW;%_viY;S2
zOu~v^3LHt8_me>{R^!zv00f3gsY<1MVsR<%n<Qc7S|L-pUl~8C;w&wG^Yi=4e&AWZ
zoG-7dO`kNKKCzQ?Q6Bg=yy`_KyUUAv_pKX|PGnAh91Z8%VQxWbt8!f^7RnVD>i&io
zU&%Pj7h3my`YT*BH}Du))K(w_j+?8T8-4NSO<Oih+_&%EQ0rc$K6miwc%!C(90#Ik
z`JQ{1D>WY-Z?zp~xfT+$Eb9%@TD_XvEH`xF1v}=B?U7CuSXOjd5_cLk!6EU2^5Wuv
zr&+;&4<4I|rvhAvnT_jwuh8v8ex;B^s0k|7kwUq!usqMXuT97EO5AlCid(}@o?LQ0
z2S_1G#+cRy88|{_dEV*u0>3N;swB&@ba8RPah&n-anB0`_W*EtwVP#%ELR$gR^@V|
z*B_`Hi+)L_JPFmH9j~9*w0*~=8_wC{21O))K>YV4-aiT+N=13fB|iktkTLE^VM%MF
z3W1-rV#kqSiL9Yh=`aLvO3{;rVOtnKrRW)B#uze|F@}gkbwA8zB{G24LHU@?SX1<C
zcv3KN5*<7SWv8D;mgMHqNpdsv1b5009$XB3FKoxD?Yovq%QuWPrkZ0q&gXmmg?^NO
z5%2esewZtzF)$@xUcUX}$;NE86by{X6O!aIwB$3_cL7b7D-<k^GRe7H6dN~;9baf^
zCRV!bN~vs|!DMCPj$03w3&o|?u3%La3(qU}l1$V}1B<Q2)lBQu@=SxL_?|C*v)jw`
zO6wPX{&^)|ef;3up_PmUfzY`tSixj}bw^;K$QdvKnVX80CNm{#9fndkRvXl&>G-+U
zci*vZZf<^TvNl$AIPneZH(JmdLjomzQK<PzY=acE&9qU%^^}nw3szfdFo;W~;;~}~
z%Yoky2Q!r#BRkR<&(cC1_6d^eSi=}&?9+A*F9C-*kE1N)4%Yn&a>}(zv&8j(f_i-{
zHEMogMR*>eu%J}jV<|X}0}6z2ip8?hIRGVj-W$Z@O}AXAH%G=nlN4U1T&q=^LdZDL
zoJlU2i&980%Tn&hMA?;gpRCWdA9eFVHyIt7yy(J9H*Y(i2@c;+UHpB*P$Lsq3sy?r
zXw)`tSfBRVDvNWYBqvL#hekYq0RRnMQkIa27(4Z4hlV^NGD~EwHU<%aF){Sr44piW
zV|c77`ZYW$XiJeJ7A?#hyNr2IDkB$zBwajmyjrQ$8s#igoMg39l(||zI(F&Bmw)Li
z->4Q#VHmN{<o(RGrd0Mhql)kRr=R+%)n32odug0X3l+cAR{c(rBu2Y`E`ZfwnQ){b
z+Y5WzviXz!CV920r}L~HRP#Z9x><5ux62s`l7ZW6g^U-{R0A2#cu5xGB&^oO?DXWF
zckd>}9qS#rW5HwT%1Q*Ohhz#aYu3t+OP(R_JD8<LvZT`~c!5>9L{=^qk|>srH%Jo3
zh%6S06{AI-YL)h-WQ-MmfN|jqttiYe9`ronIHDxD*KMbY(t2a;=&@x2A%rUgkD^|s
zGFtMyrR8qH4OW&9Fc+*z3#EWM?aiBJEj>+&UK_LAUMUqTORFn+ZXAXPO69}_WLgy~
z)hP$Vv_D7_rLBbwExDeIldiFTrBr2%4NYXNh8w7uwXm|ZeE86R39VR`+HjDS%YKr^
zQPdl0jIQ6Xt`K-P-}LR&WQPtOPtz2P+iQiUXPT9XD=xq4ydB%QBP?043Yha)&40pp
zXz>`D_D}6vGL%U}UyKl?QfX>(eCc?7ad9!~C5&@KhDg?07)F3j#~z5tS_^dQ_0W)K
zfShvxFvf7k;EbPtBgQ=T5!!2tehp6w12iY53L}#nPTYK(>1DaA=6fsX6f2c-u~dlS
zC@AX@tOOpPtc>#rZo2WV2M(N^nHux+xXxgv?jG*+!Sh_F^!#UE?K|A{<=nzzrB;8*
zb=O^a*|XmGyKh6z2wBnh@+@`>4lyl}p=PPJB5+CPoCJD*j>F8%%*f<OG{4a8wPQx&
z$l(<fLIxu1<ukS7N_$00TPQl^MxotW8lP;EfrXQ;e|+F*v*gdG);QQ`7JVNrWrbo{
z=aFm=db{?LBjN)Kd4Ev#D$JymC7wWI@}es&u`EkG*9iyt+`{sP+2-gdrg@)$rFmwo
zMTfXA5NLUSX%4{~8L9R{v#_`d$QU95#)G(CtF5#LwF-~3%oC*~N()|L{f2cXmzD`x
zx8H-O%a+wTpI<t*(c9`dEQ|YL7AchhLnaU<5ivB5P0-|t&Qd}muu-p{=keJ1SkUx{
zOp+vo<hqq|+2u@TS<+fvZ7rUtH%5b?<a%Xg5r_nTW4&&t-|H()#E=u$k#<Z$MG!7`
zeJpH$_RbyKcM8`-%On$kU`PW310WzEMmW@NNPqx{L#;}Lh!y}4IGs}4!~dEH5D@T`
zgj0nr5&$3o8~(?kavpx*RCxwOL$mq`b>{H%5*z@qXlu3Fx{VvUak!eqYVKG(=v$?Y
zMzTnMB#@y#y=xZ%fPf(f7&>S)5z)}tx0W+xgrLC!4BhvPuUYhK_&%dpt0W_p2crJ6
zOU;qy{k>jlnGaeekT3`wxqfV7esQI{uqraPbgVNzwlR*g@}TwhpT2hEx!cliYuCL8
z-}$+lw_LpQr+(zeRZ55V@7X%q{P7!p;_7FAT@@wiM6(esFKb59VbXE5N+obEV5!}X
zl$f2Ja#&_f1PpVNS?0~mj?bT1I(PHv!a}>^2gOp*jZ#IFra41_-1wfmvf8#pg@O;#
zkwyQ+@wiYcWH68niEMkNUnrIjA70fqnI?=?mKzogavvoMQ>#{^u+KPZt*2)v=jRuH
zWf3i@TCJoYK61R(%iLG})C-Rsc<{jS1(Y7@+;`kOO-HLGQ0`JE<6r^-Mb9Bko^VXg
zd=RUhb~%^?<*Y8p7P$=-Fi|eM<#IVsvpjp;&#QQH$n)&3UEk_<TfSQ`DnT?%T4QaP
zWVJ@4P;Ji7XPMHye!TgdjlS<MEiL4KDqdWSwKn6ElXE9dAVyj%YteNAU_4H8#+`a|
z%o2uSj|uR7U+WYZ<yqG64;)$aoFdblv>zwXJ+=_I{?5xUIsej&{Avl#uv;16DgBJx
zA`l>vVE_QcPH+4~Bm_h<0HBSLOlYMD2pQ*`5daR=?`fF?L}&pR0s}y_8~~7ifB+E@
z5C9Mm=t(cMj4{vi#>U3BB>Db5`{tC+(-eRMvf*-Kt+nRVT17-ch@qu_YIULSE>3XT
zn-2iP<;PkJ@EH4E*A)F4o)i{)-MdcA)ypH3mt1z_-~p@@anOmgv=o#a-?1v2JG$Ui
z3xS+moF5GOEY}^0&Xuz>yt|@*dkv@8-ByHarpI6R`d{r^BMO1Z;9YNfC%W7fqSJ}i
z&(0iPT9!f(fyz~}Q1m4C181q<at)4;1nq7VDP1m<nzee=TlKxE=172~FW5Pqgi$hx
zT%(JnG8yB!tQ@$-f=><!DW}Ir4<1=?Bp;tFDBEARt}LaqwA5`*j&O&6m%#6~dMA$c
ziq#?$+@>)vxS29X7upk}BT97wpFD0|39jq*I{nGXjjhhgy7i;&#r~&1`PKCs$HFu%
zac}+j#`fw-W*s!#=F$UEbqO*cZ8<WJ!RXXj;M>W@=G0&(!YHG1xg=$hr4eJ4q<Iz*
z0Do*Io(c#@l<WTd^5L<6=BVIgb;=nE#{trVxb1^=d|4<Lu+56)YPnqQg}stfa0A{S
z^afGKD+o97r1c%Gt<@SV<Gj?UUjD2r7fzn|*MI#|J*>^l%vd6;2`m*{mysisgxu4q
zOgfQ{^z7!1+b_AKIW=wp_;+KEma(CdHUtK&r998mn6xq~v&I5{(9pppxbQsRS{sLv
z%Cdg1pJquRDEP&|_5DJjfQV_9I?_Qz!^m>vTrl85Iu<zR5|IHD1jxt|f*~RV4vasE
zA67&44FIm|SF3d*Oj9LYL{eZ*EhgWSHu_zc4krWz;==$eAToZ+rE|*314xgx7Og4z
zH9RR82FK#=gZ%-29f5$$C`s~^){l>lROEw)mSm~cw+0teo2G&5<uc1`YjmOzxE{o=
zJ+v~277uj?S6_2&uipzAl`KilJ#SlSq$vc;`(0$vZmk-tRUA^08>3VTq?p4^&G)A#
z>bcIB<J=EQR>iG;R`Sd9%PYbYNt~$Ix}H<ynC3ZSskYpI8pVs%qpfxn$Llt%ziZEd
zdU>Q!&Z<psa$<COZlPJL=4rUP(g!ZcYGyfEM?z7hsRSoW^kASYa$`EvQ^DnzZ@cHN
zdqzhmPR!32iiMSx<yx)8n2}PUb&v0F%R(Vc(?f%K4m?Sr)9I_+<hj+>2jq-PAe$<k
z<f|JeDvs}e9A7wE9UoCV<6i6v?6y~@r)QRyTPsWLEPYBJzxT)xocjf@-I*7HDrGNA
z2Z84~LEWU8R(fTA1prX6axI9`9xyPZRhkgbi&b&-#Dm%ZfhZQMCd{H}P^^?<m1HJe
zYM(4M%3IFe0+vbXXU5t*V~o2FqVg$|S!9!bKk4;<$H%5FxcHKFn>K(Y#_Vu~8#-hF
zjW#)DNgDS1omSdwM}t1I3Uo@AtR+jt_|W|3)>xG*uvRM#)((i6;J}!4q!f~fkRj(B
zBtyw0cM-Xe5{2&s1?d)?pv0x)I-c}gE;s-qq9?Pvbg1ZE*KJNtoO9m!`wtyl?Q}G0
z1cLy7L>w#tJIgF50z?o(AR~r=2DLu*1QLPlp@ZfDfV0O|<H8UbV_K`r3k&m0iwkKI
z16gB?Aw<RyIU)ySr`8c90wQFF7&49k3<v=bIRF6}0Ly@YELp>lRu&e7kXfEPjswWJ
z;EW4o>~zLIV@x741crzV2@otW0w4yNN?!PXqURf{?!4npNVSR*WM~ix83J%1h6pGC
zS*GGz)ti{ABIf8wDP$bSS&}+Zx~?OI;D~t=SxZ97Zf{VjRE6WVJKZ4gqCrm?on=|Q
z-WVGn@An6T{vZ&t5EPOm<wB%MS}7HM*CngFofhYuNofpOvbi>laj@1A%ov;HN=TW1
zWf>q7Qp&)SQXV;cWPEf?WqG4oEe3_7$Bxw-l|eX&qNr3ZJFX)fCyF8qgn~y=<okY>
zWsc)Gu2Zd4=jZ0o*mAK5maHXU=obTIOvPEXUSpgSK(SP6cRSr~chjcL;UG-2bY*3g
z3(k=xQ8dzM#HngDMgrGeUR-R{>#cTwYqhm<*>hh0?;o~j!eZEqr6{HSysP43;JF-6
z&d*am62*`Vx`?*h$|}`DsTMf(GzwEoY=CrVIrKMOsDknS{GIpTw(FIb{fO0ib!BPu
zd0VYk*5V+J?%T8H@X_OF!I20yFPHtmb;xMWn2TCP0}O;gBLr~8B==TVd*woZfiXAN
ziBc-eO*04_p*PLWKJdW(>(@=R!_aeGLSkrU#>WpHJ-TD(mc^Bhv7uMind&#kecMXo
zB+E49xoI{Vt#;RHgMzD^040$!Dsw&d)o<Lcb6u=3hC=5QnzY_1)vMKw>ze!SIl_!!
zsWjGKXsb9$O^yaQ0Z1k}2MgAJlGR!(q&TOQkWDsfe$7vG>=oUK>5??*B^O@xFaP|F
zG}V*MF;_nIama@S5%hZfBu#_B8-z+_iQ{^%@9AC-7%^_*Bypu%uUC*+3%WVhIJt0G
z>6A;z^Ff{y7Y@tWAnH1<ljVJ74M}<Q<o;@*x^8wV?&Ym^Hzz9lrJ_@R)UnnXs?_9z
zOs8setbX~$m+!pfLg6|b5Km>J7_AbWW`jXH8T8`zvQ0yq#saj*NRR=m5m0Mm!H@xC
z$GUa^03ZNKL_t(zlrfBHa0xkMhKx1VSOW%-p+ipEOC87MoC_gPoQi5P#z8m)JVDk+
zF5LoiTyp$krRtQ5#m0z#a03f~z<?~^Fx&k!5b%sc!yeX^KXkA#&KaXhwKO_02D>rO
zv=pb}u|!~K_`B_t=V<s;C~GYO05U}4QWB9d)>uoH7-tOGnOmzK1tQZ~97o^&&bPkv
z?VH=Jl{|?VBc-fi0yyW)LGCc%gq(3{Q4@kiNhkmX0`Z}Li!mgBBCC;<Km%aVYMrNt
z9@y=9UOyZZibdhLTsm9|E;%AHE*O&xC8Fe9GRA=sAtw|TupjqIwP4G|+muN^{@G9K
zZscpsl|s)mjB#KR3_}3)fH5iU%xux~<MD|?H|k006oR1BYF8?iMy<+`Nyb`B3%S;{
z>PV7iNousQVU(PIzwMkT>b2VKZnt;gg%{m?(<jHqCakq8j*O<|mDZ+Bn~gCZ7kLs*
zPmCSd|3FYIBx#P0vtiTb{re9a0*9kY<9>gze#6EfC>%a?Xrwt-ES0$L96NR_j-sQk
zS1(r=gak|7ZpVV&vGamKdpS$I)pn~=tEE{AoO!<A?RLw5<x&(yd7iIdzkcy#q|&VD
zd4pD(C2750Ke@P29~p_#glY1FAkCFuEW~M2uGf~AmzS3BDFlU;)~eE)b3Rh7EiWwY
zd!V0aIPbjeqxIUL+sy_}r`L+&5a6j%^v00QohVK@$cFE)A8kJO;!E?Wd++_b=d(_f
z+8jW*{!%A@6s{;0_*kXHTHQ-8x#+y{4H`TZC1YDIckkK77`FS}I8A_vS+Hc7BW}Lq
z&VHI#>$OzH7HDj;*>0~Eip4mJeSmG-F1qoyZ+TYQNJ-&hXdETfAh*WYj7#pMQ5@%4
z(J382aPaC&FX%=C6e3(&+Ihjb@u1xtZM3_K^#Z4VT!v{Rgw3L?RIH>K={)nw9%C#?
z;xJAT9F=BB!+dCuz@|A9!gGDrX?Kz^3knsZQk^Ry7-fbtN0E$rrqhcwf$PYU^pGLr
z28=-hVq9BDENV-LKmt?9iX)XMSIXGsRvWWC`qr)I?bw#(Nj&IE=V{stYOJwVdmiVg
zfrONQ(i*VVYORvA&!k`+a+BBVBWaS?>XngZW&gqZ338XamI>)|L}rZg%brmx&r>M`
zYHKLV6)QR(nK%tQM2<xt2qgt7$>Vl6?#9TREnCjpamj_{YS|)@1xu9W-PXyuw7a6B
zfyolIid*GSF1bZzt+q&Dtu>Z_7?Neog0-lBjkRb2^PG$&1mJ=r0%O3CN(CPP5CKS|
zW&nghMx{0nhdzAHn8{HiO25<#UB$doW4t;(EeZucC=wtM4u4{t@h}S-?reIbfbG!w
zHd-fxUfl0-B3C+tUSAdp0LGXcaTq5CA|fCJMBt2ZYXJ#?5ej06#(+W8ISh3r=bUqY
zj`%pTZqhh9@Zi3?@7}ezxJYCGnE`8=AnqdPz#L-S0w)y6xCJ6aV3KiREkhsz0we&7
zU<u4H%)!Yr#@d}OW5^gVhKM93;{p&#fMx(dmRSp+5fvkeoGEgEG2{e*g@Uj0Y@}Md
zXZPLpk&%^NjGPH{i3?)LfDwU3INh;-))mfp!%zFm9;&oA5E3h%Nc*i;9(B9jYNa|_
zAD2?DtgMzx)z;#PJU7*9%}2Si+7p^wgHq9^iRpf?AH{K`(y8fnmhtiFsYCk@_!YNN
z8S_eosg1Kl(4RZO02WqOON}}g637O=w=h3nA0I0f3m0B;;e-436$|A`HQ+jbE)_Sd
zuB=W?P4;^|t4*;qn&r0L?$+xK;W`^OZfdpLVHj$y1LhWMqn_u5AbYx3L^sbP#$5za
z2+Avqtzw~4E>*KkF($(}1nx@Di;}F@kLH(JqvK;-x_8~Zt5hj#W306*%Unl}j5e3M
z;laa)iiPdc^ZJ9nMf8iM#~K)aj~4>zl{Z{__dUClo|&p`{QVDn5FMT_?Rn0j`@VC_
z-5>n;zs5Os@@~0QMB;+MMA7-RpZ)K=-TuxuzXAz4Uh}e_`!83&Wct3ndymc^*|=f$
z*s&visXWqXe*7Okz30JwwMIiTavg^Ua&BRfBLm8f0YY~3t@n4jl*Dm=B?x+YQ1bj{
zx!mcrgCH0TTFudVGKjJ0C~ISF=9W6vP|xJ$X8ri#1C_FunXp=NP1N(8#E?+dH(YT?
zbEZbBfd!hH8NcPW{Q!{WnFVwu%d?EEB!EsklDw={9*4O#P%cSgDQH-pZ@EzD_Nd=a
z!3u$_T96XJV<-i;)^O>6Dx(vXn^Rs@1Vp1@Zf<^Ls>uatn+2YqS&CJ9Vr5R{x#Y||
zT}R9sqkDs8DFx#$qH!F@b=~2Dhm2_(=PCnWd-k5FRU5PGrva?%%5Ja6BzIiTb$uXa
zj3vY3C{3bx{krwFTK&Y*(JYEt!IX=hJ3FOv%3{o-oN3G>lV_lRKyKc6&ed1HaBO-4
znB-uRB<!y&E+0P-t*&@RRit3t5+ly6R*6bev`7|!0FeL~DNAJS(6lzj5?DeqgpAWL
zqKjmS$RHRT`htjv0S^7(1VdSN$O)4|2*FZgfN|z0CQjQ6%r8}%lcoB&TdpdwfD8#Z
zA`lYUQ~sq#hPBpz7-K27lxKNlDK|z9O9>%}$PPX7c*-S$r$YinU@V~lU|=l)kmE3-
z;o@S5kPAkS8H_!+f8WhF-gxxL;V_CE#|0n&1nv;`kaOY^fEhqU4k&;R0}|W-2Sfs5
zXpxA42DAu<fni9-XhdkWR)izC6pj=?mW&dTGXZQE8V3S@G7LGKq6G~{6vz=7L#4Fz
zc-Ze<aNhPm`_qq(6)HB&Q7}DR_XsURKx7CU3BXiJzAtS-l$D95M#rK!@k?dLb&@1E
zguvu#tG#Y|#^J7m^5PvAIb8a_9|XZva|{5uV1*z^li2tDn{K{o<JiVhwZ6L2%EEYd
zW+slpa;dz3+FH7B=gvEK?b@($bGJXxmUiqoe|e=fzp!8_8ylZGd2-%!9h+ya;GLC4
z9XjK++S1Yzsr>lSBZXpdWMpLDgZrzMvHl=xcYB3Gku!Pk-S=(UxCsoPMY7ahX-nyB
zn%z7Y^jpjAso9x97#ktBo_l^6MK(8p;7F%FT52?Z8oT%Go|u|UbJCVMGB|eZ#MX1S
z4WcN^vu>xmw7fbyHNDzuRq9RG_3qfUt6r&)wLu{;=7~S?;Sn%7K6c5D4Hs;l{JX#Z
z;;+By^{Tzx&w8~Y8*Mhv-!MDdDEE)8ZmKt=WMwYaH^#2qy6L{#?_FA$*G8LAU3KMo
zP;iZZ(KmhmGm(ysNophG>$aWurr-I0R{C)!q{|Fv#!As@#jeK#M?^}STobY0U@=V%
zGo~68weH86O_y3poYX|wmBponb}!Yw%km_y6zg$fg<uj{rC3gokTP!a=13`vqZo3=
zB(tE>#2}816qCr*>&?SQ=1Y#BQJiWkgag2TxnM?HM|xW4E3JX!XvPFu)LCb|S#Gy7
zmFq#btupI5K_zfK^gKz-5*HkakjdOw1fCn>00@i#l*$Tzae8c`8~01TA8^+fE;w2^
ze2}IlQ{{p$#M2Z@Ge-9Z?M9=ab?(X{0OOo{o@b0@0Oe|FJ{$ls$B}Uy9Y21&-!C(N
zbQrn<T@DVj(ijUCNr5CM%D@3P5P_p5TCBPy;mE%3XW9r$X{vQ(S&kf8W2CWt=Y?Yv
zlb#SdPdlsgD|1H&tBa7di^vKb0<Zv-XQ{Qu=)@YWl?EUL<eXb$BnX6I8YJUnfh<@s
z+5!SGZY&DH?9c&<j3HVg01ylrSOQ{yr*@%g0006~paIZzICEUz2xFwMD$V+H{jj}K
zsE?M$#$7)E-~b#DwbI5I$2sL;!?SFC0syV`@HK*9D2IPt*W;X9YfnX+>D1!S5Mijq
zhxtB;)))@VT1#Y&(Sk`Dx_5wxkhFrwG6ndK?OXRfxHpPo#<?LN<czz(J>sr^MM1<Z
z<Ae;55o5>^kXz)0TyV}28CV9Efkx5*7RgG6WUV#2wX!0)Kp>;Flp;-2Yc)rU!X=ad
zf|vv*41qRA2nW$J0tUbsNWqhAaLqMWK6qf?$&<_JAa;e5YR#nsj4@yUXc<EWh(-!~
z;f_s}q6oVyoqh;h4x%(nb&^?sZP>`zq{_|WiIal6&BjQU<Tlrpa^=LaW61f$x|zk5
zmHqn<D4hdAp-^b8uC`ZN6OCpd{XC2*H<Oc-fg?FtFvjDKQMMEmtEK9RV<!rpABWNC
z$Vk}j`Hs7BYI=Eoe&N`$ky<T{RBN?YF4nS4Cy6q^ms?$9VYSn9T$c-fCk&(2c83eu
z>-VE5W=tq+!#GvOgmI#c-Mn>cr_=9rx|5Srz5W0YlQf&0oI0@o0GD33J1~ZnHY>}k
zmZ&*0s`3l~@+=ozx{m94{>13G6i%<#wPb_9FITIrPG^2`5jekK`-`3u)zE35xcRGp
zW%8t6o4w+y7a5<GN1I81uf>&0SC5}Kd?ZELtW~w9JTXfP-Cg$|*fY0$pdB6?<i;&e
zt=q`Lyk4z6>xySz^_*u*t-t!sFMsMYpD-Sx#m*o`DV52k6hgRh7<2A)dU3CxYGcAw
zbHh2xoFGm!CM>vKq|gd)ab*xIh;w5IxNy4tf#bR1Anb>o=4hpV(^|2}Qe~oCA;WTQ
zyx)%~H)FMmAmed=(8&y`O2sdh#l*x2lfw1=!6392j2Rjx6a;d5dV<I%Q6|t$6Vq=e
zNu-LNUl6kHITZ)%97+OxM{r|{o@2EVjA=_2Ayzh3q%Cm?=bt;eZe68Pa{Ap>&SaeE
zPIo|n3|TP<1ZS6jUHX!-vDv4DuSZ8OY=8C}A02B>1a5&dXQ+nCm2#XU$BrK>lnPo~
z#)RN*n&z%kOyX2&Q?1krLD3pgxz@^zjZSRae7?h7mFQZj=5QBNny8NX7!<hggY>}{
znnkOFtQQK!Dy8zJmtFd-OD-!HN?EVHaOB|f(FaXy(Kl&-wcr$7AvhYNbe8BmGfG))
zRGuT`f*~T2QOJOXi6;n#ELdxeGR88-fQBx54X3>M1V$UJ$Y?SetOC*q21o-KM8*i2
z1!jQ}DU|~f2Lx*wX@`M8V|5a?mb#se%g}KpazTJ_`gn#j|2~A_<*B3^k}E5%qetcz
z=9hD&_~~<h4v5H-fnf|3h{ys%B5TQ<cGn{S0a!q7jRB%zIud{-qL;t)W#w}DQJ576
z9^9QoA%e{_%^jbZuuKwi1U}q7fN*+K0t3X;&M5-07OVlQffNW0U>I0H%ee%Dahx;6
zASj06z;!*xk&Y178nA%GjA5?p8$(H$`o8A}0Rj|%gz*broLS)5PvUyD+^ja3!+O15
z+4ohHJHD@m{YE(a-GoGpp&!U%v9N#deUfEynh#Dc#c^71jzwWCy?|j-D_2IU86$S&
z$T3&Q!1Foh-NC@GRd(I~KrskL#>V#S+viIDikH1qnY`cW-0_`TFTMD(lH>2Xdso=)
zSVoS2%T<zizV8WHsWtW=JX|hSvLtI%Ytv)nNv~V*gDgoe+Od;}7EaD9lUEA06N?Le
zA()t*-231@WsKuGgCw<#S1J|ZdC?%8o|%clusJ#!Mxim5IPXSLqwZPG@4V-pnqMvj
zr6@|e{jfPUs@lCIN_TAAzB1@Jjx#Yn>IXr8wOknt`mLqKAn@}z-Lz@rvE#>Um1?Ke
z>a;r(qoYep3(R$*EYG!a0`HvhvBQUtK2@s0_q_F+&w(sIck6jrEI=NGi}P8O6?vgl
zsTWBFNzVvxwL35w`ohhDx|ES3#}RYgllPqb^hi*A{<AK<_L7Uw-LZAk&Wqo)^_=^E
z9=Q82|KU@|lO&Ep&V-Rpl0%`$b6x0#rqxM2$4f)ym&=eC239j&8!bEBa_M%`bdU{r
zAgwZi^iXi?xLF#9$|^LIGwbfzy?^_T9s3XM3ykkdFQ`-w?_ccqT3gm{Fk$X6G3fS8
zo@EgZ`uRZFqeo9D?d6&QNs}9knhR!sY%aLFJbzLO7C4+CW`oRSyjT!T-{aQ0j9j!O
z*KwriN2w#Y){2oOYppVX#F-#67PP}*kPMdFAyB^c+zs=Kop2=r@S~WDh2i!d`;e*l
z{e~2ByfN)^f7`b0Ns=5tb}Y*i&Y%=HN!abQ9X}{Yw?u>>yr5hquo{UYan8_xB3hd#
z5rZu3t%QU2@R(7@*s}C1+^=$v=OB5(trarYxUuNV#2U9!uWy|mDY59}-dm#XN*;%0
zj$Xioa7e0T5ScVLxiV=22%IqnAO&N@hFL0vK-MtEhu#cQRvW_@7s!kuFqU(Mh+qI2
zV~9p&N-K8y=ttlLj3Xj4X076Xj9DT?lw5F^Bcn7AxygWBrBd#CzFT&r=y&Fq<3X=6
z+L+ql*GE9|q3^^P%@{X83>bg~B8~z8hD#;@L}{92d79^O94AT2m^*aeTWbm6)CmUw
zoO6c4T8pO~d<0}Taw4$AkRe$x#tb*jipQ~&i*MVq!5B?|#3knrFlo7e3&0uU!xMBE
zG6G->8N*ZW05C+skb{vx0u6y;U=e`<O3r|6ceQPdC4;46CC?NCl$=X0D#g<JbsJ`<
zW&+0vq-<2m<0JKEwS4Zz_01xaWT&SlXJ=;@mlmYwY~6ltP%Pf^ty?Lxg7YZLm~@bk
z6Cwcv<P148rJ@uf6HHlu%4en~!~URIYuGI37Pzqm?j1jPq*x5%C}xNwW214JO4k{m
zo?KXItuD7b*Xj3rl5qxf`>nTb*}SP-sf|xgW?Am~-nr+U*ADx+QOoU>+DQGD+ioA5
znOa<J$I583wUK6;C-**Zw<SAwa;{vj?m4jk@bTlLQ#0KpS!}O=CMs8CnIpYop*GU+
ziiPQQ>!;VvmdX|2Jc^S%Q{_rEj+0`cSO|)q>yM3&JHla%&(5r?SIgx}S(6ctSXx?a
z)a$<E)a&&^u~04*j~+Q3hl5hlA0KTZS<WEr4+gz{o@EnbW5<siD+U249Bno?ZeD-n
z*ipyxR@<$xKVZmzFS+Q3r$jZhTgMOF^SMz`?6<?2omYZ4;a8{qky&C;ECl^-|NaLK
zD{wk}#n55a_^#tQE|<*Vi~(4JFwJ%!KlbfA@45S~`vUGX>(#OG(G8R9?zrpDj>cAB
z*DH-qH#&T9rP-WbT5c5!Rh1Ed(3vqhcZ6+@7Nw_W*G)Ztc<4|N1m!{@P2<21jLw<h
zX|6e!Q5=`bCA4TY>-57Yj<;?;=jf3WX{H5n-*dx$Y_$>G8${Og>+>sdqKO3_p*I-h
zS`%0RBq4dN7L+Qk$)eGE&88&K<reEfsU}3t;Wepho+xl8jFt>I17j6u!Whjs1LT^(
z0E-pHiY(fHLP0ozaXn!<W+uyZo@Q2Srj@Ezf?^<z(N|pYlTVLDpXTv_`CB;`?RJ|i
zgi#DkG)6|Xu_sQRbUfD&N`+#@5=$~wu2i{X-Oef^sXQgC9hU`xlcr%B$CixFGZ(QM
z1SR2DgF;#QnB+Q2;%+E`H>)GfYNOexjn*ru)86WTLNaKh3LVUujwuWCq#wpT6-PRa
ztjYuf1dBibmf@*f#P6}4$pQe8rH9Ue;fy1KwFWE=6I%GGO;AIhxf;5Aj4=-#E(TzY
zF$7lITq|vqHkndss+BUiw#Fjaz!hXvlEm%qDjLJ3$GAWyEOG!6fCz-ZkO1+lgAICv
zcyVce>B!;3bI0ejR*WMNX{`~_o=%r!j9EL>{KH+nr`$4}a{>aO;fWXz9dHJgUiPxb
zu;}@Y^EMf4l?7x1gbPdpazG~kFK=%WYioL)=dJa1f4keW_dd<N=l1tjUqzNgiYBd&
zv?U>p6dP28z>^FDC;{Su2LYT+oB)X(z!?~SNCq+xCo$kSfe}YmED4H}P!bK9j94rd
zi*>8Hb=%W7yZ_s-T^a0KW!eK$0Sf%aXEz5Q@cwJB^}f$S_WUz&^BEg91KZ?{H*Qr%
zfssKXA_D*nkRu7koi_8?Jj=4lXm@XK=PO_S%1^!d=AhsE;upVg{rdH<{rR7j5N_;$
z?e>!B=9S%QHQOEcUA>;4K0ZEpI-f5uX0vC<$Bl0O;;;PjhYvnFKbhKAGei|9Hh2W&
z0gyIhY2bp=7(VyLi|1$0RD@gI?Cy>0Tm9}<-nn(<%B{Vt<4*VX{*}Dn%ko@BkyOe!
zdp?^+X)+iL*NeionnQ4{VIlOp-Nkf&>Lf>{s<sUb*G>KG>^Sb{*YDjrpI-DQqoa$9
z`FbtWq$TedR=SK*5oc-bj1Dv}i)**<POIYj-CIRl<%8bNm0hlc2d`pPw@sX7E`%tK
zopX2Z-2;TSZI`n}F5^T+-7Ggv8%q_e>ty-#@?ta@IqwgSkH7lWug)%~Rar)VQu^RF
zLG5C>Od{FV6-PKeIC|~1*H_Eca=8l54|{`?gTq%|eEH(y^ziVYHD<iMeQ|!FwTV^y
z=I4LqGm(a>m|i~p7dyk@;qlox@8{VSj07+Y?M$XwH^vXY`*z`A-MT=?1xXPyE>b`R
zE{{YSDWz1cP2K8?y8gzuzy0xlM-Okk`0~9sU;Fc4`uq>xdH3;ov8+v7lu?{7mg~CK
z2#f(Rj**ZdxWORb8+VIpk>rY)s;d?pSqH0i9i>_AED#eyWU*RRrE_3BmaW;@-f_AK
zK*soCCj!nuMpdJWisq~4@O)~id%n<l0+Gzl&Xxqi1>`~qhCb+CFCF!NyRNa^ORsGw
zQ4!g#h>sIUCGrpwCc0V5(FwJ2XONOo2;LYiMBF;RZftEFa_A3}y{kR#-NE5$oW^b2
za^RD3pJPy*7<lmXb3gwZ<MIAya$i6CxS$V~-zUJbstJ*ch>Ivr66Cz9+g^7dWMqst
zCNLqm;6kvnShh_Sf)+e~aDjn*RaHnq>z0h^r&%skhov1ENe8}cZPhf(QUr{d=p|Vm
ziBwXAR)SUN(3GZ{`(~}%wQ1J2t!-12w~RahIkzFa1tQ{XGj{bMgb;#r&ijB25EwEL
zLTnr<-g{3@x6KADfDkxih#&;VjZXw<qw+))f(zsUAOs&Mc%z+vw%&Tv8r!y<$a{wn
z&|8jPat2_7)r;v_xoW+$Oi1K{GZ4?ywg?P=*m)U3DC_p%;Na22htHlq_1+1|iD*ME
zgU#HLu^)QyeEKANuIU?Y2EcQ0@#a0rhp&JA>%Cs@N8BvG_wpSN9C^Zc5<nnRObQOb
zz!L;Su8kuuh$+v1lqWzT0}upnO)EKWdPl^V3tn;&3F%wijpN<%)^GgUuYTnVuV1^e
z6TFEdb#4DjEcun4$?W{hG*z$PIXyeRc5S~u>^mYEMc@1W58itB!|%NJ;GO#qFXqL`
z@!7-sA1%%<E)S1ixqCMP80JYin~^n$psfj<1tFE-9qF!rjj}<?<46p9qde<6=O^Rs
zUayNt!(n&-*0s*=zyv*8UdE})vM7!PkwuPqH`i^wH5%FAC;R)ey6*1n7Udd|qAW4N
zEs90jO|7>|DIru8CwZO^2K_XPf@>qicDJ`<6-yz9gW-5I+8PdaCga<;ZXKOGo6V<B
zpFDp0^wG|L?zXkMuFGz>+e<TDmgH<VO{b@)oiypBNxfce4f>5~YSZ=xJ>%T<tNZW1
z^Znhu9nSf~j~{lj?C{_~x9xJZ3?VR3PaYjGKqkdvxy(D=)w)0iWmy<!v}wEju18{$
zpUf`mP%reFBtAZTbT&VI@s)e)dijOV{px2T4OXvztM!Lg>z#b?=nwvIi<s<0e)d=&
zJ%Wo%4DCdzlj+6N>C!<+Wg?l61V$VafeFKi^Gxtw#PUR>5obcoTl0-SdF%V{e*f0?
z_}}@ZuYK?P?>xJhgHqjIR@SZa5JE`eh#|JRWt<O&T?QtL)%y<)>yiYBPL>^=uC0r*
zIJRwn>nKEV%n)3#jd4}$JDnWC1Tb||Z%_J*#kww>wUlSQrp4p)`f`bj(ppCdnB|?i
zZQ3R<<boq|?t`;UD*{Rs&iQ_pXAruQdJ!a?VuVPrSTZG0aUq#x$jLh?g|#*?bOe<#
z8mO`%5M9~p_WQmzrrYgwvn*0^Sys1i-oF2T!DH*ehftR7SHJe}ZEs)uOz!JPpT<<*
zd-TozVC%}&YbU4YP17nVvn=U$I;STm#%dv@^Uenc2to)Uak-cgS?~^#rkAtRlk;A0
z7<?d)f_TQ_ao*2(LdM!k+p=lP$~GE8@YZ<S25W+AgEPUIx?Fi(I$ar4I@?%htZ~+V
zS|5T70ccaoJ}+{;bI!Sq#5?DRHgazPK}hBNW)uS6`{1mz##sXt7$Q7x;owH-H-<*<
zy>re7OB6Q#b?-e<AOh#SwJt!g-Z-bd(@xj5ZmcsQIOkeZmsK&Zi$%3w1gj+jDI)?R
zB8MB6&L`D3koDowqlX`V^wIg*8G#pnf`4ktbl!&_GUFrLbmT(_4u}vLJP!{6;`4_A
zY=*rd{Q9r|`i~ZR0gM9)L=KE2@aIN-1Y{hXcg`x#oYhLQSPEk;m=+Nu&RE3Sx{ReH
z?>TD8v74w=$=81Rb60l8X`Eb~pOR&!tw+PY*5>l;bbB<oI6ed)(l|PJ@_5jH?aXGE
z%VI5r($-wLe)HMc?CHfL9e{-<lI+>T#|oe*tEUeiO-7?&?5NkbM60S25eLqZGva=G
zX9vj-w}y_)$#lAxM76WRX^ds*O;PB}W?h3H4u_ptE~QwnR#*0}T%4Urfym@|d#r@)
z_Pf@*doR9Fn>LL^U6q0S{{EGJs_qlRs;X2J4JKV{?P9S|N*o>?-no6-I=h_BvLs*5
z7XYxewWWmk!8`8|(Yh!m<MGS4?nF_vx|n5o-pzZ~xCYAo{VVr>@Lqi7N~`PTd<p=P
z^Rg*4U1vO7?fZ|d)e`D4O0C*eyTmAJZ`B&L)u`H|_B!_7u}VT~wg|Onj7RN?M^zLR
zG)PH_y~jVl7yt7rFV3rc&UM|_^<DQ&k0Y$FaH>wmbLp+SQsm`sb?<Kc_7Z&E&vi>v
z=;&W~cOsQm%t#u*?dV{WyQrX`(3&Ce|6&5s4#NgVT-7I{3M~YSMhy78LeIu((Xrrk
zs=BT)oR`Wse>b)lQHeD*-+9mKDqfp1-K?ElZ;B!Jx4H?6r{Up72{1a%*;U`8Sh{h6
zDGudJmCw1Y2$hU9E5bwmf3WP~<L~W1bb#)RIs;siq7^VJ(%j#>^Q*vt)y?k0EJCdJ
z`gilhtn00UnemX}=kkt%;>?!AoiaOX!NT=-S~Uqy`0&?%4_}x+{#1A`l{&Z8KUjPU
z>IBiYT2xQoe?}8MJnLHWIHkxb4fb$%Ce=Y=4mogq_M0~<)44}9hOV}&Es8ygSU?rz
zc)(eQ=btX|q@16f#v<?HqQNQb`&qo>3Xk~a7DBusSXU8C95|Uwn41^JDB{@B=)+v#
zi8ilw2r!;QLBEv|dT*}wt_4}hN>sIXr$FpC`m_|^Ek}H<88vL?z^Yb>Zk5OUR!h=K
z#WsHN*1DudQNy<#*$9ubuxScxU2x-eK#E?UHgxHAL%V#jGmDm@hPvmSR+$!Jn^F0%
ztyRJjuPHtx#~hQm(sQ!rhI>NZO~T=m#OlZqMMesV_LxEcJbWZ$kZGVXb^69JO?^#5
zOWiB&V8E&tDa8nzE{btT;$i{$qDbR$X{-cURYG`d66{@N?g@vSqdqy8syGUOIP$8k
zGpcPYn6wpr78HMG?PF`!Y}(l&;^H#&XUjgUukXo*4o{(K|75`$ObXQ`GIk(M?gGEg
z)wpa}{+H4lba<&uq2k0w{k0m{b)8C)NVpt%CriOi<rH^^g!Nvr+#?bbgie{Z1l@oO
z!#{-(aM|7fEmdq}yEoiX8UzBoZHAX>#V~r`JA_ge+r6WLCf3UeJ_VFh6bw-DIp6>M
z2Kjy6t>cNjlG5b<FG}TN8|unBPos>?<VzF=wd>*QClNH9Ly=J?TH|1YvFo{sO0{h`
zX|=mfU5>c$i~*&Qd29^&U^iN)nRq20lycf3BcShwd%CS3-wDnW<ogPo&7b+S1Vn}o
zgSq2uF+%3GUKpViho-q~!42jDi+oRSZ#&eR=1o3Q?y`4Y@^Dcmn$JH%kvk`$#{Fd?
zh$l`VH9WLmtm}jL7pbN)_92Mz@2ypz3KXmyxxY3lYz=MtK&(XzuwKgK_cxZhJaNvQ
z(M+z0#WrVEC!MAz?Dc>-z8s1zZueFWIoJ0Qg$P6qpv_>z#6JOw`*W4IgoI)plULyL
zeO>LA4c@lPV#Vi&rxnWV*4Bomb@E6IwAx6ls=%Ve+uQr0<8X@>)kxoIvt7jLpJ~)v
zm&i_!;`3&D$|(5sAKB}DJ1;|=@{t3s@~sh+QtnS+U}~%@;O!Rq#t(NDn}|e9G2bEQ
z5Q>&0$Hlmiz7%@p)3{cm$V>`x!{2qlPd2q~n&GBb<v8Hm$0NfHMJgKQv9hlDls?7`
zzWgn1=&i&|qrhP`(aJPzTl|qO%|KCA#D1MlPKBTw!N%sytK><#PScOVBcJeyMu<0F
z-0nSuP-hOzh8l0BIES>`m?n*~2IaGpV=Xy(xy3*5>Ub>D9-ul3o=j*6o2pkm^ex;S
z?N%J7=Di03S2)FeCD%bAA8sR^3N=&ejxu;eS<2cxb3Z{+Q-<Pmkp*5D1!Q6AHx$+@
z2q}uMNr_j9TX>6)RDCX`lJ&yY+8Tv>hfAX9QUu<YRTs}4w^Jg^g!<+F8A>ah_6iKO
zmNlI*iM|Y$+@k9m?t+YAd7W8gO-Hej1sb+tY%x|(>KfN#SH3BD8{m=NZ2S~lBx;)Z
z5J@l?_Hds)`WEfK!@6Y;MX8@r@=|jM#jQ|>Q%C<>B`xq}lqGQm@r2LRxG0o|U`2;2
z7Y7C`=&V1yqvZkt^~Gq*Q{G8+^=}WOsd$)0aKE4`1JP2Te?c*$ECf<&J~9?&Qt5!*
zFMGn{g3sOu+fAjo%00ED3p}_ML`k`PnDu<l!<lwI==pKY{=bQreAB`BV2*ed6d4cI
z+MQMC^&QH-GlQl^W3NANItG@g?eDx_2`2+IqLc^4&A_)~3=ZIp1`DawQE}iqBt5_?
z`iadqjorD1oq4waKY#F>YE^E5zb91YCnxZu2Zk=y0y8V^0$8jIl~UaH_Vkdy*8?+y
zMd?W=$(_I^H{Xy*Kk9*#um_ysWfgykkhK_xa)<+bV!vW>EGUTI+1(s(>2tO_FbJ6M
zAbtG(hC2)MYMam)dNs3eE^tKSU_YyFJVG=KX@jqU!ylb}0Q~&Yq8d}Gm}d<0PvmVh
zKj5n#*S-e&S<AuaNX-1^rjIx^Do870CKrY!ju*GNA)V`8>cg7l+4ivTq>8@*@{8rZ
zNKwgu5DS4w+n{dv2LyTG?x`zaF?$y`*I;*oArJ_q9Gqde9Sm|JKKSwvJx`T|S>0Uo
znTPcKS<{H{U`7pN3r5|~v*Fp*t`_;lY)=)GnO}H#6HYayhC0@a=xED$-z35IX0Z3Z
zC^7s?Wo}BLCY<8%ZerlfuzHZ!D5t`!$g6shlXju)$pW>?cW&sW6NCwP@y4}i+xDt1
z_hX@G??Lx6_5GTQnQ}N%ws+xEJAV0k<R*+TYF)0FsY*V!ea)pT#*}9;`a?JCiyD36
z3Rjkj{eT`fg$qbDIgU0<Cx4IkgHG5LdiPD|&v8Zx%$l*Y^~xVX7RTIQ<a-?y{nU8{
z1<k`$^@u&%Xmahg8p{lzkl#4x>%rM=mIDrhRBz<b^d#HGKwH)8blQo^=&Xf)VKsx7
zUgufG>XnQAy%E%4|E07<__O@H183P}hJCqMUnYT%+k)D!RzPcH*fbDXv^t8V;C4@7
znlG<EZ#tD>^fNGwEZ$ZY{wv~WHF0WVrtR|<9PzBv`Xs(d1ycmn%^2%^9WoMRWm6SQ
zk<?+pS<jf{_*RzL1_?S#|1n}ufsAccAz_nQr#>yR^gW*VfW{OoHCL6r+}F>4r32JR
z7_!xkJ;r5P>TL@CO-PRn!EJTXa=%5u+fnz&obU7b=1Pu%qa=7$II^t2e4`l-2Ym%8
zf#O%-viHTgcqjm;G>Kt|?l21vm8v(~s|W`_!5M@ns&EGr#^l7v9obpPaK8=;-rXUI
zdUvBnp%f*s8lM)O6{SG5<vLn}$eCaP+))A1s`ju{!Dk<HTHg3NY?WD{I#S4TBt+x5
zzK1sjSuGt-zx9a#aOX$0oR?gslP+tDTr}L^;I%2B9q{Zu8~$Nb3eJ&Lep2H7QT2Gx
zanT5eRQ|mi)_^1xWHlch+Mq}x$>w!g`-T@oLbh;3H**_aqZM{hB^_YO{Afnng57vm
zvz^&Rpa3^>)Md?VY9Tu_<YsJA%{((y1J+Fd|K_#$=E8{tI0v@3J*;N#mmi3{%-#x6
zaLuU$%-wMwR{yAe>XOLt=@f-9>+U`Z3qyA|xg{6Ado8^{x}QijMQLSKk~ONcQSWtP
z){IrmRO?Tjd1?4|qUtgJHKw=()&$b_j|?=fOZBt36BaR4$-tm!;yk<UudsxHYH@01
zOI8i7u*UbyL>#S~GasCVwswWC-_|l?=8O9RlsqPPj7e^K6-#F|_^i=VY=eeb)n*QE
zs71{k{|aHo8F4u6BW+oqZ_F-c?(P)4eSTLoIOxA>mGw|~wS|B|z)VJhm8pfZE5nE}
zApT>RSdTtV7DcMmSI_zok5FHr&RwpeD8rMsb_$%c&CcJyKLVCy47eCzxFy3PXXG=$
zNM{Tz*Oc1aytKVN>o-5yl`~NrW=@6dmC9hJ*tQG97AW6y-JNY}50*^pUcST;TD}Tg
zTxDXe0!5mX&j&db73Do=#CzHbj1o9YU>q_^IWg}{k|bg?80b|+IvGZ9#YL$%YD-d*
z=qqTycTM+(SM+!K(~kZW(SbcdSvUeMD?i7i7TZ*+<_L8Yc`_<Y{&?z+pyD1Laa01*
zWK3m^97ob0l2Y~~Owb*g<kTqbylu5z%pXk39FwD~%f~`>TQzG%)rVe)Nv~wct+v=~
zf`^Iz#EUUihqo>FukU?c-JAk;Ih;0m6ikAT0vv*;21kczSz6&Eu!Ci;x&X5YFLaax
zUTRutS6@?l{$94$?+M4T*K&ttQGkk@<+lDQjorl&9A2uO+5338>*g-y+r>a6%r~yU
zF2c3rNgDKUW%sR&olss}zgewYXs45LAQt^V^xkI%?t7m7?kWy#cdMqqeHY){V3|#A
zDiMXX!CrBF!OZ2=<J(*h9e~Cnlqme6-k)fRJn%h1UV1W7>JaK@+B#ELoNF;b7&#Ys
z(Aup(FDY_ik3@&~cggP5&E9dHY1u1jnWk=Gb)swn4TR}|v);bccnAi2g2@NMaA`pq
z|GxXn4C0n_Py~ZN1gya*K*MoCkuvax{E0R%|HYhkmT(3M4NPvTAK?49hNh_!^iVKs
z`m02yiB&;nsbVigd5TAaI(jq&H~6miazm&2GhVjrMkphAhC=n6aUzX~bY0F7LzFnN
zT_LOKvMl#Td3kQWD(kQ(L%X}Ap1EAFWnQ0dmTT1xskU#dosKHV!o`?tc|hdkgoqDW
z8xQPzJnpYJSTM$Q0M>bML)rX}*48Jb22)d0uqOjnF6r%oNJ}14oP$kWZmudds<F_Z
zn)xL^(a(W?X(3DUN%n810QAfODJo1#sg6ZNmW;3ZOuQO_4AFa@n99C}8=~*K)K*#K
z#Trii@XW7hOoDJhR$jdE?UI_`LsKxr+EC|h(=iiX)ia#Hop`Wmr(v!>Htu<g!34_z
zdPvu%q~xS~>Bu3EpE$ir4~tVGZM_>``R^+d{F|k_L$-(i`Yqu7MNz0Jqp6by>&Z|!
z#31ylH^s!5srG1VfB(CMENhwrw|T1bm$v2Yqi&?SP<8sWrTfwkgA%8M`M;NI(hRzO
z)?&DrW(NRSlBPAF0Lsa>F2GlPo<bR>`c1ws^zbe%f!!}p{9j8Ld2w`r`g{4kZTlD<
zEK!W<id?dcr4C^TYSx!Mdei{OsCpOR=l}v|p}+)i5HGg$i@1k2TvuMn_Wp!%;11U9
z&L)?`H^r`WBRySi(ANrtr5m3G>VcP)l#-jkSH*i-Ua;{$d@j@(d*EC7nu;4D;^Wb=
zG>9PeT#ZJWk53G$i-&W%l6}^|#^OvuLcHy$b$Et@4532Qe%IFLc61y=%*>}!eA}}%
z(O{Pd8(Cp9$$ZoK9|$Q*7##@<v_C6bKocfNJ&>D=p-9LbRwR6y{$SORYR{Eirv4oW
z<OrY`P76)e=MiPZzVL}ple!&u2clMK^7HHHEj1Ow012aJgC|i{?fRWCUdH{;=YhAU
zD|?O~`stNV|3p*7^8^iVNZbrHL|zF<rOj4dV-lC^YMNYi2c8v&cES*;>$6vnoGp}}
zLhf*D|C49BB#juC7mY)|UHb-Uzjf{dsQV>YC~qE)xSGFzZovFsD64qK-6Q6$n$P&E
z8vLvqQq+Bg)Vboj;?A1BASsF^h#O48gC!EIAeA*)N_E16r*<s+J#a*>U)*S+$#=-T
zx+7v&i#b3opf6NjW4#Yi^m0)u;GTA{453-*1M&b9$~Ky~LIRQ_Pv~bA*96iea>@TF
zo`RUNpHr^MsLXU7brI)vPlH_&Fbj3<wwuJv`U}YyX<+@CTm#cTZDTz1_(PY1=0O1}
z&&Au^ifSjYkf>5hXIcJngIhNZ>h&NNMcL(Whsiiju|u)mGk)>tG%_!SYLIrCwelMU
z=(O_H_HyLS_0^ewqjj!UHNc!Yd+sma2Wt>xI+&RMJr%M3sC*Bi`+Ixa*a89NZ8DQJ
zrHA1I_E)}LqtMd>&GY`Yd6o9C9zvi3(GJ9St6%PsslOcZKg%%h%9CLT{&kyW7;2(R
z2II#Kthx{dHNdICKYXJ<Ly=o+dlO=q<lk$4GMNwahW?w_T@RZBl&}9_ne8Ge=*?=^
z^R8;UySu6XAnh9FmTZlvv-y&n^GUoi1!lVr(Q?~ZLB29VI8K55^?#h?xGp$qb$Sf7
zV_U|v|2kA|>AeTOYiy=P;g6X4gl(TP%b^j<FzgD06)>u6OnbE(=^b>kf4RSQ>Lke#
zqyTTLS*+p5zi|PW%dhQyCDJ=s?k3c?uElySEbYl?y;*D%s;Lp&D(mTi$z%O%6<jfs
zl0)(K_QmM?-L~~a4_BvG1GAu$D|=5d>W$s$rkKxwYSe)J3f#R|J{3Cl_Ks48@b`dn
zd==Dr$Zv`BP|mSy)~tAG43)=bcdlPCfN)tpqXJHF-T<IvE!qv;Mu}a*el5@*r2e(n
zUBHEnEfOxL)0uaaAc_%$@<B<iM6PHClF#;yYmH8D^(6avauh2SILTTZmDNMVWoX~Z
zzn_viybxf4jKm)UdviBiSA^@$p?Ot2|KzJ&b?O9?b&Itg?=-#N6`q#255FT916mIa
zwTw;V%>o!kmJAs*{$2P_1_dDndoBv{$?-79z|>ULb9Mcmfx`uK9?Zuz_RM&i2Q0a*
z)jS$h*)Oq=h#rK}7_$Sai5Bjntql)`B&Oau_v?<0YfdCdP)1!2F%v}1YOC7H4R?^l
z>b&Y+TvWalRw;9&zzH<hJSr~^Xe8g)X#HW?ce($_v$j+*Yyp3S<@G9($(Vs`Q{0Z3
z=&HvzX@Z7WLvgK!Ny|SmY#L`n(AdykQax~g7^#YpNuH_c6QH)qT6CgJ^t}mBM!9rm
z^Dj4V#N7clyqsDieaLN8SlpLQ;R^%NTy5$~6QU2@8bXkBvHsF!8|E@bwIst$?t-92
zzjbUm9(eu*v#3(i<H)9WEr9P-+Z~q2{eAc2j$;Y7n%|2&L!b0EKK5XMOaM$<==;&*
z$gi>fKK3%<4KT=3fRHg`Q*i*r9H^kE8Y)0kt}sLXZq3B%sS*P*vBIp;S6oJ4Ww6)j
z(43M)((6jP#E7nES=W^zZ<}Mb@OC%8^=v3!@Na=kYI0JpjKkqaVwWi8v!#QXm%D$S
zBJ`nrtd!^$4lW4pJ8gF)mdXB4(Ei!bw3#v@<@VD#Acrp-v{<7}Jc<HrhI1yjBz-VH
zhCV<wb2N1Cms1TYWtI)>9?z6RN2X{SKg&6^Ll5i3-qcDPOmuY>+ID~`V+X=5G<BaK
z&Q@aS?0|L5b>5>)Izw1<TC#$y_hZp&==AQ&9#k+d$hO10-E*-mhDKHUz*=Y1t?y}x
zV(2B%QH-mdcMEC?3k>A%%#16vvBq>nIyM&7-&*L}shCIGCyDBdRj*xL?_g#|jOy+B
zyb$TRmftooU1P_B$-fQrwS*V$-|0Du_3{q%4Y}BuEU*n;wOT$$Ut=(58c&K1?IO<d
znrl8QdIWUoz$#umcu)dTVIK*DZg2W`2@3;FUM|x79~O1|`~usMNGsETuS4rYL#^&%
zS?i`YQYslNai12MK=L7yL}v27x4D5FN2c=mA0pu7B@s7+EjQ7`Z-y&!5YmP*Kj|=l
zvRQUcO)(DL-6fbK9bgHVeOJkEgX&mFGvkM3(&=?S%?*XW#t&zA7CuE(;|Kr&*d991
zd89HCP1It;527VdS<$|2#X|Q)NMG-c76x(Lf~r%m+Bn}M(RR}B=HtyiJ-)~sTE1Go
zMgwQ<dKG<VveyZLfmH_Mys5M6Lk^~tdOEvz9aXghBSSlov$be<?_U+g{I37fR~J;O
zMvt0)BhHw)L$KxmQ>x!Dd{=$~+x{s-U8OipbGZX(WQqFpeE93f;=|Jmyf5L}5(GX6
z-e@TS2*ch=w$Ai|J<L&;T!mI&Qr(`5jOLp=VN8M4=#IFP&cCdIFV(>V&=X2WO_ma+
zr;?_#Q=#h)t2cK_Ql|dOkoZ&Q#~ef_aVM$shkSpgTT~SFs@A=aI(xhV(0)4J)+!nd
z;`nGE4T<?(_QSa-sIqc%KHFB*#+2(5S4YZB$D&1DO4E~r7tDDLcWR~g`TbQ)5$=Z7
z$)HL`N>1OBPyTAGgLheehERSb4wA_d8C)9Q-m8CAxf9=r3TEzaH*?4J9H?`p9WimU
z?4pF8MN^R*(z*`*S{Ko70Lma>T-vR8!7K@XTdIBZJx4Aw7g#ivJN)4W^1uB1?rH6k
z#>0uiaW&KE7MKg0cj}So1S3}h2p#g~vDUl}gZ~eiJTG^`(J{sea(|TU^P)V%dljw$
z_#)#z8#zfVXU%Wfylc&qTVqs=X%$%*@6)@=8`S>nte~HFnO?dN0?&NLmaleBN)>~@
zk>WrwHCe(}*B{&dxC08uSF*EcG%aK~dFc9!dCAipcIk54hL>}dD@iM&Vq*F4o?DyX
zby;IoCd1`Z&nh_6Ly_NSTjXEH!lCo30h;re$;{=8XT0k9E8-+<8j9r;&CvBwQJyb9
z?k{$%<k|CsUXJap1AN-Uj1%4P?d|Om!qcKV-RyTD;s$x2wSDFKU>iJ^2)t&4>Ta7L
ze!i!=Oxh5YnZZ*T%n@Rj?ft>|Tf92|5@IKeI7d~6iFGuwe%*9yd!FxQJ8x8Wx;Y=x
zXkDr5EGZum`d0|1sIbYmv=Y9c1C?G_4BMQasV<*J2mhx4D9gDiR$9C%l3^Nmhw?wg
z*7?`Z;RS_81`#LAVeP`Ynjm#AWN*O1a4em2kMF_8<!}F1TdE51+y6rL5Xx6GVJACJ
zG=x}*)S#xj`?g!>AqiautroN~a2sCW)pxbj?){g*oGM2X85S1SSs1Z<y?K2d>HV{|
zb(5|wBh&EdatIENL?Y)sj`t_Ti1iA_U6r1-_Qm87Xurn_&_#*-GlYH{1+1nb3kc&=
zOT9OL$o|nV$~5{dZTol>iZ<s2T37una&qE_1Ch7WZVuWx`j^j^ZnoNq6?$QHvm^^#
za9$2BpIzy_$PnBPP}kI0&0h29C7qI&@LT>>=>Jdh*jsQ9PVmwx@go-qW#NwSW#R%c
zPh}W6eV0QP-bRLBReJ8EdzRULe_+$ScdT#%<wZNi9DzQTR`u4GD`o(4Z?eH_G4Jkl
zJW&G8iGE=0XJzmat)sYCc=tQ0<-YvSVJM${Ty}2ldQJNSi*3GpsKulJqf&$Rpl+c-
z1o_-Q2uqSx0p7!CRbw-ymA7Fhm8%nUoKK4knQgLPWNE4u1wtV~1I$BRi|OKqCc;)r
zIVWGUU-}pnfBLiaHtf+B&>)I?ZP602jXW6;<fshUp*C|iZfo<>Un)XVV&(Q;@ph_x
zWLx@>nKzMVk`_WBk$EB_o>2UoE$S9=3;gyFm-dccl^R3ohhe2gTj3GYT@;LLR6VpJ
zuj0V{6e_iA!>V+Kx1A)(-5;x+g@2lOP7()_gdAcLR%AV>qS9-DM8znY0f^vz&xr(`
zO~L&_6VJgyy_atun^so!vyeKH$}OPl%pT2jd2y099!$(!qr8B^rLM56r3->|>+wK{
z{lMa65iO~G=^91kX}IuVY5wul!&V^Swa9_tqg&;O>vVjW`AN`k?L#U!@oVe3LMgz#
z<`k%?T}yN-w`u@+d3jBBm&w*gVcxyhx(s8y863)?(%rB&gLl{kgUT`l!k4`|b!dNn
zX0WN|JvhL|$_1CY-X&ODQ$Oy8w0&Q-n#S@ySW!5ICk6VcJ3!h-DLgPn@*iK3a=}T4
zvE6~#%)As*m6;Z5mLcET$<S6|vqy$L7w329{yc&{7zhSz&?5hB&X8HPQyG<%aFmJ<
zCk%4a-x8&zqw71@>-!;Xz^Fl#B)k}(n3yONOgReuNifoM6mp_o-I%oUtW7=Ys;Rm5
z*<D+CAF#98?1J4Ze5b^k8T~dWC{Y64=2=t)-RtmN9C9c*>RR%`jX?#jh-v|Z)rEQe
z9r>lmh62EOerHEC#M|m|b_lNIX(WT{+eMkFcRZF2=%r^je@+GUW>P06h<b^Pk@7Jo
z$<iqdvEr;`y^74p=v^*PcLnvnFKxq_?Sfw;ZS_RSv78a7(}S5Dbxbs4>+TK6!0_-g
zla^N^@PkZF<)=zY&xXglSTZPOy=;O8de6@vEdy5=tI|B_UBmp#!$<Q4wyQxw-jRC&
z5x#+um(oj%oLB!ASiQ<|Rf+o~TWH0J^JV0g@Oa=>)=zP3vs0SHifTaZYkbxxMgnij
z%Rg1VHNpHsSg#+(4>TAi;Ll3NKOG;_^<HW9UB2tv?YlO*`7>KA+vkrP_LoU@tZZdU
z>K+7wyJ>>jewTK?+s4`ToJh43km8+^I$^hlSSLb_1sr%*RT;|eehc1A?&R|5crxFu
zaR27_gxEDz-?8^)>+<Df-;F}w&D;%i`4{U+?*TICJL=1zqgT(v71%9Qe^De)mk`bt
z+USIwv7K9Zg;FQgXAoC^Qzw;nXOJt;+4>$(OIpkR3Qs}i?fH4q%{e|dpp}8Eq5$5p
zT(>y(CDpPo(6Rezum6$J+d)Q0Mf@8yD~_zd7sDCqMDoR}Wn~{d**$mcE%csI!(vWj
zCSTIQmp3bfHx)xHdn}Jj&KBA<9bxsdU44!9gNcZl>5K?k!x47thrD2GTfgxz?@J)D
zsnNvMg+1av`##3Xz0`;+TOvFoa!p_8AF1Q9nrGBG^ba{<YWt|R0oEBAbPHy|9xjpY
zX@$Vsr`1~3vbnbe6up%?Am`Cy%zA1!BOI3_iTbVXOQGa3O2rZ04@v_GiF+WdOg)Ko
zOd-zL^RpY`9qMca){K&ONgiEX0iNn`r+F=ONg6a3mWbk*2W^jMO2YzzB>fH4(2Xi<
zYP*qx5jRfO=ZcF*pTE*@%Lq>?jg2Pky38w(@)-EG<n>*j*P`Ep--bc#^$m@ty5N|r
z6O3KM#8|KvwJ{^U&9h8!A4MB_2FEMN1_>Ow(aIOu+S<+|B^NNaxXv?xyu7DRl@X7M
z0}{1}s^x-rL+cJQWITk`weYIcFAy5p`h1J@n*Ub@*w{48)ecYH>3%z};Vi8YkY!+8
zCaE~Co~pynwitPWAlP*K;{Kef3U|J05Wo(&S@>nl9BkVE#}?RyOQZsJ4uV8cMwGE{
zk(LqbG|KP7!+!zEGj5&rseXzar7+m#H|E~p(9m<@;yzhm7x}qni&uv!H9n{8v%AaR
z&{n0sj*f^E3{=3#HgvO6`3$PO>D1|dv~!Hgt2IZ7iIsA&*Nu~cE#1v(XHh8B@!N3X
zqVJVo3i~nVWm9Tv-E*|<{Ya=_!dZ60$j~tD*xA@40Jvy0S$Fe9;y2FRWJ>k<y?gez
zt}agUvUGCnKXeD}+_xFgB~^j(>lCqbaFD8q%mza`0=G3wa0Cp{FuP52vQz0A8Xo?)
z?n%HqM*y%fUH7{hFX%O4UdNzzg07bdhHWRXL|Xvw*Yzi`+qFp3xZ<}q^~c<;lMAB~
z#j|~&pzqQ`M-5rndvbB3*J%AC;Qjbpj9JR-c7<=<dw*B1k4L<3F<~N4Kd4N<LOw}C
zy~(@P^PiHbWKJ^a2Dn09tWb-z<ts@vw~rQ)>#w$z%|`PwQKTB_HN!`0-`GM9JdCfA
zH)l5xB+y-llXIL%b>_b}G;dfvlX_4%X2K8XYQQ8s^^0^t4tM0BuHPkwN41};WGb4L
z)w>&UbvB{c8kQqYsOac2Pi!y7ao32O)4uD)o8VS2@?!H!64aRdKN?+L0>(Mdo`Xjv
zp4}mpZY3KwSk!F7l00}#t-=wlBle2RE#hU1fosn#u8O|`IiuOnlhYQf^mD9b6S{r^
zTSMFUIen|`ql++`qtcsMAk5{_O&w7M#ToO#hR@Nu-bJHvUq$4NC<LEwBC_$+-NQ!K
z65mySkI8U@SyyP-+D0U)(NO-m;D;k8;)d^j*Gv9~b=gYX7r_)$Q?~61v<bqiHM5D<
zsH^qZbg2cQQ(-RIf?jX@umW=gMWq4pE73Gg?DS=V460wEBGsHuP+>IoJ<S8}S00Bj
z*9BydO5g_QDX_5u9PHwx4g(V4Z=r+Pw+6DMf>B%t)OU3J$cNAP>zf;q!30%mE^_f`
zMe;YD4o)9DgdNAoJ~D8qxi9<>)>dbyad|?*t&WlBSMl5Mcqmt(3VRxn%eM)16!iID
zpl7Iz;_ki})x3Q(7We*o_U=mY(HGE^!s*QdLP8X%%d`FUbNTGd?oy5i@C%2%2~nI!
z9iW!aly6p$C(6%#{T<zZ{6MrJs-vuVmkjUkiS?W<z-yttVhaxfs}9!D_FsexLZ;f#
zZd&?P_~8;85hFvoX6rSbu4Z5s`o47PP3n4z{q*q~5|7oe_y>QmgFasUZiomDwG9mP
zavprj$WZo8?oGI_vr9uw${=>8ILMA4q@tU1`e#jF^OowTR9y*7vSTtFIKJo$!XIz*
z;%(VW{j~Zm7>9y(esy*fG6g)jpUD}1+WVrW_<x_?fIV2J*B13O5om#uMdjo?j66{w
z3UR-uHqX~7vtJ(;d!sduMwU;9gC*xNOY@6;lJLpi{dKX1Is6#vczWRVjC9gZ{Lu#f
ztiAULVOzcU;|PZdE*F`osTpr-Nyi7u8(1uL_+sY=nnj6}(<K_dn06pQ$7$i&f>uqc
zg9^AZjE=p6m6iY`RKqJ9lXy)$M2F{n_~|v(3*C=SA`ch0EKF%owE@!;`zAxR>*H=u
zpDMVJ6Z%d{nvaLm*@z0%2egUGM`yCQgR@QI6WiP^_UpZ&TEpDgpFz5yljHL%&L<yk
zZUW2-nh7_k8wykiXOCk5wKlh2McRC8x9rZ}h@G2jEC3wF4w0G6rz3kSd?bpQ5!wg5
zT$aoWARglYaaEtODEbqUTWbAL(fwq&rY8y1*6zXmg`__Ec}2{o99mybIU~av_2vk7
zOc{=HtQm&%KY_`7AF_n_S!%r?En0z}{IQ|v52wbQnYnYT&%7lcFCUdB#3bj|=q=W*
z5^g4Mfa~QOx|@gGR0xGmYpaAi##T5GkEh3Xz2dUpuCydb0XEVC>p3*{P!jityWiv!
zUPv;ojfE~-n#s+Pt8z2S<$a>`XAl~D|M-^rr&>#Vh3sG0)D$ltd#yY4+aC>vo4EYg
zFr0xq7Sa%IV`>`3#UeET(qgJj%;X^5pS9G|0Wd`gb03b*{p6Ro0(gdBeJCuKjCN1I
ztKv8sn<K;kHivz-?{ptEtM^>&nx1I#65$h;M2VphGn6p*BnDP(UmTJqCX6eiADqmq
zCRCUU<2LpgCW}_vRZqyK$rueieCCd5_08xJ=zERbwU51}1m?;jhSYO+t+3Djl;k%B
zR6c{^?a5&@q`Mq~<mAbB<g^_1cZVO62oI}LTlxr9G}S*A_8fEE<*hhw`%kG+^cLmx
z!%!|lln$S2l>IYt>Na}rwxjd4>5;3|V|=@}(!#Praed5eazFnEwVh}%`-PL%_M5=_
zg;f1_n@OHUTB686SEm#&`)_7${x*<w05#SwA>;J-{;}x}+y*YgZeqP!BtqSLG*s}j
z_gR{iUU>(#xA*Rzr*3ckPWA4wuG@$nJ?DzcF%I{aST_z04h``SJVS*M=mZpMyI>7&
z{5iT8&eMzt3p4rTH4F-pQnq8i(jjt}!)T(ZGWH<e<oA&*rf@QUssv8{!K?2f@DSbh
zTE3VS&*B)ZTJnbT7hpH`R;Dz){ap4$%Ur-h8A_*geY!y-27|$Pb^Nv$S_O<sxEw8u
zO&e`dxV4CUmpc|Y{G;!^7DXoHmqZKxm7Cl4C!=TRJwC@*yHxuY=~qjE_E7}e`H3$0
z@E?eQOR7foqNmm4N*2#_%|QgvRo}2<H{a*`H04LIv}J=CN{*{50FR~#+IbpsNlqb<
zV2PL3Yfl(-vk~uaZkFO!FLI6@_@K;QVgr+{Vl^6GD6(ysyFZz2;h2uDJ(GMPnZk(k
z=q#+%VIacx@^dE*R*s^gVs;&2nGfA-EoD9L3pA!ceh<u4x6QsFQhWf2tmAc`5k@)W
z@O0yd&0wj;AOc_65Xt}jKid+H>j6TP1QE=S{p=}=uS;!pF!q(dIBs9QMrM|LdwvD<
zAkFLk&IeStqYZCN&VI-R?00s0FE74d@@Uf0t<cPW976sm-RUm7JxhZ47l}4b50=O;
zulsjnsUD7sMLv0>Da8WJ?|%ce@z$^w4YC*g4Zhg$@%!|$^ZND3dSX}ZsbqzdbPI;|
zd4I^yc$=_>E30L5_c!FBA)lFAIMa3c7>UaDG}Q_x2BEL<y=hWe(ft%gaq21{n!B;!
z9I1^oHly0)k(f`}RThlqUANa{yWl3DM`<)vKT~JTx*he5;to0iFwg0iU^rIQEbo|O
z!+G^q3|-ngoFbRhFj&(2(f&W?Ws)ejt1EAyTv>@jKatwnFPY?#3qD!bLOUYhb?4(O
z?l0ic1bb2zy1e6j)trjqxTrru<P}YI`%UswsRw294Y~yV>U@*l-1q8bM8T-#9c`Wa
z*^kW>(p&TLmda-UT{h-j^wpqj<e74PtC@2C(fOvS$yw$O;PSbd^ec!Q3?ZZinm0Us
z$#@<PLGxF*BF((T`O}12#o|ea3VG9vWz>ZqKVP@@>S+Z<#W9Xb(U076*}4pu{zF%N
z(Dcc}l{%#2d!bh2fqmTuHB62771M;e@0hW;J`V*W#8DTZ9ylutdZvyFC8VFoK=3{M
zxR!eTN<AA(Jp3J6-sLI11*wbLZX7SsVb7v;^gA0I<2x)jLh+^G9bqi&RsS2%kR1ba
ziR7XjeS3z8IATzG?x4x?IZIuDsfFLyiu@Xg?o1b59E{9mv@q*7%R?R~8frco63C!t
zISDy*?+0K>NLI`yyYofmT+tv;w91jW;*>%mxbc;;Z;MOcRZwK@*-Xfx?yiX5JwC)c
z*ccp9d$NP6jzV{<_=ktD!6qhnQX~h6CDrMRl_$##jO+>X4heBqDrq-ES1rJda0$DY
zrye){KGG~Wj-4G2|E}$UzrW%ln=ugfZ=}QkY_Fo$x7N?Cq1kDf$C*=p5Y9;&UCOJ0
zfdMs_apn@>me9iT(qrMP?|NbaW5io_G9fs0qPoypUCm*!-P!ny_6IIw7B?+?Wz9&c
ziARr?ZwBb*&|#7W9DkqQA}Y)Z#O!*38ztefyjRP=f-kn)i+kT!eNJ3SKCA<wLXjo!
zfNOhi!2Keo_4s_UIFmzJx9q3M>;|fh&Q^1}S=M`JoA6n%Xd^`AA6JNy62{u9n!@l$
zY_r*IIhH%%X|~!5dC3v@x+wN<nD@<xS>kT>rp8T@g?nIOxa)-*gq7;U5^`zvEu#SO
zw<`4o71e~E-kMr|1$OV%Z6yf7t|J&|oc=T=0NwK;y1;^iE(YmJ!*<SI)pT<%Vdm#q
zGqi}3`{&**MR$e1vXx$bJv+1SGzE5dSJ4}0Imyk}^Xn4@_k?0k7~Xy-0M=GyE<LvH
zv?-DkZsisO<Z7U?d4Hvb+EKl18OvBME>N@)+t%<Te0mJ&YN28M^_xJB`7q!!j0BBX
zh{TcE3nr6N{#NlL*`!=URjNNa`biNYEh1>oWwvnuc~KxU7wq%#-|>zPQ&|k3uXOPg
zf(G=9g*i!;8YB%fWc9<QtUSe*aSWkC)awo6qZqi89UKomY~8TwZF$X1{*UQDr-9UW
zY<LP8|G*90bSzADcy5t{Xn?5bs3?*bc|_-YwDwwDlCEc~MDuvsfpGiBma2!d!+yBE
z`RF;yJO_PBbq?dl4BIWHM&+9xUGwhM64p!eKg#{2Sx!X5tsfmr2?wK|>tw(9RMt_I
ze!SxEEC98z=;9^fp`JqVxi!)+u?Pq5bVtE7a-WcRmRl<&@k0c`cwhzbkefB7sSb7!
z_f7tL0x7j9Ph!u$D1_6Rk0H-j2+T`LLlQfpvIcP(W#$b%&JDmKslHN3urQ^lc-K8a
zS-Ol_;$bHxOs>IG74N8{DgKQDi*t#>s@Gr$;ot16WT4d^sugZ>Cl86_+aMbITa+BI
z^8VzkSO+lEFt|1y(4q#3+N&5BeYh*1Bkv<B$*8aEhazRrB6k|L6uzw^Lhhjbg3nyc
zO98ihIo|h1aQfly(5czV?qMI9NDq+OIX@&-cR>><(Qg+Y1-cBnxw4yTGV#f-nnxX1
zRvMn{z%T5=@c5ZC@~94bl!7YJFR?loiCyjve?6&BT&RFsbUL{|K^YydS)!=wy_Nuf
znUa!B{+TA>bOl8=^4}WHkFZaD)l{tS+j!JHo^xwmeQUxJBrmkHT?#iE-}BKP_=cOo
zLYp^R?9<$(G9NbfG+pmbJBr5BaDZZu?z}f)kn^e&Ltc-kW04BX?dDL+eCdKVQ7@x?
zR|`{L25-Q_OLABMyB80K=cONA2^bM0=E9I1j&)03G~5a0#?7<BDXO|5>szzaZ(+<D
zruiDCq75R6;QQ$bjmLl1*8X1BNry%_ItF+mW<54SdXdOU7AqT@2;5Rol+^7X$_g)a
zxBvcS|GV9;unmq%;vWikbivQFP3>@Uyz+$n&I2aQ!76NUW(qDeTROx5Ej!oxS+iV1
zG7XzVa%^UnQYN@@V407`Kf!r0jx{&WSmqy1o$KW8Z0~I8a&oX|B$PJ^qRr=eny~|N
zWZFc8&?$S$t(BM#KWa5+W$_UVlhK2toFKM#R#tt-?>HxV!5{mwDCU!@MXL*~R%grr
zgW2+51+~nwpIA#jQtiahzu?k^8~N=+gxE<O`zusY+8^w#$(@wM?Ip>n+0#y}wYhel
zO6464kNEOJApb@h!<mYxDI6y)dB7B3?ERnCQrsR^P@(OoL6Dp(inRuQJ?*Jse`dS&
z>u1!jd%0C#2Iv!3JeWo>a(V+HYa;KN<=ggOzhYd7R~H)UH+2RXf$j)tYzbpbRDG>~
zF(t(c)YZM?Amt_z9An86sIIM%_VU{P6-xFBr8!j(jrQhakm|GWZU98^M>Dm8I6Hta
zRR-<wJ_pu)Ip|5Qyl>FUt}>Bs8IfV`RIJiPh&Rojonp936*004x)tMW2@WA4u9EK+
zXpngUWleyh&Lk|m_M7#o@LHtF8NJZT^w`<%%9@8{EhTnre5<C>5Oy?i_U)BIthMfw
zxgtG_+>GEC5E)Tx+pd9@qHnEy!z7mJZ;*KjD?skQ)AlTnKJEKCKE8|L8dT>;$%%F}
zqv;Wz#}yyE+9Lz5X3eUUdXA=1t?7g1gX?+g=?H}O=tVohs~$l!wjX}&LztGDvKCAI
zjf%Z+uXSJOY_aVNKD&(#mxuqrr+Wp3xVp|K?&9o1rSzc5X>ShZS9iP+)4|8{`0A@`
z=oeomk}K$luO9ggGN=K{<Y&y=DCc;0@Yy4at8Wv37tXcXy+<tFM;tPyew5fP*rICQ
zUt`52p{ah0BzK`M?-EH!poDj}fYwROElOh259vj61Um>E)j6~b0v*l;Qm5u&?&GXA
zi7u?lBEx-LDezk6frSmP&MyxcrM(4w`QG8$slGX_#iutjXJ7g{XLyp~BCK~&o)za(
zbK9-OdQIl>Ev_axlA>{0NUxLYC540e3Xy}M?bIx)EcvA7WRA4j@wjQ4U!>zJA7}se
z9R9uwo%7EFmi`zBU<1qFJL$OEsJOKkc}HyOS>!0eCe8BA&GHLcMH>{~IZGDv>&O>R
z#sqQpo^Ws{B?T>PYk!)}%QLKf%4U?q+Z(*TwYo?Dl1Il>saWo%&Wd685e>6h=)}Ri
zzf7snXIL}izo$Kfo~eVLlG*IW=9vS{=1mwD)_1ujI-xDjG1c`5FEHZr=Cbjw{!=-~
zIp+-yDE#VR`OPKoL=@bAU@qEw>Vd^9a_&M|16#@3mHVL6fdV9K(6hiDl0VQtK_286
zn?|zGSHWClMDh(97AyuR2S+-0E^bGjP9vIfii_FmC1WFRE(n)@h^+$C2LR5MvMiZZ
zXKAEDl{wlZ<U%3cOKCi3$BRV#{OEs5Ja6m1z@wEzb+e&Wtr6yWYaw_<%Z-zB<(zA)
z9s+-KKCtJT`;|Mc#Wt*|U6wuVyR@9L!d2%~B#P=}Z8u7g*abaG9+YYi%f+mmZaorH
zj@+e2>g8!S#f=`Bg*CgLuU-&55Qrh2uCOm+Sv126+m-VkXe+BZw9fnbqd*o)AqyWZ
zghz<{q-^&#PpK~|s7;0ly}Mahn#mjFfU=%!2?&h%7D1SE8PR9+9x{)%r}qj!KDX#@
zYH8`R$lCz03#?{y#B(8th|;A@JB9i$;3Vt!!{rsPR<(i*BpO%Bw2QzgL~P@Sr6MOs
zx&HUu`$DBmRr}$Vth7?o^;&5PhiM_2C^zErBX~X%A@uOKNrJE*qtQ&i!OZi?B2{}0
z?)(@D87V`apIHhkU$J{0w)fkET^L^Qd^pHrBR~a0LrGpj%R*yFvY>on&d-Ce=XD|k
zP8g-QzM$$bX|VTL*f7gL<{6dpfNA9IXYa0EFsI}o(%00eFspAPiHRd9yr9{@G+XIK
z$o9wE17Ww*bn0mhM~*P4u>lNQ-zjt<EQDklY7(x-nyDm$fa_OVAOijUMP`D7ZtNr6
zrD|uFiA5XELVEHZIvySj`GM=bf)l*GwqNBO{DQ+c9zQ-(sz-dB-y!mnhha(4qxZ5;
z;*!w$_9@ewUJMh;r^`1hR-BQe(9j4t0H>P)|36}_^6_cO`gzZ%v`6U0u~e%g7cneb
zX&({l&m<5wSw5J#OBp(a*9sBCKppIZH@DV}diOiMeqHTqEr;T+yVI^(8@6tgQ~S+%
zbxgZpnYi>&cv24JP1@v}=9!t9RE6S6vj^Ok7PcvQe_Pn@)+b~d{`v)|I(ry*{40+m
z^CMlMX{N+k>!^M~O9rPbuEVUXd>za5xu5fPsOn}tY3|6cf~7)?)6Mk!ST$0z(nNK-
z?H^~in_bu=PhbxrY+5fg>4Qh|4mFawLB5<(5+p>pFDX3w{W0pHBslt>ABs%vc7L9}
zn?+8gst^_?>X)GZISEK&(J{JXOu<i0j4LNS6s5-YLku+B)D)#|%c)5(_(Tcn=9#cl
zgm4*=aB*>iMmVM3B$3f;?f6HjL{Wwax?t~*z>hF{I)s%!U(XoG89skmdH62j4zT~N
z*C*jid^$P!+JF*TD#%?aJq)73?sW*K4t!W9(1E|$x(8?Giv!}T(U3AKl`2KY34Fov
z6mHJ!1sfNbn$K!`f!N?NKu32|mlAuX;=lId`MOsd-q&dJ-u+5m9k!$4;~~8fm%$g^
zE+xpw?S(wcx{TwX0(0!q_C%@4bZla#_i~rH**nxfj-=5A&hs@BE7J52y-E1EZwaM0
zyFNwwujhb&0ybM;-IUnrh<(`oAY2}n<|ul8bwF%!EzQd3D_7w%H}O;dTRWD+h)Djz
z*x1+x>i9+b#Q=hi)v=)7HYE5&2B)Z#6I&k3Y#Y3{hY^KdE-ki;+VzG<`ekHWFJer5
z<ymNDWMy!%V;g8g$IoFt{{B9}J*^P~EK{RLnZU(%AXK8alb9NVQHCU8gPmPN#90Ct
z_1K@|G#E7)lcdZV=E7=g6GyaWR)6N1mN^Wq_k1fplBqwu?LIYiUc6{^AkrN~Pzr7D
zY57GcrKQ9zGK;4154NWNkNX&XkJv_>DH0G3Cl|5J2hj%~=WQ)97?vZd;Vywh0ocT|
zS<J#L-s)!G2y<!Gx2tzh_N)8Er?R~(C^&WGDAV0Vym{bE!JFTeI93^3I;zSu>t_qx
zo(t3aka}~id=m!TKwm_6il1>=b*j_@GP|beK&8Ldz2}_Q0>t^^W*Io*ymYxSMTFf=
zvb&jG#t<(g)T+2nXjX`!CLvjYs>gBxN9H%j=Dnw;JmR9ksLPk1<t*G^fuqL~jc6lR
zTY<3a6)#I~C(Fe5k*h<@HxoDKeP_fph6|sdhG`EaBE2{n>L~rE?;y4vDj2#tgbNCy
zEA2k}D>bEL?k!u}bO&YI5iF!vb~QLS{>Sp6d)xh43opb$KIsNR+Kg0=rg;yD>-QZ_
zKcQz%`#OA{pA+*DQ8Paw+W_NGWX&8|vJEmriFw-<O>P_ox_d@Q89E59Z7T%b`-hW*
zL*{-uoma%qP0{oNR-cWe4NciDJ?NN8#h6R)E={{v#Aw!&t|jz71BE)bs9m@G{Y>1|
z0rBvUR4T1TXs-Wq?X$BOjE4Y<iag9_-AFhbKJOt~dAVO~XKS%TNkB9|f*V$q#m1`f
zg42{kd-TvtJ!hAjd&WiRdOP44y{)iq_9&e=%|}+xj-@hy?{OEPWdb<~eyGe7x}ZQA
zBg0I03+xzlsrmYBf@lU2JFt8>BR56kN&0t(^8rWcxL_4nUO!?NSZ~kIHya<<&Lgb}
zVXbMNr<v*t43`Pr4{o3JGOAi1(wn;UFWa=Mg_V>lpFg99yf@R;RaHsgj%)h%m?9B0
zx+UE5!nRCgLO5RswmVs8_BMsRFQoBkI#!Jv|6+)|pME-o@l{ccUvw0?+93%G1tU?V
zuH6JeKb^s&Tjs8SvCbg<suvpQE=aOwd9tsGoPeBNT7~a0^9vcUHP3xVQu@`HT#8{i
zK8TCH(2(50KW^^yaqS!|TD*jwXCf0li`(LST2j~+KI$V1(i!F>X`bDIFi{Oq6DXnf
zSDc{f)tLvNSd#1IQh9J*JB8-CNmag?P(EK!c3XO?z{&yiox0&s*d=DK-rZSFSm;#C
z$BUC4;!Ee%h)_!_OH%46h+OdD&Pm^OZA17f*0Q10Him{aTn2vB_4Kwr-(K_7(zn#k
z6Y_oq_>K=mNj}(becw!KqEK_gvvA%)y!oCQx>NG!IM|r+<8OLnYIWGH;b7`#exe;*
zOmVkBD?kqwt|O>q6zZ(&ztw74^n`74eZ4+c@>INJ1N5x;1)it{lj0Zi^BfO-nwPDD
zB4NSS{ovz%>FSY=7{5)=WdECDHCtPFr~>2^2WBd`ZN#J~Nt>MoqalouNDMVsY{2Fg
zm?~+9%b%wk>c`QTDN^U^>r}dpMvr8y_3ZB119fpL{aF$N@r<POxzAtRwlP&Nk<uI`
zrEo}nd>?i|6q)gNlB?WyN`K9m->s)$g@=ycajDD_iCa;#k0MbNY09Lr-81s>?m3|u
zU!Qt4Ush;X{SZ6(R#DyIWasH0golZSY2HEE=+n%kN9v$mZCk&ATRjKOia-nS-(8bU
z9>7&bMy5+3cB~ZGOr1142?9=n1e)JBD!nQgo{ry6H|!4CU7JQ7Q`|YmY}2`=Tf#;S
zYvG!%?4Qlbuh)Ejd7z8Z7s8pC*9OYf^M_R>1nBp}cYmkKVj^!M^JiP$ZJV9^i{e9l
z&i6w_^-$fG^;WuBZ|}X_5>WPyRFapM1Kt<vhB1>c7?=Il7B*{Ob0^ZhCi%a@4t;Rn
zht`RimRB|}8C|ZFH1{wkfOzSY^2PjA<n92~%>1{`|7>>LTx4ZsXIh9XGU$VHVV&13
z*7EAlnuoLN<~_0A7n3wLo0&>o%}qV9Pl7M!F@sB%2pRU08${{FCG*YUl+wW_Alsif
z_V$MtQ6JGP=Cy4dei+;!QDlP7HKh;-B$eLE7L|{+boVH}*U!kz3`5$A%wBb$*hU=P
z8{BJ|eJjZ7S!d%aW93ra49OlzywG7FaB%ePZ@Di#Zv@I%keA~Ew6wH#p=U=_nI&I!
z*b0B-T@ml93)yZpLY}N{^NzIv?P8G^-OE>N0?M;nB1YBFOm}z4U%Kmth6rzDds|ma
z53hyr7vT-3rONc9#u7R1@Tl>ZrtxKd#xK|&l-mBZWX+^P&v7WSd$n{O89GXf@?m}R
zgZY!gG%-<FRY4U$K1rrJml&!<rdy=k1q)fX1Wmu${MW^xJ^7|vSPu&WXj1Ll3RGEK
zvL#TRx0E2r51=r@`z&zJLp^;;m(C<uy)^x^PjwG(XTMNS-F)Y&NHI)@LdAn$y3n?`
ze+o8jl4d&CZ1Kx@S2y>PYoCO4WSq+3dG7jI@b8ZOp3u#g1x?@TWIDx*k|%D6(!*14
z;D+-uT&we-`DVTEBKFNgU}n^%DtPbzXgce+Ci_5*Qvw1~A}x)8A|=vD2+}FtNQ~~T
zhmeqN5C$lzq{QfvGJ3p-45`uOgwdlpJJ-3+Kd@i+#P_-5bI%v4V35cqvtU%({hSr3
z>u-`dx*I^sffU0nupi7Xbe;SddDqSAhA3NJqTk%~(c8tWQFGW+D*X_ZW!=`>_~5A`
zAte)3Ta}7$$Cj+(iWj0d{2wPCk81`l!@6)yg}<eVb?#~mwZ8hEqSxks4`|(LOi?Y2
zkXm5bCE$ThwSFi%e5-cGjLzZp*&72x+agZNH(tX^T#u)?-}W1Beo+z(S|7CErghV$
z5u@}nkU8%;U1iuv=8K}i79!X_uafD8tS8{ntLF@vE;}{scmJ1lX`e{u?zJ$%U}urY
zJ8A%SJaYsE*O4~WF6mc!Bh$=hoFoc0&$@wOCVoX70k8GUoec}&RWb=mx@cm2PS>JF
zLVg^aJSsfj)cl9H{i2giHO7KP<gk-HJR<H$0(p6yII>-5SV96l$K~o<6Cnv;dn~aO
zoHN_~EpghZ{o3iix#YHE<v#v?@nmXAwJd@2A^fR81V5gxkaB&{1P*p{q-tiiYeUBf
z8UPAxOBOHBF5BbReo^<DBi8}x{4)G<Qe4ooz5RM~Z7Yc`KV6Sn&UfRTQBq#sLbGWq
z?)|~a?LKg-a~}-+0Y&#7!BC;sHEl?sH>)g32o$^lL*Ct8A>hq|J%=ft5sxWrAzN-%
z6TS(u?Akn88pZ-V5VwnImAU4^l`D8#*YZJ1;S~4v_2Am!aZZig%^Eg=>5IvW*jl+o
zi1)NYAYiWfE_9Fi=GO7?F%pS1?Y$rGy<^IfN>&DRR&GvaPNs8`^19EY=>#cwHYDoQ
zo)GDb7gdBH+S~n){%ziJRkJ=V2DZknr6nYMz9v4tjK9BOy*~_H*>!g#s??1gH?BY7
z?eV|=Bp#<R>n8BpTm#*GJ#8b@;dgmew}?TRx(d<%+lXje#>T?1{2d_26K35=LS%h&
zI1Q}Uv}LRI#hZx0mbdn2U8FQcxE~MY|6$eCd#7v)qcPhL*kb#5wdvN>(<oJpLcM_Q
z+28xg@K8vJr6e$MfQl3~CErKCh^wq_f6Nj@sU8Yaq|ZG8H``YDvhmp;BC!|Q0}dD!
zbNlMZ%RYwwE1TL!{k{V9HpIA!I2`?d)&a$35A0r<9#@6Z=}*U#B}tuCJ+FOcd@^QP
zj<+5JB_^coe)NKMbq=4uCmZ|<F*&tzKOFgJPfxEb@`iaTXd+g8ad~!s@(P0t#~j@K
z>*u>eqHkj$lV^@GCz}O1p)-3U!a+VknUQsgmwET6_rV9B^H^4-StgIYg(s8dhXII{
z98{6`z2(qkoaS$P8G+LYiMMYge6Z2y>_+wK{09cSd<~8>rOaAT2sr;0LXrr_f@H^?
zI5nZW4R-Oz!8Dh9R3?$bu1+5IQ1XKf8C`%p2_ytMyNCTo1N{>c1p5-Kj?gOW83h||
zAs^VeTfrqci1+LS@z(Gf;atQ7jM^52v^l$E_$rb5Bygoe-fsD{WMMeNTTEV7PO1;(
z?>u%6wCuI%a4^p5{m{yN7F`+#@Pc3}+WFmj4wJYx1;zFdI79sBNvV_IOV$#K4&6()
z2%mPM%AgI1F|yacrmJ)MUp@6;o>j=*yel1J(KeLl`pvhQwTqJ5>HFZYh_QAVDdu99
zqFKh2+l@hqx+fL&LyF$4YCjM^Q|g`k3Tw=FVz3^Bi`EDJG0~=!f&^|g;1d)%wlB>J
zJs+J*5e`Hno!b_)3$a8zT={yJ-!rf=Gcec8kto&rysPm}^QcRM_;?MM>yY;tL*XPy
zFk)T#X*Ig{8ft7EcDK1Jl=R`#`Xx%dD(j(M(AMLUpnE9Bs?+<i#UBgP#a(xQGmW#5
z!qLsYTd1XD$0a4*Wn?a3aOP-N(bes1d!0D@k$ftZNNpjpIY93srHB+ugB=0xj3YYg
zO`3DvY6DnhgPS@bkl&Woep~otRl%VS?iRreg14h3;kO4doC1Yv2p`(Sy?-UQS#mc+
zz)qfcpupF3jY7IelS3h9Z87n-GG3OHs{IP2y;M^3zk}Or(=y<hi9z=4-@jMg{O9N7
ze^a)_doJs&FCi<JCarP(m;K8WJz|3D#r7R^@9`N8<r);3hbPZ@v15c3gof#AzCwI2
zCaH5wKwwbkI-BcGmY#DWQho420ctE{hu;L1lOXZf!b?e>HG7q;!&g4Jd!5HumLp&P
zTe*!sNu*I3ArcT}Jev+PGO4OlQNp1Hhtkzj<O<;kP%}jyidK%Bm@HMuns~;~NZNJh
zj?~Ro@ayrXixKi-iwONv)`oHT=F9~BjB1;2aiS77!E4O<tJ?BxgpFiMdf?P?b-we_
z9S@5xpS?}{8)%!SW%ydo8L4saC3LY&5}XWE8RVoSXBmC|UHzLukzk<tB3-M_Raw^b
zf9k)%Z*F5({t5tY&A&1@ahY!5=o_-5k<rnIY^-GOhO0pO%2}i3v(CMDds&oOuYefL
zilwzVQO@szNx|-`*m1hnh{GZv95adxa=Zd%O*{ROld}hlAzCZGg=0cjQ1t;2U+rsA
ztxvVtA|_SqK}MVLmony7*1aL$ksb&d{P(qiUBd6LyKY^L+u~x5Dw5`~%RsEPXgA-_
zG7(LzGP6uGqd!(gGMy#U=#3Ry0rPp}!NCHI5u}^l%u(V|`x4l%@tpm2b%ql60D^Yh
zL@lg)FUm0F?_$>1hsJN4c>)h85)-Q+5RaX)$(l2wHgX;9DOa`q^Ezoz)eh{>nmxGM
zsjl&Nt@yc)wn?pG<OeB`BBGff`U$bp9rAXNcXX^&{lm@gD(l6Sm9F1F32ku4FPGkX
z<nm;aberc2I$*&SI>}!b+o!5@e^GL?wwBd=so8MaTG<@@ri8~EtYc!B4~-c-ayy?m
zW3s+oFMI)=?fyMHJgh&GbD&-zh73Wyw71~x=;-L|>^Xuog!za2Z=BkDI)C4=4F8^W
zchwLs!=iwrn~c%#IsXiRJmiWQr5o9W>eCxfyfbF!5BmGw{!BM~VDP|bm4&XXa86KW
zA@sL9f0w)_Z@iCVU4m(*W0{`tOaBy3_SSMySdvMpE+U(f-^-)+MM@jXGlU6TnK8#O
zM}4no=1sD;w!cHvf7kN733wT8aNHd@QQVRVMRF{qP2(dy9GDNF^NGUHiptK+xMosY
z^Nn_?c9}3(6P7h^Vg3}4%11u3Poiz%qls3xiY|o90|s^*$`&Dh>z>?*SB^nMK4xZW
z4BNR)d(-WUo)K?b!9MI@W52|EZjq$Mfh_+pHIW=;rM$*c%fnqf<24ZNX5qp~o({=*
zCP|dxK#?8%?g{Y2kX41fNr@}*Rb7fwwHvo#a<sE2<n?Y4r<T#;WbQY*m?lP@<?c;^
z&68hDJ27t~BZ-sQsW(=kzDNQe{W0Zok&n8D+z*oNxjGOF-sEs8z4y+!Ly%hHtJ7`@
zCB?DQUo(oKI`CMTK)507Z}l;Fqosx5ffSjPygJQ;CPLsXj-ArsNN~`Y8Xhb^nMTMS
zt2N<{x;`QXtAKqKPx}}WaF&AjlOZta9Z<8WbCL28fufl&WC4u*zn|-><K@%ff~S*-
z>epMW(g?naJH3qWliX2%=1fBK+%CwA^f#0^Ih{aADQ@H~&Q3IfQIeBQK&gV1>V!LT
z)+4aF4Ul4{s9q(&e;Xxiq030y&;U$-u~y2zQ?Fx!I~4B(7n8Ev(-6fgD(n5Vd)D&k
zx&Fcf4w6+#0(g5M;sK33_Sh5nI8NCVue6O3%)RqeHC0|oEpwI;Vpg`75Ps(B8IEPv
zcZF;n=k@qttZ$E4Wy6~8|E1(nP+Xsx?QibV0H1k=Gu!z#T7DZ=)bNi@bi3I%8S0E`
z6BzrtQdG^<8uk8YzJ#J9Fqg=&abqz=!T1)NwR-M1o4^C){bm*N+xYq2#zyTlnN@(4
zhN5Fl4c)V{h??cwtGSh{Q<Nuf#X?a))nbm?@g(*)E-vh)7@y2EQ3<Yu1Y!_HWzbuo
zImO-TkJokCB5-(pAAR|sOKXdOc&nJ}e2K+^*=SX8!EacZ9(7esjcb}taczTzE}o%H
z*_})V#H{(14C`-)tl_cj&RM6oKNP)K4^M6WxIxHEz+oD_ucoRno@mo5LCfwx&H_{p
z#)2zL;TNmBqw#}Sw{%1HNkzQMZ2>j#D0XSPENc+%Q#IlXdhzV=@GxEY2#MXyTIxO1
zSh<7Etz54nhLZ9aHsBpHQ`Xl%Utsl2=fH)!fETG7STW4nqMY<#sq1=P-0BCM;Tmg@
zC^x#b{9=vpx(^#`c}Nz_X6}NcuM@z-Q7!38?*r$?j{4!;AO|0p^47gpDhFUg*x-so
z_K>=Zaf^|Q3K1vKe{4oX+U(g2ku(_z?YfnseQcD?_C-AV(6c95B7(O;B5Ro+7+ld7
zHaHhFUV}LFq#vVjC=>qs*!MYdwR~0nn>+D{HsKCL>0hlt@A0oB*JW(H`)P5BOz;ga
z)i=(s_!Gy9i6f@N1nT|$jo*P*J=sF(=4>w_u54y=@!tWvBPL0m`s&m^@Cz)I?O#^R
zuqhAj8)=!s@YBPu0*PCWzYg4Puqmkd28T=kl{SRu!pY|Y$-r-&uLF#y@C#p(ud})2
zIt-<9z5V!vNa!Dz{OXR()=LUpJc5J_I{c%UmkO<lOk!>$M_<^fvGargpV;t0eRcic
z0?%0}W+D`MEnX+XoG6c{^!sfLibCSgAEw^e6;sspZA8yiUHs2M36kULIfRSr9BLnH
zJ7E*#nUwxzFYFjL!1oE8Ib>8%#gzA;4@d=9f=n^X%S4!w&W|-XL-Nn?a~fe6)&CN^
z(O1=bg*`kwku94f^j`oQ%?FwUus|3)oh{9?ccCDtFzhkzm+Va6gV8UT#JAbcVy4q;
zyi5fWdu;hfN`>Z@FRLvY#wuOFvm+Vac%In<mtpgSf3V}3aia8Z!BQ*Hm=Okw?0r!Y
z+mu|e_je^`416P%+$q#%<g@+Hz|1gtbYJic`{y39a8P2U^8(N#ZbX3Vg$M;zcOqCm
z(?9$E)+hP}>`9)J+LxqEUJW}XPYCx$R69+-eUtI#VvjiWBhsAaT;I)+ekOSP8EwGP
zWpSIkhLRq$+E+Ies+zTOZzzd~LYt-Fo%Akl?AxdbUquy`1S)FWAdaQ)0{0|kOtvBJ
zDS7&@QDB@XTTh^cC*4*TwP1pl#idH;<U%Kq9zZ$m7b`oxZU?$nF8<o*i}kQLhE9ll
zPU(VAkf*im)ef%(HT|LKy+{J~d4MZWFE(TEDV8)OwNl$wxW&kyqdc1NTI+Ugzl$tp
zZ7pJ^mB53i83OaGqaYZM+o<q<G4q{Y#&g`zR0d&r0-Ubx8}uBnpV1ZfHvK&|NY;5H
z<Dgh9_Zf5EvIpt;x9Co%syV~98q>CNGvQ*YtS7^0-45K#G!%{=ULfO9z_CsQC|}#7
z2V+V2HFgzc-P?7Qwz8@4U*LX4%xxPoNRm8zy0iLkdi0EnZ)*3rJKNi}&2@X*zY+dY
zDu0J<fr84tutp2Fxlf4;?vllmBd1y(8@yg{T;t}wh?O&|>qT)nda`)19nG+g4(+(|
z(e*usb!#!0Q3a1pLnN-$!ua=MJzrJxp39vZc%;>2k`8n->k=Sz0T1jw=_t97G5*^I
zpv0Tq4o7vuGrL*Io3SlHG?r6&66U|r@j?-3!eAdZmpdhk{B&}6-*r}ii&ZD71y+i&
z>ItoDJxE+7dgXGJ{iDh+liqm=_xQ4M5A!F_E9~c5%tHIGTOIU;&=0A=N9u|EMhnuP
z{6wE3+yi@F)?3i4ETQ?dId)~)IlO<2wD7Ruyj3Lj`t7_N>}o&KPF-H6YSH628XO9s
z;8~?L`SywJuZk%@c@sGmp;OLqc55l6tCAlRvu`AhS}D%{&OF(3DlfXm3ik2PXYA!%
zCs!JC%8Mb3F!kJ*oI4V91aNbn4zE>sjhMX9;;|2P1@k9wIJ3z6jFX?io*(PT<NIfm
zHhh8u&MxxT0(zWVT+`Ah;^$WfRd#?W>#iWw-CnhGnS_C}l%}1_*vT-deeE6z?MKaO
z8S4R+R{cwMWrD?hMAoCXL-Z}~*2c<n0rhiPrUB_%JbRI^J8KFjL#7QJ-nUtz>M-&w
zP|ySk4cHgfE@tFPH1ae`7-5Ez#d7xSXmK+ncEGsQcx-q*P7V_PXBPqr2A<O9SQg!D
z<=I^So*U%ymP;2eqi3v&06yZkf5FGZ&-@>fak-hPBqX>4*ogL^G4dkYl2Ucj>0mBU
zl=^zD>s5~{!A=wj*a034g4(EmS5;OkgNPNK)VunbwmHn4Aze3^y*5Vh^F*=}2Y7vY
zbok9aMl$?MUbS%H%mGH?VubW@>e`UbuF0cKZg~#l9f5GpOQaLuEZlQ_)Exa#&vmc^
zU*4%x!qDov^ILotG>Hz|7{!QFuy}hifC!MG$0LDV+%#l3ls0tcX@_r98rdZBaKcRu
zp~CO3MFhK$CTMfl&y+|f9VG+-UcWRu0nYxwx3NlS(AGjc(AE=pf~XU``K{B&rF#Nf
zWD@t_7dMChw2=sFw-D)P7^pKxvp#)#H(wIIH9EQ%OFC>^Z`li=`EV-szVNm_PyVQy
zDP+3!TBHen8ZKQ#wp}wrS8vVC%rc2%3R3rc%cAUDS_spqe@*>h<@$WTel$&x<q^-;
z8EOg|d<&TYhK|>=vMyt`Vh*O=#EOz<JpQ5(#WLm;D&=tK(Z6$|DTxw{l%4435{P9D
zi%dIa_ox|e^5k2L+WF0_aXs@skt!J<bK01Rk=7IChoAw|sYtcczMDI)=~g{(M*XqB
zf_9r$!!HH?&+f)CTuMU8OJaiSel(+cR{SxlnuB`4ioZy*{|dE}#`bZ}i%Fz?mznq_
zt)D|OX-vE4zL@ntEzDmR^WSHqmznK1KdV+Q{w#X-T%H#OHzMU^ZU%HB1g6F>efU1o
zrwXRTje4cm&S9-AWXo(2J%B1|z?!CKKC%qStg=CV^>F(ye%BexyC{UL<)wCsR2A4a
z6T5~b@Nf`8e~1j7V$jJduVaGI3o)q<VMA@HMANO$o6@=QKxIMR)q-i%{eF;&{Yuif
zYW%8!FRxN%p7-02Xqn?aXR!xAX!XcA^_^?jPk}kSFOT{6x3TwRmnCCrK<J^;>>iw&
zDua{#V{2|Y7^CmQ?e)cgwr0<(eL|c)Bo*M5e<Qt2v+&co^QG(IIYsZ$-3w##7~bct
zu1|h+LK1ixFmTS&Gkan)4FaoDCZ4aE-P5amj%w==<(8S4UEqQUxL>)2_rCL(Lk&5J
zNJuc!?;ub&^I6@q;nZmiO&QFml_%PBDWA)!ggZ<<p-<?W)`U8PF2JjDr7|JtWkAdt
zJ@GABA!DcT!#;FQ%h=O5r7@vN{#!M=tF<n%^f^<R9`O*4#76Q%C4qLb=r8E&r`&^~
z4G^SWX#kD9-Afglb|zy^_LH9l2pU6Uqua2YlBu?x@$MM`Rg9oj@EcVT-5=tg>)|>q
zOIO-RTKz<MuP5pMiYSYS98LQo2>`p|n$&8f*;hQ4cM1jC&$7;39Fji`2$5yaR-;~f
zv(?MIhbs3E!cRNfpa&skFC*ED1H(<l4PC1js0hoTvV8-7{%rGpGTkBGfDCgSI~{Yp
zi2q1}_g5SS(<)qfS7?$#0K#k|n^MUsd6|l^2vWEAsqZg(#kGQ@v-OjNI1oP9HaR`L
z_CjzuC?uj7liJ;!kV`M*wTdetrLBa#w(QxapyUSB9@=THpgIr0C;oGcpqIfaHbp21
z1hL;u_>pWh>lKboB=I-5WgC&;yP-TI5Voq)-4^J_ookwnrTl~HQZn;3o>YyVyY<Y|
zj_-?U5#4VI-v7PpkF~xB&aCgV1c9w{B8q1I!iKA9`*r4L)Bl}It^WGMk`RE%S##ng
z+zVt`>g`x!j`+7fK+PF9qUlLumvV+)4nFK)4R!Uty10YJ`%PhY2u}VU`Fcy8zLbEw
zzwm(GJ7HHExS8<YFzy%0yg<3Y(@E7eU2{RGOPA2&UG@w)iTrTj{BZ*YW?%Nz=hlx>
zqZU6Ze1<yKnav-hwQ`!Pm{UuOTI02waG7}?YfH(;4Qc(tl}E$(M~6t+JXORwB||?u
z*1s~qh<%|F-BzfTTN$-K!{2OQ=Ulg`uN*q_St$;Va=+Ug9d$HU5)M0KLNl`*yc&x5
zm*si$tKkk3l&%RlW*r6jVGb;HZa0W9i>-@tQi-L17PjBZ-658{SSNO-Dzns9ev7C@
zZ1Vu6t`l=T=O8*kE6rbrVRvq=z4uh-k5D5LUlpUUHR-*l)QR=zB2Ig4um1M$HMHrk
z2uTPs_4tYVw2AudMfK~=z{TB1u%sIpW#3D2t?WjvSq&hgwr<L@O8W8RW<61|c9IQP
zpXs!Fk+FW9+VSe<jq3_@KS5&sEa{rsfy*|Nmw~Y2eB6K8RK;G(Rnjh(8INz2>#YE9
zCUr8Lpqdf_Lhd2WHfzx?qo2yT28gD%1+AfMRLr~tPDSU%eDD;1D6U4XM(f|s%C<Mw
zDi#bsc7d9h=~dv0=ze*w=lGhnwpccQMYdf+mRYOVW81;}(bnoM1&ftLr&7~VoY9GY
zUoa@a`_(*C$Rc^%OxBD)QWvx=uP`Hqgxobm=U%8LCJ5`QyDU>lxMMpUro+siJ7fO*
zUW|^+Ix#}ER(DlG;v^5s4J_2q3_E!8|GX;UfFv51s;fSxmyMg9BbqK6kb9PF%WJ36
z8tB-8WdPPRH&0eOsVXaQb&z1eEV~!K=GSEe+ZN28!2FFHG{$ip=n10GwRVWV13ztX
z6hn3rbiH!8J{vTWaNw+h<841m)5UV*wd^veWJc4!wO9Y%>i(&l=tm@^HQ4pIAfNFa
zFmrf#@%MDqtIO1r4&>f=iFGH@OP7a_OD-<%+$dEPD;kxpFtWn(emGX@4{@Y#+1WT1
zu?-O=xa>}XQy&A~aipds9{R`38~shc6@QC>emAkuH5lZqG6;PT<3DZog%-@i0efmf
zenZ4J*0r?FdSx3c?tz6$Syq*=F9`u(+jR+({o&FoUA4W-S2*-R$E&g-O$ITy6!%AR
zmU7^pi(li5%_{bxY5f5Z3#09@qO@ll(9v@7fS&wU?B`iZ)vFYDs}!i#>y(u%5d3OR
z?!0j_z_(S3O6E_=(>huQ<J@|bJy&PzBg6dEb=SEnsH@f0Z=T-klDjX}<}?76Qv)iM
z`886S{d(wcTF(_k@aAH%dEP01?DFOmvAn2k*8-|(nj?Qnm6sMH&oWtA7&cB&pQ|w>
z+4e@Se~7esPM|`NwL6kHPr&jQyaaBnFp}7fVU%7ZujJ(U%InKOUQ{NL{nNB(KEU5{
zdz?wUo=t%<zSARp+)c<HE&-_6^F5lW9urS;pdnn`647dc|K|vKw#pS9BxLn6uX&V4
zmKTO<{hLn8P&el~7pQW54Vy1V@6^Q7m^zP5%uj_KWC16}_eom;#mTRTZpBJ^FaD<N
zJBFSu49VS&l|T>2PH<{#5n2%XM;rBO&HNF^?Qi}R8WsA;6cRr$>;l5~w(|56RWRsy
zwNpKRLcWT`4``>35l8@1*v?$(6fVN|i62LZ%chV1F%H!+l)wGW;=Q!>jc0vN#8w)x
z19Y84Ed)HWYeoA}^C128m4d4#1>Wi}Y7)>UKW58-*PN@(1d<UdNUhMII@|h6h0Z%e
zZ~o24&WkEarw-WX1Fcm%(H-($N`C)IZ!k2;t14(dBu5H=vE_P73e~Ua%!=t35mbDV
zTXzQb_QR7Afp@by#0)1z%S6c&Y|BvL;<I{_MX5Q!)Fj67`fFL@W4zK~M|J2sq<4)R
zW)9e!0q%S6e#za)rJlcH{i7x&72LACQj($)Q}Oy_IBmcUkkc4^CYW8m?{aqIJI7i)
zA-gdx|9}yE&Ul3~P77=}Qh5*NjBJDZz$J#Qx_VF}^iPhwGr2{6Xn|VJi@^}M$+F(g
zwNSzcksb|S1-%{ovL4lq4Xn*soygE(Cvh*eCk2AB<uyDBQqF#fm&Q9<(AGN4$bjg~
z&lVzz+ZZ6UjYY2A)vrN|lR)KIB;l!^>6}6z-ovQOU!#4#LovFE)%$(X)Xz7gGI5Ev
z!CCrP&hL(D@k5>DPGq{Sg*U}!ZKJm9^7*?qcjI#dzn)r1g`W(V%<<LmrOZkQ#b{8b
zLx#U)heh}fv{1ab*YEOIk0qQA1tCf-wb?mI+yL?uS`Y~%?WaimMclXfL@)*rUacn;
zW#E!LUP~O6>CT5ZOQ!9i08i1NjoG}~vOv#eK{L+JKSpj+f1cc*_uusiU*}~DAsFo#
z20LA@>QB0ejs}9KBzAV|e351!z$2V0T=I%j-L3VnZu_T`ua@RM*$6Q_LWduIRNknr
z9sqdV77-F%;)P=VH3qu0TVajsip$sVK+mNhQ{{>D9g+5+qaW`6dO4a*d^!HPzEmn1
zJSOfn-78_2+r12L?lWy=>@6hw;PiD*H?-R+u|2c2KGf9BA6G+HN21Wd`L%e8lI74a
zqn61^M9~pUf$egTh~k&RKh2VN{YSCmc>sAT!E~(i1RKR{{;Ni_;*7yQ(QtG4d8J{;
ztmTZ-bn7<<nF7!1ew%OC{<Wp5&*{rSRS$4BqURjSbVZ?J&(I^U!-4j|{~!ydAh6be
zQepysMtuFjIeF@rKy04v4^v9-<$hs9&rP)cw~y7witpgT)`82FBNdr@Y*4STQ=q5|
zLm}|*w(kC~C#yBM2iUotTM0=a;z%A?L}QD^2Smsj9=Y2+@s*Gw@6OA+(Ch1&Gaq5b
z)MA;?Ch;P&_!#$Mal@?SK_)?K&%ci5{4)xO0;#d+v}!?myD4pBWGRG9o_ab}EDFcn
zlTz*BC<8m5BzpAOpcnnvRK)r)V14mZCxQQ}!}{WhUya-y43nuKHDA(wfA`o1ukQ&q
z%R5o1O2Vn2Au4jvZGGzRY)KifugZPvCe;<PQQ-+(FBWT5wg(Dm#6bCz%D$)$bnaVX
zJD!=~Nc_xo-VS*k_3u`NTOkz^7?uH{+QMg_O1t;kPo5LY3gfXYswjQ|FmUN-L0mbr
z<P3N=)Ct)$Y|+LZ782bdN1nZBm%v5gebD$}$DkYb&d6Z42-~O-H#i+wxn;f21MbuA
zc8mpk@#?P*F6(YJjC)QdF_km7<9TU1c@@pk=?@gDrI#VCZ7$|>Uo&?#d2O&-!@ncR
zn~}Iv;>}RdBMv+ta$aDXuV1;~V+`)Y=>SIQzoGw?6KEtKL@KQ?qC*5sn1_*}us`HG
z>(IPRT=}`V1to80d?|Nn9V>Mz-pc&EOy%VN7+lR%@$@yl_nyuaoteD?h0KKLB(rC!
zI^i+|3u0llp)VE*-^cPwD1S=-M1+=M`y{VyE3qy@HmLbiB)JvnbnU=d9SEKJ#4*%P
zO<<(@J#@MCbufx-uE2fHa7U7|ZMVA9EqU~0EaWNSbMgwz0{$b+dA5yKU6P@>Vl$sB
z$ds4Mj)T06E#YsqIO7N@jA3IHC(>J~=l!?|N-_^&W?wt^6w7cP6(*;7#Z^`KVK8<i
z^YHK1;G|>lv6~fOT1gc}fQ#>$A#AfNBLpM3J^U^0-DBzO3C47yr(@;_nLpR(S)7Ka
zS-o;4z+u3Hk4U>h4TK}NyXB-<1de&0D{#kdwj(DX&Y#t}K-5EAOe^z9dvflv{Yp*3
zK`fZ^WVx{I4hdYZ5ZRvo+nd3L7L5HZX>RvsK>DhFFd!XYX*uAXSd7gDvWRD|U(}N~
z;eD0R`J1}GKj{TJocGxnrD|Tj!|a$}9n_)nC^9KAgrh@)R9Q>EKAjut<+0&y;Dk$)
zWoZnnG+6Z}!-s@l)CgD$5Ph#MsNQ#7?e?XoXnxbs=Z_nbI<Yy~ziBhG=MN2+07Cp_
z$cTvR!NAlMVe{>;3tN$i!klMvBXKal=urv#@kyy3DL1hexxy#GNEG$}#L*pedo>tO
z!P7iA&l`b8;<J!n5%G8;+eNbfGYS@CBy<_eufDbt$+b7l<PyR%>`;(_*LxLee(b?!
z{Lpw~tMFT7^ANUU*zecS=6AnV20bdT9)s=yqMAG>-Q8FuBFGQ|+%5>IAB20w56wFk
zN~4<CY7?JX8>_CVAQ;kP6&C&=rLj_s<jzAJ=t<X;43ctge1uwm`r!1M4dcdHo<O1~
z;i;K%48K?L0^Bpncf$>Pk@1`JBmAdJDVY0P{R=CDTi^_K(6i$FroLn~eA8eI2(Z4j
zTV*C@7@a=q*-!E}4X9ir27C8YfggX=qh^<KO>QjGjXHr#MN7P>&TNsDVysL)>qIkr
zOLw00hX*2hLvHDDj+?k5<-fPO;^cGU?PTdb;+kxV^nYfaYkj>B?8$qKRaUC~@>Ak1
zzx&&Ji@{Qrn{I3C#DUWsbogB}z#6(gyWu}S-!s<qVtmq3BC(|SWvpsSplK)O%oCYR
zm{slm=4fy3<m7so4pu1`KS;9{&$5|Tn@|0mm~z@|(FMPFNKZ;@SUk1i!>kdCSsInH
zO;)~+J?T!vnOc@4cfcoD$W2k1Wx$O`^^d}7bp33yu)UEgF^w^pM)xD2DKt@z?N=jo
ztuWeMx)1F_+#hj(K2-57J>nH2=HNPnwh=;}q3MOFc&P}QJjZx-`F{$L+4CAwYCn6s
z--4lhAe{KrjzmW9C&xyu@GS*3K`s9Nj~Tdx&j`KUuTYU99zpq6<T!9M8znM^JdUU3
z3Vk*8v*BG*B^Y;XzheWCU?C&YA1md(lY5-A48y|>ycY_<+!m#bg}&%uJAgF0FMHW%
zk}}Me9U))DD9r6)-=i`-M12ob{#&WiUc>O%0#62i3*OiD>d_b)&MN(Z6ueUVh?LxL
zTL~~?Ixc4aQ6!biUV559Y?O$d>x>b%4$nxQ=NCIUYdtu%5iJI$@+_3q#8^@>f1j_o
zwK<FJLJlC3va((0OK0`2eT!I2_)Kf7GoZ_gBbabiONW?Of{3JIQ{m7}b|qjxrwwpJ
z8d)S}=$;g^{=};0f1pilWXa;4o(xw_Bh7xgul?ykEJ2=bJ(H4(@3-^FL7qK6`*p_F
z%nwQ%61h^7ssWGCWfx1sU>X(FnPZ`6SADO~*5Ap(wi!{rV~%uO(E7Yj2Y)qa+;Ao!
zbnt7S+4Dxu!L#t;={ARlk!F~L|6IEkwm~)Sc46;xY9^)B_3dx4J3U!@zfD<(jzYT5
zEZ)ePm%y>%kBfD;uy%3Yr1da};qyNpNln#@^`2dEk$`-nM2Y}2E8d^25J~2=bI0aT
z4B8>)>q!s)IF{%6^xsqyzw5#?lEMFd)yZAHw|43}kCA4T@XUBcdEpn)I@&f&Cm6O-
zh;FfL>ji!ww*`C8rd9V!-bi2W-0yh0xKq(*_`-$z!M>`iU{FHBE)GGN&?;GhfuS7c
z(zv-1kj;1Kk-CN%-wioh7S~T+HvQ@K>Tqg#%kop*Tbz+1Q(*$nw)^!LC>tfe#-MF^
zIi^`HAS9%+^uNd6gi(D;BO8Ny9m}~|Qx)I3kn~L~BiFXn6HWbWM&T=}aT}s3*#w90
zV#KV>K?`?YGGPbDjq+xmt&#LZ#{;}fKaYGmfU!D-uk=ygMnVRf#AqU1rsrLGhw6DH
z-FE{=(-%fvqY%r9!eG+|x1Hl{W4YUt`%{hk)61Ra55c|95jw1LPrtSBE>=4$f8lSO
ze5WB?oq`~bR+WV5&?)o2WnZ!{Pd~$Ui54$)!nchb<){Y@7U!%xE<;&VnP$#p-zsrb
z0h4U1R^3+?UCT7%Ij$iqn0JBpb;7N4gc-tw$!psD4cQ!ccC`q?jEC&FLc{~Si4d{y
z6)7KVcim#HWguUH`B9co(#GJZ$-AQvC_l<bRRJe%1gq5{LdsH_x!N_$M0MFEy%0fm
znL-IzRf`h3u}TB7I21}YppR86C}#jntv!9Ku7!x(vnMz?LU!s}yS|a3zs}X~xlOmX
zR&`dv8<8>12x%U!S8yC^{0|?Q8A6E5*ht<IlyG~`iJnwV`7Q42R9zD_k0Of=6yDDw
z@_X|mu342xX2=c3+l67$ifLq@U(pf=wEwj95P7?u@o_bZgglMh&2wy0Y{dg8&nA|)
zpzJ3POo-fEqU*gF>Rl2V0{(&kpEuS<4(7Q}s7y9m7rS{$>dak32o@A9=s85T>G3u;
z)kp%{ON5Lsd%PD2_KF)R39)C}=4M}_ux?<KdO}C17NS-_{Uz{VcFyXl4s3h|#FKU~
z9OX79^C^00{uNgqchrHuXA8i?Lr*dAUz9lWP>@@5vWo4Oa&ymrACPr)iSHYvZP=c?
zGNLo&djJ|qY2-4n3LJ6BpmU~IihervPS6tSb?Zc3%2XcJ*6LBv+LoJt_TO1ZqePy%
zMTg(CS?KZU-j+rlvMb!YTK<8;vcEz$xS@?dp;cLu!9MAdnF<t7_w)zQ`FuUnx-;m{
zZ!KmO`1`ua=bl39EG2>zj?Jz6V!;*B-ws+{lXs8|NKBnHZ^6DE!s}>3bEy=H?}+Hg
zr;J>2Qv1pzOzY<m!OMlEaQ_vC<%sR(5|2)?glNw&)YB(~YT!BqiZ}7&ru&?AZ^YTZ
z8<`ak`bL#BR;;Lx^}~S(&}k77#-scA7uJZv3=*-{8zs$IheJG2MkwvYxi@!7tZJ@7
zSN}Gx@5X^`a7t;P>F^);<W=|;BZr7QtlQ>pU`^@MK~@9n(*xUicb1h)#C~AJoca~A
zC~L+6Vqyr7H$;gAyMqP0mV{lP+g}gAmF4SHno_={S%bWY2M|nr6$y$uU@B5v{76?i
zp2W|kBhvB2xFCMAK_N1V%B=D0nHr*3QPMrUTq3$;9et&!#CpLL$0+{9Uja`84;Ctt
z@&uX5Hi%_v^9o^|b&WU^XC8DC9O--?`TJ1AkK=EVclpDkuAbe8DJtT09ec^I6yx|-
z(7)G<%zfv3fxsPv3nYMzE_kOS2(wI7wx_q;GCq55$9f$^sX}&9J#rEjqaOh63h+i>
zRoI%lt85%q^1l2r6?UG7!3c(Pt)K)ODg}FQ>h5d4qdpYAQi`+pGv4kv<w|irsdF~l
z4fRZ(h0R-vZKcj{X4ullsek89?B>K0AcXD77<N7YdACd|*jslGazrlisrBH$F5BF4
zXBuS}QFwZi%oM89VI<aom&Oi^gUj$Xqtcm~#}w?i7s)BP=qqGeJ)j#${eXyEK-VJH
z8%&mIFG`=vlNCPGIutSvPAXL;`JNuELdFoh@SvNCXrW2FtJK@URi69Ei$Xb)mn@iK
zw+{{AL@?6_jr>wzCJ49GA)H_H3kf(DwXVxf56wDSo9_1$^Mwbxqo9F^e(Qh60e$9w
zo|3?deBUv7wKM(V#0hE}ms!jhV*p{*wxof>_1HmP^~CDxYq*q(%wZR^HjFY7_imx#
zc7ehh2G^k*YNfPjHFU27X^LXo<FD&E@c_bP8TJin`#k?XZZ?zt<=Gd*!$eeV+XJ!V
zZRiHayoO@AV0tgHr$47iq^0xC(w`H<veYQwDa^QuDI0E&`7I8aLM(@gh&+4mune}-
z3W?N{R<>C1`!63syG3;RX64-|f!@=B!EWMTUex3d{_Aului9wNQkf6h?Kf@uNeNKJ
zYU+M-|1N%Zv%1=MDV2`P=)2{Gb^=5+vnm&UTb4E1HgwIL7}8T<$e&?YWm`L$suXBI
zDF1_&R0FT;^Xt%=c|T9Hyc{A6_qNF0mJYwL?gclyuR)Y0FU6;5$9Y$kt=#fX24l)f
zpDR;h$kD~%XJvaq0T;V&P|L}ZULeBV@T{iWa@$bik49u(5v|2gW5Qn>FoAjk7RxCU
zL_e&yc$(l7@rGi+slzPaD(yUT2wO){Mg3X|xrf(ycKd}L{jw42_T7niZ>W}OAt4ol
zSw_o--Kg?mV>@9x6+6~JVLr!Iz_CN6J>~ZrBIBYj#w#Jdr%plF(7?+{An-wBRTZe@
z>S!w5pe{;yp*I~|G}U0)b+~!AQLpS-0fL8W)6$zApZ}UJ7ryk`M7GPoY?{xSOhfp_
zltzcq*yxM>emL*!=GyHE*ZaacR<bTNYAdPsn;mghzaNrFCQLyKBr4P5YUU<e%|Yt1
z%3kaIsb$WCzY9kh>!-WG=&Pje=8$ECDYJIj>MtR!b*@KmN4x8Uiw|=A=S<eA#SQ0!
z50hWP=^W_1dj93A@!tAy#(Kd+(;%-H6@*fiWN7SP2D3g?RN26Cw`WToQs%9sLT(Ft
zF(sVYYU|V)?&BRB&G9{q7H{YuVRDBCsGm15qtLRlPc<=-jkaQdYSt+dN?|AljkvuU
zy*XVy?jM~8OUCQF*SP2V#4d=rJ5N~S@*hpp1dC|7cE$w@;`%Sr3_VjKw`Y0N$tszw
zC$uqQdKkciCXMc!kH~X-i!+i>IY5uWS0tvAMA!LR&Z$*2^?#c~Fl|h8D-?)7@u5zP
z&!AS1<AM!NgjfLquyoQGu0(2%56Q95Gew*V3P@iUPpqfMp!y^PBsfgVp!PEQ>S|A5
zw)sP&LS*sk+&^Ar6Fw(a6riW)H26~IVBxhAh@jVzXHgY;_H=JBDX)2N`qBkD+Sbii
zSWifbWbji;{>3$38s&cCoq3S<tc=SS8nD1M$e!2LPg4UVjY;p+a{Ck$+p64N^eilg
zqQTo1#O!}Nd7AXqcCb@Pl72`u%O7Du0v<MWxEOQq)s;rYUO^yoHu&LchE3q@J0&#G
z&Z{A+HG{cr9fcg=OrJV}G88>2B{JUH;Ost3Pxt<Q?UDI>w5x2;8@xJn>ssg^czf1H
z&QpfI>i|N~U1-?66TgYM_|vu6B)T{4q2gMv;Fo@$VF3{b-HZHitI#7t(7ML`O<Q&K
z-$M0R1!g`J*u#uZawA&UXm9&Cg-+_no)e`orT_J+^hyyM?V^jPh#0(DK05Ggb~;%?
zg0C`?+Qv{^jhXa`*pOq#a}Vq;gsLgC<Z{W&ZGZ>Md4*lm9%0>{nV7dV5cw>Jn4XDY
zqWBa4h^?_%@($fn_r*QGh3l}(&CSDOh>*GoJx$FC2HCsSuKDiq)cono!Kdo~-#vDs
zeJCSsNn;p(bnInXt4EO3y~J2hF@0BoQbreau8)2Z7?-9fkqz@{6=(6Np2WhCe2<AU
z!VV8@8CSQ)jqg`qSew_8;IR;TFP<s2*1LYlc38Wdzhb>b-(Qf*om^H5oy_v(fv$<-
z>8yhHF%kEHtjkFmy9fKyCHFTw+f^WM8dBJsqa)V$Y6bz5t>3X}{AE|;`ggX6zH*pm
zvcm%+R;T6t@*Y+-VeOgE1^@N|>o*D3U3V^xfAeB{{n0J^6GIzrQV(@lYR^>!G&W*f
zkCFr@wONonI<%8MGg9NlvhFWa>pU1BBTo5_q5a<)gC(-7E`7YRm2zKO^0f0K%qcQn
zg?Og>h0EG_w@GwKL7g`Gc^|bc4wYgYK(a(}15Qc882PXIBsSRutIJ$~RzW>O<)EQe
zcHN_3;`#tMlMX>G=C{*~WJ3AxTrfVn1Z74uSq82ud^(iFwO>au!duW8Y(;xEJrq-g
z?w02L2Wi7LMK`xg&EsNh8a;SZ@9iVDwP~?XPoG>R8&g!SQ)iim3iP15lytzLFf6$9
z@_f0g#&L#!&@EkrN(N}h99}^fRtA$)=-j^hTM1(Z&wN?cg3<oQZH5nUtUjGxb_<M6
zA8~t?w<olJL`ZCOV>fBg{ierAPJEkh(3wqDpTX&9&O;WN6CC15r=pkDQX0xM<OG7!
zo*k6)LT94FP|}WU4gy+rA!n}s;po>WcXwvuuLN|F%qa>!B!H~K-28L(bs;TxXNmXG
zM9Ht@1qOsv-zmmg#6DNA$c6<d{p1i>Pl!*%--+UK1y7G8n8l1_NPhompWQg}#jTqT
z5%PpUGf_xeC5z9rN-IstCRuT}(1(6x&r4>a$-R+palu=0s6A8jK?1XERI@zcC>4p#
z2)zWG(ua@9-b%pIWR%&;D_BUK<WAp5jJC-}<d;GDFK(~BM=MY%ENNK1ikzhg?A98*
z<C0EA92be)?TzlFzNl&F{vB8}ZMou)2#`l)ompW`=(}qnfk?}yW*)i+zY4ENOTugT
zwJRb$p`j-obSpuq_a(jeyJ}V~Q`=0Wwo=I6pe>iVGeD=Lm`iO;B1_#28>E2(JN5Nr
zh3KB^yK6MY66EJuO9c)_huRgjjbLd&6>i;GGf^otKFpMVPEPJ_WNdjU@fjrS`XA#9
z=%$|#)2K4UMl~a{C~`;!`ska@UJ_sWiH)DkhBdlC6rN?RUCdwV^HvP}@MPL{r+rYi
zW&@cAEP$h}MGBVZ-0?C0_C_}(QnR?8roa~6g*7*n#>#GGH4b<E$VCWY+;%%VJHhkc
zd7YDMLXBJM=%R?eFGe&X>&47z#!P$&Qgz>{EkjVHj44%v)4PW@u)ck*cWK(%Vh#>Z
z)t<Q5RPIAa%8W3Py1B2TaQd+86n@K(ez^gFh-Douyo*0^gN-pQj;PtxFuQQ}ct5T8
ziAu)o?NKbL8gq#t+_(^XIij|FL>ON?yjd(f3kUWte$6sjU#OZ5hWKBV^j<YQ+hEJS
zT9`iF0j`?i7z&o<l7p0Ic0qd`Yg=XWA9`*I<!%v`VJ3(ggSgcEK%aC?7TFMlb1Q(%
z9l1;)ZLSIZO13sV-#ls~#6&@hs!ROxqH8dQ)N(HM9lpjUvukJ>H8&;xGsO7IjD|7y
zx<$if2AYLe@IO`s1QV>u%^Q2F!o)cuu=Eps^(X4~tI)SfD&LCEB5Cu*;GA>hAy{T|
zZg*>{3@%*)8_oj`J4#*#;to08tVUpIq2BldbV)T6*g<2lwP)f$==>|5qsba$gny4f
zs|aQOM^KXCv|kbXv*^)(iVY7wxuJJ2I_Rv!4o|0NhGv>P<<oiCAG~VKdCvB8PsnwS
z2_elLGj%&bS4AY3IIXoz3mifKe0g=gI{(jkjU*J6MC=s#xI-N?X5t~@1wb1jR0ST4
z-CIE7Kja(a&D1{+r@%?xAE5`ci^#LFQP@&4){IwYDhSOl&LHS#s+=bX@L|L<Dl9^y
z$waOrElaIVQXZkP+QPc+;|Y7Ud*LfSLPS|dPWjv!WN#Vbo|AAs`KW*eq0<iQ85+V2
zlIMk}^@*Pluu)Pee<8++!mcsH_2{|Erhntt(-zJ?Lw6bl`mgwdnL~zU)!4Sl7HBQ9
zK1RqSr=^~!raD$t@uacHdci1&2K#N=#Xmp)-r0(;qeOtj7a<@x+?2h0U_(kZ8gmf3
z1I0wBQ0-}Xwjf|8uHAtIVC?4MV|9Y4#8P4gIrG|smC%3b-p@W01N0)GN4R(YF+jhe
zn&yfy@T|44HYsuWpTr6LY#dFNn!>p0;#P$U98LV2AEq+SHO(rE1*W!xKQq<xkqneK
ze_0I3es^i3+orYHss`Te!AGO2cUu=*tDj!F3CYK?e3_FKQlj#L2byW4iioq1Hb;0w
z=%qTZSJ!l(TN>+$0jnIu65gr2T#qxc$ZaIUJH>}a*$ySoY$@r}Y{feK?~iOe$66kH
zBMfphOH23jHwTZrBpCDSd%V1OB}Cw{Br9E)3fbh@HrTbMJ5k-6e>Z+Y1>8|?i5%L4
zKCW?tYj@K_Gg@0za}G(V6&udrXNHTuYV%$nP^|}x$aaZjfIhK$7HJ0W`|Q#2l9G?C
zbn<fja;LsTFqbH+WpCoHf4a%Vs1RFuu%CGB2X$0*-l6(=IUmxT!ED*FuyoIMQPF!u
zI;fCaeYpGU^e9V3OV2?sd*^bI^&UGY;Y>rQYO|0^saj}2*y+v%5_|*P!s{rmS}&JC
zZ#HgJD%aTn&||)4(SM5rxM7z%z5hO6{c(*iAhxvL0)1YxYm~NpQ$Q9Yr1s@xe9Zk=
zESD>2NwkQ_b|Mp(N%PfAlPHM;?-gzNx-FZJ3yW+7uR};gs}U|fZ802`fomX<!%$KE
zO3J2b4hmYL{vH`mmI{iA(Gy`JrzwLmtVZ+6&!v9_va*`DT4_X*oBU`#*SkVynLFqP
zOycD~%Bzk@svNRJjYv*0oFGlT5dFMbMEI92^VOp2nf8`}4R@zH;XyqM1`DZiS~GdD
zWCL0u+aNj3AkKn<Jj*U&4UH;w^U06t&%fRC>NM*)Ttkmq{GUyYpIm;BVA!ps<@j?7
zrJ@%G^s+DS0quPi3*>cApnhEQe6fLBmxa^JgN!kQu~5%Ngp(F_V<JHM7wbRzz_A0%
zKe>nn2_E{lu!oV67Xk=U21-*qiJhN6W^W5if~1R*d+u`NJUShP)T3%&V@;{=6i*h%
zOOJG^gem%B*Y%|o@hVehVZk*IMR#~5Nn?bvmWC<t9rS{-4u%;3>_FBWADi-1s>g}7
z&7Gbj!CoxTwH_VnH;ne7S@!7<!i@vmx02P$xE^jqj+?~7C3a*&4p}4~ckLyLEznPv
zCD30d$%w>_aM5*X)xJVVtuV#NBdBrkJS(zUQb<hy;;~6HbH#9xzf~)sFD3d9%Y!ev
zzw^A~=C0kZgKRzoBF#6z`&6<!yAQj02*G{0Uh=2%n^11)A}uM)i4s{XyQCuxdi&|c
zlq|+e5-_1FxsA0B^D$~m+H#WFq;5;H2x3<L{$VM2u5H<|QOMslq{hI$=!}Iy9mfk1
z1zJ@TqKO<~5W!2;S95RSk_AF?y$~K^ne1#%u9OcG(y?cuq9R5H3ASM>uHkz3;SH*t
z=%n<cyO=JD21_ule+_n=ylRh4W23S<abJvN(_ZReY!a%HR&N3H2#vd#+s9||Y7LY+
z6^4j1azq~#h(7hH2#I#rtC&yUao5Wp)P0cr@Cm&Nv~{sb>}M_K&a^o`y-0rm2p>0E
zUVSP`VFqa3W1Ae`ca!P~Icpy?b}*MFP5!hMGU#A}=|sf&(vnc%q>lV{|Mq)y)WziE
z{gw4ie8~l>B=qi*X)ECDMDD&hjaBC6YVKZUvb3T1WQQ&_tw|`+P?yfUKKkr(z#D0P
zJ=!MMQ_uh^JL+1wE>`m1zoe6^MhMZDJ97{@{{nE>sxLo`N^*a46R&e#+`uYNYpYW$
zy_o*dXH3)mvHbEa+CpDhWGwv)bcg=%_guUvQSgl$yj{6W35-~;qdS<yRzo_@u9%1h
zYDnXWbaav`-R265NgDw#E)#1*HoHJRjV{OU*+P&2ddd)TF?k>u5K-Ig5Bf-ea$n1L
z0p=FjvJquRZwiyij;sGk9t|o~FB0OkEd(EP(Fdm`{QNHib8tQOS_jT~LY8twqW6^{
zMZ{*%y_ogg2Tz$0htB()l?!7vN~M?<&!PR{G#>B2%#xn2+lY=w9O7g7IknT|ABi`<
zP2bu8zaaJQU9;y5K6@5bc=N3Rq70!koj|vG{C5v?c($+{$Kdb#!NtO#e_E|D;jVQJ
zi=ibw+GmDs0H!B5EG&$JuYj6=G=qLVj(X9{(V4RREgt^o2M;OJqS;hqaJ<vW@D_=w
zdA#_CvMY!^?9Z~5^h2me=XwLLLu(q|SIzIw)#mhfh{}k_2Ca&J>t}dsP-M2V=nCA<
zp?UCxBYho38O>Fw6p`Wvs=%MDUk%k;P5fWgvLkUL*c`vo`Td}oYr)r7r=%B-uGJFk
zwtA#UrJcWSrWh(*2<cx5y-f+cyO2)uf#APSFOR}uV||j&o+mT%GUMR{Ut=VfqqbWS
z75myB$QOb{t6a&voiUdmv~x47R=t2VO|1Ud*5@=y4Mn}kk4`NCF%j0d7T{vvz}YBc
zbq06c%-I%6#+0jakr#bV2t&q#>P9X$_KJi)(}T(xVRmp4cv66tFLB>|XDP&R_PXyO
z+YzaZNVgYEB$LZ(SSROKQzhaDn?|BA><<Cj?)OK<UDIreH!b?tO(KJOy{-Ui+TPU@
z1akGgI-T!Oc6x8mDiOye2M(@+V)}SBQHlb|@)MeBIFt{l^av2|ZQ1Kysmytx-T#pp
zyXyr;WdAo`D_$&9FBjx0baT>sND5q}+y_1}Qe-l1i2NCC^D5BWJHk8U>Zbqklb?H~
z5Bs<O2O&Y;zCB}%1Yfwk>$<J!_Sxlef01X%%*e~-JPmrbZAOI~G^_gRYH011pwku@
zJcmqyRtgD1XoW>km~jn6LlEYeVldi79W$4gW%Kape9|ut_a}<u#pT83WXmTcrGszF
zww^ENZcxluOHMo;ZDcMFk<TwKjM7EVhUl%$o3=X%ooF+!{pGSmNt6B{_&6N(f1K5;
zx=Soo)kV}z9Z8$KmlLJ+HxLp#W1Z0<1T<7N!H0S@$gkhLwp=YVqHn|T@mWG$bzTBT
zgnqvVz+JoQ=T0C+K~@^jUM<sTI{1}e{f(QquJ7*-KmOp|hYz2=`PLVo9zD9X`_f?4
zvjEHUS=V%B)n31L>+EXr=F4}Nf3r(F>d($kukRmx|NW23I^KQzvndTR`j_t>e0=|t
z;4!#i-`JEqC%Ad@uyC}xIDYTF_ipSs-A6kbeDw6#B&rt*BX%7qTVT8VAO5TVv{0w-
z{b>`C`WgJ@ckb`ry1w+VwX=V5|IwF!@+TjE`kUK>(b4f-sJJ-0oc7(ke`<TX+u|%=
zF-Pf`gCa<15cfudd)E#rwC5LB9teP{wgCl`QC6?`;^K6=HE<br9eZDn#>2zSshrJ1
zk~#!JWCT)rz1)R>S^*1dWe{mCCe*^4TSLOQxR~u-+r2tH-W+bc`%nJy?VZspll+gK
zT*TBG2iDl-YMB|+)CprYf4i(Cf;DDVmBZn1RWA_IhRgf?;iCtSlbTdj7>=E>6n$7N
z=9B)wx^%U?bfSS3{D@lgAOc;N_765m+u7B8ys<r+6kd4Y7d>lLuc%V;i`CE-07@7c
z2^FE16g(rWQ!LD^wAQ3GqE<?wiV~R<DFRNA!l3A9Hd`&ri>vvae^=kw*x7mb$-9R>
z8k0mBAt0?JDDAQ+2_>xi#~}nwC`u_oqYYwSBO#!3MXx{V=ZF*)v5<-qQDH!dApt7u
za^aYKrI55z5u~juW{w;Y!B|a}IEBu$Q7X$!Ut66yBmjtrXsn@0KXK!IL#TlfQ7M6H
zxmsM#x~eQ(wzkC%e|mkROyg^i#E=MtY0XRpQJ|C+z{C*<t=2@k^PVEB#DUo}x8BDX
z5&>JSQVJ3yvlgn$rYh@v(2f-C>}+og$4}%Gi9|rwH{ySZ=RE=xAVCr#U_d~_Vv}e9
zgJ{;)E&)P76a@AuR--}o{K=!u$>iRhd)u3PxgKofn|ss!f8%FI-+t$vDup`6^W)?B
z)!B>}H8sypAAk74_m7`GS<J7>YSsA86}{>94l3RH*c%QR@$BNFY5j6pGJ_$T8b3ci
z>t<yU==6h!BXqix?ZL!>!3d5l$=XPfQes40r+3$|W@F6yr*qC)>zqSUtIDtHX0oxx
zfW6*eHk*a$e^;wjp63Y2DQP0B(KoAlb=j87t}1snr(0WF{o!y?HD{Nzyw^`DwOzL{
zooZvd;Ill7AvSHRv;ly!tdvrOWUNv@h(<B9F@^}OwbojN1c|NED7ad!+PZuHy${Q>
z_9+Re)(0d-+9;((n3+9ulYo^LtrvCWQ(U&~{l`z*f66beRw<?DM@N(C#&~OT|H%W-
zu|F6+dHQ^7cQ@-58{1pP>FYNSzWE2=y!Yz8)3dY7`TXK)cJ0QETQA-DtV)B^y1Tl(
zy>nnXxj4Gq-q^i5tyZ)4?D**X;w(nKb#3?Vot>LoyPJc9LADogd*<!4C7&;<x?X(#
z?#}g_fBO&aKVG)+-48z5-r4IgH%f0!#!sGHqSC+n&F`LE)#S4A#%`Vu*mw7e{?}i9
z{U=`i!YJ?OPDSrE>P#U>$}E2Mwb%LxN$I1->gjB8TKkK-nbqy%lljxDSw;Gz=O^>3
zT`emROrhHyjlOW}*4>@0Nv3+1GJ^^^)7q$Ye-=)RQj%C$thT_c$R8eT{OAAmfB*A8
z_tX1({ln{<dPtXjh&Q*7=cg$Kg{Ca=$@3Yrp?us#_55<yR$ZpYA(BSdR&kIGp~L6T
z7WW^Vy!yG<V@gHQudC<-?`%)*y|i`v&R~!yr%c@>pM*X;BiIi+(jVquc<b}ZIMmsV
ze_Jo_>|9F;XIG1}%QMQY?G>wfwOp0nw*aDv2vJh{!2*oL5C}k%AyP^i1hsBd2tbUn
z^Ich2A%rZ?t#g<2MPBspyz*KEs#~8pF@e%5grJluiUE*`0aIj1yzU?)r{SQOPDjJx
zxZfWR`h$Kku+C+9-s|;ygF$~dRN9Ine^>N$=7=<bLI6!l5orJmoI)GBMmQLwvdpb3
zhC(b{zBL_Br=w9nHyROGkvJ=WfhA}|MKA9S2SBu1trm;fs$7Q98I3N>ilQirqHVi1
z&I}2V1Q0->vI@0AMW7H6)_yA@LR3xLMP>mM0R^;v=pccVQb;LCB2t`Sxvc7{e~U3?
zdA7B+nK_qYP>9H)0j<IxTI)5E0E-|20SW;q0S(NE07yv`11ix7NGL|};of9xV^AF2
z{oeP#$1%HmZ##2?4Mku3{8!$4{OBuR{nGExPG={}c0PB?oGp*LWp(!C`1Hwf4BRwj
z1QejQsupD_BF?#|M@Mbj7QNo)f6nf($jkYBG8$ja7p6<y+3e(MX}d%qZPi+B1Sqmd
zWQLSkgG4C?03f0@fp*Qz3n8SGw6a?3EX!K&n~<&+We^%oHl93wbo<Wj=O<@t*bp;J
z#uMOBuIAPfAWcX8#bU0NCT&hHE?XbSsAtE|`?<TSs*EXevd#&jEwZ+4f2+D-0By8m
z46P?vue>FtP-{&f#%hHKBE}dWT8$csl22V5_V*97ymxkS+4@w~ZD7F2A$ddwkciX?
zSVaucWmyw^1Xwj~(a_)fyTAC~{I~zb-rm;f<?K6u`W?*l?|=JSoMdx%|7yAP9M3LJ
z`{T*#^2}xK;e!YJhlh)PfARFu<2PS@qlw|ODhyrMoSvPYo^>Z@l~VclKR!mV)U?}{
z%GK)j!HwPi*2(4K@%_tyi(;~I@%(Aic81(&=(e}A%+cTcoB#V(&-~&qfA94#z4FKJ
zeO!j#%Vv0S{s|}98Re&oCFsfVdF7d}E|-VX(e;DF8ynjN4Ss9(fAPTfY2dWZ0|m|V
zJJ)swh5q3EXIBZ17wxhOJ{|+95E4WIh1TkE&mft@-7U2yc}q|v2V29kt4`*t7DxzF
zh?;<u3Lyeu8v+`Vl(bR?@b<y<7ypmHvGA)0pT7Ufts5&ZR{?MA?VT;#WlNxpLLCxa
z%$sGM7FDhDzU|qgf3uZ9oQ#U5sj8~+Ni`kRbwgC<xx2br_6wT=&1UUnxcSisA02Fu
zOrABXrBU`r7Ox~lKymM2M--l2&Nnx=x^9t>7F83=iniR|&HcsM`T3KRoyk^~Wtx<*
zm*|mMp;D+cX#z5&)-4?oVG+i)vJpdyz_ISyEX$0w&z?VDe^u4~;WaU~YJEx&qX#cY
znn=f(LZaY30uzeDv<C8Ot&1X$0XY~YkYx%b0+2CQ8+5J*$_VBr8<Id>Ho_o+MjH))
zpo9|&fQTjqDh1|s>Vyals(!CfT9eYcq1O8ly5)Q+!U9ZMH_&Q>5u|R*vRc(m)rF2U
zS!b2e%ILbSf7X*#e35g<_1OQPtHdgX)O7(suNR2_*#fNs6hIOn01gpyyZ(T#<Ns~b
z%r7sE>-BPDkrT4_tztkBrI3DDrPqiNkpP7OQ7K>$At6$d1T~{#Agwe?k`!5*L9uah
z_8hRwv+V5bd~d&2WHYcgwzBVi_np7}OTT>Y@ES2xf0tJeKKgj&7xRm&6ac$O+JaRk
z%Z=$3u7?OA1g*{beK=>wqfxzDUOPOTot_$ip>>n~plV7V(!5<Ng3dQUh@`!5dL&tv
z2~zOvy=S9hj0o(UgBMBFM93iot#t^&W`ombr;i>#xpla+xxLec7!og5<^JA&+qN6S
zB6Kw>e_T~pd;15W>#R1X=a*#i&c~S2;o;$@_wRd+Sy7Z#wY9Z*emO&hhO7YFEFbp<
zZ8$ogopg~QiLr$wZJlOLAOfgZm;u(?&utf2plRB~lJy49PfmJ+5vWBzuoT!j*LJP2
zClqFds5r4C4(wf4XiLk*vZ}hHqvZ$hf4H?be{#Lv+2xr}{%3#gUzuO7vMd{JPk-w-
z{^^}NcZZYd{YQ`9c;mJ0>E_wV5s`ZH?H_w~biBRynMF9z*tv&i7gZOuVqBCyx5`q;
zdmDxAEm!T?#r?LXi+K}-s5~LWVPUjMgUk+wy-)5xJDH_xJJZYB9xWKN5rLeX9_?>u
zf8#-qK!mxe>doP_=*xtLdz0xaKlx|eBE9>Ye{j~ex3;(DPapWY6C?YR-rd99x|(nA
z?cV?J(UXfg6EQM_j|rQ)6M#X_X_BqU)c7iOzHPaqR%t~7*SDrbdf{W2qA+kuHn%bH
za#=ZL6cTE5O6?Bv*~R0RU%NYQ%;4VZe~*uz4L7zAN82BN>yK}2?^Zl7JBg9ouFGr|
zeFUJUY@+vF<1<USNu%Mg?Lym7+0<I;)6+}s^mI^+#s(>_mhC4GpG`NWO$S$1lh7&@
zq94&PR%@O0@^`-bCx_Rs-@beE{NyAh%m?F(^I1xGRn@)i$;~%E-?rR%*_&(@f38@~
zS4|zeP-#+5I}*xFhDaztL{XTN2%rEak(fBe7*oo!d_0|;E|%w4SNqql9bUf?nq_2(
zO{J3|Wk6|85;;YQk^+KgU`FZp21SwedcC%)qXC6vlsJRd8m*NzPOD7k!>m6MvPcBV
zur{JJg9uVcf@<A5*MvY4v`_*he^vmLe6(6+d6pNsvA8O`1|)?LIi}=8N)nPz;2F9k
zRTrzyN1!Y#T;JuxoC!Qfg$Rt=kWw#-_&=Lh#~3+@wN7g*j3BhOZmmbD+89&-z@*S+
z);hNyZgXO1tpaSD${Op88Vs^BAp0&c5DKG~A4a6tiUEN|5d*}O5};xQe*mNuqe3zY
z2sL4dguK7E`_fCdZ@;wj?)U%rwb$Qx_uVI3ySr@&Dq`Q_U;UXsd!tML-{1eWCWXV%
zuxdj*zTgmpagk^qL)IT;!!aigT>}7lR;;C+N|Dx7mdmZ__~i7cSuLC;=8zGyBFRi{
zYz%%`u3}_oa#ZM?jWH_gfATzAmvv^b)@rR&N-@R|g3%5DLI_1sfXhvmSB-DGz>H;8
zw?T}{&MwY(wnkk@SyrUBZo4qMDnq@@Y^JmUt!Gy!AqlFh*Iv1IdUl#fm$4Is{$Rj9
zkS5P*Rn?Q}b_g-gv($DBNfQcbg#^e36(X$vRhikijI<Y2qU^_yf1d1YO;=60tWL|;
z#{@~hGaE$=ltQO%CXp3JO-U&geFMr_Ys$)N2PSh(Q|%t^^?UiPgBwlPyz#l$zVn@L
zUtONZ6pk;>=S};wKlk-dKKb~Q2cP(^yK(L2Y_$SomR0##m4?Nt{pd-l%plKP=on0@
zqrbknYoq=2=~HrUf82MMi<K880HeH!fl1+L(3ccXFISHAvJ7~mKiVIcZCEUp+Bk&t
zV_&#6KbbXMd$`^6(_!$f>B8Z3`k(zL|KWqD&(E)pmnvqX(P%PO-9}ZlO-lRIO$$`D
zO;!w(k5yBP(RrSsu+~VLVu~7Rqu1-oGS5tqz@V0OLLIHmf4ZhAv_83*ExQht)kte5
z7rkKWq7p!kd;P&-w(;?IKY91z{f(WGSNi!!kIc3=DLU>p#P?M1Vl^iq;8aY;SM#eF
z;&9L#PV(7ow(^_@Js$ST)v*OMN;yZl&8k(YqM9w5{-ih9B(1y=n-@DLr!)9LrsYR)
zw9+?j+^(ADfBla>Xxb))c>VDDs;S60uYwm{G7cx3FTM44dwOMh<HQm`CD6_lS}Ua$
z!a6nng0iS+{osU3X-+`|VoG&gD{F`2@vLbsFK4&!-q|@gc>g<3N6uwd90Dd0idq|O
zASI3x;}WsG001BWNkl<Zg}IBdD0=;V-&$*oQAL(~e<z9a%ot6~k;rOojmeeCmC2OJ
zLuf&1axMieoV*VRNF+k*OuHh{xyhxIm{epyX5k2uI7JSLL}@funb39u!pMSBbxjwz
zF59Z?nmVS4M(0_UIcMvpWL_^wDI`IJ7fS{Ju$CGcBkF}{g<00H{uc}!!~}whl-5Xs
zTCE$=f9vMNdIH%sb;yf6*Mg~Pn}j+hanANPMq2-nd8{_F-e_Q;_1F~=7=T0(C?ry_
zu2IOD(;>7?S1*=JlbQEE_+Z&^*;b^LZyH}VtFvc6ardR?2iJf9yWjhJzx2!h^I!ij
z|K|VlH+FUp9<3JDd|`S6K*L=Gi73$;E3t9Ce_l#L1e(M;dvS3RLfy1w?p&TJlulBU
zx+Zc0L~FE4$pVm|B~nT?b<K><IYp3V*$dt@b5Rsp>!xYCt}BYdFjJllhl8rF8Xuw$
zi&cFwTR7)ZS_1gC?-ZeLL&WUlqTHJfV@e5yW6JWponLKl>=(IG2m247A6si@vsuv}
ze=4o&wsppxot+oG{<Z7ZtB*d;9EkWLcXbnEj0BQMtg|SLfTUC#d;|dyG<I1xuBz)4
zo2Fx-n35n0r`%Z%tq8}aRl+V;5<}*UXv3sq7b0rcbNQgRx7+Xcvd_Ko+OPlCum8-?
z{JBp){kW`GI<xtpe=%Q7m3{Q+Nz=5mf7xvBVBbr6c6$2itFJ}>neojNtE$P2N!+fQ
zMna_A<;$i?VMV%UjPqS6J7M4=w{571!+1DQKxNA<vxCeK=!By}KY9H8D{tNNP|YsS
z-hY2SGJ`=s_jT}HJ0AB3%KY@#z8af4`KCU4^6c_hdXwv0FWtVg`49flua}k2fArz`
z*#)Q+Qn|Ca@w{r1)*%EA$r@*hj0lF#_7%3v<s#%wh+XINUICzvo*xsybbD)ObIh|P
zY7@G+woq8JzAi*zE^>E$^TuELslRmp*~ceGm!U>8w4c9rd-dqqo3~!u7^a`TX0Q{{
z?xVBw@6FD-ft}4tn-#11>~L?qfB*8n_i?$r1kNXetn+EHDD&}9{QUa<wr3Gi56_?7
z8tp<E`s2ZTxokQ5+NXaBYSm|Qv^KBaeQh?seERg!<@2NM-R;%#%KJDTj4QGrVKzSv
zdNAsZwq8H<%XV>jwF<G7XthGEP%Bm{YRs)p0tpC7Ddmy`g_=TANLCvWf9a~q5h@N>
zvny-+dxv+w_w9GCX65E&NFb;pDr8RDpmUl?m{}AKC&R&{Z!D2w(z)W5CNi3sQUXed
zk&No~qSkqqgEBG1q%4uul%*~SM*s#iAPfS88Z(_zuwu!FBvlE5l9VJ;Vhc)1SkN<Z
zD}hN+8Eqv<-89v_YFBk^e>(!~Wp28$<!sS5OA%yAfULlU2wEG&^+M8G6(S-ENrZfi
zL8wzMLJUa4*aax9aRQ1&p%6<%M%2h0IC+sqQf7$8!^y^Id@-9_EzziAJb3%Zzn102
z5Btl?+GrkF*5PVI&<vCk3MfI1L?F>X%Cmfbf4Zvby3T*>OJDl-e;@tP{$y`386-$q
zkv)6(;iupF<}3aEKl|nvzx1WA{*QnCudC4fo&WQ1fB(TJPeTWjiF0mb2N`A7VUX4s
zBO=C_Jntc5zu!MOJLP2S<z;V_4|_QP#h83oZEs8n(0f1Z6`|`C62_Ei4FJ5>Kl=od
za;FJZjL~~<t&K5Sf0L~{If=QhmdvT`x{J%Jt*vPxO0Rt0%e#7+m@ls8H?Cb@m9xdN
z3sFWr*ERLUVyUeIlzVsYuy5y=WzfZZJ{Q+p)pgmlli?^cMpAHDesp}&8xEbzfMSTr
z8Ut9gO+_!V#R4Lvl(A~91`Uz~l~U^2=@~#Wpq(osHw*|Oe_CnV@2AkFAl3YGy0!b(
zSKex>^6}BhY}F3?#meJsakZ+dP0qgYjlX|z?b>8Gy1cj;jK=f#FNdQMAl`fB<@3|4
zrj1$VdV~J6=SRD{k{89YtbjkW!oe~1j2TXb=a<#8yt3A|q02`Tqf$M)%8B|#(F?WB
zoFZb+LT**Yf2>d^=jE_Jh+VV1y!x|W`-N}(!(X|5cx}FzXSup{us=V$Xggodf-8DT
zQzTj~7XRwc{`v3x&L5nepL~3LzDlY(J?g8^k1So^*=W;M>q`Z)T$G+yoM@VvdFwUm
zB?lC?gieXW)HKzcO%;REnlT+;oE3RLAqM8A?G#91f3=Su2|6HB76qH8Cg&KS?&9)%
zY4hRde)1RoyPI!6c=El!^~?W{U*cYxl<z&aTJ^Q=+YD`W^ZLew+TWU;e(v_x#iD7w
z+v=I^vDK=WkId@$sxaBXjlFk1dE$Xa!*LzjyjRp$M+bYGnMO>II3!6yQa|tWkBnZ4
zlDL~)f1WF%gM)*%>*iNWYi*JDQ-o?YySSV!Xw|zhU=6!B538oSm|sL+yZ%rsO<JKM
zt+g@c2f-yIrL-bL1{vGBUC(c0h+x#lWaH%MsA<~If8niv`fI;3FXwsT0<+SnH5oR>
z<eAHg903qG4hEyqa7!tzRlzaTbvqvPV-!&`f9&`3Y-F+l>D)LMlXxGMphWOKFk@s!
z2K56?$GTgxh9A8Mwku^#0?;vdoov59;Mg_M1A?-~BqgL$67NG@HC@+v?^8<s{$M)Y
z7z_rNmzQnZDy1~7r@uf%h@@Uz%@racN@5|xb-PA@l_ISbvuLd|%<+ZK7Eu5ab4n<h
ze`D<0RvB$<78v@&@z(a<v(w9lDJHmi{no3mys<8H|AM0vp)e=G^+*6Y#?%6U2%$ud
zUfWCqDhUc`d$TuR9KZ3opSZX>V^Fi%Jhq)A-*Mfu?$vvr`@>)Pr~mrbf4-P*N>lyc
z|JVO^nO1F)Ycui2Om?<n;}5PMESu%le`tH7Kd5H2i?j3U@{EjLU0o#h&1$Y84qVYL
z%jswwk>@@%O_N!ND4|(dZ8&z!$z+2?)c}A-O44T5a0=Sl$)vwpEpuy}(F`#w+*&vu
zASWzbX08_9@!6-l*Y<7S*g|h^>^6(JU)HCOk9(u><?`9-{Ayl~ZtU;d%&lgZf4$r-
zE@dznIoE%FT9yrX@5%I<rYWm78}++Rirf^IVqK@0+8DFkW+ov@idZ*3v4B!4MU#Qk
zstlY`LZp7t(^@wOZQHs+cil2{5t7M`u^|;%G8s+>Tjs{rbeKE#Wm{Q7<fKv6_0`U9
ze)jzA+V14mja#$D>T|EX4c0zCe?J|JcCX#IQCBsCx_RpsD%qTFc1>fHKG@xV_VhTm
zA<I59OfeCWb@v|}H+32gdjQZcdY7{*LZ?w~AMU4mas6NywLU&MN5ashASoZ_y`G))
zvs5jo6a5G8{K?T(l@wc(HOo%h{LZ~EeERV8>a0AQuT*ANZN1UYc3-)RfB)bgAKyP;
z=6%CBo(#Y7+u!`Dy9d5Y2$n-!)D>tQ0WKGF67E|Qn3E=h(o>3*QVBznPM}FuUSutI
zlgae>^gIbDrF7diZ5L7^P$r|U>AIL8q0O_7SqUhX{mGPD<jZABX=5<GCOy9ld!wnQ
z2%K`pLfNz9v-+(5@~wT=fAHJq&kv>pGsrTeqm$!n2m9Do!+vIy`t-r$Jk!UsCF&k&
zQo=>v)0z~E%k!!!n=XWys;1@7?u+?fd=N3jZoZf~XRWm%CKPDf(C=mA@dT0UvQy;R
z7=t7;$fD-U#pT%LN&^uPf!0bSy+GU~HrgPOaP-kLvvbb3jU*;Se@U1-cX~BnEUxx$
zT%YXjzx(ZnS#Aft!XzrlDO8HJL1&G1*7k>eLY28;mgQROWqq!-W`^L|W;sKi54U@h
zZRHBE7Dz?U9Fq^Bi?Qu|3T$ZYV@hk45+I5&LrTb?L>Vd&gUD*RQqJ~=qh5bfwb6+N
z?b;9$OI0>yS=Dvje>TmUGdvs)N5j!-KI`gAn2FFDtJMqDA0iR~;@Y+X@PqFaL?H!Q
zl=mJ0){;dbl*A+mEP@OSAP~AvtOFK_9Ahf0rW0d(!`^6fYdCR(jcd2=T)Tcl{V-{q
zj-UjQ0U0PH5m7=`Kw?ZJgs4y;#&&bsFC5J-&;H`S_H*C-f95yyzPmV|jmL+R@rDQ_
ziJQASc5CPMotJ;<Z~d)l=6Y8Bqu>9>iy?mb=f3`M<vXHBj~>`e{oa%JU%h$b(X%J#
zT1(qbihk}~)pS0#XV0JInH%<dy)5T0tX3;YytlJ;xheq|c<m*V;b;KDWVEras_Hg0
zxk0Va8EuS-f8G~aW{qmw76`R*i{&yedgZ(}K-y$oT>|QdkDiV;M$F~Ov#X(V(nL+|
ztU4~+HV8zZqqDPX`#aO|=<?#++B`;?E#^&|x{w47A|a47Hgg#<AtIp&fU%~~#`~sg
z>zIU@U6vbTASFd0xc(j&=NvQpwgH>@)+^AV3(goef4sfBRcO1rHRQNTv0BZ~2U*rR
zzBqZ-`BW$AV$4QEwZyh=Km6d+!9cz8>h1H(#f=*;ee+x2`s45a$(wJ!*|p*N!Sz4<
z!#~{J-uc3tZ$5eYV0ktF%9p-;|I>%d`D}J^vA4VbS(S#!a&$J6WHf2tblMp2W8HO=
zt)kb%e~0&<+`hR}F3PH{R;#F}C@a|7+g~l~t<A0Zv*#~uZuW-lhaY?#BD#L2b#Zl3
zxAo$qM>Aii!EoX+HuIJw{p=e*@zFp0olhP-qmAk5@#W_~|8`ZC@BIF6e}`p%FmR+$
zfY8(-Rc+`3$CMC36AcQ>!8;8Akq`*5>&nS^e-wQOF)pv>T5CwiBCD!NAgj6^<VDwb
zPB8(+#9bGZ(WBk~=F7V`ZvYE4{>eul{rWe4sj%km?zLPgtm@QxhzZ(mx-mUHJ^S!m
ze|r0+Yp1sN{Z)Ha)|g1xJ^bkO+V=RozEZ`gVw?^OpXB`N>e`*_$Cp=kZd_ko9TjGf
ze^O(#$unXRt8(=(p%D5^PD-(^7g=VAR1Azjr>7^vsFlt0yy%Yyqf{8zaYUzg_ICeZ
zwOp0Uo!-EbMg*3`LRsdlwNZd2t)FGJQeE&V#&se^Qi?1fk`)<2bN|!(*Is`4=Dk;b
z?{|OqY(58~!e~iJ68CyNXN}eh5quYmf5JNFilV2KI;vMOM$>nkAThG-=ab#R-XRzl
z5K>I3ZoT(l6$^wUF($#3Vpx00*V9pf#ISCo6Dp*X6mr+L{_=7-o(@LSRTDtB6oZOB
z#86jVT{p|+D)_X{l4MyHW2~DMf@qCDaJc?Jn)t%b_k!fA6agZD2qZ}YAVj3Ke@=--
zM2TTwU?$`>HUSfBZBQhI7-Cc*#ipr3XE!#!__d!LjJJB@&8!$&`@`>Tlu#uyNoA~t
z;3kl7L9vI$Zaz88hP%D-mh1Hex${k)tIf?xjB#sgD<#?4-g)_zSNpx8FetJCsMRVC
z>|gn7|K7re@%7Dr_pkl?FMZ+Of1;b;ef_q|;NI(ZHx9OMyz#Qm>`S+Aj)ub@|LWJ?
z`th&6{Ko5t*AGrlPX<L^7^4A%Q{q^)wd?1BqcbXnPAfI)4+xN%1t5grz2}saQtN>R
zyqIn)(h6Yx#AK8)!{L~jomE+3Q9(ke19v_>d~`Hl)}U=~IGr9`>x33{f2T~IK%>%a
z;Iqql)h6Zoi`qXqK7aV^Xueo<T}vQFLE*GQ@NEYGq+mV92%%e3`kc#+wTK9uqVG5*
zqP0j2nZbKcq>alMgax!U2x?FawzoDmH!`A9QZ*V3s%7QK^zx!!HQHvLd0tkwSNIe&
z=Zscu+wSe{-Mn>sdu!`Yf8TlM?YG|gkN<=JNC|8VMkmjYR#%H^xiX|4-G4BdjDGc3
ze`UFv*LBGpi_cCah=58I0UMQtPWiyAYMD8^HR<;h3&Z!`f4s3Vy!*;asP*Hcvz`50
zj~-us{L#hia=E)R-QAx$1qRj7S(14C{3(O9At|j*rYgT`Q&SY`f2F;xf9Dte^8flj
z{q11K^{V~KSAKkFb8njWuk9Uh6WuZxj^(N$2K31hI^qmrKq_M~%(+(1sJ@}0Gkv2r
ziY%A3Rb*Sa>nZ9x%PFMj6Z4{K0B9}EFvf<&ptLqNFf5kkXs|&_lMnCw+Hd^o-~C1p
zbW!?2D~H8K9t|wJf4%9}oom-#dHL2C-@3QIv+?kQ`(K?N?vUtc7mF?>@gR#Jy}{t|
z>{OIlRO+XG>Z`-TEM}{NYlrtAJj{#1haY~roL7+0YHdKeW(D#i+L#g%mc$&rZ>=-q
z@pyml@b2Av<MAXVQQE%p+UvvdWVV>sO*0$}opA~j0Xl27f7Sp9h{js;gQXT>7G_{Z
z0A^leabS&cQHoO<4u|u_+<X7|x4t;t*<H21>3rR`Rav7*=9D#{6@z$AYx{?Yq~&s+
zLeyF*rIC!y`ffC){)qacko8tEt$b_&k|MM&D+bnOq!bJI7=B<XPr@m#ODq5>MM?p%
z0H+jZvsv49fBoTbG#(jel`+m0nd>2{)v9cors-M%$cv&V3g(o8Clo`d5C~pu3j=^s
zN-Kp(|J)Wak^-#>U>!&j09Y^CASwh>q?A&c6s})angCJ=DaKW~T9l<Ix_R%Fw}0X%
zUw`|@?!5Zy{`Ff*sUM0!Q1?Ffs*q-KqH{2PRt2kqe<>rACm=#<>gDeC7@^tQ+t}Zq
zK6v=)@yYSqUwT_W4~JVfZoGnEDG2~10#-mqe6yHeo?M<DA8bq?e*Ey^`yWrp@YTF8
zYDigHRn>BFbaXPAZjDD1K|H@WU(7BXU}mgVN-Gt-XOO{Q5MosH0zg8D#u%kEB7su#
z#fp^5e=}z_X-!6}qUd?w5-C}W(2Pz3K7_98VhY-55Kh8n+cr%|Nm}oTlrc7h@bvj{
z*@ikKTl7c}M7j{0*4J&Q8sGZZ_yjiVf_IryikvoDAtok8TA@+EDYb!J(T~CB&P0(I
zqb98s8X~Qg_r43UNs$GVwr$&jy!ZqdWelmEf9>)9{zl<|AZo!uKCD({;=l|HSo%=;
zL|Gn~`+32sx%JY4Hu&BL?*U<v_p&Tc9Ga>+e*UbNWxj2a_k{5Bt9L*8^pol4=BJ-N
z*xB3L+1*=})hn<4%x6*>o;`c=i@)@Dv{Au}(aJ#V7nYP9?CqQ%A5A9Hz|QFGyYD_&
zf0VE_y>Wg~&gZM!*ETk{29Dt3{A9K2R$djO>GO*#P!7;EAt^{85@JHZsj+X~xb{aM
ze{%5oFH_dL^U`a7^)LK_>HNvb$<)WuYIQz4zg$$hmn%dq-Uw?1qqGpSBt!_IBVZ$h
zf(j9m&y|X8OOgQ9%EwMH2{x_oLM)84e@6M3=8Y#~BXgTL0uUfK%}ayP<Hz^^$?yHH
z?<%vnI>?5Xm-7b?pOq=qSMx7@`E8MSdU~|FsA64Ho^`8KjEhI}W!~!z`<W<cVhp~^
zhoib}%epC7=Ld(oN6+Ueh3Urh+0%<_hljoeK-IMoK;_nb{pbIUo!uV|kG}lwf5YEV
zT9sv`ls3k&tYHVqca4N@JQx~s4<9_WuJ`sA{|u_$#n~J~>f21C0wkLk{m}?OhbRO@
zj7m^35ku%ar<5eJB&9TJjmosZgUQCrFTZY$dGFo#j*m}_RtiLkpNxBZJKMvdC1OoV
zYhztuOkQMJU6+2j(n?T+DvHt8fA+?7*AAvp<wj@7N<tJLQfPfhDfke&6k3j6AjT9?
z1SBafFEaH>h`}l?Oi{RrsY_ur8kk<DtrY=4B?&Cq`F!!@$+N0yyZ|WG9}ahRcM(wu
zt^tc0HEIot)-4;-YV8tSV<OgkH%<bAoFK)ul_VlyT7y72k|L$7paKk*f0SK^<WW*g
zRS>W98?U@^<L)c{;fPRcGDaDov^Cw<>R%MR(V}c()Fw|tjb~$B2!cvw)<~OzFC{jb
z99YN-{q~Q2@!Q}2{X2JWzW@G*&JFUs>$*tD*mNdSNRVPgM<CMFHO4eS^~JrzS9Z;^
zsvjSnJU;$t-tr=N%G~z;f8pNsTgqmcbtk81o%dU#ktRgMrfIaMf~@ZvNr3<$#XQer
zN=liwZL=&Z%W{2R2q7r_V&a)nLPQ`EKy5TA9YZL~s%+}%c&v?yEOk|8Mc(=(NsO_p
zvQmNu3AD|!telrzhs~`qM_*TMNPavT0Y={?n-{Atu!u7{rGSbHf7_QNZQB@~rIfT$
zPE^yjF)|TZ>&jJ4h7GBdQcS`~P4Gb&wOW>CQS<>xQt~m9QdL>63d(Gl4)cU^adfu5
zvCk-|ZQZnx=xnx(q%fwU=vT`{mOG=}jXN(5hgpi@JKuiCT8mm0S$6Z<^|RB{+c$1K
zdiZd5e(~{xj~Mw|f8YAfzx8ka+j-u9^yu+;JpRmT10oA0R*0PGaD998>~f|u{rLIC
zYRTG`A<*~Vf5ys^$}X;!m#3FSVXUQ(KRnj4y|q6eWx;iedS#sxAq1_7s@g|pB!^wP
zdvG&LgZb9<)PyW%S>&tv$}j4hA-(_MhsTS{GvQKTYrfdqe;8N^3PEdZLmX>UB?eST
zI#Whc07%5p%dBOO3N900EU#uvt0)Q3kj`vwwc^f)ByAEW&~=`b4v8g7hV(n{f49Aw
z-30xm?X8^MqAZUxIH{|qzWOU~e4#AMci#D<3b<;0rt)&RsPe0?|Mb`Ihu<g_E2|va
zvZ>Kz^{m|9f0~j?SC@^+>S5m=Jz1JVBVJC9N^PjDyTq{=^n!n(yZg}`Mf%R|ds;m^
zIX?aFpT1jn_4fAmV9=k<E>nm`D@=wtfsnKPuYBcekKX_E-EaO*5VV?vv2dAF#)l-r
zh)8QvW8$?-762rH01%PVM(0J_b&8aL6obL%-ul9mf5)F*l}puk45_Z$z+$o<#wbil
z6e&YWMNT3yVJ93BopGku?*r=SMY`xLI9IfFlK_>;wTT^~kPD!db}=NN)UDqQBa%W8
zk+qNpwARMC$#f&{_gQ3hdBvoh%ZRilBP^tK>$|#Xlqs|}SyuG>L*I5}B&8%s2~Y{D
z7s_3wf0bsATI)5)Qz?~V0tF(%iq{xRL|m`;uP^577gI_Q6IoKIt#*RAY`s^sRSbsX
zDUptmUo75hVB{Za6Ql%I+k&(v!rW0WSG`O!@)?OQXXvZ#Nzd1Fg?w<Zdwl$SI^CE|
zCcUCxRn4=f#~`Scu|{=uO(9wqDBDo`VShB-f0&f)U;o1AVr;g@y}$ey|J=*B_X;Nl
zFt@$!twUB>Aj>Li17}%3aunpMS|LjCZRU)`AR!F<d1h3KVco5>)`k?Tx(*>IrJS|v
z_<XO|L&VlcM8zVjvaaje`=Fg|>-GGHs=D$if>v2CBZDah6zshBDMWy}?Et`N<FvAh
zf2Nz0w(AzFG9(tn5Tm?^C2{6-ZViEO2nx|y(uP>LZR?^}D5bmBH+8GER)B;8l31co
zv2MFO>$4zf<3kjH^=d9lGMO!_s%xXtxpDbqYtu{6y?o_;G^UIU#<{EzA@<Q(6TBy7
zc6RoUPESMPzwj4-{`B;i2)4F2MN*Nue{!{m(f51#e6cvZcH`jiTGPdkKY0MCKl<d;
z&#EqP;-={+g<Jb$BxocmMOD?#W^>`JU1i+yWO_YUgHU_!+B-KUpL=;TrLqlPY5VN{
zX&qG+)U2!{r#8iwW1D>0bgLEkhKc>=aQf1he_|sW*P&f-U3cM6|M*XaO>_Rcf4^7G
zW>*PT2+`T+oqx0p)XR6ai@Q7dt5bL2!tFfWDdc*OcWr%R819VZ&AsgPE%W;I-YbWL
z>l^vjbdpE|(!kkqZVKYqbq&x()dHxqvZ)hSZ3h4<wl{#MZTC~X;a}L;m{d($*AV%^
zS^d~4e*5~b(Bivw^#@N*o}8B#e{FYve%zl7r@CJro!sgVI%-nZWko#L8yV9|9+&0v
z==rj2_~TDbC!03~@aU{mw%2)I1vy_f^G0U#4$0t;sz=YvG4fzA8uW+Ea{Kn}FaOxv
zhx<Fi9Z1`>i_^15XXj5+3JA7o*cHQTH}9hDU)4<_1*EKT&gL%&S%5E!f4j&!_q=Yg
z#+X8g%<C(!0g-vRN(gsexp(LF*IL9ju}{e-2C~K#F7IoblX9q3mgj+k=T1;WV?qQn
z$`&dcs;p;SE`lMXB!pyGR7e!TbV97Mq@BtJ==zDYXd6%oX+{mGqfi$lqL5ZLb49O+
z!oi2c+{M6(lru@BtgFU%e?S<7lZY}#87qiuctMD9J%lr)G@`@^DgCn@S@=N?2a$*r
zp(0dT>wkVp0uf_O>)%oe0BB6s2b+8K|MT`9PnTucb>H6mnCqVZZ2HW~sw|V`yQ&*#
z12!6l00BWXVkm+#Lk$$c&|jj7#tMlM(vV_+1PM|QB%-^CMw>ESf4(ZKvNE&2{JeSg
zn&bA?;8k}}6Ez}K>N9=uM)$pa&sk@!-%@pg4s?IAqiiom1fdvNBqjDK{p#ejws{4>
zxwQzqy}kL;3wNukbe&twX8-%2Kj3aT$Tn8X=GL7%$4AGzTcfw$e)HSk`F5|@AB``+
z|NbMbGei}wfZ%*Kf2}fWt*UGH&RcI^-re$AA570cfwi!;(R=Nc=l0J|7Hxt=+uK+2
zVx)}qt9e-s!8@fw-PA}}mZbtLvTWfzAhxTOCWs;kz=@SoDT>j?=9wZDVl1;vYXpF`
z=c(yh0EG%&lZvufF6yoAT}vi}<YTJah7r8?S#Djutc<bNe?;ei&=B=|W#=PsOu?bj
z!H1%N7$b@(r5V^{h6s&9L#PNvDIp5MF$Zn3yeR6rPBBuEOJV>_F~rXG2E8ss5kRFv
ziXkLp41!_?K}<=+K$75^ZeABVQRTLBsXjY7sbgqdXHA(H8FlD_QUH;?uO~Z$;ds=#
z_Ja>TFiQRYe{X-UY3jq{qs3zJ{`>D=xqSKg=k6lm!$*(Clg<78qxo#*I={E~`1TV|
zU%PtqD++WD?f?KF07*naRGSTiy#imju{9X>5mu1eE0;DO?w?!4s;?%K%kO>mh*l>)
zL<YIEJ^cFJTU(RdrRIZsk7kG6q})AQ%vdF>^msD9f4G<{rBrS~+4B<*LZz;*J@f3c
z@tfRV&OhHjm|l5$6lSaDqxYW5X<EWzWJ`MC!z!fO^HY~^UfY^n-Y~v9ch0#cITy2B
z<qFr>T#KTJv!Zu=Hh(bpzUj8LVQm8^7S=Hh1P|sju-P)D-l#xA)rIZJXnb<=$K}Ru
z<MWU>e>4u(*JgaJGOU}Evx8Yay7&GYU)vtrmTzosl!Lse3Wh{3jJ3JhAFhI&E;^N$
zc~x!Z!;_;$<3iEz<zzp2_oF<^%BtvGtBIm(@+|j0ot-a4`oFdfy@(v29E?T-QfxEj
zT{4R1vxOvI<cd^|6hnwCKxmMxQswPC&uw45e|~iTqo^P+3d~I8gos+J%w!Tdgs^rK
zh7e+m%uJ|Kl&<w;BLZlpopa5_*=Tp`_6v9Kefm*7p8}*HL^cz$i4z)SwW@jp(bO)x
z>0*YUilQvCQCUo~e4Gs@y6TIu9Y^PUN+OKT#isR5;M3W%SzJJ9azn<j)uzl#<`g+-
ze<Wp$uxCk#LP{A$IzuDUww^$AW^`WIgdoetb<X<)0?24_v@y(jw(B5<L>dgOc`_i0
z5s_m`6m?;Lk&FdqCeqAnb0@)<jhkQY<v}2&lu`sh5n*MiD5)yJsIKK!=+f@x(dMo;
zrF=Q%g*m1anDnpQ(Q9inL->b(@VmeJf4#q>RXjgEIXONIo&BHx?4J({ZHg+ii}9eh
zTwEk6PESsU{eIIlC&y<mzw+eMPZh!}BvA`9=*&bPYUeIy^U3bjRcJ@O(dp^M+2P^v
z>gCupCDE<j-J{dTXv<r-o*Hdz2Z3g}+}OJ0<};yK_NscmNFfkHyIf&PMP^&ye=>lv
znK33r7W`7EQ3Q%GB~B^kSuP^R=q$5s*TuwDe_%4Zs8>o;uUGno3&3sD5K&AKP_#zM
zGYy~s6C%gh1!waDAuU()BFjSXF$LzNwPt{}Yn08cH3R}FDs4EW&ikUYMV>)i2i`z9
z5kgTE2}6uN&n*X0L@}nwtc;B@f2EkT%>+OrS(~Ml7?L0bfzylC)}*_-yHOO?#@6KE
z{(Wo9RoiBHx&C5dbJw;-WxVg6dh*F--Tt@#^uPOG|EK@)`#=1_#^&Z^G8v6V=jRu_
z{$R0MJ$dKpjg2ik=(E6$>rV)?LPZEEeP!5Ng|IvBPsZg(pMAc$*&FxAf9G{K+#D*C
zUB7zu=Rd#aLTgzRFhN8yKYH_xx^3CV&Ee(GA1@@;RYARhE_yv-;KaxbNeqKZ9D*O-
zxQ+@2U9gLb?(x}Xu}QOe;g(lpr*#g4&1N+-Sq9Mgh`tNwhv#@>mm@rN<r;G1om(w^
zh+d&E2SaEyMZiTDTEykbe-}2>EQ1K(MI-ct22qk7eUkh=B@1A$j4m_g=ILH}CB`1c
zgdAC75>1$r>PNRJDRX)jnmNj7Ys=1;oYVGXGVWKwH?!{Ix4-ks-~5|@*&knDt>WRd
zVT$8%VXO)qo7s{R7g>gg-i77LjfOpgXtfQ#F08f6a{RSTdhQq5e`MTm8W)2<JlsD~
zpfoBYw#+3Eki^y-oflaSfPl2Ud*#m4&z>EAQHN0VE1M}K$gIswmJ$n+j}bMICh&kH
z2*Q#ACuH$nVV*04K04<mFN(WQzjXM@qo4iYFI(TP+O}!iBw%c=O>S(iwT?c-km|Y1
zNcDOH)f;VGxjER}f0nA3RRv{v+q6JPnp{LD=u)a(yQo*b@eo3rg2`b#7`TAM2uJ`)
z8D)xGA~SdZP@-ZINFa(k??-C1HQCJb`~VLGKr7|EchOzGdhPCuFRB#hCntpKyeX!{
zgn|OXX?>x}ky!x?AP~xz44{<OiA73@IYEjnEC8C&Ac(C0e+0w{2uzmsszK4a{N(QC
z8&Bk;322oPb4maSfB}Ji75@ML^ndeT{Fkr2`eNcXH*$D<xIDjT=8M&0v6!C_%hq^U
z=2>3i=GGu`dwzb>b={y}e)!?pd|F?>`V^rCi57IAh?YVMv&CYxyEFT=QFi=jf7;d^
z=4GSkCm(*if4#MP<N8gfH}4;>h`8-qGFjvpVz;OlTYZDV1rc~x80|E87LSRoNhuLh
zmja-QA&3Zb%5#IjoY)69$$CmdUC#xmHykxCh{+apLns@ALFgKfvRE!5$wqGwV(hq!
zZPzs`)9=@9mlEW~z@=5^z0J&g)dCtHoK}=sjmU{Ze{PMJlo1$C0wTy=ic8-Vm136!
zz+^_nm>XzTT@zweSV*ev7P(bX0unM)@NBFy#&Q6WbvlYcB8f^&Y|51n5fD;r>t)w*
zf3V%Ukbn^plqdw1VPUKy)*6?qS=rzlKl{mRzxj>*#|OQ#RKoSLzIpSh<Kwfp-ua|I
z?A6Uue=GIrrysxk;`2K@8{1oUJc3U?`O2gl6281S-a9|hMTX47jh-EDe!jP_GyUdU
z@0$cy3Ps8+d~@~2wHNPRLtdRtrwK-X{;z*F@3bZ}YvM4OElta&^^rNbepR*JwJBJm
ze&=hinEQJVPN$(nwJ|B|Hn#q0tv92P?cHF$f1IXu-33WiS=Ou72GCD`_P!6RZ#;YJ
z)w@^wm9kn(D%iUa5D>|jwhbq<=JC-|1=?oW5PeNs8;KY+h^E*^cg@(>wRt;)`zkFV
z4cf5FS-*`+L<=SnL<l4bLZ4~T6^|a&VU!h{Wp329tqrT7j$M^Z@GbPJH$MI7iQ6~#
ze;+3fyqG)C*mbKaFE!;U0!qsA$~oV4-kP!tfFKgR)l_JsQ~H(2s;_X>%jLz%IRnKI
z1elCX$Uta|9E^#+2Esfu>+3=9Jt4mE(yO1}`|#}G!Mbx1MBo=DlgLSF#loNnm9Pse
zERupS#E1?}W>N|uQ7ao<vSeR=`88M1f8PH7U+teB%@!B+a--<=tHF>^dGDh0zKi{8
zoMoA=#{I2p{q0?<`j}_h5P$#?DoTJMLI@It1dOOWNR5(($Z63M1*5qtN?_+x%aV#B
z&wHAa4n8phF*6D%9lLHdTU15QwcTtsT`U%D<AP)8+28qlfBfRBUwiXMf6bDNe>NN<
z2x=e#wQe;45rQx6l>iDDSu2gW9`tLYy)cT92#XM*LgaOCxU8$6v{prB2jh*&#_si7
zw&*v3JMRo==CpRNA_&T_R`&FZFWgPhYeP*_yVgB^y!X*ZpIEJfa~tFFVln4rZasNN
zsC6+ZrCxmD#o=)H&2N75;lrbse_#48u6MEPHY_5oby1dpc<J)xPd@qd;P9}jY*Ci?
zKKq;yc6P7y2fY_Jt}giA#rerrZxB-?f$4Mx0?cBpDF=PGSb6Uo=X9R0n#SfCDHTJ^
zP<!u0ASEUu?_F6GnYFpm0I0QzF@_}Awuv!Gms0BWD`QPekvRZ>QYyw6e?#Dunz|u`
zJkL{t`Fv$kOd$d?YwY)XAx7`LF~)nZl<M_*het=f(by1+urVfaB1Oh%?|q?65@F}t
zrb%dwH3)(TDFth-F{*VzYok<JpGLG&0xE<cAnU&{5;3#$2?b70&bCW47;YS$eZE|^
zoe#pv=|rTMB{2&#A&8{$e|V%dz5DLFcb>j;>xm~2z-oQ}{)0U4XJxUuy?O2W)xn_m
z=1+h8Cx84$_wIc%-rRiaop%pTF2>`HuWBwJLDtVp-Gk8OMbScgv6y#VGtTv7C<SKX
zILp-hlt2FXzN^owTy}OgV_P$WB&5h$o)PJkAf;%n1t2F0K@-4ce^pHP_Ls}}jeqbD
z_l{1UymGV32g9@D{NMc<Eti3`LFj87L}*yJiC6lgO;JAZRPdwEAAIrn(Q{AUFeY$}
zv111Ii8~)woj<$i=JQ~aEto2eOFzp&n}j)Gj<}r_qr|(BceMGe_MdUL3v|b-WDz5A
zjDW0wzRXlw=|_37e_vLb-=EGt4y!v~d(Dbmn_Mb$^I&f|xN~cwLiu1%!Lgvu411B_
z^}O{4OusTf)H)wS1X7VbdS;L^E48A`P;~K^fW!aam*8UAKKx>DG#O9l=cm*2t;?5<
zHM-EOY1Hpe&x6s3c}WCo$7I<bUA}hX(v_Qw**Q`MLc;8Qe^g4Z9c5bUwOv(ft(1ZU
zi6b*}5+tuuior`_t*GfdB5iK(-hS%Yv;D_s`;QJzk2iO>w{qkBiUXyTIK-x1X>0Xh
zW3qdF_vTaCXgeEi*djxv7|>=aMw2)-U0kf3bL=5w-IFYs5-f1?OI;R0XAW&umIlg9
zlhssJrKHfce=fw}y_dj>6Ei#KR;!gT?|uI8<oImW#NMd)-GA^Wum9F>^{YPQc`Kqs
zP^&fOMnFOczwAa=fCNa00zmR5ZxPmJQ(g-=wC*4QOr$^w3IHN?U2I$LDBHb$^U~$3
zz!IVe5HT`vN{J<{$1h-lUkNosVos^++;TBHJw4mse?L4qIeB>h^DN79Ym^wHuquk-
zWWwYx?OcBJ@PV@E1Jre=ly=Tly<Y25;<P@83^z9R_V(B%MBLn*IM*tr>Q!A7*_tov
z2%o%lsqPlZHKynndDbi%tFzoH*R2xAvb5d>!yFl6LTNn_<#`!GwAxymEtks_BS4YX
z6$i>78Y%#akp<Lh)wIqPMbB7^zpOeVTGyVfA;xQ$9}EE%e^bs`mT9dAgMrO+jmdW{
ziztA)ZbB3X+$kZYjYeaPQAR24e5X*9QYmr0YS36j4Z;GVl<K;UMN&$nP<~;NmUY?$
zXw|fPhexBTtQ$w#Qe+@vL1nCWp>>V`tRd&x&CSX8|N8&<2mko*-@AA3!GlMyzy8`=
zZ@uy0;ojEPfA0DD#n->_jTpmlJi2-F`s2OFd0E=5x_#$v>*D$O^sAZ-dERR}$`Hn7
zwR`2p!SrObsZ62gr_)iNo5kGrvgzVtFwCF2ylYIVV>?UX&i2L=8^!)Kc_)tDGf&?+
zI{ZQtvBWa(t?JHssZ9R#Xq%7+cc1#n$M>$@xb?+se|D|s)muLcj~~U<8dX4sCX*~v
zu;fu`R!tp*@^Lw<7YR5u-Mb$iDC72y=hWpJo4sYT%8N3@6at-`cba0OvQud(`VBiM
zA{$6Utv92cSc}9%4zHoPtF0yg?{!4fN)QnhCz2$E%{a2+7)Y|AKA!+3LE660Z@v1`
z@i^^{wx%cb;eYAH__ez?{_B5wv)F$5-h2Pc|M1`M9e?Pw30(^ynN?SIcg~KFr;COG
z(AX@qfL!KwwOU2*dPUanm#4>5{B_YQkx*G1VzhZS+!!8S9NSD){Sp*A-<+SXAmM1Z
zJ#PY{r07$UCV8^u&8P1k9PD*|nPsJ<brWh>CtK7nAb$XBF?wxyS$7P2@3r%gL{d^j
zDB?rV);w|NSxo7Jw}1NC-rn|=>;1uKHeVK|NGS#9o2KdYddg<~$wU`b*6Z7{$0!Jj
z0f4k%WPrtTHC@cRuJ$nq8CwprOvUJQZdsca*qUClN>K@#1to|!N-L|9bjnbaPDEme
z0@yT7k$-1r(~I-vyxbgq?+^d@<=^;LZ?vJcQAID2W)T!nTE!5B69KOMGqT>)vp`}*
z1d?B3CPf6`%YlOkDMca@VF1xuA)+#Ja(<RCF269`puB2VHGm)}PO_Gi1tI_-3bXsw
zRf0;153cLlhY!EFckkoJj~^rAVlf+!dfs;_xPM~UFG~|+`1sQgA3xaZ4SL@D)3bTK
z!Vp517v6j3gzM0OG0C~@?d|v8dN+i0<I1(^?ChhD_La7P>geS7*{AO~#$r7FtDnBx
zUv6(z{iD+p0#S-2rK&8rT`+qVue4d#PT4Hi2tb5{Ye|a*Q58k6Z9778(W_i~Rl$jw
zC4Yg~+{BP_tGBl{Q%X(UcCMQ*>owqQ4K4-%rIgYbLU18<&MlTR6wV6+z*^-2aXsGw
zfQV+Zna#4cZHK+QC<?8$Xr*;@-ie?90ssprtvlyK;;QPA0%LSaG0!VhR!}1atyN0N
zTFau^rdd}f1G3fz2{Hf?TGfrKvV;0;vVVQ~czOYfQc4nn^G+dXtzFj|Q}p|TfB#qi
z=YRfB{?QM9^uyb?Z-4l~z2lSf>8ze?Y$i_M{?510&(G)c=|>-ZaAk8+X8F^1pMS8w
zf9dM2*2US<e^sNQZrU#n4xWE*+k^TKKY9Pw?VXF0{X2JV^{x&V`zJSFed_yfy?+Pe
zic?mf&XQ%uI3~G0(7WGy=H$F-JJ{>m<L2BzQpQF}%T?PpGM&|y^n5Xmdh_VwJS%n{
zpBxlo@FJ+kpQqI_DiErTvRm7|I)tUNpVo^dYL!xQ?fXr$<>il_xx6@8+}s*Bi|$JQ
z@^Zc$KwquIMP`LnmNBD*s1;RMntueD(7JZirL=0>BF_kkAsK*#m_XGS1=%wQVjDX~
zPJvZOii1VDij%+p`t!qqZck2Uov)L*`pk`kUUt=HM8mD;zH#HmQ`LL-meTfe%F6z#
zULBsCXq}4+q7Vq`wzkGxbai2j&1gItteREc|H{&Be~X2LS`cFl+dEs!dVe9}XXncV
zJ}+$YlAB(Z6-8l)2w1%L!ZN7JC!cxt(~sXj{_G<klaenR1lQ;6brbb^$}6Rqlb{k(
zN|W=UYun{w(YEbiveUKE1z$Ivwb`vZ&*awL|M;D=^Tp0)s|Plq3n2vO*GFo^b;G-8
zqnvjx2$Hha)=hUYKR-RWIDa}lYU{dQ&6Lt>b-Z(3=Q=<g8Ol75m~{dMl_e;8R!NF3
zC{vIKr|3eAF(LvH)pdPvba=Ly^*1;E;CKJvJHPk)l8}L{DF*#fZY#|$`M~SuKoAl5
zf4ts^6w;U7Bk(1Z_?Mp|f>I=+)-Y=Pk`JUJ%EfHn*2^lhM$t4xvwx0Zt*xN&WnnJ>
zq!fR(>&95_T7PtO@ad<YzVps|zH4tiaWl)ZvMgP*?B)8_t?PNFFJ@=I{oU_Axc}(R
zGf)1zzxeZ4Uj0Ulm{}WRVi8i57(xgV{O0y<RaMunU3>Is@A%~K?CflNYv=yGPhWWP
zIq&^&G<ffW4_5*2K7Vt&L*3M~rk)p8r2w0w!PVVyGdtaX^e_dNAQGaBaaMP|UXPR!
z#5~Krk4hU1u~+p|j6`}>H^crQGdjl9`T$7Vn;S(@a10>Qb=|72*C2T3T-S8~uy!?t
zE&+0itTky(S{Z@}!0|uo<@j>STf07fwrwXVSZgDK>pTLoK!4}L=H{?-T?j#Gg+#%-
z$Prk&&P!xvlgUb95n*Gjh`25oYl$eOKtzdC4ADhJs552HF6KMiTl2bEAN>$9rkELp
z!t1>c5o>M!;1B*TDSGnBC!4zclRx=S|K*?l*<^DAK$gqJd++{yGMVf>eDvZAFZ^GB
z@oz_?(d|3WJ%9J?i{Jn2ANTshC$2y7RSkw-zt^vN)6=Cf@!7k(%Vsv*9xYbW!Kj!F
z)u*3*SX9}=`?Ic_KY4q1_r_*b*&#c(SYqAuy{njNNiz*)X)YFXn-#X`pPZi&5hx0P
zfAF2({Z~Kyo3r_2T4YwJ$W%|^crctUrmbMn`mk(wCx2CK=<Ntcjt|>qzc(1%xXCA{
zpY8d@vD-g5s*0*PJsS!R%7UUFYo!^kmjlB9NSOld4QfPi(csiAAuHR+D@qq_2cmq8
z4Ra$B0Z=0`C4$5$!)P7n8Yh{_M%ip)m*V!$-NR4!h023fV}Jh9l}kI1&(&}L&Ogx^
zrN~iYzkjGA(dqe_C?u^`?ScT2h*7f4#t;egq+cn8!Fdz~5b@Vl4xUncy#Jsm%C=kO
zRoSnq>Fi>0ac(kY2q|Qnqpg0nnQV=sKo~-ZzJoj)3^uObxcSBB_bzH*8c-7RJR>5J
z*!eE`qyQ0AVgyu%P%|J=X7$WD-#7!L5F|S0E`M}wa0x()jmfPSo;&*d;pxTHur!c}
zBmpvNGF1{pl^4ZCC=V=c=YXY!x?a?0XQyZ9XKlS$ts3W&HEXFP#Tb1Mq$HqKW;@7H
z7@0f9X(PEcxi!5at6*?)xEH(&ArP?lO`B-%<lKSzjc<SNx#wSTofM|$9a3dR8(U~A
z_J0Ox1rit)C=vlCl>oKUZr!;}pb3A$PXncp6p%_nEhy%cNJA1}LPDcSiwHxCB6+XU
zR<~`-A+jbE6rmWqHT^iHL_|pPD}Nt)RnL!(4nFwsy&wJPM=?f8@xu>4EQ*{%&{`#N
zBB5)W&CSVAfAZ6BfAhBv4j&hJfrz(m-GAy_S5{?A>_Q{JglKHGoL&IH(b3Ux*#F|e
z=Vej$tNtsme66YH#a6yrF0WqNIa_q)&er?)4=@Xb#ms^vwLR|T#FVKkyW`{2^DgRT
z-Q;;6%F;U*z0dPPly>Yu*3D#G7DaHrs0;vv5K@w=-&aJ;DMg=W`PxmtsvDDKF@MI@
zYNfSaUxvgOfq)c7P7LBYm-mXqQ8+}@Ym5{k)^(ldd7kHqSrJ7adu18Im$g<bsFX&6
z#cH{hAeoti7opZQWm$@V)|QkJ0VG=Ml_@3XoG}JiyzgwLm9@@$5tRT?YZkef&)Wc!
zTB{W)6)^#bcV1hi5fg`(Uw-{(KY#nl-~FTC8IOmq3;*n&{j=M*pT2hW#&A&m`Jex*
zL9bk#pKT6D7S-?l&hNkV^S7_xx*f&5^x{j8_x6;0WernJv)bE#RAv0lZ@per=4^2+
zsmt@MsU@}k;TH!_+<DR#Tn$Q-v$DMR@X_3dt>LCDR!}>KnKSzQVkTOT#(({$^G31J
z>^hDhiy(jcum56cm19iHr2!~Tx`Ng1o!tj1-Cxcik`gclkpb34%<JVMEG{m3GPn`5
ztaDdHZ;VHePEVbvd5VQ+NUR-G<WW^gVhdn(vRJbBWNGeFwXt>o@KlRhrmhiCl29Ni
zSV2q*3sISL!zypiLuHCVMSqpaVxK<S()+4j&YQ`l+iwJ%ExKG6lc;&Nh=2XVy^kNe
z*q+UfcdlIBJ3C#?7aQYoR%Of8Dup`5WV9xuR`bRO84ZVIw1`Jh9Ar4G;Mb+m3qVQ|
zQaZbs?dHXJvi0E6gBZCe3JIdf7_!Z>TN^z@1OZZv&UvlOou}_U{(tPV`SE@)&kSW+
zTR`;Q39NN{b}>c)g_=l&P$^{)3#FnYGFkwVFe0wKg)s#|(M7Ry{VFu|$zt9!WQ|fv
zDV>ATRev}fZ<PJMwi!YK=Gn!?#o5^xUp$`8m$SvZUNzAt0o2y6ZR|0|5Tn*w7~A&Z
z{N!R#X8nFvC>T`P#(%iquZnV55qmQ$UEtO`5bk1Hv_Vi`x_oQ*@(t&KRx#NYwTh!b
zzdzL0`nENM8Wges3j#88VooV(#exKgfS^b*LrNebM5v9{$;Rjj1rSpbBGg)IZAC;n
zJ?xLZ{>tl<(bgA-M<B9B#fe0%wE+K8h0ZejE1#@Bzc_yHy?=M!dFLHY;JVPOvd8;-
zx2|7x^%8)qvbcHu+H8K-wes9E&l+Q{T)BcuHa9m956`Qr@*%{;A`)VN=o9dGINaXe
z?)7?$#iA&()oOLPf0$=EGt6dF=F}^DWGd6|zy7t?-gxi*s><>VN5jeS{-f(JJ=3rr
zR+GtiaQ}-l7k^d9>Y{11)@|3dE@)$7NHM0osI;Q4^NZ=UcWEajF7kZP?|V^-P-2SG
zs?2JoImsHy)V3`%TWi<5`#jHG^y?Nyjil9@W$(0xx^Aq^*40#1RW(fm05;3ou4|U_
z%;<i<KX2*~U6z%C;9}I)c3qcc83@D>MI<jQ04SvmWq&?N3_isuAZykQB64DF)cUrH
znSJn}OiD>mGsCL!Ixn@LLQW}It-S9FqqHQ_(5uQ<UVZ8Q-X}!#;~)R{#`PzvswWA2
z>%G;R<MG#CdHLSQAG^jG)YYJj0qRv#4z?aY`l79u7~NMk8&dq*vo|mAsIpiU#%_&t
z4AzJ}Uw_o3@>Ut8@YSbx%idTtpLboqzm=V}^0~z0=$J1|cRbFs{wTyGkaDZLX0<)+
z`ECVDA1{~3-ZiajM4Gl~m-E%C&7XQX1G71qJgjF;V%zEZah;Fo_Kkk(RQBNFZ0Xur
zn`ikGs=6|!dRh<7a3)dp>~ytsEP+(R(E*>tMt=(^j{+SFw%&BW+?-U(sK81q9VHXA
zk_=;xG#(9e&2By|sJV8%pDn8rxt*B(OZIaLdbouW?(gpd8}FDvuROS=>$=&)qqn1^
zIwXyGQJTE0R&~2<YXJ-{<(cWcTVony3ahq`k#{Cz6ux!s^8VptYs^>ny#8MoB38W~
z8GpU1>%IN`%a?YC!|}<<ah~@?#J9mH+xRp;U-q+&Uj{ov2#U;jyuH19t(nb~)>T=q
z%OONS8<MmPodYCoEYW(biGV1hGlf{@f(b}$X1oMMVPWQ&gqaj(mJowT=R0eSCL5!8
zfpWM}_Q#}6N-3l+xYhB|-lv~@djJ0YrhoNGgp&{%5wUq+D-8%lq?HCBP6BH<O6$)r
z=E(lqrH%1;%L`pB+QJgq5^ce-OcGaZH(zyNt9&?OvMh#FCWxJP&bzALw|U+z=Ey85
za)`nri0kSoW(G(iQG^f`DI(HqL~u$WcrOBkiU^q3xho>AM3aa_(RZGC_Nk}u5`P=^
zvi6Z{4Qs41CrK%-bzlSwzY1inA3l0?c5#8Io3`s!z0J+x;p2xO(Ydb3px^ITz23#_
zbiP>KfB0xT+MmtNd&A!CC!XhR?R@~s08?@ig*7Rob>ML2#`Wp+Y-?k?tLy3PsLZPv
zVp&v!!C*X|T)%$nZ+`mb=Lhe9aeublynM5bDR@UYO~wPGX*j5|%vOWVY&8ArAHO|4
zJf;31vIGVs;J~BNaJgJ`T|MXz>(z2H8Ch$HFoi%Yb=y=`sjW3eCBUX>K^Y*%)+A;C
z03ZNKL_t(QBp-qxprRQ4S|v#w%d+fJAVh1dLNR8LqEn=m(#Bd+s%_d{uYa$VYT72U
zY;Nw@OgZn%Ec0D(UES~Zz2^{}TKmaDj10^mZAj4}Dl|GJ@F{W<AWV!gk=DeVVu~XC
zWyLuut1t=&??^*LX74f$t!pOz0SCvN^30~hNZ5tug%|Hu!#pMV`s=^>{qO&9I2gb9
z(hCnhf4H~z`D?FzeKtM+)_=FZlOn(M#+%>#_P3w7^W;Y#f3$OH?wi$Q*mw0;HW~=A
zG<r}LPu#qgbn>ZrbZ|Z_tjY}70;|E)b<s6vr_DFM^&3YYzISObFn)Ta$%|s+i?fRm
zk}5f1U_wF)p0zeEz;IGtT-1#C&Vzg7GFy5tWB`LMY~qxZRG_aE{eO>_vyO$4M$`1P
zZsJDv{P&*!-~Q^|S(C;l-q6*C>y*tCP|UJ6dM+yu<Ph4bOoRt1rctR`4aagK{Ki@5
zP1OaSr!>s*lGft@r9hvdV6Z4yyABsGzIMH<n_iK9{?SD{*}XRk_hof-+Aa3xM(ft`
zYUPnK(b@-2NtO^Ih<{s!MU0b5wQcBJr?ldbgi)zXDM*gd3j!*Q{eI7YRONVMSe~99
z^!n8=Y}dbzrE`TSNxW;^#Y_zbgUMvGSN0DNj%vTC=c~KVyw+Wo#jFb<Lh)tl6;V0Y
zPFHh5(3$<k<}bC*t<TWPm>khL#|ezFi9MlcOC)RyD}-I_8h_u$=vg9j1P}y9Aq0b%
zRF-+yupprdk(rDsd%CESplcc?X`A+7fB)gb2g}(aDuX5?O^BKk2Ok3@W>!jRYnWLP
zA)r-yI2^L1)#CiKhhJQ|dL{3Tl{SgM_Qw5<t=asd>*{6Ot~y^9LsrNnLZXez^IT<=
z02rmHs#A{9JAW0t5(NnBvW>NXE<()62}BS<lfn=gKw|Xq7ui-s0$>7BT4z-mS*DA5
zxw-utzxBQAPdxqb;ApYxP_GXO0GO=FopWv5uDv~ftH(N@FMZ_1Se89u-rw6Bje9=S
z2+8^Q<Wo<bPv?ybg0fgOfB1L*=*>6Zs9pQsyB}A*tA8ORn~O-`=oB+)U{p~9Ct*of
zcD5dVvDbx26o&mly_)SG9`Ed4dwg_ubb3~=>L;#Tl{`-&6?tJ0Y;Hv2-DjVz+i<*$
zRF1bUU3sv7u$awuwl7Ea(~DVQbe?JCFzA&cK~W+~&aDRJh@Hzwky4D{QbdB-wgJG$
zL|RX$Q-21rx$UA8V6BlvGNZ&bT9VPmTGh23i%JPK_#)2|r^w<{(jr=6X0>a(&Uc)Y
zHCl@@BqmeV2m^owZ8A=wX}SbdJBKM+W306-(sm9M=6L~Hr#7-gW;RO2=y(nFLx?Q7
zcYSMF!!kI<JTM9rp`z%$K^~Ras&z~X7_!QK^nb}mpM3I3=l$vV`K8^fr1Zc1vp>D_
z)H79oaCmZl>*h6Elt298517(V-g@&p-~H~p@4ma7ohRS0H^F^n(oIVI_B$U>&#Qia
zFdA2;?JiY$f6#mQ_$YJ&(Y<{4n?L{H?N02!`O6;=gD>UxRKFmnAdPb~6H-lTA<eoD
zwSUs0Gpn4$$wog#AAAxBTIV}e<%BLSny5LFBhxRe-5qY7cB@0*gk)5b?Gx8sERIjk
zo*WhC@@8>xoNF!<lCUV$TCWdGh?D@mBv5pV%B5!#eQ<d6%6IO*5f4vREo+nad&l0N
zHq8d}ZCe(kMr|mSE}>`=41IBY|LDN#vwuPVBW&uJ$e1Q#V6$3v3am%U3|ops*g*=e
zF=!lzEGuk5NrKAE=4jH-%%i=%PLW8gvZ_r^D|>0OBhmG8qctBKpL@qe#oGLu5D0*h
zxC<$`RQnL8bJnE$z42tcb#^k7IBV+8n(S-8`FgvGHqY1AhV>QU^y1t*XEOsJU4Pg7
z!V=1eAfQkg6w;OefEh#-Dn_N0(Z(9*TWVH5dLLa(o>LM?h^R#|5hw+u#aJJsVc$70
zS*EkXdC%k-Iry+#H2Zr;`(GU9HlsXarF{|v2!X*l5Qs5~tZfsZND*RTvy@_96(+MW
z;0K?4)Ej+$GMSXQ?i$R?0a(+VyMNBRF2syDVhkJ-2P9FMQAR~hETFXsF(QB=wN}ex
z3WQ(IUrtI$i)au7pjZ+SdLL5?fFMk3wxtL$$6V{-VDMmnpB26O8^7`V>#v($KjEp1
zoGUQKuDwic+pXXIYbiK{zlG~&I-Rv`x0p}crWOvX#l`OKsH+z#hS8`h%YWk8XYc&@
zCl{v|r$tfx^FRF;*REW$nf|LE{QKYh%|En8yS7nUL=uZIC{5&D%(Z>;$*0av&y~X4
zw{9725BCrH{XUVr|G@{gD0g<YI@gKF$;nAI7+t@9EzD1DU%xUcsVefuB_HGY#Wb+I
z`uf-3{`lVX;=ErCNBv&%K7YmtfJ*7aVy*2}WodPmnb!G`LRsX>s5RZwb)ju)Nno@k
zi8iyOF~$TWn#gzUaBFL^SQJGOLMSrZELX^^hz9*CrYHg+tO;mM7fyi?w4xAvkr__O
zXswiTJ}PCScX^iAb<=e<qn4Bg!-0sb>wt^G2oaEk=rn=VqO@|I6Mq2_j8PJD0zs{8
zmL(D$90^!bF?mvoh*Iz($$M|hB0+rW$*1d8EyCV)&piFqXf%pRR?EfZ-JQ=q`)o2A
zbWJ;z^!jVB-T&;fMcw}Jhd+4vrI#K(czET~B~8jT%~v%Utfp^${neY-cmMGB{&43~
z^~1kZ$BQpcFQ$Y3=zqOW4kW3)$DhwvD^jD$<qc^=JX+<pyuJIxqU|u7?N84OYq$$W
zXF!x<Y`a!x`NipKXJh01>|ASObJXOj-r%Jdzs7ryVpP8C3ZRV4t;x>n;P66qb&3-7
zWw2X~wHzNkemGQm6&e#+|ALQU*P=mUQm9!JqDrb#3Q=jZ@qf<8hlx$mGm+I+wbi%f
z-s!`Im7gva`$SS?&oQas23CLe;X^gp*q3l0<;3dHMvGu=|M={DXLn<Bqx|&#0ijws
zZ?zF2ME}&?C*J<~2Sjjk-dL?!)A8(L3K#?AxoIL3sj95{MKzsH?>v2DelcAvFBFHu
zf_E|gx_0!;5`O`=cDF{G)$!5UVzD?E-keNG5~Y*^c-KYtXN&XFj7XCvBWdj<=~l~`
zYa6SyFmg&F0;B~sBatAYXsxxffY^D53Yb$&z(^qkw_1@hU01u*`RJ9Rwa*+$KmoIm
z1_F!>PDw;0QdwSfp)*ARAg*05>gmbResHbP8Tin~ZhtLd#~396Bvz!>G;xx|NB|OA
zwQ6R)elN4c$<3FGx@nWiqE$);m{l&pD)`z5CWtI4Nr<Tn&c)7BVX{Koij{C85+qoA
zY62%ABvk7dln5&j1)@^=(tMPnib(}AUYuYOjgkOBl-7vSHFXSp_2x6rJpW=f*$!xJ
zQCVw)>wh|m)@CA<*sl*L>%#HBO{K7&O_#Hac~duC*AS$Bza(IrnE<IMvitWx)jBPf
z^Rx4_o!#AytxcWl!=uA2%akIKlv`6biv&>_V~k}Wo7u(1^wn3s{)4~x%OW!;$LEXr
za%*dAb7MT4or_>r<fqe%&b!&+VQ=e3;`Ge3&wm_0#B0|sUE3Ty`uww>e{k=jqs{Ag
z?rvXNE@s<Tum9VB^Y0*rUT#;>DWc9NAEF2Z*KshYi|Bkxu_(&A?Y^uMVhJhcMP8C}
zUAr+Vl{WL$A~(bdNmAAC1?SeFI0OjZrR2+^4<K!`3NcoLAs{gm03#$q&WzH8YX*ZN
zG=JIvK<9&ivL@@gZa$xCqat(FD<UWFeb;sZkmord7-JO@6K7V6BKN+7D9jL}1V};x
z1Vp&bJt<5YAjW8{(uTS&8ltkuZEjJ+fA)|6(aFj2D=)tkV;GdxPyXg_UVZh|!@WHe
zdH(M6A^7RV`C;2MtNO-`>v_?8@ZiDm{(t`VWPI=5y`7z%|JYXa|8Z5leDC20|J8r;
z51)DMkD$n3m-Oyi{{ppc9<Q|Nom`wwcDC+4JX$Ol<V?n@YIv*6XQ%t2_RPFY>G8ZX
z{oZW8M5OVkbgqfP6_%<@#j<Exmr|N+4qR}beD=;xf^(<@Gqi<*HmEqhYA#M{0Dl2e
zZ4zAZWu{#m0i;D2oUmf?ZEIKzArTo+T8S2wA&O5-Va=VJw%wwF)YGEExj&itEM_{W
z0OQ`7HgCj0^Z2rmRT%P(_oRK#&_#4TtIAQ&cf4vEt>MPi?Z=NE+}a*SRF6;2N#oAA
zA6+;*{p`lAF`#ZbI5?V;ura-UHGhGu=e5tca8Y`O#Wh#1Y@Hr|F*bV8&mzFA?V3RG
z*9E&lq!fu0D$2^TbS|{*>frICX1-X=&N(_`)M7FH(Zl~Rs&>Ei&F@l%2|y_od^bHk
z?V44uDhb31lh#0#*&;Fvf;LJ-2#F9&sTRwbb1eX9We~`@0BsjSh~flNh<~C0wN{W2
z5D5%&WNjoyZF9v?_WEl-zFy-bo9^O#ae8{uwjQZnt(L8KF~-CkLqq`tBtarjs0aiJ
ziHMV+qUGv5&kQGDX4&JT<EQRED<)T3^)|LuIh?Qj`D{Mg+Ch?(k`G;oE(G6oT~!UU
zEYF5m_9|;Fd565t>#(rI7?<=b0TzE}rbrn<0Bz@ux8>;eGj}iTUS$DIYBU=5dgb)2
zi7|c&i9<$Eh>)Tn;@_rzO#6;!7c=L&6rIwTXSVAa;aC(|Uf98)w_MKq{oadPFCLwo
z0Kn$f<{NLmNvb#K_k~$VAyEP#V+EBWLu5vyjg5RT8IL~y;@)C*?tE;!W>tS1B(iyS
zKAjz(oMc5wl;67j#KF<oZ+-9Y{QR$RJhb~qUwr!6{ScF~*_&^@@xsfmXsw>S{p8;L
z2ag}^h0qm76N1e$n-!t!G-8oyV6VwACu=LKlOl>CB~DG#C`AMb#TUkOzA21qx|U-y
z+GM$2HeFm-m!#+&6LY4O%`|^7#U!K%5n@0^P~_HVYQtJkkI}oVDiMU4QAC7Oii`Pl
zb8~Y2`puue`|<SRjLEF?u88P@U&jcQR`h@K_9iiwX4jqHIp<E_@aC8!Gb3`WtgLz9
zP-KZkih~AHrmRN0VL%HT?uHxi$_9F&je*{H;l1IN)mpgGGC;SamLY$s)tX3(s)?*(
zu?AMwkW*w#8Pl7-@lMB!mz0HGsA!<T6F_Vt@Vytf@y@;Xod5a#36+Ex+HRhWCv%Q0
zr1l(u5P$(VLg480G)+k<f~GD3loUn?cD@bXN<np1q^Z7iar^GwyOQX_#^$|`Z~gi=
zzVXqWTjhLO%;#B>?A(98)9rSbhf61u@u=V1TwT5W(Z^5Tc<hx|UXedXoBrpFJlA<g
z|GR(xpPv7<|Aq}F{iXMB?w%eWpG<s+5e2>T?yX)wJF`=)$@_41a}aE~wsPT2O^Abu
zRR(x$t8Si6r_;7wSZD^o_INyL%K7qgo_9oJOO({@c9&8eTh)J#gY;Gyb>j9O|BWYZ
z|KaP!gq3rVB|zjDSpsvc0IU=g7)S>6ZHNSj=w%>Hpgo6RBeNh3AXCyx#nw9}i3!f;
zXNpC-Y_v+o$1@vA+47QE0oS|=)qYQ7FJDtc(pFVHE5fX-N#fDbVT$q62HoA=S?P7X
zi?%%5-d>$m6)Jz2&&xbXN6Wg~>DXCa7n2CAJ0c|xvAMBzd3SejJUJqAKGr%H)3TlU
zD3sLdA483QilL<L<jcoL$8|GNT4iYtDxFVf6nJfQ)ZJbpP$wtHdpnaGSNj6VTZf>e
zpviQ4a(I|Zt&oFlNQk)b5@e7@FOHmx&w`c;JTMlFA_9LRf>c@vMMP5Q5WOJOT1gU;
z2m`p#u(vWzPYUZfTE|J+NwW_Aj3r>Lwbq`ToSvPX7PDF1mX&o?Q?=H{5Pb-ML;#CD
zMoYOkXamAxo{=#Kq-brHW!G*zE~FMZ@fOqGa(B4t!2l&n8DMg*g_@GA)3mi|(_%JH
zr0Jz8NhyB>DWkCoganL{7g%QY-V;Rt%^VSlP(?x%!3A!ecabqNA|Q|uQYR*IJekc`
zE<AGi$<K6pOKlraahj!_PG@{_(zflw(j)>e+{T&X!a5~>V(Df&p4L_Eoh1TbpP0_H
zo&p3RO`4je<)yNmG;M=IT)cRxs>>g~{>Jk1YHNR~C>lVCBpFGXOsa%fl2k%SEyBm|
zzO%gC-&kJ(=E-C(2<Ju7>vh*QHusN?pM3hcG|iI1>eVMccXxOH+VyKgBMxu9_tp<z
zQ)%8;S+H>Z`ZZ*ywPJ7ASBDhZ!~F+k)gnL$k)&|JB}TPHArl?E1?IMGj8sA?1~fWp
zbz^^RMGW0eni)~oC6O>nHX&-I7w(bF;GOSxyYtx;AtD6H!g_B^T3hcUODQFxAW=3&
z-3G07C(V?kx+*()7C8WWLh!!&%=1sItu24{b6@<?kAGT}KJ6+*0t6uNJ_y1XqX@t;
z3ZbN=`Fth`K%xt6fMAq>7@~95w;}kZZ3usyNg?{c5%}S55e34a-wEDtZCyx=S{V%=
z9PG7KIq2j{;Og?|=wQF!?~TvS2ykVz^4cq}MdI0XGCw;Tbh{WF2e-ERsaXl45F1z5
zhr{;Yf9HQ)84m8>|7)Y%*4Fyn-J8+j`STkG`?JmMbtUcixDgTV&mNpyzVgw@p__j+
zWgAX77Lt@yLV?v$2pJfa0t7RjS1Y5zh3!F};dnOnA!;r1Tvt9=2pQMMFn;vQPyX4R
zDP7*Ug67WK4}>G|5D-If1hED5kqk%+p~j#w3V|F!0r|jzA+qocGD0h5%@IX3hHO!^
zbz{=p1}6}+S!H}MgATU#IJQ2xGxmRVRL&Kd(et?-kIR5bx6|e5*!oVA_A>L?r!Gxu
z+&wt!_4{cTli{$bXQ{$|KUuA{Yw=+0geLF3@`%!|bkgNsc9cZreZJs%C~8B~&vL1P
zlD{z44Gwy1TI<%OF?vZd)k!g#ix@MLt*$PY^Y&oxtZ6Kf$kWU=kr{&XJNJL?P0vmT
zS;nqv$^xYlLI|m)Hi86`0D#EBIR}I!Wtw$>ux%Ufd?K_~#+Xc{!b6uxEco8Qh#`9L
z=q)z&Sf1J1dXC+0ceJ#U=ebg9!Cw(V0C?w{x}A<^M>fPo*ViusA|WZMj8ZL<=fE6^
z2mzh908tC+T3;2jYEncO6&inKl1z86J^swcAAW$y{oax%4gdgL6w_X>*RkDTosS^`
z<h`!e2AWz3ArL(WU|8HXhzKb%1R`SQ5EzkM<I8zDofkxm2vQ0J&_<z@b=eG-)<5&b
zmo~Q_F<LoB=Ymmk^W4Vq;bGG>3qlS(RB1mXgN6`3nO=EvbYfc@y_bI!g#hPjfT*N6
zIel>B(aTHA!z`0;y!pm(X=z?oE6XeA*3X}gPm5V|@#1A6lxNaOj>wX<0LjQ4;dFfb
z`~TO!{pMF+I(KgE<mBYhM<1EZ%9FD(BaSCilXW-GU)b8(GTr5=7Z<j#FbLA=-M!<0
zy7efut|sFk+~YT%9}a&8fBoH86D>x)j&B<8Pof}p;G7R3_WF57p<`50%#kug3eG*e
zoEc$~B)CdS>UWc>D3~LlDC>GO94Vpes<z%)$7!!qRuut`di`#agy6jM3|>m*T;shb
zB{@b8alw$5gx<M4%LqUU>h;sKoBryreA~McfP4M)+rRO(-}`^PKZ+3}2?mfr3?7jr
zkrWCcR&~vc-8^foD;sNDCzTKY6d^cgT?+zz3~02owGsjt5Xo8J?;8S_B{Ip(&i?MO
z=pOFvC&rvxTa!}0_2!$;eD;}xy<MYqO#G4e!TT3J_j#G9AHMd=XP$j-KAD!&`MJ%F
zPAC7g62rWlA0B`0Eq8HcB|QDYqlbU;r(b>XGvEE2ANG6QxwB1O0gDHxb;o33B!uOQ
z%Uf?$<NJ9!IhhLQ?}^ZAEs2!k?XAs&<I;v8DUK&olgU6_*IprF3|U964s$sVK)F!j
zsG4kkWeEqSyj7XhL2XGw>Odkj7+oYp3>cZ2f$Qi+CR%@Q(^REt_w>P}i>YUb0VBrX
z-AwAAwsunBjlRCRIePnW{y?O45$dU5X3Sj7CdH%CP+c)nk{fn$)X$CJ>70E4lu-+p
z_vD4u!SwXO?YpN>KXIj>w$l!~=nhW`Y2>ieIlr<x85iERuHf}<ziM2oft2wamOH(s
zDYVAbVRnD9)1NiviJP6@UY<`&Px=?-4GtW6KCTAK!#-t2F%B^}4?aXhD9Ymg-hB)h
zU39G@l7TUzK*s6p?D*(_q9?X7IFt~*lSG100%7FZwGbIYj4_}wQb=PoLR3<mO(sp<
zXl?Se<C#66l+q}@!1f9hk$tcg>a07QO~xevZ=8SI9xW{gtr?)Tt!<m&U7n|maXc9#
zL1474Yuh%27(!qIB_%O?iY`VH#K6IG2+?^j5QGeE)wU>`y0NV#AP^{o-n_7N<?$!m
z=zIN~5t_E~-aG5c`8-d1Y2LvYA^0rs#2AAMKQln$pTUO#0Ah?GdH{$3HaH}X?1S^R
zuC#xaj0}uQ3(xW7bfS{Zi(mZZr=R=0*2cvU7^1aCnqGI1ce<yir$S1tl=psdm0?~e
zIDd{By|>=BmWY^LmS<HpC&Cc@rHk9$ZvMm9UrST7zOjCEd@@>IX{u&@WqtSlA)>f;
z?M7e`8IYtDf|(eE55cwNWHLEEJ^=Qs%gcZ3YpWM8UH#toeqfC0bh`7R$c8BcN@aAC
ziw1<w7{ki?<_ll>+R1~x{oT9$L2q?!rPCW6AMV?7nq^5dpBXJvC03S3w|5Wo;izfa
zBuzq$Nh<rjd^{efS+8;47~^Bqi2?v?t=5{r)pc1njnJxT9WW5d);b@UnL6E+NVI=V
z%dW|XJ%Fer4f><nI_o?~N|Fp?EQ-1)YtmXEFhpaN06^rXsxLo!<;!1xaWX!1ZvNb}
z&+pv7ySce`?fSW!w~i4X-l&2L5LuC6!4QJC!L~D%B|<17MhYCGj3f~W84-k5wrN@t
z27-_>gy>_4%t)w&NR3X@ggF2}x7&X~fU2qzV^X8v`N`W&(<1YHKKuGtzxw|BAH4V8
zd*`>#b@R^syZ55?N2e$E5B5f@D_{DQQUeLN_{dUJ@kej{&HEp{w!GH;H^2WUiSAKC
zid0>o(!$nrH5m4F=Bwh;6BobvrGNR(PkvNR>x-Ap|Jy(ROZV=*nk!qfU|)Z?7LaAP
zlXNgoc-T#Mj!!$OPNna4*_C!Mlbf}(HD^hF>;Ku=H^=|OfA`WK{JWRyCaw+oMN><J
zE`nFGjqEY9cTpk%o7DIE`FPqk(zY@r49Hes1g190-iLa;ngtHd^Q4G}Q(u}g5h!ev
zu(=rY1*WX#;N`SSJClM}QmubbD5<H*GGT*Flp6N>yOX0B;`(Ui&fz3WIZ4t<*(@)Q
zj*pK^QXkK1qdI4kVw7aX_#_)Fv4dV$n_T7!f&A5t^|R9lcTW!1HkQ=Xp>NCid}U)@
z%b!a^_i3!A=yir{ecBzi<YFvOr>EeY&^iEEXLH@@=Voq-NaPrVRCRw<9UUD`$0t(6
z=v%-L1vuu&L2C(un#Q%!%LppQ7_5{+Ng1_D6XSi`wALeOlO;*Ufh~{_5A8S*QKAgN
z1`^hlS%0NFDh<)n>INATK`lUd-h~*r+v%;YZNB^7dx}s9jVPql06v6>kr9{#Fb9SO
zdzP8m2Ok+F34uhUgi3$FXdnTEJ_Ht|RJwld(*EHIf!A8*S&Zz<avr^N&Ut5pYn_!z
zk}93Wa?TvRk3<p{ZCH+oyzt5O7?4qbR1#SrV-y4kL7*TZScuxXAoL4g`0|s_y)YcE
zINy3801ynV^(M(Xon8cJt!3cFj6&PCi^0zr`4c_JD8^Qai0ps8bCMK-L{Pr=8!M}K
zcJ8lT+^(V@EG?-dowe=8<%>VLd2_U}Ui)A&?YRX61S5igNS-60m*g`=y)3h~t*U8Y
ze|9!jCQY+^eSPiu=boLO97E)+yVTfdRHA@G@Q!h`ej!_06-k<=X>QEX{=xpv&Zyga
z?8?@BG6v%1wWWWp?Q?lTes+>_(UGl=PNJAjDrN))O{m)7kqDv{XgdjJT3`rOQIV1u
z86g@a+NN%ua{!)WQ&$VSVUs3E6oFf7t8$Lfx7Az=kFg~Yq$ZPUA%(NvRnC^JQUvJo
zfq3zko*tj=PUfdh(Azg}Ub=i~_twqx!!&_Luw#gpIZA)+iC7X6vye&&qor?Zi9o=F
z6kR~1D3nu%g|RjYsXz*ur=*2cXr%0P(uam@TFob?qhar-Z@=~QQ%@WnKR7r$9-Z5u
zBoQjpCO;WZoaei{`;q8)eD>~7Kj@8C-~0H^)Y@6oe)fwmhM(gI@N)(dd1RElb!laH
z@9x`g-rj%PE%y%VS*eewwPQ~Law9Obnrt_fO6#-n*-zek<JS9ct#mqrTsLhQYmN~M
zA0i3@(lM?nfi2EYtMkL9wXQk8wybAV?oPHVj0cqqgbl375f$d<_wEhGOXjerCt*lQ
zMwpUJn7Sytgxb(FOxCxTFKw@vWoUd9QJq`sx!8Xy5=N1f;1r%Js5My0@6PMjN`&5M
zZ6)vLX5`@-iR&!71xKX1BgZ<OBRQlth}-uM>$bkIzI1+NH0);cVxDG$#<hD79@HM@
zo?RmQ7naAf$?EC~`x-(RpB)W`>1d^M{>r(ssg=Y;`Q?81SAXj_fAcqgd!;v&NLMdz
z+S-3@Y;K&puzBUuITh=LEaMju%qyeiJn6ju&dr_M2N=wi?MJrHUs+qbVA5XQxU;j#
ztS($|^I~d412Q&MadLdfu1&Q<=Ax`-Md<)ko(Czs5RL!=LI_n=A(4+xNH)f3t(8hR
zQnxe8@*a`WCQVG1n9S(KVJ6pFCps}ENlbrt^ZfQ-E~48203ZNKL_t*2>U!QE4Tein
z=R`6D4q<VC7Tr#Ndu!Y1)cd&b^H}^qDP>F|g=R);LkJu^2hYxi7+`T%MyZ5WD;wt?
zx$#7Ixa0%GfZ$P($+JO{bU4Z^>*jeU&$`RYYrWn89@=KPvMh;2lIDR$>llSbkV1bb
zA=KiHnKU6OV`OWs=ZBVoi$7m$qm4<D%mrwp_{y(*>&35pef`{aj1<FydIn9?G)?2Z
zH^w|1y#JY+$XeSpO;Hrn=_JNa_zxlH8t-h3L2K=-Bc!^njY+aj=Y=nQ5hSgwtmIi<
z*VXdM(((BC%8f^Kntk)X_zll7ME8FXHyQ~UkRwA-B$h^_qm#3`Zl8Ph>2H7Q*Q#o+
zwc6O&c<jbwufF=qU^rmJ$#gEIRz#AR2vLZDl;_<`m#;LfZJTy)Z*OI^{DbfRT?}!w
zG%U;N=;U-V89(vZqgxxRy;S9fltk~mb-_olZL75K-Xd~pWNI*xkQ!m@%C>*47SdS@
z-etz*S&}4XI-gnR0Jv_d7^5+U+1sY}-sgF4w8~S{$r59vAYcwEQ89)XS!+#1ZClnY
z{?GsVce1Whl9SAQ>B}#zZmz%j`YT)8o1<Q~eQp^;71}Ds=wl3wF8Hdo0T>9awGcyH
zmr_XYy&xGG+86_p&_<F(5YB(OxNsPkT5BbRBoZXWaCsSs@+^DlD_{P>%YXOyQ%_&L
z_J{+1^ob`=PS4iXH%FtDM;^U#c=X_fFMeTZW%-}|SO4`pZ@>4|uYEo3^pZSBMD>sQ
z2><^IAN|Q>e0Wm6_3k}sI)P)>=@xZ7I4M1-<+Zh@wt2S`y;mwE3Cw@yXN{|f@!E}R
zAmzL&2ZLdrccfA=1SAsPo^R~$JpJs`RsGwSuRR~!KN+q6tIN;*v+eWGq<E<jx)or7
zg5XB4@P+xiZKt8ME_x30=o1Qtq9#trcO`a(T1f`gesyqj+Sb)G*M|ONEXV$w&^fzY
zP|s*D#SyuhB4QjaXX$?+O6kh8F`SH_r{u~!irSGt%i+DMcy&5)CUKR|QEy*3PeKLj
z*;cE=yq}ADcKq12t*ckIP@_zAqD*X>Y>>6S>h=4+wRxUytaN9^bocBK(#9llV0w>T
zy7;T#{^y_n_1}K{vF9(YUpRR%BQ56B;^^?mq)Cv{x=+n0<{y6#A;jj!S|`soH#WDo
zE)Kd&p`m_v?cC<&BpW_Bnd}`rXstyFjjL)``RL}wWO{lmFbZTu%(A@GAEjAWNF_)K
zA(7M~?fwHrpkrn!m6DoC=iPp{H%hafNji&(bSafmT1vI(aUp4?Go9v1-j_*=D)AAV
zTX?220BLRVyt{wAyt=-=K_r23F-f%e{_s&KfN=o?CQ?!wA(S8`Pz28;<n_lN-@LdL
z30!;#i$_EyHKEj6BS>b!3{vPc$%T+Yi1<)434rLLM}bmFAOs*ml29Pf!Z$D?pq3%V
zs;(C*%OpvZQYd5-LKE=W&;8QVpM9~nv}OY|-oc{j8smS@z-K>mp#lbg1w6Sd%h_x;
zpU<b$@h6VdNe+JTlx{a`>yo3FQc5F-OZ`9igYP{5`Dg3NxTn?j=BBq6Q?t9f-|3CM
z@}*yPUI0r(6iDnDh#3HwNs7Pv^Y8iqOUr8?|MYzuV5!&P2ujJu+JN$dkMBJG?B}z*
zqm{N10SSL$0lV{3spI404?np1<de^Cu5T<aFB_$f9z00X<o^Ag(P$-2vyVUicsLjm
z5&?Ab+*bA3>7>)|CEB!AnI);WmA9Oj#M!#7>KOg%>Uu=L$V-D!U_cU;YmlMe>*i@%
zmxYu@5-!>rS(-Fe4TOj|pU;I5Dlx8Y>#70L&f9;{uxAtuM%ib-@XXS3Z+9Pkbf5do
zbGtituU)+cHhlA+{L}yKU;i(?(dxk;{>P+~6gD^%A#&pu1z|5Gkra}okF2$pLI7h3
z0T7VcS{H1aX~_UpRmBh#DGeNBbgu1m`=@7<rM305c@aZErO&3*n;(7XeEjua|K@Nw
zy!n5@%?F1Mo_*$-^XJch^wCEQ{K~5@pPrrU@9!TR>^bK?_~4^QAARK0N)6sqYx@*T
zuz}DD0BKZLn6j+P#_!xekj$lx0D#s4^L#c-Gtq?R{P_)~uq@^#qw%C%TV7^&RyOse
zbL-DrjNgCrcQ$TxP9lko<i1XhCQ{kwH?Du~-#Y~dI)@0M6>b3NIXMZoV}_B@qHQ@a
zh=e#W3KlxV93dbB$k28UKCDO7??MzvMren=Lf&bnb>FOQuFZ~5+i67wnW$DMIj3p6
ziOL25sudP#XI7zQ2pAo}!NJkR$IqR+yi!f3OWp3t>0u-_OuPGg6UQ(un+xYh<AZ;_
z3tKB*g?^rnk4lqe5c&4q-CnQz==IC@Ztn;bwX+@B7cl{qKmGhSKK}8KqXPwC0ZZGS
z9vo_A=+oPn|M76HefPaLg`l;yWs?{s&B4K;AVZWSNq=Q^eRgujz>F|2XH{GE@?K<1
z5WKBfsVvPrGe+-SfXKmyh(HWJMr(g<Crzc4F@_KvfbibOh)V0>@@m@YmUSb9l#+gC
zyoa<Hl>z`J+gj^`$vTomWxZ;Auzs<B5Rr3kGMQ<euCH$%A3o4p>oi$ta*CpGE-ql;
zQYvdLAuRH&F~;a5F?v3os3gDg==I)csj)uQf=GBD0HTjhDI$~$?@tD~u<w6e^m{o%
z#mK><Vj-hIkeUJzhK276A|enX5&X<Lc5zNuN;x4IVg#yfSif-Tktd!o-Ci3gI1k7S
zfmr@*QJm3IYn>zs05EgYH1qlV?CfkoX`Ic^y#Iuw0D12b8G(t=dn*LS=)3*yz5RRN
z`mJxC+gz)P*_DeIh(jk)<!paG8V$8jAw=JLXRWu^xtbp~eFHEbA0Pe6pZzBr<?C<!
z<idsXoh%&=y8T{;g7*#&@9Z9aaQoh0egE(3#(D2}QC^DdV|2b<-&ozbcek78Qc7d=
z>hkiS-_5(-*|fCQ@9gXx9vlt_OUK8jbzRq0&CEaqf{ZaC1d2@T1+jnD2<$@e^I~R<
zVGh%?aX0TEGlb}@tE;+`We^!S0<hAOL-5vl=d88HXm6c!&RZv?R3rdE5PI8;miiG|
z)bShN`r7KosCH$Rr#EigK*T&xSJzjcd;Y1H|KiUgxf_pO+*rybx2c3qY6PH<$fN@N
zY^8%YNdhsZ%J|koWFdcOp|Au1A%r9f0LGXYS?WYc2}B`A=l!H8vO)j${hhLHPbbr}
z@$`wOpZTLd{^O=;UV7=Jr=EW5<6F0G-MTdx4LjZL{=x1GFMO`9OXpnMwrgu^{K>St
zpEufucW&3?X<J#gF_~0wa#q%5$!rBuRaup#C(}vWaw5~%!V7=Teqnui^>Ft@fQ;6j
z9-TsTxfafrMxkU3b#-ZVCG{$q)VYPO&y>o~lJ?wZuO7d@uNv%ihUkf%FqA<sI`$GW
zh={-lAV6x=#HdhEfK*TlJ_FZbe_o1FNS6e62n`8?m;niauA3^7)BPi`bqDYqrPr<d
z4EbfQ>!xxtR*8Rnqb<H~r+YF2MnH%{*Ond}o>aD2y|8jPJ6XE0sXJMB(6LcW=JWob
zD<pbU7q%XIuv5;)v6wZ!_7~Qc&#x^<A8+41@Ij@@Bw9um-Z$X=?yXx9*k}YP8y{Wd
zbLTed`5Zq<Q}Jm95~0FrHT&S!M>jwG@bLJ7l!`evO}&4#)W5W~-Rb5QoJ`4cRo9b<
zZJH-0(E@}bMM{zq-nrmfTbIE#t|{5o$X+D^QVCK>Apj#&XrryTp^oY=_n93dv3h7M
zv9MM$y30`@C}N2$6RjE86IE>;d>KL;qh$`>TkmZMPAYln^3`)&mpnjGl`i-Q41|o{
zM;kaWg8F|4T2E$n-nP!xF05^AUB7Z&qG+p{L%<j~2J|2T2u56dUPzEiQDq<#(g}kE
z5RAkqme*F3USFgMV~E^Dv?ze26w(lAh?4!HLBKJ@7$XSuQk)gdS<&`a)-K+-(Oq6f
zmce_5urQo*&V>*j>S`jOG9*mXxTbM!?W>}0>$ZO`tI65;@Z>mrQt3tlMCYAAu&z}~
zR!xx@vweBH@@E^H8%O(lS(?20#*bFl)~m=z$HyE=>0WekHZRkpm*ojSM2<)hAv8^~
zzrWw<_3nIp`!BxxgD*V)7)qUY5XV!k)t&o$wdJh0)Ov^jLdf7eNeKvnW1eM3%4=7z
z-2Q*?W(@AVci!$8d3JjG^>6*w_-y*jGtb=o@Z*!y@o+H6^Sqi*7OnZJsTsj~XLPEi
zB!o_$1H`&+>$>(anmntmTj;D1h1O;^o+1G7;wws`%$T~GGsh$`N?=vYfzY<KkWxz#
zg7a--ZL6iKtJ&p8w*JMx{BJ+L{nPEMS9X8y-^;VSEUMmMaO>`!*Is?uHSNmM^2XNs
zk6!)TepkD|Cq-CnX;R1#0$`Ag-be3To>TN-w5dhoT#JIda}rS~8NCgJI#J93gf0YN
z(mGMvjAyf=Y2W?+-)^11kSFQgyE~ux%;%0CoFpbI%ku5F-`?EZTpBF}@9*5c?L&XK
zdhN>I-T?r(;EzvFu3UZOY&`w6QbX{um<NP*qT58NelI65Z5_HvRuxvO1PJq_!`_x<
z(*gDQ=U(V_b2~52t#1{s8PBa5b=Q`AAKtn@9IZq~6zTxOX&%dEh;rsSbea?*$EV(?
zd*eJaiZG&9;ObxlDe~+DdiFM807QRcMrHwyNFWk15vj;bQJEMdLIjA6h9W~igp$LN
zYgp)n(t)D22lGdjBf}JYy1M#cyqjj-nNf$>&QZ28ILU}40*9Kk$XA!TgwS7IpSwn;
zn2ypFY&!EEdU>=n6>{{afBsrZIF-T3EIQjZl`G2Nbs5O}RyDPB^pp4B`uczPu$|s{
z^?QH(V0IWJdx(QxCsm>vpI+SBl)tcMEMaGmfOI;+@nmvx<_En&<06)@yn5A$rM+E4
zD7{k%Fq=&QA+oQk5|IOhvYh)6oNc`g0x1fWCRq#t<3ljG5JJcW3XNKh9-u!ON+YYP
z3Mknl#27+E6bz9F7kUzAP)a=oZUNvSKn6I3fGC-{Z5JIkFvhH}uRr$q<B`MO{{7i(
zwm8)PXJw7W**eD9G|i$mGFn=B?1?9|)=k~ymxoOOAAf0SPa8rXnj*$138f?<WDK}S
zGmD5JEY5MHlE78b`g7YCU-+f3tzOtBnKezF=Q0G3h(vPnCKpAd7$Y-S>t?e#$52<*
zY&NT^YBCwmiz)k0ZsAaf4uFZ!xz@TC5z^G8X?o+SC+B7PlXu@PtI8zl{lkO9lcS}j
z(MvDAbbq*ipp<cqEsA+lmmFP0%V67jetL4q(f`tyzT_C9O1}H*n=0$?A03ZpQ=O)1
z)_L;j=b!n^3-|XA|KczI8Udsvh#mldQIL>I3c_K(G<}6zlkNNWKw^Nxh)23Rq`N^H
zBt{6MM5LuVZ&DfrMoCK7L`oVY1etVqN_R`Y`@YBV+p&LOao*Q;e$v~`?SIZAJuBbK
z`G(eO@8iRx<(4{2L^@1hY;4!AaY0ut0tH45@gs!j83gD;>44&)wu;yi$7gvcm}#PU
zG^U7-y~=%JPS8z3j_>w|6PV7AwknqIeH>(tj1g2w@RLfuET;CGLv&N^r2kpDL+5=w
zDW{10JK*eS0m&p2_!M^PV*o$=-WhPV`caxkrp0sr<l^7r%}_rrfYEggO~F+I2Q0XF
zBz1|Cz9*4a0BBzx5HN=us3r5laO8dxg1`mm%gx$pFX+B7tzm++nN{2bO|B46yRFCG
zwMUiJ9cIHZX3M6=#uC6<_rU2XdT(Cw+-)BgSUf8?C}5ir;$d+|q#~Z9DUFzfpv!6?
zeIZQqKQguUD{6l&4%>KsN65Ry+l*kya_o9VzR?c@#+L>^C=DZ3ZR`vMr#xU6(835>
zuIL!#?$ybi!|l_gvPfscCF$GyJJN{@B&Ija3TZ27lpH>)&>aN^$rGts6rszKN5m8n
z66q!DK$H^Db@+)DOXx6d!plk6K%<oOP+n41=l4zGjUzH_or+=mZ+;_4&5i;*M!7G6
z!6{VE)jy3|QVl3rA%tv>QCGs-Ci3pV$>#X62I2d(+|}+cQE1k$qN88OCG_heq*o<0
z#x525EkYVaUA_`=T!K%GeczYQo~IzWpUQh&0WL{bngI{*`Qf|vgY?{{<uypfP3(x2
z_tUV{wJPbk7Uy0`(B<Xv($dX$dsd{vXTXuf(}7jrQ#dPE)0+6JVsh#D@6(R~ZeC0n
zKi){?X}$KKZ;H?-{FM)C_*^b*m#kp@P7g|%!&Y9*>s!iLA03yJ3C|tc-q@WBkj{Oh
zX>+c+G?-OlzfjR|S@7u@h))TeM210xX3Zkh2l~k{#<RDJC)33B2jz*Y<1&6Hlmoaw
zkz-q*{`~bVrNpjt#oWlq$jpqT08@AZ*2>H2IczXBc=HkzaI$;{krHThBrb>@>)#<P
zoSZyFnM{O9MZh^^mV!i|R+An()6I%Lf*|tfD*L~tiz#{9E;vU5j(&~pi9skp6%{Vu
zOeKxfnn6W8*~S^Z(s@&F^bO8;Jpf6XJQ`Zsp;&g1jjH@V96_dmL52MK`XABD#IB_&
zZ#_MM>i_ybZ|1}go!8XJXK{R*6|TNt`a{b7aikP839jO3%&3%tyNx(cC9$ZVOi>fZ
z51CYMWqsOo5!$7>l1x-m_Z&>FnxTHh>U$hNV~&ESY`&RyI%T4Ci`j4dG6A@or|0EC
zSJ0GRj%Me=0BT}I1%_))%IZ!%ezeu=ktmi<u*dSN>tk4V3l0_O^%fOz<fPDJ^l$98
zeWXsRFg@7H5!8iPIxfsr(Xkg#>pKYPoNy0KK-%gM+R2!H2~>r6G#L%kjuW8ZG>-QH
z_tBL+K{t34)>z|N8c!GPfN|K`c!NWt-PefZpTCf2Uq;7UZqFJxr@99IT?{E?c^&cR
z&=DDdAq8|%Xyh0;$L-S;U3l~X^~=wUG?K#&YaJGT{8yz?zWNYg24($^Df^mU9DN-m
z|7UaUjPdQahrdIO+#N1_Sd;||hQ{u1B&hJ~xEAK-#85Bq)BBtyz+X$|^1#-v`P*UJ
zw>q<H-9Im$Z%nwdGB+&79`G~n2|rI$?pL-^aPtyhk!)1MNI6wdMF_kYqh0Q3;3xAv
zj&)o9WMI2N{z>Y^VAk8|{pp~K)gURN5vS+rZ#=W3m7OFj9?x=wI5Cst1`c3sXe0ub
zke|;(;>WFA&@k1}egZ_8a&cX%s(gO(yy$^+Sm5w*HadQ{XJ4P)hsJAZ^S={_k{3jz
zaB#}c&7M1O(*8G<RAMLlTqC47KdL_XD^c)E2JxFcP8k#OE5>P&+`x;i#AVeHDYvJi
z;i4+qz<^z(8*FHyC#;6AWMyBYPk-v{=5R{7Kb0e{e}ZdYFckpKdyM-dSw|#Y5y7g@
z-iv3xSqV5?j5tg=y-=13G{0&Q=gXV-3mkX*5oB%SP8nPI_8W_;pP#sayq5-YSV~kJ
zs`Nzxk6lY^yT3mYf%s!D&BM|R(E(r4e*z6g^UABdXG4zrrGNRzOZG}bkdUw7oqSHw
z?`%&uN-4J}M?4@L%oULg<ylQ=KDrP5)`-9a!8GXkpUHy1IxZEc-)JYWGYe2!wk&;X
z@LI!a0R`PZxI^sth9-AC#swcDJKLJrB<l}RmtfQoy(!)3%HKAP8(hT!|LPX`KtF>Q
z0%=g4;5w(bFDFS>SOq`-{@ug0O`MUx&O6DFq``Qp2{^fr<&i=Km@*8>n|8e>p&&{+
z55W*^<b?sYaXKV*yB?BR@mVe3VruOun1q!c9#D%sHeoUZ>hO!6JD{$X3o*a9nQo5N
zS9dD7@3GQ)(}fD1$)W|Q#4h}nhK61g!|NQA$>jCaK8-S96_6Dk6No!5s36FHh1m`8
zH~ZWGd&F8&3)I}e&3>ob;ca~J)Fn9M!`#zpYiH~1E1UBIu6;7g<wMwURtegc<jB?a
zwM@t5sfVzDs^407#O0+v1G)lj0Y6MtG1f^sGV&LfHN&_nOz3Fm<9t1eOomMp>TrQY
zq2XkR>uIs=9QT(=#SlIpL_MTi%p4xk;0RpAUgEdeu5lK}X+rhkb(J&ROYQEgy-gi8
zzJB5ZwZo?aKHJO7*t7wQyB;gR)7Cptis$LW=4~IueMLqD=@oe}FU}O`So<=1Rf=B*
zeSxaIGmvLvFM?1#>y-sFsjw?5(0_+he6G4o8Z}C1VplSxcp;IKKo_v-44jNS{Q>}B
z_ulsdUBO%PKu2D(?A+YtW_KG=DzDZhUsKc8ASu}cg2=|>MWZFd6$zgcJ7z4V7&eIq
z6FaN%56gPPk-Hyq_u=S89Q#bo?$U2sJP%XV&Lk;cy(s1jEn|z(W8De6mj13C)$gyY
zip(K=<++v@bXm93(&87W@8z9XeE_sny)f!k&|zwP_-D{7yZ&{4UObWQIZi|{)iZyB
zG(M0#i0*6D2XF!}Z%sUkQl5P~E+5%SC5$d}WIH8J3@TDkY@d`ZIeuCC3X6#-iCK+B
zktxcEfm;;MN(V3G52b^+NJ-Yv$D7}ml`yPwkGXL~&cKmV&mvw3*Uzqtj{x9ijh4tV
zF)={Wu;bz1zj0%y^WrT01@II<vrVtkRj4yf8PB4K^ZN!0xI3Jaijc+*j^1h;Np$*O
zqgZ|F!p*t4xsPU4Qld)XL8u+d1JpEeDEP*d#2uuWtN_M%&xD1aY(&Qp++YBC5!sWP
zni|ykq)iQ`Lh<8#pZV<Ofq@h*MFo`T=}Y=l6^aMq<3t+i3?gfv({m-%*9|CuuFf<s
zIx>b(Su_7QrF!Y`YJ#8fz+)eRWJb>a+Jq<oD}<&yE^&M?N<>34hbmEmYgbzg$}?Y8
zM;~6n!7_2`AudIkC+KK8h3aH=ToCsm4aGx+X48ILn3n#OI&}$L4!uc+vN0EbhJtHm
zlYP2g{hQmyeD+=2{$UYSTktB<WDZ$}_p!Z^&u<G7O3)={v+$!~nwv<+Xvq8eXr<l%
z?(fBSJNxZdATWDsns4p|Ycp<&HsiLu%ZtchIr={+eAv=vJIDE9ZwB~x_9usnH2Rcc
z<>sW_`4Kq-eLV;qjX5L^v$`wJ4%HKqQ7H4(x_7ng+%%*E7mX``*zBl}jQff5)0&J&
z%LCp_u2K8d%?FPGYzi5le28*S2y%+@xWm*x5m{uECzmE1IviP92|GJo)7-zG>d%E4
z8O3%{)6!INzevY=P(V#G<$Ge(RfQ~D9b`L(WF6EE0|u>h*ptxhL+nR)^F7DKJcdQ&
zJ|o9X)QgKrdevmV$0O{cRv8;Y)bOU2w;kg&gHu1j{$dw9c=(;xC>sqN2W%6>Hs@rg
zqA+HSX<DlIkr->uPc*>osht?X&$rd)JXyl{wE=eDo7>=wphmi9uNMMUs_e3zV=HR9
z3S|)yz&LuwoCIN@t(q5tMGq<u!krbIZkheG`V2$qd6N)TQk^_+u>V<7?8;r?F{<>5
z^;>q^WnbbjqI#|*4gIbsH_>UMyKld`!$4zew9gM(q{gh~?C2N>D-QunytZl?7#pBc
zWJ$Fk3Hj=!!=XDhbULQG@Iy_tSn=d>8nJK(MEa{%n4Z+_1_baEGDSwSNSGR}m$3D-
zG8X`k5hIU}-p)_QfV;O}uX^>dlQ4%qKibRi)$YfocJS&fUpP`YzpP}n9+zhJ27G`c
z`I(UrGOKto2>n!Mk7iRKB`}mEw)D@W-{uhh0Y0u|Aq8hx78u&~t4)vb|H)c3{IdXn
z;9}BK(0`VLaO*koxy5nC?OoK4H?(){94TX8A0<je8y>}4XwLerG!5*25QK4N@#fB`
zmMXK&snhPh6Yh}7q{<ymYT(tmDBj&XHzGEsKp732>Zd~b8cL^pRtz@C0J&mHYwfZZ
zEUOA)4P$-pUe`P7Yvefcf?7qhH1~kooey@&WuwA5ns2?u4OC=#Lj_)iL*&4`YAi~n
z(WAKf-ApL@Ka1msqpn`pk3nHKcfWw|t?t`@FXT+$4jVh{x3K@QH<+qx$xQuGc1UFv
zwK)BQg&D3I_}3RD+41E57x|LLiSJnh-3yJrjUP_+dTJ_#^=i5-99dZ5+IC+6mZjdm
z;a?bi`xT(cdqvUjO5X9ctbPOs@g<QI5{yj`4bDSxdF7*it}^(#AP{>MA1ex8Rs0}q
zAAGKCR+856Iz9?@<7Fz>wo{Um{|bl8ei>j&BFz6d{~iXfBa(b48-;7=uS{iig8s6{
zpi4k``y*dYwxp=&Y_oAr&h+6VAh$6p_MVlA6EhFvI~*;esKSw2l%fzGlGC2!{50L@
z9OSc~)%oafesge1`>4IEfcx+kxLIP&YYPIdRZ4TZY!OcOBV0vZ-0#894@}D{)L6I1
zi(^^F)Yu{G57*Aaou~R5m230O1+L5f%n=G@F*kFks1m26&KugNQa}hG<C2RD*|2T$
za1%bdy6bqA_Nj89p6pdT2Z3Gf?HV?%fudBdMr8?Mk+%AAI|&0HRt72M7rfj8+3;Nt
zj&MOGY#T9*RO>EZC2cjdx6>^@s*<@#<#CBXOxnV480(XiN}I!NaTKdpQk7B>c|un{
z(1uyFpbO;D(J-l*pH3+-@GF+|8Xw>Edr|T32h}f5j?OFX-Zt)*8?4TchV0DH%Q>Ea
z$A`DeNlMA_c&#%El(gOG+#KPuMHS@@(;95;ane599|{_hVVpBCc7?bk2%1>9wm!c0
zi%&~S?d=QL*VRit3aoufoh>z2t2=Vk4_kdKpheUfyT2|>FYE~1-R@V*oL@{BlX6oX
z96jtrRy&%u-A^a3oSl@9K?p5;Clzb$s%M($bg+I#Ja%$XX4)okC^<MT>=&n3o0%;c
z2JRz7gs0QdPCi2ORSWYAxP;%f<a7kN^$eD3Sbbe?b>}YH>nq&CrgcmgCE-^>$L;Q~
zlHK?}#Q>1;-`B0LUw(y#JmclK2w8vxI{RexIWBO2<yA5GtTe7~%x3?72alJkpA3~H
z<j0;RywzVCPw_mH_1kCKyj=WrKir82+nIEU>#P>J9>Nd3JMWp|mQ}5JK%~)Tbjlwo
zuK$?dZW+r27Oot>h+NvPc3x@re(G7EHR<r8iJw~loEefZ&9ie$#8v5{QxWyf=n7@-
zjCa1($kTWH)hbeaJGBjALAeg@k58M2L+@X;e-o%_zq%N)X*<{RbjUTVN>-^*(RpSS
zL%}xi1MAQ0R!i)o{17IkZFU|xhzH{f!EqIAmEg8w%lr+c9<SEJdPmfOJ5Is&(&*Rw
zn%@;Ifam`3M&^jjk=^DRd~pEpvfF2^V=+Az@4&sI{R+0i+wK<NadPiIf**VE*f%|<
zQ-Vw!biC;KBbJ_^KHR+e(b_tJ2mtOr7Z8q@S8QYBpuP8CzK6U*pDi4I3~=yuvRJeA
zj}(1VuKWU5i=XE&6v{$HuWy9eRvSuL91m3lk|^m0ubcAq&GZe#=%UhehMV8TJp*A{
z92PZV75<T=BxdSZ4dut$sjsPD4+@VW-u>+EBuZtVj|sxV!<#NU^(ljYX`=`6#T59p
zKQi5rW77XUW{_YjcBRs?^-txHd?i{6%0-o><8FUMyrkEGIe?I58MfN$x?<a8rSiN0
zh__qiSF)c#bz|LfO-+I>5}c#~873rTm-7ha!w%P_0K1gbDzP)0ls8Sr;dKSv+lg)e
z#6?BpQBhkw{{Tr#<?k!8;{atsOu+JBo49)qhX?v!!=~cQQfZ{@g?d>PQVWuvUcUa$
zmj3P=XT^9hyMbS=-bZ~p_miod_&cb~=W?6en~!GVyUHrTZqdKl-Wj2;5$THTO=QoN
z!n$}zI58?>K3e(O`FYXD?J1h^Vp-+Bz?Q=OFiY!^i1M}99ZOyh_LExM1Ze%;bvb;L
zDvcew^t3T0`4tr&ymRHZ;W~XtfTLXuKlW9n61OCy&t)xwAPtL!Y(o9J>BWE}kDO;#
z7azDLDyGWy6X<x&43MJxHzC3$9Jmf~27O{oaLp=#Pa8Gh45j~Ay6%zV>gM<8ZZ*xF
z)XdBM(m{{QciuABH|ebp8(Vt&sB&o@1ed@tjpIi_<5$I(ITFY9Ul2WE4+e!gU&Weh
zkvJ=}i^E9yg=lDum|7+Dzh>Y}YDBEEK}QXbo-J6YPmJ<ml2li}`1nr0c+{wm4D&g9
zo;xaI3)(<x9m)hVgy|`Ix-{jb{p!!_M0FaHDO!+W6|Sv5hTR$E($G-j?mnE-KAqag
zoRfp*X)odhkUNwkzyo8*r?4<Y=l!1*p8A2aHa7Ltbb-T9UKrp?Yye8063!PE9U3he
zxOIQ8=iu$1W8dy}Hko+5EUp$mv5O2~{C57GIg>;#gM=kE+R*F^bv3Qo;cH#oQf<H7
zW4taYWQ4|g?kn1|#i?XMdP>)Sk^70xVh_OemD7F!#6kr8^uc>Smp$cI^Y!?8^|;x8
zw~N=#mG_7)TmLFr58xonf2z=8ZSS(7jqKN{?00&_3qwl|3Y*>U+6e;t9Pzdxx*+*i
zbZdH|Mbm^`AvzP3Gy|h%qMlo)^I=RlTiYeuTQ6y3X1VJ((KLh#f|G5y&BUb-jI1gv
zsz!aj7Rh2l*AGNr*=kpc1RAIM=Gtxlv^pR9moL!f58MqcTmyI!(?13B?tf)ACms-9
z(Wm~mF|&MA=Z9#h_X!mku?&*3R~$bVO?u8&7{{<#H94OsZ{sScVKZOXn_N**H62>>
zlOW@SYy%baD-~N-!EyGQPIOq=F4w*?<CqXXP2LLLha|I#Gidqp7q8&8{ODJKd2xrF
zDYdsd&ugYr<Xr%1qg<i+3F2Lb?Q95?1oM$(^le%?xC>P#Ek$mob!HdgQy_`X9S)J_
z?E}jxp!PGnhIt(l|B$E++wO0y;T{s$`-wB4Y>3BuEDdsIl$ug`^ebbpQEB{?Esmw0
z+E-4~O=edE?R*R}UbSElh)BG-i;R?{>h-(#eooSHDFD|H1jp)m@S9JP#2xWMDqskD
z6kXA3=+WWPM<;3PICM<;VD!zOqNKUUPm4BY2?YcK-`CS@Rekl1AcCs*b=z8QS|)~V
z9!;It!KIoAibG-HA0xR^Gi4$|O;y2abE28wDq^6YWSfsAq)Om#Vb^yLJv}{|-ad!(
z4KfeI<^Uf{`}arIl^ag(f{`I^9DC6oNv*`I>jd!}$swlg|6ZcnJf4mpH=izeNZ+Qx
zbOjdAg{Q^G@d%UXq|E?5GQY!H0aVK@7;X@;4kz~LrFTu>;~=Z9zzyLWCIk~{^jm`1
zkkEfbV~Q5=61%}FIpsKFXs~dfAwo#JC|qJ2cupVul8pX&GX^;h57_J>qS^`NVQ%vB
zeVp7q4tnxD>Uum~UPN|x-mW~R6LWIf@R&ud+d9~sv&#7H|I!My_HtcYZuRCd>5x3S
zZkV>xS#0*U@qO%_L<TqnwTKLR7(jPPQ?Y_a&+DoJlF4MrEw$5%kA0(llTE>N=t6jO
zJTXB#zM*J&k~)8&#YSROttq9Q^?*l!=-xs_cm-~Rh?K-@wu7T15Al9?clY#fCT0sh
zdgW(*=OUj958eWoZ?VI=mX|f%{q}4fIPHX`InrNSqcH)Ok+f|mg8~|(b)}?E%jPoP
z)(NNf)*XTHOQDvB9x3(|T4%|YfUEcZF8>;P3fXMGN%{OkBzC$Ko9%lhz{~<)#~-=f
zI@8L~bp06l8#NDgvx1)+7f-B+M39k?O$qi~UyBgI$uP^=mw%9J<8w(5#|f#>gV6|d
zhaSRR2R;Lk_sI%O(peSC<;YKcIxN{<vR$GZI%;GyFEK5#5W>v#yfj7pe;o*DGN7JU
zt?^|Yx!x@mx7Ckhzzy&?+<D}MKf;sSdc7)VQ33nx(rwFTq7RZ=Kkp?l4Wl8z9VA!(
zci%0z*G4lQxj<<gIOMY$;ksK@Uh$+bdiT^!X2~ZoajpujX1)vpW+g<gGA+XMLqzc_
z7d<E7{{%Jm;>%A??<5x5m(DBG7X&8ed?caiX*OuD(2~=8ZO|B~Sz73kvx9?!Ow<&F
zXs6M?Hd3^>^2%yNx8u6Bh|{0hqP+qNg;FoEeRJ=!p)s<bLHwaAQHn+3F5P5kN>pF$
zCyxcliJ0O*j2BNiv?v!FNwB!1T3^O%aVZJ!!})%i<)?h_fMBa%uC->?gJk*!#EP?#
zIR`hKte(Z}+91v%@`Oa%s=#3Ug`rzxk(F`e{Qz)3vF*G#yEt3LtO_H!J_ejS&Tgs&
zUUE1lfrw(lm1wjk#^!sl6K81KAFdYQb=%I8Pg_HcEUE~T-;^U#cUM_59WHC@!;jK(
zGh))i<9fJb$CnH%3-dm0%{Mz!)sJm8;*^O$XsV|+R^I_Us5&sy7B&TKyK|B1r_UhD
zc%{&DrVF&3u`l_=**a>}vOszLMIgIkwao7rB>mUyukOBSMB5mmOe(?lmhtW?QJ!oO
z@8ami<t(x{3n%Ms(7AWY&()uXfV%~*ziBNgHk#GW-)KF&z@Jl7Gap?}5azcAM64+B
zLu+emsQWH9ayF<l&uVTc%bBBigK2qr&PcSTt)u-gB1j*e(_>lr5fJANKW0SRr-JFi
zzu;(!VXKLeemxf3u0UdRXzR;ClJe`gW8@Je^gr3i3MJXC=(e;p%sa01WGHiFfMxg|
zDz~byC%(V%^q-4!9%Ov~GA%9L_|OZ*K$x35YrRSA-xz!6<Ka>7x2IREuDr<V6mX6@
zMKx`%w}xb_290z9Uiofb*-%BZ_g2sPj4%DZBowjkd7fYdY*enS_DY$xdf!~!T<P6C
zT(7LcgEmnZ=d6zjkrW4>aye@mlBCq*@ZE7(koT<p^0GvSNQIgI_wyMCf@PZPr;qOE
z1R}zn?g8~YZD^w*;e^{So9X9qiQKl_L1?nz8n~OeKb#*R<)aVhAtwggDXKB0h5eO;
zgnW=jzr;Kzn|?nQqrjy2rLtbTm<6jVA5%7Fk{<NmOKNl|N~Q84?v82dM|%7#OR|5N
z9Ir=En`{nKLOg?if6vi(u7=k?q3Iiv>D}j2JnD732L6Em{@?Wgvm$qQPyLG$W1%hG
z4<u}V(A*#ZR`KiOiM*TJ#T9ksytlgq%XfIVpW`Rm4-Xmsd{S<=M!4?Yr~XN=KBlIU
zT3&QnZ5ny?#i`W%@2}|MskvQDgob`vA8Z@K_Kb<x8@;bg=p`GxMRL!ZqcqHq0o`0I
zMPE^e1V(O|U5EcpUW^_zXnUCLK5SdqP*EFDRf)6%ir_+7Z-z7Z(s-5F4fk1GKE{>6
zdsRs=Lh6J^YS&ycwwtGsNjOp~q7tY1#U?I&8SqedgX)>&B7E7bc<T&>Y#e$?8|vcG
zeJaVI`%wx;DZUz^-0vqj?$bt&Lmz@Q!Ms>;I9*0{b5sP%6<lc4g>YS^U=ZXT@jB)<
z2AMmcq0%jO$#fpj$BZ)WhDRq~4H!l<xW>K7PQ)pO{0kxdJrp0GgK#WVQwq^Jq7s_d
zhXvl?RkI94>g>+``tzrxwwGx5``k%Od;43nn|t-VKs$H$b7X4e{(j<qe4mft<3*`i
zz~SZ0r;Urnh|<yp)&QQG)Z$<gw9mYs{96|Qg6X5ujZ@U}s)fi^vu~K`c%p#N^u<>|
zOH|}GW$woacuagzdV&Py9>F)MP<6cXFpKIjjpgRcG}5WzTq)zfVPQ+J3Xv16Z<Usu
zUM0*lUdJ(;k)VDpcY2Bj75?{dB@ZaBCOed@n4JNwD{y81%CC4FBSUzef)SH-S`}EY
zACoO_a1y1Vj8$_0y(0l>gQ3;fmsoZ0f(}K3b}u3}zteD44HDE5UR#kwTBVsayIBXh
z(#LIJU(-_2dP+Yvr&r4aJ~*4-Eo__nB)KjhU!u}7Yg<}idRD-w#GY3M&(ntY6WpH$
zb9_l+kM=?=-OdR4Q)FJ;BODq~0IbM?x<RlZKckXRfxe7}Pw3i8V_)l3et*c`AhU^+
ztEA4A^pCk)zTBVr^dQYpuGJusbA%9a*3{B+AEre}x#6$)?H}g(-)uHIFRjIOxNO`^
zO-;3jwU7A(xOkkL{pq|(lI?kXygGjBI@Vs*%rt2I<^lVU;=jv>GPc?jJV4-`AaVW6
zx>y@c+K-1kT*FK4z<%$2bl>{HH7dM0OmjOdUP!GybN2%;jhB+5zgJL3$uKm}#~?<t
zn#?P@iw;|I{Qxw8&m3`qaZQcMK)wkFK~<xkaTF`4u@nAeWG80?zl+BuPlYOFyP~B*
zBY06&_FrkfChO3%dvkvt1*)*4`^6%la|Ldxzi@~{L3}S?h01wg#0#hd>sSlDthjLS
zWQhQa@_PBW{DqrIHIw=4@uzK0y^HG`Lz1x+y9&mH7I0C-f2o*U9Q<gcKcwBlA=TKD
z{G#7iRqcf<#LA9>FB5<2YWW|h`OQ>BL?=N$kKWaT_tQtgJ#YN$GvG9z<~G({0G=co
zgJYS>-GAd^auM>i!N5h1-FdR4Wd0l^?~s#T!G>!S$KmGiqPSPA&@*$>*Wd%zsQP;T
zONaGw2Ca$u3J%n1@iLgK632B|S_e(b&jpdnhvaaK=dNWq32Sa{W~E>NaX0a?jGRl3
z;JBie9BOJOr;m~WFPf(JFEz6z`KVqfZ_uZIb4DWvG~c}WrMhzy0zqdRE16xGn_XI#
zcWC+x*Kx?^vzCt%85Xx8bVYxN<L$y^{>lCfw{H*1?(th}1CFLa71aFtLoYf+z6(dS
z38cl#^KKk3?5C7#q0%;2fjc-wW`wO!R;sH&^ep_^=Wxu&hpQL^-D}+{?l0==nU8!X
z<*9T7n21KukdhAP_@3TqRRy{{-L3eyFJ-`5JB3B0t}{u^P0X@sFIqdVI9JU6r(f7Q
zI9u#4lVQNmGX&>T;(zkL!;bpkWD#WNYXJb>u67+`DrA<0RW9(APN35$5$XEW-s)X1
zgsqIN+`duE)o&y&B_&06VW7$NaCH25bPAlFu8&9a+WpVEg;HL!hOND`v8mLl^vy*X
ziO+iIz@e~Y7quk1!Z_WSoioXF`K=0$3N=n4n$kX7`X;y=1N-FdKy{R%K5o*zhqZFO
z&NRCfk}XL9Z)oL_^@*+A*Gg=DWnt%%U}N_6q#|FN+vecoRo_5riu2RoL;vFrfcE$5
z^pN)>Dl>9z@nZ4&f_X-cL*snm*iNL(Bg)x-)+h6CXr<}mLU+CixEgtCwq~+tImdH%
zpU^d+?83m_!=W#RuSd@jXMU#^-OkPt4>X`W%-+nVEJREh<x*4<CUy6&XSpI{3h2VC
zzI{`Y6QoKwmsDo*K3cr7dAb0AdMx&nbhBGA^oz62(vlKuGvCcr$~>PnMC0=O{5(Zs
zzs*Rl81Q(njQK*v1fru=#s<>eg8tj##m7YZBlXQw=_ipw<?Pz@Y4yWd^>T;5gQK^)
zKU6~~eV7O3fy%S37EGu;y4`Ym6=&IX-u0cFccsu$Q3;#n(89??iSs`UWN0j3x}qBz
z8(Mj~lN%Zpz>zlL3gL(R1T|Ss@`*>gDtBiyCV)1fw`h&ru7-^lgWd9V%P02HbxQxC
zzUtCco!46VuMb|FSMe$@Wg|~!9@m^dC^pvtwrpa^7T4BO7aQl^e>JtV2Ul0vMkGG0
zU)J1+2Kz$Z0a|OTJK5iGUCL64;vz!ySP(H|w|#ej_wkY-w}iJvgF#}QYlZq@9q%a$
zAQ#-h(YjSd@J)RjDY%g6Sz}Xo*Dtn8dU<15Ni?Qqui7pbqSgD55)ut9o{r!vc1MbB
zj>7dHF*e^6dsMtJnDfZK`c#<EBcIhR#>spgnCHGVD!N@IWMpvF($f=mbanOE6G<>L
zJ$<CV)Ea5q!^V^!Hv~~bJ|{J}?Z*2+7Ag0awPE&JYTu)BDyo1^ty>hom{l^7cwKyH
zgKI?8MlChvsjMPqV5HYuZeOh&x?eGFs8_<~X*LgYuaFjEPG53IqX2eOOAzkF5i`93
z|DG6kF!}k6+x#xe_h+iH`Twy$Exgf7l@ag>&ynbSp1@BhJIR{sZ_g`=@lM|;N3!O9
zJyD!iL#baos^hGy{f<_JJm9L%yi848qTM8o8G;=S`B|&x?Jb<jG5ST>!2u)~?T{!U
ztX!DQ>lEkNvge&>i~zQVmlJdpGfTs|Mp3aB&8rU<TE;2cQ(ceCoYigjtIMYPBxu|1
zh3%=E%!Ew%|8z<#|1iKmRv1;r5K%*Mvvm~IaYQ|2$r*6&^JQqB5zia02NQY>qq}Ak
zImUSw)f(FMurXqQ$_!E!(*^$uh<tmPdHPHHw5nV!WIV+AmK$(2?{GhEFEdCiG(@Mo
z!m|64mWms3UD<hY$hpmJU^O6y0-6jBVHI&kW+wMt1VL=YhNh;67ZQn$!jzY{s8q0w
z+)SKsQk_o!*+VZ>spc9nf&TmK%)i-amf5wu&s%Z|Q3}t4kB7XtSUK)Y3GcV5GONn6
z-EGBeG=4TDngdQh_AwTRM3zs)Xs?v5U;&qZ%uP`_H@~#nGpvkYAFv}MM^>7W68cx}
z9vw(|Oir4ab`o0pBXc%*ajQ+A=9ST3L`E*i_&msDi}Ss4m7K!;QvkMi)!(%J&LSlj
z7niJ=)`EPc^{V4yfb?nR>A_sc_!6&_hbvDa@OnCrpasA#9r-4_C->7(Y`8f;%=Ig0
z0sm0G*P`klqUirtKZ_54hS-FIx-jY7NumXxYoiOuvcI+%D^w8ORv&@=VM~ge1anW_
zPaNj|2iDZ)$Le(h4P+{-3Ve=_QrE=z{Aq-A9!K#tJ1#vrkz=;2Hiu;r89Mk~G~-X!
zZ6#0yT{;htcW{AUT57EKb9iiQ;8!Mt)+_qR$bK~o9>j2-DQ;+!o~8s8o<{U3>Ug<j
zw~9X4jA^6AhQoA6O<*1DAhJIl7b+@`SHj=GbU(Qr<ZADA!zflwP~E2CEbX&o34IHL
z*0j|hxN5o32dYL{qM;B)@0K~k28weub&9pghzaZTEFC;*eF~hfaq8G4WbF0q4XT`0
zAu*!aW*O;n`C~6z#$gXRNq-b@1T2P8F2dnpLrYFe^?i6(_bk#VLczGOo2^RZOAy43
zKJ$dkQc>rxFCP`Y#8P2bembd9T0)gByt$StJ>Ae7jU6W&e<f%?Uas>`dW<(3{j-fR
zNcuN#)C|7yYr=U=5+vU9<)0FmThua~ZgVo7pE_ZBh9*wg5-UYf5WUd+VULGJ_3*W3
z8X{Mc9y6tznb27zWvuMUXXCMZ{a|_{er|U+AiHE^P<(NDA<qo;pM{A5Tf0BAD%4yO
zKy~1Ganrv`PS?N1Ii>g{1g>|+Hc(y{6vOH52RxGx1~Ho?)U6ZKwg$ENKH1s?0`7sc
zyMhj$HXottC8Z0pIM}mui>q)fsC3XjkRxU=^yKitat$*TYWHOOZ7WA6(9PzUj+2ue
zqvngiJB;`ml8ibwscHc#PI*KP{bncx!*UcWWUzw6NwD8YHJDCZ=|hskK2_2^Y0yn9
zAR>Ji=k|*;@2NZNdItU}Z24q^CjfdI#1(oRc+De4o-OA8xb{Fm#4~FQEJ}!)n42$%
zEgFdjuD=^qKSzCL0bSjGW$X=|r>vFRKGdokwEM@rDyEivYW&g1%j@y<wveB?(!+#(
zKfCxjq-H#o5e!GSv@(Oznfu&q>tzCmM@M}PTcWzKN?tH0g8UhG#<oAa;Iqc3QKuQ3
z5@V=*()PAy?(Xq*`u5Aahg?!D_t@C1tVGoOA6;n@g&&shF1tn=PKZh4?yI`{*#l1}
zDS7u-sL|XwU0Vry@?M8Mm(}EcMPigNSx4uI8L~F6Q;K&Ip){d(ub!?c3d~LDPdSa`
zbv&-GEw?9{c8a@S#i&L;9Bo*bO%>#immq!Dg~acC)6k2d4$8q2f+S^(Dj+g6da~CL
z1?|WLL(QDOp#mH%xD<5HwV{3WVp6)Typ$s%R-7K$v$Ld^qv!1Pp={RH-)Md~1jWl^
zd>2erAYwD@6MGGZ!$QN^0D3uGE#t~6rBILuzXxXMPTaIUf0jqzgU+`z9sBw}aS0Bm
z<BJ!A9*2LUcVScrDQPaGn{`EIaSX3)rP#9((nZE>V~oxIHlyp;^5~jH7)7$wI>o2$
zE}}^3^9X0H5w1X{?aMX&inFU8+Gpdz*vZJ7x#C|pR+{Zfm52To0L;4GLgvk%S?6pq
zUpwC`1x7oO4KIOQ+#c8$gTD7<ilOQ}F%h(iLtEnmx}2rMeTU(^Fe6g~eY2NJ`nRQ4
zInb6WJtKHC^uOeB@nX-f9u*oT9!BMhwvKTeF`+oqP9wZW<dM_wjSSjRfj0XDSnX@y
z1;{Qw6*}oy_ra$((`EfpcZ~8Y4I)-K$S0j0Fhc<!R`in0Y_tFNw5=$eZM{H>Yf?iW
zDr9(UIs@m~q?r6Dq@Rr$gPl383D(|4R<t|I_A9eUmJOoBg}qM3+ecNH$?sa2T3^yr
zL|-%#EEWrthFQhYhYZA-xoDDSX|rw6Rj{f`qr?{{vh0W0M1+LDftil0%RID;(h)9K
z{Nx_w+2Vmo>)1EnufNYO{tir}Ry4ueyWZ{ojZW|U7twsvyixeLI~Dt2mf8C%k|G-I
zyS@?Ck+ZXt6BQMeq+k6NipdB9TrT&fM4UT>hT)7PIEI$YrQXMjetYA;T(?)>S+-LV
z^tC$)AI!*WZEpNN2P9?w#!0n3J+3c)-2S)P<?n5L*~ArtBqHyXBg^Cmt}JJ6+aIr%
z(Kd*pUwZ_XAuKe=l(Sz$s_L<rHnUY_*Xv(nl0^0RRpxFrGivN|BotQE;*+PG5%ZYb
zbyxBn(;gliIDPw;8+1O@cwVsg{O+1&*;~^+e|tFR_zLSc9WXQ=t-k))WuX^~=lFpP
zszu`|-XI$Euj^qfUuNp#HyIO4IR35FdiIAlV9_)q^Rx-Il~Ba<Bc<{uMgN4;VZ?dH
z;aPs>cdGYodn{YBe}UEWQBuiaPK)?+yn&St2!hK5zHk1uyW4D6IJMBY4EtifxFRO$
zz16+87G&@|30Ph3%7yTs^wes1JpWt^GtpISZtskx@**I3JCfb%c5_aPBS=O2vHc<C
zxUKVtK;!pW@(Wyjik`m0_q-el3HqfH9d53vA;sB<Z-WtBGSWWZ8@*j_yWLBx+eNt(
zV7jDf!-0>zp;eKB395WY68nT2UO&zSi;Ig>Yw)vQ!vQD}hB6D<M7HjyVO-oeDDH1I
zD>$#LTy%fPGye6m9?;_ZcvG02W`7Wqf<Vj{)yZ$zoC5J*wC0nt?fy+fo1-MSoc1%X
zak0t^Ng`I|&t^B`VY~}YYDy@@?zHy1s9}1d_srYLI^0o~nqXD1y{YIQ;~~Gl6WUSU
z%N>mXFL2-)DXU1*KW*D(r2n!`U;w#SA0h}YhxcxqNokg3A+)5PvfpXCwbnH|>Vx(E
zkM`~{oE4+7E$%7u-Ox49%DlVV*@@-9J6)O?ch#g)k_))KC@*?>MD;OMKl&%Nm7-lr
zk(5Z|q-{MiiG|zni;TD1C3-RH_9$0PHbXpK80ru@dA^VEGOWB)EkIBv)JamNr*8zI
zaRH9p0f<-HWBb>M68-IRHlj4~wQD-p`7(QCRS=TYa2$|zFyq$*|7<)pQGU7NejHJH
zF)IaLwNcmMXEL~<V75(pHdB;tj46)2YY1^ENDH;6q9~3n_+4KPX;5^i8dEZ|`Ai21
zkA`B6p(l(_ivJN6m==Fe$P%&!Gq&oo18_5cl-KIi4c$Y)@!h?|^63-^O!RC%d^iJd
zellKH#ybt%VRlxGP&WBKn(As=>VwFT5lpo3w*bp1s!K%;B|K>kI#f%0VfC{R_B`7t
zJo6?(Igu7`MaFw~d}9bvPz64lN+Go=(K0Be?}ggOmnYy~%}h%h8v8f3wz>xZ9;oEw
z^R^~b(_5Cv%CNeEJS!fajm9btD4%0ycT}&!--Y++TDr-5#MU&?0c-49(&q5)`e?e~
zCn-GtMHt->lCHu($QIc~gQ*-Dx%3DQwnuER@%PzoaLFN?gq4(3D?H2LQ!$M6!whtr
z_0Z}39~gML!Y~~pgJJ$vIV<3>`l+wln5$uEd&{Qt`r+6J9W}Tkda6>^2(j5Y>K96=
zXR46E5Qa&MN6pM};L&+C)pd)+9d>JE<dE0?za5|1-CKE-jx5s&)mc!&W8StXHLs=2
zXj5NaVcp)v#$NGWl7>RPN(iow`#ggwT#Y>ziX$@|iw=^$o6|aW1e`a0)o}QSSvV;?
z5(KvHa_f}*Y8PZ~2TNrF@2(a~-)fn*pG_i>dwqRddadLv>rD${uIs(m=Kg^dQc`<~
zIT3%-bJe<i94sdi3pd7oLUSjxW^(;DQBZVG*rCkr2GCQqav@5JGAA+cRMn`}Il&aa
z(uUgq8`LIdd29><HmBZC4sm9YUR>OqvvLP;*mnkxLg*bZ=7w*>N{4eg1NR?XaqaeO
zj1*<bB8lH}4UiRLgcxox73Pve{bplc1%V5BztRdQsD^yqfKLWE3n#qX^AKLESxqK1
zA0LfWktI6mEsg(OiQPtc3j3MzS;v8lewE86+m(YXeUYUG_)v!VY!?$|6v?~WieVXw
zRwE8JoQ><7^SNA9T^HO8hf@39|J$*(@b$H@>sUGbQ3~_FD;)c+B=l0L#y!;w^Xo9N
zS*pH~Q`q;b=|lg8yY}%U)P}$4QJ7;Daomq0)pH&v)mJjIWKNT^XCmL0pWiJwry7aV
zybujbsZ>t|)V!~SAyjhf5o<Siq!=kiX|zYwmx=vDeT&6-GbIb|;j*O@l%<7q;Gz@w
z81(nmq#$NmiIKry-!G|(-;mIKXt^7IOL+y7T$#N%n}IKC`V9JsDHFE`tCS|Zfz3{6
zi!{x8E1ks`icuz{X;fTMJ0n4X-&&~CIU%+4{OZpO`vdH%IkoV9_Fr@tEa`M%Xp}&7
zE53Mof^y`Y1nA^+|84U-0-B(FJqv`ZZG#Ss6OroawBEMPG9;~urTr3z2z!T%H<?LD
zMs<<tJSC*oK!U5DW&9a=Z~PD1nIu<`(~CY=6YI_19LL%g%Q$tEEQ1&+vmNK|gz&^q
zfz>{qJPu!#K%SFncbh$A7G|lFrUUHnNc>cZJ{~)Luxjyn@9GrPbzMl?dD?fdXyR{u
zHMa3WVaLnwZM8Hd){tChTz3Y4f5uSV*v6%@_eyJBUHcGQ+rI(Q&WmE6&eIeTl$GJJ
zhqhhE&VDK7`nSUF=;3u<i;t_bGhpc+$>$X39)5I0?S0piIY*8U7ofNeq63=BV<f5X
za%T?lyNi|&uOi!r_1c9g63(EhscOl+;tr5XRG47OWa@Wm)M}bomyE9GyQaQE1SQ|U
zIZ!tbyrI3{?mWA64mw;<b$^@Nv3Q8@eTC1t-*Lakss3jo%b1|?$4mt4NVY^LstI9n
z9}?qUCom-{B`GTDdyBGr0NLgl!1YlpPshJ<<3?wZdEHJ_>35$FS9n>Bo2Qi`9o&vZ
zY++%c_3>iD_JwHGStvt|lFU>0`MG-C@p7vJe4<ZVje0y9ubdTw7;2QQkn_>m2?Z)h
z(mW4HB#o%WN0W6CuK91gA~|;^yTYOH%B#^q6;%lY9A7`~>`lA{{Ck{M&(~b)uWSRo
zoQCa9Or=@{J3U-HG7|eqWU(GEx&!RqIN0Bw-K0ws*^=SoKkmrf)22dI#J_PwS=(sk
zAl({UC0+)x{cP!;$~b#XGnLYLvw!SGAfbxz^|LyiCIDRa7V#KzixU{1(^fy-tlyW<
zblx$BAo>@Ixe1ffe?*|E1Ia|;gKyY~K=<EMUbvNlUGXiFY5r0X5%qO!CGgHliQ4{{
z*V)N2G#tjd(WqOpSI`|2J#(-co}RvHCKX=?3z*4@^)N-{xM>oO;N3ku&YizKT3#8%
zy?$)RzVr82l@gy7W2@jVORBzKcNU^JrycMO^eLV@iT8f2GD{4V1(cGse$vo5626zx
z0x9+>=2}~s!8WT{=Ct25l2%Jyek2mfVeT38k<=_EiF4i0UhOt2Z4jRStEiJ2VSu!?
zLX5$TZK@hTdZ5xcboXiYurE^@vlj&=Cl1+IW#5xKNSWV6N?yG7_`&r@6dyb6@twyk
zCDmnQq!AxkLL$yz9e^tZwu)^6ei@FoJ$Y_Oc4I-N#{P?#TZkSbo)3nKvdGDzYY-~o
zsBGgxN>WO}TA!{KoR1DZ`sHS4Wl@)+h(pTpjTE}1f&Z*n4@qGTuL?Z1Jle2c5qZGT
zHwSrFU1vZ3c;)gzgBaNZTeO`iv2i#gm$7qMVf`t3z0T%33}o)fyn4r)KHl5n)_tMu
z0obf;-I;az-ycaFgXl;}f`MI58N3}HGZT2SX@}3zMBd|v_DsQf&+a+oz<>IC@rQ?^
z<-b%Ny{u`HAmN#r%zY_d4m4oMd&r;oeuCq+yWy+pPshHcS~)F#m-fY%1VR6LiZ?gC
zm)mWkJ%}yE05w+z-)Ak}E^Tdo(}drLJ^cc$#^`CEpARA8*`6oh<0lXhh<%=@WUY`C
z>7zn9`l-A&n4$y7{Q^e!NmIxAplV6l7+#BqH*z2h!lW1<_}iBu?(Z<95=mjjF_Qip
z_s$O+E1<QYv+2Y{|HwbAFS8I^4%I8`8UZd(SKHFpKmgUbQlZy+FEXs*($Az)WY4%Y
zi}U<kS)w)MqV++VAPj+s7bw*vz$(-&)$F{SMsU5Va_xQJYF4M%(An@X^|tfAz4P{u
z`}Qzeach?G^>X0x#UA@W@8V*cTZs%+ws_!eWc3OY$%O0IdiQBJB{VQ8)J~nfOwFz^
zUKpMXR8+x*F@pKFsx0DeI<PT%*x=H4C`ZKkKU-%jyS46Lc*wG=i$PK9A<c%7+|E&3
z;Yy^m-#-->LggAyuk+JFCB%=WdY|Ji?Bfpa#wA;+^V7F7M7Fa_Mo%d+w=)RX$zol)
zeNFqn$y8-Zf8`~vS2@Qx-<<@1kK|^gYY@}|lPQw<1xqpP1bxr?t?F_#T%n@Ei%lr9
zz5UU$e^20$*x}k65nc7#@|8uvduEj`SA3W;)2h1?DBB&2x-brcjv>{DX@l`9|CuZ<
zT{CaXU@Z}N+crQ`sw-d^gLn{)*?_kw7$bh-zh@XWICS*kI?~bL0yrdgIwMFY;QIt1
zMYU^xCRNnE@xHrwgJ#I|2(wEE$R@=d{0&1z?<p(VNA2Qc9WMh--jW$+xwJ*pO61Sd
z|Cuq`9~^opS^oK8VwC+QQ~cn0LP8yfNG_>ueNJJ!`u0eDY2p5enW+0wv6B1g|3|dO
znx32=m!7h%80F*WzkoRO2}7sF=>bVKyXu8~55I<{|9(e_I$!qQoZIBCPgZi6!WE6u
zWZ5v0AqI;2A0jyZW3J~{tBlDHtR3|j7CU*+2Jhmif?)-a`V+SBH2$|cHC<4@Ni<RQ
zDn>=K>GfGM%lHB~n=XZJ$=vsG%#CoIb!<HqEZ#9CF>Y|N%1?UOc9i=!E<i7r)aXm1
zvZ4{$e*O$Ky4!G(kb*;LDtoO5@6LaU2KuO~lj*H3EHqM#E?dFr5~5J8v_8<gQ3ls}
z+x<AtX*R`dX8aIHC_c-JC`No7ZJkPyo+g@^1sz0AeW?PRSLXe=d$@uA=ABOAasga3
z)QH?Ng}0ow&3{;IJ5DM|7kG0ZR$0;Jy?uq!xa;8N^01xobljewMm~E;<@^InpU@p5
zW=(rMtA1(|TUwNwrF^cm%jO`M<?oM;de3|_*9ukOm8LQ%9Wg0a!<HqhQ(S&pUG`td
zGFD5sAme)VioN*94{wxNNh?SE{HXPs%TJbFn}016g@8ZLaK=diLTP9Ze9_BZ4`*!O
z#3l9gysj7iGmd(yEft2L3WWWh9{i`jlugrxhYHldqdxGeQ4?`DO<QMdsmQs=WuJzB
z$FJwK-a-*SXCpk^Kb+c5PL=($wySe*NT1e$riY^;&Y3*eiiwPS(A3vWlYnD?88nc?
z+0%MsP-fG5jzlBigvvS6ahACr6p&SIPt&vW+PLj6q$4A}*>$(k&%H52l_eigX*QIg
z^Kgy7|0bkQ5SAqqw6VA=&>yL+W!~BXd|%1^0?Wue_V()lzuYGE4_m&vLhWBd#t7N7
zEN5vtj2>REmm$Hcg9Pa4y<~4dECfI4P7r(;vh0i)c^Uvmb{$M#9(Q-bCa=Att|4e^
z5Qp67L2d-iByex<`f=UgsyG?8ado(mNP_^@o3{<pULJQ`A9XF$<YVr6U0-+HAe^;(
z&q-yjQ7rM%@?+lP0Ii9n#M7hyb#CYBMfHPCXU=4a^YVSP90S|yeekxvbgn~04wv~C
zheFTnOyRH%Gb6&nhoS3Zh^NfS)pF2A_4GB0WT<ObM)d@d_p{=(3|r{D4F7r@EzgcV
zDMpHs43Y3nFN;TGK+g<FAB;!Qofbyar4`>p`9I;aJ~ve;t*H`eKYyF9XuJM}6l`Qz
z@Yz?DNy<Xwzux6Z4@c`Cu6q{;&AR9ALx!oCu%#-YgqFg#@l=Cy{)N`NJ{|q92HYc@
z<VYn>d2Nd~aN&o40_N2-FQt>uZ%?2c{=lKo@UF|`F#$G+>7-jUAy7sT3pnQ*8uysh
z3gn5oFB&|(C|9>J;3<eGV2f`^t;lwda}D)c=lK++#x7t~N~mCiBd2g8xpFSmM+S@S
z731yG@3!OzIjZfeIl!mY+{hzc>E!kGTI$s*8Af6mh?B`p*-eGN2q8gbDai~BAjA|!
zI1vI_%b}+qKe1+;VNh%UsIgm-ZFIbK5VkIvuhZ3aSJg`lRiY9q0^?XrljLl?ULmfl
z6M&w?4<R1HmgvAR#8&S}GM_Y|7vUa=jb2ozuE2f13}qm!*`bnS!Y?nR%6zQ`ei7Px
z&bX$ipt5zt4-UpdpJZc>Sh1~_9NE?b{p`(^Q&R<N$O7mW@NQs*oQbS(xjk_I5&Kxe
zhRi1F<kwqZ&Bo)RRI_0oVf`vP?QNR$!+7j9O`Ca{n~k@~Rrg%xoMz)Ssy^jSL+MAW
zGnBdPB`LS5%_Qe}Xz1Kh;QdKgpuRza-=QV#-RwXLYbunJRTIT{^C3kVz3MmMsDf3N
zPXw}r?3?ceX#j=fKAK(HBaNy34oDn?g_%BM(^m%`_nC`Y;AiHpSLs6F?X5sx6F+4m
zidfNKTK>)8)-_Qu9+gY#qLkB+`_3vgSGw^YRq}V3cX$6Hcs*i7()%H0W-G1qrS(Ez
zdFR8}zH0mOA+^lqq;6Q3t#hLc#Ra7&IX2V9t|&Ed0A$*v6b0RM|C0H4-PzBcE#>RB
zxZCb|jyQ4&CP4u&o2kV9t#@BZ{~C>n9k_L0dpG_W+xfeGc0YKa*XrpK@XyCT#brK<
zBDM~1fpQ@>Hg>p~E+Hv&d}?lLVnfH<dwUdyWXls>hJ;#>7<<Iwpw#mJ(KBfh71r}9
zIlRCq1|U&jjIhzt#lYK%JXEwBr^U`vN1!@J2PszLvfuXP8HaXDL;C|N>h5%Ge%^gV
z`g|VMOI0V5+aB})+liI%cFGa26Q_xvZ|I8_$h%d^xA1q^f4XgaI?5OEV47xp$kYlv
z9iqKgG%B)#*=>4QK{EayP3IZN=J&qwP<vC<-Xr#|*rR5sS~WuLt@f<-*rN>{#NL{!
z5~`%A)f(|d?Nz%didt2B{h$AfU*6_bo}BZX``q_+U7xoupBY@T>7Su2A1}NIPABNv
zEU27<j!uqWge;qN<3H^;GAz3KIegRnCg19TI&(F4Rg)_qZr^4N>RL$qO^Dm0CL|Rp
z?M<P))x*qFHmna<co1*pRY8KM$^QYJ7x7Ex<sJU4pl?DNmbAe=ipi4o6~U>$Nu2R%
zNg>9iSxh@RNh%!MpMrnCqS1^)0Gk3*`WkHE>7|lM_)KWj<9GA>29HervzT>A-9--)
zsFxx~7+k+<`AOy%Jb|fhr#C#8u4Cla2g97hGA>071S8wIomW@SANw4gqc_5pOD8Y?
zxZgT5eO3cEJu{(4+o##W-7uU(QYi-^l{&1VV<fUdGu0u~%ObE&n9S@yfPAwUVGFCu
zGKqLITt@p^O6RNfCK9Wc`ha-46g$kkSre3P@=_|9inMGPl{8FmAzGU3wHv5q$)La$
z=<vAAYsHZ6(LLAt)}}hcq#mAT#2bja=WAj|WTNXc20=OIo`x49?&9`)6vPQNBYbcP
zimR%m*-Y{y5iNuA^R7Q>0BYjiy`14m)&@Hoe*HVS5Mi3MAV?Mpg~_@E!XxXr20!f>
z`uT({91FAyWuhKWtyj0ld6&+5$m$s-FtO+p$UaDWD{CQfw~VBHTU+Xl=5uDx_*JM*
z={HGe{LfKYJ;+nYE~{(nm!}qwBO<R=PC1%}<|1H($l@PIqj5*CfY%a_U+0}}47A@K
z-`>u67)|Bh{eDmB@>1sL=#lEtAAd;}N@qABSgd)G3MSqV#JYMG2Wmbb!Dn*L8q<r7
zLxUXMMn;fK!2;<L63mt!fCY`icoPX@z8ytX(c+O*WCa<-PLzB&Dl{E<R{Vw^Ddl)v
zG|#!%_UwW4Mj$*72wxwpi~gQ=vzq7JAd~Jq7_|02Y;*caQLcMQ_(sZdwDTvazEqjn
z!sf`G><8yddv<T$3O-UDs~J&@H~B!EaMDeDs<UDY+p99Gba!(jLwEgI_<q&JZ7{ot
zB34ooGa{^TpaI4&W3#OWD1%5`e+89Xn7jMgxc5Wh{El!>K>cI3OO^Q?LNko&i~Z$=
z8q;Z{nasqsZ-EuE{0BnuH#HL)X^P?ZO0$rGv*|E{7M&T4Vr|&o;d&&spYaNV!TN&~
z#*1s0&_DrZouGz>B&RZ4ij7La66NSWX$^fH2QBWWC!fD0#I#JGEf`4HU;ozWC%^x>
ziO932W>E<SF!dc?ttuNgM}ns_KjCs6N+jO+<GXoqTSu2qzvscKbHC4;Zu%^)eA6?$
zr_nhTeRi;t7v!$HMgmod-uN@Oy?ymh6wjjxF)@)j&$<)TFhQ@!mXY<Q>u&81>5KuL
z5G)7YT5JCoR?k8*#b{b3=QTr%)D$`ZHOfcxpiJ2kP-ROgwm-Wvuz$ON$@0<{cHi6x
z4GX1`;6SJb1_y=g_Wb?pAp2u>IlY{T<Vo(6-+w~Cv|_-lul2im6!Vcbou@;7T$?P7
zY}9RU)gJTO%(Es)8c4?{m+><m{?^1Uax^-NC@--i*=HW1a=2KjIfe;7PozsIA9*bb
z4bcNE%aLmlPp#dadeP(mN_K*$S`38P#^b`Cq)4M_G1^)_;5DHJ4MD&)NO?&-Y`;O(
z-=W@tuy}%~fN?S6cs$iWoqXi_o(E>O;zaGc+22Gu0VZQm@FUz$!SQd>LU8z1AP%cR
zAlX8vFG~9=07HuS=n;(m5!_6}A&8G}4}=#IYs8tr0zD3UR@QR6a7!`ODu#3gZ{J9{
z$uj--24<re_%tMhm$*NvqIf4&qcZ5QYis2r_~fgPFGjlTm~|35FIG5(@&Yxe#fCXm
zB4qV5O_&sTpgAlPW-`Sj#&_y%^$B(hVF7OC3Kgh+>AZr0pwK(qFZH+}v8f0HK<iFV
zx!g1hju`^a(?<zsdS>zr!9zGI+p))&kb!{r*kn3sioUh_+8KAR^sQqICZav&Z+%~e
zID0`}fi~Dj#(R<E<p1kGZto$7(Sm}KC(3PS;Dye|x7Nr=7WoOSxmIm|J~C1g9Urep
z_ZTA^@5B%*!!wGIO{V$i0XkPr^RO(K*U}GUrw~}%D(@ukYSs!#I$IBZAxES_mJ}Do
z>vFu04zFpwzv4S9OYm!YlW;>4O}*1-sn%>EBycuJ{(e{uf0nUkjp-b1@y-{Va#!&$
z6=kZCk-wt1zhUw9x)^t4L+Iwun}PO~QhQON#MOKD*8_c&7ZV_C_X-F`uQD2ZTX|q0
zei~V{q;{@*YtG#IxkH2bFk1dO4(@CWWk*ESjOCh2r{gM&yxk@LnBjv8sGYDc`YfAO
z#h2C8Sv?p`QE01f8d&J_CyupA&C<m~$IAB!v6`N~<qCg7yPywgo~#I=Y1BQ_2`EOP
z-6-PIm`l!>==G%{;8G;4Yvn7`zw6SO9cclmySko$jQ7Kz#97iaBFM%+)ZaBKyBCn8
ze~$Ys#qEKWmdgEwwwbxvDbLTJ0#7GrWoGK0T>hS8C^zpvGR*_~fESbC%qHrvPb~%r
z3!8Z>J@5uhDw#T!h(0LjJhzw8vEZnY6%^o5im?^Z#-j=a!g8{+IcgR<!w;h>tYrvv
z_5b<%9REd$NC1Z@-`1n^dup9c4G)+v4>pF<s;WYp)*_a0a}gLjPKqy&RE{VA{rgvU
zGglYAc|3D^suZ{XOI-_%)JV+~9Wr%B-jDcXXP#&7K_j|WD8&N{A|O{jJv($RmhfHb
z2pOI8Yx+_Na4KD{=*okO`Q+q2VKeB4f6c3FVPPR&-1*Z#^y)Wwu<c83@acd4+#v_b
z{xAA1RbP+!k1j>8=>jcnu!)+v&0@8dmcSN7Xy{2q{jc^iHr+QgU8F|z%y;H3vVsnt
z%V|yT;pD}a!#93?Hx&CF+~s$(bUU%kd2^7iM9d3FCbHs>b8+(#(|Kb^XmVbB4T=(f
zYYO>J_eSW~v)LMKyo?TnCJ#$EqmdMA?&*SjMo%SAp7Jg_h2>>X1tp;i!8e4Y9;b;O
z=a@zu#MnunimtrM^xDcNc>HxaQLL1&&_t|WER92<bdI*b!-_<&c__VNx2mz0ck|Cf
z7<?5#!thBF6TV%ej!~@zG4*46y^K)?>ea|VRTTx9-;}N&J{c-z@S+Nf3zR>}=HOO_
zmtrcG!v9vycwtMDg6~$YHPyD+eaII3R<y{D4rUWvkrYOGO_&80nxsU^m<7?#epGwI
z%weObAg9pHG*V^=*N2p5eX0GgGAf31DC{HPL0RSiioT8EqD&aqcQ_(?TIpWNu@zW`
z?Ti>>HTJ?BEj{~GRCYUstXF?vF(LFt;5N)RTOa+9u{ZK^K0vv+@F|-@#BWf#u=MP-
zX!bmBwk5Cge9wM=-|vxZfkPrM{E_Opr|wyWNug`5z7OpuEwugz7dPbO9G4dcZ4T5+
z39+~$4afBc8l|L)T({C$<xfsdqDCmMAfiF+B@n^HpzYGnjB#`sI3{PIb>#D_(T613
zvRU}XzZS~O{rAZBthgY2_GU~Kq*mN)@b@$s+Hah7{)W2hc!)3l^1Zmyx=qd0uF+|4
z$JV#EB*B$9ZdY?UFQgJ~f<}rr7JypJGH~!yqiLq|+2tOHjk}(@e57`CHC1hyHBA)a
z<Fq<3<&DTRb1#PGczNa{yW+l&w_iWO4+c6z*1C60-?0|#_;p@~Hcf1uUNAcNggi)v
z#2Vv7xfPx!L5$%jbM4L_i%MsI8XmDLH2QDuJ>O(7<2U1fZFT1-k;WTWb>JyopSavQ
z96kG920mP*;M2H*6f}JY7tGUg$d>GrfVx6ArRFp}KbcmuD@k4d*s5^!yXxr`T=V#`
z)3|)TLfvt*b4wVaU^>ujCY;5#k<qt>HUPJDwq4BD=FQ3Fa`e^J`Pj!Mm`aX*74jXj
z?K(kz5bXrOuIALPE?Ad=>zkgQUkxM5I|_S?N?}aSYlqG9OnXi9)O%`C#|Cv*XV<5w
z@1dXXFGrphadw7z((ByK%naMd7>nm;W+se5jXs&s7b}Nsx1TQUoDg;1VC$||I)Qf6
zqYo%y;RnzPbDD2*EL}gsmGj?cC3ef8N%1gzf?BY*#?XJs*DdG3Y-KU1u=k+8=<ZEA
zZvy^J`(0UfwzRZ1bT-|B>}ple)a}=HNVL9y|6b!N{dVWpcar69=ERKOB4T##7rLub
z?Xo{BA8TWy7O^vrod5K(;#&-l)a%elV02GlEs<w^urHfRVwk<Swm@`=Q}FARMlL1M
z#ZLz-s;mC^H3py^i*0Y|P6?y>q!dO(S5pqf4z(Y71kNau;@cZ<OS*kyY0MjlF%^IK
zerV^$L~{5mBO6Xg!<B@AeiR`Zs5o=*PQ-|dMB^b1%=H;NgZp>H$7y^#0!boKbQURw
zR_yhFZ`j(g(dXGT(?N<PnArf6CuW${yd^}q2KMBDZ4W&ETO6vyrs4W@tzVCB#f<25
zo7Vyq<)UH3U*jvp9=>hFWxwF(`aHA4obZ*2#%oksBP{f!TtixB;y?wtIz}#W1+O@4
zAT{LYY?Pkz|KDUD^F&AX=!m>{XMgWczo85jGV?GT<a4uz@BR7yH2YoAiAMgL3f-I?
zEI{fkxvb)aL5|^mMFu+~^(SQY-^iwA@(o{$DHuMe67BoW)Bj#QNhk;!R<!Z2Yb)rH
z;X`&GS?ZzhY$bWhF~9zs4MZzE$8~GuSQ*68aK(NlMfR0Vdm8({6lb4XnuB@lKMB9m
zY(ZQcbnm1`xW@R3KlCr}@ZVS<Oc0=WkN_~HI}26t8LfHLm<yw_OuhKA1JNaEgwEd}
zR^%`?O6>6Hfa4Ep{7yuzgc!T`{ddXTQg|Y;9O(%-Eu@NXBiFq&K2ozeKJuk-7HQK7
z?b;}l4=D}42PH%^=6xyBbeQRm0qyIfw$>z2feTE&HxwGmY<elKpLM2Dik=HU1N=u>
z2|8{EB%&5dZ@b7V!Q9%zDQsllqvGbseGTWp=Pixq!dS$vp<Co-qzLt1U7g<-yE5aN
zsEwqYME^AyM^XZetZ&U^=cj|~dWYb`Y+hbkT3XWid0q5w=hLRh6^8}R40>q-&6e=S
zgyCgs7KN}7x0?GDt`+78TdU?^U>%X1uaq&4uu)>;w}ly!==?4XIp{hX@u^jTg@zv<
zl=W)<3~OC9q4y?Nhf2nMrO8ESk<WaIG0s>2x>o77)@)xb`>05n`Rw1-mKpufVrv-D
z+D9d_N;fyVi$4bK7In*<syAnA_ausM54yeO>+FH!shlLNK}wF1YUtGlfGOx$Gn)P&
zuhtfDa@E4ym)&uh1GGzazUoc02y}LSAbIk4Bzk3$S~b_k>Sy7ieaq>bcPGtDU43Q6
z$nE;6l!DG<o6aIv7Cb7>ClQy|)GV2Iv2M|$-c3sE1HB^l5h44f$3M2vC}QtJkM}ZL
z&*Jd%h1SB0#}$WMSSSmI09?4TCWh@Bj2N!}^D;n~GT(cq?rOX4hKz^9UR7bNFtqKf
z7(MHd8$CZ5cAyNF0GB6=oo6p*KoR<~;&0`ns*#&55Lx6<nB{$ZJXBP7^VbjHsVG}q
zx*0ybZMO2%akTnyq=^~h%v<^4UpdK%CL=BP+V$0Kr$Srj_Dce;3zTmR182mhJ}BAj
z!s0ofbej2lpZe|!{$_@lm&($?N4c=(UX+P}I)oL4Z}<jf5#AULUbziXnqu;94;Ud;
zHq#w?Sr)P$mSD*9@3Lgy?{U2+G=Q$qr)d<3NJ1R&L?xYtXtLp+q!R=Gx}LfQH3rg9
zHa=_&Zk8B>zQ_aapVw9WY*cuxt=qbgKJhfScPh{`OPf&&l8lQNAuF!rcdN>fLCXD8
zjHdiFUGv`D(^ZJKLqe30oKGKZ{+pzUsTpPZHBIQ=qUKMv%LU8r;}Q@!PM-u^50~)k
zcL~+WZ|lGDs%R-S)z!bh3P#{2gAS(oosbl3{FHk+;Pc;-MJeRk41ABDf(v@ea)6dS
zll*tq*~#5@+5GTI57d1!69W_+RXU|d9~O?asJibJkZ;l%`1Jhy_dlQfz0u>r7+L-I
zcYD_o_I&fxMOloX{5`@la#C>~2x4MX;4@!0&_Bi2i0I1@V;JnkV1}Id-)BAVc^={F
z6>#`*41oQn#8pV@>qw_J$VwrEQF?iLMJ$<R*#mmclwK8$!!P(UMf<fY55seNX!&bq
zb76tMJX#mYb+6ef^hhLiNYNl{e4w16tIXDa#dw286QzNXs*~2KM&d!D-ulK)A@0!N
z_XMyNgxOuYFz*~JoJ$0z7r%-+H27O2vsowu)Hk%We7=x8h?lr-x4+2MxZSGEz<+Gk
zaJRx_aw}=wYYN}4HKK*hJl+W}-Lddxh1A+o!$|bSWLiVkUr<o!yT8A`TkBV|>$^X*
zGvr?KY_F%k&!|Kpa{C{%CZY3QHx^^$Uiyk|ShVcEWeo`)L`Mk1{T;75^P{|)^-kjn
zFwncIQFjkb8up~o;hX8esAd^WGX#tUj<{M#L-6Mod&FXQSXtFcBw3PQPwMc7s66sy
zCRozw^pmBIY8rEBXpsBX^Kqs#Vm)EUQoKkpMiX|`?_M&$0QN2WD{l{$uW!a*qkTK;
zuyadGnzAwzchw@+h{2XFauj<^>0A$(R4IqtXttX^Ywno#YEtgI-1e@E27@ByRj&@P
z_E*{%mrKyf3IMXt#S7IU_fLw9b{qhA_w%#*CeZt}*?Bd$Q`c!H$}l17Y4pfNxcy}1
z811YpD9F+v(~z-twIbiCpw{WXGrTL&elalbXE*M-dKovFEbYrY_(^n78W>I0QRu?+
zCaB0#Z!ryhQsS?Mnq4jHox)<%`axLVNu~YG-&Vg)VC|&nrRtZ*7`f^DpL1yzjuM>C
zD=w?&+gENJa+#%NnqYtW+E2hOaG`dirPHn9ed*V6dV66P0?VjQ$F-vtzl}0&_v9Iw
z4ko!n3A6!O4)cMjDz^|IjLEe64W9wOKEsLb0KN$7Ls17?A3+@S1q*E)q7*`pNAud`
zCrCY_awjukc1_dEl79b+Noy1b;o?uk3m|d#U|(Ts6i3(k*CfTxC4Ja$U07f6`t_AY
z%<(IeoX6k6k7Hyc8iUZa)cf#%MrvP_Qd^%38IV*GA$6+bgbX-=JUtTAnKUQQ8q$G|
z&pL2tk`N&YhUfZ>Ba_~k_S(-*k-!gj<hys3GDTa4N<_A}HNMv#LcOvR3NsZ5k!*tw
z4TkV|;(rdXQe@1^QKv=qwP@5o6X}Sfi!erNvHJxfdFaKM6s1Yi`Q6i4{GhfCa~YIy
z|Mxv#>c-2S-b#1Cosrpb+DjJ4Ll-ML_;&a$9BV@4D!{9SPS+}6{!SCb<yD?;I9_jR
z<vp46h>f*xO@m2i?Z0~(Eymb^Jzrn&yg6EGw<%Dfj%a5mq>Yz;f3fJ>G@}r7@f3`n
zm9)(*DJjgIWC=YzaogB{x-UA;i}=s17fNdsY1XeQE`<a1ltk?sO=4)x$y63v+PBzo
zrXsK(2IkNZ25K_n(>SCiOPN-qs3nFk!4vs(_-09?-WZ<?uK%%zR=iw_UJT+$&`hW)
z%H0*zvPqZHTI+x$(>3lQL1#(gJnVjW8g_K@XNa(^@WhWL)4XZDVc+}pa_37S*o1$E
z?HG`T0}8HtH#axi-6=i!mbXjGx2?n~%L`8;FD5o+80@bPDvYWvG1F$uDDrO&;C&az
z>L;5&R8+Kx<{hlHkN9&Y)OXzHSoqeYX&&VI?lwlIP;;t;?Kw3lNGZrY_MT)L0h8@E
zeI30c?YZ%Wx%x_N{u^tHg;>boWD&#Gq79mWkO$|Vvy6$S=DBNH`pHP^8FpQvO>rxh
zewQyM`2_`Hk&PDO!Ad?Undpiarxr7W4PR}^kaVU8(aw$gAJZaGcX^pWX=C_;w>hi(
zNKuebOTan;J$r@Yj9SY9u4k67R;b+sSTc|2EXMq>h>StT4R>&Jiin8F`z>Fi>;Z*a
zLEzB6f&2(;>DI`(f%M?R-+Oc2z`^nI*>qC}zboUJUFBkaQPGKVld&im(C&k&>o}Wz
z)OmZ;8jv3qd^CN}F6Ml0su~sHD*1az{5$yaXkCF_jo-6}<0@nZH!=JLJ2nVc5s|R$
zY^r4-{8UCl^}Jo})`Z8IU<My3OEiDMP_*R%qv+cnjq=ImCy1jd(DI})hkU=$@N<f|
z=^X(!Qq-<A^0$Fo;^m8j4UEj@2$fE&iC3P@+yW@*zFr&bJKbMYI$g~iOnG^o;&<)o
z?P>sdY#L=+@r_B3_hW~6i`|d(?R8=PC^kV*$oU}&vV0`{)wJwqHa<}GDcK-hq<O8T
zE-Vk~Qbkz6$ZjAu^B-whqO>eg4W6c$y0DB7vRh%zki?N@jOn8m>m4EmG~+81FQim#
zJgF=h4f`;Y#!J(jBUw}y%}4+N?PQ?2ItlMrSS86EkMw}?Rv}(N<&QY^H}CndMaCJ@
zK2SkDD4KvNU#o&Y(HbaKPb>Z5h4mS$)Wh!yZ23t7C0=h&xL$l~Uzhz>dNQDOab>cH
zVBKO%{KqI>3iLHPp#H0=C)<Ws2AtK%Q{jM|u+!Wj>C=zJc_GW90D;UB2KX!TdjIfm
z<hBML!r1yoi)z$>woktY+g?+oo=1h0qe5m+b0NcXLYxAk%NT$-;>xG;B?p-9F7)52
zsNJ!Vg}Yim9TnEXU<bfNE^`nJDaN=R&w#~;gF|v^%i87EBW_;bdN2V~*MyC=V7fs2
zjt8lRzmT)^kZ)A57)45AcS1_FhB1xZ&ok|nh&4WamvUS^)f;tl3jLHdUo1k~D;D2h
zJ!xDIPO{zuAmE~#g_7ARMi2M0<Z(1kuVb|}NkmX&cg-e*GxfcUiWLth%|<}q^ZFe2
zet8)i2`^F-qE4WR8Lz@26|%`}4#?D*xU~`p)jyKJ*9WH-1_nu74S#{z<Ait@x9)F$
ztvEt{7RnI4L^@<aV$BC62n-AirXnyw9}_r<IaYv@^R7xxU*&_SoB$4j-Uml1)AIYb
zVYmNEv=?FQ+QYvl-AiuGTc&?3Eb02B-}lTvxxC)p^$}V(d!!P*_dN*_Qf`j9H;4|P
zhgaO>fS4MSjlCvf9NivSPhqK}7@m`uu>lQ46ycnDj4wa9+WlhUdI{1%;c5}AQsMN7
zWC$?vW=&s>=zj@1<F^&sY{?LqcG>hFxwM%INS8EvBZu^CM2i+p9>I62F1v;px6wth
zvvgm;KiJbZ<oMf2``>|*o7s|);bE_$BE^n{*&kcMClXI$z$&JsIIqYPORf4PWK%w7
zr?b5^aP#l+&b{J@z{tqUvvEiEN(@FU1)zKR>Gr@KIOz%+q5jfNts1`ntACAkXD1=%
zbUbn&>>GQ@ax8b8k9Bov26<h)yghsAJ(RY1y=&QV2JZwtwfiSkYkkm4Ed$5hy^V~u
zFYTe526b(JyE!`;sJdYfMWzP&8(uB~XOfS{DF+xy`D-jh?>bi;!t}E?&lF!I0c-T+
z>$0!S&(>g<d&?cz&kgeOO{2`{>Ps76&p_|W-Fsg9ZGaDqt+}>4iAm>TJLPn#H7YV9
z5KNj>`y5q1zo-8<jWPlk-eU2ZiHumL>Q!#QTk~?R25l-mPxn_AA0|%;>3*nsj2KPY
z%$;;wd3)Ck(X->`(8Z6Zc{F5^fRv9I!sTnDHeaIuJ0z_pG~FZ1TwunS-4e<#R5p#k
zH)P=w5P`M*iWIlf6surNH}N1qdo<qnppC_=SfJC&4AZc2rm!Mc=LS-`c&TvyOfO~1
zMITf|>z1M!NfGLe(D#$crd=drkE?m_vbug97d7HDv*wX=?26sf{j|pkD116tnC!uU
zDjBdc!3ku_a%`Iqn0^zafR&|5_5AoM@)qP^%M9k!cC2M@Pa#N+`Kz#;x&+s6IlKc3
zGZYssesEeE$W_tAjeO^L_q{!6^W@u2J}12aA%pfObJ6eS8_k<uFY+wrSW2_VYEhGn
zq@!&aucd}_&t4`fCrp<E_Z574q&LEAC1iYLMHswWi!glTNd}J!qiQC$h*gghG+Xbc
z%cMCaQiz#8MDpUx(=2{;eaoC=!osinQq1z2hvoy8{U33{%72HF*#a-<G2y#6#cRzJ
zwgOONcrLa_vX2{5_$5BbS#r=+Gga!<ZNLC=JvQ6)o!s7^KTlesfFzlYkN#(W&dmk|
zGeW-<4O)r|p;}Z&cH%kf?b7?X6Fl>!xZ_K|GF`vqv+}0Qmwk9-J;GFh{B=9cwDydR
z$ppTx8CCpau#?W20^N1<_TP8>dIC0G^KH4A(?c~pb)4PR)uAbR)MFYcP)$8Oi2m!+
z@VT>HEEN`WrBykx1yB<ezK}#dDp9o*wrE{kT3Ql!|MRQ=?&ipo+}zySX_mwY9AZS`
z>7PF_x9OZN18(CyWnE#UuGx@Z{~r9OE#NZ>+4LBXi42PQU@MdwE<3kRWq^Jzb@H0S
zAY6dK9Ul^%)9bkT>#w8NvA-l+dfycMS&Z-Ju|kjgm2R6G$q9gJSaWyUbe_Yo*`p$R
zRVgL5<Kmc&B+dG$(K%+S<b<1_4H7ZFd!udRUSW|Ue}Yd>!|r9Ud0}l%E%u1es|#K6
zge>%MrO0{Yo0z7Fxuqw$`GL(7Ub}w2J|!vu_ruJ=90`@vg`msJ$Zicl^0y@!fBD09
z3r>oZc_pQc9Z;M0Y-iz|oZ?UEvKZ=lMt4pIo3(9WlaA%!`m~0i0a`n9XL?tsy`{b3
z=!M$t3&w!**{efzH6t^`^}B|1E`Kjv?%;`}TWG|GdEd*JuTXiCxi>MeZ0SW650%i9
z)fhqBW!2v&$#;j+=1(m`L#dm0rdfv~sV)4LzR1Y6H-84cg#7LC6km|+r|U6-T?lp_
z_v^{GDir{Kbvtduo@+Ga^fC*-%|bYv+}YIR*7ITME!W&NGe#7*hF@&#GUz{#F)l;i
zHDId8+w7s!Dy5h80uP6TR0q`9vp=o$;e}FY3gdVPoLQD}x8gywhAfV;Ial<Ikcfld
zAzd?T&IOnwG0Ko7J9^xA3gu<Qlp*P;JcQ7Rbou&K>96O085?6L<L6ncFdA<pH=2z!
zU>+vcskAcqdD?sqq>`4#DMp5<jmiO}hpWrCH49#dNr1A7_26nfvaf)Z_5JLUMP0jE
z?PErwqFRynd`QI(tjAlO2h9%g54ioKW)LFXP~hqP4{MSYwr&4bEh_;5xn2|Li(z7b
z2Q7^HpkeGZS)WZ)+RzYGsC3zS8E>pZqQ^$PmWJy}B7YQNCzVN;s!e{5))Uqilgz%u
z@b_Jx?8Cc4W(mTBrao_{g{cXw@ioIu^8)C+kg49W<G!60#sb27R5p(Hxik}A<?F#k
ztpLrqh>m@wT8-c*V9W`89p0|+>~<v~tzW-tD&ZLA)E~sgc-FDcFgDKD4-K&$F?iF%
zwBBswTla~?R2UT*0^lq6^O__YJJ(3*3DtX`rafp$KpV*Uz?4Xq^h#=g@*ClXSJXds
zTRC3P#|nXzOXTxHo^N2Pkxe5(|07=tAo0!B-|i!F8s~VaV)B2QB;Pnl|D79DFz_Hg
z_V7TMa0<rRJrz4MwT(8Bb=Tu{*#oG~HPrgr`<b2UlelG{W5$hbgTaOb|5xv3?ZU$N
zLR>%GU$DZ^_YASxkT<+oT#AKE?i_<fIJ^ac8EG+!lYVIu*RzK`y*=K)nXktO`eHXN
zDb;ZdEPRPl8mT%?=@!$a^$4>@qBrX=Quf0vsqR$=Fnx(P&qwn;jOJSN=xL8<X1+sH
zNmn?IkXL2uQ56(s5fw9(;80y9)L?TjP$Fh&{umU-Q4_FTvB8`HpIA*O;51!nyJ=EW
zj(HluM)h_)Gbm_A^@{~2+n`qyXtxywy)OS^Dpi29S9XAv(7i5=Fi+3($mR3Bdfo=A
zig)7>*ZcRI8f0x$pS7uiV;t-DsP!*9m)<t>8UAya;guw91(rHBtHmF))S2eos?DcA
z!Qox~9xT!%>gN1jM8eA5dnI6U!)m-wZS8acr9sQV3)iNzmUWpGe-{kE>#U-p@K_|<
zE^1@7z|a29E)8*wT)dF;M^maqo2b=^ktLf}5fdV=_h=lMQHx>T4r&QFbrWQRJH7or
z*-mU~WBOjb{)<ZL3GrH5L^RV*-h)ATOOAJ*2w{dNBd-sn?zP|k(}g*?#njqLP^_l%
z;PaPGd_790OY_G()bmOP5}K_7<{U4UZyNxChrvhSE#KnOT8FH6VBycd=J~Zmp$f5I
z8=}s#9*ydp*GpW80VW=}7*<L>Z_*TR!{9MS%$Z`8O#Nj--`9%&;$rz7#Q52KEDo;w
zBfb)Hc^Qy8i_WZnV=>?wqT?f>U~={#WhyoI@O&*rIbDk9Y&`~Q_^qtnNM>|QbfFbO
zcn_Jae$4t>8gVEFcnF#Fy97A}1<8Pkg`XtT25XgM6zacBu<CGer%t>N+26CRfb*|H
zr!@Tq-V44F<zc+~x~>`2Hz`7))3&@rD)HZH>g$rRG!97cJ2?8y4!JYRm8Ng@`tlTC
zBDV-(r5fWj3P$1pM6YRa%=Z@fv!%N5a-~gs=_sYUNc&_nr#JpQt-^kf7D(0E`p%U1
z60C}ai};*ZYz*FZlcP1oGAf{~+MbCH$|?8Y#e@~<6a*5C&$q0+^8)4HXT1n$F6n(v
zr?=Mpi_wTf>6=21DU0(2J#4+kn2&=Cmk<DF8a5OoVCb^~1Pv(0wA`!4StwNlDXk$7
z(^gX9p6NQCyg^bb{ykdls!S9SyHuOXX*!D*r+;P}26b~dl*83&8XqSb<6yCp7B(;F
zN#H14y?@ev1k(6h3pV?`ZgeADNH_J$qeQ}zRAWHYneTL%*D={x-GGFfkp!U``)f&C
zlaM<eu^a>#OtJ8Cj-VYMDNBX?I#?0S49FK3<0QXw2y%8-X6X%>EG%}6f7yk{5_ER(
zw<_pS@=7iZ6lHMt?%m@HEt)Kd5QG<5>*sT`vwXhTDpFl+uW1k@Qze_tRi@5tZ1lDK
zwf>u)2eOQE707q99-$vpnd#XHli&&u%|^o9HaAIu3jH^M;-5s1<6~){7ryOW!s6+=
z-D&+*PE{^qwY8Ymj?Rvj#VyKJ*$Q(K@r#wxaeoP0WlN!%clkr`F&q2HpZBi9mSP$v
zd>#v<!PYf`9PUhC>6x#3dIcs+oZ7Y6+AJ@01aHlOIm!HNp{JptEZ7TwNm$IP7K{u$
z1cAjJYEf>_?Bk0SnGscYWLkxud!BTJTg;e94zG`nCy_d$D51T5(XBmT=L9p`a()Q&
zz0b;8?Ok#{S$PtAy1@uyR-=wLiWuaf?ZB3oFhBQOH8vd+Q&9_0?MR&P1K%~?)Qc2<
zQ`BrLt<UJGuTQ+jFw9EHLXm{bkqK=I$n-gPOmI>st%cXlYzNbbu8WSWB{_xzK4vVk
zjM8~1VFq*nQ^swt8VGXz&(;I`c`4#(J}Wm+pXA-3_6#Ebtqy#>Ii6h5uu)@>S>I&p
z$ql(g=fd0~^l>y=TH;_Wpi9(Mk0wnDv7W_1Fvwh94?%c)KOK{Z)j+>}!@>lN;d=7Q
zvmi8$sf>;F#R(5fELC*O5eIA-TH367^Mm-<UTFz(9s=6J&G7rfO!!tWTB<czAN<EV
zHqC|0;}BV(YIrTpm}`!>tJgS`rJ<Q+L0@VmDj>X-T&9yP@MUHT`2`%#HcYV6+)*@&
z+M6WnU2%|<e$H6v)GjHf^vnYYh*;Qp-1b+a^}!P((#NLK+)R~;zXNAy+_XFy&tjOW
z`nO2mfx&YzAzs_=UC4*Jev9Gl=Av$%jM~n50ti=YmO)X>%s_7-Z4G~HA&Gj2WK-1N
zGiJ@i={ZC-YM9U$lO<&eQUxXvo?@6O_#iCfEnJkVTp&>jC*(*Z=Lr067JW<_l<S|n
ze_m_Jq^n=ONV~vLE7-i&tbsPK0`*XDA3H&52`fvbU9(W84<8M2sD(<yoxID`{`fqt
zq0=!gFs-txHFtUQ=g*$uL!%+Xk@N~FaI(&yZLI~lgbEW+BlvQ8DMC0bno5Jrc%4N_
z@G&9TuIH`jkWsMP58MM;+H93GJ9tS^c4qluC3w+2+NCUVG^=EmA_=LJ(jbxg*|SgI
zL)f_4W<pdX4zATnR=yPfq;f`*6GH;qi)s9m{>!Xyd08BU6Iw3}QC!?TPLTk%?is9i
z1~J=+Su!)=et-zzNgKa*tj_)HoE*+2`!p0s-9kZOvCa!efT>vNbm7L%z}pPUP=nR@
zsgUt;nhwXjmcPn*HO9Hi8T=TBbS&nLh4P!d6M`TrL`!9b)4RcOz_zh}lZ9wD*^o&i
z2#7f()k|jfU5*~DF()~__fnkG9?=oFHa&11;&i3Qj1I|9`zC4r#iDL%!|k0}l^MEV
zD9z*@wtfYGn!>$7Xp)7PLSGJ5*gU2ssrPzu@p?!yz1-3+T7m5tHGo+BEbZ30nD6n|
zxAL}vvpr3{IO-xPZ0Rx}f6H*rcy2Ml_XqX2@uGxbjZL4b$!fLR^@1k9FvXSi>Y0+5
zy`%#9eKPRRj34vDKHFTGo8F5Fiohw}aIs0g5CpO_Vh%S|m**Cj7r^4ZWUp~9Qt5KO
zKxv*;sr&nRb?37jg<vT6^9d+R^gf2#h)ro#T95_Tpqt${+YK}N8hdajj}M913!hPI
zGkI^WG*;}Lw%^P*2#JMwICq)or7LC7Rmq_n$2{=)EByZ`DrcAu6v8X$=tWYK6_p3*
zfeK~ljB3?bDY9RbN{KlAx%{8tB+D#4R9OH#^}(3e%$hjOaISg(HO5EAs8}^*@1)(2
zxk9J=73<3h2(2{AlpA?J+ccd83=hwHlNI3oQ2fSZpt(;5v`%O?Q;d4_!)83Hw28UZ
zMU|ANc;xOssHUO(X_Xg0*f%aL!X~K&92D+2(n4c33}y78w8bep%;`iVR^O2<?od=n
z)z6G-41JcdI;4CoP=KVDso{rkUkDjh-E`vGD72o{hCRj1go=b#+UdcEZYzX+^LLG^
zPcE=jb7Oz=Y}({>go2SNW$CBz2&ZYf-3V;~I(Q=-^}yrDcE2R=iF+a)pE?@=F&C~&
zE<+ltyN+XzZ5VZjs-6wUofm&fHeG+l!6Td}ppVa3BlWByJ&W!ulQTqbk4!#a3KvMM
z@{7s53+n8UhAn+Nj;NGQ(+y~`3to&z`(n#&c2&I`$N0I7*E~FGEEu2QXVJ|t4KhhE
zws{m<8@%BIgV?fJhOO$YYAk_);?0&?)(>|1>12e&J|7<9U;(zeW!#g}MsZe}Ri;EV
zcq9lP<4tQpZrlIr*i*AU1!9X`gee~EN#6N+kG%Wvy@FiJvTu4H_a;6q^Br#??i0b%
z9z(Pg&qeJ^lET9U3W*7%l#Xb4MnQ$q+pNM(VdXIbbQW?;nY)*14H*QKLJpvk+MJ6i
zi8SQoN2rLR(U&bpW8G#9DFXi!69nnPHL9le-dW$5Q!ya1&MkQDsn`h%+gRB81})4j
z6lE}&m;ftq7+cb!ugw7cDBVYOQP&>2%Y?8c<=>wxV0V2r{MGtDNwQ_&RG3dc1;|s)
zcQ5%~I$$Q2o&y!d^`4@-RYp1{vKd24`?e2VEk3{ob?@CtdHIFd8kYMZr}N+Wh%GFn
z=I7m%JNyn9IikV!z!vj=1R`e~<)nRUL%Y`~EM*cD9mwupn87-cZug_JJ+OMpTP8>4
z<{w!5ehJ?izf@A1m-kx~LV2tAG)P2U427vR2l*Xs0>BDZR?g12HR5_tzT+HsX{ma#
z3A%62nm}1;>G<!&0JK8%DEVDV;O64gLY@jr-BucU-QAjDSu5|BpefgXaOUjpyKGmG
zry9IIK9r+&>s7}pCeC*I`vm;;*N08Ft94%zULU*&aH~$wYZYOYIU3xJS<QJ)oN$P_
z-0`yqI+mE-ncb@|7wZ`#9pMRqSUVzin<5v}m%0_cbx)w#dv$()un`%SOzFJjKmC2P
zgi(X>M+l6gXkL;1&)R!}*7;RnFKxWY7dSk44b>+@n+nSi<mTrMW|8*Q`^P`>l9e0=
zm5t`K=s#N%l`iFkN3?Oq`S0>9aCb9$-<klV+0+sgk?l%hdw6B*fu6o@8kO1c6<9oY
zE{y=SF1Kkx0vNhXQI_>Ze3OE#cwahuM$p=ugx{6~g3m&TCKT}MEgxZ~Vbbz^|1|af
zBm$vmXf9%Cmhft`ft2nABW1rI1z3{fSOyeaZH1W)*gZebn+YLi0@z3SQ%DDpM1U*3
zmHK3H-yp+?vt7johniF-dh5i^eyiE;sn^49kA>aJg+n4ox{Teq>!og`)}nShX$x;^
z#a5e&gizVp!WSu0a--0rmVUBrS`U;~f@rqB+eN!cnEc+p>yXz&{}cUoJK!|IBbaVS
z7Ed;t$=hTgSeYvGsVq|nHA3lH0}!DAowuOM-Kp#+%xGyNS6AgoPGFObq%4c<8}MfP
zS0W9r*%aY67hv0W2&IKwy>9cttTpSiQ4^KzTy$`r*{zJOnqm}J4q_pKU&%C3u@BPx
zi%Lza$o6PxY*U@eqCg*HULLcz2|SOl|IeQ&U&F>}VSY4=<%gwzPkGsZ5Tb5IZobn;
zWZxhZU}@BtJ)N(sARr0RgDP(etCuF4;I1{xY%viUuUxEKOu4i6cK4~fKPpJAToQWg
zb9g%;12u&1aOp9Jcj37vFBbQ|okF<gnZRB1z#f}twkur78)GU=(JcuY05pseZ#?he
zJ5!jv#9hm>3T{C&X#ogV`S^1gJ7JXYl;U}Fh>9Fe)pikf{-f$t;)&hz0ytN=K56P4
zA_QYzgh<hgc1;tz{GQ1q0c_MMU?%Eif^WW#rX+W>M#wHMT}v2kK6Y;l#+H)cFH4fr
ze)y4r>O^-%qptIw_8lL(1n{_ES!4|fn~UrYh}c+_!Q6ZVq-RSRMUqM_gk;=y*1JmQ
zKOu%6J&cHFB#$H1lO|LbWAQeZI>5^6i$=D;QiwC34p6Q#uQv%8Oez}wRxeX|vq7x(
z!#5202WpPnZr4DPP+g__hT73=PWcbL?z-sknM3jzp|c~<yj`^4$!rkl7vksjxn-HO
z!cgrruOt#U9j9E)61f{RSSe2LJu&g3Z(wMPwaJ210w>gVyfg)zUIKIi#syYaIk$gr
z{fkz&6Luq>{axgIlJ6@@L0td%YACEOlxUXN(r@M+HM?rSA%=4~UnL^Kr;bxU{`<0b
z@1msG<H}bNM#jvQ;YZ)x0-RQcyRlVi`Hvn-vr%3E(coLwB%j;3&gzXv(vBqV6W@IV
zV-z*y%c3~mhpbTBW%B0<lHaAL_Ng21@A@{GS7nP#MWH&InsQw8uuOex;09;Hkk|6&
zR%Hz=8u>Yu&+bsi26}6_kx{1UtZ8BwIR2eR6j`~SPVe_2R4>oWcET*k%D;Z&)ssde
zwY>ohO*{mKG>eUbdgn1He<M{aR79GSY3Bs!8jsLYxlbc1C<Q-%ecH<%=RiP-mc28s
z-!Ph6k2804JLAIEe?|sPCYzD<d@1BbaPg^ynhw;U>6st~ai3S83R3VovyTWhm$Ni0
zPLNR4KqfWfrdi?)`{Hz=|Jk)ZV9pxo{R-6iQ+W<ts$gWoSFe~+LMxFTQK-O4ac=C!
zKPQ(`&fad5_pV+KLBSN8H&NqHn7Dz@eIEVo%LS2wJcUa0j|le?u)h<g<{`;KMaqIB
z;eiWfTIpk;&eqwxyZiqxiewGI{T<F`0>@gYu^%doD_Z&mHqn=5-6j?|f-Wn7SA$pW
zcVuA@j)30~j^eAL0T=Xjj66k3o}u!H5Q@Bb|DERmVQctd8W)r*<pFo@rb6vbr6*jp
z)qmws(-`sONyw8XV<;3I8XAgQ(*vDtcpr%#x7gAZQlzwleDbG$e$BQ1Y<Lbr{4E}L
zAP~mGv5gY)Y|j3pl%V0ib-<4!IGPR;+zMiD+KY;cBq=sGHl8O178B@Hi{+N+SmAMT
zuC|!f&+NEcyCM5KZw&_|Db`>{Z}=Wg=!Pt0ZScf&un(*C1Ho$7Cb#rG9^Wjgr{L`x
zzxx)>f45AyhkXu{P`&vD;$y~-K;mu7(dMI~C?sv*VSk_f#i~IC5CD^x?<25Rj`BF~
z=?~g0f#&Rh=2-A!V`J0(@t$D!{HbqD_*y880&l_*s9Z$M&qn>2o8J$;ByLZz=$zO-
zPBx*}&tvg=Z(x-)>Qwk%%X6r3_v8#?&~*2})RUwPu_LOlynLu4rKV&pe9`-r*t;>N
zQSIsUj?RFnfeau7R^vX{cZ|6^XYYgQ-R;k}mnAH{?TpXR_BW_j5$c!XMNF03=lASn
z^VOO=TLGFP;40~XK_zOX?s~h9wer&<$IT;97OK!1J}HtVGUbMz`{A7=L*93LGDIyZ
za#IZ)rAvOEO6%op3)vdb72zc<AihqBS+#rz3g8Yqf!hMVKaD!3M`&mFRRX1Uld6Tb
zXm_kk>dbwmMPtl<`c2>Md05QmX=hT*+M~|PJyx}w-~X>I*+b2zyyUcB_A+?5lg{kT
z-WA<B_cW!S>-i%NwTP!39k3^<UH{RkcU?Nt>sN?o!TEL4Owt@CSqvwFy}UkHiM}(f
z_4B1n1{_yz*LH`B#aio*J}$SVfJ)qjmR7$<9u$=Le0s0BH0Vn85O;h~=E%%XgO54r
zdY_AgqJ*D)6glL4^^3p$FO7Bp%B-qq%wd2|-=rM+Tt<rK!5xo(!L9qsh3ADT*YB`@
zKeE2E)BKIsYrmLs8934swjbHi4o>!uqtjhLE~{9UM&+zgO6GM!LQzp9m;(FQln9QY
zh+y3jG5v^Oi!vVNBHWiPw|Fw_&Z|l2@A5pywN{fOIiNVl<o%p}0e3~>$y|~B^>_|P
zYxvgc+(xEl-DK8k;hk6I&<{iw-a|sw9Zfi6P3CV1njX{XwO1x@He0e98R7b|@i~Tp
zaNt35k=4l#-r=yd7A~j<Ettof9mJT(vEHw2@8R<r&fOMPDC107MgeuU<)Y!Ee;wh#
z*gN@O`?Go%a~auT{eC{zVI^la>D2Pnpsg{!kSQ-Ag1Oz3c7{xlO1uB-O!T25sblSg
zZw=LS!S1vJCjHa*MvnBs6{s5Xu*}e^4ln{E%fOjnXa%n=qK-9hK~8nJLDH_|@37TD
zoB2nofh+a8#L2@gSIj2zrUOP{i&ljro9N-(gX4E3ngaB9h}6{yKFsb+MzR}>8JB7r
z<4DSulr(=jKB1s`%}u8x#QjwP)21!tiX4#Cd;6Z0BTzsgvoHm+U@c(vu>T!>J)q0j
zM+&+lRv|BHC|o9eD3+p*k5>+>_{dad@(PTS8@6~d5GGcUM;LjIKTB0pJh<QIK{@yD
z4;u&{@vINXpwmUX<(LU#6jfkKEs+sXoKfEh(eX0kP##cyj+FXJ3&XGcuZ6cq+Mlgj
zVn8T@1}Xn0qI!%X%T}B*^YHgS;FpVPfn9U+;#0YH`QM@%Df?vHwDs`X{-Z1Zu>G4|
zrL|t`71N(fbafgwjfk9SSmTn-S|Nt04wcz)`uT7A`#_hXJDN<Jrus&&{e_{g^)ZY4
z|7I`oRBq4AY}%)$DkdhDuc<rbZA4)|cMIf8BKKxHua?5pj=Dyub=R;{LXWF=dtbbq
z+Ex*&@NJ#gYs@+7eR|)yuql6N@w%O}BPvKSuXK>(Pur6mhCw_n;72c|lk~r0(mY1-
zhVZrNCk9&JvV!j$A~r^%H!|h@I#DRJkVCNa@V&t%J0Kmz4}2kzxcyTgu?H4Td+Z~d
zvn4m@O`A!oVJqAHTe7uspx5MOM#f*|p`hqB+(V=FZ#Vr>uc<BMPpBS^+xq#I*e}Au
zE|-?hU%b3(y*-jX-P`>`#sU211!LiY&#T;3LnAje?B?3nGR;|E$UbK&Ih#Bp7Mwtz
zDEodY?TA!tYt141j>k@xCjx5K&ZbJ*GION9;|UU+hV7Z<*n-~x$0N4|)C4R>DUrZU
zJ1`>TEmtsH&~Vr`waZ0VrDdIsoG|_i<3kV6x4Q|Fm9#`X9@$7h#U1@yw_N^g#O3G+
zg2d@HX*4lX@#K#<!A2(6Z-IcfE(~Wn{G0K#!EqE{pT%dHEE|`z9~iNP?=qwEdQIHi
z8QFL_;>NwCa2Vk8WQ5G>Pbe=3JMRJy2|rDov?htErtIii-m@K-0N`EDe{!(iXQB$|
zT!i%0v<aN_<dX7(at@8wUw^=ao-ulKEA-K$@itM>F^olYJrhcYGlROEt58gtT63-O
zDDd*i!cLsDIe)dt?N_@67J^4I&>x&_Jz-NNuQcW#!3jv6&drTSUzy$N1MBXz{O=#$
z;CB7Q!DSHT6f(9GT7%U#;nU|ygd#s9M1;Dzgk;+5L$|u7KaUNw>=a5NEVWfYTs8E8
zcRH~y3FEWkO>67yDa5O&uv$HmGWzx~9vUi@O~WKdrbH1D@8JkOB=;#J(PBq~@h>7}
z8g(2OK=SLrrP;~_bXV=3D*1JsC(E7uof<q^|FrZJ<3nL&lu;v+pboJja|v<9%xrc3
z{ac-)E4X`7I=hY2?|#-eY_Y%|zF+8!?7TOUo{FN?bgiL=5G=E%k1LPR#s#?{iF({x
zIiXlvam;6AgYfW|IVTK-vU}#?;h}29tH}g_%e+vAppRcylX;!s70othhumGwz+r_*
z56ko7g9(I9h~DG4{q{PhtF_GT-CYBde-md`V;=WAPBwwi>pb_Eb<5kmu+F31?yW<f
zCowxV%Q1T|MsBD5t|gv176d$aj%Z!n<e|}2aCkTdN<dW=RX<M~ZZs4vE^fK}d8)uB
z%Fb1ftk!9uosM|+4l9QK8JFc~^`NVea7Ve^33*>}5mD{RuAjwblvB3=%i$Wqg8^sW
z%V)nf?U`yQK;;*Ma~tvDnezzWG0s~)M$h_eQdbqs{ts9DB2%)NoI5^o-i0k-uPQIt
zU(g<y+uf5Ed24u*p@Y<Z6!qKLy`mS)|GSvJD2du>U9>9+I$jm2tMm2Q<0!Ir7SJBs
z#&#SQaT-+~DG!w+eid$o&IeQ+0hho(%3X(#9SX&+j1$kiEF^5*9A$s}h}y-nBFXq&
z=w>{L3yOS4e}f&-lER{;`PUN49oJ1A2mRfK=?z!At*1iH_r&d1v%crZs2je(*CzQ7
z#{aWOD*5h`=Y1-p3@VO42HXiyqZjc3ZhZ}Iye(m;M!Z!i@L_=<_$!{>5I~WWUY2Wg
z{ks-iOzNc*{{y=$?bZopn5x-p)2|yxhdh%u_vGKr{yT6vUloy<z7rBxU($A!?ZVB)
zM4ZJ1UM&@aZT<Jh6m^AohY8EdV~4F)O~~*Pvv9Nn!n7)`=J=${_wOYA46@naFd|nH
zMG6=~%#aB0IHz=lV17dXQXnt2P~g9(t~UB4bfBvuPwD35(8=9PAlV8B<4+1H*2+3l
ze`O4R^q!B0il4w=@%yEt=?ni+FxzYH$?jt2OFtU~kH)aa(ZWwz6$8ZU<8@_rwd=B-
z*Kh1^aR%KWu?34Y6fB?4g|QMCYA{#1MDC1N*L8-kXjV^rZRo$j0VfE>JaYFnxBmc(
zZ8!w_&|2ZQwO?|jyCz{kp`<{Y?(o>QldP)Ej&8cqgB^UDL<c?P*t%&Yk?rnC2?E|k
zB8^v@LOdAZc`lM9@XKRK2c$mEnu)a%PX)!7rB65Y^_6tkuv6=zufXnklwE*kDj1RR
zNlOSyhg^j|cgY8a!}gY!R6E7^CcZF@xW~G+Sn-8y)l4Ma)6_DNWurPO|FP5ozLi7+
z^p)yKlEmXox|r|rP~cO?>WSE{7d}he_UvKe)oTk|QqnSqpAcg*GtGQV2Nn(wHQE<#
zo>*_4T%-^aPshE|BdMRJ)AW*ioW1^SHbGwzp$5{7hbI79njAi{)v7}~|0?sJX*rLH
zA;wQ|&nc^=D@l0xWhowT6F|&mjgBdd;h5SwnG;nrw)@2;wKW)~JImL@FFUW+fuF|8
z=aVVzrN9AJO*A+n<kG2ndbZ=t${f4^!k<`NLcpx%G{NGwkI|^KgQbLMV$b{x(sksN
z%P85K3;>yV(}G#TSwx1kk8K@E@@O7F_2$xt;n5$b>Q28%XY33mir_LY@HHW+JajYh
z(hZ5P+Sno_pOK|zab+VmwZA0^!;ARzs|;#$X++J1N6fs_4~a;<{2pSiH|zGVzj5A#
ztMWF`b=+>x{~uLn9n@AAt$nmWp;&R3wm_k{I}|HY+_j}had$W^QXGnF(BhIpa4n@c
zgyO}56EwkHzPvNvy)*Z}%+Ab7=A2~hwb!%O^NVD=NuzE5Mf-l*qV4#?ExT#4N(Mm9
z$;{PCQLr_(7;vh+v7zd#8LLqqSEA5A^-A7OPoUU>!iH+vZzI0iAd;a<l@%}R1&l@-
zp3&AQ=LJ>1*!xte3Ab^WIR2&6I=T#7yDDi$7*RtZ^#*Fgl5-`{*p8WfWI`l9$T#3{
zgmS!3J2k!J)qztMkzMS2@G`<19~jui7S5>&2?JSHfW4ouE$?i`G1>4l@fUp1Gs>q;
zNvjaP_VQ}$Y{iRc57`RfDl(y~6169qP2OJJIl9UslJVZ!!;bV3rNb|hHwMqO#s7Ge
zlfx|jVW=kl`g=IfwtrQ+*6&2k_iZ(8ETs&|V{*=(5b~!3VzOKoZJux|VW4@?by@1{
z*H>ApVTS@`++*AYu~=C#PnbtPdr4aFS089#A_MU+6T8mh>SU;vQq(FPG2JiHrJw9i
zU*zd1tOsqWze~}o_A*(0`ho$Cz38nbhx>a0rJ6=)Psf?>_;#~WY~k_CB6KeW4E*}~
zxVLFo@+@vj)RY^?qv-`|hChHlRd9RedkK(j^ORm_$&@imas6NWc_-0$<{br%mz!ao
z#k3-@>OK$8PFU2PNq63K31&X|Ut~2EF?cz|?zT=Xy(MMs$4c$xx*M;3@ImohTXcB@
zBh}2sR3Z+$Bsxc<GCa#$_PfsSU&W9-K0Q^frzyKpBu6kBU6xSaStCIAnMCCriM$?l
zk^b8+vg~WAs9u(WIt&h^6h^2oSs@pHm5^do?@z9$?^<WFbqQYB{h^%KQAwtZl^ma&
zg9`1L_b3WEYCu}+T4zfxxyL9_S7(V1{W<74Bq~Qo7>VZQ5m4kk?<A7-abaJvOVCQY
zAPS?uwfx&%sD~UiTPmRc#QdxiBOhOh9j&csG*?8v-h@;X9Lo>lQ3}M7U^v^penvh#
znL3lGm)jLLV-@_fuCuGYbF%Xv{$kzpOtBo=oUF+(Z>vU*#y2&=nf#l%RZI`e!`+<q
z)m?1icQci5(5$Pc4!eD~9v>6_%}d0k2=mT5`F1~sA%oXj?dE{nnkg_N>6macv{C#}
za2!g+?T6OW$3IpcA?t?r`Ufrs$`317V1c5<cfk=e-rytM2SPIgCHG%cY~deDD%Q<#
z``C+f4XyE=LAamrY#f&DcRe?98=+Z6(+O+#k_0~r$o9`~J~`EMrRmrsOHvz6oJ~q|
zPf>vM4+?Z_Ukh|2UiOoM=I1Q^W80Pn)a%_Oxz$XDMU8)4Uu-EeEOi6IS!oF9!?Ank
z)J5Lr=MXQyUZeG|8n&nJyWe2y<6{i}^Ap$Bd7^y=JzQi-XkSL$<6LIY&!N8%z6ar$
z4I4C5Mw0lsKJ^qHfB8YhsCf#2U(lo4fhUN)P10?`MQIXi%!A#KuyJ75@Bt}v=aAt$
zhw|vvWenMVfIgJTT;2^2-=Lt!NH>}vBkOk;eceOQvlTKLy1gaRp(0UPLnC&>HtpuQ
zZP_N1d&TUFHb<iVvS$f}>+u0*q<$9y`S)Qf_W^76A3$XU>O&~%V}=r|4i1wTw&o99
z+#4Oo@|gy{1f3`BX}H+FP@a<|)~m|Bq^CgE;K>q`CC#7|2c1^WOEz-s7aaYeLe@j}
z#&!8*94yUHUFL>~xai5GTf5z=c8KnV?(YaPsCa<_CdOL0$B@m7!)dV}ghD@4Ya80_
zHhRYzPa#`>?d$zrem2fKD7XDpRH(t|Y;3KExpW3Q!tb^NmK-T|8+7ead-pCqnGJW7
zzrVFtTSv3R=V}*h5@WL1F_**j2>hBT!M2<^>9Yf(t_!5ii#QR+>HC{qoQ&~*=*$2U
z10G6=J@O{zY&g-fc}gnJ&O&bXElE$+$g!&KMTs8oQ`~UblA8Snn%=BB16tj@JO*EJ
zdUyk$*fw?mq@LgJU}E3bWw|i*GGIBZ_ui)pw(Awjo~+8UJ~X6V{^`wo`?c(ICB5fg
zwa^M0rJKEbj4V@M>EYrkCh7F2CXWJPrz)aMJ&{j~R0|XzD^FMB0#wnN#ywR(>tyJh
zFnx#mJIBmBwLm=K6UvO{mX__ob}qS=rlzz6PnO@W=t$+pOLsj~Oig?!)YS(U5@bO5
zL8zcfQ8MknKN08(5}5U1{^1Fg_=H{+P^{XbK#hquOhtO~oi4Ea_EV>YW(SbZ9ZL_=
z;I<6B+paM!>)k+AA1a$u7(34HXA69{_8R9bf;Z02oVv_>+UbvAm~QWDX5@KmMg&bw
z9MKCcG;_akaW<mAD*DZ|@qzdy3hIbWw942fNh}}!J)9WrQ#fbEIul{&UkuTH0iuc$
z`98S<!i$R%9a$sr;^LB#Clufy=T(X=S~XK}AEkoKzN@IOW{)q86C;Y_gEJF^-4}=T
zZcP1{Abri3zzRO3*K|2NV9ZM>8~IqzRXTFwCC)?lt#LcZV$>D|UoaZ9=QYaU-)m^!
zn25q>w&680Z$z8gTJp2^5r_2#b(QzME7^4sHW{-Wk;r)Ajbt7<RRpr1$o*=`<VklY
z+n;1&v7_r4pBxRO$|Jzme`nTA@qSzhM|lADR!1$$n+UN&p9qbsGd4h5-CJ1gV|?5>
zxY|6sdhvbIBxSoon7dL)-zc_MVF>jtqEi+GLG<wn<K|K*b`5NM+XOt9rx6)%VzimX
z=M!vwFR^lq*cY)F0i5R%Uc%7)xmiuy5h6l5rV8a{If*fpaZUw+j-%~}puD^P&uulS
zT&U?gjqsxRIF@Mn;m0-LK}06toL*UMi}Xu?<-O<E^VM9O-wobz{697e`+rFkVhr1S
z1_={NuBi*sJVQ0`ZGTB+VEb&qVlFBJ_1-3KElBl4ODw2Cr9pv>-;FdRnF{qC52TFH
zBGGo}nZ;sqlw*s@@JVQ2dp=XeCeIiSXHObe&7slcc#~U4m?D~$Fz(<E|CGv&ul5o(
zbVCn{Pcfl-5<{9@7SX(Eb!l|+_eaY0k*q4?wDEL;F_myTrzyNP7X;1Ss&S#sLjOUR
zVr!k|-!@q;qtUjUh3KaCR|Z)l%aR{{Azm_}VauWs?XVZ~SEJid{PKf8TMNf~&0a7T
zySuAWrbPNq@&oeqeHH)nUplH(b}&}v>fxp*XeDGXivsX7nwHQV35Y7LBSFer9B<i^
zV=~?>qf-G42N&&~_=}W^RDI{ro9&sz;PnWi;2n63+4!~q^-pyk{0tK8OCOhIe|WrP
zj^P#J7YbG;@`Vts`dG1W%Ikx=@mO<`-zPFLOw})>@>C@qlj<1XP4oa@&(>zcY|=sc
z_5h`5A|P`N-6yw}*Y`^8_y1Bf&DzdX$YzFrSBrw36|^@yRu&hf#HC;VH*+`S6%-JV
zTsk5-zRo_W6qp!Z!Q>LKnOd2WQV3!AB1!O$r07N;_AKLW8J!LbRT0F^po5_sVlDb2
z2s3<P===oO-|x8pM;p$lfT1e^?I(Yx5_w{;hvq*NQ(FdZ7A0%ho{Cb8hn&Fm2Bl9%
zb=|A0=Te3?cdp8(88gii`J{88Q^)1nDep+rAlNw2D0T0zPZvn;chX`x<7tkw{c7}%
zF2ZmG4@}uesN{yJ4ORL2>HJ<`alDm<|7~|T=&y#vc_4p@DRaDuZ-S8m@=-W33T_Fh
z)JNa`Dt@MXqBAZ@o>ogp%wa>&$@Gulhdi{^E04H(7WP$Vk0YSpBW&QgXYOyhR`icj
zh3YhLF}vKt3k262s7KE8r?Ti`#jkPyZC^k5@eei5MorB-{RKaFkSq}X=2_^__|AYp
z&C^bjjo;zr8*I6-$We;1Ou%NUaI9XL+bMW&BT=;~=g*VUxS{}HKPfR7F=9Gr!G2mW
z=Q1R^U?ia|^Yh6;FyYJaFOXR*{_==-1)-t@cm*74Rq>aD{;RomaRSgt2t<(&lQ6GP
z`Ekw?UU-02_wjS(H(8FCC8YI}(=yXM7y*ggoH0XQ$!W$GJ2nP{PXOBG+2PhPHTWEl
zOLbo0dQ#c!O^$ITi1i+}@G(z>cDcrrGD54QDOKA#=A#&nd7C7@n0N|m)1b8QW5!+Q
zdP8MxW1MbNXEfR<q$0;IYzbw9xV!%C^t<jL5X%TxB@5Ln%D6mBLdL(XdQy@t%fx~~
z6xyQg+O&9BIjT8z1{@u^0e~g%ghK<wrYpCMY+lvX`>(P2dww^;e9F>Hk&phuOMyRq
z?f9ki(OJo=XJ#E#NhT!hKq52!CNJxh70`k--QK<C#6xm2V#AjLDNoDjzNNk|MLE)K
zrZGMq;gJ2D+LQD+JT|lphET92$^=MhoT5P7NG}O<)j_MjC|BqJHa2B&_wimlt?N?9
zwQvC&mOhiWwkn4{73P+DxOzD|s<WcFg+kxK4EnFVROZLiL~JBknfQo_``F$jRj@eE
zxMi1Z+%r91&{$(wNcW1v^zoVwzgFn^izIm#()~N9y1`hFI`#NV`SM}nP~P0pe%ajP
z2;6x=P)yVSF!m|SR%OT|3)XRLKsJYl=Mmx-G#YO7Bg=Qzc+o+KUQ}~wuRox3LP1Hq
zGcS&nMPc_~cgIPP+D?)C#~H5nn^~m{6Hn&tQ^l9U>@h4O_3`oXHDhDE1eB-P#a>>*
zlG5TVs~Xw~(DxfmT4uP=ry!x6cfj8YwjR}HRG!=h9L{=)I}4h=llm-T${B{3y(vu&
z)z%746VK53=3dm;oF^Iy4f7RF2<t8GWKVXil6vuQcYPk;2yKl=^d0v21s(M=0-VW$
zpl6Q{z=b10!&IzA`>eC#e}#^|S(B8;meP`^_LDADDK+8RtLDPgm*R(}k%7LN@2kr|
z2El-2zGbV=XFd}@O>|-@BWEn~A2EMF5ukn7I>K7};LOg*ollG=tExwwTB?3p!abtJ
zfuZ+$sAaH`gIM6pZ<F$(9mc0iD!&x|_-o+Bj7rFJi1horl1zwi(3=pIoGKS@{WyDB
zLP0E9H_HNF5VIOZQ=fF&twxc8)o4_>Pn>iBq|$Oo`@HItu+7O53sPNh>;N=h^Oviy
zsds3@-#V>$6@PewW?AvII@XJKS9_|(bB^>3@r0WFJe|j{wX<8vMzkT+!Mn)24bHg`
zo5O__3Ez9h*|6AGHnCZFp+pBE_h-FS`0~cv{)(1gdbJ`}JA~5FRH)gh#r8+)8w_}W
zsbeSUA_5M@VTJ&26m8X0zcIn+df!8Suzedn)$r<}cDb0%9wY!|vDSI_%(gC*o3i4K
zhQ=$hoe`OUER6f36vp?-6n;>DbuDbUH<d!TX$;ZcU4jegB~yw_cvl}qB^$Ugy`WhE
zopNH9+tlQeb2J>xxwhgtWv3>^K8SVT)tbdOmhuR7e);d7C{Q{bbzsyvQvT3VY6$I%
z4zr?B)l6Wg995;mS1bOkr8aLz{yc@(Soq4fk#@>u7-7mh(i6DZTT8D=HOBixlgjb~
zxZe92aK*5k!W_)~T+xGPRH$_gq70qcO&vRZ^&8Ym9B=joOdV@#8mwgQq0;BT**Pj9
zxt-0V1$U?OKoTYICm>TOl4Aa2e?D+(x-ql~GHsp+^M)Eh%($z10-6_X;<gkm$mZSr
z4AK>;hK7$dRNh5!X>h#YIhE(&z{-%<yI$3gL`;y2V_T;0CTE{m!C~`jJ<w$daJM)7
z$gBD{Zb#tH_Iw+g;)(LPj~Omo=<~qi0C;ovW3aVzYi=!}$y|Pm*ZFXNqq4Bf<>@&8
zD3Ol~eCk7U?cUQH4Y|X+u+S&))q_t4rRiEu4aK5=<iLpKdT8d1&A+(pKTb`_AyrL-
zUJ@1KiKVD>DpYsqjVH^e)_3veDpbVVlIRM`QpD?0XoZBbZdwEwN@QE_F^hp@UP3H-
zw3MiDDyFP~(YoIO{aLHwe4kR?hp|7TPJZEY?qGv5^A4j|*kRDU43|J^AR<|Btq67X
zTJ6&7ryVd8NR;qS!#7uTHMvMAI7oks_VeKqDz^_nw-wVh2m08g&o;Yn^B>x~@7I)N
zPB*vWGv{}yg~ZDK7?I*yB<cY#dMK`Uhwhw`HSBihcT%y_*+E!8NKTJ~mR!OeKU`1F
zJ@6!t;U&E=;u%ydq79Tt*($m#K+N6JW5J0f%c>WYUq(Sq&hm7Ijd!G8lk@hUNS-Qa
z+7@@1kt(qVUgd_UwBuWRwbpk+DeNyQUQb0PA)x4fXiRZ|O#PWmxGfX}EhiE0@Vv*A
zCRK`BD5U+|E(2E{)iM^~=<>imVoc)F1t$rYg@c?j+5O=!<5j1eeP6U;_*MH^f*yae
zPM<s703W=A{V9COO|--JGS4bm*WJVGX7!<V=wN1KE-&bA*XSke`^G)Q)OFcOFu5$R
z4R+Z0gEW<;z!#=37njEUJR1B2($`*=ZEkw&_V#dj+St<k=o04!9}bA*0}rh)wjWQr
z@5d6Q@AahK|An~T4T%89lp+3bSeRYjgInHK@^L*X{K9h`iaUKjw;i0fZ47<l&pNxk
zZ=8`Mt2C(E&BerE==h%M`LX7F737&Kz}zxxlg*6+8r&D7u!6Lfj|-Qm50JUvW9+^j
z=5+18S=hPSY!q?D>bkkf$za$;e~T5^?5dGrvx}|{bRPH97Eibd=a|^hp0~|z&Ch=Z
zPl3kmjI+gcU%abM-8q-ge$kU=Qzt4WCh8mFD4e6yj{>5McX{p`NLB1hjzC7mQYihq
zS3poZ200auAD_&KRcpMV3(8;%bM=04^Dn6G6m<y6^S`T8njw3=pM6YpKYDol>}>UL
zDZWeu9(w>9{@bK7Zb0i78d?duzN5`7<2bxbO+YPJ$Fb`D9|@Fqay)eSctVRKi~DI#
z-R^O+|0i`LeB3`q%HK)kCDB;cI$}ItCVXL=MkiH`$9eVR)T;ACs3;syXE^g5YLpm`
zVN>FXEjrwq%x}e2kSf4J^%9rdz+H}n&IT(Nuuj9IEVyi#-n8FyrQ^DwRE(Zxt($!=
zuSX^h|IWK#=Vs*h6ZcmXGxpn+cb}$zzwh?H{=F9wFZY%=u%r1h%vDO<p3pMzkg`_h
z;hL=p+|(O>(qI%|ix*aNem-L%q1^!OJM#)p;Ai++HtXkT73e*~Q^w30N#sJ`{Xp|x
z8SoKzu9Ze?oJLq*L6Qymhs)iY)a5Ao3}dj{c;78W>1$}e1L0_XAFk)><P>9t45`2)
zSmIc9Fjn8ZB<wNL<(J2LZbd@R8m^^8+!J;TCAJeI{uZwLVw$yhLHAYh<Q~_^$w%pI
zI%3<*BjZ*hW6ND|O*F4PxVjp&_|%b^90=!SuSZqysLD-@9A|FFY}HgN8lk75Ix-*-
z0<j`2oFUK8Vwcj&?L_x&CkUmXoks~LY+fDgKqXv@=TIgKTp9+OgkN5N%k@KO6K|63
zzExh^yH==l)=$vWid4z67nJyl+Y|c6scb$kp%mP8RykM75YaAi3G>@{rut7Akh$}8
z>^%QICyWCp$2W<$En9JJc+M>;$4PYxl~B+m#@6{t;c+A;|862wfT8+aX>qccl_27W
z+^;m!Tu-Ew&SlqE>~+|ps|I{3EyPZo7DS-T85Yk52JFSN+~sTYB$VS&#MsP|UH(SM
zkW;H(+;1p!nTc%GoG$Y8wjBm+q;Y1ru5|K*N{&#Hu*nZkI>aRW2=H`tc72qM*+xLu
z>;<7QC6M>;`l9~!2ckgnl@&^mr1HE^@gq|ai$=)%oeK*MVX7aF@L3ncGFaPHD^;>U
zzJKx!9v6h}tJwZ8F;%f*>8F@#%%^c$d^rh@jei0r!9rQs)L7HmL-xRuDZN)9J@lWK
z@Z#dU4LL+{A{6sI?n`ZAvNnSC@AxM84g(u&3k%uYBfCyY9EWGw8Y-%tsieK%HA?%o
zXALSw!otj-a>pdk`#Fq!^t5pn%O|Oz0YOs|W%#9}d=zi?yB`uCP}M>S&ix!Aj&G1U
zD1|_Ze5pV`f#1;3M?S#y58e=52w})m?JH4ZqQgiLd5GAX&i3K?CqavO26{3?jg6Bh
zs;XKBmbt&`*<@FzL3>T;IGdA!Qr+gd&hUYo88Ct7rOd@$8Mmcmb6tqbr4{l9rSH_7
zJs`mMzb}ESUZSToxf>Eh!^fL$KPwclJ`i&?)D`4}PETN)WpE%+_UG74X@m7AP9A84
z+E2Aja8kEz@EZ7`{3wAc+A{%fxaEwn`Z7+{@#zqG08^YPRu8o@3x_`0hL#dK+i0L;
z1@X>=_y{@9YOnt>^mB;awsN|?{F_3;pUJ}h-^ijauy8qZF<%kmDJtSFE?L~)J|4`C
zg97hF9wxx0*qk8abO3R&%<AIYte>9-+2S|HGkKo1o#n#Lftgpb@nZGe5{nrj+q_%-
z;2S%m+J+S6yq25w=5Ei!^$S7H*m>uR!<E3hRUfJu1cGiQXd}Ku3s;6yS3tEj4i{y+
zrUmw`*J(Bd@sBut=!CeFrOk3pdhuuRV#1W#i7uo)q#Q-|fK`n=_|&c_ZYYyh+P8Yr
z;O9DG`_f$Epi=@gJYt2IaJ(lBHA1-pL=*)LujXuK%V!*h#7!ohyhC}<zL;dBQs{(h
z6N{@k8JiS~wXJdC>vgo%!^)d!&o6Fn42sDUl^{&GMWQS!rI4&9C3ZqAwCdS5_LIgL
z@8PB8cGr9ma0sBVB<&&hdlDimH1UZ~s%XDfG9TMZ7M11UGXG-LJ8)P0%NlSrE76yh
z${(}-q5|$;IFig~u^BOgnAZ&2>j^`dMTCjCF4_C;&E>V69E?l(`uV-U<iX%KPDnO+
zMi1$CC;!R@d%oX+h0IMEeM<7h7pi@R@}$)1Uk)?`lKs(`@d{zuB`=r{?mJzw&lfIc
zXIpE|!N5k<c=yqrrp4osVfW1d$_Vf!XbWYDTpa&mp{wP7vHKLbLmBK?J|Z4(AGtC<
zH;W9O=nNSRS&7pSDgVbB^e_N}YexI+qGq$zIev=;xREV4cb~Iex^$nT3hBO*bm_)l
zXMqTLk54)(uN91{XcNu8eO}vc1&$lrYhX%buN@B%(7dTqHxph1B62z655b}|x^4_o
z89h$Y<EQn@4oY8>li$gOX`L`-j@iCrWvKQQX`21UHV{gB*dXv%g~t7jq$JGp6TT`t
zjo~)7(Ev414SIM=$_6hk_LfaH)b6=z;S})M);J(2h)b;7XDbGGb7A2r`sJOSba0T@
z)gY^3=Bg>YndZD@*?Iic#>yS#<N02}V~OP*3pR?n(cx>ho-zl{r9Z$@2VbPPu(7;&
zbCNYX5+rPjJm3GYWNA$PK`>{mv=?%Hbar5Md3LpWWM#=~vBN($9MUb5L&VtW6RoOE
z(eKTKwj+r_{Dgq=`Y!(g@_53^cs;nieY3WPs?L3G%0X3lZ_+GpwtAO~!FLG8m99YK
zSdNGqyc;!(68J@lH2CB><6}WV#*-6t1^_G*(nzAPbmCP>f?4<xDrQ5#aZW}bs>~E_
zpn>W_;@ket%Pm!a{W0!k^@8bIs*Ii}>r0!@zXA<ko@nVjSLI`&etX97Rcn<1yYGut
zk>B7u#k)YmM#4#PGy9I<h7JL4+h1=e^E~$_{Kt)6a{h?p!+!yR5xY2T4MycoXdMG=
zV6&-PU4zParITiv)2lrPCvxX%BB$I`z2m62@dpH&3MlM&gti%tlqHw<HesUjhB3{n
zXO0v0`IOrahk=fcoUt7p?SVm}*(*{D*9gt7>s#Q+6?xF%$GbF7o4|D!S6Ub6PW~#w
zme$lzSh%T#fFzuFt)SZg+#J-$atlOqKFlsia)vP}$s!ArKkLzT>QsJ&bHXX0t*QLD
z82wbqktf#<-;!Uvqf+T7(Zc`QzXiE2XHTCchtZ<;Uv9P7>uTP9+ue`i)QQTy6FZ}y
z-*A|YZl_vx8tA@0NtjOYDO(oF79XfC!6N^{6{is^$vGl#`f797ij$!DItcKALnR>d
z9DS2-eNMOc{l&S*4zA96J_VSN&G$~bsuC~va=EImsC^4a<OCCB!*11T8gT_<)(;P3
z$AkB#K<{PDxi5V$6D~D{=f<Tj5A0$}q?Tl6Bk+F2I3^v#jPRp~@UDD~{j7&y&uKY2
z*T}(hXJA>wLgM2mf@6?|Heey`+5RnE$bB@*PeA6r#`Ur9q9oST6plPxlKCg`DcrfH
zM*4V%yg1%{T6e~Uh$3Yk*0U2R<9E$zd7a49eRYCLu2K2s@hseUI4fJdKiAVbUdV?p
z2Z3CYSPmxswr1~jGb%8CfTVxUl6%^pn(K+`_lc$s^!wT%488UOJZ|c$9Q*E^E}bq`
zd%^}7TT|0*A6IeX#^|YB^B#Ae=N#_rY<e}Y_n$gt<xGb95Ta)8m{uoP)B-yFlQL|g
zXA2_(Ak=fu*Xf!6+0TY)k-V51nx-;Ta~QKd-P}}gODg?L)sN~<Rm>AmczHFy`x!EW
z=1Pm|I#apnH26RO6n_$?w%hfP%yR<CJX-la9vv}hdGJ%y7GE+f`8MtN<e<t%mO=OD
zB56#^2El(-yUrrUv+7%#1CjHZz-v_Y9ow*YXxV-?xb~3b8DjC{i&1QfQ!5!P=(shp
zra8q+*xC2y+fe>ei*rTW&E^HlAT|60#uGf6_Z7#oVjD5Qc=|?XZ&>#^e)rwt<Kkmi
zUF%698>5u8-;o;1<9rSF!PPr#&`*>hWN&KxaboUanhc>VbN4XOvo`m5fU+`_xgJ9`
zgApcPbU&1J&y~lra3-K6;SIh_NmY8DXc6d`03)Z;;HPBJAzbU}5T-L-L<|_4`vnWl
zz8K}K1Q8&sQew3Y6;+B}Z_tkw6@`D{ccHICjhIkbGmuDn{KiFl%f^9?^<p*2?UkgT
z+`B(UQC*C~=i-hy3T|@Lp0ao(vVE}84zZ?NF%Aws6|N5ylFa2lP|=_;7lYVM1r4t_
z&V8@*@Wshq`bq4vVaVP67n8m_`=*9+gD3kD7rnr|o2xL&dn39GUh9HtXEyr9CBX2h
zs`0hW>B9sbe(|m#_8~+2iAtN%Kd&Ug|48-ZsY6^mYMK}0wznv6q6x`yi>neIaP<Qh
zX=wvz_vwvD+)G2S6rW~>1Q7x+K{ZOX)46)KpURQCu<(0orB9Xb`J{DxKra<0S~!z6
z7eMf~M6aamN2)lynlqM#CnlHp3|j1)N}?L=IV%8`SIRv7p~r{r+5p{qXZCVO>`1?s
zlu7ly*w*R?F;GGz8$pl6hWh8Hovq&7QNL}(A+<}(oibw->hHM8pJQ+|QiMmDj!D~*
zz$Y#)(n7!@G$8d?y<XDMy-cZcc@+L_W<aTcbxmWQ`>U5YXAS(Az86vq9Ys(nS#oT0
zgZB;#KRyq$L!j7ly!qpZrLI2F5`wagXm0$Pc(V0{G-WC~U&L||U6Bddw>2;6jMty^
zw1#ozn5pc{-1@0(2E)JOs#MV_1(kn9hsQ{=VYL#6lX76|zW#nv202@w>$(FkfB<Os
z6`Y66hS}yQ=W&LzbZY)0;&9h5_m8&ckvntmHGlK$mhmhSTxx8Q-pY+{{j6B9cIJo9
zyBg`NF*M#0e8U<qF}3f4+3|fr6m!YLXqZWQ1P7K7AJ#bSKap?nl)LA_&&#?~DbRk_
zH6>@<Ab(I}1iT=;D;b*!7|g$EU8m5FEOC!$f`)lB;F07qGe=;)CfCyG%&56Sr$>*;
z7`Z|h0tDt{PylATspExVv}$tIkQ;meEU`1ild_Q_fQE*HYadq<O6!O+xBMC^(Y#<t
z&<Rm}Vp`&V6Z@#u_kp--<%$j8wZdvrte4%LfxTpq_w%2Pp@ZQxPT)9lrGA!l9WRt)
zSeYl}3GSO3Rx*@GAAC~BGh392=4Xd4^YbZ525^loj=@lw(SD7ttvbl~4I20&Dm4@b
znyK@F%fD4n=I-JUxR!ay!Ex<6i<h|}Dxvl^Ic$b?>{6X;2`Agy*_>}5Y*R{KzU(hD
zM9FeIW_hQEmq>7i!43Lrxv=CzOtIN)v$7nxNQ%-%4-7g#Ur0(A3?e4Z{r2JKC`5AH
z{PSCuVGl9OxVW*<?-3>}v$+Cj7E))GVVBa|<f<7xH|S*m%+2fKdK_4rLZx#-9%oi?
zX>0Z9)ndl$OoC$^p2rA|5tABWss@cUDTf3)SqwkfeJPw(*_U5*Ust~Rg#_(-iGxr>
zU9iUiSfH|_kC}IogdWJV%gqyBo?dNbT>r`X!$vOV2P^8FGpjfps^=>-r@WHC2yDM#
z3;W>;hhPG|AJsjm_=Sa6*@xtkU$Swj)&EGJmIJvwOp28`AnzZ_k+(M&X(kTF&czcq
zA&wQ^$X4&l_aO80#r6xBz8xGY^@rN3)GkPFt7nb7^B`LzqsbW^?O&HQ1)2#O+RtO_
z@WpYR34}R(YQ1HZ`7s+bkPgM+h5-|pJkrOL5FmY}=_j1d(ZWw*{P;VZ=D@GNR#|Ue
z3}d2bu<QQTME872zUsAM6Vga_;`)02?Ipt@mbs-x0)3J==gDUX%hURSw#r4_8d!lH
zY0Q#mLKP?^%1GHQ551Tqf-31J8Li$2@$nn|YW*46Ffzs3KNO+iF<&Fl$P^-y$p5IL
zeFgBe$OdKq6vezN@ixyEQ2HfDKG&Dt=B?qMYUJRweXSVSOxFe^{TiGWh>_F6mc<Lr
zdaLwm^kXYnsyiOlXMUC;eMJP?VB?<W7g%;v>qIl%e(#fkXYN?pdby78+Pbx!__(j!
z<YMW!Hmduu)j$4h_^0m*q7gbdt?u~ukOyFpX?48K(;Y~JjzhWM<Sn$>goJ#s{mUZ(
za)F2~N6!>2a*ouB5^SLzbOhfV7m`?uhMcWN;FJb{W^NN#B-jo2E9pboK7M2rtoAnW
zL<($ITB`l2s1VctPNOF$XSN;ffkrIw@%_L(CW+_!OaUSEU(cgxRd%TH{5U<JrL@35
z9mDp@18qS=tm4}85n6mBUu!*W&~Wx%ncB>`14v-0&U0?<tuI^}=j^q|5kex?c4J7h
zRlej^dIiSNUM1pb{r9B)o&sR!D7f&6t+daH7Qw#BWwyFLtH85JLVD*GTk|<A6fFK{
z<M~V=vi-8c&>+LyBLZuZSR<7~=@Vc@$xEXoAQx8oG;jZnXN^?5kLjPvWNEM2Il}}7
zu~q=7NNI8O$1(5Z%z7Xp<V1H47Cc}`w=;9dX6`;TqfE&7aF8MWhG}Phx{>bV3t3s4
zC_*!Cax1X>|LB%-#fjwv6m(4G1qHr@V+9HNGlyRORW6=p)0GylzsN+s`xTt|ob*-Z
z>W43x<O~HmJLB)ig=zmRy|xBQAkv+_8C7-d|BNN)$Iaw2sgtO~?c<C1sLskv-e_B;
z8C7M_J)*M!Yi780Pxw^xW3aJC|7}Hi9&R59Cc)9ZJSBaW<<OrE5|`N-`1<X-=5XY~
z$s9S$Mi5t!W>?9qgf5K6uN0jv@#O3LsClca@^ii7kiGu>-U9JPuRU*>i~FbZkut|a
zp!V(eZ|K1|VZrS_%)e14PeM|2UC6(vY4ZEej4g-&U_*5*;J$v|VJ6LKo?m9C_Da&V
z0p4z7)NzZ8LD^+wDly|EXls`(IkLZhUxex?wy;|+-`iVgQ|avRsGKq`8@>E^=6%Ev
zss*u^yx$$)^O3;mK1tJ>@x+G&UK|itR2c;z_XZWU<3!GXc<p)pW|6&d`;L<Q<&&Ql
z_1p6SubYCbk$Lyoru6O0(1nwm30<|wBA<B?euB(EVSJKbkjmQHH#J<`4*KJouC05O
zuEX>6tw>_5ZkrDuTwIzMWcB5%>VCq^K@QFz&TM8x<X@E;#A1*tCDpf^q&X7CKZ~i}
z@vRJtM9eH~!%=QplqQE4lQ$j~68T}*KRq9Tra0P2q|~hOeiT{}f3da4XuJ~U#7Fv@
z*=IEuO9xFQy7wT7?b^1zy~f7#?uY&EJuh*_`+t!x?dKO<rz4zfB!|-{!@-e=6^N^A
zyHIzNxRHgWiz~>bv1~}PfYsf`*tUz>8jI<-*Dp4b`qzqKlU%1TO8-EIQub^#VrvBe
z7yfYq0HudNm$djw1Mc!Ntu`oM)#X$x()k8Eem)>?Tw7BO3%s&#UXR$NCmu|({vKE7
zrtoU2=G(N8JsrW5Qq?b3@~KU`vTs<T8jZAeA~UzJ+tn3{qo`xPqW>Y*g$jHjk1`c-
zN~z~yiZbElL?_c1BCVVzpCxmoHPSBxE?DR-D-*nn3Y?dPJ0C-@y^Z$?+ElUyl4xQ=
zvuI|u|KJi&e9`c6`aEvyFTuG@agx!wevRMW^id*brRntF`U272zww|ch{aE5A2o>m
zWjk~B$mxY4LVVC4wx2F@I^|lvA7b&7nu{9@SKL1;>`hqS3wYdA?T+zphB`O`wLaEb
zj3m(<oM!DaZsDZ8bMJDW#20;_AY$xlu6(U6+S(u(Rj5Nf?OiOSJL9prbKfb~=nRL>
zU+?Owv1RLGi?i?Z2V1<3Sn8kW7+rE`%=L^uCYkWJi&DZ;;U3XDez0a66LlPY_T(S<
zhs1Z>zNJ}bvrnCs2Qz6*THCiaK=<NdZ7s?+zSe)GWj)3Yg*X7)3pFDbeXrFlZ-&*l
zLr&7KgUfaEnnW`{I2L?jaw?JzBM#KQ@!pB@9`UGoEfE&doYkCv9S1k}?wiv3i`C@1
zd^}y1ce4bSgulzMgyjxjnzuO4MJy)1VWctgy?u!2S0_wBi=QZ^8Jn*FF7%3z!sAG*
z9Bll}SrE&CJ$cJG+-*nEI2V>Cn}@{fdwU+U1^j0Qv*~l<`vaP}S0~QC^b&O~o&$No
z?cDZPzaj6qIH|FsY>ZzCWhHMZTBjD!>}%=k>wgo^R3dqsnU|fPWf9dQAGOOA9Vyk}
z@Nq)(sQee!qFqI8<?ikw;JKK?dCBNJMlmL7jV_e3Q2()QrKZEvVS<EKHnnoXX+V4v
zX2=rvB8K5_4t7HnaTspsTUiC{^1ZgF#wVLhq-Ypc^zu*(g*_^P+3V{H=ZNLzhF3QF
zoUtWzc2TcM+&g3LMehdjnA4J1I(1+uf|2N_KBrk1;~7(@23w>lAoUfRMa?Rn|CvKO
z;%_uHGw+|$un`|;Q>jL;<G$NlG8y;fGRun?*O1>6YDC_Pntl_G_WO4TIXJCBBCsxC
zX=E(T@d8zWIcVPrZAXgN*PAshA9T$hs)0)0mbIOl`j|u+UF6gRHBME~<D|)bw$Zmy
zr@?;BKV2%XFW1}-a9a*k=ST%|m+yIK5f0azneJ}Jb4@dQW}#2_%+4AZ8qUF+N5`R>
zj9vTx`ZQad_7kw5$Ulkdy|3M9**a3tArVl!ehn3~6{LtTsrIl4n;~2>DQidhVLk5L
ze{-nYqII$PtTm(NO-&MOzzQCzq)FA+Cs(vC<@4=%U>h9;G{80BmeIob_4Nwd3BSpd
z+63EJHS>XMFNj#wzCEt3_cGX;L;E8cveTD+u#PNpOfBEdW`tg`C7S6I^m4Ynj3RJ<
zzv?h+_ICpIuzzdwB()ypL9o1an-EnsW|$rSI4(5ML!Bu~-Ilb3JI0`U<@zY{aqG{}
zMwr|{U<@CCRS{zgJG|5*&%)@hxHUns`j*>xS9a;Wt(f)rzHd9PFA*q&<Dtn4RTinJ
z9LGyKqENg{Bc94#^G-LCx>&aLp17?VQ6Ztq*j{(8&C7vfy1v4trks~6zFRi8tm5+q
zOIKL{Cr!=nESXn1cgckMQ2aCP^MtApp788>M?nBBMt=+Bi7S>uO;uRHPh7O6F7`Z|
zNlQO%hhIQYTN74<CcrPIN$~5HF14Z1oWts0v4L=Z1A^f_J<E-I$J^#n;+v2XWrI1w
zi{AUM*%0H2RRhj(PdvFd7#I>6eN<nVzVNl3NfP#bMP%mqKS1()7U3?&mlXB>&e!el
z-#7z{O?Gp6n2u5p(`+!!SW8%!<K7H7;Jij;`3h6!DERTp5P3TZ?^?CJJo_qr6786O
z9_M<KbJ>|cK`-GmTItr4-fED?Q&a7Mc#&SCRM$0&osl)6y39k{VdU#i)pK{VHBqb+
zSh>9GwJ$DATjNu;_Qjf>b}TM;|9nKL3U$yC2)?EF^&8?5bz|zUC&U%L`yN%1#EK@%
z^!9lQ)hDmex-2}jpSq1Ed{ML1{ys(qofcqIeey~%rwfTN{^~!X>@mNuJ1Q&NOfa!o
z7P^9!IURXwxI*q0fOPTlZ!>*c_1%Ipkdn&4T6@BqP8L@|nwrYFw*p!<Z^?f%1sely
zl@oC`<-_9Xw!_WK0RzD}L_#$2jdI>6(-kx4c_VT`xp{dmjiyB=ky@(A!MfY#@xwt{
zn{pQ8SA_~{{%^k<@IK)Smn-j`oOEcOBdM^3FDj;J=8F1EX@qqKUTpdMoNRKi=om6e
z`MgjSnZ8#__b0k<(}`}p=)A16k2eAq^%O|rb7PoPqk2eKWm$PTM)aI{sjeig6-04L
z^NH`V5}!hgb0SI_&tn&K4W*A|qumA$l><55WBkYOmL?r`zI`3Zk^2=s`X`P!OY*C#
z9WN77?&WxH7F00S<cIDo8XxOZZ-p=l?QaQL8Xwceu=(F+keBH3l_DSP(*+7q)~e3B
ztq`izj@~EMc_On8yPu!fR-s#!RI}C^`q@iWKmYSvNI+&L8;Z}*F=8R3B&+>O&L<R?
zs;Gih37f<!J=oL^AGiHtyyk6Vi_gg}Ptfi%8}fE|sJLB|>kUtB$nmzO*-zgKE#Ebc
z?fj52PuMY321|2)|9rBrSc0b;2q<Bkxk2R-C-ZM!x7a;$D+jOmOryCe*q>VkhXh>g
z)V+k?_faB|-s1Z+ud^E#Yry%U&t@qk^31oF65ilDf-K1QaB<bwk2d#`iPbeU872KR
zx%Oj;%$htG6M8R~Ok1uy1M{^At5=Y${8UW*KXiA~GSmJt7dgm!6rpA%d&0?I1cT}#
zVJALG-U4$R?P2wpM%Lm`hOs~c;sK*{b6w8tHy=4a)a;$nGhy=j1ngHWH*TO<sZPyy
zwDGnu2f?zV&6CPM3|jGV3O%ivc85jwdj13@si3_RHgjQ_?)!9}5SUPwns3nOy2|wf
zJx3qMoU#`E2I~XwJwyS(NOKUFJ3jVa?MogZ%NNCqX|L20B5ibqiMGMh_jaAtH#HSO
zg_)I4x|(6e=F&5-Ppa+JH5#ibE1z5srIft~i7(jy9Jj4(3$$7(2}d+7BA{Z1@r4Ol
zB2So@Vln8l1Ps>8HwDK%qYH&Sw_}Q!bmJ$qR@cfy){FC3wV6+Wki|O#Ci{jPHqE?_
zOGNQc*nB=^yX|$>ikJ5c=l+i4S~>Ll*IzdFQTpTr?9P(zP!c274Gv_JaQ$-4cAk>8
z>DuLJ!^Qh%yP-y^VAVAJXudD1`i9I{mYbXZ)D;zHB(-aK4L29<OiXF6)0HfUBjnPO
z*kl!I%Z*68tUFQ21eI~L)F<BajAxt`0XEp%WPbyZbm%(z5e>m=dQs?dIqI67C(*u3
z?zf}A)r*s4uHT^qofICsq!Oq{v)NKJ>x%T)H5pGI>FEIo>F>H7t2Jb!>G#n>P~x7S
zNzp^X7inT@nnqRCc0kyM8J$V~<e<Y<xa^}UCx)ok&R_*lqL@KKX{0s|m83m!j8vLH
zB@E`~#*UkcNpP!4dkXH;KhV}9Z8HD7Nbv($>+QP?bJTFSw36@RJqmv#x%)4^+ZjZ6
z_Bmg#Qh3WC0Lgdg%_{(5PwsM~x5h1mdgfDhEgDOCTy@ETtN0xB1g)@MIW>DINw}Xp
z^!26u`2yJMA6!;F6wpwGW^y}aWrG}rO{SfC&YTJja!^O+4+SP7CoHG_<-Nm`>+6$4
z&)E`PSGt0ITneZ5pMCrlmm{FTC2MZ9&%*=e%6hjg7OpDf)3(2i2msIJfIvApruM=n
zP#d&g(AH7Rk5QAmUSm#3*z#e8QdC-|ILbr;4qRCJt^Rsgw$(2WOz0D1e_O7|%jw{j
z+31UGbs9k6`y4JW<~k@)VeJ|gVhOs*suX`|N+6u2KvOCET<SSC+2?g%$j5wx$3Q>L
zqz9Rb$E37CEKR5H?p9qQUnlU$9HLJ04|f_i-L)^PX3jZH8TFQX3X;Vg==F!hm~5?Z
z08#yxiZ}`ym)J6MY^1AgJ3dUlsc%Xi>IrMi6kt*ENcc1Tm#n&)lf)AmD^CV7q58|J
zjcd`EWaC&xFK=m}uZ1sL)^^^L)>y1>OwsD`E%a&XR;rEBj5j$m=p4g}c^d7@<{$Mj
ztuzgI87)_#Wrm?y@Jwvp<x;E<vSrsAV9Ysz$^GjpnJ{hjT?}e!^9ouV9X;L{&_o%3
z6IG(de0|ow;xBrhl3(|^v|7(S#66AvR=((iFgheCvT+2}FV>_R&Ia7erNh7dT-X&d
zd6uC);dQ!L%77qPIQ_;o4(_<)&<7*DZ|^O5RN1qs`Dj=q49WUXPZH6#QQHRa#8;!*
z6euVu6X_-L-cOCy69_vcp?~p$&5AygB$4`k<VnpJZUK(ORVjGwZ7!1Cmq3pn|9j3m
z4Jv~vgdzBB8R}35LC~>gtDrn6@8z0fI*Iqn>!gGl124M_s`T~4tOt<Yj}OJh>Cv)a
zQ*npDY!5+DyRjCbzJ`_Q_BKG-<YH=JF4fqi<c~1v8*R$!cfvID(ujPo_bE6SL{&eE
zZeIowW>;frcHZ|v3fQt)LE9zAPg#^-!6jLI0ztL{zGB$Vx{eef>@?S->Jq1B8JN05
zl6(?>xY*N396w-D?)f<!>sfr_4ihKb`}aKR3oh<16&_UyZ;9?`rWgQ9Yd@CICED-k
z@pRwR<n5Vv+uKY^1kTI_Abp;B*XwUD>CMc$x_QkWhU<dhqtSBu-sMm8246ns>DT&X
zp}-Xkd^1v{)5#Oa63i8dj$$veHyE%-d-Cg#Qj5h*DB08*4!Mv##;5UcCV@Yv)b?>h
zWgxzad5<!lRF0zT7-N9CQLlVH8ok2rk7Sw8BIO$k^TUdkdLG+sF~SL~Uj$e^B%x85
zi&Aq7(#xa#x8Ljl>*YfRF%Kt)kdfoGEtmOLx66by-AjodhZ4Sa@f9T-oH#^`93@J9
zl8&kJ!TC0dkiGNmY{Z2Gk!tGU;`Yp5TBDqvgWywMJnwiVZ#_W!L%0&FCt+@LYwK>A
z)L~Aulhj-%W#W@3D0)!dc*ui*H+$f{OHG}Wz+q77H-DJy-?Y9GqexV*pbfX9aEzpD
z^7WEXv&$m(FMa*^a)uQ1d_s)aIvRo-eIX`!foSIGy6U(3Ghs^YNAcAO=y)-b<DSy~
zCz~h3{LJj-BSAnOyb|r3jd4!T!GYPfx&Hav9TH+{61SwSItrGvEEG9nvt`1m77QMT
zjycb_k&~dPc>(tp`o0Pn?AcZjJ}Ig85{fq?^>e1aaklmK9lyUo>&DF{>{|#bwNODx
z;K~_a0&d&1<?rci<80&KN|zdM8<y20t4H-c>QwUi1n?fL`|a;&>?kVgN2S{6r+ZJ>
z3uf1Z7|-w}+j-y?sMOhuvP5ZfDe0cnib)Asl075pEj}WdcTYD^G0qD!)}J{LJ*6h8
zvz8YT5IFs-JgJGw#qm%jDffivecS*43U31S!@_zl3H>ipy~?x-9T0n^AcI;_KLSJt
z!gttEOLz?MOjNe@T%EUPgYJ1zK5FHaU#_+M*?%ioiJt2Jd)@zYH3ml1gS-+CHt`M}
zAbjN^!<NffBy*Le8T6pidR_Tu<+x)1@z76l-+9*mh&J*&BFVJd@A8E8HpZmm*pBD3
zG_YUe3+NZ|XfFAjj#-K?2>onqckMtdkkqzc!(5Pm?ZBGd=8vttVUggf?$*0fgybX)
z6w);C6#A-5ne!6-^td!1m^?TI;8iJsjec`**Xzk7=v9MZ=VJ=;A;mZ3=tBBear$)W
zFq;Q(Fc|!tVT(YjyAkBtx_JGImfHo8JZ(mj3QQp;c!Ib4$6=j%Ir@x#hw{fKkNzw7
zmJh!TDXO&R1Fl!$t`E({=0f-)xn1r8HzM8F{-!zji?%tpU8f)%#_B))zXx2~uS>^Q
z+G$2N|2#uJY;-^V&b}ZDzGzEiGrP?SqnCp0;>aTc=PoV&=ECYcoBkc2$Xo&I11oOV
z>Zm#9=EY~450xUp+xdW(p#e%@^a(`xyLPgc=6Ib$X=^(lW>Y?f_2zdX+Dr-0m1guA
zrHfJV03qpzgqkd7jr*92D%S<Y*J%gAH4wJ8_lqHqTkRGXG6+CR45WP|IcTnVKcy*I
zw`}&0Z|>j8f#t2&-00aqpxvSo#0RWNB<vNZ=yg_{L@zN4c&sdaeAFKMduqBkCyRim
z?QFq(yF7fYU69zOw1;IK$5U0Pdlb6pzouwfnqAv+CG)uPZl(_ixu3x|<!N)e98I9?
zKKBGoh4Lf@J+$q+NC}Zfb9cjMP8Yi$T@Ckm9&d!=M}W0$$R(hUcwCWjD-dc&t}p9C
zG;jO%(<8l6miKp3hL-Mq<dm+i1%Z>at`*}>0Y6(6TalM}iye1x8D0Ik?uXy&)UMa6
zMIPosk3^OYw`oAgO0X-i_P`i`tjWK>>q!fEyWhEQ0laJ4Z+txJ^X>Rlc|N*AGonD~
z%GF&VA@f++4fq|wKqXBO$;EDw{?m)Qq2=|Al?%o^-hAdwo|{L8fZzFSt@}eVA^z6m
zM1Y%l!u|1xDdcf(<=ks441D{|)qkuR>CkaRubd9$Cmh$%aqU`5gqGipKYFoT4n!@9
zG`3{9N#A57s_4L*o4^rE<D3&<;6UVYL-TH^=7Srs47BnB!{CsMV6gdW1o8?}ytb1c
zG9Z77?JDE9XqpFW*<@P*=A@cC&J}C24Db88<FI%x9}u4P>ob5pAt!;Q-;{oQS@#y_
z#PH?gYS5{~%9NDj;__uzrLp9tNY~E}uS4MRI6v?*L#B1YVJGCo4m@aQdDaBh^Ua+<
z0|L`U8Z-ZPpT2c{)HFl}T)4uLYr(-g13U{i&X4<1dW_e{%l?l0c~YL&zahZJcFz@|
z)G-VRn~-#y$h(`&?|Ntmk%8Pj6UmzZxjsgS<d*o_RUFFYAFj}E;B~)4<Oe?X4&L*~
z1PQn1%s8?oz)rg@ArQZo|K4j3-UF&q120$0WX^WO%%_Tht4UOytHbqJSLQM<qw{c<
z=y4O)kH6x+0v{Lt@9wJE={+#6T*N+al`+^1v5n2ag{Qe*M<}yboSNp$U#H*tKkIMG
zze$F#u5@iDoAaQ~J_;WN(JnVByR9M@uah&(oqb^!X<@!Al<CqZIWl*^!XrG`V@$}@
zJ9yK$cv<EuZ?WsH$?y|vq)zu74r*B0HmkX(K)kA|(r;MJg7<RQvL#@S9Nb;&J)<9Z
zb(SBH|1iUNfB$ynYy>zertA&Q$wnX~A$`*at2>mg9rx2#*DzMjCk2=H*RFT|4{u%X
zV9Vi@LGYZ%14|iSWG9dxJd)OYcd;nLe<85mguL6zZ@U}pfsO#j@VptiIWODm=1$ko
z*1k7tBb_+NWn+iH?{<M)YipY!b5c1LA+FP3$2@S3rWSn;VUxB&N2TwrQ<i*PsfH3&
z5H*<Ne#ZhAT#r+<7T2-OU8{32w;l-qfxG|~O9r~U8YXAj0hdnSL#~aZr5E~^Zc1B*
zGl`W=(MxN0L-(T;BF~`#a1zzMyV)mja|AqPXA2>7Gn60TdU@S_#fH2}Qj4^tZ{8{)
z$22!KsoTw(e`88f&lz%H&UpP#`Ed=@eVd*ivRmtZ>S8+=fz#Bi^szLe=JaPrOEqe1
zyreO0Zh}e!Xcr+!%aFAu&5nJ;t{VRh<lW%o?c7Ll^JMqh9B_7t=-lpp=+QLW`$8cs
z0s_q!fjZXAfvZd9uC;UE0^EH$*$do26TRh-*Ezl`emzxP@Cq@b=HC&)?Ub3mbpK4x
zC^OzGTb~~_GIs0!^c@*+F{8)-q<d{+?|J~e7y-DAMAeWC9ZfjWi`IBVoR?V|9>CD7
zJ0LQ2pR}sp`Z#v8bHkb<@#T%>V-?SB|6It4F>q4R9lZQ<M2&puq+RBrxchu!%k4Js
z0bUaiiSPAH(X+GO(p++%)s{-S7260Jx+idYl^ZO<nDhhc>THSMnr|4}wX$zIo7W#G
zgNm6_MGk1h-Di<9r~Qu){iy$b#szTNY~ej4(j7ED2Aq{X9@}+41doS6H6QLP{vTKG
z8PvodukkJ+KtxJH??UK;7<z{Q0g*1fOAS(_R}m6=r~)F=rArZM(xrE)(ggo>5Kwv(
z1%aD$&z$?>&g{&--JQw)@-5Hve2#M)&k9!lrQH5Y%6<e3Rydk6vGkt6^Y1X%=ifTd
zRr=$tj}!Kkn8ASj$>2)I)ycPzNui)E$vxLkTifwxR2%fmmm8fvyL+eB9f#wTn136J
z7~3DQ*;&hr@S_6Ezmj|Zy5GGx_2@j_zxV2`bx085oJaG)UD1gb2h^-Unk1f>*|FaC
z^wz@TrW3P&eU#&YCmii`)8DSW4L!dkRJ>+hd3AUb)wC!kaW)}%yI&o2K5+1I_abyR
z;^M%qLH=lg%^*;!Me*;jQ^&vlV@xh(a*y-|f%m<R-baEL2c0+nF0Q|jTRgpda7A&m
zBzv>^wDI;!V!&e1cJ9-g_<Qr2iW`X@nBD;S`;V=TViVtB5T8Ba0YISE(6yEuJjdgP
zvO3TqRi?8cgnV<{<Hd9~TgU?vkJh}8L9<PAx6BW{0#<xaQ=;%&jP(Ql)-%k{U=y9a
zz=P?Ehs#^me+B~neG%}d3x0RKPRLt*J@|@k`Bc91y7AVVd@1;J-*>Fi`nb6KVEY!I
z$vBgD95T_#-eUc1P*UM`>Za4`V&UMm^+j;!d6s6D)AY4e@VQT(TT{px;l01gem^L0
z=P*|t`@9}4x%h|CjDPrYRVNcuP8qrt&Fp?z*R~bhan)wsTJX~&=V$H$T;koIe@+vu
zH=Bd*W-Y5moxiUXPZx>gUY^Eu%!GDRhW__P6o(&(9sc}l_09T<yXDfa^Ga6H+fmk8
zUH%LoL~yV%ak<cRGZnBW6}WUhc=36ih+Z6O<MX#t=*8v0fu;YdQgFZ=<~!Sw+rih7
zy|2A8crC-V(q`yF{coXr!F(-><P-NV43EF(N-izDoC|0_o2rky4)b35Jv(egVRcI>
z`JpjxT)t@~^lCrg*r+;mEsCwn_^|co_<-os>yvHyTcW4!#nzqpxBhPW*8uPGyqTuO
zpWAes9WDL8Gb2wSPCI|AF#E0?7sHbF|5t+UCFK79D^EF|oY4O%LhZr$z?15mt@eVC
zm96(lB3|>piWnX}6B8|D%%CFVfP=6E4D85DN&;w1l8q)0S2%Zo7z`d(Vn=KPRAf2+
z&07lIa-wJPadK28$!$?<#77qV%)xj|h!BPGxs+;+L&#wZjD=XDS@&wE!tpjW>kM4<
zYHa;zgqf8wc3^Z_n%@-x6f=o2f!|b6sMbHBY2h@c{mREgKmeO57TqcGQLU73K-`-o
zZaCBAC*dK84Xns@A|iBYYO{_p*M`;ndaF-jn0!AURA6s&m}^fQ@7T#af|*JQ1B*F|
zBI1>}H1cTSP!z3cCf{jL<yRMRZbQ7#F>fd={EN*bJuLumsd->IP{8lyZPm~P0nAWS
zDVzC+|1H&ZRJhn=e4iS<X%DS#TW&d=R6;DsC?3ie@t4&Q!kJ+0AZ5D{)FSFkitB|I
zYQl*sgoZzk*zA)Y#%a$)H?L-yNmsl$Ho2&vY=-^WlfE>3UIr)eTgJlha+ik2Qn*?A
zkFoS<b|cl%bVnPyOL-))ZJFL5dWUp~WejBcK!#5<%jF9FI#)E2X_rr2xRw*#U~Q61
zojD~pSL(=mRdOav2ND+6>@9pWRagIgKcKS0uvX95{1Fud6G2ME$jV7?mLBZH)aEL1
zo}ZnPq9)-7<TL$Z*frf8<1;2o9QvXaG?(GNu;*L$rDWrgj~Zn((~Bh?!zV~C@3e>~
zX5Vl*jzL8WGk=bFcmB6eks?P4#wlyJ2##qPDa{I|oc@i>EN7ykst*u4&0|VL*Dtt9
zu&UBLp~dLw86gmaq5uLc&IRH^@qhp{S`8{z=<83P5r=|{h!U#nzJee*b;YTvU@+Mj
zQgj7f`-NEwj9TOo-fkC;wcAqoKfM4BoTw3pQ;$SFL)#3X6iz4W>sZnvvdBycz=q;K
z6$s<|k|qsN%4FhHGznTL2+)NS!B{dTHVzktSKx5CiVPe9KSNwr7C!*uV<%vTrZ6&f
zNu#I#zgKdT$cEP*G5`Z858{lS6S+nrES9Mxy@G{I<POj^igN+EL=&U4_}FZA_%yP0
z?}R~tZkUK@vZ*;fxuRIc+zE@H0Ly^wz$oRrkjR~kc1{8e0>(j%;_~iTJCzw=5c^GR
z7mj754-1>Dz1U-=;>fb!&hgoki+9=%Q`h6Ge^v$|&GcE8ezW#EgCAr7e?i2x&f(0g
zJoPBQf~qKCNFkgG;Q3*jJg2V0!Tl&JGmm-tfBVsSpigil$0ri367dzl;vh7>&%)L@
zOI|W#M9!l9q0q4pf+L}{_-XDeJlFLP-&ghHQSQDnaY9i&)@%33p~%Q*y#g+!rCI3!
z4vXq?z^*dnaw6lNsjGn_`*JSB0V+=8NVG8t`I8wjlq&l>*+WelC8RQ8yl@}%D#F_6
z4HeO-SX@RJHZF^SG$Vt`!p@0M(_G-E1R(~=rR7T=36Uc8$Htb$g=dvBVL>$@5LG@w
z7P*H?pJAnr;Q#c@t0^D?93&R7-8t>r+-3)f86YAash>1wZW2Tr^JkfLuIzsNa8KIX
zq0#eH`>B!9T6ebHx|p&U60Y9;)POLuFUzG@<QX<P!&1)01`-WHLbUom09}!oWnn4~
z;43020n8aJJJYTFDNY&Um-Z~b?-y=pdk;FRRiq|Pp`^TokJQ+|2rl@jR6trlF+d6(
z1G9t5g!2Ouq<lvcti*tF0jxu0uMm%wBXPbAnKy412|y$QW>*nmaUn$-B69xU-u`IT
zcIXi?fW$@av51OhRjf;6j=ykr+vo{Z3*M6rm1~SY|NFC|0xnVn0$_-IM)vOH#h4Gy
z6{S?M(6LEi7D?MR^~Lr>ZZa;16R?L5#zht)v=~GHJz67`BGZ^>Um`0iT!_mY&~Da9
z8W^qMz$9urhwxN54y8nR_|mAKA)J_-6Iuf)tq4lOMve|eCSb}cO3Q^1fQot^mxFSx
z?P3-nEiZ@2Q1%|;e9xfr<+Cu-CcJE0DE@z^4)0T_?rG0<r8YcPnM0|6^66lzE4ZsJ
zZtRFx=6rV(rkHRrG0^kd!$QcgkZ7T+X_QAPkdPrKgRTIOWXPdWloT0g645_x(+t$C
ztpj6P7?k$&zIHXW$PYQGu{FlY>;)fV<+Gd<#(gR?E!Mux8H<C%GOjFh0p>Cp8zhZE
zt$JS)tEiN}R?f>)nUgiLYbhIS?`Uzs6>D-b5DOiSc1cJk=!S|fsSS1YX=wFrf1-E!
z-x&(=Ji>?D><kCVkUOFLg$<(FFZP*=#Ybz$85(ids?R9$yx6c+!{_PuAWqSr7i%~k
zz=4b`Gg5k9`%oH!SwsCXM9GKwnmjP=6C*G?m$L6TtPljT0q#Cg{X`KKmqUOA$&R1q
zbV;5RZgvoHaH>gocWr$oWO^p{MA^=N6q)Aze3Xm^35yQpV`#M{BO`mhS^ho+qe41G
z)z2=yLx8Aq$MNm+vHOp=_{6?#R(x?vNhzT1aWDD+fkJ3jXG9CcK1oTFa!n^Dc>dj5
z6F}QPAT1q8ic#_u&G!ltpFZo7GVd)u*%)11Lqymj<gax$9<frO<`;Ektk?jkr0Zwx
z{(D}g&eUo}?kb7?TPuSt{TsKKwWEQ=Z<5A;Z0Ol0N1rWvxDC0noYi>|Fw^U?zFF{I
z?H{(bNB21pkS9uRFaG<NuJitzTSJq_3IPCD(uS4ScYHUqsH(Ur2&M#O-0$NnMOT>R
z#TFkHuGe@0giu-rZnwbS9Fcsf^eyUOz=#^Qy*R0Fpl#+QJaDESd)E;2CewgG)Mp%K
zU}U^}#TSmMAdMA=#zM&q#eH;)2#~5NlVtk2_%8;3N`esZid~tJItETyJOmCEPRf}X
zZ@WLU>~w!U!$2~22(LW^kbMmHSsXb`MkDoFhWhl_NwO|2EdsZz-~0IMVp&ugfEw{w
zTRpWfGh<A1gb-i=6lzwhYQaS;JFz?8chsiL2ItU+MWQ2z_SAgtN+$hA`EPJtY@#tI
zud5g3;@HUSHAAIN^g~w=jks6-U&tO6WipC#KGAq7Tt9=zs9-89UDN=I6MOH4k&!1r
zV{x%+6ZR?*fitcKw0BBvTzJrZ18x<=gbE4E6|sC-FwAStFLh)v)!9)#?dLkS7Ir9O
z4}e20nW3WM{z%1BwN%qty0kY54^-oevWoTi!{Z(s_6~tn5;Ogs5?3O=4_Tp#gpy1>
zXiSKadSC+({?}8>L?!BUW4YYecJuVsj;3D~sCm-pJXMKGTi^G+eEv%+P8o$I>#7MG
zhYUzkU{Xq--H$-^I)c>z6)^hUeXS<0Hwd;|oO(7cPlEblOTc%4@8xR#qVEq^<k^d@
zmHmW!WT_SZy}Ty)A!!vfzq?UAoY14$SdRVm)8k@q*IoDF>voksxrt0h2;*0$g`4Xa
zCo@wsXKr7gaWZj~xj?)o53K_B2S=0wCJ%8a7^bRX6=$DN&Vgd_u5@A2e0<$Cv@e^J
zB1QA5A`aqHXu?v8%wY>oE>q->fPzGO)z_UZiBi0#%d>lrZMI?!ew3#&Yz&B~jCd6*
zr^+za>m~?~)cF>LmrybUZR=0P`cRM=MqYbre}HFB=)X#h%BVM~DSlM+vqlw+{(LIX
zgn1~4_c&w=oDjoh{PFgMnG^4(l|x^d>rlmZ7MSCv+{88=y0YSAJS?S*p$-O@mT-h`
zFPd_?XT0avJr>Kz%)FPk1B-4*khaaJF3VWY*_Q035)KlxOL&4(jY-F4X2#y<vKUi~
z?$w(H2&j0z5xwUMLiEi9qmYIMdfj|v!@&<LkV$|x`OIYW+eI!ABL=;dv%#Pu&oJW$
zCQ?S*`7#PS<^+w4xv0J&_$%nMEJ{GdygDhy%d)V|pAm;<FSX-~E^}&LdPHi&;7<%T
zvPYrS26YnVgAON#3s?RO*hVZ!@hpBFVUAF{98gmv%yDjVhX@hEDwCX^OmV3c2)|xv
z32hJ2y3YP6xA%ZL0aJS%g;Rrr#6`l>v<Vz4)L4`a#SzJRgRNX@NNOWB0Xe@p6J=@q
zwk6-t?u$1XUy9ZZQSlQZAsi(nqV<W5VN%-}45Ni#ChL{~6wB$!JctO(VBea$cU-Ja
zMK9`uPXnkRmB`>>rv2~F3TmbNjW3Z}geB_p#ZRYgOZWxG(EE&#EBi~^r^+YtZ45;h
z)^1YT8B+WJ%xg3+r=m-}1@yY2^aSogXOjNbB;5o$DXf_=$}$#bY=N>%Ffs}{)E?XJ
z__%w1e(vA)ydqM)&(0=&S*R;>`nelr$a#bV%P?wD??IXQ_dOZ&9m<#)&)BTe?RgI2
zo1c5?n#~0m4oaxK80$g3_++sM;QZmwYcEC1i|jZipxAHlh?^6H&4R&&c_qU#ORS7}
z($8ilWk|#f7?Z@rl<eSxLxLQ*Oa_!55t!$<HH)ZlIEO7+LB)MS&%LjluOGhbeSNXV
zXjxXoCnNk@`YH#R%!uO(hx6IK$Sh1}IdI`AEX>E?p1I|_WR;Rpif%uUmT>#CTa%3_
zzVRU)ZmoSUc)gp^`>d{eQjf`euIT66y#OSa5#h}7JGBA0o7kA)>o);1iN>fLd)9@#
zVKvQE9f_g92Z}7v-U882KPDt_x|QiqZVoeGQ|HHLq50$Uk=mH40<$p67bAgLxkc-Y
z!4_;7(@+EX4Yud^TZTW~vnaLy@3~v1g6H{b()UXY6*g!NV}%hw46Lkt=-t3zpCD{F
zoBz#XLhk%UuDjFNYYQ`L?`JtqkPOEbhG(o@G<`S(K2qJNU;6C{YawH4RV;cKX>71)
zqZdDvp)cw0uXQ|Z<2BBWHkn;pH|K&i^+C{>jT99ef6h*Hce8D-bANKEb`<sbtFc+%
z5W!u;+arHMO|b{5XU68;>~IF?R3Ja~FC@-}{r-Ir^Q8zIUI}JE?gxbpI`pBR!1-E3
zt1ZNsHqOa=Jkf6)rNza$(P(_i1WGG3J29e9jpD$EF_AJ1Hc?c(=vDevnG`17myDsK
z1Tso*)w*-@%(s(&^TH##0U8`%X2f#Pp<D&DHsEo!FxBG+Pr%U6K47|VBgf30r5R;$
z90w2%!a+crLGBJNJa%C9QU^64DPuwMgcyV;W_!Q8(t;60ih7LJ-tX7ys$c+7MHJXO
z;LAK(;rK$Apu8gG58-#TbjN5h9f(>z4c4wnevz6-+u8~9uojV>$*<$Z9M5o4X-udT
zC~b~CLOg}$wraQZbaco?%$r0Oj!K|XF$DC98t@X3k?e@t4e`AN_Zpa^#Z<MN<x$cP
z`5dzMCdcm0n(RNZ1;2Hezd|IN>tig^A4Qy>Wl^(5zolj~wV8fX{bC2BxR`jmUBv<w
zo@q#Qt7TAmobkcqkkr2=6EyVYWc_5_Yua>{yP`tvj<C-MQl(VQfjYVAR*7kb*<gBQ
zBp9WR(tR(<M{%p|kmWXyYWR6MpF|A6nI6KXI%B6#<#Z#S4@DGt>2C(vakKZ>+VCzo
z_ueG!XLwmW?oMdG{rMZSSsvq1AFL=Al5>|Ma61~kF<AlHW&w)D!^=Q+TcXU;`GzyK
zqFU(0_|jp2ukpt;%7)b25W~tkEuA-f-GvUByx>Etg!TDECu1WcvvlrC3A6NGz6IRE
z;MR9Fm9<LdEK2vU0GYPG>_))66u~$;GPl3u5sFKUF`6?|dIVEf$oCSeZ$zyOvQuZy
z?cp<hlcd6~;WfVbme=l+*wM3{Fb;MES0%AuLZKQT51e$^B4dil>}zuDa5g)TWll5e
zpi0dNtLycpNFu;Rv+Luf0FdU{tR9~J{3IV067bFfqa%_uO)^WtL%qx4yHEOdrUzQc
z$MS6iq(n`a9U~5{g%(u^b7+<pa&l&{KwlSxnSaSN(c<)^8ENp*D+5+_EWUFVK<&%U
z!v<E5Cf;)fr<WR;7e*(a6h^1a)%q#Y9ZN+L4Wln8U{GDm0k^g^LuN2Sq!2dlT%#aZ
zg-OJPx(()RFk=J(*c~L<RA~lOInIdmE??7sqBeFIrnl)j8GULggA9{odPZC=dr&_H
zB(?CJe3e|`Qg1+gAef0k^~sE1?hr*RRib;`pc|JRLuT2}u-8lw4laMTY9jlD4ZZsb
zS|`VK179a??qwe_f2)oShY3d}=e_^>jEjDHKK))t&J*8p=|@;(->4Q>MI@&Y0fcim
zGSeQ18MTahVTwr^OmR1e-F$hHkY+T?capaJR+=8A9hu>r79{f0rpvvl&1ZYOj<m2R
z8PqQ`9#KPHOFoaV8?2_6_~3OLa)gymODppc(l}8ipLI!hq5$<hqcUJf#U(S&x-tCG
z^%6$eA=7MIwd1;(*ctOno@hjZi8-!wNO_Aei$H8f#uyD5B<NWy8S5wtmB8hEj~WmW
zWv^74WDt)W!f(NcUcw?=0I>I-6@^g6NW@4Sx;~}Jymq~C;bSEp(WEtA>%*l4S5-0K
zQUSpoi2Kq$=kpN9LHhcJa(<h~no=FkYo;^!GgOB%NO2M)i3-;IMZ}4SbzD&(8;l;p
zfK!~^fw><IeKHGiM0vTPA$)qmRJD@q^<Syx6Ech%C}1(QCuB&1NjIt5+!6Mubn+>r
z?m=0KiF=u(MAfk_W%2Q<f{)h1P}CduPt%Vd)*N3pM<1_DM3X>YFx`1?rSeHBxH~%C
zZBatunfwO#A^O$W0^J6)W)kP}PYg`W*zi}Dn+tbr7Yd5h4;;fnVUvRC*9Y2?q2IOe
zXHUGa2CK+nVHOC>V={W=%L}k0g~jWI-@JE<7p7_v9QoC%<G^GuRxdNKcr_h|*+lB<
zivMQhvC}RMDXi(+LZ=5xlUKEVq`ar_86SOE$S3ig1YypU_w)H9Kz*VGlR{^Z8(W7z
zDLz)#>MBNdCBagT&o;ucR7mf078Kd*ShK^ulV*MYbv@=N@l(ljGP0`m!ogrE4Bex8
zMs^9;p&l;i0cjGfh<Q*k%;1}N!ufjO$iHXSa$pJa;%a6I#+2+(Za7}gYaPLeEeFAc
zrTzC95@w~cI2bqp;bcsS!{f(~Z-|M@NT1B0V*c*%Pg{<$@5^#)^Z@o~A04vr3??pj
z?(G?{m~@Z9G7}dqu=hVD5_XfycXN~=V-`t8n~bx}lg#O5;)FGsKNn4X=7%%GO!A$m
zr97IbWy3~)shtxKho^D~=!qF}`6kuNom#`fzD`D4*x$irVqzb6qXctFg$v8pZf8~(
zdvbJUAl=;V%7FUMS2Xs%NwCLXeVb$ARmr1ck-e%-WPWb?!x<nG9E&}edAvmI%QP7*
zc9>1ysCZhMgYuX5^0QZNVy0{`wtUJ}Tin0O{#!FP7R_z0L^>@Mf8*hJJFjCD&K8aS
zu62hDg)DfAnTrd@7Q~IWR8;8f9zMfm1it)GnVOjSk8QY;f<L_CSGz7tmfVtkr^d-t
z!g=3(Sd!C={b^Hp;pT|Y$Vpcip;<aPKI3chUiE#g=!1NQjd_#Tc^W8v?&Yh*+IOg3
zK#!(>zV|h;wcXzWokX~V*uQC?QoU}(yg*^0!pM{x#&(NA<yZbhh}GK+9pj5-*>8t$
zMn9mnk{vV4X8m#<^KiQ|#Z3iv%YPem0zyL-BJZ22Ye`Y`%71#RHsPBelIw|<n`1+s
zA$)tt?72Y3gL>(XP5+HGIK)3=cPUdV%$E}T@6K1g7OatB&`z}y!#h48n#YDrrAc==
ze{>QD#*`;~jXtJTwqqyeB20wCv+`h>!$InqMg>U~4E7(9qvHkk>2>$56v-;O^Oe73
z+&#&>klrRofFxoFiuk<5%8FuBQbfHq!sAF8XbVdLMb4nN{2Y9*IT-E>ms7Fxh^UE1
zrO6aCSE)@GoE}&fugxG)2p@29G>QvE%B6;{|0t#@#JoSakd|>|n%Jc4=M~VgqMvdI
z+aB=$c-rNeC`VRJSp3<Pxw^o}A!pO`yNzyyL(%Tms6+Q7#MHQ0@`Qhg55f0;>uk*|
z%ZpPg3iAtW!5inbnrj@t7~d|Q4cMr>Qlr0j-=+NyIG@=*yO2rcfVE6&0!L%7wQH>c
z$2hzzh>PWezu30J4NJdX#}9lp3LlP8!9?FtehI6f8!JucCVv#plgb*CBw5-<cE`PW
z#s^6LC*j=XU?w5KUi$w1bbqZ$2r5Z~^PbKWDf9efs)l-&bCC&e%ZQP9kX!0+rV;Mi
z#)>5FJoqQ@(XGJ+PkX9^*^Bwi194G|3I^q%X6$HeY!bgT<^rRFNZ^@Okkvw^Ff15y
z83Ou5ouBw=-J02%qe|=(owokcthjEb8RKzYTlP{g@q?sHdCWt&2)y!f+-E+As4G!P
zNl{7J-hhr%*2PH<NzQa|R<U|j!$sc5sLa<)l#q7G6lKdYKFtD1c$X(iZx-;y1fah}
zc8|U^BW8Nwdm$f$0w@KG0;*Q7?p(H2(>}I-Wgn7xD}M1&KDRy1TZ$!)M!c=F^P4Gg
z5_|{Cwj;0*iAzz%E&)#-Si)jkz%o=&E^>LGr#p<n6nQ^FC4aj*C*)qFQg$juZCe$c
z%ZPrx37<tc42Hn4c<NH<JaELA>WqB$EmqL!MI`yjrTv6<JA|pyK8B<~(-}+HHQC?b
z6B0Oq+%PaBmRqnbnIA}wjrZhVrK|kl7`XW?tzp9E`Ja<2@Rh9!sGoIAo-XQs0#_K@
zBzdcvNkhJ^4Kl7vb!qAO&xHa#jXRG{5fO7T?Cj={*f&ST64XyD6*Mt_)wd141y}cG
zy_r^fF;d1$y&Td`T^O`-b?;Ai)61{Hl7XMrQ(XZMsnm<mrMAYGr`Jmi-KrVCKgvIy
zZ)n<=fWkF5kl?z;jKRMz&pd5y9d-rAYBOI%B2z)1AKZO`Rc($5s<H?DfC;qUhhK#l
zZ`_3$k;2V)-!!L91yB6Km_0RLraTpSuQZtOQq|lQV29gUYY6KV2(HyM&NcwZa`DWz
z1aPzWM}}C9u(6^&)o+Q%j*PC^K~+?+iuN?S#usOJTr9byjG3}f(2o$p-KF+6)h7M8
z#_>^GJp52oSlraWUhdrl4wuxBrYb8+DFR=6na9|iyIoE(xZ#D#;_=iIhf19455sZu
zzFvujO@90MfVK=CO229SxStwvcd^JB+H%Z}?UH>O@89Fwr@K`2exmzXKU%Z1cXbv1
zZvI}Vsu?|HskKpBBh^CE#7ZKdBC~O96kJQA^?qf|PF=6GpwRRXXR{&4u$q}Scf`rR
z;fCfJ=NdV<G>9<b0XWZihz}Qut({#;Yc$bv@I4pXRCAK`LJI9F;xCA~%3Tp1<T^!h
zk|R5r@3-YZw#WB@Z(bipcgBD8l@P|KwU$-H#7JWyNxkewizkrYp2Y?W6CX-sy84UK
ztJ1ySN?=}LV0vc`2h^6y02k=*gp&l}5d2^0sGGBVI>tKAf&Ddoaru~!Q#8QK^hp+}
z*+m|1{!SFCsq&DSdSPjiU+8E~p1YIF=JPih%;~!hTS|{m(tw1U|MqU%v)jMDJS6_1
z(m@3L?kM18<9=whWwG-6S-_+0oHOv?^^<o`fPzet0x}~b`SPaGpV(A+er8#2MRkj5
zgN{z-FE25e7z2Qv*+4XZpg~ua3jKJM`!GlT)YLx8{Pf7b%&9YFYI<|v&ezSG(_$dP
z3}Q?2ay8ug=fQ^5)<uunZ{2rqX>AtOueICSd{T~v%4tTXL<VeOR6=@G6jMingc*!g
z?b;6L_KV8ON*q9-tX7!5J+hZhBhF3Z>$K&VSB<Iu2&sr_t+tHzEpb)4Ym6tECLSMd
z!Qpg5C5x6zbZMuQPeU$#|N6oEa91kL9j>klibC<Zu2LY$O7cm9^vYS31hHw3=tuzd
z3K1YK9@%3U+q)<)m|s2BIV1?U5oFoeHfjVdE4m*LN25lSEu+0YcXU`I$LsSZ>H|HI
zm<PRLY@lWegUnJH)?dEKKS|ZYO+hi0NzjjpJ$m#{$4{?wXP#uJycyAZ<P_d(>rxtK
zb3lO%)4xkaxJv4(ge+@$@kir_cDjn&*?Y=J_YaIcJIhJD+fJj9b_>y`Vy1sy=4n<D
zY%-mO4+e9dARp<$gjZX3>pm?fTHx3z3o+1&Mv}a(3O2e{<65A9a@>a&MDN=#68%UD
zZV<Vl+B-rQb#y}o2#P&D$<pJLqm!QZlg(sn*i$gP|MLmZLH|UOJh~z78mu1ly0(Un
zD(v^s&#&)R*roTH3EU^vlIHis9|K6sc#p?UqyRF)lk`*a!AaZn--<8$1qIAtA`H&M
zuVty`U}dMg(3gPxd6v#DF+G+NK&nlD@`8Sw=Hhzl#V`+?q1;5ik6D#)N0B#nfNf7Y
zoja((x;~#@uRrYFUfLTdjjE|%6}96l*<|x?qqI<c5y0l$5Fp0CGQ9Gj&VqYR=gZR8
zf%oV3i%9pW-95#C(61kblu>2`5}10(BQ%%t$(}etq?X3i7{i{<&~E`nG8NYcA^KDg
z7IR`hU6j)_<sAoQouEpR1~>NipQDgSe`h5T9iewje=cl<eX1g8q%Ezq{Y*P<uU}F2
zKDsBb$U^Zs@C_c<VA97z(9HsL_s)*#|EG_;TZjGY=)jzk?4K{4tq3&(pqR%R9OkwI
zA4%PzNzlg!uyFR#&?<*2AJzfqDI$*6Y!y=HnF8HtF280OODoE0rC-<~WC9l!lKk%*
z5|$bV4>O}$MXIFQ=1@#~`iP6~KV4ZWZ|vGb_LFmale|K?njjTCL+%VyTpq2MYgzfh
z_)AA#=6T&mud=mQGDY<UVH#~e3{JmPxR_q749u{9<LS2czNBfnR}R7-=)A>j^-7Fj
z@Cc#I4&oy4+{^8Tf!|7yr#^dlkpFLP;^ZS|=wW%gr1vdevLVm*j#5i}xlPHRIy)lk
zVRYx;vC6CJr`PW>Hy2=ph4FHF%;)X+Wv$I`p8wf1#4MHl&~VLGlfazDCQQ~RihA~o
zsY|sO(NRP)e2Um(oH;aO6@C97E<Us*a_^|nM_h?*KH=toN~BbQg<|_ux^eu~nZQ3&
z4yQ`CsxRQSch5o+7hh*%ln69DbJoH?1l_ql6p|J-ZyK6^r@APZVY^H4`2(}-pM#j+
zyEVb(`%PVfY{jHN#$7KANxi2TBd-qXRPG)HiI#LPcumOCPosVH3-a%`XGiu`9}k#>
z{|f=j*0bGv9#Y%y&)gd&3t~s}gUXNjek`1BH`z<GeAK97d9x5p)Kl+7ij@<a>1nMv
zUp5&OjB+V~AR{1lXP%(0BlS0zBu1I+aDF0HluN}=x9DXkeq(^S8>SbjCMBYw3Qq5F
z#Fj$K=PfG`HIn38GFm3nYpn^?Z%C2+o}K(lvRse!BQ;C(_3w6vsTZfe@Co+u36aX1
z(EsAb^w>73JyFt*@W<~-LGfMrIpfv$G`#^2y_^hopO+hbUsqkE8P#i(Wq{{6kwmt&
zO<xZac9VUQj*{HRjIGYJnj^5vY4COxN!KwH1AEv*k&)l(_Jso!qhO>1Od+xNHxBec
z_iM0A>g^5FC2#o|2^#|_&<32@YbPc7=iKAzg@LSOI}>S0a2h&TvhF|79T<bjye*6l
z1%3rlLB#+l(Rzs5uXaMaU${(Js{W`MGIv%j0m+naMMaoIv*!OMSyjUx=<BHSKKRqI
z;G(rO4L=<2rYI;L)ifV{P?vPQ(Q?fF+8g5|UR_g4(SG5am0p4e?nWrq_<Jb^mb0#Y
zC*V`o61(Sh?i6MNEsZ;dj_ejT7>gQ6u2I#t#>joQWk}D<aqas*;eey<8pLBnu%vO$
zl^>%%il5YN5!aw`eS;f+e~i36mFrd2X<W9+1|O_+fAL_SBnAkvoJZywA%v81MhPyr
zhz~5xmi7CI&m$_Uo(4Z4)RRvynRs5VPKVa{H8;i?PB_%V0{O=NRjl=Vx$1$7%K=9T
z64u&UA(&6f)&+r~S!GBHaB1dN;-@dElrR7Z{y`egmfM1wqXZ(!OAz1mNumA$L_qm3
zVm^vJTXoAAFBwb9(YZJD_KX^=BT{;d`DED711DG9>~{Qq8Ku4Fv*(F@_ce4I=A5PE
zf+g5^{$2*K7Xn*4TfZh*1t$UWVOZFb=s1!rJl8x+<Lz4+v6UKB8z)WuR?>12rU?fL
zIQc+f^NE;EMskO=&xU|aj}Hiu_{M(W2|+04`j)A(&J4-9JSe3jV1uO8$w?%fzdR_h
zc>s?R3m1OpL_)KrO#d=lu20@HIK1GgT&sZK(dPBy%^mh5?B-{$L4|Jl<sTl)4c)H~
zLvVxlFkdwoLzztvvm>VsRtPuOFb6h5V>Z_CQP^ETN(N~E`R(DuRc<x#?(55;U)^&b
z>jdw@!r#i!!9@ZBT-(coGYI67bZ>{6X*vY0zN0UD<tYUvbv}i>It~bK75(8Au{B+H
zIB*w69j`0(gGwPhF|ppNm4xg5*fX~(qdQ+eU_>6O8v4Cj5j9v~)pg5x$BLO>=#(*H
zJh40h9rQ=E(+0?ULu{r@rL&d{QIUuk+?MUO--wDMPkmqAj*CpQEH$Sx|DIJn_k)W?
z?d=j?oB9>Q-NZ=7k<J*;zheTtR?sJfebTa=Jz^~IqL~oid+QPVm{ngAJ8-|h6EUFl
z=Kwkv9SGhMAs?cF>Pzv7VJv&^lxJ34uq*GMdM9BWou4~fk%LkoI5B05rs=)*wo>v~
z9KuAr1P-ymB8i9)J+QFk*yKAIU6Y)}a|rj5LJn}aDURzMZv;($`F1_)#B<?QnPJZ#
z72A$KSwTvKtV4^vxf->!AyGq&VF_K(PcHX-B@yQ}{Z;Vjv@L`%vtU>lm3lYrNWPLh
z<jc&H69pH73bv2)iOnD{fTKP+e<%iZ?yV5OBi37gp>J<Z_ZFiFqQ%s!u}6_mv)k>i
zEkEdD#Z7Xy-sh;UYI8@qO{Sb}ZF!ob%VXwbHp~mOcm7S4_WEBb1pVD?{AbMZuZjFf
z@nr<%%a2ocUbe{-0yaST#$t$j&Entda_t}3XGf8>|Ds<7?YciCS85C>ZTZ~Z!dhIM
ztJ(vfYIvVXY*BR1T0D-qKJ-R=bwl8Vdr8{U?>aa(*7p>Mu#qBmyKZq+k=YjL(Yx72
z-{re0^7iiO8sTTwOIK3^N;D~R=K9j2D;K%N5CM2v0@fUFGMztc9r<;<Y@3}hEI43o
zCoAxe$V{n+)gu;?8WZ<b#VDOGI6AiM+`Y76#`EqgcZDO;d=5;#Y*|Gb0kF90D`RD4
zRb{suR5r}~SnO7L?CQddIfyp`P#>XXbBB$iC?VsH*8-~?pH}cb_b<e1#gJeTLKBr&
z3^+?tXe%=e)!b4p8<Zg(nUbugcStec_WR>fPV$}1@I`xtiOtGVITU-iN=9~c-4C)d
z<SHOW8G+=n!zec`2Y+)%V`)Kh>a56K4(w7Gi=J>9jjtJ9cJ?{4yho(Dh!93wsLX`~
zRSq0fh^t}Ea4IOS(JwDqqY+W72x~1z7RM|gIl7kZV_1CHH~h`xPB$xs<cKM%i-S=?
zQ`88>#fXg~bTl;72r`Jh9^-CTN<)Z}_9a2bl`*aNw*mn{tH<*!r}X;go;V7ZH3GVi
z#76m~;TBFHvU({0uie!}h^J@+Uw^Hv50vfirKb$P9)B(B&)8UE3{_Q~&wq)8l{)eV
z#}5x0zI=B4{mJoF_QO4M<$L_*@kbOGPnZ|y=P|(!f|kjm3&MBUG+nXi-@1pl`F@nT
zKJl36c(L_DmD^p1uB{JCN$;2G{`7ks<SE|UrO4~>a`8ZuU5%uP27AMdE0@GgjeUCv
z7*tUf@ou^E$b`A?5q~*5_MU17|0=&s#AS0ixD(Q=ZhRUkNpm!t%9}IKlx7rRL^(mB
z7Q2FOMvG*hHVQY};=dmv#;MKhEw5r&ImIzqb9uo4!=RTL9Xu{@==Qgb;w#as;8R^$
zq@k<TR#a5LW=Iu+@RG=$cUHzkj}0z?dnc6Dncjggm@C6E@*7qqajWhdy|LLXTulpR
z-a7TFF5P^Y_|nk(<?lSRgW-cF)375xy@X;vMfD}Qv6zcT-dCIopRaeQAsx!8&S~(a
z-mcLP`~@vpE~1?E5Td1+SUvXA??;8KEc_WMA9lSJBHqfWfLJi-tf=PXvR1R~C*+fa
zgm+M1ARrW1dHXJ#y5X*;lqKy)!!HOOl`2zL9BS{u4#QfQ5~R=4^+#{mev6)RvBKw2
zalA|Llpesh5pS12R7O31lfB)f)jayXHeIEP7OQpT-MHLCmJC9%&PD*L#boCy(OQos
z0n&OKEMiYh7!!aLd4)t^7l0+WjpHlEW|fgf5DDcT)7EOb&+G%pBG#z>q!5<E`k8{|
zfFcHD!ud>+PFl!6D$DA`4_YV>HSLWsESq<Q;jnGc|Li0Fu^TmC+rOzT4m`PGez@`V
zw+BawyhPC}MTWn=^556*TyG82%kL0WWh?F^C=dr3V+0PvY{JH#Ifr5b6z;Xj{@dYw
z`fvE%XNJhhli{FNAz1N}TsPB*f!m?p3!Z_4`zk2~GT~u_QEdx$06WSy41%9b7{RrK
zVK!UKGH`w4=vh=`N*oQBNK+qQMoA^Q#mK1F_$W?-wmvt=8QqIuw;O1wvE@INtDfS^
zex{X$=|m>OaSVXI#wCNK0K8?wD4?Lh7Z6As6PEklSh;a*>AKX+BEb4g#ZU?kvdtK1
z_;fded-2oF8}99sPs=Up5<WrAe=7<w<z@APBV2w4e8I?8b5~eWq&iI`aW`QM5fYlp
zsTT%`*8_pha8l#jGmsynhBi@MCXB~pQJt6v`G1oeUfk?|jUUMtwd20%?^S6U;o-f>
zJD#}YafeRa*V}=q0~GGfadj4vyj!Rpunw3eZ7&sYuoEImo_(1|zaJf~Y^fTjp79>6
z%1;EXv#b0_9408DLs#qYN&uw4=_sIFZj(T<6J2i9sR1o6WK0%l4Ht;w#9GvMATjJ2
z1<VYs5{HQ>Z;H4NCjWJl!+H$gE^NJ?aC*+l$Pql&Fu{xIA5!0cNlJ<P&>x1N265^|
ze}X{RQGG1sKlmevaxDe{e*Nf64q7ua8B+$TNJ6~(OU)z(&>w-te3gX~!(P}l)%YDe
zE`4f4;6!-GGva_9w39b>Dlh`+!Vu~y$xcGv6Z_Fv(p0g2Bu2*qD7g6>{T+QJQI9td
z3RiT0S>eWlQ*u0r&R_NtAKt7|x7}``SAOBw(h1$dppHMkb`AGK&;2@s+btD4Z{+pQ
zSl8zXUOQ)X{5u&*;Etkxe#$ysQ$n5MqDx4^{3m(!jt8NT&{{suPkEBU6~i@^`^59<
z_oQ}9xO6S8-1MT;0z0KQ12LdZos<6vv?z&3y>AFNt$47ZP{niUS`?|<K0#mp?m8-@
z@W@hTOVE{i$b*AgOxf-u%h_@UpW=K1&=T6Ov-ah&ws|hm-y724eXCv!?@zD2wgu(!
z8Bnv8CP4&I40jB8lT^RAV7}S<&URU)v)bfNupTUL|G)#&Pk+39>OREVZGAWKJ*@XN
z*J1Xz^d$l}*gy2i!<F{MS@-paH?>Uv@PuOY_$t#okNeF^4CQsIP%T4-s3+L@Q<>^b
z>J<H7*giBH-OiFhTU3zX>?C`57@u^iI*MyN3F!HO@RwAf+=o?;W5OVJ>w)ZwfOGep
z9bqb2Fcew<H4u@JjR>H8e(J5@g;bpQE<=^+hvI|DNxvls1F`qCODaR^{bick+Cl@W
z-*bv`{RwF8yg2&nYgc@A)pmNUFxuuk>+KhUJUjjA+uG)18^%7nBd<jeUJP8Y1!g&U
zChHG4A5o_8O8`w6n=WKHM4XoxE$nF|NkU0M3Ex?QMnn^Qsto$|wT<NT%U?@q^3lsA
z87!nZ!h$CS3qfOnHBUDLV;+6Z^9}_*7M{f{-ky64$9_twQN!|6#d}Q~9o(O*#F&@^
zuBx0mWK(%%7Xz-wHs6J`-vzZUt|v}>7&>&E3<=uQmC(iDeWY0%mEQ&-o_F_;;@@G8
z+vz8$?*^WB*;T3>VA}r;hSoVN*U_I{UcT6^iNPT*`aTd08M*R1_fLsUm%>{r7mFRW
z(uiU5Y&6dP5Lvi?Y_{g*Kb_!C8rjpT5ND^Y&5<*q%8;~}+RQ5_(M_$t1m7JVF14qn
z7WfwamL<jLTIfFI0XIAU*v_+XwI#h8EhT^XZadSC)?rKozYo3(^ckNHeW|Kaq2pkk
zq7Y(z^$3t+!S`5>(xNBsDGdSXbyP@>0<QXlZU%KrO;^0Rv?K6%{pRLojR*5z?$NT|
zR+WJ3S)1xB`LM+KERlP3@;7!UlKhy7uA8T*E-c3IHL<dKNdVXet6>;b39*N;6MUk1
zt~(G?rC$6?#m{rkc2!ATl{Bz9t4$gXa1=rBQX^DtE2xM+ID}caYS66JQoM<8ShhQ`
zx?7OeqYs)Sh=8bK`=PCV#7Ie_ce(*(%R2az1o}s=?DylV9kp-Ly(EhbxjIp?i*GlM
zZ!o*n+a}d#zHm~s!+=S(_4zV1B{n_Cddmmz<t|srrBI1n>I^-frGH8T@*4cO%pA!9
z(lp9%`S188alZMkFT-sZ<7bap8DzL4Ouc}36<Zos7X2#>R-uOa1tmbULM<X7ovlo7
zH^>Oa*qNDT2%(nU-_3KBVhrEtCH^Zl$E*b=El)gDQ}RiXnb>c%uAZhriqgI0!#$rQ
zf#E;PW!fH0Z??zM>P0ZwO;r^>-bp2)tK}ld9Tr<ZSXu2folKd<`1hIPzl^aQr26RT
z-yD9G2dWvdlevcC8QLcNZU0s&;t$(22xGFbtU&Q|qDYqV)~5HRj`T2+|GcvMFp~a!
z=)=VC1OfA0j-ExmPLAS|nvG=7MWn8jT4g!8*C~OjhEIphUcpnEKLp=9nD(TT_wkgb
z3suZzk*Zg=C&g%Z{F<Tm;UTF>N3!v$)Qay882jXv@(Zx^B4*6DgRGw>ajwfqM&P1t
z9oBnX9|<c<A-_~n+hAcQyWgs(Ia2m`Z{{Y~)~qk-cC_V$my(I8zR?SSI#nsIYF5Qd
z=7G0AH9>-S3lP2OtWD3{T7^(p+Ne0s{qQK==k9L>v_SkAr~L<@Q5vuQQpPU1zdx#S
zw=2C^;sae@<o)-vY}<#wFM_=m`Yf`Nhkg8xvN;k}wN5<B7?F`(AkfSSaEcUK<TJN&
zK1q#K=nIYgIA#pqYK9BBx|9-^LT%ZAE%z>v<|)@~#=z3jXynCq*_(-YM^~o1&0)@T
z`WF^9BwuVX?g-0~rS@eyYr1pn>7qvF`(8-wMz@NRo4}~T{lTiX7w<}L(06YRlL|1h
z%AF6uNadZG+3ld8OH-4``M<@%ZZ~TJ6DE4*V!)17X3GOSo|lE#_VHh=AphO?pW#T^
z$37&}wne?XWvN#AvW>5aU&KTUJ=gnQ$6E5U{nbVWrrjjAcJG+q-8}WqYH`bo5uNvh
zR))rFuL2A^9181*qugFTnV%{>mH5roBHxzaX{k<nC>h<x3*){hKj=#neLtgZbQfHA
zHizotj5g11f;&3L;}LlW(x&%a_yYn0rp36mPXpzsp8OtcY~1`d@crv6_e}t<oROv+
z7r3*5ajb*L<m%k1UfZ~ue~1$Ie|H%s;7=CX0B$ZiPf@WqVNTL@U%<wkP=B0u^iliK
zC<^Yvb(zL=@Vt?d8@1|XT<jSOmL#JopYEFuE4`!Hk3*37Z-#%sm19Sg)a%Tn^-1Z~
zhB+>9J$mzgN`+n@P%Ixye{Bioi0r|Xl8;~*q*R79N{T>Ar0dt?QSzQ#I0MUYBGqGZ
z$Z3ZdBrX}a`)nzF?y}Ah2jiV|fBrS-Ya`7^8Z8JOeA7lmU{gxL7Cz)0{_WN*GD3QD
zHJD7<0Y%gcQFG|&F;v)lOY8bXO`?hYr~~e_Q@ljjMezFFF9O^oeFnMQn=Mfb(HbPi
z<jhp_=UfDioyCXZF!Zk5Z71+qp+HFQkd8!-rASYpUd8eeq?pMG+RUj5IK%@YGEyA-
z8RPz>X-s!0&wj^T$==?UvtsBoPKAnEc>nyz?2!_H`AlG?R;24%KWzTr<zs;TnWKYK
z7vB)D!$uGp2^}W|(*PR?Em}`v!n+s{)YOhn1N`pJoo)j81b<nGhmED#3B#{^G#0mp
z`y{COV76`Sg$}T?^m{dJ2llBhdoRKEf-zZ^jg~Tk%gQu^k5^;sTTC#&IOlQmPC$E_
z?)>p<R@isjLe`Xeqe|A_i5HFYqk$(89{1Bq4tj*vl3fFAt_V=5Il@xstZhtCfQ>Ug
zYle-ZPf%GM$}!p;RClB|OS6WhT_8by+DCcXpR<4Lae}8_tcC_Le*3PrtI1DTP6B(4
zsBdX?1?TC!pHR4RGo#ad-1vN(Qt?6-+T&qbYizLr^Q|uucc!{$<aBR`xOT~=g%S)v
z2D@lK#Kij-&YnEl+%pvyaK{|D+Ym#0M&WXX4ENQEM3KELmrPwtkHN{<IkRdl%a}Oj
zE(QRJRTfG>sRl}EshV1Nj1+$5Xo#(~u+XA4i8%Q6-6{f71hLoCS*`h}edmc31i|ox
zPtLosE_2K!e@ne$Ijh2t*>#4h-NNFs4t0n?L{|SZ=>b<G%E?`VFl`zBlIL$~R-wZ@
zX6G<({C4ba0fN>%_j@qL;Bk3S{0tAy9h-yT<Kl1N&j77mGBfQt&iAcwrlgPLUT#F^
z^N!FYUE~6SJ5qz+0_w-ADHH=2W{F^?FFa5jC|W3N*7)B=Z3w1ynYQ4Tjo`{c(ffES
zm<s2?YUiO*1g^#01Z;=}Vj0KdCn&}nl71AueRuh0ne{L$Zz<eaFCOXe<ue2O^7b0%
zXuD^X>a4cs>~&cQ6z+2eGR?LoNChb8=<acAbo5{BW*)q7@Pug|@DPKHk6A*n6}G3%
zcb@JwdNkeqU9iO)OWxldYt!5zmN;LXZ_N?ea<p5a33-h<4GlVy9R<-34+?#+?|heh
zok#G&DUvd$f8MmBW2{S~B;UNErJHBG-ibM=<L9RLc6qTP8R1C9ansY~WhU4;9kE2x
zYkqwflF~qqC&;tqY2D`P#mXf=tL4O-?uwvVC9yCDb^dqE&fE4xW{oNNl;MzOThYdS
zqWXilV5wV7SDCuh+>-W^;G(gy^$STj2+ko0x7`{Ehw+gDYXvS<&K8&62kimSuWpfW
zaH&nWQJ2WOs4jp#(t&Jz|D{as#I_!DoFSd}-bli)p77BO&S5|<I|nlD6o`#cl`4R4
z9ngNRzR}d4wHV!(dz}UmBlk)4$65PHH=VXTbi(-VJ_sJRurPm$ZvizP>sM!&x@G@n
z`?wbUQ~M+%9Kj;DD5%w)TklE^vRpCG2&yZVB^!+temF;)MF=V@eNQ9~;R*-{4_wT8
z-6-F=9$SD}sfq;F2nMC0JKvtN-BruYdnk`hE-PyTnh89Kfo7^iPwkJKYzynWu-zXo
zDlr~`2;{5V>rGDGz4QJ@bCpLzuj&;4c}SnfhbXvp)^kp)RX^2PLPoUpEV(IOE>tVt
zoKe2RoaNqLcK$sL8>tTcTYr0c`)4Ke`1IZbnNiJ)nr!BAGn?$b@4<45v$L~>@hW5@
z%S)X)Bk&T3l~RanAr)gfF~p1Q@^#Hyp)HIg_X}!zDx$aZ$3mxd*<S-Kua`HDb9cS}
z?jG^XX9D9%Q*QOn_q6XNKo*O4Q~!1Fn!f$polPGnfxgZw;?hawZGL`sq*S2g1*rq@
zhO*N1b~GAY!PecU!V3R<VJ*?G4XBU0Wbwe2zbs5Q(uEO}kbW!z8Oz<5P&&jR!I*sL
z$Lgn@p~ve{PUaW$*K#epp|A3J6`P388P|SD#SQ9hwDwapYyscS@VgDZ-*n?D*hghq
z)kENq>yjliB6e8B7d-<0890+Ppve~bd--jt6%3i-%AYxiaI_7Omno(}QQO7hI?kql
z`d(bRn{C}!HPx&0<?2e}5{xnhFi*5{lQq_CJ(YcXNz65yGaDT9bX&zhs7+u`(_?AL
zi-8w@yeDmz&QKu~-$;k<$q33L$turmv5=}>M=@v(Mdq<Q1*=i1#pVM6cb{_0+^I>Y
z!f9E?;fDST4AFfHC`(F`_!$yuTjx}bjV^E#))<Ggh0b}<*DILq#=K?2Y^VoRzrC|2
z_NeODv;28^((X4kEip`9VMxg#y*R#xCblTUnuWL51RbMa?uA^~8w5YD`fugsw^OVD
z>NVRThj_=o-5+erm$PlGpQXI#3sphb12&HMsg~)Mjh}9YiZ>It>+ht<REZ6Yy0=%Z
z|Ce`yxx9B9bNd(b7juJot$16H2g+Qo;5ozpB6Jly?7AQt>Ho4x8apD8>LyLo^wlg7
zJYm{7=(#+cosV+?VmnMgR+0WbZQZhlpz*N`4W0OY74=7uR38~cA2pVG{=5Qa-MpE3
zx%{wIb>~i+Q?b@)-nt`tj3(-9L}TE+rtl+TZbw&n8rUmH01qad>~qu9xMTNfU74nq
zboF_vb<6DsS+4#dxjUMi2cSxHG?wiE4VIeGNHb$(U#}CRBUUEEdoEXa_xEOb7E2kn
zjE2`-N9(eJ^XI;cX)ArTV2+L;-pnAy?z9U>5K=!GXR1qc?DTvWBrv5K{N2K1uW4}y
z?+xXw?R;d$GKv9dyBHoGZA6GkSTn@!MK4*foQk5dEBgTNyIkoqGx{+yGcpi*K_M+J
z?ippDxJtFh_nTs}vcsg6lL+{0IP9CtPd{{%_*5ikW>pCJSftOH^R7bRLQxuxFk7T&
zxBe3p`;|S+g_|b{igsc8ZumTJ+cdpnbYF}*6$hpF#SredW&p(T!5Nu(6B35sC_X~@
z=|A;X;RMjM>0WGb5o(`tDq^sfvb-(y(K8B?00fF(d9N~EYJ+6fP6=e=EDxAR%qiK9
z)W8is0D+I*936fZ%lGS4R8sl#RX6lzQ1K70{HZ+Vcy`yMDW(=dE*JQrd&wGD{IDP4
zIs4iV<LvNId;dq#Ihy_5?#bgoQT7)nH*IrGiP>!b4^3wo*7X1N@i7nv5+ejeBql##
zfRwbfbax}&CEX0k(KYE%T4120ba#nJx0EnoAl>)=5AMfY*B)?v&-tA5KIgn&sWt##
z{Y|<bteyFcTQ^CMMG}X>e|gIFzv}nGm56$>{&GFUJ??kjqaQGk#M8&~>5M!5$A4}T
z4}*_iK)ttV>@pEKyZUF94Ur9kFQb7V;;+VY4Nk<Ol9GkAQf|s_-ZXXA1R0VBKpGf}
z#vk0lZf5p$<_0Iblu$qqg6Lb{EZ4Ukv5)4qiyCnp-WRJYvkMH(HXiJkZ;~(<ld4@i
zi7$@f@UL=UW1`ST0Z!k^ES9VQ2b2lU<uwyvogq>+Az7z^xOXq8Czm$IN&x>AFf$Ae
z^x<ZE+PR>!O=)=k1i&oynZFVg?h4b1T16<?K#vGWH*mKAbfkl)*Zcbc`F7FCaVo5w
zO3d_rf)GHQMhZY-g%9$tORKnY;<V8AhsUd!6hY}TnxSc>n=;pdzA+9QED)(HCq$*E
zZ>jwE6n5EOfHg;{py1plced~g92~Et>P&ZKHS!p4PreLEz(Iw@K&6`@JFMs<mhfu<
z<Nc4Ny{hP!YGn>0U@a%ig3(Mas?Ua-A7v^Qsw`%b!G`2A%>fc^TCn!GUXCA4Vl219
z{GWsHhkuJANXr~(lwZc>eV~i_LtN%BUZA$?&_>55mD(c}SZVdOVfK6DY-($jslJwd
ztdkP7q~4Y|w5Od5tl^GifXU#H2}cE(wWtX(O1WJy8hmH?h;>0c{;oiy;=$GZ_N_kP
zO?Ko9KD!>kalzfb&fv}Ppnibw&A(C4zuh@&AvJn2iAPvj|2d%KB>URg{WKUL->Ukn
z0YwdVo^~{S?`c-p(=x=(Yxcz2p4+GID@^19lZ;!w&1U5f^&<&j-(g5+g%HrZMH8>T
zXga^Zcm;DIWQ)}r&q38JFIxn6R^sONQ6adk)yb82RrJMxG2;7d%BD9sI{19yQv5Mo
z-R$A4tdmHdH*PhPggBl=rDQDC>eOYArWfmd+Z7y+o8s>p?GVdtE1+$O?`whkhx=*h
zDX;n_94`Fs6;ZiMI4qs)FCY{_I|Sw}`n{FqS6iI*eD5L1OlHf{wyHvHy583mNkuU0
zW{A|4J4yQJH(%v5aHT%auUI;kf&X)N4jK5_+sjf0!~@y+@l0lHztT;bFHl+%Yn;y}
z?eeD(WI!bI`I`#IPlcu4v_Q$I?DYq6jsN!ERj@&o6cI8-*0A-hCn^L#=|ba%#>a^O
z_1h{~Vu;VujSSAIF1AryE)M$oU1PJY+Us(ff%+cT3F^BdFF9td%y$kBW={9+_RmrN
z@4Hfx!P7s@2IM3$C*DKELmVnN8<Ei0VsUWpxNpZZ8{m&M9U(-f!e@8q8RS7X|Mt1#
z$-N7Um5)5NUcy4d0K(q@$i>RgEwH1LrdOmB_#^kToiS{Y<TUx$&r(R|H=S2s4u?``
z@PkTd2X1dSPxuCxa~+P9eYd~75B%Lr0jf#doedT-5z8S4PVJrsiD*kgZCIZHo>LP=
zS!kSczd<#?L2=2TI*kiqyQ*wqat4tfIdTyt_J&t}-mt$d`zWkx*6)XP^}hc<$H&WT
z{we~Lb7xwU-Is92C%aroZ>%tT$T5DKnSAQU!{4MTCW^B|EOEc>g!lrx&N(%l9IYf1
zA^vxeZM{`Zvt#9B;=o<<M=5)cPKq=%6ZfUQ7bS!)No_@a;uVO#oLgaGEM0)-OT{ra
z07kN|;zCz25_q~cs3q{N=Y3@xt9dF>W^7=2xSTh(cP^Dg-k&=3<MH+A<omU~70u*<
zrQNkp9&bz}&=cT~kq65WHGss@`<<NtnhsX!$%2H2p$k*=s4&2_pc72WAd=U?i)zqI
z0(0gYeBNcUjR1Xp63L|){o@x{EQZ~|0OlJUCT8)gx>c<V59dZG5%K0R)5u9NW}!9r
zSEJhIZzC%U)$ZKat2J6a9FdLQhK7;jLbjlSz*U<mpKq*?)Piie<(625^dNVus{>rY
zc7-FR@cX`ui%wz)AYb4nNPNzS>X0e-->V-Es9n4{TvN#*EUuERkK4&#SK_vwZ^O<F
zuNByvha!eeW1!^zz9igGr^{KP#P2rBA2dJRM^Rk3ushczkdfBx7FL|r(k7MN-jWLy
z>rlBw(KU7G+sarn`ADWn!xsZzk3}SBJ!j-m-Z<S8E!e$&5Tay8`E;~GuC{h=>ZrIF
zxJ7&PgK&YNq}U&mv9MXa34r*oOkm@4jLDK?^`-&vA?V={X=u)-g0pt93a^Sm)M^k0
zKGzm>(w}izHeI;FFtHdwJri(#V4Eo?dAA=|gLAxny89|4Wk{@DiEJ|&F2x-SA_r3L
zu=S{9rU(HUga!sI=5xt3x4=@D)fKbhUA?bZHqvQu{&dp~#LFz$XBD?^!QT0|Nx26P
zrdIzH@wyennG~&B`T3?s_a!N}Cm1C6JlYClkajZHYtXXST#)2jbF_OZag`?FcM$~t
z>75w4r~5JhEvi85#y}bekYpzI9d$PbV0~W=bE19O+Azc2OluWv0RoiE%P;mN;6p-l
z8UCy|rwhim8~YAlT;5%%!oG0;!-3quR(B`ATGFtB&Z~(-`8nim^M-y6iaX?TGfABD
zyb<y4x6^YLieEk+Vi}Od^{f3kec5!IBQe|<jLpz)I0ebmKP3L2%PK2RK7dZQMX>=c
zE)I)xls{)E$dJxbIDLKTqFBp>&x3OoQ=~cJKeYE&@*F<hC{6f5UH=mu<G6KylULDj
z1W+F7t}#k2se0$!_z58Bx^;7Ze}+?nIK384v=S*U)=`^AINIDESdunNyz~{$%c(>$
z)$N9)+hmnOFB{`F2M4l-Ll$+#>G1J=`Y>1=)#=}=gAV(pW6Mj>z@k~SERAvow>gjs
zCOX;6?R6eTojg8(18~Ip?9JNFt}jwBIn5_UKw$<KYcq_zNp29Qg%9Cmz6*UfV?5`d
z)%#IH@7eHy#TkFDo!U#DxuX(XNPgm_)$1Nw*CXK@zRd5JM~fuA+yuI@G&gOtllc<2
zcSTe)pT>WR=?3T@D{?Q)e9Zj!BR&Gl5Z0M!fM@rLky}b8QMUnr%NyN%e%h9sF}vdQ
zl6bCZPH+;wJ&!TRyS_Gi*v42w>33e=oL5fszUXpBW<8gBjz{5h`IV4fBJiRevnG5~
zr;cfdO+IBKPF=;D#PygNa6Wjd2UC*>qS~8QTOEBpbx*y8f*2Uzd}ja$2zZ3Iz@A1Z
zFfxw}^Xd)djx&h%Oh*X4pB<3EX~Y5~&oY8itX?;Tyto;QuF4~oLM>0=f-%ifi2|~u
zIji|e1dsb+*+8VaYR;xxhsb{C)dI=G`u6zv`SsP!TP_=Fg51R!p^5-SI1XmZAu=ba
zGU|;xgZ)VQ&Y2O8K`%R#45%;|ruKv>BNTfW2v)Mihiu_~Cw4r8izx-LKK59#N~}3l
zlniAvY+Zfg5c!gcg$1q!=Kh{3<1*ud-IYCl^DW5u@ZxJ?6Q;14sk3PNw4t}}@|;-I
zG|$M2nJi|&v(t?H_qVnCi?~Xlp#It24uNFKUk~6<ngtgw-w!cxc$`8OZ0YOe#h-<`
zXEn!oAz$2-FgM4TAsKjuqex$HoEQI1^hdB0N|yj&k8&tptUW8?@kAxSJ$U!4J+t10
z{mMU@5_Dm+eNM@p>Izy~W%<MgbT}b*_~=5T^M`4nt3M}s;(g&{9@nUr0X04#N~;x$
zUz6RdJleZ7jCi}i3<*HPdbtl)bcdnY`#+78&9=cD9^wThnCktxBV9!W=P2*2$fRe!
z!cD4IY~a+NAn}^>P*sB0#)-;M-@)52YH;)RPwtD;r45MJh=}{m6F=|F^wZ~p!;>n+
z9VK7n3v~=55)NDRI-QXE`ueuBIY>G@;WvA~%UGnLV~{>zn#wecv1h3kOv(sk`4)yH
zg>U+1ZP^5NcW~oF8+(Zk`Zqj&KCZ#W83m3>t_}iwZsDc*IIMmkR~*kcqN@N<{Gf!b
z1RG*Pb8lG$oV4lX9mwA62NTDX`E#&SL5Z83&g-eQhr^`KyM5om)8^J9jW@<smmkXB
zKAl@-&oG<4nTF9nZc(jWAJGS6<b(B7bntF}oK`IND|aOS7#e-l+nUXVY?{S{^-}YB
ze|HKvBJZaK*AQ8TbQr**Ar;gB(IuS4*QVUq`T6#uwB48A^Mitg--pw@oMT2=9fxde
zzO|eqSk7CU%EzkA^NS&(-cag3fy7PG$%>nUg~7p17%43pofM6F2Mv>~SX|T^v}l$H
zK0m*HIE$lyxW^EtCqpXV=S#GTcitQpJRGY>d~JANsw6Y6uRE)IJ1h#$YJ$!flN47}
z6A<7)mxyG(TEEQs!AW{fS7A?y$qIy((C)j$hJ(|;MjL89!wi~?$yh0R+W>1xK6&ZA
z{VXaNd*E-UX}ia(f;&5j+iN#hSF;b;T}Dw~>8eTQn!H=QX_eY>t=G$~VgQMbvrmEC
z4@Z7@`iI+Zd=6n(*Q0Eqm7puUJ4rwZ;hbR8WsMoRWW@IGhIxae4aHlCpcPx2QV#Oh
zr*>a!r5tKfUECIpX2!zPYeGP@gLGHp@WLRh)CqU#$FbP^)hRMY*Y<@cs^66MN&^ch
zTvN~55`alhe~AH#5#u&Yt;W}c>C?{-ByKQRxl-utZ>&rqvE`%WS;gjvukV|wuxA5t
z3**C^UrGLhWj=>QZE5@?;Cvay^XlmSjC^?E`OV%br^O8ie;F0+ASpv8lk)ZtMd~a=
z7~o_;+a9@MV!9!F?yX+bxpJD*n?~NV6pmdgLOAX9*OZYGH8F<4&!u#~0KbY;Na@|!
zuS4IjUtBacJn1oQ&B=BY3h>wg4^(dKf4JJ(q>{KgXytONHq4N?x;v$sEXXjz<Ss}F
zon?VnN2J<LKHqO^1hD6w6-`)ltC0|_P-8Y&wf*&|=;TV9T2BeiKlJ;x;FE2A*#FS(
zGo4rZop;+c@7p_C-_jZD<4Kn8G?rg*%X14hsC{osydv0m@V&fW?vjQ}v;;G!Sx37`
zM#!T}jiXX}D?>FLpZ@{7j^#e_|DgW*C8I5bM4psGTACGzYfckEX<jkaQdtcP%XT8~
z#RUW(yRr8bF9{vMOq~QzWDp?Rmwn^A*IvP%LMqeuCqfjYUrzTPwojAZcZMWv(YO0u
z<g7jTRJ?N<wOhNLIC(I@$XA%SD=N7AV|L#O3-k=U8mFphj+eMzDll7H4tZ!tll`o0
zogS|<E+qR57B)|v+jB}TzlYHz9Y8Qxk_gl3zLfiiAY5N2X#g2%x$^Y2XdMIT6@L*w
z19f!wD^_p0H}5(==ewx<cR=I$Mw;~#Ru55lXfosRgL9d-+sg07KboZ*7>sS_)$rtz
zt69*^KB{%H<1)HIjY<3RJ;^km>c^Qp1M6Axv~pQefhVLwFU|B%d_rL77{B)3wTBy)
zf)KQg%oW?yGbuIWl|lEm*47t;-<7bibbXuH`m?2FevZgQfb_Me9jF(XZOEw>dm<1T
zy>y>ydMw)QWMbf>wI5QB@%|fK_Rt;VQ{If@&`os${gB#PU|{3OKz<8!f5t_96Z6Zu
z!oArhd#b)714mV+t)&GPX4+@#$@Jsg@$kNDzwwm>X7Bq_0$~YK(8<*0m(@{nRP3oJ
zT|csV$sHIlUWq7asznlpnrl0bYHoh}I&asrbp$H@qT2=32}Hg#+Hi?#O*I@Lc$qYq
zQ`uX~(FPSp`#PAceqZ|`+OzyxsyS3f86c&kxbYxRxn1re0~mnOQl;?AL2R2ln(yl*
zZsobmLOd>S&Vxi&+e5s?>2^<{VID@Vp`Ab4?~lsXt`FW8-0jidx2sD85TBl$CKX7)
zeAs1aX(9JsY(xF;{r(xSJYxZC324Gq>R&HcIc2_5xJcl=oYmD;CC!oro}Ay`Pr~jo
zgz~AR$Nd^iMFD#pPeLcMa5gl!(gDm@vK`*2ekWxh;BZ`Romg~vAQS5$#5%m?<T9J(
z<>S|ox4!Hk+lbVpjww#Zua0t$_<&OC8=^hx!d$oM%7C%qp)nC7++P|&zPkOkbQRez
z$slBaMSc@TuxMGNMES?yUa~HRFhu0-j_M`tsmI|HcG&zssI7`Ax-Ga2TmDB@$mMhf
zW?Hz(VEzzY0DJTj`MO_L5E3AgcQjeH%k}>5Xy#FVA;-ti=l!llAlg@Gcllb+TUQ+b
zkD=yEPL)>tPHtpdD8rtfC*)*k`2E@a$GflAHi{2SgE3^r@#7H)7o-BOBAf{TWRUa{
zY-t^M3oF>myO@Z+T$RR7RoUX?IMXm~+MgA(>kO2*eeR>cY?)23o)I^r%+;1@X@aEm
zJ)(V`QOV@tB^nseO!+1NT{LKqTl7&5d?@vtyL5j^4O*h<uLzHceD<6n&EN}J1{s(U
z_=e12H_%owMzGaZgD3-K6aT#$wYh19yg_Kq!Vo0Uz1x_aYCvt33U820{j4XGIoJ98
zV}o8ag#uJ6uge6vKYQvWFf>{`5lP)Mbx`im+TUXAmR(VScnRASe9R#c{~;>uAJX9D
zx)Q>OPoeu&0dLM@{(P8}GPVLigb#v${LLUc*XlG^LsWdY;}{|;)N)Q>v7-P6zT#en
z8C=l2p4_3zP(m_JaSP5Jl}?eF>{I|!$Y2DZSSM;5iy;zj^tCFfJ!{5%O&S7${>QU0
z(vnGW$j-s4G^TkGNob!m<moCv@|=0n74*hOCe(i+Ge;49U-!^^9M368Pv-U0V7vaM
zu)*JDh8KR^YD10Zxk7G9m)SKO50-MUvGmx208dBc<)NtL`zk{x>w=4Do`^nLA%0U9
zsG(I<_{KQsX|!eVe2;RzCB7~l^kOY__AdU!Bc4aXU!Y;KMyx5im>2K_<o~Sd^lw)W
zkDLypx|R4<-}d;D*Ywny%DhA!WXr}`#JT5WohfryRP7*`@^*-^bURDgxw8kxB3r?F
zU@x>z!JF(BFAarX$dIMzP(*%E;sLM#E!23fe)<l_qK)H88}yXW&jqjdNP0EkjB}Nf
z?fE&^o3{PQ!l;4FZu8K^&{>x`G4alz$KE9IyQ__&L3M%F#mER@!B=L-Kt?4vt*iz%
zn=}&|o7X@$0x`i=lXK|Pv<mZTHF~e?Us6NuF`6nQC0=~$fPA*YN|O8+2&A@x00@cz
zG^$3g3YApG7Rw!l#tesIleM?jg>lGSC>&&gUB46+Jk}w>QVeo$GvFP`_2&yH+<fSX
zV3KHNoPzCtfAa6saeGs<^3XV13J=yk6jp7p$B_6F3=LvZ;_>cq3FhALo&;}mCcs5C
zvfMHjupUmIG{_jS;EW3x?YHI48n-Q5^`)2SuqJL&PimV#Dc4eh$7@BBV0}#B=XpZO
zpaVxt%`_rCNlHM0RG#KPUF%Sb8gmJkA>R!Sd=96kC^mkM0l`xq&CdzhNI{Gc2!+3L
zluVaow?9Cu5gua&`@`heaGEn)gu%fjl;31x_zuSu!BxJ8?SWi6Gswrk`j2UNNpt<j
zTaHtziLSaf!nB`Eo9blsvlS>>U5kcd;hyg??)u|RIi@wwm=r?>zv^aAOcbb;>6BMZ
zQzo}<MV>ujFhnLaJOS{s$Z#t?lhvGi4RN%RVJiNdFf3#a`}e(KMEr{ItG3T7)=;Yq
zX?>*OT<eJ+g(*Yfdicn_xRe}Y)J&JKeCA=&Jf-Zwe}BR9C5VwSC@WJBbs8iRB8aao
z_}ID8vkG?L&Go1=DzB)hb{g|mNP(ZN>gqm`nGi<mY>BO>t$?J_39~&Af`B#ll6o=l
zyrN|+fsQVWy2Klu2g>Y;$ehT?AfXfY3Ko*+1Wl~8dtRboyC_{lEC@qCg$x;~p%#3;
z(PjDwck_Evw|yQg;5tZ4Ay?YM1)aw`z)}<cZ`i3Mvj~;epcjgelO}*VhA}zEE0OU(
zoA6J^AycGaB8@<#LY>bT(|q!`4aEbi9uksN_F(hsRe^>&ocY&hsyw_;*Rfdo4&1-F
z9ecWu`s5q2BjMTt8$gXvL_wU?JWa(&56vgiILuPXhchz%;)rj%zq#M(^^r<biuGYm
zy<-eh!ot}}51tk<ghDWbHnTW#Qimc>?VnFNS6%5Pg8v<-snbaWw-_LU*Ae4^Sfm#`
z@-PmDuo!X~NnFS$Y6S%i9x3$A&r7oDZ*1~K8(w_rjs~nm9E{Aidc_K)WHu}@&OiSf
zYvGPX2+V^2lk78hv524e^A$%WI~1gtGP`Kt;;bPOsAm}>ZHq-AgJmuA04bag>a_dW
zZSc=k-9{PD9W$jQCb9TZ;I!{%RMGbKml3Q^Mlx4TzIBBdCpR1s0AMAcCLCC89qlIr
z812dzO__@>9QtgY7y$H8D{)*MSHmnMRav?DtA^Fpcs2{(uA=$Xb9S3k*1Xk;Nr^T3
zRGTN6_?o)(Vn(9v^KEy3JZ%h#M=D5omMRdT;Zw!Px9K{cBA<SY--Gz8#dO=3-#+~<
zSnc$}{5>+=|93{u2OtkDBq)+q+zZ14-VBe%{M^}G$T`Xym!-_?VQcu}@81yA`N{D2
z*Dl8!X@eQQYVyH#*6W4TiP6*BmWD(ep>@`lt=T$0{@$(>e5ldCK8TjfkbmoyX8te1
z8f85OF&~AnLPGXintR=Bxd0y~KH0?JReK#-BYB{T;z&2Ovo_fHXo6h_+eMkVCF%D?
zFjSg`7?($efsh9T3isQM=rYQ&NHeJo{zq^zG0==XvYLi>P^f@Mp83LBLccPnHslS#
z)|#Cy6K(b^VV|ApkAq%yclACMXc6H87pI3bvkYJ4xILdlQMu3o<C6wp7i)Y^40&pY
z2}kE;J39DmItdm!bo={q!hWUy+kIz;ArV{Ac;56<+Mb_4j>@2wTGc0@jB$FFi<HdW
zScp&wAC1~#wOro;_a|YAMe4N8-*4InFbL}wFxh9&3*;$kREAlmrU47Ed;fuND|734
zMpF*^hFu2EowKXwgfF?p!rN!GT^slE7M*B$p*#OI*<G0ZhX{&rg++M{o-zvU_%X#P
zdxK}Q_DF>koZ(f$g;ae6#K!9b56>IAJ|2cKGVO~Y(nuwMAJ^>o^4qbcSHE#0AYdyI
zj$y62Tiy1+0RQu?gT3uP0ZV+0_y?x{g}!-%jRp7kR}NCJW&+Cs5i&t4v2cWi8tU-k
zWT8Juq!kwIrCeaIOUApKP!QHuE^a^ovDQl9pQWyBuwlZ|Qp%9UI9W!ZoXrOqLS>2@
zz7J4#&wu7u8ZivaUAXeIl8Z1SZ2p+5iVgTo)c2z^+U>)b-BXbu%J$|jKer|5Tm=N8
zMrscN2MLw1SOAQq4ln4#lf#ON9Bf~nPmpR(`(nc&CBNHSeQRoJDm}iLRI_Y48a$JA
zsQ+@}=NF93_Hjbq4@c|QEi7Mt@tsJI@;cwT^JsdSHeAdayScf!=T~o7Ju<S~ZVXPC
zn5@n<UWOg57LkY#AOEuF-CLUZw=W%5spQF5z_OZdv%sgSt83&m*JNZUty+%P+a1u(
z{A&|-GSu9GGAi*}!$Z!vHoYK@HbPOmx4JZWWAIJ4sw>XENV0!oUhP~2GYzgHuPeLf
zSx^74yUEdXT<>6Lz_*)4X}38di93V+=AzUHqlKcPv~^0CO}SU(dFlpZ19dB|Jgf;Q
z8hlc3HtuVm*I%Ic;#cb%z0s&7`upo3y$;w_&tXP>=UdhxGPLZD1nT7PRFSd#8zGSD
z8x6q1*(cWUPRZf~%9Np8AsctFXr1$91`boqgt617*=w<9VXRe1#UKT_gj;-vf_mTM
zsId$^SB9Go`G!jmta+|lto0~ee^1h$H~j3Uyr&;N6YpftF81YCI*JRxK?Vw%e-N;H
zPvk7*SF?EUD<!C^((!r<gNhs)`W-%!x!TNhHQcM-ai8Z_Bal<bL6kz5IrR;Kg`BBf
zEpNP7h5b9jXHXzu862$XxErbDY8TC!$ZnY!cU_v9U1Tcs2z{xE&ePJ9Rv1%C;Sul{
zvHOAwbRy#3GOR!c>WtVQ<_XRM!(g)_z41y=wRl`B(5k<a%Ti+jYQ6*H6&^Z3UBY)b
z;|LiGl<?9T$n2q+yX}h6q!zl3i~$kw?plj8VyRjqW))JY4Qr=)J@p)za3%uriFr!E
z+JxPKB&$yJE<omAg}Mj9N`wgEpMjOgv0hTJ24IM10C7>RG(v5q<U;ryMlMyT0E%a7
zoiH=0nXIAO)z`PvpG%tLT3V&D(IsRKjWv{3Fh%E`GRgs@DS(lwk~mNJ8B*Olmg<QR
z+dpb;c-+Tbn^#5s&aP(f^eif+mmu$a;6`b)i>#`##|*AYn&m$i4Xji~(m{&)Ha5tQ
zZ2)<d9yw!;c)4(XB^)al&{hFj1}1>qe3LLBg*6uM9{+@BPzWq(t}M#}l8INw>s|-x
zCEoE<4Q}jMrz0et#B#<d<7AcaNiK^$8M&7Ab6gD5Y2PXE1A@-S=S)>5ZpRy&gKsYu
z7BH(HG*%LHNlT6SlVv>3*HgvSPEO<I_R&5cyn>efn@n6PHmG3^b<2(6!znWMH}&t$
znAb?e=4;muv*S)t;N^h1w{vA%4YU8eG%5OB>#~yYAX|&BLbNVTOBFO252cGM``Jb$
z9Psn}TnqZlr2Gpz`x2jvCPlf5Asa*MB42YuRck(lXuwq;Wq6}+?2!G6^7@ec>0TE_
zc=z#n!R>z9=nCxMXm7hVZ^A7yd1r9Yn!pA58ZWzw>{rRjO3Q*cE#b(ahn<Vz->vMI
z!wgYyS|Oad-vo=)%S-4sQ?w+=r}bfKGg^XHBItC+*sbGY+3zZH^)5-`?(bixPSGrt
zTOVJ)X?$}t4cFQH>y^%Kb>icd2eAetY(SY<1OD@REk#&FJsBy}F94KTYT|pBAiqf3
z(+%^ZOeZJ9bj6DVQPPiqvU)rs;s@^Mg(bb4JJY9eBe3DI>uW=Cms59i;j^@Z6`Fsi
z*2drOHL0YD-3D?M<q<1XEgkBnYYi*?+x?HHdFZ~(SV!|JDToe*v?Ma81KpAc9<7c4
z*r0{8d<I*&r4KsZh^Eia&2770L}7esCvfpsg-a-)FNGJJHHHcijj=(5GJR8TFhXTL
zu54HG<=C}pavzUOLSJ#c-s;mqR;8z<eOSZ4y@UA*)($F9Np^RfA4nvu(K~&BRHu3}
zE#Yldz$K&V4?*lfl>o->Wi$&w4lHYs2`pryf+c9hfXKeVrN?Zf9b^><zE;E*ep6yp
zuHPO)2zZ*-j~M^V9g6P?ObZz%ri=v}GGNuteujHcw(}Y}Bosf_*}bnu>9^iYrJcrM
zAhP}Y9?d3hP2ST!tMXD(5Ea6Bg$O0~aU;o4Lu#p_ilJCZpLs2aB_ASEODPoR)9dK}
zX}6B2X-V`da+@QIiiUc+Sdg(4^3UHG#c1QTd{AGDIjX*A!-EK76`foNit2GmjVT!A
z$T;JA-VX8ri>73SsTEV#0~Hjtr27>JViXvaE#*o&sN1!i{sxi&{Atv3<GZPgS>rvW
zoJKq^Rrv?8i+0{n4VL8un@4r{>hf#AZi@xS>esm1{b1h>03i0gcKs1zyv@_PQS|*^
zJg&|=8c=CO(Uh~mDHYZ3>9zIu#TD^pZx4^A6VtFTi}jW~k`7vgB`yJ>hHsG9uS?g4
z_e9aXf7_UW$k;+MNh--b(q3C$I{)3BI`ZAcMknO=qVA#FY`7+ty>dmY1C=!F;Pzs*
zJ+M;s6=ZWbR5*-G8(Q##3f&sf*s@wbEhAc5>JVSD;wqSx80__jN6x(T+>$4=U?2GT
zC@TEKTBs^2Jn*u!ZKWHz`S$OkX^>IE5tOcPo#RVC&)?dgr8RFY{@xIk(ai&%&<7S(
zuma~<#lFeCUP8&7e`d`$5nHH&)w~b!Z(QZiKqZrzxUD07tLeo|?+-iAKixL5#jcwF
z>)P89_aAtak=bwQ=xA&4MqM4aM=MNt+XxTs1|Nla7UZm*yqISy!sEHI)3u#wuDZs-
z(Kd+u`4$gOCT$bP@nq4pNiaSmXu&349M+%1RfZs=)MAPu<)On3a}@djGvx~8|0{kA
zed_M^>*2556#+M0l3<Ria5-;*mJDL{1~3wc-ewpb$^JTq*I+}MbsE<ad|U6DeYpxl
z7bS@YT()0&q~%03G!*;uX5m|c7~gSLqqS8Cz)x;U$9GE_c2@7H9$MF~^ICn6Ow3Fk
z=pS$C{ppjEl8olY%Cu!*ScoCcB(q|#R0E1<G{=nb1Mhalz7a;h+ogSF?9|ZUvg8P{
zB24+Ebk6h11|QG*e7%tw0(BbXWxnQ^KC{!q0azraei9pFRR|@Bi8KV`2=mysGJw0<
z)!14Wtkr#SG!;sqp*%Omin}-YF&BGLdJ%9yB)=THj6!bx%*Z~h3u^TecstHn7#0rw
zPnmNxB2daWS@Zwb0qlG=X{0Dvz+DHr!wPvMyZWPpRQe?Oxt*Ap;iMWhT{_$!2rQ6M
za46wR0ZLL%4N%BtEiA->n6}qm$Qp*SK{WelO3F3jftff}P^}M3jj$k}5cEV|(Vn&W
z*yjo@LRj!L&39UCAhnAE7>xXG^vVH%L+;s?IMM2BTUx9DC1Z0*kv5ky*IA_2#zDP<
zUzdG(opSU^OBj`0Yqv`^KV<6(BV@F-Td2U0x~W>h?{sxd=b({-e-4(9wl88yTv-AQ
z7cIn;BV15bE{I<cQh;rp;ZkGHcvRNlre^97v6b1h>-mQ9&m1kTmzdh8mLn-ijb1DO
zW3iU^wrsTjd@5+M!F25s+T+LPDaJB-UOMev)Ih}8mnr^mC6y5y4EzsfXJ;45FMgr)
zmGsDW8#8%2#h|Yjti_bq-dfdzN+M)={@!-%Wm7K%PNukM)(4webOY_|GUVmu=jKKi
zs{V=V*gQ8aa5B`S=sh_*YZkQPbGuufd>*da`JZXEW9B%sW}!gjtXyHJ6c(parN&3S
zE-Cg{#X>Zazw+Esr=n7CJe8vEQlmUc;sI_Iw3(o<qbZR0c-$+mPH*PZUL|-uxE6%U
z==gW1AM)@7+Zd>4DFNfa283R^H=7t5Z@ryZti?Rcao|@F8mAOc=3Z7_!fdQLEF&@m
zC_%PLucQ0P__qCRSVLCdaP-weTgGe%kddGy!^|;zM_wZ2G6M?+d<nySO1c*WQVV`W
zqA_HdI~2w?S#Xi+y&neY^LQ<BYqkb^3}<j}H5P1j`NN$zQ98Xh<?IvS5hzOc#{CZ;
zEeWP(q@W5XHX?sQQBsb+y$e?ow|C+DRn8W$OW?FrT}LT}Su#>uy*=?Wne)yca7~{v
z`~35r>wLbq%rI|UJB;Jmx~ISk!T6)L=5gwFy_@89uWzr$#!nsshrnvFJaY75rC@0l
zp3;b}R0Hi=C;<99^s@rH4q`enf`au)k2FB|rK38Tbb_=}U=(Wnf@|&J#216ZqHnu>
z&y^`@)BhtGD;?Z-%kT#aK<dIMb7T88?&@on)huWzAr8-$6ei58A)^x3wJg*$@9Yz-
zPh1JMn*QCo;=@Pt$=$xqlaYtjDh@NB{9D*SlOQLf<dkMiiEj5~q_<c!a3Zh{?Ssmu
zPh9`RxOqnM+gP;w^59}$d}h%!vwl^XC5)3bE@zm$a_P#=x^p^gTi0$uk5N%`tkpz7
zaju9!!SK^BFdJ+cI8TF%F@+Cr*4D4A8O%lYxb6K-hS}|LFNs;mwGHI_>g1iGiJPz7
zUeA8*m@uIYsN2`nbZ!7KOMJ8zh!L<<N>8`{#F7|-;hALW%xSf(W6pY%HnEycfvBr`
zm!-oCMcrqn)A56^oRB=aDpJr-4CkEB2Kt5={VDTcV&DETOtRW#P;s@Nl~5Ng%_J3w
z2W-#iGtX#z2=s1%^XNX|q(p6e_>ABr=qJQ{_(esVjfX9>2?J?C-F?&_&+-B<TQxdP
zPCK1nUx!Q|IMVxUA6zdud$hMO*(7|I9JwU=P*50;t7}|Y$k#({#nm-cwskAQDfAsB
z5dK9CHr4;r7`s;qrL`0ma-H!WwsGC+7L3k26aaM<1V5&&UUvRdp41)6hGLDHJwIA3
zHRa5ZKiwo*gSiQn13p)=ImN{{MuK}RqDJqs7|5gnlmHEotmf6KP1^)ZMh5L-cDpCQ
zVx;a2%Nrw*3}ois;2eJq-VDKab^$S3jqkzCgRSa^yFdVHD@n!Gzke80*kCvIe2z<$
z#5v(XhC7j;8wE<4pbkzZ1c0BuE*6B)ZS9DABVhX_?R1ek-Tk_}M2q+B+0}(2{oTRU
z75e^ujZU5A@r>kwp-&-VA`&buOxxA{TRdurXN99uD>kr45CZ3G4d8qEIKB254u<vA
z1Rv@bwEAL5v4f9SZ`U3!I<Iy*O{zI7rSRdE<L3c&S6ldoPT}g^w|#s1ofjxy*EM7X
zRw0;_m*F24A9)D{h?QyXLW>(bKn?XhD@XiU4hum0wyrn}cut%!3arQJUQcRV<hOgs
zJkUsRT*td&zbTiOnWgat6$&3H!Q6VVFI9u-4w8@S`V=-Y+u+R7HX2YMup~kO;PI`B
z0SJ|Zhnd1%Rv1!Wvz44}JDyLoU`BaUOKzQyX@hnds-Vx#J|v&oQnc6t6}Ua&j1eHb
z(giz_!W^`%#QQCDL9MSYa9A=Ve!1nr!#&vDyCLwQLkg6dnaq!G0X%_0nJpNVI2ykh
zef$tB&Wk5G$N@|x3#5_(z_qN=;hx0OWAr1@WgwXyu$2NW)-%#nGNuPer2*h&n1M3N
z8%g0uf3z&Ta_9r#rXgy|xxY`_Jqp5GRk(G{>@-}*11v;y$w3KLyn~fKBaD!GO|B~<
zVq)Xp3vz!=1+DTQR+`kbS;x~P$?KjNX;Q4+KKKfvPz%nK;dDuZRJ*&?D2%Sqwu0d=
zNg8Czr+t4-)br3G-PckrzsX4I9W3!Okxn8+v|t=lq(N3L?-#Dqii)m3Qv<u~`HVQK
zEi-~6Fd0@};{FFm^hER2)m2dv-FzW65zFcIFiR4>pop6D-gCmNZu_c)=fs21e^Ynq
zW66d_a{*NE;B#|vrQS0J5;s@aMO&D=g}vqL%rlLzEr#m0x^uEtDw^5nH-$en(0X!#
zPT$>pdpCY?ufJP9+LzM3>3DZxgHK@7&{iLXWVC|eKxwc_)@L<Od^=AXn%1s|HwWjC
z;pp~Y-OClX`++$oz8<Kr)?_VLM00gcM#tHIPA*5VS7jW+EG4eimDg)!2;QXEc*1P5
z6V0^Q6`z>@b6H!e9a6Yo)g#n|)(-HtIx9=cJP+%HZVPf`=X^U@UfI>vP*ME#p6r>`
zSmiqG1?!%iNb7=g?Kf8SbVG7;q-u}L9M5V0%yuAsTj%>4)1kOd03cgu&b!As&Qgea
ztaQ4i8dbd{9&lCDd4IT3@ObUVdp*MS@MkXMa4uw{((FF2^C}K@A`h!KsgCDKB{Pcq
zvBokgKB!3@L5gKS9YHQHU?iAhL>=;o!BWC7F*PywI9>CY@wfncIA6OWGR^wmHh;nF
z(zu_F%mv`H<bPCJe{ibmAEqU4eQ8`QV%P2l+!9B=cz&s<5I?gq^r==~eN`lrf++ua
zEiO1<J<_QanDvY-0F9{zTdUK;2Y0Ee&i{SwUhDXN?`(tNdWSrCo=Dc#&pNhqoe@X#
zc0Ovn@7f9)8GF}vpHUnId<EOuU~y|d-F&>fbrWlDI||~(c)h^BZlSW05xjvwW>N+w
zHNkY5<M`DyfBIy$Xuz{RmBf!Yh5z0rVnpiss)?E!PW{o%VQPeQQ8*^O#(qiq_#~3G
z*Z{7EA9gv>EE>+F_QQAQIEEY&Yp!w-44;Vkb~!P~6dJ7vjG%CQ2g}MQG1IxOBJB>O
zOR6m018)qcjqc{Rt~9=6!PG6Nb7s#**S34!^z7Tx9ABPCul<%;qh}?&%x<d}43n1M
zhsZv4n}0w*1kZVUw?FOXA0MD*I|R$TfB2OlfI)3yWS!w@)Oo;oE?KX=P)my_eY#hW
z7W2o#`e3ikVZXZfFm_p&r6Zr#LOvyJ)439Xq7wzDU=NQ}t(&cr-uPFBPSWtQvdz)a
z)9av)r9%u9#J)9j(He^mQmt3{)>lox@)KumCGtII#sC7&OQRayYcVHmHTJ^o?#`#3
zxu=%4m<eQk5Xuk<jw_`2i1=mB?e><$USR2{okTq7@R?XXY+|Ha$UVuLj|i;0bZAr%
zYq`@t*Z9iIw27bjy=br&rIx{FuzG}zMRvN$lF%9(SyA`xd=tOCnVvmGY9T-|G-aOp
zF4D};Mj$8T5}70&dMQz^d_OS<yD#bV5zdLo^Bu5<Kz<98V+p&hcT<yFN=8BoiMNz3
zn7XKg1P(M|G~{&xKS-jKHK8)nR$h(RTE$>%jc1&GIM(3G`lk~{<W|<})M3ArUOcn<
z%^cUJ!H8e?tNcvWkHa1b=%X%aaIe-^!cx$j??6X~^Rf>wRkM4MW(A@rsOI$JP%L(K
zl`|F9s38-IbVPceQO?A9Yv~E2q$PPzV>MSzO{6UB#dr_;?r$_xW+v=7L*g#5^M0xO
z@^M$<aV<Z;;9X)K5Yfj`2%`8XjS~tk;(qFB3&VyB5a6hM4MZ>ZaMe&Hh1{&OpxYk<
zJ8xS%uTDF!Fg6O;8zgH=j5I}xyA%k-_d}d2j+o-&O?ntA2@?<7f@XnM?e8fg{|jX@
z_rf?KDDg-dnVN?EDt6|l4KpXK;$v5f6_%1g2JBXJn>Q`lXl_$+5Zu1M(Aai!uEx4>
z#W{we_Kq{Mb`vos8nRwiBR<*?;u1O%k=ikmF;j8URLw;rZ8n`0d;=#U9BE-?S_2_>
z+Y^@bV1v)N+%he!4UJOkC)075t>ip(2kibron9hF+M3|E0D3FyzdDTlv*yo%<i)oO
zSY355LavsBQq1U&Ug0wyh7*tjSt)SU9859)U43;<F1ezOZ)}KSPMiY~y@A0lX~N%l
zB*@PnjyWS3fS^9%d))7rj3pp{Ev#6@7I2tv2ikNEyN+M(dAJ;^mIL%gFUw+=3x=}-
z59P^9C1hba?YF!?+uOz<%$GtE0Dj+|jRmK}RhWCv19J2gn^SA8*YV$(0~%+9etdRz
zzny3wcQtm?8}kV9aX{jCb7jO`-<3P;Z!Vbdp?y@nMfPq(_@1{P|MKiWCzWDX{&!w9
z<vCptatEM3T8ZOn3A}WZ5bl&w<Ag>?In3qQ%w75(t?cEk9c7ogUQr|_?({nTyvS>q
zH3&pM+}GB2&p@>l?v~>+oZ~P9pZvfLf8aVbxVdI&*y~R?HU$Hrc)g3Z4mM-(<%8#R
z(FgyAt0UvJ+kY8m_vgEhvzvA7UEM(;OVJ_i9YW2)Mnp=y^DXDr;vr{K;nxQG*zYi6
zmI-Huk(1SHca>ZMBx#uRMaT=S%Gb=es^+W?2+8<(9x0pir<ZO|`_LhmurEo+n*}CB
z%opr6D~(EB^1uW*DI`=Wa48ufWSHkBCM}~Nt)NsH^p=O^<AJ>#nRb{n5f?yG?+}H>
z0@R4riYEkX>nM?1z_IGnnLNJPF(lx~fg5;uWaTs?V)~yHm>1wIUFJ0iynP_mRB|_S
z)c`Q!4l-aGZXBRhg6ArXMNU>fuD}0$CZN69h;y;=oTo4Oro(pvTY=eR2MZCOiVaTL
z7-T5)B;=E-b3ueMxBuw4IfXsi&1GclQmJP#auvP<wXUc^b#--PNj>B3s{btrlQBGH
zB>lv2F#b}?M-}4_c~hpVc^Ce8L*I5YlCc)}XzGL%$ZYr+0V04&`W1{RkugDwR<GAr
zW;}S<*dq&>o}03oh61phLTB-G(-r?xb=)ZD$|P%@(Z)1r7E670-N&dGv)=7?(=hk5
zozlM&*sEIw^K0whJi@2YMHt72jfc@cs~^E|2E4FhgbAAW4~I%u6$mL%GQ<2I1tU_F
zB^@rQ$h2fPfPiUjg%cGsP1Gus#HObGHib2;RyFe5Yg1xt$Z6=x8h+B%M95U2Z%Z9~
zT1EN*q=;X|4e&mM$C;0KDmBxi@0(vB#vPC$B?DHsSc)OL<YCq;OX2+Zq$FIMj}04*
zGqPWll<5l9#Uoc+sig(hEuu0r|BD18el6PnGNMXH7`;L}PF9V0*f;pLGqixEm2!pp
zizCI9N5nbs&WEXlJNQYrzGl2l1kzxUX6PmUZ$No=JbNPQ?J9l==z%sjz942(EymCl
zN%m%uR_NaI^Zlch-N<G>0?ELL(NSfz7$6XhUTyXa3Bk0taTLTO5IKoFx-tX?E4LIN
z-Wh!2^`BN0LgPQkSupLS?dZ((s(;^<PZJ-x2?ymOHK|6d>v$4a*FWI);J}iXZVm)f
zZC8dS+MS-;YU=bhy5QN6hSkYs*zDa*q|1swjMg$p<Mu;bmj3H5+AJDIsH2xV-gEk>
zii*>`UE|c>)pY=Mc_SMqZfy@uEk1paC{VNf0RNji`kQpF3F?k{GfVdBotH$OCMhzK
z(%?w%h&A9!S^V0Y(eL=}yO-xn7f1qIyMIWV-W(q8BaH9y((osTNXLIi895Ouq=pv(
zAb#{(DsWcJ=1Nv1UrJ$dDu~@i3ohqGe_`<wdbFJpWB4L1!t_iAVQDnC)Z)4ykbxl#
ziC?u;w_kQgSEJfIG4*zW>sm)JCQ+oNkwbW8XBzVwm`fvi1}Gq@mLm)OIs=Ocr+1*G
zm7~4sGOD?_map1M;y5KbyQbQ6d7e|Z?d#k;r03Nvke%O`KqV|mkkn>JICf|?dUWc8
zNI_G^E>vcNS|A5=Zz&PSXt3;7ywo@zo}r(Nsi^OE1pn9KKag}@swXD7rleQ9H9&KH
zq5lo@IdmNak!8J88QI5Er+b+sS0fIifA_8D;H~6qDE=2)L$miquX9#@5f2hl<oSVB
zjW9^NIUmM|6TjT?<VRTgmED4u$7x1Euao%GWvio!I!j3drWdtfvQ){$V92Mq#4RK)
z8g}cV)^*J%(HHyS4~3^#Kf-kCBHn*t)Qd5SLdRN;i3oxuS!!8m2nJS+e8kvR&u=aS
ziR5FXL8QubedWyl%}tAIc6>n}_2mT-Ly+I=yD@)eVU=h0oleZdns7J>fec}C-)}Q>
z6^uJ%KIZNVHCpO67kc@EeyQc0v(dB+2FAmL8zv2`-d3XpmrT{hDWs7)#NJ?A8BGx5
zj}Fu=wc=haTcwmk!axggaDn;}$Jq_UtqK>=7ku^sgaxIPooaN+Z?lXa92`WQ-{1;2
z%pUnHrKJ^(^(jDYLf>wl55PTLd7mFPF70_qgG#wg=T^kUo+rrB-=BrwUc%G_mQ34G
z6<X&m+UWhaO=_Hz`Fi0MGCDT8)NkmLsN!rasnvH)*YYI%a}7ATxOj<CivirM+Dg4A
zOw6m!>3SPpqKI|eNZg;}u7O@&d^CEa{j$b$9lkcgL1%xbJSXX&TeaC)#ksV<Ds<FP
z;5J2*jqdHeq%zV*(0pcP!-4VVjOSh|L}UZ;O8S5@oo}0Ofj}p+qOTlJ(Rm~mLgiY2
zDk^Yw7tUSz*tE%9Di3Y-RR+`xL}^@hirIQjF}?QWPVczwjF&2w9IDVkzp{V(MUB|w
zRd?^^^-u3$x%!+^e8zaWqV0c+&91!J7p8+_@Ypm80qqniY9dGx5QYUIO~=_7pEFbZ
zxca!Rd3Nmfu+`rAXm;;tlR1UON-7-+ZeyNqJTw<l=`MMfELf&PDG&i*Fw!Jo^!T-a
z5O<e@kpozmb^NpSj2N{PczE&F%uG!4@(K#fOhiNOVo-UfiCNkzi`lvV^)SdeB|`g)
zDPlskVk6bn)iEgwCP<rz>0{{#qm;f<hkJIQ7OMu%Rs+$PS-6%;@LDwHU1c4xk{Y^c
z#r$<bm`WPF6Rhw|Iv$Hudr{s)c}&qdkUZ>OPlG+fI%M}f`Jam$UO8MfO)`c^>c8gM
z>DPJwfwWe?qZK8>*JWq`IGGweD}G}i89o<Mq|NT{T`$1sAL=CT<U3E3I?j=>Nc-%@
zCjip^TuBe3FJzKT_9jkJP-N2*(J6A%T?!fsW>eCn$fA~RwKD!&SErs2J0FGXMWm@u
z$xvv-#?N5BzpE>uX`&7WuhXo5!S;Iv_sj9tup{kbZKtgJ@Q;QD9(13;Pi^LzcMHh?
zN{zCZBlsb2iKt|au*)IRXUqzpL6J<1(V5Z$e3M<mp_2c)!&z*wSW7ho@2d|*(6Yv+
z6h*_<eEG{A-Xb|FxVp0Cp3ahn>8L{=%!s{dsik`R`Qd8aO)a(oNf4-hh$|PQ^FF{P
z?5kF&7KYWAL`PGZw0qj~_rzLyfMTYLBA3kf9v0AUK(+YOSzJOyiO^-OMz69CGhC<+
z2E~U08d+_pp`MgFhyK;E_TmLZrD>a+*B`hIOwDGNUcVIO*ak}+*F6nZYosN_4PML3
zU2Qh1s;H0-<>0j2+4x{EcRoBE<5^wZH1HQJ3ZSK@&&x9{?qiGy;(2bh*pl9tjgY2E
z#KrABS>M{qaQ_5kfE}h4vL+@bHn!So5mi@D59hTe<>x*lNZOsQ<%l^xLZ(USs+QKR
zRJ%4hDR%pwp9C>OqrgwKUFJq-&Ob5o64=!(ST3*`H<}8vv61e*W38;H5j3rX5<0$J
zbk1({G-5A$tsv8fLiK+Ta(v<Y_^?t&Re-S3obwjgtVXF?z+C2L<SHxJGO-~}sspUM
zi?tm1PyVz2m0Z}+;?@YCZ%B}Yg*?=w^tpoet_#1V4G#tGF4eg<WvQQSFT6@{Xj)Q8
zZK<xCnsF8&S4<K&Ok#PrQx3w6{w-D9>r8ZUk^vObN<*2+SpS6uZZ>ea2;`Je<;<w9
z2HbAC-5m-tNeWrWjg-BGSS|^am9gxq7g$@oCH)B8D1L@zi4H*WGjC8>fg+I}U&dIY
zc_N>#QrKvtt7~c`FvUT4QMCTtT*hQNP#r5igFXYu0xNobRb`$DdUm?%7W<|)X^26o
zlDY`<%`aS^TVAV&TwXpLlRRpCh2YLLz9!@SpcEP!|83<^gH?%^LKc+xpFfqn9;5VD
z1gO~J^XTTab<VHr{)H7PGtTIA@_r_9oF`=AKq%%JYdNV#Wp>(sd945QsQY)PM(anu
z|ADb74y?OfUpnf5^n%#_cA9}oW#nsY^RG8$nS~^oVZO;gHb5j(ESy_PDaEpIC`^eF
z|1)<X<{n>&wx3BqV-BO}x5u3-14PBd__rAb>J2z13-Z~Stg5C&tdwLm<1tE>28K?v
z?U9vnalF8|bNr97z%NyM?un*`CY@3y_S2_iFp#D7*VdzeX3L#gRsq%a<!X6#Iw~7O
zv=1hG-k&6`1|O|@*0kSb^V)tl_^moUSWe*NAIiwIQH>V<As%(;)4t-oA3zmULy;iA
zNp%u=J*>%3_i{yqQgL6>ctn3QLHQK6{$=Bv{Yyeze_ylo?ep^nRLiil0A13%IgCp`
zOi{4NeW7tY0d<Mdq3-(oBR#7f_H)bL5y%y}PxCeYRI<{ffC%NOVVQsbv3%HAw?|Pz
zL&Rovmb4aJcC<8ZYfL?Jt(MV`Xs&jGY_|*A_@ic#u*a1qpQUpzgVk0~URISg0qE7R
zy-<M~c->w=N=;XnkUhslkd9cBYHkiT=b}v&X~<cPDQun<9lo0LHch4@j^@P0VkL+h
z>e6(0ORX-+stE3xkFGvS+>vK29=5JsJ;3ZUud)JTn6#qnmt1DN3LDszT1`wR%hF~X
zpJ9NYQ}#HyT3*m6q@T!c`Z6-8<Oj<}$9+Wev7$!f@XCQh4Ddx{wcVzrXg<uO;I({_
zqj(N7U7wK`#9C514TW%Kuxr(tn67nr|7m+CgG4siXd)C82$CaZ6^iRA$X}Vi=(lHx
z@BGJJ!_{$4<r3<<)*7(6{kBYA=9^%^7}0N!X4ZJ0iKOV>+{m;yh6V<HW?y8O`V7)b
zw6e*|D2DAzJ|%VcA72=wJ8of{^q6GkQm5SfIXqd*-<az1x3QKIcIb!|9@~7;JOJST
zCOSe~Mw*O*X<J%B75S-@_TRxR*wQlU9S&h!aU&KO>z7hE{4_XyKKuZ?*3H>C|6K6F
ztH!U0VJVOg0-~_F{-azIay;=C2j5B>N0nxMnvDeS+%H5rl3E(h=m1L_CxGI){~u3h
z8P??chyBr`WFXQ=jg$~UKtReNozg9ebV!f7X+cUFNhKu)(p}P0KR}udkaoc6j%WYp
zIG$&(_hLJ)d%Ldd{GR9MT;D2DCl6I0|Gx6-ErKFQEbDf|RQ4vW^O*7efaR|7{=Bin
z>+Is=i4OIXf~fzlSIqI{5vfgsOv96(!F+vik5aGi<RT?J#XphO#s+IhZ__wiCQf3y
zL0s;$u}6yaM*RDzc68uPsLkDRMb@~yxa_M>&F3A7{7nGW80X7Y|BbNVvjBONtI{PT
z73zPkrU>8RAf}4u;eu#0)!y10*yALD=yW55f7KwJ$HxRKD=UqxB$&vFddz3fJL02X
z9Cgp)iEucIGU)54-v6kX!PKg)-TJqmTcQA;u6Yweu*%*W5cK2kAA0ZibC#czQ`%39
z?Fkm?82}q<LLEj%jISg*C0NKKTm4#1s$_x=teQ=lch{U_YRV$R<ahjWN?(WJtAJDV
zofq-9GCKm@gSZA4ksrizlri&C)H-}qZ%%993@)Gi2eR9fnf`9iU`kddf4YgT5=t9B
z^)q*E%=|Mkk+?wQ^G(-5I=5m5ed)x->b&?si32D)bupw+;t;4oUrntdTF>{*EG(QF
z${}lR5^!Bv`%JFzj1t!eK}L^~ht@B#dg0POq1m*TEv&)WW7kZTH|=#h{QnK0!+*pX
zn^_R1o%&(jC65>aFcO}MZy^)sWB!1b7kyZ#$8{m9@<Z}N>YoPAnIB~GEG^9Yfo*If
zylif4<zs2~9E&D@OElIMk{Y2w$675wmXtZ3AeObam)we;-Z(zruShSC8LnL`c;gsb
zvv4hutr@A`Y;2woxwcRr(|3woEXg|Ur(1Gq<bBOG2@70|f6cK8R%qa9>2O(uyxH}i
zZ5@k{Q<b^82&lJ6>q%D98c(MMw3dB0E~U=*D-E8k*e}m4{8+(D9HTEBYli&ATKp19
zyv+TV(-nuvnt4dMHGh739zywu4N^1BkPI#wD!kaRygcf>KF$OGvJAi=@BcEo>b?HH
zEAsp+R^yCw(@E)8!r)O>@0w;gsp7(ZV=+(92IjCSw;<C-=fxZBAn6L&2_wou6VzTs
zymC<z5#l5!-5&q<<NRgSpa^2aVkVuojgp;sgK?Z_`IMP`vzJGkPbEH1O_f(c{Z}}C
zHeL8vPR_S)S6_++gT{S@k<`me6U(5P3?zwsMzX;bb{F%o40@CPYI&EM_ZPHoHsr6o
zVfKT!&qv5qi&8d~YCC*@e{NfQmdvUa8H#Ta)X*0)Uqh5b*R%aE4puvFlYwi~&Qnv_
zd)badZs3Uf{>n12)>|W~SB1<TQviGLT}XSNvKjDu#`u8kO!@Vf0Cg37Dvocg5b(!<
zqr<QXHxsfF$~h^eh{X50vDTc|VET8s0P0|$^d-cle?V(CndljydWZ9O^qR-t<DDEe
zwLB}k9AR_fo4ptw(t%yg{&^X;Eqq)O9^KD!iA{7*9fjP#rx%`;t|$~f=P6K5{>+P*
z%_<)*FSZ}f=W%+QvcJ&aVhXBh0I|PVpWQ0#hlEZ_W-Y9-1@gbSd+XX8=mED*TmLXS
zn4MH~_G!jK{)=obKnMFTqNyBRqV-tZ`6bhXgoJoZcD9}(PBu)U|3A^u`OC{S?u4vj
z0itO{@|3$o##k{gxp7kUjFA)*x%u**ch;#xy>V4u?n`sQDi)c5r+MndHv0B)?EQ8H
zablaacc1-lECYYsT)848vEns1cjG^#w-`Oq7H84CXFu;FK=JTi=PXyF#%VTn4mD83
zYiVx2R|(wVu;M_%?c_N6zfeB>4vtq=p}Y}$cCY$9fh2GOX7^zm>OMH}ADR5GDEx!R
zxy|Ruj>bTr%L>tfT?9o*nd>r9l5g_Fgbq?aH8mCPo)<V{;g7IW>^yXTNLBOzzpuq4
zJ?^x!3|-~|EFi~fj8qWeT|=H7d^5%<!5MUm!@rh)o>FM{m&W|gQ-#rXJ_|ZxHOuIQ
zg)@I<-k&Z@c0R5$IjZ3;Wzb*yWB)y$ZYLr_U6;Ag#9VfxEXt7rpV>90%J|xAKPeGu
zn4c`p-}Zb=`r0+3jGoLj*UW5DB|EJdq4W9(-(lJV;$MqTT{`Iywo-qWURq35z{}D%
z{wK<7aBNk6YQ0gh$Q+jzGjAsfI`H>@$F_a6>nX8Ku$-|3`S^}Lf|cagHgwc_%qXQ?
z?rk$q62GfM-e#a<59%(=rW%D@<a_`<xC{GKNuq3_{U%k-dBjOfEve8YQ*;>jBN*jx
zNR8eBJPCuRc6PV7|1-7}A0sMb$ScVi&kc0Q+Gr0;#0^oXhR}r>in8}-%sQZ7p~RBQ
zhaE=yGKn}uwz)_+g)?%_Q>j%yiwb+`^Oj0nc`0}6Bs>=TXP28l&B6M)rZVWHfFkG~
z8$I_2?$4(-@_+p9f;3y}RFX89r%;9?2|s}Ww~iy>pmyIz;bz6A+T;woSE7a=ic)m=
zB&`SIaHzmA5;k8Y_1=&SP)HOmBx+A%F!9mxzlJ#3w$W`4UZ~6W>gg80BUF0Hagi&-
zdm`mv-Idp-(jA;)p#G)=P=AL1`Mv;++6T=kPl=_81)ieV8J_TZnvB#GO#5woTG=eH
zH$rnPJsfRUZxnc`W~r;dyob5NvK$;!o!ic*(9sjb55>I-AoB}E*Z!`|dunGU-zY-B
zI0X0!y9L?-`O~`}2zylrT{;=<-h^nY4utEgY?#rJ5$eC`JRe)U4}krgxjX1#StvFZ
zpqZVWRj_hMk{-^OzqUe{KT-r!YkdP!biqR-?ynMIgkzJQu5IVFmbZKand&ZMT3YR{
zJQ7w^xGyA&K+t|qlS{i2*=;W>flT+8yA2{7Lt>CN(3e4!ft<zDx2#e9xxHilTX>X?
z?+Qg@e?jNpO9DaeS4X={g_A0{3V*ytM|7iF;@eUoF1PC$(Y}sWf8l6_hcQ53;_K$p
z-hM^y&lje|Y&yy?90+G%3brB^JR7UzpYZsl*n+wXSGu7gOmzF4MK;JzTku0|`hZYN
z&i=Pq%&C95d%F=oS280@K&B1SqG}mKndtMFE^2V*)S=o4L9@}q6Jsrz^GDw%CCK+a
zmu3(}{Jz>)B0{|L5-mA&cL@xa6^0u;>1=B10PeMUQ;X?E&i&el-v1iGcC!23DaK*=
zhg~*n-P^}Z5j=Vd>!y4B!?p9CT&WeqrI2lAV{}&k2jfVxiOIX6WbW2$BHFDIjj8ai
zw42qt?)9r@8V_OB;wUeyZE0d?659dh-HP1tgdFDhpuQpu5C-S1SjjN=^;ak{fu+8F
z64c9lSxR!m6mC3%k@YpwM^_*}c54>TqU%ZjJbj*YJTc~`FNTno|NgIx)v_uE&K+`#
zL`siiXLA%NlbCJ_KRf~fLw#ID^I15Swp=epIs5PYf0otmc7G+~w4xqOoj<Cj*6lwW
z4c|NrRjWS|<fj2hNq(9Mrm8#0rt1pQOnsujjHj3LYqa+F{>{J&8Q8{DHujl%#Q%~@
zMS7XV3c>Xal>$%4G_|5b3x3nBg<?Egjni`4B{U|N#Ejp)%7sQ{S@ddfQ03!`zQ)h*
zRjygx0`AJOd)M+Qj?B%<&CqHkHRVc-w<n|2a&aXG8u(ydsbNi>e{S(_wCrhv#?9z?
z$N4P==rlrnosX~z^N@L(PxZ`1G{%`oiL94QE(=8)rK|U`8N#-(DA)Ow*te&%n68ea
zD-a#n{nOb#T7(F%tGJ0z>^@XE4blYfswj}UDSmVMyo=2t+u1fp!#mUVvgL4i<KJ!G
z<v)tM4}eNJ!`~g9q*vOyIC7$AF_Waj8<VExDUvT<yxsh<K%*GBxYv+^3z@+_NS{y3
zeD8Ke5PINly1n^FF{fb#*Xz7mQ~Z0C1~JPQ^_JI&<*QK&*$eEAbtNHPNJ|uqcd?nw
zeSI5G7MBE(mfA|CnK9-B&EW|be8GM6Y;N;t9q3!_xkl-LA8TxS7KAm-JmVmtH`lnk
zbaV4ri*xC2$PYaKniuf=7gCKw%x2y<TAYj~c6tQi_bal8YPc~giT|1K8>`4|c3(!X
zce>1v-~_>P516uRC)Z6FaowolI3~D|{PSFesK6qo?;3O4B~*QzZ=0MaH{)ng<&@3z
z0>A|y=UZ!h+$0{~awY<BXh_@n7#yU2EUhBq$|(0e#ij8mEUf1(z@yM5P^%2K!+r)^
zBYqmR^+vXbM!khVSw2M+Qz!?Ef}6%(=E8J+xuYv1xVup0$*^CJn0_j$Ca%&c(Pf0*
zXL+11cVlnKdH3dBPf09}qS@@sFMk;Wh{v@yH}812>UcP`ThJ;dGH`$lhHV;WNykpF
zZ#Ud07%{4Hw&suc<l98H1!#tC+8@Mm$h|vp>Y$7B=GgT2YNysEzd}yc)*>ecXKd0M
zuflRfRq}$)zE+iMK1p~zIY9=+ct?MfejL_Fw(HFC#T4FRHbQB{8^JTvC#mJh`=+%o
zz{^9#RL~Io4ayX!q+!le^eIeGM37bl-Ih3r+Kc!8*YZC6@+N(v1ots8v|hYbI?IUc
zytk@0+J79oHC4W|uKKjn01XKMTkyVFoc`w8=1LgWZ9A?tTqDtTFGnK$btL84*kZpJ
z_FLE0dUyxzX9HI!Q|BoE^PMRzA0H(y;p0nOxuwo?IYzv}sh!h}X^qUnYa*q*psUjI
zJ<krF%?;}b@QO>Lmm`ouM^vVOZ}sj=pf;}5-p#>J!2MZgTVV|P%1+3ZL@pkjPi>fH
z(@|aT460tnaQtfFpAxkJQH4k4VTm1ick_S|!jb)R#~)qZ4`Qkik=|XU{Dj}Si-JD*
zUzl9YNwOdb)!)tVU-!Dv?kN`j!0mYZws)SV%=|ONI)UBw>k`nIp+4fs+VB4%zH(We
zsrtyYmoOjCdgL#r&>j}v1}}5qCvC4l8<dQy$owHHAEKUH(0-CyEeE2^T43|V1UgkT
z&Kwzxi440h?Hll#&rUluTL>D0>Uc;5@KuSSlTekU_4e~+X0od|kNzKJ4YmW^TpHKN
zZY*MB`7%uQ2DqI*xf(49Gpjy``_C>AM5D_CRWGz!m%~-YRpqtmlr^eUB?%JFDd}O>
zg0#QLpr=ov>guQ*_HMl;u>Mu<%lz(Y>(`Z*;&$UB!zHUz)AOXoL-@5B*~wy5XAZ8I
z1aUBkSI3IFG$hXRxzZpD<$s^Lu<06`z0YNWSzJ&3H-VLQBCzRy=3x1@k3w@Np7N|N
zkOCY(IoQxqYSO})gyj<*C%)>~@UGrLg??U~umtTwZB-6c>fccN)~Uti#lVw;Lp$3)
z6{H2JX}DkU<#ID~f`E1@K7*Lc0QoWO0nTAUO&xa-`zu5l=-XvlF)si4Udrj_qRy@L
z2GN$x3P8R>LlwUKjp>DJ+dD)x@t13sFni1yJyuD=*J&_8@VxPW!?CWrcjZxzyNj;v
zf!^K?jZ!X||3<}g|9G``HWGnKN=j~BW=*6P=sN=7Nn4lGqiy_*Y^#gQcdd=6KvTX~
zWVEzz5P!zT{)@26!de<~R=+Lov>l(Batm2_0+KoH?d>)EvuLm7jD9hVDYD40jrFT5
zY=gH59ogV~84qiC78>2t<4>i~iZnND;73oFz5jra5RsfHTXfN%sjr1)7MgXBdRh-V
z*fw@r2r^)paH=iGxE$Jp+%PCV8G6s7443ok+tU^&rvlFzmqt^2t}o7Ss$vwApxtKx
z3XaL~YBwO6go)emf;MZT8#NgMMRLrLX2pmzPnv9VEpBlbnS#ag2n7X2PI)bVb8~a{
zICsIskFwuTeZyH=ee)&tAG0he`Y#@1o**h+D^qpHrn<YaG30FZ4`JP!)n+_7&mW8l
zakmyI9Xw3Z2S?G>g?1^Bz-TaS*9XAH&=SGL#nkcj`7(#8(kvy80!f`dv%*tV4r_CL
zlvI(IaT?jd(ankf@-?c~c02_;2;PZ*V0yaZyoAIRU2Og90g`4%H8c<Shc-4ax@G-Q
zdTG)zdT7~w^)Z?}D<TSFYB<*vx(e`j-Yh+K0q!|Mb~GQc)tr5v{<Zzy69WX@{)M+L
zmX<%lQ!V|=AWNg^64O$R^;~2QtR33bG2&d6@eP;RIn@f1QoqOc;{^Xm0hMN!!EaW&
zt8RZ+?PD;F$2S)ja2@LF>+6{5zdc$qx2>3f*#$pa+a@z1y<(Ae%6TGut@x)rMO4Yk
zykCPsufSXrZz;jQ=7)w%r2q~kwx1L;?oS{(Q=*CLl)`iOM{$<^bwl2V+#=WH1%!&R
zeLpw0SwkUUG9C~yerQwNMbqs=(APw|0(N0s5g2-xoGuDBA_HT_6DWG{EqtP4-Sb6r
z6LyhCXwWrpxp<AhglT*V{M)!nO}sQ<a6Yb>Q-wV!3KuoAj)_eJS__mxB*Am^IA9*V
z-5mls&iXHt{xvc@rL9OAkhI^)Hlh)=!2G~Kq3+l~Y8(m*XT@1Cezsk&)YH+}P=@w1
zr#`0N_zL+2N9Q<ZE{JoAlTUm7a3n|}QHQ)TCnZV5<AM}Yn*@GWW_dD}vSr-3m^1p}
z)2C0WrKf(_+vyY_T7{P6V<-Rg%F!9?D@cNt>Bt1#(1A?$>Za@?<PZjGyv?mrChFWO
zNzbNMyK}t^xJ$OS(qSMbnX7{VE{$*nfsM;P{i<1Geh0(KBxvB_-@bhmj={*=0`*0g
z#yDYk`6EbV&&>Q~Gej30t3lAUv{wfjq+`a7i@Ys%Q2-7!)-wC;@r=t@cH=0#eVpx*
zV0fQkPfGq4kr+En_WBl2&R_+0V_Tjg-^;&!1qB=kV{oj8Meddcl4*}w82;rc9a$81
zEvHr-zyByB{oaeZv&e;ps<4Y*Uk_iZGfeG!XBK1E!+?bF*2Oq&YKJ`8h=1mIhaUs&
z&&cNi9!vXT&x=_1S0)2!0f~(6(lj=hS-_rbx1IlH`(N$UimJV0<<y`4bKAXrbaHa!
zpos}93BFmlG68|*M}AP-2~~Wu=zNZHH8-yqAET-5t8spxYct)Vr>YEz^Pm!6m8v9f
zJMZ(my#`J`rOz5t{`lA}((YdE+uqR<SVs;VjVvxKHAM36Q_HT#KfT&}J3x!ysiV;D
z-|39|ac=i(0S*CW?=vKG?@E30&yB@{vo(vvhx_=qbTy61+7K#R+AVnIW9guyu&q29
zhO-$0J^g5&V<-ZhnVk^Ll7G;>X3+COW2)Qgi%scyhdl#Jwih{`Jx69gPLghB;wv{m
z{pkXI&>DK#cPqz-C@cgPaB(2tio9Brz3ZX1<xbwYIXapYn}*k-SiC7veo&BJiW^Mx
z&9rD71l&7gj2r*Ypnv<%w!VFdxOiVcqK|(!5sO;9ZVeIp=Ytj6DB?y@)%cRDRKj0r
zX7^gKw9%bqD`eIKywBx=kw5g>QnBjI4;FS*ccj<~1<YYOa{U*p5i4#DD=t}x5$_+9
zOue6G2BUpPCk3_xEHC}<j{aJA2!_dNQ;DOtsdSzyy_N;ZQ4_`Eaadf<ku!Yd`uo84
z&;>UnSd}M)eZgP?@wfyMY79<<?qL59#{>4z=LzwXu-Oei;Vh=zgoJ(OP#?gg9I8?y
zx>tAXoy(>AY$&MJO9G{l@;Q$B)@69x=#Bk+@kb2@Q{&X4oIliXE=c69MQ3V(50ms$
zHeuB?u<D@DH(fH4`VTeykHAPqZx<SL*x}vuJpFk%9(z%ENlisjij-9A$Vj)XvG@J9
zMd51+*JE#sG|q>r-|JPqBMSf?c?<nV1dl2FqZMEQmq$#&do6@G*XL(rV`IS0P_0Fy
z)IV<%WJmp|L-lsU@uBZHU3-ScjN-)pnOD-kUe=>-)b{QLy+0^Lrm0^(vjl6}>y*l*
zt+nXP@sZl7!TO3l%c$g$e@^@b(&XfyN)-uPfu=f)bBc}Gq2CV>{6KR@Cf8+Cb)1;o
z2*QHmg)($j)Hq+eOtaCnD!tvJu>rL`ru}u=L5BvdM>6^yxlb9!;9y!s@JUpE&urQN
zUxEGc9#Zz@OUxP3TKm9kGCkh<u}$k#AojP~^0Ui9GRGf7J3Q~+5pc9Vbn1!P^PHV3
zRgkk%lF@hj=rJ=h4}4p0?P#VyA(~A#W`~s|w;i6XCCFZFJ2aq{hv#)NgHhW-ufWu#
z-y=a8Ic|UE#2-J_5R&nxk-2<-RbM47F}K(%EvrOmD30Lg&fjq7Eu<zf2g}~9Qr!QO
zz4zLvIM1b&en>Q9+I`-ta-MT@t8drQx9v*F*p$q<DnJ7a0M(NFFz`olt+v}H)3%6)
zxX&Z8xWv}dzR$<oGhgfvvH~vjL57{zXQdf&xRi!AwSwwtyUcs$%(5|i9B@>oI+sF}
zW=YS8jd8!e;>kOWr1zo>idwxqONf+(AGA`|HL47Ru>Rh{0_0(1(4}F}xzFA7>Rrgv
z(RtMNu;rx#VAE4;EA)yL6Q<*5`;?X4x=euJASyQTk)lN~zFf*m4y^*=xQSYbV7S}W
z-@YsiKupQ&Ybb8wh`-aWRVX-%&U;J+auHH8t35!|-)e=vs1lTQ`Ys+~T@Gok4x^2t
zj!w0Hm5;QeaRqA}9+pQ(1i$m8QBWpUfJB0nc?wm4IjH6jFPo~(<Eh~trSbT@4(V1C
ze#Q@xarm)w`(|LawItBTciEsS!ly1TdHc{k{@7AJ1f@2N_Q#&i=O4vGOi}7dyDX0b
z*S0!eFg=LX>7Jh-)#uidLU;ywkb_iJ5PE5x#$^OwK&o*JXOg88Hu`=nM+T$>42bJ?
z3;i@e!yIm#WzaA-Rf9~zU4+m4_K;`urJxR8d&Xm+5ad^AIMxhRL!8cvCkJ&dO91YE
zSE)WI><w$eJ~SAfE46iYc2+!P-?@C=VDJ2-8as}kb}=`%Jo9UZghD3ilIuN+{y2%n
z#1R{()tS6=y#2O(y<5B0gVb4jrH}a369g0%7;j+%N&EYi9)c~Ff;16SZBh(Hh%@(Q
zs8#86h&>{^{+ej*GN2U8#4$GAAJoZ&>g#_Bj_FfXA&7$2`MCYiXmN4UpW!YkNkIAH
z)i<{U81u)hwwoi~XEH1v__{XkZD6rZ*;@k0M6Br3+Jh6KrCn%`tAHbyNyk13m|Q{%
z_59b`r#3fshG|_lUR*{R+i>j_?l`&EH~P4F!E;HfYef`#vo7<O8To0-uPGis@Hr2<
z!J1<C94)0@NZnU;T#~W?M~8QJ7s*;O`=@@!4o)^%H8mRbtc!a^6#IMWMQ942cuS_%
znv@>z+!BBPHG}eczH+mjnoL3<QW_KBedF$P;+=wO_b?`SN@~SMprQ>e>}p5mXbxQT
z=eA*Ogi=Zr`^<t{Q+-a+Bh^efjV3nwX+^gyJrtd{jmzF48QRKy*i+1xU@@PF3u3j^
zQD+Y+l*?BiAL%j+wHEpmU7USFzYg;0s-G#&g@nP|U2uN@wa<69x&&}p|Llv9Y2<3s
zm^z5(+uAC@x3bbqxj9@0&dWP*VwP_<XYY15L<q6yfVQf#vN;914U=POOx;4_%c>N!
zsu<~U9H@xSH}CQ%g?5cC<@r1YPe^)4dRF&mJFlkiS3B>KKFdKRi!ObttRPRx9QnQ2
zAxJ7B2eF5``<b&f2AFKh<xWb!#Zh3T{6dPr#srDq%d;xWKX&?8Hzvq6oE3BYDS^f3
z3J+^nBljRB7~nc9whwZEHNEVq$~ZQ)*q|xvKp;B2Us%-lldwfZnM=cZVGy<lY%a-x
z)>Oh_gfI|gb81uH&^rUI6<lr;l2p(;Y%(`Lk<S%wUnZHYgH)Hf+Fc(Qx9E^)&3vT)
zs-l(#Oc~tnfeT8+2^Hh}Jb@(iTVPmeGbvpP>K=E^a!X(o@cUx@<~Chb64UNe?;Nh{
zPjqd0-!HCB0{gaNf=JJ6#H9laa!ni`zQfhYwV`-$P8*|WOkG(Tkfx`iyY-dYt}Y1|
zfxxDZl%sz9BbK}WGjhE^Bk*&lJB+A&151w;Qx#aB9<4vOsHP{6)X-PlAHBUkPaozK
zv9z>I7+CuO55}E@Ntya%>Zx8o0e#Vpe_fC+Wcm6@bd3pzb>;_b`fnUC#xO-(2B6TV
z*&fM5;rv20q-^17>=XYE4!-fwv}>aFyv>8ls_k>bAjX+I-t3~my;k)qZH>lhsc8MR
zdAMjly>r@Bw0-^l9_C5W@BU0^*4S{jn^u;mq-n)0nqOrB?(O+AowF=>K|+}KGgKvO
zftz`1XMJk^WZTLq18X!@P<-dO=u&S6mK@FYD58cxu^cr>-`Y~iWk^;|gJ&)p_?zop
z`=3{GKeVBeTblj=U=FZ}UaW3Ye~i^=T{1W0#8IqDvezP(?JXTQGmSw6luA7lmpd`T
z<s!D_m%IPavgaE?Rkh;#*?E@nelP!de;C#;pK_`2KbxOw1CZ@s)l-N~vP2!VxJvq3
zhOW3K4I0w*<wi>0{bCMXVTJe%+z!VuqOG2p79q?W_slpbBOsE(arI&sveyZ~-O<S(
zLQFv(nRV|PFaFBxNciQ0mxnH~QG?H)f!H}<$V)qb^2G$3(*t$oZl6hKb}Cu6)++`-
zkF7Uujv_~@z+Z3DpQs$}{r+<aS)G&2puqb>98!Gt18?|8Nj<hlZuhIJCm(o=%&?~d
zeo(vD2ow3we>GIWs=P2G!E~Lu`dRdIjfd&~*uMO62V8;FhBgY-D6I6~e!9unuKB5L
zO|JX0tTe0GSTH7;Rw(#l+X@iQtB7zdGGVJ1>3cj0XqF#pmPFUf^YT(hLL4k9;VHjx
zuuNON2$Q!5tCzJUn4k^YdfhR9uuS+5{Hp}Q8o+unU$l|?=CtJSboNN#^j+3Oj9ozQ
z2pN<Tg5S;MMca7C>2u}cx(xWMWOV!M#)j)cVAYnTaZgo{;kD>2mVlE(?g3j{!{ePY
zCoo`4uPymhX~J8ekV=8?(du9Z@ly|{zNmJSDhGk81eOc4*W+DXFzRq^o>ZzR)*`2D
z=j_^<uh1crX6@hpL?dAS!uLsRGpXYF8M}e+vL@$$`!hQKN1r%^=TFF+SdU*&gwN&b
ztS_~QZ^{A3@6H8F%yH@FG$P<S3O6N|T8{u)ek{FteBt<V3H%bZm{pY)Y?ajeykQaj
z!MMRp(6C&;6=^Zzlv!+J%vV%5gSMXkT|3v61XbDPpjD9A@@uy+pB=l#itD1}i9T4v
z!NiUi!Ai{50Mcca2f2UpIn|_UmdO7dO!PHlL9`0^sjfZZcTGiRWZO9T<2Y}f2w>WQ
zAa)Fn2y?AaG1;1(TD1%Cc8?@u7i9=5S54-EEj9hOU%a(NJ9deV5#iN)qE+TTJ$AKc
z@SBy}(NX<bjv(Q!MOsmLpA&XkPH}~e(6ssRgl~JhvnPNW1Worew@)7yJ##Kj9`62$
zi2KLH)GVnCWx41Z+Oey96<n@A0bp%$2$!Vy;>q@DHvJjH<yoQ{Ra?nDJz)j@N?7Ht
zT344()c)wS3b_3wdl6**NN8M8UB=1h=+0{0c?j~m_=QZ+1D>s88y$lL7HQt{c|Y@I
z-~fNZ!d%mn^Zx$>>;2$KT#=2`#Q5|)|5|>2`2r@uRb3~dIL`xQTKOn7h0<({E6{vT
zpKFND7R#M*->+)P`hVQbpP*n@dLkZg9-;%zcCFo&`t}<c(y;Tmw`rZ0J;_weqX#|8
zd+3~X!_e3?iY^dCWj9*#yYw<S27Y{fsj|K4eb-bj=`UUk?B`|8a&Uy<xb;=MnilUo
zoQ_Yz1~Bgc!4H53$8%xIpCvzW*u$-Yw?7edeH^KpMVF)15UHSMlaa#z#4bcNq;mFq
zu9cRrid=6uU2kJ<ks(P`QI%Cf>BB|!GX(7UL7nHPRe?6{9%mcp%Nb_2m1cH3wo~On
zQk}W#3&@cAP<e3~9=QXVrq_-=>Ty0yp;Ux8i(g#~`T_tytWZ`1xY;EY+B0wzPTX_q
z+?B|SH*Q^eR4FZmK8<ShX78HF`ZwmGz}g;qha)Bc_^P$d<8p<}`BB9~<*m9E!v3>y
zt}2eP6|x9Anm-(xcA9qUVG39H&h))x1(yJq7ZMIpAS@QR^?0Y){Jikj553q}AH$HJ
zxBEVXq^tl5*8|49(orXx*<juBRXb6vz(LZbQDuRpoNNCSxBpccN`9CDQfB4hgxKKw
zMu6C~)z~0$22?t%tPkO8-8e)c{H(o;>Lir-A|$1vBr0SlS{J^6%bUb6k2~-8_XV4^
zXHQorCNBHn#%bpi37f-;B~w@<pfK$s5*fEq3=GE8DvWR?i)vt_%MQ(0&-6zz5;b8D
z<Fqk9{EFZGCg|I`sh#W&Y1HyE(b0_9h%_Bj{gJ0kp18)llx$6IxJw3hZczBzo>;*z
z-=OMKr9y}<wBaqnt=XvY$U>*0Rq{BNHa&S`v>v3p4t@Tbm!t3W!~vP&hxq4yd9>aK
zwD-J(P5e#@>Kt8<f=%h67DY!h^z_9x-J*g=tm^j92aYeD0^H<K<lf#fUiNh#8zxM~
z%RkIv=m~7Oh0e~E(zj0C>bUhk<UFJj6cl_Azi(qqsH&(%0jrzgLoEBD5nl64o%jFw
zW4E|v0Hc;CbGzO*bj+{e?c<Xt9V7*4n3NdSp4a3|T30uO1(n!;4hsvUU2<>!=ec0x
z9Fg{YrN&|5Yvu@v0L=HN5_NtonaG<js*dbJ;rN?_MwH4uZ^cJ@TMCoWf2T^YvcJ^_
zM7?ma%kJ{chNX=EU0XvMzB>%0gU^C<NT+w1#vOKX`G;|TiN=RDU}f-K{B8`jvr}AW
zLFKyhzg|fC#SKa2{L=fybfGS6(f!f~w-Kg?!Yz!=`v*eAMmYG6MJt%w64~pk^v8{@
zdzKe{mRM`>?Zf4!#RX<(fue;y@2V+mVq3w9Y&BetH{4@?qNm5Db=t)T3v<YAj10@E
z3fOpZe`I-QkXAUJ_j2@DPt`olMn^A9zOxMbn$*Jv{&L=4Z^W>q_tn_|O_iI-hRU*{
zFEu=dwCN?tyQXDE(Qa!uV~1c!#E6FgL?ReAsvt1zpl>)Zo|iE2?!#J}(pKA<!o{~v
zBir4*oq5qV=XHq1Cl?F<o;{nW3$<*Lw{HwK4CU|6jzGA0!p=JEEzCA#i9nc+Tx11z
zLlQxVTq*O)sME)rqum{O+kq0=^q9$H`w%hkXKj3zfNz6U^COSRx*u|Yd4t))A+2wV
zs3*c6Qx^OlL*fLqW;VaHW`h!OVRq7N8gx(wi1hewEX$vd)j9#p6E3F8v-&E^FI$&{
zE76WK;*Sa5@eC{mN~H`5?y6RV*}bFV1@XNL7Iu37xfA4BYgQS;HeNSVjNGS=(MLJZ
zDo&tJu%v5|6MsBRflsYccnfcw7}<9oJ~Tg3Lf4)*kq)H#_;KRD7L$>!D5Er+caF~Q
z%?-(zLV@#cgh50tH@-9-Sw(%`OW<sMq5!L;qlbrETq<f2Gj3w$49Q+CVf?~qK77Y8
z+&&9jPDv%mNWM#`*MEB(WAhHp%C#eOl%e}4wa6T{ByyCv#uu)`*x|l^Zqgk<9m}9&
zv{$u3sc~glOjbl~n~9;SZT&I$YKYM_;OyifKqSTAd)NKecp6bRPS$R5e|1%iv?p?v
z+N&@U#8P!X{;EaC)Pt7w_><dvKIt~Rx160ev9GOke5cFd=2OA~rbYkGC|qRrKH4W@
zx$Bu4ugsMWI^JtszFVc}d}4}V3A~_ucJ|3%%hDq7V&YH=bH`Ndu*ZmY;Ef9D$I@(o
zzu0$tpKO_8sarsHUT6U-WW)JyU-RlY3A3?L6$SomY;KzSR<QCS_FPFQzKUzgI!I*L
zlTRdmXr}Ck%*6zSj#~<8f_OBH3=9jGQ)NGX@+iGE`k=T0u0+n&kB<XkOokQy?Yp@W
zi&=}Si8hdw(5eoOHH{<_`MekZZej;$P$SU!7A_aXvn|LO9svnaR_%`B(nv`fd$HO@
zKvdoJY2x~!m8o!~sXLz>LvS?0n!q<G0DD=xJ22absk<NUyj;E+y+0oC(kBLOF;K#N
zb+<(?+n2A~J1-LMF~iBUUgjLN!mqzw%-&;>9<IPv<JYWWhc*wDq<U-yz8t{gk55Rr
z-|M{enfAMM?7S<7m!vkd#I(qCN<11|zCX*;60`(I(&(9lDNN#K6%!WnKTrEcJbFC*
zrm5-Ts1;+!8_lCaMO#6Y6s^N9yaU_(TBA{9_O4vlR)9K_iLDkweL-670<!*&c|@a?
zm$19&K@`ha@Y`3jaYPM<F9&d~gOqV$DMzte((e4?*K-Q0WI9CTMph6V0~|WoyTEs0
zRJ!Pt@QMnl_BQz0-}SzEMdz>DR5&CA{PbVww13#ykg3O!#+{Dv>p|BFDAw_Cf@%1K
zqlF1|By1f&+eUun)ueRGddJEhJi*8MoK5TlPN;S-&-<iJgyz;n&j3L5>3lypL*W}f
z6c)|;YDJD;Ux77%RjVt6?dcn<BV`<3&_faT1C*G0q9pW7svH#N$jJU0Ct6dd8lO#H
z0j554B~++a(%(u1^Ynt3dYR0UlQ#2z)-2axi}g7#Yk^-J=onYhhYH*4RNyC3u;N4f
z*}9;{Qm(=xBojN{IY6yVvl(rl)qgK#WoJ!Ny?8#SieYD4hp5W))$N1Fl%r`eYqUaG
zSv!pjt4vVrFJsQHyo?dU1Q9{NQ!ormt+AhB4_(Qrdev$}#gSQKw_!ZvP``lKvCi~4
zBWfROug(ytn#4bkM>%cq7wRc66qzEH(!9lTWTab`HNy=6tg|k})@TH+zxUzTT7Tx5
zy?p@|=r1TlblrNJgVZrLgy(*MKaL`jYE86GfvS`sc@1g}T5TAg)Xy(UWtepZ_zC=}
zXss1#?MUHs_G;;?u{*yCb1`*j&m27#Jr3tm@$aF016y%88Xa9%Enhz@5XU}mKHlDi
z|0(Jyr-%dM;srz5#?v;!+l}cPp^9l6x6|%N{RQIEA5r^k5d*F=S9d-gS}?^|?bs=a
z0=+;4^en0gXPlDy=W8}?;Z<KB&zs%P{K`0NO?GpUszKN&m=sH9Kb%CBm(ymKvzdHY
z*g$D%Q&5rZL4hg$jW==I$CitsE;k~DXyA2Bb_Ot~)bx>~xIv@bjWOMIk@oVzP(tuW
zf2i<#qI{@Y;K6U6QhhUVr`YVdFc24|?D?798XKJKJu54b)|AbuEpY2|-FZ?z)ZM+l
z^CoXw4)wfV)AK-687!y&CZy~Ck_BWg*ORdo?TC7fW75t8wf`}7x>{kwpW9SC&C`#8
z4F%$tOz|drXN^IphAx$Y_f^34bzH4$`+0}v$o$Z7i^Y83d{Z8M(Qn04MeWdXulb$L
z>TZ!Qi_0t(h>WEQtaoL=(r;^2|IPJE$WX{Mwx3KJJhY^$j&PV2<razuHdh9TNJ7Fq
zy1|E2eN`ZwI>=z`h+jcwQ3g%~es4PMATA3IW4ibtE*GHwftz+VU0g&h>RF;Hsl9ac
z4<pYK^8`>-2v3j|Ye<s(QwSvvxKur@{%?vN0u5<{{3uMpwa#?SN!8_wcBywS!qPD{
z%>AsX6+*%IA#!{oOY|(vFXx!|oV!ZMk9F~?A(9%V2m_&Ee}y7evfp}HOVeHlC~?hi
z+B(9}3qSzzsj;8Ck)53FCx}yrkDHq9ncORUtnn)KPN5<zp9kw+=ut|*D(WCBX_U}s
zTuN_mZwUyfV%D*!@XIrN+t?Y^XK&2Kb<3bSCTQ;}Mm1*}V?mnZm9nyi5Cg8EvNBz+
zxc>h_4nIh;kP(hux1}zly&uc^c?JgVISJ^f16;}SjfStDGm({-F+A{TLB^*zG@Cut
zsak&6y&vmM6(QN})UF<9XS487=iA!W7PjfG+!AoK{c}693^+*;^X>T@FNohZT&z#6
z-69j^v-omsi4(a)(^PckAXX)_JD6QF4VoE~(iBl*QXDd*W@A~|hI?}-&^4k1NNY?1
z#rBi@YH<(5@F{LlKd#R%)=z3%%tnjLU9##coy~px@aUb|Gcx~xh-USWcbcve7HJx;
zO(NKa6f7R`s83`$#r0*|{n5?eyW4A!o5xG9vABV?tSofxS|`u&(G&{{u%jl+jU~vF
zp})jGP_NXVa7){q_Djv2NghZ@unYjOyhj-Q3nvxcD6<~jWqs3FJWF;~?ao<35R7%}
zYk;}9YUx%U^Zn^^K&$&$B<O~ZZV7kkBNTa6gA(k<L?Y=nF#Xmjl1mqM@Eu3b$Kz*n
zJOmDG;({N4%Db$wXJdkY`ILOlAf@{0OVCs+N13g4%uj_a4So!wOx1PapaJlFD0XO~
zj}C{G1f#TR4QD$K_p#8m?X^afZ^|E8ENu728Jxz`yY%Ae3ry!Z67v#%ureF8VOo&G
z>{>3Zk=4RWo4jIXUM0jWe!on_5^!-i5SD`^I%m1-uey&h$n_yKM`yO*hOzw5VSRHa
zOFhk?kV;oA`y2a6)%;-vUTmp;rDHL(KFT*UC1w3F``e`1fFJ!V;OeASd^rFLewK4(
zMVa4I=fhUmzYd@IVkVTn;hxyoGIL#A>cyUTs+DjuVHb@{8Eq)~h>(p=*v(1MhK==~
zPJd`Qgff8Rb$s^(9RbteHU9*PkTt4dlL`5U;{b;zoP7lZjXN3y(wK4TbRSmKHt4GI
z64T9z%&j<eMfOz~Esty-_o;3&+oVhqtHt5(+PzbuV4S~}@Z>SFCIKZH2+`rb2%XEO
zT{LC$bsux&)G0GEAR${fE7Pb#(qr~WIJuLt!J0!Hs6%upUG#rH-q+POwze*#&kD>~
zpW<-*#Q&o3Lv-{NP@ew599OYIr*NTei0YH1(d27;z1R@E@7BE9ywRYL&|{RztkD+U
z_Zih9J1W{3(8w6vN{%f;*D^|p<uG~P$@~89wzjrb+7;_moPTd!H@}E2l~{|3QQKl_
z)z(LAmYC}o>0+fryJ{osu#V3FJ4P*-81(NHoGtg7_)`Sv{@Xk%TgGJZ8;;maQ=_RU
zqYsb%agCDUotrWjoG>Olei*(rE@LW~ju2Iw(o=Su*)--JZ_(FB%%*yM=h^JhG)VRG
zDG@j4KR+Lc>2sLFsP@($E9%Pzocx#`vy=vC%w9^*Iop(ZdZpPEnHExs;zx*S#6ni2
zq4ubzZ*{<q_vrg)^<zj`>9)+u4^w^a^^Y;0(amM$-no>ua}3`c5i$9L$Jt$<AbvUC
z)jWa0-EQadne7306-MefHg61}Ka8q4t#h|P|1Hz+K#c{t5iV8B_VxmdhILYVSX&k5
z>gX@l&Vm+4!Lhu#t@U+$(AfC6k5X8TQTwIDTOj{`cfPy7^F%B%for3d(B81>9SWt=
zRG2(SJWgxMjPo)<=7s+2jDgl$s7joZLQ~vNUDZuiNJ<hr<QXpx41!&uEx2ER`xQ2c
z-I*tpp8En++|O$5WZAAUB0Hee7T4^P=m(4a>U(qSaWg!wLE+P#w-?k<j<Y%Vdu|s9
zx%B?rbXzeC2?$<nsAzf+C5)u&8L6y}o-CVokK<&w5sd_`rdq?yzB$F`afU1R61p-U
zUM=6a-aAhIQ~&Mvvi(0c9_P=~s!#f&RKL0Yn!o(&QOQ5dh=Cr}4j&JlZXB27J-5-<
z-<X`1y;(1{Z?rHo4?LI{0Is$G8HRWuXot1~?&<cFi6ZE{^Lno@pe1uWwCeMQ=ZjW|
z@_M;^l|BM#Cy_l$qo`v-tDY}u++ZRQJ9Pck-cxt}>T>8(O6ilr6CCP#!f>kBtdw=8
z@aLA7E4GHTmj&@DzX)6r)0XOJF5j(+T8)QZ-TM~vqa;oqGvZ!<;x^yIti^Chm?Bad
zS4_=-J|NwA9QQeZ(tTI>N2P9vTwT?zcROx!WH*P1nk?UbH^~4cHEVQ4?KseVxe{IA
z@v!ULvcKE?c6ZqI#P-=mp9eZHzfkWpWxeLY-u{>y^`c_qtIuq{;Kw|2R#s#<6{P3|
z`aK7wkO;`@C$IPtRjK?b@k6Z?K%)JiZ&=U}A9hGV&7%rDBQ1DXME*=0tS}d+$AX$j
zzbFy+^75L-cEjb2In@8IajG|&@w%N!7UxPHvaX(6TtuJyE9(4-Q|!f33~;PnU8}I5
zJ)2a~<I8*=Lm@3q>vn7?oPa-FjSYr0+RFaJ07_8-b(uWD=a2a%mWHVMU6ZrHo@k#L
zkKH$h%K`sSO0(jIJ#RaCW+qm!8k-PTa+X@(T8@j4nE8;X>4@QL`y^~EoL{K&$h$aB
zFs<C`B^;NG-5!yqZ$GOz!L-YA<`3`WEDUZ#dO4aB8C9V99rYV3au#JJ)TM6oy(crg
zkY?GIu|87v4%XB^&@IyrLNqZJRp~ay`udU@08W>k!Ec0tcur2G*yi4O$D|&5<1fSf
zsNtgs)-U7ZLf+nlvxc<r*4kPhaJS@>az7;IYw`CWpUyJ#$;?z6ub`j_hx#Na{2b|1
z<)W#<_i!W3ay{^r(Um1mM(}?7{_a?O`JVpE&^0Dt<H$Xe^Xlm4{j+9^G25{2VA%N%
za9A{Pwdy;k5v~`&O-}?%i+c7unzeD!!Q5}lED4%g51bkT`W2)pxb%Hbl1%aSF83oZ
z;NAPJIOl?dxH?wvhpU^X`FtOt8;xWM6kxoHW`YtJozFZ`O)u_@`r=e2swlPvET}=j
z!5mtm+6QB#Y`<&%nDFu*i&m9Ai`R+*(sUOco;vkytkI?fQ^8nY6Gjh?A;vWg$X+mT
z&zWA|Nny;)vket_3+1R)O9v=eZhK^JdX!$zq9a({r!(8$;67J<AllrKs4L0ErVj#<
zq|efJ-YWq-!Jm-cRjHo3FZ0aCxx`<WOHYMa(9TQFuETL5u_|Sg#r6i`iu$+!VK3yj
zwJIMGPVz>}66r6+x5^+ZUg$UN*Y?4BYBj4D^HT->2`MU368I#RvESeR{T(<(-tdty
zw((VkQ)d!rac1D%%7F8s^woo~6n+>8+ADCV3u05t?;_i^_Q|Jmg8sDTP$?+T4&x@H
z34^gkSV7Nlpx<Deu*@vcDOEP0cKF##CW@HJv^cyvShtB%?+A%JE-n#pJ>U55wMs-3
z*-sq-@A!~913tbt4@DsCh47lj#_R2g9&WA2qqH%#F{1wxVF}@oPzpAD3ey_MR6{mD
z0rs*@wvz4BEn=Mn4d0HI1AnjAW(X(im2qQ*C_q?o2w&skD5xYCeNF}5(hlPJQGF-g
zW~2^{&S%;(=j+xYK2b@MaKzIonTjT#QXQHPNZWcrIl6TWRcZ4v!B*sqhj(2%RU7eR
zkgbLW(|+b-w4_^EzL)hgSbL*7n{aeJxv-QHEEqQfL{M1z){555VLi1~LQ*1QSv~#K
z0dT#)xVXp>Up5)plmeurK5CxHD3@wSb$|2l!cNLO=l;wP{>x0O4+y$)|IhK$h@-%4
z2l4x@+DFY#cgc<2n17WWiqDF#-}=^-=;c2w6|Honm0RNfEa!H3baM3X+4w_8bftl7
z$7QjKBscdKaf=egxl=o#KrQvwaWmv-dz*)^mCbre;-|(F9n~a2?C;gpNu2nyvjc;y
zWK)akl%BVtS8mbr{g}@g=A~U({<1w{1|8xt_S~ebSa54N%ffayVEZ(3?Y3a!2Akvi
z1SzXAMi>+zs}X6H-LS2(Q9O;H&D)9d&Z@Do+y_1=nrHNwxNJnth?!Y#Tx3sJ4_;1Y
z$@rx#uw{Q+83>?Pb;OT#e;ZM)33!He;mQ+FpuK)OVEGioWmnq;39^A#BRWwGE50S!
zbpqc|f~0if{VG*`^NUgRypFQb;{k%#cT9yIq}RRq{{*@>>#JK!_kvsqe!{z*|6}_A
z#-WS|20KY|U$vtMn-W4_C<oZyH<^n&J<xiJ7Y$CRVFlzks)_%AS2eB<h8D}Fu8vkG
zmTyNb&#$o)^_;Rwi`-mkY)@R_e>-(lI0#W@3>iHEDJL0-Z%g_~>BAnf+-xd4^Z%&E
zgyP?pAX<%^+ZYo3YKO1_*z3YBnF}tA4{eNo=H^KErK)2DKhM*rU}An{dxCE-KiXs&
z<PwL2r2t_Zu(oR9>*sM#V)W}R>XUm#;4D-*H-Dns{B1BU9~nA0x$f5W`=w0ZmwwaL
zk;{S<=q6C(hU4QxLOGv-h@r%VUs6E26*gbCdBHX6V<Ad}%-<c!74mV3b<YTMR(~wF
zJLrqOZ$>RHc4yJ!T(U;JXJx?^i6E!lPLqJCHvu0h<(5`fvsMXW2e}%cY+=jFYIO_u
zvhOhS1*YCP(TMt~A=e5sujBTdc_j67hG=%;=ul`Z`COz@BsuWxY3Be7*#!|@Kiy9P
zxOCSX$wMnOsG_%cACn|1;7P~fdhOKFr(*-8^%K$Y5-bxsZY#TfHfZKy(q0bY_O>0_
z4j{+uO5amQ^h?SethZraNSpW|gcpnljPnFF(?$Q3%Mg6|4*B)a81s^y7PkmG#{7a}
zJ5g9L$|GTf%5R+~h2$N6U2evyRlB0gOG(U22~2x+u{a{k;@PR<Cpxagl;1`yT%|ru
znR(32`&BfwAG1*{@DhiAGxhWDu~tHXU|>g-;;+qo*L;Q6_iPtu_@W=?0~Q3y&5iSF
zi`eN2=r2^X-!u2$0y{3;&Vd~K!*{xg^^0x!b36&jh3T>V-8~s8c(oLAKU4B^7{A)@
zmdS<wkJ(k4DER6AQ{DL)(Q#7xffAN`cOOw{S$G+>OB_+ZK=XfIh_6Jca*}k2f56|>
z<FCS#&11)>Za))vT55Jp_!yEugaw#S>oY4q=Bdnw1eOo7?30x@3=dDYG^|Oy>llui
z&hc-2p|qwxe);*AQFY~-Jz6rc&<Rf2=bkFm<GWkP<ihMb#iJmYBm``r%}qTuJ_ZFm
zLoWP2e(~^euvy9Q@mE~9uUmmpx(~45c63XT>(lam2^$sbzp#Gsij&ken`ySl_r5B(
z?Nalks4>3dc26rMCU~AT)*wdYO_jMLg$pXzDWUu8#}<#(+0L62=;IITR!@})h_uF%
zANGQzUnzVE=U3P5wb5x<zFfovU~bQyT={=pB^GlN_0n5oUYsOt&YgQ_CHVnELgAgW
z8fN+-==v2^zpJCp%EY;-!DNhmf3RUCp<DES96!f9e@{caj_1^H^zu|fqsFd-xUoD6
zD$f0*gB9tCN5jWg4{_!4enqwCymZG!+1<vu22Cl@#TSRH*pBBd`R`3jHM&XzSJ;k<
zPsf+)Q6QXFEZy*a2)MPp-2~E?nG5Y_>+shHbIeEX2JUx&v>#68%*I>HwoxckD9#7T
zp~>J0Ih9rK#!vVid<HsnthBK0m-_mO+o7zW{7a|YeqMLBu&0_z^^Twgo>*L*kfbsC
z_+S3GgzLm~P?ha=TW0B-i-N|x`3zFcu}1ylh({q{Kii|Am;0wKx4;g%vj=_MaMyIl
z*!DTX2A7`b{o{`hN&1Xk!$C}gl<Z&dhX=nkDzieMjZBix?K^`EXmU7OSyd4J1M*c#
z{>Rya2G<Rx;j(~}bw%o-9HhNPs-j}H4xRFre{meqLF~sdI8bDP&_(&<(}0KQZKHHm
zL70kkRq)tNv7uZMu<n)-H&C!@$BMTTf-RVlpQ+3FSd0!PB~TulZ{hpRj%nf2&;=dL
zyg&>&ynZ}H6fzm|o~eEhqGP%i6~3abS`0M;3DV(G;0a0)+Y<+qRw~&0KLJx^E8{AS
z1=pjzG^V`z6MKe*u1UvvK?jWD@ik7`jb2Dk9i~)tJ>&*BIQSwQA6uki&0LqGnWX+3
z6e7}}1f4KB#|e)uNNTGrFD^az_UVz+;4awP+Z%e%6!{&(4&JbWM1GG_z<ICF3T7oq
zdC!9%Plz`*AJ_BBaqp?-a1B`xD7dXeR4pMvi{|(;Ov=Z*cq(Vl{;LpWh3a2y`Q|=m
zpFFbQ(6<d3i<$6kF)^g2J>DDg_3_W#LbvfP#E)ri`na+%@5cEQ_v8$*#^p<BOvxK~
zrB3Xh1WeRjZTBONZBzzpoJN`llbQyPwPY`M_r#au^Xtl<ytnpmiHIeV*IS2E+2%^V
z*KD+*im7yLBeHgIw5<{e5@iS}Vb#okN*MY!4fYmzid?J>&atPr-4@uYnC{rA@D_{O
zu!h&XL<;vWwRn}ITVi64=PwV#4EJ;Lsy>r+oaW?}^61xm)r`kpHfa}ITU|Sk4SwKr
zTq7|z_1xz~z{Nn-67bW)a4bIsC&Usl+p|iiIOoq2qI9s1BM`zZy5Q)nIiAK6bat?&
zB?J5)n$G)?%J=`{$38|JRF3RD^0tz_GfGC3>`jurw>w)LDug&FvvA1fSP92Ik{lt&
z%&|{)_V@n$@cj>r!*yM+*Yo*!c%rMD{bLMr6T9}_w<rAa@LK5X?!p`NxZI004Z2d@
z4#u$lf5kruINZOHzR%R(^;6g8{a8qxUHo&|@Tisu`+0jNMgr0wPu4wPUZ@;qTF3Yj
zR+4E9^TAeDY^7vV5j%*}yi@ak{|IS965^6Zib?I=gn{Ml;6gx6A!>JPmp|<i90OSt
z<f9GM&i_rIy^1BZS6-Ho1`>|)ffFIX?`m=?p?J`zZZuZC&sg>raht32c<vb1twg(0
zHHmGwkF`=!QsE?*rf-=XMVmk9mUoSzEjB1rtjuIe_DwUG>wA!1_P|sw&hMREy|x3Q
zqp^#H`dNd{9-vNM@K2DIw$&aVbGQ0AdL3_@>AXTseR(n!bzE2tHF$h=cJ-HK_>W($
z`3PZO)q?>xTkQ^tssa?Yrg`%A>A%J=Ab6ZKlR!RdNXXxvHX_y){Cug|h&@~R&*>}c
zGCRlH#wLc7)pQ0*Uzp#=H`MHiI_^xo66^iyI2vapprdhpK$0%qJb}@SMxG(#i8dpN
zIJg6y>i}sCARh`0P?fo{vUJ=_tEFUo8#a|c8c3>zJ=|66XcIh)&c%n4(~!kOaJQT+
zG2FxHR?S}cPJef4*3i98WjnE(_0wWw{_Y&of^mBVh}u9K_c`9B@o`2{DJ=}^J#}uP
z#}v?r+fW`WR1JTD9HpF2WlPjNXlC^hdE@kYLh`dSOv~4!Be<Zw9E+~=GShAquhh03
z$=a+>3AK<zN#!3V#XRVH7Y6b<ZjvvmMOtcm`^j;>R@z21YQXgcGb$<)dwjE=p(jj4
zob+3<&6te)6vF=$$HE!y;R)(ZVpGWdHXxWtX%gn18y2b1JW{e^@R;noumPf0{2{_W
zpDImQp)&b*z^G)3k&)q0QA^}94(D8P-4)T$Xt}u_aJsc#uddV6Z7<Dgi!A)44sN?8
zIZA|kgp)W`!`g$t+=K4#hu(Bq?*4>wFK<-;U4Wn(#C2cFn<~=h$a%uatu-1)fW)5e
z^+6}Bb`Ao|;-WYszOlOEA<Ihd)CXK`cAcTRz%bVSbT&RB@^bnf>@by2a>nOpbn+&e
z)c}P3?8`s%H1!p)qVPsEKK=Av={{vKdi%PoAAj<snBn7L7fHB2`ajnV^uRK|a!a5b
zw5VOU!qLG~0Y29rOuXFPr7Z=Xz$zlP3ssdZ-W_U3jW`8gjYe8o`)JL*D7GR7W;q*R
zz2-VE7AEETV@X0gyMj~de=7X@S8oIv4f(CpbD<&oJUM>-#ik-@3c{ZUe#e-bWWljh
zi!$jO9F^IU`@Z{`>WRtCUo0<px3c3lxkp^2I(KM~ZeR47wos8ucGjK)okg8|5I*(!
zwuz&(nmEokHpvyw`QB`Bl&&#7-1x{JEbzHB+j``-R??m7%uLO!&EG+{*bm$9uCh9G
zHHWx);UxUqU8BgDPVm@%*`pq=3;9=+*l^pau8Sk3%E_gb=@gj<_gpe$&NimH!U6&o
z@<FPTnKwHpw`TcWl3rg@08Y|PjJ!DJj6C{hHGh(IXC^kVdXBT#g1su+U)%g9g<>D0
zPV5M?AFGUufQx;q_pBjQTtLoJ=%Y*;&Vbpucp7b4Rf%28&-bLnD|u`P8l<PHeIkOp
zeq~_+rN_L4F;0~zJ~k-~>dFz(N#&7Ow%)MwAL*&>FSt|HHtIKE8#Ok?b;8uV&SJ!%
z9!baNgoy46KmxDaLi|wc8`alP&?3p9A2LrheFa*yR-n-<Y-;R+DpH=W6zf%T_axo^
zQ&CZ!!%yexCpdf>2rpK0CXI#-Sh=fb@bd`xXlbP(zu(ca*Z0Bv!q9hHK3oZB{g3yM
z-}nZ3#`-_(oYEOU75?Csndu}@7iBcTaNgmMS0Ic=sGd!aMcpEQAwSSmmN6Kj%uY}4
z3K_8?Ptf;a6}}OVv08XOFx{zP!u}tRul-9sUcx7SUu=vz2FuuhbE`-x8wD{A0n-~r
zUMjpa*!cTJ{K>_VbW!;u!4CRU4KMWvxcxZ2@hyzn8ZaOZ`A}TmNk1>`4zAJW?JU2c
z)n-$<RZ1UkDzlURM3x3@YkON%<Yv9a<98=v8d|>Wh57l1BF5O0x~h_NMD0r{YI~H+
zLbFh@d|LgYJYCPBpNKr!9=aZJFa4R#l(zA~WX+{vr+EohO{pM%Y<RZW3;L8dnA>zV
z5ClZTY5*TNwzfUAjW;}+o8@?uZ_ib~c)nRz=?hO1<J8iGy^%`ziprjJeO-WPrZ40y
zTnxK7wcp@Cj^gM}ZU=j>p-+m{PM3)t$FWz$Hm={wzZaQq?*9oucUayu(No9w=S;e=
z-d{BT*&37+SNSKRF0TQDS(ohnewNk#3hbD33Id<$mLgh{`40ar_Di!;_^B6oQ}qDH
zz}5QY<-je@1S!bKGuq-;?F+mWoAt35i)?nSW|Kla<Ja6nm?-vBEV7)D{!8cY<mP>E
z7RmefVpm_Y%~cI4v(ay&4?bZgipN7*fPgVpT7>1?TD7W@if`r}Mrm?&9N|dhh9pLI
zA9%3#-zRui11En>XKb6ywb_Gpv&@s!roZV3-^q46;>Fon;q#u~T3VF9>W@#EqT=iQ
zzu|WiYE#doj0<&0*|_{>N-`kog_7dG9?CUuE~pZaB%8D5lmjHJlElGd(hmk(j2f?B
z9G?D1n5Lw3q<bY|sKv5>vgn}9np#-_R=`c+otK!+<;A~3U=t&MHu$nl?IhbG>(-oa
zr*Ru3dPkAiJ~^XA*YnTre4OjrkAD)E8xkA10E6HVfMD+$c0Qs<NmpBiHqM=NVG|$;
zO`&U8C;bTf>L#aOS2Up^EH6s@=PDtdtV4Sp`HJl<yn{84&#&OOix#IVdzBm}0N^ZP
zT8Q+G!lHU!P6RHu0q4K8-W`6WB3w+j?j@f5&f4+sA67vIuW^t`$kITU)I0e8JiDf@
z30d}VMFbYyeos%8VT>#F<&I<0^rMIRurj-Gd}zu7`S?$?V=Mfdy+LMwv+lAnQfuH5
zo5291yooot>q`inkN6&DorL!XUM2ApF79rn9^*-2PNWLb7=AtX&@2J4q3GSdJF_@W
z5?V-b?^fVF3c2fw^#9m~J<KqrG&PO-H*A-Aa@uHS7rA^mWgG6JX2={3p=-6bPEs8=
z)*&(b_QsRrV=&%gPbYjcDa?xVNesEG$&fIqGs?f}h1jFXrcJa6o6TK7QR1N|;)!Tl
zhJ0F&q)(U5v#R?OCDm+%#wJyR$6sW}_IJ(w=nW>EEpkg<G}=;-6zYQyO!@Rp<@wp!
z+22oEC~Sj0L5cZwZR}cs2fq9>iF$G1^?rR3Cxky<$$dta<{lO@@F1ETyAc%ZS}Ef0
z9+=yQRq)i$EyKqK?)(GJw+_zP+vR3bF#*1{OQJ)=vZ;T%FLNSjQO3!9E%G$1Gu-z1
z!K4?GDWux$sa}=cKEKzn;S5j%&&DQK!X4HmtlRwEWsi~|W^SvW+uWm|weUi&>)>4l
z>Y1z2+lcad<uNi=(l6uBf68oj+c$jj!<H;~$mqX64mI5o{hSI|p;Gucv_FbjKmPd#
ze<aR#Q}QnBwj4SIL&dF~^Q_uwq{j?6W4l@%yPUGB!FK_~M%4D_>$}s_eGks7CS5F6
zG^IQiy2mf7=`NSXuDraDeZtWa=JL6QLP}KFZ(ANG2kW~pT3f=8vB6!(+rm%XZV5do
zY1{D#dP}cM+A{;>l?CZ2VIACw_<-RXp_kwZ_9UDTQKKY86r;isok(C`?MCNewJZ<5
z3TRMNAMTl%a*9`Wb?J@nc%Nk-APV;Bvp>MBReEk{toeU9nbcg8h@RwBjd_>@jS6J)
z;Sk&bzTl+QF$&hzGFkyPV+?twRADbT>r(YPgAvI^Pyn(s-!-Pd;y}7!P;M};hq<5E
zvz_m!dbLY65Bzy|=5}lBY32CyUU@S;g;h(>7;b+Y*!c``4L_Z~YQ9QUzsd$p^`uo(
zml>~BXXj@8o>C#&aA8I<zBVjwMe6d2nb*EanrK^0>SWOECn*2$=-9|jv!W|$0f?Ok
z_Nf}(K(S=irXrM7M8MCmASq+TnmZUf`n0t1a^>IP4r=zVJDn4&Q*8GKPb*YoX1e#1
zpB!OrXmTeE)kXiE#G|2MMJ%NMrPf#%{~Cqq8b`=jd!p*fFU5)X)e~4!W(U}_m<Qqt
z%cO6-kLa+E%UP-V2J4dsp)!7~+gx~dqoAfTpv`I`UnVq7ekz|aMoRpfWI!5>HRKE4
zFD8nbID3PKoLAAEV{`iAVjxkTK&K%2=wq%&DiPn+)_7MWdFf==H@(avE5Sjyl;0T<
z?N143d!4`l)<n;?gVEJzM__EKzT7BkZ<!0<8M?fa58jV||C%e1*1_D`uHu<m3o-?&
zjU)^eVCnb#V*I(sTA}6@G3p6Y88lV*jMDcJ^|;3l`fgR(8Zs|As-bWmY0MKCrQQR9
znw{^!5)3P2U=uL&`Fp&>kDqtnoNbNwjQiBJwY_$EFM$mX3ZBxQ7H+&-@BxZk&n=^l
z{m@l8<-)4FyK{83-2ZwQothzLC=CIi#U-Az9$$y+wt^!Z;5<we(Z{xWA{6dQmD>@(
z@mGa<VEOOXK|_TV@!}GjM3O=}hy+WSc@a0_7mEU3W3%COgVW>OhWuID22J1`7A^i`
zQ0Adl3ArVQLrJN7O5n|HO8tk}VMl@3FPpcF%zqv}4<8FLmD$ESzkaCuhzbxrxnzg}
zGZwowmy=h2-_4((QY?HUB7%Yxy=P8OQYO7S{Fi;sFoWx1Uz*3>-^Q_!e+(4mq<iYJ
z^OC~&lM*vI=w>(UZ|fBqK3eu!3t&mZ#OZ}_K5QKhiVh5|h5j6fO%iPL1`Zjb_VDwF
z`p>=|#xgbSyJ)ATk*TiMMqt0bumgztp)D%1da`BF4E)95v8f)pJPai?fo7?g;))kw
zruOuWf~)?g>k!fI@(hkWk0#Ogk9@QKxN9QV<{-Z$S=%o*Hs`Naqz*BzNds>nP-T+Z
z0<z%Gk{c;)v-+QG>7CBl+OF)}x8Es^f`q=aNUpY<Oy9&GYgC6_0h<D3)wRCW4i^%I
zFZSPD?i>^b-;N!Qe3{7vt65W=nrFBe`UYYDes=B6MB8+n1K4kfrD8z(_X`P?@1NMe
zv2+T;JSNU{cdOR#fa&<l!b|U~FlpAVL!6acF1|hTc3Ot;gIMk-(iNMl)BF9KlUv?b
zt7?%aa}txoC+^KefSs%B-&VE}Yk347-`O~puC7ff?>kr?+fnwJ{JZ@d)YH(+XVT8=
z&$}u_UDyAu`_9R(EMQ0TG8W~jc`L0LX|)d2v1r={hXzJXNenez5k%89(sn(iK4Z>G
z4t~{08Qh@^6L0&6XQr&aaa!qrpzy`Zzb-5_&wqh~Z%nQpIPRQbJ~E_$tPEO|D_Kq1
zJPL9tZA+^nkkcshEp$%F`5LMxyz$3UFi<eoVuLVC&+FRi0r@$VUvo<AsfBUf78Xst
zukMtU{BX7z%XR$<Q@FPWk)XQ6Hk-iNQ~a>)4$;HAPq+kbgqimFgwmiCedNo2%>LYn
zyY#udKbGtfkmXX-d8(J*OZU7M*=%rlFr6{!ce%FcYLUzRPd!=Lu1j8_I=9X%g`0`}
zz8_tt{<66qt2TQnUAjDi2}5B%45L^UO}n+(>u9m*^_DPm%B`PC!^+}}+IL1Ua$E_?
znwuqtEFYQ~jj?RDJB#^1jN!rpvE09BX0l9VJekV@GN+3jyoRZ%r|W$sA_t+WD|o1m
z+bGZ`+hy9HvjOK~s_oEXS<`^q|D*UIy~(p!n3or#rq3(#XQEu9sR@J0kx3_nsJ_=P
zP!2`*zJifLpH}`s<u#+xl#a!NEw{6Dreei9LV`A0TLSk7%x6QcmbP+=sWY`c$WX_q
zTe<^|0w9@dMM=rZxP+g~bA&q;#mC^-VcUdB!&nh26%D8YC-rxBAW4JI&nM?kPAfiL
z9jl!+$1eWcv^$HfZd@Lu7jyme4YXV2(Fc(Yjv3l+?QE^aPjh+&+qkrI=wo{2qeW8`
z*|xU0=uF`>S_w}z9MMBXCK|sc^WTVm`vsUI?nylH;4oqZE>IF&+a6bkc2|MMrrOi;
zt^=&x<|Ag!UdlOR;&AuNWgtZDx)taDl&UCaer-sc226}_5}-)(0WDrH6M57CQeF)v
z{B|Us5wONcLj~I8ItscU9?t`3f4a{##Ckx?VsY0}{fgtZZmi(XR1s%G+YA1Hz`t~#
zR~JpYick0VUSIigBsMmA->9MKC*&Sfx*s#iJ7zo08r-AL{#a9}R9#?je_A&%*k%g2
z3Hg=eIQ=I->Vz9IWBt7VCS@=IUc(|K;Ss0n<r22-_K4DAhsUNSgC~BM_+zzy-3(n4
zZ{!T}*UN<WQ%*e4^%vm(Js1Yq4N>N?_wGCta2k;ScI*F~_cs2yI^!zT6-}a0ZSf5t
zV7y=z{ja$S3;x?c8^0Z@`*?ggzR$d>vaFmO0V5Y0yXjJJb)IcB<*ou3Dqj4XI)!hr
zOpH#a8*med3&g;C$g};5<98Tkov!&UsVAqHQ{J|j43%ojA<nZ#IcNi4AL6PSo9w=k
zRI?AM&JXm|yqJGp?7O*HMn;t5)Xg)zRTUgr1P%4kLi>AezHt}Km8@VHR!p|!AA`wD
zV-=*k<9U0_j`W}+6ujQE)x!Fdza(#!Ov|t$f`b^Ent~=)LM;MTNaU9hBw^mEd_wp7
z6qu0~GcDC&gj4zOjsJjmNvC5e3;9w8kIhU?r&1en-cKHoLrjKbMl(%tX#C5O95i_L
zw~0NRk37zA?mEB;6-J!@1-ahJQ2|k5r(4sYe@=hOP>`>CIwfPdsZ!|<yZ1GL9?3Ux
zz9USW!Z&msa&_Ep#=Sj`!}wsyKDWSpi{C%ltc&pVLuLoU1OW$_uXsoB#rNi}fv&#|
z$>Ywnp>Ik-tRimarU7$&`kks|<(&JjD6$2UK+Mh#7rF5YyHBmf-|S%*8@b9q6a-}`
zsgu6xJMyyanTh@SPIops$|T{^t(pREsIq$HA-=m>>Y2=Y^9gBjlz(*k!nOS;@7|g4
zFg9Z7mW+#sQ-D8y?&&w`myN~z$pYt^-jO~2v%hae^%%=dKj|pTR{A#6d=ndz``?~?
z8Y|_x4>#+iCqbnurO7LGJjy~msl@;E{9m7klB#N<)aAPO)eh1B;F#ETxYf7oTBzc=
zF_SB+$GiL>JW$q7Uw7Hhx3(=@2pbgZ@X5?oy8}NY%?LQ7eDi}GaxG}V`Au9fH0E$s
z>+3sRp>LA*6>jk~l=|eAfrB~nzq%$<b^m*X7QI}WznCSwR=Y-emt7w>9GHF$WP5B$
zT<sC~EyPykysbWeXZs`OyOUQD)O2=Y1a_<BIZ}n(a!cGSO%1LK_V7$uvMq0*$q?^{
zV7^UN%K&7kE)&727P-{nG?kNv^(3uRIk83I3mb2vCJrqD(?4>}qjwz-O_#ewE~eyV
zLs*$&lD7OlBqrf3MgEx5UbQNcW~cJ=fNxo;k*Nc8%r2+Ldh;u@z9d@PyNzy*%|{!9
zDzLUK;{^doJWdh9&49iGDxs7MkcEZ$1qCZWXmuukk>XhE0_po@j{=D${$~Tw$VTgg
zXu2v}*?0bz!CgnG-shKcBvKE4lvOu3>t3zTUxjYZ$iQrD;$y;{;z%L>0S&=^qK{->
zzIe~Zu!qT)OYG6GngoZMwuJ@SCNVv8a~JdHscf1e{G>j;79XyM5Ie~D$NOJMlk?64
zR;OQ7O=I0XQJ*6TRlCvY=!uhv(2)HpduuXDEqA)m!cs7aV~|b`{+zZRL-PGEl-uoW
zY-0ko=+#C3AnP$l&)cOst7p^)g`uS|ZHrqI8`$Fm_17bow$1itGc7LH)!vWg@tf$d
zLF^H%%OrC5NW01J(Z9r5&b@Ei#hw8B8S&#{(7N>iZ^;=&)qJHy<gqx{ea^w~;zk&P
zKNu52Wc;cmiAz65N@2cd_~UM&Udg{C=sgWQ{5#&`9Z8t2Y2O1~8UwgW(>WQ9>*nsm
zb)Xe-8R4I>hP<o8Y@E&yzGF{2!h>g;TIE$Dk5&e!*WF+P!gRRJKgalTpkjoGZ@cIR
z&J&}VAe_-nm<(=c{2NhyxD!5fg7rG73gfah<)BYPN3&e6p&imL$21yUk0E*4d?A?U
zf5}h+xgd*=@(&KJBq!!7nrKYP&`nRT@18sUFDJ;I47Qo!2}oL^2(-8w^t$Hz2hNGz
z9IhKUE1}S8Z)tgs26BB3R}$Ahhz?g*vO1Xz$*1irzb<-<L(x33FooK)3p{xGBsEhS
zru^_u8o01c=rOj)8Y}4dHwrdIjMVDmTnYQabo{|?Ol#GH`+*gIv7^QLzE7^ZV<-K<
zr_M9jYYwNtTI*^*UiR0N+H(f|WQ)H}Jl>hDPN%dYCcI3+Y)|Y$aBx$-Sh(wzkL;M*
z`QM(RpX<(w?{}Cq;1xwyLJrc?xd{UW?IBXE<re9Gw0lg$l<l4ka&z0s$~v3lIvYz=
z==zy?$eq5%L9~np-a+hpIkI_9(jM_Yz*#WInYi0bUE5n&-KJtGAH#<%kIGQbhZ8$u
zFN(p2?&#_!V8+?r_jkO0|HjGc%7NSLU4sc;f4cSE<^Ih+$JQJJOza|m_|f@CDNHz=
zUWh^MY*5MWatMd3u0=}O%AwLlpZw0_eRZ<5RpmuXxgbYoegOGpW=mW!6G=A<z*Oeq
z`Pd}bw%z+1V=fsWYgI-8vto1-dApUqT-Ib4HMToIa@zEMyu$oK6GBQ?!hsxEpPC_R
zF3Ee%KDFA+r{HDibz9C^9X#gjH;eDA*+cq@{|iwzocO6QC1cYCJ**Lkf1{`E`_qE?
z??C96l?Ex~$4Nn=Vz&Z!6rrOl@@$wGv0Z6*XC_STbloWGe5cJ$A?g%N19m-vzBCkF
zRq+9Mbc<_uB5d0~wNbw3cjZ!3Y1$E)HuSy!BQBu1x!Hez?Iqs?c;4T`Z{Tk6gxw<z
zI{A;$T8H8QsiFRa*YP=>8>^q=3*(<D^~}|WEKkXak;c~zCLmvZ<bTi3KAiYz*=8*s
z-lt^qjol|dG%yq$ZrgmkLaqF9|GAhi<x^fhU`KhsmztbM2tw+{9-P3uY?4wJQ^ZgB
z@Y$fSe7qs|^bwge+oQMN!;SD6e8k8JakJ3=7VgXk()#Jq!Cza|oe>t;QyY5jm;N|&
z1FZ|4?-mcWkPbCJ?m1&yqfVR=Y|Dm**v<<SQ8{vt{{iR-Avc?WcMCXmEZZ_rNoh#}
z7_plv_4N}n4<IjjpjfKdZtRX1oYE~<SY&m7x_+6Z{-p1@GKWz!Szkvz_7*E<phbXu
z8PYD*BzBv?ey@S_r4nhT23*MPdb!CB^0jb{!seD;$BV`Vd6KVX#k&DAvIcWasN^MA
zp?+=K*E(dp5$6YkbKy1k)^I?zkgE_tf>^FQ>pk*EkU*QUn<;f-65`-1{HG-?PgP)k
zEy*%+_|FRtZN^X{GNtIgyUMLp@Hr+p;Wo=!`VC#ZU-V*D_R0ClKH*I1am{Ra@XkPS
z5AGjmV_BvqiL*`~G_<XP)QaUvdS2Tt6Q%EOo6SI+y#N49S!_6$Zf*UeP{6W!V=+Vt
zwzL#cK+oh%T6KqV6XR|$yzvb##BVin^LRcg)>n<4-x#^Wfj{IfECByR*CtM0s16q8
z$pw{V{Jd44b2ooS=AW@5Wg2Fpe&Uakcjq?^OV?Bv&XDpp^tiF!fHL@9D1GH-<t?u8
z`cLXLUCvh3WAIiX`4pEI0a(yc#wt!JwY7n=$Xz?oRhD>lf{p65y@T=k-7?;!Xon(C
zoNj$z_#enr3qQ6Cznx5}ZIOAB%OfS3bbfx`AV$lx;$EYY^o=fpB_Vbqqkiy6d^rlP
zi#LDFQcRv4G5PZjOeM55nD1*r>wWAoBCB%A9kFf@73NP1Y`5?N&#d1a-iBw3Cw_|D
zCSI*Feez|ydDQ6Op~kv-6&g9c#ufPLzGRp2ly|OPi%jbZ!VY*+I91Vjq`p(C&sdHB
z$TxMBWwTyl{hBWWX6<fsT^mLYBD*&46(@n`kshhz^#sU*iW@3AlxjqwzePQb-MBg5
z&B|LJ{t5dXi`@abF6}N0)y^;GH!CVe37#EVlrg^EscxIa`KaQa3nMPoi@!au-#+-3
zz!&E2;U04zuRPP{8={VLLy_@-IgQ5-zXCaU2}nDEY+lmkyc+?_#2sol311nk-Hfs4
z_rEgheATN!nv1?Z?LuQl?Qv)69oKpdn05Y8c2W@0C;`aciX>l~n?!gvZi`uH65a=W
zrDIG#9LeAoJkyEE*%EX5s(zoLcK>5jvYSQlZ#a=7(6jK{3M9;NP>_E`K)b%K#j8k?
zcd2JQKpJ0Nfwn8y{o!QpgMkemW1p%2x~JE9thy<eg^+R&-zXvIL(~gc)K15Zzy)EZ
zi$$*U8E}?}d<enO(!!krsLd_h!B0P^cxBh+@oaCVh6GGWiNCpH1Q|%@w_<5Me3BU`
zBp*6!6(vv1-!~ygQWGNPN&I<8ORK0*jY2fvgp<>`5xHf5I^N0?)!4rPkgoGYnhR&i
z6~x0qgQ-cENrn15bGI05Oq(Taky{?A8)yrF;KD96U~c?W6p=M-4c$=zqj^nkxx7m+
zIz&^_X?fzShUsBH#H2C8rC)QD$I7j&u&t8GFYk*?;sOgZh0`Sw#k)S&`-aSw?5asc
z<Xw$3Tk5Jebskg)J}eL@7PGtI-5&M#@A`u(XZVmhuMj82i`JajTAf+pq@lW@{TK4U
zTTL48-+g_KjYDrkEH-gpE{fU4Y?bZaorEZ_VDZYovWK)C{?&6`?S-qIhJN@lzT=Sl
z?Dm|6ch7V9{gY`SE-xdmTeqL>PvVbkIFnqB?6lkxcq?WD*>&Fq@#+iiuAS_P_||sS
z-)=d8oMu|Q{QoQg^5N!3_Zj>WFc?CG6=9_vRFUFo-O|buzy!p0`$!_PhyRSHDlw>C
z4p>e#y)*d{5ab#h<`0~2Px@J`_4%C~OuPD?znIKKbc3~W*np0^R!~cqUDo9EwEpLf
z2-{M=t5Y^GW2`V#WNJ(q6aMLIPkmz7^#x?$0NnK#OAYQhBmh6G`j-gQDYeo3iW@aZ
z2=K(wbSdVy>BT#l*;2IdOmg!s55)gHlK!w!M`ET|JN%)0la^Y#9rw+<AdBZ7so+_-
zfsaVoS#G5^6mptV&zVHQyM@}87!TDt@QT$4jVpLM2${yH^^$9Z{Mn63digc}k>1Dj
z7=KLb66xs!31-jPr|1ko?}13M`dfdh@H6ku%V)wjmcNBf$&7-8=tQfD!L!PC+Zs)6
z%q^qe;We{mtIBlCc}f{<vt2f_p5>{HY3&b}B-8E*#82LW)0UF*+S3_(Je6Naz72tt
z7^Ef7Z+!}zPM1oc2A?~8+)M1b#U;YIiCmZGw6|mYkN*9OKNg1qV5sgW^by{w5;ciG
z?g5XGeFx-mZ_csp!EnG-qqeCSu5$mx(Pk2LNKX-h6}yM6BGB{mFp}AmK5EL1ChM1!
zEZQ-*hz<JQ@SBx)FG)(8)W{6ZXcPANTY~Mw1r782KgKUfRyH^O_4#(%Di92&Sc|DU
zo{;FSExK^;Y2g6#1n#(`qF-mJtY2W?;Cq}U*EM~J+a9&ntD#vgZA#&Kea3J%&nh9(
z689&hdr?w_YO7UtS4e*?ak>}-=&hK$K^>Ohwf1~zsolNJ?)<!1a*er^8>;D4b@%Cz
zx?cQ1N3e?6P0EUGgn#uW9YIsC4@{7LSh1E2asx@uO8{W}k-+&?X|sJtAC9^IOo9z*
zH`2rJV6qxo%4Wl&Et4NY;4RHX8!U;9rJzA?rUlEU#x17rJ~EETny0rDVt9_QkD##f
z2eVD69mQf+r{cv&yc3prhKZUn<&$U-&~nxm9%{_R!O6>D>6FfY7X#&~tWZ|XHxQb}
z1>JlA2#`11Y#X%QWlhvD$NhJr*Vla%a)0I;CD~GyaMtYXJ!j~Mt(ei<Z~X+JGhMlF
z!kgCb_{U$=A~HxAJzrCi1T^_$Cqlxn@OjObt5)-aTq?6ObMHiVwLA?dW@4>sz}(x?
zWcYD`YN+3pS3Z4NFmVpJ>LAuM2|hr=`qAM0={*V;n4^>M@>SBCW1((^?B09sx3+2L
zZVi?)(WuPOf4@Vok*-yL7^Zwn*2e0|+?J!Usi*Bvxil|J*Ie{qLM#ZtaT?~HyQLbr
z`-J-GWGmlh$L&hR1;-9?kYw0l4SKllu<o?W+aXytDSO+0eoi(_<Kp7j4zG-l-p>#M
z9>A(oZjh;Wzy0_|1Frq7F$hB`%1<EqT=PX`qRfc7jsDw#+^neth)-gNv}<8ta$crB
z@68w4rC%lW!YHh1K7P70SC-`8fIsY;3gGxIpb>)%PD_e2>9Dy;_8K~nlWr|6kO_8h
zY9|ulZzz0(<y{?_2YTQ%wAhRmEw)o&YV{X$%XsK(#9RN!c0Y+{T3X`d%)iOp2<k~u
zB74}td|AwW5mbHToq_OW8}Gp)Xz^mmd1_6h6F^jtXIw7YI#^u1q?*5YYURZjxa`30
zv!&v5M*MC$D`Nxal~*(5+vGn)>tT?M_SbYD=~=on%2npOD!+0hSbHl{CS3q8Ug%GS
zKMg=<nee*OJ(4WV@xe4XlZp7-O$HwRFDZzu&-Sjd6<alU%%Uw-IF?fMeD<l{$)2~{
zom*X+eh6^vh0M>t+76!28zYPrDCdu6$Ly+$mX@sRmm8F$1SHegx%F-cQk&I2lFDFA
zhUIL<jt?lfBZAY|4?Sy+vA{94k;+{VV<j(5Y?GvOhWYU{g4{RXgo3ac`evL@f83I$
zB)9_3dlS(96FAtdMMyyouxt}fFJB77a_*{*(N-4r8~3*Qq?OQQrW;K7VBeXz<x*-u
zAU0Eto_U1Of*M&;@&qmZ(iMbD?4o0>*xu=S27wN&USmt&VBml|>dD7+l6zh#Mon@D
zSif(UfaZPo{|LMWKgR9X5O>S&Vys`B<{XDmXkI>!?{ur^3|-d<dceb@ufy!50{*;%
z|Dd&;^*)2I%Ro0Ks7Q~bN83zzP05DHHOtM(7cT$#td?Zj2O7&{4B9tZQuCzquPdb$
z2?v#1VRGct&PIU`PLKuAiME3anhH!YmE)uf7-PD3G*Ng?{_U68!-<>0-s~igY=b8*
zdI;rrEM&{>EbCj|Yw`>V3NA?xo$&HBYCpbs-x7X?JtJOeLGQxpVE+t}5qs0c@@zk6
zRKqUMQB=*b{2Xz+&k2&O99gKj*<7sa=Ju|0tEGUic@|*WYRfs-(I<Ze))f^L6~iKG
zJ8u|L6%Tfgk1zL+bsZ^PwTn&pjtta`vAHfYo_v^;HMqwk01;-*sV0Ltz?*IUj4Qe3
zM@L~<$)k^>>N*0?w@!9b2q`C3mF`+mzMa2vcl@v{DUJJY41yMlLiWz-AeUQL=W@i(
zecW?N55THJWiD{$rYWJrTX8hgsw0F^@cqObKA@s<V|Ri%+CUYM<)pv4pYjN#3l77e
zrXSu&m1Zs|rxq6pL&DyPveqx^9q}Ds5^QgVQ*<61$UvYx>K}Bue5eZ$JivPH7h?_H
zd*b9QCY0=;833pFT3FtwP6iQM=GGp#7M<t-yk_Bnz=Zz(WRvCb!L)D229~^uhMD+u
zzkA|P(y)tXbNW)TCD4XZ*c{~Ru&%8ZeSC!p3%bV0NXo(vV`Z`S6IYPt-h=eAik2$J
zG-4WTu9UGc^+6dkYwQP|^b757<GS_x%V=|{8kXTn5NhtXcfQ;68t7Dg$ZEca4KV@`
zhvDIKhl9G`j-{A&l)PD8XA}uY`zp^Fv0=qx3PQ*$Ri1RIe1AhdFVu2OI04n=Q9a1Z
z^u`q$6Uwd0jpa!rm#$))Y?=e#p1q@A@{|4tux){9X1YxGgIRJ^<Au()Wy*zL@($9y
z%}chh>#Cc|)RYj4fBMcI{Ni%}ADrbTf=~BDesi1#GU353?ZM|VoX*AktUg$U$9M}O
zay-8|?OZ2+Q}(a2=XW;l?$+z<NlW}Ddhx#c5KQhUj#9zT&zU-mh3^;tgzbBu5bu$9
zG*$f>a3^+%3Zs(=lw@ENtqr4V&0LA}CgGP~)~k?DPK?!H*N!&9P8R@6jkW}>yj_#>
zHu+@Ab$dznlEJSpY;H74p@jq`F%39NHmtLn<=XL0<Ll321Eg6?-6}F^Vtk|v<Tpu6
zheW5)bNr=;h)Dp6KdSpq+Z`6At+u#sNkB4aE#$vNn)lx81C?2zb}9RKNA=h5MEyhu
zsQX&Hzk!XO2&mnR0lv0k6>|0KQl=x64N;9TwGO%JT*dmf52lqI^+YoG-<mYx;7+#T
z+o^v%Y|T7}tLW~P8GMQvx0i&|-+EMDZlvS(#hRs8d-86^ZqVfOgqC2}v9`Ufh*lL=
zB_GUtmj3<^y*NozggQ6(TxTR<H^a4O@yll=r}eefQ6nyuW8k@ME1)*a#OPe#KGeoi
zs1hM<m$hpjvWeLVdBe%U!KoH5n^(B)k(yga7vtlu<SU$A+A$nD71+G<TS=t@@-gYN
z<bs5uxbjf`MX}g>CK}T5_nf?c-|CHx74-ACg13*0_!;-1)3(yjmArQTl<Bo#uH<QU
z>2^bNwRokR9RMtYD7y`G+P)ByPxrj-xXY|&)h_CIj*0=K6zUpVXO&vj933;5BBQPn
z)jC;qCpR4LDvka)z;E~H7M{0w5{CEtK~Y8|n1Q0FPm|AHMs54y`h4l=W}7JfYNg%T
z@r^CCzN=*B^rNztkU+|#HUEQ`mwE-E<b<x~YhM8xJbv9xO41Wy6&7*%JRJ3%pY%q&
zhOGYJ89!Z>EITV>!;GIu^ngXm0JXZFTZb;TG*bg6T-tJ7tG7g5iziNt0s_<E{^=-1
zmM{Q&^>Mdw`ESHdREWsWjdy8!QVIGr;vmTV#K&gpLYtF<2h!_{05@<L<B0t=oe7!O
zoT@<8rXCI4d3-S_h+w~0%y4PxbGcu5u~e>h;mc0hHfz;+J2PE9ixKVwwUF3xVejZ*
z|1V@dh+~<Zk%Gl~fS<M{(`Jg%Vr`%C)v4lUhqo03;-z3K{fOix9|0OejrkJ%&UJxx
zImucQiKxq{-<PO@LVg{WcWifq(9ohLPaxdgs{OjQ9T#Tl?x<GRje-61m4n~MJh73Q
zRH1)YJk{<)9An@F(MKkTgyM^i4z|QZFn7~&OWl1tpKK55$<O(dZ6ajvJKvVp%BzSE
zPvGR9vvD7Pct-k#A)P;43lG9WXcu0+y&nCB{R4}4@e|~JNzK$n|5l!t6CLSejZDDh
zePEd<8^Pe~4c^U$*6?HQhZ7YQVE`LOo+&|;y=0nul=KF<=%eT*WV5H0>13ITU(+W&
z*N)?X(p>!}a$Z7}m#9;NKhAPO#>Vunjuk=qYuP6q<6rAp`v!HISuQ3ZPr*LeyWCCG
z({d9fg}LRP|EDX>UNo84Nx#MsRRPd|&e^ihRfo|}vW%B60u*(eD<_koLdLvtKDgsO
z65g83AC~T(r7K82o)H4Ah*X*r!J7TEd|I`wgu&z4X>|4R(NUQ8;fsz1{Q@M?+1GCw
zH#$*I$oDlc-QM<KbMl&(kMNacU^psq*0vCT+INJtZ$!RMAe(5?Hr_(i0rJDp2CCan
zU~%h3q6xJe_r;x`1XWjWpiuvji^rK0Hd9>28*mDbt5armsldO^y@QBObk6^l%F_4m
zf$4D6qVwt6pg*c@H^fn&MdSQ*vnJ8_biF4r@co#$c#-oLQFPlemvY#-hj;KpXczEh
z?v@{xkA{5QJw1K__X}i-^5Ppf-kO@o6T8DppV&-Aw#pdLmmPVAeo=`yTH_-wuA*Y?
z&X-tv7XQ{Ka%E2rt$789Z=l%dmpV|W@MACC<j`4~&kYmogJJW}+bURkG@jnLO?h`Z
zB1kd(eDSP&OfAaCyJjS1|M5wi-Pz35Zu!;Jg_*62k{w;YuEAqK!4)-`bRW!#eJ?k#
zE7Lz9o9z5BAy8OX7$o>0CR<;Rmd)-S`he>Qf9%<{&|Y|x)oD*z4g<xQ!jjalB}}WG
zr3ObW4-?-6qBr@o(GzNyeG*)%p{jYKrw$)ryohDaoSTS;BJUTpR9fOJ^W@Y2SM8c~
zo_+^byS9&2cL4NAUZ;MR1_=d&%uF2{a(qpLFHKOBW<!JKJ=Fz${ADb!d>H&a%Y%=v
z-dr!yd?wHCn|p6UT@EU1P4Y=JbFkvz{S<q12WQYzCXSB|%bg{yf*`U1JL3*)-m&)4
zYi+#s@plehL04q(9i8d{stH#n7c@wcnym;4@HM6&&jkF%1xhoeZrzi_gb<vdFvqwr
zO`X0Mr-z$nHBU|dAg<w!be!l06qHKNR~8gp9b8PD)4@ib8C9mGQ-f6AyM>s(RpjF#
zK0)$uBBK^pgk`8>QgR9`CWVB|Ii`F%FqeZ#pF3<2+V@i-pZ@N%k+(9OWIWCR^gJt?
zTGYza_J9o?UONeX@6<mty_YR9iQ?)%oZ0#_l)--k&CzWu#%mo@mUdZ(!=*A2x5rep
z)CZ0_%HqIjOgAh&h3*>@GZ?h97S&>PsDsm)>AYL&eP3-cf?TKOLp<sh>Hf*eyvX~J
zg$HR%ChiK##7YrwAkn@z+ntb3G57uo%3qI!0yrwWOjgZtg;^PnQhgEAn=UP1FQQG7
zAyPwUhRk}p%eyEmxl}1#-NgBbeaV0Kso%z+^q*E%@UMu)o5&MlX^{?^!Vq&OghP%a
z1f~f}$FT|@d(CN~r`5rDu(JAYu|;%9>~SX35Y&Y~?u+%$a#vR&A=i|XAMhS{KfDBB
z()%p-IJA1GZlErPiH5PN9>Mr^Tz=v>az?!*{-pVOn7VHu+GOgkCe+~r-Gd>j<yQ<2
z(*fX~{}pzOMcdfExBYlt`Sh7&4qYN2Vs~|Qb?f}^Zt<Dh<9I1HdMZTn^=C0swiXtQ
z<i1!<Xu2-~Ozv5jUdsp>f|EhCZ8HJw8=7#21dv)OW9^VmU-<$I9*HNTtEc)~Q#7xi
zOdWGkd}Hi=CZ(kKIapTK|Fa_$aYH8<nY_H*(=L`RE9M#q3Zz7xRmD@1AFw)k%m#US
zZvTxKe}3<46$<rq@`GD$^L*RIdW|Yw<ni9^;KYf~xSejZV-J4o?~{xFwpM_(ojJ3L
z7YCCUi7tW_71r1>KKru!6F~+Y?5Pj?L`tLiL<Pz-rA4`yw4Qh8Y~VS#^1(YV-Ae}^
z)--eXQH7dKH1Vu&dZI?{O4)q>U~oCx7X6>|(G=H(qiUhVs?cX;U-l`%W&6hMGvn^O
zzCrv%;v~03#dw{U(DLH$x<25p<gS>c{;0QBE=-o4E>BxX8-7zXB-L2?*SuYR_0>l4
zn0#@2omq{o09^Ize7|VKVE$|kvbaEE`~Aj`R9L#c`#7yBPpQmDB+DKY$^xb2!C2n^
z8K<=?kO^}{Zl9@Yn36j{Yekc<&ecv{d)aj1#0iOKV{IqD{ZA&d?SSBX5x>KYO0`Y&
zC1`#NB!`(O61XwWub4HVQudlvjR=^bTfO-ySm$A5W1Fh6nNy31@(Hm=T~h@m9iRR!
zo%i}^KnR~-3nFYQ0P1h!3Tz<9{f@Z&py|nsdyS`;2Tiw#i;J<jdn?m*hwJTkF&0YV
zby-e`D*otuZ%B;+(ubao^8+x2=qh0t#$ZDDee7#t-ej1QScrE=y$u6|d;-EPoOr#D
z^KCs=eiYt_OI|5tdqBrX@+{|Ufrod(0weGA8+3cVluCg?on8?*83Q%$U?X~Q3g1H1
zy(E!Q*_sUVKXm$7PG31E!aD!>1`G!18J*w?UPCk$UcvxP8Z5N{<T`UV+!ZvMz8gwO
zr+M&xMU=&Y=Sf1$&c^$7kuq?ctY8$$q+axtQA^l91{O~)&HEA_`S+$1fn9JQltn`9
zY3y-UddXOzs=Oh*y!Sa7%;`R}U=pL|ebq>E9wt&B&B_-(7pK>VDozpJ#P`SDmJ;MP
zVjHj`TrL1hmZI*qa*tntj4OB}KIZx@NNH()G7%k1k-_e&K1cEB=&U<S-ZG7?6V7Bp
z+|l$+s@*)@eJ;&eFuI!3NtD9CRF)fr_v^&uU7_0K1twVG$6tplR??M-;xrwyndd<a
ze5a!m<-3E-h+Svt>EFNE83e(8D;B3pzD0LU7LZpNxIq{XY5z3pBjbnP;!-8(h4Pl=
zO|b+j6^?m4e_8kPZSBR-hP;^xwPDPEDKWHOh{ZTIr>8pa?xw+(0&C!4>L+l1)X+$$
zcswEs3(p&!nVdA`*H*O-{Q_7Eo0=GLk1!mGl1Ligp`H2Yp2rsKZPWi<e6@Urk^4oh
zBmiIbeK0P6oB-8%n#k9uP1K)@Uzpy&q<SB5%FTG8P-H({bPQh?h`Sz*7Ypfq-fZ@&
z^3rdY8>2V68OaUz+O~BYOY`<XHK&}`TyETplOhlI-{kmc&Og1lz7)_&_n}ZVG;De3
zhIhd1$l!sV;2m{N2YqU?ujAodrh>XxOTg8stHgXOIx^B7t*WgmSuulm#8RXP6&B=k
zauD$?=%&V<i09cybHSj&ubaSixyNFFVqn0wd<8Gu_@T8@!dRj2TLs+;$J2vx{JYyy
zVRJH`-phkOI!|XfwCAoa)vk;(QfDU`IlyT&HUJwTcwd-=BH`j4YJKYXI12@wt<Pn9
z|8A+Z7B5r-o_QDvFKLAlt!5t&m1epT3WAoh<vpN_Oupziul7S6x;Oc+>RX@mM0)##
zNV>>*HlqATYO8#GK_x4z*PP1MZ1mtoOZ6=8b)jnH%?S3ir{KQ8S#>siZ@?083O+u$
zj9QwM)D-uLrR6eLwz0Zh`2vuK>zhqUmnQT$%r+ZLNa~D@VbNuWc&m&5JPEXs5>8@^
z_mOoggo(*(H-5k1D}~by&&qkvbp}?V@*2SmZf}ID(b?QWq-mC6-WnN?)=lmC<J-5F
zGm2G6KYeDqD_$}J({vI{oD@dGAaLf)JJ^RK&*`K>i0wVl&|9(o7~t1}ipH-*Sdmbi
zbx9$4N$Ng{qZW2nEaT~HuYFl2lO{9Yr}55uv|9jEekhUJ3Cc|9xQ02>fs(>tZ@%9G
z)j0Da4#~zeP=STCAH6Ak)VW6QE;K(Kq5>D0@KL_^;Cv!2OUTE9tYal$lIXGMaGkP5
zMfam;6gQ+DUs$g-0B*CmU$(*Gn&1EJmBQ_lVLCaydN4tV%v)R14<#quP{tU|g{Q|X
zjhyi?vBenJt)hKAH5Rk{is?nGM^)ehfj1^ov9BcpLXu3f2q&kfl%>7lSL@}4Sb{(5
zn$ExBa66axChzrgjQ(O3Sj5*_#ZO2b;yMn6dG!>WkiLd|Kq0~m?y3pxZsF(8t{b4m
zx+Qfhu<_JYh^F|V(a#W4ph`CXE+Ykad_y&=vQw%(<(Uyuac}lzhy5Wf@ZgB-UyKm6
zluZwuOw(poHGJ<Ni6egDNmf}viSQoQ-brP#BzWKqR+SM$_DjQbgykVNw17Z$b_RFC
zo~(}#`T&nYh*Ra2G2_Y^kyA{Xj0^Dg1?jQ<`g#kfgUqc`FW;&V(^a=)2ShhN5lzq6
zfJV>!44i9(Xi4?fpC7FChB~u^MS4a)an|i0^P|Xkx9eByG21lak1;>!ZnH;?mE=gc
z=H6j%Y*s0D0fRbW?U8|m-DN_?cqZ)mh6O4c2(}H~PZ?89fllqa*o>Ja6n`Tv`VCjH
z!ZtVIx;n28iJe#V9j;^V7OeHK)Aq_7)k)!Y^Yg&b+QQ$H)c~2o!tB^rIGdz|$0QJJ
zsGd~m^*7Dfj(zMHSmnMfJpWhj-Lc=%X7SdE7v9~`loF;R+}hN3whvr@SeU4Amq1{-
zM)h9?*X2Todz&0J^VPQ61y@k8cZ4sovK=1ec{Ws8V>gX|k!$H53u9+zS9FyN3Zlmr
zT%28X6P+|KPh-1~`cwUtl@?jAr}`JN$0w10^~u@63D)&K?Bx0sLt5u?M{oy&OTon&
z6*2OBx4+5gui`TqZp4Ik6iIiNBv4_{#i&{GP-V^?RSFMBL_isf-mpU?zadQY?h}zx
zly`C;GHU+CczJm~3H9ARJ#3t8)b8-erMzQHue+~($K49>Q5VMJXMDujISM-i&W`f4
zcUJTSih};CzI%qrX-xA^kix(bQc}M(ES{P4hAi(E>!$2ed=n<s+jl<r2IQtn36}OE
z9O5*gO;a#Fo#&A*hkvr|IzvJF`6{x$bNi`|Q+G>n^$>0GXozxko}jrdR0@B#(1?ox
z4;`W6NfSR;tM1PqdM#6=q<SkgrNlST?}Kx{d#GAr;4mFkjEosSTrTy7Ny)%>d$N$s
zb222=#{?8Ewws&W{J9dqUhNXc|05-niRXTH2QI9=G;YCU0N#yBg0w71Y1T9{Rfthx
z73La9NH6J2+!2qeOBlK>lj!~RsClH~)<iUujA>e08j6fEAL~o<WlCcdmJ!UF!U84&
zG(75b>3jM^)-L3HG1OonY6K3Aeo`PMPVK_TqsPhv|9Bm?iUf|ar!e1ki030Mew{56
zO)Zw$R5%GQ;R$R8cR21W`ANDoWHaywPHJeVu03k|bUWKC*x-0=7v$%<+LYfThr~&J
zL~wtF@C-kBTdDs_f45F7u&cA9@fD+t0W%LmsHaX?vbPhQx5G3|SIvTCgT1%BJiSWB
zwI{@+Fq&V)fDDTQE7$Dd)AKDE&96|6q{`*Np%8uQqiBoD%_d)L*RKl9LTqE1=dI2E
zG~fCrONJ1UjhT;1Gc%K6KMo^{&Y3is!sVk_a0XBZkhq5b<}5uZ{Xp1jw$D$2frN)!
z{d=7owQ=C=2?Sp}y%ix}FyiW}cQZgnS#?C%N&D>~fF9NRgS4`;x;<Ryod=Q%6*+ye
zMuvux7(}hiV@~?SS4d~nsHvfd0UZ())8<HmP(1K)+0@*84~q8w+6-Sh#o#W-w=nNE
zGj`R#m{$Y}6yX#lXezJqkvmomX%HWME-{IC1XYhhW3ViTp{U5M8rwO%nquVswp`c0
z6(BF7uCC5@02vuYygbDm`AAmyc^PE;>wiUg83Kn>U7VSrSp>C<e6HCfJG4KQ<N4}k
zHh2y7L{pCYxVUzB8?wfh<*UsDLG|oXX|3u~I+{@Q-{Y@1u@vKYM99?z5b))Auvmi2
z!)9JIk#E!5=jeRve7zpD?6!pO<E8*RHA|be%kZl;uB&dY^C@E2)#diA+|?rS0&fQr
zv^&l>Ds}sX*e2YnR|d!5J{f5%@Hbwi;MwxIdM;y97`3;*>%+!(iankuV+TvD%pRSc
zoet>Gr2CSTJe1^$Eu*QaR1B}{tP@~g>*eqv{|(nMfDNgWOP=n`lzB<zuRzzJfj)oq
z!zV&>v&S8WJB~ETz8eZ1fu8exnv&h(an+mc$M4^OT%rB?{qnIR#{>B9ZL7AP?(bSM
z*7M3rb^#d8Sq91;=yP+^^!n6BH#&#ToEikibD|ec)uS=hWbxK6hkl8>bt8P@_!_~_
z+%(v0KD6nWj%UTrT#WyxFpwW@?ZX!1U+`Gg(&D817@wyvSls<ZjCQk%;U<z0cT>9?
znlG^mb2QBiv?+gh`w;so>;vi*qqg`Ff!2_*HtR<cEK8JGfsE&-4Z%&;Jnxp1;7Lgp
z%_j0l!-bXfx+7HeK9(6UyuaA)4KA(*<u}oWK3KT>>`u%dCNfM8aHpW@DK@;10`f`U
z%}KI54yOJq{54FTSouml>F03PW6Y3=59eaD8F}0z!JNRfBK7DySZFsaS}x@!?*GFC
z#&IN!dK2bwUVA-Z!8j&8iPaUPzerrBiA!C@5SVSAc`@p?=aemCkYyrB?m}lwiqINV
zhYpnt)FT1fH0mH+7|{lbZGk|NwBpG0YVG7p`1h_qx=BXLP0k!QliN6qvIF>e<n<mv
z{;I+Wtjx2sO2m#c6~Yz<y!6c!qup3iPFD`j!QZMAJB*%0GNI_rrh7)^>}K)qr;(Fo
z-8RoV#2EonWmX0=LFS|=Ri)4QI#M2026mzA!0JPZInGpFoT7U^?W^op)TNp<PIP?G
zm|ZN1R?J_YN7uD60(UTCd11B4@c*OfyrbFr|34n1L@A+4s9L2)6|vQ<O>MPn?<#6<
zSM65pS+giwvnXn}RP3m#DEWx05u+N!p1=G3<Ck;tZ*oq~y*KaI>-l^>o*WD#3|wp~
zA2IH=Sw6i)oOPa|rV9Vlk89fGJLp3k-?K;2@ZQH|M`+rbOaxe6k@9a+-#}F3DQD#q
zvnOMnH=r&$!pzjuN9wlh57uz9yjl<)fL}t-&NRn(Z)HVI3v;BP`pe2%B=1e8&R(yy
zO|_F6m-6E6`aI*&hAl^=WiA2*z2^Dg(c80BN9O`s$oxt;<&<2q<FGhO&j!+mDS7Z@
zZvLI}6l*esKO_!Dv_0B3hGLHN%HB&AaIVO&sIS$VykU?nmehYMgT6jp=xH7VxK>xI
zH`w-GJeyVuiF)x#u~>BYw2H-iHKe)OeQiKROTTEF8LuEhaYjiQ9v(hKd6xYVoB`xs
zM0WA#)KR?2j-jW@c(VE>d=lD@@zdA@k4*l#lKk)KvE9Al5y$JLm%&$Kz-j5l<*1cg
zZ0Urn;&0+yL}ybjKDOsVB<v1Qu=)pdKV2QuUG1c;wH%@*3T?0#eZZ-k+-*;nG9Gol
zYCbQ$<njP3|JS0gVg;=a`TCAql){f+9Ija_S!3T<d1j?dw4RL6(}ozP$RFEtMQ?qg
zu+R3_f9n}u$#_dX?sG>g5q<Rb#9YZv$Wx_>n%?o_)ZG|?3d}DxpB-GD9S4~x0PI=X
zu4Ih_kUl}+geJ2c6B8+2B!A9>yNrDs5&p4wv3aq3b&?v`^7jFYWDw?Zk6f-0R7iu2
zeI0ja;4*4sW28r<uUmA!lE=jz0wGxm6K73pu<MJ$SIWhfXBV01W<U8#2_+%5S8+j|
z;)J}&6km+iw}l!VMruM`blm~4U8|{8N@It{YoXiaRCrRnLQ(@x&!(aVDQc2WA(v@K
zVFJmZH999yZB=$htq4Gk4E5`-8vG`Uk&Y!kp@J0Ru}~YKymM_!*oZy3+|Zc4v<Pa-
z8K;6Q`T&JRBBA^_xDH+IX)OE!?<=Hg#PJQ2*MNUnBGTmXmz#_%fF9Q!@+FvhrioL`
zG{a+-0Vur+zmp{?2|P!I(M1+cdoRm(wsmZ<+zY`KH<j)463UD88#N)atBB&tqwlUI
zIew*5W#qTFhr?cvlBq;f*)t$WbRikwADTnRQWC35qv-KA_cD$JnL%f3?OAs%&dj_D
zj2#K`hWJNrqy7PsPUFjb21=KO1|0Gs(M0f=A48cGQ?@H=XU(E}@2*Dwzt#y6LL^hH
z%TpT4!TEV|^{aJ~1baA_c^k$q@&q&1_#eHo#c~lHiD6^w5eZBu^S**$_3{t7!s9Jt
z7lF;?{e1~FLQFCQmO+;a%ZyQ?8qD@rp(6cGQ5jLkJqhdy84!0EpU!xCrQE<(e_CC%
zT~|6OOluLk|C;*|Z_L<1ifN8-{n=`{OSW*Jv{LCzVC44vLC=su(ShLDT^W_Qqbh<N
z(l^(|X7;eNGP@y_&*LG73g{H@UF}(ZdI*X`)Fhjo-UU@PH75}zc-e{!e>7Bh=x!w=
z>RrR_IKXgL%J^0JaEXHA(|VN7T6OgSW?sTPL{ZN0b%{hVgKWj856muQhDexdl9iyP
zS?uqOgDJ|?j(7S;8sQb<N!@ksh|_0RL|yyFyUm_4_+|Qs3JQx>EJIWI;wsQ8@*Ci9
zHX0K6gpHkz?3Q^@!(k8mlUp~RNfq8Hy#t~RS%9)6WI}D?94U!r(p_DKYlB=OmEZDV
zLX>3X^|;Z77rRGeAW6(XskTG`thDp-sv}r1GtV|e?oSQ<Zwhz+bPD>{!TkEy@cZ|(
ztVMeEPQXge<5dxFT<2*|Fz)mehc~af+waUfa?#@y*z;7#l{k3fEar!HdvkFCoL2x>
zn|<eQrIGt%)3vH9i%fC<wzn^SFG#yKY;R#?s{`E}pB{2?3OloEsp!=HR(bA2=sv+%
zb42bNx!t&qbJr6+{CXE@!k_3b^$i{~6vsml-LhFcLU*l`4oYOG#}I{o!AMnp$CA-}
z)kJkLyc-@pI5;T1*st!IZPbiLwk!h`7b1#R`;)*f2+?8WPIc&Pj)C^a-3krFqjM%w
z>f!QE<NFFI9;)4}fk)&}uKh^5=;U&crOC!Pw#vZG{~_ua2llZBgXQl7eBt4kxkr*P
z6@IjTj+fVST;xUT)C(xW%*CZO2y@Rs7x{6t#rB9x4K1sZ<j7EGVtkDpK#J*DV5J1|
z^}{Fmvz|W0$32N$bZQV@V*HRT&E%vDp9V`<-lT8f+#f)6O1bAAzd(z0ZOAX9WeM_0
z+0DsRTrkIy)e(I$6Bg1>7D9Z>RFI#6HjLNwt6Y&Jq+4tnzY3!vyc%p5)`7b|iQX`k
z4r7cZA(956H-k{oejfmQ2-4r2_(orjq~d~C=EJz+biT`1pn2oy?vm4&Q*xMaHwNQw
zBDpByI2{=gq-?4InA=DpzkmHdkuY+&Dh*8GH8}$^>yO)yx_tXMZHBiDY`hh3R*=(G
z8wdyOfbF_qMVJava8a$$G)EOlzU($x;-=%LK~e@Q4?bLXzXDne6meFiNR5gruQb|{
z+Y2T!0+le;9i~BLs!_3w`>l$_&vnhEW~;?Se)I+@`ZVjtL!)73`JBp#I8k*31Cqb2
zCr~-b(9&oPWE3?$EwDYZy6ss`hmjRfzgKApTJY#MTd8nPan88{0<un!w`cmnHqe)j
z48XMX4nc4q2(|-w6Zqo`L^l0oo)+tpt9B9`cQSXUiIr3s8EG5r=c@yA{?EJa#jsHU
z6b~C4`>of7XuM7lNyzQ0`&+SMc`D=t**v%^HZZ%nmq2otz}^2FxvC?y*~#{#e#(-Q
zKkPNj9xO8HcXCm@R;$$6B%PELX2=vVEatX3!Y=|)=J3RPe=8=$LQb9SUn{02%I9k#
zFfwc$j7ZS$nk!eZ6Q&q+M=04zCBtBq1G_KwcTR}kix|dJZUQGul=TN~iumchuRSOB
z7Tl!{*OTL`o~(t=VZ3Z}2S!HiF1M64^Z)BO+e`736OP$U;Qj?3ZwVvIgD1NIftk0i
zank@?Q$aQ*crJ*B45_;(KVI01#<97>e<BXI4^}Qk9M>8wM&h^SMJSgKGz~7v*BgPI
z0kA`CYUpaR@AzbMbbrR&doBL_^lKi+#c38t&sopX$3HZkZ6ES3dM10BVn_1qA|A~L
z0bd%P_OVu27JY5KBWpD$NReYJH1oFG&&3ezV!)vS!u$dvLOupZ`29C;Vg0GYKLzoP
z>XyoWrzXNU?!%MD#h$L7juVlqeFFou0X_cJ^^M{I_)Q(v9eQp~p?BZla(E;L5yvb=
zc6lw&hL4OHnAzALQaW|0%=X^uJ8CHf7a66Ip;x`e9n$~ScS3EQt+D2nqwmHu1%K=p
z!`MWZitN-hLg*kKJycxpM8zsvf1fwgY_H8W{{352Qmk;iwzRpqdAP=;{6vMxvLS^Q
zO)YTNkfZW0VlrFlPfwRK(?h&y7o;vGnB2j~>kYFPFwI&o>tHkMz-xlYF;0LPZ&zjC
zI~odL?*6#1O^_OE&TcQl0VViX=zPH_m8E#t$VXm>8lB;9ceq9TQu!+*7pZwcLd$F$
zoQpSyh<*r+r1itALp4~Nc>deyVhj_w7PxZFU0GODa?$nyOl8Bpno?9}S~?r10d>mu
z|5pU`819fhwYbM)MaxOZ(o^VsA*KI>1+Qy-t6Qs^D}O13UV#dstaVM%;T8#mo16d5
zU(GLYfv=RYHRHs&=V<U_ZmJafqw>EkWX5FGI2P6ZrP3IdeJt?!)d2n7inEglOj`=<
z>CwXzm?#QH>ZI56)T+X*#d7*nQRZ*Zd`1L_Z!i?1e!@I&Y?FRF1qd@B;TjUA=PNa?
zG_X<ibQOCE*wgi1&w{q?PU<X49r4Pz>}&>TMmpLu8REQGVK0;rUC9IGy%jcZADI{N
z;&ftQXtt4XI8U26uw=iY8ypfQcxwnw0^^cF{g?mZMF=@m8||lVL#|_FBVulT&jy%1
zh|Ci3b#dD|xhxA@T3O-Ib0<@_%oV&3`F*@1!6h~D=!4&Q+zZ5buB3`@q^K+7W@i)T
zM}q)0J3TKsv~tV3@W3xa{2yES6N(plUHr8<Qy-m$QvK#IVRQQ~_hbi(G@Kr|Fp_Y^
zd3)E@=}%oyE=KIE4p|xncunv}o>T%hZ&ni%!>5fT*%|qWX!XhW{@;)4Ys*xJxvS~F
z&Kpn9TPiLy)}>&U>n9$Rd`Ov?aH;J(-`pA)$ilV<Q>`N6F5C>Hg1ptCJyKD<*~Tb{
z2j@$KYv!{Bohoix);Ak#X4Bo-H^l1Qk~<}7VUFsums@F3kwg?o!i$LA8v+6s{_6^E
z96dp=g_XEyo>3}=9D$+q2AM6i+xBl_4*;&{_7-MsMY0v;7s7Yu^2~#0Q)(BZj*6>p
z#!mfQ*Vm?PN+Z#X1Gf<>0Y2R)1yr8{|5FE*{8Gm{tZ)-GR0UQx>_@Y%xp;MnKkEH_
zHKEEWLmI+ics)}Y1`dP(=pIB+ooPToNE-8}u{c~9he-Bfwgz>QjMp8`m7+noLdfy&
z(sOBGXG%mV<cd$}WH+2pk4#wRNNcd|I9!<%*FQ;Fu)!Cli#1hKt#x`L3`djvhvgE%
z^&C!Cg<s=WXkn6{#%*Po(>vs<rE4(}?Z`dKy@;FFZ4pbL0w5u7EGQ;QPl*_Kj3&IU
z{S}paic9t>%n_*qCCFF&l+_e{`CAq*aO1V(v$~NSQXSzOwLFqL%<=b0K^I+T!*Y>b
z>UH6XT@p@p$#-S?R)RG(I%)?JS6c5-&Vs?(dHTv*Aaql0_sRFHK`2BNT`xKEF$wk~
z(1ukFu;w}kf|*Z06`r)c=8gm`*F{oPM6{*y4Eo#;1zaR1mPaaWj3bs@i(+_0vQd{R
zD)|<+c@$!C=@Pd;zyGswvQsahdP-t<dC#8Q_%&!rW2FTH?yRcHJReL%B3fcwVcdFO
z?viHxBIw$yh~7&gMJ3sTl&ZvmGR(Y%c`onrct<oqAA}ak3!MIhw$g<%_lMg_w>B}E
zT!#-Oae@18S&s6%P*SdF_h@)eJ0=LlFe<}5f@bGZqt`)30ZsbGF`{?l7_P%*{5OW1
zS9VVOCvy;+H2QBm<6y_MuF@;$Vr$AXP0C3+Xc^zo5ZyneVX-&Npn&(Bmo_z3Rn2CR
z5+K<2or$^fqu$2#C(Q)*T}3FxG2YwgJ}NTu3KMWKW85!adR^^?=R#F^6|@yj<e?I9
zzjmE-jOPP;w!^>tE(WRDx)Zt7%H&|$8}d5thuN8siLn@IlyOe-oUWMsdq0}5hc`&m
zi>y2gdOcg^6zpJXVy4`j7>kj3`~Y6l1h2?B9Xz($A9l5Rk%-6#MUjdu;#A1Net?gn
z+=IQkVX=inn|8IL-R{j#W<~T!isjwd#IzJ@OQXoWIX)0$;w6{ZF%JYCOLJUJI7apM
zv<T#NJ|JTUR-0vuCzyzx+XrWTXGbY19G9CFGc&CV)*=&Y_<=_=54N-=|Ioi_D=jYV
z-d!skd=bB@T9*LqRVy`S|0+TLdrlZPFi`O8*`}Y6l4=BjUV0R=yk7+DX>73U9NUS%
zF|$?CuhM>X?VL{$kANc%7!oYY<WI>^@k8Y>X6*vU=YQ)6;e;YOMU;L0SGXJpLf?7C
z`3~I4@%h~dk2`Em`?w5d8x7quMQ~t9dRsnx_Z#e~iOAQpb~~WLhn4HyXflX_UO2!!
z{HkuH4t<1KhsU;P_$@_d8{hjkkvz;DM^DT34&hNAW#yTIC_*5OcG$`Ixr~tfY8otf
z8UVtn#^t>EK@|#tI_R?85Vns)wj!GXm^HMtUePzBIHR9LGkmIh#}mi1Y@nhwuc?~!
zLFnPu53<ZgyPCbcc)17iMMUr=F`(=AI4pu&vXZhIZKM6l&Vu&c5Tkm=`;fdneVq;U
z2nLO4*rQ`dga&^Tmnu{xX%-iSCp18TPJ-e&5eOa0=;M5RS+%_19!RxQ|2HE}%?;dD
zc8osxVa?(s-1-E8a=w|Z$O(CY7U=kUdW8Y!uaz5q8KJ2VNM$jz&Idy{<k2^1roZc+
zT2<UX0*5@%cC|8M5~p2Gp=WLm`aM#>Qj}u*m7*J8{ZSy)CWMeI0!ku=zz3vnD)<yx
z32kd>IS9Y{r_>z<V$np&<4WWsg90MTTSE_Kx|XNjZ_Up=H)A?UxeYSCcGJKC!Q9Mj
zuNzLIscvM9%4f{>)nRdZ@<Lb!U&Kw$h5D_a5oe5ohjS72SgcX3HzR)+J&Kz<2}P1{
z6(OC#FqxJxxdwljPXOSPGG#N{R*$&%AVsdL(GYd@54b<m(^&J;-&IjA4ejOS?d9cV
zH7lI=Sj7UVIM&?bmdJ&w0T<q;Wl&VhN!`}Pb1*I>&c=wIhu^Juw-soa%M}Nv3FQf$
zXsg1yw+YL;U%$R=RTRsiqj1Ah<Ht8fHUggP_6|W^Iv$oz0;u~pz52gqDa3uB4~T(;
zIi<O^ZhY6_C9lL{Z%2Y(yZv^YB6d{PApa+-4Q0dyJ`g7JY>Wn5F76^l%hpebcXM(y
zq_c9kUNx{ftH7v5SsK~?0|NF}@hb6t66M_f8EQT?#-VPGG`O=WOC8&VuA{lrHOaCB
z)cn#5XT2vtXlrMiz7evjb>ZaXuluLw;{SCSd=PnTZr=H%t~DEmf|ITkGsgrM+$<S$
zZ0_lZc@c4XyxY@Q@?Yd~aZ9MbnCIjcWw2eiKVj-_^yx`URxrqy499O?e$BGGIIS%`
z|JT>^kR^rHxZeAGZ(8K2?_{s<_$bQ9O`;^;#mvk&)p%h>y0Km<?0mxgiTcWXTObCE
zZf=Z<a;qU!5LWdmM5>!ws{`Ssh8g6Kp1a;|>^UvC9RBGweGnN9r7vp1b-c6DE=9v*
zsdPpK_WE~s)$_Q-$e08p@%4{P6+(~KkN@iC4F3JwEhsp`bMN!4f(uXB5KFvruGeg!
zmCK-GQ+J+&7?~}QSB3b1c0|^<dOZ45lnC3CH`jTPH#W9hI(p&aJ<DtNzQI$#<zcaL
znos>|j9Vy9$Wx%wA8q}ZU;7{O)p%BJ)_yPv=S54@)`QEn2i;mMUEO)G%u!fa1}&Y4
z6pX5?8p-Q6#vC@6yx#zK3fbMuGoV2{<=^}9iac7NM<N&?LZD#Gu^*Bj_XH!tO}!j%
z*%bV>bN%kU0GH<Grrxr50VJ{<q|upzUGLYJ1s+1uY3;-4*3|`RQCeCVpm`}5@}B>?
z0C@${XcSJ3X!G){H+2C~E^Wa<W@cuUnNZk-*q1m_Z`IG=vmLY?Ny!mm%yqjMjV}<m
z@l!9TA`p^*Q2qulYEpr!eSDk!n5z=@6rwVMpOM-6gqVkC^{)4BO(VLWiUE(yt_B;U
zLYFsl^1N-B$z3vUsYiwRc|Y%|o9v-emn@hp1Zk#VyBJE4&|^IJ_b<N)g}VM49$%XU
z*0CZX_?BxboHCLa6P<c$5FCZT1g0RENa_G<og4#cl$NvS?dcvLcB#m`JdlSWLe2Kb
zMQqF&0ry<aZB<l|{TV54Q@kQJJ^y!>wcGePzs$dG%A3v6rnzTrc0$;0{*dSz!H(Ys
zC3lpR6ibnsMZf73&uc^RajdkpTybx&2lmR;YsBIyCz%3!ntCFG#0Y_MYz+aCnqL8?
zOh)wvlFhIWt=s&5!d*duZr#oF>7PhMuvJ!*Ra>{=MjG@of#=<pb^y!!&G`m+f=@Dq
zopL2Y1v=$KGrzCNmspY0iWOOiJPg?Ry~Ld5`_(x#VtxmAg+9CbjaRywb}$X9@9GQ>
z4nJG_v+oYH<v-WGDlR?$v=x?ke+y{7k8Sty+KvkCHV{BVEz8QiQ~iuB^$SkQ2>RvH
z`X>FPuN5hdA^-#%ZC1|BM^jc=@BjPo@Eg2(dShgS3Sy`edw@FwKf=e`i=B;Q!c`m<
z+Vb?DLU<6NmXg_w;Bhq^ip&^Hnx;4(3_5Fm_zebogfxmnIMvNQq+ePL1GZZ|&D(?9
z+QCMj^ALizEZgO|$mKaI<#uMv$@51YoQy{{iBw?M^A=WqEYJsxa=S}xb~bnw2#u*g
z$;))Lz@UMyXxsZd9UU*4Y9W&5ilz5n1@dE^4JO9!TX3hbFoC$=+0}QI1zIsGNu66U
z$uU&S3g4)sNuAhK*OB)CdbkT(b~OJ8N2vs-?I!k2E$@Eb>Kv13_}O0_me*_WzSG}Z
z6waeb!ll;#=Vr2oK+e4z8($vcGcCA0S{cqkjoM{0yCw%lD*8}fUWgb&ozYKb%-WUB
zeF}5ZRsLRigXM&lVbmw@44AeNu}`?-NWK*?b*_o8BcdrwwYY%1Pe;^`SD6_+;|ytb
z`<1{zaK^h&-jJjzRx$|G=?lpcBp>DcJvnnBLOJPP-^BV;DE-ZcthBS$)4DEtZa==o
z*z0CC8hsvS=>G4SeOxTvSIRiUBoZg?_;304NUE<M>TcPH(W0X=Pq1Ve!)h9DJoiHM
ziQoV{9-d`l<;V-rt59(vO;o5<V6n+)4HRcc(yv7KVt-hZCb^R?3s=68W2-h|k!$Rx
zNiGPeCg+IQEQFG1P-$e_>xkan_j%b$5JPUCu8k0@!wopPud+B&6Q`%MDu{DMuPhjw
zg(zx<GYl1f3ysw^ae%zX;^O-%T023&)!ckQK68$L6c8vTE34Fwg1b@W6;G)oRS}TB
z9ivE$F%xEf%H3V{@8-W`k=~z?6JNm~DqC(`x{8a+-ob%@Xa-pzlYVMi$v*`rB(hfP
zvtR1pJ0(oU&cEkzB89y6&HXy}80kKHS-a^zM&vl+=W{Azv(%@l^=&)PMh6npvPUN;
zCyU?e0Dr2sT7>TAUsIdMI9r;ns_JSx`^pHQu7T>W$3jl+=K8U2yBF6R<GsHt$4gnU
zvy11rk2PjPHRCKA`^P&cBTFs1hF+(F9L#eN$M2Dom9Wg}|F|JgVaRb3X?eFtuDa2R
zCDUEpuV3SgSDRY~Tb(@@KkgunrltCbEx=<KaK6D;nlq4sZT|-vPk!$^+ugWt|E#ir
z^>O5QdXN*(B5+YWOZfM#^Ntq6B{pk%I(E=0!xZ{AB&^jFA8WK)e2BZRo|>lb^UwAg
zf$BBhjjtySch)^j|CTs64hJQSvxWgUUClFF5m_#1S#u5KF8)71DUim^u_Aak`;%Ut
zxGWhTa~g0wFfnm(wVZaz5AO?KM%8&VE`E;B^OYro--uAmGfvg0Sm7)YeL0qwD@$ey
z5I)w_RCx&7?A`R_k4*pctm>s#r(>389ZbcuIO-eRv;JoEO<QN)HZN<&YpBoC568sf
zcKteZ$D~mFC#o=INP5V)z+br=ulpXqx@yz<!_fIighup-W-s|ak=>&Zhm)-zi=^b}
zg2&@`Y=y3(cPf2t1@}KxXhJinc~OI7VSrn)BqSO}p$jqj!Y=xf*>@R2FKVdrrDHwf
zH6u&*>S#!2?Kr&gZXCS01?65ZNkXWnp7N~6w#`JQL5f--%KRkIq9*RK7=H~<ygX=s
zxF@qE;DtrWc3B9wN}>~?(Ns6M;9F03u2(?bH(;UpqNDMg$Ujj21pB%psnc{I=6m<i
zzy9V249IKvBE;9wQxQI|0y^fZagQEexSR2srH1h7QTj0HqOAxJUdE^U!U&BvxLUEh
z7(4<uU=P_p(A*!9DdP{B4d(h*b&#LVmsXlqbN;*udT9T4edZug_%VKY>AAy81*L~(
zrX*vPQn+ighhAztlePM{1(I>V+h(><P8}nrsgpWNg2u8DI%MD5LeKWo?Oz898J3c)
zOv^JB)5L*_mYH1{<Cqt;=m_HRuC-5ZxtXAsZiQs&16khejNNx)^Fp5{TKa?W70pkD
z<?p=+@#6{YmABT_FqM7Z*-M8JF~H*}Vg?5^4a^b-Mn0~Tucs&~ddaW=F?6x=Y3#J<
z`4RWPqLI@T_we9|EmV$!HlD~r64Z9p{o!DA^B8qt!|^GNOZ}rU((ohZ(5BYKEMWfO
z7E2X^JkDtGNZnLQ&*@ZHMugu`IoSKZo}<-jVAl<33MZnSJMX#ZBhGh%r4Sk|gMPSz
z?S;Iv!Cidme|@1uS~uXw+1&KdQE<a^pVwXvT}wF!d}nk$0@$p(F`k1T4AoyytGpL4
za8x7j3S8c7R17=bpWATmZqA><^WoKn6S91<M{zJ=^@Uat$~n`ME%w$aSndCHaL#Oe
z2ck+Sxp!93BobRzp8YfsQw}X6Lf>XzM^uR;#n_)Z14uBIxvv;Q6$j(ia`6*ob1{y*
zI$Lb4sqw-VN;g=01x#Q5K2uu9M#&T$!{JV_@>vD*7;#okdt^~!4eNoP9IO5sQC%$q
z0sbl@?tTDUiiX~_mT>-WD^J1Cl9+XNn7wXy8tB8&b_+0LR19&d8*ORWYQ+pJy^ytr
zmTA52-o!GIpz<O85Jkwi8*wRw9D)l6HSg>f8nQ%1d!7F^ImP%IAM!fiYu&V&)fQlJ
z5{UGdVMDvNJTCZN>5`r;_H}=!Kd)4QjQLJwsklnT)O%Z@n^~GVEF?&ABqCr+dgxQg
zJIS@KylDlp9B8;fsmJuBtAE}LMKU8`JKlE1)aSo{Dc0M2=ftx#%spbxNQhTwSDhPa
zGFZIu<2ZWbA~7v+_8GR~rW|{&dAkqe%=<0fMw{s%sZCEgtO2`H5E@Q4|7^=k*5bof
ziBUlwnZV=v1LsdM+VTntAPYS7Mc+nPF@l}`ke=Tq|A+blXWZ9i+_BJDJS-L{SAjR`
zniZv;9n|?dD9YWhc^MoP`S1hY4eittP&G`l)RuMzdQw<&6{YAmrl)xaVwnBfEYMr7
zxBzt{5>v`VI(M^Aksx1?8~P)1h*zXJ=58X_DO><W+uwR0i^+#O1cKt5!d!1ymgC1r
zV!)&&oq14D>0br^hGJwQ4oLhe!yH4tNS_dsTu(%Y=Hx)4BUWOh^<oi+I1&i{XW4t}
zo~9;D;KFU6Ua%wlERW-S?=qGTH+XQ+db?}(b@}j&z1!m#uhq2^ZWT<<H^`|&FG{b@
zuNn+!SBmqxJ+5Z2w&_|(@z*v_U1i3ht~x8i++1;i6kKeopPPHE01-4H@-oymN6(FN
zc{gc~ocbN^H8)#b7V@U@M7jQUsNE44`JJ#N2)p-U?*(|*B*5dqE2|=0C5fRwoDQDj
zf+(DHqyZg3g$;GR2dKi@`}fsNLC+Z>{HrY2xBakAS=7B|t#^WFwzYNqPcWan)@ktn
zY0}Vl)5XCanFauni~I2DD$lV;EKYxs=+XYc-?45*53mXJe~SFSYLY`=5WG}A?Otil
zg-ed*vYR`;IeFtX#^yiS>SceMha^)Kc~7Q{o@^zq#F?9sYs2*F@g(s!<Bzdn8UY&K
zn7gE$FEy~(qr<?V?1g{gZ^DLaHmRf8_MZd<6j@j6&otNot>&%`PdhT*H@TzVluJxF
zb}g8hN_rMLu*sjC_$elMG|4z&Zt*{nSqv@~xz78ww)yYT9ayf*D-;bv0`zH+#=1}R
zV(;HH&8}>#cp_dADoy`E<in*2mjy9E50~R)L}6!n+$;q4SGBf*Ko{4FV_cw(&8*I<
zx!xVrD;O5&WP!8@>J&bV`?%m;>^~sEV8lcLK^O6E%nNyKixFRzopwso%OZr!;f=xE
zoa<H}mbcZWfx(X2Hv*o*fc=Ik&&W7`I*DJKKxH;J_eZIyFLzv@nif$*NX?ES?owr>
zlW;4aKC4eTG{}Ex*+pKI$-9UCF!71Ex!3Sor(zk<+YalO&==4s=EbnO6lD50e7FrU
zI(X5(7;Gb*>ZG$VQ7Fw7=K#;~WKC?^I_ne;oKZnijlfv!LhKkZ;R<y9LNTPz8gtXi
z@0up~=?h$cuXf{_)}$D5_xZ@~O__|c@(Lm6bf~|+Ebm1UApM<>>lH~xKF!IT_wWI{
zv?m0(SXr8p2zXN|Tr1LO?);n-CPuppR>ohg5H2qbF4xm+dqbLJy54UE<)cEksf=kr
z?L+rb9|v~Ma$H`6lMFL^@Vsf!M+&PD`%Wk$6xqZt&;8+pA^LBPPBh9l;^Nfm=*sr#
zf8rGN#58uLljDJf!osfP7Gj%V5Y~nv>n`AoxWZnYEK{c8EXVV(7g3wUKjSB)y}K!V
zlZtjd-BB*DpS?br*^=Ix;KRVTwu#Jkwojk3IHxu}FT4oy4<(fEqK&UxK37-Yc%44!
zI#$kK9C)#?U7I&)zBRK<NjFa%OJ85}UNBe!4K0)&e-j(p)i~)5{T!7i*%IKdhJh0m
zSilYIX$b*dH7TZ3@1lt<wyeMvf8W5stG<iI%dK9zW%wDGl^IN)owOC1%xkIIl+AH@
znS?f1zFb?IpIHSPdU1_y7k1}!X;G(}C-~B<KMO%kGJS6Ddkv6`emT_`t&G9+N700<
zufhLp>JOa`XUb~|3=~h)P8_*wfMh>7h1cU6&d9z+e9=}XX8QvIbrNxjS2`tfoVO@l
z&KUrs<9Iw?pfZt;qvS!J3Dndk5w=++P~=(Y+P9deUbgt8wd<prEPj+31y8LpX!pV<
zIQ-|`9%w@tz`bHrN9ZMbuYgrSgM;xw$eyFZ3D=INxh`+s%*Zc)c>@7bFTSJQxT>>_
zu@cJ?gYX{zc>n7!3MWwElLEoZ?}G5lGOTVn4};Nz0{f8!r0I8|*D8?oU!foe*stTi
z?IVYxB7ZIZH~4L#O0{TyJK59CkhiUeaeD=7bOUe8N4mP?dWPWQsmqQ{hGQ0bS9oHm
zj&^0b;PI5zBVOX@?^(dqB`LtmVf195LG4}9_)ExC)V)kx;}BVA@X>?V%d6p(#s1yt
zQZ{Ii-dLLD3e{pxv-!Ug7)ih*XzAt3Hm!&_7<WTH0{0}oW}ko;+{Fj3Dbs^?brgeI
zG4H5E*gCsu5{onVAw}Y(Y_8o9Q*e;Mu}}+HK3H{pEs#0@1FE#sQj)rZeOZf})R$Tt
zalzo#vmMs+uQT=9nZs}q?9oP5{vXj4Cn!5)#F<I#_jR{AUe9Z(n1WYE7HVStI@;#>
zmSWn~WMXdG?)9{Axoryz#a;V7Q}Nq7-%rTaB7~~ws)<_>Rc%6J&HG-f-Es!xb`ghj
z$s%@Fzv+4h00H#qpT7|&e}a14lPA!=3;zxkR(Hpk8>@9WB2WKb4*KPTm(v2ljPBne
zaI{-t@W(_n`}A+boBaOl5Bw}eaA_sE#eYN6KHk;&QjNB&9ztHrPf0q1(_M;Gj2M}c
zMzT#6Qk^zHE$l>!#l-v|sB27_YI{t&R+BD~lCHrPu<d&qIL$a6q39`6a=wD+5MgT3
z1Xu0ArRCDe?ZuG2<E1&r&|c~drSpBav)=o)o$h$-_G?G^zRyV>45%N4qBkr#<9$cP
z?92w|EH}JlJ$0?^Y@UW?r4X6%d0dq~Jy$qCTO;d4POR5etkUsC=T4WeQh@a%UsKD;
z+yoe)MY%}5#uy8mTkzIYGl9eCwW~b%b?#@rf@IiuJX=<NZZhNWdS+nAe2j?gDog|E
zo*Ul&LB1FvxA`S)f$gr15wI@%5;znA+u0I5jwbDt7M|NiozG2=nJI=47<lf!a#f4_
z6a6#hONc*VbC2!$s2K7URI*V5wURyK$q%5`ku$4y^S-9D8W2&G5S}>aczKx>9IyzZ
z|FHQr+yI4)v@^RUPG@6xtCD?FLJtGi7pOt8iN+Z|B=nRT$Bh^K*wo74T**!ydt~Xm
z!{guG-3{J-XgQ`SGq;UohsC*aL*<_xg}oq&4>@4}x_paL^;>kHrH7zG?d{cXz|>hU
zA$hs(CS-jaq=Cq+JHG1X;whT5oDICrR--qnx%VXPh8W#Iw2x(9s~2dAO#(>@M+)>4
zt}VR;@NrRVeq~nJgv$3QSR=~PZ$#8pT~&8{%=)1*$93^2jSJsx!S?!=r1sD<%q*3P
zrpW5+vr<SW?cd-9wH2?SP+4Js6u5oV>S~<wo#fb&975=Mf1qwW5@?38eQBGK2dUS$
zCvoq(oiei%WQ_Q-+df=(gOqFF3r#wUMemP23#V@J1gKhE;8m%N8*}pD#KOG;vV^y)
zHsfOZ7ZNv$vP2?u1gg~3-k2_Y)5FR~9Pdn345=zH_HveQZ51|b`HlJkGB3#bxlrSC
zR@ON=N7o=oC-?cbXV~(Z83jg2JYATj{fNNk>RVYkQFqCKaY$w+hz)Hbgga?buJ6C2
z&+yLpxtf}!><qDR`{|G7J~0fu(0>bD6%S1+o>u0x)YP}VQO&iKyIQX>I9bs2#e9<~
zE}r;zxOCY=bkHr0I`;rryRcYO$K&&4k;3Np^XfWpOzN%^aao#O4Oe>;RYGIq$G9w?
zZ^|}0CVwh@P7(V9|AJbzO|IO4OE=(Rx$mhEx~B8Pvx--I$>~Idi)c{KLW1antx;BM
zOdy;R?E%+7tOofL*SxXlysGC%wI6K>V7XMs=cC_t=gdm;jlisQqn(h*q{t-Uz~mH&
z+M^!Xz*YFTHWe;><WTK$zw0*Y=-=>|F4HL~jYFZ=roP(4ex$6tQ=rH<vp<{=CCB$!
z-iQ2SohfVnV^0%Wv-QnCLi^!l-mL95eVf%JHkR#fIP++3<>l|x9nE$Ashh_Vsj&zO
z`ojfR%qQn`UciO6w~odhPWf2MZ#}5VYxE*DiV(io(;G!JSGo*tLj*1;e!fDluFhQ4
zYA#;>{$F_^U58+sPl;_`*o;qmWh<<@OL{h%+PsVOKDRRoCnp6$xH-&GE!hy>l#m1!
ztF8T*^bXeJKOhRfQzk}{+WW0QoK)GS)1$zV20zvTxHf~i_oHwFOtw1ugCMmJRSG!Z
zB=f&FVUJw2D=MySWRcMl&>p4Y#MkBSUWfGx?P!U-|87zl$ER43POk2!@zSGUv(qeU
zzzS<{m-CaJf5EYIl9%cI=Yzh+jqEcGFOb#aRtL!fYNS6~Yq=d?{d{PlM=6f4@oMd;
z$@wM&q`t2CpVmU}+qHmp74IpL?8AhWH+V*VOmgpdi@qVid*e2rC0c+`wOo^%bvFRc
z=UHF%T`1G>XGY9|tDGkm)-QUETSd(GdR%^*<%xIu^b5bRvc=-Fo}9sDU(+!kp5u9Y
z!<6*Px9O43dR$!xPo{n1SAVBh<VyS0i#nYFGps7Npf3@eq$*nPgrr5qAk@xLXFI9k
zi?7ryjz7NpW>t}F6zKAhoxP61z>HMqc;A+m&{?y2;t2NbW-DnUyxXx3aASRUHipb}
z?|UMneo}Uf>JG)z>wl8V`5)si%8FlDSWtzX`yumKzZ%lsFsEmd&9;=ShTY??so7)%
z?xY}C&-6F0u{av73T|eJ0uM?Z!wKD&{lm%FE<yyJtbHNRh4`#hihR&!_F#Z=qD%Ho
z%b=`J*w*>n%;0IlDMHwFyZaDjtvx;9A_oxY(?btV!~7#}pY|L{aeHBAuP&~3l)N-f
zx=JsQaq&YA;pVJ6V<!fe7gr06z4btCdA(zi=vsDH+lPb;bne8M^yCrNrhp<ClRAz$
z6IekwGYZzftP-f1t8Vm0k$1hCA3G8*Jw7`UmNlqssSI}N+v-}{+>2Toop{z_y?9=G
zb#51NIyb!_gXBVGGv{1<J-P~MnTz)rmZezp>-~>#)BqB92PJ66_|m)(jX-yG+qk-=
zoOD7{LoHl0s6ptuQsY0VF1RyDD|L9$HUDx~&*sR6S-fS$6njsXV(;U2N{bp(W{8t2
zgI#u4uK{N~mwkYm$7_r={V%$TgA&eUHM_zFy6w5SnSo%C^TNl0+rBEi`X`4sKSCl_
zE#+uTCd{M?XGNj{Lyc?!Ak@Dtj#w#it=i!mO7Qk8NJ4%GAH6b}yQ&6-6lbcgqfvfg
z!c<<4ZM!V5{If5&a@5pr{@4zSO^RH^=XW2ChiWO1q}nu!-Vl;?vriQOjYC{9ysFWW
zi=}MT655*I+Pq9npqh@4#Q(e5``*BL@knnPlugU+<BI%e10UY00V)R^+@#%##jd?%
zEZ4tXKf84j(Qvu^`Y+|uT{p=4t+0Z7H=cu7w(n{={`vxCTq*t(>DOpDtA$7djTnrI
zV(zsS^xc0|rkhLwpXn<_QA3ooF0xono${|gmk26Y_Dh@gsIJ()&waflce812!)7)V
z_{H(o$1b#_M#$j5RCdtnx*mGWYIv0<1dS-{T&gJCz#j3x<!Ax6<)g0ccY^-#udhv|
zuAbcAF$WraZmLB?X)^L-ZA%!r4sIRnL@N_0zNprkD=U!Da6jXFGtR~MGP+etuD0ad
z@23@MdW@f`sh*9wsmG1gqO;yvmvsSNPa*d5>u^te@?cN6IU2}w7P1m`PHt|-^eB`G
z@|A6rExe74H{m#$)j!xE-x@0LSF9gaC$D^$e@!__o6~dnXN>mZJb~hLHRAC2a)4RH
zZr`RafAZ?D-Njhx<$wsmn^4lziOW8`jMc14YVm86S329tniTgRZ6XBc<+(wrZnz{7
zSGj#d4*$k;1_9H(AF%nVnGz{jfar6H2fc+0^`kZKi_&~qPuY>(XDcI)m=Hod*S$6-
znu3{U{y{-)SVCi?4@j2xdN{t8?E4zk(YsaOly9O$Qze=vOaQmgL}K!wV-!xRQrzOW
zoY3AX=k&fb>hfgq;+)uWvIjbZ!*^Hn$`*UijyHjWg@uWQ?clxx+wFDS;GaT5Ul-QP
zw)?!<&g+gbTkMAd*q;c@KK3A3DTFY1FtHrGIOe;VxKcEStD6)-Yw@R@OiYXoMl{LT
zee>?>3m31$z4cxDQ+E*bD~Ks2WkiM&ryCQgl&hXW)LKD4Fc}XtH8)ka|HdiHsj)RS
zp9?e$3y8u0bqVuOEa(4-9Wp5CyZpOA<TFoO9MM7<!pY1R3TKE{M~#=WzQ58gyk&<p
z=|)2z^aiNX_=&P8hGbgs*z2otS1$~|4*(AvLr-#7J)EAbSBPs&ZSSv|{J~ioUn_G>
zDtP%iiBkeRKV87@-I;DjF}idt24lD(;{cwh%2fCGc79J_%-j9=@-prN34Ir%Vc)ZS
zhSnM*qxGNHmm`{JxcDUqM`v@-nT%Dq?T&r_HzpAs3V4W)Qj6Ygf!N`xN9ts;txPFX
zU;CSd5*sw}5k_&#!ix=Cw4b!0YK&<NcOTC0Zyi)X>eJ_sdU_+(&*4s{rr<vVP-m`d
zPT!-bs}yhY<h$WJiGtDXR)qB?>Jgfp0upV=!cOQBQ?c^e@om=jwNolK*)6+;nk6~C
zNpV3@P#U$?ekiu7H@S`YI;p9?ih!=v-LT?z6gwU1_Z`2UmVGWTq^YK+{BnG=j|3{j
z7wl;0jL}=v()1YJ_()e_`30f^Kuz*dzBER)h-#8GSWNzsJ!7obG`L!<se|#xeZ)OC
ztL^F~nCP^9iN$Gt@Az|mz6WyHaN#pu=BBH!NomEeAJ4={aY8T{Qi#~FZ-e=b!<m*u
zNnNUsmvXCbYsatg`2T1RO-dvs9skk^durhM&_ee@_-$#>Lg^)uBkCL&vD6C1NiiRx
zOV9I^!U9_^kCzRUP6sEi(ts;zrQ?6#+je7X4dfxW6V{TSDRu=%Ayi?WAB48dXQgRm
z<hm^CrIM4wvo9t#I~UW8DA44!(#rR=B|9>2n#Z@BIv@kvK4L1fgKhA22e^r~$M~_%
z2f)GZw7^=+V4VjZP)>Mhj9eB)V_9V^V|`gYeHMYe-!ljIXm4BJM1%-_FXReb_+?Jp
z_a815YfXy{AFG`o4U4p1+qpPeBq3d5?ArFeZt7C<6z#uSq8J|Kqj4)mT146tY`m%W
z={x;H@JAgGabccsKhJqLHvGF0EQjuJXU*#0ngjxGJf}(o9GK5+j2LH2yQA1uko2cE
zM(E9OLf;uBkg!@&^ByUA=FKlS;K2C_eO43~|EjK0hcerk>B~2^PZGS%E+$Ns&yOtX
zY{E}sl`d}zJ`;MbWAYiB+9fU0cRrYSwL_@#W<LFQ)T_`ZUond3r`h-t1uKKrq!vSv
zFlttrr)CcUkS6xL<}~{;gQI0G#<C~#)^|xy{@3LkhhSGKP3!@hwSr05U5g=(ajR{r
zw)l+^PrPXD7K}k$+Dnc`)x<28#wg{Hv019M$lo@vA-J0P^`Ujp`SPsQ3Y<`GMhh}B
z4T-)bPxAL}vY;6-Y)$F6Gh&SQU+`{}@SF|YjtBza_jx5Se&`&5djAZ%$KELj_fHjC
zR(oX1(-M514{lJrU`xn!QmowQt{TT!Nj6N8KUK{$emIKb53=^m>Hop?Vbzbt>1Rpe
z!-C=jj{}mkizvm?A=o3{dg!u@)w}FwjP-yhY|2Q@pgO4GRAQa)ePV!E&&fmBRodmw
z;ne|f+wac{#`&LA!fs{?O$@sQ6PvTIl7`us8kdbE0rw?#bvZw7MaUa*D7Sl~X;v6a
z-Wg3QsKU5}glMJTH0Z1oB&UBw%7lB~9$Y))-R`w%e6{m=mPcHi6d{)Msin%}Hr@uD
z0usOmkGbcmB?0++TvCjtzClq@fmgjz##+G2Xy#w@rYpl3w#(2j7g@p@s=sy*O+$>y
z(Cw;Kt#-xEYHAvo?}g&Y`XM7_(WejD6e_27_UnBV=eD~d+pTRVmH(c)^9ZlC=o=TH
z7Q$9hCE|AI$NsjraFT(tH&jitMfxHZK8@MTrd^2AL^MT&5ckwm1=MrG5DYNCrENfl
z)>QI-P0gDJ-S>-$r;qHO;xz58vx~A?^}k58oJhovh10PsMm@IymycarSO?Z;K5Kf#
zmEU}?Hi>B}TD&=o1D9cwvVR{~Bo}2mjc;|;#u(?n{`j~SR$X=NMd%;w4Y+&3$;mjz
zg>C4TJ~r}b%~tXqBUNY==BJC%kHC1qnAP1q{acKO(b(rRvA6UuFA`>kez9=ARKJ#-
zsUF1SEsxZl%9YNv>5bYwt@zW!ahX=gWo;c|w=$0E^w^}wB8B2`jH<(=bFlb)mKQX#
zc*dJuVc>&e{-4>|#h*V>8f2IksYMCf2h$GS7h`?rb_o;7V=d*R{vG?wk2y{Oua!DG
zA?_c9^m{tw!!;UmxfrXv%xHW?lxlt!w3qkD)pf$+O=y#ab1LMm;rrjDkZB27LT%NX
zDuLCf>|Sm#-*&9I%XX1<y;(u!<}KQ|7d)S9I~g)2qN;O6_&-_*1*CrRvW^_5n$Gz;
zTtsg~uAljpOdxxXrO+QEYwDm6$lfyj00Q<S;P=&$CgbDUGVhWgn>f`3<I#SQb=MzP
zg%hZ(sR=j>+Mv#{m)0W_dI&FrHoR)F-JVx6T<yw)uv#wtIy5Y$DYjg7@-#0qG&rVG
zZM_FwekaQ6CpacXQ*mA5K(<v(W==e5_&ow{^da;{?&<iIw(=F0pUwwx*$qp#n~=Vv
zH@ZjvDw{9~-mcRnhrce%ZdR(Rp?*889Bd26IJ<t0O<*5}{&r|*WPg_ldDojC7nk(L
zfA$Ex--<LSNWP7CTQ&N4H_X=Nv@oNhlj7hiNC7@q4QEF3I7_Tlpq`q5-)!*KMAcRp
zJ%;leqOwT+GymwSUnme1)FAliz6yHDy3IRr`dSE76+8fueTvrF#~$|2Ycd)2K<JAF
z*;Duh+`bEL!(DP#@r~8*np;Oq`bM<Q7M>M2w!cD9vF4pk$l+6D8*_=4%U|XGODvPp
zRpSY@Xf%NuG&a@~Wjp(K;G{z(P63Lo^>?YU-q-1CtI%(*Jq2QPdp~zyeMw*R!sJ5Z
zv6f_Lv)-aAQ6DQFR!3c$G<F4VyqD`6m8bQ=eW&XcKR6RSY)?YD>h)TF801T13u|g_
z6TDucxvM#)R9Qc=Q0k<*OAQpwqP^8aIHc9IAQD!Vvss$y-wPt5%CM?4J@cF|iTus?
zB+{iZ_7qa+7$6;L=|n<ugR8)){r1Zimnz0*!zP<t6;yJu6m6yYwRY;ncV954eg#z*
z-sX()P?~}D(V)^QA!mA2dQuX}RG8Dq?6%BE5d5W(g-QFw>Wqv2baJ8F2kZKhi}N|=
zQs8na5!g#yWDB00V51>VRZ(MbDTrw^HaLkf>(e|C2GZHy-!=u(mMvroO-B0@kegMv
zY8cBPwE9IF+aIe(Wb>SqzjZdq4iy#_c0Q+pi+y7$YAM_X#nGC&Tf;xEYqa>$lQ1d&
zP3Udd+77xO2X3mtOXBs{GO&BEAWKlLY)H0nqRK%{<qkp(=ER>m#p=}U<hn%{M9Nx}
z5S2dxxZ=)0WDMxJZAzkdRwf%3YS;ns3{7?N4EoDY&ct+Y-YO|dDH7K*V(kAfaKW)@
zn7b=V$#b49>)(#aJP);J>rz@mk5$lnoQ7OG_Ya8_CWA03i8GAt;7GUbQ9Q#PvzBK8
zPe!*658GbYIypPkIY5_J3aY%Y2i~hiL<|nV;qLGlbGsN(z`!c!u9-EWOls#i51?Rp
z>}7Gay?99>*LcKn+GIuUYWARl)xpxKyGbqJmlQ*cOfbJ4t!HLFxn#V8=%LoQWp(>{
zI%`&@L=1$SoYKE1I}5>-;aGHd;oY^E>hN>TN7tjLH{AbXZSBxi%4f8nH+A4wq71Oc
zNMaFSLp}OjY8t}kzkcSuxGkoM7oz5NM6;Qwe5vuCpX;ux8K?OoWc%h#`y?GTbZ6=$
z&A`o2|4N|lZqzLm`ivz1e?#<7V&aD(th70a6GI?fjOkYHb@2#=P1i20#CTWPO$nxt
zA3R|!)!&Drx@>ZA4{30uP`Zn#R!gALgFo31hb|7zOAgX9Y46jQB{O<`b`XXQ9$++v
zHoMo1XU7$E2C5e`U|cSyM5L@i7z~6HqOJDmiINpM56_&3UHEJ<E^ajzRN!Z6W`CX^
z(q8(9(UqAwz6H(H==~3_F0=T&&3sIjZedVXcpQ}~?Xf7rvL`jne)D<-Cm@Qq=bCsq
z$tp@qFV&!qPonAZHzjRuRYN6_-x`0#BZ{^sW%Mjc53v*H%g}XJZ$_-ADE)AMuFZVF
z8O>Atk1NyTx&!h;SiiVkD%o?`RI<U0V^RJBBy!M86Ehr;<FW=9Fb#@wc>e2qRL)`;
zEW#E4STpFg2}P{!;nq?Z5SXWyJLd8?c&p1ip328oG4kL%E3&=gbN*s?sMYs{8Fd>t
z+ILuhnz73}PC7X6NVuMUB;tEn8N^wQf%Kj-59k#}%cN3%b+k-fhF3)=jjf8|<x3|e
zZ<<S$CS50(-qw%iF8@pHBJ>vCtkpG~P|h8zR4RD3T9Mr8dRp)rsDjs4nF;V&RS?xs
zbKWhrqnU5ovD*=fl+nIyrqpSqj-ujOt)nmVS#KDVa+QnfbH&xvK$NdNn3f2%2@VbY
zKoep;vn*31rGf1L3rk#EK*a?Db5(%w80WH)RKe!kCS<c()J<I+D5wV3d!)5SpPHIh
z9wa^-zVFoVt>F*Q%gVDZ|I<Ut^pT~_4gJ7ZernfG`6Mq~TLfPBTUK9C-KjV3$MbAG
zQJ$TC?5qMEFE{C$r^%nhvtH{{3?$xuIzg2jwT!49UH-N<=BuKlZ;~H#N`%pK+<Nc#
zGoLj2I>A73lJmv-qxA+$2bipWA_H{H;S&_5EGaaeoi_OfkiD@CQ?Xx)?MO+2=uUqk
z9AX!QQNP)$vQK+Ch6L`IRZ??Wcw$4h9+!e}?|68AM_tXF&=d@np3Nhjz(fqSF@=1Q
zjeR@p8)osd##HSftsfGx!?_mA{e=2Xp!`Lvc!YfI>4d!GOZ|=*sSrmbDMO$qmh0PE
zk@&e8&!r|1T~dT>N-kqgC>N5|UzX8#B20vUQj(s@q!r<T4o^~o@gF$q=gqEl*fVY`
z^yWC5>X#o4P9l~<#wO864=upc$*@f0`$e2o>#g74WM9<<Vf{|*`|e=d*c17YZf?ni
zh34o}54h_AX?^hnbcxcLd{s0mVSMo8<b@)(>;NdccP}|JmP9m8D~1y%_XSCz=P-5C
z(&VSOny&0wvpqcpmkN1nT{EWh%b!SuUd1M_SQhabJHLJ-qbBJS=AI&k=ttyh^bjrk
zUxbkL=$4iBS3IexA=m@G=<Lt~%a-FWcw!ejpRc^jFxRr7`3)8ZmPCP2q2V_Am5&vA
z?)g9`xc_b-41Bgm-(R}e%CoDpRglb%I-dms0Y$PeW8p2nk&%S*nh9slcfLtfq*_cT
zi&u?y=O=dOE#XJ5NX-Lk_p_2~85uBbJ}D<;US8g^%(FAl*+1x5#`>r-xD8wXe>9!<
zBh`=h$FJEnyY{S*y{}EibzQp2=Hgmq@4YhbwYRKf?|H54vNEr|xk4y2yb+Qj^}XL8
zKL5jOoO7Pzv8}$|p~kH9$LHf2OoA&G)c5M<&-u=%`?^P`xjG8GoXsE8AXWj@BDuq0
zy0We=9>zw{P*0?h?YAPg{9R5zUEkq#5fzme4aQRz4ECwD7(vEVa9J}_nrzUP5}Sj+
zSIySO43gQj{n+&AWbIvoY5lI*@Cft~)4Lf_sBQ@i>G{-i{Y-8&9>-PY^2+V&Le6BF
zUJh=`s1CHZigIkt#ROt73x!iI!R0X@T+(zJ9WfV6yze%N0GR5cX@?<kO0d8mXHZ6Y
zi`9L^>WP{jm^glOJ)HSU3jfAKf9XPgzZktn36{ro?#li-&K{X6FvO=dn0+-zhky1f
z9}j`jUoQeC{&Q{7jpiSoVo?S7%f7*&U_TP*@%AtjrKF<+EjnvKF1DAvzefy1>o<`#
z^|f<xxZq0srmuJ~3-OQ~UncY<pF!Q*MKis0)s0t+@Ych0(mJHfQgjxBv9Vv+%3d9_
zE_t&Yk<$t+7xz-$R@^NhflPTHZ8zA!XMOfR-&fOTgbfca%5m!NyQOFgjL?%T?Zrl*
z{r{qb4w`tH*l$wEuj*Q+yw^?xY0IxWT;6&0Irj=BkI+|@R)!&U=acaKLD;arJ59->
zQ%Z<?^1!KTp@JuxI6>mRog1@(K0}q4?CpQYNum&JmGf-3hZu-wKfECNo16S2wZFTq
zj-8!#e6JftMP~~;FZXQzZSS4bnFLX)ucA>Mk1zfXsBaiVbFxkK1=#Ku7$Ad*#|vtz
zFdh<U?}sVcZVkShIXSc-jx=Dp^bC>)jz8C8nG=1A3|eNGRc8p%(OGonJH9iy91<KX
zYDGTuscBI5pBo-K08Yo}++9Uzj%v6bmF-gozvF|ihp)GK{{Gm`SFt$LUf!gWx<yTI
zFcVR$=}X(WTQ2R0Yv*#7@A};yjf8jVICu4i&<!&0ytDlHA>?weRZO|ByZbjTu<wnr
z`I*l{nKPxn<osETWSl@HrM#M*Ky8iY-|?+KcOU;IsYsIo{-saotgGhv2#O!>_HIWe
z2KjSzoQ1pzjRD>Cd15r3*-6UrqB|;kvK8uW*K7(vm*x0Klb0*F)a2j`AB`!?Snjg8
z@VRmKdICm~V%;*&y~4?s0<3di&h;Wcuv)aS**K60@6ir>4D%a{EL;pVb_G%qVO$l@
z4>v>F?*^lY`UueYzkw|!u)u<QNSu}1x*4Y<aFp|g{LxjA(N)t_U*G2dRQK}3oq%qJ
zbbS7&@ff5}ha&Ognd@#`WHPX?{#%lqrZI}QU&jYWn?Y(?BlmANZ_6-p2WqlBj%U;t
zp$LOj+fW0eI|Wk|G)S2>`))1pVCCv6v|uyz_P(TJe7Y|$<hjOr3*Hsr*1ruNJ;)R_
z^y7zzF@Ax#Jo=qW<V=LhAOQ_<h{I<Jw0X<Kqc_4kH_kAaLc9w7n-V19c-GTwmuxNn
zBd$H_K~?`0R?d3f^RJtX(1gM37wdly935l-(E9*ArH;iC@uw;Lpp=v0eOi;JF4yIY
z@ry+o%nY)`=yM+sp~|gi#UrVycKwcQTg54=3r3wO!5Lur161P~2wb%gQcXhx$&rjA
zyyMgg7b$u1vQgOl=^IBMKUEn%(cT8kAhLxzTBbk_q9VrkGm}WP=V}vaTd3<KK=3%y
z*GVf%DsYn%eXavYX@`pQbf5k~{Td1Gb(JZXJQ)$M2swBS%TUKlN;68Ra|h<w4)>W^
z(UiLC<hKXcD3ic-U6tR$*jGXLi@-eH=?a7G;j7PE%k~&0WORJG=HpymomQl|*%QK~
zIv^!c5QeLQllzz-%w>&O;mTePr=}v~CY3_KS92tijMUR@6ps49Sjy;v5mP>7v}APi
zmV4}7m(<He7yzgtaa<WQbHLuvM8n~G=KZ$&bwC|z*49F0>(j_yCnM8+{t1_L=}95G
z#j}8*j&ajZm_ym-^#&{DTf2`NcGQ-gy&pb!<lUU^PY0QQ^!;~!*BWA|wdfZX<gMJa
zwss17PIIck6v<L8)ndmV{%dFp4`lz@g!k|S{dn~(XF2sqN-&J$$!l`x2;m!ALHp$R
z8=-bil+ioG*r%CccS3JG@s1H*i}?3eO^EM}ihWqxkzu0N;@0h%xvM^5DN7z#sokuq
z3*6Cvv$X>?%>&G~g&QXUjo%G{c*D8dz7`CgMbmkF<r5^6VypQ<v~JqLwy<#a01wTJ
zYYXn3hp_>4HX^Tr8=5MmpK6UGF^LkD(Eu>2+WFD`Wl)A1Hu(GG53B05oZqLXI!{3N
zWJ%hgQW`Wp7(o$NDd3@fEh!mmzfowAX0NoW@sS1{>I8n@vZ**y<8c@zO>KG9n%RT;
z{Dfq%<%;=1JGdmLa`)59y&JmvV%p&0@&l@ZlE3{d@nibrNo#yeq}H4;*Fw%ZU(F-b
z1RRFV8w@$<0UKYwa-U?iO=-rNsr?k0rY7h3$n#>p8yTHy7U~YW4pT|p2@L|j5YDS2
z0qna`${-L7m6AAC&oS~Sj}}4#iReMtZeiWTMBR{b8ISpm4{pX5dfX|%(~_Nh9+N+9
zer7Sl(#xl2*Y?{U(01!b8Gdxj8;R1aC5nd|7(9TpeULxc2~bZ73U1UV&f%$Kht7@F
z@X0%B(4Y7dSv$-KR8PG~>iT_ZGbS%TinBkMQ`m;V$Ju!4TY$VQZ5;59ma2=af?Jn6
zDhA~;R<_)-?_grcd7szjYeQms-g~&fQv7Z0weq1!;hB$xnyLGX+_Kfvr>L{qv!~V{
zdfs33)%SjU-(z8ARWk5atT-jA=ko8pl*+q+?jNQLpN#D4ONO*IrAcT*4vxE01Sq1s
z`(Q}`oPvW!OQpbLK2`n{nzXctcr=G&GZxRm838@$Z7ux~W7Qa}jm>F`m|1*@wAs3!
z|3s%0%^o`^WzWoakfT~P_6=qE#i@=<g3*3}w(8we;&HhK-#16+SN`wwZG${zKVZhh
zialSkp9*K@OComU0&eqHnu^(;YcG1N)WzY!w!bbW?Vr?<f8O#pe$}}tf3Vr}VYACL
zf4-Wz`rX>(rODxS`1fD)-S=Y(ym~FVR+>B?Ds{oOw|=pfpb3l;A*1dnovv4vtqdbw
zjZx*?7Ecu<ciC+hrSL2ko`XdE#=kNh`gP^F-JzuVkn>F{|B*mVA@>5g66U=gz43OW
z??dQr`RT|*CVv5sqx-Rx4ECDitmP)yPuKCLjFmUj!3Q}Xgvl$^-WwQeiVW%OJP#Jv
zUUbr^WhBHgL-V#PAX-?|H{m|8Mo#agZH~{(D+|H8+O6>qlmPl;-K&Vcxo5N(0)umt
zAgKMq8?$-Q(?d&Fk`jW~@mT{VSu1~5LLOa|erPC?VI3Nzua^dZ)aGPq)gZ&aaEurK
zex0Co08cA9_*oi+Z{}5KQ}-{AIL76Z!g)h~sHt&B4RSmikxd{$5d+C6o%3R_C15*Y
zq#CY7dNd>+ZfFOy{HGl^Z-S2Mr6wd~Yr)BZn5&)1OB-F!{E@=!KQp2@;#J%1C`(kM
zBJaZEIio;;1TV{F{JF;E7+Xf*Krd%5OjGmy>Q~)sL;R?Z6znOU&BI*aSO8=KX(Pg+
zCEXC11R;7s8YMqyo%N-mXq0C7E8y7c7oT{mmjiThsUhCW!Sw+lHiz_Kdmi4IW;k<H
z^jMHqTOLJKaPM~Gmh*s^IXyZVK*-t*!?8Gynem~u4c|;FRxtChoK0Y+kZl&;eM`cs
z(e3Bc9h$<ql@07%oaWg^h?bt3hbkVPr@w|~yNVl&QRO)YV)uO1_n3f7QkS>a_^$dt
z537mo`hGW@g`NL#wz-Mg`d8i>;<#=iM(*=T<w-G<y?RnB(vU#^Fdv(KM3^2*0Evu_
z%~>4F<bf+D;$IX1kQF3uyrFYH+^6q*?r2t|81>}8D)FL%3ZEG}tuX)~%jRMC-)Yx-
zWErNOsOXYoY}Zdt;sZ+?aZ-1Eo@cllQJz@g57#kC^7A=nAK7;tK0UiQ2or4l&8rPQ
z4eXA;UmV(BHNAs&md*eA-yi63Vc`q6&b#gZEVid7-@dT;_-c)QWcKezRL%Z@+DG?^
z?WpUsh8KA2<o>G6i_Ng#A|b2K^gm=6^ac8NSb3rV!lmyjI{mFQKRabB70-UmYvf%}
zq}lR$uUz}F+S#Xas60PWT4uzE-d<CF|7+F<SC>!;pLs7IuD1xo9N(Wa;P~w)?02tN
zh$W)pjY+HjFaef04ajhVA<vL+xVd~XLXOUbK2>G`(&dFNTq5hUmdIA$DC!TJWGUv3
zJ6)Eba%aCdJ+^rJCgV#Svlsie3DDR>b8DY0DZgsIdima%clE;hKb1W?pMu1SotdW2
z%IwxvHw7J^!iwUXoC3+*A=>W+ml4*kdES()y=??~**r<nO`X<>d%){^IVxcxAmY8b
zfTdEt@-GZ@6x`fP)6dH2s<{T!Y^lM!1rZU#s9Ib!U8}Ve974*lefUyC*vpg#-k%I1
z9glyhT*En@-fXvYIM3?Zz!qD2e4DdfetSRJ*Opl=C-m#YV3b#4gdrEfOE8?j&GO7<
z3@}VtVE;kx_U?Y4Z1FTuFXLrw$XgQ=ll$wd7beAuCqI=WDfrH&R(%3Y`Jta)_m*UR
zKpAt~vrePL^ONKV9tAkmHnmclKTuTs@rR6q7D(Z{KZP{?#2nJ0xPBI>zYwY{yNjhC
zl;S|wrufUm@<$FTTvu;x{V|rkE+)NcjZKIj0qkM~G-ST@c=`01wq0B3U(0t7bp=mO
z!R1>`pK<}6n1zR|Peq5F>N&kr)3|-NQ7u2TK(riebQjC@KU!o7Is>obP`}Uno?}0d
zc^1@#|BHIm?v9*!UWzrlyg9vcxZDrfx_!C1G!$IeJ)nJzB{j^R_xdSWBSIJ_MN<<l
zQ|a8cnGWWkNmKv9GA`%(V81{Zd3Qr=0H=&6&XKDIyp({Xg4t}<x%c7Jt#%AWD14cH
z{9d$mhJaj)faA_dR{2dLwe9!K=z_!!oqoI4r*zv`vquI2js5JCU1lfE!rV`tJ}+JA
zPNrU-B1u2rzZG~dcO8DdX(La^zZJ1J4ZC&v)7C2p(+s@a@A$WUP;0(3bQbU!rgA#^
zS$pWe0Dt96uy$j;B}uVj*xk3U2o!-Kp-TR8_ldhE5#eK1o0k#=0{|MY!Ko>KOFU0+
zItZ~;b6JN!eJQWK5{Sq&D2qx`5G383g1rzJMsSw+TxEjK5}Ngg^*wio9+F;t*VNY^
zk~{9z<#frYb4yVE&@QuQ#DK_wCf4$k->0?l+hmcDuWaG(_BgAblt0J<qQH-0F7F{J
zd=(pQ+i^u&FQpok_ck<lDzYXpS^|f+f{9Q^Kl&dGk+&P0^3~@Cuk{X<!wfZBzdf&c
z`fnU`;^&jKF#q015H00v^23CNkEYV=-Ltv)vM}3&&!IBiy}j;htK-jX*!if$H{S-&
z?(J2EN19pv@yKBc%fIN@*+miT-;ZRHUa+PSk4<TIk>?<k_tli&V>g@$Qf#v~)By`*
zuz-Qg`YBXp;=BLBsBwS-PREXMvq>zgRCJzCE4%dD$iZh^wuCVB6n;4hz1v@&-f(w#
zIAny>#WK691JWm2h_k)(0&bOB(V*7k#VpA>vOYdD;eEWxwWe+Yg`9jwo4XMZ(!KU)
zNKE<9o|tmz?+XNQz;w32H<dN7t!geK!dD`$NSzb{0t1udw9SO$zLF{DD2?g+UgCn5
zO9X15QO@L=M|E2LA;psShn%2GHyQP+9K;iLpX@^G`2Oe;hJC1^`ao<vr-z(vaNXCY
z41=dzkl~&@4sK{c0W<vjxU;iX75MBxbMi|*>+++6UHWiCMF1p<@qr^8d?0snJ*12J
zJ+F)UNys(T048Owb^lx^rYt=&vF?Bi*!}AH_h%EH(`vICdj7kwls~l9sXNhgX5FvZ
zt!|K2ZS9d7;J{6F1my2r4NBB&!6k5j%DsUyj$mx!mjTogl9q<kejj|V8OOoJcQVBM
z;6krun!}sA09tT!chtpUGsKc3EM9({s><uIQzwA*3uk@|EtJHj+HMQ%pNNM3b-Ar$
zDild{EOvji6?Od(sw1va>{w#C6=OFu^C+{9V6ckq;R{P|=9#=!*<#Br%DbfeWiix+
z$@A9BN>5ZC4!L$(9e@j{+ei|Vy|0{8{jOD~@9rc{<?i<F*6n`z_FszxY1nwdowx6`
z=jR*i8puRf3-;%bo@~3|z(mVZXa`0xWlN!(9?BoeIR%8+RfBr2XLiIXo!Rfd@^f2o
znW$%JjCYzfyb{mLXOkHH?Ju1<;9Zc{qH~Tu^Fvt}V30k&qK?c%#2nZ)Zuv^L27mPd
z0ars#Je{QXT7Z|Jwl43@FXEaBO)7j$8Q#XGV8V!HO%<pzl1*W6ux%r<ZlD5lm0shG
zcywjHeHaXwR6z&dcYb*njxx)wi)H}$SvtOcQmsRz9qfGkTc6A7cHNA~*~-|&e_!zj
zD(X5_vRv(~AaA~w+;E_mpWG6u4zpZgbFCKmrXb0MnGnS4A=(2+0y>wM8wDJ>1nb#$
z&Pam^#^XjBT{y*FP+294em8VZnK=T2kbdxrK1pOwjPDHC(OTKAu3E`3dAPtwS<3D{
z`>gdVW19O$p=!}^&4DCfhW?sH_1$+Ag!PcRIy9u472C`&#0Y|brgClL<t4dzQa8d;
zfq`r`l;D87cV-Q7*6@Oe+~nG$;ARVQc)Y3)Y1&D^^?$2OAAgOgD2DuW>s(hj0CD(g
z3UJ|BeP3$C^RNR<B<zzf<oq7zNI!Sf5paIA=jF%O_%wv29~q4xPx;dGSd9}^5&jKn
zx<a11^XdLQRed}X@57D)b9B_}yEDR@-+Mf(<d&|Jd&@)DN&}-GZ-VqqbRvNc1h!=O
zEy=krQ_|cuB8PxXgo8Hm=Tr)q+d(suzw1B6$}EC01>1g#RBMYV%&iI2)<02|K4xft
z^P`QPlUSO)8SW2ZH?7^`Cmn$(+&bbT=Ez<P>Gb0J(cinD^EUFAmgbqwP+m4Qeqyd%
zNi*}6_Qn_DGs9WLepLV;QsN1Ysf4j%dIB`i7dM-%-@@T2FE_ATU`jl|@4L@Fo&^Dn
z<@qog@VXA=#D`Lbb|*&IuTu$J4FL_&SrOhMlrK4WUdDQ$P0Vlhq_<G0gKb{-w*Ucj
z^SceVN6J0_j`o6;hw={|ZESun*g%QtUmb=!y>-N6zSpaZ!+R`5Jzs=w{&JJ|#di=R
zmyaz!Zu}tSC*5?9TU~I>*+gCUb}yfL{`+~rggURz2Bam-<*2be?N-7(BU;W3x)(fe
z&8wEnqW?f)uj1i+*fT*?w-gpM$4p4WKq5$&?@LO2*5gC%!dELFFWUJ;#2%3}O1D+^
zUvvPHYFjS3i*3#<f7@D=DK(%kZ|XF*v0%FAa(x2_iWuOQzp}Jk79e94o(%r)9%-)4
zT+e$!&5_~I&sP!gVD%jfo1YyVWuLFXe4Zk9Mkkt<EAj_xYmLXKssec)uSUG1TMJ<#
zc&HCqnX7!0Sk#EYw5gODEgZJ7_jCb_{F?6h((*bRyatetn=sV$v)7)5P<8%x`d!zo
zdRSj3+|gawG{IhmYm~Q#wcw%v>`xW&iLZD$S1Oi2dEq}U@;eusbZ+3PV}o(4x=^W#
z6mRWYb?#vfK3~)hk*Annj-Jh~+cF;-MAduuh(!&@?3(rs@P>;U28Z+`O^H>3WCSg6
zt>#Ch5W%`Nv-@c9p0(aXQE-xs;_0a|C)S!(w)<pyT6^Uh%%f^(X3nU-E+b#<J+Y>_
zdD<an)%WMm;X!AhEI76cNj_ZQtEIXHq^Lt1m<;pFp!!Zm-}+`xC7vD1VX9I{IOvsg
zFo$!7*#+&s6kWC~hC1?gy_7tYed~398yZ~!!kXiQd^|OwR2iEF(mPtP&6|s)>K7n8
zhz+ELBvMr>p*&bPjgy9x2x3=QST)R{u;7urOn21VY5e=~n|lD??5y}?^?v%76Dq&<
zmv<O%QQDnbSDqfRTDRN*?Lo~Ctai<3Gk;^3iC!$8ID|FxdwLp6Gzhlp-ZHe0{0=>}
zAhzSoiqgVysBK3>oV0C73Aov4Go`11X*48MBRL#*|NiR@dT(!n5L3YBOq;o$h5eMT
z*138^P8I>OA*_KJK~yVRKimjdjD&2_xiu>1qplOIdIi6!&+Dq|)9<dm^mT{y2Jn~N
zub>bli{?#BgK$(V7@^iS8rt^@Y<pC&W_2^Qv4Zc@DXd>N^;OK&_L@i@nAxWOX8U-~
zHg3lDc$8yi&mKt^Q~?s?waw)rB`rI*V<pn9Zar*c%=Qu9tH)z3muU&3i3d3(<_*Zi
zQQ9K{T(wa)@MI1$T8Iq@ZBzP4_qCS3I3a6Nd^`;jr&DOf@(bv5@r_Tg@A~xd(}&(0
zZ>dbO1Fw`0{xrL^1P6DdRffe~7j}6H`6&1wNkO8A(eGMZz1zBwk}Spc$ge<;Uu91}
z{v+z#4E*so?t!8O;!OY=0&_oFP@?LhLE+fd#l()fWbDw%3cM*xS=FT#tg!#CZ_DT5
zW^aW2`zFZd3o2qeHK&(Ayq?Q+hMK)rXG^t)iR(2@CY?2qRu`!NgL~w8$29be_t5iB
zZE_ZroXL|k1d$L*eWBK_<9dwXtR&vw%X@_u6s!$5KF?(?2jK)NbB-_zWz|`zj0{v8
zE57^?517+7bahcRX*z#Z--v`|olE3u{ph1-JAC)*E?dId*u<ne!a%OB9Ow7Lm#Va&
zeZqCY&{d<Ny!e=KkyDY}dYKb&=_R+Dqvg@|#3kE)v1RI_r}0=}ske*?3dkPV-ma9p
zNn&l^gFhI^G8@lxkc~>wD7QkncdC@j5WQ}QM!IQlWIK0zl%9g%lUgMDV=tp_DcbjF
zm~AZI4E*l-#xU6(aD7W*&(NW<vwStJz{^OPP9o5}R7yJB(kLS%qtnzFpjm7uK6&?Y
z&g4vF3LM+eFpqDD+{YOMuUK-w(zpMK%$25CvkZ!}_M`Y3@U*Fh3RHsPC}EuZlFQS&
zmm@gqCzI>cnOUV91yA;tVs)NNGtd;C`Mu<q?f#Sc?U!%bx<FgXt5+KmYzG5JMVi1K
z0FCsz5)}v&OiRnwhUMhIDws-O^^(aTHstVJT9Gv_U}lajf5IGyiCI_4>wZ)5r^QR5
z?w7i{ge&$+b=mw-84DD$@3?%lP^Its%jTN_C$r6;V|Vzr&hN&-4o9JO;sVdKA}Dz)
z#=ay=M%S}x;mzMJ1W+ktG8YhOY=jsQAj7*zVFF8vg+`?sIsvY(Qg{@B%u_@mlw?Ye
z4^9}}PuZYA0wkm%x6y<$vUG}+iGT^H>Kp{Us#5>}G?3E>1y7SRB^XMl5#|s#YF#?p
zQ_ZK5EIEAiBslr5h%ONI>(|Dk?M<%LKR?EAe+pmgS5W(x2mpR4#zTrIgx@zfcLiTZ
zD4zYU|E_#{Ar{gDf<_+ttsOH_J^31nYj56XY=;X7)HJAe8jwocAbxz{Cm+d@j-Pvo
zNJ>hYIC{ImjryONfI`ZBg)i*l$A^{BlJH-@&UXMMFs;2Ir{5V9gEh=ag!Dm0Le>_U
zHB>@Xjr2qFp|ujeek-3;x$Kduq!bhsnd7i|uSyX@NMJQKXUw^Dg$dwza&r@GxdMUX
zqFW?BrP18p^=)?FROwi#MueP&VWPsdaxachOI--f%!Ss~`o`nR%1Vz+JF~*~I*Zyb
z7;Pzi3EwT{^&R+Iy0RKS)reyP(34AUrpS(FQ>KjKm?vebvJT`~cKnHGh6cTA!uiN_
zC!c2sY+3<AHc<=zWcKO?1=Kf`F_0vP5X)AmDf0a0uBz3Fj{dm~xIV>>$NO?LUgI%}
z5YRuEKoZXu;{b3mp{FUDp14Yx7M^5^O^O<byj&e$WaHB*Ff%uQDy1w0&_1>FoIATu
zsysX4)6v{6pQ{5H2)ggoqBz@9Ut6L6Y-;h)&OHoYHYO5{pRQaBB*FT0`L65@k~+~#
zW2YrYPK0ik>)I;n(-4q#GI?M-+6Z<1`J2J2;QDSa*D$j|fll@D@o`;U-Hiby6TaR{
zkQngAov`3-MsRZmB`uz7Sy(0`$XAgj*ZgV{5Q&vW&`=ViMdNa~k~oP7u^^&Wfc&Vi
z{{y9#E96B-DwL9tOkOVsvD1*USZyO;ZY0V8H6kw?AL;&E-x!vCPiLqy81-*q;PLgt
z+1+15O{-f0$;q1{ri=H0SUtF;6n#0fPp&_i0iZ*fPNsBU@yt`pkxPzM8@q$Fj6;&y
z<2O8e`MM*6rPoA&g`i?PfLru)gbo9MnFQe1H3xyH67fUULzQGo%!L#i?m5z@GnY0J
zQW)7oiNKLn1jz(;IWxTY{IR6WuxT9rOD-s+)9~P*+)nVL;9FzrVPwa<z)Jd0#V0mn
zGAn2Ak0yFs$EAF|?x<FJE?+1US9SLEfF5TKSb_QR1w}GdJ1LW5DnLNGYFf1Z^Ej4v
z``3P1Pqpg=iw$;OkZ>_FF(GCl<ARv}N}V}oNoLB_2TwZiNx?2u7v@V8RytAnShdR~
zTss`&lN8Xgh+Z-4ZfoP_+Btu9eY&x8Zi)m?tgbL4ANXb4JHaK3Z`7B?owB}Nk*pb8
zsW4M!&8Xy0Gcq#bR#uppn4}S33XY$s<Z3VLpZ<Rzl<h`_wt&5|;mR_PB3|u2h5q=m
zxN1a^r%|3R8-TL5R*m7iib04q5%2*Do*8lmJTOwwC#G5_fDn0EKCExfTf}|x5@&XE
zb93S{dD@b6n!ml7#dwcG(|3E=ALIX3i*r}Kk&U&%pu?``y7ll!*+YT=xnRX(D(8z>
zcLF-82R7wkZW{n3-?&pp%d9JG=de;hFhk=_CntMiHVdluT4hLsiRNpL0j&nF9dudD
zmFziZ?EI^i5AE&vUg~hvmoThwwQgbGzrXizI2p>I<7A-R**%J=kv;W;jia@rqqV%F
zaZ`x|b=PClVOH8Kkewn+6Fwl}=21m|sdx`o98%BH9KkM~ssY`*D@QjDYx;N!Au5T2
z{Dj;9bSPgp|2;o6>wQm=4q!4TYsD~E`l6GfmH7)5pR-=B7CcJ0a?k$lZ&D@~O}y1W
z^gl%hI>@OcM2|q;d7-N)M@=(!EKQx0#-5PFtR-8)pSw+_0^J|UK|1&RfrfxvdMmMo
ziQyM^gMw5YM7cjjM5$kg2K3nSfy)l%qLXz_IZB(^2AGf4Y3%BfeK?Exj&3G`?MA0d
zGSXvxWYk_P`C}&JSUjhz{2AmVV*1V5c0lh1D;ZxBEH}R+Wu|u~aE+kjka=aulu^uZ
zu}wX%OYmub?#hIc^+RTx{nn=*Iv29;BC$wfq>lFlOm9V+ql84&nvfnuG+Z)ldMqgz
zjEbXv5q{b2A@Sb-SIG<Hs(ayTdrS4lckxopKTJ}Uy`~&ros>8I{$%;+tDaT(?}#@_
zfg3jiPp%dB$m|NmD?n*M?pIRi|2Ddryl-*k>C%Q|>QTc9zQx+rhS03x?Bqdf)=n4S
z`>@YzYk69{w?9vg<AV_hXS$WF1ur=x4nh<_!LgzdxfSrn!o&`FZu1#e3{Nt8o1LBd
zPzeKwkCe#^Jld>NoPAnO6Fgr3m>^*saP-0f(ZG>#-Uub{DrTe>Ae_X*uTwP4_$svm
zEa5lmVil&dnly)dF*jjOUxD^xekOzf!JPjVS!mM(ttckwl^3*=8x^`3=BVyT3ZdYY
z&r3G@usiRy%{*lC_mI>g6Q$Sq(mQ<Po!=j6s=D>Zt|vI0akjmga9L7>C}*#g*ksGR
zQw153Ev1)X{kjpUMV`N@?=_Q?nl4&Wiq@&Pnl#DNP^`ixV@`zB@f``tbqzvr(_<ru
z0HVh%=kvhOjuE<mh^3>m)*PEEM>U)~Nnwr^&KN`|X`dr!PJ7S(hMX$5e2{-g{gw)n
zT*`;EqR0O%1u7To>Yl!SR6L_OeO*Q0AZ_RO#vhnwz^Xf%U!F2<H(by%g*<2ogmEl<
zM}FXWahSR&MFQg0IGW2pv0bK2Ia*%2LI%2<Z8$gw=?FrXBO+KMs7RykP`B4-TOVIY
z{#$=r@SLXeo$@3SfcM~&a90t>V~AY~Q@}E2Zlo#RU>@rh9Ct)k2ACV_l*p_`KF<tm
z)3->bag}*iQ)XK9*;IwGj)siR`V<Ikm{cEgR3^)fjCeM89XyFBgupY}v)nzjWOSx6
z3LZ|bHJWk*Zlsxj0uC5TTtp3#kmi{8o_UT-4+{U37D<DS$tqb_Za;WLBY$xHesSLX
zkWrr$VEiEWBxakDwWUkFB~tO5NIz6*&xn=yxWtr|Lwtpynx2Rv6HOT_#3vK;_-sS=
zPyg-fho3rU8N!NxPf}Z2@lrNZ7Z}Uxo+`bUV*c4O_1eK4yUi4aI=}J^Sr1YE_4V<J
z;t>rgGXbx;`RMCMDzCgyA5}o}r*w{6diOE{ewFMq;LEh<g`mlalR`O*h*?*%K0+X?
zx(hqzUq!Ad&rBRh%I8dO=jIuMAUYdXb39-tAu#EK|9q44eCWt@F?rF|)>a{Euu-##
z=2F2!s1+a9F;Q@$U^>VmSMk%r<Gq;uh|Z3Vyu{C|lz`yghf)0hohBD%wZ{BK$2;sK
zT-Yff!Ad?RC3Rn3(<a))`&)_$3x7K_qOtb%ZQVL^Z9t!f+ay1K{v2=A<ti(`=kAHv
z;0!v`EmXk!TxXUM4RhI5S6P2y?m=%oo=QC_qtIRbneRV?3OX=M-Vi0HLLl}U>TIQ8
z2qa<IK7*)>kFvVtr}OwTko57zcnLENAIu9EaFzg+tu#hQ0-_-~up{P(w6b{>2*02k
zhYf8hfi~r1LK$(IwkMs;P*Y3GhT~vyi?PedIU_`GU$Z|ROri9>U*Ed+7?Nc}Y44oQ
ze|DvjIgXmQ9epGh+-4;35?{MGUb28rPb<e<r(G*gN}h!oY1(N9?bD2y*9CJ3kaBG^
z%WLeg`)WPwIA8<KgCf8g26_<K@+fPIo?+=`J$czslwcHII-#Y=R4x8c_V&ra90yTr
z#Z-`5Z0A*37f7*t$dv#B{lZ)y2zXhbz1BgKj#5X9w?Ti^+%0p1iCBp%@LOX)Yg@d9
zeof|0?-)Z|qN9DD&dZjsHvBC`Fc28)$n&Lsza>inioiBF^G~+acVJj2Xfdp`L_#O=
zs>}wL{UAUjM1l-7%u%Pn&)ecC?{Syk7U-1s4$4f1o(cHeO!7XNU|(|dQ1GJ~VVMJp
zFBWJwOMj0%f<rt4%t)LpjWd0xbWp(QTHlH_|LD4soGptz3;X9zL4tgKE7n~`Z7-dl
z(ji0Qa3yu}(Pm5OesbVR-+q57k^V$TYR;`r@*{uuSW<x2{^i;LK5Y-!n-FRej|F7F
zF~=_U_UGW&VY_i2y>+R33~~3>$gfeGA@{i+D)`mZxg1*QhB;pOBP{wnOr=sxdk6`^
zJe3nQ`)_~N>+x$ucv5|p{{FUu0b}6Twp%8EB=l;a7jb*6DJa;`s26c23Nm)KQL1W$
zti>~|`{1Zk<2t%CHHPOj4G%YMyuI~v1@pMaTC`}Gd(Y3^TpB7hSfCu>mS|MhLYiV5
z)%)e7LwhMzL$s)Z3>vIKRlpEkYMcon09(?u^Kwi|Fx)#4jPA}5vSdvGTly0MoF+_k
zE2JJ;(o5HsKCZgWShY)x7l5^3gH2pm8UhFL=~M)RG6g3+@>D|QgVZ;N?xd~LZ=Ly8
zJ&hW96o*{e1ggY)gf;%G7*yc`6^mMInwe&$kN?ac{R?YfE~7iIl8)SriebFeXKOoT
z+Ukl&W;$lK)XT`q=(4MP=;}BvI9O_PqvT{&3)0teY+ZD%t6njt<8!)s{S;3uVbnY7
zqV`9uylHn3XBHL{duuN(Qa7dr3@W0DXpfbT(Tn5#+=!VVfFq<!QCgNO0va3sH0hiR
zK3?{#06`j$5Qz@8Hdc1wVRSrKft<owY|@ZY$Tf$J{;vC7LtBQEfV7MJ+x{+sI7}rn
zCL@`%<+F!uefzU$DhWP4QOO8#gtWP<At8+I*hQu&_yxxkt;W3NndRlSf@D7P>Xn@~
zR(Btuc#$@5rB<~bYNa38Dj;{?9$?G*D35urZ8enC_$O}}Grd-cy%$F`Fmeg5(0Qw>
zitb=gy=`sN?^E@NHDn7al;pLX1zcU!2Bi*|LbZWN#=+A)DPF$g`1<piU{8x!T*yIT
zE6&We5G{?Rb4iYZ1$`t{^uPJJ?m63U`n0!lqoZ*kIY$vS_v?G<M&M^u_sPjt$QSbX
zS{1Zd@E?DCW&S!5pYF-}vTXbtt<(ofy<J`5SIZ_;aYP$vmgn&{^<le@8!fE9h5b40
z-n@>W6F5KJm|S_`+bgqyi~rk$d)DOF<4>hBpB5YhGvb*%Iq$}ncG)6|>$(WL_`d1+
zQ2F4?*|(Qa3Vug>br=J8c7KG~$?DbYPREwt0hJmsFYpRSQ(~hnm_bMT<Y2;BDmJ%H
z&(`o|Mm+HSNmGV>!UyUB`Rc!FXEk1%W<m+=qxIRE1d!zPdN7=PrUsMeP|?hj5ZhvK
zYHJe5ur6y+d%ltL?S+)g$`=!p@+FzwqpM&%KspvRwe3}N{ysig<1r$?1~Z|+H>JR#
zV7w7qYq-61MMhBQkFb<KD3H#a=O?(r*RB+5S~3NN>%T7XocrlgJ2M{${}S+N0GWRf
z6sPSyYVBnE>JzpX+<@P?tzo!4(U&KG(Vz%<^2Fxd9+E2J!QuE}_M@(cuSb)ThDKsY
z{990a_3Ze4(amkf!T#flm{82U1|7}x1Ak!o`76a-LK<Wg0B&mcp|t}uacE~@(yF&I
z^TDz^;DMqSI%Y*y#?gKppq|bNy(a$_T<?q7<xGik4U46r0W_SDYG+nqInq~^Cl)Y+
z-s{xx9U`79di7?{E#<HxgnDAU4JD^{JgT{_U+zpZG#j}wWOjh*N(Iw`xoFpz4K)W$
z>F7ptK=0}Wp~92&(1K(~d`|;)wN3%MyT^2uw<3AD2v8cN#u7nN574N2z+I^>U^AM=
zf}yh|!9t1ARrks?t9;A9>H_+s(ZpKr{sd6MX7`d_rIh=Yf9VuHHa*IekvNk>0p<X)
zmb)HodMQRlw3M`~Kat$7nI?(Rmott?4To>_oytTKb|UF?7Zt~dR?Vjll@)pPEa+q)
z4cXJPLQcofk6vLCF;{mZmA$RgDM}&tjou4Ym>T<^9)ABC#rb=&j!)$#-S#%Hsoujw
zl-h*)k#d{sN{c_es|<0MEMBujtxr;)tbg>hS1bfTo3@TiF(_9HQm#Zm(e%KPrrI)6
zHW`aE(Oe&6qmkB5jL##iG!E0tEJILYncEx`iu}5x!#{p(Zf<`6KHfb}eq*g}YYA?=
z<-Yf`Ksl^y>r9BSBnOMbWplyTfS}$aFgXQMGFMyuiCgr`@oy=3zz>Q)r4KPL)FIQE
zD15&Fn%Hb31>58W<=dxL=FPZuH@{5l8AAj-0$>-(<b|U3O@a6YvwG1v3j6AjE&I>>
zqJ75+4+6h*Cq`xfjsA@y;2TF>sv-^IO3PZbkkBz1!YWFQK{mW-!qDp7L`BMzi>uYs
z?kXo0RJ;lOj@hmMjAFCRVKD{1dqLl&{_aC@N;USjXGLg|ma=)QURdmiW=&Qt%Z>EY
z*uyr@q`xMBVcM(2(78uHlDLNTmfJnv)wj<eT`9kH5Tdmn>y#Ua@!Np(Ol|04sS)MY
zYEa_XEVwgOx;j_BTtvLA3<2}=yRu?~Jgw1c-QuIGL`$J=MidiMST@DBokI8J<G|EW
z^Pv54X721^fo5N1(+WQF&6c0<|DerA21`5$_)k)n`NvyECJD!h1rvK@oXc#{hVZr|
z>gdN24u`83j2nvhsm<~uBsVZyj~HsHt!{&0afNY=)D&8hy{u}dft4LjXUXw!f3w4D
zknT<j?kR|56gV~y0_=a%DZeo#i5l5^TSwaqu4>dI%+k=q)3bh5zW-LRCLv{iESSv*
z6+D)E@OH<$y&YVuw#K@$bKchYN}|><I;W(OfVhAF1SdQc?&)q=)+Bsse$p=5Yx{|L
zT;dzKjNzn@^H+>ymoN~$P~*c{0^|blk`i+=5D`LA;y`ps;(aKZpXx(WUXfHhf3$E}
zgDL@}oHXTK*=(;r7J24#kbI?PD;XA-0=9c#=DAjs4U7KL1@&?>M3D2=z#qL_%HN@n
zkuQBt$UT#hIV_OC69?&d>)6L+7z5zAN3F7b<0E0I;{V9U0oc@c1fjfrDAy&IoBFWB
zb#@3Pippv2=Y&I}j2ZIa@FX2Bau8*3439VvuKkGz%0+1d<AOZa@8=+c9UicfJHIh3
zw{vzIeMmziLa|^`oJ!V)k(<Gv3fgtzC`b}JjBB=zP3%|Ct{4&p63h6>D@ld2N^{jV
z7>7GB@bVB6-!6|(%;z%K<o)o;7Nd#8NR+yAAfg7pUD8lpaDhY4Vq%i#1Evs&hWh$=
zc*BMtP07iR29rjXFI|ld&CgadbDNG>+uD}XJS_Y`Qiadm^NmvurT%?*_YMNXU+mXk
ze;bwJo^5fPEmAT&w{u%$=%BNln-8Gyka%eBI{q@bHi<o~NWOS>LZVe?$Q8Eaj*_M2
z`Km{vrr|3%P3wV|&^RPanX?v=NPIcKgX!GVz~ZU*oDv<LHU|f@!>O@zuU00L6fg1B
z$?^Um9jrn%eW1D<y$XHU8BC5Wvtn%A%(h+{$66;yp7<?|@j1&1+>|ihIJrzR8qf56
zTJCL$5Ww!?O&FDxyjC>vrOMu@KcD;*REgDIQdJTQ2q+dEfi%j!k=avac~k%heDj0i
zrFv23J5}-Ms=+q#_)@Z5nqeL1_wQa7at?D%_O{>PdklorC9%vS{mrTcYdR5{0#F#j
zL{g(yDxTXq!W}{>Tk)%+F$>t+KKNe&8#QTye6#}g6X23FmiY+Ns5XZZ-WqvucXszz
z#RQQvZD6Si)O!j-H(rs`00<-t2-Jx2q~=O4S=gKvuix*`O3QJb33tbM5Hi{MNvqR;
zo5P<Io7zeP@|!{LL;~&sAXSG|sb<c6pi+YG5&g_IHqS``WDV_%;*vAnQ|1S5`wdq?
z;5ndWV5(CrtP3H96GX9?iorl+kTw&S>k10GP7D51^34F#PG<(a+~o&0^i*3MSSe;o
z98TysM9EO=>CPL=By(4Y-8=xe8}so9BcxUj2-T-=#t_pHk|(xv4`S0mE!?yO+^#gD
zxgPdrG@{zHE{VD@8_nME9qR;mK3$5&W6RyPPfy?8+dx0IL5Yc(<_{$>bK6o@FCAZD
zx>{xq_aXL=n|W9TjT&lAQeqgYR(`c5C0%6(+gXb9W|Gl}2n+iu$n2cw)X&%BXXOMQ
z|C+FUGB25Y5fKrAASvuOiDH8ZcV7dy816TJGHvX5p0>!@GTfi<ea?gyDrZ^8tuJK5
zT&O#e-uqkrhq?;JFO0#fiSI(rNHU6ao}IHow0JY?Nkr)j=1K6Ax2-!8F=nvRKHxyW
zTh=hoFXW>ZYrBwwSL$=Gr`V743|s8vcxeu=Z@XsFd%1BbZpX3Sn;$`?oaVbhMfR~;
zcq$)8Ufs4n5C<w1=hF{rl{x$LD{YCDb06tYe>IH?Se18Wc<+s$6cfa;W@a?NCTgV-
z{7oc(rsq1u*f{o%pn918805G(?k7`QmtiOsy~%2A{(NB=yN|`~@8jHz6W=v&G7}Nv
za;A`uRGdg#{G&##Zo`<QC?xU1u5U{6S@W30v2dIEBxyo?k1w+5I&o0`X{_E1JwqT`
z3$@+R@hpRr7-TQzr8dg6ufagOrkKe|%jGF@h%HzRi39Z0E`Kz!2O<!$3QOSF=aC+#
zJ#}speMdg#dyX&wCn+dfhJOXq$elB$M)gePIS{hW%H$FquJcfiM}xQo+x}P;_;NaT
zsRjrVIAep``T`&{sW-{+5Yg!^{Kkih#+Lfe$cI^lW~ZM76(jS5Om>6r7q-qI&q@*)
zg>&o6w_RYV$z&O-LMP}UmIY=G(-T@;gwncFNSs+w5DU6epcZ;`G#$_b=ZFruJ`MX4
zaT_0BD3(Z~vwUdChiAnQMpErlBmjW}>9A*{<L0t;QDCeoHxZODnM>2o2}O3-&~DN$
zqMOo%E^QC6wN+DcIPwD}BwEc^zKI<wb@}ikSZT52k8x+68y!M^zhl-v@!1noLAM9c
zdNuw;sH+o$*3xX<qq7RI`L*iY&x&Qp`<(*0gEmxA<GsR2+hlKVZ?F&R>rnt2o;m&=
z`4Bo?>)hB7%J9J8u4zzp4pkMOREEFaX@|=-zso|@hA2B|xW?Z7wUf}tXMb}#1>ac0
z8`GEo!^TwF6f{O3wbBbxQd0hw%sttwS|oK@_HHssep)_m_g{g@FZ0f)rH!rifb8>E
z$v9_$+(_HV68{mfy7ueG(&m<qB9bQdhrMFKy(taaZ|&dG-b4W3$)HL-2!H$ji-gDa
zo2zkvjqpNH!8C9lsRkGn@aBS^ki^4l1j*p&msz7JV+m0K{~$&mpLLNvl|pksNo^?;
zScow#QciN?ehN>#|NFL&jcvvucj{HG>9gq#>dA&{e3}hYyS=5Ug4G>FMwy{sxYwFB
zbDj;eWFBmq50&mSqheSHsPG@qC;029%B-4x8vIl^Eio0bc=*XIDthyGS)<%i%4Kuo
zNM=Mtesb-*pAc@n!d7l}#4kX!KCvn#0Vj(`hXEk~nj0A1502Jj=%pN~0QhV2*S`6q
z+yBf!5l<0oT#at{i6rv(Od8I2XDU&slgUju(L}HFKPWsyWok%PtrEEz8q>jy(nC}B
z8{+6o4{wtI0Lp$wbQ(djwAm54#_eckY#co1YE$8$b;^FTYwAPMtmIB1KB@3@PSP$J
zjC+Nsa<F`5CI@ZmATY;+B#ciGS^zF6@PJf#y_1iLo3l+;bGo-&o{43JvjtZq=RL^G
zM?1ei^c7`8DUt6c(R$832<PfI(@JGFB<IkY{zO4rvqXyKq#qdo0`!32p(2hyPAly<
zYNkr9l5cQj{KOUYg6{9jIeQxK<*EF)hGtg@4_#@r?9%aT{n+BlM3XXVdr+|Kw33&#
zaB*?5zh5=M?YBAI-jc<qTZHFwY0Bw~QvnRt%}~0kMN5J_{7n3l_!$MHPix{|xR<+G
z)^#>4*fkpezOQdnKd`4+K!!Ld21Zb_Uh)lmso06#ojFhki#XooPWxo8_nCagC?v%P
ze_wUOAu<J$+SD%F!WwBkYZxM9jX(4a5p`o9Z$5#+1zcWSbTvwvm>z4C>x+rG2|2UK
zqhdZu>8>ou*v9ALDIBL8-770A_wM={!M#?Vin+T;YT(Xl<v9K5dnLGArjlcLl+ADj
zJ#;x>-<AGn14We$<6d=ZWG$S7nxtRK=#@1iLl+DRlS$Cd7=B3%)04h$_QnoM98J&F
z9nW2gaaeQ)a?V~D+r}VyxXEHE?!H1%D$!2xurz66io5WpmCG4(+XNu74UiveZb<*z
z%R1*Eh>|ll4XOpMU^YyqrSsw<e*TS8AO;S2PwkI<<KdE`%9zAF77r(H&Jv(yDbq;J
zF+Y#O-bDpy&ZE#8f1h4*9O>3Zd8Yxx91<qFM5|lB4o2!LaS}f##sSS`5)x3u7I`(u
zGR6!V06-`{@}y1(FdZ*z10XKxZGMK<fAY1X+F3_~i+gTK%yY6wmX?Vdjs^-tVBa{I
z-`Vx2$z~UjT-b%~_g?mm`M!1h7PlF4AmN&F{m^n&!LiU5aNHoRv!{_6Pwr@w4B+VJ
zpyVE7B#~}=`4YgXlhG0Q;Rni{f$i{0_PNy}77r-A7y>YhK4ZazJ)b!R-EVkri^NS4
zfaY?-QC>g#ZVl^=I&?>kEFs(R{266DMDZDUy{!SCrEmZaaxQWp-54Ko6KPt3JfYVG
z07MK8+*Hlwq#IfkbVOPem`Gftlo7LyW3hkXFn`LOdH&fvt49b+RLBq3=A#vV<~QeM
z-_?QE_-FXW{|-&sny9BHwBNYWG)QWuRG0k9p4eV}^V<&}!|-i@P2vpRy!q8D=LpS(
z)q9ZFfWG+QFIj40-_T$qjt{x=k^=L83L~=-i!1=S1ET^1dq+P+yd_exb<@&iudrPL
z^7Pr{boSe~boVw;6E)W=0srMxfnXf7VTk3#L07^dv$r;w>H&EZ9mn3XIRoJx;m4GW
zaYP=BY=D}&E+kt^eZAm`3?jQynlu5^HL+1C{qWfn{}z~CbvaW%y^XYxD~u;G9{7&f
zFo~Bz9{HU0A04e@LZPi~dRj_)MOkWst>1V#P=K9yqSFn<6>OVs$o(A6d<=5awzG)`
ze5n_?V22qba%6DeSM2{bm4uF6*|@~>m!I|WI7C@BjZvIQUp9!~Y1nD4_?N_6PPj#J
zB*(_oMxOhBUV`}2RL*FS8X~Ss6AahPQ!v(f82OEmtp}g^Np2*M2jv`NWQ@)2kp4%5
zs0D>!0z%IR26U?d>!^_gn!Kv7{e<ywY_Jbb$57C5k-nU>a)qspa<G7s?J*AUKB+qh
zz|$v%)CP!G&F6e`SxrNis?OC8tPukY;`-dpzmpLVlzIFGrnQ~RmpI5Lt5fWb|F)P5
z+VWL?+w)!2S5)YtXqMcEMVk9<tdo5MsuQaxl%aFNTN-o$TPF@gS~!38UcSE|ZhWU`
zwh?}GAq*}1ZAakTqJGoskm5qauk|FOku0Y4w<go)z;+a2&Ri0#l(?R7dD@x|PW+5R
zNJuE({8(5}^|bf6P{BWs$JEX-%@7K86oEt&r}jW3uz*#3gqW5_a+nkj25_OsfVnYf
zQ#1r%lfOKdWm=8iaI+sSxoDAoROj)0XF_-u|3SjbaR5l+;l+gnV99-J<@}Xf!$!L+
zQ=5n~GXaee=V*{;FI&pfnXbJ(9uM{D8pJhX_+!*xd=E&`fw{T)1BU8MYrVjaU0rL(
zSJ~OwRg1O3?w0;t-#<3GzD4~8K|gL4Blg@R@G~UV44>X0=d^RrRB;p`6r7cqY{A>=
z@G81FjoHoR;7b7E8ijbpd-*l9v5<)Z{a4jH7DWpFF3!%+c(}6S%5(3h++8ZJxCNNh
ziWlnGy3q5#v65$|^xGjEe5h4U@i_xeI(n9VoziTUi7;6%*(%nmM#VcdI3How;De5v
z3Qt&ckMyqZ5oKuz<0Ta$$o`+zY3z(6@JoZzlcKl)l~2ki&5eda_>;AbZGm=m)xgVl
zlm*<0;snO+zBr^UIxVdpjC%rwaTovQnoa5%3sFCnEViF{;#ugAGd=FP%H=B?e`RL!
zrd@RDzXo5s#LsW{P+4iMQ={?y3zMxv4+DZUx5EN{{2Fhe9DG^tKT`rv#KW68$OwoC
z6pn*~^zCvx8;U{=nM_AbUB^KLc#G?YVF5A<C?%q1;!7a`nAEle!c0hphG_sa2$#Ym
z^1>ayRw=4nZ$4p?mq?rVPukLKDM58dpDd5pTs>RYiT((LI)==!luh0GRNVpno3?(P
zB;}oy{q=3FI=Jh#0y_C^*_>kj^PpsrfbJGNw#8cljxG0l#d;Ej72f`~%DzGuO#1Hl
zG|zWWnRzBx9hVZ#ndgkSg)Hob-exfxpM}f$HaLcRtA5(E=x%H7{zVh;a8y37)>yNw
z1>j%mhnJ;L&SP-Q)eIQ{8-z5y<QbzESJ?^FzkpU@<)lc|W#J}dzlv6=<Lbst4&R~A
zi9e<Xf>L6mH*dds$+4y^IXZjL(Y&#eV5gRqkr9mee5>(=Bwcl)1*4|KPTg*0WF6@5
zp4Z7vN!oz`FUG*Yr;SORl4qN}Au{<5jg3EV?{rW;Pnj4j$$lhL9T4M3RsDODL2sc(
zH~-qGH$zTb%AQ9KZfk1c++o@FnX5)^q8MsTryT(KN2~%c2?wry#RrH6Q9n>HLM*SX
ztzBHSxW{}oDl(L>%`^LAY&vLGS5Sk0eo6ZL{$&}gwJR3r@M3d7HbDJr@-C?==2l_D
z7Su;8jCV((Dwv;`8x~e}1Zq({zP`1$<Sg6I7BQ&oE_?o+oXbG_x${=7Xtw8$Ac)v)
zF_?TyCTJ`N`k)?++%F9LKc?O~s>%0{{~t(9Kp2Pu3X%#6h;(;@q|zM{($WmLK{}*E
zT4_n?22r{jVK7QM1`HTA_S^gO`~J@FyR$!bcJ6cU>~&q&>%Oj6Jf6x7{8#!AkaBix
zuvgwk*>@7TvowMf77S7ERK7PhG!(KE1bdQ_vrcC;Qd53ga3sPVv2a4yH}nY;GgvzU
zmI5q)qyc+33wMWP(AP<k6490uJFzG69^QX_sJArd>vpZ(1MXQ?l=EOIxw7Ez9K*XF
zJT&p;pVDMFU+Q>15>Dji&s@Cuy1h>O6I33@!yCf#Ac_&!j!03EJ@b)Nf~L;P(8j;I
zZMd~B)G67zb(yiE#E6c%*E5WLrBMcWKqBS4NJCrI&jeh9voI?39=#*x%9yK_GjX2j
zXw*hF)4?ID@>^L8O$0#{rGcOv>HSox4~rM^)+9khD$EdvW2Pm(`Ng9z{0fMq8e8#X
zn1tH}!rPH=oc2S%#m@x_CFI$WF;T0=dd61*DOPDvCZXwKxdSP_8pr6h`i}M^fM&=5
zJyMOfp*TX-iMM&vSS=#Ioye2m1U;9RftcgBXsd|uVC%j29%~BSpRjA^f97-uYnuPc
zTW;Re{HE|It*C|M9Fau*aJZtPs^yxykB`4U^P}|C%L$oh9WE#Em$PJx^RmMJGa}?n
zD%mNf;S7s7vVW%HbZ|j(YQt2cz_lgveR)Fonywh$A`g>Vh8LzD-L=f=fORtDc2i?Y
zPkPzEm7RVB2Nk2mOYlH_nbc1$H9it`B}VQ+p1^~#+Oa(e6f~b-h=oumGfoAs>w=_S
z%1&(^dA`kDe`%#p4QksAj)X|VIPK3}Ea%dC9|$l1h2SRE%!N}|a&z+o-gL4vnREfF
z`<f2yhbj5r!!ZT=!cx<z%HH*dhgOkP>B5wW9|e-rOkY;kxioA`*YkrZAQbtG^2YU!
zY~N~jI_e%jWKkIz_?@I})W8<WCWjTq3HRzV%7W(E`sNDO+b|V~|H;Zu0x{lqe;aF^
zU4uPX>3Z6e$h(B~>dW>47pIOw2V9Ql9d-P(OeDV}_4Q`2uNN2h43krM2yiUHb%P%D
z{Dwv*m3<o`@71{FVs&&zhdzm9%;@~6Ppf^E9NA#gXGub1mGC@FU(a)MXvQN5e~Jfp
zKbDu^i_LTCLH=LwCi<B3Npl*yos-Aky?o0gn_VxAM<Zv;$rRe^4Ez>X|6XNP+VDxH
zZYiB2W6@R;bw!1qSlSY38mD@w%-Fd5hu7mNysyJTiit@UwD_G6q?{U_N@_j`{=CWZ
zl_GM=9*px;-j7F)IHcQEh$GsvS;(07^=l4}QuUGsT`L)uAs#|byA;dy(&n@`CDgL(
zf9WY>{)GJ^O6?bs+XmuXoMOQ&Gn>{tR^1_lWOF*cr~d{Y_Q!t_XcBskXm_m`iZ1^s
z@<dI+$hf7o>V4^p#<)P9XY)E4mXxb(xRKd!O=hv&pW?boj%j%+oIMDnX2AWIjkkR9
z4<b9y*s$mJ4v%@1{;qcZ74n_o)>3hPXC6E7l12s>tV2^4o>2zCaLKh8&sF8<U%!S5
zG_{N#A_pK}xMAK+t<rOTW*|rc<-2ID9u`on6t}R$_XU`PB6*yl+f0)8_b9`s`4P?s
zDH2cjzQN=+dq3uekT3=xF=PlMGoB|asnX#|MY3^S%jdrAex3RLA<j2`YZBk>PkC<Z
zEHdCtM}hLahpcf}?bJr?w<%f_*3wcxC5#s8P_-d9`(+jTwPkx)U9#jsI|qk9ZU3d7
zKCJU)jOu?Kc*4`cWTtPdPy2ey_o&@{uAdsOC*gy=pCu08mCiDapuIs<<mZ_;YWPnI
zi=WbN@nyu7t1O%(Y}2??B$QPNDR8DHq@;u|)O+^OK?cGA$D^HX?EgM|V|D0d-pNis
zY4=1rw2+f{UIXh@Wn^S*?t4bQ#I}&LeP4S@%trUETx?>-7R=Ze1YV4T)Z=!_8ygv>
zFyPW`4DgwRC%q73dokBu{A~0|vad^21*1zPh(gL_)W&ns*sQu-`yCZ`6t2gH!K1`9
z)Io+?|EwdBIr~9H98sy1nL16Kqm(ppDk(&>UCU%dO`7J-iMs9(8}ArjOd<H_mf|Jl
zSWT}FYrG%Bf+sEW5vD6m$iS|sNC_{5#J;i3V3kwzng8{YiC-r7(3<Uq$9)0sGxsdm
zk{30_N0Dfl`8#0wL)X9)RcDwa4#-2T!z$Xi1_TD~S=_Uvt_@XCCWz)JpTS}L6rG)4
z*9Ua~TUFq2P~b{$ecFGklJHeNXuw*+)`p?fYLtPSoA@Wo7q9(SLi>arIE?p+qfaYX
z!Zb&R792l=#$XVcNBJt^{_8t6!|`-L4}bjEw<WWV9Nd(?Hncd(;OFitBaC>`<~F3M
z*z^J*z*U>C_pf8g)|A_~HhiGIr8|7PSWJA5)bMxsfZblbvN&JrKc#^v*;4(#H05XX
zaW9MBN3>QEeh_B2{h*_(qoe%xWe>Ggsf^vLXItdx|JJR>q&94~{$|w6ucrRde56VD
z1F5G_Z+ovdoHRGM&ZX=#JEX*wJc?<)JPJ>MC+crG>@+`cKk9OJpz1MXVKV>O7ln@}
z`wyEws1<Y1WoFLzevLSXhJ+=TJ7g6oKN9Bo#t0%ILD7j{+>r(TWLrwrGj7zhPAJV*
z+t+0L>|XWCp-A{SInJcSPsSOnv)ZE(!F2-Jr3MXD;2TFZtx)nm((1_$4UYWw1x}Lz
zge5Aqc7`|=Bo!MyQe2zSJ^LxF7$WtWo91R$g5<yPdzLvTUqFQxcV-8#9G!5Ta4ZSb
zYxM^ZR4T|h@s(dG%$&<k#_^I;!ixVH(&Yn>q!BqNxG=qp9Md*dQ-+GiAZ^WOD07EJ
zhQ-FzY^8bMiXKHu5Yp~#2iT0V^a0cb*JgGa7FJ4*axHHs_<f?e4OGaJaYEE)s=3A7
zSwoAKdd8j#l4Lo*bZXGg#1BfA5&xb29-@eGMW&}%?gcW7#s6;s1&GXr?QORnyiO}A
zo}z2QOnbHM7`*2tC9f<xj5DpdDg1257Fk+^@DpC{HsW$gje<QVd6L@+s5}>qh%_W%
zvj*`Q6Y#0wh>+{dp7oq;;GOg9nKeLtcL+NKd=n^m1x(t$GftVy<}I7<W7}XUxtZ+v
zEzA9Co%*L^2`FfK)MbCbo-yKnmhQoWXp4nh)uVCwlgTYfJn8mRWDf44uC#|%x8eew
zu_je7Xk5O!fIMBKZXX7UbN6?$R_6^_+U=q>TLg;_+ZWHDO%8?kye$6L!5<D9G%T0M
zEd7x*&1DE1$l<BuVb0LW)b9<xAC-c9D2t66^+kq;vo*9*g$gT)Kl!4bC(+ST)08JX
zn@IY7u2y=f*ctC{{q(6p@mozg9GxGbg}FHcLH19}aX|HDuU<X}?5R{u7(K*l$eFF}
zMTgx*?_h%B(&a5OZyDX#afguB2mhPVvwr9>%yC|4fxMm9FP=&!I1NLf?Z3FgCqa<|
zh7Gd=<qV#$)MqLP`}i;@_#^q6_V)Lb;to15AHOFZF)XOn*jvGo3M<i$MpHkX6929|
z7_Qm#l_Ex}(E}KGeUPO#OwG*jSbkdM)<RFxOIHUMJoC7grr=Z3+%M)TzsukNXgKjb
z^&hW>7d%9=Qt1|}By^;*>#z+w|DEs@bVhLocTC8oSc&i3hHETR;Hm$8Jo<{1>p^cR
z=Y6LZz9lKak-dMij-oI*b&g#Ui+6Ih>+6+4<8`9g*$rUVYm0cw!Ia3@h|_<Avm{~b
z^#8i#(L+XOcST9;Gs(%B%q>gw`o9nE&`8+Q<oN$Cm0kKj1G@BV{r@<-tqVv~OF<s&
zKchu7kJA0mAGC<~{9m^u6IbMVIkT4J5%}Lv1K0=N?-5ng(<IEY)xge?5!_ww;)wnz
zMJc%<zI{K<>S&^~8$T_XIH0D6uM-m|s<B|wM&!Or^>78fVRpQe9_|v*osMYjX>5U2
zh@fj=M?z1ibi)^}UJE-e1>BfqEmRfVzH9zAw$~63?EF!jc12BW^p1PRrLrDf{>otm
zteX|P?hGuvySB|F9JbI~^le*v=`(ro_B<M+rt=f;t{YkUuI+(DX2m$#78G9vhx4)*
zl2F2L9Q)0Yksy?eqcw)s!4Glcym)d|=e+%{qisCT7gK+*Vhh)Ba$Z7B#ybKeq#P#J
zmy)n;lRFgW(pf<~N9V1c1U8M^50EWg<Yuj)PG$`Y{WgeoPi@ghi3RkX=LIF@7t`m9
z%k<5gvShKo+f?I?yPRETsZc+5Xu4lIfTV{9bZR(aj!#6LeNlu*3t*<*9<^%d=nA}b
za2mSey#bqrpX9bFwj*b@pX6OfAb_)PO|?MFUvyy}bY5T1+OOY5Y#&(V1@_ruX6Vr!
zE`EQ1AN?T=|Fa*sqvkNm=(ACk1wR>sX}ri-Ms+@Q@kewd)1}V_<Y?UeAu|><7IMAH
z?Y=ANS`Vp1X74A{anku7BCB>Dpy>67P?GzC2ryMn(s+V6<yOG<-@xNTNx#E&Y^5*>
zK^QaYZ+iQ7UMK3?I6#Ok#r%1j%7sxXe-zm5TYcx-zpW_hvVH3<nCIQ#R1ITo0#LqP
z7rW!_e*IRi<#Fikqjt&CIEiIs#EM6$zJz~ouH|tUSKxf#Q?s9zWMa2v%f#YYlBZp_
z%gS5Rn6uygl?N?*nX)lAUDulk`kai*(>t;ypN+$k#g}UhfF$&&Dgbj;J-<(exqZ;v
zAn7|ChTx||$6)q{?xr^5#p-{aT~aMK&(ZQOp)*|qmOST8Py+$@i(Eb@(EXf}+#L75
z<Z54Cw*_zXhOWp(rtNw9-LzQ;A{Dbn-X<n<QIBy;@kNj4w(XxHZimT$<E^HSb56nK
z26&*;rwxF(`hw8x_?oQqdY2f@h1qmgUyo=4Oa$)!J$+n-KFLxwy@_#IxY^0`nMNE4
z&9A(bxSE_wG{Z#08zfJ1%lx*9>2rbIVTmX7p$5=52aInJRMbWi=x{{q-h?3fB$2uL
zx-24KJ8$K#lWYkauM@oq+Y(=HLL+UL$28in0F5!gI}e2nytp}4KH%j=po);_OJOkt
zSy{n~AB^j6Wd(E76tVC{^6dD4*zEdFu-&}3!heHa9P;ckZ>4z(x>tVGx$!}Rt?DjV
z(Es2eBH(KJI%fQ88gORCj!`u0&20;&+>s1RkbGt&w)TPki+1PQRh!?+t?dSJZUAt*
zJ-)kXc0$}#RoS_r*M;cnIEn*c&-i08=lxZ~X}K3e@w)U2zI&2JqWiHtoi`>ff%D5&
zYlML}H0b4(J5*K{>$A%qGyj|5TYJm~TR+V7IxNs)d~KL(1@_b!#q6?tpgzuecWZQk
zs%qb>J<yW89dkPI??CTsccE<Q0bu8)<Vov1X3X2z`_dP18}Ikdb?~bxh>CoR@I#h&
zZnzQVoy8zHH|u)FBoT<4o0yTCyjy_mt{u2ElN|M>N_6k{LXNI@&*oNRLT=ujry)dB
z&QZo*3!fTK3(cB8%mr-V@-`rmg3Ik2UY%>Kz!|PL>PiA(+r^dFBLVOtUvJ-N{d4z*
z)6ZBRtj6TtS>Nts$@Qz#0Q%v}UN+DLw;6f{eer8L?@Fj;KQ2CCeM9m0o>AaQo9Qt#
zkmuZNL8f8I=`V72E7`X3r`o^K1*d4&nrKWI?+iZZsB5ChFNyhGBI$W$B6-yuGT(9B
zcV|TpwS&>A!Oy1<0Qw8G>!1s>g_59%%l6zm3C^(%y-PK_o;CAZSd_fo5$xExD7jmT
zm$VXEf*wyC1y2?qa=*~OY}EsfA2wm&b#Zq`E*<t$fe%oDNC4CAfarprRZyXuSe>Ua
z-mosxDM|kicWd)qTTOSXl_g}frbjuRRqK|D_dprq$K5~{LX6+2ZTs$gy!J6JRTe59
z9x%7+(lB^R=mO`0h5HApAW2=v{+=dE-i1kybs~GmMdf^Y1u;|f9lL1M60$Sh1Oqqo
zoi!S|V5j=}BPS2oT&Z^M%Gupp{CE3*8JhRM+r)p&D@Wd$i(H?btXu`t-y%<dccy1I
z)ptjV{JW10j7+Fj&u2YpUcWwqcdS?60H!x-*u{DdTt=s(kEc-{(;um@V)R1+=nGrJ
zg*xIn^=j@J2Zv<K@!Y#G`k+7KH`7Ll%k=rXk$IH!T?69oq-*gbAlb3I-G!P*w5*i?
z$Q$|(!Ma|#*EEFaF<#t(ML^W-dW6fjyG!B(g*mM5+RmgS!<#tjV0T8O>XF!;OnkQ}
z8~TfWxPppnT1j<1g*Pp?yM4cZA9FpFcYQN#daL2$w*kYf3iksWwr0?OXZT~D@Pfg?
zEG*>!fMTe+j^p8H`{T_BEQlJ`N0l3jx#u<MYql70vFaVa93lW9fR;I|0=fWVo5tv_
z$E@RMi#-=vX_gWk%Wm9qI}y!&Q!SRQp;0~7I-Lg;W4ZYc2w*O@EBtyD#;ZpX<J;(W
zHRC<)_kJx%HeLveAjm$-N#?xmB@6c%hpk*i^>Bb7E;omm>vgk$6SA)B#)yE?dGRf7
zGi1fE^HomzEigxqzHJiU9h5w7Drs{W+RFk?+AzcAF81h>mQ$MX&g+N3^$KvkZFYGo
zxUB)bNWVtrLO_>y^ZtLnJ9KT+Z|-qn)`w&6s!T>%H!8U>on~zTyX<+`>F#pd2DEyj
zCq*>^_6~S|U#dx-I7zxs=i2%oZhB)z%K_vA`uhNET;?Y4;$R<fy-W|C<hmP(F>CYt
zGu`R<z2kVg8uKH)6FEeG7lw(gz8OsKXbVMV`u=T*A96--<e@k9+Rx_UKA#W`dj6~O
z+v4+KaW+TjwmiQ(BygM_xY=9n2lb!XsNx(0GBNMGYJ7nMtMMndM_FX+?c3L<{zJ$>
z;OMXOvd2bcjK*!Wq7jJA<*s2{{Km-ia=7aZbGP(#b<hm;DCRKC%p7a`yQiJ#7MRQS
zelE&$yQ^*d#<$DkhGbL}YkUE!yB4hVA9tLpx2_4B-K21}EC${#H!XFYj@U|m@{g|a
zU0-Xky6IX!nI%O0G&^6{!;I9C{R1NM&Sv#;{rgus&O3)Y&-F~NWaImx&MnE#K4W!V
z<`%cXUAG}kL+OFAcU2RW)Q6a)60wih-=D4s-hD<~otxcVltZZuZ`^j>Gy-FnPUZwV
zH|O@Q5!ah}=*AA#6)5ZSaklNhrp}Y9di@o2ZkDqjD&p|L$~7nEga)RL2A;<x=h(hQ
zbt2CNoz9SxtViw0F;PjIXE#gLe$C0FZSVj+L=^Vg_&r46t)6%|qWPqFBRyc4t7Dze
zMSo`a9~84CXzG0%C3$h0cio5fUb!6GNEg5Pq4&sEM4{2FYioi^5;+)yLc?HicMrt;
z?O8dncO8%35Z=CKQ3U?L3-V42H*fCFoe(Bn>%x+#r<ikgT?th~G#q{8oofrHKM(!0
z&nmIdz9NMoMRp>lvrz8+j}dK~qfHNTuWzfn0?FOhuB-iJE^55nHyY2H5x%|E0jTo_
z?#G{bJ?FORp|cMgSH}4Da#2TH$SSWb&$iWiAbg7sG&M#K%u0A~2f%r~{<rsjwtg{D
z9-H&O?Q_fF&;NfXEiUwJWgPvx+_`N-U1spqQG_mvA+e?}nVTs|+tLPvt)G!nAi$#t
z#`~?Hg!iqtJ3+Iikd#{X7fVni81(D<DF`z0=RN~2GYFT7fiXYOsPHN&2owz7j|T`Q
zsR@I^IY2x?U|tNC9-x=-Vb|?NT1J=f*0qt&@+=&6x$8!4ZnGLL3jx(INY}BX<0O?+
zX?>q{4wd23K1haB`lWnck8U%Jvca$V_RZ)%gBSspr%Ww5EnaNWAo!tLd`mFHPiXdm
ztMXBjrKBU?O^jb@Yt>Q5b~YCh1Dp=u7Q6uwLSnfoLO~5+B6ghBorIq`FxeXdGFo_L
z^CGQ_)98@kN}C90N6@6mLmW{4Az74K1TKhi;zE?bPC~XIjS=2Squ=by17g!wvA1%9
ze{`yFv8!~ltQ)~<@%gelH6bylf|xqdPcujBuaDLd_m4;>^ZZa}s4fhkl4<B<&XWNl
zPplEpDP*pD69EA+F?|NV!AW2seuGo+Ez<EF1Bi(T7u$(j{W727Ggje?LBJzwO0zSP
zFFVR-U!4gLPYMDmd_nt0MYat5iyJrLXAp$=4X-(+1w;-vP6iapu_~nmmM-KVc9;(2
z;_Uk)pw2h$Y}^%SrJVy9(U>$hZtAlND87dw-s(55(BV+r?nvHa!d<#+R-?5fhd><j
z@sj5H3?CNMi{{jF?6+b`HwMMt92W#3-ABu+-v1Jodrg*lSLl~$@>$6%2Ml{@Q)I}c
z&DWHM60~zOAUI%h2p;8-S0iyG-meFQ0$j5R^h`|XoyWfza2WulQnwrOd?`HX(8v^!
z)kChJ6m3ut-h>><$D{zuxRHuTCfx6F+_*tX54ou!0kKL1FPZQtf7yX(-hd*Rm_b1m
zDNnu#^5~4kXm#-#Qy+gh&+sXFR0{@0@?mr!<XWIm8s}=oOr<8ebO=Zv<QygdqLkxP
zxCe4~^_)Gd7X;cLMl!mZ%qN=UVpz9o+G5(8V=({RY$HvDp2$~nz!n!7$mMV?D0BxR
z@rdqghJx@C@#OHr$b&)p;QJ?D@vq@yp1>I~!|n@3-3f>wpKm(|hpopbm4Fh&s69Fz
zWoPuKJKQ2F<551_bKK~lj0|}!VyDpxpKX={CsPxb4KV!5pj(P>8OtrQJ!JWgJ=LJi
zK@so_sOA1d@|l5YFD>ytxu!lOBY~Q_QJzIgW+4zS3T*jMy3s0!XLNLgqO4V4pZb-I
zUOGb=7T=4zZjwHO9Sf7&L)Vhn78!7Kf-nzp{)CeV0b@Sf+kBAzSDdd9ys#k9OI%*?
za{jpjP~t2l_rOY)d7YVAnHi^s!CZFm^8+;jzD1T{x%x^Zn;ibYXA%a({TCfn7UcIC
z<?PuEiA6cTmwhZxu{D=sz{9crQmQl?X`Hs{sEgxOuda19vnTcvh1g~1H-o(j!wJPv
z0$nP06GFoNDYl9YG=|6HvN7O>&MYa*5m6E;0q^)CxohVYgsFlip@{686T~)lg<#(<
zpG9TQMzTK$GxQ8JR~qnTyn0n6PH5*$z)5WY3BEr8$=Ci6I{^DI2s3DpT(b0)lY5LS
z#YEr%0+&+KkW)aIa47LUD+sjUOG#muPZ1~)UyeGeu_3U#Sf7}?R;i#2xXtM2I0dfi
zwyT%^W{ZMDnN&1|gevbpdYU|={EzU}|5r?)9<Aba7G9GZrdgCJO2dV4oMjx_(mDK^
zp$|!+fZ#2qxr4H%-hla{Dd6k3hrdtbe_YTm_!8p&=E)mPW^wa;kZ)<O2W{6G!IlnS
z-<MC<NtFc|lr!-V85tSM067Sp1i~T5VI&8CmvZD_>rwbqRU~cQ=}`dDa%E5L8l7o+
zr$^49#@)w8E*+ep{kZ#8!B=J}Jj&k+zsOmndZ-JuRpjbr5|!jYL{d%>5SE}VQ)hqa
zu=42<IMf59ZItV8jwpA@y;_WM31=WtVIi97V*yRti9Q)utQ)fgy14XPK8kR?=+K1~
zDj%-CaE;Y@RN(GdXAHu@DKfCVz4>AfoQxMI#Hu+yBwvX0f3M@*A4j(_$?_(YVJz&j
z3Gt-szyH;>-OMXjouV>j?at@&_A{>3mw&jRAO^WYEs#7VJAs^Qo7aWp*=4}fg269e
zbf_Y~`HC?!Lb|<xvNkl+=L)`HHYR+TXuen4a#ABqBP1laZZMZ0$_^~J<nR4NOB7ge
zCafWeEhh#%pf`1qg@yGq+;XyRrd7+(!Y=-0)X)!Aoj0je7MA3dI$*}jpN;&>bq)zn
z_WMX>C^sLs%2x6uDS--<Ij8(_%?YIKqA8A<KVfQIw&wZ(uBLP@5X7pzgU@a3x$@^o
z3L(4;iEvl%te*dk7HmzOT-?6M-(-^0MTm%j7l&D8qezY%e+J^5zkA%BBbugdW%+S}
z5X2yj$B^jM5t8^s%?JmFy@+4+JzF1>*4g*3YOD_X^RgaesgWuyOyszU0*AVS%1CG1
zZ>Q=>)&S#Y-Mz+sEeB5WOz+=49QjO%Jy^O4AsM}4ms?}maQ`#7Ag9eX*Q3lNh*JJ^
zG!r<fwMC+<QvlXx4pL(JNvyJzOBd3b%^c+39WgLL>Dx4#0_&Z3dPS88Zi0_+nkSF)
z5^fDZ<fRhtGrn;zpM6~%2JZU>zX33?{{{}g&PcNLF-SmpDmy+y{u9q!e61&WBb7{*
z3E<BRLAZ8J$N=c$cP3BEDy2D|unR-T862N#IWKUKD<?h%D-hDZNFK$$9_Y=(q4EP1
zI7KhJi9#ZOhOF9#-2L_wpB35lxQU5rOI81hp9qO!X8x|^$dM<<1_>~UwcYncd<P0s
zUB1i__#%ENBD&nrSCfS!uzO!9ahXfOLe=X~+p6;^bY)?0mF|-OaXON=KUFvfBU=r~
z1N?yYoL|H7sWvX`0zUhw!+eDT_)ph&%XUKqa_@CFr`FedcZ-UJ_Lp)PuYqfgyDV({
zOe2PARMb9AqrbeMa#nv{*>byR7&tX17c%3Cz=~>2e8rQ(nO>iYbC|DM^2P-G=`AX&
zfI~58MS5R8W$iTT3^V9?L;o|VnA*EqKY^T%tLq=EdHxiX*E_)Fl<K$`(t@(pW7o1v
z(&rqyk~4O7c`sfPIcb}QA|cN&|A-gI{R*qyQLoEH(Q(fvjoMh8#$`+acxvQO4zWx+
z*O}WR75UTq{6l38jqA_i92;>aieow2-j>$B(v%q;ZXUykk5Bens)v6`aHsXsWJR)Y
zkoGdVf)kY(Ulmc(P}tzPc~;&h*LwNLAR%QcGee6d6s)wUZ{;}3!7US8#UVsM`aH<4
z5)uUpB`FJK_{Nju^^*S8G;khtxF%z{$Q6J{bqP2uAHVB~FuOf!vZ>qs&fJgtM=|ee
zrzQ1~)yVgIgfEN==HUVR1>^p5A%Pp`^L4s@I|Gab=l*<G*G;*=QHzEcrorXYyx7f2
zeAnHc!;c2kvNw>=6PLRX&zeltg*n}zzx?%W1w)Uy!B|RgDHL#80dLD^FW^|+c~nUE
zgm1|Jt(eC+jCJL}(-3*pL*4MxN-wFHO7}IR6u8@t7AxhxCgW|<?1$d7vl7=028E!|
zy}g9=(6f$r>~Nj__@j<b_C>imzw!?nTY@4}v_rECf6Qh6H0b@@P(1tWGE!9)T9*H}
zwVmhST<v4viD2L%a6%@j_Vu<mWrr&$jHZC1*>BCVX~pGSBs!_<s2irY+E`<xpAWuA
zfERe!AXu`o9wF&-RYdPEc`YNPUQ%ZYfbzdKb>+8uG2$UQWH_nC2(1olpz+JiUQX`r
z9N@gm9~!Ae(7Syz9R?P#!YtD_xn3o?QFqx$h8N07rq$lS!QUFxisa$P6*>}-n|uBF
zl16|fPe!)?xlW~$gE*H>XwQK`LOq>YaU1HHP%~!r3c3Bds=bTvpk~{-1lStd<Wo!Y
z;y8kPx}xCsT-r|!O&(YB$!6iD2Hdg+dZFSryu){<`N+5W4Dul~AKC_$3*Ej#aKAGU
z<J2Ah{+0)PbnPwts@@=Dr533YXr1R$_=tstsVK4n+wYds23ss0)PHUEs>&p%&|X{b
zwN_&p-FGGyh_tp8WhisMbJKGfPyRNq!>OsUdzHGP*I<s{OFq;%atxnhqc)4eyh>ck
zcG6KtT|R5;yVJJm-gLe@!DzkClSAU|v6oJuB$h+qL0Vd!u7l|qm#Go(?=RDhiskc~
zvh9KyOTlT?KVDklJmv-wi@c7|;D3#p_nH=qL91Sw8Z)<2k+jbhNF%{)N^i2%2+%m>
z+%BSSM+1yEO~56v?2Wl+I`G05u(<!d)8{DX<hqRC%T&~cr1J?0%r_T%!h<R&rKB2B
zAK?O8o`+<#e^rOR#^~iBfzy6qs3Mu=3xy1L<!UTO&^hh`u|oEk{0wn-N9817{`~WC
z7y5{5`oc2*i~A8-*JTOSN`U14f#2DWJWiQbiBJ3G^ARDY;HoMjS-!0Cd@^NAmdjNW
zoyWhN54Re=H0*CQJg{uA#ngl?iz9Zot`&etOGld|F}1g8b1Kr2?s#5qy6K)ahLl@R
z9oTdD*cAnbPYiY5(g?eFgH$Y4k^>E#*~)~4!bIfMr{6lrW}JIIpkO4xX)H7b<EpSm
z)kK9-%N{d+OlGlv0eY5-Te{`0Hx~K5O5sbGp1o5&DLGQzOomQ6%!4hbR~_N{Gb$F?
z3+v8x3Aovmya76ExSs@UIGDkg452%Tths($1yt1)cQDM223Bd526581;+K29P=%Z)
z`3hWB->uRI-068hnRV~}vAc+2rpU;i1@3H}pU?-asrGqbj)~1K!^|%6p(g}>4=A*5
z$;<-Ip^`qn{qT+Xd0Ro3rT20azJOIb_YT?C^h>5sk>!Fe6qa!3uIp1(R&_Cd_jA9@
z&9Pa7%v^~81ty2e#NJsqfjlUAGD~KPE64B;T$!K~JM^qxC#L5iy^(=1^yIYnlR!-&
z^^<T`tGClE^unKKa{b*SYACI=+eWdP3EJ3=3)F3M#6!SI`eqc{VZ6>v58OS?8D$Dt
z`$?LcF}pI5I>%Q0HWK%^1A$o`Bojjqk(rHAAWz~0&Qvjo^c?RocLSXu4(TWPwl>>)
zWffYQY&=72e<GZX+Zw&xQ}LOR|KTu8u$Ft5ht8<>-^D897T=3XYCU^W+TtM}s?<w8
z#V78uGXZnyqO*$qJaz-ratznr7D!$f#0MN$u|9*|&gz-XafdH*a+2x^HVQN<VEc4U
zo0<T`QM#G$0z};w>CBFLtvjp0%75tWUC+<hu*NMJh>5BzSwWl=#Cjc_Y%OjD^pE5)
z`Z=Aqzbg32)Elb?OG*fmk!E<$eN8PF0*cn2us5n?@QahqCI;X!+63e+Zbcv>W^#&r
zok&V5!ghC!@~&)K1BdUZ(uU>Pb1eg1U3M=Ii_PmTJ9qT8T1@dNh30;Hs!hw}B3ot3
zeJ$J=?@^9$Hrx@dd{%tM_rAlV;V&lDUz*GgZ4oCkRChox>^{W9>k#k8C47&TQe(;;
zKD?xjsXRx%QU+j^@ZNnSbA<O90kOhf8SP5D8>e!cLX#I4s=p3*+Zd|I04G+Q$W&Tu
zOf|)W6md@udojwqoPHnwl<)`cY=xF(tfz^sZ&G2naP&mZ)J_gpm9dF|@#3R~rFID;
z;~np|96x~z_)GbXL7#-CUlFDb%`Wpi;JB@#1{;COSfB=r*RoUTH{ADX!^&H<SoN#Z
zi!&W9)Y*Rcu=y+a9*DZ#rNK}~=lS7#(Bp8?3}pwpg&bkj(YLy@cnQo^z)gBadF5#9
zyc1N<urlY^T1@>Nu}c3=!|zs)Zk^xHz;#FFh6CTa-0d-wUR|;~H0?cK4GIj2=NhH7
z@DL`IARxpWTSBUgx6NE9NAJJszw4h_EE9zaBk-DS&yLmFG9)j08TzePZd*F0nn>nF
z0<|myR?Zr}D@~$99vZ9}%dL`(YM2uUU-TvWmE%00nmYQ}!OS9(yARSs?NBe1ljaGE
z)NQ{^!hbtO7q8HDtJ>wiSE(m?niwB=$ZiI20j%a5_UGg1BxQ)IjXT<ZVhfnA29sTY
zJ3&kh6I+a>RsF)#vvwq&>*>s3|1;n{>*blE#M$qdl}o`cBv--j<+gNXX^%%gmCm-u
zawNQN->9l3)StF<s3HO^53yk)5&{PY2j!&yG{>=-H(m~S0BS~-^%*Jmm<{qv1OF;O
zZS_-^IUhya!L@}P{qlvQ0qdDnz<*ykJY??9lW>2fI&E(PO{4z!Wff~C%h>yjJ=D($
zTKGA83}e1MvVUB<Uv{Ar85bpGPvDE`n5H%O(RGan4@@*7Xd7x<`z>?0hhZyP-jgG@
zh{JVhKfl<B2b3H-nwu}jL`%Xu?y7;iy=rtf{cYf&dMHjinBRgi6a>Ph$RNKQI4NKZ
z9X)^0`n>(sadp?Vp~2Po?ZNilK|DsO{UL&)$bXeO!yq##eS0BUL*nXszdF9(s?w7Z
zOiM~Dq-pYMs5bD5a0S@u*`Nn5TGBf)mL!1^T+{72H$Jz215|!c<zmK8tAOR{6PAX?
z*<H297oakyRE0g%WBs?8=L&V}iDY8QALy(WH+)l{4`ObOFn@BmKfqW{y6=G!Wm0r=
z?}6&x#6*I4!4DII<SYv8|2jNqZIMw*Vo^?p#VO$%Mrqmce++X>5maulHS0vKS}xMr
z?B*^MqUU~msRnLl$O3)tW`X%C|Lw54aR9X(lP*{q*N=LAWW=L*;q@3=>9sf|GA*Jb
za_?uPZekQ;&#I+sciWBw&K9-wH;%n5y%N3GsV<#vyUV=g+f}h+rf@^LbiH{!gT0Li
zZ}ID2-<_)-`N~U7a=$m4KG~f7x7eb}er2vS$MCs_+P^5FGTege=>>pvX2rK=vinH-
zsr(pr`<ephvOmyVbY-EFq|w_?P7Z<g5gS8t^idmt5F?DT`0yJ@`3~`F8Cx-khx!as
zR7NCo;MI2M1(nEqn25q^|MKI`+@Wu9aLD&t34*bGu1^J>Lm`=)SJ*C~^ZofID5WJ<
zVgw67z=jqUYOMN!y;f(Qf?bbB{+H=#-i;#RMRKln3u!OdwjUyh90q2Kyg!`j7^s`@
z%SJZ1kL=(ktG#r9zK?~-_a)PSo+}eJJi^!7qb5QSNAmNOM9Ej;PvAQjza5Dw^GZz7
zQ4uWP-L+odE4#W2LG@7w4;-+x?hI?DdkVjUgj9dg)lGB-V&VOo%+U>*U}nPYVTg)+
zw&A`kLqUuEt!`ljw|zck+AN=nR3)FE+dH=cUEDN0VIe`xnFb&eng>04$P@}%HigYr
zaj0hf{h$9zG<+Dc@>k!*`;#UE8=LFZ=_WejLeaASe1lBv-;gCLcX@~nG)aK}+efU^
zCvfw44Y+sc#WblJyX+;ZG2KIrtKa6!+n~$%37lzH-_qh<gV;N*Y=WDvP&bMGv6N7a
zZ8TISA$qP_@gkK)*#oZUc@|u+WgXi{jnfVqSf3hOb4`yVU>MhOIPB0<;Wxd<#Q8ar
zD5$=%kA*vWZqYDH|NQ6%5pYC*w|<%{&KS+41{B_jEN?aXN`G~!f*){^;CExC&PK5~
z&&>6?L(<zWJOYSZv+LUJB(q{_1IXuc;(O1YH(cUsQxY=;++|@3mubQyha7Fq)sp^J
zj8E(Y5R%nT9qbhiQ-2N0N78=QB9MpsZ8W$9BKNzl_w&YY?qxL!?}cyfFS07nIeTjW
z*+OpEg3A*h5)iinBFFt`DvX#R$!Hkfg}gAbRnNQn+YjXP>YF<`&0P`ZUA5$wngZwt
z4kP}2W69R`7WM*Jww2CF-Z}lXr%cvPPsE}9h<21jvasvF<6!Z&#g-vn_Xz!vH=z?A
z`HP{~A%_pXKB*!!kE9K;g)k{Stb7ApetOmIyKPBgtpuSCQwGU1Qha{#9Hg>ac45m^
z(doIj<=uI8$Ig}0*37fG2;MYK4U5iBuI)(k`**w=wXw{4gjn$|6?Y7dBQX{F6glQg
zB|{?NO2QC30LF<bN|gS}l7N~cY~!y|??fIACL>H*+jVE_`CnBT8+7?Ou_OS`B)t%0
zbZvF*iYYHd@TIqxGmz}%`$pW+GTL&UTGwStBo`ve)mCjeDQviMg~g7HxWClwFCNEZ
zE?+{~m$!{Fr!MYU#%`SarJgQi@{f2cXg@{7*g9kp>XoTN!sJ+4t2^(xj6kWFp&p07
zwVRTA-Z+oY(|OL+N4<5J=_(6sVmnq;URuR+7teltBu_5$rjJGW(@Fp;N7y|)(%9s1
zx#{+FnCp?e=iO2mG*0|@=G{1H`eO${`5zDED|^Ol3a<P=Q|y0o=65R%6IY%fio#P)
z=4K4#P)(<zYUNa&6xBPAP_?(kaT+d-)RxJX7c5xirvN{-l48J>i<8#^*sPQ@Ad`gu
zGzs*sTdTwgXG-1qS~Q0W@0mluI%~sB*CMUaCj41yU`B$HAYnfyko6XIK5_TFr#ytb
z(CayfVHc7U$qa^MU%tAWN|le*(qw6+^sf+<;U@}I-*KeKyZpth@}ng0`VB9Oy+}rE
zqry`xat(j2HnYi)=1l=Gkb!~+98C64>89E2+VN$kht(3*|LgeoXFM?Uk@YgI5(J;|
zubf%Or(Wth@A29>^NP&aKHtp4#g?~0Kh$bjJauwPXow001}Q`r4H4&ySXa1xdQ`F8
zAYf1GQNSa}xWrc3KtNv5Fi8_cP`vY8q%$zUNv!gj#7WQT*gZH9qJ1Cq#vQpt&CJVp
z$4ySz-Fg9H!D{LNPEO}rqx+bCY;S(BY((zuYD6}w8`B>)bh0&PE=1DwIu?ui?tu68
zDa4*sJBpA;jq*2R@nyyK<<ozPH*zCGs7Ul&Oz#d?uFND+$S$9qfsQe&e}_l%pf`Pg
zSr*z{5J$_<F(6x5Hb_V3J}(X~F|N|Tx7~JD=h$<@Z(HDKn^5#(Ol3SaQe+&KqmI~_
z+^`+X&KR{(!B&7il$87HtH?BvPs4FM>Y!&DaCmNZJ8Ra|(R(i<&a}nLNhGJ~)Zgum
zu`xA;RNL&w29ZL?T<^!;bZ?P?N!E~-T#fy!imI6WY~XJ7W*k611+EuW$FWvMjzPq+
zTMR54H!0<<3<(nx?eo3~wlStr4yr-zWP?-ZXCdWfQ!faVd!5p3UyM_yypVcs88Fb)
z=qjORoZ&uDS4c?yC`Nww-&fn&pE~08lEQD;(nRY_&_C<;QJe;<si7_Iyi{Ak4rx-6
z_2PHoKvbwEWun#-d?`&Se^>0W=Xf<F0`9k;YN`AF12!(*0o}TWA*wr1#!|1utnIxm
zEXF(0D7rS(J=L~2)~mQHv8wDgTCq>>i9D13u9l@0saL^T_Pi-7KL=4Iyl7ap5gL!r
z*xA%|YP5f1H#84xxpksTY+;M$7zX_qe04McFb<!p>j>LKa${2oRWFjYVwVtPR8$cL
zxue58L?A6ZZsoF+4E2(}Oo#dEB6}5;7_jQWx!?3|eHESid&8`r3j%aL+y3Gs`46!o
zz@0xhG4gRb0T%Z5j5kL(;HsPO*-gXB<$1sjiiF@$T~ztqmtWBn)a$FY9QQl5Lw{KS
z0y#a@LgBy2A(5rY&wKd87w5xt70L@|pS^if^z~bsehpV*f$mr8d)$4RSFL5`t!%{f
z1+~Sq&dGCGg4MHF{nvEqe7lEQRF9GyQn!b$^2^He`4kAZA%(A0Wy|jOjm>7V;E{oX
zjCJr;l_(tQFFND8Yvkxd%F90q<Y*{(0^D()_q$)>d-BuA3kdlH{}LmsVY6Y@#^Ogm
zR*3lQn72*)?sP}k)-7Na9CF2>*OUG9TA|xx*`0`BXk`u4?ZsfzGyk1H#2)qU1zuZH
zdf}Pfe~cpI<kNBx^&AnM_nsaSswAJpJoj?`D8?t`Q;XFhy!@t5nyX^yhE>0j01Ph+
zQgxRDxrhp;<lfJk;uR+^$N)GxFz5%5xqlzwzH<Mlg$bO#sC}~H2Z8dv*O?ED;YMRI
zk{Bf5y}FufD|tfGuPmn{LKfP`CIhZ{{O?yo{8-p4B3`9j+wS?QY@_^?9oWisAGBoU
z=9pK*ti@leNAUur#Xuw%UQ9gzjO1FGr1HM+E0s%D&;*ygygK?@V<nQKUNXJq=<+@t
z=J7~u#CCRXbac6m>>CJ&$L_uEgq%Xydf6cTZ8xS9ipa$t3(72sq#ktL|ABq_9_lZi
z%sqACbD%8nG=Xl*^4^37n;bIqwPs`TuvI>}H~;GL?lSU7LEmP^u(t%BiCvBjHkH)T
z#3X2!Qoc(4&_W<jVrb9uA*s+Sn9{Vx|40>cq!CXi(ddQ5Hin#|Z@k4(563fE6jPIG
z3+vQv8HZsnMpg8<4wt%q){kFNx4_-ao65^|Ed_O}pT_Jbf04Z1knH4@L$p(o7o}*Y
zG*Ol*01P(wu>=zNgnHnyl=N=>v^tDHuzJteT0My`6#N#C2<JW@ZVf15N<gu8@t?wG
z=#N^uUo5j_A3lbZ=gcj_*Uo~MPc%zR9p9A+eSc$DE=!nnbHqzXW(BZCCN9MIJSM<p
z!F^v?FL0lN2W(LEP#*Fy2^vTH#H-GtOK09aO+U=fUBD#AahV#ZHtBL>o||6sIX<_0
zujuRzB_yNslY!+tg4sEmOOwmIwy~0Rn^2DE5StvblflDDq#x#N?7*rn{P~t*`EOwK
z%{F}l;h##=l~%9Pl!gngoLUU#`&**Ll+=fXz6Md{y|W*7>l;=7;9GB%?u$4&$Bt)9
z0LZ?+zPZA#X4eCtb3><epCL%3kBtJBQ!%~!mbFOWRS<Ih;+c@^$?5sdmnmi22|6{w
z2L_gErQib}<3mZGGpx9To;Upv1@E|yodM5dIe%f;%#IA3B^X2`{4KKKF*&oUR;<l5
zryiArZfw;lufjxeky;F-LC8y*iS0cc+V6S%EHvlos4Y7X==D6^FnZ9jRXM7W)9BGg
z{u2#WdG_(?lt+Do*6RiA(Y2ZB(~~1fCpg&~13S!=I$@FW+|~O`eXUj7AZ|R#JhhN=
z5}u#LS}AY(RPIHrI>HI^_0t0-qGm@@9}1drDa=oNslO<ymS%=FiYUJEZQ8F_@mp4}
zM*U0H_?TY}0EZ6M?RTef*bP?fwt9}l#jg(1Zx#EKbBklWosUoIPAjp2qJM*mDKdWs
z$klTL>HD0$0+3f!oeop64w5wSW2R!531-#`l{Y~GeSbMd7n2A7LaQ51K<`BbbJUYr
zEVe|Btwp8{c_XF~UET;ZN)kO8pICzfZm=k%ZXpL~E&s7CSr<1T?R1j1|2QLtit4Ip
zyDrz8#&adniMDwSzD5f@5x+(ln<4kyCk*fXM2$hYnC3Ef?Kl{>qGs{9FR|$ge++ab
z;AS>mP+;zU@N;L?T$qEQMHFs(P0QRqS*9>EIJDH++nAK#o=s;ZJ6r-8lV4W07e;03
zzX}9Lgpq%Mtq_(DiS&Re<n39Wn_Ug1zewIKx$7P$z{%PQa$TKRV^M|+dzQ6i)^RX1
zG0|motUNYJouZI$=PFKKxih;f2_U?<yX8WkCR(8mj7=r5Jxqu8abGGl<k<34nY71z
zZK^)zHHhG5Hznn$5_eXh5X)f{*m9|1J>WGKV;ZdqG|TQ%rZyTB^Lbs1XR%by1xX{8
z8upD=0@rp+B#1uSXKq$r0H|@yRlJ7vmFRblEtrX9bX&*UeJ4(tM&f6hv#$xrrHafa
zhD9s+MkyG0qCDZfY?JP;iFEFflTba~&N6wXd`RM~eOqcH0r#OAjJU2)avAUk*47@P
zvVB&M<8rlm9<#|0WO^l88@$TgT5hBjuifY-M4h!&q*6sL>Q5Oyb#SzCOltg;mC^2$
z8WmC{c0CbXBH^>XwU;_Mr?AmV@qXkR9|uWYM@-HQ=|%jJE|e-jFoUM_@36-!J>6mA
ztRk9&3FXJ~#wJcqRkNEM+@*{kf!HUBp}_`>5U`44=c887bteZQ+6-}3JZlyyJf`<Q
zmu)sXtm5o5d4_Wq*|8~uHCGFs47<vwv(MGsZFE%{U`|DLZw{8*Yd$)89j}r*B<q%+
zAO9gkz?!&}ldzf&`8KB#Z~AjYy>rEHI#G<F1GvzO0}<%f+e`y1&!n%tK$!gYN2gRC
zk$o1us%-R(Mu5+DoVEW(&aA47naEAwd1{@rDRoiI!U&Y=v%7N~gbAt*8o-^riVmrc
z_MHweNi9M?w+qKB_~IyRXue9Fqh+ZhC5dz{&#oMs=Nfa)&M}cx+Z4HMNs2BfpG|&0
zly+&1ABz=^sTW|xvk*`Lij>#O7wwk8q;k|&^>ZTI45xEjx?mg}KE`NJ-HTYbY-|+v
zO1ZLqCVrLsAKxa`r#7?}QXph?M*O)qKbdMYa^RWIW!8>BYNrap9p&CfhMlf!Nt0gg
zPRRA%n2x}Ceb2znrPHd-nXkLnNrgW^a&g>!?0QuPuqG-?N>YGg<E8|DRb+Oxc|-5B
zT7(GbI(_EpcfOb@U^9PYBEVS5cpdF2QRxp^(l??%J6#PhNbCFBT7_lc<!&?lrCs9S
zrFxwka6-7^kDM7E_dm$UfnNAbXuVPRQD&HUAwiwos6>o?ax;LG<)q((lH$2tAyH5i
zqQ&P<{ItO`0_be@yF0_4@n*@yu4V#IlB?OWI3X;bM}kUYm1C75ZbG!q&dz=aW<@y?
zZUq~rOtoU$6-ngjaCL>TeoFwHBItAY+1g^mew!D<>6#v+l3F~y%klcPX}ix!KdYQC
z!pvtaOv8Wup_(kyWPGkqc}R|-FS}X=dyMx=;LYzLpvCOwb~~A`4bvSFwy`i>lCf};
zQ2c}GLu}1lbR?NYn>IegpdHl>HNWauK`+l@LBi)oE;)G8_5=ium`KLjzBle|%cP7m
zo?qBJO>FFmr5^4+6|7<vP@N_r*VA*s#%9!;Hki){Rd2|ik(<9a$M$2x3O%bOeGq8(
zu2V7qb8_xrq!_;>N=-r!+}3PvnjG@A@K&tlu*y|xYW<9g>n3MHUT|cr(xkd+)faoI
za*@<}4I5h3R)j@XdEw<#(iHuOiZ^7ps^_LKAU9B>w1vzi=<&bxI>fS&WHrRDgIP%r
zgugPgP>L+`4JK7g^Vr)YV<*wRubbPZzc7%;@`Sz_F_24_V=;baRBx4lpl!V7%rP4C
z*x8J<m>InN{wF5y_N@Ogm0whYyq1t&KO_XHqqR9lEBx~&Bz^^O9six!fo7#Q4ZQvy
z*3q%pNB*5lroC3c=xIcSs5yf)Zd7+_bbu&-hwh>SYlc(v?<hPF!To5E=2t7;6HXv>
z)tK(Hi}S*>P6e3hJNZy>LwPO7CjCnOCL)|qEi0=33*o)s%ypd&76)%b<^2AQn(_@t
zLtP!u%i}fm606vSHg8043{}jhXI4^_mFHud*oap(`v;jwrl~`(xj7dsC-Mc8(5JHr
zY)E@c{`*7E@tQx&jRps6*I$|{+AM(E%u}leRbNMpb((5NUX4`7Rh<Ps`;6=qP#CwF
zE_>BczF+pXa<BeVWzc{=jPZjTl#%+&NXycz{`r{Eaw9?ADpUXM#js>&j}M>D&lZck
zhbkSSqB)s|i4W86Z(bI!M-MMJI&{oCzwFt}QSF|cyea$2Jmeup0J8CdM?lJeG9tJd
zs}-d#*dPg1$EZ@D2H^|B1%DT?e^m&^|7{$CxDYXD%WJ%TF<Q^_g|td@Ya=mQo&EWG
zK37WG%#P;#qMhc95w&_h(!%9y6Hk#=P|Vd^herWIz};;~e6~~p|BpfISe8mL!AvEP
zRkO$4Z&`YugAz$(hETOBaM%DWMl``id~Yw^l0QmU8Gi>KbznM8ps0H<xYzB(7VEtA
zz~4{2`D0G0Ga^sLj9Oe@yBH+;xm%rQtz2LKo?0iAxMcj0L|uTA#>+;^y;3vkTZC98
z6(ukHO?Cg0l%%ZrT@HNbV0QHijLnK~<GMrIH+fi!D0xwS%UAxUlwp8-9CJT)mfH#=
zo_u{n=gAL^z$qwjabU3ZmoGiiHP*t@({uZ@79>D1WcK13*7x&I1X*=O1*N=J+NO`#
ze?(l#t3i!dMUhw*HND2gmj;O*h=qfD`PWe}L-^jPp4k<$+W-9c1lm8nFgu(c%1r)W
zi=u(4kwe3r;SY!m0PgNzwStgWHq45sYF8d~px|5_BSA{E1#g%^iUc@_js}5o_vYto
zq!H)^$GC4*CT*SdRwuVtsnWpTLoaze<;CSRwJ*d6_rN<tH&^YC`GSP9#<A%_tB$+h
zsqA`sx?^T7pU#)P#~My^r+gxc^@!Nyf)a|$V59v7o-&jGQ)2P^N*OHH;KS}|zSDN(
zf$?F=99W{}MGY+jXzF51rLPuMsap=S$FF+G98z?!Q}d@kj()LiTN7}LU`w@YzB|ft
z>A10c;hDK*Q2rbo%&b7cg?iS4N+dfplDL!tn%k05N}Bm^>a-Q?A4t(!X^r9hacq1f
zli{`a2g^U&-7o5;L(z)2dbyXE^+Fmel-2Sz$D@5q_jEtgnb7O=Ktz%!d+nuEpd-6|
zHugpp$SQglBElk?W&vxeYu_V&Bwutqh*1F#S8hwGRGB<T?g9^1cd$0M_0B#w6k^%z
zG=+meum3jxe`q@IM=JmSjT=Xfh+|Y%_9ioX?-6>VV{h5nd)`)sjxvr!R%XT_bjVI|
zWE}~|K9uaeH{bj7!}m`(<Gx?7=k>g<#}xpIhOQX8UV1bRW$Fq3tRU*=g&<Jrk<X=T
z@eW*(*iCgaI;kzR1cSJLkmG*mfzZQ;Hf_5(P~z0zdhX`d*6;OsG>z8A=!^UKQ2Yt7
zDti>Ta9IuMxN7I@rRoinw6%*^UGS#WhDoRF*HlM|s-6A)-BDCsZHkA-Q1=a&Nx<~e
zODY98RG&Fe(ZH%h;4U-<fMtdu6p2)@s9op3#{_o2;Nb*D)Gm46O~nI4X&(%YN@jKh
zNalhEKoA5joUeSUu<6(K)T3H;0-D50_ob!C?MgKnw-<~l#+5?1({#RlCse8DGD_4-
zBVbZzCM0yEuoNstl0K69Q~Al&3T2q}Vl>!_{oFr?t6TVJ$(kPp#Pk>G%Ohx<_#!2_
zGnx*@NRucEu9^#imM)LX0JjT96|j8a?R>yfPmXUU<%Ut`4Fy^pNqdJ8J97~`iWd3)
zfoW&^{XL8_Wy`CIy;MTmm&*mfQTJX1)=uW;Z2t5Piq=I$jlI8rDkLKBnzSHtdCAOH
zpv_vjsotrV(#0ES1I0pHpP1xG5CS^HRmVT|+stxtqqeQ>i&O9D6JYn>hw`!mSqaFp
zkSD)W)KVPz)r+pP*DDV_AQY4OG;|m!^Kjb*e*K7v71%bJkNOjFv>t?d`0-C-o?(`b
zq@J9_L=6u|y&SZP-U%uOA(Zj|rvryed(VPXHYqiAKLF<2{ru99&V`+r=;jnmQ!w~m
zA1!|}5frtENh!GOE4UOK1qf6iF1OfrICQv{HK?^&4kIFtw>ra*^lan}V2~6sF)iG4
zL=uIiC&%qU_5>&qfu)HHtw!~L9;Cm>aCDX}+)S{WOZWW?0#5A?=dW%Q?t1)QUkVZ&
zLPQIIyQEa;6XBicgA}v+S#!_IO6&RAX><qxH!hbL-0w4Py8GaZe0yajYvig%ld`q>
zDvM&{&SAC*K|NaX>VD@{?aju`alsrz(FvaXbK<c0sJpXX8cxrF#_7v5>jzd_Qlvj=
z<u@naku~fM!b#Q@y5*#Kq@%Z^-aXTJU*4DrtZXT)g)}%Q+nBp4`v0{x`n45o$#b_(
zPlwjaTJZu~H7_dOD4?;UzN<si_O>pjkaSLQAcAsx7I6Fg_0=~4XAd)S9cW$O$fUZ5
zeBy75Rnnae^K<%n<)HO{eN8;Y2JyLHe%k`6%N9`I<ymyb(zA*QyzFzP8cLSQ?6NPP
z0ekI<CUK7@R8VHfL#C)dv3y$l%uytKkAL7wQm2?-%bDWciQe$VB;3{@p!?`}i>+k~
zR_H?Tp{Rv&rqzXz!`|XvVj<dF>cM5}56jnF;BzauGEAgmL8t41LSyJT+|ZU)8yj=p
zAx=>q`u?Z8X+i_wf-arTsE6JO5#~(|uqW7~K!bQ+Xk#&Ytl9{)n?$Vbzi7WW8AJOt
zS%QC*UHDb0$jwy#pTBACt@`gZn`CR_Uqcx!-am5x{=GS@&iUaz&Dg@!=T66NgJIut
zEkBe?iOCsw<v7@&1f5+rJhKOrQB2X|Rk<4?LPg@b?@0VMzema$>f^Q^qR+_z6;Z*J
zI$@I1aYJseCb>K=-S45f^fNmrH8$;*W3($CJ06-bmBbJYf#CqV`yT@h*z-6qT1+wC
zcG-;TP$9<H%#p!OS(V)&v3nUVVJwas#c{6-tFd)?OB;)1sWjB~J6Fustr_Sjcm)@6
z(G@A>Y{bshFmP4XQ*wxHzd8o2sFY3%jk$ztP_>}$Xhbd6d4^#E%@e~?vZfw~7dssx
zfBSmXi&FO_Sa15~&)b0`USMDIUHHkhqK2NrR06?AdauUx!%Q>VuzmF>c{Z{V+4>v(
zcY^OoNlSld3-$EeIBh@4wUaHsD@ID>HKGEAw1pp>KhBs+dc`WZ5e_W<CHcUUSeeOP
zw<O`&Ta1FWad5GzJeq#=UDlr97!HqK{4lJ1{V(IA%rV%c@D*kSQcc%zxO+4LrO>3r
z{+8tLrc56uU5H<vJqK^LVeiV_ThQonuO?5k;8vT`F|Ws$0D*za9Rl#0cD-)adEPu6
zRWst6GkJWUg)Dt33_uJTwY>#5`Cm+v&-yc~E~p-WJM+<$4`N!yk5mT_G%|7!JUn%}
zjvvgR%z`UuQ}-M<XGa9;izY$IZbrf7KX7z-sM#-Iq{mQ|U0PjPU&$F@L^s3-$|HA;
zMqfwY>@CiR{W&w8zFHtHFmLvi)52L2LQu|MMOTl{6Z=L0&J)gG)933$AEa5MHur2=
zLu{;No9g@y3a%S&)|;X??4CM@g3i7hZ0Gqz$eUFsd%x*bdUg1?#XC}|;t?KbC{^a`
z3hb@+9#u2<!@3&q^yY#(-c1j&G2Wtk`NOm|Q&=dFH18kY>0&N2;EUhcbf2JAu~8AN
zomOVq<A@VL>$oawcrOrzVn_p*^<mufQPrCNsq0yXNi)2i&~NqRCt>FTPwp2`J-+Fu
z@iAhi{P1@#20{BMDDSmjVL<W22GzEMrq%Pae10}fn^T|D%2C2I)vl0Q9kI2Xf=;Tk
zR_ySLUfIuht9<<vvmRy&vtd}5)u|q6HvBlF3+V3Q=%)=|RJh(MMpMPUFk)~>Vx|uK
z=d#h|;wnre4mwqJMiHRJj9T4SsLqIZ=O=qWPn(1}*QlX<{7C57&QV6Uhf{`-SQ2WZ
z{xIuq(Cw8sEHYyZjRuZ7u~b3%GcS6!d&pO|WrH8-n%p}6INk0uv2>7iKP!0pk!C8m
ze}2<(h0Ff&>dhyL_7JF_Qr^~7b717n?k{i8k2%N~!jWMOt<0*%{d*p?vu+2=nF_-{
zT~rGD^vgy~^I!YI-^KNFRgM?gJ`S9GZ`X`$OgBf1_bl}GERUZpFZ+gdavZW3){2*|
z)EFL6O01wLiI|o>q{y8=s?=fkkvh8gKEMV+AS3AdoiYk#y|yvUzxE}4)zRC)MMz$`
zf_Kh43;MpjD(W_yW7p)(TeLKJ<7V6T1@0W9n74;jpP!k#pThDhW&0L1juvzav!~IJ
zs&G_%6q>4kCI=5oq|IWzOf}}NY2zyOFEzfK<=pp<#;X3gMF~WY9-7LQc#BZ#7r@D=
zc(Qedb)6oSUilL<W<pnn*!k0zZ-PFth}c43-^|;Q-Y*@3!E*g@<m>{jP|hV^UhgLv
zDkRkL(evX$zY!`tO3CW*aK0{qFE5K$bU)XO^~a$8a|s)|zwBa?c5ey$dOqy>Z)w&1
zWgeiB>b0}UWuYcRMzFB9jKdD@1LC4s_oujc@+{%#E7qIfZUW-4-L%C904&%YjObN9
zPE%nA+l3!}R-so7yEtCooE@tO!CQwdEIf_aA8XwS^wMEU8A1g)Z-C~ZNSVJ0S<Vca
zBcFte`G~ma2;<G*N5?bD)98p3r?-mHYx~YSF=RZ>$c>h!xyPR<q$U>tA({IME3-}n
zG|Vz4R7Qy*_|t7-ss|UF!@%B)EZdx&;8#DqT<;Y7S7)&@LFoxG=F>VG^<Qc_f)=N{
zerU)32YJ65JP+(#{;lE0%5t1p=yN+6h!Z^b@){f(BK-DV9LjF)LjWF|^t^Ow-c5_`
z4vBH9({5shnp~9x#`OZg&I7rxopT=6G9m1&clgjjUVMKDe~~_v3EvrgfB%K|fQ;cN
z*JDK$l}TXt+xWnJt+3{Z;ds02^@jQL<AYd%C-(nQP$!DxTBwjhLoN+Mo{N_#`IC^X
z>7&)+UfZyxyF?D2Uyi2V+TKiREB)R%;&;}}&noWH8kY|}{rSDd6-ec!egbNXa-JBU
z_9M$G2A}G!%LCwA`3m28Ie*=Ib7uEx+Qq$T#ywI;E}US;X7?R=^@ioRfG!Qaplur`
zJz^XwUwX@ZkRe=6p?Hv_7LDbQlFxpi^f}OXXKb6S#phN?lU{|HPu>mK58|_$$MmDs
zY}F0b24vqm@e2(AuKp&W27J!&s7A<3d!&>e{D-RS&!CU-eha98<`GWzP@zSahN-ni
z%b{CNV_{aN+-ix>xJpat@<tc*l_r}*R$P2M1!eA%=l<`1)m9-3VPqzI8Gy}j`JKNt
z@eiG`(;0QVKGbTZX`%jE`FzAvr1R3TTwbvF7MrWJvDy~^S1I7AYIP^A2PT%$gEU%K
zQ^CAvOXI53&|*Gij63>6JP=FOP3nVcG^M3E4V6=4xTT_jwKz@vm)Wk$%CFZaw{?sK
zt4}>uzP}*&2Kzj<uCH%+5*|hH&zj8`Ltly%TW`0soDSMpe&|ZfaYrRx$jaKbCG6r|
zYFZN|wd*o)A15sZ{a7i5tY2Dk#(bW^DxHj#%aDVZbht&9y3JCM5rx#Y&wO+CNq^)j
zW8X6Y`8nK}hO05`gBIj@n*VdHzz(LJJkKZ2<~y@<Z9z%2KYmf0oz)lk%vN2<FT_t^
z=s%Ps?He+vr*@|$$|k)x)9f_Wr1RtYIHlj9_16H{dTM*;g1X#-kz9B>k20(1sU(bo
zo!;vys<M82>#(ZwSV_Z31B$AWgV<F~Gd|>;@(%e5LmafwlbEoFQgaLGg&}2D#6ou|
z<4$1EMfYi6?19RsJMfW)GKi?N1?)+@dfa#Vs@r#C-k}?(3lsxfJUtucw*LDr0*<Ap
zA724bcHH|9RaJf*FiGx(l=11~TW1<gj|tO?^oyP-?eTkVf%eW9FhX`jtIJCyq0`~{
z(Z-MqC@qqQ?95)B_9|<px^I>7th90Jrq6Z)fTQEza`M5-%II}+G?uZr2Fx&nF95ml
zL$~5I2DAG;pXYZZczQjJPQQ^ps|NG1K&2#}K5Io*)&KZ#?O<gXWVjz`|7iO}RQPHk
zjx=uZ2`1Et4lIgsYQ><d%ur0Us=pYEku+)vH#fwCb*>jF(s+6+6-?)-rKim=klyQM
zqc54UX83!pyCGoCRDP%)A&&PbDLC)$#mk2*XJ*;uu72q_EESQ=nKSkDZ_5P|X_W2_
zq%=_)>rr(G4gB$SLn(f}`PM_ea}wTO&j;#=yVF#*SNmoii5&dMk$%!0^UrC8NaGlL
zM4GX<z%jVP-(Bc3vx^$1r|lIBdULk>gSD+A#M#H|a6i92L3DR}DeHRwdV2h=a`bLj
z6qPR)bWIj0hi}%QkMIqhhk#(`g~)XlZ~nP{@mx-R{ss16Udh*f{RjUc5~8IG#pkqC
zdZ?KfXT$#T5kbI)_`eCFe#{VZ&`d}LA#md|=67!$V!Shtc&c_@-Da>qKKWd<J)^Lw
zjH4_#DAslYK0MA@t@YH5TAczfYoJT8Q!Sh~jt#A<4$vBRHll+AMqv+f@3i1%tY$Vm
zH`=#4#<AH856AkEaM{hB&F6DB*PbFByMGUwd}_9Hwpo-v@Aqj_u_%82sxW1pzgaDW
zmC8}G2m&HDQcPCnWrB{rmmUp8dkGQHtf=j7Up?Ldz#T?&@S-8`o>Wf(24{jZe12o^
zCS5euhk07g9qI_Q|Dtq2tdR1)Cw)qoN4gVt62qcHu+wGqmro8$wvubKo;NY=i$RRA
z_H)myQNiIBHhzYDr6=39UOqknUDRSKpK+AuoKAra*<g*2ZW^K|Z&6vdY39L_>_>&<
zQ7s&gzM;V|Wpb=k5En-r!mlEQaY>e-A~}2q4-Uf74nVYvKuY!qrw!%)pJ1EA<UfT?
zbAsVN@N*x=KTFGw;5&VM${$!remZzqP5LzDQ{N1AAgx^H*}bZ&Bc`SEls+EXZX?F$
zlDnEdpQC%1)xRJ)xVr^y2+7zXSU&Cl;P{!Zu2l*#s*DLGKZzVvCz>VtKgE3{Hi_`D
z-bM9okIe!?&SRD{J5;D6cxk2HTuT_Q0aFkv??BIeC$4gX?;U%n$`k&V3)RA?n(E4g
zL?Qp|;ID}os^`>?x0q?ESJEnSCkWw%k3!8XOnhHfHgf1ulMz>Z$<DQyWLak4ZIL2>
z9e|J6`I9$V{zl>Qvps@CHe&m7#7P_K*TqM((0%}9-OD-=RpVptBIPe8Wtr{eQ{2Na
zIa_?6EFR_$--suKLQ8}Os;Rd&pM%oa*<9;M&E{NI=}>TMYe$E0>FngB&gfIUwOpnT
zxxF)wS8{{yQTNw}E{$hpXn1?q3KloRghQzUncpb{XLz96S!-G1akpa3H@I9aGbJP>
z2or%|JjrJ~Nlaa#(gJC-zw((aus*g%MVCZ?tZP0y+uGXNdw7219Z7Op(X5bAr?z&s
zQUDxJxl*wzUCx==&1~P%)<mT=v1o(eo+r8ivk5H3KD{mcOWwJ9e*x;O4HgcCEF&ax
zB+hXkd~BQj&klw!a~nm40*x}AZNf;k0C~?yxIl4Uu&sqUH0<AGodj6d-V_pdhmI_~
zv$^hg`lR!sStK%c08-&gNw2wr!l6%`#AU<&oz{^aZ~iX@bW47wML>FD+tpYGPW)B*
zECZNdZ+ynwG);?H|6{*_9eS<%uhq*vC~A9isxN7(SLwV0wROD{dGMipxZojB_1ofV
z1vm!E_g<B>utOM}Q7#R0-nZW7_PIznLlJc_BGpx{=Z)O1jA}GX2BmO9`v>OKJ&R}7
z0!HHM>T6zxaUTg*yWBVO##;{YM4?XbiUAd{u_=z2kjI>@n-f`SYT0b><;$L6Sy4$Z
zYqJtoOMmq0hpJJgcybAhUI8G(%ifJ^!djL$O6_>p%vrUESFFxiS>`JF|63^TWj*iK
zri|X_6x4nz6#3Vrm*UjO!re*x?58CaS6V%->w|yHB&Eh2P9crkc5Qw^f~=WRGXJLI
z8*RsV+&=V7X}Ibo{#;#n5MXG|>NnDFX(I0N9}8g-EKY()#u@Up5LpChyGA|ac4f|1
z!PE(B`*F$Mv}Qd~-l$`L?v0waY|o(zv7Bq%^Nx!8VX+b=I*o4$TDiU&>g(%s>V3=T
zl*_5bM4&T@3}ET?<Mj0Qem>EE4=sk{bjqtZwx0~aF*-e5*xz2-LZM*4WGAQaYhM~n
zIUQExP;jx|<3HEpjD>+>m)u9PoCghL=R-&Gt`VMDjoa4r)pmD0#t`lTG4X#3NAALb
zsL#Qz)PDTXLZ|MALb&XwfoaY~quWS`51Gt8SeuD+9U2YZGzeo3jMx6SX}s`NGT-?v
zRE)TQ(JvzX2>z*wc7<^@!;q<BpB?lUlZ$?~5)ekT6LJ{oy#+FtF?^+TDtrSp6%<%2
z_{S#Q-@Yn8NI9Nr%NXPsuqOJ^r)s)ug>9lZ7wG-4PyM?2RoN5%AtekU7rZx{1*!`}
zS_i8muZneO;!d{b3-WXH67-8s*U?qlEJ|twktn!ru2RJQ$XliFWJfsJ9WD6Yc@iyd
za^Sv*sU{&b0g%wj?(@WaGk@s(3<c)sNJ|fz=9SW%8Jw7j#)1^J>i>n|-5z0b?xBuk
z9)`7*h-%5Y5`G6Q^UKS@y8N+hup%$7_&dtFCcAhuI6U*nydYksI>@gZj&gA=yMO<F
zGXfLrw~-fweYJ2o>A`@ptIV99ou!v%z>~(A|2rE51_x1AO<!<?-{(K7Ud*WN%-Th6
zh=7!AzEYIwMiOtqWy8&7fpk+7F&Kg4o+bT+V4@55{3OuZ{~MXfzwa^0x&`>oX9{NF
za64vT=OEnTuIs$~)pmA<UD)xI(1vIliT&bx6I9AcER2y?w`%VC&{YW}T<m<RbkU)w
zF$DPI$`_<3cA{3!e4>u3?D9U?sR<A-DiN&EkcnfIK)aJ%9#*in&HCP%`IqH(5+Vyu
z?nKDKVP6Ja5;85s!Norr@-^gOx#04Zvh|ew$)Lm&#l7q&H|3x~>E@L8M%6!jb5cFe
zbXISTQ=7%oYW_S`E*xw(GyYB@_;`rV0HCF26BVFe!pj-1J#A%AELAn@VH5F>dylDG
z`1eiymC{W>!CdRFl!`Z+_T5Wh!Emq;@vllsd`eVtZh-4^3p%W<UQ+zGXIvQxQ0xmX
zI%);TXWRt3qOwKkW!T8c0{@~ydQ1ZBJ70q-0}3$`1SfI*HQs@BwcN|MnFGyr|D9}F
zOiC)tN>hvImqnZm+I2)Pov+UkP{fF0+8Y)gkO@0-WoY~-LXwtdPob@8_-Vn&B{uXM
z%vx!)xh0u^R}Op@8(}(RuxIod$UrHnTUli%#2mnX!VC`2T|EOLoF2X@)~cdb5XCe$
zHjz0MkGO%ZN*TR-_p@?_w3vHAv^{d9;1eP?prfrV;^^2$4Z=~5<aBO57OJhm(?_%;
zP9lzl?b;&UWdvT=_t!twi(crO>vE^i@?U_&`B|P#r+<-XKIw|GOgVI$L4?D7n35ei
zbb-<m+*5i)B{@;e@1x^mDW}1RPYB}Vp3i1?LD;jorRXCEclENQ#8A<)$?cu$dDi0y
zm9u}d{+_`hv-#BGeT;bo`u#!%@(TUW(ly-&I3!dSWj-Q*>$sp?e*Y9zc_&qx*z5N*
zgLV`QZCb88aPuPjNbujg7r)T~!APnX&lCW6K_>$nOZccgA;SuFsB2&fT``yQQ#J?p
z)|s6AQt*>YcEy{%Sm?@0JJ8d^$YRoE|09pM$J-)7@XXs=DrO$PE;;>BO(!#JoNs>n
zq$Oi?HM?+@A?Rd#=;{><TTHGocn3P1PT6>(e1w$VuDx0IfsKThBU=KtOdB$d^Z@mk
zmKtg42?WX#rN6d*m>m-JIYtybaNmu;+Rbnva$-*?;)9LQ579dy$ogRHy6NqL7rf;w
zmZ*EqtMLrl0Yn6#BUy?^kv)AuIp$Cn#lYQe`<85P!0sj-*{zf0hYMecYr?<SC4@XQ
zFkw|F4OEca$G0C(dxBv|weC`&wDd%z1K65IcmCkJHgN(YHN1qoY1%9dBFAE>c4wH{
zsGac%!|T(zKWwaT_W$@Iv&w#bzF4SRiN2a04(I%Tzd%U-*=3gUbsaD3Y{=?p!S!M3
z>z%6l<QIc~6WISG_`eAOouRvb$?YyMwbD!`F&vP;4`}}7Y8nGPTiz>91*fN0KJAjs
zlqfM%CQl*41v)YkJ=MGNz_lxI-c9?2MIln3RL}eoUK2PhX61$emy1o)L+}1MZq86)
zJ$?2oKjScJq0ZQj01q#(t$-)WPERMTK|65gdAz6Du<%a)*%<BHExbf+O7!N);?1(%
z6)QM0iDC75fnvMUOV{Yr;wI%>wVhn;zmcD+Dy#J}xaxxLCQ$k3#i$_@4SC&r0>4B*
zxkK8)w~l|eKKEE>O1GVHmgPB0uZWW}h3j1$gN+s3PY-1Leyyx*3qcqsucAzXdieh2
zTCKzwt1+ndW0a`$s9!Xc)QtH~Hci8x#X!;)-lsf)0&E_h=p)?M7G~KGtkX@0Nxg3s
zBcI*O&gwUfgO~4wAA^U_bAr5QEy>xSdISP|&Z{l_%whKDb+4A+)C>-~v~lZjq*7Dh
ze7ujPq|T>{j_mUqY?-WH+4IQfi2!H!qt}|jLF(Bfb5-9fyyETEl#hg5+;rB1qJJj!
ziZt}J0DAUrbFXUF4gR#4CMgJ$CadAcas<;#gBCHHudrpF_pFX|W7Dl4UF{JkmBzz^
zLPNqr2Qt}Kv*?!a2ZUk}r!PUcFlBl9keh2p=QzJLg4k~$wBEZ0uZn>BVm|4oTZMS@
zR%{wd4co|EO&0ggq?DD=De&kvOnHm5rw)7rdIo3pm8SRMEJm|avs^4?Jmn;`^g`t<
z?HK!k`cJRawV;2}G5kxv8ND{9ksc8dsc4?Ed-q&d%ILX^^SX}DZ^?X+VoV8cl927o
zka2JP5??*FEMA5r{10+Vh%oJz0CW){(JxE$ov#$}3igREp}}~fSV0=cR~q*U_RUfd
zXfDyoBQcRT+i2@lC%~-d)Z_Z)RswA(N}m}qvpY@8c_N-s@RN0!yorTO=@3Fg<6`Mi
zx7xk5zD?hbKT_>W+*UqZxF)8ua0b&Bq(P_}+vf1q-DodCwV&Wox9q}GVjuU|^G7>6
zMaJuk<wL!SCM-k5)%4wsEfJ-Yan1{X(l-1e(3TFfkuZ{>EY}mtsLjt#???FVQ+@tw
zpiEXuwS;tMx>R}XTstSD`xde>lN9bLS;h%TuoUc1SP=XAALT)zq+DzBC~&%3Otmc!
zYC8&&-S0YZ-DZX#7W@?2g0Ku51_bej8V%$JuGqGSjpd^(+wi^7CyI))4c~xlC_BV6
z&d9pnC5I<hN+!zJ&u@5}goZBeY-1;K<(Im4vp&OpWc_0A8>?c&w!M~PEqSjv$u2ja
zL3+luFY2|`au@hBL9<$rtT>G8r)R2Gr@E(mIz7=5yn)-=iQ2hHAm`vRrho($>V~wm
zdOABg?Bqw0`pkaK?x{dHWdq$D_l685utnyH2nmzpo6&-+k<sBumz$63j1V?%{Nt`g
zF8T1&)d<&NS*O1;T3QcoEh5NcjSXY{|MCA}UiZ1l0#53B-(IatU)vI}aHohw{vn53
z<ST{bK-Z&uu_pa(RDQV-rgdvw9)4=S+)&pVyXL^_b`y{eI5H<V08nmDU&&im>6y)Z
z2r<3K2qyE-QtW`^;p<Z>aSYM<P5}r|%fI;hs<?&0HRKVan5yl>((J4-t3JU(9RCZf
ztbk~eJEx1Y5!i>Z`<=h?K^)`U7L$!^W_$oMn<yRPwpL{&<By0;DV&_41&uaxVNwEB
zDQ8IYtlN@Yg>69wIF*MS<`bw`m=>$fuTIks$?R$I(?XVFi+<nxI))pE!>LH)^N!5x
zgNw(XOV)qs`qk26pHMPqQ$FARDNm>9i|IE$y+iVbE!X_;YU+&0&o*YkVJ)Mzv!yc|
zRVN-BBj#<LI3`T>94sY)gKZBMx%yTMOmpLxvU!!m4kPA)2u@EuL)wk!T-nm$%qA=Y
zpQhZLQ*c988}V%=5X9c9E>s0{Yhzm%6mL?0{zmFPL8Ci9(PfrNiiKTZQ-URI=B;P&
z+fNJzg@sXEL#A{CB(B2+;S(MPkdsw{_=8_=Xu?9is*<S^gjH>8-nOr3y_6_>plo8O
zMyxoSnFwJ0U%X?mM~=jj&0WoUV>H6pjTfI*vHb6ofGo-K(eZhd31wMUaMF9h+B_qS
zd$|F)v`K6+cCJx`>XjhjHkPH~O!6t!?Vmpn*#81I!sc%mg_3BqyKg~t(xh_GI2t2g
zns1p)iJUF=3?FqHNI)l#3~yIUy#R4sQ}E;lH>sZ^+98d(H}f-uiP%6tFcS+A2R?<~
zw4?cwu({zYEBO~~sdNbCh~-Vrw=9H@uq5oo(oZ@Lc1IS~G_FS2)C^<ul|#apz7fNc
z>>`)%UoYL9N6dpbw9d<Mk$3G~<1^Bx_vXIphG4;DuJf11Mk!_G&bp2Vyd%H2m(G^5
z16;A)M#a8e^z?71OOZ~_P!2yV;A~Nb{EK)vINW}7-JH_eG?Ricc>-+914nJYNfzm=
zon&rVzwm>m9SA3UD_sp{wAwhytB6yr`U$e$0>RB~`{P!;Jn$yQ6$vT6!=YO3wVKsZ
z?B(OSyPeIdbh>cu-6U}<f>;IWFi(8JWU<j&8+p?AM7d9v6vB!r34j9Z|CLObff>}r
zCg6hgve6Vn{}##~<HrN}uPpq0t-RK@{n!jRGkdFeefs2m4j~(#iIpo3C(WXC{&OAe
zsRqZl7X6f6$(g`2I0cqIqB%!(o{F&E{43`vbD594y68^Jp2}{gVv#vpY=Xo6vAwOg
zod!v{A6JN7h{(K#q5)%I*C+aBrSmjXB<gGc?0HuJuCK2mVja1@s@#*+=3i1PMCS%<
zNfP*Ca|4X^oaW|QL+XR&S?^kXU>XvNx<Z>7wqIa<Q#XubQBMk@vI^Tnj}PbMPf}Aw
zln$n-$~{6^p?ek*w^`b$5bCv^2GBD1M~ug^E2d@`#`|rFfH8k!(>v+~3hFy|lrrtf
zhPC@GT{9byT4^vf%MF<ox+J4;1xk_~26jTR;Q)ekbdC}PD_j0OXPSgchd;}Jy$h1L
zvssOD{3f&UQ|H#O=Y}9yubcV6oV1ZKki!KzZ9d=6Xv1!a(4tRn`y?$LZeVvuu=7jX
z$@R3<N96l+;7?S9^7(FlpVn^Ua|223W=r$Unxp<E!vZ7p>tc+32~17-RkSLHt_&}i
zN>0gRRpj=SP3yZd%Rl4=%3!RDS{*89kiLLG8e{lB>L*WHM((}%h$dm5GVMGzmFSP!
zby97YbWy0itW{2U?1(lFgY);l=JHhkR%xuYq6Rh%^z_U{gaq`ZRLB;mxP;$aPpeZs
z7{h}7(+S3Po+sajtg67DzHeHfOUB7N+{MFdYx}Y$4CTdf`u%Y8YG%XLu!D8a&?c4=
zkxKVTH|M|m>RaYj<KQ@AfZF&@XSIa0NrF_1Yq8(8hl??Wd-8E-iqpjeyiq~moc!aZ
zUg`iN+3&S5{gF@ObS;OtygpfKrXa!=jB!ES&Klp!jd8_W?`bjFdu*kJ36@Nb{OC%)
zqJM*OCW0yKkd>CXMlvoM<B5L8(HZ<l#BFR|laTI#!Ycx6|6vj;1noEbVx7WqOl4aW
zbpX*L0vds5TqJTk?XF$%O}MXhECr%!vHSoP%H;<;xWZ|tqU85j5kMb3?!27sleQDq
zjvpUqYTChjxTH9mYYDYPF71!ng-%D*K3|H9*!kY<bKP`v=l-OgNv(|yZbdlvW?kee
z9X8StPa2G>Zo6JTQN9^-ttn8ZqNY*9O-zhFG8m;Jz{ICnHF;x)gx@W#_S}CWWeIGR
zD_;bFDRrxMzmx$te>DL(tt!8R@}2gN$TIg9pZN~KERif{)wTzl&p`y|V&8D|_SLBC
z_FQ`pLsl+<Icpr*Gm1+nq;=G!aNN!HgV=U8aUA7o&k1%IJE*YNO$&kqerdfbOeIt6
z8T<RUVg57|)Dt8A{MB6feoahD1Q47hsR}^e>Z~l^XUyAf>bx@ReakxE9wE&h_*@q>
zA$3G9ftLJp)ciT>T9h|IrLZf)Q*o)=%>WA58y*_c);<6ISmc4M|0CUWm>Sf@X%UOX
zVe>OdlO3p~c|I5zn{swFv<O=;llbbvk^Iij)h1hcfymWU&S4Php&Sitpw=oXLJSMk
zpo<O@^DjN9i_q6r%Dx0nUrv-D7soR`N|&p9^3p!u-d+}-Vr;hx(`kPY5!_Y48Xr1P
zbQrfBs@kjcpH0$0Kxs0&FNyrVF!D#NeG#?x6-r)i=ZSuxu${~X>SUo&YU>qlM1s<&
z`&E>-yH3L=>4a)(+J68d!@N6p#JhaauaIwo<=t`hqtR;YMa!zm7^GU=BcVqB4FQJU
zGm1`^p|^k4w8;s{co;ryNfeWafcdKO_pDn&|EL%G(a=l6NfMgmyk33ZJZ!?5p=%J8
zf1k(gcB^(jD^Z5V`C?^#^k6uq_c?ZnIS-y0W0IPbybYw;Xn?7I`{Ds?UyL;^T+D&x
zk*M#>V#&wf{kN{#^E_!EwlqfjOl}wiyUgTmw?}NW;|mX?Wo{>YGpC0TFj@6zzj<)a
zmzPFnwBK=!p<LX<m#59akD0I-YhgtN-Ji<FU7=H+6{QSPUi=WsGw!EVi&Altyqcm3
zyjG@qZCt!g@CiWPjweV6;VIjmvv5wp8ax4QoHD*0QmYTjpSch3gOU;zYjh%PIZ|c8
z`sbXWpmOxZVOsH97!j<Er1j=o#~-%~70-e}uA-aY`}uquk(&X-PtW+4e~GjeQcF_T
zNtR|bw@8j8stDfmz(#rr*6Mw49=!`fFAO~3+zgtvPXP;!ZYC$@-$~mgb<wV_uS{AZ
z<~v3{JyO!ima!YgVl6gG^s5i_llf@EY#HH8SE^j6)6;V<4eU;Y5KIV7=uYBESumDV
zuS;G_6yg&|VylH}Y%&xqMOs>0Z{9hVm$#A_9Yq+E(#7#E@Iel7Og*VGo}@EYEzJ6l
z>g8BP0oPydr2-nlUq53+)zFf{2{GU3sbF8}#>ANvX#OF_gYq76+>h}yjInD8mt>`z
z3qI*iV%5q<>JzcUjo$A2`+L0nT{uw5Xo{UqN;-BMYm(!#KE7!0`rPz2BI}@$;<#^E
zIp_~)DF2s4yE2?Fd%4<my_8|x58~aweg%6xf$jK>FAXv9s=3K&+m$$PpXZS`<7Tcd
zuFCt9i}!`C_q+Us*IO9hPNO6j`QmzdG@S;)rYFT)a6y9KZeQ@>2d_v^9+;cI@AO|&
ztP;c&w+QodP!ERna`KO!VvioX=YTIlaXo(RF8Yn5aG$n4Oh!T2`qha@=8vus?KjHD
zz-ey?I4ny_gq$3wjE~EAULMsFi>ml8-Vcu1)5(C*^z@nKM}g}yyN+O|qdwjOE<yn~
zxP$|5`{?=0A%M-<8Xg4wgx2E1tBWy<VNjB=#D>)FDU&DB$Dk*n%(YA|Z#%o8T;^}r
zE0+FOnY_;&!)qAte5@&w2~aw<?**F262$uhZQzK7sfP0L>v(Zb>$XUSkpOl~VM!go
znI0Jd9G()S;cStv66;8(`7<o12TJeC|B>sBQ%qr{hex;;6EQmy4zH5SOga%B>K+{I
zZjbl?v1k~dWdRqp>w$8`(FJ0ZN~e|%YAL9U_mcJR*}B!<tUuV=9Gexo0_MQ{z|DE(
zo-}Zv2805^HFGnjYQNT|D)RCOos4}gmhZX-wmdwp9ehlkz(wizej~c(_p(M#*+Oq0
zR}7@rSSSujQ~AZ&x?9&V&r9k<j10c-KQ&@VKRr#d&R4nwV_KDdbAIB4|AbQxEaB~W
z4~z4R(Z9<opIX2Z9B3qL019vGup1C5#}ywek(ZPEIWaG(FSr4fDK6Zd5**-fTV{eI
zEKV>|o_h0U`@`dpg<Xxj%Wl7?m`#5a2~|$N|Fx~J6!~wJSg@5>Zp*MiR$98%)je!@
zKR-(F!lisAC&kpK$~8z)eo99M=96oRw^D<uYNbZpkKC=4>LS&n2iRJp50o4t!-oGZ
z<HEbcpfXscX9y8t;mBi<9c8z*qtJN*Tqlj7rEkZnYdA&aa%txG8nyY%UsqHxkg~*=
zuaXgROiqPhK{ohcsi>SzHWBp@uDdUzoaN#tr}1yCds~*a;h=jo>cSL>i#O_!{b}*r
zLN@epVK1Jx;a~=k|55o*C@<<?MnT?8lkv#&PjT!DY&w(+nT2kMJtcxdQWcHzffjnz
z&oY#VkpFq5pXGXN74iLz+ki3Vy+hC#W+T=3;dJN`d@CD?;J(A>8lftgHy==C=$8Zz
z38rCxJa^iUFfw3!bAF9pDLj7#8cX{RdAlpd-*pai9u9DEc7>sk$lq>TdBiJFM_)_}
z<2OPALSMXbw|58;S+O`ehy?a5SCnNO&3Mp2&&hr&);!H#7i+$dNvYzDe3pn5HRNl+
zo80R$a{AnX+wj4603c}&2)yZ!?|S+AzHmf>Hp>lkC6^u8=$@hcUZ?$rokW#Sb3dN-
zQbqOx=r(zBYlrA%aCy-6T!xv5&E&dI^wGw!mvt*$@7`Am1ATpZrx!3<lZ<#(;OTOn
zq35Gj$(fX+qm;I4Fr>O*@7haG9HUpkmkcw4<~vV;VPI^oZ1ES$1SY3o5s|5u%GZBN
zH?pz|-$wo$8A5qEsqdz8`Q`HX<p}M}1zg|(o2nG5pD#4j2>D}rW3UlwXVaMFeSGE*
zkSKUMS$y#Nu^;Ca)%pDOYo8me^bIwKpVC*@E-}3j;?7G#6()5UtGNaL{@)bVYhU-U
zC$$s6_1VtldhK-%Xo>Tx4pACI-;Q<LirmG?K`h^)(S}vie2U7-5f{trp7RmtQEy;#
zuHmh*?Sy33Zy9wqUv^|gb|E%`N(mAejc_W>HPTz{{&yIEK{;#LPECEOdrz2yV<mES
zWVC>JNAVVTZ=qq})-^I65ES%d_L*(%O$%L#DGc^@^mjz4d<?u4E`{E~S(I9HdJ=sf
z+pIpi?q3q5VwE*^muE)g24P#VV*nSoJ+e9AA|EL(;?FYjY2@5jE?nbhy#{}2X12{L
zA3W?8$8A#!{?bCxspMZ9kCp^r0%vT}Z1qQZk(K-nCCw6>u3*TMGM3SX#1s4{k)whd
zW=yrLpecr}zxzGa=}ju%a5D7tof`ZO_o-$T{CZH-=)g6KU<<nUywuIe0MLJ3k<AYy
z(oNOyIo!5YJi+0x$J=Mv&B4RL!%a`%u+}c}3~hG-Qoy)UzPdfZ_ElSW7EwTI;dPm?
zY<pdJQOeI23C&bQl?#r6lGMaR!la><X{*n)`0QuetIYZ!ui(a4##-3sHu`&E9gS^`
z;fq%efMVPj2U(&;`43It5uV&~<r@h^g)4WjuOf*rxk;P*dCpl+pT}y-5m=kG8!t3P
zwYaOZFMd0B#!{C*flf0mfafn?c7DDp=F0>$5g#6CoXnTJGiNhT>zYdj5zzPxy^84E
z`!FgoY8+SBlfoq3h?Ca(k><x<d#7~js&y%Err5OHD6_q-GyWx@k(V^UbVA8Io58cz
z@dRkTd$6&B?}*sS>#Kf9wEOJY6<Z;1SC~zF6NG3l->4|XR4-F=YTb~(6czvI@e8_p
zUAG`B-@@Q<d^}7XqQ8iY^C_G1ecGK?U6~Hk|M0-(9-T@$y)~Oo?s!ls!ZZ!(mu_tN
z!IHsAwm4lI)}s!TLOGnqENM0UWf*z<Gzh(HGArv_kYrTdCjIn_#u$a9_lY0!Dz1G?
zDhbU8CP|aT+;hTws_$6uF0@fCe*3E;lVLRbT&n#f!Cx)$AYaTOFY0vlPnXE$_li*;
zDIO*;6|!V~n`mXYAmY-5wFwjY`qQ5jTIKNVtO8o`zchh>(Nfs|y7g9~kC>+2Q0o^P
z8%y?IzKDjL?Z4on?r|f~AA_eP_+!Do+Ydu7InLON`oWAx%~1#E|GFQ=!nB|Nn)!Me
z9kscuelmfpsloe0u~z&fr=rr*5oi0(tVC~%9u3@Tzdl~RMO+9gp_BGOQNH|94o=1a
zE5e9d@^L^o#=^CzwY{BNRvQCj6VVgqS|Lj7(Up>y?)>Z-<~-lAmOV9Ybs}ujxwp_2
zPw?mM^%m<D)y-jnB~{|8+AdahgL*ffZJA13#COnfrt{{%R*Ah=;7pHdn7>?Bu6a7d
z=Y+`AgoA69-=XUKwUvHc{v<m6MF`Hd+y?A?GZ1j{>*FgP9v-qHDR{1nvjvnr_W(h&
zVeBet%=a3^(AE7;AXBCO#Ory2OXu@!YHjIPj;VF`M5(Kx(MZl>WSEZ@Csx+KU#1vo
z#BuL_`ObVZSa=n^CukSFRGKyTOQM^=9zjm{H8Bs3V^Dh$q*Y<5l_VAJ;#ijEJYi}E
z++VqeR}9<V{n7g@-QB4{OMLSrDrc@aq;|yP&bcC2H?f2GS!$unLJ3mSR-MhLJN^Wb
zWGWBmo}0psCiD(HEfEt>VTY=f$<c~~60AC$bC@_S=~#~1h;y+g(LIY;5*d4t^O${9
ztFv>|U6Qw~>{UAp$D^X<)4kJC)%O#?G#!HTb!bV-NolZ3ddX<EWv=z)P{vdLbhG&D
z&%K1_Od)FM(WJU;QN0S1+qNBh@gbIZnZW4@py{YH%0Kn<(_U+^Nmt-_=h91g3$FBF
zD_OaE+X#{_y@Zzt0dmU1sr>$W3N^m|Ez1v&bQ$hqJw`nI*0t4Hv<EGm@7y5-V*dj@
zZP-jIHI1}?h+Cfey9tbH6$r;SBvX^qhK{l9swGPA3A!1tBo-8-%hVavF~s5}gO*$B
z?d@oW1}%6vUDDgocrvrh)n9yfaN9Ov3(H#<uih%SJlDy<{&)*So_xIWx!#S)63IpF
z&w4T?Ug=`nGwIMn{KZ|zc^`8DQ7EBNb7fOBxz>+LT>hAxZhDm9<AUqI^PN#kTPOHv
zzq+YRLs^qhed0S~?){<!b=U}b>`4;67EUxxCmX%qlFaGQ@Bb7b`fYGC(mnR4hF)@)
zC`AxlE8WwJpZlrrJu%jGEg@72lB6_n)<et&)P7%z{_Ud0xw+z-EBOGgm_mVK0sd6l
z0a<Z6(9A8m>mbSuaV|-u7ngk6O`@7KajW#7Q9|wjALFn0;@gp*wb++lf|bwsP;$yo
z+JBfsVo9;1e3Z5fKE>}zA0_4*F%o#lBP_`XZMW&J+H&8MSWYyV)kZCubvWwZ<%}m_
ziccwxYK5VygjRl4S?2>;5m{N=^KBR3n+oRFyq3@&O>!m;5#Zx~y$jq-Jc+zKZAah1
z%4CSYS-ci4$p2tv11bZ0={mP<&4QSFilgm3ABdf|XW7XGZk~cHkG#+W$kJ2K_Ag(b
zr94RSf5r^cw+Xh)mat5Q62<-Y6N(=ZX+QjK*0KM*wkc|PYzEkJaY5jMH2H^q56bbT
ziVK&!uDp9^_U_$DNeakacr=YNY-){Iid7iCIemLm+ah$+2b|2C7<&1_Fl(OpX8V+4
zq&TJ;jWk|STRQwo$)#J&R|%#8Y5P+@8+%GJN){m`y66RyzfZ^AcCQxb_o!wK!n&cT
z_r6`f+MWeB86GeBmI;%L8Pnv-^G+~sUKGXV@3=WM(w^#xVSA~Kx)N;2$<1o4!!GxF
z!9{}K#dZoThGQ|#V%Hc$m@6=#p+l`{>Up2&HP?!QSMW3n3q}T@-<f>!L(f-$?eQ$T
zA?;$ZN#o>hdTk?$<P=a1rKf=2#Z32YcMl#Oo*e)yXyY~Sw385F8->>TodsO5&)0Uz
z*=?%eTShrv&%vbd+?<J^rcol_)#uxa=IWWZt)UEJpej|oU47qL%V_WgGVO~1DFqyE
zWhJ;MhhXcneudk#m7K18%`V0$##WANk?H$_?xJPjhBe;0%H#4x^YFqgVos1W8(u1+
zIN<U3>N|HUAGy2uJ-7Lk;-ud=jXHbrBI)c51j$t4uOd7{+D0KnB+t3oMxT3!J9;<@
zc)0YwWtnOVshzbDd@Yy%z9KfjKITpR*SUP#_jWZ-YH=a#e8qwAFsg1@jI%~@qNX8N
zX}xBL)cR4ARK)0LD+mJAAV@B<`^28Ka0MJFd;8BBw5~n?pDZ&?2Zi51gL$}Mx9H-u
z`5oEn{NLR!N*RCoIEgOF6iduT=sNW3=}K&9V)0}}H_4MEk!11ozWtgxPfx2Whg!)I
z^Y-wiv9~~5ecR<i2PteaFywqX(~9Ni$6{?$@Q7#r2b`DR_<3HLir$rb#jfHDAqKb?
z#1a<BV6IPX?u)7<cQf71y+fcv@`s3TRP}EA`SwOr^daW4Hs$toz-q>Qd|1>+u}n>d
z@RTwfI|Ktu^2c+~AV}`BxHn`P$C}<5ITNpS&V9C(v=9a3db~)G5aXJtbsvn4KGgzm
ziAYDzjm_uJVXR+-8J5_8iE*}!ZUN%SOMKcpv1c-GG9EpkIByJ2*yGZDL6-~LBI4Ee
zC1dCE<7j|((VhLIS7)erJj9{LfX1eHnVO6?6dO5EK&ivHuyRDWb;bvGkeoew)FcUU
z#r=M|i!u*iizmCkb73)JDn(f0+g)$E;$4Mv!#BE<G8L(oi&fcDUC0Rjdj@0)+vVrv
z_+sVMM#A{h!B2bDW7DeU3tVlj?9AF~s>S?jsbLhqi#WX13?DD<eVeBYct7e-p?dwA
zT(4qnZ4Hc?a2UI)cQ(l?D$aQ~MYQLOVN}0HZ+T3wM=lB4$wi!l!?l85E&gkl28V3;
zPMeXXdvW01+x$x!O=Jn?0Fw;oeAJ%L8<wch`3~Z+w3Fa)YYjxbR`Hx0+GXWl3F6+p
z`<yHWh6aB8MST61Lz5Skum0Z%AoE6c<>25zucEY+oBeALp(4Vp8h=rzQp}eucJms>
zp>e31PQ@N%Q|@jF&or*A@ucc<lNeKX8l1$D69gF{?pWXeOAI|^n7*pAk}p|dcs^pM
z{Y6hR*8u&BmM2K$lG)%Z&CSp%HZqfh5U6yW$shuU^XI_V0bvX%Y%(i`BmMj~jw}7(
zoGF4fZYXvYp^}qWV@CR_fn>hBJ$tj8gGWr#I#94?ByuU!tmE*tUHIOM`%;ZkDcjt-
zPB+)YJAmESv9&<0;#}jhr-~917r(mVMcPj}R&vG4kZNklj-f~W7mr(yb>uRrw1s8V
zetuM^0v)aF=}Ad_-XZ%xvPU{E_g{aon!!8LyzX0jE;`5ndRRU&QbWpp?VY1NI4WM-
zilQqvS}L`FblJ<e1x>T?y>-+}3)_H`bd$bB`U4sqFCgGcQ&opAS<N@QB%4ar(;0ot
zZoa~Ati>(v66Fi4Zh|EJo0YfoM33&EQ1kO`5#^J%RR<4z@75*9_>_sW1@f+#9xLZQ
zvN||;9dRr~9(QrVS|nbi=HHd;2RDihXkoE%;=50ny{BaUgzOI@fck)MS)9k3<9?ot
z1aLQvtxZ_cDbCll(pb-b>GxU01a>$i!~8Zy{_mStXz4BVVaxU<&N3qWr3C&ApTmoN
znnuT+jO5oFbNkV#!1d;N2&tjm>eoA@x~ao3F4bSq)HWy)=^Dh>vDbShedr$d<7B>c
z4h!aPj;dtMd!99*S1@9D7ZG6@=J)BE2v|NNa-AgIn2JE%Twn3s+OdAW^epe~;M1m~
zXT}z<_1%ponkJeWgSFU`4YfV<3^3y&==i|P?!DURqh{bd{c7W82^elLbedU>H!7;A
zvh~?rP%mI-->I3G^>;hUlLodw1k2^;%lNHdy=JYMpF3~gx%^|+ahx(U)4J0J01n1<
z*lGXzVZ1SdIVwR^{tHINq*6>&6h<@I_<f<v>K4H*ElQ<Fjed=HYP>!ipQp@Z)!-GP
zmp^`Jnn*EirqOcbB2>{KQqggKZz4Wr?`Pqxi+C4L`bJEHtwKNN<W@?R8nl9@qY^`U
z;bldrfeK8t6<AYzc4q&G(aNfK6@W#)qe6&Dh!`P^*yG;vJZ32LL*cY&?mDmKJwdQ@
zt0$_gqN>W5E`71BE$(HB$oCpLc`mZIKaOhfRvd_7-oK~N?M@toDbi~EefK^-U~qx6
zDVHp-LGR)RH=6|<UOsA&J{!U6Xivn}l`lOZU2EN1-w?91VRv&bf|r?C147>9Dv>E+
zFgLrelp{|isPbikDCi-H_J2ff)_pooH%*eLmBaUX6DC;?JNCy{YOT*E(O&TT%rmO_
zvn^o<NyAEAV9uPMnr*ft%GJRkmZ7pzO9OSZyzJ6BUm&SplAbZrg~j64sQ^sa*VZ~s
zo)31mYtJgY^meRGCcdQtTs*`6G0;J-gmg=RFj2-k_ukongK1}du-nnEG<A@BEaxBT
zytV5a?cfmYFv&WYp2iw|`loNSJQ6A~GJdhsKED;X-FXuWQx@~XCSM47h`)Y?70846
zgAT#pZejU$ZNBUS_3|M*lq<{hF%Z9nDYh}$SBWZ*0#Jdib9+E&HAJkzcpXeEgsxYV
zgBHhauj&k#UgFdT0ggA+)G~ug<Pg@GvLmHXR#Q{comrcLk$)LnQ`6IW4cHM{-hO|N
z_&az%iV8gvvJGY$lI7}|fqiU+NxAaZP-=EP{uf9!J(|D81FVV%hoh*Vh=>SFUILjs
zy1_2`(Y3$^ApqE{>$Sa}8Zb`IrYIj>OQ4UWAmR_rD#@ee#m>%5+AoxOr6uYpoKMnP
zW%QMg74st4Y1qknDGWxjLr1!zMw@|yh2`jbmO8q0P-xM}cg=}4E}hg;el~LczW48-
zDzpN0Ldcqp-(mBXRq~29rUoDm1-EqvVk9e@nD43w_`NaoAlAV~RNPhH&5u9e@P`}B
zowB`zK`H9qL71aV$k^YofMMYlzJcp~j{*PfUv<6cG<q7sZy#$uEJRvm8i?Vi@dL<y
zxeBijE$%OPIJ#uik)&ioFaC3{LDp!GmI<8aZQH+>p>841YqGoQ>zRmd`6X@Hw@{s4
zLPBT)Y{dI`ko|XGEIccuHT~c6F?PAfvq=hPgB;v!uc{PXb&XL<un?GP^17%HvXI1M
zjh&6vQ#ZZwrauSB2jZK<@?6ZWK_b{&<s6nx`s@~5bmWms`qA}|o7K05xe5nwMcS<!
zl~^gPHrQ{m|CC};ON|je&o&`)OrV!cccQ8T9C%^$#Y?McB&6h<Vl=809bF65>;!j7
zhm_k+542fiqOX_p?7~7NA2a0eLn(yyP}ILPM;R#Um$={5%=$0ny<?P>apfeY_ARMr
z&T4IaO+FAnE*ONSg5{Cxxf6(M-5>SKS;5~{$HR+@_z0RS1lUC%n^3{PlepA49g+qR
zq_<xeav9A?5o++O!Lt_Kn?^yfp{9|#imOd$*5h<%qtbS3fgS=e48u3dM1s5^Q?jjU
zTG$sFmbide?hAUAm(73v<7U+TKbp?_k;?b~<A=mCvNIwaPD02*cE;%#$Bam}?7jCo
znIU_x!#lI&P-ez4ieu#nA>>eY#<698@6Qk4|KR*`o%_08ujljeNammLkS3o#14qh&
zs2H&XYhTAMjEJE8U82S3peyN!^FVDIL*>}RvDe&}cp}P7aBPZ8frSE<xnbvX{q=;Q
z#TUIp2wDg|A1I^i+WkTeq?oV!l~5~_2gHu_DsnN)Hm|+(21ky~lYr^9wK!zyW}9Nb
zR?ysRwUFa=8vB4W&$5YTH|Z$>i)s3*-f!RTxBT;$ml&Z{f-v{p^!_Y8WA8dPl`VT*
zZg5;dnt%KF^@G<bG4!Z$$?c;vWCqk3AjNyrpe6b27_;n6CYpg>Y!Fz5W2vZNic2Zt
zm()lJ(9l6$8h2$wos(2FjhOqk2*SMh#HJ<|eJ9L^YDRd+>1rWpcV;vqhauu3v!tl?
zx)XSa%vHM=kb|nL$4%9tpr5~f0bFLnxr^M*#c9jXZHW@p>9-(-oq%?rBF|xFhE~Xh
z%dF)sxdm)LahjWISBA=?xg^$tJDZv$8Y~=4959on(mLTQL--qXkNvLZ1INob&`awd
zs(Dp3piNx(hyL|jOQ}jJV-%ys8k8z<FlYj%bbacYs(8F^E%rJ1;1~ct`x&Ir24rWb
zO>8`?vO)!)t|3S_n0U>Mj8ygiN<PF*80LSF&!xL74`h=rr&6!?)M;MiOiFj2eo*>1
zma248@i5$UPt(AFnglvr`H9nnBMZ0e41IbTA;)>lz#f@yLVx-b{xQn>J28DoX4XOc
z?SQ~-1~CNRZE{!a0ij7}VwClOVE(DgELX=<G#_R5f(LHaGDeZ|*11}HQ@8lh)ERw3
zSr#0`aQ@E5&Wqq{r;_FLG7FkTQMNaQ!)E%b510B#iK)B$<v1nfSd?Y-_RHMW`+1I>
z^pV@5ZGVeS!L>sTUCM6_l&6_ki^#r1P%Ju_er{&R)46+<x8_zGTb_;e`PPa0WtFmF
zGf|q9h!jzMMj3JP#1wk>M{mIgA6(T_@g<=r>SpJejT01%WLp-9UJ$JcOhRl9P%<1I
z+11@W%o;N@3r$e-u$`NQYN;61O^0t{rtND)q@#O$2-^`Gr=Ges=T~|tO@Ojd>E>?4
zC!0b7AteJ9Oa_vrAtFtb4?*<dU|{~i8ww^FfC%oU+<O^M6-DR29Pcz&shrmJIq33=
zNlOWJy_&0pLS>LYCBZoIzy<w_AJ!VQaA2e@a>v6alKkWCWm9V3MQ<MEH1!N6r)6o&
zv%kK)L<{S4$@oap7l`fnk^Acih2V0sGxnKe^}&%zKFum?rmi;`NO_UAj47zM3&dD@
zwwatqYKlMSZQ*3HcW+ZuKY{N<1o+4(RhZ#mhA4U(Q3R`PTGYLpD!oNKk>Z-0HE!V_
zNunA$bs)NGp}-9sFH7!6pw5iXcR>u5zKf4XUvpozUq7#ZA>X0tY}4^MbkiZj&Y*g4
znP<!b^DUOz6Zp#4PmD<ZWkRW^xDh>4)hHIu2`4r*Kw98)$cil)%vFZDR+cT<c#76U
zKACe|QY5k3eG4cZPEQj<h?&rQZT{#UWh5o|y?&&LCidfJAEk{++p6$)#Q8HYNfuCz
z7bHUfMD`6!lvt6KMuTqgNro`SU7nNH`1x~(mfCU2=m9mMvV=4g$>-Wl$x(+eQuieH
zxedRN%d?~F<<u*1-t|5ZppK;AmhoMsDUD~i3g^MP&(#Z006&En$>=&49X&P-)K>4K
zF6wHU82%lUTwl0)c64^NppJi!{_q!xSr2H(%R+%zuDN;9<Yn?U1~5WBc}+VauIg_g
zyB0F<F(H}li+V#%y@Z`NF?{Phmp@}b^@z!9XJ@AuJk|=ZXdyZi^ihFLQ$bgk2cwGT
zD+ivPo8jy;)V<(%cug&)!h{IaAJtrSo=!KpbMWvwW{>MkvXwnF54!$7``fYP@=t2S
z#WQ$8AqSUih@l{`;0eT@;2#AS(2z%CFql$-B40&R+hkw%m{a?^P5Bd#06DDH<<UaW
z`F0Djo!HrR+D-(fwJQ69FB2*6-RNVh=(RDGcv7&jSjq$4?@y`adK@Adqx_1IGCg?;
zw=S%8pqu01oB>rEtgdFGELR&R0hLSQ1nxQZHa1QZzlp>w;?Cvqt!^@P89=`qeE|8l
z7>@!)W-}XTF7lAEQ8E_ACzIV5C^xmH8EYa)kwSEmU@jg%1MwU^U!>jVB#k)2-8S3u
zv|+fmcM)gfAC!VmN6^Tx)rBM688h^D)s0e`<)(Hf8Gdp|!uApQyyZh;KyE5m$QE@;
zi51($Rv$6^aOwkfD;VHR!h&cmkr^>DD1GD*(|a+2UFO+l;g+5G`I-3)({_;a6a$>Z
zmZtmLpIlR|RVVj@_rm$iJvt`^*ju+dC~n7ewnj{!=2dNse^DUm<w5e_-$5;2z2-jM
z{oBuN*~uIGZ9<#MfHAX|9tpoGjHRA>M1O;1^j*gQ!)=m~WsoS37TE|heUUPU8|p6B
z)(RB&3g3_o=YKCcH%@-?yJ-$$*7<4l(MsA#lX#u1Nb7`!ywHA%FGKcD9cLv0{{8EJ
zZO(zPO_74~HSCC7FiOTs`~FCfB*r}O<DUwZGA!Ky7(z?&cyTdYi<jI}CD2e&j9;WI
zK98{yOvP?HVOSRJ(m0u8dr!5%5oJ*5VsL{;BZm^QA0)23My``o(ZrA3@95VgCwDAM
zqk1aUW1NX>=(42?2yg2Dq@1Mxj>Ye`)8EpqZ+De0{~J>ytxRZkf1;h?*CO$hCw%_-
zN}lY6<MnBmLStLyD}iw@XP8e?GA3~qP0gyObnz7NUdMQlS)=0$;F*N}g`qI9j2q!}
ziuClE9&1CBfG_O4wOXjo;*^a{lv{2RZDtTWd&c9}_A#vdT@f{B=}G6Wj=+gL=%_q^
zRvMn2sw_{7j6Liw@5T2>5=WhvhQ{Xi@L0sU<i;{O0`>Xh=Mblj@9D8kWz?@zL~8Gv
zm|JX)(%tA^yi8c?r#n7*g94DvLdLfbqCM&gC2Wo4SzlCEHe=TE>`h?k*|v-GuJd+J
z)cG94OH*m9Yut28ON&*+sXYh`adLOKtwdJ5=gsrQ&TrA;fNtS@r{rR<>rQf&{bVnQ
zv{$<1I*$Jc$Ke^qdpvQ6;r2_j?B?3ihMwl>I*X%P3)7;y4_}&(P(4?cjUqDg0~4_-
z#6O|LG<~k_Qw?59+608Mqfe9kEK8gp|7~v=;Y264M-$vBTJHL+d)g{|W$$5_$H(Sz
z86xtpq|#;g6M=6#XWJ1uPLTNWEV;PvS+>Ty=CGR!jB1#bBM##!X;0L)nU(t!nS<RY
znQfgHW0T;|9nEZ|D2W8}<pGj3R+hC18qkf>7sXakR-hY-#bCBuuXfKLhP6zm2kuxR
z*Ay#$+`7@%tvH+XK2IuXj6#04PA#-=k4b&Dc~<GCZ|%)512s*Fiq$-gEXdNkZ$eB%
zrx_{9JxsL>1R=Ig+68IKpxqv7`}(U_78B1^jf}8;x92|{p_aR6St`7A=BeKL8BDr!
z@G(#!h{7s%b}GV1B!nJ2Wl*}WxUPVClMv{BDAla<=&?qFhld=f-W}869xQe>qos5{
z1~@zZj;UK-owiC;rC>zp816-uK)Ut8P~P`h2FRQ;`l{)i^RK_BaI3>gVSC5wtY>mQ
zjRsWj%b5|Vf$o<M^jTz*Po$O8Xo|P}%(rV7IFf0jQW1>ZeP9@sW0_PXCByH`l0OqB
z;;Vb>%Ffx~{K}bi>k?3H#9_~qRD~D1{rKSBz2$shB!uqtKM!yZ&!MxJ07LR%<ZCd}
zk5dJh7@qjk+)sLH<@w$;ubzBVt3<>kF)-LF(}Htbg~^z~p02DN*z4uY+(}_FKC(sY
zM%LU5VKO_EcGk;E`%TjNH|G>(1nZrh`F4PT7F<;plOO3_bigV5v)O;ex>h1@(-U)$
zbf5|cL!a)W0~anculG}pGDw5s8{6Aao`DCEuergnxNhnPZ=7x9)XaKEAD=GcaI0sl
zaX;#7ILNaMq}=aOVs4E_oPID?><C*rd-VRGJN66eT~it(i5QOU`-2G#hkVz^O2f6c
zuSZ2|gu}1TQK5g=eF%4t4~n49mkEX1uG=Qd^Ms3w3!2m5y+3_xCD%T4Q?F?;QaZkX
zaHF3q_;7KN2bYJpG^z6UXjUQ2%{}R5?L3*ULtPd^V%Tiwnx|$P97mho&;Pu6sJK7C
zhs#5;oBPK7<Vfw;6oB+vT~1m(KKbR_KuBq7u65-V8m^z)rX`OCK*R@IGYy*WtE=ID
z&INL7SuAy(XD;`N3TFd+?d|Q5nCR#m82=;~?nRE#oYe5}@N}&ume_e-SqAexggEiZ
zV+p_~w<bC!UQrE5`Hpm#m=wQWd?tA7JBrRK?Dy$f*X-~<ol@Jyk7p%a`(rC@fKJ$y
z8r0w^Le|_|TP+0pPNb`DV@aTbn78`G+U)}u<c+U%O6w;S4V^g#)eFSYRi=iXCil6O
zm250#`||P<s$)U(b){*~PGNx?G;v?LIUYD%I-nGN@Y@P@X;s3+x3{$&KfZ6H@|tUh
z>$?|ila$NLUEmzDL7o<KAC*;J7-t`)l(kGbLvIV5EzXccEAy7q+akW3#C^kua2Gw*
zgp$vxl`hg3wCy-eYzJ&M?*W;5lCkplbYrs8!_`h`9x~!x-rhz2CF~eJ4%vP#zsU*r
z{C-4|_hjOEWc2u1gPwxEao?!~oha6S=V(+h;CPH8mD}>KUcl6wvEm$?m(H@A0saMV
zbVtjDHco0z)(Bw-+1ytjXHAH?x5UUEplFY)AmpGk`|vBjUOtdAO~LXnF9A@Gy{d&p
zSAfYt&^i9@!cV;Qzx@7U;7!D(Q;96_{_AGO0O28t_7}J<6bYq|%wI~fSK?TCq_Y(D
zNZ5=zCcPW-wi^s1B}1^pP~Cz}`X|G{5M`g@z>ClE0y`MbhqvN*Elf=Ena1@{Rhezh
z7Jk5ac%NM`fA?#=iObK{N+E@cC=N4Yz-tq1Z|}HdDN8_`nR=iXxOEH!6Wdsjfd)T2
zx(u4L*4UlO+NfjJX+0j;1G)c?uarq+L5QKYR-kp=%SM8<f8WyVObxLwaDLCsR~47I
zC-bklw#rofo?u+FsJb&jOB?m&m5+>{``uT3=gp$uy*c2&pWeT;tkZ&j3mH+>@71`x
zqUqbIVCMXcxo=h&h%XA;{hpppM@Tg?$+kC{_HW4q0%cP6o-!H@<?sKpo$mM1J*TbX
zH<^;m<S_R9)%r<PRZOi?t}A$H$E*(f2(8O%eCAB}M8J=TNa{2+V7*a90(PwDYA>_D
zYyV=9PVsXQ-6~w(AHNbzXvBElp&Ppaosa!}F+0XYS0v}Via!%kJv^T4lJUL%x)pB5
zip*fMEjM&75KHK>#+w%M56S}+nhRg6A?>Ib#!rRzdWu)QXY=jRJ^xK_j*#c+Or&K1
z!z6Ose~Zy{zyvD<i{iRbQ&-0{Xrii>vFuxN{!8m2R#IB}<lj;%;60HXx&Nh!4e;J&
zpSJdSw6wRjCO{j6KUBL!L_~<^f8(Ojdwld`v#u)wh;7Bj-V}*}RXYD|A%x_{<BJp@
zbEP^O+6k@z0K|uS$auD>Kw1XYwN>ez+7g9epYU)C4n8tBl;?b*PLizJrnw6<g9Z}V
zBa(@hP2)5KhpV^j2VLtrFCL7BZ!1PgkD|4*>3fEu=<peKgofsOKSzH2{DKmXiu*GX
zWsB+H;*;~snP|u0;8VgGkPq|ZRtj0hkIeP-3y)A-zBHQFnXofq0SL{aK<W!|BPKcd
zRz;F~cm^tfF*985eIZapxSpGv=eH63;b(e6AqX!@ax3WG-@h1pqJC4&Dm5JgO6}%+
z%tKX?1xRi_Vp>L$;3C-!&<bVVG)*e^@s1CEvyDwz?~LGZ3I|)}CkRMk8^jtyhGgf}
zstCmrv|pDusW3tf<9nKN@{4@&Eo84a(Si0xsMqnA9-fU2d@?wCiwWT04Y7!~{#jB8
zJVj)^_k*74Fk(`;0Dg(@C~ey*VG8TeF|B=U{){pM&d;T)?1t{z+XISQVKC<NXtE@m
z)puE0F&W(gkl3f13~%Y76}*fz692J-5ZpJw`qG^OX3pRKQ{qgn94iYHJ^MFkN$-X7
z_g|UoI^wuKymwB#?!Qh*yZJSLso!{Ym8qf}u~`txv8ki4j_WVJ=rD9}3}j6QUTm&{
z_FyiM1}Vc^>nPCJO*85q172Q%d*x6BEzHMm4J&0Tl&~KC0qzy^N&qRKLf=8x<C8>?
zC!>WmPu-9p@9Gfqd)rC8V!}MU--lA+FinYvXQXM^u=D{EX?w(YFl%fMze+|!7d_Qv
zGv<5#E#3{9H<j5)e(hqvRm8Ypb%0nGeYSY8v6`9N!doYIi|KO;54leh=?@b_#Ba$&
zi~7z<WxNnT#_LR&&Uw*jJ~)m9&9;ea3PF^kvJAimLJTo*C-S5u9(yuKIf}t9af+2i
z!71DCzR~37t+04_K2*nXA^`MMvv0uP2-4%E&>D97b<?=cy{2y3eW}z+-p?t*<M&s(
zT4wvbAFL%pn?jE$O)lC|=ks+}vj@Ewxv0JCjuS#CKz8#icz(}0<CdsZo6qLiEs&zm
z1;Qa??4PykCgI}Of9|F!DkxX)O&sPjdh7|^V*_R*3l`_%)c%RzUO!Fg;71`^uIhT+
z+64>e_`r_3T64uN@3qyiF4i!oaC1{jL!5vuJjr0gQrbA9k&!zL=;(Q0eV=v7cvmUf
zj{X(;;O2424Md0k=JT!v$JOd;{ihkwhIgZ_zH1Si(x`^o!uskOUlcN9he!mVe^{&z
z0?6}s1s|*ey_xB$shXy#+3IQn>bm3k;7o&_%vz$&IQJplu{uZ@;aq_Cn3Gvc;Z}+`
z*_mg5Sy?><@C#d~H-R3}jis@?e8)m>SNEz3hWHpmql&EYR;hn%VLyNVdLbXSZmm$8
zh%68gj0ewE;u}!RJmO-!Y5!FMn=ci_su+8!YzUf-<Im*@y0s<3n8D>O?*(HBQ<=e#
z?1$tG14c>*53>i!LkGem_0IY$CMQd&bu_QPM$~m-v5(dR>1b(ftclU|z<mP%5iN9t
zH;6tUQ|6_ICM$%WX&(y5{G9#8h8OvDJ-}3gPf;YJV9OPfCWDYf=W0uniJ?Y$(ozX#
z-24rc=i1gH*)JV<8S;Djp_2#!a8w{E1oH3SmqPkKAG@B)aN)C?Zsz;1vC>w|8*b3V
zT8f=1{kv2WP|uV!a(~<zdQfcW;(hea)r=hcUt<0HFIV^;r$ArNHA0+Ql?iDABbgkS
zhg6W)i(bMV!^l7?&Lc+3B!=?MLI9AT(oX@Mh1rkv=HlYn`FMN~U-Z1fB?gYG04SZ7
zyMCj;g<gNv^i&*H+Y`bSv%Pr&@8(n26|LnY%+BsYY`lh=2#?5kvK&%Ok%m$1;fFMl
z5PArRTICxIryP^;T&EHIQk;-E1oAz66ceu;dB?*wUm!lE8%f29GA9H6-I6{;-$;ew
z0ISoQbj0>4(jt_DAuZj;j)yA!=139}8l#+LFciCD`7F0WEK(9_ZZ0P^+u?p!nDen{
zQvyc|<<~Dm;+Yu+BDPs<{_%^Z3pvd~rb)rG351lBlg3wVMR~o;2h_Fh1<mz3d=0&?
zrC>9E{@7%^YZEO%RjTALsvf?OzE$BoJJxv%k&Z|sjZa8d`O5}2fhlJ#w~Cq;9!ccq
zoqFU(+Aez((5|i~*4xn<+=GR<VzU_a1H{eEU$_n*V|+dJ0$WO7v04WGPJhiEx|A$k
z9D3DSFVbWH^O|d%uCu&mb-d~eyk6+)4A}cS^p~p2P3~(Gis?i|w04V#&4D`Gl&1}*
z+unya8Xz1}2tu?8--AnU{|4&ob+vcpe4K!0ylYaySox_K@i?_4!YkbOS{|5#b88=a
zE?jsW;QNQ8%}eR|r;(lQ?b!PI(k61JEOBH;FR;G$!57Q=&W5&NAO8(#2Y3on*LJHB
zT??B>qA2svADf$->)RYK+X3ys!3y1|$c;D&L<M?!*3E6-hd<Z3Oa1B5m;Ya?!(W)1
z&d$vh-dlMCWaL-B`N?7ZHpbKW)ElRa>mtmJCxHR;`)sx4*3;JdnsG^V?4RNKBp5%k
zzP<fH%nJ((3oENG|4>iKJHS~fmh*Q(O$nlpb}&uO1s?A1_YA(-u`TVfC44AlF#xhH
zHm2n!{9<axqhAVeY(%Uw2*jIVaD0Bb_f8;@*GEiqL^opPu)FCIgOPn%OcvG4)ftCk
zorG)bR^R!5P0w-k>D0f6t*-VTJnY)rY8VCZ{t0>e3U!4ES^LzZ5x)Ux?8pHsOomga
z;tWezoeynxZ^v3zO`K)}!?8DA7auI;!uAL6%LmKVY6>deTXhFZs8}(RrvH?cXUY@#
zcJ$<MZf=hHE_64h=BA3v-*wWmS6=65Knj8fhNYJ!70P7D;IZXM5>STnOCV;OZieCn
z^atW&44xa^%fcV*d>MDi44`!g*zqQVQ1Z6u+yKGjp~#|J>5v!q87iK3(_<+_8kv(w
z!V0OH!q;4!z8r0~tdV2!UZ?3*#V^6qwN|>?ogE=?{#Fg2%y#(}Pq&5kruL?{*#m7i
zBpm-%yA;L7ZI12ODG)k*wQ@)0$)uni2bUYnXS!7_R?IxRJ7>7r@wZK7u5YXtuGgbZ
zbT@aN&?#}|3M!C^qG}j>RX!Kokn|UbeNs-(*Ig4O5eeZHF#?0w0-KMI1JqO}|AAsc
zrtA^Qyv)#u1bWE(CVB&*?9I1t${Q?m_`iiYIQs_%y46F#ts&y9;*Cc!JftDy(?3+1
zlsSx8mS)cpL$BCwH*LMC9%8~Y2&+A8a&lE$UNn0<Xr`IjfpR}`vKuaq<l%{NF>oH#
z?3EEMOEWRle3n#NE0ji>#`Xy8qu5MqqD&YpA1@cft<OsR*_st&psHvxh>J-p2nSE~
z58>EOdj};*pNZUtyk1O(0cAPah<&GOe;6i)VGYP~d>{L|6>J;cE_CPKvc%})@QpLv
zwmgSAT}fg(3i;=mRoLIR>WZOz=n~&u32F|ZjRzol(~UOHKsoFmYvQ8y#qzTf`IA*y
zJ?I4C*Do7H$iGkhDO^fE8yAJ#T_{CZR=znn-vKK;OySEeX8Aejcc;Y3%H6M=S<FiE
zCuh^0$B*CuB^7G27J{`T!^@ps9Y+-N_ne>Wmdp}RgwBAYvZedFTW*B&p03c}bP?Bd
zJRyuBo{qWYY&5U&QkgbX-Qr`SKfkSKz_t(1v4m_oQQiA|)>jt|RtSnE^hif$a-x2<
zgle7!tXNx{ut!L(QZ3-`3GNMA$bWYIZ=zj21wB~bD9llo-J1h%lApQnJ~X>(vGgHC
zym$w-1ElIL;Lj~4$WoxBgC51(EVfD7poHgy^O+4JN@vWy(zaS;I%#ygNsg4!N-1h*
zWAu8FLh0WzP|+*nzSPiUkJ3q|*<!bGF`W4kT*br}#2x#i*djiwa^|3Gzb_ZHzl0gV
zM2`_zi?6<1UG-r69Wy%ev=`cg2A5B<7?e&Xa3M^ctIj9Er*O*dA5Q39dtUc<jj2sd
z?oeOvEJSQPSTNrjRq53ev`mn25x}v<A{L#@7B=j#mRuHjW_izKDBZJSy7e{0aWQOg
zn!d)6YmD?73*bqaI*X1pNYwBGg9bmd-PU=uPB&pl`YKDU)FnC)LktL2*|70L(PKvC
zZ;ah-o`mAMH6Ud7pnyvo-4IV7W~k}23yX6q8-Lf7(nHNiLm>0`0P|G{+xltOSV9+f
zo;l9z^re}T=tkJ%h1|1%xtmsH?Chr-3b6CGb;rmyO*#dJdfKj8+dh_75qwx`tm)r?
zpf{ZcsUz6C8yp4G(ij-Pv9C}jQW3`!B=B7i@=&Y1O}&E0jvVjBe0caB$VouEg5Dk=
zP?oqHve2g9DmT|-;n9Z!Pm7L5rT0I-&VKp|+^$lNf3~ynrN4R?^+rGXjqts*?)8pw
z#x%yv5T^%orW{RE5-UX~itzYuoo7L*ruH%Q_Hm10CIyB!V%bC(NNgu^pDL5la?E45
z$NakdA2A6Gk|(7P0qAA~v?+Zk51Q|u8>~zwNhbf30L}fROa@2MR+fG3*VRpA(GTLd
zDM(HV=g4#DaoKo=Z`={437Jyy6Wj8S*e+c79PS$aQO4TGpFyteYa}7;WZSXx=s|SP
zMLJ)<V%uR4Fm;+*b&<GR_P{gz(9-+k@EtDHYQwPB>skx(Or$M_l@#sD7oBCgIpc1r
zFKra)af$)ZQ@*I{ONz4k1+%U4nf?}!>to{e)wY>2k<eM}n82f!F>7Za<dCMR%bzLX
zs7u`0*0>ipto5*C_Y0Gr0_^JzNuBJ~8JZ+-03L5ymz(bvP}VfJUFmFUxblz??5Mzn
zTH3VoRD21g;m=HXI^K+-FQx6grNUE=6c-dCn<6a@oI*)Kci-PvO67br`0kEQt4<Qb
zy!MnoCiVqq!1>p=CmtwN2{^G00+IIcT*C>x($?}c+ngnTL|JjpjE!C_qOMqi5-X-C
zK-}<$5_i!lnt^A_gT_|D8~Ig=fxo}B)00FdfNkQx5zZ^!$AyOvI18JrtMkX7gZBU-
zS8?dU5lUD3yq$P8GNGpw4g~8RXkYR-B-_g#y>i-i(_F4OA@tFs4yI8{0LyaUeQu;7
ze&;ff5Vknh03XV9<odVxC3djUuNGZcI7c897jo=uZ{K?Tl>)fl148f7tF2ML42m_l
zlWo#;t(E-cVwuGBbk%M~dDf34h6Ekn^g(5m_`p+08s3=hF8emYnp~yZjU=B@Rll6`
znV=AOvqh2c^Vk0_3tlqPc_q1O$o#gateN_c3d)fg0qsNU&7w?b{)Ec!yl3B^Q%pEJ
zMUNXe7wBreO_H2-x&>2PQ1hHy4DViS6)j}@AgtGj^e?5%taN{xrYj&7v9z<YR5gmi
zKb)7YR!HS%P%RL8e>jE5A72R|lpV;(Xfw!2HOhEnq-L8xb-FWAqRHBW`86ISnrddp
zvvpV>r$isn!$@%hJiH7L5(qU3eJPX%<ZmLAiYGy;4qT?NOZ!(EM&9Pj;OQMm>OOSN
z4p}%;L3zwcaav2xba`)|?d*7b#GrJ2F^?z<bY8>38DtXY)@9OcW&dJ&U-!%^Oup9-
zY`(8!LiG|3))WNZG1f-e6-}IB%)RAg8T)uSZdEuIZ@$aIqvmmok+IR;sZgjM5r1%S
z!~I9kRfI{_c-R+GX81RLV~Lu=v>~>#rs5IX+2(8mN&`5!qR|)s`uT_6DU{BpR{1N!
zbeV<Htd-mk5&oMpb4I+gmHFC?HmdI$bgh2~>15O>{47a1o<m;>!_*!`nrmxj43ONT
zw)ZQj1j1(Swrg>|(Y?Lj^)i@a+}QjAKe@<cqV-=!ew;J29{*xzzkL9O%?36!x>1|W
z32!H<G4b?q&NUy6IS+v;XHXpIlzyhP^6bBUKGA1gXP#G^*GH(UX=nTz9@|vkcKL&~
zn4Xm3w!{zr@HIR}>7q;t!Ok6Vcy&SuGkV&wi<6xxi5A$YT@ap){`$Z*TE$9(S+hVg
z+w=y=*AN<STQ#lv=|dy!7FW<><Qvw9K*VYi_&6SgpxXnRuMYLDQGOlmzS~2pM6I%H
zo>+~kkTQnI*$^Sh1BJ{Of0K@Xmx0_7>tlfW2cz?hsY9ev7b%NQYGiE2%y<{CO>8hT
z4A(;;K!Dr-cxy&-B*D%?8ozCC!UA_Hh|>t7*U-@J3;J77xqu9t``7JnQH!bVJf0oy
ztoF@BY=${eL|>sU_BG=_wtmXDL-$yF0W^}<mtDetr$@1Gn_920&W%~=EL#5^+d7V_
z<=w5lv{Nw9==8|eOxDO@J=i1Q!)iYazH{5;b&&f_0c=*IwboeA<^is)lF#wh(Fg8F
zeaw8JYp?7qrcDZv6?v^SO8Ja>RTEa%M^;?{+Xe1-5HTpW1-ZbTy?*Wm!u;KB;*CMw
zewjHQ{iO6@7OqM4xUZv1m$6-!g<Ux@YDiL#Ay@^vWg%>-b*<}YtY6xc6~H)@<KxDh
ze13mX#T)_g`|RlGXy9>uZ@3<5bPrkI@Yqt`=d<u%h0%od33eQS)GFcP+@eMrP*uF3
z>PtBfN3}NBGb+^7*4#91%(O_n{mI^%KWd<N%Z;PRmmsaf)KMtA{Fp+SK8g|3NDup0
zeT=yk+%mE4SLC|sW6Jt0!wz%Mqu#8h#S8$M1<JEDsVXX%4KzfupKIPXr53C=ys!3p
z5##>~tQ^Az&=D5e13k@i{Fl@w-qXExnY|nf0$wYr8b{~RS6+Dztx|C@WaZG2M%}D&
zs<$#f=jJ}6-=(psurY~DAT$%ywCq;bA#o;#aq;oWU=k1%sd-Z)y+>~CiIe*~7gqhu
zDUukv7n7aNd}5>$3?rPq4;~D6UCJ+9RXv=!K1E%sqFzQQpvj5Ro`C_)l`C2KxyH}w
zQN0&m#tBY#W^dxvTrtcLHrtpqc$#4bs3N9~^X`^A#T;UdFZv_`mk;bECN}PJCUvC7
zM}ZG?_3oWzey4YSI{_Y2Z*C^1jfeIA7o+lJz{9z-y`|xkci>X%LsSA8DLrrLQ#oY$
z_$+i5s_81I^s*aY-St${^$E26JBUpdC1-5wcE4;ySbwGGIUD!qYJAASd8W^I?ic_4
zY;37BU$vf4U`?Q87?)IWxCfSrqSLXMHp&*jl<zZ?|Immcql|=zH$0t{lzIoH7Jv;-
zT-l6U4Ya>8GqbfnLla7o=dTi;n`EqL&|{lMm`RRl`O&Q_!o;*JjncWz82Uy76vjPy
zs^Nuxf&HuUEN6_!@O$vt876gy(2i|rxkfAfOOY~t^P~Cd;`F4sS-Lp*n7ub`)9!?z
z=2WoPbM-k-oMLkmtj`gC_9OiA_wGe&APZ|wqO+`0m!y(3F}iM3PCOTQG}-miiK6W0
z#H73YR=-^WqptIjg){WP)p#t*D(P5CJ|PEQjx7DL#iR-B6b!&BpT)YKP9J%Rb$$N)
zebXNs2%@)j`m2_<zI@mE-H-v*yrfPB6yFL(rp3psABDidDK$+Vvn=%GPYnjzs_&Rp
zX_mUTib#kMLbe?juFr`q6E^PlD}P_-(mou~s#NJ(Xl&xqDEA{|tEQv)`Sn(C=)zw)
z?y%0#&9h`z<CYdFF8T9y{Mp^wN^YwNl=sV*;kF0(is>rp1)^6&o0vMBxU;q5ZRkGV
zaoQsv(3aZsm#{mK{uK}(VGtQDB7(t}t0R&+?ZH)_w(Sg@%AU}~ETcUO0~GE&z+@W$
z^)+d$U|?m0eH=;u1hu~0nqlMW`(V>=m6hccyNr;#L$b$OSW=PWsl=$lL0QSg3!X;*
zTOhRrVI5ux3Cs3?r62Dl{EbvuYy4G#Xlg$66i>V)0P-n(iBi|cj~~0b7B)^Agv~5X
zF)m7ii4@$yPTm{etC;H$AUhRi8>l3h3Ng3sk9B3r|FLuzC4$Wwn-+=A)?`!$!WtS0
zXb+(zU5*9A6|}sxh1b{Eikj(>@ub_qfYUzxCQ^q?RhD)vGfRA%r$6b?Pp&wHO2;n4
zu+ZEY^CO9gJ?f;P{7nNbX*rz9$70%m3e!YKl1CZKSdJJ-5<Y3^$XoUOpp&0b-T|m}
zOm*tDN7_)`a2y-f7ZL<@DF|5Dd8C7FXxMP)lYoMIBS8*t(-v!Z=zEJF`;w`Fc%XtS
z!s(=>43ti*ddk*7bllEbBh$~-vrAahE&<#?p$Pisq**eQ2;6|K{#<M)C*ztIU3eL6
z%`HtthYivo2;!yi?cKpX{Ze`bFEIWSod7}>br(owHTTHUu9ubAQx^I5r`$C&u?SsG
zSFdT3c%z&V7%SN+xLF^N(iT0bza%x~Np#aAA>q?3rLP&>?~AP{bM?JmrC6}AkSl=Z
z8@^+_SCJuPB=>gatJx6y59ff-c{i8RoX9^=jj7T`F%(6=U_CDsq<e!6zahh}m#aE%
zg_4`y&O_U{z(Hw7&Z-$Q#Aju0vVH5bmHfSp6A(yc=2)XgTr^kAdN_2KM*}>zeYCbl
z+k)|p%+`&JPvXynePBZr0D<G$A5@?$>v(sk8@|H<jHUBv=F?yB2}#U}?;9;ZRcJgV
zCwP+*7?A6F#TmR5jLmR%QSJtonHQ)NrGH&(t8t<Ox>Z$K)m82p<p9lL7MM-9jBB4y
z`M0C1ichnNSKC$%Qtu7X3Jd71wP3|*tFS#FE?siPdR;%U=|SA4U{RecXuKiJ_ch1>
zM7j1P2{v56Qt|b^-hvXBUXp&Gv?o;}K{qEZda+a+%%h2P&dIg6w}2jlQ2({f`-sOa
zva;kj-cVy)FG}8O`c7z8K}M!I#m1m(o*7LhDO+1`9<Is!!4w7p;cFl)*1zPO(<*tZ
z*&XA4FIaFM4P-jkqk){|X4#$-s&31lT>bg;3#S&OD#I-o`?JTpFJC%)8!{7EE)^IC
z&<MRPzrfJVGCC#g@ntQPRm9HJcDwOt*Y;l<F*Y}X>5gcL-|3&@9egs~LyM5}U)Q6v
zp^pcHEdc{&$FYR-FzeHgP;drl9bv$uip(;N`<vq|fz42x|FBWKJ|CYj^Yg_4om($q
z8;5ZFX1Nv6dwl$@@rR{>8G9n5AcG{p9f;ahzFkdalx^g!bTEr*U_~A&ef(FuUn|}<
zq_7ZNa!|A|QLrH0LhPRNv_hB(2nGhKV{HzGQy=^8eYt5yv*{EwBqP>0kY6oaJ#%_V
zEBP~>KFa&KvA;>q#O(#3zVKmG={l@orbrvNh%p#qXM}6u`Bm*LWIopWY~qruYpb5k
zusH*zi2Ga$rx&P$QNE`+PAuUTIs9_PYufzGG(l!G&D^2ucAk#Jt(j?t)8Aj!i^Gor
z-M8gxQHv>N(5}9tqoZkhc>EXpQ}M{^EkC*VKxCP^9Y8n{Gt_XZxlaR6#9|3ot>K-7
zgO|`Ca8j7N%^`AjF-adRr%$6XkmnHA`@UKS#;zeqg37a@GNFOclf7qSMJP+%QHC=s
z|1vZxZpqT!tftCV>4gra?IB@Qoob-{9G8NE9G)+YhLF-MA+=Jl!8^7x`lOf2_4&_9
zY`#3<eHl<Ww=Vjw@og5P0V#+C9#1N3q6LDOI)-~Xx=Q{u8nZ#LAie*yM(;t%95p7|
zM$<?rQ2*Xl-ry0B8(?79xlu(zsjT^%w;oI|q5dYzZJsp20h;K)KJ1%k$B7CHuRHiI
zL`IkW<|6mO12azDs`lo9&;kRTY@l5gDg{OkQ*+d>g8lOm=SqbqscF=)BS?uiCc=P4
zOGHHH+%`Xw5fVgv<6~(2Xeo(DOr!LHyL~*WqGx(L|F|<-Jz6$md;I)^v}Ye1?@hs<
zzsyf6tUA_<LjNSB@`-`0wJVWN1S+y@l634jA!N2Rl?k?RF<jVgvBDGCSY|3-map#?
zTq{D>j;sF^Y}U*a(+SM^9+=vi-IHN%Xc)#PCk!hv1Wp;h#-g7W#8e;l+}{qUMl{T!
z&fZP6$W6^E_sFuscUk@-Z}Upd-u+`x4Xyvrj`S6CDI<+Z-u`%*bP|m0&$ky!ZNaWa
zek>Us=^)R5Hu!t&*i!(rJcDdtPRq5wEu)wDjht7RD8e(<yoGO+M+Zz9Ja}+oBB8Tj
zt*TGx3_tJgxd96SztZLeRCBCJr(@J8r8eU3*fc?ADeC&$Mpa4B^84#;)HR*Idj^y}
zwYpS**mbRfc*h^TvQnWD5g6D~J6tN9GLLfcZ~#CAp~rWBE)a<qhq+2K(n`ZX%}3T4
zLe^{TZk@ys{NK|Y|MvM$Q!rfY6qfKjNi3g+j%zj*O!&&oAa^s7fq@hlNTVDYWbHR-
z;=a}kv*xbm=CF!}2dw}Ax|7F2O2JmSeETfe@DucOi|71dA6qnW`#3IJKJ2i^QLp4`
zC-5rsLEOdf%;6KIzt|;n4x%y9tz|a2eJ)>$m->xFVN0Dw*!AVr$A8In;lNK1T$)PG
z$%*S7*)#@5BQvM5CZj*DP-Q12r~O^eMf2t-<-Z>Ft?*qc7Pki>TUa{(nJoleHchj(
zjr3TSj08L~HXTBEIf}r7H}<p7xx+pNws&h70n+&zUm>lo(`3(xHTi}7J86|B*-e;&
zU@ak76Gg#n*9oD0;cSW+e$u`m!v9D)X!-4Hd31h>!ZjdzVJ}ig#pLIEWb>E(wiXvH
z*sh%_fCC7C>^WE0(Xjo7sdlv-Z}K|q)zA)Jw`?qr@O#iML;4PZBJmqVRTj|yUKAD4
zM)uj#!0Hk2*eLVXII~%tZo~~4e-eMv7ybN;x_wQgh6c0Evbju66LX@sT5CqB!0@2c
ze}>{}_?t|OugbhpO>k@TSg-*lc@%{2HXQ7nA;A1dyflqpi1B3uLy>aK`V%i&F#C;7
z5mNCUkeC{?IcSXe%Sza8Ph5P_fd>BelXxxIXZo-t7-Lzyc_MvjYR~!lV|Lr*z0AEB
zoA|dlr*aJj(zOddFIxo89)!TjI5Los%+$Fs^~G~L)5@T>CmGUwq5F~DY)J<9khpva
z56#c{`{%Dfyz}T3Me8VDS%8N4_vph?^&htKEZ~1_e29eSra`l>(XD<E=_3ZV*9Kl6
zt+9LY@h~C0ZSSXp5r@t1V=LNx`pQ1b_s3c_)^B*<>yc6WWtg%FeLDOx{Uk2Yo5fF`
zoo*6)a<>q*jR%^SZ>!__Y5k>$=AqZ~zP=4_?sLigE&Z*a%D-z&^7xu@iZ7cC2jbyf
zGRLv;kwbTGM4qJAP&Qe^$PM4Q-<@qhZb#)=c?)S(H1^&J4m5Z{D4@IjtWC9+)6KoF
ztq8qI6HW9yl9idy*B{BB2@AOVJ1JIDU1O5++WG;C^h;v!6SyNAsFiixq%v}Ez0AMw
zvj75@2E9QJ8cRONqU(=7d1#@AwkN!KdI~lgEU2H*$V57aul~K0?u`S9-Epj;uzrz>
zeSV8by73ux9d2CLVVq*6*EjvQzH~t8^X~62f3$O6#s3l){_|oVHxnqzEfcg39QBAJ
z%(M_{34ka&Rbd+&iHorZ-1IN2<`n(@y-E|~q@9-b!25C71Kc{J!t5vNhO<t7ZCVt@
z4yz8=XXmIf-AU^z*6<hRFLrD3elBPVKV;>-X?zAE-=19@m%ZigjuC!%PjeFA6}a<-
zFEvcKcSr`ncgelqlpFN1E6}nJ|EL&5tV3D0UjgZrJI`Wx=(ry%h>oP9_A~Eu$wRQg
zm$?t!YUZ;T24%e}#cJO<ExcW-Qri2m*#*?ZzET&;2OZA`a^GxQ>&!IZ&m3akDZKUQ
z4qX88FwAIc`&WQJ#P9f*aaUt&6E=KPlM$DwPjFkND+=4357bMvTUp%asOva;Z-qMl
z^_ti*_$RN=WqhHvl~C4ro{c(62BK>{rvU^v{+Ual-orU*NQsMCw3l_R3fhitQ9AjR
zOHe#_K#Bb+l9rdS3feiiUXj06)l%(oeGjl3_20ME>co$c3+9(Pl}=J#Pbi&_F8n(q
z0!5ebwKrD(eo+uh+Jg@q3cM1VQ1prMydR;iAg0cRLYkA9DU)L2(Rta+>#ftkl)J&!
zORE?TX;J=hzG%VO3E62ksn@E>Y)KFLS7MwKqNtezDW6?Tlx(>z{4NqqC+Cjomh5Tu
zz+p#%$gFs01@9>YsRL$-XYUIHcvXAymy>)enH`h*==;kOU^%ewE<zw56vh{>8k3+)
z51Z8d_>|@AqA)M+sl2g?F~_r;julz)!ZGnN%paTipCXpc!4>0BU~4o*zEt#y@LC`<
z$vfHH3!nep+&N#WRWtU7=6kg=!3y4PdWf3O)KS2B?n4XHc7NZ^V}JRbCWBfwWQu_g
zz#{R1RK~PAD`2ciLEv~43@20Qz{$6iZaw>n?^WvJ4Fq$q(YBLopk)4wsP}_pRhJy3
zW`vfDxh`K;$usWDLmICn1jnXpZ8dwEKR#GVx+`yOTX>kmWQs^v!K{hIJq0HyGe-wb
z$q%RB^vpJRO?y|t`;UrX?4t@;6TSI;O~FUELhKBlXG6Fd(uV;f!%FyFoBsx$I`KEX
z5xbMj^W2$~f=(1z;Q~XV%aQRfK>9r5aQS^`MV8PQTto17W)!_)X~lCm5+}n;%0pi$
zWHyi<xl4GqOV-2u<XJBL1%7IEYfD(%(y($2t#!x!EflVu(K~g|KsiROUA2~AZszV_
zO0CGQS6%HQS&)ZXHfb-Eyt7pA=HG4(hQ>4Vmm#-pwrv{ZsiWQ?-Wl?n=qk!=+U%Ml
zMNN47K6iaE`H^@l4*P~X0-*BhYh*jfjWD5qOlekBQMUM7+(%Y7?kj3k#C`|F%uk>O
zLe5QXOTWRqMd6o8)GKcuUfKjN41Kxyvv4&+SA4m3{}|{=m%g`1*lTM<onX1c&gT}|
zE_zl@2-nstyF>kYC3vIKn6ok9dQT{$U}0T*XmR#El+{)M7i#(d#2;fIhen(HM6Iv?
z88>L3{FkY8y3{6m?B7oFabm|D8z3^6F`MU5_8XmPperqfrBl@xzyt&Y+)2g_b*d7>
zcuXx#_fd*KY7&9_dFbSX5qTg#@-x!_Sh%mgWvVLjcP{qDkY{>V9%(0fc3c<~SGN$t
zn0S|(sqjf3LOc!pbLkY%H}ue`1N)nz*){VXjb<Zsq89I6qyai0b6F38gy}Tz|F`eW
z61&Iw)79HYGF&Ht{l=c|;b+rbHgR((Z$I_7RT`#2jBdZXe=l5ry_Q1CEnE+Gc{o|7
zwh)!P-Ck~d|2wrjeq_cw*m;EUba$^E9mQ|a#=(c6BOn+&3BuFes17ZaM2cEo04=pJ
zhrkvkX?fp&-9!1t5h~0R!jWu|fBcB$lWt_h>UuH%M;7dydi=OP?0ioG1s6);gdj^S
zTGXa0hTe&!hE&$yx2>7<z-g$OJkJKAx04&fqQnnF^^;CC$020p6||5SuC__hJ+HK4
z<FS^bAq)BWmp_^G_{F>ApGw$%Yg{{21nTuTLtC+?)d3h4%L^J#>w<^h*r*9NAhuCu
z2<swPg<*j+#=k7RslcRI5M|7Y-IfRBhCim*93SV)J!SUa_*+`K4)^LQY^$w-O$Be7
zn=sqU!i@BDjVz3w-N-!3f#sAtXE-{~rfJG^56_m-n+q^!496r$(yNhzBx`FEKjqVC
z!vZRK6A7W*Q3o##w0GP*F}0ZO!w}aN&#vpMSd4riqsVPXgtw*w3PHB_VF5$}VI7cx
z>~4t8t55q)P+$WmJS?anvU-ioMZumQs`saEBwF>^!pmQQ$@WB%hp(cv1U)Ug6n}A4
zm`Kp`h>`uT$R{t+g2m3c1OOG~Do@sUJ=U5xrFlDL!N!gz$KRh+0j^s;M3Drp@)!4`
z@FJ3r)wPxP^WLM{Q|Q^&y3=0O1mdQpqWn1{v-=1=(G8*Pn<zjDoxO&6-g5AMG|We?
zy;#oO2C6Jt?qIH&?e!L{WOH-u(2m91^sFk73p6m-bDSg5^uu}FX@Nk{l5g-mrv+o+
zUVZQMn0XyHwzjT9Q;j31iysn|crzVPhNi4-64X4tb||^QS%nc(t>KZG;CZLvnm+#&
zaeaV=Et5GaW%%T896t;8k&&#!Iy*JESYJ$4$v)@Vqo4Di{r2D{y12}jfDhQvNVRKW
zIzc$mhKgnMKmoU2RDwftJgaansGPmi(9&WyU6BP=hAEpEE)9MbNSvRapKRjj-@)FG
z=}wa%N0A<#`mV93j&uN7X4Xb+$AX68H(E$hrOUypOdhsg;NpJBKWVc)brgPn{2Dw^
zzUTG#cmEnIog@dQFLox^@&bknc{|6<+&-dq7+`O26(1qVm82F;?<C2|q;Hh0%Dcw0
zMh+ZCd_d*$)Krc?E^mBRPHj}S^K^9%p%H_^$q<+PjG880(2qlpdlQ(J#>8T*iyjU!
zVgbZ^vT?*d-K6!}x6z2>LaR&DzyJb;)t&aoY!#TDwa(qv53CEr`FgT2AK}k&t8`pl
zfq@6dn-9B!b}-YPmRE~kR)$N?{;;yr?r?|rRta?N&RH$|{JB2gS=`6?H*!kcb3GOn
z+sgNv?j$ty{8$8aU=@Bj^_rFVt5b??OPf^T@^ryqKHTaYy?_nau#qropEENT5<T*g
zVJ=!Rl}KzBeX}PP^Z_beDQR@NhPujqb8Z~c9_rb3`TR-Mh$JPhcQNS$-%WhCd|9?r
z%N;^~kC&%o=rG%aM~j1_V=5QcSa1bZRB~L2kKHg15i+EL#J`(y)BRl#T<mBf`l=qm
zOR;oO?nh0nWo9OCZ_+Y1KUj`n8FJBP2Jw(1mLZ?p`R0R&l_YtQ>XGkm>)rmicuyTn
z!lq+l7^&gK@Q#+C4&kj&EvHFLDqEytTZ7B8vNf1`tA{CYN#qNNd=l~-cUng2;7T$5
z_&bgw>B_iB=eJ}}HvS4_JSkm&%sQSoE1Ow~%&{MU@&*F3zvNE_0(e=b3=+Q5+0%H9
zUiv&n8}U2Lt1d0|-JhQdk@+ansx6J#{hFJXG6Hnsf43(mvGbNU9(*&1Q@|}^b4gq6
zCWH$&K+iPU!}An>R;eCX8G4#<lR9d=>IzXtq-Zcl(ZRfMGE3k3;=w}c&xEBIbSdhu
zuOAxzGCZ0w#s`?WLz)%^qn_+nN`+qbjE`Ua_VWAF8@)E##mXm!gM|?1f860jqw~@K
z+Z@H`-n0e+^MvzM;Qn>JP5rPV?>@R??COHGD@~dS_vOcQ`kE|*4}Z3%OfD=!vYi|U
zkB2dmz5y*h@hbM^Qjl4biCq_>$f|68^O3l#&(BoL{FxD={0{v{d$$qx737F8pj@$t
z@9xjktZ;w*y};rxUX&*@(}t7pUCrR<hlz=s!p`Jq)H{9Hh*;11Ehf99+i(z>;LWFv
z4m8a!g3Va(cIllZ^|AIlcBCN91YvgxyRZVTAvWFv0yU&?Wc726=?K5Ke(D>C53u|n
zM*3bogrKF7!E*B#9@6a?AmU-s7UX+9<ddO^^AZ=V*M6prIvodklNAqZ>HwOE$$c;|
z=HEnl(#>v5-OSxo81<{0KENBUED!vSj_%=F6ZdSS<CK%iH`F@MP2qan)){em*7OD_
zi|=1>zJZZh8L1Yp_59%cgO(3;08AYcw<}`3aXa}}gq`sI)T5KL`TS-WuuEMscYcYv
z{AsIaezrUG_Q~sU$wxS^NW!^<BmS4~=wnp;PxSBuS3W^`(dIfU?Edl0<Yw-HTA6&y
zZgkh_+*;=$$K3Fn>%*_yEe^R6m(wNfdp&c%Qpm`p9^dg5O-IILo^Wp*4WrHEo%zD|
ze>3TKmzA7fBzK+m7#D|ogu7$b-6h#+`=6+#QMpMuupD!1j`x~Ma2;U<i%)wu0vAxh
z+gl$zZ<U5Drtf8rhW)K5>H0ZhAFm?4_R7Y+=0_p~1TRZxz4^o6J^2$(4A;oqKz%-o
z&KGvAUbHaMZE%aB=d?b{EIH<W)~>J5Y`*8vo(bImP*iJj2j|Bax7vl2&J>0u!Gjt~
z&@)F85RVNF38<!NK0p6?nd}zbECouQhz-2zeL{x@q`BL39Q~<EA3uHi)X*^J5g3ZK
zEM5aC$52yKM}7D(T}!w(eJj!$u6d7Hk4($o%?1Bt<P{8n(mQGl2&hH9;*5R-(NK2A
z(&3}KHT0=4tt`K&9mrS_zf?OXxECzVO00@(IUCe)_XHsyzD(P*cigTYb3--zW$>tY
z8Ns;7pQ-ID#<bPyCw9zyPZ*brj$c9oR=)UjaJFGn15NDWul{#rf+NY4AeyH?zP?3g
z4XliJMpsrRjwi?UjWkVeO4s=gsz<-<Qu;~gm{^#1e`u$%j5abAvxBLT7ukT4AWxGb
zCDJ%=z@nUEyg|GeCQ|yM85CRDNk-XayaSUhO;H0q+&$??Tn#}BN205LTPGFuSPU^j
zV~l~y0^nbv0035~-|zx^e_1ttwvR`199mlECjlq4cL~=&CIDZ<aceadfgMBz@xq&p
ztO9drmuKERMX5=!D1n$|?OuA@7`7^go1~yD1Km95Ivp6VZY0XXy#B`zY$y@^;Oy)5
zrqWf(w1@df#?Hl`xA#jF0Hh<s@^7o<fAmpPg0Q%x8{(wpf&&v2i7^8SJaLPI&y)p^
z{9^8EL)f$Wf&$($<hwP?700us!kqdbnVOV;*eONz?VKAXcJ6Q=IJApDrxs-%bz~*o
zyPDcr#f20jgRM|3WA}m|cyWa#^*G9nWteS>ktIedi`fwhH{vsVhwfRnhY^gJiYvqb
zBEaimiZ#Dt?fd@9%r*Z<(|P|>{l9<w5F+9j<%nz{!lCTY$zIhFvSoyhz4yyr9oc)m
z$;i&i-W>Y~A!LtZ%iernpC7*d05`|&b)K*5d0mg|e$P|N#t()iD5sx-n^p+U>l+9G
zF7}Rm-uv3Lua}MUHz&0g5*bx?t#4`AFjHyZ#NdB;1m;IuHJ_cZ6O+5!yW8NVS2)u(
z@UOu9h3`d5fpD$coap6*5ZG2)JFeNe!LF@|L#lP}m#$hqhjbN{`EQKZ)bvQ599^GE
zyBD}^1!vv3TYU`a{?X!zvbD<QRmS`=d#;>d>uj+^hVFR)!T<4GO+mA{vC(UxEuuDm
z%vSvg8-&30kw=^#6v`^?y#8x#4ZH^d+k=CuHnSLvd@Q?1Ca03g=DWn={qFz{b1Xs2
zZ-#yR()ny{dUGz10{4*icJc1Kgu}%Y|24K2**&4nPbw5(2@{0Kvg9EHgd&LQ(v2it
zk6JDbeGZ@5aDIDyIt*rZ_IomkYceQVz8l-5!#=ZvG~?*fPkQ-Qno-5GZ{vZ+$WMnN
ztp~@qSF}KOL&~?DF|o?YnasY##D}M9%;w!@NE&6*C+6(K(L>;0+F6iq(c}7U!-O3h
z#VFx=o-xTS+aAPORQ!3fq-UY|6xsj2FXzXc7q?DoJ_`nPGY#<H9MS3)-nmOLS9g)U
zfb=!f?Y7FMMcDi6cM1EQz6WUJ++24T{9oHOU!IJ7J3aAijFLR*N^r8%=vuUgnS#^`
z7EfFW_<LRRB__9yA4N7kAXi^`-3EOTu+Rq2M-D7JxYKS`BzPAeOSc@$dm4tFK)?wW
z(tFlFKFUC&!%U8l0wA<rx3Ian9dnUL6MM*lI6|H;<RfenYu?8KL6$5R(~0IUZo*<n
zn#NP!q}}IpN4HOy!ynO^)<MTyy^!dP`$&#nS$0`j(4tF2T*@s*B`?Cbd*Zqai(T%Y
zpZc>)apLOcH6t;TUuHLNXdx>(En7=y!4Pf&S&-okVTQ5d-3f!sMxooQSJN4dOweI|
zy<x=uqCHT23lKy9gN3V`L2y|fqsxqkrUq}NRZ|R6JsQhH2KAmOz2?Y_a^D>kUjwRG
zq=Z|E_-hPd(ZI+hE%q5de}>QD-vX)YL0p#C?sjpx?Oa(C>4-wgm%MHdyHc%)0EKpv
z5A5oQXvg;PnyPhwr?2QHhZbRoU$2eko!^jGCaUj6U4b`V^Zg^zw^*>@IxJnnDLhOI
z{!_NYqh5Mkz*kzzwyI}|%7_W!EY66w5{LF~b|pU+2#5h-fU>ErKHBq^C1-gg0`hl#
zLVj7-D#U$W{e@S?{)O+)XuMAf7cq9ihvUH)1c&m#;Doo;YOni`X7@J?d$zpJ5_t~g
zEvJ}iU=^iTTnikE0FA;g-m#N+3D`Wh*0kat_}m2xfr`=Ue_uEGm(43r^Y`Fj`z~&-
zOujKvtN1yeY>CQbe`C#;mEQXp)x5)@pJs^9JKyYTbTN;>h=e53f5dT%--U#hef|pz
zM{X^@V|QBW+6T|kOuq^6sx{x@u{UDpxPbl!xDm>p4+!y>FzDX%=gXAfclN%BV`AFV
zoWH!8_gRa}I-mM7Tzgup_OfC^g%7dA5jGlho^p*utt=f+`Z^e!??xtNNlS`IPkz3G
zcx?S$MS>1WiNjB3P?zueP}J?N|3{W4#LjMKQ23E}^x;FN3~qzThbARNY^BucMc-&F
z09`dq(6P0PsnGAZWd-twfjfUJ>G4H28*j93C*;0j4xj->sVIt26N_c?Uhl1QCZAQ*
zt>^Lq;fss61<$AnWz0auE;iRVj4Y_1IKX)6s3*(&uvA3TY4>p0*Hu5Qbk3Hi4}9%U
z>o-%DR6#$mR%9R4pN|S)bO5jW1QG|pufBCQyOo8Urv@coz&<?FtgfCO{rgge6u(c6
zlueuwD}5fO#rH_U1Y~>D)H0q_4Xv#me<?1mW%5~xIX<o#v4on&2M7?Yz+$Yj(~kp`
zII5NV7wgQFuKV6ddadOp^^t34md-Ou*&lQTpYltf=pN`p->&@E7f2eS0^g(nV5U2t
zE;HFqueO6}SK>>{yBfir&Xt%;yK-B9#b^e2gn#@fN&+QlC;o5~y8F_3-XPXCK#nME
zz2w;MX<Ym7$ao$+v|I=5^$L<ehS^x7FCjplSP`L&%f}vssS2=+kT8*1;G0ILn-}iV
z+9Ji<5Zrx!H4;KnsSO~Y7Yop#NcG6R?6+wOkC5Py=Jyjs%wr``14<i1<!k<`viR8C
zPqJoB6fB;sX6QT11cSdPyoU35<(J=v`T|krLX);7r$5~?+*ZCl8=BsH#wswemii8@
zi^8Jg$DS+86Vu>9gBGxIVA@<Us1+HF*B06avuEK5DEB5HBOtKu2RiX&+67eJ@6<&u
zkT6gx^95C7It?np;%#S~nv==$rxWx~ij}SH?>6>Ww=!!t3BSeB&?H9RqkX39v%JtZ
zFVF7Yu0_px;D)lUw=LwvGL3YN(=VZNVPh$GnAyyF>qeCx*!YRg-hPE6y9(QXUHU1e
zYZQ;`>Pvzv_<Mg*!0rC6`#yF}Uh^y=p)+pOcTr{7XL~<19D&C3S4x0d^or3^i6yD3
z^%%AG!<LjcHlAHux+M0srMt=d&2P**V4NEa@^t<El~Yy2);R-bpY1-@bZx4l7Q^cp
z{40fi=ZZ`yP^)1^YHBTGw>u2&E;CN0{$y?s@?uw!_rH+vMGeAUV;mO7wtr2Js|qk{
zYpdT2%RWiOTvt$e=O<(VTQzeT_w&RQNt6z;YwAhE)&wx%s47>-Fj2E$_SFL9of3Dh
zGaR4I3@YieCVu;m0%)y%o@L(m9{HNx|LT8|=uYXY&eQo;(M;8bf!W}^JRhM^&E1Y*
zC+#xBo4wmm0N803v7M`{sjG7~b$OvQN=+f%pSPyZAnF{gbaQ)Ea9hi9o}J($r^#B-
zpGfubQ3GHwNfZP2W0K+d<F;}uN&H5cSvE@9oO101v8(tgQ2cc^2eEg`t0W<9q5>R~
zB3ks3@^d!yxV-p@3B9f2*sF{2FWHYgmeUVoN@SC_o}(Rtznt=NlByieEWb{8{QX1N
z2QLVjE^#|0TeJ)fGQTuYmA-k=%nP-(*x7R+nE~kO&Y_<CO)J%#+$HOG{r37(x{}M;
zy3Zl03cUZFj<!oh)AIt@?!!Ee=fatu+xxV@iRtWi=1o_DUCA75vZf+ofgvo%4a!D~
z{}xIB`6orD@S3Kwd&Tlc#?zbj`dW)M+Fc3Pd9Vqeh2&@po}N<WgbW6Kfbqz+R91Rl
z`2%rwO*bm?6|=vZcWg3Nz!lv!Fe@zOwksJD-zFf%LHqWfv&Hh)_@Y_M$DscjOdC(7
zPV1t!KIhNSk%xlw(wlu#&aIpl&(+w9{Mi)xs3*EFGJV&kmMPMwYaPFrZ8Zv^BiA)M
zwsM>kV#wg4?RdRxtbz~OJW!_We1vhjK+${R$FIpo<idJWqGVYhAOGvkMUzF7U_X!<
z3My+b`m4$qKRqMX7fcaldR+}ABPG{Y`}v-DM~kQL5vxrYyBs&dKL2S`={F}R8@>o1
z(USfrrMXz~F9eu6a6D)EGK83tSx~s}bB3_<G*>^kw6J}AD&K8s_iKJ?cKN$d2H<Yd
zRI42ETmxK=r%8Wvh<MG;v~Kj=^9do#>nG6_#5DfiWSCUwOJ-sgkdjK=OHB9r3CxsO
zXb0oXiDKuCS#Pf86Qv#FZT(W6wwSB?!LYD&@AxD!iQF>da{_gCytkK^ljM}k+ylA#
zroDxP;@aD!+hgfy%d2+a5VWp|0Jl5sUh0H%#FBV1vQNU)4)LQ|nR61U!kgSM6PgB%
z3nYQ+u~^m1-<_vfqAmf<lD8d)x2H_E`&qZ$r)7pluisyS_rTcg`Rhx-Z7G{tT+Lqy
zF=y#ECa+N_yAe+szRhNu7a!|WEvSyI2MQ0u6k`G@$`~Fj8-!@V4ZlVLim8WR?&jGN
zA!NwrW*k+Izcbp5>(v%#3}wt{Kl#=S$z-S1DXPkkr(-*Azf3Ax&~qZqbc*5RHE)0Y
zgV`LDjKA=7=&8o}LLg@z4cn3VyX~Eb^_g(pX3q`m<};M4xfROI`xsLc8NZ_Q?sxit
zR&g9VcX#r-nJ9xgn?!#OKrXe(&Jqz55U<ajk{jW)7FhGsbHSf_q7L*)@jrX!^rXM?
ztV^gMqe>~3)e+*m-#JjIzV-<>=ZPzwP6L&cau}UVT!DJV#hCQ92KQ=zd*G`3=!SE1
zlv3(rQSUkdSGVWl-rMW_*$I7S@$5xoDT~N}6}cDkM;a={$j|ox`SPOA{a-~eUpbex
zsyKq;!@r{M99S{S7mp)kOXHraxt)5ZrESfn!7@&dCr;!GrkIJv`wpA3MeY7;_1S;?
zl*m<(6-VGWBqJCA^ZqfRc}HP_c0vBl%R{B|8N*ddzn-?&VKiY5kbhXKQu;{DL5ok+
z%l(z}L+}$E)(i~+HEaq|MI{yX>+Yil!d&%(;zBCd?_ftgWO!VISblTlo~><$n3JnT
z`C0#-?=pYuX<&1y#H`kGNB1FubEsY?gFS3Loh@~DNtIT%Naok3^MQP^Q0>`?#NK;z
zN}B{FUQA%IHs8z3?W#!m&ic08R4cc-(Oz>KGDrXaGCeK;4Q+m-l7V+R_d~zEN_;`E
zJS5p{zxa2t^De!|n8eykh~Me4Apk@C1r~n&51g^F)a|39qVSdB6R}=enwP{3S@zhv
ze!Vl8J2@FJy1u`)?Eb#nZ`fyll3Ng~iT*ITX8TOlKfY3)M^xaR?#9AaZUXmEQ+bdX
zOdzC<va1?M(CpWJ4X3J9<O(8|f<P5$n7UUwo_l1FPJCYbY|I^dYN@K2{_M$e1D64O
zG5mG800D+wLzP3!5nK0fq9OcCy24(xQ$7ca+!hLzYl3vz>-)fjM@QR60AIk*D43?(
z>plArU!#Dt|NcnR;C*{z4#UJJWaQ^O+0+BrR$&bw!yN0DsYDpyqz;#j&}R-#&MUls
z^G)-^JsBAm{9o1jJ+LX7Hduh{8mq%sDuEa@?CuKj2t~iF1E$`|rE!d#lW^G$H|4#Q
zT`)1(Q;hPOlPF#7xKe(xv7_}mVnj?k;X#3Xfsog2A4nau-LsxOPur7zL(ube_SCEO
z^zsz=>tqJ9RXWbAI@?{t9>%Z{8|y2Dhm*HVg#QXpM5?Qay4Zl?H6!<ZTpIX8uh+TQ
z=7lV(Z+d<Q$|UKrl=4P0Pw=sSM9~U9p*Df)(`R{02z=Tg>(7XcN5}b-G|wb_AmCUC
zesJtgv7L;F7L2j#k((RTgTyqk9N-sz2Z}}?>3x;ZSqh*Yk(m_<yf0NN4K^t=T)kYb
z*ETnGoo@f;^3rrW>PxoRWhpFq^b@8wtQgv(^h7n(h-jLT8}<U|b(*^*qSLbtzQ@Qp
z>(d@{pdbuQo1dDIp`nDS;)O!83-7;qzRar>Ya*z0@afw~kWmJ=`46m@yr7I3fF@~w
zjYcTu9fdL(ett1(2)v)_^S^T!)>%{(^cERLYjo8e_r|xWw#v!rSZl)OuhU+0x)W-9
zW2|6a>~uogcdJ<HL!XOIlK9BLrA3l9`F7+NAJQ+neG+aHB9pe}shovcR>(;{lrW1p
z+4C9%6}Z&JLk7{^a&43-`i%1j0+i^77VV+07lMd}!2O>T)msUyhsCwu$4n4%(sG(<
zTB~gbxmQSq+p)C6jozjG{`7m9s4QP}m}nfF=&j@s{R^kN>b?h~X%<%YhpXcv>KQj3
zyC=KG*H5FE>8J>C_*4$TEJ3rg%{(I7c5`!{CsV^i{LgN;ang25me13GZ`nR)6`WX5
z@3WTSGIvV4b2ibq$JA)S(WfV*5}M7}dT^1(-+VJi+j5}SdUmcNL7$d&I~ezL49o%S
zLm*#gomLJ)ned}e-zmyLeaQ3`nH6KL9X#e&RhaG(&HXNnMZ@==L7FbHdT(A{4-W8W
zaYuy9P~tr??~8j+vcCdGBCeNk(!)+-WBu}R4t9;&Z;G@UgV8FbWk!w8#<$tehr!3s
z?L=Pl<=1zT`)6?z(<tpSfsnb0)GuCN8Rn*T=0Q&Phyv5rX=vJVq|w!ZMe9j0PU^CL
z__mYcW<w85Ag5>#>l^4v_#T)d`#aQj)w74`bu=Gd8~clzG6RH9+Y$fZ_IdB^iTq8D
z+F^VTST|(pHG*M5u`tS*;B`}Cv_npqsLgb~lLZuQp?l=!1~)E}n7jKQRCTAnqwdit
zJ4BYIH+cVR(*v@;3s_G4y`YtVxKQ!BX>pe7Q6&R<Y%)y*t1JS;uDFW`lhG}}(&3t*
zsbA3W-Sj9LfIHS$a*Qgjf3hrsWjqDhaW123-;HNN{7%&%gm^?EM23brXyah(s7gRs
zEeU}YldYVCP+AO%(LUxt^sK*lZR{r&+FL^FK!sx6^Ulme*!w2^__UL<OY?Zp|HSiU
z-XGf+6T0g1)ZElKKOtdIoW4<Q=J=n0S5*mN*Z<TVx47GVV6(chfKhBD!)CQ`{pZZ(
z>E0V>&y_P0yDh!iGF$7v(^FGT4u@-VhUJr6Z=Q5pir$0&Hn&2{8E0TiCd46+TlfBe
zy#v+PD14rpA9K`%1Zob-TV{x-+0{^8#mKog!%D523(L;ZuA8%D?M`doGP)Z47AWN=
zlJ{AG1f4h%vV5am8;4i<uvHH#Wj5zEKh|?cJCxv~?Xr2py3^lF{V~nW;JmJM$c1ld
z`eIzLtp#h6b2yd4fj8o{XTWS1gO5km$KJ=WnKQ4=s{0=l5eB}dnFPomJekp~5%KR=
zM-qnWs?ipXTMMw1)WH8MecJJ{YD%?=x9dCLg5Q>lWC?;cK~=5H1KsBEz-@Y@?d;Ut
z+@{a5$jugTwLgC!)6ApI=X<q0EFfdr^aA{<oJ_WY^D=DHcA(GMVyJ9Rh7<k@=;$db
z+QbeL)1;GP&T@Gv7@BcdEOu~vNwq-Y@0Q))mQ}TJ8?%}WYuh(#w4WkA_PxZL0*oGK
zO6dx4Vog9duf$f-^ssAct(M*J&~S3)O3Zi-o=o&YbMcnAd=W}P%n3`HIXks6>$U#h
z{&NPkwv<mfA>SN2>XwiWRH|oC5;1C%_gbBmQm#ueuXomXEDX7b^af5D8@v~H>gyA}
zac0%I?CWEoffnfssUffGOQz~QfWx)xg3M8+*O)(QXY~A6|3D}-1utV=h3Dm;zR+S*
z&X(KeRy)Jl)V;9gxeMxoEL);SNym+}<-QxJR;P`FV|pgvKYhpBqoe-`O#Fruw1Z1}
zhXv9%`yk%!nGmzs=>KN|l&G}nsl8=YE1RaQ;VgUBw%F&+8be1Krzic>Y{PNvAJf(P
zH}jGV8}@GQy1BV}x(P=*waF&pDAFa5r?)I(gRIE&A|so=d5HhniDdrSkkRmzcyZ#Q
z=JIQXXh~GF@5u{U@Mr^Uk0(l#Vmt|GfOYDDfy)!3X7B_*r)@o3wevls2To7t_jYz8
znnpgawo9-!NCZo%#1=gjmG2|zl^-_isE!Q_lIiVz)Tu~i!At;Ar3jozv68^R-sh3d
z;h8emBS%x(f+6(Pu~^g0zJ2%Y5;y-6Qm6?$#U58$3-`=_PZI%GmwIat(!T*VJa?P{
zIHQQci=mpnQ(Tk^-_7{K;)tc@L*W9y@F9t1+k%kwig!p<7*f7(PJ0=V-EDca@|Qv4
z7jfqNbnfvELUAr5Eh*)bCG_gOfec!aJ&zy)F3%n8L@r7lAS4?>S->_06}3Qw!5*8U
z9iU`%3PfFWW+3GW^;cDe7I0a;1Ca}R=jdBoop$-PoH>BT`+8!icE0KC4>g64z3<(C
zIQ<HN?&dGMgQ^7OLN8l?mqyTIG4<!Jjm@Hu63R1t#tH3z`dQMLMt<Vi8GI6+js~Mr
z<uD30&A03*UQ+}r1LqNI7TI*amyWss1XS)vNmo`^_eM88%ecFO1_;+olNp_%go5fp
zehe^H!4-WHmZ0REF%x!E)o&xr_!`=pI?7r?MuG>J<mz$4ggOX!&WpIB)^|}SqmpcP
z=7}b;f_b5F29!207)>C!J!_l23xxrb_e2)`m06!Yb7W2NG!^fd7Gy&8$J_;`XXw73
zi1tXT<YBoGFH?SJ0}($@CWwZj;dYVxFC0Bgo!-jJg_lJX&Gv`?sD42o>G1%jj3H3s
zCZJ$3LlSL;a}h`rUQq^2Op9^G$LrF>-HQ>!X-`g6%KI+;?lWpSIGihUKVMDClDwwL
zi4AByTx*=Q<qUFP{3|AV(nbA7p6O*!c)6j+&I??p^!2jzH83N6A>v{~)VYXjJxbu2
zEU(qwIj=9Ph|3rn+P5sv5D^x!b!)K6*aYvr_bSG$ADDvh;>KBLtoR<YX#0x&^QWb5
zUq{)6#&Z*_v{n22%szQAk&N$A^okeOM6rwu4h!bRHXH=zh)9Sq2?)r;2J2d@Dr#<w
zuD9%3|8+850`Dl^la+l|dKYBOT2{KpyDfdR-lsu3e|3J-{WSw+RPPb^v{UWn-x=}a
z6^+R<Lx&OemG}jz4|YQ;{8=D@U=cg0nrK;i)_1x04Ad@QCpIydLWL~%(`yw$8{aL8
zmg|f?U)qaK&>*2=Ryi_r0|xH~#<Ff|$gGjpFMj~QNnb%`gM&pVDM!#?&Cdx-mf@y@
z3(m~*-*<8Ir%z<>ncILilj)gR?yXcjdt#VX{#XW0-0bYD#smwUe1f}O15AeAXE6#@
zYlB{UU@*L|hqH4ot@TPLHFv@1u|ZDtm;aoe{U?!%zT4>K!8WNpGum>Dl_+5JIyftg
z3)MaWnooE2Vx0)^nfgMZeH!sk9k4DI;#G%Lpa}UpQ~K*{=jmuxx+qe<SDca<u3$85
z5BX1#Y>ti_X7RiCIVd`1sgv!8rBewCJZwr6w8k5GCxi7^34{6POPTCRvfS$nhR)?s
z7(T^{9>UF1^X`cfYRn)Igg;@C^s#l(*gJrKx1z>(%^zCIeF<fyE-a92W?7l#!22<N
z5;#0RJoR=HEcLxU&4)f*5lZaDl-6jC1;;sC4+!ot=HjN)iGtyC!G}<2)NWR5<wRpm
zMW;bR-hDLr9R(~231vN~RKVt^KLo@0BPtxC%oZBn5Xvq^JXVMH05&vx9&t>3EHEL<
z7xPL#-DaunQ``c|Hm7t8Y42M5w7EPmx}>bA)(F^YU+(#0m_?}jK}wS0q~@MSPMx)D
z!{b6~)=BJ)(9o_S08Q7lnl`pF6|VgnkjBj!T&t;*)Ok#KCxWJ_@IjKKodZktii`|R
ze6(bTf8fC7X?gC@$>3j;su;AwH6S@0JWBbaR>JzZ6MIiI+C`NEDUv%8;MB))hpa1_
zbRkwqY4EO{!9Q~=syr%gnaUeiir19-!OHlP!btlG+Q0?&@mI>voD)}&1#}SBEvu;R
z+#fC+qB6OnKXFDEA9M6D-Id9r>i1$Ii%k`nB<(Hpe!TD$DgzVOe%SaGdSHiv{S=HD
zkO`;>^27i9XmJp-``Q0XSy2$*^s@#ovWY}r!~$2N9``&7&Fb0bOa>2Pu2ZCft3j8!
zbR%7f+gRQq#Fnjzl{YKP3}Dii_56ls3rk&OApWWO!7USNZybek&De-c!O?2{5lc_s
zdY_@pn^|P?i5ppJtMR@C5P0<CQ8eM58@wGRV;GUHU%9t=r=PKo$(MMpcT{qP-K>pE
zg@gT^JwC4bUcvlB&qMgN`uxf@FKzE!+S(4D4~fm4dbpmP%CMVP2?W-bHytO?wj3qR
z-wu0$+#KKAy_-#@prv%4cES9?m}O3js;R8&jjW6hI_|sY{6>wwz{|a3?Ii77ze_#6
zvci&)THAS1kL~Z#O<Uq&@9sG)uN@y+@TumH^;P#f4Tw@R6Th>XsV1jBIS9DpxPWx5
zJlUO<DDHV<A@@dU&SO~gwd{YrqE6hba(j&<e<rI^5$?AcwOXDhcCDAqw7%DRljXoV
z<6j(-=%FT!$nN&cBCwg13EW%^&&L-gAXemD&qvurK&_{qHjc?-<)VP;ev;OMg1(!t
zOZtxvM{2gtjbmm4l=OAz{?YKDvI{R(jZHnShid1R#8GK|9Y4TL>?;bUmYXU^s}}<2
zz>N`baBOroPQCSo<HgGf+r7snj5^W~tw|TMY&mv(JYJy_pr78(2yFhR-FS+qMM-&H
zV0ZM=zYnTLP*N$7A;Jbf!Mg&l0}E66N_p`Gup4>WhMv2{AoBy);(UsUERzyIvojOJ
z_bNl6Xf5*;%5OiNs;s&%sZ7%+)bRCnjDY3HS_0=jjn+vgDOg@OB`s@+sQ$VblMgZC
zAct#0YXXqjd{b`t(ruf}f&n)0g^@LgkF^6AD09}>dJ7t<YCV*<21vC6#^0JD1x$jF
zmGP_}sSr@V5h9321e`EcuZk{>t&vrpuPz>;m}0^Vkxd-~=TML*2{J^8{&@V40=g)T
z{y}B7UJtw#-O%vd_oHU|^z9tbI{`2@_^UQJ0InG^33NffIlbb3OqtHclylnNL*sWT
zt7Mt2l0lKMGh2UX3?^?|Mb9v#^OD!S-G{V{WB3nw_4=}-xi|!v`*7#f%(9RrvLG6?
z(1LYKBM)~t9`lgXi^D`3JOOTEmM5&nVKkJ`_74UpUm9${*?t$3`le!+Ki=rl(o!wZ
zdPskJ!Erf5Yox-xySul&v!mZ~)eY7kqcyuIquY@9dm6vwovgGbMItIggV3+EYt?2R
znGP7HnE#yIqO&<@`e0y)b<(48weqOnPSBcJG-ghyt~Y=k5e2rdVJ-^6iAJ;gO@+en
zW3dKXp=Hy-0mqM`$&-a^4+QdRMsom^E(U@ZQE5aI48`}m26Sl7`6pjtak8)oDM1XZ
z3#!|fvGo+|QLexAmo@*F>$gy^ydf63oZQHXPQnD-xaGEd)hAT>uP4qAI*F4B*xI?y
zb0cR(eZ8$*Oszb2DipHBJq~z@GuwFhJe%k2Kej*A!CmL3+~X*LBY4=^{Q$)k`gvC&
zy%k|2#ST`J85en#5S9Vei9M7}?r6s0gf7c_t$1A&J3fZmMA=eEg;*!isXhHP-O1(|
zD~GJ!y5W?YjgS;W^Tk0zz5}MMUIBam67_9%fBtr7-g|4Z<v0$wQJQ?ggF?x*^5?Fo
zvYG(DEYSTqZ3LZ$OhHfY0;C0RGGYSucEQ7ahZ~!}OsT?6fNKb`H$38GW_bMd;lmJ+
z^1z*a_$m!eJe}WP#fxEv;rW546^=YIxLQG#Z)>PxdryN^(0dSzsjDU;B1~Y)@{ZWS
z#lzvl!@Xt~4*nx|z{^}f)MK$@qCNZc$k@v+U@h40`eNwKoc1hW_(Vn!8sPLFTN*uL
zlLFLKC(r2bPZW?QWzuOLXI#!uP&_H%ll<wi<m}wiTeV^^Q=@2oea4+aQJK@^ge!ym
zyk0PbMe<3WDE>OQl)l>MZ~0f+kvhg1B|bU*3ON50e%(-*rjE)$rmE)4n&onIL9c4k
zM7}ael%bovu8V;2%E9{uT}TtO5_i~J<rrK_jc^w<fdBG3l!8foFX^H7LB~WD{M8Sw
zJ3qumU2aYkgWyDYk<iMmtKGS;q(h_qk77kDTfi@g9!YN{+1lf4C?p!{kN1QCE4qxm
zyVo4uUw@|8dJM8lk4EF(ymZ}bzosr2&Xja%;kxkhx(j5*xyvfIi*>WZ^495yac?a=
zZx)}5I)%ZIhBQvRIej+DEFlnxJZevZMV2*InYd2_wBju=kicGJUy(*jt~4)x9P=!X
z91rY%z*D!>09&d0JRe%9vVd3^B{1P-XNT%T*qIpcsQW*#AXQyvA^4O63y6PVL2bH|
zE>b}-K<PVHA?CR1rt)X!(6<7gAYLBMs?DKVX=~2V*6ST3@s<kLr9o4bPR(we7cYOy
zbIQjf$TI!>@bHMaP4A(Fo)NO(nL1qrz9$P82#yaUIpa0&e5d@tlvM`*X+P5Zo<_f*
zgpGKfyW+c<j(P#0@k~DveHj12`?{ZJ_~jwG0{ASR5^rwjW9n4bbo8yYmA>(7;O)P_
zQ*9C96q^1OxmM|ad|hGOiY$>dRCN<mG2eP%oq;2^<7Z<9>iBt~dK6lTNhX$1jR=1v
zjSQNSni2+TbAv|K7kbxpl?%(ZBx{wqpSaHmbIakx&6Tkr3qv*<Cx*i}Ru@e^{r*ZB
zNMN-m>|)_y(%90M%cj&7!le9$xjQ~~VHD^0ros>f(Qw6$a*eAOect+^NK_CRQ4(4D
z3;&zNUnPI#wLsr`xTJa;!~<Qi^J(2w<}E{V>~)@8w!1TFY#vRVlRA#h^l$BbN<kvz
ztEoM02ED!NhsV6e_dMRMcx0N>QC)v|1q55@goMH6+JlE2O$B9F0w#~q5UL-Qytr4I
z+~%5U<fm!Jku@Ftw`WYZ1)zTG#whgqDspmOw9<F6)NS5-X@j#@2$#NL4xC9}W%sE>
zSOdUq)a{w{bvtmjr=_heb-l7VZ0K#ZLHb6m=wh+?GKsOVy0z+snB)i8(d!w2fsrx&
zrH5c$Sy@4unlkrCbYePOOH+tt#a^_e=qdl9<NyfN6BZFh!{erMZS=79rx_Cyu-E#s
z?VyF+3nYwm?u+tVblIaUE(;4XI~FL%oFysI#{&nw!K68Kps@|0d)YJ;ApO0#p!Fj1
z_F_3}Tk}TS_j=xXwo$YHJz$2<#)_w45}Vr=8F?HWyx6nX{CL{)a6512l!-y&wrjwy
z`8E&NT3z|mW9{g?(XE{O8(mwe#HmZ>gHqD_>RVVdIkcg2r-|Fn+A}eI@69Rx*%l8|
zhiqxxT?*PH182q!d#`!1By*j_Rd{~c)|0W6ib@+4(eHES1U|3eSYWiKbb&WSz|Gqc
ztI^>hFbES$`3i2AjBHLna;Rv&mWfcyOWAT)CitgACy+;jFqT1LFJFstzMam2qJfPs
z!&gV$S()M<M<=CQ$;^3>_#iWMMHmU3Qm{Q(aC0uu#&qDucu}O~P(}KxxYc`a){Pv~
z140#(WGn(Jhoe0K@k?*l$ngYC%pZGE2J*YmeD0r>y_De&54?Ns<MjO?djS=6)cU0K
z)ha9tSC)u)1Q$_I7)c(o4fdLqRq8IgUlC8VBuz?D<JV+kX8a)E7`1%0Nio*oc&a33
zLP!5DlQ5|ZDBE(Zl9+-*H%${oSg#)o9=<!%tMAb!%M5V9zfE>x_%2W9D%+O<a7BEw
zc&emVwc~s6+0pclgd5Y7C?1so9u-s+X!)I#^5n0k=5A9M$>kzsQ}&A|?*1Fx1zYwb
zR%G_;jt>iSkSFoCB$sA}utE`a(E51gr_3Q?@*DA=3DB?y6<D!-{<PWF8;hpa)~aJ>
zZIf&x!%N_3nP$u$BwA2-#U6){QtZx-#P0rW5qtcY&&HOR&#N;yu=6z=fiVGbd>teq
zU~5|UbIj_L<!DzBxK>mTy{CwY1}S3@C=w0Fk6t+3PDt(CZ7dtp#rKQy6;30X?3O{`
z;RNM62;j_tUDd{ra38IfP6D(8XnI%7X0s*%5dV^sdk<ri!{$&}Vy-8a@^P!nTL{a%
z8ka&J``Ts~FWhtiCv8%SlCIU;Yd%lf)P=3e9#e@-Mr$XmUtjbuPUj)?>sVbvrfE;A
zMqNC-kfKt0TxpH!KWMseDmRR*Us$x%BjpQdqU3_93_1Tc(oc`O&J)xp-tz%|rAYww
z`o#K3WxRK62z1Ll(_oSOJ+UZjIHm;7)vg<jeEMu+B1u?eGVcV;qW;TnJXoW@U1qvo
zlqd(7bDNycbNJA3`Tx%MXQT~v89@Riji2=4Z*aso@<w6oVsOgojnwtg{PhAqUBjSY
z-(oX`w5Zd8tlnfiD@%TI!%tES3iuYhGItj{W$9d3SC{NJ-{b_iIJ|OLY}9hhKV+ax
z%fH8h2dRLXa+TD(?p*X~Sfu7R1Gm`eX<cn?#KD5Zefy>Lci~o&)BDGZV-~}<B2|-8
z?w9*jQLS~CaX8;|T6O7@^-g*wTm_Z@U#LrwG#oPh*%Uo07`N&)44el7QBw6D*Ww$c
zJWRe9tK~i#0-}}m{uO$9dXM*ZE_6CmxM&k4J1p&bLc8jIHdkh1cWv$ZOdL$@bG2zM
zH^_13VZAQcFrlNWz7GPKRN}ZGsP2~1>XXp1VBbt`q3!#~2x}K7K}HsB0xp*)99=Lt
z2PWn8YuK>=-NWT{r*k{N_gLHa-{icJ{(91Nj!j^ij}}c79PEhWbeYZ4=}j-LGf!Kr
zu39huhZq{yhBc&3N{f1K56<nX6TcEXXc^PRi=`3FYx6TbqFhF{o+gyH{-Xu<uczAm
zjrSKP0*Md43<Kkh$==j6Yll<oKE$$bAIe2!|Io!}%@KIyhX!hrz43qOyn=TJkx}AJ
zU8!#2=W&VNVa9}YT3FvLB%@@Z#I(T_yQ}_ot$VH_XkY&e4|6&qOZ}Nf5~8)L`VRAo
zuCT^cZSv<c2t*rcWh1n}D!+F}Fq@c!x(<qufP5v+RQyAx<X?RqJQQ5BY{68Y>oW0v
ztZzMA>!Z2(;y>V@0MF%P<+S_qaoozfy1L#+Q+9vP{vEGis2<>;fp22!2w3s?!nnW%
zgD{99`Za5tEE$9n<QXZmv!*Uzzf~3vO(%hRXl?Jl30GA7ScO*cSCr*t_2pAJxk<Bg
zE;C>t^F1-T*|2{;iV&qGCnwhuo@_Z9FTc5-|2hqlT*3YewsBrxM@=`Gci_NUSZw<N
zUEsMu04ErWrrmza#ho`TAqtlG2!aJwWjrn_Q#5s6kkdN7gX&+e)}IwmgOZJmag68K
zS{BdA8ish}^d)<cBFXD8h-%pFyk<z4XjWsmsJt-$7M+1(Y2G-^HyTFPRoyhQCwD^|
zvW`WxfxKU1Ga?{xePHM1f17Vh`B?Jef5itoAp{eNO~}Mm3NyaC#lFp#!;jXzcmHoq
z%~FXJ!))^j_qmn;Q96%lC@Jace@xyCZ+eIbDW?ZNv(=c&#lTsb-8}1l)s+5740FiN
zIh{^!u#LwylnLn&K@4!IKX(!Wp(X~C==Q7o+CZwRm1gZT;)<4^9=F(+xx#tQ8x;z3
z?w8|FRl$hWtYP{3{R_M1{hy~dJVuv~CIeuiwBXV~<otEB9nUjjI@*JnDjb-LZE2`*
z5S`}tI$zu+&=8mWYourNLRqkj_<N8l+<Zbyb2U9KWex3SqmdXmRXsO7zejPw{c-HK
zFd*W$uFD@(J%P=4AubeIx0vV(k}uS&wR1Djepnw}X<K9%qcD%*duxv#x!h}si3wGG
z%suJ(^Rz0fwP4%W<*;rrlCn^iQ<S2E{K7m;Nw1SoM9YztoXYHWV_^7tdv!9xQ1tXj
zdpKRJhwG%@`#IxrNOKMs)0{_J!p1XRO+ZcK(-c|y`=?+HK^DI7$kqD#kF~h2F5_H_
z+uomWr#7?9q*^KWVMRJN@KCIQWAd#qE-$#y@uSlvKTbN|NR5O%D=m5HZExhb5uk7s
zM57_7+d~`xM-8g>mzCyMZPo&~)0;K$xP-YFpsEdYbj1`6xk+T;UHLXJnXnk(j*6nD
z=8WWR<os0z)9pHws7xXWW-RdXR6Sj^rm<?(;9E^uQQ?CAjOsnZnq$l7zH0|g=D<0K
z+ub_#`nFRFyCWb9+c~|w!vfau0zOh%q3w4O3`dtie|`@&jd#GPk%E|-3gjki`7d64
z7!`E+7GFNy>j0*qjQ8IuUmnl}gD(9&76I0^kjhxP+{!_HTob2Bj*~&k>UBG!zA2TT
zk9PGfNu0NgB6G^1_0%4{scBnZN!e!EvBQM34K1PzI>wMmPgUIb$35b|c%8vZlm)+J
zeS_=Zrzz_Ox7*$EOf}MAnl^(_JvXK`<-}=KisF&M?59{GPt3#DSiqxve&DcVJK>2?
zy3mulN_pj140SYJAMlC8z0oXk3uvbolY2@mr6Yb$MC=Zj`1b}|1hl#+7Fp1jigqF-
z@s?8g>9iNZ?0UMmYwRMlM<MA)x<ABYhsvLPIQj75!-U`5-Sm^?Y9~Fc_IX3bKfc#R
z`a5Xa+tgF(zjc15k_)^&B8_QCWZ`1bAS0uJV`2WxW&~ZV2#9q~n&xqby^bCm2dA=)
z5Ixhjiw(Ubt71OQE7cDo6}I2DOso+&jma~<xA^J=B;P2&z2i5vy7Iw`UzkhY>MC}>
z;`&tmv`!##poZ&Y?gp}F6IxuGhN^WQyI(ce?G#1LmNMgLr39po_9BUk)%wg5sf1Ed
zl-{2SJ-P#^x?T);iiuR#R(mgQfI!B@1GeFpw`WN$-Wy+xK77zoHPB}IYA=r3Qo>RB
zxBsDg@CA(;`ttW-Gm=gNHuL)hR9k%DpauS9;7}~#Eq6N}G7PDVvQZ6(+2aY1k(hSK
z1j)Q2#$%o#j{y!9m>OX-xpf}r3AFr1z8BM+BIGB#kENSiDl7i0mQ9*#JxC-DzXO0n
z1&cTmHuv2r@WL)N|Dvi`dwzDlx$$R%<)PtNgKwe`8TJL-UD2|p0`nOq6aIOvL@<?7
zZeF~#r6Pt_{?q8mK~b6h*ov;nzaZWLWt)hG{cj^5zXKM*izD>~S)LcY)1H=%f7e)Y
zuSd>$r(=FTYQ0<=HWBF#BdTAD{_l&8&&{BY#tE3g-5=LTdRq0DL}lGn`+icQ*4GR@
zDOdRvee(%&Pr=CqRDnwdZS$){HO{d{J&8TvqYY#8Jk-_6Zu@R^f5?|HrNEff1^;a5
zdo}X>xNBgeD~dDpx_kRr#At?boW1~iH7<R%gT1kxbitl4#J$j&aW%lDAEt+YoNtgx
z(6Hc>+~4Qidso!p>3VuH#F^#InhQf7*pMnlb2V5dM7F=#(pjn0H@tM4KX2aiJ-wW@
z=7mlx&8ipoE!qMuWB%Jz4&8K7*hzTsPt+*|1#sYdzB7M?-73GSM)i1If_9rkeU;sE
zqf0XnbJM!ISKotv=_@VONlX2CDSCFnf90qbJbPDfA??+0CMcAfG}_SL)-Qdoh~;Wu
zG`sTn7!fEOLV_XS8DQ{Zrv*hn@41PX6D?9k#YdWAJH#A_5Q=gE3aaT&M5^}OLFt`d
z0Nhi)7f;L-;#5by%xhxl)J;HoZ2k9n_HlBuiT)EZCq?Ib?ZID3b;xO}=Aws<S_5d4
z31oA+#G)v-8q|nc{8f1UUCbnfGA#yAh9wVVr;Nf#gm`8?mvDW+R%;^sVjKlCiN8_#
zO$3*lzOYK>{hBU0PiQ@(rNk#njJL|n4-hq*{EE{uOX*YTeh{)A%_6wQ%96xbvC-s)
z2MzebP337q;(s3Pjn7D@F%hzklVeG<Pz@U{7D%Ad(mY=8{*uJ#Q(~1~ah}J?Xcu_5
zJ#WJ3tV2jYBCNt{^K>5I59Z4K@Ik+~iXjG4C#c0$Sb7W}ETW^Lkt=tawbrsq0~}az
zg>DNb<jJZ*khWyS<mw5zn2%_L<KzpQ*}?p{KC{0+BTpa8s8$>rtTu`h;hJB|5-cIY
z#8M1FaGaJhH~wa9))SFqTPlP?SiC*?{e*NBsxqgbWMkUZ0&uga7TVc7qT`7tfg-;T
zKGf%4Y7?z}@|rbEfVDa!39Z8oJm_5iJdf2EiY#mY)9_PRnHZs1QTh~<+^06e<4GF0
zIEufVZt^)_i0HrgcmxbZ0;+dGt5I<KTjlY+Q;iPZ0nBoR(^sKP<mPlDTx;_9Z}e)D
zwz8(?ampuj7@i_GF6R+T*TotUyrr^9JPn)aezos5??W%RKV;<nisu;+hQfg0u%5Ot
zFe#bUdOjfmTxH$V^4PUhIQt4!R@b)rEN>PB@{W;_pOPoi(gsHx(56mk*RWwy%Q6P@
z58vpT4q7^MxiC^&q7su5;gyvPs+byzuDTShp{s+UcveLlJqa8ur9gc}#azwM?(WI{
za*qib*`Kri(YfzMHjrr*f=^}fPGI{ugg#9{`s!><sNju($2q5sfr0122A7}~Z&l9o
z>-MGT&svGs-5fE;mpo~tTb$}3Q5;-sludXn=6UqHPy>u48-Dmu=eXK`vi@)G#8!-9
z&0}KVey%`yRigPu`W-*-UzgMT-iIJj-xr*M8(!^8-%Qd1z{U3JfZOihvLa=xou=Zn
zLZm7&Buw+Ae6DXR_#nNxSyzyjl>E~J^PaSC6De;x`O{6>EU567kDXvWpnSUMgeF+X
z^XC841F^a~Q|>n#1Jc(80B?M{dV!{9H|%cUE8@kuiP?l2^U?=Uvi@LR`k*%QlOE8&
z>9Yxh^EaLJZ^2C8tpvUT#2+(#m6xmtHuM0w%E=wwoLO8KDu{%;cOS%@VQYb{xePCt
z7mEzjKUpC7AM|y9_X}oMS0I#32vTm+$uvgYnO-|<k))sA(%oaJli7O=B|HisW$%d3
zoghoTRi%GMtZ45K?}nw&92Uy{GjwS8ci;dFk{Zg`A`tlaj$x!edM_Eq)l8a_f6mDn
zeV}445oAn`eF7&@&5utl>fke$HF2`yfn)sYCNSMFI@L7t;FbKpqMoe$4@IIfw)lCc
zBq>@_cJ_Y9#b~AnvM#92Oe58bpYq(p-FebZ)dSmnk|1}xzCU(q=R>S&$*l;Xtn~n}
zqENyJI6F-&o}j59&ScVqVAuN>lBA5vgn$5~cRl&=?|ygmTUg#{z7TUcjevamKFm}(
zYk2Iw$U{0lz9WMfW96PVnRL9*n6zx@11dJx3#mK1gYUE^rC*+XqW~dk1FT6C$<UWj
z4J4jFTxYDBq|nNsyp>rrxg`0$W;viuq#AH2%bO2+fXo<T7bN9iOj_n}2qR~lLxfZw
z47~BszKkSy4Gkph8R23f)WFZ#6SYsIV_CU!d@V~7PW|zqeX@nI-ZNc4P)a%DkTp&p
z9}||zD#-et;a-))MXtey-ZzC`KYb@OE2f#sdC8keE4CCV9Tx)Niiph;UZCzdqBog}
zTJ<S;2B&I}z33yt$&j*{dZ8L+3EahXiy#$A-gsZ*Ib{d2JFCgyC39|a;{fht7Pjoz
zIxp4}{CVc*`u(-^j`L>Mo4d*}6%S1o(s3TU)IrerQB}b%XM1}A{a@T1X{!oRS@snV
zxfrGDF6S=oDi91W9aKq49|Lps4i;{=*>_IoZ`+w3i#aZ>(i&Z+G2QsIG(_uMF3hQ^
zfv9xhl%t6BJCC<_?q4<=0v`eAUwOm`%hH0XrvrlRav7u5yE0)P`%YbJ2XbngYn;HY
zT%Y^4L64XaM)iphgg@wquCSU6w5#sLMSJtifXWvMiJ6IH7`3Ul72xs1$6`Qa>{Y%_
zgv@CDANO5`tDC(mtVDV1(Z%r1MjvMu+Q8-D%5<#g;le>)!aGeJ)VsA}g{Y&4Nv9{|
zOp@OA2OZ1i<<bn~x0`pt_QJ{b`Io5eD0#h^OZ}OkK0<A6YSi)Ejk|}3$JU~HqWQg#
z$7eC1BA$<oDBTaf4=iA3_im(t6R+0W#<<TvrNuwY{QZxHGfs&Z?BB2~{wD(7w%2=@
zn%$QEtYJGtzg-9Bt#;QRo<#DLgLAu_y=Rhtr<I>`)(5>?<x6jLcXqnQj7O0^d-Xsr
zyI`#F)Ys0Af->bJ6L57i*>aUo%;d98t^R4p66v!{&o7EA1AIMSV2tlmryuu^TE5Is
zcdK>l_kHt7qWFm$W80z8^_tu5(eQAlL~V7`+2%kwjmI!orilpb1cbrTnu>efSW|a?
zFCvLQ^rx>Eq4$y6?+3u4<zt3zD*3(-tw|0YVitzVM3A88Syjp`CjT&&80KzfQBG9`
z8SV#76mTw6VC{DIa4*S2@Cmlw0ivsiVF&AiuZoBRSnO8Qg1&$ZyGc46TZg~oGG49r
z7y&6GVHZ$O?~#Qs{k>$Vv3h;)zfhGlPOO2x&P&IYKO5bv<2hWWM<bezdTt<hi91Ah
z%T4;qS<*`MK!HCKSla)h+(+|{I1_2X+xh$CUf?>AuppOt{tKR+7jH1>v(~L>sa94q
z(r=Hc{Hx4@PL!pgc4UzxGK(gbv9}RY{4*{z|AItg$Amc{XF>I&Di~&-zlpqEGio)v
zy}aG$jN2C_oS(26tIJzWn7zI>6chD05FP4i)wO+DN83Q|hx54olQMMOoFRQ3C*6Fu
zZaNLDA)cC`{aEZFeDOpf6EtI1Hl>uxiY#IFwzT-kRhCq=`9;&b1G!WNLYfF?n~VSp
z8Jm}C+#fh#96Sl<WK^8-eKqiD823t(%ix;uLv-1r^AkwpYoo_9|Bz`pA`7Z{(E@Ve
z(SB61#)EuoX67b_D8^W(sGdFG-~PS^XAS^~3HQr(x&QVbmPO_rxwY8JG$cvCpN2Xg
zrU8X;6NRkLJgG~(a}*IkjGJbj|7kD@hZH`iv@|V}iF~6*zxi|3Pfmv`uXL#HHy=UT
zDpDB)>!p(V%c4Jw@I($l`#kIf2kIeyK~DefK<>0x)y{jmITUh=#`SBtcpb;Ru>j)w
zZf<r<GW9|6)f>t4Tmq7q+v8I}G&Pd7bMNZJ?j9dDUGkaY%>ut~orf7LU&FHay7^^j
zj03u*LKj=KQC%ebmGQWbZ8ZIRf)1%-RsJ0-DD2KAZAvWK;m_RE6JwQP9xpF1#kVKm
zD@lt1iR>pfH4WHb|F733%n&vH8yE__q{wn#>YHuenJsM$$y;gcf>AewFI^YA@x8op
zwsOxNt*1_t25>vjKtVC;&*$rfN;|?|8HOW#waBU*|D4(HwPTy_N{I7kdahOOHCTX~
zp*wDK^&Xce{Rhfm6*t4n$3N~;iq0s&1KU3(L+A>y&X79$7uZmgs$&jRm!*k^EHtI5
zKbc-X9*V?D6oc^0v`J2Ft%-=7`JXi_U3AA@fv{yWN6XR&h!3dzs+7ZxaL%m1)4uE8
z-#1S{qI5l~DC6?<fdW%QLuve|)5BZK=SXGq2n8IIr2D@=%Wvj=>VzJ>0hL{?hli*0
z^Yb(LFab|&><iC1)qcQcP2<sHMlKFm=tjx3baJMYQ(2(}*oIzXs4-R~A+AR*?>n2@
zyc{^XE`6MFpRw^M9@^nG`F?DeTjBN_)9#)PX?w|p$0wn0cyTLKTu)00sJV$a*rARP
zqFnYns@w#ug1SyDCSW43P^t>-chTV>su0pSr*^t!%%`xnVBmnop2a#*h_3<6`QYwA
zDd6IJ2Fne?NLH4V(f+mOru7tFaMvf`cL<yj0mU4JsGJ+Gjud3->5IjlCM9VLA2fk%
zagyc0o#xV+J`r6AYm4)$jSNS<sK?SD()qJ{&|s*EbEm)gg6J2lh^`7zG=kQ2Fan9v
z9LjGY1@k`(qkqyV2O`InSvk6T!RK<&KPX+rJR*|Xg3)L5AiRq9Yo=%dolu&3aS@@;
z?`KY5QRc|7;$oiG{fyyP9A`@jr@Qm@25&GhI3?$QJ9lSVPl{VVl`3%)6M;o?M%twJ
z%+rscAu~$S<77O^Fy(6j1=B8<&A<&L{FCOam^8q^7Xq`kEg9;G+prg139!l|roy8d
z^3QlD@Q}5eU~3CC?T8m2d$(($0UY+7oYdBs>3l*@ARx%4bX1&GhE>|C$H&qP>0SXM
zKY}o*uCr;ln)N5LxM=x8{j^voym%VKGJ{|l0o%|^ihu&XYURTst7Kt%diw9#$C|T)
zseq@}a6Wfl4#S_h=q)q~6(I$0C!p1L<b(V`HEy^&x(f&@Q#>RJN6!52^B78Av{oNm
z(YNPTvL8O4v&7eD@q5?*NgN|ffYh^jCQt_UA3vaK;QGYO>rJFAPxB6&qIJ8$D~-M~
z#sfj6>-a-rrcqX%ID{GUTZO+xl<V*sxPJWc%+2?5roOMQs&iz!@%G$n{(N?RjwuS<
zPqWGseFMJI*v@4>97o0nnU_miZ|S^P<0YC@?hAgu{8{VdE&+sq%>7=W?NggW#~;bj
zsh%O(aRCsnfiKZlb2m5r{kk2BLFsJy>y27m&zyG=@#zge)#2;)TYR#(*CRYX0o(aO
z4%xkwiYj4W@58x1rzksXYkO(UiirR|Nk?Bdi%ZP$7A?PFu6ImviS!4q0TxqF;JGIb
zxA(UZxa}QT28~pc(d+AL@?ndkc|9i-<Y1pu*`B=7>vAhift*uwWDito6OVk-$)xqV
zyeqFbFUBXEXntDDbkd*Gdtd=%H4YNk8R(r{*ic(&QnQwp*_`>g81(!R({_*_2_F1}
z`~j9S)3$u%gdF6YaTZy?O$PWzyAT0mR-LttpV8TORISK0Ryw@OYch*($OzjB@x87y
zPBSm3Cq;xO%E02@{tp`&Sq{yA@iJZU(P6nX#$=S(PoXV4ClpU`{J{Uwbl&f5{%;#c
zm1wm_l^C_EB&GI<Rij3!z4zXG?}`$o#Eznb*8CWuMyOpYRz*><SFNI`U1~h{_lM^n
zxQ`>b-{ZQk>pWklMa!RoJ|NFt6R(&LErHplo6-E44b5FQmOBqQ`lXh8Hk#MzQeTCs
zFo0{7Gvi4G_abb%>D!eP^f>|(#$0mk^i*oW)M`)D1Oq=kpq=iVJ^`8Z_LdYc+sA+K
z=TMq!RE!@L566Mx)oyE0zrb+O3gwf$;1FiZrV@28d_f2Q4yC{Aemg^aIG#j7V^OFt
z3d}I?^j2t$Tl6abbL_1WgPMbembTQ)o;C>0OF@@yUlHl!1xm=<UsV0|jaw)b)V9<5
zDQ2z_ZCGCtn@@{?3cVB`&Jm(@c`tRh<bB1?_OcYT;ioVs6nez>N(i=@=UuF)$H>t9
zQ)>7kw_gj6YF`)ETUV6W=K5Z)+VMT$v&c0Jn;KCZ%71WVXH4?fR4VZIkC9xqNzEt9
zogW%IkMDMVX%qcz!6T_n<fxnfJp2=;C1_*T?dD2iDdH?ig01!Mari#`Z(?T~6uj6H
zD8VVgU7+O@>ZF*UxAOB)aq@?<t)#skM1FtMt4e7CUS0XLv#qV;bDaZ5y(33U1t09#
zxh!2Z;16}61h@4gJm*M0e>;x$dYv0b8=D&as(XO(R`S;e3hB}QU22sr6|L369YT4q
zj64It_zY9Ba*8KTg?dU27SX=yPf3Hm<}O+doYYGoVJCKU<Ztj<41Bj$*q=e}pt8wM
zdHb0ToZ)s$i>&W9>PU00xv`qmzTHZo!P5`Im7({KIvb$ryEiwsiTV#j^iG$65u$_-
z)KWdd$PQNshak=mZ@z!jmSw>nUHF><s=7MpV>{y%IvJq&%{`{~VV<P_>GtFqMwjUC
zSloV-XR9D2ek`Y+hvDMZ%Xh>8W3@>OEf)&k_&o!_L8TViysHF6x;)FTcY`C>GvSwA
z4iS)IdZZqxM9gx6#P^j{Z^iG-lK>_~3v+Wh8Cc#3Q;A{P+^^Eu^q7b!4l$83Y-2)J
zJ2BfIHL)J>aAb|=hthj#4Qo<g2!8y{@my{Dr@=w*j;1G9cgf1u&;3)7RIEmixU=BZ
zo{#PmlD(=<T^c=6sCgpt_?1L$Ss|~5%9{3z-^NF;tlKUJ(XBsy>MZWd{awp2odmAP
z73QVCPWqD_<u&`y2grK{{#Nb59vt1D7`F+z!J|bs`K?+ngnGU-Hm1{R-B)d|+vEzs
zw1uCYZfvC8-PG27shYxSX=!C@W$Ft5;3S@V2a|SGRHmhQ|CCPKw45h{d+OXl*a^w|
zB-Ji|ZXsqY2GHWP|8;5o=TmbUhrRiEGzyrknG*^PIQJa`4PW6fP*h$04Bv2_E&O}q
z?&&FgbK~8}>76n%@7I1s&WB+#>&QZ?uN&2-N<jAZpLYmq_Evh?FAw^)=R4q`?JWFN
zo-F9(AVw9+F{W1#?$xKOPGMgPpjMh^Dd_fM)XQGI;_#-u*H%AiN*fGm8=ld>CpZ^P
zYdO7k=j_ucM-0T;AmeMBot}P^lvZTtWYAM%)VD*@6xskjo7@}qUdc_asQ7a**r&Jn
z;RCZL-;-7ezh;|F`(Hk#|Bw$aiC1H+#w4k!7zcXxg_3h=R5!o(oLX-#-#NAMHa;=>
zclTB<)ysjprO~?Ie{7cAjFMFA4_Iw=^tkjgDsn2RBzZ6o8IX4eL`E2GkJW|wi?Q!-
zsT%iX7dLo{B#kk<qjXS(@AcesHfOFCG9lGfRX;pMDy2V#m*D~5{!A5o?#pYP*j_2N
zc>ni*Vp6;flM#d6#rfa3iSJ}MLX|J`45HaO;H&P|@Qc&-o#$UT^|SIrHZEiR@l)?O
zqo^&&3RuH`Zc)iykeU|VZa*2~m7ROM*3&g!yV?(49(&whvlRISi*-LzbbrohT+3c^
ze0^O!#oIq&>g8~}Y{7X9pD<m&Wo}|69lB>RGiqTh=E+;ER;5xnJZhB;@G>|3>(OX5
z_0QivERUI$28KUPzK;j5TDXv<fo652<ut2B1`e&{lmd{ccJW%w6y02EuB<!hdCCKs
z5AVbh6#Z-yCjERq2Tqyg^vzE}Jl@<v(a1yJe$*7;KCbN>@TZG`*O4niyzJtUyT06S
zdlWeRyyS(Ra(qw^(u@(}P{AwzGE=Y^*0YNyMZBX8+ovCT<BxW#%h9T<Gk*Wt&O<yG
zYX7$9)6suPnlIzC?Z$C;9A8A2x`wy2I`ciDdoKVu=>Be8Ip>Eefy5uB9CGfrJ<xl8
zRwBTtFIsOJ{}Ve^4QzC_tLyI1EU5OT>b7%?_4$)hE}GAXjKSWx&(qAmlkO9Yjb9u4
z5@}RbY_|zkyEk)^T`n%P^76V9v2c)lUViPJ{R*1B68-KVWPW#dJyc7HCfy$&9=!dd
zyft|5eU58ukpFAhu+3TS;RYeM*)Kgkz5*iBXL}ZG`)IjvBe-keMs0uO@haTP)bx4q
zPnfv=W1Vgb>2IL;FD7#`wUHO&+L5Q@4Q}|nP8cQ^-3VlDT&!mCKR@_eG;;je)9lT1
zi<5=nRY{4=Fn65(pfK*^P7CMgQs}>3fyjfz0*k8<xTv`L@!cK{<*pPq>9C)@wW#qK
zk+PkZy;(1BZ@9M%>w~eXYMuCuve9IfI80?)4F2ew86b!GBCn(<uc!%XU+H!bUXX)*
zTqf&zh(U>|yXt0lMf@skIK)e2l&3!0oDW|Le6*QI@jB#7XGilQTNNMWMZY%dQcHLS
zO|a$gyFRexg@=p?n|dGW@Z0^1n3$e-+Q;A4c_ObDrQtD=wQUV;t@WAFyfEtJpFjWj
zh`aOn`ntZ`Y5Se27!S;@FarEuauI*dJ&rlJ@k-+Tp}Bf<bvb)eo>@?K`*_Yx7N|PM
zoA|rQQONq2kdp7SMC_{?cLsq4i4?+!(UgECH}iz-WJow@d=5W7)MFPcxmP+M*<4-U
zo|&060!ROwQXPTRM$Oxc^<)#tI$PYq6SY<DwB%bC@MM%ec1=oom$1k_{0f&tRehux
zaJ@o!G-W`i*uB=GW1I|nSBO$)#zV}C(skw2K>8X81a16(ZnFp%Z2p%IG`w}RLTS{b
z8W7!JO%f9`)Ai_J4o_B7^BGv5p*MV9YrC-=F8cxxuGyKnl({=VNM>!WKhurQh6sr>
z)RvAq4u(~_OYC_*<Ez>A^n8G`SDgYs?j@q!%=9Xy+q~=@%xG$>QL)GTNV8Y<T4?$A
zt$q~hNI@cWp6MOw@A&#?<3t;nqqsCk%Yrq@VieSU_7{Vk42o{6EMYA-r7-pOJX)-9
z*+{;>r#1~)dQ+q7L_f^<@Bj0Z-8~`8s|RQ3JsnPr8M}56H8P!FTsYvm=;2s>jaN!@
zlviR0tK_rjQgSpv2ioX*g&NY8yFnVD5=6QR1^3^!kj-{gQcwRM#lj}tlQVlkFMCP%
zTaqn^3ZDi+t~SPTvj-C>hBbY$92IN$E6kfGISi~b3+@v1<n7p`%t8pv`T3W5%118<
zFpiB>&8iBm`WHpvQtwPvt2J>$c%U}C`|1^G)lP(N`uB10PF~TBhplhlKe`W%J1u~+
zXr)~O=U^&c_FRz6A|8vyA}A=HtlnHN-85~E=3VU-+$^>C**QCJUdB3To}a>{*}%ib
zQ^wR2{-c)Hi)S6d2i@lxJlZ}{<V~!193v(cm%?n26{YqXwpu7iLZN>kQtOU2T?rAG
z?YF?Q>Quyx!z?g|P>&yUW41T3vijis?cKnJ9`8)h!iag3kuibQV?kiJt%4_&R>#de
zE3>_>2I#HH39!LeVvVl>{fkxT77qhphv@?HvhI$aN;VduS>r4iv*W}NgwUL4RkJHY
zRf(M!JDrj1Wdgfat&`T4b0>TAeTklyDW}1AGL-id`TRXa+@&qo0>y?!0Cy77_RIHN
zxp?c=&=-zZb`bV}=|G_$GQ4#fDgE{wy;bF?apjyRloK-{;<@)2;Pg4Dn8+D}`f2Er
zSdo|L=GZG}32N~?REUfCh0<n>E)J~kbb-e(r&xuo0bFzk(y>rHAa4Vg2<II&hv`6{
zVDG1Wn62@qo2rSdYwg4~H**p-?t8Dd{Aww6)|?Pk1lc30qi+Q~55S)>yk!6yffIlM
zjb&%>Q?OI0r#(CvE@9KVR=hW4p~>SagXXV3Mh-SMvdV<5M0;<$Psv?Gh=;ige+tk=
z@w%(VCAwEj8Vr4w=lD|E)%Kl(v|NePK#!f4C_@jQR(}M9s<Lw{@Bs{#3>6-wWtr0W
zmwI>Qk$FTgVsb??KC*sL)tF<>Nb1Y!`V=|%$@ZLdJ$&~iaduta<M<ZO;+HIXh4QFk
z8zQ^V(5)NC-S#lP);gKa_C~qW3+CsXuYt@+eXhhRyN8K}yZy(<L?SK<)E{}oHeS6H
zAOZ3#e=)zjDr78I&F(t>Ci<g8gJF;}(NJ_Ndr?X0ZL>tRY^1rYh4?Y?2lBJKd97zw
z&b!#_oin&KLFKVP$??FTe2Bwvg)LoHVuaLl!@hY@LsKK=l6=;-&#ZlFbp3S@r@We$
zfjb<D?aBl83V9-drp~6^2#b62YOEnt+0>-Plj^ypuE<wXxA3t?)sy$6uf)0w3ma$c
z)?B{+SMkF`n0MRF^L0ilQ7D7oZ*&pj#}FYYXt5RUJ)kM*MLs~UJ|{mUWBywXPGCMI
zWypLA<-nUxnXs=VHFaG5T)zGAAxD64j;*41+l5;Zr7eiVKy-V^=FjyxX6ceg?&;=C
zdpI!7rQ&g(sHh;8dBT<49p#VDp(MdM0$t!CNBxPG544=yE!$@V(thygv8M??Wk!2o
z+(<YLPHMplTtW}w<{aH%z14-(ANnQtrxMzQB4AH=j}iKfL2i-{p#HEwL>6~cKEHUY
zV~6!4Kz4ry{Wh?yyyLGyqUG+DTJio~P-OGKQQe-$Z^5BQIvjf6h^es&u9*~!cX4M{
zVQP0eV6q*-JN!k(a=*)V=_GfCaR1GQE4_pJ5FR=%z+#FgRD?NQP`zs4;8`G=C{Q!!
z<=i8Ujz-(@Bs;~m&`(%QE<pZ?Q;4U?v~6Q5ui0TDg4k<X!Xy9T$>naACyfO-9p0%L
z8j>kw_ioj|PtE}?yruVyD%;i1;K;K$JY(F-{P}{!V#F0N8<YE2BzI8+iHBbuz@6hC
zB7_&f-p|OEc7y1T)`x6050#)s;zLrcoT3JSX_<W=mxA%pXNo==z&|P(xJY!WoIi;z
zLx!^pgMb~aUXH~}_d>*m-m&{c<<*la7_P7C%l+m$Ue5xQeg2<cui%)MWFteen4Zm2
ze9HDEM%i>JC;^PQ)v<4|8bkQeP*RrM`ON4y(%7D)E+333|Na6zAkw)tt2FVvp?n8I
znHsDM4A(y{6=>jD<sw;lwr~X8Oh8L`<iwV+!21CvB4qmry-3%8LT~LHFe?_IuKI7m
zQoCABB7?7r40E1UTXcl{BRnGk6DjK&7-kY$s_)>-RY9N^peap%8F;?MyL1@c#KD+x
zFyJ=!G&T$rXT7xyd3SNO5+G$6`&(udOQ>J3TG%}y->LAb5Z>%%>xBId3^$iHj~k^D
z;vO>5V0<thUhwb<&LRI9Z%O1%Q{-yBKu2DFzB;ZDLj1FT)hXQBSyZaA)To6!Tgv^c
zfGzYx2Vv{+)2~J60r2$F0|`eN#bOY3X|@kJ9dUWZ2dT+~Vt*_)ESCF-wdw>(TlCB?
zpVOn^pn5G2^LFhKQCBZSnXwP3E3je!^CO0c+7Bf7V4kT*5DU`(Db&^OCF5cJvPBx*
z6&CFOZ>z{WaUy0&+Mem{J7TY@L|95OsXS7L(n0jYH)sEN_~$XP5QdkZ8t9SdWO;<?
ze=t<DD}pi^cQh8@gdBc9W5yG@HRNcxIc-jYmv&6*LeXlVy@n!aeOTjU%r#a84_F&~
zQND~gM@tS;3|>$}npZE^2nVaclGBR1?cEbjy46o01RGVb8)mtV4-S2E(aXZc1mHO0
zzDpRC?ehSQf>Ab&{<3#StdR9$+v}B~Of7-iA3-#m04FL*M6Ggx`TpVhxT*Q%2@vJ@
z2>0$*57a~zwxVn`n5y<hdSNZo*xRBdVhpTL=ZG<&Ce@{WA1YKZQYX_iJ<QdJKmCWU
zr@0Gz82Ej=AyA);rl@zyuuoYYl9la8V^UO3je-wW7**hrsw&bX)UosCF`zz~fC|p+
zkRpqsD#D_@AztZ+3al5QhZhR*$+w}j$sG^h%=7CQ`zC$wE8;&WvHByjZzPyUf}-Xq
zZTN-m_sBo9()+(X_M*<Bpxobh-7zWVo?$DOxC<SI;jzK7i0A$|8+ndZCUdBwbNktU
z<#4QULYOO!Ax@2vv8=5AeD*?$K>^I801izd;=nbJKjCmFxw;pVD&&N@7wwWsfpKcX
z02{6=#Ly52o)lG7!f)j^b@YPo(zL;3hK@XL4m@g2+Ae0r=<8Z^@aOyRn|}hQ)|Xi~
zmz(>q_sN6V7TbgDTmKuzD#(pGk)EH!J)AI`dE~-6EdciNnS1~0yW8p9o|aQ;>5XQd
z_9!nV_oO$O%5Bv<AB57md<`884P(`Y&9uukR;k-6hgiToN?`X8`J`$goNFHST6>@w
zqh;mG0k8B_e})%7OZ&Cc>Y-f<bYxc@DiOlNtAfwNe7`SEM?Mb|ot){zE+n<Xk22H<
zyN&gQe9S$Sw5Fqs{);j7=|ULK2~lJ%B%xbSM?G*3vwEdUw;EYv?2yxxZ`Y>RtE*1Y
z7f<~ara3X=5fl*-A;r{rNC?iC3*SID2ObS$(Ct#@Ycu2FOIraI6|~V(mFS~P@x0nD
zpWW_E<BNs0dTn@feY*lQM+07}o1QNX_kE6*v~LLHo$cr{|9>%o(g%W+TKb@aBDnqZ
za5!*`Wmit_(Ic;_6LXA%RL-}J#GR$EtJ7Wcmr$IE>3iPPC0OOq>?O|odILQt@#4-H
z0p}7Ro|AuXiU5AH%Akt5p0x`t%^i9$zvkxYxl?eH2yf;yfY(eN60~skiqjGzuU`;}
z4CS5prqgx{I>y@|sL7CnQ#kL)l7yqmHVv?2TOg_Zex6W*UN=~oz~SEBM8qmswXq@T
zOv+(0^NW}hVWVM`&6ALZf+d7f4`Gu=ut{i4yL(N?K@sN2+l+88JXZ~F@A>bNu-JEv
zGn!@QMpwf@-Rpj#5dnN)N7BQ-ACl`0I3m)bpjN}o+lSSejZ$bW>ex{#swDF6zuSA?
zK(HbROkJjn^>Yt6)Y%Z73Gnk0F7KEr5eq7QnXvQ@CN#W-Y#(1R>yE!N>k-@={O_KL
z9=#a?^FndR%NvYs9U`VGG@5mN;tD2pp6GLF`43Ka$m;5uceJGQj@7dL^J4OUo=-7`
z*}WMvK$oX8fL`dSkQ0&U@Ll^V!UlF~iCW>iXv@NuXkZdCxCjV(TBXm(@zT)WDXpDl
zWKC^&Afl%9aw5pBJ!u>)(o&0U55JzbY2T;#sYqYq7EBOt3^GFpr5U@<p0er=IZ_Ma
z^5DTB`Il({Qo<m?x9q8OJ$4-?{Pdvnhs?JGRh6i@Vb06Qr+$_0-fsFEDq<*V<Qt(o
z2-tq+n4u-Jv!I}0WT?2<aG@7d(Zfwj8l+^RR6Dl)s5Ptl-c(b8%nZ<{12R(nj<Ui3
zJg5;<7<g>Jf9N447vd{u$xrWS3wtFV2gf9FB$wbKb=RW(?Zq55;-mbvWV+(n_jgL4
zR+fOKZD79}T6NRO6uxwm=>P&)#RNzKcokU>NzNo7z8lJ0>sBMEL72bVUe4jn9LLKm
zFBJc*tgIw5#990q3sDlBS(3Z_6GH|6+jAxqWX7vV!GIDv{nvT(PY?0GZ<2C2^4G^D
z?Vkwh7V&uxl%oYgR|zVrNpZhi#Waq{elv}{Kh~uw<LYV?Khuz@r=BEUIu?il*%CR3
z*(Oti7)fvoR-@t8hLg?hce8!tK9dt%+Vcbw>}K`@8oa>KM&qV)^D0qgJpI<!7se_1
z&u7~GCjH(ynutjaCqyOodHBIYKBv$OZRAj<cj3=t#~m#nZ(6fFLL>WG|I~SK4}W9g
zRsQaFGSpISF!JPN*=+Z8%Vl14#7_KGw7lQ-?=5}k<Du9_dRukSy`CsNPB2!5L0K0y
zNx<*%0ufznT|J!UVeYq}_2)?4j^_F}$(j$d($1MA^4aMvW6nph%Uy8%hL0pVVEp*c
z`B6m9PAPvv9}uZj-?U0pRnN;UMFNw}buc~0Oj+e3liM+?N?x8-la7*TzC2&TERZR9
z!dtV*Zs7t56xGWo#uz{SZr$$)KOz8%mkPcQ729Y3P^mmAY`H3v%8`GMw*V-#XH<da
z5xdLPFV+2<!JC6lIJs`a?73<gPm=hO#>Xt%?{0yGU!KIkTmy;q`ANBT2(iJ7W}A7p
zQ`(M@dX?YQDLosyKtj=JN0qOUohtN?_hmv5G&&l2X9yjnPzj0}Aw;bXX25gcAl9ZY
zO?TUrL25Mj#qriRgtO|-ljjK8+tGIz5geI?gbfb>Lcjxep2(q2rx}92VAZ8Go18Av
zAyaOYVNTiIj-t7YfZuos3*p0@Z*MBF1G#ev8kOV!>FwO%*h!sGkucQV6e)2tlo()M
zOFR1-l3HxZ|Cr-3hdnnfdGsH!PkdCk66OUfi0U!d-(r(P9bqVz0G4rvQLX{^U{#gO
z=Ayy`WFOuM_4XlUF+<Z4NnP!=%7tXTx>_6C-Et{pcFudj60QV!tTvD@_qT#odsf&h
zVenw3Cxbzt;VHBU{Oz|(Z?h5|B!u*vsh%+zWAB@nA1SC3W(ptm(-qlvPN-#w(z`S7
z7C&>hWUeWKmFh9YnWF=rDsltG_ABYvO4&g$%_%tO9b`poGOZV#8e}*0*%wbj0ozA(
zr4EU|)1wJinkG?cpIG;BIAlD$N0LIJinz7DOmiDij#B82U5*YHR1y>FiPrVB$PNet
zs@KKXk;<Bema8t6h9Du9{C!!uu*2m=&jh|BFJZ~He>(-N5vOMvJi)_-5GH%EJ3Yk<
z+i(yxNi1|F-DrunFz+EkWH6hZ2zDpqD687T)7qt}RfCLF1a~Xtvk)PC$t4LIeE-3a
za`qp=QrmCcV4-)P&z7UuckZo|g2_k>Evsv4WWR)W4o(80pmcifn}n=TZuY-(H)r1Y
zGUq>cgRL9s-*8(!wuF*-c~1XZ3E$r`GlzR<r&y6jk2oJ)#W3`oc(PkCY7R`ju8O*?
zh%=wtgse`?bi<@s(JE=*fa1cYfCB$td^bI;DpL$cs_=dLemuyj-gH8Av_Ljc>+T?{
z+-nwuEvk`7|4y`Bbedtyf4OyDAF3)ckGz2Z4oS($`mu%4=T|@L1@8Iua_*WoTfny~
z<V(F0JcO+ArNc9YX669ZxCWi9ny0{fhmkv_KnCMx=^7`fZ1S1?k=JiQ{M<r5d4yE{
zF1j__F~^@w<_?18lsw8JP2it9^Dw9N!heE_C><3w)AtKropaXaPJJiC3@{GdFVT05
zPll!s0$LiePr9RM`CzI85@uO}_3)<P=dMj<{p1{F?pN2dHm`%kN9g?>?j3FrFuBs3
zXEI#|FP|zI8R0IBJ5F%*<Y%|u0>{A1<M8$ce}<z%CUkP1-n-?_i#~zs&g&LR`_=Cb
zEg@U$%nW$X&S^NUFe|Cv%dJ%MH3<m`fY@Ci2lqlcO<x}K$i3cdm1~P!g^vKwx%n-3
zEl12lG;m7{PXT@N`o%IGf9$~zlMNcfSz3A<?hqvhB>+Vzgm9UF6pajx%t#^x5279^
z^0E{O;cD!_M0I$^7-SzScpcBN`KaWZ89A#xDc&qDLYGb*>W!}c;1M9qmc;GsocXlJ
zL}@qV>~M0Cu&&~7Ik^tMba?*!1Ir_Pf0<+2y$kmRud>PQkW|iJSm_@2c(i{MH3OgP
z6I+(7BsZk2%=6ks{pi_p6lI}c0Qk`(*#Zp}qbU;eXd3q7%me%b&j|gB7A012;QjN4
z3FBFZqk6$)*B^SIhLe(U;PPPfv?%hJ$S$6ARA^mb>-Bw^2Bn8jm_0M?mgOlLqmv2Q
zB(^7NxOgp+5_jOD2+~ytLW`3|Ft}fHP5UtSw6mxvS_AswgRQ~#)?iuop%6pBfhOJL
zO)GCE5tu0#=EG)66@@u5%bhIlKnW+WveVge5r=ay+N2P}x}s~@shQn(!ds<%S%S&0
zMn*IdzMV^Dd_esZMBNz8?iu_Rb=y$?_yQeN_PB>jE3Dm$jYxn#QD&4z5mnEg-~iTT
zx&w8ji`50kegecCv*x!KiRIZJGb&~Kb|XIaO7Eh{){sJ>X@ii2pQ4_D;a4ZcO&sHq
z<6DzB-86)6K|Jd1!RpnOp1D$=&zJD?LwEEk_boL(c)CFX<cUVuH|Z<v2wIkDKu==G
zjTES}Y^&ZJvRlke*WDN4=!T`+5|M`QH&Y8Ezo698L#UAc9P4)?Q%lQ(lc@_)wFK$7
zycdf-Zji8M3yb!!_2Xj$XyE>JyH?ZD^}l_4jh`?qq2q*Le(5fvxYOR3O+<|cw5<_(
zbg^(xl+nqW2Vbm9$yR`{By$S+mDqmQ<Bt}oit?Qc;5ne$#zv#}jve2I+T`)%iu{|2
zZ51Ib%=?8e$?_)8{c25waX(>C6CEv`oQw3=s+Rf8`R|{dkzWyWw^aZQp*%6=4iggi
z7x<sOId1{&`?1d!zeKL=8rQr&7*tK!TY|$6BjmQIUd4=VnCcReD&k_QPZp&RLa#C$
z++W*2FQ7Ml+ZAW4GnHvHqG$d>(y35x`mgiNF_qNTpUj=53yGVaPD$Cw+IC3`Yc~Am
z5joY~V?-U_-5%gN-<&g_TixmnaV9pdFm+fzI8dw1G;rL*m^wWeA`-m#nV8$(Uj(n?
zBi%gPo6oeM<#%&;PQ3i}>+g8%kzv|-sZm9~=BQKsJ8BQu?=0oWKbPm~FApborOoHU
zlUCw*>KhhX{dR{AwpM&#CfEZ(Tof@8z|SwA+eNAr@eNaOv)%k$o$5SsLN!G01V`@w
zUJD2da!RTtz&r)XICAu`dVyfFP55X*!3dy)LH4f}#k@vbj?7uZ5Z$m(NE*sFHi3q)
z&5j&-HXc_nF59nr8n+RY-dGOG0Y%YnwgC!OZbx}2*63DO{KrY1&J#kYDyS>l?jvgq
zIRX?!c)g2G65{B+b+{<3PC@LRbXE<yqbOA3Q1X%Y@ci%ZfG4^-Np>i>-q63RK<MM5
zotd>SCV0iJ;HJi)#^7`=t4|HgMXL>yTR}aLW1^_?y65WB7Tkz;@+zNF(J;5zpWS-^
z0Wxym`m(EYc6#10@HU9!E32R_CvwiBSmtHA(J+TK^icxcW2Sg{vB{8tFBW+eP@xSx
zr+fP(uv{tnLi(N^fMwt%O6sn8duP&M3lu{54y){Ii}zwh^KMr+{hQv9B4^~IN9YE#
zNBaj^*7ITiq?b65i}2$=9x|r{Qw&k7Tk~J4o3jFZnFwCV2qjE;WOuMO*fZmxP9Y^L
zd)YS%?||jz%t9S8rWl5hed3qkp@;QVu21(pO*g=|{P?Z;cHk|)1I*gH%ybR14;Ik=
zNsWm3Aa@<(_tb-?0SDcNUro%VtsNHd-_7`9Y}P3|{N@8ej)9iq_ib0AB-e6P-0eZ^
zh7UD6NY_YL?^8ca9~vO$-~+A3lWCa0Q10yk(Rc~nYwg+F@bcbC*SMGshdSg16g<+Z
z8y|wOjykUnBH@4JgTiAHvuZky_BLKk0m$n7@hwh&`51_AJi@n9<d^Awn<{QrPa-)k
zSsAlpR4GBNC)})<D#h_)NRrd!?I-cKg3a|!Sr0SCODDFJY!TF)C^t+Igd4$0%G+r~
z<b<i&{&{)sABgGv`*T>vx(zr5q@|@@$5mJh@{<>D;4A+s^<$`iMhk06J}8<P#F#fB
ziWArKqU=cz*ZIsfk}7_(N7mkShb=2O5cBNvb2vE03o#|NEbLwQxM1;FuLhgfWROLw
zqxpMzbGMM01cGEYxUaapBC-`*-%L;b^KT!qA5K)0Hg<Ikf>L1Ta&CjX^4uNiAGJCs
zFi(ra4UE4rzV(nuirl<9tJ|6}wq9ty`9(}!P3hnm0}((K%6z^rxi2VBH)LDmZcxOG
zy}8^PON`;V^XTpB7;N}%(KE6k&{Zxlpn#uw-)KXpJ;FBH-GbABkKZj1ndj0uBcrgQ
zvLc#Q5`AKgJY*?(TwH5-I6!sOp^amS*(!#QZOqP2zRIc^FRGugYkpZ2^4z=^(+JW}
zIp>wI@=owe#gWL6^@1%+Zlddxl9EKV@}EbJ{|IE_(ax8dlZ9`UWLZDY)jBPZv!`=R
zybVY~><#Rb#2d?ZDi&P5y}eVkMn9N29?cvKzEs)*4E62xc$Xq3ePe<fWjwn!BN-ft
zs(?NLofoYZot&J+T)jqUl-ya>fy0hj1>br!Q^G`W5h5egT&}iK+@Nm3@kM1<9`&#j
z&a+0RY;ci~FN(x;SN;!JaQ%lF;DGy+O{-9@Ks3N?pLZrN=;-nAhf)8WWFj@$*F|Yc
zc_=s1ZFjrnKjuU=dC+@30DN3R0p}R*`{rhtX(}s8QJY+(qX)3ui=Np?<!En)DvXDC
z+S_0h)KzVXPz`feIztLIqn~eAn(Z!bZMhb0zseHu^0fL2He5f5sCGs5>YC4yfj1Sl
zEj>R!Q7EibG!8U`dMCyA28frSN?p|j{mF<A7bU5psKMk>UD7<TQ>{2oxE;cFls*IM
z_@$yUsOF05{e8F0p-|GOj@q2^gAf%^hA<dNIB7jwzHpZS4kosxrOv<|qq3!Yp)T#A
zA^VjcQpB%NSl7y)Vm5TLG#Y=^YTbSmGX+pD?6v-AU!U$@`<!(?uKjFgx?hEz4v#)~
zid6++o=*o_(j;&>r#ZkK^D9fAJ{^tuZq~%d%W;d5)KNE{T`5~n>&-Z1t5^5O4>En^
zqtnHjJvOXa2&pbh7?u1jBxUVGA;)DB`)7XlB})uhzuZ@(w^sq55I%xTr@;Cg3JNL+
z-))^F5JK@Y`)|s^y^Y_$whp@LPi&i>y75JuOYHylypQ(0PgSr7-}3@*5FfF#%Ci9B
zqbjRbtr-IJSh49BmRZAg?~u25o~&v~y)yCkl-~BJ6I_q}5$ITTW<&vX0Mi*J2|g6Q
zUl^bGabXwd>j7+Zb=rB+57ZS{E7#wLGR{b$r>QRp6<#&Z5SPaj+E^nAFN+@V{@xy%
zC7I~^2#`yLS-J-1AK=g`gI!t3&hv{u4@%s=eIA!mqeZlxK}L@XA+s2-vI*?K=tkl$
z8m-(|1xRxD1+s==$wiR#Je;q^)XA@7l#UemzIkoP;ju4RDf@QpP+>Md{oO8C{fBz-
zIzvI>`+xrySAW1@&`+&J1C7=by*tl0V#v=r+iuo=p2J^#q^pzv%FZojQY@zM7YNop
z)z*4`KBL`nw&=}`_G*%Y<$OyBeUp}A?mGSs15_Lw&GQQjlCq^4BF0&L4Av3Xn5m_C
znWlfsm975-tS%?5ud}%Bd};VcP9<R7k<Hb$olZw8oLqG9*F%Q&B<j0LtS&0^9*GNb
zX?nI6&jt6T9C#p4udC={x&iuqq2iq{a8@8UH?tTNKOg=V0*Yx^c$V;V&B8Dii<62G
zQW+4z)$#R!o6J2i-)6R4o}0hNl5p0oH)L@4ny*!%^gz0Fu|jVtv>5xQZ8h>9y3Vqq
z^J85F(IaF%&kWzwwN~+e4d1D1vT{$PXvBw7>A?}WX)ZM_XA$S!v!&NNWxy<FcfalA
z$okSxa9&5<*OjIYBTBCrp)VEI4>rdA(0qoVOizvvKabFeSZ!OUG~LG%zLq<97{Whh
z+acS8hz7uk6d)*+JJ8ZTnsJkt{JAH9F$2U@!YcYiih=-smIEP*t;l?WU~rt_nKps%
zR%Hnj?JII<Dn|=ixV@+~=cB*dF1i>_YO2e(A@brjdkVA&DZLLV88E{)6WSJ-VV2HE
z^*c3f*R#yij3bOkUy+qriGD(IJ4$MFfz(B&`e7O@B;0nS!blfWLSn>wl`hWVqGE|(
zHRj|vy)_yI*3dPh(>GChFA#S7kNn_6w_F2~X!6iWJ+SSd)~K<qt(VY}B;~*Eu-YS+
z+r$_Ba^&RX;8m^GvkhkNc9_%^u#}fVH+NUB4r0ius@pHeV^3r2_*msamQl8OJibH9
ze*tmW-$CujD&CVn*FA|-;@Vo)UxXh8gc+JPNmFusCj1EO5hT5dTD(mSUIpsE9vvxX
z2#)kIM~J?Wm|{=ll393=R2^Y&*W-z+p5a5BVD3nL+pnOw;l2oa|4P31jr>E7GkXDr
zFjEId%PEwu`kMplylw<%G==@_Kb6eq)^A@|8WufDHX~L<97YX)6uKuAOFp$QZJRyR
zVkVU7@Pkvg^#BxV&EU^U-U<h({Zm%w2juYn&kt(_Te_EzieA|BQO~RE+dm?=*cc9C
zVw^MxwPIMed(Lp5993RH8ZDH$(8V-Bn$5REOYHWfykE)OIW1+P*KG6)sQbUMSrfSe
z0V2Uq9_yb6zVIC~-Ey&~a2l{z{7)PkIBV8fUw<^YkekJCZDMWhY+(U!Jw41IA0=J}
zU_SddfBu2><v|aMnaufi=9;`ka8AQ&)34rMS?SW@XN>Ch@`mqDifU%}+Jfl=OOoCd
zzJy{;`1mc`;)NQ`!gokl-%z6T%2dhGkvvK1g+@t-F-Z%9TO6A*YMefAqs|t?+{4lM
z^tHHu<hER(6KyuR>}2?8<i93(lMA12Y_AEmQk01XDllF6H%pO2{nA#$Pb&E7s_{~o
zYJ=2HI0Kb7E}VHs!m3sG$NBY9nJ$=GW4l6aLaB25a`D7jJ3r#^OzsK|UyM928CDqR
zXuEQ>au_}@oYNxF;$lGQ#iIQ0b9A=0YC`b|3BI~NMqDE1-Qd9q6|B|wpul*Fl!ds8
z65@ayLvRu?klDHWul`#8{=f}7H!rP9EHNDG=5{Au2&n)msS;Bc13%u(bED(vbt?+2
z4F;Obsa(W1&I}Xp^cSNobAm9va{nFRXJ7#7L^2DU0igs~Y?#i=TsG^Og&eeuJ<68y
zJ<+MPvyeQTqR3!t)|3=%hCXew3corUm3uJ|8G1Avym-$4fDfsN)afvZJYFF`I+s8|
z$+`tsz`rK0$_#uoO5$74UE=r5aMDBIE`*K~vOE*uedx1i`)WatSc$nSpYiuZy?O)f
zLlIqnFzKY7twD+7B`KHNdBD;;)a;wGxQ6OiC2%HpO5n|ai<nBXsol-P(JNZ1K@&l8
zNPES2HzqCDbS?{!J<a-?R0<n6^lhYpzSKg9d&N`L0pXu%-l(VC8AYW`m0@qwEN5;z
ztkc@cfIaM)n06?fY=6+oFda`7!v2sd%`ZyAWj#E_AYWw)ql+f*+WLT^bTlMwqT7Pa
z;F*2tQ345kpwK)VXR5Qa^Xfr4_C2lBvqM%)7<kFxBmDF@%BH2YX~C_3Q(%|n5sUe{
zb?DARo4t)q>;I}Euh#}0gioWMIVIoDKI)p?z%mOM2WdTtxI8G!9M5?gc^V8)s=z@>
zk+23@;qeLr_P{xDKFlMJoCpy`Q-M8z$1AD5Qrfz(Suj^6;!!K!mf=lsOuH2_X|O12
zH09X%1A9lgHS@F^!trDCJl#f%=&lw+;*4KDCZK5uwkZ@sx@cx8Ii*9QnJKC7X8Cl1
z8pon&n~mFw>c7Hote+CFr>7+jM-jnCia)lE&h5g&bTqWGgPd3EnI)<tPmem`5Blac
zxMvQs!-$;GUKJkV(y|{Ov3=7{F;`>fDmT1Mmk8mo&GsRio^1KiNJBScCoUDDC{%~{
zRe3@cC5+)KYAe@dDKQ{J`FCdvk%jVh>u?KEJ>HdZnt#mF`%K9MgF2~vUnPNm<4@A{
zaiP|z^WS2yAiFuQLixl9?qk*nyp#a3chg`_c=O<3OF)h}Wt49G-hcaTXA2c>pf`d)
zQR7-btIG>Lg6r>(LoLUXYt4jyVw01do|`!_JFocOLi}fOj-47~M2k3>33<P+{QepC
zX@Ok0JecOqnuU5=NPq6EaS1LvND7lxZCQv48J^uuiqp<tYP|sh$@@&;Z!k_rG$i|m
z4{1hIg&UvlJo%K0I0Q%|zW)deo!65!`Om}t3}qh0aEvlIzZ?^_x|?BDl%yIJPp8o8
z_Pcpuf$b7^cD=oS{d4r@aP%z2>X_W!AO~%AbGXN<jAD-z|6qa<JX8@|wv^VK_5HYW
z*>~;VW$OLi+InAaBhk(lj#*JcDG%n<Bv4AG@YN03?(P&qaz6;x)eFso?^E4RA<NCp
z_m8bD={6h9))@++QY2M;i}20iQs9MZ$PSAjFYRix`QBe8?cO0i{a{DB{7jyjTPFU^
zK^31Y4tL(L5WY{|d18ID3J0Z1PQ#gkr<A6}lg&nvWdvcUU2+K=JALbgg;UbG_f782
z_36lH!SgR7&u5L5&sS90IHHqImd7HbUJxSBq?fLpNMsdllH5PMeIS(kj3s)}M*8Jj
zyhxnOdj@%kZFFLko-LmrD^i&VCAcTcPD;R7NYZ4{+Wv7VIXFZC-8YioGTqrkBqgGX
zUz@)oXuFAN>^<1wF%ZgT8)Z|F3Q~-dT6f%wO1yO^+%_B|Yr&w)^!C69rjfyQ=f&Em
z%ti!gsHHT|b>SON(csvTUD&U~jIoT|B4>$|f`|_I{sg)8_0_I6JZRHcZm#pJa{X5y
z&gXUT{^g*H-&tjx^17qwF&|Mh=61ItO|h$cyY%$JbTciSJ4QW}p3Aq4rzralg~KGA
zOkHwy-c*}fQ2l4rV?kfq#@St~no)50@v(2?w<tXg3di~Iec7sQucd+Lkx@Jtrv^qR
zCD&ND(!T8?XN&OJI_zu6V13jX5^{5m$(6_)2W3FCAeo!^UES_#t9H&?MueHcgAh;8
zjJRK8tLA_K-TP+YXO&P=@O@jw?b0B6$<0{QQ1+Ic6`gY0W~G1+7|3>`#(_SyjOh(o
z)kv5u5m6FmRiJof@LLrL(`F@7O?qc}*Kw^?nBStb^F+`q5REd;=DcJL^0p$Bis*p`
zJIH}6#-sMX#Hy#lHDWgVd7c$Rq1w=XQFcxr1pMf8pQ(xC`Q;i!T2eG|a4SRgrD6SG
z`PwgZu+@I=il5*6XIZ;OcZr1X-GVQ>zk+WsYbW1-QnqZF{|wrd4s2*=Vo8LIU#V`d
z*fNvlCW1s~b}Ojfk%sgQsC&id9tV=b7I4jV1_9xJ!g%nluLP44lz#^wZ-dlb{A0vE
ze4*W#zT{|c&p7zaS0B&M6>b3gPSy0iv^z!wU;(QvBZZdbCr6wMUccv3n>5g+YdhOJ
zQ)0)c2HIl-Lvw6Z1T4O!at$6F9Q2S=k#1(p%=GbQ@nQs>lhZ4z>p4dgEEv%ZdW-|?
zpycFaqlrs*clVy8K>iXxf{j2%JVi1b5L!{I@YK6DEmdl|)|NyADj&g>&6-YADW*We
zGpDZz|BoXT$8S<6qE}0+fupx%7^q#^!+Zk5#D{wtavP<0Vpi}Mgi3)<zboAErhpZb
zm&aP@|5DWfe<2`w2nWnd9^2d7yfy9`yq-C9is?y7Vwgb7DYKnEJ6%J#q3&c#zD$~)
zmzU>B%WkPF=_;@Ek{K_8gR<x0f&x?-#K8gtE2GhvIta2ZrSqcV$*=2GRa-FUQyHg;
znfb?mKp@8`3hH7F45~ZZs{Qbgs^jLmtRSMw;3u13#s6ywcB>*U*H+ef8I<RraO&Y6
z><@da<Z`D8^}rj(TEs_EBLAsMbpEAmLu=xecO$~MNtdQg7!-AeScq4@!YIEE>Dg*W
z{<%5>!cD1Y##^Ox7rZK6)rVZlCN$HP+}k@l_uC~@UW*(LS;He+T?5kA6(BX%oqMTl
zNjObIo0_Vp5U?T$2R3Td15w*_#Ttc)(Jbd#KB*jd3|9o1P5y#f27ZLC6Ma5>s8A_~
z($O%X!&cx>SSJ}h`nX+e5PZTORBWKq=Ne5-tYKh@PSOUMIAg}1TJvLt4J`??jTh$T
zo7E}`4j&wa*hW<2%KinX6wJ#=|8<Kzgh_`#XOoqk=`7Grrq7s&@Iz@QS4J$lV#S*T
z7OuD9g9Vp^Hz&q&*MeU60LRv?WTc@6>Efy&vRp&x6^mWQpoY@B)H=wpimNsVM=Gcc
zZu3XkthEQ<PrVz>j--XY7&@QvI2_i7OXiM%6V-2pmN42b-r9`t_THzuUM{#=&H9~k
zlO?xTjM*KzNBW(avf(V4ZC?AQL%99U^71nL&#%~Eg2|2ymzuNS{N-Zs(rE|9i|wk=
z>tl9m*Yw82h{8Z12)Vv}q%zl>nbnjWNR>k!eGU`xMy{Z$%}_Wg{^FzjyQY;NXfLq5
zEvjaW{9K<5d{)}-7Lh*iMw>e&s!ZzANhUf_YdQJz-(H!Q>W7G^96h9AG~NE~CuIf4
zbK^MZxwG$Rm`?r@1Zhl|xL3us@F!sr!fUl&2F=+IGcp*c_65-ls~L8}>LkU1Ua7mM
z;Xo5Q+TVR-Si(wcw%qUu0AL=fptb-T`X`{r^HMp@xpmi?i4a*wCF~lLnwV|8jqILh
z06`r30Dp^QS&v(-tipXKP25S08LkYX=t)id-ZHt!gU}(;n|g@*h$I0{Swfn2<U#G%
zzkhsgCakZ9ZZTEg3tQmFDM@U2h%zQ657kLYNx6O481EaV9PYfDV}l3oD|{$*%(%g&
zEOrJR453FmLD65O*y7%)Ms{R#AV3_$jvKPSchQLx#YTo@6ODw-Kuy6AcnZ-iMabx3
zP^%--)JgIaALV15sV-fpX@j{ZmHh3L=+CZ9NCo#;A^ebssZ&x#z4C!}<o^GNUf042
z=9~!=L9Mz8LT`k)k$t`i#J>D_C%ToE?wwU1ti1n>Pi8r=0N^AHde?GLlI$Z=2U-Mk
z#YzS5&$mQur3VHz^6kUr!e)2QlC4-+SlYitoIG*VppI2G1s54iaH%OmEw_$muOjme
zakE-A|1M`DS#E<r6+$5PzO><M!x7M89d&<vlGy$8x@s*gts1LHHCOZqcXSxe#ney+
zWJ*o@CwrNl67&S2x68-9_Sc~Si<SO$Z~yg4=HyAe^XRy!XA|74te?j`e3KbNfy4je
zx&vMza%jvZ*9};mWF{B5KNWUmR2fShc6}LIHes8$p}gqYw3*m^1|q)waboKdYivC+
zq`~tbaR^2?z-a$_CV*gaIi55XE$CVC^xk^Uzu@~R1#po!*Q?f%&!0bfpmDga2=c0W
zf{G4Sd8oBM!23V_E11I>vx3BI*tMO3A0w$ZyUpT=X^<2gAVHh2NKVp4NfkyCpGDph
zAtLVH$6ry85lJyBjQmA<&HuFv&CB#VMCBzGQR}-ss>4YHJ}+R2G42glB(nFxPhq=}
zs;=brv#5-#xB~6_`;!ySeZOi)wRL2P+mA<R%SkleX~H!%Vm{C^ItKQDT+bR&pE@rm
z8S;nfexwh-%EC*ulp={gA%^M>_|zGOyCKXtPfv{r(QH8ksFaR~W3$a@0BE|LX}x?)
zp^N~>AWt0iSst~U;6>aA3cq8CiK2<VEHg8^-yBj!-mn3Pf*rOS?<*^K@R>!K9BAAq
z{JSg+E+ZoayX*?ze^ii%#A~(e2I(5Jt`twsS<H0=_-5(sx2vcB>bi<kIWMd>Ll1;|
zGKb&b+nH<H8XJXvEQaj2`N|I_P}?D~X7-9z{<9_}_Z(S>K^)jg(|mG8J7i|9Yvm)5
zkoh-Bd)9N9S_vH@6{1xfpHNtg>;u&1BTR5lWY9toGZFI?QOGW3MLoB*nIR8`Kkbdt
z`&axi)Eq7!o~Cn}^H;dJZqK^i!vmrVXMHq!#7d@qozDhluEl8LsP3sr$3s`oPOYi$
zz3Vk7_+8T#r5A|F99L&70i2ekC_y4Uv3pd~(d5ec49ZE=O6`%~O^@n4-}!(bUGQu@
zcUeaZMrPcBdRzjv8NUd$q-PqY4bo4l!`&jm&^dpTN)h~o6k&<@J5jLb;dS}jSuW=r
z_kBs^0oL7ENM3qZCiK`~XZEpoTRp+zi_a(iZ`L;_DR1O;@yfk7C!N>EbYmVtrDBB!
zqAlgyoFGLUSP`N&URaf!Y$d2Q4(D?18(j0CeOs%V?+g53B74Onpmk{!_&@~^OV(^l
zE%T-7i6zy(r%r!HuWYK~ONaZU3L>UgGze{P!iEM#9A8vQcr9EMeya;=^*=bAtn>OW
z{uFPyEUa#}TUSv|>6e77UvFBxI4wB}haXtqAbl_JO`TUO<ZO0Y<_m!Tl`ET;3MpLr
z;bJR;r+%<-Vdrd0V%F8~W^W;=as*s3vN1Q;y7<CU0v*K6@n1z)Y;nP6$+*-0<#`2P
z-T61^S48?M^?W<`$U;qBCV!0F8PbM0W{#(@$-%c~Pb&D#)$+K;0)(yj=&ux(LQ{zO
z^X0M}LoFr#7bV9xd}a*?%qqM44vES+&lI7J;!^->nN>D|nv9kM9wxQ8urP)*FdX?p
zgex0f$7Xp`uNfNBAV5*|;&)=OE6V_WtAf8<5VEtqjm=6?wcn&s1jfCuZw6Q67MUoY
zCoAbQ;KBQYAu<nL|E*6g#%J7zrJ%1a>r;5!FP+pS0xdI@;W+~D`~Rud*Q>FpD5t5a
z4JpS%$rUlGLrfgSspFj9sV~6x(Zu#&qku(Cw4B<sNKDoZvDZ)^l<X&kJZ7b~M+$DY
zn=}4bbI|~-bR}D`vfdqN=!7Wv0UjZd?9Yl2*3|z>8p`$MBM?Tw`jVyT0)1a~<oFN1
zYoQasQqREi7F}^?2v$|DekZkvzIR2_w^6{&Hv%57sG`s4^Ape2PUg3s{R<@Omep8x
zruqpwuJ941eI@CB{H=TOq8izPUc?bjV*!3u3lKaQ?i$V>gQ_QzOu;zWq`Ed#<kjP6
zRm7CjoI@kUb6D6?v{Y9!QQsxYum`L}QPlE(;3L0yG5rdt#;_jtEvDt{I_4S+YrcBv
z;Eq}8P6mXXIQld^=KA7o?dj#BWTm10LxQN@XCqqYbEK_+qmG!NDa#|)h)X#Dxvk$Z
zfQe}#EHYxSXEqIEL!#J5oRe}aRN}s2(y2U?J@_tUW!5a^&k(tVG!@`1mZr?8JDi0p
z7NL)NImUKs9wQ%p5lkkFIWcOrrvdN30VRFDodDrM`s33ite8mvT_W^j>A=zXkDV4%
z<cGfE!Lq(%(E>CoaV#Lf#?VwE6yuhLYsV@qqFQv*QilhTI&3d%8^u4;nO9;7mHaeB
zBt(iijbM{spGoO;@_)?+OS!9syTCz>$fhQNrY8Qr7)|b<nXhfAOI<5pfT;?zb!?G?
zbsn)~w}|9P?cJ&-fix5EQ+LzIg)_`jTfazw^r%>XVW8pYgE(%e9(Q?HLo7Rp8vNdl
zv{o1lB6&<j>_PgHL~iX#sM=3hjDu7i%H$o{9n>`E3tfi{+0U(&EiEk@E%2tjh&{%$
znF()vkK-lPm$RP8d8+W0rq7Y?uctOEtEfMnEh8JX9;h>}-v3nJHFNck>TS=;;>5ya
zQYQ1L9NbUFhagqPSZ3ivB@ca!09q-7gUdxJ&vP_NeOcZc9+U{YhJrXZmpx!Z!xamr
z1Bp@si3bjUm%C!lBO-%>xZz)Z1xeRI_Ff+jE@tq&p`HOeotvCUKvHpwN)gy<pZxJm
zB~=q{Kx69ZFykk>)ftCz7qNeduXij>4b|N-yIW)b(Qs8)A051Vs$v~_wO(Si38bGA
z4hGrXp8fc$<o-si#kP_S8%9qKzuHrs8?^}E>SOTCycwx!rx5;U4ZrTcxjvDwnwAZ@
z+#AV_znk(hz0{V@x;e~KCTEY0bu+ZR9b5n93!o#tIy*Y}%&7Wm;d*gVU-4GSf@Ds*
zP*63ul&rM5iJ36Y!`sUuazn3k0iDFN98oqAdbu{({LR(Zw|2p+;KAMdr`pzG=h%vh
z46eM*2c(S)TX#6TUc>+D3gx0^*7mOA?!My1b-4&LYdXcO{I2XgPKt@Sg+%+GI7H-)
zl&+5XIEUYTQ=o;5?q2hte)qZa%XU1BN!u!7_XKcflW|g`I7_hU>IBlq&LcyIdWP?T
zBamPQ_uHJ<y~{ulCrlV!92IoNd_MXJJF541%hzk$V@5?^23PabqAfxVwwx};efsP1
z^x{nJ5|-;^X!U<Ioo77N@Bhb<ag;bDj!{|JJ7lknjEL;L*CBh$N2tU(L{>Q{LdGFG
zd*#^27dfJwaO`!mH~-J?#{a$>9v<houIu%Fy`N8}@XKDY=(~{SfD0?r+b1p#E^g}x
z2SbRs7P&jZNHvcmUZNW&=PS`=16g4YOOaqw4<lc3n_Gd3A-bCFkNf1^Nn@xT?~riK
zMCgCl<na<AO$vvGTKQ-}Fcx}9)MP+N$o{*JgFt@)|FWwoJ$<|*qb4Wc-LteN>JYhu
z>u2%~_{gocYvd4di#K54YgI_2TI7uuDdE@Y`A+}*c_r8h7aFz2Bfcgxyp;H1G-x@e
zwBVmaiMHhRoo#Ic<on&871iNr9d)75Z<6^V824aX7@Fj>RhjO(uuyHQvb=Iz_}Ow7
zaGBSL^$3y_6$W2C^}LYY!QF+wr#SbgV^<Ha7CRyj>Zi{1;0aJe7PtR4Lh_ZTFOJ`*
z7#gNTv~7D!-soPP`zQN$e6(%nHI)v9!$L^tk?-GjZEDBZ+u=I&Wl7Z!dRsN}(z5I#
zm4@8dBQZ^5Rsu_?UPwhP?4?98eIM)=z+Er>k)g6EP1i9#S9{`w233+S=)-@jaEHe6
zk5_17fvDa9g*(cUC|(>>ZCoZO<gTRlxO|3HC&+QIX(FT1jy3DHY2!o*VsyolBfdh{
zmKkemW$MYeGClmYK=gr{YG(M%V-im6^uI+YcM25Q4H17!@KF4qVWhz_8LX587}Tey
zgYfpxf1H`fjnFi=KaW8vZo<85C*|+vq3yo^h!A?sleF<0;ZJN0=7@#SZ<6%-Tj`l*
zGzBI@yJ1K8YsI+muBLS9q93p&)D8(ofn(TfU+n6nBg&7MJhl`^6&k%Y1=Oe3DWA{b
zgsR#($a_^&=&1VowBr}7aE<_@nd!w2x&_nht;`gAIs5l?#I4q|Znh2JyWZQHz55iZ
zMKPc2_C^x?w-A%MS9$x54nqp2r;tFX8*@p|%|ku+@ixs6^fFyu3X`NJA1!{Liw-W|
z(5jWoeWKhR92gl`II#wv5h9|ZD<>#FN|=Fx;S3rVq}Wf@SDtHD3M54-PEY?iUVhkq
z=?)yunzd+p^3><Jen@cgnLoQC8k&jOUCx*T0&nBfZ#zZ#dlDw6{vPLFO<$dk4~eNn
zUd$Y|AGIZA<$-eq0qGA|@FU_W&8q%iuU*TmZc(*SIdHX;$NhD3?X^11>o|U&cwvpZ
z?wU=JK|xy|gjSkA0WlJ>5)O>TJ3ZQT)SPY4%_06lK@26iisslJ?>sT#LwXYKua)Iy
z<pYpwDb^!6+%j&}+arBudPbIo7S||-%Dedr%8WA`@@AzPR+!z8dPL8`;m;_|_5|^5
z?tJy1xX#C-pIet)bPPr1^7)4ouS<n|>JzvN+tLylnBqB$0DbZ=!$T7j4~RzP(;raC
zchWd7s@HxtUY<&_8x_@-c6m?06MTJl&`fBpKRh_%-`v<miuqKQicmz+>S1$@Fg2ee
zLtTqc{L?fU@mu3zs^pv+rHxn;hm|h56zVW{MOh-kzcYb@QNg#Z5o_5AA;I1(26flz
zPZLQvi4p_=&z<tP|3l4JT)l)eX@N`WmxXHXY^XjdJ7>Ra{^*`~Lgz@`YYrMs*p0jW
zWk32kzSIlKjbXe<I?0p7(`#f5==G@!Xe)l+r_{GJG0Qk);+FhC6MFA2JS1l;neJLN
zap1|ypQu6k3sZIeTjx<Z)u`KlfsHEg#(FF~#FWVdWJSD4+}fyW|Ic%G%>f9j+c0@6
z-mUg?T_lmCG)PJ8r-!79ylgHFr9X^wCB^+dfy^?VKDeJWj5y-XQ1h-Ap>4IHD!y4K
zq2=}B;AF$SHg&RjFj1>qI#-{~vGpgB!3imn`!l=qw@S1^{zzm}d963=>(;5~s7)Ob
z;?_<maJ5L+aVhlIE9BW)@ehw!YKm`oK@m_|$AU;myH45#i!2WoXge}u%y^}n)^GWK
zq?YIHP)$XKahX9Hgp;9f-dpCGyc^=dU|KfT?LpC(v}SB@)%3q)*{wR#;qSGX9GhR9
z3U1CVPxzabyK&#iL-HGU>e{NkjxW<IT9+S62O3AT;wZm-G#DJnes&$tQPMfdXl%;d
z-@=t+=~GuqUtqbDGy`Sl_hIK{`dZP}$CYK~ko9B!uy@6?rOj{cqoxM@*`hRoj8v%T
zW^9bny6~dqT|R#(xv6PxN6`m`PJ{#J*YW%6{jq&Cfp_IZu!nE{o^6LJ<gViQ9{$e7
zj05UZf`j4;%IyE4Zzp?nbU=R$_yrZ1s6klh_>9X#8l`hT*k7{Q^Uk$Zr@xz@(*Xa<
z>!$hx4t`Mw<Mjl>d+)sZ-&^5%2Q9X?8K~N=Es+iE*xQPUHT{wgxYrN-GxLLH&;uS}
zj_yO26OYrh$|pQzKq<>Tm1uo|*<Ul@{!t$`yl5D_&xF)ED+yN!-^;t%`Efdo7y1D)
zA)FJvmE`*;{NJ3n;@O~ihnbMv&=4_d1qAtBp1>bgO25x$;8`#$97`m8uUJ&M)V@02
z%D>#aI?lb?YP>4ET1;u*<I57YTMLp0PThf@-Worex_0orTlLL|*e}5+yASM-fo^6>
zK0Zhj@uD~X=(KHOqH8Jl1~{ON8}L}^?mY>@fk^((R-SNu#@F&w=2dnqNDR0_y;G!@
z6(VY*sC1|PW3F}zF%S!dm7^yU84#%Z2y#v;RaeXB-WB)Q*;}vnLZ?DSfqz|8*pqLQ
zP}GiWwmt@}b?Y6x-nxX38w323i^w%@Zfm;;^Bv`u6mlt_rKNa;$F9`MBI;^;b|4B-
z&s_~z#;n4xk<u9?j$wO+bP7qaBYXs{nXe{}by`>`2Tl&-VXHT|*1Zq_g%k$mJ-MgS
z>{~$_yINbi`Zw~b#}0T&)jA0Rtdo4dJ@wcSp!*`+N4QnE`_3N@2T1xfW9WN+#XZEb
z_$50nlA^RB?CQ910_uKxINe;AW#n_XIzL#bbj`kgjZs_;KWXY(UuR~$;WeW9XJelk
zQhfJQ=6KBKba~g}o?GIt{1fSU`8BchE^?vqvRDcAcz-PxWI;9+se3Gx0is5IfO|=$
z&+qoK_VK??ZhJjX0C59G2NPI+RJyFjQxV7iT%Ib1BwVOGAVNlXWyw#m<0|ZGr|VPw
zw{F<wAGlb9lhv?6iMw{0=uklrZMS#`ZA#z7JlKb1M~+gO>CoWizuvjKf-5Aswcg51
zYG<{6x#fF-=?L$?d~>z-d#dWMg{A$(3<xdQgA{5#$TcVd`p(XRjSf#Q3i#!7@l)-G
z((%v`hPbN4b3Wi=>O?W{?ps++ytykizO<~{fGsIcQSKOQ-4>gCG+F=`0I!qK>%s^L
zbK;5Uy_k2}7G>iWar|jkI`@%ASg&|h@`_45eP|a?VytkEMR|EMjP?y~noBGr0Fn(I
zj+l5JS_61vWXZ^8mdS%2JId!-TvHpbf#~@xhqwQNGf)3ubpNGm4Y$vcWSfo(IS(={
zmo%WD*U*(_Mb2!ILuE;L$-9mh7a370-JdddZDe^g9MIP~Arp0D4*LAAZP&Ckv$bHP
z4v+~Ru_4U(YfUaOuWpN~-^c<e^mBUNvt<d)T_DQTbfwu5D@W(V!FVf;`4v^XIAtu7
zJN^e@;zZ?C>~g4#wLxKLU~X<iEZX1M`PsMQW$i{pR#tvfykc^Xi!aeLZ*-{a@x*I0
zCTOKAz1HT#Xo7FPub`Lfa?g97w<*af=z8#B+G3cZiE{Y2=iN|U<DLEkra(-WS^4*U
z00>jeclT!q<fKiewc5h@`85R2#~NR6ttzT0Yv#X>@UNQcf!$+QdNPr1{Lxj;In}KA
z@7iPA{B4z+)B$$f@J?3W^vU1{NO{lmo_o=NqpMH$?GZB~_{)BPN#{Q~yXXyRB))xW
z>^gnE>Ib|C)_i)t<$HzQ$dA0}Dx?A(H2<DAPBedOD;=B;ll{6|@q6vo@bZK}&FTK?
zKR;jMMJOmC>^J~dviJR@K3EXD?Z0Oa@F0<sGyJ-5wPhf(&hZRCx9UL4%=_>>qVhBw
z4W;@&uzMLL$<zOal#ZXwcQmIY{+&3@U>Qx1gH+z1^W*pYI_(D=9cIf9d~Hol0Z5lr
zjz1@qvOM>w{Zf<pz7$SrR>i{7_SQ>}#vHsCnsJ5nGYx&LI1?h@SC>iCDAToM+RoPY
z9_nUBy+5~-0$a#g+gO-K`aXuk9Zm6V#v|wUQkb0$uv2f=(s34MKBSm~LftepZ){M>
z(YWflcZCoi^KLLV$2D&?93}h%I^$BhzoAB5jWke_ulc!gR<YtS6uzzX>5>>0^<NPV
zCSj`S{^}H{t0%g-_4r6#IY-2~y1C&fvIm)ow}^kY_qAfvg|;mNUH6+-mWnPNsxK~U
zZocFlwSZoXf4Z4#YBm*5azEznBZ$Nx6q8=fPa{wSai&v+9$*DZN!|hJ<5uqIo88|g
zy)|{p(myb}`hxk^ZS~<jE+m^L>eq}w(O$TW?2L5HA3WY-wH}Rf`f3jxoIV6j^n@;o
zh5OzQluAji1zDv397nO|r)S71Dl01~Dak31@IEl`gvFa^6zb<b5fU5^9HZ9?2@jut
zc!_g=w|cc`2pkXB!vTono6ZPg8yNiRvLZ2fLSO9!>2m9shki*Qd<z7TU~3jUWZS}5
z#{A5<?(T>)sb#MOB0f)gkGMVPdV92Z?6FU1{CjfxFj@(UAj9d#)iTn@cioB=ceEi_
zqk*tlzcE_H`t(6Q!y=tgwN^SrMY4F2{^mNXfii>c1cpMuRgK!iGQ8Z&lFhkJMps9@
z^x*MY+Ovv5eI1tg>IA33<N2}qr7bkyk&4`GWOe{CSWY2_-ks0g@P(=AWb0U>CL;%p
zA`MGnPwlfsr6QESql*%2UIW{RIO9oF-jZBVT^Z7!V}X%A5kfJ{PgmcYEg|~-Ack5S
zb_<ctJVOe&aTW_hWV!1qc$_QB$^s?tKRvtx;$=HsO(RppyYb<=Un}5mpmbJva^Dw6
zXDX*%bmAaN4_iKE7hW{xaZBI8kgCP--ruloIT~m{^vc9pu<c?vek-KNM2%6<^lH88
zg|n+GwX++t52AFr?~Ju5pJcxY4p=A05g0;L9)PYC-e{)nA>mvkm0el+)TPBrrS=bu
zCu)SlIsT`x#PMSKJ>sZ8$LFP_LOB0t-hfhQ_w}{3L@Gi^pEmx!jq-Ho<?9>aCnKTB
zb6m&`HHd5UWEZDD-(T*_VHWN4|MOVAq)ai#36UGDm;fpyx)PohtdCnCbRTS7t<wQl
zerAlh&xz*1k~wg>4fYBfTI&?=AX!{>((m5ccC8)jnkX5Onzi_Qe4%oD0g7QtJe$9N
z=Pz1B^rn*dL0&?m<a34h&mf&F3ZpeO{1vrSd?oM=xZcoEHdnVFhlO(f11+W9*g&7*
z@O!4Z0`aQZ`bn`*tE-!u8qZ*nzX5xTI;^zy6CWwwZ2y;v@_%0=!B*K4eK%rjd&{mZ
zqSw;er51UB91BB;yQME#g7c3r6$Du;OUpb1J`3`=`EM2LBO@cBz^M$T^VMqIjFq5T
zDU=lRLxH(>XU$KsYBGaOjLPxP^P=7;N!e)?wmcDE?`RoxPC*4ki1v^UK$^@xcZS1n
z+;Bu?DzA$^?+-V_(mm=zOH@0gGl;--#U07OG?BWHk$#ch{f}kRHx+vxb%-KaC%L0O
z6rA|ONu3kl24lFt^}~x$?lnJ+&0(aR2@Lx5A=tI%0t9QhVgXDWlff8H$8;;@T@See
z6}fXPHRrI^4Oov@=U?D1sVq^BSyvZPZ8qp->0vwGEMxX#0K4QK`@NT<5fe}9=4fMn
zJ0Qd^9n#_@!{EsN<`{Eu<F|EEy<~#Xx9MCu75}o~8c$zGZtsvXtK|UZN!g+D>DI4;
zBJm!)H+2O#B(sNC!e9B5#GnF=U|cq&-$!{|bOl!0uKBj^K^y@DQX;#csHn(2uL))7
z_P;W!+r)WS92Gb=uVWN))%??TXfqmFZld{a%IfyCi+nC-q=D*>L8RI*vjfmF2nsA6
z(_|QkA~C(>kU}m7WZf(1c2Gq8J2%!Mou2UDM3FC^3>4~+byUxfXM+X&3-MNPFq5?`
zyb;Uj&Td%_z)e3iQTVAMaG2~e2fV^-vRIxXPeAF;O>Kdi$P{9)qhdb$S(F9xXr<1z
z;=_}mW=>xBvmrg}p>jQ=me0BH_|(uJebn5s9D4M5NZuO3G;xIbHkz_LwKf=?s2j?h
zx%_JLC^Fq2POmNn;a(vfBY|8?q)_XBq|CO@s=yEE@S5f_Gu$WVlww7a#01`b{dre8
zv0(4c#ZH*MMNDch7H1`A35O>nrKk3Prc()HKCkHbPvH)`sMrz67bT5(8g;N<ltPRm
zCy943@L=adP<W2Nt1n2;&%1kY3f7tP>T#@vu-lmI<_eoRE{0%*T;ZX?6Jba<epixE
z5CAmD7?^9;&IZKP85x&-&B!zj-&w$JPh1pD-d=fOIl!aC;)vm=<Inu;D)j$lg152F
zm9L=Svr?BV-Xcw^p|Me1%`o(cSybuSx}-iEVOkg=OJ>8$#)YWUd^{C=F*jDIJAW`T
z(slXF)TMU%=X#l#!ae*|P~h1D7Qjxn<VVkV{6IO_E;stlmhCP4y%+(mw%=dmU+%wf
z9-G0{PygfD3L>gBs|YP8-B3ac4$jeCt<w1_9-TXcMO>`c8&s8a)mx!so~G<KeUTpB
zGbjlR`UavG#(AC>4`|<eQTJ<u*1h)PqC~9qKV$*ULow`d8Xh<Q>)AGFn@i^|iFd$R
zA>=TOUl`R|Q6Y&nsGC$9Vm843>P`lHbX(~K!)tmHepghWv2HvF1yO_Q^G(#0wO<}o
zsYLJpTU<PTwK00<?%jtocl4=4@+2}{r<-J#=D@*g&xE(2P*Za?z5n6Tlzj*2LTXD<
zX%1SG<P=5}tOdm)l6$@aKgXRJHO5}u$Wr2Jt}GJHeSl}9TCCOC$G40HeLvbCXfxdI
zxZ1YAaKHL@Px)wY$cl~rwaFLsWyI_B3GT#J<J+35P=Blg4MG9~1NW~n0uN?57mLy$
zmbz-5Zz07Al7sYNaFH-Wh_YBCH7D50)^C*zwk1s_MI9x)nvVixZzG^FT%*Xi@X~92
z!s3h9bze#>pbyRB8soo#aPA;l)B~(s?Ar&Focb_zt9&s5P7*Ck^c{j@>7U&hw@`nH
zQAsrt$2T!To?T||Yu|HxK}#i%Gje*~q<%y@hpgP)On>yD9A&(nO2TNO+lRC>nw}mg
zg&L)O!OQvPN*Dm)CW7|5?k^!sCS^>D#>M;j^pl(O@B0+avKquTp|PZ$3flC+4gy8K
z?I+8|q4PaC4?)ZlGm{wzbo*HRo7oJE996>3l#0C5;`1Cd9<x(|xcfW?E*<^uZx!w~
zJcz(Wmfj#|xE|=$DGGk=WIVFhoY^bou1^DzUS@Xh1gti?B7ECIgKfPE>D6Pe7p?DY
z@eJ}yAOf`@>c5_WyOhsm!25hBL@vJVku_ZJ_fXg5R&rU%Y<+#ny-UR$7v?HF?A|>@
zN=PosH86)qr)4(CPE;gR{>BSoa@y}A8)Z*pzGWJV%gD2EY*63QLF6ERSlcpYR9uIv
z-8L<p222Ov=tRn5BW9T}!`6M84mbT$J!~UV_`A|QpPuZm>NJa@4rQ;qSz;h5glwo2
zI(KxX^P2Mw&Ugtea(d`8ecGUG%6CXxdiooyRAATs3giaQhjNRyZ5R(C#&pdQ{wx2w
zo&H8Xl)ba<T?u~{+A`Y+oKE?(d-xD!IT+i3({-Vl$3&>}X6ttH1kdQQtEFUDt?;#t
zasB{Yrt4c43Ist`8e>V!y#}wS$5=*Is&<4D@q>ecVb&S0jO6MGWfh}MwIwAb<>lp?
zMlpd&NlCTSo1m2)R){=!a^KZ5-5eD9t=YG`LZC5}w+CUI*;G7ey7A?}@*I`30U#xW
zID3*|e`z(ouAvqo&&q2Q7SdYf6$h;;%Uc4mEY@4JTIV>lp%<ayVZ+e$pD@9jWN!*&
z@hZRWq!@hp>0UU7Szh1SF{+wEuL~L!Xd8x9*|sa*MOY>IIM*E|N`x4=up+fw<VR{Y
zQJ;Tp2gz?;g3BMNM>#q0l)GD^fN+&KB2lsG3+G*ggg-k-Z!ihYV-B_zX;6rFc)h7W
zY^`kzT-$m728?AZH-vLE9aWWFPFknE-A<xp*|ED$uHhdXer^P1GDU!l`nHj+8(hg)
zTL}&5t60IJFJ>5!3M(UKkN-XE${3iq>yPto<bS2ms1kj6y0=%bxt|4SX`f+E#J$2`
z=d*v$N5YLXjd;jxn80q$Nyf$l-F=7_VLnVg!6_<ee{rnPF#p9)bN!&jV`~x9rCQ4Z
zm*R{ihw8~#E(9qF>F4Z#4N~WL2rbLR%$HbZQ)-BK2+m>{;yMg-R(gt&WyX-YRt!JT
z73dz0Na?yyZ|IcDTMCSc>MoS^S4am76iHeC&f4vonD%f%)D%MqF;C6R?_>HckxEK^
zA{KCoFvQyjOeCD-^%eA85JwI18)PJRCGrlL?+^&#Z(|Em>-39Em3?2UYy5AL#n;z2
z+UAFzg<JX}MNySVcbGFx{gi3u6R)jzj}6}x1^}Px(M9qKfI@lMjO8721iik1nrgX?
zVa<%JZs|812YF0e@L_GhU4f<j!#+@);=?=g<AM1UYdSS7FoQa5cldw}y@YQBj(+P6
zQ%N)h9`)!kN4%j^KA2<)#IC$lBT*lI66o*m6A~iJa#v95sngc}Dh$O3ss41d3-724
zoSzm0_BLJv0~|q*Gqb;yjgy$L*iV)H-inCP9pHIVrn8*Rgkpua)=sREJA=kpP|$8R
zD9?wa_-6$#N|#_laYOuIWF);X?(|5hzO|<mLY|@E3V5~0i*jq4qw=xp3JeBTC60_R
zDqlet;mvJ+2*0!BOgb^aQ9Dluece#sP!~@tRs$F@x|7B_Tc}`O$Vc}c8_17q*gI7&
zo{aTa3N;rR-uM#^5>M3FUH;EEKWAur$jGl&_`=KUs*QvgPzqFp0iMm^*7W!QMCy3^
znD_0y8Os%F*-j>jt!jbpo;P&(v&;NvUl(Lqa&wJeyOt)kdWDkTuuPn>`H`}%GGk#~
zOanBSPrV~0lTA3D{`Yl~7{ZjP$dJzAdxt|D%i?bSaH@~sZ<#qXq*tgTQzd9J<p|R<
za^aDRxTvkIH7L18(|LL#RhZ7YQ6T$vLBnh*NcPus(^El;m^9g3eM&WG7;(Kh1Z$*4
zdw6J#cC+nSp26&ljipX^fSM*3$rFXEc|hg3<8tnbXDLXM^}xD%a+oth7Nz6dV)TP0
z^03_g#nnZqd4p){DTcZ4{Rh;jz8oKdd4dIeF2NtImKYwIi`m%NSs@Bmd+1bJzXg_l
z9pryo(j!M{_>Il*b*_QCe>JmaYS~cNqIcn}va(_Vb^ApZ704otj0(VQ*A<9=0p^CH
zo+IhihgT@L(z$|;E+65a3v!EXdWRROv<166M?U!BV#T<zv0<5MwAnVX7cMunHy~Au
z51tv3DnL|+I2U#N7qK#Xl$DuD|8jmh?8S?T$Q&)j$K6)zJp_VX6|)~bl~X{(Ix1&$
zsW*w4mX6NQfL6?=6{R=35z;0d4%}Y}cSKp+OlK=9ZBoz|S`4W3{4t)m+2RlTbNcTh
z$vpqV@NoAsmY_|i?(i@VIIusz>^Q}B?1g5}o0*c{$IZBv6G#D>w|0K>=o@5|5+9x_
zfp*azQa!FB;Tu23F_f6?Iit@TAAY7m4MB1UX$)z1wv&2Uvy@h#dg~Dlup^tO6jt=j
z?D|V>SpIJ)CFv1KOtmzV<FKX&y60El+yB`4W1HVW6P@zb<oo9298)1s%6o|)Z@HpE
zn}stz(Zuz^EH_$fAuvkS{%3=&!7G)9SQd@1#=06trRmz8n&}Ov<IWpKspdm)PL5=j
z+>{gyT^?`StMRWwqq7#%z>F&6Xo<7sNL!rx%=|GU=|@CTO+$TX-Y}xNIiPk&DrcnP
z$p2%*M_a8_)3UF+8=uT7$DtZGw=_jEs5aWJrpRZ2y*B|e?6k^F7h^(I9ZZ<iwEH)>
zNJ1++>#;K9%R*JH0V{;xU<7s+X$V{y;s=+Ts3TGSOEF<_t<uJTCZ|e<T+CAZ%iyi&
z_BsRN-9Kkr&XOCJ7lZasbL_wG{CSYzN2uT!CUaxFA8-D?^w3Q3ReBKijnO+EBaw1(
zMjZv}ZduQV6IyW+xV1Wrr2*Jy6d&CyE3-nISu!G}(Cag>N8)Zc1*Ig8{9L7PQdTz9
z`l<P-J6xu{>gia(l22kcR>Jk0aK`9^_@OrK6m^V{X$GUE2t`k>L5>3lc~<Z%h$EN0
zHj;jL#>fhLeId@q5N<m4dC}1Y-5dXaKE^qn940|fTtXGdi14|{T+UvdrKITDc<2iA
zriZYi!!3DqG(LPvkN51cf^^4F$G+41coTCCTUe8+axw<A+-uvbbrU+slS%ttA*idt
zSWLG~v~R(oMkP0lT4sUs)ivh!Ig<XvSGies(I=O23dXv36GB?RImD!DLNmVrZ#qNK
zu#2R?P|cW!hEZKd$7oyv!|rD7sB-u&C}Ew@%M!i$<`Gto(?#+Fx!PRoPh2hvC1sT>
zoMZb3zYf&17LM`fXXWHr_wi}xMq9&Y*A9ws#)H$sCD|kH;k7JrQ+i3WYdlwTade>1
z^R9B#W7vzO8}`vxxsAxj0ftHbW|>Q`kjR*K;2>fn<nQ@;PMq-BN(Z>wvZiOR)5GB*
zBjKuO5HWk{A$f7Ut+cg+**!hQ@CKYBv+T|<0sH8x-^4y6r@H5LAIF<0-7GUn3n>5d
zld;EJ^pTroG3Qo8PV&z;`G$Tg*m@9&MEhTOp7<wx&Pq6>8<oWF>cE7oPo=2HEktS{
zc7koxPKlKU{E~x8z21q53I0U>TlnfV5WFD6pvLHjgFo!LMw!ypr*>R#f4Yvx<7WYd
zeEj#F^SMPS9VZcF^*T0Jp{ng%H`8@1<m~*}ikmL}EF<Ig*5*jgL;J9urP_HvN7qu>
zjJ9a89Q`MDP6!C+#F`ZH8XQ?s1bLJxBWmA%4nDVv4uVrdekUcF=Cz%HcaM-!OK!O9
zV7G-7`xCpcgNzUA`A-U{sIXx)-+@$2JO>e6B^_`hd>CYXqZszb+GeDL??c`(7x&{^
zx)QxJtj_7uWHAtWR$EVBdb0misrP814(ea2QlQFxd-B9@RCHKQLkKBNjD(CJwErYP
z>Q`S^tdCQ?#4Q{1`=or|vYI!g!j-ENxS))6Trd{*49R|Ob*Su{P6ha#ZUAQ?P!xE{
z{+(N6wc=)Mzihgc{9c)^wQW~Z?cq+`o1vkN-qzyudpET*=Pt2szn+gGr}kP!>q1{P
z7_gv4M)v}YV<=yW<LYa4N8K$!squf=duAm_`Wh=ikvkkWJ*0%{xGBZCKIz}*=Zt?k
z3USTX`C%8aR%G}*diRowSO+wths^V;2fxLDR5PQM#kc&rc3?lYkM8d9%ATx2iDi3h
z-JkV+*zFj8wYy-&yO{3{RZOe@;V!0ktO8NMKli!OV3NZ5cz<q<@Fvpt;^f{@L>qr%
zp_K0X?=5xz`7|=974qw{3J`3VHQB-D(CQBuNgqrC``Z5z8+X!23De~wf0FpU0bzGU
zv$4T?tsU@rSeN8eLceh)=0LtJqVhhEKm67Dz}zu4_uRKfOpxwDLB8M+nMysMFS*1m
z3Phx-XGvCzNU6{mEjp9&7TSwb!)_>xNKL^7YbIlep?W=ZxAyP(pYp4%j$`b^Tgr_M
zt1N-*6!(CJPpk8BhVB#ew-y$9^>Cfr?+Jt+0f%w~gW|vVi*4e0Ua(h$OdxjWouONy
zUB_P2D1etUXQZH@A;CvtPjJfywo}HqxGu|<$veo=hQ?4WcfKLL7Y1W<i~q?r*3{I1
zkJp6-C)+HOA4`hx28BKE_V$Kufsh&L98dPX79cW(`7}w>l9VL0y(}Z$c#;xSM9iqg
zAj>|Egm9$w_Vy0A1Q79UV7PX#V-IiMe*O^u;;5t;dp2vp_?k@dATw61CcI?wmS^GL
zEf79p4{~SFf9?JDYW~{yi?l|bO_WuN+BNxKG_mcSuv?pEOmB!OH#T+v->)<jtofUO
z7mnI)k0gi0Ki~U%;^G5WNG>pApECXMMW=kC+0S!DTjg|T9>0ynVz=?pX*@$h<?jc6
z$zRQDH%2BU%|7S*@`K;j^md^XbaSCMDctB!`-@0_e~H8btI{18Np@N}XSCUkms7f2
z@vRd`HxP8A9o+IE@dk(sivGDF^3E*~fS)boDl~G_dlpf5uUztJ8d<=9Zykmy>QB_M
zye3(_$nR)*F+V>0CS{yIv)*R&D9AG0Fx?n@UiC%^#`6yrlnEw3hLY=~;X8o83Cl@S
zRZ*t{KKEU@kv-x+RLgW-!>e1{_gskITTWSSI$tvo`0PLu$e>*+lg6=8*BjFY`0FqX
zC06>_d{>C7K$#o7A??eC=t8>ia*Ho_o{GnI$-5W*#O9~4D{1vx#=TPs<?zSJR|%8w
z!2$*0o^_<eR#SYFci2zMFy757vXHqk@1XJICeanadyKsj6mR83b?;EW1RFAzgswQV
zL4`LEQj*1aR9<mCPlEAax-kZjceVNBv0^%DZ<RVjzlbc$-`rMeKH~QYnfT-xi}`c>
z^YaYXGYc10xkOKkxZI!5ph_jXfAScEx6_wYp%%IxGx~)b&~$n&(pEBa)1vC=kGJ33
zuxg`2)g8z)p;%aB@~6>$r}bCncBm(qmXssq)48^2Kth)czIC$2=KK$U$rVXxRdwv_
z9$oFeQOL;(&ei;s_ovAe?k-$_5GrEhG_QdW9>*g?3YFMzKlV$Bf>IB%s*@vjzP{wY
zZ?uR`ZT|Mmk_EkuZ@v$Ehy$67W*F9I@N#MORg}v07^rJH13VWR>;IbmW}J930i-56
zLkeY+Nk6M5*i~f}mjWKrNuGs*SXOq+A09W23Rl)ke|~rb)k-Cx7`T$U1#{ge!b5Od
z!OYCRe_d0feOYT@370MOXN1Ewpil2Y2V1*m9Enou@kvkpUXrZ1BH3>%Hqf?mCs1?W
z<fP%&ps<K%U+Q#v+HmLd)w#+gU4z}JO>(u#`o7C2mpc@sJ0w61J#Wmn9lU(3B3o{<
z+-sHNHgK)IL`ATiU0hc6MsDUOee$l0v!nV3wA-1U{b^$GpzO1$Ln_}FSEmQ=)?15I
zz84u#r{EWw&CL@;TQJtL`F<0Z`*4+B{r$jCm@`@$gLSj4oi@=G6&4<Ldw|%%;XsZP
zNG<WKx0+bv0U)uSnVv;SPVEgfC>|Z#B~+k*D>UKY0Bkil%LRd>elUXdoRL*l{u~1~
z79uXTT_kCG!X)#GJ+0X$wn8G?@I7Zokr$WwS94TT`7c7kB8)w6zua-q>~{}usH<CV
zrPJWp+hnu9&~+UiRXP0!GFs3%`FJ2AzzH!s{RfD?JXxtIMV3)~PX9I@L9c}@ZV4oc
za@9?HgVShB@b(WhcGL;W@{x>aU4zOi%C85>({_PtUM@<Y)$P-yWv8*fr$OJ|Kv=-e
zLhq|rugbD-#cQ%#!`YA+Z<!ll8V3SzlEA10U3T55Th=gVV0|!gOfcVF=g-$Nb%89v
z)e`DQGR%Klz4dtf#zU26Y1qv@D@Q-E)_?3;&%>)DBjjbxwK7~w>+Ni8>I@Y8kZ!D+
zUGMqm{vI!gg*&6lc;8FB3{(}XOfrT@<UDDbt2fCQk53F~kshomy6iq9UPL*5_J!)q
z=*4R92|t*Tyc^2wB$>~3Wof#PkCD~_6gGnyF~J$5ZgKZp;^@UOas*5SZ?}}som_s5
zjjf)OJ~Z$(b#G@AiN4&DL;aB!ygc{`^$ym79Lq^$*UEe5jg)yDshR+FuYWMZl0-t0
z-ftBlSASqKB*!$&!a@-hcAoHEk?o$5{jIOx9>fL^J6~QSQKvur;jw|*F}tM>T!=xA
zSH4P%<78Jl30k$bKdpY0N>X+X6`%}?ALoRdF@=jXu$G(W4Gj%ASb52NHb%l)$1BT{
z23qCWMr$lBO*KD~KT+d<m!LhT=xA5Y`ZB~y!5uAWh3k)#X-K0K=SDbcNCY6rIG{$K
zL^!jfc<CRd{|TF)CKUGcwWK@;@T~_YV(rh-nh*&LR6U+Pg%dJ2IH{L%cRTOqYtO<!
zzc9hH*N+RdT2%6q!<6{&MLgB`Muqai+Umx~x_kvD)*ElDa`&+^aauATT;J1^cx=I4
z1#WcnP!n)MVja6|SBD)L%)M*V9FRx)I(aL#(^5L!T@K<o`T|9+rgAGlMZKn-1cr>0
z-x;DRAcq?5j4V3X-kG17ndu>_JZAI%Vbt*?hrYSJzEhFU5mDaBI?$7l^uxqBQ=04<
zb74ZYI24iEEV|dK&nBS`p(iK7U(<?z!|C88%<m&WuaoAbS>b3xK6omQ2=<Y^%^ErU
zJ<tKk6aNon%R~?rxnuz^dEFP%wY{s0m!m@t6p+cH>l!?PI@hpdU0{o$QCCyPS^neI
z1aLZ<_ov0yP1|ZjyqF2f@u@WhU9~Y2#~jN%xw@MntsnFDoAACNa83wOctHrP0*)`y
zb<^3SE5WuMml(E=&P3Thf!-LbKO!<R5?sCngF-~LTkZ4)&>&9f9>kc?{n05;;68{a
zG%z&WQ~W(ML+PqLQ9Dwf!INB90VVTEKS~Bwd!QT!>-FO3G7dx_&0dsXvNI-o%^qND
zS=9718zxt+eX*&^psU-*$0sO+K|k)Ir|_Z56(&yMo~B(*3}+%WX9kXN@@dwa(x#$P
zgw&O`wgb2v7!}53NOqi91EacI>0H0o$~YVm{P4Q74F9Xq{OFa_19;E;zs*JQ_aBX$
zo2%QZn;984EAmEb4H|1(Yc%9WlWdkW8>c#=FHYOETVnooIV|KBaXn7sXX2rM@<s_G
z`<%CMN7nmCuo1el(t!89I1T87^xGL+Lf*)!5eBfJ=&*3ld5DG7z1{>fT~i_Mg!z6V
zydfsqwsSI-oVT5OcS|hD+EijIXvT<<#TmQuFI^p4Jk!+EQ<%gohQC~#2pN}G3`<nK
zriQbo+>9dU+WUB8x$TxZNl47se_cyMS?U-oq)gZOp6_MAuT_=}0ms2?tDJWfB(gV8
zK=J2l;X97SI*HUzMvRRgb!?xk%a?HR{Cas5^GyO9=Z@g^CUqos!8WoFJ3Av|S-Kdx
zd0k0kGAd$jt+3zkm-nt$>wktwAHyinK_r&Z)mmJt2>&>x9ZbKsjN+!SFBvtmk~+H?
zhSg4?z!~);%a9XBurezz+sJRQYh-Uo0qlS^(=DQsR<`jvJ$j=#lht`af~1>P(Y+$`
zHf8!KubHWxRpR;dzaa5E<ts^iQ&1NSoUG^G<elMkUh-CMNccH^e*_y)GNAbUzQB!N
za>jN8Whh$oAhnvSB}*adp(Eq%;%w#`V>4^Ae!pav-h_7?kip#t8)Mws)@}7!K*zsj
zI{axDI!EQNm9_n}O(7{xiy9-&b$6;#@R3HPUHf^oe;;QOOL<xIZz3x}e(P1nq&pSW
z*&C`iqxdk<sxoN$kHW+5kA-TP=EepRq{7GISqXf1*(_ZXjI{(X8tb~)pZ1q?qkk`_
z?L|aIlwLG9)(3(g(2Mm+c#cpG04Z#H^9Jd$*4=vpZBR1lc`bHkRg3#Am;Bw7lG=(f
zFF_G>t=3S{(yO?Oi>heCkgFx>_v!t3`tnC^y1eg}X=9ev<561E!jI-$Z66X~nkE|R
zW*HOm3}n|1Nz(lizt&SeTW`z2Jcd5Isa>+P?$z=^YF$TS=C7^2eF;9C2<&n(dve8#
z<KaRpftoGxlItW`qnnuD*-oln{}$(o7dym}@JH6d;(me_Lc5SKs5}*JM%Ezt@Mv$%
zvgEIKaKeOAtPVOo7`$q=ehZUle`fH~fogeYB2G%DT1ec<D0`cDm+Q!&G;yl3<8+o7
z?d};qzgv_v*X7BU5Mz8D!0VZY`zqWRxhTA*9@a92BsfUrh>D8RYWIlGZMjt-LDZ4D
zOoI$%&(1^-p`t?6zBye>g0F7ui6JFlHm8P<X$FJ?^1}({z@{AS*EqIq8@>FobjZ_A
zn)R)WF<h>S4|eOujMYl3KZL@QL1v=M0kl}xH|t5HC~CbXBs>FhN1Ljbg1)pOt)FO1
z^LohViwajZ$lu7vw}zh#GR?nx;%>x{c@iZFPL=SHS6f+jZOPRw;~o2BZ}JmtlmnM^
zxw0#Qe?3MW8L{fGfV(NgNulYCqPOV!E&q7}M*71;-_tLANt!hl?PcXZ(P@<bt$rp&
zlBk`sozj!-^REG}pQzMsf)bTiW`>5p*9BxOUB(y(^ov8MTK>`Z`84iBH!rWfKT}#j
z$+E5d7Q&ql@8d?y*P9K3BCJl<ytx=#>S}j#fMdUv!Gp&3vyg8XYiCrxffgL@L+F+i
z%RwK`kL~o{FjqSM`Z;nVS2KKPptkLXg2II6n733(Z4@AGp`tQ#+XkcP)Ia3<L59FE
zST~j}mV`Zhb@=Is2-Y}X5ZjK5hn!x(Bin5nZr90g5eNi>l4quqPRiDddUuMlLMo&F
z%@gq;S|;+S1ON9gaB(n|GlH9+CykTOV0|m=j}5XEjjOj=3%9fR)f7SOB6uBHy88{3
zn&{lb0w_hm4JF`x-?}v2Q%^(sNQ`pz)k2I+a(zO)Eu)Q8@Xt!G3H_Qm0a<fV<}?rn
zpXw^?*Y)m(>OQoyWv1DtnlVNZ%HfAOzy4AHnmRW;!%iX>(A^mW3xIUV-H4}wkvka6
zSR2{3EG_!U;U`DhU7m^=+De!xPr=Gszc|POxA2jrAmMO9MArQ+bl&>%!(Kg3;MVxH
zfG$XD;7c+tr`ivqS%IM832to)i-ItbI^AckznFb<^{?yA)qkF$Y5Y5RD8AdJZY8-c
z=#hh9)l4lT9&XO~U@gtWRyrI>YVGG5OpC)EIVcUqS1C_QpX)VU3cFYaddk-@68e8Y
zugN}py!-dSEzCwqp#m5f&A+*KuCrnvjQcH_R@tBqfi_z(b%0Z5_o1}=Bhwz=am0J~
z@|lJ#KMu?XS()6h&rmO|-O2Ws6I<r;F=<%-@!ld-L81R!s;hDs<?u6TqJLLKHOV&{
z(Y6De6^WwUJ;~DT*|v+Nd&+0C+8wrlPGX(eNfbLDbpUVOg1Wc;_v-d`nH)NMVC$f6
z`qWEmdS<3Mns7Sp9c#;8H}!i9Q|sOWrl`Lars#9gNddnS^sNd}S^F>kzZHPU2`<Pe
zP0W4?Q8Q1x=A_Z(=t@(<!Tq^zpOb`wS(sC20B-WS&n#r3t_i=-n+7~v6kMwku;s=)
zsY$y3zFGju6kI-*{c$jiS#TqY<o<^Q2s;5UC+0e{)@+v#ytvYt&(>JJ<KGRwq)HxZ
z$9VXl>sAjLIG=r?HJW>j$#x$NsTJ1CK|U0<TaM*^L|BhZ$$J+(*_vjoEkE=3cjW%6
z-BmY+sC+5@?rO5?5)NSARk$PAoG<U!@ai%M++S%eOBDWXDe4Z7J1D3iR6JSKns~*i
zIJ(E<teVH2BgVA#e3Vr|__Z~-<VSVpnMvEXV#&(pUX+fg2CFR+$&Vdn?!d80Tq({<
z_QM1Vp&|;uMoG4J&1Hi0(lM`nffWU?HrLxlpaSE)(^q;~z~2*m3-)KRQFG1LjetOO
ziY~JzX90q`DcFBw(<R$0{OkyfG(sp|)yOD@?|R7RrfZH3v!b<@<LFdLOoUh3Lbyi*
zKRKWez&U#JEFN^lh-MABc0sLj!CWRDEYK|u^-k|NS~5Y;F%BXlto(HxY2*3qcYP}9
z2BTO#S%B80v>;2fhrVt5$6Y;T!WYXCX*{NOGS+Z(MYV-~TX`^7$wbprFUb$fu9L)I
zeN+3kfTeup<2lSoj;j$E%S$Cgt7o^xMOjC14lf9cn{^dseTMn=P0_U#Wm`$sxgi^S
zV#+PSduFw%sg0ElG?pd5*+xo49Z*utL%l*`K*#>#^j_h`%9I^oUG${ViH0*QON#Ym
zn&msRqTGjCKWr~H=B})yfq!cNEIIgRCf7U-R842=jt(v1y|ejBY#=AMi|}}2>TOgE
z>Y%t{8Ch|=;u=pUM|_O$wf%$}Zcj8{QjlOQ{Afuzpl4?4nrVCdiDZxIDlP6B(sdY3
z08GxRA;r~@VM!v(O0pR;8Uy>m7`|dY9R>N*G8cD?Oo@~5N3QP+4Gp{8KmI0~$9S+o
z<06=q^lEOE49Y>U@#FhW#0&&ADgKEREz*7eW!l@T(ij~YB5FN{s#dfQgIl5X^=$O9
zZGr3qGc(u7p~~PH`gT&d$^eUT?TrFB;CccKsxJKCt~z|fCQOa`83|12PW9@@!@r8Z
zBBP_CBhC)ZRW>ihf2w)8K04k!vY)+NpzD%lk&ZWL<s<Qs8b1kGzD5^QOgkbRP-!9D
z*UzXe+FBz2rXZnCk&UJ3V3835dD-M(q$4Uz{|aiB>X_z#N86vL@XI7|#;hEe_L3Q@
zaiVD}u{)a`aAA*Ht)eTQC>7$PYmbV78RX%Ug1TLHS6a@x;^x+=RF12myJMGo2VJ<U
zf4-L^c>Lb;s_=lQ^W2A(NVh#zTcj3-by^BXA8K}HL0;CMJ<fa6Z5`z)Q1=f<PWR*c
z&lz%*QGxZt<~Z}4<~ds!MGIg+vx2K8nG^1(jDKOblH?_`lVr&%w5A?UQW<1G&dg&@
zL;hKd6oWJVV1E(XaeVJ;>fWdj58V)}^6>$1ywP!4-d1vSEuu3<wV#E9u8W&1T$NbB
z<0-`&Y^nHS_g7P&y6)mvu7h!*MtszYf?|wxP_)dS$jfgsy$2iSD#yTFW8?GLR(}H1
z3hJF1&(Z%XP5md{mfUH^X@UL$0mhj_mk#3jVl!JI5JcD_ix;|9n46r)ufuTvcx_-z
z5GMKdy=%KgeayJjP5c|^mt?BAu7S08ftBQE%h;`)hrTEjDyUgRK>C<Owd>WvrIrz+
ztDRDK#DBypO?Lc?J;1aBrW6%c_1jwEmx6bQyrgI9e>yl5Pi})}Ls_G=+x5+}NU(Pp
zeKiGK5~ELh?kNH_zwOvIw^V`~5}n@iztUgcb7k+-)w{EiBi074i$6^2T{xJsS5Z=$
zot_@EVi*gOKfs5!*nSbuHDpe59rqBt54!d+ASNesXo)Tw_!l?t@$-43wV>!O>Vf~#
z`chPNyaPQ27XU7&_9p#)YnJLU6@1?7o`11!-XU54VPu`hZDYf_XI{d&UEm<9{cD--
zjiQC`_gRexL;t>*g6*BG<o|6A-(&we>4mO&(t6alQaE!=s5ScNnk8ymHk8=vin!P;
zQMmx`f^QNUfl7OL@Qkj2`yCF+&+!HtKFM@+ou+@XGGhYW%8mV^qoY~x`(R*v|7nm`
zplP%(orIz7$EQPuaXl7-<f`v<7jR}RlO|gs9c67LPj3(P9{YtHqKrO&c`E3r|A4m$
z@g>bM<NmJCs)vCK3)fSqx*sL(r~37Koo5{J;>96YdO#KOG67NFb&KLUKHR#OtjT)j
z{8v`VF!M*nU$L#4iwWc;hzA8EB*y&M>`<BUx8v98<KHo=>Y};sXwr|ZbC8JMm$aOe
zCY`KL)g_a}P>4HTV<pNG9&;vh3Alcg0qx^)NVF{c5Qj7JDk9VjN>ll66l5zryG7wF
z$dzWy0XSnmnx<b5Xo>_S?}{IX;TzR5S_uh(99*grupp$s!>OtNol`kiyCP>Tbjn9K
zu9rU*br)vaRwf1(yK%jCjl@=WUB*{S>IbJAxps6cPq2n<p$b`Ix%nyS;~P2G?!0Xt
zrqI5lX#)Y>SW`nML6VB-p(U<+qgl?*HwtEerZk(|W$fObnSs;>l=t*$8(cnDW<Gc}
ztZr$Rv^3q-wRr9w5v?qDPx-G3>IvV(+M*u-$mIT7`aYWwv9*7`|8IK>yqcV3WIea(
z2;V|C{&k9_A74I=@s?FkesQ=6hUb^t6MW#IH|e3sYEWD_fJpkq@Hs(AHC;NKaY_xK
zG~s_=HyzObMfb5s%14*$%^xD0R0zQL(AU!xlQsl@n`KTtg~eBU6@oT@*3=6>vm(FD
zTAAE_Oo1_XHNdi3zuYUY<e(yP7x?AiW&PSyG<vV`YQIb6{BZPY6O_&@Pn4lFtNU5Y
zruA#I2H%mp>#+N)7HOpMBDIR^i@!+%E3_n>?~b~PJa?Dbsv15b4N87{hDKlQXwTZM
z;0s$2g#633{D$8??$^$18=l|xT~H^%q~!n2f0BhtLZhExr^BS|!fPw4s<JXz^5`Zs
zrZ6VK!4VNmd~aQ3Ev(-g-_2(~qI^+mQah%~k^D7X<8`=CzVeIWbT2ogR;G_PK%1OQ
zOTz!QcBzKZnEusFbFZW}=8!H$bx?|7%Vd<V=W@o%F5RC=rTvN@-T-cAh}kw>hsY=H
z)<pPSje+zQVE^Js0f}*vXWgG^?6@rJXqRSBNt5CvLUsJ;;rHXc>YMFf1iJ4nEThZK
zV{6$qz`m0*VuF%6jqH^wCL7?<NpdQc9!)Bn-3E^5wG|Y8zfFsa_JunW-Ww_#j3e}_
z`{CqPqD*OlTA%kt#YP^6-ld`3u=2m}_upEOHCn9$SShcf=YL)Y4#kqW?`SHkmQR4U
ztm<+3Ckyok=GVJe++(IH?R?yjqbe7h8B_N6p70u>$Vr&H3$+bpR{*xSdSn;4qM2|_
zpzHq1;)^fo3=8-+sx3RZI+REj>PGVL&w7VX@&CnwORhnuaUma>l1~0bg`KRM7BkA`
z=zzzrat`tfgJ))2c4FQQ3PxEjO5(8KXMef7%7!4J_B7m7?ir=^_2v7{{i+ZF*I^RH
zNyJN44Tx#RBM6>_0}%RBqwI%)IpZT2VNw6db8+#Sw4?N>QF=UFZb=0iPLl!2?FI|5
z>Qos?Osf8?&Zc(=e|O}OR+)wP_BB#x&$MvgxCcUnhowah#ya9SNLRg?jV?yRlCjGu
zUCPAtbaI-(PuUumaf72i6GsKFc+gNmBd2Sn!>H-%^HHZ3pj{oDq1RUD`p?LCQE#KO
z`6K61JVu^`*9D<b88Xz(F(p6npGB9{k(L03gPG_^c~Zx_3o;T6j1sJIvD<?etK-e%
zOY<Y;%YTIi<>jmrTn#l}w}am%e1=Q+o_lZobMURU*pJK|kUGL%@1hXDt%<f9Ep)IU
zA?4D=W|$-c6?B|ol9+T?>bWOI90j<=)(2R5uA(17AM^xdnT=N!#blXZs<Uh6DKcvb
z#i*tQ5(lFU4yJ>i1@YGE;F4eMo-86xcDr&%2g?wJ{4n1qPX%2&3pA#|a(<wr1T99G
z@q{lWFSG?|QEkB89O3x@*L|#Io0ZDgDZbiY`U3{I!w|*KwEc)<m+#|x$6O?YS7(ss
z^wVA2m&-WZP*uC}R6TmVu&t+4!YJ0}Hp<Da+#w6+`ApE|Wky4uh7nH_BMnNdRfK#2
zv>09ajvkK2sME*n4K`T`kaEZp*qijF#mOP>QUeVnNn;j8C<;^|F{%)VhG{(`S#F?=
z3D9x1_+Z6B##Y`Q<n^9eTU#@V$In$*KbbQBp-XHFzFeCqv7E7$tM&Eq$-Q?^p!A3?
znYIaP3>GN09xZE1e?Z8s(D=avou%&H*NO5!>mJ+r9I9+=#IW(a{DWBFoi)P51GaPA
zvRl>E16NI0rEQCRj>z-B?Jo!z^iEg-7a)@1C7m-G5~}mfj47CPY;}`Ao}<tk?(Xh`
z?!c{G68@e~j$R(8sGM*8?byRA?Da0ZPf-rRUYz%-cn*35g=PQXVy*DG1o>s)lK{&{
z06pbA<4KySyaMrm!Y>DiDqFr4!Ogi#QDt8|>aA4H=jnWbKq7VvtTbL=*9L&K`V9qB
zQ_h949mkT|osrx~aO7fbZ@*_cJ3L+{L+rR}ufD{`y)m!+YlC&)Xu~?SLlZ^~M?oz3
zfA^%uWxt~+vJEiUCeSbp{w#ia`XsldabkbA)k0=`f7QeIXQXlIP*#<&+q2jhM(-^v
zG;UVmFN6~xxdL`dM(=F`o13x#j(C)klJe36F8(xaw6Dj4>(|G=W?`Z0b~<DI<6zxn
z>9FlAtICdEUEFY1_{n<@i5ziNl5oN)a^U!QEzIPnGMkM@@EOyiV*l@FJPegGnnuWv
zv)f6er_E*P`f4)-g~0#Obk=W8{{I&rB_T?_DUnuET3T{S3WAI-snOkagLEmK1C$n+
z2ndLjNY@aMAt0j}JsRon{rTa$zhKvPz3O?+c^tM07vyx#ld*Ph%%5=whjn;w$VRBE
z3LY2mRU7Xi+n2@UPf~X=0qo+@Y*MenXET_4`B<l7e-Ka;6@Tz?iL^6=>XQ9{q<ULU
zXL>Y}vydv+eHGZ}N?uNq67Kmf(ZomPaDs)TZmOB$N_hB&!x5T-nRICU0DV1tN@-`;
zhy9GUNg{M@Ye~FQlzE(9Smjv<6!tvCfMFYu!an}J$~%?%JBa{QE~YvF+j<2Rk$Rsj
z@bQN^<Ws$3mvp~$ww~HUY3B%bB2~5Yb1ymbByP=B<9}HO(TKZnym&}_?d%&NYpPVR
zY$=*LS5M)3Mn{#qJ2EnIOyTkhtYxk9Ui$f=v)wmB*=Pi%3J-4eWPU!r!u{Ug?=u_l
zZKd73FQ9pHLhozdOC76_N@r&zQPPoJmoBwuUk;^i*3&jgWyumQ`=l~;j&Hv_^FV6b
zBq<^0l1o=v7;f%55+Ie7PS|b+MrAkuqMt_p+6FHP>VPx9?6VBAry5m<Z8=BZchp!$
z$X|KT{V>6~Sg^z0ysTpq9aFIz0tCQpM+kBRp1WS#9unE`Vc8`>K4iDdbY<MLc5=rr
zNR_z4&b4T7ckHg1r9p{$OI`c<On>M3aR~*-OorY~#0uCfsWZ8$&yfh_h?bX|RV_6T
z+xC&A`qm9^Uij8=(r~sgDQazN_C(qSA;$Yd*Vo;By1{H?qJ@e1kz;o-i!xwj98MTZ
zg2@rp;8e8CCnF^ZS5~x*kSm6egfAATvSIPLbz$qQw+GF>uruOzzr3qudVu!%Z17<r
zp3TCGJT~f7w223|v>|?5wN#gnI7<yOxj#KQxpBsaF0TH?>#DuY^U}}5;+iJOISAuz
zze5h4Obzr!xkP_|MY8Sn2C(Ym>go{^WUw}zd~)yIs}cG|UX79&pO}b_%Tv?VF~1gC
zKD1zz``$t}!e4fGbac;0T)9iE;Nk}dvi+#Q{t^zlr|FBrLazHpgU?tn?V*<!nV|4=
z;T7dpm!tLs>5sQ`G$y}UdzOb>#k9-1cbxb&OyLP;4SV6vj~csxbt_lO;Yyw?_Nf0E
z3PM(<40Yb=$X{Mfo+Kv+Q?U34w}Uc`HFzWks~0z;gS({qjLaXfbKNa1`yRMEG(HY4
z`4l(I-#>q^*7u&pxCPsk%n!U3lS!_abgY>)+O(SYk#Lrnw7l9O1`g<flk}pi<(rZU
zyWEBEo{n-%GJwrR`_1|JmvJX2UR#b50qvDaBRX8j*{J&-Cb{JJc*o`FXYa5qeMQN*
zb=?oeIaAO(dB1PcxOyJ_zx?f>Z5q@?y5J!zb2}(Ap$!_V6Vv0vA!-_`NQ07T4cEn1
zxLBFK9n#y+>v&{L6}76KS6>=WZbAO5b=Cv*DoqG@na79s^t>w9m*8V_)419@R5psS
z*edDi$I9f<`f_T77`8p0W011oDRb^oR-NK1<4RX>IOIo7Bf^sdVjC`}&V+D=DnaD+
zJu_gFWh4;<gz%~sF_27{H_*sQtfnXG5v6r!QBQ-6g#c-!f$GKFfu`sY{#bk_`t>uN
z(m6m`U$cZDvQp1@^<mbSU0Oxo^}h-#`m5uAa)0vJ)gFqmXZ&hB!Wa`<i{}0rFP&uV
zaU@`T@cu>Q2_D!Rne!X^CO6xJw=0m%dVqfI?JFs(NjZ4h-Tw%}8()y1q~LMB=66^P
z(Bijinxwl3nje9S)ZyG01!tYNhPO-8mJ8>=^3dpm?AisW`?p78`3;kv!6xQ7-{DOk
zfvL?Ihw3@Td#Wahd`0DThf{Kt0nRzxc;p3GjCxrTl4zPlO>y}A_~1GgcfEhP`X<Du
zD*R(w+?69+j8~?fB7{GIyGASV733`^M~uuwSEg?Bnal0*VWwf*;MlvfVK0zu3(EqO
zg<0{jsor!VQt$_(`bO9L`IqbgZx`F$zvfZwEo3IZo*%fl*G@aif@z~7xrJ$s%yTdA
zqn|fjsruh*wr5SYnLqx)bnO2-uwjp?rf+ckoFO0J2l%EVKBgh!ig`ZYU3V2?8{s=T
zD_Xcd-di~dyIlS-KT^+=P?|Bk=G6g=P@wu7HaRpV{`T^wr^cxDKF@t%<6D#Diw_f^
z<`8XkY~sYuS5dr+vx5=gXcIH^+U?ULOh&5Jwqnm%9Mtgw+!9CXZvSQ9f>5{ZTMK?<
zRMW!j0z1THd#GMdV^hklkts<;wGi=4PDUfGbl#=c6HIs&J`0xnVwjPVPz)p|``+o2
zi6p5*L#&bx`pjCsULT(elI{JS;TRKgacg|=66<14r_L;7-vCw@OR$^-?-ujCqqv_|
z^Tr7T&1qMpBAEBQ1t;vg(d7b1T<?@Uu-y%A%}2wQSL0xkU?K5^|K;IYW7mbP%qPyZ
zwgYw&WIxz->%@k)vX7c`2j)A1o2M3xIz2WIFyrX;th9Q;LntW<md1m$fp6FIK&~`}
z=B1=$;+IESJY*0ER@QrWBK^3ZW~Q2HFkUVW20{jo5S)M8`~ttAJ8%DOc2~60Hwn3(
z<(KpEO5wHwjY<2YK@~XLZcy4;f+?0;HL`0K3}gL^WWB8vEu#R!gc;0cF=00@{?6WD
zv_dL5l0x#c<uiFHCtKcWuNEUj(Y=Mx+e^%Ou_()znJugRm6?pYN+`Oc=*g;@Z6npk
z-?H5?Z9eC-U<XIob)TITPx^>JdD2d?K+TZ}>6pIbaIzb*vee)CmCqVnMPrYXPwo^e
z1}Mp0ScSr}jI#hDAKP|XmXNX)iIrl%rvcM05-4tRB@i-i7|5Dn%|u9(#6Y0JLP^D?
zQoKk>5P;>nizfw7s#Cn5a!y%uPY+6x<~iDuM?sHJZAb1w6RiF8gv!^f>9`#`5SxA~
zyzC4|UDBp^aJ)YhF2jjeo~Rr{)qY5OqoNn3`8brLrT{L*E!v1ED}+c4i<F)#Nd&tI
zoI#iWwM$8M@3^(&`tYob`_>Q*$G+9Pi|8A^!leH2nwn7`8Ghv8gqHJ^lRYXgP8DV1
z>;%FDTJ%}AXROu9OSn$)mZg;01am0l3GIiN@B=Tc->MqyFST=Xprpw>H!m6`F=5*j
zMIkp!0B|#EQp-IjBKG+qx2f}-+}{15-c_(zSZ><bmN3RhK)EameElcC@P6dOsKN1r
zQp%Jlf{0#DwXZD6N6u0pCxuuwmQWz=Cw}u<hRA1_*>sF|U7Cy-NWo@@?#tVgF(yhi
zbTxcejQ(hy7T<k+<cAJh*|n^K#wfyICeD)psP>iH88euQ#XPSYNo2INeDFRNYJq=@
zmPDPQ7*)Rf{KzcGjFn3X@ss+kp-j?`J*%oGgwns|=J>Ha4My*-qzK)g7&~XC_32|S
zDe-M;yvnJ`sj}BVI}Cz6I;BaFtGjA>W#)J(EGda5Rc|(UjM1<4C6%P|u1%LzdJkY`
zM?fPp_36x_>Q7*&1_9R>Malu{fF1~;F74ee(jN?BOxSME8~5AY+jOUg+Ya7G%bN}x
zJW>fUC0n;woi{jesVUqF-O_liBabk(9Cc}2c;SA*;aW~l7&5>p9v250`_}ipvrE1N
z7P&dr<DddRZ<zw$yqAPANws|~$$&loh`lat`{}GRMUkAm^s|<4p`WI9gtG5;fdjNr
zY=eFElQMyXVjePfXv2(IhP@n%WoKZbh`_LD{Da9&AGyNMCqWwV-x0Bkj99GAQONDg
z3ikcN6-I|(yv9hBtb3E8l8Y?z<it-_)9G2!7(cGMDhONPFMqXqWVeQv1cWd_o9N9h
z3;y^}u!~6a8<+9(>IREWSMz|A#%vw!zOf>l=O;loSF(3q-Y8PW#N`37<2)K2G_`&x
z&!vt}8?>ukJ8vtq^dl-u86wTnR=b@L*XBO}EyR+vO>Y+#6z)C?{>ysI|4R#cUSD}A
zXI(0#Aj>3%`dB}C^aeDR)5&a}p<CxMIN<CO^oZxycX+&UmXUs6x^X(UGXBb#4Ij4!
zz1x@LKXON*NE;n|4QgbeD`lp-d3~b>mEPho+D|_MARv~oCA_=jdNldO+J{T4-Du>2
zbD@as4pHDdAub%Bj4p?mB(Znka+gbE*vps*jq@QqF7GED$^dA{kd&b${}Bi=&?Cgy
zNz^=FcC{=bo+3mb9QYh5dQ97?B9M+e(=90?YW9C$@;L8JjPlpMd%vu=lYWLahF3eu
zzrl#~4FnnnI=f^z2uA4*5k=lH_}WX^mi4_!8#Lu{OoZFk96&TYQQ-ulBpL^ch;_9t
zstDW4#zm6kuK>Jtex`U<;o%c(AWYUve~!(z)Ys&9pRB2XJv=?oJ=7aJ)EL<wYQv0A
zPaa`r4g;&7V?-=WiIsbdZB!(O`2G|tdks)hMrV8`!%$(BNluiTTwqyYviM40%kM)$
zSitT|<n>W{2?cQEj{{?YiKR@re5LV8`lhGA?8OQ&uW&0yYnJK5zRglpL=x@?n8RNd
z*#5BnK>3yv7wO*>%wd}(y&bpVqZc0ckPj6_B>b2bO$>Jyvl$=>_v9)C^ThiIPq+Yl
zgn_2bP=mJ&Pferjk>v>1x&7u^)pEU8^j*VH9*=frLR5+;GeN3es$PA*c{!KaLYVBk
zMoOT#Y06n>aK^=i_O9WyU7_jM@*0DKr;gM!q7+(vp5FxyEdb>B0OC_jw{D)j#vtWe
zZ#=dG3}-zhz9@u?u+$y)R}qglo|0sc^X&-<UK=Nk6QRB9vs&^N2p;9H)>lUm2sA?E
z1ynwq(rL<(O4$R&r~whQ`>MCkexw{%(*>k)Eg{mIjcFa(aSN@QP;jLr?^gHeJJuj-
zYmSnv9_5&TkR}s|)U#)uLAx5Xsv!En$Q;dYb_arLp)H<sZI*rFtX11w*yCxu{zFRG
zf^i%FoH9uyHAR<Jq_3(9bPR6#&IQ&9H=!m1>WpfVvJF><|8qd2#_Yu&1;{1>D&bvL
zlQ~rb*@-hX^BPq{>PWK>FDmBu{Ds^;hn}8S%=^JFC!T+TGBsD%3k*Ajwm;2(x&7Tm
zoT=-3NlD43OOxk8edCFb45)Z8f1GlY$ajXBKRG(L`r;Sxb&Sx9O1*4vp>r%J;7E3%
zKwEomx9K-2foM6?`+oiqOJ04jcu}`Bp~?V(Jdrtb17>!R$dgOu`aut7RhcOb5*$M1
z1vE^Qk-MM~K33>z`FEqKp4i`wdvq`Pfn=XBUh$yqZ<ntw#-!{Tww}y}zM(P-Ng4JK
zh1<dRswGSFt^Ory!8n=SzVQRyl2rXZWZJ9Ys?Xa&divYW?M9mS)v+hQ^xm_ejIXAV
zVTO>?OJ@wj-M3U#JQgEI4{F<WpMV9{MjrbpbmpCKMbH?4MAm#tO=16F>zD;K(dO>=
z#DYLK{7l#`0i#X96JM^zEHmLQb<hp)%?cXf?bfaWWdkQYF#cW!EITEYmmmb@>H<;j
zLmWX-IehUagM%QKt}}3L4X33u{bF!(b0a6ta5#Y(926ylC^o%)YOgg)Nbq6Jnu<I1
zEGuet$CRFvU5fs%Ohu_JdAsDQ0NMRQgDoU@=nRw|4+zuIQl{ip3f7<UUJSM=bIV~x
z-FlM^h;u3IMG8owz~IQ@&yA4ZYGM(^GXwaZAW)Z1K1vaIcXzj)zslrM@+<>ZE9?1U
z?kA*G!kb-0s8ZD*_vtG2!`(0MNQnm;jxQB%mKLso+s&nJV#CjvKOG$eqyL!k32mPj
z)-(e9!$sHEmNykMQnzbDVQwHBZP&hf-q=)=>7@o!_y~>1rzC$p4d;h&2XiCAteEO*
zS^8&`cx+Svr&VQj6`3uS2q7y;R7xC)BEoz1Z^T*G-PJU2qz{)h+qU_QByf$3A0y<K
ziND%wbvMybR^`*kIwHe^9}y86t*;GQkL5YFn0>d(AKh8q`AhUd4Jlxw))lx~twhvK
zc~z`Y@~smyuAA}O7d8#Yaz;elHgI>KWz<)l7~TPKIz{rkr$N^@knj+aYVJ+MUzZV<
z|2GpDGYZwF^*oYdqh=;mB6}R*qHJYi1tIHsB+6pYWWaB`YF5($k3NCKvBlws!{aCi
zzB8#&>Xx1dh_`=f^gSOsEvW6NZ^G|_s8B1_=(BaU*OiqL($^d<eE%KTxixM1w*~O<
z@%h%#F^=;~>;+q7o93RdM>S@NWZCm-KZdfaPgLj>`I{gyCXzyYb$QvE#+hr@b+2??
zk{p|7K^r5;Z|ofxeQBL(hkj!|4Gs^nSgbCsCrTQt3Rigq>T)GV4M!`H$tQ^znp6t*
z9#8U!Vje-w$&F*M7Haf?dni5Na)YjJwm+KbXT8C)UT^fHt7fg3H~T4AB^!)9V&eK3
zaOMp&v~TmRZ#H8)LpRq9nT#g`Kw%bYl5H13qQ{lQi+V%_5=t^jYO0HHIE%+VSu)6x
zOUho|)~Q163dB*SNq}5|*5h=6mjh_`_M5F4um@$k)*^6ss&t0`C6Fvnp0^nyk}aAy
zx~^QMN>OBMoW-Tbe~Ot~2)uGRO9w9cB^QET^Y6;Z%0F;q&(s^1E23}lM!ix?tDMrH
z8Qq40;744@)uqlfv9oVMf2!ZR|0KT7C4ZftUzi`}Z^dj=N!zS@(n+iQ&t!kob(3D9
z1C~OFAeC!p3f|cRN5!+>Jx(T08`s+OlT<aF?$o`T;6kP4Y!Kl+6V=yn{)`YwO+ep?
z9JH<RNfnp5Mm@rs>k6joLQG-9^)cGb`CN5+%ptDr-e}I2#Q|At4HbejQc(@6tbvUS
zq*bQkRn&w+@lx7qqr$n+FxTwFBb#^gjYx}(c>4-wifzD){XdS2HY1O&(0vQyGJO)R
zMq$*SSv+0lPe#@C65tWTO_IcGh479bDX(esAFWZsA84gy4#47**lIzO#l7XD77H3v
zlSICf(v{dL{zf+5jXX)NA2Dx_%Sqh!_<HFhZ`m}r%Qaz_>P^W02<#veg9!H_$hliv
zFB<|oW`G@Oa4`qs+9)+PE7vi>^cD=8cMLE}&;>nYm-HQInq`OIc>%8Ta@<}1;$O+w
z9eN2SM-`GKC!xt0cJFg^vZWd(C3)CTi1qj6d@zNbuGyXYE9X3{2|Vhvq2;s#zl*1J
zowV)|etzU`b`wv0KM8puP0ffW4KJmJqJlNK0E0I_2Oz0bDJL&P_mW7_Vo?dwuVshk
z8@DC%9j|r{Y|7H}C`$&yzp|wJZ9PLaGqEeqNxkR)oN$+&q)U)i6YgT3z1SuK+rOW?
zj}C-KHorZR8^aEp^?U9YQc^~|frNwA)m<i<U-^ijw9nDA2LUpczvC)7?-62TZJ&yE
z0Sw(!ldv=~x=%de#1z!ZrW)atl=uOJtDp}iA`V<m*I^2Om<I6JbU#3#tC7Z|y4}qc
z^`t)<(S}0qm{!?)AHVszVwbLE2STdn!33_P<#*rb9?!$cb8Gn^O&9+<a|)?-t8*WN
zI^tTkWmtPh69#u`-@uFgYclB|?tcdOwTmSS)b$}LF3v@P!+(Q)>*X1aw2rnlliP3h
z`}_M>4*o6og0K`pvo=L}xu9dp@Wwo(Vu(zBnWIz9%QEoF4&v(!YZuhgzxl7s4lZH9
zG!OUzA>nelzjYBL$HRewCI4=&`_(73Cbj+{>QbP~NJ)$KqZ3w>-CbwE@Uwo|;KU1a
ztjXZ)pPHF(XB!8ny_)K!=m+hZrl$TKEn~hxj##Jiw;7d<pR2YmPVy;OpWT0DL;*hY
zoegWv+dZGXT<s>-c~O!(vr&6bCKY{j9<rwX#p#VS_Wu3*cdjiu7ibVCCdNm}hID~b
z8^^Pw60l)#243LUJeSk}q_5Dh&F}glMu`XsXH9UfmVf^JJ1g=Fc79+=U-n8=udlT-
zl|?5;Sz~Y&neEQP`Y_;amr1k7QZ{g<3N~1Pb$v^k%@$7L>8ETb@^T^Hw+qHh%-_+n
zk&Hf*W_<2zJpawF?AdT<;Qqe?f6PNcXZmj`U}K_=d)whV3gH3-Fv^|&Q;@LRTBOtC
z(G!b|;72NlXfMbU-@MY58Efw^ywUIVznqA=o^aly5glMhad6<+zbW>mS>flLVKn;U
z4CH7A3191>8=pTmxc>E@bK%|l5R@@QnGNfNQX)}%p7E5aAw(h<FYbjuHo3}IRCfMR
zw%7}_(Mk<%o`|l&Q-EHNl8?Q*?G!{PJ!v3kq-Pd^C?a}}+EVn;b{5(H6|=ovY<rqn
ztvQ$~_c5uS%Xpm4<x7V}RMmPHTgxorvT=#n26<D@)!!%TdYjk9;+fT07@n~}D^jbU
z3f#V=v6Ns$!MJ2<FDtft(<;<PbTE<|#gIM{OokGRp@oD4b@YvPNk|hyexh^&;va+`
z7MHg)YvlrNe7?2bp0IY@Z0dwD4p*k!#Xf;2;Lm?Mh@npVRZvJldU7y+)fswQ_U>2r
z(w>LhCwuN_e6;I;cH-0>XLtcti;pFeGR@nj>un<~M_)7)#jI)&D8#{-qPqyCd~2tc
zx`@eJ<bU!5kj*~ic}JtCdpUph(tJ#_Rl;cd^pgD*B%$Pw31{>%yF(dCIJ@aJ7+3JV
zZ#91zP7-G{?aTZe6%3(7WZOlPpr?GzO|V+#nXm#O`|im3ljM`G^}IAPNX-$1;rZbB
z;NVXqE3*n-jyuIwRSf+kHoR+F#hA;g`)_Sv*$6Sf*>%4VUm9D}$R2=gU)1st<f=^w
z;`KhfllpZ6;ayfbVZy_r{-0`|SkhPU=fojBsj{yP;Nr{*!A+O@2S*mo>9!Yrjfs>}
z>?DG5zofYdO*8K*lVA~dSr~@<C#yt_vs~OiKYY0RUiD%q(;3Vkmoyv=`h5T9<M}J8
zq@@5*)upYrZ=L!(H|H*qZ)%!v2WvDLJz8?`_xE>p<~R76uP0{QdWDp<47>U_;~<4$
zHmgFe=9yN}!2$z<K+nIg;0KeWw?75e6oPQ0trV=+yOs(8*GQaT`>s7%v<&MgNGzGY
z7=cB6;c;>fVO9l>CyNvUZ?1~uuSU0P+kyZ3<0nAcyRTqK=xvvkbIF|ck4)whlg?U9
z$kC3=@$t<(Qj@+V0Z+MR4J0|#8<dpquk~A+b%ZpIYsYq9pJ<DPt@YW~eLEU$Pk*yL
zRJJRg{KMoUw5B+{VX}4M@7&)%8)v~u4RT`->xn5?%ahSv<Kye-Xl?#<Wy!pPO~Bey
zrcm~0vXZ733*z%b1HVs%`mlq&RuHvVe`bq2Udwi;lc+5yNPZ!g(2{ZO2Y1Ioq*YS|
zomP}`hAi7}oYGDT<$=Pc*1$}iz7?s2Wv&BueI8`A75<9AJ6qkk*;HR^o6yUB2QyG{
zMNv7Qwtk4JMS7ne?;A}?KiU2ZFe^llp#|qdl!lC648&6wP>`=pRuH?l1v!6bXxJ>B
zN{(2rl3Pn7oiY*disDGfh|+qag^}vCqJ3+lS10aD2<hYGee~Yke59TGz9Rb0W)25F
zYa1y_w0z)lQ4=+%;mgjT)Dt13$)wGl-6SC-E<5EV{ty=}Glre?3oeiY2p@p6l{a>C
zTW!r~r7S)yK0K{KanTi)=FePdA&??MrP}yJ#b{QBexOa2qmSxk3pp)a;*71=IzQ(#
zv8;$Uh+-)>E~JEdqd;|%W-qNWTYYNET8Ip3)8x0jEl_FB3SoE|lQ`jX$M7h&U&hTg
zIp&Q0--j0AcLFdCez+|_@-z7GwD%W(yFZki$^cfCkv4EQoYJfz_v)Xj!qwb?C8)dv
zV<#G&C2KiV`o4LFJrcOj{>SQ`qS^}O(<2MM^l#&xGc`z|>T?%S#`{=o{|m3I=cUo{
z4>?u@?(mumk6eg{XS>+RJ~WD3?g<>mW9+6Jz{5LVd_u%c#F3f?)SoJO#OOs)mM$(_
zV$KYH{)F*<$f(z&rjqHGreIg&-Jd(z{nX^u!bWDCMMZQz#+Pi+x{08$Xr6P$5kMGh
z4#7rPMpv`8-)@bJDoAm?R~ke*k`~{e7HYwmCG(I|lSW<5Y{!@<+0uB@+gsngP(s<7
z9A4nW(sj3{v3d@8%n|+}O7YzhJaM+g+0f0aJR^-T90FZz;~nxUq+hN1<ZMkx`Ih?w
z*3P(0{fZtGkq}4qe85d2t4Iv=N6S7tgq_OR`Npo6(Z<(!(1^Y~Zn@GcbJco#0`2ha
z-xg{Y49S!`26=@wc`jK4zd(De8SkR98lB6xpp|tGDd6g2uf3pP7+W%U)E?>Q|76gl
zr^i~iG_Url9(MZyee=8bv5kYSRy@=kf9HecjaYfg5TCZ{$+Aeh2r58l9x{<$;pVtu
zI9teBdVs&k@N{dc8vohbY~5nUba~40nShB+f6PDBwq<DGWnXf(%zP7ZV1S>;((<E^
zw!0r^901O<1{r_+G>h%sf?d`3{_i^iJvJxG?$aRRSFoxjGs|Vun2vhyll@g5DZ_js
z`ua}arPgGL>_gB6(BgG&Qav>iCI5^CxBJjCcz0K+v(s#39mZ()aB1hX@xd9ZoSEr@
z!|5H%u<JE`N%!?~miuteiMc;}cZz?sy!trfGzEYabkEW8!^L1HgQW6@>uVJ@l+;|y
zRbEMW9MX*j?v35~nfi|Uko{fM)6boOso%mnfCOqm^N#YedbEs9Iu~!kL@-VT0?Jep
z-=ws^p$_dZ4m@Q45u-Q^1hb<>%FVzF$IJn_bhOod76p=~AQ44e`$-pwC5k{ip&#}F
zr0NUyVv;;N<ebIDptNQ(pK=mH%}rJkh5g&sM+o$zY>}ir42iyfl|M#kY7dlFCK&hR
zQvUtz8_pr6+NDd9#BZxt$&;|mgVK6Nu#d{7Z>qhIT{{n$)hM>d>Xl697a$+X1(PIE
z;nVz;rr}PlC7sc@D@c_(kQ+2bTd#`%JW+UY1S>MjGF1sb=OBf$WJ&D!Q|E8RELM9%
z<f{+MYLPZWAAgA^(jMv6xVt@-o!73YyWM}lBH2oSqHhRuFB!oOuZhP;uXa==)d`G^
zdbLoh%W6MZ*?g<XoAL9usPQbJQJ{UI+u&dKeeQX0l_}&e^QqCTzauJM4BQUIfGtA|
zxl$7=;as@O4~qIXB*NqyU1N6$hjDal#EJ%A-e@8|zTkYPCK3CFA;v{w(9N%AD?zgG
z%OorMu=_LL{VQvN$;+i>GeqcVM=;KNx)a^paT2N)2knW=g~X?*W<Gd}kZ1UJN7d&o
zB_ubV(VL0ujy>V8>-n|Xkafe3Qvd|2O(~ok9u>^k>hj0?(#5wO#RP2ha#)70t5{w&
zTJ~>>({-4iwQQD07X^1VG>jSJ<i2?~S-_2Rhi28Cv0@SU_fp^fM@<Tm%KP!&ZSh0e
z#9})!<&=xQ7^iVj16y%<o$n`(-1wwAUOV`UGi^N>&I~nU?3~0HJIwNW0Lb;&A!Pe8
zkjy_!-6uNEG&p0NPMSMoY`b1~Ipg#FthE!KJjnZ=vT1y^fQU^LogH0SFH)7z$ELPa
zJ<jRTgn0j)S}c~b?W|k)mqTU=XGT$9?_h9|x652pq0aZufv08kuYYH})|M{&?OPB@
zU8CL=VM}k5!VZrY=}j@S8w1c-!}~&o1qJ3UUYqgUg<sK>K@i~>m~h$Zju7&${*<hy
zizuy9jkiIHGD*W4#<ni_cyR;0+ZvmF|4hhCPDb_xp^X_seS=(_35ThU;M8Qk|DKF<
zVvzjYyz2J2iSrP9zuOdzSY@?81ttt<uXSJx`|@$d+IS2bniMJ-4p7^=%(+$i&c!+{
z7X{z6<V!BNs|apyHa-}eik1%EziE0{ItKlH0%B(C9(?z$#@A7}9l0)<d|9N-To|<e
zoQn-#ajl-vgKAvouEL6w%=yM1O<bO`%baV{QovN$rexScusgDKncsQ7a!jU2YF-_4
zxjk)(|4ZSjyH@V{7+75|SP{Bf$9=Eb`0gfM&^(89XyS!7356W5gv-q-{Of^t>>o{o
ztwq{}Z3hz$f=HfdA+x3YT?w9SS^DO=CzrerlIC3JnkRef^M1$o1ej*5*{Af}giM09
z8jyZDDGr5PnP1z9Yc$4Tr-?<#Igjf>%@0m9(A0U&Xz#EcAeZ6|&+uDX?LF2{4qGbp
zd2{}(0)Fe91W19MwFi=WjAjLx1-G%(#Z~krE9HCnL6u7SXtn>OG+<*wq-{7dhq?b!
zw^hVZe7dmR(%%xE4@R5#7@o#3cqd$c=w8HX?EErJ>SG#y$;7tiB)U6SDWdp}hb#eF
zUX{gL!G8bw48Vym0*zDjIzKEMwQZ7-kv5K=`Pvb-($G50?|yYPm!PJF788qujwnNm
zOETW}P=^0z(3YMnhKRD4<v!^Xe!yAsJ;MY$z+g4e;Ve-Nb9hVm%9pf&$&}7F;_J}-
z_;u@vO9f)b?3>p$vZzz0Kqzvn6aj-}KB|#~Yl``!0b9HXm&D}>nrYywq4~Ye*s0KN
z(|i_5uEWl_x&c&~L!DYBfj5lpy$jfs;9yKQ!mPrr!g~3i5e;EN@x{KXWzZ!(?r+f9
z9tE&}y>M9#JaMoZb}{C{A|$!t^(=xy6vECV&pr|kx>+<vwM6};Xi%cS>R6JqKg~Zm
za13q(=2;eE1LMVt9yUTB(Y*~HAR9o-i1Na138=}TKT0dr5~B8h{k(o7t3(t}sg(1Z
zVRbx%TAk|OM+@jFikv;kNzJB4(&ZNe(M#U{FyZ#8BO}FnwAwIrJQ%_=L53fs?@Gr)
z6w_X(A&%LV;$ICiyc7B8q#YlJ5J~e!A5?PzXp{doTT#-oVd$A+fsl<DN)_#{KzbX!
zuYI`*L+?YcN>HX#2=5m@HW9^DkhrRmN@>$>*P>9}vW!bXDQ6G>dh@|GSmIwkur{3|
zLOwIA1?9CP6zz-Sf{&^wi4k#F+rHjmSZ8}%rTn;uMafk+OHH=*a;W^hRXdTL6%8RE
zW(4D_xci^^jylQGY~wbDUb@V+VYk%+`NovZBP-+8uyLa)8rKQnCdxGP-O>-_%6j55
zg)Rh~64~G83p9xku2vKV?;)Epffp_GXjlJ`SthQbK(%MoL0HL9NHo_w60zdi>Y(qH
zx%pg^^*a}I_sUXiimIycw=dpE8M6WpRAZ#?cemm?eBS`LT&!h)hmTy;eZ`z7mFx;r
z842~mQ6F)yT@rpG(O6vD<tdGh#k|Y3qj%P8{M66$qMQ)EA71|Mv`_ZSwAUNj+uKVf
zC#$O5{(ej4>S6Fkp<4^)%Xq*>3|gtSFP0_^j`=#F>#r659TY9db{v6SPIG{IeFQ0F
z6=9>*F!Qx)xZHIZR2dG3EpHAJg*yI8-b8jDoZVm+f<ejfcJlV0!sg}8@v%>%$;+)-
z78&PVN@dh}tir#g_SUo0v$OFxZ^~^^)&nXFpSDUrCpQ?n%PBm=w4Cg1FKzt!cpffi
zL*$LBe9Q48qc<hj?E(3O{5jBr`sH45)`PBRe|_5BWTlzswHM2B*I8c|s`OeG^wNAP
zh=Xv_$aYJdgx*oKe}P|`5|a^=>kYzUJ~?VP&C%fap{AKMDmBJPel1vqu(*;=(=KW+
z`oUP}>glK?p(jHObGI<zf6wjg?@;PLY1_S&UyY(Nd-`ZO#|tH08U;YmRt#w*5D|vS
zV|H77$)gzh=dPif6-V|PhOU)7Zp}VneqLT7q6;&D%NNNf_U+F?t+Z0bU;5i8_BP~C
z9c=B$J%X*w$&dI^Bo3llB|8bikIvh8$n5oAIlF4ld>K}MtwjjkA8^_hYG>3$i$HAS
zjX5;X;;&eeMiaDOQv>fi&*wT%b#9u0n}(wH(-AQtxue4dV;X!>f`poU*Oa!u^T?qz
zU9Sb*@)W(I$nsU@B2$^WD(0<K@;PH?e$T(37<ZZ!EXd4Q74c67+O(-QxyiuPu=#Ex
z3!D>_cS>okx0BKZ6kU)oVQS*XP_AwpTUD&tq<Bmqmwq&}55V=KvM+~)>@1N%Y<XQR
zO5`!&BaYM&Ep%QT=aXX8gAA7^5GxV_4K7N^mnbq5Iu?;<&?xc}2*<mJmsIo#A7TO!
zf%Y5{eGekzc+1EK!g;J21zL^G%QL?55vp<V{&3O5Q==2Ni;v@^7IxBSi+H4`4uv9S
z7>+vxHL+K?YhVu-c3llzZCYOcMWF;99ds;@%Mpu^tHT~r(Y)*NctxZXA*UwgE)(W6
z=|xhxJS3X%S3y3jq@=j8hn=4U!eEaFyGP;7De=<_yP*cf9?dv4O2i-uqwoiXN$d7l
zc4SPy7HE~<6*bbn%XW+=NRiBnkK`jU8BVK@Gy5aa2T(KBr^kDXR}ay0zMqJMjJS+&
z4W~S+nXdjmD=pRLb$fFi)=)6&R~@+f_<1mz*`kBIl(($hm2x*k!g8d7_iQX|qfaO7
zFr8vP;AU-GXQ9O(?jN{lp8P!3n&+-$?KXq#iw0wN6pAHe>vhou=Bxwc{~|kYrf=Kt
z5?^kg0DM04U0pZdxA!pZ%|?3$C&$Z6W$wVXX|&|+%)-rF1-t?z{@&=|I?fV1FF~$r
z{z>_Ri<9%MiA>Fke)yZX;Wa^y>&V+B)+-e5N|yEdPd~7hUbOIaRN=z%25Wf<mcFm{
z%OB3_ggKcrs#xEF@ZOznoCD5<ZfdIyYxjWNy|#-I*3hMfqU%Y+j&p^Z>(1L{O`*}p
zzSB*rA6`yoJ=di>k;VOG(rfeEHJN<M_$$SHZx7RXrl^Rld+;Nr^vh?%*V=yC@7|XW
zA4V>ZSA)nCt5O6HZ@c$4YAKW3tCQr}%PP@Q3>{2Yt7_K+)2u6=NV}@Y{(ST;%J2^G
z^dxjEmf<ZgBm<JvFeOYb9eEUes=Ncq-#rbHm`4uZl~SP8n!gA`k#>C)=B#`H5ymjO
zdpmK)!m|C&1hLHDcS%32epGp`9IN5`Eb|w|=;LZ(D*W%o$99=s);|%P_<h0~zj`SD
z+qDhb7|I3nWHj_5)e!pU1r1IMv>N?@dt0`_95Z_~f(C(Tq8F4~YcZ;wR{0}HH@W7b
zYb4*{FS#yg19dK!r7zjvcJFxovK%$XP+w)z9cP8qH0QEXrY3?!Q>g9Pla~}1^FgXU
zy-7lwbZ32wv#L33^C-HWfZvXGhOOz`>=rSbNg3@tNnXp5>GA21(Tqy(9bo_lq^rq9
zG?RF&Q<Kve2v??cP4r>rS`|ARL%o{xV`EME=_@D34fFTm52p$lBk|x*)CL&I3D}sL
z2AYHg7$Fd~F18JlUI>ITwfq(R{rDFj`b0>r7yL2w95cf4RZ882BvEQnE)V%DS8Gb&
z&d65tNfXdi+FK7Wd7~1B?%V^IIMNV<#)OoQ@L`We(;BGRKLvC{C~c1W)Tn9kpySY<
zym1-6HZf|{d)zMT4nK`()ei@rEG1-}mK8bYNBs2~!XGh8{}GWSvFlNlE5jZ?Xa4|w
z0VCa5j=t_ZX};mKp;9CfV~=;xaOC;|RydAGSHH6t&G6Pvi{+Z^^<n_-U=dFxDy$wZ
zw0buej?py)YBBzDYuZ+S##mqjn+fJlLtrHru6nzAhmQ(E2Y9jHEZomdPlw`btK-!k
zTRR7~V(iEBn{~j;sTOP+^t7LzpPys)mN1ww^J$VlSS2r}7#5gBb!NeyoaYJ=l?dv3
zpLk*6&ed_hf=z*Q9^gvS#gNX$X!HFuaEqb{+x8D5$dy0-D;7G}ad=|5PJh3!<J9MS
z-u?T6M)20pDm~{uvuEAeomb#m2Uk33U|YXyc?Cw}E=EDus(;vaGx+WJEfMa-{}H7$
z&9v?R9X6i>*7$+_ox4I||7ch*|5PXhZm+lA{sV5gi_U=&FbcP+!zvH_5g)zDv@^!Z
z;#xb-;kP@tB)qrS+unsi-l@s8g>xYg8{T?$Gk3e*dAZ81a1gH(Qv5lq8PksA*FoJb
z0Q;E=x8v_`*B7oRkV2i;@Y=cvVpR(H+vtUVr_G&bL;RM&dM<Xj_!L8-!Vu%vVBC3g
z@qqOl*zey8y6D@ML^eP%bf5%!)HmlIEMX{Qi8mFdcE<YGZMNyb#rAX&cs17H#UjLN
zl)!tk!`$PUZ*G?b=i-T^*M6a4<)qI5MaEDCLT57#>f&#<dVMPm={)<Z>8!>DZ6{U@
z81grD-f#4tjOpYNqOCc+jHA<R??HK-k{W@g#n0QZVdi7&QSZW$yj`lVR{jw049uie
zs4Tb|^h@hLwba_4`GxQuE&uHAz0>X2TgGAJ{ApXJn=4IBS+=LXF#djCEFt6T)YnA~
zYAEyRd*)+=yl~OBKfK9LdC4n_sWZQ@%5XLN?_QuVj|>G{)mn7&U(VIct~4|>xi^mj
zDbYGQ+IU*VW1+?gQ4I~meREact=cX3_Bt;gqx|hV_C?C;YZ8hL#M?FODvs8AgDW1U
z8sgz<N||lqL=q}XH;^i2h!?yPHy^&bL5v@<bA&?@^vFb*o}#n|0`xQ0Tz0%!!w%Zd
za2+R;&7CL7?pJ8swY(`OFF~n&YgHoXf02mo_&l5<ihq`mVidFgcApgTqg$BV>ce0G
z-S_*>miYPSd)8q&Tqkc@%^Ullq^ePQy?CrV9A^b-C@yvQ!^UCE;)qLoTCsy59vqNE
z{buBV2(QrQG{jK-5zft^rslBLrCs?jnd2FYfGGVNfA|aJh4%Yk{=!o&A7`Edsi8J6
z9&7L#+fX%V^r>2GMK^OOXT>3%-iZjyin7qYDn7kevA#EYa=1L@q{sOhWqq$#fKzqA
z<bm=l2*($f95ubBK1DkHI=3D6*OJG5Mm?UVMmZ^J-qiLiq;l-`-4eZW<)@+3M({TN
zKDRF-bX=ldb6-OjY?67)zYAc3flG5s+B>cb#2j?VzQ$ODwvOcR+Tqgi*Nw5Ti$u-X
z*x**;z6(V0n%ANJ`Khe?=+B?8a=i}==gMUjEZYwfEkianI}7z=b1IGsx)ZZ4fNSgr
zDeWIE^;Y5^@4@P{Q#StJOR#BqIXKp_UnZ1*&-Q8Ke_X%Iv%Onjyz_jEzVqL6=XSgV
zk(eBm;hZ6Gm2P=GHP*EIqp*3t$-NbB8L-suzW_>IH;F=R8IzGGK~IauxYGxb{(B+d
z(mvkQ0fI#4LNAf6m&9*w7nQXy#%?xm&;R}petI%U4BuWuwYlbhbbh}#f4gOPyV{ow
zKNYgL>N*2{1E){$<$bSJ{`bPDw!FqET%^8da!&C2%m;^80H^IZx!bEF;@Iq)aqv9H
zBUHjIw71GA1TRH{Xww4R)p3Oan8zO)9E^7e+Ai>?lWo<J^E{bbIL~S*Yj{o)seH3n
z6zVNsq?xdAbunGu&>|_Xv2`g?=8$xLJ_ab;>H$g8F>f5Ie_~og?~apGyD5>euV#3m
zikHd34cYlviJ<qfuPe>t^qD=czz{E?o+fS*Y4`W6;cvYyA{2N|G$I}TOA><HwF_7O
z+tRUUzv!9FzTdh3^KOxOEv`NIa{WAz?I$_WY%T;$T_uZX*<MI-^{N{i59{j9Ap_Rz
zzX3vw*3?|FKeBcq_c-Wanbuo#zS>NswU5)kBODp@Hk6t4G=gKN$A4!br1(8wiK}dh
z-)Etwy8ERrtKGTjo}yb*wYZPSoktwXqr9gDzK<Lpzm+=Pc*qpC)vL4f(89Fw9Oda9
z@>NDg{&^4he^YN_x^9*Z7x(6gt6gy}>d}CVs;&(yQ*>zx{HRy<%<aE2%Mm9Vt@}!P
zU!&=p(%yI;%ubHAc^tzPl)e_@bBva&vx_YY6B=~8Tr@TGMP<PLQ()V8+?S)rMy=xX
zB=H0-rJgs)%K+sN9at6!JJ2kW{T>WLB+t9jbrep1##+7z7dX5rkX>QWPWtdll;$JA
zJ1kn3p(h$Mxc?{RtCJUsomf~pO2ULoqjWKa*~``M?4VdzPFfb@lGmkC_(fmqoeMeI
zP=s4t&<<~~?+XuubFP}AqN7VN{*Ro^_v3-KKZQA|#TrEJK%o6Kg>eHWB~^;}AJD?K
zDGWnDtI247m$BLruz1>7$#sYsKm|+z8?7WVvJxnDIAgqInyw<=%nTnET2q<pO}<<8
z>{*N7N=(Khjy}rdMu-)8U=!p6ANmzaTEhU(dN@Y0*TewEVYSXs;!VgC&TChP##avd
z{eeZ~y5DvvRn{HBt*pzzv28YFjMr2x=B%}O{mW$wl#wMdJj2x9EVJm}jUND){5yD6
zRbEltXE?^Y=szVTG`Q-In`k(kgN$#^mByxK+bguG1;8r7V0}61>FG7yucvEk4X`4Q
zb-8{T0{ipuqR`V*q(AsXrH_q`4M#D^Jd^JTzPzTl2w3VW2sZvtKYsIg)u+SA41^Qr
zFI=tr$bgDHhZh!Au<Lgh&;Uj3XF(qNzwO?(8M`-apLAj5A#I0^?##^0S{2}x_;d)o
zQQX{MhK1TGWA;ujC3DX=Ha=V5j8oiR^}W`)y{b?Ea7h18`y?7sc}E3j?$54&i;dbJ
zm9Sm~PWr2H@e$2UIl<^oz_xdDT<F{>ly(KAvEUY3f-g@mjRE$GQVv3|Fu$O%5QMiy
zL6P{}$;^x~`Mh<@PV;v2f*B_h6;U|CX8LTTGVo6!D-~nb;qkveG7nxuAzYp24})>f
z58L?ouw2-?!!z@;s@QxR_{skM1Pb-8ij9M_*3cQpkGl)is9ZQdJ$q=glc&B^7`oYV
zGV@6(2i$ks0m+3BkJ~+B3E|!>b7T4`H?=lkX{c({z4P*>y0+uAcPv!fF=*+hnPbP_
zhi;g*<j;jdLaYlIpEo>GrgWNY#k~FUFJ8A;>-ExFJH>{%zQp4p{mPbw2!BLiRs3_x
z94K6taSuv;qc=M0<KFIi;nrwT@M+?3f0Iq=xiD#jA+Si<40*=WlOt}IVg`#A+USb3
z8Ls4mzTnFC9T%VD`8M;}m!akp!nY|$(Ef7tg~9f(ud~k&*V5vJEOYmEarCg08>GMG
z?acP|hu*!PITPQ0(pf0Aes*p}n|k|%1jFHanVCu(Ud~T=w^A|^d%KpN>V|CZ`p3Uq
zx(Dub;y}Dmn}~t+V`Xt>>&G0=ng4rsduU4Mfntwh4r;1WN1jnVjdK{{8<bWPY1juT
zu-RePN0#TH`C-=`I2xMbPh5k><u&1jS5^IbyP7n@AzelSDY5#&Q0Rq%D)van{&H!;
zR$IkU=`mPskp5w67MsbxaO#x#i&LS=AN}H}4M;?izcNCm<&SW26Vd78KYH;W7a_Ji
zE{REi6NTzqN2ri+RuJjNTV3UPOT<tyKwh}A!eEpTqW^H@5GhTLa4^$MsH+~s3uoM{
zuu(NsigO5&#LDGr^~Id?JSLDyixZE+O9i#vI&w9wgrM85YfZ`1<c5{m66uOqfLk75
zrqeB;Bf_sD!DG8l!&lHINK%~qrXWT|J*d)H9D6a7!1+%e6R7KhR*yjtiX>L0-5WKi
znL!kxVWDR`rNT`MpULw(j44K6=qERZF2sK3O~8LX=l4L_=kpI{E}yx{%f4B5sLjs7
zrY>1J?N6^3*?gUafa`nKWY>?M=Fh=_ep@qb{K~=ubg;XO--k_9Fz&*MTIG+MM9d;;
zw3#*@naO1=%8t~2XQmss-cTa=6<kdP4cgUucy0?J`3$L~aa3<wF){YT12G_tpzGjz
z?xXveKbB<I)cqzUuAYRdjA!6cu^ZZ%P+CFZD81fUb`TGi)Mbk6j12EF@Bv&Dy8?$Q
zD!9SD!*$N3c~1SAv4B13`{|<;G4bk&Q3Ml~#}hY){k82uCqq-$*W6(j4MiPShnkaB
zL((TIh!HwF<5(#*6k1mAE9Q2SANVe2YUGTro_ruF4^q(IhYJ_ilcYVuMzLYB6(T)V
zmBzfQetOfvt*za{Y;VN?I%r#i$T9C7tQ_M7_)vFAe($JA{ZiNZ)iT3cpZ6FD(z&-t
zpjWFORa^y^4YqZ%d|BnBN0+`E7<{yTNl$TBP&@3D`@wT<Gx|cK^Ha>JxLX=L>$kn)
ziWy0$WNC3tPAUwK)=T0Yhi*|?&np9z$va;47sJ}s-yO%^kW`ugU)V)`pE<U^u$C0Z
z)-5m@ieG1m_I>sw`dD70_#=)eB~4u<s*abq2L`KxFm7FTy!Uz|lP%Eqrx+2=8zK$j
z0>Zm98Xfaa>>G@BL??Vqp3o(p_+4UxC0jchn(o{&{O)AHt;D5Ox7`ePZ@Yc0?EXh1
zsfRF48)d+*_A2-r0DR)HFTOy{NfsQH0Qt6Gb|}T(VeUg*w8D524<pzmrQu^qmAv<)
z3H!CJ2_L`UG}fmTH!@8k8EY7m18_88lC7X9WLKx-26uDWx!`L)=MQ_Dt6eerC7Jgp
zkNdJzpF;tS5`+=2C-!5kR-=(tm7bm!@4bhKcBa!*BBNWtJw48y9$27Po#k>B>F%k8
zg?qDz5*)5YqWpH?pRjd2UoS&qF$3piH>n<DY|(Sr^9h@OL{zaKAK1M`_{(S;CbBJV
z$jKLIt892{j3m8B$oRA+A&AC}?%L#ykVB%+;uJrS4M9oB_^L>%IC2${c8|0~-mYJF
zYp(ei=x~<<!n}!%C0`<#lp{$<v~r#^;3>Wu7iDMTIPQK$>xKyC;CKXk1pOKlmxD*S
zFJgQoG@*_c9v7k;!$8SM`v`86#ADv!LuvEZt=YUe7kv^UldE;#n+|I>-E>s&P|<!J
z+}^!g<y$?Lex8H%HPeNl&&BRBRKxL0ALJke9(VvOW58|RB*@<@1bfqd3|?7Tc5p>3
z^SN4EM^i&aShG5e8>EW$_Fw<>XLD0jIe2bju4ilye|k?>*GtJtgYnOKm*x?BiF}PL
z&?nm9Y9KpYQ0{oK>v6K&<?<cG#8H|!-=j9dy6DD)#j)QUbl#LubY3m3?H$G{TphuI
zqR{o_UaSo{dqsQuGmZ{#te?iY3hPA*el;jh-jGGy%x24<3Ek{d2w8R<A8K~~{NOsx
zAOY3a!E*eLeMJY0Q<?S!o9df~hJtaHVF%qKZS$W`&$2ErgDEP+bCee<nDyd-HLwl!
z!NR{j`*x1-HS0I8U^A3rtF$u1T#5iwN%*70qN{(ke?vXMAvb2W6KqNL&xMh_cA$gd
zO=V1sk4LWUb-8qCiBOL3b?r@66bAf>uBXtIx6vB>Oh~N@CF`F3GFOAzRv-6AveW)&
zYC!7qkbsil2l0yV3DO-<rK)InXzN}x=hiH_(Eb#e4Lv-JwY1%aiZU@-Zvfq|4DA+N
ztec~mbHXW+PkoI|4MC&CT^j)swpYdmb}(*rl|NB`0xf04(?lPvBq-zQruB3<8mLFs
z6_Pe@dlN$a?94eG-l>qdqMpSuSi{M%8ZTI8?Ki-U%u(WA1RuV!P10uUyP3=>uD*?=
zqm!hNwL}>*x7c6DBE0@89KdpF10D(41$xfi9sT5$h$YJoi$&p+>=WC%PH$n}>fQ5$
ztz$p)iPFRSFE^U|ytJet`V9qZn6<?0T1oS(+3C(ki$<-$x)qiDpOm>X^jVZkAZdz&
zEZ|@{df~RZ^CTOvz5Q1N*tLj7GL!SJR!WQMrsO;ar&d`sMMUw;0DHtnS8uy)Gp_S!
zJ-A`U7HprtX1(%&e+QXNFVeVV?&~EJqhsyWq^qmWqD3CV@$}ez>;};gG3nR(!C|73
zVJ0~fT(oZoW<SOJ>~+kvPbd0MpVm#HF_VVUFu?xB!!f)nnfr>#cjHC+@?uG4AxYhp
zsFqlEy(xwgnFP7_KpuYHGC0XnHqSIZ9LNm$GpPS`<in7-Q#QwY8e#~Vfq^*h(7{r!
zDvezXw3>);Iv5OQB|U!n8DjH^Cs~LjB8Q5-Ea$Z{Oj+?2bbyy}xQ{_)BZ*yESr`??
z5VFL|V#yfcFya_brOrk0BTjYG$n%OQGEaapRnc+e;5pSVpyr*MtTTP$x$trD<?y-7
zI<qOks9&R*ptlx&Rh6$fZg&?n0O1xF_d)#$Wy--2V*x8qk8g+}c~rrlEmTy1@NxA%
z_}SHk+1uPCbrvj0MflMr3bc5jWG;8aBrSq>`+3)PgIXJ!+fG+$qN=1V+X94-3+(vr
z)=FnbcYXmn7yNcFaSJlvJUm5nS(wdo#_Y+|^*Hc<AsW8UA^kg!d%>s&CK$Kw<lEHT
z3eJGl3HMQ>@K$t2^b>FZyYu8UuZd<bZ}Mb&dJofaz1Y<rB=by4Do2}}x2&G;DR=Jy
zYV#L`FN|i}MFKoRx4<{??JA89M@s3{PV$3vKz%eSMqO{~l8zKn=VGK-YuV~|63-2~
zZ5P|@9>a39%6T8-p0SD6Cj`M?YuM=u`q^g-Sr(4LKZqVbOWtyjmi<4P&O4gz_kH7`
zR#htqLVao!P3%xx%}`R*imk20DnjiR6}3m~y=%`<D~h6usFhL`MbXwSt=8Ut`Tp_y
zKj*wp&Uw!B-1mLm*Y$E^_jPt2ob&FSr+!UMTHCF^26YzJaTuZw6;baeW~V!V1D#!(
ztreZG^Ruj`qrk&Wa3ehpvmJUN*Hv2~Y@Rv0@Z?j39;g}c`o$qd*gxFPnvBaf>(7ex
zAlG7S4E@Q2jy&Yiilmo|gU9c7wCyn)*S5uO<RMA?70V-m(u;atlk$1pznvY&NKxsb
zqYh0jKC>-0>QBi9aop&fCt&Wmf>Q#aTVEj+MAQDnRKSM2)h3cfA&Q1^^xF+A1~Vqb
zMEnYrVTl^Q-C12bm7ysCt)hm-3!=eB*~K>rcPa-PE`N?4dMv~lA7+g-Jrc{LpH|%~
z{_qg7nKxz?<bSwxk?6v#n946eo6!X6%?MFad*{3H<nlYGX9)Joqc(E=r?yCZoh&iA
zEnU;-LpJMsNB}J(FVp5~(Bo4q@yGV%D(VMo1c3YV6TU{gdanh0^PJo(t`}xeSC;PV
zemXx{yVy{>xU+ah?y=Lywta9_DiC!@S&vVhDb*Z~eO2HmAfTklSQSTn4j+9HN<$;4
z$ZWSQwvn8zsAq&kk{J8P*T+l4B1@~EtLX^<_;ii4V?idFx-l`J$@in!o$M%3nsEj=
zzctn82ciVFccpnu(Z-D^_0ip%L{ii$s{zu7zjjGih(1c@jV%Lq?V<;gHTRJBq=P@=
z44zEfe`{cvlfyzJXzNhiZVa#fix<=)@~`rP*BF{>P&m2?C1nEJlR5y6)KxxVgHa~*
zIciI{YDx@HCP17`(ZuHZ(2cVsep0MVj~RWg<&AJ*O5<`X7)-jtZ|z$lh3V{izJVe&
z)`?7jJBw==we(32S}@`<iaw(d>mi*9uid3={Nx*q@cU}%B9Z%ckF&~S-kqKKF#Az}
zI#H%f-4Kj1Gb}Q_qS<Qk`kgz>su!_Ix)p~DUGfE_CWq}aY+!3mXIwHapic@nz{{ul
z;(Ru3x=5#q0Lj*$F1*=i)w(AnQ<=Xx8y_a>Ms+qNd8V^vv2a?yiEjqP)o9H*pf&%u
zk-I<f?|jHNaC?1Wkn#n4O4j)9@i%VK5?#465_qe#yM8oS+D$&;-}7!vL9K}{tPEJ<
zr@AaAuz;OYU=rb?>H}6kfv-l->TDI<aS+gSJK=vL%jlX-nc_X@hemdL0{DpEyA&5a
zklJ^zgq0AA(La-KU$NZuAt8;UC_m*84^8})Jo@d>&MNa|o1@d0yS`7As^<vz7H?1@
zHF>D_<_s1+@IQVNU@Kc!SY@whIZ(acCdZj)j9>$Q+$rCma#O>e4o}_5G1Z4T=jd|=
zOp6=;`1PWQd=4}Mp*%WYJ;zvT)cvTodOLMcm9-FiM=GPiXD({|I)zPmmDPuFRFx6X
zBrjL_rLf45bL5z`v$x!$s3u2Hd3y%Z7iYf3S6x%W?*INtoStk>fk%=SPcZW7EhF@=
zP9CgcEkn+DK;yknAx(W8?S=?rJ_q)U;-`>1WuT~w<>lr3*o$9yn*gc;Nl^!-cfrTA
z`<2WRj|xcsd~4v@K<jdX&=B8v+p8S2lo-y*_Pni3M-1-8(taD0qOdSA-ALb?WB9{u
zzBj^@h%S9SBY@?t*q;|dl+3~I{Z@-DPF3us-dau<`vka&|KAFnCK{yk?KV+Cc!2NO
z#&E_b?8WZm%QWoi3U;5qFPV5@W0(@_Lqe-?Yj;K4KlwR0vts!>PS*q(>t{DOeyN?}
z;6xH2DIrsMO!Y_ML_rqhm`J}qQ8)2+9Bg=A2KK5VzphoEff|ic){&2ZQB~yoU=4?)
zlG5HQxPF)FjY-0ROp&NJWFmM_&F^gbzd?`t>^cCdsptpz8W=7qYL=*~LHd<{Gs^zm
zG7FYh(#PRm;^2#qt<E8J3UYtK)uF-wu-3v7!trdYw19u@C`OVwsm`eO-@LIJ7bP_<
zsrCi{=gjr6zyxUIWDZ<o1h^8ged9E4_`Iypw|}^w>plqxWq#@Lt0UxXsNw?%bWc&I
zl_#o~YpHId<Bxl1?x!_vm`%g9Ba*An57}yfzRsyl%STC=Kl&-7=M75?hgv%4br+D<
zZvVP#$eireufxJlC8g}F#Xzw{yn~c7s{j2<#YJ<FRxYw7<}4CQ^V>-V>#qUtx9*&;
z&)a`Nf)PP9aYMlkHM4|yk2Z9Cx!9eTi&A;@n+FH}O)fZlr2@~kgc#%TCb2pgW*YF(
zY0?UX&KI%Xq=ZlpvS>ap+x^Czlk@b9B<g3d9jpP@dwsJc;0EC1*K9xOf;Ao%#kEm5
zjF`5RHh`K~Bag?JAk?r-L|0N6W2*Pc*mx`5&*Oi`s79Bz&8fhQDBH{xi1X(|Esi`B
z1`*=3kA3dWw|3~ul4Z-Xy&0jgPv$(DL*r^3Mh@oNJmiWtUKXDNPeRno<_8R$Q=9TN
za~aF+Z#wN3e2B4c1esZnkvmcCmJ6K$BVNuk&c@mZEg&zdsU9oV;dHapvP09|_zN8<
zG09?#Fh4iH`SSGp5h&sDmK~(BvxX^GmR^(k$=ysYEgsO*Zl9Pu|E`e7U+))}aS7UN
zcRJNVvpQwpn3_8F=Fb&lP&(HO1Y>7BPWOH>KknRLz&LFV*W_;NbB?TBgxs^UwBRd&
zM1Zp)#r}4;XF{>B9SPb`?OqWLE3R#hsICtk=i0rBnt3uD=8%28nIE1an5&VIgCVA8
zs6whyh7Da8rrYZjA^k0XQ#XKbvkNDtmY$w^7_w!<n$mp3NbR3!QSe$p+hWKjC+9)K
zs8cfqHw_}@egHxH(?2gsRq~zaE(I}J3h=Mdh}cc9<YTYOuH~EXth|V|^hQ<WM_%>8
zq0<i?dtFdCks+xyU<`m3F62!61pPdsa`*N`Ui7O+?CQ-9#!yB{=Hlxw`-k||1p(4@
zGBFG)I9cZKQz<50rMUF-9Zdj0-7MQ`NfOBcc(Tl$U}ne?OJM>lN~A5hPov?E%e#Cf
z2x@>JcGxPED}9jN%l&s^Jh=+3qH_SG1Efc%(E<SI;1|Ih!DXsyBEN7us{9e)rRV&L
zYx}F;IJKw!>LsrKJN{KC_Ups*?ciyDcg4Z(Dq#U`l)pBuV%@YTplXvFVDhV@o06%&
zZ93^*;Ip}}RBnc32NV_-oOlyS@^;EhFOdpZU9IUs`<VCvY7?Dd4qel~3lE)<%6IpD
zQ{lPwqHzLxHMQ0mxflzw<78X>fIGPCDYqMwjJ$!5ukYTMN%_xGBCNvB*Qi|P{8fvt
zsNa}TYDksOGcX2~J|Q|N&Ckq_QP<a)rtb$nfX5Fk_;-#e25o0`(mSu+ftAw_-XEp-
zKco0FDc-mRcS1&k$-2;eRNaFj6#4#K-~LTttl%})SCEpZ@t|$|f7fk3N)3QbhRKwD
z&*D#t)ip2&v0B$rmW=3byFa5jL$H;6bmE*}d`$KB)$e=DX4*`7PMzY2ER=1L4oj~>
z%*UA-GI%M!w&uf&JZb-z42oBzq_|G7p}`w5E!(HRDSFc4#h$b0T>~}iAIs?vZ{=gx
zY}DM%wQg8{MZN5nBm4J-y4T_30+k9V7=VIiCMdyq{&?Ph|Mg7%u+(Vn#>tjSbT8Ps
zPF{bE@H7u&I1waSBVsRI!0^m!Dt<*eLSe6u(w!R&ciZ8o?*L4uWoS)U47jiW7Dc@S
zN3S1$rIY7;y}JJ3_N}}a8Yw2ij4<Y$p>%|>s8cUk%dpsu{BpX?KC~`ip~ok%gFp}&
zwz_A>W7~1CCnNdik6n|8m=s%y#H!3t^^7|@1~W;NV(J4%0RW?0b)s=fSMPR#mzX+h
z3MDu?VS=dD=wm*97MT|o&#{l6KelwfRO21IEc!i=)Rler^80<Yiy=3)`&29dK>$Z&
zok$q`i{HFcGgW9KZ7yxye$V0b#V=py5P}lLR*F93HV`*KBn4ef3jd;YOob*E0i;Z%
z@U0{+0F+UU6`jpb<d3{18}2lvPtiD*u7oY~P|J+pUJs_x$J+}hV|fMz|9G8FTk05%
zEHj12t-@a~k)l$r4z3?(H|-q39jy{2Z!tS3HgNU7ehnuxL4g-L)Ku?-)O7y-b1BW5
z0ayG;i4j))8IU-l#~m2{8k&mW4h)Sxk7^PdR93t=QGhw?lrd#%Yh~KEJOhtJrn9O@
zgTbErZ|^;_rkI`aSZ%j4;7R>vzPO;4RY(0C87VGcw^+rbwHW5$pebz}5yKHbp!-Y{
zj;yPRg=DS>3Mneu8Ds|0XBFu?1m6>NR<5s$#gMp99tln(W2-tx<_S-yH+QFg?QI<`
zZyl9wc7T_3(=0+){)8Bl_Z>uD+qJ<mJGvW&^%%l*{p{4R)`iEzQ?&+@xWJO7D{+c%
zet5mdi6=A<rw#Rk*)+4qvVx*5&?fLL`L+4gersjx@}O<WNAKSDU;!?8QK|_A8-rnE
zX6HRx2xFG1I<T)S^`u^V99EW?VTE8Z(-w?MHobL1C8QYsvU!TfsA-k2(Sqy5sX+7Q
z$#pj|reqpy_zM@(+`eKN05<wg2DU;_40?9@C&=?)Gi|)iaJvf|cu82L+DSgaxJ*Zx
zp+>dTSss{imbW>{RbMX-kc*;3co36gzJ5)`RGa;z7kXv%aF~EWQ^Pwc9p{2O%21$2
zv+6AR@kOb#>Ym_N+bJPxiiPl@@MP&2EZN6NAP(Dj@8(H&lkw|}Y?zQ9j8-kj0P>KO
zl$1$-nzkVzg(rP#+<v#KZ7h28JG@^jZ;sz!Egsy8nbWfqUu_;cU%KndH<MK_PrzKM
zO5WcYCw~aTrQsNcAvCv-3~e7~!R17qWM6_J-eSTL`c`NVTE$60P3>}E&a<$dWTz+|
zhxKmy_IT?*!`06EI{e`RXGsn+cdP<Ku$!|a#eC!_@Bxfdh+w>ECFQx`bLGVuLU?0h
z4wr#drdDSwdLwJMPmp!!4L>ykrTY5*LF_}UyUPg$tW6B)IXxe0p=^Y4#=gC&X{<zR
zMzj45oJ#B;x0FmaC3js5l6<*bu6BNc4d|iRr{|xqLYvZk%*U{>(qMfzuHE55FYK3N
z)at%}!j$C1Rk)cka7<*u+YlD@iZ~}Ek~I1TBn>UgkDZV%7fas#>g%`)6kvI=eSb^a
zyS_#wLdtaor=h`X^+p7sbP8cqHZtG>+Urjer*!!*dpu^*H9`PBBa&2!O%gV+SUM`n
z>(j)9V;w8;<DYNY;;Lm4t5u~Q>wQ0mGSO&1TWMZNest_#Fwgt@bjc+{v#7GOB+rB}
zxuj1D5*FyU_u(ltfX+_kfNeDj??=DI8dG69$vihTD?LdGpbjtX7IEvu8v78+wVRTa
zj4UOxYNuQodX@7sh{y<>D7MBWe#5D=K;ocJDjk^-1p!iY^^Je$dUpM)8Ikm3r$O|X
zHj*MdK^tja?iu}vsv=Eq@?fDEMzqET1`4oN4B!$ThCx@lH|u06l&yOzDtil+>2W1s
zTGgDZMkAk-eCVVSLBK(ZL*V}4AZ6;VZB-;^aIav^LwOIg=plnWI(EOo5OsWXR8_x5
z<nAL<yaT^Fdvi**qe5Kp<@~;4G*5CSz?qlPU3d5X{Q3W70{PgCyPc$dQv0NuPXEoF
zu3$wZ`J;$=igs=~q!mv1GFD+-f_X2)boiN$d*Q{?hU?GQA~EknCqyN;B3o1J`^-$!
zA#o)N2p()as9B`B#yiC4UCdFma=y>IxBtRo4@e00Yg2ZD$=d^`m8XMT-4v}~N-)lD
z`YEQSiHDeC!U%P`C<OA?fHMtD-AX8a6}D&)#hohFziiBKR|I|6SVh^kSkuW6=2T@`
zYAwb&3LUVVtL1JK*B4-%AAV}4!)A;?)zTHPV#6p0=jm;@o}hniJ4{#yjvjFhtD<fW
zr3*GJrODcjq>kYS4uA)eZkHfpciyy?E;aWpnZAG1s$5;;t=dynL-1(haK3hhF?~p%
z(a<53opBd@p2{yKL06eD^;wd1r`fKq?b>$Qb}}oF<qqqc&`E#kBjbq5w1cy~cbkFb
z*y}P9k{r;$Du!1kd+&QWGH>7Ejvnx~9Bp{iSs#j9x0iK6hx*VsXO48&-}$N!2MxyP
z7D9kwA`hnX^9#@S)_N2ki}#m29S#gSo&BA)7<ei}>8;GjepZ=qJ&1^X$;S^!yRq31
zuURz^)m4b!!~7LXrKR`pU)W%eE$Z6^uppcX<;;p93*YwZMQB5Ae<-oq)Ncw_bu%Xy
zoquy4OjixwKKBgSj7D~y{~VE=a%3-k%__udI<C`+5~r&+T4Q;hzkK%hTdZQR_tE_J
z+4pNbr^o$*AFA%gzt+(>c>0z{kdHVEV~TkXibxxOlmBV638hB@!zWybUSB$%VF{W8
z{5KeTIqILeG10IjkX?Lnn)*aAi%CqoG<1|t6Y9uTG*%DOf##V;U;mrRrAWO=ss<CA
zY9Pk9Zff)3r$(#H40{J4gQJWYpW|XeHNfhzF?Gi<9#aIj<<%?3%YIw=Cc{UcqJ)_N
zidrHgpA|VNwW-bBQ86*cdqM)(^1Cc-6V{I=g4{4p9PHY+j7!v|YuPk(Z(ai^Ds;K0
z4zFp#g)~!BWp((I8Q7lo!-mgL8LaZH*Os2`KH)Wf3L6SAHj($QcX5a9Ylt{!r!{gk
z0#WK;XVi3IkqR*%_>ZiRuH`MwAmS|7jZzJG+($jCYw!1M+JKFl=8<G<Wu>#EEk)1g
zG|!7^l*atMD;B*VN#$<!5G{!Q=Gt1bNw<&Fppu~$S?=(NS^qwJ9VHx<Y{?nJKwZYd
zuF04jO2#-Y4_VLMomE(3wD<yxAR~Gd44|*gk@Jw>1vf^fqzB4cx)tnTK33**@2&3Z
zYohHsK2xO1)~SdWV)@wpg`19`*WxYI+^2s~FIMTd2U)~L%M~iVQ*f0o;8mq1KdXGD
zM~rpUp)e=?N5$mXbvP7nD#YqVV%aV8#*u=0ti`?Gzlz~61E(=S(5)#EL*{ZcI&PNW
z5{BXlq<(dMo8hNV+-@(|YdTsH_g<Vmaduui2A|yLRaH?_EW(aDrDuadj7=_X89s;&
zZ<D!_Y>YwHd~m=eV#-kcCht)DY$$-fg*z<<P)f}*05ZL!6lVFG(PDi{3IrN%nwe?$
zK5~)uy=DPI+!VeZy7ZO8X+TvQ%Nk4tHQ2Uyb@4Ko@4i6l3u@y*<w#t|GaIyg;jJP{
zdSQJA_@`7DHY!9{|L(l2V*l}s*VC~&*PK`EY_xDYkmIBo`r#etUD#jXNAuqq=ak*o
z^Y(}D&T|U7+ze)vEL_YGCODX`imK9GqHw<#E=D&czQj_jqlwOxbmB#$jG6P-*M~uA
zfLHmSabLriIMa+*!J$;TP^1CvkY}LmW#WjGr$G&N@o4OEJd8(8iJ~Q3>@i<_KlJQ5
zLKJc(G4A3L^NTe9l+hZdgAhHxp#NIKMd#+`1fH}|jp7M<P|y>=BQ!BI>piti`0MvM
zJDLok8BLoIT&d3u_y9OT8!rAatfg9X^_uWcoSH)xlv<HM8w%8{FI@#F?mBpL8B0W8
zvS5uX-als4(%fKFJ8^dZr$oy&XbI9z{PQohkQuPq1Sln2XU8_lRDgXHK}OHgx|iKP
zO2KUXpE|x)(V{3@07ra>FH>rUypJ)BMDZC8PDwgs3j@n%ShjA=j}<O0st0-NBo_<f
zzjVAVmW?eW`LA(uQh*cOQ5sz1B!9an*s%Hen?V+y5y`p!6*zt)SksVeV<=xqU3EN{
z@lGT2Rc*8BeZ!k*9N8Q?#oYbI5W*Mga~;seHWd)xRXbhCQ)HG_?c?zx#M*$+DyQ)z
z(DCr^-{%&apEE%84%UyvGCrqm2(pUe!<b1=Ym*hNF~pF_m?37;3^=V?gvx9fyH~rQ
zL`!ywZ0s4={t>-WN9Y@9^b_u9fflH~P{?c1+{%L1QFvSulp#4?sR>^O5k4e(@5E~=
z$<OY1A`7Vvhu;>WPVwTQ<h|lFvUi^if=cD+B1)MxN1;*xC9U*X!q`_(c|YvuibBM@
zm#R222%6HG`O)ETb4?8}Y#SS7!||ds+Ww;?)=+4-r$0a2+x^R7myb4rs)bHnfDyWS
z*bBPBVyf?GRan4tuaabm%>VCGw8yuFd2-d#*V~A{r6#X&-65MN|7JE8TC*k*3oo}H
z&5<NHUQP!Y6iXOGH%uj2jiG2)+I}9E>{RDOzOo6~G4j8ELm9unx3puM0?xO#)=@7A
zodGY~*Hd*7dRMvo;g4=kw@ku^Gc0qwTP@$HxK_#l?$Jm((fkg-V_}eVtI)spFt@~#
z#91$KHJ!J_@89_;vnH<QDVMu|uKnH;Vy+2OsUOH<@f6?iyT;os=x}xarsVeT$C8_b
zXXVZ2{BQaz@Js8pPTs-Tz45n1@XGzil;ntC<9_usCKYcc@&js_<illS)YEas42*np
zjVdVi*3U@fNiwRWezR~S6AHZcw(Z_Ie(~a)e9tky`Tl*1xcNA^+9lwyAqU~+mckll
z4uMGtgt0iLYxIGLUf+5;C?e_rmcl-JRcZo+N~f5ILNsHAu3}4V1r!-mtllxJH!^G4
zP1mltI%DU}<Fevt>)!}$<;M?I_DA;eaIRhBU~#)HvuY`cl^!IyQ3WQ;W+dmt3^#p^
zp>nu;o7cJTHtNxm2tQ4^nf-qca8M43$xR>M2Eda7h^+vBj&*6Pm4sxv2@>hl%*!Ct
zDtzVqNPpXeFpXU`VOVdXdr}zb^8Hd~5tfDAjc(5Q`pvkzAh^W<kAydE+seFvd#k*y
ze+?0;(*1fkKjlU*W@}ci9()+CEPrrV+w@<T<utetv9c;I&B!8jM;%&`jk3Rn|MJJ(
zrZ^5NSMOlxXC)l>UM24?jAEK&H)2dprF5vdz06Ard{|Aj8}z5sn%jfp6skM(($4DC
zIhZ<O-j#>YM!?vCoy}`~N)reKsca(Bfg@$ex27cmoaxMIz1dAkQba(xmr%GEqfBjm
zex_2X-=hRH16iqD$_r9k(zI^j1Ls>!4vSc500Ua25gQ>BoTM>|f>Y;ut+D5Ob}9{q
zsSK;m?>^?_-0i$b)##HLsl{aN_Cu^-jB!HozL%Y(l^93XH!G6u8+hvdxSOTG|JX#!
zH5qb3zvpYDm%It%L5&v9DYjYT<pfK3)Zdoq>htRuy!~^7f)Jr>AUoG~Py(Pv-)(gs
z2miUPxR%=yq8hc3J)G;xbM|jCt)lw6>&ItU5JULZqZu_NRV9VRMsmpI=_2;W!1RHC
z0~nLi&rc=(Kr51fH2zW#&<n9;xo&-%g$gE`qwh@P{oEI8?kSr#FKXeL|0!vebFJBb
z^9=O}nV_)@ifYy#ythtm*8WS=c*vK)hE`B6xQWr9pazlCZy$-CN-~VhKfW*;6As5R
z*V?P2@wHL9Yw;_h4P74T+^KR^p^;}s?Mi!n;Ll6<nVFM{{KMtzb4L$676MmJ_UDi?
z;&EXyS`2*NjS6}C!j*%S6)to5muiKt$e@hs0FQTjKb5!7zO~MYjg$8UaM1D(t??ci
zu~=(2y3IOGTbb9!8&1+Te&+-4Hczns2If9hf|Vo8EE1q~wZor6{xoC%O;NBth1knZ
zwWH75=es325Hnw{pnA7r037E!#@;ZTs}QJGGSv@VS+%64el1G`W|&klRXX4}=rHW}
zKVGQ`<CSR^$FfQHi=`??%9R7JARU`~)ukV^;S6c4?GZ!VLaFqXb+rXVFg8=ROWKX@
z`^xLzJgGO)z^3ullPkw-H?8u3(PLC-CT(L8qdcHKbOlG6ZkBi;&8eO8a_Yo)Rwe$8
zCt`9_eaoa)y&*Cs2eO(=fdB~Sf5K)bZRO8J9Na8AtlJHLkQ3qT^{dr2%xX>FZBST%
zuiNG>Qq)Ud1ApQ;`K2gu7i;kP?!i)7wW4qA`p<VV=BH93cqZ4iI-f`~;)?6oNAE@+
zirhS*MC|r8H0324^WQ2GFVKLuMAD`Vph*3_NsdU5DdcjtX7J9B&NL_Gly8c`PsCmE
zM*P&=rHg8+`uW36;PaTOxDU{BPz|Wm<k#6cluxjvqnf4hS>(0^76Rk4%P98}J}HM;
zw$ie3(jjb&^u<b&UsdDdrvxDO{EV@}1_gO>XeLLilJVmiMFIbjTM6QEn%OyB+vbKh
zOJCcF<ER(?h_1T%C>)d#%o|P@DM0huGT#Ti>>Jzj6mP1Zvzt2kb_<u_MO~I2o5BK{
zK8-HU`nHP)3EaT;C5zs}%vGfC0Q115hmU}{6&g}Y3L@nh6T@Hq&Editz?8H!HX9CQ
zWo0JBQ-+s^`P>!5M9_7?kqJo&bD~HBk0$+l#;I`nB!6dKWmP4g^1w}eThB4);_;d1
z<-roO>f5eLm4ccdZ~aV=<$mR|hR|`H>!LdL<?aHMI~t2qp0hEJ+-;uJecKuErZedG
z&!7F#_HuKGbYSCSTcGjYt59DmNJJ{@?!Jv=zU0Zk4DByr8mU~X_L!BiEVXu9yNEw6
zGc#;)p|)mhpMNDxKU$pWZ0~pr#3VqYUfhZ39fX|>Mh^z49etljE2VL5x;Nhw;uqvO
zXr}F;s-}cpZds7Q?78j{Je(ZcTRZ0i)>ykVe-FD}e(!kdpE(k*H~;PJzt62{Kj_uo
z(G$Zo$7dCrTkmmoDC&Mj0Muh|@D~=6IBR>8CfG4n21O&8jZmX3I{mHFtVhrFE~+m^
z)J{i&wtFrU8ZHMov6mxbbvPdH{h?WbC56}g#JAR1zdot0ZCfNGt)O-hgfYe#qDiSu
zZi|p-S%ucTlUS+N{`Dk3DB#Ypwbrtpw&sthYBk*a!y5kfLf7z#Vv{)~O)~r=ND{d^
zEknKQu;vU1Y~`=%D}Ooo?Xh6$gP$*muVR+38ht3+;j(fL=nc)%ure?D#KwqGy#{kM
z4<M#vjjEd<(@Oz`8|L#zH$!DP($&zop_YHwRnrcI9zT?rTfNh~&@l2=^-yQ!NJ6<?
zzvA1ME}!_|5D)Tf_xA3@&X}hGBJSeMA+aRxvtRspK@o_9rI@!c)9)s$Rqc6pi_Z0`
zV)Y47G>iy`w(UU6Ngb2H?FG7RztruKnK56n#-QSB2(k1Wbd;ZM!mHFqxpD{;VG=1Z
zu7~R)YQExNv9BvY$w4EBAsKmsVVvp9FD069qCu$<R2gwYS3)OexeAEv_xBd2{S;#Y
z>UjoL%gcZn<f}kFJ%qq(J7d>>3(;UxLAY-u+Ucbp6I&!H!)`juF{S8N$!2~tLLCbL
zN5q7FL^`r2LZ9N<9X-^fcBPoY0-K>EMKeSMW#g65*5(84hHSaEirBVpSzzV_!VaNX
z>s^QFzmE52-b7G6X|-NCROwNS)wPjf(wZA35D4bA6qmh-dfldy@U5s`rGVgIrsr}w
zm}!HV!*d_hyguQV|F2H|*u>%_?1Gy(L=a@Kkvv)yyzfTw9joQI%iq60=xBT|@MwMc
z?@)YaP~Wewk?2g#{EBql_b%5<B7EZ<>ZmoG;_RWC+KsKvn>UJ7MU*?6ef*@`o2>2L
zNJ-(u!kNuGJH`dG6^i?Tfz}dhBNTIY)}M>t@<jOLjQ`@8fxSbmG%62|b%WHor-Jhc
zWSvr3$9u6?)4&+Sr#F_`0x{o{KCRwai`QPe|8vq~!fpIc+Faje1PBt9ZIn(1nusLO
zeIZ%_e9bX-GyH1Sn=Rt$UelGzeQ(5YbiSPyv!X$#yN-M`#+G*}-Bb9kjDE2N!;-=Y
z(7ZOJotsu)X2!2Rd(0pEziMtJX5hYSjoj6qf8;xwC7zcX^4NpjqL$xVyqAACk;O5J
z5oPI`94%nImZ^}fb8t<~1knZmjyGY7AI|!jF`SwDXWea9AOk-%(t1uRaTv81&Z<F$
zwquQa4Fm$Sb5LK1PdFY6GA`Vfa)eUD0nf;RYTUM3%cFJs6#rYQ*6gU^5EGA+c~+In
z-(rX@j>WRBdEPrR@(PDh2_UvwN=_Q-sF-*gKn=l*mzYp<v-7=q)3Jq~{eGv`(E=IS
zN&<_1_@EBm*aDOD&t=_YuKoF2-nbC)8!B{N0aw~A%tD2VdbyLu17O?)ro>)bLooo<
zVNdbeP`p_jHkPgU!-rXTzUDJW=xC&91Q!6;+HZ?whYV28V;6H@lZ^Rvq$#86j!sK!
zAeatqW||qolR@U8jbY+QaD%`Ty_iHX>DtM_yGo4iEyDJ-24zwhTSTm8Qq22QyEIKu
zf41&b8#idgXKN=f<iKdb<0pkjtGI1?yq|Gq`vsOs5Gv%v&+3L*x5yBMgL8g@Ghkxy
z*zD(WbBQpC=nPnt%2WKyIC45`POC7h&KpMl$aGZwXwiQ~xlODr+9sLhCayg9;aeTo
z-0+962&ufPVaqFw>>{@_yc9ck7GzjpQ^>~LPa2|e#%^N(kZB0aTgg(^5BxH&lyzsx
z5}V->L~v}q%3lU#Pu2M^Eueh!{5{Mu@N0(<9^#|IqWMwp(G6<pzA5Vw&5w+&J5Y&J
z7Gq}qy9UZpA454gIr&7&9a`dHsl06mhayXbNyQcw*&e@{{ih^3L*s^ox%=L9TwX{X
z9CQ+@ut&XJQ$KwOEx*gG#3VyMJeaOeSHgN*qVyY=3g08yUH&tUjbRQiOTW4T8X6oY
z48>Etf0$nb$M4=?W(=KY#Vj~uM2L<uPot2P`o-}6ARDm7H0Sx>g@VUYT&69Hl)FAa
zJE5xK@nX<snkUU-!^vxXr1djv&hl$1S4Y3wg9MNg7RB3{RSD$Z`00hJho5d?DgED-
z-21yf+*`jaJxe8;_Ph(&&f@LaKdF3l`GqvK`AFTu@kS@=XOlm4pf`<*O`t`)*e}I|
z=@q3q?~qUl41De%Xg_Zz6#j&j%?4>uA(An&yK6u5xbB3%$qT7m@WpYSFgZioja_GF
zfzmj#HWNtDb7EctbK~V^&q?p&OYGlMo2liDhuG0CcySRYiMx>Kp$B`V+nk2%b91Lp
zKva&SR#3ST0AI6h%V%|fI^dglXH89=c(#SQ>T$P70Gq~>Mmivf-V>*ZrUJy6?pbes
z_@$aEXAR_{0;Ip;VB!DLE`~rj{(B^!xFnn!$+RT(00BTynUT31Tj<R7un>bg2Dk~N
z+ZXNT;a=4@BJSb>jCG&ma`46rNqNcnlI^Kkb%O|+7H>vb5s(l}2QPJa)kom2&Q*th
zJS?&4zQ*p19|cTYw4v$dY(m1HiD6`}A7j4yBrqsc&?NMv$->Z)yPOOc>~{g=;9)$v
z(DICuQHjq8Wp+&<2P*)Jid4#?1Z~%C4qL}F@ke}9l1kJ7)HMsB>KuP~@npPLIH-ch
zCc?P!=$M#nxOal!nmj-%u@t6LigGfpO6LxKGYoi7hppXIdgPCAc96z}gae)GjX&B+
zkH&oUv#XY6s03#We`Bu@^qUV=Up!RwyctzozX`q_3geq9`-MGKRnxQA5vDj}2+;l#
z2p9<5F<UrOk>UH(B{}lh`T@+^jdS-!;!W?Xf}Fd?5fMbMQN1!`ZY<kGMX{i>bzhC>
zNNQISe+y;=<l_jbM1{l<l`z-&nEIri&f0_J#93z7lizNi=6_zH5h)@C@oEB-!AlNu
zwVPJPWae9{*n-DmdHx-iwhC)eUS_X1J3MlQm6Q7V`Y3&NQq@e+<pA@i)sZI37MB17
zHTu(_d)x2^Z?yf)J|~5-GuwARDe2RvKNL3CpWo>{=bH_mcS5Z1kr&*|@=-i05g(xg
z-%S}~;uNkn>#)aWOpNlEuh;FCCPBH}62vjaoJhqwtVF2d)zlcjM7TN^<^d+Yn>Xo3
zlIOQ~=O6HM{BV{P+g`Zjw|6=`T7M^JzHlT?kSzY432dNF2i1B`_dP={*M0fLc6WC>
za<&$OkLj_y6kOq1K~eYVh+`^C)oelVj{L&Hy{|j*GMrl0zL1iH`pwLogyzwqOdU}%
zJ4XY1Y@HX;agVRU@{5F#%Dt}VN8h1ul<q0U4oy#%c*GUp@y$z1lz+ME%!#Y7bFQxs
z2oR=J!ZtQ@@}4}wtELT?^KDy6JqLatzj3Q?A_S0`V;2`x(rPcx);E@-4eS=*9a9UG
zQ0t)hOaBuI-ur%jxOo1Z&9m#Jqvs$5Bn+CsR>`?;``NkSO&#9CkG0`OC<c4V#<ru~
z&876l<l{SUC6t}>9qj5yUq^pj>Hm7h{;7u+pV1l#Gq<bv+p3cVIQV!@1zJBce4b-f
zUB~ujwRpr@!YQH3fkD@fkyY0m9plAjg3men^khBn+Cm+!d{xVi&>Xct)0+9<tqLy!
zTRQvpI$rkLU&rIS*dg}12dy`aR2vL>U-)2-`A8gq<eB>v)A})vKf)aanHVZPWckr#
z_@?lI_UJzoCN87wtD%vK<q-XT7_mxLBod~*eV;NVVKI%NlE^7N22~+-H@ySw(xF6y
z0p&Rq$_(Kwoo-^yFJ(%8OEO><Q`y75G*4ms@p;Dfs~Lh~oybllkdtl$;?!}kOh8Vh
zEyhZ?NBoIP-_NsWu8ZM5EGf9+r^o4i691raIXoFb$~)mg0-#$_k0P`ch(<`isF9rt
z=W?p;m%{s=3-sqWGgo26kMc0>s71LOtzzyvPc$9SOl&`dh7fKAF{lz-yH8l5b1$Aj
zuTe?NK}@JiflLLUx5lHg*$QEX^ut{RpcxbmhB(vy`?U+(qih8vt?l{<G!m+EiR?Gj
z<1L?*f)Ya}%*vT5fswrLC}*$7ck}zm<yl(mJu99b`o@NQ({`@3Y*y*o^qnKmfpiz9
zNPJvJt0UUI7pAlcb`0_2#-5TmNV6#0eV(6R56KFYAw>?m<NUr?`(I@HfWS+6&Oc6R
zKgXuk{5OgFZ41v|dIofgxg#@5jApz+hAx_Bo%E8z61TpoQEEj>wTQdj^=yeu$<ruM
zR_nxGF7ypD(?GojX}>)`k1XKy_w4rFoc+!DKKDf%xRgGM;8ELwwWl3+0m6w2beRm?
zv{bT|&Mk5=H)u}1f{$CXf)-zH(=Qep)^1uf;O2k+IOU~)uP=aM;+q|Xq>4_P$r-Y_
zGUfo44C>ns{vm@s6hHk{be?0<0bvxTMVPXFA-M)QdV21i(;se&lPIt7VSqp3wzd?Y
z|D!cOmhx~`GGm$5_75tZ=455~Lnks(5&}=r?uc~79=mA_4sC4h#>9x!)*H3X)BP|S
z>66cDjN<T{v2eu865qsF@2b3tRrjU4OP*Ikh_+_>3@t)Rjrq$!EHvur-*sY$L7q5L
zTJP>}=j@Or3xFIUN&c}#`Ljt$x=+unmrd-x@1T2>u$nr3|GWVoE6nrtG_0cdOB)5K
z-;t4|$KOXJ)*mN4pv}BK;B$os12{R$850Mxq%GIDUwB+Y2m}()iseUt|17ey)pg2b
z$NH4zNWO){4S4el=xl;YTgkKfTp!W}Sg3TJ{&)_*10;((b$h>5Oz`g$9Sa)HD)4=~
zpqe2{#OD93&@WIMPrCA$U}NJ}dN>o{)KGzM)i$=+S4f153Dmz){QkS6UTmZU;xlgo
zuWO|qg<)zmpyeMo&425kjY%fOETf`)NX#!oX4^OTg?1J^RRR$#>fUyDgZ|m7gpM%W
zUL2_KGk3!S<+{A#cW>C?^qv;z;$VF3S7N<yVhcX!I*O;`i#8zNw!H18sox7!R{Ee~
zW7At<6H<X@+!+!NDvefngziu|TB-7`uLl2{v{(6-)3uNU6u3KyZ*uMU^WktrN+d!$
z;^Obbc}GoyUCtTF^X2NvqYaDN4)TP>hg^``14fNv4YN@fcPInk229v2Y~~>cK?)c6
z87tS@_tj2})2HN_S76NyVG<8J5(La5{g4U6gNarwFzHAScKoK$*zCryt>qu3(ej=>
zmJbOv-XraVC%hlV&8spFYMU32?|Yv8;Z$Web<g|t_g^2yslL4Nua8;vXq??O>}_+;
z{+}6?;(;h{Cvbvw>hIA}iuYvmoq#a1CzjwBYhbSKKxNTZZvq<tm*0()1EW=x?-l72
zBv;QVrqyYR$^DJ&+zqXOv*q5j!K9N&v#Ayeh8|zrv`{$q%gM>s%R9K+TD<n;d%L^U
zp9SYtl@CU4%Rro;pX~8B&vWuS8<iJ_KZ+=b%HZRXmEN%{Bz^l**^g(z^%?awSXZ{k
zWgih%B@88ySTZ*im5b?rQ`Vv_u}?pHNUjaK_>#Hg5G&u!;Bh6R=1Ufi`k80Q`|2FJ
z=`n9>5HwUqR*Hh9f0K(dmJ?t)uZa2AzgO_&ia&VAl)3BJaHYywe5oAFC1ZhHIe#+*
z@HX$(_&Qy@SiV0LrGVX$j<Ml1#x{2~lSN7gfpEc3vNROAC&~uOjuzW=`m1U~48H%1
zwJ*i_X#a=_36dC@xFk00$o$x{Ed1fY;kE_g=Vp0gD3IM{%<!rtUvkyYl#kJv=E-*S
zx|JISml)dw@?lt7IY%I>#WwY|P^%@7l4L$h7K9$_6ejOqjV=HHeyxXvYnV0v>NtA_
z7mP4sr!^jpux1euKu=p`nOMUS&K>olcE9tl0NWh~R|{DC^Vx2r%=r!Yos5D$x9ES{
z7iDQgSyEg-IPUC79o`HRu{vKa^k|TErX!+y`}F#U)${Dux-MScdx3A;+A3Q|lneKL
zTFhgTd&_wwmKU8YvUsu2{Z$mRMB-53=!E$sm^m^5aYtAPh+c;4tfeDQx}B5z-Kafo
zEbm%*i;F~zA9jhy5io6pn(o6_tZ_b$Ii66M7NY&r?LmD%yOG^osn+Ht!%xxLmx<;U
zwZBA~8Gbq_i8CtuW^4zIxeNeaQJu{&W53M0HFV!?4w|NymB%{Veu`{0e_Ed2XlUaW
z|5?#FL8Oi<@nsJ<DpHfmkVi-DK)JKC5wo^-hkL?`^InqbL5dxo&9yw=&8gM$=a-f-
zN@C=e&iIJ>1a}p}cUJA<FA2}7MraM1N6y7QYt1dXweYy$zei$*Jh#k#(-3{hUbybZ
zW4x-t!66fVAz!gGA_30753Wh79{idm;m_y0!f#Gzp4Sw9RqlD$X2YAk@?+*<$lne5
zg+gSS6^yEUU=~J)j++Iqrc-lW&;KmneSzmaKgvofx@F~Td_;MJ^k1>wD=_#KD4ip>
z-)~US+*DVRRhh(Waeg*Y8jztBC}Bg?|EQ8!+N$GLnZbY!C1<jvi}(C}*mLyX(Qfp%
z>QS27#c)>0<%2nkwiwzJ+=ty=nFN<s6TbKTu|8WleSh`k005Ti8dsh{19fKuJugSR
zcJqxB+yu0|<QcTyIO}@3AX<2niccVB-JL4a6k&ZO6^!06vGz9DVwrgc%&~I0XSU6%
zdcrR1)fbChcGmHCX9~DL5_+ELm0%BS3Z_?gydfPLW?1%)M;CdVw#1`6UXvI_3A)gZ
zdj9*b1#YHD)Alef|M=`&^*~^J6_j`HB}euwxZoWQW;e}M8f7mfckOj0lj#6RyI(0N
zM0Ig~MvkM?LnxF2O&m%mq=)L&jp=auY|I7mq6M(OaB@{3wuB{!H8c{Zo|<1Rz3bd(
z-%576C#l{-WPFCpXMjsp=#FCMPQ2>!gri4CRrbz(rDORk((9VX9Jpg4#y40P*T8Q@
zL<HhsF}m}F9NJE=kQeXt<;EPXZfIySP=W&>{J0oPWI2PS+etTD%7+}L4co2|{C3a-
zTWltKi7{iwhjen7S6)`M29pTjqrJgF#W6+?I^aBB?0#q!rdeJ$l$3V85}flfL)&pr
zD$zV|3$fWj@?EeMfy8}KtZb34*X=)l^UTpyX)=fzz&Md11n36|H&+7(!LgJTz>Ph{
zAyCiRj+<RJv)1MMkMp#pO3815HcQLP*u8f(6(j?TKn+fSCG-jH2(#Mrv+w6c3*EBk
zNtfr6zrTHb`FvZ;!h<&dmYrAu)Yj^HAEYA>vqA9VtM(r#Ygfg$DN5Y$L8PXxc`*3N
z?B1{Z5&UD=n{aQcBV8ShxH|D#+HoU0hmFPaal0!e->(M$l<mZE&TLNEh9Ca37z;6*
zw-jzuQagMpUq|w{#xy?EvaUayOgm1(p6mu4{wb0OZl5AK86e-c7R=9QJwD$fUGoe%
zX!A38M0{$>URBdJOV3-lSj2lDbFmkMrNoMyRY(*xM+yR|^aSFdLeAsX&9a!wty(O>
z5;0@8eYQN9A7)NkpS6;obG?JX+R5&d-RIt-RMj3!a8i8NhrHc=(((t#Ahhu&P@n{y
zK@$e1w}#PERmXQgD9=f=<b~TM_FnML--d3A0D4-mDEW)}=E}@2#04jT4^2o(Z+dZ8
z6c8UHN6F+6ND$F(B=QLun?b}P)O9vmRp^*iV`LNGyJDa|GQn>Md{`6)^Z*Vpf{0O<
zBx`3w1VlDlbPdZ>6s2TBA##kZf*NR`y4mXLUn19cp+T?n$K%9hOX(#=D<1pb<t8Un
z@4xcw3J9`HAYK7QG$Md#LumE@_{zJe5&7d@I#w1Id%f8B4<eDtNMA9kCmcY|E6@aZ
zosptrga(cFHCI{$){FHEqq^6v!18hHn_5q(jMT*dd{vevA?Be#s9j(f(+YZy>;+Ch
z&cv>o>+7)eL+;)vF1(9on3ajs#NWQJ0l|3bt%yNhL3jpx!Aln5t;DYa5oHO6vH7B!
z46G%D**i%wRQDiEP%yf~BfFIi@Aum8rVlmA1Q@0SmnLI{cKDBNohZ}r6O(%dQm+X*
zHcK^e3eXgA7g=t)I*6Atxax%rq;kY--$2Jgi67w%2yhO-IJ_#I4i7UKrz`N{%kE3b
z2Dd`Ym{8ZUiufTkw6K%2rgD(n#clcxq@7;~4>a`+wGl!)5ncwr+MK+6WA}_q1z>5U
zqvPtt);}|#+cc^MEe)XTarVHVYS|$-J!ToE?x(r>SJ@`)wrgj_I_*~xnEk1)^WDkP
z_?40e?^?TVwj4<bRO!06BCV77Re7~M<A?S)8&seF)j)oV{`Yrda+~A8Y3`eOhjmnO
zu7~_2!RGm5LkYE+waM|1=)TLpYUls<ZxS%yvFqz|@w+Qz@}J&b4bz8n8+ESMYITL)
z`J~z3c#882`HEyDU{p~poWNS1Q1FLm@7!w5Vrk-;<E7K2;Zl>0gtC*681&>A^~&Xw
zgwpRR@8!UbaM~NA&GWUCd;3dAj=AF|3b^!BVKm^NO(RuW{HxOO1*MO?KRB{n<5Z!I
z{qTmTUf(&b=53c+*U510+@zZi*7(Ul>1+Ajw8>c{$EFRv{NU8z=EIPQJD01Xo@TS=
zk(1nmurW?!_8Z!&&o9F(TDRxBCYgt#?LQk1m*9u}rSl9bR(xl<F`cBxJca&thd;tQ
zE1h{6B=F_1gHMhlF>tr=svMy!BkhLN*q9Gud~<U)k11T^%d@1Oz{`ts&$9vf?xS)^
z5vT3J`px)~&&2+0Y0hU?+Hp%UBH=yKCoqdz=|vdvb#lft#1Mqiy9>gDt)xE_zp8Y@
zmH@w+_OHtQ*3HT&KimDlj(CbOL#QwRV3nW`wS}ZPFU!b}_pW~8|MVk1HfKi+`^fEj
z$?TKi<A={CztyHov6lNPc6@T7ZIW1AVd6`Qn8|l5eV&H>klrG9e3Zq1Y<<nXfP+C@
z$0R*2RzMr{?R_`AY~T9(!m|3Ul$;&eI*mKF0k0h4>oYO8B-p~3<kwEQqk0vuF=|^<
z>G0N+@I<T1FFKzUGpnAq2XnYhf5To<kSWUH-<!7b8yiX>?+%X*MlwE$&E}KQ>=gK8
zBASjIUm{{_ms!7gPSoH=`#jdJuvkg<=Naq}$<PFF3@MKwkps#3jaqHP_iB#~6Xa-U
z#2BYm5$h)5F$%C|xYHv~E8^8iMZ4A4ifo#!81gws-gs|Nk&XPj!wL};7RN|t_yPWK
za3BXf^}?xsT&a8Ay*(YO{MvjLEW_02t27q*6UV`7Y>#Jl^Hsf!HF2v<WvS^s9daid
zR?yQ=uf<0mHl_)&j5#%ftDhy7a+*yJiA9W>ls7j|uA$d!1I^oP+B_EvTlW9`nCvT+
z32i=ocmuEaHe|Q8d2;syb{#*DAHQ3hQ7Wa=!~MooXC52KC=GFm*g(O|e>b^PJjez%
zRu-od0ksRvrOlN0qXd|scSSnZROz|*Z+7LUpMUfn|I@qouj=Q4Z@}MI!GA-5uN1uH
zB{+6_K?E6lH$tu<V<aP(Ywu?XFTL=sj$-EOO(KcyQICJAYk-pl)}z0a?P%pZPqIh4
zs3<+~Yc9KdJ%&^0QM<QxdHdB}^Y)De(U<`}FW-<KgDYumk1vmG8r}`uo|3%#KVD|=
z8T^In*ZIXlQZB*Pw$49*U{cl)yyAvE!k$fpM?bvtyXBeooc`?7DLDhXR)fiwH0!6k
z>q9A;`$rq|^vv5vSle_77=5i>@yJ?}<K+hS?>6+y%yZXrD)#vIv*rR<)~viPC$`W3
z%>3huo9*dP>uNFpEr$H7X#F#g^=Q~<xRpE;x4i1jLK1<E)@6_oH)7QF8L2`e88S4#
zSXunXF5m)JFOR$mh*78|)`gn_Ebn|58JXat<^yB06(366#tN64)i0~0y=l<$<rtzf
zTebi&sAra5X4IkUPd&s=g|CD@puh5=Kd1hRb*;)nDduy{j46>2A$!6<(VoDcF#^eV
zSnI_ljV=Ub*1SXnQYF!Ln6cjPPBmYf7ZR=H4pZl#(`%_J-VfveFFlYS>FM7--+5(A
zDsLhV3l-eY=Es_zvg9|z9wOni=`?UrOr0JK(<A_==}#A+wsu=87iO)j0@cWt6;?K*
z5!KbdM#coy&8pt<qH*mmC``31ke0=s3ZKqUkzLbKrXMoqc_pls10KW5pkVQa0X&#~
zM}r8jh761Gb&xyMN263}m;w5Ac*j-vRlWd6QI<#8+d-+(fK7M=1A{^$dV>)#uH)R#
z45({_4ZPd7zF9=gr-YV8A<|zdmL{`aWh((T{OfBqV`=iCWqDN>1aJEDlp@FRi^Ma)
z^R>#e_hnl2()0aFK(3A%zfXz_+^+yqLO&p<t=t%wK#U@L{N@JNV)XSl6t(xz;1hdY
zy*6x`-P&+hxZ8-M<n!mjXX^`HlfQr8H1Lrf;7fLzvw<1mGHP$C(Ye(<OIP4-75!XP
z@Bnd*_si6knOvBZC|_)T1U(+yV)}t6mPWyh=;B5-vO&VnY;l*UsHzkizQxQH^FAaQ
za>idA)lJu%bp)UM9je5hPj_F2^KzEI*iFD%1msn?9{fbT@LC8@3<ajqD$mT%RaaW5
zEG1z>*8X%c^Qr_tQ&r0)te&*a)Jkr9V&pBD>sgmRc0Vf-q<W=!ui0k0CVa!jW}4Z<
zBcnOOSoEB-2R&(ULu=jQHmd~ZeCPJU>A$3qz4`5s%Vmd0Z#OP3#GMVSILPfC&|D|1
zH3za}Z+YS=>kY4e3tdHGoOc+)$^4%Lgi6+AHNX259q(QCsGW`8S$HnN`FG+Z$g1dU
z`*N^n`@xaR`MtU|LDvcv7JM~LzDc0l^rJ2jc#~~+Wvkd4GwBAURcn0sXM6kJi$JC-
zLO{cFwIlvxul1~J#N2Ramzku6)1|Z}?0x37hHh`S%i$J0p5U~1e&(KG0!p+NwV3^E
zRV^aQ*y`_Ojb!AnMp^pS4nCy7Z$r+PkAALip_*EY#=o^6QX=`fcXkg$j++CvwO0>$
z|65f3+SAeH=jgV%epFZBUhE$|I8w6k)Onht6uXucuy8z3u-nXLbn&F&#V)!HtGa(6
z$`n6i-iMN1_t?;K;P6H;g-H)gZD<5xQ0~2A$dz=>$hc%e!xml91tp$Ot!F1g?f7rs
z9iru0U2X2~`C8<RrD$4gqjqi!d3|R~C7+Z8`FYCwPNwHbs2pAT>%C?^Bczqfy?R(l
zHB0=|<z<Vi_2n}%WuO}A_RAalVYDAUD_B8*dd_#hyzINf>3n;qCa3IB?X0+2+2X*%
z?%rFalip}Eb3d!QNaKC2Mdkt;@u8+K?>Clx^FQ%E%)K)w)7&%7!{csHiLsAQ7K~yM
z1Wq>bj!@&EF@C@wU^``*j|IG!t9Q1`1PQr?Vx9a$H;XRzN-L8>{!Tu|{$BuO2%Gni
zaR=9LJ{`u#KYfRm3(aJeY1-BnrBX1R)XL~p->>=+6u3$`%}qJJ_~O*nWQ(GjI;*FB
zFM5!sy;>GiVH8Zm=uO#ZTUqTuW7C#HKb}54`q5wi@cZBY(Ytp)JU^c^sY$zMbb(gN
z+BTD>ET+?`+P@KRzl?_coxi>J_{o3SqoebSi-)t>$sc_ADs%fc-+TYjlZ0Ar&Ey|_
z;g#vGdHIFC=dxuVt&i46>mSZ~K89X60k99fQYyyaoC}`Ia&qPBwZ+9<?GRC{^(h^G
z@WIn}@90l{;@4k!@wJB!4i~G7$#mAvrXl#T@BKKWF#0eGn1piSQ+G^{4$ptxI}fiP
zY`^xgmv7#@(e63sIE*XrhfD3JC{c$XtCcT`qL|&hJ-z<$@ssnD<8rq1*-w7Dp0?ll
z$&Xf)_HN#I>635z<?2dR9NxS8`0m|qx$0t%j7H(S)TlLxo;2;)FLtgJuYT;t>mR@R
z`p0iKZQT!x<O6Ub4%3~zcCvq0mJJv)jL8}c46Ut-qJH@B$=AN}jX(OMKYIV}2W44m
z?W%gRd*y|qD5gzqTwzL8H$`1fH8ccAU;JmEoNnEZ!{L{|bnxH&*MIn<pZe0C|9AiG
zfAZ?bo33xt@Y#R)U;L}{WAk7Am%sTt|IV+Cak0}rhql;{)<^53^$&k)Js(3KVz^WT
zCjb^9X646lb$|cC{Re(E6t)3CGGrXYM8CN6t-J64<l4c@w{E?FP@bL51x(v+HTBf{
z!H2<DvvKT&5~sjY0!;5e4G$0Rmet*@?P_<s+27lqPOBJ_w#XDil%}2t7I)r#NW#XL
zqAIRkfAQmQexj}?N9TXX!x&$E<+a-%`?w@&ti!-B-MG0^R;MSYi`8<uT#SA-P%4x)
zihY>Rrp2|r>07VAGTSyNpmphN>{o-(0AQT0rjxy<nvEf`Pp8LccYgWq-MjDKeec72
z_wJ8_2gIG7ePdl!HO3aMEUT()oGt3IXliAxL_EJ(0Qx`t-@ku;_s+k5^62jW@~{5g
zFaG}RfBp6Ee*W`&w_lk3%m4g;_Obcj{@4HT^Phj^{r9`y{KD*8-}%MKY52Fl_?g+2
z=dx)Zt&i46>mSN`K88LKv4|^*q3@MaO6klE*xufH_0^Yu{_{H|0nr$zh!Ut|tN<(@
zKmEzw_kY%u`=5XM)NkzU+<yPwy~ScRozBX7no{y%48st^fWSi^0})6divBzuoz8#p
z?h&#t3*6nF&bB61U5$RO0QdH89vvT9W3FGldF$2<rOcBjN8kL;H<!!hzxCPQ*xx@$
zKKkXN5W>JO+_?F|&6_?ZpN7%*UAHW(aYow0G?gi|l-hr3GUSaP3$0yQ)+fjPMIT>!
z?RK})U;XO;^S!_R!F%t%dwhCsZCw@>qIRw+i&<{8E9$B!in6SgGp=yca$=l0yI7o^
z9Zs6Y`|jO4KU<zZF%Vp-UAgn&!#l$mKl^J3$0y_c53wrc@bTe4{wJTk^|AlAy}Q|M
ztB3+PK68I&=6>0Zd!00`ijtC4LPZ53K%(Lakl;NKYZfdJ55fz;Qy_M%5N#J!ZTbOK
za0s^RkJ#7u&fJ+fhegVQ9Uoa#?)kmKNdLu1=bW=_5N7io-|mhlgFOyCW9;9!ps$<1
zR#hmah(K#o6a@e{+a4YsCQ0(};gNGzq5)8ZS^<AVBq3dvi9o}j?oE@Vyms~4^~0+|
z==Hh{P0Tu-t+KBqwazs*)XqlVVsN2t0MR$Kq~L2ht6Fh90~MqcWxqe2VVdd~=;@Q=
z(dooGJN#|<;Mt?JR~VxV(oVmmqF78uUdX0(Qc7*iPH!hk5@XVCk%F*ULg;Nb%O;aa
zQ`3KFd=^sr{g2t{a6TNa<}15ueykePwkFM0qV-nyO3~{pWt1`sl-5cqr4y4RNuK9v
znguCb8>Wl84Q<`jb-lQM|A*=PnO2zVcB@={@zw3ezl^{A=3$zpg!O!0_qVtA_YXj(
zgYDAE41E<`B&51-F67A=W3P^j3H|xpjrV^}2oZdHK2V^P>h?N_xWB)j72Ugc?m6o)
zhD>)tK<B&xP0%8_Ko&HfoK7D-y_2QoTem*=@crAP@$;jjBL!+u4*JT}^}4RtZDWJ?
zl7vla+ceI%05QgZBBBq~l;mWEp-~EhcRtue>DJo_+Y<F;<gu>m4h7rHrc0AZtrdR;
zU*vh?K?npz0_)bU>Sj?bSIcER9-mnoDjQZ!BxyQD8KV|DgH-E0*Ll+I6h@gWEp)1d
zge=X8gfZ$;d8yOs=tNoPXV0fA>j&H0)-RtOKUl4&i^X_4I^OE{2wN9UKL7Hgz1I#t
z`}FI8ds$}lwCwisx8Hm7^7f70D_MVa0@lSu=$%Wwlhf+O%?!pE`^WyaE8{==b@+JQ
zR3QW+#2B+IlTs?JV~iN1wGKenbv>QUoNXxxAwWt%3<4rzD*=Si3Jj5o5uC%K=<i;+
zT9##1RrA>lk<v6(3MihBk_YdDAR>&2A!0-#iYU&Hz?|O|rHB|~@Bn}^Mu~suqeP)W
zDv&V*2?8-lK?*bqLLm{QKoaL0orokDAq+r55`<7vDgZ4cC?lmJ5<-CyL?9KBFsXqT
zqaOwUvOM+PH%&uAq*?A-zgn%Ftx&YqE}dP+-~p5Z+}({Ie{%Et@9vF9Cd~#qrFU+b
zz1Oeb{dv^QFTZ;WPoIay(%xje0gs<#jn#(-2baq?{{NFP#$Fv46MDuNV=t|X8M9=J
zG4|5p&@;x^hU3sP#@L4A&@;x^hU3sP#@L4A&@;x^hU3sP#@L4A&@;x^hWisl7?$k<
S-}`F-0000<MNUMnLSTZC(&^~{

delta 368645
zcmYhiWmsEX(DxghA_WSCAT3ggJG8jF6bc23TcEgG(Tx-j?(SBKyK8ZG2v(#4f)sZ+
z+|PTS_ndslb>-S0*2>KPH#2Ky?eI(D{lCO9l0ZVCtSftqdk}l}O0lzwN*oWFWq~T$
z588Lsgi;*ci~m3o@SrJ^c5~st2d$S54`O`X$Tm&=mly5x+$rWGBwmi0kx&wOsuw)N
z(-x;nDsYF><j$YV$I;wmBPa|xT@qyg;BxhKy;R?I|7*SV!kWgMoK#iVCmR#O^zSDD
z4?qd?|Nio=F5Ts=etxtm9YvCx_mm9WBqJ(2A7%9;KDAALPihM>{(i7n&GRX{m&tY$
zt&8qh1!E#I@97>cYRHt>!Tazf`*wD8bQCZ41m|%#`f;~5`{985=jeE*;QGzDyGTBZ
zWF74AYv=8pIKH0m%?_T(H5CEy&~JY)Q@hl*5rExsw|bm)amiTd+Pb>Xe&iY{v-Qc|
zbEX<HfF6h>5=Cy`#`$UH`MX-DykqlqrIxO)SJAX@sQs1q#_#Ss{NwKff^J7EH<`!Y
za2BhU)gzUMSrrYYMBa{*$iVIm#+{7cv)a+-u0PG04nrQ@CRhy;+Oh%Zb8g!kdddya
z)00OE-Se}vv*#t#nG%O@;Mt-F1M+cgr-_v|JY<wNgR-EUnZ34?Nc&%!WtE<DWyB;T
z*;2EQ4|j%l%L^UHKF5MCyOZ%GEiEm#wW4>mW8cn#$?X<9fhT?U)r-fgPD+qn+ett7
zr~k36iuO6>;~z@kEuX_inc;&~y^hb*-TKqb1n?MPc)yw4ab9S6TWBcYak{N-Ke2s#
zVfb|6)Vy-&R@t-}o_!sb%@_X&>wJJ=V`CfM41bM!D|{QK;=Nft@V52Yl$Xcr&+9%$
z`<~}Fu6ZK<GlAf78}~>Jp3ko2g!*zk?;v(GrS%#Tu#x?AlP#Dz{_nW9<fe~a#PR<6
zgdMmUadvj5eEb(OVd{~llzH)8v)Hp}I;&0G;CoxjchQJl%Gp!V(9-|48}lu@_aE$X
zP3xz}haIPz=%?#wq?|dc_CL7x*Z)F9y+>FZh3^0LcHXsidhO4a-W_+YHymEI!G%?T
zQQfjRyXKuF?&xQC|L3}9c-CD)(DSCt*B>t%ua4J6g@qrAf!kv9>8EX#M^gx7t0>*>
z)aPlxtnzU!Z-UA3>S&en>3ZPm5Z35^yvBj^tgy=EhKDHkwk-(d-QxNA`TtmO14DF|
zZ+zlmWn$IqxE=V10XWb-0K&%}6YKY5>!{-R11CqTt#}lClZcD8)1=1x`-vxSaIpKD
z2mHf=Uai;F5xHH{pQF}e-)Bp_C1<Db{pF!Da~?=}zhN$NmilK)d-lIS`JeF&Z>N(x
zo_F#1m$%b{ZehlB|JOjRz#Mnt<*ZU$)pyPlU*ajzk{tm53!oy>1)OhXW?_3DIcH-B
z_=Q3_8&}<6y6d-^1CG9b&3*sIXWtjt!YtFiY5(OL&wg0SZa*+%XHHbmZL}+2K+IWP
zD9O;=Ms!|B+&%A&WoY49ZDYLa_uGX}&zkuFI#{kpyraC4YlQjajkX?lq&q*RPIw20
zjx*h2Bz^}bgs*tWA}4$=Ywdl{(&_2wj-DN`&~aJ+)zIfc+dezRZ1MN{MTNVM{VH4Q
z0oQ`-aogEAyU(5?<#iXvTd#v^V-?RaYXj$*PrL+E#i9>;si~>YfDMsvqxD7vUg<vj
zcI(+Y{r%`5?+1jTXhhmo`>{`}`*8=NqV@6!SZ%e1HPCBr>~XvPJ-KuH4^G)AlK02!
z$-Bp})Ai*4t1;d4|D0^!vR9;2?)HtCh6V(#C<b+1J-Bz?t<o!JSk&v76MvFi`MwpI
ztmia@r@9MUyKI2jclsQ!wLi;XVD0`dZ?^nI&B5xi@0Fs#X<WK}#|>~t{LudXKea6w
z0uS~N^Y#y$?2jhTH0~)Hdfx0LcOJP`re|l1A`S_icRXH#q*aAD9qsJwEG;dc#eVF$
z(CT~h94S9kdW;?4k01YM?Pu!woVMkf_1>~!=P~@8vw8*&Z%Sc;`B^Ye3&6vI;a$BU
zIVmZU-$Ezwcww%3R@Ql2R;adjQ3?2D!7n@?!d`zDc&6^UJ#(K_{*GE_&*OJQ*i~()
zZ(Pz-hm!CO=qWb(3a&LJ`mDFR%fr^2zv&AL3n7diS<f&(-IO=3r4{XIeARc|N&Nah
zQDSp*6ZVW#x?Rg&n1_AqVM8(D_Pv4kt|UNt+e6av@2>OduJa#ur^7Ry+<TM{yRvwc
zBKM^#K2{vyZI8!x>1>V7<j(t@|CsH(N$>QS)A%ZUxd?e)Vy>~{qWsy+4-GIo*t1{%
zW6i1&*0|y{@+{(cm|g2(MeC73{G(Url~<+r4s++jp*^M0_esZ$tBxz6{i}h;&KviY
zlPloq>OaRNel@tMuk<;*gxNp+hrQ3;rSKO^Q7LBO%IoUC5B~*CVq)UAp1ahb>lL_&
zJMjniRo4Yv-j#p<g`KFaw)-7+(RhR<m+SnNcIJMn!vD{HkK(LPfhnG+NRRoH#}mG)
zVk&^vu+K%A;c9~|OqxXB<*<IiZnfp&89CEZ^_J~8MXUPd0g+2i8Uey(-(K>ys;a7g
zeP&<lbsh#~MVykhi)$0Sc5_4iBLmjZXtQ)$_>2a0l>TLb3UE7P{`y&49nX3fKD)^j
zbbpEa!>ay2O;HNC?EY#rcUtv61I8{qg>ibVE8E@HJAC?6jBk5-dR!Ou;Vnmn$p*Jw
zD0mctZU@86mFGgoEr^Qr^mN6z?L&LOt68t(a@jg0(L-k17aW&`(Z=_ka&z?6Ba794
zso-{@vqQ7&;auhE+=zqk?523=|E9jwsk62}(nELJQcpQaPk{LUD<}7YF=qUa|0n%N
zlKk;6{htzHv5)Zos|sG8<X3dj`L@MgCGj5JwElE?+kV_}T#=#oLC9jhWz`OdiH6Px
zG4v?^#HIf>d_gs$o>fa*NW%=mG9IF392Q@)_<SR#P#1o!@zvn69OEdyFdPk#;a(#x
zG0iEZ>aF(cssCaA!f5$(Farf1`PRiwP?G8VsFaVbYoRGxat8ODfQJjrw3mq0aGxhn
zPW@P<W{r{WD-O$BN2(a(Uo;D&2F?W86wK<V$bRxtROPWT)DV!Rs@ygXsXr3v8@?Q&
zq;wY;86RI30>*bDB|2$RP!<NR7Yh)Xori7qK6l;XI<vBfK)1>nb|*7ZNvY+UASRp`
zN4>u-M@BIfjf8uA{)GhUIu--Y({&afhyBY_<Wg-r4}n&*Z@2d(eq*$eZ+#EOW!f!u
zE7XVv0=sh}p>;3ff06gMU(8$i^OL-aQTa6XGjI1&APdfrKO;0J<J=8kCk9Rprob^I
zS{k-~YB~0Uv|Cq|HwT?<XiTBn)6_JUFBpGN_wpaRdj-o+lzH_`#Os-9A?v<x(X@{m
zvwK6Hjv3%rO-X*nGg9LVe%ZqjzR9A1C*t;RF}8OQjAX&J4F1#)#?3*ZoRGu;Ax#P$
zec{!?W+ctJL8YSF3dICMQRV0sI!~E9r!9E03h2_qeF-P9JQm9?laFp{J6V<NqP5G~
zue*BD1e6nsqA92tJxXa7h+7KdBq?}kz?QJJQz!e$p};tMe*E7ATMhDG?7Ef4dK?~O
zo29i+NBjw8EGx}h5UTGPxYKv&&M0iueHI8Q`g$=`a<|7~#jJs?#Ekrb(P~J0;qQXY
zWeGgCm2o$yE{VSP_)vWD-{eHzn!_o=f$123mjIcnswr*G>$s-}dbqKFWu=U^6%Kw5
zH7asHMLD)&RSyxPWIz8xIJsD_QY;eam0Z6@5?Q31K?0A{cpQ%}$I&P8HQ_a{gI4!n
zuD7M#Q<r_Cv!g)Z)mr@&EA&rfY#JRdC<)azj%Uhou!pBS61Dyno_A8Zjfu(S_SBY=
z3ateWk=qhBMrQHmpJsnffpIUIZ$gP>%(fb)utwV{I$dpD;Sa$1;tC|NiTmR!EqVkq
zf<8Gq+j%_dV$HKUmOlzw%LkQ|2*hCttGCLZK{syhFa*MuSM3_HyJ}drGGIN(0ja4=
z2Fr%p+D>mUNYgV?O0Mf~cDAs-57{d~%zX;KEPre;^K^1Uj5}M6y?`|srnr4q7L!Uc
z)A*T~qZie)yO;~XKEuX%>%qg0WjrFmTo_5iC4<Bu#?cSLc`b}$<c9--a)u6qw=&*|
zyjA<q#|Jn?fzT&&0$4_}+PMCY^SKrx44>E#F5M_o3$tq+Y65I!>c1S`>mcart4~tj
ze9{!FByn37CT9tlq%CC)l#q^cz#UdDHgprlkXmByyr(yO%6z(9_s9rJRe)&p1#`Zo
zGycu$kwiOsH<Uv{vNgvDLJgKyRcG1>N}L0lD&v17TKj5l+nIS+ht~33--=SOv}Qq8
zoH`l`Kem?$`j|wVsU9^~dJHg=v|2Mqzggj;DVJl<z=6Yz?Q%XB4JJ}v?}X#Jy2HvU
zzh&%<l!tU&zph9*<cH!gy~7t88IQ=YcHJ9)1Gh{6zPk~atJbdI^SC>7B2x~kVF?BD
z>M`ExmCfqaWkkAV>?p{M*8Jj?v%avhtI?ZTi?|$}coU=fT8ns*B}_0fZ6k)yz{AX}
zYi88Foh0}ZTi@s<YsHqkV65&*JmzchLV?ZXhn@QO<NAeyQC~J2<`|l$Y#bfu+*opu
zRk`{TrU$!TagCfQSHhdk7^~!JSQNhknn|o9WR#Td6e<oFpoNbu+L|$cf{@F@otmc*
zc;OT%jkI5~e*DgPW1qR`_$O<mG1Ix?5KYccrOGOsFk>nFG(%AYWjuz{PTp9KQU5wC
zz0PDL&L|X02i~_s<}2SF>Z)RcW^0~9Im<s-_JYaZex{;<V1hz<a;iUzn0@#Nyndj|
zdhL=SB@557xtJM@S>YFucMFc%(7c(PdOhLV<Cc-uVVlC^4U>cX<S*LvE?qD@k|jOy
zW#0?&P-QrG8A`lkYjdgL)MfQRDL?x`yM%+=JnCjEmbrt5l_Zn8z4Ef}(iT#jq#GV~
zVb`k0WNpo%o!SIH$~P0$THVD0G+9ml@a-7Hh3QxtRn^zTGA5E`F*;!RZWwl~9hzCS
zUz{C#dJM5X?(g1LUqC&wJ~y+bRMxh<pTy%sk)d+d=TtLqF@^sAj5Wd#sldz1`*r`(
z>*e?ad*rA2{-4D39kvKJ6a{0s&vIX{Kk$Y{CI@h(P6t(GN+^bSke8cq0HKP7Ms`OQ
zOU`$HjEdACrZQw(+WP~>*;J)bAL`~kEfDSaR2%qk$AqAjo)%qk<z}X7mg+Z^29%1f
z)pc!b6d|9%4B<@!YE>-QSol3y0ZzG2AcfO5-q3+`x1~y6*?#d@BIQE8GL3YJK|a+C
zdA5{DH97hbHM^k#4#c4dK$^_VyZ<-Pj!S^cV`y!}Jqti0q{vdyzZWttiG2Y@diUO*
z$Lcx0eDXS^iE!4JQx5)&8~Gg>MAlM(LruWxc02MC*F(xQco1Cpfd(03vfD~3;E^SB
zwWxlc{$2BPFpc6bu-5c9T<1!488&qJ7N<@1dbQFI!`hk-adQe(z?lhRcX22A=<f7u
zhbFFqF}V6O{9*KCi@F<sez<k}*_i#>?gh$%{FXXdNmHTQ)G@7mUZO1u7%ljPq7_$r
zRWXjP&H?XGsm7Z`bJxl=yxW<vI5U|3xLnNnfdRAWipm$X0{*S{djs-X<q=DJLl0M?
z3TTZ`=glm$kU|6kASielcWFESE0Hyw;?qh!hhn{{eDRiG7Nb+xFN$k<BE=C>+7}Qh
z^oN|h#KA8b1wM?V5D^YGy<99TZYYMZtX2}kiz649rMP7~r<;>qq!^|#c6WZs*HUN_
zg}G}v0=9O$Oz<sms?_kbWM-`yGCzr3u$wq<eR!evS$kCrm~gh`lwt_aX!29~4`4In
zluHatH3kTkq-4Xpt%9;hT}BO3h$Nx*>t!&qQGeI%{_+U~`=^n%RgbtPTdO>otg3wh
zXNxP0c-{Hv`fuZx%r}CbOrO3?x>dg3&$82`t5|U#I}#>wQUr4uVHwRkKl)54y*<D7
zUxR<X{!9np`fy4lC=`|soy>i?NK8yDN#+<QAf+LN7=e*Ig)YUU`oTFf03NOf`XE0B
zsQiODPF5Agl#ZyG*~kAu7rlg}=stGH7Teuh_Nr!uEy)vMftIMi&v2tk;bHb6;Pu3x
z_(aFI-MpQS$M0l`&5Qb1getgGN?`+nJ`NNxfJy&pI=Pv+QJ=rFO|5j!@apGReV^Zo
zzB|27X#DR?{Sn*RHh09F?_+aKO1XM!<##er1(JTBkB$y*%hlDb9M(f*Hv3Loi6lH}
z5IfxUM(yovg(oOwEB|2Sb0<S|mRbfjK6|n8V{xqsWVXinq)lvO!}mGAPURXf+V9@J
z0~GsYlg=J<_SI9A85Jq~b>RN2IMpnQ6fi7|oBPw{hNZu|3`HK<euW-Ik>U|e%=i|z
zu5Y_ZNlPy!+S_W72EjDQ{eeveW#x?+^;prhqXdPt3O)ntEw&?@`()IJORs+tOEn5R
zie=8Q%rlAzGe^<Uyu3j_YQoUCP_WeqKwEMqWyVNNkW#WRarWKDd(>J$N<jXVM9rZ>
zvyC%dEsLeCO`Rt3ex&LI^OABi2c}d&wL#;|@ddVYUku^b*qS;+Df|jo{d0Ejz29DJ
zZCZ;j$>>r+sGqR#!5}0g&YoO1Nn}(p3M&m7DISnt*IwQ7+eb|)KR>_X_Zj0hKnw_k
zwD^zrW#ZXCl;SgPDgo!x-kE`aXRla592U=~kEBCbi(^x4-|u@pq1|)|%OKMjIV#D*
zw-cBkR2=@UN%8#9(BS?K&J5~zsA%X$yB+4<GkNQMXV4r?c^Rf?sfQ2gY465DyxM0u
zNYrD+db(=2=c~gC6EBWVg9jJ)0X=hX8!rm;9dX1v`*yp#inAFzL<dH+@4@jg-_>-)
z`u#F55#t+C?~S&pzvKpI{mJdyX$<o8X^|GTpE%=)<C6-*)sf-Iny|wX{My<t#3Rrs
zKjRAS-i*|xlfs}meL>meWZ3HkYy|^e1?)VYF;cAUW!Hm?$Oh_DACiLi0E+t`^@D|b
z9P~KqSj)<*haB(w>tL=S9%DVz=5JF}?y7l{^;RFS?x6W<R-SH7mX^}!79}Nf3TS=w
z%JrV94^h0Gn|-m8vdXOaE}wTQ%;1&+I;`Txm2O-wmkIIth#8%J6gvM6Slq1}S)(PI
z@7J-3Nv{IY4V$&<vB52q0Q28pxKH$`CThKo?=%}>PPT?5FY>-kTR?hf%RZ3#);HHB
zm<8=$Y-Pf2-1XgEUHREJ6BD%<g+=)JJFEP_j2hs;h8?6W$2e}gxpYV{2acDLAEHH}
zaFAi5#uqRe>Z(RA;$!4h&Jk=pSUUQB$JV-Ws<>kHBFm_et83;KAZ-?>5~CRdfkHDc
zT~KJKsp8N<h4rQquhEdGw=%eqyK--R^yrWX$UwneAbea55G|FM6e=nR71eW|kQFlO
zm{DC_?Hqsq&p+@iXn2Om52fahiwTtT_FPX|N+LA35b$)%eM-sLaI_Z@@-2?xMBiW>
zZ?|K*{UpTv>%0ZHVjEo;c^99o@3^^3@C%i1w@+K=R3XVal_p)Z`9R%AMY)0n7u_eR
zEJ_NozeKvG|Fu1M4v|IKRX?-!1Rs0)x^D5uOCT`%DLFcK%$8ap(*kiYdNTKc5tV?*
zgW%f#T{A+v>>;V#(Yw#uJ-_6$S?BDpx%{<Ygwh9DAzD{p>i3L1_<i8uJQG8ry!F60
zvx%49$t<tbl%u2H3Rm06i-_!z6m^;@YAaZ~*26L={t&bK!)z~z|9XmlFbYWL`P-5C
z`gYsN%fzLCK_E-<C2s>NtidbuaeMnx)7)ode(F-;DFS#n)@_UqveelepFcS~jFkSy
zGP|-pjig!+yx^iT<KqhH-knW>rw#Rwt3&8}81kK(Y0csa2_Ojt9>fG7z<7};iD6;2
zuzm|NrrJGJlE6)rgjRs=2t$K3PqW`GeiLujVN3Qht~9-A7`O{uquS@|s3MdlTTmMY
z7j|)sD@N{jB<r=eH96cJ8dikVr>2rhniBa_uQvds+IcZCf=r3>^lGtbtIZERBW%{n
zk>sF9%4*r(#n1BA$hiHVv*4CWKPRQX)gk5N1}ZxI;j7@Jdznu&a@+2k_Pzt&>Tlj!
zaeNn=@~!ketJEwMM|K<yg{{EM&zX&YltbJPibHKdBeBtSU8|*qJI4!JY|t&A599cR
zZTSvRL%KaTb((LOdUDuvkPxF1!RbP#*(%l-D9#k}J6Cs}DteA4wE^EaNZj!*FkFy@
z*z#$IzAOuf2&lz|M3oP#mD}@F6wrMVh7Fk|rOxmEI9__$*TCN(5~>;HioxD`*<#QM
z|G=ajcd+c^YgxdEI&!CPI%#KXksr=i;Aai!6ES}7W~s2Q5ENFO(C8qBSA6A?UfDN)
z-`G~yz(x_<Cr>~Qt45Bp%;HXB>WyUk{#kg*O4hN>3^wnDM^TzAd+(%aD-chZZ+$j0
zrt?9iH}XQnL0pEjcS!410L`cOTH1N&fjv|UE!@I8hcNd?lq(Sq0fFkOx^0Zd+glx=
zT+>8JQ9Gwv<&B-<I^N9rqEQ%%irTKrt#rlN3d0t%Q~&!IdF;e(au_{Rl69tzi1r&w
z#_#=^i^Pkz_DU)fzbfkIzrvQVFQSi7;T6Yqf{GHK3lB{9x|}IS$`CwXf^-aZFhE4V
z^v?3?FmQNEX+DL*#WgbY_zHONL#axDeU6Ac@z0FErQ`No>f5!f5++t%*}!y#>6VaF
zD5srnXpmy+C{6tNoqVOJ$Vb<y!erZ}YFr%7Z^7Ivlhi0BiQ|ll(^c!ob<+ldWap`j
zVuhxIU@j>r-k7~z9pfo7R5b`0f-a@xV5Fp@h9xgG7~}jhF4r^;7OE!~)>97%87_D1
zBsSIAjKoHbp<x(Fo!clUzkJ<p3mY2>b9OUyxL%x!RF;jEn8Xz8>jqUlnI;h;(O^=?
z5J-weDwNJ#iqvo?FM>ekD&_q_bkjPGALo;jlQq94&CsHA=~#aXg;G07y&P1YcqfM{
zHORwhN`VB%N2YQDAy0#Xk&pmVt~Ba$c6My)TA@Z3>=?&xQj)>NEx|`p$g5at4k=SP
zKO;W^5a(d{_Up<vgLIQ8;X7&#=>#gtNV!3LKNL<WF{Hs7nhK4~uu>+IsH6mme#eaW
zrSA)Eb=5GVQs`qjX$uZ6(GwLkwA@-}E`;dG&aujCSjra2AzNDyk4C^G<#&nOtZZ6F
z5}Hrb3XBf6>Ck;QoeXEX!g<rdZ3XWBZm=Rp`KKfjiD8xzP-={tXH>#z-wEMXK14c+
z_vc*KnacRs9C@NRUua&5I-738dlu$!8k8RavZ~uHc1dPQX4(xJ<?6|Q;i~CQ_<qz7
zOmK^T2=uk1X88gbjtbx>C>+`q!w!!YtGOwDSSD3e+B<af*U*N`kVW#bl0_0sG%RRp
zl3jR~tu9-X8#>NS-2Yp>4C4S_nlLI@iXIa?Zt#B`_+9vZST+wFCr&cuqN`oFpKO(-
zp_0$>p$}bdU|zy~V6BJznEepA7QAXd3|}BC{5b#B{YYe@qz3>~^i4@(yrdBIz{1sf
z)xEt_gIFTA{7qV``g#f^oU4kB)MgYXNKW2fbIkn>&cql_GwM2X#**tcIxccL*sI&9
zRM$IMZ7X3?V%|Rxoh8%^-*o;uFSz_eS=&sylm949lZ;-wh-$*McFCf?RmPI2ZSckC
zSFc+JZ9*RI7CL}3r`PIW$y_HaR~`F={ki>1*0Dgm3?rQ?b*nA5#t1%n*C;YRBsn^M
z43z#DRfX4bj>-t?beq|gwDgk27YAW`akG17UX?csz@tvu&!N95tl4*b+gE4==bjI=
zpM)n!e@Mr^PdZ$z$oCXB>T8OM?@MxPHxU1ZQ}R~};&=mm&sAfip1|->*3)7!4=k}>
z(oDL-&X*sKj|eIAbsafSUm4Lov{P1hQ}lmbqE(6$p$-pJD*U~x<+EOR$JSv}S;hJ?
zwAel0LW7oR3_Z||YxnozhxlN+NHm6!SSnl@>KwIYKRR3;5*_ZMF1l@~<+KdBEsup*
zFtN5Jdp!$~?#%W>%u(Ry5E#Rl;2)|sWpka+w<XhN@j|P_fE-R5n+XsV2Neq6%Hbm)
zKK|p@bk#9pm{LFs42P-~>X2oE17$fU5{Qbos-<!)o9qM+`Dp~+BuazCg5~rSXr`Af
zAcOc@P-?Hush{-zgX+WsQ;o;>^O@@%`Kvk;@hJc&6IOmIXq>+`Mi+;V+FXxJ{fE)#
zS6E{HV&1#!COBb(3<VN7=#v%+!!wX*p?yAOQHpDEDwcG`2*vLAvQqtMkhol_gbc-g
zJ`S4gA3dPUErZg(nO=K0LCR12#~mdtv1J_WSS}P}U<^8OZd5T0aQyV>D0&omw955r
z2|ZA_<7H4hSE0@nsgw9F9$K%N>|AIy7fU-9r_OBK!_d5O@0?e%VAa^Vy0+GyymB(u
z`IbDLqo2sJ;B|R<quaq^X}w$1V7!H=$-&?s*|si~X$n?rSX5>R9_0cm@c>Qa-9z`h
zElhYq)c0Uf&Y6WFvB`RDZ0xtH%e346&OT5RV^jh=1b_Bo>#&$K7<px$zxs35$}$Pr
zJPP5~y6V(SB(0hSf1b7JYeU0%uqn(Z6>4dp=WY?qBNwr;44+bH7S5*I98g4+<DJ?Y
zo<A*@e#bY6@<<8B@f*4NhK4q$f3d!KMpsAJvWR-|#&q0Y%9XEwQ+GM%@7B}B1G=gi
zPBFL(s{+gSVT$Y%`A30xqAdxod9&M~9=ET8xmD6ba7f}SYghKpgSgc#7!jXz0yMLA
z!goVNlVLjz1fl}TONp!ZSFIlRTOv!1`caRuI1UKk-B}uxFHF;X6obT%xq0>C1@MLI
z`<+d6Zj-OyBtZ(ouo!J{>NP?ElNj>yPyO#3rx~KoSgI6?qbPiqc1T?Fk+29VDh#}*
zD!xzJMSWaeO>CK9Og1KaVO+z&jF!mrF40_h$r$=xo}!;IK%;<k+Cj8V(r7y4#h*ns
zM4g~~iifW>N$D^)HP|eTyNrtT!ljsm5g8WueveAB|AR#&YF`sq3|KuF!1qTjia~pq
z$iR+eJFjCqdb+FoJj8YOO`<=^VJ;`xWhwc%4Bp|($VYyqrjXwa&R35u@S#K7ib1p)
zP|C-*X;wmcm_wpPlwx2jAC=^i-*J^=Hy-Tf4W6lE`43wRp9Y{MNeU?UzKk`SM3ay=
zh~0K<C3lCI&Nsu4)|{7sHK%%;nCJ80@Tr}(qsU<e$&{G+i?o3Bo;*FLz#z_HDt}HH
za%2;^U`~|&oaz1?q*3STV0=I9C`>4lB-0HWhzdDI;x`D>-z7Fph~oAiKSV$_h+$QY
zy$<_t0@>EwiRGWW3KD!Of|tqQr`)L!vKtHxn~?cNO(xRwP2XsMaA}(C{C4^ZscxQs
zlS3ItN=ghfw2O4BAM^_O9Z?^CJ$ht`6L?=?oF7hcVtLRazqLD<8+f%2tPc-k#iE;%
zL5%~NA~R5GLmg2vK2vR>hvNIo_Xly}_+w(mfN%n+Cwn<C=zGD|?k!MiNEetUE&+-y
zMvD}V1S0*0;|EBvfaSM;QDK86l|j3AcCqH~=9W>S*9<91wU2)fri!IEQ3d0G3BW%F
z$mbV($n2Xefw69yL);&I;qRt>%MxXu+t)g!QS$?Z2?zxFj#_<vO$@w6r@WbAf!9+a
zwp^9C4r0=>j*xYvc)d4!PTYHaeLh>=E4vcDS_0*N4**toEClCp5^&e)txTyZTmqSN
z!Zj&5X@3h+*tPDpv?hsxhwj^~cF`2B*xe9$5m{O~<XdT;@KPaP+iv;Lu6+%^q2Bf`
zO0)G%aIV@g#O*V=D+N93(wBD7p-?oqk>JO1;=)%Qe?!LPpW2!nbiI72c8UXGalaO#
zlQebnfh~yM^3@t(SgNL}Lk<aNsb%HoTGm4Z%SKaT-(weUon!a}?xL$);-zI1piaXK
zoL3+4gwoE>)ot(Wfxn%3zh~x9zP|C=9dA8)ssvoUG9dKFNs36XNO0(o5fAoHqTgG6
z;0v9|FSM1<!n3`uMhti#CLz(8dIHrxu!9v~<wc|VTT7m&ef!Su8<X8S`MKbwq~F5N
z-I$(Oq#Yd;>ej8J5J7!~&8AHc;SUV_cC~s(F3VYPT=eq47Q&=`V67m|>gltxH~FQ-
zXHK=g``57wOG7SNtV;s>{F-_4ME!GZ8%dkLz@?t(7FOTxD~>79{dgvDa#*#q%uHzk
zB$wsbZ8ME+#}snhojBU0fPnhKBwXJ{*#}=N8B9|RLj_PsF3}2jtWe(~k*HbZP(xa5
z7V9rOS*9pW&yOn=-jtf#Bu6q2j}S`E*dy5w*Dlwvuc9k7T9KJeGFxg8&iEI@X|~xm
z2wL<XAhw$8KbW@!L0~m71m_SdtQxQ-3AMD9tL0OOXIN`7vo6;()DljW<~(!dcQ{xo
zr733VXBkfI5t{ZuM;pw9O?WJg7;t_RMpAN`lz(-(@O57KP1wizH~Q)YMl1O5#-U7c
zG@quZNI}R_$VTd58c1+a{d{m2ejJ@?MlgXst)vv0_|_zeloXD`E(Se#m<y0pP9c)O
zOtQj-a`R3(E*pHaE1k|@nHxBCst2JA#`wKMrjF%5G`#KR=(i}x-3dNriN@K{A?ekV
zz@*`lj26mr3)|SDMxvPv{*1!G5eyov0Z%zNO-T)+<rm}hzZLfW7iI6gwH>x)Or)KJ
zD+a<`p*f<0_q_+r_=Qq@1_6}lv`y$QLs9!CsR&TeEbw_y#$xI;Lnp1CM_LYyER23Q
zdlGWvs8z-m6qs_{+YV{OvlOaSvZgo@67nG-4{Oq`?DzJ-sc8Rw_zPzuC0w|$L@z!*
zDY7i`WMUjbVJpHZH@|eMnn8I*ZK<Z5cIX2&NAr}GeQH*9lWlIB3o!AXIJpAhnZ-Lt
zfdXPnIrgx8)mm{kCfL#JW8+n-v#w=tW-KJgO(<*JtiEPZe$;-w`Ea$`XnQwq2jt)!
zE;)BEds<t1EPej?QKY`9X?rngh+FsH_-!~=`H7CLt7~HT{p=8-FFq6sr4Vvy(ji6Q
z<+(M&?~u$-yYDEMfodNN3$Z}clJ>Qv>1EDVXO7**`Q<OQ+<_eP5*dw%0FH2_3Z(8h
zO`p113!b;a|91OkP1EAJM&s62R(d>I$0?;$lgw9@-Cb=VZdRW8b3bR(8xB{T*L^M&
zf}khfD%Ka<tDAceC+~wAVHn#o|IO3GPUrnPWmB_7<wdOlz-p8~_7@HI&76Gaaiu@X
zQ&K_muZcC}VE@cb=zJC9YjSderT|MtvnTI$!~A~w)4xud*VdV{GX0Giop&ft6L<4>
zj&ucMJm$-CQ&ztldW99Gf~K2kOxn8@(yWSH=dcs@2U6v~>dI%#N-ZUI?>n_7GcA1c
zc=a{FGXqSW0!V4T7=H0et>1Nxg3-*8NOC;I6g`}0Dircg{MDXkyT^T>+K>}>_!qMu
zRUkeRC1!}cY^itqhboreDPn<42q^E!p{C}x)ub&+sv}F~=F&@nDB5qK*ill0rWLrz
z6CWHws4^hw*c&?z;g3_1ckdT2nhvqZS>CgV$bHFtZ3e8r-b4#8v@E|3bjzRm%+zQG
z>f(b(ZCTiY#hP_=urS0%YfLF*i<L9?uSRfKgB!%s1Ajwjm;J<<#Nu9?8TlW+m{5C#
zt9ck#WomfAT^N+hqrt?*A{+IKbGqr@W;=v1J_?^nMUY$(aYm0p&Ky&RVk&}xhXdNt
zwt~(xaRNG}5=i*SPUDgDuTXm&$Z5As{cymO!yshN3S6x2(MhCmC^Z=k2p@+OjBHY@
zlFI}FAv1{~<=8rYTdE+|Q8W(cK}=lrUbYxgvPMS2Dloc`_^5tLSeCOQ>Oydz;>%S)
zISFKskyPN}KPoUcK6tO-ZX_pES>+LjAXvd0Xj;dZvKaFxg@i)q6*%N2uU_I4#4&A@
zDj{{id(K|I^<)V*>2Ic^+J?fq-fT9Ai}&Gf{lJMKfD97EkZ=Fcb9nC7gVBTiCDcIZ
zLvq2U(5avG9c;~FROJZ?G@rgeh-3qM=n_ByDX4Tj{?j^`*i#(H=v`pB?syxA!}_lP
zrGS%zqa*8^zJ@Bb!lpVuPIMd)E@Tj-s%U(9w50k%YEl3ATynz+ztH$++$h7COg~+O
zUL^|)mK`toJMFl1J&obSpl1GUZ<CSH;oI9ky|P$|DE`NH<&lw>p4NF&=LP8@oI9RQ
zPSdrt^e8%lE7L)Y9^&=t`YF2q-tfKz@~2{Bt6<KoUU#$S%hS`RD=Tv4Nix#b`CngJ
zX&1#6Om;2k=-V3Q&(zwaV>9FAN4kYzFCnh-8NMkI_5X0xv3S!THvJ9m0e2O6P%aka
z=buC1fjH%BhlWbug+x2-xwN8`mwxv2vF*kGAS)9ggyJVOf|lH1CKx5iDZmy4%pqKH
zNhh2~=B&zQv~{$R#@So`B~FA{?beYTU0*$Yh0C{p^xFAQIm^Zz{W0Dkf<(H-(dy52
z7?D}6rbhUwR3)4HvDfvW|4se7LkGJU>sFz9GePV#&-%+aW<$?04faHZ=X2)MA@EoU
zs|&QK;to|EmVm2aQ6KO@`5Og*_NE1u#ChYNO@ArOl<|v63DIrnnI<vK&3y1l@Zmay
z4#uu`C?nOq;){5KV^a#n^>#6ZK~bm3q)FCW557{(Z_6}LEkjxZm5WMC)K_v?=;&lp
zh#C+t(Q`R-E+#psC*o&j2&L%4xcuGHc6|lq^JpCO3<p~v8Z5t@8D6*mtljl5C`L*o
zIErvce>8DeiVx!_J9Jk}vFb7<#?2kuInZrMxwfDnn+_sR5N7B*-XQw>zhx74WKpzN
zJ-8d*Q9{OEOkDEU23wWrF%HOcxm%61-zbpw|JZO#``hrE)=`H+zkJr4og*z=uT&yB
zQaPjZ3^@WOoo7{;_S;qou*?(qn+JDZK-&voCW`oouFXz_R+ccFZ0LBS$kv*7kX&Qn
zb<;OhA0Ogk{d>-GuF+c;GfIbqNl`s(FONfk#3L#OJ2+ITMGT#1<$FNRkRPWEi4J)`
zcydy@a-)tnTa39JYdX9R`3r~<6}JGyLj(CS3gYex4X44NeTmr*aHtWZ_%D~x+LXHq
zzDdtDQ``;!{W(F9W?i4*1V<P1&c)~%+Gh%JcDB_ek)Q$hStvAI1;$_$G#ak1m_hJ#
zbSr|u4}_D$rGql%Ao$9UIxqWXuQ6d{hu&C|g8$s&RS+iK1W}`&Ne_qoAi?#&b5PG#
z$SW|F$WaCj2+{>uaA1{yazQ*OC>DIlIh;|>!Mh&f;FmVAY8J~E#!z$=IVJ0Fu!{MF
zAEgL^3OKbNdmTd1@tYq+Gz>B4RJ&|MjJWVLTVTywaauK5wENNPO8AO1xPm^gOZ0wG
zGS18~4UhQi7GF8<Kpsa@)Bs0mBJ;aBD1o4($vE+}vmP++;m+Y;op|@wm-b&<Y;`pX
z2HtYq(}AvU%cMu=Of!7EmA&&kJZ@*r{mI9;5dzl!$R5|{s4qLHQ0{oWd*W0Y_g*_+
zC;ctCil_eiMGbAc>BMiTy0#|u-L}mIV$CG|_4JfX?}N#13G0ytba^Xqr8-lOW1cvl
zFgIb4u`}Sp1C=PpbzF<glUe1=I0iQS({$N93K<ky9;*01>ePA3CmiuV^<0Qg)cguN
zQC&Fuxif4uX<uIPJu7`+FX{~puAKQ(<7cL3+02m=G^(P4jT;tTBlI<BhaV1&ek1C8
z+N_3Kv1POB@zV6B;>-=QA^A3lSy4LyZG~N99|?GEFg??=wENN}xT4BAhs{hq#w_gS
zXcHyEkm4i&Vd!+@nylYqJ>en9)^Uyk&pexaceQ@o_;jWo*QYv|IytT2^baMAxszmH
zlirbq9v8b&FIbA3BfNCkVres2u{FLZ(56e+66;VN1C2K+v(Qm11^HroGzsImGW(My
z90EkYIj^%;94J$dqBPjM(t+Ep7WJ-}m`on-360QDoS%ssD`)93B*ZhL)4EpRGO?w6
zY~HwBp3Wm0t>u#(MNb|3z7bJ`aDdpzd`ye8)mA&tpM9lQ8W2kjQOZRM;}$Q#Wr*md
zBK=h_nRDN!4@=S%_~VJXA;XFG`0sl(KLlVBS&xy3;w`d0$^ssP<l{2Z(nn5za5H#l
zzH$?^9l7wfOb*5Wv6WbcDnso^J5EX_1v#ZPi3w`IaF~2+z-?vuCCM(RRpaO%1?|;g
zTK3X%l(Zq83@p*}p=~erRRl3*EAxH@+)?ZA?sp~32-8>`n#{if>@H~_WOFEzTsHu&
z{=@cKf-0YMOVS?wwT$H$m}IbyZZg<8I5Y?8IV-jdVrf(jRQjSSck$8CK>CR?I3<6C
z#$nqzn<o6FULcKmZs0e*SzyyuI~P>6KFL`C+bY>QBi>QnOuFaQn?S9p&~dsz1hPn=
zUPeMt9?z#5p|LZ*sCZhl(YGzfkpL&E!)-9ia|24!7}`DaQUrOLf}a3vG8ltnQ63WV
z?e`4d(C-OR_w@O2P?+L4tI@OrxgU3b2=6Y3DQXv(E4cUz2%Kj$D6UB5pO#}}gzt}}
zFz%=1o?l*19a)$YXO4#Q+?--#61PMo+17sH&NA~*`^dC&66;y3vsUk84R9O9jh@eL
z9~>ME@7K8eI%8YeLF?9EZ%OBR#}{1#mB)U;PyLERY&?crAFBXA)hH;}`<oTl0bW{t
z@Q{<8pY@hAe0GXZeB-z3D86jJ%oa~#!v_)J=C-wEEo=UUx#KI^Eu7s}_jRA6@i#jP
zw%$%$BTg_^x5Ks7mv_x<^T0BqKK53B{pM(Kbaq%qkgbm93%O;H$LYqqD{`*0h_fBf
zi>}z-0hLU$%(OkaAqB5@jHdxOwFQLN^|8T(*E6~e%J!+OiOgLdMD!k-3CweR%E+Ss
z+PkV(tEU_lwV)?UsjL@P&3cT?(><@s%9tZPSL=UywjW{mHWOtCIRZj2F$%lhP}<Uv
zj#C}rp||^HnGr%D#mb{-SpzR*)5i`48?#abE&0b~T3=sc;M|xcxBcZ;#XX!aeTR4d
zP=a}T=<s*5tUJ}=ng2l+PE7C0^+VD7!4u?P%AH0Qoca1Nr{`wjx95wEIN~YaqO;QO
zE*~+DI<Kgsb;mqiZ2)f<*mjcF(dxSD;(nQD;nbeYKy(Ff`b?R1c~>PdDH3L)Wx11L
z%3lkrHy>Kf_9v?$S@1+3(NB<xi@M>|4a*9?WL1jf;rBciM2PcVhNuac$M_H-wq+5j
z<3e?>uT28+3aM1TBSI13m>25tGM5{?tv~<<(NcSq|4n3S0LJsCJjV)XC;y@&3`*&w
z#x_4_&r<wxrmKs?;X|>-8vijL?ub74+8>>mgowm+^17o+$X2k&Vbmq5OlS<Mru166
zeiJQgR!84sC*1%XAu_i0PSM?&ZTavcDP}%6#R3XtvLz-aVI6R5kE2w_sitKfPGm~{
z1zQp;pivoO07kxLnHO57^jA}tpN`7>@-&cd(wJaZB#GCPmS4er|L>b?fwnCVdHs|p
zn#M9rOBvmkMFQKo$Yi)jXfMREGW1=dOwk}J3TmJjDEQA_d#L%+xGBhQa5B})h;J*@
z2#PM*i_3a{PF<2a(FShus+%`wBSGQNu}02Us=zA51`2EQ<i2PS34E1#uDzE6Tw-Y*
zTU-5gocB1`VqOGN_@PoEQ*+RY-K&t0pQE1)@9&3g4Zu~#m^C<}qzV!wq|UD0<oi=N
zCC5aC=De0!d->_9KxAv7j(-EVInI+Bg@U9ft`636BN=h12&rt0Ud5nrg&nw%^07d^
z^P-Sv0laUIG|f!GJ91J9C?J$=xd6w+okLC7)x%V5*wD~PS!K&@jUs!w2J_veKBu7I
zUHroR57b&$r@YN&SP@;Gfpwm?hGS>DxA)2HkXvi(yww;h8=dy)Fu8%xzkj)N&le!a
zz<E1OK~tE6xKy~k(_LsDFAv(kFYFIzm_hOnfb3e=231zxfa%N9(!=<fF*T2YQT5N|
z*WyM?4RyctUYC{Q!kn$^Jv+G>rlEnVmgCUUq>;=>-1q!Cnh8b0#-d?xP2x-|4zHuj
zxybtc-NBw^o!0f_tn@fdoP3r#;Xl^~zfMO#3sTrDIgZQlqzIeeQAGymn0Z?T=t=}O
z02Iv1gacQJoDEwHh2#8JdSY@jb+CQZX-C%90}PKifoZNFrkn2m+vUOC(CORR&CQDi
z!*!kY(4NU8HS!NaCBnOI`npAqeFVfRtOrZZs+BN=o0rU<XJLbj)nkBu>%x2z9IjI6
zu|`3ztE)d!vr?U*$lm<t8R01-$ip^}xpi@Szp21og8oRfj0{ezdoSF1u+i;xOwO=d
zJdE!ot_IG=#+yWbz1UOBI#RRA5<O$X8rA7}wJQ7a&1#4BRde!D^WR~`x4s*>TV;WE
z8O4E<ZUM@Tlmn#55D+w<wSe=dCN^FAj`*x&;8S)ZZ&>|@_VKg%+m_Dt)g><=`|q0b
zUCUfecwS#W(X47zf@U}_lf;HN7iJD>D8^e<2QEisS~rln5f$m6SVo`UCaMgDb~P=7
zvV=U5DVIsL7A->4>(1-8_OA0`IkCG#qez(QbxxPP_hrAxwcnNDlg{xSi?q@QSB$7%
zffq$#)4v>vLtjwWXq6*|`8nPKW#b2jNBpbkRFM0UP(SV%)z$r)qTek(ZX&g_f`<D-
z{&L^U&6$z^AUjc`N>aLXW=CNNFX$W<E;PugOS#c+#-pl*3H|sHW{~9MB4N8qT2Md~
z&?tjV;684z0|SRppyzLISQ&a<oiABt^y@D;Zp;k%K|W`IzyCY06}>_L@k4)#oeV|q
zn{u)%Iu!jhR`F>_2uM^jXa_x3T^`mx4-SIum5d}Dx`T`~n)jClP+7lkAfw#0RKy0%
zzaiz8tJNzBC=7xYpcuL~*uga3Bljzv$FoFBF)>OocQd(IFnwY7Zk<}~uOo+jXs{jB
zm3ixz0eiKEOyJn?ru7ydkihKbhx>@3W8#ovQa3~q9k<gLx4K~yi(zJPO5q1=&C?^-
zX?W0Jb)%hv@B{FO+NP9s?c4XDhXTsm?UI<gtg`I<C{^A2I|sjX+}E_}Xq~!AL*bUX
zsKKJUZ8+5GpRUgLW-zXfMLq5pz36G_^H`g48k}YOeK)$#hT;!^<|6xMQqV;AZ`d&{
z<$v!?UEV!VI@R7EF}n*nY`@r&2l*>;GV*~?RPgbeKSniC6$yRG0tMlS;kSLr+-^F2
zLjYm>J|s<BaO}b!bww}O`B>I@I+G1Ni=lnv6Gf?Y$XAo0%7agExOq|eUEt>V!S!-|
zZ@0xps40;TycJjn&b*ERL?+&Phx?O>Y8FNIYUfq=>A~jd(n%S68$J^Bv><I_h)im7
z9(6n*M&X}Ci}jXG$|>Kg{71eC&+(f*)5i5<>S>Ev!8X_OMar}sc-aV_hBoo<YO0z+
z&f}dHYqx#~`ThAnhc%w|z=xV;Dz$|7bWLW>N?xTGngHYbAK{y{6K8)MV!o+5=oc(L
z9@+geVcfRhOM?yw=Koo9ak`r?&yS}+`+abd9}-O|x(HjZ*W~>)hq$=4EGi4nI4s;d
zIIok7r}>mN=4sAn@d=hvA$mX#uP_w8UxTf$iY{(ttUvAg#+prJiOlIY>QqW#;>_k_
zsE|^{n*hx^JJR;;jIrVxiT$LuEo(XO*pzMhvj{pK#dhj2yli1U-HeX&{nbykxA*42
z&X-v9GTw{7K4ljMBHquh&^1Xa5yxhW&dTC#Y`&7@)z~+6JNe$=gs+w=_xGJY!GM%e
z567kuVH;(&5P@Z1Dt^Sc_o2%A{n1s2m$mz;{Sz?F?h9|V7j4$^Q)~Ct{E&HuFCn)X
zO6}GH$-t8pMyHFIYo!#iW_m>x&LSwVszqYvXwy2b8oHT6-DIS@wLeZG9c6@^Qe^8^
z9Ju2&lQ6C`T<Z(_bK>FL++$`|{tk8YG;eRY0vD%wBQ)b~!v+@S_bSNQH)e<QM^=F^
z>I<M4O#O-_N$G{Uo*W&i)2Xt8aXBu|8*y7G`cGR-sjn|R1O4rxdl%sbNzzOeKW{I`
zU{`Bi7Q)x~KY!j0vp=@Bwl0Vy-rfEw{&3;`*4u|;rP=9Xd1%iA=B6;rqUb5@=;>tg
zi&#zI*xSL6>IeR-uOs}iXvg@LZv3Lc{Kr>-Y&XU!L#R1R)^;G2-t<>_efe_5PuntT
zRQt$A{VKX9Ha@}(%b)cNi8J+_iM8df{bc6l7t~Goehq~OXEY0f@C=wjn{$)}^A$1v
z5A|_-S?Y~;ym0~Rj1OY|)IDd!FYSVStu4;i1mWIQQpO0wt2IJ28i#RR1Trcj(s>yW
zJan1aofjYsaIWC~q>&s^q&J^#R;9xg!$m_y<J4rf=dI_1qD}c$sOf16gEAvKli)k`
zHGPp1+~*pJTcpNTR^-QqA7*U#**|<L5dhrfwn*Mn`LJC1;E+qw#TO&C-NlCt*v-m{
z^>=%~e{GIxm(xZRa?A(p;d5?6rGEj|KeZH2v+|QDyuuU`??+X9E{E4>BjY2^zSNAm
zjD6>sT-M9VJc<e-8JOq_;fcx;@#H&reRi|+a|JWE?!(uLb#JGSfcJ?QhEZ-Lt(VhN
zE;2*e?a2ei%n2V@@Cc7(s-;4@&)<0)+O)n3lRVAOjD6?KmHviSb?cV{8AKcK_>y#6
zv&yp}o3F|_QAm`2aXZ|d7*Wg3h@YBqpqvoXE<WMuepN_9K2sQLvCkAED%7UNi(;a+
zP#>Q6r_Q`wpDkG-%v^TjXr;q8Q-2Ju9al-L2wGfPQkC$2I&2KH_>wd;w4JEyp2whL
zI9e-oyyIi7hduY_Pomd4K;Hp470!%VaiV~RJmxel4Zv`AV&}3?3w6tB(QntDh9&@C
zcfh`0C$&gJq5V5N$Xq2FHJVhizpv<*+tsC>gLG#V_Lun`tM(R0#JmH~&$QcK3`(x?
zc%=*U2hnTbac>OPn2G1AUo3xWDc{?|^mKCk5O3&tQGeY2@JFT3-3%DKanKVLtmu3)
zxY!-)y*p-1tf?RzjCDl#^c9$71gh{8;t-MeAvNPj^#>(5EBdJkd7U0~-i!ebhtKdC
z0DD6_w{-y{+J{0AhD=VP6zn%!dgSg-LWEysbn@M6oI7~j+)5EI)knjmQHOOrFI{5k
zhb&Uu8X{lq1QH`h0@l+)x=FZ0V1C(+Vv-%F+`Hb@vmJR5<;eFh=#=xHAV~Zmf*<&D
zFXkuWq?lZc!SAb`x3vCZN>UT?Z&S-BxN9m@)Ss<6^v`Iyj8nhDP=6klFlzNXSpQ*T
zZ?(yp-?`1R%%D$|UiZCfjqi2jKo<&%*~muWdn~2EDyWaz0-9JxiY%#qRJ^#*d86d+
zY3cp3a~rEU0W}L0`MSShO_yqb)cxamxMO7s7Txo>rDWYOdyE_UE{i7bW3T|Zz-qld
z)ox2axZIUFWitc2Thd|8%PI1VEMOHGkFO8PiPwdMlp)q&Yi)D)8dfyWkZ11Q$|*y0
zn(~f@xBq6N9GF`p#TvKxB>tRias2qq{d=7Ip;x-O?j6!n<IQTrUeOtxbjbMBVDk*l
zxLl`jYG-{W!7+8B6e>2p%Ke)b9Rtm?S|@MUCNI9l)PJ-DO4J*dvQFcq=GsZy9%KQv
z%0LgzW3s(*1TV<*F&mkxh81-)Ui7-a{5M$gF~2z-2zV~YC4&8+R!kc=|9hDr>@Xt{
zv_DR(T96X5!5xOBM1eVVZ6w?KCx^FcoRbXMuaRx*s6P3K&T=K``BEKkSL>YBlFDq+
zT)@aiU>sMZdaE|VN_d%+k}~_7hh1Ry1p5!Qc{loj0V&^OBYT$yp=K>U$PlvZRDvte
zTzbU1{ty%4+xZmHd0**$SKf$;{QuB&mSIi4Zyz6>N{9l2bSX$nmq@38bV~l{j!`!$
z-6b(lDT#rCGzb$JATor>kQy;?!su?E{a-xytG(H=9XobiXMNAlSM=^D2+d$d=?q*x
z`+HDibmZ}gK7#phC!kMq+4Ms`E!GWNxsMOM%r6pXFTh67nZ90uv4HTGS`2AvV?tvg
zp`jw!!}BfrSlUTh>C*ASvFcYIt+Uh@*8&cfrl^N43Z`^q7Eq}O<RVaOn0X)rZTJyf
z<7E~he~<I~XZg+HjsGOApk{e!4&sOFeKoiTrfrE2+@U>9PO8q@EfNpCQKgl6n9>ek
z<?t&tKVcdEm=%OU8N6Y-WhC(E=En_ebpNy39)i_Q(ZW|sx|!X~J1{&_oav5;u-g+e
z_5Ry``<S#@mqIK)cmN}B#TF-#82hv0<^^ES!MC@g(<a#aaD$v^03CLP1I|S{YSStr
z>%wH^epNlc8!;4p9)zkW7BSfEy*!HUKm18T6sxK8jvCgrTs-iK@_~&|7ym>eD0416
z{tviy)2rJ=AuJ?Q)9o@oWgjQ?a&~iBzoeSLCSR?L0t#)r0B#2J`bLv=A7eUk4y`EF
zPwm@mitc$lkH_E<GrzwUxp{l$3O#d`D6z2@p6KBdLh&Ke#=Fr%g0`(fyTQ1Jbg~#k
zZoGpHHlBx=G@mDByW7Z4@4~Xjs>WE|nKGp^K@+_$romARis9hp=Hb%{*m>l>%R4A(
z><kTM@Q2X?ZQCEzhAdwRi&Sm(M;`8M46Qmxvz>;Q>QDa;79!@7)UZBviYp<A%Rg9M
z4ft#!qbc-jeUw^pu{WesVynoT0#PU6`aM155Q(XQ&bQxP<45j0{E<KnX|wS^9_)m^
zV512aI+&ZMUFdd!^0E`Vp~0SVyHPXZk3ken;w|5SoY<IU)*LLko*%808T}*w=j7rX
z&-b)Ai8YsBkTbMb%f_n1Wr%&DbssaQe%}Kxf&JJLDG=eygpfcG!CBzq7u2_5cRnsh
zHVOsWE(cj`IWgtkG6GssS~YkWqly6YKqu`x@3<tWxse9NkMEbRc8L1fN!QK_8n}PC
zED;3QiK9RFQ()Sr<V_lkE`O2&w;nzeAP!)omE~|FXFqilJ=>NaMQL<Zm##br<&F_^
zjyrtHKybr4Xz^TtynQ?r-tI2K1)^6KCEmLkvYrX*qJ`P_hdbboM<<b@%J(T{R~$rp
z!^8eHipW49<v-(=AO@Kg3dfvVJDDP`R5;+1PfxpdEmFIAWOIG}d{lxw`j<%LUP^9&
zs6tn7*xA3FVBxVmyPVx!{RTvnbjI?*P131--{tD$C^Od44%(m*p-FAo^_EfoUQ(<?
zKBc_v@#Q6IIwHdSU}br2jfueN{<yHKdf{mr8}~2$dDG+nMAy=d$m64Lw~m4&SxEtG
z1SZy95U~%tK24-8TktU9bi(Xgh#DxV8K0y%hAOyAxzv0Jp7E9{wQDAy({V6lS8Cd<
zS--kC#-T3<2M3WDr`-iAk=}vCu(M?Xt{O`-AF0EkNCh_O;J~Xbfz%K6LIuXSA#p=w
zeSN!hcu2Hx)t|-1L#S)2|0uh16!4H&?8RDo1zSbTaSp9gXSXo3v1Q@B^<8{o3F9%$
zt#Y5Yn`B9UV>`p_P0X>8JsKoeLjJf_a{Qz@+UmzkMB&oKVl&B|<G$HRG>paGecF25
z_qOu34{&ZWD%a$UTRU-hTTwJN$LIWzSL;ts@(Vv&Esjsi)!A_ppdvL@GGK&fER6nF
z)%A0?upX-&=Ts)&!;ezJenAZBVmdzYVGD!~2|vchz&N5AR9KP<A&>RUBTc<L0qxZx
z$2;lZteQbQ>S6hB#-M{dvc)WE`iF{^w?o_ec4&Lli1&vn@7GI8#N^%X{FRPxU8WtU
zbw4CTA{=suwEt0(P*AbPL;?<FbPgQZBg`1%*qKbR%!JIU{ej1#<Z*%Te(hEx8SQ#*
zZ}Ns3J#ntSZulFM=)F*O2cIBiftt<L+uYBb2`!DeJpPC_@2AY>Pc~0Aj5P*fzuzb4
zHM;kVowKH7!O2WuJx#OuIGIe4BW2->vL7G3tthjTc~Tusg5L)&)&TTUUdLndGm`q`
z&po$$gHc+-NP(2feC;=&bWVbUn6)?;JJG*H9He^$0>rz$lQ9Hzk2##u#o+jTN?KXk
za9q&5(r&M^@2Wg8B`m@a0p?`F3^2)H%Yicdnq%1|yTBfS0)>b(bpr8%SPG08Lx}(N
z8>t8ysTc?`Hfj_AWkJ=Qw0p?>Z4ZPgYj(OnCvnbSWc>VQO%{9*mlQG6Y#+6Ic^)ao
zk0r1qj~(C!ZqV)5>((a9QF>bFuO13Zs)<`<!|fG6BY=^WQ02oX@0@Ko8Ruy@NY!%7
zawMAjucpGX*%dpNybF&#R_1?52AeeL&%64o!B`e;%OHSF`Gi!dhY89Qi73RmSXS8m
zEW{ZLFsml#Id{DU_I4DBLqBzR>bs>QTJ_ZHmjtJHW*emuvrWQ@FZDfHS3FDtg!7lp
z@7`?;HC?0=!+s0vC0eUcj1pg%-A_<65DyLR=r3o-f>{ae%E|*&t=nv;(+_U8*78YI
zJ{-2D1pp>y4FSK|1%wN-m)|#S+D!_EUu)7RrE<CGCBY;p!IKf&1v7<dL#};a!`CS_
zQTg{1J{60QIK`#ZFT~KuSHA8-3o+_4sv;%TjmSBxXz@j9JdF}l#&E;&lT^%fcszBp
zZ9j};_YN7u$2qS9Nx`XC9gAUKOw5A|`TSY$yaBkqf9~wCX%<#LP7r0f+g<aQI!y@a
z9pN7m{;5Y$rn$t!*SFcdq{*8oXIWwQzI)+uQWP)mi?tD91mYw#bf&>n1g9<B%x3fG
zgbX1Q!0_evCrkfTomPCa`Sp6G(Zc0LMeibA1>0SvGqki*&h;dF^zF^0d(oT8w)jZ7
z@LOQ*$P&l>SVl9ENkSk$nbvW%K*fNAm?u>|SB%`D`}Ge!w@54x38Fes1@5Z7+USpX
zSV=O>HQ_ZwP<-?5&nr*w7@>a%I=|g^u)0H8#zkFj><9&dDCUqsz3%koGd{Y}0GE1r
z=v;^Q9<eYnZ$Z9}=v{^c4p`zcm$6WxLK-k_AObaa5#rielzD*E+7g(>O6ukFh{y&Y
za>zllPCAqv2PirY!o$oj?WjvslFPiA@JB7a97Q?h2El%-8Rhe#iRtNZ@y(^L{%zu=
zF~39RiuV|S{h5=KXd%Umrtm}b4AZ;q?r{$UogH|4eG(yHFjYH|`8Tc2Sex9d4`^~L
z5`5qy?X$Qdp7VrGe3G=6l*hHl!D?J6BiAm|sI{?VcFKJ$g@h0nWrC~Om@^Y8!p&Yy
z$+jf0F_}<^#j4VaEyrai(I(^{E!;jA4#PqMQDDvT7$qZ>(XLfvFCyiE0ju%sc72vz
zZI%p``RA@2RJ6Sv!~`krrjmpNz~qDHw7sE-Fsy*Wa5Moe7dM=UfWS%<VF_~33&Ubt
zhr-)7OUyzafzx38bS7*1ZH<M)O<y^zDMaH4ne|x@wSx7J_a+C^4iANl6KIpmLX{M-
zW4;xV^WD*c7ZuCN9kv`*hM#^i;rm<w9`tcd*Hp^56c#TQtH*Vab)5%%&yneBh%gEI
zYk2z)-&_g%WG5W#W{{cH8`wJJ(IrfzqOyl3qja|eVNB-<85>%7zE5{QNwpL5cXM+S
ztAJEg$a=<gARhW|yY{IPBZg{h1Rvw$n3HS;KJ4jze=^!wTklTv5|DhWSDUu(YJwuB
zPUe%3S172Q%>10`p9*l#J^E<(EtLzG-WojL)bM&zTEB?bU5M*0`Q6TZ%d#E))81eh
zNiDSXt7qGCi#bF5iWa$%d8yi{qyf45Oh~$zuWjS+^^Nm)GxI_-?zx#7+)mU^zjjZH
z#$0l+;K{M6t-oz7hhvug0{9a%VK_n4Y%_kLUdDF~Q&F#P`~e)hMuEW_jiq;oKWF3C
zpIFHMJ0|Ok{*aHv%iY?1N~H{2h9aY*10GWF%KDg5!7RIG_4qN%k3geZKK1(ZQf_Xw
zE^RA%dZD3)h6Wjvm6M-rd>!jZ!D7n&Uu-*iBrFkb9ZmKfC=S)~?|t}#eYlCM$ow@w
zeSW=k`SkF=W5Bofth)5l-0rfVKj;k3IETbUe!dXw-)9lItp=X;w}*z|_O?TJ`jkBr
zm+OpQ6f=lT5La2(iieLf$2lpe#Z}W~$x@{UEHA5KAJMrOZK?j#vp$-#FczW9WaXlS
zbMYpwp}Kkn7+S$O-7&BT*kfXP9dZNA^QN8$IX%k|;8-77Z!v0tGMx`XF!?DFJLnO<
z&?sl}|G+_PVjRxMWc={3_a)|a^K3p$g<d_SYEfa6BAxm8Ua^{u0K9sclG%-coZ0V<
zPTi-BoO^F#(D6h98R$R1<aDN-f91L8J~1BL|2LRK6HBNWxf@UORDhvNLMrQAByzn$
zx%_pb6fho&z74lE_qL|EJ15yN9cj|k#Aa7l|92nh#z|cFq26b%-NbT5^<TOMB0X^a
z9gQZB+uLKLqo{VokA(2{1jJYj%)74t*Kcp+mtbV$3_foLF$!=O{PoX~cuy|lw2-Jm
zoSYtrU?v3REIp*{3q6W_J#rY*Zv2Ce%gI<91L%E1$-8AxAmlyecZm|A{gd*VYaqhs
zduY`^(G%@4Cx{4~HwSwh8EB1`iO_9;Neupy{@vj+k%gKnbM`#Q8!J^><FVN!0V7j!
z;Am{et=pIjj%taiih5hO1=saRr8)OA{S?7YuqkeL8<4LscVmANUGWI?JTvmqJp(v_
z0Fr^_uJgZ6o_tsTy$3b|D~R~A{Rg~P89*sHEZ@C+<!<=#l(SuQPsiJ*=3|@xjvTwt
zWrznwY2!ly{<8Kri4`{+`_A;d`#U?G_0rn(mJRdW)QICW1)0Fkj=-a%(a83X4LYI*
z=+p2-pAs!?LO+DgeOrX^7ZlohqfZP#guiW;E<dbyPw%i9skkk!IG<yWwpFJPZ=Y>U
zQ;1%#mNzQ1K`{DPwJ=>Ue+@Xj&}5H1ni6d0`WD{T9S{}_0PF;Pr={aHGrLrrEj<1x
z@qxwBFX&fjx!Ne!YPBP%Z&VrwHQC^@O|wAxBgob^JI@m?cGsfTzKtfIIR!vJVi4xh
zw@E`fh|ia|z9|1wjyxK9IF?MZOLKB6G9hA^X;_n|-7L`Ey|MYGv9!WAoosM@NNCkj
zfwc>fOVho5!r#MMyxra2h->lsv3U^iHysHPp6V@fNu`1fv_qbSv|7nydjtETm0ylb
z$%ILlPAK1+RAA#0L4zy5<bjzs{RjA(i8xZrw5rrh+B?T(-9v)uu8dPR(ya7)16{yk
zfAOjc=KXi+0@{BW*a2|aG=(D<SFwQE&eoeD^^+itIV{<zweRg@2BZDi-t2}vyFvqD
z06Dz?T{h(dT{c}a9#vvN9zTRkjLKS^|C=>MkUwF=vmA4qFDRdS;NsZ217G>ynD0hi
zP6#6ktvcM{A9z$nsRbsuu=rLPJ-_X|(YvBuw9jTis2o0>_sNynxv$MY;rU`l4nwXI
zWZRIxmTW^FaF>G4=*cL_^F8%kxfTj1@FWhUk91$F=O27|@YR7Rwo*v@KDMkea4zo@
z0j9Q*!)s`xRD=Nv0s@;o;XUIdu$yq3FYA!lM4-!n&S<^aSh5&t%oQHOWYIL^?KG^u
z(|4O_e-#=M3b3B&c$ai2UxY9I*61h=AXys=uWpORqqcesl*11?m)O5+FSOgT*80!U
zFc25%BPC7aL8%f)2u}!Y|K|wN%!4#d!_uLLLeD|F6aYxk$aXq-v8lnMt^P-iqj8E%
z=9B^=prNl#irBXiOFd+^_E=_@?pu4qR!7%MF3x=89NOH#m0oO_Kdi(R1L0*4nt4N8
z)+T3BnK3y>_5v2}O9Vd1^nQBI-i*BqW5JcRP8dm~gEK4McGAg+QwTKDAjoLSRAe0B
ziS@_^V7<mC<IuX~j+0HuoP{?ig@7DEB+K5LOUPQ~F9+FSdBmm+7VqhGP~~!pRiK;C
zyzDU031_1`!z!_!jFLXT!2W%|O<CU@eXQB#Ba^3+%&yEuOCBRL#nSxsO#b#Pbi&XO
zR(bkG%Dd2+$zmQHvs|W26><%KIv^@=Ff=p?AUL374;5!}d2>MXby~s*!(fx4-0{*U
zerT!nPg>lgT)Y{RgOf3ulYT*$`;()C{1czNWgMHeE>J~>!6h}OCiXvtZO8{#{65!n
zRnH-MPtljn`03={kAfsa!?M=&6nSairy_31{eK&Snnq7{d&4?5x|?vsj0xF)yK<+1
zXVIbJ_(}0=k%^hZ<C~z{tu4I(b_%in&R+j(c!1(OZIHf6)hxm<LHMVz+uHTY4Gfc%
z)nwoA1I@^{LNKDcH+E!FR=LKw?{8Blp)X!HE92i9b@+Eo<m*`mUirPZynOYce8RQ4
z?_kq%5ycjgYDdps_eyiZ0K}1ri3c`zyris)o>_KCm(%{*=8=_^pI=;T#__GL;u{`w
zZA2pn-R?{rn$)V2wqzjna=p`#P`;jEYkj^vJU{c_U#K<p%F&rX!K7>ZTRpxAOp5eL
zU+7YXX;szcde?E&<1rq62&$q`GP`=w*{WcB$hAxNUIJx_ewW2uO|Y}@Um&Y=_Pl2a
z?Ag_|M)@niftA@38|jcP;p|Yh?ec>-O@&E8{4YfF8xNgonEAXKh@Kg_O|H%J^?c{W
ze&}sS|FQkeHsyxzO-BEF2*fi_eAF=*{Mp7FW=tPB_LepmQlT`zxG__jA?-7#n<u%F
z+&ZT7US;ixD=t_O;fR>60elGk>ty`4?yGk#P#-M(lXHuf8_5%pvlfDqz-6pTo$Hes
z4(+%!=w$-qxanxviWwkx;IcJpptS*>7sRCEp9XqQ%6qx@zJp#A@|3+6Q75KT0Q-u%
zCammZxJ_o_8||MK-At_1o&EN?r;_CGScW~BDNf5dm!HIVN{aZq9I$xilKhHKebXoe
z5)32SBPXHyPRGDl&@|P{|8yLYdE~LaN&lP#W?BaCbL<jsKnOjtMSYcOM2*b9jH#EO
zDM!!~=sc!YT2(OGC=`EVE&Az@I$4Y_`<@f)^VGKzDao-+lunR+;h5b<6atJ*QDMS8
zRTW)^n8Xp%ke9Je0fk*-mf3dd{u~JAdz3~TPXBuEu0i?XDx?S*A{vyO=Ff=tPuoyT
zzAhiqX@bdv8=xC96Yb~+INQR<D%1pxhd-?{TeRqdez&#lJls9rfdxsj{irn#)#nk9
zRf&`Dc)Esfr;;ka9P$;shVI<Ur4eA#=ImQKV2S=t_fJL~1bng<szJSpU(N+~=a-_6
zR$e@iAzNk*2h-IWHUE41Ko|bW9S4CZ`h86vTKQ-XQOstQtc>peuSCaB&9A74v4KQD
zcvBym{`_5tSm|)2Tr&JRdi|kQMcoQ|G_o$$Kg92_GW07uMGmiw&VUBp#J__r=Qhh^
za-m>&yKorqTYzHVRX`+|B3Eqa!0vNT&*dJ33~7`VPe4`!mAL;{n6aqUMh&Nr+zcj1
z7n(}gADQOEN6#mXsu(7?cRdfD2Bmjpk5qLaBe0k3y^*Kt=52#Awi87^rp5)6v>B9@
z<VDPJuRiK&)MjNVCg$<;Uso}f8Ky~USH{N|RVYL&0BpN@Z>|sXDp4jSE;ZYZ?|Y@(
zr=$@&*o$jbZCU2+-b=*kW{V@S%9>QEwX|>K+0CW258B4246jj&VIaJdR)oH4Rqly6
z^|3a~{fu^Sz?W%F=7Q{d8<Va3K_6^GvFr1Av>cQrDQP_b7PqvxX!{hq+ZPcY7WRkz
zn8h0)9fV*iksV9Xm&?lzN)`PFKLqi|_51m9TK|Rl%Y*CJ7e|vxDuq7++!Vg|pC#T1
zii}L4zn(5D-dE4cGF+9k6B)JmkU+tiql<}KV^YuIzo$!}qBTMeGE)C$`Dpl~bAIqj
zfN4%{HQcB5Zu;$LDf_pj^Tpdu>01<VS$YdB&H$GkH?ZBS8{Ws7`6pC4Mwa0`vi6Oz
zRxXX5c@|_jm)Ikk_*2%ziXAZq;sIG)<1+*y@$%~Eeu`cyGOhSm!;C$89sNgYlfQ4j
zjm5VwN<`aHpU;AZmehT-l<x=Tt<wXF=fX0>G1eF&zimPtf-+75>^CdX*jbLJ?|6ac
zcyC2ZLJcPEE<5ODPtLTR_)2ev#qQClS7?8c_=2{d39|bhZM2>$yFk9_Hc@8isXbd?
zz%49()@%0OV-k1CDtD2TK*mT*pYT3%j9PUI@)6jORhDEXBqg;eGc@!A>>_~9%w<Lr
z9l@TyYE**Fbg(Tjbt^?lLN>fa%Yl1Y>bdd^-P2DmEV^v8hu)sDIam)8I(0t%O7NBz
zUo<KNAY>Rh6PCR>u#i}XzBcqjsJzPO-~rq~oH{vy4d{!G&(ovjDiyvLpHSlY6G`3F
z_f^O_;$Q4~lShw*fc|ucj}UdzrpLETw|f-lXL4m}FSez3NxGnnZQUu;IRR0=?8@z*
z;eApN+#L>D`F@2*TiY}mvrxt+U>7~_T^Z%aX0aF2QxgyRGYpPpSEwMr=C8g!)Y`6a
zVjAm3>U;}x^TS-grU(0E=;L4GS-6l6bac<Z<KDoS?7zjY>g#Zol}>6wD4U*uy{;L1
zeXio~lY?Y5e~Vh1ynStYfVx*-S|w62L0vYA;H$=Z7CtqkBx-gjpzU!(x6fuh+6z_V
zEzA8zvN(OLC4H4=bDMSC?6jinot`@Dn1D{Jnm8Uonw6hQs4^cy&NSp-JMP*iJmYEL
zcX@GVW>XNU2n#cLCiH6qQ`^3QpWQ}=M}A{s$<R>a=?pvBnl+CGHj3ujGd}#oFVyLP
z!C=cmm#*1#()f)qu0(3ig<fz2E~snvcd#PU9d^YiM@Yog+NiVmN|_=5J&w7hC8zXY
z+oGC{An}cEpDr8cYrAH1h47>QM41-4gK_v2;PoWS(WPbK_U>-pX*-g>ICN)pP?IyE
z33s5t|6x2F$#NIS%*<+$?Cb2sV`uiA#mfz9(}u3LJC?o%V`?@eY5n0@6^~rByZd|a
zEF2LZDVEVhhpYHg<<47J5^mi%;2e+A{d;z%R`*|*jVzX3DH8w0rJL%@KBVzo=-!Eo
z+CDb#zlKOFKXqSLFTDEQFk57|F0!?B^$mBix^;Et4}|;4+?<Pqe&RyLksAS>pOg?7
z0aeF8vcJ`0spzKy-3deP#w8_FaT-0g<4HC7SM6UP#`hTGWAjlp=evjR+gfpt-s^bZ
zn@N$|t%=*_TcmR2_TO=l#V)bZ35<n1lw|cr0KE-^K+fEDZ7%Uzt?2!eMaEk58h7RQ
zGESDtz!NS$dSM>Uxq<Ye^0!pq-dH2!8R2>MVe8em%jUPo{U<E_cr+{a^!5TcV2Q}P
z>G*mWVwFzG7KnL7fYtLOMN!jF>dKy$r1Rn+C*70q06uz$+!#)AC&+7OCek_~Oo_!K
zi|qj`5vDLtL2s0&V3pTq8;G+{X!Uu|MolHa=5bxj(zCc<?HR4dQp1F?6lq1suy>{(
z!bZ5MV)PqGiaLD0`fcwO=4xA->6^6M@^U}WBrMi!`R6T?A|V{WSGQraI9DT~uc=l}
zpfluX+OZvA(he_HfyC>0Qn9+|i3bZ<>$Zi!C;4HVyM1jZ=fw2PgV=wXh<sx)Cw8Fu
zVFCxU*a1wIXV;6dOw86(9}B&MVn)t2A-gHyoi(l9Z5r0KDeV%5906(svxm>1ssfKO
zLIw=x%Y%iH%P=zP0rG6ZK}2`(Vga-x6U_Fka1$YiEu%r;ugbu}<lZ~zJce)o+C4px
zg=4cD9{ysRZpz2AGbgp&Gwsfg04g;l;@ysg)2!j@iB{VRtzKHkhJ!6kx)(W6-p9=`
z#V{;?j;FA@LMUQGG+)5iwvHOjk9PYGcU(>M`6E=5TNZ+BS=4hvL;g(c+AQ^SDSM-+
zeoby29Q?Q`*x#`OiTO3wZ*QEto6I*#%gOb0_Sy&t7ahvdf@%1%fwpgeb<0lw?M>HK
z%Z_NDKkCE69+i(h;=INv3s#Ky<n&hm{MniJjmFMp$ZHNW8jr5KN<gS*8d3wxAQNu}
zZ$AEmC`Vi{Mz11<&PV;2yivG{n=8hwtgJK{sCj8kBVXPq{$W(UO++G*z{znlKD><u
zFY;)&*a#Oecq3(;0BPL#`Owfd<Bn4J(ZBT~`x~V4HIi%0w)Qmiw8^F~AR2srwt);b
zB~*1A5AQ?I&(A+H#ACW=#t!oG{?0*ht!Zh}{pb3;>mF^JF6X18DU)BHj6u*Gh4ogx
zjiG%C_7jZAjiU^6dpw%BoL*d;p8Dt;gfZR6k(YqOAdd&oFIS8p1$};%H@P?0#sV(K
z-@5XU9lPELX?-LH<KzZ)pTo!Q((Wd+QuF?!>YB#8Cnx>Kx6u*Z5z7ikWb#F=F3mH#
zNXb;L`<klc)xRA?!;bz<{*RSxsQa*FeG9U5YyV@q#l8NOMp}~@bwr;LIt^>MyB&R0
z<FEp~1lS7&tQtb!OwRU7#TL!g>S%}?4Y0=L|06lniYb)g5GCFrrz6Jv-sPPt`ToIS
z-`yqTFl<p+$E{qbq5o9*mc91W$9$a2%`5ox59bdu^Ma|-Wy5+k;zmza)7_n>UJ=qd
zjTGzUePiP0eomlx@G7OoT_JYuhn#Qk4f^yea0|@t!`z}TT5taXSN8ZVsW0d+aZfjY
zfUwSxRIbG!q`)YL0QuTm4pM?SI02>x$7LC&qRB*<t?4jD^&nYpsnFJ%*_V*~F<A;v
z0M9EfOcOMhXXdL-ukHV9383%ULc+n97tfH>Wbv@)hSyl&cxi5aR2hmFO(k&v|Cx_5
znU{hsn-po&im4Fc-1pPKOAqHIp{Q>6l7z&(p+w_Uv6*~b$40FrN|K=R^i!xm1m>Mo
zqRUW5i;osfER|8y#ob8AiILe})y#kP(%v)0YT>*u+-%vXC=s%xpb-mlwoVl|9cL;R
zCo$E(4z-Zo-VP94o~4WeoePA*tgr&Kfo!ZsqDo<_FamCl2dqS}Mrqb!T2|S3EVX#n
zKsJx2Lzyo|&U!xWR5$^o3c5G)!l6)5gS5U~s!)OU$%<G|jV?4!?fYjxvwO6vYfvXb
z9ydgLZs6nZvlueVT!V}Au?}Cwo4uS*&nahhR+P&{k^P+nkv(h=mH{x?t$p`1_<Git
zsWg_P2<1~MW!*Ky%q*l(R&5X1?h0~yNP`BSD&E(W4uN|k-RHrSBkEIvQ^~+hU`1D8
zu-t~`Wu;@g1-|;kc@&&_yFo*~J9oQp&kk+kM;nD{3fiY37x+bj^T;l%dqO>5dn2M+
z8aD$1(etxW?JW|3AX|m`5E|+g6r-w|FFTwr(!BNPckubv+0kFBW?B#juXbcm0AD3B
zr=$y%No3W+ao-RR>h@Xu3UTgB!<g)l%cOlc!&^KsGPbtV4d;y}?e;h1Z?11_twzeQ
zYUKKwQB80@4FJ6~4*2d1;KqyWqTQ7$Xf!G-E*7vQEjIwBIm%lcva_@Mh}Eum9IJf2
zF5)Xyx`E!AM7nv;b2;`e*?T=$I)h7Js5g6;OnlwfFf|RfH8*v!o)JuKulHT(++W+s
zV`EGkT17$T(+@RI*)6=kpj49^NAe^o)|`b9h7&PFPJV}fgDZo}cyjd6z^UTn>X+h$
zIon6QJ^(THxMs@@T)%qk;oVjafk?!KE`f(FBQWp`{w($^{1?4SnRxmj4S#_lX(sH#
z_7PqWF0NPc1coJL87AuJKQTAAGnV#<{!95=Mj33>Ugk4Sjzpmxan~8UPJirj>+LD~
z?clD6vRc8S43icm<DvR&iliaohv4^A9Q4xwF*k7$HIqMkwkR<*6S;z)u)I_9K=2!~
z^v9s#S9R)^Sbw$j+Z9dC&q48yRk<H4)^Y`0RTG(#Szsf1NV#14U-+r0=B@|m?~6S|
zs1+6GoEi5^Hgy!3QLpNi$!rJ_aO-<>dV2$$u~W!yEsU<i#hsJq<j82<o${jR3ZDUn
zf#m-bXK6dvx<B82tU*oS?SK-?&wbBfh`L0y%ymi8QobLlQ)_#dr=^YXQs+>`<D>bn
zjr_mL<mLM{HtS1N7HMBOS>4@%NJBC(y(jv%GX!xCKZ%K0EgN@~J84E{SRV0$-K@|t
zsM4`n$69-}^v^0{nl6Mh53OTR*<=8(2@gk;j0r4YCKqZu6Rh80Zi=+hdF5PY?;}@x
zM`nheS3G+a5y41&s=^w}Nw59JSoE>%ix11`WQ0k}Db(DItU1U3;!#$f7*?=h^cZ{v
zMtU!jLxlqtOY|ZLChPIZNv}S}n1B*SF2hK?toV|goSr!#6aIENLzD1TdLlskW!YHE
z#nv1(mpi<yt+nl47v*S6PyIpjNCi!hr;`AJ#W~P=395RPj*rG?kyb9w_%22Ih51QY
zlQrCn^;St*AmWrc+0UHY-jWohiLrJiio2>RBLiJbQq{!V*51zA6Zr+VjUME4^F8Sf
z*`B*SJG~)W^l&;tI48wO;eb1$u2G-RJG(oPH^)WY$3u+huxc9-!kpvh<h+$<{~Y0!
zN3$F2le6pdyS?%yJJ3>(&pvOpJhshUtV3K)aXh(X5qScs;e+onE#hGc4fUDj<s#Me
z&HelE$gklCYdoN1(!_L&nZ(&hpGK#25=wAMU+AtaZ{@k8(k%0W0>EkyagEoDKAY5(
zwvXIjTjSBr^_*!^3_l{r=Rl2P;~t+mc3TykMsg*eBZ6jJEX|I>Z78ue&HqTac&>ip
znk&j7aCh<2ae@)~eqfygcON?GUxFwa`!Iu(ds@x3n?76#sd2~8G-P}I6;ppZE1(-^
z5V9Qj{2Y~rM^FLbXXGUOZ>(^Vx}h4C;U?k(t*sKAY9%iCj-Kd4h;3O%4MdNBt|YpD
ziFrK7(kTOl8+u|ap2~prYAdVNrZ2-W4jm8TP;b-5#u)88$8v&$E9ZqX@+{O@_ct)Z
z?MP$t*bxv>=cI!QS~+}g?=QaQMuy-Q)Yh9LpG^I-1<I8!<Ch4Wf}DEEkbmsVynbXj
z8iD0*|4J`i^wF@58?9fbs~CnHyt@vVee#|uhQon@i#Jb;(`Tm0pH!T-b2%lOoZO-3
z&2#cm=0G@9YH`l##cy4tP^$bQ$$>e%@>SgkmAYyjdG<QkNIKZmR5f{dFhp)RaZ7Gn
z@s&4*0*IqY55y{PNWf{uedP}xnSYQ#cwD>PZW#)j*B*`&#_9@-IbgkVKISL>4wdG6
z5lbDLs3JvtZy@a7mkx$po@ac-YKEK~ZQkY`i;X_6`qDV7-Jb%6a!Wt-xQu3k-PwqY
zb2Xex`;2@6xu}2uv^$0nefjTf>~^ldh~`A80QjGDd1JFf=>m@UkMcW+m{w*tJw#+7
z>a+Wi*XJp|D)$^lMR<Z3^aT3y9aCO{*pG6pYLd;ehK!B}pzLfyFyE@oir%>E*!jAn
zNt?c#^@|bi0=Pmdgv{xg<}xc6ti^&aRW~#4HT~mv1I#9><cN+wDy{%j1YYO{v&23F
zWNU-n<Q6wkR=Z`bC|O*}2)ztT96t|Cg%ZAZZl{13HX&rNog^m#JBnZ$>p#@*&amGT
zu|Go$zC*dG^l+c*f~P~QDdXF2*?oN<2zrJro<iGdee2JE%Le}q#D4AYo!D~C(>MO$
ztRdw>HR=3VJQwtwmRy67AFM^Y#`zqG;h+%{5)c93@fv$mAkZGGMn6>Nj(Cx$i2i!$
z_xHWQ`0<?x8}LWnViCT&ckhqQR0xhgQC-eQMBN@=cGU~XIvVSmvR1n<lBptMC7kuN
zNYd%(T1XOd6G3IKD<?s5qhTJ<Q?>FD0#XZ(sPY;eM&xHI<Eywwy%voXGIOoK(s@t+
z;?;&iIB!^CbxJNu7O{rF>!(6@O;m8U9CAuZ0f==MI`!h^56EqW%dKEiw(YC1Q#3xm
zk5ur5HtM&@Rv_-IobS)xY%KpLicnlwRQwk-A3xhgaqZgYFY}fpZgfG?bF%YYCm%O6
z`!;*Ux}Q=BV?Omoe0+QoA|DV@xTQD#(%)nK>H<IOFv#sA%VP6KKxtBpM=VV@M55OP
zbr(a*c-^j%g~^B$HlgOm$XhaH8|&RsN${sr9#Q8w`>&DbcoU?LJm!v3ff&5^h{uI%
zNex>Z|5PiX41PHpaUS1y4c(?f&EcedmBOzCQXEZ8OXcU4!kE>i0Uz|j{E~^0<oKb*
zlPSN*$lk7Y1FEu*WNuJ~08u8657K7Z!OofTW47?wQ}o62hG{G8&sWpVqN;Ii8Nt=R
z{lxF8^hERwP6`;{UnkI>(A>qtgGQ2sw1+XD86y4A{eJt7l|~hYe3%su1IN9c6Eb#%
z(}ly&uR@i3`&wH83DztFLxXqd|DLfJCrc~cd_Fvc1vK0&uM9%Rp*;88u3N{7MEH$K
zcX@wRE!}=>Cq;zyp;#GQ3a2$%<0#&Ma$7ed?aXtExV*oRK34$|R!IGBl@8V5Kdo`O
z?@uRDp=HNB$b3r3>h_uTZp?cgN~UC@QIaZ_Z22O3=x--RU~Tf#yO>yu;t7nF?VJq;
zmPjdxSv6Y>EBZd^rPLt|@^iA`pyUMo@{@BJ$q<Y?O-~DFkjK(>N!=&ZN#IG%m}~{t
zWk}DNA*_D2sa1&+(sqTWI2gYRm8#=)G38C+p#=$nNC^o^wM?5_sz|tU?r}Y7n}+{<
zB7taFcr#T3WWHh<gABWo*|EGz6=S9atF>u;L7FLM)U<B6HbdZk!EPp5KPDX(diIQ<
ze<nUI%f8@M*Lk11S|T+Uk1T-bxw>rRcG@*}R5KXYPw8cJq$Wycz89q*SikSc6BF^^
z!GsYd%vw`boWaFX7E4haTy=+p)<*3vc-_`dsZ)msc!DwJ&?0_7NSYNp;*ZD{(K_0C
zjDtteBvUzx3Ngf@Ej-KG5V>tA*<Hn_Q+wswSXx4=1OiS66&o4DQCVx#@3P{gpl@KB
zffwW>e;n!WgB}dfQ!B`5#@^#38{!a3=QLx<q6OIu#C(u^&Jx5Eylmqv@HT&~jD#nX
zd?dXMc*T%wLH<3vN~g(gAp`;J3*NqYua-5yh_Q|xMB0;Cgxx*}Y)zZ_v?DS*1X89U
zAUNB-ChN~}saxjKZ-s2@>OyuF<xl%h{?61MQh@wnWuXZw#E8M>+rRz5`L~kdlY)tR
zY`FrT30VCHML{$+i`}>Xyc87K^=n*AABKhjQsm#$ewe`{R^P;QN_PLTux^_67@Yhl
zx)Wr)Ft03Mk&ju)jw_z19p-mKy2F@eT-h`o!7kS;mBTiaj0zx99J6>GPr!nOefV{O
zTh{50&4Qi2hvk&bTz9A9XswIskaq9qu2H6n^kK%35Hl&%X@!ENo!!oHc>3p%o~H=_
zQ>MnaS%U!O<Mo0DbInSBI|lsjl4)0O3#;mXk`Li?m3eushpkipjx#c*+uF=pi?Qb?
zCsIyc!NJGdp=eTsz{}e$y~k~>t%099mUh0p0CANt9A-eEEEDKI$_k(6{JskoHFLyB
z$%#;^G)h18FKP5`@-)fNaB^lErX$({ppAII-cM|s5Ol^9F)NUZ7nHOK;VO>v1QzEP
zrwf!6WLWQLx)?P!H-X)T;{!>v)r=TMxhn92^NrD*j;4+lRZbc2LOIB&U+>rU_ULU!
zrQ9e^Re+PKh0f5{K&<0iEwrRwvrEKW#OI}i(hHge67DLMe7TCXHGg9nLO2kkomiHm
z0_!9niF6onCa0BQ+U?3#S%`3bg3ZXnX!-diTz`;zDoIBPFKJWbpmc!!3pb-;NDqx+
zjbXGR+5_8pY$xx7O=aVHe|LQ*=;9a9e;v<F4Vx9p><cW6I~=hT7x7vj{3P&&D#bW8
zw&4e;9K>Xt`rEX(E@8E^^C=rGnW2RxR)JWR@MW$ThoOPFsr$mhT-Pz_r?L#6#=M{Y
zy!0mhroPPp5AGP*Ab_<8PA4k!hypHvd5i)V;l<MVcK^KB)o<486VzSt|0J!)T}?)b
zpDTfTeZLfF6-ZML9UM$Ki-$f47sRQ#+=k_r$a0$(ckP*uTz@>^HXf+VJ6wT{&N`dj
zf#(%Wn>s;ca7Bm3pF>Ey@;#cK6}1RF+OP8Svgzy%ig$lj?5Ueh^yT4rxZoWa@Hnv&
zCTjL_eI0J~u=n|!6uq)Q8$UNFx7GZX_VY!b>&=S>s9uyGZV=+hlBAh5COC%QK+VZM
z^<7%Dv1ZI&KfhW*SGtvV**zG7ke$w$&3p-T?~C3)xDP)MlB=6xS=>38nrfk05&3#F
zB&5$LUNY{M)YH?Zfv?PPwbwU{qq7?j`Hqf`(1Y{FNsqD&(l#Far}gF02k8>;E`~Pc
zy$&+s$&q!sGx%K4^7{AK*4#x>hwreJ=dfeu#P4$ZIlL6R_$E(8G!GRtPw*_SB{+$I
zBu_5H4oEtUqaclQ5ForviDCuO#&iw2fj}36@(Dvaz8QhS+?)y7#`^r{c$%VUX=bBv
zE)Tvj%pS3QJQ)@q5LqLbB=ia&6#gtNxaWK~C`>2CN*P}l<4)$rd1o6`=OmwKGc&4=
z*mXqQ?m*7Hj0?|KPt-?<v>tt*RilBMx_tYO%7?&Z@0Ul2UHHCb9#?1qpg3P@0N>V-
z7|osPj`{8LZ0TyMsD&p=rWA?~$3h~WwszpZKb)oXbAzNxNp_XOPHk0b>2|-&=e%qX
z6oi;XNR{(wNACZa9>j3hsO8p2N(I*B4nRV1jKYmah1k{&FC`OWS+~at`*?YK1s1H>
z!0qH#i_LIF$_~o{=hMPPISvZt!R9E#%!!|8+$?Xy1N@GZZzgU9Qii^~op`cq%MpWZ
zwu#)=z@V*n5A&SE+K57xdh%Go`~;qPYTzA$y*!C<FaGOMLn%CjABI09ZB!$%6zMJ3
zE0Jyyil>z@XN8iY0j2=zk)_+cJb1VWfCi#ny5_f|_g0`SxuvBnP2zaT_@(ddKX|M3
z&QgDW%M?qxCNfeZHs4sg3KW;$yXd2l6N!WF_TOCK>~Almo9uhS{ld8#R`&)mROU_j
z2&FlorDY2DWv&{9cUW6KlhIkFRkdnXoSkR6tKnXP-)y`2R;!GOj|d1;VF0A0r6w{B
zc`&Tc{UVPGitxg~1a!3ykEQBbcthW;<n?l2M+XYE%Co5WhjD}i138)UlEOiup13N#
znQ(dP_<W(xO0#wa$lub%kemq_th_Z##3fmq&{IJOpEhXZ;c;FbMn=DkN{146<t0^G
zaPlwg=gW_cww6{_i;X&mfYCN}1J1MmJfh#kZ0J96MY-)jW^=I8qw<@*!JeA87Jf!N
z`sC`SE|wlo%W1NV_}q0e@(rm>z)Dc+q~iZ(xZ#h&;R@Vt&(u=`8Lly3^XdGbyeQyJ
zt#x0nasTM5&-v*CtuBnIl$oE-YvhsN>VBA%NQQN<8@(~tyTfdj-MoOva01)B?Z`NS
z{5-!XmzSqo$Cj38Wtlq;Z(k;)lkA~$Pt(QX*bix<;~K6Q;{IMu^zKuDh`-U6owmQt
zqM^xx%osJsRZ1%zrp>*$A9u>YlxOp(EECrU$&2rkRyxh&jRq5}<W)q!kf2c(fI{?Y
zEukSnK|;u|&|8NnPl<-<rHjcdyNKg9fz;4;kusW#(s_9@e$BGO8Rf&ey!$|y%y+l1
z*N|ObpKXiW!v#ng+ta7VO7+n9vpY8@z}01P?8v>G(}ZMHp+oBAh7bLi@)iESWz8Nr
zkT(QC0G0QO<?9!{AJYX<^P_3j$1Fwr854Hk)A8Zc(;(`X4Q<T);$Nrp1AnWCiS`B^
z9}L@{>q+lK?^WUloVSnQH$TjeZ!6T71#_#O*dm4xHTn$Hv{@zKuC22h8yiEAao2`S
zTBwl7kWR>42C4a=Cclr*d+~Mwo7a#-0$GARAki{Du!)D$!joBHXgdG_g?5S=)zAvi
z+i>!Zlvvv6njqB6K58`ZcBONvYiSf!*?uY<v2MV33L+46Wf|O$E6IB?Iu{?l)5dht
z47>&Hp(E734(LT->6nuTC}gp*oFvOSd~HIb#_#j}4H#u#)VxwtORwU8^O5jg!pEss
z06{oog1<vq4ObgyIdfte=R<E96^3e^Cz){y*|i|rY}y!17PG|n*ktZdITa3-c!m5{
z{|pQweD|y>)Guf?l7cYhmBn6;?29+bU{`yY7k|EK5R%^t$B!E3XB$6sib>oLvnd#I
zzzb2y$Ss<m^=j%ez1>q`oh0HgDgV5F4Cr2B4?P9z?$DAa<U&KX`{qM;l^1L+ZP_z5
z*<On>=~hSMI-)NH?az&Kt1`l!E?28Dt*`4ACH{A=L#iJ%R9-KxPlrWa9t@ivEw5na
zJ9Y}D%L4ICyk$IZdsAFS|NDrwinf-j4#vyMP{g*2hd882?sEyvaC}zZ30`(rahD(S
zHZA%#^7<Re_Er`e5UCgk&RZ)K|IDAVHQDxr1O&8nwf6nq{tDoT;D4$c%-PU=<Fbai
zYs>$Z6B+FydppK>0PG1AUma9blu5XMb^ZU2w-e7yyR>Q9-8b>$8RO>;EnRI$i%mAx
zM~_(9e7)^O;SvGY7$CZ5(K+4M$D5_{l;-ZLjGd(w8}*yT6MC}WU^;a=lGSwldN+D=
zb7g-rNG8Ok;G>1VtjM8jUxr}cx4FaCo4+S0g)#gXBn$mi<mu^UZ9MG!B$29e)X%%}
zbZzurfU9SOr_w|O4tIUCycM||`CofcVTZKFb?oPz(G?18!1K0_kmCzf{{>RyCd2-w
z0Bp3g7|}Upy(r&1gt^dclXyG3e6=mY=U#}y8Cf~0$}#HY@Ker2=)Oz4y-~isyj{+N
z#CM5;m94u}ngFRiQ&fo3md)1-_G^CNR^(<0!22}1-I_up?j+mawk^$GnBQtv-ITPx
zNgyX6Sj)>w<N&s3ZV={Q_e{A&g~Je6CUvrwo))yRqe+H$qtDy=WdT2FsI)kAP_KV)
zHSKmCxL%6H-~}q3Jv0AFRRH_zu6<$;kCtv=*4hkxd#i^%JxLRbz572iu$$YiriHLs
zDn=0(?Jq!>1+%TE^SD{1+!!<%Z9LFh0N3Ggdo5KC(8?e>_mx)Ta{gE>85@g!F?|+s
z{<i=b&yU&foW~)4oUd=|;(dwHK2WGHTw5ryHwx?XVHLk|q*1wl1ErB$toFw{Z?R4;
z{L=_^7cWkgJpV0neE_v)7j@3?>1gX!BQQ!Hn`z2~W5Bmpcr)RO39NgAk!<MPt~tt&
z^DaP>HIg{(<BOM(OO@(A?!OumX%-5NKSj+J81S1+!eE^A<ZyOHG%wBv9eIP7c<Fo2
zG%s}qUmEmG>tN2-8kgLI4?YloR~xBr%UB%lDLcem4?-a0c}R>ED(t}dv=O*D&=o<9
z^`b_C5QnOzpq@+3q~Tie)W->TO?Cm`e7w54pheucI>y89&8FAMx5MT~6&d9UXz#Z9
zJa_Q^l!P+mzYFsxwODLcG%cY)O+juhZT4&O_d<cXFiucQHzhQ`NMNib)rFovQljC0
z3D>YM-s|UV`Q=FWD=`8;wIDW<!kT&;4h|S?6i9<0kv2{uS12ZU;;8%CHeNQ1-vrF7
zdzI5BZ17?7KSx$mgf@ba+r4Z*G5-#6Q8N6kidrX!ORT>E1%lZ`#~%@yIA=w?0wsgf
zjmrDdj*0Jv$m@+x{*>{iP>qTmI8s;4AdlQD*QJXQidlIg&8cGwD9y8H@mJ4VZuH)4
zW>~E$pQcsZPR`u;<9238_s)PJbL&vH-LHO!iOo05zawhyEcS=q+TZLedrl+grs!2b
z-$!A1=@Q;VH)6d=ss8w4gST^ezzu2HJYPBRihe0<$JOp<A3q4a5tKBzKMBhdEb1t+
zgiDJ14*q-)UkqVtK@`Bu-Xwn@zn_(?%CV{^3V%VK8wex$%%KMYVF5*%;l4w)av_rs
zCe71RjkEs=i-cC^!;<zT$;jvOt%|xomWhFLJWWuzx^&`zy8}#a)xs2bM?Ft3Q?ukx
zpnD?ZS!%|Q`A9gp6Fzmm3Ts);@lQ@zvk5c2oBX((_ZhYq)h-Y$%9iL6<j7_7Y4!O#
zXPqaeFYYRY@L&_O$eaLPVzM{s6v9=s$?p<!jsMki9h8ZwJ`kX__L^KS(=<pI8|4KV
zb9ho$ep;}do3*NV<tihX=*&sXPDn(({5mJfW-k?pi@@~RzZ3@GKOPJP$vtBB)xdDl
z;|HzJeO)S_)Yq3Ws_%f+Y&Z#HG!JsA43;4k>*tl%@jmZ#7`1`c&z~n1ne}c~b?yIa
z@5Xz>_DdI!rys+*B{PM+e1qo3<N32f+C>~bo%;0Kd;VTk3$fak)WK-AJv`=5!nn4g
zyng>}?c|_g<86_AkSd#!jg4otk34~%?zma%%g^`(i)`7fmU%qJQjbKu!;gOs?V}IJ
znxn6ru-Hm`Q46HheF>%GpyPd#S><@#{rO~VP0eMrPm?B8*w)6*a&~FynYuq}*NI^_
z>Xz*A{0dodbA<P;R)&n1Q7x`Fg@gnKO6YsGS=|9$#E)*8DK<@K86eCG3zbe*1*dy|
z72xwUdBL%<lvMG8ZN&vBzNlY0Uph;ih+dhvxiYDU1c)NvrfY1QU8VsyiBL#d7TfMK
zjl3TvE%l(khH$cNjX=rbkfI{~ywUxA|K^#rYP?ksx}iDROQyX2;T_(}j5fh>SKL<6
z1O9_EKV|LJ#X*d@U6ek{lM_RR$|C#oKQP#eu1NpQ@?@TjJ5gSD%hJWpd~@GPN~m%t
zglF~yI2rN<PJ8g4%KTQ4^^KqdBkwy&*RWeW8UT8VGAk-9;?X5B8!Bt;`SMu_CBdIS
z^AcQN|MHcvjz=X1No6gF*&n!=i%%NNFzZ0Or^BnA6%h0jJpL0SEPgiRBw9^s@VH~W
zz2#qA&pFnNS;PsdQ@g0F%n9pez06(Q%L0MdbU#mw$uUCWa9L>>SuG=j!>-~z`nXW%
zl+k4&#vEh;9~XH+uo|av>VINU!Wp##XRz5WTF1?wu_1TvH|rF_9hD$v2>;CEa7r~L
zm1T=_6>2<pklm2gyxd$QYSp-~c$$w<L&E?3_6A9IyUl(zXul;88hz4#6$<?Eat0or
z|6~wB{{aW>;v0Ph(zk2Cc5r*c*UsaU2<&wr-*3$}JLNvYM2*^5uBK@#>gV$ZJA<bg
zVp2@PoT{YHiFJ{Oui2gnpM)e-4#p3MZ_0?0ehd4;_l`e^T9M1I<pFaxCtItP)lA~C
z3x$#ZXLY_XvF2S{2UvSan?Zxo*c<?LHQ%dx(sqhAN60~gjm%7TUzpxU$r7#*h3ct^
z@7sto*r?ytIf~;~(sT8Jk*dEYBG4Ri+SR2g7el<I*Mh~<fO07iplbe1ftdHNEX$@d
zPUa9)ip7RCBfm&%$IHav)ycVfL+)n2A%!#D9AiJ820z4JcRtH_(>nTUA_C}zaUHD@
z@(Y8N5ci1?F+~2H$)chyC}z(ALXaObrzkeL_!qk#Qm7~VcxuBzp+ea2C7~AHcFJKS
zrrb3>Ihiw;Mk#?L{?M(dQQ(KA<#<{>Q^uQ~tP{Kixcyl?aMZ7O`?J-Z!Jh}#H8?cL
z1tKw7OC6^Egxn61v+a}jF`V|qYw5?rUg*e{AXdLjS#djB{)M2Xiu%unbQy(rl)z!<
zt0a8R`QIlLF1^gsii+l1Ei(yj(My*Q;A-MV9r)uGho5rcA}$*`&=JjfI~k(D`BrPh
z<t5HqUfM2lg=OiAkHf-qY_|6;mNr7Q%z&?`$;QBAc5jfT*-g6t7~&h+v*_llRJ65A
zlQcMH-h{WiSH4<CEZUz>_g_POBfUL6Wwl%I{_nLTW9#edeSLkc(YJ^F`=xmsL3R(G
z7~?~?LzKMdZ_l<Q8n(Wx)wDcF=2Kwe>ltG$wo;H$1dIz6wqrAZ`&;Ebxm#W+uY26+
z?~hxZ9gWa_r7Wj#`=h8o#LsU{^*;YW*3FRe`O4|n$@xEb_1;q!HQQi~B5^nOe5s4-
zXKg&%3(?WoL6}3p#sPj^Vt1w$<skZJ_}Zr@A~JHprvoSl;1iaK@%Y8<|A4(|YaO2D
z7edUK|D)+VAK850Hm+uhS~Y8{UA0=o9<?_$YVR$zx4P}UckP<rnzd_F6+4QGqJml_
zh^<C=@_X^*AIOU&cdqL=&*SqsI=yriVdVUnnP97VKw8aRQO;*a*x=veO{Va{ThwIj
z9plW!MGnHZ1&}DoB48|3{{AGRRhzZCh-ZlE&0E9Yk4dFKRei5mi6?_#OJC;1(<y}m
zQORCVZ1PPIiMi(%^?I`P2SIr3x;PE7U^EshR))2^7w`$=1E`qg&eUfMwrveMD&jik
z{_3Cl2tSo6gH<e2r#^;EA~_EZZ<#Vs=5>+g3vLn(v{w)&PG+NHPIZ#n`Br#%w14S{
z4ng($1OS)ol$dRUhmGZ(;MSoa@9BxYJ>M-_DRC*aL8FnoN#OE9>S{_O4aJz|c>Ohh
zXXKy)Fl{QeWR3pZ_{Z@Lr>ig%O|WATD>FmaAP)A#I}XDsY~Ji_<5m~F)M|vO_7okm
z2tz%OITOWIQAC*8VIbjrlbQBRN*T>YESu2`LV~eGk}NT%eTaSE<&@Z0{`oX8acocf
zq;l@M-db-|WV5KiE(3Y=c{5FfXT&=v3S8`GT>Lmm>b|iN&*gaVoY(>>Ez4A(7+K4w
z+s>@@bZ@eSJXSW8^sUKg5Z3y`i0r>C+YC0q7*v8cpZxvBj0|QmSWsXmxtFzsMk&PQ
z`d{c+%!Jcf*08kT2tB4kfM_sVT}W{Z$f4~&pF0z4v~cz<xGFigEz^$M1nx&4c2BN(
zHc$V01_t{YxxKzjBi(E>%I>Bn8u`FJWcpOByL=YX_`aAl)~v8jB%zsmeoWMrEtehO
zxD0;Li3<I<JZdEd{oNN*v+61d>3}cdVRhB&>CnG$^UBm5omY&fUu=XqWi42G<Ql8N
z6F0M8i5AXS-u+6FFj&JB`7w=;yYX2Y>Jf^OvqcYWuBYY46Bv0qq6gJ~exT)P<KqCS
z=v;7lxqyUlX4U02`AimiJ(c$hF$44$3R7~9PGK>JFN~f_NJ%oWV~_yLN*iMS{E8{v
z&oM)Ebf`tM<IAOYhcdyZ_I_?|pL-1UCJ1P?&n_pHzn%(P-ptEB+^iC)n2ToBZ}U;k
zFUex$@#kFw7uVCp8r)lRJ9sSAQp9mt)6<9Ci7J~eA)m!Q!v65}ggk`7{{tRVQ#rJE
zcgqMQ_4P#sS6rufgdvRqbbxGTgN<jH=M}nXIw_OO<Y?q)xbnhG9oeW?tMtT1-))pU
zmHixp<^*OBmIF?=`>_!`wtQ`GP*C1E=NYln-GUG&f(LW3!Ek99$bS%UfZrT3kUeku
zH*TH=s;1=S$<M|vTlg<R3`kM_@Vb-=?t&yI569g_;31<ph{Kb<<FiPm><Lp-2e^>J
zS2i&H#3l}JH$NT?N}|e?M}%X83SVHxIU9r21oB^$DXz)nSlGm!zyzFMpOVGP^Wo@W
zq(CkDi68E|oSH1Ue7%M55oVo@{l^nGF?DNqh{IV1ZRO314RTN%lYNc>he0hYuLm=x
zTj4^_jypSZxN4W)RjnQI1NN<q+0P%*-yfmwiBHDH#~-sW6DyC_IO!{8jaM$X4BU@r
zndQ6!&{qf(D>H(ep9Rwxz)&8))vueOfcxoHu8_@-d!<^+6*dod>0Zn4R&vWPS5Y6D
zdX^Vp;)nCEIa>55tcw#jO){8w&fJ4Op!chSdraOEYc?Yep_+%N%O;D}XC%-(e#=`}
zC-9}eb=u+d-+0)ufrlp7aM*K=>TM}o^T$iTdxwyo9yfZoB6&SnVrX}B$6b3<zljEN
zu}L=MHcHZ|V&mp!gt;$7@EIkIfaOuCbe9jHYarFo;3sMP-m*dVt6!jZu%(7;r^l+e
zq_h|m6A&0ExusM&_B8|E-)yO{Dd)wXwv=#?)yuJ*UL7$js~J%nQfzX`Trc0>BdUP=
zwz0_^qg%{&#e2)&T_0}4I_E8s5J#7nxXNrdTo31IB&9{uPQg{3BE{2Z*HQ_aq2Fr1
z9WXMaI<-1A%HA#rgnCtV7UbRh`}dBpYtBM#9$M(ppO8$%Atq*QyeQUxnbXwR^uDO%
z;N?rd-Z}xlxwU^6Y0&2&ti%(`iA5RUV(M`8tsR#rE|f~xUxvR$Vnv!ZFrK!hHoSI_
z@b{*z@7Ya^N=RCFiUUpj<8577J(ZBgocn9u>Z(RB_V2mb&_Hqx%83GsNIV#3uzCU?
z{{C#TOSPh<R-dQOXE|<m{Bt88SuisQPY=^b7hnH<J(y09jA$mq=IdZ39DD~H@tk~`
zcwwuFbZuWs7#~&p>GC=@Ydlc6(5>zbivkIVjCR+kkgP7@3RIi>eL6p@AzvTg=D!pn
zqwsUYGtWr7=eZsOR~rNYM;}giF6ljb;qOdtpq*<UHuhmr=rAf>djraC^_ttjEBl~!
zxQnP`>ovO=SIBdxKJ~9YQ%j`)j@gACC7Xs~Ie30qwQ|Bpe}p^#(#R|-Rcy|_ebqf7
zSctps<ma@;k%EjspO*|67o*vsWfLO7kkPbQxT?$*dEbBx4fO|Al{4KM7<bp}pNix-
znwU=}mg&sg%CT4S@k+mJyV^1z^Vv0bsBh$*5!cqx&@CG-!>Lr%V{ronaRrre(~9D1
zirB4_$5kJ_aTM8QNbiM*4+eb~=TamsiU19!jDCCP*6S0734Gxu2AV9DBF1&oL3}aw
z*j4P|YUG1-7@9P%Hu6H1EI1184Jz_!w9l-T`y?Do&?&c?Gze0|dSf3ih3m=h+>r4K
z`;US_H<^$X_PaxWu@Inc_@s<bTTqaab{Mb9=H&81fc`uQG(WS@y<xr+RI=Merl9QQ
z$!$PcCF7{DQpQ$g!BUP@@@73b#K}VL(Rl+ZRLtgGHU~>^-H?elB4P`qn9Xj|pjRhS
zkowCQGvDNrx%gOT%n$6n2za<Cy*nA^x)m$=aDDv6s%b$xRU60%^5#n@6xP<TFzlWv
zpPHIf<}|y=K2Y=&>Adumn0CD&k}O=iJqjpAHS~r$*fn&SlblxBClLDY9--5KKPlbr
zC|McUv*~q*npSrEj_zO^;r-6#)g}Nvt<!%Wt6#h763}h&ksSjM48SyBsxiE_m-Hv6
z4FUd`AMEl9AiMHwC+10~Y%e+vI2QOw+Tf^x^tN+r%|-83HROaA)TlwEo^g0}cXwa@
z+Zw-_z~j)x6WyPAYEma6IeaE>AsSS#`PnQ*6!!)0)gv9!*l8o=@%VV>`N~;n@WW+^
z-_6`eQt=Ljp8v(^^lBVwYDuWz0VyV+d8-#3__K)znAIJF$ViAE@C77F1$5YAlai9s
zJw^539tqrq44fW%grZlkJS=(xKMEZcO`hGgr}cT?9u@mz{%Sk_fH5K5?Qb><&t1f-
z&cI2eAM96}yESj3Lj~p@iN4)^Tdie1hO?xPlOEh6Nv=0-GMtW4dz!G|u3s^gkSVz|
zu(AOn{5!i9JQaLK`aR|C`jQf~;^Fws5}t&ge&ScgD}Ij=u6l$Yk<WpYazQmT<!1-&
zjb=U0qnvyfvWtx~SQAc<KlqM?7&lxlniCg>i_3*apa6Lz=PwLy;;?n{yUxk8G)MB&
zwC){a8W@`@337~&8Le|^^c7(eoVtW0saimL%Kp;tk5$o<ET6W1b${x3qXISknWLYX
z?gN#*x=G<8NNd%`4)qTdy*=J|V?jANUAemS?=r1~ENz~>V5;2<1{PyTZ8~sgW?;x7
z3$odP3qi#n+!Bq270=&kI<bO1JvaBCRMg$3F=w998Ivc*PCcuI5z_Cpx=PIEl+Xa(
zKMHyYgfm}Qherxg36U@9nxvV3S1U4k+Kbsdk;4twSBC|$B(ZXnsJ(P+D=;6~ZM`?3
zNmjaGFUjBc6{II|9O*rVv8X^FM;a2&z+jW0x?rOuJJ&YWhHG5~&=_4>b{(<sMtw$+
z*Js57S7uKe0^et}-kfwNO6<&nfycm*k*d+Ik+HuihWyd)$CeoWim|tNx9jnk;;XtK
zc7d<9=3WV@xHs$?`|XrM;LS077Jj9ldwg14b`1@RleXK%n+)RKpxjqSkx_Uuvd1la
z3EXN=d9q);ky)HJh1gYG{aa6XnN0A6TZ+&=;%P9UTyi>yg=~Y(@7kyBee?nVU1F)?
zjeUxZ8@pT}$Fj*S*5${Fi_ONOqdN4IMhaJF3_22up1Zk~{pVj}@<IVoFCa8(-)CXy
zALKXvYXZ-`LlQlEepdQmR*K%9LH|AMRH+C9z;sVWE6!NK1!$nRZ?VZC$bYR9E^run
z+VyGQ=%7)1!6vxiDL7(SGYM#O&cr4tcaTeW$B8XLZ;8p$3R(|FTH(GUrFg9VnnRUZ
z-XvmZE1g0ARZ+>$Pqp2(FMijW`Jtq*4|)NY>y2|_Yn+|qPeMr2uV64ih|O9%{OW9+
zD7C(^juiFVj`p1Na=)6@v*(qWo94<txi9sy=Cqx!8z#P;RllRYwp#_@I?StT+b;wx
zfsN}Ilm3N?8{=^`nqy~Nt*{b!F+A@>s|!wo%QM{GD>!|r8UF-#BEc{bDN~V0UGPuY
z?uU*2(z~p+!<q^+)Q3EIXV!l=i>Q|m!(u^Xz7QW505!d6m5Srg#y`nhxqco9&FD0v
zAJYYk%YH1ZH7kLocLO&0)h-f+0y~9<<>iiViEG{pTlBSiQG$2F>-OxM+$5XO#b(Us
zHy>a^iJLVTnyb`9)WdCBXQ9b66>Js_;n?zbIX{;=_&Mjty)WJK$2hHEJ5H_J-rZ}r
zuz-QH80K3x<NeOCGqINXtG$`ff?m{B>(yEa>c+>ob7eba7ia{HBSKJpyQRSUzWj&3
zPBORav%|yV)*ioJOxU^=q}7T1w79w?NMn)8hTdE(yR0y?8_xfy%fqqKJ-bYV5P6an
z8Ke`xo@}gd-#mv<s~lDva9uvr3fsm2T*K<Y31w%|@!yBo=1VZPjmNBmdUbxth78tC
z_4HIWuj@GuVE6E_N9VM|K9lDUEKq?R%2?*t{|b&%)~axjDXh&Bo&17R^09-38jLF{
z!v82B43#FKrJ$;|)1KFuRDIHski+*(8Ro?gD(cks-WV*-(+am19oc4A7pK@P61T5r
z-%U7!fy|iBqv~klM*bY^&yo*MyIZ&|Vg@Vct%x{aba%$8C-8P?ujPF?*uZ63$F4T%
zY}CiT{=E-%v}Wz0|0CuV^=$%5a(7XB9Z_N`yC1JD40j;&-P&o@OJSmPRnkVtj@re#
zAPZ{0g|?JV1WwR-iO>sOPZ-_i^1B8`mH=GnJGX|b!LIdW6I{hVMLzhdpm0#H3uK|q
z!DJV3cx$jSB6&qUIW5)HN5#}d{LaGAg@V_y1;Y`@_EcP7&;FIbP<cgyJ)Xb&x9z>W
zcV-bDZ$fLwQFjM*5dp7fuJ~^8MI=uZ8x==8e7OzQ(*ajnp!8b_YD)1LOnFXfKB0?I
zb&g&0H#?JKJAM&~^!c1(z=pi~ruyhLBDe-<5aOI{VVunMK^3W-7SHO}S|Gk>bt0aY
z!()92!gV{nYI`pYEiY6ylW49;NiX4~X<ePkudq?lL0dhm6omL#pfAJl=XZnFd>|Ny
zoN=~;Q}S&8FKUQW{Z&f4Cdi|Wi_Hx?V!fCQJKSHew){Q6W!L4#VF0EHfUYY2Hx9(&
zx>|<(Z*6`gc=+Sri`z76B2Ms|uODvF!^2Us4`)pnV|t(G#@~r{B<YbXUfJuz?Q7+j
z3^Fwe&Phm2a{VidMNQ?trPK2mBuxlG+*>(#!MBxk6}^JL-ItkeDT+gqYyI=^m5QV|
z%=$U{#>c}PxbW={Jm0dv^gQ_@4U}keRoE<Rhc5iea;m!z|Jfy0vU<Hvx7(`il&sv^
zS`GO*9Cw8w{J5nUMl3Ts%vgQcp4hMq|DOJBqb%ORBu_8fle6f8IA4);3Y9_uaT}rA
z^X{1#->W_{SWe<-oo>mC2E}HqxSxR;6}`Vq)5x=PZj`y1(LuZ2Ky4g8!1Y8<u+%c2
z@r6H5_iJd3{BI0-qd1Lgv-bCvp`Uj@3;+LGU$Q4~%+dA&Dr$Ae$Pl_WI`Czcx^+YW
zG|O!H>^v(>U{u7!_W$RICJofch5GjsHnR8d6)XA+<+>5qpY%+!QZmL)!n)npzYu@x
zpwbpMMx_!pm^?M}l`sF}VJIzT6g)$xlU4cI6-p~euR0Qe_l+cEwtA{HoddfjDln{8
zfRtj$x9ih%_mi7iyqa%iBB@fmL-u1@Z?Soq?2GU^T+`2e7MoHChRb2|mi`xGYtsEX
zTdjhc*m@n^?TU(9s-RB|9PGP&+2)TAcydNu=THJ2NHRcLg_5PTxX}ml=k+QjRp714
zTHr&@3j%q{?ZO<wcMI<nvy(C7>Dgdm+!PLMf!QZnx@wkgp4P4GC(CTh<6-A(4^aaT
zm`K{{c_#*gBe;Lae3|T1w{W?X1*zy^XA-u@Swd<eiyK{EsXw>%?pqFgA!v^d^F3J?
z9|BaRf&tX3W6yGo9$i#ukHzi_RA1wYYwH4S5;svaZ%-qnKzL@dT=ZFtE^+uPSUv_s
zGRXq!Zd@@D?!7Wa7IU~dJ$eoI#*OXq5}^>*sei9&*`c&Fox5_jv=iD4X(yWNOENAY
z%_OEwR>{Ip5{FmTmPEU<*~rZC*d@6Xkr5`;fFR3I{%&nmORLd;;c<HD)fKbO?dH$y
z$_0sW9NJwawHV@G%osT79e)uiK60;FfjJXd$9t?a{<q&E<gg8j8VV%NFYF*>FC@M-
z7C3rxix%99w6m%vp%db#WO<H!jpf`BKx^;RAfo+PJ{tO~*ye}HN`5M+{{C;DQEeGy
zABgvvZ?|~M%*51tI+ubO+O5Dc0;p#xaPUc_uvA79#zF=*-NLN&PK~n#p}xh9wQzB(
z_7pF8W&kd{OWVwCmETQnEL}<*2Ma$wJP~@KT(N5v>JoH$uvc-rIsBq|s=RWb*31o(
z{oTA4tR?EydckSsF6y#;D-gD4e0eDV++mo{%cWu4bpbMg*$$X?_7mmjyMhTf9rKU!
zp#E}gNsSnl=B&oR+3`@r$j$mwkL_GqAw@@tEPA(k0ChAha5q+XHb07aHk9999vzw)
zr<Ekl_j?R_y{GrOMQ@%2QgXOq%g`q4lu#Lp+l}FIR0}1wF1UAzi-#$0U0ne+r4K*d
ziiMWSn#{%aMkwm!tc}GKZJw#Fn$9m2SLt$0Uf0Tw(Y7wXR=1&5R99<w$vw`YM0h!8
ziJ*+qG_|XwRC-_hxyUdc^FW>sF0OL1?s+}QXX5tkMVG;Bpm5IEkD$IKnIP{L#Fid0
zaFWtVN^+TRD|@=P2y7VNw|)iklB+8=JUvkeX`R*0IMRaZRY|Gj@?Xn0uVim_`a+ne
z9A`Ky^o)&#)x!2if%=B&#s14(OcJ3vcN}`@-rad6bom512L7ZFLDml0V5CUwKo&i>
z*Y+Ix`gsCB{S`hTZl)d|Xiz(N$XQUgv8c49_x>j7PMLY|CEPE&*$-G!%c!qk>o)4-
z<aDg~kJ*z7+8)u`y9#cqB+lB}7{6ZZ4;dx*`i`WzVxLg`#;qqzOn69BSKFRC)&}>6
z@`>Z>XR5`T>#}pu7C$y%b)D&*FK!<*Sv1Ogs-_lAS~_jnA1u!X)&>0lA4<P1qDM;L
z<nraYFLn4iq5Y}69Du=cfh&T(hZE+zPOIB}=Fl&pm!AF}4}Z3-1%tawsI->y`mSUH
zeK(FN#+T_$VMdR38(5*!&cClZ_Aa=%YU&_b4Sv#ies4n-{|%&?{;}g>gUa!I_S0Bz
zbE%)VXY{eJrM7(w@y6DMnKcx)xFVs?cD`omdvWwR6%k=I&jTMNc6$jF-B+mlozxW+
zaRqQ6Ku@89+07g);;rxabCb50mV!SnR+h75n#YWn0fD}I?qU5eMVC#fpoV4*gp*hf
zw73x-4R-MT-`{Pp6oPl-bv<*~KN>U>vSF<k#o0vnP{wdzxBaD4$&93ZW3T?X+9kf1
z;7fQ;#|qT_g)LCato1FbqAm*^H67gku~g*xrdX00(RaI1BXcp{=OU0At!|gpyDwoI
z{uiq^n*z)KDlR-3)Q5w?SY8!9FMs3&8p9FZ2K5db-Ku8HW3)L-_?Y5q4E>@J@~$t0
z)Huj4;+bbjJZ*rr_%Z5jcEIl)=6}P49>0V8-me&`jR0LX3SOUxX^}+%ss8l?t2bD*
zfv4-z!7T<6XiR`EYbl%G`)_Wtu+^@mc+|SpoKM}Sf0;yi_{8}qon+xmK3Rwwo8fUB
z@K42X{{F_}{*@Ejt@Pjb;8q#45)jjY7HQsD`AfpZ4q*z&d^E2n!AUzZ3$`HeVP}2~
zbA5kj1UyLuwJjQIq`IPt(^o^UM}d=e0rX$9`Mt8ziKT<{%Rb2<NSo7S`8>%>^1QSt
zIVNo`<YDY%bAIYXpZ6ciBa>*Vn=I9>5WpjHev9tPV%g%CQR&+&FgJI|d?9gBbnmZ;
zFDQnQPwcP`q?VcCh`-k+l!{$-Q`cC;`{Tt8Fh#c;&sSw*n!9olN**)h*sPjWFVbe&
z2m&Xt96iZP8u5KCPnvjA%VV~spj@{w;di)!DDe{A>t6EaS2D_H^SSh~`gbmHKLPar
zn##1<DO+)O5<v7en|L%@_WX0QLZ8uC1~G+GZP)VeMuq)`yX21OFd5%&s&(<l&S&TX
zx}t@cawA<8IP923vplWtl`4-RWr#}BwUOH{<EsvNMHF9F<`_2C8*Kb!p+{t_%uYiW
zdx*8`Q{clz`wQT-=K;A&`m?%$uGOM03w+X;!?xV9d_bGI?8x@Tx(F-5*HTnxztQ0b
z(cF@Rm0>X+bxiZ(;il6Jza|&?vJEA`iEa~@rkTQvd|n;}qnk;`Qy{ce_dHI0bn$%e
z-QOHjqlSYE>J=#io7s0WbShkf&o`%T<>4VxGviM|iq^2F<bucKqh$F8j``8^d}LGm
zHou5MTo~3cu`jd84Uy~vQ7QWNVZq~J_yO%Ha2CN>6_vF)jX?N<|GupMC@lk!wD$2m
zK|&hYA-R*gr9Jn5?RV(Xf_Z<ew=Mp#7zl2=eFlI>#IXaBe7ng$2W`mZ++;m+W=<0}
z9J5YkH`KTZGI`-;dp}WBd?=Yx2B!48Q{=G%4@nyfIP~jTCBz^&LIbAxgIJ<7vB5;o
zhrlm$cm3L1Qm)3ny`KNxMgWkm4&Qb%%l10?)@Z%NRXG)oXVlX(lpn-?9FaLUKIk(N
zrJLB|G89R{yA_f7h+wBE3nPI9@2ZwO)=6H5rR$n|k#5yDYrC!bDvN2E`DS<a|JPRK
zi-ELMza71H?>S-0+~A&MLoVAk-@!sU-%9~|r&ZED`P0TPNGX+u_<>!jQ6X}d2J#HA
zy*`4Dr?G;h$**9z@77pYj(Q_-ou52!fsGKaaQ@yYJSn@nna-=P2=I^{qksPK%h(<4
z``H6*C`?x97W3~6e`ZHFTwCAManVEiC+4ig3pQ)E+uQyfoS~Ac8XjE|1hH`IMi2A&
znOSi}4mtlAco=$0{REh7!K`YyJli}7l=xhy{JCr_X(@C(ru(57h4^~Xl9<mr<KTJn
zM}GcqFJHc>J}>5DU`u7w`SLq^<vlvb?4kPMP9ydt-D?}m%&yHpS#m>Rkd%ojYO(Y-
z!*~*^c{PgO8;VOSSp`CK#V)$%8cv|{EY}rrr1O>snt94Yvp~#b-X$sCY>e8JHGM6s
zlWiCHd0fP2%&uxGmqYU`#6o43v&mj6zhTm{sj0`XzFd{}Z<SMxIM!yQu~LKX&tg%>
zmCU-HBkSHZ(Pa~-BCT(ViSc|HaUHIsw{btZLavxn35GqwZsF5;R=R{ay4#|889oSO
zNDr)2f4nWv5&%ow!TJbpX`ldd;BG~Qh6}a!;h|m3YC8f{aSXdft@d9<U|cjaVV5al
zR(3`2`yLPognfHH=}i>ZOi3~qhHBKg^d8xw9lMc|*R*KZH%{Xs;HF9CFf_VAlici7
zqv0`YiqDeZ>{KWhUi~V*GI}eSETZVjJ}c}O>Gm!ldr=_}x<9H&jh>o83wLjqhHe&D
z=A~XLqXr&u^RYa;a<reGab|u~{XvnEY=+vkFnmYz{S&3WnryLV6C!YLW$nIWH<Rb6
zcyi9V2x9`?QhO&-j5+riXFm0m(`VqI`K!jW;L`f;CDktC4Q;HU1(#l1gV|9pp@L2_
z)|fw#LH4)rEcG#ynnU}WSgd?pw(Xxu>pAo^B#oAL7=*(K^F34+AsBkZLVsba!X83n
z1?>4z^&JO1#64WEp(`HtcD9WNbJnB>`U3VQ0{X8J_ALR|r?Y$Io<?LXEUJEWD^PzU
zNG-1cMe$?2gB0itQ|T}%L^VyW-v`dZ#schyQxsE}SE~I&W{_#sHCHv_XsLhs{+C+S
zH|3;pGjhc1pwB7!(3c+(!b6AEW>KX@IVw~(-*K^3MHyk67{p`p9aP+09l_2rLYapU
zS~YDF@u_~5a`l~7I3b!kJ>x3UtLzJ=VlHGDANe*L^fB-lhLZ_0pp3=g9~PWa<O7DC
zD^_BOu=sngin#TYU&vd>WOvVNz0K%^c{VIL8g`W6cqKK~Cobn7cC=Z>MI^r2>${Xg
zpS*i)WGrDNiuV&)?wuQ>Ak6tkkTCK;Udf7xUh5(u5)v$0m8~8(IydRvIn$(Mzizx^
zb}4aagl+7WZ*fx>nF)nuU+5Q00d1gUT7B|0TRMyN`b%Ma;$i}5ei^f7-_zZ~s!M24
zQxA!BAY?GZMRHte(eJNFt1?zP>w0o;D_nyL#~Ic$TyY<ItH|+RWjcc~@}uaJw+UOC
z<+KjNJ%wgdpA@@HXw;e>pqQIu2=a36*5kg1CAmcUL-5|jc1!JPujGdtO+p|6M(Z8^
zUm<*~o92A2=azl7)ZLBn&rhD@KAlNkp;(Rg&sLIdV6I8CQA@G?A-6Z%X!h@30$qWE
z)|qA!v;9+b*oxK+Wga&+-g|#BHQC9(<_xXn?wv_W0)`G2q84~Xog|xKaEOK{B)sJL
ztCe7&ZQ%Vy?XpyXbf3gQoHT%IJmu)rw`XY+D!b=_tm0goyn(hwj)U!!E_nNwn-<%|
zig#8ZHIl1-@yCBIkZDZP&$LLZqyF{Xk_P=kGL`gtZK_)-ynWv)FRS|$YOvs!*~~tf
z$sS=7d^T<DV_cJ`JR7s+(tpfk*>$~)b~;02UgKX|&(~*90>QkKfUVTW=so9-m;3{F
zn01sRivf6A<}LGY2d@r<d$FGhS#D6DMULvmD4Z@mM&5J<CTW}e2-8Tz>O1S9${?)d
zz=5j`r4Mh&dr)j*rk9DY@y|{3`uIFc3g8~6KAY}yq{DoLGscF9GE#=2YA<BBj3tE5
zuK1!Pzhy944aWoszS;#&xCX_^gfpWmq*u1<%AK{Q$-$eE&%s=v3U&owIIW%8C+yu~
zKU}ZeJUo|>z+XCaNZ&;vUDo+VNU=x5KoA;))uAMh6``+Xh9jtg3*-8<uU6pr$W5z*
zyZ_ScVJ^a)#!soD@_mA?Dw<~C3?-Ap<QQV|w&CGl9SY;XIV%mk*qN<$h8R08<%|Tc
zGJn^Zp;oAb_V6(R=)%r&#DDJl7{5;Y;dN96(VKZHbXejs4)a!8Lp}vBh-}t$#Tln$
zL{CC3PAs+2b}75=We1#onoe~jM_;R<Sj@T02PVGT*NI`$N&4Ttul-Uui?Xbvvm2{V
zC*y7e13YD*kJkBWBGbVspCAk3hYB3Ydu<}|MG31~8}H>J<jGhgP`x7mi3E*?C&(qR
zq+3`hM%c2(+OWL&9O?Ob->s%EjyqoOfZ^?W0vk1*Q_5by4%ruywqYa5%`sNHWHr=y
zTS+h0G#Cr)7C|m#Z=TqehyO@mMLA&lG!!lZ0X)FTn_m}EYCLCB4-c9K{`5q+ztM5c
zA!}<R|Mhle68*8HUoj9-;p|G+6QNDD+Phdb>TK9^H1ozX_ryhK_z}t$M5Fz%-wvF2
z-aD4$Ihc^!tT@Bi$%A*3TP@~IA!8x2gZ_)n#N<^fieL2G#2QfpUS~?ej?z{$MuoKt
zK;ICC3E)Y^vGhfj!Fb!hZ>iFxX})`Y=luL&*=KR31K$IM=vrN<ax0v6jo6|K7h_~C
zTW8?T&LkMZTw~`&jS)8)CU3oY@D--te;~_(?1x_E;MrLB&RPurpt%;+2yw88@!E+9
zfW8+qOVy=W(IA)?!$fd9(=-GTe<!4Xgw5Xi0&OwVylv6MiUrrcB@8*Z$9sFf!7O0z
z^>}9x3l8ji>xgYe-VF6*x63HHKfsi?`PHzW#pJA*(xY(GfcPISJ-^*&atoLLb@SyS
z4YMo9{xwJjC}$<K%3x6Lj+<Sf!5tFC)0k`S$lkH_?-!?cp<Q~szuZsD>kB#nx8;Ix
zxkPXZKLaDyuMyW-W)kT99-82))p9M!igIM$Y3A~(+9Y>Di!o-Sa{ztF{E#3Z0y!I6
zB`xY4qXUIXNSb+wsn7_7ZWeloL3%nn2bSG2E1R3*Rlv=`R_O!tz^8U4Yeks&D6S8T
zHNvNam8~3yEcsjbtN-2_-cJEE9*%E%WT!)eE>oKBlP+q(79Y=)`U5t+E#2az;&cAI
z7d+tT%g6f`=q48;&+;xe7Okp=s;M|QKi5?H#Px?5J0c-ng&@kdm1jRyL$AG9$X+`A
z3eBGH*k@Va-Iz-p=kko_@p=kAq_+e1dUfc_!|(0q5pAX(SVAOQG4gOAX>-K3s~tba
z78hzVq$|YQi^8jKIsk7}@ymQb$5ae&U<lz0{xWq=O6_e8EAWBSrBZaw09vOMkeQ?*
zod~@+KbbF4KP%@!D&tO2m^HOiMol;$vczZG-T9Wp2~&Pc!6!5kE72~113NwgQs_^(
z)dh_=`weSXa?AN0NPw1&LL3q-fBI)(w_sYriqB&8y75&vhg(W>v1oj<kTr3@=&3pn
zn;_H`LiUl20xOCox%{2mykE-m-TEt7yWjKVy%9ax$E{!)+9wW_O-4!OuF;gZ*75c2
ziq;Kc%V3Sj$=B*QRGBzVDPo+DqqvEOv1TPnw7sBsq{FH)dO&`qT0zFef+NzG2LDlB
zW{{#Hs&~OZZsM3p;&30n)9xlwE8)tBlmff;`6vi)60-=gD34`)iW4L(7gRhpq3HCJ
z4!O}3JH(tjI%OGCM)mRhOib+h#n6{)Dunae29sZfpppt{#2;T#gZ#U{Gm3p=JyTUU
zmct{6yX|;20tC0N)>mbN<jl#2v;*S9w|@VR!89NtI0F2s0YXL1vX$Oaa*)5<6fK05
z?3#+feL-zmUj-HIr-X2!<5`qP)&e$-=N5-;kLQ}3$?!3UTTG>uAoOd%IK`B<PJO#t
z0YzEB(N3LMT*bzp^QG~VC~&%8vsSrv*d_6}gIXCNxk`69{uml1=)3+zMTlXUZ&Qz!
zd(X&Cvb4Rf>(9?UYFgEi<b@y;)gd}>JHE&54W?VdYdbEqijj<31OgU)o@h_Xa%5FR
zXGblvx-~cd;)j2*jBK8|d6)03X~^mM-nK>_qQ^(2Rt`^op+i2qs7aJBUNlH(-{EYD
zn7jrcpke${JGB{Z)g7p7?8l%2ds{W_ZDyA(C1Z7<Q=>KC*jYtbUb_GXZ}EaQ5B)<r
z-bqp-y0-5QmS@ApoFBdoEn!gH<bjl2NedGaO_Ev0W~=SWVx_ltPRNeO<?Hlo!Dcb?
z6i$;1ycmvB(7Zp!j3F&4yY?QQ{Oe?MG!BT1!{G0JJo|99fu?WjJ%@#$!u}l(Z+qV_
zPZt`#6*28x?!OryIB9N+t(noj-yXPg!qj~Ui~N!g=jxf47RfRm3Gd*Y(U2&sx0Qw+
z9$P0X^@N39nFS#3rC;n7_i)uDdH3@blTJb#I&-}GojX(YHbqEdR2WA>Uyn!9mII(B
z3-K=qRUReE;6KUlK(tuqJz)K*t?LfIcrs7uH$mcIPe8c4SFEdBmg?(Sk^&pfNIKEf
zil|4=H)rZGDFOEt@7+d=?!fDBVX~=%!>-MZ4J))ngXX*bgX9sKtoDyAWuJ*3KttXz
zaW(bde-lgfnC`3=fr03%Lz@iGIsuOv`zyri%H;M$M3INt%t#MPhC!30uG@r943nd9
zMS*Px^RE4SEGn`gnD=3@8u;lkF>MVt-;Ke$?3#I)Lm;Zswy$pL>W&hdLXi)(b-Zit
zbZl>gyI*#g?WsMRcNeREUxw@zGNN}1aGxk{9e+M7NSvduM;WAocp792jL{0iMpDuA
zDkrY(M{Or0;Dl<|*`Bc6N*78cSHY?Dq(gtRg(q+LD-}M`ZsLN*XcJ#+ZGy-Dyw(U|
z7+*?_8rP?a70%|y%-JLU@Fs6o=CMa%E>+JCKUR+$)9;v5D(S1nOi9{~jm?uBzh`is
zG(9|Brb>B~^Bpaeh^0phoHG50SbKa27CcU`>H~wiYoA@TM2&A9Z0-4X87QQx3`Y&4
z$9?yUJPM&9sQM_>p|SrrvtQDb$&$OtqU)Ywl`2@r??<IpK2Fd%VFI_LZ^wUPqPW#h
zTk5+pM-HiDNfItQ&>$9YVGLXluwx3UXQ5d{{Dmzo@K>DswJiYr*&@Xw9OaVqlj4?1
ziFFFH`t_szoQeu`yqV9&w{1z%-1$l4nNlf2aN+qMr>^279X?emTnsn@R8iKri0tL-
zm}o1!Ht8}MxxMx1yPWWNxbztH>V<ggkBQ@{2gV|TSgVwlI}W$=ECbu7qh2$_e0bF6
zii{dRKG*`&Ut;<k8o9J=<>`+J+j-8r3oy({Nh>^IIlD%NWYQ~T^y|-)ehYA3WI(sH
zv{kGZ&?_74dG~AIKObMOsKLmr#;-C`G62DhfxVwp@~V@5{q2r5>*BNb*SQPvUEzd9
z#;m?{nnrK%4AO5hMRG}c)<P!#Tf8^GyMUN;cyH)r32<#KlEzq1TMlwp9GhPbBfyU=
z+e|&FM%ijH@D*;>OmOhOV?72gtyR{ErE5M=#fbNN329ocv(})dJfoN{^!;jK(}1t<
ziGI`KK**K6MXp+=Ian(6YBhjKTE<Ad|Ew6*dwxfhrlWm-pfm6-o0!ed@1gdo@7c5%
z{wP`!K=n;}^u36emX^MqN(<ZAbbXRMi-_4e2*p5;Vz@||I-dl;koH0Nv<&{p?93xk
zX<xb|IJ%g8Lky=EK*r>it7eDZU7|yDj1Y19DNNE}9pQGwKg$6|j*TkxBoKtOAwpK1
ze3`Foc#^#}SNdM(qd~QZSN8jTzr|dWvPMRLkym?Er^IBdVWF;i;*Tc<Ufzuh1(yr0
z`VJ)K#NONh0!h#?&r?DUy@G*DX&{5w<reIp&Xd|HzhTf4MlfgGAuz>>ZhHIH2X+wS
zljq%ndS|f<jy`+G?H>2{K%y>Zf`(!Tk7KG1^L0x=5b%Rp;Ixawz)~E(3Is0~CIb=8
zVQ`JC^LsR@p=eha`eyvV_u=$@d?{HxP9HL%mG2$14A-)mvU<aJazYs->}R)J5G3%(
z@9pljgd&MO6QgWA1C%@8Yl|<iM-oL8Ik@y*rGpjAUK3=KeA~|&9ba?>@9Kb5>6>``
za6zV={LN%pg>!gQB*tY>sK>I$0_dq=%Qk2_Q51V31)FVh3~#3Jf5=245eBWz^eDF>
zFh7{hq@vGWN?%=#3Tl%-q9#jxtIqL-C>x&_q`*&0``n<c!NJ@Uvv)zOBgdpF$jP`D
zm6X>yzH#l{j(8tv@e_ha-TUfqJZ+U)u9EbMdz@Og810kHDZgcxtn*uQ;0XV1{e%XF
z;*&cT#1>tAv0333fa>!p@L`SquQqi|oul{_W)J#D(7^3rY7i^<Q4p&2{&>etfU3G-
zEzH=12rCV@XVF9P{GNc>A)&jmr*Y+{S%aC}>*6J)${+N&+8F-?RxmcZz4?A4W0m^p
z<*_v_SD+VaIe5VmIGm0GX4f$=Tg2^AkC!MdA>6=EBXHwS4C&FT#(c7m3?1nVY^FE>
zdA(F8fLPbOyOe#HZsF_cHn1U6nUJVKDyzf3S|{gp_lIuAP`gX~RU%egkh+5fsKj4a
zGo-~rO{i3l`Lii9naI5w$IP(#>Le2EK7Et063D$yL8=h*Bs~XedRnhB54-sx&lxsE
ze}G2|VBJpb+)}?twSs}hySMjwZ@VU$bI-n>win*J^T(jJ)=%37)!)%>O`csL1`fY!
z;y<5R+VXt%SW2>Z*TL_^8^~zV#XW;G3%%B==+js{gDn`2XIUCj1~5y9oNu{H9W+f|
zoa~7QMXSn$o&(%Zj!(DifPV{3FZ0A7kw$%Wkz_M7q!OBFIlgnhNW1R@?y5>}Y>Nw*
zPO~Rl`_NHDO}#g_z;Vdvl^uI`!0r(Z1~0v|XTmTN*{t*7RK<XT3exv^GGOh))pAaq
z9_`)K%-w68-m4)3)6Sl^YiU)qHRA>;(2%^VP2;AuDIjl^?Wz-1D%#K$n60I}F+RMF
z7<6#y1zv`oRPBvd5NwT?o7B@gEw$F9jC@*nQh`r*G4IYm?Ar6>_SCoMFC%A$gWn1v
z|Jec*<6~pG%Iv5M=6k>EWXwea`AuKB;rOaFpx?P+cTkrWBDBNAZJ_>$I1{UvzDjKs
zj&2+H4eTh{y<v*1D>{QdeT0-+SZ*U><7wyEl>?C|Zt=ab$IA3MFBvmEu1L3H;A)>w
z>XqpI`fz?25Yp9m3HRwN`CZVN^o3^1SY+g+6bAb}es^JP3CjfWx&6l03JKi%G7tIK
zDWC7!YEjfU@BxKrv0O|NL{@3ylA>h0!ypr*K;X%7<jXch?i70cwkxfw_o}<0pamDJ
z>HB`D(e}vtUkb3kxuUs(U6Ge%h+FT~I$sbf<B{M`go`(O*M{>Dc_3*?5Eian8*IlR
zr;3d*B$B3~(>%VuH2B#^Sd6(kIW`eT{_D(jyP^GsYGIOS-AE$2R49vhOi398LdW3+
z<X3kHVwxK`P&yVG8(MapPYQQxPI3RgOwg!6Ri&?PXU1krPUH3c%}B(VT&4)j*<LY4
zerv4A!1ndnJ9}L?g2=6p5SvqpwkB#ex_-X;4V$1ctFPZI=~kuefzuMF6^+(5_>--<
zS~a#}YRRApOVkz!Ptcy2fs73hQh<Wlj&|fyP_%Z%@FE}UZy{XuUUENdS#(7UzHZI=
zDU-(SRlT?!<>Eq4clc3_fpFi*d7N8qc2oSDHn!YfhGQVPW^QII$noH&aXPGrjASBf
z75~eva=GBQANI~XsINa$9ybJdNMHRydW7vHk*1cqpY1HULs&KALTG>{bI8ik-h39j
z8l98Anw!?52<Z`G-oXpG8n6aDpc^46N68|QuTUSH>d&a@c8R;P+XKAkS~C<<rLjZ8
z5XGfj#CWzWI8J<lwu5+t-b@kLJQaeT@Zi%=`tr22cDa+{tfiVYD~H~qR-(%+S!%$)
zgHVi(!{(QW$eCzTBp^S~Y2MT{@b9!ZE-gC~ZFzaV<;5_nmPvv?Jk5!!@nYNFK>VU$
zD22&E2fx}M+}(wmwehOEZq~xKwKkR3<@c)#n?xruv_9cnhKT3cdiw3tyaEKnYJed4
zSE;b)=Ir{vEqjY>c!t4jb$y|BouI2P0Z2)Y9E2qmbf;+yKm?{Qm9w_Jk9sG|!6fAy
zWtgsFZcwu&KjLPq(2LS!mC@pB`uCG-%%$D3-LTb^PdQ57!o6nGWzjD$mEWT8q$;gS
zq&~OiVNc+G!I?H%-%7IR?m#CDaecIL5;Yzu`0Sg;%!L56Hpa;%aKBZE{G+G$@-3Z&
zEn?hKSV9H}gS&PtQ;uDy5Y5+Z>c)CvhT1=oT5)~Xt`%&~Ntf0S3M-;f@-r0TqXn@4
z6d!*NPgt;hvo&e^cF)lSb8Ws%ekH1%6@HDj@qbOnjsHm>O6zT@{#hY`xHlmFFQ~nh
z=?C^aG-zvC-YKO6mbnPe;9AEf$1RhyC&uQ$VZ8!^U0-L@W31ct)NX30s5^|sec%r2
z@ls-kz2EC%qIS9QG=&(Kb7Lv=<08Z@KEe8R-wzjgHhN|z%=amYw3JD<SoUT2Gu$8w
zeL*|=&)5n@=dF(Ij>T_<|9#W>5Y+N8OuK5JbsuMSD~45>J!eX@!e+E_Z!f#`=v5gw
zz!V9XqLk0z!Kg8DR8asTgoV|CR!M)N5s4MvTuc}BN)aD~yTwYZVCScoD-2TJZ0>;O
z^K030(h4fCi=T$8IZC~IQ@zn4KkvQ5<C_@qjq!PPd-a*c@X$e8Y*bjTqA;-oZ7p}Q
z0U5px<){gRii63sx5AFYs;aC8WZ`m(lo0?M7H({^xQb0WG5%B1uSHFHNXUw4cSCo_
z_SO%{BQ<Oxryrv0`|Rh#VWELw8L6U@k`e}HSzqW?2nL%j%cPZU0GW^EV#7u~qQk}x
zJY^c-`lX<FTGj&#LiMegC`!!JHM9%CpP?8ZIVB6hqABFd%DEMwGdq(6SIm<3MSvP7
zcI=eNn{ZiAWr5n-zrM)6Gkfq_1MCTz5H=f8e7t)c4=2_s)NNivF*n9MnbBpY(qVt^
zk9)_<4*VC1k%BHS|5npg&UP%;6=2A}?ApGIkynNvI}5em8l{H9k*se>f9#RamV>~%
zJ!aa-Vo>(U))<%YU&-#PV?8Tg3m``izR3mxd!=%-Kg}$9bBdr-R(PCCqm;*KLq?ls
zBifB#_wtakSxx}gP3m*%A=)%g@4u`upKF>~{E{VOpxaf5L9O1fsE4QvdOCV|P;iI&
zhb_}<X34>Wm}+dAh0tLD>iExI_j!C>x%}wd2D^4v%ih<@df67kw|-hqfH{LQvC@*W
z$4jSm&7NOd@s#+pH=*};K8rKENoYOOPDhUiPd)ggA9*yZ@Xq69?N^`L&Om?e+}kZR
zQXAhjGKZ4geOtTh3KP&U3=DL^R9w7@^GqlqohzL(*W>f8^W4PggRft*ri*?i_I;}X
zvkbkzSzTfF)k4S2e1Ki20f&iEMOu|pb6?!N3667QZxfyNpEfZ$v-w(dNl$L&j~w}W
zN$nHZWXd6Fs?0nJ-({|QS#&Z*3lZVhi;Zgz&v9pe`ui+BF)rel{)fAN8QFW+OX_&U
z?*b+2@bPsw6dki`;eW%&wF64LG%B%~guw-umjvT&Yj2ZXox+K#u>cFLSZ!?>b)mwT
z;KZD~{c^{wk<XST$c_=49GjJOmtiObmu*+dX)45N$b=y3_2I?Vd~n4aw_2v*0^)AG
z^xtu47k@kBIkNjjGTncC*58gBOOH)|QVA?OLa1}Kb2<thvl{B6PeiBjs}7;VYMWF3
zW<@hQXg!_99I~Nqq-*kEd_wQ9`v)Ia*_~5FKyj9Ek@l#5^R#WF7QG_xTOL#|<XMG?
zxKqQ@q3XbZ-ERwjl(iWK8(U;eOvi<V{YSE%Mwq1q>IqA*spt`txiFG=7K*L|^gI1L
z(y>k4^8d`erQsEYmD^Zl5l?UdN+lsIblfa2Xzj+NK|EvPb=h1*Tin?)s=`$4g)MB0
zyevvD$)6_Hw|3W!W1PW#8~qv-GJhR5_2woc8(%J0`$8-AGNZi)9qJXFihP))14%JS
zxvvsEBzZ?qG5JH93Y6^QmThN(AJlYL(vED*6kBdRC3~(yk5$70%l;PVtIYdugV_*&
zbB?uavRi+~jGi7Sk8Uq+&y{-Q=~mRrSiZ|39dY$N{3#fVB%Ue0x6__$Ua#(KsWlFR
zcfCsJl=0i{+GTp_m{xmtCVM4TQe-$CDnxtP^?f?@g%nA-fzVMevcO-uw6IpF*pRmc
z7Mw=RfdxhmzB*>7C6oiUi0K}we6AEV%`B0Y)&RS~-NwntFnK?&qAJsGe(e-uhV)HP
zrrpZx;f-*gA*HdvtD!!XMSktm+Sz5=hWWO<Dld!NYsaT}-mmF8MsDs;PI8blyN-JA
zPY<jqtJLy$wJe_Ce^C4ZDX#v|XrS_CeDXE&43f30Bx~@<P5}>q>eU=p|3@r@91A*!
zW5sHIS#6ELl4q&f9V~la(3bZ;Y1&t~H~8Ye-~36{-qf^DlNJglzwG-~J=x`>WWr*|
z_*|g-IV;OSmK5VS_;)3Hd1e)Oe@#E)c-<;{JD<iZb91x!AUs{(G=R>dNhN6&Of=Np
zrf-Z~xc=>hFuC0WI%T?w8mgl<dQiR3^~^)=M=%T4=fN%oW0nO!vEp4ttvt2W*b4pZ
z3U>aUA}ZHzC$CU|XA2C!4TGj5Yj~>pUz|=}1E+xy{7m5CboWP*4S)&&FsXRRS<G$g
zbWecCz~xQ<@&31=`}z4^{(^<>O$CfQju4*1G4{)1Vh-rA<mS+<V#JRwiF)$G3PGPj
z(Iv~kGUS>iSzWNe397nq?wwJ_iR$H((7r3n+poqSOz|&~daR6i&L-r$)pz^XdphfE
zOzMqYuY!RyeiZRLpmmi=5o49>6<)gCZ+fU>2bEhci~NCYj#hyEQj0pNeCisWu!)W3
z=hVYVq5;t6e{c&AkDsEAF3a0Hbl!XBb=XU<{RN)|&wqt{ynMU*x`@XvOf`as?F)XS
z+~q9}l^_wPI!6-EA&0?!`NUaEG*pOyX+6Z2`jVV+YKieM+)Eq8G>d(xDG)~UM1c+U
zW2rdWzuB~1dBF2*!Q;aj{YuNPIHB~XgIFy4qhvt2zz1p`Xypk)Tt^O*HTDv+lRx+9
zp$5wrMca@mk{e@Ze_Is=tIdN`#dqZ?UFn0oRT<J*)mbC?w8RJhgdX+y5fu3F$0&NM
ze^DD))1f1!icsl7&iyb*vBkA7Y6Zz8N1g-@k~uoHF4!XZJ{#c4e@Uu!h4kWh-anTT
z4h6Q7FUH2MSFd??BmI9q)A>7Y;?@93hQkSCzw3tlS$3Y+Z<tiSym48Cqc_ey0YHLI
zR{Z0OXt+sJN+)!d)DiKxK_rlU#y_P<3i}N4N(C!EI$jmGr$JDN^;@__cNlmXMFlYr
z=tc?Z+6dD!Eq1O=1<PVYdH>qiu4Red0%Y_tYQ(Uok%Unrw*6pP;AUM&j57YKkg`%j
zc4l&F)PKnCy+Dr+sgw$`zN^1yp_?s=?OL)VE!F=mB1`n$zR7VPm-Rgs2f;ETgkA=*
zv7@{Q6yLC*8d{YJU1FX@{m1vfj6kp(bsR{{o4GZu7Ww>94A_JoqI#OfuTC9GfpPoi
z)N0w4Evl0g_Qq5bufzm@4=Fz?p%b-E!6dCXn5jGyZKm}vhYE9wMBWJnwJJ+Co=n?J
zHOE_Fs|Unkh3w%`F2M=6GrMd0I@53YWb7*_+OD+y{^4Q&c9dpVO7a)I{&`;3<n+cb
zd`k9WNMdz?z;1QnG)5Qj=f8mrAphrepjXgmRY%6d-UW$<!pooSd<(rTSq>2lu|bO_
zYsi4)+`8H)8Ivs++{g){&vV+mithlNPw~nw6_q|DIoxhe77M6q<3=WaW1<9S`T%cl
zCf(cy2O4RGk^n5*mHN}4g19rZYS{Zr!sjZhO=vi?d7r<7M@_&kI^bvx!0fxymj;L4
ztuMoTR>M?q_ynpUtJDrY+sDh-?kCG<VtMlkN4xV$tuI1uWkXTJ<);`?w|(VM$nD)U
zS~5V^o4R;!TRqO&F1*f|RQS139@qc6VzV6?;`?D&!=qfh@vicJn;{XtvD2r(UDHEr
zG7NzZJJj2He76yBbOf09w6qB#5Iw!O>#bHdbHiy5o^>m-G9O#V_-D8(eoswSx3-Qb
zkD4Txtk5z(2K`4CaJy<2NAOLAChrvywp_N>6Wrl2{f0l=+e^;BMILuR+B*J{1``xM
zmnrQY%<;h%VgqF&J!!DjBggbI6)4>VKW%*#KYDzgG`V-Q0XqN=>{~leU;}rn$QAU;
z@VNk!#mx_H9MipR*N@zW#*aP-7iS;HzYuwugki(hQiE)UD}xXro+z6#8<TFtAot5&
zzq*mo{DAhyt=r$fM5`V@eKw2pstxjMWsOo_$LVYcY6Rq)q!RV~b=E4bCm!NAlFoWW
zrjpE^j;l@rQ;r7OlQ^KN`s%D4<?L+iGcr&XRXJiZbFs;%#r3%(+wNhTF?nJgohaZO
z7IC4D$2o!>&F;^x^+%kg=i=^{I2WEUxe)JXvUx&0>}Oeir(q>*ggR_3EHg}x8CwD#
zrMIbx-&kuXuqH@(zReX<W|n5-ei=h8^cAkGl+DhYXUYlS<6#9V8ki>7Fxf&NFLf0Z
z75|T?^NMQXi}r9pkcc3mBcKqBbU})AY0`pp1*9XrgY+gtmC$=PAYDp8LX+Nm6ObT8
z0{-bBy-B~h5BI+1Vb;o8vu4hmv-kdeAKNGQCPTFnX9FvCgK3Kvq;XQYGchw85dm{w
zvk+}&u!@j5;xUl}S41MJzFV1evd_-asqzFzOrZUmw5#S!K$T8cJ3}$eJ4ZTdX9pFB
zTcXU%hm3Y^GaW20`Dudir-m+GRzkw^L5XifS^&a9ea8X<M#Ms#_pLi}q&^;-dx*7q
zEn_fiF5E3ln`^svdn-c+jf?&jWL1YZ|E*X_OXl?}KHcK5rI7Kxjl|0ho~_ju7|%4k
zFtDwwkApF7OSYLRNOiPUGPM}f34~Hf{PYfT$7#fjM%A9}olOp=H#WbzL;8{-pSU`T
z2EbP&5sqiO&6NgR^*8phlZH%|T0hF`D#{<b#6B@EoknVuj$>yW(n`1p`yoXT!vd84
zjKdVoNCR(!sPgiG*xm`9^+)Nf5~Ni4kGh?HJ6M_vQhlmd(oB>bP7G4irb?2{7oJS0
z)D}=GPKA7}#~4zRg0)i=`ez(w-tGhddqG+>Y2+cqFEIJ6P}(G`cmX9Q%SWuqR$TPD
z_F-ls_>5*k_7tB@1*9AAart9s`TQ~NqvwQDp3h~%W0^;Le_==`OUKICR(G%4?ZvAL
z6uk=)nWwpSG%|O*zp*jzAu7#f|6He`cIag1<ILgeft91Bt+!rP$=LbOLEkFSbapWR
zXMBtat3M!_vvBQ#-ga|zI7zW?eV3|RkDtRhO{)kv+m$vJxj~-E%+KST1$1EpJza&$
z)UfkXGvLxw?_LU-i1$<-lXC#iNXKFKYqP*VCqdbzpIYmwtmVPXRB?FBSsybrQOdOP
zu_nE3T3h9o{fv|pQYKgZH58Bu^q1hznL!ivFU=!^ZciL%U0f0(?|Id!&`|X&s46=0
zGxM-LNP6N}234XuH3ybE{<+>n=G`7n{M}h2(HF7SRCC|gZ`4?`?UIz&rTMq9DXjE#
zN^WO!(`r7|nT=%pepCt9N6}5W7|lH1U7X~{KWQ0(V1so1;zm9fS8~AhrD%IlAyz#M
zsyowlmoAT)^6$W-@qo>3*H+cxFKY2s_(8Id41^lCF-Jp9D)qZoc)_VLEk*>y8v#Ku
zcP9`&|BIvs>2N|3D$pPPT3sOAC8j}9S7JR`c&1-%Ek=06^>@%@*jGj-3k0D`vPZm%
zwE=zj;e)UQ$bbH!v4FsF^PJt!y_36sxb3wCsci?#WZ@mpLyg9qGk@5Nl;;M@>`Gv+
zV2CU!8x$7&iNSaon_c)OI=Q!!HVh(^60aOK`E_yP%~MKfvU%adABx^OQI#p){w5HM
zdc$XQ)4DaDRk~E_$yRtHnZb13wuBH(a3}!*7*4G!PbCWm1CH%!5W=Lbk`n<zRv`%D
z<W;0O82TN;OakS+`;Kp<+^LNMo*g@o6D5GZzrBN0TGTi1P#K&IY(k`vU|tYHPCga^
z;v{@S1WN~VCerqWE&TB<%-LDI>Y@N3AKi~IdKS%0F#Ha19fxPCdF0)Wik8?;Pg@J+
zG3l(2qyXLqxiS*s0iOSvRBm_Q5k!>m(kph2_Gb<Z45&|AA;;HFHTYlStuMb=M%V7o
zd^*KHE=F4-;H46$y0>KA$~G)d!WDRE*nCMq@(8axJIz89;`H)~Ma8M`WMK@3Z<H>=
za+1YG%KzWeMclTg=ESV^D}ipe=iwXCGscbtoPg4GU`^)hWcl*i5x6B%>v(0wJ~!LI
zfL;3^L!ChE*V-a`Jj&`x@Y^W`&p6s|@vX#csXC<Q#ZRGxwK~JiN(BngP;%$x)6)mZ
zbsj<D_D-A?{fe?ICf^X0!7roJ<x&t!>d8OH{eyKEtm^-j-mc8vVsCdlelI-AQ!iOO
zzX5!nCoTRvZM0Y#L$=Mi+ZW1oNR5#6%&h!|t=+T*1dx!#NCjLi{c+@PH4Zp~-+P6{
z*sj~!uJ3I6!BcPc$M1#GdDRvx))<;B%FGW9l&19cv`|OJIDP+ce*9sfPBk-V!YMAL
z7h!guc5C@tyW=KD#&=Vwu>&93?)J*nZv#eR5-gT>g8Uoag)-+Fr;7B?S@}KRGMkeT
zoBuVeaF$Tw(=0)6^2B;8Jmo{svRg~e5I(Ibc;5|emyW8vAtDl|?-YVKJR^A&A^+%a
zto_2uSUk~QLVzMZOsY=~repY=K}eM3@fgIku66H$92HmR_GO{jt<3FYch&C4D;%)n
zsLV4W6sg?OvOZ!n2WC~|G=XFs`5Qzi8Vm`~KKdBgZn57Jq=z?lU_`VQ<aKA{Y3v<r
zYOQlOmwG+t$pUg%+He2d=$RG6+YGs<WCQ|=SqZxLc3p-3oLjXwiMwDGWotMc%~CC#
zysLMgjq~$8M!Hbo%b6;2%+CU11OO^3C>%-*eroq}>iyav^M^Pm1$rq3vwHPpnCdo%
zW?t2>``%Jw8XprAp|5L~ASqY)vGjM=&-Gw!R_ae)(CSpd5uLr~5{hJS`N7AsoW;fR
zmJ+hj0=jU`=Wi9qC^NO;PYEChOWC{T>7l`)F&}m$C<$qGa6>ApaypoT8UWMeEp8^L
z{E4%uGCzs*rP~aRVpI7R+6$E^&!n!JoUHq%O0?qN=z9Ei{jHg{#UWoE>Q3S|=)ml*
z-MvHt;xG_m{8jlZ%a8i&j<W4@s{*ys6z(_yyuLFZ3|0&OZ-A>!;TS=f_=J^8j>x7<
z9zvq@Fo_KerxJJDPug~A06zHL(0o%*S-HBh@Tq3Yr~aY5yh7sIapH*QLp45lpcw8h
z*mCv<nPxJ|Z`(!CV5&l%{Wh6<W=M^M^0S9d`N`#TC5nq{hk7X~vCLc~S8%RKPpc6o
z!s|Vb7cT4Q2tmjj8u<`$z`Hp=$6@Bb4hAGZ;so|b%NU82)<36C;k_9tyK^bEBor(r
zlD}}j=UtFKP(Hmr+^*@gFjL!TJuOw$@Z2u0-O=dT%v_(s*SdOKq??=D{=q_vo<*{?
zqCxyX4t)w26?gu_0*xuTQtbETm`(fbCq#M@%dGWD0(Z^t77YpzR!&c-Z?nBJ5rw)>
zZA!m<x#M{$8&g*i0mRH_mN<(UhA=;Y>X0S9@|?_ur5N(kkD#3r5;FEW&boXXT_uqC
zwV(WSz5S|sl%oO$_ctn(xd>bA_^@o_D%O0t)HTB2@%_cTl*~)IP}6{;X}s&a!COY)
zJ_+~B#{Ctv6S4>q(_NBX=VovFWE#0zF5c0KefyugsU%$iAV!O5Q-jksN9~?Uuq=FZ
zmmrVC7Q%}Y=qXd$LDDZJaah&63-O>ehHdox-#GN=$bv>m=7pQ7MZuW;ah9N?R(pI$
zU_D`PuH3pWpl5t63|ao!b!YeY^-@>ih#w8UgKGY<v)SR{7p#EJzOR^00~yLW`&aZQ
z*u76Y%SjNx{ryKKDP;t7<LxCRV`M6kjfQg)vg0|ux!SXC)++v0_YCuzA|dZsq~XDZ
zkx%={LLmw%I3APUmY|MhmpH)OK{S7Dt9N#$ex^fv(Plc`TtcU40>|U;sqI(3=A4j_
zMwmUzN!XQn=p4I6XJh?Hiy94)$+xad72GUxZMz2A!g%zn7F+#4_%jlgEOf2*2@`$B
zdrcjEPwMq^*fHrokvUmIdPlk@=a@8|#3ZF}DsMwM6Xn2v#QQtR27^Q9A{koGpfI=u
z137*mFlNqaZ-0MjIH@8-8eRSRVJa&z%+j1C%yyy+`YoIx=}1F#Bw3?$IyFsSmXoGN
z`zuhNoLncEOqF$-95^Un2$AJh?SnmeUg#E*gemmkf)Gg#v4NshKy2@-pFXuuwkezX
z&OOg4CbnKk(bWw3Mf|W+$?i@Dir;8k4hFYzPdT09>{QZ3K(x8#j@5c^-OYB|5<==l
zbR6QryqKS;6K-AZ2qKU+BwNAJ5;L;H02EASgL0aGsxm1qxaf$*Hjq;ixho9GR3(|}
z5QPuI*n%mbARQP8UkbPv!3N<Zmf(NByR(Z|f1GSjICmWN?6u?5P<yRMcCEG3V#<#_
z;%=90Z%?(EZaRQdtd|?fwdtG~;QkQlGwcgp!#(vA-6Xd_RG|o1)zL5NQd9sHu4oJK
z$K)Cl@!K=qc9ey$6=Gpx+GSI?IqRB3X~s0Re1Z7@&G{mGbovQ{f#b7cGN|t!(nFOI
z<Th??aF<;dZis2HH;Kh>7nk2<ty~>i1pIb2QAy?%)}3zNtaO_ZoV*7o*3qyn<$FH$
zw-4JmK0Xne7~%R5y60W!B@d7?;WHwi3ZJb9X0zwpKOy2ujDXPuYeS*k7iKT|zpAkn
zS;N>ru6Gsg9N6i$!2R@uF(hCz5hX>oP;gb6#Cw)|Z8RZ>R#ZqVrkoQc{Dtbx?CN-z
z3l3hr!!qN1eaOQfv@W@rtIikb#*smP)O_B=53Cj@>;`)7J;6381Cjcr6||i7yfK=p
zjC2)iJ<7yoF4+vTN?Q4cr^9RUrx(iu2@wKezt!k?MtVWl!bg0h-9e)3OB3uoPgc-w
zzNu$tcNLfEVv671g$7=nI37*gFNlyBJ(AoK$9}~;idDKj2QF9^0<MlNywR|{<IC$m
z!Siuwr!B~-QVyd6F!vLVxsS>4>QgPT4`nV^fJRW57W~@qoG(zJUj3!`#hO;jAC=Nj
zCB&u4UpmsDI7QXhobp;~`=_Es=Z4SGt{nQq`c49_3GlIOafM4??}yBzxqYdz<PDFu
zjH}iNiFy;sNovnWB^S57{d+tmkX24<NakaSzAc42+-#PBLV1e2Zer<X!FVKcw<OVu
zc3T)L{W7OJLF$=I$P@VEM<)E+c!Shpb&iA>u(~u{&WCU8eYQ4G_)9LD7qC3G=2RjP
z2ALxuPVBeRp@Pf5Yw6Fc=S_YmBL+>1?Ggy1$S>!f!cSPJZTkymnRfVD1^~Bhds4eh
zDRv@YhD;zLLkCU_RhY=NbZVPGYv4y39gAtK;oqJ`X0ofoDGTZ4Uzeq+!i$Tjbjm!)
z6KVR1IK=NdJ|ZOT;>F~8Q}Mn@b03jTOb8ZmKkkIyWif%MQKhl3W~YZdQBvZ=Hha}@
zXQS=5he;AtOW?^2g%!<+50CJRTr7A({x0a-CxDT=@0hHP>Phv;MV{9Z>s@o0oKAsx
zYTaXn>PYHsaw{W!zOkvdf}hZ3cT`z^F(6%!i?0MO-<A0y?GUUOM|}kZC=?FLhY&f&
zb!Rec8oz0rv1oc5<$rd4mf^*lt)qhWa*px$x><?A->zn0mC5yXDetD=^@hLW%Ita=
zxV=bd@Hu6vq+1QlmoSl~HKO>N)TWpH(OCisc?V7mt)I-uC>_`JdQ;S2bPw9dy!8Vk
zeOLCUst%Y@_PGNT@wKX)VUpn$VKU4QBl^fP-5S3ut->~pq1F00r;28`({&Au)=@^4
z<s+Nj4Ak9Jogz)pkcJPBQ>I5Oe10AQP@<KMSIH$~ferOF8N~{h0cUx+cBs;6`}8vv
zof%=JtfQlt$`lsUCY>N(4uMXm+MjRntk3&PtnZlOm?^a&FUI}NK}y9Ly3WJpr0rGM
z!S2XT1oJBgPfv1>W3N}t>Z*TDENEWSMjWG#IMoM=7RGlqa($BZO@j{3ZZ^(<pi|r1
zab=c8uJSZBJS>-gJ%)0VgyQjxmk?6J=j@C(lEc=PzjeQ$G4EZ6AL1=0N(U-Z(QLD>
z=rCn%LefZuE=;XK@fgNKYTt^gSKqw+|3LwoNpgP^LT%xis7cuZqQT&6HEI8U#~!Y~
z(OMcDcy+nxUWJROCg8m4geAM{m+^BV@Xt|uRVb=rSy_{v<M%_;hcZ6b{cAF}$VDF2
zoWZb|qt{=S<ZIfR4(ckyQpcM%LiLoaqaZ(oBsiSXM<QN3u46GJNV4Ggyj*bbB*r(u
z15)}rb+|rAJ^zQpN0WVK%k*rX9A)6F^mccF!$%{5qe^~e8mNA3YWC9iaK4^i)c=p|
z1KNhssX*VLgRo6M?98-%eu~qJyvKhZoaQdpV)!$t5u4v~>O^?y6eTulkY;mObdtf3
z>llvN3rVf-fj^M^8S2Lk=s&JM3{Rq%xn}N*ntdL}7f56>rC#Fj?Bi#)m<u@{EmHEo
zB}YOWE!P^r6)+WIA`&y?v>MoB`dS{&$wlQCRhAwC4VINtVE^_cwVPNzzu01}fj3Sz
z+C>CukKB9w`>LyPF<U)m?7b%Fi61OhKqaDI1uE7*7~b~2T5>dvF0>$#t1pd49!{wU
zW+g13kCu4EZ=GEvJ1ohfazDRVMV?8_KN$Qq_{no{Dv&t{`E2v7^1d-$=U|8!*a8^^
z5fhx}wCLd@RnpO^T`sR0@tdpUUaTYhmXbNy|7)YNEMt^G9Hd0JETvKgnGGp=!5yA;
zkGS?5PLl4~Lke8}5&{(_&(k9tde1=Nl!5YT|Axi4(HD^l_XIY`MK8ieHjQ7OMx*7<
z#_}%YlK~r<UH%crf~J^O5&b8jv8?j4cP&ymiN5JWreF(LCe(0w`IIKGCbW1H)A5UG
z*E@0n1RYA|IFtGJ&&aF;xp%!YiwBBL1l|A_t7h!_x68NNvq8Rlg#%R)e||7b&n_sy
z=D#c}arKR=mI?2$TI8;muG<Q1IEZf^Pn&WO0}vbuHr<Hn6@eXU$4hXwA)N4|N)b`;
z_kI~Hh*{?eaq=0?u%iAH=a^Q>Ar9>?H|}T1ZfN^F+P|#b57QUfDRQnS6EGcLQ;yWv
z>+`H}Laql5)OmR|y;F4tonywBCPj|mf2>6ECgwc#baj7@eVf%x{clMv`ElN7Z~vE;
z02Ksx{w0lYG`pE*qP34~G*rdjuu$$1i4t1@hwVB@E<~99s3tzRlNY`vMNxGxe&pOr
z80WX_;ri41VE><IvxC2116s^vRL{EUVM=w-N!||2;|BrPL#x(}GMOEsn<tmSzT2lq
z^tQFl2XG6Cg{z_IW#l7`S|@{4h>XA=03q#}x<wv9pW<#XT8D{qoW9Z0{#sJ|79lkv
zQ8g6>g%-r|j+XDkl}hOq`S&gwOf_tP%ljS2Gx%KgHc2F&KLp>LzrkHyq3xXoyx6xB
zX-p9_76;WgzdKHmGUsdOeVfRX7~L<2f5l4mFAa_7aCY65HN*t`^bv6t47~uS_EEoT
z>c~c7A)Ot9{8KaEx^@~aDV4HkXw@%$jbO6xo#-<HTSG%vZQ1+siAw#=RpA1ZUmUIv
zEy|{GJZ2Zgw>MikXOFL!^ns(>`q#h7d;VJ}HPqp=uA=_AI>W>CYVvCG_;1?w@`&QX
zJE7gLsF3v3u}cRp!I8IUpzadDC@1hDf?Tzl*mtaE{`vwwLP-b$dC!or<-}6ItEhT?
zw#Z)^9Q3f>o138NhNC$hM`W?zzE)L9M#`pqrG)YV=lk_7IfIOhqt~uHDQ7S^^d4M=
zD@{j6kT0J#8J4QWP*awDa)Xhih!*+8R5~!=wa|KT!PIe!HOrwt5(b3$l#)4%wY0g?
z*x844h-4A%*qyTgOExa97@V%I(8d~0i2CeMO8tU+lQfIlGSJ-dr2v}5(Ij@mvKM^(
zL~X#|5X~p!Z{x%<#SK!~t0a?{XqjxRZw!_-ER*HsBx=CuS}!!9>P5m6*B2)7*~T3W
z_7psFKqdO*ls+5+AVPCZ#_nmex+8drss>(=C8?;u1<<L2T{L+IPLvN>dl?APvA9O{
zSE}5|?_fXI=eT)1^tY^VpP&k#8tUH3n3X@9WqwHH9vkwZjjD6wqa7&Nfc0_%^)reH
z=1pYbSSw`qx%Fz8p5Jzz(oO<-S+Z)5edkr@_XooR+|&b$x1#NbM{{;5+lRO39k*xr
zw5kF3%n$AtzaL-D@0!h-VMaaxj!*W|@h75lMR9R@)he3|=)C`njHMQK$Mn9hGW!hc
zm>DmmUD#liseM_2t0futCf_h42o~L`UZ6{+!T6GStzt%SW|;F~7e|)Coq<P1ghDA_
z=L%km-}(aUZkp?=w(cI5rKY1hXo&#dsZ9#@5f0m<xz-Jqot@3z%90cf2RiCq@5}xb
z&57uOyN)O=WOG#v8h+u2bx$u>yKko371@6Yn~5StFiYvxlP85BLiO)4z$0RyL<&i^
zkGQPkeh9}6Sz-B1wxM6&a_z{c6QwPEmDE#S)&}@rOwy^B5nGT^m6h9IB{$Efk9v@Q
z>I)sG`?k{P7l}}Lrk}koB_oxUbqy51RLoXE57z8zQI3l4fbHJUGWFtj*m<u*bOb~`
zr{zpzi~HFB5ooY>+{)13?&;x4Sew?fs=s?*QaMWLIe*l5eNYLoFD;DpqZ(ed<Lo%c
zoB{E+)?dEr>hK$tO*1u`-u}Cp``ap(SIs$DQ8GW6g2aj*<lXKLEyP%7S0KXP?^OU-
zCwc8nuGru&g-0>vL|iXBtpw<qJy6E6a-<Qzp;QVqKZso(ZfTVmO!az;F(W^l5Zz@<
zR(~Ukx-SPRQ>gR8UVsR^vg-t>?kM7z;sJ}W4`?!|lc#dz@0!<*SQhvn=qF(drHiYT
zR*7oq0B3&Hf6FnYmxWbZ?DwLVr&D^aUpft&=5ENcECyt<A6hjl%u(t<26%T=xH!@9
zX*Ay!Mx151IV$fr{gD{}nLPJy&txpD$BdI$*)eALBzKu37%B>$zQ+$Bk<6&K0<kH5
zR^~d%tVCb}9V!oBy${p{$)piW8>mJBCCgZ6hX?pqqw{BG?Hs;Lx%1fM;qKqIqn%55
zc?(jFIO*Oi6`ErC!ZZ=X^)_SHrzIJ%vQ&2i6BRB?BHKcZDX{%Kc8jcW(CW(z@S%tj
zOQi?uOx-aoMzzvU?R}gMp~4yie*YIMtEOs&i1({EPMY+qUL1<FUM7E?Sn`2SfiI?+
z6aW3oZZ<t5JM(V~s_w|v#FnoxM98A*ifq5iD9{i>X6tB!Sw2>?qFEEwFovLaujmX|
zBRLtMf|lXn&ZYVE`jElDg=hJk+v#b$#}e0LA6)I4nIdOhm`%kCSV-c4l*b6lC+0+d
z2!*Kck9Fcmx(JgJyDn`!Qf-ygIMzG<RRiaE50-y+&PdFrx!ZGQr{}NCJDO@y1K3$t
zsavLwi;BWtPvPrv(IB76o{6Ayo-_Rh+W)%30wnTf0*-O_^%EXT`FTR0ijF-Wd1E=;
zT;DDZT)@=T`1d@Mg&SOedczNh7VPp+IS2unH5S}GY3bDFEUV)hDgWSAu0_T8G))IL
z&kL_BD6yCQ^46xtxWu3|Q=c^aE!>DU`0yrd;5sfSwr?}VV1l8ljb^M)iYvQcz!x&G
zz8)EXEtCV?{qnuGUTWe`p7&7qV@b7H!G)d{RFuSQq*iye46Okj1EWUu)jI<W>dD;2
ztVC3u-)|n<TB^lPioDq96-k$E#yl)8h&Ga3R@*{;;J?t_Uv&tYR8PkAj!bi)$aR-6
zmqm+(PepW(x-MCKT0AfD)UlMl$w?2hTTEm7WzQbo66EN-T=m!N{0I*(L!x$O%>s5N
zmjg!Tt~~P2{{jmMd;IJ8wQY5G9LMpLT>9|fJT)v?e!e+P)nL1#;Zc081__=(;Ca2}
zUBfwCVvLfCs^om%+R|vKw>)-$|C+Chh)TAtg`Mq*C9rE2joM`BzIi4c_`F_s<Mb2V
z$Cs`|k$C~<F)TMRGUppO>KXwF$*k6ktsHBE){{7y8$e^F&1DuWe>bGuJ>SWAyMTg#
zvIw%ErfeHW1mcGfnPcgC<QB1Xt$2)}jw+ZTOW*oYFkY>#Ufa+h@DJ1-`Kuv5^{-F!
zb2D74KT%SCLxS|x{#pA4Q<k2n&x1E?ycItT_X)DS>+t_8dAnq^*m3jwPuA|0+stS*
zY1on-unYKle}D1)t4-?32%^x`=8oB@O{8emXEhsqdbrYNAZzQB7Ejrff`W}Y*EKmr
zK5y_|m;MYJJQcEHq5#2L9+|b1HKv6}NEU(yQ|$|tpEUaTv^N=6MX!37RzxU-4?@u@
zRE71YWJx_226|NW<9GLF<7pmrDc@a#y(anz_{g*e{JT2RcU`!tmARfiI%f_aOkIJ9
z^=gz_*5OSxT@BJBb$BG6go39rs`ZRpR6rKN;FbHpr@4jXOYAGKba;6UCbVR5i7<eL
z;nlv&*i$oC<%ja&Ck!U6`QVV}4eFL0DLC_=I|+LAcjmx*gGP8_LxZj^e+G|j&-`m3
z85VWcUak-3eR_lVbMeIPKOu?5T6TRl0t9FjBK90~A6k?ORj^^#qh#yld|u^9{ak^I
zR*+;JdA1T-y**X$Z@K)&?4OR$^IbALvb^*pp`$S>m?oSwRbeaSp}fQ=C;D2Xr*Go1
zw?tUY=1Oc$y;t_s+S-o*na2sQmUd9U&bn(yqK`jwNVqpEBfPx=uVuPAqn{_Qa=Hn)
z-Iwv*Ju!;$4mv8{PVGcxxVZdvebjc<s`kHRx6^>Oyo)P(+`qC3ncKtHB>JZl;~Fgn
z*<4Qv#6Cej&b?*VG&xq%c0<9o)=dZ)EUiH6pX)_F$tr9|+?B$*q$Co33$g%oir!X>
zA_N%K;U!6|tZ(@mM1BKqQf7LrUZtXpJ1h;nbh@v($yt-Emj64qG8G9?YybMUFeT<v
zGk!fr5i3IYj?zkrD%9B!1R>gf@yTPSxp*S;t$iE<64vNUCWm;TLS{+ih=S;3!3dS_
zmeCS=)$U?^Ej^x47A$F<wk!bd28{U<_IfCXGHRyfR!%RD+AfZMPPqE?-`!o)Ht&y)
zQa3k7u5IW2Ot(Y8KNjj9rAFZ5*!P7|>PQ;Z@zQSFpKrHD9Vhjx4yL+_vC@|-TA^SS
zHNt{|-g+w&<Ft|9jf3ZVS}R}~-@QNmN_%TN>srH+1MAp;*~e0!f1d;U?h*#OP6^(p
zN}pt*C$V&YHFKroJLB%didDbXRj~gwP@33@zg;l92*5Y4t{gQe%5{&LCa!;;uI+M;
zQTUu$?u<>9ZN%t4WBWp`^o3GDZi7^nt%BhVTuHl1VdAUtq*8hjbH9P0hI*ueizgYM
zFO!9F>WkoauY^*k_!L+Ht~75-x{-p#1qdv{%k}z{<q9Kn?m%Lt_sgY+WKL;|mKVD)
z`LF-apPlvUk6Up@0a&K9#g5DK$*u<T%@d2sPn~qWf4)!eNPo^1C{u*vU!lA`YX<=n
z{4wNrn~%&`GfUw9fxD@^?6YD*mu=d8O-_Nn(WF<gxOpX}7Yt;;sW*iSt`-05m3i{T
zX$=3#-JVSg(;2U7tk5td5hQH*nu`DeNyBU@=hMX}S%x!Evi5(k>I?CTyRS`_Kf66f
z@8P&p;Tq^r*H-J`v@l<8KU^Tc)|;ru%0-A}z#LQ9JQ!J(661eL38zuzXxgMuz|s3H
zK1HTc@N6>$%L0V;aCDVO6l*jYriCSG_|Cx6EY*Eh1QDgI6$1Ki({sMg8Ml2x^3htD
z%UUfx^8p)5Ptj!73;SF)+u9(#k0G>{(BK}7h+tx>8zifc4A#C%qY)q_<Rb2zPwWMK
zdoagJ70N2703-f$;gw9tDn{RGMq8n%{XD9o(WT?O58!V*UMqEN-S5VwIVv1m@<28)
zJpA7hwHea6zDnuPamBDv;vava=w`!RCvwQv!a`hI+hp4%OWx<)*7GdUhgQ|1(h|*a
zUlh$jU_!PKEvN0iKz9jo$+<;_g$|F~n*g6r@_lnswp&rK%3=2=F=^>X@%>>piIn_&
z{B+Z=fjgO;&!S*SwIYN*c;4a5C|hC}p*EKyjFK2KM$aRc%C7VpTCA0fuC#Q1LUh0w
zUSehbZM@!jJ1j+&aG<lsF1y(q>Sl?86s3BPzYOx%5w3EVvwhSB9z`8_SyuF?)jL;K
zJA;sQ<JrBCdr#6}y%6Gi@ULJCa9j-gNLwvXrky(%z#jpVe;S@@g>pt#hb54~yrzXa
zC!`6YQ>p6OUi`vh2g$(u$u&8r!@86OF_diP?@jE(RCMSqb8Q5E&gNn>Nb!gs6G&z;
z5(a<e>1B{B^>GSLh5gxO7_xtLUsUGmU*Y81+5pQU7Pqi~n=_$Z-Ju6gsgpi&XabL~
zFI=xI%LWbfl4<@)O^4!7fS3QR_=+(%*`=!Cg4S^b!3P@~|DiJ{^k*qVMFvcsEFJ#m
zS?{2qj=tI0TqPR>mCqB0f=2QCOJ=U?`N=<Kzt8P%XC|VHJeHg2Z@c)fk{$5F5SxP+
zHx6C`ZJE5PbMSlGXkF%~p=?8JO27lek}(e}H{GBL0^+2)lB47XDoQ(fW@%o@gq6tS
zDx@`FS*xd2y=2INybK1#dR-eg-tPZd0M2az2dvIBn@hN>VAf*GZ4o#>dranNPBiO!
zZth?UeXRR*s>me5De+|{|FlMF28Wtfz@Agi>4okzjjl;0-zkN6eBSsgz^yQDGG!nm
zl7yszK-PhYLbgoA?ED<?2w3NsRyJ$>c&n+Rqn(O3gwgt$ctf0wY8^=ya;fmW@qu|y
z+||!T`7PtziOC3K4yt=v8^R-(A3>W^_aSU-0y;Eop^naE&~FLM2$*VA5hG&ZPd{FD
zTa@O1hV;?Y&`5(3vNr%da<)DJ6|>Xw3S26x#p)$*v!B9sX}z10dpSeUl`^<_hMkSw
z!8herc)^}u1K;=lX2@r+D)Oo*oS-j@l;9nb<g3v)EbkP4z{E)>YEJXXs}}dlHQ4HH
z?#liLGTb;1JlL2_L03Jy16gguosK_Xqgrk@*Dg@t3hDK?dIAGb2BF*82(Jz!GO}0!
z*?mXuDCN#?e1vhPiL6txL>qOiZ6ZTYNLW+{^h%T)&7_<CB==5;Rk>2xh}mM5jgoI9
zmgB6ZpapO3&9$*#>psLMaZX0(EID<xr*{W`f~B|yUA$SmITdyJ-Fkt;?+~WdryHEx
zIuTwKRPqE}0m1-b=61O}%O{n*`Iq0@3e&Am_WvE@iPJ~ye4lBf=H?}TwYkT6Cfj-V
zd*cixch8Dt6)Lw4Rg>;HW3$7kL6E@g5@{l~8DV_8Tv5An@BX-W@`Ok({@h6fWzgcj
z2W#_C)TZ-#!AMr5BkitB2TvCfaWgMIF6WHW&IgsHzyPi_#me7nzK}N5a6hr+#u-7I
zbC^c0?lxPkwjfa?jS$jfsiZ7TO&+Y1^u9c&|B^HO$2?7i>esJC5s<(oM1eQvg93b{
zU=Yo$<p#zC$i1bT;iiuAE;MCK<WHDr6gg!RR5h6)C;A|5;xD)HDt&d;J*h|u?%}!I
zWXgXJ2Yg(|y4s_(b<$z_6xE;ekR~$Cr%N~2n|-Uu02Za##S4tj%?_}=2%o;{3X2Tu
zTB4r5+dIgnARim$sox+n!SP0QrKdEb!aB9tP2+06jU%b<%Zhu$=g;bBHT!3~fo^Oc
zJ)Em~RKo1(XH=Pbcp*hJ1(lTvc(R|!@82ESK=^K;l*Ajp1NeW-PVMNJt@8VOANRY4
zEZoOy%o!%=TbNA7;vq%7{pTTXtr5ZD^6W)oV!^Czo>Ztp#ZWjRV`NS^I&B^FHC)ZJ
z{AnMixl`S|=SPD`vL_V1PSlszdgd*ZV@iCUh5EoaR=^ogEruFy&c43RFyoK@57l@B
z6okJM5y`K6$2aVGnwEDyY;s0ItT<&=z_&%JtD_!+I#AKdRCh>w+h)unV_1HtFI_X?
zewl)HJ@ldy`{DHA?k<Oi${^FME8O4R-~IITY_V+K!O^h-J61XkcQb`|f3YCz?B2ki
zH7>iu&ZL9Zw}Jl{>S>q-grcC<Cg$+~f)k@22dy$dgE|QqWWjPeL&A>p=|lo<n4R{+
zemq_&NQx&tH3ytu&PZ?Y^wDu{Ay_RmmGw~I`DO4k<pjG_ZKM=QDo(HwXZM`lv0MU_
ztXh2djP3riY_c$p7?N+&k%Qm_hh)vlGA8P9TYf5x6P|c-<RBOm8VkqkuK~_@-XbR_
zqnQ_FYN}~4gWNNxY0^F@krE{+Dk&H015b{fe9bkhn5#e&2qLtJr<r|`j`$vbR?WY*
zP5%nK6!s%j%fX)E!`!>q7N5NW!jd3Um;ZQu`Jon9HNtX~vagZ%xFR}#A~aNy;neS`
zyBabJo;7}YWHfef`Q~PPXb6~|w6hdr_C~kzZgV7P25TozMbDR0jc*TX30I-pS{Jbl
zbd&AJ7d*IGqp<PC%f61ItCNM%Z-i=IwvK@it`G9cs*Ln0u(L^_ezY+ZU8`<E6EHA-
zawx1#(`ux-U5DZOhc5r@p<q>Kid~mPK425UJkhZ-rPbBd#YotD0&h@fG6j{YOj0VU
z5>Z+Vt*`nqhD@JHzGl;bQr{{ivkS~Ki#WHuqvcaj^28kp8R@9Ss?}3<PKwLFHB@=Z
zHc(QA)=+|B`!Tvtiea)LWXx;<Oo20X_Q}dh&@``H0?g#^LAj^RoFfefus7CIe|)FW
zDM(Lilz#a>WvT!i=;tjsQR1;hvDUY>`2>2e$zMg7UE!G})KK%*`XOd?FHJ?nnaEzz
zQQv<{`*ycW-(0Sd3A+Tt6fzh8Ql=spEKC~NzB>loq0y@4pt7D*EA@zn*Q#1DMHcdK
zd6`&SVt<P<wjW;EfEAZ{;nyp>2CHSB!q2!_UQq{9f24K-v>PjEe@R5!i3Sv|OJDZI
zS!p%dP*#+CE#B^)`FnjlC<XM?N^3zpM6Sov80mKslkmnybPjg&{*zkQ-zUPq2mh&Z
z>~3s*U3WP?uU}XxuJPYgT|bpO8&;BdOdggPT;O=js+)yqcXotiSl^LdS2T%uC-MxG
z=o-%&pM6>iNVb{%`o)m$^+tDTx$S5xKmXraT6I((R`+~Nxio<<dZeUHQf4T!Eon4n
zdD_vhk=9FP9(Huu^0Amv;~M?$u?!yb-g-1S=XS(>=W!l=ktC|Vcsa)UWwoD#t2XBq
zd~Cxq7B}G1xHuLdaCovkD|5LiN^#mC-q5MOz&kz;uzWU?BJF~1G{>`VKQe8LlE~Bb
zf)It)2`ZINFrq<uQIbdVXRVKG{8>I92~<_Ll4h>=akKvZ{&BniHNOZ#9U(%*#0EyG
z*-RO4y-k!WfT^2!iZsNIQr%$#D@qg=DZ`a>QXeAkhY*v}W*I{sza3%Ip(@HRieaaa
zNB?{PL`B4M5)|7`N}LYLCu^u-hNP5jO%OFFe95eZgdW=X{hQpEWrj>Rmrge2>?-m2
z<q4I>y=27Sm*M*v4-Nh$QIKeZ<p|-daIRYi6++53itR-mH_+F&$zdALv>$#Um|OV8
zjtofX?7kdimal_e{adN~2`6D?H+<5I!n2D3Q@^t5q4^S%@%4@)?3rclR+e^6dDTL*
zm$!d|(BsyVw!NT(*Jh(fyXN1|#w96=g%K++?krnKZ!@4?q*-5{8ZBb64~@7IY%^rP
zyogmkFJ1IKz*maV4<Nl5Oi=Oz6?-FIv<8dQ>y_bhg%E>uN^^%%bbatU`Usk(EJ{}3
zlZrxDZL%uICnE7ZA<H5K(7S-#@{+r(FsF({n@tCU_(tzKt#RV*R|fi&U51WS3+WZ{
zI)*%p4Xj%CCdX>~{SngN>EIIEQ#7q-gKDrR3fA~3V{n=*)t~}zxQBDB26sx4y+L|c
zKoUDM8|dy{;PhHs>xbNjiG##@##*WXR4cX<0e#K`{ySf8kh<AWs^BxPWty9pfzx#^
z)TLunO0D{!0FpK|a&bDl=#g8%;K%!`*e(tyuAFq!IBZKtRu7RrxVc4az<FPW)>2p`
zMI=*>W?nAenj-esKtkT{?>AR}x@I)hn26T6+TDfm1vW5oxIf@?)HNNY^cV27x3t`g
zWRaoiYx0sN+;cgRxtyv$P@4L5o7X7n*TCB{eu*(`5npgN%ll|!{YG6cfg7aoDufkE
z$fh2XOdFrrl0_1)^^gCc*D`VpEGd<2r9;C-tZqcadaO(QHb>WBt|Ra|ps{29^8#wM
z>ly1xR4_QBrc`uy<9F>YZV|X?U(^SD&kk0;nDF>$*i!2aalbTXidTA5oI3oBGr;?=
z+}EEo#p}6wvq51kN1I`7*9Q@{B=6B*_gVp=(hv7YlAdj5QN$#s=ep*0_}hEzTs0r;
zWO<Cw3^ir0a_fKZUc-x2KGPx#r#E}b&b(TR$sL)*YQDvnaeS9S29E)2TwZsHsmev1
zP~@Km1_0x#y`Rz)QLlE)y`1|qxO<wVr5Ey6!dh5bPa;tH&i{t{{pUtv%FR_)y0vmD
z7H-%Ehbndbvh-(^a5B=LXZUCqG&D|o&URM}^z;<B8P#b_khV={o5xP<Q`acx4aJoW
zU*P#hl0^>JUb40@auk4_v*{PP7ct0N5y*#GrEym;`oG`)Z?)&rry#VI^SMwGHJbxU
zFa^E>=^j>RkVgmz3gxS6&-bUqfxo1AkIxiNca5)|WVkF|ACJ3B+*Of7II&w%a|jPR
z=z94Vlw~z8&k@PdDjHaYq2G-hzQ>qN>*~iki%YD(p?dGhhXLZM@8qeioj4KYkCQ*_
z(aRC!&w8`+B`H;)JNGG+l*^iClPpeU+Lv%LYtrvMb?B($ti&EQn6<F4%U@>uS3?8?
zmHqR&|AvBLN(%8@+CL+%tp({avs8qtN0CH%FI<8eGDt#FbyO)qis5O+1x4jVh4Y`7
z@;xNiNth((QxAc;RQk0{9rLLUZ{=##7}+16_G=Pgf2^bQ*%!;s{~IRIm68;fmX-jc
zuFY-^^Um%&M{^6;zONR_i%X{A7~r^eX{_2_E}YgkYdQEkCZ8Drb9jFjnYGa~)ii0<
z+~DznGF#5p&0SF9rPzj<6w7uqthu4q!}pNJ*$>#6AM3i%gM3{-IvMkDk)VvWE;ZD!
z<sy=|Tf47THt4j>boaT4P=EMfXx_aS(Xx~!y<B}#BZ|EYNb3fBZ07f^Q#Bjgk%+0E
zo2fx-gl=zU6=>cN(V&tJ)$VY1h7Srfbj`KOsvPyz4Wc1|Y9ZqGsE1$d9j!ea$wOqL
zRha;dvXgdB=OkN*euyYUBV3C{u^Uo2;V~jdqaDj7$YW`##>MU|MDRleW5f0I6HG;J
z$*S&`u_OO(CB0&Ods72&+g`s~V2^G!{#f&^$uYY?5Z5T``a6?9=XN&_Pwd7QjbuKa
zT`n3e{(5I*#6))27T=j@6gA>?`R(uh@gp#O+RW>=dMMOgsYNNTn#Xji2mMRd;@ESD
zhu|_XA>aG>HaJ}}eY8f8(DUE?NAF`|#~VY)j^JF2bnU2jVm$p&2;6i{XBAoNYL%A8
zXecv?2?)NT>iI$yxpb+|R~Tx_^w;U9^veZK5j(+WzppHu>c{gNwfWiEM?i)wla0Wi
zze{A*f8D}!==q?txTe0}ksGw69?!#--VlAR&zJkmANXEqY$^t1DlD)3mHz2m7st?|
zuHc>1;SE8{s=S?4dW*m2YJ*8n*(bxf%3M5|k_AgS@^cHwL_~Ky$HkpdDDoEF{ga|c
z*SJ};%Sp3qMIT?k8U)F5HrIp5h#GFdt&+n-!`SK}n?L>9ebEgcvVs|rf7b6&y6MZV
z&Ql(ZgP2bX&E5k4!ou1tiB$^)Ek=w93E)nxlYU97Ah3#&B)!Ya%d5?<6l<_<BRWey
zbaZ~yimQh8oI#t~o{&N%l$DpuO`dL#cCATpqtL_ju;BakUmlX`mp5}~Hw_2CSb_ak
z37oKhZGNSH&2W3;6`8%b*TXW6(o%`All%GsOh!6h=<+YTu3LA(@RXPhA`{iuv!m(v
z%>uyr@UE5Te@~2MEr?*jMX`d4d`S<bK%CkF_h67d)Y+XEzl^biM}aJ7#IiiAtv-{;
z4&Tkw5&jAqd>1;rqLr(pH#`&|?6&vHA5JnLxX%h1wk7+Lsu!iG8YM0l&C3hQ{=tf<
zG$s6xhzMW*)~(+7+)?9kGc0s&PF{z<<Y_QbXRvHXJOBb+bX@b_jEI^7;vZ6&#Kj~s
z^WwHPodj~mN(Oq-;(s&z(lfr-%*FivjC|DSjNMLW`G~iC^Yfdv!hx@H1;mdy3yKrF
ztwivTil#Iw?U6K&Npytn2(5RVk|2$n|J~$%UOvrqmO!WQyN*1I*PW~e`OG%QB&?g5
z(pR5hN7$DxAqoZOqSVPD{1OdCfIWsi!a6_yB56$bO8vCz&Mk|)`Qk07CSrsMN8`35
z8csQD&zX}eef**QKsxqUopamsg<Z7$o6y#craHT}#;hzpM%m@C>1e?UO7RbasDord
z!XCIRQIq<I-R%OMzsaD?(XR|1cTNgWuPRCJ!(Z(1Hgc=V%G;3U#_9IId%%iTRTfZz
zX-7q=H|;|S<)g>kaSgwBoL}(4%oT{9lFC`XW2ivX_n%wn0GNN6iJ-&&y+HfD&7e}=
zw0-w>eB5|#a&l90FKBAVC;h~j=VY^#rQ_l-28-P`%JCKp)|5W~UBpBs-68P%r6fl2
z)9Cd{{En;CuZ8KXpi3l$TIl@<Hfx*qcHh;hkw$(fzsA5aQCP>#tV~XWVM^A|n!M4H
z62FCigP)BXfpW&ZY??~#j@yXY(d&gsj%BR^Lo=SgbRstfDA9t%9pm@XdHo~SOfX&R
zCSzVqNYF(sMc})@&2z`<#r9vQ)EPsaE?h)#uu8wMx>)W{O^S8yA9yh6#_RccmEJFM
z!hP(`ulw7}#wK2UubyJoslF%q*(>L(LY+C4T)1ojE8v-B<9zV%k02S^qHr{>xvt5D
zme<ngDRtl-<Q7VEPs-zZ3pd>xV*7@K5$mnJT3jNfC<`|*6~|9|8NF#BfG89+W|KO5
zs&>#BoQPMg21-jvTYg~(1*O#7dkA(;3KN**4|zT@y5;{PKoh~=c6fpRmLqp%jJ^0m
z!B<mO0F^iw>Ky}3K9zS!grW}Q6T@WwkeVt1%)vVBb^70m>h?eBDS5!z-l@g|j@n80
zc2m!0-D=IR_}9sAvESvzC|XUwXLqf57xO^ZWK_r*EE^1_ecB0u5=F@qD!qMd6_p$@
zXpeyq7dCOLny<g<NseZlkjzebAT4l5K!}tRc!c#9$kJ6Dc(G$8m^)^pz7uM>@oVvx
z1h_ifHuJrVlre`ZF+HFpdI1SBtY-`cciEn*f{0@))0B+KFVZGW85C8*6%ks{!*Wq$
zy%|lziUM@opNsj6@m0r$4NQp99+}-*oqIJKc(_{s^mv4SL1WVKl~<c*T)>6r**Fr=
zti*IRo&0eV$#EB)b@@%@V63KEmSN`Kw73VX@|gJ@Z|u+Lw}(`i{4FPsQE@cEn;ml{
zftwUf=>-|FuiQqds|x`X;P(OR@{O~8e%JrS?!}MC<2#a>Z1-XZs3GQbmg;lc&2!xW
zW|DZ)p7yNBgi0BM#i#iI4ni&gN<Uz4xL~kyEbgu>N<CHp)>Q(Jh0mwzlo~yBw1^%h
z0Hx{(^fM5azU6n6A3TzJnyEq0-bcg6THEi^NNsV9W3p^LQ^hZq9&WzB9+HvQfSD|*
zICQLHPjU3aFiP+CvhDExcVVoEM4fLK#zIE=IuX_48G8kur(fj3glPYWBmj|x)Z}yv
z3~biTHU+NtkGldV>!iUPrPN6l>Ak$8;jGD@dO|SAcqQ6`|0$;4HdhW*adLf0anzcV
z*4K~eFCL)}XdB1MTp!?FmMjZxmqTur%SfL^tjuHb$nCsYW7*KkoQ&o>`C~B!AvVG}
z4bCq|PYlgSM|(e{mu=p|=YJ1AHcFw~=Wsa0Sn@*{RB`+;R;84POVRNeNE{B%28ong
zw6=)m4e(!Y%FMr%YX9)ZA?@s<RQbH)`gQii9s95!GrwZAYTX|3dy+5}1TNJqAxS9m
zq&{AC$MN(N#S1iyc1nNP?kPAaE%cTk7nut!n-V)qVXlK`Qm-KZBLe}IR%$<^oPgIr
z0`}1tBfY0F3{jRuR>a2!B~z-R!y^mmsjDvE!w>I%Uvx9+txU}~35rtqthX7>V%~DH
z>~f1y{mgpLPt}?$f(zc%yNh64w?xQtqm?JfbZm<CQXfY1-H^KdyS=3PJeH%Z=}qF;
zb8!6EDzL%trnN>1pwE6m-n?*N@SIVAKgxVHleZ=8D5L6gi<_5-_^V-AmCUvT)y=d*
zHODq|g#!v7L16u0Wup<gqN^(43$LV|K`9r-p^)36c`^$D{>^8;w<~s*wvIxg<NRV;
zs!^{hb0Qk_&M>pye0#=iS>z4quI6WgZ=(sEODu$=-YoP3bKT+tj|DN0hC;cz2svbC
zttma%e&|`-e;E}vP}kK}C3RkwC&LBYs}vKc^>C?q2R_;jQyx$*jgj!ai*&u(%xgVw
z>A<3z6W8MqRB|Z2OeTVR(5rGL69Ot$C@E;jwpH<6mb%~5r61*CM9-55pK^i_KO1W7
zTMM1m4`)1pb$-Iv2sYW?!A`EH>2-`9!S{lm{UnPLFK+NUnp;^}@pRo=`uFeO<TZ&w
z`YMJ~K<T|M&2{|?cEIKGYFU{gO_9aC$ti7Wys3QJ-dOds-g@HA#l`L}UrRWGiOF&N
zg=*QVg#pc6H2)UZ)mXC8$#!U_WP{F6Z^EIX_BZ$JOFpUrbRCYkhxws}yuB3Vs^3HT
zx_QH#vGpUGC$dR;v?Rl<+@R<r;-41UqT{rR1Mgxw$|ar{&g7z4`*||V9}v(2I^zcE
z_Naii+%b>1p+@o;&frz_18ruvittns#yA0`hsAYAD0yww;@%O(xPCSV2py#OEw+Id
zuXc|(d>vC)ImIp;(>(W5jQQ;<+PmnP;$1enxBdR~jO*Y|M)#_@U+*q=DQ+fYt~qY^
zcNbbHSfqpSut=QWIw~^3TT~SA86Q_NyPC?dAiMsvv^6uL>65o+ZC|@}fx^bc*xD{n
zI0~-#mAY`fVo0AEkKnmnUheuN7?GPBOJY)G`vyDIU#nf6w<y(Nsb<)3liZC$o!DYx
z+CCg=Ft@Ddwt4yPD^S<;3hwcyqmFd4uI9Svpl2?NS3?6n{H9H(R)Rp&&x0*wUPNhv
zr9&(?>#G!aI#=e=7vUKHhvA3s#SO-(+{&?M()0TLJCD`N#%4t&j{KA8tr1*A64H(@
z`V6f(rnHW?8}t=K5-$v06<TbbQJsmmhO9L-WP@5~DYbdz6yAPTNv0859*nkR9lE%1
zl4v`<;8E_W6Z*x>!cPHcDD{ZCn4MSO&fyUgGc9|!)%YpcZvn?i)9iKR!B%=EzviT0
zN-$11+?BPg9VK+n!uDBEwr$fGG#ceH#~m~(Rx8L`J0m9741hM^xdb|^y!k1oce}Kt
zk!-gd;Alvq0G}*{c^z7fR7~J-i;K?JQX{Hxd~f=jYQtAEc>ttmaXh?_-eNe$Ij%#a
zii!|gy^&u~eruSc%&AD3+s(Fx1=RhFmbNY{34XOC9n$exuZ2y|)Ujqq$L6nSjoI`^
zN^)9cpd#WYcShG@4xHv2UbBQH)>!k8yp7dZPp`KRMs30>XwtqIXocDp)}guK6=aI`
zim*?d!CofxfS<R-W8_%rs?u%F;`y?E$JNZZxrt+8v<BN48|Rn0XecQMq4Bw`RehOl
zNG}2m>*!XC26x)tA)(ZXR)BSb3hLz*tXKa170yEfWgQxRGrZ}b&@W;`sycKCP%!7{
z?yUYj?x)h<8=N1%^mjko9nj2^R_vW~5@hylb$#OoD0Ud7v-j^Gl(M}1_#M0H)q2KJ
zI=xEx3T$BNyO%{>B~Q{1TW@G+zya%Rkg2r$8T~2Kl0K)Y4Z&Oc|2EETUB&w?#I9QM
z+Gmlj&Uy<yDVBuV(Xirj-tSog0d&=MS^P+}K^0{%_?~_;-TeaYXpQN*@Rb7nCeQLv
zYN6KvVNFvMbXCvQ*`+9T%X+%mTf(dg<jUhF0OBl8?T3WkFY|H{$@}7F$^gP}Qy?nN
zAAl5uv(%i@U#L&-tte{Sp%i&gg-04h#1POsS^EVE7Ak$+NE!xw%-PR0A05ibWrxyH
z#Te4f{5&a_c$io(qLN5}_+eqt!i_b{tpc2pGK>kb+NyAQHyjrcs@{>wGTNo_Z0~0G
z_G038Q|1Oo(E(gvoFquM26zUF<nqp9M5#)vKVCc9+S*QU{u@V9@H<te8QlEcp=a6M
zLALm!K70^7%bT@(hZ#$`zNR<(b@=7xT2uPF_s0OwR-bK5{dnKa{{S#S&%SK$`0DF7
zzx&RA`!K9HWy_OEvdkEVB-eX-=c~^hpVdtZyKQq+pBYHX*eGefXqsBalcgm+osFa3
zI2@g2#rD(V{Xz_mf_nNOE#^^y(9$Toxz$;QFjw}|<!p{xrIcLrL0xZp`J?B~&yHtT
zH;47Cz0f^BpU!*GRSR*ESz(oB%qSshMHQBRCP5~&svUJHEt;mtGeTlW1|T6OP)m%0
z>=^{Hi7g|iz$zrg!J=HmwSV~Li~XK%j!!17UnX<$xhwmf?4r$x`kOC&>B^Ox)%$nn
z(sXjl%I;#hI5<AmIu{j0ArLH^r8Q>ME(>F9M#DjGQ7`iD=az2!4Hgn=L5wkMZEwzh
zmopJRIh!Z&d0~^6+;p<6C<;SFz~a3ZmR?ofc<%X6KmOq8-W?y4lFu3hSLf_i6ZLA!
zE2Wr|pb}C_lk=f%n)z(jG)-@9yKSNiep$EHX4h^#ms@-P<9AQbX4@N$?%9AYgb<ux
z9jOsl4ez3ja^AThNXlBftlQD_?BsZVba-;uESJm0L@B*e$6MF7t_9SQq0IA$Su0>r
zS%RWtm89r`G6jiniY~+$BO(yda=F|;JUE?Ax*O}?{)2CS{ab%u5;BlAMXx)^ZKc^I
zA9&Rq2qFUiufI1Uh4fkX2z&-5{?(_5pcILy6^t4`;{&OPGMY@A<-E$QQ8W&J(X3)v
zD=R2`R@e&wDaBv!x-pj9#vdN;fBNaC@4oxKZ<}k+T+OnqEK66<JGs7g?Q)*!(d6{^
zzVVHR_aEPS?#93Uo4<bT^)JPUnYA${79mB6A%q~oZ*1*URdwmorN@tVkB$#cPfxct
zx9{Km^re?y@ZR?az4t%-Xc6##_H)-;)b(;wFQ<i7DZs{{cX4M}Pfqq8KT5$Rh=k~3
zoGjZ;r$fpJVxHyRN2Lvh*r~cHMk2jf*8OfTGdjl9_y9;-8|y_;a10>Qw(VlMT!G-7
zb8XuKz{=GW+62fcveu+EX=Ml^0LTBJm*cZ3Z|&;%*)*-BV6BY^uJs6i$O5ek8yo%B
zwIKwh6%qySB1d3pTQ8B7O(rXaMTCv9BI4R$tR<q90ud!nF+>*?q0W>&9Zk2lHmA#a
zb@W5Tm||uW3a|D)M69*>_P76#6y3OSqh2<D@+bf4U;O!Bu5GLX$b3F~|Gi(Vt*z}o
zdi?TBFa6(t^KS-&!S!2zFTC*l%isOMk2~G|Gnb$Fyaq$3+v!%F@yXnn`26jic|GZG
z4Q7jRZ&0lD)u;D9Dyr<!{Yl$SZ(QHmxw27Jw$IMZ=D4i8-c?LXNfQlaX-2cD%?ev|
zkIznt2owdtZ-4y{{?+$?bUJ-Xv&;$=nd&GU_WI-5xDhNGALh+}&RVrJ^z#UZjt`o7
zx6|ugxyr{U_jdj4$nEVPRz+2xoc0BKWkJyowbBfi%bsBXq)dVK1~nqMXmIT2kd;m3
z1*K8bf+!zj&0LE_0MtlKi6C*xFj~jC#<k33gKV;9=i+wH+@nwTgv$LzZGUm+!ny6I
zXX^KU|DWoNQsgLqv0GG;=;Z8F6p~hpW=4QW#3)&2V~7NLty?LD!Fdz~5b?KF4xUnc
zy7#as%BEf9RoSho@nkd`otaD-LJHZ&V6&TTB)ddWAPgZy-$I`Cdg~XjTz&H3?r7;t
z14?3^XGA0tTi+(16d;01jDX4zY6c|Ate!aMYiEEIf<(uE+=jLZE&)ifzIN@U7Y-jh
zIvI@(OAU!g5+I`{Qzb!Ec~Pth<$<MX9k4X8T+WuKrzfXpr_FM<Sk%rXYgST8iZS{i
zNJ&7e%(jrDFfzA{<63fSa%(z8RzdIhU^jRdLLgx8>n73e@tFhj<*$A7g%@9QtrVu?
z9a3cm>zinQEA|Fy0TLJ$C=vlCl>oKUZq>O>pb3A;PXncp6p%_n4JhW6NJA1}LPDcS
ziwHxCB6+9MRyR$<A+jbE6rmW~75zA+L_|pPYkwblv78<r?tl2v`#=2Q4`Ymy;zu8S
zR1`UfptVZkL_*ut8yjms{mD<i_Lbk=KX_W?1tMO5yLPR0ZCRBuu?w{T6QZ%%d^`ex
z!^6XVzx(9jgR&^QRrj^mzECfx#b&;k&o7?aKApAY_T~ro_c05F#ms^vwKeSI#FVKE
zJHw-svo`Aavd;4^l%;nrdY|WoDDBvRteVNVEQ;WKQ5gUTA*3W#x2uSlQ;I&%@|Bx@
zv8+vhmc<wsi-p#Dbr}+41OieRIWdT9UEV1YN8u1ruP{=GxLhvtJkRqyF)O0zW2Y=b
z_^j571(niBFk8%55+pNo@FLW>x-3f((AttRB7j6Iy)vcboHND%i}$U~w6fNDFQO6v
zYRw{}>9h$Tsj*s-QV|n?c;~fM8ZmKr_0>0jfA+JV{NW${{&3iLZTPc4`?KrUpS^VP
zO21eA^<V$1UZ<R$oo@687S*?Y|L?!^i=SV<c0G!D<>gnN?(Qo2+#06pda=9rxXSn|
zUwyNv%<1e%Qk&;ly_D4WM^E;jxpl)9T=hzmv$DMV=<(Et&HjeW7O-><GiUVKXd+sF
z4~E^k^+vJL>{^Z>vmk%|um5Ilm19ivxdAAS+k(~g?VX1y-Jee&k`gclkpb34%$M_7
zn2kmq>0OCg*18L#*N210CnrwSG{wR*B-V~8@}R0Du?4U?SuELmvNUz6THm~XaH2)c
zQ(FrtNhpvMEFdO@g{Vy0ewEi}p)y5(ucFFiu}k+h^`2VJr}f&o>u&{|%-UQRYf<xL
z7Ju;l-H)HT*qly|wl7@VJv~`WXY0dZR%P?WB86p&$!JYREvB^(GU)foXc3R1I7q)=
z!EZ~W7l4!`q;xu(?BvC8ZS&#dhcR+d6cR*{F=Xp`yE1x+2m+)So%33mThHEqetPfT
z^k}b>XNEGZEg*XD1Xj8|yBMQ@LQNz>sFX5@g;G%x87%-w7!g<A!kB`f=%U!Zd=cv9
z@od^LWQ|fvDV>ATRkz<Cu9w}ewi!YK=E-O@Iz4^z<mqHOpUkGqMIC(-KyB^H#vWq~
zF>0-av1vwU$D>}Eb-P)ipjTyo>%(rhD$0IE?9HTffgA5YxQ%Jn1VMf7{I#9)SDXi0
z#AF-PDh_(xZeLsLo5m1oP{;x-2*}KdIi;i(3lbm#f+EEXDS?O(p*C748>1%_Kuk%9
zP;0HV6%lE5zdQKiYj3U%HlG|Eg2)OLCla;R0{kl#I?L>@eX{y&boBmz`|rN{?z^18
zwV_jGPxp4OUB2j+a{#W&;_Br~lj&*O$P3RsZ;ZKc;Q}hz*w{EYIIF74hY%BsNQeQV
zPr$=|e`{;2)9K7+v!ciri^aj-L7wHzFqw>*Q>W~ZsZ6)~<`>?0>-`U^D$6q*^w*B|
z9$$XtxtevqT3Z|T?mszyby2mfMs=;VZrZkSK^qf8iZSIyr4_ZUpN+?zbK5C#k>|Z`
z*Nakw5>u2$WmYTANmfXvrfHbjTD#ia=Xvg;U$rP|B&}8~d#5!lm$kLos+y{*s=BTL
zz-C#~w)K3P8Qtx6r}Z*Kmu00OxEQsyZQEv91_Cie5y=Y+07_|pLzzz!gHJIE$clA?
zh@4m(wYsfhW*<B#lTs4Y%&@4v&Py$*kW&g)EAQLFC@qOJbgJ^T*I&86`w0>K_{TrK
za`~C6>PQ0LcyIN_aQKDSUcLMA$F6n;b=50lfaRhtdz()mKWUb8jP7%r4Jm%%`K#x*
zRaq<wV>bsn25ZEBp3Rn|@<th@@bzbR%Fa+UpS5kbyP2If@<8HIbj%~u9u4!XI|wle
zq}=MZUTpO{zFmOQNAvlScXi`xk-Djy`E;>p@|&+_U^dp)9xW$zV%zHF(`7!O>sPv|
zRoTPQ>D)DwCeQL`RCQrU%ki>r`V)z&W5<iRV+o{cjt=;LIM!M~c@$__u<@n^=K8o&
zMg>+{=_r|?m1G!mq~V~SYj)FdLG`7}-E3Z!$jzGBJ7*tI(EUx6aDQ(P*m%bTdf~y%
zUE59`9sWE@T85-CFG`b_)neJqo238-m-5WC-mNf=F@;65jFGq3hA4dP()qoEr`DLy
z?Rou27a~@FoemkjST1+>_RgQ%>Gy}n$47bI5fR@6qipTd^laYE)_)c33?V2o!{OG}
z&ZT-XQCe4JxhjVc0c}XqJhTpww6R31u_gkdjLsBdnF}T$v6=A_5QT-AV-jXkm{~#$
zBCT(&HJWUU;swh7df6S4HYuf$+Ta#Phr6GC^6CA5`}ga{ClO9UWJJW~U9B`A5Rq0I
zfH(=P;3$nh8%>e@rE}}U;ieZF&6>gz*%EERFi#Q}O*>t*V5_`8V6rTRR3?b6ch0-2
z+qHS#&ZfvLDRPLyB8aQ%C}svoB2k186)7UpD@1TgA$TtWgo+55SGg-9twfWEMA5gN
zd;aEsv$u&2ds+F&wT2ben3JTGRyr_(g<l7<){h=PJ{^q^b=|a`s<W}tKY02GBs$j?
z8Fahds?!-wPNuWP{YQ_7gT2Y*tkdsYf96Hr(%uK43@{}ZQCO2QS_cjnu3R3EPdC@M
z+U0UQIV|%khFBI=uh$z6*Dhba_M@M^{b2upgD0o+jq_KVn1Xkd)7r3SH1&H`mf5Pe
zkqyQ_`0>xj2Pf3+MV7#T1RQuU=+Eb~wq5qR{pDi5wl=WV5Mc^|Se8v)Ri(Dp7?l9)
zx&dW?0FitMf`E!*^edGlaV*QSO@R=tu?oeQUW!hUR!SRdNvWo8I-Raos&4AYvazv$
zZ8PP(FSE?I!L`e7x9dHJ=+w$j7Gh*z25CZy4pE`eDS=OslK^32jES@+<`h#D;m<10
zNm+$aICw`IA~JiIX=q%%*6nd{%qh=oN{obUs9$>dcGb^Qk}tmbJKz28_xru!%dfoj
z@WG?q-3M>H@x{sb?5kh<dW!tcTW^1V<!fJi=GKinAK%$NH}&;mt>1Oa&uugiVrg`*
zES|Y~De2@>{doVZUs#nHumx7Vv1_BNPfzMEef7(SAH9FB*E4>6q0Wn9{mJPlgrrIi
z7?_Zdf@iIb3(#LHN24VpzWea*u*~M(3mL$m3!6A4B^BrkMfc<Rq-9~G!8kpCtLwO)
zz4*-+|Ha?EH>uOm#4Ea5cdfE{0*YDIM9*dAfgD0pm5Fda#WW~2tNu`qh2J`Dy{X#3
z)0Fx-p3{05pcLpb6bu#xm#)Rx%Wqt6mvyJe9^4r<Ydd!b;l3;mPukh;)M(u}UMxIP
zCR+QzDajl{1aXTni*c<|O%qyw*D9?zBw<u4Qwow}^n!p&W4GHeAXPbB@0TYh`<-s}
zOWXBtW9eKWN)qpCH=3wkueY|g(J8wJ`-e+ETTU0ZpL?S{FSAJ-LWJV8)GMNLt{E?;
zf}k_|%gtYEoLiltl`%P@bB+@jV-tHq(UwTq7FGz`*w(&@(X&M62p|Z5j6w(oF{v!`
zu4X|(6(TbkQ+9Nlu<<(p03ZNKL_t(hB|%r$Ow!cN{@&iBM-S(dSyTp1Mw$>cCk{Ra
zNX)F1($+AuB0@l`bidzcNsHOpy+=<jT)dEXhDw`6U^~O^`sQRhYTM<!X%?+7i#{u4
z5+Tt><$10$N&t*fRMiQ8$LO63UWo#PRoTW$Ko=oq<OCv!ph;nf3?MQ3_{(f7A^|Xg
zD6O-qj4b2XwA|SG^6!50@-xpq+CQ8vTGXpU0stm!a_3yrG%Ih<Z}eEF)47kF7|XIF
z%zL}LgJH*qB|>sO-ne=5Y&@-95R}=X{>~r%(c5powRFw<?|oc<buNaGY%U^!qf^YJ
zfl);boP;G^*xr2fWVa2GDD=C%<zljTaJ0R1>FMF=;mPT8v3%yjMalCNQjr%1!RAIJ
z-hTf1WfP9(k;>ubxeE{X_GgpH_SX5xemt5KM(3GE4!uq(5)>t(<lLfH4%oSj6e-0B
zE=44WO%nioOr-UHcsynho7*-z0oEExBr{4}ttA<4tW|B(vZ$0$gD>(daf&QHB`u;A
zW>&kVZGFp0S);WmLt-*zjW7U6&?e&)>bgxpOXn~}YmBv)MVi)u!aOfP>(oS+$jnBm
z7#**meh85z_pWO#YnTV8n0rQnB2*N;H^_rBTQrVI0Yg@Q**l-y`Q(#NTJKNJ&d%*z
zB&Gl2FaP}3&F89a@8J0C+SN<8D8K*x?=hvHyz}<gzwwRt-g|F8IZM7~Z-V>Wq??rZ
z=kI<zKC8Oj-e6dnraf2X-CpO>)5FjTMEC0Lul(Y}pSNQF>TkbC48D|ascu0|K^o>}
zBBbinK$^6FEo!AjXI43hYwO(<eeg*jXq|6Yl@l6`>Zm!ABhxLc-RW<hw2K2@hh$We
ztz*}YW=F@THwJ|{zftTT<(kWcBrFQG)~f>(A|*gC2^3wUa_*T#?;jk#_VwFu#e?HT
z!`kHC&XM;gb-m7f-IfKZK@&=)ODO6DeP0~iKis!}`n1=*gY_~dGNz6g*en*U0_%}7
z!<Hfuwvd9W4I0NG%L-dilAtoPF<9$n=JD=st4JhPS=A(`l|8q%Ezxyyqc!gzoq5Ma
z#oGLq5D0*hxD6?|wDcj4r>seLJHxf%=IQZ7;-p@-)?{D!oj03Bw0XX=Hmt4)$D=dv
zoXre>fV6G>OG_vtf`CG0P)J(>0A>(Ts2G(}MjLCKZ>V1Q=zVlCc}__nA)*$=M4%Lq
z7Gr&khJEY6WSPzi=RK2W<lw`6R`2Z|?maokZAN*<O8X=T5CVg9AP{2|S=lB)ks`#x
zW+}zIDokc$zz;vU(;0kmZEda0bz5Uz_Q0Bd`pmW7wIOE25o6$xI3S72j4~>6Vgap9
zh!Ft{sg+tDQy~0o{&G@6T10~w0L7As(EFHD00d!Lu`NZ2Ip$jTd%cHyd#vd7FMs*P
zH{Ud!Zo(56IagqeU3r<Brd|E_ucY7*egoIdcsyyEb~YV1^-?%2W}}^*LA#u#7zTrX
zsw|7=pTG6vpNvjMCq+^G^FRNWmoA*Mnf|-){kz}!o$pwqT~jM9B8f#9lqT{n=Gxx4
zar5-#Oews6?V8c{U~j+M?GnicAAV?ya(jETb*+dTA0Jn}!R5=B!u0t1<qLz7sv@sl
z@-d!`#)0MaH^2DvkME90XWgnl=ysBS_c2BQR7xinYi*}0ORKZYG|q<<$|6@rt>~Vv
z4Nbk21V&quXfsP1V@yDziG17iH#cXqSy2=rgd(%`e1Xi0sMoDxiXs5Qnt)bx;S>l#
zD+<9Enc<X-)=DYoqf$0{muLBMS-0(yQA<j_eosVJb-+boga}ANbeh0wQChiw)`@@!
z#wZCnfuL44%MytWjsz^Jm^>*(L@9WY<h{3Lks#i@@$7Q36k+e$=bpVe7z|>P#e8;t
zXZzm0duxMUTQ_4#Z@%%y{d@Oj%jWyv|K6*wyz=<rqYLNGX;QAPKd;GPHGTDquV1~q
z^PO*fXZu|B{l8U5vnMB`aj!dnc>j}qNox1$gXv;HYOr>GU78RN7r8C3?>sYWTFfSU
z<FmpVZiCSo5TzKKw$WKWI$3P5ub-WsX>Dwdnq1ZCz4G!Gc=vIP%C~I+l##i%w!PRt
z7^!xdq6B>&>}G8($Hz|}^_5<P+C<jB<YU;CXpoo`Y8Hj4lB$$KRNAb6zx(k)VpDWX
zWVKmsc5S(P@@Qt|r?c4}krderOe(mZ)%QMnq<ZUn67HiMTOFEc5v=VVot|y)tZ%HB
zpWfdmR14>=HX?-RZ{B|9=fC)n2#(Kct2JvnnvBMPF+iT1Ix>-}%Bovb<MH^`vsb30
z@oYX)9108G#rWIW(KAbb1l-)&9Bfoaho`gI>`ZuLZH**RN-2PMZDfBsJ1fnAG-)!D
zR!)+3F`u}mwn_^lrz9dkT2M0*2_lNtS}O~Pt#_z^ImHBw6hd%|1u4_EOP5+7y;8LD
znIj1(U>4FqfRVu|iHJlh%ZoO&rYHc!HS^hWe0;bUTw`<wJ~Xj^T}jw6Mo9pP6{!_X
zoFp+4fP_XZ>Pe^D$t-bl)A?*!H_2qtDkVM4Dwkjp{L%*|h%6~dh^Y<E#nw_`vO?R6
zm2e^wBv^TB0w*CPRI3=22rCc;qEh<Ie3YV!Nd+-poL~}-k^n%I)`-&9%NY3L)#sjj
z@#Si5E1<PSWvvZ=u5Bq=n~6|jzdE3-3detwN?|=2&nKg4y{y}|CP>|GNx(KU0a8(9
z_wRqIb(+nmr)Q_zJ3H%}8#>nqhX+}fDMcbFw`N(-5=3Q;F_wXBW@n@E>#u$Bdw=t{
zMP`nV&Sum3=H}+c`fxHi6Tz&=PsXFxyUD>pXY)$p^xX4*&mBF&OP9`F+UPxgaPJo%
z-W|2HarxHmt#k9)Wb5MPfAd%W7Gmh+b`hN-YJKt{ia>BJ2ZOqZ&ZiWMqFgrZXH`Nh
zA;rANOH!_F)(551X1bW=hBzTfs=A%v+zJ$j0KvPId|7k>q^TDn#;Vr`BxV9&gha@h
zQJQeYU{HjAMjHTVeGpLAWNq6{r(<nYWUe|z<mA0?n^pkwJO>10tU_Yq%t}$@-nS5i
z8Df+GNl1Wz2v@l$g-HX%7>!lhP}@dBR2I3-Eo%7B{>eW+K0bQw)mLH+y|VhrkAC#}
z>#rZ|?xM(xw_gmwk4I+*O<gaRSFT*ni_XJ`50CbL_O{lBckkZa-roKXZAJedSJlaP
zAAR^={U`tExi|g@iu_GU@4fRcQS18YLYvO<=wxks^X{X=*?dOMWUQ*1H_Lo-vKN+~
zndd1zowlainM~)1G#r%9)iJojQkAJ#7IouNN^2WE7u+ZJ-rY`c29;ojwouRn6^9qi
z=y(Z#AV8`~f(t&+w1WeHG;4zsRxG}042vNoA_Gb*(V{X$@rfy{xO3yComG%JT2wgm
z#}l8$Oa~QU*g4hatr%z;Ulp<nL%#mLH18W4Mc1*a9CUoki@Metu3X%D`uO3st$sxH
z^!SW4ZV$WBh0~LJSFQ~Kb=|`L;h2Pt>2#}qHOM+%`-}@0rDIrJec{69$<dRc(Y<aK
z0VYjb2a3Nf*bO43NSshnR+go6p=lQTPaoIQ*=%yg(HWy=v+)lf{rf?+^VP3>gECA2
zN~z%6@ySVBFFI98AWoRH2BORsky#M5Q6fS}gjh;7n@^l;06;5)K+Xkd+7Ln%Cy+va
z6a}cYf`oubV2~qgBQa{5D~7VuUHS3#3MW~&qqEt`$*5^OQnOgh8}DL_i8+Rd0tiTg
zM4(U+2oe$zCqYH?#aW&iPQJ{tr-w&3Z$B?4S6X$}w^Z4mF8tYKI@sJsl9ZAUZHO)e
z-?nX4^|LI``dD@<Yb|+)yvpmau*4XbH7)@be`Tgf89@MTr?t1`;QDj7&+S}f0ZnQ!
z=yy8h__U5Oeg=s{Mo@^5q9EdLQa`4B%ahT>xi&?oH0GIY+gdmlMV1$~*XzvZ({8u(
z^5)Bj$HxG$vAOZq+i#QV^txSP7E(x*0LWNDrN|JO!C-wo@2w3751!neP0pN;O<OOP
ze+G$co}G;+N5{umQ4;0Xu0OMXc>25F{CmIn0S^0i@9@c|_wI+7l+E6L=dG7seNAh1
z<NA%=`wyQ!-VLEGj3xw|Wi~59+iJuj)4*PnVNTXoRwqRiLrR?Lx>kw^5{fU3X?<N7
zRksbtWVFe0J+IrisxC><J0|8#E1PLxe~L*+5hBEZilE4?(bR;MpdO=lSydtkGoy$I
zrxa(?@y5p5<;z!p@!rSd(J7Nz<y{fc1;2_BDy@jnXyD|!RXLu_g+*x_B>^FT00=`8
zj#W_<WDG&uECGyGnNkv>OVMjZO}!`zd+q9#M~@z9qD$w_KfZVWJKy@&rw{Khf9KQ1
zd|u}HlLrsG-R{P4{d6*3>-WxYZ9Vw(-b>HF@S`97NdE@c^uOV%DqD5z|Nbxj<_q8c
zZzRLKzkcWAr)MW8r<0hNQPKM!-0$_vk)L8iF{T^m2gxtDH!qFMgd}BLmw+d}?pEb=
zI(2SUp_u^P$#~K(=NlVU)lscqf1;*tx4T|gcE*j9j$UQvH0<5};};+NA3t48#Du`2
z0a%ztg9z6EUMorhq!UJ$m;e!@PGkvel$1P+D8dA~AZrY}7=$$DIGc}*sA40tCLf>7
zJd^RuYuP4*_Qz1~_bm3RZ9}AeT{p8unk^gBczk?VAYVI2kDfl+>~&*cf4>}Exw18@
zYcw#QFRQ#*+pyhE$IqH(F<}tfQ3Xlj_W8@tJbk)1o*WZ}$c?Skbm?Z1mC@Gx$57*+
z<1*TIs*RK5lct>*Ys#VmQ_QC`O0vDR*1fV$U`|g@_MS|hztLAn(FX*h6-}no)5F6;
zTZ58(OG@#oOHe>(cD3bPf4vJ<8;Bsd5LEpC@b)G#mS)$T-#O<_-}K&#_hOFBjGQYo
zD{HPIS!9Vtiln3#*`};UyJ0{Z!`*NLUfDn|v@y^dFT6LrvRVr_ECaOFY8h&^tcj$k
zVpWk<EY_Gc<eV9E#GAhHPREOvB@4Y^$v}Z8fY?OfdoOU~oqO*&fB*CQ10sS{S_nl%
zQs@x9Ak<n(5|Ri5xX`k<GEI+4>p5D-Nm`^?fuFJjthLtKqod=KljCwWYr3kou5Rnj
z+8Ck_0gwn_vBzjB7YA)XSj;mr27wf<&9dzBwd+D^p%ZU0?XUEPTOJHhqLcw9=Q?O8
z$%?jXOqZ6kc_K|ee@#hBAxIgGZ6G9IjJ&`yv-h4T0%+!lNQ5d9q6jW<=e&!Ikr4rb
zgpfKhk>k;9zIx`d3s1gK^q0FXpyD)3i=r4G9d%u|ur!Il3%7CRxUf!%pIN$@j;BrC
zcxQ<K*e9l#HdBBgq)Ah=ys}(XleTM7h;!%8*G=`4*WXxKe_87+m1PSkkt8EYlS!2j
zOOi?msYUqY-FH@22AdnJz&x4E1>w9b`~BYf=GM-^;ge54m!?^=w07x<FWz~ubNR~U
zp%HsG-+SvPuc<U2s4Q5xa^*6z(^|2&8*4)f-QLcARd)ywLL@0%aEVb}S;|BQZ-Ke%
zS|gQEiUEyIe>&Y-TN6XCNHZgvrXmtX$u>l-^uj%o8NBm@UT;2|B1D8BSy=ClNgL~Z
zWGSU26eOy)Y`UPeF49a%YU--Uv&aG16N2~k3(r5XzP|E>FMj2xKYPEdeA-ip1PDOj
zeGr5(MiGEx6hcW!^Z85?fJ7I%0Kq5$F+}IA??Uixf7cQ?lS1@?Bk-f$A_|11K@q&)
z-aeBUwK^K^?>_A6YN^PTz_pdp!S2pr&>x?i5a8-)^|c?r7Kvxm$^7JGsn^5cIJot-
z&&^5@h1k5bF&uXP*Z2OX)#1|p`+sGW+uq)|^Wa8wc>46_?#^uM?1qwdeAtQz_h<V@
z7cPE$f3)W&ZPkS%j-@0el~7=H6ha0@r2xT<=k@An>CD-sJj3y9>O<68<hibWun;nC
z4q<%tYft{^?I~T@yol!Z+xx;1cnF9gID*&#`bY+(h0tKo7==KNpn!bfzz|t@1{tA~
zvf+p#8bh`$yQVd1Zi5qu*{n7`n56<cdl)+(f7}WCCMxI3%;@>tj>lENq*wGfI<~$@
z(tc*X@YMN9gAaC3`h!8*!(=#Yn^~%GFi6%K?K<2YJE6&YuRNl(t3|rf&kmBPyw4Xr
z4@F&Q2U#vvQ1X|?y1}LXy4JdLX^dWyOm$LD<|4+-WNT|H)x6t%c+$3(NaSf|+sF*T
zfBAcN?@mvSm$Hmq-Bu+^C4>-COKk)RCIJAEgL4iDNy;=UfUxUY?|dS(R>qi2rNTpx
zNG$l?z=$Dw@aQeJ%~+n;#(IvuUT?I#n&-JvYQbL-LI8N@+oqe2X9qUKMc3Ca0wN(P
zsf<z`lIOr2i3kCmw*XNK={jGRvwBiSe-{-RWs*$yE?<A<laD?`<UxPg69)hQF3V}Z
z-!E)0Sm$Gi0C~TswSl$~LI_090T>py4I)B{41tK4IRr)|*ZOK+P3I*MBZ8Cy0klyl
zWmC0F%Ns9z<>jrjj~T5TqjSM1xpivuaBr_|+XW$q9;vh+kwHTUpG~hkIykbOe~sQt
zib8;M4M0><93AgpyLw@HWte61jW^#IE-%lkdUa*>)W+%K@o_n8&z-v<gz`)}$q`wS
z79bg!BOH$p|KR`lx8MH8%coAQ9~~WCz53W}Rvn#;8F4(Bnyj~V`powBw&|@*y*P9B
zB7-2E-g$T!P<I}M*7al@gu8z2fBE6k(qDc5)kKR?zwmAA{ZSOe4xIBL#Qq@9C=^DO
z<s2zPq~P46%b5`-NrJ1Tq(LvK%aS<)imGWw!;uoYsT=E^b)5E#s;&ue)F1ScBn0Q3
zXYf)g=UVSQDakQ%hzo|aB=pYZSw;X-P=Ao7z4SMJ{kz^(0Nfv>-~GqmfBgO5|DzZ&
zl3)M{#NZK05=o&DV%;>%*vqrlx~jFdb5aQbKoNp-)^#Az$ACsV+bAJ`0g;^bgMlG%
zSt66n+}n9DEPHzocM@Yxt*=Wd-+J@SXTI>v?!yO0>zMch?}PU*e(_5(Q9pU@$1gnl
z+<Y>vrt?!<n?;d-UWs8|f6ey}9<KCobu~Qw;?=!B`48WC>4oqA^-l)<-rU)?se#4*
zaZ{KqjD)a!Ze{z8dVD`mCr4A^{9O?`ttF9Ce0F<l_pq`dNQ&dh)MPRcH;q?_7(-Ub
zwP7yj0Vo$r9MqGouP@`0sczOLb<kLnkUEe^Ek+j!5d%hMX5c1zf02pK+cZ^a+B@E#
z^f2`dF<`_Pyqih=erG2oUK{93Tcfx4=KCUT%Fs;x3S;JCHYu-0Lru*{Np9H1(I7X1
zr*rlJP)03W-j`?AmZr!1x9%K2{lvvV+D!{~(d`|T(#T;^oL*g<j7x7jSMo-0P`9qr
zK+1RyD@DI;ORaHjf0&&s2D7$0a<kKCSLV~olm2CSg9AsNkL#tC;efKT9ETX32OlCL
zlvR2E;e8AkU38rzl7TUzK*s6p<nUmZq9?X7IFt~*lSG100%7FFbr2atj4_}wQb=Po
zLR3<mOeSs9YHjkg@XQ`iN@<i{V0#6M$UfK_b=DisCgTc#9yd>&9WAc}tr?)Ro$cD-
zU7n|maXc9#L146<>$)zNQ8@t^e_h>mDBGsBoh2X;D1_cRvwiXUlU?-vLCy$m*Lv@r
zb=7>Hr~Nc9Fh&SI%ZnIeaN$z}B>ogW3;+;g4ABEX1hBy&abzEyw@t0JWMp7eT6m5}
z#}kzlFMZ`#pMLI3S{oNbV2IWlY5KjTyyzVt9}6k9Qr`Q;Rfc(?;QR;Ff9SpSwzEXU
z?6N$o>p2mI=+B=!+w0{&dHuCCH5;272Zu+amDRRxH&!<u+}}eKmoHxnEFuGvltM5Q
zgYY4^u9{3Hhljhser;uCV}0%1`Aa|e!H<kFMbVp=Wj0J1P%5L7T(lr`#u!#NwqE@D
zH;?uo?mW0NSn98>uNM8Kf5W{UTTQbpY3DPeWvayL^61usy?i)o+b&6y5Mz?cK|ddl
z$7$AYoj1n#7<Hlmz*?)dCU8wtHEk=jYC8uEM6$Ea2WF<CmlBDt?bx;Xun!QGq@}^A
zvCcZrk&-0C7|XILtA?}|2n^8}B>)h)t(yy1FMjP?FHOdK&dr~DfA;x%_wQ_NtzW)!
z>c-7O#78%(-~vQeBv>$n;BBzoOl64>iinW{$0#F7L_$Ucp_Oghj)Z|AqzobY7$P$g
zDj`y%(==fY0MP3d2vFB`VoYlEJ3oKBZ98P1&u8EI#y39r;KTRcd++r2sa{^(zjHTQ
ze{g(sfA`^NZS||4e^Y880p}iDjw=52t-t=@<JVT!d;jJS{v^?TN=T7vN>p0dX0Dco
z1D*N0JpaVGZ-4b)zVq{+R@3I(`P2XQ&;G)_d$-}*RxH>zodsmsE0O~9gonNK-r;eP
z>QwrEk6mS#W^$`>w&5(vZ~ou+cFf^_`(MBOhyVL4O&iyje+FgSNQ54OSF(%jF|v12
zA_1G!_XqiS+O^VlG9(PhR$v6CF3CQCX1taK4$kwWjC)gGnJN(|Y@4t-7xWpXtl{A0
zv`6<QC9kGhp-@s&n`OcVn<zEx4<1YoVu%}~)!TcMEafCgCsn($GCDjwtVn$_Ym6#R
zCgmu}%JEUQf3(aF`aNxOnJWbHmo_&}j`!~z?XGVwtEofZRrC4k=7yI4APL>)v6`Y^
z4B7g$H|)s8SRGG~!8xII0I<&Hx)|hUZpujH7=%<!T^}6mO~*%4#OOP~5Cu5q$U$of
zg4)(~(aQ)b#u%)WLP;65N)zLK*LKz;X_F;M#(^!6e-MxCI1o{y48aBx)>YYHwKu8^
z(em0R852P*KzQDT7`RvT*Veb*eeXR*D1=57QfdGnLd3`jOahn#!-74_%<O}YjFN;v
zB2q#nU^I{bLLUMPQYzgzb$(~>h`?(t^DIX8RW*;^Ip@5y!FA5cBuSOdVl`)u-bW${
zi#9Are?(sR<a!LqC_pNSERZn@0)!w?kPs|HV_gvX#V>#D$>&}i4p*J;JP-f~hR%AE
z<VDes0G+iAyqHnwx^6M}86$tD2N}iKDG`yqcTSQ*kO;~*esgv0_PzV-=g!vAFD);t
zB%O8L=7n=Vzj0%<xzYGwGVQqo1Oy|3fJmMrf1sD-Ge!L@v$m`2X<&bHGFB!{vwUM?
z{rTsfogN)R<gB;c+GtdwfJ5+(akOzJTV4}Mnx<)P%)!p?&b@o1UjOln+w;j7h*#E^
zx6hu+6Y{g8l*>YPIy#ABI;oiv6tto4f=42VR-i2s%(TD|>ar#!F)~6lN_1`0Ip+X8
zf5*0|7k0xYO^_%8ch=U`9HZ~*xfUK{M<Pf~Ce=a;XT7VPtvaO$(B(_wrC)h^eDYv2
zKX!uNzH#IHh4T+?-Z(u>6KDlHhG?0i)SieXAu$W7lrUQQwvh+~Oi0lMM2bQ=Wmp(%
zqmT-uka<d4NQFkqqDUVZvS~G+9FK<mfA`;h>*=SSI5^zjJvkhm+N2~AD$^!E8c&?(
zI}dgu(c$>y-S<E2kJjG%<o49sS=)W#D=&qAz!Bgd7)a!iQS$cr)dvsny#3~_hYzZU
zd(EuUhttNfCjq$;7&=Y1mrAAe$@t{w@4a#JgSS?TVky^cSH*^7#L|a|f`D|4f9pzM
zhx6n5^l*8-XHKuL=owWHCTDAmyR{31Ev(8BmFC6|?k<g&&0b$m!jO`TFeRBV^-%T*
zwV`R4Y@A&=e|Dp)LhGZ5>ePDQ#ZHkhilhXm@K`~k!AgF2-n>>J^hfKf`5-qV50^<?
zVbLo&A~l5^>vWFfkh&ml-QR1vf9A}_^6Aylu$Rrtd73S?u6ww@-*}jNc8MIESsBkJ
zYiq0Q8wg>1a<DW^N2|r@i>Io#Q4$m7R|dV``0d~Nt>5~c)&5W-T{^dA8@IW&dFss8
z#q+0BY!<SNUq&#mj#l!d_~4xz_ipWCFc;50cJ}ne_4P9*?KiDEIho9wf6@gvFQ+!N
zAY)sXM~8dtx>PG<E~|Q0Rt`Yrd62>j;Rp~QgizNt68Y$aWMhohTB(F1^@>rJ_lcA?
zX=1X(WJWIzGr86}(TOohVtQMr&n~ZR<b%=Da9Qe{NQS^6EDq43R}9W>pEWx5J}&$`
z7C%r*8Iwq%nbF!10te5*f3ve823XvcQ7WO;>gK7(u07EkF8cs6Ab1pH@@y$d3XU?%
zdU;;tS#M=!z29GgN48n6swyIpq<LV`IYyxoq!3C7wRmGDO-RZZ*;?!Qk!9fG&)3>$
zW0E9u0lFx@{_Ed)>FeLxICVBgieW)LgSKtkw)Nf{V;&9Of2t<3f7Z5bTbAW?I*IW!
z{zJ&Q);k+x&{{j|2&rirW0I^WUi|7;K+@{!YM$jyQ?INpAC3<%UVBWZ*|-0*f9yGi
z=pNxlBOwEFWC)7H@@RB$bka24bI(5g-S7OPx}IyTHa9mPzxMd6ul{&xxWtH)>0C;!
zh$Jx)q7VTo&wJ-De_U)k+qK=phYwdrD?j?--^LI}%fqUw4~~u}lkpReU)|nZ>!&I=
zq$GOhtqVSa?K-7}_ZE>;BU6KkgwzPz)VAxikj`50E;A<2k|Z(H`OG>8z)f4n7>zN^
z-nNbRKF@QbRi2t6ON^0%fH|l{#Ta5_tu+yKUDb5>Km6<8f6aPINlr5J)vvw0wzcu<
z>p$K;yEW=(XHTtQs6$uB7=4U^(FI?3HUI;mwH9J%no0`ky%!`SLl<K}652?T2*NoR
z7Y^f6YptY^M1rIkuB-r2o@Fn8{cAsZ<!`S)_4K97k2&D0Pds^ae6qf=H5#oxcJ<oc
z!TyV1`SSAWf671mFaFDS-hS^J-~3iu^piYC-iG>*001BWNkl<ZMD+`Og#UkqkN#*f
z-aD$^diSn0Mc|kfy|RhBN0kS)vcBFnHt!YDd!<5>z-)ffx|$d-U%L!a&g*JvX_)7Q
zR4RsmM8ezCt^M7npMAP+e&@pF=Y#tvqm6%c;kkcyfA;h<DV}eI?gUt%Ah^}5d}jV`
zSG1HiW#3^QeL}%d)Wiw-p2VI|tI1NmQ}5oGc1``v<)J?s%dtNtbj~gp)Hm8oaX_x2
zh!}?}S-KRZbk)fij>gYZa&aC-<4B<8@LpZMI-NL^xZ3Ba&z?C=LIvyD)@#Fjkc(z^
z`1s}Re@hp)QKL+BqD*YtY$@w}-5(5mXY)MWT<y)u>4TF!NL!P{CDVWW{JG!w?mz$1
zZ~o5p$Dcp9apq`$Mq13L<-y*8Ns}O@b)TD2%r6ch#Mb6|k!M?*TW7b=E%lZ|OM~9}
zsjUl1Hrzj&Jlx&ytVIc}s~cDQ=;q~QdVDA_e+pzo%(A=~jMA(pq!OfrkVtBgcK@Ct
z&@r==N=eP6^WLD>AEjB}B*kJPT}q{tmQpQxTu55!Os9F04`h;}N_+(87M`gLKw6tT
z@2#w?ZES23Nnl(|5-q+zdK3y^TmXTIl$1sYB}fSr!7~YY<@)ulbK8-?#YeDsL{w4}
ze@d-2f@BuVAcanoTnH(Ih>s+b0EjMn6eyJhLI4sZ2?YWzd;=o_Y8hgzn`WW1Op-(?
zg+jI=v;m*};;%gYg_ruv>o!2^9W0uzF@6d@`_zRB7yuUV<f^J>v)Oz;pH9c0IZ`J%
z_{CFty{v00j$TSBjT|ly{_qdK_xzWhe{Ck?zE)?qw!F2Nng<VdivH;9U;S0*1+YX!
zfyADHm;r#9r1;A}`@Rpbyt4ku`ybc<%l(2QC?#8K1IiCSxqbcFFJ*b5m9`N931I=d
z^HQn9!^4k0yz%6d&u(pOuB@yWr4IJ@(=@q%|K4b{nx@$&pL{Z08WIu#6nSpzf9B+P
zQVjZuHeFq1N$PFwEhi>%w(067M!&YU5fL!*^3o_UAc@+w$S~;l@-%I#Qc5EU7j2C!
zP1?EvLPVU;=RydT7}s@8-2!Rn?P%CH3YJFM7ry+=@=E`~4*KZ6_`-7!?%lb3=`z^x
z?SJx5|L1@GKlVp!yZ`P#CPh-(f8bDr$gNuxguRqRQb>|MvesG(0gNF8KtN_|U9erI
zB?HuT9Yav0G;oa3xvuC9j!!1b>l-KYGKPRkpG>DWKK{u0_?y4^?cs2E<HH;Kd;8Bm
z^UUefr$7GqV+Q{5tFIiN9PRAv?Cw5v&VBgd$5*dD_IafS@2Rr`3MSY<e`p1OG^!^|
zRW()X@7>>(%$1D*fYt)@d^Srn(T4W)=}o1uD(5Dn@uXT`Sz&ilwaxic8_!#eKYa6d
zH?I{(k;G<lUnd6>sqFKcmv-(Rg9DvIgwP4M1n4<A3bruA$Y{}a92i7G92f-)9b%3U
zkO5@qirtTz5e<3}1(Ffkf1$6D@3qrrVAi+RXNSk#w5F0wRI8Mn(sZ_s$_4;x6qac*
ztI;w9j1FLT_u$<1Q>QMh)|2USuXl937fB7%-p<2`W0+R$nbV{3?!z<Nt6qgco{kSH
zlVuS3)}06ae(&m)3wLkb6DS&I3po%m0hK@f{I7lTv!6u=3cvzZf3`c`-P6j@=eIHc
z;&85e_q{iTp!M|?lNcq<?(Uu-LzE=RV0CR{c67qPj4-cebyxTEeq>7!ylq&iEX_PK
zM(<sK$iaq)Kny-cYi*IHQpy-Z2o6Aa?_)%z^>Ae^EqYbc3L&MWPmTAG7Nb%Cz+^jX
zeK1)eNz~S>)(7hsfBOd!Ip-#mnbzsX#@6B9zSdf&$wHG;mZfuX0Rxv(S!)Slk!OuD
zMkk5U^XWt-`NgYO`lIF6`cw-d;e7yzK02j{P%gYb8RWvgchT?V2sI-IkBWti0zqmD
zKo}OjD~O0dh(z$Ib?oAtu9R{@FvJMd*syWt{9{i%VS4>8e^789kQoB8{B%*A(Nb%j
zBnbd8bKAD_`TXSMWI<`1%}>1ljH3W~?-3b+iO_p11jgulgWlbpyWjckZ=c#)uglrR
zb7zP{k*I1mAB~1us1Ty>ytCF@>s-T+n!W*;4-XIi<WK(}8|CY7{QS(B(?ymJmwJPK
zhJqjN?cIK`fA`_7yMOt^zinFQz2ilBDYB2z`EFx#?cUuxy*!sv8l%@%R+a|6yw{se
zD{K9|d-wKs_l8T$hlj^a(=>I%%s>Q!j4>euicIVUvDOIeLh$o)W{hDD)01&8FOV5R
zbk;R>Q)C%L295x%wB!)Hb>2B^tufkL=bZD_NhuWxe*h4K-nOIVL4*!<{I&0Vb8U0f
zxGKxjYuBzJVxFgK8*9%!|I{mg{-=@LwX5egmvhNoDxpY?0Q3==RDe%cI(U;L5M!#0
z?<_<Xf))x(01!e*q5xowiIJsFgp@!OVsze5$}(FT+`50S>bm2}^kh7J;^}Ap=#T%n
zZQGY$e}4I?r=R-d=FOWoZ!V37MX$HB`{2bFzt}XDbFS;U_4RfBY}(yFG`g1W-D<|u
zuC{DrGO6L{q-v^)*$Sk(wkk`HrjxGYM5eKW7oYv|#>(2>gChYlT6=tO4AJFUI9nNo
zk})*(`L)&5t7OvT7J5EYDnCiOQ(w4r_`!~9f3aT-(GxpiD1%^h>?dRp5rGjvfYhjo
zQK6s!sh|{m2CiU#T8dFfmjxGuhJ-=PfP_HTOBKoK&H>n_06azM73*Fgzrsz^)=tJc
zk#BV65AF1!jDQgkqR@?{{k@~wmTPBL_hv`SXSQ^a^_B`7#biDoEcJv$kLt|!<NNoj
zf7v*ev(`8M%=*gd^_A%3tvkCus8pFm%gDm}7QBCO^JWA#8bQj|M;H0jsm*3S$IsGK
zd|rV>sBl`(KD_zyjgLOsJKUF2F~_!TmX`<Tx6c;6+=7!Sd9Isg60uA3#3WjPP^3sn
zQo=hITxXjqxYo55y9U{-L_jJ*3Mm9&e`E?>v^BR>sKLsB*)bBUN5&EhYbB$590h_R
zmdG;Ent?r0-8I2iA#^cX=HR{c-iF|$lIJg6I<<Y?1C({;f{(yJ$mo5vfdeC`zo+$N
zX6J3^Y~#ZE=Ju6~S0swAZa4&tfnz`qB7k7T#pi_tsT5TPG9jHXNC3e|jACVde=X?`
zM4B*$$ZbT60!T_B4S|Lz*)JLd97BvTf<Q0DN!gy1-C%Y7+_h`Hl@(+eyk`gt!zt%n
z2;q^gCITu$!nCbxTh}$dE}O3DnyQ|hjQ5TX!)KLlBtUfD2?XmprDWZfi7{s{oUQ%I
z=GNxH&ciHC-hAVyYwH_z<b%UQe~zSdKe{-ZS83AE@&q6vM<j?4+P2);*(v({+n?O}
z^Y8!Y%g;ZKQs)KYc&fF!eg9!&IqNTX9wLAcGB{6C0z%-JXPJ@m@}-NnKDrTud+(jM
z3nNdCkH7Vu-yWY#pLyn)8y|ggbUYp|E#-M$&nJu4eBCyTV7)Us)lw2df05?^v1z)d
zX?%<(&l>9%Ix9q>wV92l2mrkJijpWZrm5%5F-eRPSeJ7kv|S^l)KY}teAimrX{nle
zcHyz@fAKH>t50sdfA-SFd-w0=SzeWOe`)FFo!hUy`ig72)#a7V?Tw$l`Zt4~c7cz|
zu-MY1kRb%XAQ`=n-nl%df9Szz(}>o&4h4DVB%)9<dK(CJqL={)T?oLWb)vKx&t_%Y
zz5Bzz**<+HPtrSg?!EBB7Z3K25|dR`_4eCuZ*6TYkCub?w{P9@AzZqA@!`W=0C2${
z9v@x2^w`OG`gx^>;A1%t2;D??iBf}pPGGtw^pdPDtyT#T=1IZce^ymnfcnyNFZO%6
zotLLJwoBKJ=hlpR>nr__Zr&e`RwE+{wTt06kJSo9IdcUaC#A^osW+PbIM0kCjHng3
zCfGoVJUfA&y$u)uk(iNLfFlx!L`*~~GE-D028j>>BBP<m5D=l_aNt@NI-zt)(fa=U
zs&ZtQf=}1h_Qwy>f2=n%Y7e_P$}R>c8IeTb(2y4S+H#K&25TF0*UA*rQM!t4G4G?7
z2lu8zj{d`+y_OPAWiT>}&US6>$|`tW1@gXAZ6h81{DZf?HQwu{w_p9iAMelhf@BYI
zsb8c@)Z^oG+gtLN){G@AmJ*OoCpex=j!yhif2nm5D_B{(e`Lh+!v}^?dZ!RzHk$%M
zWM9`6A_oXnHTNMn+j$!VQWPppvKRu!M__Ovgpdmq8g(2!z+g0#M%Hy5P_jpeF@%UH
z7$Ola^d!unlnUGdz$1VRa0mfWGIQ50I&NT$+1S{4{QC9C;o;8x*=)8r)c+@Cjm6nI
z#@M#)qBb&G3SNEui6^wyZPVtL#ZLhre|h<#HiSO3WsFf0N=ZV<7;uqh77;^Ooa0C(
zf$OsKr_P>x@mIdNcIGU}tZkb-mmzpWB$A6axhNvV7@5IZH=E5lhNiA(vsqo&lgW5q
zPT7BU3x`5<08E6=b=Gx=kftV0(`!#XIj^dpzx#Go*Ct8t@9pj#9V{=8UVi!If4!Yu
zrHpHBS<c(0;^-n;2HQ3B<D)%}{#U;GRmTuj^8HueRM}wX;BY*f>NHKW;>o9<f98c3
z@9*sW`Ct4M0!T>^JpcfsAR(0$gu_Aq=;-LVXP&96>eA)QckVy9`{2RN+xM*XQfnqU
zI6m(6`bwz~BO|0)rnM$SWClbKf26dT*G<`WUEt%%{CGMST9Fhag>`N^n?{Z(C`R8k
zEdfODgLN@Rj1eLmV+3II0hmJwby+n{BN11Z`%gY~`B#7KrP|FL*q^`qwlc~4AAC3-
z&&sMB4pxr#Pv-OL<C{14jt@_r-nxGMF$CXFO)4lM@LkK%MF#6!h*1a;e|=c6Ka>z!
zDM`X;B?JLS03e9Yd*?z`wSZ`ngckrnl_1H{aK(52{dYeovfit&zFy7Be%@POUq5~3
z%+cZD0>q!C*~aE32s)V_e{%a{5On73d60BGoqS%YAq0S!RP!K6bX_~2Hg#>wxt-6N
zU^|86TF;&dfK<}!4coc_f2Dhw$&8#;lVW*Tw)J#c_EV#UUs@J5RnxP>AFk{FO&{F&
z&i0>Q^dFo{(~D=^ah)ln(PT*yLK_JqA^-?UB!!U*0A&;^l14~SNk%~!ISNF{Oca@d
z6r2&JQlJ1pZ-hv}oFl!3;nLhbzh8ZamaeoRw_bR8@8jKWdGIr?e}CMJPb7Q6B!xf}
zh><-qb#r&Jw=?J!5APq^hIj7n-o1No)Gv$_bv?^cQ%?6KMiNv`>y&aW^wIIW?6~ck
zwwjKHJs0BCHlW}y|HJ?Jmw*1f`v-TX-F(o`e&vZP1&2`2UwZDj`?v49&$^g@UeUD8
ztFQmy&b^Q7ZlX*Cf7wORbaqnLC$s7CWHu36BzY>09M8r_CkH}c*EaL{G(;DITUZrQ
z3<^*SnQD_JNn{2_CAHCsl3EIr7lqI!GDc=ndN!Yrrzi9Ics4nn%_mJeZ`-=->drQc
z8V&%&7$HPr5J&`=nba5q0E<MwsHZRd;Ib^++S<ypY=PBuf6o0U9GJxe7rRcHrh~y?
zb#--RWrc_q*i38f!mQLeTh}w^I{@&(d+$T^%-nTth%QDSqeqNVQ?EZ*-P~H-Aar7k
z(MTAANBaL>dXcgLKnMvStry$puf6#4uRrs}mqu%+KxhD{>(aSy@nkfebv)hw|NiyV
zTTeH`5fjtRG-FJ+>F$n~iD8bKBPTYTnZsea+n73XI%cMa$s>l}>+`$azW<(oUbpkd
zv##rL-7m-%R(h9@m)-Vt#9Tw5(M0(SZ~{)zgoFqQmSYq=A=q%9Oy_(?`DS{0Njyw%
zKs96j1$Y2*ldz2@k0WPl8<jsuzg=ilxL-Cw6-dbxXR8>n2U^9_zThorM$H?<P5IgI
z?n&Ypuqa60OeDB#de<C>ikf@9&UUz4b9pj+>)mm`RIjJtzcuoBbVc;VfyeX&PzG~%
zcjvy93ACX3_R-`w>Cu|MQ-NdaOYY4*5693}{?QU1MBRwN<2QLiFP=dj3qvY+cO3A0
zPeP<?q}9vlX~IJ@oCsfgcu0K6;)Kt(9aV7=uE9g;V}hF~wdxcnGVB9-u|fOQ-2!9M
zKovS)+;ToW-zDqm%-T78etWzMSki+jrhhHe<~#(J)zwu*0!a6#mR+XD%qeN>It_c<
zPiF|&GvYDW2xDl|mD2yNxg*It$okX|<kG)!Y9>BuuRf6OeFr04o7j2$xPo<~gLI0V
z>=BuHHL(eEd&u>UcUR!iH-F&;veoUIHuv@opIpa9@KT?enyOnU6M5SUfHF~7P(#PG
zXwaIdSvR_M3-9g8NsERi@|>f6Eb%juHeYFfEG4^?Op$k?IY-QJNJ-gT-rHTY*Q%uz
zedzJ1+giM?%O>61$p6MHXe9DcQ5`@xU6Qjz0^eu(u>t$}YD9O4^QxAT!%LBN5?+jN
zW3)J6dsAt^V@t(e1B|bLiV~T<kAVTShzzR8l8A}L#zvS3T~X$VwrMya(8^xmS7lp5
zX=Y`80vil2ex!}wD9(to;(hKhc&TFE^O8W3h`cc3Vo1S^38Y%j7j=ehmLHduGq(df
zYh`ZL);3)&@8{0Eyn{WUbmz&#E!8sekwKn4zKtZCC!(Y8nl1tFsDLw{<J~;&r#V8d
zN;=>={7_RU?SJVad$)JX@;yqAd)wclX2Kk-EdJx+Wgoit3(l#TmNumIy|3)g%6;z}
zk!enIjnD{L1=*DYk5@m=LZyb|6P3S`3gg)7#*r&iJU26%YoB${)-H`t;$l~$()p4}
zl~>Q+JjIc)5U2<w_89O$F0hVr4}Kl{bsq9#Oui<}pu<qcx7wk$8d3Mtf#faH&Kz7+
zH4;fo;{M${E{fGicKuw!Gc9^|+uhN^BU^Xq6tu?7%&b$s35RzD3yZ{Zvr-YQr^j0g
z;waC8_R3ho7r222#TyQxp1$e8lYQv-u{G7#`r=-mXJf#bVCTPuOub}Lq8I<Ay>!t~
zK|P}Qxzguymm#cXhWb2n;vw$!;q%kVCeMro+Kb@ZAwd#$PPVeSch4=aoE*)}gD?1M
zI<TPJo36&^Ywaf+74^>E5vQ8JCy&uhP2yPws36VP`1%k!<pkQq*U+f1h5=T5jZRdz
zGw&Szj{4vLcTrfAgyr>8bJX2wmiwqX1G&&}-NOYs%DdJAwe4smCOp_NU%%GuNi-H(
z1PUt=erK6`9!~Gi$MT+mL3|aQM0pj*hWGENS0hN$-kFW5lw~$}t@aK#97ed`HMp^t
zn3DOF;R~A|BmGe_JN)e0Y_I>CyRyakcU5V*x9<Wp2TUPTeI{CTg=L5;r<|w7-Tt>h
ziN5y!2Uj8Q9lvc%U9GJqutZ&H>Yszo>fp3bgQ%r^<WUFY)Vc`F_E{6|?j$MdLWPqu
zv;_W%!eKpCaW4(PDpb70H)I;=2JaG;Z6RW;W(ORf>Yr*FK)=e;Q>(J!1OFrxZgw7i
zbpv1%nWzZa6nL^WH8cc7L`2+Pg4)u;!NvBA$m@|T3|}F}wYJD`Pyfx0qx+pK+H{^0
ze!SM{k+R_qWFhIyu<SepSHsAdv(;bzMc9kYEUi;n1_4sIaLgC}QA6%skhlN!Wnuy;
z>3di*xd8fSanvT!dTT?F<BLmXM(c>5K;2UZa!S>VmBoW)a_l5^H}?bC5+P11EI;Bu
za;AeA_?Xdg?C;qm`+5e%!SMil8*L(7w4?`K*OAM(G$lrvit630;^$-75EC9JMaB~?
zE-;g2#{N{2$Y>G&4c~_YX^GD{;~p0aS=#1!6#+k{#>2tsQ1k65d<V>+a&DPSfRZRg
zD34k1^B2{r1D4dCA;(S_x5rh2cZWQw5+QR+-Mxz_gijHb^T=Y^qPwHj)Zr~EVQJ~Q
zP=-Z*rPaAXR!S;wd^Gm%xe6^Yo~k~E&lKNM6J%5hgIkjw2Ma9tdo;{XM&4u2m-x1S
zV}F0&Rw$&51cwCl$7#RkiVB<poZ945*G&~WMsk(=W@e}nH{<Mj1VR}<nd8tyggpAL
zBXKRKK6FNu2!N~8I<Qgu{EYYW6{7teS`<OAaURSx6;ComTiQ5f&R6lCbAy&<tdiEA
zxL2A%`OATy2$Np<Oy-IHGN_55c4n&GV}jXQ$Qb?q_6AvP+M8M}FDw8{?2vc^142kV
zS+YP9%}R`(f*@^lya7XblZ2k9%<rYSA}ia2&XlQV31kuYD5>SZ!p5MCL`2yA&y72o
zU@s=7<3`WT4qOa2__WA`{|&#19fjL(OZ@6BG9ey^>pw*EN2}pGGTkT0;*L|Tj4gIy
z>rEM!PV}M8gOgk0q7~2?5h$O>*Ju4XrDsz5T}C0%oa#hP=jlxM^#)(>+qv7kScMz(
zL)Y$qD9b3P!_%DiQ=u#_9o;o#>iti$Dkat;+fSIYH0v>At2}TDN$2Mpe)o5V7oO!X
z18#wVa(KZl&FxLHme%x~=0mjYWXrT^dV@bLu@sYy<?V{-9jbv5^CYU5M26a`Z{{?2
zCZ*Q#lz6rM1+uaY$oMdOo0_~>W71}E5&Whsh0Vg!n8x(`>1e+B{qvPg6SLvt$m8`a
ze;KKYY8!xyfu9MgHxi`#wpt-F(tI5IFPzT=(Uh!h*45F_SXaP-5zLDHd_G*I3hXZi
zD*$^-6_N4NSU~Qg40)srUr-^Jp@A|6;X>0~+d}+LY#cS>7CFX@dS#*L@Y7m<g{#uz
z)w`cN)%hJ4gL-<NV5%bn>rsDKn%CLb_CZaTvF`TdHz$M7T2HjRy}beJiZ}SlpF}!;
z|8z-Z&m0kY*E89|lU4EI^6$?tphi$O%^m`rZ6gai697$w)!+KPhzL<~&V(c`87r&p
zrm8{p$0fl1H^G0sXzx2gAksOkXS7CP!OaNa08$fr;|pa-Iil@{$lR7=O5VoI@P^p#
zaD3(dtC*^yXiEd8ap+3ia0(KS1{xl1hnywHPqaaXdEhC1VP8}_+?O^}e%tXLVX<UL
zV<s!}090`<af*0KEiYTYeHX41LhL$c9o%S_%CH=-bt*iis1+gWFZg=a6_~j>?9yP}
zXeaCazC@2_QJcjzSiDOi?+?>NMx&5EaoOI^nHhr&y`FB7WVe;#LEC0nhy8&sw7YEj
zzh^wNc$7XR`JhYrm2Ii|X*sHiHrlG-1@{XhYQQX?lQ+Oq|2#n7lnt{NAJssnjKkK$
zlSf;~rOanYkm<7A<bJRKvQ-`}3om~^JlQyk4R20MMSu7VmeDMWdK&$xZ_5)qtg3Ug
z-{E=wySny2$S0kBj)}_8o=E>*g}T2!$#h?EU2R2}gI`p_4V3BOM)@N8KblzMv}Q#2
zo&a3jvy8(sT<_l-W^l#i#eCb1sb{<TreXk<S6Oa&=FSJ{!DqvYP5#{**yOd3K#o>d
zwHu#5d&U@=XFf+F#!g7Uno*_<r^6)xS230=SEh<uG~#f?)oh&iZ!{(wuk9g|FRYiy
zl-L4yg0ZlEE5NUzyFL6^H_~q<;MF7Kb6_X2FujT@g1W~5&-Oi{_Qg*Zr~%%QTnBn}
z{r)H=%`56w;$aE|AKh^)=u2lPKWy1VJPdQ!fCY6QXI{HDtTm)KQya+82;<;gKAjCj
zQu^Qjlge~~Q{kWVK9bn(zWh(;VB@BuE^saUNxaIJ(d*%jspE&W?(5_GTcEL_0U#E}
zOD)|vof>YDpuGZ<_SH{j{2F8ta%ICy>MT))IpZNB$YtYnWr0q#LE-?dt+j{C(~x&q
z<ez6M$fjwT3bx1$^Kp0Rlpwg476)wz-8#_K(!fYpTx>#4s?is2Q<OZRx>M4lHwuiJ
zTk6+bwF}<Z|MrOVDDrwMCm_E7041J{qavd022s`Qvc1d2`rxBA5V!pA`h&tNtt7+9
zjZvGMpF_tQ9K(b0tL`N!q;{AzkCj;sR>jTTxnS06EFJzLQ|6k;tb-~l_CbxrXh!8P
z?vK7?zwLZP%Cff930yDPOACwNG>%`~#olq<AjbtTS|V>RFI(&57J^oRgPQL9u^Bz@
zVkuAu-p~V1mo8h0yzl-9kQOuiM^0|6Cdc{V@)k)%;fP_m`%sC<N}{fK-7pt)7ZtWK
z9p5<K*t`~I?{ANbp~v+kUzCuy($Brog!nx3gOj`?LhEAf@D-cUe04rcWJFWnr*vV8
zmt=!ggkq!@{Mp2e(fXeN(v|q#+kOktqc$(Pc%`}+(f8Hu2Ko&BiqRXAgFH=Vjc1Du
zVhS=|Kc>?|e`#G0pPcx#)(EX81=+(}*2M+rQ&~CGXgKW7bw$E4TQttGq+Tz#+1%|n
zuZKOP^nO%eye#SlD|uFLSypfU_HE6$jP)0aQ<(kTI`Kth;=DHnmd$U)72~)h{L;|}
zjKSPz`LmSTa!o<5={rjW)~oWe3%92<-9nxS(&H5w$uAzaty5}%?*Q#+C?DL=pbF_p
z#@UWWCv{Fw1#xRGVdET2XE`v^3NtbudZnT~JUp3J0E{Vey3;BsHCLv2=1cOA?(K#_
zN}Aq<%}uUuX{-PR$K$}jd0Yna+*no!oI;tM{S%>`Z<E^bQd>Uu^O7=lw)E1oH_~~(
z$2w@4-U;iL&r~0f`GV8WAu`2!$i`$D9Fd(+d0Jv-DTR8h=<fPItR*Ep)ODRG@j(cL
zU@FeGy*f$w{8lP95XJVMwJ}%8Gm^%h>OG7iaW`w;_#D9N5Bq9a$aq8cBm-ko`3-Dr
z3};NAfSjCDA0c$}G{hbG4w<z?Pyy4pnn@IX{Zb18d7@5+6`0>xcp(pStVjo!Jp?0N
zxu@Lfz=rHs?U0}g$Npz36;$x(K|fK>#L9ZB|HDz?L7GkAoNL8e=<O=IU{S=cq{N2M
z{XQbW2%yueRM!uWvmy_dt=ZE8EcI*9LJT{I=jj}@*>f%%E`vil)!vTRA1fHxz9K94
ze#eo0-W)#JEBd1Le<K4MQ^j|Cz`tKa6W++0Cu%>o&NE5Z)2ds;4~&V0Cy1Od3_-Hd
z+4aNd@`rDxQH8;;cVw1<Ztc~j`RX+ZK)k(G<Hfu88c8KQ`tLmjvIWwZr2gVybJFP)
zol@ZR5{vY(#_B}le7E4j<6<WENE<?h*l=ddDc25riGS!cp6@bQz3VPI9PRfEeNVF=
zf1=bmtNkhZcy0X}g}%$KT0Nxvs0CR+doLL<D5NZ_r@;oR)pAGSpyQtV1*`;-0I;dj
z+A&k4F47O@%}JfXpxFdzgRqglseOY8itST?L*Wz)iE581LWsbLddDX8F8TYn!8K!8
z$b@Y8(NU)iSSeL`c%<(i@n@|+{4ALWT0eAfFkUMRzc^ezP7@mft>{7e6vFYOZ;rZG
zH-pB;)Jv%_gE4*cEX4P$<4_b}hD<Rq$5+Cv5=JfVI(OwFcq7+qc;itbgFLomR|U^F
z2h$N^GIX3%fBqHlBJN^48~BSNH*PjIl&IqEXo+wBx6L?&hWUiBmZ8;wb9F(GgoV+t
zd_eCEo0_+|x*5Xy@AmcsaADGI;*z-)s-Mj($Zhf8AwHQgX_*KY0bq~EflC>BbKaVF
zQd@{LLR|bx%2a6eKNAwuEl+-y(t_$29xlW0D-$x)8R+{r>vQo_izeSdGJg6j!Uq3H
z|EU8-u*dbwt?F!Ie6zHE1yif`z4M3F#ZGv#a@-;7zs<t&_Rl)opA~eHyLDxZcKz48
z;itjjvB*zRAs*|CNPrnlWKvV}`x|R&c+3T)WQcxqN=W-Q-mk#S#8ri;0E^5Glc2nh
ztuYm!1e&k=@bc{?*!SzsN1Dl*n^%-uRAd$2V&pQ4P3x_K`DDVXB_v?7cEs4o?9+*|
z<{QGYfkiv>$&bd~H%+`}vNNbRqfYrJ_r4FPSY)u591oQ_HZT#npTkwI(4@YgUQk8b
zb$wqv=Wc)IuU!zze0DHp$|;vdIUFrA!efBaLf-KNP(Qj0s1sA=uRx5r3D6S#izY}D
z)`etE*C!$n><TzB;H1fRMbo=$<_%kjp_*$IRDaL~Ua8;3D$R+*BwdQEpEE-S#G}aE
z<sdjarCVk|shIHlr+z2`HdXzf6Bv=cc12_n|E^9lH!gUJjT5;?IDRVQv*Fl7$y5fC
zl7?$*jj6s_v;LB#uRS1OJKVhc(fAmH8;2<m^OYI_rt&0tz%IV;oU6T<c^?QL*wr!#
z)hu{S#e<0lrC`N~j!X3t8pOptAj{3oEh5<D)PAM{DAQrvma<JGdfWd;ffa|rjt^=1
z*e<gSK5e1;9~JRmeq^NG={vid04{&<-0)!Yc_n4TOtNs5G6xj=k8scf*R9Eq!rL)I
zPJh=;c|}GZjOEp*guRL#CHw+Q$xK2ZcpULzw_P*OwL;;6&C7Bi`N|7jW8Hbf4P4Gc
zH3fmHhm+zvF`B8J&)c|S1$IT{3E}Nv@-Z}D6UE1WzXv?5-;II@Mw9?5Ek)07!oo%>
z+d453$Jg`1*KYTR`^0J|8lMMA_zl9noA~?}qrIV!wb5fN`vML!&RmiNjF&L@-zO!>
zeLOE_a?N5>X}QZVJfXGb?g1JE!6-Uq3_w6gnbwa$gu<Gv<0RQ;bzxB&xR@Rreg)je
zHn%(qS8tz5u7B7EXNa?@X(lWsS-wqeUk>U4H%HpV_kU(r{;ZN76?6x<E*!ccM@w2F
z(U*N=(^tR77RFSZbN8ysrUiNb!}zr0Y<A5l94#{VkzafLj4VSK(O9C0WwTF+p*wko
zgDoFck`33YuTu!%){S&M{<~{C*=R&y^qA25wcGob3yX_j967(Js4L`pdgNwPD{J1S
zWsSuF+~I=5I@gYlg8X0q^YMS?DlY#tC(Z?v6ZoNywuv~B;@Zu7RWHF@RU{`j9wH_+
zBR$qezq;024I9>*W3iF;|MWxVYWd!dO4Ni|{${RT(7JkFdMLr(I@0RDBcB`Z`8m9=
zThF9EHo+|Ew*J*l@3G`6pA$me@;5p=FL0iyDY?Bvd*H#alhZ`(|1-c>?uj+@SFp1b
z;RNDxImweI1`qTRbzQzNf^w#pLZ}cJ#bBdqZ?0113r&3D@CxGVk^Y3whC|_pYKAuB
zY0HS4n)?ml8o0mg_F`cv_#hLG-stZ$Y-HVtn|j%Tz=#>xNn&ZkliPos`z-r!-Xg(l
zs9afE)0{+m{z(tf=Cs)>qlcuA=!Y{qsPGHjF;30S?Ai9!DCdXo&Tq+M@P0{F`OWLT
z`ZXg|Au-tB7`|*#Sm-~RW1dO5=p*?uXWkW4027vj>xRkBq}IkZvH()Yd6?+QF;9}d
zZD#86vsBbd?m&AdC~{QXFa-d657~puQ(3eWoazlLrUOm}S+WMfMw~DfR8sWE$7M$m
zP7PX2wcQ691u(G&!3bP}L4^|nUc24w|8QoOYgw0C8mjvxX)gH83CdfU!TzU|H~i?7
zz{-`7s)x|^roS|S4?Sg9X$GJN+s*S7?$dzNzKa8f1BOOU&1A4w9!Qz*`WrZ?3>TnK
zRL+?EzUeA$8hY;Id2tec+^H@VzHvizyb=-&n;nVoxKG+y2YXC+CTtegmN|A)(^Bh1
zvy=m53csVp=jX-6nY&MZ*a#Ur_xEluX~}`dRnS@Q-(TN|lyS1e*MS3doCRAUvI6c;
zq`S~LoWh?_gMADw01iLahDfhk8yk`kY!MfZ*25&Zd)Sn)3_;q-bLFb%5F9Y9K|t8g
z&r<eTjLt_%ujs>r_x&<3rtDQA<9jo^Ke&H?B92+tBDr{o?z$LV*ZSq)=;-L|tnTyl
zPLXc;FsPot-FK4-9#Ak~vAkTbO;l-NVZCdj8XDo_ko5vci2obOX<PW*{hh(HUdWCj
zWQYA+ZpYFnuZIS~i_cCGFsG{jC5s)07S|re4JRHLn;w8y>IK&Kw)*erRCHaBd&`>J
zFc2p^Y7dV~*DZh6{Z(krEwFv{$B!R#R|7thjORBA2??Xzdd6)51CggU5(?BVlJaY<
zzMFcq{&Q3SG+l@|-8Lh~bTRs*HcaltqiM<u=@6{0lONtbzvhgOtfd31#n#$_k3Dtw
zOss*ejT>{#^N;+D@S=~ruL`EK?pAMCgMg|~cl7Sj$FQ@F%k?VR2=9fvnUoX-)Ag(R
z(tf#(i|9*EV#((3|4kXEr#~yk#>Y@jgu`x^N@1-)A^}^(aIT^Tp8<O<Bj?T=e9klS
zFM_#4U#XL8HB=1BTD6Fb++LFrn`4w|Fk!v4n_^?dwS{U9w&)pn$}CQbynRz#A#&ca
z+`O*$Yk{XOeAXL1Y9NEdA1F2E50PCzs}ub|>wf6zS(N)>I5{A!t@<5mz%=EWiVk<g
zFPbq9P}-Zk9dquMze7(??HJwklRmu1a_A2TcCY+icW+m2=58q)d_4_!Z@*1&XE-2X
zM7OHW$RAOM){1t^iV0U{Rg6rV9x!4#Mmzl0?V)WP#ELh_Nm$(++51?O*kgin0E6H0
z#)GswRoJxpy7(B+%w={8o-Mf`5&SIs2tRrN;yCV<>F)%60@Wv+Ya?f5D1|)nrNCAC
zSv6i?wSK%4THWa9pfVwWN|gnpyamEs@Sg=g4jMTtuqT~L>A1+9mGLLc)FA<>==nKj
zGy5`bwWdK1>MnnjbS0ARzexg25IVvp=`Q}E|6Vcc$wzyb-v~Y#29}3P$Ik)``&58H
zQ?k4W47{{BSehmX8BEL5@#%A>)@Gauf1<)t_>0vF=2tsg9Sw(hghmQW)!MZx1%Z<k
z&TAfCrgk6KUc+qDK00v*>-PHFC3l+;kU=UgE)Lo-b!~0N-lA_~H=O0~rayRhF0y=f
zo=}jRGC%DcFwVv@z)|d@Jv#D`11v++NfKzDT1vJ?h*bPo``7zHF@#nEp_RvuoAt0e
zq#zAk?|F4AFrLqd8flGBpw4d!)t-euY`0q*%J^^VWS}pD#Z!3o3QZXsmK!ClSC_=H
zL-ae^fq{6JS@z*XeflX)h$LLYF8y;N4h}l9W_1syMQOX_;tr)MFy&4HX4J6Zb8YuN
zF8sVm%xHZKMK+C2=2?cZnCr&l`_qauB0ITUtAage?4wa0l2QK7h-=Wi^Pn9fWs#MY
zmc8%WvF^M)Z|sgd^R@{&c`^tRi(4bW>Eu&C{VWhV^)+%V;+o~*pI*z3KI`FX|4~$q
ze_l46IHDB!T%G3|_YJ^=!%A#hWd^eGeFkak=U!rr@JI(x6VRpjGj2kKY2D+wQ3b!d
zf)AH|L^UUC>~A7BH!sJ7y>xVRd{8VkvOy&4k?78e?wkE^32%Qjn~wWF@Z(4KN4AnH
z3M;_HkM)11AML!o4?w<gGsLHS%(SvP!2B`&<BPj5Jav@yK&GuIWdn6RQMuls_t7}&
z#d@fZzpGP)u*5|y%l+d0$;sWtW3A5Nq%w&=ETr$|W_1)^6vE|Cs;GtvOV+vYVj2IY
z*GF<L!>FbySAwY$9bm7pG@h`Le1TCw;3^ntWe!l*Xfa*~LoGVrL0nU<$9=oi<#Azg
zT59oSE3JykKs_@j^A44kWAca*wYm;I)_`dPKHe0$D2f$RE$@fvm+6`xc~tB_dlq`{
zKfMaO9<cnQKWv(9(#b}z&S#$s8=T@Hd{^fp>V`0w$$SNR9MAU<=YLLL&#`Pjfrt$#
z!Qid~4ofB$?fo?oTQc?a0E?o{mp_kK{_WXL%E>Q`0|}jByAv<;4Ag${a6ygQ7`N~2
zrf}cRnfKztJM<Y?yvH%{k%88T5}pEK#CN0^jA92m=a{`?mO-Z|m=tSXm86|4`(g%x
zfQq5aFkV-24M^Rp;L7gzHX0A|R5pmUeSg<O`nkNk*d-C8<>z4T|G>fA1CF6#At7ID
zyY6@Yo6;(uF_W$ZL#mJZ<#UdGG3kI*&=b?#y?#T6#WVGJVbywlaY@C!<u^><y+GJW
zB~bz;?M&{Aa|adb#eLEP!6qEsH-H3um)eA3LK&l4pF0d-ktz2QXtrSJ2?^Rx6UFw(
z`If5f@R#`Gmnux~lq$;gmPnq)H8-7xEGWD;z`~R{6;V6Qs);x(eN!Y|8UZ{lP)<ue
zbJpuV@8dQ(0z-l5H8-(A6Xq%%Jssc@o<qIu6t;R<11cjX#>S{Z`FbHanO%~^er{^4
z-J8z)qcwT<yrRN9o7~mq<-8BYuA&r!3#@#MTcpggE_0=R3wM7}o}T0P8~1iQptE9^
zt*fJR9#;4behsp<)}^EY>HH5igqhnQK<nyK4_G=tH#JKzH0JLljgP-jAdX@E@`RwA
zia=yU5}x>U!XM&LjVM7h81&n@L8)W3A?ojyk;VVsB^trWh9{hYR2Uo!4L(CDq@<Sr
z4L$s3;S#Ii2XFZYg-1$^;fZZLH#vgEYs=;ewIRPci@So+vEjfQVz0H32asFE0{&t2
z^6Fy6-khns5beKRZiV|{z4iM?-PfzhdJ6EATjz&=#qMe)P9o1_<m80C8p(BhJ{FOI
z4B(<%xeW5<H~y%_;aAQLQz$c3tMHO};Z{X0Jvju*-$hpra{ZORgZl)P%H6%tgujX(
z=vD!jdJpP=dpZ`dl{G>6G3TkAn7sV;l6#HS!|&w(VR=Drk&mRT|Mod}`pZ9cQT!E?
z7^bX>nJbdhr=H-;gKv7|i+qNa&1(LQL_96((4@=%05zUf_~KAgc&$J|0)2W=>>(gd
zQh&KWs4D=q?>)2brW0B(eXfdk%XMh$WvRyZcmz8pnJp(F0FwvvB<E`YCL<u5P3cP}
zLe_)PVm~V^{mxgtuYe;JW6R(Mv-zp7Y4iE{ACLOAzivtOYL6j~Bjs88k$NAW|2aRa
z5=uBM82LLNwb^c_?{&;Xt3^^DwBjbaw8kV9Rrcqu5M?cxjOATH_ms2OkP-h9O+}K5
z+CyL@m_vF^X*C?D5wum+>+OI)`mOKc;@N0pSiyg<tlWy@lZqFnKPu^kOzXL&oH)4?
z!hw}z(mGKSBXerk4AMqk^1HI^Sq7>rjTHs@TplAgE817m<pr-uUYKp^y<9WYDqjL`
zUJ0({3X(ylF0Uk=;lI0%{ZWkcI|Yqgqa(WQz}NM)rKRxm-Kmk0kyWGiU~*5|cpO+Q
zow8=$MxT~V$Q$BKp~Qb>)VE#Eh0LIHs!#UBjxk}HvS^6Fcj&N`e?rn(E35S33*z0>
z#P?vO1FZ^!1H`nc;b)#e>m_sl4tel-!xDr^4=iSVM*4Ha&wr1&df?c_6%n!M@Y|{Y
zCwOs1uIL_?Qw*~dBtqf}<(e^4MOYPQxfM<fYl1O)5OvbE4K587(6r`3nL-puYFd4S
zh7*HRSjk43<y_PwbvVDG$*@d~_fVH_4RFoRLL70Uxj@QZq`lC0g3UqG@=C8;7mq5S
zwX@yRP=DLCW<EI4-1TWl^ZL}>@6g8XX7Dtx0sB4FAFK9+?)U`z+F9cyVMm3m%sG;4
ziLX@|sV0#?a8Y2<9}c7RN8${es7WT0Hw2#izCHNT&Tk}jlf(z54a}+<>|o?r7>1Q2
zCEE0-tp_F#2f%4s)YYcM2mA5S96`X(rVBv#e=$xgHN`pz4s*@V&LWIQ7uwukZq?Hu
z>V6m0cvnrJCQJtT$HwpeCV<-&zn~yXaN4}M1pNgl7@@hjxUn}>XT4b(z?(Le>eayW
zELvwZDd!&Aa(?32%ae^Cd^yb$&2+f2C@Bi{TK`n2u^~#Tq2L+U&;9xwARj4>8N7Me
z9ErMJRm!8bxjQv6VU`Wqn-O$0%cfqu@9ia;uPN`}Ku7)Ejz8Opd`i&Pe>v5AqE8!d
zQWJS-|K@FMERv<`f<^b~i<gY4Y4MKwIE^g`KT*@nB33oEVc+bw5dr8#=V?XG{rtn7
zjo8rsj$WaSWcWU6ynDA1K<gEndCkT~*ZSr9D~0zfo2WSED;@4i3TSg0M?c?}e6CzT
zj)g-LJ*Lk&Eu=^=h5;vzS52eXC=ir$7bsN`G2oE2In!&5sa5oec`wYGjje%_)yjY4
zp<?OB9xWOnzpTYWucePaobIv-jWwW49w1Z1&}PJxM=)Sg#aOlh0l!-Og@lFGa6MR<
z!5cFF=`4kxk53s@(DeRF-Zg8Y_ErOjsw>1U@v3*60Y#6st140*{&_mST9hO8t8u;8
zIBOzDgDvkX?LIx=C}*&oT0T20@_cyoe6YV+mMakc<T2415i=TxW;!QMl3Hl<&y=qW
z1EO|Nr{}$zWYoR$EZj--Z7Od{+{-~YHX)R13AR+0Ri~)Rs5^%G&N}HrWaWyyM}2oJ
zi=z{EZkW?@R;;Scs)YxR*xGq)RVkMs_3fzU7+xsgx2EfaKx<2QPCItdw?l{h)-{<l
zX4xcFuVAfpjM+JPp_mYY+{^WF^r3b0c+O}jCL4CY5=+i;D7Xky!dl3mQ$>mQ(_gFM
zKb5qPW8Ip~5ji9^Hen`t^c=5LRTQff^Z^8)XdZI#j%MreYxb7;XbxD6I|%JoBNcKB
zKP*{V6zAuDPjHnUtLjItb|x(@USEInFE1wPI6>g-;0q;xM9!$Lc{P(F>!&GSWzM9r
z)D(i2v%mjA%M8GQga5K`ONdYYdaSWD<x#?NWjI6S-R7nf@(B9EZFoS1h-U+1nc<*d
zT5+UNopj0=g9mqgNruAuB_u0|wA_8I^%ZwhFuK0(y)s#)6t>zVt@azK<mt22+vloZ
z#Kg-i$zp7Ymsr!yjJ~C83-J+<sSG9Z`+0)Igoj@>9Z|sD)((u%`nC^5-C5|<U&`D$
z+x&x_-<jg9-=1}!I{4g-wW^aIaa(_FZ<|{LWl^g|`Gq2d#)kESvQT-Q8PG>5?Z4aj
zcsuSLxjC2;+;zS+Us>Xz*4Q9Zke_d65o!7Pc@_zuV_`iElij!t6sHYrtH!hme<#;O
zS!XSx0F<t!JIOP6BVp{<>3rdzF`-@R5W-;xe57%t-)kK$p3{GSl4e#rkNVCno0m`e
zY{R^~?Cp{IHDLCiTWNv5c$blh(mgeeq(5xBfOdUZ=4!UQ>#kO@-zVg6v%=M33pI8+
z<ypOIHpd0OhLw_SOTg?NR>Y?%=#%?z$=bR>K!8!B`Lpus&VaAs=sb55t_e$V*Y0KO
zgB>FQLS9cq0k`#Lh~R|5M?<$me4m83Sb;bM7#yrP_V_XQ`;+9O6`7VkYTq=Ju;Vbb
zMsmf=x!iZkf8~8<Bx2$gzf_ewk07*HqI&ksiqt}^Ta67Xmqm}AfQo;!fK5wFKuH;v
z4zR|TBaMtXlrW^}+pv@Z)DC#_v{DmRm8wsb+!{uqsqQEL&If+091cvY2&5X%k-V!b
zVDxsXU?-k*L-m~DqVadAm9B-o=kC(eqbVVnkeBL)t`UNNNb4q|PL?S%&selO<!@k!
zj*DIHXwjc<%KV(%qP(2mys8nNJM9z_0P1zSKYVulBcWT|(jwG2t6@Ig@&h{5|26w_
zRxPbs#?R^G*Bk2k^-{pE-@k^R`SP(rjg51RvU1|}v+fsbbs<*92>ubv3PMfS!s)h`
z(negb+Nun_8pPEetM=l!H*(RTq0ilG`71%gX)hqWQ@0yfIS9Oo4bpcSgw1Wia$+Mf
z8Gw^H2MTT_vbXxX`0ORZcuA|v>s|i0+aK3CTpFI0`erl^n%h@R&E=9~&q^szk`3`Z
z!2uigTFg6pY0LH|dA{Y9De{o1@jv=SD>TShnkV9G&rkOOKl5$6JoYyKeIZG<@LL?U
zXJNfkjvgwXps`<yIInH5{%$;nH?U}^sIUSyw@dQ?$z-v#+=P?{NixyQs~MzP<#s&#
z=nN&dt`XvwK~PxdHKle<*ZsG>yuH%?qlzt}Ro%P+dlE4UY}?8@$%W;m1=IC)kR21U
zwGo@J>UBJl#rHq)Di@;8L-BP`OZRxC>w5MZQUB)M-Ng^T1%{v{vu+C_xhztS3E=Es
z`m@X3-p<y&ip|oy1eP_r${|O6!|d;@Bz@mu*Jl%!he2%??YB^EPLe6ilBT!9c8tCx
zN?)hx(_#Kxm`O7vk6*Hq9N1roPB7VNnfNjqWd(lEg%wAf?9;4=1Rnc@hdbMcUeC%L
zcjVCg@a+yqPhszfJ_@`1(;8HsKnzUtGqrLa9kQr2woIzMWxM#japEhG%CZX9n+Fbk
z82_Qp2c5QVeqN9X4NhCC&&*_|k+Jv?tEATY2uAOmM-~@do9b&vY%rK2?p~Q@VQ=4Q
zt-wGsWqw(Y<4|g3MuicFMSul_dA)ue*5`7uHXgj=ynWe3iYk%^e7@#9js!Sc+PnGU
zi2r@%hY4brRZ-6y>U$j?vL+P9#%|3x3)<ZDf%cF3?T6#L^NAy*hu#E=2`wGZn_}0m
zG{r!bpuXQC*AZct_Xb`dAgG(budoEPw6_EM|G0InB3(@xQ&WnI6>j6<W@e<WjG3<&
z0_Qk#J-oMEwY0VQ?3pAJ03J9eekv5}vs5{8;bSUVI90m-2#;NK3`H9DBZmR~1k9FJ
z`u3KEg2X$GL?<e-%r8cy)Z7eW#SQFIi+{N(y0>SNYQf(iv}LIVR53)u_GuDsy@r(#
zx(Y;}+4DDS(g_-gSq0i^t*oh1Y@BRg0yp?s+~+uo-zs-JtgUtecTwf5k9bQYo79L<
zh`QW-9qJrz;xZ)-IHx>%-&bUc@%Egkm~2ebt68FnjwmfH%zRJC)%xLlo~!&pUizkF
zEXr?ZJRp}cL7%KxI#8q{e);x;qY=B0x|#kGr|S#72?=`pEk#P2v55k?+K3_paJHs<
zW)FW)zgJ9Y|N4TY*4VW2R<i_oKw2-$kews=29qBUcp8BecsaEfl$M$%-MM-1BlP_y
z;qdP0=Ynng$1j{Z7(&>L1c57?mM{A<2zcdmdkBUh95MPass7AY`eZ7K5E|t@#QHI}
zoeqJQNA>5>VuQ^|9xE!<&oAP?2$<Wah>u9e;=Eynf~6|dKIxz5%m(vn=g5<V2#M#H
zc!_j@SKk|6WToT_Y$qtXP;nKJwavN+ko|=Eex!PXnPrJTh+}WyqbtQmra~^#)YKN%
z(oKUWPlH3YcGao~oJ?!>;<Ifp{TMZy%fU=@6ZUZ;=W=#psX-E~D}<l?S{iiHH^#X>
zO(M#q<VH-By@oUn8YbL^^YNigjF}Jrc(5=5%;NOpvVzq%-S?;z^eIKg<WyfD9c2g>
z#+Q-6C5bGz<73kH!rS0udgOucIO|Q(PE}_xT33KDLWy8W2!FbYxTI|-;q%ymzE6nH
z<PizRmpSQFc6+EeilPkaZtnmqGtbd7LrPNo;>R-3<#YVcP5Y~mqdh<G=!hAf?3XIw
z?&A2q?VG`JM3~clxWZM@6CZH3>SWE@&*k487Akp9?xBacxO1bEX#@k>3MV2r8cTJ<
z!>3IV68~ptdjmT^KQ7YCNXzcCG)J=OJ$<SR6)PX)ernY+E5D-s2-~@#0^!uz%L75Y
zmi)t@T_pz7CM3?~jMIW78vq0fsAMb*@Ky1w-Qq)NkwFZ!e(D`w;U@IDYo{us$Rn6k
z;Wlq#rR`#^Vk^_+sX}8*OGj6i@S5A(BuO_GK>>lgpX1Gg7L>6^lV&XeCl#DfiFt{C
zlG60&{9Brxp%)eN@}L;Kuzu68a3i5}^CHdPuC9*%w2!`0)V<t@3b<kW@^*q)%<dO(
zV6x7Gci2%Es-#ya(;RR=qko_p_ZHub<9m|{i+tea3i1Q`Eu^fX*zqx@9b>9xEvG${
z0_Sx~CN#`|mAw=ejm-+}ON<&LdxHBm^_9U{XN6=wa;73~C`MZ?Li0HbJEmq2WwAJW
z4f9{M0e4J2Wi_N-91vY!;5%4hYw7o4;Qc3R;xZZ+N2b0n#}rTkyn%gITf1lUPW@Z;
z+=(+VO0B?M@D|j+7$3iQo#)lQO3$*q%0P@Kj;IxSS6JD=haXt<7g06h#}L;DMtWD5
z4q&Mrrz1OD^hkcb2nA+h!o%}&o9n^zvFRa$nlyPhip8T7uyES*ncO7S<)1L&U0t+N
z_rXb`dAoxurqOXSi~1VAAV)9GFhet8Q|R)75^HFv&`ugZMv+jDP9WYRMen7pYrdmu
zQMsAfZv=ni(=1-~(hmqkfiKKp1VRz(A|U9f@%}`}!)IT~H2Hn);MkvkzsLUZ=)J>e
z9E4OF8U;xw(A`Ca3+*yDqtn8}9mUSE=NPr)h9KWq&((5l<G~|4qn$)Lhvd<t@MVm@
ze;WBLzb2#|@fn%AGn#wC=Cs@$3mK$pTJ-DFN-<%oF$0*=L{Lu6aXku&k$(;!tA3CW
zA(PG(@qA0Sh5{WJqb*{68FS2}{G1t(N3@#;_`TPOrxQ|}1s3jw{bkuz@?l>)!9qN2
zJlI!>sH5HOU-7sAA#J?%dmIay1G7Ao!1#-av74y#%;xo?{35Rxx^mVQ>T#F8J0n>;
z{T~+(vh`_SZPV6=ikn(QM(p%2FI!kZ`?QP0_OxwoPL~xTZ<iX~6>d(Jvp&zsg{DDa
zB!^z;3L=`IVDucg-);^{s_&wX^xKz<3-W&}N|k;(eiQ?IKzyJ1bR2pZB&f507}UHj
z#auIwcc)~Hs58)fZ|eNtYa_Fd0kAfTxpI=FRXbyYXkM@w(n6WnUamND55#)JKKwK3
zz8|l#YKgc!oyl1`yuUnBa$L<>5Alq$du@L<wLORU`+Ki`=kk&?ReC7y*ZHi2m=5VB
zscwPv52j1N*~wYH<*Zj*F!E*_X^}E_B^Y`3!Iz4N4N+ogZApdi2pZ2p!NE0hccSLl
zB$XW>HL9irJFfc1{>?rLKBnDj5QZ5f1pHRsC+Lkcq*mAGwqW?colX6=v*`Gw<mDV2
z20luyN!i?J`YeAl&~x?(Wh{{NHufGY0%frdz0m^7*GVGQj!BJ~Yjj`eS)`1PI&V8~
zqe)3mH`1}b_Y~(ALXdO}#Vr9xS!uZRH=TD1|41Ly;Z*PKO&+V%KjYSOYF%EWBUGTW
zwc2pgxn@h-A4WfHPb@zg(nN;3`5tYjMTMOHNoFr9N#!`7H-32Swf)-yxVL{}?`{Eq
zN}iE(nSiMcz$oIzQg|pZ{{56Z*wQ8c#n{Iq_g#vU6B;?rWcRhcd!5l;^;5HGd-}!c
zu#<Zgd5Xm5_nIsz3$w*d>)k|-u@KBH;!M_{5h+z!!9PLm&M2#{Gg}V6mCkljbWm_+
zSVMQGjiKX94x4NEx|^ue|9{_V*55B5Ke%=><Gul)?8VV?k}I48VMQ44pvki1BOMJR
zW-scx@?FNQBKNO3y`;5oV!^P87KC<lPO3M)Uf12TVK}*jmk0@)8Zj?zY(km3V}8NV
zrNiB>dW9!enfG;1^~J`#<bN;io<s_2<&%W&ChU5cW&1K-hOMss{5T`?=aRIpWS)hY
z{Cg>2%GKCI+Cs*UkBzF>sTDfy7lIkQ<F<5D#;3P$@H5Mnkx+JX*D*2<HG_qUhjg6&
z>_o4D<>Y^5rR3zqbZ!sa52iOuSu>^{JE|Yjn7a#<1nLEJ;0aI!z_7sDuAjaO9!>l{
z4my-0{pKq!Ke#@-AvG0Teb%{qB+<Wkd{IE&zJh?Qdd~FsHot*$^#H(mef!qBIr6;s
z_86O$hAJA^<5M8r6|~IFd~em}bNQ_J+ORWE)I6|n@mG?n(f$PzDOvGN)6>llFM}*@
zcsO)0b8dl{jCFuD`@lWT3;26Kx7~IeTCQ6J3S%IUi4-_5y^fDZE3~*L1#s9)0b0JC
zq1bKhK*QIFLGrICrr@@blILp&I}9HV^l2H7j#-!(eOSx0a}cr;gN#f}hb#@7oJkRM
zVIA{()LB!523i>}ae`)#u9Prv$TQxBq?=H`@wdg1OQSJ|zmo;-R=n>@k7IlkBGKIH
zg|YGQ)@{?X@n$bSE-;XI(K^uykPUTAeGw24khe5L4mF0AcD2^mN<Q;%>7>-?UOhY&
zT3E4Eh&(l{k&p~Moozo}8=LN<SLbiq7qc)7|6pMgxjDLid(bT%e#!r`t-dxC9ZRYw
zPz=-2(VO6LZo69>epMcIKKGxpFcY<wjI)3<RkROVRue1+0TW+xc@C)nMzma?W6;{8
z=m&F;!xe>~%6s)T*OwA$B*8R0&J#A0HBh$~IwUi_vmpm#t!F7t^@u;4>S?$%zEl>@
z5_8NFZjG(it?i`8z~7BRWBWg?&g)m8VdNZsk@oP{ChbRzo-9yHy-cKM71hxdzJQvZ
zzAnNz!0Y@sHVt)X%Yg$$zcC*pYjaO+7yK>7Erunz<L&P*mccerWU-_%C^jQy7izic
zNKzW(PBOAsafhT_JTsyv=rDaybOBHiWqh)8TABj#hG-HDNc8W%U6D~O)`HR#sSBJl
zDM}roM=;agbSw%{T@zxmf+x{&WNOzu?1=`-{O0y;s=h#bBp@{TV>etQe&}1G_K;kD
z`wCG^|B;X&RqmTc#QZVIyIHcCqYL?vzb@!66174z+G%u-#FoFpQ@+%;N6ONgb8U|N
zy83<9>wEl{r_ygT^o?40L`(R|%HEtxz~S{-`9#;W3o4TT;%4_5ZP3Gb_ib^#hhn0z
z3Z){PS9#K$I}T8X95%>+XRo;|_4h<c@6f%K))_tDeWR?WBPIOWtl%nAtSsTJQ_gxv
zVNM4(Qk|4HbCKR*yo%z9JCEbU5zxms51~@l&f`0uw1vp+gVDXY40b_xUk_-ln!Xkc
zk^-5jve}FA?ZY|2=H<kubuDG2i8J8*Nrv<vkTRhLcqfwi#}i^p_@RN5u4qgt=V|(q
z#+eZ+v!#Q;ckGpj*UrxV_V$I*L@~BPv~I5mdeyb<F@#()Bt07f2c_L%>?n!^MSP`4
zrFaA62{>QA{P<vJnRbBuGN)y}$*z^lBD^Q1T!}cn`=ZZM3QLVKqq8p+s^IPjM@j8K
z<uYCYDSxCY5gp;Tb9e2*5ZnnGsw<F|q5N-35x72$ncEr&@ppB-ueskA%%>bLyj!*Q
zPA92Tt8Z;@?*^Y|7fX$t8aSJGQGxsCTZjORIAM$5tGB@lD@zN8ZN&YyQBl^A#@5Ep
z=H_6qK9N+u#eE7}>&<)ei2*hY`<@8xTq@w)1(>H)Wbzx7op6RSGBSQ_G|R58K{k`3
zi=%EvV#9-xV){rWd|c<(5OPiLbZJgYALI*#@r;<CF_N1Ec1$iNIw;Qd`z)K>4-D=1
zzv|P1B~VsUcY^od9()dB8FFgm8ry2aZdU4^kG5W|p`!>f=6*kX^S&eW{dU{zy#{n#
zD3z|=4ERrAKDjVAfG=LSPpByg@t?kOp7<hr*7@`A&+k%<EDWR_Up12VYEQywSFVPa
zo%IUkr6nV7_Om+gI_Ivg{`}TKw`?ZcdO20k9C&zo#y6V#d+?|K8<Zg>=oNw%s<N_5
zp2bFE2&+*>Phkt;N4OJXz8t~tRnG(ncdCrD(v@${uXk!Zy{n7lJ2|mymJcMO`Sh%=
z*KDF{stgU>SK0y(PBwq&Mg5{oKJyQj|7QyX(@bIo%7lYR{W%GqU#&#NxBpxA!N+>{
zS7^b4yjJwaDowS*o%_S0g035-gx9}WR&v;<z_>GB7TvV?Av1Y}RpkT8XX3Z&g>~#|
zw;&~s@`Ckm(0cf7jR`t+PNHsgC^l-IZb+S=4P~cK9rx)tNg0AQD@B9zo=hezXEzO#
z9g8k9hd3QHJfqB*;E1m=B5z65iLF%1YpWzW{Y|J<Tz=)#;G3D2q=#y9%Gv2W9$33l
z@$^IZp3crLW<Hx4H_gH|U#$`&S&r?P&N8kLv5*NL)HnKM{B(kkZe-9j?qgp*kLS%F
zf18?lCx!p%()zsZ&8J*D4PedAtsDkLh#lil7YltX5Ss1ex^(=xi$*odU^`_0_qUl*
znn`7S*x-_u5lU)MQc32N=D7+SMu(iiFR71XWe+Jjz6RSnk*dEKX*YQ{gz2Ki;el{5
zE%zs_(0)m>PrSTV=5V8o9n}niVQ4k%DZ@CuBgU+zCOLWZoZElONyd(5j}eHsuE3;T
zr2!uRhlxOa*DJ*sGB=SGCntNH2z4cnRC<HBI3f3Uo2A(6l&@Z88vMOT8SlI+WW^sG
z%;!+h9=y_6Q2m1aG47|v_&bzq=}29}V2M9YhGbc)p92YmO%$JEA}x-FoGou>lojJC
z{>xGrRaB|)cBp(g7KYj-v4NJdEzS+c$8sv*9>v;Ad(n&hKzg+9j}k(WZCpr1o$Lgh
zbL*ux4(2y<nwu6pjPVJsMuyb9+`2^R{)vB;`D*A8<v#(u?hy0h%OtJS?Un>i=)oFO
z8mciAYN<4V)X?n5exq|y==MMorB=l^I3+lT1U*d|_dJvY5e3m18`v*Zv|4G2!-0&l
zTJ^R^%bOQ%f;b#;nwt1dq1IGSI5@Qvh<*G@0$02gA`hkPE%W{xa`QDZIz;ub`}NM0
zOeYr6_^1^&7TG#x6-l@MJpfTs6TpVF;>j;(1>eJ0Cgn7P&gJHV72S6y7MkF6A*0;I
z?5oZ-+u(AU-9Uywx<{PndqPa|XTZvtra^bWec$oJ?fMoKsz6d$_NXI#b2X?tfr=~N
zo+w$nT_*e!&2KZ`h{^4^<3o?E*xsbp_`KnYlevZasoteIyp0BOvF*=J&2!D5MO)yn
z#K?gQuY5LI{x<d!b@`?drZ4Z;!G_b%)3i%Vsl@;u9vTIlYOWv^kse<QU?U^P$NQw$
z3JxO4UyS`2wcA~73w(vsMS~}?9{KRQPvm7=YvaYYc^jak?lftqIQ;B(9k{<ao3FO|
zFVgq+ay+v<Cm~j;a?VuxZV$X`H;GF;)POy%moky+ujM9-@6Hd7|NRvlWq&cF#aGT3
z+r#<fmA#UaW`;>9XbFS?5(2#pqe>ngD{Voaj=~W((E^LpS4p?c=(kz31N2xUrpU)M
zSV=??>-6f7O(?!Dfs#G)xdY71{--84ii{vo*I<N>V!(hw6`zb(e_Rh#wL%$rJ{1t@
z612Vh$T24zm%#*ouA)v$nW%vgjRhYtmEifKZ<7Omo)AYyI4ccEywfE7)#Hv6525N|
zMOKL%dYOd1k0~7!(vq})!?N=5_qhoSld^ElHq7mE_v^DU-jNP-80ST_K~%)7Re{3P
z-Z>p54NAcM!^D_GLO(NnZB+0<jkXEAtKSu1yL8ko7VI;V_7W<|5i&;ms6`eB<HeKp
z&R)TOTN_b*PTv70<i#Yyl_90)JcZ9WK!GN@9!2M#L6$&LrQ}}t3}v>%Uka{vCJN<3
zKB7WMm8s%?KbE!<t9kttfA=7ifAkG=7V(U{b`I9dBTQ2DY2?I2O|z5s(GrZF^6I*Y
zB$jVe>)EsgtQ=p|J<a6l6#6^lTn_qM{1gI7+x4%slY#Q5W8=lu5}!L*wcg_iVH&bh
z>|pTl@mWCT>5JV`^}Z_p`ExumHlctXKDb-+S(t!4hKgXk>%rWe!1|h=YN(6Cl2op0
zT%vyzM_v{u38yM~uA-ui?=yl4DTO{W+kT8e6#E^G?O4EpLkkC|iFF^mq$eQof;7Xr
z9$HG5BLIx!p$eA=jhM?}N%QMh-->6ZY}zBjf<A+scFX+0qsWkTs>>U&mHOW1-?)Q&
z>YD<_;+BVvg!iYNtDPl06U9JKgoiPqTE*T)Dszo2Ey;~?7>%HHeqLVEdzT<DrzkrD
zNc)42qw`Mst6rX3)ZG}mmyLBRTk|7mS?|Gt%}&o(kNc|(dm6+C$W^{*!T)2QER4OZ
zc^C|y@p5)^YFn1a3NXv&00rnF9!D5`X@Ne;f2FK+jxbj@xf(XPhmmK_%(7iC;#s&G
z;qZ8`=G?*p$j#!)9}g<^M#+NWqKKpEJ?CQSyX%~q<3rEJ@Ky8ri_^-QPIT|#0I3Af
z)EwqmO^zlZnU*cKDd2{!bffKz9_y2dx(8yyK6<h+S$70fFS<#-i+lgIsLI>ZDsp?V
za5eVG?)GGJD5pkNOy`;M9(+^}iL6HH{Xd${GOWq}eg6ZI5T%uFq>+|Rkr+sWFcCzg
zl^R_)<p-1j5(7bcju1u&6A=dz4yn-t=^i27{onV09KSu|*fAdMe&6@5>vg`)eo|h&
z`1s-BbmGPiS~r}bsm|hoij!3BXU=zE0{c|e@BZZ3RVw7wypv|mx(E}xi2#<a=QlAi
z;yqQe81%I7*rx-yt)N)Nm!79#k3g?MujO*cC<Ok~WiW6TPR+osJM%FWJ;ey?n(Ip=
zvI`a8)4NICe$S&QcZUJQu6eE-5MULQ)hA=S)^|vlar*Y`8ExsWK}Km>PDR?gr)hVy
zL%ge|&K<6=0l4r<{Pm{y<!`{@>aQCEh2vUX!8Pe}Bx>l7pL|PAO&uoU)!I(?pc6YS
z-ngprc4zx0CQra1df@lz?=9no2B^>O+k$f)!FK%C%68;t#h*qJgkU=ld#~J|75tf0
zX10JEtMUR}iUeLI&rv9VZw@oxvEOb-!TPmp=gMyB2j*ZwwptF;Z3mPeZjMzUlMOcf
zD(DNPm@Kqm_+|oQebIJUhI{NP2Opc~Te>-pn^IsZFpHMv2DL89=fVFNUe?^*MY^}J
zQIVhgVRgSRHZ_Q?(c1)UJ}*(-H<BvMFe~;Kl-N2xbWwooD=lLANXNC`2_#c~m|}^c
z;ec3(HK;w%`SmVJm<b>um|4DLO6Iax4p|~wscCKs8enzK?(zH0Qb#WSM;zXxm!c6|
zV;{A&;5&QVQ8P&_WAAkvQ3~6FyX<#5R>RWg9U`}B(cO)O@mVySDuLoXb-ua5d|#{~
zb+XXtgaMGRofz3-o@z+o7d6*<xcS|Zy_jjuKl}`oAf96BTfiWx?`kD9)0aF7Y=DHm
zLDnjn+q&<`ZC9)Pq%KrTQFayKMFh)ae{_LX4dz6AZ~XiEp^4hjc<u5clzQLIDCgB&
zjb(B&m)gHvPMUM-_?4_=apJFh6?S$ac?Pj=0+%1c3(or`bHid{{>VJ9Mi={WM#=IG
zDAi}u`M9+f0r(-MSe)p6`ZpM}YJ86F;zaDBZ!YBnn01B}R<)aj?mzuOQbnIX33YQ(
z()Jk)MRjp>AnkH!$E_6K2OzThTp)=f6&yWSnAg~I@!hZE2v2TDllNR)Da@`R))yNg
zF|~RqS6H2K(Dl}O-k^`^1HHHfX8Y4dZu!-Ij<AbeKo*_jP^F_sBYCd+FX;UGg!_u|
z&GbZjK({{2nSsn8Il_E-XUGYKhgpZnum;h7i${>$_t}~{)oRFEc28YzJfF+A{@&O)
zy>=i%A7|<lCk2rcRMDopt9(1Q%<PM}5n2!9mAk}REJ=!zCv|iB5Op(0-wqQ*Xt-*v
zM$I$=0YQh>s%(%%N0!5LH7wckZOT-zvMWNWIO9~y{q@KTx8tVgD~5tK8XxUKt|y4~
z=Mfucd9j<E={=*PlcoK6n83qzqMecsC8qVEB3x=+!L=lE?>95B6B`s182Gx1Av<Ee
z)FI?}FRW#k{rZm6tHT+OFpu?GMj;rU=>-6|x`u?D&a6)xW<98%xMS#(d;fTh>nE_>
zZRil(IJOpf^t*q{%i990{xdSd!=+?+<4mij`KV^_HBCIGG(XfMFz_fV@-(aEbiE=h
zY+Cr>7EduA&-XdI*Vzjtk2g~2fruvEYn8^a^2f-U)CljG<SKH-Eu8hSpHqb8#m)fP
zokYLp?qO(MBO3jfWNWH9f`~p50dhm;2cfxnih&nLea~Pawt{0^UO1d&ZphxU$t$^k
zL4o$?R}Pipedvv?MglR&-Rw{uI@xq-xqGGmgyAveJXvWsEPBJOTZwF$oNwb`C}rlF
z7~_CKlgcKOT8^5o`(Jn;V8*;Qfs|GTFDl{0xDTZIuVI3k;SY*`Q8$`GZb<1RjQmOb
zYD*w<g@Jfv2;k;;Hj?s(e^{u6Ioz{o`v2oXlMwJWxBgJZKwqS)7k^MMjwK~2>_{aQ
z(1D^T%I&vToSsO4oTkiNK&c?!Uy9D^hOhI5Smmp%L<9706ei$^&yfac05UC@l(M6J
zetI;%qCug}uVUDQGcGx5FwIY0MNQ#fie(kjGM^xVa$zn>c_6vu{u9$Q=J72*e7z?L
zoAT&~!G0I}Q+w1%Me^Jc?i-${2?@@y6ixb@eG^yK8QYkB^Zjc%7O{7oStfl({nA)!
z;dJ~hkT9}Y^1X<|lwKkyAg*jEY~v|>KZAh+&vO5;BSuhrnZ;Ew|MR=;^JPEGU~c(n
z$t3Ex^}bKJ>XXKnS*60<(0obc5Vl)jdU_H(+;!l;v0B?6v=x3qY*_2#Ci0UX&o{g;
zM`vd6mQ8o8gNyf;(hK%<6uC;xi!HIde~bvWRq%RiRww!-Y492_GQolRX6+nOY&nyY
z1DCm8MWhht#_hjS+9)~pF&{ueanSW5Uz%u!WISX;cp3J7Vc-en<N6Fz)1@hdrOy`i
zmJ$!>vyFv~C#52<D!7a;|B?9%-S>gcx^BGFOkQUTI@by5k8`!^(2dRL+IJJdB-cb<
zXj$*}SwZ;GZ35u6IX@kCOf-7MZYX$rzxJAn@Ht*{&A*<D6`Np5ERmKESdOj0ei(Kw
zD6?h@$xm7=N;}<}p&|Lv5VSKGKbFxwroh#Fy7CF*6*R!D7=@<T(AELP$*fR{ybvRc
zGXDa95WhULb>*5LvA#P0`&q-&*uq>-xaZprTad<k84)1*yZDm`j|w!TY4lY_u0|M^
zt%tQgqZ{4zS^Gt@l%%-(8xexY4|C@qcsC9WrStJ@CtqN%11U88YGgZb=4>Vi*nE0L
zy$=1H;ahF~CAhc1XpacD5nn9_rv+!9w!=4eshP2etD%1}mf&ZPSVQ)ks-KE2x4Iov
zT+fZ!1Lydll9G}F;#0-1eec;Hzga#R30}^-RS!1?9bbt#gr8DuH&?)?W~PD(n+=gW
zxMkm5qbDY_&2`{LHxm=u&PBQOMFI7IvjgICe|UoySbho|(q3Iwpmj~FN6nsTvm#?C
zy#^Stg6X@KC_dE(#6=3sI+t!f)x3O!`_4UprcdLrcR1G3RgpzPf-7|Y-237TxMXKn
znG9<RGf_WWk~Pt|Bm#kEIL;`p5{Ju{t8Ht9yu4c2CN^ob?~c7l#JX&OQMoPQ1@WWd
zvA$35dZ|wJFu8QU*}S8Id7WnVIehtKFBSQgzwU|HFvfsfoCE8pD0hC2@L)2|`Uw<r
zRH8N?a43DnNP17RiTdOe%T(>)5E6D90`Il8$~!;aa2+&|0BjySDHB`2GLypnf~#Uh
zwBAtNWg1WthyED2<>F~CK&kbJPSN=>=UvxcB_A9K30ciO4rB(E6CJ17#GAvR_~(s-
zPy9N?-&%USLHXgB?tGaFk+;s5xXA?^lj6dCM&Rslal?qY-`ANg*+Lgx0tMM>X<(l@
zRy|o{ZosJD&4i8Pb3C6Ec^t&wXkc@5`qjt2K6D!>-T$JEx@FwsGMLojZ>Q%r|HPyC
zdp9(u{?1qpNzuU4yIy^7Xkom0busm`ci+oZqra=xNvsc>fwuX_l{AQF^GpQPexUIy
zA^Fa>`A8ylc$G<7QiK+ls?!&^TPE$wwcI?G|6CP%^)5Tx33W8z*vJSS79!s5pQ6y6
zUBq*ko*+y|N=By`-6il)Fa>;7yWIZl)5dw;bYhA_C0>j>G8i>l$qBLfBb)2-AYTm_
zEKXV+=Mzk~EEh1`m&<m`1H@1n=DK1C>RH#7U&=Kg97!^k-$_7O{yl{m6TYHvd3kXY
z-|HsFJ4Z$U8zAIKOd*cZA>^XLJaLl2J7wD0J_Igl8kBr8h&KvH%EJ_YtDAU8Nc%9+
zD@p!%rdXQ0ub0%FF~k>gzM7_E!17pk?eP3K{?d`@KS^Ls@6H*}kMUV)Bya;YfrO6o
z^oH#z&nNv-1=gbQ*w|!Czaa0RL!vl{R^g)1QglrHvW@U|XIPH2vt8~T(R1H=b+x`e
zt$DH!g<4oxB*xQRu2)}W<r!;m8IvPwb^p^8W=;tl_<i2WLugLLv-X&xMd8Tbzk8lm
zf#s(5>!q=cgMi{<o0GPbc09$or0Vq)b^^1gmzm6AFZ$TAdW>DD+?NQiFjC%*RsuKf
z<Z<i5MaWcrL)D*9Btgy($}P<FC3BUY49+x1cq+aNTMTyb+S)HT4ZkkbV1aP~7rVrp
zdicrw=*CzTX0O|@t#O&Zr<6;@_Y!+JuwTl4xWOF>_;0nHB095L4w(UTu$!6Do4<Mg
zx4<vS9`uNacrJA`WD>Epc|7yX#%?Kh`Z;EQSrQP*gVp(mZyh+%Zh2p<cwc2^kdq{H
zJd50)K86fiBn3jpxBY)S@BZE65bhjwJy$E+Q*n$R+Wn({mn`+BDEcr)lRe!KE`$IA
zo8+@$gKj2LxM*{Rw~jetiAU>bv6chFmV^W$YvU(qR}9ad;j(p8p1|L*?G_D{7N7l<
zO^xgQr^II;pPJc-6(RL~A>fL+-N5>d9JkkNV3|2`gOT-49uBY2vah!cQF*rrPS0XP
zwT>qXW`@CA702}H4qI{7L)+$}PC)rWuCGS-=Y?hm_mKt#6?f9vmcWW7g{`Ua9;q5o
znHIZ1M&aZ2G8uBpW!NoUrTf_#<atnBu@yP>Jxa+q$NF4S3cbpkx5y=KE{KPMoL!#X
zzjupnB$_|N=Y2=97!wP~2SoERfQhmP?ia`U=DgN!nx%S6=+UAKLX}Ijo+$E``*tFK
zH}{+C`YLO|x%&e>8ProTr)<RzN)`x3v+&JmGht{t*yzqd(~TQn2CLTW*f)jw9fdh=
zQV8_GhgPXtI8M>vj(4ccOLHdm^QnIMPZ~61eTwg{*HIdMUttzAujT8+74SY<a03~o
zm0UlZZo65_*!HomOf0{l7?JvaonJxPi^}kygFUp(HxeYC8M%bUj1bjz!)*6-#4y%F
z2jO4k{~wNLaBA^T7d7MYx$(2(=1KS8#?!m=3Zn48y3RSKk_JNOm^`~=PKdGa{SJ2(
zS^2kav`5FDqw8%`|6a&VX(%d2hBw+tczbzec$%Ti1&!rAWepAWDuGr<K#{f6_Mvr%
z=zcF7GN8n=#K;S!gQ&N{+saw9TQC{8$cX8bcD>3#Qr)kd&QT6Muqho*>5L^s=u=AR
zQXyA!%_oW&+Yoj>^JXjb+UUnlHDxlorKp>r@XuaRvRDuuX~(|mQ-AwQAJ&cCwo}Pl
zN%<dM?i#+&lO?9%fLcs#?a^E$la9khn8r0f`DJOoGPz}cuSrs;K?C>GjR1<^!`R~r
zGYQeG0SNX~AX|z{^?a(-xIk4;Wc&F1e*pXzuKf_5hr+njUH+5p&r>}d&#ogF7bqE)
zz36Re`Sy2{M&7(Fn}6i)`@6iLU&1$e;zH((?j6%UQTRatG>!U&w9K09!Oqo{eddIn
zR|_rQhey#?O$A?Ebm>au<atCxX~)K*>0ZbsGdfplty4rl6n!6a+4_r(>uqGk?-c@j
z%sVq+rxku&^}I3stixb_SEwMu%PEMkCRA`03mk{_=w*e)MoWi!dNK~;`ivv4(c8HG
zerfG7uBvB1JPNW&gbhBSCobbXDe_`}Q8(``OST;2nuxv2^I5==Ym#-|<s}A#*;Q-|
zo}bG5)bNAiJZy4OOeneK6w9u$KkN{BIQo?M{eDn)^e<!Nbd<JQf?MOE+GtL}r(@M^
z+U6rP@vh`O_2cySQ~SvM!ydu@HYszgIhNDO91MuuXVIxpn~$!GF#{JuD$XMHgFlKj
z`UkJIPE3p({KI(>EF>=vCcT@_2m79J1L24nHx$ZShBuf_cX_=ti2L=SURZl-{VOj=
zcRBs@ym{cr^Lmwd`JYof11>kO6>hI13?Sa<t;((2dU$h5xL^+_b2O3;-oE?FrULH@
zOdRH!8Rg+*vc=LoIbi4;*5xBjH{=bdNC#=|@R0^+k_eL$A)a4Ul)9=ox#DjV^r`M|
z%=l{)UR2)H1XJEvpvMh_wAHPKgCpjZDCh$c52X$o`x|Bd5UIPiC~1$uds&X2a&$4t
zW?1^?Y1x}veqkSv<c5>!j3sPhaiC!!SzRWT?sIe2=Wb<ot4E8K=6Hs63wsf`Y&K4w
zouWx?u>(I-rohqPiv#l~$>rZHAPNtel)=(jC3B%8?8_ea>&p<DI|*AU|78sJM;v9b
zCvnH8sCQwj9nM#V=8=I-<*#Pg{TuZ-129_L!cvlM{q<0!tb0)4*^3LI2Wo(tMneZB
z6jz9*Nd@VPz*E5ltVbgF3z->v#ehK~LoKM-1^Y$3z%41s8oIo+YoN@E^_4NurnfQ-
zD4oV(SMpTGwx=3vYa^;>XYn6++!8|vHV)6G>s*Nihxbwzem{yYDJ@Bg&YPZYe0#Io
z)_a5sJaw#8YGhh%S#@eJ1+WUWI|Aby!bti%#acRK%ei0iuVkK1fbPH5G!}jYBE3B*
zqsx}Go{D_r$m2EJjBaRhcwhxez~AE7jZ1Npcub+mLb4S_r<Xt(&UMnU*euC{Xk*Nf
z(iXW7lZG$_LZ1Eeen+4-IZ1etbU%)%3P=h2y?CxpNFk3fLscU`0~r=8+nbwe-+=<f
z^X}R6bKtm;z^>wH0<6c3tPx{iQ);VY8+~^7N6ioZS(spsCOlDrVMLzwboD*AHSu0L
zvEOcx>k=}waTpjF_#^KsmA?~L&@$p>LS%yEAa^8pzBL?wcbJ%(7}k%H6_#41xg{sb
z6D13ij!lMUirNBSbn-4Y2lgjWZ2Lb2(*JzizcfgLZp9+rb-O%+zkPe^p~*gKUOl<1
zXsiCspV0cKl|!n46cLxvHUMD^ef`Cj&~Zz5aZot#;MpO*k(d$ztN-vr<Gp~x<CZh(
zmOZg*V7m!wfm(BMs^j3|;0RWKrYfhQHVS$E922>JBnezCyw$kc*T{2F73f?vL&$5W
zsTp)OjBQ;%Z8_R-xIP<$r<t+%t5EFk*J;gVbtBP-eSPt?_7}aSGZDv|a{cPfk&94C
zjQ?RQtw-O4&v=1@T`*B*zG!gqTQXez6Rf|=&L{b`>XaO$s%i#eukLqzwQ16Fvea<>
z?>rk!4}??1F_=q^!Q{H!m`o1lA0?&sQj+pMFx(UVLNboIyc+pe#H?|?Ms&nAg?L?E
zZUdL?8wxmkMFoy+fkvlDXQv=34jSXi-0>zT;UdwbibLs&yZK6pqT&;2ne{?C+a0`i
zeiZeS-?R(%LzQ@u`E<D+w4onAS)3LRi5M~saCn!IDNz7%@aYdbsp-@)SfR1AKjH|O
z;t`5t=pKOh+@lNtHKt)9DNt-ST^nC9dM=^%p12>FlLs*Z@g%v)=qAq-nkyYcf>bBw
zrG<H;2~>P#i)GPtx>3Hy^f2P8qUhnx8{hP6Z$yg^+iv}+WixaV5z-e%lCezm7V-me
zN`;#53W<!S@A3z}cXSn_X+ni;Rzb(dfnT-i?9C?3`?n8uWi2c+RNjQny`)twe%m|j
z_inj6v{!)xU7I^Ztoi^uB0IoVwRKI;QDKbbU%8pVAE;}CUr8r~>$z3I0_q3dEWbD}
z{t0H-4mi_ytB_w-wCFg?g?Hb_5(AHIrXWXrk?em`-7Jh#Vx)*>ZsmH2VK&5{ZIA_l
z?1q^^*y&G}%`Ixa4r+>wp6*<&qg5KzUv)@ki7gH)8dz-Xxiq~d;pC-McQ#tyL!?w=
zcN2WR2VHA1!FA|D9Bofp1$t!QYX%<0;tcYJdxM{!R*$7Z3PBvEh_CO7B6}s@WQI;9
zA=up_i)V_<D0G~nz}yGAHc|M{NN;7PXwPQ(uuwth`d_i9Rd$#Ot@?}`j&bzlpajjb
zVL=7qY(kk2vXG1@XuYinB8x8!sI`-*{0TF~>d?0>*wGbU1G->(E^-h~$!X^gr7jgn
z7}aiG|MRDv_ro9OlP8E@0RP2s!THGLVVKscEDli?eljp0&mMGrPAr{Cx6eso^bTE-
z?Z4JR4^E6v5Rd2DkbGM5SQJVEIbva=7IF3Ow{FB2bMMRwxB-J$#r6MkR^}_)0srCY
zx2QW}3g<aU7C21x=s+CGp6{sn9J&ww4E-zA&=E1~`pRvSN3IPJYnQ&i%6%y2n6s2S
z^lNbx?C(?ai9q#9>Ls!>p*LfP&oxPALWrmVed)z=JlR**K4UEM`jEWcsfU`DYwEl_
z%i$_r<FfAbG{k9XjZtj+aMLSjd+&T({qkTKx(}tuyG-$(Iy;=b*!HTP3B8^`Cz_}c
zM{NNwPZJFGWBY(7DQQt(uXjp;vo>fyZP+ETbCNJp&bUF#O+$g{+5Q<ogdcWa`R$)|
zCMG6sX8+YVLuh!^mj)hvv%hNViEW+R%KKDm!cm&bfQW;tuutZ1AFE7DHec)=Hy`V&
zz8_Yt9`!86sRbTiUD}Chd{Gw2sL^c%R~ibYiw{RLvjN$+`o*%ogp`y~3YA~rXMfgy
z!!*^@jB^rbj~-rjAFKoliOUn~v)0|@4kA*hrNO-v+%qAB%TIN~vh7=QweWO|u|ulK
z>j7D8MN&?iO|URr_odJ~d^Y)7H^+yeEvf2Jca~cMHR##w0uL@0g#-(y+yG+_WpI+o
zjnceFF@PJZ6IRLpPZVpsKIOd8GqJb<@<9*eo1&u7<^hPAb{>@M<5$f}@sOVk<m!sr
zoCq4!sM}4XTT<;RT@(poieBfLzm|%pgd&AzQE+@4KMmF`!&qQ2xk$c9XuvF<jFU&1
z$&%Gj#4*l#{jl~4>f2nIk?imK-V9zN8>v_wd7vGk$6W9!8!~GnH20nIUkCW{aB3I)
zONMq^6l0F9VP#^BxjL7<YLnk-Y{Lb(Zl(;hE*`X{N3W$(-l>$kwCJ_X!)3G%|8FSa
z7hxsEbY=EO02J_+fOW&uj2Wy%vQ|8(Rd<pex!3#N6Gq0cqhOCn>&u*8!DI!m_<8$_
z0jS$+Ep}g6NHn*od!-}X&#AO2yUXU}V5z+G$FvsHkd$b^k1iciCZ|`cAJ|^1(VwQS
z{-&`0>zlh|5cux|tR{Y=2>X5hg(62124vyvT<k8fdOJ|9-*rw_E83hG*3AszBAFXu
z&V80>z7j|0S#+{!dHW@S1KG<mH8X?R1Fr7kP8JqYq0GkBxvKchfY}y5ThF!lpJmjM
z9iixNQMXqbUuo-@y^}uUHmJeb)`#RX*X2v=vWlpWLD<PSSOy`?+Tp3=V$BSm63Dj8
znuA^vUousbk(@gkw|6y2G_m%?1Q!@J;-OH<OS`33Y5iq*6FY;?<XPx=mp<(U5N{uT
z4;B>Fpex}>yY0iXrY4o#T8mO+dqu$4!{ghG8$Fv1A*QCj8MB`^)+~lj@8&JyJD_q-
z1~DBx64nuP8YCsS?Q6L%4j7@W8P<Y-j<H*b%6m@pt7ZC>%ej(*9u**ck4L)c*L&R*
z*-QivA%_EP;T3A5Uj4n2;xOR98(Y39j830l%vhK_w^35kU?ajqD{svke*iUVY{~z#
zSlC~5y={-LA0&oed>e?by5NHyEy5gbeY4Em-p0;f6JhQg7L;fvvl>AAE%2VJi|Q69
z$G<nPz4DrTr2FfRXIn0&p#7VhoB#W7KBi(;3AKH?P(`*kZ@j~R=Ig;jc49NjtGWMT
z?UCk3s?FgCt3yM)u1_!!wejt4J0*Vwuc_DalSF}x#<{G<23_gICb!+yFWwu4un;#1
z=x4hMy&_sXj;O~_&Xa<%EZ=KYlt~Z*FRPqWq~tZ#k0o(?Z62Q?%8y3+R4OJ_&Zmw)
zp^oyxXsc$w%gX{y^+zxEN4jL~PJ9JrJG@k{I$KV=^c9HeEZ`!(#X>UCv0BpN9h(Re
zr!4oYae%UW+`6%HFpjC-j3AlU7zwpX^<x<oUhJUPwd+vwC8HO<8A}eLYvtopeDj3<
zyh?|})yR`NgC`1q5B2DHmClPE6>$GSiJ{{=Eln!L3VHxuJ7m+^HE6wNI!%3}JK>S5
z2qNx=GFYT24vd7MyT?gydB7g;oI7EZdS$Sxvtm@VoE{V+#R5E>d?b2Nr!k@a9wmNm
zCEW>w)^O6T+oW`PC3VAhvW+d3|Jpd{wFfGk=V1&G-xU8Hg=^h)9HGv8RB;>9c9W3<
zoXUKLQv&V^`8HV<bfo$1o-(({oI90^7=PVWm_2tw`XkL#if<wEp>d$R0*AS&$;rQr
zaWdMP^RF_D(f1_n!(}MD%|&=<IGM8ekn#b)Y)CAHbGY_ZiPK{V4-u#nT>1wknG&fs
z8F<p*9=lP3cl6LqVTGBq@@5jvOC^D}DAWIh!2pM%b~L#oOK9`sgG(u{B(YIQom#~4
z6{dX6-T43$U$G3H3lYE7E+vdW*b0gqUhw?++6&n{wRf!cqXfG@4-Nn*RYmyB;!HjB
z(E1|2tM$^Wbd+|@7guoFD%0*bA;C2p_q9TnUnzlVG$Np3a^zRo`xw5~w=ZBS{$yRW
zfcNdVw1vI)L~1&x7-@UT18vGyE~<myhE8&P%05KY&IYSwnu%QWd|K^K6WDUG-(p#g
zze;|ZX_6KR6snv%ZWp$EAF#8}=Ek<GW;J2*YGu&bShq_plt2|w->BF!vO!CwS@$=$
zprG1{XycCAh6_`(+|{4*a@}^g#I*o<Zad;+j?<+2RfBxIFe>fa46aFm_fWF`71w=0
z*(v)qG=zS(Gg%yG{*14B6~aZDE#SnJ@oyXwCHHjB3Lm<5C3Qz>u*)aJH*oFGAkX_D
zORv}4>~n1rQ$HH8?g{@Hay)hU9X6)%xh?F_yE$m#WJrKZ^i2u#y{gD?w0*TT(2aCK
z+tE9U_SrW;SN}Nf??j>?@8dDc%W!XC|5;}WE(|sSp{{>EJD%2PrxZ}Eq$aEO9Jub5
z*SJ0Qut9Y`faZtZy{wVoy`7oZ!%d80kNg5g<v4N+DJiGstzbYLDry6Yp$kJ6pFVxE
zt4X4u&DqRE*u^b>ak%P&*X;ss|8h&si3W*b;ivZDD?=OJ_BR*lX>;ph+&ycplj0mA
z7cBZ`Y@1M~k9o=vcq-$@?rd($nTYdH1ul`7@U%0Wnlp#UXrbLKn_yUbmt~z2xNmA6
zzEN?!uUc){9XTg*LeGr3KF|w(xiz4Twq+}!dQ1lXwPZ#IV`7+B$pTWRlK*pWTnSjX
z08UD;E&nc^5r5`uu;?3rmnYgT{(T-J#JPKNFs+Dt-eK|A<i4EYQSN5p6Ggz#n%Oeo
z8}|&YoH(3r4^aJzdi0B@@K1CU{~X7Cjy&<V^Zv36q;yG$IBDzEy}3&M4?VZ3NN-?S
zlu7Ri-_=!m)f-jk)~XCx(|i>lO}1@I8XaMrU<3a$C8zn;pp3SNH{r6!`9GY_pBNuc
zNY^<R#|?eiCdyjW#<x<4_E%(xpAX?kyTbVHf}(=9kDwcmOg0^b;}X>IyXz-;awqlG
z<HvMea*j)#{>2m^{s5%;?jy&%z}rKX#3~Y@Tg9i@zu}BR@v*>zn_l#+j=8?A5eqE*
z98vBS{$B?8_;Y=Fk$zP#T*aI$=Vi6(C-gn7r0-GjmOaUQqO;0j#wBS5PLg}Vpk%y9
z)=EzNi<AOybC=5g0et_Rv?w9cvLbUh*rX(f;m>gWLY6V7Mwil4u<gbkhvIUjr637U
zNxOkD3$9OGS03nBHgU-$wyRbpX(guEJ)-$#`ckPFk1ZU<e$h5cQI#GgE_&M|OM8_1
z(|orJgSMs`;Z8lrThsiMSa)q<Ck{;;*W5?9$l$nP%M-qX8@gq4Q%ZUV2z3YwEu`aH
zFPT{p$K#wVZ5~WWl`<4A<d37HAmilKytU0&6ho)=9AG(thsLII_aPlkRHdN@f9OT5
zatJ+Qod~H38;1f|r+ZGlE8*kZVsHYUOX;iM*+sgqDleJnX!pnn@ymAi@m<YZzbFy3
z3JO=F-q*OzQ3@>^Ej?=4@~_D}MVhRp#=nZJY%`Jk^+ad~w&JI%F`S#+o&TRXDu)n4
z7rGx0(9rAFi4*MfvnkAUC^H_l*fPi$lq?R892hm4`_gb%svWfN%ww$^G>o7I<#cwp
z7aUQ`4nO^yt>5$#rB8l4KAZM7w`N24552Yvtt;#>N72y2JA&@{gi|OHn!>4UTUI2+
zvp>s{bhw-%I3(*~rZ+dO5lU<{Cb_DtDT}~{g4YqnBR%~G&H$ap{y!*wLe2n|F_F{w
z$WRNN!=i$_4^mp)+TPl#|Ab;!@a~o5$T1oUY(ihX*c{f_Wo%MuqBiW5Y`Jbdd{z-@
zN!;^OG@8eKXm!gFux+%P9iL*Qv7pGH0+;hs=E|1ur>EpijZc9RbA5fhik~pS7XZaB
zxTF0qZEq=7s($K=xx&fEJ{fg<Da@U_)95?SgUTPk)*kY2e3(qGPBqS43;x(CDvpOQ
z9(tR;thmeju{<+^;FkFeSx=yT@KaIt9;d5--t@s~mVM~j3s;OeEjm++sU#hy)ct{n
z!87TBuoz{stj-1qCXM^yXG59@5d%=SrCwG(L6&5I==>9CU;)wbn;*t*==$5=Qgx!E
zuMe=?NsB^?R^Q?yqt9gyeu!fsRg8aI(Q0G(hkp>yS2AQI=k?pK5_O}bUYbm1@$)zR
zgoin#89mEJt{BCKfA)G0nx=h;Va0%k^G7dRKimGn=$5Op(R5l5r#<z`bXfsPX(juN
z&n_XR)7$>#{USvW5ka!W3_1(cf7oOZxETxfU4DMvOI^&#!c!NQNb%nB@C&S7wRJa<
z(GkclQ<a5ut6Ir;Km4-n!J5k-l}oAF{Jq9+y~9>JP8zG^S}#K8g+QabZJb~-y6gy|
zeXUIHNd}ICr<j-czKG*J;Ev<!srVS{{>L2~mbggM9D%=DrDESDFEoAn_Yn}Jb%os5
zj@4dsKCr!k;TyCCxF6SBpcpCIWlne@wO(DV!2>I-g@YR*73JcSGk^25^Rqua|0uvH
zhZWZ2fF^*p%q(egq53tNEIE9YTfdk}bg?!}kFhKw&bwYp<J(4nU^+3u8JbVr2t;d3
zSBpse=#Nrg;4_USr`&z(UH@2T{Uh`NWelI~N?1N!{`R>R-CIY2xSdN)pA>2ATME!y
zdlSkNZKkgAHa%iuF2fJ^V|oVPel8iY>nb+d0%d(xjc2`!|3N6Yey1=K?37elU0FHR
z(v;&NYm1RpQ%MA978TWjt47T94X4-T=6*P$zcr*yNsq-9#{RzLIv)7mFgu)P+e2ys
z8iXwOnu7eTRWzLg^0Db%!V0x>kqEo0jETip8y$O#79WxXyr!GSszxbkvyxG{HujBs
zgGr-JwD0aPzClYjGPRpkK0*9-BkEjMx|b&ECoP=ZVL-sH;@||!CqiT8)1QnzdZEOi
z&2J<oFAtNXXp)xY5F)}wTc1gRN&XgC4V4;Q!EDVD`B%<p)zbrNZ_IwI9b36w<rFSb
zlB;8G^(Zo^<-BC*3|?@$Mx+S>n|9B+?ZVC)nTau82{J1%weH5@5$fENXyx4et3-I)
z9{dO3$hT8JdRuVzuN60g#p4!={%&cx28DThI)!;ggb83;E;bsn_4IC~$(f|>Hp~WZ
zFYO}K_tVUj0e@${?NWF0dpa!T;!Y#JlHZT!^^g$Er+;V3arq{n$v~tpDw8^;(N_Xp
z8PXX;))jX$V2kgzcjdZ7_B~J#%PL;T`Yxd3p&MfPCfppM^-PR2`Nug5BDH>`V`{CB
zM!Fj(=nb-J%&F7iJ`Vkz?vqgI%2$uLTFI+^R?&(0I2koeN&gvFPKUSAE7yv3i;^k7
z>pZ3OECI-n;qocZkTIq#Ab}uQ2;`grtGqf1Rw)GCCA<M~utaKHFL4@PoEzA71280$
z{LV!aO={C&C#wC9lC*w;tzPq2QGF<%nLELvRc%)BzHm_#sH1Svi0-{e_lSG{F_^1b
zA&xL80NwcXJjKE+iIQH_k>6v%veERU_Yu<hA)BDGXb1${Rx?^wV(SH}H*zgq;gpUv
zb>TyFtlEXeVJRZ+8ZJZ{32Tvd05>#Ii5@K7ls88nI)nwc&wp2gCO}^}7Um%>DYT*z
zAj`Iy(KlS*48BHWkk;TKAoK5SC`VXrn-=X&ayW5lrntc2hD?%f>>l!kyvdAcSt@g=
z&L&RwlYCYg>D+u&zvA5D)k(nbyircDEJH1sR9vcfPo>0HWTve3gRD;g+Jla)dAsvR
zL62|wQc+J59LCsxEC<o%Tx3e%&+vFdLLXHqd+$kV07F~+!bG&bNCo~=_J}VBG{rWm
z#*Xa?b~AeogKK@o0DV9d;`$^rEJRX*Gn7r}r4JtI%)yvlg5ic(7GJvH$Y^CVu&;gD
zP6wSOSlZ=%9Fr4F76(L={zCj>ity)V&=!Bd^+o<u+*N`Kx7t^qdIl>RWsF^S_TLcg
zKMQTS-TRLWsa^XESd$ma*Ve#|GAb$>Fg=7X(WjBoO~Oe+(P5^`;X50#ad>GjWPsc;
zjET?A@Kt3qITPLNF_27BL0lQT0;tKqM(ND~Az!eKU04Gd17P!+cf<!V+wu^x&P=|P
zl(2IdE}UG2!^OW+$$WMDTkkrf8GLZIiK8B9hzLatybUx+y!5g7ne~Yux|ZPce1$F{
z)L&~2t|>c=@4&WHf!fspV^;nT|I@uDV$gQXd+LiNyJkn7Wcd1F!)#N4>!K^s5u@6m
zHF7dHu~{7nY@&mz!!P=;POneAfn(xsax`fe5*FCP?Mi<2Z`7OU5iM1{Rg6W`dWUJD
z|K_BC4~-Vyb>e1zHa=7cJLz8Dkc>EjM@rg4RCjm#!n`q;vECP_4oCkv(Pvt$t)m(z
zD>k6<*2=Nsl^MtJ>Fhz=L^n2iO9zbO`gbU(`wze>C4D(`xofm;BnZ!jbz&k*Evu#?
zzBh*r9LEd=rzwI`pcMf~$lY88Hi?Zt7e$Y^DBK01<-_8`6aHbbNPfL^aWN~=#xzPP
z*#;hTl2G|hv5AMA6z=$J+Kv%5oh@5x<w8bHA8ndL7;<B$?03X}c`T-PPw(;y^(-iA
z0PsM6m1=PF(`%K*lHNC>d=`hGD;LvT{cZc7DLA44N{-|?U=>?W&#jANct8dsiH#Xz
zU`+B;TDnbnJ6#q3K$n>v!NKW##)E?<nl9B*F~rkIh<M#5K7-?Jqf=Ox{*NW-R&us;
zhl{4Od>bB9Ql8ycle(Ki)m=#fypdPbl+*fEf#!LkRWV<ns3f!>?&(zcWJ0_lsmNDm
z?T+8p&345^A+JrZ>nl{=mY2q-RU$;4*S8xfl<g(?8Ib{TvpQ|_zthJ8!=)7&SqW|y
zYHV$Oj(z^ON`cI^U}WPkNFxFNs=P~9b0Z;u2Hq`{u9&_!|Gj}YUt!?_I(CWKG_Xhy
zGK~=O_>&X;p*a3yE|e$ET$JqE@k=_RHp?P}{eBYRG@sd&4|KvGGO5(RUnj>&<1|a#
zE_ExGxMN$^jp$yaCoztK7b}wx1zBNcC~M{^exC1i%jSm*zYU>KD6Ys{wCC&T>7%#O
zgG!}n^HB2&in@JN@D#wAOhd{07BHtIe?c;!<O%b`sdr09-_+EeH+xD#dLJgkl?fgV
zRZA!~>h?hC3KN+-_TRo?9-_AmajAUx4wmH@b8^iOGL&%?6Um_wVsKeqLw7XZoQ+5g
z8j7ItI3-j7N2f0bZG3wS<z>TjN|>~cZ2oOOrLMhNODYf#Q2a1Q*1p`*<52L_Y$MbI
z)pY&ha>S{07ISgco^+Tuo#!3C$Zltf%Qk;+O4K3#z(i&m>gW&q%dssJ`x=P$X8IVa
zR*Cfr?uZN#6#79tUqMHDFiT4OZkYxvXGc56_vbA4Kf$hn@!sNYxH<ly6SIlia!V_~
z9Q3xnu>1sgnZ={cN6M^Np?x0i<0?M7oH^MvWD;3?6dw%kXmUba1C(wheYO2^cKI<R
zbqKdliVSG9+OI%vZe8uxm0q7U%}%c$G+)mUFTL|v@5udX7XnShnQVcHd4+h(+58(x
z@wf2Re*4S+!62bu{^U_K$?k6wkzD=#sZwBl6X@ZluZDbr*eFHRPJPUKu5ACNKhl4Z
zWPVvLUSZ~RZuWA9yOUL}B=hQb%ZWxr^~Kzbb)E9%(XnLzw|jc+BkosclidFCdl*`;
zhc2|-N+D}W+*2Pl0$tt&eGrZjpHkBg^R+MU=ehH<>|xjL8766OWP$huAZwBIVdAwM
z7Y#rT0%@k%@eLH(nXSfcOJiSULJ(j_5C~(R0veqNj%3kUcH}TDr-&_*{x-Z}<#FhN
zvaj3=3JOAI;2d*|)yoH8B#E5nzxZZde)QwAygU$)BJ0k(abF&i1`|StqGdV$9W|Aw
z264?iNeVsME#;blNW5c+kK##%Jc}pJpLoegc34V5zfB*4Gf7*iLUTi5BVgwOUz34f
zj9jibB~Njost&t8{?r#_5=cwZwH29ffiI=TgeACF9>^#ACS(1;(TW979f~-nPGfL3
zG=ucU3#EZrORrAfmwC*=_Z>#~XZ2%${i4!jgu5Z`mWMOkjEh+h`jMf9ge14I0fY~P
z@BszgHe!gzqx>;T;hg~7;7Od^M4jQgw_0h**jls3IzOfgIr-aeUHjpgDoTR3T>q(-
zO*G!DHuuB+F<B>$Y`N<{!gR?boXF_m=C)tpgBj8y%Ud&nz(Aifu~I}ZJ<i+%b~Q=w
zz=NVbEb%d<2cOb5vQ~|+t24*}?_(o>DY+V%wq*?2W>U`+LS$GD7J>z=?vdFeA-A+O
z`k@cTk%lHZW+8HHRIM?;_AiOT%!Tpai78pd-LVw3;j@x(Fl2)`UF{Mu1Ji4RavwiO
zS(6DN%&xIXDc`=~BOHSBK|UF2+VvzLQ2dG8e*aCe<fY&3O0K`3a*zRV@J-PF*UxVq
z9R+f9Qd3heo1^tf5wbM@W_4{y3RxE3TGY;R>k}`o-^Y^h_lt^8hQ)jKQ~}G*C+@rp
z%#-sAu?u>~4I1k$H>H(*4nr~c*Ys!nTj2*6*~AKA6=H+e9#S%L_mn8xzj#Yj?W$oZ
z8xNKi3$EJ;o$0}VqhmBxv;Wdgcb5f0Qq>IgFo?K&JU&)XP|)+$=+jl$w!7XzeA*Qz
z@70Ul#I;TwVJ~B+G4|dyv$Gm3{<LS8W@!T8<xI#rqcwkhu8Z<8R?53VHe_+H?{<ux
zt#+BTu8S>Qs3Q+epUKm1N2Ob>woY7|w3B@$R^V<sQ8*x$&hC0U;8(b$&YLvLtEs0~
zk?0x9;-UWQHpv-#4DkANNnotMWv5vFVrDuIbGg*ED9K^3j*z(Cle~O?=<zIX@cN?V
za^n=~q*V^7Xz)D5Z)vGE1TT2o1O2soEf)$7Nfu8FAZJ^|IBob(W$T(+jdAW*j&A}_
zM;Na=TY%H`qgET%=J1QxIgSk)RhE+q-fF8f!{{V0Njz;(=k@UQ-ek*Vm_ujHq0nsT
zIhs5CB5!Oq<h;9&U&vT>>YwnVPbejVYFm5F%vQ%=DFc)LG3%yK_iq^;@A#wQXcq*`
ztX0Y^MPUpTvaMxx9|@<uy+)08IC$vyS|yI(0Ma8QFz^9gWF7va|Jf&EVuovq=*P+z
zirWs69o~F<%G&BKvZql#$fM|K-5&ZF_TO_Jj%);1&;AIt4V{jAPJHy3utFJvOuuRH
z;#4wX?Vxtq2^x<pun`F%=Ems1ii<s{F$e3q?bceq2NlhMo4KjX-jRZNo>=hei_nn*
z9}4eUN@<g>=cIo8a`GZ+5NDRDtnlpv%K*$wqsI7O(vS^!GhuED(#~Q9At}-qcAnx@
z<iXJ=TB@j<zI4h+5=k8qaf&O9F?C8Z$C}-4{~+y{C;Ief=rmLxf)c3+0f-iY@kzCh
z8K0aBOa7MOjQhvf4%Zt0V+`SQZpaD(eyrKMX6Un7O?K1C3EVOqEmO5hG3Lk)U9RiL
zV+!d{Ec#YY76$i)QgS0-LsID(H3ZjmDF<KoiEZj4%pNdyzd>Rzqx@3%9+NFKKlM})
z+bj_@rWeR%1d}A7$wAzCiGzxIc{jk6AQ|2FUS`_D_xW~uVzLy8Gho;)EU6CQBDY+>
z^e$&EH9+otcZM{$wXi2`irKM3IpT6XK5sgLB)`r`&}CrzHia(}zP-@@601i$ww(#)
zMK0|O?X|TzYU$)$0zWZ-)9;@Ka2-uT8$*tfM_W}hC9c13OM#+E$ba3|*JPr4$4f4*
z^4T=*-lu!08m877?$0%c6yuRV?PHCPnRj{Lwr4g{GEqf=Na#A$*<A|qXTCHYEpHvK
z98KbOA06FZd&?qPd_!oxXu;ZQ1=af{oe)k?na%QB&K|<T(K)wCHS)M}^QP3*nyH)T
zV$54vnVIN8!uPQ(bCev>7E4p~FIg3nM=6$>)F`p+2AEMda_|G@#I9h#Kh#gA;EAN&
zl-IW$A}Ewu*nXVwA;CszY$+u2dsEQy3Tk(u(nrL{``uWf1O#PBH0*4nd~w#XV_<{X
zuxD$!mg-+|B)l+axj?DD;@yltpYdBkXa8au)UHp`pJ7Hv5yQi?S3xbZ&%_ny`ZmT2
zB95{(F0YQ&_b>Ws^%kknqgEB0#M5*nOS+GQnHm!wp`2aVv$grtBFmzWL<jiu`_4<5
zgM6y&dk^YPR+1(&6L^UJ{+aCh-h#_sq3YRi6){7LU#qWjB{RKSuGkQ*JN{JmSGx)h
zQ$&m3v`T(|op<<#%D#3%lh5IjKie3am{^6Lo?b})N@L`H-$rhC>}C2XA9tnKKhFHm
zjWfi0=^sS>OCDyQnfq+)dPOE^>)*dyT)-;7NkNnUX;-O6Ae3NlPsDW7E!h+0-m?{)
zj6f-bhLK0AUSA8F!<2twvh;BzoVV7*Dkd=}qXj=dWqIRBgvf$|w#>6~i0=SLDc2Pg
zX>7*vVY6pfEz23`oF|07zri!>fgfTliQWdRyKo3cnmJ+9zK-#&i9_=j*ZtP+Um<6E
z727sKA8Mp5AKTvIr7Xg;w7xApHp>j9^G{ba|0)xXNQ1Ug{?H_!Q`PD&^A;YY2^OOH
zc?9e3T@bF5rB0?#(R4?%ELsE4JTR=jJ{j~W$@E_IaSREXth{6*ylktxy+0$qepeYb
zTw=>zP|eIWH<HNqAc9$B=a=MG`G%ByA>UhsZzZa)`Ztvdx!d2%T(z!fwt;~Ww4D@Y
zbhvx@6h*!hA&n7{Vv<>8c29PVW%ukWAW<N)<yXo*m20SxPkC;Nm_*=X*qb+N>R090
zWi5MkH4Gzyz~YKRR?q26_1Yh+0V?>3V&OCadTl1%XDE1v<A;pw8Ji}5N6$$&*Q5X(
z3#y`+3q;AAQG~|E@{=VJo#vI|JsuMJ%?eKi;!NMqsL)f!=R-@{q;9vt;qtC(4A>80
z)UjT(w!(flxF~s?=m0XmU>Q5{uVPu%N9TFO+P8MLu+#PBQ$JhpZ}^AgLPC0YxE2pc
z{N~pmU-26U4k4$b?TGZiuz}YBi=XFJA+BQoXnqFaq_MLN1l<lWWf23O0u-~IqD!3l
zCj3&8)xGhuZkjd&@*vCl8P?X>_4%7xYtsC>u9K7C0yF(}Er3oG#~QPk#$N>UBh{xo
zJT~cz`;ZI8ZS`0+n_Sf6tvSuD=EAnc$gWS8e8SGnnJ(p2t>to(sv|d2xPh}Zhf{W7
zKWz3AH-8YE`k>`<b6fTNEg^DW;kxu^P2YFk$EfKoC_QBnM@@;_8*guJPVQ%(g8P4Q
zTU-7zf!{ci#c4j+V66~;DLpZ&!xNe2JDS~6mAsC)m;f#_X<ha_FVyJQWC1WuY0uLc
z)hzo#5=CIA!r^*JQdPr-_2<Q&7&e%8VNfsMcs*%s=7QdNKArn?r*&|CZ51D$&;Al@
zLZBY=3jF-OswIqF#IvEPr}Bf$Ullg#dum5TDS1iv!p_e(o^1dJo7(|l5xCLB>YvR&
z(2ZgKE=k!;k1NK`&Vm5BJlwuqqDLyUWb#dC5)RnDY8^wzww|p7XWAMO%55<|*>4A5
zIPIgyW?oqe^QM0{RvG{2AFJ!?8g#w%?*Nw^el@t+vUh62<>HaMOf+8_vNrkbGL*h*
z&Lw#XVG`EX#EOt?!=!U}jO%=T{QP~q9d5De_(GUqvMTxV>|J@24QvK#mVOVjEn=A5
ztAXJV({hsN2d;wV9`8b!#8~%=#V7JsT6eSGXOI$0+nI}IetehiD~T8CVRi_q<tn3w
zT`ny*+@~gKRV*Y^B4`d2EyjeU;FX9y1P>_h-t@;Zq?l9ke3nZ#wN70XK``k5sbLi+
zE0~ly_JNE$I?6ecetbCsp=R>zEfN!nx&id*EXirPi2Vg?e;7Ew&TU-W=8z05ae0~G
z46`t|AsFbqqa~w49>jt97EHB4rR_ht1<{lgn5~}bY5a`Z?&>$>eTF62!~Q$Q!*!qU
z+B)EdcP+R>Rn7N29RT-WRkq4+nQJ;DY>ck4Q6j?p@YmkYLk|y`CGFe-*n%Jaodw42
z4TYfducvOG_P>8l1wx95-BjD?+c_{Q<4I=3#vT7rf;@*|T|F<$2HoABgJS4zzoXb0
z4RyU45xo8OMce=n<(}%}&sNUtWJPo_bd_A>FzGHuYNtiH!qT)j8G~Xr7V{OH?dyig
zzRzj}{bELxsYqK$rtr<jF=iZ(Q;mS(Q{42YNkYThR}kjK)7t~$%!!{L<OTfL(M}M4
z|6jYKJ=5eQv+{KE{Ri$s_lgGY6SMn^+HqQQBy{fZr%J9|MWJlcSW1TNyki=Ac-p1b
z#?q4Irwo~(@NiS_W9z5ZKQLGGt<eEI0u00`d)x=e6S#ATPoQ^NGQR&XBw-)$C+oOx
zkQnUoli5!6Zt#NhC<(|it5&;Rv|l7fpXyG#DJ2Q$F3dvhMuvIC#s)0n7gKx^p;XLM
z4EyW2`MfQw`5=hny-@x|4fDZq^V#`Z<7zL5*Q3gl%-ruhII%Y6Js#YJlbh;C<BZJT
zs%7YB8XLBGo(xMD<<{ohhG-1|slG!R4WpCe9@SL}$`M?;I+R>D4VngR%W1!z9oj#|
z`eL$ueg~U7Xi|dAbQU0{E}3*`S&wK1O-NFSYJ`oS!rx6Fm!}Je1a#eHjCl)Q4)H)q
z&+uq%qRP=Nhi5a}msiyLtl99L*r&5fs)N-(8>xTJ>ue$wW?7Ag1H%phLIH3!f!162
zd~eLdxg2>L@g3cGxbyc<u}8k~(A3ai@zh_HM&Ga&shZyBLx1(<(K><Tp4s0V5r+ee
zCYd^UMu%%ZjU<%0Y(%pfCsjCF4vr+m+C!}I9tPQjA0r{D4WXCQyFNwx#?=w03yq5J
zhrW))<w@+%zaCsePhz})`SG8>T<S)>Yxczs(O!H%+4FMgKO2$jkjuetZ9tvA%8UvY
zBpWv<C=~gmIE1L)Gm}WR;;5(aG=Tc}2VWe-7yR@)jCrKOk$lsvsB+A6T6YLAEsr)E
zs<_X2SHF#U|EHCq@I0|yaoYMhy#;O8O5wN0zqj5la`4=fB+4+PfKShs6uh`Y=QMe4
zJIfF7#|4vZ<~}aWRq|GH$`}IAYpM`1dbE@;P<1nA)@H84{5n1~lr%4>ylj|mGMI?)
z`8J!Ed^QlaR+jL&$v_KY8Z0W!P%LFd;30$;_Ggo>`|OC{11AiFAZ5Pjo3^&tA}}d0
zy&?jZxEt(ZIDixbhF&7ao*R|s8P`p^i?$A}iKm*`y4Aman2V+JFD_x&o0%E><lStQ
zxV#dIp+7}Os+Aj3sEZZO+Dh}-H=S-B@n?zq3bh-wJm3C3|6<^5T(3HS(Ttj#W#nkM
z{RUHhRK^5izgL+@U&L{4(6*mN^uF-hF8>b7VDBUm`c@TBptAfw&QaeuI(WR>C>;+S
zSn)o`ETdaxTE;OKNvq0x(|g&YIeZ<CHus3VDjt0z<Y}@CiLXTjx)LAf6E-{vY=>2I
z#X0Dbzf<FO8$)4PE{R?KI;lEZHcO4Ka(QUtAAuB=)CPaN{`Sd8dh?LafvrZ^REK2I
z&hk;DR%0QI5$G0;esC5U-G7sx3H02|1;@^7D?ybz_Bb2IjKIWl(uVF|{CCUQy?b3x
z)W`wrQKLRSZtw8de$D>h+s$#R-rT;VKBKi(<r<>lJrh3@#OjB$61j@Z1X5nvMi^6`
z)>%CR5o-x1iB!e)W8T0cn#Ip|VDgjG+!s6N1;Izzz+vRZfoHr<T^|aHGFIuU+mEH@
z()6h{lJ(%F<A}N?5~Fz!t4x4>d%Tu+G@*%rv*`Ulw%$4_sxSKcA3_O{knR=`knR)-
zrMpw4q`SFDNOws~cZoxXfFL<Eh_uAeHNepP?&tgbo@YJ3wSNE2x@+#8J16$pXP>>_
zudiqsX&O}A16UUNu9d>1GCaeq@Yz2yMl7fXGMR=#hBr&XHZ|_0tWZl~3OqD6)_%WE
zN%CwCr4pjQ{%lq?y4mS$@BqPVJrRcZ6lv{=-WQfpu!H@&4i%Fl%a;biG#SjCRyiC~
z0k>ywkw+^M*`tcy_K#OBD|Y^BXZ%v40n_^bKC4(GRJ%FRbe^FO|M3yk(0r<Kel(WN
z5Z}HGvc$bJ3>Q$C(3W%kGd?e%5YZyHIbX@vOyYYQrdx#xQzX=wfHmr{ao|+%Az}4V
zG2_PqsL7i`8+3C7dnl7aE)5UgINxt~Z}Pv(=)XRU0Wx!UyB$8cl)Ec2#r%VqSQHrV
z$}gK{qb+>8Ahr9uU%O11FDU#aT4>Mp{aWUDz$G&1);_!_+%u1to7pmz$CM)O9d+;~
zk$4gV-HuXCf$sExup;?4Z3dzxk?ukFgte+lg;&5wh7tGB06RKlv5WMRj2vzJ#A32%
zbFs`tR^dUK#gR**YT*;|d@mIhj`B+Cj$o>BYFMP2!u!E5o@0g)4l`#Y{m_CZiq8Dh
zY@0gDi^CI|hmEJdEgITx?U}Q#qTZDuC4gd|&utI;wO_#EauM14u5|{*)2ang3T(yR
z>Zz|6wzuLYch6u)#SLx5vL7`#1lyr8q_2N4Ozk+C-cImI-R*U!C2_Bam)~AYqh;VM
zqR<1H;xE8(&4Vj`mBUuS84>8C%D!sBVXR24pVPH`uUKJDgpc_-+GyTR;0vE+jbN#`
zgv61DD9p~%ZS>i>wUyP$U|+&&n`?~b%DyKaE^@KPuWTBt@ZGDZNYGO)NQMd*gxk+i
zYN(*gpajp|EX_T!4AjU{r>7&5QzX|?<OT3xjO)SZ$9$2mLP+i#VWE6`Y)^v4K?#=k
zCW=F$(xPxEn|0$l3|12X7Hb9u9&XpPx&}EqI&y<a|A9&uE`m2#H)W~{5#g#jonLs3
zVZ$DxyG@3vn(!Kh<4#YQ{z!NtOyoH6N%dabtA8)7a3HLf?*;>14kV9LF8m};fj1o-
zULJz20f&=drb7;H!fK~y(TZx&)fAp#=t^4Eys7_1-1KzgwcEm>*H3@EpCVA*XFWe&
zPdnJvJPGrYaHaTOJ@fQyBz9<Q#G6#(HaBAN;O@FJX-oo{7c!7Ug4*KfP+?4&XKQ5R
z;>GFM#2BAN|14%GqF8h~IlEhqi2!BVKgl?wL);1ac`Qc_SxSb#7xHKEZi{md<V0}e
zTqfY+Ca9&zE0zBej_+vl7H@561bjsEUdIf5|N44ldbh)cY%UAo%ad(YR%S_G)MIWk
zTt0miufX^^EVakna~4%O!+oP*Vb=!_aa~BlLR{9qU@MT#d2F0K(;2}rP72sjprdSj
z=+Yn1s8$M%%yB)DRN_0f@ZlxWW_DdR6qcZ*rlR)8d(TNfx7_aUFD}YAU~PaBTxB0h
zYIb$Ha5}Vm`kt-?`f^Om(OBEQvYsR&DtXep{sjxvPE?G7I=|iT;c_JxECx2MlgP%U
zgw_x8SsvUB+<13CcFwN`a01O%?b8dwfT@vGCMT`2TW~RrCk9xyfhZ~EeOHsoIgQ!B
z)sw|_;UenHEI<%#%@@st&Gy*-cxrZcHS0~im8t~U3%vOKWG>X526K9jXE-tWmJ01S
z44XBh#PhH4Nf_UxEk7AtIulLPFQ?cb%47Xbcs}#(W{BoL6!DOzbRhJ2R_24iX>gcs
zO#{*#g$nhXci1LtEiGrfCWov;`KSOZv9|sz!a`X|cE?ZRB|lsSM3w)26qT&%&rQ*2
z<KReh0At-Af~ub^$XFF^^GH^bLzgXw-lnK-_o1-`s@rIYSzG)$sCCF~ck>c;Pj7B>
z{P;A{O*t!0XM0z%=t1Sg%ckv7caz#PJ`i8&=JT?WGz2>sLNlQuZ>5hm%n^JFqwK{Z
z)K-_yCY_#6QjKFu_UT=*ywS6I8C;Ue4eL;l(cf}>4pC%Nl=Z~UK&+tH1IU8w+gybb
zwzsUw*l$@<8R%scrDxhwH9_{_ablrtjT#;v?$P08d*`!7QaQ5z3)s!}Y(n6_YrMc6
zc(n2OHXnJtGS{;Vf+yZyPD_5sIPq+~>*dcSvBJ&}YKZ<uj7-^X@7M2H&WxvntZq+J
zA<SXiioX4s=ycHXhn|Ts<KcnU;@N=hOi-m@$)--iZ_B&;u077{Hi?C%-D9_awK6T-
zz-%$HaL~8Ax!K<%Ksy}-%B1XGou4zQWccRwYZS*_(qL)5SvntN(_vwGYa;<^bxvlG
zz0h6Z%Me*Zp+XmHcEZXqSVPf6&u(_Y%lNT|!kclnh9VENp8e5EcpY;+v|v$0ZsxyJ
zn5=G%5GEN9ID=~sN}ipkxK>xsrwncGT$Rt@#n%2N%U6N#`plFMF+~k8grM2TmbOVh
zIgAWCAMQ_+kMmLK(0%0tUy-1=tVzL)rQ3>SFCnwU6@L+uO8?D$$3(!(a?U`$)q5?c
zC^cQ)$QcWu|Ai*UAgnNq<MVc>@W(T>3Y)y>Xg!R`RNgP9_zq7s>EXtrYM8qA$I@&U
zL=u**JsVLXsuuA}kcK?<mvM+3KA5_{cRGnp>loIPZ^1bl#OMx)06;u*)6o9cu7lvt
zJeB~BI%Qw0I`l6S`<IedfjmNTXmWbUx1P7wG1jBN_uUEWg}0<lk{72DC9^6Y(6r<=
zI=e3>eoG=tYV5}|d?)OiKT<hX=N92rqud663&VGr%aVQy<;K#+kDy24x3Ytt(um;Y
zAYxqyu3fyESgn8d@SX2G-nkx4tKze#mVUxa)zo}N&#rnO_`vWDx%0j7t1^p0=2p0=
z^_F=(FzHy6v)JJgbpLN)uk|;LpIc3LX31ZXspA^XZQsVpWl?VpOr1|k<N~Fv>FyIH
zBaMkge7<$_fe+X4P&(LD2#+o}7D!_Hjg3Ny!ag9j0(dErPnTmd#zzv}-<QkMzmxOx
z&8NM75dQH=y%GzZf#HNMlJdg3eYqB<K&=7+Aas!TvTwA9@Ay^l@bJKMf+vJE@?u8Y
zBfg4BHRrk*xwBhLXbK2li)dfQbeq0TfB)u}(^|ybGChWlHKjxZo$hnKT8`B^ME<hq
zKap4&#mUJhT~it^uugynP~<N5;LUy5WU1>oDWcW<sal1YfGU^;Uep-oh~J!S*ku8a
zyZJ%Qi@TaKPeX&c=fy-s{O;rQRs#@m`}3<P=V9Nli(f0Um0(+b%ZW<FB9?;=`4G()
zM#qhi`>UNtXm2;H<^I#QF<`GXFrg??gKu4r%*Qr-#x8Aa^_n<F8pVmOm?=J0tj}Z8
zd?))MH=mQlSL1bh#i=Q#4wel5my`v7ERwGapFX=(i{P{+lVci(&EWO)zH=9KzkqZ|
z4lNZeA3cfPcCP1Klc5R*jIg<_!PJ>QDM_`CWg`+(L)2eXlu~8DAn6&Kgp76!bh2d5
zIPy9`Bk6f`;ox@%OG`_8cRrlH{(v4<>fGlE3GtXdTB${i)54fUq#=zmHe=Sn6s^bJ
z`}!$I!NoYIc7a?~$}wF{wZCI?!{6GWT@}UVM5*Q_u#!}Z9NWVh7FYpdEjMms?ZlQV
zdh_omyeKIkDA2OT-Kogm)t{S&_HHNKSs}T!aeehcEhz1?m*1rQd|S=Vo06x+e+=!v
zlg$6nMp60u-12)GH+IL!%MH^qAaKLw_;M)f;>UDl0vy3U|C&?&*)bcW5Th#YTZlGE
zRxAqS1$G|YQr+{d`z=CL$h#=)kf~I=t{E%xS8~CM%;co4f=@m?Ro8!(MtmO_LW%V?
zQn-%oO_Hwy>(+qqF(-qeO~>2)0*?pVA33RA#WScOEOUmN#$p(;<iYQ=IDuSfWU9-1
zv>~}XmA&QiMQ0(SA@fAzwi%@ld!DC**9feuL#Ha$L=sVwN&>dyrPEyx{`{z5CM+`(
zyK}#qzM{$Xp{o)hQ)WfJj1MB81dF&HF1mrOzjf6nw?=T9$M%c0!uJawtV7m5FW2SM
z#5T~STIK1TJQ2px!GsxEVgbRapOcdsyEv5WFL<fPw;6{MHH#0rbThzQKRuoO$h5B4
zEMXliG}69ltE(yg=jb)=0|p<JUsd#QkW6neKuaoy!cYURe{*SlqYe7k_kId>TqFUj
zlCE41`m|t*u>qDLzxdsw4U-AQ3@0LD&MFutqE#ov3Lg<K3V|CkG6P}L$~FiOQ~Md1
z$r97jh&nnn@x45zn<{xxETbjT{ocRgay;)NM7ko8p?Y?&bA=8^2P(~-sqD1W!b7gc
zKpYZYbl=xlr1iO{kCfIPeay{?II7gfvS(0BP4WecI6H}4lCLxLbw_h>*QerGwyIb?
zM^+>IPVY3RQosETf}79CzNV(dT<RVsc|Ehib$5j^_f^OY;54(Q_1!D7Dcg8Fkh;Hc
z4n!<&3~eq`3T3FnQ7DBJPgUN%59d%}W8y-{u(6?M$Y?*T=|mh&5{o`DOW#e-Ix~mE
z7S?-UhembXf%_*O)xDS<?adj5aFl<*{Vf-K_d}qiV{1+=u3>_d*X~|goIRdn&=EY}
z;w$aJ(r8(qlgTXcd3MO(>rln#u%3mXYGuxXKeh_4k)fu`?N+%wRU`G}0#OSN%=fyE
zNxNQ}`LrsaGI4HMveCBepj!&XeiAp`s4r?{ydT{!e~1iq@Y**r^jOy6+^6yYNY}#@
zg}fxd;*sV}$g~W<h~?|SvK(aHNPX7JT7qKv*GDbqt$8Y%<~X{rBBu9B_7a}L<qB14
zx!X90RmJR6D#q9s+gL+AWqJAx9GXmFuCbQ*RPkO3xnnUL`SCGo0XtCuC{Hg09QNS>
z+fgHwJYZ?tCpXoo1}6LtnhM}u2<O>!%Jl+MHf*YKlI<CXhA7rM#_f&kmSxQO-m{Xr
zhhwH}LWcMCrin4KLYH1IN93-*9nE7r3312F)6zXGdLe7W>8HlF#c@uPhD~Pao9A1|
z(x_qj8jCS?8Bkq~SbkEMY^md*$d<xX@oM@u{RoEQ1EGucryKZ6huK1c({dL69?Nq~
ze(GIe3&~2m_wB9Jl~xAn9*PXGad~0n?esJWlYxzxG}$%CjP~mOjcs_8UhtLwNxU|1
zvSz<4)o&hn2P-yw$wj!s{Wjh_S<BVU!|(X0cIaSsWIoU5yh5LOC2`}vk<!`BnLn8&
zKcHjL{yj6Rob+)=RV*31>B|>p&#hySg;^#>{=nNqiFpGvlM_7Eb5a=KEQwrdd_VfQ
z3n~WD2JQ{{<$SX-dn|H(m^HiG$`wQA1?~0urYjAM6tYI=xL3wqu9WP4pWlyX0aY%h
zj*;`a7#ih<HKp#jm_8>mny@J7T&Oq96&AK;m|@q}5eOF4nO&{VcVEuZ2EemduZBU|
z-Z1-d$IaiJt4(NgEqeFOjZuctKFV8kF(GH=jKX~sWgrj{CnR6A?jQ(-3GR8;j5Ib*
zo0f^?kALU)EPa=(&o;U5UHC49edL3yo14qg-JFA9No?R{V}2lVAkI&7-Z$MQ)BPpa
zSx~oF?znhJqUO7y7P?f!{TwKsS!dhVSkE?yCHrlrHRH=~J!ScKGZE7VKT3aXZkqeM
zZ!O6m9|MoMkHQ}>8P!aK)T2-!ifEN{!_`uG<7ye!={Lo5y19|wy@ebUxoI0k1)97j
zM1q#n8_bBouk|GOkqz;5U~mf}+v{t!rAURMR*uG$N`a^R4&LL}hwcg^EKcRw&uyT=
zdNQmC$973|XSVvOFU@8n%dLoCY25z=%p50h0pfMAje}RFHB*Wz0EBr%H>+YXO%yZ_
zzp!vZBqGcpA-u0QR1PAK!$x+a%1}h&bu%_Xs>|hoLlLsD(b-8Gbi4k>{pzoM_tjNV
z_sO5jS+B3^c_d2X92WJ^uCBlKEYWkVU-RN8GzJ(?W+<GWJJi+QZ1I2GI$4R^m^bU&
z1qM~iKw?y2K$i;+pnI+;Pc8sUXWL=eR&0<aeHlgBP|5bH0<ngr304JoGuwO=B;V~&
z4|A1Ogmjerq9?Ogl#et=<tVfa!ym?E@Znw||HpyRo9Gf9=J!$CfB2)eVXl>d>8L^q
zS}zWvBs(elKKngXfZ*w2S{l^d$bz#N2*H0U*W@Ak3=F*3*Z6fWIJ^hbuzXF`_sSZl
zm+!L&wl&|$Z#wYKhsX8d=Z=ps7tl-&BvIv4FZ&^>&)XIc(@Tb;mr&IE6A8y9utRbf
zI_p@CCGy`vYlf4s_wd;W<E&Zm>GkWw5BV3TC(mayOfUY<6uE%G8cOGmH(0m8P6Xgo
z(@b;rcYCel!<(wfgL80i&>PNT`^0j^H)02?Q2Nz`46dXCE1RLqKqz#@*2vEztRRse
zvqdw+{isu6PjaGOddU}lxEs(mb*yYtl(KGXRe$<xgEmXmIw!FT4kIXPZW6S%Yg0aS
z;5B+pTvXraE{6HUhxGUQ-BB%o^w}ET%5FZ6rW-$#`xTY={YO?IW5vpnAy_|{SlHSQ
z79Peme%P(n(&S<IahjLN9s8+r(t)Xx#bD#iV_k60U#hEBFXj2u-M_8SeSdFz5T?qE
zSDx75p@cfZp<oUZb&1!6l_&K6)EpFN)E+I}Fyw5=JGWP{e)ALsYRwB!Aaya8O(XaO
zbH^^dybQt}G=8iF6Uw|NLJxl(D!dso7xMSir=_W>bi_?GTSHN-)0!|@ldI$x_s0~w
zhE}M~-r1UYMo3KIN5Q<m-~SjJ7{P30E9J`&zl!iw)DldR9{$$dcRpU5-H)VQjAaq^
zE$G|CWWJ}LCbxXrvH7vV;j7TUuuTL7vWj%BJy(~r0xzb|%iHslN;~O6ZXC*uYUpNc
z2$~udbIEjq)WuHECaW?9<Q1;{Ty_QsJJRdEiUI9RMnu^yt^bi$23OvqzErb+J@Se$
z!Kvo_VY}|(A70RT5k?wC&W0E=L*oR66G)SOm26eAunD0mBr5!Y%Q`ml!j?J5s%343
z5Kcvy>bgs<e@vb>N0iKdE=|^q8NwGYt69l9P!Nl%`V~2+Nk38Kna+;;GXn8%dHQZ{
zG=XC>49q_^mvcPBVZ%Sf!q*kXP^Nnk9gW}A+IiFbI9Uf;)}6Y?&h8|VY7c7zpe?&O
zp#{3E8>kCQAMs$eRr#1@(^gQ7<C7kr63pTn<06vtv2Tc>Ime=PN|na%U@vC%`Uo1c
zI9<(QEwtbr1fLAew$N_xbPu?X6FQTjzpIuVW8LCqU~ON1)44LFxa{OP%+VNr0UOZJ
zHf(%1J@l~`$A|K8W$|`!!=POQz;Hej&sY8Z1|R)%rEn_oZg!%g0WxTEH^&C@E|^Y}
zNtye^wq`xYw;oa=(E7SjzD8GRZZg@riZHAIW2UY6M<PRP$b`X%ktvF?$>%l(-~P>}
zUC4JXmp9MPH&fL3@Mk@FgWpf^hBtTYhk%SL9mE_uCzn)LQ>(S@v>F6I0306O^B<Nr
z63haTH;YB5ljfd`>H<tOaSG=%g!Je>Vb~J25I<SG6~D5UWKDUE9Ht9#E)N6VftZf^
zzY>TjwdM9^?|R^5F59_D?7}|5#CIE>Cm}AL{FxLblSAQs&6jwCjh5)2gvU{yrXk+2
z_txQ<KgHmze$q_RY#T@S0Q-=5aAKJHak*wP{$-D$(3qd2zy1Ew+|7K~LmE5}IAW=L
z++;Vq?$a~%d2e&glJ{WC7}dn(^f=djxB9RMJd6Xkk2g{oBR>`?Kl)aDtp3PN9T!nF
zz1Zv|Ts1$hYrVlIYp+sDv`!@<9DO9V`cNb$wED1)>=vTI9jTTE_GRcrc-feUL6<X5
z;dS2+7sOKc)!##1XGZKa0WWWxaXT$xPO_?qXu8OHNvb$D;e@;qbHRML?}C#+k@Wka
z;%{$Qeep-F8;c3>=GrNZ?E6wI-$Z%+CAVdYYayYbIp+HQgqbLg9f4ZC5t}gw=^Yw!
zB?}8V<5ht-VwFq)yZntD5|WbXIcDu{a_DX#!XViHh%aAKQZ!(vfL-~d24NKEw{+BG
zlXB{Om-6W5Iq~?<>>l;W;=V_RxA8y11Tzt_KUiN)4_TH~W#1WY^3`NqJ2b8A&-b--
ze}C0$E10bkVS`xTUFbR5J?&ZV=`!z_u*yoi4*~>fB?AtXUKDY}?n^@^e9`G}$X=ga
zOWi3vUMonRC5??;^z|+8I!rkTIWM=If~saC5$y_Q4~J46zPB)=`MM0)<Jxa7pJ4eH
zfoFuY;GtB;n{yNzfEGCa#2DGKwX4neRHi@2eV9yhk4+RkK^n6rOLLgaa+Kb}*iZV|
zFES;4Xmc-H_qdC^;_y^3ves?0p!j9#2U*@PI})e`KgqG9QF>%g6H~CSkhPzvi#=N=
zd=R$2_Qs>ZMz6);b47@}hG^Y)<d!7wtUKwK{wO99fOx5YfSPe@DYC3un=)>*E^r%E
zWBv2$uRiS`VYZ5Y1Q8Sq<av?Om<G>^TKN6C^!G-oX%Yq+`IRZ=3MVDGMwhd_=jL=t
ze0}|qS1$bBfj<zP`2h%T;Mi>WukWdOOQR5#!4<iRCHclHW$ZD*rdkIF2PJX_O#0q8
zK#i&MHMZO(fYiIIf%@t`eG?nJ@>XuToZGKNI~=Ar%$eiPW`&O)%fp1+Ma<$bzfa~)
zjAi^rf*LnsgV506sA2CAO4YPCNhd)VSj3J^^?UHrBz!5F%$zpVZ2CIJ*5g-Gs=Y~F
z!E&hD?#%BQZ=i(r3SvN(P2#JY9^RxpnAW39CXcuV4(4jSZ5p+rOCk<#F6I#)##sxe
zyPZSlKUlxt)~=?~wcVPoq;?(STuADgV)1uAW*gcQ*b?=FhtT==Gi74AL2brh#UAx>
zzC&%k|7GSlk&=H4Q2mVX9q~NFMrQ*F{xRCgaly$zYFb^v++RZw4_(#Cm-fjcZ_6&r
z#?BLfvXjr@VLn`0=Kd#rvbcJka;QYw+yq-)X!Wo69>iQUGzvUeXP2kkC@-sArr~(m
z*cm#B1g_39y6=w7f=;)ZSA!6no4<_mZP;u+@shj4;soaY^gFb5yTd(@ret7~>`tpm
zb#{9zN4mf$wcKFjol-BY4<4RO{S}5!14FyNYio%SjmwP-o&HFgz#UD)tou+#Zb5JJ
z4&S4q1jnYLqLk^$LbCBjD<{Qge?^$I-+vBPh~e11*p@R*XEuzW*&DMqR*UowGcPj0
zlaWhQX3?@V=ByCN9!p6f?2*eVwIHLVq9XEz;is+Hl$DL?qIEbjLA@Avk%r+sKy32D
z$DERpQ=jji4|ko*_kDe#(nCWZ9b&=V$NWO!qT|CE7qEUYzp5GHVRZ3^{rVqV%4YZ<
z>?qep^1i}DGM{tGsTnw&dK+=80K4{0LjAz0`h82&2N7-~)wH5IweVwkm+~cWwymNx
zJIm5!=C{RyL&*JZ!bLF-W0+hI2xy`@rlz?|12++?H)ap>abszC5~nApJ)LGZcU)9Q
z)R($Kk+BgQ>t&MnOSAzOl%+0G50iNDxVYVyzX6am*5fRmIOF2;;}Pn7r&{znx>^`K
zqf7X0e#YwA!L%YQG#*e*MxXNDt11628x!6SSE5%PBrS=Nw$*TB{bO+eLQfo;DYikz
z96hbe`EM-;ZMenrm!D0JMXayv$siUZ1Shwc2tP4cp*Qnn?Flo1CH+kh>->H-HGa<%
zt5FnYSqQN|A7bv$5xltg7hOY`!AaP+<PbihFgI-JpY$k9vAhEn$flIRDx+a8FEihD
z3>0Ye^`wKHMh(u~v%?6Waec*M#=e$NZQNv8R7|WP*uYwRP}g2(czd~skVJmk4*Rc!
z+g5=`L;Ws;-GD)!>nqk21Jm@pgZ!~e1<T$--#m#&B{M^Vdjnx*JG~&}MIEJ5t*O65
z-fQopWdf=+d}9Aa{hBzQIbNB!^lUnZy2pDZ3KSP#c%DKq-wfyjWM333qSfBBdBP0z
zNaRe3MCf;N6^>aMQ(wTVZA`;sx#90CKA^7H=FX<HxJ|gPv70eVr{=f~_w>1mQsb$S
zg?JBSP7(RJ?;GQXk9b&{w6^<uXK{W_Hh%RIUGozsCP8ynAdCT4ba=>q4+iK&GnK#W
z2E2%P!x}2idFIgyDBBDe+DmLGE8_0hF+V*zxE1z!lItHljQy%Qb>_$W=~NBtT##ub
znZMlT8|tb^uKJtMgv5}N?H80Cj|b+o-1<gD7Rrj#w;S+UXR!QRDnW=0H>_nX7Cb${
zq9;hym5lYt??;6Z`>JBP(PJ`eVwY;uZK)~j_SXgN=0nARTo6>B&owMFye9MzGp<(F
zc=&8_`cycoXnSo#<_iK;)*flBUNuu`?BH-WAsL{Mw==GKUy^JX$eWE~o=#e8xa*cq
zXU9(&bIno9oCbNyP=WtPisewAtSAy`^ddf}(TqhE;NgUrTRJjxU6`ryp`X8_x>Cxn
zx=56|cJC+Y!&TbAcFe`t7k|M3V&Jhcv|oKI7R}hX0n5CwGoId}2msRwfHn8dc<BZ-
z4Mm?l_3j-0sOvpZX2$TF#RdYw#RoayPlV<aD<;sMcsoJs$oy>5aGDemGo%l-(S3Bf
z-4A#qE1EsU?q4+iiK{CnSxiO#PFYA}*L&DRc01NE{~|=!O2sp~!@)N+RA2>_Ol~(B
z&C2C?P)=Z|lJ(dG8Ev;HqNqoW8JD*x8j3$@zrg&NB2z;!U|%&YIDA|ArEXwjz|HDo
z3|JO&ej#<`T|(|@c-Xw!c0_ieE|_d>V|8(Ja08b@$_*6i9j(xC8RsGNkjJ$dNp~o(
z-b5uIeh!1aoZsEP9xI+~0dWIfUgcqL%~;UU%Phz^5FF#~dVdomzZ=6y87&*>J{F`$
z3czuK^^nlLwR-^ZN-5*+4y6o$xp-VY>pr<wle#@P9h&USZI9()4oL{wwB-Fo+jYlM
z%3%-vYd#?ujw}~(gJ7+p)F7#X(b6EQ73EowZdwpnElPgr*9nEH3=kJ!KhWa7dVHj(
zX}g?!oL35^p%1IBZo;G&kinz}R$^VYxP&JLZbGYtU*FBtRbu#*VgLksFteM-uB(%`
zz&aAIEgKkO^nMIo`K>l9pnrdL15EThB)?%3Js!`M(yPk$YAzkw9sZQzx8q^xkHpO&
zAxVic@^jWE<HvssXUmweO_{MbNGfQyp!1cs8CeqjL`9HHxR=PYPi6JvT;FyPQ@b`7
z1;S(=NcQxnu;G5qHW217kBN_$r-l2A(N6};Cd`ixCknaqvuNI2ISN&T`NxFQx@!(a
zA1d4g>osnOl2K=85dq^MK2sS@6sVEuw_ouhRSQ<`w$)n9x8|y8-*Ew5HOtE}Jrtl!
zBC|<9|7)`wrlC>w+eq(q*V*vM>h(bL4f`@YTP^TDTrXzI3CLai%Juugy5=2Xa%kaZ
zc>83te)?0KLSInb^sZn`hw<xaVJD8<*HhITT3qFv&qDA<Z%)f1+%Gp>;QzcNj9~Gq
zN5osBJHHHv$ZU)a>k(TViJGpp?i<uK^NWyQl&I3dJ_SscDD=M=d136WKra~_r3l4_
zhhCQVpWM3R7hsey%HpREW1#@P{0^@ruAFUB35_hiXT|$mG(2@Zo<Vu6`iW-`KqGRy
zA&YvpbBChx@+8paXyM3f@fW`3%j_O9%)YG;UvR>fnCN6iV=zhDqDz?6d==57L$KaU
ze?}*Im&#I|cycyKjAURIqNg9S#56B`sVqcB>EP17C;{YNXJ4-9*w-Kx4v+Vz%a%i3
zEnDW0DNB-Ns)rO_I`?yxP~BOW%W>DpkyL=%k-Kyq19AmXQ>Z0*eZx}8|BKg%&Ol2j
z@C^y?lrY6}yj8%=HaIgOSDdP^R4^1*%fz~L{baqUWS?K<?V9W>$l@R4$D_d2z`c|`
zdm5xH7Enx|#i|uM`17?nf2?)LQ%`DfAu-bSTdwvchB+usL33-X*ZJU3C`QmGkDlZM
z`Meb3PeW0D>qZK4$+H=)(`g$^Envx0;i&s@^>N<j3A#jfTkYFebj)&`$<HIqw&w(C
zJt1i5#6;(?0&@O_+S^Xa<EiEN82MSH=PsWiOJdKK61n?{NFg6|eS!@wTDY4h*rHmE
z`qvGe-H(1+)%HBNIbUybyFan8mbJz&Op_C#cy7~fqG6u)r#!<51(gBtcpdg9n5;yt
z9~IU5!0lImCajk#2@VOye(^k$V&F@zs1(IO(w7nwzWIYy=z_B<8QB}5aqYLUs1A^i
zu`dL?aZsOS`cD!2i+!WGh#v=D#qMdbm#v(b0UGqT^J;+gs#Q#xZoE*@Byonx7tGab
zSTmf&=145y&UZZEM&SWC)Ea<hJ^!Lpw&?h2q1D}P#th+W3gft_st*_^Kz|+3(BN9B
zmHrGtZMrr#W})FRtgu&0sRK&I5ZUR@t6unJgA9q-F07Te_<KZKApJ5YsN4H^**l`m
zV5y-Vcr;BajR#d@ql}6AH`Ud!%T$FpP>b;}-W#NcF$u`VO#&aZ4{pwWK7?i&5#HXf
z4D4-BYOzNYT`pvhKOL3G4#!+G(3uy~olDPvI;C%WNp7)?4=+%+9&@92yZ?^w?+>>|
z_k6PK_S+f=Y-&%L)Vr9#%*-)l#C}I6SNY&);wa27$ol=mpQ>YtI}=9-J68)=I|XlR
zYb$pF=bZ7LfIon@i}%3R{i?*=;!YH#khciTiLo+ZFm&jhxG4~8Y^nF(nsHfO=`R!C
zYKw~-dwf`x0;@eGFRupkrT%x_=_$~Pkm|~`3i3Q;D$F%BX+J(eyR8%Eo;Cq@G&N4O
zu!F39*l3ckw8Fp2%HaT(0Zi$z-4Hc~k2=v$fi#2dPOh`eh3mo{h;%Fgy>YDkBc>x(
zna>YY0aIjAyJe^RlYB!Xg0xpS@%h3em7gpf@7s~Ihw~klPF(0`n`BIu98^g|#?7gv
zC~-xTgW6d5%8^q%)V8wd4Fu1y`|uk(%rl~FiFn@fa0iopi1vM2kLs%Tg|FBF8?%sP
z6nF)Z(}@dvr4skNe?VS$zC85ZZ@oRsPFCA?uW{-QH#;pszaq)xZRO%Gh32JSRET&`
zW%<ohI>l^Az6ZQyC-L6o@?Ci5W!&`4=6lKVZ9J#*(w!NwVK9EJ)=ui>6`g4zCF|}h
z>5gnZ-V&!huUfs5daNrY-+q$abGBz~>H!=}1$0|BcZrC%3RYXLIDa1y-M=TBQIP#G
zj4d?HNnP;k1oL^>8-rBq(}&w`qj@N6j<ErGzj{?{R_Ot3K8{M{=BK&#@k+C^h27E2
za1_kpGj3(6Vt$6=>*n_!*nHmYCUD~q0kR$|<b?z%$s@cv0c(HW^TCGuCov<S=L*1!
z0{tauc8r6%UT7%^`tV&n#pUh&#l=|K*7M^L7Rl2AxH!^k8J5Rh+j7~@^}v<je~{w+
z+T^q7zFk?UZ_^8AGz-ED@-X2@UI!VqYBL>5&soanx48k^3qq2QY-$0b*jYdGO#Jh2
zQp+TdTgL-?q4QqW2zDCx1*?PwcL(6&Wbnox_lOnl?t*R`D`@0JVr^qR(o4vzSiKLO
zRf+Oh`<Zsj)xD=cgSe`~eXva$55u3I*K16!<knx*U*2(~k#hY<=^b?YCu%FuA8F>g
zkXFJap8cybT_aduN9RQfo4zT(JXU1`mZ!<Qj%|n?5l>j_LdkV2HIF_fj|sr!+A~sB
zN^wtw_e{Co=5tnCUiM;ROIhRP&&ea1$ss-xbv`=|_U$kCf5pm3s2fmZD@I_-4l+l9
zoW1V~cThSFq+>97BI$(DWwnFb?Di6wm2eqQ=pcghn+I~mQ9PC=hFQxde*)PlHD}|#
z>qonODStsAv@aZcsGy2jk$eZ_LHCp*G7{eEa(v`7slphs@?P({Al%8x<S9gC>b}sy
z%>7D$`u0G$W&8W`ewi-K?Vr%M(fJR}<6WCpQb8A!v=U4Ctv6DSv_I^3&oqwh_J2#=
z_t(|d<PjyCD2dfwC9xl0UnW6kKMimiH`LpdG`n4C19ou9X9QIiW!ZHA(l2h~8*gQG
z<$mn@B59Q`e>KYOcaTC`1hH)k-H*d<74oy?m6dT&Q&Z~#tjrOdZ&*zY4P9Mbtv}rD
zz#;eGOzVFzJAd%EJ@j)T$_M`y&wtGh&P>ESjYM_-axCR7HEe%<f(z(x{B%6O$qtRf
zh?g`kCu#5UuxBr7Gq9@!BK18+Ix6=3v}Zj{h_(V+eYSe4;G}Rt!Fl8D2?yV6?K0N<
zld%25g)_xfSIPO^*ncl6LJX#Hvi0AmlJ)x^Q)9bT|FVE9)CFnIQ_EeUz8TD+b+}aH
z#1j;nx7ltNyFB`qqlaPNsS&vKxoV{K$o2;oPTr6>qww(%&u2PbAoG1{DQjDw#j)Ar
zMG=cC4Q-s?idHQBI}?wy2bXV^gGAi0mHFjn=lNP{rOqA=3=$b?G&{++IoY2hxlthH
z@;XFVL)}R>B}f{a@8aH@V!83-d~)V5+c7g1Q+LXI^re5BbtBrxefGNCYTGY=2CB7f
zo6EAa$AD>{`*I7pz&=bSAxm*s72AnJZdR-|A{3v(x^5D}ItTTJ(QhhSn}>xAOGl>j
zZ7JJ^NPC(TjEqjmO8Ca0jt*{$^FXpkt&N42%0IsagV5UsJe7M|Z{zSBsFp<3b|RoO
z2CFUJWHgvF27}Sb?aSih%;-=|%mZT*OGpT0U`NxRDjE$~ZWn$Gfs5b6xB$Cpqrvna
z``v+`tdK|d8He4UEPr|j+#K6oQ~JNozQ4}D1qB>>;m%#xqz^}L9}z&{b(NFFBmU0g
z?Kw-n`$pr#@5eiTdGCWiJXL*Zpc~@b^}Pw!nxgMH^1;0b?=@c$)LK8SW9hv0!d%kL
z^}L*)PYjsezsP>+HGHJp?D4Z9%8G>%{<N?1zzYtCclpknn1bTUZLbC@u`uU?-u;>x
zB!JK*7DT<r;qB`AGl~xRQvqt9$Y#fii&gLSm`~L0Q_NC%ed*L7p8G&WEVmwDI`66F
z9E?hExX{>~Jyu=K3H(cWWKvOozCThm>tbaWWCJYY7*CQv#{ferR?Rdx6FA!0>Ji+E
zW+XYBVhz}yv%iLwuVIfZ`(6Hc8CI=4m1=pS&@K@dk;nv4abP*Kwp@^2n^)NF;Il-G
zKjJ&KC({U%wk?|z<`nJ#-4||b6@3GPV>~~Gi{2^a$#p`O_`vmNMd0E%h&b3B(vmH%
z1w{M~-lYcFez8jT#p^58jf%tVL^kZi8!T$e62#?1(<w&w5HLz-lCD@)3R1)c<nb^V
zqfe_v1w}_ZflqS&93I$C7#)?|J=N2}^bSPqm~+B-kIf8?4e;){-F#Uvzb0mWC16Nl
zQQwV|-;^ld%;cNCR-KGIuY5{pEn5&}3rsK%Pw;@}yS!!6)?g-7vgpA1GVPe#OMZnW
ze?f1O&}46Y-fiF3$w6s8ZX-O1y$+#k<3L}wMG75BEqG#nWNe!FUg!B29@J=FtnY-f
zKk-$ld0s_A8jTz53xZu1diqt$GlWzn5#d~2@XbK)d~v_)sI=Lwi#qC?^E3>QO@s6{
zc7Dhl_c_Y%dK~YW-nMRZyV~ipuQQ^QhpzAajS8?R!#<Mo$t)_!chyD}X82+Kqm{qe
z<@`@H;-MiyLU`vwaNm<ZGjmtct<SE$VF_`-3%@*CYM7*)DToUG$%Ly`TZh?mu(Z^P
z3}Vq8v#e{gUOnnW0`T^5BT*M%ZST{^zIFVh9!L`XvSQt@>M&da9kgV^=XBUFtLb+4
zW6{pj$B7K2+&e3!D*b|=m+l`-cGx^|jn$bIf)-9QH(?9ztxTZsLM)x@TUS<vgZ^!H
z^6@>hS{Q1#vLR8ctyQ9e2n@a>%8Bt6)FQIV{jL6fwyG|$S3pV0zv~<d)V1KcZ^hSf
z%zGMtJ$!wTwQg`k!pDQ_rvmWfFyErUW1hLcF#lGBX`UqENoBSBRvV)6gYB$iezAJb
z!dq}vV|0#D>d!$<UEUrTVNGpp<O7$vqC0SKm8nv5DqSuFET7l4{ZqhN&h=hNaj&I#
zOUUPpBP~txM3~p6`j-YU1eVmjcx$;mN-X!Ll%lBMXLxzgr=t9S7W9Z9<kV|A*7m*>
z7PTN`UQzvm*jLgF&XdM`PZNpqtR2era#-T$pH{Y!p42$3)Z(r%0{zU>YVY>pgEJ=^
zz~Jch-NW_KXONHOW+I<2W2N%}=bNUGCuR?+h=`z;hK7JzDmw>2a-Da51mj&b*nXlx
zHcRe{D!Z8#O}ySN7@-h~gClJ>qun=IpYEqVYp#mYiG&f*&3eB3-7FhUhX14Qe1=8h
zp|-tCq}#HK1HmG-eySFD2IcrZWpt=FqIWQc9Cei(VQ=Ua_kUcyYaS&nNVf+iAx%B}
z`d-n=x!u0*cCZ92O+Hj-S4!<AFn!P^t4DNAXU4HrW3HX7GnVP5o5&lRGopzhoCc3+
zhYNGEP4vwcSrLVU?%Zr*;UnS=?zOBj4C9E2uo0xmLzsP@$J$Tg*rU%F@%iGuQw`Yx
zi=yN^(;^kx4$(E|E0&`B=GN94Oo8j71Nbs2a7nI!?q(op^Wj!Lv}>STV=|cFt4nD%
z-`}VdJPxjWrl+!18!q~I3KxU;zdU}!7Q*{g7fMt~w@I*}0~`=VOvc&L(U?(q0hT@c
z$Ot<NsCCz6GoBaTPu<yLPK0K?lzD89ve!G{;_2zwugp76KU3K-zj}Za!U*l@?fv)f
zU%cvq=W75B;svv!!GD`-TIXzyfom+uXY@rsro!FH)$Vl7vbJq+ZOhuz_Gx^-NV$xy
zEH^ua%hxgEuE(rSP`7L0Gg|HfZ!ha$JudOL&uB6fVm~|s8k!)|Z7O`r_USLQZmxNl
zZ`f#PX6R;~xo_ztw9d6I<hr}hQ+HGgcio)-mfQv=^t#laXUl(=U#@Q&(^#gB`(#%r
zsF9X~_Ti~laASD(n61LDmtR};T4?UUS%`P?;rS0nxZH8uC^-S6h{A_Yw%ED?&`1%u
z9DTF_BhF3|D|(%w^}>K(!i1hD^$hAnMUeaP-+{@t%4lPew{M;dyAAVJXP3q&9{Hiv
z<5NHYX}1rh*5$*cO76@`WP-DVb^gOZ&;uf9^Q_bC>gu;mr#JrP0SAWt>T+jlO&!&$
zgcNXhb*e>4Syj{)jUOVhUHN;I<Mr%j#_jCN2(GpEJAT>aPUmAkXCltU9i982&bB5V
zviR@fLSYvx-wx7W7Zq-#dySj!;Qm$m!U2+1Jq|A}zLB3t^!3mwvWk-PP%wz;5%mud
zvUCN`H)j{5E`U-7-XK2kPs7Hp;4;UE`3M0y<Q>vcrS~V{B;$|#kw|)+r)aX$s5*Aw
zj(`Zntg${@<@aoR#F@dI;6L}@?_4Q$kF{m6;HSV*xEIBhu`L|PTv!<4>MmS7oVx&a
z{+wda_^jEcTQP9&=*3%$kBMfDETghUw9(h+#wdw-9h#b$NkJRl_#U<O-yL1M6`+%v
z{4O9QURYXU`oiH3`Md8JTNyogxB~rs8@qt@74do;(J`1bB$5z$|E6ww$VBx`C|AHF
zTQqNee*TcS(CnZ#ui7Xs&of1xU-D!?Px#9`L&-GK%#m+VMK5vD8R$@D+T>9O=#_U#
z_^egEx18=_Gf(1#e0aXxv6Tvws^*HHMJ3#KBcAoU=ie9hA(z{`Zlo<*{E!U+t9MZv
zCdhK!yg_DSt^xIrrm`G?z!!ZvD$P8;%v4yWtSpfRzx@Y(2%7l^#-(b?H_>*$b!apN
zJ<4|*nMB`zB{=*8AE#koXB?nHM8iZHD*e^bhHU%?zk;>JKjcl1#&WfFbV;ha&Z>{{
zyHgcn1(RcjFrpJ+cwK++9yewgIk1!RdOLkgE8^zhWd}cn&IZl;k8b(Tt4gdTULLfI
zv)ky!*Dd=$CMi;Ak2ZdPL2<(Yu!~-p*Z*+^G3}i%tj3R+q1KL;L}|m{At@95wllKC
z#PX~oAn$C2gcP3la8A40y8mI^RGf-zuwnN2?pCksPuM5!10f2VZFtW34EMMax`~l#
zl**rGd++V<ss@|B5u@<hMth_c#CZPV>mJYU#V$iV+5V)h{yZZCe+n5OLs<24#*^=r
z4YrxJ3eG$AP2W7kPzsCQye`(8HzwO4)#nL!b>#pK2?+_`@tz-9Z*cHnWN-{eLL7d<
z_>v1h!*8py#@5!>`j!^=`PZ_bzeA3L;W<r+_02XU*Uibv!iLdMXV$!amu&2#qh}mS
zu4R5}A&1FRPg`9*A@V!mi29ZU?hW^~v2?U_^n_CASe1RW=Ef&}%JffsT=s|1jt;%Y
z83STP41+;OM^8s9)oNiEx2i2FNj&8(L@ruxH<C`iflOtn9kz6!sgX+ax9Wso!8P6M
zok3oxfr-$8FoK+*!BR#@N(y*j-z}4s>L=1tK4twq<H7geCjk5i`srUkM6=OseSL8c
zUgGI5qS7kN2d|X+q*^MT|96Eb!A%+Tx!6j-{?mwOn^Xg@Xo(O1XWUA`TNOk~|Ic&(
ze;$?~Q}{Rto>ljfcAa1>gFj?v{MQxa8X=e5)c_rxjLM&EU20!zkug0b^f%piBi%W<
zxe{XcxyJ$U&uAE35f~`nDl5U_LthJvbvZ7-Nw9mkx_I=!q-}dsL{`kWRL{-LjoRHW
zwFA&d5ZC$iOMp+FRuNh!sjjP23;K_o$fa*sH@}z#7<totI&Oj<`tK|w(N+^&)c~Jk
z|JB<(WCxsHVB4Pt2~t5bIyRa4`1n9Ao9XVG*}BqSW>?{6EjN2rQjgc;Qlm+Kb+0^@
z0H1+Cqc}YGaMQn2udiFoKEypd+#Un{?4~?YQjb@^r5>J{U7<o^<%_bN+U$PEQ-iWm
z;K7Nb;~_{vC?h2yF}?doN-N<9oQumX2i+d#-)*uSw`|0>)vlhG3gxoEqX=1o9^gO_
z$Up}bSCsImMb8(;gD&TT4(B0j+u6=}>WAQt&CINRues}@Mj@W+Is>@Qa0YG{fvX7M
zY6V0iOccD0z@={6EzLH4ILlZKL@ulO^b%?g8?BtB4{3E@7<CH>0Yc4PT@SG~;u~OW
z;)v-U6q8;0ubE)8?e1w!HIAed5egI;E=&LY{)W`UM)yrsx8J;pcW(ETlE4vg#sZwh
zf$CkBw6}kHdu3j&(>|=z&dtqDnbB^)b7T(?J%aZ#N*yhJ3i!%?K9%k@^!xq=;XQ`r
z$>+4RH1O@lfU&Xr-TVhONOg5}$K4j!WA5s8?xIus;eTA1<Hv_%F#F+Z#O%Q>4_*h{
z?>9@fx3u^yIyS2Wb#BCq1s(hjIsn0qU;~-m{PzdM(^md#%<EAbTx!uDz~ci5ag)j~
zeq6}EUpPBE<7{4<I`qm%<iK-GE}AI7p8)28owu|g$(tNE0<04|+@L|Wlt~=g+1uNj
z-ArZt*NlN1l&gRK{JHw?8D=*nW_!rnl(Lnp0U;8((-<(9>c8(jTqQjEBzJK>HXY?u
zdBa~h8ZFWce-(ex61LH0`8$5TV_P7>e`({j{%#2c(!Q?q@(}E)=R!*^z_RE3#qE7%
z<$n+Cx?Mh21p>b^LTgTTThva_XpxQrlB2|ac1rE%8%`OhVCLnmjkWcZ_fyHM9=xuz
zv^tLHe{e8MdF3?Td(+ocXBTnUe+#}eT;kZbv9VFl)OTxYYO1%lck>P3Z57ygz#|gK
zPrd%0(Rq+=<iyhmcz5obVBk^-s~rLeA<!F|H+CP(UcKACxVWgX3OaJnx7-E=N0+Ue
zYW5oKm3{tB=jo<zQF|{X1$92$?62O;HD@hENZqZkUd64pAYfwo7u7bMo6KAbH}A6k
z)de2fH@|cm>8=rj?=FX?Uw(7E{k(C@bA-j^)AP)!^I{G_%m>;1zK=1x`xj3wwt~n*
zE_p3F{x^&>GBQxs2=lLgF_TxkU!Qh;w4g-3kIVS@KQnIqe`ef&jR<z>_m&pnBpK_c
zyZ(HUkcmStGi2L*AgEkpU|^7W5Krg?8i4(}?x3^%r6ps{9MJs(oj>1KfwNVn7n49j
z(0SW_t^eZU$|au^f-mK~t@gkBB@mh89T#&(v_JM71y;OQ1M$R9euDGK^!D%k!%f|z
zC?wz-9F}`}-B~lA!N#=eM*4Sqj|t3ooR@Ij&oy^Jp(rZT5NS*c`+rS$i*4?CD-yPL
zHqD0-*x+p9qRIHQgtB(I=nMesprL5*zHJ9NO5JA(k5>uZ*V)~z?r>Ge)lmN3P=u7G
z(X;&cTXZR$jV|K<%it7(+@$WHK5!8E?bnp4<|?wOOyBjJ-SvY<kfu7bjrsQnV7BII
z{^NCi0*%BC_@*~tk1yODU`So{M6u&QoH|Yi5&+<CgZ6$$0i0=%4__#qgw;x%{id_Q
z35s_&NqaxZ1!`b{Q_v^*A-%a{>!|%W2yEDe{fn$pk#?|U9(ri+dQ^m#&%t`zZE)CF
zYY;(Bk`e+o+0=Z8^CbKsYtNm1`r?FEt}4_5_naRuotsrNKpXqt3{R^I*unt7mjTUM
zW6HVXb+aejxKm8;|JzDnHO|BC1vvNryIJaXVY~|*WRFM34@ZMZjAmeOK-NHD(_<f1
zg=4td@4yh+s}|?JKI4LnrXNaDM|gmixlxlFXq|q8u6!kn)b-Ff@PMf6I2-ov*r0A~
z1RCeffO~L2fc-DXXJlsfSdo8IaeaLarnzi^R&}B6&`@9B2a&xh&gRZf(FraHdD+?i
z4~TqVb1(@p^8f|MxSg86VIEx1*tP$^neaXnw2#Zh*~fnK?M*s?UUA@Ke0!dms;6&h
z<=@PJ|C-(Ua#*kHOsCFI2e{e)Z?)Qp6Y>(xCGA*yf?W1q-K;Qj?)0>?t5`k$Z_Vjm
z|3dka_Ha4<f7@j*uf-s;lY25`Sj<ANS*0t`N4dg11b~FRvR_fV+ZORNO}%9s4}E<9
zx8V)E8T<49uGTc+7^VNM+iazOfaw3Tny&{<Jpb=vP$XEVBy}$JU}T%sR#7#%vpKY3
z&MPJKJJ$AT=`U$V<v0{FI_Voc6nvw%^~}UCdhj8TNPH|!X;nWQZGn-q`wTDu^Jo35
zzrDTa9%uSOPoV#{+wj655qe>HCa(x1zwAewNK_K-+tANZqQLQR2RFg<9zvT;jK<PR
zl(@`JcTE9v$h+sZ{Ftvi38PgFh*}wSi6>bpP*4bFiiIc&-Q~Qky?$#-Qws1_7gJ=S
zGyPl06lW(>zr7!O$^R)j_@X}f<+qR-6k<$jH2-5NKK<%1_xd{VI85KzvOgQ(Fld}r
z<A0{B{5dR}Io=LfH7_QqJ6Ci1Ghkr9Dw<T+hCmX@I#Ishqx|vDC%qPU+{y?<2EgMi
z{+A=98~Lj*;<#%66F1elkUx^Ti&WW*lhb$H7W>sfaO8k&U2vj4EkQGZM3Zbn)XueX
zA*dRrku3L2WgzH&6N}Q}#DR@EjOw{u3Cf!ohSJRrfPjLhWfY%SS!X})Iq}eUj$;o2
z^a}eAO&n;A+*A-$Vfc3rTy}NoZCjF`6q6b{lLxWK9)Cop%NwGm@Cn6a)G5zrMUo80
zEG-QMT=j?QZRqg{8&gRnl65;T9_r2J)1t#G=UygDgZz@g9kSNJ?@MCSe^&MF<?mx-
z`M+170g%~=Uw?iv9LqL{+REx$4>!8i+COtIsMU?99V3?@`$Tr`uOWw$G^l|quQN^O
zmcC_GR&*NhIS=c9xrnN01Tlr`TD7nEv~<)3lr{s{EUYWPg#9|m9qB0he&8XRtNL3*
zmOuyPiOv%=K5jI~TlKdL(lQ?zAYd#2{S70as1auV>P6JxC=MNzh9Gz#{nJY(1_m!&
z;!$l*zvD&*#+M)CpYG&5A;6+YBO*w`jG7qz85<R_ZDFn>P$~bMydD*^{8Mt^Q<5m^
zf{Ms6G72VEvdl<)NH9UkGCA}!Pc(KV0|P_l2ZpD`=B%GSl|nGT1)~Q`%UauRggztv
z1^h36bz1tDY8m5;h<FrpnRDylsMlN;CJ@YW4GM7lv!*~Gqp{^LzggRwmqz_VF?>sa
z3L&G%N2L=67o8VM?XgU`+P`mc)WL0Ngmi51J6n{`_|%QwOpuohzk(xWVHuLnA7bBZ
zm@~b^{OU?ThDq%|zCuq>DB=YZ`3=AteY|KQ(ZhLTzC?dZ6oKbz*%^H(#Z->kAx?}m
zni33SN|4rJmM2dML}!L5_x8~pCjEV*_=TUZQ@Aw$fBf}^Cy@r}B5+JB!|+K+a3m(t
zk6G!{0?jTcPG<=Xx-4H+<tP0JFdXLjo>YmJnSw2^HJohT+Wcwd#%dfskeP^IqoTnl
zF#F?)^#4QCS;sZ`hJAY*HOY<c-!^ap0@4i<qf1H!1j*4QB?4~rNNH&S>5`D{1{o!t
zf*>HNq=4Yd^S<wM|9Ss=o%j8@&g=S~$3e&mLJ|-T?yv&&QA$Dx80XeB17SR44N{~I
zX^+b<g7bhbm;GMmfkD*F&tu@>608)vEs0zfMS6YYgoe|7PrCY&CD>?)2{iQ|4vP`U
zb82}}gn}h}e6W!vQNbCd%vex02t=I=%&30{x8gA>X1>AvPdD-;2!IMjVJ8-41QsN5
zfKgif8az?P^?@Jux9B~z7%SzKQ|8@I)R=2oe{jy*?DemrG>KRUMLan1WNsP&uPHLg
zB%zpqBt@!Hkat6X;GoDfIzZZlv<iX{=6ExtCNSHmM8k_J=MJ4fMG|XN#Nh(QDD_4~
z91S-x5KfWa*x0jmcSr^??Jgw4q5?3f1wde#3Em9iDlO4g`9L)@ASZ`&Ynn|86V`EW
zQ%8gfKxzY)54V}uAeA>1$ojvl@Vh<l2L}NFhEnv2OJ~wH`knv!N1la9+0(hF@7Nx*
z+eM8QHn&u92z>+rpaQu}Fa^y+KBtT_OeqS15*CpceL__I*)vV;0ViS`+?6&|gUI9h
z3{)Zpl;u0x(`G`n)1Md_jRI2eF{5NSM1F8Tqts58r-htK0+;<-_NP5YrHNb{LJ)no
zq3Ad?KAtH(9%?irl0rmm0Uk~a)#94Hs=OC(uo2Wvo2bApdC92184Co_pxQBiYW`o>
z=fTYVYtoQ{*CBJd7O9w@t!b1{kA%%Uf9BmemuVR-_?fhtyF|)Io{(6cuyXiTWHOT*
zz=4d8OXmSV2p~ej&;}iVqmm58A_cZSlR!imDq1gi!OR4$eMO4wQBepEErOrh7YRy@
zujsgjeVczi2O+dN^z_DnoO5@vKs~K~1XS7wzcw%EQ9HUfA3n9{(FIt#1O5G_R8zyE
za5j~U95(U-DD8JpA%Pldn%~OVVHy9w{AX3FIAhtNz$E2c1quPGkL<$290&jo;cceL
zOMRw`TWyLRA_U0WbGRXwD4ZOiz^-Y0VyguAi8WOTrYOrXoZF^<@fEcp$_R>$2Tw|>
z;z(gwATNq;5=qGOiWqQjUt-)7^m0qNUx`dnCB}IZJNl7CWWqlG?l+ef#MQr;#>9n1
zeD>k89~B@b4i{^fo|=adIYe8<lb-_8e8f@7I<ye!?dASIn#=!g&gXjn=u8^=3q`<#
zjo)DUrw=(p^y1n6H%u^KtiM)aoI&{c;58pGO}l_Ajak|?dkrn0^w+Vdv|D1ex8QHm
z(6Ud4!k(6H=Tt%I0V?$1reNw42=Hj_TS~owx27~$%Iq~H+uPj+Qk%+L{?|a|s{PGm
zE#!Wd(dg?{x5^>s@29ou#F;iv*tRhXelyXD9RqlrI)P(~`1bPGuVdzSE{n1Kg*pZR
zDmVk7@8;RwSYtld+xxAZq011iA}DGwdaQ-g|F`r7NG@+^!CJk(5CO?oGv7o&0fySB
zEYHfKqKy3SRxzdOkz|G`=?28Yo?}o0vzgUvu3%IdX{0D55^}dy@YFE^zMnvXh%xB^
zc^sS=1rk6wW*|pu89AWQU<5=kE_QmXRZ3*lj^}%tfq3MQ;esSUzK_w`1fIoXG*YK!
zsLz0%An(%B0y>Z9$lmS0Cj7@Y0rH0FuLC`;@Ax8yMn4Bkpn9SPdtM_t#jLSBgXH0)
z;c6=6U>g-2-nFg4ea~6fM*JS;l`i`HZf~f*WH%XNnLpep{HXt;zC$^nNwX+B7LCTo
zptyha<MsP3${ceutR4d68+0&)LLvf+oCqXy<Df~{>t!bcx_d<yj!1OhfOFX}X!VWT
z42pIXHw?8k4#(v?qLWj_9yIa^BFe?F5DqnRZ?k7cvK0vR3Kp_28km99wthYB#By4N
ze$FyWo8YHbMN`H}7rA0$43kaK*0U_vDL@EqcQOE%?O74thfnWJYdnxU8~-=Hd2I@?
z*s!1Cd=|-dx%BgNX*!Z23<qGsT9Rc$wb-2#QtiKlpbo@999TH0PM4@iHBl`MuY)RZ
zLFq0<`ume+3zCgva*~(yJE%~f>~2@a{&d44TMmBO{rHyps{3?Gd`9~_ehgoIR@DCR
z%Kz(#%@IU!+!IZ#sB&uT^D8$;rd~+8@kz2UP6b@XtjJ11^6z|D*YM-5{X9)1!ep)+
zI>>0CjGKQ|?wFye{<mv>f2m5%nZ(3MIaG4;VFMgeyv@#u+12fVg?8Zu*nf+B3LYxh
zogVd(OY{3~_?m}vph;A>Try5sjrsx2d88oPIF;Q{Ku}ot=mn{i$5>EMRJGehA5Y>X
z?-`c^PST%9^OZ<OgD65^qw(-|>#n6_U|@)DL!kz%rhaCbm>p0>+;N2^Sc~3s<i})f
zUw)4ZYpH6-cv^7sggb|-#c8JxNs1@nOlKXU3}h(K=q)lDSw+(?p)`xHmHRih+de~6
zal`pg>lg&jq9X#0MwZ2UiLyMTtSkV6IJJt%bQF#)aHSVGY)MgWuwLSsxnU41FoU@y
zul|J-aalMLpoZ_OK!adtQXGDt)ozv5zd&az3$wwfB+M+TO-Q7Ow(?>Uw9WDx6LHk8
z2VU}et_lOGSuDykXyVRJHno(6AqIl9O4GvlLu=qfOEyqphA|<1vWAAkXZ1Sz>rP{b
z$9%q+4%qkaF*3KNt>*VQz;3tFG;2h@uPn(ZnNwZb9&h7A{2&LT*p*%o(0yn|-M%r2
z39WpD5}}5cBtQ%7Nnr+}-y!*BK|!HNF*-p|Mv0KgdTC=beVLx#)<=y<?vZq0mZv(0
zW&t?>9!krO$BW4k-$fm)1fqE=(d|SyiS-|m#H0Yhf#k%~-<)dH48oqqsorM2^fNZ)
z7MXuDccf<$^(|zEmHYfZy!9Eo|4J7lIrv!qYiXNiG4o7awBVbg{zcoKK6L;NXMMhD
zNHowxmu4vXgF&yZsiCE%p(Pv}XKZD&W&hO64#hd}e6IdtbvocM^N#(soRyv~Xr#z#
zB`Bcj#;}I}AvPJ&aiK2s+XmsvO>fA@=ELT8y2L%>syB6<#mI?#oXz4uHphcG1_-K5
zoo7`BA^ix{Xu3ZnU;BJ><8cJUpR*lkst^K1IZ#$QXppdGafKG|aOqQN5)lIjJLU|S
zV1l(xDx^<~98m%S(fB`#OK0qk7YKhcR}Yjp-$@u|=s4Y!^(SV?;EK|T2^^Dq0I?xP
zMCO7u(#(rW?uWEyM3!)IAu!A=B^gEcdtEl$?)Qe*cQ&kqrd~bza(@q*T6_A*nsTF`
zxH5m*adzX!`%_<eP9uWp;SU2}5Y9%UhDoWuQZY|aS=MZOIqIsK9NAv+!{!~?kb`Ll
z?@LeZ?^F4<tpB7fLzCr<SM5e(T;|UoxXh3SOsFD9_7OzmeT%PrZe1`xwBCH?+ZLO7
zo1d=Z)fkC>bAI^1R;O6dup6&f5ts_J2Zx5P{%{l=V6-mTyF&w|`!SQhxmgW=(Ad!O
zeL`X!A~5nwZ~}?t#z8G(s6UoPdGhy5qlC3oX+{%+D~g(w^XL=q6}lemhkj4$`JSbT
zuPQs(GU0^YX~EG+=}H*faQ2Rg-3wFnqF}*GgDunzjkjo_?toOvH#t{}$Rg#$WesUY
z6d{3PhO@0A4H!y}QdYLme2FyZm$WO(yNCFm7BkPQL8vUwdKz7uweJR5OffQ=+25~7
zOIK(450|ztX8Is9mSUMfm}NXfaQtJzyhqV*6;h~VU%VI<APtlHZ#GD}{WCveUl*nF
z5y}P8(<f;^S@ko7TOB37u;xru$p^z;Z-aRW%gcEtTT8><&?*54i6BtIp_ApIX{|nR
zyw-3GfK{;W(Frxs8Z4MT9$16}^rVY#J%46}wD2YeMs`mK3IVaWs5E80dYN%O_-r@f
z7Mg+VW!m!+Ou;>i9a^UOk0*Dw4%`d6&!V4&9l*?phL6O0oG7AgMlEg5OH5|bhKz76
zbL?n$92oks4?>4ERB-VU+&%WO+qa?euOe~QT1ee|*vPUv{JJrE{gi*TzL+qP+`iCP
z3a8T4pS8kn%&t2-<*(BOc=RAE+tT#Qh0NKb)3twZG14R`a&2pCGv0wGP~^c<Yp|JX
z5umoArL84kS!7l_tgK8=5zkt#7;lqU!#n=j0WYF*wi*V30HM^AQ$Go!4mu;Xk|;Av
z#0qs!?E$d-VS4yYriASHiO7NpJL`B7Vw~n75ZVxzT~J(qlf5ID)RKrzfBtW3V_2U`
z-R=U@VHB7}27kS6QeQJu)+I>-aCiqR!=_#02Z*`oy!XkIOg8czzM5ruQinli`XoU!
zEoF&%_cf}yjJNJ}`!jr$@JUKBn)6PzRWH}hu-&`QB7BW9TPouUa5FgR7G=ghwQUrB
z8&OHN_CjM`^3)%NQcRj<8^#-_-i2gB;>$1@5Ne2$)DG>ySeX|aNblGa*(wA5SQx2V
zNfeIlimzZ76^vreOT%$-I$4rB*(fU^k#4~Ndq+LPBzo-_)DY#AtOPNPIaqB)#2dEe
z>Ps(88zMDE!3N2-vr7hYKQZ?SFxTiRSoBQYPCv5Y_gpso2u@KEgo#3GARo&FF>uYI
zeCzvZtdOL<Ad{%{XIjGU`y=(ykP^U}j??0X0(@X%XW@&*_4ZVbtmX)hi=L92hhAl|
zcm9fQ`0MXI!3A~x5n!AQw<Mn=W4XV;(|jnuW374oY^N^j@uW5XaqK?<0Q8<b`PaNO
zh~lvkDH+!|9v>3v@WNQql6nMzc_*05lv}1Iv;cS9lcy29ms^QXdm;(o{2{9ylh2h=
zq;3;|id1R5v~Hjlj4_&?UgIQI1V?5+@b6~a=u08BP6?>3^qA(^r=fehcPpwVfCz`^
z`+up`LKa2xf9uGq@g0*?#me{LwK&T{t@VIl>x0mAYaISXL=j)-(8hC2Ln-Q&oUuw-
z3xBjs*oEyFksb{k&8(X9nSyx#;`;jhpnMm`Lq4omkt6|L`o4ta6ZScNV>!ZpK@w6o
z`-D1wteI<=ekN*$cbt_FH_3?<=Y+zA%Z&H_=)M(1S;;4hEh8^2DB{T{zVhBz1ck)E
z)k#m%cg3Q#@Mu!v&}B>+e`QcQ4wt~0Z4Ht~6WfS_nLma@76ise@C<?_#PYDg)gNp5
zEiz3NJ6>PfmSRO2veS#$p(bOStkSB~ocban%0hSfd9=J%H<_bVE>2!62LH1nMeRDh
zBr=6p(0T7u2OtYda7VEB&(hKk3Yub105H<xr`0IH3j-ZY7?h!?B9j4YCKHHsItH~*
zG34AvB*uX@a#8G?eU<~{0%X*aEj5!TIxWDoPi^9uO@Ej)*SNgHyg_IqFVa1~3?M5`
zbF(Sm1Xcb_qDEpKGmIflMQC&~!|kI_gZV@k3S0vwULj3a59l*JT?QJrvR=0t&<vOu
z?q)cb-=_$|M8dF~7a2e}BG>rtu<TL5Z5{eMYlv}-UPinh!wfAzB;9IpDI0(V^iSiF
zhB=G$)|{`5Q$@9thC~MJ(b_T17sW*X=CL^@qBTrNEA6tt{OZ0fbn0tK?aCsDyKMa}
z0SP6oa*fer?C;rMT29g+dB^RiCR(=0FJmV%=~Wn{*bM|Et%4eu8YK{opfwx_iHnmo
z4#EvdKAlnzNs`yN4wo6<EkbWVI1mWI;Pkw*KE}Jm!2aJhPfCDo+liWRkppwuQ2%hk
z<?j{=ss`->Rr00eA(u|-go{<_Pu=8QhQNe0R%svd0Y}zRNDKtaIHX@)PyoXa5EzDW
zVI1>aWZp!HJQG&i)`|RTffcNid_b(AMM2$|=;;W`AYbQprAPz=d??{%%<@9_q+V?S
z@XWACQZ*F+xaM|sw_4xIVGNV{XJntV1W$Au@9~bL7{|mD>DeNMCnzBsqNwnw&@qox
zyVZvX0cI0_kyLD2dK#@(-`7frb(G^jOffOT_Lf(^3Yg!61|(04DPa8Q8KV|ebil+<
zFKSN^w5Z+qSx79>Ml0?iDo!xC-idqk__HF)>T84{-A^yAp%9a##$66my3~l!x=r8<
z>`>>kzel#+HD0=3U-wSe)@hj2j*L7|BtJRvL~B^UK%~88dbEV(-d<xf-hn1>F{(;^
zeH%2!DPDM+3$>@-YV3waM}glbY<}j<8zrWo5f6~WUNTdTAGWBW_f387?xD@RjZ$qt
zPI57wK4>LPnq{T6&YE}BFGq3GeQ*D^d#(PN5}>T7#Vh0R{eG&Ots`yC)8D!WT2;Zb
z3lH=2<wm|bT5d5a?dFaLP8p`TKE>QEk~pg&z5=ZkZ<}6nS68+<xJ+b0W1iLOWvn^I
z!GSwe%0W#_!lO6)o5n{w<_UIq*3WnKbvSkt4y`>7Xf-mt?9Sfvwu;UaMWCl7N=?-N
z)a3!{qYdb(-j7-H5P@br0J(M-qe-2vxjxz+6>QjVUx^8eXT$-A<v-#uxFRuxup||C
zEsuyEoi&~gw;<_-*PlsY3D=Wlo>61+RFFH7+ioM|u<&FQh@w^vL1r*ANpci(mXu)i
z;Eu0;2`BodQwIw%1?N#G@c^w5)*=RKc}gutP9iR`z6`?5Ius~EVO4oYH54LO-ik14
zM<83CW3a3U%bfUz2#w+1WMZBayCRLe=c&tm3%Z+p+%qRp^QaM9P%0nd99sX}lY<<7
zIyIW!9ZY3elwAt@@G#Jyc=7i8*ea#lr=M*U($DG|i*&U}x4G>SJ(pNQ_+*(M<cvGD
zQxN8w24?0nD;=_xNQ~oXt+lSrl@K-`@}eTS5O21vaKjQwe+w~5$ilo&Wu#Ey5AVtn
z%ILG>p3>C|E6@z<ycjg?H8TlzYJ5WpNL)c`r}2&;N&D{Z(KNE2&kSysh%~HomR5&R
z@I{7lrZ<>)-}!@$lKPSApN5+_XP?dt;~r4)0oUfG3J`t#Oc`0e#;IYDM0?9ET4pLR
zTx878N>{G-#0H~cVIKt+R3R`lLTj)PLnmAAjDOKI3rQw~?owZNUi`^<Hqoo<r0%*j
zd7E)dHv>_kJ|DZ*Dc;C_hq!tB*d)8oFM_lre33X4B3SF$#e-$+7eP?hi5qgl)GT+1
zEl5fGx+q@^<XV(CBu_O{DSrH|E!9`{RcQM4oh^a1kOR|=1E&#_7#fi4Yz3NA?Uhlg
z<#C;~an<H;jKnb|F<vnyu)?Vo&GJr$#^T$?*+3kCSS+g$Ksw2bG@TrfJ}-UFZdlPi
z`*r$jvEHI5@j=idwpNQrha5pVbjmZk;B;$FyAomr7>Z0VN-swkmTt*rWK>A`Gc@>O
zl~!4s5R-uffb!Eg9;}LliP0pum}xqw7aK7s4jubE)7i*vEI2EmI3nB@nS8-@gZM8X
zK8inWoqT@wIe~;`{EO-q&mS_<fr={C%G%P;hD~1|bVrF2YF2r`Ky&PZjg$7dMR9B#
zn)bb{ERl{|Fd`y*gSir7t0KpI<ENZT&BiR|wwTnh>(85H==2jDm-Y~@{Q@4kDNE*^
zw54S9EOJ87p<_%`n*cB-NprSJhElx~#n(6}lL7wlY9_SC5Y7+05Ty2_vW^t;Xo=(K
z;QYCW$gmy5TamB&Cp^vfKgulX_K+|ZYAz7}WNqDEModTcz5SiG1)65Kj`zeoU%pjz
z05fSHiYrO;dpX#JZ+;kG&1^)z98)GVLK322oSF;yh`m#$F-0;#H7i6-T2`J~mk@u{
zF)mY1*=c`2Pu$Ohg%8Z?W3I^eP^t?Jk8c$$%gqdx?@HTDTw1!$X|AgIQgX1WjHixS
zR%#4;568Q}!U#^UC7+=kB7$nGMzlF%^#3H>BiH)c{CRW8mIV3JoGF@G`(yh36{baN
zCADPNCk8~qbiy>s`tDj<`!8n8h?o_JgvU2)c^{nX<wYhkwX_Yj2zPJ=W$`>QJeADL
zb0nvDM$}A>t;-)u(@VLD54_75+Ih$5X=~GVN%9^q+v?n}pXfu065rD^q=?Th#Qf5l
z`912&K)1F@D4$Q{fCEepmH=3pBf9}~T{tWwQ~|Y%e9qS9tDd0#s320`D2RzXnd_Yb
zFvG^3;3Lvd9jb{s8b}x_%hzb=UzJ)&bff5Mf=Pe)r@PI)zF+Ka^%2AM<E6k7(NBLx
zE(QZKfYum=OA^=5nZlb{C6@yfW}+zV)P{9Vc#Me&5>*0ik%<;`ByG3V0dXnows?4H
zyon!($kD$F<ubVH%K`y@S~%*L)W{ZpPfWkhWd0SOzAs~91|QKsK^El<qptpy`X{k#
z*7C~kYh5g|EX&FWacoyPo_9Gyslj3<2u=trmzW4d4cooAH=%<S`>KKQcPR+_{B^>_
zOT4uud@Q3@NQk57Mpn4>g4~2v_=JHyAw6NHK;E2FX!*lZogI37pZso+&G+ln=D@wl
z&D)a#K!`Eek|1C$*!<^V>k9v4hNrcdtF{WjhQ_Yj;!1XOsH#q3+IFfSWpuoIP@;Q$
zxE)g$-u|ZxLU^zJzjuQe8~~`OlApRgvY$#TdY{(IZp$yDDo!=nG4GJ<J>2Z4{zPzC
z_bWf^zpc$UT$m>J++b#~(Cy9-Rq;W)|76Nr^So#jc&NEJ6^#je0ajc@vO^&nKfqSN
z0$yPbLfD(o=JHZFZIK4+)}bXZFyU4i4YVMeASA0Z`f_sX;VF(0;IyLuF0}9S?AbHn
zA&H+7I^Vm2_2kzpaO__7qwBzx|8{R-*BtSLHH4IdageImo+ox627lh_?CPW`rHnlB
z9V3kB?MNpMZk7fRY4E9$W63IrfO%WH>tre7#*{T*KdeGqYS3Wm3m+-z$CfHPD;bv+
z-8~x;DJBF%qD4#?-3yh?ptxLwRhPk>pKCiyBSe861vhjdlNmLV<@GAH0!HP<8qk3l
zy!va&-npZSz=P4@??p-151s+4C;q0q4`p8WP6*L%O!`@$x)PyrznvcIraWyFVp(eS
z7^J^y!^<0n%H<1qb1?1}&nmVInA>(&a&E}r1i<+F8w)VPM1Wic2_h3U-^(i*_VQ&`
z<`#t~*gq!km&o_dBXf8(7$zpletvd#2mS#7Y(%sll6XjJD;^U?1kuUs_8~v6EE}2V
zKeyb|gdqQY*pgQGB{ImueJd2Oig&Fr0ke?FafIt@99q_0S3fNf-%fbyuatD0@^=-{
zYGn1ZVA&aSh#Xw<*WE$3kDYsOlep|6P|0)_!qM(FDhSSTa~k=f0?w6~?yXSg*5h7*
z&$C}2+@I_jO;_{(_Wt{<u$-<zPRaC7ua@TKNHBPapHFRI_=f?(KXzVpm3KTMV#DKQ
z#O1}?qnqAetNx7<y+L-LsP#4e1gCMp(g<L$n_7%9j;5p@3iJY<G0q&+UnxcGPDl`B
z0E9X>iVQz*)J@R?MGVU-LP8flrBrxV^}2S>F~M6hl}H_?^NuLgcim-JKZV~Y>|%%H
zV>q$k_`h@V(8MT2kTLa|yi($wacn<gZ==Hj<KbJp2&yBXdzL)JCPQ6m2Dt3{@l4HU
zq7{Rgv)V1IQxnWC_UH)|B)Hfyo8e*|7!Eg+y?e92UHa^|z#z~1e9yS2TYH2bME}<^
z{#gsV11N6YGE|;W4YW2d9l`+~b9t<m-`*5-?DTj<ob9)7&Z4i%Vdv6$XYa;IT<L(X
zYy3*zR_<2aI>XDhHq7ti$2WD4AzjYKei&2sW48ZJ(TL8_CH7v|`52R`&kYVX!XJjf
zvP%*K3|b+fNv5{z@;C*Sp&W&q)EPm~me{79XBtD4kEEW_YzEEjxbefScYD_b;;bvO
z|NHsuk9SyB{hQBRG|q4H&+bkbU2lnqxVk;U$!fRmpLRBH;_X)u));z^I*agnl3$ow
z*NXMb{}!uyw+d&EKy|O<c>osteQfw1Tgxm7sP>eX1gpRga+=Uq_ZjZ9U^0Rq-*UU(
z6}GfE^i^Cyo$+r1T=t^*W|tWGm!SYbO{iL|Dz~nyo14!McZpOJNuD|yvQWEmT7aK@
zNtn0MUi+gsW6~K7D3%!7x3f-&UB@KwoKcktDWLR8qj1xPQ^pkSa7H*a11Jw1>u!@A
z1?CMK0cgYZYnlx$E!>##-t+%mM3mPq8I%H&K^Ns;+m&8%DmNMO+^>hQ;!uvIbK-Xs
zf)P^^ceq6d&M?Uau~i&h@{(fRbj8JUlc4_$$UFTar8DnFlq<iYAAK*fz=SWZbVDq&
zSRsdpV;6Q*gu)YBw5y-RLp=8!y8)znQKytE6{0kjSi+7W-X)IMNM>mP8bE^bpSR)_
z@3^{-X#-WYkH5WC`p1>>c91{`KotY?Ort0!?z}uGf9rY&|9i4?{ZVAEJYj`i4cFDw
zMOjx3L%c>YBBTgm8R^Ob82uwQh=D48If0A60PTKYVbT=Y#>!TS?->nu<sz*}c%>-i
z(A_vnZ@rsC@25#A2EK1={?)1~!Q0twJoZs~@=|oPU)X)`o%Xv_so!C*fb$v^uh}w{
zxg$Z77$6ElGphRHuQ7}<wA4zb?^w#Iat1%G-Gt&jgunA)4EM{y7)$gl$*7lBo|%wR
z3mc%Y%|Koz#P6RXa{lMe8OWxg;Zzpql&%~9=a+PZCO`{-JE6HosoR=D4NlL+Ej-LR
zM6;p2*2TT%O!Jd1plnd)qI8sQxCgI($L>J61B}xM)~Pu&R2o-X>y5QXWee`?K!et{
zK%-v(6`a7{gd_&6Trj;`mcE??21!_O*Vi{+td92by?{G&g*B(@W2;*q0@RGe<vlE@
zl)g*tpGTAjEwmgv^D33L*0Pk`1rOxStu7C!AvMWT`*(%0F;m}%(9o@)y>WkiY~wpW
z68%X51;l6LXY!<<=K`4%IY6y;SwW11EcGHO{ExitF&(y%KB`>kQMPtfHz@RQ)0nNx
zo<hdUae@xl4U)-y0=TlNQaD_PSiF_ndlD^GlHfOOqQwW2fgo)Coabje_6sfe5Md(0
zf*<TiXtxv@0y1U#_@4WR=Dn0@A>;kEbGCd7h5f>Aws{Q7^~iP%e+!r*ekXSUXUE)x
z(0}=bS?YT8@_rWkH9L&Tm&DV7c#73B?m3`q61bz?`PL8kND%QnXL_!Jz2bo9M1}_b
zy0ldhwRPxPDeg@roJ%tFoto)=&-+46Q*BJscF>Q3$ObpTTePltkxbP@fXVsjYq*uw
z?|a2BdQ_`xJ!Um8u&6?&*!p@Nc*YOitwLNG#`U*9nkRE3GUcQWD7~aw%%hZ}Y2H=&
zy1M>5zbYkAi<wNdE;9-T*@8^U?6<l4mEdhspVPBlkoekJp}4}@of-;$<h1@2xgfj~
zj5szY2v}bp!Ct(*{-Vx+O6Kwy8^Z}uL=16xv#dn?vAF(EVkL-Cp=LE5C=cn73@wCX
z*n^t)No+{L+^n?BRqLVCS~LETymJS34vQSqOw(r}@|++<?8NMWjNeg}gAxLPH_|Qz
zgBW2?6o4`AD3}ex280wi1(%A(EXg>5gJATC;q76Vvd6B(qJwiCaIT$P^%>iCvoDyC
z!VBg}<;&Age@ch2BNPHDP2>koU{<=Od(n+ijQSb9HUcEX8Ze?~ydlJAFs|#k#0x#L
zCui>cu6=Yk0&sd5q4MheviBbQTK3@|FZ8W5><IO3jrK^>TGHC<+2t*Q&;WY!!?h$8
zSQrTn%xZ6am9L8pH^I@u1AkQ6>WXrnNI1Q^a?g2C8Zj%ip`WL{^Y?2JruWUY+|}8n
z%VoRbpCgvF&THbMKQ9Vt{=D@k26O@R2TQM~EjmuFJ(u1S<acun9J9!6RkpT^5H75y
zZyj%4q2m*M=<h|V&E=q(og-((<E1V-htvCsIB>XHX)$fgRd`Hv-8C(M`U|eUj(e8c
z)iqu4QAM2N2_eht+ICqZ%<!?6_>tyA5<QfOlll*Qz&G(k1TXH15Xh0w^v0rfkN=JU
z65%3*tYh?#JmCGsU_*3IyzM?hjh#n@%E&l>p9z^i+nPSB-^(Y{a@j?O4aXnmmgy4%
z+(S+-17z44YR+V)H)A<7FnL=ytraP4NeY01AEZ=`V7q)0!tGkjE!{83;-}qsNm~LA
z{ZQHTpy)hRauh!icqJB$SYDpamcg_3sHt(3_>%0K`OmrP425bb916zLT;_9AL2+x4
zt)e_<Ndg{-L>ouu-~Wvg07iRZ@sVj0l^3h^nM@X804jZw33srCFO0;HoQKG4YoQ1&
z`=qRz9CNrMj|wY++>7tZuW7Z9g<8;mhd}Su3B8PfYr!RA><ktTY&uEgnSIXOiC&Pm
z9qz90hC`Ou-DQuWz!IFjcCJNHI8K5m9=S158r*I?HJx|e9*!EV5;PC(F;-*fXQje6
zKfIS|{Z^@ropMrvPo}G9CYRJu!F^8)Ye|p5C<iiSu>f6*=nEx$>?nI^Ub2nZ5%JvL
z0(W+FoIlA|_Hk|Zu$U|z<h$*!e45^fPi=7aI0-Y;+~H7mf_8NO_X4#Hb`y?vljlEO
z+zWaq|3#HqX{ojGKl_E=jb*D9Ve>X$z9kS-wWL$yRC6QK+spoygM%v9wVy1-=jefI
z%uU8C^gI3@tNK>rkzgGOz(<_DQ$i)pL)@f^zW}E1{Y|-puB=ff)FUMn@3NL}pXJk>
zEHv)9IA!us_<<5KWQQDN<&qs%AC(#8^CCGwT<@A<z>m`@M)O~_b%Dy%S5GQH>xyfR
z9U|)W(7y(Y$M1x%{ZDjR<PDv&i_^ozF`I#<0t8n?)(>W+7@n{rp5i|6XLN4K2^-9o
z;F!1#D-SiI7R!0M7nQrYFT}Jq9LkF}17ZUVUj>|}8oLf3UOW%_Ws=LNzV|l1#V?D>
z?cL^;YQXEudSBQdkb04x4;6uMT~X+kW7%j79zpBHf>aDjbL_42aWAk?sjWJZ00zIm
zQ3bJLkQt#(@g*(BnNKJt2ng<>zEBe4)b#!5WaCR_*WD%Ps|>$XFHzN7Rl=!j>T|6_
zNr>SP&&m`kk_OYfP_@9<XRjK+z5FIe)V1W0;a^upk2I=sb4zp28EZA6tbYm+IykX!
z4`YH6;iSq)fKh5pyRGEuiLL-zbT`0qKGq7ei?v}7vH&CjTa>$c*oECAQeB8B+2=*^
z-F(|+rjRCJIr@h(2SWNao5V63B~hI8ge&i>j00ST*|9eypAYi9C&aL1-Vut!W<r+s
zh%Xh&+;1!sY6@PTUa@ozI4yOCC7-u21>VFa=*~Hr%H)5OxymJx&9yxt$)223$ieh~
z@rOOP5YEW__APd!w6$0F?d^~M{{5a(0P}sw8r-8+Sj5xt*2wf<krvun?N<rWGL>;K
zN<3-G93=)Q($eDuMh#5v_t4S+(T9~B2wU6d+9<|SH3fvHW=ybgi({&(k*G89lf9AP
zQrdg87&!p~okjLARY-dZ5nCHIP=L84BjgUo0bzv&?>y=diPtZ@7)&%tswd`Idp}mT
zN@Wk2?0xz1d`JFaeh(B1?tR;s>;tR|&7AG5-F%TVE6p&B=BE;WDRTOV&U?#mRnv^_
z3wwOktVt*boSG<4;a8U>5E8lxHiY#clCc0;gw9kNb6!?DDKF#?f}^A57p9R&<fxW@
zyi@Ltho7ek?`FsA*E?bhn|<GeZ;j3S=)M~hlFEYab#+rP==ntd>dOLWcST}@6V0fx
zTKGahZHDet#mdAJT)7(b-br#US0O<0sq@SXEMD2iKRT^$kPgUwTuYZ{D4;^()jM7G
zk`t^yNj^kp)}7J=?W#2XfPq8s6BOiTnHMSuNfPDM@`JT-uw~ha&D27%o~f$Jkh$l<
z)WS<4A$^wdbznlh1O%;ULHH$}9taG@-3?91*jDG(c$3E?Y)g*x6H=N(!r%%L@)f+!
zUfy6>J*x*}JAox6kHm+}bSuZ^iilo-Sq;nLMFN$-aL#%8Tm^t($C54v`MA42@eQua
zukZD+NCnGt<|}iD@zaT%JT>OHIw<J4S97Vux$UnY`>nQHyz_U$3dXbU=xhj%8dzfl
zzI=OzFlv3-vE%z`t9+^a8(jGtERI58;pk%g!{hPi`Vy03-vlT!rJ<$gBJS0}RQGaz
zNv(aSri^-yBy_39s32CDh%TC4=11-)(l~Vsj|Qev9K#|k0a%ecYdU9F;N2sY{p$K)
zO96#)bQ#-(`N_R6&iZD)MdMI=Qte=Cd9|n%|33Oh9w>RY+wX6fgs&a?`o?^;fxSyW
zS5)c1L<4H|OJe_g<HliwSJv-%^T)`0KVTT+J%a~v%8Qu!y=u>yF6~q{i}Z1}!`0%;
zn|F@`ZvS3J1yDQJPrPOol+ifu2&`OS>tgQyg#33rGh*g~y!{^0zg8S`CFM!{{!hUF
zbgOD9e8Cp^JzO!MzMMHf_y=6FJ_)DTNwxe@&LSVaRiA<a!H7%bh(w{j?}K44C|eKF
zJ9)BJlNd>er6s+E!CBR6ERC#+RRjQpptaJtQ*Qn`%yY>Y6ec<K_&S>~RKcMbxz_H<
z1mOZ8mi;+G1)R2PR#whh{bfI-!Mi(8ZrUSsw}m-Y?!MyUYk{L8qmkdc*qx2bfR0vA
z%OKdyj;t0ixBzg${yGQg9<T3n-Th6uXdbYP2?i3}LriBeruBCjt+2{S$l!8)mB^Wi
zC#tqeRgc*5WZRx9@;x3gQdn?Ph$%7w3r1rB>+X)#;;eRA&Mi`0tmjQ)x}lxu!S_V!
z^vDafq}r9ZrPN;yU+pWF4;phWx764V+?Ng-lQ0cu+K`DK^loi-3U42U9FA~cKKo%3
zeMcu*Uv@|-V@>TNUfzBm(fzQBr2n-Q)Nc~8eEa6I*QeG&v6kU<ch~WtIszwf(N_T+
zGIHW}=>IA*RfcFTT`sWIN+yQNveP<<fu#{*k(ruTe^vZDX{Ar<10AflHb4C|qrc{&
zvKDT<Ko6xK6MX(+-r~=FU+E|6K%5qnGn><)1$LkLGA`RV+e6!suKn<joHk8AVkjd8
zfS{%a;Oe&$QcO+5c@?D!C|=xUy8-SJtZ$A^WQ2~<3H83IwGerD9%tQQH>27MO{W7d
zDcQiQ)7#tIbtGmf`&V`z-t^h*j(9Y?DTt*;%vO%~%}ofD4a>etu=EcKB*5rvGlStN
zt57VEM~|)w5d+5(PUwG@WxA+GB$QFjbk{@#k-+J=*VY{mw{U>a7)4&KF50kC4aVw3
zqN+-6y-;88k{1<c^ae*J;6mC|YAhuqiVMUWK2a$QXeqt&>QFlKVXya=U-Ni!;C~=3
z-tvV>ZmsGdfM07<?t{(+=8?J9qnEpvTD16~m`5)=_S@kIn$al{U27CRuu5s=zi@CX
z=IprqQ713q_c*h5aj+6$YMH!2<CFyGNu@Vt{)R*NPu2Zhz`ayuSQs;wAcHucxr{`m
zgc(|@M409+?@WRC`+9k__#P-jK}@8C1lAN>UnmhmS;~{ge5qcK>2Ziz9W0hs@Jx`>
zJ}5Hho1&E$rVrr4RZfsV%~IJqxI_P?GJI?4K00Uo`>z26#!?7^idTzEl{v@QAHCEn
zf**+pAJ?b+wifdQ!3A|U(zuoQl+z+7vJFMkv>CWt12%}G-C8vWBFeFBfC3kyP}b6x
z`r;y61}MprduAU-T#O677rO}XeSY88y|CBLR#aTGfxJ>kVo9Mzn&bX=DnMoZ=cA@q
z{$DkJ@D|%V@5!L(<0(qzFPqIERf}$qi_mb}ou=_bl2j+lvvVoci0uuSc&;8@7_T#i
z=gmrKPaW^MMl|6^VfmJKIp6idkBo3u*pJzJ0$t$Cneu0dpK1*>7^j_I?vDbgPLb^C
zu4)94Bag^Hohp<|)ob?)p99~QJ^}I4zxjZEK5fyHx?aW~l#CZe@&t#CyEwkzdjjId
zJUHM3;b}d(i<r7(E`L<UPI)vg(fywF%38cE*mzF*8>Yrf$h=;559b*iPY=;6ZvuJ_
zz_!pkq=(=xT+9yrPWYi(GH0G|&C7hvI=Uc9e%wf6MrjZ?Dvbo&%u?O8ETb-3<l!qW
zxPHS}e=K3zP*j6uNkhZETZzPSS&J~ly(!Bpd)n)MPpKG-pMLrLcpv5=^ZITL!xrYg
zc#KqajOb%piT?QxPIR*6y<`Bv;9k4CQz93x4)!o-nLYZ<&g(BHbjkD-fH%+3t|u=v
z7brri3j*X*lf(MGRB~ernuC5;e_f|B9Oj>(SW+iwS5SH{#^tgQ{`GYA4t9k(H@`nn
z+h5$|sE))YZM`)x9+DNd<k21oyA8t0qy8Y|viwJ^H8aQd=94zhorQWl;M{F1jE6a>
z<mKWP8iTPceQ_f`$k10)8t6MlN0-&8<mS;uW<b-nMQKUl^2*A=#dU#0j?M$bm)rT-
zt-tfnrLHDpS>;jSBz-j~RUoEKi(X;-L~>m;;70D8OfPn{xML%VBIG46(A7S0b7H<f
zeB-03*jnD-&=IeNEWPGKa-aZ0AT{rk`h7|&&pCJEM)#9d3IgTbx8v6Y2N%zIR6-@I
zIT&n|wTNC&Dkk;|CF?f{Xizq*(uVg`2~!{Vq~LmI3|F$k48mm*U}}sU2uvaz0e%2b
zC4Mdw*{slHKp4W&tcNHCTCQ#rD?q6&fR#(jCLgv*YkFnv8^!iSoHEItks?ET3PM+c
z0RjdjR0^9S-6I(Kbqh+#KkOk8CcSQfU}$=^K#K1*y5A(3Ez=b8`enSZ*VjYf8g^Ls
z41n4Z)*{Z)I7zQ$!aUD?_=dO8Yfk|MiIa@*Z6%t;*kPa9O*8GT>|-MgXdP`|;mW~5
zh8yG{Uz>YaXQ;di$%6ulDP!FreU9~@)mwMWcIUf;KjJ}>-)ux02{D1|e8-2M+~nU+
z?rbbN*Urwqr3Qda7#ElXNQD3`ssK1Vv<?I$>T+c~r*j8k2*^N?G;3-N!VmZEbpvH%
z#H7J)7B7sC8FzEhSll{JWZaW)rd3-Ci9K>Pb=Np${#={BA7k0b6wks=M-{?p_B@U2
zfyUNggg>_^$M2>c-}YqP`CmyC&_#=UiiA4la<+!p#)f(P>$|yMp5(&A9^`twlc&WJ
z5QUm0IKRf+h-haLXx4r8zW~G0P{m-d<-kl*SVUP7n1pvo4whf}AoyW9i3e1UCZUa_
zab%=vTFm5wCX<>hE-f^DbjFP1SAZPmc=*AIjHyO#r8MSGOD7QiH<eDaUer9qRqQiT
z`@7T30!fQPh4}MB0wADA&>gt9o}X=B^A{>9{S-4AmD$RY1`HLb15?q3lj?VqPY6VG
z5WD+Q137bTXCKnc<>f;Jhyjq?crp$>C_CjfqN1|VZ+0hsv2VU|v?wZC!PonsN+DD)
zRFLa|`}4QOp<Jwlj4@oa?2@Jnc%<kPPfjB@?U|MX$7$+z)68ob)gu9c&_{ot^$>m%
zD8*1X`nBF={?Edy*C4}4<5TFXE{b5XJOMhBtVQ3q#fkbS^rhO66@aAW6E%HD0|BC*
zAJQW8U#cTJRS|serO)PAg4y{DBvyPu)wxr&j}DQU=In3(RFsn;s!vyOiZ~t^qKxm)
z(cbyZ@2ev4jW2lq+f`~bLnlYjpH=vI8Ae0{GV<@4P6$cby2$nB+Kauawu^o1$Ga;(
ztA6g#YlWB)A`G_r0l&@I<$9LM=1dG%2pLx1nUKPT-eUz0pkoLitnaeVJb%+Epfl@d
zkt#7m%_iCkW&s_aBq?Ej`Y!WVKl)m1@52#WIGn(}WumP$)YZa>Dz1Kn+u9s?jxm}P
zmppn;(%H@Ho)F-iFT1pNZ^4bJ>y^r%F5{S#!reXc3PaMe!VH#*?=q$h3EP=E{JuFF
zSG@w&yv&%T9|x6CJvU2>)|FLS?s@`)ws<bso1>o(mKhFbwp#d)R4{fkZO`Eu^5g}{
z<FTJlYOuU}Y7-<1gfD8COfQ0seKGxfSr`jLPS(A{B3F7>vWJh_S)jlK2z`RUiQfR2
zmI(XnN&W2S(LnY)bQ(&8B@Z0Lon`^8#@oRK0AM&%hyly^wg;<{3zxp}P|Y|^85x<7
zbsQ(bFXvYXtq<!!cKY>D#kAIEzdHUr5;bxYtoJ~0L+zQjm=7TByD|+PnEex+YrE|u
zKPS(<lW9{`ZuMR7sbrU_)n3tMfp&Fv14wvCL}nOwlc!Im0(Q@oO!!@@S42>{K+!^r
zqNhAy=JazeanR<W4&+ngR)Nd-yVo{5oqJCevA?iWwBXas?c!j)i$+f>7$?szKfo1`
z1qAS;I9?>~;pUuwf5e?(UbWAN%P{87Vcwpms-9nc=NmElgM3#j``1nKEILrmxwD32
zO6AdiIxF%atvxF?k1zj<A7TDpHy#E4dHGKqBj$d2_DSwv=k55voA;f6hAVnSe5qPH
zwz7yCG%5lkBBa^h)s0aK4jmre8kx{@-@g`9eom7zR4pQpJDK<|tR3Se*vUPa*%Z&!
z)E6h7{(dxOU~lG#kNsN{|NC#0pMtaRtvPdzN@bqc{|tc!m3x^berqJ$6MFpmL~-&<
z#kW_<ed@=AbFu8JT*=&&3iJQ~bYi-8?v4R7!;$v7L@D%Qb)>B69w5^A;KCoF_+??L
zfi?(ONXq{aL@&ehRN)ANlPAoDRDCjUe{*$OYG-oMa_#3ecSEV5+u=sBE8H{ULHkK5
zc7&c~I)mWF)PE4y{COt{Dj<8FV+V|;3HL2oe}s}CR4<%VEsuV32>ARM_W43i{2&U@
z^Fni2CY+nBbz~A=2+M$IqE4GXjBmF-^F5I<;_eAfEaU+Ca9U!x`#?CW-B3ePst}rd
z7P$$Np3#cYDX-aPC=5Yj@xg!ZFs|2CmorE9!3nF<jNj~Rj~1iw_=&u*sIV{DwsICY
z<X-wj(nfTbS`QbLcFlD?8ub7;s}D+u)PzF)SUw&7rk3B?+RBY9%N>rcD@>PcwE?7R
zO!UrO-B``4I$``ymaQ8<wmv&yUzbs?+_lf0mnH4in_pInZIaJrvEfoqUNrp_W>~%W
z{s7p4|MK#Be@^@0e!Z*z@5~&M7qE=e*Hopyf4qEj@oV<o>i$}PLN7BPHNfivoSrhE
zV?B6cZ}{z>-JQ;ZEgTV2TDEqe`r#%TbNTXC_ut09U!8w3pZ?vIlO5*j-`{Wl?kE`V
z-v0T0tx6n|Yct&GXEE{+9O~7!fVun{^h6_jD|K#`0~9Qj7z;^d?aFoZ)8HN>dT5bZ
zpk%2xwJNoaz$)RfFy(1p<8MXG>tUKXrEWL*PN6ySekx!3_PNc`vs!wLvi7kT9(>?W
zPxKpRm7u>dICne5P?ta7!3%nF{^*vz<P(!eKU0?GS`_rN29v5qz9Xw0(cqnRR~4*M
zB`Q4yh#2#<wuJS`bAJBi@^$Q~OOJPyrFjZ##{h41Gl^s6-F_!`r;?&qf#8)Ek^Jdu
zT6a(+mmP)im)e(2<85?5^JNbz#m`M$z51}p3G?QdkGz!VAf#xS_@Cbj%q)+{m3?|$
z`Qk3w^Jk2#0L?_ynrx1xN^Fj7p=9(7$V6(#Jjs5ekDMt>N~5D)tnfiB?ehe|7c+jT
zYax<GLt5HK+ohwPPL(w7gwXej_m^-2uqKOAE=2rI4Y;0~Mine9O5su_rfq=t{7~$Z
z16IeNaAYk(KH1uUkbYUb*5G9L?qW9`;z#go6P-tQMUT*!5(;fKW_*XSQ7?Hw{vABJ
zzD60&rHl)lq_ztvHsMtR;44YwohGEXDwv+yM9~4hJ-Pqa{}F}#%F*})<n8gv%fFLU
z->i2Bmx^Duzwz^&rz;8~d3V2Qi4<F29|*xzz4^=^yI)w^9`JdUeAClR?w%Vl>Ye$&
zqO-t_q6izNu9cZ4lCdwJkl^O4kN<w@zWMJB`L5Bf$9$fCh&%G?<lpzsE4Mp2@a@s<
z_P>J-yT9)-H&YjP_BB|~ua#jsM9<m59eUz(^?6Zi9w1n0ch>tLgMbPW&!sxCv~0Qp
zI;wqt_2TX<ljF&E(!p+?hpEVVC%XhJpeLE4x^MPT^%h@{1=>U}n%Db$btQEH)@=Lg
zVWaRnTo#9R*H+xaUpQQ;3dD&5J1fTTJD&a^{f)!k3&O@rPQV&0OglE-u%5ElPsUm&
zZd5Qc6bjV3yI5mkObPquT=puu;tB*vf=whEg(`?+c$C~U3TtV7!*K2duK@3GQxfd&
zo=u5JhQ6>-^w1OX8*GIFpNB>f%I_9I6G&1KX5JZS(cUl?^_Vv5{P*%X^j{x)3nS5J
zI9ML%?g_P|Mw+V@9?H0Pb}kCiPwnlMuT%?rih;p9>}G-+saZ&B%Po8UAi%s)Zw$k?
zT0@Kq0l_O1=Z0y51FE@pUx^PF*97qT*-;tO(dNTu@*;G5If;8|M7e(9S+kQtuB9(y
zRP|C|-L*%~;kdvk0UVy55h3*bZRfX?mWMtA4vbOKS;Y=A9|ms@9ZLw+Nl6aw&ZH~r
z4;`KOnnxDB;bBEhMalXaPFn!3<!f>v-FpnFA_|sZU2@U9$9|zUk=ai6lvm9H+pcsn
zp_*LQ`EvO-_92#Ow1*-?*b?w(91~d8dewcc+AjSVBNxuzb~##Hui@oIHN>lKDh@&%
znstYOph33WK17UsvY9MXu!gTYPt)5@fu`C=>;3X|hN`c&XkWjW3M(8tRpWQn`-6%0
zwpBV->Y~g{8Q!R=sBw5I3Z;Xq!ll)mu8x<d1*>n@AOC%@+uGW?g>2n>(0fR1`)n_6
z!w3=)cs5-8;bv{a4s+q#Nueqfy_!Zv8AGL2h)=XWbNR{K3(S!Ve1{`WaBz&ck?ToY
z(2Bgj9L9v}<pm7*Egt#c(u766Vlj(S_S`{~0JY&Da=5UDc2492&5ZPf!@+9=f<f5Y
z$h-5Ko7uCGu;3pC^iC;yfA{2NDxOMP|5e@3;569!bZ=sNkk3EY*2@Y77DQ|{(V_6W
z<)0KqOcmvUqawCRjS~_GA8!F+wBZbovi6=U7{KHhe7z)}Qb1K&R={;%4?`WB7?i{y
zz%6eI;782O%>gF15<qXb6vCVpp5I@>lMgQ~$EjMN_g1*w3rhkGUR_q|ZVR#TrdnHU
z?eEY0mgxvMbId@yb8_~HoW8GtLf}I1S^^<58im)E8k*yGSC9za02x-quOHFRCQ_6J
z8OASu{jS@voIfa~xo`F=6O*(Hpd}m%BC?+}C*q8FVj7iC_96rvS}I`2J`t29<5-G|
z>0<xr=gUc;FFylIel@@H?SsxXchQ!X!H3p|RA*m4vY!Y2Iu>a-yzc!yd<%mFG8dny
z0SlA!3;{M4B8=e*qrLcwIRt`N5rJ1+*v_hP+|t7Cthq8nQ}b=tF|8YUd-<P8nsWQt
zEw+kw!HoS00W_tU#`|i^4>FhX-(I@8xmYNUZS+qsi=F*~q)Gi<`DC6)esjsNQL^#3
zHuR9#@0D`xMiui`>weQ$3yuC1ktx|=1VE5NihqX})g~w9EiL&D7LeNmvlTX7><caH
z=>^DOh3oxp3_FDXVFJy*5Yx{)B`^?a)dCGXsJtgiTyx{5_P0F)j{P6E1fCC<OSenk
z`+B%_&cQv`JLBRT4(7G7rjY|>X<eV4&Ak0n3to|1m_h8>1wI=Cx_e$5qOdyww_hse
zOk@?CURwp_B;Aos3+QP(IN81{DxGpx6<DsE-fIX|Z*`T*#;`+nT36y;ZT)B9#K$dY
z`rcr1_Qv&_iM6G6cb4etCw(#Auvu?o<FFYCH28!6s?DTWH5VfBLx#$7OO#q_fQRiL
zXJYYowY%Zuo4)rAos<Ybj@TR4L6g0T>`;TZgwI@V0ye(pl7S`H3X^53S%vET5@bHM
z$&vl%0B1-SXvRDgv_%{p0+F^qR~HX_YS5bQcS*=p+H;qO&gHSx&_~<(Wm9)Xwe9O`
zVTl4mhRSgExsGROloh9se1alw(Vq_=8lTPt7t~z;Eu*s`vhu+vY`7)XS2dWb>XV$`
z6)mWjBcR(324EC~fSCla`o*w#f&W9(RR=}&f9)ltrKCY%X%K-0r8^{~k#3~B8!nPd
zcc-M1(jC%`fGl6SL8K%XSmNFHkKfGxcV~8H@7~Wj=RW6&{)IqFRO~_Jqp~Ofnm3m~
zgVH9PFvu@AZW1Y$@l~L3*e++EyEiZx5T99oY4IC2@C~+;dSu><Hdxp5pad#hTR&|(
zhOW+6F+uR7DbP!xJYX~PT87DR%4kMV=;vyh$OU#HIea3|vr2n0EKyN+V(0Ht^t8w3
zlL~3^8H&pv#nVQpNfur7=12j9DCg&l&FrXVO8@<`N=8&`eAJ(8om;=JM<fCu$4UaP
z5xyM(eiHG|e-eRzyWY@WRtXV@_4wUBv`kR~Nl-Zu{i~?Xy|%X@A$ra~nG+JGmhH2Q
z66$++&_TaT7;CnN(TOBg3D2h-GX!F~9el>_;SX2ZKR8jtP<c>0-5dblme=ST;*bCK
zyQ^S9j<*X{!pGwN&)x8k-RygW*Gk?2By-;0uRjtwx%%Jbeopo)@K2Q5j*_@_OxiY=
zdf+0sdh$i?*v!QfUx9utXOsuop%^QJXZy)yC6yh=UiD2S%`Sd_&AuB7rA}X8%O?9?
zw;PwK**`GZKqoN&-QWno_)$vpP|U19&O2yrjP1IkRJ&oKrlzh$If4Y_%4C$3_JVuW
znj4(F>!GWKYPosS*XJB#$L)oF1#m>rsne^G#B{#Y$EMgoQDS08==@zE5=UI~Y~AqD
zvOYb@>(-b!U^r2F5ds~3+<l=q9hdzYKWdWCegD(nPL#fPDgywI=XEGW($p|I$R4H0
zi>>>OF4c%z?}iKxi78a9OFzOtvssh|I*Ohp6C$%<xM)|<|3#gQbnMvtiaF0l^)+Pi
zv<!oxKn>n$5H7U3>TzYq^Z9J8nJ&_t^%J}OAE%naRq?yy?uv!<sb6B+{^b7@d6pKw
zWL>nyg`?>ACg=eeHjH2%DcJ<=W)KE%)c0$*oRM}fxIxsw!qmd`D+{6b&qGKHY;xqM
z_x_rk+oW4`CKW{)mXl4CwieQV+QU77P%8+$|B1;hes{axp^*K*t^U%~b^Y(sJfF+&
z*Tz<;(jWo&@2GRv#&x9AWtF#p?t9b70Zd|E;1w?FR~trwZ=I4*5{u$I&d(8(OL+Z^
zqAJiGE2jrzGingX?sX`kP=@;v{Oe5D#4-g@)8vUG=nIytddzQwoZM_fNs3pC!Oljv
zt4%_|=L?F@e=q5tHkXjK$=CmSi;prA^}u`e<Wo13Y^W^ySpU?B_CU5E;XA939{z7p
z0nRN_Vg~eC{tDETk9}V(QLvaB(TUu{^f2E050Wv9r2$)BY!aq+m1UzT^t-p|oWftT
zP*4aPQ1ccR;4`LEz!mFrn@{1s>x#zOnvv;Af?DKSJf482%ed?2<M}ETU$Vlxg8Rpm
zqR61f74F0rwTYeqLFnpC9eJQ8S?{AdfVxdqc2KTV`D8U9;C^>3RQDFG_m^%N3Ozaj
z0|V4KW40>SIy*djKoZ(~&G7=%bp1}#sQ|H>)qed!R*2GTzI_rbWvABqCs&8G5f7I_
zQ8(P_C{>B{OVe$1Q=jll4!0D?&24Q3n>Vuu(+7TBoN?$NneS_9;p9_W?R#5JfaQ3+
z<13<WxH(13Y2_}-$<Z%^kEVNsoV*Dozvm3z$V|y-FZdPsG^@_1v8MDe<!|dWQH;2K
z$E?=0U2->4Y8P1}@Jhsj87`f)HrcUT;WC>)dm@9U5Pmdn=?{~;n*mh*OiJl;dP}t*
zuRCG>qO?=h0*0da?$&eP)zrc@fh;-bi<^cGI)<((J|ZFpSJXEYglWItUn2iLgj%h$
zx6Jv|7ABpMsWCy>wj)+D)?q&;L_V0uE_45-Zu0Mkg*6${5Y)`=y3QGo-d>5s8nkog
zR*Ci1(I1_3Biz0I=dA!DwfAwQ@ZqXLBH(FZ5y=xhW&1s8eI&MeINje2z{mG=o1fNZ
zf}bV`B<=?`5XW?ZH-}Mbg&CU2a_UfH9t`-<tuF*RcQ^CTT>c)0*<i{`6M9QW)zE)5
z!ofxjUr=KB9rFpI$D*tso-1?qA2MfVVe%(9V(aIWfOCH*iAI?n{;!8=&u7FcyThZ+
z&+0{aX9|o#`j$0k0hLssQrz$2>Ch8(1c8{}`l}v8pH$hxzqrP;QP}GbSuX_cN`PDW
z-rI-Pzk4@h3%3`oz(t7!SvkyqNteOC=UW5UvfBQa(Js~RCIR%?UK;B33QNxPO3X)k
zLtfF};OcUXZ0`xp^^~#qZ8)t87i&qKney<6Nl_^Y?KJQuF=PP`rnK4uzxoFGZr-2^
zwdXtF=8n6kt;cWtCCGvD=GxV(t>Zg_dgSZxH<B0$9dC&GG^YHeK*M!vKCVbJiyscp
zk23|@C`Ph<8(X-jY*uX@J?+4C42XfVqIPK9A5U$Wyq!{OOK#aQx%Rl5QEPb+`i=Kj
z#Q001_#G@z)FVKPvch3QS5-?{z0=?7>nHOuyzSbW{M}EEvRvmN%{HaEOJN*QE(JI!
zNs~V_`Ph1oPfyk6N6_KHyj}0cNQyA%RYlYa%^WX#CeOl~R*OFO8OgLYvr{Cyn9e#9
z3KmN13i%jio=F7vzK`B@#CyE*%j5~l$eG7I^%e;+O9433cR0)~tx*QwsC{kr4iep#
zyY~86(z|h)JCORSe<c*HTo@1W_|+#Ot44&+rdL%K!$mTVlR*wYYKOJ(>EHf-c{63d
zac6Fhy`aDQKB4!>makyozo&V+;EUPDqc^ZS%)+3XtDBbJ7EZFzG=Hem+#QuLG6_dL
zRRaf$ARe&8qCK_Wf=cVd-(U03%)j%x>>%f2PU;Qzg700{1wW(~KJJ-59+^J=UF0@h
zTgz9isY0XI^2pT=$YRXOHK@okl?#k;eZCH&qt?SZ4-inZ<X+mMCfu6sJj^lQ{l|gK
z%Z?Nbr{=SfqFvScC=bmZ9ne6JO(Kj5LCM1S`U1$ZZ#u1vE!v+O{DCTY_OONnITXf6
zZ2VqU7A0a1)yH(Zv=cTpro%@(J|4a!A^LhA_<VfRty^aTz6($*3_M@iI6eBc{4iEQ
zS8%Gf_xJD8$$a6%L+10nOR#^C$L;onu93II^C5R_t&2F~iV@Y{&!;EYUFH)i2?M6`
z1uqcvlDOG)J6VMZ(=GdlLylm>Mj~SyBN1v6R5CwTb!RYCc2hSF7GfrVM=SN6FNgBt
z#kHdqhxuYw4w3YCJVIv*^l-c{@RIzQZZ9o|M)@TIWS;30(*Lx#)9?O#g50Mt@E*~8
zs(7;&AcmCa2AO%Wl4V~_GTCc{t4N_wR{*-&UGdz{$7;aU%ErUPj#{ByICq2O*>2N9
zQ0vMXa*yQQ$?*dH)7n}ki31Li6z(e`1|t;}_`ok6_T08#WGtoOR4v~EiSUoa7?ou6
z4zgpzC6H?Gj9B4veJWGS?<qzyeWVOC7#im7juC4yRlYxesim=K#ZSDQk1N;iW&&&(
zw$FXTRR*oSnA&daogVxCwtyve1oxk!G?8L_oW{y>sFLDaMZMo7t3kS$lLaQ199E|_
z)o;+~_0pH!e+yP~+#ttT5mY-{(jTA&lcOXZ-~6pw%Zvu;iuuEP->iddUFL%muJ{R&
z8^cWbtWIV&@+4bqdIfg&)eV+r8lX8IY2&oR&@ikp$f>(1fHZx5-d$=WI5p|cnXc&6
z<sO;0qnaT~n|DWhIGmc2rvq-sqUavR=$?L#oUQeH-XA}%tgs4b4*mqssa1p;IYzK1
zT4}nWf~`m+MrBv}r#d2qLM=DgJY0;m7f6!znFvkSHlDWydhgEFf^RYb;BHakX{B&2
zf7;02VNDd&*YSSdI>&-hl0@%!mrs7v$`y@RKOV{n+^9gT%F5QhC}H$82;7fb!b7b1
zHb4-+^L4nPF2jd?S4<W>>sp3T3Q2TjrcaS#_OY(I+K&5QE0@4v29zS>0aDbT9tO8z
zRR6VdqTn$o=(o9ScuqjLfkmAkq3;VVhJRmAulN9)SOW8j;=$fNKCRT~lfZN11~RR_
z^8S2;4vC>6&I1uwhl|{2NioG@LC<5lGrdn?bWdnnkM~Chbil(p((2cV=&Tz+OmLgX
zw%<?C0SSAqUWDOE7)C{DkDj#I`bmZq#eRZXC>;B7G>w#m;7~w>_;<U?srf$lnQGMI
zc@??o_rH&ikBGBgZqaQ0n>3{Ar_SWz;$ARKld~8`Ova`huK%Qh-lB5|ixyJH8+YmL
zkq`jf14b?81bt2(yI0_<2j_RO=R*_o!Mz9mkA=^%!1W^VID|>}uzA!u-<TL|i!u;A
z!W1MP(3|VX0?=gLWQ3xl?@9*p<X7BH`1xz}nRGv#%)<%J#1M@Ay9$J%jDn6Sk4KCJ
z&*Eu2gbuHrT?*INHux+IxIhfDx-uXORXtiz^&1?In-S-BDo!vH1F^@qg`oU}EHYeN
z<^9@v3f`g@$#6XWMM{=T(*D5&ES}hSO63+9*LBnuP|Pimk6t6DBjm)YgPs3!sZU!F
zY<09IS~c=<MwcIGXDlz>4&$$Ya&C!AlELiX)+-}qPS#dCSL|p8sx#b0X<A9hbUt_M
zY%w<KNl&L)2r&`$5}N2Ntop?)AGF)}D&&Dk=q=gxc)k9)YRjEw$2YG}`m8KA$z2&n
zKIW4F;Mn>WoqjiM;z`k&DZ>ftLcNhDqg9KK*o9fX6n_`aH({?vxg#JL#Ssd`<^w$8
zxptTWp{@caHEJabw)2$Q8sYk6!z!o2Dv=#GO}dHGHWLyp#4_%%kL*zcC}km|el!x#
z*W|#pfW%X7;luGk;q&(1+T-d=59bf-|4@v9k}3yu9Qs!b&9K?87r&;0ABQq4bn8w{
zx@gW$cjPmegS<zRhF0nGJ^dZbCpBmZg?-T3%Gf}WpFw0`B3Vlh_so984#Wn`_chnw
zWnCnKgvA^OlnCh<8kh5zr~k0O1YQ5$e0|aL1@~2Njq)TTgMM_|kBwv3)po?{Fz5(?
ziMIu*X(Lk^`LrJ&E@yy$VRV64$-$4?L+7rK*RkZy$3*%nC`H%aK|yN)z1`g}Lh1^7
zyV~2kyB~ivpR*PMTk8_{|GLSp@9k%3WF!1+yBU4I%hd#cz>J#3RR@<r!|KX{Q#oj1
zrzY$12f+YfUn2%QJ)E9CtOuwO!T>X}5NOZ+-@UCLi6QHxMCPL$l*C`quG>)v7p0~a
zRzrJx9Jps*PHa6r(I&Gt*Y!Si2+WnkeGb$l9=C#@k7u4zft&dmQ@jy^U_@Y--`P@d
ztBA?T@Odr%PAyw<!p_HXDx%BP(Arw(!XAG=0qn4dW|3CD5@b8=^}Ptk0H(!HLPi<5
z{}YjHU?_v3fYx3T;au|I!QnQtX_^ehhMAhgm`&w~N9Vt9!Cp#xuS2StKbXB9f0kd$
z6`1fghMPOoh~;_spNjNg^FgIbWfkQV>)$j@4n}j(YO1EE$JE^7?tWxWZ+OV0VNI}l
z(?YD-qMdA+!lMX6mkQ|%0DK4X{l(`CVB>V1i0v;8%)7s<bbbEzv?B5vZ<c$nsbSa7
zqO@}4P`2A5_g=Lm4XF8h*J&ba1Y?U_vKzMWpay%{g*^^VG|zCU!q-u!{@*9htw#SD
zGN?pqIm_cB-DSMu+ws|nlh->dD?>(wD$--RFasN?rT%o!8tBPN0cQbaX)TzjB8tiy
zUn+4*nohf4p=f0h&JSJuosS|grsLw`3iI$eTYcO&MObU=gkUK^UL?!PW{@znIIT2w
z{`5Wm%wuri!(FF_6RpKoebTKpe7l>jA?{a0;^7(80K-X|Q&0fQsU(HRYbX|%_Y2LW
zNHRo+GKD5-@<zmq1D110YlVq}%!TLOWcJsH*GDE~>9IB@L0WN>sJ?yJkN}!v>Ib9M
zhWrYc(wprxl{VyY-aD`$Y>BLF`0>GNkb(Xq4o$c0OLp+F(EG_GQn&CVIReNKg_ea*
z>E@;qI-;Kys|X5Bx;WdGm0kX`O5*vmu61QZtx)vgxN$@cDCE*{pcAyZG?5Gsnk2%7
znr)*KC4Es~#vmsCr2MN>{HlBKUG?~^t>EkUGAc49w2#o}Y*Kd3aM>kA25L|2My@!9
zviSG1*n|*0LyD3{Uk;R17J6yT(q?`gYU;G<g7mDyBz@2e4G?8=ZyX!{#u=q2Pe#{@
z*3z7m(bK~Wum$G4I>@<S$>5ktXo7JP5mwC4I=7;s9MG51f?`lH!s_9t_Vwr?u)A5?
zcNA*gool|LWFZUw@QVp<;AW`;raV1;%Bd?5)^IE=K&!TRWV17C$y=L{nBbz5xpSU{
zt)WdPVIbPQ+I4>s*1(BrwGfr%q^;EV({>K_`eP;Kcz=ZCuxGMp#h}@{-$8$=7n@)5
z)agugEo6Fop+bNl_jopOPo>PCI~4@nBRhw6&&vX1c&al!7fY!pv$Kz|qn-KBhRd4+
zHi>nodGYtwV|{WE8iI<6p#~~Fkj>bj_xtb17x(>awlLk<6E-iphpYLz=egXF5Mk0`
zCA>lGe+NFstv<IW>$spi4=ercFK_Zi^c|FyQgn}2NW}oPOZT1Zx|m0fEIrw3e%Y9$
z$S^59A}kp^)PDMPA@C2F^?yv6j)h&J-HmRiAr0_dluoueq+=1;(#*o&mg)`X7Z)Sh
zl-N<_@6pcn$+-~!v{qk1w8q(FZenBelCX(X%=xIMX4pCKJrtyBKs-1luG`_Ibz~D~
zL1zJU&&R-L`U~;vZaxDah~sme#5aue+5PFIJ4@k$DVF*3tXB?YpxF#7YePaJ2?07)
z_V~!umMS8Mh_cmmBLDEN;I(p{8>B9RD;%Az|HIM4J#RAhtb-G%&rgEpr4WnX^xnyh
z((S}Y)L=-f&_QSPkK=qC{LOz)oj(fUj5OjKK1ix56%BZZKIMk!(<J*438RPRJ1U)s
z_Xck?nWP%vDlNODWeSw9jg%+jFx~L58VA$F+-1KKgVtlc_~7`LS6J^+zG4k#;mb7q
zz7EztD=;yZ@VmLPcWZRYNkYG=CGN}4Mp~K*`F%oX;w6SMpeQkjv&hMg^Hf}WBW{=W
z0)+t)3E|la1Eie5Z@Yq?ppqaX6f67F=lJ)-_D+<FA!&0~8dN2GTf)>?5G;BkJE!V;
zIUau7_!M^-tqs5UdJfT6ri?Sa%{&NXiv~G8+cKoNybHxQm347u_q>}Qm>M2d*Utz#
z%aUT>M5kwBaN7AysOjqYZmPOuU5LT{-WiZAP3!Q1pH5AwXya#0Tw5BwtSV-U=YK~_
z;Ntvg-nph(;_;6^e=b7&YI<odskXq#=)~KcH=9JxEih2@bt3<SwtjY}4{q`7P@}^E
z8Juo4;A+c;(u7KdC8*QD#8k;u+b&`Ma&V=AGq?PErIN$Rz0RAD->|RE1oi3`W&HtB
zTA@WB<Epe%L{T#mA0F6JcX-X{+<MY-dMehof%N}d<N$BTEFy~4jgQLb8>r&qdAWOU
zyi2!`mgzf>J`zI0-M!y;Y=e#PhQp-3r;7;-zjU7EcYen^dlx?BD7tb&7B<&bn@p%t
zftu>FNSu?hIY5B)F>ia?oh_WU*#joH3wkg5*QBH}e7Sj_ClcfnY4d5OHtF9@NupTJ
z>lWiZp(gVi$TE0rrwvK{Tm2zsN}R+b>{6p`aNxm#uAyg`dYcqS1s5p|np^)#&1-7V
z{^bvX78!x_!c1ZmBg9Z=??eu3W}$j=Vlo%ai)XP3#<Fjuec2!-+b11ETmc*xIy93d
zPhV5gYAD%hhGHh7oAdgsMaz8}zM5wKIYS$Kb8;3K(7Z%qWe2_Li;E=1wynvy{?ObE
ze`>5y14(HLiirn5E-ab=dkagRu-{(J-Yc~&FczY8v}36MSy}_Jw1rMgTz#^Z39j*m
zp+gAxtn9u=zAO&5?)UjEJz!hIH~B4Xd(f6Aax=db*5ZWlc)gPO#Tgl@U6NMbZJ-kF
zcNV2OPNAi@<4-3D!9la><sp)i+RvM0Wh#n}o1;=BX8h=FU_H$dAKvH3^%Go9Vi5!O
zJ^LAOQlio)YtR(atrq=-7t`ORIo(eX6%_|0G@LcJuB}M<S^0ZB82HSfK!W`x*>dr2
zBKLY|YJ2<se(8E@C$_2uE~if#hFxTlL8!SHa1aP33L%V;{K`ZqCG`?&nRmwTgk~jF
z>!`16)M!{MP$=#n$Sc9_sBLR#If}30FlRtrrbt9PsKSmzYC!D}r?^B6KCG;+exHQ7
z303^h7B^<r_Q!{4GeEfykA06sT1uAg2KUXjDc7M8jRAU0zqLr^N+%X*?kh68FJfjq
z`FV86txYJKnfaH&jCJFXMKTiN7CST=MT0L`8)h(%5Nf^znF0tE^3==VMyKhSnL@Du
zBrW>$@yL$b3heUk9gBcO&TlWL8cMvAU-$_Li6UniRVwe1>uINq6>lD)6fBiO*iqFk
z;wz@%O{cMkpWCZY!!!Sb`?M~YK>=4x@qbSIm*<k<Mk_C)N>02Dwb860!?bi$%Ppit
zppJHZj<QhP_nuZ(x#e$TP}A8zzOdBIWogauT1*X<Cn)^S;@6ET5oJvPHqD)_`0FeS
z9*UkjTvNck8OU7|>ojPYzf;7r3l3B+{{>riFB;<j8&hr6<em)tEL!FAfn`Lf)-EOE
zSwKg>$hfN16vhASMyz*U<A<BI9?^*f-wvD)w3<rzA>uK2wlan93IngsX9a?-c(ZR#
zPF?4)%}mvuM+)waH?~<ZZ=C_v8W^b0PR&UgqvjJa0IrcmkA5c^qnPBu`OhhSdRJO7
zOJsl9Aj2vph*3?JzK{JxDlAMwInwsfbyZxn_uGS<HS+de_~j-Bkk@2cp23m|Syy6C
zGCK&}efjp4g^Q1Seq9Isy{|pts^fYtJVVCpe)@#=`6V%u?uTl>b``Yf6|R?Ig;$-<
zuvvcgKKNoQiZ1AC_VDm<#ih$HIJn>?ahHb@yQZ*v!%|U_9J>?g9R(CxVnbF&V)rs8
z1U!>{sh_DDgh~laS6$|Mbn-tBMIEo@QAoPS>qv7{bYX9bmtw=MH?tJEnX(X%=zUnl
zSh!k}X}iT597Iy6WE#+RWwshDMX<KE28j;4WpoQo_MCNXtd1W}&Z}%9Uza)VrJvG7
z47f~^Fe(y_>0;dvZ)%d?h5hZ?@Ore@mGE~*YH~cf^^AeHnMWrr&dbvXHa3Ou=r2Uk
z>UL%NxL7>6jJBUh^@&FD4V7f$A{*3|u#<`!_I|PUm^cOBk5#13wl#QbW@u}NV0rAK
zvpFgdV3wq<`E#)2I#jbT6oy*N8RYh})kqDJ$w8CDCyM_~yT$d{dLn(_Q<!dPHB-{m
zAt}-fnxjBUmd1B(`MTd!_v>hq>n~YVht2RbNru~t-tPq=^GZh0O7?NdM-Ebh%p!Yy
zwBM^EN(CmHHZo0mdi6L#;ptcFZ>*xi&W9SS&L00@*7e+DV3miL%sTR4W@eIK-dWZ!
z=NA^_6|6f2W)_)Iv^gJyabuQceL+S0{hBWSrrQvx(M9U~h3ldvtBSv&|9YYBKlmd%
zPJX@p?)Z0m%ZY)*u`SO<4~xYo&~i$p#c~88qNm&EY~6kyI_k@7mQXbKX`J<R2*LHW
z#+m=*9Z}+UJF3ritt%?!pXu^_<A}2Cy7ln(Z8|{TI=9S3De<eeHrF5P9DQ8f12K&|
zW7VpFS+Suoiah2fK@4NE<tr~%JmcQ2-)PiDo1a}ik;;Ln%P?WztO8QM(UJ_M9<k<L
zUsu&l!#elvUu%ZuzU;bOpDL+9Cb8%SA3iC{%~C7V3a+iQ&fFD36^s`JbQf7!M`mb!
zpu0x88DU-P2V8>i_Pe-^eilkG=&VE?ug5fCHN{Gw)z1;WE+m;o$YE&7{>!_w+p24K
zg6s3uoxu8&a9P@qBnHJe==l%uw@9?b>jvGh)5gLELOB0PqH{zYXy@FwaVaf~<C0xY
z)~<%2s0(g_-%wqhJszlTNR}+r`|a|wvBOUMb_^2<zMLbFhC{ilOw~2aQ>kOhgfajY
zrMp{Dp@n7$+b;np6YiS#@YCMg4P-jUMz_bo{>9(t2Z^U_`1$SaH?OHcVL1}X=;Mo<
z*fdq**kWlk@$gySKV6s^TdsfdK3h7}w6r)Gs#wN>H4VOtmp5>3g;dk7!ep7OPV|PE
z3T<+p%1BWpvq~(j2>5V_i=-{g$<P5Q)wm@tEOghp)NyD2rTN!VB${?OW^v+nbL~Ag
z#X%iqq+kX~^Y~dDB4Spj<_2NcalNR%;D*?5n`TO5$OP!E^(Eu94}?S5=sW@1yqX6C
zL)RDQd)F0eT<eI#5#89m+7oZDoo+RT(k9Iu1r+*N+s`q@lgEy7=qVQ?A2a~6q8J5|
z`VZ_JBqbmg6o(j+)s_=FiOx2kgXvjMULI?8ZT83>yLi0>6qGoACM+mS4|$G3q0vHS
z@&`*Id|H~NB-rUK3g92~JRDAO?4j4-!3SsCkEH=+pQeRNkC*P4SsIpB>13cw`Yn}8
zva3T}gGlf9MK7hT8bOiP0Q2_l?XZ<GlMHW)8DjGlVF&86CI$46B}Z7xiFKp-{<&`&
zlXL5}rMR~fA6sngikSb^H_nNjl1BMN@wYMC)A@~i+xxcSC(E^G2oiJAoc4m$J^%T`
zj@ji%LJQ@>zze&~kD;i{0Mb$WP4#)bu2Q_K#2cgzw_v6scZ%gm|9+jaM2?M_Ge&9y
zuCh2dx*m|=7+ZXTz`-T+y*d8-=g*&gYoT22^{7hDk|U3|Z(=M23kt-1MLt}@xF13x
z|6Li*?NvrDITAD5YuP$6;ZTrl7B}IL+P<J^D$ZiMUpHtpsQXaXp}Wqh2nisFxZh1p
z6dG<__g5jcn!igF_rJP2*Iu0PM|-{?Sijpd_20_uxi9|m7UfO7L@y^AC<N}-ZlrIx
zD>A*@Q0w3ji*yX6a!AI?I&9C5hsBunX3B-3N)Q*&ebP=hM0AUUHfIlxMcpomW@dz;
zf^pJyO_3Y{d5PfXOeEvzHNa2CFn33ocmE`^@f`nrEP#yKa%tx7xkUzWJ<S6by^p)B
zbQsYqYaL5Yuc8h;`SVp1ts5KXFVB6ENX5LXC~_v$Fv1hIRUuJNEABbFq)rd~wQO6e
zp?x(?3TrX8c^k?@N0ZhC-#KrqJ?f(Fe60gSOB1x42don@PjKD<4WB{^8<*OwJP*d4
zhW|9rMWI)Oy*77FfCDm-u9|U>d%|}SgcdOnX7r4~oD>xlg|8Usat2?E^Sqf#CD{qd
zP)85d15ya>Ni>)FRvlZOx7O$YIHm-WYDISyiv39^4wuplTdIqIffhtKOD0|F64e1M
z%oVbZ4_WQ-H6sI(8PVhhnXo3~Jz}S<7k&1A;6V6Qx>ua!OEf5>hDFGU!+2ZRl+q)(
z460^rWlNj*+pW&mMaVCR%j28NB{3B8G%XMF2=YD9_4KsJU;*TxdCUm>iLb3#8gjB_
zl9n9PUMvNjVT4(T7drB<7~Y(_3U;aNmxlfLqtouw=_U>|Ifh_C%3GXb`&C&|jf5L>
zW(_)*sy80SssF!uSJm|>mC{NCvqc}BOtTs58D|JZeZ>Rc)>a)YXU#4Dd<@{`B#RyM
z>@hOB1mmI-X>E^LYLZ1{PmLEVL+qW@5@wc>+rq+6+OzMcd_@~zR9ZtMX`0xYXPV~9
zaWpA5v%nJC8<$<my`@*r(J}v9z%T5*sp}&7?AdvLJ}+O_{j^+4h2vaTM-C~k>>IKW
zqrp#`llz*rwQud`>ABisyx1}8o5Gz%!I;$JaTZKI-r95j29|OrFjaeFt~+g#HN_%O
z^+pp$fm<550fMuS+o~g-DU=xFAU8-BV~|o$-A?GMPG>%fza++Z-S2s-o4A_a{?QG-
zn<6)DLrlZhSDeE@4`XnNqoC)i-iIP-m@>BB)9QY$)AF9X-#dw?Kdf|kn53XIZ7Zkb
z*%m@+!Q(AnfCU;(81e!{BCSEv<3m|@IkpFmYOy|}k<7X<xew_$_7?f9j2w1=+tnU7
z3G^2J@TB^E&mTRIWlJ-3P#mO)NzL;-Ng<t1oGN~~C1PNg-ZaCH{##8vyQ0!ERVFIv
z%Gi)XM(K{kTp~mAOl;)pPouoV?QEjz^i0Bx8e-FCWCXZ$d9FG4yhudS>}29({}^F`
z(*{szlutfGY35IUaFn&N1~LMn3XN6r^XC#z^1am3x7RKY%V7cVr>Nd%c<<$j%kvZ6
zL+sw)N~*Mww&m4of!LXXq-OqGk}kEp;1lXzw}S^~?U#fR&?J;(f}(7;GD%5rdrM~M
z7oK5-_7eFrk^{2e0lh28s+sMPKv0ijT00JWX@EbZGhWc052SGa;Rkra{ADd`G@|Hp
z2CA?4MO`<3AH=^V4o;?Er%5HBmHmdtAtzasMu;tmpFt{1{{l<n2X1{0wQPf2Sz3=l
zP4La`5<K|ov{T~laK}bh1*^!>n;=~@S(76`fyV<14#&bQUm_PSUg@>&?H&artj;RY
zQ@%V<w0k`+Z0}#(`vdU)pfExv2-tzRAAMU(xnDX>&(XXYb&Dj(5<etCADxN$h6@TJ
zEE`?NJfI2B>P;tCBZ5wXhm!<RIT)Cj6&|cU)Psc4>oZ_S2JHXDSKB>WR!%@>@2H#w
z5D3<@&db-nZ%#U`o~*ynfh_5ID9YG87q8jBXxcX$<mct(+2!Z&{imjSXe)k}b923D
z0{IX>Qc~$#!WMHD^xT<yJ>|$BE0s91GZBlsc%AC@e%<*jkBpvkvSjSP3=DI?6l{SE
zV?IxUcJvhD%O}!F6wFv(F0m({BA%9pQcFtiiYQR~Y~LAh)_%_9iAI*&5eY;oXVi52
z?CtGNqC~SCgi3Y)a(;$Z!r3$Lzd1e%-mqCcYCf6-t=*El=*c8~gGGOG8K{6jY`NEh
zp2YpGOv&z>x?az%KxO}LpVq2b8!L~BA?0jS{z3Y--+bPbbT|Lk+PURwMDY*o>~mej
z5wrnDG@wT`7?|h}z>*+9C`o*XrmOp>hijLhoi%vy{q4ER(PqGz$zl-Pu=nmIcD(C&
z{_o3UT=Uwa{q#(e2H5_Sfctoghvr|MmK;F5EX~hRXjFsUeR}WLr<!-{c8K*1xetHg
zEf!Fv<o6j%R)(ky<Ae`R?7ddAtC3a*r{amB7D_2MDc>kQ)4tdMq!nA|mZq#zBjo&S
zvfrb?mYG()U-Oxom<IjZQiB~e<NEz02T7*>?NH5f(!j$ETk<qm^jh;;ktcY*ZtWF&
zn3s61#?pI*OeKs*hC|i`Jstr!Hoy(y`MB_H=9|xPrQtA!(}S^w88;FPb6F4>#@1kR
zJZ!3*EE2d%42A<D8&A^$&zCMwi_b&A-PFc|igEVy9`IPWxH9GYN%jU*B*{#!xs6`k
z^t2;!=R@UW9@BeX(+lB<nnrqr-~Y<AT51>fON=9y(=3)gK&(=vvOg=GSm`MI>6Q=|
zKkZP(r!6#XfK41=0Epq6*!0J{!iT$&?Jt&gH4FZxfA9&cf}d+dL&omXvZJ7gOx;*i
zMgWN`ogo|KGaPXy+TL*($h#L}3w+-_%1%n-4M1ffqGwWV&5-?<x1Q=pm&6eTA{|st
z_<~-1K`H+y)YMUjTwqbRYVX_<PwDHdRGXT{(65ocP-zlPPT30dp8*F(>tRf)Z9e<|
zqF*t@m?<9zL8qgCz^6x;LZb4~vDkk{0=!~$J=GWbjG=H*T3wZSa6d8}mVh>|)oBFV
zx^c|Edn~WU(4{x7Y`^F5hI8MiqSm2CTZ7~C#jX><1+x2aYEa@~z)v@PjCB_~3616*
z<G#+}OYJUXnk+l&xO6@JwNZ0{9liHZ!YkSObN|;uZ|>YK8$D+3W;UL6XUe|@AZd2$
zTseuEV^i@m&RmWgW~{kBJhW3fY#brrOR%iL1{dVV*VS_EU)~|%?Qn{(C#d$RFMOeG
z1~sRvLWVUu9iz1^{_RFoS`6d1?~p7x<AXSeOZWA{iN8v?m5_oW;pl%TQaE5TId)X`
zvY%$_-n73QTh=h?g0F@h{6+JD^f0$=+A&)y`pTA-W^vM?$$&a1`9B@b4$57D_obYr
zKZX$nSQGv4-nO^6^=LDF2!q$VI3)B`|6*g3LrDQK)<bTa>$>@KReBNABj7$Y@#>F{
z>jJ<|`tu)Ger%DmxLH}*ZyqeD?gf2e5SGCZ968z7qnZ5*A&^_g|4QBgEY?o0ba<J?
zn$Ym!KaQBC3gFQq4x@iqqx3p<<ay03!=rr=^s;n*j^S@AWPV?ziLJ{t4p;u^{W3s-
zLX;eKjOx-*lZt-?Z4h8u>7=w8>;7Tjg|8cPI@gj5s1=AI?~s?V{C@F|(k3R@vS;na
z`*;Er;#|TOuOKfEM$ToxmMr*8UV`c8;6vE+zk%Mn)+e2Rj}p%n&xrY0+|hcikd!Pr
zk}nME$TDm}Vr@Qs>;jto@Fj%Z>nz{!c5iBXc1Ha15L4nVs`pZ)_jbYbVZ5sD@n6(g
zDPjC<sRxeI_<!E0IvKK>bLXZ{TfKqT5xN_}J2lt-`Xm{kl#p5A1q0g<$j&S1g;Hw6
zf-6=i<i}jKAp1T#QS;yYDvYo-=S;bD(+0NVfF+sq-sX0U|2m2K4yr2#$oXboLu3SE
z{=m{sT4ZUI#ArP&v+1%r0~we~nwrZ$ePej_14e}v?w%p|!*zRdTa6R(L-M5=)KkOW
zDXed&B6H(;D_|oCfUgyb1rtY%l6-S27~}hBO$iMtE{X>?`!t}19I!t8^FYKt+`G~t
zfh4+5S%i{bd0dr_{C(=%g5AA@s9r{Ce`WFln=RJ!ier#j6p8!VzK96(0VBz)^wWm|
z&~5^Dxls|)qO8*YxBo}1UyJb=eolkfHgQda@_bIG^nL;~f5l@V>&Td^zpyDMOs&u)
zf1`cn#sR%Atj6rLQPz@|vUYJrRB1dtjTudi0u}8?&G63`R<V~NaQ;rGZhgqaP~#g1
z6|CTPqsaq#6)n1%{bIjXhss994S31epCMh^+a8G={OXL~i602cB@nw>W^F#)!+%cK
zBmj4}h&(P$tgyLSfr;9)v(?Udk>0iUDr^5JX<ko7Efk%IIMmeNtUn^Gtir0aN|SDX
zPGr(0A-6lUUAL$xu6WSZ#S=^6^Pea<gcx+oE$o$Fl6nH3e_pW}8FTj_XE$Vk>msld
z(R*>b^nTM!Tmt^pw2oVCW~Nd~5yP;xU~(i;{K-=sw$Mj7*ua@NQ3hMq`$+SuLfRp`
zD5-lR^0$3#wA3pKB`Rh@#?Q*?`j6X>9$#+P&cuN$qTq-5qo0FMbJPbq*tZHrL=b6o
zbvqa86k^t@%NltomLis-UAhbFK2y@`DFq9qq?)Z4pwY@8u}xW!l15B2ibaf)9RnUD
z1ePL;e}h+=jD)lJS;|M*VtDG?HeVg5njv}`Vb&uO3`*>$6EAV-2RA7OZ6lzPX4P)f
z2w?ErAOHHcu6C0;fyD{Al@{?x3e64pv>#|Je4-15xnvh4Yk3QS{-Hrr`Q=E>T%^eD
zT>E^r@jN!QurRczWjz}9hp^wdRZB@*JKbWCW+%HU&^rNrR9Zon87sg;KFL_?`JC?A
zO`8O%m3|bd>%B=Q2)^iP)}fwUmXt<+0nm4;Su<m&q-h{kjZgON6zS>G8YyvHX<U*a
zA@3!R!Z;?FuA1wfd0B}{NyOwWoJi~lzvbSa=68A-pDkUP(ssAMy`T9aaQ7OaCRn{{
zafPt07$|prc5pkB#F9-ylhRbG^`UelV=>bS*U6o;PbH2?(kXcLQMQOF<%I9E62N}(
z6GM~OM!!^3ZcOoSZLf)o&I(l%dn^hY<G0G<>hVf8C58of(EPhj<JT4;SW0$t2C^1P
zLjEG$ig+o{YtDAzgdX9XxJe<K1*6rZMK_Z~^)zb3Bl2Mleao&Ze~Z=2m<eWC3#HGD
z@xavocx03(Q>2mK;`Gk02KS><4Wm!9jgFO2({iT?p)nSiSvAjMuD^iUiY_DUes_X%
zBM0lEHuE$El?sJ06FU|8y!u`J+x0CF7AOn=ZKWZxX3TUJO)RR}C{K6y<Dj;7{k3jH
zC{N}G6x&2sFR#6BOtZ7cH4z^!u7dphf~0;a{lPyK^E+vp+S*QrGY1EmYPt+o3fS5V
z(i<b^jx9!FC;9R;PCg&`ehyiDjVD4W`(qg#{%ySR(q@K>b;m9K7x23?ZQGaM)P+p$
zz2l0vIvH5d!Ban@_RLX#cdwTdm%&n<bFRghSX5X%NBQn}`fkd--eLQE_P;yt_Kt5#
zGN4+AX0LSthcle2aKxP_ahQzkfl+xvTaQZgQ3uysl?tmBn7*$vPV_H&Lf*6Q@3C-F
zRgn*k=0+KOJ>Xa@5P&1prW05zUh(nZ%uDU5K(Qdh$Tkk4SHw?1g>D8ZK)}VNmho0(
zbFwihN(|L$ieRdmp#P%0O4~9`k^M=T*n9SsMvR@Y%TdSoBV*!d$o=L~$IUt>T}SQZ
zGzxOW6_$|b^?80eJY*-zF1BBpHh85aZT=aDND;DkZ1GvkX}lR6e&P(jI-H-#m0SDG
z&P*4oS$SFe{b;)V&!dTJRu<HM@fDHB-vE>{5Pf)&+s026$wl~{tziDlN>waLu`6Eg
zRN>QD;UFRBDeIim%6@z%ZaQ+Hz70d>>moIn5>EqwYoMpJK}XL*x$Ama8q<6!O5>%X
zB|e0`N+RgukMBGG#|Q*PD5>=ZioK#D*87S3zZgzS#~I#&)<u5?!))gmzCk-$aKUjS
zJC2bL&y9l;Gv%o(+)p$nmvWo59zF%6X#EUJNfAj&=<dg_z8iTmslE3FHe5VVr>pM3
zI^nM>vX0oxk7HVFCMQhFaUrEe415fBgmyvy!KP|YUa4#2JHA&Rrk&ZJKerU5vA4=8
zH?GkU5D@U1w|qc`fVH}Swgw!w4q$v^=9Q8NeLlqmX<6AK$7{xDt%FDRPJKs#FHvBO
z-^=(I1Bd_J{s*h^sXq}0A_UpSD`~EgugN|^y-TO=_T_~3W$xd?co<`(v3_0~Z;U7?
zWoqlcRC0I+^ZJA6a<}=qm?>0G1tNsc4uH{1^+=G_%kO~nv#7YZN!?=W==Wyt)|G++
zS}T?hO6+K>lyw3h@;p+<Y+Kh<D_%+SL>zXUJ{)_#qY8RxBskjuMAjUAu4}857Ft!h
ztIqQBOvC~VURDxoq-&c``ih!xW<;Bd=KBQsobS3>w0Cq&3RXKD{VY;ue>swx<=d!l
z5HWwQ9!gC@B`MH}WChIH>b)h;bTUS)r|<onEs-ZM5#$~Aof#%u@t2ofC`vNxSZB^?
zY!u;YDrb8ysqqO5u;<SeI!nfS{{mkr`UN!D#~&_-4rBVDrBAjlLf7xAUN)$${j##*
zXwCLhIn%Pv%2aJuT%ow%yX8qm?Qn7=Y+g7Ol{-~=n1dPzlbf&AiLGi~_uV?LAV@rr
zS-hs4b<SZ4=b?(vn4muB7_(jWdCkYjp{n_xvK)r_J7^pZfDV<_BMxuf2z>kEek>}n
zI10Fwr+YYg-e2Sxh{KnRCPWS2e4pLoggIc2!}U6}OQ@{j2N)eR8>5MpoH|{870q{_
z3lfBU9q+U&BV~v{F6#e1?p<Fu#>U1@OpGK-cyxQLLp)*>4O1Iskqu}HQkIp*K%}lz
z|8Hq&X-x!xM-I%6jh&nEd{P>s$h$ONw0>WXJoA8uN4~{hgg9~|BnP3zm4<B~uX;?k
z6Nd7f$y5?r;6bzI#8V6{R2f!s36Cp64ceKI^F$6|tAH+q*EJ>u#6BpAIo~vSHaWjT
zQ6heP#fPO@iXs_;O*&sQLW70Mp#-t3DKHV0LUaQc0l-=7^Ew@JxRZXTvVs}{wNRxO
zs!_+LCw^DkT$Bb|If=a6^)*t*p%0JIB#~o@{B`;t+5J-EjEV2#o6M~&R8oM+^D^Qh
z$o8-haQx5!oPuxLn<t;l%*ZnG<e3MhDO<+czoCbuI1v$|xUVIGnbJ0qbcG0PVm4{8
z;wZR0>E#rkD?WWv;J-~peNYVhZ<St$&gp<z9r7VtUdnjZ16G^HF-vvraN!Y*G^_S?
zXMq(YJGbV@pkV$~td3hG1i0>tl8k-mkCxo~s>_na@q*23G5F%o*ZO;3aqc);H5N-{
zjhD(l+1(cF7Gu4>zMR^>EGZfMb@OQ7?C<TSzt~Lhg@~|^Y{*gAYC5!1Y(|@C>Y(wb
zr+c6%9X>`??va0ISy)sV%{J!Kg>+%+Tcqv(hEIi!Pc3;>WhEhf>;zA?pCAp)E6Csf
zZ(h|=4MZ5705mcpKT}{_J_rsEgiwB7?m*(7Y0^{)7z5{<S67)#IjHo&ot*}WyuAF1
zG2eHLMS;Vs={0$^0#cmBgZT!|=zphUheeev^GnXU0xQj>o1Mttl1d^2jriVjwD)&^
znUFG~amlgXG5zgIEj4L6^a`A&y|9>DyF`M~{Mw0MWkskaeU^-gVS8%ILXf68z7u#p
zou8N)Z}@g3oa?rDUH$`TIkGR&)+U`qTGg?9PV;cm&j=4CerKmJYp^C@D0{s8p{Tdb
zcljF0P4)lEc5KOev3_@>-?}o%djEHcF&?>-Qb_5jZJb?j<bS1@C~la@DzpC$OIl&l
zLfK8kNE<x~3EwG=W+kC8hXx$QXY%Jx%}-3(vz~P!##*2L8WKz9@i6W!ye6u5%0)ub
zu@!0!Kk}iHUtH&6K=nS|FsfA77)W@(PP@8w+CWzS3zwLN?>@+aTPW&(dsq^+C?Jp-
znv1H2l9xG`i^>XAy}d0q%fcePTz84#Ye*cWSE`~Yf&D#fjLjXopnI5>c<A0hh?T`q
z7_GD@jP|jkpz`*6x5|8Fb5kb)3uf#Z)96QFCBI@wiu}Trc#!I(M4mZw+Vfy9)Rf1@
zat>xoq99RNYH@<a`xuWrSm8b%kUr1I2ml5B*Q@m|UPJ&o+ytAlLH<9Hq7&UPBOdk&
zj(w$>Jc}4{Gn{oPoNpkK6gGx_bauIzXa-!42&RyJD9*d@TdpLjMy6{Iw~qu|+_j*8
z$H$XtVHEQg^<BRHUz6hMKODuelgdE~;;N+HQ{ZJxSkW#IkPASfG70>KQu8{vprJ7r
z<w*abAN0T%`VSc$N6r~<thr(>=6_F|_+Rb1X=Gd@oA%f|-SeMEOgL@V`91Y`){upU
zfC^m+NA(^`WKwu%&%x$BMa#;ycO!B%_}8uZjyB$OnEaqZe!<t%)15Qr_m`Hmx?#i2
z{oc)At`tr97-6EF+!Gi3`@>#qmi*L-BVUg$`SyX*+}SqAm3`PXqAb&-t*grunaPoV
z>8bG3_MLOiFqwfJXN)v@#XLKRlT5dB&6_{J_*03LRov4#;%p(PbIB36M0n${A&Ym>
zOQp06Ue2kes`~Wkv$5o~IB)KEHhoyc6SPvg$y%}%C@d9HvV<<AtqrMmbYJrolr>&m
zT?Lv{j7hB9+r%X#qGxZCkz<ttL6i)67>Fc7&T{ix)9W}^`ZOji^J5uZANk7?!Z}M`
zeHU-m18xdA`efO^NyvJVquQh+XO~%(d*|6;P*dy!6{Ly}(Td?Jqc7H!F1-Ew_u|j3
zWBXFV2w$J8*ByH$h!?BayDE37jUtg%0GP=r(Rg1tr899vNQs`NIe$n(lsQ4lu3zgs
zt)`mq1O3+A+`<_{jID^$@B=G4EyKbB-LZt?W%#)+a6L6d_oPp~`AAT^;eS27L_~BP
zPwRo!Y{Z3Al7G3g#bdLzt4u~lwyw)Ygbm5C`kuBw$S%igV7K?#71D*6Mw~t00wYxm
z_AAGlTz}|GKNY}+#I7^a=0HUkEoh|28jhe4(QFxMItDN{PKrCYv>c|Sh3|J0gav^N
zfKemmG*>`SkX@P7;GOu#W2qaIrPh|WGYD!xT(?U*=|qf*2tB4D@GgB(BJfNsG#mmB
zWqJ9}FFh*EUsU3&G#+>`sc8(bjz|j=ytvdV$B-&NQ2+P!2Mkjpcj{?CMB-tp_a+yB
zpFR0J!;xL4aU`)Mu~?5%E_@yHi~-BOW>9WlFGELGoqam>V-2oTmXI7I;)HCXf$FSf
z#a?uWO|egsOt9XkmaTHg!VP|Ee72Lxc_}f|aVPX=pL^E)wA+OVw-yNZP@Fk6@dFo$
znI^&KHN<6^$JQS;HE8b@LS-3Zu+d3I9Dy8dgAgWA<R|mEAC5jN$mzbWu1?<!nG%jU
zV7v(LNr)D@__>J2$LR2m3i)Cup{;)eANqR!WOVkJ2?t)gwq&44Ya(r$74vvGNV5YD
zd{ks)4VqOtKeDH5li!U`aFfG^3AwnqKZI@7uGhk<)Wjb>Goy-7>40<Pq}I+UN8#l4
zMjy7~PvsSV4}8GS-(}E>nh~F(%IzGHA#MvRE<NpfT#H9u-D~!m$i84txB5lt`?+s}
zLES-Pc1V+x2u*w%tkYPGCcbI3_Gl)O)#q#tIh3tB>Vk*2TrWz4=iaGcLC$)FZq2oo
zSyIVqMcbB7F^m?x2yJ9roIH+=l-R?+#p@dxzsQvOib?_m1(AmJdITG|=r)eA<LFiK
ztMC=h)YaBmTtNxL=+r(e7K$_En{)-39=W$0rAMex(AU4kOUao@<j+01J6yZ|`!^!M
z2R5;=x6rs#!sigzu#6DQQ;pO{zK_L4?kqRQ4o$6HHct6}#&d36Az^%FSKNbK%n^B@
zgAGZHO<MscDEm$Kk@0~vJu@BPUT?rXv2|EdK|V22;nIE6pC90(TSfY>$;H9x)4wkY
z2Y8wXMp#H~Zv4nRG7|S}qS!LhC@ta@>wg0|eb5*}tR?$+#qc>x&4=;2;JuB<L$w!M
zGd=Jp;JT)jqeT5CJ7@BH?a_xew(-pp_!p?&O%0&J&y2!gg;z?F{rQ-1)~~l_rNt@u
zUkTmg@w}SIBCho{Zq#amwTTij*)P-I8fO6}$p;1#-PPO#zKpWO;{pW)Jx3z4lhGPQ
zpHgV3Gt)G~v%;~l1tC~`AC>z1`vV+7C<T`m9dCQuOL|{aLpn)Ad01<%uIFtCdM;YH
zqI!TOB%N5%whnD5tXA-KzkL>EniBg|qIJ;O@2%;TUg0!Rp4ff^b_Mr9VYFoQF(LTB
ztpNZb|9sOLe1u%mNpxS!16T6T_ok0hCpk>fsT6bT^3gQXrbP+H9?!>~{4~<X$#L{|
zSt(ddA0(yd3Ab-|Q`O`L>>zm`%+6>?KL9C;keKviM$0(ths#JG)4!5!YvRG;1)@nB
z7u3+!1C5*g;Se^qP)XSkJoh&h%j6nl2}}8LjX~1W$m675iUbwiAZ9R=U1@H`6Rn>S
z3J&TYl37T#43MCT?xM<khMdNK{cLN^PHQb7OD7ar_tCl;#dt7!nHPWah|IknQ2sEm
z4n^Et404W5)g}uue>@?t=?<j_Bp}ctTsPfCBv;#c*<X-`mhSb2kXTI&s$jHh+*i9C
z8*=hQ<fWVznoq11HhZo-`6^me4NRYF>b(W2I(k-4_>UIs8%-pFZamw&{BHlv-&yj$
ztx|FzfqbBLYJ!*|JWkJnCw1WNx-gH1J9A=wKGnZ9UuF2GpQ0UqzMGPq#xC1^PzE*o
zCl<(WmBps%5hQbKv2)h4tHx*CmcfW~v9selMoEX={q%K^KtWt?i@syadV53BtZ$kV
zqs@Hu*`uSdPIxyiYsJCQo+kLpxPnkMF){Rrw#!#w_`hrCM*R&j38YAf1&NqjEqOWN
zJ*_`>*4EjSAP<y{oK-56UZ4i>Z+8vS)hp0RZ@%BY;6)b~6KdD5)y?I<ds}w1e)pXL
z&BDjYVZ3;~CiN_5itE<Z-@Re2T2*x)bpG%V2jwml7s^r7^7J;#9`ci7*q4QoSfvLH
zcdRk1uBaG)LR%bh2Y4>U^Edq53<W9<DmbIDO?}VHelhPV9Cvs-R3l$*<utCQ)3wv}
zG-Du<9nKjOq#&-$Dg^xO1ziBS#rE`&l+e?z4V_<WkSkmk4QvPQH|laHgzWdgZnH2L
zJKH*rslHQZ7yfrw`s<rsAMJCfPAO@6#7U>>o3}!BGRc+_fd7A;C34dmi6IJ38{P1A
z#Pb6jct#)r0r$@Do&4CAwPZ$xIse;g>$OychQ;uP{fOF5LYbcNi0c*XT8Ua^5QQ_f
z&qmVJavC^Pd8z&%O=lg|<p2Kvfy6|V5z;Y`1}SNf5|A#XQ@W&kz)h*6yBQ$)CI(1K
zjF1onhKPWGbjL{P`t9>O=exgmcFw(X->>_+uIKf9Jo|L6L99?kEw29VZ@keTaKf|6
z1%58Gr;Ws1-ZYQiul*c%&{L5~IWeC=T>)p)8N2U&p0I(03bCBt*v(J2I94^+U@K(E
z;Zl!%!_(w?-o%82_%9|dF2rPk2l9$Mr)PBTw58>#N6rn5Zy(q%y!W%b-Z2!}zV_GZ
z!n&;*PWT4Cc32p_0mtqllqn7@JG}4Gul>^$Utw(t%5o7Dy4IpG{BaiZ)nQ7hIlu^e
z^f@^gsp)3a^ozL!OAu(i!-y?j#PUR(-_0|1?IPyhi(<(ribk{6I{GYRHpdGGJCbZ@
z_d>lGm1y|v8&+Mp%6e&8srxH_?5NO!*De2LAX)J$g<KYxjiVGqF_rKsGIez(Sdx#p
zwOJx3j=SoQlu9y)LhT=e2kScxfLP*BDqA15wKb&;QqLgCRrh(ut~QhG<=*xSommQe
z$Y}^aOU!=<FWlKm=cFY_?IrNC6eUW2uD$at)7|Iz5;zf^c0a91q{>Vt*Zgl>v4pc%
z!~t7Tb=|~>_<1&b?P#;!xO6@H!NYI-Hg|HT;Zq{}>QJy`F1oC}c?K)Y0{xmAU#csI
zd?a%<vn4))aYy#9LtAF{{BoNs=GrXys6wWV-&+JNhcl;9gFl?nmH)e{z4v|d7C3U6
zRO|5&|N8Z7=<)}c$!>kMqOBx?kMa>|_NcwY2|p~@P~W$~qDk`w^^*qOhu-uFmCicI
z44zW`MZRbOf-@P(&8o_)dw>(xz^+=qC%u=~en|;aR(5MTmw6=cf?5tL0CE(-wHGd?
zfrh2fEc~~AuqAIFl)9~3Mm4U%Di5NN3uB=~gBsx78s|QShQ4<C``O7~uDvq7|9Lzs
zst*EbPB8squ{k>Q=nCC$#YTahGrwwV1di8+_V0T7FP$cYnd{g9;42`|*{Lysr@I&L
zhak~!b#>UDoYcqGaV>N3iv`$$CS`2VR`3CAahfxg_486yrCLf!%`WWqr@-XQ;q6;O
zk}Bu)rdhr}N0?LRbzBjR__g4#FO1SOlQ}rwnm-?!u3B$Ibl>f}<!{FJe<O59_Agz}
zcSd~EPnXl3_RZe}rc98UeN92Ag-=YO0S61Z8Ft%+D6=WGigjm$>6$WjyiDhH^Q=*O
z3M-icp6uVD+@*P=ZcH*AvIzcqVGK2StPNUX^Dhea32rpqTv^#xldk8KLxfd@+nT$@
zC0?=-C2-rsJ?fm_wl(%UI~G>;p0kaA9GC*FfsszftH1%WpVXvqvS&jd3+^v#Z;zOO
zF~W<j-V=^b?Rtt~{IlaoOmJ>nLUpcwt5CEqd;wN5UG?pc=n>`sxF`JIDe=YHo+{!o
z`m!6YHt0CcE6PAG2zdcQ!YrX`y8Y`4QI0L)p)bh~a{x_2$Jb%7-H@I-C#UEagb%vg
zzH(1#1M;luU^7{(sd0wT`HCxoC5Q-Gv;i$CHQEq7d2tzhR9bL3yBGq~iltrI)=hb(
zqlY6Wd=~rb(aGgs)2cK{35j<b|IJb=eKbX8q~Ssg)}33;YAA=P#$+edK0xBe^XI*q
zgFJi+q}QEtN)i3mXS-czE?t+oT}R`em~MNzfZO8BD)PtQK%KX1L!q_LUlp_BjiCce
zKlmCv1f9u^aF)n@AZ~7z4#d8~I{hgkUZ+aTZxte&M&Js{hN<`W9v=RyO>RA9MIq<u
z1;Jw9=GC7~+06t*PZLwdO^(C|eaq=38*bT6;S~~mWA$;1<!gMKU`0Yq2d*5H@Cn-3
z3GhNb#u1O{06oV6kHfT?`#SX!ezI4ezKOOz5pd0I48(jm4^pgo{ist*HH$htjFo7C
z^EFO{74!h*%?O&1*8VjzN1(%5YN{0K;UT{DSP}g79pytDrHpt#g5eq0CNuF(vo}b}
z2IrXvicflyNtnNu`DtQyW+$6k9!FXIDgg%T-P)D(*CSH9b#LWEnbWOxKmrAs7(da1
zu{q3{qj1jH=6uc8X^`fsn;RAxr1d+`1Zgn5E+nE1;@vpWYN5;9Jq<d!td^Qh4@0+c
zzl{heUqoPMJ3&urUxy8B4Pi^5edUq!ogluFopZ@@2Z#3F&&Xk3{-9fguxdew?pbzi
zm0$AfXWNbhDgKgu*mkeml;EtHbWkZD?eGTAK#e(O(kq>#tzzD_?Q3iH$mSzF9kK-F
z-ZBgBQo`FOFVGI;yB*1c3z#SVJ8ia6+y64dt8p2Dql>cLiaCabu6rxd+ylnAvKGUZ
zRi2)&M%4<3%mJku@2PI{k2<%{uM1G(&xhBSZwd9_M*&4?o(rPQH1W#42t>n)_vTH5
zLqX_a&-GY-NL#y<aCEko18C(AML0_>*NH-PcC@Lc`rI#VO=!#dHlYycj0rDM7d(&h
zH?BZD*{YmA9OOz|Poiekp3cO0rlnPund#$c&V*Ulz!@Z!Ph0s>KO)xL{1|!`^it6%
z9Xre>mWzNKZBLW~$L05-Un-+Huk_SQvt~HuM<5MM(K6iX@7w;8%6WD8a#g>Q9{A`*
zN=@2p-bj^PM0An!fz+Y+KR+%7_{*4%cV2+IV0PBDT-^Bxy?Ix*xb?beA3@Or>yed%
z5+}u|j?fBc-bXKMcZ=p%SedpW`!6W}%<Ax@e~|-`=FVab8X>OLEmOzJ=oiCYfA<Wy
z%%>-<J9EaUK*FYcO%P7DB$ZTF)fAG>e^~UJOWq4*{ic^vGJ;OCyZ?aeme9?`yB^?x
z->r60Mso1!qeXE16rmOK2Ze8nM9NQC!&nQ`^v;5AFN;c+99kxP>Q)~Xi4ned7O98(
zXWDnhxv5i5BY>gSM&sVh=)MAcTecG8QXRT#QO{5?H*H<GsCY9<Q8$1!aH>S(7OOn2
zF0GPKxPynu;3>Zy7;$N--l=!OZEygdEWpC*aOOWi`drQ<e6^|YRtN-X3kem|crAG<
z-Y1;Ill8Ze4Xc1P;&^TI$OJCHN<^i?MT(y_QkrMyZ&CBFNtj6dKaM&(HU83rrh*3F
zG)(Lw{XPw2$CHTUhT$0>!CiKCoHe-xn0pnsux4Gn`_9kbWa46Fi1G@91yuk7*C3N;
zRyk5yk^WCuMA0lmzkNPN!j(6XRW*V$wg@gHl9`!_(Bf#0`8t4AdD|Tv9ANFmWfcpd
zyQ=2X*Dx5!BPeKg*QdIv7tvE?$Pt)tIj=EXPfFH#g6Im|y!9^?>htjQtV*Aqq4^4N
zeKYBcda>m{i#cBi*~A7m0N8EsmKMW|b?2rXnu$MUPC0jlvP_nimgcGNuWn@|OrK{$
zzKx!qE-$})_<3@tn%gwv)LK`l!61(Y;W+8^G~E};o6OEd<>4*WY}eE<XFBQ-@1GF%
zXN<$9$>il%e|Hgv^SMBc78q*5SzokSSJI`)L@DQ??cDFGbzapJE1>-#<Zz+Vf(zfN
zD7T&hG%@tYlq0^gmBDh}tW0Uwk1Wr8KEWMM9o5m9&B(|oY3N|=<?StjNchzIPCs3>
zvZjW=GAQsxnOm);!MbB(Z6lU!VxT6WO-T6qe_=`a|E{HmFEzHSJv~At78qnA3X-Z7
z<oXJq;!)~B>jMsMflswsWqjQn1^J6z_c$|>Z3$qAULkS^5}`{=x?6d?z=a&Hw+%Dg
zEML68J1>w0Y8q>Hef@p@{72}%vQJPr^Y>}NgC$bWxedmHhYNq)x<xAzgBm+Vk1Tty
zJfk1)et<*h4F7zMSnk4BT7Mavf&2V!pYj-pF>JVhWXm=HI0(HxR;rq)sltDtR<S}W
zOR3eYH2>#<Tz#|{YiIo%4+_yww`+2BZQ#!295mqp;plSV=!xi`T=Ya*UZ-O*4U6vU
zmq$mj>t3x`vX1G9OwiT#V(>X^YD!8f^3u6X2(jISC1nSAMagreXDO-;;o(s$2t+?=
zkb|JLv=)FF8cCw?kh+XmIYrG0MIog=tNUTG(A9Ctx&DHYz3Dhw3PH-=kHwK-1*kwF
zh!q01raSUH;Rgvl6I52R#<Pb{&A13aqO(Hak_0Ad3~~>PjfdtgFIF}-yri>{KR#xc
zSTR!@Rf4}1KB8YPGlD|lM1Sg<%W$AXUy(zLG{AHJ*VrN#^yLLEMCZRh6OY2woS93{
zZ9{bQ{3Tz40{(4{OC5AHNNw>)Ij$@5jIn~Epg1PRzYdx|M&`FPSMCNinQ3vPIVgZP
zk^VJK&WmY(h2Ktpu6e4%Eg(<q`<zqpNtRrap5y5UmXK=kH~9_-EeD}&dPYX>QiG~V
zD*-@dqN8T_Ke#Y8adi{YfJZguKjq7dBO4H>iBr3Il$??-C!4?jr-nJtl#&sjfI0`n
z<k0rzB^xQ|Q=;tE7gh4lI_iO;o|BknhlbhoSXNcb&fwN||FEatMm!VECeLj&EGTf^
z32>s(V;_<ftX#Q48(QrksErR;nEqc*+e4s7EZ&8-{IP9CXq}kkpZTWGs$`Gp$&h(5
zncTJ&s^qKLYNpty(H;|?wlX=KiS2M2?U@D8fr0sAzTS=geQWXxU5gw<4jf|<wI7Nq
zEr{68&d!fnNj1nm&ny;bbNtFOD4!P|?H_H@`+4B|qrL-$dPmkU;~XS4G#4#s-wK$?
z1RP>3MiPQZVG&ScdpmhJ0_)yr!e+;g&-F>1?@eqvWqg!VtE{Ym-7c?hZ?8|Lb>ijY
z9@Fi*VCh$a#d?LPI>t7Xb)%)3c{LhspE}cFXPa*~S@~8?1rqN~HoYuW^SI-Dcyj0b
z{*VMa!(-7|jc^FUSRdJk_?|cd<ZIbtO|6=b-~8nY{l@U_ALiY6F}yAV1r)3c`M}fY
zAiEd`pR`YCo}q6kKDBOn?r>w*B9(j}@0Pk=EL8_W#zD0in8PR?dhE|H6C%7-lcA%}
zRy2~@SmWX)!yXF%Mwm$VcvD#O`ZhG9bm6+pDcEC5gt#Q$k(N+J%}MzQ&?DI6;cz(W
z9(;mrYO#TjWJk-rgh+!U?6iiEe@ttjWny?(eeUh%s59_WB49Haq@(+dR6<jKLRbL`
zF?Dz-UTI2G{HB<y6oqOv)^Z3erwcy(vc|~<gBF2FLc_D_N_D-YCu>C@qH9+Gw7y#V
zo_5#5pIXPvS5UEZB_5zSG`WQwJrQjr0HAtRQ>;`3J-DPv_@q|gKDEzNze*z3JDmA8
z%9&(k?=CK54XLOm^l9)SQof!kXTl>SgsyNvMdBXx)JMO*yZsy#CiNGO^(P18y;XQf
zG2R^09F(U3wr5sMDZ!qHvL~nRe`O`r$jvjWW^9A&WF_1D!N7tOGKF>MSL6H-)4&|g
zAyyc2#iKEo^C=xy5aQdG(Hkz;iLdNUN(OdLQX<3B<lsn)7Q?kI20{`ImADkP2(sfx
z|4~FLAhUVZ@fG|DF}X84R*itK-+=jYm<rc&1+NR&IvUOQ!u!xTV_vq6s)7_f*O8uS
zHx#Akh)hBB00L@BhWbk3uOr=xAf%QkNkXgkO8M?P6sn=&gh@7J0ps%#;Wxf*JjZL_
zpvA)!e-=01Fci>a)-3rFQ6eb_53UoFmI_5Pf=WqhJd-xIt-owxo|F%(woLz!!gyKa
zYqLx0M;B^22hB&Y=|D!boQSEr8{H_&oif0xoWaQe+q7Yes*JO1@neYhIEIBVdM$@t
zokDGL-)CpCV-E9sif*2fv%zTO$2xbqg12K+TVEl;>=aKAv!*_eVbjy_B={%TxwiJ0
z6zb0L;OI2q8|cz!!P(ye72IlKV&w~yI$apsNWJvATN_H<opUOmH=bYM#>4YldxK9U
z=I|UqoWkaJb8|y$ftUHL3XO9^rUx>2YpaR1>{Yj@JciPTTTT07QF31(&fu__hp*#9
zc9+L7{T%3T6Xh+I_dAi1DAe$Yx2=co8(Rsi5bDibs<WkVu}lJ7WZrMrayy{T<Ce0;
z>lVPBi<)f*YX`NjPAH2PSwuZJ)W1Ea1c2{fB=;4;o?_a6L+2^oKWi4VyeO87peC^x
z^R9JBkP;d!F#WhVd9ct)(Vz=msdP0^Pi2F>h8cbzv$<Bur&>KS0%J)C*{!_8Un}9E
z)v77PpL(~VsXw%*>$#j-zt%8N?F}2u%JTUtbv4C*|Ks2e?as8%H24@ugm)fK)&lyX
z(!d=z-mwgIssB|73rpFo9w>2;+E{lDUj?N2a3PS{_OQB%VoL#Bjt$YdE#-g4f`721
z>=F+}ok#y(cRrnXh1-4R)hF_(D|`H&HS6T+xR)S^oKuu2*+nWKTqW6vYl?NY$^Izk
zCk0biz7bgbqXTnP8WisxI4nvb2Z+*Ki89eYbC!IG>bSkQJZdm>Fl-3y9jVI*Bh5uT
zVzxCSgVw~8FP(x0;yFqKyaU|tY5)=3m@V8U2{qD5;p_*=swua<#qM`E{KNzyHg7%r
zyu2sT6Mo5B?lm}aTovg*>RHvBKxA?ix{+>DB5l=b2d^wBIDb0M@A_5M1Fkziq~<7Q
zxV483nDLvBs|UU1nFB~D!d8Vi4HQCyN(|Ahtm??qX|Yx8#)vrJ6?8RuF7vgub=)Z;
zEHGbc^X%*ln+*&@<+o^lsp#{QL=@?pnOkB36p#qWFY44LuhQSPLecRLjWa0v2Tc|a
ze6bLjek=b!7drY?J%v=jw&PUm+#HsnWxi?V)0H=xp?0|MkwuBT4A$z_bAkRck#IRD
zQJ&Z~b3vxnH~&rU6;`dRe1hZ9@F9f=676`fq?7A2`KPj4TT7b>bxP^%?R)7gVfkbq
z<IFbk`_ISMu=eNtvbX<IrTKkKEiBA2Pp<G}<de%Dg7HC4)^HWTW&Kr>9wz=wE7!2R
z)}k|YbB)eEJ-ZQ6mU=X~u+THP9?0E?Mng#5%Ni_NWOJYI%La)5nZG_t1a3}if!npc
zo35)5mN#qL9M{*y<=V2_e(lE2u6DWg^_n8j=C;d7_A!~IyN~@6EE(JD(;^M>%VDtM
zB2G%XbRKhqmT-W1Y_5QzYnGvtuGIg)%m8Nu*GNU4jN&1ehr56IylCeaj@p{etA+7M
z*r$_ijH}S=4Wk5fEOzUzz7f0BJnnyYCB+9JC#s;9!-0_fysCl7zT@iXeBZqKez>E3
zRrgf+*JxoHk%!pN6VbO_1A3y#w4kB~n`d`__qorOUC9B#KT8S)rsS>@!X3pwE-$%o
zsI$n~fSZD@^CnBIPv!8D<^6==7>$t&`AKJGrS+qCX)dED9p;TJFbl3$ljkSWgh?Zc
z;U-G2RmQm`|5pVX=|JrniT%2cnrd%b?ne(g@AqK{15%!F&`-CP>!ug%uiiHMH8;bq
zvHyjX3y}RFQ0*iN9~`bI$46m2%t<~NA(s>({pYt6!aQzMjB1mFN!IQBYASG(b^{mx
zJNsFgQ>e4zpZw+Nk;})mUJvx$U8>y85~h(~6jYe4sYpl#NfqI*;P30THCXI_hJm3h
zEo5Tifg?K)&6X6^Ka*JzZ{ey4>#cvtEUWCD8P&d69n%WExtSbps2ZwwhWi?XL2&*^
zaeyE+P{-mT>Uq(h69b^lQ#HHm2eMtr-j_8^Tk<i!r<UB?9p?*^&7^7=hCN_4cTtfG
z5+1=5ld0Q*OAWASBb%ZH!#T(52|ryo%v784nZLQ@tw<?Bzu@}m$cArX3OkTUg6K|?
zhuwY7K97HHVl@#-NJ4p{pHucc&1A4D*X00Z{bG+Kr!wYmE_RAOw&`u(+1be$udI6l
z!Y_&TYd%|WVT&^WK}+eMe8T2K#2Ap-H7WE7k9J;;761(k5=0Z*jdK{u&4E=x<NQ&3
z1@LpFqRHju9xsXA<`b>*Rou7nnkV#8gTEBNkblNuC8*L|L5<c|n?0u;Ll+C89ba@Y
z<Q9Ex>8efRrgKTz^D{Xkgw2+;q+%Q2A1Nj24zCGe#zh%AbQ}Z2b}1#AnAB#cmXo<5
z%y`eB0Kiom6*Dl^tWUv*hp<Pt>cw;nQl<8<Uk0X>53r`#<!>}&`45wK3#i>Aqij5g
zf78sO)Xl|eIYO)4?{2*KyXHj4>6Gc(mIJWiVmO_-cr{fEUGWf*mJJCE^xbe3L70WC
zZg&3KZ9Z4VoSY25Yf(p>*>-ZyA_v+w3tkQbrifhZ4;MD6HcSNsx0#6;k15Xn*r+Fh
zU;1c#eFdS%TiO55KhV<#tMkV;4mM4pMaXo)pUEeTTq5r$?hhSmwY1+K%3KXWm8J6)
zK%u=FSqDGK4+h|Hvu6*WLJGeTiDy*F<P6`gx}zw9Ip*0}j>zk)+zZA0JMqOtpVb0v
zZepO_`M?#o;KUPhrP^(=b0(%bIVi6CCRO=^r+PT&OpFT)Bc*AvGxTu$!j=SVk!fIV
zfUCpx0eo^PP1KG(KFgd|^j_knnCu^5_<w!7E{yt$-CvV(@0xtyG9>kcBCj^5%_-+H
zIPmzG4Z_5*H&(WqeyCLWRFCJ)9y9QsD&AjGv-+``>Suu{o$HvvVu~kVaF`C)Z=y&N
z{(M>3eU)rgy4S+1LuN&`pIu@}w-4qBN@wy9#ViNJ7KcrDzOxrojbFgCX;fIC;0WIx
zzUDhN|0}naf<+55`@?*SXP+ssgJ?tNXBv89%X?QF|En3ck7i~j|M}^PMgi5toD5mT
zj_s6s51G?rJZwksvLu?V`z(;tO&Ma}NrL%`9jvTah!i-})E+8gk)3E!wDGP+imeh~
z+VKL;3*1Ls*|u=s^(Ibs*y5iupXTEQWjY^a+vDzhc{SVTJy;g~kg4e;TC`_Mg;Bna
z1NTXD^xI8M7mGB={r2>megJ>iM{=}b%12T}A@S_2ELH-V#p)T4@sTpl7~MSSw;p$H
zvM0|ib283Q<3gS~OYk4_iq^5Jl3fpefS_C5T1=;#PC6%@MKoWnA-spz>*q1x?o9p!
zDMW&9+vGkU71mJ3uon**r+y#D&d%JeQrg5lT<=VKEF%^m{8#BIKpBfKAi3G}r1J|q
z7rNG+uy=fBh(-wh(~-D_TQdjZxvEf=nqOckzFEtUF40lSf4g3kerlrh1NxoEMZ*gw
zwR_%XoWW<GI%HeFc%X@8JuQ5O6iuL`rE&kX@tR<MV%XH|X8$JeUq@z$R)%=2?Ld=*
zmohoZdzYW!@KFu}*w&I;QZ;2h&b@*6#ekW`e2G)gO9m!JM*4XvY?2IlX~|p1Uu$%W
z7Xu{s`-Hwt!M=X)j^(Qm=U1=X)j#TpjvgI7l9NE7Gwn_<dQ$6istrzW3AG`Z$@Q<f
z98T5&uaf1G!i#DiJ$e|e11(exJ@mMrIjJIB{%38GF|2P6;Fpy}Z_HA_@32n#AqW0M
zbKxubHOKXVFFXQ7;U}WQoewJiGC1z;%g3pQGPAVWy%g)1$s?0(M%Q{q&UHvQua36f
zzVm#xVYD0j;)k|^*xco6&CcF@)?F*d^ucRof(J_(siOK_hoiMk={K@>=l9;C*|ZhW
z(P1w?h(MU}fCq}Qyr2Y|&?Xaeb4zC=-eA8g%ZKKka;I50e?}~cZi+u-S1k0TVDI28
zJx$IhBt)WM4B5m7Aw3P6cMra;TFq^!4c<l``ELBk&It8_O}ea4uX<KZv=|6%4G`jk
z7YL^v-)B`ECA`x42DRdu7`2u_p=6|_W@N>MReuQs@ua`D?o&ALCkk^hfp>qeNyVVV
zD^pr{;z4ib12w@V(?4aq+tx{1`f?vzk-?E1%PM4el)S=V^Qp5Pzq6b5s`3Jc0y9m@
z$6_P$oD@GugNXQaph>v$U}jsZ*`Ko025EG!xZ~9N2Lt?{f^!!`9qK?uRx5R9*;uZN
z=VMNwiI(`&ErPiaWOlxTg_aZH!&2GzxQ)mYABN#Y=;7)rhZ%w)xJ?we<Bn{k__#vJ
zEfn+~kU?C9j?of`yg+f1T5(dJ+<jQ$`KIX8hucTz2mAXek9^IU7#R?jd6{;B_lJ&>
z1x<4pgQ}<A*tF7lBe$5n0OmehX|0LINoU~s#)+^dokMXQ#fwb4?dsRvm9EuI3yWB{
z=7b>@VSFvOF#9q4`nH%wyL<wxZz&JW&3We2u^*sbROp&SB<26N<u|U+sT0#DTKyF5
zP{ne9)shp?(yRUJ0ioW{!zd(F<8^<ic!t25oYhczz7l}%@$tekV-aV4$+U~01i&&b
zoPLg7P+`|kkC&i_LSIlLVW@E*Ng4+FB>!8){CcJEtoWQGvRwbgmT!DSLu%&rb?@*=
zkig$m?gsS3hg4Klgs;!*c^{M(l@qd<xTH@Vbey0oUH2D(E!6v`NdAz+b?t(eA%Vv%
z>Z#bv=SW=COkMfsd@SGm#>A`!aP69Q%USQ~>3rlSW7#-|tZ!Bru3M{jHp!;d&M<)=
zS|xHS($LtKLDQ!cEx->)lE1Jd5kzX`MC+?^Ok;CWU-AsIy>LOs3;No0-D0WgH~q1*
zCS-%8IEtjfczrfH>(8|bj!r7X{<c2fm%Hdr{!TwbmHUVGUfF!;{Z1bQ$h^3rtubD-
z&86eFE`?lwOUmBoV_(wZqjXndqX4<S5Vuc=rk;;C^S{W%s|H0h=^dWmuNH)^4kwx-
zrugrQE$=2|FKrgTzQ-hpX6PXes^<1bu&v$HM-PQUif<C<Tg&gt^&3qvL$)cK{e@j8
zleM9z1$PsRBPsv-PaFb)e6@<<dNaPN&@MxL?8k|lDz+b8c)GKmIK2M;-Pf8jL`F?z
zYm%2SmK&%oh0?8^bC2j>s{ajQCd%G4fqH<Vb3Wu`P8QW9SawjyD~}8WCZe$d)1f@@
z1^oxJZ7kM-T!$<~xh-6Dp9v}6uxWj$<EuK}g?6E&!@A`(oaxfw&Bbxd$L<0|gGr}-
z2K7(~XVZfq9F+h6EeLdEk0e+kUS98A{9Exd2ol0~j4T8z9rh5$+~l0@^+2QrBa(2b
z6~JHU4Ae+pZjlOB?)_!xo|_ywgMUb(&*~_DKJ1F@K*nUs4d7A%CJKFwq&ndDX*%CN
zK~BX>DKuEd)rcyZ2sYxQDOFjMprCM@SoqHY0X@*$0g#Q+Fa?0KWiMli^D5s7saZ8?
zfu;-`H{r{i_wx^xdiT)+n_u?N?F_CXRyzKXnKv`tHzV@(l0j8GC;U%fqKdEiXxSS@
z-r@Q$O``QRJOtPw()AK$Y-s}r-rs|BWK#hu01E~yMxju<!^3TE)eTC5S%1;-L$ip`
zj4b`95l!;p*8JgQh-YAXU9ld?Eg$bqSCN8JWBWL;d<$%59}OBJILKvJNevjdboCUc
zotK>@S!?PGYSpXA=C6#+&2H#0(pPh6Issn|nYxFZ?Ew{$43yUQOLf&dEEvI8EpzRi
zolZ@rJ#&6Mrsz#<DvkKY&}qL~m^*jI(8BU7u17NyzD;_w)jP9Uzrla!xADKtR0^?g
z`w(C{PL^K-*Lw|B*e5FVeBK|p=sg%Jx)^`W^vuS+?tE`@@--7v*ZJMpyGn!3FEtqv
z)Q*i})d2S$_8VNtIaKKQ&dnuo^T;}jpUyr_p`VhRj74{ruzj?@pOyiC>|LX5aA2@V
zr@59pd$yL{fZ&V?Zxspf?-d38`nOe07Sac}q|8YvgEuDq>uxL^HSZU6Bclc7L45<P
zq!j^zT?_V}9R{M~YJ{6ylu4ewN$-@u%y{xj0`J@3Lh#kVQSgX;Jq05$DZGRt&heKb
z*LEKH6EhFUqhM;3ERN_?n?EUUaDIU_T%YoEpRDe7-JxG^au<}MEbrohGtB*U;!Y(Q
z2ra^Ld|2|+=@`~^4g>C(?g|V!_;=fs?QP=Zj`;8Tfpg1yw7Z~~=48NT3M)@c5$7z>
zy>fAJVR<_Ym?Jm3&JqC;cPgjx?8(#1{tCIf9{)Sn;7dIL3LU4%M5>Qmb7+(5e;Y-x
z3-j}7FR!fd9CVeJvKFzbn$@YPuq#LRm5^sU{18p|ibtv;o^lFHDA+S|9#UD)gZdN}
zU3l|lm}bN_pHNRLryu3HWo5#_VH&_7P7)rRbDd@^aLuUZ{7ZNmb<q>rx^8GuHLd`B
z;A((?tz?Spp|P=lP~geU`p|Gg?5xmdR$K}qBO$0@q0aaZ>MU9++TCp9|5OHEGYt~r
zf}X#%CbJ<f@<2~fj1j+hlgDC}YXfBo8|Q|KFpJvbMCkN!+NNwMwcl<nuSNpBFn>7>
zW*uCzBE=+T4`>EAGejndW2qaG#U&6}5=IdgO(PNDPFJSFKsKlbB`a!rOj;bKSpJ_;
zL=rbM3%koLm0snW;`s+GZEYRDC6T)yKjiUeVv3ZQ<NVGm+pvndu5k@@gy6PUD?~*m
z3?jGp0hDTs?69(@D4O3EcL0XHI!*YlZ_$^Rty7iNy<V+9l_z~)&d#06&@(JND;f&+
zyE(=V`v-rmt^NDgb9;?)tL3ZGmni<ZdwLzdwYSGRO2h(QxRh)pN~Io88)g}+FXA-m
z$Lxu^ADp*iV9x}zElVHnGoZ}ZcuU?Y(3Zxsw3H(LF33o?EviKs$OBN8aEC0-T_p_r
z#ClP+ot|T{yPF#y|Mhh{3NioA*f22PK9CW9E!CcEoyMx#K54<nu2fYQpFX>XVTj|m
z3W7cVUDD9-qO~)P`}r(lJ$Pp4><W?bJ4i5l!)fE8W{Mx-t1@`^<XUd-QM$p<>T7of
zCZ>bEJ(bT`8>eERH{kt(UOQXgobMVn)sgpQu)d=uD{(RCUj_%-)V$>I0)xe6?32J9
zXLMRhcY-S7G8D*Qie{0UU?QVp$bVTLjZ)H~p-Pk`VGuK6*H>(^H{*#263ta<{L-`(
zp|cd%Uaz=G(w0_rU3jl!Y?ymyw@Z0!0Wq;ZH0{E@aWFa>2#KPbwsy{KBogxsoa-F)
zR|&<7%Y7EjECpdR-ib6+Rz<F~y?ejb$A~8oeSw-Qj{meKm=qpHFFlX0lD)0!nz=8v
z)O_+}U@ZSqk?^>t!nY{WUqJy>nL~tkM{>Uk++(e5+n7HlF1Z`nw{a%6$99kKkIr}+
zn7Y-*SAGIylLZM-QzEBYLQf6dYWVJJfz#{whJ)ZkHbhoKs-Z8cuAxDf|Akt5gI;VQ
zo?%t}{E$tg1GJ@U(b8D3(>VCwa$?ub!dz_GS<i#i2UY2MYM&t~ttuKw=V@UM{zPD7
zX^@a8Jkq-dd^FKt3&LrH491NFA#+Nza6aJmsbmh1;4<R4WbTaMvIB7(98}YpV&v*@
zp=7m3PSVl;8Tpi(CxYPNoT2Za;VJS05K<g)g+|6xPqj(|U$6!Et)3c=;Ajh4jVo1=
zN~xx?0xPztYsV?5K*FIA6*~f{;^Jb$_Fk08MWeliGn7)H9BW0H=RF`wj}YK|^WMvF
zx(Bu>9<HF&pR^<fG%_ww7q}|XDp=vD6}{EsRIm;E2!`WAxo<xhiD7%G#RJ*kf7pAT
z7^GZ$yY3YYmXea92ZIc-SFNW)6Y)aXg)O_HhK@@=g%jypH5XmlT+i#!;N+pNKPUGn
z)mC4hD^=NMt2@A33L9pKzU~-2@SRO$dfC?5>DQdWIRLDBc1<8Z)BYpGdcg>!5L%{c
zA_cYZ|K@gMto%PpMJphznsdsPDYDNLDYh>9FMZlBX2lj}X#7I|703$a;iaER24Zq8
z)TM9#1pxkzKg-lMjTP3S3kA4{=y3OZ=1>Sk)zOFvPkIXbJLa&@6Z>!VMGD5w&q=Z*
zb`r4-1|Y1NV^=ZW^Kt3(G=tv)10pZfATxFdQ#4_XbTOUAxLFdIj=5<%hef(nYPpw7
zQI8InN&U&zpd?{@8GL%&>iz@Up#Z+2p7(blps%f+zUX-6;@sTbhE-Xtk79y}3Ah}V
z<&bP2Al^xOLZVSKS-;FXS~bZc(x3W0G0`#<OZ%NMqLJV%p-`-OS8-7Jd?1{k6!PEn
zdmIo;(uqcA4PEK)*s+BBow>z|yar}a*AoKMFsXzRB>%48!%(rk0Zu@S@<e&ipF#fc
zfF@zLrN@Lz=TWO)bfkzE9o2F$Aw{Q@exF%VSieg*0zYOcIL9H`5n^@-g8!H1<?@9Z
z_zF`oW&QD;iK!K;U~Xo9e|3EToUdvx1n%nU-TyuBx}Ih>DuQ@#cCU3{746RZwf`?9
zyFd2#>p&h0zuzg#i;wj7MYYi{-gGgAUSN&E|NQ-!{sI@f_ZI~efeZWo)h=geBzOA<
zq5pcZ?z?&u`TxkCoqfVu<cu?DS%FP6F7nk#b&XD5!zo<w4yAFjOMg9IZ~2%E>>Ber
z0y<GaLdh(t0i#lI*2s5wJTw_>fp!QS|FR&ai6oD%27*cwmDMTWDjj7N0^lBaxQ#d0
zBOKjCAxg6k@FhAyFXDs&&Lz18hv+j=kbI-a=2jVFIEfrH1r8z9DeZMLum>(w(X2zy
zs&SRFT3L-*MK<8jQKqi%*b5=hltS)2#c5hb&a76iUDSh&h8Mnne$_+v1_W;m>VXWB
z_HA_hSs?03^|s)N*6Xj=vx08wy>COFIl3EiN-!pJq<$fR>Z()psYEHW^&(L7^S5s0
zRO`6m52?f1!YC9<ofu;Pw;W$E7U-7jJi($)zhk9JSE?&W@FXgxRIZd27mo-Hb1}C;
zkVIkc;38EsUR(jG&2XF#6lx*Nj3Is6?5RVy*XLts^l`@|#*bt5%ecL}zc6h)woXTQ
z?<wdbXE6r)>1q|dy~Q)G8LUIrtzSDjX`g&rxYlnqsm{<e=LSmM_-cbt`9og)W(K2P
z`YbGh7TEbi_PZsEaZ$Cd(=9>yfTNp+%V3KV<D4>uhz8NI*}$oaO6R7^>PN%tK0ncw
z_4|81?hYrnx8ttO)c+*OJlAjPE16`1-Zma*^5+v8l05N?)kv>F*H`;Ayp8x+__HjU
z6gTe^>nI&nnkaxsph<K`dY_Y!)Vfin&dmgWp~d*XTPv#!@<|*1P#3x&Gl%W+H}Pbm
ze!g$XX!u3AMdHU)GE+r0mf2t`H}m`HLz5Z<GQlx#&=!W+H)UjVwP*s<^ayFUnt}A&
zjU<uz+Sze|_P^Ufo_97jG<bV?;es|dH`{x$OoonIao}Cid4Y83(edRcrkAi?3^u})
z@Yw_%@f)7#2#_0XL0PNXG1JSJ`pQ|~+fl5l@vaJgl#RI9ydhvURZ?(v8i)nl6L@eU
z{ZdBacKUu}QDf74j2`sH1|ry8>qdSuc0}4Dc+RJn2p1$zybnzPE?c*k`7Rn}#_UD9
z>5XZ2ftLdA<#3EYDtzgI>s<YcBy_;XhWJJM0J~aMkw-imi=FUC&~lTtqFG^eVgXx}
zN*{qc!_kiAo%_9uUgLK=R2%9iGw1t^*bfFVr0H^tQ4J?AHqDGP#1>T7HZwO@cd;vr
zGA=(VD}S_B0T=(Wou2a7*2vskq{?14$kILe2>cJ&<Kg2WKz$0F7u-*|2DN8tMbu`T
z%w4W2K$B~)=8H^Pm#qY{M>2+gwONr(qU-yu?KKvLHHuSzOz~@>7o{XAOdb~gCjw0q
zz>O}{Uz_w<jw~=UUrU_TBckePG5`dTxl4!!w5aB9{C1e#G*8{BvE{VxbWSR1v@s!(
z2Zf$nPs`W480}2r-?$=xTZN$B4(se10Xv(x^*?1N`!n<dB7$*YKdnax2p8VKn5NaG
z;KRQ0)`x|QyE0#ae@i{Ljy|xP;}54Yx_aFGa1#!sfrEoU9HuCjxgR2pM?wJ=5cmk1
z`k;@A^J^;ZRsa{LGpDPtMWevD@&e2tS{u;MRvrCxU+fT`f(^+2FdqGjfQA!~dC#Ph
z<X}tvrZ=~(tqr@ji}8bfF|svoHX7M&3*8(t#gZ(sUKtDqJA_*IG8UsG<>%+ehhy6~
zFa7Cs{a+g84R);ozFuUz@d{ur?9Rj^=A8k_&5|ilp$0B!%7odf(`<I2{#4(%x!#Ob
zNONM`(Z@Rw@I5%(y4r9S*m|B><#@=6r{%DEG+Mecu2ue`)qGGmSCdNFDW!SBZH&L>
z*q|)l0!`Hubc{!>sK>jZ?_@Y^=UoT<+1)@XsNe)vtr|2x5ip%JHh9}yjhK@a*O)`u
zVM+w6+FHfCs4>kGzxz0KrhRoA?HjPpPpq0y$&CQYuz)U}b;{V>Gp*NjC;Nn!?<*&R
z{o?n^Vq+8_gPWUFQc}0ORk0nD3o}wtR)}<c)Q+#_gidXP<s=^$yPYwjz>-2KnRN&o
zb^`c%JL_e0GTrbVR9FJ%ju)zzpYHd^7VdtHg=uvDTQKD3cW+R%%rA)hK7F!XvUrfX
zB@IY*3Y%a->XOk(F$b)xVS{A4$`Yh&2I@$)gFN+=lKM0L`(yshDmJz7xD}D1=eU)@
z{;zI2tj_1dU{nvnIR)yM|F||0S$w~GJf$5D(N!bOe1wLju@JG;B+43i7L%cK9|T2G
z;e(({gXE7XI!+Ye^}J06YCoh$t19DG$$_HM-(|GKbe@+N57-LvNa4xyEcM=14XOCA
z5(XZgO@(fw&TIJ4Rd!q@aujM6-$|HmF^hMY@5<9v*-)>^n2tBNB5F^BJ3Ev0B-xqu
zK_H^cX{xUKPXIKG5Mj`wGFy?KP|LAOSFE4E9oiD<if}oW#sLqo>SO0>Rk?Qh04_lv
z<j4B+3VOV_%_=W`Fx3ZUoUqcQFa?@Y9KGj5A6GZW682x}u`{ngID-X0evIZh_Z9my
zOCzxG<V*dn39joKro?b2vHp`SDPp~P2u>12tWO;$iZ>efMP2NCv|3$qcT7BMe1gK`
zd;ue3vJVsy$$tow!pPDgPHw+6cz_Qhe8c~&gpD$qUFi$ih^8c2><z$V5b+C)zx(&N
z6e3a@E2POD{R1cD+qY%x_E1HY6I*xp^YP+O+WFEsXnO}Lr{B+_86Ps^F)4r<MQxeO
zG$o9Q<Q|4ocS{^ZIP|fqg2&0jAqp;^g~2rNkR>^26jTnEI1UdtDurGBDZtaqmqhrF
zRe}7c7$+XNk_5W_ZA1atRw}FK6hV<+I|E%=dP?dx>zFr7j#jy9O29$m)c5ZN55p&h
zTt#yi+x=^+D<>=S(DrDtqhlAOwzhV*f;E{7owZ*A9E2}`L!|Pix0~<l??E1aPwc6O
zYDG-P-lA?-S69y$LLEf2f&IO`K8cxYXsH78)$Q5j`po$FyEmqE|Je|^Tw+5myXJ_p
zr%h^t$p^JK9(ld{$xyc>8Kt!9UgJJ^J?^1hp~#`<WH!kc_BH$YgP#7XiQ%eYY3*vH
zlnLtgY>wTjATN(&i}B^(hEt-(5xqREa)m)`wf^7UN4Vx{x2RH48o>I#@APyV3-wz3
zEga<I;hSDovB$e3tqvT{AM7u}e!d;HXg8`h2-(bYX?&|`uC*yPe>Qme6Z1s*`t`~7
zsE_1h=C^hYM!X-}pP7!DM@FL&xdgQLBbgs%FTy3IV4{Bl4F_!-#eAB72zh%3iqaE5
zc|zjjoR3<%|2OY&dJUW*KD|basB_t<#)<uWjQa?WO|fN;uS&IQf~#r=%f-9=H{<=9
z**R<#;q|L5tj(NKpe8<b0p2hNr{QvujUEf;yS}3@Y4{Lgt?R2^7ihCr<I-MWic`H<
z_gJ!zm>F6GfrM*rD%Q@KKRDMFAdEiPV-nl_c)I-19!k2bO9|jPvWkY&GIyvl$pjiU
z#n#d@-j251S5EKl@9(|!J>B`YHdC&htk7=GiW?)$fFY*Ff5b)%;@*_zsUYtq&d(!r
zH2+D4=1+K6UMgiQ(M9{XJ@u&3L|?h2TjqlOz}LP(DQ9D(=Tgn(123n50Qe!ViW7d}
zOHaF8mN7hqC<KrN!sBGqE6y26A69BIX-w&R5y(V_xcLpQL)H7Yh(C6Zj+PBfKSEO)
zbN~02;wO9Gf4`r705OMyqO78c<#1JTxmdBaoeK?X5K#ngs`pZ|{Oc`~$fW<|BGTqV
zVFG);H8DRv@=n8GQUNcRp-gzm#A~mt%IY?7-$%ddi*f;|aQZ|TI~gsy;jY%|WI`8B
zjf&&key@8YBeAzBYQXrtPev)4P!9S;F}c^~{TnHVND@6of6s-VIxT@jg>t17f-n|^
zH0&-IuicUyc4bASUyqDvOodaICUpb675#zZa7{Ppo17v9J{U-)v<q$4xNX$*jn!~j
zl`d6^#Ras!XPzfKDQlSyTnM^YI;!N})oCF_4B4*WnAzB+XlFb%TTrC>fmluo)}(N&
zsjWeek%xy<2O~kpEWT)wKD%ww@Q0PZaB*o{D-}|<hPO39f7@FeBAGA;C`sA*xtaHs
zLc^(k?M4|qE+P4wC)<};A7@E!_L+x+jOCVe7jUswjIuc2xhWMC-}5)iQn4HUMTGlC
zu4*DTBX*NR480AZO=)o31nG(l+r2p{@<RGK$e5;A35^R0{<7uLd!Ln{Jg+}Lk~v|#
zeI013^TJYtsq0il&8J5RZJ4p4EFAAT`B_&<U8APKOnWm`_<#`E-uucc{asq+#nsrY
z8t{fc$Jx`@uk!^xdei1JIP8^8rlA<YzwzbUepl?KXZzx9Zt8%6G)YSfIrAE3jUe74
z=~0^{-=yVz`zrsv>)B-O-|N_}0A~e)ZyZjdD3s4S@9(?Ap5h)irik4Z<!RyTt%;M4
zxeX9POI;QEq)+|}2pS$&NTX!Z2;qpe18y^c)6B)Y-Gd8GUKVTBRB@1cIJL+piwSxg
zx0WXrbfL^QilSfQas6$2nos|os{5<Bj%a8k+Y^7)7YTc;kO-Bze?G^ubJdVb@ZXkd
z1a^#2!}_D1x1PmSG(v8~G-Gp~0awoFT_2}IVv4-M)ZI?)ecLBt42D(BBA>qN6u5&W
zGR?~dUankRo*Z2$SK#5~);CVwoh)8r+(U0&5C&Bh&IC{B`D)$o%I`0MamV2I53c3p
zi4<89tGhA8vq(jc?~YH;{Mv~dO*9&W<twt1;aID$jY<k$)_1X0g9U{OpOg)x1r2wY
zq<Vv*ytNNltBA!K(%lCw<kVvU$R11IudUYJ1Y2?sh41K}zIlfGLecJ6oauXLOTgud
zL`S<@?A*MAhsz@#3ZjRcV9)nNY)TXgpojF-q(+fcdr8tloOsVsQ3F9<k71ibf|uPU
z>d)Uqa)O^0&&d7_5$_T+el9B|Gb)6g_hr{+#6DH;aVmPCRQ4GArqolT0bcL;Ww5Z|
zC38OiKmj*#0TJal>Z%c$Y!W%f?fcE$hStvQZYh%2{i>XBGW1=IaG_250rCQOnmx~m
z4V-w3EW!t^y{gMVJ=GXYV;x{OQMtLNhD|UH)$eNU<-xFacHql;zA;SyV8>0(CSFmO
z_IVQCT-fb9+dp+Uy#qHkz(+v>S9wMff##AKD|HGbztojqthp!yJAH$7+UvQ49TI+_
zsz1-7gA3}bUa&k{>3@0tck%N14<Gn{am@?fS{h(>d}`1Narn0|@j$*fLa3GyMOjP4
zVO|H7oC>kguIse0c{#Er;P&19Hy9DfXLm|BW8K$uT9c|@x7EZ`3;6WuQyi&K27C0E
zpMKn`Ld>{$)^TjAMQJ+^z%-vJ-cIXJ{D!r+zk_nDE@OAe0;Gw0le`(DSjKt<)^rDZ
z=x97hAozL$YxJK2wffu`7-?+?Jh8b7)C8afi2{3U;K59=D9q<z0c~5~@RDo#5#F=Z
z3#<lp-4467!g_QJfL&|x9A3l8$w^n3^z3|o^>mf5ej}nwJa@he=2Is&n@#LD5J;=!
znNt>M;G1r6cH_0sxU)5d7*^U&wwr;&9aPP#r!9l;*4y2?=#-rF4F@mIY#-6!%fGFj
zR?U-GWNxT@o|@%vzDgg!-r{#AiTT*eEY8%YN-?nXE(sf5PpxrwMr>7Xs!WVvwnlns
zI;4rn^KQg8V1^Q=uFwc6^J#N)b1i=C_S)Ifv6t~%>@M)FqH+brsBQ4S9zrJfgZ<9y
z1*y+#Bf0LtZf#x*aQF$hneBaDrEm9`h%~@sc+y#bnwbQ4@XP~@mXcfT3H<$ot8|Q?
zvLN{KqG1f^a%`(<@s7~U`P6=t9gP2FhIgs62WZOuxcl*{Z11iddN(>Irh-<dzc*p>
zwUJqcgfbS*<XtXbg@;6Lqc~VtcEYV}tF;+XTr=P$(kO^71fr{-9&+Thl;{1@bZwF<
z_%;C&_W8Bqh7N9>Fh$qZ{`FWvey6wu<M;8!kZZs+Hr9Kt;yczsHT|BV434*D2R)ca
zosZqGX)|`-4OodY#<g&PZRzcL#}}|6zKeGt(matRq~MPoN|dj%=9u=0Tpte2WOYpG
zQXJfK#bjH?*FbHE?b)}o{S{W#iIPgXMC(?!vJ(XtZwG*XMT6(VLiqWT^!xg{Byn$o
zIG|Z+Li6mbi1h?r{0X><U%dLxqcHLNet+@4-}+0|*d}&8i)8LWU-NEGn_G?f6PVxN
zM&|>_W*@cCfW+9z&EetF?fGJXW^KMKAX=b}dJ%^ss2nu~v(aEW){Hooexk5KM)N&5
z3;(kRIz}#OO$g#ii?0j)CZJ+Xs?5p;0Y0d57Y{x_XAUE<+&L0P3kwUZo?7NabF{$x
znl2u`1)D~Up5t}2G$-C~&x?dG{kL8H>hK+uw3`~~J4|hB?6rx;w4d+CeSft%?~KYH
z+nm_;L0Qn*BnzTTy)9hcOgHI0XJKSuU=H{=2p7b|JE-1no1$*As4YN9PfXbYfquJF
z%$v7D56I{r!ty>o<cfgC*sSJZyVFG)(<w|nr=gvQCDla#&Wf|alxCH$13im_LQ2}f
z{gqe*^ya%e&w=K?a<OiCwJ19{@+iPe|I|w0$1e?EHI<4cKEqw)xOfVVmkDwj1r{KF
zby&u0sOWQlcPj^W@Oe{PvzY;8Lw*a3gfW#xPYu{dH#d9B!6S_Ht9_@5>o0Ehs(EAD
zTlQV!IrPYEENF_;ax~aE#<XxrA0G!xGn$#_56R_819=-<F#<$I-?bCYpZH0C+w;m$
zn&D(>_I;ne0-sye&39$kOVq1a36q%rC@)7=*&ie@I5w6L9gs!Lzi9_Z{^i@**t3>o
zB(_%VI-l!09qqb3ytw>lxDe`nwr;a@ihTA{jl5D03#cLfzP+~-NZ-(03I8g%)F?XO
zK_mG{Ri=7Qi;Vo01#h`$4lS8{eq#Ics@*oD@2UYFo1qZJ-J<_uS_SQ7I&PPGdM75R
z^nO6ocGF>J7p}%rhP!Wo7))==xu&NQyi(t!30=WElq1saeZF5jy~w!U>fhMq|H;Vt
zdGOi#W>}E0b8$~-{<@~%wE272%^v8pl{`TPd|XhC{Uh73!6pN}<J@c-*Ns$uyYP_D
zXN=d+m-QhVZPZ;mkIuYr`!&B3;tI?fg)ZtWl7<D!O}9^+CDbwk4A6E;{mXc{O8>Ou
zX(K%nwi!BABrw$9As+=#PE+vEt(hu&V-+%HYLgJSH%7@nI(qd!A!tLa@VutwL@@qx
z?USrp(O=2ezPEG-oh@eQF)fP{Ra&*wl9_4XdiV8fLi8+_qINsAjKkui78oY31$uUP
z*xq&9uH7kpx6uM%m0&Nc>h~=t;<1sF*6c4cl)??jz6p%wy(W`?W1xVi+}?~4=U`Uh
zNsm%6?Q+h6(rFn^2J8EpVD?;Jc%ZG{h&iRW-pHG~^RJ>Fa4D6mvq}bf-57}4#>JUQ
zeOvlJLI1j+I)vnpsOALMq$%D%RNa&b-Gpf?>ITTO3|UkETec|85O!cgO=Cm+w`ui8
zU86T}Zl`8{_U9p#sF+vNW{6Ey3pW{pD^Dj&&z|S-BVchGvmEuUMC5K}`|=Du6(9|M
zwT+mld^JES9pOLVsACbphX-L{VGi^mCZ%5Oqn`W2Ol{NQkBXxlKWEjY!CM{LTDXN>
z69O<*T0{NGu7*2}36dK`6WV0=pR4HYkaj3No>!8CY3fmcFfFl{7IBzFWOtG$S$LZ6
ze+V1}404-}Rn<{C2$z5ppX@V5J{&(Bjy(NITNP|BjA^T>Qcg#5?@TeXaI8p2r*I~G
z6S%MwT5O#%pTHI3ix`yb2-!8Dz?X-R0S{ODsNb|b8)$3hnPQKgnEHbmnhKWDUR89N
zZ?uEp!C>urhnGIHZ`XrF&KJIW277vrmn$0Y{-`I@6Bxi9zzY-n7(T~HvMP=yC>9_#
zjK$|5Sg1e?M6iyw)nFaxyO#H(r@{Fq{$7%(ymV{?<Vv+-J=Av2_tR^_Kt|3-08GkM
zH_c=MgL#Zf<eB9Ae>9zkKb8Og|BsRJP8?E>k)2VJ?bstLg(zh2q~q9oT|!yMo^hxM
zg+tkUgp*N6IypiPWgHwFTlVkz{BF0eKfulHI<E72zMhZA<NiQACD&E?LN_IFLO8V$
zOe1}}^&<!MP~|D*h@i0NRpW(+{Z=jcHLxFu1Md%Qhou@v^PrA)dQO?1?Tl=>L%mc#
z;9A#d%dX9!x$5$DwT_9lSz^^w#o+UzvjWk`8u>Q>>bz`AE=v8IpYU1uiWbmeFEMYx
zt8sQ;xKDF|I2I)JPx1WV^#+9{rCbU{<)~bQSnQndFKz3B|KjV|L$FjHc}jJ-$z3;<
zYA<+lS5K#Sxv_b+o~{I5Uu^d^m3v%bBv1kNGjQcWs)&BynB)2K^KrU-Z}ZZ&qrWF#
zkL2eC?m%24y<TTixp>OHjw1pwNn?$BZw|f(`%IKO6{?@Ul<6M3YM&+TRd8S#ejari
z-tFWiC`hw586L^l-8(ttLJ8^ng+2Xste!%vpuP~5WlRl;{0w)$J@ecC!a-sJ6?ytG
zok28HzCUq~`i;?AMm+EM>YfeZ(<P>USeVFAI3QnmjTbxr=Y8c3QV|o-r#KttF9(a*
zjph5@9!}{#&H{?vK3H5E(_5$JeKes&Mw#K9Vc7FSzEjg;EU@K$Ug8xv5fIN3%p4?A
zA*v_knBX>k312NaEyma(JE*xLImbp-T^B2SQ~639`hoI2hpv_zTmu=G=bQHr!ekl9
zoq<Gq{RLJbDPzQp5){ac^&eQg&Y4VgXqiGe`-8Y_PN3n_Ih|?Uia#9A*tj_{B8#qH
zmZ15f_#cDGT)6)5H{NJoss{NY0#-Rw@Y=$>6QRtvr***=N=0q^MThI+d(>Mmsr*5)
zTIKMAjr!iF;czbO-8iNk3tq<O-w*VBxKtr7L?x_}FELy)lmc8ZJ--$-rQ{A#b>U&;
zbfJDBo$WSsc6Mgt-RT<=Fjr(#Yl&uH4euy`vHFe&HodNvd0)Am^*YkF#u5aYAK7?!
zZDC8}eG7bO9SzW|q{x!e1nNxNaJ+RIP8rdGALb-SdXL0rhJ%wbt8R34W3G7$l|$ot
zrPr{Yd5U!&Q1-*xfb+{;o-a<YUk-w@xfLE}6Dfg)oO$KquzLGieIYGcBfP6s!8^ZJ
zYdR{hkpOO7%~7=%=I!lW|K8^!dUXm6#fl|WTc}NTm^`G3*83oq`YIcwj1<-DH6z(y
z_Cq?Z@Tfg5oti1alvk&idR(;u{c0Ws%JkSN{75peGjh+z_@)ZaVO7=jp{I6D9umy^
z`ozZX&%_yfv8IFHX{pHQ#br7xb_0|5<~ED5rlrax9)7t-(_vd%W8EH=D%%NE-P_+r
zJ%uG+J95Vx!5R_AmQk2zRVxFsyEP9d!#bID7<yjzIpM3fBjB}!ep*olo_Akc1I;%K
z-!|9*HA%wP^xlejKK!{z+?U*uaP(9C?fr-&g;{e`>YL}6{Mg!aEajw1<Jf`Haklv<
z7EGFp#(d%6uyo(LTQyww*njBbbA5f5cA0<48t3huGkxnr>h6DhWH>22k7@A91NH7;
zjp)hQX*fvxpVj$v9}mAPI3CD>uL_V_fDw!V#m4brwq61;b}082fy{TZ`DNIO95&M$
zOn`no<GC0O*?4mHGNIFex`y$_rr~Yi;o_gRqxQm|Yf+Qk@7#CM!7X$z=9W0U7_=Rz
zzIiLK1z)P(-(<Dz)7jj~inet6dq(fVMB}RC(}@qEIE?S~J(+l3ToCVBnn}3^<~*o@
z=y(=BHddC)kP4NuWDBXK9QEHsarb(@$b<0O&V4DP`4vAU9%jyr>>o1IH?GmXl3)7w
zFT_23d-lBf{F44b-ueEh60G7k>2*|Vd!t**f3QgcoTMaV&fy{W+$LFlq)#eS*J|84
zl(mMN^4-L8q-JH%k^f&GJ<AK4QhC25;Pb~CGCxBc)L^0LVBrF*ejUL$uS|MlZFdDE
zL~q_>a^Z9dEpA3uWwU927m*D1PxG=}x5w<&QU^e|Z(*A0&~9^~qusmTFIIlH@<~;q
zj~x8;sSR&eQB8&I<Npf3ES$hf|L)tXUgcSFJ^mq;tJ18`wb{nb{;ZFMxFhP?0r|<h
zS)>HnaSt4)5bw*kjl3DpY37+imzy`c?EavIO60M1#WS;M5vr`x!jAlRl!#n2KnQ?9
z*s>4o;PJvn^%9r*9x(jLkftL%bM%zsQ4QN&o(Y~Mw{3Wz9o2%&`_pB8jh66Z!fAN$
z#_zVqb~UH3h|(PSscF5&5x+nnD$sat%w{+D^7CLFRmTrZ)U4S122xjxICMBa5e~J<
zX56ANV)559dGqyMQA7M7;LGfH^><+5uZ5zKUBK}0_m@sTe#(a&FJ^t|>kSDG<{R--
z>;64CV|T?(Xp!&~rTVcAQ6Rz;6J=gIER*_oW_=yp{nmNS3nynWK|YHC9@7dVA0$T3
z`Z->*mK<o!p)YGQ;HGe5veUC_A1x@e${lcx+Es#5bGNK(Pl=`!VqYZoNhj*FkNuQr
z-bE%lVu2H{DD_Cmez}!oUZv!ZE#Z41!b)6MoZiJDlg+QL)7@t*se7Enzt)Bgl~@gH
z>8vP{C<yZI(El9vp5|2qKsnZEZD~2;IR6VAHO`*wReAV^h57obBc_fHlNW2dgULR}
zgvrAjv(utalE-u6zdz<A9oE^}IBKz18q;ej6k&dh)rh`!JNAWtth^X6l$NQI^x~IW
zdrw=SG|R$wsx)jnm?39TKbou6pVYBYxLq(o48zMjtg$?IMUX@Rjl<3D>SR~5uLOLS
z*ZCZ#v&d#r_|AWEOy-uC^!qBG)zs7+r7Kkhid}~o8fR^#279(iaOU~uVk~*3y&lmx
zeijA}_7uAh4nI*+MxhyR={M(8LkRwJHCOZ^eVYpz$48tzyFVVEBT%ir6|En(&pl?p
z*fTOZO?=>2>^!dmuG82%&3DA?bw^ZwyK^+Y(zR+qH$j33(DD<%2iixfosksi_%%ri
zn#PnTmtSf@Y14Ss>IlUk41DxSXfYQiSno1vt+!WQcAtLsocJ2lN&QrLJ9ls@9tFg%
zaM9E(vAePS*@BhirG>+-XG*8j-DhbI;pdjC18XVm2gZPjL&RadJL)8iK<umsYpwOT
z2D9GW?V6CgLjRgH>DQY2<9?jPwGB&iw-k-YAnWFoW*^G_`j%rhZHDQsG$ldhO~Nnx
zhPOFOAisiFPLMv0&=Ob%Dn>yR5aoXF(neS)ziseVrzlg0*|PV>m}O7fw{NTRUWAzX
zuPRetiNNtQ%BSEPkWHM8i@E!1+r{CU>PHiy^MkRpUsTh{dqz6E9s344{o)J6aKbBN
zZl2WO;C&X+cKY`4_`Wf`REyXMAxjkdE`6ALx_*gMO2>KmTF*%DS8UVUFPkTgGq38)
zo1GV8(y*nTQV;ZmXgCMqso~5GJN&fX307oC1d!_n*0Eb>_A{MCNFxeYCx`Hh&ie~n
zc~%AdEt;EqVS9hq8lDZe-HJ)L9HIp?$iRMr3MMhUV911{*z|Z%P6_A?R4Mkcy2H=P
z!rF=97XJJ1m^gjCQGxb|Zz@On%#Rg00P?qq(c7e|z@f*ER3%dBX0%!@jg7Zxy{ANs
z0e|H04WbqZbFb*E!&wOuYK!V5I(Gg)OU<7*+sw^e*Qv$!R*qTyz>T|^O~_KQZ6ZyJ
z78}?^q89$CQT~T@MMg{i$X3Y7?usv)DB6GXqpjq|?=Jkfe24;1TATA*^CFPMG}I~-
zMcU6$DEHkM7!ZHU{7!SzgN^pNlpgTK?2WXkfeJVU3tylq3oOQB!{{SGAz+G0-zINF
z*=VQj8TKF;@-GB_bmV=uU-$m}K;wAg&~bNq_Frs`eCZamM7dbs?Nk$yLofd2ziR^<
zxQ({&)3NI{RSmL-Aq$+L>29X(hq$v|&6fiDu1|%9+p3rhIpgeRd<BG1jX*Cvz!r-&
z8(#hv+!fSNe+m*rEp^SWxVd2VEn?#jJwr)^(#hSVPfHpn$1Hcc7JI;T6Nprbp`Vjx
zh#BxydxfHkumeA>VpYVBs&_%|pq<fe%$~2Zz*Tm)i?DZ-J@1=7SXO%MNW>RvkzeYy
znriR==SQ;T!$!^rZRW565k@WjXTSd4u-MtoJ9oMkFJ10^#rwZ6mIb?olnr^G{+cgM
zg&>jbjzhQKj4C%rN!sd0TMcj<@{x6*j+_&9$`9#2)K*DL@F)NWvBJ<|U~7Qmwta^q
zs;p1zp}BU_r1O67LD%sqUqqS^2$B6zE;eCVzvAkC+(!X4Q1jqIdF^@UDX#V@2H4*+
zxmqtfoqJ9JLU@Pxm;Nmi#yV}<wZziW(nfDD1<ni_JellKI#@X5^iO5g&A1pdx8`bJ
zGof)hSwC*LSn3Lgb?jlM^*%_Y$2@N~^z*7FSj!q~dp3Mm`&^!)w@qu$mZtmhk$I<N
zxn=!nc&x)?)=i*59)VhE_KD_>ud7wq-rSi|>>g{`3&h9xbElOWiC)4^s+;z!d?Z%k
znlQR%C`G47qa1<kjmy{W@Vn~7p)Tj~e)>?fV9StdfSkVaqK50TUTA?)W<*1M_{YY^
zLXpvtk{#w~m^9rodW>MhYAA?J<m~pf3nc9O?sqjm+6KCSZU)@58M#1n^S~LH%Wu7$
z1u9L3jPh66KVZ|`=^l3A+5ZSkX@L6r!A5Pz&LNnfJKNr=_+lbCI%;BEoU8k`JxG<;
z&-pgac)Dg|*Req!%G9SBf^x}y!0*QM!h@^K3aW<-{aKJeEs(g=VtRkj+^P$%VwV>P
z6>)+Btc)a2jotO`h3?I!?LQ$G#Rq?cwc3iiC6y;?YcW+hUqii+$5N-K7b&zK(`&I?
zUF=unK6C!`o(%GN1xxrD8VKg*Vc(Gx#SZph4_h%QK`d^$TTKfJkGA;G8Ec2x^HEi3
za}!22K4+hk^Y6NGQ~VAkvMP|5h1=Zz)>Ru|zB<2{+2s@A!35IA>ic+DP$<b!o9XWZ
zTm{~Vi}88mYQqlxP&w7*rIzyGg`L1@CHkroj14BoCflF#qH%S5cOXDTy%Q~S^j9&8
ze84c%7f(TVz)!bF+>PwGXY}sAE3!1U|GJd*=RMkCeS?@>KjR3kT+}tKZK^}jTU7zb
z3eWr}SO~bwQAzHx@;0UwzGlt)k}At+#4TNxSHkkT%ZcN;u_9Mf@r3yn$N7>2NLmnN
zSm5WYy;>zaqH((L7c?kT*ni9)^_rE}F2VocGFd37Y$`uqNduHSmHAS!9tH0_tj&$s
zMPjMM%Nm*jXQ;6EpCrKTe)>Y(tbqTP(q5|t;_58(Qwd(wbw$CQ)}7r5Me7Pn#Hmrv
ztVE$@!nk6}dHHVl=TE)lbf}K&1dW=G9F{@A%lcpKPu3PMVKr(O&&z!$N%ncNtG&5r
z`5_Nc2@l7AwwdGd3q^zHCIi!`Z&QiMVX32o-zckbrRkM!EiMim9)fp+T)-@ZLf$KY
z`?}`-@>EO9$`{yKt{f0#(sBG*@}WAgdoN}7uxR!;QuBUh!w2&E`XAsV7_gsKxD;Y(
zDIi)Qp9gu9mb#C_DJ9W8kR(h!ZwaKV_wsvJ9Ua!sk00s^(?0$^GjlH59hQ7Akx28{
zL?WFHWZzSV5x+c>p{3Y)11o(;k1m-F!|d8|T<g^0k17cD1Ao>XVtg<4KvM|BRtY1~
zth<@QkO#SLuc@D6)=7U(H@4ZTcZk#m;+RIgLf({#sc~~|1J(;~=Q_mhVQj`*i+PvR
zIqw8seDa~ePG<{0S$$XVLk=ATn@?c&krkdp$qGZ{u<wnVQ|bwJK=HRmXRUN2k?=zw
z@@~YFTWouhe9PCuNDXFa+1rE9*3ow$?poaQMM7BE9TG-f@8+T!6+E+~vg(ToFR)_+
z>`Woj;3VG3XRXzRA_>$k_pL0@{*L_#u#!eSaCKplxqxXVf-rpWnOTxZqF|*E`ZfLY
zgwKwR))AyT4nPk1m}TP(ESPcPSBk!~X;bP8jYD7L6hTS57j4M-rQ*W%gkxN#eiNs;
zjYV!t*6!A6j&mZZdS0ES8|eTh|9zi*xU5(;!a4K-v1&kPhnI2%ao_b-D}C`Sv0~sl
z+6l(yONxN%*-dcbtr6t5SEC$mpEk0tLbp~|D_5QYdJCQ0??TOU?JQ-naZq2L+6l54
z*NyRE&0O;rkf|f#p`=PyZCZ4loKW0=vpS4vq)t^SbjxI)nNQ>RM@UBtTGxqxu>9_u
zwkMtTWo(og^(jtW-A{s<$>Vj10##ls?5gvp$2FZ_#-whqddUp8y5->)+kS}Nej;ex
z?q+8WT#A{VnZ(ttb)NREc%BZ4zGg2o{F1fhIuXur$10#o_&(Mq3p$DCxQ#89yH^)D
zmVV#FJy~LihBAzeV;Y%i#NMQj9n1H}!K83hW1+%BRQaO!1b1t82D608C}!=q*_C8E
z@s`MgvGGGN4bxd9@`s);!sp}vd}OgJo_oMlX)sJ(sns@Y0$rqmXfOA$cFlbY_Q%<O
zBPAaOs1DQ5uPq$9U>g0UU?|YsI~MM6+=#*0Arqu43oZWRH1umTa4vAL{fX>_{C)yS
zW`-OOlO;}6b+=#QLLTD7UFY3OPlxvmwa&vK0R+)Dq{7R*Z~wlvu-H?bdY}Ei1YD~p
zUA&C+vBBo=-FBi4UZ0qHt^F77HEqE;Urj~17P8I#MUG@g@r`)6e0;+Mm4JULXNdo4
zJuPR~&A12L^O1aq`ukp1`OToR6!8Wo@vUjtOdy&8+P&91+jZP{(6zPa{nQQ3efF=E
zuk-w@wn#nvWT8IoI`x|?nM(Bi1Aq_QtjLDr_G*omNQeBFQR9@zB_mPBOu4N}5c|*?
z8K61%Xkzbmx@@I<!45G!+72+}ttq7r!}0SQ@=E`ly2+a9?q~mVCt22PmBWITWNXT9
zgopGvO8uIizK8z_1Dz>h1MS@}mFCQs#`o4YLT7NHq~6d5=Ekl59tPK9pxOnOMZb}u
z8cqN9mX%R)#)FRPH)5TQ-+#7^;-6YC9zURQ<$f?$>W@<a2~Ifqb?4R%m~WckUAnuO
zDi?KlGNhj*>{ak+LJ%nm!{M`hv>0h(xbv^7UFfUErdf4PA<X*Z<iKpij?HMXy3r4i
zyKQXd@$Ie@L1kml2YQDV!0bh+celVzvS=YNYSH1H2=`z`rdT<ncXRcVy9?my5+u)K
zq?fg>nOTqJy7d{x$(i=%Ry2Mp+szH*mhh4$eJ@&&VO}`p@1w!bG;Q|ZNlacHj{$Lb
zD$U0sA34s{fQQY<o2CC2)*2p9^f_&)@oI#Jo%fOtHZ;)Ro>aI4BrtDzzeOtbnx@?Z
z{;)%0q^hb)iYdno?eF4QVbqggAdy=omek=-m=a6MHtYMdxw`sH6$D`9%2boX>>eSf
zUuJ^RS68>JFPJo@6vLH08>=dB@TI6}8pNap69(9zBXx{;i|}WaAh@Sp4L4|he4JJH
z3D*iw!aN^71P1DOQB5?sT1j`0V*Yu@YQkehWAX0yH4RO+A3-e5=1eoJRjXLXw=Fni
z_IXH<((wZR_LGrGc(T{*)GO4((!GeOfXat+FIN~C8h1oV$)K$&MH7Z-aAJP{&~E0v
z*xn$TUQg$$lmXpfgn2VoHl8Ve8I|)x<-3za3ZASBWROBTEmYdu!4Akh2jFN$^n4L_
z7S`yDnrREHLXL;EF3{rdskJ|j*eqh<?b<SCU(~hACSSQTswwniaWT~DDmqz_7{dO7
zvqh2#tP)S&+LshkOc1I_sY!OC9tfNK`EX7xv@BStr?_<)L*~4tIADpGDLmO-y1J1=
zewh#WJ|uWM+5&1i>j>anWkNG$=Ho{PPV9cxr!x(Q8HFQ#J0Z5O$e(+K<!m)Zl8Iwg
zX{$TdIcCfqk>U^I3N4>@uAk@3EWh(i6AH)O6*p(sh0;O>VdOid9S36YJZfmE6fK0*
zVfJ5seNS~G3eLt>@=Rr>$(SsaUbL^G0v35}1Bbxb=0Q|*Rb`MGf_y<{4uu|T!=wc@
zr7|CMhn%csnW*0pQB{arMk`>L1$~d7QV4&ZD_V$O@!9AMCJsxjwJ~QTEQ_?Vv%-);
z-&4BVcY0K%MulViWt>|GHR5@sQQ(y}5EwY_{z%7EjyLUS(i2Ln*t~tcWCmO||B(;p
zqY4X_M_Rin&!^~<^XqyV4wh3~aQ>q@vj!m|%6%}YxzAdjcJEQHhCi=)tfTkrrYoeZ
zqKhBbF*1hU9jj)lDG)<-RP_4D@FFeEXS_#4)}_BDO6yR<++{a^-7%~=P$#qGqNH8I
zR^JI>sT?~cqyF`!@oI!-x>*$fuE)pmI@Hm7pQ*9=-E`F`X%H>;<;xde_M?=8CVI$t
zAo$EcLa9}zYn3w+eS5fVexYdPmPWNdLAjU4`M32%a8G|If0^%_Y(FBs6T9wRsGbqt
z%v;#_w2)^1U@i}Ec*{qkaTcassi_*NtPgXqljr(X>E+M?pPu9*1OoNedA?|V7hS6%
zDQgJ=I?JDzwXn$u4;DUDxmjiRx6YMFs%RS28z63#(PAdZd#;EPYi>Q(klSUR<Tew-
zg}X59*i$K#^M{mIw{oA)>43EmHiW<#11-Zw;wRU9Ox}QvH^?6gx_XcGr6AOxxI|NO
zN%p(m&C3@|tt`lafIRG?WD9<O^;wr~&DcY$DvSElQXA3R#kWWHp33nfy!h5GdX!3?
z45dZw7xA6RrLXuLNEy|at)gU|<QU@qZSR;geQ9p|gP-*CK32bQvM`w}_AOmMusv!1
zy;*Mj5S&+CL+zmdAUH^gkzP|66<gyGo-Dd9@%X0h*9%IcEdc-6mc+e=jqQkZmcL*n
zLjajLKdLT|Z9-lz`(89T3C7s?PQeB=v7Zs2LV~v4c)IQ7ue<n!26EDtnEsb>ZG)2M
z7U(OTQ`d$zU-{m7NGNo(Vy=H`l}#l|e+3Q?J&|Q5a3!Ny`?^9Jt*PDw8t*;_CT1p$
zdPF?0avoVe0Wh08_}0Ql@Su6~Xl}Dhf->5Ff4?F}gQvEw;ZRL1ox_Yl`k$M(PL@14
zWp;aXx(RSb;`GFCPi@HeFX%l@iIB9;##4Fav;>@&lvw<C{0&b2>3{Ld{0!cz_E1)6
zg1`86!`?(#hA{K#yS%L?<jrwL@fT22{?11F@MG_8;4nx`SjY>FELb`K&RLp8=2`MU
zXdqMBgr15CPH9|0-V}yy4#qY&0i~W(#vU%!!3v3eaOSc(J#j|~cD4?BEcEb6JWZxB
z7390O-=p74I^!d`4Fr_`^Y_*Cltdr+{iKr6lOYqpN%tcGUEJk^qtoIq>rrPJRsy>x
zhk@3pmKaE+Iw}w+yM&%URhS7xnqJu%-}X~*Fb_nab(vH1oeaLh>-bf}Mj<eIc8s%Q
zuMCr1t%?QVjfC7StM$&{1Zf!-+^nO{+0?x&3cJ=U%q`D_uWj)4gqat0!R=~u3Cc5p
z);H6P2cw~3A)Fn-L~bOEnxvZKOj39P^q~DQo&Nl^Bs5wUf9U+l-rcIHy<QZ&IcAtm
z!6{UyHy=?`en%*%h!7|sWSWA6(H27GXiL96YmErnK3QF@L$s#>M;@@<L^CMuTLSi}
zkPd&GjxfoSCHcj3PZcC)UD>GrMLH^jz;%JueQoOf->@&|x-W`(qU*8B<G_j(YK+_|
z(cMw-;_m0T$F(=d-?xdC4E?g=zB~C7$>vTDG@Y;mz0%G(6>sS^V9cCjeyy&9oX!-c
z7ZBfq2IO>Q4L(U4^tIB;lx1%`8oU7Sw!M-(;0y$N+x^s!k2lb&3e0?Gjlk)e=TwmJ
zL-6S-oS=Y`D1O|y*=8Q}|HQe?gjo*3fAL=Gs0%F)ch>#%i^$7=$8qVIZ>qX#m+G`e
z+Z8=4_bTDU+#Ox?0`Yj=lvmST8BxTCq-ca3;~c<S09jPjroCNT1b5uDf1w~&rB)5t
z|GP)eQmhtvv>8>jlm$9VqW<+B<l4*&8>>3a$bTBltx~=(wEi5}8Wh@7?6ofJrzz!L
z{BQi3mbPCavl0C=Oe<Y@ETEC=61~yHRddD80!#CD$yFQu=9F=x8Q0j8`qyWJQU=X*
zqxPy<X!E_PuzB>YDrNpI>l}~b+N(S5!Cs=VMI3oMBhKTH*Eu(G#wJ&`#>Tp2lMpSc
z;RUpRDj&;07UGzIR))`FKSpxKMbh6kGk-t0(&Rq1>)~7*BX#QBzOq>4p9|?>bY=3L
zx41Boc!eN%IboSoFdiY0O$lfg{~L#$?E1G_LjxG4ZxXD?og}mUAS)&wL4FrAXkO+$
zDu~W6@paA+8pn*w(p`)TPE4&-WYumx?xSv_Uzc0Pl;crADB?I|(f(^fLch}dLeEXa
z^PQ-heuSW+AJ$;~pM0dOLFZPDERmnZ)i56We`bQx-lw}Nv;T@tJ83xeP$83G-szMQ
zbzF#SP3hvk^f&tBLC?Cz9hdv~0H*}U8L<@9m97bMnI|b!rHopJPAIM4YeA`|Yy*TT
zVgSGdJ)U_)t~sb)esn8N)0rt<R-ChWs)0zWU2SJRF@C$6ff(DkbhJ1QvZ*~*OT@tJ
zgeg06`ac2tvzl2H);paC6Ns1N;K&1x*IAy_gZoo={A7&me{tDn%f`Wm!NV|?sj!O)
z+Ar^F%Ph$}5!B|Tz8oo4&`fx`UcE;+0Fbr*ggr(RHTkQmf>0S!n-r4{l#Ro%s;)BO
z?T#8U9yR95nb)u6PvC51y!!=l3I^;)U^JTUMFcfEh6JAHoa+Sg=l3LEc2-|RXZrK(
z5)UrHOtW$iE?O!`fZasXdRV7@Z5}rj^&LtlGzK=Z-5b#*M%b&b80DnC<ER6~T*Nd=
z>kU3giWcB3RU&D@`hC%EjHJ)7JN<a!YfCg$DFh7R$Qu6z-Ny~x=DWFT^l<OnJ(jwQ
zd|`20KGi8M>{wsR8bckCvb8oGtWboCLy)PjsjA4|Y_Rypn{1GpfQ3lkjDwUWm2jr*
zEcy*8nPk4Bz>`XT6?uAK<^lKyOKdJxbfIxYyUn!S#J9W|u~G9;wmbXtY3eajgk-bk
zfm0iONR(Z0)@dJ!^w3nH4PSa6wr(`Nwzk0_PB3b3R13oG1^+#p2?z6ttR$IHf7+?U
ztkLcWHuq4>*DTWLY-oU@!4Fkbyh{Ag=10$OM1re=9Wo7@yeN|h^wm`lkN;ghyQa50
z!xt3*{zNpkT{hnS^KWi>`QN|8|8Zz?7jLA#3YHkrn|!)#_x(mzxc~<><Q+r61l6V#
zYP!a6xmw=hjs}S-rbkJUNK{q}jXL^!7@@Gf%9T+gNwUeZ6D%0DBxorRpI7}|*|`Y$
zSAV|a(0NW&>5BlqH$4llL*$58;*=%=lJ2%gYzf|%scP#S(lH<3s+Udu78WGKwXid3
zE|!tO&l;o8Y2bR_a47Ky{6-S>8_N%27)9Hms0HHr?-6J3o`bGLItWh5Ct#VHZLWhr
zv}+gfe3U+z4i}Zw(HtbVDPWa#jI&)mVDv_KxXT#T3nw6rLjV%8Js08qD^3%oG)+%t
zkfg?45H&U%l*%+zAvF*ln=cjExnO*WZ-G{qnMdVkFZR~BD4qKJ#?gZbS^R8qOdY|H
zVwj=$wHRsZX1jA$LbUvotf{BBcDtAzi@e5k@gw-Q$1)X5$%_>Z5mEam|F+yU?Cjn#
zp@J<`NB}2!bib0EmDWOb5oxHyNjVQcM%;3*MovkLC}ZHP_N68X_GFy(O<GQu#H@N;
z3wddZTZe@xO^f3RNt;XLPld}=b5AzSfG=u7p;P-omq$TW)lFQeRc>Sl23!l^`F`4a
zd~z@z!!#l%jNI#4x{qfn!FWNSjX64l|1W^DaY22W3L|S-xvSFoa*24bU+?=PHu~~z
z=4DoC5DZOtHm>B-IsM*$ov!$RdzVgClnvc;3{QNQ?ZgufLGCU+jg4EtDDsAJ2>lYq
z6Z3uG6?-ZFRe#BXxzI-A!m+d@p7hl=-8rVa3=~^$s{X_bc2I`YD*j^1Zs37g0mSXA
z#;1<Ig?xBu$D<osK#Em0@+*C0{?oduIKKz>5R9oNCPp=Ue|@V&?CW-bDsQ>RjWVVf
zA5|_J7aH-AXP8IrcMZv$Rp(Nj>uzsJ=)IGqsL%LDf;GoxkJYWg`ULH}pjQ7nYQLWT
zfDcI!c<2QC{9o7~+lP7D3FRJOKhMXedP3Tl$?*UMt}-kP_lw`?O`7Sx@qc3~&i@Hy
z3bSfA=pit(A@i7l01Hdx2#VoOd4O1~Q!y#FgDN!YEFhmQ+5AtUrJXD|{HHi4gdgmq
zZ{3#Un4ZtLP=$%|4f`+C`;ys^st@h%{{U%GVKOkLf;K#sZ>H;XAdAEdfR-M<Csv<h
zJ=<ysGs(C4G>UEw!})%<HmVHcOH2FkSy?M<&hCg`*>2D{o9RT|Ph#I9OKQj)zr%f^
zqAu8miS$L=qwSQ~RgU$oWVQy?HjNLU&y!ulLbp_z69tsqk8eM#6(?KXduTZjrprX@
z!ZI6HDtQkt?(S)4;oW@*5Gr0rO$A;XdnhQ}-+jEPvgfzocrXj>E(ey{u@AUeO^y2j
z=MS(e2Vj}hwC|%^wYm5m2_wGWY7yIqX;VQ>Evj9>`Pu(=N<Kb1!g@B-iw(*1Fjy1v
z;#@~_|KJ}wjaH8!FYG%-m(fcXVs^U9bah5EU{;v+nw^ku9+SXF1eCQbTQFTa!7S(W
zO(MO_f3cqFVWdncDZ|S$l(*(D;~*4_bix8v92Bc0%%)8XPLIO9od@@QbQaJTvmp>x
zT3a0Hp<@EmnC#Dun>@v})$w`S82c5m>n?+|qP<GfU*kbBzI(wGO0CKBg#r4)iD#fW
z%}d^BXJKo;4XB}wS50Pat`Dc5=ZiY1g^!s4+yA3ymcp5WnwxDc4a9r6m$VS;8GC>2
z^_IWxJ?(96BEwh({r>jrW?m51nFu6uFYmB+QpnZ*HDHQ;5Na>`fGRqj;khghwqLW&
z#N!fYfc_JbUu0){_Bja>9SKZL^GaO3NQs<o>TCpV*(aAcT*LmoS&(=p_R+p}6nEQ>
zPv)0sKnOif1<%6PBfG(w!Srsfuh$3g25-Kg3q7co?t!5aqF2S4J29S;;(-qSF0@7r
zCv{$Y#}xYpStylVi5bP2WJz4{7l+ZfL(uN_o6#4kmj|ZBgjuggc5ci*nGHZ-So4`m
zB!CgrICA=N1x;o`^~0mH(^v_ks>;<<Bur=*HY{aQulkLelU64Exo}AnvZ`{bL7%{G
z?vdipe0OjwBw2{<Q`?6I3j&u*06xHc{Ha@YAf&K_OV_Ewy2PkXJJ(H{Hnf<su`Xv+
z7JxvHS#u*%^1CAp<t1W2tPgGwDl2=q0Lmwi8!oo|@0_}Py<|8A>K6T0)KW|1{Jm1h
zi!?+P@iH?*aUojPxBB_tmb4vtz7VG1nZ7(TC!KOeMxb6yoSr6M86O!(=fhQ0M*c1k
zil#{%qbkV4%Bkr|g7s`?;J?4lS9#YMdz0$Wx0Bb4B3kn=(YJa{mxu+uv_iW9yxBB;
z;H$OLaf9yD&;?)CoFYtB7y@zeIh<6rxBARf4)e((uj=o{SZ`k1YxQ>Yt=cCl(>9+N
zemo$yhr3G^tA-xge(;jxLd@twyAPKNowB30{`uO~R9IS?X5LQTIP23lNj2FIYK8|N
z(@$ge3CcRDJ>f<&B-e%&)K)Q2(6mD|=*9TYp|k61$X&LpPcF3uAwxf^!9%w{`1>5=
zdcW`isxYRmh!ZQTTaBwh&7!zpTD%k_QZ2+sG4t!I+>0No1$VKndavBn_!g|pB{+W)
ztHDWs-b7s$IGZe`4_|C+QRM)Y4DS;@;0REQSl>$?Idly8DVNiT1z;}{h*tC0c0KSD
z;+Ib~8*eY46_`w#sALKEB1Fobr_{df-sb}c45WtWYK5*fSW<#%Ttc}*H#L`H*EEjD
z52E%Elu#G*2}ZWchZ;x!+`*^x+PLCW3XDl$eOd1N>uED!hmLt|waWrPgPQ@ajBw4l
z<Pie(2c0PvKmZy8D;hT(eqEyfIlBz;sejKyooDt^0BpzGD0*mU<<jXt*FoRh-V9KP
zouGkcA=E27h+wSeLiZHK>!+lqk@Fd8^aK7b>-EWu63P05@rPK|^yi!7+J2I1Le2zh
zTJNIuU|vduFbxU4l}gqBe9e90bS)kBy!kCZu0`7J959s?VA`7pwdC*Iu?g7FwcgA0
zQz7mt^rNs{Ak}|0bbihX>?7SX2RtyxR~O9ZY7WF22tjti;M%Oo?7*<NeJA?;bkltQ
z3n&`aLe!<R8cgt&jAIg7h|$SserZ#xn(6xJAtq0+QP=>=A}oqcG2>NW>`bTzm|I>A
z#9Ri@7X|5!4gOTKn#nlhtQMKV$mk|qxfN<TWz2lsN3uk0_{UeI4={%odNWstRQ?+w
z$Ey*x`#?M}G<0iW$i-bv^go(U@{_n7)Kg--8O)`~*Gx)~8DcG7Mmx_0iQ$Ja$D0}c
zeb+!2*W{5R5B4!3;5fWQst_*m;@=pwKQPdS{swyXBRX$nn`s+h`B5=qNq^ynnPMY9
z;KoY6n{QICf+S_PN7GHGR3)?dfYw|3-DwpYH!DFwnK|k^lJxV5fi>?7Dl6R*k_VKH
z9=`l+eZ25*$We-uSs(Q&Ha2$g@86Xzj?=eE($YN4v@Azbm$APH1Pe{EG9TOvZ9oRQ
zv=ds)626sf_8;Vj@V6hol@xiIEUU(!wp`a%|GKjnp<oPF2p?b*x3tnlUIV^s27>>d
zb#|b0oA+5VrM7X58I?5|sx57826ffnYaZuziIs?5%JR)!5@hN;2(+VYY}iW>%qtjv
zZfpb<vs6|djXMDAqQHUOgo1b19smaMD#%~+UPqRw4MlsieA!u9oc8X{A(f7Hg`fwt
zirojE{du04P}%lj*jWxIIHVNJ778=Owa!t65H4H8=IszQlJ2I<^Dxbi{On)rYsl}i
z-K?6B?ZvMn&bGc7S7o`IH~^=O{3u}alm!G|hRLswWVuEK%LKLIkt&9OTVt`hMktyk
zQ;zS7n;Iue?URuj{75AgOI-1xMc~LE$H$mLDW5($zb2RE(NSDe@c!Axu|`hJi0y2v
ze>^PWWb2DZ?)#=jB1FcWs%GWCQ(c`-k6ZRt|0$8whFdfAqo?^{I);el4~%L9d2m+o
z(&?u4w>Pxua3qf-jq^0%ystDonCu+r8#GmIVe#^H<WVU8y<8$Ol=NEO=osXdg+}u;
zr5v7fhS!RW)RFIdBD)WKC(0Dv>h4>Xc50P6G00(;wRp@f3tZSMVOY`azVr6{4MQrg
z-jc4_Q$gViAJXOe1axEW#4OVlTxzJ&voy~eGW4!d@(Nd5`>G=ZXy4h59QsP$Ci;)s
zg|B-@Vzp31Y@%^c=bpI72u=a?k0)K50tVt>b(}Y`r3+{Y_6c#sTv8V+V|yPG8s*%=
zAHDNa|FT}~&u-d3%jipl@KH_<^g?CCVveN(K{0MFyRJ^Gy3v*qjA#(K=Bux9SdB>Q
zS!DJI7(5qy{z3ux<n5Y(*0f+}m@oIp{@Sb0&Z?UreYK;RI9WpviSd9yv+pE1rwv-U
zqo1Ix5xU!>DZ(!kAhKU0H=!9T5-w=z4Y2dGCtsH#?o+>iT0+|!PIq-@GLMrgzap$A
zB1@aa0k1c3?^LJm4uiuEyv<p;EItA`st<X~qsbs}fds^{L6Q<3QL>U{*@la{cVE1_
z@0XKY4ZU$kD|qkcF8f0X_&mBoosr+Pib0TFN-IH|nv+!%r5=*}$)EZ1LPs*k)vDg`
zdv|}3l=*l>Q4gIPs1l^#F<$Ox5D<Q?9s^~hg1<GFZwcMWJt&s$YCryc?1<rr!I1(P
zOc6#51NlJQ9Hvp*CwdvJh<#wBv<VXwcqE=f(&ZwB@HG&iLZtyRu#`$ID|SOhEazKz
z|F6K)5IIhk(enDx*#nqMLb0!P_Wwi(**HOJ6LI%!@i{khzoDXFju-aahEdJ&tf#A!
z_5XIou^}?AJUu;8^jSMQm)#M3)Il%lxY4`|K+?`r)#8aKa4se9eU^h3v*LuDl)|01
zqHpC2a++2>gTJCYLeV!rfRkLVZtn#}tk35Z>3gNaT>1=KZI&poz`2c5ULgvW!6>M>
zql_xnx3^_71^Fb^errf(-}oR3eL~-!1Id5?;lld&l$5=^jo<y>aqbmUto+Z^6=Czf
zz}+$U!Al0c+<2h@UCRBL7ju)_N0{26*K$){H8oT}-Smvpif@(gZ_QU2ekcs~x>ak^
zp*YH7vJ}Y=+gP(jjHY@cYZ`ZsC}$fqanfL1VdvJ{$HJ3;R!#z%7~X&k&LHxXym!F#
z*TLb&`%j`(Kj*X2MVB6^SKd!OTR%Uo1?)S5Bgsok!&ColKQ|rUzv?gfSoLsWZFf0@
zMA#euCQrK4{?WU8Z-$;l<9Lpwfp``f;)A^qMJfu;)w{j6u0|5{Pkl5nT>qtgqGn(@
z5Z}<5B6{(7`{n_A;Q5#HZ4DCwrKZ{|yi?4-+<!?{OPf(lYONMo6LIJmJl4k71?>LC
z*fd{F9v~9^UBb0lMSq0PP8FL;mfcr5Xt><{+3600Wdi<6z&xgzaNQ#AATcgWH2@;J
z=CXWD4Ee%n^lQ~sW|H_j;$(Ad)h8t6<MGOEjj;W-&`n>swf*F8;rf*g{1?KJjfzV6
z@o}PMA46-;CCLcKt>M*YR_1-Zj(}UJ{KIE(o~FoYxl1m)VB`v6P3atWzz5l{hI6_w
z>i^BYuJO9Y>;0SsxA=hLV`66{X`@I5Ia6IzV<-SFqoycyl&ZRbDIGorfjY6HE(dF_
z*F5WNnnpA4eFfZJLb+s>s`bWRKzj1upndI~^;{Bq!%5fl2JtYoES8P}Fn(5#gQ^KE
z2%ZY_Bhf5SAwS}iR}E7lbh2gH&;poN%6`$fQEArl*KiZqoqRY2Db=K?4ULYDjw3uU
z#GLCfK17*mictk;i`YvTe~+kK<Iu*`VlG;4?k`?&LcN<y@;83vs+v`IWBWd6v)%4^
zPj~0ll}CdKAAH~9%aJyKOZZGHgVuaig>$sHHq0qWOe-JqT_=n&I&&$OUx(qp=QcAA
zKf~CyKgM0uewu&_7lm`Di5k8iDEUOVo}+1K$D(e(x(O|l7~N@X3^YtUI0NM}usPD2
z8@0)C02H|yy2&`H_o|R=Ti%e`7B0rI?h{Hu79z?QtCSKUV-Gl^L3G7@X1eXL;B9z?
z03FFQZm_aaVmN2f+yw=8%&u1_iapR1yTjnznIDYod*z4{h(kA971Q(#w9FUwr$WB(
z<(Q+S`F}R9?^KMY{xW4ScJnaQPZFdsDG)9jhsAfW^eT(Pqpv+0@a&pHlR~3DcBe@%
zp<nYBG10#`0HTm?R#FA-brfrDF>0Pz4%?;Bs33()0w}%GpSM!4)DEkztD~gz7wjgS
z;&Suy3fpShuSSc#49Nw50X%KZwiZ*9lC-w`FSN$7PR{--2XZ<$6bDOI58<fZd&OQq
z@W}sV>3)-Y1CwK_aawbYw?=54A}0n)ZKQ_T`%7#F?9Wf|CPJV0JWVI+%l^Fo7%ir#
zXTnkc?OWq*zKFy5gQ$k%2opwJ*#i0!_GaI8-}4dr<5507o(*I*>Ph{MpVU~}#W9~7
zkRBxy@$ziyZ9;1^@o4OPSK}m^a=w|;+;wVZv5?Q8=Tp1hQOsuOwYGM$FnhYNr#NeC
zZGYYl9GsuDQ@Z?JND5AWyN~nE=W#g9DCJ!F_Dj3zDDt9ZEfT3<glW`bU}k16swYoR
z)4PwJ&z~=(NeO`WZWGa)i221uiIFb|m(AqmXT~8S)Tx>b;z`cON5gq6X<eu5&7CZ6
z#gUYmVD+PAp=r^OYqDHW?8?0;s$K`K@p1qW2VJQu_G%=Uq15!$;}}d0m?1DHJ@M=B
zf^o1|&TnN50wFGn3KLZwExtQTtJ|v`_HstCoVVjkCp%s>VKm_U?BF+`3P)<Hi9<{H
z0rBF2Um2UJO2}(NS(mTYUzGY_E=<^b*xDNjS#jr%J@C$bc*x_#PE{5v)7J3B1wh>u
zmv(`@g~qJQ_b0Gg+gL||21@VQR=Zp!Qnhz`+1O*LjiRdU&rfA})5ztF1XgX-fjSt>
znggu@Lr+_?e@|NY+z|UjaK6v9Rrvsw+~)P9NaTOmAZfa|j5O(QR~WHIE-%={W#oQA
zc(0#BNLufms=$mW1J%+Lb9@!RDUi4;Sf<19W5OPxMtr30BkqcFbkTv>Zo=M9STmaE
zf;R`RDF_un^Fv-w%<@OP<1L+drLwqmnD#EaK)=`pTU)C!(P43sd{#rwd9*K<Q6QV`
z1KSHIKi%C>5w{wd7}nl6@MU|#6sLJ(6qXlPmjEq8K~WF>WAMj?y91F4aQkQlF+q?K
z2N5VqUe`lGGW>5aAjwBhWnhn9nRV<bdyrvnZdSw&-<^$%!R_Usfn{rBW2Xi$Kn?kv
z|3DAw7@h7`IcQ$4QyyZK+vYGm^}C0mpga+Z&2qA0hdIy6V^tX5-q=A*bE`B5DJT-~
z%OxIHVJ-rXbB82=QA~ZZZe&ytmw;m&fB)i^tA}bi?Qu`_c;Gagbi27y`ky1lqYxXG
z{J<40$V60ezjghQ`1LZuxPZvlV%>4+{k*BOae<e$!KiX#v(&5mb@$nlrh~d3f7RAx
zV1PC}wbp(#6Qmw}@KjliFIBXD<yGc|YnRekWQQM{x^{Zi17=Dj3row9y0z&ipB_u3
z>7#eDvMz&Sxx}|@u)|JJ)C%FB^U&}z4OmMi1)9P_i$Nwg27`gdY;bShdfC#{)e6XU
z2Cgr;szN1lpBW{-Z^R{ke%8nE99HG`UFFe@5A1yG)aedZZls-~4ubmrM6LHaUs2cD
z68SZ7Vy+HYSzF(9#i_TSJ0c?yaG1GLbNPsp$Xu?XMby?oH&=FOE=l90lQNy)&=Snz
ze6rfhq3CVv-9|U&gAB)b8*|2%7p<>rblauEzSzEj2?@~J{R?+QTpf6?tp_%6@bRui
zo%~CVFH}?<`7@}@fmYK{0rn$;WjH~P_ReBn8jxq{eb&o&wxe-2f4&5EY_)?MM9O(D
zNDeeP-x$bqP*+yBa<Cf8QB_jTObVUwmmKckj|0`s6us4~G(d*e%SmVU2&WP~zcF7M
z^KF7aFjBn8%82FfkJhAb{?v80IcAzW0;LYwpF-5STJ4`5Qz+|vZ1-FL-Rh$`nf?v*
z>#((G2mZiB&e~8U%iq8%it#oR`)Y!;m)EqH+i>Bp8AHRulvV!sS1anJ_0DwG%knT8
z>GU<G)bE74OEjE1N}-}(nqOD4aj;7i__|bIHr9eHfXzF?m!}kNZePS}@d#T+dLMQD
zv1P7WCC-G82EP`HJ1HB`)=v4sIS5>6769`?Zlz^q!QH1gR~gPrFWsyjyGMSUQx$Cx
z`%iD#Q#}@1RZ>k>j%+lepx@I^6~IgvoB;(AJ1aswg~92AOolAfnHX67BzF9Y%U9vo
z+2&3~c=h)ZccQhKdg+Hk|Bf<>qYseNhT@#=b`y*^P8Z~~-A7yJ9+*EC4-ke!Yg-_Y
z9C0L*NIsOBA)LZWO^TFfWylB{^BaC_^&d{?WFsjWI!s%SGi4X^!Ey?~1f`Os__aN_
zF5aQN&o~hGKzn5XjrNy(k#1Ib-@=@;LsriQqhu7dCQ~+KHXQc`C14nxA;!s|^@=?-
z*B@+D!+|5t#a>Ph2qXhY@X-|CzZjpbQ%pVVtO-(eVjkGk2YouMV!2MMFU|-d($A~9
zI3p0pMa|voXgdd+(2k=Puc^ZhbtSdgX$@Bt&1V4!_5r&eGSDv)ft2ITQ3AF@Dv(d?
z+BKs=XvGg#>ao#s(MG)3%f0a&?@piN7aSX@SRn+M^?lJl0`$k33A~Vg$x7QBLV^ft
zg}}v<1m`n3U9K9bIehJ4NpwwwG7CILG>i`^&W2Jj_udN$uuDGP5vB%1V3f;^{F+R&
z@4kn-)0+>$Y$UGg9k^GB!h|H#&2HSR)SaP=lex~Z)4Fmjcsp%LmV#yzhFoVaD6g$F
z6c6?D6BH&N1H}TK7z{`9{kzvgXA5ia(k0M0loY~6akq&;X{^Ku%q1?SF_-^)q`tI_
z5j_OC-5I-Fq$|Df3PlXSD()PfoOlMBy?L6V3t?B)P6(|%4icaxAD2IP#BBui<J}hA
zA$+)>h&JXHeZb@XV&z*3Vj7Rf3tHYLMz*f-ZCv94Fqr!=*GN37%1_m*K(XFQKv)#F
zzq3=!5qlaN3$pX)V<QcJeNzd(*3wm!XJ`?7->^3!&Y;q7uG2lv2B9PrXeWzR8sTF9
znSlKLTS-0Q<o5aT`JTr4?X#yIjrF}I6%YBq{Bl!W7btDaoQ{THD;PM@I4wP2GdWu@
zamPFE6V|8grcXg5#>6za%I&SNu4siQ(kQtkz}i1aDlztk&911%F}XWx_vz6Qw}!P{
zt%ZdsnHwau3Gr;c;R@aC)rL%XT<Q@cp*1J=4aIMSF!;LoA7i}l>{x2Ny8SGVMVLkl
zh&=JIU&tC7rhV_+VTBeSU}j+%e)hg``uFdRcKZjp7u_1%?mtGLnsiN9y67R2I(P}~
zP1KgHZ_i`h01^R%$7dZoMEE*2+`Bvc9Ri^q+<PL&JJS)m10MHuy5dR|`C=bpSfJFj
zJ#oXRy-urNbP;peHprclwc}LiaQ=6|nkC)^fke%5PIW|n*|jTn*F&JB1mSTB5V%Vy
z7o(3?3WiJveU!7!n)y<-KmFfX0v+m;eu9*UOi6Yf?vnlm$d5M5!|1t-1iKQ|tGaLB
zsj4fnR%$Z5fGIs><v)V4KDg28|0=V*S=!o=jWhn=rcF8f`I$BgXGLzbsF(~;psmdC
zSs+Qc0!v$QJy+%>t6<y%Y#lAN)QuNn!+ugc_eJJw5F9(9)6O6+QGAw~)eyFrU9qGV
z=h&f3rT>)`+|<1kd(hJ(#m}BY9|yVIBkW9_4-s@w9O?5}N#N41mHLMfoF{UkGFa?w
z5Y-pJQPsje>d?(_6FaqvDFV2$`JnYVa(BdpS3M$OCkDzDL!S`yc1y~^j_baNkltrp
zO3LjGfGgI`mWh5~b(L24{GY%}N0`8b!i0U)7A|?VP4wMItP&sf8+!zWhCf?4Ata;!
z|NH=!VUKE1_b8r!YwzWhy-x%G!=arQC8;%a)Zy9&|D6f50}pYG*BLaqv$sDbsg8h4
z{l14j=CzIT3L&q5<O0EO&;RA|xmj7T>&6(;<{`)<H|DJml0!$t?3TTP)G_2EY)WCf
zNwG57JXhZ*j)7t{8w{J57-_3n54<OC$7}K<0Hl>RVg3$>e`#GiCWdoR{|uawGcq@?
zpOhH&*ApTEXja=)eM527T{}gFw>8LjY`Ph0fM1JkO3qgk6Y{t59NRyWyPD!*KLQLF
zP(R9x^Im96YcmNv``~JIDMp{5&j(-1RG#5pPr>nT*Od$`PaMB?qYl58VRuE9FYWuv
z%Hh%vcq$gnsx5W>DO}7V4?9x&cXJN*zNIO$ld|^)sLX$WU)lY))^fveh(Xs?&Q|^F
z5(8YD$0<`xj5$>NIyTNx2?j;(Jqb!l25eJBUq@|w?42AI&PKUJBB$Hn-M;gEEE-|p
zty`~jQRBE?<9wr?B5})Tqhk{@3bI-zM}eSY+#YyQ$L1Yqv^U8J-nYn9{iSB1IC2=F
zTnM~25fLIq<W6cFk!Fu;!Pwv4O1y?fr@HX2k=F9Y#s&!QQ3ymjG~<*oDwpR=Sg3pj
z@S1F9))zSb_tn3rNc^|i$D)FGu&IZL`NYEOgw|F9>Bv~9s)%rCycOlBE10OuAy%F!
z!G;d*eF=q9^YJ(`sLSAT_;oF6Fc{!;IbP#;Yp5)>EBYGWIw%W>0z38Hf6%2+U#6S)
z?j`Wwf1F`?uwgg-cox+wi@iVVaz7KQp~RyT4b`4KfV7AuwbhJhfdd+UOy+7krK26J
zp9Doe)tQ+_9CIazH}oy8^7N0KFSB&uyc7?0F}JWlxFfgw`x&ZK$$@-;_5xZcPE4d%
z`MSAgk63A3BQh&yrQrZ2GgRhpY%qSHwzT@6afV^2i!?S)8d=>~%=pSHg+TpgS2DAc
zmUL2Ch?Ctw@8Dt$#M{9`T42elcuiO~M?iDu&{?J!0$PW4P3K&nVPX6qaZYiaj0_K)
z`^RSACj*KBvHy>zvkq(W|GW6;5fXwhLK+FF0n#N>A`O#Jx}-}wZUjV>E<swlK`F@z
z$S4_KKw3&VM~~*Ye}6oGbB$};wfht2ocHTc4yJtv(1946Xvd0pDh;LL$Bo|l?vLUr
z?KGiXJiU=nHpB3BdLKcyU>I98rQCVO{eX_+TuU-ID?U|WG}RKUpnk#9Hp!c=#6FVs
zC(Jr*Q;##zGMh#r#l9DxLo*qYw!}j}%;}IVu9ZOiqQKdq<M<tArqKoAY*_`9H75)V
zC#M6v1M~!uO3@25g@rdd3b${(T}2q;GMnV2r$t*!@eMTCqLu_cy}O@k$8O~0YWKbK
zd&9RT_G9)vXNUSnLx#~1DAgxoT$*Rl$2_XxR9p?s<WMTU$AiIQ^k_QL@f~w#8?BQr
zY3}0wA{q@5l(15Vcvnzt;%^ZyE(bEm+r=%QhLlZL^Pl?_;q9~cx5Y-t2hUQKhSz<h
zm;`kmg|p53!Tkr$w=TNzowHw=m;ncCwR#>aqYP}R2C>g2C1v?&ByvX5B=HW!tN5$$
z4ZbF!YTtNmP3AhKH+b1e6x$vHjS&6(7+$;MrQ4|L@GgIo*S`PR30t_37cKk5>@MaZ
zP*qzSS81JShyIhIXHL%d9RKT6J7W<jEzibFyJ$=8e|7rFN$>Zq$#Q(h1A;m;C_aq_
z`*yxrXJgg)u(G^4x?uiBn|K}d5iF4pb?&GhxzM;cxOEZcPZ|+^fAQLY@3`G%M_g!V
z*><b+Zr`2q5!O8La5B!7uFL-!)$ZSxZ_ulgf$M93yMT1L$9VL>w~(I5g_A2ai;#mD
z9k3~USXPK=rLj-vj%l-Y;by<>e_U_mZL@-jxoN<~tL9^+(2Ii@3gGk*yKsBvVeCn0
zHuopy_O{gGQ`c(Dhyu53n}X$(n|Z?M?~x@`6O(`|ZG^@Mos+J4ThoQDpI?aG+Un-{
zfhG2Q^+)5krZ=)SSA;gv)O?YfJ_*)EP=j*wkgekCxH#nb5K>l3K_#1;ugH(wMw91W
zms)gKUE+P4-E2|QK1xAX!fUQ{Dt<gVU@a{T17N@s(bnp_(S*?Rg@n+P-UuR%wS^yr
zJss`+;>|TL(w=|eLKJPuut{2WOCh6U511iifz(RsR<^3nT^9)Y*#UR+xdnMSnV|ED
z!nL)v>Nu^ivPk~#uNfa^)9Ig?J$|o#k;@DG3i9jae~Y&A#W5zpK)V-aMTBO7k|(C#
z-K{k}8YR|X9D}WBR*P3_KVj?9kA{=ywc<nHQ=7BRf!vLc=nTG4rRDdO19e<h6&n4&
z9<rxPsilr}V%{f5bmcq0icL%&3~<On{!_Dx9hc>ZGrqrnqD9h?8ZPZc%#c{AH#}uT
z8mTKiyG0&5c0$S(N!1No%a#E_S~Tqty3fQPKjZaZ*%~AWX2($@P*N#|1nD0o@I{$~
z3b@$_q2)ihtJoom%J^I<H&7S{*SV!`T7Y~G8rBW9QYr(pF)2S@>3irSV!hMkdZAJd
zymq{vlGe6w6|%JEnyA8e%(+=`Q7|RTs&YC~=aaY6u9=?=?ziu>`uYO$6SELi^n1>6
zQmWusv`RQtPjtzryw?-4I7%&0eP}FUIO)8f)_9YdpVfC&$CA{<WPMz=`(skhRAHK-
zG_k#ub*cDYi<ER2tK5?*w2^@)tLwu#=5xy)l1#DLTjV-A#s6R&5O(~&3`S~l@&pLm
zfJseRj3RHvPVdoJ(GJks=`O-m<LbV@$6`vSt*t#4DGo`sEB}rQiPzDlwiabd<?GXg
z()GO6<Di)$v{C(9!{{uqm-&5(N>3?Fx3#H7PiS3g1<j^m2bpR({%*=}?*-}~Pr7S0
zotA}tvgmI|u}OVZDoUkI)aBR_N#MajB^G(k^Y~LuHGQ#AXC1&%e|_NLXJ;5k)H_h+
zJu14*bP}@Tv`P8B-%+`)=}&`irj8VWbK$AKw@r$JQ;3bt=(bSk&D!$vz<irb<Pog<
z_PBIs&W}CPoKfMGI<HNbSmIBZ=<`N5E(0o}xZi^)`|<X~dm`1P$E{DuDXmqcZm%c+
zpR39FDGMO(h#T;k|BD2WgBNEg(_A~XD<Vwi%~{T?scfH(ao>ZPedI+w(*G*Wh_skO
z$B!lSviCIS$;!6GVrb>oFUtn&x22pPru)J;y1BWz*;=Q}fnNN-();Xkp`AMLN!Rpg
z0(`b;U0W%_lMYPqM*UTu!ud88B@<3N1rNb<*61)O7l2|I&xj1+fM8_9NoVbp+&dEJ
z2HiCZ36!&&$SJ+h*0Id;f#*WEa|*Wy8{pt-1WWMU{{;J<L#*0V?Y#fxwN&Z#Bsw14
z`@T$%<9p9H!g%a{Nv-BQ1GP5{6B`Z3Zo8JURN+0=Xt+M=h`$~k-~gwxOLBL0D>9+)
z!3QAi^<*|N5l<MNl67&AP0szoZ$OD?gh1h{Y>|GsdfV9?8Hxwl$Ek!W@F$b28aP}=
z8U>YM@iE;Ga3GPCW>cQ%T2%t0^>avK0mGOsx4;7gSPLGQ32^G$h@`a@yx`9F5r%OV
z`15TJSc8K>8<Ux~#>&vGR>6A($QsI9-~(yM>~OG|N>V72@5D&;s}W~b<p&F~t5lK9
z^~2RcL0<D#`cl>b(&BH(M9N2>r_$b$v2hL)3Rd_B$_Bd*Cf@!un!=A&)PU2+srGb@
zmyTFNqd9%H-@E(P(JB=Y`zu@Czse!jgre~Ee!{3?KlFf4#vfBg5c_1CEk!2`bfYhN
z3*0g%O=5_bmB0zk67flE4V)DiTp0?=3D4ewDIt8ydz6LP$yv0q68#4I%nW$`Q2IEE
z5t@9iL1_=(*>Uh31Rt7wW-=GLc}-u4UP^6Ig}!sQ^nJEJ-v&a(-`4plLgl37L(iA2
zJ`c3(YLweP_Vjqm1foXe{Q>Ou3nA+$YcdU<SYtEnRnQe%k?`}4j*VES@fW=H1^IOM
zjAZ3xDB$9*;|=p~n}o$rQ1RSDsbe)eu8Vktn0C0Sol|QD9(#r`21l1OwQ1JjOL8*R
zK}tooPHFP;(=;1slRIVD>L1hb6SK<st&42d<nsGC-%uKYvf!;XNB=5Jb5VfSd*_CZ
zg1Of+UMD_Y8?CmoHa<Q-bJEaX&pz~a&Tk&q6LnlMZtAHj%n%ELyXD}5wkoHGU{`wZ
zdDR}%y3^govXH{%8o1LocCe+pi{LQ>y#-P^+8gz)d#6DPw|yuSMd;!3(a|<Khpwdl
zcsYNY_OyuMFj{j?XGF(5=;qY=$FC!{mpux1LwSZ8E?TPCrY48dQs$P0vuOJrMhUQ}
z;Za2P{UsXVs2bOVgtKQJ&XteDJ|;}J)nWZH9V~YS;oqL5zaN9~ayq<z4yR85K!^LX
z6N|oBvLTyfi2Zg<wW6%5J^2p)2qoTaG}#4L#XM};6|p8ceFsxg#($u~3?7^^1929R
z7u(Ju<4`t)ITI7}W=-=XvEUe=67;)jA1wPxGAX7oRm9U>kRygtRWdRWRrK-FvHA(h
z10}556-1UFsSK(P^mf^z5Fm{5ePVejIU9sE5^g8<BaBL9QUo=o1!H65QtfAmfav3B
z{P#UMnKC(6ftbya<sMtf_suGhN!uA(K4|e^zu~|GJu)&fP9~Zouyy<KPwIZQv>z8&
zX9JTumiH68-eJvZPU=|9IcF?{E&QO@I#$>ESrla?Wo`S#q{3^ex4>e^fdt^$n}lS2
zaoe<VGUY>j(72lvsq6sCkS&9TW3T9KY8njTC|l~(1y+a$nzv>960nibGB_e^>6cZ5
zgpHv|iw$|m!dR{G2c$#;*bS7x?w7M%j6%pInN@Px)5k+YAn#(9v9?1)*b5!CBxOUv
zBHwy<L1!WFUt|cNB-Ow*<VbdIVc|LgIJ=T?Utl7O3j1%|VI(15r$Jl@c|GPYw%rOA
zX@6D;-I+vFbl!oLGdab@#j-+#J$ibnsqjYxcG>zU*n@0Rv{WhdX*AqFf$D2293^Pi
z$!jxYY*RpyEgCH>r;q?=jZE>ZqER9qc>fcsf{EL&2H>qbvrSJ>Ma7;$8?RCifpeSU
zRxIL)2l{u(#%RlM-0<-5y`9-!>!;9|QE*V>>RR}haT*uu2OmwBzjT&wM;>~%(>W@l
z+l0}^HuY_X;E+fpNc_*;Fq<-QPI>UhLsni=vTjnX<fv<DH+Vl=N>0j*9?_10E1A@w
z9mWrVG36|egST{S4Xq+iAsyfP#%JHYDZd{+R$YDC=}Z^nt7{+ebS;}()5+%5N*N`Q
zTHi)YP1(PhTPoEI-G*krqVHpy_TR-gM-dU4P-~n=E5#8+2??F&nq|82s)n|L$(vdT
zK8$88SV$yBI#79qRWO_T!S?4rV;QFSs*ilNz-OZ=o=>9<O)?63%yFN=Y9Vlz@zic>
zF0GSq(o_R#LxyvNChd|LCLXba&&(PtNy-Nut=`t=b)L+=a;RUooz0!1un1f`o$pJ)
z+3wijcJH{|o;*9fi6}*$gKhE$D|eOwA`DN8K<nv(+hP|u?FB%vTG!bd*?TcnRZKu}
zYo+_5G3YQ<j+TAuZ*<+M%0q8Fe$f+6*$eG`(<qMS)?`hiz=SBN+{gcxRE<13dVKlJ
zpjm4Qi^Hz$fd+G%zFZwLHe`=@cuV_>S61xIqS=%+JlQF9(Xg*j4oxU-eyo@3V(3oX
zQ-oZ|$@%%NW?JvbZaW{}0OzA0Q}^ZUI0O6W-z(C(PWhPH^#*tTNT-v)GY5=H;N{@1
zTHWaQ^k4P|$|G4L4FTGi%9Xsq8-&}`1B<d6XI?w$9DT4_Wi8wxV1eI>x7Zu~>IKj9
z{n(GcQ$DAh06456#>nIQz}0<015}M0S7eb$q~Q4SvKL-bUC^MhS5y0+A@d@?URclC
zxP_eXFORSNFSzeRN=^p<4g+|svFz+nJj?s~Cdg?1SUUUO1%jP|gpB`+2gWXETa(8K
zR)e7sjf&lZOpBsTh@y1Z)|31pL#EiIM0TN11^hU=dNle_RS+Qxr?qyOn>4K)3<d)c
zuI@qt0x!6zSTpUyh^_ucd3@#VN&JNmwLTFvUa5`?!x1j}F-sYI0Fc>)MZ+a+-cv?f
zKIKhU(qQ`miA)};^qh3GX@Ebv`7dOu)W*Nj_BLMWuHNGMdh*m+=x+-zbPz1yyFOX{
zI~}N(7sT?(Ni6Hh<HwJ6^5kr)jusxLX^q4mS?jLa`i6GO(#NazKt1GhIPk*_Y~+gn
zwAPi)!EDT?U^Y}gTY>k67lW6uk1jqTcGyZat;IFFE~$|rTs7S<zeP;QP_sW&!0G{V
zpT#PE9j`Bru4m5o%z4FWnEVzZh)sU-4Ta2&(7Bq^$SNYzULTZ>Bw*u5xL2sS&u=Jh
z-6o~JM)f#B*~*l1MszAiEip?M?KM-)O=xZV4z~|I#Wfh7X+SBtp#O=YnNUO1Llv~0
zM2&kw`XpYm_>#(;q4Uh(th8#X9=`TN%oAs4AzXyn+=HG2pV1{QLk14e+ehVV0dKt|
zO-(zlx?has-D8%&3>Fd0;7`zDT)0{c!sOkqasyITgEc-=e&(~Yt#S%-vw3LloyT{}
z6rs~T9X)A2z)WLxo1eH}r{qE@jYA~Ao}Pr7R5?HQ@a}(p+@p-8yjD0WED>{ac@!Ud
zY}b@^8zwpV{3^)dsI;^+(9VM7rxNaTV8H<R{PbT*NqMlzA@<n*+jPk7^=&PSxp7D7
z4aWvtm6k)B@nC504B->a#u^SQgxa8tYkk<ek%z$GWAHP(y1UAS9IX!Qk9$d84=mme
zgUJm}hx)B-uctiwZRFHo@NB9%;PcJ`y0m`u%^x4LIWd#Z?<t{5UlP$$T^W`x=0d5d
zvoLKFTwApsy*aUf&pu|QMgL90QM9K@&@NsbwJ?3B>i3jBQv`Gb)PjN0bPxiRWYmB3
zS^&H7V77NE9MeG&D$L}+5meZ>;yLAG<oLS$z0RrNO|Gub#$QWe^>|#R=*V7f+Bkha
zErzjoK4`9QN@Gr3=fymqDsy=ekJ<1o-}?I-uM^O*zb7gZ^W{}JVMAqv<v(T=Il9N0
zY9WNZ>&z0gin?QodGL1hT5Y517)_L?&ITAYU`b!Bz4-V<3AFx~^WIVZQr9<?fc$zB
z2cBo)OJN=mO&gI-l>s+-R^^mM`?rFT(Zmn$eQ<vK_{(4>y67W8R|cm=p+^#zf8)eY
zC=p9I1@tQ=9!i<yBfJYn9X|dE&0~`)g?Tn5NK0D%q=KaQ`?6)3VHBJM{;0I!mj=Kd
zd0z=u8hJmGm5?qcF81jG<u{5^=?zxv1Sl#)L`k^tms5&$=d;h;CH!5S>`?jhKN3{9
zvG-$nx>7AwH{HV;-9=*eF4wmf>gOFO7O%Se`g+pFQwoo_Vm}(weCH>u=ugs>U*aFC
z`o<D`aJ;wYtfYS63*2374+;zMZCnFKwq9PLZgFt*eL<PR_mo<T^ebLz^wq`uRgHzC
z|3ytt%x~lYU#~JFgoeB%`H9)y4@Ry!J3s!ZO#a72Qc6Cs=_(s8Qp3JHDD^$a-&8&6
zopTWK4Q<%K{l{N)WtW;<bKnBKhM)Err{c-A*R_Ns1D1>3<pNKP%dfAm{}NXKoNlf1
zLn7fOibbwkpY0IxzfzTHKcrZ{UTVt8n>1^wijN<8J4c($%Qd1M&mt%1e|qXpQRx&=
zHJELsoIruqCeL#ScN#ak_|Z>D8L=M(T4l0pv^a_$I6u{BDA%;=LDOmEJM8fCQnlo~
z7I|G%GuI~cw+Hsb^*Bi3k~<t&(viPyV5(W{I37DLrO?qio2bJr`CRe9raXau%q?~`
zsJ8QT`^DThHzNJ2x~_q(t*+~j{e1Yo?Oq+u4iS25IcE!xRpJx0-7&9;FvOoW9Jo6Y
z7Qk{os{G@vKQK{-gsYk!7)6ECOCd=%=-2NR5RQ$%*3wr?4wwx-j?{+q4LLj>S+E>#
zJ}$2D?pWte42JaHnQU-$UucbBmENeRaB5qS%Hn;%wpp{1>Zc`9rb^Z2BA%ijuUfdu
znk)RSGW~@-9s;PXtEeb(ta*Jz?ab4W)ZkR|&h--YD8ZO5a#H%{BYe`78^ZnCo~z0A
z6%7IEhlDhEd^xoKramN&;<<Q6sR&8*|INEAbMF)K`Nz9f^?xL-PtyruTLKn1Sflfx
z_PT4IEdU4CVfev1dHRq7Yp`8*I!Va(CY92v(nrt3h6TXH;U0Lh#i9O5Q%9t^B;v-z
zic6>gm88qjNn(%q{+dAt{86p=L{b)dPr|AjA1!tGumx88%r&)QsKV-LAy6*JJJFZ6
zCLZWFd)eSYaS-1@uGn=>(qSX!eG}?}5=Zze8@^2rsks#mIJnM;L#^6E8L7vF(@Pn9
zx}eIU4^U=e7tD;CRKmY4Ws5$N)X4)oRW3ZN2D2w$g?m1-%e7lA04P54ADo^Z`OyV-
z<I}C3T%U~GZM!3n&t43zBn{Rj$5MX^_Y0v@axqOH4!E>+H(YZ#4yYr-Q>m@3_B4@Z
zp(ynxb-`>JBMA6uXZ`73&oxse$<|7{7V!s00E1qu?|4*9lsCqi+Zp53q*g}Tte&*E
zNYw2FvDZ6^@uP;+N;f{JISj3M7U3^5WnXL$q)3?GeKxIVfX7V+w=SjXs-jW`--u3$
zkB}_pBn*1Nu^B&?2a;AV{<+B0Bi8c4+o3U1GADo$N(+won=z1c_aw&@NL-#bEZaf>
zd9owfKuJkw?!q<CM>{En)j5ZtY7xZ+rXBI8XKu|6c8b>ytFvU}->^LUWLXURYFWM4
z_Hw9dv%zOlRMQ3uE$qa^RJdfKPRu;ab@Rja=L5HH>GH1kEbjKudqF8_2wv|`%WcT!
zckoO!UHzoXUi<V&g2L5#*WMgfG6h%wxBh^-5Lf4CF2r;9YOCeufmN-UTqdg5kG0?P
zg=bMK<&4sBuup?;3OHOYbQsa$UMaq4kXMir@`7i_&UJD}dhl-7JR~Em-Y-tY`6@dS
z2e&gh>3I2Ndyf1;@oGgqt;$li8l>1T8lGd7dbq@HQ{_5Wgr(ACjih|a2c$$S^Q3Ic
zEM91Xz`MDHPM>k7!mP7-O#58(jfdDFTo}cwTpLt_?;xi56DCE0c(R@4E0+D+{x}-^
zM|*9_qpP=IX+ma6(re+a#z{jD=c-XT)}I;vEH<%`$>_~0p5OzNaT@6IJ6D5}fweT9
z(u07BW&zw-I8G!^FhnLp%c18G3TelpV5&UJv0m#TKU4qL+l0STwELo8(`lluCPiNa
zrGL5ZAqkch8Hs&;JjBXt7rwPTmTsklNB1x-PgJRJv@BQXeu|m~Jx;7>Ed1@LL|=n{
z(UbM8^fV>BG-%sMuJgD~sc5>$Gdw+D{=NAs!u>D0T4wLxbv;pn5qEv@fWMEOgPsln
zZOp~LqC=c<qJ-Sdnu#VgZeLmcTK9Rr3caZ>=4~Z=E`KDNnsem?K+HN}!^GsASDoRM
zT%ni{k^X1Xoq;#RDgJ?ht&JTw2T#XOrEO%{cR6$ig^Q0{pF$)0Iu1=fld=&4D({8O
zGx?=huse<$-7bMI4e>D7vx7ysIZ_G3-Cx*=BIo4yfeU{}<I8E~A{`w%GE(#YtzdX%
zmdW@76#~PDO!7r0qi6F+N6D<}*oW!u{Uq^BQJXeJFd3*sjjM@`btEW7gZ6OsPBSmz
zoWzK4&M6p-+1T3m-k2Tm-vmNnvEFRz2M?`yeoRYavW-lLA^mEC{^MdW8_Yw9hxWry
ztrI&tm1a{NXO~&8Udbqg+-*=W%byB@uGPY9iyn<i4dGl?xB<V0NY#HXv%d4ZN}C}1
z|4t{wixtzcq;-Y7!2rn*B!zURBzmk=g$>yKsEFCBM4`*D?#AjhATTuau72dG!bMsJ
zmY|l(ZOnT5l4jX5$2J?{wC>|Ha@9(~w)x*P`~wf{{(OS`73VKj{3%n$d1|Fp%7vhx
zW&AN)E(|t50#;CiV@KevxwE~&&~UB2=BkqpB&BNB8i}K(m1#d3+ATV#J$OHu*>o>w
zIr}P_EgbuflD+H`P=m>0wwZnFD#c?Hn=@)+@&gup5L>S@C#(bpM$V%lIQt!1(c&m$
z3EHqXjK{&kc_V51c1eCAeel{Sg!_r5FbSJVzGavHoY?DV{P4%TbFDbBMXAw==*-u=
z8%&guJmp!^Y<fzAV(}W;U;g;n#HI*2UW%qI{V4_AYb$<$xJ!qp9h@w3)+pC^63;5W
zTo3~{aN<Fu0Q<l=D!%XSc&pTfVk5Ee<BM$gq?3gW@c*Xr7PHN+4J2$SB0;d%zIfCP
zx{UkvvLqe0^5EZngE$={cEv>NVnHDz;Tv~htgStRz7X(kj62yDilcG>Suohh_NHZn
z0*(iGyGkpW{Vfp6cO}{GksK&0>Yp&mTdgXs%)CelTM_S5k9h0f^e@E6X>0~ntn*8`
zH8I1m!l1n>Ez^GuYNNAoQ2!)~{wVkN4>)HYvM93=A0HnJH=E+FD#u)wxYlcylo2>T
z?8dk0CfHEYMUQ(%v;B%@=L~uRJeSr0e*Xt{F;d-Exu^1Ti|2z!K*PT^ZV!Se6sKJm
zzEv#9f@nFRy4zR6u*XlbM&donNRL4~bc8hd{b=kgXhVI-cz$|)&P2TpVShySb$E{v
zdA?Ta-g!2AyRcv>Ht@Y*JS5GN49&3ex<K>kE3Astcli@qVVuGGDyBcDF5<O#fbvG+
zy|2wLq4AX3B}xlKC5HHau{X=ho`ZKenxBcErXI!(R})$AgYw;@LeH<itUEoD{7L1L
zfg2-G8klA_><KX2nWRvSCH)cLy)(0Ne6xbbp4xP~bnr$xNS*j=_Wr(Hp5{v^Ntda~
z>S=IOgY1A*;WRx<dAE{?)pZ#YK*C0Ew>PuZb-RCY6720-4HKgOpo1={7O(BR{IoMy
z<H(Dm6v*E+gZz3XoYM%iL8&MwT2CLz`k~oqpzoWuhWL&r^`ctn9zO=_G_UTeT(uVG
zn%vH?XXJSg$@_Xlj9fkRNg^p%NQ4!nr0l=E?mYfiyRoNmxg3O)o({PN&X1jYErrb=
z8GAosRqdkhqGEG-#+ktymTu8n-|l6O&8>W=e*v=EsLf;H=*Znx!8k5zwvHNSOvq<D
zUbzX=SBQy(#jC@|AnW|FHuQlKETIqGa(a3?>QNS_XJQ{aaP(UE<!XT0lz#Nvl1FW%
zpT1$uJ)c5}#Sh297U&XSXX7~Fvt47kNvsmITBszcw6rv3fKw}03M^J}kz~@~4S^2G
zuHcH|{Ao7ROGAM<>7<R0$w^amh9&=lA2Ti}DGIbihqmuV$Rf2W^;<Noezw7N2&$!G
z0HwmAB#%9Aat3#rf*P$_dK+lr<}*YpT8a$XyC`}22&D3y02PC!k6L$Hl`~~!WL`{B
zDqVK6kP!}uh&SFL5K<fWABa$ej+wLRty72R@u~=Ge|wL<ss!G_Z|``5>g@);uS)-(
z$9j~&_EW#4eR(1lSb_SFh9FE#slbx;P2iu1-<%!W$dljhiCZ6&KZYr)!=P{R<P3cf
zc6Kkd{h}Gi>4CF_{TiZ#=s`EguC=kr%<vCS4(;sd`oAF;M{B47CMe~O1N+R4udHQ|
z-y(eTZshOkLneirfp?b84c8eexd-z+K`Z;jR6l7W`?qu!_6|O=il%3$2(WR|wVU28
zDND}{w^_y@f02Dkq8YevF8MZqfLw+{4=0iyhg{S=tOQW5s(@t^>50E+dF?f@B3$J7
z-G-w8_AD2*6pT>Wsl;C0?w_6A-i9I<W2_a(_z9?D-3GZR#wAN=a1xnfk|T5pHsk_o
zJsKwY!}%%(*^6mjS*;H|$RfA264>mx2P4`s_Ln-eyz~)eJoxVdk4m0){KNQ#mb`(*
zl#f}T2{mB-g9qOt^!8$?cH#P<8OhBxva@~tZ+C6)#_MHu^V8WG-zn+@i;iGXc*xPr
zv0A!APbx+nd*ge$P^iWtA956PO~#a#`>ZNPH#4Kcl(mdB<~U~9E%@rpchK(U@`;Yk
zUK=G}T}MR0N@_*Lk)l$K+pH(k4ou;8!Ed0Y$q|sq62N<KQoB%-E6cnPgjufKF{-=~
zTl;SO+aVoux0aw_Cb*#k{-HN!q>=}D$OsCx!JV>g(<Mh>eMzAivNo$JQ$RfH!CJ6q
zz?Y$-aG$r148~uJJgv>LpG;f?JPeV;{H+S{4|wpfjZ#RcMxS@_Vp7c^Cd%Gbqjly)
z0l3~<bgKOE(61-7u<UF3+{G6T{O?^@jiS{e<<PeI`=hCZ%F1Jo3>CALB2=szWoA3L
zWtdEHt(Ji6ljWJ-z4>A8yzAW=-;mINg`DB)9s`Y;y;7{g^#bNPM#2C5R?WBxdAB|j
zc(Xk|l4mx2vXW4vNdH`bgZ&T0mKI;I6(FLh75qszX~nPnCZO0mo-}Jp7cC{}K7_1V
zajJ=fEM?2Q;$I)?ZC<$hd3KK~shrt57L=>7KI75-b(pM*re`@N12Wvjw&NLiC8^t#
zHn9c+KdpJI<7jws)4g7k^F>8)e^~+Bv+V1M;n5Y1(mTfY+B$0a@RYOKwyxw~0uz#C
zH5F$!OVRjs)r>ESH(VXEB~jDZ5T9%xTu7GGYr%9PGk(3dk&)gCix?~+Q;F6;jsT}V
zLjE4%C7DTguzd*=gs9$W#RP#23vdx=c6BzYy{Ud;ljA)uCO}Kd{{6!E#hs3~6?bvm
zw`XS&5Aj5W(mz=DYK|RBM)tM<VG#z8r+2A;PQJTo&8q0!k*w`gS;GA5oX{+xl7?8I
zA(S?T^_@JATI_B1?rq)+ApUO-pEj(qZ8U(?o%;bE6A;5b$*Svx%MRgZy=?s4CqozE
zt(;$BeRU^>k9;lYzCT&aBFx3OeSh>Trqq#A@Y|>2R4||MEu-~$s|ofTI9%b@$tAzh
z?fn#8Q|o?qB}X^e^G-(hxm0}}wzE77_80=|`!j4m?DFzSb;}3b{K2O}qy?|Pn~{I6
zj|fAEvg76#^FY?KuR8tWUBb}X3{*3;&tL6FLzT%DT~u?vATD%$ZcOv<CW&<NfFpz2
zjp3@9I?l-pmqU<bIq(6TrEeb!kXFcB+#4$-qU$l1zdETG*>`t0Qdx=-LV^U&F9I*D
zcw5T<s$mpn_OHO_WT(}>D`DX-F8htepy4XN?@h*~RZyF0dih-saQp$dbpXD<&XsVT
zYCzxD(JQ9(w(__Qt$*$)?aYV?o^NtwiX0M--QFI)T;H3wZ;}MOPhFZdu-G<-5eos9
zcC64{o5Jk;;$nM1fUl3I&tZPv;N4Zaf}S8nZgAE_ZF7&CX8jsJDRrfT+YWS4YOPc!
zccEq<{9@Ps=*}U!2n4nYDUd@1HG<LvRj*PAz4k(uSJRlyb4x=vST$Gd7i%vaf9f{Z
zAHx_G5&52*X@HVh^Y0m^Jd~#gnyuRnnd>L&#bau$U&Tn2f$B;bNj{sC-PxZ19C}Q&
zi0ll#Dh1A1f;VP;&lU<PYHI#xBCocw0HA+-E$Ub?Jm)Y~Yk|$u+LK$|+TVM!B+~G;
zI|{0T165Zv>$lE@&2SiMcW|)d(t1C5Ig#u|W(5NX9XJfpU8CvKUaF8cJHlE9=&gS)
zR$p8Yh_-pbWIvzY*O{%Ka{p`5|8yMvNqlV-d$V@&{vz{fnEoePLC9~H&xeGR-)auj
zg);sdFN)-r&rGC2e9^PYENOC$TRd36jzgjKH8NV)GQUj58C=#pMQWMHYt+%Dzh_Tk
zxa@F&29U<M{qeW-v?h_sk1xYjElz2_B;%thc#E4{sXXgves6fU<q9C10t;MU1=UrU
zwpgO?71NM1a+{!eF(oS%pHn#AGH_!$ivKGx^=e8A-wPDY8`jt8NsWdMtnpH@v(z8$
z`8qGJo+u&Zu|eE1ec($ho*z_qSu(I)b+H0he**2OdM?J@5hqKDHlHOhRl?MFrwmHe
z(0AVXBpT|r{1Q--FrK>hApgD&jB-VRm$Cl8j{4rRCXt5}6yhICVK~KJfkGer=EXqt
zCb(Rn@=~)UvWA7p-jlY5MHbm+4OF(~3VlW(E-3k1e^pd$OaJnJ!_7qdGW;Quc?Bp|
z5&rY62Z1o8u%q>U?#NzH;;213D&_DElWLyGp=VH2AosCf;)wT*LU2|nr?b7|Y5^-x
zo@Hj{S8Ifg&u^W>TrIEcKNH*!4q)3^_~3tgaC94W2^ZR^S79_eF)itHsP9$N3H!Sv
zF}1TZ_XlHrLhl#?+$h|P-0cMZ0aL}J3F<0#OyYix5&li*LwDzQZxPvzT2Gw4a)rH@
zZ>pT5<|n7-E1QLUvs}MKdB`)W1jicugTwhI7M@NahOg_VBy8U8Dc4sf)GQ<<Fux|&
z@j37zvqA1*dwMF>5M@)9r)Rh87GN33QT<4sYoC>-6rH*F?FZ!X$kdK&CP28Jt-Rek
z<JL^-(<u8&yT{|~*1$p_6Xy8R8_~=b>*zA~XhTn_tddVKv+Y!z+w64*^6d{3*RI|g
z*Q*=(`sfvV@yEtY2X$L!2g=N!R5uJpjVI<hh4RxlD9M!V2IT`0)AtG!lEM()F7ur_
z^~c8_5){r?$C1}F^ZUR#3xM=HJb33_BcFG#=g+69)63zZJPWhQDU12+ysq9azG<me
zztD_~A0hF|McfEQqD|_;Fv<TEYoZx`YxTf(P&Kav2@)7Qo^c@>Bj0)>ReyD2q|!lU
zo}!rhr{^N0W(1FDgt{OBgp!y_Rtk6b>UP!G^0_oS+Fo>=raQ^F4j{uu3awMmU_<^F
zXRoNhfA<A@)uHvlbjxFUM!?0OBwts*$*~Y*p!Ql@-L6FK5-k@BLqJNaRtkZ5f~|sa
z0pIUpnzy!|Xq->sc&XhL|Gj)dwZ>~=?R?Re1Vc3#hrw^I9NQrf{uzaGJ%Ok%KT+1D
zGJ;si9AD|#PWA%epJREhac<`##zuCRqORFpR>xq9+pRBN7qKTBj8}H2MvLG1V!nTL
z_+&na?7xH=qz^NaFmP`Aee}`<V-`^HV0A~KxtX{^5kFpi{;7~oRN~nuUw=hQKfI-{
z6P1>^5OS*O6wQ6gPl{bZBVAeCNan`R5ZnB}Vp>y8S_VOY+V8*E4uNMv#r_+Il!%`J
zNMk`3S$~+d7P^o~t+Rp=)6+&>5qLk7dG_9<k>qI-uw|I1T^CaIJJ>WS2CDk}BTraz
zBJQjHhVX%f*;=B@`rhlU!agC3j$Q1H*SGT?;D!hQZPbpoj-cDiK(i|~jGQMTcM1fd
z9_QWsBL!%Av#&yfe0<_{|D$?lo-Ghlv@v4A_YXwg?D2c^nU;7SV{Yf~UJeVs^s8(w
z&!qIujp`&SW$C=T^;+nKAz0}SnVDKAoI*lFJLmVn9aOwh###C|-H_*?U}WfG&_>b=
z1+p41of_tl%bA#%p*1g0_|EoH-6zDN`>coHe}MeezZm4jA>BqyeF8Xuw`OEf9D2Mw
z{=t2J-)-Lwd6$zgt4JXT7T`{PnA^KttE|18c*ME}Zz*}fE3|HgY`MGbW9ew`47glY
z%PtM&Vf!&WeABxo#e3NA?rttXH?-C(f{|!zYwp<=Xk8K<aTuk@>QLM8_GD9hn_|6~
z0Qex_3Cdj;t+sAiO^yt>rR2O2;!ES^PVxuo{7Ya3y#bZS)3SE#$qk*m{N2I7n-4k)
zmYf?ranCbv<|bn9FgkY$p3`wRXD;t_7AongQPN~mTV1y2OV9r$DHc(d=lXIKjvf}%
zpo?pE#!+Q|HkPu#&TS1j{0cd+B~cop1O)I92$fCy`c6!r*i1&JUroi%yc<Ri_owwY
zN!zZbhNqmD`A}607z2a8&WfQ?8gp|3y`D3}g9z<PVZqN1yj&6M^KKG(BL=+cG?kL%
zHe_E(;94o)`8Q&Q1IN+Qa?)kSmIn)k-r~l_Q>S$pc%(66>&K8Ea*^dtN@_F@z`~{a
zwqdJzrw=mA+K34Xx{5lN+9+|X>asL;&v0W849W;0DJ`@7+W$r@W$u+tQ&U_-H$FKe
z>_1p8tPrmB(zW07`+&!ye2&4P)dQ3S&LuVB4M}ER=x4t(k=bGN9=vO#u;2)WpHcnv
zg|ZS_-b6CqWH~8WVeLFR8d~-fu)bvsV-57;CVXZ;cM|tDwIrMTN;s1MH$PIi*paC(
zwVHvH){>k}9iLnkF1cfWe9>OcsI0|1UprMfVYY}Zrm<LliYbIC`FMJsu5OLP3b*F7
zN{A5zALt3L1D`Tdh8a8KKnPAGr<@6|)@L|qOK5)Gnn}*^c@x<2o{3Wchq_On;{T!z
zpK}x-iYd1sIvMx9s|~%w%HQ+`@U>_7q*Dkk++prEPls<n5Y!pi`*#i8^dZlevB;Z4
zitTCN^mMwp`Ot4Jw8unr>E8IKx!bKFAt4-L$H!o^Ye##qB9r^esT;T7(opKE1>BZ`
zVHzjiF~d=bG;jIy!7D(Q#2tIJ6cTIIDM|Xx&Ea%JAw=9x-U8h=gWFH6I}pHx>a8N+
zU9ieUiC?;dNVuIHALhwUuKUr3R&SrpVEQ_V&nMP>|NhmOm6Z&d4(vc8tEFzx&GHc3
zv}A17UTP+)187`&BpY=5uPd8@!QH$X<8zJn>%6)gLfW9N06Ng;dFoR?BO82f`D2d6
zd%~xtNCI@o8Gccd_lRTXDk^G#!*vx6DTmEwgq%A|@J`R-?X#*pF~myy+}?)@&tyU+
z7VFi50;N?O1EPF=W1JZE?_;U`U@PuK>x`8Ti@lM;^#<|<A|_49PQx$D^XhEENa12;
z@$Q2oQtJk|l|Z)rkS=}X_41vUr9!ms1Fb%7i1u@uhfIQx8J&Mi`Dq0WauI*4ab{^Z
z8z9{L{KOD3LD6;*oEz2Ex0EAFO82&iK!0Gbs=m@pS+}uB%zZBPQ}o;V8DTQBc`|8R
zQ9@o9re?EgP|W!K+!Z9Mz)>ZRAY~+2UIC}<sg(~vhNl`9N%?tbO_*%BO7y(4qJqQ<
zIZr>_x`=Uzxv;8A{Vzc1cU)x%)P($nTaNc;RiXOJBRs_^A8uy>QbG@=s4*LPyRG)(
z-wQ#aHQ(_6Dy8d5qZVMkL72Gi?`-Au_4V2Hr?S6Ccf~ZFDW>;)V|v+3$sReX!1_ix
z(z;&)?)Q2WmDQ-^a%NlB$1U^nAC3*2tY;}+D_r)p-8H-Sf9!$35#^S&hs%AVI{Z95
zJM}TT{~>Mdiry38%`&r!Odz{FL<L4eRW;@GT#koX#DRdCt%8IDa^E9q;4`AQs6w5U
zZ)bSjt#Q^(!tB)?3o|yeE}AtlKWy7&!KYpj`1SL3l+_Q5>dBqc(G+r07VBpdh8uKr
zU>feb4#`I)vy?mOGZu}+9tDaYz*w&_RA!-*T*v<Bg<Jp0?M_N;G!GAL;1?Cw?iUmU
ze<YSS%VV`mFu_+hdv3@%ccUp{N^FR4W_rwp`HL^8y&JyRj=I+ZPJW|?@VBmV_nsL{
z0JWWg<HP8ttI*vKR*6IPeG;|iUV>a%e_kYp=~is^dA`XEm=K!}rhoQOse$71Xqzhh
zyn<G6frwygm~K3^hPfSkzGc8t;kvkJwn#qypK9}_=_Etn{Em?P^YTB22HI^qIUchC
z2h7E(to>L&ipPA86=O|R`nj$2ic8PkrhujV9a=i?#mN9BBS}}^t7eMKuc?|mF~>F4
zm6<h+e`cSmMbPT;554T$Ga}C6+nuzQM|I)w@-rGw7POvvP+QRqc6REK2j3hGb5E!t
zvB)Z-v(8W7tZ6rT#7?6{RrR_zE7-o_$`quUK>B|D=At9slYKedFjM2(wzkHhN(3wr
zJMi|bU(N_G+z#YjV|qZ4Xuw2Rn{-UN1N=0L-^GHT{W~U+O9P*$L{;YV<Vy}Cwj({{
zS9X;Li%CH-(RNkr-(L%wh(KU83_l=0#kAN^B|9Jj@b~GcFCkU)oS8v0TjYe3ok@k+
zvLaM1yv>mjVTNh4zX&H8S{qwe2OR3Z$gM|ip)Bi!*(<uAi}<s?Hx57k?c!a*r~kuH
z%Qd{9zzYVlpbhMk3X(NcA>J4^wc@%@e|%hpt;+p)iyax}I?eh9Ij7C0-j&@`g@wQR
z=VH`$N}+k-#u=vaQnE?zH+PTy%eCp*Cp-<*q20Qb#AA~5v5y47XoZ#7JD^0a6hX)x
zM2L7OK*+)H#Jy`Q_W}vsyki^ekuhn1xH@&LrIsQh&z^L66HJCDzPFXI$0Ds~s=Ug?
ztZKfv$IcVO!z^~9sEtF>R(!ykhR=H3%gL#1yBxBJ@ofXEoM|XcY<&R0+dBklN&svz
zl{pRauD$*d4K2uLXZ`H})YrIvEVG@#UR|9AA~A^G@WfU93`B9oHu&ToACnpo2O>yt
z7Qgw(WDEX$2D`j;W6UQ|E&$#LxrEpv#z`qh%m00g*3amwXs}6r^4X&EOY&fId}-*g
zdfT2;=22*lx8B=#SEodDIc;yyw8w%<WItX!+lh=;5BPx;;+1OyI?vd=$kS5bTmv1e
zOx9n9aRa7zOH;pXncC|?2{NpGPB-P$Jli_4%4H&1ew4w~JB}JLZc)n5&?-v<DW&zj
zu_~phM%Xz}TzI}mIo|N*U%}WHp<`+7{+(g1@dmTl&h8~W4gC>a!d=um4c5$&qzn1;
zxQgAQOek@fYAqJnqoEQTX0muad#3AM>#I)5Ua>$ez0~N}&noDztIvgzWzFYM_S^O4
zfCf?Vl@sJ@P{azlK9rsg7OxJ?oxgg34z4P7oONwRAj-RiW54F^4gS(#w$gkbjZYc!
zHrHB)k$awuBasL*YvOl$@M-AeXgy%2_T>n0eLW=9c|HS-gktkTg8p`Tka{F?KBjV<
zbkvyCkXs{tdBZ%>rWly-rg<Y5Z}UJly_+vSd#mJL1=&{!xpp>sx2}{o@~0G=^;}c;
z#rk?lwG2@_6Y(=>XotzLeSL}Bz5d@MsPA3~0zJjQ6d*}3Ygm<7NKnwXZ9dgNvDvhm
zpZke#cf2<spQqCk9Vu=dosZUIuJ=Rtva-uF@8a`I-n#+4CYxBWs!CEpfeum16GpD9
z{Om!FcZT=8GG{ibJkBTcQ3X3e-h29n3cd$380^qmGNSp$ZDClFZf@D3fw<DRkt!jo
z<E3xL0!q5MV%8_;fsU^@|5cND??C(Sv-kHOzN`mK---?HoSik1j9S~SwNtCF(6qXV
zXU)4Cu|;m)?Cs4zWiK{0^1(f2a}j5BmQPrDGo*Q*^{!hX@)S6qv#{FOyM3lG`ALy>
zTqKj~v!9@}j`9ar4qiv-hoiN+N6*zq^{4S{5;(iXJ5)aW$%rk|!)Ilc%-DG#hz?L5
z446Fy=7!xwOGY!1EvZ*mZOv2e?r!euF*P+xNwTpagB<B(tWG<ox*NwW1WUimG1#`7
z&s-mtRH&%FaHWTbY0V9<$Fw>c<0~o`!bLH;EPA?$g~xf>==|JUQDe1h=}30=Bw_3d
z^ih`*4TSUqs^`e_<x9k$$YMU+#n{#vm#G|}-*@`4G{tnIKv}%Owl{L7YI_r}WE00Q
zDzw?Oc2MjOq9a2<iOgtdMz`c<zf~AdZ=mh!N(;$SVyHi6H&v&rw>@i8HmhG(J*N|T
zIWY5Y_ur9I{loMjf<o-S(vLf>ICOpNbMvXCV#xNYaEi{r5O3}>zTo-=S=m?pb;x$$
zXYs46!;;-gQJe+Nd8?-n4DqS7nFn!K+wtfqtsdILaFYrtm?y*j$t8JL4B&fL5WMy<
zhD@~O7r~p@Rnk7T^T~V|?d6Djl<T5Lf&1p$7BhQJNh<2E`*B0ULVZ-c()d58)2-op
zg&rb+bKaT;Ru4V?L*YonI{3V7`x9;oa6Fpu<yjTp1+NYndEx&4hn)#CJvorncmrb0
zeAdS1j7H3n=e=&o&i2=T+e2QC&i4#8d36UYb(2i^Ds<1aRT3F3lz(S>`=$&MtFFO{
zA5Ehq{ENQtkkz<q$CGJ1mGW?RH}(r%%zL|ksC8<+2ZfG3#JY&MkXvwGvL$)|z#k@C
z-QDG9)@nT8EO7hMSzN%iQR1vM_;F!IHi<M8xghQ%z@9Y>vL_%vCF4ups}cD0v#)k|
zveu}4FjV}=6K9}N?#KWUemW$kb#-&Qy*&s!Fd2_uc{2-h-e_`_$r}Lya}Y(9T}WcT
zu@`=Ze&(?KZ_O?vtsXwy%qTd3+DwdXamqw(Q+Ovzp#Dai^?*q4x`48Amx_1aTNtvp
z@2^CYo#;4So(7sB5C>hC)Z(Ugkt5G19mCv8)BB54mH(9ZR@2uR^YfF0dT*~A{m7)I
zy&1I8PV4(PJhc)Tnp+~qb-cmGF2$#NS_UV8P%q>K=RRHjFQU$@o-ZE=%qHj39qoBT
z#Tw>ltQ?&%GN2#jsm}Xxu<NfX0;j}rcF!+s4`eCee%1%AslIZim&K1+{lT-LNtMlV
zw)b<XkLvSc#{pm2;>uyq_n8FfVwxNwiWyJ!P9^24>?bOmF4Z1Dr2Hv|o>DozkC)25
z9ix}eMO@kEW-6Ax?qUFBr5p%`n;Q2ht-pL(W>NkJ;BWG7W?-dG`E(~5SHT;Q<8C3N
z&`71pRO7KpEb|j;TeV_!ZJVv<O5{)hs+{3hv;(|Wdl>K9D(3P^R#irR)vvp6UyJ>5
zcWaw`tWG6ZQMp$lALQOl?@;f{^qnL(PsrWPoQS2=Dx>+uNoXe^3mjg@yuL4;XIx>`
zcN2dj98=E4mM}PgBdNrx@*fP3weXoyK&<itVvmW6)Vu1j-UD5BW#J({Oo(~`{N=b<
zrg`{#rK$r(Pt`(6lKhrUPhL>*P*#PozPD;lv!du8l)CT8Z1$TsC=>G0?lhqJTJosZ
zEFp#P3rrGD?m7-6aZGq-X#Dzg%^$(++rWEo9Q3<<7qO0ia|IaE=Ch0}YRnb94R3cx
z?oLLn$-|5&)>m#$BZf+RQX&jdY;@}RRhPi+qQZp)aBZRBFH9#}VexYoTM!){sV@HQ
zi+5*hD?{JttRLt_q`w%4GIW3!0Wu*F!=4yLjr@x_11e&pd5eYeI{YtBwp_&LT*RZS
zO9bE~bLRb#8HATw<GZADYjR`}tW2qF)cotmp-2huzD`U!s4MjLIyK%;Dyfi4FL6}n
z+K5HhJKi@$ft-t&-tOFSlK0fVjJG;wT=)aOe9So1o7+2EHMcy8q3E*3-1>j(g~hnv
zPksmiAgqw*%1;eRHewez3LO1|9}M>UAm}w;**hVk4|k6*{{E5C8lTp-7w!1g{Z~w#
zdrH(Gz?6D=9|tmSFtTxM1&ex3cFPal(lc@Y;Iw<jRwjx23l^df>&wvRsmXgomwzp^
z0^$%ydt{CnFYJb_sg}EWi<3puy)<1YcI)K;GO9JndzQaDxvrF>HQ2kMByZZvHE#-?
zbZiyfYaIH!*^3~>sh*lL)M35GPo7Mt^0xH2mp+f&6Ta-s8~gQ$f<V~XRXooK;p3!{
zRCp0wqeRkn<5Bxz@8j+6Zo4A`n7uap630wv+RGIE9zlRmJdZy9`I{tvU()AL8*b|W
ze0hIm1+IkA%61IgoYq#DU|z2DDoa5LTzu<ei{Fdt=29ol)Y}q1!vV9qO_|=Yom+Jz
z-?Dmjr4qQbdSZQjWY)?+caSuxp;|30^oKX3+M67UX|!W`n8;2wq=+v4YN)kV;ZeUb
zZN5(}>^r|}ja=ll9sZk;az7_e0QCz%o?l<=?QP$1QwY@xDxJjCukEi@&P~+@Xd>p{
zHFEf47wlvT$?#-Wr=#Gc>H1StPfbV~|1><r`$0K+!M^Z;GH0k5f`eOrKY%XW=smms
zgb%vBg07f`CEI}JC4`ex@SmBj{iA+db+jHQnvY&%u*VH)Mx<zYrtGvs4E+JvhTFri
ze}3NHssx@nxGx6#UhFNlEw<R7DUR+<`q^kpk|D;Eu&xvFJ!0P=g<dToH(t#hLUyek
zL0MxN$8{|Lh{LJeum}zG^E=MLgp^i^!-Xk!5^~MgPB(@Fe(ICpiQ0~4R+M~3UdJG>
znt|)uY`mX9(f4}a#Nup1O+Y-r{fDKsIb-zR)dHslZ*eE4tpv`{_p>ZjBx}N9BW=0g
zs|6qO48Gz-rt13VKVdz@y=weFBA4a1kT-La0Yn8a<@1ymQ8H6La&hC<Et{o!RN7{*
zH#+tRK|AnFHF`6V;DhBw)dVFiq2`qTzj2#b`>XdvM8QN5KR-XT6rd~FqF!S4R-JFV
zBTu6V{+L>R_BdmEF(V|XW!T<E4mE6FSV2cecXh)|C2%OAD1jS-T%05Q@_js1Shc^v
zX0D0LPDbf5>u-~;qZ^pbgjjd*W{Gg0N8(0S@*vAf%GYfPaLX7R8m-$+@03AS)uykU
zthWLE-HG+W)cM>FfU8N6kF2=mVaCeP^Yn6Y8e4f%57x$n>)-DhaO?<_J)pRQ{JR;^
zV~I^|d-lz)nKRZc*+T1KKfB?|8v|0@ftwVsh-9>$D`%7PIW1ma1e;GZK14R&2%$$F
z_VRjaxC`vmR&4+K2hQTX{zST5mZ>7y>Ye#kRhhJ&q>!XS1YkL5Bn{H0ba@jtRGuKf
zSD>HK=cp-_;{CfFgN@F3@zAC&8k*0}Z>b1CQx>6{>HTqZEM*^+-U$wPCrMwHl!tS(
zC^`zL+&_s4OqL`_<hz7L_{skY&n1DfM~0z&)MF^|aDTvP{tKwZH#o#zolhyWAQydi
z;6G;8gj0#=03E{^HH^&>Tf6^-FDz=Ns!bB<^)&}pE67~EFAgfi$Jxgdy&$4_A^o<K
zswyJ3t*uGWl`|m`Zm^f?m@JYv;)GG7ym#*%b-ub*lt|G(ya!J?U1ncS#LSF*<#^8?
zR`dLQPNhj>c}4P&9yU%&l#-<K_^Am3Ha6`()wACT_%018L!jCwyJOvyU+NPCaO+N}
zn%nPe+b-ERuIQYR?U(4IBe}uW+1<AK4?H!CZULS4Rl(z@JMD`#=5lwvZuR!huWr`Y
z*TKQ4wyb8l`m+c%9hMx046iRGJC#E*Dx%IewBKH)-N5v{Rt#13qPDT2nC9CX<lXU9
zty?Z2E5~75Unm`?!P1W|4F9k0XHldoJx9UKpbPZgZtU_m+r`_#)wv$w3}QO=g~i3J
zrFsA07_O*~MD}?z9uCvkZ#Dd8Tp0e<KQ$j0=5|NQyWV{3lG=z^bZh?l$>4t&Tv<|T
zs{8PN>$l*77|hV8|D!tVG;lA%f}ggzx%mUoa2e|QQ5=%pN`S>;&rDx>uj+@Fmi!-0
zXZg@%`}XmT5)ffSVl<;c1V)dN7$IGgk{l`B%>e0=&LQ0;sibtLlz@cNtx~rj`0V%M
z`2((3yRLJ`d3@tjt;S85+^*i)V#piED7f_2+%`cpl8puX$t;D5f8x%jEcT4r(wH|i
zLdlKtsbL2ufbEY7y(4F$CLOWN|4ftvzTRAO8wxj#fKuQR*>#J39XX@$kDE6#2uBi#
zEe-s7D(nI8xFUiP{p-LVm$UMP8$oqIdVlfN7*bY|Xzp%nbPbc*E=ZnoQ6C4xg2g^c
zZLBt7!F<@pt`xL#^IKxB^HGBn7oD;as!iIt;LUg2pTs*~^JbZf;R3)uq*zK~wS@S&
zsn-s&SJWsB2QE*^5(gFbemZ5zG?&(Zk}i1eA;%<EM)4c{9utqu=C_66*$zX-z;MS9
zIN(@@;cda6o~mM<BS5urg8ri>B|aghkI#%Jn>`b&nTyNZRL59K@~v&<^pCx9q493t
z`+p}^=AH9D+9MDDuNPWK#uk~Z>Lp*f1Ags(AeR!fZThq-;^);ZUA2{xkf;Ze(o;2*
z7DZ`AUqOPdlm`ZVtFV=Eeep-}^XiE7exORtl%s)n-a3T!>b|nrAd3uDy9VKJsGqLq
zoBLj4Ff$_e;=iSl{EBs3PpZc9M#oYAp-^?07iH=oXw4_{aZTvsMx}SyS}L&I@79`T
z3dPENz1cxogluo^MmwgdSfaSZS9GDVX=1T{3=@wQnYE$qvOl}C4CZ7<_ByDT8dSIw
z%i(#Ahr`&$da9Z%kOsvLs*(sSVc>n<mn7Hv`E{J*M{C@|l%E3{Gu=O3$Xh}e61G2W
z{;zGQ-~Xs%WiRCFzqyhu^I&h$JTc0%vuiV(z|T*hUBQ9fMD5-vUbvP+>pXSCgzLl4
z!!$G*(=+^x>l5BCf6t4ZG0N1x<I4a_`a&H%Q5VlILv;DhTy2}@-?@lBrf6cN%FcSv
z>D{Ub?b9}FOET4j9~$jSbKnp#TFM#~K=6h$wngq8Hy=r4;wR7^)CPWJ_San}>wEeA
z)~F6LOSj9@$0i~8U5qy}Z_|lH4sld1FM*DDhANvBJQ41(el+i$Z<B5d-O6Vz<<=q?
z+`&x!klfd2ixMx7&H437e@XXqnkNkavQ%({8(5Uawt!Rmz2R5Ha4B7@mFc(n&Nkv+
z97ZtBTURdEKduR|Uq$WC??tHwzNBOEBguMAVySxagMX)~e*gP>8mb9UE@Ot!IP;=!
z=U$K}?Op6Kb5Fi0h^4=u1&tLE+rxfofbsF=ag;M{i#!kL;aVFhz6a~S7+IPqXy%sC
z^Ol#CTNVv(Lz$Dhe1Q~Lg0JqIHBluKs>jjYJB5TtAZhrcgV`dh&vYJE+BnQehrH{t
zG|uL>ivw9a%IPdSLj571zkezsBXiJWdX(E%8zR_x5}-X#tW?u@*clGe1`kl;C_Z4B
zzLo!e1U&d1aP0OUXG{$7<D%o$oND6u8z9h=?`^^>4Hj}wOVJy8kraE&2Lf@jhr4&=
zp1?^JRZSo;KY)elwcAdxc5l7okxPo|wDD^aZ#|)!jAziLO!mq+k)9zj#dNfyorWwM
z1;25w(hOf-g_fO|FIK0%K?nZBS?u-`(XPz>gea^xkqWr{@4kZRCAaL^+UB3WEdKYD
zMVk+G|Bn9l{n@*kh{L&_;ip2b&;AbD5Rug%sef$UAzNw)OlB!!O%WX5A&N_=egba*
z#F0{9#6HfyEqGnaGGVA1w`hs}vzb1emPR3z<{1kFR9Z9M+^>lNi7Bq;xcfq0v^lGD
zM?&I4Jr8Qi?SILfsevtaWnMu&iy9|?tYj|_yi)5rQ1gB+W;5>paoslP@59IsTPae*
zwhQ<)KOb`7&(N;zJpT;8@Mc%Q)$Y}VohkDE;KhA#^C!ryhJgVS`nV5$Uvu1ifq!>a
z)7j9o-EeIzQdHhg#`u17vjHeo6qo7iZKm|iFtgi`x95F5N0StI)BTf7)#nNaV(;FP
zyg1*U9bKe1H^yJw<}8!8AmKGH1~Y0iPa)_@A}9M~^ZUWbv%+4S>Q^+F-{t^;Xcbn&
z8c6F0?w9&kY&R|2VjrHN%K*iMbia+mozfSGRuZLPgJS7YIb((wg(j2V?ODGqC8lge
zL)_?ph!KjqjBo4_#dp>>SF}DlaIg5%`R~Q<(*R6d>GpJkZ&~IAwn$0SBfqrZ8e+@g
zYlt&F`6#OHco=4M*+hUX04mPyS&8DFWBtZt@o;%6r@f%CHW&3CjR`a0E7ne)73Hkb
zz*JWQ>@?vpA;M(N`CJvSdpIf4Xpx*0!(<F4=`$Q61ZWfrNAA3(O~rw0O=B3lFE*kr
zA=kkLY9YT5q^bL}httaqBRk>>bv;sbRe%TvavtZik8l_uHuS^$=QHTzZ6@A5Pn?C~
z(VtsK-v^@K*<GNU3;z9B30aPhQH|}(FPI{GtVnp*Xt0oLg6p9Tv;oN4B;$B|&9Uj3
z7IpIe`SZg?MW{+)F$*tmZSfDxix-lcr)@?PR*h~H!>M0{nL{1($z&>O<9I}G>^99q
zD)R2MaPR#&5o||*Plwey^ih?%^0k@GkBZ&|?ztHnoGap?V~rLb=_<-{ZsC)9`f}Ut
zGdK`Nl=3PbNf68Ng#LK_Bl<$(Xqx@ox7HVTX<EK_KjgE2azyNXlzAvwsja7O)!pw7
z^t`d{4!OftSacn%et&p>qmX?jS@7D_j+1_WHURIX4U53N6ZiZ9ig{oAE$cf7%^3bU
zt{}D6Mo&*fIyw%(0;Kw$3ZTbzn?Au9EKcU&G&K>;sBx1Vm09F^sfGnv`pjuzrVP9w
zdI@~)RkqGR=nU9&-0>X6LoXkRIc_sL{ZOAS<Yl>fGjr7a^BRtO7PX!(Z#MYpuf4yh
zLFJ3k=Db*+l_7_V54v>&Q)_Dv6HHn6(lS_2X$1WUdVTcHA$zl)Zx5Emz2I!WMLn}2
z<^T6Xc$F+>X6<yZEgdMATu4X=w{(-m(@j<^bzZ}&3GjwkTIY}}w$$$>iCJLrG75=L
zl_i#-W>9b)NGvjRY*A~ti*tAwG}4tq76_RfsIgXfLLHRQSnBk$zLEVJ%gF05tI8EF
zOI1T7kznCndMme`Pw5@iH+QB#!*bdw<abKk9+(}$V<X<R8+U0$31y2FU2FCjUDfr4
zGKxTRByuGoDXd4%?elshh}UctOuKI9I&M!DNOW=Vf~GiMGf78+%0Lr8+2$<jf#U~T
zaVn`O5tUA@E4{qo<UV%pT6YTZNQI1y;h@zMTGu_jy=@8#GpIi^?k#SVV?Buep?}_P
z4JCLXC^1j>yNpOHahlWnBjMDxjAYI${p{2Vf)%7dq1;BfSHQi(fDZ&90^46hf&9~w
zUJ;od@r)u_aXgC67C;1mihEnQ$N%nj?<`YsSXe|R6<cYGlr8P3m-E7upSo}sqYK%J
zw}oNxIa|ubyB@!=SJ_sLr(Idak>7lp+fD2KJz%>H54A$Bj~0*9Q!jKPa+)JKYPf5y
zL40Sk%@zPWWHhti*HZQX$ou5CD{mP^#IO;P$GLGF!rA54r9ESKRtb}WBqeGqM{2)G
z!$2`(v8bR5&c9kX!QD;_kvL{^Xq|PwpaUvX&0wA!`6n_26+U%wJgH=j9CQ6Y6tq1E
zAvnhE4Duk#H7;vzMnIv$J2{A|?ex5h7`A@(#h*2%?g)~{=NMeMT<`B!Sq1HHD1*fc
zPPzNd|1K7oo0;u(-2IL>RW{nSg|CTuc*34#*)^B{o=&+M0Wss5>xfvJpjWFRs`b0b
zstMT+aKi0g?RW7os^_<Yz~u1>n1#UT7~NX3ZKu(JQPF5NOlhgqRp+9&BAX^8LMp~0
zK~9zvOr2NdN8%0<lAlQM=WicP_ixrP@7nnLcXz{J@nS|yi>X)nUD#t@J}Q!y)0?VK
z7x5Jo#-S}6WLiuz>B1Oa^uFth&pkA^Sz<K>54#2L+nsAs%TwVMQ!cqvc6W={y9UQM
zGG^&5S8~{F;|TiBw;|-`Dd%EIB-k-#y#Lpy_B#WYdja7>?=`>uJ4^e{rF08#Qxlmr
z;lv@=JX2L}WTP$^kNLNrDr^|!4v7*>dL$(w>hRpAjl=+1x9<>uhD=N`v~U0I$Qj!b
z*EC(v4C?ij;Nx!Pm5JyjLFED#{LRzVEYE{0dAOZ4B6T5Qs04OueCNDZ3IE~Q9}SaC
z=&Gu;FkBbaV|-K9_{vTtq>V=mAN2<ocqsKl27EeM2SHZz!YEBIjEpHSRp<Iyi{LWh
zGr-sLn}<)HN5Zpc>#VkXCe|NypPfaLcWs>Kn|{uc!3Zq-LScA9B|?<tVSz|wTT@)j
z_1%X53yra*9I%d_7`&p~0aYowmRRTWd=jDmM_|vK?2>pqdZ^wO+F(v*R{i{^kVE3s
z5f{jxM8&#CsFsCi5|e}6Nv?Z;txlAxVkskRO-4Us>9>Rik*NnMM5rkB)$pVWtqzyp
z|CFU&A?57pHhr>fxR}Vr#mptBrkP^|s#4iF_wn`p$i1I_(Zlqr=FSolClHhDsj<d4
zGfg49TRI8ADhdTJ{{7~B@#k)Troi=M9ta2LL%ObDDkJ@qB;pbWdMjjK<KIXHG5NtJ
zuN2+5D;-j_vhR<8+ALsywUw@Y1PB+dT6L+w&(rqGjeaCVoG@;+?Fvf2+{a8Gbb@A>
z<*%<F6@%FJ$OQiPpGnHre8Im@Z}JLXQX2L7%kOuQWu6Pf-!Q7+CC}or6dBKagDW4E
z7stGrn8u#wvNGb2bllaqJvO8Q{QgL8W>sV<sES{sQHGP(12F;naT_vsbEdg%d6G=`
z)#f6nMtrp`HP%-Lb0=fJofvPLGcg7pC-g<+l}!Qxn{9P<=5Bqvx)e_3fNWqGM86c&
z{#p<Kf;I=+hL6Yf2e^e>@)gHzCBPqtdGQnic1)$UBFFB8sfx4HzSVn>ub)Kg-daX*
z%UPm9T?RlmZQ_{TkkDni+rw6=+pXW_PSXHYoPA5IN)qz-;Hu_sh~CvK$Rp6x^WmUG
ze#)-X(#PlFhaLLrq*)7FVdG*RGN*Xm-=C^mpLZRy=DO64JN~jQ&oYFHrrb67zq$1{
zcLwOI?THr;2JZpdF|+LobR!+o<u7sP(}vIhpE+n6E0vsZw5D{GvXgo07+=08o^Uw=
zfs7m;6cl0}1SNV3NWPqqb826A;I1)}djB64inDg&2sC2ZvL3o|mrCZ@?$&bAYV9L=
za{Nas>eGq~Jy@J~ylnHMhDluUK!?jn2xyj@b0R*kT2y^!yPga&$=(TyQ@NxByQ4Sl
z_~rsFgilr1+9tmReBM0gwB^Q^Ke*d~FSjJV3?P|(?qyVFRZ-(Wd#nY?@RE4`sDTgg
zbbYN46Y(;0fi+YiP6L-iK>6ryqyIV6bM$jjNInwlw(m^TnBq1$9ghT$-4PPI(I0+0
z^{J-q^4<AKcRP37u%Y=`;*gax42dRfE{Hf7rQOcFu6Z;5QfqJc(T3TfKoJ|)$rNWO
zh@tzA39h^WKTkdYWK@?TRAiZJi{lC-A^^3yYG={F<fHN{85|2fDU#s<N^on`$oe|Z
zu9W(AKxcb<JJwXmr#QE`v5XYS^PaT9d^H}+JKfu@F))|9`*D57vA^JBhbGR+@qBr$
zgc!n0Z(|w$@XPdtjO{C`fg2BcWb-f%sG44V(di&IV@2Ylg=A)<+xI~q8lGd9&qDVm
zYi$6-#c>`^_R0|4)D1l$YL4Oq)ctNVKM^DuFF7c33uKz&#Q1ne1vv%;{n@IhBCJeo
zk=#Fx7>0C<J&f)Enfr~2(J2>%hUDsh2V7qSNeqN^Z#v(76vLgbuQabScZtdhqO<Y%
z{Of1;*Dt5H4<1IF8?|1>A}$6#&vKE}@rr~r+UeS4f=1Ceio`fYzKVDh$w55w8?np)
zRvwA*c3JGX$*tQIrTq@Y5~l*!lAn&ZAHzX(2Aels-p}5=R2<1qLJx>1#is3%<jCRc
z--z-6&jj4#R0y<jcnQaQ0lbBPsjg>YCd<O%q0_7^R&sNl$9WS*yoIjjAu?lG1=4}n
z>zjYx4(<L?q%O<Us<=O$5@`5u?$NvY0@pZfM*E2@kV^Jbm!XU6^2^bo>grtCpZp;=
zV*hSm+)rI(@yi7K?n4V}v;lFz&Q%kKYZj3ouDDh^UUI`dWT!*N1V!c@vMUL|o%L%j
z9|OGF1%HZY#9XmG?-p?=FB7<$lTOLO0`hu#L^p;%Y*@_2*kVC3Mhp-Kml=x|ae&ju
zwOy+m&y7SA9K6WXcn#b9B+uOPj{0!2^6VB)VZE0FM%#9(+-TM8>~^<Y4!Qs<+o%7&
z3S)imc~iry3rofy4qDevNpg{k6#RGh8yi{ftekus598?q*zu|2sg2jtOJmQ}0uy4&
zE-}-z?iO!^=5No(0-RAJT(#*!F<5>d_sJW(H65(GeWbDaI9+$V%6S#rYc!oX;fizl
z`C4hjy)>;i9i5aZ68(mNj4{!trQ$XriNR0GI6Gr|%3^}yqe~Hz&>B@9SGn`E_wE_P
z2&+Jra3XxL!y)e?zNS9!o!*9%@5dT%9^Sb27?wu+aU#%&FgCid`EBDpMrO<sIkHkK
zbonYp_qT0xcXzj`t0=sp!8xk^>wl3khSZl{JDjB36llUt`~Z}5D1Z{^Y)J5#G)FnH
z>QFuUexog){#v*)Q@3rcj{V8xWHA&W<CH1AK)MMyFs+Fw+i5ZJ2UD~7)l{W3^MK}x
zzBS`ns@_fhnC%P=HPpS{!IbDiWFOm=jz>_9KnX`!*xw-zRA_mZ&4ZyeVq~j+{y;M0
zsnlU%w7Q*sq4DeY=-(I1(4LV0cE0aAt-ZshqXJK@<`4~a#|peSNk)(DJVsR%nLlk2
z4iT#(5LszB(2i;fWJL9watCr-N=$idjAv@di`|I+kmeQAfe%Km6@&gbU=@;DSwa>!
zuL^)7v%I49tKVGzX@aDQpU%emmxp7zzyJUU4c#XqGHfz%)5Yt<4eu3pTV7hp1V1i3
zHtP;z|J9c_^YCzG_afEjWe;|Y{oCgsUCmvgE)8e_>lf8i?pe-X+%zyQd(+i$3?F^|
z^LEof-s0e{=QQ`N#CN97{3XMQlvWE{*4F2TqGN?NwZk%E8YoN?JGianP5bLyzFCHk
zXqgqUnVy>^vX)1mL+)t->M;%h5IS=~6NnHB7G7NM_%y-f=~ZFe%QScKC4Lc`BxE9|
zsx>`lrG-qv`Ar~dK%<QiP~m%odDef?rWd+(|E5)y8wmB!8X5a8m+;j)zuR}9Jgq$(
zd;?R`Tw%0T^~Y`1OK&Q_6d>U)!&-dMdMP_eGVYvOVx5K<PZpA`L%-%@sWNM4LrEc@
z0AO#W0QG=N_r;F=$BSwIKmS$&FAOKXR0sLhU6BT!drnK=Y;=fRI<ej#o3912pw>cF
zqrVC&MpS|ZIpVxjWF*6fU+7yw5Rrs&V0!Y~3Kexa4Hb0~@O?OJoGp@&Yia?FBdu@X
zyI3cQu{^&gbqq$=cKfbZ9~E6X{wje#%Qy?XUb&D7>^=@&YJP8{H*I5-3Nzybao1~9
z9Qii=Y-;Dq`SBM21KFR9hmX(9DW~Cwmpb^ycmZSox!BkY9bcQ;b@Op$PFfk#jj8UP
z6@U|!@BpEaLXSbd#8ZHUcU&y!*UMAWXJ$I-j`LSz)>D6e^v!pYNx2<O<}e_(<B4a}
zHxfCve_8*SD0VnnzwaLM01{3I%Ic-t=unulKUcLi<D<6ona039Iff6qSRFA!+<Me`
zj`Vd}7{aQwb-BekJn^lMogs0#Va)yM6lX`ks_(uf3N{T+dP&vUx|0s*_Sa`8Cr+-e
z$7(RsJX*&!2r!h4(9KN!Xh>C5MVuB1<#g89RDpTg{>lX6B_3C^KnMY*MBJQI3Iuow
zPy$<^N0O1e3ggNxOCVh;=&h_$62-Ak&*#8f{WovMr<B+ciOjjWU3AOP|B4eZ6?ETI
z6DNAyZ^q;`L-xJ1E*a3QX}*q23BY0)g=b=vF1bSe2R?<Q-X-dO4g$W3UuR8JeJf6|
zf@SDnM_4&gy#GA*j*Pe$2QW+T@Y4V!zw2h`;RBR_fZ*5m%*BO3ibC8TM#<Sa>i0|e
zLUAgppw&?v=d?Ll)j|gV3P6*270ydFJe)4?M7>rKceId`hW3S>C0Ken9e>fMc2xdI
z8>SYgKr*y3ov#(tKNrx*OCNM2)NnU~zLm<J+uJi%RVPqhEOn5#_szOkuc_}K0g%6C
zCMwA!6?!wlNW=9m_Z3p6Z1^)V0^28C^FAyUsUN16UkI}m%=w1?cb}J+x3;!MEMxz=
z<e>S?x^i!M2VKGQ(cVt-2^JVbM@Pqk)pM|QcT(7pJ!J_dI5A~Wj#@D+O8BCRt~yGt
z0@1)f8}|El%(~d?p`#v)((v)|Sz2BeL2hB_BtKVX9dW&5ko2+j9Z8*nQ}<{ui?B%@
zcPtsjoCO*2w-!2x=J}(r59#~qnj2pIrgg*VT@gl8{+N}s0V&RYU;qdMmcRGW*1(+;
zpKzv)ot@QlcwA|c&IKEM%e%O>)^O0oz;_C9?UEX6Rr6=<Gn&}9yBo45l&=Wgl1fG{
zHj?>VzfudxEki|V6j%rVEe_u2gQXQ~zVFRCidhcz{U!U0>~m5A{xv&!fx((^J4<VM
z4t?}})B6t;dYCC{EPhKliJs;i=LYqp$L+#l;Cc7?WR9uAh6F3l^xfF1b+G5U|Hok4
zz`IGB#k59i!T6=6#uyPcF4%YwY;$F0WqoY|o?$~2({l-dG`^)5Nm7#Ey)Ow(8dUW-
ze}=RH_<Qj<0fan{ymv1Nycsu)PrwibVf0vzBKiJGP>8MFKArF8a~9=yjMxq=ugG(^
zqWXv&rtwJWh!qm;TO5$StY$|`J?CC_GgQ!8M+K1sh{>!ilVc#NON6t0Fk2|d`Tt@9
z<{025`48TaV2{jDg-$#^XvOY{zT{dXh=$e<m)!VVo@AJ&d}Xx-kq0YBLv9jGrGBl>
z1(|*PC4=IO8Krqj&C8XbVOr84KkA%Db#Wg+vpO2ft+cV3KQijfDi?~cKhjTYp<Rk!
zi$5Lu^}B|GN{HEL{C-RbaW62sJ{}_M!Ff2*(-_FPAV?~0jz<)i`Xx1%IRZdvFQ=Ui
zz#jy_xL`}S;nj1y24+|2v3KbHt%r=PKXr8jOF0TA%EOMwDrsYW$AiH)cQ_6Iu*lcj
z41RZ~`i`S7b*$!pPWF5JcFND$k0YQXF_U&}TDYTgCCXqjO=$@up$u(d6-@i(kViWk
zV$5lbi@w~wM)XE@@c#aO$jbfXYhzgDa$B9K`xG3Doxo=SSh%5;j&iS@4aU3`Gb#Cp
zaqnYWRJ>8B?ZMcm*wCAOiGRiSZk`Db(Y&IpehKpA^e2T6Iv?fa$hS@7DZFK#(|h7%
zj#&%lO|@vM=A1|*C)DUnaxWytB#f)x%Urre2wF!80Rfci*ME3pM<pBZZHSo;*_4zS
z@`(>3-sVu2eu1L3`OUZ8rI?=^EPETJ&z8^!oW>7fEA7?JKbKp22Uyo(2O9o3`FKSy
zfmx`vFfFAbEN@EK(jozP7U$hL9va^abYy%8mAS+dS7Qx(xLL6L6?_z_uTCD-x5e95
zxW#y-I`qg{oDV=qjLS*N&1nl|W3Uky0`DbEO05;l*rJ<3^6=`{I@X!xRp&kX8x8+P
z#ICRR<`2t(D%?q>Ts3ckgQfR_0s`*_$I>P`=7L<O$BY$}l#QYR@clm?Jz{I7ytb3=
zTo~_-0A)NDhJuX1-x<^58U_GM-C|%j+^?a-q$EHDit$}6Uv|!XzhBCtj|-^;f3+5D
z-@$Gy(Cs_qQ-7K0J@}Nq_1_j-1BHhRDU;KMn5)Tay+z9w92#kahs!XLW5#Lt9@E3g
zghf@lac~Y4XpY03o1Mg|Nn5ST-_3XbVPRS&n@~+z*4~s`zUQaWs6L5$c!=<-lL7&B
zY)QBAW9L_CY@Y7)jbr62aRt)>796`6BJ~+<G?Ha!g@ccvg)ki?=u>KeX8~nDxc<Bv
z-NWk9=OMK`XF@}uLjZWikn=Y32bDxqhhkH>6ir>9v7v_xNw~MRvpOhSVFF$;+Rw(E
zcDRHa*dP0jG4J8nPV;rZNp{U}&rgHbA;>gXISB50F70>OlsM+4PZ9F!oGxA~@YeSq
z`hF9A<Z1rju=l?^THFLc{QZ3zr<R$|3j6ji1fo+5gO<Zpk=S~hYfyXKL5qa35A9M;
z(p2xa;+bRr4p~G<7aN41)(rlElqj9y+Pz-ZFTy!^86kn?XUCU2kTS{CYNsAcR*_zn
zVNdrXmc_&)e6HU@IqG{C?E#Gt)Q{Kb^fH3&jdmR)2S%)z_qk@Y&Jh4fR3a(<&?)3I
zx8S%<1YW#|{B2EO1Y5t*F09#lM;@oE4vlDS6_Lj9u$-KpCXS%$C}&r+R2@(#l&h;2
z;_^&x(*$X!vT>(Zp$Xsb@+^8n?CfByb*QzF<$LIbXl9l8DIEOSXdmM%Q-KzqES}X~
zYuo5ZPKCb7BLIms{H)|4E3+rq|2sGX7NPq+RtN#xA>5Wo&tnOF%U*?s10rOz73EcB
zRmKipFjG*9f?zqbp&U_Qpky}TZH`GC%*E9;8pJW*+Eh!Zu@;rTevjqH?f2Yi5~>lV
za8*#`4YTOLX{V^-pWYO7q0a|nmC8bat8LJ?s<=w>n$;T$8+XOLT{kbeQ{Y*V$`r13
zM`1e@u`zl-(i1!<x$WTDuCfRPykV@N2M9(o9goVJMh1xS#B4C?V?p35nTFXUxBbYH
zFYssrnrwFJX*$86Lh8kVzUF<u=YUZ{Oz{;dAfef4O_ZgH5HyT7BV=m*oNH|>`!;l;
z+=;~SEmew19JFRr?VE>FyUTA6V7zIT6*7S>EC$wN4><1CV1>%9mEPMp<(50#tN}D|
zokJ&kRVA%JQQz(@2mx};c_f5Z$w&qYp_L2zj}Pqkl(~GhASIl_3kKgKARj99&q+ws
z_<43%WnL^4X9e9n6eiD1sBV#f4@4!3P@Rmz?M7*jYkw^<N$Ah-o&B9J`17$fuC`3A
z{*%ODdEN2K!+TkV%s;6$?}G1p{uB)?n+E(TdzssS9{6_A`xk9)UI+x}{U<0c?*1c0
z*1!L0y=WAmNrkj%ebpHK`{r-M4Ep*RRK&pCpMA&gX5AdEJUGGsIjHgtjhR-k-IX`;
zwTu^Ut|qS;Ff6@U*GB(BE$|f=MM>|+sGqd|scjM?xh0G}<}!GJpUmQ@NI~qB8Ltc~
zdv8F5HNuT&ddkesgK@#|W59^Q=K4!ktAnyCFPFxio_cjk4RQ*_BEkf-|GLbK9jo-B
zK%)t~uohZ$>PQdpV*pG#k<UI(05L)!P*xh&94@fnNY2Gv&RV>Lhkfivp+e+OZI3dA
z5}0kJCIF!E=O`?Fxa4iIt;l84VOOacFBmq0STa1PeO_{<nXSm&;=CIVK_S!Yf5ykZ
zbs_Ee?`_gU;o;b^eQA3#^a}+Dq0JY%T395(wT@1PSrb-O=<^utpS)ma@_8h`NcQL`
zkNf5H>*Xu#2-LVY>oozN;nkQwJ3W(j``vdcBsuLrZ8|XwNrUOXT5PJ<Dp-*Tww3gZ
zs{iNbF%7o^vPXav=;R&ia@0An1E0n`U!ExJsk{*t!lKS{8W#06a?N>70^qR5mrexT
zL8p5yDI%g&Ci*gLa4Qj?-j00*VnL-JvRPQ--1c_5^6ItT(MyNNAd%NsAHS#-?|!S=
zeT{caKwq6{OGhUdHR|zV{lk(iSacL4*FK5Yzd+Eov_}#n8rk-qzq=us!FGA6uYHy;
zZz!k{9l-f~ZG~}!@FJkm-nu;k6be3KM~jnb;Gra#a%pBSSAqy$U2_yeC?*SS+?XQ>
z!xvsk++4`Xh4xU&{WH4d+GO_qs7L2->ZTR-9(c6tc_T@HA`*dG8xcll&1a|9vNjD%
z9(TLr8=vR<P8&NWdgClXv?^hw<QdxlzoW~~|L#6hThx}!4O`OTWYeTQ$~E`@gZ?gg
ze(x*Uv2##npfJ8oBy~#jZyCElcD<4IJNjDGDCcD3{x5cZZB267>~8STtN5t^ibyA+
z_1|$Wr&E36+3mZAzZ0Kjl7*OLy8f&$2Y(b@Uy6BvIwzaa_#M-R6hp6m$5g6jKHBz8
zRPY|ImND?g_tF$7+@amktk4^(x<W5iWx1}sYIR>gI<)T3zIb;RGXGXA0rqBN(L^~k
zPNYgjbMge^7%&I1bwTyk_p&_&yKS16aS3|H%)u)j6dEc$HjJO;w2stfssd$IV{2h5
z#M714xwfT^oUy~@$g^$j=!{#j*Hym;q~~4;i!N^&D3^Q_+o4TV>g?={nnd4vRsZ@F
zGifV?%)|QmBseDn6fVu4JS~@t)f?f&^IB>x@}7n}QZ*Y9&>|Ou+S5VRO+ota3SDRW
z-c<F?k*RFSeeE+c$=nUq_|<E9fbtXX?S^%8SJqh~=CCC+_kTk{VQ`|j#+@$wa+3bR
zm_d34Lab079Nf{<R<7Arh0a@fc)$|KO~tn7J~>b_w%lr0H<7@m_psm^B82r59A-;w
zW6)}wb7s81{V^om^K7<AA<Hy?`|<h7*`me(0EuMo|4Dq{S39FWra=4X(ft!v7J#uN
zHea1f`xddcWtwS5nRMqPOUuMnRZ(4y8c8}lH$s-iaFs!XM$)6u#0k;#u`@z7o`%P-
z>Eu;oW56l1<>{VD!@243vIK0zf!u0-2QqJEHUd$IsbKVTYx`)<KOaP8cL%}+DWrws
z*8(^QNt?bDQ#7KuNFpb{hBqby3Xj7oQxCs6{Pz?f0A&X1nVW_t%Q-RGGpTP6B>#)(
zJ3oH&MR^bR5kLan>lX9P`B8PsxL&v%%i2GxvDK3dmhW8`iaN<T632D?p(I9XI}uWc
zev64;VvV?~I5-KIS>+M^V}W*~G%xDEWvkoV`vq^?cYh4TzQfNM<KQCBT+*igyEd@9
z`8BeTKWaDT9)6C2uCBoy-n$uDOcE^)3}wAhvinhl+J1rPq1_Qac$<hhAHY(WF;RQ*
zvj`t9SP?L@Z}Ft8=%`jw33dLg%ylO*CYM7Ip%JFrAXE8rPE-*^V%&`3k231_liFBc
zH#IjO=^4Oio-Vq+VXHZB^9a-|kdbD5M+wwcN1@T?U^ZvGn2oaZh{#%_1+_U+I+(Fw
z02B+qdWo)ZMJ86?ci&fYvF>N<y`g!M@{wkcDiY<8)q2WFMWXwvZ2SzjH>jQ2E>eK#
z@%P)?TaWgWS;)Db>b(2%{9mV)=b@zJ>i}G47S)1BfbOwsRxN~cCFR~1<*fwp=={}7
zlSk`a&UCUgT0DDfy(Cpr+P5t3nP$KHYEqj$TocfiwieJ{cjO^OUuMsH{=}PI`vu`A
z5UZ392gXI#OV=o_6_O8W(g&MV_YHs;c3R}FwG@p3QoZqs9H-%u@qVocftN8*w)|;@
zP8Br;jF5|$n~0@ibnF8Zild-f+7dO&k}`#n8>GtvCX~60OvCvzFXk(>5|tK<>J*NT
zkI_9``0kChpZ_c4pWeT-`920=%bbKadk)zPGrk+?3=D)K>asPs-wFWk`ZYMV6R%qP
zbIN}?&qsV^gnMKw27TAc-@w`u8`f+k^6fX%$NcpD?1&VZqkOniSC<S~ap3HyC#rzf
zB_&X7xtQl3CdDFocX_U*<nU7D6?Yj0E;<Yp>g(d#Zb0>LR8z0Z9)MD}r^89t)D|tL
zxGM8ouzQpDHKVS^&^V!;$N}!Ucj1Hf1zT+tME?n*F*f6s*Gv|S05ETngvwHl?tfno
zu6soVSaWQ6a>y_+tfJGa{UIkXQ;p0Zf6+z`kgl2W7^sLXK^0;qtb%|*;c@^xG%^T;
zgE$;DUAQWhLqM&;&QFK|v?BHq89|42Lel{T5_Yu(v?d3VP6^QC*?{?*7B&_<F^X_h
zLA9=coZ>AY&5VAZZIZ^6irGCAV?Axe3&(3JY5G}`Zh{ABzZ`e$=``Z?)m(r38q)7r
z;O7uGawgf>V>sQjJ)3g==|~cd!B5f=q;095MaUa_rD}&-2v+lk$;lq|^MT1NL9D3&
ziC7$hF=o{l%i1b67zuVY)T8T|7<;4Sz1z$8vJG+D7)j5)$-;q19HjX(>=(3FPV;V<
zct{_KL}uu|r5vA~EYf_eYmO|P>kfx2h{+&b=3q*mpEc&WY!v%g%+Vd9Q7cYLh_t40
zY=YPlifUo-Tx&BiF<~KZ$>YPh%<s9H6v3sJ|1fN8qea#81lZEDn5_BwlEYpi>sGXm
zcHZH^)7d#O#<n@9jcIOCn2kT~$7l+x@h@wqWlA5A>1^le(l;xak4ve2>Aykn+a|Z3
z+Pv})ngfoA1qU&58}#?DN(xGzJnJAAQ+wPP^(Bs}CYRM$bhRb$wO?l`36d<Re;lMC
z4^Aicgo(Z-O+1ga2;svN#crZY@Ul<OtV;eXItb+KH_FCRA(v=Tb&X4gNF8lNbc^Bm
zC(V9x*%m8`d}zy3;ETK|g|)qOQ86(wQPB=RMdqQOy<>M-6KH%E2#zC{-Kf#`U-VX(
zsXdK(f%rF~007i~f4two=4C%YA1IgVVwG)(&squ8S$epd>m7yVon9^-osU3Ju^8Ae
z7eNxkj|E?1eN`|^bw@7js7#LS@nd>`)hE^=JMS{YK^>*-{x>hdu}N4|y6&UN5ehvL
zXF}MU%=0-0^!@3-{eKQ?hRg645NiY$c-UeF#bLIHYPN_8g>me@&WIwsosWPPG_*W}
zXL(3MSlFmgx6{0|oxZS-Kb{3wuG{hAIlWs40+J6qs!Cq160scsf|v!OvcIMRWwBEj
zK5xQB{m&)1iJxPzk%qARat>=QcZw!uVyh~_(o--Rfde|tkofy0GB%Kpf(+kv|Equl
zrFPDu=OC9vkDRK%U=C)*p^hn0y%jU{k+kl-0LT(gAX`}b_i4d@5!uE7tQY2uNnb)}
zxYhI64=}a}-W&vA%s3*x+$6^=LM<&K_s7$e^b7lM{)Q}H_7Kf0>W|>hujjtYac;$k
z^E_KZ@42xMj6nwA@03bJTbAJY^p#F**9e8kYLTnz#t;3Mv<0N@jrCQ5OF&ekA!u%O
zb+r&BR4_9}J4FmZr5lo4{fy-7?Wo88s_9a>x*m};@T@jTu50;J)rT?e##B}#Ok-U1
z!8aSi-i*0~;A7`QH5QRal8A2-eAw#e?ryZSw6o&Z?i(B<vWil_C(KmwigUQs#M1`=
zZHmaoHU|@HA}Hj9z=CD9U@ER!*jn5s9})>7KF@(i_QFjBF-<WEBPfRbQ8S8TB~q^e
zztI;lv&nViO}xYk98sXvUN{Y^yE%$55*|OEcjZ4a9lf(b*vLzasyi>Am^P~EMJF~d
z(W*so;~y$AcS&4g_L_~lYrAih+uF*viS{Sc<<+k|8f&SGoW(g64(h;da!nkf*FIRM
z_VsZG*b>zZBfzVLMK({<$~O%MI7TwWm9ijhbSQgxg7m0T^^g=xMhEekPsqXk`*BV5
z$PCXUk|NA=9!5aUWUzCZwE`iUdE9I(DT#kFpNH+zpF@ESv0?lr7T$bA3JhhhLc8Wc
z*4kpBQRzPv<*=;`s`EBMi@Vq3Gcf>p5@z0D?RuoCkw98Bz-XBruMju!v55R<W-L8~
z6-Rr2HHBH}!=BR{eY}|Tn<u4>a6lm;=Y66Ins;hbAPg2U{J+8O9^L(4F|*|Fo)2w5
z>v<-?5_b8`8*<do4F!i}hZbY#X@sobzPhC&wR?#QwP;oQ3b>}fJi&eAe{BQEtS=>Y
zs1%djSHyv0?V$(3H|q+BlOSZByNgyqKrhVz7xP!kp5)k25dF(kr{peg@0ei-DuslQ
zNFa_YrHJ$LW~pE?3@s&c1c+K%#m*2{+5WR~Y}HDO$*-{e|JA!#0HFHjZ<F=<CJp`F
z>^qrdxXr0W)d%kc?m<Ij#?mZLcI9ym6H_TH@J4R-rgTx(Zwn*7--3Rh^yqCkO4R|o
zxN?6#vkwSWi682sW3<}(CbImhCvHRczM)@kwj!pc8XHAKn}&!te-L&q@szPuaxCA*
zwux&Cs#9m*@MI{c)F)E|O3j{VJO=B3R!_lTVYk0ZzdYNmpC@>;_{&~@3R^8@b%rtc
z&!j>b7ExRoY9mun;%%&2zLm{vZ|Xk&a(-K7U4PkVV$ylu@4RNt-WNSXIW!az!i-a^
zQQ0Z`(EkF29s~I~60!}o)gPDF6U`YZ0Q#|7iu@^lj7lXBkxHmV`bgqvPKnPyS^dv1
zC5V)TLz(3X^x|@8@K(!V_Tm#BMlam{#dfcoqUxH5jo$Wi@xu2g2nQ=Y8ugMo-NQIa
zsy}4v#lbwNP7T>7*?uH&tDV<<hlsOVD#<GBB>b92BNbtM6NX4OU5Uq~#i&yc_{SyY
zBHte3=OeRlrsDu98px-KY+^?;1kNISqW@(inJh#RGzbEves|$v2QcgVDHCF>5}ZYN
zwAGV99!l60JDOlRbw0AI<MEX3N$ZS6<((|shDIk~(dcSS{X`i@O7;*ERN$2hTc_U&
zZ9?ZV7su^53cE`0;F)fk;{hxHZGZ?8i`qF|1A)~S?}eoh)wMTEaB~XmAxb%&7-R%F
zPiS3|2v32c5*mkV1VxT}=Lf(6Z46&=Lce@?(_lpa0?U6A_z4%6_BSM#M;9%aO6ERJ
zaqm`KT7AUm*9+lHoz0v-42aX^zs_FqLz7#wfi0eMD67MjLFH!~kI5?+2tuJ0Lw$q*
zZNi`5ji(P5)n+<$aulR#uIi9TB9-~JRX1e?KL#>kw0H{;?*DtINFmw_19|_jRHT3q
zE||&pd{tuV=rN7Gd<O;w@}vp)i7ij)%KW$WcK%SGh?BS+D0^md%srYGScbQY#-UU!
z+J3~Mx+fY@%zH_`7;N`yPFiUYiwp|u*iM|PVjX?PMhqChQ@k<CEE6gB>Zz@^(L0t1
zoMOBQZun&2O$EN%@-`(~v>s<+n3Zxas2QX?iYzQLsAu}wTl6dDh7|0fxtb*vq5Wyx
z7+cGxWMdQfa$*o(Uhm(L*+!#)p6q(fhn0M~nwy1a-gzg`Br07D74xdfW-n87>S4b0
zF*j;mV@$)X26yODYrcEO6xVg}<u+8i70uyQD$Jz#y<Yl$mQ<7-2zi8Pjz;FB1s+0M
zge)M#S{^Tg{%L0fTch*wuHdjCl}$rqe=T_`-hAsXFK2p`V5!U~ZFnpCm!B}L|9)K~
z0uP5<iDJ91s@kJN^!bF~P`1@1Y?s1Q70&*WM*L|wyH`}YJR5-C%6@hDc^z!j3d~9a
z9)pQAAy^a^#=xHZA+zo1SxxR+e6Qaxav$lSRdEyEA`Tv_Ekw-n9%H8l9cd_@8r~A2
ziD*vj;A>2!Q8{BO7@NO*788D7^9-=ga6Skilnq3s0ZM%OPs!#P5)dkdHRdwO{^x9b
zD#-w_VQ9>(g0RI{DAbh~{&;8Xzy84eU}R=J@2pn$%{1i_?YVSEiubg{2w=&?5i_1>
z3#ObNOb5t@%0X0;pA)~a@9PD~W7wHaJPz-8M88~|M|IUFqC5GH|5yS@sLMzeKeO-q
z#zdQaYjnWz#KRM|m3oi)w7-Ed&N!V9tmb3(l&2UQO9djv*1s7^y3jfTK=@$%2%;Tu
zlrI(4p_1g#Q-JJ5I)%wBm8NM;ZJ6Ff?X+STZKBp<^RuQRrdY00@q(%GVr<Pz)YhEi
zV0GTHv;J^Ax}O?0tPo9lxYrR*&>W|&z`s|&-_TEBEM1e}XM;zHTsge-FbMep(AAu#
z<}N$*%2hysQCx#AJcUxh`RWHoMviL~iLEh}vB3U%X*C^+(<TDIJ|<4D`XnKV*nunF
zZmDLXyQ}#X!GY-)k7g{MYEq{{@aX>h^S=-cXpAh{=(g=iT)!jEfN_={>BIrXG4raj
zx`kpwe1b6T>M)|Sj{qP~*U?zSmD{%HqKAz<p^__~I3M!W|5A}L;8a6Qdtc&Yquv~I
zF!Sgj5TWe3%(?nIPR||wYN@=o_DM=qlowTMij%O_9J6`HQtAfc+H8tHNK~AT<po+1
z!&cF<*1GC|t=hVAa)gH_b1L*P=lRJTtBItYk78lH!YkL<A#364@$tZ^AV;-K6~NAk
zy|8hCW=-)V_bzsf<Q|6TIVq4hYwCC`Ei`)k1|c4OJ&=ZvSbOQqMvT(0TGY0UHXSWQ
zmib=d;B82h@-;D(|BHLd>ZLIdh7U?fxI*vLn(lpDn82h$%a-X{Ncu5P=vbEkpOk$s
z0IETUO}PLI^$Hklp$Q69h%VHNZY2Wj0D~n3ww~hPu;u>+Gdpa1%1(Je;;BZ{F3Jio
zLUw|<ZFWt+zt+iMHP_!EK2qUhm>V~CO)N5F%fEdoe>3sa`Kgq(ds*}d@`cKPA^PgV
z-ut#{pqVgsFAcwRg6hy`9kR%(ojvoi!<Hh-B@OzC>CHbpffDMM@-&9{)t&3M$#|)9
zZEbDW*S?1O;okzqB)pG=)mlcUnfDVSrj0}KXz2aJoDzUd-lp+vgE)=2x+I}01oE~}
zW-4Y-=z^VyFuN$1g|8kBtg_2#3O!Q2_T^I(*97*!OH6zM(6;MnZzlp2o!?0m%+$3;
zlQM+&3_W6H4(HhsKd<v`{xfOxvAw;oyOQSWs3-0}ve|!rQhXC0Xf(R;2Z<ob(HYjA
zq>Nqh4G(*?>o?SuOoJYxL9f4~61IPRa>>JL9`8nmjR6ZL6e!+X4)nA!8c+W{*2{oB
zsIk&{8_VV8e(M+>UJJFwq=ReQ=T}MT@RFa?z3o!oM+?KlQK|?oGG!qq+Pf^9s+E%>
zUCZ{@<=?omXmt<4oP$C92<dE1>Q9_7zO49adOB1q$Fr7GYbi6{tQh;c!=vh3m450C
ziOy_(d6Mlhyt1bg^zTN{0Bvz}NONPfAl}cV_e?ViFPKiyr`!?W;_JOizosAUe({2k
zh^92NfjN3b#qIJ%2ndPBXi?p)2*mHmhVdqz8V~)9o}Y@?jnR^kU1ICBiMbz(@nlw!
zbDcNr!^i2AD%DM4HF-Yb0&<**O5Xa1q#6&Z^Z2KnOisC6B*Vt&eUiAHbaq_gRFD9u
zu-Qfb6;j7<?Rd=BI}BP@$gY5QkO15l+~?&b<Zf$pF7_oDgs7*mSSGLmQXxPHk@g+_
zo7<eo&fkB(1fIYCF(}K^yzK3u!%2{@%zlvTgj8%VG$MgTs!Y=q_urK-g+xd!pZ~os
z$z>>hc-&l0*8Uo;g(uu`4)aC&h((dmKs}~YV1>X>SM@(P49pa%@}K>!<WiEI&M0#)
z`?_OlStq76e<c7oke;+6{rhF7XpdXERwhV1u-=YZiuc5$$n;xx;H5OClWCuRuP_~9
z13Z#F#5{IJjjE6a>_&t#FuzrqWq&OGbWW~FVwxtO&Q=Tk?#-9Gz`c5M>8q<-WAQh2
z8ZZ|Z7gA7SRK&*Jp_fG7)PkLZ%TsWMnP9#qHhyUm_T^4<3qM6JxBLlAy+W^!8u``K
zG;i-YXIktja?*=gCn4~_YIHK1jH&)O0Q*`=F_NW9N>m=G6e^#HbH-p-J^{@bUpPnE
zKYoSF4M()WQwDMhDg&z~P1d+#L+o;_qY{)cI}E149*@NFORy?GPQV}pixvFc<~)Oe
zH@CMt1X_w`?CWb*ms5SB&it^%!}$g)yM^)Rzu3i)IcsLcM`CG^6sRxl;iQT1vSP0E
zlZJ{n7-v0w5@H~c4q68~T+BPX?f%plvGA=4t@y_Kl^57d4zX&(dM_z5_&fMD{-H(y
z(f^wX!b5|NwZ5<Mk>z-h%+cymZ1`v4dt|-C2PRLhKo0h+M16L*6}dj85#o+vJd^qe
z9A7Gv;2hJl@MX5PK`074%)$YlQq7*L#@iTtkf5j#UFxBAS7T4x$Ir@uN&9^Ce9pHZ
z(Sj*E3CR;F5>JvMGrnsdA86UEC@czOmC)#ryX(=IJ+oFm4O5wRwts3T$WchNQPD)P
zABu12ot{I(=BPqKB&#PbVKi3@2p?>lf?W@_tx_h7xEIS3UDM9dN(Y|jd&36PQ#owG
z2#;}?L;iCj+Ab~w*UkUd&_U>*7m;R{vMdeC-#W*ly!MyGOyrQBLZnZ~xtm;K=*fA)
z#xGBq#vlO7bS37cEyYic_1?2|jtMessF^zrM#JsDyZh?GGiSS)bqg`}<g?|#Gi`y|
zSUVIgAeazm_I|7EMW>6w?9i9m^9WL8Y*E#~h56qSHYrAGWl|==;Y;+0+;$BGD?m*X
z%T6cu*u`8;3)r#L=*fQPm2Lj>5Y&b~`Kp7OP6{8`>Q9bV%RCca_^f6-$?H9D=6jl!
zc=#qsWHG$-o$O=(%Z{PI3jG%c=487+b<PnnT%NB2Xu|Cukvvg*^oj@o@_ueEd>p^t
zU>qq-a`OW^n+0|l_xcZgw1tjVTk<M{vd)&}rwGRx$Rc+aK>&@<7*7HKN<)>Qk<&MF
zQw0GXPV!~8iUGm))T0U?&oGxc#at@6^JB?v&_mCg0IiN6+$ci`Ao_b)3`lP6xx<yK
z?Na6t1AP4}gf*pcX*j(W#_em6Z}laYAy`d;WK~si-@0=9ADZhG>8!WS7aC{uAIN;k
z1bv)}$?})^XJx+E-PckvZ3J>S^76u^G1yDAqA*pr?~+xNdJc_|p~qPb6+SqU5HDeF
zcG?eXmzXH#ok&_zw{EfX%hDrfX0z1HNl6f*k`=EI6&}|9{mL*w#QLO?w|xHB>cdw_
zI<6Q2LRKsLHFSQykX_r1*Gp~`h<V76KePFMCFk_iM~1BNG!BYQb+)<W?plFYl^mXH
ztXYE{HKwdcXi&-QBNVH%9UQ(47GW(3y%O6AawSGDwLPBvByaf@Q>kF*{9$ePLZ3Ko
z6lCX2Dn~G96i*0?;QO{sub=|6-;J<s=jRY?<^TJu03EQ9@v5yUBe*kLC8<h_>3!MS
zPvF@TTyiiC&Mc`kK&4u^)(fDf1md;p!^ov;DiD6<{*neY&QmX+cR(WNuMTg$f}Nc;
zyd~;dX`iy_=!hT4m0$!phgu`x6%;}hze<e@>J?vAN=0U4B-=krvL-`vhl*|cCipse
zTc+NUp=JKV?&0hL?=a22X5y_H6V9EODXQjR7gNGI;>FU9IsUziSy4DUP?IHebR|MG
z0(4~4ZB8L^z#<z12G=y<o!A;!cT|U(jTjm>wb?@7MlhVk=cg%mbE=M=5+{J+b_yzh
zMrW%vn4Y&WGYd~?NXo%L$;A6QC!1$?qJjuCYO-&oPBXc6gBpOxR)NtYhtU9@YL8DF
zAM9PP>O`cyqQBZPJt$sEIw*Q7SvE5a^3JX7GMw^EhO*7lunAH3e6SXDsCo+^1ZOOM
z;(5&el|2m?3@}z0PWG+8#J8JfiI`pRUXOajDBk{JZH8M<&L|ll*mrAnJUaJ`uagbk
zqVYeIK$~+`{E!nGz1+EL#kk7hX|^Ep_^FYCt$v2%vE%XlRTD2(QoKzs4uv%FnriWe
z@_nur+U9%yj7TS0qP$8rdVxeN`hTnaHl&YKx=kMjzx2uTiqbOj)aEY&ooz{eq<FvG
z+B`0wnXHUY!%(Mk79m_rbM~}y5$IC>$IGo9LA+2Xw0RJur^#*eT=n3;VI0**5FRq&
zD;ktKH%oXMD_&jbIe7U!6?=^Au(+UUR#7bxnQD$mLsek&FvW%mi{q%tvrs(HY6q#m
zs~u-cVmY5K75>4VcV4pSJNl}ui#9W=qA~xSZptz5X7#aH>saZ3Z=2ogE75VSht?Of
zd-hU16JPF%wHykmlpP)oKKu=9b<X>sq6A_q?Qy0g0W^7dA--$G%@KO)_{lFs_yVSh
zs-FHnVT!Vnr?z~i{JANk?u1@JUPIZP$?50^^iFZf(Bn}L4giuHP=Wbm0|XKDlMV`4
z6D6jSP#C?c!j<2_h;!N{-k_I2<rDy3avZtus>kZn$vKLlqZB>Xw#<aw0xIK^IJ`zB
z>$6IB_twHtHcL63_Zd5owT`M|d>nCCRD!cQ&(+V?r#I%_|Hd5aJ8c_Zw_Ux~t^Gf$
z-a0I*FKQbeLKqNXC`AG321%vUAq5Eu0STqMq~nkh(kYSx1JWgoG$_&`-9wAy&_nYb
zf6w!N@Adw}1<W=3?6db?vDdoSy8C{JU?);h|H(`@iw<rdpfNP@*|yiGocPMAh~RC0
zcW8buVbd*yc392ZrwpL|7J4sXWZWjr=Pze(LcC!2bWN3@?xu7mm`m@3{V1^6Ds-GJ
zbePlC8sbMJzoRzOJE)5Ks|SkgH5(5m>t#3#{^qYyy*Et6cZ4DSJCX%69mnQ7;V#dg
zf9a|+{F@>g^&^RYbi`T9QZ|J1WAT~0NfZ4fBPmwM)0f=zRE&Tgm?%gNJIbUYZE;(-
zmiSyI`Q6{rsot8L4NnL+Fkv~H90UAu%PHl=gqBvmE+y<4AMb}5f>g3~Y8szk-6w_Z
z;!vAe&(#w@6~lIbSf>Q`l-+rgPGU^qPN~_M85R_g8;5$B9+sf=p*6%mKe57)l5Kfj
z@3^W#CPe0mz(e5D<eI+>6h>daSy+tJmaBvYEa3O5$(rVOC+qy!OKyXhh*@jhqnNL&
z;d!)HwCty9y>(*nTMAip76v6sXC}wEaqT5>UqD`S$*=eGUuHRGOVrHFb)O#kkeD^e
z_5C?Z*wc^6S}I$>7wJ$n{#w#?fsWrx90+?3wgEgM4?wf6YkwdFmg-85?=;VE|KZEG
zsdt-rkRieosGF`V&2{$zooJB+H>t7wf6{Zq^(=JjLE{cXYEi`~ElGg;@5d+XB@e?R
zCP!rzE2sT+WrZl<<D;IES#LL5(vW?fv}Aa+>L!T9tmkkZ90~!*qbE3@D<D}BvmWQ|
zy-B#hawC`)@^af^%a`(w7Zmg`YWl)7YO0tc`FcCatuT|1!TRaDN*`C#uCaxC%mNW1
zA4d5k<#2IBA|BHJd4K<C6Lf%CnF=YMD=}<*`cV0aw~)wj{lDXpWjBr#m!^96oGyf9
zX8bO#|AQ-okD+m&8MXIfHpMH6TNJ<5Ft`CPMSY!O7-VWDOX66!O4i;#w-ax8WvOe)
zT&qG}D#FK*;2NM6tu|JW9RC}(Zj1@8!X>Dk**V@E^KDY>4tqtrNfnZMZw1Spsl6gA
zm6y;Wc#BpeMWttJCX(n=YxN_DOl3);TWUbnIH??(&Es7tx4)J59{RRJS$*8L;*%>d
zF#j>|kn>^FKLM7u%RiWXVSMuW@fT72)J@8RA?$+<g}v!GtcZv*9mA|<!qMCn_=1<I
zDh5D<-PkfRsCAy!=%&1rU2iXY%6VCELkrRN;@l8=S<JLR{yEbjT1Q2+7D_>#HdkkM
zQlrd)^-jmV{Ab(hEV_737ZP+K_Y5%8emVN+`=7`~@6Bqau{wf{`Dbb^ZzkOZb9c91
z-RTrH)X{nRIL*RFdx9oD+L#iHs3zX|^i<g-xc%W}p8l_8L_XCC%NNPC58{76=x6z<
z`dejmrnBU**heX8O8Rgug@4NWXKO&k`&iw>h1Tkyu?7FJSo1&X80gbI6+x!to6Piq
zpL92r(_H`iB}H<`;eWQVe8<J{fApKZMVy<V=*7QpbNOb@^^H{iH}2}ZZfwB+JYeI0
zRs8=uq$Dxzb-2{=mkmj@_wATfb&XSfdG;l;2>kJD8ZcY^yLfwpsVpxy`hI1E_L)l2
zGd?$&hCx0THms)_K6~~oU>_gY@4%;PK1*u&^V!b$q(sl?>LMgd)a}ptFmPio;N2yd
zeK0c|p!=W|(}MON_C?`97!Ui~uKGbhv-fCHHpmcpHT<Aq`0WuG!hX>5SAO;2>;tvq
z*<o56a24O}gFCmd;D7MZq`9u{WHhbyxT0zAg7?h=5H2Z|ed&!k{gdG&^|iFMsY(3G
zzn=#wv4sK~sCAqSH)Y=LO!!|)c-AfZH+Wq-`RswuJ|YenKV6SAy_E*`;#}asuq3AW
z7{Lod1a2$P7f((;xEn3{9SwphPN#sQ3;@g2%4$O~U`_GgoBj<C_}hB9=NoFY5OBGB
zyH9evUmfQUG9(JO9>l~|{lOuT*k6^tS^f839e0hLug(tQfU8v}ub<SArBp-5hmO}d
z*E2biTVDV)*yiW1$=MG@sl$Yrm>A*a<$K+6bil|*<Kv5|wwo6w7i5s;+iL)Q`?2lp
zBT2x1h&0Ac8go(!_)l-81#G4jj&1Jk?Fr_b51?;v&~&Le$aC=JvVHd&Hj;9#*8Xi%
zp>WG)vzg=;3UykKTmS*%0LBx*OjHJ(AOp6Mj}I8GNT!j;ZqjJfzm1U$0H?d!?bLey
zC#`vn#pLENE$0|0VB1VYww{##)Fr_b>@^<^g>2_|ijsJ`mu>{0E)SungWe!dLpb2I
zpC2{*6S?CCMAKg5OYH~k?(VL3$Eg>ME#6));s=6%{u|ZFXDkDCn2PkfIm)?>Sv-pY
zv5%MFOQ>p6RQm7QkvM{df#<2nW{a~kaAkKcfmZZoMcd8AgXX0>IcJ@0kC$_iivgF4
z(ibD(=q4_|D1l=|K#2P`MEWcQ1THN4Y?p(lM|9YGC@fqG9WiABT#KU)VyPX!FLS7S
z_xVXJ&d&Nwlp4G5)=h08>CSr;qw>xobIu}>ZP&YcZ(M#=O;ld>RVG`UCrO_t<p4Lw
zMZkt<fYs{&)E_X~!`Xkhzx*?)cWtY*QVRV#t`RxadRhmZrvakrs$iGy6Yzoj!z&p!
zi=)HW&Az_iQ>3mp)j|CI`Eil-^}lugx9cbrD(>cQ+|9W9_4qG|;-g7vwCm%&EnV<C
zn03;fjHCJCOA-5%@(W;oy}AmVh%|_nD@g;oX+4{-)_HmR54!_)x>4stp7-TPgw*Eh
zE%aN_{^-jfy0f6Vy1G)391g@GftcE#9~cjdfQzErL*83~oue>bzvXfyijxr?fN831
zKI~HzxyZRa&%tSTpEAPy5pMJL^u#PlNgqV2Q<(*Trv=c14~ZX#DQaJi7fFF5pJZd2
z6{S({$JCGu9vKnMI~Kf6$D=Q#Pm0w+x#jg4a?5seCclnsyO^kK-Jkz3sGcp+P*=C!
z!j=barx}TNnc~9SSkBFHNy&%7+wR?i<N4txq`Lpst3~jPM)u%KpLLjA{W1X$nt0Ah
znH3;;dp<1vz4y7)|1Q$<R_kC+)9ik$M6kbG_T}>c?Mq$1e}a4)>)E_IY3Q^tKTo5;
zAc~}(@sbD_v@<@ctMqpO10A3kFV!RGUHf0<T#PTTz89N-*jB%`cXX8cA!~9FrRR6{
zlP;j|RrcX2coHmleP^ng9p<Kh=08stFIyIU=NyodYhQ-N86oqa=>O9r0lQNHyH>!d
z)v$OU7|*NC2B#~r`j9u^;3K+WBRj+-3<ja^I)k#E{}br+VZT3qQnz6yXJI)vn>hiY
zW|C_tP-{44f=E<Q?m(d|1#UKg8&cw9>39E5&kx_r^kKm7Pi@Y1*@qx-<L(02Q^0lO
zc7-pfTlP~b{nnQVB-WC|oUV_aY>-y$&0RH^H2+(aFgpDI7Nt7>_G&{K_<2uw-eb<8
z=w;R@3qvTVo}^o^Pv<fUy-#a%{8wXyZ*~`NcER*0rcbZ+5L`jY@4K{~^DcZm-m5Wy
z@Xa+Sz~RzY%WXH?ZLP~6s6i2Q@>`{r-VY+VI(51|9ljbEo(I4F^YEXZ_@_1&PbxgY
zCdaTJ|A`=QBn%wQf?@&|S?lFo?^rE*Z)9!Iu7eigRQh^M+8<o_*40???8D?DqJA)j
zbOg%(`Z_Og4$cBN_@_aaUkYsafyC*c$S;Bib>Pam+vj`B^&h93KZC-HS0ix&#~)3u
z-sz{tISbz??igM)n0Pr!AAce7AIA6%SDST1mfS0S7h;4_XK}Y@ao{`v_upy~^Z&bc
z%N33%^Sr25QE<`YY!^$0uWv64WQbMQr4E6FIEiHfb+1EEr_OIkVL-7D*brZA0OQM`
zK)AcP9ql>+O%XeLdpG~A0SNzkVSJbfs#?s_H#@LUV-PeVRl|)HCg**6FN?;&5gIUH
z6AdPsYqy-s`G10t#}1Z`l>AfhKcz?-S3SW#4jZ%=KbTtg+$*Ukfi}>jzF+1`8~1t(
z4)<`_s|z%Mh_|J4j!nCuh##F_H4OXjfjgX=wSZ6k>=s;Tk1;jj7=?TgCE81O{a2Ch
z@O4}j^52vSGaNyFBO7(!*?)U^Q<U=?i2G+k!!W1+mYd7?fbjYhG_H*YUDRNZT1|N2
z7>IH5Gn?L8&K~#v|Mh-_2mkLu;f)59GtFJY<AzF~u8RNlG#-cFdG*hg0X>wE|LbcA
zUw+zIcW67$2AYmOhH54Maup?)Wi*RnGlMg43=BZzp!9*!L{MhzhoLd?P>4s9oZPR(
zi#vgV`nmnU3eGke1onG@fF7F(g3U<JkmtxV>cC7dL;vGqg#OVg<ahiX89dnAKiov^
z1h{TPxE!)v!9NrrBKXXi(O2XZ^M0eA#+KhjYmv2rmC~2LE-l?DoF{b|Au{7e%(kaE
zfv~UQW{?&<`F>dVul65}{17TI8@O?8ErF71aVQ5)!YVnDb=6Xj(vo`r-UckFOS7sw
z{Bx&}k&@41Zx+#4huzWv&D<^SRc&qRaQM*WX#C*7k4!D>o(L#oJFZMHcROzMXSVjC
zb%s282;{T@2TlQp$uQ%>-P72&4&~wgtnY4FNU^c$p^!vk{OG)wlvpBG5VHHQcUYEq
zzyKpruu{7|G%)@SHk5!NV*F(E($%cx%^vONrPlTH-b=J}!{Yo&-9lx~G=cqHX*r>V
zbb=S29=J-*r#*oYeKJ4y$g!)SQ0NwWka_tLYMnc&uCB3Su0BhZRmURkvGL;keDh8X
zQ(Q08zl}THHH(YMXCt-i@%@<@ElUc>nl7t$CGaPt!?2*FFkB*)2fNseSXCD?3n^mK
zaJX0#E?*=BN;Vt&<>NaUT-nfwXh?qS0~v#|Xxv;TJew-@f%NoynhNyaIE@+dnDB5w
zs@Przi_lQ)Pqg$ntPuV+Tu6Kjf0OTAO7VmKg_yeMh8cc`$8Ykg@$_*a@MGX>a6|+)
z{Q%#9;`@T_E*4zK{lEZg92r^o1~-JS==Za+x?T?=5Ju=Y_soqdrlfl<;lI$$X?1tk
z$XzQd>o+>$;&aTwO7yBf@E#_#(_=-!0_hV2BXao|VUW8F48#aSk<Lb|TL^P0e+`!V
ze6C9~1dFOuYAg_&9<Cq^u;mf|ls&2H9Kh3zCdGQ*J}|%trH3xO#_IY@Arnp}Q~d+C
zL6$4tZAdg!gLn9~87G=ovhiSP^LlroogT{g4&waD9D0|LoDAnVJ|h;H3KB08niosI
z@Y*W#iDl~exBl|-=O6gKJ&mxLdxi-8{!aP7=?~9<em5l`910A8twqA{Jc;Ds`ky)>
zReU&n3{XNWEG;}s*w#Yh%D64O;4w8tr=kK`Sm0}zhEYv!qq2McBW`ld@FOZXPERlq
z{iMM?wESzhmWuwkyy3Db%Tu-Om`EZbXefQ4J?Xvl`w82P*~8TBPwid^RUdn3kLww=
zz8zQ@f5}5cPXx$dw_D-U$xtNLewR=RU(SaRMnD)g)8OXqxq;TB^cw>sV&rnzAaVw}
zymYPqpXh)5xAuZ=Ix!Fit1jZ&-%6A*)d}Q(-&F;t-y!`3p=EgVY9o3c!4l4H4EN!U
zi|CPqD0K#6L9k*N*yOO~m;-}la;ub`0t00raG(pqcsGKN%>8v+E~_1b)qPFv>%t9*
zL8jYoz1vk(lGS}IvM&*&6fZjr<bDoOI>cKrDNd2Bdt__9GB7SZOEg=3wtk*xgCj_O
zH<%wgIFL3I=WFp>m<F7{H)oM0l_d8xQ?wI;{iEg+Z&V)C;qE)vmEQ_GNe}<iVjc;n
z0F|sTy}W(_ReE}QFAM@rgkX_kF_1#P$mkLhR@XXjCJYYvVR{h}K~l+d5+|_6D2RTa
zs8uf%FEsi9iW2#mi6xK=;|yd{%3T@hq9U!vlNTVxqJoUwV^D%z@Nn^3Kb);OoHBBo
zNQj)@FW{28o~gWAP?XL{<fBU9{w)URLCmeL_fYRVTRptYo>vhJ+E>q;(5b)C|AzG#
zGSpz>N*v_>luSlOX2iSeJj`YzU>q*6HO<O68sj%EwETA@-?!;9l7WSQVg2}%MD>4H
zONG;E5fDeKdEd(8feIM~M62fx1R@*JlLv{SdPWohdm>ryvO3mg=j(a92Aqa4kOox@
zD?#DE!x|S};?BIrJ_lxe$EH=rwpDjBJumB<lDHFpaZVZ=O&Fd$ImtsTIKHS+NVT`K
zk7JlSVtPozmyQ*DrED2U-%k$>riYS=WFLqmd8PM|->F}(uKPC5Q)j?)qT4b)Sy%h|
z7w7${CZNA5Gl?!uC)EthKlYBd{ap&CS27O1N<GTL%ouZo;@;O{PnILZ3XVvR2n=M{
z`OA}{K^O+HOt!TuRG=Wr4OM{WMGhrGOjkr>!?fSrq5B41eYZOOMc4B?CaZSpMdgcb
z)!<JliJ%?6Qt<rFw-{Sj?SEIQ>f5WB*u=;sdWcU*l=mSWgp}_`D3BZ|?X)NDYQ6{8
zm^W&L$Uv9(Mmr6t<w+T9*wDQVOOgvouTEDlT&}<0+`ht;jND~Gl&ZKWQpXZ0F$coD
z^AK=Lf_J0NOf1BEISWt8?P%Y{4B-Tt!XX{aA)c@(f<hiyh;lK7OMK64F0A9+Pl(3e
zzT(2PoCn6@Cc^*&LNyvwpXP};jL!*_D{8LbuydeoX;QRkOb^rI>EETwLW#FC=13k(
zRC!;z>Qlx^HhX@Gt1uo;iD-U6Q<lgG$&a_oSe=L&$>z96F9Ol$oq4ijtM`ivtL$V5
zKbWAmiK%d&!Nk)*R8ZZc^js5$$2KOBz($B0^swMJ%9pVZs4<WlNay#u*cKE3-bh!_
zO~q~VPPNz^>!+#^v~#olmEdPoe=p#vWN?1~8Vw=K)qR0^qS{?DnkVpxKS!xo7K#NG
zmrE8GcmZb!d$(`ky@4MBUIlk~40VP?IpReihoQk1*p{q!Q8BaS#e=qAwY8nT6&88E
zpd{tblc50nU)&K0k6e2NHHTL8UFHvdJJ(1?ET9y`9i*=RUL*BPcJD=fo781*X2X35
z>9!C7HgyG!2lArrStuKAr}u=!?%jJ$7sIvP!{6xa8Ms0B#S+c<%xmL_R?93H1se-p
z`1Bi3E+UWHval#Ee2Q(YtPWE+0y35E=g^3~fB$I%a50EWdOvM~cAt8#BcJf_!87Go
z9PN$4J!uvMfd+~YG9}gFokn-N^|R}4@i9RQ&FnVk<q^flF?pXvaPvIXTQ8@<Q|!VW
z{pg+|U9;zDaZxsQ<dg46l;6>eRnGJKQ<c0(gWV+Qvm0ksMLMsmnK)^2<ZoGMeHn~B
zl%5n^e{B^3H|P5i5h2Kl+&0NA2Y2Dgj&OFHUS_**+gY0rZK9tM#0xjxXH#Rq4IvN4
zHZ;_sDp)HS@HOBtA%6bULua{Heimlv!1yO(n$ZyVPzuUGib&S7?tnw>OH~Nqa1IN4
z1qOr`JahnsFZ1Qw$gDP;20PUBV`ea$38V_pBGZII?mx0%ZqFa55r)quG%iEzvWk_a
zP*)w|AXo8=;oDBPO5j#Je*^?{F=a2U4hLNIjV4ZBw72?)IHDIw)cwJ?;?O|1VT==h
z;S<yTR*CTqaRN3b9vFKaOgOd3-MhX5eN|Mir<qwSU7K_w!5zi5i;#|OX}`+@fGx)!
zX>hv$oMXCIo2RyI%WKR2WZq*C*_hppb=M_qxo!6^8ut5LX0>`CV<%c?Cl$47_hO2)
z-M5mmUOdaks)T<+-=t+`Zk7n~AQ06sd3+oaC+TyulUTbx^3zwhf>$S>ECMk@`T)vM
zkc`a{Wrj#@@8!H7J)&Q}3$~~PxRbq6E!E>T_3z8t&g|L_&)kNg(Kjnf#}A@DK|{>H
zdpk6Y*jfxWQjC}Y>x-lwf-a`4GI26sRHLIRr|qO`&Z+g_=aOq7A5HA%@@G}dv!ZYv
z3lNVb8W}YremMJysXKPQ)!ojwX&B0TGhX!ZN9)CI8grE?E)SPwJoj?|CZ~$p_J>Sr
zQasHVWKxWM^WAwww_?N*XT_uP@`rkh2o=##rbKaphDyv)f{g51geLnKHIXNJ1_`*_
zPL%4Urac*W|9S1JrjwCD18MijK*i?5(Ixmf=0+T?E1&&~;B!y^T9Wj^`9f1HDno&;
zo0MEFnT0GN-c$AE067pADFf?K;tq_kDOrf3oc4<+30KK`HjCd2JjuzC;bPZ-G9X|I
zVqZ#T7M>#3bBBd1Hjs|QD<UN-)XJ}Pb#p1$10ROrGDy72{cUYNkjJTrbTrCIUr4j4
zTf&A3j=+CH+lhkYB8Uc6C$@Nu@Pv_8F6Xm{zdojI{2`G%ZU<KS`WIU~)}yo^(MPRV
z`dU8et#RQuZmm1-Nw9Hc2~T{kALCoyp^#@OR@b~A)}42l!ZbaCePEP}w>Mn~S=#%=
z+qrf}N@EGu!^B9RA2?$D+p#VT1{YV!^mXmKIibVrC@%UxSc0pZZZyzd&8Ip0`LpcT
z)XL0lC!m7bp(}+&4&CQblv+5_;I@&^&Z#0kk&d6Lvwpfe$5b%cFI?^pTq&BcXlYX5
zpQ0|oILyhMUCXau@_Z4w0UMascZtuGEkkclJX^`83Yie)bpUvIe*$<h6c0;Z9B;VM
zIi2qOxv(u-I7aDORD@B*D0ZB)kM3l7)^oc8^i9<sc6MfUZPhGFPq-ha&n~=wS?1t)
z-agALrn0KU=EA~q#-~6}t)QdHB1Rt2Tvx3fmz0OpJr?fsQ&O@iC)R+Hxr}p0@o;b+
z>gF8=t%zu<eC+&Qf=vvQz<J0}jv>;LPk2Up32XM0VyBgN`2itD;Iem{8AjvE>YTCy
zWX8qGx_$Vn+dQW_kj+)%siN({5EOjAv~c$B;`!bja9G6abmGGcfb6vkG2xHBS2nGr
zTbpZcDsL+*rH*!Lhd?I6qY`VDBRwaZ-?uAB;1=z-nWpW)mAp3~j&8g7V|Ym-^{syP
z_9ri(w%tYNciu03dnE0Y>XiSg9SfLgaX(xmL3#O3W#^Y><k!vFX+ep<O)T0XXIGbD
zq`l!Mbx(snP1>?81{`%phEGlEtuoiutE`-@{CYv7#1Yn6Pndr&&A-jAB_|M;N2LOT
z{zf}w`)!1)OB_TZ5dHnasUJt$p~`%l`*L=Y{Jr>u?J4ArhxZ)dP)!SDK2Y!^Ou}`o
zXZuCkhy1sAbtE88HZo2UBi#5oeJ<<eIxT5$TLPy;@TWu>^FX?*{g|R`zdu<>)Jw+e
zUUk1KJK)rPyUblXnT5W3Lph&~(XO=SYkYz{{aOe<$v?R_bI78?83E%iDgW3T$8s*y
zodR~Z&%FwoO?!2gDI$R0^8Nw29+)3ZH$E|qi2ucVC<C9tMpPbq)Cw4HAo&-m%k<hg
zNMc*mWjN~_QuUlfHX1G%TltjCSU=gDxjq$lvJ3h-@;w1D^r2?Hq`Ps_`nVB?0ZCH`
zGIqP1uY_6Fs#$jtPd%%TqZ<KjA6I^t1?i`Pc(_P4Pqk@0&;!M`fIjikZ4cyT8MjkV
zhQ<b4*|oTFBNX7DziL6!HJ}s`3XD;F3xzXbWlz&sm7h)cjeqL8u&iWP$!FGgaj<c3
zLbfzCT56=lN^+)Hx+K~nzG-XO(oK_#i}${R4#-f5h%j_q3Jj_?<uh3zv@KZF#GdY7
z?S-e2r2W|k{C%abuU_d@lyO%&QQUde)0ClSK`~YH=Wy%ncZbp7Y4V>E?h8Allf29G
zQhUssQt3Vy)s$N7Y0*N63IT(QqRKE)gaqlx5iuzPBD<=a{7x~hsD6GZtfWL`ls)RH
zrs;hQVLtMSNkfB>q#zwdWqE92kwJ_?%cGO5TCJ!XAb0VlpmS@w`{~dymYLHVs(i(t
zT6S(Losk6YWyvKIwqV1WCXd0!C78?Gw{LB1Y!+MnA%C;_Ys$F$zup&Y*S1PkVn<-l
zywH=fpR&3cN^AY*=@R>O$5C%kVQ}ULyG1(QbpQ=@JST1uvu@Z)y}2trpUt;$QKVt5
z;Ky*ga-dO0aY-b!(>h)4VNwtUR%UGq?(q<r28BFU*<CnIrYFKZ%-w~iHX^g1GM=aV
zS8kkyu2zHy9-4C7l$PDY;z>NNGZb}KC&j|RNp9a_FI|S-t{g!&q@$9)#I;QWnV`OB
zn9Q19@g`C#6xM829-~X9^G5qIJ(b4};k)0O`M~1#^$PFR%FHV^>C@^sU^~MEbYlU#
zlgWhq@X*eb4B^5;ac?Opw?D^e0oQ01EctAwllqL^!rGUROi|xI)!lSDq*6!KajmBp
z{kL1l8~OO}zL?B-m_0!!&3p>i1F(8q=?y_djSx@N577q?1s@TEOf+3x3Ka4~4HCn1
z!0svZh(ucA@^@soKQw>D%gzjkJ#;BXGvY7JosGi-FJC*ZNHq%iFmo*P0{#R`^b=#T
zjLw6q5*nBN6uGL@eD+ZTU-WY*jl`nzUH&e5b`A=DZ5{oFQT!Kjf!er!%UAM*@2x5)
zZ>@H(62-BbMrVe`wJk~ks~1hH@VBgl01-Mz?D}k#PDe#eT|GzKvwqC}8jPj){Y$z~
znx_-A^MC9$>+mIk5Ya@y{{9~MiMre!<Z;3GN?AgQpF3(c4hP)a1r8N~o}rrBgyf-D
zP<Uq&zVT;Pb;*-)Feo>L#-wNkbtEk!olLwXd84*=g#*rhM|Mk}U)=>e-~%Thp~IU_
z8y@%NzX_{aeN)EQsQIfsurDTc(nLlG$XqoUt<7#Q$nlPrZ4ku{to2$Y6;!yFOFv0s
zzsLbr+ggP4?8)Va#33*_M%UCrh)ieFdkcDORt8)geAaS!QF<M9F+2iPC;3fLN(3r3
zsn!&nkp+k`&4w!BtOaR+R#;5AneWl~%dCZqQ_tbngUN{`bcG+ft{-Hf{kcwUm`A9H
zQAKBsgo6Rm41J!VWR>J7gbIWxFF_6;9fIsIMB6)4;%GwKHP$9(2bcR!Rzfh9fS1j6
zp9=agX3wYXdqcy^H?)!>w|{Z)9~CG5P+n%F^qZkQbvnjFRse5IROV>k3NeDGIEAd2
z{jX5cOD>0TGZ`{$)=Zp5h5)u4L43$X!h@XwadBiv7KIIb<4xP5`WSO4ECLHX$nl(D
zNR+N`$u(DCI2<zXC`{VG?G`j`Eo56a3&H~AL^irsM{olT-n_xZg_o5OQdwGBIxe>E
zw&ZXRZ~XbA3+U;2N)R#(3=W=L`ZKB|u7SmiIy!2s`OD7ZU=ItoT74t(2T%{8El#Xo
zUQ70l->P($?MdJ0Q#?0}g~?g)kQ%7}=wvJwO%8>9x(p!=jK~w?{7Htb*Re6Q4&!8w
zVhT~=&pi=RPN&+iun37)P<g7aMuLxHn3A-L&Kffi1aKMqJMi?*O1?%Vgf1B&QiQbl
z!j|6bQ!-NLhqlY08h>2M5k|$hpzfAizq?A6dsM!vq{7|z)-+Cq^Synzy{|s50$(E1
z<0?HTAR*n@Ygyhh2ohsCs9VoYNS<y`Uf<F}Nln{4H#=K4@hZN(r^k7<E9Mu@N7;D9
z)&c%w5P#X~`xZG)7MP-dw}{51RX(AW_+F{|QAC9Ji)Japg~o;l@@#VQof$KucV*~b
z2dCkeNPC@3F?S!gQj=28!vprwKK9;@7i-J{Dn0B`!R{4h!AEj79N%>Da01OB+1cs4
z@BgGz20BwDYk?QSb+{dU=sN{C0e!N7fg(ed7SKmU{_uP`?8pRU6U;Pxy&r_okm~`c
z9o_Vr+u<Pisl1q&iONhNe`IoUGIYLiyZ!csJR}-ARbeFQ_qR$24?^--1oqH<?>M~@
zShWMtQ@66^r~@nnD!rZc`HH#!2ZP;(Hr|g(8B<QVY*>u16;XD@3FqUN=+?tNBybMs
zwI$cjvQ}tnBt0?Fc@mXqQ0@XMe?hCE<4b=CltIL*yl$$}2<Oi@mFwR8=Bu<P#uuGz
zKprZa(Lz=0)(}khO7JfFqW$wvzE7#=TU**%3O!=eXHub_e5)yD`Bs&NM)bwTkO}=4
z46WFAr?;)CeL^FcuXfO5lzQe@Rx+U6!+*n>jGP}Lo4Z7dbr;K}fZ0cnXvWaW0cvdB
z0S#3R#?ogb#ARW|Q7Z21Git35*xH%J7&q(>Wq`iF&jOE0v_(R{OmTc+daojJeU8Yi
z*UHW00Z!dZJWiNMp^Vh)?J`V<xOxW;IU|D?MJG8`D6a2vrVHx*<$Ags9gr>Rc5@LE
zQ$x03+*(uPeWV-5Amy_*y;HHEJWIhCt4$O=2^zPwN#Q1jD)5T`_3%fEcR{s&+-LW(
zuibLEfr0Pj4Eb<=hkkAu>jsV!TLkyUXjN@C@a3S_0VZ~m59R+pj&xjMKa!dreHUmz
z<F-Fw9xbGxqcdUq^se61B0!8;qdOs5wRVEDKQX;-5-y&(Uy#Df!=ubmH)*iV-G3;7
zLD`k(vx|Y*Ys8DRP*^=DW+%({a##4Mx!GrZ)?e|l*Tv?BxZqd{Eu&rng}%8$KFUi%
zhFFLHnM!=+r%cmIX2K5k<lnQtIm`$1$J$JSj3RdZx>l?e>R)=l0Yp7_LN%+Hmb2ur
zR1NIF4mLq@*&ABa40R-bB@%g?dEV&g#7b~#b8IEctIB!H3@`Q1Q+hk=RI$vs3O#~`
zE|ufvErmj47zw3bgOHRIY;0`3dI*G~55M33j@N#w(WcScj}oEnJv8=2PqYnNcf5t~
zTdPrIdmWN;tP8431B$qsG$eQ}j!t$iE+l-E>mvPR^m2b*Y)Jd<jC8MXM<#ur+K&17
z<`F^EnC0>O5gikOGSi)oi0CD?0>_Fn>v9bD0LPu#9@)yAuHLJiOh~j^GQ66nROn?q
z!!u2?bYsujnX(Dnx|xOr<?^?xLZTz&$=-+1DJ>2}pgtEu2f)q^ggeu4msdC*hp+!+
z%l?$HutB5r5DP-8hN70E4~`6oewbceLKA4(rCVB84I-!OFWPSR(7WQ!OT}@1<6;{N
zZPZ_PoIK~UV%yxzn3YHEpCRk(<)8p!GgMS{uOlYrL?pJ>hny|em!U^o0dU*xa=Mw6
zzWp)mf6)E~Fq?dvON(KISeL6&?9Mb<VfMmaq&*>Mui;Ktf?)GwJLkgXob0aq!Mx)}
z!0E<p(`huJVrT@Xn%#-d(f)FN+7lzNmxz`M$NJiNhw7w$M|5XYJEVNUK=gdi*W_y5
z<muq1KR$8geB)D^*jk}1N6*3{G3?w_3w_-OqXjmAzfS;=S)a!qBs1alIz~|^W#iX=
z3|;F*U%!bzW>C+mqHG~UffpvX5243}SN~3+H?=HmXE-BR8Ib68U1pw)Bpr$kL0ag4
zQp%5iThvcZp!gR-7MM6AlH<`N@>ro>skcaZJTx$Inttk)YEn&73JH(G<?Q(zf1$<0
z+IvXAvx`czhR#sDCSa$S@f2Hgr*knL#vPA<8E~3I6!1Jy*&+w+A)u0T-^%h!PLrRW
zeeCaLb8;MZf*CH4?v{$Ae)a0E?}Iox(bKhos}c5vr;o!CT*A%I3AOT~7*`$AA8M-W
z@T8Rc6ckuF6G=Seu_hZjVq)}Z{W<tj6^9(S@Ucp5`^ZGL?w!p{ko!rM5(AcF*tBr-
zkcPr=`${<RLyV#0w(f;R;>w?+KQi(K!mwT|Rww(2(AefXm7}6oUn~6@==?HcPaOWU
zqy{d_iC`MDM;<nR<;>VXOdf7cl%nt%I*(N^ug;I)D<Go}iPZjd<<)HEp3xg?q#bZ#
zK+#PYR3KA+$2)^L!4p9NT~qui3(XG3P(Wm%n(B<Qlox$E4T%Q5?4@-Cb;-=RD;kMh
z@xh5yHuo5CO&Uwe`Z#%A9|Q}Dl}?0e&DkNrb}}V|!$B#;Uk?*Q`4Rb}jm>3!&A$!2
z?MT0K6Y{LBy|s5h;QFE=koZ0UHsDm#{F~O0PKV*jQZQ0SKl;O}Y-Qi&7tVG=jxU_u
z*Qop6-Aj>OQfud5*_fR?r|J?eNp>nHwn-$<_N|4)JWmJK`ybHxUy*X`_aQ!Z(}`Gq
zds{eW^-}We_w|r&$7O#SXi(qhg3YC>VzIkOnKC7PpYfKO8j;x)IG^q`gaF|r1RTMh
z@Cr>Q;>+ujKF`6y!0|viTb302Z=Zt*5M*-6DP|v*O@$^t%jvtlv`ta9d}>9@s0F2H
z7MO(>KO=-dM)W<fEH_RmR<EcR&x9{ec_nUFta#+HEZ)*W#|;Q^={OY*wHr*!d2z_G
zifmSlEIw9~GTvWlTJ%1@`UM;fOI{B2#aLKA=)uDM&<=yLn$wet5`rtCVavnBLJ<s)
zC-3tF&F#-+XAzp06{iGSu1F(q{8i0{*7_D-jFdSn5SI|~c05Rfan5!BvhB8Q7gf7I
zy{ewZED@m$m@#1Xq^s;Kr;0I)Ic1vPv~<?Xj*@oBv1?MoU?uN>E8wbh(SP~XF!0AJ
zU=}AZxwgp@z1Vg;pFZ|;B;r$8GCPY(I&(2L2Q*omy&yKP&xgTfy(^{iI-~NjsF%5!
zgS(rVyNk<BSdN#mB1hkAB_?gHN_7?+wQl9NrRg)jiql&r!(=9Yv@X<h*7n$!SK5YT
zgp^C4<x4jT$DqW3?g3DD$bVK&Cclz}Rbx8B+oR<;@XXhIcq*o%k*hV2@Il|Xg+xX!
z5H5p7C=0`Z7OO`6V9&BmCyYiE<0w=ordJ46Ox4CnW-a41{S5MXYD8XLpmgS;_IUaM
zt#k#3u2VOh6eq#VqZ1cf11iH&pd}_3gv%YRoj?I&>;OQw;I|EP7D!8MwxV%i>nrqz
z?!6O6Sl4Nhc!F<&_|3$&PLOTPpBo(fYz@nLX7X<boju&~hnglIe^+#8Sw7+@PLaXI
z`$Lz`<MHiCeD7_%7MWJA@~5Mmxsc`9DP#$^icR$Si?p3vOrjVb;^U}5l#6Cf`LwS5
z+Y|x8EDnG<FSn%6$?1gGsJYR@*RE9DT1}l+G~nB@&iX@*FtE8`dG`4B$#SlSflpQJ
zZnpnrs#6(W5dyr&C^&sD4<jwWRlzD;<b8mlyIFYvcw>T?M_08k2A~dsGnDxbVUw2<
zUL^Dh5(3m_Fn-vEx0BG$ET`ue=5GjU(%{kbQ6ODL+rHO_m|jzCgIfnBYFbk-{FtX<
zM5T{%BV@$c5Z-5AIB@AbLq~kbt+x7!`9p{4Qt;qW9?gTFs2=F9#OoJ9zwh+)laU_z
zz0vW8>gdE^1!Ei~9{CvBGShG4ykI3lU>~ABf5VdtEpc9zWml}Hnm({ZHHFm|C3C&G
zLkRRHXmSL<q`p^bK5jprF<6xCqe5`7>*T);!i!N)#nt7%yPmEj;K%9*=9Ty?#LR4_
zzVpam8ONuSI2>ov%402nSUoWgY^5!CY`d7nxMymt<6WcH*GowxU4QgICE3gClvfI#
zQ5|QwxhmeF=d=s|`IbT+uI>GZNKu7@5daQ;oSR`5ZwI=i8NMCRYuLz<Vyi*IVeEWr
ztYdv*LINsdFY_B*uWNWND%)-r`t^7NcFs4{43bz~QRE4jNqqIS6fMYFruz_oEx*t7
zb}fmGd2-;=Lb)NA^i>B&J;%@2&h8aGES$E-%}TAmGbpIOUifqQDYF%>EPY>7EYR=K
zqz3nN`PHNskevpi%uZ*wL69RKIV1Y-CB#DW92SW4B24a3xb3{I!}M&k?ZkW7bcF(T
zN-IJoKhjWZ#pdaj$SenC(582F#DAZn<znjbgpN4i2sqC<-J3gdYuo=A9y>2AF6{Yd
zAic>Zmi)B|G-yCYg`uD>HlnW&VA0PC8s+|+mF_BE_ucBTDAC0NH5V9kC`^|xt$k5K
zg%D6?q?|NVBbV2oWYwrWIs$`|cEb|aE~e3|%!9VKGt%d5(w}<{(K_-{D3@y`TOo3U
zM->#sAdy>UzdsPDCD)WbSu66A5hkbDB}R>nuZY5{flJ|T*lRY{ydwr8xI_6kD=b_l
ziCx@YXFQTGUtSq04p*P|TnP$k0vn3td@;SRFqOxy{7(1X{MUo!=S`3OoWD%av(P@*
zn1v#epUpDQ!d+57!;79WxmydLf%jy|>jZMUZLXHRNx1tj_DC7dPg~5sh1=a2Br!VG
zx6p{`kqvwAfpZVn_OCi%R}Se}%uX;+&)GCbt-~KB#0ez~k)u$RVJ5}pbNxg^>9X--
zuEUufCb|MqVuz51$`FV@<J#cXj9C2?fu1nt*Q9V4dlk|l!|_zsXhnpDk;9@~H%aT8
z_+Z`Kb{F#{hE9OBuV!aCoDS9G=|myF;HVd&d?*DA13L!yr||+J8!aTARaJbz`6xn!
zfA%Z;=-8{VN!>oBVjZnGUEL=jwlF(mf%3#D$FMEn;$j`6m-b}Nr2^5X$g}}AMwsOf
zCRq!`m3ar9Oj$EZk7L{0u3<0Zk|BjtsnS31fMA80)|oSbh3T~m@4iW)-aKAtl`v>w
zk5Y3s>t#ui)CBImdm`Xbd`xA}|A+(8f>A9%MCbRUr0CjWi$H@IaLuCY?qj<nn^3;E
z8rTUBBa+*fXhLBeyDf?oIw{=-Dhgqm6jsWUFag)0wDX?z=bT*=xpwV130jQG-((V1
zl!AtIYmeMO^EK6H?_|%*%PVYn&tYu>!R2=3Rz3N%1E7yG!-G=DLAkiNd@*}Luul@<
z&dWO>_6cFqd>e!hX2D2aY-s!vc5+}Z-|!sHP-4H+ky0@Fc2i<?vdgk59b`_)`)#17
z^OS~oOz2KmZ?77ezr*?Awo~(A#|$#(y0f<S?p<PSQ=*{$nnXF#jk;J!*ximKD`MaK
z4^k_$fd>_d<|8S>mBv>nE?&v=@v(m?3WJYCkM0PPVdagBjf%;2zAOR>1a>p;Qou&S
zVG*IQ{Eq#@uoq5k*F&Rqh^DD>NIaj-h4c30*43UQ`R%Gn<S3}JH#axe*VpOPB3y@B
zPG{|SThCTH<JmiN+74nvjLdEBD|X#K4=)1LO$-|kxlAs!DA+L4XMC@|`yDWnif*im
zq{`Wr$M6sAG%vm_fnZm=4T<A-wq1(1)q?FTlfjy6BECarKGTg~2GIPcq~L81?XRN~
z@kU3U%nG6cb-z5dHBS_I33KuHFCY0jLGAt^cOqMV9*79i2<CS&C?!CEO6dHq9uTnU
zbh|KNxbQPPwr(z~n+^aGo2~a1F@<&_+qij&>|w$QgeXFIUKaO5G!#)EltNags#cd|
zU>QBBoXC(@UIt0kvM%lf{j=n81DnC10+<2;slt+Q{3IpOA5Y<&tH0!NQs<qmVb>gT
z%ziDTipt63g0yU`QMCoSF-PkFqV$&+jKgo~>`R!@$;($rf$Gw%A&(l}LapbKhl3O&
zn5{^0XXlGoIp>|n+JXYC!)|t^OCiXLf&THo&o4cIwS!9xuPW{An9tU7gVp2Dvtwnd
ze+k|<kZ@7w$F!<Ha=`9j4-7$uG3$MF7yOt}$IhQ*{5D+@g5}EriI4mNFdq#1IytPj
zWcXRI`Kj$#t07<!8Z!dQ?=D7o3~sJDV^@_|4&Wg{^sH+WaAn<6E6sw7>(=rg)~}N-
zy!V$nd|E2%P_<UWhlj4l(a-lI%91~4Wi+(Zb0SWq_}ua}7pLFc?cibw4SXI>@}||#
zCUb)`5ko6F{dIGb)e$8HSnN3LJ}cPfsya6{@8^6$#>D@WbNWX^HD<Cw#qQ~&d;K-e
z^(B3b&uX#}D#5VXPmS?-p7(=gk3K~?j+hK0ojja)7v4yDZqGiPGkiGT{N3gH%?McK
z6>pQBQDQOs;QFaD3uV5hXRUre^UC~AG*ZX8VxT)hVh4j27*Xbi0xoN~TWWvG(xo%Y
zk;%&xgheF3LsJIg?zfJ*49F3h@0;DJ7%5|`)6J}VpUGa#Ch{Hr7~?DOnuVqI-jlQ_
zvvNNjj&hl(KiXE)be^BHgkBWL;q`ShA<Ug`z^ey|lq+_fmK-sjKSd^@SXYkEbbqSS
z*!OR5cfdt$&b`&c7Ql0{JX6|zYXEHG?OsMz{H=k)dFN*J$oD>0KCEgA&9jq^L8%`b
zCCPh3nM0Mv%r|FKgaijBPA4z87SB(|2ac2I&cBXpLOVo<i6+Gu%xSD<sp#aAJ3}-V
z@0*3u+@<{jCtg}Www{ymN!<26kVvp<&KIE2Qem`xM!k9<0)Xd<=MEu*>Pl-1Mk`|P
z&t#GG3CG6#dVgBvkIpwgR2j}rPEKYTMDa3Z*nx(TMamDf5?;<WHXOr8dF&V0Hz2R)
zKXA4(M;J^w@#=3?WsHl|8DwGCQR)#v;8H)y%hsb9r`GrTf2(ht-ye0k7t)8y;M(n)
zdpksHs&H@sEg)R=LG#`0^b{?HV&<q6q3+m);ZArUJ3B6`9E1>;Xh_sd^>d*!{*T@F
z2KggYvZ%&7v0A%}tqFuj>6o?9r-yf@9;HXtovcaleI0E$`2KD@P~=fuj`u0~;N*B;
z-yi}}%EcutE#b-XB13CgQl-*7J^~^YYmuLD>YITLB!rn}YfOy#DztM)ZWOBV$TQn<
zaiz#PCyXR8kUZAdv9Bn&Cki?y`PtqjjDx%BYx@Zyim{KUYn^kSwtPfHQ<Mdqbrh4B
zsDF=I73vjG*r(uwdid@9u2z(4^mM$%DIePNnWc!P`4mwwuEhb7Wmx+~^*c_%2>tQ9
z@%H;dfKYEx72Sy5UyiAB)UUq(=^^d5Qr@F=GZS6?P<Nf(jFF}75Y*xC{D=(r!KBAq
z0WF)KC&=_>ZCq`ew}l2aTNERFRd*p$xVTsmaM@fkEr>Hcn<-oW<2`#R%Ay2})L9cF
zt@DtYdL6RKKhKx^6^)#RiiLiCYpayBbk1Bez}>A$K`BovpWudwjGmnJE#VZGl#~<{
z6tuwyfAmkGELb57DanL|MMZcave{{zlLlO05L!Qc=~SGe#axf~F|C&&&lO2NxXdih
zds*K)()z&a>)zLY&79&M-Ar%s)_la0#e;=IDEufkkUuJ~hC9p@{?4`iafgT@tC$ub
z$MJ;`EAj3l`_u{8QTgDYyVZ+8`a7&A3i%aLyV<nF2_XDIi6F+{PsvDC)8@ywqmC~p
zKC`~$Ld){umRZ^GKb{k?I#kJ%<3VsAW{N`TSxMTJDU{(HcW`WY*df@VODTy4D)?X3
z;1<R)jruLJsxIZMq(=PFjIz%>e!W%!SY#=@Nv8-zadVnn+20$yFxwD?`R!RrJAFLx
zWm^>BxzkQfy{(8AFH436S`BAk4{S(<$T>qF4ChFD?6fk+a7@d*H?wKLCajk7TO9Ep
ztF81~MUX3^n|yY`GMe3~<fsUm+WnxkC%|+WQtEKL)~W5ta?k6eZ?wXwDZ(1ai#b3E
zXE-GPI4_x+i)AFg`|7(EhaWl%Y2~r&^Cvwn^L%N?S%MW?J()RuT>Cim^#6bfx>6<$
z7eCvY7`57JFywYCC(@JS<vMTvJlZZrN}vBIYP+2!DLK(WM9ZjQZJ&IAxds=IYg%wI
zg!1(X_QHZO9lm%X@BacC`l(2#MO2+#8cdEJ)3~s?X~S?Y=nUDz>uH;3DDJht^y}k;
z(vn91%Tx86HT9d6RcQ!4HYpa=3db1+PoSXqx?uZc#|WfPMwFD1Ej_H8WWd!bs3XQz
z;a~`5J4E4|LNrvES>-xQ*werCl#0B3npc|wkIsvTxMP|t;|btCAea4|tYx!~*E(h$
zr+9vYQzz-xB<*#Z#XQmVw}$%}vI$g%{c_{mS<JKpLN%sebe&QyPL`gtJt_xNX{-AQ
zEf09J?kqWao(%^ancQ^WZvJ<(7VVZ9c%uAX@<joKigF~VAE6GV@e{Ud3^A%jjsm-i
z*9kms+QlS*PKkTT!gK?(N@%;#GyTt#qhBYK#LB&FIephU`BQ&egT`8rx>t^%+WqLT
zA&J*+k9sr24JA&olXaH?zj8JvtRd{W_9Oo6*j?=C<i;9G9A7$C!AGJIYP1R4uXxaJ
zC<!0@2m|e7rs&hZ4t53XA2Z55GFeeS5nRY2ijZ``(xLv<nD*YHFm?Ak+rDGnrpT`o
z(bb6Bbb=?I^l<18O=lMR;rBJNL;3uFbl9_Wu-WNx2$S21=<ld7LKyZqaXXm<)uKuc
z#hWa#7)>?TB65>I+l!r{O}eNkBm@O(-=7FTD7+v4>Y~aAJ=q#=!^L@fp+|-p&&%^c
zhyV+qOeU{M@hd_g6eg$iDg=u1HB&85CBDcvni(*UsrmkW<08fmbQMebCTey-JW54b
z>*<NK{$EwPS;5L{_UT=hbq5&@B~>{B7Cxj$re{ot#4sr{p&+TRYNxjh1$9yv>nLZe
zTBfOvqe`)21~oz4;huvqRmx<2VWH+0yD}hQSQ8+zqbO!0WK}as0i~i`)u4>doOl;&
zr{D5eJ^_tkaxf_WQ=TDwWNYfMZqITjw=4w0VE(zVqR;;=8(i#;s}2hoT^6MDK}S+z
zE0;&~B;cb<nuo<+j*f>18u(DlJW6@Tqbz)h14~Q~%5*cUs`Jruv*5_b4ShY2L<7^+
zU<L5$i}PIP$!sbAEwIioS*z+Ebw)<krcJUO%zyvp5(G^Qt67%4F1<^aEqyS)DwU{o
zHUkQS)775S?ExJS_I<YHJAJ4*n<_Je0qtO=c{L=CmN*$H;^j4ZMD#(&X71=8B(e7m
z$blF4xCy;AO3893;Xy8u?Y=PpDX%PV6{X7ZI;+krNlmRx7m-WWCkq>3={oi`LT6F1
z%RusUeXuieeCYL3fAv>3U!1*CKfme`M>m`K_lycDuQ(z<Vk45V!(e$|``XQupCzQ*
zxAQ%|H)g=2HYyU^wBVSpTWUXIEmV-~$#MXwsj=EYroq733*MIFFkZm#ZSb&awga%t
z7L+7(FNnR$O*Z#IkBYMLw;ETsgZ5>QCdqk&yA0OFxS>KZ$?+j58AaBpfi+<-LWH|Y
zuSjh2I72A+5N=@nSut2;<y+~eNNB_Sf}>6o{stsT(hXuCK2oMI)6vn2h<tE;ldsr@
znJv|yKkHG<e%<K7j|*t2V0$6H3r;u+8AgmOKgTdd#0;K*g<y(wzJ39itHaThauCpT
z7I%{m7V`nR_iw@;NF5orUGt)i@=3C%Im<!Rfj*1!mNLd?4!oNsKkMYQ5iXj$SNuDy
z<XcJ#4MSA|d!_?a|KpdjUR!;MrxVsw`DTjucfFdB?yHmOK#F7KgW?`a*{*|uY{{Vc
zix@7U`L7&u6s<KA-_604$-r0o``TjSe=)-E!Jghk+#1kgL}CQJJ}x<z>F-%3j*pBR
z)uo3wlNaS<=r-T&$-&VEOCB<Q;-iYvOc|qOvy98vicK4{nL>``3n;4TQQhk~J*$mu
zpQ{7K`pL>fAk_4(NleglooF4R7$0tC_e+i1Q4(5GYeXidGgM2^B{wh^Tdn?a=FR|%
zjI}v*sM|#@Nez#*1=;yF7!Q^7%JRsiMUv+|PX*HV+PHKkiZTCNWTKK{VM)}Fh4{Q|
zLtn4OAfKW_CLU%H6GU<PXGd<6CfCJ;pPDq}@ki@>K+BPE1?p3F`cs$acCf4z6oJtW
zoN)8tA^lLk$FcV<tzw^e5br$0>yyR$OS%xzpf_8SVfpnNvSqg}MMy1FStgLohmXe6
zrcc#ElTU@Vmz#VrI%V-u<(s%#_GEoVf;ZBOt)eK@-T;qE(2(zY3&x3QT^9qkJfTvi
zcvenDfXQTyjJcROCM-U4@)b2UBsPjPNnTIc@Ow#3;<i48y;97}Zw%Xb6S*VSP%QY6
z?h`!IXSp%!Fspl(p-aJRW~!u_8jN)IY~@D52-3pE?XVhEYJND!FM<c6vc#=&Xy(M1
zGvV{P#C}1V4%%B`=M4kqtA3Xmz-iG2@0*(w;C3HPplvueXTw8jzz2z+pPtWNK$0X+
z3Jmq%G~C?)meIHP)!61P$JL4(H|D1yBtudcBW$m+nFu2jDetoIHM*?M&R-4qvH@Oy
z_Ycx1kA2bEJ6=OuCozL{eQ8HWNBJEZce6IUX}W(3#H~_kDB+ryM6G}|lG9DVZIU_=
zu=eBEM+%vTl=25XsSD+foJ^-XqF&G-WjnPVdJ!4tbx4fp`2~_hdcQ~A@1m&fVDv?k
zcvgogmsgRw&`7^J5w1$07H6rIMfRnJu}P!rRR=jeGq$R`7+tnxtra?N@{#71nkoyq
zngGP!v0`|57^j`t0`^H#6arI%B+&v<@S-tRSHcnYo-H*5B2ujbCSOA3j+PEEDb<(u
zyqLG1P8&@_B*_WVS3i9e$^PQl%id5k(Olo49~wxyCmyhK_{;65X6tSHNY43>;hO`!
z+hhxQ?2A>b;IEmOw~D=^)<TEP?}I8T%k~zUFbCaq-skISclQ4{A591wU;YKi)sGt`
zIVQY>`ogbI@_es4!tbBnP;3cJZuVkOl}{6e7yasHlqQ{=ZdZXTlmCyVvx;l-{onXN
zVkpW$q(oqV5+YI}U6KPyX%N1oFlp&{=+Myvq+=jCx<f)lL>MJQiBZzhCH;SX2md3^
zIN0mH@8=WO^}a5tz(3cy*OS-vyn#<P-%TC)7h=fSZfd-~&-Vj_fr-5QufQjtEg`!}
zH>zyIKE^;*At8xZz!-_F$;hmf_IbPauRHxxZ@-^!+hrxO!|M_lYyzWQXPuwkc7C1o
zW%_hON&NNK?)|A#*0-+%l3OQDskuoEjScm`=H;HKc)k5i%EsE$|2tn$A|?sc1Z<0B
z#<uDTB@BOH4q4)kD^NvHb}P7Iu8y%O%iAlDqId<xW;KzeY=(}O-{q^vO*ZQj8&~{_
ze2%iI<|I6QPghm|v8N%I_g&i0ur7ghv*{leCGrd;k`#=P<yKh?zbi6Lh-=&%6nS}W
z2=f-9qB#;3OB5lzA<}zs%b*)4`2_F(m}HYeg8iNlTLJ$4PGWoQ^iF;+x9*k!sQ5Ka
zeV8}W2VoNy&DbhkH=lK;f;5Om>+(D-dWE1FU0ft={+I;9>=nPWjhQ$NfBw3BfAS@5
zGF7x#oBw}Kf#y)-6iq~f%bZM6^?2FkpMznCuNb#EI^T4vw}QOimafh=@B{lLnek-#
zIhyu5ZR1(#j#~E@>WqR}?V0Y^%*?9Y!8O@A8adC+GzZDwDi(;=)+XG?&hh$B%W2@n
zSI7CV+^=u|1+Cg!nkQCD6U~MOOJaEL=Fxt~BeWS~nU3JvR}*|~@0)SQCQqCFGLn$N
zzZK;JW&J<4y+G|Bd)$%{)R^v}aEktVcI;xwgd@@#jbHZZuaslETg;1lmlY}3PeL0U
z6C5#SY(Nzva%;VX!Mo!Jky%bWsoEXx5wX>Bx36r#!I+<VvCKb94-d?QgL?kc*R6;1
zH>jZSc8?#El@Ckh>vbXGF4&!86}{{@62%xDCfW0v)%Fumr<o@hN$n%_>@oT#I?1r=
zwUY=F7`oin;(oARoO`#V;<oVk0j*-jr}FZ0Jv}{diopWz?K6eIU7S9+G7i|pKG)0=
z)2RX`&A7jTE5AuS-i(og>ZKd*g!7od2}yT;lrl%iBB&F|75DISbE|pIjOru3LE~yk
z9FiQ5Sl{z$2V1xrq*+VA?<h5PsNZT?nNZ*EwLipT@psVW(IgjL!uPrLj*}24ww?B;
z--Kh|8A>`eCAXYv)}?^cveI{TInpPCD!_2g)%>zcphwy4|NN(hF7Gmi)+xlPi~ZPB
zGpq0D&{AvXtcG=W`2!c9tyhNwe{Fdk?ESkq^{5KkNy$CE(88#FOR2jjNcvSqR2Ot?
z;99ufqeaL;!w9ICcwE)Kn5CYi7?~H{dEGI2heg6jqHU*39!{(M53df?`kn6s<E-LA
zhyDLyCq6t4I_u@oBD~(b`a66*2JC5FXU0okfA=eIK~7%eUalL)Kgeu(9{c*Hfv!NZ
z_wx17RHn4Cm$`iqa%)e+d%(ABU=Ts;m;(ugk;6#j<zA|BLRC4HzB8FPr+bC7&UmTF
zZq6>;iQ0mC*PA%1H<~(nl|M8A@}fgP+YmQQj_6r|wukMI_}sdN|M3^nBCa3rVaKwr
z&b>0;*&UN-pLf6+E4*S<1L{Ug6dIPt?WXWi>`o{tC0Q6bF+_x>6kR`S7aQrK%nO$`
zAiz%@-->=DsA=oyQR}s}%2jQaZEnB~UX+zb!^w8CK5+(n8>HbJ?b`u;yf|i~miv8K
zN&6ifnfc}xcI&9&XxnSpaaCiN&e3+DjQP;ZtC8j6Uu>F%-|1J$CIj!6vDv>m_ghq`
zn5R3d#YWAt)++c0MZg8^k){78b{vE(25v1b9}T^B;)BqvD(r4!ez0CnSOi|54YhU1
zvvBXpy>fbG+a(d2KMO?LJz`4EPMwcAQjiSSk&+ebmdd-iCgnrlPJSpy>_ZhtqcZBI
zDYlBz-D$^13frNs6dT_bKiAf`_53QHk0t=spu}s4RHYZvZ8E)cJ9c!)uCbxvp^^}d
zcia^x7>V^oiVLmcHq}2B{{4GtdeG?Pr#MB$1VR1=oi9@4Y=FMb&hNT^Q{_*GUTS?3
z6!t6m5hU}g@`JPBns-y1lJ`i5cRABLuD<E><L&i&j`B0l&k>%vW$<UdlA$%>ns%SR
zZa7}rwFG$-d?#^3lbId`j&4=WHJn186sc0|Aq?ddpdrEo)UQ&!pKnj|QduUD!Fg)s
zj7B`SQ9@;wYhA!k!hQMxq}5w@uSg+B@V6(!pOxwA7dg+hW`2ro#p8U7>b>PBJZXQx
zr1FYsU$`2i85kPqD=5;nXh@nVVyGIUi@f4uq7V9C;WmSLwVB$>D|p0S=$ld(MTdvH
zn$pl1cA?sb`Iu*(Mi_$@>D{aJad3Qe52eEis<OnqfD0&`pFxQ@z-M8<XHkeFoQ5VK
zLC9}yp#wFR?9!p1`Vb{`mQ~Os!vNY92WkhIMsYZN6zgjvlc~B|dzVzv>8+ZhBTt@X
zP|}+5Nf)s?)Bi^*<F&+UMP0OZ>Q|(!-!rv8E61%u8MhWfPvlgp|0HI<LT+8YIb^68
zh4h*LT;k(pSvfg(<mwOjgOB(jD3T*0*_DCIi<Yaybz#=WG(_Cibsip|3>jm1cs>2^
z=5Eu<K1}&$5xqsg-sx(tY{0)=D&Vls%UgiY(W(7Nj89+Rq&)kX-P{YRB3;)I0*`eH
zxeB+W`@h}OI{WO=tWj@5(D}y1<6)`u4^urr-iVTZs^{_phOLS{{Zf?Il&|e=_@Bh3
zZ^Jn<v(p3$QEkq(%ay*V`?6OHX`TD``0p0>h+EQTH<?Tl;*-9A=irKyt#3cXXSy7E
zH+t$jS4#I6x|Wf-g4~2_Z)C;8rRE8i=LS?~rrs{*2zFU~s=?&yaCRm`uJ7OR+_@=W
z`|`^Bv&TwLOD%8q1h6-I{Ww+@K5-k~&y9SRx#cD7^G8E8ljnscDgp*o;QCMaiTErm
ziQ9!2dlL*Jd(X{sw5l7E#K?qj@t~n3m7~%$-KoGAozQjNFGrY@F(tt7GD8KriQz2y
zuNuY%NReT;;6D-iUq{v@hTFbYKW6|AKa`<$I3v2|)pl{Euy#+pL_>qGi^ta0M#B%K
zldJ8BQVx|nQhtjg_OhPy9=RssUN%>Cxw6h(R6h{+B?qJMerz%5Y_UzYRc`JS8pgJ^
zqxrERr0HpOeK&lPrT(Z)RDPNd*C&~*pr}l~^U<T{Ec$~1J_D^2whx6xz<?~lZXS*<
z(jr=EiA^Mr_)<TKP@JbYisSp~oNT~Z6#xC675v%O1oG6f_!(Z5fUR;7+ia&&o}O;n
zk@2Y(7Zh~6=oPd^dGx-y0ddH1$1RCD+YAlGZkhZp5bg7`sSy6XP34C(u?DGJW+pR)
zshdREVcA>Rnv%$8G?Dzj0F!$!%x`D&b1w@y_NgTc3q4QHn=|Y5uWa-3-Q50qb=Hs<
zIrQ-pu^9IqCUc!5{5jH_i8?J5BB3mG%OvHpwL)5mrASU~H#OLrs9RE>4#esp81F*F
zo~E!+M!kZrh8^(_2@Nz;g$QaglFRXv+|5!+x+PD+NGggNYcAm42l(Z>oSWiR(PJD%
zErJLhQrJMPB5&A%f&Q9Pk_hw2ZVhyTS~J;hoZO0;%tj|ehsA#rDyEzztk3L*0KG}#
z%c#5ESErp9X)vF<Q7Ny#Z)nvNcCW(7^^%ig64Yod+Jd(FM>$BkCQX*Vrh62#Tpn^{
zX}e?}I2+w_$&qpZVmsW|b@ge3<;1Mf>Fsuus0m$Vya^)FO<MvunN2oM73_n{nNMO=
zQ>BLQ$M&g?7c{VpCaJu*X^r-vV>9b?3-t0zL(@Qs;Zu0gU%MU|p(mfq%E~HK&X`H8
zupc-o9iItHH&j=5*^1PD01u}JxO!GG_#xm|V&LfEVGX?N-<5&&vJV#*7rV@T#x~r6
zU9sGrn!D$>dbMA5<$kg~iykv5Kk7}^3ObR!EFRk0l7qq&C{BoYm(C()nMwv~YDH{q
z-?uXzMxL_}{Es%48;;2J%_ltd4ZgNe4F|X$|MVD^&9!~-s21`zTzq26hZUOdmQ+8r
z^)vtknE;*FpUWI0!{G~PcAuL;9+y`q78m7Jz~yG0M_b}RS2I;ti()9MMhiHr(JVfg
zLRO_J$<SbZT|(sWsub)*Vjdf`O6<0O%s=LQ<RsdjYdx~RzKS<I9+Wx%tEUyPck6oZ
z>D8j_#qZ9vEEd_IlJWPZJb{1u^Baz^T5F10)IgUWqS?~6l(EQxotQ+K6f<M`jVJr<
zbCMd7wX(HUB~9447+uz1*>}6?78heVr!^lpj7XHtWSC8?RfHhejrrOB=L6#2eW^+d
zSs1h!R6|)vEqdNeRe`$u1q2yro>^A*!zzJZnkq*K$3dYAg`<R!DWd}n=7##NWae9&
ztO2d}p6)PB*GV<COLZcht-Pdia#c_GhRs{tj`}t_(pax=&~Zj7X_n0E*oF9X!OUdV
zT0s-|H_4%xm`0c?1<Zy)5&ce7^eoNz`UX3o*oYEntZQiLsG<9&gPDM`Fj#}lZFtRs
zYqQ^`0rSIKx7t7mKJ&Lt<bCvLZaFH%<8yO!z27PeoH$W$AhFi`$A|Lr=PIwCE?w^D
zd{sB(r-}rpIG`E#`-yBjZC0NpC%GmYLvc#g6E(@Z{w##@32myTp=wjlrhM+c<W~T~
z@nuAJNh{ktuepnhi*R3i)PPu?5oN2s;MxgU5$TJ&siHO+(b44A@3#oJvoQIpTWGC8
z{zkjh#pC(}HT)WhrLLXB)Pl~VWIlu`T{;odH;N*(vJmoFvT`I*Z9R{YR1aJWrFC?E
z1DWle*YGf$Sk9OZF^T*ApA0N%uMohZo_RXxzYHU_$x@lm9(|^ZN<!;|gcohq;D1s&
zi7u8<u+;OCP~j=F4RwWZf~kWc85XY@STo7kzyJWEi(`YPQD<s}vi6F|Q?DvXQNf5D
zih*n$@D@yY@4|`ud6vU<5mj^Tdxv^jzDrjt<g#*#o7uV>+egzw2Fg6Q0iumo!i<@4
zqho_mQMs_a>~1%4&*9;rssU@ya3%%#5qV!Ar)Vu=9WMBY^F=tP+Hu9-%#RsP4JPP(
zD!cF%5(r2-c#A#^R}H-uD;#zh;$HV^(}=b>UwMw<R7Rlc@RIcHNBrUOFGj{hj=J>_
zD&`zJvta*NgPi17oaq4kPgK%mT7xdn=B#P2O|4H$^@<_8i7G6Nw~qxAEUxh`?N^xR
znjI(IdAU@Pr<MYP>UwXCBwQ~d<}NPnb5E!2gBHF=ov>W=(0_J49aAZ*u!7f#{iiCH
zpS9FMUjDS>V(j_nqs*Z`&OB&{PZWu`eIi=TzvAFl?j>lpYZnF>C@TaF?j!k#=4mNN
z5%TvW!B46zuo%Nd#Vk{K;~^i@P=PGStGXKs)<T+Xzmv>R0q5PmTQ6VV!%rDOs|H(6
zv_KSG;OV-3?9I&0S=oz8;N9g^%h#gyVcET3WBXa6XEiEX7t_FHr}Sy>Q&&3)#>)X=
zi;MMP!llj44#0yY<{-batd{p>QZHT~CbJ?z6v0iMtvU97-SzS_(_x^_&)Fvv9k~g$
z$$k@^<8vk{iYm-ZYJ8-^t&L@iXrI*zH+_BI9fl&>QPbj$_c}e|O*o>WaG_9iYyb<I
z{ieN|d1-f^eFMB7Lc!jxBPs(AF2PRVO_w&@*WL!X0z2EoGC>z>iGg)><JMT^HtL7W
zpNj3qDm)I3O(62x#M<&aFP^9NBu0H=kreTNYwP!D>4`i|2+FcL$fK^UW^|kmOTL+P
zH1$h@M0j*PU<=uKQU3H|z=sRdeKU0PWJgcVh4MBVPyW1BB*ixanuoeVJy3X`qxGe#
zRt!md0|5WUeP(l0+IB`P_Zbr;#A9>=^)>dLwST$n+FU2l$$NBoxSf8g#KKi~tIfdB
zg^exJ64i*4*s-^>W0f+(XbA8>V1xatvp>aenD<6Q>`?=pErBPRNdq-8E2kgqS^N(b
z+eXpo&~Ka$qSG%Qk+?H9f6U3oMVWpc;CL^Y31s$u`O}NGBo15tiJ4@0YfN`!*#(80
zjfoYRnP8mEGj*L@PU`Ug{#^77{^i8jG>^}{o`cwrgR>NtO%07*PR)!bTWrY278H*y
zgf@goU9K?XW{=ns$~`Vz0Ku~aC*dVssQ7s#FdX{~Q92JNsm3Gtd35c&f8r)2cD7(Z
zaANDHiiTR8xH<g|9%EXS{*R3~CJx3I2!&m=HJiMsE`&J8l1;fh$UBp|y0OPHP+O?r
z{Eke!%Qd`wFw^C#uyUfI9PCsH#ZF4l7QdtYzlY7V8m<mb&7F<Z#jytEN<2q+anPvO
z>DY0ZLjO6)W+&vwPR{oaJwDk2ej9*&F{9QJ_d^m6{w^AQ{8ILv)!IZ)yH$g+m7hoc
z-bx(U?6JoEJV&hZI+Q3M|6U00L&=X0lX?@2bKX9T1ApFy4)TZU(Z!L!#wIiV4LIDD
zipyvMSKu8KmLoB5?1Hxc{-y$Zlf8k*Pl1KfVy;iC;&<WUmM;a~tK&7s!+~+KVydTn
zd{2jx^I(H>0oUi`R1cPfY|XR>G?``pi<LOVXHMPUs;9t{U^)abBdWy@{A?#$sGgd2
zcrBG%_<uI&`(^pSuUfU|f5bts#%sSii>IJhR4rtHZpbCql-8I(2E_@fCoIjNRfDcr
zD9)DVgOP~jkk|(+xDZYQ_<bHfcbQzAd%5^IjQkl?4$+<#`c2L2?`{p%aJF|}uNw*_
z0miS>6BlI>FZb*9i2s~BAHM$f6dV^!&z)f*QV6F1p{&Ykr97@Esehi&XBDG;Sr^49
zM1Qs<Y3Q;qitXWMn*C0DqKaA1|6<f1f4K8^Dy+?<9AP+}>1_zC+~<mWa((O~b3H42
zi9yu_Uwz{OE@J6}){}K+QNJF<Dr+Q6dWuW7{UE;!X|_b<d&J$;u$qcCHby7;u(rFt
zcZ|x5Jn>UD)LfotCWlgzVnP$k%9Iy65ciP}iuqzV5UzjH4fG#yQ3zRKN!ehoBv*Hc
zI=CR(sT0+6!mWU)van{m&H#zLZ(d0~DSv~fr~X7(5MVEzS~2_PnQ1JalR$kXKLdRI
zSEVqk*_bA$j>%(7+J|SS>Bn_i()egK!d&I!HnE9y{q6niqlxo|qc?n@h31YG-Z@%X
zfjYqlDsve5<()_1@Zw=|uH^CAri<;%?+#Ykv786W@dChd4Jsqn0RLvTwSzFMj_6{D
z&k2Wwvxm>O-y$4naT{j{t(P|9=~D{E>if91XF7g6f7;AMPEU7y-%F2}eXLe@@2flE
z_pM^tYLmO&(h^<sQX(c!5bgn92T`6S{ALaASDtLWr~v%k?|2`@5Fxfn{@1TAt?b<2
z3oo1BQ>q1Y8&JA$xCn{gS9+^?#YtC=j?ZrB9T$7;&<Gm7mc*dOH8Z)@U@*_30W*IX
zT9v1~wI1R5+>6QBKPV`zJ5Rsma}CQkV;U}@nA@-fb|(=e_F=(1XYk`6aFuu6LJG-t
zx7w-Tu9~5qw)$iwE?ariFBf|qzp#t<M2PPY!PU~dpd|Nh#KwE4#X(7gfa-s1UmCb$
za5+KO^cFtKsyF*`CH=P(RpKE`tDy;_)5Qc?zrW!y9hYD633_?-xLo7k*D%FJ!<@_I
z>r>#u5Tw#kbsisC@U~qdy`3xhnn&TY;)ePu=>O%w&CJa7oq1&`wx!Xn(iIQqO5azB
z?IE?c{OOpc8;6ER_lg5)w&}X>g+*nbpO31nKHr{oc+R9HeYw1Iee?lAkFu_gTi~AO
zMX@R3+F-<KRu$vFOn^VGuLS>p#XqlckJ3cxb>MQ!m#Fx^zdJU++*+mZQC1U}Vx#Gu
zihNxvIpyVj#+r4x{)K7>lqsM>JBZlAt55O+V=IonOa=k}{?O<A$+{6rg(>ua!nF%Q
zS@-7yLfiG<+$&)$N|1$G0kWh4&C@tK9M`gF^SZ3M&g~5PyY~&qDP@F&3N?7zIR!4Q
z)u<R$S<sj56YeSD#3$R2aSWKpMTsMwpyBo+eK;^Ez<p)wIiSIuo|8^E*uB!*=_m)8
zyD}O20E4v*zcC`V_Bs@tMm#=Jxjv@4{{B=NM1%%nE8nZr^Mg_%S#1;veLwpucaa&i
zKGmbL1B4%^duSpYDnKE=Gv0aiixtuXlU@XJE~eJ02&{)5=DgKMpmhvZpN7kB@jj8E
z*;gM1?{_H^fm|T~j|x@Uxv?~|<#aa`aA3MkE{$uMNU7-~eOmD9<|5XXMC^s)i!iN+
zMm-1lwRq!Yz9JpY@C3XBPF%?f&C>9q+h18#Zt#FP*arKFD#Qa<TK&nZ`-bj-t)8lY
z{IaAZvYbHxCn1#7eK2)~(C;&`J;#S&&$6Q#i$w8FwS6C9Pn0<*m0thMyg&O1s3J?5
z>u6r>o)XPRe>__wNaxD5N{vDZso3u;0)_{-tdX<(Ab($fYq4aMl{b9kN4{Z#Dh-Z;
zBH_{cHXGt(6x4`tnJzpYW(HB4FDT(_B5I++x(LSNa3LNm<lb@`8J;fO#<XL~bcp+;
zJD%hS|52xNpS2U3*6UcbUk2!+>DBO&!Jr5P7m;vq>z2BIYV)7IpIM{I!hyS%Am8uC
z6eENwSew`D$<SPD$Jwu?+;S`pN_gedC$35S_Rv<w_|&7&VXYyl=DMKcpDNTlKku!9
zFoTecLJCj8k+=qv+bp*DpRN2FHTur8$*Ntv*0j1r2KT;%G)3y_@&GT0#8#obZ=?~<
zBYiL)i6y;$qjV~D0k!h<!rS`gC5eJmW&^F2g(&F1R|9^{4?KFrgjot{(b~utd_@NL
zl+anm_Krn@;qx=UjHXFwC`ysGvFpw|+RCJ>IMdWZ-<%f^{+88H@;(a7Y8}oIM{9k;
zY_vEL6NKWPP!e(W2|%cu?>0rwsK#qC<aeQ?4C)qO(!nPL@+jH{kCdsi2@C&)pVL*T
zU#>Pbh9pm%pVui1a&yp`xA-luS4nMk7&f|h2m0(?cwbFlN8Wl{R61G7zj;9aG~m(@
z(2stz6LQ&RU_b!zkh#7aAD;7@1%m)RFyO=ijHbT{!SwY5ThmQ;_H+a8|Mq_rGhvU%
zEzZ6ktPWQZ+Htlbg8V6EF?s0f&Ip1wVq9bvJna*Y4F_u32*=)yRTgchf6_i*jHcI}
z?98s9)HNX$RNW*(H!q+h-#$H(FJHikTF)4hlRVf}XMO!KJtqh(u{Cp~N@MTH!|#T_
z{~i&s+Ia+kHlg;_#UgcFMBw=eIb`1>>=l$H5r-RkE@3@)PO!)obF*@Ah=EDe7Qz&&
z%v#RI)T=HCJD0Cr67DLnjm&p6<hWbaKMLn90_mm&oZv%b7{@H0%Bcp0lhoC?dD|oH
zeV6mCw0mP??2^$F+Pf8%>U1hZS-SUZ5saBI;JCZ=<W7#yp9q)0r4DaGdPY{j^`Jwz
zfTP#tgI6D5*rU1AkQ)q`juZGV@43M9H4N7_DDP`&u*H52W(Y=TQr?2R+7(kDjndgo
z87rb9zM+d#No9lNp<ZgCEZTi)m=8(s#)@q?%})5b1hjiBFN(W#kT*3r$kyDIUV<qB
z0V5vhNRbb%H~N_b<#A#$N`s)h?w!=}z2&Jd+si3-856!*^wRG$f<0AZM^e61cQIIF
z<JozZj2LH`HW`Y}W-A+0A#Cg%Bxc%(npQf6f3P=MB+4)Ki59S$)A<6D9oQp%>|;!s
z*nKW3H<%#fdb~UgC%*cUqQ~z>vd00SQx~Q&Ahw79hAkyiWY07*v$+#)P{#d(a`UN7
ztKlyx<<T6AijK}skHB`NN&F5~>ud~8qfOZB`pU0q`e_F%<qfz14H;2apTS<~QWWV-
zEQw}Qt&v|}DT*nJ!o`SLrK`jPj}HtDuxmZCbmCxi;iwBEu&HyU?1yW!ju-(ium2;-
zzMlJ9@Uuh$!ds*u3t7nC`+n|i$e3TJrw2;T?_^co_WQ>3yQAIbvM=^a6cu$(78Nw>
zR!e>DKB3J%j@ciKsh@xH3p_q|BB#$y{cRsnUmDV<VZfHpBB*g7N-*m60%>eXOf;4!
zyMJP9PnpD|M%!ZrY3i#_g9EZF8MKtGb<26bbNdwyu(J~wmJ;IPUovcZdwGc!sr#SY
zi7MKYALu>xv$MmSaM@aA+v_SO_8*Lu@a`OObc0XOv&`F!x7Y;ylDqs>+MhRg$Y|dr
z2!_IaFpqbI92+`3O?Zla_Vg$;%}xLQospAag_g*JNfD{Q^O^|UkwCBU^c-Q#7HChL
z@XZwu(|oKCckhmP7D-)(b%T?T^!WsL73HP=o=;@6uc+KCq;7$rJcuv$8<4)849NIw
zsnc=p<Rmq=Lf(!4UTbP->M^}D)(;ynmXsl|-0?Sv3jguCYMS*OeX5tZ%MZ)kfR)k@
zz(rKCW=^~7_TL2+a4cS#m85%oei;!F0S@U|&5E;0D@?Zf@W&GL@BFQozJZkK!@<E~
z7XORIuRmmra7yNe7=yM3yKr_RBO`&nnHG<4*4hokY_ZCl*-e0-<<8aVX-j#UY<;Vh
zRM}<{J56_%w^Wi)><*zh>yz<r7dHm&PK&B>ZaV)TX7?2MS}b$6<}SX*3X0jR;{m(L
zvHd@Xe>)_f8yfChO+D|V=z?^|^8bZkQXj!N-zTr;01Fd_79BwzN8h<}If;1nz1wC%
z;PjGr&}uFMjQxH<?1-g~Inkz!M7bevlPvOax*h*awvY%syDVM*u4(`GZ_79{=&F5j
znztX0(gKDIz7YM?r?tnCv;K`%f60x}nAHCY&5`t0-jZai>nnO}2$QC_qD6}cEhBh{
z^2*8*<4X9{X}Gmf1yFIVItHR8zK7^uH%izm{g;|Ak=AVCwPlX_#X@_8qV$hBES&(5
z3^t1y)0eOH^|LkYe{Bl(b(32>)+nj*q^VP;1MY?!_5as_$<EHo&IkU7k&mxp0s+wK
z9o=E+zrDOoDddao9(jA6q13u<(+#smOB;VR32`aGlEixDEB#v#nM)2Cp{$8&5Q?}$
zBNCp6uN;0fM<@46OurQ+auZ3;4=0171vzOGU3s8IkAl%Px@6FNGp5ysvQ|k`WyvDI
zvFN3KX>A$w3#VP{+FbkDXc5&gf^5r9LR`?9T5{>`mYwNiTeIl<TPwX;8eG^L$`qKO
zf(H+#B{LP|!<^!W#|rDjOURUrUo{Bk&!&rkzW5i9)t9opihpNiXB!hbn%kS(y=`{}
zf4;37-Ej8AR@y&eq99+49Ut&lhr;^-sQXSR)$55z|Kuoc*gSWZW<x#YfBM!LtCAHJ
z%yA624*2iCD|M~QFaI2YoioGh^IS%Ae+-+Lq6q$*n(iN8=C7b7B{M6_YhKXjeGZH*
zG~54rW^9Z-t!ItR_NJnU=@Q6%4*IJXuc~(6xE|?tGWsRzWZPWF<Rt3UPS#!xuwU+~
zxxsY@H9Y9FLs^zo_iiO|IMBgs2cc9Kax-!_m`yy`Ftk9h?;H}$0-?_4-?V~uFG9hD
z(=n#tezA!uC7oW<Nw|o_6UK*GLOc<=W!#RK4sZFJ27kgV&5aG*n4g=zMBXOqfnxW2
zCJFwVv2J)uKXdaGL^-HJB<2G8dG3W0E#q2akB(H8Lk~=Nc#1}+B*}5qz4Iu~XTdih
z#PjV%(IEF&l_8}R`#6|wDSWS_a)Mj-gCI>P>qA3HPEpvc_A#OoR1#$(O&^6V-hZG#
z%<*t?vbAGKl4@J(DqU;G)K~16>-yro_}%`c+{G;6KRa0glk#F${cb=Z8?;C@2xL2R
z+voahe4jV(8riPr4{+U>TK-_#F;oQ{M8s~UEs3$D<nQgdSy#ZU0^AGcc5eS@A9uKY
zK5N%j^Z9dP+Qb)fmhGLt4Au~B9#E>yezUud&-C?c8UN#oF$tJ6%GhMaX|s%<EX~F2
zzfhU|QkCaTjj^}8UnBv&ZW+A}waxaYxJ!S7`>YAKV#0}aA@M}nC*k`Iu|r>_A4kQ@
zdL915c-(4RQl8p8`FxT#cYe8ZVHn+eI<L?b|Hbk@8v#nl6hr90kpV;^z02icaT|SF
z+bjWG*I!T1UEQ{>(48{c2`E9{BOaYPoyU4Y>-mt#Y=Z_JKWD&s1&0M?KG!<Mv%l)v
z({x`?K1&-Dd%@-kN0+AO;ryv&NZtj}enW_3&?9>>rQqS|?C=w2|K~x+&(6LY@UEa(
z)`H+q2^K96g%C*v6?BFMi(-UkAeUngM<l;B^RM?GH*v(+s5yd~siF?iqKqdA|DMjc
zc?nih@hswiu7l(nRkkOt4fluSp(VP-fAuq3=&3u)XvyfAjpku=f^Cd0;<2#soKD&P
z9VOa|T3v+ve+BpQCmja#*<S4jIm_8>N}k&2*{ONhkNha${*RgXS?f(ac#`N%Hdl;H
z8J)nxxrKOKb^AY!Qi(VlS}sgx;Ss?|R@Ze^$>|HA{fSY2gYOq&w16p>CV4!WjU<UJ
zhOkCx^Sa8$W^Wa>;06zW9{wD>5A3_V4mtw!3XWaula$Tsx$JE570-0YLMO|9A)+;^
z3wcMCZLG(Y5>VxKct*;zWbmH}tH&NImX8Mf>DMWGYo3|C6@pIm;D1V{lvDMj5=*#v
zGXj{#vo)v^TjAGTsbg0{@#3yTL@)~b04sN?_lsNqQDoXV|5w1|W8HBeN^<e>;kg<g
zi`r}D5D(MGJP`>6-0nC<=7)NE<NY6xnTSSB<24b~N_&_7n?Kz8(&h3LIO9g<&bH>#
zGjft>P*0o*?JW(T1(_*hZ}~cL53$TV-XQ>tjS{hCe^)vlh;VM8EPgk4UU#`?6#94W
zHW&=LWC16HqZ6*$Mrc=Z{?TP{4#}$LZ2}Fn*IzkC)PzT$&KCJrm&fTrp)qe^HB^*M
zj2u^9w$Bot0Jmf+rVS7tBKyJ~XQQ7IZaMXLtu_ym)H_cszgSWE*tY(CAu1U?HU<0~
zVEDn9A{ENb!mR{j#Nacylfo9*VmNNT*(`gNGEt=A!<M7l$B-tpqSmW8^H<}T?1lbO
zw@IJOqrO6im5v%w*Y2<b8C3As5Z6bT=HGAqb|{v^VXM;CJre$hzZR~ZUhl69yF7oy
z^i7L-$?PUEJ%1K(agbZn=kuXt0h`P#>%TDh8YZ+z^G;{k*LUM*DgEcmCr<SSM`1@t
zwX!qPvi{2hUpcA2R#T<PGLp~%9-Do=|7b&CDiw3H&HFlm7rXi7NaUn}izLY^7EK)R
zF+01(efd3Ah>A{z?WMDezegC6HdOb$X_Mi3ShpUyp*lUFkGgyyFTGSDECUKN9*V2<
z!PuT6_&HaJ6Thm73QIokM7r7T5K7n6C#FxH&NZLzhTMQGXkE?bUQhwoixx&y(Q693
zIPp#DU2E9Ex4~P?wl!PR-qo{=*GsZz-uT};0jr&ruZFm-Q7KCL1`+dPJ}-Ex%BX2t
zh&gR_`V!yaj8}IE9p1b1;FCrXTJrK<v9U3o0(4ah)7$#fIF@z*461W46!}j<n`Gnr
z)MW$bz|@wY`7&!%l>>mpzEEgrKyzTwzAlk3aN_O*Vi>dzJ00C8mYEJ=<~zW0OI645
zz+=04Mk|OwUj)Rg7iY3bncpLUdG!RNcwbstX12d}v%LIP-2p2I3HrC@mToikt0Ixo
zPde8Q^!2;cn_Rx8r`VJGs3DWDE$pg0z+MGU;9l(Z?^7(#2FjAQ->y3JEZOmnF{Jmw
zNZ`rYZt1ZH&G}{dt0Dc0{@KwbiF;Q4^*=(G2D*HQiV_?!lhf<<8B57}Vb1i_OgbKi
zz8BNVA16V_561_e!q}4{FS(?`@pQ9~A1xNuc*Vb$2f2(aLpoyH@~75&e`zoj{Uj!R
zI(PQ{9E%$OfYY{4e!;@^!nZa*|Cy`55->k<ZtK>hiz%Enh`WQGd@Wy1lMlxXgx+}5
z$W_M|27B+`hJW-@S2?22q~0X!Y$5m?wTFT6ix+Ziugc3RJeI>?I-KH|Zl=3Z`H|YZ
zn2*--qK%EoMwY6uQmRLg8s+O7i_%x}AA}-N<Tk454Zu{CpFdLAF1GBDn}&6d>~1CR
zXub|p#cbXF2v(Bn4y*6ruY^b07_gFDpa0whoSmh7SS^z}nIf|Sa4k(q1&!jiBENEk
z(Raz<Y`lh^lP+IewI`;MO}Nx*y=_mxA*}a~S>VYRZ;RJ&;?#<>c%(jskZp+AG6Wl~
zc0AY%1lS`JpJ{OF+F@HO7gayjGSj`F6p$7+G+`4;OtKQ7d|{rF_{YQ)L2)ljX)vml
zS{=z@E6?+w$wXsv;`kFB<+B_9Y%1Nmk{5BGtz@7Caq~~FdmJU{d(nH80c%ue@{By}
zks7yY-Q=|w8{~O>Y2m9v@6SX&p(I%=bW~g53_x)3b3*8i4IQy>=JO#3BJaZt`=+^9
zh;FmnqsfA&?i3g(1gA+VJ$?GkJ{W(LJ+V1_dFIjgN>B(|Jv#TR!<)XqGi8~%HRZ>F
z`|#p6dDC{sVF$f;tEn$`3XA(Qm~(X0AbVwTHFP&u+zoIF6UG&ymYSdU7u(qz8oKZn
zB>|jCp_BR#RQ9ST$4yjFpRYC^lgH<Z`ThO(y-7<~vjN{m?`3Ko^@W*{oJ21z+h7Hi
z${re=_m268<sqoGJDl+t_HLWD*Xo>hw(EOiZipV^8>;Xi=2>!gwDUSy9N7N4T%YJE
zk*(Y|!Fb=)#23`&g8>=Wb^JeotGUY)G@y5%Ih;ChNRqz74XxYvu_0A=!>lBWWg>*r
zfmX@$-|l}3j5G$0*vK!+XtzfU)=D}?rrPjWeI?`ALkx})`;2{i&U+2i`C{#&tymJc
zi>OOIrhHvc_&$O6w#MuVv}#4PO9A20Hy{1Y$nEqy+E|L04LVr)-{>V0Ld!@G42Gr&
z-I;%7{7B(Fcdbbq%!pHb)uC?O+0%KpRSib)JO49&{f8bn+=$;_aHW0Aqhy7`aG!nc
z`(;!-w{w>l)o|B(X<Sc_p5JT)QKZhkvt{=AowMyu2zG1R(jQqpySa0S5t;r)0LN*j
z8D>216%^OG^w+wX%3jSY^hYxSPAcLXIKOUN+OKA+jB4fXUE2Df6+x)-3zTNi^Ucs-
zoiEr{1{j9=hPxy7L$ZeHeW^qp;{VF$rqyZ<kX8yQq&`|aH3Tve>kiJrDeZhjvK`$w
ze8iMkpH2B}MrJN+Zf2Y3X5@xUbEns&<V!Z~E(2nrA-m!}u5VlzJFi9HSK`q5$&c)Y
z+LEQcbtW-#vNi4x(_5vdEng)bar)4Q3Ly+rPUNzz&6Gvn8~st#<u(0-`~sqz>Li~v
zAg`~&Nb<P-BA61OGOfF>lQ(PLQFh#5!=ILMqg0rF<C;l&f)#eTl_)t4aJusKHB+uU
ziR_~Tz7Wp&iqr57DHs*tMk|@pJkPVIq@yfkJkwC$*f??jo#|5S1kzBe^Mo+4xcJ4N
zRl!ovT5qz7a4dI+MY6W`y36i=Jk@I|)6J%^(h+e=qEHS7ZlZWNiB_-}W}`J&MwgRN
zGXE9=;f2!6rQa!A!=Dh^FYB6Ki@RG+?spHH2V4)G_Sw+ct^v<7Orl^6FE|H$Kd1`O
znfVkB86Zeap2@BLXis%ZYWNm*BA>t)djm$C17|X4b3^GeK2$Q+c^_L*S63q+Fw$y+
zj*?Sh9L=r~RIXuOS(6rE{pObR=`v^-A19E*S(*@fx4?WpoPr-7&Z%ib2mi><%4E&W
z;aO^|$b-SRBmzT@;7t+ZLlME*<9m%&x=hS5m=O0uBc6g_ZP!=g{0_{;wmQm!Lt}#+
zj^KysgZhti5Sr@0cPPoFRmmrtMRgt`m+=rHmuUXk-PYZGMtl>hv|6M}IQ<<@@3nQa
z`*P^yb-Ta)JqjX)%~r?$R{~>Os|Nb<IbuNzJ*Nu-z;hWszQM;uAbq0JqBBx8`*z3n
z%Fmy#+Ge-C+WHvC7y6En4&?_k9v0HxD;nndQi!VA_V0@9;(j~rIwlU7clU1cEdMjQ
z3k3@o<)DB2RB;a^TfKL#+WXri&H34tjWMIn^(Mk~2fE7OFqk_2(Ic)FS{+(jtR3^W
ztb>*oK)mQDczYw}@uY_WOf`NxD#Wn`+{&jG;V^hXyqZk-PtWX6VCpKOw6#*21T!Q@
zg%ad2eG2)hBq(HKjAXQiNm*Bvm4UJ<sq5V@RIBuM$$ZvX_pwA>3`-T@XhIVtWtoHF
zx)9P;4u<vp{V|XqO*H@Kd-Auq2i9gYpMg3@vkBED1ic9o3cIx)n1?3o+o)wnD7i^V
z3I{H&yrdzfQGi*1p;uK`1BMns>qW8HmzXxSv0qIt*DCZ+y(O|21d7o-`dV7U%o48i
zDQOlfS;Bp2T~$>b%1ZOgHz$?0YQ+cLBcv)&ai850Jf4o7ku{g$e&#<Lzr(7}z}fx^
zzOm_Juay<=CY8Ig(G;vqnWUTNuSI$$v(`qzeGR9L*2;~AygAlG)9gm_0nA5T8-kxc
z(H<BU&YY+UCsG@G1EYCg05f~!u<~NW#26b0#rwWT>%^dpS(!;TpUAU|pWr4<whmj7
zn^dAZvrif)Hv-7blq%TgGfOD}sP)~t$LDd6hVU~pJ^T9wY|Son^1K|pob-c?hWGA?
z6>sK0i*%<>VB85iLc?%k^n&qL{MJ1W-A5fn^{6R&Hg$y`LG6opi#QFRS~LTfJu^l$
z>LH*^q`j`MIZ+dD#7bQH>KOyt#G1|RwSLnMsOCG)DmH!;u1a>fzxoT1uLY%);4bKD
zsp&6pW^wJ5VbfqhkofLko-$;2-qno>uWdWo1|RRh>)EEa*vQL;{r!QKTWQ;~=GQA#
z*DHD!LC42^`14g9)&9&Ib4st!iQe;hEcx@+dOul(GEyj*jY5P)yrZkQ1A(zwy5c8V
zMNPypudvQD?UxG;z(i38IT%cx99o=(h|JN3goYDq_t`=P6V?3PGq@5@B*boY`>%!w
z@GsT>u6((k^+CMB_V>;<w?45?cf>)eo81#mhqlR0q`9vP>}^tG5t@TP@g#>M^hhY?
zhMW@`mPDOIAMs2yDamyui#*gI)6od!-B(e%;TNsS&ZzSUSmabfC^5%yi{ilurF6D@
zf;DvBiEUm@o0G&|0}6@-;Q8#grJOxK1y|Sd)m2a{l{Vmg>7kJP?#Kpzq+zAS@0rin
zgTliByLH~R292Bf(v}@nTR)z52CS`xET<?z<w7KG16Q-1|7PySt0a9s3uC3}HE08=
z8=!M%xeU-LtJt5{myC7BR?6(JaS(O+4l@upgKKyV4UKS!tued+X659Bop6k0dW5Qz
zuL8sD0(VDrtQWS!JlT`?*3<c7Jk5=)H_AdGI-KL2q`NTUe6(=4*ygd{R*b_^KvIL<
z%8LD}_s~-rkxrz8opZjhjG&;*s6&IwK)-q#z`y1D=BvbWV)JZC|H?$$^W!b=D{Gq7
zlIHlrM;yLuC1W}RkzW>6V^cq^M;xpsbGCS#INQJFgTN4>w=nW0f@IVo3Dn4<(Km<Z
zftqU25D!qQdI|0mS#vWoTCe9>^3i${e%xG~KJn1**q~ctoBG%aPkd{7`fo|^o!Bwp
z4~E;s!XiHT?a98o4rRkQE@R$&&>B%9N=D%Z=Zv*WvU~iP%Dml`s9U&?+ZrvXu)ZA6
z4;%~U>P)?Lak)iJH1C=aHD8D}#;YbBf}v-!zDr!FQEE!AA403yM{(?Kcty9y_VyD{
zpGl;ErKM4TO1obq9(B*(-59-lhy(71hG)R?Dkpv!eHf989v1rTd=p-hhoWAD!gh7F
zmtLkSsj|gbcgZnw=<+H+NaW*0wE4@*1n|6AB|NQ?v7r(TSFBwojV@&?A(C2G%I8Zs
z31^?lBOi#plWU(Y>wx>4GxMD+4mLFyN^ZB2pQos874&Tkd?<J(T~2&I6R?nibH1S0
z37#)2^Xe<BEjhS4>+|aIP5Npwp^q(L3m*x_M7w`u6F~T8x$~D1s`(GVRlhg;3v+9B
z$)oW}{;Yt&f`?$}^BY`r9tp7y{&%YtM1Q;EnFFR9YC8SPtj+ZjALidz2McZZ#An`?
z0x?-tF0&~|pD8w0{B_|31k-9bVNTfD6^2h-pQf#s*flHL_eW#P)1qcyez(FLP9V#5
z#xDArg1mi)35G!z$qK*gXLY7Rz0ox0rmk^BOfD)K6uT_#^r{-w^P8633UXCpO$={b
znL?Gcpg#7<cXZUu3GJrWM^prfc3^#QBYT)N<K-gc2*xe3lnw)+Z`uc^wru=(ee^2s
zS1SvIbTeWCAqs3hy0n*8NTj=!-&?Csyxx#@!8i{hxkB30YS!Y?iB~kpf!8^!!TJ2^
z2*DWZTOb|<wC^$IhikJ{{C1=dH!uN1l}jF%7s$S~`Ryt;cyN=M$;o60O>OYUc1DA7
zm8i>K_Ga}B1qcu;nxXe!ef~U2F<q7*f3;=Djg>x$2P_nv?avuRnFx%w(#<etwukk<
z-&Q$ommuy|MqO?l(BlG*PU&qpTVBNd*)RI>G~iquS6aqrG}solU|GF0@`bacrlMwc
zuEu~>#(nK8y~V{-FdcKw@1WS8aBa+_E_c0fdMhgaL3sl(OK5#AgO6d?(VtjcOwzr6
zigY<c{`Sp2#WoNesO;MhI-XUR_48^f-BU|sSNg!NWE%FsxOO@O{`*0f8u<fF^ILNN
zuN`PwV58W7PNFu#7#u)Bd}Cc?gvF}K>H84g5}e}ba+a&USE_xlgqNB-B?$lB%Qlc<
zF1rRoDi78JB<**+tvhyJ57!07W6Ru+!w2O)6TyddiRt>mg?!}$DJlNxq8BGax5<~b
zOqXUov7QCD;qoZLd9)4>Hkoh8!+OjaXm+0m)zL4<UsvUFswH`yU9x-|Q4>~pByXJh
z%;~)Mu2wj#UJxga+)Ct7c{_fQ2^_e5J_9o6RFg#d*ot`HTZumZ5M5{GgjJ$7*(@sy
zhuPz@h=Rem*8|jUGMH-jFO2IBi`1L)Mh7NHNf`q9AQ56|ZH<}U^4lq2rw#sJ!YUYB
z=drZcJiW#L<O$Zr*=yhI=0cWrobc9E)JP-NJ4Dq>LfQ^@5a8kApX!xvo8DjmAZJC6
zzJ8ddlHTwCeKMz>?EEecQ&sk!1fKZQPptFj_gWAQ`gJYLUaK&i4Nelg{?X!Map)Th
z@K%7qtTV{N*Vk8qN99}+p@vxxIcN2`Y_Nb)6b?tf)z`nLR65?yprvQa?_we5I#*M5
zx$-<{KfX%9(%8V_{LJNQEhSe9xLkfZ_t^g-PiX`s$#&c*uHrVrdS8U5BuMtbo%3i^
zS3h5Lz+1SHyujLtsx7mVaJse)`aP7PAn|5c1HpG^>eKq_2Va-UU(=;OK0Qwxe8LCI
z$4;-^_%<S<|H+eUm<bse-K&Fu2S*2jO{>_Bcr6>EUI0Nea5?JZak*Usw0yDE5x@Tt
zBPyXTqWq&$M{UtMD~{{$Wuc?^1Fxvr5lxk)2Q<9T#>i}R`}wL<nO_y$a20-HHch!2
zB`P*Hqdu`&_1S&<@cyj_Mcm3eg}TIv7<DvB94e%ML(%jlchAMf)WVm8;yJ+9!v#DP
za-qQ;AEFt$;M{g(cGEMNfP#gV*R;GAV@FF%=9@pFjo6p9w46tNv)%+>^}Z>0VFtd{
z+3hrov$Nh-){DRPoU{dR2k-{Zl9m`?>~fGi<3JUN8tq3?MJmW>)Ci2NHVPKZoJ721
zL=#!U_D@?SF1pd_Cdug!4?|X|*251}Rh|DL$dg5aZ@!VfL5(Oq00P`nT4&#Mthb_9
zGVLl}@cnRmr}V(As;T0(fpd7F&D3|XzgtfBG5cMKn2?chw_@QEr0%G2V+9(W8*w)R
z*3WH!|6OH+0SwNTB=Ts1;q3A?G$hf<*!s_9w)rE)gmImkY5`78)=(jZ1ehWmtg4~i
z^9t_-$#{%(VTywk01Y>$9yfp|zt1MNIZeB0Q!gniyOU)tx7r~P^HC{Q)tL4kb<M%`
z!Sz2`RXQZOA<s(`eUposR+0wdS<YXs*7B6oHH<JAyntV~V>%jr5!L9`sEoPX0j|#l
zBQYxRuRr@x$PY5zi9jB!ETuHuz-w|uiM;$!;CSC3L^(+U-p>9z$(&#mEXZ?Hqkxv`
zgcu__s+;%O)+f$ccJn{`FX@qJBN&T{V6b#+k|O`hZr33Akc2Nmy|%t_%T>(G*f{;A
zd65HIV(csByYL&X=93ONL;pyo7^WM?*AGD=()B6pCAGly#N5T!3o6-n7ysN8sW~pl
zU-<nmVix%1U_z^u#EEY_KgUeI_TSRF`e$K;<k+SEyc_X+E(q!3z4H^3TfS)m9o_QE
z_D}`Yxy{XD<>#BGnV>)h46yj<`+lR<9e30^`^kZKm5!dSuT4_8Tw%Pi&PJW*Ctt9K
zy}hJzG|twT4C;^(32QHJ=&NJR3A{R`AI|Ze1I~qN>7%J(5k}^w>UD&M$7!+njJ!(Z
zw5XgSj<+bYOi&p(9E~E!-;ue#sOkXsI@Dt`GrJ=TUl65rJDocn-j(0SZ3GxVgcu#@
z2RInT@>imiO_j{&TKzl5*TE|foX>jCNrQy%xvR?)vM&p4Eh-<a6u!#|D)ZEQ!~=r~
zb`s6p(3p$BC_^e%8D<4GxtQ>NWu*1H)FNLl^*RwX8jOv;K8i$_V;us~WwJ&m30s*=
zZF;ZJDY!0mIIWtAW*Mr{WqrV%!U9|sgS!wnc^<mLh*2aIn!U=QcWHH>2t=-#xZDp(
zc+|)6?D=2G)mY1g%toh-vRIi$PrzK`6N9z+D_Q@e*xb)vcK&148+HfeHlifyyx3Tp
zQaa;Q&TP#jZ~qSYrpX^|v&-OO3OK%Xa4RhLdiOl{^Ff~izrtIH0%cB^VG<|&zgJ27
zioBFJr6L-i{1|F$YVsB>!t={zc!q0tG29TS`=?*=(4)Swg-iXBlI!@aoi|`kx+71T
zUp3dV6)-sd&f&ith*bh-GKNa#a2|{sZ3$RUzC~Y@;G3Z)kFZK&U{@;%F7obfC9`6i
zXH20NFJPC8XYL}kiOEYR4!!#(42$vpd#ejJ;@ha(&pBahIyEoT;x26SWAeE96LW2g
z!OFAZhu*EoC?$d$f5F&gN*M6d6za$D>iHJGU58TgVSAG?*ACU{hxJ0n_)jUDcMRG^
z-qfEEF~WNn^2nLnUDSH#pRC0VkoOhZ(o*a7(8|ND-)1=0Tf~Y^yH^yd9%wk~y>x_-
zd>4W%vS{_K&*a6_)Z;nTa!!oCcg?>Ag;E2L{y7Rc95-%V#Z`HU(?tV<Hzz_XiC>zd
zvs*Z*q~6vi20IiuO~aBoXSXP6RV|wI#qI7~F9S!zR27vxZ0|+POi;|$*5w7hI~Wft
zsWVl@(wBzcX;s^gj;FO`51R1`F^9chs7QMYr(1kZFF+rTx&Cs>hcvnQ{gCrE4|QHG
zQ`T@z`c$fWDIOs&4hcZoeebIK$$xxR5Gl6C(^F_#q3Q&dkLX{7+?7jlCo}O@X~O4B
zX!fJ*_VFDmdOAZYz6WCU)m}aeU4b8vQFMlrX9r)p1bu+R{>00U4l4<4`GEe{i@fR0
zl|iEk9BFE4sne#oow0#NaU=gVNLIK4H;7(}AMrzmoyQYcEC7zU{cl4QpeC8dU#UXu
z1DK-)7`OZ<FWPp1o&UWBsJ`)9M=J=?;^x!9m&tzTTTQaJj>nm(@BL?*(pXtg*WGAX
zWp-R?Zjx11-qUz=MR2ZvA|}~C9wEOoL%x%!!_sxC%1&`7Mp4e*y+L@2Ci2~*dv1b+
z1JJXAfPUr&3XytUO$49I82en73akjuKS^SU)n|>G)@~Iy6qNb~9$ko*h#Wy%8)}*f
zlFrd6W0xn8^gC~7E@E9S{O<Scsj#O1Jc`hIs?xRG@D-M=Cc>t2C%C;@EY}Xg0Q<Us
z!QS$}Bzy3E7&?lQ^*QzLh->#ex03LVH+xP=djmLsBmBu~N<f+Q@0%*7EjrLB>NxG_
z0!+5hK{@L1yk{k$BPN_NUVIx4TjuJV49yNt)QZAU`GO7##6(@>Pto7cZg{;ky{E5#
zL}&Q!8$LM@X^QT7=J_~$OijPZ38lyl>!>41OeC%0d>A2jXrdl7aGrNi&;N+fFgpE9
zbOLxMJJi~;4(V@fz%?#k{adZNx=3NY{CD+D^YifskPO-B;^Z66O>sd<6sN#OtyY9a
z)G|l-SUR=HGCNngv>z|A5-tugpW~RZXa!y#QfeXBj}M^;;*9+JY^r3#UJ{jF+cOVh
zkHg{>X}Z2b1#jA8L_bkbFVva3xul|erR4sPrn3%f`hEZQBu9K?gfc>5Fp!1;(#@np
zN~8qoknS8QARyhPbPW|4qeE!~1Y}5~bW2M=@9%RQKmO+6j~(N_ulu@Q*Lltw8nY>s
zIf%?Iv4##8hFlK}Mh|cO5~q-PEE#gLc|$SZ<X_(?GaOy(5DEDXjW~whd+9XtLIKiq
z%{}*ob%H1Jr7!^;?CVL~7X0^@US^~)0V{z8s(~T1TA08Z#l!JUV;DzhO?!T(+}=G8
zg!5=KGmf;BNrdoM#5Tk|K0EIWb(`7=OfR*46B?x@fhEZS3<}hp_HC~03sur(*u?1Q
z;zB+kp$g{cf#UJk@RY@FL7gcJL$)hVPZJ~=c65j9T%ga@W*XAr4I%&r`;+<g>Q_v)
zVaX^yEh(xPr=UgvM9kzqX<%J;ufOK#<|{u2cJt~^BABihZlW`f_dgeIz*Tp~>INHd
z1<w}uMgU$%QsKsEe&ALq?Odabn+3)phYX0#h8%slxw(N_NnBay<C<sg0a|rhO7q*>
z^A5m^9u2XKcb+qL($1HdYTTnh{ri^sM4R3UW$QLVXXz^SnEPWS<o~O_zmoBP50J#o
zw$#Q}@C>_PCI9Do6ZhQy$yuszU;t|o*39=*ixXY+D^B%&Ea~rIXZIu$un>R3yhrpw
zhwB{YZ2kQ&hSr9kXys+S@B2O;{UKrZBtTM9vi?35%O&()m$)cC%GUpJxOS<qoGj}0
z_b|X5+G*#xTb(2ulRA!yD|}T^{SH)v0SUA^!aifX4Gdz#LQv1AFe>`s-+y}J;HiK%
zdblv?$h}iG>lBlwdlP~x)BYamG}kyh%-x!60(jR|b6TKphn-di{W@epWn=7(G5g6o
z?4W**$R88&xhw|`>(32eId1xBpiHK1j4b{ll$dpOpMFeWY{S+%zx==qFeL)|$4z#G
zQ%Gro|L%s$GgIouAXqke6m9L=DYQ7)s3M$Mb=<(3X1)!;GM4w8--g*=Ku!BIdNVKA
z-=IMYV5!}wv~){7+_w*0C|PLVlVntI{c(CPOL6i=Sj5ETv9^qnVed?$;8>^UC3;LY
zWcy2WBBSZHcH;p}u^Dgg*9i~Y-nKApeu1jN#DUR)cA@z3Wg^Pdd1))NU6g<ub#`6`
zBLZC-1b2vbvx4Nvb5`fwI6&Pt&&<r)J>&=cj&Zz8zeX+oHTM_J|N2F*S*DufzWv68
z62^Ql!kgm(R~Q@;mc0^hBb$1ZQ|8`HpMfU)s<NE)+#(Ws508aRA?j_`c1s^L62dG#
zCK+Oir13p?!~5S`7Birlqn^-4m8Z5in0YCH;g#t*Bky1L@I2i$)q7B&+ns=}9`|Z#
ztE+1d!7?6v=jFwY9we|=(mXSZE0N@UvuMs~GIUJdptC@z;B_UME~q=i^-_*GOaa7V
zvtSAiW44SBEvfryGS?85BskKb|1McivDc3=$Yfd%U7P|KZZeC757^jIujRxUTN>|o
zB?p_(ZZ`e*(XApi<kK`8%(i{Kh{LiZ1ohVPT$l9zJT7%4!K6P?I%xUFP(M>2K<2Nj
z*=d<m1uO!&b)t+VMictp@j<`m`?-k?Q+b?n1~J3aE@rK)&jggpBsxR(0F_yXP`(ze
z;ibdkm&MT$USZApc1#Eg*!~;8T}-x8?l_wGk|^7eM6s?Q={>sUzu16wrKcFjhc86^
zzMlPEOtFBv{)d(nr5SHMxIOw8kyv-`Yz9}gj8E1rb@&P=iV$~l&Ve7aACIc@2Z;-j
zP^tB}WJm+MwZ1M(kI9MJU%9edh-fOR0U~$0_}{>A?3J<c_9-fW+-+m@5sP@Wy=sco
z&B^4aj>~Oe`G8tmN@v=8Qn-+?w)bBj!>6$^UZ2*5goOX#*)~Q71^{{}b<nyIA$9y+
z$hG|x*BwDj`|x1{q%W(ia>;)qm__2Y3xDO%Pw82Pu8WJy*-=_#49sr>;_mJea&wk%
z1#mc<J(U?)2~cmG7xOv$7>m96p1uQ@lx%s?<7JLJS(T*O7@XCujsO^+39?T2@a6AY
zla*8~e%U4UaryF*C24BDw%yHvfmiSfXD(31j>On;NT=+aUq)u;!D6#FF!y7PrdVhR
z5&D-^dhtQFX1m29hiBLlfKc|P+oaove*Pcm_X}DZHKAo$7r4@LF{pjlxscDxlA)kk
z?u|-f=@}d-BjMp6W0G-uX>oE2ogw`T{yvUh^js~Q;5oruz}$C^pSpDNx}<Qt%90=K
zBdu!XnkMS8#%ype3lSv*lBTYb0q>zgQW~sIzh!!7<WEwAh)Q@QDWVA{nCdxtkpW67
zKCZ1nZ9~2+?Iko~3HzZ%PP6&>d*FFG@+(izxHxX*#77QQszp=3BdYm3{$78`J#blc
z?rR;fo_?Cn%xV2choA>{;gnh1aruqut)**HsLprB^puq6R8Y1O2*6*b2`|w{hQB5f
z#t+@J0>QTXXl=L?=uV=G$Atq<MxQasN|NEBYHMqQu6#qTWJ3mT%TX7d)O8kOA*%yJ
z^Fc*vY4&EGzBgKWKa&p+%R89UXBKDV;iYNg#%gro#Gn*8Bu5sroL4ta;d5FHSdD~}
zBv~<@amN2sLWUa4<#H8-5yq@Wr3N85Q2fkTe11PnF&=%|72nOzPi|JBE#=p6x>kbv
z<*GPLsiQ(grTV=_j2T`OuOe|!^kV|9;&BH4_}qDpG=GLUW%RSXakll>RaHAJcz%&1
z@iCbWXR5DFLobf^9S-N+jy))&mV9(kRPkgoRA^HW>&zM(l@j?&7>)U_|3iZ%6ZiuV
zq%NVtg9A=Q=29woj!4_C0kEzTL;*!FUuGZtx;x^9s;{m=`(jr#N1Ez{S6L%I0KQVH
zttKT=S{aNYD>Q8nvbh<}Z*ev~aU#C|eT+Bg=&-klr{Dr(H`Jo0sqpabb}8gw8nwe_
zd0X9i9rr)g;!t|E=42IIvkn*^(7Kwo6Lm$uXB^QGq(5+VS<EkFXwV!+pzL{FOk2NT
zWdl}|Q=jQf*(9%D_(?swAB{d`*8hZU@c9>Zqk4R`bn*Uts+Y`e+SRP^;5?iHov1z>
zHw3lOh|%F%*jZOgdGTi@ag3L5l<B33Ce_qW8*FNzYiZXj*U%VIm-)@}tM8@;Pf68M
ztZh%lgD+1Gf_6;27t#!D@=@w3UaqeEnv-ZWFN*}=iagz!23Gak8yjuW7~Bhk)8El{
z7&Wc@fC+I!n$+75ot>h1AQrVrFA?XJJW1g9fkMYFGP{b)9Cd?v_2~+j%q1vW*%mzN
zLu2x_l$oh?4blN{$n3J!-`Bxb!F$VLH-}AJ><&dG%q(wmCB{&?p#xnpa^V1SafgU9
z`<u1Y`lh$#n-~gUY<WKF{H!AC-Cyoa&3HA=jT#X(HC1?4vol(lYph;)xa?WF-2%`v
z1KcbjpE@xZ@lD_J^K&bUg9u37+#WA4uX(3`J0CZJ918@V5VRb&GF}g{^t1{V!~R_!
zd~z@nD?>mG&774XfZe;vyp|ylKut3zB`%r(h|Q?rhZel)uWeZ<v%CA@Q%DYQxGx?I
z)^svX2X_Bu&cR5rF*F<gz2%sUc*3R1>#YJm^rkT6oMLDku4!PB=xW}{oo;L@y*rz6
z53i!=P1R>?yX@@DuZ50QzBz{sxgvtF`*<;O@4$>bJdH16!T0DO&@hrLl(-tpK_2_~
zOUb2?p`3DO#h!Rdh&an{yQU60>#G8u>``448fUZV$wPc6r*E&yL@Cd|;ig()%O4f{
z9~$0wajCt6AjH;;59skAYC_6rI9GSPOaP!giGhShF|R|dn7<;c*^MIw=nm9uLd7rX
z6_6j~{=;H4XpqXwaKig2zHr?kc;8faSU+@y7*HP1b)HU?)V3#p2UHM(kf?!h+gMe2
zq+FWrP}JHhz06X9Fkw|w(-zkT_eqJDVnIkp>S<T1w^8Xf>};jReH@5wNH?Z~*V&)}
zdfY(W{^(%O@t(IhB6e_t)^twT*(O)XHkQKKvaYwc7p~)NCXCXI;9+~W`Nf4^ZL*;_
zm?4HGmYEPj%^L9&0jaYL<rwrdYY^G76GR*@@Zt9lEC@gPKJvh3kg-I~@&p&a$yT&$
z-PH6Mm$4^VzLdZJM(G}Z=k?K`G7}1T%9;CZMb+_QbP;ZI4Z!&F#;*SkkO(hzHhb(J
z4Q-Q_zmfQcK@~rFNNafJv!szCjL;oe{SmCJ*r1EFeZR<cKlX=z(SzR?o=mm1kHowz
z_J-2A<8#JKE(8wF5~cqB%};1QxSqUwgR1jAzI?)f<>RIBh*iikeR2U1b%uuC;pji;
z48c{YfC=`Hc;NLv{KZx@S_eA^tc(iXq-I~;05!)IYO^>I5W&58F<paMZ~+Q~*8g3`
z2RDlcd;7qQ{suWvRjCM8S66RsZ51jH(xK^3WGd7u2f$#mUYkj{%cC`QZv*MW9zA0R
zWI=8r_b`{{WHx69&oI;~1r8L>3NcetQ<A>>^G(f?^a&Xm4-Y!;ZjH*laIIKenL|r6
zsySN)LG}%r*so&x0Fwm#6Fr<{84(Oav;^4Vg!}vs2=PFW<g~h_gG6P3>zW*)<9N1i
zsC8~;Lak#UqNY*_u@#^DE|mUHHnQE1sR1_xW$}Qu{reO5sq^gYcuW?xKksijblm2`
zt~+E`h&oFUT^u+L5@X~ly=-6X+iOKNI=8mJDxbWVI3BxOGr#fXe<XAIQ!iiYkzpy#
zd5k~9SDp|2q+u$%dHIA{yx&h1PXHhb*(3Ph$QnX<$L&qL;!wBa?X_>`@|FQG4+4F5
zR=9kh2vjFjMZ-3x!HMiFUZ*IDlBBezN{u6bV<D<^ihXa_b!iQ(K!tS4B?6g50+ewv
z1{TH7S}bZGLtz0F#`c_0{C^@39CvVQKO(CQT}S^5$p3vbXjfnbNcvu>Xmqp%=(9Jg
zc%-(Ph<<WjX#Y~*T$?`7f-hvZUFDb;^>$}`)LLq>$y+h=og#q<Jm;cAG^%5AYwrOe
zxGmT&*m<jK_BZC)XQq(L<a4c^ebz)7`pi{D0zQ;3TTF!8Q)2t==cLgpFr`o<LOVP~
zq;VAlB;ew@zweUxuIlg}__Ss?CKSn`a!m?T&~<{9=yOdg(yH3E^Jr#W{F~Sw9sOmk
z2=O?5$!Bf9Ln2RbLS9=L#~vzD#nXp5k4?LW;BuUyHAsJAY-bez%5fYQKhTX;NX4eH
z`=jQz#|JW=;mgZ`r%EPL(qr6Bm0<N1X=Tu7W(LfJ%m_!wYS5cUb)jvks->#0<Hs@1
zJu|-qC`n5(x^E+sa)4w)|4<+FV)M%io*wT5SoCrNLezCh!eiI1${T@>q_mU$x7_tR
zV%HxKNf`*M4Q#yn_vfN|N7NO~B;$vQM`=(M0$1x}@rw)3gPZV(!qU$lQ#$ZKAJc1x
z-ODGr&w!)8Ut9VZph%TVCu;-(^%CXQct4YD(CQC>1jQHZCxS@JHU($fw(|@Vd(sJn
zc<mQc_iL{{`o@?>BP)A@bG`0!g(&3M!UrX$cTQQ{aD(TKXGg*$i-ms6w?MN9&7CPL
z?Qybs*0;cPbCr3w+=>E?r~YW2lZ5wLh<P<;)98URVtu-8l``WUfPmO3Ur3OP1l_%3
z+Jd{MUIL4Y1Clcp&{7G#$XJ?{!JgGe2RDAkB@R@1%JEO$FWn`|wgzu@tQciv*jP2B
z{L-!PVGk_PG@5I33Np~{n?&U=w99v@4^%njbR{F1XGZGtfkK~zd3-2PHZagY!PqIb
zqSi^bq>I@y16TuT-ZcOBhc@J-m?`*bZYc!Vq$^2-{rWl3RRAkfO4rQyl|Y1_>!|F_
zM5#FBw4am$DVK8D830Ghv?p`bYj=Kj<~r}rJ{UK665+X_S+KN#nit3x_gIZV=g=+h
z{`jZ}jG}1oKHXiJJFxCJZmgm1TsyDRKk=V!F#nORGyD@z{|=!9qZFIP`AQgl%2cLZ
zf4EMgrNF=jbE8R#N5I}f6cK2dppEbIUuWlsj8Ov5)hJ2s-(#kuEiS4gx_O<W21=Sx
z(Rcn^dkWApIn3g%&x%j|?U-A>HRkj{p`4K)B9Yc;jORnj@I*D$`!anHXMK7M4&f&s
z-ksa!{M7b6@j<Jr;OsE*w{}$rO5xWg@3EpT{>e!xwOJx)JyqX}=Ui#zlx7xR^JX0z
z>M^Yy?VTY{GF_Dk7%j6!Lwt5XFh9|VcK+|DPey9%6ZE0t5Ne?a5U5Exj{yWKjn|qe
z%i5h<yU4|kY+p-mj2>umoqhIM>V~WODz)yK#V4!C>38>5bU(|W3byci<v^kmt18`)
z?u<?_-Z5Iu+*2mtX2@Sak9&RceoD<TNEDwu1dfeC!ZFE1S@7&fnL;WA!US_mv&@-J
zDg+7uKDIdG{826qZ_!^7QeAR_ND+aN_^FBVJck7%=t)h)rp2$FYut3F&v9RV_n*+g
zsb`y;_lYHrZ(GNLSJWfAuRkiD2M64)0BxgOhUyWo_}aF^E#ltW`%Zp;n#cJ0A2oR_
zYOp6@UV*)86UcQP(RPz<MJ`qQ^GA{g9f8-VZsK!gn@;R6)p~VNm)m{c&cNfowK>sG
z>qaaw$t5D$Y47MtQc?$3{~Z$h_+U$^zO9vJnulBfYh8K2@8y)K9ls&!Yx3DM{xm%d
z@XhS^d|DlH9vWN>bAT*4|8InRzQ~Z+Z+^9Hm^%g~i5RqINo#QCj;x4qRRkvW4k;S%
zNW(CpoB5dkNPx{Ze&&8l!73#^7iue{XsWlF6YOF1!mZn8J3x+O4KH$0892=~X-1GT
zPfpC?S{mgw6}Dua9WaY+C~S!h23DTl>x%v{tf5?9JR$A2`-|aI=jFVRHZX0ug4_#>
z1zW~^U%V=Fp5PfiX3TA@szMo+pdo)M%L^oe*U?&{S62r&d$gVZdV$`UfG4<KVD4px
z9z_gpRG-DJ{haUZKeX%Z+0Li-?YYKWJIaOVgM9zZFM0|?R{czu|B4gYCnuVHfH$y{
zlaq@}Q){av<9()(f4|ey(`B#fWhYf1=aGF(K)eFhg2SSja&3YaB2bC&(h}Qvl{Zwe
z8bm4bZ%M<vIs%q|*l*Aaf$?m~2GTg*v8-jO>Bu*Eu1s9^T^_FVM}KSXzHSp%8~Pd5
zk`-SCl%XjNa|TLS?`&YJTR7_Bcy!}zp_U>;S_V+M8<X0Uj4L_CATC7$=7<jPiN?HV
zdS2T27im^eYi0>Mqawn+6uunpVEGySmw#SQ!~**T!m%6^`l_X(SS<ET35F-g%IuZ|
zhse=CfHjOaCY(Cws%*l4jR?sb>-^NMLiQ{Z*84K%MaNJIC=pS{f=R$e;7|zLYf#1T
zSAVbjDqITtzTcZFg-S_lpveR<C0l~BK!fez9pam2-Z?i^=(zR4Y<e|NvUE2}j9_qZ
z`?C8{5IBkLS+|m_fYKpdEJXyyze=yz#aT0_j^=A5Buy6-8q-z9rS2sDk0VBv1iqH<
zE%PoiAisH;S1x{@9uBvk`ypar9ShmmAfBxN<_C2gr5Evp9FbX^hC(&DNkag^zF-UC
zJf~JkfCA(Ky0hXNiq!^|J7uSAx^f3Q<L{aFcpq&(OvTIF<Dzt))L<w1!t&GTiGX7|
zyXuSm15M??y|Y&t(BgDCK6W<M)zFkEL5csps`4AZcz$b^x3t`RqDrbDLf+E)@@d}S
zXvJr_Uwga)4zKI#&iqrLd#y&sjZtXl3?HRpl9x%A@15r^H-&8?BX<+D`d3I)Ce*wo
zn&oenZqM%){s)iv<S<&U-C1APa{9eqn=~xu9o=?Ihxbk&)9pA@QGBAz*)p-+B7{aZ
z;HqJFOr8dNV9Gyx6OnnSK|qTtZn(euIAHUm=f-I8(cl`Z^1;EW1lHC0cB}Aq%g^p=
zsMm8bQ;A!iG`Z$fi4_lXs!z4<&rY&a_wP5wBI+c$Un2iKMnqHdZ)27O+nSf)gxar0
z+kXE7?Q%UN&u~g)o6lY!K!%zSev;z7AcoDgF#>;dplA6l&kRF6SgAxg<%B#rbY8)q
z_HJ%$xm?fvbWt^`NrgWC_=x=Sc$gdZweM0X#hWOid&FdV{q0R%#gfELBW@iSZJ3%#
zxgE~8CKgIOczx!v_hcnm;mEsM-w|2YmX+v>-S9NiRI)LlMNE>&2L0*j-x%XNX$7<z
z(!g@e)Nr9lFAZzpMkvbKY}zPCwKS`_MpiGjd~yr0zm0Gk*8VD!N5&lHPnH+@{fR~A
z=Z`NwPTKW~um6v#7n*Iit`lr827s#yZ(2KV7*y@k;LBx<db9rL-{<(-P7<{F&791V
z6u`+63`)holLN3{qt>|1XO@;w04}$lCs*Gf1wLo8f>9(!4Hu0b=f&(NK(FczQH`CQ
zPvEMYX_lNclKbatvqNfsKeyC9G#%IS;ds%L5ta>Rp{}oYb{q9y64w0dk^wsq2Zck@
z>`S9!x(zX|&I7IaF6AqxCYh5hSt)R*d<vPmr^ZDPJ_0#dslF};Z)uW(V`YP`#JI6a
za;XGDg1{Ap#z%r!dZ2BXTi}JUQdh$unxhpAHkjN`=#y-1MOFd%)97!)!x#h{&dlae
z=3SKb0^EDMzkfM`jOqb@&9cvC*HsfXW)V`RKzy$eL?Q_kz#wK-T`swI<T!CA1t&#=
z)|i}z`^-UIRcD$EVU$_1_8Ia(Blo)XZguVsb*O%JcfHfO%l|#Gd~!MX@7~$jdPAPz
z@zD@zc$WFg*+&U)oG8cb(djIJAPu4mPEn)!koG*Y_+|lZk-HgW{daLpOwY#1N22}a
zCi$FF2l9Gxs}hN_rDx&wqL;rIz~Q>?A>KmM1XQ3iyd%E8*y(B4Po01NWU4ZQU@WAl
z(o`fDTC+&As3GG>h-kOwlB~2r1gSv!Uubf%=hHqP!Px;SgMHVxhaJ`1HKLuOoWwU_
zgzq#1kN*(f7a8$P&Vs-+sjXjxhqC_r@zdMPAa+8Hi%DKB0j`n+fwBmZm1<f=sMUp1
zrAqDbv)@FuwKg@2zbISU;>~HG$g)odq=4@U#!colC@p+Z6WJ!CNB31@;Vtc2V0rGT
zS)9Fe{ew~?A{Di4+-zOmyAPih|DBvI<j%DQE?q<%*cW3|oOb|IjD@GuP4h;fB6+ix
z&wg!m8QsA_LL8^(=Z^Of%s9n7sz{cxK>B9!&sSXp@5V;v`o?<jUvD$z^|~rlWJ>(M
z!zgWMhV}K#_e`k2z*BbVG9N`XQ+T(TS-g;-Ra6RbNa1*T6-@<#M9?Te_`dW1^p<A)
z&d<kp5@2@l2-E}3X`zHm_!C2`>e3WMHB;fSo_wlQa28lVdjTYz)d`*3NJzk$MpIW?
z)rYH0u{UuU)`;_6yuP5g{f!FjZFp&7)O<a75utvZ;_%}@)R`Ad#hLGGqf-()sMj2_
zBZ0LsiFaLU3fLQ3Lyw#Sg??l3#zVtNby^uy)0a*49qv2VCX+i%K21huS*#y@8-0((
zszg4S-~E%8T=-_;?drGrqnje+LW^Z^E+efi;3goaY@u!PmyaN4!#D~?q;NEf)s%dk
zkd9NU{2vVxkWXIj@7K!3z`Tx3RlhWU$@Z22GVERCjOilvN?1=BrP2E9=8b~DwQ^ZX
zu(KI{s1A#I>cIDS!&=kUyW16SgJ-lg-p>&3quKC_QDE{6SZ#n~r}gva&sgmA*7^T7
zYK_s3vYhXtwPjFsO<v27XSgXrMvy2d$$#N^<350lV;0GN>Sv<PZ*$^3Y*Cs*01*S<
z)3q$mB`u!p6E*PP!FkipyS?Vz@vaxPdRBl?)v%V=rIy!uuXqA4)sP%OuuA8@FE3W0
z&W2C18ya&zT|vHGzWKiO2N0T6{)jZ4#-u4Y?J0t66;K}uI4K2Lhubyz%P!^}^~&I!
zz_>l*MJkV8J>Tafgwla2_;D=ofz`Ka>H(NVzqtW&5<GY2&}>e)7&n;Ui`^Jw7$pd-
zswVDP5-Nwp%UOHEEZ~He78EE^ZE0>be8^UZjGTE(0TRlYkTwF~kiX3HI-5Hde?2tS
zgj>d=hPj7rj7S9|H`@!F@njen*+>B=#4_Vms6yK07iUsyhP;(fnak|VE^z?cagMI3
z&6o0@FqTxOR5_|m>Jzmw!;R|rewJ{Yzs%2_Gtralib>%haH<P(f4%j0cE0m+{QO1c
z{U@sI52}!=l_hHFTK0|p>f#X+Bd_>u`^4a=obD<GujB{|f(>R|@1oMsH%s3!kXGNM
zP4Y?ReP2o2)~p-|$lAa-89sXTlfy&{(b8B+g4pv?iP^FiM&|<Y3|IgYi4>Sh!7K3Q
zqB+8O$JC-!kC~--(E2+UQ8N323zi#$2V#R-GR$_(*X2jr`?5ci!Tu1|+_{<mLxoZ^
zn@-1QsmUu~oU5txKI*?z`)Hj*5BUppe{R1e-Uhk2Y=-Z81*la9%(AOqp3S-5uDb@a
zjx@NMVdwnUoB4we4sPFKR7<y4FGj8T$BvtC|4`g5B;Li{T^9NVET9+KyNW7|Q++~h
ztfAVZ%)f(T5Vl5*ZC-zW;ZwZtr$$-*+)LOAFT$pl{$_6CpvH^Q)y-M6F*2f!l?GL~
zVDsW(|DvMSvb$99^6Qh^AI{E+aBxDX&`F<Dy{oEnD8V&x5x0!WZSFTO$|r$Ehh;yh
z5y8x|?nX2Q4d*Xy_1>B&XDT#bY7?SQJo6cTW#?vZ)0^m%qfayb0XPQnQA`gr9x6Zk
z?L0n49X79e(xc_uC?b2anhMwKxc);X>ay4obY^Bq8_{jC(&yoOku^+<^=bSR>~VYD
zS{@kZMCKG7axqM1ua5q?zi(Y`dbgfx-r}~O$J?*@rgW4biQUK9EdBKQ&pl$IU8I_7
zcuq@Ym7~a?E5L!UlWzEG3+2=L3_<9M!PVMumQv}bVZ49;=#CJrHLK3GEgR|^0D@kA
zC&{Uw+Kx!^>|#G|=2kftQ~PUi{r(Y-U9;aA9Lt}dS(Nl#7cLU>Slx^R)RS@MZLrAA
z(|~6>HMO-a`+L?7`^G=G!dmFbiDJ$xloM4#<p?6_=}dO$BJ@36QLY<ZUvpf~g0fU7
z@xHjuLW5I$2q>d1s5HBDdxfc~yS*nfdQ~_H<#L2|$zcKnrK&2n77QkGN@-n1=pCs8
zhg1>oN81l6Q{jqK&%{YO<UwZ{3<cR4%wKEtVMV$UO6y>?ZD?^C5}M>}>lK<6lf;o#
z$l&u<U1HB3U8)O0b%jETkli>zxfq;LXMMXk-QKo6g5?<wPweG%K!z^lsAiv<3LHA-
zC;?)jERus45NhEIhU!i{D*-PWk*(%Y@b%677dmq$d*$KU^Q*L!ZE~;nk%Sc8!E1|+
zT|lEv{ub-9P*9pi0%M$;8Kb1^f#yF;?OU$4l*o^SnpC3NU@_#2#DbPUpw->oU5t@k
zl?vu}Fzh(-^FI<xFoCL9I@+X6_+Odqw7KLzq&-vH(bh%ov}+HPg_zCZ_3MAf)33Zo
zrEgAzzyu&9m!^cS6Z$FT$gCKk9Sk@6la+>CjX2*WnIhNa$mV=Ic&o+#bPnt4_&Rd1
z<)$~ZsuT6up8U)^+)7-wb9+85YKXeoePi9vt~HbWV{Oo1)y_i4nMY?b)u-)Y7!fzO
z(I`k;R9S^$q!y}6fioXTYRm>2fH|tFyzWy^>|kRdc=*#;wmD>_jFE@U|H-}aN93ir
zz|=3QDX)B!Y&kzvR7rr7*|Y!q9=d2E_vX#hyd4S@(M?S`pfGqnXLL!FZUe|f_W*lk
z?<NeC67)AY)!b>0>E=ZF2|&lW$~SLX?mXJ{)Rw*N1x9n%u9}nVs)G4HZH%N(ez3+o
zr!AhEN*@RTN*o6ON}4Ep^6TsT_IUC6r$-V%lHw+H`p(_iy+DlUAnFH<pvS_QnZ=fZ
z5mi;Sbxw?vV1Xq7mfU#Cca}l_I`CtbrN2Wy3VnbIu`b`<hG6INzj!IF0Jzd&?l|sh
z^}-O3t)q1ACy)JO+Fh$i0^=47g6~dn2|FEmmVYH%C0Ax2iJuL`jmccmqHaH#i9amY
zs+Bp3Jl6tG2B18xs;9@UzIE?1*EL9CQ6X1Dz<$Sh{~GFO@6-PN{$jg3+D?D?=yFP3
zJ9w+Nx71(zvg=E9$l=kFJ?-p5F2joCaJBOJ`QbOR!kop4Y1BE;-<N3@-hF>|<`7~Z
zDWb-vE9R8_vz(S+`ecAr6006>Dw^OsDp6j4(h3Nb3-2xp$H-OcYlix83vK17hP)a4
zxjxk2?itzZZ>URJ)FG|SI4H7T*4I%W>#E-{1Tbku9LETViq=e83qF1@SvLo4@`(up
za?O_D9b9GsYA|4WP*DEw*_Nm;7Au)Qx_h7YS+cLhBJumxh)JNCO5~PERF?J-C_*`t
zC6tJc6h4ShId?t3lECe~j6+#8oakU1qXq^TEol@a-o&Tm5V+T-^Hf)5#W!e`aYVu^
zK?IGp<6yW~%!dYeQe=-rBxC|U&cOg*Phyu4w8B#7AtN^x-R?tUy8_}u;Upj{Bqv;8
zjM)?*;k=3Ah>Y1a6@S}A6+zbB4`roRrcrwf2_|1uK4^9ZUX_+8V>fbyFw}%iFOoU(
zANf2qIg&I~1y1mcfS|;Div2yM?2h>IZq55p1?GARa_2amE%+coGj0SdDip#bAh#Om
zSwxZbL1P~n{;f|9PqsXSrs+xsN^%oZ+;e(fRk#x%MVMrt_0hYc1Vo0zKBl}w;TG$S
z#&#qoO`ZCmv>l>WBfKEsa5#uPnMPHIijDhuW4JTM<VCf#<J?+7Wo<!D&R^m(M;YxY
z{juomTKdGY{8jj1hK*`Q7CihO0Th(PEa-M$>%MEy*Y{2`ZE3f1#}|aI^MUBVg`F0J
zQm=o!m`BIV8{S7|--W3N;mRS5`Eijl5JReXs~*j@CM^^mhMXb5wE}7*Pn&-De8KbJ
zaBKg2Z(L@TqgzlVP+I~pF~08_V}(I~E8^ZB(`0v<6P{d4)F^Y2#NU%c_BF>Nl*m=)
ziSZ6tji?>FEy*jM&q*|tr_K^00=p`C(kG~#Ru|k~`#Xu~YivDiuIp^3+wlk}69RW&
zXI7TI>`-b?%gc_prgo;ky{W3R6{inHgTdVE|B>Xs2CcqQvCAgv1I9oPii;q0iYeq!
z5K?qVPX>Cr)avoTJHww}VU1Ov&1p!BRkN@3G`rdf*++-pWuN!uap-I5BjcxSWF;#=
z`uA1mnTvP;>jU%Z;dd;ky9r{?V?8507LVc2F(}qS2D_z>fWJQ`Cbea6W-q3>SG1*X
zRm$5x1N$VUaR=0|%jaH61U*-Qk`q-sVar<*>HwBp)ks(=E0?pd#6pxJahYw|-PzTW
zb9q%~J9ThYVeWD5sF!0tv6pHY+w^D(GTVkuurGanW?t%ZQxb=P<lTOhI)&`@7YO%C
z4Ju7{yQFppKAh~pmwwuPxeT;oSJtf(r7blm!=-{w{#FRf-ejWuug6m-az#Tfu4jjw
z>5}vC?MPSNy*uO0_dBB2Za<7;DwL4Ib|lyVX%os1R_6sNf^>$XZcA?}epht1ep#uu
zbiJyrCi_^w(Q#RNI)X+^PNxXy#vj?T0#Nv$7ru+)Y$^EB50`>YG1@nM8!~=ZjB%_F
ztKD6%&bIHDE_}MKUIM6J(!z72xg+pa*OxR=#W`HvS&%lK{yiGkwmaNL!PSI`-Pq}+
z;;4VU&}r0$%$*(T(9W2P^p*M)V9isT@^)5;7f~6cc}9)a`saIl-Fba0db|7yb$boe
z3{hk4cZ*TL#Z*pG{*vrt3<sWi9g)EAK*^Qvf*%sY$3R?1Y%{<8-`OGS`KN^rPhbFr
z46qiE50u!*8Cm~OK^ft`R>3I5tK$~-hxIGX-L-imRz%EBt)!b6CZOq7mh0{+^td}S
z#z%pKVlg?9@d8=J{!T^5ja?kIlE>LgXNBk)cY70WL*9@B4mI7$drtYfh#IMZq2fvF
zWfqHzLHrywz)kh(cN=lCv5>)7V1zYj5*kWzT7^nsRl&*zOt+X_%{gocZe_|!o9D?c
zY-n7jS*r`f=vIx}t5t#pkTJ=5-jc2D)rPe-^Udx4=Vg!f8gy%+s;Kw69soAP1I8ZN
ztSW`JzypC}5Pf2~){Lwx$rid_SVI$+@sTt=27C6NzG9C}!ea*so}4n^napGVwVL07
zg)a-QugTrwkxtdzwN-RU-lP7qLXw3WdpyGoM!22-{-@;Zdk&73qlxgvNe(9yx~A)@
zt4C`U1j(tr)^yN7@e)h}U>?oz$JsytVn7+WD)?K&-#i-1AxM*>y2nButr%mF17ODD
zOiZd#SE4Aa0QT-<^kZ@F%g3ChY7G!wwcZkDf^_@J^yevv_c0)F5BtzV4{PSOa~dgm
zf!?m4Yy_R{%`(!`3GuY)4qgCq$Alvmh*^K}j-`Qo97nk%S5F&e<Ywc-6O~u0s^}Bg
z+5K_*><N7pJSGc#w0I;eQHO4(tuI@D?P^6VvruxSt7fE9jJ81AUZkjJs35lY93NsV
zZ@Nd<C>C8CW;sgKg!0#8wH%#FCsHHlT540_2FCtOB@+tvOqMY)D$eu<<ArexFy&~6
zdtcO1DJm27rRkt+e)oFmZh@lUX7|b2f%%76v)IFyhP%J*OhLc|spax^%6BR7uIo2G
z6Y6NtlufOo@On~vUp&AqB;;%;oubVb@z%1T9D`?}z)?E6RoQ%_uDn6}a2DMM^bk?C
zJC^3yVUXnB=qZC}3QuK%1F+vF&bbCahOVBYXrd+@0|Q&f?HPZG9^_LX<H*PsiP6-N
z1JK-tw?NaQE;U}7b1hrQ?y;2rtWoB=_;i)*V;f`b(tqpR{2P@LsC`CcpYww}I)l^2
zrft5*0cVqiEq?nC2>H@$7F-%kM$SK(Ry&GHOPkSryIsr>`isk+u0~IZ(VzM#PanKN
zg<OhuS`wMKH+}q}2j`U{5%K-|m+5*Ol?%@Y*ttt0!k?5)&hw>Yu%8>+#5+8W->cLC
zPY09So$C5Yz1EwsBtcanxp;^b7%Wgp8Ox!WGMp*v2ds(CEbNVFXg`wiTmD|%$@~t;
zdBn@P*ZEF3iiG<b7-hBO)R4dLvsyv@S#-Huy`eDmI}i8Oa$_gdrb_6Ff>6(-B5;nP
zRwYF|)kFFQQ`=tawC4skGmgTSfszCzrT1Ykovh+-LJHNRVt~dt?rQYs39uZ-l6;|z
z|5T_?!%^!1-zwfp<0+Az(|g8Zi?H6Vd+Zlq<G(CRLj>8y)Cl_M^wdhTamFZ7dZX&>
zwQLXXFZG!5gHsJ>0Bn|8iT_L|vzCwR9UlR0TjkCWy;VMTH2s-QhLzp}Z=)%Lrw9Qf
zS2`=z9CrVVvi~X@j=$7MlXgcbz?wC(RPbNJL2@hGRnTnrB^p*cuA)r<x+%ug*Xw8H
z<mqo?6KH3*w!d(`*xJziB^hO3DaQJa`VWjH#{7mD3~rX|BMH;?*^&uj&z0G1&(+`2
z1rPccl`ek|G%-lWK7M}9^_ZH45Xl_Fl#Ha{Xi7Ue8_A7mf1UFb{~n&=a&GS2q@Iw8
z34jFwF!KRN9%28iyIVNk4a(&BdS}>B$oK`^-QFO3#I4?Zz0pbCAxDK8ct72;VBeQe
zu*kz##H(ARGoJ(F+zDPh42-_7d9YV|JF;}GS%5WO7%i7M9vs{a_|p)Z3~ElwapyGn
z8pU#*^Z=PU%JXM+6UYvO<QC7cBKei2TB%A``?y!L*#$Gc&+Q+CbH9}?f{3e^m!B_r
z7Lm4{1-ai*W(x;?n?SYv$j`ygW}Q^Utvle1U%t^A+g38M<*oHhi*>Z#yQmCm`ADxT
zD*R;=T{%RlIQCNRZ+tjs0B3z()$RUl`f%ouhZH@<P@P;!6&KKM)3Vl~q5kYwuz^t8
zxQ1)ZW&xoO;VE2Jd;3BQUnvsw?%q%leFBiz*xldf`!p{Rf*Sh*!BZTCf`8KTn_5eW
z9QVx~m`(`hJ8JQd^23|2Po5NNg<C+*eME{yD<5+Ey6n$kWs!313DJ`j(jt<m&olNT
zW15}U6WR!z=R(<sU(O;$SmOJ;WaAXdH@`nK;r&O<$^yU^RaU=WYoLBD%`Fb40fEwn
zQK+7=N&lues^4>Ckwt+;xX+*INp%(a#m??b{S)m1#{Y<61S>&08Ji>NalpPNziiFr
zoZTzG-`LJe%>GFYMTf9YwXBB8!dUQc&uRY7T%ke_%hi`||BWz-xsUB%E{MMJ-}-tX
z>S7*z2e2OsFaOYP+}~sJtJg_0|F_^;gGTSow`ccKp|(44g07p*&DlUg^xWs{Y|rF+
zz4Nj!pMuYIzJ2__H4$Ix=3+@_AsASlJXmP*XwI%$Kx^_n$P5sGf=4~jbk9sOd+i}@
z1}O}BIaSxyotHZoH^3nDd@*1rNU?lGn4}LN09LaSj&?~Fr`u!c7qf3ITmzjZp_4Ay
zvT{^;;^TH@&b7S4pJk&_gW3Ts`lPU!ikUZB%Q^l9c1C=mbx0QCmFp%iPfs%7+ihcn
zom)6;LOvXFMACqmiIA(XrbMNd!$nOz0sKTf9qs}yi*EJz5IljL&lPyd%Ph|ywIwPb
zt(ny5Xy}x?9|(*_dGHC?V*b4UXD@6isDm=LG}tT|AIOb=W|Elqbkvw7EFOPSf#mJP
zlmR>%IUdMYg1~sPYo_uHC#fR$^RRvNR;AB@D;|!ogv1<ECda@8*n0ea0mE_^n=to}
z`ry@|T?*xIhxT^9!}_OalVRV?%yS=K`j@XW1XmG62kkTiY`NAm8bn#TLz~}LEi4Ne
zb2?wN2ZQrq_inetV;GY;2$(FF;{otn93`bb%afOHSn|lN?{=8<=k-r#499#-oMD({
z&Tz$}GLgXc+pk}5yDj{VR`F^1I~Z66`VavJ3qACV`T02oEx`CXHgmuZ$h8+-X<cs?
z-rZhom=C3DE!v}Q<ID$gNvvM~*%>&_1B<7svE>azW%JDNtzbP+lCWo>zc8^eaf-)z
zo_Pvh>G*1s^Ur?y{^Lj`F(fsOg)sjdrDyk>&9V&DW2uZh;zKSMyO&NhD%xxEdzEh_
z4Z0)VQR?HY=V_nydS?hnXrX?7&a%$3IC)DV*hdNjg{R?h7z^Q<U})xyf$-M$@96PX
zdTTt8S}c#VkabYXV<0PayhsMi9a*oaF&<%e*GYV5Ta9=;ZJ^`Uy%=;n?`Jy7+nBq6
zi>2=WWZKbw*JYWa*5@Y0SEr$;fx0*Xl5VB^HMKQ0E;Uq<45BV4sK(1ZpWN_a?*h93
zJG*{Kz4R5}>wbC_m^a)uF4N={Bvt$HjGs|Dz{~q|eS{&?PGX^VhVG#ixTETw)L2BN
zcuzt?!q!&ypvE(Y*Lhz8%dptMgV>e!_E)}D&KjlqU&RcKUpmR2ZLS!~G&MFfq?J`7
z-jbax0z%34Utg?ZP`~mAxt&-!CAk`1fX4`KW-Zt7SgBZ2=scqH>T06{h;GFl%U&L8
zo0tn{JN{oNV3Mr!B6n;d;EvZN8V@hGp)XI}n(Ft@U}c|xwX^)ahGdz;AA=mr0O#`-
zw{)}2Jtl)~sngx9c}jO*312CR^o>Z_`WZi)gC!b8_g~;5%~;T%LBI%iv+J~RrL%DN
zcQe*b=JY6z-`kVU^2Z3>hnEc87)x;I5Bt?;_q@$atZESP2;Sb4`3@gDuHhUmKg0A?
z9tIczan|+Plhghl7kwH+!?u}yxM~2<waFgc?Fc5u86GH^_k;V={BZ=KvGpJj1o0LG
z!l7pOf9h9{GoD=TDdzhwwfh})6Z3Zj9_}w-sGsbco0!ZkE;?gdft%dF03Sp-Ix<31
zTf5YBOWEuKCZN{k`sl~<vdT)&^RXjQJj`uWTv0&@K}E0%ts#-eQ`-r4drYA8PS<_}
zv;0bcwTbx;-_hRU=Kj$3?lt*J)>A<~RTPQQgATsoWpq-4ZGCAf0ZDlEyK&|$!IX2@
zPwxcX)seSaX*fD0V?taNk77KsqNFO{hJ#>3Mu|<u!9@dP5nA+^e76b7M|<{2(7+(g
zMU2Dc6VFmxaH_h@E=52)prlrE^~FkO!jgSoICLlVLh0Pz-4l2#haei2^C0g5qu43?
z?!`1{f#uw6t>PEe-=8VwXvQPlMisyq1a25gDGp}yt|Gdj^dMr=$zX=Hoy|H$pD{@R
z{p4Qt%n6?AVM@X7Q1qgi$;)4z*jB8U-<_Ve?BSUSZ55ZF9f%{cOWgHoZZrk|RN0?!
zaV9z|0v+&Rb_}c!jL>2sB><r~O7jF#jdLxLd7y<R>&i;s{bswe&Hrvc-JPI-K%TUE
zOZ~CBw$$P3?_%YoX!GGi7y(+(lAMm%nm~m*shom_B$-;CygRi#RCjw2ESaSs$PqVw
zV#+Ew^Tu0>67ksT$2&##bgCX9D$}BY3ET38r!wm!J~lNEO*sm*#|Tk>+D8^nF;h~h
zM~gf~>PIh@rT#%9#=MHyu(;P?8vHPefi}@pO16zO^)~m@qUt^=_18z$O#eNk7}=D{
z*4Pf^c&WakKuC(<N*6ou_4SR?3TT%BkFpWRt5v#)K05X3yxDWr_LZ<(3h3Z?8(;6~
z+Ho;67LvhCzk_k~9DRW@6!F3O0zs2i?U3u^I(q4w?|M5XbFKx_079m)E!fkh8r$B{
zK?*{@P;?S!JiGU$%E=<Y-)3z*Q95`jnBvXuy#MWntXWpBC8CCJDcdNEJ1zww^7y`9
zDZ@gg7-6vvVJL`4bMm#p!T$bZDZeh0Ne9vX72u>pB($l@fl=UOet?Yd>613_qp3NC
zL!;4bswo@(De$t#w$1ea1u|e=TueLC#C^TC{_dr(wR1Cl|2}s2mv~9?VznOF!pG|N
zzR3JUDimoC6$Bw+rMlrl?}(c6@?}F#76Z*CeP7KTy3O4jkD2@Lj`u4%mm|SkFB$sZ
z(#M7`L#(S&GZ_xXQ^k&=G;1!!GuL~cf>t_%&ReT;#knN%Spy|a^k%+(z3xDi)IbyC
z6dbxs;#cYFDl02h|NSiSeIuH~r}XD~^Kiv*DI`LHM#HbDrHXwtfP68Kzf>cghjgbs
zGo{pl-tzh=6DWknP#@DJfRQ75hY8HL7F}kL$9RHL0P`iq{UZ8X@UCc63jjtndEMq?
z-#>1*v$H$@t<2Qv^PBgb7?j*He!)c)d5~wImOZ59ZRg|TL!Xey#h5n`9SU<K=nk{e
zD9z~RCJLWK+T14qa8qnX`;__UC`)jcc;fgsxC9*?W$$LG1p$Z>qZ=4_gPGFggm2(S
zZlX$$x-}zbd7LWYB;yJv7~WyW6BsMCU$s!K2!rWgXasuLQcSqd4aUB9)&v!2C{as_
zRf&mNoZaMv80&e!<6VS*EzB$mz?IXNUe0Zw0vEb(V<T3>Y^l9<D8{cL)`e_TjH(yZ
z750)`UI2+l1hI;FOc$X#DDGfx86Ojh8i)ahiS<M9>?+be!#H{w1y|no88hFDr|y9$
z*3U2|C{eGBC?D<CyxFgDMx8pDdyX%>?6A>PRd9>vd2V)psnzSXyZi6mX0pP|6QM%X
zesle&_Qnt_aPu<R@I73&YsVOKl0GJUVv#IHB0ofEI6YcKJSr}D)ORrp0<mCG2LC<B
zCx}Jp{QmdltOH;9Zkzx1#P05TCE(3*kQUoT|Kqz~)Y>xd=BK^$8|S(nd}L;i)B<Nc
z<!j1<lL!!mLhmKxTb9I{uH+{Uz>q6_5%Qh=ket0UH`7NiB?i*1#|cex>Yh*&cF(kw
zmssjlARV%2a`+j=1h-vBnBs~TP?%<ya^Zc9?~-YAfR*RPzU$rD0j`YwT~8RaEa5Gh
zi`T~JKe}%amW^<-2WI3lX}w9|CZ>5IcVA7W+_2-0>3vZ#-do20Sf6CZ=G&M`uf~}S
z4Hb0o3HrtGOI9El(GQ||Ed}){D{G^mLFhPo3&&dniiX2kCt21HQ<B4|YG2nS&y!q6
zv6W4;dpfsA-;aqi9x8korsX4jRCv4EuN*ja-f>4!c(Dr9*jsjcH_m#A=+l(A=Yb?e
z!moWCJAcwQ`>mXsGfD{@uXD>vN(o6$I^ruldQ+h;xlc{GftSS?AKmk^s@Gos7n0_!
ztFw!?D7{3|@k7siPP6HUD9f5lMKpC@MKMjev1VW7=^KA%>7Ogq%J(1Kpnusj;EC=w
zFWMfA+Pv>Of`prAIZzN;20G(gcU;sL2CX^W+3sz#J-wDI=cOi;vj)PVqa_wG<dH+*
zaEm{KiL&nd*Wy!o7uWTLkEU!r?Xc<Ao>!T;Ojp7k9)b*&Hx)s)s7B#<cif!uz3%r?
zeq5IW4B|}{YP36ld^=8`_`Z1Y;?jUynf=3z3&BAwi@(R>uPOyR7IMlXTfOo@ulnYY
z<D;Xa1?9&fe%{_TV!Qp&Qs(%3;!4)W9Cqy%x+Lfn4*@PCUDOdZ0Cu=|P=}vxSM5uD
zUMS(e)m}YntVBplfT9t%KUxm|txFyl-1o9|M#g9L7ulFB|C2YtY>Ulq`@f?PT3g#l
zSmZr=IEVo2-<wYZZCfuZ6r1d}786xkNFLHL7uDnNt2Bu8M0INk1Rj!DkteHDg2F;I
z;fAF2FlG|Y{9LhNR>AH_<dYzQJd${MkBrh5!zTAF5Lqc4ABEf3xUk<YibqJ;)$qPS
za3}>E&LE%Usi!~HUrbZoui@=4746I+7NjH*OU<E7K9q^vj7>t^kEa&UGdYmXtb@K+
zcofhsouRqVglVbA?6k2mIXk?_IDmwG5t)v+{AuiITas6zBdG8J9Lh}K%ZGOlhGe-;
z4<`r75cGwTp*Sgn)>ti=nJuUv-v^PtH7auKcP7wYj(P3@D|z}@h9z)q2AJ>pZ{ONV
z5{Kniu3sAer@`XCG30~Aht>wUt$kE}d^M<T9<q6YkNS6TySj%RYybO|?3s9uS(a7a
z735h=YBoMItMx$Bgz5w5P_dc2_Xypd*zO-!p(=TL)u^eQLRp$3)cMz)yRADv2diK-
z*L4_vbbsm1*UW;eUse>9QBZumR4&+Pk|!VBgD*QLzB42p(UU%rnt1YU%7`7`bCK>|
zNGtYt{FYa%8U_zPT{s&uo_N)oJ1<$aYab}-ffres9SkkrE#cD6b<`@Us!__{YHrU*
zW0!(aD5o~*r?RQ)PdWOe1g*&F`opyA6rBu(x<gIP#5tMY&YIAlDyjXgd=YYZy(7us
zq~@7Km9h>fpK>s%ZQZ4|8usZaKW&_i(LKia8c-AJ?$!1ZqgNRQ<qPu5ARSfC5-$WE
zt2IQ*+g1YbNXhc?e77|cO#TT`Wd_8q*t<<s;w{jG;Y;jW*|+p{dHs4R%`3gtYwIg!
z$M7!t?of8o?|ftqyZ?QRf8O|&R?5k$Vxn86I4yDJeZ=5h;oZTVvucNzg}F}pKxJJi
zGAq($8Y6E#KqzP%?|?C(ya3pWcI76zo9;RSVb;xZ?2(|$=cZa|y8huXf%_VRQyA2m
zQgO2HT%SK@Ddu#2yAIwNaN|+&c2R622+mDuEAgL|>F+;ZdaYev`#3e|&vd+tiop<f
zeTafEBJfC%Fq>J&!G<qFP~cdR@%raiP%cCk7dR3bSnt2lr_}jIRKi**TR%E+QBo2M
zDQ3*@HlU3o6MFaCMaNjdYGq5ADG;?4Rk&??KBygZi915;m@$};jD~Rp5F<QbAVPT%
z=-``$m>kSR<NlRN<(~s{KPN|@{lyjFL|p78F~Bpe$RF{->(Ypc`l7o^e4nSF>F#H)
z>vf9G8_|<dT!z*WAld;IZBhXiY)=vuwC~9(9%+FHPNYPCPid#;=DhO^watzDi8D4a
z)LR&}5b*n-kdjf>fcEs%$uR!u-spFGF<jN*iO?AEmrpwI)xFJ7!ry+yykcwPQ!}I+
zghWs?hDx!pSRqodB1(UB1u9F`qJAKHAZC#<c^4`)co3p@8bbr1?B)30Wwv7BHOc`3
zg0k?$F1FOzV+4|{`!i~}Ysy+Z>W2#^r3N<fRz)|Bg}|=r>zsrHF@nWy#gci{CS#Si
zvR=g^o(y&wgR_c(++(p?slquZmM0X|^cNq#a&>)4ety}$w;=sqVlvjZFR;2qMTiio
zrq+D{=NCpEg@vHUg*d@<f?@H`K7d!P1aOg?_vFE`1FZ6LN~i&-U^i4T%Td%;(unnS
zEb^e;ySq8d_Ci^ul`g)zxfdUEDk&KlM@Ed%!VJYeSL?uFWatkh4&D`+&iii8p0?iY
z+?{o%+F$?fyaNCoxu~GcUt3)aVNaM-1_-{Wz5?kIAb;wzB%hD|YN>Lfm3sX{bt2ob
zFl4hMwOpGj5>+UDSATaUd$-Yfxqn`}WGb(J1-uMZ&zNpkP+}L$s`+JIbTHS3R)^wB
z{UWcSDCF74FS=@jGNyD-Hs(KR`y)#Rx`_?%$BpwTRzvxfmEix6rt^%a^8f$%A)9iH
zIHEYR35T*roe+nTl|8b{-rFT3>&Py$ijZCQ9$$MM$;vvmW1l!i_W%0b_@BFOIgjhQ
z&gXi+U$5tLc>46Iw&V>s2N~W;I@jNTR&sl|Y2x_oc-~=U$qb^I6e~1gs55|0^uKNq
z)C2yOrAv=A^f7#_T!b4Wh*@b6xu)8t76tz8NY;*_JFDb7_EtXJRRT+jl|KV}@8}rL
z%-uC@t8u)~?e~iD5w3D#i%SiK;E`dWAxrN2Q29M9UPdj-@*4-YK8mMkIrG$_tZc9C
z`rD(u|02KvdTeHNrI5n@P%>tkON~w%(gs{~Oz858kBQ?J6wCt%-MH31sX@ZU;pW)F
zDaPkp3i++O0?$@n6mAD;{N7ny>Wui=)VZ{@@(hHoz~Up-a%Hmkg}GJDbnh!E{;U?r
z=dm~}F6P<swc4?ld;7>3shYmgC&iINVp(nUKxte?ZOqfh#V#mlB7h-}(2AWi+6Er7
zl04z{%a&xs-8}RWX?w1`@F8+F^Km(gstPoWcCxV;-*-kl_qWJSVbg22M2`3`%$FjO
z7;y`Nl`iwj8#S1#!;kjn)cfxLcj{3Ae%Pe_b#3Ar@)^{8M!&3zZEHL@eE)TU6+!Yv
zRK6)et);2eZ+&Qbw`Ezd?c^M|1X|d4oSZ$4<J3QiP-Q)|EAUF<$o2SOUl}WI2rVLh
zQZm-hT80_AFm`D(O;xVe<M|0Y0=-xVJ5?hRcW&Z{liwE4FQM*#szLOThWf8>YMQqB
zHwZ`CJc7}j8%{#el7z$G`P7lL>~B|}<I_$c5bP8PE~0R|?#Ktr?WPStpx?rO|0Uj!
zWchFX><oL%^mOB)%v8Bac23UKfj%&JCUxf(V$s9=;&}|(<M#c@n$z!01&>jNdAHeM
zq-CCey?zg**kbjOVl=R)ixie1+`y*nD~!^RixEk}4vD%SLu=*!I;j~$>hQ>8qK>gY
z;2{qehG|91G^HFy==!P;z=*Aq_zRNQr0q<69p?WiWAc3Zq{osHa*t{rX(NgnPOMAg
zL(tR~p)l8X7@b|Tk9DBXz)YTp)XTVH;o?P8ty^!!%iq^%b3XS&A9WdVK>KA!z7cmw
z7s5D|g%glNzsJe0mjr}JIle%N?A=XzVbdUeBvfgg)9nMjXd)Vbk*`vWo(w9v^)z^R
zHI4E2lCH21&j%iK6u#1bHYgZ?l`UJF+j;l4xytw^p9;=I?^WiVEX-k0;XNbMhNhN=
zeV>JE7CJV?V0_zU9RRjA5?khv)R{+aof*T8ls<5TxN&py$ow<nMe-V<q!CNwHHN09
zL09?AQ~&!&$diY)o_y2waC=R~q2Ri;DxTwuvXLRat=CUVRZfwnYlP~KAQ7gR?h2Pd
zY(3AWH1c*Q0;f=jgCl#|ov2UO0bo#G#&9{5Vf695_~j#oDn(*D)hzD-zoX-W1C+=0
zP`waix4W)ftxL|cWxoN^&yA~Qt`5dhdU5BBE!vx36r(m6$tk7g=9~R4QCP<l6&shP
zC^U1y=j?1PwVo;=DCBF=P0fcZLTX8taxV^S^G3tXk5a@8aPs$AIK+Cw7vU-Z{G0Y!
z7;Sm=oy`|62Gjc>V}Yd;q!qr@j{0PF*3F<97-$!8eteKsIxFQpBA@Buu#>E*aC#g;
zFh9+zIJ4lMvN7FK64}c!lXG=-(cQ_&m_j)_mX=q{^*_|qE(4prrjw3HPnf-B1Q&5W
zqdp8$J-O@c1G1EEkS0rf4g;PjzHg_z`FX!yS1dhH7UO3Zl=C`Ua<I4+_h-LpunrQT
z8%^-8uM1kcRM)+&tIx*Ctg4jR54*vay&m_uA*@irZ*9(SYa!U?ba+fN__dEiU7LT_
zvjm#xP!FQ$`v0K+f!Uv$*}*!zGw4&t*)lVk`|fvt3|Fa=X8QwTaDWSNnm;A^%5P+(
zW(C~X%LeZ}PO2+ly7g4v|0vEwq-}qXx%EJ{&3Auox8`UBSf@;V!=JqWofgs?f4AYR
zgB-FN#HdZD#D0UBo+61W!+g%?Xn;KQoupSWm(@@JH3Z)rbX8QC<F}S7u8d&f<7}@N
z@=FqcXZ!9HnK%El1hNN<UP-XUd)InzxlO#SOK>Zi7-rSKwPy|%oV>ie0$dCx@?hO7
z@$#?BTHJQ;SZz3!HAs1{{ylcaPY$!j%N^KPNE)!0T`qqTx#(fe2LDHe>rVEI4daE2
z_4Fx>iMgS)+m@B~z<+UypTiVftn~~JLrIs>?-;+6fYF_HAXU4Ag(PN_+^SwT>KlWT
z#%)U|%2G**DK{Q2+7r#W{LMaNDC>!6_M6e!`R<USJzw?^q-c#G^~zh-WC&D?gSkic
z;FT!XwpEn205pmE#l5Y?x7*>rP`qAFQdZi0F;GgDWR3=EIPse-8M{mZ(q34sr`|(B
z`<rU$zJ|I6AYk4k@LtyrMTxtC=(tTD1y@42v2ZHkVZ>S?P@chx<dgsYaG6p`<DmWi
zH<U6FI1|qlio1DJH1CMk**L4ry4*^mn+bj&sX9)dg+Fr8^y1yHxK&+OmB%<+reZX>
z?<!=o%;wr-Yilc$^-<#%lda3=nvj132M5;&ikBl<fcME^CvVeOncVGp?*MR#7yTkW
zQ)<&G2!}bUsga$X-L7+9z|Mkhw#~TSZdXwn#JJ5Hw6aAOrg~Bal2w25uTbGq<EaC&
zV9gIK11ZUKH{`Z8Qs8Sf)JQcBt9ka#T<@=2CA;jt^ZP|<!srJp)5m|KI2|<n5(i5A
zNyx?k!4ruF4jmeT9A!S8jK|}|z=k^Md&fTdc+9+q|5bY;xMH&(96r($ULM#w5^Hb8
z1m+Spi%5w^1MsFCE{ViM^!;4@+*`R*I-Evdpvb8GuoArZt=l}tdNq>o2R`VKs+SeS
zj=z;igHF!Ce!aTA^kv$yz>vvuT!zR(y9f%7kjy9wsw+EW@gXn{6u+coZLNe)>zVQO
zu;S$*`&nKVsAUH(Rz^f0TQxsAnmL;UwaVue*?!&R=f9rP#wf|+NcxQHob~i3vOd%E
zNin1M-3;%^`8!$#1ibYRE*`tPb2Pb!-7naZsGRvQRo7GHq0{|kH9KUd;&UM_^AHeS
z1+&qc!Za2ik7ccAtu={P=@bb0S$jD_pj^KAt;P?DTQgdiOhTv~>#XdF&|R@?vQC85
zn8vt;6bhzqBJd)VlARN7hol*N`*N~ZQ=xg>xz&IDG`Fq6cjWhky36noo0R|7zU(Iv
zYZ`=akKZ>nSjyn%EHD(a;B&^G1w6Dy&xg02Uc~449`uJ1$Xy;R$OmtI-TK!L9thbp
zrX=lTEnV!6_IY!h$?l{c$$qK;HXY2{+>h4FXcabA{rz8mHHSHD!9p!uH-C5YDqihN
zngz(K(d)A^1S)A%HcvYu;F6P5g?Ac@3j6~CY<3fIYDuD)9-E_sL6L<3VBtp7>?{c0
z>XEkK93-4;W(8lyiL3qhF!Fbg3U*X(7r#dXuWig~6rfnbY{dn{=g5`Dx49jADIHIb
z0M4hFtVNR~J_mswZmlnmwcYvufsqD<<Gz*H*z@BvSrAsW43-MlT!INF*GueeZB<Qz
z2P!40<oKwKJn4-pt|Fi~)G8%2#PLvrM?)}Ll_DGst|z$M$dHisZ!FaqjR8c@7Xj11
zw_rE=Q<E?x!X$sbq<Y>XXnat)Z`VR?Z5jfN-A|KsX^9&Z&(D2{b~|_%T4L1>qimPO
z6Vq8m(L|;Ex@Vv_fp){2&?I_fyBf-}n|LN7If!{u29)op8UV|EN_tb+bIV^RdWyL9
ze;q1+FEBD?&bN^Vdqbr|51)oD@@^4ZDsl32zeFQY{Fxr@>CjBLHd>0M|9!hO1F~xa
zW`z#3=#vuTz;H0|kr=EL7&U$uWQXt}(Mg1OR8zxA7m-pzXa%<6a<XACUg2u-oVgAz
z)qk6sVXkQdc=25Mugq8#^?Y&tnqy|VQ{60~^Y!W3^ppw2jyg&OF==1>8{?VOxBp#|
zZHF=MT@h7`b)1cceHlScGpRJOj!4>S`7jJD!|g{hGp(~E)#VdHGJm!Q(j=>pYjLb7
zQS`lob3VnSg1086%mWp%o`-jCa*BHCrRG~dB5>aaT+nL9rZ1zZ6F7~;%l#dWTt$Y(
z4_?{w^YH8Vi(+q&zRDvaa}BogaF|lmX6h7^()rl(P>1}WsO7(=83q3{w~j|~XPqs(
z9j@^nLshz4XwLB!x4Xsm4lYi=Xjll7F!TGL>L$-^bW=8pzt5CqQm7kOEk)_S`)aL`
zkb?XR-v2x{KeD5ce;L}@;r3-&9!Rw!WkfkN-6Pn~hTCD4S2O3VyF_GfCr01QZ!MZP
z2dS(57o3ue?Rq1lJ_(yVwvJc4QaT)c3OnDLm@@ZK0QaH6hrPf7erx8CJZuEm6r6I#
zI@ZioiJMyW&Ten#Et$!>lvd2|dd|vB1Q-LxjFyGd0xDo5*1Ksl(+m5=HMGCO)xZeP
zmak@8>?Sx?4*RsSKfbI$Up%#)?4|9%{<PR8+Xlt$wRYZ(IUI6qZ->1W&BGqgtZ%hl
zfr!<sUd5x%yDt&-#oWqtq0#k~jq-KCW};4^NKSy;bzm>18jQLrr!E!T+i}Vc29}Fh
zrz#rk>D^JT?4@<<Yx<Z>siT*6Zw1t=UtNp`ouqU=0RlYyoOTcI$F~u%x6>usjv^PX
zV%sj%H*Vfj^pi{MpTSIm$-9c2i$fP{i)2f})#aF`+|hVhoRe@0RBUR-C@(^eu*I6=
zyRL&b4;;yW@~6~DySlz{@o6dr9LU<?7T=0q_?V!5e9m`YW!t3Fvq39@T~EJ8viE=1
z!M6r2?epGarS-;K9z;A;>o}KbRC>ew1ZVUZ46YsdoE)f>r`G1nUmwpFvIi}9ZR$kQ
z?gX4<5t2Meh}ruiq5V230&W<uLhvFV>q#H|!JPT@tLIDtCrcca@DpA@u7@uI4O7fr
zOoTOyYl~vOC%z%kY=7z+oq~|!wzd%Rm3;k&I5#gOlh8+pAbuX$h4n)+CQ-wimD!th
zGu_s$`W~v^cBLw)jR>c5M6f;La)W1)6UQd<yezROOGy$mHQq)-kWZpQddOKhiF`0O
zAU85Yx-)qhlS+inUNQnJs!+~9Fy8jH=J^8%MizY(6lu3-{<hE{5*ojk(^6aMVds?5
zB%sa;yG!;VlYySoy(6qH^SRUmw8viVCzR`B49`CZ*cc~PS`H<B22(@Pfscx^XcNvb
zDaHA&*5RIxsW~O%sr<#AJekd)yW!SedM!sir{pg(`MmvwegSjKOC4Vd_4*jPZK}q0
zRkNTC-q`H7-Y5Q-i>0VBGIiFm1~0}r(45j-HqgjcdW+@dx5ejW^{ald)L*&DiF3b%
zuhQMRXEfJ#{dmSvb9hLSwcq~FAi;$1E|HHnd7SNsRanx`4@F8_@5hZ?g&V9x-of07
zOx$^*LDL|0s}S%!SC^Aib#iv5<0Xmy*QEBEIL6E)Wko49uLN!CEbfN!H)J7w-qpR9
zCVv{~**LV>rej5}rA8mDSQ?`5@<#NjOzh{;vYE)d{Bx&;Wn#_u&oevJL7exkqyQ_p
z5?LBg`X`v9foW5gq_K@XHy6u0L`@Mv1xr$%F;-<%sP-9nrb5?ZE%P3Rgqm<gr*TEO
zF_x*wk}M|hea&=}AmR2hoDxQ7U(PX$v;cmNWP*Yv*qGrgr^P2cMXViQG!<AD*9^K2
z0Ge8ACLC+7PRSM5k`O%71$;+WPa~h@jrv?4G^tn3$19#5)@)Bz**ZAezH+T^s<j<(
z%bh5jl>~0F%e9l1Y~N{S&RnSXyTn@Yf>2=#DNH2^i~AyDp01i<I?PghkVT11gF;2E
zvN_Oh@7L%<7gWbvhs4)ty1x5Q83MUnoLZ*py1#Sagj<7xYtt__T&}Lf+m7arw(;o`
zO&S!DX8yq|m>0j(1Gf5kE6k;Mk=#-v2Z#GQCjj{F6wm-$2v`W3s%TzF&rU{@$z~)W
zz8;-VmD}y|e_VapV0ZEV(E+~Y9mfn(#TWE^y=5u^ZnjHvrv#0BJ?RbaJJ<7V*DEL3
zf67+=n6LlF@E6mCAoU_YqKGW@d8tIL5AIlGhZtlkXO&oT5#LJNTHVQTnFkhtAj#b$
zf7m_W#SQRG0@J!b(Gfr)q|z&`nLIAr)>I{%(z!X!53xp@`5mRkD-t4l6|Tk);#j?Y
zmL!rWEe2K&Ze@<4B<}FT^=epd_Lg$zo^<BYK<${Bz(9vwq>RDaM;SdbARGmOX(41e
zj~TdRt%R=mDJaU)B0)>|5%8}A92|4l`iZN!*2L2(+#&f-@%kSgUp{9G7aH<pAc_u^
zyUj+)PxTJ&8}rWN1;pr02rTrZW0dO=t6S*K!KgSUe9<Io@h(x9D#IN{Fb^29O81VM
zE;MhWS{o&)^qFF~-{6+D0K*MRc*KkJ-<gygB*b`mIIjgliNkeL8Ng+o^hoii##_)B
z;>x@x$;pSvs7%}@UK&2*8J6{V1)>0qb#^s@_36Bmc>o`s9P?DNW17aDE&@VA@3^c;
z6hjvRBW7k}A@1zo6pr8^iF&v{p;7yd0zz6yp#UZ7QB^>qzc5VLuPN&(LtEbM-lWp(
zg1~=|kB?6U<La(~?Y2txmP3DvU<W&ofB*t$`Uu#bn{E`#xfm~uNI*#C-)8W4evNQ>
zcwbrRdu(y3Kv}X%Rbqi!)$};Z&QL*8JlRGqn|DlHJkIvMmRjLFlQx&(Zzsvw-Q(8o
zi2Lf1?Df~!WLmZbzyhvTbEM$#urce{I_qsEF;8kz%#;lfK|#miQhgwY%e=8z(ZaW^
zzsq68z)9n`O-I)ut50dGA#aO*#5#uNMv&=G1Z>a7afK~Lhk{l8MzO)3fS@W#fvH-L
zd{g<J4LI3{uEfxrDBf5wVL0;O#mG=jAUJf3#1mnN4f=hnrguD>)UqchZw<gd$&G#%
zWZ|a|;RP!5%eJ(V?9?W2%0(kyd6lKMZ~WFREc>W!q^if7!9|G^8)iinsgm>|(eN&=
z=X@Z|2A%iwBcdRQON9!-)9_Zps6gPnrDY0}`9MAI5nCJHxA0M0-Nf~NeDJ@yPuT(c
z5zO0-nOQj_#t(!^p3+%4))+^P_Yq)U#vF-0#ewf%8;Mq-Rb^#ISZuAe)DHeN1%^fE
z6C-cz6$#^sf55ymIRvbm*q7?aaEBJN*}d6nSy6>>kT4^u;F+|YOW!Bn_NZdca)lZa
zI+t9f?(G?ytVRhZcmzA!7LWBU$O96CdVrl(?Ysim)}L`6F*2I}b5H3q9T&X4PR?ut
zv}CzMDKf>$s(}Sg=B?lP>RCelTqTZD;7fDB$}(m??r6Rvk%9N*!QG>tb;CQQMOy;&
z!fY~}mENBv_vQ@OLs&}$on7!<{pXqV3+Eu*Fm6Qd6Ytr^(v2QnwWdcNu5SwYVk=*v
zBA{epaJn~%FVUE32T7M5MPJpoKLam;x4BC9$I!&@L!ywbxqKuel^`J<m*r7A`WDPJ
zusReL+E&T&s{|2$;-Z4^s!l~8t1`CuXuknP+p}f6&sPREr%zP_um5Wup19c$f3B0Z
zvmr|k<yQe~#>7wG4;k2%ts(EMv|Sy6_2U6STKPbCo1?woqXD<r8ze`JXJ>&XcCEir
z&`iYHwP8R!&-02y4ZQS~SB|=I`kFiyAEOELuaEQ3CDh!c@@zdem|~)#x4*$+e$m1%
zArVNrPKvVX-*1sAe{vm??jJvte(G7pij~kH`c6Dxmk$P_J_Ox+py2xTh{J~^u_@UN
z6M*qpn7W;8k&<i2k@+-hdo2<uSsa2}swx`_|BNc=)22d@KtzoKTq2;+sseB>O@3E+
zhzFMijWqd##|JMD-zV1+%rZ1+Urma4+@4zdVlW=5Fl0*1O#F!lY6XYCTO{T<U8G2R
z$-}_eJ^t_k2bAcBFiE^BMXAT+8Pt;NYiT?u9(VaD>t>y1CV}pt`SSx{OA>qbCF6PU
zcGH4%O^x5#1w%>a&2B^a9Q!<m`zSb>=HvDdNO%f@v?u*zXhO(O?#m|v@0HS!s4kMg
z?8+s+{1^ls2T`hCzIz1!nxBMAnYP@w?^VAWegroCLw-aPQVhWM$0sx2J?Z44Q^!|&
zEmp>}oi!8eSMx3Q56G{9j&r<|qmwv)3jaVbuxJ`Zr;~uZ!_VbfMp)r}A@=^+%4Opb
z4wFrUiub-(UQyQbi?e|7Mm}E%>^={tXp%Ms<;wo({_b#!_Si$^ME?A=MboJQ#;mFV
z-OT=iH_>1Cs4%4MNv=P3w8ByZlozR@o}v^AFb$uPg4MQY@gfB9cU8!cc9J?Lyq_3J
zm7@M8Wy{v9MVvpm!>@#M+mg4seS^hE`%6;y7jvhZnslO%?^*5a5C_pv#N5!SY-{9X
zjCbZTL?NVco{8c;K~wZ_D~{W`m%+n72`?N*g@_AYrr2mP`eolARb<r#^TSav9Yzyt
zzt<WcqD=l47(yQd#IN)w#J&sUmu^q5{*bS&F|<8cxSGA*1Wt3Vr(U=DUQSHKvo~F=
z%et5Ztl#kiZD=hQwAYb}50#Y@@7}w2wa<Rtv~Z2TnvIJ)idQ^a<rTT|Jen@F^CE0q
zg_UTU$j{BDKmVW-7Q8Xjzq;{Xx_HNQL3$G?E(%8&x%U_VrXu0sgX2f}cCJ8-+yYou
z#LkHI3^7%eZ%g5@#%1DGG;VaHO+I*R1T{0y*c-Fs=p_tx4=gpsb=)sF_2hwV>rd<t
zeo~gboF2Iv`!{-*=|W3Kk#OX4H5Pxh|K+&tVEft<ceRTu5MzdcdvN+>E>dTwyR32V
zC9y$*p1|;NM2XKmIXS=0nGtLtUeNq%;B5W))B`k<H5H+4A$53#U&?0Q!qrkPrTv4j
zmX!C(0feeq_xgiFf_VdBx==9h%C}6>#L#2Z?_2-gR}bM4f}Dt@#^lU&;A^pIubVe2
z6;$&}%!<Bwn(|7;!iDbUWD-(bS!l$2hStZP0qnu`KO59ccJ>rfEREl+3u#6}6hv+m
zq#!A1R0nJTiccg1r9z}D=BK|s6odDB?M5=|J<geURQVy?a26{eOQc4)hU-^!<F)IQ
zwH_dP7KabQxqv~Kg&^E+-e$iGa4!3W@MOYHRXz(qH1+hHky$xPvE7QiK|x@YHDm&!
zy6UO0?w!)i9PLdOm6?>~%1qVj@WlgrWD1gxEIf)bU%MUll#{cSge&Cj+XT6khPS2M
zJg_LpSM&8MKZmmW+N%6+l-eB3szUEK@@Nra!s#jMp1K+x=OIxvFnW%V6S{CJ{#pzj
z1BJ%wrR5I=6WP{E@4%oHgBF971~mXy17%-m_Bg@jt-1wyIR$wU_C<nr*CX-ooq0(A
z@y#&iTPRT`(0?^(On_OjL5V0yex%XUAuxP!=z}G4lE)yIgqib`wi;;{bd)bkJ1m4Q
zkJC1b;ktSH^3$~jEg1g3$Am(kO$^7od`o-Zsa2u4qc+L<;U2cFgnPb)089tU8O~$r
zO^)B^<@I+_5P?8zZsEuIi;h#O4-PolM`qGq^7l(IBZ>N9^z>rH5F`jegU|=2w`i;X
zX{mYTKnYPe>$bd?8TT%n7-JI?PHFj=Ec#6cg#NraFp?6sk|x@d5X!wnf&816#upQl
z7KL@dQ1J2C_g2;D^oIWeP}@WKdrK{Q$5|~I>^X*pRy=R=pBy<I3RibX>c#gu%P7vQ
zWfCRBziMvGkVbOVP$_s03(cqy5xsA8Fv3*{S%pR}R*!0QSiJP=cr#pr91@!jOZqvE
z>Ag%%<(@Vd=#2P_Jt1$&PaceD=gwz@KH}`!IKS-3b<R&1>|?S5gv80Xtyxt0Q<*Y)
z!P8#6+OA9RK@sh74Jch_zuur-2tEpeuT5<0d%i&<h==2`XQNv)e|M%CAQugVIj8%y
zfLyjDPQ)Z|E8XsDf&GH}=_fl4ci)q+laU<h`A0YNRipTMW~VKhC5%+}f)+F5dXqsa
z7Rg0b6RpH6l^B59MYca(m`cfURwkST(uG8S9U2<WsKaE+FVxP5k^4wWN;>IK_lMQp
zSv-tppda&W#%1)2GYjw3#_D5%?S5*_*81ysk&6S7?11I{XE|6t%VcU2+o(h)dLqw)
zI0S0Y<dsO^X$c<qH)>bsoQU<?4U2C(i*unQ>#eI31!jha<2E~5&8st`-&SwY^^gZB
zaP@dimV7>4#_oL((yt}!4Y2V`+jBRIh%#YDOFeB%drrE-|JCCK#n+nb31^A)_%r45
zp&yc}{FQOL)1GJ(x-a7mDNw7f6qp@W7QAon3B`M0*4q9}n4fl-Uw03DH$NIIr*_`=
z(~|4?18g5<ZigheIiojE0<h<`N7q~C!P`fl%q=$*nvSlblX0}iCrFB<#FJMJq^jC|
zDa1EEj6Oq`Tw<q!mu7)uJK(33484%Tj34Yt@aCZTi7?MXPx+;#gg-kmovux~C7<{M
zB04SD&Br{0qNgMYYuw$83}Q+o)`Ot^9Hr$7kUd7zMUG70(q;kc8t8Y3gcw{z^ypD>
zVjTP*Sg7$QbFyO&eO*9Rz=!*S9ye;5^fo1jw5RGC*-5+7<(&|vzYGtjG@&4ZL!Oan
za#gaay&F2q+Z|mmqK-G6ka5{Qb-Yk5`O&KIS%b%KqK>aBP&9j3>-W~yz{&_f7^4Nu
zo;wR@x!pqiV~kek;9Lxiq8H_ip$O@c;v&}Q0v8-8c)k*6E;p7o-&(yd9ip|mcD0ll
z70zgw?ZFR?j#BOtvF^tPtaxoVrgkjyUT?DVto&s9uff&Rv%0nAG!-Pj;IBICOB`@>
zO)hPOvZJ_f5&eYbKISU5iXj%ENPtcb{_wf@z<Z<`|3t^=XwQrQC&W&h>Tx2>DwB`f
z1i27pwTwL<F1rcuIv{R3>@!}7*vQ(I;>)i-QKLwte-9-}gI>3l(9pTHdNR179StYN
z8<&j+HFVJ0zEQgNPVQ-UpL;6eU0hVZ;(Qcl{Rq!}{w8ci&lo79_`epR09?4W5J*q(
z%+{97+2j7I5lO;P!+58D`%br^ILkuQDdO~nC|Rap^37*?Z-3D$yuj&%g;G?>jj6ot
z9T2la`r{8>t(#t0v(T*|mGL!*=VJm{#tzT3Z-ms8?>3q}qJT?0?}Q00xaRk#zIs0O
z4%#bG$3E8*Kc1gbuCV`S8*phciJYEwX&c*4ZFzD%JhgCu1$J2%7LA}caR2=enj_nG
zZkkUDUqLE(Fa3^AE(HDGJS&~W`WnY}X}_)A>-`)%`TC*vztD?yr`Q>qPGO|e9WeE~
zGsSm=)`u}muYq5hl(V({R5zl^3CKR7K_&uE_&Dn!VW=xyrDzuL*)wn`sf21vIL!;j
zZZz5702%2mCzx$%zrKn6w^g5xf)5BjdJ=#BC~#9-DHZ>s#Dw*cT?%x46YnIIOkY)E
z<LbK-pECc~KVrrou(T|04q75O^Zt~??0e3!9uHLdd!E4c3nY!oW@`rsq*6t?;0c0-
zzN2Pm^=ESIsW0H=<cqtyOfD252Fcv>!PlnuZO~nynnI}OPwj4X5qW;1wymL=?KC-i
z(6^h$rVC(q`Q7So()bNZ{sQaj4|F6}=?H3mC;buTBbUo#{ME4e^<&Y>T5%a;b5li>
z89(J?L|3f?%j2lT8|tcF^VZ7ytZk>Wif2o}D#TA1258c<&nX6-?Ob6$_>@gHmg7P_
zvevZ!HahgIY<f(Ye?M~oYagh#1y=lC7c8V>^xuyKc8Y#}o_CwEr*bq3(B&@+yj`_#
zK9w=`q$MOF%AfP(e?SP|eH_h4gt!{9=*s7fN8~d?{PEf-P)aOp3JgX%bLX#SJ}EAq
zGsMXQbPSWz0wQN&S~ay=rKPD8FDa7)AFFbt4^<T}Nb<xI*~Jz(JK|PyYqD;_Eu-F0
zj@*oXBvG5q$!GH_=4w4+P}9jVd*~n#q3G+1Pd59glF!(l)twa(PsJ_hw4UtW?D~B`
zT|Z}V0FG;(!rxXlI}%7{#jr4#*r=eMu+=jHT?WJy(Sc~V@*>)!W*n~u7tbU5M5DTl
z<}Y%@k$@gwxDp!9jaDb63`D5c`dsTg*}>*rcJz3Uvim}qmsu-X^3$;<VAl9m$~b?w
zrptWh-51J>sEiH^<$WQt9=(3Ap47$A+-(ESh+Oj+YuWc-!Zi@0FT13yNJ10npg=4I
z(T_N1Ey=v4&lq${CZE@?rN5^|ld!K+f(CzR+K)F@W6E|`j$udrwOBG#do;)E!wzZu
zr_D*U)|-D+Q5t*?zAbYwiW{Yxq@f3g`EFtT+41=IntRofF58p02-wtEP5S$W))QOB
z4^tCUq3II0ll@Rb_T|cdix^gKQv)dIlo0;*TpxC8X3Z|U;c(}u*X-Xxvia%OgowPX
zk-6HWL^1-yY${$QOb0_xem7V*aWG>toGwX0<oC}M5G|u{exs5-tI9vW(F<@(J*Lt^
zf$|28XqYRJVL3xj_&bVsp-|$--b{%EFlj^I^sWAMT3y7f?P_Q^UO`?4s0X&@ejb%2
z{-!te$Ja<ItFrwY2|ljscI=@r^Sw&1!U<1%?}H?Dw!MYfIfMQfv)P7VGf^5$4Z>G7
z?M-sJh8<bTcV=N774h5+SzNr6)9w7kh9Q6D`SEfm>IVp4<VRepgw5F5zAyb}<u|^b
zd+CKa(ye|%p0#~3(BDgduYVxHS6_1h^sJ6z5V=pvrMQ0lbP&#Jvnaj-x9$yY$8Vh`
zOZzB-s*hZ~ye?O}Ym#xP{JPmTVc9~SyN!Nhj5aO_IKuP=7K4>#50*19uPa-aSDeY2
zI(4;Ca6o9Y%84?DZcUE{dQ0V;ZaV%p?>7%xNu77A@;bgSzXW=(fB4k9lzCyubHtK;
zvT7!!ancN(!Ry2mFS6PKcYB`UwTqpq>V0rkh1rcv+BE#9<u|xg8BK?DnZj};WBY7s
zYELU#Pr!GeEAv0yx*VQW1sPLS{lS3M?7Qmd`epPpTt*twuBo$MYqX;EJYyl?XpMc=
z+28fw59d4mHNfiC<S+;XuX`_O9FC$GPUZR#IhJ>smuLR3+k9{9$F(bYpzMYXPB?4(
zXsRrB@)0U*6-}>UJ0?P<!&Z&Z5alt@yA9!LXooe`zCb{Ja*m`<Mt@9Y#6*ijkplE`
zmdlyq7NP3!ReEgfs5tUgAIlAc|FU3qbWip9cns{BfU(+}S-oA1D`Kw=khwY(!XQTa
zkx75R)7HRCDl-duXf!=ePQ~#vC_E<-{z;O4e<gii&d%`q!fLAcijQYVe0oAh8NQ-n
zhoTVG#uSLZ)+PScc=){3?m_a}zW67{_VVqN)zC2L%bPLozXZwU)1&1imGi+|$Xzj|
z9uuxr;P?I7$pB^uygrTdEAK~W8f`!CVj?2d27YFv#V6>_)5WZphN9eV488zO3-OAh
zwHWK}2?tH@j<7}TZM&SH59)M?S|<WF0AB!&V9ZO^N@R{AX;1ABPm&o7q>1%^M`ANw
zg%M_|!MF>riXGiD5h6;ChC*Nrl*yBgLLLBGOLfn2_R+4TLdI}`>6E_<k!1|H4VdL2
z!?iqBB3VX8ai^OJmZiQz@e^0}k0fUVNK$vubAl&>r`$LhaldJu^q;16KJ{&q*o;}8
zAPkiaaX6#E2&$aX%_7r|=NV^?D5vM7dXE_Y?aqxm3lB_W<IB7MG?XJy6iC&dC4fUu
z8o$Q0U_f4nt5=s)VcN`TC=nlbJ{Y~ADB7Os?Cc!1;LG1SIp=?!jT_q>m+ocJ<ex6L
zLO*(5_0U=xpQ;U}O3PX9TSBx4^1rK4xH$&I!pzNQ4ThMfvI7sYe6XP=f!m*2%Tk%T
z9<6{(iu;S3n%6s;3%-P1#Vem{VAjE<xz+!TY^A4b{rM&@JMDpqcp4j9Sy?FsyzHhE
zZo7xhs%+Ob_Vgfm5+;66^P8|GaaMZG#3OWRF<I4{l^Y1>b+4x8W`TSP?WsqY22~Ob
z@b$XNw5hEvU~9Y|WK>lO1;9+*o$Q%De+}s}WTXjbS(-eKrgaVsyqcQ^rh<>-O?AwJ
zw$l0BUzPu?6uaygSXVLAyG+$fJUQQ<&9f>TG4(t9Jz<wO5T7IEcDysaJyj8}@;|$y
zlaE|OTa)?CNae&6svG=5+ogJX(m2Q$*kS8_>_Vg4O3y6Vm|Bj%y2>%T*e6U?EVKj^
zmOJFv5X#@+PzX4LV-gr<Ogy_hn8szFEzw?kyL(0<d}rO;!3@D#wBV-zy?0+sR4FXN
zDw-Pi8Iz367tRjEr+^Duz+6B&XJota1L78081%o(tbpwd2O~iam~NgNdwBbYTZM(1
z=Due$RTa&L0zC#FyDfg@j1(sCr}0W`>7C%`qxntzwv49DZv(|LGR4tdC#~ld_=T2*
zadFJoZ6jiikkpCWWnw%j66r6KU{A{1I66}vVtb*c1th}f*EA7OlP@Q?qKN%FNa<gr
zv>nKXfbuvd60Z98G)uHb%$hEzZ$<rLdwmjy$Kin%BRw?XDce!i*=YS6hLyEug7BHx
zeIibso*O#9xd5Jb{VCeUQ}#DhxN1{s>8)s@9Gpzhi5<-~d|Z+flhI?!4V6I$mO%CA
z+sc1;mt8B;<HWnA`G~RJrmu?pOmQ}wf$dxd%dn?o1;sl#QeJx<yR;#c4@fCFmJPK;
zZz;XAq@j$k4jD?DYp^eSwum8rkh&-US9xp!wXzN+T>!MfKWm*?v_cM0L{?oa@rZ)M
zZn=5aWO!$))`HiA0?)0X(~hl~Z+neUje-}M^71AucTdAs=i+D=fUDDr7^H1a_{O`m
zBq|DY|7@+gsuRxtGI7!3!2Q)!i`i{aN({r#w4@Z`L^st?0^@6$q7rwA9L9ExF$jeq
zVGWA|vU{F^)T%zWhbYyKl#q=F<QVt+;U;FzQG9SWP7Q^BJ(5;a8p#f$&x)wc)|1rR
zVWT=qD&a9t-_&BV`X|rRN3Aww+J1PI(iUzDKE)%eKQh9}g<)Y_-!LUf>i5IZ{D_Wk
z%NFvlIm7#X*-V&A=bL0lQoJ!o=Rr1H^aC0IFaF~*iu;SvQ3!lkg5{>L`HYVh+TpU*
z>jU+Km+3UCkaWag!0rKU3TLtncqDJ;qsrk3|B_;x1TV<T%Ue<{OsQ-Jhvn6=agX;m
zmRdldM{+8-g-$lIw%Fe*^f{qrmv+6r#9xo@uM`5u%Yp!*4$yZ*9|#5bUW-330v@^}
zGL8m9Z`{4hMswa`26##-RPu*HjkZrgf~|?`=(mS1LKKM7e;rrOddJ>A$HFG{q8Tq*
zsxrl`h8$}lf00odgS9s)LSZB5VBc%Z%+>%F>ujrmjz+XLO8jYQTA4lNm=aRv*fepJ
z_qi2}uLDIL`oP(i=G77i-;M!l6+AYh{pKlLySL8j%d3Q*8aif5VT}!CrPBV#2^w1C
zEinV6*sdmMgI%xwi%{3UC(9P^<4xOuq32;X(HW`KU}t+VH96^hZ+FeF94hhRZd(8z
zYcIM;f{u>#!3SMmpS1xeC&6~FL~ivy%a}DzS$345?hs+3=3+Y};{{%1*Y7Vi6>2);
z<^3Nj$QhX$8_Kwt1G^J~@uvGNqj_0bO6p1Ku&KR+Zb9DK_@J}?IBoTpHilcLfqN~A
zQc{nVZ2!cu;)7O7%+FUlSrwKXN8{{{K-{nsZjP57pttIyVaPlx9<Z0s89DJeDGt{B
z?5H)erw}dzTpBN21IdaluKUYdGXsN)0YyUoD*KGP7Jr(2u7lB=|K(FV(X2GWUt-9e
zS34{(y1N9y5K*wNXSF7b+v?H`L-{GwkDmns{q2qs08zoLc`RX!VT+lfZjgN4zY1Mn
z(1uq{B5md$`_t-Idx-3_5IpbH;hswLNxcb~N1+}aki;m!tw(BwDoWC8w&pnnQ9VyW
zyOi5KhVCeJeB^imHGjyxa_5{@Uz?znfBVU}SCbM0l`*5w^zryP*_pNq*xRy&)5Vp|
z(zQPJHuue$Wc`#R{#>XWw}aUGqU}_&QNfNu&XW$YulXsHGb1f9N^u=~r)sgxpUr*W
zpepK|F_aAe^xh3IRT7SRjJv9RMns*{f8+Hq)qFlF9oX2<;bHt$4{&MTYJGivy^psW
zdqdUg&{*qL;;?-l8QJpIb$y86=mStQ=jW?vK;0gj3mjY=FLhvZF3uGj4*oI72!`u+
z!HAwNLQME>(unI|aL+%2Qg_Bsicj9~5bjtWx7WZ{BN!GCcs<)d|DX4Bs%#>!VY&PS
z{tiAH?er128!{cqnirRd6gKX4d}GzIvGo)G;n$16Ad(JsrvYq;cq;NgMs6fAN;UMS
zWBwNnp7ok_`01wS8TMTAMV9<s(kK%Wnk<KRp+2{W2Hqy<JuScK+Mln1cpZZvB}%Zg
zh-w8K!nUOzR|rMB7Tr8BwJWoMXusk7)ieIiC}&3ap=mGCV6iT^=fvcD!>#?``PL@(
zWerf(uBg<O0c&YaY5enSH(?0J68#0F{qf(6GRXx)!(FgPH?UQ3zM2=$Ts6J4bd{TK
zdfoG>LjEX)gfj8^$oyj07x96n?Zmi^zaEeR)-yRy!xYbFcUL+^bDrw6-(!<Pqpc2|
z3f6M7N4EX9yC`$yI3GkC`F$k%om<0kli2F<rI>ya;#N~Y6N#gj)#c(MgtFL-pJXcL
zDGJ5%x-!i|MnND}NU5UcZN~9Z*t4l}`5NFyQ=7-gkCbMxAO3Zg4|~44H^G}zL<5NE
zHTwFW<6PQqTfQu?K$c}m$vVtU?(6tXj|{#xyJ^h6r#RPi?W)PVHa22<GO2L=AGJ%%
z<*}X5UTH;{@vOILQ)TVU!#J^Nv76w`C&i`#F+F*_f1>EyR<!M{#R@`-qv2Bc^{RA<
zvp>gDS+<n_pC$NsB&;(X2(%hZ@=i@j4puIB#jSR+90i2IRvabzD>Wz*ZERZ&q;ksT
zkH#^3_i<0>Jw*bKN@~ka?#aDm3;OX~UoiiZ(AB<gY0ALuc#LA;zrSEStKr1nMK0@d
zd*K2*vJim(Yw^jleM4G3eRR-PJ8?}Re}7e#ze8$>`DkhMV(v8738?b*y}pI}PWlw}
zxG(kd3*2@_?}3TppXssF#pn`EFz4grB!3<1eyDgc+ZkpGHz@E*v}+g*QFEtcOnns^
z%@{^cPazGtp`Aojz(`ET0$v@6Zdjgfjxl%=LVG}XLP+rqAa)~pEvh}@0PYH}Y_>2l
z*rtpi5Jg48vQB_<uiDrYRn>y(ffzHbV~y<lNwiPicEV`k6juJ76P7=&{*8&0ma>bS
zIXG+^D{E2me+%Cn-Rr<)tr|ao&;e&N#vv$oxAjAC(?D=|%!rH{y-Ch;qu0>0b;3w2
zgwYX=5D_fXYe@@W<>T(i@265J6-|Fd&ktAWN!PWd<N{Vr1azFU6t4e`O=+6f2QD8@
z&!u$-noZs9tK3ZUFcC~f&iygC2DchS0v*ie0Vf^vvcrxsbPR69GePy|4sB-+hPK*N
zp-+H^_2<*i7ranud7Y<Lnx?*|g(Aa+glF-`DVX9bNHmj|RTld#eI$+H>&ntJ%zEci
z0(C(uC!k8AT211u(qRzA$iOCBTO=+VqJ0no%M7X}{9Sa_SO19Xv423q!HsE<Xqark
z9OUwWpl5|XQL+${D5#95PPCIFGfi@WN)a+6#)Bw@5xLDKR!(w%Se`bNmsKX;Z@yL|
z1JtgK|2l1a!|YUL&qPSkHoRGhP1w%d9!d!UEJG=U>sL)6Uv_r_Wir^};(mvXa%gSM
z@JLs%h-W@{^Es1Rn45C1K>;=rB1j^NvT68$pnEr-juEpk;m@xIFI3jf2ng}EDTNll
zDP#V)K+m_Db!{;WL(;FLr?quWnycX!6vU4&54)Mk_1t{xd1<b(E<tsoq<#`CBCXfJ
z^~LE47{s_$xMG(bpu&$~$o_*xDL^NK?sLDMDB&@!`d*mXy(q)qknq+>r1&ZX5)wkB
z34i2NR+W3Fl<Qlqp~VkvC}d5UB}q|A&cu8{0~Vd&(Z$%Z8~z)~EUeVu@BKXHiHNBW
zT!wwXtic0Ug+v}^r#p^zv*yezdh7%J;Fq(96Ob_awbm=Vi{B0Q-=0hr)c<a<B!H(?
z(R*aA!TzDw<@#dxoc11@qT5!9`6U^B)*mbt!ldP_YiTsGW|FT%dWW6u-mHQ``h-oJ
z`G_sNOK2oES58V!4C%jJ5B1Ad{1Y@@G9V_B4J^qpG<%p%Z2LHwj5|eD{{$M=Zb&?5
z$nz>+N9R)eeN^wT9V^=2(PpDUY1mfb+Ah}Wz@Iqg*3(_^kuo@`L&GGR&#jS2jc=Ls
zFFF%vBw?o5nE<<>ORQ}_YHChr1VwH}%Kz8Sfg(atDrrzgtD=Z{pi7pB{VaL8&!&xm
zM+a8=!B+%HDQe~Nl;v<W;H`2yE3T^f-)bJ~EHB1d`T8PX&@kJ_E9<iepI2u(1Vk2T
zSlhTnESsoiSw56eNATwv&_r>is-`*qKoXzZHQghMO{-5~45ZQW5DwpHd_l2)YC8fW
zLKKBiT7FMe;_T6};8WVWL3qdH<QYX=YWZ_Wti$>vPU#&74?fKjc#NjbUVD!i<$idX
z-G?Ga6`5_nv2se$@60WQs*2iYylG1FZGyzW&-y(_YhJ0P(;uIw{%+iERBkc2dz8YV
z5&dM~Ae80)JEnfymv4$nokFX$?YnK1w6&T1vMoWT$pR|>o0zbBw^#N_on|(eur>WQ
zLn>QKzp-OAQp<D<6i8~5_dhdQr*Gr)<_<CcR>cdBpQa;0&E{wYZr&$&|I_5F$VtO~
zKjQBLTXPYErcO-}+gqI{Sl^YbsRiHV0daimNr%YgL9K=F`Eg6sT$-hVF=GQIKPp1d
zA+OPCKx_rA_7&1%-uROqBt$y2)p|Qtw|=$}IQ3Rwm30A1g8NSl3>mfiV4}Y9luhL~
zEE+aR=TFNiZd1WYKZPfXld-+y4do)mg(kmxK5t;mx^U-3rrHYuk{AdZXN7wgq%TF{
z)xr0!t)gUWft<k&Umc&6T!`vpKRt>EnLZK-(_`O*kf9CY@ULG;w<?vUJPSv%y;^rm
zw?7++PHrCo$SLkfq!@n<?K`z3l8BEP4kgr)L(k9>-5lTg#TY^#3H5HwexxV1DqU<Q
zwDS4)lXCKvWGA@GW!-${U}!d8cH7dBm8*R5;*=T$m6_QxLW2Kw1ei|6k#QqrfTf}D
ziXkbmzxrmugq?hGa82iXTJxzea;(hCLE~?!_Vob}(01^9;o3qXi=v1$j}e*5GVA?V
z?*3g%)1zWr1zNbX^SunV=S;D$-smwIdJ?qzpCH;3W+FfUZj62l5&K`mmUdW4a0RGS
zrcSJ0m8lnAF>$9;^L3#;IgTdSXKXsh61IMK_ZB>he;eDMpCOyvu^wF(ckh9BTv8J2
zA_M^hZf-h^VOlN*21kIF`m_9P_W2fGo~oL+|E`{~KiJxvI~@d@G<7qL`%WGX*RK&U
zl^*7}H4PzIjgqjXuqK}o^JaHKGAqr>`no?#Dgzy$H=_0<tJw#SIX(t+0T^vn&@KJP
zz}Vc_+-27LXBu@`B7wnf1oPn*qo!ZJ9Z*&KtJ_h+1}(^)V5wz&{%j>O{+@jB+P?Gk
z=#v%Rxa;4a%zaPhr*113DUA+hG$x|$3UmdGg}|$u_B^b78zg4*c}e_xU6Etdd^uVu
zuk&zB+Kw>@QtLU)3Hmp}G`YedZ)6V_U*yOH;i^FFLv@3fr2^IthUgly4!4v5$~>Ic
zlRHiJ{eyiE{?h|@T^~8;+k&oR>ElK7#krr@t=S*_v1DtS&dmY6efk1$cR1|6rDIV#
zl{)4FG^#H`SGy+#W);o{Gq>IR*hcS$k*4IZR3IZ`L%it@e?{rJ?Q83~GWVZYQcOhc
zC#|k;G*gC5$>{Q0)8PhE`hs78$slGXSM>SV-Mc#_SQX%5UknkY;n6v0diiime0Xcq
z7a`9<gNjjl`>Ia;2#)(!vmEM1q+Bk2A8n@A6EReYzVS@IH|^U{qxZc$$#gBGR*Vs<
z%tmW9@|jm|eQPV$b2s-V@&sd6U)|U`@9B2({Z_Ogdz8n=zn83a4$uArR$otoX%9$#
zKOu!P_=!LVE%yA7we9av)_Gl^t4(>iS-{@!wOEW~G-n9Zge)h!{&<L-Jvh8{A`~0A
znQu0Ef3>q?t6ztgWifiC(0Sgkx~eMYcK&3bM46QuYO}M`utNViF805^jgAJT4P5LD
zxrEWIoX5W9#7I`zY&iq)n`@F|^axk+4CH7IHcVr&W)yCS^qxL4x|p|Vo9HOmiitg0
zB1PjbDMd}p9c$$Zc1{@;RfJMu9LejXZ`EioH4G=C00ULTB+~u$K-^o`Y7db?>wDA-
zU;0qGU(gV6%;4mY_!@s##q0}AS00LVV|Z%46C#dayG5VqaRZ3aq8y$4UHhOX945`&
z;T>B4F|~_?l7fNz5F1zg&f}H%G{(+vqNG>>QEMQSg6LfO<u}fwt{RnQ`J0D9fSPiq
zcKgUb^JmIS>p&;3-Df%SzIzS5;TD9=dOqsytb~Jzioou)N>d>uBM7^n0p|etxc3uZ
zwpvbSSg$840)a?=H4>pEM+pMF^pK2vCC`eL6+6iWVJO)y;ESuetKSu}(B|V36vPm6
zUkSq0jK4|EPbzLAipYAg+wnw;QHc>QjcfJ+1w1W2CuXS`81LEM56XCxsr3d4m0@Ga
zbd7+9=_cT0@^JBY1U0IJ0MeK`*)MTNM>Vm9ny+O6Nn&oHnoe{`RP6BXSTDx(vNz|#
z)2t3BVwx_wYi8#Ho(K9tysZ4GuWXK|6E8H=snBcQ-1NagphUX*M{zQcCCY~l-Yydf
z+Eix`a$j+kdxSbv?}RINXlBW|ogMb4{02efp9_9w$xBJ4y(m<wwtvgGJk&tFvj5HF
zraSOUInj>ge~}ovL}<cG2BeD=h~ABS_Kb1PqP{$J0IhA5V*7)wk*}<ou<I!NZ7nkZ
zUseNx@(!)teI3sYfFBBX4mGo+YdrAcDs$5x7-A8Pho`=_(kU_gxBVnGXg~0X=WH}=
zPX%A?&4vYoGNjtJy5vwnY3Sm>fNP}z;8_JWBnG^+js)Ocg>}yNj18xQ4*$kZZ#>n=
zj1G(@NTgzly@UUFO6WiCfhuF9wNsG9bj)l-EOv=6QD19_9`~d(Cc(R*ELKJioG64l
zM%t>&mjubVxcZd$+hYq9t45-S`=o8l(NWoM3wd47uGG(vIjffTG>E*N2^9u_G&_7Q
zFf_}ttC~z2LdQoBU^zqtdeUhzXA&~k_PBBf2P*h$cZr8J_mR1L`M=+b$E0iZ^L<eb
z=jTBpDV22VOp!y?*$waM1PN4#L@0>{-{>C+qSA5~*A6Gu*<Q)1I*j{-Y^HZrfdlJE
zf_P7c?N7N8;>ux$Ui`hlG!B3Vg?^x1x=C$ViS1c0Jn>O+$RL+(3g2Iiehy;3Ae}b(
z8T{SSib3l`PWbJ8T|uvl)&1Ezy!^=pHOhwMUE0M=U29FS<9U@a`$FUCk}nYaF#9H1
zSjytn<^0_DSJY^%_oPgTSMb4MIz!<LvW2^GwAz)I&BTw()&_noJOT)2F748>2>$$o
zT)8{!m>!+f-uX*;O;D{PhQ(>5_}x+>34OiVPuC4aehL*5c*6FS*N_X{9{v)=_a?<6
zR0K5M1StJ7eD_)RR!RNOKMOab(-tR<+=-&L?6uBzC(4F(aNqg*3;iYHTN90Ls8{mI
zR|ZHbnz9^9)J+ant^%2-i&WyCG=|&Z21tpY<z==Ec>yxf5!_SU$!gnyA3^%R9<gni
zh#<Yuon&d8{ZCj~X+REg*3f}Lv?t;O3rohGCY`mzT;URuk_<dtDd_7AfUsZ!{`Q{(
zXTb5RftHpQ^-nss#dX-=1?#|-KQfm7OA*_^6%Ghq13|&!b}u|aBJ0>GZ31WI|M=OX
zi*Y&{uhmlc27fp5sPIOJ45A<!@wfE#C)NJ5l)!_F2=1<wO`)D2wY7~yX$)GgRLiXG
zcJ1n1TuNOO=KNmjDbEjaaDN&160Y#q1JB}B!CLaCyDn{qIoIp?Iae!dyxIOz*`~qA
zrwYB!%~Qb2OTHU}Ki>ap{C+U1B4J|w=U++N)po_@zru4#mlH{ela?mmHPqC@mClym
ztE1;)Bc@ZrG9`n#Ma;_C6JB?%U?2Pk$@epIf{G7<r5T|@My&rw(^&>I{eFFT^bmw$
zzzBgMDM%?uNlZ#gLSmx@esoEPe2Edt5S5nE3?wEwkuDjfqlbXBh=72AbUpjOc(&Jj
zv+dr_ea`1x*Q4$|>7jjpg`04(X{3m?SxA?ptsYxqk57G9G1tF}Pd4fmEuOuBEk8$e
zMX*5X+_94sKgA>U)fgiJ)MVvkYx4<D``P!TWMw<*kr1LhUn~?WjoHJ78j1+unchwl
zcxLW+zt|nQSR_{%O`z{Ds!e{6PJhbC`s5i4AwbC>!D?lT)zhNIG@-n+BSJiT?|-s8
z#7a-EO`a_}bS7r})1C3~^pd^ffzYAMaKL$WaXCZc^iyNVMz_R&MMHYxsp5lfFZRCp
zIE~@2V;LE4M7vvx>D+?9(cy;+(C|@6imY6dZ^n>4sn4!`c?L$`bM=Q)6CGY9@Z&*W
z`sgFC`|vt;Jo7>G-fJq~rMdR<PgjcV#m|(61J>rZl#*$_uX*2pOyl#_eo1NcE}8;J
zz~d(xlPH7-r76EYfAnUWDAGTJc&Pk+@7w*RkDkpXdkxzvdeo*>mF1PLf=G$dK5kxL
z9nhB~RrDu(Voz752O@I7-{8w<NXHku;UHpTLhDEqiI$#Dw8zEM<U?(Z2tSjspTLV%
zA*~cbV0JHAi4avijDcVwB%uw;Yl{Gz<Mqlo`C>vmIXu#uIollO9N;kFGInfsB`R}R
zK=4bQZPUfyi$|^su%9C%D29;zX_?g`Ll7uwFM}+{3^_QGIG=8qSz9VfR*v|)&^faB
zc|**BKlZ}_-z3sKZKZK^%H{t0bYtCUpaTqzpPgPW8NuA+j*%EFCr3`pQGhYLY-GUA
z=-_&aRc5G^0Fl}psUvWiWqX$<V{}?w_aF6%<H(RuN!6RxFPTL2H7?i{5#zs^h0GRL
zMgNgbUb~Awzl(dl(A>DJy@#-F2&@k<ew?JoF}C_%>P>&e0-In3nrzAm6T27s>F{^a
zq4HDXO3%vu_!k{0T;klds2c$7by-{h6xI#7Cx8|UGp+id`N-YV8TYhAh0AxRF-}CT
z3SG7#h;q!~WQ^Ca@qyJ9BF)xuc(hbmYTW_(QhKyT-V>h2?|~je{H?QBHLBLzj|nsL
zXQV1)WAggxZq?a(1LXGez+ZpLob$p?BrbQWE*6+C^1oI*unl2&3fwNfB5tgM85tOG
zk<F2~>YlX~QiWW-ULsIXQ1`9SlQc<!T)!=?>DO6LLh2Ux$;s?@A>Zj-zmBj`QjmLN
zT!KR77o)lLb8l09;g1Zkeo?K(S(F>TRAUZY?CoD){iuAJru9H2RVvf|iT1~Odt;xX
zED`et1Lf&F_2Pxu6u>+$c)wTWwU(iW2hQ80>G-TD**oW9$TR$Wo_;TJx|aNP5tEKn
zvb9uHP&oTH_JlkQ*f=6B28XZzk&XN0_3I6v1I7gTRZx>WYW(@eczhaK#Kypp6jzis
zVl+!wnCba5+OtR2oafqMq=W}+H<}>Pn;bX+3dsV{=YXvdpmPLUYku>;@M^#1Zst$P
zdB2;s{w_pH*&Gbu@ku+uEkDRFVS?%XN?9COpFWZDww@#*hl}Ew&CJcnyTBI<TU5l@
z<d?cwJx%q*k`*gGwmstMcK9mWeCX=z&YVrjQ^Y}LehgL6($D_$b1e}hPvycjlcu1c
zc*o1K5e{vj8(gwn6w!WC_~ka=J$ax0SWUsGmmlk{o#%grc<R`{6ukXR<n24YIBM-{
z!HjRKCy;k_cl$CI_(CPpesV!XiiOz|r6?UhV6<qF3HsF_wbhlS=&FxTIlKK@(^87|
z9Ykk@yyio_KLxelMvsVy^NmT_mjz@wp`)(7yI0c=0QYnB5d99-XvrZobEmMzYkM%}
z!<||Mg_IRVy0P;eeod>dY}rL~4jJO1mVP?X<Q%Hha}k;hsFfwp3er5*<}1N;<n}XI
z;Zp_Ca#A43LvmX9q)4Fb?h(`|PF3$G@relBzt7N<Td`4(CS>LytVp9dLG8QjwLwWf
z-VU~ZfI&NHdAsGj|H!T?!xTmc0xm8Zbu%<8UJ&>rRZ&Vqw+M4~pPOoIhE_9;oi~@C
z*hrLf*M1Pt6|NuFpLL2IVC2F5%WVnV@9No;B9yq|4nuM?LPkj6{G}1&0ZXa-<tSHA
zPyf$V&+jIEqbsua@wB-k(ifN`+xSLd**IGX=;2LzduOBaFp!0u-0ieLxahE;em1Gj
z(~Ap-`YXq(etDrNwE_w5j7>%QX)-?cAzD1Y%fc7R&vwJ!PLJIIYiG>oY_$U8HIdB1
zH`s=Y@I>s6Ge}^5DEB8oGpvMnMFh%eABAlFA=M-ee?1HeDjB0E^s(R2q05yvvE$<d
zd}NP)3pmkVsF3C=CTnjG?lJJxY)<vg`kjBvj_cPGO#GzH>S=oFYEJbvpGK`io0&3H
z8TF6PE<X-LF`5E$E#WC<4@j1zLa!MGnPzO>F&laMR&TKW!QM$&^1SnDb*LQIZ>swb
zh-}}5_((Q2B;fo)-Az<IANQX4D`vtJAUsq-E2;P}se*(ZjMJdY*n`Q#aGf$K!d|qK
zU_E9>mSEv5?__G~igOzC;kKyonr(XtS^A-<tkA#rXQ7`=2la08Pdcd=FUD`p2F*WI
zmb+lSyr`NBTRjg;-m-44L5#2D&kxqI(|RdqvoAv(a{m_JQOq$I%p7rf#`_v5QlAL8
z(StK9#6Q({K$Co2S*q5XN^WQwS=>aJnnsLE6Il3`hU1B5PD~=*neIhdTJ_~Da9^nX
z0;WdC&2%s_(cs1+Lq@m<dEy==Bc*Ax5Z0PY=^;J%NJU9r`47%>i<#$%X`pL$>!&)!
zEXN#!cQ>7;T+Grs<TT3yn2^A2Vbg*8`j14eMt2-O6mb%A&EZxO1n)P^@(p5qsSB`t
zuOZR&l+Z`0CAr*h|CN@Ou6B<kR<+|V|76vj9N@A!EfoXOe@%7CPkJ>}lrLc{iYf!r
zM$w}e0}D%j5(f$+&Uq2F_(n}|u3m}S4cqUn_w$IAq=D}OeEK3cHAVnO>CRNEu*KLW
z7Xzk)p?EF#7@LZl`a&sFO05?IgF#zAXpN0^25YUN&omLHq>~2`dk?nf8dt7EJu0up
zFZcVmNJ9;ywR(kcbG+7QQxkdgdn+bdPT;8b(#NRJt6>G5h1P_QRXLx~K2oR06Jm|f
zQW_{qGzbLdL$yms0Diir7|UNyR=S($9<aPSJ{@JJLodk+UB*NWt5Ik8?V*hrK~~eh
z)gq-d!4gtWs3~BE5a}Vd82o<cG>JgOd$*kAB+X7TZXV6<e`t{X{$uug<GAHaLbIq*
zbN54_#sZUzd9EWI7ovbSLt-H)XcWJzsS^dpvj-R0ya~`;T#UR8gqVD4Q8Z!1rBfrL
zZ)q~|B|>~ieF~x#?4{J71?m%26F{n)M7L3HVVaoy*-adc(E~T!k-a%Jy?V6aJjhXO
z0q)-<sy6zTRa(NOXVdJ)Y3_sEl5SVz;{`559m|cjl4}n?hx*FUC%E3h7W2fWAWn`~
zAK)lb!2pWvVj5(?B^jTP_0}u7llSkqHe@$xk_8~;O@Y%TDOqm*CKV-818%<-Ki@Kr
zYJ9T9FH5BMO%O)ReE%`mI&8Zs&j@AH5J@?(^K__~FnHDSYWa)2D)@S?mI$6m32t04
zYQC$)4Er4t&6`Oa(|Aw0knX{l0PWy1Awpx#fI+?o<$+Bkr;kpLBsIatI=lpue3_}~
z<g^>L$henO?z)E2gCw(eQdcD5Xxhy!JLjO&U3}Wax7OB}qV6J|v;FqV->WKtI`$wg
zkkfnTvE#*^MgO-YP6g~qBy*&@sPK&|jrD!<4%5PLkyUO7MnyD;mN=nrYR0fNaxtI-
zD6UZ({Rz3a#;GbSL-6U0;muDq(u|Qz)v?!0r+D$WaS20J{ZnNM_e~$v(XEv%si~!<
z)!VCuu3L0TU{RKnJNdiH6LC7j!|Ry#F4K-ckNro65++k0UhcPF=(4<tsA7v{Avjjo
z&S^hlKa7cle7`SHuPQrg0b;3Tl5=4PRv%xx#r4P1eaTD@JfgFqIf4I@NhTXy{{>PN
z4^J>Goe3bCng-af@)dri{K@N?iu^iOXz_2_6U%YMPSlN=^ngO0yjY*zelf9J8a6Xz
zEu+V~Xr44s+vT8?qyL(T83LM8x^6Q*Kdq0wv%z5z6#o3VEQfpotJ6J%qb6r3&}vp;
zNeU2f7fguxl*Yf>Z8Ua&&4b+Cv->kORtl#^`8qMGQadc2a>FuBp4H!ZRVZAMgxh4Z
z4Gq~|EaJ12b5MfS9;2^)YQx7sE8b5d+>22V-sla@q*W2;Inv3%GX>Y)oso-(wmwr6
zHVp-?T<=W}1UN}XS5P7HZ@3cBU%Q)@Ve_8xNlsa}Hu+4i*J2EeD{xZ3_J)A-r3YSi
zgfu)!L1g9F2TgxYn(>}*dNJlgrkR<Tw%R1ywH*F@>E&qsQ|n1y+rc^~NA1r2!xmCd
z+g@khWg@Ue)A@Wg50fXO13ZkfIkKskVCqW$+x1~0DnXLS66a(sw*jDQNF2W->t0`b
zR&4XC06z)_y<(~MYD^w6?|ZgT8N<X2qVFQfqv_!ZkZ37Y@Uv)<WfG+hy(Lnc-6@9x
z1l~HF46JJ3CGJnEL`Imy4L<)3&jZ_tKFq~-k2osj_=*b)q74^7YKHwB@(T=AeXPqP
zePYZ|%~LwM<$iTbV7YP;9l~8jeqW^pbnoh7?|{%WX08)YHFAcEE1hORrBPHwwSgAl
zElCz1^3#|b(I>?K+of=1f;dFQWFAv2co>UiN_>Fmh)~}u>O1_>Kh<AZ$>l)oDtSwM
zVLi71lK$@23+W21+B-P)zE^O-J)gxSR9-WU?07?!X$Wu~`(G|*?7O^6s1^|{)k-G5
zPEHe}zXx8Q>VG3e0p=g3(8hHJV54XmpHV#i(()`_&|Vv2v4)@Vy%8T?ov!}!gDEcO
zO4l23!Cg1JIsIYT-@?ELUWcHPfkHWMJ1)rA%1WxJX=T@V6rU$9>bpbbW|*itzg$Ad
z_g@hV#!Eou9p^_mLy_(vfAGCXc78?$RbEwfu!J4v9+7L@LKdchxrW8@;G0F=&vOzt
z%xygaLeKGbKK-^+n_GE*9<~7Sphr15dat+6PlV$0&SxFBZ4fs!ijpYXcm!~=Y;evT
zHd(iQM#3n^YxbL|&lzD?&Rh<6{W!DPQi>nNlJ_65U;RL?o*2I*;{-`}7D4nW%ig1<
z_;F^6n;`Pm=a&vSeQ4YbHO7i2E9+ZIEr<F2E1Q-<0&ij%DiW2Jhfg~%PaDF*Y+H5?
z7zm&b)h4Q@@8F6SYZ0fdhfi^g(yKzf+@<7n?!w<-r-19v1o2%*70T&a55>EZzA>O`
z*8s5z5;|^Gjdxc`QBElzW7I{}PIK3I)CKo&$UWdBGpuf^()Inh^Ant@)i4=`65Ph~
z<t~{zjp0Q{zAgWdUF-F}vK$-W-LexZFK=GA*gB+qzEqdZA+1Wa;?ijGp&ZoY^-3&9
z;=i#TF&u3L8?t{P{J!IjPn3ClXO7#hra*-bA`+~M3Edf(W_1!<^S`COS~NwB)kUNu
zDhjB$*+g5FsXz6xwRTPy_GLV{Lk;RgU9(DeDjnMx{nK;z|HB0TXcH?xN38I5Te#oU
zf(cl--Ft69$I1EQ5Uh!j0$CM3MLbYcY&LqQ_k{U{CT;6ne!BwalRAGUG(c-G-}b`Y
zJ*d}S{WCAd*$JBzwv%R__b`0H=%?N-+Lp;pFWb+lY!iROU9f1m(+LS6<a{L~mdA}j
zaO~ePo0nBfbx3&is>!dAXa@VJm=5yR@}`IP_V%_rPIy%1c<0I78L6RaKG^fL3kw{2
zU-Bf|gQH^9EeB&G|G86^0+OBh+tY@02~TNFn2m^(Vw7KLG$JWg-)p*IGrWi-%E;=N
zKFaDZQf|0e={Pko8jqy%d=rx@=^ho34sk=rGQmJNDb~-0!o{gX3sae?zaGk5283mu
z^oYI~{vU16@y)(iBR$gqd6n#B13@9Jy^#=v<M~|vm}>*lla2;BY<e8lvj=y-OC9OU
zf#*|U9MqJZ=E})sUj@@X7Nd@h+P0)NFgOrODKL`JqP-2GFf>-&s2=?&;$%%pCsM=u
z`SWawj`Lm(PaMC?V^Eqe_GwzGu^ZO&w*J`f4-*A8@BW?G|Gu7|=#LzvNy>Bxmczj5
zasnVQaOdvfz7iAgI_^5UCiS_KlauF*qLFoCPBRI*z(xJW&X6^6<ap{l?2wBZRa^e?
zvt7{kk1mMt#JK6j(IS-{7g=RhnrdUP&H(F@{1?7ON6FVq6Wa#^vuk3EhnC%ACUQ1$
zl8lPyb0;SXsvjmdSk^-#mX7ZZC8iu)DX^KiF>9mf+7Juy#2PisY^o)U-v7481yY6X
zN2?xDz?9kG)KNS~QRGh%Tbc8A7y&L2qFkg&kk*qI`i@wsjP7v5?eSe#gP?qWtow&m
z7zHKuwtkq!;RTd-ySaZ!?Q{##(P5<{|E8sFikjQF32ofC;t|OkO(#{-1tV=;B{c*l
z2<c2{7XcvZH%3eGSFZg!j=^7bU|IRWOtlyWxxLD@Y;{C&o$nRjDz6iop$5hC#rIET
z4^+2|ERv#hkMi@8ztL(@Y+j`t6cHhgZMj^CXdb&*h+&3w*+V%UY=HBCgEaO6Um`CG
z_65B|h7^4kzA~mQu+bmPFX!E}<1fvgta(1y0m%Ob@uM?p%QjBV&M<z;P{x7UuWX1c
zwho2l#2%$@6|Z6o0%gIdUY8t+QGYd&tKIoxtN~d`@v9viM3+z6$xKWqt;$`M<=G{!
zl_MIt;=E!0K1T<f-eSbppG9n1BX*L!txlJgZIvSa+Ie5{uH#ZUl2#E$q@+solUvIe
zpaulv&>A7EPlqettz}eOJ+-km*O6s#x^JJbYXfXDM_e{<2CwvNRGqB>r=_c<z1bV0
z6&0d;$7hS(bf8GF!SMP2J$`)W(ifUWP#Fni-xq!IWC)4-aL3R2A73`)3XIY=+;`)j
zzePHY^Ph{^uxR?P!Jc2mzCAH8kgx@Q;9h-|!KdzfDCB+<#Ehk-*&u$#pPuT%`ZI#I
zygat<zrU69ZK?S52Y%IHD4EPOAC%>WgZaR);H~u)gVKjwOm$+tVBp_ivQ5Map4aW1
zw0$;rM9z|MbK;Ul%v@2tY)=P}1Jj4Cv$L~P9%d8YT<$B#D}g#@oAzR9IsoR|a1qVn
z<@%3T+&3q$|Lzt(IES#riCA0CqnA)VdC!@zdEl1OCOWZy!7zew0!xRT{J1x!hWHH4
z!Uw(<pC1FVb4)rPP~?Q6YhpU&AJKP0If`=XG>q(`!JWD^)^CTjmiMsgYO?RH!ND8*
zX0hP<a_5w>XHt9eGjrd+1JgfV75cm!^{T(JY}P!sLp84kW_e4X@y&h-HraaFv08YG
z_HLf-kGbH}&7g~;=1UjP7nY=()F9QLyj`iM_FchWTZG;4Vm@h4m6KU`xepyplH#-y
z{lCVlCLpNLzMAaLce|(y1RB1;euF#VU{BMkNzGx}w-S`wJ7Ypl2*9KZZF~uYU%ZzN
z)miKBM^sD%GR>YXk-tH$u_7UVk4A#;KPPYqz-lM>&L03f8GCDJquiU=zvnI5hX>r>
z1WlqlkB>(rfP*@Jr7?WkD74i|;x{G+a@uAgJ@~%lIWg7gHH4b<Ypw9(U|{paCiAO8
zc&ewjwM%MerVuUyux3Zo6m?<~&0+9?%#im1)1n04#SmTRzu4uEG`tlXwKX7LhBBIu
zuBoFOk9aEz``I8C23fI@K5FWAFv9p=cVzHWcKzUvGmzbXkR3(Qoc+x$f)mSrs$X=X
z<hqjLdq*sbn%w&#%@p?ka)0}uW_+}IvA0Pcc;aa*UcYe#-XpYIcxyqY_7$_+9L_mZ
z)0xIH>Bs4`xU?j*hN0N}XD@m2R!1nQ$hUuAU}Z|4xjPGq-oXYm&oYszJdyN9Iy7nt
zQWPNZb0n;o@VQ((Fi2t?f3lRA9g^F2ak61)Y3Z8^iYE@mlbwSxD`gZQNdw{A=-tKR
zp9h;WR&M}*sor5@#=Raf-BP4k0UI|D7bjg78#JM?I7Z5b;~8Uow?S!LQ|+_V0XBxt
zOuMhlF}*{%Ue@2Ez>l$Le5P@<54jR9CUd314<kau!IdQa>ogZGa{ppieLICk2aMVM
zHu~$>Y(s8l#eH~%)Q?*G(zzY}*GBdhaJ({0mkjX0zZS<Kra)a+MCkX>G$aJI{X$}z
zm)4-B%&y(#TK|syjU(31u|;9>C6O^2^=!9c{`8{#B;zt;!C&^4qEJd;{U_y2qGr*~
zl!x+vsTk(?R9|(u;xiAqhe`_5b5mL0{=+N`JA>B)xhk1y%_VGQjj#5u_V6Xe_$cdk
zn*m+t-(d$!Ng3INe|ll@Olqnkh>}ap2p*4@L+2__xvwiQ6}7fH&opP`0gJIa{?Dr~
z(bwK!MXE{cI46ht(UP)_JFY_6+1cHDz9xen%8Cy;cX`|V4s+`WZdj+HjR9yZZ$}5q
zD_%-t63PzrqkpY`eSLkglLFNAF7{JC&``U?r4IftF}C*u>14x|plQVLmW6iskZQ#P
z(h&2(vb6D@jE04KY_$OO(h3ubNlhIpMNo6U<)2_@kU$HeEX98!s-eNy!Te83b0iKB
z1l1>>$jeV{uajCEU_G9s_PrZQ4-<B%HHI#0DBROAb258E3@`5LU{fDAofQVCShBvB
z?zMFI%2{l17DYXc+E$BBDPCn5***}K9_}+6eW;@1?fBw&0GqU*2#zXO${D&zUnr@a
z>dNc%H8BS#BmF(`sp*IEk9XG4zbNvEC8#%YucZpjsKA&yk{TE^&I+f_!DC>~X+u=6
zM4pq|o%9-})&jFXRV)?yobCXtAk>0;b{0p{e^3)x3;AMwddOn~tPu#LUnz8SWJ;yi
zp@Q;W(pd5qkfkp)Z0D-?!_8zdZnY?|`Fvq6!@KR0+}{VjNS^F4M;r;wT;61AG#Zt2
zN2?L-QR!t5L3YQTA2+UC=Y?LSzPo27YicC_>zMFe1L3<!KO~wM4TLGHq9iHl=?ep)
z)cQCUwLZSpAG;SugX0uXRGS~<8Z$+ber&)R;t3~JC5_pgtgug4Bh?a~kvsOhAc8E?
zJo36Qr5<}tJ}vYFV>0A@7(+by^K)%&v2G&Gz;rwztigMB7GYacBCinI)Y@877LyQS
z&lA^HS^l5gy5Ob~pzmJvHYUJY{1weGEr*bBT^EO5bIiB=R)h0QU#rP9St5Hs4O^S}
zQyEoM;xGdxp|yC7zIzyIK9Sb(!Sx0~T?9Y=eI>+z%=4gPK)=oL>S{b9cj?pKJ@B%6
z1orga#@xCJqi*n;JHyRiFu}oc{7#!(sf3HA?qqBwi-s{!Cj1KILyS@)2llj)If2a%
zk=ftI$U&|9mQOWCHuhN)HK1{75U^&KWaUO^aKvl+wW`W`c-!H>m)kT6p|M}1-xRwJ
zu)CJSeYP0h@++2-$s-dU9v-G4)8mM=()sXBJelA9BfL7u^-wzX8Z@DR_NlN-q%kyq
z())*;oFZ^~RH_oO+B%P(&M~M0WV3GWQBtsdTKaH{qi9#-PUiQyIRzDglp5UNQlf-1
zu(C2qlJHPD+nwv&**U!UrJ^3yi;|ZM_g@*n3h$<8@DE+LhoUu?M_j(gX5JrVj;{0o
zp^^gKIV}n6!i+E<C?yq`P%P#L`nHY?%9UU`0P;LNy^xXHDxrUlZ7}|?3U}F4OzqLK
zX17rwpVU#YK?wY@sF88<Q2gFhVWq$qB}=^-UsMhS2@I24)i83wi8!OKp~T?dIk<cK
z|AznRo%5I7sPgEm<?s71fFkD}Np3k+8hqOU=V$0+>waBh<k;iCq>#hSNnp1wuYHFE
zY^a>B2%V?NhzMM#1&zIWL_D6v7B!?wmj%p@V;Cgd_Nudow7$!b*fpbk4Mb?zSyII}
zn5kJ%kz*Y2e7qX;ISkwtV0{}c?4V1hjFkaJ8kXb7kt|>`C4;h5hVZ$}>F!lRrSKrP
z=CrUo2Ey%6T7{c`-#d>TrJlO4Zvx<liSnSjLhaRi`f}7vEIPt<#B1<KvZ4)ZF}i7K
z>b$qJ&t@C3X6lH4ma3qQ$FaF)5qmdzuSCn{^SkTBSJS-Ng{Usl6^m`$&+CcG_%8$!
zaN26>P#{Y!7a2vx^d0+xmzqSW_r{OaVvX19X%dJlSt|nSuW&1&<M`N62Pmtj_$?||
z6JSSAadN5>A0H1c2poN}Nv7UBA(AI+V-M!Tm;c%Q`@7ifE};ZSHIjF^QsS9$<Eb^m
zX!gVTknIVc%OkDwcgfsJP5z$ef6l(A`?k;G1{7RPK>1QlA30y;ALflTL!P4y5$SCb
zl@ru3G-$O_q=F+d&1MJ++*2ld=L0{<7Rr^EUxCSGf2(Wx?Qh1q7EZp7h_RN44`^i7
z&!|DM2*jjJ0Scm>#f($W1feCd1C5&L;<iT*iyVu@t{}NVnUcx$RU=nPu`!?OK54M9
z1mGR-yv&b{DfR`)nX~z^hLYi*6_Yk+BSV1p+eO%9Ied6UNDDx6CK!pn#@8?wO55`L
z<~y0JgjlDpg%j`Yg40atrQquRCse|GWI$8~DBA3XA}Xzh_Os+`v~Q~K5Hkn1-ghtL
z3akqyev4J)x`xedaGEUCXaPb{Uz@gYLeET3pX|=z2F=Ake*9=msIm^T-(FNk{k=Fp
zC07I1C@@*!X3xuMBDSs=v}LH{ocMA+$t-lXP`FYC=Q4rJW~+7OP`UDAUSeIx^%iwh
zs<i&tU<P>;6>Ur;*FZu_<gQr1ee>k&m=7PEZd(JMVb3GJKobi=$fN>4qMSq1<^koZ
z7F~Qqhm`M=JSEI)Li_nlyXAe_K{)RhsSXGy3R?q28#>+<2yP{J@4J|RlOpey3qfzZ
zd64+6oTNaK8F{u*^kFaNyfkXt9<@>aa#_3IEok+muC+D!a!aUvFTS5)$aW_@bdUJ0
zrH^&5K1H3oJK7hXKs}0Nh$P$p_^F*VG}tB9|C3Yj6xb{*H>VY)rbJ{wk&JK04JmQ>
zE`VCq<wY_X$eP2&Ew9H|u%XBXR{zGArmxSNe*@NsO(nR9aH1mdX0q2DJW|cdzfZLe
zZ)?4!dn>c%{|tL?SWLe;eTipS&RrkMPcG1@;c|b^C&{-j?~4PiAy4mb#9SjdOEm4H
zpKSxrcso-eWp3^u_38-&vPN}fNad0d0Qv2azCD;fV_g|%4yRVl6_40i99rG0qBmib
z(#V8q=W;6vU|@W~!>)%9UQUh@ujxp2Wg6lF?h1T?@=;0^qFLZMRFK6LR<^cYYAqCR
zH0<P+Mf;6ui8NUflDfc;>9%7M4a8y<!WUqzGX=%L{E6gI(rtm2@amc|n<-#-$pG{3
zOp=UTD6vPW8-0<Rq=o$QWl>W{%RnfmQq5$Dzsm<GByT7#aEK)K|12)ktj&elr3X?V
z8bRRw)9i$!X*u)<o_aZklU6lIDAOZuLCfaJbCM>N(ih4uc8fMhG^=U?Js%?<Bc@I#
zy@&FV`HigAI!UHRA;R*G3*hgChGlY4VW<YV36nQ}=RJJ*CBb9c<I_8z!(CGUV%giR
zo=IEO_}s}EL)OG;SXfGbpPC~CKmG{4Pq#7Y)nb?>&+?ZXKEM?ZIPQ$`MDRRKN&^{t
zbQ<?*mm?>plg56?72{$clpqm79alOZ2tF86bzq)T11%K3;+W$MXw(Ql2j_U)Ddj9$
zm$v+XJW+D(u~tct^bKk2NU%?kyX~rA`LzzaW#;C7E1F5w8C_oK369fGLs;K2)tRu<
zxeIFJiTBNmZ9O<ib6iH&hR7r5H!E!!kvVssl&;42{6ybT<WBPb@%e;o4(v;8%+;$M
zWATL`9ZXG}>ZXo?{y+c7y6@>%O5#3uMYSQ!e1A?^iXc<Sw1N8=kjdA$_%Lb;hr31U
zlFZl2iwjUPc&3<4aI$2a$OMwg`SW(7uwz7i<)c*j$d+mvNgVdR*VMa!X(Sn~R-IR$
zWaQm`$Qs=-{_grS(>l`je|2Re#v2@O<9PcdFaM<sX%^K0)a6mTxWhJ6)2jqB)ecsi
zQz`h^U5PU%mzxDm`3vFtD>iiZpO3TqjL2F!yo?sUirG<`bJNJ>d976^CQ<8A;MjQi
z2Ue097A*w}{we0}W}bEhHtN-o+a8z&`NOFH4zCqPF6orqgk~X+UCO7z!TEl*SE`Ft
z-AvN<DX#<2<E7aK+g88RMBB@C^Cvti;X9497gq+yq~7o|(M61VrUrs2#Xe^|O=_%e
zJ-_g7*=yyw*tb3ZW_t_6b$CL>w{O-E<P;VUil*RX7Seo7jY+@iFZ;5H|KHOb9-cpb
zn$72P>u&Le<~Pcu-rL)-3KV_3BW-s}vO_yDwSv5{=HaHdLz*6#9vE00c?`Uml@er;
zMsY3+uNb?n@~(2TDVOM}zny$_#d>I$ks_mHjo<noNmdl})`Xe@oD(&?cl%fDo7zgT
zTkz>7xwW>#3-9w@;M7|wpP4x~E{Edqy5~#Zr;5)QZz?%zLu0s#BVSgIT&@5+Pqu!X
zoUL^?14SuRh*X}7B{tA)PcDBu<s1r!wtdor_FPqF(EVvLJVr9_n=yl3L;ISBD9MOH
zJDm-3T~B_Q8)n66cWazhRa?@Ds0PZf(KR>dK!Lyt)#JhPa4_tKs^cwH@FVgfB1n}W
z*G26>OF^0r?qUef^F^l8lv;3pd&sT+c+?iyuG~86=S3h^)c%|<4h-mUX+&*P(OLQ?
z`c+lBcYQGa7A}p!{9zlrot<~}(Ccd6titT+*}_FBv-ibp&J!T?&|Kx==gW$ozrmhP
z0b$OE9&(L`D@BFR`7C_LW1#F*Qu~c;y|@VjAk1?^US-{U)JWpWj5P}i#gLG3?yL=f
zi_nN=$;i}&VF%E<<gVr8(ai&4Y;vA*KPx*wHH)6GA5oqzBcP$msq@#el!owb%+B9`
zC-~Ly`P1VeAcA}^`AjCdajS&>T{ys58sCZZfoP%<7}x^ACFJIqkP?zaRu{NPS~x2Q
zpZz?ry%+@2UUSaMxl?p~_HuKN1n{pJVEEjUy+YwvO12SEFH}H~um~{d1E&D$H~ES&
zA(N0VRr2<o)YfU6nZB7gh&IKak^bd7IYz1t3|`OxiP_@?Lmfy8WmmtTB$@fplK+%g
zz!JJ+O?~BjemZJDD`oUe5vySPgy76D)_cDip!*eKd3iaG&RPmNnC*uEZS8w~A`KOJ
zPyV^$KeiO#PZs<*R#c?)`l-#&Fr+SccB=oa9koKXGeh*-{Q8Lv4zL)A$zoR9gVTl6
zbTv~|Tf2FzpEY(h<j0=587+txCAIPcx|imD%NZrDBZBI4xs9zSr8&{0J9jF8hRN;J
z!NPjt-03biIp8*IYX-;wJBhZA2(Kh5v-vhZI@Jn%9P2>SF<H@5C|v&j4!GrTE+Rgx
zOV03tCVO@Ud*RQeo<yS?Ik~yYp+|4-$z~-<JPg7bUz2j9|G2Iox?9<YH8T=r&Hbyi
z{qv`z{5KUSqZ!F83o^=)gh`+rVCRFfphdF2mgLwpkg#fTK0u%>`|Ypf36>f1TD|$z
z!Z**|3q}V==|E<_aqpW9#->(8GpejwLxZ0y=dMKjc$sqG1~|C~EOSb<1fLTanlr?!
zDsd|zkJ%+&HI=^)nhRMaBTV(llA0Z-SZyrhO$<ZJCs{+Ci2z)x&RBhUd0{!v_9i`y
z^gy;Lr`VOl0GJvjHke)G{e61zZi^TD1?mvj)Ji0gG!ZwD?}cn5rj~v#^gGNSQY<o!
zta)sK{6Wv)ur3HynY)Mp{O7u#M9bwCU0%f5+f=;FdjDLJ=#cf2J>8zFTYQda-MEZC
zKm3(;(NkRZ_Sr#@*NYu?|Nms!)trBgNv(Q#dZzgq0AM`SjRCI#8Ldt2TRbHDmdSb<
zYE@5fT#y2;M1)>oGf?OiA9lR6lMEkrKi(e}1<sFiN{lqp!`ePaoHx;U$#Kb;Mq%#P
z@)|mjYXKiuQ5xRJgnJ*0kp|Qj?&<w^d_WG2=`x}4^jpH3P?XP_C}lbzEC?Rsbump6
zcLy{OyjOQq)J83Xz@c#MyWMh>phqYt>m)5_D2F@ROxji~b==gNG*BPP$;tMS)+Rq&
zLyCtOueC~dXmIG2#N3X_d)QAn-CG>1i=jLI$)AH+O-TuVj$GLbtiZR091}Cb$XVA*
z@&@=n`|n7EwdEmnSkrh*K0aK9pHK5XP#b9=c#m7%Dxpw}vy6*CJ}6+-VT7(EB~jiT
zR-L<gtj3Re0_7YM%nLv1_mC~RFD&y_R6{C&Nt!_RxVD|nsvMn4Oy!-|skryzzU|9^
z>>K9IiL+rxyjP08;giS!kD?R@=OkHz0#sv@%$8J~5AALw9KMtZy<98kT?T}`R*u!n
z))!-(6*tT$ACR~tS?@TY(@C8x+44A<+m2I(;{<8(N<VZWl0=<SWi@ik(D<IXBn;!b
zQOTnYR&?0p)8$0aso7nH<f>$S%UZ~ka=0t-)V(KN-O(;VCUMP0D%*wO0}rf=rNFc(
zd}xfEvXNyX1(YRqqA26Z`M*JbGd<#Ko>}Z22~Fo7M`(y@nHAO8T0n!Wth@u^H=9=r
zM4;xa05#~5g&B>r)!=&;N9;2w+Z5jw%3e8Y7b#mGJ*V06TCYR%?k!|kwSjO06D2qc
zVl@77eOfNR-iD}YkA5-l{Nq7j<DnF}K((5@2{&}3XX!+7B&u8hD_UFyMMW`BfPGQA
z&Vl^-O@ByN`N;C3qnwh+;QG&>#jfuMGd>#fNBoW9c@c7H&(-Uca!CkM{v5d0*i3GT
zh2++V|GQ21Ly0_$mLw!<{NnmC)d(RN-K_{zlp>@DB(eB2sZvvbLehmE>$<o0J=goN
zO_w@SD7;&%jeHpcyzyZTju^WSR>Shcg+=Q$vSezBqf0zzbKB>ehZlliOC|6!ziV-E
z>TQ%H)*gBr3|%ID&QBX3xLaN_v#A2?hp@yw@SdBS{=tfdf$4kH6L`@`+(Bz|16kqj
zQJHmjfG#;NFl;tuxw|`MS=2l)0~r=&n;J_pmes$ZfdbxO4J6j~cPmE;_02iw&UE$*
zFwm<)-qbjw-;ttwG;CHnJ<u^t%Y}g^1x3a7(3Oy~r?S~TOTq$9rcy(txyc2OpdoGX
z^aPZ!JA_c`+?|xP84<K{kzvbWsuFe}_IkN^LP6!>%=anJ)GjarSJ+XMsg~fp_Jbjs
zbN;h$Fc5zK_W=^e3bF%$_E|3z&0Rikp?;+tMLG~a<lUP8#pnYYW8)`6*Qo}ESII-C
zZ@~)gjG-3<-CUjrrZ-QQMq-+^N>6Q{<X`?y8PZz4jrM8NmUo_p>Vf^u2D=8h<X%El
z!4!5$Bk0i;E3_XIED~gg=5%18hQiQNkzhU`L6`!h;4V@Cf^T3_uc#$R27>syK&sPS
znr?#8#?!nAOGLYBEM4VQDjt?+Vv^G`0m^w8xKNEBIg*u+|M*d$P%L2v{I&$crF0Vh
z+`m0nSv2m6RRrxi{ZocXu}HCe@wXpqyi`pn?<Tnk6CMXWa@*~%YV%P~#a`31YHI;z
zKD;-)BGsv$npC8Q5x;UbiAPXpqr??b0<u#Fqd*oLsnkS}JEqRc5Jlf%;hvu)n!WV%
zXRy=|n&4XCMy3LS;Do8^=>?vMLo#PZ_dw-*>hj0N!KVg^`mh%gUIORD#gUQz#gWtP
z)}YWpS+2Z*cf}t>=?ur$KQhab=RebgFY-b~!of!iZ__`1s4!FXF~}k)eGa=MBuZQ*
znWc^q5v6zegCB!OrlD8}fq=v&J$-LT%lp$xxTsd?@UtrTvFDQ6lCQGvD%zON3fe<<
zEBit;qv<q^uIY5kVTk`FjUuNE-l-<kd-D1ZxEVU@w@KE%M&1%r61aL@4+vHURlAm}
z{`GikHUXc+8(l$8lCu~3rn0%WRPL>OEfTiW6S;4kG4Pug4KB>>%`TbyJ_l(leQ&HQ
zvVnYDRy=Mjr9*QFE`cnSyJpV+`YO6hp2vXfhBretqprW~{6VQL2}ePf!A{1UDvTKl
zPeC2e-qORw;L{4|GLPaBfQj-$v12O72B-@q{i=l-*iFIh<MS^5^`=aHp-hA{1#aGn
zmV7xw7Vq3KRa~gABm7QelOV-jTpR+_Qep%0%->ZmdM^_|@4gXqDB!_G@5{xNhDuW@
zL50HN1LWV2OldAHBT=;A^2_abCb@ykzu{n(F8(1uoyVpRA|PLZ)xGkM4>cdc9LHQ7
zBngQ3tWNBxD4r805Qt^9Wnd#GX@6&}e~^4sP7m0e(ghx$j;2*O%|f{USbgqOhv47U
zmcRV_oJO;*N3;sq+ub7CUiJc7>>+#E^JnB_6rU!!r%Be>+R=#=JeA!aYi*t4*KcZ_
zTSs2gV&0n7MFR9#57}&Qug+=g%>Rc8ycFp+np}nGLDl8!0`DSnGmO`AW@pRAYLCAn
zVqko!jtJ7`;a;s_ZCSQJD$jMeXj!h(6mT)UUoK{81l8#DmnN4QliP?@2r45`!VzIJ
zd36=8BQv-gv%LIZK3G(DOj`Fv=K~TQ=Y~6w%=7TyLyAa66&_D-ZxLKYc?tT_KuKZ`
zp3Tc{!O_$nB#9zrb{x`*o~JA?my3tbi*Bs1<Jk=2Z*PD=TeF9Avthe^>`D0yRCD98
z6Vvia9Qq-cmcOGF_TzCRn7i`jpCX;RZMmcR*5*z|zt)!iEjY^Nq9P1$gg|=GrwM>$
z!b@LOj&F?%aOlGQY7h*?q7LRu;Qa!91itG-KqT;%pp5UJ?9ebq`W_}q9g&hR^3C@R
zZ4X?R&6a4|c&On%Fbnd}x4mZLSi01&W4_l836egM_YA;MnF<Nr)f4Uw>@Fo*LCo9d
z)`8)5^6fL|EkBtnKCk~#HovmheFN-+3p!vaM){5iY>K~B_BSCRbMdP4{=Bmb-AM@o
z<9cm1EnJLuUy=c`-&`@n_q)E?ot&L{(LS?TSyi?D<*{wBR1tSI6T3zMZIe-LEy==x
zW9o)R?#gknwEV+cxy)L+A_aIekxecX<QXHyS?Iqs0!$*oa`vXgx=SCV0C-^2k}F=@
z9Mt|}YIY}FIolv?Ho`kVpR=ynuTyyHY3xYSLjBEHgmO_I{i3P9DXWlHJkL4v<t&g_
z+E`QlF|7_Mvcfg4BU+Y?WcREsgri&;$~EN^IB_ET;Qt1obi<=`RtP$l*M(mWi)x<_
zbZ8<rs%ctgENQ?9mvhPAz@fS2;~CJ0oS00}NJjohNphx{QEoGpFnlwSCMsWKH(WWO
z+-uZ6#o&E7Qg3K)3X#QB+WqMEU{k+4Vo&Z5(@BU)xhAcZnBYA<jL?zYun;wSQL6Ev
zI136yHZGacs;yh{^X`KapKeZXs-;%Cpc$R|e~~*C&%fK2Dp^efbfFPk4bz#@5@4xC
z<c8Q_x{D+jM*epwOfnkv*^;NLG$Y*617p9!<v3n*D~I3bPCjJL_4KG}KN{6_E`8*V
z#=ny8lJf&W;8KLTDY75rsD~}cxh@0}>}1+<d2tFqB1Mk`zsoRfu6xH8V8c2#bu?0>
zg-B1*yiE*lm;qd0Ep^T|pFYXELS%D&|DXI}I62`ENuSKX4FCMIk_E2#I85rjSt8~c
z9HQe~C+E6P_hQ)tVQMUb{d(Nb&su=n2oDuZ>~EM4_ry;gcOG6I{PXxNBr21ygedTx
z)IEt9y{CQlCq*K#$_Cg^GuKPI9Pw_Wd*7pSIe#(vU>*oPo*J1Gdf*Kw-1=gFBP8PQ
z8Gd7I>g?~){h^In1~$4FJaF>l@_ycH%c%D0&FyU-JHFytU1<TVf0>{HPm-CLIiA$4
zNDe->j(P`;tub0muK6G2Dq9a=Xk*lxXKk)Ayk;7*eX*HTUfw-CxXVsmGQ0Mi=>6hC
z<1KNZ=MQkv-78wLF&*76mh*LVQSLlx`)q4ugauBT4wZdqUA(vE818#WNlgxKKd1b!
zs9>n|!mj&Sgmg~pE+IRaM}^hO2cG{_N4B<1Dw_0rA3sqpc6e}*YD^|CHHNK?{yH#A
zRVUjIdLalp{vvVf;^D}S#3bL2d|}26ekr(%2|(t>4lUb0(Y&k}@|O6Oct;{p>3?!-
zinyWq&o-tS%AD)}6>QLGGC8P<{6ZV+fc&9w6fb;OAx82U#*&;v0mF!mF?O!BZKz-v
zHTrx&g@mL6<zqM*Kwk#rP+3vmpO{fx%>*kv&VGi57$-%OJ_z2*Q4(w#)i=6<K$rrV
z3`2V_<{iGn%#9Tv$qE_pnKP_oOo#{&1vMDTkTg;&g>(bmadmToM%n+QscERIJ0`eS
z0n3fI@`P1>f}=W;p*M+Yto-WyYlV#gd&I?KpSMKpQyoMCI1WTL_S042)<i%-8a0HR
zIN`usq$ai~%Rg)nS^YJ#*mloM6bP2R%a{*laj5@$bTYlr44lb0{=57GoY}S=6SB`I
z^b=}s-!K~92xQJxZhmoiaD4o@YEs|1?C)98^4M@Pxm9DybRR{9+N^Vq586`Vg_;;7
zw(vtpJ>%<qq6N2)CLYXLi5bUnIq8WCkT2rAV6+UF5*NTl4Yr>ZF1pzX=*iT+y{6`P
z3*#(O{TjlTCYbu9L_}TKHRx5L_+T74KOsZpQwPC~2g)leeW7px_ae^9E7t#+nG!#1
z&S)3~D87Ir((h8Q>YJ7~Rev(BeRalbTa084Xk38acjE4WMZS?zP*&6&R?`k2q65=K
zspNbMpBGB~&}6XcH*p^YKu{3Dudq8PTV4Inr8oa-c+|TJ^sfci`>tKt#JO`ZeF_wD
zA}Tmguu**Jql8i$&Gwal#o?h2I$3^9yw56^$sk01FbRpQxCNR<M3x!-mADvc5B|51
zYqnCPbx2-y0#uk%hE$sOPafEkNk^&%$?>J7LYetZWe?{o%E=p&xYOzH8nP>xvy{tU
z?Ou2!GRv8H`<Q#^mHeJu3Ce-w8UV(+rc+Vl>K-q|+Dm4=eq!*X=(7g>LlwpL>K6s&
zjz)vh5~X!n3m#H=)-BaePj59K_xuA-N#AC3HVsPso`*|AD_26|*lEC+oIHGrqw=j^
z-D61x#^ro@L~&rt<8p5Mf=E8(Ku-O2_?hOEf6YQfKI741vdWLoXFt3z2tn|?2l_N3
zoRwa{-nWK}!^>scmXm3Bi65UY&bNWf-L#5ap2s52d;g|7$M|2%3&q$p=cp&iMvMZV
z8@#3T6s(i?7K49l)o4BZJzOZ9mxHltKRGjI;}`A$QvlSh!sN!36yfva=5yi`17h@t
zl+{s%H~$?E<TF#`LrDQ?{C8i8_z1fSumy_iX*%3p>0y|0US1?^VfRKd{UTibimbj;
zspXiq8VoK|QgR(dh<d>i<Q#;0d+op9;hflF&IDu2*Td(MpIMs^2y-7H<)^fx%l@`p
zCgTrJCIGQk;u$GB=xjIW`3IA-it-8tAweTPnH$)*D=duNky6zJs_qa_H}t5hgOTxj
zq(r!Puw6%bWsP2vfMu$S;k6!FjoRj^s9Eb<<>jI|3IXfI+u^>e4^Efr_nGxJ6JIn+
z93-|@A0UVJ!i1gpcukFRFqqI}h0VD?{?U~+W`Lq3YFS5jqZ8qr^zGHUf9}34$$XL+
z@Gkf3MvBlO_m6Y`7k>M$Bcd9>aE7Ud9DPdu%r+nDU43lU8v+$MhNaUQKnqf2CWh#B
zC@yz~IHRdFsSCb`Y_tXtDK$YptddC}C2|Qv4Gu|WKjaQkhZ9pZlDyEaV2FUIz|Caf
zRwv5t$^ux8aLgys#Y@>R%n>bl!(J?ycjL<Km^Y;IC3VgU3ea`2F2ezWoQ9-8f<m>i
zIaJcgh`b_TiSF}|jUaD}2e!{{9!oCR{+aV9uK1BSKVFAtEjW7-+>9=cw1Crop}gGx
zrbY*L2kf+KUYt*`g?cLUY$lV@fTjQp?E?v*9puw?oBfG>|8=)pNT_Z^nCbS<vysAQ
zR=kOlg-r6G?beVxTCBC1JKFBIz;QXU1Z`ZEGH~uMRfC*FOs)HqO%S4J^SBrA%9^wk
zj_bi>o)=qUcy&u#-!J<<x>?drF7@9z?(@Gxm;V&oj}<o~?#yDK`s5Wn5@3nNLb?=V
zSd5Heyif;k{Kw)$<yVE}Paj{Ccdu2_efPP%TvG)5p8q9;&zDP_l>zHOkSh{<x%RBa
zQ0GorHk^6-?smeM0cQtV(Vf1_;ErGuNIdV!h9=%}?Z_YgmeOpELz@4oiO9r#L`jQh
z1M)uaez@7U5{&U*T6vm}1;Cr~67{*XsV=u7{TRr%0g13jKKb>&nHKrAJKTwIWR&%_
zcty6n(%G$1tUD_UP9Jd}%|KoGP8Lk5SypH0sDb&S1{n`{5ttzu6v^HRH^hpf-q0ko
z|5AF6#B3#Bp2VyC*)M#e+&;C5H%mV;Gm4uU=}gGATuxw$*Hy!y0ij-<G}N25U_t!)
zf#P25@qYWEKII&zE<JnlYb?pU%BCk3t*K=$VNKqDB<@zuZ2r=a{c{*#?Mn;{BqD<1
z3>;f+yp@OQWic<umdJ<rtKc%j>X5uU^6}{UKjbFq<m99~o>>HIxk?=RWwP30{NMIY
zkyo?xLopnV8}l0Ic;Dj>gJkKCiDd()8{R*fg0}sejwZZcG#foPda82NY5VZ<Y;V%b
z%3VlGK>JIMS;h5UU(-S0V(8*{X8UXjxIDjHQMq_>xu<g4nP=M+jGf##@44J_-ZWFk
zaNgRq{xRiJF08GPA)&H|uX^F>AC7)=y3u|z*y-c*T)K&zyzA<`NEz{Xe$mwv-saLU
zgE7R_%Lrb1Wc|Csyh=}yY<f8-TiA8U>>4HcEdS|NI_B-?L<qD3H}UEk8Y6(FhS8~-
zntFPADu?X;jInKN2)Wpe=ZW}@VOG4`$(|fD<uUDwr)V^2@O!RgkIpHW6u~MgDw6p$
zN_>gNfk4x&q2y1`T!|&gQR3^;>wo?tQstf0-B+$sE6IrqtxDKDM&9)i&(pVbtuVum
zaL7N3v{;1}>E9M9)g7!*<3*2Z$O1<TA8QoAg{5O7?U$WeF^kujgG=X+*C*%g3WiB0
z+}a5J(q7G*uw{}g=yf9Kc3cgeH|E{~S5F_69FPo=Lsjk(<aLGjxI1F%4#Y?(LnbiG
zfaFbDVY}!S_w{KvSKKF(%=^UPVNP)ty@#r|u@sL)0tUYJ_S)$z-?*)3_zH8o@A($k
z7vq2eFQcV7G^Cg*iPSKDNc6RJ92Ik>;a5uMw+m5PZEr`RY?9oP#=jiOc6qOn)NVB7
z092}xEolKVD!0r-WO9E6k#T$B((7V4skrvt>BOAO^u(0owF0;D@=8pMBs!;iY%N?#
zIp?*|*w|asYcUN>9Py>G`@$V4g>8*UbF-7ng`kK-+si)km3)o4%C^1cQ)2;Mip<nz
z(ZmHR<P9Ys9A?*YFvdF}4nkwdZkibY4Ew`RYn#zysd9>V>rA}OQrdUgRNkRzI{Dqz
zVPtD7Wjp+E{*BH(9ZXPwDHSiA-hqpSOS2|`8?p=5HDz~)Xc&u@6-x~uk{}6@GWxOf
ze3|M-Z=CMA2h64}?|fIVWKA<>qPQN7@NSy5Nxuu#QdOOaj${(~T6)i&S6cyKwJ+1R
zxiJ{ymGw<8<Q1+y=^&}B_T^nxt>nY{@};K`wc@8K{lzo-peQOkI0T)X{G{Vpn2|$d
z2=xd~O;%8&CG{=I@D^i|qTN%JD~bgaC#nCc+zsb};D1#MLti~P7rsKlqO->C#O?5>
zb28vX3Fj~J%!U>&1=B!;ssdmK-tTg7ZKa3(ms7y`{P_OonZ#tRCv7|XRo<SS46O5K
zM`UR4inBN@d0BF5Cd}$ah-rrF^ymO(SCUwF!ui(rMa;d@LNh@g9-+tK!Tyf8*}1uh
ze@EJ1=xOtu6LEOcB7Va6kkWejnBH$CV@g9y{J-Ez^5LUvfrO&5J%CMI7~65GInd-a
zQ2PbSOiP8{N$Ku&<uK8ytF#jv9bgQ5SeBh~-V0MBx;tLASWc>KaxFEINFq0CcqDM1
z=(^fj_VDG6d4A26#ABb&KQ@^kvgnkH{U1%|8P4YazJE%I)~7<XV$|N7s$E*OYqv)1
zQ8i-kRkxzGf+9u{T8bK}+O1hDb`%vwtp=$XTT%S)@BcV{9_2XXNl4zg@9R3x*K3H5
z?nrKGin15@8`p7pYHr(cyf}Zx(LO~A6yNlybeFo5y3W0E{2HLFi%`)rD`aD5+uVAw
zvmJi<m*e~_nB!OB@N<02&Jg3>w;yV1ELk;#cKfNwU64l6r){{c*+Qbmu4}Nj3=2(X
zQ=?qsbPgv6mGV1C>wVS&2DL89kL$v;F#`cX`xB$ZyeH^|i!;-j9sFcqGIY4^D$^B*
z&oiN<Jc%;F_cg%w?nOq!;eYh;UQJsVysj@hbPy~DICcJg`=nNMsG=flYy0$c^-n-m
zRkGFE+S(UYHoa1(;tz6E3J9^NNzK`+s?yNV^WG`NxuxQ`)Oyj{>D{%z&(440CWH=D
zPfjEtH@AX<Uw9TTY=xyl=#>8o(pD;GJ2U$Z!ita<2pXUZgzD_#O$K2u5QmMI#!KYJ
zSejctN3ueJ_%M>tO03jg$)|#Xt8uJiZ8Z=sqOs{L{6<WJzgq{Z&kc4#t{i)xPvw=p
z2S2R_Xx{%+s;u&Ykn57&jDtF&xR63l#xN6IdMtYs=Q{d{2|1@wNdwN3HP)eQE+PG?
zP)7EoJ>d2La!YMl<1H2=W8gtRjcXJn-{4?|Z*6*r{nu=Ulai_K2s>}-xDY{fH&&I_
zO()1muN_*TWX9Pz?wB_y$SW!;C@6qs;OEy3++eX!)id<aIE3){-7$L2puoVj;LDLR
zqd%AbBqH~Q8xRnOxNl*kHXbBY*Aj!D?~Y3XGU!)ZC*4PgUmHM_*0GETUEh9Q6FZXW
z)^B<`C;C{8G$&egkV}mP#Gn-%ePk4NoNot55}GJF1?@31@{j2s!j!onH?bPfo~X|?
z`mk5qqCrS#VvQVDp)&IT$MfK4w1;qIW>p?9Qc%h=3l){s&;+_|q%;x5C|;ChDuDVV
z-&-+=LOp~f36>&3jG*<rLR0when#7w(6Oy~(!^EYprF&AQ-L-_jDctonZu$(KkQCn
z0Vnv+L0;#{8~FQS);;dM;GVL`HC?wk7~w&W5H}<JwH#x1_arGNYo87Bn<<j~Fd>P0
zieWW9IMhee)$}pe!e+|lneMwy;1YoNMpZWNvOC1K?W2sQWo~e>2ei^e9&~eD9*oFI
zl?(3Uz}Qx3$>chNo`Hd8D!Xdy?k|5B=kh=NR*U|V+Ztn}<C3Qj8HFeR;TJBG8yapQ
z-`->AIoLTW%dY_o0?dZBqKFS~^~y8~Fx8m#Qkrfuu4Qt0qRYEZGZt##p$}=kt)e1?
zQ?q#N=&L6(ZH(%s8qIwx9(uKUyjncxwHv7^?DuB~4<S*f=k1GnN_Ul1+_mV*?00By
zl3M*)TJ;H035)2wLa?U9*5@ligRV&ONRbM{?{eA>Kd0m|NN^YSp1eL)eK}Bgu~sA+
ze%|SQzUqCx*LIu~QPT|6Di-Q7JL{m>**Be&1~U3mXWg0`z8<!zUd#&=zQ`(F+uN%}
zrQf}+b>$lurP|I(M57EFXz1}Ma5jZox)ib%dU$wMRef?ygmk~O*C@EFd4Dh3-*b4H
z%@o`|Kgby#N;j%$XlpzE!*vw&rQQRL2%ed7)srM#2@P0%om>QH3$4(1A@9$G2OO^7
za`AjC)+pP=KOxkWV{8V(zGl;E69}%Jp5W5KfgXt^RCqNuH7$e_J&`Z3k8K1rJdHNt
zx`Ipy7EdH(?8<om$YhnEcKEgR!C8g%4yiE3(7dU!5cj9F%Drn);EY<cBC;cJ`<H@f
ztgq$i$BtIM{VHIPPOTFoRb!805M$MqbhtvMkV;d`NWV<)@(Z5yS@kXh7bQEneY7M$
zM1h2Eh?45<Q6$h?EH{|o?!w*55y&X0I!!4=i5T$YiDt4XyD^f7Ae`7ros^)}?+}fq
zh?eS%g+9kzMJ_$wY0@GC4Un(W8JwyzYb~=Z&FkHez%5_Di@m!Ke=S8@IdzfMNqVib
zDCA||>iT4>rw;4mYWF?w!hK!QWbP6Cxrx^*R5Ep*j`&=Et1&!~=lTicp1jDqhAqC~
zwpxA0ZO|^pJhIpNQR)WS8-MX@k75yI$4^uAgjQOE7n6h8!Gees)sy<#+P&es*-6t~
zusq%YN5J+MDE9+*`dN-8Rq<z21*{?7nH4m9mX*?ZBgFkLbB9qaYn+glinkX`m&>Y-
zIUg`R;n(e@$h$tqtBY)L#)h|%gP|c!w{gGu8`O{8a@Ai#qZNsgyQo^hMEUo479!}L
z{FBbv%dCZ8*PP^UBD)a}Zc?6P4~O>|J$Um0kh$Yknz75P5ky*I@`VxWc<hiHYmo*@
z*xX*dGAqZR=8k4RufP|4>y7pP7|Feqzw!?1&}TwjNdtO#XNK3t{cV%>_xu{&p29aS
z#&vB6mwX;DH{Iq^kCF6ahyE8O`OX3wtIU`*_AZc}Z=lL6(WEGrGnPMt1`IuSOFo?k
zw2T8*#|jO~Vz0vBn;ylP8Oe>Wzi2Zju$}(tKK>^0`f*IgSm*k>4+3$M4AKkU51bz0
z`^X+ZR%5ZgT^4Wn#ij*yPX-2-?0<gsCBP9FsYGK3q+7+zRdH_Pj;W#AszIiQ&>Lpv
z%uEe+$-d3vvvVmFACl8eG|=@Gl}Sv%sA^biM{WCRYqk4ZULl@g*a|tH@pqYI=x=16
zp-vR^aXKD{Q_%ZP&^mjZT~Hoys}!^5NUzC8{t5XXz2*$}hrH6S$BPd!JF7=Wu9r2!
zb-j*Wf%j2usoV4O^It<Q{!J=1c{f=~30Ro-x)er?Gp=-;g<p1}=gx!YPrg|I^}gFP
zjuH7sr{H(K4eDfX{6S@|^xp~9vkBJG`>HI3)*WFOIhd;t+CYHw2hR(BH?hQ`sf{}-
z)1<>NH-@H(WkRZ|E&D-VQt!nDfywrdma&WUo%?5-4PzneLc#M2gK1CHBoSg*Hm7SQ
ziJZ3j;m)RF*y0hUPpTn1JA(~?%P~9YwF4!YE;BJ7#lXc#`4CFEKiyif0YY4q{^w>=
zQ3+c*^{1xR!g$d=iAK91o#8?J2H?Gh4(zqSv^C!E1w!QaW*%tD>YY=vi#Xjad7Mq-
z;J`RSy!$MGO*1o#y_ho*q~>qZkLmL`ujKE}%X1o*eGez`@9)kNfH?`3j<e2i;LPq~
z0yzC8B7yK!hnb`Wz0PAcF)$R|)171_f7qiCBPEH2IiXC{U}eLuac{D9xt|GLE#c$N
z;T{y!TEsLg-sk#THOC(-qaP!+3j!_vlc(~T=ehm*Bt#s>MdoY_qYf>h=o6J(*42F}
z^&4|!7Sj~_xr6W!P^88&6WKa@m}8*)v|N}`JzYqYG2XQ=#6qa4xn|DJR=idZ`rs$m
zqORrV!_5PEh?_uiv5bbc<)f}Q6d}bM+)7JEw<_MD!g<NAHjMisjPOCe5!~u1-tu6a
zl?(C-yXm0$IGi*3YJ)pvDysb7lcL<Yg%=@_m*cos`?2-_*$&QF{f>q1nq2#Ca^}Uj
zmYg>`8=I&YCRwvsMn71*>V8=L<<cxk#5sfafRab;x1^CY*$xV-<H+#Zv~H2xA)jYd
zuoS$z)tQv3t;HGXt+%?ez;V~HJ@H0P?f<B%8rZ9j@!?Z4nX)WgydRA!6#NB@H4_~2
z@2Ww&3QPdTGT#<-Ddx6jKiB3R#ip9tIu`4>wesil3qM2tjY-tOgLc=G*Tyo|mUsl1
z;9eQLYeBQPq0C8JZ@2hY-^iOxpJZQy;!Un$`TKaDftrT|-<Hgf%Awk<KsT|EwO$Qp
zx+jurDP_~|7CLi6H@-ls$i*2jA|!gg9O?fmayWqP(!4IW)MEQIXXK823YZL<Q67Cs
zjm=VnI3Xe2oWnf4nZE44^kG6mbPe2>=fLqo`(Xx$rghj<8H%|$^Aj@xw^Xw7{@5!(
zTBW0{Tz`Ueybwur@nv)MB*=3zww<nv%xIIKPMcMvl!*`hdleNqi}0+Mfo_Znh{}pC
z0C@%Lo6kb1I>x=chqokD-GA~9*zILPAT|~~ZA^`!5U%wK1p=-hKOfw8WjoX0#3Upn
zZYL=PdpRJ%ugXl6Dcq*bmw@g#N#lAzX8BamDb}9-O}&ct>?p?j@;9bX<@7h+`{J}I
z!$9KNc*cNyLf1s`i%D4Q8`F^@3lv~FxppM^8)VmpI{m`o4I28B%f>l<1QQA80;#1L
zPC|!k&uSamA_D^CVFw$fDH(q{w~G<k5{6nCIKL^_R+}RdP5%}=Pv4H#i7Gh>28FlT
z28WR5ZI;iEj<7M*saa-IFimIUALCFuY>s>iD0TzEmV>3`*)8|gBp)3CAhTvp0a#nw
zJ`ZY{lhMx-E0TQ2Ud!#>Gz(8l?IUOyf5LoXK!TO+!STgxQhP_6|HGa!xbSO8k&#ge
z>c^mbp7Zq4%20SJyHuirMz0KOkP@m3Gph39=yX36e)w7>qx%IR0@k36yl7iEYiq44
z7Gxm*H2BNX|7hqCbq%;5JrJObPA!6Quy3n`_?%5MRa;deU)4_-2i92d>OQxG-?(ih
zCfEfdgF+L9n%E^Z;}_pZVf2MV;O0?eV6vzeX@o^e;VGPv{tj$L*b44q6!&n|G0d7_
zm-n_?I?GlO@28<*%ZaBZy8d276REJ)96h8l9_7FW9n5V8x?l>GHDxYet|G*!Ql)&c
zS3tW++JD?&JOlFlA=x(`?8H*7Q$()0prDmb-w(@Mp-PP0x}XRoH8rIme91ESd(AwH
zjp7_^6g;0;sy9UQ?#*ze6(9vBFSHS}+JRwVmQ%@fiVzqNSk;TdzbbG}_C=XEV8CWp
z_{QdhF2G*C{IAYX)y8VtU?#!klbej-P03aFn>)*IH{=v6V^sges9HOy#1CMUIBc3F
z<-Y{aPL&NS*illEL(=k&_V%#95A0=Cj9fqTU<a1y<qe^Zor^0|kz4;>xpbV&kzo8W
z-?`upUF((<M$QovA0s<R<nDYKGUYCBpIG|1R{^^?9Puu4_u?WFo@FVtkuP)m8v>JQ
zg0=8T;r!+RA!Bqx8Z8;8Ow?S%k6-*7u<0op@jv%#wKGh4cK;qW<dvIDyds6S*kSuB
z_w$#-y6{#1X<nrBevOwd=5eQXE|oD%WKx^2!5{yY)3p%HdP3>sgIlvQjejUIQKm`d
z0MOMKvJbwR;qPJ_W6jv#_b%&r`)gU1BquATE&fu-!98DkM=aq*IU!L;n4q#@Ru{Cu
zBAHcBeP1ap-S$E_m^RRPpF()%7pBIXl_{RoJbP1=C2!C6|FQ(ngx3A&Rll6sBb+j&
zf5I0YP(c<sf0q%uai15gEO)g5qv+PasBiKv`8Esl`37aCrkGEk^g!d`)JL~2Nn}yK
zw(EbdLF!SPEgdZC@I<*7jA|6gC;V2&Pev#!0vBqaTTu6RfGq3Phl_29T&x0^eW-0(
zSgX&YYLsgP`73lI7a-34dEK<?Kf{ZG%agv##Hahy&!jhd%<jhh=>$T!lDD+zT{fq^
zUu?wC&08r2FHOp|tDwc6wM&{Bz$A#*HC-oVUVoo$Ion_YdnBS-#s_0s95(Vmc~zxg
z0+GNLBk)t-g<xCV)T$g`q<wGFHK=)Vi~Wvr0?ox{Hyx<0KKk>wes)9MLA%pj!z-Cy
zZ8<anQUaO#8eSMY3k>#y<I>O)Z@>?Ma8c@cW}5|eDdAbhn{5-Jwql8nvZGbOZC*;r
zf=*SlmaO7|^Djs-nVF{Q_7RG6zkVIfm(tUve@;7H{NpMKqqOdMnS;)kC+=C>wuUS%
zE-wE4yV%t`^y)`U<bu_cAJP)KgkO6z_|4r;&Ujd_ZWi|vkd+(xGqrm$BUR-~$l=@R
zE>X`yaGq!j*~IQwUks7X2T1>vF5fVFncgvS#CQgWvG0I`h3M<vHc-j<0pxnGhYV|~
z1%r??oYSnaQ8LJo9n}oi7Ngl^G=2ahgK;8-_&<0FJIM(2gpWc~i+{$LUqh5YLt=N~
zv<^@oIx+xe<wTFQj>Rh?U*tnZUUIwY(!3Yk*OR+SHA<7D9wqtLNG4y9&^lpSLg@@s
z#`aqh(Hh?*$<zu=RxwJCU-;oT)iO=E6NO8ZuZS1b-$k)or4AO^&Oyk1hh0_f=g)`d
z(`E@5e^L&1+WK|7NWRd}868etQcqZ;xgwL7WePa4l9b^vr@*b-6QPY379@hqR{a^0
zyYeaA9RIM`8p9QDIg_L*OotD@45pj`&XfFP*;Bz(3BlYJ5JU3tiQHBv+tz!DKzo*K
zncIf0;L8fZAPb3f|6;!I5-jcBydh5*B;Tb9MSMt9i{!peHbg_Q(R}_sJ>4O;{OmV?
zYnjO<WDIO9{GCZA7Cek)b-C(rTe4V?tF>UHp~*z(!v}vAWHfIA3q0<<cC!u?tJ8#T
zv_#g}R^J`Drlt*htzrBlaBeN2qSm9tZB7OYi<YXJU(=xM{e4hSGPY-FKluCa0a>}?
z?IJ&%VnP}dQYgl`BqEK=dh31&RN5A(Y%c0|7lLgcHd9h?eQ|eJhuj=vc1POdlgHhS
z88{fj=0BFw@X_%g#Q5_Q$k>W`zr`~Xg@}xSZufYS);RvzEvQ!blnd457|kF8_bT+#
z<E0-8(M)7DKx4^oQgUV2TOq0QwM=;yPS?UYI?j@U1P%)t!L15HSDmr+cpKQZYr*4~
z-X4COy`vJb^A};~Pxg%1b&_oxd9vMRWunUt4#vU{*C!vumsSb#qZmqtJ!C5@Dkj}E
zjO#l(Mq}a`_Q0(6Hu$3&8L>8AvKun2AP_tJyp9u+pFPt|AP~&TvYnHXl!#Pl{4;5x
zJ@@dlUt_xt-ajWTkEtF3<qasm8bQJ5Cq=GJPCH@FUsVflaYX)I8Xq4&KRc!=u8!Es
z<7ho=q}N$;Qy8fLm&^^fpyO^0ROc6R?Wf7cS?AszmmL>_)&JaCs2;W?3lV&!|Ii84
zos<Q)e`oi`czJnw@{}!vto3qSj@TW?b<67_@fZUu#i=gQA?XLe4>A^sy-AF|p8bo!
zY38Iizy^~yW;>tsC>Q`PyKu5x9x8e3<z5Zp@oCg8sf7Br_F7#@O`L+Ew{rNQkBp+c
zd<b~C1hO281dH@S`d0T(*4ET1jlWI5o28<IDN}J!FEs&MR^j2{!5})%mvAz+tCQYi
z$>KDLv1+mn2Tn3YhnnZMr;Bl2sMccy0x>h4C+FS1^Lwa_usc4Z8_G@H{ZKY9uc!+~
zueQ1$%0i82nbxL%E$V609&s`sLoJe*$GsJtoxQ;wPC9Qon|oQODLeSIq{dbGZQIVe
z$n{qR2Jfk<aUrx{GO@9oq+zhq9<=dxh9yopMA^dDRARO7ZNZ5&kM>PnsoohDC)B+w
zQ4o3-8-zFgmG2K|_HRJ#)jm?ELRGr=6-l3|>2O?z5OUfmDOq7?|B1WwOkZcZr&Fxd
zO)%C=F7P?KdA%W2sxDg|dX){J{lcUEr|Q4xetA@VSg1`Ix_$8%{A_3bd~$L`+FNiX
zt~h}&BqUO4GsvG1Xe(24B>o*QRNSzlVLYPyofC>}Ha1&}NV})s4M|oY9W=wZDoe)4
z$;YGklT4Hx$#|{|{)h&d5{F2mqo)R&CE2#8|5k|{5r4hP7QRNTFjoS!*W;ydut3d%
z_@d=WbGy(Gl5GLeHy>B$80g$E!73?CW*n{TCu=pFN>@O(hn*g>ViHIK313lh^NN!w
z<Whz}D#!f7*_d4<APG}9W-Bqy?K#gZiEPS%uxJ>MgLjP0q6A5{RDnJ$nLIY$R~BhC
z1B1(g#+zgc&iqk>+gF`_nEK)KZ@hkNYDP|`p>9?y?e(2i=4;lQ9`wBmm*Qd@&2J8I
zqR`k(Q-k%*fSIQ=+Igg{tDwUGQD%O2ryMc%A0~_OCWeJeeXKwGJ^fW~SSb}l6!q&$
z^16ld!OF|-j(?qmw^SQ>n%T4TS2=)`cYiM0W|K~0X?%Bo^C1?;z$i~A+h<cv8upN`
zi^b#1e#LvLHuA)fewU;H`=4in<{j(#B17p{sK`k7iT{2bos_zL@X#!t@G4V%Td=*E
z4>nt>KhY{^6cZ?jzh-7;<_U8vD=VetALA>P5y9c1S>9e%zKQ6dtss`Z-D03E?s4}@
zZ3P%7w3E}-EoY$s)$?<l-!vz;g5;nBdwY8xvS(nA8l0g47u#nYrz0K5ptv)q6m}GD
z47;}T!|Gc5@Sd#;=Q#ee>c6qni}m$MME6qqQ00<O`RAQ^m!^%<_K+3qA<=6&GzC*p
zxp~M`I6J$b`g36tm!-Q6Sj$V)Nx`4xd^hxd(eCRdkWys8CKs6bL8ol$tIyg$L*R5j
zrcH}DS_srP74i;%Kvxgh1{`p(>k_lF64}9I;)wdRk!#EX3EKVLTO_J_wpj<hT){+^
z5^(YzqP&U>RJ=oJE0O@+GiaL8foHMbn+p;a&%fMU?ThJTa08k~x2#(OH|n(2E#0Db
z;1}Ri@H1XbO6Zgc8P7N_vQzLDV$1qV0P!M5CefKEJC4F;a{gw?kG(y=x1+D1&ZhXl
z5qg?vPMvXZD8@1txqF&9TL_eX@F?+xy@qMqR8|R^%Z-w@?;K7Qj!Vlfq8P4GNecL)
z+<9nf4XkK^pRC!uHC1a^{3|6w++gTR0L%xO@Gcg*iGIWmqr1ujtMWI~-TF5@IOY_a
ze-Cm}N6FxV*&n>(6D?I%o2JRSgR@#e4`JilN?f{@6z!?m#rXT?ayEp`dq0Y%3ThAg
z!q>DRu~2M33mtpC$-II(gp8BU<v<3*_Yh@LiUOqZLWJLDO&>Rw26KO;aC$$jJ98y<
zGSQ6(ss1;^@4A95yK(RK+i`A+L1Vpcs?DmmF)(>k#-lGD*UChxqraJ`O$4!AefbV4
zd2R95m*!8%!~LrKTjTQXZ8xv*e2(?OehH+!UlL!=WE<)$m8TN+<#;!aL>D(?%wT~S
zy=a#Ls8Z9Ous9u@vPxY3^^Ww5JgfDph5XpY7_w|3HBi1K0V2z~Sq^LopC>&7`!X14
zw$tdiAHpBUM@g|aKudTNTkn}<2Z<w!C((u~^hj1j{;{rzQ?gtABnwvnmgDM!cxl~)
z1_MZ^e0Tiohp;|hMh&D?tT9_(lo4l-y`(lU@ibs&rPVaMOe{szvHs=9dp;x9>%&}-
z;(lKPnlxv(O!3;#0`w8}`O1p;(StXS7OS4%L5vX_c<|CwqNrn~t3L<*R7!>}aWh(T
ze$$xb)Vi9Ap&&sVyR;ucx*^8ox4E<wKEqI3ntwV@M=!elC0>+#vk(sSQ#Pqhh>ik6
z$i6{tKKL2pedl3;zCkG)xI-8tb)4Z0E1St^Z+thY<`KO4(C2vD2IE;acb;i?V!DCy
zEp^+>Q$5{1+!zgN9snO%arK<8<_73{%YO-28-MG=TBXGLohSBr-B&5BB?4(9S{u}Y
zL0MZ`T8e_0P=6xM8Q$2^GP@SA3hcwd;hE@!ZnkrKO=B*v&J(75e<5dR{K!ahaWRiV
z5IDX>BL&=M6crUqqwrHxRNWUIvNyUzWYJ}A=B(i8f;y)2yd{{rdkKuctsc4*-YI08
zved9wwxHkBNW)ZBDUm4X`?=4Oz0ch}hb2@m{<L%)D#>{-yoiYKK;8#B+V=nLkHRNl
zlMpo_hs;B6q@`!@Mv_we_n?;hup3Mb?JtgTKF+2-24pjs+nfPDAwM?`*`m$uA_v00
z1e0vbnqo_SkLT;{PY3+(1SlyfX&$EO;!eRdV9XFT7O^lVq+P^rFW6T95tuCKkYBd-
z?1krutLZYFp=rJODKO{^zfH+Tc31ZmRHFT$Ya~^bSs$xWEcu1WHsl6RQ}ZuX1!lBR
zaYe<rB{)T~D!cQ=Oewj@KKS^8)$a11H}c7$hb*5D)t_E{I1ShBBn76{v<kCaqp__Z
zet4s__37fdow(S4|8eUCYwQS)N*9X}d;uZ;6SkAG$+2(>JOJDJJY@`%11ASav&~Rs
zloXCx4u6-v)Qz}fA+?&T$>0>7b5~E03U}3hP>8y+P0rE^!>St}OvU4KBa|q)m+y$E
ze>g*-7As$g9>ZBs3cgd99UuDYxoWv;A?fmrac6go;&-A)nD&cD@31(?U>gPX4Kj$v
z&zr><f2@^006DZ{*cmlK*jM@DUiM*vnmp>`pE%P-Wjo(197&T6HOtiC!8PEkmCc{l
zjkS&k;J0o_hLqUMq3T*y(Bickrc=(jl%5X--^6PzDm&O#QoRhaRC2*cTgvvw$TntC
zN%9~a)TR88S2&@@@5H!rlrwKWLj4L^nlA2+0lhuIuhx?R^Nx$=hTB~K3y{-mD!=B8
z5;2zm8=`*#WDKm@1gHhL-4b}+1LG!6V&xQe#rYBr+h?r`^9|ZGOcd!$W-JXO2p@tQ
z@6^{$bLZ)kJwPl7K88d2bzs4^p(mLO=S`QGB2hyFt%op2-NYgb(N!IW6wTsZ4+A00
z`V64pf)Pu~mAVai$oz;*cfhxujFgE_O}QZ}E4zL9*LEh6C*`%xOv?8||Hs4r4A?I>
zs?wB^U$phaoF_RY^%W9e*C{$-Zn+hL74^4xVxbSo7Vq1uhrq7zW248j7}9vl&3afg
zHQrUxbz8{2e)7Ji52-|r*H(kjsl}Eb<9LAR;>O}1Q`fBwQb-rI*YHtBHsuf0Yv0H$
zibC}$@nMxuQT3}C|Lp!HU4|YfjaF<LRbIvZxN?OOMOSmEKM}H{wo_YR-Cp~WMv~sF
zB$7x}B?V2$weOv3aYUpyT7?$K>F=MidPgMO@=bY?NRD2DfmvVhwHX)~z?`;%!h!E3
zds^HdL1@W6=K>+VrluzFH;LTftdJUo9yV-}!>7xS*J?uQjpvng6a2$GJ?|k7&@IP}
zjg9PEs<PNBj70E^Z<-jyv9=bC{~qT5`T)cyv9hveOc3nZLC0Gpq0;+T%c1Ww-*s3~
z4%7s3e?)68NNLgWh0|3*{~X|a6JuwAL^wy1IL@6!8lEKDmu;lSN)=U2-oazf7l7Ru
z&>fxJN0=bQ=(+2zo#Ju$Q(f)Pb&8{PwjE8)jcN)KgKD>&0m;_L^HU}|cJeMm&ATMg
zTMyIKYd8~Jl{TfYzDnRK)j`B-Fzm&P7qWS;9Kc~?TavlBx6RKiz%-PKcbN_SqA{P|
z$Lz6lKBj_&=Q(nSY1$`rCAqpi=x+#V*#{_JC?5)~o=Pt2hxs?gl`%$3PU7ycv4JCl
z-MUfl;0#=6Mju25`mD|Q8_tZg*7b}1yb>3sJ$3vw4(s*^p=VVS>kPvkr^$A7RuoS6
z`Sh44TyNHrk(F@)fZ{8aYcy&GtT$;`G<?g@GKW=e*M!xg?}X4nSP9A@U=#Z`^oSi<
zlj254+bWPluP5bbl0{}Qz#{L9Gmft6He-!d?$u`YY~7E7I?b~nc(3%p<bK}tMh@R5
z#LTbDzWM&=b(GtnuS%%k=M*)sr%w4PT}JmzHn1_hFsM90>!Ueen-fBk>Oz^Mx=zR>
z|A5_0C5HwrDmW!qNxgHFs;f8}H(5696DVH$capt(@|X!Zma)6G=3073+Dv?}T;oHU
zxZAN=xaX>`fDZf{?#-SqIB4=PeF&+jn1QXd(iDG`akd?_gnQC(b}Kw*<WY;}{8&*Y
zTdb851X8ahfGt+<i$5cVUKe}6u&XmiRR4XJ)s=irzx%~E@D{h<;ob_}pw$idu!W7>
zxx)f%YC{}@QB?}71$LGfr&&jD1>T8H4;i<a@c}(pmV*kQJ(GipVYkPLXM@&wP_Q>g
z+h*w1X`bBW-WF(4-5VSv^o0{~cC$Z!wuT+f904=3U9O~yha#3=eRkE#rZx_mqd*4C
zq;65XLD|F!HzIHbjG9)B`8fx!4P0C_MH2E%O)0+g{>!1SeDqwH|IO-+sM~6>`I^(*
zj~1OL>IpE7r|N3>h6zQ6D-=g$D4)cS4OCCo+cL4*Q2+np1SU_zn$Y#RaD{20{_9tr
zHei4LmNVD5I>e!M%53#((F--^Kr)w)R0K~wDb9N{ZI@aLm&N%_^}Sap%Ay0gQ=_Qg
z``crZ3E+FOBd74_^_D@Z{C&<G8C^IL>09}DNPr-;!UzenXIF_#=NAg%(-gfnajr$(
z-)2?5*RWB2u@k3i0SBdqYuq6|GgRE~095p`$>jX(yh-@w4h>hU(w{zGN%(Mfc6MKe
zFKow439res%*<s)Y}6#I5n_s~DkSWhQJPW$^^=CFT75ZO>Yg^{5Kc&sAIzRrr8T&0
z@{V=nYOY&FsBx2Jf~*PRt=wH?!c|X<|Ga!`v<%JcnyZWwT1@dgGe?V^r`s3kDHr$i
z^0-$h=uI$~HBP3<CGMB392$799gXB=SLgGv)t+$cc2)GdOfYLMpVQ4pu1#*hT12`Z
za{vcl1()xdf8vcE(2T?MgmRR0Di^BypX8#X8FqD?h)K*VA&Bsm%NP5YEp(DZ>%a7g
zLH80QdX&H0L5tZO6<9f^AP*?dSg3k;QkDOa2;L7PbZ!$4x3)rU8^-Yw#^PdPh3wnL
zU1|*W>gz=*Ru2CK=9j{a1GA@;9-ph8%wO)#AgV`_+(SL@iX`5M3u^KY2>7uTlvg@X
zr(t`GJHDwgIPwUuG}^J#wWU<JNg{!jwBkv34mRaZPAC31$2~WrH2@nX14NgGnc2oH
z(HrpXn4aAd_he@G=l}MRpw%+?QCF_&&fVKhZk!1(6kiqiUD1wv|BiezRKxW#C0et>
z3?;vTmI_R}yV-irE!|#AP&4+5+HE!qo(3U>6!O-qbPx#ksOm~Mv8{{}VRMBuI$rBW
zOkaGIKh|{zLP15HQv;YO36=;wr|u0ftf!)l5jD`ex7x_qf6sXf&7EV3U@jZXzvqpE
zZQWPfDY+Q75K>19)kQ&i*{h^*O0iJ)hTlc~g10KNn%&;i;oK;pep!#taaP_Jn;5wk
zv=^f~KFsMpT}b6@Li*AAo-`*L^}}yhbjxKEsef5*)L$Vb2LZ`uczMrO@<jTp>{?<1
zPgff1uvW!w$NdhmY@C#T)&dSFKlZwUAY4{~j8T<;JD8@(Bi6ZvZSi1>Z^$1glXpeJ
z6=0zLEU5@<jF4icqmC&q&Zw~Hp<x{*nr6(Fmgc9HKGCx<bCxk+6-y8>77Uy7j5W^X
z#lvZThG{AA1B{3g@^^yQDls4k?@nQg6rsuc#p$jVN9+EP;i#yF+KzPIYvWi*Y4lB9
zxhV-T2|A|fzt}@o-^v&M(jY0ZC1mZltJI_W^*`X-0poLXZIj@nWGe~6ELItoPbHic
z|4NLR=2Nb4&ja+${{*6p7SRurO?s4W8;R!x-e|oGGA@>xc^G^()JDR5e_49LZ+%wa
zR%W67`Udh!ta|@SozbMO(X_$DS9jFCPh(icMcRpPF`YduPRvOy_a`{s7;Y<nq<s#t
zA|z-%gxde6peIwad@Sx-&0&0-XW?+3emT>mr6nAkUSSTtCi{cfu{7zuqc&>WJbnc1
z=1a&C2cX&>OggfEv4Xa1U1>cZS@<W8?MEa@8S9?gJky7QH%n^T6Hj$&Z5Wf5#{Eo(
zB-cW``S<gUO|qO(_B&ae`75DiQ&YOBmAf<R>jZ%_Uxw7OG95PIVrD0p6Lk~>OUXsL
zy<g7xp<+;y5nk}>{i3xI=PIr2H}we_8STzgz%FdJ*U&5S@A3uHRQ0^Yg5vSuMl}nc
zsbyYsdado98&D=w(=iZ(gz>COPU)+$FKj<vI=QYC?}wR-p*~&(4L+tAUeB_GK>SPU
z9n-W!B{z81A{swq=~T6Tc)EYmDwtUGQ8GHHNu=Z;{)3z4?UO%$-hC4B337IW2?!Md
z(V)@^G+<cU*j5aMa^}?W$Nw(!5?A$KxnK&;Yi_I@GG@HzJm*d~tCLIgM|_C?8Bt)2
z1Iub~xtK`vdwc(IxKR)!`Z!x+%5)O^I7mK#1S;C-xz4yth*h(t8myOoz1$hY^$rD2
zfT5>Ar7b&9_IWqE(5)n)X+}Gl9@suQ&G7JIyH;#wyFB&DHuUq^AD_#ySF94dUo=AE
z^ZoJaZk8#U_W2W<>nxHD>q57}I3YJVUc1CD8ByNR<cTKdTGX>3Fuz>eO-tkBi>)Pd
zV1uNzmG;EQ%~UiGjDOgwCtoI4&=rbcen+I~wY)d)KpiC&gykgjVF%pa?*Yg+oDeq2
z>!rdetJH=2p{(FQm&<77GZycthE!!ea6)n&yu5cLnny=ogN3^YYT$YF;-#qWu1dnZ
zzj15kJen8g+j!ku)$gzS#&?n2K}}E!_0twMlJt9Vmj@fdrhzeg3}sIOn<H|wnY=Bv
zJx40V`YLCKN<GO6UeA{E0lnw7>388DD)Q$3`*>&+Wf{eJ{#r#Xik+`b=796vONQ?@
zf|I`{tAx#J(}(ra-_kShhmew$PL*?}nVZ;^D-X!;5a)gkZ+#*8<17)?B@%Y+zro|p
ztWv|LaZ<Rxr*~O7WHc@RE;(m3L@Me#yyAJ~>9U9Qq^P>zr*v)#b>KC(=4YHdi6Ekz
zbLHF9wOBq<``Px=#XezwVz#N{uh((Qn0k3;(r%p{;Cm+5k&GYuz!h-&3i*TLw9)K3
zXeebpzid*^nO_YX&l9$<ke9>P2BpXLJf3CZGoEmAQmz(1;`_R&&gf*V^;!oTnEN3?
zuhy(%!xHnna=NM_R|=qg8MZ|ykl{;P)1d&m4%>5;|IShLgzO#I7Tx8qQw{TlQb0bn
z@1sfAxj619FR*R?JHFgr*dnQtj&i}m&O!+B)Kxb1$E2%sMVkH-v=1pjTjN7+6~-EG
zkFUzfQcGI&pnZURjPdMYc5ZH%y3^|Rb;e6>bxcRxB=gNaEdY<%-%TPuZ}{kv@i}Lu
zJY9t!vM55*I576>eE@Z5v`426UTk)p4_xkaL;$utS1nlnf@Arzj(<a{U=vc+kqilm
z%_rZ3QE;1((W(s8Y8E?G?5$<=&8yxv;nvFhXyx5{v`xLRNS|M_UkCnO<FHvEXY#C9
z;^LxV%4-|gYQriBd}!#OgYm>Zz_~n*(bm<$^$rZ^n2MdnbpHQ(ft2)b9s7?B;)&7F
z=rYkE)MA-{)`oq*=tx#&FUys|edX6sCzRqHS}K9EN4T#Ho0VXEY5tdzg2GyWm5Doq
zKB_1+?;KMrT4{Q|yyno+k>>MvXHKr$5hMh*0bbq&pG)Vui8OdpEH1Hhkkzd~sXZbn
zq<V783*<Z=o&o=~E=TD)&d@|d+j=>wGit-O%8aMnx?FW95pPS&$4$ykS21$i3+L`N
z7Om$;v*6}6Tfz=+$uJh6_#an5i_Hglv>{WHGu~C0Z~jZe@lI#Vir)#E6yYMz(l85b
zW$6x0Cf@ZuoXMPYq#5J7rhWlFV{R_#tyzLAYsSTD9Vl0>D}+0_+|4NQ@b`C-SAH%k
zQ&KR>6ljBZ|1`Hz{gLj;QXGdRLKQ51fP07niUvzeV51IX>2Jg{4|x^V6%F6DalR*$
zjBHMuTB`q+R4)js+`kcWpPCdsL784KJM^uQnoHMDLU!CDf?Zhan}UBJpS-H7s(!7k
z6Y{2O5d=4x#RJ^C4e2=93)YI0cv21}dnlZsq_k!Bq+#}aCiX#%cy`k6(DbywW=XnQ
zIZ|pQ2PK`Riw(@`pL$;`!rcdbSp<!~>w|{~sl`Hzpk0=PHi0M@4?Ua+IC3U3#?XB-
zxa!vtP)RUC*BZHBThzUFBaMfFOcEE*O{V=`-l?-i-9=YK9jazRDb%1Fkj|^V@rvvV
zs{ffMEjcK3N{>SNlIwkmmkq(mo9Jbz>%Jw9T&y11BLu{9+miKBmTUDD=DCl2N=1qD
zEJd!K`syavo((Iuq?kMe#O+#@LqnBwhm)_nR^8%WVvC0u(NftUDz`>Pfc-k)&`#C)
z@`S07n)v2{d%?4{q;ulLz;YMPt9tgBOOkJJ-1mLh`G`Sv9Fm_{#`PBjC!z~eQR8{D
z6llm345b#D*BXK?k$SFBXz;qM)Kc7xaN_4-qw}n_x__$1_9alc=_lSf^qgt-<p1{n
zup*n6l@>CGe$R*CRTW`1%1X-l`GxwUqXNWkBDI9e#6W;zXsC~5p207_UtYXzL6MQC
zd)r$QE<-1iR^f79wgBn;ILAJ=zD<FXX*9*K@P6n&zstqm36M>++v}>#Vo+K<fK2*9
z|6ZHo0bn8%nw~4jBaDL2dxSnSapnzppf625XMHtPlF-o>{JQt*<gZ0HON?V$uporP
z!sfJo;+5v3do1vlhL4Ch*E~K}-U@ZG`A?<F_~8w*dE<aF7W3XU-GBe8FPE4uS4fwC
zRR8tG*}bqduj9R?@#tP2e5{{2eNpuB%^M$4{d-lwm|#7gaccD!J*Vb1u2bdF`4_|y
zv%0ah0HrNlwd#p;)701FKjMVHabPQ9D`>U_a{0rmDMfBq>I$K6lr&oU7}tx#$r3%3
zS*qLMpir}N=Ei7TXek(~E(e+J<Dwa~pwz)+kdQ%`e<zXYF3SSxq0db3_i_F;Q8tP)
z0Y-|(1QRh^onvEQ=FY9g+DuKj$Ti<i)V%*^=Mc{uX+w3Tsr`Y^9W^z=mG#;j3&2wu
zt2iwQ37a-5u%T>bdcJs3461Yi(kw?HOLA|3%%28t;6wueKd9`^p_-jM>5JTxx{yit
zz>dG`xTEbO#MCEEQlqRK`~8TV)&Br-UX8@Gss?yg_KFFjU3ep?<#M;Lqc!5hv2^*}
z!tC_S>14Iaz+Uv@8VhDhOYz(3{+cz{B2A~Ox0PYEvJJm=7vD{+HCo0*UgDVSxQ%99
z<?{5={E9kiv1Vn*f5x0j0t-r5(VrNj@8Za+!NsvCOCfMN&FbB8KUXhZCJ7)@!D^)z
z_~4qTSgm3@PsZq}O`;#;gksf&_<GvH>1Nv4T@|dv6`0uG-qVycQgiJzFf>p+f2cG%
z)Eg0^((1Rx^5j~p2ygo@KqW8v14F2BYwn%<OxFbc)RCRDfocRl<%k03J81{zfm@!p
zZ1^nRAh{^GKA?msAkSbFFXI8|M%Yy;D({#1Zw6oWGdm$)LEkvh8gInL=NMJ4SnDoY
zbJI(iXWe3`%=V|{fBd9Jb6yapReSFWQB4}#R>x&gZqgYNPl;TC(?=)Ge@b5XkQJ@N
z*QqE%Ic4c9#C+2fu4$-)HwwIkM?CI1@o;nJc<3%>6OU8$wlTHZLp1=kygI^8dGHDK
zgUsEYv^S7|^vueV{l!*?ev0>j>`^ptxzpq#;g-vmE>|rDa#lNqOUO27N<v@kn@ih~
zeS>xP%fHi`*r=XUR#9rd>}@ktxo|F<G)-otx6*J7v*C99W2G;pIvW!DrNcrL?7=*1
zzBW(Oc~vr#7h@~&&fXdzBj?t|Wj-yi2Cu4W@n)u1p&oGVBK&ep(|jm*Nl(DnbQ2%(
z58o?I2foHLPVmw&e)()kuAZJxZHHNXhAW*dy8Bi?F-Z8Sg+9*WBm1wDs)|Y(oudD6
znk9Zzq64VAE88pAlYHxg-U}*+9<<JSen<>t8EAMsTYg(s23gJk2t_&kNHBlfFfR?$
zFDqD_FOCt2ODZ04v=~2ZlcS$rN{ZP+qo)=kx?u!&|N2X5zhM-oomREQcj4xe{TZ`s
z@MoEKGkA4)v!h?qIs246__uJ`ba^=4$!sLb<6|4{jbb1J-+F07))bGTAXh_4d^gFo
zJB@i0LGWN~)PtwM)MG9%SnGJcVfSM7{&4AIj&(3qRE$R7NrX8a;Hsw&8$L_NE=Gi$
z9=N3iPvJJfA|SC(OLsk{yExk#Qr<G>Rm^4XeE%&{fk^A)5{{5zcZN5I?{<yt4lO<^
zN*A4+GxKk-UQ_2NEcglwYHR{ka~Suokn_`qX%7$(173f$O<@TMy7f;hCOuTbccUp0
zw{*(Vkhl~_q_j|wxZV8EKi&b&z?nI4QLOqmCiU{dTjlR!%!%V}rCsFjw#(gP;5%yI
z&&rVKtpR(dGjLK{c=?-&n6bMulwGGTaof3I1!}IcV|RW!ae>z?n}y@bbQ8@g59dgc
z$4BPC@%ArJ{dC3wD%Wf~e7}A|s4FW*POo{q2qexoAg1Kx<kUQKHEdbWzyuYXY5RUi
z$RT!QC8BY9Lv;1}%fJ5)BEE`JpFiX=EViqGObq-C2T!Wqt<?6wprA<N_LlOckEu9o
zXk>tl&`A5!^})=_q-hi<w}wkZ<z#a>Q3T-XX;S&S<9)f!1YWfqEmbX$I5;>c#&GuU
zcneeG%GG^dmo-<oCqC2}6Wp_^?(jXEKNp$$bdAg6omUzWUYny2<Ba;%QX|LAp?W7(
z_hw?}E3+n&<ZYHXi2~j(R;#STkJDT2xoY0-g#5=1j-tBc3C~&EupXi8t#tTg%}0Qe
zstB<gicI2r^IehpAnKdr=f+i0D%RNVjKzx&=#d!t#6T7I_%tovM0Mk}9~ymMwA0z5
z$hnD~IhG+jP7k6j#P1|24`tMI-k=?WZLD8=iI1aDFTJ^?rk(zwg4<90>r+d4YN(M8
z&4XpMKfnHUL70;$RUh+{3N9I~<@r9Kbu53>d>`(HCg#t&3L1U7S<XzIukTXurt(*@
zw#g1TtA=LU8N}pPTG4TVzj(<T#a$fvqr`$Y{QD~*%`n5P{-nPRRo4`l=E)hz8SJ_p
zsvV(ej3$Uz3~x=qeK?(#0Em<89#uX2r=)t(e!ScD4)7&j>HU&}PQ@MBeQ*QPc|j*B
z7z+`OR-FerVWc2S%S5v}W{5{TR-VXa-)x!1#z*ghq)B7c;Gq-=xH|ym91EPIucKE(
zvjP>-+EBHAX+Fw*94n%)+acE(w(O@;s87Vf5^JVLOD8fea0b4oCc-}i%WW}(3Ua1e
zd#)}~(RC2qRg&2~zf}Q1Am~d@R^_d*z9@0queO-yuU5U=+WX*7-K3N`+5<NlcK>{t
zatl&wEAE~c{8DyCy+4iv)mtW>Hoq=5Gr+HsAv@k$p^|@MeRh|p9HkTG>RIu6*UtC;
zkfm9U6isaLF50QIoa2(w94`*U#BOThSqGR2`8wnU>)~e-SFC6Nn^5m>PqO6H7NxKa
zSG5n~p;ef4WpY|9Ui$$VBM)~HBlPM6%GgDUE1?_;3C)ebSrg#xKEZJ?6MB?U_%Aa3
z4^)tn3i;xWTQs=AM|8F=Thl<=S1BjPLO$0>u^Tov%08cf^9r-*mp7LYZ#q@9_FE>S
zTIc=*nXL$$3(o?7ten7tDC_j<*Z^n{Sp0iXWh>Ss+a#~1VE5o1ui<lW?0C8glE6Za
zq94CE*>d`~e(_=;v}s~~e%nBchq3j0XuRl{;@|IXjlcou_FJ7W^xB+j4JIwk%{F;k
zx1FBmRSW!KN=o=KewBkB48fV7kJ#2_&b0&+K@gL7BcS-L*AEUt>Dr}!vU|{)yhW`@
zW547WzIsDyJy{*t9dIcGn!h##k~#@q)l(j_R(fFhMq=my_gunq6m}PTjSJ2zHaW6k
zo|<wrihCWfzluLPG7lz69WXUD&Lne(Rqh8UqDvJGGgEMeE_KiEUJj(T?+XBXM6f4r
zxUC9=S(;W(9f>DvJ@WSKm{L{ckhRnodo}#b<U?pK3gl0Dx1UYUwYa4h7N9*r6{cTk
zup*1t?d!+@Ut-r#clT}p-o2^8Bw5o*`t36sD+8ClI4$pv{S^O(-BRVZkoc$%0psrX
z!Z5!yE`EEIqAJk)^ZF?TuTJ`BO|yui2yP(8{K=oTn*6&cM&Uf7Zq$1txg;-MTHHOB
z>9LdX@W+gpK-ny8bCIiXr@}thPu?`zF{PojlSNpP`b#PHM`ZD2!wP-O0#_bT_19H(
ze$`~8ggT=6HSXIa!W14HG!}%<{`pvxU>&{}#mh%w*#C@DyyOE+a@?|H`2~x$I;9Yx
zL5bwzd_+u+7w)lO{Eu{pY$zSc%}k12rZsLb%@wfTnIP={p?&m5(OFIKeKK9A5}l*;
z+b?4uCOAHXQZTcoYQ~I^LnW~}Kh%A{R@9qB?`!wGk*K)(e&<_h<*f|skBnsbb!t9X
zg-0@_iCmvBZWkMI3^U;pQTzRTWTOYv0LP27(aQ-AAe1O)?!9gE)%Rx1O3J1O&Y*A0
z?M3kvrgWm_X_-qg+RH_T%}s25QpV8qBQcFre&`B~{T~WCTVZE3y)_RfxfpSx@0cTU
zqh+C#j8OdKDs86vvLTNYY6ErUip67P2T#k`D^O*6sUHa`?YhlvF4Se_pW5j+f%VP5
z?<!WE;doO%%Ii{o1tBD8@$sjz;a3xM9nBn0obg<-nw2Mu7dGV<Hd&)b$KmKk0ZFWK
zXym)h3bUcDt?SfrS|)3QpnP<{z{nbm(C!lY)F9rs>(+FIUi{s>IOS4!#HT>dr{k`B
z+ggHq*!@hdaN+!5n=ld?9{k-4xUG@SYcjavr(pN##%N#N8u!3nrlCLo&HoZ~huobp
z6XmtI>rc!3<NH}=Jp@$#-n)?!Z)(%Gc<V9Br0sdeOUvO6_4T-8c2^zxzk`{tT!5o8
z;K=)O3n<GpQ){W-B0<T}`=+KMB>}mOmj7XM${x(ODJu@vR?UIgPidB9px}PEvX6P@
zo5XYsMxSadGzn6UC$q==hqrAD*j?La68v)Pk~F#Ab!5;ob};|8Vt5=xj$w>Tr$c^_
z8is0>Rg{MlU*8Kl9$YctQ3?(H_O(CnKNk;EOs!19SRvhL^Bm3)*gFFb%>f>kxg_&X
z9<;Io6A3yfI@un<dUivAU!(FRC9<2E)De8PJ!b|W_Pa{@&a{+fm2`EVe3IsqH;0$j
z8?@aug<pxaYC5`+YLH&Pqf&4~0K5u{>3lag;OWViZBoibwMR~7R(VV7OX&HaeKByk
zI}vASXZ<B)&9TF6Be95Xez?+75~m=CeQMgPfRJU8FifF}Nwxzn_6Vei%kTz_8f{GH
z;&^Y;XkpvW>3Lb~hftPBzgm2^{_cD<UbY@uarBLp{kK!>P0cYlL-p->WsDTk<%u|D
zs5y?wwsJ7%setV1l=pmVz{;t2<o-!=o5(k#=dWz^n7?Y6Dst2_yeuCywVJ!p^LMXT
z<v5W4Btq8BLl$}!=n@$p(_&h$5KiPS5@7lOfx4p`nVGPxkoIYhd_!3DkLLLH8(di{
zGS~QFQdC)2IJvc7tB>($LgpXe=JCP7oYIszuS+R1m1skG3Az6FIK<@*!<6O{7*{rX
zL`x?fv!0*0D@<&?Hc^wjU3B~@k;^C+*dtGsm46M3e-q^othS{}`8-vRg&_DJth2q0
z1)bxs$>q~)V!uJrO~{X?f{l|R<Iig9_e*?LES^xSQL>{8_4)NmGIn8aOtGNoXmFZa
zqTUSid<m@|@*C}shxW$zmt1pwp2R1mS1Y5gJwf^Mi#^wC*jo&(`nByz%G9bMEy&Aw
zd7@xX3F{QF5PI3RaIwA*$_a?9Z3P}>!>2*HTOQWSZo%J01O&iY!&h+z5y?_VdIaCW
zz=^9&uE_gJ3wm;F|Fj}R^b{s9E$&=2apKq+tE?QR^Nm4>&Bt7ynP6FhwZQc@Prj8x
za$Vzcga|$9x9EDC^ztpkEo`n1zb$)yQ__(V=@NMAEUj6wF{qGD><y5@RE#Tj!?ITV
zOmTl#u_4O5*BIV4eGO~}aIol`>?=M?3dFSKy6i_7-ky+%$m&nR4-TZLxUaPO60n56
zR3}<on!kaNo97+1cqwVDyEmc}>I)Zp1??U%f~@HYG2`!XjF!^TnLNvnM~!n);y%xM
z0*UqX`r20sdJOvwNqT>*iCmQBCLCRg|GoJ}{*YG-3gv-<H)xasWQZ--R+bd=%Zr7}
zg|GN>gT|&{L#;=KV^5^nCNKLgPk_J{^yM0G!I4(C7y)s(laT4mOjbWj@yrC#`K}f(
z%tvnwaX_vep9$-*D=Wbc!q7qfis(QfSD*4VvU=^8X5B59I$oXmunkO{A$wtd!KQ8N
z5BH<&p;z`l+#A3>L)~N+xb%Q$l|YN8J;Xsmf2#Mdh-9g%3gO4B6Kes=1zrbXx-?Hl
zoM@5z`!z2jcQ$wT`$!k_bUb5K*3yaHYp{>pi7_K1K`s5rf>B~1B+v%f=>(`bF4y;+
zx8$0F?OG0J)|Mhg9><i^Xxw^{g(;qdS1$yqw4dNN1Oy1g(c`aAL2O*K6K%e~8a7>~
z2A}0z-eB8p&DA416y;bwx7LuS=falr-2}op>IA|5P-l$R;s?)!E2P6_mok^LQTAi%
z^Z!TFd4IFvzkN8=h|!8sq7;d(irAw?iM?vp)|#zdvu0G)jM{sBN>Q;|u?f{yYR0Ol
ztu<nl*lIud{_y+(IVa~N?{(j=`?~H(Gxxh&;j7KRYQKyOl18`|s&7{iz6EvM55N<>
zf;un`*UR<dcJ0ph{C|JN1^5+fk8QX7D0}axO3_oGPlvVK9qN@RF_aCvBw>O4h<Fd3
z;>uwQ>3fZV=^y)EQFK{~wMFG<QP0{GtgdTYqAiCwI9|h?|K1Ln|GxAZO3P{mh)?Pj
zib=9ce=^zdt_HEV0bDmok;+sMWU(f#5)Z0J5@PahYM|6V`K2dgVQvae05}Dlr@T>&
z#|aQVY_lm%%>;{WT}z`nJ)V^Nq4k3QmJkJ@C^+z8$)_cDr<FJLZ_r}>(rjwNt7RH=
z9@CD<M|xg`Y_>d6+R{8$Za1P#NF0|a`cjhzq4f$ThX^<iyZe24HC|$ycX0DZ21BjE
zQ<QA+19rBz`i(d}r#Hc?jzO|{C$cT#4w%SR#M4@>rD#a1T;MR+(t~{&8v=sfzhl#G
zJUNynldcsTCG_DikiJ-+P3_|oNtZ3JjVTsjDO!LDb25CA!oYwtdZ&DKinv<uJ~f6;
z_Tgb82<Ot7hU;&5?#_Q@*HO{ciDE4C%LmM4qdPvErB!J1TqKLrA&VNla;$0$5=MN$
zSY)URs?zGs^9A6s<8RlCK1;<S0bngPN%L)|C1lu^yc2!~9Bra2d{fm#<o>Cn^5Xgt
zQRQqNWn1yWSgGJcn@Ivl(!i9*M8sG(DCr&|vzLQy(a7|Ej6_}JR(9cVg-hEm)~q^9
z-K{$Gu0w`fLbXL&zw~g<Dr#!+xiqcM)^FcT!dnwb?0N)6pBgY;6dO+upds7X67&8q
zK9e69!vgA|i~;s!BvZXsa9eBQ=Z27zQ*qx+0jRRDJHc6?!1s9VBv7<J?lHNqa0Vif
zn!AA~M(7SH*^g78e8QK&x_^b~fJ^_Hn`?hHJVUy;GBh2$W~||;z~5{4zXdQf8mYJI
zCiPM9)<9Pgblo7}i$acCK*x4a<!a6Lkt|xe7Ij<V7IF%IstwWf@Paz0^C3UC*Ew(%
z9ky?qw|&B~gtS|V;<1Asd#mlw5qG#<(Roarb{}A0pMLW4va8EZny#52*qqU;4ga7?
zf=C4ix5z1<Urcy+`0tI^nL&+a1I^9NF>TZN!xdOF5|dI95fC6<G~66iFYi_{p;I=t
z1gs50PtSccz}OB`zey&o27Mm31hY+&Lx;p2=-wONSZSR1)+>VXcx)l|UUk+Hw{G1o
zH%tH6Ae)FzwkIrXFUo{*a9nn!dP;IuCg7Twps$E&jK;b8qz>HjG}?_VxTU@JPfyXT
z6}qFPMJ|>GWFS!HTDFR^1{y>3CK_ee0<I*D^T{s$Q7>{Q#|9G@44Cdl6B{|SerP3Y
zTG(_Os@1wXA@eei))1QFlj-XCkn>v7IHb!q{=dcBQI}VmpYX4r#)#;!9ZW9R5PuGq
zqh)V%Mh92m0icaZmQaGb+f61(icFHs13tL1BES82UkGAiUI6O1CfJoTS2e9f(-@M!
zmcml<CKmd$J%Exw;zN0qu{ag}(1!Q50nD)Nz!+o*cYJ3PA~1V1Bcsd$alW~=m|N_P
zY#<?-V2t%SG9l!E{@ZWkJ?-$hO|djos)N1I;dBruP5<<in5LSqSnGkkPCq#IX}=N+
zCwXm0iz5;V$&lpvX7(7a>RvkIvTaJA5t3|AfpeMB<*2>Y0NQp`WCT7Dv!~+LB9v*`
z16hVO850mv6`Lq}nC8CiIw;(VBTEL^v@lz$%_HGa_%!kT-U6ny?pOVfJA9}4HBMh|
z>_e}QPrA=bt{1ys{oC!C^ax|~v5j>Ny~PM^@XPXQ>^_*_z8o&O>h<hA+c1j!i$Y0y
zcTDbhdO9ZyxZG4_1-6>Wz7j6?FO~{W7Gu*JoNPi2c96vbQsWsVeg@iU`Ne~s;wjX$
z;S@YL1-~6g44A^Nzs*9CQ8<+@l5~YU<xUu(xTlnNu@*;cS5*@f>CZm6L*{ls3Ae06
zW%PETkNnyW$7Bk6`={Gdkg3)hJ*lM(eN%ACZ{TmGxQxc`=`o#jc2f@jAY~@-sY=Kn
z4@50mzpmdG9!{#SwM=cDPy<*TZM$`d+cU31iulBOZlo<~15G4)OgVTb$ax?0iKXqe
zHaA&6(Zuzh`Qw|o0E`sI6hL}LsDe)v1tWiciq-_3lG2kp`_6-2Hryl+_~{Unc2**b
znD@4jaE}(@+mu)!q-O7Ievs|mJ9}?$g!fU$yx(VA(d_LYJ>c&`C*-Qey9^rRp@xf%
z2?7aS)-q<7)ns|FYJxX!vJ^E-xAvH=J;q|OzxQ5-G!CK7ZofTvj+~yJR!~tfN{=C#
zKU5SsM1&`#?-rGiWG>!bJF%*82sW1l$ka`5M0u7?oX>C9FOXP5OBO`T=?ubC`Fv%$
z`+VV~i&!YhIssRS{MTyrAzBLFh>u+=7_I=#Wgi%+T%D6@Ut>b~vR{1n`NrB&y6{w}
zlm1Jwgx#4Y5BO3i2?G0v_I>cr7q=nb`3z=ij{_7}3+p@H{+Q_h-vAY|Fod`*M-qB=
zkXkTU8z0wUdzYKB^3T@}5WTho&awq`Rit)+35!OyPVKLjg=_4xn|S0A1LAxqjObn-
z@$`oq@BI%Z?rqk?gIOmFCPNJ$CI_U34FtN6!WRFnOqF=wl0{mTCma`Ni8aqAcg{b2
z^|>20?^>_5Qd?B1@#j9~61PPiM{QKBjCtwgWBtP(t#o&I%?0EC8!=O!G?jCp3o90?
zV*jAnk8bQUDScwLp7!3v+JCjyz^nlV))#OpAFLVM2o^hC8gx%ULR;JIEvOXdy|c@8
zUjng|L_5{}G1J`zrF-&PRB5U4F_{a@bV4yulc;y`@lc+(kLk#)P#(+KNe52v=97!;
zP2ycE?=ToNrP4BfL!fMZd(`P4X_cEH`I<g0C@#MiGB87#yNCj4dT5|dv|xfQPju}j
zC>;Gf6+o-3$<T8SBxh&V;S~=jRm_RDP*bE=bnkO^#z2qduTTA}Q5FlLOlzjhDfT4o
zWX^Vwl3sv$oqTj^x_|L1CBV|dC^myl{Fdv%WDuGV(R76vq2`WxtScBB2MP~&Ddn&X
zqh^e_-$X@Z2kAtvo@Fs~UH%qFTn!pUo=z`)wczyuLON+AeJmTE*w3CPtlXS#_1hkJ
zpV!rT@;H0-_L4+#iiLWbvsK4PCV8F`y<&f@jOvn&A8~+z-1n(xGC|v!Vg-|>3pxdH
zrdh_(@ZLbcHC4qd`&(ON+rNO^*J}3~_eqvR&$SKokCZGNJwzaZa)aD;`vJv_r;KAv
z^AXvU^=)n^mS{VnoCQTEk=>pdMXD5i3c{OUn0(KvdsDvn6Ga0jQfVDX&kLxrxCi&e
zG1j(_Lv@%4bFUs0iPJ+j;Wr)ux$+(ffSp<+I}o~lSf(sE5TvT+sYe6iRf~y=vfXG9
zW!$^C=v*XfxkWtyFli)`*3gVh%AIceEH8?>rmi5D^M&~r$kn5*9PWkgH=>O*wG7JP
zCwPx1lus!7Qs1^yJ>43-o<7Wsw;q0=H{othB0xEKdwU*tV5U6>6oPTGcZS){#iLu2
z5m0};&#klS)y@88L7UlGlQo^vw9lk=DvXnxn>5TnKR>8pZ2JqN3Q-O{1VuIF*J_6=
z7OaMN7Y&v_P*x~1teo=xyJ|Hv+Y!0%ca1BWWfQmWj9eaYc1)LM24^2+>5r3CQf^jK
zBXo=w$X-!KBIaPNNE|Wi?$8-|<Zg$UhWdcf<GOdY5?z-U%6_l~i@PMucpi#v|D8MC
zRGvC5ADp7ST@J4ew40Lio(uoC{xW8qQl7imt~=;*W+C}O1_Sp4_GH(f>TcLNpEXIt
z)+E?-=j;9gHy78fGfZT+p!`--;+m3=s6}AegooU$nYdQj`-A3rZz-JNJEJH>K+Ei&
zH$M+_6z59(^S0I~^72JX%jYKay|?KJ4LyR%7q28)Omqbt!Y<eL#-8zjJp)L|$200~
z#0hS`<AiG=blajVuZcPYxc7U4ttYN@a#fqLbo4=Lj*cTLlz29I?0Gc#<_CB9?r+~0
z;VWtt3|tT9!<OHGo+&C67zvakY=E(=j*-9J?Yl%ybYocxmjTwB4+ejrdyk*mB&L2`
zACmMrV!p}T&<DBCR0`rb)5fPgCAS3j=wRdVip>2KrY3imQxjyk+AW6td&s_oQL6%1
zFuyrtqHOsz%3wCb0#bmnb^FLl@1h|?hg1icCQfmyh$SCpJQ+(T49r5L9u35RI{n`%
z#3Je5fysD=$4^n2Ci(6zs5{1>M~6MrGX=VSA5?`hpo4fxAe)^;k|hV%{d755Ch<E(
zkK8`-t`Q=2(orDRtXfRo6hBMV-)7n(fxgGWs~9b3b33+Viw+wMry+HLO-8>6)gL6R
zU+us38&2Ac0C1SODHieT4%9MOB4ih!51<vRn<qs;4=TU0Q9E264{~2ERaWWiw(a)3
ztPmPsAO+yH)exb|d_R9^9Uv~pYTu21`<?XrmC1;qE@jPZY7wgxYjUu%ufT1iN5KBJ
zz@;w=i8RY<P`6^<YIRTovQNkl8b1d60N*7=a++8{og*RX6$tjK50}5S@VOzYvQ!-N
zkt&*rsxQ(N-O5uTnzbxs63$!o!sUkxH#}oOjAV?-uovn7Kmb)=dvh76ybrRf@C5(U
z+{+?qdUv;X4Zrp@OkIa^TS<lO|FZ-D95p~c{qOpU^Rv4WbqAb`yvBA@NT=wkOuvW;
zkeXL1V#%@~(SNNl3G$eai18Q0E~$eGX>e~>0$*hNY4kwTE%N>AGR>^%dYP-n;r_7+
z*}B5q(P-u04U3z|*k!j7`v>6{om1P+Zx=ec=+c;@m!g4~6SYrAL63mHF()v9Q}teB
zotLpQTZ`Pgcek*-5l5SWL`-V`^r^F=9SJ6XS9-#EXHAFHteN(fIXPj-A6~=Zg3PCC
zEe^uvLrD`oIGiC0{wd3K{NH34vFqyJ9(v5T^I@xV;Yz*|iY@SH|L_8v%wemF0c*U~
zz;&P01{0OT52p~POWP-1ZS-|ELM)r%(o%SbxpgTAt*;HzxL9d}$m=+j$o-2`6-4;O
zLCa#OxPer8(zNGh#Kr#An;x7+OVR6k`OV}=2h91-^z{11WrtEuP@689BWjg|W=*vC
z_V{96u)ED~)w-;RA&$DB$n^}n;QYDuU=ow@Y-_VTDWo{cUoF9mU(l^GxVyat(=<su
z0lQUC4)BKU?mk<1ewXX67twaT5kV^)P&S<>##*`DCOCJek&iIuvHUqmkN>~AK?aoz
z{H5Tz(Xo*O;?i})0^2eD{Ch^G-n38f@d-nkKg;RYUx2Up-!sW_=+cwJ(}T`=yUFVQ
zSA_yP`7RLbopOJWmvdoxf@rc-aUq*c0h<F?^45&@?}wjVyXkZ(mPu)CaYRa%)!2B=
z-#ycRp3CFIPkyow0u>lb8PeX9ZiKr<at;MzSWm?lfu{d8))}@rt2MDE-B!!v(^Y#f
zQT{Z-QX+9os>-rh4hmRK$V{(X$UFeW-$?a04y_>hAfA>5BZX-pv55!GCU)@_pqER7
ziLn#?<we_c!-g16gCFN(H3*A?7abTKveG0{NVT=u>xksYeczXLk|xz}#Rsgb6MA*a
znU&62|3kt{bEKY|1-TVa`@4K;ZFbrpM_7SAPR+~ZDIWF-X_{V<Tr(+9m2Gb4Ta~L8
z4;N!>$Tt!k3%oPscbA5&$2h*M-`98=w=}c)`?Dd3Mj4|T65=jagAmke8XvG>F9W`g
z4wKm*AeG3~V~)hkb#+)MRFPfNLtWPx;?>bVBhtu7#Oe6amYffdI=1!b*}NsEQ^1&2
zRBk2gMO|py(75Kc>qP70$^U^eIB_E>wJSB++H?h>aI6)r$=UNfx{w&2>6swKhTIQ%
z4DW78#=_F6$!TE-`9ynKIF!Gf#@IL3<U6>0srw|XyTx17)`v3F#84NiMpPicG=(W6
zA9mSK;>FnDpM_{kg|4gg$|k)-xgAGIH=GzC06#P72P;7~2Z=Y?cwrykOJ>P3fsHQI
zPGK{ubSE^?O_3a1`}cwByMwEVQJn|ap9^uTemd|4B?O^RibPM0^n{Yf>jGEg5zP_{
z4&9*{QstR^016&*Zjw0Rz+7j+$|+t!HU4!}@e}hRw~dmE{t{nYb^F<#QP&SJ6Bz;e
z9T)>&DZC`KxgzsW*;Qg*c^`DPJ^Fib<waw@d%Z=o9UUDZ#U+I$UUt8>8QKR|)9+E*
z0{cr5NmIw(x8lBu6BjzdPX|t)?}uN1nxF4WWQBdU%xqiZNS>ocw{-Z2KadVug{r49
z{w4*m-VDDs8JL|h3k=W)aw(8F!aIL=BKJ4kp48cn`F%otLj_IM*)fAb(fX3Zx%6rA
zIBbSt_W@3B;p$It%gM2KVEZ;;e|7rz+j%{Nnhx`S)6aPsq|p214fb1?gKXRnBVQg)
zzR}AO3(Q^8=TuR-GtAnu5PH42>Om@+m!4q7T2n8J=oUR2+cLY)R_p8R+ZNnq=-wNp
z{5U`VF|`;<nyC_JLx<S$YiPX+-aA{kq-KzlZOO;<5W6nsIplomYS?eNy}-}5|4N1<
z()`~$q9-M>U{ltjMv*Pj#|1SN<zLZ!XLrk~?_3pzI#$nqpPmbA+5M1#RMThvT|@LC
zGEI#z7#e>qzc<xnhEwpd%XV?GdJ9rR^29LBXp~g2PxZRDxl6P3=HGCmsm>l8x3;+-
zy1n-IRcxVSV_XV1{7HAd8h11uEHe;6nbuj&I_NI%SG1F@|FY|Whcq3W#Wu7mSNBa-
z;Z2vqFmHJ}fDA;zjTeSBDQbTFd?}q-l0>M<_7$LL!NP^gPFNoH_O+@Bywy$TGZ_Dx
z&#t=rzTJ0k#@D_mrLtD^MUa?`B4&CT`x_ni1q_ks!L+={rcm$0W0sg06FLzx9*_I@
zU-T$Y^V~vi#cX{L8(e}?><CN70-Y`BaIe}#2nSXeYpUf3rd)5cHA*d09sR>9r#FHW
zli$QS>@`l`_3g?%3bUDX@3J*Sn`v`x&RGAjxYuDT2FDv7U|+Q_nb4Ml*r85HUMi|S
z@P5<a%{NO9H^bj}1VlpsemI2uk0pbEY=4UXlI>5vr6^x?bf5m~zPJ##DRLddC1_Of
zNmgXBA`29CSJ-?O{`I6pB9mkF0u<`Mq1^e}%L6OXJvZ(a-8EP{0J%Xk8iZV`*&&a5
zz^{>_nsiZHpnrjP<R2d{Jg+5#8iU}kc{3Da;^SAQe)9K%<(<^DiEj{0TZe|4aO)~i
zbh-)fv$xo}8;ycnt1CH(H07f=!Giw*7j^{dCVolvujBK>dDzQwBwux3<gq~ieTw}=
z8B)s!a8lcpPJr72p=uD2V<|dN^C^1Pnmb*dfVZw~g2mRB`8~h6dVcihL({FEESJ2N
zM3gn=7+<iE#z4xkATaeP-AA-0uc~>dRtu8_V%7_T@zwf<YZLR^wf)Ww9UcA~<1WIK
z>Nf#uzKQFYC;QhaVm#1VzGbMYW0{$*rZaxVwZ)+^jNnGfCZ@CmH<SMAI6-w%R?~_(
zb44KCS%0n9aN(DVqJ@b-NcHrrYL&6BunDuROf@?}RfLGRS={w2cp=%9hOK6LXRX#^
z+t!Yqhl*^lK~mwqN5Sng{8kI;u+!rh?|kV(v?z}6pZ?t-{rML1GpY_P>Yr7$#;?(O
zbv8V4uvdDF&-0L5gi|5Oc=n_zAKx;TW13&FDtlEQOq13b=+YK|yXCgZfVg&eQCn`-
zp_pM(dufG8_D_B{W$l1DV+CBEmR$E9ZS`J_?VhWg=7{K_P@A&WtA6al+CF!4d^%5*
zd3?O-yH3YWh={=Y=PWrV3)XRZ%viXf$XNC2IFpf*4QXH7qGDUm9&8Zf*!H$qDb{%J
zl3qc_{BUYQBa6^Rwi9iP+$en?RoC2+qt<z{zJTa#d?HQ+q^*4I{WYtTIqrd*v&^WV
z;RP2JoANo#4%FFml{6{UHj&>qSb2-VFuz9cZTKl@P?f#;_T)p^$)3kYHeM(A7rUW-
zykU%bmZ}vFY}yp4rsF6&sJM`GuO+}SkDs&OiVz*#1aS&T0zdgE^-hhp|H(oWgVoqa
zJuL)I%@<4qWM{U*AN6p!r8jC)V0&KQU=_FhWyLZ)P3kwHhczPuiAtEAop0wQAn3Bb
zJ5J&Ix&$0+`JpLGEdWo>3u$F|BtRqa4C6E4Z0B0EtUYjUA}m1Rs~Om2J>(kXQbq|~
zHBL)?kuKQh`krLA!xa=Yu6yp-NBCU_A<8D3;2JYS4We)*2mSWa@((~pC`BHhQchiD
z2yw&r!St_<tjToP3bIzA9Slr^IE`t!VGee(^r!PP0RKrF$P;iDV?XPPU$e-nr8^Tl
zPw*33Be#mVuaD(QuKzk*bLeI3;rLTvF(B45ox%dTc`~YoW>>HmJt`}UUY!>4Ue-W;
zwsy+~e>Ul;%$f6T)$bK;ElYu-kcDKnBNHP?fy__%HnTFt3VG@6CBQ=ngx4A9pw)^n
z5PS^Bf*^jHYCH~cCQx3At0o8ds4m!J%3FKOTD(;i0_+Pbc}~XSNMCrM4bx#^gpq&v
z7z6#_ny*OAH30%dsfbRyhlnQC1PBb$>d%!BeS(DX!|5TTXO%4FnJi9r=DPhyO}YXv
zO1)fa&^<QgS^mY3Y566yBbP5vGPB{3n5d6<Y8EKDD&HSYVG7|_R@6n#3vulYx&z&3
zXq9_|`E(ckMu_hU*32L1Nxg#`<jiW{*Rxn0IPw`Ku=0Ls0TcvTgeG62zU33%RQ+)l
ztIC!~wm)xu$X4ZS{ls`f8=5UU!%{V$f}gqfq|UvFf|yh7VmHZ0nL|U)Fc$WqG4E*n
zVt3cs&CSjDKDi<0gcCNQa~InGUH-8~*hY;;<Tk!m(CA79zaWZlK9FawMj<RGigC){
zFoatx{8fX-A4H8=5iZHiN?b~hKmnucd<@Bas)DLqijPKBL0=d`ZjEXa9^RC{vDDhZ
z%s1aK>21@@Hvn}RW)V*N<J&)|RQ!U$<Kns-wf_l+k=#!QtcCviMIBbubZos<cvjtY
zwt2n~_GiOQ&MP>4#+uKO%OoWxZ=^W_@RqL_$PTL>aBu&JM7&7XPb%HAY0M=5p`1lx
zC7yiqZBS?~u7SM-d(XWhe7#!87<mwOz5WV#H;P1xzVBjcAS~{u2R>8o6&Se;MTZix
z9Vrh&18sc9kCCS2X?3;9j{s&-t2+xq)~UA@^G1%3kKY*@4K|SGlS9)%PR_15r$zJS
zLO53`7WvLk2nYENuY!^AW2wdKJq9Az;uUEGYnacIL=_(5bRxAFT+F4tY*;jmSsYFW
zo)m}ev~{0zU(A!Rgo|NB3<s&>c~Np0fd~z4x33p8U%iF#_XfVsL+oC1&td#YJS6Ui
zYeBllqEuXGhUS9iE|`wc%e;*}+%ShvY?)KaF)v~tdE435>=C2HcgKFiEvh$-Das8}
zhfO1rk~ns~=Jhy5xp&gA6ZMofG)k!SzpA^@Evr8b;9vfOEZ4rQBC-RLVcJnF<Ei5k
z2g>*E$^4WS42(#3;V-?Hl)ztBgCjK8n1J*D6AbJ?hy6aAC}}<Fv56*$kbg+fKhiKx
zhEX9@AIpr`E}pz**4$mq?EC(Gh|@$3;0B}Dp@^cloIQj{ye@IUi|!A&Xvh}Hsdsy;
z>wK<z5g6o0a`Y@thpc?wD%h@2#`t&@lgOj4Nxi1CM|)?US?Q?Z8RelMhJ$`wGJ`$k
zm(%cKLL2PEe>eq0QWfrlumCyq)zhXUxdlWGFX1NO6DhU!HJ|EFkD|q4e!BC&mdJxA
z*Sm;wHkHebEhFXAza$+|Z$4-gtV)`(q5i$P<88%4_4fQ-uA1&cYJ>hTv|j*gENb5+
zP@;smofuGnj>Oo{KbUqLDc_d#K+`85OjM~^dv)}Nr)aB#Q%&M)TkI{Pkp6}}x(H5s
z2-HeSGw5(Z&cG>|oyPq2n|Bo8N$^CQZ+CnXulGAR*h%d%l<ApMpMp}*2aTU@tZ$2M
z0O+5oCQy(A-aUSl^gvyo5T7a|6cz9EBl7!O=tq7j+yf4%lb7p#XhkM?6i^y3wS+7&
z*U;_HYy%|lQRW)x0-imxVTqTAk|D}8DVgM<S#Y@XFUuq=i`#Ecrb*&qAH=c;;%vM=
zJkrfAJKfpph2{b0cL_f8BY3PYU+fzZAnh*Cv&n$99)o4j<6brmmbTz<Q7x?ynS>kn
zfB;J;3Nx9)DNSr*+N_oqlq8YBZzVTjbyWM(Y9NrZ4(g7DRWHGejc8F;!~`c$Du?W`
zYY~Lm?9r$KZ-T~(+PtOK^L_(?g^3zb6QpY<7lj1YQt27H<yTj9-9XWsZ}(1i=C@jo
zJezHB_f-04(QCW=E<6UpGG1@i*U+Svr@@@c(dR9d(|#l6pvyc0ci5lf`Pq)aH$zWy
z3Bs|^lLJ)ZXQ}5c1>4gkf5S-ln$#kxK1rN!I}N_VbqD``ia=Y%)hWsUS`3&kQVM1p
zoNvgtf6#e2(9&JxcC2zveZ6$}_psz@;(7ynu5zt)eS~P5_w0DTScA{0vsHYkOs8%u
z7R*R`B{5vDv_x(#o|YR$Zsw?5zV({DxVqj~x!$b5oT*efO_)L;_G1q;==;HdROpGF
z&2=^66k&%64fF{nG;v?9CzYJfIdomrUhi>VFB6on_NG5<&Kv_9U(7&9k~&FP{sQ$x
z(o@HyJ(lc#+v?5^rsIV2r&S9jy^bFttf|+Tu=4MDyQs1V7slG7mh#Q_g!J(+IUjEO
zkeG9g^%o|rkRybc_(8Y{{u|1tOekq(<q<@~<xxo%tIScvW*j(rm@E@e6TB)xKeU9l
z4TCI`6|~zAS9F-aBqLBqSIy2j83MU_!4#7~fb4;+f8osYukCsrVL5r~Pf7r^40lDC
zLd{U9s+h><%{7kkpizQ`MP@<<#Q7Z%{Ovz=W_7cG48-Ysao0Sa(?oqPD@y(1Cqt-+
zjC*&xk0Pz7I3bj46oQr8Z|PFp1QtDI)jzT&ZTL0!6)ALu3Htp2k4<mjdWF`8TNdvL
znwGSIEzOH$;+G1I?D!;N_-FyN+-Y;eU`-7TbAXwtlY}Cn=B4vK5bu&_^=~BN)Sr|R
zv%fm%M*KT?Lr0y2DGwB=xCN`lyCwKG<XX`U)rdGFzsEywC7UxPJw+R2vYMN-O;qg#
zto|uasz!%~`E@FIY=+8jeH=jcJnd1UX3Fik1Au#Q-pR4yr3U~$JY@`saRdP3wM{?z
zrNDU~FW;nbTnx6#PWmLnhhnXTsPe#}(l`!E#BEpp+bcs#=Fh0q{~K_Ql95E0-brG;
z&qyN`5(g-S|A+b@I;0!(A>HZ*0EQj1)`^Bj3FXHlB^8ZA)@oxW_vSC(M99QYnLm|g
zaeC}l1o?#Lc~35Rj|}XgdYckh!*9Z<-iFfOw^dE3K{1W?(w!r&iPu_qNEApClJHy?
z@q+Ku4S=e%>*SbzY_0+ggHA;f>Gxg;bGuu))Y%qxZYWc7-}N9DJnFSI{2?ES`-~D8
z?9|@R3pw8|;kNC3usBU^IP?--MZ$_C>R4nkGYUxJ@<L6~>(niVT$n3oD{~^LhUVTh
zGy8CEcD5cxOBTizGa+&KvB9o>K7fEo+ZsuXKf0n`ytco)TW%DA4_TBB1ol`L@L73u
z)Lri$pCFc7BJbXe_`4x^-W+ktAwD9``QYVtLV(O&PFvrFK)bm3gN;UFTz&ZJy_AQM
zmw74OyVK&48y*hlGcER)IG;<b&(8LCj#U?m`Zi^1^R)8$;ep*O;;0&N_!Es7!R-DK
zR5=~5^ojIcX1F?0x%lhm5V}#GbE`zbGvD=kpyA_jM0Q}r!N~RLH6znC;ks94B@?x=
z&wZ71y}nb_LCP8)EJmD7h<E+_lVsl38M16|;ktspK2Pa7`^J6wk9di>9(C!xhd65g
z?0da_#n$z2tG;u8KF8q-n_+}dHeE5#cKHU1eLwF_f(V}C1?~Sin;R^K9^4?@HD22f
zZzGw>=7Xi2>g~J8QJ1MZ^WBq;UUG*i50C$i^^5Sky2jr8STw}yHOt2)=hZD$((r3^
zs$eotU}#7ybL>Rj7hlN-Ni>p$ag`LoUu$t+em`3&I-ea50wxvFW%wc56kjmO$0hWk
zY5AYsgbTUX`isfgO)*?n<PXz4eEqUHN;KNqry=TY$dlZS(csx5n4kzD>4gkOKDut&
zVp8mf@=p&_p=kei?RDr_crN1%gmqFs;4v4IL`9+Li0Ie3!1>XmSRPo0Oj6NT#e%O<
zA|l2#=kL|`!=|`qr#n^<NiJ*&=Z*8t!|e~1l|0Uyy{TCJL;Yx<Cb<{ieyc67)Zrub
zB%94P`+o3=PiOK$kyW-EtfExtpwpxgqq|XYu-e;EdDpmtJTEw19t(0}G1h_kb7jdV
zn|Q93jkl9Yi_kEVQ9RBxi6289d8^5)n<qHegq^)5EL`v6lzci7=i8*XqFf{)I2oM|
z=8-vSA2=9LxD{c38v={oXenw58uER}1W={=CK2^ER%tkpS=NdSiwkBu_{LYo5Qy<k
zv4N&NkSK$wxq*7f8%jSsPu76@xqiUV-4i@kzVS@jwXg~e2R!{DEJ3V!B!N#g5QadY
zYaz<F6qDJ5mnaZGN3J<C1Dry*Het0`BQnjbo%%Q#87ox2QlsvC8%u4!8L|eC;Y@pp
z;v~tAGlo^k_}v$Cf6kpgMR&exDl~f@Z&g^{C+Qbc=ntWw<RN#ffuw)OSwDSroA856
zGEm*8R-Mx=r5D5`EC|CQDT+bUcXFU9RS}CaxY;3s6GHBnH8s=C{N@+uT|a0|_07||
zuXmC~NA9<GF~BvYEi`%y4UJj`SNFq1tq&`GE-uD@|6bU69mrm}t`7_8e072ec767&
z>ZzfX>r91BE#Bwyg5YqqU(Y`Y*$?RQf3-JW;B!UK@`Fa+VTH=;xw@5e{q;(v!^M6M
zoGjJsFb5I2VS2K&aGs%ZK9kdR$WM@@V{xJ<ZNx_&Dp&pW7XyTuAc?%Hj&`4p%HoLQ
zOAfi6In2o%nwt<nI;BbNwti>71I(<$<z`9e;kaJX^iBtHZdwI+PR%C2p>pkZO_(rx
zc(gvLMJ(~kRyt_*imiLt9!4Z>4hLIPC3D?~S9|A+=f<-wz<>XoxX*XLb&$3V#I;_@
zKmILG`yunyzkd(wyUsVIMjrV4zS=lJTq7t!6qsNFE;R6%L{aTL+pPBK+{2Q9$s}+R
zd_swcY56rBv65C@-+AUmq$Tcys3@;+4iO&8C3=aAXV|GqbcicjsERn7S=w~$hEO@D
zNi%@?(SIAwZtDo!5-7dPHy6XAwv?G`W!5PZ?6b%jF{=7m$8I*J-?_QNmaAm$=I3o1
za{q{MZkEmwmLaWj`xlHw(Zh*W#`pg!b{%I-)hX|PFaIX4eEydpThJlr9r4MU66)MH
zHcSDaP<vh@18+`IHz{N(_ZQH0??5zJowM)Isq06tFe0VAqo^1U8uVdD+p!r`8-x8H
zL|pI0vb3khW{xlXc6xqK%f2&1UT6jqcqfjYDUDGAuaI$C;SFf5ntHMBNUsZz84x}$
zI-n#G>o`M)9#|jz^B3Q_Yj4vMe%`8}Y|VYQ1o{Wl5gZ>te0uF0;!D6?1|sIeo;<Ew
zF#Y)}P7@-$>)G{+xHH_@$BG`X9)xHoij91ie`aahdHg+NgzIwB2vqbjN*nEE^7M^7
zea}Z3?6DW2=4OKVZ58Oc05;czhe>a<kTFMLi(fe7$)|X@^H_ZFzj9LlejT^XeYtX;
zqP+JHE&d<{AiDp1dv+OYh(!GNm}I4I7D~)C(e1ZZ$0Kj*RA%Shp#H?E#&MU4-(**t
z%fKc0WUtixfr8S4cR>%05i&FOK^xfCLeubhgd^EAk7OZB0)1Lk)cCJsiYV9wciBN-
z42&ubNkZ?@0ZN@Lj7WuXb8{ZD$8I+zwCOd9s}dm}6FoC;zWS2<dJpk$L6Pw#*2=@l
z>@k=>9q?DXlTnk5QhFh4C;?g8@Ebf67#8ZceHJK83r9k%34kxw1IQZy^OxDV3bfS#
zb~n-2iT;}A317g>PKU(+N(p{e0w>Q0>YJO46A0}J&b9-*8(c^izS3zlMEkgYq_9aN
z`{r<SzWzeUR&A+JDk9{$jg5J<iQG-{OqUHBqn-JtTZMsIT7p&<w+vbol$eblS8WMk
zEzNt^Ji~~`mkpw|O@e}gHQ$y~9yZ(8JL;Ee$J9q$E0Gr2&)Ndgdx}Vii;Ig|uqIuD
zyZpDlxs^m5Uu}2P(4w5z4_5-ZEN$BZHf$Ep@T92}!%C<h;{15Kz3$NlZd*y~BxN6^
z{bhEzdBOA6@iE!bDoM}2wDeT5sO#b==SlVC)z!iQH$s+vzLr5n0?T=K-gt%kaw%?#
z8}Y}@=i={Wu=nF)h!CU5!8U^r>4(l34mvL*UVqw9;$zC1Yds3JTbaC9A1*l9(7c@>
zgnia#yU-DVAmD2mtX`X%Fp^sp71g~<$qT_KzSta|p0=W&cWnE+HQeFj1FDTD2l)Rz
zTL7DJZB8<#$(R&!aVrr?0;8zR$*pD4LHH~Cwepj%;m;I0{Lh~^Jv=sKU=ZK_`}ZR)
z^p14Fk-*@9=pdu;`>wNrlg^%BG8v@z$zG+LTP5o=#MwV5V)y0oWmvFs!)xJXq&TW`
z?q>gAht>w~(|Hc=uH%_4hqDVpRt1~bP}bVu{O>mm-tX(z1_yt%G~0pwzTx6|L1W)P
zr;KGQmt?};=FYrdlrOlUOFrBW6qNW1$Q3#<pQ<;R+ihlJ7g(J$Jdq8Y2;Le#)6dF|
zGoqjoK}72SB`KL5Z;rlEL+UWU@wHmBGCU!2E4@LfIcv0j+^?uATWVfdmyov!CD<w3
z4ILNM*2uko(Ug>VTTgv$KWk&*D$+iid!ci0&1ms2^>8p7o)u3J8)0eOYH#+Ca!YyV
zv;9|pptKQqw*{A0dP~QkOJ2THAd%P^_69jDh0p^QrVI4%4_OPQC4g#a(h{^8wKPV$
zPqE|+VKHCc%Nfd44K~EzflGhZ-RClK=5=**BMf#QzpcNjR*6{6^08|Rb@^NfG@g(J
zJkI7<1i4CU#+W7oBxtFSzZ3EjqWUD_Mm@JC6)h!!!t@e{6j-Znsp$gO)yR<6#W(?3
z#O$-#LQiUTo?vSMl{^|jO5TqYx>V+=pz!R3zK=bUfBs?>U2j2<g{*M?=)J7GS{Wb}
zK+}uc+#Cd;%o?%*?7o8o7)=~0jCJrJH1crbb6GVHTsm@KD5KxmGdIOi@;}}L(xV@g
z5VDJBSC#zm@nmPQBOqciRIcvfKp_s%Em+Ku^wIEulgvB@kQ({uHFqVM=6^5d7Qs-y
zbpNJO|4Ywa#&B+8r%)*2?duGlhL0j`qd))r4GOf*nk+(R%(Zvf+c=%ON=e%oH{p&V
zxZ4Fr{8_wa6=T*T-f{8;9Bn+@ngf+K?fg3wdaFbGtZlAnxY7r6%F0fADJwIJuyOn0
z{7{hW=g`vd$s$R=Q87zg>>JRbf4|^;zh|3aKR?>w|HA|e#(c|#e9xY#S(k90K44o#
zmx(nVg<_uJUswr!sbyH;fAbvJ$D@X+Qa9z0&JG9+z5cgQJKA#I;U9Bc)Phiji>kqe
zY4gdATMBMQ*I2kse^{SE7|Q>b*!1e)+X*NY$$XdN@h9Ez&``fuuYNTf*iB`)H&_n6
zOG&Bamt615_)L)>M*UyTagIvlWmEN&x*ej*KGtaQ)zxW=(O|m5k$T|}tA_Dgc_4c5
zfpEL%`C!hqpIfsp&KqAjRgdr_6~oA$!Ce?*-YP))+yn2ZiyjOJAt=2C&6`>}l<03j
zUu~)K9f-v$_bx?fW-DLzvlR|I-bsa<W`Z;Ik$#sOZgmfWZZ7ef)SJb0*0?H#Iak*d
zX0Cqd6Ffp)F1=w8=h`{15gBla#<Eqiq3&HabIwe(8sx#!t*IIp;zonj2a?9Y(kKA!
z5Z;t;(=;8NIIiS|aMb;}j+|<6pSev>X72Bl;Aw6!Q<wJZ=CH_d(^1I8)OT<PmJpL~
z`T7gbnyRjBF9=(~nJPS}M%ibwcoLCGH_`UVa%<m_Ty`Xpl!9b^Ab_uUAv?S6_}L@1
zV#4pUrRV0^<%!Dr8Lo|%{@>Rev<MCKS)!do3FhecED$4uFm&Wo=h;~Cyc<ms7<Znt
zA1;6*J=k8KOs_8r(Bodgv17AjUn~a+4MN=Pe&2@jVDb6mUm&lR`86e+M`@E1qp1k{
zjK4!R``va(XLnfr)yfv)Dw*WGxt_1vn(4YXNTOAE#VS?+%Q_R}V44IBkh~AbQyj<D
zu<8gCrbFn!%B2G|3tyVwW}K**fyy{`{!15RJLj(}7{v#Nm`=eS5nN+IF%)A<A>WzI
z6JeyQv*z&ujR|{PiH5nxxcWaz#<=$!F45s@frWLfkG_nBgew-sj{56j!Qo3k-!HpJ
z(3J+>!b(qr0extCMxfcFn~c%Xi6YS`AoE9mO$3c8^1n{RtEyZaYUBHw>}Pp7T8@wW
zjqy=WfmYvn06-B%bznU3Qevfzn@}9Dtz%BZSi}!?0)TcQRtNPHI%HAt4V^$3&z;9C
zt{$~Qc3lA=r<$$yt>y_7DSwdB4?({ob~G0e{{rqf4bA#=+xg%~W~on#4<ip}%EH1B
z9as>wTw31a^-#E5$ljj?k9v%S^VQ0~V@$`!MrE<$uSJ!I?GN+Js*h$Ai4k2@Er8V&
z#Kkg+B5~i%l(8-1WWiH1<ap!908~XuVPSfD9^V8Dk}aSeoF#V7%Wf5|8cs+@Jha8U
zk|@~8bWw=GD<o0i!gAQ&b34W(&U_G8D{l!7mOEmI-8bCIM?Xp7NM&XxU?Hh$9Y3@^
z+hX!3CE~9Wt!5VKwK_;~R_$<g(S5P3LK;{eKj6N0xiJ%zER!9N+OUl|10{CSzg%zM
znO(dJGIq7fet11chs|xbClc?UoSfh)rRm>x8F8A#U*qeq&*ZvKhqqk~0EoffucfFv
zdRRc^sDe8A2N<oq-@gr`%l)ErlC+OlbeY5!6qHkwK5J`bMrmfMtEF|4(?N|smF!i8
zlj?|1xDhLzbgMXHWe=}s;beTJ@hq{)GLaqj<{6ztYds;6Iw9tD0V<vw0Pf^(KU*9^
z%o!`ucct?JyXv+Wyyprn5PHy%tG5K?0Th`X3qF%Da~YGH{WbZe2LvkhZx#_D=<yI6
z=B%m7fo3c&#uhr`pk_#rmBp*Z^k0@+2exF!>EF9v;C0>v=0R%Y?oq&(lTd1p->C!d
z`!hhZku7Z}Zy!lD4v)5ExQ}~skegHZ;jL#GieOzrf-p^=2#aBa&^>f#V)PLxUs*tA
zJ~e)*ZyG>B89|s2ZYf|DDiLQ0Vh|t69&~T`wWyHG;(}4MFIhm*=4a+<ah%O4GBZ<0
z`8A{GI>Pt%oebEzuabRy@U;;8zs~Osp2cHxMT29247z@d#$S@y#4mOhuadh@dC%V<
z_9j<x^zxP_OCJ$Ku|nF;j3%Grd-<FmL5;L|LVaHnk84TIg-a6Xd3w4ef_Sd-D^IN6
zp}Uwb;oTdv<=01Z^a)nxXweqDBwZ@)dvV%jQLy^L+N!;5-mGW->{kJn=;w{5H``Js
zB!>m)282g0JvVj5d6Y+xu#d|Ms^&k)KE8x}#w&&uf4YJAB9Gj}s4%3V@Ya*YD<MBw
zMp>vmZbHpmdcN{L+3(D#a!)2RS1$_^&VTs_@WqO5S(Q2ExVKD%{f;_NRU6AARQ-v!
zTBmlTM{$6zU^PY+%cG`C^2eF0sXm4;W!6aHt@YMx`PI}uEyRKw9w@1F-JvLjRr0^l
zhSCFgxylhvy?ayi?v~&*)fYp18jPcLDc&Ij)}&L3!?5$gQ_nTbOV<2QjIEf00GatD
z3Uz!OM4W32^Ci6n$N5VXpx{`0^NNgw;B4UAiLAk#5ktfF<~EW?=Bdxg;o)~7wguYC
zqTcAY{=}kSqlY1vM^lJGrm2EZH=mI0F)g5Z9zxvy0r(8ZF0!wmm-L|HQ}<%fb{zM@
zgI3>{l6hR5wz(s&v^pkG^2e^(&2>^Gx2;H&+auEE`V^x)J8Ng(aV$jcyfEU{fn#kO
z3_cP2d;d7Unfcl3;V-ADow@eF>#KQp#fS3pxrTzm<)xyT8Gl|80@mYDHGS)zvY61L
zO@yh@!-I9+6lf}6<mTB?{|Ct!eCmgvC-<3NZ6RgI&we#|JmC%@b*Jf+{jsScB)K<>
zX|_w$^U;sb*X(qw`jUEafA?1h4^HXLV!JAh?<P+xRsKg}bmZOJY^PGCEF2-&u`!_`
zZ=T0c0?ecOEImxqpRHrcw?nrv=oomPFrrt2MdS5<>Cqo|wfNoFB7|oxZVODjAjPk&
z0HX`7AzL_1SZ^brx=8x%HgCDD_WJ@#1<TM|98R9x=UpiWA$hB9{u<eC3%@_vNyX)w
zSeRv;8Ya$mO&@!y+!I;0)28V&6iCXRXs#tIpAHalua%<sHrnc))s!`u7{jacXcV!6
zWqr<>JX7lk(x<0kLBh62EpWHKJ*}lnOOS}+N-jCmjex9VM2n=tp#9c2gs`e|KLsm)
z_*`9VpQyC5m+gq>XprU-80tv7WQeS=E#7lv2CbBH5)2Pp($0=&F-z<OI5<z7{w~*}
zKBrbeoQIVixBGjeQT|U9-D=B+5XQ8RXq<EbWplTM!li2ph>@2Caoy)FtaK$j6r-tB
zbp5SNEDa6P*eL@Es^O3H)UpLbD%Tz**KUa2xW%($2hoadJ4b~g+-*xf*N|uWv9t*p
zWXnZtjAM)@s^*Qd>hLu(CQzybgB|hiMk*4f6OBUMvEaFpeCwfzynMbcBHJJx2mqR}
zIK{It>G8qgi7zCued>2i^exOCcX2fSJ9MS6!#A*Zc|~X^uryW@eX_t*CGx!oDrC+7
z4N`o-o5td}hW_lR4RLjmkqqy9XYh24y7i`@ES3l9MW^|F$OS^NoUGm8{1D2}6Kd3l
zN`zU%3G=@Uqdt>^zVIQUa;4dwp4K?jaL_ajkgJL~7I6rF7^44=mvdxDm>;UN05YVh
zm2v)IP58a68h;Bj>@|yL6Ss4Ir(u<d{dLnN&qi|PzD*(w$ENi0aA+dqGoN*R^WdtV
z?ETo$n~*G*>EH-+7U=$klu^l+va2#$pf`2(?O-*eB6U1KcB=#Ndu8RS(*7bw0)trI
z2~c$UJ>*t!)V`^!){w^?`%Lr-us8Oj?%~-lW^LSR8)UZWp2y8<R;y=7ogwqAglKUQ
zvWqP7tI)qwR@s4e#LgR!&F;>`*4~8hIGUSl{WuYr*qATtD7pBVM5^%St{t?Y4dH3F
z+C@#jNMCE_)ESTP<|z1$^2UgpYQsK&N)N1_vZjgpzXY3s$-aSv<>Il09vpc?ZZWa(
zqtlT<tQBPw)|bhNGfhYMFoM((h!sUww_DscrlQdBQg`d4ct`c64VgFC2~?NBDFy%(
zQR;V`&bsa;mJ;2a=-uzMOBb1W&DL!~%2zO2QC$niUWo0Qsx2%lIz{^@y=dBvsIFNd
zMRQZLBu|spn(#|EqJ7)hN0@L$pJ>L$A6y0hK!gl&qJo<|QEjD}qtCO(800qmQ|zK$
zd9(Z^o$;4sb*q>9Iu#+Y6p&CD4YOhakdE3ylMclqvFwAQ)s6laXvO9_`PK`0{UUbB
zs1K+_gdvjphHOb@#`KE_&OfPH<@-^0Oc!5D*Fc@s7kh(xr=5SuOPiY{lD{@Li>-hd
zY75g}_V@RvME;u@E(iuhSY$q5-`!vJNO}0ucW*PG`Nj~eak5e0OYbYsXiE>+L<5S>
z#a%kL2#I)X+xT=j^>au%OYpN4biRR(fIvtAL^A%FzRqI+@&Ty#lEa%AzKff32y(di
zS{dQ0zXq44YUZ<oG4?>-GM>c4-s!-Aw(WkabPoT%%3WQEckS0-KL?NhLM*e~WOwCS
z9uwMpNdE?MAko2DZO5pC68VfmfzP_d(mj#S7>S)I;18Az7#uIaXzp@bGPdUSJKjVD
zY+#;4k|JN8gfD^skqO)2c@UN&H<zR`$09Sn$h$9rqZzr`g)30bUliD!IJ-Ch)hoT5
ze<y+jkfR8CwWua4Zu+L=2y)k>1%_`D_WSYODLD!aTtAiz{kM}tKFZ7PyipYv0FVyo
zZ@MlQ^VCW={KLQG#8F#(T>ZE4u>?U6Sgud0z6C(Ku^SbcfF3&`qhFQWcx=j`&Rpmp
zX%F~Q;jvWsWru7e!W)HBbokQAo5>RT&bWx?+fwZjF32v}qMmRNQamDCU|^h$&qs;Q
zR=mWogwXJqn?F36%g~YM{?Wry7;La;NH|~&yi-Zt7b<DNc$+L06AXzt4M!;3XPQ&;
z$Rku)dz>@MMgEGqi9}J5LajlM;NKOVOcl7I9pAw!cv6kl!d#}Cz3ir(lFEeyA{N0;
z9d)y_v(LuB-vSitw=}<tJU3p!%`|%eoIAET**!2KiaAvxjPH13c2x;&+`qu;%Zr^R
zQ-{if2|_dh>O-B>fCjM!LqvuDTKx0eX;3fN=$#R3lEz4FIb(TdB)G=Z^%13xt!vP_
zHQmd=PLepo3uEPjf3Z|5q^Xe6#cgS;3;Ag2LC%0K3Rz0ghInYhUeXFDJe|dhl9<GQ
zksW(D>S4EY(WyvagbL62j$@^jp0Pf};O+S1xan}g7RO|HT9MiY#9(-rttcs6ykS7l
z$Cf5$#`0Lj_n$(E`HDfGciUPY)Qbvna@QBM&(F?pp^vZjgl9ZI2fxqVLEU+5aVv)6
z>8PTU#P>&OYf5pgUUCHtz0Sr$!f!qeHq=$gy!>qQde2>t)!93)IRSY`QHN=IG9q-F
zx_&YII-9g%x+pB>A;hu=FE8EOeU-J~#y}XjG-f6+`4i<MH{>84!N1X4(yp~;{Y`O=
zESB&*`E#yiY7Gnz){BSgWImb%FN2xyMVOiN3Sh3*IPxk<opjUeJ{JW*n`lPbYFxU>
zK!fLB8>X~iuE!1+t41VoPzSN=;DGL&<Q8O?B9LZyhP@EQYWl(t>HTR-8Kk>@MDa7_
zGJg!P`!N(VFeKb-T%6Mf6m*KWqKEvD*F3H1yEy|c(`~@|0^w2QrC)0RSZk@T6Bnw7
zYY#@N$%W{Wbc@~mZh5WWFuX+uB};AU<AM5edEnlhOVDd8CT3p#Q3C)}&5*D=i1#G;
zKJV1?4}OS5ZRBU6#Fu%TZXH}u5#ADNMsp?r4<kAX_MfXQ;}85!>+@tqtdldtGk~f>
zi6{U7ULO30Ub=B~YHaM5c|t?7R{WrQwl*A;wHh)N*d-#GH+fdmS3T#NH`3lLh4cM;
z;0G;*G>qsieVVQz`wt}o(5~7#Yeu2Ej0{Y}hzJs{Pq#``5Afij(r{wlp4CLtx6z!o
zIO%{0&e&E<Q63nR5|067uwkRdGp-`vFkO2hFr-*e$K0yQ#57pLTsoq|awzsuK55b2
zamueyP*e=PNZ5J-zD<)6+y-d$QYslCi@KpyQnee@F?<ltGL)-dsnpRSx-=-FG41IN
zMOcd-Z??Oxi4JCWP`g!U<c?o1^Ey3Xys0jlSeVx`+deeF68c-4sv4@7cGMHp6?p@i
z_KAlv9_kK^$acXc$KH3o*$XytB=muqV~Z84P)*K7#*=-Flk^cCb9o%<3m_@2cSjW(
z-+!J^_=ftSp!i=Ebq8N7!E%+xaEV(`gkO+_jR)q$CGUKQA6v)pn*B=@M2|a7{BGKD
zgo97VBr0;b(nrQX#rn7JpN#YBC*7BCTp#+5&aD<qVil)u-ng@6oA+d5BYbLRD9nwg
z_qfJp{>9Wu!FCn>x6VRuoCPP92PQC2=I2j?*`I}u*)F9hCV`2BDB%J3&$h9kI7DQw
zksw-UtYvF>jY1XWUZE;dnm`5`<NICHC(db4wLdH7En5`Ge(x3}=$WroRVt;dGadh6
zqA1vDiKX3jBG}*#)l<yn@9l|+WPca?A=Jj7rST@g;&AU^V)}v8>pV~(?EgAD?{BvL
z_>ae|)e2Igv_!3-DztXdh!qsEYS-Shwi*|;SB+S$O=zvwu3fc9sVZuP*50)@-+aF3
zeE)%Oe$2V&<mBG>eUI1c`FuQcD2r{3dce~~apur)7!4697J)yN{*)v)8dD44@0jV0
zc3Fspqq@`g{Sy*;q5TA4qHpTbP}B8r&+fIY-*)ma*MA_*9$DW@TV8hJAkr78U|ofT
z$i|9kmwJ#;ot(kfhE4CQt+pq#89C+u6&;{$irO*QpB<*(I@mo0WnXNaP=W4~jY_Pv
zkR?bDD_X=NzzYe;xBa<<Fng<;2K)bl*au-yk^_(=8<Dnc91I+9Nq_wL(lP6)rfshg
z315GdTu)Z{Q+aX4;RtL>1P@=dh5!Cbad&@LIRPs4qy7ZqKMw`Lpeg~D`zJm%BLhw}
zqyOYk3~c|TdX_kEwkZ67sBqnU^3}I3k2iYvS-h-_+F;yMPI9Y}uYhnY<1NdE1Wd>c
za*~Ded-V~j)O182+*dR2BIi7_yc>idVLOeDeEf84XNuO)Gse&na8euxs^+<XVh96A
ziG{&4f+K<(-q*MGx@t!1%Y_}sd);b_x`mkNSj+Jr#~j8fWcdVI{B+T*litSR+T?EE
z<gUtuC!#x%`x9tj7PxHwP|~L(*W}`O*E7J3F&RvnSW@!(v-Ep@{ZFFmBI{A{w7hOY
z<FO<*-6P_q(iv)HwXo1h{|0ZL$J`2?flcdmJVOm5d`||d2uJMg9X62U3bkAfJ|)@W
zR|j(d(cX+Z6@|skkA<y{bRIMxEv-{Aq`Gubu$O=H*ff*se*<E*ZgS>J?3Mb=XEw=5
zmb3DabR%|3MqfL<EgLep98LA$bov+Pq3yZG{j2HAWzI(%qfKnt_$Hh*i1y<r`KHQ3
z`uns*3anZwOc~YHD%O-DI~LIO#&{SV^9I!d6^>Yzk22H&Q{&9Sz}`5U)$-koMI!10
zRo*+<tNtA6Q=n3*{guIO3Lk$D!hd^P*CU6YCNxKmoL0|&Dg^6sSfZ0Gs3xsMVcBKz
zZ&PoaC$Dip4L|Aq_RbjIYsOf%b8?8(%-ph<H0%7}JpHVMU_x=;xm%n^@~7-yi>EV-
zT$nNUN*&`Zt4u4D<{P9?$c|OXT2gN%N9k4pduQWKJ^)eK%{y+>``NHO@M?Rycq8Ru
zk*}#*>Um5{N1y)$Qwkydc|c+E!=q{E>0c((gcL73*4H~kxf_xZ2C-|KPUms@X64iE
zlCH4`yO@0F)KnafZQqwZ$|{bF7;Pb&uXOZRHHY=x)vOr)Y{<J+EMs|C`kFOyEj)qA
zl;}2W9|)vTv9z!=;r<%sx$_O*BfGoXcG2zKdOo{j3hcdd>l`e(m=NiYs7Xb9@`oNP
zWj(YUEW*DyrS>4E(}kc6xZK0FW6g7Sy`uU_L<GpV!#dT-v+_EL2Q5d(pbSXPe1QSQ
zO@;J<DjHa<p@gO8j~k3-R6+PBk>Bp=FpzK%^rur_?!f<DZxlT#G$aNQBRV2oo~(xT
zLuI<3C@G~?qM`SjjxFnao>4A+C~#Tz-Cc~XtdtbYpiN7tMEOwMnkOMq9$7S&!Cdvc
zRebtB!j?6y6ftu@Uzs>)mLOg`-X30Ym#;B(E!Li<J$$_on^#?3aIO?!pyP6#FWmGE
z21qF1<rPR9IDGviG{LtAy}*YYU$OQ$^)Q+k?Fxj3=L?L2$^KdiOFU^6`?&YJm}6MZ
z7ud<&5*>s&3=ZTIg2h&EzmUv$<eX#OZMb3V_4+jjebMRQZvo1p&VUICTSBqK-Nz^0
z+5in7-q)HRg%bbD@9mZ#fhtk;c7m$}04Yfs{PgVinDrX_lC{jH*Q?X5<;+#5PPZt=
z(-#&BO5PdSmRB02B8ppCn!lgBtVWDW+(Fugf-(_y<#fTFmG*Zmof20}=G@V+uEe(D
z!AM9)T?x9wILDb&MY%o6^0xIKqLg-a1IP4*wU#@S3Y=Zy$SiXE>e<Lbh~=6f0PR$<
zVQ{x|a7`#PU&j8refu>g9LyoYQaY&eSoE;pkmY{9r!PXGkUjqeNS1@Yeak@}gdkmp
z7TgMWs|??@tMj^EJ;0ApzIeX*EQZ6Zd|Uwm=ZX2g)=f+_nWzGbdO(pEeemixRp|G|
z{cg#Etz{?4xh^*zmheZ9dMVQl0D6A$uaj0&H(N>ui;}C8tBdUWp3}`#h{11Z3K18J
z2a%>pL?wOM5uJ9aZhYwz6;%GNVmr2$-ICn4e!5Pe__?~SiUn#hSVD++so_(+#xpa&
zg<{)7#k*5{552u^u?1W-Fv)lx_FMWIEc*^+V()(xQqqF=TgdbYB=NbS4S+<J9B;(k
z**0L~>R)-<^-q1j{&spw_m;_@t6w-(bai}Dp<|JV`+a)_lgqxRe{D@SwO?NooRz2L
zY>=g)+8#FkeGWwh3?>c^b@cZVBv-Lhq?T5B?X@#lN>EAQOF#J^$F!Y=ahCBZE9#8o
z=F@!id%>AI-XP-<$<C^i1i00kl=nA-)$xX?$9CtfV%+-`a!Ft*<jaiV3kmX&$Z=>5
zx1zPVV>utg-`^QJbec=pZm?{)iQ|Jr7S)J|uLHfkTH-mG(Rsr$Aq;Qpv;;;Z?uVc(
zDJxQ&Bf<2{7B*l_B6i&|_rTarY!VUd_KSG8Y^h?b;>8L~{T`^B;J+Mp-wB`~0vYQ?
zMMf_##b!j&R157lH2)P6Us1HtN{wUWB9*L?q{GH}f`qt>X~PS<-C-ObtBMxOxJdPG
z+r0U$WKzC?E#wPWZQ^I51^0bWmuGnVIAEc=-?|!R_orOBY6xc2PfRiJCI(&vmg3s{
zCiKuv-l}=wM;aattUlmyL#dGFgVHKY6*`MPidYhKBy0^jzib_CojRZYR)Kz*_S*VP
z_R?1(Cb9g`(<IG0%tla-JtQa;lX_dkOuBiHLjatVPDA%J_$CLe5JY+6Ulx1AYe+Yx
z5`Q*mYfWSRw`Ml85b{G)OV5hj^;;djD7`oRP~`!Kujz9Tup>>)e{8+&E^9w+gHcmB
z8*DO%3K$T3WWeTIoyVs~^H{=Lm3UB`h%gwG!!g$=gYFYc*T)5tU|rDX@gsZoN8Qhp
z#c}(XDBYH00pSPDxoqKi6=Ig(f+Jyt&}6;@&u)iri&RV@xz+i^S}7y%IEr*qW0I>!
z`x{`76eL~&C~ucJW5F~w_1w0@s|ed3+tJg^<j<hI{130<n|Xv!Ydj1@%?9hfI~>3G
zl&;qP8&-9UBO{|+rQy~kdrGWJAOBH{;c3=x#<1F~Cv$}Kr__NKzxwrx$rFM5mSHzR
zM*-B)hp!zIN=ob^g8f^o-NDfHv69}RcloftIz4v6K-=p3tHssl+S892;$U7+pmt51
zgVkLqq;b~0rJ<piC<+QncezIfsrtGj_5S{i>u)@QO%Rh(Uk)~@b52jFa}2lH^iDg^
z2iX&gpq4*;Jg&2A1z|@4C{%os^`y<{BG+$!{Lzyy1Xt)<@=LW)Pzj7wCBTOFRtSh(
zB?`h0;B<7z3jh2mTb_Ga%@Je2`_oqTuBCXR=Z?np!7xFV7I;45s1cjJt1SN7BlZv;
zOCa4wG33>G4yfFExc#5RAmlOQOV(O-)CYEuTRby~42ZGQfRxyEzg9R1+^{d8mVbN?
zYY{sR%h!S^1pd}%Z4;`FJ?Z#jRq`!;nph5?FyAbQr`+pJW}c2b`+Bi+ZNH;#!8uzY
zQt7^ZqCuCKXpoWexJ}=cldYJW%faR2`D3mOXOe95FKMDPM??R<eZILv<FgXx{Ppt)
z|HDS6amDbP+UhEW%F*gkZ^UJiKCO@aE2#?4ynp{M+qOqMlbG;x7uQWaB8P<22UH&h
zdjX&k$aLTKZgqc#hO5pGv&bXO%UEWp>wId-n8RO`e;;|0J+M#oB88D1ixJjBPcj#S
z-6SCiuUZ27CKBu|#6SJ)NBI`97h*w3$5-1>ac}&f3Y~cIo(RKOoiiRB`gy1`+H=e0
zi1nIvI@y~v1D%8{qrZdUPF?65-`v~;2;T94Icl<>t96Ii7mgcKM@x?Qp2iFc9DO8M
zdC8?R=&L|Y$tEXm6}fU>IHqnZRtF7-*Mjpmvc|%bl|YsbAP2>K$Cc@{PXtaspOU5J
z;NOEUtm-(OY;%8mN8tLUYX3I~{6N|MRUH@^-(8TPto$cSOwl12^;Gq-hbn-Itq@a;
z0P7kE=YmU&yrw9Aps8qPGit^UV^nbG)$-0*T9P843ny1$2k}HBRGT!RxXu#SlMx#&
z6D8gW$x3*`k~*uYi(@g?N`>RF;SdHPLLKPdTls%#*a)d^a3SU^By$A?c|w${1@9`d
z7ZyvgC<OoWXGNOItlTIEoMKfRvy%E3Z4FZKeiNRJpM4w7TZbP4c=R&BpjC*qEU4Is
ztwg*|pa*HAWx_9Hcpbrr@aK@OzL9p-lGxivi!aCPJ3Jiuw8Hb`t#-&^a=gBW-D4fe
z3tRm43rRbHFWc<7ZTBm-GS%94*uVXsmY|hSlVaPVhZWGI!0{19@v>`1$Mb@CqKYcS
zZqdFKRBuUK|9Lc)ocyM`t8e^htgAV05TV%q0c`1p$!KKm;ZzI8@ulud>?Yet|5r*X
zBTpD=vg+S$$G>X2$gT52ynkP!%x=t9LrHt@$q_8F5%vS|^nh@yFih$qMVu?<%=(wL
z>IP0~Kz)T-i<h6@ii6hUafqU0upH}p5!D<1jgdd?*}}D$Z#Rd=ad7^*x*|Nde?9lv
zv~|(@r~D{oI3E|7RIbj_x04S|IGw+g2~VG@3xKO~2R)uY^0?S&+nxLru(iRXajSJN
z<x}90sF9Su4jeb0&JhCXZD-&JvB(QWD)i9;9iQEO367pO0Px6OnheHXC<j*|iX4Q~
zD{(Knb1SjaSGwQBtoobjj=5FN%owDk{Ppc2wtOtYY>D(hB<ZFv{Pk}iEUV@C#=5bj
zvHaV2Ea;vO7Z)cQeml%WjHK5|QD(?c303n~xA_=JQF`+g1>$+~(xF5n6X^34Z>9(U
zGcdf5GZYvK1zQ_fRqrs^*sF7&?S@U^oIZ{D-p}1<4SJoJAACLG-;=5Loo&d*B0%ZX
zYj0${LC<EW)g8BJKD0Z}_lASCKv}KA@uAry7q<~3&7bWKsR=Ky{(NRuM{Rxu<l^7*
zLQ)PlJg5sSPorl9?;N+{(+q_=XoG>S5?%``xkt=n?dOo^RLUdnvB<Aq%KRp1<=;}w
z{@B?z-pyk1udl8irH$E0^A0(-w@Z_kolRfw=PO7Z(B?|{HN_X(Aq+<B+etGf1^kOi
z7J*w&cv>5DL$SgzQWP&0OaAfYS1bX#I^Qy`ND;dj^E36%tXnumRIk!-t2>}d41#09
zRqwsW(<d#oAl5jThb_dy>S0RAYVt8la$nR&Z!O_aF0Urz1Lx9q$x<gCfhK?BGZW4k
z9<Hs%*1kZzx(Bbo%00p=6Mz@P#H5xG4x>`$zKDb;^$42=;|c|oM3B(9B3gzt<E8H1
zd{2tLh}>AgQU~dK3?uoDfNgaKbstn)e{2A8fKD+?k>9zM!#3uL(?k{(1X22I{2Xfk
z)4sh{G=BD<YVNx}{x`m{u&}$6oS{%5FDa$5rcRhqywN-Z^N_0u%T7mmFOG-O=X!pm
zaO#7SQJSEcRMy?Wp=UhbyVoYm1RTSZM~->+Qza7^QI$?D4}7J8Nt*T?)?utpzbFA8
zYXS1MU*Tq$^Sf@c5MIwRJ9@%bu09#zA!*ip)NO>0ltG6PM<#EmYM)lH`TL()xnBSC
zotCohAF(<=izB30wdjspIy$2lTW%11IDK}HBxR)a>g+|mW?m9wir95(eYeK+3F+ZI
zMH%$++Pdztolv0MetAon`nnKHI%OC91tV9GaA;TM)KqOS?bX3wZ~9EQ?zocA(Yh8M
z^Pfr1?DDUEe7r{6VcOi)%+6JIxe~F0<l86u4Vpn%iY_$2DWh)w{7jG`cQk^jhw23g
zjyPctmCNH*#BnkxKj4+Sr@1tnkm7v$J^^mCTn7Jz0M6#M4SqS^;@o)rv@czbDi-0u
zmA*fZ)veQpD>NtRFJ}|^$k<6mETZ9|Gj{HW!590UMl5V+heY!K#u<q_B}o~OP_F&M
zvY$z)@u|&Hz2f9`h46r^ccx?(D=k!uB?w_0#BE6Q+UoAmitL~gAu(0{ymr`c<h0%Q
zeJGPJu&z5nT013D&dMS}(&S~Db~6I@x^k1z(<ET8fGrFX;Q}4IeJWby;WfqgAd4*&
zLWM(2LHThBK^S`aoCjySc7=sBdHk`7ANM=h!L%BTiZBj`(qLbi94(F=!s$)4L|8LY
zM*95U_J{25PVXZ&$2bfXBr*uycO&9F(LxLO-F`Fjn;{{&me+y;89Dl9B?zotvQ9-;
zURm)(H8Cm4T8p<J#E3V{+7h0qt7-Qh$!QZzihb8FxMC{w3N2HlWwMn27woSE>0LWm
zgwRbZM4*ivGa`E%yzC1x#8XSglrNo2YeJO?23(RiZtbez-b@f_CBlXRGJGlqw+3hs
zj;4E}f>M?^8hLtTm|SG~{$;@2tBO-bVmiV?l#yLzZxQPosZP-@(L+F{-v3|@4NX9*
z<avlx2Gidc+8-H76x>{1<_-qa`%knMhJ4J|s|)`C|8hEsv4<r#yAmnq`dpm!jW2%h
zc*CF_>5V__oM=5`YCC0V+dcTx0T@kx!&#27iG~kU{V4j;PeFu<3RLOk42_Q>b^F(`
z>>*DLymn)ycX8fvaW;{Vl5%U!Q4_U@j{N<77=vy<B~>w$d)AAfH)(mb?}7iWF{C+;
zf0KmzMk-J-cXeXnyF9({+WoyYh|?%O3F6weJHC8$xHQSET8`baeZX@d2+*`YnU!rm
z{`V)@_6A+2og}{DwQpm~M>I?1Guk6~#kd+@18Zi*x{8*QlX}w;_6-BCy4LolsqW#j
z#VgU0b;k+Q7Pnu(dDHPzdoTs&PJw7qL2DL~5*!L5Ro&^>-F0f+oCU`^N4s5MQe&|S
z2E3Yi8>TI%eAgzYD`sc=z&Y(b|D^%5fa7CXGmU6D8QYwix{MwoN?7Pp)z*2S?S2)w
z_bcK{Ixz`<@L(VGE>i9(W*wW4Xi$U`E$y}?XlK!r<rsC;^w)+QKSeX^k8x4GI!tU1
z#>rvQ>$V^qxR-eAz{mH6G`s?XbD13TBr~VI2M-knCAhL}S%utzY)}^#C3C-TQ;_1H
zH1ELd%6=#^YdDG5TGx9Ss~P!?v2<7Jp-ySV<;}0r)crYHXzR(7cR76<NuPL%o%fRs
z$|rn{a~uoY{L`|G8MQ0+@eL>$S+R1@3C~lz(`7M51*+`>yKn&~^Qi_`jkjfrsMWI-
zg|U<LnpBraOg<@qeUWHQbqPkQA=36JVb7)({2Ly)H#t3hVE~EzUGz0`2*dfzK%z)F
zd}&@=$FK(~P9Faa9TP53fPE1qWn_F74pHt}RpmknCjE;~EO3|jUV)`g|9#AhiX`;Z
zu!?=+0`Oe-ijwH;EJT-(2SPEJw=gyclxS&1Ztnz}lmJ%bBVn-o%J9g=2BBE8JK8nQ
z`VdjsG2iJs1!N>7;FoAhjIcFb1~L|dPOPe9i~O+Tr`TXh@JLX#XA(||MWgo5x{ONG
z25^s;@{L5*-%t#7W#)Si-NfG>vn!`34_b1|nA^Wt?!*@A1<Z*ZT3K3JHaKZh#S(){
zOQ;ghfYym_)_eDevtzkOO9UG${8UQbk0_4CnoKLDwtKfrz?mgpQI!a#|6&7nHDs<&
z+eF@t-?$MG_8@2TiNq9e)IaXkCjICt;dfOcyt^~Ekd!ggiblnrC>8}jA#<<tvUDHV
zlm##~uif|e>^$rnr0;XHLnHlGY9FPfL?JLJ;1=H7zB2v68Na@1%kk(Z8bln=Xhjyh
z)TcnFY$J}pm(ybryk3-q&ev0TOhtE}|09)>kXbmEJxEGA<8Ic_N;wPt-TM7!A<}nd
z*XPjb{!}t3D;U=al$whOCwvDK$H<Z<n#@uoiZRL?9NV>-dsvnfWV3;BxA(KH`$riJ
zkWDo<)KvSu4jJ(#Hiwe0T)Is@NF{kwqGZ(lyU%JLCy)W=rfW)AuZmJ5<QgE@M#ZiU
z9A(dTY2VrG)dMRg<_a8pGE}ex=;QECjH;N2%Rg%aRa_xt=q6fqNB%D-+3OYN-*(CL
zByWzjD6p7wbMT!v8)PEFyv`idq`^TzkQ@q5`Vq3c90T`V0rM{Dkw$`E)OfjQe#!5R
zQtnTg;*6*<$V$2`j^Z3L(AsIP<qidPPOW-Xq8x*uq|{$geSCL$Qxp(1Bp_l6_~U#q
z4Lun<I9dFYrQtr$fPzt~o?`*lhMQX%(Z31Rdv}@QAR;!C=Wf|@YuYlF{xiTvUZAT+
z2^QH(Zd(MssC|k;Tg6f0N#v-s(0xc{m_;OnfvU!^2Q}iS4N<d-v*rM48uH{<2|T5U
z)<Umk?l6`AtRm!t$Bw{L4F_6xQs_OkzL<kCm>0?_dFAR^{<&cCHZ0nX^vE<{B^!&4
zQc=B~#5O@NYxqY*%4M1W$#Pbm;N%xS4S9B`%cFi{x#((L;(h<V*Q(gSVe5F^u9)e2
zpi$yn<A&!h26qP32m*koyb75ZHLRU*MmU)95(7^$PQo-oYxu;$DjD<)A&e9wJy18B
ztHpoMZ*z&nbSh|3z{1VtI_sW{8cH!WA15dBrs%P2=4u-AC3^$<LxgZltrB!o=6AVh
zNC-9wgo!-u>YZ+{s<@c9fRgfv(O8=oCEt9Iax8PkE~J8mNouDL9YQQpBg)Y~oe~!)
zcLMN>OYH3Y#<k;}lq$Ip!EgmJ_I~_%qYF73#lOu>n&+L+lt-}8#qhYSElft5<!V^f
z0s({hu&Ikx6VUvLxY=qZ^ZH-%R#si*m^T~VY)H10dNy|b^e;^kaM+JSlJ8N|AZmVb
z30YHO$RUH9{=*2Hr@mBuWOI*3D$Cv%eJnj)8(c54>t^`KI3R+e`<8!i?6T^1Kb4Kw
zGb8Jg^|B4wFZ&E=^yD%NaAn)A7rD1Sw&g8zRp&ih1?(eJ!pQIA9>mgO<qmI=KU0Ck
z;C80S&|KE;3vu<O3(3WE7dCI#`^sbu1Q&m>b^ZDMV0#X@zE7y9FB;{{H2aSsdaHWP
z^&@q3xVwKe70R80;%YHd&^O2Sjg=#HSAF}J$qORR-5h+Rx8aG5OLhVlmEupfm}V!d
zHb~&wfarpju8z7$wo_|nq$yz+7^0b|>|hz{e_r5vn0G!TdwxVH#+S#r9ZVJDb#x1y
zRZw~%LeRT6#J;$=gl0YSCfMNN?nB4~3nGxg5=U4lDTp#?iZJYP<EeI(eF_sPU3zNs
zQMJURPw{XrO?yxf3Ly^1ja5;r@byK8f&fs6im(U|R^q|C(BAKU--T#4lo^HN<U0|>
z6hxX11))jw%BW|rq)X%}ii|4-lTf#gf<wn{i%|bSQgQr5i;I=XyGc%<vrF%(+9Z6V
z!15-LMSu&x8uO?j@-QJVv0?M&g!~$>EIAt+eue1PAHq0gxC-uR|KL9x32y6KR6vj=
zeb8RB-ZLv^92U;?ER-rinO{MM5>^O?m&Oz(DEjPVh$G}Uf`yCK!p%RFU>xH}@#n^p
zLNwV#H6w#%HQYt!P|^+|58mgWX<&rbKW!v5!#$pJ#})JfN{+*#o}vUaDMq9|g%Jlm
zzSh&#75{`C8Z2a+c?TObyq{AFSVe*>ou`rL7tt_^#9DqVhraGj8k32o0He8MUEMG+
z2um!eL^x5MstD6z+=d>jGX#7dr)Oth{MdZzYlEDT(1(PLONQmvW)Is)xb(skG6$8G
zJME$2%~8LQnEdpgjk%mVHD}gF5Hbm-X4iM&nBI)M=G*Q8np4W<pgdsbTx<3qpiM9;
z0ez-S_}vF4c)2A)r?UB&h7}1_=hl5(sc`7CbGqrICAM&SF$f){eFHMk)is%z^d&@U
z*E^rPy?b{#so&-&Yy_$MQ6~qE5)yNDGx4jr{&gHvC)IR$#^R@ueJ?wE=5$S?UMyzr
zWNBjbua>b)%8&{K0?4o4J&_}TWGxONIkc=eqBuC|(e$U#HzVCQ1v@{~=r#VRtFj(>
zI2^<)%x&YWW<j1;C|Fo$BF4JOX4(|Eu^eZFzhPq@i^IB@wk7fU>~tqr)`e>hr~Ww*
zKucX-z1;~ob)1tF&lcI;Jg6yAlA3S@&OfzYcd)e`cQoxh1UP6D`pZXpfF%Z_CKi`K
zi;%bbj2XBj-7#>Y@H^ghq)KoYytWF>U!Z0~i=?R&mI+%}sBY;Q4({9?(Tc3NGhI^m
zo`r=cBH~;ld+^y1g}Royny#~RJRu5EOtfg?(^lYFvg_qx+UDD<#esy{$3asNcvSue
zZ!fRJ)yYvFU^r)cOxN@BOdW+H3XKwn+iT7z_t4&(2{@_rmU(6Gr%h>v?iJ7~YiiNg
z<-)}V&u1>#+9=+_pXa@mf#~9}bZGhUoO!;jOy4BkM&aapgM^T!eSPU`#zKgp5w1qO
zhtDgO9#qvZVe$CELsy?D<;WSH7dZJ9-KO;01xav#{S}@N!yy(99*2#gqppb~T3hEI
z3#YxK^)ww?*SAee8eZG;=Z&)15-r4b%B`%8UT`seMvqmfoW^>7^9^u5Wmhx;^{(Fs
zDWgtf@#jt!7&Rr0d+gtQvtK+;<D(1Px9JU1!-68CENIl=gq1)!3i%?kz2T{&0+ybO
zO&ZALSK5bbQnBf;ZTB-GLOr{srvIXNLkBH*VTE$+=!PCo!Po3Xts8Dk5;lCXLs9Cm
zPJbC%1iE}y2*fTDI<HNmRv4;-JZq=yG+)hPscH;PPq2rQ!<VAZ*!xEUC}1%KZq7=9
zpW$3q>*w(q>2OlBRNbP1z5JDDBts1Xz`9u*?OvGPM+4rlmAeAMmvvqr!p1U!cZ0DQ
zEPZxNSs@mI18WaetYEK%XgY)z(?tDHa=*HHt%EDe-tAh^Tk;L?g;E5ypEs-bF(uz!
zCzKK9uG^1ot4FcV9gW6oE6L@A9oe}^Z@HMJh3)^r&qnM7T=#s!|GvD~326QaDD8L9
zvU}hV988wnXu2p=tnWpAE@X7RT+n%JHD7NEL!rom%+9(SQ86}d=Nw_JBtv6ToPsE-
zSdV&8Qb!%rz#ntdVta~NTZ(z|ZR_E3oTL66fhNCIZg1{sz$l=(YVjiagFo1)=@Y8z
zE!$nfu2EWCGB3f!w5gK77Fz*^-rjrC*mit?u#End6z1qUJJa0Yc{#?S?=nkTZnNc2
zaf{+z3B2?jDFs(B9YJ&!MM7XGV{_w@Ft9r>HpBlMvD8aV1ny7ILEjFPraip5<-gNT
zK-|f!w@fJL>Mr#8I9d`IQg+>kUjJC%TyM>t9hd=qtprmjC9J#_OYmQ3a1=Z;Tvwf*
zezS2;rcqY*$uz-5F_Re%kt5<nCK0dwx#69?vjyxi<YcPx=ZAvO>kv136+WD$n!)~b
z-PV-&6w(owplu(6;>1x!MfKj1M-)M#o@Y+z+LxzbItBD3HFB{**4&Tu#w3wcl@2RO
zZCdDjxkv=N5Qw@P@VtL#B@-j8K|WlcV}Dj_M)Z7I#t<+G`|8;OZO*m-v%gKe!gfAM
zC(IJRLLxWb@K=DU-x(^*GB7At+S-3g?QcqUS^esxTyv^VMO4A#r#Z6Ktzm?rJ!CL>
zF>AA!{xjzG03y7_PgWiKLO1IvmRITBX-IkPephBj3Kl^Ml-G_M2!|rV2|10nOIMQL
zsX>z#%lv68qjnRGKbwm@OjROEmlz~0)evQ)zfMVJe~kJz)Q(Gvrv=y^Iba-}optp(
zgbGoEPGv(xuG4i2TF;FZo4G_B%b6pQ#<p)PVuST?QwAsR&F<L@_0d?yZZuugxR1!k
z_DR$qtrGY@pdsIp%`W#n(9r!@f2NI}25_9*^ND^!CA;+|j0$dPfrGrPkV7}bK2ug{
zcECO``@n1b$xT<6MsBdDry9j-L2nq5e%79sz-y%i4d&R<+T+gUjZTxFC3x$8)1JRi
zu5Tpr4|?6Kh4q}yZgBZ?BUsaCZcElzCuJ-}wO|goI@B<m6`XN1&=vlYq{96a!xPM|
z38UiBX?#HndIRO4>fYojuZz}<e>WusovQBgPfe1Ql1VLmzuP@;Vg<7AFA=!$2dT=5
z;V4|3w*FDHAa371kX+Eh<2T{jf{7FDx9CXkPTFsEMPreTOMR}_Q`2qxMmaQyN1hGW
z9y`F*ESt3KjNi#WdCM+}SMIh8+kO5B4^<{ZHqBz4_R$WWZ?PKLv%8qUnvYav?l({t
zLb{e{Itz+mYHQp#@VCC*HM>Vo-h0lSbbD{l_<liToV7mjKRO{y?X#H!ZZok|MtxNu
zijq}yHL#UR5yb;b`DI>a{Ygvf0UZ7JX>b5)&(vN!rC|8#?9=enX?2qyaMlA{oDMTJ
zO?bok!&0zct4A{+1QCX(KOG5Wo;KB24uw<<2fLi1tr8AP>{<B8p(th&#+PwU!w++N
z**`eWIO%X95T}yawoai;gXqG35TGU{2RFDY+x&rP^N$dk6rd<yv7ikPyE=hCfb{dB
zv=*vd)0)_o@4b;91XNef4sWsV^iR;7^BqXl%S?h-G_r5JD&=&j^EPzMTbzF)veFO^
z%I)lbNex3%DCiy!Sr4S|C}+BhUxc*YGfRm6<|}!YbFLf6jlDZSJ*D3CH=ZNkf;_)B
zc9O++M|*yh5Rv+TqZk6}ZIBU90x*1B@@fsYtB*TQFWm@Iz!Cv@4h~1>`A&*=5?#Fr
zUG)~H>5G@?lYVA~?=ih0hRawC=5_4`JyO`+TxYs2jF5H>8eCcb-HeG%Kkkf7Z%R;P
z6aM(5ga!S}OVwt)EK-|;u;{S=ee{Jrl${$E14f#T=JQ&_noN{4i>`zNGXql%fBKSJ
zwziz%>ZZFhvP$S+iVWlaKgcU*FVE9p+n(fO%=lE`>V%hUsAlSIDrPukIGDJr&<bDW
zHsMLiYR~ocd$7NZ>X3r(PE*C`xTtz{rIuILKAeL^8*-Y;LD!kSA?-G(qkFrpj8gmk
z`{k3G9Pb^KR}BIW84&k@_-;!_&$-l!4bD6pGD^t&+t%|IMKj;au_%1&R7?h1k`vyq
zNDUADZ@X6eAQoXrS$bP@wb6HOm`qrcQ}#>V^vo2Mx`@=1q)5g0^L8)usIB;zwW!S1
zo_u`H6YZ(ZmX#Um1#0*Qw<EI-R%|S0!#C{qzh^PhSC@#ROd9|ucdrVT`mQD~dgomx
zV_!t_JP4a_Tyk>VSz9xhZVYrU`2FtD==)F7o`3t7<Fd-PW*zkzXRGe4N$t$dm3WO=
zTrb|3yZSYqla4Kjx*MEOb_j09)M27|6?3qp(YyR+S7*nq*H@z}H>T{YBne9o9sBcm
zo`RA)aowJeg$s6pLf-!Qlhe^lYcapQ#R~9^ht_?O^z=77a%+=*D!0?H<G4xlV&=}8
zcCN@SCh@a`cSFhSWM&6&dF^|j?mjoyzk>r5hXb5cNHbqq;%Ly^hmi#Y5=pa`py`mz
z9W7Bd|5&S7OLc}#LMS?_)9{a5KY;gT_BntkL^$p^Kh*|ewa{NX(Q6)fLpQEXb5lof
zROG0`2fReUfr##3Lk5+Hv766Xwl7MZfIr5l>%aAqmI1yJ5EGFSS!$Ev>+eDPP~fs-
zQ9eHAKW%9*9q)dlN=8r)+1|@-k<JFNT|o5VyH;hj^0;T=m=b3r9BS?uy+Xc<7}bH1
z+l(vb0NF!zqQz*-X+dUiWTB41PsaP27Uk6=U72TILt$WThZG%NQrsgN5>*%}J@T)e
zPDY9>!2vNYdAv`IDt|;0CWjyr35U{ajXrXLmNX8*MCcAp#O}p~&>|F_J)Yc&z+%)S
zP@lWOn)U-&3`KTEX;B~F@7n!=1?wt%!RUP8c(di<00-$&pj>B_K>Ab1G0z7gMg_t+
z_tW*1iVctad|tW~Ld0kH7YAs4pfAPqJWvqpxRvX>xm2j7Pw?G-ubki15M=3vxQZ4{
zFh)4+LCR*tQ`Hm`dtF_Fq4(Nk<A#>82}ouZ2G7eGh^M*E3M+^Lr40dr^V!t9GkJl~
z1p0bsse5gOqFwc_Z6jv}bKV+9_wL@6Y4crRpjT4rDYljnwQ0ugdyog=K=hwa{{0#4
zp$+ue=suY{6_cs=b#~mjiYWU1Ph3Xa2x_gab#(c)D$A+22=TR2*h|OkuD&)^O#6fT
zJ*ywK0(3QI+AC$V-sh*Q^<Wq`N5Pf=N{3fR3(Aq7POtvZWL#nP*HZVFquKnA18d9S
z$1p9|@(&K`eF7R9CKE>k$=?+nHZ9ngRF5z0jxO1*POdJMdffG$9wn=o7=EcD#Dtz7
zYB=LBR&4|2_4;;&DpX_g{fm7s{;U;DnkTGlw4JWBovxXsgdwKfNNwhO1$esvJBUNM
zX!rLt!j|YXB&N}S=f+&|DozNkr}(~4QN*F7szQ`o+j`T<tMU0DEiEnER*RCf_Sq7r
zr51XJgoB@sS(={&1NJ4aaI!8xJ#0JPmi4^u!B@8cM{%F7w_UGW%LT^2B^Cs0!42%l
z?;`1uD<fr~&OwZMtPGA*s}uNY2uG>GDXjj6CSVgvkD$M#h7B`F4bfRuq8+~F&S=gQ
z_FEJn;*9iQGI|jl1RN%sB;krn>}8+>QDca?loWB`AcgVeA5+Zp4c)H}%@WW{&xQ9T
z8YI5;t9G4K2UkigD?|<ogn^#%Lh1Y|d{v)zqT#u>O<M76WpFwhz~(Qs&+q{$ay!3(
z0klm1w~W0T#zoH?tEu)iR)84?g7FHYiP(8*Tj)R3-oi7G5b)!mue_0#fqFq;jPx;S
zMt@JyN2Dz8gDQsaxb)D$$l%c*=MFlwcvYodVjK+wn)l<#xZhrzlu?|f1=Ih)uRP`F
z_&IM%EP~mx<#-c#^XsQVMC84@6bB>xxT(6FNB7}4jZ)!d3arsnAv9akPj2p&f&qI`
zHjC7Ux(jqf=<-6nA^`@4cGi$&%!;(<3bArxh&~eBLqHsM{6@h;<=`+1?n2T?`sba$
zRij|OAAj(Y#7eMICTdp6J^ecM?t<niI|W7iQrZK>wj(xz{_kOFm1jo+(t;j@jFf}n
zIG+pA@eJKwJ}oB+!qTg-&*wRpe-tr@Kz4$Mmfdj>qK$pi<Qj*yyfgfmXBL<FVAYnf
z!AFH@4K-M8b90OPYyX#Z>Gg@CMMkDo`y?S0Q8mAi{p}HivRCb06G9&DLJ#eL!#M5v
zaLzSuYyx<HmYT5i?Rd$@@a=oAZR2_xVol}vt_ZfH>&xpCCH+rpMd5GWxDIMCv9K5s
z7~-CfUAmL2T#VadWaX|wAUco=>Zy+ZGjnPCB9?{ISaggcE?w?JBkNd5n9l3whG|cp
za-&-}E^%2Ko+7tmyejRYNEv$OZT2PdWh5MN3^71($u76yW)!zknr$%;uk{nbm~PVH
z{?NP}=WVs%0?nJNe2v?EBguqNbi4l792o&^s}JLN3V!0Jn9TyW4%bfl2eS+W>|b~|
zczGCSU@Sc?kIz1D0tf9HjjJ+eQ`5L#xW|~kLY<S+6pjg}f8IoR7RY|_F5W2Ezx=>_
zyABS-*gVRJ?e6lK;r%#0p0&kBC}{jy@ay5BWo9xghItT{Ba8+7c0T7{HGvgrN_Js7
z5{+V(o!0a1oXZgcs{Hi<L0zykJKVw_naMI$S6zoihk>gZU#d6IbOdoEw@H72*_Kw}
z5u^_jNVQ|ND9dZ<qO|k(v57k9mEE|E8vu9(r}0@(Npe_VDaXh%ECCY*g>l_j@sI#5
zqL|x$$hfQDExTsjV{HXTVY0hQYmpGgXA;+x6)}EcFTP<dq}gVR3qhv1oxNM^Z$k}o
z9A|gAuR}3d%+f+m#A2x_MX56pi50g(Bv@OSh<T#M6|K^*`^GPY#21y1{ba5HNd76I
z5)eT?R6Kau3hVW7Z~fG|6F0)1kU{$5Wou2>5ATK(nSI(U@5$k%YP*)7SDz-MjG0MS
z>{kW@3>>rEPt^x9hG%LdD!TcCFb>dgwK5|kL+#sP_EPm;3FtLr{lq0*iNd?#d3@9`
za8UYwKgJ-nU8%rhPSE|kL#s9ypc_KkN!)dY{^?7bk4lUb2ah=P32XGCcKvDJc_Y)G
zm*}>1NqW2-TKai&C6mvY;_26D7;AIRv*Lu_oANKShi%C`P3k#jh;Rw$kF7gB3$LC*
zZ0m$@R<&M%U4)J=oyC%lu>6>m*gLg%BR<~8>rOJ>f;G*`aCPovaMI&&AjA@~Fw%Km
zhEeC?>3P`mBW<kELDXfo@v%cm$K}fT4IU8B+w3jJTED|H7M`ABy}r2en`RROIF&#O
zmq|;0tk1y(^b3=w2(J26Q#2-DaG?8_`bYI8@(Dqy30lR{BEfR&C@o@)#r;C>zDH>X
zAL7sbsyhlC7BxfHd`i7#0Er#rVH)#Xuk>Ntma0bGK+7uDw`Fofp`>ns<2Rphtxvb^
zS2{<Gb2YzSGMsdJCTK!^L+RgD!A8B=`|t5oldfu;wfx#f^ElNsR3=>3F_hW0E!^(y
zX8T$O3*)fN?zj@(ckAvQjAu38@5tkNP2`WAx1+6O_AhmWtdWbe3(%R+`Y%lOlyEes
zW!rhe_Sn+VN&D@)SChXybwxc*#0t>Lhl`WJa=QnoEj@nTm~<v?dfx|eDP5`tRlv^N
z#ROTulQf|l9+GZu&7T;wSM~)s_XbuL|B)JO8snMG-W5S)cKbcczw>`;eYb*Dg;(G1
z>slJbHgytW+J(K{0J%_<3Ofi~9ExP`pV7Zd#-WUogJsZz3Q-M?gM8v-L1W~`i(Zc+
zK@?~{j{f8UdM+4hXtDXTxmnn1XXZ`b{b34l&`kW5t@B-Xytnh@gcL~JTj(p6cfeB)
zow!1|oM*=CRv(>HD`a$@lJn#h%BGirl0dY>TyfGZz2XA~Gap)}IDc=Se_irB1|J_e
zahx;&*I)T7?G`gZ8X&0Kd*g<<-%-Nw3dSCHYCP0|O2A!Gbgw29^W_IYn4eR9tRcio
zdn-gz5Yn(ne;G#XXDdVAfO++>ud|2Z5#5btLrfw>`Eq||2tn~G4CT~t7}xM%Cwth7
zZm5O9ZNmIxwm%(|TT8x-<q(2eMloaausl!0)AMT;T-Y1eJ%iJ4wjAFA@h1Ly7ikd<
zYTz>x8jYS@kS2ZJjRg}CMVRx(l7T*2*<66+`0rZuZV2MsMo99t3^}^fg%QQ1Md|7E
z96W}*5v6(x%H6-Q^#4i}Igg)wCp48%d2}$YejpFidnr;F4Zq|3oC1rap7jb4c}FaV
zrXfyqF0HNVr?7&o`txp(gW+1y58?T2{&R-`u~p2@%-T*nrUn9%-IpC5JOo~mM)Ab@
znHx<u>{$^@dodWTJ?8Sy(7x!axu!)-gOtP)bY&=Jps5)FW2a9L*R)#4u_q`BX(>km
z!TukyzwUg7y5iH-D%kf%FFST4-xaMj{aIb~k=ePWA!$zqi&^m-mN^_~J1BPr<`~xt
zT-#c8C#+lUF2_`l-m(IPf*-Grj33!NZe65kyIvrO?tv4Zo!zy~7c(D9^|?RI;V09_
z)4U(@d<)KeEFg{~4tC2ItH3rbc#Bp5+op58+;ndsdvQkf3!Zx|JE<hx#)H#M-aEjx
z-`x2?+r{|E$l$x(p`n`vx?0bj)pVyD?=jWI!e#v2STp?7ftS}eVovFh4{^iQ%G&6M
zj-g*|WiLMhm9iHHJ@`O4)63vCI{ym-HTtUP;U8jv#79N_J10lJ9@iT>q>B@<HjpO4
zWhx+(ki=Bo!*hOvfS}yX%G$CWrX}j~*hq43*(hK+b3yQw^-6_=&&+2^NlW@yiTPdD
zvPh+MtK)EbZ#Har2B0L`37B*wlcKwU10yd^YoQ84q4hfuFPJ0aN1zOh6?9a;Nv<C0
zeunZ~xq72->bup$)_h&ibPSM;yy1zc_wp9h5EFEEVr9*KRB#h(bN$l1^t+9lwak}e
zeA_-=V>>gOHTY%I^r6yR^KPF;v!YOr)2~j$5GSUg8ywFRKO@2YruTM4^%c=Z9D=I%
zNvR-|qX$1zy?3uKI<7AXe%%FHS^pxesN(yoY!?&37(>T<aoD<7zw7|6m}G50T1uIS
zP1@hp>nJcf0<DuRu2rL{Jhp*^==#7_y@IKY=b3I0=4K;p{jSJJdK{;2;jdp^ga>wy
zkZ`_fC1*HZa5CSqbEn1E{=9o>cHqrM)>{RG+jWA@lZPe~^_yHu>c&J}1q<e)ev&0Q
z>O2m%P59Ya=v$W3a^NlN1!$82ryvGl#8yQW+7^M;M6{I4S+3S^>~!@uva&sxlG&Ti
zIs28oqS1QUk;7^2^~n^#(`o=GriO!DTn}_+A5AeLzxwD<a;2HnZ<2#wpv^hQy8iX^
zTWWJTJ4=Wf#y7=V4|}>de75AaioKlI(NZ;Wu$Qx#g#?JP0<y9v`2RX=t^qsSEwmaM
zw%E=OlX^e%m-Bxfeh}RBk!F;d_*Opsp?CIl?Xr0ISvG5CcGi|Ux6*b0%0q_7z~Og7
zte^tc!tz_&k<q8P`d?e`rwN-D!I#0*7k}rMps$N4<c7SFk>F-Tw-jIW2yQ@s+4i{b
zDM2mO^@YIL{<GiYz1SmrH1MfquN%K`dAjHHDEm=z<+01Eq}ag-?GP;uA4g8Z?|Vf&
z=f%q{Eu~GPV)*8O%d=XWX>eHJbIFOE<3K9b!sBLRNyqXm<F?1GjmHO*Ya`}}$&0}r
zS>3qzZ71!=sc;vwwxjhP*>%PNRUE<Yu(Q_Uw`DO2gx{h}`Dy^uLcfqGd7t;*)wJqU
zo9|ZnTjRiu`}ONZ>u0_m_jtMY76`PO8WYH4-Z5G&{+pz6g5~8$N7)2MRO?;@a1ueg
z{X`%diPqLDC|4CLuU%-wbeAA?YRgM1LnYhWn}D5#g*c_Wx$BD+uj_&9zqGB#^7^Zz
z*GFx@skhm+@~oHVrtCiLbqC<_`U8P0y{HeIn_c2ygJSQ~G#UN=cK}im2H^Spa8oc3
z$l@95+s&)Jp`it(x3g9CjpSbolZ3$*XMI_p=KTJ>;cu#Uo%s@_nVC5V)fsjAHTJs!
ze@bwqnXif~L}z~iL1kZWA0O@9UTTT62O1uid)E~&ykhpp-hq>`eW_54JsO&A=`6%n
z;ac1dQ{MU?Sn58O6o0ZA@$byhb>{W{eb$C?@+U2?x6W*|?y=n+PK)WXkojVrlYiTg
z(k)Pwd76KE!5jbDSy;E1-P+t^HT6mHJvnBRc%^Ai31)vzrnj^j+s(i+zU9WT%n9(1
zaE6uN0lji}SF+bmt^GL;Q7~Yv9jW7UGNCM&m|vS@{c+hYd%D*mb$u|I>~}pNdp3c;
zyfnH#7ywS+w>1Q8EHAQM-MCJsK9s#OyDp!*_&#w@+739Z=Qg{J$zi>FP5_z~^Q4*f
z=bOwEGo2{36FMXaE(_7|BNLubFo3ji{_4@Vd$v(epejj4J;}!?7DAq)aLg|VTMoiw
zJ<&4R^xsQvt7`DwXumw)KXsGNhKz}KuU^cb%beT`J`+5_1_hwov%23bOloZ3V{iEO
zcicmI5}Kkb@g)N^gKrId>64q%`1<rLj%6ZLb2$61|JK9Sy5|B{<F<j;fUT`q<2<AL
z0Oga}X0OZT!wo?c(|Nn6bJJ%-=B_wh&yQt`$U7N2Zv8j)(Ato(5p(bHF#WkNV}eUe
zZ6`p$zSP{+{Pt|ef_8c7x75g=>5ZS0jqPkp36VdL6ruj-B7%zjACd`T%)YA+6<Llu
z_|uWU2N}e!RQA!|X^*!L0Qr-_hg<jV-Fz$4Sv7xId=8xN_%C<Ax9zf#mS}-o5Y(s3
zan$m_e?#z^)|B4U9bbY(_X%Ks>grO$X69YzTAgDFjnz4C7rCEEWp1u8<<6iAUiLl0
zq~h%~c(A@bW>oJ@q%(c?yEwn1XEX|RG;}DxIg=1lyqdp5rUk4@EkY$_j9&YAa9E*2
z*R0(9k>e$hQ~Zs$M>%(Ix%~Dc2sR?TH$VI|boi^94A&pqd?^b$!o%`^p8Z_?uc$Vy
zqxV%4S|s#H;T3@z=iLmvWPBhS5V+^G<KgQ1su9X~S)bZ+WIBzxgX{jj<nSmFqxzAs
zSyzGMgr}g|Kq{_Q0rgVm-moNFz=XGn7y*3ri7iJ)CkegOGuj0ZhE)G7f%v}2cKwSn
z^`q32Ep8(t0;iephl`8)T}DXi2p9Ku;}uR|l8i9rnXxvr%I9pjAYX#OZ-&XpF#R?<
z3?tv)i^Y_jZkq)R-;Boy=-j<8!BnFuAYB<FO`!m2WZ#`@IMxr#3f#;ZzM3KoiR<~*
z%j3B;q{}lg9*noiv}gHF1e92qrikSXPxcUnhSPIWz49IO`I7rvRR!?{eB*u!tvtPv
z^7>0z&H7D;czM80dV)M92_mnX8UO}ya*}ukKKHnEk}w-wnQd>M6zW_8GX+@u@_ekW
z5)AN#>Xr!T+P`c3iD9b4=n+MKOC8Luok@53<yjufYAh}xX-Z@6?&wIdAMrM@%I>Xg
z^_zuDrG>u8zNV8!+4E0<JKeSV*9+IHK)go*Z}`t=@3l^MH2^EWjoPe%uDvLzv2d=S
zb8NP>QH70-G2spT^7Y#GLiV-w1@%HZeio=R>xO1TVfdnn>2Mf07up`02`<@aTu5dR
z^N9zsgUPf=6C<Umh;j>m3<`7oVbvY8TG=CjT1*7fig|cYPJ|xY%YNJ<B9vlKEsB>L
z-8mS=P8;0kt{_2&BW#32gt$1VC@HaWaG|b9;z$l+)jW!HP!u>4=Ivpjm`4QEfWx&2
zy<>PNF$8lTZ3BWo5~9Wq=9V*-{-k{n%8O<}uAXNTIcZ6<F~)?kq>=_gmtE|%X^_fr
zJ6oQBD+0R{Cwq!-Z>{Sps=Da&K8VwI2{=|dS8uqi$LHB9y%J@^2dd9p=C0oGJRj4q
zFLu2m^y2@!3lDFZ#^<$pef@9TP+zFVq-opu{`Zy<=BPAArtqu(t|1E?dS$-)kZ^(V
g{0i&;=gVDy#8KL?rs<M=2}Jm)C_F<|AYX+154o70wEzGB

diff --git a/vignettes/transfer_learning/unnamed-chunk-17-1.png b/vignettes/transfer_learning/unnamed-chunk-17-1.png
index cf2089adf418409e903c35790222d35b7a90f29b..c007a8aaa66cea2fccb915a84f878392bd4d7ddd 100644
GIT binary patch
delta 322071
zcmXtecQ~8h|9*_tED5SnJ5*^=u{Vv_RM8kIHB0SPwLNG`j2h)_t5qu+6tQCOwnkO$
zUBsqp6~(Wg?{)poU(a=}^Urh6eeTzN->;LF$1;IuNkjkv4bM$c59FKV*3d3{A&-1F
z?(-(9lIPP{V%x8!Xv`BIU}>LWO2TsI2z}=)yG7c{7s2;737Uotf;?Kc0uJ%i<m9|D
zs_R6or1m|O@6gmB+Ghyfdv2G5d!l#E=$7|V@4<Piysz_~)Y@&k3b{UxWtm=!H@gf9
zlN}7Ns$K&#qU%pZ+W+^-xxF~v6Q95EfAcf4ujnuSZ+5n~BhB*FR6RP<!*t!3JGke-
z>d$mSXMmsLFPXNBHIF&PgHXTe5V}{rt8JkxxyoLgyK(`n@#h!6mf9`{*IF+Zoff<<
z=W{|{QPra+I}Ctde7RP>B|A?0ThD?9fd!=#KdYB_W>4oG4cr4ig>1MDpI)Q|n~658
zna@q^I8*gS?JYV~?Z@=(2+W7fFJJymwmPfP<d<LVn^&BD_iB|p*DD+obh)XyHE7|v
z{w7VbVJqs{D8T$?u7ja<Pz=~Ok&+FVEqDJc`GC9Wa2Iycu$`Mgbog?f8~7d*3xqBw
z%hVUV`sMp+d4!w$2$!a^HTA%M&2h&)A@FY@SKvOUpH#`hatCAMl~ZoWzK7D@&Vj_x
z#=`k@f|Ot5g}kWpNxW(6;cuCilVz%(mp%qrk6Sixwq0hMTJAs4%?t4hXn*7QA*+c8
zD_AnW)4>?f{y8_eWAATIkm=TCM;g$!&6&G-^4=#EIQ(xI*w0d_{<Iap_C5s+ay?-7
z2${0F8q8WaNP2KywwB!}V0SBaxcR8^+3&jHudSyuSAT$J;{3*ZeXdQ*aqhE|-)U{T
zbGf@GIYGzo8<x-hDyU!nKDWB~ZrZ%dX&UO_6L`F1;7om(oThRvmj*Zu_xJ%yZ+5H}
zervW}eJ}qxC#qPZ-a>j^891`aoxSIzd|IFqFf|q=+Im280nQVwe0`@cR%I^7Tq^%M
z&jP!<Tl4#ASIelq#>^r4s{u9&(4U_*drf~=)7p->E43Zp`%GiVw&^_nw5a$`mCVeH
zQd#}(oB)1J?=r0AG@lNwUG4%_ts8b!^u(FD^PdVky;-&7AhXaV%S!+K`&XB`$%M-0
zvp0LaR2SU_iYqD08jZEDhSV8DzrE>t+FIc+gW?7)3S`b7v9y-K>YoRf`(=6<)9+n%
zaW^jr#f=jG_Lck{>r3!IyhwY0V3~A0dw%KXbow#(f~W`#11(iWwI{$UKUrTy?HV!k
z%I>-U6?gE0(ksYL%Y<F%h1<l;{+aU6kq630+vnF@$H$wBX|T0h5kU)HnVDKxBG;*-
zY5B}j{nhL#Wckd=aoa8NSn+pbF3NA=?|I3B;|=%+Nx7|UWAoj=R+WNXsNvHDN6gFb
zjcd8^{Nsq+&@Mt>%MrlfzFeAnnXI|ZyLr5$;;S$A{)BFC^5ve@1%^#};ozsf-|*h)
zgEi0QD-S?pYwhYc{G7r_$7x|_DAVvj>B0QA)p_@vR?X4T4}JuFJ#j+mk$Yx0*1ONj
zf8zUGMbPMjS-ajEiQ%@c_w9F8eln`aH>|Z?Y+Q7g<ofl^xiJGY@VRAk)8Mn;B_V4&
z(A8tci>bC|97pAU4^;lJGfwOFmuc6G8d#N7?u|TXeed6gT1Y6_@VeP}!!C+nS^1~B
z%F&jJ|J~viMqtxJ`Ldba?fBIns}hwz)hErDX?r`St=o6Fh1RaNT`JB7+JYAiLjSbx
zCv%@wTO8)*dVOXCx*bFIp0^!-Q>t&r7t`eitmdoj1)a^E{wg{A1O!<h*bRrOPWlfk
zR17>;*$*Gy4rn)icGURbV$1RU)%xzYg_fljpQ;C~f2tQQG=nF(LtU4f+CqM|c3lTm
zot{|*k+Mh${zEDuzn63UcMNhaqgBo;hflRyd^H@X7KR#*fiH~s-M{Gzz*Sn?pB_KO
zBhlk2nU+&#841Qnzl$M>D^hv=)kb~r^MlbgzqY2@!iB~6ea(M#4<59xpTF9;)#ksu
z7yN1~b3*Zy{;HEWe-tp5pmMSS1dp*91dQHQzAOX&UhJV;cfMO)Trnh2c{OKW{b1xm
z|NR=<26&uK0$KC_h^9ATK|!@?!Qf(?G54fwxwhKH>eW<b3-K!6iuikRe9h|D<BXa8
z)n=O3NdH$BNB>S>;_*sdG^ZNS*>0Kex6!I;CDd<D<v*XjkR{Rc$vwsOS_Q>*HtyC@
z#;Y$yDxTxf41zhKBNRhU-!Er!bZSE@uRIf+CV^u%fc%c*YtOc;-)qC6Ki}71&K7d{
zj|?}zTB^W3Rz74jt?0bqKFJ-@t~pmc-9MoL<>dxm{l_k971C9tvTtW_`n8;8^KRvz
zlEInZ&2t_5Uek<!$9?{-d2rFO%pJO*dwJf6E1!?M_))I%E4;GZ{Az+x`Gue20o>qH
zCBBR<5uoGpzfx-6`Y7R-Kf7^xqJk=R%xOtzrk|>>_}h-Uc@Xvf<Ur*(x$P`<i`@0^
z^355A=-Ji$1X+cVR^r@+-f>T`*TA-dd-+(@d{7U};xFdzd`t2u*U^gZBVy70H5aUA
zpYc^gX6Oxd_pC$@0_f_TY{s1!7xw6CDedAQ2e>5Nc@VO?S{Ztft9&sw9CTW;;MY^+
zr_`~V82{%2DH9Je**jZ}Qru1;-BQ}`yBY<EYuiF?#|$7)C4sz}n!JIM4D>6z#;)Gu
z>hgjJIab-oU?kP-RUf>#;{3U}cF}$T{GKbZ{Jm^&Hq0n@p78#6#AR`s>(fP}`+3xS
z0LYrUZ-^D4C=N-AC$v5Denq`ivb)fhYjbs(+x9ES@kH*;A{|K6@xeTQj8f?R0XYSZ
zjKA>t-ygmY#!&qa3xyklLf$hyZszq**+X}>a!)^41$Xpz<z9RN{>*A#p1-!eJnU0W
zu5bIQcB!G{eDU-Aa*3|)3f^|ymv*rZDE*r5S<iYyoZoVXUv-sVl45;*$8?{)QYQl`
zpX}bMeSCZwl^ZZ#H2in}ylAUpuqLFh%y_25jgvdz<XpP5N9F2n+r^cl^Kh?}d??m!
z!scQY#cg%^n|bf7-IN~mXFKRw!-Vox@7K!oEh&%S{j7?{I|DSkty^<LKW`6R0pTT<
zfg2A(hRwgaRC*j|9}=q5Q+?z@%SnHUb8;&uEi0AEb=4Fez{u96dYjkI*Zm(V?~CtN
z^k3~x!Vj+w+FRDLy?18bpBOVLo;TAC5!cK1s%I|RnYq4`=hR*umgfE)<vz<WwK%vj
z6*v!!s5^F!4L7cAOL$zBwlxUk0hWG)zF*dqPuUFQ9WST3TShBFZe8s<vQvT-PvHi@
zX3y#KtCMSMR-RD@4+6$rS_7kqxdBh#*C{lx%^v+cANoB5EV(=GwEb#pS!kE3Y;?G`
z^lagx%6OZvZNEKv+~9O~4UppjwXZ<h{?t-cS{^-Rlz)D4Y!GM+WN%h21FE~{HEq8e
zBvgW@D29kz0rgk$+!v#adZG;nD-IQJOLDx#k32%BPyUir<_=`>>kJ83Wx9{h7Z(q#
z{J)=>Cxje1oal?GFEwm(8;mj=u1JUH{JA8B5k)VP9G^9L<<7tt59d`52Xwi;5`3Q5
ztwj$vAGWu+tK2qJzH*%Z1B{%vT>1i6(OWO;u67xN4!ee})WR+d0x#LW-=7EmY%3Tf
z5Cf-!6ffD*lolfne$Kb39Dmd`y|q`eqoBCGyCx2rU4<L8ZM*w5yI^*C+b)LuW=~Mu
z7TNplnwHJ}gLLx&i$4bM{CGK1BMQt>ES!#X1&KBdr{x|mmT)qV0mUZ)y)##*`IRBd
z=dD%KKJ}W4XG?39d#+1YTwccptv{6<n-4{-UhE$C>pnVPQz8Ccerw_3e|(~{*WwZM
zr@XSkJ+M+!Wpe({1b6x@!1yG1_Q1-ISa0P;$UV;qdH{6WSuv4+4oC={e*TYlI$6Vp
z16PKvn*4SYUh#hW3@~1Pc2xSx?SGFf=d|Hg=&xM^|G|aOLsi*xvNQ|two&G=xy!4+
zJ-P;SK_l|yi=Y2*y)L)s?KgD5D&axfXqNKC>>j<`L92}0wGySJBJRdT5fA@0R91<~
z_R_iXuP=VfuOBIex~;M?o=-XIk5ZiUCb!l3zuesaabi3SG?W$hp;n~n4|D2G=h0>%
zmr5Szzrq1#rURjq;O-vBo~zSm7wzS_!IgcjDO|t#5{TzN_n!Uzo{$r=p)h|G7qn1U
zvcb^0QObS!mP+{-Tdv|7T;=y%kc!u-RWspgJ}r1rJwfj1>F1ME->bi=z#jx)_ZpMz
z!}aGIasJ&IK*;K<&o92(`ee86A@!%1>e-(tkH4L5`CO~Bno6rQC)=(0?~fLvtPb8y
zSa=;N<^*0+w4HonZ0T~7U!U#3zZn|?PO8(|mX{|s&ZbU+SM1u>p4|<JT@JdK-3s07
zF&*-7O?+|Ur?}zB<wcTV%?l+uowac(M21rR&w;oD%d+7)gLC!L47@NH6NDEwl7zTR
zN$E)ywjZPlQf;td-`{6%*l+y_qF(nLK+WXv@^C5D2R#Yf>BQ|bb9h96;ynlYunluv
ze?H$p@DvagW<9$;<+@F|C;AYV{HG@SJ{i3=<=pWj?o57}>(Croe37OJv@M0skaM^E
zDSZOSNP1I2Q5xOs(c28N@w@OZ94$ig-~O(LeH#hAn0y8T%?>~kEMD1{E=@lA6&X{<
z{=QRxan~P7`)enOBNxswcKV`?0Ti#rkmpQVB=x;=cPMrU6+U^Rxzf$<1gff&Cd!qy
zLK@C(mHkc%*W&tyX&^LxQ!0<@yoNYOb|Zo4t=0?TT$TJU4LeFK^>W~C(>RZW$r}K5
zQg2yQ-%ug@c*Mf5_V^gj+6%~j9sANqZq)^}t_K8CnND8#Hk#xpU#EEQ3~sO=DTSHQ
z-R>>_3;k~?-O1P9eU9f(QqXKac(w|9wS#+8Xxi~<lRIe3PE^GtjW5CJyo>!hX&tC)
zIB(cYV8h!zXdG!tP?S^*Y_k&?o*YOJUIXl$-6el({%GI%>^mUKog=I1@I-H?Es5l^
zNU3;sI^OH|9{4gnZNKpTDEWA29ylwzo7-H!m6=FRMOM(dNA^NIYiGKp#7%*JH_8H+
zc`jy8V<VpN0wtpcvNakrk}9q<*#XX7HCdVI42nnHzh`WOZ!x&Ky4G3}ErfrINxHoT
zRg6jYsk-qco3!80zt%D49UfNWj?dD|ET(~fG%4p@v(QnoXUKP&Ud3762pM9F)|es3
zB!B*}B@goSfw^YYrHIq34Y7fq_!7`GKBJPB6RSo8Is6QbrEIi)#Y<2Q8vuGX<~{Ym
zR6hnz$!du5mGYZ?_8E$>hp>`$s<&+(eDlpIOU+;5Wwz#_X%8TLDSys+HdULW_YPSW
zMr)0Ijg#XsTB!71SmRAIqAAl(0vBk}7NB*woCldYr_U+~M|1s#oaLDrh2iC?tyCwO
zQCpS7N0<_qn&d}v+GgGr0DIQc!dcw2CvyESL&!^=i&A-uy7fR{#N(QGbi6ocp*z|v
zw%$<IuV7o}+vPd~9Q;x@#In>xaE-$kxV`hw6~;lj+^5~QDwV`kf?vojSV;0-V+EZJ
ztW~t0#af+xk(#(2#s{{E$#=_in%G6>iu>T3Hc>Z|Rzmhdwp^uKfHxASsdkNKikF`j
z+8P^u*T3~1QOS^E)$vl4$^i?z$tM-LwJzh;eo2m&5;>3TVzfRKF@&^P2+O2(HCI0E
zVRRR%T$q&i_ctzol<9DCc4q&b3{f^VU>jF6<$9IjI56(OAM9_(mnD?s!$V^6pZ=yz
z{P~&KslCN3aLZ6t7kG$a1L@SYKKS}vWd)5hrJ{N~ij%s|(E0F#3sIDy?+oUNoctj0
z0c;Bj!{Y~ePe=lAd4Hh*7iB93wX&O_+S{<mbR)eaH6G^zq=Q5Hu#Cxttzbau7>6N3
zAw%E6WzN2-u~BNJV<3HU@S7VxOY$K(Zxaz`vJ+L!;cU1JKn=TTaQDTv6B7}rYf*ws
zX4K#G55!WkUGN~#?sBT87AzZ~%8#EuX~~2b#YBhexonuO3n2AY9I%tpc&x1IP-BX1
zF-R5r@PiwgR{%z7A)0n@8tmdi2eD18aLEwEet_E1GbFnBViJw8gTbxn;{ZW{{Bqhx
z*}RXD)quqfwcKdg?vzXi7Z;cQF4yS_yswa<P)N-8B3UPPfiEGz`eOdOnk=TJ`4-|#
zh5E14@!+wOBG!#eF{cLoE-_)!K~M1Idf!DGZ(Y5oKyH|(ftF)-%heC#NAf!<AzP!b
zPX3{gpsk?&Nt;(4!WQXDf_=)89LUz)i9!wC1fq5E&J3!OyQvP+XZ_tpCVsT=XbgH2
zw<jZWt6aGR|0a{6Yq4Lg0@aZ;2<afBdCZ}D)0@!V{m_UB$@D>6K^=+rJ?BYsqAx%V
zzBI!}hI-aOJO4Cu<~yMw8dN9>Pf#^lALX)Rszl!KbkIEWI%lm3uXGen+P%fS0(AqJ
zZMT$q19psKSgtO^kw_p0YX=6gr_A2>?@)vjnGiy3phUPN<i>3C19MBcUcyvuGzU4H
zkK$e)tm1b;<{QE82&O8u{nRns=J4>w97Pfai;Tm>qkRRXOrO&{#Ar1qf|=AtrG$!(
zjtECbmS0f|YdyiA>}iBMyC3NxKtQQ8%0=jAbaw_gnl0Kk28$%Xe%g!s5^TnT7GoHs
zh__!;*esm%+k(^<G?(hKJYSf;epc`0Z|Bmu87n90CZUZl_i%L;N&9L8)p&3>@|*s2
zoj$cPeR<EUS{#NfJX1i*&F{Yd0SFWp)*1g~uus#;-9eQSu9b{n($hAY0gCF}GLgtG
zu_q`V1j3&GnxBl6vD6sco)@XggMh^t8qp6(1_lZv2-EA~8Rclvs?D##H2R$k76V}2
zF&$9x``8<eAJ2?%Wv?l`sQG#2X$r++u~<)QTZAT+AbKmWJl_=m^k7i=#92C!D}E6(
zv=T`V%?j%-R4^lCk{1Ac`JRa>u-jrAVo#3Xy|+j#wi9)jbRQcS&)qOEGMy%=J)Am&
zPdie1&8Q<i8$)>w9Ns=KyYqs@Te6=A=D<~GO$ikdm!)syVPewO`Dyp<Er}rvmPCW0
zo<)_TERV}a=%i%?Qo+4Y%4m*uYlXP$NSgc&G$l_beABST0*GnkEubLJ&&s+76Eg(U
zck*gEYj}wKW4JN0fg2pyNqPp4As-<(c0@ar$w;idogOK-_D-S!?xSLFYfHjyx_9?e
zXC6GhPN5nlc-Q2c7cBDq#<tI}^75gvfy&W`3W$xC&_}$1ak6a;%*bWT*#T^9q;|h?
z98C~k6r*1R@^FUakjC@G8b0w@vdlsS-(<GfRp$`k#U~4RQg{^j4yDgs8ZCM)=gsSS
z^w`L}$N@@KhHP7QGh4j?Xf>;FBMG4wUgY_?hy{O#5y@dZD!Btq^G$w)I!LsM6B@X0
zBH`nM9~+z<9BiL`;-m862aoM8)dETW+Y`cy^2tMhzb7%X_(!I01p&tf0_A6vZAzQ|
zbaz-B0u6mYll7f$w7=NxDKCs-aUf1y%Ng<d0Uk-DHYs-A6>CzZ$w~n?lgRW2@h0jL
z8uqOGtoEKC2^A(GxjxPcs{AlMkO1kMUQ>mWK(doFu?M&3^K3PL=GZEe_m;$c^)Y8r
zsWm`RTv1rh0m07;hA_i@WH!8q%JZYzNsnhi^zHq4Rz6Rvfjd+ij|<8y+u&EfrLx$w
z{vnDoyVi}?i(uyVuUZeTT_hPh!rv$nCo-z6L~fN|d+6r+&d7SD!foOee(rYjAobmk
z#cD#0tWp%7NNlJ6yJDIo!w4<CGN-IM8vy3>!6GagW1jP5PM@#|)sq8}OYo6_#f&Ei
z$xgS_R4Lgr?(>UL`T9wOMYLL&vl!6Oi{Z%W^}-g>A^pCBnX$L?_;|_Sd0=L^JTHWq
z2ExoVj+48^t+BjHAPx37sL+q4s6RGCVD-fjJTMSVc%n`jsyzM&gqKQNi^C4EUF3;O
zDJOp?M(1T6%@KDKiU3pccxs6I_}K}lO2Tvy4%8-d;q+LJa4=}r!X~SOoyz*}%m6(^
z^ZPq+1{=siB#m5)w-?F>DSS<`MUjQ6fnmI!GE}Bx{|Ety#r~5juujGK9#H$BhSVF|
zEZHx|3~n(XYm$TWJz>~SfWd^|PbMQkscHj)<$*XjwFjekC!9eb>O#}n^ar|SY7EJu
zQlDewpWDgn+<eQgW?5mvz-ru2cw!8*FDUR`H|5}CmJ-VOmaRgjmZ0VM-nZAaB{|;i
z*7@qJ&eY6qYIoQ(Nj_dG1O*d01>})_3OIe)mx>|9$PbUFN(Cc;xRJGkRp&-DUZ7@b
zJXp#~c*gzvhdQ?=maJ9pBo`W?w=P<?RU<}dOINp4Bci~&D$3Nb>%RYOiP2leK|h_e
zrWT-4Qr+#6trlaDFfa`^8us;rTeSJ01)iZXl?jO(LzfdJeCT<5)6eGzUQ-%B6*1kF
zt`0dEd|RA@DYgL|e7A0WrI@gi$<ecxzJa7v)xtu}%rFt9Wv}@RsUxrh{g}-m+%;#Z
z!A4=za;PzN608vX&fb*uA(xW?1r`4zUewX7BNh0k!@J|{spaA*MotTnx=6(^6W%ys
zJzgz-7SmCqs8F+yChBhe6m-n6Fu3Yy9WSNC)1g~x3q&A*npbMj)y>Uq;(b>XqlY`k
z{PZ+`gvEh(Bkem|dxSTYkFl`^eJ8wM%fJw<rV2uE-NN(1Aas08ffQj`2dkzPGiTok
z2NYE2vsw2kNlr62%0pdzw!STO#jfH~XhGO-9~y!t&N{X#vpn2{(_&fpcqM2Gx}li@
zbKcdzF~A~2$WUx3!AwNcBc8^1rH;mu*^a)Vs1zloYRk~6mrX{4RZE6duUi-6<OSF@
zhD;)~5~;cD%(4c@Cnjd<3{SFs@Xolxui^AzFE-VV)%WxbR3AITEsOgHT;oIVjV5aR
z4z?N&)}ZPgKe{}H^w&aW<JD5R4@j{VUtUV<0{XDuw&;$xAZjEOQ*9!}J~v|&o5k4J
zKmjj--ptFqfu?zk(9u^DE_xziO+&v(NNt@kqS&w9!B&lbuS|Uo9gr+q$Hohmh4;se
z)XF19#8-IVKwpD+_#ouzI3aTIzS!ZNTMqLK7Le!{E9HaqY-CC48Bns1<6iP$W*#UY
zEAgr3p2t%%Rf-9(18SRr`m(IkV^$_39n88ceb={>uDqE`Wsp2i686a$I)-7YPlWYz
zZHiGjypL6|)^e0lZQ+(55(bHfe@OZ95{gp*d*R2CLwCf`GCAYfK3r$5R>$Q9cm8|q
z?Ii<oaFDqh){cbT4IDqh&?|l1%k%|CQ>H=WUYSNQUf(~YQ0LK8aD5gfHQyWa5fbhW
zr<N-QV@i@4Xa=2yzDXuQqu$L3z;%-|v!l?GjCay*8)s!wYt6ir21!UT3-$eAg3zs@
zK_TMid{62>Hdc}v$CIxAK<g9!Rj+$$WsT`<Om8L1QsoaRK&2*0s$tsXbj$#POi=EA
zl0st8Of4@)Knj<6P_=2gtAR|e^Ah7Uky`ug$~r!QEG0ek*+?r-H+uKa8KC%oLJ{;A
zV*l>LeALE=t!Wb%p`r#C!vi#dMdZ1sjB(PpI!oqcIoFS=k6l<s&>e=hG4Spb&?jzj
zu%!NIaf*+Ms2i1^r#-KYJ)rG&mjg97@a9gBnTw?`2}jsMgZNHNjEIgNGC9)vu}I$H
zimZ29aSzkUnfcJ7Ls(MMYxZ>VYcw22%vmXu`77B``LOT|v2r0uY*j|jKaK0Ifi0FU
zJyM%8OQDEL%Q~24#L-fwAO-Z%uI_Gb?ws<?a`XX*o@&+t!i|*xq141U_Fd*=HxIU(
z9(2@@(cnQ2mKCFe^D>78u%NdxkDO$(Ug_%)z<fVn*hh2ZnTVji{+Z1H=kJ{-DQoci
z+(&ahj1V)H*(i3YaTn)ZGIBO7U%bvoRKs8x+@Im_fitt>#e$qvPU&GZ8KpS+2)QaS
zII}$C9uGPEYdV1O6C>dA)CQ*r1t(~WB+z~TYbawQ9y<`kfn5CY{T_=s5ZGu`T7QOc
zgPa-yc})>c=f!Nq1E+AX;UBznq9*!Zx_9(IDf&!tMLazN&PL8Oh@ps<_)lF(&J>SE
zfpZeW)wRqp{hkUpgkbKW{4;b8CN9ry1$L)j4LoU@0@4*ZjYIrDK-)tU)>ItiGTvm^
z*wfi2LZ1vinY)n~M*XIo?TKa>4n2swEk+At(xw;Q-o;7xzCbEo=j^E=r-&eq7hL2y
z>tMo=dc2X>Ee5@Z&;*mGcyUnXbr6<*x0_c3LGiepZ|va!sg&U&(fq}CeWNUneTM}T
zKM7f0KrO*7%ZQfWtOX+=S)=`e-G^I_c65_O{URRa+aEIX$j%kJ@*yN>&UISfXUIWg
zedmLV@KJy@gZ}8toCmBSY^-wS8NDoa&N^kdWornXL=E=CkGKlK0(jJqm&NAg68D*!
zg34kRk&^OhyTyd2cp)V)%(t>=Kt^Ve15`W;IK_k`IXJK^d8BPn7=}Tpr(#E74vDWY
zvC#Qvuf(_hS^EbBkh5)!zO(J5?90*=a_78W-eRzc+Hj}P9b2wQ3m0{iK4eh!#^6Zy
zk846+g4u(5E;mU(y6fDeCI<0uXDge2tc0qD^}5PG)h)BJ6*HBYDmDxUVb~C${9=IZ
zbs5;W{JLD4#Z8D^oHTtg`nQ-^rcjeHLetU4);Y_B|HBV?5K~fCT>e#h*>x7ya&#n<
zR(f_0yEeIM&yT>58kC*c60!|^P!R1Kfe5+Mn<8VuBNJUuyYlfiBQ9mu*Tcoc;174t
z@u$JcRw+2U{D*Px3dH!r$Yhx{X$%2`y#|6(WL0)=_pBI1W}8x>?=s5jEc#kApP-G-
zeO2@MgVSxgaIog)J1mDP=MN3WbnWaH_S0i?WF51{r4IoRhzF#NeE4LHKAK-A?$b_F
zvD+PWDLknspIjay&lD$pd-mbY$d6BBrwKN0@0#i!_G2UUQsbawaS^hA8W@1V*|2Od
zHJPR5Lps^pP|b&6P2pndMM(jVK|>e(0Eaqe5*leVt4X}oRIxLyJtJKqEehifvGBc)
z<cPlIU6apHl!CsQ6R8MR8x#Lw@v%uEoDT+nCnlb8zaisZDXpz4i|Q+u_J3k@_t}^Z
zejp+m`gu-o5El<`0>f)ESAha^eI#abCz6~~QDB_C=6SbdE8XJ=Eo$MPj|XpXXi7Ky
z)5dTqL@f@r1=UGVco54oqV*ZJ2qkNHP0YaMQ{<a}HjiqZG?ONyT!e|U4%7Gu^$dWt
z${AntrhL&V6}lFmjdl5q<$;F<;fqqTb=`_G^qSHmnpTb6MK-V30b3Zov9spve##8l
zBA2$nonRlIX~_%krDDs8L5Hm{|Mkahj(hJdiLZJlcV~z)jB0J$eHal5DSc!8q7n;Q
znAqs}5~^)eR|-;wmZ`6c<Xuxt4;Yn#_ZDl9Zb)L*9YWe>39YTI%3A`1y-Tl){Ml`p
zDH@K;)hD`n^wtRgiwyhe?Uz)bF?r>K*Hnz;)-!d<#rZ66oS@ZDeeoC0G+HoTY+eS6
zfPN!{4SU*l+cs`%)QRo7s%xe|gb)6qRCL_z6Vwq}yW9CK{H1=-uW$LQ-ZnT`_2e|#
z-K__VVS~JR@%>ddhuN88=~JFZ1N!t|w)NN);WvsaXgfK9bm9%Q9n%Xjdb(Np)s^JQ
z95Q0@K1;OqcLH1hK~^$Woy&Of_#SVP=cuGx@~rPzC4DxQJnCZ;A)T=iOU>NFqdx=t
z?%YU=b)&X)ju|f?WHN?W=@M|wwu0kfABr<DEb>o0)ivAkMzvu$96MY+Nq}F%l!Ah-
zc8E%)xfCexD2J6L@id9t`Z{<T{K<@DVLi|$7xT@2Iw$x`YZp6JbgA1@4gM2VEO|vh
zV}Jd$7noi+XktR4yDKU6KtwEy;56-<{F%R~djP?bZl|ZU?a<9(>pIN<{sGrYuhnPn
zsvh2TFEn{*J6dbmB@|HwJ46N5kXv^zVp}a=ZUZ1`_jKB?D3P0qqc5dLsvqxgO8(n4
z+{Stt%Nw*_c?Rn(FCHopLfk$jjjvl5lS%nQ%ZEE!Ws>=7oZ=?B?QjkLSGLwGU?XP)
z7V%<?+=qv?m)3#s<72V!0F*aF-X0}Ifw+6ynP8;D`2#f)mS~@)=P1BFL9J@bO9RJ1
z0dhWbls6C6b$-Z6bt4+?6;tMxY2-bq{wYF%8B$Q5nthJ48Z_3i91Wk64a@6A4A>UC
zU*9d>9MJqY^{PlVt|%S}O4KzUSaXhHKq#gvGC^YFEJn;!&@MM=BG%(!1rE>Xa9OPQ
z$Zhnj-kil4XPmi4&mByioAU_nKX3=ol3HELHiF^2BMFy+XWWx020z(yZsdrhZD%qX
zeZ*gt$(xhDq1?f2iy5QGy%nm%HEo_7mCLhG&?eESIqTqY(iwFF_Qkj<y}20lzsEiV
zG=2+x;xs02C8}G-+YL3I5wm;J2cE&J|BDw(dvt`HuiyWOLiy5<4>Vid_zHY7kFk{5
zXXzJDny@9c^A-z;Ed~2@&A#q`DGnbqipfe4A_%Bc@nG<S`Vu=$CIr&EsC+*ZD@*4m
zD8glGdiKS1Z%2eDSMy4<g>c`zg>NH4ySw{mV`;{JU*Jr6S#<CIlV6xc3zMkDvh)!D
z*xiMUN=gkRQ{GtR&s;Dz4XmNhY+65U;L4mzlH+LxpLEa<d}H+BWOxUYIp{&S>`YTS
z=uH-_SYuT&`lC+>&lmV~$Nv>$bD`H+J<v>+K5QQ-8(iord@C!q5&Sx{O7HHd<rdLc
z4^5j6vDFdshw2=6+zR+08icr4{!xu!xus7ZHiYQx?zYqm5h<nxcE95PC9(!nK};ZC
z+Bc#umh_2qsm{d=4usS=5Vq^FGXct8s_{Ye<@bO2pC6vjXo0aSF)I-Y7em&4EMh62
z5!laSM6PEFd|0up8UJO=mpOiPbfiS|+d(hP>C1yO{a$i!=|kW}y<~TaEp((D5#)Cz
zH8eFKi0h876D7d;qxm&eT%GW9=>6JRw~txmwL3Zoe5j-N-CB#WRh&HX!@%RDsuzZS
zs&^xtO}qsPirwtu^qNCbyIY*uV!gvNRDWqr^J5~(rs}&0{9YmAS+6*Ghw4dVEZ7KY
zc~z7cQj_QLHFjeflyBCds^5Kw*SG>ciE9sSuIGi#SkMASIzn33#hC}P(Tb)5^uli`
z$#36xp8dFDq2m>X+qC<x8V3o-Q2q!{)abJG8A12+Mk86TKlZ72#;HjsFsbziJVqH`
z^AV9!0Fgt~X^Q1dG~yV8|H4TcWAw8$+6ZcR{!<|}!<(cLXNo+NYhmy&*d#+6K@*p8
z+KOFLu{*GcqPs^!&P$P;mBHJPm5J*GBdl3fV;Jn>FvxBWNTXA<Tro?Nc9zlI@UgdB
z_>GYSM`f&>77`})fRN~kH6-Il+Q5=p_vV_NvW&iZOBqNpsR}(`zg524JWl%zRErZ$
z`)Z*>Pctnlrp?YFB!$mlP%2`v*(e`->{3Q>2F7^u^#{$`Yb=DL6Eg(dSZIqJ{7N^+
zKH@T?1Us3m2P(k#3o`B?QFwc;oUHO`??y^bgz9gj$ZRUlTLpr!&P9`0XFQfC+>ZKn
zRoR#kH4Pg`kuf9DHtre91vN}#h_U=HNe^#@Z@x{X%@o!@07_Dy)i3`CVZDDDJ!>J5
z@gMrD*r8TuesmHft^7H*BTmq>Y>XDBuH@Y$u*UHQ_USg*h8hu8{A?^8j5R~pmK9%o
zD@Rc>8urfJ9{G=FE&q_B)Tu0Dmhm;wpeb4bjY7o0<)C6F#<xpTjBxI0skp*`i4RaA
zDR^-^#t=D(D--*#Ak}B^8c#}$T0U_4rH)nV{@brmeO*IIzY^STKkQ*kJj0LHmZ0L~
z&m_{g_mm4tn+3|JW^a67y&Sio!=eO|{*+M+?r?}sbrZuUPfb@G9c{dn&v`UASc?`Z
zR8sFQcH-~-(1boB;K0`+_<t0&!=>Dur&+RO=cHzWHb0LAFaILhj;*2p<<E5nD{eK<
zW@6pXF%bE0xXOcoXUAWb<9rA2n3gky4dG01z1_S;?&7ca!I>GUJKbbF#v;_RsUWx^
z^E=t<h%C({ZJDW-I+yRh)bX1kTsWa8zA`+lkWyTp11WFJdCazW+1(!7^RbkTmzkL?
zk6D8E^~{ToH|da(yzzpf2msAP%Y*=H+}53&b&GTS@p88PbyY&PDSm}o%PuLQECXeF
zC#s*xn1u$DXY%lVdQ+K@psI-Pj6E}vSO(+GBnF9$Y2KxU(7Y+t7UIjeM>Y*TjEir3
zt&q+Z$L~$Otm0DENZ&|Gzm0$%+_T;#xzBK6orI<cI@BF=gD651@K)$1ouw;LgFJ28
z#cEn0GgCo~9m?ZLt<9gl<o(~%7m1LK?)gVSwvqEcY+Z@w-mkdJ8`FvIFpjfTTF@Y`
zsim+f<)q}!__zS?Kyh-NlM83+*TFg(Lcw@50X5K^+O%oP7WDPGcn+4*o@Jb1?mUxh
zq|iv3-l}XoK11aKx)o@bWTTXJ%tJ)mrVrO|v5&(7hNUadXONVG=8PV$<7oerz)iHw
zfl<A2C`f;_!NZsqP1uWSbmlr!eyrDlnsfd5?h%~lyBK{D3-U*15<)M(#-zL#ij)Tl
zs$G8t5|i*eNLuvqsiGM*%Z%8y93@A5elBOlih@iL{hR?@Y?IjyC1wKF&AE7vzo?k|
zI*Kz{zMCaI>jws`xpzw&Oc7pNQ%15t)svp;C$jKY2C~|AE3!~H8|%W8C{rh2&h?Ud
z$?8{wzkU>mjbF#?iv5Vw%kHe&6snTrn3vyP3>kt;BH}1{uTjZFJl2)j;Qy7lC6?l<
z%pJMZ?hEKA*3oQ@x{Euh(dOL-BN49TOOzbWBNo;!*Qlhw5U6fZ$(&o{-wwOyJh-Ug
z5E;h7&jV{{WGCC6BJ_PzzIp?g3#Uf>=$^dpPF@VUUV43%lG*q!<<ex`j^uCtpWifO
z+-5eWUB|mw_(2U2l&2~CNhXUNM9zAY9v8U&tpgB<BveH6Vv2P~H|#?~LWZZOcaM%*
zh#a1RtBW0Kq&iFJ^<Ay_b-T|4(2sw#vVMzA_L=~5;e*;Pu%b;{!TVIn(ECQHucQci
zmPCp?I60sF`dG?;WLml+YJ=(>ijmOb8XiFiwUcz#4%#Yra6G*<^}FO-TxxbyDEM9C
zG*C<<x0at`L`!Q@+DIC~m1Q*D&!d1<v4=IuX|OQCeabgUFF+p(CAiFBbbK+?pcR|`
z#?pHd*I>2+onl<PUS=YuCcb7j!u#IogQTADMG4vn7da7G$#jQ8xZ0nj^Q(g-jWzgL
zVYI?UCO)z7NH92$<hm^I&vfv1SGe^7{+zSAqIT1Ix@?SMqtX-VTGXUnW%>)875#HD
zPbzVB4S|=QzCJJkME&$y$AFsF06l}8qv<t2kdA&feh@)#Y86toGS=Ec^x4fQWi8sr
zHlWxL_&XELeha(B`UCz5*Hs#5pikRn`@?ripKgFEDo}$vzo3UbP<<CEsZju2AV%gN
z*D<Fng!}g!UEhX_UDR?JIpqbJ5Lh?8PK(%aBU08=;qR~Jgi5k42G?RyqH6S2L8kRl
z<EG_`AH9b{@Ha}J{9>P|W%QTh)ynR;=!AuhH6P#UZ5L-c+BH%VC`bAJ9gcd4v5RYb
z0(<2%7U3S3XbTqbp?^pnP7SzxR#U-1QDZfx<qljCwQhDK^nX}c4Ui{jr-rN>8a*`@
z`Dr{N!_5107%0xSzi3=%m50+7_G^EeQLcU5R+$y5*Yrrg+<BLZN3!067Q|$X^%CqE
z023li9@l>ny&0!}Bb!5yGc2MT%v_xpUTBU-Nbbm!mlJ*%-RTzK2N-Pgic&0et;ETw
zy<!YApUhSeN#)k5wiZwS>8r$c&9mgk1<+(RlfLH*9(ZB=i&k{S(T?jhypS(_!#fE=
z9X@E^8bOhD^mXDSO6J@snBUDpijgAS*R8&_^*qysaWxxNPOUFcW4R?Eq0~%hY?L#N
z66pCM*{^C3=kc-!lE5Cvc=`p|)QfLOC-S~mox<*{2=Ybr;163C9&bE~v%{P`Dol%H
zJ(dZbARezM)ku5$#?P^2rP?yt;t8qW^#t@MKGc1}ZPJp~7p-SBxavkfRfT9#4_f!X
zdSUnckx&K^rD;zVEtvHAHJ>gb^;2he0*3)o)h3=d!|pSHs$ueCw|xiBig(<C_l_Ko
z;X-x>KW@lNah`=Xnjxd<B{B14(O5OX$FYI0_yqWu_QCyvztceko$fduWY`nVnDL}f
z*E?AFAdTi+k*#}f#CLO?HXqw0e_7~?W+c9&d})mo^ZB;3vS;?kD5hK-7Co5=A#;B2
z$8XE!V)p}>Gsc!VAq$*$4ibIHPBLrCVT=CPG2jl4_QX<s{dak2OMLvBIRe&`g|#@z
zY>jV^&W4TRF(1-b@pgn;Tc`c`5#6%6By~6_J8}9I>7Vmy<(?gLE`p91k6Mt!$l@GB
zqNML2#k+cJ2^Jz}vodbWyP#KyfHLNQiMOX66hKs_blqqqA1mHj`eU6r9mHgGKikWh
ziFf5)`R#_ReCNT7$Idpz;PhxrgMOJCk!9Jo6)l=3m6i3>2tlnE?gfp6T8E7H{t!*;
zvM?zRRh1mwWN@3%Z+a!4l{&j-bFH5wIKW$so(1Wm{2W2oQ+ardbU$}3R$QsL?R~e`
z1m4+iY>W=f?Y(U;tgdbfVr5yP7wXEmx>6CW$xpmLBw0kP7H3vP&IZj5>Tz47Q#q%;
zQEyry52H>b!&pg4jrp{`)-pc5JCNqanKK*M^r_03EN?*0k50x%=F$7(y7af~1y~Mz
zfIJ-y9W3C(i{TCR2>$Lk>Jz7lW1BFVZ~$7v8Z>j#n4NL`31S;h24WgRC%uRZCC+ut
zb@oAV@|LoA83`&V$$g1!oZ_2!XYm|mFON|%r9pRgxrobKkm%MI`tu9rzy|{;mW=t`
zzAEKjs5D9BzgT$$_r;Od0el=?M0@Lz)&XQgwPgP;#jLj~CSi99p8%cfYZ9abC{$%M
zt@SwVf<t24LfG3vmE0L6zQ%<$!oEIX9YYN4x=Vu*sNBl5w!eQzN2cp57vc*c@7A4I
z1Yo>O0=6utuk8;LGeRbpUwo;$sqw_#GlfKkH~3sB2G{oF$t!)tXwC3i?|JL9x?E)-
zu<2QN`Ny<yZ=}4nM+mU}^_5T&AUyN^SZQEMQNyRgZlUEe;4FuXuj(Y2C}ns}#};8?
zLNd~(b+`v+GC1PR$(>HG4i+#$A<EumclWzk-YWZA{R2JjCUG)9<JERDkKGvUOBm?X
zORS;QtbZu1s@3hYu!`*$#Gxhm?rsV4%|;O=^1(C`$X-gj=j37F$}JrLE8XkG;lP?9
z9D$*x#2U;0K-2l~7RakVmkn7THRo_Bd^>X6fr%LwNC`ty=8z^ftf-MjNw9mc$i>}<
z%C6}_;ApHTgdY(z&?v@di_mL7nd}^(*0cRNmECAgm3_fh^P+4tn-yMk5inG9dADjv
za*Ed2A?DY?fs{m$iw__KKkJAil!`G$NKolzO==S}M`a}T+<2^uuf9Fo?8|dlXl=H5
zKmPR~FyL9>UdT(;d+NMCj$b?la9@qmZaa8*RGt@-eQ&Dp>3k2;RG?lzJ=4>g*a-}2
zI(FZv+9R?E72*0P7TT(`K&pX|?oI+N#wL}KyRNTrFF|wfavxZZQn@%BHBh-aUENBn
zpYWijq3%9h>RN+yQD6QJd`*1^t!thJLuzrb*~JZn4cJHrIx;=)s0!tHjxVBRI(BHu
zdfQ0$e&!VzA{5Bcm;INXeJMXS6&*dP^nk`Lu7BH(rw~_-(`;aeARW+8Sa}(=Y#|UB
zmwUJ5nCa3ZB!H~76ot5Wv?RpSV9az!j6m-oLM_BbM<nGRJ2Q43(3^~;%qb%tMLf}&
z22}fNMEhUso=&%utTyGIuv^^31FP_oB84y-aQNxbrI^n>a_SDxIJ>W7qsPnPm@uF5
zFc28p<V>$v-*P_UCj6fZ%J7pahAG~oa{1)e`Ocs_aI(#(#MB9Y&7*yE(3hZV{_NQ^
zv;qi`j{EZb>HFOGm*;JlTlF3bJOwPMBqWcaZGMKV&fUE&#z694$FdqKsW9VsQ%X#;
zU%d@=6`oFlk5$!vg!0g)&ns@8%yUjOSAtyi%kApftRA}sH;$m7kPu?1ze918iLK7S
zKzrDMty|Y#3r{oiju?mQ%E<PM5C5^E@ThybhO9Y5GoA{cxd(i#*pj&slr)6#BH@YE
zh-jVW6!a5y)RU9()|S@L^J9@N7rZ(h#sK-BmJKROiyHX|!{x9cn1VFg933@o7=DmE
zE|vi-6pSCk8Aw8~yPz8iJBJ5K_W#|iGt%*|EeyF^eDQjiad7?M@)fadHokt%S0p0N
zh{>j!EWPxtyB)aG=<CY)TN_9;($_rI%?>76Z%1DPlb4DAhm+;eSLSne^aonVIP8Fh
zOHL?er~2j+0jKVe&M4+{aC|hg*A{Y$`qyTu5~YiXeT>h9JI^Ttmw&D<POkR7fu~R2
z=ZYFkxZ$7Ah*H_8Foe<j$XL}$MT5grifCzqWW~n`uBI&uyIlMGqjyzne=S|dyl<@U
zk}xyZVcU&qwiGo!P|z5r@7H##aheiv7+~hT<-V~C!3h3WZ}iU&D4vD+k+ri>umEFI
znKxOn1D=MEf^aMZbu;C6?U%LY9luy8g*InFzyo<_Jj(vCE1tBY)gKXj@n@qXH?+F{
z-|c#E<l;vcMtA}lGb)}APY2HaEO`+hHkc@#?e3!a$;5?w)1NW<S=$=Xuh{KTr}V4N
zM$VnsK~q;Lye^I>t+*~)HDx@hfM4x;2o@34N?-{2aGFP3t(|5+8VL<j2)&qjP@gst
zT6@T|0;?_+AIgj(Y@#I_{(2R%8>EZ=vC)uB*D|BsFOVN!G)l66%0k0zZMGexGVN7+
z9h;F-EoyolVILB=4nL!&n@qSo4)lpNdO2ev8bQ@-u5&r$p$EbsWpxJt$uGxTeM?1%
zvJ}-s@a{nl+YYn^f+&^GeIZ0{Yj40khiNJh8(O&8H=_7mg88$Ad2EV!y!2T^OREqw
zvSRDHUVp>@R<;Gg-`%0Dy7w`H+1!DPlrGe*Bdi7e#&zDxEghF$nfBG_7Q5Xu1=4;y
zrTTwhw`V#MY>!fyn0x$zw7*fB*#h^l*<#fL1Lm>OAZ7@i7eYgpiOEak8d>0;jk0-S
zoS@#4aqZ|^EJRyO@nb?AcZ*+`aM9w%Q?=|^Q$H}7|Jtn)r_S5EU3UEbss%h0k*_sZ
z$_jy8&7~?;WZ31=*4IPE%u~wzM<xQiJRs(3+Mn9T>PWARPU-7_L@r{OJK!zwu@IQ$
z>RX25?eeH0`I^W;4ZWVTl;A2cvL|?uN66u1c<epX@pobkJiJ9cRb(P*z9YCIR-c8F
z!lR_bLV~iZd76+fO5?7i4#tAT?xyWy^^?oVf6wv06%1(A+(6i>GciKnvs%+66vC6K
zB_K6Ch7ET=U`v5}A3(g<Mbc0c)hfGTWU7QGj50oka6iVq(Ut6sASPyT4vul<CL|iv
zVfQ6D04)b1%YaKT@xV&n=lKcU8TJ7^PRYQ?wwKs`Y)iiUclhIpVle85&W5baUNyKk
z>SRv$xOr+$-pj$hi-WbXFbPHQ-1!T)lep)WslJWRDgc9ttrgE%mC#V)Li{|rm!-`3
z_ee4xiHV~4)&zHlxnQQZ6yVW0yt)ASy+W*NfnAdqL`e6JKs`q)X(j4ut%t|Xan(F;
z+`OAQ9OZrwfqf5g$MEQV#!@fUEpR8Cw{YJOzx%>Q=$@n@+&sOOy7aPN;mU59=FJh!
zrz|BNI>5`gCT^;R6~+HDp)jF-=HhZPVLmB9Ivt^<wGp7wJ3qj|^i6{!$U~E;XqEdu
zcN0Cc?8)KN9h-3%^sD*ruz@d+r$Ag%i{{>oA=~J6Yxs;Gy&bh)SXIqOGFloiIS<Bn
zoBrcRFqrFGpJ&ZWDm}_DnhriJ)Xw_R!$MgHYXH{$L4w)P{9Z|L)YkF5??01~dn*|S
zN(p0wM;^^b(Zr%^;L{_p;M!C!<cIDEXvR9*?)t;`>3POOpgrLQY;x%9wUx(MJ_}^+
zx}v9eG6n*Fs7OHu=IvaAdG4-Zs4>CTNX%O?F9#jh5Toke!~54p4{7L_niIT8w^Wfp
zQeq;BL5t+XKt?Ut&*G3&!bu*MRDA2?Z>L0!pA0eN<f-;ehz6)zE+k<fTYlrO!c?Xt
zvv@`{4RFYH&c_+j89K8AU6k8Hr#d)bD3cOq@Z;X*bq3rj<LK{v(x(5NU&`+K*~jQv
z1&jDv)<yjNeV(Swz@BrHnLLaM#HJ3&b$BGq%xul9dZWhBX>R7neLMU9*ivyt%QVM#
zs$?Iwp6}RALKEgwVS3-sdESlp{wsTJLEl$5y6RH<X1>ns#3$+sfmM+$e*E*IvE$@4
z+UrVnYY7Iq%E6~c6Lt%H<kY(Ou#26Yu~}k34cyh<!BLbD=&+?`PPX2<*9Evq+ro^T
zg$ObI_WMWkUf%mmw&{dV_^1ALmoLualhU?)M}FEJDs$Bnv+j~ELV<vQ07Wj4XUj8;
zvf#7Va_XNDWhh;xz3oJrUpq%{n9L+^d`QuqYge2-?pvPZu0I#)%za<6)$||wsf@_#
z=_#A|)p5O5Tj<62kCtiro8N)fP-X{#?Hw`pmii00jv<U*#$>+=ej9!(=kk`eLD$FX
z|BTYmEmAxctDWZ0=6lmT9#~nSw^oK^%bSD#@Dyj)Oo#LhxIG>B-Do))DT#X&{d9{t
zrjPPVZ7B<x4O6l9&uf>jH4YaR{$d7B_jOgfpzCvP4Un2r(XR*${v-{kKi}H<hxtOk
zSS=7gxHsCpd8g3aj46>nFQ9d9cOqf<pFX}^^=MO`=UVp4{{CenHFGTFJD)+k{PV{S
z)BSMom)4SdO4ZSi4a@F*q~k#%xw}H@qM7mULGwWdQ3dfdOG_ll4-IT!38JPy)1yra
zM_YTdZfZY0ekRnNoB4q5T=7W#i9=U>9|Kx!w6_Y*lT4FLAt-Q~;OQ(X#22`Cz$zYS
zz&K`P);asMe!Posu$=;y81_csw?`oPix|#fji<#nXy4!Q-`(yhtHo9ixTWe}#AVGu
zcZV9PXlu}W(hiR9?khGY=PiuJV+H@3Vcpabf&<8smM?Qj(I_BkR7rzUKtP;H?EbsJ
zcPjI<5=_hikA4wY95mW#Ot$*qfnW6?Ezd-Ygb%j33mXlVs`>1{oeq3*MtqmtrZ!mU
zRk{4DrOEg$lkanr`&Koqkm|md%U~2)7i0)iXcd3nQa^u8S{jg?je5IO$HUd~c3cjA
zjqnU9NJBvmakv7q9E5*wj7F7|xOA^1E^mGP-)P<2YGUv)I{kBcmnR;1QtP#y5JJx!
ziWAyxw2%(FrX}F}pV(HlP~nUH^h|RNTUMLrbjBkDS03v37olQ81?u!jLv_ElIrlCl
zh>M7@;+tPw?n+|OtT_j4s!zs*Zt*#2{CvJ)x7`OJu>l>JqVXb@1J%wJ?z!GuM;rgY
zZ7(ho-^Ztf?3wngd3e#>ul_`&5vd_q1HSk56Zg1g`VpvJP|A!FJUiF>0{@)y%(aK5
zq=XI;Gj7IYJ2zgk<|y<}wO1CUGzk+#fy;y8iLFtu?e_s~+zItTh-&Y+T_>d0y`q6e
zVC1?UO+Y|s>;I$a+W(pU-}kT>Ic!K%a+pKu?W}}^n6sRfL(b>(IcG0QOvx!>&N&|%
z<$Rii5R)WR&E|Y+G3BtFzx({~{R{Tk^Yy%+_kCa2bzk=!rMw;gbE?%a>(_6Ez8BD$
zy=N`Gq}YJD-rg<0FNvzlomMW+x`xjk!Rj@YF^9Xg)Z&cxBP@zD<A@5%D&+Ix;Y>J)
zuFY<0usQ3S9N&1-PbGytdWFh_J8kE0DZG(4y%pU}Nni4$h(5|8>j@f(7_d4D76bVy
zs>geqrw8lC0>zCJ#sPEo{EY!<S?t>kG_tvOs`hQn16%stt#Vj}j8HkpdG}@3RH&u;
z?}bcQ2CA?xlbadsgpW>ZD9WT0M92{3x|k53a~?j`%C^b~`dlH-ia<HCv&ji_$$!O(
zT~}|tb{+{jXEr3xwBvYn)l8qh3rI#81(+D6CafV@jA#Sv<KxVk2@t2mh3+E9XFp)u
zv{=fhw!q?P=a?<`#sxut0AGJ}x;el6&}bv|1<Z1-(3Sl;rx{wvg{6bpvG&zJ6{F8A
zC8b)Ko=r@Np4JyR6m8PW>+3)X9vs|X>-L->-^;?>cuXfXrtW8`>t1I8F67ZbK8d<l
zdA1Y>hh7<>@d8U_*!A-ZiG|hC$RX~_qL4*_jnuu;VNyW4wnQDb-lavCe<(1N^DZKO
zIBqt88ms162Omecd{;cYYhIB1R-*o!jpZAG%ujhdTN|j9Y%S#TRR0Au9hY#-Zzg&h
z_|uksP&V_PhoKnq)%H3N>)I6k_N}+2TcYmerq_akAO0xHrB*Y)zAs*RMUuA`!mCq}
zF1TEF{-xowk(PJ3YuDVXlsFP%=6aL_^P>wE!;$XAPr@$1EWy8Znh!Q~q0Y{Doci0;
zEmh_xs!7j$$x->GP}sYiOEBEByMeg=OHHL7+r$ZgPb2?EFa9zYAZ;){0#DFT`cn<A
zk+8Z?4~Fu{`waBP9Mzf82(rB>?8C;~&ERh0{t|rPLZ_{<;k7bc-@57r^SZEi5J8JU
zybd`w)!$zudt)cbV31Cti9SFlx?9axO(jcsTLM5K;qlD;d1>ot^_DkdD;u>hoVK)F
zS-Xk+JnU-_Q7E9EsK0;^)PLwhYzwnH?Fjq5tTVvNe<KqNVlIX!q`Ra^-j?kgcJcPO
zqw<${08Ua=+Uipdf54o9UX6TD31~msTbkP2@WR^L$DPqLd|0X|(*Zdn#QJ%~_j_N3
z88@@)XiMhPW3Q#iJa;T(>hMrReg&H`^`Ss&#;|k~Qve|suHENq$W)T;(^&CFzaJTr
zl;k+z=W3;`sIH;EYGK4R&;4W)e4Cyn^J?`GvVbw`ZAFC>&TpSv>=_TVD9KX3S+<`q
zq>1<xl#NDTWoExRTnx|qTp(*|G%q7B%TTrk=F?Qq8(`(g<NC&It;D<^@{L`*AE$V>
z9he+=0dUIkv(~pnhTh88i4l<(h9RyZ>Y-!XPW-Fl%IB)A%yj?D60~;R0ZAgtP&b?$
zn?2A0*Pm-It0)Y^nfZCmBeEB`J{sOg2_^Zp;8E{#I1@syQZ*a@a3G*2?|WuXH*@Ds
z+t2>KmqM^)gN9Gx5Xf7}u_-s(mRY%IA1YxJSSnF{IV!7+*q_{VH?Ur9*OJvkJ#Sm0
z%?Z2}h<4{R?Nd6KFq{DQxz*OOX1K7b2R1i402705*y9!HvlgkR=;z<s+uMg-9CD+o
zu@YLAtrpsW?knd8-ds8A?S4`AH|VItUcpo~s)Ya0X$TRb-qr9QOm#mx9W~-+{nmUL
zK)=vJv2SlIdAP^U3Sf&uClShCW#!=wtu@}_=pU{B-ey++h)OdicYSkX25aXz7lc_C
zmy}rUt!$19gEI5-YGB>zHub1oympD}TaY1`MPBE(JM|<eHqORdw9F*{(kG(ID8?nn
zZsK<L8A|#+(0OM+cjf_iXIqj4`H?t)*z|0Xqu;~ZQvy;Ve<;A@1gPCY^F2N3{!>O2
z_2Tb$sCZua@w^DzHa|O7k$1ak?}6yqt$`xa6DAOe;Hu85fecLKx1QCiXc&9c$%hf@
zbD4V|bE1rk7i$oLn4|Hn4f>-MVRrb<CO_c&IwLH?8T{U0<?|}PYy$3Z0Ei|C=yJ#N
zpwC|NFS;($(*+i*<%JNG(#Fg88t3eD#|}4)OB#$HSl6XD^SKOtyF{<vAN~&^`1$N+
zM0~pI8(Yh|YxFn>GQGUre_()>9>lQPCxYwqeb(WT$RwyJBJ}o4Q}%X?le=DjSwC0h
zIJp0d&{VQkFT;R01~l)20(j613YlDT16(X>)Z$E->@a`E&RR<1^`>uibs3S}SR#cQ
z2wo6>`o=694I20TzD^W-Wai<=m=|31<AG3_N7Z;AsD$NWxWO0*s+|)O%S_qqhZjYd
znSQ#`!-0a9+4$7b(1|cR=-+n~iu(5*BOyf7H}EcMI8J5XpIr_pkjx%?)LK#gVrhMt
z6xH7K#$2Bop+c*Jv;(l*cHWpudUbccVg4T;i}XT?>3DW@>b;6jx`EAv&HHsy23i`f
zeF6ou$MdPZ)-!c8<Vy>Qof3OZeu)P+X(jJo#O%|Ilzw0P7=G$)mXvMTB-^a*s2{kK
zm6iR!`%RK#Zg#+?B8$)T-wriYr_?JgITaNPi^P-;-GhyZl6Jo<u~y$JeEbNsSCIRH
z!}ZL&Wp$j&xIXZr>>HlxGB;vQotIzo!#1BJ=8*Qc?K;}Wrz_zh9t&l&I?`CW(wyu4
zbe(om<Tg^kt#p+%m+uLwy4gtop^?PU&WBOtfX2p*PYQtaFPX!L$jDCLT%0Ry%1gUt
zBRcM^@IQIXe_S2Bg8qc-as#^Cictbj^7dUy1StnWt+Fw2`Zr;s2u>#_XTA8cORGj=
zt@iEHI2*G2!1zEBCpRp9)`i4m6^eb>U_4<qIvE|{mkKKeZ^y*`Y-590jK;q2RGQ+3
zv9-N4%m6^T8fLBZuS7j0K5h02wB3UCb_fJePuk~5{M}dAqgzMiZMcMtQsg2usNe6%
z^wa4^V*T-KRv8~XjL!E>rjahozcoSd#qwd&odE7r5D2t5EB=-BR>AxI;-ufQ{|&Rg
z*D6lKAK<PGbjl4hXP()2$jT%927`8E14=k^+yJb7W4ppE-*JmBox0I4eD>6UtAqx9
z$8Zqq$i2K*Uwzvl+)=TedrlG`t3f<;mR<^7bd}j|cKWosuANSo*K742vc3lHqm-<m
z^vT3Yp^K1*y=D0MD7!A&q_V6|LKUat$oNann7Tf=SfVe7OJ$Wwwh%(*X=2K}D-FYz
zfWq~AMi%N_N4}rYrYZG2?w?r=YAS~wuF7jA>FhhL66XxcqvtmsBrAVXYsJYYJw^EP
z0EkrdSic4?Ln%raCqC_=6vTuC4}OB+R!P*QQ^hoYq_Tuds*{~<+d}$q5b)bY5$3YE
z{;1_q_Z)F{!0-DfR<=e<wCJ0sFdi!)*WjOSS(%UN(-D%6wx?4LX3y@qzwNn^TA9Pw
zYsv1rN--cYc|=6g-6k37$~R+Q(~6)(U)tOo_i9<0RwcJZ!b#Bgz!8P0fRmN|>nMBy
zH3@BHs^!6^oF?nSfeN(CMe`(OlWi<rS;^%Bn;mNWDlZr}q3$L>sM?T(FUi1_;SWQy
zMKgrM&>yJp33uYHy0s_+Dm2MMui93olz#3QPo~aoq$P1xvN^jy(S`L+T9W6$*Q-t0
z0gVBz`QGTCHIXmQ8KNZW{P8vgFg0cwx4*y5d}OIkt6N+FeaBd497G4{Ob-cv-dD@6
z%p0xrq{r0i+4{rD@#ivJ_wc~+`jZr-t3CJHEh#y!l+3?{cjz-Ax0~hugHm*MNW*a*
zO0F!=dG6^OP@VjZ2Kmd=6a0)tj9Vb-f|&@3)}$AFt6vJVwyHSOv9Txk0FKN1*sJ#<
zN*;@@hQb#4tZWhVa+79KE1K9h?@#~Q-n#5QMbXSN5GR{Gdsg>_^;ij5a{}gH&h1Ub
zA8ZvX%@hQ-)tJvSajUa7gFh$8Z9iQ2Kt1c7^5qnim53&%533Gct9)2evfp0NPRFhI
zG;;3TUIcyMHus7;a^Go-xexad6AicONVWwjf@B35B#=9nj9_t_9JvIZRfYGG$P6!(
zL|J9cTPXIZLT*6VuL7{8X$D{yYV%|=Gg6W2&;#l;eWAD@;>JA@+CW4fYbI|<@eRUj
z)^wcUU|Ck6G9>tV6Q^KsRFq`}iW`QMQ+rtr5r?Tb|Lu~TMF=D05b_J$-W-=fwGxIs
z61aXO{4Nfc=#-#Dui5FUp>8zHFHbKvevjV(6p<K7|1mHC7>KFU(Y@nnwJUpJSoPx}
zN*9Oz60Nph-p<2|D%B^~8JKE(zi;}QB!-t1AMd$a1-ZDA8cG(l(zUQ;Nf}*fLbEk*
z;m@uOaDyW9jz;(lH&lOScB^uh;-j$kbezml6pc4^<ffQ^yS4s4GGFZ+XqZ&*+!poZ
zv_owxw+jwTO}=}^`O;JJKYkqowfq)h)Xh{<?iI#eZ^S0T+ME0gg1Mq_xOA!z1)OeM
z=M8znJh#$gDfp*obQ+TtMPm?!KU$h3&Tnr1J6tL7NW+gGHf5WEU%_gq{0nZ@N!?MM
zz8BNu#h+X(%6zqug1WZ5IX2^_6n8r25EY<vdlrzi6#6i(K1~!jQcbxBGHH2>s_`X1
z;?tqTwtOBfW1Sj>$F(p(m$?&)+wT?E24yhIgXy)nLFkUXG&CZMf3U+d*6&V_u*U75
z67etJJ`+)ARgb@h9L&@u-a!n@P<@Zy%>~4@pm$><yA0*#CTR`Ws+g9Fybd+c_1y?w
z5&&6Q_*}4b;i3NGWn8_%{%`+lZ_JNz@oMMoV}riiFJPzk+w$WGA+z;rV>Lely=vUq
zUq8y``uG6dOQ<T)6L-4ZGHuoZMNG8*wy-&Mn&7*?EO8e&a}`y`Y0gXN!R2js@BH5V
z9m#cSzrNwSw$Kex=8C$UeiX-d&!S<%7;uO?VN3aI@onMLqdxN^MXkM^mOtPS%0RN4
zrM9z7yOY*=dMy!Q7E@Sk-AYq~JeLZjC<~s<ajB1fZt0MQRp?75QhpwH2D1%>8XC8U
zdgS;KhEmBpg6gQ>3S#A)oQNm9))tmdyQJGYh1tz7=s>!P1Kh$r?f9sjw+NPaAXABV
zFQoOxEqmPcV0$}Z1kUgK?5?15As5*YRQUQ@aY)xa6x7MxKZo}C$#F%4JWh!S3H)u1
z=djU_xW44!BX%RCWy(jcpKwgNf;JAr3^LF{i2NxX0Sl@-FLVWpxQl$%Q{yYwWQ5g>
zw&ie47#mE#?ZsF=?E=FD?*;(vkNB87B@Ai=5zcaV?Utjkr@OfmxWbI1ARzkk@MPTj
z%~R8;m22l_DnYJ3!%O}&i1cvR5K&Bn^<eL$#@gOkR<7?PuX$K|H^WwkPa}SF2x%cN
zXg?kqfmmLc?mON;R2YKhq)!+_n4uUp@V4RQD+pOVlOuTg5v>eyquKnS1H12V23kv#
zn^gEWPdiTjj!B;#Pn?<g+>NI(fSon`v@1k8QA~?vnmGG&c1k!q0Z!D~k9r%(Ycy~w
zHtzKJ;2_w}GPm|$s}sMTZXM^=iDH#gweD-StL)}Tl7G8(xr4n!G^H{Je~FG~GI9rw
z1&;Rp4lilBinjvzN~*s*ruFFbi?V#<%#vanDKFBR$NAA^^Vce{DiKz&|7RRwMeZ=~
zl2>M@ID_Q5JRBPW)tPg+zyHWchAsu&Dn%aBReH0@B-Er&YQEx^r?yY@`|c?_G>dN|
zNjU$5lRS1K1<GIetAaEt24nV;r3r0sZuY}o<45&|2Y_@&?9ry|R{Fw+x&MTV?P3e>
zi5KSwG3w?oIUa6EtHnmpsI<RF`#p9vs`dUrR65;);}fCwFifSl>4jPwd>FdBLm=wq
z*`Eje!OOc9D@V$Y;=WA$1-5kNBE{waMsqKQk%Y(BLf$HSOb3nr6hdfUK60U80&?A~
zbGsUrCV_sw-x97$6vpk748`{>>7I8;TSsOu<NI$fgGO6N>hyI-&5#%W_v391V_hcG
znZ33F<#$CmWn^Tq=`S6xqef?iyo=fy2CcVhW$IBmbrNYw@rm5-bPyOy1efUQpXudd
zIISdn5a{H-sIDJVawJ6G2VeRVk;m&$BF6j`$cLFuHu>esCM9!y78%bgc+0!Lu?FE*
zd4g``Mzcg#;U6{2hX0feapW@Ek{(-TmX$4+3s$euFq4y$D|1Ve-_1;6=)+xQbWuB;
z9NrrU!JL<6U<=2&z6FCNoBU^Z-~3ylHcqLjg)B?kni7#$yX2({d<|gsd0+f?ZwUAU
z{@C!HyM}qH1EC)Mo1#*L`*Jdo`DY2d)~7O&qM(|-VUC#+;PiMm)XVQZ0glyEvx37b
zudCx&yUmc5IW#s>-`X!#hQaMI_v)A%7eYqBeVnn>Dgx2eRMU}8Ivn$PL$0eFd^pNc
z4qq+K%=IHm+0yqCNpjelZ`B5i20$c=Co?33Fub0pWYbJ~e|$=i22R|5FU>f_9c|78
z=(OWOr5)vgKiUj@A$^eSKo-UxTh~<8C^Fcjm3xwMD?qHUZLmxuvH$DecwOg9?gNs7
zMCErQwd14ySJa<AeZan5*l0?ba&g<jh70n!he#KpafTx<Z5rw!|AFWC0oDIXet(-9
zkt+bwQ<g>y5j=x>t)fWBBpawBYxJ7G&Y-vtMFB1=cLH1UelC{4GdsX_N9ayoA?lfo
zYjDz@dnRfGg<ht^3~yr8_*eI_u^fpJSIh1>!5o)=4`nj*j+qs!sm31fKD4AzZ>y`?
zwE>3*#xr$Co0v^1X7oNV=0+$kR{Lh-{;PE~{f`Z>`w8A0pjM5HEVd45WQyF%4Qepr
zA|0L97C0(i^Z=_zJX(4=dvo&!od0B<N{F>1cfA?w+S*f1OXJO;NF;J4Z=Z_N86~W~
z{hl5-_ZsEOF7%+k)u_1SKE~2#I`VqmHBnJ9sIc;vXxmt$Z|;V`CT`wU8)k|IS#E#o
z9xjt>|AP|5DY-<>UJ?5kt~`d2;N+BnLt${w?=r@Yc^d!3JvXihHK)tAHoLEG;gi|S
z#Xl+e8n!q@q#J@A7a#FunCLU_#Dj;Nl28cNov)$J#=_2s%bYMv<nw3OIR9hjgdtwy
zrt*;)Lc)-g_)owDQJ+34zEdF9N*yHHbZ#Vd-3<&5N9HSi{T`*6kEW-CqhzO^p(^nR
z4b+BgowMn~XGz_+E^lt5xql~t!My&P+l9H;6H}cYp<Esxh;*or2s~Ja{n_w3I#`=H
z`zs9`{M$I#e9_fS?M;93*kpa<P$JPuGb8UE&!=sL-<v?359TQo9gDME`RFi*fZ*}w
z%HC;&h>3mPbDlw)Z&T?c8q0$)seIUg;v*)r1)^l`pB^kZ=!-}DNC!NurwKr+*EkYn
z`#D4Ge2}|%``$P^{R^w4pR`N7^8TdcY;%)v8JydMAv1|YcH6e!!t4cO-}n@Xyh`aw
zT{t?C1#(xtK4?9^-luoN8(Xq+;qSmsh-(X;H>i&D5T2RR4{ot<qoK)$>lgyOEXwYd
zWK-M9JGC@?8VMCCUf)d_R-F~gE<iIwxCUbQ?0kffWEcLz*h1@18ICR{LoF*EN9%s0
zKy*N5Hmm++Ji8-o(a&~#c2<wqIN-r111=e;yqaPC?;TZ^)9iUhAg8ip7S35yLQ4xd
zBoh)qZC6OuwWw#?VoMWtuqV8`Dsop$9Rx1=GfFXh4}PW&5&?s`?WG74XS+LHQb#AJ
zWa;;53dK^n#$!!D$84`;fOPN={2KWY_TkxXb&0p6{BzNPzeP=dlT~Dm_71JdbM?Tn
zMQ7|)Zs+2?yg<?-?CROB%Gl>iC3AkoS65}ZaR2)pOT%mA{T6>RCR^+_Qln5Ju~X>Z
z=elIB;H0e|hl=bf*P)8u5)}k9d>6N*57E2!x-vz0yvthgu?ymLq5KE)g<I|2Q<!aH
z!OKv@P;meCXN$WOWDVqlFH8T2a{#kk|11v23e8X?jV*kN7r81XBs3^HgxlnGp($tf
zC#R~l=<;Obx2VzT`H^Hl#v!@Pz1Ik>o3kFrZwitJv*!}KuV4j|f8;$a&a`NeGO)Wp
z{TONU{B7|URG>VUrQnTQVy>>?Nr}3K93$}J0WRw?o0<xpHdSNE!&BpnXkhguD7s)u
zn3~{-=(PBXsC$@Qjd>ISp>vbbNbG(=%4cH$TTUV*6wgCIH7x(hYx>+iU*#k)Fe@Ch
zT70%N!NDCRb~tf13>@xOa}a6r8Heu9yR!tB*KUlhZ+dX*y)G>|!&fL3ZnusO_Yd3V
zk_%*jh^V?cn3Q4r@KRhiP|)dlQ2Gp+Is(y|E4Y~q>R%vfYnF~IBLs@oi2eas0rfxv
z1^q|hbU@87x1I%YIGk==R@Pt65iTuI;v;E!p8oXj3~;v62%OMBgF_dT1o~roa(0;Y
zwE4a;$$xXo!vd|+AM=~)@*7-Q$vsUTK96O1iR0Ar!3;=3p8;>5rmOz!J#S#@mIE&R
zT-<3VEuaR_s@etWT-;@M&wr}EA{FRjK}c;~{J+CPW!$Kl7R^B`ihMu{lRh70HkU4b
zYZh<0T769(F|hb;*_CCdt${a4Wov;Emj_c&5eB8ZTy=J}ksjmIj^J}?aT*`6MrY2l
z$q~l~^uStKfQRxq_xa;5%+EOtn&BrMlz20JZHdj&rIAT_J`hw6L@$Ie)>8!GI#sX_
zc+)Mi_L@$2)J|45_cjA)dFnj{+9^q>Cg%tg^G}e+C_R#6CA3cwimiL*z3{BO9iH*e
zMfxONg;|MWMtS&Y=SY??ROo06DrC7#KN!Ol1Ohl+H5`Yisz5%K!AR#s1YLQsv<V}6
zi|3B*w@Dps!q}NTGB$opmUKHtINUH|yAVPD?*7x({<dG)3r0CPnns>o6Y#VHOU=*1
z-k;M}N3>kOQu~Fg{|i$#mhbPFNVvtY8mVcJ{~R1&m@m>-_z2o8qju0T<<9c~7stX3
zssd)Ot>V8Al-D$&*p`$YPnuPz_XqI}ug6>Exbd?NPhLxa9<9?k@hcSON&J5F9y9he
zeJL(lQ-dQUT!Y)m(mKbj`=LeaUL3!+i{WK6AErme$%9iXHmx#aC#UNet;Hrgs|rr0
zVEX?u-iN4Py)!p!OXiP9=K9?g83{}XM5Y3j!KPA{GhqqxZm6owhIWN04yw&4x2}Yb
zUHemr4y8HanpURlSL#V|-XLF~zs6#B^`by@T<I6CDd4p6?9bHNc@vwCr{@?S2#F>u
z{uDBNK6<>K9~u6{>5d#d#2TL0S=V4tq;I7yQel>ulJtUP+r<Dz?2?Az>U;r><Gz3$
zp4CI7y7YSg%~zE6iz7X%4k>*uEpdkDzjeJby!hIt?wRXy5YW@x8xtMbImZtlQ8~Ei
zA+0@#JtAPWWZuc$j=zV@E9j=k8`iB?802?qDKi=+tt`bc4Q}^s-13mRe#GvNC(g`*
z1nFqKv%K|pp*QI#-;dy{3TyHb2zg+?Is9^4wm!4!&_h{S<2n(UuOlt~3cj_L6Mn8;
z$!s2F=w+GiVw>!gM3cP4OgM~L`09k~g4<{|G5YblFHLyF(+p%{9wDfu+(S_e=A9TR
z0z!W%bqv+E(TA12)xGv6cWIvF3*-^R4bWF&HIQcNhLiIWyAk(IAc<pBE5OuRI_;{B
zy+1i>1dcGb7>w;{QmE@Qn~~G^-r*0A7lfM=&t*&4@V>jcyRY-~2?Ho*?L+T`UvGMM
zKSLD&3HpkKdO&yi7Oy(0D}xN>_kWKu$W91MDS?95b-J;=cU0}wRMcXR{z^`?J+gni
z+To#{X?SV$cn6k&ka-NT26*e8C2d*+-|6rxFp6}KW4Bn_j6%qMtS$&dzCIvo3cDrX
z1QV7li^hjs&7|Xc7xb%A5W&YgVKJ?{9SGl1M2J)kz6_LK#Xg0;yf!N&xYj+-)tVCW
zUhYgQ!307}WP%Bd<y^4oN~86ulT8ILD31}+FpsY8SxkSXJf9p;pJS*Hu4`iIB)S~Z
zS@cDx-AQ%mTVQvm7SdH)aJ#J`1Ee0$2h)8MW|{@dP>-KY9=EQwPL)mJ5SB}cY+!~e
zF=+3+E#-B*d)FX{$~-G854u*zZ<I)I(;U%bHov23lc?l$e?QN=+V{n(G-pxD1HeSb
z7<u*wICFr$5Y+=Z+OgL4wijRm_@jtE1|O`%`M^OX=TT(c*g35`Y);SKWiZnP+2vG|
z&1+v-MyuKP#*(pg%PUQLvT6AEe{p;(@6I-+uDZc;+Lw$ZTE3}!`l}8FYCc~7aR+~M
z_D+mUB}&!qKH?`OeK_v+zD|H2>qRq!nUDsj#%u1x$)gkCbe%s^F6D8hMFn^rvw5_z
ze>S!6elKfTx2@P8gu@YCK2_FL*GF&7{-<$zHmC90L{~Ma0k>;mDfO<ifEQ%<5A^(!
zs*zcR;i8a(+~B|!_G=as2Wdq)TF91L96K^MryG9l?;qxeh>0+^H)uoLnn=+;7%GAA
z3WdYr*q;OZi@H@<VuL36cFQ^S8`96yo_>j1P`T70A%Pf3zOZVYwQo;|vwfgek{kKC
zQn0kX?S6=U7FTs5Nblc`37A6OhF+d1+vV2`9ZbqyR7l7zv1M@~9t5u{x@q;Y-g*Jk
zV_~JJkn>mj*%?<!dolS2pXa3>(zAT2ZEx6Y!w~%dq>oZO+sd;iM^GmE#XdiT(p_j!
zW_pPljbazVITSlM6q9g6XPP4DwMg--b4n|PV(p&MTsfVFN*5jw+to}Rd3DHMHBYp|
zj{_h41d;d~=$%Cb1XJOz+XbutuO%od`jZYniDtd++Yo=34x&q8xfJeqc)Z&qu>#}*
zrv#M=-9+3Lbm{9(@4NTp$z<&$$A~Yfed1ik{aT4ue5J{^Hfydh89YGh2J5e?<~fV5
zS%|4SyJ|a^zFxk|6fSS!^DM$w-kU*z5v<#Xl1=PF>(UWh!Y}A0r^+D^p%I7kBSc}u
zHZ@%w1bfO0)#;)Ja_N)`)z)RC0y_zL=t!j!Xsv%itu+M<f@Hx{IKk%yzE%X%=LIWI
z?BqpsD#KWUz@7r(7aF56N2~5tJp73*{?V8_e1<``hyRqhJm{CQL(r9m(dHo_*PH8h
z!$Ci6D0K_`3ki4OxVw;%Wn%YzHnwogL*%zukzczcsA(RkkZVnaMUD^nQ05-n<E43j
z8OGxXvb<?AaOWy-g!pl0(P?tjZlLm7HjCLe5{*zhIU<j(y^o2axnfGJB?epPAUJ>C
zy16_eemjNHqP4Oj%WYKLkV(*1mw$_&UqG!kzs`@)MQW&09)J2&eD-qP^&lBVcV753
zJ3E$6-P!(g_&f=BM>!r~{7pCMeR49^xHcu~!EAC1nd&vlkxpIP;f48ZiWWehMrur~
zPu>ctD>yqOH%h3*9`4GNgiL-Vv~BuGO~oT^J6fxLCT}TLF4BFvm(|f(pYKinS+HLP
ziU<$a>3<I#G;NXnY$tNh(|c0NqickzptQkd&==Cqjqyk&E*AjV7(4eUWa43#NP11{
zeGn|DeK0sb7uT<=F-z<x;><pauc*9=wH0zMe&!Vpx~X*uVY(`UZ{5dxhiyVKb<#u=
zqu)9j)h=gcNlYDYoI1_zhc&j_yrypZgFk#Xbcik*Rf9Pa9_ck}{e0vQfW7}5Zj%Xt
zB)-1Spcci>Gz#!*KN5uI2A>@aNCQU%BJqTl3Dg<_>Te@hpj+c{Cx4%+0v(24ooC-A
zx#3UtPh>vZNnJ%Kl}xC$te>5!lwVX6dnY@)$G;^TxPa8Tnta0XgDt6?3vt(E{`=1$
zMc+A}VAXs9)d(FLi?)d^G8#)x7qIWWbH)4~DB_~!5Fj8u^Z`nzd!<{uUW_^>mphe3
z!u!axEo#0|P14`0xTNFr(T<G^(hTbMCfL5!0V_gEt&#He#Qte+`hgF)JsFLrK>ZP2
zl|y}h*|!6l=Upkua~bf|V5N5zg)X!CqW8QWvC8|kM)3_q%GbO6h6!qhofOO6KJ|B}
z_YliB0ALFcN)<#lrUFm!Wb5);yrIXg8nuoM192M<%oDtm1HmG$$PY5+g!Z<E2_?+h
znD|w3b964VNrp(J9@ct_)LPFE)E=N3ZB}_@KrY2*7y==RtEw%)Z_dt)@(dzgncWN#
zrjv!yjp=@7UEE%G0?kJ^V(V=Lxi}NPEC8FgFh};(I_&U6Go!*&9PYc^m4ORvYDLC(
zxap&Oz=dYY4xho>d+lqn1QkJ#vY?`y0z0)x->N6r%E{`L(EO{pCgvIu9o_OXDl2mL
zd@#|NT33B~4O_w_16_6AoBD2#oS<}kL+(x%XT6P2LEWmvPFBbQB#gsZv1HA-1W-HO
zq`X%FfD<Zk(z<!Hw`Z*=%+;AX=4H1>ro10-E;MqJy6DR2oO8pO7xdxYK#`SA{c9=S
zwG|N@s5ij&vbE=y;ltoG(@Y`3$8}prIR;|R9BLW~!XlNLq4&U0WOmlF1;o@x!0RO~
zq(I-k%|-`oNs0B#{as*sRXd{$EdF|~dnv^e=C6?P04hA^_vO6E3O#DeA6u|*q7`py
z%T^SJjDVXJPBcGM<n<@Vb(8G<AEd_BD~$elJtttRXg(VG?waMFwqE~e>W8^YC(jj4
zB!dh)=!xlZ@-RrQ+u^^15cFv8%p7nOh23<ImoeOnBM@?{b8YoBN(T&qq{IcSiUQ$3
zDk~xYqUxXeB7fL!cZm>RzjI%bqi^4mgfQqTDx9c2FR;na_wVGcmn_l|gh(04cK!LQ
z*YBw7Rh9rhrBOOSU!P*w`eB040*V45SpN!g$T^<Z)*nQpth1o!rf&ZUnpOKoUrKm;
zAP2RIxTVd%`5eKkL)ZoKDxk`#8#PeVQSEO!l28i&gK%>{7iy5l0t#-_H(CEmq<kDS
zKo$DU=n_;_h+>AdkR96u=I7w+TXr~**Rtmpp~}t{!w{dU!|IcGJCDuH(=ksx@hWdW
zHlet}srGU)38v~_M=7(ScMS2B#M=Uq)=V|=ylQa~|Bk1sfs)gWtK3@tBN%*}|54X;
z1I_IO9~~iMs*f`q%xKO-B<<ZN9>Pt(Y&Z2d?B<y=#-P)*62NgSEqxp;!hxKwavC}&
z?G3S)5dqXGIf1}Efyg)u9W~FN@rE?v<U^@yf**!JsxMwug5ow0Ha||MoFjf8r@O-{
z(W1jPkO=_IlvZ2ty{QzmIp^R;^+bsTE942p-r-wlmguWgLvmZ!r7M#x<KIZ~k26z@
ztGy(eY{J!JB4VzI!;&I7IE2L4Hp<o74#Rz-Vrf9}31NqNvarSftkgg~QIjJ!^6^+?
zT*qyqBv&buv(!`{nL#~0ULEn(uO1`1v$Nt~tnb{|osz^vzwfV<JGSBsy%!eHn*I2v
zP!IKUu$3<=BQryk^vdzcqcH?>{nt@frw%82bt9^K^GMZOG@QvYo-SpOPOB#R9GLS{
zC-i)jr{DY|R43C!0TLOE9=$P*?W!Xuy!`Ea3GlbFtXF_$We)|KiB#W6xTxOk=T}eE
zB*ngEzy;R~oqa!cx>57AvSKCpaLggjta*(hJ@@{-zpeR`OhuUguhY$yP3A&qW@1L(
zHLeL!A8-nI5nUx<Go-ekntA<rO)7l>%1B|bR_OyT4KqW8*54{znHu}^j^hTe1{CQG
zXS=RT^{lh(rjqu%Nd7OgT+Fz!ak~w_dT0DN!kv;{-wvbFamCHN{O5uIDCA$>$LtBL
zmp8mG`P2DA{cA9=Qof{t$q^pLh}?kuNCAY=C0!H8aBpHj`{~~%T42S-MMg$P6Y+$(
z(|u#RJH|V+Z)(GzG|+I#sO}#}c2YQWXP^ccrk}#A2@hHHt2Y#R>6q$lU)5>#9WKeb
zcugIYcA4NLP7}r<M*s**|6MqQ*wWS&K+E6{CEuZR^{tVoZ}?*q|CfSMBYIVCIfbiU
z-mMe@W~-e{Frc(9BN4gX&ARt@$X#KUAXRCuB!@ZDo9V92qu%E1hdkyX9Mp!N$Ow_a
zjFF`Q7s0Uhg%YW$H>>!gwubUiP;2JKXYoCfkEnW7HxWR7J=j?pYkdEda6;5V@_l)n
z<94{a5!BsKqB>{iVDB&OW96|cV&W67o`_^W@{2y*-wm`jT8q3c53BJaGd<%}^!CG$
zNjBz^jkG`D*G^dJDKx%_glpKTL*0`FKTdhA4&9p`cz6X=RhT8BG3!J0*{m}cw}xlV
z#?}{X0T}-aP>~f^MpTwOvqy<xYwKZD?D?I-;D8sIIN!lf?wO$97z!4~;227#bEHF=
zjut91SK4-4Z#vrRid>{Q<<dMz4dwW3b-zWuz?)|XR^h^9vJ%^4JN;<>FYex3(5gF8
z>Tj`_b_d}7R<s%qMc_~Pfbg?~yTZk&p*z|DJmmV?)KYai4Lm(NIXhH4qnsTyo;|`g
zxW|uxZu}&a-yeb~2meN8Twf#|k-A?Dgu6PODNWQ2_R;q<9nOpszl{sA+f2Z%5ve?}
z$H1Ykrb`Gako`b(^5OEr*5ANzp`6&pfcCvGOZj?on|*+|n>Ou$#7AtaCq=)I2CluD
zgcg2DXNOuPMuqAyX@3$AS#_RQ{yzV+)vM;k$tCuW6Ch0%AzQVV-R_rx?>^HFSp5C}
z=s^{PE>5oVS3ixYd_yNFuCkRAW8f40)L7$8@jsqPXmp+2I~k$H*UnEN=JJgM{C-i&
z=dpeie3Itttfk1M!5K&ozPxddF51G$g$D)(g#=*P^n&WcBI?jDnQrY7TGj5*w7A6L
z;&vN`*M_`v_IQBSJOQ*Of`$gs5~&(mXH;}8MyQwMPwv``4r{K>&T=dK<KKm`4d}L>
z=vS)DlAm&12^=kJAbuHO<z_{F>7SjQ4bdO*g=T@z<qtQE_v<W%%&4YPg-y?c(hk0q
zRCJ}7n$+(Xd?fpzSyEc62)+JfMi`9+bu!1ImTODTg|`Avh99rXmN9z!*lorDQd7>W
z?YS45wZY=)sm;Ra;j@Z*iq8wHKItl3K3J1*v6I?Sxo0lbnVp(gXM`B^<rfj>lXMgp
zt>k96yDd*57M-3_-{Wmhl1~`9TP=I~qX-p-J4#P9iTi5-b7vp^{rhto42XyDvY*2o
zS!AIiB8mYPneONZQw^<c2@;A$M&jW+>)z`ysJQ57J^cnO$LOybe@@e4F>Ypr?~4#I
znBa{&timoT&l8EP5ay?e{j*Q3MAm$X2<JqDv`%Tk)JxudHBn@qRgg&P$e^dUUn3qv
zQXt;$2w5FO=z-M7yDq`5+zkS;f7dDH&P<KZ%mQdSpzP<qn5|a-);et@e0MGPaD{9@
zUR95C6+^II+q>qFG~~6*1$mOmIR`|<#2mlH9F6Ue$z!cI&v`t~Br5z`fA>$@L4|bU
z{DrN1k7~QAuo?&SKS5hz>kA6fqP|Hu&(m#VqiMa0+Uhub*Z%W++scff)H~>qtf_}N
zfW$fBo|JHAt*I)k*O^%`v5vNg>QWtRW*)-_Q%<J`WOt%rx*#`Ieb4jPc$rTq+lTO9
zj~OCMpMW_zIbd~6NN5)HxyVgTCRC;{2js?T4@5Qa^}IV?FVjRDts-K-R0N6NAwJC*
zQV&Z^N^?w7SJ!a7^o+7C!(k-X!tCE5P#0su4ncn?OcWJ$!jrn|I-@cQx(RQ6L>F$n
z1;iR(QoMM9W(IY7$BM+yt0#Lla}8m{C_noGuUmn%3YEOs)q753)Jv&Eo6p9FXrZ%l
zmsLCn{k5-q-!CA{d=p0VQx~lkkr+&5rAq?S*D$tnY+SAVk+D;M30il1&6O_NPeE=8
zg~L!+{}d01e)G0FAZGt{h>LTDpSRNY@@{XnpRomsgV9&r{jp(W`1}HoeRmy`+}dER
z0?KjlF8W+NNMupv)5%hDq>>k9?tL5^Wz5UD7bOGUZ)k13e;MkYV~*a+F?Ve&q%rS-
z$~uDsCnpENITElhQ2|}m4^cI3#cMw-3Y4k~^%Yw8+Uk6y4jSJdPwj=pM#qw)X^&j&
z956$O>nh(!oS$qv^u^Mm6H^j(_U~W`HE0_Uo%c%o<w%djIrZr;15r#BKa~wicp!pI
zWN&pNjVyUNQPg4;zlb3&ARypcp?)Uy{;AbmY6S*;iN2y18X~y<eI1?gMVtNkc8qf5
zj6LB4iJ4+IE$NB8B22eM#D!(}=b=-8){U*(bZub*DX}g{-2_DqI0M|v^P`C9=sEOI
zhd6zTV9+SE(#=_92=w01);xW?3!mG*x#BlH@?`L`%xXiZl@rIn#D~BjZC#xC>F(6p
z(a{lbd<3Y(0=3cMmtE~E%8mG0_TC-Oue&6j11ab4;cYPl^4`s3<Jt|F<7dEYC(3u%
z<(ndshiQ;z0ddD!0svo<tXI3{9`mh5Bt@zE#=Qf7v>0Q6Pkq>&Es^%j%wQC#HUz>-
zYj#<yY%qV@FA$^3_YZc&Z$s(@*Z8J;!U*0D;h+0IX7Vx!rj>Gu@Mt?bgJdktuVjo2
zYQ(!N|MHei(qe!l$s~qs@c^qyESfBfFj|Rk6?`qGx7z?0Mre=qSTK9`q4fDMtbKIQ
z@HG!S3#?M(<1;OM?eR+}_z-3IsngYo9du$G{1ED*Cu-n4B$1UT>umeroNiX0wPIx<
z=##eU%S-X*>hbjR9Zz1U8JJy)^Il$N#~hIfZT_ucRFEYFqq$3J1Hv{NXL83*4*m>0
z$=6Qy@1@-4ggJ?)ppC|Q3dItw0*=~<vxkA=tXi`4^pBIT=x;ey+=4T4fYBU3IBL*O
zZ*BIyfwhM7KlUu`qPRcP>={f2h0QJK3LYffJQKbXmFT59&5y_rUY66#ffVH`<yH{e
zYzUqoL4pS2D};6mu=*$*ShK@~GwDvnbj?nTZ3Lge=?HDTJt}S=y2s>J5+%BSOzrG+
zksyM4U4~;z2Y>%;k-B=Ldg9LF2nz0>#OoiO=hAN{`w{BULLuPL>7UW{nD!y1xjDPl
zTEQKN8rpcx%bIn5rL9c!)6)#$FI|t_Fj&iO`^IPr`@cWH)P&MR|I7OxV$R|r{ODJ1
zWm!$(wD916`MPng1ZSshnZ+ebSI!{wa%B0&vrzq=M0WVXDu#-|_wN)M@N8fhURH#v
z7BQ51HIiS52RS1L16dza|2aZvvQ_2~=ekYMdnFJ{a`)TiLUW^9S0qciy`SRDry$pl
zu+sVm^_sv$v{KQU+ri&Wa#urT_D=BiN>*2$nKpcEs3a~Mg#3DAe8=~iV<vFcg(W^x
zPbvjn(pvDeJidB<)zfJn5@r;lq&9nYyklHFbpV{rou;o<U%aJWx`249=3W{I`F1BB
zs-kVl2vX!!Z>u|s+fie3N<neta#{HhqHZkE{Q+Xn3nL&XAyQhP{1ZV(@cf^;grsIe
zE2K!Xgn4gaPV$)e&&~bA-Q$yu4Zqmlt%>%-u4!V7QjuCrL{E3&#+M3np1VjVaoS<f
zl@i*LesvHPUq8aq?pcB1C&iK6!X3!w<#nCROy4nI_abeg>h^~;cF!#=#qN%%(|&_0
zpsI3}Dg7-SYy<NSRr~bl!+GIgYSk5}Vs+%;tf<npF#iOZ7;Djs#Y0x#AC4pXc+t8d
z#?712p`wcx!Nn}}+5vgXTQOhph8L<OkaF2SG$3$(rQEDBm3#cE-;AeJ&@-_Ptv;XJ
zP%#;LNJYJRA}+YtOiH)bd$ysSGSMgvOiY9k2eSKkh2eC%w|7bRj=CC-dUgJRHuRk3
z&x1JaR}Vt-*J|?=rtc}}PT`Q6C5u}}ZRY*2jWonG=$Yg%w|~Wt6H~<R{dgEXQ-yip
zQx!dP<JH4w>fS%?EaUUOf1_s=7OWPyfIbmSj|*@IUE_C+X9ZuYtzeyJNP65O0fdOl
zFG5D0or+gGhs4D1?WF2v>F2Ks^Y!`Vv)rw*_Q|K=nN?oAOi3Uqr8&uJbF=~h9Hie_
z&uRKeY2}7Sf7A^2{>}f~5Io#^NL|B}F*&BGFx$qXHoT2xrnP>inZy@dnfTMSvbtEp
zeP>oy^_W;|$v8rO?CJ58$^^j=C|V5-2kfQJCq{w=^^0Yn+JN)An|+XibTT2M@n10Y
z*3b-Dcisa9LT~z5Rjn;pkiAHb%eo_}a~SV4s`M(xp-Hpm$F-JpmW!q+)26o7<uN(w
zwGeV_!0j9QnMpUU>*ZH7!^UT)Ij<B^5ORHQ4OPBQw$Xe{A+?*+-?aY%FWcDZqA^cy
z)Sm<nI;!jSCph01i8g7j=&E;)2YrD(<D)1D4=VOj%FB{J^yL{<)0C<YApdA1`a9IM
zSd0@xL>L$lfL+mzXI+m!C7lIDcrYV7XsG6~$-v+D$Meu9UiD$O&Vw{X1_lDvy3Chu
zayYB@)2!kCA`8cE+%o_cZ|njZiYqPI+04pG(GLS?t_dZ*Dt7N3$UkwQcFYr<(1Pli
z`B0GHV;abPKpU`EdS3sOS1?WE`?r0pT>wG+2+=80_9q{N=DI2(f((Tm3uj38p9xZ7
zFGVb`Jp4Uy8w~a5aeu-9r31O-s@D7|WMqB8l+US$%t%2`-vMwC`Xp*WWQ;GJEb`!2
z-8`!<EOvc+HX&)2iB5LHTzf>x$AEO<BvwTBb7Z^B+MkB19!j*2-5vUxwzl@IRfz<Y
z&hixmy`F~1<(^NVhXG`o*-0-=YkzoHW_ee$Pf0=$<#m&6cZJChPKOZ9DGb&l^lGV(
z9t5`=iqf`mIpt>+<%S=KqveGd%JaG8<;=72wvG2J)8SSv{z(hWa$oxNB5wI%TN{k#
zA-sio`uS_wTqZSMl=2Jz${&0eSel&kk2Yr#G~$h)4eE>5LP4W9Q9U@1Oy{jI2XjuR
z1kIOLAu>QL(Z{b&IeyZW-?MT#J0vA*?bDn0y;uz4XUlC3=u=45??FugCY}EjqbSQa
z7Q?<RZXE29skynin|~JGg%{+`a{mn$(oheFz$TPdHhwq68*#f#+Q2!{wtnL(e6zPR
zUa-27sLF4u-5F7XUoC9|ke*P+NjuEe`43GJV?BH=B<$AfD^O?(9PY)f=vPd)7^)nk
zKePAfs4MNIaHtrt8P?e9mG08x94c6@AqbZZx-Q-|gf~r>7Vh&j9`<ycrwcwSSJnX`
zxU<AdC{i>|pJqEkrs0EXZDH;HL5W$XuUWt#)_I{%th;n9L)6PBLE&MyXRB$|GAf;e
z!~LgC`;)V+Wre|<)g+mr`owjpsy%IN>Z{o(aV0qX3=rJq*G$ex%*uc&KCdWCBarOp
z8jPiVZbGt2MVX2j0#cejiV^&Y7NdSjMbe?XZ(D@;c)=+oXFIBGLN|KRqQ76m{JMln
zX+Zgk+~=zrDy$Bxg@3pIu-%)Kz4%AwR<g;=T?B743U+XjPOj#II9*vqsK}`-OcoIp
zFn<<_WR676u!9)XYAG*G!=!)?9t8V9Bx`^mEJg5LG_C76IXK#(`LU?Rp0v$1cqY)%
zqwtyqOrA54yxfCer6Qwd$TO4AoY2*K<0uzbbly#>R7zE)idsNoVRME|5W?N>0_cfi
z9-)fdhTgCcbH973{EdlxpRhGu(xvc+?9u19I}OKF=1Pp1t$c_x7XbPs!Oi6Rf72~@
z?J)Jilhwuo1&?l)Qxn5^gbd+ux@e*c)vudGBr^2{IP^O9_4?8D_5uQGq&1~AXDlzh
zdZ=Y$MxnK<50=6Pdb+Gq((2Fch!HoKyG8b^c)M!0wY8Q1Yix@4f~?V|Lvwd(>#;r9
z2J+#K!)dUxwBauB_wPd=3}Mx~&|GR`-z>b@mTXxB4Ur13-f@GDn_Uv!3AGu*HsN0C
zlJ>Plq#Ma&<W04>lmD4)*gxwe@e@u?PJ-=4+%$A#TF3iw<&XTZq_$WkN!T}Awr0>n
zv#5xF(1o8DN|WQQeWNA4Wu7d94PBZ^QPTzCm+QRp)__j@C2tKCv%F}Z9XSSP=q6?h
zbIzHz5o|wJt%ye>v^7)r!fGBvf4?9O6d0-yE27rX{u3ub54OtVG_j-d^DKE9p>rd*
zwZB<59%WS@?G#yKN%)yuLa+6M?b3yp)rvQsYNiDL-cJN6E=C&X(kx^I;tv1)AyDRe
zffGheK=G^>z|NhUPw&WzhirowDxT;;&-6^sUSBxV<Uqtl#Heh+^BUR-ZImzF2PP!~
zE--}H?j&BQzwoP9u^JF@e7sT3kuG4!*3Ue2`b4<%#vUJvn%-y};1g-La-X(XY>WRv
z*<eh*hgfK9?A*O8P}_EChu+TNNF_?xO><}SmT(6fOxGqx0hUwPvGr(BAuYq_E<L^c
zbHVc=Y}>MSeM?bbhG$Svxx>W;en!YKzo+ZMoiv#bwT6-TF)hV($-1d2h<+bW3WnOi
z0vXH>xkPwr7$FE;dBT@K!qh*S{_IUBkpdPKWz2XuG;MwFURbK@lm%lyyr0}g8w~6{
zIo&(_i*LIV#YDY}-tMK<X87@bX)0|8@tDKeDGkS)nkf04s|5yydLp{F2ywTIo+-6s
zwQIVkNyT&a)scGDZi(tLS?Tj`xBPmk_LyW|FJn}m450NF^1SBg=TCosrK_%Hw_G2i
z@cFC*9;4MNk)I#eC?o0A9z8#iy@IV9AIn;cMA?rQ=Eb1KTO_9n&H3M&fF!24q1e5j
zHlJ2PJG;AqylO3xRuZ0YVL)VU?Bog}$BPAdQ$Ah9I?<%VrX7zb`&iw2;Ot!osi>ME
zht-GG1JjbA0gD;JPntWPFQF|naXOmdPmaQlM;d8twQXHM(b&PCHs)VWywJhoqH8K&
z`%y?UagoP{gU(xaw*Me++E7*jVfieAJt~;$shy3OF4`jRZroT_baDd>6<Z%o>s+qF
z)W7(TM`2>S{c5wbIpx#4Rb|)AaGQHkL@IEHqwB}md3vMDRHeB*Q92O)K1iu3<TK){
zEfs?JU1fOV(oKdN>+Tp!bHA{wgV}@kk}r!fR;wzewBUOR0<%Q2z;w~|RV+N~$z+df
zG275e*=$Zbd1gTdvFG1j)R~5Vr90#A=ZfK&f!|(*T{03}N2Lq!tEs3SpOOP;gNHyj
z7E{F@H~WsGdL2d%`n^myKVZul;rZC{+irQs*=x&VO%5Y-+R-Ws>K&Z!9v}Y=#**GV
z2*0`a?tEGzeHFvacM>}TC&Y;o+8vMSxm{v|@gI%{Ll`v1wd2NmB*Ljw0emdkp9Y7H
z<My%k4vof$YB#3^*;$1hJS{N;#<1;!9VU&e!k4-nGJLSF3u!3kjO0hP7Nq}Y!%MHs
zzR}E+d&%`FuoQh)Q7CH!qQUDU-Y^p=p!t3v^rGynO-#!rDBV-54~*4@S*%UxQemmJ
z|8@L$`+haF$GtucLN{q+OSQ|gQiR>yGE`9yR5`w<bNqwusc)+_KyfPn^yIG0;GdCx
z@r<cHFKBv!)JwjkK_ON-OCNW3I&P!!o|#x|!%RDYyt8+((b~`*-i-giCX_ewX0NyX
zK80V3Cf%$)TLDf4-XCvDrwdTV3I%5eV70D~E^_!_%bb0}mS2SLs7OD2MyIHq=+#k{
zjP@ezw@%k#U(C4x3z`|)h9WJ!zJw^PeuGX!GaMU=NlpkiH)qBTldcrPZwXKRd4GER
zU7NBn)>KHLG_+r)SJXoO{uv!W7i9mr(h#D)4ieI{@oR0B{aBJaH88;N(sL$^R!HbW
zt0R@|qy|o!@yfW5$ZHzQ@P4N*OufxocKmDqS9-YExqSeuaPw*=sKvi^4;0D)o~7RD
zGhdXQC$-x|T*@??A^&2}9s8ly@;C;45#GvL9u0-*A@m9pA=45=-&%fUTGZRL(?s11
z?Ox+6lCO>rs?!D3h-xGUi~j-x7cqEj*v<8ObRW7}e|GJIMy6;Xdb?<1>3pj0l**>x
z96oHW4Oj_9l+Co?5_-J7VBy({{&~YV<s;5HB|)-#hz&=BbSEm83!|_65R}C|OZ*;f
zAYLiEA?oN#);APncmJ8+CWaGTDU?_L2d%c1VQ~VN7HB6n{~t}~9?tat|Nl9}$Z3=)
zF>|PSJ0Eh$DHd`phm=E3kwZ>%K5k_(l2glB2gRbM6XwtmiWud%vY10GErwyvey`8(
zy1v)7%U}D$#q+g2ACLR}e!I`tyucyADjl7Ig4fbsMMeF!&oJ%r8)aJdZ{(_o<@|&_
zTHo4wOJ*JJQ4UetOZ~()_V+t`cb@s^ql<cY?s9aU$kpiF6Dp&;JH7Yyl#%+7ww-XV
zz5|&ruO%@qcS;U~#>!uU>pv>Dz7+Mfh_%=?H<!_q(K=FN)hT<G7s?}{ZiI#0R)SZ2
z!}s*;Hf&6UhS!QuY&3@1V$uCUkV_l^cJ8m#-h^`$I~ky!^u5QyZ*Bg8{O$6%B!uif
zb<Z!pI8N|Pw7g{iw^N*pnKZY8i~o!~0QCTMaNFv=7x~g%LStQWvgWwp8H(#TNcEKk
zuZkQkJum7Hx$v*H;cDol4UYxrPYTr82kUgt|8ddKqC_)6M8dp78+JCmIrH<c@<3=e
zBu`$!S1Jgojo(b*Lud)!22Vd*zT&RSbrUU47-ul3>^HQj8&l(q@zM=XPl9d?LF92k
zaV(<P`e1!~H5_%=Z*oX%Pu<${(f<Y;ZD`g9@o*lk${XMG5iv*CU3D+=$a}fD5GkJ)
z<0|}yU}R)cqGX}Y19OXy7s2B`JT1P?OcB`1xMn8>VvqAoy{9yWQ(<r*Jy&Id;Xj<G
z_nb@gwZCL-tx~n}-Hj63IvExaYfu3;#S~-Jf7y+>F-ZSGDC2d+GON_&cn|xFq9!IN
zTG&G<>ICXWzvlVb5k`N|jRnIHU4{YjK&pcaB<t^!ODb(tMO?yMcu&SlmSmk;4`xtX
zJYB;R9~WT|%5vAoMuVBJ3-i~IzL0KksGrOQ@K_@yEZAJn3wgmB+eT?19;`1>HSs`*
zXjhnBR3Au_bA<ahZFQ1c01UT>{84;D=|D7!_Ui7X!4%Ss@K>$78wMw9;q;)=hc6h6
z7%#SCwZ4PGEyjM}Cwlg-5G2j%9KIg6w!U98mO>kU*L)u=d!1tU;EpNGDwv>u>qKi?
zo8aAmQpF3tvqPArxgS=1AUEb+0Ssp`mSuWWQQ&?{ncQ-EI5i?pO4RD;5v`j%zd>J6
zac>FcA~V|Go*b-gFC&M>5B6yXtDDJDF&3SsdM|>6d?dIuOp3iCv-Nl{%dT|%I==&P
z+3~x7knwoxJVb58xvyKf`<G?k2PKgFr7AMu8Hm&O9K^GLrSC#N@T0K#H$tMCN#hxt
zv<UzHi_u1Dx#w=5tgptVCQMG;*hMwQxiCp@sK$0l00c(teaUn45YIQn4YnH18PHeQ
zMGv9xwnN`6{}9|<2nKUvI_6R$<9HF4St}7by29DrW{6G?c1nZ$tKGM0#gZCDj&9)*
z0?@$l2gQT*sBsT>YNOyUMkA?>R8pp{=5J8^8Ebx1yLK!t;HJPk4OsDYR6PeR9hZE4
zkjufip0l>Gj*>#Hj}x$h&*RQMj8y>H<i(q^t<Gwk!Q4sZ+W0f!EO-}bAKX;8%VJ4O
zlg&IR)TQmk-6C4*8{%XOBrAp(K-g>vU#_^vH68s}4Ils=Fo97g{4l!tJROFR_``9-
zeUd%NVKF#<gYieLBmqU(*hu68K@pWui9J^xLA21gt;_ifH+=+I3)B`!WJv&`7@sLp
zPgD+4^)FRqgD}*k@sQTihx%m>1)T$)9CAe_xiM*acK+x;?|YY)LP<B0n81I=AMlW6
zZ*8u1cXnp?+Lr4{VqQ2~XJS)mlcAGW?HvmxI{{AKTOKP^X@O}04&nZa@Mtv&^JzOe
z&hk<i)xLv6&-5c20b7t8{}Rdll@}yya2F;UiI6aAZ>ys$?Ii9VY%J~|Lqhybq32HJ
ze4ewGkmId&W#>rD2w`WyGi2O{A4ymRR^-oP@)5UN-NFgfjYc(kh;WR0W$~Hog&)Z$
z3hzdH5_M2?;Ce67%`OZKnKUXsQUbW;)1z&Z;-a9wGahTW`>^qtrLM#rQ%rCvvJXx#
z&y3LX#aI<vs+`M`lh=^YjKMs6df9)LRd7x8NPNhGg<Og3CDnUaY*mP>E^}eBTqUnG
zBWP5}J=cAMYK&HZ_<&WlUlUBzsGm6fZc5mbQU0ojdu%W8G=u5Irq&Yf2#qKUevRt>
z?$iFQlrtx153(|C{>&u%W!hVz8PwE`Cypnb=^#_RX6gp8q@%>NCR~QkA1tpe&f$A9
zcKZ`btl=d#_D+|wr|atYeJ~fm+Ir#;&xWy-y{7A;sZHfBT$dix#;GmD>Vfq}X2Rp6
zg{?>5N5`<<zWovVuscl&CHNL6%DEMB$O(N0)OuPk<zHBq<{Hh9QmraZD*SDKJ>EVD
z8uq6{`|5+HPzIZhCo(qZPfTe2p4OTA5Kb940iB~Ra~d;#0khbM`z>Qv{}Y3FL(ezY
zgrKiykkiyYA(nr%=70j0NaH)aI>fQ7M7NMr4Q3&`iAk*_QYg{Pj?r|X*1di_I3&qw
zY(WHWV{DiE#Mm%3+0fHo&V#!|{@ukq39JZAGyyGB)OR6e1?=h~Q>fCZYIscYLf}mK
zXW@r$Ffu4+lGa~t;|{gC-8C<-FD4QS#lw0pY~Y>Q8JR+XzyCH9Q&6d?rPZ2`;B@$H
zJ#h_sfnyFjg%&dTh8WTZgWo%q!y#na*9;Y@6q%ZsN*oBfRHYt8Bj#T=QL%~k_LeQ_
z!yzqR<WtYv?oB*G?@k?m9ctY{*ko*M?lVTHEEEHHtdiW5T=W8!Y=3tv-&K}d5r#Y)
zn=CHQ%Y;CFc?!xELnU%;H6<Rcn+8LhbjIA?P#A@eY4Sx+v48MUnN7bnAAMX8t1Xnz
zFn<-=@%a~tEzO}E)Jy0h5w9YB3O5qniIe2&gp=Q|9$Va<3f>rJY4ZAxs1}&{g4Sit
zP}d4xX#aQ%(vas95kL8x3#ajRcJ9=DMD2u!ASc8!OD$LIG#}U}f~E4YjBBZ(LAs%z
z(a=VS5%G}Y{H5*zz9%7z;iVXJ<a0SYl?*RNCnkS9T%(42p$;n>E;0j)F}z}$z9KXs
zS-TDyLdo-jo;!(>Wh3X2-IE6|QludTs2Ap8+_E+o^xjar7}Wlj8~H21A>AJ7X+M~!
zXSDVko(M>f=rjd62`)WL@eY$1lJ#xsN#C~~t4}Kvd0hREed~qKt63uBsyo%)ojoaw
zke@gF-Woif$h(5uUOwn7IQwApeDOusz{yDkdgc`fr$nH);+#adFmVw&FZk^=DqSl-
zqqU?IGV}C%cAk_eSwD9l^&;6}giH<&VEqfrJbR{Q28{&+1Mo!YPhj85SXzoD(bW0@
zyPbcjMGNQpIpye#$1Ee~ePO9;cN|XQ$lXMivBw0Md*;})acUUP?$jF!Aq;>|(AXT!
zw8jBvUhzCISF$@LrL`eN7?lQI<NCTVxBoA|g%IL}(WAHaWXvOH%hUa#;Jep3j;Uia
zB}CHI6C?2X@Of>Ak;ZTmBVJHk2yA>a5afqJgSv|yOOYpahk{BM5{%n&t7}8cT?8)6
zd)nXnAOTZlfpj^~NbbJ^!UDIS>*|h4L+Gqb2p-p-yzNU6JUNRRVZT{OY;Du<DYzLx
zA%=vMvTpt?H~jNfm<K2*gQSCNsv}!ads`Z3G9wsv?tDbUqLIlX`KkK{p-U?BuL<9_
zwdvO{FEf{`6Zo-oy6y)sh#E|rSPD%^A%!Kmu-uYfh0^{mEj7Mzrv?)$^2O-E6?q6z
zo6<bTnS_!O4U`9)`am}GnAsh4*wqk5t+{_TrdLm$0+X9r3^XjYV5CrW{JHprC)roa
zmR!_|BtcQ7A~lWU^<qQVA3h73c4q}ZENz8_DzXB7YCVy(-xJ<m6Ra8(o@5Wh&_`xH
zfl!7!dy04!vRYrhm;95W^NSwT$5c!ksyQK_tsaWH5Uu~jsgGAe$;{>Ai-ty`!AVB>
zMa6R?VZFBSGxnO_x21=YyxuU}QkE7eZzy!iVNs**Vu94xI&B)13*@53l7(qLG=KNX
zffc=f=rz<;dqy?uqUL`kl91)FumA&7j(pt>v(c|^!zb>9DtCGu=g0Q8vUjhZ3b5CW
zHje{TNu`8(Q_G*cPRP!Nr3^MeO#H|skVqdZZrrGfIRVE_$=n*6O5}2Fl^H?ZkITmI
zwL@lP+lGEjPI~rOZn@9@nHju|zN21dt>Pezj%X`qH?pS=|FTW?2fX%Y*d_;gVq%V<
z%_$FpjU`2$Z>(HXP7hNC^Mxl&yNX3E976?62Tbb>bs0WT3<XW-j4q8&gwkYkg&W3$
zO6wSQ2pDI0zQgv`h6???l(?ca$wT*#Xh3hbz$dxk@`B7uOm^gdLslClUnkEZ<35PN
z;<>r^TWRNfZz_o7JH-lv$ncTrbQN`M0lOlrrhbXRV9dvcDB-k<qLULN(BtRvz|mTf
z!1@M5{3M^~aR@||<V0x7YT<(_bD1WZ1y};bUFN(y#x`(<8tSg*)ngw|ZHd+x1Bi|4
z|5c*7s>xg``Zpdbxacyds)ReO&oY%ng>d<t5j7*L@$tzoJOn<LjH>gyj*V<IhPF04
zG-$HW;<92Ic2P!B;`)C#kUvq4p-fNSztXY_SHT*<LaI++&B=%yHIShCDAT4cih#XY
zT{+Gc57V!5wtmbolxaNodn8}|tiU7|5`?R3pWA8UHPU*+XkqRe4oqug{~O*ViqzFG
zrZ)A=LrA#oL{bYc$PQ*<xWJY(udIfz8alv*AqwAE%EN$m5?*!*MJKaXdmFIDawX08
zMGMTe7a46?^+AuVYdpJ^Uu?H+Qq=HX2YZb9mO6K89G}GfGoeWj0kdt5*t-Z*??HBl
z3La-84mo~GE*_!TQ$4Yn+PcI@JYDW${~DrSt_6DeD)*fpCrshdm9pnoOax6A-`yy?
zew(qm_12J-q^D|$!v9e;^ae_^F;ADDRl}<aSaV%-jgw)NhO)w%w~-GIwpnzWg!A^S
z-HaYw!F|;RMSH-etneVNL$<Ja+!_lw3#d`nwuYU5Y|9<{^aWo@PBOs2FH6=QI|(Yr
zM)hT8s6~&K|LWoZ#EAH>*GB;(MDucZI3tmf#xVy27ImFY@cxe?h^Ab@_?N+38;Lq2
z>cA~H<zRhrE!J!!E@OT*BA&f<8M!=cVr259jym;kuu1CntJWa6dijAMdhRg&;G<<h
z>hu0dOGTjZ7^Y=nB$c{1)AhY~ACQo`n5exOpX;OA7J1MU7q$5Uu0EQI<F_>9F2)<o
zDolr=+<x8q^5fhTaUk)Qu8`fd(`-Vm-9{rbiQ3-wk<0>K+Vq&Crww(mNUD9&#!Mu=
zoa@;u^$Q^rv7rNrIoaY8)jWi<(pL`zzBLt><RR=X!d|a!NlpRadVL+=yYMrtt;RUb
zsHix5Iaw~h{{q5zre1gvfiGqDlz{iws8SR^RF&n}hk(-^KaaN1lFYGO_tH4RV%d2d
z$Lqw$%kK9!HZT&FW<`2&6er;iZOn-=W#zX-4XxEin~+v)PIa8B<g?_t11><yuP*$e
zot&2d&4~qZTmC77--WCiu6jT}G)*R^6)cpqobLukWHO&gYc;H1$$nsjX{~7vMm5pP
zkpng3VMuD+Qe1n2WS7>0MDE{+MYHbemhj!w)_Y}-JPg%a<(hxSuflTDtj}QF94o46
zSvN94a6E7<LzBKHlvIPiPs!cQEY2+M{J1gRTCbR)fY@kDLb^hj&lP+`@}p`S$Rhv;
z)AAo6_&*su^XLo4iQt<~dX<POxfLad25ZLwi2Se#cN%{Cdt$i*mF}Rn8?(E!z3p|l
z^~Rk+onkAJgSIA|dBTSV?q=HD#c7_AI`vgnH!y{l!RKA71k>5*ZvZ}~8>bS(W0a0(
z@;pGF)wp{<p^0DM@n&%=4@~~V#cC+}>6FJV!4vAYvr;FuU@6ztCWdMXlrcGVs%3Wt
z3#~M;+x_u@f{fRzRf4ruObtSowk|sv*lKy>aE{hO8r8QNt|li-c=dFYwFZ?cAvIq_
z_|29dopQ($<S5A$<3jZ5%VjMFZVK;QgHBkvTO!b}IJl3CKVT5rKQ2)kHWqw10s{Xk
z%?u)<D(Z^*W8JIR<tV+p(WS<4AV*PWw~?F}LfyQK$2$h3bMC0y-KcI<AqdZPfOA0|
zPc6Ubo{5{Dys>72oI$<fP2qkSbdsYF;};lzt@?>^YA^iWE)%u0x;WDdephA&<rd~e
zi(LdYT{?k{+<Isu8DU{|R$rDZhu{c%e<h<=<WZzj(7TBN1S%$cX=#14OISv@$KTD`
zzzO~8y6LaR*tXY{jjQ-ln}TuKWcQG8%f+q>H9|b*JigL1Vq-Hk7b`ont$rjmJs5Ma
z-U~0x)zIdtH0AycBv*z$A@Af6aGGEC|4vlh&*IWvbPt*Amer#>+5mJ(dXRlNR<w$H
zxoe%9!=7Vq2^EiT{{$YF7aiMQZ&|i$B+$k^$3AJ~7uVh|h0Do`PIoXt+{s5_)IYB=
z&rW5`Nz@DG-Q*ViPgEbKF<)@v8X8>8{d_R0RMAQxGwo|vV~VhaYKf>TEG}<+fb92F
z>epnjO0j0PNb&AV(Gv?@^|l!a%+tS;sGr*+w>d2*exOEQGeemTgCgDwq|6}V5HyO{
zJ|Ir=RE{fOy8BJg;GQR5m)6sANk3_;F?_e}QF+K<t)|0>wu8TFuY_RB%bKL=nU%{S
z1I<!Gyk@~E!2OQwXy31CVkjMD&oSj@?!B<L8QbDv&j_^<K-6UJ&C|VtD;DNg)|V^d
zY1B4P)Va64y(1o$U)z?}VkZu_y$;rZN&N6cbRZ+un3RA?EarthXvhrC0CWHI)xUME
zI7;%rLdT=f`ZnfnVJGrZR7LFbe!~75@&;wVj$VFO6j_N&S4xMO?f#+Rap11!g6FR_
zEw=G$Jkox@GTd^HR!3MWh*x@qcDi=r(Nse#>NY78Q1^>Uk$jjIGES0jOQ#|9Ke*Q*
zzKv<Sg{%!QC{haR4!T}g9tCe0cZ*siH2@!ur*3KXYl{`smi|TOW5F=ZD<Mlrefrfr
z^&(Et`=FwnEJ^EWk;ge=2T)re6=AG=Kh@Zi65I9<#v%UOKH?k69Z`+bGXe&y6jaJ0
zX})~<^>{h?AF?GCXn0CGU()BmlzFKA>rLq3?&{@g|GEuzyUL-K?5&zso?l;!*mEV}
zP|SHD*;J~}>sEbk%|d5UKi?PKCMn5>>+u5YE<**_(&;W<ZSTEI+uyZ*le{y2DX*0f
z7(}fFxav1p_t9gPb5`6DVSDYH!Kg47w}${6W&>#?BfXGN^4BB2YFvlBIqpDbY`>Ql
z%J9EY?}hm7hztox4*)BQXG!G92Zv?ZmW#rW)-k2X!Juy<L++=pNZRw|yM=RUn}Gzp
zhulL*aryQbgv3mD9>17h5io!gfYLSHG}DE0$H&!<AtgY^_1AkAjnUnxct<mNE#?qB
zPze{+SEH`L&-uV;r|m|2+hc_1Xw|N_Gh5w>BhE}-=!NN8q_+wfu3=lHcfarDj94ny
zeGBfLz}GcDRVQ7<Ad!8Kq;23Vj8nqi%>GD598a9=NzkuNLfN(W3dMzg9Qgjv&LIR4
zWZAbaSbyF})`854wz`KJab$GOTth|?8;O^r*p19bPCkkf;HpsS#9XT|EBR&2jn(0c
z7h8CpJ2bUApo>&zq_4a!%x++`J@?Lvr={*SBq2s$xloMa299`q3$P2nd6T4i1Y(@J
z)x<OcE{#I^V+A($KPOw)0ph|xQ=6}r5%-}1b^e;dF0G-&P!z4FFNGRDtiV}xQLys*
ziHG1f4FR?`B~zzmHVH<kgnc^dApD8(lhp0)`Hk(9{KGp-irv2<!9w1Cw9_jp`<7Bx
zzIQx}qiU8R!JO!xs`uwPf4P~^)*0KINP}>qpzLsDqEi2y&*SP+KF^CCVn_Vc5djD@
zEqwW}AtP>&Nt7Ey+}i3G2bS8fzsD+4!?R`<Ba%G|%`f(`kR5IJ%c4Jk2jP!~-B=O!
z2!s<s?xI%Wg1BroPk!)XrPLKX9~9A2o8DSmyJM)P@4xt<t)%GUaSrg;NlO7#DgFY!
zw<^0NB79&nKOb|qL#|t8nZXAVyBIh*aQ#75(pg7S6c04p>Yw%_!{#2~(eGyg4!!x;
zO8R(5<k#bdJ1w5$R0?GzhE%)Wc!M~?*jZj_Wm~T2D$slIGNTgcL5jQBoh!MdxW$>e
z{CiMv#=F!H;2vh~@xzQamyh2(_eJ)}sqXCze}nFu)pM9Ek>bmzm#JDVkLpqwyNXIU
zF#{e{(*_60`@RI!M~ch;)c*5wBXaU!;GoZSfBp{<=bwDj|DR$UGBD!o?xZDirGI35
z4G<Ytxx|o^RMM|HN@!B)zTP9}kNs_vExaOC<?vhfw&eczI1iCbr)5&ZI2N3z-09pn
z-U8OxZ7iY1m8=s+7(O$qO7D8c!;)xB7V5Zt2M+Hhd$ua5RG&7p$k_>ZR4vz%%TpH}
zdK!o;b-G|x>pcSi2<zzk$cZreWPbYlc;#OmpKLz~f{oG_LHX$qI=Irw*S|O_`U=SN
zB<gp!po{OlG%$t!6hU#r0RPKO1i9QJx_yZ>(5$X$rOWPhG30z!^QAG=&!S8eC2T=u
z^oMvY%=wf6u^PH8Ov?$On1jRPzNUdPod$L21;s-Guon#@RED-pVON*-w~RS1)0f(N
z)#7r90JAS1HkWL{u~Ct&Q{%KOOJNY$wwJ75^;-!2PD+^CNZSN>r!Fz2)m^I^bbFkm
zQq|S!-x}#-><$i7=pAcbFiKfr{;qiNWEY2zo=;0(V>rem*u^zuBGLTmeB>Y*e#KhZ
z_uIGd!<EDRIr_IE3Wd73VbL3=damu~<&9Mz*-v=`<{&ZNyq8?ptP9-Dv#{Mdb_qNq
zQ#U7plXMXKGl{ueC{Yqwll&Skmc8B`(JO0E>C~N<E_>I!xD`|5hSCxXGWX)MCB&rY
z(~Oy{uC;@%k-3qXNkIDPxfjK0*2gC{z;X=MI8Ip(gI7V)YV;jgQp+_rWu_z-Q)t$z
zEKSJ6tS|uQ0HicS2x(_~FiQMs3bS2N(|@e&G1^(DOtsfoGwHO5q~}Ha3y5BDw(Lxp
zY)(>!;xm5lJP=Ah(hTW$#i4w#Xk&XD&}VAcJTpLzU}izq4b`GI;dmk|CcL%gpX>ct
zxe%?Mkr6rB;|pER&SPW6Z~0;_MnzMS5K-AI%gS+`FQ)g}!qqoCi^{Ujo2}=+<N>40
zH6ZSJWfLoN7k=dic8%5BgabFpfLdEHc13WzlWfY3Lt%Ld6?tGiKS$>-Y<W=T?*+w&
zLD55POu*{wL=XWwP#*&6qYtE>?e}eM6;GsPF4}e@#z%rkAKf?cj{J#;qNxE{e=EM?
zvv?fRGk{a-POrw57dUU_T-d1jK$pku`cFo{)yJNUh6oAORr?NCC1C(43tFqHyucC~
zrG#2kC0x|jLfjQ7Z&FV{zq(?&;L+s9o@$MD)evKa(Tn;Mf9@z=;7Z})!hj&t{L|AA
zIV*t+IA3qVkczEyYcS&N)A=hc{ibZ69KfZ3|EGsGUS0r0L~Q>B?}oOP*TpDd^YF)V
zI*y6Y1}_i0Tm+dSEZ=n5HT;AcsWfOhq2Q9f<<m!sAM)~Mqm+?-i*3!42UhLW+Ph$<
zuOl-e)4QqPjAVWLf|4Q4I7dxPPpp#@W7Ts%pPBTXSiP%CbPrM1A2A#Q)FbGG-LoZ_
z#cdXOr2bdRC91S^z~<K0Wov!uToM!humoFH8vmo<sDl0DYJamFdA+MumHZs2<`*^M
zywo^X$jCBwcfsUI_><I>?ViJ(9@JI_%FWf?)#wSkOS7mT&#EVm-}uMrG<1wtj%@C`
zGh*zi!71VwrnIn43aH)Yb=!^2B1T*6lePL$L;qyAzt4~tHc|#ga7ZDmKRg^hbucO0
za*!04nM>>XRNPpBwUXA_Q+NpqQZmC}&CmC{*DRudY2$z6Z);B%@>+{cUDc@ORbUQT
zYOI4RKrY$#WXy%8cvsgPQ*2qHCX&3jMu4}nnSY%TJ3w>AOoY5|9F0vHccYhpV6iSK
z*Tf+|*(tw6d76`=dmqSWUj9Gfe?Reb7S=7q*IM6>^#T><dEEDVw7Q0l4_-jKB0TYW
zD;#4kYiz5p(NNtI79L*Y@T5I;XZ`vMVWH7KVxp1E#OBv;)bKIX^ol-@mb{lST-IWE
zUsBbI!kKFe@FD`+J;Xg+bmqC*JlEpifY{t~(|Ueegt`c3T_vYTSXF?WD9~z!K79N_
z`l8lY_#c!mT6H)vPJ>(Bj6Y77I|5u?V3BC(<7(si$+V0=OH9-8Lb>uyPK_TfZ>YfH
zbKkdJH6z3EZx`47(%Qt17*fj4-s|8De+z-JgPp|Glu~2E!er1U2yFja#IB*$zs^Ie
z>IID2cCR_DsOANdND&44!MdnZoam^^Pbd61zo3#Nhkc|5yeJK${AfLKue9OCe<H8~
zq^bfPcQO6!=i-Du|B?ms3V7b#`fVVv!eUVqe>PI5+|O`{#Ay_%=be`xD2GYh`3|Yz
zb^b5C!m3Ah@Q+=VrEbww*T)yP#YKv3TKr_8)?d-ep6(1-btNS2!-6bSY1w#+Da@3q
zMpYF$gA1Ak!<Wtqq*=Pka$7)e1|bU`+K%XY89>dNHd4e~LXBjd;JBmsV|OY8JWJ0)
zjg5`Q+$^%rQ}`u%0|Vq}o;I1|#72676MWcRY;~nH^~0zg;x+M<hD?qy$JCB0rR*Y<
z^}nNHocga4Hc@IjlHE-TX@XuE%q&JV5&(3el#c?Q+F0G)|55!GQDlh`8ah0wg61-$
zB!bk(-1@c0JchewutGYwPK+!Btk8k<Jsq{vzZ8%Z6**m=QFEMUA{N7g<@Ei1XhEDx
z`Dcw4<88u<RFS*>d*bXYMO3sKRTi0_Q8$vVj_WZf@(_)BEZ3397+(H&9c7Qf+?g>C
zJXc#k$fpSV;{E;Y0?({(rzaO${~aXtVEf{jH0_&%48RgFl&Y2lyHg$1)Rf$;OR8&@
zT1!3N+=_eEU@VnOY5+D9R81FNlL}f3c0C77VlI0FJjLq(U=Zw+Y?;KwTnjk@f@Pso
z!n>|NvXQJ$K2St52y&<y=f5L!a~p|Qk5_Bk`5!-6LpDz(GU~+n5=!v+IixYbY;<_i
z_VIPddCzgxE~k9!8rxng;>PB7KP!Q>GzXTfyqU_5*aO@OUi<5-%!7qDsfQHQ?s6Be
zg{rhmbRA55va)PybN~H~LAq;@@p~b2&|MCM_Jl1hW(3u99Q8fvgic(Ms{5cq)XA+A
z3!g%w!B-yUcYvXNT^kDv0n3np3w|KpUT$0TIEUaTr-~x~tF~WaqU!62gqp18Bxt14
zaiInAcVUcq<R&`ExZ2r5T@mWW#Iu@ho-kH^uo&BWi8t&iRZzRzR^?bndyi_j|Dx8g
zKt`vwwyCrQbp-M7+_+hJ)F%T?3%MBc%mD$YsF+Xjo%>BcCx6-?$W^#q4jHq)1kB1r
zqrz;96!^QLI_PxZie1AKdq$!g+hyv88l4!^+VF0JjCd(z&?u->7f^`hhvZ&6JZme9
zzz8+V<#FXcyfn-GQvB>}QuyT4^M+D%G8$C(ciA)uMNTy``*L_QehSiJ^{}p*ums%6
z6A#x{qQSiK9Ua+l1qpU}s~}?y%T;_Y;>;<QzojyKrRA|p{e8uNz9UiRKvx>T7L~ZO
zRc~eczI-~_jms0`6j;3qR{*#2PlXEc<s~sallHF8LePYUZjWkfjJUD!{h-H%K?cT#
z>#7V?;`s{I%%fU1MAa|9tHia$PX7tIvgA3j8eblaF{e>OX@f-}&<J1OHR!B|5`1%N
zCy|uAw~|^yC&M8`_tx9g#pK=oZ{JYB<sp_;J#(i<-c*7gEM4#33H``=Ct{SY-2>8>
z<w0P?ELzx)p`4qlbZRUB&I;zx|BTyiLXFO8YWN^7&rVG=h7+lwCAUfy`y@HZFy*|h
zoZp<c?%tUV5a50Q)^hc8x~b?Z)I}`nkMWeD71CnnafJ!h+;w5F?Bb}3kVlQLuzXP1
zEk$X;2yRYW<*;!2!%`p6Q!2U$mTeT{D%$}$2;MawGjMRq*@{awxY8GNU0mig3aBP&
z%^SDe&B>I2W5sS|$p_>6Q>o53f7j@1wC*k@Z?A7})4kfUW{lx6ER3?Vvf{0j5)M>S
z0Jl~rL3Cs#$#xzuGj({70qp$OyLN#^j<E2^$M3z~N175&I7JDHP$*6uliynY>(I53
z=l)#mE-*D7H~IDO(|=Is7jY;^Lkd%WRpi(Mu#Se`EC?tyIoVU{GhM4>$2&Rc(fJ7?
zkzVNUZKN*J(TpotGh{7XE1ezRnQ!b53q}Ne#L)RZ+q$%ByBuFcW4KO9K3z@<uWbo0
zkE+#aZ#z9TD9jj_p*uz?B`7IG4&rQJSrH?7yYKGEf)s>c?a$Nli&2_GZ$YynxOv0!
zs!zuxmg8lPC0thZQ^_UIeK`eoL`z8iMR@4>DZ1Q$sy@0fytMw_7?qlkVjNo3NPEGA
zdG_!_HS;A~o?BjvTk5J2gtJb+OK^Rcq?evD7tRCH@V)dExop>+9Nv<tBYe|%utw9g
zDeK6|$o<0EqgP}UN<b(}{$CedfDPj~+Y`Vl=neX{b!2|&e$?|maFop4)VK$VI+}gq
zd<=G~v?jsS-q4tS{;SkoQM@CFM<vtJ{GOAUb??tf!@Ix61R)IPfzvKrd3s#u_LrUp
zK!%cX9rfrzCHe5s<W_>_(||CATdzj?l7JRQb{MrWH|;ew_C5;BmYMQWqmRewsBBMR
zNRQR<{qyf{2={OS`H*Fh`t*M*d-JXgl@XeWk>PLulA<FOUS69fhm4{psDzY+X28K@
zX6qqp884uy2Z0{BE41S`y<&O(=MPN;f=RgfT0y0sj3BUA;uBi;cDDUPm4x8)aC}W+
zO`%)XrPi8a6g`wmiSbl(Mf6p1zWaOmB3F?n*-{*hRj9gBd7x-|gj?$3Q43^RC?j@g
zYV7nRx<*RojIZ&aWt_qnHzd&JJ1Nv(5S4?aruubz9OZyUiv0}N_3bN{GQ>M0{RaWx
z!Bp5T<3A47eqd`hpPZVQTuxV@1pOdj_2vy_ave|(4%AT6<W(J&2KCw45TWmYx`9=v
zJN73}?1X*@%^m-xEKxIF8Stdt1IRw%+xONMSw$3QI#}JQvzWe}A%jXjSes8wN=irx
zfr($hG{^@SDv#vK&<?k{GU6PBt)A8fT%|iAmGq08$AQ%}w|5C0-}QR8HgrUFxtDtm
z)%(clq<0OoWwD*1d1Y}J0rTdS@5XF})3u5d-i8E!byCT^+`2ghOt09VfuvUuM0KX+
z1H%r%3h-%ysGMJIgKfX&l9~S_LQF%qDFUa*`>l(b4za4OO+0n>-32cG4(y*w&zn^C
zydF>_ndFLBaM;%S)D>wF5W`fOP)<jAD#4(APirk-zT71(UBYC7d25g4i)K7sO!=oQ
zJxR)x9nyA)YwkhMstf)5a?N7EU#Uh@UFdCN+R<=Ag_m0U-uy^^5s^iW@jCD4YyIaR
zENu2KJX=l_)&xC!=i)SU(np;Gq}P;tMG!P~_#eVzp+M~JdTQH54DED;lZ`%V?4zx|
zET$&VROz?Imx^QPVA;y$lhbJRUVdVX3CXiMp|of*qBy5#eihh@7|PG$^zqLM{J1Ye
z7VYX8pdnz^<revy1&GweX&pST6r@Brn*rHGnKuq3b$LF>7z(MDZ!N32sQu`Q7Zbp}
zSS!ps-tJ^_m!&x5NUu)rqT)sj9h?*@6Kb6X_+<T+6z;wH!viuoSTQc*?{IMNVEhg%
zkD%5q+_2tXnc?yluymwRV}J5Y!^t9Ve>+wy)iO3FlAc%(8b)@8l7bk-#lH)bS9{xZ
zwZm_RyAI#HTr&_sRzaM;-IdVUHivT{B0%dD>WEB!K6CRaHztpT0@fKTGm|6EnxGp*
z)tm%3$LJGBK)~cql@xRqG5n>P`f$qb5t7eWqy6<Mh&tXQ7@GV12<Od(0#RXSpQf6u
z=jQf?8gd-zyhlo$X#Y0)KPzL`a@!9h_3cTYoULVfHs<PYRyy`?MV9WqjH7#a9EUj)
z{@yu;6UF}q)f$0OQ3TVQwZ12J)8R3IO6eBz6@m5+>zQAi++{b2cSNi-GAoD4VWAGq
z&=0S4f`m>@uD>4v-Wf#lB*s$t)cZa*Y%cb*=yVEeqncyt2qbE@r)O?;28>n1k5Gt2
zR;>AgK0ZiEIAePfnC%x=<}&aJN+QpKhuvsI#1A&AwI*koSF1+RK)O)IIMsO!4D{Hn
zit8dxr5gUy_-~893k_i!jQ`U=fe_WwmZ^IFBgs^twC#-a-uug+!0LR9<7JzqG<V<f
zvlPT5o?~YPNX#r%i``}OXK99!E?>IIGl74Fu-(nJmKO4kX@rgQZ^2S`MLI#2pU51W
z<Ka|PTiE&ORu8;pesQa|&P9s_L?-0jt8+_9AzVi1aO<`Lci3Djkqb*<X(0nrA^?b^
zgU^ZE)V0;=iJoBP%LiW?^tq+0<$Qj|{)jX30#wV$mR?l~Yo0+bN|~&xIjT-;R^c1s
zlPK2^z1V`^X7+t)Z3_g|YP7+7m$j~Qt97H6D_c(d?#n9?^&?&$W5iTA$-k6*S6IN=
zJ(;3@?mO4$^RkBVfk%?95#k{OYZH22Vk7TIGv?OTm$zC?_AnRpUA%3}gW(Eiv_c5~
zLPLhNqn~cd!Nv?81Hs1*fpYA3%RONYk<;GC6h)eCoq*h@BOd;!e#xCVU`V!*eF8#t
z{3>q3S!cE%O<PG3-pPla>BL>@lDLy`eTvZF&a7}UFqX~;C;hfXz)0*#@_^KkoS{Np
z)zeR*?VhR>TomGWxA{HmJDGY{G@W^FRCC#haokIy?)Fng=xP+yL6O(Nd~W;4wxwDP
zXKs+jpR?FMX!IYn__8!M4H!_?exQtzrzGxRy80_JuZ@MZ{+m5%1<&QJx5WMl?EynE
z>Y;>&lv~wt4!Ne6j9Th(PMpXrY2S0w{C9+YtEU@Ei-*NQ<>u8jGG_jGn~HhGgm{jw
zFc^(e`0`o2j5N(eg{(JbZmm3#5qpR0-j9Kdrc4e{<S}H(V7V5wS)bz(4%MeE$OwqH
zbsEs`x+W|OgV1rv*=~+2biNX0!_EOaxz}+{tjJ^5*N>D=hmKH+SR2WY0Fq@=5z8N-
ztt1(FjuSHV6V>2F2K&o#Ccg{3rfzsmP1Wr#h_JLz3$x~YQVmp@GDUI1EFaZB<L)+O
zm}er>woBz6hr+FD@^m&zowByt#VhfcflHzgHpblm-y+s^U)&WG*iOHi$xQFNa7!{I
z>Pn<8Ok2^F=bNfg{qz4`VzfhA>s5Z?(3ODh#2tn~EOU%L)BTjf-uUSGCdoCLT&zAO
zb?)Yu(`SKnnHX&(OnaOe_tBlZV<qWA{OzQ6k8nDJK#f2Au<HPQY0%x|1DzKDg-zWQ
z266IZ%aQh05e^N-yLxR!90(b2eqGCHM;q&|COe#CAN<y-EeT;4V%ReTny|s5E+zl8
zr&C+baU*qpvZ?+@9y+`dM+u@1pJhC-UAjzdsVPmmOvpwb%ekyIp!2N(;Tq)?PV9U9
zNqV6?#NtKc{n|K<D@&E0uV3lvX!)7TGJ^?{$Wv_}YJRgEzlEO=%UdQ<GjbjD-<=ST
zbP>zvP|7ydan<54w!Q}ziCm1QSe|L2wt;(cht9USe9zA-hQO<AM^Q$cUnT}E={X>E
z`Z4g+_MchCFkn*^qGDud{EymlLv^=JOia_FDfmh7J(0!93gmT>-hnXUSn~RHHf!El
zv>lb`);H{K^2FYOIYkMw)`wJHI)N!bCMHLFz7BazFU^OGe#k;?BX`0NxB5*EC?=?c
zwtu=uGThY9V8F+@P?b<KU*2_zVArQ0c0T#kmQGp(Sl)Vh%Kb!s#EA{h(8=A<BHYni
z+{DO#a^fxL5b8^SBCqd-Hy|Y)YN*%vF-y_CDic|dW5(+gsi#5#q9Pq=5Jn_S(XI<7
z(VQ)vh+zZcUxjKY1I77;w!gejJKU#zroEWcQ0o`?qlOB9Gve6qtWjnAI(YaLU9<k^
z?o#!+K@F$DM~0>tIM7pJ6Fz{z2bBg7z)RLwd&n6IXB$%#7VvS*>aJvldpx)3N+Tfz
zi$mn*I>Z<tjLkk!V^aVL&7M-;w^#qG5uonw?Ibo&ZIUvZ<$U0VYV*RV{aqkKi8`2*
z@p3?&iF>lJI1{nAvc7WQqdz!*xmjI(t4$ZA3~#OpqU1E-%EcyoxU9gCtH^_`gB1A4
z%+|{CzA%7!Y%@fK+U1`$&8YmN+X73k;j*6bShjs7vE*SU66fT~Mm+oKV%HL6e&X%&
z@N1@O<wTO6*wgzBR>w;8{RJBqKJtZ?dD){VBSB1U8z^#C69Ro7!zaRV#y%msQGcGW
z5gIPHcz%2X<k!YQo%&WprEVjsR8`F&+sv~X=`l-{o#YYd>?^=mU3V8-HR{iUxT|4t
zm7ydp4LQe?hUnlH>GoK-i>(DDNY#eX0%X{*ikXpHM&q4-ZFRd8I_n#N=wCykf~kL4
zW-u~_`$lolYny>d!SAxiJjf@XMx8__%6khiY1b?vdIcj^rWlT=ToILbC0K?<nsH;L
z!(Ey&_d8;#J5v)!l+&)dAr<Y9EfX53Q%OWZ14-CUhbQkETEA9I;ulv^q+Cn+II+lv
zx*y#}LKz)?TOoFfJe^=avr<N}ejgKmK3$s~uO}s{r4Xisb5vCr`zswB7Jw_x69uW~
zs*Hk0T7jM~hTpz8d3jmkgfEY&{WM5!ep{x!%|W5*@i&~F29av;@gZWp%fY0&{9ujc
z)kq3=KM|>=)@*G2$!d4<+}_M&?!wN_j*D}LZCTj+bShwZN=!7A5<+$F#TskOcFp8j
zd%j`#Q>#KZ)W40`a#L748;nJY7wur3W&-qU+_%2DQ2tJ_H&evpY`3`_tOH~bc_(99
zXIdz{Wj9MhOU`#7q2x?loUb>}>;<K;%H{NFY)um8?t|u!At0G2Q+6>ED!*Iqm2DcE
z7+&ENyCk7A<SYu|DnM~@i$Fzu`fSBXtuJdz^+iLJ?%51xBT7_s?%B%Mq7H!XrYBwp
zj9lgcAayT?A8cl5&iK33-h(+Ch>1CsR8N(@tgV&h%$Hq)O?DZ=yf5j`kGjh(+`$W@
zNyNrb4A*Zd&(EiRI}S1)iOEG#Vndc1i9RQwVEfOq6_f)8s;D3D;H4Js3a<6<YkmyC
z8~^_e37{Q)`Szl4eJehmJrhZ#(2MA5UQf!_6H0|M=HrX;wtVGwmj#`yuSw_j;QEl)
z7Tgxxg!_kqhfJAexbC?8o&DgxU!Oi6_`4#Fq=1bW4Oo88&o7NlzE(=s9b}qB%bKc|
zIC0)ZIa=4;bCj<C%rU{o@X=hLpL(rW7IKsb)*X3w{$$@z)mcbNkCCOn#B;Fs6|04R
zC&Bh$(vof@*Pr1WmMn0^3?;cU$Fr}|6A}*(_U98x1jbvTT>Rd?u!Xn1=`TTh#%3%$
zyGjchkHM2k>=@eSxZeD(<QbS|kz3|N%MR3`x4g`#UQF8Vo=gc`HW4I7FW2)&{2OR2
z!!`%6FVjDyrnDt*CnqK~hfzGsL|3MZ6e<yUU{NkPZ^X-D$tH-wGqIC$D=$xDMb;#4
zKbUQ3V(DeteYN#aHZ1+|^0ur3uhNlGDG^{e8lsL{u9}fs(z&=45dpCLwz^l?`>6DU
z^*S`TcRALi8p}aD!VTuR|7SQ;jZx?2aKiqdqRGM5wWCqK5(lbrN6an0;IMa!k@nmX
za#wV)*4OX~$ANE%%99X6Os>QyaM*wpPv*-)gTHsubd9U&ppKgCj_<(wf_8?tXx!=d
z<)z(NJ@SZQ5fD1y=bjb)f!IlyM4j!NWPL$Mb0|2$r8+XYG+@!RApyu=?`Zpx_hi7e
zdpWxF7Wc>I3?;Ggf)Evv$kqmjCpQ2z|2F_RfYhH<f2&w38jw{ZuXt=OhFM25e8rU+
zLz$Uurg|SNvDFcG%0-jVP3})Qu3>UyvFFX2?sEl5lkF@q-3pXW1?DU&ELf?sWv);(
zEy&I)FT{R%8O!rqLlpc;dS+TYlu@~SA?D;Cevm0=pOq;NeEwZ*O49Xjbn4>Al+^96
z9+}*34Wk#ozkXtM<Un4Pn#Xl-N65<yyPBTd=7K$aM_9!TdxZS@*~S)-5>>EA2g+gJ
zTAsWy6n=uMz%Raj83vH(HXXdCjDht)(s_rNFuS(Yl>N26wKZf*U0E}iukYicL_O7p
zg>PQQCMG8AMv6yHREnXpaFo&L$D5^Kr7m3j<LaOr)dpj3K*GmHpDcEvNl8wz8l)b?
zp<(ey<O))$S-Q|)Nk`!vQI!yKNXUAJWf>-Iy&R>J%O$7n*L6^<1wh4xo>}qN>T}9z
zVD*;1;PKC;fxz9ck^5AS&uo_mXFlhCn~4O39Py!zZ5Q2oN6Np19^n}EVP!@6<XFf;
zmm$Gvksu3$JxHXZFAt|U^%upM5d2HNJE`-?%J=1bcj@$l*;nBVd>rLf8**Nwew-FD
z5)#F7D)T&5R7QbW@f;!?qOiQ!P`gbMdwF-6Vb|W~f!g2LDDrA+OEHR^j(q$?s0Th@
zDmM4_!pS>++}Y!K*jKP?*2P*k6)T_S=bQ6(gQU*XnSZiGK<a{Lq?dhV7g~!aH<z!B
zaeU_(_(JwAPrX+3!_CY6jgtz3cEe98So|1=xCSh0;SOqx@L)gujS(Umg9Z>C?hcfY
ziuc!DgHXG~_N4f0>N5l#T@JMAG}$(xXsV&CrX>?z?B-n;WiKZI2B9&E3(~ePgR$0D
zeufYsc1`PtoTR~&5<bI~^TfBb6K}YM*G#k9m=3wZ!0iCnKRz*@RF^S4SwRl{Gp<_1
zS^~;E8&vz>u}?Xc(wl$(3I~9#TqDKEt1d7@DdG0RE*a+-PYM8`4ZZ)?zmDbX=o^$s
zy8S=VZB7QYzq6rKSrXKztw5gcDFyOF7gb_|CdTra(F}=?8htfQ&GBh(!6$#KnXQAf
z)rZ2K1x&uU6xZ(R<3>?6z~*9#&-$o-`ier^+P2h|=)I=vXxU}Htu8I##hm4aDXd0B
ztV1&*MBY(bGI{WZ1ILHDE`G*}MxJg=u)I;dy_C4Wx4}5vMD5WIm)Z}vP4?;SJ8D1z
z)u7ELrcOV}+QV8E&P;mvf@5oWt|!9ei9-jbTq1%@Y{8a}IWn{xLe0$<Yz0Ivpb+Iz
z7-3{#bHb)0axkIfmXS8KtL4=9pvgDBUT^-_gTexoB*x+*ACcfW4lGisi|?}}K`**v
zlzITCNLJ0>q65l+RKdTxJHN6V5nqn?R~2^B59sr@8OZQY)sqtH$%u57UvL|CUc)^E
z%;-JYAtCOZPY<6o>MlK+>qQM~e*;=!AdzY(E<JJ@IXsLTi1v4>me1;c@OhFvcuq$7
z<?W|;E<gg0aHq)1=Y5o3UZ8(uvOsP(LGN%aU{P^dpX7f)Wg;!NC)e}K;gT?CxjlV*
zBO_Zf)Zd3)PwqZ@4|-I~-l3+nVcgEr?8Utj;oN8E(qm#mxMY>^yPYWIm=aO1;>#JA
zs6F7*DX4z|iT`q0i}iJM<I!LA<l!&iQ*=UGIe?6$dNkXL!iL54`K>bN8yP9g{gmTh
z_--D7$y@l!&pH;yaQO)GcDog2zkPPj(+~u@W<4{FEe_`vHzrZT(ybuB7ebl4&rO!+
zR=*QN4)<aXmx$r*DAdm8e7_Ev_T(Fbv6eV77(%e!O8(3@B9sT_<M65VyQ%lu<Aq_0
zz6Mtm2oyCozbsx_t2O=5XQf^leI)k2pOpWP_f5D0^xvPm#b5?jqD!VYT=z!s1w~;?
z+tki0$n?>rlJ~Y|P-~EBY16@XY_p19X@E=<*M9q9V5n>8A?kqH?g8|%1Z}Tx*sQOB
zZTx(pEtAZp9|O_X86H8=8!_iU;P7Q9@)7@H<37}=^RdxrvVQdG=L6RRFu6ltB<>xR
zpUuVmnXz1abnj&r&aKc*O^u1Qk%=YAbnUOLg_%Te4dt{s=XV(2uIY3HYjow3CYR5P
z@eRGN>&!-oqYh$|lIls+histT&-idvQqy{7|3^<Cc)f`T9P7i?TvYPTK0BPCz{v*`
zc6Tb(i(fB1a?Qfjsp+(E4Gs(W5YUS@#7inWT4B)l8bih4!9$iu<!zdnJ#8k=yL%cU
z0L~n3nQkD)dox(T)$C@st(Efi&(S|FGj1esd0<i460HxXEoIV<zQGel8%{reV6S|X
zBa`gkr%d6Cf{0&$f^B^?UlC&<Ku^!47|K$4>HFa>jdr*`5_{3}L_&g*Snp!=YBslX
z?$3wi%DxOw><Reujh_?IY{#r)kLRxi+~Dhx@Ds4O2_YwYyb-&!2~5!S{gG0@l2)u}
zY8nNMr9=lW<NeinDs^d7@7I1HX=BLcj=Egq$WH$^8NS&iYQx5UWb4t@#reGv;89G_
zf0OJMq;gr}TwzR&lWA6-?KO<1M^ch^CoAt(&)hl240ZnEUtd$3>(?;%w4T?+xCdi6
ziZf=tv|2d2`}=Lu?}apJ9<{j;(A6Cw0z<v7I>y#+NYx(GYuOxc`>(c(Q#(>Vwtrlu
z^C;K-QJva3*gAPGy;bluLFY6nh}re!*O}%dZp<0Xso_7P?*G~1V}1s2yr^-)1kIaT
zfrOcU#Xvo+K;ylac`F|r+bitJ$_=)((9dU{O5BC|{`LSeH(~S6)>mA5!Q}EAu|(9+
z`Jh`$7{ivvuAGdN{p3Bs9{@l@_P%+w|6*FkC&#C3ZzTe+34Z7LyH@3QywwSzEbh}K
z4)d9C0mTCNc?T=#Z^@^<R`>q|2TOjAlGXY@>wuNK=)*7`K19z-kA|BJ?0rRL;SsE1
zrJ&Q3B@sxq<5A>ThtGv>RR8<BSHu|u4;@%1EX5*z;rc1$Ic?FiFqOxN{bX){;%AC$
zm`3gOm|$o2drk;eNM(-;pLT1?eYSfLNcwuA>S}%#n4R(sA)i%4_1Ax5pVkm}nsCa9
zWAI!(acUy053yM17p-$#6?405TJ1ku9~GNNEd)lQ@mMGe(@1kexjuOP_c2)B)RDq+
zJo_;`M5Ri=#q9CvNK4I*J5A%Tiy+5IBi3rXh=!?s#9pHcKsTmd4(VB4B)(}s*qQGy
zat}Ev>tL?RZJhSsb&0Azw|++(xMqkl&m}VY7@wAn&e0FBdgZF|dsWc)mTnfh?UOPj
zVYEC?Kq8QKmHZ+2Q0pTE6_6RkFEwf*!||zu?`5GUVS_`u3fEu_sooycuuS9b%oiz^
z$aMcyG_sl4oJ^WB-ro!r0B}oel8N!_|H{>lYZk@88*F>}$sHvnRi*DYuVCLdy^@2<
z`YVbm`n$R~_1=EodS0xpEy?SeCM@q(GSNyUh6DF-IHq?s^l*J1wUbLb*xDbV`!PX(
zAUpe8X&5W*XCt(>)KtT|=elMF0Ge=d4QWuJt55hy=p2D|b9xl~7zp^R@vC!ER#q7X
zD#-hlV%!C~g}+-;6397FezCCus4>sA3@5PZu(iYWTPHm=wQmbtRP=#cb?4zef%&BP
z783r2MmM0JMMk3-bMbBayl$0eLXc4sCyR4z9UETrcwk46cc;c0a*d<2=`u$2zn@kY
zeUW3A6;GRIjl%Gu`YnuKm!Tb_jiQDmSf)^U@B7zP&~HHkmpt=Inw`akEaPYUKyZzM
zl8z@~hQgMeRpJQjiqs{{;~ImH!e)>gZ^C<)3^p@R`z@%$)t(JDtu19A`2fpmOpVu1
zNj4(gu5vu~{%hx<(1~Q}=b3qY6&=dtiVCr<8eZ2jv7Ba+$ZyY51x7IPLRw^R1Vs9=
zkZ(}6&+geH#rlxw;Cj)36a$cssS(+{_#z@p#0T-R)ea*nFKVavxBW79E2Wc%FUm_K
zL7Oy{vw>nQ0qf7zwdMN9{4U476ga)$UqQq+HS?$!-Nm2-!@6#Jj$i8a>jrucx!x1y
z#1Phm<4}4D?N{#HpZ&{DrV{74KMx$D>VTR_miemnwZE{=s_|TtLr!HjNiZeKGOB%2
zIzaY)*sau*!!UR2#P;&5#MS?Gs$PdI)>n|{P(O10Oa}3x?nRE*tSe0Vsrv@B^OT0i
zgZLPCqhwS)m4G@x?(faVw#RXo_y}-knB>}Cf&XWItn&ZSbe@51{$C$XV$>=Us;C)i
z)uvW$shOfwYu4Vi_Y4|Dt9I>CwMwm0VpIH#id94_w%Qe1Bep017ms-Bh2+kCf6w`x
zb6ri}*S{)!(By-D362T-&6~{=%=^@A_T6Jss<)~iZtq!f5;PV}Tf|k&X4N@tn4MBn
zfx!Io4LlMFRb`f!izxg3SXZ!D8d2HWh%rUS#?FNQjAX0X{cuqBuMIW5n!*Z}m~A$G
zyTg)NXwpo`TF;`ZyYLA=|97%}n8)Gq=ZVD)Zt0~&x@);sF@2la>F=iU_So}jEf#@|
z!|J>g4vxai$Rli6JhCi<AXc~k@}-n|ScM0UGi+a@eSMqPG2)7t$A4W)MtBjYmd?km
ziK8l)@1rYT;?KsX;uA4!EK7@Bc)}xw>!a(<*^v>)1#DA>u4Teohf+SO_un73Xb$)T
zx3<R1P*r&n#3KusI-5}JB%0HI8u2VY-Fk>(|E1{7^~ks~gBIf3E#O#%aa<e3LDAqy
zJ6jFV^o%#5wAM2>N(3~D_z-|#<8h*g<fPpG@2T|Jgv2ykN@<pYwyxlxoGE+T+9knK
z!G66OTtmj=S5;QM!rlV!Y#XQi7iAjz<_{=sk^QZOIlox688g~EE57H*5aAf!zFTg#
zf}Xicf(fLmIc1C^Njbcq$|*z4@7hYr>hY&}2aT+4ar9>(E`7R}5dPAaCjs4&VT$-L
zSanl)u+q)t^*?;aD>C^4ZVGV_bb?2T9>N@*pu6HP8T*S5Bt)^o5%Eb6WAeLW_H)R+
z@rDO5$r$=fBS7koP=dxWvvN3`vc~M^G?pQxx1NouuNRt7gmBB3!-v!%@b%)|dfS^;
zk;*R7U4G}z{{~kOE|Klc28fQUy|$Y$Rc;8MAUvjVa1$@g`?&nCvu6p|wN~{FBW$QX
zJonDW<GxXMtH;@I^A9rUt0F!rpDgt48Uc#l$8ff#AzWCfS?7`#=j%NyvR<ePIhxJV
z%R)tY`JA5ezhY(3tDL*oBp!199TKmGCB*6WkGZWZiSr4BG6rKco;A{3^Eh~k&?H(B
zuz*vh<Y*FZg7p4vIlfxdqTBdqrI1P=9TQN=aNzNai>N-Y6r*d}5NC^zwE(~d{=8TY
zM?WN7ZDD#HA9B_lM|e^8^e6ToR2N7*Rfi`HSnhrbAr1Z*qN+eI35+8trXY#R-HWH@
zcA7SC4%kK%qAxVq=i$r-xaOKWl+z|8(!ftq{8j2sf|kgX9*Mi(=;|uGW~Z*caC#_m
zTm`xhf;-ol#fMd8;@td<H3fQ?^(h|{q4o}fccdzeQ&9Y=A!f3NhEn$w5ZlBXi<1K8
zHV~;N5@02h{0HArZCyt|T%K>Em%0;IP&84>4-p3A@ju`-M(%?+H2uesb<47@Fv4}m
zD?LF=9gT}lD}v=tsB*#jrElmBzHj$zp|Z>KE&cog9}yeA5x3`(jdJrjiYB!4eQ&_F
zVJf^Un<&6A!SxEdI$^kq%kv8_L`ni=C>jc(q};foPj#_i%KQ)XVf*QwR+_=ii^WbB
zM54rGvgs}>aB#@Uo0+!n)k_Y0s|(+`=cSk}g|~lmSm-I!Z%nb~cX<u>wA|-By0mDU
zpfr@RX3WkrD<Rif>Z@n=M#tDAx#|G1dlE`oOXFlbAjzkO?oUkID@`1f#ZporF9e0(
z=dch1DfJe}ouK&X#lfpZqST8sQt=f6v4yUi4gC5&{Q4B{yQthzXh1z;Y-MjeytUyb
zPV*29Vhgv0>TXaT3@IP@d=*0FVCX#y1$Sc-qPB9v_LC({64WX2t_FeJ3h#Lj4C$h5
zDL3$ViQ<WLeHe5oO`FNqzzplR#tH^V$IF}v3-tI$9vpGQT`4fK{D~AM2sXsI3^l!5
zR>MpziTblS<_Q06Y(TE#YhB{TdiiL9II#CCp>N@d(i9K`Mm>e?G%UC2zQ>+KATo^P
zOR$nllG>;7wQHPjU+9~U9+aZp8@U9vO_3E(;3yO+jV9-}U~}D-29e#7GbprdI&1bt
zDLbT(0i+7ce))22*F2%8&j9U6tg^d^tF-YUmwR6ObzVePqtDg9vu2ezT)wa3=P*<c
ztd|U+#1|^&u;|U=;{Z#H3KojIajZqb->M=#X(Q$dCqzgh507i8i?>_)?f+O1%5ia?
zBMr}NbKVK<E9^FK5gBF~Mhssn7uo~I12j}qRg;C~ce%-tiC>w*ZyMOIOq96Se&UYz
zf?KNtxT%-FGr3{u{RAJ^k6km8K>r$jZxgOLKJY#m;<W{X1p~|$lysFmDxm*@iE@&J
zgMa@ZfE|6OzP=yE+l29@>oeT{Bu(d9yClw}c9J1QUCH@DG`6BfW1CVAOqETwmc&6g
zc%$J%-n1s{)b4uUU{==Qh+|k#Roc#)w$g^|D=gb#Nbj4F>m3cpFAcU1ZTEO<jABzI
z1lBsR0hLFyiJkyox4WYlrG%nE8UhTE3N;c>y<IB&%q`7-JmPdR;wR!+vcv5$*u$-)
z+-fdc?syS&V90F<%#Qk~pY-x>6sI@{BZ--tsLiBAF;i}=tRxyDsko;uuYMEUGn{F)
z9u6E7%TG7)`ixEL5anGtEO_Ms=Fa%XSRuHuBUI=xWaO=FBLx|9RbRpK-Vysr*rF#k
zaA|)pGGc4;!OD540>WB9>oD+_TbZDKQj~TN>^l+dl`LZ!K+jy=1g-b3T<ya|sL+a8
zvgm=?&)G%f`G5RQbyEfi<j~8f&-<J>a*OnU*1`ZX;V<>Gm-C1JeziC64^kXEym^~G
zbfT!L-}lYWNQ3gPbk>>_hR#4WgpYQ#Zi%C)d!CqQ?p`oZBZ9Lln<KjSFT@R?f)6CY
zlmN$=N^Tol9rYlkBn^`%MYLj2u&ru#r=qF@lVHGtK3E?fu?2`dJybaJiimX4EIJ9l
z{?8uqr)Hc~+Q}xHRwN+ja2rcbF+=Q@;@XI(OKWL{gpHJklEf0&QILqj^QbAaabg79
zouIDdPD&6!h+b#c-pyH2cf{FcAREtG*zNOn^S9+0>@K~c8v?IW?}DY4Mvau5^Z59E
zC;>fE(*=)nLmID6eroMI(E<{sT{FhhufkEz0}RC^a)+Byw8H?R&+kQM8RlNZN%3Nk
zlc-7d_T-?%yd;nD);83@?l|24Ds+jiu_*iQq9MWo9|g6ec^pStoERilp{m{=p`y9F
z)MX`>MJ6!Pr~7^n<8;ydoICM%FESx9TifJ~woe!-07S!1?PaucM62Jt|NOyYJi0wj
zM?`FYagT5%7y9C`)G2s(u&!uHIifZ4+E<TILGrtDKMxkkg*4wK4lLc^n}R6yw1K}>
zX%OnM(=d`sOX5iW%f-%)E)|ak3KQK3#q%}Rs}%ECJ&3vTQu0jrC;kbHr|5_s)5b<2
zu9pIavMrfD?x;bSbU=DWpA9mT_=0|U3+-;KFw+-lcnb2~DRY^SSX~qbV0>uyYojuN
ztiOAVBZK!v(QEdNZJx0|)kx%1yVl&fh+$(r>|jDBE=>FmFOZiGrR_w|XUj->F`FIj
z4cRZe%zQQ?VX5a!PPu7G1%$V6<}+5dxxPgPnCV(kd`G6en+*6==?H=H3^V&EUbbAw
zsopcMSX<;EuK9*}4_q9ED%Xf*O|>!F!cE|QbTNrK`%g%<4O9kvzQ(P@sJ>-Hd~`Gu
zR*P@O_OfI(6l5MHP)iydQ1Gh%*1jh*{#X0p>Wae`)Hdq7dbY6(hJqYnagToSZnv|7
zfV_~=riCd=MIBa_v#WV=do=(hNL!jyKrIUOM>}PU*1AhM^5$r7YSU|9C4YLzDt`Na
z)S9LA(%OnP$&If8PhnL7_r4<%%CuWf7L)BcBR+-j)js&*?Qk?1#)-*D!8CYQP<ndQ
z#Euh6Ia?ADuUPajyRKltLix7blX|<(cY#Q)H`=N1vdS{c4%U5HYnJMkB5(F)kN-hb
zDWxet1H8YM5K8EXc&{?9jae_R95dJs@%!LET)K-}8`;eH&ay3F8HJFutiUJ(6^-~y
zTnigSAjQwjM?e7kg9&1rJD<f&J6m!A+V&{ae|JOq-Vw1U^Z0j2iI#f47}X)Agmbp|
zY+*Wd1OWSrs*ar%GNOWWgWlVI+flo8X%5=mpy|;Ame$rTBzY2Q`(B$-O&SG@#D3d!
zQ(@@k`SxJt;;>nC1R->}om>~TOCX|j)`g+C$B|dZp{+wB?kk54L?c&-dx&+Z<=AG)
zJLNxnk#7@)j6SDKG}Mk56$J;UR$`CfzcNiyP;t3hd@&kFL+JSHtJ#YyR6`-+e7o-|
zbl1bdLH|3qQMof8*Y(j#-$jsElkR}zk3NBDs6>>0meg()!fm(B$LvXylILAZ02yj1
z?>FjMV3dD7K%j5}$Gnpoh4E9eKzby$zE&`2h=CHt5&NA4&i0)fToI|9x!)ArG_#4(
zw){gPbVWH1q{F53Z8(AU6-vpLlNPx#A+qI~PL}KsxP|K1h@R+D3N>rKeLC@qr|o<?
z8}L!X|KVH2!8Bzq9}}GFmWf|QuK<m@V7_Dm`}pfk7!5UPu9H`JZk5{hw#?lJlz(dY
zKRi~2q#emX66+_R90>{VGAiRxgLeQaW)eJI=i>R5LN$;k(djA@z4r9mV&|*<m&AMf
z35L19;~RrMb0ts{Qr`n_e3JW0@mzJN=|DJ$>xTmZlsnkpM54}VFZp?NMQ8di>REQk
zVK%=?%CGStt)cA~*>qjV%Il`$FPwv=Zys%sQXQsqGi*q-&I&`ejrrv6IT$y<QW}ZO
zLhV<Dea18Rpqr~H9O7!TA}ecsE#6qrW+mG+L<h1bGhn~a+&B&WFRr<EfnB)S!;z2g
zq;d}NoM&}4;NJ$H*ZhKyuTfFnnS&_uGD7WS>FP3Ds(IQxt1M$x;Z7&MM1z@!%O1=m
z7<F_W`U_j;z1SV;)AY-HqL_H)gomj02UvNQSjcJdVOkio0s{Q(%+U;hV>}v7Bv+y@
zJm~WFC!$hVD1p&AI#KCH2^_8lR_m9a?GcpK|IL=UA=~ujKt^#D{4|HKg8ka;C`zR1
zo1w2|qhe>JYg}LnwckPfOX<&mQ3kO8(%btny=h3Wvan4H!yndUTLphxNvDU&NR6vW
zO#%%v2F0>Pt4#fSoaQbA2ofs3_WZl9*n^eX-L<`wV9R}I?>}0u1iQ+VvFw;(7y<;q
zb3=j4232`^w*{L-o3M^8+`RYYp**P$J-WRH%MYdAn8K!7*w2r#CmR}lmYd{)w)%JZ
zpN*KgSj;AKF4noTUx|-XsHe46Z;rS=5_vWe^<y&KVwY~>dGXpuvp9yWRIK1T=j#6O
z7TMRHsk{R3C7`_lE{}eAbR~6L4I;MeY4YZp5l!aT$q30&=>>`OqB$TxfG>u|?-o0)
zA||f2&B7^-H{pI;S?DJc<%l-<d+G_yIg~$1x|Plwgob_;u*H^Uy6KTqgwjJtFMcM5
zXz{_INTHT6atSrTma=CVofjzntVcjfmT3R2s}p7w4|7V(g4Hu2Sn)<52$2FARtzI}
z|M&aIdA6BK-B)g_UgBfGBiha-+}<JwdM)@UGSTOVEc{m~eK+FQ%druo5dF640w)VW
zsmsbNt?!k~x{ZkwL!A$R4#)$`ALUhA^Z@!|n>c4q1!3XI=%UxaqzZX)2oI(2^JQlO
z4UjULPZe09ew&?p-ojlk715hdnn)2P9QzYxNvg#C*Y&?qQcO?lLm(X9FF>Ow7|W8-
zc79phM(p~obra)#I!=A{GyK9pEIPoxinS*-+LV8dnF1=+Yf=ID#Hbl(u3P=!3wHx!
zNCUM}9IMn_zSP{wXf>G2qtVuBYt1@RQ#V255U$n7(tTR|QP6*Eh*l}*RbsbzB*f2S
zkI!f)u#MtvUa~}KhIv+i#Yq7)?eR{-s|w?qw`7=q8K^|3i;@VUlZ*H}pXA)syZK4p
zN7<Dun&|bMR5HI8VCH6)O$JCYqN=fAv3=sV^M{{75Cq}jV-WIdKBiE6Gm#IuagVyC
z4FgD-DWVUjSWpNCJyLoO;|JQXGAtKsg#U{v!?8hB$>*f#$5r)JE<F-|m^ZY?r{`fu
z8t_wF^^c5#iLmA8Q^+0S?qKzCgQ&#$Pr{>N*LKFrO2wpU^0_G?uKL_xS6v1AWmJ@I
zdiVM~GR!xHRQjM0U`{KeZ9gE7Hv*cO7QLL=cd@v}+=`IJC11BJNdbzM*t31#Cz_7@
zBkp+<J0hu(L>sOJdDiz4*Y=LEU;4XyKAfypJRh}av0L|OIh0yip3<OF{oVme`#5X%
z7Ut^L{RJODp~zF_!K*PS%Lg~271ueGt%lxC*;e!B>cLsO@=DS+X@aXOSUXCa9=Hs)
z{}f&MA3~fR&kF1pATxb70|HSx!DOLQzIN0=h*|Vvpp}6TzqoM6@|*Si*jZDy8QZoD
zdl+2VTFy~DO)dBFbFMM-)@P~f|IW^%IF#@8j~BD<Frz;=uQ@X6NNU{$IT1PIn!9mE
zQ9nt}JqNZ)zQ)m0rc$eZbXDYN<~;v-u^0I#46(%N(ao8R&O2745UU_=)wq8<H2*BF
zE^SEn`XE0Yo9G>%-FN@Rc=5PheAtmTXeco@B%AI~hA~GmokGaAjbOV2RuxR_hn8jd
z>op$i`>T*1p+de^sEofvNnr+1(Z9EAcFjEmqkFaG9@YvAL82M>=>SnvPfq7^c*iOj
z5gLxI)lF{zfcvWCP&THElACK>))J!CQvW#Iym_km*$-Ai1Y=~NL$<4(lmG+J@>oI>
ztlXWS<+uWMWQIN=&k755;H63!YM8!1&(b2=P^;&(b@~NAl>E-~4eE2-cC)8@O@{bG
ztY4Ls`CDbkfy#}2M}Biqma(n8*XE=qVpamfmw-s?cdvnlE7a^Sb|WGpw1^UALN_{a
z+I-e=hJ&z6{N`OJ8+2d&ZBeo0R-Bre2lzEJq}~U&Us#^N)QMFdQ|&wkV7uT@n)L?v
zoL%wo%!-H4X>7j?h2kQ0$bG;W^Xq~f5bPNYOc97Ef4h?NhmRf#@|WuU4o5WFL-5h1
z!+i%X*Xx0aYaa}KD-)X-Q@^V_eml~hGV6c(J=N$9EEvRi#?eMaVo)xewrMP4i1WBO
zIzgSd%%X{u$<HlO16LlxjuG-Ya_tt`n)7-hWtmfFuq-LwqZNE{6@lnp)fMDaiR6Xt
zTE(j=cdkSkW}1)V&{uB15Hlf5$!kBO#fvEKUFzQkMFj`_x*oha);pIw!QwLL_}V`I
z+N`r4*uM%ywD9v-H#gLNrLa$ARtC^LQH%F~NbNxFrg<T8UJoCdOs4=iSg=U>)2j!c
z$Y8>;Po(mJr^6AaddkdkD=Bzoo=!hj@r8g=r*qQxF^{EZz=bGFM3#vz3|3k>$6UU!
zAjN<JQPAJs^6qf=jH{Sxe;hob*tt`m{kd#y3w?35E31b_MJ2*N{^s=|cz<|PW@v>a
za3Xrm*OgDeslQh`35V+2TReEL*>~gXmQY0&u1O^-kz;?Eaa`m)XPB2R@8hH8q$r}E
zg=<pS0$s9wHMxlJOvdeXshph1sYcj_D%EgWV-LT-<+WFz!~CWPQt#Ba-VO-D(DWp-
zZItY8Eff_qQE}0+ZAX}L&Y4KarM`LHUeO*3qJ)rqp~&pX{ouMae)Iwa1sQ})Sk{h7
z(A)9>oq9XA8DP0*hGIql*DfIm>MuBEj`5}6rO4%J-KPjBmopanzr^Ha?8IpKQ&I{x
zNgj^m6C&>KrRd1W?g{?Ny^>6|@6Ms?a>7BNwCv|5(K45P^zIDmrS(hUdEo(OBO#me
z$2DK4_r7*YHloDG`)VY3xIdjSU&TW*Rpi>ZK~w+%=yuY0rvSIzs&MOl$^GODIgUML
zC7QmtV+2-27<_$tbRjGretO{Z$iv8O|NP`6GY2no@65KHZhn7|I1qX@!fA~wAsHe(
z`x*W&7q6WisAI#F=@^(orDOXo3I-C@ZAuA!szDuJ$^7`^nc~1^-a2mwrsa<!SC8K<
zYDA3abw?>68knG6$nhj4Co_i4zNMYae_Z03F(O$;+)rfP*L=KUs@c;t^GLd9D5;n1
zaD2qi_X|Usq~#!7T7s<jPN^>J-c7RgOMQ!oBSQ~@n6)wP%yASc<dJ&i4_SGS%=w^g
zwqu9yff~iU3dYakzHkC`mK+6gimOcCh_^q2@HfcgNAc|(3o}X^PZ<lfDs)WwFnb-H
z2uosm(7MENVbA$Q#A#T#4(;W%<}x^`U@Y%(uZ9;D`@sle8Y(RXf9q{BtNH{lIyedS
zt$gUu%RuGF8W($8AT8C?scIvW5)MDqIwqO@n;bDgj$9|G%0Q{IVC0aDbH-0jRojN_
z<&9Ah3I(edj-{Pw`z20^{oC8A(rQ(RQXQq>sE$`z0cI#ysUikSgBBGI%it{lY7rV9
z=&;ekh3T<3FN2X<zZ^4@@thQS+VT$ng^Pl(Ny|%5KH&pDCA6Z7=m?7yDSicelQXH7
z%sjsRJMk*GYSSkXXWvc4u;6MqZ(|mTglUgonQas=G{1!_lwth+_23x?!vzQswKH$z
zk6s3@oD>E?YBxI5>4hha(ww&4Z5u&QG)s=5X^I}S_*ZvViJ#XEe0p?$o6xq^xbats
zfpeJwnJYg%$6S`7J(Rd&n`$hhaC4hfUV_>n-2$Q-?Q8x$#H!hx!cLB&*KbgFIko1U
z7;zPSsfbt{EsU!qySn9d@F5ZfU~+^iC#h+Fu9Lu)i><!G(0lfke&mXEO#x;bqj!3;
zzB`p;Q9`N|H>TIix3px9D7;yb&d$FU+-v-xi44Mh9quaTUIYOci>t<HC^Y2&msO{=
z-|ySnfw@@??aWac%!5J-BNBc3_xAn{?jz<GIqb73_zaZ`pmL2*5QUX15@gjzSaUN9
zmKNU?Bmh-$FEdsI!Uk9kv%6IYtj<@0XaB;UDaeWyU(Nz<6+QapeBs?}T1n5O$?c$N
z>j2%e8Ws8A{--aORjtr1LVh;6F6*%Qy*c~+>4&}kj4LKB;#5+SB%_)-`Zoen9&fow
zqn<+Vzv@L?1QQz}h;>}&&EeD@F)I4|$%zvRNS|NzLR@&u-vtZCG)+Ux2wH#-;wB;U
zcr@-dm);6|Vlp~}4?^)kDol<mbh!kFb`NXksHwa-tXpOAAOmQ}(8mOciDey9O3qEr
z`6L!cBjGGQ-TO67*xahR0E<|`7X$QfdJQRTdgq_S?@S`T-b>`VJt={#%#=KSMM_(2
ztHGV4M~bX;ebEU29gtUp>ZUQy{~$8etNG%IHeazLS=w*d1}2B%K>67dsq}2O8XZMh
zMfmTLUbqw1WGa6A7pH`;POadb5px5vUDx!pHNX&jJjEPhd&JMAsbx?UpDs=^BOU9(
zz$bZcq`?aDrsqvH9>3rr`klL;GjM)Tt=DFF3tp8;<_g|0H-27CDK9)%k1ZD(_HSwX
zTyg6wmh&PKF`sk=T-huu#QyW0!%E)$u)>H}cw}9I2B;^|j10-cwfLOkkU2KBWFSk1
zH`*d)keEKp)!Z}fSs3I&21rm_y;qo$!o^r1)em7Qd$^aAsw!osCEdM-SlmqK;?5Be
zfEvwMya3-xz_{?tZ9M<~Q31CpLf+KdhsLsrsD2vEP&TMVyfCZ5q%x63m7Y{0mBDzV
zI+s{x?atzeSW7he7MHdy_@pAhrFu71!Q*l8MnTWQ*yd#W&Ov^Qj6-Idu%4AaBDkE|
zxV-gqJGK`*H%HRLkG-zTog`V_UBCR9d?n}UHCw{2<=#oKj=Wy?Z6%EwA963HVwis1
zAK<dj(I(ndtGDpnE`)6qO8=>8g(*iulXDFKC+&r4hhCiT5sxj_=-s{uEVyEwE5<I<
z9%|2;L1~noX<kZ2?R&?}K|1Jeve`tC!Hjp&9dPKm4q($OZ~%nVrr1me&a)zf-QSTa
zoo_AhJ&^l%eEjeFYe#K(E=Okmy)%04XaNk)Xw2P_l=rteMQ?nbpgI)XLhF%Ke=w}s
zpe>@_--TWCm279PA%@m>-{!xPWIw?NghiMwbv`;|Yephh!5b-PoDbqCXkW~Tncl-C
z%{f>wmIBHK7?|)>4)0Vsx<)Wdow)C4r%Bx9m2V}B-S(ZM8oBcM6Y<qk!z9z_O~&JS
zKl4`j=&#-y&i2_@2xSZ^JKM$BZuT^=eKC|q3u+6E`Z1$TO27H}<*pXn;N-)!IdOI+
z;A0`F)j#4_juN303$MN|dL}RMyd7%K%5qc!N6{VeKKvO!-sWC25(q*Bv~YS2PNuk6
zpv$lUj|dVc*ol04H>(z5%bhk4v3-@U$jh_!J_B5v_^K4g7B-+-FQPU7sv;TljGFyI
z`JCbyUQh8*yf0+&dvjkem$17?)5lwJ1dP_bPfr_A?fV|!YIrY-(%i~joQ}XnqJwv{
z=S1ToRfrM*FSHgh73<tgKa~3_7E`x*yrkUKrBb&UOF9L&Dv<DnzUz|&a`l?fNA)N!
zWMrpO2<YB{B-jT#Xi^xTDfjb{^8SE5`ue$Vv1&r7gUKnGRO7Pt#o2X)CLpSHbd<{`
ze^dNp>$@wCT%P=?IcIT$QO=qyD8&a3R@>B@5jR>X0CQ}x2%?2bG=|K^v2@ekBXmir
z2#EXD2X73X$7P8To#!_i9<M6Mz(7HE&skZ3ToLhY_`2pT8`Y6N-sMrX!}5MhxWR)M
zV`n<b6NR|g+Z=?zKgn1N!H+VbUa5kLB|i~nNgVSRUzQeYnIwQIyA2r$)hs-lf?>IJ
zz3r@g!nkt;{^}7gpnpibF~DrG^JdtY^|=WtxHzL1{#aMx`g(l)=5*ZpX6@$e;wrG^
zR!VR=BI~XSt|sR+JUh$|!f;o#$=jBWEwjdb_wUbR7dO`HO?sy}<;{2ZfPlxJC|;Dy
zj}AGjP+cF(5H9ci*SeWA_+ZEeX&zi-^p>lKdG6c=Vb%W2;gjTOqsnt|>MuB4b`bOU
zQ~q<?>R#A5w!o3bq|wJ*h=EG&bm#!atYKUa02!8Q^qP;HeNDPp!_9hmB+jlKuMH9s
zC0{eR{tR4KXOi)lXNvr~k}Cu0ZF)y+!e_RL-k-Xl7+OiloWHZkarL)eLXvIMJ}rd=
zvf)OJ_()YSYmVfx`-In0Ip5kLj|Nc>w-s_5wkP9rwri^rbYkvIDNO(*efO{TimmZ$
z*m=<BQ}YLqDe+2faG%HtyBWPJ5R`Uit_crFoqC@7ljjC1FQ*!Z??mIex)+jApUN0@
z2ZVE?odg#Sn1QBlR<kC_E2)!VD&~1{4Tp$rUT9WUkk&mQGdOFcOxWEFn>9;NxxsZu
z5Vt|6){SBXr6|CXcDY1rD(`GrVe@h5Rbb~LF0%Bos(DOk6<Iooz>s)GlH8}BUiARF
zDU0sp5|IwC^@Zo-I3s3C27AyuFd&;b_1U3M;_hPCOU_xXurM|+LT6=OyB0mt5J9z0
zoZoJF#VNL1`9ZaGlbO{MkZPL~M~^wwi-Tpe_Ko(P%5&wgQ+|pic@w<a(d{g`V#JF*
zp7~a})ZT2Hf<eU&VZf@$E<$D+;MCw!uBYhm5pV8c0CK;SHR^1TRF#ez@HVlSDS;GU
zF4YHPl=4)x4+f^Sh<m+f0=}_9z=u##t>1G~@fd)OI=tWCWdToU=3KM$fxgU)(FVJ^
zb+S`LkzUw)?NKCGa_7Gq1R(j*Qc^TOo?V^GBw@{WBNAl_AJ^XrF+-P$s`!uIFqfNZ
zso)8Nd%9)Xf@KsSNNW4>Vv0Nf6aE}!qE&Nz<a>=qT;x@93`Tao#33ffvSy2vuP^tA
z&7a^;0bp~cU(HRK=DIUZXf)32=JE<{t#T{_A*JlS9IxkDHg=JxHl{Ju6Kvv=dY_B5
z+S@vj8;Jb5R_t+3xomwUd)X+Otews_EFJK=a639mYWvQOQ7Rum(O5FwBOZYHI(_c@
zE4ydGf|JmRuHCW4Bc#R;!v%w7%nC0(9IAU84#E+Xv+`hRbn!yKgit`C_dEpva#2rl
zq$<$>rxU^ugS;(fVx2%7lZ-<4u(<T$lrz7VHB^*vNrbdi@Hn*-jF`VAUx7F8h(34<
zE4{^Q_^<Na<>gr<+YULHo^}ObAPj?$7K`<6PW2);{NfQ##EA?|S)rEy%y(CbwouIZ
zpL0zJ5$To3N?uGG*50JDm}DU9!HW8=CTtk9VU>vasJ3d8X~a}%#{5{0D<{Ot!3tUW
z+d8P|RyNln?z#n?n!3?!VoFU7jmkSTd=+O&{Z@E42Hn0RHz(nr9kScULoEMquGa=-
z$N`_W*Ac!io2MEIi66j-fy&L%`MD9&`6_<pZ21=4iE%|fh5^hgPA4_X2Heisu*uZd
zN7^|3x)=Q5+c9kSU(362#jSHJ^PKtM2v%%U2*yI6DYcmL(p|*pncsfk!bKK(5UtgY
zziK0_93mM_mY<z9WCm^S^@W6)>3)L~vHBV_2;m=6G~C&~CB1U#OAZ@3N#omZNPhRz
z-L)+>0);zmK~quS&qHTlX+MC2kK)0(6+R8|n6CB20_Z_0#r@OjR%QN^@mt>rk1pr2
z8;w85Xf~vNyBVv(nQieLIohy<9F{^-0aY^BJ7%O0v@pEn-j`xlbIz$D&<|i9qjt`C
z#Ev$sHoy!^1z53lD{s@{{T<1sKn`j^(F32r1Dt-tg6+mU+;<~_OU*t#CMpXudjH)_
zo_H$x-Z#r!?8^BiADO^)_`_s@brb+Pzq#RRA}x-s{9eUN#CS)FIyH-h)~{8AkBk`k
z=qS3x8EDQWWE(pP=1W3T{$2UwD!iqL8wpZ%+~*v~)LE!+GLc5_uD|WCoRtwxK~WLo
z_WECOZ+MsG*sPv=%}LknSQ?K@NHvLgr)YBR@$s>*!j<F`vV0EQ`E?*JQ$&%LfIJib
z@3!TC^&Rg%o#(GBY!anj-yhp=4MZ&9E+VY3a|%alrS}YjVVdE-BE5ZDp0Df>tfo8Q
zn#v?DI}IeWc<){MVRn)}F0K?xz10w**^%`uYh|%y+}dALY}E#>7P~kf6;zQ?nbxK6
z!?*qz9r2Fj51(;^X?lXLfkFLT-h0Q-eq^M(B46j!S={$^a&Dgdo=3%C{3xqP*$Qz%
zG}z9HB|Dy-Zl6q57}SZ)=DXw~epo`l%L2J9JTwy*&s8EqLs|=KiwvTio;;l95$=f*
zKl?X9MEZN}x)H(ngWjuA{;yRhPB*gPiK86>8v1FIP42VhZTliV3+Tmm9?7Vu!8)<@
z)8{W=r_i5$`P$trg1891!CedpSEOx7KUMb0`Ik}PU1K#H5cOp<!N8ddk(t4(L0>HX
zI!H6^pO(BmiJeKr*XvK_9;h1c7kk*vi9~qS<m7vF<O>_#qXQz&!6SN$zSG3+TD}Z%
zc35PwPhm*#Afh>Sb1JfSLqu8Tn*+pj%JY#r4*n?0pl5V+52@)gq<wU=nKO9o5(S_1
z{Y-z<B7d~e2DH(qr6m<d&?Z;8aBx(4=l;vc22ngk8E0lVSW04=j<H{x7IOosIoY*U
zRaYy2&+eYBhboUUfU8axn@l|LJ1ZP(t3{;$MU!yHF%|lzacOQmEy=EIH(*F-Ff;uS
zqPS4QEFMF4yLV7&7}Fj)tYehM`)UlP!z(OoCl_(GRm*Yj>S-z`g8Ht{7A%hB&4iYu
z<i}PzlM7XxB<%zAEmfynJ%uM{itp1vA9K(vY2ga8w0<O@i2aYqFfViTVjc;FUTu?<
zF(jwLY&0%L6$a>0GDmlws4>a^Q4o(IZtxQui%Uuq>=*#b_j`x}u&lPr6Op~c$HkF+
z3sNAqnReLq!kZrvB^;jnrVj}tp9~edKP7QIV5v)VfP1*SNsXs}ZnY3lcMnX1PR$b|
ztnYVUT@jWdyO%hsG~(aVE=vsg@MNzHIe-A_c0|Mu7`E766V@p<viGIQ6HraNVY4zw
zJ(vn(qh018DIC;Nr#zpKV*o9?%yI!GBq+aRw?%`g)DgTZQ_n=RrnaRwGdwqHT>Z4b
zAm4|)nEZ0$D|=$ze1C244SmUwp$`NWJXQ&pcdo3-%GYs$Fe#LtGc5|AY((jfr+NT|
z%}Cesx3hLTgj<`}*RHU6-3Z}B=}|vrgY|Hsy4gclA!D7X3}gZe^=5+ZK#XT+kl*Oz
zd5j1m>WA;rQg_X+{p5h>rv=TI+ze2gv29Df3G)U(Y&#l{!`?#0W()qL$!uyBor?pB
zQ2a$bb^<b{4XcJWigDtW7SUpaW`AMTRnt3s%R7dg58m~BJ#bYob=9PF$dn$Jk^suJ
zNZ2#!>qKJsPRHxcqQVGaO*qd^GO!#>Vh2GV{f!YM3S?Zgun3cW8(?PP1+?7&f9oaY
z!i9VQRREYno5dEQK8xKIrMW*6B3co`ks#sL)?^-5SL4yNgilUR#&vmxc`z9m-v2<t
z1xyMcGfm5v7vq=_nfKMrHo6tHSn@hA=q+|HG*`?|KS#H>GKB(2+8|()>{343iHMaA
z@?(FU12am)7#I4ECFE4f+}lEaVW^CsTb3>!l+&UBH<YT92XG28l~3AM*jL9a(?>2>
zK#5Ck%!-2_ZSii{wn!CFy7t&`UZC+CpVCBgrR4;*NH?!IU7CbnJoIeaQA2=#f33OR
zxiA3uX?PvJ89C${38q0PwV0(25T8g4<&q|9N}9ao{!4|b&ddgMFJxkAqGqqYkyg02
z6^i}Qe50D->seE`>&XPp{s+^L>LmhtzwEz{9SH|P9)4Qra0NgeGe0JUNfA!8xapy*
zf!$dlLVnyZ%^{3qrZ5zi0ih1^Ggd<Vp;vX-NFJopW{TqNTSX#|NNL9q6=%M!n5p@b
zz?&At*~KinZXegs{uuCIy5}0t--!8i;1+YuD)2Ko@9@al$;p@btg;ghV1Fr))fU(n
zp!DOIHp`Z#PfEF@@EeI%T)9e9%i5d(6kZt;AAQLsAa7l_IpxZ<V!r#52z}yBb@*Wa
z+gew-$O#<dJ@rXBa?A!XhWsWiY5_!Tj_4SvLy0~!{by7>oKLi{sPg)NTQErmHBEkG
zAqqvcRw0kbH%M+;#N)!cJ0pa(!<9!j1+Deht%-iJqL;Nq$GK6Vk&sS9DGlYFXEihU
z;+(fZvLx`!7NP-fe^G?S#D3vquBtsNpY%X+*g)*?x&<ibi^#Tce@<AuO=Ek?w9qNj
z)D<7!$^&8$=Z26OeIPYZI_N@oN3<haXDwoBF+EByve`7%znT@Z)ktIIhy1>&szG!d
za}DIFk*-UWTBlW)K8-&uUR%VL#j(=I*~8;*52ZwLNim(2Zdjqq7Mh!D-{vKtn&XgV
z@VdG>$6RxF1YQ#*&+p^_`d0tE^KRUBsC=ho-Jx{A|8Bp57MT}UU9T>>cK5!7htFd@
zJlb&D-(fVQZ~xw>2y*-==v-uAo$0})dY>O|me9wBneWUA^1H=r%QPna0}K|2{Gj|c
zD5kDuh_;=gO@>Nm`3qH+`t17$Jv0vTv!OPmyUoRgAZG>Rc(S9JqxuX*0F<|E`r$lL
zfG93d+(PTGo_Dc*KHthVpVo2IJYb+=AfiHZ*`{L*wN~g)J4G%c3Pb02Z^=cuNTa%B
z1Icr|=dZ+Rv^3)A%~GPvb%?KCjJRh1*vE~kf+T~;%ih0D%CwBCFX}AX7cwLE`+mGy
zqm4xXDf;vEl2kVqbG2Lmy1Hri?+=@W)&wkJJ;JVz6tc{IOyBn2XKD|l=LZm<Id0?6
zrV;<nzP-Eto89M310GHjbrhA~Z|dsqw0c=vnW0ALKqz7ZeqlwP+&4H3SFo~vIhd5=
zCpk;NS?E?~Y#(kGDc|^C9TpL-dB?{$=PHPsMm&*4_4akd1#v8p`+XSKe_MgnjpEc`
za9-GpHPZd$z1_XZP|ij!Kf~n71<tc9Z5Jg_zPRS71O+bzm@!&fXo3qy3g!@D1+yk3
z0Wfg0Q{P;b&Mv#tW)Eu)Lg#a8ZNXbHgJ1M>5-2^gQ|W2mr=*Ebr7(pcsg1w4!V$(E
z5evK7$4}&+sQbuC_P%b%nHL~o1Jctc^#{sD#K+3UZU_+)fb%d5OPqM))>VyL<F<L3
z)zoPz`K|xOm|!mc3$gT9ZO^@6ZORGI(rsRJvuc}wGu9Bj!x6h1A-R0fBmG5U>shmB
zVmE}33wrv5|3Vw_#rV+aL7vmhyM4pGH2H!kfH<bp>=jjOY#vdIO1u%I=iws}S>4ws
zK4EoDko(5BdmI<+44Zl=e}UYk9zX!#0pX1Fb(jVF0-61Wk+;HG!^7jP=OVf1FL76Y
zgF7<}W?Q#ig)sg<RX62%ri!ib><bKW79E)FsmhN0a@g|Ry=*s5kUzD!)-ixh;=fYb
zsC-{F<YXw>!=OQclasA1z9tXB`9_9r*Ubq?!SmJv=KI$n@Rc*tKE~WcY)aADR`5w@
zpyQ9HZ|P|MMfWjVTn1@j!0B+H+I8Ynl<HA&Ei=Ifz*Wkdk)DB(H!fth0&*v)(#qD=
za6f`7eBcR5EJ|6Gr=%dA`iX8<Y{vfCw7w=&Uzpqb>m6h_CJ#2!_iaU3n%_%kDl}3k
z*f>1%zX`?>4-_W`MJjJgJ)@AqDbz?3hMd+M^2t-Zo3qWqxGyGQ-Am|*ymI2xv(Q<{
zIE@qu|GIC%F4uu=+HR}83s!en{Y$h?Fj~vAbH74v9)@!-AAea`^dc6TS8IQ-1DDV;
z!r8Bb2FG>9peYmy2?)6_zw-?$riLIm^o70qOs{Q+j5HlV6{Zs-4L4Z2aD8nJz-y5M
z!wqcvD`MHag*ljC{Cy37r%np4dA?VJjljtqsyaDrS4Ej_Q1;UIhLXu@s6rcoBP%a*
zj3K>-QX0zuUXH||#vZsj9FX6bw(q{jv>>6q+J3~q(`&Jdn97i6;LYKCRIo91*0GdV
zS^3D#hTgto$#a>#=KepAkRrWY{l`st7wf)0vUKB9C^sNUn@0rhPuE;hL%T}oqs}h4
zeA$*5nLfm`GlP_VuIAzH!h9U}50HJv6S!o7Gb~D+{K=N_N6FEi)<-5K7anKCf3!s7
zHjTQPe<?=`;^F5V6rMJgtiW!zb%@ODdwzwnpx>Bdp|j?G0oIYFqlD9Z;O3G6$aznc
z?>)aA$b{6e@O_2;uWaUG(!w;WjmuSvkThwto;CtHnNXXlD_^j|tFFCyC`jFYB4IKj
zSg>&xu3VsRJO%+<cr0~wUR|&65o52&i(M~HG#ZO=(>Eraof{R_T+O-lR4#`5kDHE*
z=r9-_u=M<Mn0Fk-Lw>u$@jFX&hPVFvz^1~HjMGr&n6huT9CH*+L-PJz9)cUU=GT0t
z^z$kJL`HTJ1(4^5yE7L~5p^|HZtiWe)DUW;Tf;1xGE)*4zbEe?)E|HWT5s~>=llRN
zl1rnqU$=W;F~CHLY>MPZtZ!+$(yHiG%x&|_f3#Jd0R;#52S#55qroSCJh^(bQFoyC
zS)FM>t2d|V!~Z6MoauRocrt_>&y90>(vc4N()IPlQg`FR(R~p8ez%vJ`$XP=d*S-}
z`r*yLbO-@TW_`2UZPh4G5Cj3~8tu>IfMjPyH)pfyAI{@wjyBT>|KFfb=etM-4mUPm
zd;pT}p}QgJ=2v{gzlgqK2a@i0-*7g5yjBMSuhZ9WYTg$DcM(@3Ui)6M9!z+Yv~MlR
zB>l>8>@8k88oGV<1d)1@i<Lkbw&xBX{w7RuQS`cWBrauZ7BfZ1(?A&f61DGg0_4jl
z1<+eNya_4up&vCa*X;eH#)}yYcd^q65H1+<c&w6FP4kv4dIP)gqy=7<cwdfht0c`y
zlh5u=eK{-a6%iD9+!>$>r3u*W%=s*?lNJypu4l~6G<?4-5wonyOv;VCKO(*EDw=kv
z+8t}*KG)QY4{N-LXkTb{^h$a)6fmzqq9BFgSG2avb^lpw)pT=t{*OP#)na=7c8V{T
zd|b&bi1<gt9D`Ok3X>1>9o|+fD!26gKKC#siY0q<iddx)8ynl>H@eycT)g|ztk!2H
zCSg})V%TszD><rNiofqH)wG!Kt1q6}tZ8OFwa!^B@nEZFm$T_V6;95FXHR9gOaWv8
z$2g-bCo!@Kpaf*VtUfiiGN_-K5=IN|&oE740UM2NJe{#v2>YBoSVCwPzG$evXtt^=
zB0^{pe#(1Kq7Bt~zvQah=(5CE?){1)<vcD13x>4BJd9&1f92^Zm^eYoKHtNoXvq$k
zm}F0|`DCd$9gb1waakT~4IoZqqmcPu(T-j_oSPwI36vizLtwgs34o{xiYzukWpVLl
zo&c6!@M{+0oG<z10qgC^>UTWP!VKg$Ep>Ac#KwMe)rM2m%RKN?CP5vLtL0Bj%rIpX
z7&#2)e#Q0({=0;Pe#zNS9IzJ85{R6+T^q_JLH{Gzq_0c|Y*Xy?^$!5q<ENtz?<;(u
zs|4`pjr-=)<aemHOa(v)@1n(s3kqGGk;+;BVrVuvz^^;RHwHbEv|lgf#CeTY&$c0;
zO>@z|01?ayiu7!kE<d?woqIS3+JGMx!-zvGlHyiXZNkamop<m=!S@Y>g$3{z0xh7G
z*FC}-J-GJ`HucOm&AUQfa6LKMq9(u!yWVwDQvzg%gsQ>@SJHix!58#>VoGhY<Uiv}
z0-m8@^uf&BfP|lz)w2-A2!=sa0fPQUA_iU34j9tDOiWeOY(fETkr)zJnSV80N!#yQ
zZ&xx)s3T(o?4jJH;<C42^w4uV@%1~ay?b510?;u<!UU4elz=I@v^`{C9~fF$oE%`&
zmyiXiPbpO_qVeAMSU}gZbETn5k=kxVx-<pqXLB1qV1f9&AY60_2kaw+rZBD2_sBz^
z*~%Aehr7b8X8OohB;P+M`PzB$si|Qgg$<BsO#O7!-K1QmNEl>#>Z_2IL0-NDVk>cX
zUL9G#tV3K&wNOU8Ur^lM)t4?3%gUPF_6H)d7Y()d%QAi_zBkgWW;t18{aep|+eyD)
zuH-7YeUh_aeiIVIe>jGqIjPcX@(3gc8B0IdVrb2gy+y?=Nv94f&tORSU@SU0Gkpgz
z??Kba#+9}08*z0xwMmp_$d{e1bu}-}zkAmvL!lkAGXvlACWOhs_6xvd>E`~4cpK?y
zkJ^>Q4_oBQZ&v!K<?`=yX@qm!a{aZ<bs3X`2HQwG+N$f@VnY$QojK`^G+juOYH7X^
zQyi*~KB%S42KE^=v9jpMy5y{t?7erTx_GMQs7WCkYh#<OtKRFRNwHF^fqQNL{b92#
z)rS&xlp2zz2jHXy;sX03Z4)D@uGju4|5jZSQ~HTobH6#VxXowK!ebaU*zk}Jw9K0(
z`T8EWlq87k!v(_jt&&3+o}`Bw9N<N`SUd5V@)UF&mPQ>q_3gp{y!rZ5Y)-~R5Fd^A
zG0LT>dGYl`D(`x9^I&SNI^zaK&#8TQ1?SFW-<hx3>`Vk_pjFgCJuX56c_wXL`vN>%
z*1r}fT(=p5lYt2cdW~$jCCC#A%xjT)t(J9J`6JD>h+gTe5PiSY`O^5wzvLjc$y;Cs
zdB6;fkGA~~uwCmRC*GO#(aYZ}l^@sQm8;M`U;BP)<>)ZSy7pInV+dt5*vQFwWcZyD
z%&hLP#eSWtZ-Zs8&wDXb47ev{L-5uBG4ZO^dD0hVS3E;jVy(pG5`g=i@1`sr#8>;-
zLE<Tt2nHX3`HKhg)C|FIxl}(jLB4p~4Xc9k4Zk<<d2tem85d+ILsyq2^`v}vF%%cy
zdVaOmjQ6N(dllieSRk<ua8;@jihE?s(~5Ccw0dq5q=&liJdKZ2r_rWmOgX6v3*z$i
zrVu|wqX{0BStAdL1Tn<|O*GNz<*U2kzaN3HdY%e(J)i2noI4o})2M%mUJP6%l8GZ*
zyE^{VEN~KtKXumK*GE&E#M`t1ago~^vZlD1@z36jL7#ZB3nxY-ca_*6TTeJy;a6Ky
zRF}-F?Z6j4vH;_G87saAjayxxWs(G#b+_<Js(^+>TW-k;wR@`Ss$^+kZ94jXgnwrj
zQEYzFeRVlfWF%TwR|H<)yE+N3?hKOTbyCePpL)XS7MXZM&+~D10O$univy^m;3c?M
z8zREy8ZXpsG!MSbFxMFJ!)(ZtQ_TeWiQ8J>#kST5U^~Hn_52SOdke)?t@>R3kL9w2
zNw<jMqXS??Ra=`ennP9M!#73*o&wABtf}4pLmz-L@gFOSobFqXRvy$_dj&H%A8Z)C
znQ;mjUG+h;d*+e6e+`zvJ7B;PF&go7*FV`-FTO<BFcg}H!k~c%B&h&Y|BO1?iB@dZ
z4;a|Uzwif@bU>l(R>Jd%gnj@chQVRsR;J}?M5`hv&WRkAU)o{$65*fQ9H2uyY8f-R
zHo3nbEayjQD;cfY(y-53c<)c%AgWh5?G}1o2I=@J6ak-_nA99AkKWIMlH69#Z7yT$
z6u0wRY9_*@BIK<J0R`jyTy8b*UDaVh2GH8Sr!7le3%hL^r*fU-+rQ$YV_evZ!CDiN
z+-^MwCibMUhbx}F2-`G;b3cH=9j`>(Wxe(NVqvbtTk0#>0TQV+#_{Zp_xi<2^(4(B
zZK9nvWT1LUk(><dBYk&r$i1-7EFPLrQ<GleF`8`3?zD%swv&Dku#Vud=Y~rX(`)3L
zUPaSEQ%NL4s&-`eLYtWs{~!+At=v_;+g!bi_0OfQP_Hya^t{A1@6F*(%RSonX0UU!
zb2F4WrvD7}kntLTGT&C;oETR_nSc=C?kl2WE7Gk(u$9HeZ{E!vzh)!>k)iZnvPjc0
zs=Bz|+rxE<1rSLC#H6H1S?Xi_Xy@bp=BB}OtU0<%DmqX^zPqby<9}EB^*xE3K_Vuh
z*~95%8!=VEEBUoGD{J@o{NMJ}XA9hkOtIb)l_|6li}kQ-9=++|xRRK9#Hg+{rL3s>
z(5(^6qdy$I0PH2>nxfED@3H+R8u@!>xaR@4@7Ki{JmH>>`l`|Q<8>3G?eU{dMFalQ
z#fz6_ia=6YS6TMC#-}c%QSIpR_*~KfC_hx*8}YeUDRE&yn}lxaTj~RGbp;zXD<@A4
z5SU0*>3q(Y@5^#z*oZoqdo~&i9zWdmhaA*%JiX;YE`ed6Q8-L(Cm(_nZOUt!)KTh;
zKn`VAR_$z9Qo>u@Y2rekNZby$&GL9p(N_9CfjGt~|1IU!GV|Bf4?DCKdo4$|e++-o
zN34E2>TY)yFAT2z>nk`Q0brUC=5Q}VC8}bCdmW@d?kFamUr+q)R~cKbMJMA1F9z4H
zht?SYEf)=n7@O0j<jANHQf=L=U25obXJq(%TY#~ugN!p3N0p&1=}&@H6Q}Yr;`~ev
z{^dk^T{7!Hl{KvNeI=$MJ%D|Zi)bE05I=n@ob(ex{(gZe#h<qF&H*nPqts%a1FCng
z_x6{tgOcV-CHZVMnYQ%mr4oLD2gSX1Nc&iTrocxjsb#6puV`%6+Q|tNGzk>yj03#=
zjIT)$r>6iEDx2Pwx0<t=og&VrcV%4JC6@^m%<QDaA)$%<hI)|*K8`CeHF<)VF&TWU
zpThg)mVbPeZ8;{z59(wWr%baxITI#(D4{@0O3w#!NX_7XDeKW`7L0CgP9FSNwUWfk
z%VNg%A<zQdT-~1BaQDk+kD2E3nQ(emK=NNVB-!d#nz4RFHkFCmqt>|3?K$n_9!iT@
zbEz>~sQ)BLQHHWoLG4RRUP1hbfxWB(`J%DeL&bvc{sYEbEuA8S!IR)%uf$?cL5WGr
z4PpZ_F{2av2w9}>hx}ZXSH>&vtSUL>^t||qgO2*piAU=aen0u)b^W{Rn<+}AOLmT8
zrlymc6PcePLo07qK-sDJER-y!ax8{yEB@M$*TqYr2t^JJO~dcfrUJzG-`<J`kTar4
zMD^B($nFJrl^9Wd45m^B^zgLp2*{@vYW=9c3+0O8RPh2p*<~|*k^^CpKdiaJnac|3
z2RzycW3c;<d~u}s$uQZe+5$@7?;COrNv4{IDxl8K%{&F+*;H{X;-LB<_XU9WcQye2
zR^EyBA=@$9LXKu4=^dhZ-net?Z}<R5L^$aZoybvflzhLG*2qb_9Y+scfXwQl-3=S&
zx_frKcu3AN<>`zv$GteKw>{jkJ8RcRS6b(s*UrIh_>!6Cn%(ACC)>RV>8SS4ti?38
zL5jO98`4E%X|Ny7d+jqM3zv%jN7Gq8G}*Xqc!adHNJ)$iX(Xf_-2)z48U>^~Z#tD0
z7;P{ZNGaW&1JqH{AR!<OsnNXqeRx0YKiKZw*L9ueaZKal#&2J144){z^~VE|d*zoG
z0!Fp(l7{kv^LL<Q3nnI(Zq?Wcw<|2H!}A&oF}y<y*Onx*<zX90^CEMJK^NFvs7d>Z
z!eKY|)Y0HCH@h?coG($$TcSpk@(<PdO^qrjgfg-njAO{y8HzESeqw{y6Qw`&153;I
zrdFee#`f$4-L1V-lbGqXcYF&;@6rbT!^7*DUQ+Q<liv8K-u65`p&-EpZ{IdqNUlb;
z`ad>rQNLG(l*<LnIZt}^ve<mtV4rYc>T|CdL-(zjXq3plRAvxS6J};3@Gw38?rFnU
z=l>L@PULh^6524_il(ju|5Z0M0r@9?^Y-d$v@Tu}EK;d+pGbzlLCbtou81E<iF<Bl
zCan4Le!Sb7nj`kPL_*o_VmRz&4UR#mR3OO5lwfv+wYgxSYsN%(e>Q_;^t9XtkhRtm
zo}(eP{C3v0)DRWeYk<$<=XbKwzIih)tl0cAcdH)F)H20I^MTV4TjUIXx|t*oHXxg(
zz52R@P8+_hzt3Qjzj63PAJYjq@Zb>6p9k19T^GKxv;WuC%7-svXJ-#l9y23@;7zFi
z1XG*uFLRuSyIJFj<Ps;D{Jt}@Fr`?_)`rDlH-xj{mXBwi;O33M`hJ8P*S%=|!#SKO
zN$>iAgBt?!<;GWen1#O2q315?b#r9q1ylHnzi?)pFgU1YZ<*5k>1zrwQtNuUj;|=N
zU~xtKeVNC0QYOoomBx3wLLVsW@p01$15T;~#EERVnc;xRa}`16zU8+3PX+H3T;}WK
zQzvMU%S%MOuUob#SS5#ySY#L*e>)BSqkxx}DnusY4u{>gDz;zZ3^9arGUK#F5gppH
zOxzL(#1R)KGZWF-D=Yv~-lykBG8f@m%`mQ3WvpSaNN)>!-PyVc`l7+6DNS6-Cd$S^
zT3(ph5(?v?4oHji=7=GZ5|5=ElX(TI<50RN36BsI59lhKZm6|Z6O2z#@96Ag|LEVj
ziWPKB4kEWvyC+kMQ#;?<n%&32vx{9dB`<AYzjT`2QxgBfx<!H8<3->eJ1DM(ps^ZT
zL0Bw~<`Xa|hRrKhCm8=5r6^u&d;G)fJ|kQ}B=O(c#WfpFe~TVS#Lj+FGc&z_rTX!P
zXsarxVvVFjz83b_UOJx;@<+Vn(#!mbO#o|U;J!6w9x=PcT)5|pMwm}Vw*T5$#@vDf
znq&U6Bljc~;Fua|FynU8xz{Eo$(4+C3hIC=JF{ipCml~K<Sc%DsMcz(vsLJuVQl6D
zYMjmWudXgg5Rqa20M|7kZ1u_`3wLzdGx!POQGfG?(G^6eMM<r<fJ6|SZZA`D<)Vla
zk=7<S8fMfx2lWk3tw0!4uVCiS-%Vk+w{AJnALjOfQ=cMHK_1x&8PY(NSlhnBScXI=
ze3thJ9V3MQ{z7>n?DcBX;xOTBfj%Pw<tf+l%vgJL_<_4P`bfRr0|dJHMR6JNjkTYi
zh}6vOI;bam>nvOP>fFq7If_@8hc9-7N{u-dl=`AcVtSR(j1=soiP#jIgD*B6Hs;4{
zhP47O%XecF_rF$9fo<{a-oiTY<ediA$>S$Q|MUz9_R8InN83iS-2tyd=Y%vfIiCN`
zhlN5MolOEM2uLZ-gp{Gw6!-TfKd1({>Pc;JPTMQR#KB7kA1JmB<=d~mR(~3W35%nQ
z-ovQuJOU0B$g&^F63o33#6S%njOoLo`AGqGb^>2n%hQ$QL=D}2T+nCcPV|IAL~y`2
zsoX6)L=+o2pX%Q97sTF*<n3RxL`qjGPP&e&D|dlDh={m}xKWFDB94l^>p~m6wuFf{
zPuz-xp7Ff5+)F9ml6b4aN#V}{zKHhFct|W6mS5>BK`%3W#YN8a?jj(AVm=jQiv|L$
z3n(XjAEe*mOXA|hbWwZpxl@6f|L8Vfcdg7GG#y%ohMlw^Q3X<-aaA(7^&Xh;)$-t)
z6C}9O748wDP_oMw(n1Jg7n5wUL@(ZYctAJX4>!=ef09^OSYN&@ZtnV4N4T>)XUSA<
zGa27E?n0H3FsyFBaPDr6)*T301kf0ZaWP?NGEqTXRrt+f!W655J7y-n$Qn97n1qz*
z{_dq0=*d|8qAz2=@JM8PU6FZMlErq9Jwm&h2uH;#nYsOX?qBN9H@u@6ER!JdSVQ_u
zROG{DOEVGr7}zuCA)-KW@u#ZCE21eLE7Ni8;J(ehy$|xHOsPF$ra9)WfMi;yKwEXY
zN4tCb3Z{rbyBZ}i(?@2{XT|;S(6YPwperEF*4`Pm<5aw3!Q=d+W^}{A7feT`zc^CU
zX^9{Imw*KuUAFEoY9(Mt1X-)t(UD-OxlOeK96gRpZ)4&rgIZv{y-1Zl+x>;}=cR%Y
z!rD(yx*GRS#&dT$3>8ZNxA&A7i33Z$@vdqx6pQr27tPHdS@f~c+d(zeUN7sFV?_?t
zjr(V+D&xU5M)Km5(dKKhPa@ny+?sB`kd2$#Z|QG`ffzD73YG)qR_72uksqS2o7gvW
z%(VCfm6@Z-E7}nju_Geirz}rQklD9fG$))`7s^3<AW8yeu?RHBWRh|dMUlveMv3nn
z2N~6V_ZvHW{k?seX3{Hd{W#A}KugkA3zkPtSMuCxJey!&>2zhsrRIH}_QD9~l-7g?
zMqb9@M8eTcfw}t=%y1ue4GKC(v*CD~uAU1eSc!rYd;0Kwru?`Lp2{0b^z*(%{JV7F
zfQ>>5yRyk}kN`)g!^v|^_EeNz0;K<bO@vR=?6KT{QSGDOXZ93#*juhaTD8lwbQ8zF
z9w^HjX$cPUyNBr--IcC$E1{kX1z$(3UMVV1z`h&Hhvj3f8fMA)E^hE;UBa<k%_qpX
zL9=R%M{Ra9+RdX&pCL3j>~$MD+z53<X)w{wV7ooO1!PvS2l6;0&O$<eoljV}5urPq
zv3N24?jTHmZN%9=YU|6g-@c7vM5{tK{tu#YtpIgK`GPqMw<4s*2Ygy3Ge;5Euk}6~
z)R67<Z6Zcl#EZ+DJ-<S5+q7Z6EA;`4&-aU5E|ev4#2`_N-A|g;>AP|II~?37KM9iU
zrA3wqK!%Kh+@8=f(}rr(pnukH8B-nTD-1r7+bZARz797^o9(EQI`fFghr7V`C7eFr
z8t+suj$=Jtmse+7E7K`NiK=4~x+fC3;3%_ZEP?bl3MaBXaHbREL81jT5A>6XkW5%E
z;RtM&YYGG=dwK4|r=%>=diC0{%}vUOhtGftO0r5Q!<TBj;>t)OIw2$L4)w+J{WGUk
zn=h+P`?<iyf#+1Nl;=t()+=l_r#s_{5QMtEcZsi4Eu1Vr*;+b(x3t<B?)#FixoZVY
z^J{Tb=;kN@wxZX^mG3~<=@<Iz>du!#>QfS-S_bQ;0wMFX9QL3&Z0u$|dA0TJumBK&
zyA*3V@lnPm6f}i@Q-X0*{nJ>Zfmd2f9S%UXWs3ArSU~;t`7kShZ{N8}8k4cEyTq_&
z`;pDCh0yJy%0wHiCC5mK8@=>GQi4mHCAi3^L`LqQD85sO-2OSN9Y3k=InJNX0w$t9
zs1^fZreB`h#e(0;mDsKoL*i`X8L%?l7F@~$KHzTTgg=G6Ih@+pf*55u<-2Ae#;hDQ
zD%w0k7cubP?sy}5G6mThJQjgcV9gUdl6AHdWX8K6N#&dJpbv>(4$W_6<_HO;Hr7tN
ze~NtTfOtTDN)Ij<JUAD^FEKbKk)_=A$rxWMypJaQcM&v9Zq(kn6pB)^2dF9Oq*B0f
z!TE6T#`gYgL}%BRW`Rhjmn^KT`sGH5GCHc*HklvBpBDoS1)|Un%Pse3wf8>V%UFKp
z2119-WAX0%U=elsHnFpD@!;z6<dw6Y;&|u5;aE?5vnTIOL^xkz;o$4V^Gnp6;>~ZY
zz~=xpK3=&TDn(&8`X84Z5Ek2%S73zyrvTtYz~cGj66&M_!2R&x(|@%&&g^}0Ep-NL
zx-vvBNsOL4Sf*@h#%@l~e!BOb;*vA&6j)*h<44A)fk5Jtn9s_Zj7gCayz5>Rp4DjU
z0*mJP>Eq_t=<BveB=+>*pPnzb#f9XB8lq}C-IA4;{+^)D#!uvd$7yBLh#bHEg4Dhb
z*nb7oic9*Yt#zoBN&k_-lL@)rxM+Gq0i{Wk9s4+s@YUyx{T8nh4EV&_T?4lxEj+8+
zLtC=&$j_&Fe5&JKk9jnX(CR_Cme6y#<>hG@`PK8`|Gx|6PQ`5~-rXu@yNG_>I)*E(
zdfDhUPD25=r$*MHs6>>@rEvH4lRFe(Kj0a0D?oC;Z*v#-(S~!S=yEvW@&HO2{9-BQ
zkh@uaRxcUN<j}o;Hs+z>%(W%9{k}u_GgWiva{Egm3iQg&!t$YhNuzssC+UfCrFjh3
zAb)kTM3L9jPD%kf^qW<bPqk{B*?;#<D@gaw)oY;3V(|;w?%@Y~hfQ)5T&@XGf^_K|
zV;t5zN%8m4Xx=z5LiN-uMHxcXaahvj{(+$O&ubrFfyL0@NjAS8HfFPP;z*uduOxCa
z3>qcAAsd6daC9AtG=H-{i*>Q_4F!Csrb-9(_RD1qegE=FlSu5<dV}zX`rlmR<ttRQ
zqe+hemR^XooB-z!Mjr^A4rlj#Pq2RgBt3u1ARj9CJRz3gQ$D0enZT9Y&!{omj6ltG
zV|_gMd!CX<z58`j^K9tiVi7uYQT69&_VTCZP7F(AleOCii?3p2KQmxWT{IlHjUi|_
zN%Qb{OdI0<4ShHSYhL@cgfq>r@;0)S>W49RRU)SmKO)OH0BEI-lDXs_z!E5p4g9|S
zx%zioet)TWx4S==+nI}I#(6@goLlbcJKuasHd?2<tM!$1iTF^DB_S951V<VO7bdfp
z{uTu-EZ)o#6niEjYEz)F+MMgZcwWpEQE8MZ=2p#iinY7fEO5xZH{bi<T)CI9A;lA{
z6)Rvxq|7`p`G)K_5xIJ7Oc6dr*~3l^m*iu(CzdF!y`@WzlCd$Vg(^WzoTJ(P_{p;W
zy}CKRIrWE)cVR%5N)t0aMr?p}W8pSJ3ya@CqE5Y93%7%7j)P>}Ru}K^7DM27FWx0+
z!pA_88xz5Vhr0jO@=A8K4_;!r5{EuZ5w-vN@tcWZ7*F@Msqtimb@HTtF);4m@fOG0
zhlk;KNr8u!zP?7LX~FsBE*9IUuET|iiQ&!(otWp83zmz`XIsJD-G6|tkkxDKr(f`-
zEtb76$C~S1rVfjS!9r1VmXapfHYAnMSzVp)=G<?`M$yI7rkpd-bY>-#PxAUHm><ea
zyUnI&_DN(8taKPln#*P<BG*x6DIJ`*&{h7f!!6dI(&gC{gB7)MIQ<K9ESK1gRk>u9
zt~iM{Sh{>vt?zU*2k0l0Q<3HiOUS;kg(e&P8$r%Hh|1?l53BCT(SOCJCl!?#|B#cF
zQTS}(e7+({U@&YlT4$-=Z8}(NDrl`#3(Bsrg7Gau8C(zHoz3!Ax=+8+BvEdA&>bG(
zh4hIiX$>9)E)T7m{_zx^bAjpaj2E1!CSrBz$9=-%KO3e_fSZeppopN?mHJrJjb2J3
zrM<P3GpEnY#N4dP@7si_`+e5?VC+LFFRF|?Rw{N7Bx+gO?(WQjzpe83kZ#3#X1R|~
zI!R3w{;e0$x3*~Sl?6_W<_dizLiI!7iFmETU|n~Dh}bZ4;&|Gtv_Q%?c5Y0iPQrew
z$gyu-FiEfsaI1fizp&B{t3u>6mEx&wuK!pKc|}g}K3aAL(MPr1R)bXijO&<ceYP9Y
zjjqHI;!1jfk|4R2(A{ox59nUa`cLBTJW_XJ!f0Md6@13x>=)Lr?3E6-XyMqi(u6Q)
z%6M+OyS;r2dfRjmiYC<0qe3Ff5{=2HBWV#vG8kvT`R?L;;sE<naLEUT2N&zTBDH9~
zz48gYx|(S)>D}w3E9O^RZa(5|q{~2LW3^ygQzh8W<hWINojN>_OKXo4zQEk>@XFkM
zfnBU5qDfY9KC!iD6druk`gQ#Q^M|ON>oI1PyFfI{iZSv2pXYJA0S-bxel$I-I9w@k
zK^p<E<mnG_?3oToLY2Hqk+|lwt}eM+;VF;3HZ-QH5{%p1I4c)LUc(6%EX;T|n6st8
z6SiZFOi3q{dymQ9pQk_lXBu$X9kO*oi}2Z#)m0m2;OTpx#Mz1ZKHW4cuM!Ju)%kr(
zC5T%&v@J4MvoM1+;&sSKdcH@6Q)8m^`4jLRwI3l-;-W_M3YC{7#k%-gd5j9~WftpI
z(HeZT?#yuKYDcWH5RiWSr{brvfWMc_7$r^5p&m?ug&>>E*pT95X`Mt*?(ap`mhavn
z@d*!f!mTma;c*F+q@)i>G^q?Rt^ctvVC22!_cxt8r4EW<Me>4LA)MK`R2+v<{?<VA
zb%-Tektt2#@uxB*mK0CSvVfl3Z_M^-wQ{X?>BLS-49s*YOHIxyjXC36oXF|}ltc+D
z%j8ai;_doW3AXOG`fNYrJfP7yozt{vZT-(S<1`1dzn?GO+(oPfkG^R@Vbh)Xv-mg5
zNGLs5rjlcToHH4VDOc_2QMD<U5omoN6G6!<qeFFAZ|+A(1)-<ic7=27jTT`eqtwKr
ziDHw|JnG`_dZRc93tRM11A>C(2)&bmxaO`<UMk&kMjZQyu&}&x5*4`SQgg&=TU41&
zr1Rna2MV^pMU4Am(@N-n!C<JQiOKVY;0;bLuTo0FpB>*p^dDOdpZXUY0jwL_O;W!s
zMQi;ll^xA6*0*aO!)uc0N}qfd*2#fwsyLoW`j+6kGvH$8wkgjYxI!s5!m$=4ijwUu
zzwTwlj{AZho)5zb;~hxlKb<n7>bspbp2F4D+K&#ZMT&Ng&2pQKS0IYICONG$m*aDb
z2|EWTs3v<?YTB*)!GL?<ephu3*_;ye^pt9SxBGb$W}OP-|0?KFNQ+?F$?0|iYhCS1
zpg7MlY}a^nI`$(;*)z@0NYs}~hu$P2DL8LmnTU~NPmi$~9y1Gkw%}Dmfb=&AC)$mu
zWsTcobF_yTpMMbI0mr4~`%pYrEcR4fzxGR}vdh|G@ekukNEs#YWY^kxA(Tc)I-g&`
zBaWzIunjZINGZsbBFzikWwaJBwBs`&X~Ea7XjARM4KEd0c!+dtr7X4LCsc16*c1wu
zIAeQj4awQ1s#1qgLPFk`e+!J(L`K{BIQ$CM&e*VH36MxE*3NR=xa6Iu!Y3gR%<W?%
z6**>Se9_595upztC}SvHwtvrF-u>@4--uE36DkAsW(#<h>`CZ163roS=bPo(V=c7L
zU-<=zQD$g-P+GM_FNCVug$?gt2c7!ogI=Q-n4*s5-**UbgqRvV7?E$lUSnK}`|3Rl
z<gNCMbD&xoY^}>pz?}_nx8EK1W}=TsxiH|%Xo4o;8bEaR2e$BtYGj(XnTWLcN0t~3
zj|v3|l~5nj5nszLCfz`VQnoZK`OT>d(IAzFCg(^SOgB8DExWz)vWCx2)7^pWc@IcV
zxs&SE@bD9qg#)~Gw!Y;Rd3r(B!*?7}t;Rx<d9>99pWKZ7KY?0wmZwc0t$TF&@3XM+
z#R(WaRRy|s(|=$CZgb(QQOcFXJ(&fPU9crt`aT(T$FW=LAp|jhS+)yfcoiX%uLts8
zAT(%xMSL@h+#i?5m$RUDb}NjS;-kg5u~uuLzT(}ourNhRFwX~8EOfS5KXf0t@nS92
z;<L^xzB4Z_2&Zry6<MVY6|V62gfR*z9cgykC!m9OMX{E(>&;w&Wd0Wrr{bzm4y@0{
z#cFkAPP%T9(=q^a&2pwEydPYHT`n>|Xi{AbOgHU=euQB*S6kt0s;*BPwb1HiY5Xqu
zF1n$f?Jfxj%wcD1CC~R4kVp;LfXE~{YD7Rw22;6FUIL^pM94%n*+|Um^`CL#p<cXc
z!0`!$B#^@)aJ%>Eden3kqlt`d2|Gk<C4{#yy07n@S+5Hp9TAUoDCIW|9GtyhL){^q
z(i-v|XQzvr=laau&-<ZXR-m`v+&AMnqKo3KNPZbT=r>}D@x(Rzh}<FgVL<jRF|q79
ze>C~dMv2Gr?)%^u*=`hI3ZLjFp7H6AaexT2(BM~;?BYgN)`61#j@tv-X|+frf8UwI
z1;YnKGdU8)P);ZfPQz?$>#T*o5q{4gt{UVxyzCnL-D{j}LJNm-q*(}vcZXkZ|N94Z
zTgHT2qxVBh{kIn|5uvRLRuND8u--{axAO7JFq!u2+otf~`%BFI4pvoGa!;oQID;US
zZ{J$0cvY^AOh{UBJTe+!6BmztGH#ko$&HuJ+iFm)gMaQFL#MY2@7EOBc#Ss{_hf?d
zLEhGy)T?S_f!0Jpa5GXw*-nLii9tiSS|v97;BGVEZ|8fkkA*8gMN%N?+-yNe3#1YH
z6aF;sU0GF#lqF{KO`~3ul$1am5Mve-6Biv@G22($FX#(tXj(-#&oY+psI~pOcr8Uv
zrA(6iMEwH~n{AG{6{ltZnKMb|`USU1a$Y|Y*<Oi5|5Fldud#07%U)k71Zq9$zl4@!
z`z&dmthp$B`08E1qD~|7#5g(eNKD^zMK0a#5nmi4k#IZsmj%X);xQa>s?2p-Q5j@f
zltqb&i@%@TQoX&My*u8$Ix8#CFXGX^TAVu1R1s;a`)<t1@_E@amx)OFLxJ_&!ET8B
z)Qwk{2uVlF8IjY#fP%PeD=v|LGaDsoien}-k&bb>O6<yc_@bKCR@>oqQ*D)7p!R!K
z*?e_7Dla?6xhzVrI8k$8>r^;iUA&2{&VX1C*C)1>kZVzy{=KFajyW2^;q9%+T<TSN
zE=uM|_}p8XLM_(3LY-YDooOkwt=AG?oW#hE+L>cA`-%9J$-Cbpp(=hxV)w-Bv;)Y)
zzBi{@#pC`~$08ih!=c-yeoXq*ejeODwu*bt;-_!xG}n};okPO`wWBMUgbL|j8Zo{2
zgo(XG^JdJ_sw4Bb)bYyW^G*~#jv4$W6>MvV5a|*<M47Dcc&@O`;oN>mH%%O^_Z*am
zvOOc);xlM|PTp_=wcFQcTuBq74ZDa$+*{Y8kK2;=$5WcN;qKO}UBRKPa<p^=E?5Ff
zV>7oPv@G!Pb|B!n(iM8`QH))Q;KOHSU;ARxfZL0N)%*3!-OA;erFUQ7NLLgDNTmm*
z%gXkp{Q+@j(n<jinX0n2Vh<kIn7|Q4zAg14;;~|1q`^P`aHFRi-We0K;6v!Y-82{v
z(C2@j$q&(gI@{vQ9MRy<;1<0mC#Qt;!cJhZq~&184&C+Y-4zMB&qP#+%Y}B)`dsS|
zw*K7;tFhvN4pT?`zVYcilG4#caWbiKq3Go=wflR;&!h)YS@Rjae?*tgB0q_rVF9@*
ziOe_;mE?_cVf;8;0D6^U$ryV+`i6WDJ9l56ZFehA1efWR>AW}VD;zMhA{sX)O*lKd
z3#;ACO<EE1w-f0T^vj%@J7c}SxZ7H2|4L1Egl=2CSL{qx6HCDfw;Y4o6aU&U3Nih1
z-Q`9qGgk&Rc3<q0MDeZ}*=d8+q~^^184N!*!lU?=R1y;b<mwW}y|Z@z<fLhZLczrG
z{*YzlyIm0OyBO@tMzw$B(w3h+yD7$bsfZZ{kBxjnst3+3<%<T{r9%BK9<d&Tb2s7@
z&$8M<tXzp9d7U1R&_I4oUd<z64*cvqt@mThcCJL<RzkFKV-j#SY}fhm*NO5jXBk<^
zzqXd-*iHcsg5agE>7hT_mwwp|=oH@u;s=dbdVyp$Y_f=29s>OKinHabvKJ+9O8?2$
zZM7;?NUmqU<4{$R8Tmw9=uZ2)fcyRtLp-5e@`#K1N=NYYY>3-jZf+MEn=EK_$NG(s
zol^S??LsGroUuN4BR{=RLw#$k_r8mw^-|~$w=J+>653+QXMNq}F7yCI>NdB+BuDa?
z`X6FpE3<WByyMIDtR-Q6m120r-8k!gTkVtH*v#ly{Dk&kDGs&mv4Vb%34R)J_4KWo
z$>@S3%V!IX2JDX1y&3|eRUoaVP-zjlie==<yc1(IsqBAtR3iG2VgK|ectliVBQKe*
zLEHdq^U{&Y<O9<dZXgx5h6PzYtjG^lp^JkF+MBxl=vf9sF^lIfIjwAuz-5U&qs)5b
z@V;MnY^E;AH}LA_S<9RhzZH$h5#-lDtj&Uk$6Xv@_<UCv9G57q3@EMgI<sDecm#hH
z67_);5*Z1%NF21U7|pBD!)lHpKCOi|z{}#Uu)C!Ng&8sZGangGJM8`^P=!c2aT~jf
zCr08>A43H|aI;F|uhcK4>CS0TxJfGoL={V+VV%vwQ<ih~y;@#ZrfJ(#C&?u@?;J#s
zeI+$I^Z}W{R5JnYdm;I%BKuw6qO5;~(S5~zy58CUyjR=$_|^Gr{4cSL_NpNPpxeWG
z&fU6EGOcv(P84z6OFLSwxhF_D>g|2d>6c=m)gqO!EmKTqWk;A5%XfS(Ks-xP>CH@Y
zpl=}X_Ox7ClA--|dC6DNBn|w@QP$nPs6TrakipnQJh(H<0NMYzuVeH1@aBBiuL`J-
z$sC&|i|^lzmD-)XSdYvXKQ>eVUJjEi4=u}A1~b#~fuP*a=mQTE>c{kJLO?`{R;e`b
zk6vn-8Zv6HO*k|C{-AipNJ)o#?L3yRR(K8EsGJ>+k7ff@bQ*<Dpk{fzJ?nG~{sY-i
zPCi_2f19}pq$PUvb-=yC`!x)XJU&l!qMV#npF`(rz{dUoKlnt#(vueG<<csGZHri<
zF`WlH!`h|~yb(WUBSO0r!^37-22@G@l*YO&*sJ!uSf25X8fD|=OQLoLGP(jZ#yN3f
zAZ$9VeovZp(i6l9bd+faq(_KTBP+#e#2>S3c@KK~5(<Kma#dleRJB#s`gKb4{mPR;
zWjc<IPO+AVZ$vy-waUP9|1tyRHZ;~qHZN&}%JvzluRf(A5fW*Ym_pO&LHBb1`^XQ&
zEtq+ARVA$Rj;R6`jD|aZLQ;tAnQ?AM!pI9b0;3Y0^2^(Fijt@?yZIj_P5oI@m6$2L
z?98`GHae?_v_4q0(&}a3C$Iwb2m7d#@!NpsB{|J$6#JJfHNawsr+Z0xhzyn=W-*B^
ziv2qGpwf|>tLVeE;ZGVy2AOS-&OI1QuFlfYmY=fmJh+@~KS}?*6*pAXJUVsESah8J
z+?j(}(`cuEMhN^fP&m5nYNhdPXh+-JxxFxkr@Z=l2%=qQvb)Xpct}pa>_0m0HMTz7
zNTL)XunxAj3)H~l{WzH%#hj@>@S!a!Ui|k0ES9!%wYT!rH7?Pui8`MB%>-K`M3^u9
zwC`!Xn=`X|SW5!)BwhlJBbf<%dW*!9(+Rn7-dw4HVD&&rjKWl-D7~oDG)<2nB-Kma
z_zsc1-jF5v&3kimL6QJUOZa*tmE*CHrB!_STZxjvQh?*F+?L~TxKE^KM9{BHO2U{-
z8!6{+bIuB^nzA{Qcpr4prAo6&QCshN#inR+ZKW@c!aeuTJgjj}SCA4Z=2!l+&|XI>
z_`5`{B3A_cA=v*V9jG#(eR=6cJ4p&r^Z7UbQ=B)8YEdrcLOZ^m;nvOmmZ}$ivPHt>
z<@%U|0KgMD9+ru1w^(qmBZ`G7Fzy0nUEw$DnERXSDb}va-UQ|XC59gQ0HTSXU!6zV
zla#2l!|ZS?+WikQWaTp~AU|bXq&IoCo;_F=nR%mhv6f8BfGA4%9-Eq%)e%Kb{=95R
zzQB%}V5x!P+waZg7;~jK3cCPEo5|bJo=$-fPhi;`L1|a18VhQ6ATS9erOXXS!vidG
zbU1A@<l2|kz1Xxg3*v3rb26s_nhpqu@-SWm1h7iW?SFLaV>+>45(pEWsOVZB^##NG
z=;$Wxs49QZT7Sq9qjsiFr!*BZ;>)UQOE}w=ctsZHRvWi+dxTx&OZoGo<Zidp0OW)n
zQ0-rwgAxnBz2b85DnAL%f0c>*jGX8PR3OgBkN9aXP4K(_?D)K!N})Mh>j~xEcc@%p
zQ@2(xxd0!?#jg$zBHlih;>|PLWa*m;b-BIf)W<l1epufF#(hSrJh-xac>L14OQQYd
z+)nwc{%663g}O=~bU?m*{xU@tZk<x%01r1646d1sFQWH`nv#lGlt;WJTXsqt<Z3;`
zi4{gU^5b$fvxJtj)%>o&P-U!d*NB4CVA6s1!PghcqL6ejNtTF~X6@fM&zC0>WeDe?
zS`}qeI-7|TnX|{lAs6dI{IJC_c0zn%l?T7^0%r2rUmQUPkp1sk+hTYn(_aV!ms#G<
zEtD`qRX?w%83V2LK2I<q=SMM08TWR1^E@PImKWTjR?loqMgCpQtveU$<>RT95)lUw
z`PvbwL+AqS*ZsEb$q&M9(85!~&HLDGyE%w0xPq4LgJzDc(xb}Y_sUSWBDlF0ESJl2
zvAQ{i6hc^h`lJ>e?P1?pU5*08W?YJQZhX$eopPY0q{8J!epvX2Q?aU#JeYT||EKp8
zx{_7k@@jW<T9*TF@XaY2+n(>{dE{Y7#aY`bT-+;D3zyn`X2mHC#QyRQdu3DXuz=$f
zo^r!-vVima)g@EgXYare&P~}rvDqjKsIl4(t&0)!@1Y4H4gq12AvJ(>e~)o}*DAwG
zo(W)k&Yj}-j$SUOBgII8Fgni3f|d2b1H66`iokKcKTJB|-?`;}605h&MtjW-l)fIh
zmy@nYFEyUlgypGFGE;4T?%m5Cc%r=r*tZyy%<)HZ=<w3(+0io2iVOEy$Ngy#q@Z{_
z*M-dTcegt%f>8_Q=g0tw7IQWKLA2J(G%7MwRkI^wGoJOtaB8$5<B$E8i*07z_i?l~
z$wyo+XUaET1wW#a<oP-+s(QC4_~xc^X*8e&Bz89IzKCT=LSZC$$5<Eg<8GPze-uES
ze#z5wh9B%W##B56$N1jjXULxM%H&Vh+nL5*Iemhy+X7`8ur}aXAbwt1HrRwyps0#K
z1D9)UKuvDR{PxHXU)o<FXb|CTE3r-#;G;iI!AT2ydJ5|1+i__Eg&vK}kQ;CZh--4|
zwYG1PyytkRojRqO9=kAcUT3Q>BSrJKXYBR-qfI7w3`S=J&1f5~RzcXG)I(iMNoQ!W
zIFx5a_~*3pZ8G4+nWKtNK*(ONANW0J$qSPH`71wzA`V$munK{Te-AB7&h?_kuI4fR
z#NtZLl<+Q}!H&wON`sdk&N%d7r{cx%Wh`EEzke$#v5Zzh=XM6$Sb6vp<%?<e_zm2%
zKD4!LM)_dxx}XP9?u&V?vmY0BoXAmU5xLmkW@|#sAVB55nkyT3g2JA%fa}wnv%zVU
z@^Va@1uNO`I(5DjbU}!YVC!nyZ*u~|mO}39l!LFm)c+;hk8MqG?bDhnexGp??~@M<
zqp>(&o5V%FC~vH3VzmsANTs);1Z3SxjM@*xNgy8FHlul>13o#Lp2fi*9qQ36UgM2D
z|HjwW0nJX^jDe0;q8;(Id`?4I{3wSqIW{Avh1S!-XZnGjn9t<eSWr?P!u1sCMWmIP
zkzPlXy9Z_zI?|$~IF2NI#pCQen`Qlu*@9T*hsm-Iea-TMv(Y<*>*D!Y+e_^2<Tk=G
zr*+Jw(rNJ-hMb-vN?QZr+J&yh2#s5rZL3rQYCm$R-!Qge4hWowpt(>uYT*`(QWBnP
zwB)w5du@zc+g@9;%LVSsQA;Y!b^CH9a$`tQNv9S=Y_i&ZSbbm;`>R7yLGI=jeb>C$
zgx;GdcFR*I-bQG6F|NnaQN73K0*l`-t@qkysL-;ci`8pHY9XLnAZu&+&(mTvh*?0B
z6C*H54N>95*H7u{pONa>$Y{vYvQjGun|>xc)c;S!Imq`}{U=KN=-yp8=cHC_y^PAw
zTywKz=?X`VLOyDV3clJ`svdZPXp5kygn_;w`&7u@@clE&**T2&d|{lrAD%BQEB0+F
zJ8WnowQ+2o!%@~`FG2LUjA8WD7X&<`oa$eQXxUJs+;X<F{fI{sBA-N{p!K_$bJFG~
zwq$OFo(o;gS2W9E#69RlFr;|#ah6SrAMRtpWK+WMPw4910BDLiW4J5Hz82rAKMG*&
z{Hnp8jyTv)1&KD!CMK$`02g)3hbw0df=;mgOGu)Jp6lP2ZnqbAyO{eqM%J#dkd;99
z<%WJN6a&2%`vZwf<#?#6uSlEQD~&@VM6MR<Goq%hI12$Gu{|GtC`2t&CsG}?M*LXy
zPGJ054-7$)3&v$rnMj@g>Y0lxgQG}hxa9&#0eQeHC*Bz4>dIiJ%yto>T;J7M-Oo8t
z)t~h<76K=n5cz1i#0rtAos-KjZ+%#^$Krns661o6UX<6GBMqx_`>66XXnBGZdT%RX
zG5r3f<8J&~Kk%V;7LDxQbHkfAExKLw#(7nV1xy6~S3EJkE2O7aQxXjij3{050t8M*
ziLoji#hc@%UH=3OV986PjqcQw)6YWC<sgicb)NieYIe6?Z0qGHFbCNZe_xT!YgBie
zSTYYP-U%+=M(Bb0I-jbr^xFM32gG{_!B%;=PCXNlt?v+3Sue*dro-T-x77!0XO6E_
z);_@>&QtR}x3_nGK3g;Zi6ee1!K{8&o%(6u_u{D%&I8+Cfc7_#Ch{#TTZu!Tt2;i7
z!(lhXYhHeYd9T+w$htJ5U8BF4*Fh6J_eWGDwy^n$7u<1;XIJn52?pQ#DbYvq;?Yf%
zfA{)}n<~u9hNxl}541bL!tc)A3B44G5)s&J01DOSzWovo%UYJhZt7WCSacr7VvGCe
z>s>$D5`C=IFeoVhE`q5if;)sOk%fh2$0gvTd}qh2yB0Hh-GRi0h32|1C(1Esv~V-;
zS6GH+5%+#O;@e<XlRmopW*w`8+q#_1Z`}3}NqERVLSK(d+j|%C3YxLE^Z&Q^Ez?}L
z7gJfkU=>8V(4e(<#Xu&PJL%ARfvE!sm<M_N&28v^X8^ZIr3se4D#N-xaOofP$8~7H
zv|!1xu>2~ExATSGM7OlOtUdu5ewkwv$IUKM2}Bfc%Pa`!eZ0V>5TDP3#P22y+8&C$
z&(v_Kp(ALNSf`~mq!35D%IDV6lO{l%XFARQ{K=4$9~_K1z9gRa_^GV+Bh+sYXd3Rq
zhQxoevoe{gOh{Vnj~ekh*+c9^9<+vds(?<6owAg##!aMYVz(L!g?CayzwEbFGcDO_
zEsVT&es%sDEBWZ^l<T(g^aSd?ir#NJx$8GykGBatj$^hIzZ<f<mZ>kW^5fAiAbwW(
z7uR7fz^i(gb<+KFUAFC{#JkKIfDwG=gP`!-^m~_kNQX-z&K8SEY@yAzEzNTEj9BV^
z_TP@<fCynm+o#6OojMJKDnujJOk+Mk97XgHqoFNm+i+I<cvGp3qGT{e+?lV>E1B(K
z)>}XpKZQ{4#<S~C^!>JGUs%s=#FzVC73PHQ-=%V4Rjt3Nwld26zB{IW2JDz$wcY#C
zX|(ZBCS03k+8|#{sQ<VfZPzi^WFDr&h4j0ZkneyAAfR+EcRU9C?CqE4SKms=(i$#Q
zT40t?d5w$juUJ`==m%H~PW;{V^{4c0ux|JQ8@MrepGFB;I9j^=F-KOoyxCJ3f|upx
zR$V5E=j)32*?Ra&I}f}D93L5!#a_<s)LxQ@pmT?ZrO+$4+lqJcG~t_Rp`MtPuvKAf
z?abe8`uF9@f$;tD-R-JEqjPn;YaT0H70H1gl>T9;i=!kzaS6-!Y1<(iGigcw+##}*
z7^kVlH-Pv{;cKdxBS;vlMYys~Bt2zh7hk|KBKPn`(b7_%?}-fXs0J&vkrMy5^8WM4
zznhMxlTXZK`0-`56g_9J<<kmM=a<uYZ7r^T1}XXEK3c=+DcgN{Vem$~xAfr?>;Cly
z!srkw)QXA8(~|uL|93cGM0vSKPYef`dizB^(U}MTgiGLZ;NpLOb=$tm3?GGR3Cj8K
zl)6?X8R(XgWPti80g~89c0z;3u0n`@lw!JXu-T3Uyc2WQ<*|SJ6}c%dTM=4DjvcFZ
z?~Xe>1<WSX&^ZZ#Pye(v?$18@O#aD>u_0a=A-xt^va32gB=su)lZ@JIw0C-p{cApg
zE~RYqJ(1Y7{4og=b2=HDpcR?78stMSjWc+;)+?s33S7o6fSC}qhcdh5h}bovfcl>X
zSLInEHL)-H*(NJddHSb@R!S`&%3+=Sm$vL=dUN8Gh5^HTm*i7&JkObJzy5lX((v!m
zTOj$!3#_$+NYo&(e!G)6nngyem|u+ZGnGaR=3IzA@Za_KTcsWTI$>lKSI_uRojTQ_
zP^zwfbjQRB1oXsY#^*eW0y&TKmpEhCcuJ4#jtV*1>iC0MMAh4~9u27Yc1#vAQ>9Q?
zhJhTm9djsqW<L63m0!F432FEL?iNW)<oMLUwl%2fS^v4NDy(q?F^$^wvwg+TOLyHx
zV~8lrrlut@LZKJ1IR5NXhWC4=xw&pnxj>S_Ctucq?)%&O<tE+bp#0aB$w95l{%Fk3
z{v4*{{$^Mo+m0S~-|uX9<^DuYgU6(q{+l)+(kzj)+<2<*Cx$j+PHbkZPmXg~V~vYO
zn$u8>aQcNWsMIq{`DG`wX*uHNYOecp9X3p7jY)hnL-G3I?@^Y%clDCGO`#(#dB2)l
zU{72CmY}DY%#LssK2>QxwV2HjFx8520e`MmW2*kfI$TV*@N(=B20gB19(_eDlhCCQ
zygK_T;}Zq$=){PW<Qq!zl`!cMteu#5=?h06La)j4{>rD_+QD<MbHuFc64VpP*||HP
z|LPL8ve?vg@bV3Az0Pe!x88Wt8vGrENR=L7;{#*Kj?<NU*KICtwEszWcn-A}3WtZf
zG?!nW4dSMuBQCC4?+(`!Y#3)KY-+KSd6N{6%(O4z@Pm7j^xM{#d{3ah2X>5h7n<Vl
zhvk<N2p>O4o!1sOjmuxg9JH-WV%#XOC*I4!wC+;W?d1xZi6gd<YmAd%PK0uIr5OnC
zTzxrv?Y`LTVJ%OZFKtEeJvY$2t&HW<U$(I{ZhnL%(r%kpZ`fOvjY6HC(&hST=4cHA
z$v=`&9#QsY1#(+ylg-NABYf#0Pc<QX4$aa}s-f~k$J@ku1>U`&mffCj!@`faxkH@t
zue{&9tqt`%45sghU_PpMX!1JlO~L_suyewT%gA3K?<D(yx8=)B58LE$bc(T%4O)oD
zTcSqkt5f<OP#+Svm&2Y@qOdR#abz107qty(Y59HSbAPkzviywy&AKa#j$k46#(KtU
zQ<BebHyQ-dE1gx1Vt<JIUWD+G#b01Lx={P@|EOSlr;DpMv&%E?!7UxvTWRe6`8F?K
z#;R_cZ@Zvx&wXZU4xGZTj?QX<o5gaBu)YmCH(Uir3P6e7Ui25N-Xz%E<=)>=U>31w
zug{B3t*^?Z^FriuZpBn0jOwcoaqQuRqF#bg!z62UW16XElRd$NC$=}+X8e5%)+PyW
zB*lUYefYQfH-y~6Qwd-6U8A&_L`R;v)OsE!nV74b_KI~GKn#?WT>3ZH30I?u|6*a-
zF8>ajfaQaLT1b``mu8PBfJ-Qv)VggY@;#gz8z#ln{_h=Opzp-T28~0T;Z&T7%w5%~
zMs*aH=^U<vI!Ke^%hpeU#oJZb^Hvo2-|rr8C)nI>JE3ZYo2RCl7mM${pQ6I&+^Wk<
zd}&B~8r(cP4=R_f|7EF^G@M}<^MV`{5_v$@yU9KrG`#)rC0(mvu9&dm$oW~w^~L4Z
zi8{(z1g{FA=GIobcn}e~;9=@+*@BCML*@NQa~OX;U(@A9XB713EV0I5%%(Yo9knd|
zVGkwf2^%^u-vwlQ>e*48^zGj<o+>!gZ+6X=Rea-h8R0`2L`e-j;i}h-UVgNCo+_~N
zgODYY!-e32jG$b7WIpk>3a${?E!t%#+y~)Ak_So)N{+`&1uWgk&EhDsYZ<iJut-1$
z*2CZa%Mr<Z&7p*(WLR&MzPpj)eP+n9$$&S>@KjuVXmfyX8k;~r!^#}4wq%R+{EJ)Y
zq)0cGs+>+{tw~ShraI3WT4W|jserQ;jQfdK;I2o!<7;5p{U)3qn;lXe6BF*fo$#z_
zUTkx(_*$sH9S|A;+YXj3G1wo!+S&?P^xQ{X3$xxI?|K4PEh`dbz_uM8VVHbbchPQw
zQ;~Gv)7`fufp9BR&N_0)m0Oub7TzY6wWKL8-L$-5QU7t&X~zuv+dxki;Qj=Cch`N(
z+I@Xi8*y{AetUw-owDhw<jT=tm>JR$)yd4n-*sgm&-i+8YN67xA|DekQGD<9?FT-H
zN-pOjTUy~kAF?+Z6il1ZLzI)ne3Kw5x0R)vG=y8*c$XI?ZR-q|!7xKjiSYXS_~f1l
zxHx|xdN;Dy-#3Da#HN&hS!{dC!bWf0C7;xH+n<}sNCPhGW*WT(<;SV<aJQ5)BDF8h
zb)B`WDkDMpSRC)#%L=_LN*pv<4q&f}RtZOh!RqRyQrnO_3F=!~CuPN~3cv-*Mlme>
zm*2kY@D<<V;1LT7Hf-+Mn+CvCVT@U&zOlIAB(U#F3ng~{7Yw|ui-~c0e~I6yVMoa}
z;GaKrV#RvB^(<;GxAxi{)%Coky%F1&2xyP!?38P~PQeuQs~?tp{!dm)asZbFtep7W
z*zvP;ySAN?-gDg1ZCE<;yP@Q7|GUoso-@j>jz!7if{?!`?&@qP;qrqY17~yMW;k^j
zVZdi~IVv6_SpcYrGiBpi3VMhs)!}8}){*<_dij1RX93KpVLl7V{TpfK)4RjaAd+}=
z;S3QewR&#Aq%dOGI&u67#J+lCCfI4Q$wQc(M)5#>Q$I24fc0rU9=UWzT7>A3f^2`P
zV8eCDs$#^&6*gy`lb{XT`Lir&Y@^77n{I&}2r%IVu0WAi(Kd0^v}!(+DE2u;zi@Zk
z{10Py3MI+Nh6?YDcQrq5H#dV-CJOGH{B93)Dn`3M=8MxN1g?%QZ*3G-YQq<&CBP$u
z?nsU(OO|~B>?WJDgnJ~tuu23^&IIdWO_TYBd}CnG=aN{L*ddvaeL`R58$KD^`u+Fi
zY5VFeu-$!+xxZSy!>lf&ry74QHe<84ovkYc{P8$h26mztW4*Xe_~=@spoW>&)9SN1
z{`-edKH~Iqx(C(`TcN}s>QtMXo)NXMMx{Ib&F@3;=Ea(ks8^hQ;jYY=ARcodUxSAI
zwH1d!LrwkhiB)N_-7q<lPn3r28Loy72I2j{mt3(K3+<QI+=+$z4bz&-W?Gnl_S%yM
zo$&_2^4WX@A<`OUxby^H3ghAGa%7<G11uW_2~u%FTa8K0Ru>~00|dTnUVNDSc7b30
zdT6~*P9NoGSlsmzjaF0)zc~KPu3)=p<-q(WEN6i^vn$K!1`Rw=xvuGK<y)ePe?beJ
zWOWBO1mUQ-noG{?L4)|u4hovFf6eRb>vnN}l_$a%4-fZz-#aCpt<28IPgggGhu?jT
zxb08`X09Jr*SAxCxKV8AG|hUWp2@up2B&J6Y><GeokL%@Z;mUZ-><%zxR^Bk`6d8f
zUe2_EYiexb%<^dcFzhH%5J_|WlZ+5>UVbg%VUkRFSOS2Rh-hI8{`S=$N!dTsx6v-)
zHak8w^ZgqZrqURs?&ZeD^DV|9w3*D#^Ifm-qfQ1`WArl8SYpS@az5E#+JUCm!TkoW
zzfxQYRO#T#xuXS(o|=Muk+lvDep4EcU%5e3wa9hqR@W8IRl`PQXYa7hlvEBh*yk_g
zF&AX&Rc{y*NqT6cxbmqQ-v$MKP9|ca=w5jqwLb{js1!gakIn}xZgNX0Skry`xV9V(
z#{a1Cx8t?R@~lrp>)p|<enNr+Jyhd~L8QCo_6&{Fgkbz1=+26+N*k@p(X=^uqsMyA
z&~|B-tJbD5vzxq4Q&cR0J_Z4(>Nzwoevw{RnHgH{T1B6RvE<6hTNg(JhunpAbqFrn
zJJCp)1!T}R6xS2rr3*wal~q)TztZ_YzH($5^6&&|V^Y^vUxiB&sr2Y#<`=*5Lx~vC
zjl2s#X6(24ei{3OJq%>W&45GNZ;vj5c&=`4!c<>C^lfVhllH2W@X`SKnoI!|G+%g}
z!OZ5BaJPSYV`+`AvpSh|%R`NUnN|wQ1xK>I^wD&=y57Q12tD|hGX=Av+jy{B=bk)@
zY)_Dwseu}#m7=Ex!rj=eFLwqp%5_7<l0{y#zIgm%t}Hbu3wOX=>5-Np-dhA7#OU9&
zvyqFtK!rJxT<VMD3_O5?-Ol9DOM~wrf)%QT8gF=6f<QX3rapN`g-c6gX^ImsaoWEe
zF?JDF9C34c?AKZ|nA8;Ht15M**2(Q*ytLUIvwHwHuN7yk+v7D|W@3_Fvnlx?km5AX
zP|@ELna)MC?hR;Y<+z(~1zR(S{2lUW)<5w#e`TKAA++yjt^kC;R0Q~@rdkBENedou
zJJuwz%&1c&vWfZg`B`hR|M#_?Kl<Y%{D-ko0(XB0tA*t%{&*BbHQm21fG?aVZ>2+-
z{)^5Jam3n0zD#&oQ(78DOxI>w7Fh`=C-<zv4gcdqft(K&%>vWywV^(@GKq;LXS^HU
zTMAHjUH|vTs&Ifm+*RrWXdu9zGCze4(pZuBTLVtOO@L409&I^H;4hxi3_?~Yx$Z^y
z6q7#PJbucW7No)=jm``MfqEo2`*y<WYA$!j$IH`f;!%e|K*N}hm~-lzGNMEmL%zNE
z7Yb>Q9{1&mdh1X<N+EfFP(^n<r!2*e&#e;cK5jc7mkO{mzMd7_ds)GZC5tR}V5`>&
zC#Vh~=(DN}Gk@(icKVxS>V3f%xpuDB^xG_t6P4VLjdx`x$h(rwA814^g?#Swdf&;F
zAxM25#hSV-g+>GS37*~IfMTklKMt?xPmk5r-1ZZvh^rgO2#XZ@sK}*E4-5kH@kd_X
ziAGu_15pJQp^~K@gT!k%d|}^^9%|58UddD^<CgNL|8=;FbhCY#L=iD>oGMeZPus)j
z4Iy&vmDJFrdLN~*Ssb#O)-OmAR>dILwl5zyr)jo^GHPTX7Rj7I<tu8BlP`qIIms9I
z#wN@*6Ror+7xzT}di2m5F%FBCnc>Ynv1QW&5c~1<`=R*a=Kh`MoWnv-u6uYtlQ<V_
zcClXd*E}cFFR3Y+5(ch!`#V-Tw5wjiJ$(FgR6FUJbT^tqw0;>cQrQ^M7bHCu>v0#c
zw5iO0^j`9DPgEX`yZLseFRRhV$>{2~;-`4te_k+OrOR^XK*u_=bG3s*n#v!y=^$t&
z0V90xhisxVK7>T<5IxazK7T$jnhKc95Bp4Al6*?#HP=o@DH=wQQnYixoZbe38t&4Y
zaWcls>ZsYt4}RG=9Yz711MHP*D(+dN82VY8{(_fh9If0_{Df+aKFJRmNB;(_bkLpO
z`O;TK6PMmmQu#;9i!AH8HEM~7gKAw>042`<2KPI1GOIp+Zu@9_DPS17sGBSdHpzS2
zgRjN=r>m}VIuG9Up>Ws{Yru`<j{RI}ERW}AYH%6?D=85g-f@#6C*oG6s!K@{{nFId
z9UuDHSm&8UyZ4uqop_o%9^}FO$)zkQ{`*YR(S14*ogM#iF&08Dr5EHkzta?B^?{Q3
zh>O$p+>*Py-w9|%q^SuzqG&;&W&NOu<5??F6e*n2BP;~3;)P-D2T@9@KDerh9PfZt
z|KA8=|6=@DTHGjFV~%gwq}5>i?Zxu-wc_n=AaGNzI3RlFK$f%x<z4eod^6Z4gdcB#
zPn84KtNWLy<SYi_hHyBa7C}Et1Ljdx^0>)>njRt4hq!XYJ>5y;P*L;mGZ}?#YT^$O
zq%1reC!Angwf+Ei8LJw$>P5*#Eq|rgH3o_AwUr_#bwMxGDDn*+xjp~BOG;pypF)gV
zNt^a^{A&iM7#?_IL^(so^YU_QkYg>^=>R&+o&zO<^(5A%@I6SQCso=s0=8<j!{^$)
z330xlV2;ayK^nK-I@Z{pR)Wx3ILVKI!1B8~;mSFuE+nccNn@h$wcFzJCd+q3pm-%6
zT8?{X_0Cpnw@1E9rA+PrejvKQMyI=Ft-Lm6jE1Gz9g`<wy2IBFe?myOv|EI*c$R{c
zHL5b&5iZ)@XFtAhvN0!u0J5*ayi^1mn<2@OzrUB!5=V6m`@iV<Kbp?+p~?4c;~NbU
z8|4U<7Le|45D96N7!A_h0yiMSC;<gVh=g=X!{`?2FEJXFl5UWG_IvT{FW8IwzOMT`
z&d+fix-4#=MJX(mvcDG(WLlj!=BqJQotgeXXa2_-Tdsd4y2N*FEQ2j{Dj=vNz*H9W
zl3B@$Xc*DHsl51bX6VOEjvG4ju{rMO{N;@kn2&;8)P6LU3ADxJNnpxZkIuU69aflT
z`yLw&Sp+6CJLO{1I0b#+8gSm2)0*qnu8QwFex|Ro|87r`@~1tI&I@ETw*4LPZJVhS
zPa)KX;Oz=Wc!|M^bD02-`*$&|A%Py-<r8BT^|_er)v!C>%cO?RFyo}OQ)+BxixnRp
z&nB~~GV40W4k{HgQ0X7`^{p{21ivp_^;oKn%NF>=k?(3jgc=;$+S=0MUC4V%F-x7o
ze4?V%`?gmaYku#+Pyiu!SG*L6j^aTes^E;*<7*8#(=V9`Ix#t@I%z_Q(u@1g4lzLG
z;lb;*o;44tv&n|muZ_Fm%0%STv(vDaG_$st+<225Ln$Q~yVzDwqr0_Va{(SFEbYes
z2XHxXP_Vw3Cqeb!MaMkOp6rL4xA>zRk&yfYKK4>Gi_oGDCJ>0CZ+1Aqk%dxhX^w<)
zQHN%4)ACBf_f$)xf^8E7)8oSvP2}J#clELsCXi9aW*65w-TSDbeZDWZ4i9&j`0Tap
zx$|BnLIQ~AS5`RCwd&SXw;~9p&0_@j0Z0Bp9V_v0s~@(0^VY;#$g7~N!T^f%tdE7~
zr;ZjLlDkgLMp6D9OW%z<i8C6;%n{jHh?5tT`GXr>t&h3#bn`>g2d_4^x5#AoeG_^P
zK*8BsXN2Y<>4mnaVdPL?y3I0!I#&CkUr*xmYNye%WaDK<&sECFalevS`szapfWmmL
zoITugHPn)sy_uBp#o2*FU4_U@=|6nn(fEH!JLNst@6zfFw(Z#L^7F!}%Rf2nGF=p+
zqjXd?Tue#Ne#ySiy1I^<vwL)YhQ4-4FHQeIl_CNkz?a#nSi!;aR}b`3wXe~S@&!L>
zh{$~3<~Pg#Ur*1ryA+M~C~kYJDX?Ww%7hDE6d%w?0j11SE8A?&cNa?}_A34P<K1W?
zD9nuFulk`7<C?W9<#u-Tq7kh7#R~HQQR%c^molw^jZYywu{r?NW7sm@a;Yf#?j_&g
z15dT(&NLRZx{u1P%6}6nr^>oJiF5$=P}R>11JVo612u|rcOn0_H#UX_fTFFjcij00
zk?-lU#372R8aaLYZJZx<GR+$T_TpAnUFr@(-WISoG1Z{RI^)A9{A0k}p#qLXKX(6!
zCJeZB1e_k+tu@iTPVnn#=dG;_7&Ljs&-9jqI!%`t9>+#<ULlD_OMqaE1#&nIcG*z6
z>UWFShNIMCcm}-O+7*=QK+BUF<1yFPc)C*6d})pKVJ5pmrq%vZtV#=gJ^g|EuV;}4
z!36o-5nYEYmCm*l6&cC%YVT{c59!9{D=*yf@-6;^@Yw50-%*Y$5Pxb*H8ma4RtkzQ
zgXMv(r|VvI?T{HuH>Qd8C0of+2yAq6c=NxF6MA}ynUQ7rGk)cX0}(tGx@J7+Q=e~0
zt9}b&Jc~aXq8}pfX<##I&%mXAM{bu*g*B<C=jDD!-H*nI!qD6Yk0$d~^(51hJ=?wl
zS64kZm%vU`3!HxW;6pUe7E<OYMw^%%q8&$gO))|w@6A0{YK*~G1w4s!AC(-U7%g#P
zh}n$e2&@qNY)qR#3fkwmZ<C}J`DOXDRpzi>EN)erWt#vLgouu(`daWvJ2wDgXy6ls
zJKjoLyD7Pw`qWHDh-k{`N>{cIWwL8?Y52+5i_8#LL6kpmq*}pyQoXMK%EKVKHnVFu
z?smA(SR`QZ-N;E7!h!zJTF>_JK3!gui_wVZM8FplLc@0esa^QN<vUYZQ4vAysBus<
zR<-sTPf7YaEa8%`lk`(Lfz#F>_n;ZVbXYNP-qUj_<xfnw;|c^uhBzc~@SITljCulg
z#&ES$6xGs<kJ+ZJ33wvw)BW+5(_2x*rQc;lYo%wK0{XK0Q9(I+``R-=-mWGGSLXH~
zrms#m==vYPLZenGurjlS3R9xGx5yNoN;IqJ_mDB)gt(IKg(i9p+MXudOiP(bZ{r4O
z(-2~Fe1N2`jkf{sAMW?Jk4JvrA0KTlTn@D8$zH6)cVwq-Uu_eDd3cdNfp&F2DY2U8
z$KCv8v$30{YxrqbBEvs@nzDz2zBH7*%}vcvMWmQ5>^7!OvT{B_a~y@NF8imG>NQK1
zQeL_$Lc54Yu?$)#ZUsxQu40lsQ=*t43d$c_``E0`)~%hVzeHvlLBgT5@+8)Z2#$Wr
zt|{Ko2JL_xv%FN@lhik3buT`aD<R$9fEfVUkW0*{U-fX;0X(3AyNv-L!DdpT&2%e4
zix;`uU`r9)_ij;RYKzF~?|p97t=m7;(~F0JNxR$s>^du6m3GXV;dj3s5MWY}gKGpQ
zObTtan(<&Xsvz3PmbImjxZ{1x?7bzKs_dKJUe`y_x<;wmHbU<OObg3eRcCbN!T@aL
z!d}QLrc@B6+%G)cpX+?2H<t7_@+5z|(Yw<DB>G!IIClIvP_VEs6$2xM8kA!O!iVkm
zU5VH(uhdl``fYqP)%4p6HMBBOF-I{wZ77r?uatxKZ9J!7{X#MtzjPdh<AkqPTMcqE
z%_-y&C6_X6tIQPmx3#xscs(sN2b83_PXC_wD5aYv#5%rr1s>e}gTxA5`9d*~{Ak;j
z;y+~$<o@y>2|5iW`+k|#EcB7O8vY)rECn&;>8qqMj}^{xnz$@z>U1d_QA+0zoo?K2
z5XXdl)#<YzM7$+o-EuLRF&O?c?x!fKsmC*y!TG%4$HsY~-^~E}2yJj?2mEjjSh86p
zq1RSUdlfT0N#1Q>n8v$n{%~@+#fxk;C@uQ_BEG7?PA#(6`!p9v@i3zE&_=<Ej2%<j
z6{5MA_QPhi-BKF|jKD^jNa!#iFwZ^ZzmdCpvK8rzs1UzBt%QM_D7_B4q(M1oGU4WK
zU4>whbbGU`EGBF=nib6nl<)7)BPgt_n3R0vDVPlwu{7#cRT8p<6WJJ?x#JAA=zTxX
zFh!b7TpF)UG1IPMe76PCHj$Ak+77D4y>X#)mPQpcZ+(QeXXsUYX7$sF=Q$zR#i_L#
z{_kst*<Zg|oqzf9r}JBgKBFXWxMHqaEmTMT#b}SVb=c>|uV|)p0K1QeU*P?-3@i#&
zYHr-nUTsDC!-J<CH=FmFetmWJ9L7r2$|hb$4K;QnxWhlU)!U=#XS`JOHo)WpD}_%x
zehW36NYOg$`vrfLz3uox`mTBoyVxPA)n$Qt_K6J3*y+{P)#>)Z((#!H16yd=(bP+s
z1HZ#{8Il@IaU1LN7XWfw{F^L(9Vqb{9JP!}!v?9#tEqB$#tb_;!2aX&UVA-Vqbf_f
zg|yIj{bCYoQfC(I!Qi5j&=_FT()RU*T*wR-KY5N3MPx=F1f-GWWY2!T{9yqJWxmdH
z7l2vR61nk7*U_&0Ag3VCzteMm)A%s%hz3pjOoB8t=_ASFfRG*LoT7e%|8N-dplN9g
z3M+m^llQ-W*R7ECKyGB(0;EE&JOcv*(>HyLZ7JI3Oy2~_QBn0e)*eu0Yq5bq5Uko3
z3x6huEEWbv=%0^GJ^#P|&lsSY@rz4INI+yLo66zidZzB$-fthuE=li2!SPFb|Gr~N
zUF8S?mc%X;$A+7oPo**)c2Jdv3IwN8$glgmScUDh-=iFi(Di3`pqzyU(BB4ETMEn^
zzr~yV4fuBxbyZU$Cwx;XsSMPcyr=<#WJJBK?W;7fYyDVdblv>RqZfJbtcA4hK@mSI
zZ@WTOi-WuGVEth$Z7nyO3orlL1GED@5J})uG`~9!O{dkW*POl@Lh9h!1AQ4hZ<l@W
zc6UTWv&*y=O&O}Hgy&jOQkrA2)35}>d~0+p!{2{gHRK=_#>VO@b9?!L^RCxzhDnX+
zt)giGrE-VaczSnO5<xl1My!~swm3VL<j4M+3%#N&z1eHL%iLqhkD+PN`hx@P5yQY#
zK{4KlyqU(x((IC9$%%1%MuT>NbAZW=504?A1wj~z9x+Fs-Br<Of?8PA<lXV{{((2L
zpy?cHk7ozgXMj?c;BY$|K`I^#fRywUtZ~K+y$`=vF){@8>3PV%;|^^85BB4)YvTwf
zZs)+?6pW5DC|)%b6MEpAhif)!UI|P#C#TTNC2P*w#7!MW-H7EdW^4;oAbHn#t?5xR
z<1mH!{esx;d<#|vK0b;881w^{C!Za05?ELEYz@{js^-xO*<GD-NYw5s_~k6vI!(LJ
z<PU!ik}*?J62vMGEc1PoQyLSlAUt*~Uw}T>{lbNNWN_wvqk8wzp`{zX4BT<DxirB*
z+CxFiO>lj?(72yr#!#a}^A-Jg6K5Q}h_%@#S~rtRAuzoiF`v4X%LKBhqJgyyPvK1O
z!M;x7dLm*HN>3dHbomtbB#Xgx7v5}RQBQ$_!NMoR`fPbNos1NppEULJmpQqc<>;Z?
zo|O=Kl<b~gTeOv&6&o`X(A`>+KG_<Z>H|mmfNLuc9cc+`E7W<+--RC;;7c>%>X@k8
zS&GayY)lr7Z6Vb={pXj=M<*qQyv<GvITO{7#Vm`}^L3x+=s0~fCy3xl;LL-+WW;0f
z{ZApwey0A<kq8r3ST!ewSid434h#yd(+UD;wFd5&Zla<t$vn`2&zc;x$cIQ#;tA;j
z$~Wl-b|NKcF)9J)1vSX-jcQS(h3BpRQ4}>e4lc*S;^bzk4LP;f2n|z~?;_xD3YKAn
zHYad22$m|Zv{nX`b1}b>%Vc!4_7bR#akj8s=+r0VuW=yMU_1v^8gZH|H{@B_806zK
zx)ZyUPh8;1#b*FDDj&X6+pfa&R~a0w@GLZFvCrVT*oA+j!hMFCou5KU2|DHEpo;u<
z2ABR{S}~c40fDbtcu1jfO$j9ls8Ji5igIy1o2Pyri<W5lIQRe&Dn(dj9isRLhM&X5
zW<?e0X;-Xx7BWUbo=Z*(g}i0ZVtE*O9eB)~)Y(KGd=D(WAqXS>+M&}^sR75^o}w^%
z1BMvzP3;D?zaBdce1eY$Nz|Qb=6W+4fb7quOYiaQ3W@ug_@4e!QilHA-<ZHquZWD#
z6NiXL)xfBM1fj<Z-%&spS)vmpUnr{^7XOw6^(+0fH-33hZ(d>P9J6CX<=mEC$_|N+
zDWa9&s027T8484WW78QeEl4$Hr-FP`OF5O<S1Ud>bNjmw&%dBRrur+H??=ihrE!+;
zW8uDeVz4ZY#osTdET<G}_#$~GivjhPI1y|Avgf#I;aTJqKfXm|^528Hn~SBTz_v6>
z?toEUVrfpLivMis@(35prSIymQldjG<jV+?c?p5C$0<dLWrw~m_u-X5ZN~?n5#-SA
zF_QOiMf&S&$sK~EMDf02lSm&9(AbtBuh-sxJ}B3P&5*~VzaG)+yQJIQWPG_VBMlrX
z&5dQIvF@YkB*syj2rvfjuTG<~Wkx3(Dp<EZdr|8g27zwde!NV<63aIn0U-hw<6XPo
z;TZw6Av#)vG2Zgb61bcCic27!!BFteoWrYwruR1&_wIWKD~nH7mLE@Q{<^&-S9q&#
z)<^}tNmC%`^#%aZxE8GrsjTehY@7w_T#j#J(mJFBg;O$^{#ZF)dS5n1Out`-#6(jX
zDCdob&mb}e3)E2=Q@y`!1<%#NuB=YnsNX=Op?3Yoq%N^Mhr3!y9WfR^c5f#O&xVNm
z9iQNvaF~8-b$U2TDqe}mO0M&(R~&yYh`T2@nTd7F2Y%E|pXT2|w+Ne{Pw*Lh3}Rx&
z9ij&@k*7K@Fzwd8-U^`ACB_QVmVX0(tlK1;b>E1gJ`xFt7G@d!7>6#7Co$TO1F%f}
zX;3M(et6O_M!fN`A0l9{By~hFlQSV;wF<*Jb$eOWT74FLpU{C7Ref=}`q-u7c36%|
ziR;~<5kg}mT2O3+7r7l^#>;<>fVCu}Si<@YCqAEY;WqAbHt2-^9$#~P?&nt{JDTUU
z+~87g#pc-DYWm#BaH!yLKyn=$0u;#Ts`k*LpgnmZEFWm>Hxe1@Wl%H2_pbA@)pq=w
z%Vw3!3`)#QW#U3?<w`Iyh(I`~HBxKoC?WEFhpub6Z|3=9BZ{(?V-M+!x$piNRzt&!
z5%ghvSnnTmRMIdrRSuLl7v}#_cs<0=QkeEf?4l=8H>Ov=jw=8f=Am)80xWIWics+d
zqlVRI%8fxect?T6(3$VFjvpgIj=x^B+7b=x2BbU2Oeb_0i$O#bMV<bk+*B!tXf^|a
zuTM^X_YBM)|F+D|i6sbasu(gR^5gIS$<HJiRzKP$t^Hc5b?KpkGZ1@LpIv)^Q0XMq
zAR@Hi!@R~NGYC^Yk9o>>fv;Yv81;A5ud0OaZ(a~|0O!`~Z1hEe*r%2>T^?KGnY|1A
zX^S2ToCW&G#3U?BInQObaPqgHzv-`A4n{I3{W<_ZjMVS&Kq~=*@URT!*_zD6`+uSb
z5l8XIS)&W??%k!9QVcFy;47Efbn>1c^1nXnVpWqq>bY}EQ;O*TL0ob#x)8}UPegOL
z<jTYNN$YAb*O9)ow$0$%n+o#aF^Dm-elPUq&vP+;r&LG6E&Ya{#-()r8w>6v%VG2H
zX&DnK7<23YV#J+Oh+}Jt_2|GTU0OekIja2;D63;mx0kpNq&aaq)`<l}V(oHH`UC_R
zNAVQdQ0m0mVI=|Z4oPX5`{RzgH`_-wQiB_UDm=Cr#9_;!nHI)CRoyz6TBg=b8w+(^
zPq~2qqkKk)t>;NMu##+UL<X^+>*m?4&1@LyX#8a=50W$Zf)!%zfiUI8#Wa#^B6n{E
zh@*He`1>q7KG=?g+a!Ap4z+Ue%GF4ckOfK%{P1P(f&gKZ%vL-LB2^1@-?Gh89dsvJ
zS_YQx2`I(S>C5T7L=VsoW<z$`AGqCkM*=#jh;fg$wSv8TmJtU>Ispl#oc4&k&>S3A
zKE|Vw1^r#T0itvL$AT}*pUnEk4b4aCwP<wc79UHLMB`Hh&6B#s*5mHY;H4zSX#=ri
zS_=uT<^U$=)(hRYH0}o{PO^HOq@$7Kuy~%tKK5nZ*-q&w$f&R)F@uTSf0n>L(^3B$
zk~S|1g`ctIWw}55<d2VPyQLUnI+bQOfln)y9K^CP6C{J#@BjM=ee~-osq|a^rC){`
z9DP${jChUYy7g8ym6>ad2TfE8|4P{|cKNQY(g6sTJcC(L>u_yp@Xfwo#M1x5DRu)f
z{YFl{?1d?07NwPhob}TR`Kzl^Fuo(cxSJlvt<OU!R})`vbE+Nfn&P){dks&M=9(k5
z#Y_7Or6(O!6E36NFvj2kSq!PRzvRF8QHH}~2OlRWAmC}#Z*SE5fWTi1hE`HbS-NVv
z0FmIKZ_DI%vaT*So+USHZ0N#@pkQ@E5Iup%wrE<dcYNYkF}gH*e{ga05(uK8EjWm~
z<6f9I=xC@G%@_%hcRG1(G1hpbiK~%pB~;6WtFOCftx>?)6l~nN6Z4sgU=uadzjjhl
zYN19z$zEs!QjU{anRNbp1L=<MlmjgSsudh~e3La)Um$;6_Xa%t#mzSZ1~bKq=%<;y
z_-r|#AT+D&!dLc7={S4>o~iZTYT0ks0|R3VaU71%Y!neWCgR!INzSt*j4DvRHJHQ?
zO`8|W;pT|V@Q4MRx@8k^mo;HiH>hLeLseSh;TqkEof)Ksx2i14QNG*|{w@hnY?R&8
z#bTgJ*=5@Gl{WwOx=55mg|taBD{^+*VcItM{%m{W`br`z0I?EuNlNzH93v}T4UI{R
zh4OR!hx#?C9)LlEShc(cu~ZFl!W1+ZNfA?r-iz4{wi7j&VD(6g#%dA@|F6v=gwEd^
zSL4^$@2Loi4HtucAPn`egr5VA_`Z5yGrk!vxKW*GER+V9n7R2BZ)ejqJ<b&+`RS4z
zAX)wbOD`o{H6;=cQO_UtYrzYQKNZvbsoOPyUq{UYrW-RH*K%i%C{4Ap`aaw$yWhbd
z%YkQCr^RNgtQTRWT{mjHgY*5QH906b3@c?KLhWGJ*<dWjZark=;7ARq2rH=%F8KO)
zWOYw4QQ=2;!%NK>Mg|;_5^>*ULiV?V8&?~)@J1LjgD-rH{>bU|b(Q2_dB!cD7vhi<
z)3m0;KjVjm91rI!4;Lpv$CASae|m4P&od07i#ZLwR8bw6tTf;&+Q)3>^h%lJHo3~{
z&CANA4J5a}F1+}UPPrsN1}jf!eRgvbwblsSZjxaL$dx-Z@JH_Tnqe@sn$ijx36tQn
z`m~F3-9;{8)*t9|>48@xHOYxg<)6KLg9s<mngeCTN~UA;R8%VW;B14R8>On~qWs^Z
zl`7nHSnVy_?c~-Kwn{(LU-P~!q_iT|z!6>Wdnq{)C*BK*Rsb*qc5G1$Mfx1dmG!av
zkI!Gm9H7-yO7$<R);_Bl2!W0+zC=&Q$=xw9=T#nJe(G~44I7`q$qK7e@pi-fyaH7V
zDDF%<rnoTtesHLUrP=mI#=sUw{N3^F{sn*G(!GD+O#{&SuyNgu+mAB<iGq<YWQvlR
zTQs+pBZsDe@-d_AZ=(whjB@U3-YJ8P(d~!+f`Xm~<T%W|+E403zTj8qHZ%@TM~c6J
zj4rU|8*jH<L0j6HdZ#eKKtdA&ZpiJx`+q}PZcp23bjRowSHObaZ5v<bM7X+Hxq2VD
zBzG)Pi=^Si7&^_%M9TE>ioHn7u$idYtRZC5Q)SEtmfeAj23w<|S8u>nHf#nW%=RJ&
z`*)qY>$N&o^j5++a%|YOj#9We76q!7guP0{vvTsXYq&UIMVHlR1j5bI{(R;9MFELW
z@o{3A7M4f|RoZ)%z~v%6k24>oJ*787*o{QR@d6+a?R0>HY(6GLBh&AaBgr`Fv)+nd
zOUNRyufxbohp)y_oNAkn_WArdyQ)lX1h?bPqS3Nd*mX4CP6XY|NdxnL9o+M{D+8Mc
zY#^NuXjD8isAs||`!*&A<_PDbOf>8FfLS5HB?x;CdWAid46@bLSV)htvQF%Q^Z_Gy
zk_CQYs<BlhrtcZw2p_5mzTG$_TY2!8)&iJ(&WJz}>$})HMRuj!;1mwN|M_>W#(?`9
zGTGs0A?>dP$v1TGA~~<!c4i3}{hyYY)M6~RqR~%&d&ss2x87X%xpmx}9<JQ~zVioW
zXSmB!O*aF$o3+QLBcmwvOl{0&Xke&HGR9~Qh^K{(w?lCT?LYN%YRZu~NS4+%07t2F
zn*NPtuq(}Dlz3`uz}^+XT{=QAhXm(EEmAOsR#_-@uZq+~<m{}G_^8yfgL>0$7e<=b
zO+AzTyb3E7e+Q{>wbK^QCE9%IQF#sm*_X3Aeqi+_!eyi-Vq{={yZ7J0580)aSA#<W
z<~&GewnA0vYb8qZGH^^P6G2`CK%In_XQN^<!ERga^M!kjfZzA7DfUL1<j(V6N+3Qs
z@b6{Ow_J;u>0s?CwbXzS^x!s*D#Ae2!T0-{f4g`rd=sA^;S!3z>0Vj)SOl50EieR@
zC+E=3_<mTpC4>I^`yZugwVcf;*W1nQ?V+OGjR4W7k%{a|>!CXGs3c8n3jlw}YyI)U
z(AmBJU*c~(ySfZ5if5Lx1LK+G#pz~-6JJ@8obOfPFD<1;PtOAb0*xEy{1};Q(ie#8
z`JtUo6GZNcc{}E+q$3?MVL`ArUHXDsYkC{-=HdC^*5<J3eOgdeOflA9#M4CWS3B!n
z?C}NenL10ZSmkLli|k{8lECqj3X${;1TujW&xb=BdEmYnZ$04LOb}iG^KG$QB*`(f
zLWHTCZCy#gqimN`FsW27WoX*mrf4|pnts!=EXPYFD1x1jAymlFWP_AEFLQ%AqbV!=
zRqgb}&s5zY-DW15ANla?{!ceH*pbOhe<U=`UR<uJs*;Q1u&`~e0<u52$AGgdwLn`P
zyv@Y%y&tlX<Jn_F9k*j0_hZ1-Wf3p$Y3usp?Y?V4`017bB#cmjfWHvN3|(50zP-Qf
zco@ReTC{SJDm^x6frpG>ClF$P8B$&t^;Dx~E6h!BTamp#PFT4+&ITU-6<xMO^;`vx
zEEqP_R4h!M3L?S)D-K)HgNv3sDNVB=c;y#bjZ@U9xbL<%;HhA}@OuP?w|g`ZaAx#~
z9~TFUxsp93{=$I4N_adkU<hO}f%V)<V3jKzG;KTHLlERsR_iUrL156Qf6@>@@CHhi
zBsW_?n&g>;xj-sxpkO&Bofpd3Z{>*TjlmI{dxjkL{W1WX&}yl8VW3@`p~1I@P-UOl
z{ex#uFN&}!Dt2S|LqbB#PVnp*eRH)K$4kWtzE}(zv%JtIx`XeK!e6%VAH0^@6T!wo
zFv3plSbw5okC!YTZf@2xd5B+)@m>vWB}@j%cY8EGoH;a`ZL!riFBx`tJ_$7>2dAl4
z6G!s@$Oi$C4o4bbTJJ@md-2@%!o6sS2qzOQyAa8WW)81%ovKgpR86_wpb6Ayb9d-;
z((-}t>?VVz6A^afAE!9~<c3!^>DO1D-yw~3p_rvXYy}pR#HY55Z=}*830SDXxI*K2
zy-+a~wLAa6eo;cilei-c0vHBuZHLVHHBIOO#~~1}i+i?{YpC^Jp!B~Lk=H(3{_brR
z)R*`6&&~V77IF6R?lMizn`~EI3V#<#g<B`?CyO2#FFp4<SF4bglcpjb&)S$L9xFI^
z6~@kq8_s=(@(^8b^+2qN`NBod;7IatM(aHoH}Q?Dly?u$;!1o5rULmC&q-cjmbmuH
zCKou7uXf-tIk%ifk%N0PCi7MhQ;F<)dO+rkVdT{)B|4Z+Bhxa#zp?af97o~Q`Q5C~
z0p6s~6UJ9)>xm`4jJX=Fp$2M>3c;I~*;O?C8EsP16+@<;Pf?XNut3%wnc&-#yA>df
zcL?Pi@WL_v`KPLnQZR_Th0+h6zPsN-Bf!w5?46X$19~L{v-A|Ook+Ev9QIh$&SZxu
zyRs$e5)(tjERfwHvNG+>H>beO*2=^7!dm8cytGXYNYyTiYgd3*iQa0pQyh7Eb;5j`
z)<}<;G5<|QJu3>3NhRmQqQyUmB+cdtFfph>qDUG`o^WDKA<BLwmwb5a7g>!5=r(JG
zKHwC`yFj5fpOhS_p^C(62T)s3%v)wCSYA4b<Tio!d!}Cv!UdyS8fJ|Bpw=<Ki&IJx
z?_J$URTM-ZdM(0LwV!kmPl@5!h(x$7DD_pW^%5)0UiKXL@AN+zVBy3CrJ<@tqq~K+
zYBokhxEs{?;=qPv&&G@~{}izFgFSuLP}wq?iP&&RHyiD)8%nT7hFt#J9B~bTP5ulN
z$Kd`*^TKV%(g^+DRADWrWvOhEPKPYRv1=kBbe<UrLD{&1=b*{viaTt`-(>h4@#cBW
zBE!2wAxh)g{19^0<INW)M4@&>qc5-%ZPOcEgoR6AS>O@;4Aag3PY2L|F-5|@Fs*h$
zHc1-43i`qk{CmEH5QOUaq#Zba7(LX$v()r`_G2Z6g`)HO`O4BjZ*@L|fr1W<vT6BN
zX7=;ub9G=EGtAh-v2M&{a9Ayg(vfDXLL6MI2E|d(tpFg#lBE4OI1w-@6{6kpvOoJc
z44jqN@5r3(nISmt0PEs@WK~1i{!=U^Ar_V68!F=H;8r@l+5O0ymbIHpw;RUmhiqLN
z6$AbspNDERB1H}c2K>mv-u6{a-g*wAjiAGAvLEuoMIWv9)b`_0z7X$U#|K-4{(0V8
zCQ$vyz+aL*sL#TS=;#EkHskW~^whR_)Ea%*zjW*^b#O7{0~o)Pe|-N!;6&^|$p$WJ
zWmo)k^x@yB&q2zr6z?e)6a4Q<A%jWM{?Jy@R?#Uc*;LI#bV;J2luSrBS5ICK9pAF#
zV+3HTb9)BHwj6#v9hf>2b&f=ZaR|H*osW0fDtnImwBP&E{oe)IO6yHe%hH-<gNx)l
zwt1hn7&RVUZvZyXzH3du&qmd3`dm+5Mf8|r(v+x(z#c@4&72`w4gE;lvw6m*!^V{V
zx0;q<|B;iADg0|A9ww6^zE6Ljx^pvG;Bj{f*1X31AJzZ9c9pC;i{Mf<!m%0rG<ovA
zS*=xwqFB<zaY&!~-@cH7WWmPUJ4$Wjl>DgV-#EVjQ3Ig$I?3hzgrbnff!T`54HS?1
zC?!U>Aju8oHjb#l#bque%_l3Ya~ZB%h<E$n^5UYU7CcN*C`8(F=ZelEB^&5jSlhna
z{+(>470Cx&ogGI#yKfkMQ;P)>Uqhw%d@<FJ-;sYX9+fo6x7(eG8F$5(e%D>~hZ2}p
z6#bS3xV;9z`<Lm4SQ0s4G2O#RZU!cCx80h4d&)QjD!CXIno_gT8be_1?%q8pr79D<
zt+2zv65CIZHE-3=nHQ78p^XVW+s~zhiv}Cfq@75)oF@C-@F<Zodc$|{KV_BctwGXi
zt)K7-2@BXTIUJms#VbfBWTk9>pV`T_-JAhT3{F%LaWF_tMjqmGu?L0sR6r`Pu&0vA
zRQ{*ZyBxRJj&0+lMT^>zcP4@)hLh#Q^2g^OTq~TM!qy7c#@;D1jI&AdDVd+&KxKks
zjY~R$Uhro@`L2X{m2L#nvx84Lsb1-of}J*~7|Ih1B3%9izeb{1<GvWVkD6tKvP=Mf
z`jUpbhGum-ZvCf{5Y7}Cq7>~ytIw7aylAA;T+B1pr$>JPEjcAfDE3RxdM6mmz+?Fh
z*ILyFR|rD^CHq&>%dnPh)cpo8@cuq%#;RE;?^T=251M*TCO$?%elImRnyln7cJ9>H
zxRFF0ToYDD*0<Oc31#;3wEy-B(}k1_!B&V7B0hGfeYU&;oEEvECXYTBbpZdK#_Y9#
zpRJtTjTd}M#0)++mKJ~7cQ+j{m90ZVLtfuR<uJ3|J3G5L##94+bvig_XwxJZE3nvu
zBb?uJpX(<W?=hB~&n6qgX}8f``DJA2;5*@ZPf^=m7($#&yZ&|IAF8tA48Xx~(9wkF
z^CJW?25+oa082fLi7GP43G2o)@`($EgT5NLx!i;o5i-lIu9kD@a2r06zP00#qr?EM
zjgrfkE!YfZn_Ah~-?pv-#hzI2^!5}=M&hDr*gukDG2r!na&M$tX?=gA1>Ex<-k)u4
znfjW4vm*9$!@#JQzLzbRz~xGa-{n8oe4dQZVH2~EfA_NY*KQBwd1)o=l=01Pp<0zp
zW#94o7Uan)-~9+|Hf)G0PS$NO&b${P!(MbJ=o2}TD=sObdSH+P+pM}Pa!q+JoSb_q
zr&`AG<Fa7s>l~>lc@~P1TWR{dRP2;vERgZBRt=`+r_cbu@x{;l3N$LfQ2vc{LklnY
z?>AVr$jhB~gWy{n`J`QG>DJHm%*C$WacV0<pef^N41pS&a34GWTDXE1^h;)aza}-#
z^yjA_1c)a7RX!p#67{RPKVENA1VWpy4gWaZcg%@rvo6&~v^ZOp^Vy^|L3HXq_7#=p
zu9I_RB*EWpPajw(0AotHS^Dig`wXL;Vy1kGiPTdm-sa>0j5X}r77f&j;4VZvsdZxm
zm29<8IHWu7CY5KUvKfgI{_;b_-LAk!3m*hxS%-k=^*r64xwI30x)x2~c`tl>i8V%v
zba;Q|d4F~~F!h{+b!$ZJlWK7)JGl<XlTimr&OjhpztsqI{f+*M8Me$*ZQ+|fCYaiK
z;QnZBpyU4hCLz0>%Vbnn1ZtqzNxRBq=lX?Q>>mw<mB_!dp!=Zo>~CXE&{b8&;+W_e
zOud_!-Zyq~3EEG1uDP0a*3bNIDrLRNR;0cyuS~+Sk?K5jY0`^Jhp1)Eb?`VmQz5Qp
zDZ~UR>?YzgC*tE5^gb}qkXyrhK3ho9CZtIg(>ow8Yd+DbG-69MO+0<YtETGs-?*lb
zBJP43hpr-7DG~?((*p%FnijuuDv<nwl=Y_GOP;EP(RZUSSLoeM^y){rSe=Hnip*0~
zm%a5v=ma6m)!KVE?F-K?{Sxke^N`+fDNkXpu<f;hglw4tk45Rl%K`p)#ZUPI(qJE0
zcTbm0NL{viT$3y@Mt2&yIcyKc2?_Ed(GY@=@UtF7OY~So0>zVocS9X_E6;!{9h=>o
z+4B#^{sDuhb*5E<p!xt4<)ZNy>IT{+wK{Ux=LD6yCr-+@3sT-UU4Kb;=NI`hT5v+a
z*a7kC?Pm4y-Cze3(N3oB1Cuhl0_VJfaPl3y|A-h6YLQ{q1SK$TWiRTI2vv6Fx?~Kc
zT=6q34Rf>s$?L7ZZJYx}hm88aeR$Wlrz`5rSYo4O9p0J4(}d+wkIzh~(Dr&nA4k8-
zdc>_(2V`IQu5n+R1wTf^h8|X6tkO6$1EQ~fn8vrqp(cpsti9}-E+2DQQ0`5aB1w)?
z87bd;Tz*{P0~t3n!BwI?z;G9FB?%?!`~{Yx3=1`Y^Q4e16ckrY0p4dUBqg*afU5=n
zWqQd+UEnflYrLgt5CMS^Lt)zGpWR&LK87U;<Ta#;_jYpD=elX)$+@IEhV(nJy3Vr&
zwf;Xf&|cMr^qupT8Pl#KtB!3QwPP}wBlj`(=9$y({Z%u1RDUD|3Sw0u%a)Oa9s#J-
zYz!y09c3<adxohrzVmDzZdRS3ZP=V)x&QtB#@>$uhmBO#a^XZm%OuGZ2O*BjVx=wQ
zc>@B)DCJV0%M}){(qk*dMw3MyyNgCDE_R6f`mL$3)?|M3YgwvX<I_5A9sU`HRm5|?
zCKlI&0S~H%A~YD4DfKx{Yf@WP0MeqRXyQ&Gf%mP^f-amfGq5}DyB9gv{ApO$xS9IO
zR7vm#w$d=dXxf%xbld3xAD+S%vw=CAS3&k}l$(e<gF<1}DK|9M7&^kFJgs~=s#5ae
z+c<~)=G{kganh=p)CO)IZ|6x{Gn=20;3m2Yzkz|qBGJRBo+p0rZ_jmx0H<%GXwwGo
zHl|s7)`RQhv}>Ni%7l_3>)$oyystcF3tC!wmbO8Zm%*Gh6?L}Bst_m>C^y=5YX}M4
z3_YCQo`~asv=sz;e3nvehfMHRePgid)|S6#4HM>ov|YeZsDYZp^E<b*y4w6|Xhn+k
zd7wvc|2u&&F%T?=(=iFiv!KUH%6L0E2a1vKuH9^<n$xFB;`HdH)PTitxc{ZJr+Ql(
zZbZ#)Wul`n#5CDZh>>CT`=2uRk8~H}#%?y3Q&8X*R;rimDrm6s_EKZ}4EM8TZfI(g
zzzf3}W?>Xm@2H&Ck#CN=()kX$<|C;QonZ<EEc#*(vH<kcD_TIj2+M#Z4is89=Mg$j
zvKDImLtm&pc-7JHdrW5rNs~ba1by;Ll#`B~|KP$}_Cob-zA6XWA!Bn+>x%L-4#k@<
zAoo=dSz~n3)W>m`%wY_6rokDjf0u0F2fg$FhpGv9G7y&@WW#xewCChrKX$f;9BLWR
z3w@xoa;q2S?gLUW+V|LyPm&&Kv{KA^UQYEPs5)6db1nx#<w&Dxejd~(Z?1Or>|yO8
z=;k7mG75NTeDJ?M2@*H2I&Dq4%|6x#y)4Ei_tGeNhOcBOr(+WP+R3bu-pa-|0xvCE
z>{>e$goPcJ^xm)xiy($vo}hKbrzooDw?`GW?*KTk9az6=iJFA92ukCVI%@0lYH7h&
zIs$H70jUGh_oTm(W7#u4&6sCE&UR=4m6uECxq9c?VK!5<VVj{g3WA9LOD<pMD$vhT
zl&mt)uWxGn5Ju%9_>RKhCsTncLYh2=BYDTRh^QBqrb|R%igWAb)wvwz6F@RsiXZ%p
z8Q)L|&^}%a-1<$cA4MP9U`bG~x@2c}meII9%o=V-HI!?N1X=Ox+x!WU!nDO)Z61t>
z-nruElzt(y*Li{KWyn#R{x#e0GOERGxtB8P=S?u$<fp)EG+Y!p6;-)L8Alkpce)YK
zU*RsAArMIYZFJ~tV;k9VKhWV@O;1{~?(z)@xb$olAzhY2Z1V|x1mDm$Ze0kS^QGM=
zrRS#S$z5^hX1@6~_yRVgN+}!?nuzuBQ&Uf(TrDGLj;$cApXY)Qtd`oBQ`+ifhW0T|
zb@-b=DPf~aR>#ptD8<N@|CTbfys*J;fyjoR6hR>VxS&FDr==-2TC&%G4Ln`Jg@$1P
zv=Ki`)EQM+*&3!MDy{!Zd~C`#*t~9Kr=3B*aQI(_m^EVL=Y0GFlmy&NlP!_=XMwRl
zv~s7ISkYeF=Se;`jhDrxq@Iq?QjX}mPyd*qjn#bDP*dL1!1u+EN5)%or}*jF80?01
zHLGz#6)`?~k6E(5Ur6NESF~+<fjx0B5JRtEMINrv+iWAG4u9D*9>3;3i-9JB(=omn
zj08#<fGMqIP<7!qT~R#9)||yVIp|uiuK5%HNW`l0n#q-tiOB*p2<B#NRA`_#woivn
zkwumoJ(Zc2Uy}d(=qd`r703%{CU?--lRb1o9;2@&WTk=x$4-XEWP$6Gno+=n$GEmc
zSuLqw0eq489g>J8@VGu8E4CObHBa?x;25Q)9YVMr8)WP)4bI5(`(;m{X3zi8LsqIi
z*#F_?GODrY^WM>c4C%H6CI+yUt6+UsieLOe&_g5g-|fSH7i=S+S>{rfv-GF&o-M-^
zRSa!&iAWMHc#Vx!8I}J%JSYK=RE0u0DFv8LG!I-qlrFz-);?9`YLesnkA-!^R#l~!
z0<d!|-THQ#by|XY1yOw2@Awgx&=|#2nB0vXGv{&Jsxr?${c5f?x$BUD{x73b^y^c(
zg^5_9fyC*W`Ksp^yDvf{t6WpGG;v<yD!d~i5#r$R{O;5^^rUJ!*@TJ~U~p!|A3b2A
zf5pNlP%@W+HZE)K_96FC&IL=9(13)OKWAo$a;3Tbf`S>{E!eCqqvex`LBugA=o8<Y
zrIGyuf4Z9_R>zS--vG(1mF4ne&@#oR9PDoU8LdZ>3TC`j9beI>VcQ8EH%BmEvVohb
zDsx`rmXPO+nueAHJ->edl9)GVT3`KsckJPs_u)EWct;!AoT_^xQ~GB^YNOuuSC<dZ
z!?RDnK%jOg@q!}mf$!aH`#w)dS()@VJbSqX^T<8b@y#R_xj6bT9K;B57kxDDlc0>E
zyMstXaLLlRL45#6fp}BVkSiBcbUJ7b49|rsSgouquDq+5=H~@yRFhudD^Gn!qS0UY
zFl)@pU2BHuZvF@>y?Wb#SdP#6Qw7OZaXQKo(s(r2iP}OsuanWOHd$}_l({o!oUjRV
z8t0QrT^Zs<NybG+H`hmY`)o#36G%=bLU`zg%DK_Q<*3{bqBL8M1~_iM*gp7eSe<nu
z1S0ih|A?+zT;aeTx?K^|PGHY>Bw~tZ|Gge!k8yNh5ti;QIJ9dZGfwwM+Enc`Kj`jA
z@zUFv+1c!2F36w59-{Qm4xgHhA(5w_cT0u06vo?S<v=RLOs*D&22>-7G$2umRB)|#
zuqbKpX!FdUi-!dAi}<`n(xtzlo7$ei?^}ZIZif|GdH}I8O!HFIRR0{k%_^KAO+shE
z^zVuf876b<KYl6cj-dN1WI%?=vo2w*r3+Kd$bJTT)?Qexybf={uv4X@i60XMbL5M-
zQZV!U@2Zix?NY*qs&odN<^8C&rQ?I1rID6}W7jIe!)59G_RJy&8Lfh3j}DW~f*D$C
z;$ccLCjfq8yfo}QDwvjHY(>DurPIHZhNOBjl3UDc(9v!-s(!4Jy}YPDvUM82h;tNg
zfV1H8E)yP}INQpqzi>n|*rnM;?D1aTy*4tq4922BP8e@LVa91c8p=gZ6J4_JSgDe*
zd)%xUmjsyr#gsO~UyxUlGg;K$KEkFrikx;O0`!@8(j}7ef4jrnT(9pK^PhG(m7HGH
zXnd)R4c-4jALZS#@+9G=ayN@+v5m5rszM6G3nh>xv@G;4mGIP$vb5aWEj}SX<6qvp
zwzRN3uh|?z7j=YOT|dred(=#~Zi?(f7OY!2b}YAT%4I5Bp4wV_(K<g(($sJ$d|tca
z4&*lOGUjPUV}iFr!8Gikj}7hE{0uJE0W6^BZ20D0hJtoP<fy3<$sC3nlabxlP1fE|
ztYC6l>^JuPgRuU!>P#Mw3<Iz@bF`|`l(AEVp_+W&l0XzhH;$38avEtoGrdcUh)YDJ
zq26wFmMi_l;ZXVXH(HIO`MQoan_7|^=<~df++Azn78Vp_QF0o2^d?|{7@iVGrvZ)*
z1$RwUQk!)f>a!l;nKjzhR4}+3qy?+@%o_?>?x2osezY0V$rrk0@9lJixjB|L2Hh2s
zi@td7%kim2UZE`R$kEYm1ZulkFlB3`R<bF#x#_ArwxwPq?8BjpeiGS2b~zLP7?s)8
zo^Py|PS3l5;+x%uE65u;)HMqnBuRwT`SU3iiHD>rV&Rr)9ljY9lcMqEcNh~LNsRj-
za#AV_4O^2yLdwC~nQgQLjh!7{A8yw!3B*>mz5{{Hn6Oo??1wuiwI|;fs@zgZ2*;^U
zU@ZMG-Hg#8+d*}nc&ot(lO2Fiis+>!<M`q@!M<v%$0&qkFg=toPkW{4#;@`6+0z6v
zzdH{C>z!+Nz|XH$+;|$eefD5IeSLIrymEi*<}{L}N_SNi5Fh+zW&;xeJ6goPcuC04
z6<{<oH{}}@EE{LbU0{WKe;a7DwJ7?B7%G;=*|diDv`}TP`GUy#iWCrinHSniJ_VWv
zarHw!$~jHPScH{qSTBT_E=^4>MzcpyT4Oocqjoll<q%;;v{oxiPu}9hHpEaZE{o3G
zlL>RzPn9_|W+|^qxTxpWyVm0p4eL1h$SZRYed4NqmeAO9pVqIVP#gd+x;o!(9<2?1
zmEWTHaQu67;4&e>&ku<3l+l_dk;fi%m7-6$Sy;2|SST4PW%nCcB0D@-c?j+9w~wR8
z{W&!h5Ev*O+^jLyOQ&SsWa1^AJ%mD~a)%siW(#-D1TPgr4n|JU%U?jDax1B^7#+E6
zVeO0?XG-c>IOW&8@1TZ{?V3Kx2twj6h{@yl(bI5<lFbWjMnGPTNq?t?3!{Qo?-KMG
zNwYJgy|IX~e%8{<ucpJa{CSDXj+3!m9lp77e$L%D>jPhTCCt5}{KQ*W4*T`1>)bLg
z8{PwB_1xs^WGsA9;Y5R^_HVJ$TFrqaBL^y->qPaxUp(=M{kKN7={Rc@Ji?(ddmzeE
zX!v*;Mi7c~umd=9OT>(ez%p-r&yj9m9^r*E$J3FTcID<w<iy`MFHRfNjnuW|Z?qH;
zVNs0xe%R`#vmbB{qU6gqN~r`aYZ(GGlG4cER$c|Q3+<>G^mN`tWj^Daax9p&HF6m|
zAVK^v;=uk;9qa{@aBgPA?<6w72KmCb3hR?A_<}M4F@Q-R)|DiVblHEA;g@kmdA(gN
zQCt@lWkss-uV1Q4rZJmVN)RmbXHkbu?bf-Xjlgx0Y==z8W_fqyo?WRy@ES>=-hy6S
zw{s}6;46l2k4+a;kK^}=rlARg7EJ6f=CI~hey7Kga~9$!2H)gL?nOmKVN9&Zj^KxF
zMQ~Io0AoGVQ4a_RM49q9{_1usnnaq17;Um5J;mq;JmNS~K4Y>2dA8U+{rqkYj*q=1
zg70syuC}+250av1Sj|U=wzAD1ZV%1Jyk_#^ASUsj93C|ZB)Oy;VTH!09LhbN&l^%A
zecSjtFqu6`oCwTprziZfXumkNz7&jNPWsUcK&%qqIw=cUI#34)**ZRzsVFUQ94(mH
z3FYr^m?LKby_qbXQCnD$Zz*YO-L~i_z=^DAq9=^ti-@s`+!?Mu><<>*v)UtgRlj~E
zkY)B-t78@w7e;n-T%>gva((uDGRluYXlysbBbky=qGrtl9p4r){_LXV#;<2Yk)Esv
z@LvtF-;Y)$U?FW7E&}eeAN&LF=FbH#8vk1Cu=XtMqBdKHjmxkZ$lV+_XQ<I!rz1Ln
z2ERbi*(}qAqo+rrN=v3!>n;@G>vEs4U|+E;Kt%CJ;SnIBc~k6tYy2OnBSw%?f((vF
z6zYhuj4+&Xn#IOfkCGaHM7txZpGCO>e~V80Ceow%m|*<sh#}PGZkh*dAS(_l<7a|m
zvd1t(QL``nr{XW9>W<+q%I+DK*I$%U#7}-YUg|9(ull~_%{<n39iP88e9^e^NOReT
z&I@@O;3{sq)J~cfX10Lrd+s3BEh4=>f;6T}$D}4@?=v@M?2c@#EawF<{lp!>6Wx*c
z`8#(_+@{>$^>P$R(N>8v1-V+`iD;iuq(eGbl?5U8Wk4=BA$TPUqVHpb3`j@Yf}=jG
zmne^zwskB8|J&BXm8y6o)z)2?Y=wRz)^PAduq=N5ZzlFMWV(C`#pN_8ju#LV#1Mf=
z#cVh2T$0NjIR>qo1V3$Rc{2m(BB+A`Lg;+&emC(pmv*DdS>IWzB%68?gs0ILPYI0e
z`jv>T)P5Iz*fsC4=f{e0>Rzb&okN{Mu?e#-@fi##X-{?}<;Eko(%|O{gYbo;iAq;h
zNSbl&bI_;$@9Bn)U2l($k9#ot9zQ=!ElypB_*bO3iBCU+l7;b(fEK`rFbEC^D5*!?
z)NsMkq8~ls)D^<@<oa&>t*G+dE6?qgcC9vWYZfo2J810TYz#<88qB8^EM=Bf#we|*
zY;Hy(Q1uK}f<taIJeQ>|?<N}ldt>Ak%N6j!M+8p(s`3MiW$I>jSn7uB_vg;`Puv>6
z%-~)6b&Ax^@4a)O(*>qLUuhbbH9+1LIPzqF!$Kzr!#;|$5NGL*6BU9~8A*EXe{?j*
z!?#?0;G<PqNc8jKT}#B0|Ed4#3J`ROc@KYY9-Vj!RB$DueyDpQ1P<_Gx>zDCXO~u1
zf^Ngs<mkbTm?OkH)HdDx8%7c3eF&0}#t-YUOpvp!r<sNWPRH}f^O0dVAfe6%<ccbZ
z`?OQ#g1jS7r%U`5H)~+j1-yO=M_eEggqObxgUnWKJ?}OwcU_HfwXm;G6|Rb9FxJ+t
zsIcPa<nx0YkCeI@&o|(SM2J@RQr=kpy>1WwoN3tzWn%4598Oj>w#T0Mp7?>~wa1>}
z;}78n&4;N#ZmrQoL#aB;9?F&e<;!N96{>k&iFe$jt>u%(CmCSchn$bdh6vq9%)&!+
zcF-Db^=Wos`M|}lqn#@4XRFbq@tbSIM`|lU_8JvKs?(jhG)JN?CmAn~_J>?_7JA1v
z&G=r<%cYr6CpM5oW<IVMY9$wBie+%FGE9cGw2SHi*u3@nc5>s!UffQDq=^cTFSh^r
zdtDA`RyNN#W>g64fQ7*W%ErhX5ojUF)5WqKV<P4YxCH%0i^uKl6(;DaA^nZ!u213n
zdMcNOwo=^R-WfbweI$B&(I(1yD4J*w3v%D?O~;ZIjyEAi-lyI>dbB76X`!*Y&q8;4
zFAg10qrA*>>R^Ucv4#?oD<raWMT8{|5^|xPjiI55mUM&for-K2p*#KmeowB+-nsz~
zqQGsB>msA&M2tPbNL1wNe|3jx|6y{)WMzwB)`)t;3S&<rdm|)0_F>f0_N|E4=_I<S
zGzDRW9tnu&_&BAxdyVV}!JKm^NI3y-4xpfODp2Q<>9n$wq_84O>Bx=3iIX@{9p!NU
z{c=C?K0<L_X<y3Gv49eaWa<7BI;?;phi60(IA|inuvT%YikVFVdzaq_mG#aN1pyVQ
zWz@oi!@U(tnVsqEi7!uB5^+9ajuiAjNR)?I6{<KxGAixba`Bn#!2x@Ov1fZLkd@uR
zRfMcwTk`jsTwBw;bf>=z9w?pMkm7w2tD5h+cM<i3f{t8mwGwHW=wY+{WR}=;KTxjm
z^lJZft0Uy@<gor5=>Shb#Av0O>GThMKNbcY1~)twmYmedxlmowpy2r`!wF4XIk$%M
zl|6yaWrq>%+-Is->=BCu%?L;YfCEib_oySXGpnXHd1tQI<CN#<B02Z<g+D<JZ`8DY
zl{%ZNuz@O9IKdL%S&!UeOmZ}cia1GGK{oF1Q)>vdAKF+i5%LrBnZ~5N8tUqljcPaf
zrSSA;Tfw?6dTD8CgXp}@J1e0juso9KuI6ju)y9?bE*|_rMPPZ9?mwGLpfJ1cV^s-z
z#Q*VhF8)mZf83w5<S?S<6y{LKVa~^#OAg7HGUgaxhnx>X7?N|OV$7-JEOVUmshlcA
zF%#yHOpK|75bo=DKOXme{RMkGu3ewc`|x@_U-~t3!G#rR7KU%D=z4Au33kf!%l}xX
z`!BzXvrU!BKF%#Gua)iVnV-M3)V(?${Tj5Se^k&xCtz)FPA>FhT%hlUiVeQ!F>TMQ
zkp&Mwt?>9->MZBpgP+pzu*r?Ic%89A`L!S2acvhKY)z2OD30rkOl--q#{?LO|MW?l
z`*t7V9@XAgQn4{gs=x2IzYP3k+8o+r4(1cTHoQ}kqdw#Q8__PN*33ty%iurGd)iKL
z(tv(A5O(hifC#$tWP8@nRk)3N8cM)&YRvIdM{0FX3O|nV7{gOg?1rW<Nu3|QmB;bJ
zQlHwV`wN_}$jlm9IT-;+m3<G-2C_u1#~uY92ckpTBOgB0)hm(5+RV@!3x_)QMcXwG
ztPERIaU{T5_bTVp>Cb@?Jk{CRd88Lj96h`yA;v7_L}E9sruO?ek#;>!j`j$)x{F&&
zOG}ujgNPri>zL!+KSvY_ZSzS3<EJ>|yVR&-eptM_xQX8kUBEQPaUoE{Gd>?oyq03C
zzy~^)|ManGdt_u&u3Gpsyz23f{;`?@QjJNEt(=JgR8*DxWA5$aF^g+Ld5`lN$B+1R
zk{U$5xw4}G!^b-wmW^;TY3gHEaM8VCvOi{4ygjFU47>N6K5X&e&rcfcP$@<D;G}T<
zaC9(n7Y?$f4?tihVO3gYKY&@vFTBXUVPj6&YcYBwUq@I^p>s)bY`B843!%5SJFF4y
zUfXIB+4*KX+hbN{d(P!s&5(F|z?DPl?H`*)iw`hMSf;bZ(K~pIh{_+Gr=l<(2A&RO
zKp#1B0)c!Cj(mvwsPhP6)9|t4J`#~%x3Toz!r~4!ssq#gX-aeFKO)1bTH#RzFSkva
zHzaxB*UR$T*E?{UjqfM3R0sEyu9kn=EiBDg&2tOp4e^kp0~5&S8dKtw3<DWMs%l1+
z?mk9_)XBy@hU_PwMCdFcdUO}Nd%JJ1f2j2#2R93DzOa|=dMHa>-T&zmf8)?Q*i4p0
zcO?Pxm1uz>I`H$gg9SmeGXog~L)|^ffi`6RzOR2RIV|RlR_>nuqs;XQNP@Xj#$seV
zK8Dq))%ot%;8!(hS;=JAoj8gUF}JDW8Ou+VKPacl$j-uvnlu`Z;d&3Y+o$8j*ZxkU
zwl02b<mWfb{AzjYH(Hq5zh%@rvMAS&|Mw)K(*lO+ehki)yqPK|m0xh>Q@zWZ0RG|D
zZXP-pC+g_v{PuPhYGi%1xu8Q<S)e06YCyAjWN;~L9)s=S8SXvLnBDLe+w=7%cQ#`j
z363Sau<QTCW5^DLB}#}fc`7PiVpOcH_%ZkAGPiyi7+~}yM}#N?l&Qa6|8vH#CgZI}
zRx!QisjP3u1#~48gxb@SRTGhQ=VA4c`JRgU{a`~8H+6Q_hHJZc!RhE?irRDOuvqO9
zGdW{+{msC9tU_Nmq<=(Y@#JLbbmO!k8GgII?;3V=MwH3Rp%TvU<pJIV>hGRcGdFw`
z=YnUA`!p^!mFZdF&C*EUUI2zY2gAL%Ft7as%tM;Z5))~5%VyJ`{Y3kX*%Ogr^W&Wc
zf{w6`u6(;d5)bP!Z$JRh5AGj~T=tU;HKf0t7KERL{3mf0)u#QwBl^|&ly^_JTRxiG
zhC~*LvUaJ8Q>^oq(|oK^&z~wSkPaFxwM!3+KmMgi%arC=j;LFwgN7R`qjE7D(Xj>c
zSD;p~y_hlI%rD%_q4^Z=Prj{BK!Hq>twsN498>xhF@lz8_v_z=U&bv3zq(l&c(Pm2
zK3FIF1Y%#69wd~s8ianh@R~aMd`<b7ZNbbmKJPCsThy3`lIOM=1*Mzx=y)><j2MaW
zv(p?BjOT&z#IHb|Yg5lig>skzY5G4g^#Z>;Z)451SIQy#$1zJ`gRb~HX7La1DKiX0
z2-G6cdzr$+z}_d%{-F=W+inkRxS5>C!$ZED<qSW#);ESVcj}&IHuU<vHk10-FKgni
zU1Q{sU$}o9tN)KVuNq<;zpr<A1mkS?-@ESf6wk%)o3$mBkBz=%isG1;QWS#CAL0)E
zbkvG)?P1dkY2Tnb{k(?kC{ksRdU(uTVwznWvgM8Hgh#oFT^;dFo=;s0*#4ddT!Ixb
zey<wavXeA=XZ;(q|5vkudNF#n`K*xhb{+8@Q#Xi=-D9%f3UKyBu+8qhyH{gP@*k)j
zk3S~0OXG<%kui}`0G~`JlssFW`IW4*zLj!2<C@F7*LoLib!Yzb&e0&gWv~0LO+YNj
zNIq@1bP6t%?fw8?4>1ID4kq}yP}rBZmoh;82{4yUn@vO3=#VbV*{@$&n4^<q5xp({
zgUYE?`-K?PrMetY;s;-*cLR$1VmUs8(V9iOP0Z(vc@1ZUxiBWwPW<}DvpiLjhm%RE
z9qRYd=<Wp`&eeXX|Gio(b6WE0NR|jdRrsqtee~xrL_mudUx3EqJ`)9)&px`lQQYjN
zryuxUqW3bgeo0AlW%N?J_JCPc87(O#bw7!>+*YBYUCg*etO*26mejtiR4iUlPLFW6
z{*ued7;^K<PmB+7tVn;di#KazpzB}<^Dpp#7VfApflTMWKeSs)?ev+-xN~&UuDzA?
zilK`7uhq$EKmN!_fDuh+U}COoh27N=*H<6h>%J!~BpV#~a<>qnoB3sjbD)7ss7?iY
zgUatA3a?$?oezzy4Le770Du`fr>Gq!9*a&ByDB6kz^8SXYb$sCJb&?Q)6ZwMON!V}
z82m*V`QFy8;lzYs>~FOYvj%S-{fxjIL|`7GAI|P1g%yANC%~voFAggOC+7zFsO#CV
zzB2a~VtMJ6<lB2~wERVm<O7YZx~7i{utY`n0slVC!x-(7Sx?^*1ooL7z8tUVD@W5}
z7Z$C+yl~sx)3Mq`uJdooxJWY!p`|Q3_Zbe1NMX9A>)oHQyGrA1**06ru8<_~z)GNt
zG4n&Zf(7pG8U%}Ktz=svy1=aU<f?MMclBclt%L;nKnBS{QHodH2OiY;q~JkB*JHu{
zN3FLq{8H_S`xU@=eICbI@W>-1s`bBdNc;!PWcpXzO-U?qA@$27m-`6xy={x(E-emG
zZ~kFD5@{yy`+-pPur@@dh8b-5qbszy@*A6QGK?dc&Q3Y3&dz7Qwh{?0Ty_v+d$7Gk
z<78|2j@;83ANBun%zI8_Dxsi`7pI(eYE*zPsCJVM!%?v7Kx`1Um+P1ex<?Mp{||HX
z+fxbHJ#89kM+q0>tf|paTI~SLATBce=rSnsX&iN&wl;)PCRGWJ5IRfp3;GLH4n*b(
zE@Pq8tHxiqza(d6?JVlakA^<P0DUI;)U)k{$7$1>UHEaa@Z}v5vwQFqw0axTmxv4r
z3aM39Fn`t{I07e*<rV_cm_lwbOxe%TQ5#fKtvx9ugYnPd`h>*4CdFm%|0gB@WznRb
z!8cFmF~{`=?||W)a1@F}c8xwb{q-j*DmK<8&?c->O^Glpe)F~5A7)Q{-A2-zpcm(a
za5od~5Y6OLc=Hgxqkr@4e!d?Ay{sX;$mNBqoMW^jtK3M1L3oImCut7n6=AQ8n48l$
z<E5E>ZY`@-&2A?f+}GQ0OCC6lFhx!99Jx;8%<+-<(Wc^0=H<8H3y)9#2A06DMmBx3
zU5EW0hlJmXdb?~V+b6fTi=DuZ@)~9|u0LhuR~)a^mVq&4&prY6{ZLYA+7pHM9E(;N
zoYyS_c7x%gMkQ_-)@4)E_1%`+;hGa2L3QL;J>ms^WWMfq7aKs`xgY#ujLux;+qMrd
z&G)W>mRlzSaEv?`f>>RJl1E0m`i|%I0H*J(2|n_qeC)fADhZb@kaWI3JJqNb`5#X<
z1PaCF0JH!N&NfX27>7lI-{w5$HRd|GD-{xAAh&9d(HV@m{Ah581r-VIwk@GJ)-p6D
z-l0k>*c2-u+2f59tM7^_4020Le1926L|UFtP}Yj8&MmyeBKaMEKKbN_hex1c32AGS
z)~ovNf%cud>WBwn-iwM=^LQu0z9HVv9CWtk%4_Pr(an53+&t~o`AF(#G0;n?#sK9S
zYS#Q*^M9DTSKJ-494(({XvKB%HIt^+mSRtv`Zlg6yc(95uOxVNboHf6-+GE1RY3^_
zE^G&X=&-x*$!LS}5i>Ma7+r|%S=tG$yFYCm%wQucNkUSIYv!OY_wN`tFnckF+~i?+
zzgc5qoWOcPRQY^?+I>gm@&H7nrFb|Rnzz1oxUq&=JKQ@x{TYZ3y^1hD-wo!m^$jwe
zH}yUHN<IFO!$@=4v_nZso`Y40T^e<|rDzQTx#ji?-An1`dxNWftxw$!qyl~iokTDo
zOWHm<dG+vYW#tzDV2u<+50J^*4I&j`9Tu0wT1$65io3qnu%$j_C*SdL{T^0?jkKQ1
zmBP>`6L?jOxNou=96{I&>olEn2O832L<aswoaOljDKp+5#HS}O2}>r5^3W@Ph=9zj
zDQ=C-mRI99DV~W1oD2>5U?frh`A5FOLNIwjNSzwAT3xGM+I-B{y2*Ll;+>;;b%&3f
zT)(jws|#{okTrXufd3pH@=;c+Cg^RrFD4ZX)BfBbm=pv_DkdL64*pjfY=x6G60fsQ
zg~mq3N_Vz~B^ag@mWhNX!z8(zuHUgu$lx|pC`=5k!;T&TQ|ajIwNWX?LX6zEB&42k
zN+wAp1BTs)OHZ^JS}sOiy#MS5vw|wXP*LiU0}f`U1?<=JpOqAro8*f_qzN;aC#*kc
zn|6&J9^G}v5k(zN!Wn(^srB1irw+7DZuMx67Cg5PXVXCJ!6|vQ<n-^9$l1@Xu6f<*
z-+xB$(q}b&!1>&%#ll16A2Cbag0;41)c_mZO&n1;cdZ*grE!^?F?-<yUA-Uvw(*Ig
zP}aysbVuQ*3W8;-SzQNjh>HY9wFGdXn}1R@?)mLz01*ggQXIVMm96{QShTv;RvG;t
zqmg<&jZYM;5#f=*TfXxwcY)X7D&KUPwBtbCveylt2tFuC{mA^G-y_*}xAXdh&vO3P
z)9SPv&RrtKIT!qLtL=Hm5jMFf{oRGafn*NnSOC#;HwA9Gu~mm8kV336C%^yv{uPX_
zeJEO(8;Pr|3BzcV;m<=8Z2_8d^>&NEm1ZxLuRY~3a3SRB)S43Q!25@NA_sj+(rt4&
zKqISTmG0r24WbV<_GSTx?%v4sTb2%Od!r@ymHb*U?*hxYCu4O!4#!UOO7tpoMHGSs
zb|V-lFzbe<Rj^@BS9mb4=dQWQJ^HGIJ=>r@REYx7E6|24xjB<h8s-b0JK5Q7>MzvO
z9Fl4S3=L%OsoL_gXIp>U9v}niEzW^JjjaIx{nb2w|AXU`i0eBkL6D;p{{UL#hgHRC
z)%8f-o?beYQtAC_E-|HBPP;mmvpQSloew_^wYafE4O#hK<;Y{?rph8sny16`PX2b0
zb~i#b(*v<?OCu+-S;kMsF54&Gkc-4Dl)vh74DRxE@GL7*cL3>A3GbrKeOtwg=*3|E
zX&sNmJB?r-?{Yy8>f$X_tjs?UrzeZ5!)|*Y@3MYiPqT;Bw+5B8*wJ4BQ7#I{^J?rF
zsdI|N{%x3llBj{Py3m;nKH@!+=R|dDXQrI)9frvv=%ghtWxx31FBA@tC`8u6Eple)
z&)Z*2VT^FXR{q~7Ob!+XRH**%hrsrRX{e>Fn9Zq^;{$aru4m5}a2<`xV$23?nGM&b
z=I5u@es|)rxVyDAk{4X;eV@J+BEg${!qA3y8Pyvw^iG12YcdG;=I;-}rW^JSW8BkX
z#Fziwz$QM?c|eJifTFMf%KP-^)Xv!o=I}VV%ZSIZ;(T6*Pc6rT=Y>AG%GZ=%pp5&$
z-*8xFBXtsT6e**eqf}4*jvp`oT4vbi|2HL4XykZ=rEcxdQ7a+K`w_9FS0H#q73+<x
z@Rq_{)MqzvW9O^krvnF@23Z*P7Q4G;zQA@M?y<QO{16wDzzq~LmoW&mlJ|`i<4vdx
zlUb1BqUc-3nK+AazGmIyR(>bJDE@eLP(r|wafpf2;T_Oj`?0=ADM8g45W+zR+~%gh
z65?YqQ^zlN=f}5(C5~iqG{fn?wT<n)=zL6t2ZDvi>Grx~@c$yE4@;_YkRMehP-aj^
zr58v_B%lFUC0nL6cK6Z_jtjaR?$?-0nVd)2n}4EqTGW^z>Z0Z&ezr0gK<}_DzOOAy
zBKD&yCD#jWUdWq*y;1K#Rd)525^=uE^JBhsno~M*Q<2w8Jo}rzAL<mPN+&|xF;z8;
z{Yzh;|I^(!i*qD&@>{2nS7@<qdW8zRI8<4_KK=rIpIfvozwQnCx>efGRV%&T<G-iB
zPF7C_in`Jo7z-;doTKfpb{Es3$R*BB3;!DCchJX&!5?;cB;HVcYGV2>*qhG)iMJ|A
z^tS@<+bjP@5}Yck&Ap2&fJy;^;dzl3SFcb%+8y}x0nNgoY-VhBd-!@-(;I5Hg|z7s
z*XqjO2=wvq(W$<FQ++r)brVtbFp2rGJIkC7y9%PpN<4WvVUi%q3|potBkOF{^Nw~z
z;Drjv1?9HIYGzX}7Q~IJI-5&hh5XD7RU-tHvK^R|Z?GoHJp>N<<oWq7%t-Qq$f7yx
zs7~dJ8SjtOKmSGr!i0YQd@#BZEsbbKhJ`LW`YjBRUHbqzFshIG?rN{CGy)Q+UWVKQ
zp3wXmLRGwJO5<~ID}%6x@~x+nL%$l*-76U-QpT>)!|2b0fB?9|f4`6RHa2#f9F42Y
ztF4jAtBa@^r9PCrZyjcG`T>1ynjM+FyOb(5&=!8qcoFOt!89|l|MM*OBM0;{m$qSW
zvZD+PU9G(Za1DV}xWC6sJ$-%W>5;jG*k`0?A$8@c%1C9g&STuJj@+(Lml&?Ydj_s$
z>HA@{lXg%&<6-|vQ$b9ECrM6_R&bdfd>ksWlmGKi_(y7R2gcE-I$#}$E*`Uhj!Cdr
z?n?FmZvyPT+2w7&z_%b-Hp!u`-geNYp8CKgr!$(Vgpyy8Ph=A6XVg_>zH!ronHyRP
zW<273#Qu7{#@7k!ftMV(=u8iqY+VZq2wi=W3(AZ=NdTt`cy>g2k3uhKs%`doKN5Y=
zH!zmr^}6%{=GHww*n+d1-Apuh;?;$n<oS`VUzFS<-o7&I6d-p1SUv4!ZXd-M&w2J<
zGZ&=(sb(lsT_@6UV<U?zBAs%PR29oO4R#=Ms8U@ZNsM_T8Lx4f#yNiGZbzJ+nJW#Y
z)O`fw=%?WHt-^Ni^Xq0xlT03)7^aiY!)I>f2OeOMZC4w*#9zMRjfv4b`MtN+uq2Xc
zltEsus9=4G@Rdu<I2D|ebw&8z$e-QYRGgsd$RTCh<B8*phF`vGH`6=a5@k${X^&|?
z`O}rvmUY0jbD6fYGZV(F2)*mfv8CjkW~b~h)A`Aq6xQWhC~q@9G=}w1Md)8cV7r|t
z9g-M}yuZQ4hQT*^bLAzK3|8dJapvKnXsJw?9M#AF+;RDfInN_}R}il!p^4&+Qi#!F
zH5xb>h731e%BDxA<2R)fML~7P=ce1<m<xWfm(}1~5j8MI&SN`j-C<KT_#9RVQw#2F
zGim=M&*^*f^C!Okly-zro}HrP&0yOQdJH%rC6E3fMgZ*&^cDg8>gvgmA@W$*Yhj1+
zUEFiu)XD9834@vLVm9#GLN-eyV14tOC-REK4tI-~n=dPJNfxHC3cO7jkIj`joY31=
z`$BW}k%HBRrN{DO7gFO`8x+q&3+9>}D>nS><&9_1=N^9LAS)?Il7bd`igHJFFtY7p
z(r{~iUY`Bot;H&JK{nS>Yp9#KUFWf??%oshp&)APALexDtjBJ=<J0ISGc57M|6mH(
z?p~%H^OA${|0^VmD<g3gObe;Afm8@npwH2Bh5&LEC&-}8GIac7lhaI?nRkj1j(?yY
z@=tBsc*Rw3QFrm|_nvw51u2Y~biUU4w&`qERXdW>c;<F{HSMBA8L&TFB<uBJq9bO*
z#O?vq&o<KC%n6BXBE6PaGuY<5YXt8XQCOO<{!g>_abHq0DryUSzIr_aVwK17jo~c;
zHn)^PKKD=|3O-(D>=qUlv3plr8xjiwjPotW;&W@fqr6|ctIW7-U%@DgjVNxRN4g5;
zlMBe>9OW+bqH%*Jb^gS57aA4!+4^3q0hgN#x2JIiOf4-ZWg2Nas`$rvviQ1KhV?z&
zMUjzS?aTX|NeT4|>mGg$*6Lonp8d7FkABRu-(qEQ<ePS_BQ7M~VNEqknQ3C0hq3h6
zZyO=@LbWi5zs*TkMli_(;}JRwW}?QnfJocckGg(j<-FAKTHT1!pCI;WR11L3BKwgE
zl^_?Y9%-`zK}{>OaD{XZ#C8LMnD%*dV~GIvt5CN1xMBD%5QbKRtjTdCMZy?v#E#2q
zfxmTMLz%&bBFERguTMOGbWrX7EfO~#X9M6eTut;T5JOD1g~?-6aO3X!`W2?9yge^~
zVzgQ;u$egh%)HpLnYEqdfgkl?Y5;s<XlW}+2UN3C>U-wdJI!u0rHirU4KQooR58}`
z+y(Aoqct5>xCDDqA37ya6~TYs6}{N6wWt+RaI-GV4|G+W)x|-e)KtREdid4S_DHNc
zr_rUy1PrAHSYlSk-<2A;xIV>W^hUJt?#8t^+<CMSiHm-jYB9$IuI=C&Yb4>EovkXT
zj0a=XMa{;_IT%tS-P5==47a|Vtgbf9KZe}O%ZqP&2VsQQjz&A>HgA3@2Ya|AHSSDu
zKUfhpcbb?=9ORfp@!f0pNB|*E+v9xQ9q9atm~c6-bW)Fw%Q2x9|4$05c(!y#I%~W}
zRgv56rVyaZlU!T}>LVwascBJlVfMM+IC-Fg?U<g6x}s+AD8&$fNkI&a<Qdvi8d}~-
zt0_>U>`CTz$KVSO{EbxlI&X0o#o6dAE(v8OZ-lDt6(S_sAIkB8QikY)t9Ny@tyM9q
zXztXL!1kVwm^0cbIT;}5oJ?AyD!`rO30IB8P<&0Mndu;_e+?g*{`ttAuGu6>f?c4|
zA4nPV=Z}Ex0WmqP8mRhQZt0;P%TK?^q$e*~5aZmC$!sQdNxXfaKsa$F?M9qj4x#5_
zj8T_kPb%{dfSykaM0ZEF>n;i>V+L*?-T(f6mbKM0E@j+sWAjG}H|p&xXi$c=1n7dY
z)l>`qU4MmV4Jys^e{ag=<Nv5WMM7LZBkJum6h|f${^yv?<=Dar_&zAUvF-Bo;wilX
zE3G#d*Q{)~G$e6NO!1XZ9UtEN{eFI=YL~^Xg4Ki&gkSRgO5l$p(~KI7j1Wn3oe+z4
zJB-ne0$c0X(ss25-|N?f4vY)zFoAsJoSLzGn{HJblbEAFrxcn7t+Q>{cZ_Z_MTb#q
zQOMsv6FUm*%mCrK6_1Tz1jjK0v<P@bq5gEgkRQ23^!{|YR+r;yan%=sc{6zE&9R-Q
z4JBr3jJLdtJyj~nnCom@sHJihETNTWRaAHHkvr>`Ax@6Gt3xS2@LKC3_A_fu(7!7U
z^V`Qwyn(D+WBx96ij|tww}lXa^mu_*v&3v<d=-OjocD+Fi;VhcZo_Aewh{A#>JLoW
z<5?{c2s**N!1FfXSD-R*XDkq)%WaT&<3{M~Iw${)V_`8%3~L2{o$>zQH4C3R?CcPY
zt53^#yHHy!j92Q<eI?zxq`{85#QLVU`+8qbQEzOFJki7rsP;MD7rKpH<FpKb^9USt
z9UtsG&X)uy?|(9KTVQLNkFVsBQI30LuUmAxoHqm$qiebAh^W&Q*XMcm90Fz)2U~;W
zUB1>uJXdP-!yJ27{cByg5a)M4cZp)y{xwWZLYGD`tfOT@Us2uI6TE0uJ{{A5l?=G-
z%X7cfe5-9`u-k&S38IgK0`9~Z^**Q;L<h<e#xy4go&gzK%rU>WdtBs7c;7Y5lXb7>
zO7`jYUFPfp-!EQAHsv#z%68_HR5=*Z%kWE;v6ya*?sXt+H+q(QJ>%FtjV`V1S-@qz
zv0caDPlWFqa02t^?rTh`{qU{f>#F13-wD&f9gYh@eR?WXotDpHk>@+K5^yz$;N0N|
zzozr0{O!K7mbDaSC)7ABX?ZsMi|92&yI+D00E|X}vWORN7;X;=W_$`6K8e^34C&U+
zNEX99$?ob;<!U=dHb>@HetkRRQC&_;ijU{7`fm6{`TR1gMSP+WJM`J^W{q9VTIaQN
zAKeSg-1PKeSH6i>i=6&JW1977`|8I)$k5?`Cx@C?kKHq*>^u7IFPd?n#~{cgvEQr?
zSu6NvXxzr-oB=Y^jL<`D%`b8PtffHR-7Ua$T&f%T6B<j`@M<S3dA@t5jns_cVp{1+
zYD^J>s%V3z=L!i_B8SQB9Db7x$MAMVBh&rhbamCL?`$RT;Z_pscLfE7QJpm%&Xf3a
z=1vGcu<76!32vc#`Af^SsSR7wf`zxgTD@<c6w{B+;(Tt$drw{Z<{hXr`i74HF(1ZB
zfs*Ui9vLx4y#@<)@9WdGs~cz+<a|gk)sOtG_=o5Fv;XdIn1HV`-+JGQ#D=yB`UmWX
z)w%S#5oVg6;F&pM7K}5PMP9cS{pSf~Z15p=^Ac;29Fw03wlXFTN~Ly0<Xgc_(D<7f
zIKl*yAKzf<4Zg}c1WE2Smp*6noE}pJR5s#D4lW>2FrhQSllj68D4b>aPK_Mjw+NRs
zh}Qka&Qi}iGU;oejqs()r+^qbcQ;b2NWSC8N}w_TO*#$C|0E|oVj;$SYvtETSJ!cu
zeECJ1uYw#Sl-Dp^xwhkbWN+c^0EMYP@Q+&+iMHt!H(JYB2jxiixNME@4hGbpFT~g&
z%JwepDcx8b0(52s={5Nh>-&q@QTch|)w~2vKUTvVb29JW)a1Z_<ILp?n3(FzpnLF|
z>eki6eO|h_)m-(#iyuKm$B#(gUR?BpkaPwY_Z9FEUT<+heqD1yvpzDscARf2YKpAx
ze97ghMpBD5aBhkCU%s5Yih*$$v2tsI-2fwO&skYbh1U2~_qE9UHhS}Zh>_^ITb;Zl
zRmVd$K}ws4gae|dT9e;VZqH9}sX>Gz<+aayPF7!eU-|Ny*N?O`HVHOaLXbAroK`a~
zeU^GQ{KutD<~jA3b3(TU!Cip|fxA;n|7iOwYJZBvGzK7@{uz<Qm~{{2-C9Jd2GgdE
zh>5<OLTyLsogdHQnxvJhzj{h41$<t4zh%4~fT*)m5w$jpDLJ}+i~V*(?l*nV#FPi>
zE|VhT9ON!&HbvU>>Yz}qCX}xqZiZ*R!yPZ4p7M9*U7Y2){q35XUEIm`n1LyyeyR&9
zrg$sZ-l^DmTRoVpfk|lAX!#G>GE=zQi8p?iL4Tjq-;<y?s}Wif+m1OqJw4&txgHy}
z^d!cVdvc;uoyb-p%^X@HzwXFJAC79fNR(W6tnvBeVQ+U=l8x>h1FAyWH}ym@b5)TO
zyUU--2rw#lyAwM2nD&$a_y0QBsJn}Qxbejs3b9U*MPCSw(O&5v^inKZf?vw^md-bE
zOR-}MNs(T==li?YDtXmk%Zl1Eb9DB3-_tT!!akgWvL#+za%19o|7KH$vq0z->rKzo
zvp!LesEV)S(ifYRvi0dxSkY}Jh#e&^pX{NLVWUocMuj-j#DaX>X2*K;h~u0MBip#p
zD(H67CJ#RY#`-?zm_^ERsZ-4iT>Wt)F*QY%Gr6~yp{kCbwFZfpmJ@Aja4vW+E*kP+
zYTt9=W?Y_IMz$N|Xc{fXEJ_c*qQcD70s`lyiVw?kK9njS3g9$S{a6Zp=Vg4uMamzk
z$@oolb7BG;`KkOpZd8$tTZA)NQo+P08wV$Go&2I<{?Ja2yN(}9DY0M(v3*iP)(8NU
z157D(db*?SDeFi(Wh(q`ZH`Y$5GwdXm~W1k8oQKnxCDCNPV#G9zv(jJ@wi}N#IEB}
z?GE!lwBrx5!a5c&codsa?fU63jIg*x7-z6C<2$sR?57-u0-Kmm9pA4JG+KDBT>0kN
zVuCTGT&63dz-Qng=|z1smx3GQU=tyDn_ltn?au%86$>r752WI%WNO}QvnZ{Dk76~1
zZjVfjZ$y}_>ce>2`1l4Q+9N^edO~dE(ZAr>Mk+viwhhw%*%<!pS#r2ui0mUnv_r1P
zK&txIwFmb%m=_kL(o)14s?zdnvb9BLFg0PmTz<&u+=G2J2Nw3hc)@e!eh-8f2Cua&
z2!ph$BI|BEFpe`Fm?V>zqR$4pVo9<`g?H6gz(<ozDhH*8{)4c`U|<BPfitzw-Av2Y
zaWT2cZTyCiC)moUEt+>AO<)Nr#=~pve&vC6hQT($gEjD4HJSN!s|i}Vk6#bQZ#$2^
z0(3)Y{_X&C3VWOn9@)&wlc+Crm2q)S`VaBrx0xH~5z~~6XeQ1BF7<c~g}qVi#QOT^
zTJ&LsKvF=qzmOd_nsscF?T}J_k(`*!oCA<GhQ)d!4}SKYRbrygP71n`=gF~m%P&@I
zUIMu+4D-6+5pQ|P=c>&gY^GqmYARP(Jo^AZ3F*C%`%|PI=G3unQBUWO5ou3be+-(?
znv(;9_IqQVfBB&wQoQbc`sdec9N31SCAX`X*+?z(`jAdybMa~VUjD++P1N7Kh-&N;
ze~!-WCFC9=<b!7~@lroEM`ixULsnQ4gx-&)rw$6c(_F)*?~t9qAB5RDdNnSC;1zbu
z5KFpSBcfI$fP89ZYr>&6$fsN~@t?A;P<TuU{L-;2<Y?B!Gu%cM2!M)7Hlxc(?L65C
zJtJ=y>{tb9oVe6}t5mQmxsEyaQhG4+D(sK+-oPDmgXYn&qrV#=@&axX^i(9CB)v}H
zn5yumAIv@&&{!Y?uItR3AuJ1Q>L7k0OG}iG+-UIOzr%si(f@Lo-ll*H%e-4MN1Mh-
z)z@U**JTJPbWAtCzr3yAdl=%3a0v_z*!b$_axf=KR!ZwzH4O0Vi~b9sLJMzS2R&}!
zCOKU0^K&qfScm%}Yzz(WqwiQ(P_8C=e4yeiZQr!k`K18B8O)13+f;8+osozioiVJ9
ze0Y^h(yiTb$~-~KzXBU-aKK3CFsegQCcb8-+h1St9qL2Rwp)r&YtPL`W#49s9;G}z
zA_r=0`_sZJkAMA?RIriDPoa|z3~WZ8=MGTgDMboix^za{od3BMXG)2CW6B}?^`k0i
z{FQ1vJ@}@++9gb|F_iuV^0BA}3!@ZU{G<Gocv(?ZZL>#@)$H+vcRsQZ*ULvV3Z;FC
z!{N1-W_FGiiz$G@y_Cyu%Vu3`!hDIZ5w{!NlOz^aoh6MpfHzlTJqSqtm+943!X2+Q
zm52M&HF0RMVKC$~0+Gl%LjHTowPQo#K#8nc89Oi3W6H0Beo}AsGlJCwq(K6&lpgnU
z+_N_(A-N6NEZ*`X89%n2JH6P^0JnSq-vzL<uKA>T=VBZa#X;@QQq|{lgHfevc#EC2
zvO3~boh!tRBog4CQh3}8y*{B__=vkMlCRn)w=<|uFf-ZRDI|$+L7=*FZAdxFK6j!o
zddsa1BNC1>qb(*5-Vm%KIDP`e_P__Q<<b=teN2%nFe66VZ+d({*s2T})0wPI)t4B4
z?>g3op+aIKH68fsh&~1h{zB03a=SWB?LFSFNJVCi$194(G%IqlGu^A2s^<*(M!N|o
z6|Vw}-Mqfu3NvJpaVt-qOc~7h^u1j$d}IKVomwuVs%@D5?X6n8`0Pw5G06_}U4M{v
z$IRORg0}jf)HV5}*Zi|1srzNSrVF(Tgf^xnsaytZ*(;}Aw7`&rz~bhh8Yl_tL%oJf
zo=uxdeg*^vtnIE(Xx`;kW?|!gI$dXGV(6*1`}AA8pq09MH5Gh|ESPi-?7$?Uh3WjN
zQv~QZ?|qz>-1_2Q>{ry|rK+xehtQ3Or`d5c)TP}<jfgS*H&f~$^8^BOq1rc7UfZEn
zW(C?>iM^UCv2EYD(PxM2T%$OXEeSRDuIW9oVKp;GiF*c!nQlCA+IOc3&Te^gmf;CL
zbu~T+iS;H*M!sQ&^bPcT-`HQ6daN&OHE$R%sG;Z&Kk@Z^AV>|?<GK3&nSz4VZm4`%
zF}~2;`@2nbn903df?$XO+?Nl5zQTC=(BNM#A^k+-asX%BL6?I1)ToX9C1dvZg7f7=
zpWzv+?b?^k`BWc;$+^jW{l2W9>A9#RJPv0sIQh)|a+?bbPCbsj*Zjf_(@q?3osQ?v
zZkr7eU*;(_?An{ZxuX7!1?*0;wouvn@+%zk@SzzhZ~c6O%eSN{lhRyS-t0py2GI*w
zIN!cC{)jWFx%(%w120>_Y79k^GHQ%0CX>v|aaTJJf+qA)SOfe2uO%=wTB@84Ycy|^
zh3|G!3N?RjP-Zdr_dJUVE3TO!98JjVJqG%GMG$Ev`b4SiRND|#ZGyI&M1LLWy?UBA
zbqqm}aT==C`hAzvb|$*VqT=I{2KK1+@ns_kw4$tL*N-Hbz1GHXwjjuH*9dUQ8~#xl
zCn=R+BdZZUCW%ko^tQr}jYu5p*n0oD^8RcFhByW>94061@b|{u8#0yiJAIfEl9JV3
ziJc=}Xzj5IintZ6+4j?V^8(MOso>b+PuS|TA^FT>SlLT?(uRPW2Dzvvr7Azt-VSUM
zc>hj9rh%=wJR(jU%)-VCPmvz%9W@Cb(H*)YH!mA}+d)oOH12|n*N*QJa`Ej4kf3h+
zwwm)pW=tdi01s09JmcASOT}+wpDn=yY6xFav5*Q+MwZX8xrix3E9C!YO5T}_aoP+;
z30;D^8A|7a-PJRa=TEj*Pxr@9Y2!4C)74*LH5B;WDiZYwZoc1mUPvP`4EUH2e;@AN
z1^EcD!?a4PC~#rXVN)TR2u);3XK|7I7>p740{%zvpwEB$)}YK$-l{|oa|d-)?7k46
znqI9!si;Xw!*$lyrqte`?I$<R3-i9)^^h?kBB4p#9PxtMYpUE`kr!~%^_Dw*($K$_
zBFe`<R|Yygxe<h^Tj(Eykf$jQ#24wNpCZRCUrE>h+Q0iO{!ta;Gn`;sU0Ui{Dui*%
zxHVoj;}L`0y`aCa6A%zUGxW-qomGsIt6{a^tx9*fzICwbzaSD$C{9tlX3+K@x2*~m
zZtnNF2AO7d-Y&pWep{2dY<6#TCA7O|N!Y59N@N#SG>+BHqmLj--sEY76YS)En_2Ej
zwa4cND5}lvzdF@YU6NS`r&gJxu$bQhX>GTn9Da-wV?!Ptw;q}>5|-}QCXIOInL}B#
zkHs!$0zK8#`1q@x(rHIon8}D-o=461A8v${d}VG4_ry-hGi6xUVJBcIjnzxS*AmiL
z16FxLLtx%YoV2H@M&C}_uax!2$8X|Dk<ywrtf-R9mv`R2B)}#}Yf<_27_3MBc$+Xb
zV5rJP@K>l!eYYJrhxh_)qYrcff~?>7T2g#u<SWeaCNFT5asjZ|KmUC*V;!MPyEJ@l
zPPf6VqjsqnoONu~wYy_uKlxVJ#Yo&mx2_o_z704eIaXwKHY_uuxPmY(AEd|GR%wJk
zTnpy54ox@CIzW#tpkpxqccHowrCBr_B+}Juqb!jI>wb79A4ZPt6WKr3<@q%Vz+H%C
z+}3KZOXmkpVl2dl+Q)2IgGgB-dKk~d)nEETwJ)L&=uJ2JpVYM`FfDv_d|>`0=s#w+
z_Ac!qa+Osxj$!u9kdF!B`u*v<3M2li*By7;j*fKlv`b-+F#r|uX`0JJ;-jhIYt41k
z%SEYk;9*d<Vb@{@uQS5JaG5+CGTUSJx0tjS+)+HS?r6r|E!fqKKOWfl^K+jpa{BM@
zX%^Q~&k|W@@RvUC!HcnsH!jz0ilo2@hZ(*x5-`;$x7#=plbAqIBJD%M8>lKny!#^q
ztMhGVhi88=6In|=Oop7l!k*6j3!}^Co^)K_`1kK0b!zG+r)(P(mlvwYGJQA~^_?ME
zKl>Jg4Vx+XCk<G@uWDERoJw}q!^9M?cU#+JkogM>XA3yL_2t~S;PN~)b|mmk7v}Hb
zfQp<ZRz^mK^D;wn3WuIF(tKRAo-l(fw}U;ec-D6jBjk480aZEV2!e$bF$A(_>RXt6
zRIzd{!NZ;a(ys9qj;f%6$kJ82>Cfyerdx?_Y^=#ApDf%nXfK%wKGg|vV=8}os?9z?
zO-#ND<RmTcMT2eBX<6#%tmG-b&6}!d#Gv8e-?X#sq+%m%dg5+QUMlNTkY$_Mb@xQ$
z?w}IP!el>KT{!l_BCh#k-qqo&O(qg!C}QFZ*VU7FZB^@xs$Xx%=u!)v1jAVKU+t{^
z(zvd(Ae04=hZc5{Vt?f5n{?uD-e*MbLB6@7uM*S*1;PCE;B8J8qbg5`+fo+qqXO#`
z_SvE7A%G7;nIEtEBn18ZG>SXt+T9+<N?!!#iBP(!X}9(kPFR$gdD%*&@86#`XAl^c
z{-eSpUiIxIZrvsWMi9%3ZwLV2B!A~pNBU=`g;1Lf=L0op>kBP8*$YG9&i>1|$=zdi
zcREqgG{Lx`7ls*IY445`lL2*bv6SiKE-c~VO?_P28;>svd@HL1CmW~xt$6^C&{Gd{
zeW4`DRp<y}6rrhyh1>2$XfV1l|04>68K#;(d4A#W#w?6LOg7h|bcL)|HClao;cIKr
zV*MLh`Cm2m#cpuML+R)=b#)fPP-ZANpH2E$+sLB=<_a0r_41nX$)fO)6W41nvk&yY
zm5N|+QdZXhEL_~>XWat5vlVXgC<Wu-H8qX|1S2W#u)8<I8aO)Vioedt{f8_^l`+Tr
zT&S|MeLu)r<>Si-JJT+|B$f!Mg8-@MMg$PUlke+fE={{^$cUCQj+^EJ*7IFJvUM`%
zpA}cYQdV+7i=Nh%Kz0N4sy<rXU=e-axo@xuKlZ>AyIXE+0KHHRUG81N>}aK;%#u%>
zP{ZWE$$Do4;{zx?PPZCnJjK}3gvpETV*7`^Bb^BaXfsmk{vR(-=-lKa%vfv;>0qI+
z?)sMhgyyM!`8Mj($gAVuf1-OWzI?P}C7;W1qL$%ENx}=#8i`U%m?^)10cY(-@{yfR
z`Cmakb%?>Yw_~@CFHibn#>>AK?hbDS-SIPkyqZk=XfZvfAYc@%Lg>6^F0Z*R<PPQw
z2_vaatd9aZj(0+aDIC8ZC)SUzX6XC*X2u7P6DYmuBdf7hKAfXSd6K5HxgM_imK<|C
zepF}T1kY%r<m>usY32wF+z%n$J%R@;=z#5;)VaSCUNS<@-5W5ZJ0RGjIp41{p4Ko7
z8Qo&3fgDij`8dX0<tKb|*5w7K3#*$Q-{&Idn-hW~-p8AC3+m4h4;O{zvfUwF1FjZ;
z_5(NmDnisCK$}a|Txd9m@#drPgN+LP$*h5GvJt{Q<d$dSj>a(W>x%}>#g7Ws9b0Bx
ze{csm<;uKkG?KvB7(6%-KafuBEx>99vWl_CpdX^mq$^Z}vMPfXVt~#4@8BM%N+jS(
z&1IV$4oLLO00Tmrz<LTIt-2%fd))i_8;>}qE$i1jD3nagzg|67eI++`?(;Zi2MM9w
z*|~XtAF3n`{8C&5GV$P_((9%+myKVJMio_t6_roBAu;3Te6^9iy5Z6pTx~H}_+aH|
zp*g46J}OPfayvNIS{WLADHHf~IeMyZ?*iwd5MY#k{qTHNC&SA_9n13X1Fq_{8xl3h
zAts@40QF`#IN)DGdeb3<)~+u13j0aHx^S@qS&i{#Z|xjW#*H|EE*5n7zzjNZ(ISUN
zdEbEf+PAP^Ntb3|T~8TLck^RTPkc}(B=i;m)OC{%F&U<+es*u{)V(^==)kZG22`-s
z3SjV9S|TqN55GEZ0k6&H`K<AWU${TtNN?-*LBP)9$;vnvrn3k0eU`s+lhtQzx)Ko8
zd1c8DSpv(N_^&l%qi&wY20hcr)E|L5@tB9C@_`@)DL`VdbM`l;Pb7|8m~1((+ZKNv
zd0hXZ$+6zb_j<v4r_)<#RDG$Ru!R`^g}fYunKgH<o6qI9&;fr_E_CHUELvkLT;%BN
zmmUVVrdt}=gI_244;j?c_Y`QW1unE@|Lff-4p=Kar7wShG^-;3>0=|hH=Sl8G2_0Y
zEc%a#4-8_%DrcCkltL+)D3Vwtcjs*zIl|1`o-T~XCPkvjN#1IGm((kQ?)o|RC$2vO
zf=eVEzG)O&F3RUTF4DAx^+%$71>e)R+gcxZ6i+B998bkm(&Hy_>&__npetHq5umdz
z0C`h+^_zIYz3=Z$mxo=$IAt(~#?s}}L}X2VK*v=E*cG%7`@UcMH^0sMG?yRSkY67q
zf2(<zC^FC)p+(te=dhsG{hv!<@bdCDFZncW(A%_YwOw!9<`Uh@i`71t-_!Hc@n}V|
z7k19dG)})U{Xi9PbNNR8b*DCPjKD7pwh<78;8;L%CTV%B+YoL_q#CAK-$1YoT=*WG
zqEYr#<}S-6YYCy@SJciRo&6Jtn-=i1V*sN-Y(9Pp@DYxVGTFZ66LRE>gGk?C47sCF
z3PtwUMwD?ha5b1D@QW$=8i$Av4)V^;{647VVFX>(e6<r!E;#Or!6s1lJR*-Nt{UV*
zIZNkVQe+*3o<Uy~lW!t>=f?pVxkpj}XeQ?$5xXn&Q(Z~B+bheJt4(;s{o=dE5<d8g
z?AZq+Cn^&NANjRlP3KQo$B(4SO1Fz$u|nu-z(T-4-0+#0a265SN$yoA)zq3Hnj@7;
z8q#i0z3FJMzhso%IipmVdD0#NN{R!mkLEK}$O{dh44hydNjobXt^&QD?SfIA*QVb~
z4;pPQGufEYh1CUABem(~#PPJ1^NC+Py;Zi+G6!)5RA<WamT17c5dWTT8rALny<d!d
z^`05?0mR0_8#4<;Ksxz;?o3OCT`2<(ghtXG1gX<9+bWaW)U$Tl1)qq+8}5GN;n5O@
z5g|rfa13eZfaQZG$wVD((#)d4O$(1ntIkiL?T7%7Vy?;&ZO_Ha{{<jYGv8ab!AQ^6
zH~x1QNID|`>8X8kcHAedGH7^ud)O#{0*JN;h+Hom5yA*QZq8$2@iNn_fJNQw*DrI8
zL^=TmJGYvdX&3ZQDYupt2|`IJtf_zdU=@SARp*Z{9c=<epw9>^UpFoF!lyz%L|Wek
zd4<gm#=pVE7pP(F5WqFzNZYYm#oJtB(R2RPanTFa8*j*7x~GN01G2@I{ejEb*#b^5
zQAYjno2pelZpFimA~$oo=bsFAU+YM_ocYPGmlw*tRyewcEb_nHD|pZ(#`bRqg^_>1
z@MLv+px{97->)ey%)fuTSyP8hU|V15xMfA>qJ*@})yXY!(kO8ZbkE`B1q95~p4QvR
z<Mb-T?>aFpu8sNe6R&>`Lq4+hYVY(N;w~<bTZjULlFwu!-CIb0S46iixu>HqCPdKP
z3ml37=in?qXa<|5cE=GOKDrtxf??5De3Dq@Ke)hUq*8*P(!%lrG1d5V?mz@~8KN<~
z%&GYw3;P@UK_Lr3yCz!d!^ihrhL9GDf<iOJ_F!+sI5x<!=uTa7N&Wh$>Ivj+bjK$|
z4N^6zKTZhBlu;LO*|%=F*K-x9RD^8JtcflHPo;w_ZjnDfAwvwvebN{?rX{obg|Hxt
zP=T)G)dq7hEn)<^vnM~}lnTJxtz@0}YS-l5*um~MVXZ1>&wS<J49|&OtH2Jj+RG{T
z*#+ohumgL8pU@FrmLO`d3}P*nsTdTpfrkW=koiU*d>;;R>)-F}(H!cJyPXoV#6KIX
znf?s71|LO2zfA6RVO*=EtUm_MD&PFpAhJISoVS-$B>~{;$gHfASl(mrFT0Ze(FG5J
zW}{T*b=iM#)_i33cOnj23(x0=iBsv7+}IZmQ0#R72(cH{vO>Fl6?{weggP0U<fI6_
zoSA!oebs>j(o^L~*d~jVypZY|e`GKxEXlw9;XfLg3lnvQN!H`&LMiXG_Sn7#$s{DC
zSJia7v;6{Mng0B9qiC*oQ^HWjjQYS(f4*Qowhs_z1uXV-J=|rkR(<Qg!5JDD&?n+H
zfcqHLaUE6fOq_XT718NhIun&)ECWF50QP%_Bc9rXqvMH>H1k%^JSJZcW;Iq%E)ro#
zZl`SioWT5U9vhv@)d`$`(6V^(=*&dKtN8WO`tuOai>08rIb(?O=g#kk5nA61YuR8y
ztb^The(x>l8nbiX+LBbA@gxt;7Wq%|VKI6jw+qxae7DR{xf^T$qC=Fo_thJz*$##B
z`n9#Hj4aUGD2ztoW~fuKG6<~o9Gy)t_XoY7p2aAY`hLK6WQ>#)eGwlD8X3&MWbt2s
z9<0(Rr@yw3cc{^I)WQPQp*AJQvY8_~3n3#@>CXITHC5F$_P%0qPxjk{+gcmn2zH2n
z8GcR<zA$g%tN6$dOviEQ=X~5}_2Hq~^wthF_azY#8g219pnDQ`v)^5d9yb|8ea;bk
zDUNxS`@l_V7#3_c=;&iaD@{?mp9(A*-^OJ#Ffj+)NX@}hU{ww&=bmaXTGy3d8&$Hp
z@VH+r5h5K}y9RFS7h+hARwV^3B~R`d^=WVU?`|vp=?XfaDwOm_$I4*yp9-;$f#q6y
zJzhe1;<yb{gm_Gt^;IswBn^Xo*reZuM_}OD$h89h)qwfW2XttZ#&+lNz@&21^RTwd
zot4<N`jTQFbknT1DW6yw-dxBO0pPRUh7k5q^kqOC?D(jT<?d{CT7VLhxM(|zXSFDy
zW&jsb&6j%Q>NWmbY*5bRd3x<LqWoU`kJD4NuxwU$v#o1V##I%7lcgRD66yl7bCa*b
zjHmm@U=_WBw?{63EOrT$cikdMZ`mzxCE`g(|GEn&%n>P!ahZt@ep^&An=zr!W3_5Y
zQNH~QadujO9?fphFh#0HYI$b=(`l!f{XC@#L8zh72FENsXTqv?x?F#_Wi02CJc=rx
zk-)1>@_>G(j*y5KIjBrV>tVVh&yu^2Z;JlDyDWDqEVx#qHL;6pPRpC#&?=g^q&^C4
zZ?lmgx@?~hFre8FJ)?{d6tpEzJ!yN|))}x5f4AgTO!HB#pV~VLb1EkL>V;<86Kz#0
zj7<}}-$>S%e}_Ur@hSZ(<^DfLwk7DE=DCxUZ?eS3UU$atK9eQv0j&YUm$nb@XH?rF
zuVE^~7Re@Qtm16YHt}wpRy!^Qz_+p~S-VI?5fz@c(f^pqSwY;lCq))_$%n?=|CUsl
zOo1x7l)~le0e=dqy9!yVKUg<dos-n|Fin$vAVT9ZiRu~~3Aa&8v50ZAnKJuEtMDsS
zR2X%{|C^fNO>)I7pikb5={uy&Pb5R8fRvHz-7)BxBOs8k#m+bEW}-~OcN+WqX(#8g
zj|*nbbn^;r<97se-_(t6kNQU62waylH%)v({lbo$k?%}UcC#sTI?$^Z-}fvfN-c(%
zo}7kiEiAGh_e=v?DqM=0U|@c=En4vs0cuCgc3XP<q6YIKT-8jPj=@G>|L#pudgx6R
zb;ONt?@=Y}{lZ=NfGj{68#cbH;#^rRZ74SA|3lco!gr=$;-k-H_c@X9ln2bH9q7z`
z`9L7+6j)WNgrS+l>2E+YztbV(kw316FI=gJldMOkb#S-*&QPoL8osg{(7n=U-B~Sb
z%%!m)9FEy4O-MLdDz4YlRal8I_s$aQxK;cKZsL;xqi?piyD3|PA4x6%uj))Zoe=)&
zD%5*jH!KDn@^HRfem$)7whZBWSn)bl#W|z`(uLpYDnmevxU-E!OC3KpO&^*7WyAph
z+ib8x2_(g$7o71}iH%ULaq7|U;M0E_yW1O5m_UvCbTmw{0ARDw!WF|`Ps<>T@8_@o
z*&h$WXqAN6Z?Acu0&L(zY?d>AYEH|tsJ}WVO+ySCZ$s?}R;_2AiPI}+?Hq^OupN_4
z<78wAIG@+G;_ZISO&k(;hHN?L;J|G%1j!C;TN2JZUR?o0e8)@lPT6HxU3?<#Ycby;
zrY%YG<nGAh{S%Sf=!3}@0-|VWgs^lT&S(6PN}AZ+gYN=Hw@xQKFu1Y3doI!S2kR^9
zmyL+1UCQ>cNNhy(S#O_a;3l1TT+7=O{=Z996uFzS7u9neg?IKXGMeP=jI6ULOyV`v
z7iHXhE&se*XCf|P@<%a`yTmsv2D|$!LS+hhK3SvePx8*<>F)t9s|l=v)qlo@;^^0P
zJwtR=1P-+-t>vAyPH&MM9{>uU0)HBfv{4_DuaJNI9@>NoWK_O@xFgO8^#>o5O2EPL
zB<~s?4Hwoz(HuudIy#$8+w}iK(|Lze{r_+L7)i26PBx*AayZ!|yU47=QBJlfgk&Y-
zjl?k`vS(zSV}%gMD&z3spzQ3uM>fad_x}FzyZFOj=UlGq{eF$-ecz8;hXrO3DDu)P
z`fP2G8lvX3HEjdP8_rJ!0lHWUn+?AZR581EDcU*ggY%HgNb-8gaRoPHtxqhe)zLZ5
zje_>F&jK^a2Jkp~>v8cm;9M+UCvi#kyLcV7;C(f4VCXI9#xk)a<;r+mdO@nqxl_aS
zyi_c@rw1G5T;}3zT^=k+^~gttwt;M=<>_PGDw?SH*P+HKqBTGt2tfI%H0uvkzszSB
z(tbW%If#Bkq+l7;0(pZLyrfBZvP=gC0(Oya;?KV+MYq1}S8l?U^w4f~VnXN2Ty>P`
zE#USR!}e+B#5T;2$d@bxuW)8tY3k%*-4F<9gP46?gc_+%Ku58AL_{1Fl4svG^o}i|
z1}AEQs3~hZJ^)GgBVYw83}78=M0v7aDP7Rk$ATi)kA)Qahz_^<@Mia%<d6r?=i-aa
zu9>l@SdK^(@0@E5Qm`udMxYz7spGcF!YrESo0!ThRB^u+?`DR*>>Y82-zVxt@45Os
ziu1~HhNIiZ<F%<F$;ruASSGpxpdBf;s7+g)U`Ty>IxrsyNbZ2WABskJ`@51-^UPuH
zKm2}Zx3!*2M&p_1yFFpuuW*(J;EaD<ZaU!jxO?$U655-r*0T4#L751gbQ6Gc05~hZ
zSZoGP4sOS1yB-B|JRWSjq933KC}CXx(K?eH+D0mh-*XGLcf#&_$gw2Sxu`BWgpHqY
z{7nKdlV8%(%xHN)4|eW@hxSvpwGINHuVpCA8uh?4C|quuyt52kEMIJ@Hu+w(x&r=}
zep&lDv<z13cCFtf4pf&p)3wd-9E?~I?}YUiQoFj^7R<|UFoAE{(kyX)bAFyUf5V$4
z#d<ur(Yxl4^ZQ18n!!kv?%4LJeWXxj88B9xbxpVp`j_SJ4OO{2#VJr`NU;XY%Mv_y
z>A{5NXxMLk*f|7K-P74tSrYk4BIWCf+dfKC;7o+sm?J$|h%WuXq07-fIsb1KuRJ(}
zY^9QM>uKc!e=G{h+_oYL6!AJ7SktIy*WZa&4u?gR+RIb>Snbh`1>O*07GOQp0Yq;_
z%9IQFz8_XA*1N!e`Nt)qMkVHyypFt)OB(e-gytAIEa_p#)!TR_L!SpNcOMX%$$2zP
zOiY9UgAp%=5YuPddivqxr19|Dk>qj^0Pl#)7f_3ikk`;@slY^Z^4x^H%wd1)Wy5Ma
zKd#31=i9yOkW!?sQ~F=R&a(hGAPwShYQCXfBOz@gU-ld}ekO_eNZx#WcM;iOszz6s
zKAtjnve=#$O1pwnSQg7yx4bdX56&84T2iaVbb|jqOwouAh76&!*3wElDz?KauEI>P
z*(ziq9L%e59-Gk6BmTZ8nCmYe29XEVkcP0RQ#;HtYtROL!O?CN$bk9-7>wln9bf<2
z4`62@AJ-*~wOe=3JnEj2hbX)^V|x<^wfmKzuRoz)o1BSgA8jwws~{EN&&Q=tE0{<s
zA7Md`9I27YQOc!vuRm5zWBc*h7;}Y}=T3Qb+^X?LegA!vmPa_abKs9UIr_e2Ga`cv
z(M$7t>!GO|6HR)=$qBI7qh%2tJ}G}_q3>=v0{?JrrwNbWB+V|q!B(3Lr3euYITYEV
zTZ3wS5CxxDJA#}g*=4LjK_#Wo*K*_oB&^BjAh~cIJf~fdn$Hfh&$oe-bt_fs-1LDh
z&<*EK%lNL&(z402U>$cqUCNq7^=(Ly0v)yU<Gmj(I~VZ*fHu{kZM@{db&QpXdcFSi
zLR@UFKgrl3GT6@A!CM?Bwp5+)m+QtP22rqoOa||VJ{QMi{)@@gLzP3#bmwZ9JUPC^
z2uF(ouvy8Ns3#Egpxfkr+oZhYjUN$7;OZmKQy)0NdpK`1&CeKC<RJQI;$sc!>}So#
z&uJrI{&f$iRK}Jd*Xv6hu4$Uw4W-+n*6Kx;Cd7LT(PU@~<cT_ee}vV&$}K2JwM*Wd
z%nrO|UkVZn;CqVvOn#_FO9`BJCdgow26rZ7QX{1?#9EYiIMSl#Dk{_+f1+3j$1%3$
zIh8Z$Zt8KkW%+qakr;rQ#zBhlz)=XjKp0dL@C;`K4PN`ZUTG4z_8c#r=EyrcxB#k@
z_s~1{v*5<sTVnX8b-eWcJqSj@q;w{T&lY;C-G==DG#sx9<U%}^`7|RLAKm}SqrgZS
zm|X$|dBE5ATAXQJYqTIu%)>Zd>HJO}$ivO;gqjhulHeD&MnQw&SG8H!X=#I%*!}^s
ziDdCD=PBkyhay8v%t!L+Uwr8Jp%ri2H@byaxjG49f-H7$h2%9l2%iRqbudqG^S6x-
zBD94T*3q<GU-m-ib15n53Pa;d@pjw1f!F#LmF-c3=8;7mP;_a6j?AOy&trOWx{~tQ
zspR$_B)Z6e?jI!gUn8f?!iQRR^#1_%)Ra%V^-dh0Q5Mg}q*;HN#zeV_p`7f6UklSe
zuEwPWyW)#mKuOtVqkh^g=jPe@zw__>u1t2Eud%PYcK1Y#BANZbQhRGLhqk!OqHW20
z3d{5>unslIj5A!15p|D28v&9luZOoG6#g{a3AP60d)9Ly|Mfu7Yw=Qp?*I?wEg#Z&
z`n-)CYc9?c)TaYzXScRGC{+E=KBN*RxBd~iH9KXj3LMRbGpdFaEeh9Y1e?fP`1#p{
z!VkPoiOuBw0citFe|~~|LUoDW`6>Z)0R`4&lsnWnj|E&12p7lv=R4;lIYel->e+7j
z#pVU+gupb={i@)PQ1>%HC-1`)6mq}af>LwZOE45!9D&263qV32$Wux^LD)LP1U)sk
z@kmsIaZtR|5#$NW**<ku>GR6G^ie}i3ne2Suo6p{C-H;Lf&3T#M@trK+QNY@1>r7A
z)L}E!Di#vW=cm{^RS6m6#nppI-3EmiJAO)R0@!p&@7zllg8&}UHhHB8K@@5%dB2Ei
zDkXV=-7H|AgsNq4_8S$eaw}m@2-N3mI(8WD^<mK_CHw!u1eN1n^+^&|P2hm715C)*
z1ZSt|QuzopX0g)-8HO7l+xd9Zi#&K0lP4~?rF!1S50K9m*9W=_`;5TjRkznUxn{p0
zG36HUsB}$K6s6A4^IOvzr3X7LZvJ+kiy{L$#9Q<6Dn_Swz2Oe|6C>%pPMu-0Z$0aW
zAF(hm98FMnz5m(nKBzx>Sa@3!sjjz9KXaFR!vl%<+b~=%@uWyWqIJ}3ds-#Njyg2o
zM3WxML9YS*^~qcN%dItj)quOzhhcm^N{s3NjuJu<?y8SBu2pTfAq*t$BCFFPZ7->P
zU0GY<u5GnadgHHWk}_XYSCO>uc(Y~1?b5dz;9E}LnOFX_4pZ4y86UL5rX}6$+Rv7Z
zeir&igUKi@W@YLv=EP5$(V!{i-fMZ#em?mvY3O^BFHCRBOEbORTXV#REV*wtUEU3#
zhc(nF?1ZC3g+f9}tyy+6?9Cp{mhftJEUNnM&p^4PYK-YK>xoS*FlWA;rWr|^Z{j9E
zsrDWW$+6H4=I(;}3wJERF8S0qExfD>eI@AZ<Sxf!FHO4WQJq@_LTZ@XkgqssW$q9B
zj~mA8k0Qy;$jkd5S&9IXsz@7Xx$wh;M!&Y$+cR(KUF=4x_9^;p5+GZmC|yR{U`fgF
zmfSm_>-hJt{!23gEfP^@>#_XxYH_QZo9XVf$YgNEC%Eh2E6QV{RvMCUG1zta6f`Cx
z#h|n#HPx1`txc;{azUmx#VERh|FmqozzbWQwi-CWP|<B;U}0Hoa#RVFEzk-<l=}v}
zQt9<4N6f!R4lXV#T$~L^@T&#}Dp*m>PLsi>yZmBLwSjaFr`8Apfa8IqZICBkYjwWb
ze5PszkSq1YE$6uRE<LjOY$p$LnF6M7*uqdw>T(?;c{R21FI`$oWRo~pjQ#F?x5c<m
zY%+$=2R1Y5ufF{z2$&Jm9e-jZd0h?NrY>fQM?iz$UfDE%Tq{Uy9P{<OX|?3900f+#
zb;knab%Iskb)k;|{F?O{g<wJ(Jm0=n-U?F@tWRGQ8mNSX_cPFx=j#|w$tH)s#B@@J
z(snBbobJ~)pDkwVsl;m1vd=M3*kp0WMMC0i|N5+)b%z7x-DvcA>)>R9e6_~k>CRTj
zsG$x<ExNV;Qj?&!md6t+msde&uOl}|72Ngm!so1KKF?Mf$Ff-vlJ&(pEL7D&E6Hjz
zWtx3G1?5sYqu(Ycg7z30;kdd4+4f7+KF%~!!mfGJQp-@%<cy%g-|yzGXTQ~n)hvAJ
zsXR7U2Wx?>mxyM|rG_%Wo!PhgpuJkKw^)(u1D~MUt|TM=-kT>wmX?-TmX;<5ibm4`
zM_cE!O@dv6SUo(b;d)*if-ZzVo~6E6nPT0&MAWll4GwL{6zgP}!)AtU)BR`UYQ^{H
z?E{+IQ$d+QP<hGQMb+i|G66q6?h{#BhUDP1%)q0Um;J;zcxDvEvsD5D_Eyig&Ly{4
zK&PDp9R%h@B3l~#(#6GNWb)uzEEdjPXjJ}!iezo1p->DWZ?wH-q+2E#q>5tday(M0
zylD$WiqpMd!y90!pNGq5s)4kTp`R0BEh#C?KTX;rQN?)$*T4c-_wkGY8$c&RD7*>a
zwmrLle&}d<U!U8D2QsN|krN_WpOa$ha{ud~<;S3+Ak^Ti4^M<UoB!>((1yw#)yD5t
zf@RG>>h3bOp~O3@+ldwm{jD`@Z`i|p;|2?omL7^pO@V1?_J%E&99BldHo`qE?wjns
zBk#{9&j<d%<5Q#b%?heb;!;34dIgZ(=#ZxGnHU{S(9nmyn!+O2lU7&i4GlntmKO>u
zho?G}6uiOm$fjDb07IW+9GAaPB2&;Cr&Y{>9b0w`^bLfI90e#&ZC&gyZ=avVZiA4n
z(qa#}jqsywS0lpSfU)IB!?h{~8J&KIl8^2b2yg?TlKN3>!eK+b@KOE`uwz+oBV03#
zGi@51s|<qLu7u}o*RxZGd`QU1QPTsBa*j{}UT(UPJj>5FO}6o=_n_nEfygP0OiX&F
zmGcLl?2I&@b%TGP3zI#4273Fo<YNLjeWA$dcK40bGNo(@4|v6T-MXaK<Zd(Mi&|+$
zgFy@}#K}@(xzI}W0yyg?@c+EJM|6d1U>sz{KVuL!9ErFsNdsrtoFS%UEryR(DxAh(
zUsR7I>L{or)S-s(3?;sr`FqNNf|f){d-FW%yt}$fk|aVP%v?GQ5uN5J1M2=Z8>OWb
z-w`jNzlXan)qbd%bc0S4R9b2E2`e$~R!3-0OiF$|RlX!J!}wLeu5_%V{&rB!d-k)W
zF7F+2@DzY2FZ$n;=mz}|4Zi+T8{UNZre~(@wmQsXz6)EU{H~JwDDmXGMgxU?&4%67
zp%vS<YZMW2lI`Ffz1|PR7(OlRzH32WRxpQx=nOJI>(-GvFxk_y+rAr#oSWsT{?or6
zZ=&n#=m52sF*$1LI)ktL0)D!wpEH<->p+clg7GSw-kQ<Bqb)&=5<B914t4|16W`Xa
zPW#txE57vPdTsrS{p)pI2G?tv(NrwfD2_OkZX8n25Qbo7yBvcE9d3xAc0lV=2%*G<
zHd}0k6v|ho+SQ_iEZ5TZ{`{9iY4?0B1>b^I2Z}U2y;#K2*y^JKzE4i5&_6yFYxR@h
zzXABA?v-2h4OMcuthQd`mPL8of50?hd#Dqw)1P+b6Y=q29?s>s0a#xizjlgm-J(<J
z>-H7Tmgp}3Y##b1PD~Ncg!QRvxWmjw)uuMXN9m)<71mVO=?phUM1ND=3Y}fI_dPEJ
zE|Nxto~8CV!%Le3#Jw!-(a8fYwcJN`oQC*+{e?4IVvIC8<ZtVzWeM@DY#zFfKJv`q
z{2FWqP<U~6;pq4IW>@@tk1^-7skyC+MKnr{L!Eg=6z>=6krH>Z{o;AAGCc(=q{0N{
zu*g*$PI~t0@sL8w<rceSo#co-9lkl>)5+XDDVd>dQh;?j^qyalV)^Z~`^!!5PMii}
z(TiK>Tdo%ZdJH{u7$=xSulw|+fj2eabUPFD9i1=sNf1i6Yb&1qF>9mq<B=m*EkDUt
znNr>={4T7_QLhsd^sq&(MSIIu5@k&BSX)h8{>S|!NtK5&;|_x2HgzlmI-L*zqwto=
z8htg#v{cMYcv+eaZ{rc@?R9b%C1Djn_IXm?L{D}=Bcp%UDq`=8VlaaszFW?1>TKOD
z(0F*&e<LCjvk?E=3B(`fY*U7$B)QAu>l%qDOY%m#>jW-NOo$l@xf_33viJ}BHtz8F
zE#R^DZ)vIZ-|<W%>Uz*SO@O(IIoIwo9W2qSPQq-&q&VZ<M{#I>c>mlT4UAJ^3a<=G
z(&@T|!Ctfz8U#6)!%Fv*RVSwO2f$`arfKqGYd903b++)UfB+<jOMvZV;;;+VtOE-K
zf_hYejO@hb$-F$1)jCh#Zoc$5-{eg%ga(X<^%kP{>QCXP%wf!c7N-NXbX+bv^>_g)
zf3jWjL8VB2qyC9i13Z7i|NQKq_r)gOsP%B;WL<(}k^mB#=AmI#fusg*31WM-E<8WM
zAGr9Tx?!=Cvt5JVgEJzuP**Edm73oj?jE=UK7fxnP2JB?oKOV_u-4|+!0Of>jh+P%
zna{GfjSa*B;LZLh42Dl!f8Mi^apT$twu1dUtpA#f5SLOU*e&@n`3a|lrAm}VbJ@w}
zQ8p+&ogA;u64PsgM5yggm*$)#*`y__k&%PA0-?&Dt04+YVri(z!Hm1ZxyY8&8M~db
zb)nA#I0s$omn$j3=;*?w?@0;5VB$fIcB<ki5E>4k^L0f#0x|GMBM11+g4myagq~&m
zo)?jcDb^XTB8eYmNS9fhPur09vWUenkqE-S5XVjH36O!xJCG2a$ftV2bU`hn*QJ;R
zcO9>l(_x4yS5OOu1HaS9X59hvJ;83fnt%#yTIG{xc?(W8rjC<N+vYnzeW!Wew>;N}
zh5#?Jiyh(|4dwPneHou;&i8(3@_B)G5qAW1R7TTu;q{Z|uR3@It)?eR2NGmkT~|7q
z*ueRQRe2r#{4HGdRO)rXI3I`CK0eQ!w=~{*yH{@#k`@z2UfEjg<f%D*@%HyGa<q9r
zzV0k(ZCgCoh*us5`x8h2F>;3ddf1{-9*BD1C&Nvtq!F5sEFvBzS#EDh2V-T1n7SNu
z*)B1g?`K`FHZjRJ?wVeZzln$@yYM^uMGT4As_APD`b_rwgnrZ4eyGRU$x@V%Fx>B@
z;yq(GY97rg(N|_!CvVy=^1}6xf{?L1A6RwZQ|nWk4GJzzb)~3^MaK~0UP5vNpcXQ(
z((ka-%ffPRi@$iq!RK&j#LBH#qSJTakV*JCjpSYJlDwBR`3d1<^5$kCDSSPUHefr`
zCw#u*8rT~Iv%mz51S8^Z44Z?3MeZfr%boccGYWw@6Q@h^AD(`g`Sk7n6R=S%t-Bsh
zAfL~^l?O?x$oggwdkgY!kKBOe`Pt{<=U1*#hPAv|GnDqgB2u|h)Gq0Dhv*o<t(;x_
zvWiko1w|=E+a8n}7gdAb`H@HPtDG!V;=4)-h&;b+dd{{I!81g;=sQI_));WwyAsq6
zZ+b_@4X3?nYQzJifA!q5^C!NX6W<iZnJKlWv)eZ>)d`}i(s&-z{sc&peS0AIu)tSE
zsrfB=H;S{*O7fL#BfrlJ2}{7&XK!`s^hS8^aGv!9`^{*95B<K&^>DbeVUNH~k^oDl
z{>e?fJIZ6(S#*IYDq$_}1<F(i_cw}~jEcgrCo6O9izjl;XG;f9%4?RN<qi1EHpt}?
z93NNtqCsV1d;4(8nIXDEhYqQWc=_2NE2%oUag{<Cti;QBqw5RK*h1yS^=O?OR8FC}
zK3UMzR#zp<jBGE&Lj3alV?CFy`;P^mq!;==^nC~tZ-$=}%y4|M#=Un-$IMr1<1+gk
zWnoSNIzp6~d=uIN>wGblY8i$WDHfywSVmHJUgUGGy}NQh@>lNhCr=sEzP7B?XQI+Y
zka#;_%a+qtzS@&vNbOl)Vk4!j*g_geXrYmg%xZ^be!jC;UypV|2~4*$gnXt0X#g77
z$hM8fkWvMby_X<k`oFgUJtEsR>88Ju5}@MD+HYRH)~y$$!~FIg(U{Vw?Ln|1uBT93
zU&5`FY_~3TFR#2`pMR|qe-mLboS$TCy}=Az(OPdPFo`T;WrbqRhQf$4PDnZjN#~U&
z<L<8p*8fx+&u>!POOxIDaGe!_*pp{melflh@UT1W#q9-NikFFh!(Dh}!A0^#xrAk4
zb=_M>o~W$5$$6$MdNA-Fa{`Kv${YMk_N-Oz6Hg6#ZA#m9kzE#^cFUv{e2YqG!a5Lc
zG%~@E+-Ic>oJ7`(?_9}daA|lM_A-WoId8!pbOZ%9HIk$aVi9xwM;0JwStU;fo8klN
zc&_-{X2!Vg$~&-S<j^Ntbv2WR)R4Hm8g}bPwy!FbFkf<WMamrwx}|v;!CIBt+Spi<
zucg}N>CynWJ>28rGiX!gxOcG*aB8TwWjMo8sI*rZ*VPTs+Om%aAe6Z#zBogprW{kO
z?5Ew&@(1Pd@rTAOkeGgt&!x90gK@ejmc+kl!CK;|bjX*AU1o9b+1)jvWq)}d75}pz
z*-3KzDhKJCoaNcgKk6s4Hv7HG#{YdY#`s#Q4<QKGjLX2Bg%98-7n2TnrInh#{r79Y
zjFY@IHwH#oM4KP3<B-=8-)>Nfca8GQdrKMW%_ZEH36y4)%E&v|5ia|_gten`Nie<d
z*`wd9B;v?W*MXxFpwj%pJU_G-<res_DjV2K+&;@#5=%@S+l${LG0BCuelJ;XC7+*f
zHAU4-o`}A{A&q<W^Zo+1tfQF3PjUA`%hn-33E;FD@0^B(7SXXBB242N;Nml7kvJ1(
zL|n&)bPMvfKqZ*OBv8|c&s1$I3Vorp7Dc9J+Q|m_S{@Qk{>c{#Ig5d7RXYdf7Bty*
z&HuTP78wER($%udRTL4oOc4)U2D1wl!Zf8S()0|>oXB0kyM2g-T#xD1E;70})`C-a
zC?|^IXO(kmr<QUwl~Pv0S3xdoaT&2RjnZ6O%tFBH4We-VD-N~dNPVd^Qx<*L2QZu{
z6}Fh7;dyav1!Rt3lkD;1zDBlxnW!#L$EKGzpF3o|@GTkN>`;0zAb7mesGW5bF6i0b
zV?uD3lroZnzYT|2SB|^b!Sm@pf5s^!N|%ZmaJ>?oJ8S?BhPoB<N#sDY;rL2B@A-a0
z)+G2XMalGA6kNb_dM&4pE!i3=ms|-~ys}j0W=&sN8Fcoa3eh<Y=@m1IxO>BIguw=U
zu!rQzn@$}jEH;<#*djc8%$-vi31HB`t(mk%lA8{P;#IOTt-Po59VJ#j4Yz;Y|Lqno
z?_Ck4B5%8-X?d|r#$W7CpA)(z0&6BSuYSN89+dMI07)_Z8)->NCtK%B^8|3i8L1Xs
z&e(Isk4;T)5e29LFZYwPv8XEISY<*ie|h1mYg$smr7l5rq6p`WH#^cQfq#s+;~(F)
zQO{p;R8av_B3rI*Zsp*WjxoITZr_)8;PrTbaP*V3V)6RIF^$yhDLCh<#8+77^|OUO
zPWc2F{>c0rDOCaq)7r{wbaZieV>IyE*(}I<rql1RtwdC;-6^8?{&pJ86tiBUeQLq#
zw;p3yM5mOX%zaKx?uL|zU}kk%SNwQFJQIPluiQ-~;Oy+*Qd1|tN2!Hs!};9R5f{l<
z#8@-vL*_0e&c@hYDI0VW$b@yt9&v=P>skXbkiMD<?!N5ChIju?XSZa_uNK8Sa?30>
z;AnG{RHrv%2fml&*C0M4q*fljo)E3P$u5#f9q+<=>4E-21DhlQYi+Gdlzi!g;W-0$
zj+1uRQP2BQ740(}?$(6Z@RPGQRVB{7tFyDSTdq9s)VP9LFwCp+ys1!=+Ibot3S*!b
zd%_m(zW*~E8glgVsk2KRt@X9h@EOFo=hFg59Lk(F8WSTJiV02~BPGP+?q)YVq6QHY
zQ|V@18klr<41#h8?P>9c+<HELFgbbtlmnRvTDn`^ul;tPpq^%VC+OCz6uLPk`*{gv
z{2TX(*z+yPDXgW$h(`MIm)v~ahk%)0#X$0jDn3yEe(XwkMae(~TBDYo=$n{vPaR`<
zzw2?;2a{5yR?cD<Go|0*m!!S5)a!99`_!Fb=4<M@XwEmo8JxkHH{DiZ2>HX`drtYA
zFN~+_HVdTQn>)cksJZ-|?$cz|3wuRF$Rj<$D(UiYJb~Dqthu2Nz5_po0mafN1f_t2
zu#mTKNN65P7jCgKS95cL_(KZi;1KR?QSRnO=Y(mzGAbs}(ev|?3`P@NZBN5(cvPG5
zfelTn%Dn{GGD<Q+YE39Ts3?U!977%Yk|+=ZWZGhX!Cs_|Id-9_BN~p<i92&+zVvW$
zZ4>p+&^i-i4&!}<SRL?hV>cZ*E=(?OUJG-HbCA|jV;*ASpSl2hq{weGlDtHgy<aRN
zX!?<yyD>_U%XUHaxd}~C<C~%nn#<Og`+WFSrg@jXvm4lAxn3J1vodyRhToWfG|9*4
zz*f)C-(@i&PTip`JMMt+oI4Z`9cSQVo(7JO&s&$%@gy-_AX?|CE`n;iVJ~h}-A}%N
zhVm9zj#GkSj3tDJlP5PP$sm!G2vXaY9Tf|_O7jykhV9qvJQ<dbK7X`jnb~;dY@Pr8
zJ9`UT*i~8vef=+&Iuxc;Vvaq3*L)7(?_$4HVtT_?<F2tSYlYYc_4j*%ZO|LOPX<k5
z6iP|t*1hH1fbn;>HxEr3J<1%*D57Z_UT8Is#o#Bst0v8-CkzJO9KAjB_XGZ&EX{44
zZNy)eQxv+qaxa`&_fTI85<NIkQI}xEugb&67kGR)-b`wjIG=seeAW)S_nFeGx|heJ
ziW}fay{s71+bOynyPZ8h-8uc2Egp{L_)=;=i1}GE4E%ABfBVN!mI1Zbl@`e*1dUL0
zQPu{LPqaZ#pQ$N)&WYss5J|tl%o`dbk?j(AQY=9b><z~{mWB8rsRO=tx%WoIL3OG_
z;t{&oFLc)8yhRqdcJx`#tPf0(G&N1zVJeX(8afKKKk+O%S;+LUI^V1juOYEddedGO
z^n)QGfT5j_kk+G4nLZEnh^Kt-(NXNk5y_;qNWAWvCL#mbtD*n0!|4x;YhUl#%zt+`
zj|h*fQP-Y}$U1Op+zvG<PW+!a`&@MP7_m;Mc8wn-IW`4m#HSTuiyn$lQ?jw~=U!zQ
zOGFcccx#uqL$EJVpODH3wvg08Co!DHm-aOvk}|IseEs_s)bBVUX({Np-S~H*S8UL%
zESR9!_a26Aha^JA-H={_<E0Xq9yYYCE&o4HWkb-QqdBeQpr)@dg4Q*JtYpNt2znw8
zX%p;NP|N0W51zg4?RH1gh!mrfHF_8`=+dxPx5kQ#o3<^L@eKh-pj<K@o6Y~{0LY0H
zd^Dp&N0TemG5BNDnufac$G}Z5G`$K9)6K{ate=mR_y4W!PAJ`34?*)#`MO!(|I`rM
z>OE><-gG>-ZgsXcH>S6Y7fH<Mjvr2Nz@BV^mV>~HOfbr*s-&da+^`Wc68{J}8_8iS
z+!`58?)M3BbErKE2SX>v=X>DH#?VavqZYSHvx3#%OzLO3dL(}jGp2CTcSDRKz2-E7
zgM43<9cRO~hH!2M((qtwYYRl7^+lx$-etu>Z``;nq`e=C5i<E~{-@IL>YeTc*D^Qo
z*R9NrkN+r^M4okTF4OBWb21ybu5U~G@)z~l=Lj%JQOm0<-d$M+v-t{?zn6?>Wll*;
z(=E+cIGa&*dF`FMtU5I71Y1t{EWz$?Eo#q-of$>2-ntPq4-Z4{f*;vaH4m46hs>z?
zA*ASNem%7-n(L%;T}Yt}N070YzW1u5Z@&NbyU28d5e-#v2VNLb7;zBI8mnR}am*?t
zkPiq)>vt(SB@eUQM}ZA^cJqJh@dL==`|GRkFSAj5>1%R^sVsc2jJWARO(T4uwNIt&
z{*0S=rG;zD<Uad#2<nUJXIxPWW7p>+`m9rfzTZ_ZyiG2p`B&+B1(#=S1|Afb!@biT
zfx&oqU+vE-9g$rfCt-&*a8r3OO=XPujpi6+@eyo;fAe9moC6w<dWXz~Ii;@tyu7X3
z8@j{^uY0?U;xI6VfvI8fx<T`j;ki*Zv8qgsTg>!M9C2%zRwMq(aM+?3k)UvoFDZZU
zkA8w$hY5W(S8j?r#U-P0!3-Ckn4e$0u5;$lyK-QIblZBlLrezHMr!7Yi@(oqJ^U9t
zx-ztJI6FJJ+y`u{PM1R|x>z8=L+CtQFjEt^;`7>x?fWir`wE;K_rEnPb#!a@6^Z^;
zTow|G;U4bQ7lv}dBGv!PkGwJATCeiM>mFQBRMFca+G6MCKlW6*T+CDPE~g=6mq2|G
zjZoT~)Mc+StR%fFw{3V3sWs<1@+%Yh+w};dPWx~|3;`JPoo1LHTeQDrml6wGj(4)~
zu;`YgzAQ+bQ>5+~#lPUR@+U8@6S4zN$VdFAs*MH*c*p-@elfpGx|H{D?eT)i>)Vi2
z+1;;P7>vhce|rL3@LethO<w?K&HNYpQb(KO1;yu2hzUQB=9kXK_C{0za+OuX>00{e
zHgecEZ3(bH@!vZkGEHwEF@?8+mO1<;c$%@e2e|tllrT*>t!0#4zFp;*6?f}_YBaZP
zGoXCFNF=F(Xk`b0KHz3h=Twm4R8aHQZdm`|Zyb$RDu20Zvs|g2>MOZ>Qpsq&-g(J3
zW}eSjkWSHKtjJS1f&RS7>yiRzY|B}VY6QFF<bTYV*r&zjC#Bf~7sp%SAS5`TzOG4J
zq?lHN&3**^MtlrG!!P_q!(;2__!>(`g_W_QcOdZ6)YRmX?VF%B26U<DP|7CkW`Vd8
zBFqp#4Yb<4i8I%>VGD`Xui;}9H<vEIjr+t@vFF&#!$?|viwT^W$`($CXhAv+{{992
zgrTYva?zO{S-e6^NH=~y-W$Oc+jHM|WtZ9I-ws1{>3=E_dX*Tb2+Us=n-@Uuz*M83
zUW_eg^V!@%|53SRK;p8a0^jgzk%71<1ou*xR&&}V3{GUj(u4<B<g!p`;EO4PVYig;
z1v_LQ{UQe*mE|Yk_~IJYJ0znmCV>6rzTWzrC{51#r^dfakQ}tN^GmyH8~*r{xf4|C
zlcM~=$|bLd*LCRW?`wD>cDy&TE$(T9R7+h$q`<poUN8rUIhj4al?!??IxR{R(#53O
zLC~ePwDu238?yleaX4Yg_hJ4^j=S~_cfT$DyWOtE{Ag&&F|`1@UGGH<2j=Tdd+sCS
zM229=PMF>k<_u?glTRSrgTcU&8%E#iz^3@eE~M^{ldCEyN{q1^^gz<SypKRYC}5w+
z&vD!+@*Wev>hxB!{ZLE0zSB^p6AZ&#FSuA;^l1;GixJ6xAoKEZ5u`}S8D@<`8`Z(|
zcSi>AXiBk(=RbW@78M96?Sm~)66<f2mH9@hr?+1BVwgjoV3AP?iC4nl8{SHHJG*y}
zcLWMh&(lP;ct{lQx$F5bPBNT$qQeHLmwSBN0QnJ2pD}{&a`d_AJlm)}Zxu)n>HWg)
zTKiR7(2*z^b3;L>t;(GWC0!1B!VdbHj~6A%vo6N7FLubY!1;{H&_elh)&N_<ery&B
zAw#u2Dy4`Nd0oUuybs<!4mWluq!NU3tB1b_<#-GUysDWWvu{=c$a~Fx2a6*r^o05S
zZ>m!h6HO<x;c{+G0BKA7i^B40Ae7|3L28M{#lSicf7)4wzw&%k65X)d@Zki#4(EGT
zCAZ4TArL?S@=#0NWXhkkwHNrXY#m&U<(uCP$8x`{>?rK@@u*6twU+7Jdw0CE8a{o#
zWp!oObwHbwp$94%l$!5lfie&evw?P$JsT}0g5y1{UizU&_pc1G)X&s6T}6nE&d+`t
z!MB3Gwmt9^M?SX!lzE5=*Bd?7?<Qw3M_*w!0YuoaJ}ZI36{|p?pmlU0wBDVk!!gT-
zXl8i$WMSq9ozDAv!}J{tYJ)!@g%!ecY>Z#Mc#G2%x5EWPydATKzlZ3`ML##sH>IGu
z^qG9_T3h#;TMR|ag1;{~_C!;vmC6*-hv7H9tFAK=>U70Wd`x?V>*-ZhvFV#*Kkjk+
z04ws;Mn661bGGqiQ5MnF31K2%u64hBvv)G@2I#@^!He(MP&@b&1qP4K&rS6n>!8FS
ztkR!Ez}072DS<0B42pzT-;zPs2eYt#`}O(WwV85Jrf_!U$P;FF{A!j}<~{u1O}K2U
zH?&WV^T~_JeykY9d+F0)mRz4#L1V!C%7-o~DpH6;>aZ_dn8($5iTYPCL}lb&;eGy&
zQNdZw(mE6At35GA+oTL+DcBmV#SpRG;0(De>J5i1CcNPA($BdH_wMz3)2pD}<{;a*
z-jJ6drq*TVQEd-{MM1^BKqCb;*B@TyK#Aw$pcFHLL%XYQ`O0ONC&ygJQGkv{vBLI_
z4VuyUS}D!a<9X(<^J;m%Hbh1-vKB41lppEe;k#y@rO26Ltj;AYE-@k{ed;fLLPU?6
zE+TGV@JFKDAKWu4$;X1Ct!&%uko(W-Q<88pK`b`lDGlB=bf+B}U3mX*jAjC@N&bS@
zl*=jfrP+Z$US0Dy2|0A94_JQ}=*|+(caPxC-jd*q;jOrh1Dj^b+s@?;w92-_qN1XN
zhL|I11x{xkSI5Sg%a3QktS=Q?g9}ipNa1ycp=eK^noB!<@IF`BO#Lc**cV$m42s=^
z0z7cWm*g`F#`_EsB=>oR_C>hxt~`C-^YU7f6>0i!s@~rake7P}^hsnjHoX1U9`(e^
z)teJG0$LZgr<#2IF1}s(djxp+k#=5BAGrH?4~NSilKT&S`+<xBcTDhy=Z7f<l^EPF
zvJH7}oay4caC^#P+RYok4Ek$M=MGNVKQPTcoW6J-s4!cIGeg`0yR}<qy)6qXJlgbo
z1#=1#oUrXYaUK9;1hM2HG;3zQP&FSTo6Eq;D!;A4iNOw@9=}!IVDgbW8p(EL3ZK6#
zp&&wiqhsw>ejbwNMhMHTw2zvYxXf05pIEZHUgomRO3nD=c959z{p3H`{7BFD>bbW5
zl?V{kzyH@gwtwXT>bnUF_1r;QEn(eBjE9nC7@cgc#|k`1jfqt#C%z6fl7I7v%H)-L
zT9Co`O5?bzTrwxfm|hYRQY>HHn%y~{U$&{@?r;&W=9cRT?)QIPaW%*5c7`#ELEV73
zRu3QWJf7W52VIg)v8wAG#P*>*Qu?|b%-&)UefLcqH7)$DT+G)Y)`-&;VvD<I8{OYQ
z?FTO--*BI9%2D$%Lxc{0eoS{A^|tc=x6!^=wHHPFv6{$Ad7HJ(Z6iVKhVNWwR;uE1
zp}pDqLZT6+2pGZrPl4lM`x{~q^+NLZ609lW`L=R!4#o9PVx|<13Yr^18M>KQd8&ki
zEmXByof3k_;ZIS;lp&Xvi2bn$EeEqMn`p*}z_gE+#Hs0fQ-+I=1p0BbVQb91N`y8o
z_`6+-pxcq`EWZa*?Hq(4*0{l?EiG6f`Zk5aG?Gh_6L^eLE23VBZvW%{PJV-G-gPvZ
z%v>)oBBF)5WzQS(q9sA=vTiZYAyWVkkDqNesG(sO0(s={r9z&qw}Pgg*!pB7k<Y<w
z)KSH4)E3_s!KP|*BkLZXbqH2EQ2Aq~!Na!u#9QvTQ!DaTsO~XA_59>;t89CY3Tw9u
zpPHHiRHpg{0wrJ-S3a?`Nm1n?BU4#yUe?#&yhX=CUEsdrGoC4UrS9b@!~CdEjD9z|
z8;3Ms>AXxIXQ7PuKirMVZU&EFlP6M$WB&m1#kuMQsA}wu%j7fPP|()OGUuOjfQFT=
zgM2M!1b$||<J-H2zYI-Do)<fB13W;IBng;4o0R}g4_#2?6N!sikn^sp;&=GUC(ygH
z?93(G>8YO#^Lrec@37W6&1ZRh=F@w*Ur;-^Z|V!<$yXgVP@Zr{^fv$DfVAAd_pn<<
zIu=RP?4Wn^0p<3yFR=v)!(NTe<I__r6Vs{%GKulkrc@ttiyw+#SC6Arqa`Xe)dR}w
z@0f0>{#J3bhTYU!3EwJtrI(c9L>ya)co?Dc5<?>j7^phV-WW~TIel)bE$nno)m1l)
zCg!lxQDTzN#5-8NBvh1F*|Q{Ae%<h^4%!ER-9>G;gpeeK!m@4jc<DVqZy5>4v?NEv
z{7(0W{3;d%kiwH4Gtj;`<lqCh1+<uLX$2+aY*hquJ$l2RI-4aat4>|)v_7-ljVXhB
zKk2iYva$lpkXHIxn@pU=IsE%J>dVSgd4idyl7H#ntQlKf%@j_k7O`g)l>SKH)|p(M
z6>vc0Y(6J6tB!jKvLc{gbHNg}h?M7=>a3Mc_Qa18VYps*`k#KA#83SS0P-`k$}ajh
zlZ7_5LnLk1SjGztN@0l6_w`e034%RMHPL%LqZm0tjS6MX+*7f&E3ZM6ILm)WrgI1s
zt8sV_f2GQ`H-Po&7)DpkG?ZAV%1=w21R9Dw@n~A-x%NS1v7qo18WoMv9HH_vzGPJO
zM{vE0$jn7x_ci;KhAa~T=wci({uS5dEa>&U5@(9GfxAVi*LglWC<VU)SEpQz_I8RI
zwN{W=29Aqupdehbdhsna6`?1zATLx%Nc0OPRIr{|1GHuv{<IcHisxV*ZNA0Nv=(xU
zz9XM5O72hPM6#L~L3rrN84IU-2DF2|??(BDtZT<JEG#EMng+nnB<P)E)#a-x^ou;-
zN+ecCCM6WiBT7@5V+mw$sh;&WM?X}D2M;n|q-|}vO!P}u+F3tXPMP<vg5~Ro3(9N6
zYc!u+oFD(2JXk);PHI&bkvskN%eY8XNa=FWyZCmrv&#2}ca$#cz&(3U%3_fN|J);A
zDU=FPGE(~O@`eIi>CG1h(-)`B2jfvj*<il$f|$PDL<UL!o9+O=?J}>@cHAwHCmYPg
zc6nQKqrQphrxe>UEtg5d(%ELZ&&8?R|8kMpuG;}47iUo;(`S<-5>|~9uR1#FI-0K|
zxLdUtsK4NbknuHVe4qD<mn}H+y%w0|@R>tvjdC)(0Cf;*LB0=`d^lN*gG!sd9cM^c
zF+mN9eO@x>I~q_fMDDM;JN8~tfA5oZ<tKA!XHEL~VpMl|(>o5oLNvGfO5sy*z9U4K
z?02h0;dHCbzkd*@*-Q^M;dWq|S#2)Ze68iI_;+dRNX2tfZ6oR_%7(w5A~RLc6eZ&^
zMRlDGNKOM8<4JYfHCp1SoE%)8rbe}4&2*|g8T0Nd5)!>}G3#l#PIpVUExxcL{$YCs
z<*o3X!hU=|+t;fx(4o8!v#CzB>cZDxk+KHi{8J5BM=}_f0L(W!Tt@}`92w+D#h2BX
zGd&nlE#KMI)wJ7;-u)Mbk%VZne)iU7p$q~LV!@RWno+@-+G-r7LDU-|Ow@U(9I=DO
zL1bRht*UiG#oliUv*8m-xsMr4Qze=Ex;L9E*ibILmnelws!ImUuk2+g_cIGg1v`oR
z8azx+#Dp+`cR-IW$BoN}WlLf~E9NWV=E9}jcS|)#Sgv?reMs@^X={r^QT;A7a#H}W
zs>+m+?LccnfiM=0Dt5YmXF2j^ui1W9=ciGH@`WCXcPfJ?fn8S6N}L`2RE}xa!#_9(
zHQd^#Vzt&n_>ub!c_IGUn}1g*W)LVr?cNAo?xDC8kA(>5L@&koBb$}RJ?f{G-$xZB
zZz;ULE-wM;(aL(Jsiq}pFYJ1MzRM7xe2QB@LZVxEQj)x~L^Uw2G?{X**Bcxj^tD)r
zD<Fgvl5v#KZ{tQ+KfFHQI5~Hhz<NI5Y&By^S-KI&8%y|JmOGcZjX$5;RX%N@cFgqX
z3R}A;b)!h9m$lMFB67iEN13tJfyY=k|BlW#_`%j5*kd#wz8))QpIPy|=uISWS^Cp8
zhrhOkhDp~^C?jY}N7Icqb2<UijQrcaI|`0?2qmOaC@w0!syHgSJiGDij}Lf!-T@oY
zj#SD2Vn_G`Q(|I27D??lGO<L2HMxE8=d>F+VDP&hVmBz}n_AFH-UQFvwQk^8b(18r
zm@rIqI36m$m|dPe+gKZ`^8nz~+Mz}?+L!uel}$@AwqgYw`Y9DrKYgF%^D${luV-+d
z+E}2U{#}(zd1IE8zwlSs8=`ZwnH|mDepfKQsLUDp+ujnKILzC?>w<6*gYhW3%wM)2
zduWMLn>FO9<z-IJXPiHr&R5ZLor7Wr{U0`6?dV`A1vo0L`D4Xo4p3q9A)8eu>yGiM
zFg}<@>alS}#%)OftYF0v^&=0k=|yOz(PtwC;)KNx>x+-RUnxHkyZIL(jRft{nyYsQ
zTx$_6+HLdF$v99WnO{zi|8uhS<2udjrs*tF;8#%^F)>u6(<Q2HbK1uSqj4q{1@i=^
z15(xIzd8T3CDRE9;7bBqYNdzF3+tWWp%z>UR&ystnT=w_>VjnS-`EllRnUN6zDW7r
z`T6V~VI5sd?o`Vq8b&EI2JHM~y65x>Io!Z?1A&LukABq<TZaLk(ktGlJgjC_=dc(?
z^`y09DdtAJM1HI3(=a}S7?|X9*>;PFK0gdJ8%dDj`sASl+%b&;eM2uD9ZF<gr00Yb
zT)(rNR%znD{}}gY9ywy}kw``+If+00Zqk(>$@Q5+fTgz8D0@ey(?qIo=ul6PDW*T2
z`BrKWyR<-<jx5a<RWTH)Y<kb|q|kN}+Jewqp;CIHD=JB&rg<0__r}1(^49B(x~Rvz
zccM~nLFE_NfYC0s`=j_4iqzjdP8P<lt6e=4?EXPdoyDthPRl$9iG21;Lh`+iehwSf
zfH_oHexrfyz55EhtSsZGv;U(qFTe5c+^6(WMH~TBZs2L75q}doq@kaGO_+;7d}0;2
z3enSb%nWv{8vg-LaZ)sVP3M5Crb|7tP^GuywBu|9m?VvS!3fzozHfVan)k?ff2mFF
z`>=5Vx~lKHUUnAc7Kce{-CiKiI~k?ksJNr4Dvy^O23LqIw0<HcyYLuLsVC0^7t8Xm
zR05k$x_#U(7Ol<~C&#xR?n!Qe%=aAbj*m>iso^>NVEH5g>~YN5Cc)cgc;gcrcwEcC
zk_*4R&G94tX7X$+*}bfq^_MrE45|)$?VJ)F?##buQm+u%9~;t_vN<#eMVeju#D@2h
z(h=`goSs&d6xi(bp8ru6CB*e7?Z1YaL?@K!bl5iS#JlQ#ML8|Vqbvc*J@>%3Z_Ss7
zsCj~Q#eHpPM5#-^L_Y7G?LV2UQB~p&Py?QS&F+L>f<RP`%`d50c;1kr<SX<w5J~=<
zTf;R#`P;DQKUv(nU7)&-IvAI$LsXF1i_EH&^i>;rFwI$IZ?^oMas0Q>J$%6u^JB+X
zYHF?&I!4c4rn^*$<(bD2<|})w@6z+KV*fUj(8!mcENxjji<b5Zqs)xAycP0IoJ|23
zuHEFBw98Ib(NPR~>F-^mi+|y&(VELw)Ca>|Y0$&^qUmva^-z>rvd3c<x4ExAJwNM`
zI9}ac_?Y>M&nR?pNf1%J`ntpcHTwcM+C5pfSkWJJszxSQsqojAZQsD^!d&y|ZDTT-
z7BU>LxK4fQwmgTdQG!4<W*9A#TnDIYKkI#lHS-8Vf|KM>O0%e2TnLKK!3xXZGS>R7
z%vn5tGIr<@g|5pA<Lh2Ork{yvd&nqu$RiY6@akSbeE1z&o;FF2dnIYq38S=aP8{l?
zNrkjJ4x&!T|E`3*ENbgoIQ-G}+b8<Jb~df>V6`D3bpku>m69U%*R>w!v**AU)!z@8
zAPe{EuN{<W?bp&%KJfgp*&!cRWwFf@IWuNvRdu%>?#T~IPjwgpp|32DF%@YL8#U0b
zpqsOvHk*<<er&4#<wLOm>`P=>{=_!_$WW9RtH6otC}RMBNcRxm(NWgF-MNj8wGT&U
zC$2z%Usawye1g|mcYShGdmEUU^C!)X4Hb`+r}s5rSfvE+_<CuM9J!CuJEk2C?2~H&
z-;=XCa|;=*hsr%Duq3V^iPT-|Htp|s^!7!5iigTuH7XRO=df+lxxm-o>07Qkf9oj0
zeFO6?aslL3#85oXWMXVX=oe#h;$O_jzk-8%$Xszbum`IBAixedJ`6wd@sZPFw&S#Y
zb`Nnw2;sX;Aij3@oni{0EsNs0Aqd9!NceTRY2b)7eQ{1Ctm6rFCZ%=eS#A=&eSdA*
z^f8@N74x6(_LnGVpqvx`%M)@mW6D+yXuvmKgHc_kt?=f(;q_C{W<^P+(;<J*5gikL
zC8s@DNT}F}r5M1`&_k?)Ix*%|B{1TE=UZj?9@0ZZ##;JzC|3woRs3fk_K<PcInMI3
zvNG>p5R29}DzB`%W^n$vxZ3om9N)(ez3nS+mXJ=}efJi3W&I?+L)o1zbQBrv$Jy(b
z%>Sfay7WZz#nkUHhmCNS)TH<?C9r&Ng~#E}X2#-?{X2l5xHuK|K=Zb)ORV&A!yUcS
z#8g|kZ+o!BW2jTLSbfQ2i^p=O6o^v>9ZHc2i<q<uQFcw++mW=4(X{V<G!lvz68<PY
zD-Dh-Jss5}3gpWi{HZteko__fuo04m7?AgbJsB|3GV-4^UUYPD(sB@^6S#x=F9EC?
zfS>l0mVxrHCDN8s^K}CJUB8nUEtm3gf<g4RZUw1+*V;w#ho#;1CEX<*8>Bmbye8#i
zunYDU!lhGw#=kYtdcToD4IV<G`DHSN=pfi)h;aT(m}zmGvzICrL?FCB>fJ!RDa`E3
z`!t<IbTLzJ9dpCyJ3{eV9<Y2g&P%si3=%%W0tC!F)$JAcH`dZk#5w2~25XrbmvlYV
z9?T^CG~RZU@}yf9^tK+p4AiXh*p(f8T5B3KBKTMR7yC<zT=bF80<&SRl>EDdLbc-3
zrJzF#xtPh(%ltwRFLkQ$mN-~XIN>{dJ*^)>SCG(;xa+&^d{s}ej?Rdx@S3uNju?oT
z0s@0ZsdgLkye#qy1m9^l(e>zMEu-F`mo;}xhT!G!Mved0iLJ9+v;V#=Y=(1F57ckS
zy7TyXPmw<T0Us7uf;fHSjp%?4@G4A#e$Agt#gDc_1EK|#`Ip|9%MJ*mM-o&(<JaLP
zvCv4nD-$Y9fxU&fE`UqL1;Mf-p9%n>VP!j@P)5k9HGW3@jCXeQdkN1|2p?0`dPkrX
zV><6Azfsk50&w<^sriEJGaUd5Cq+Zd0Xt1#6qxG!AIfOna40E<ze7-q@Z9uILP7lS
z{Ol~420g)VPdAV}18P;v>O$0aw&u<@c6T+Lx?oSRQ<ohc{JO@Fhf_#S0u-&Bx57XD
z0f{i<<knJL#-{AJ`yH72e^Uq{!@b8w_DX5NCgOEzl+qcBwx#IENZDgCDFx$h*fRBi
zvH@Y<Kf?(t%3*@<Wl2363>njJSCV6#Y#(!g3P~l!4>ue`3B+SN%X$7<+L9%iMP-`U
z82{x{1l;9^gK0Ym`lbMk(xl{FUTsDLzY@MHYb)eO+i*B&{6aeE)+wxZ)iK+Zehj_i
z;QMerZ6uc!!oVvHXSG!_#_+_mA+TajKmC@?M~*B;B35SQc^UQKX~jItZL|e<J%2w5
z_A0H;^A#IF@Wu#5s`bQVDQX(zZ8KjoXfvtif~MVA)_jo&VPFAF{;Ov{wD^k0a0!OZ
zykAbhVq^zy%?|Ju9#jO4W8GSWJ}r3ysxGcElWOS}(?_5i#kEG0zp(ZF(BFDW5fMSp
z$qS?LbQy*C=VtGGez()pv?ODC{==<(@c;6t^WwrNOw;Dz^{BB(W@22$4<*RB$pPH|
z(cAq`{;TXZ8fSulr4Z%o6zt?;#x}jZ#L8RR6J_3?5A({#|N9tZJ#JdJo-E(CvD5=)
z?wa7S_+BG>+^?nU5_9C1cTJs!O4L-%!9jaFJ~_a7eYmEv)XzEpnR;h*dP*HlD|0Aq
zt(fyrDcYlCt=cS6CE@CzJK7?|CLrL-(RS^KXsHZ`=PYCdpdI0jrHF|O%dm|euzB<3
zfqrnCC-$_yS_0p+RnK+dTEZQgAoVTXc_j35?x;mTS<E%ISi3C7`PsL^d6h}AO#DKH
zUZu+euQTT@qu$GKz?PzYc{HN^U`u=&!b0JE!%IGK0bg?eXp#O#XhL%TRDZNTm`?z$
zrHS>mXtcOBz)B_ltdwS68rL;md9k~>dt@}-!NX*yXymI%ul1W^**CxsKT%dY@UN44
zAO*C$d!}``KWF6n<(U$A+G@lU3r5kgIAU=^K|Z?xe$w?~H~Hf5WI(wurfh1NfFr{%
zZLB}NRND!qr4+hxc!$_2&FIJzUK%M#Wiz&HfVKuyL5kp(WH|}Jr@__ne>9!@Khy93
z$7fDU%$#b;X(Kg-%^`CZ>Mh66^loyFLUW!oLt-R{k~x!{HF~R&vvR6N%3&)rhny3V
zQwZPJ=ZEhfV7J|Fc3s!=dR&jk<38hAx8iiPRd=-IbhN#)MPN4=m|zlN;6xn>@|%nP
zT)`gYF8y8DZqc{2)rU`agWLKyR<^c)iSIWBIB>$-2qn-M%#^<buv}*n1qT(H*CJ|w
zzn$Q{JJw}wU*l6=LLKdTzZ|g7JX$^iNL}yONi!Q8FgL`Izo0q2!I7dVRi4>v(A_;%
z`MumBVQ3KST|F-W$;sBN{MvjENhn+(IN%l8s5M4{4v&YCif8$|*+h(9zF=C~SrCiP
z;F_PiSYSCE^jDGj*EbXa3X!w@W+k|g4px!0>}qdh^b53!w%kee69l7o)^IOmgKm?a
zYiFN2zjfgZ*fPT{mG*ROto4HNA_(Qt7-!&7%h;|bxKpVD^U#ql`ko*{X1j)4TPHxv
z64eVA@kj406O|&oZAK#%T>-eYU|Pj}dHG8B^YmNutl*}Gq5gc-b_;(*xmSOFz!VIL
zdX|*!6)4LmI&QJbFE#Ago!yenL*kl;fqCVHK_Elt_?Tk9$w94`vK$y|KfYSOyk&|q
z6Z^0D_gx7aa#uQMAhl(b0YG{jGBu?u2$L_WK3^=qTjD;s6U7O}d}|<nV<G{OyuVvh
z0CUaQmGk+ViTSZy%@Kx|_l(*pR}s$vgV9Tx*u)snSm{MnMBA;b<yHbI!g~P6fr9a3
zUhr9r)POGrRfBPD3RqtC#j81@a7~#fdGNRyHg=>gNos7UWge-Bwwr2t?HzC1orow5
z)SBNkxm0@9yR4z?HPfQ~!V1%(Hd&1GZ&uQr?`gJ7Y>v<$HL2c-S()nuK5=B4QFwId
zA7C@LHEp`L&s?ErhLYMH4|k3>i_XU{0i`7o1OvG3&*22A_aJh&k^}iUU-U`0%}2B8
ziSVxGT6*g})<vjM3TEyYn&CmmkG0G0sTEwWHfg!=z|rUI)-?LMdtYY4{pS>g^KWN1
z`YIOE%-4tOI!vfD*xvVEgl}Vb(1!%XyvyfL`mQHl(MH<czLCs!=O{&*Lri5~Cp{;J
zzRou<yyZC?kp)_uerfMPSCSFN(D=m?c1(zllxwnUJJVw}fy+mMdtfFr$&kE0c=Ts5
z|4PNLr46Q@1wRI;hQ;>6)3w#qF#il&!mLN-hg>Z^->(8Ia?s}DXFGfQtS#2^@?`Yk
z+*UZ-i(uhK;uow~vI5piMNxEQBv!f{QD*#}zyE#w*Rr)K9)>o-EJ>N!Xw`A}dyZZw
z2mndZukAiY62C}<I{GZPS1lo~7+%j1mDPG$;$hN1b$jBQ52*9&VAQCMDS?(5HOav9
zWh3FwE?GTQ2@zAh90Ob0-Cf_^^_Q?E*b@e700j1bS66n8I)>>NR0L75@J|@`v7IO#
z4<XsxoVLr3^mbFZ(8knG?dPINs~flc1D%mba*AdMmzJ=rB5l;e1Jk?nrR<^S6T7mc
zn}cxD9c!RgJz(l<#B<p9Ab8GA3!`F0#qg3}%crs`yTC-Ghes;?H3j2uz%AxGx>E)q
zNY8xd@%T(sKnQs1qzt2^uXy-SN24eWEw4=qIa5*SJ{LdN)LjEgz$q}2blguf|M$9D
zZ@N=^lx>x7os+DqjBy`f@*Q>_J~MN_3cwxoTo2^y8+KvsgyncEzVxLrL~SiO!T+Wu
z`7d6YD5DMGMTo^+L5@2^xs~sl@44Fd+neW*EfnKluI*Fm>dMKYiI_Tjss#=&a^AJ<
zs>o8aRNzJDj(^N$!2XRSNK0OrZ<Cf7INcCpZGIUy;w566VcJ!k+VXDfo;RaE>8?3y
z?mFLqR~bRsY`2x6TocI(vKIZjWAf_g{j!GCX6;?@>Q2(5yvpR`XYxNK4*Q2k>Kttj
zTWze^oz(6&@L9|XG*0>GDDfpF-IZK|r;fF}gM;?{%SyU``Y6bg%2CG9XD5x<g!<nj
zHhrQe^X#}^!M`EY2xT`30>0juLHnlhix`RCs;814HvHa)NhB=!VY}1AE5AAhe*qWA
zUn{%i+X#29X;#e3#^FSXkS=uYw|WPcg?iLXtjrq*z9F-6Au_w=R4dC!Yi+R&!IRcM
zlP;LZDMKhD6T3)ftrDLV7n`O0$2yn(TsF@o(SZ2M{5CY!tvEUv2C94`-dv*X0^u0A
zZGXpiZfj%mk^<1wvOsTbcl2R~owwk)3xDUSrK4X^RQRcHMlIFIQZr%XewxtB7Lr!L
zDWP%EV+r{{W&Qt}RLG2s_-c!?|1=S5RR&neG}Kvw<2ETu028b8{}lji!cC3QcFGmd
zW_!Ak;Tt+Pj+*7_Q}&a(T5jkd5lt!9;M}x)h4_?|HvH#Ls8&%$ofC;|s=kxS8}i1%
zw_~Z~$U>PnEkp8o!+bdch^lfr8}Luj9nGk;#0Rn3+nb7irK;Cc+m^CaiJpvVsI8nK
ziAyln71ICH!{#(1D)ew_wd|rb=dW@L3(!=d8rNi)rvp*sO@x{e$XH#mEL&4ZPq%Q=
zv5s!%B|?1psru%QU6p$@Ru1_oR>X~C!w&yMLCMPt(i(stdXqYg=!HyCaaA~=m)0#c
zR6-;%KdI?KX3SUq-0SMgu$P6I1dl^fAvh)U<k!5Ad-3aR^l^(wwz$*mihFdpj^&IC
z>vb01?6^^ofhX>ANd#X%s>)=Ydi8Rj&TamCQ{LYq9*yG*7Mk22A)qK)wL81>MOsfs
zTPqtVL&~Dw@>wQ)Bm39k^W<k}aMv5Zb{iU5Hc4)uB-|Jd4;!ej%Negn<#D9$fwkiv
zejvB@Km~GTM2ApSno>ud7bpY#pwq+aE}08%lY%${xh@%e|Hrxc5<@Y(w>^P^1p1eW
zgu0&jp<wQ<Nda4a&?1l5)cid@uhlZK<RFu--5{Mp=x|VWY9F(1a;!lz5Iib6BYgxo
zd5LvQ9McurpJ|vL;$lH2Wn@oF`rKKKg}k+(jehLrXq_}O7Jb8d^?rbcvlS_UBTrE;
zH&7XNxY3>Zq(42TVlt*-FKGV7E;`8uRIk)7^eAwi4=x+t^-x4l%;)9!vMgbSCYUQ(
z-SdcYPN5PiD-ZR&45N3i-SoVCi~r&WWd<FoYHVm2eg^^rsgQc$0UuFi3Is*%2NJF(
zsLS=;^9mSunOch-s=u10?1OCA@0MS9`}9(Ypz7XH$<YY*BenSV*BW<Sw|-nC?dNAF
zxcuGPpiq`od;yq_bIRX8e$iN~h2cyae)czUHwebh89KG->UYwMGq&Zl7J6oGWiB|%
zB4Y$kW&c^MCfV2%kW1*o!Xmb36|2Fg&dHa;8ZYXfJOv8b8D;C-_i-$(e$7bsyo}}(
z<y39I1Ef_AKVtgQ1F;gnyLSKlonBwr8x0QZ_RKX0V4joyV#lS7{9f8@+o&uA=@fy!
ziZ08CB}G8kN6V5gQolZl0U4eJ8d{Zq545R&Gt)LC@$e4wkj?!~zTnD^f%NqBs@~W8
zP>+FM+nKT?Q*wk12PYCDx?<ub4EYhl|9N$|nWEGyrq`#Z{r=?EHWHUD<lnTM$oL_z
zQ<|ORW<G-*A+X*?Sbbffe{Xqv#l^BKJyF!H^{fLk&g0BkaHx)SU`qUZjLd)48eBN;
z4y&7PBxf9Pxw5A{T9#MsA7XLz&B{{>3TDsi_g9)7Cg&>>Rk5ZbpzrC~FGsI*XBhDz
z=Hy~&cAf$mHXl1h?fEd-|EQ`bHqzD0wD_MCg--s6EU-9}2ocpg+AABmBC)u#9pKQ|
zCS{c<Kf!hG8p^8}A^)O&rQS~P?W2L7U>a?yl|+#rTMfg`L%+`On6uR@MO1}FoClFV
z6;MB8&&`AGP8q|I+GuQ3L>w%D4`p4PV3W`FuLNG?X@gzc>I(4*#X)KGM6AOL^%e7^
zqd%Abo_^kq!79iQF-0vR;Xv+sG5kAzc>&0*D4#s3Y2JqrFj1UUBx~N=Z@B*!A+4uF
z(zT#D`T&hhOS;~FEQQsVb#K(r7^cQN6JEp(OE2CKvE_+CV>#%L_Ej8%HsV??0psp0
zZIA@zfUynbwE;~7q@&HSoI1@Zto$pU+J(NVs?2(xa)1oykNvy1OldbcPb>EiGR6qR
zBN9Yy`xg%ElqRDc7kXIH<XMHwI8t74=95!ga`JcLm5|BCnh)|-F+T3T3iE|-@BnDW
z0(CyqMk^uRXL&o=zLtJs2%czUvKyqAhSkEe8XD?|PC{C02^m$Y!b9m_AKq@wRPTl5
zT(Y-rCCt&6-6wM4n^LkBzXVB#D6St%y{zV};dMVOOx52&PCX>>r-Dfc$?wcBJ34b^
z<zmMt`qs*CaW8mZq&SaV-sD7k)P8s<_UO-g*YHMS*tdC>op;^>HFJKlMsg?OjfwL5
z>Yzb{{@m#|$Z)yauM0oKL(JcESoL+GyE{Ymthx}|*_Erb${*{YIug*~&_9wwkob9q
z7i(DwGI}rj^+lpzYJR{+l4b*5J7~b-KVH$(vI*@ge$0r=g0((g9y}#JxZLZiOZN{l
zOHQ>$?w@dzD}mq7t);K9m`>7G`a1dg!e93P3p*JAc1}U{eq9T7O+_fuKZ>SSprM+*
zF_Wi31|g!as2|<G3FFa&!<rX*JpEnoYg3DTA-Q)&P-nW~t2{6qp9pGs!po?%Z^NJR
z5D7j<dO<$~Oxj#>v!c&A0Sw6cbbfi_2AUsrC(vz2vI<u<<U?<uh`Z&Nch%R{`BIj*
z9$>A8cU{ig0Jj>=mn}JC_y^~%y|<^8Pw+Dwyzf^7;u(=BmoavGayyLB?ctqKH_2}z
zkZWjViPPBJ*BzBKn7U@)lN@7$2nH24JtS4<5eM|&P;9`q=CL2jXnJ<Sej-Aq+{Id9
zCK$mmGFN-JeOna!_V#CNOpB4f?dM4mTYhrC5Cp-M^56OYAdk+We>Xg(Fwbi}3+Fo~
zQBw)(c8N3}wvo_}y4>j5uMfOwFE`!@k?6JVH}$5ed#vWHD^RnByev?zeMv!|P^2kI
zA$RsAvlu0S#50cz@^$s$=g3vPZ{(*c_HC#9NBqCJ?GC=ipB6Y_EysoRbaZnV%ROF-
z%MHYymX_0Z|Ir=A?=-id-U!Tuwd}1B92fE!1+AFaiKmv5W6%`xqA{EdwuS1<2jxQ(
zmteM$Ev+m2^Dzs&pOTWWz1_x8(&oy__V4hfelwia{WJfi=aR><<h$P(4ZipgLqT&?
z0t}2n$bOSMwxih_S?tdjmc8K4X5m*F7z(S9sf*Cv-KEXHf1MTRNZM*yCPQ520nEr`
zhd9|`X2KssHo2_Cr0i}pZl@WwCdyQmM3}7q@!N(MK(f8P_Z=Ik$(y9@{*HyS?RTzD
zriE%)a}NKICnrqBBD&kZsv4h-hlNS}-qk5cb+bfVnE*bOVv|fxHhq#6_`_#na%K+1
z_jv<hYie?@I^tU+mEe9-BsDXzI_gBNSBU}iYvbDHrwbk+qnAqWtW|NmrLeEJ%4qI4
zv*WazcOzRYhSgGJ5sc&bERT4=``Ja)6{BdK0gIC?64&Rpl0zQwf*;?1$f|El&Vm(|
z&rE-@N{DVsnyePjeXI?;R#e}VNB&mObJl|E)_~zhzRf>=dx?T+n!j7e8E5Xwaut;W
z#CXei$mxG6L)zoEs0iA!!}74Y*Z1<fNeJXMwwWmkfJVfauy5tsIR7v(C)#EVy;gt9
z2x6162$KVkG%Mdaj?vv~w@f1?dwmysodx^WL$iVE9Uar*wkC!`LaNI*_wzQ@=hP}b
z^*mz{o7nU2H;s~0|HY+qI{S4hl)LhLA_ET4Y6rck;)!Rz^Akd+7I%2%Kp<6c<yA;V
z&as=c=5w3ty2W`YLv^QWP$Kp=*M%B~)2$@jWIh+@Gg7q&a<AiLIehhDu1Rb@#BjD#
z;;}cU0j_x<f|`rgv9guJGX|0#z>UPQ(!J_^CSOtkB2CwBK$GS0bQoNDgP;C=wfmk2
zlIn(%jMM?B7YCuR3~28UfOO?%;CBP(!DB3AW8fcDQ4Py*bN+7(kD@^}F|Ra9pX@aY
z(w6*EjNb8(RixyZ*SDi4=_?oOmKln1N{<tTIPg!Fi70A0j1;;T8n(YhUr=^$xpe3F
zjHC}Z-YWrQZtN3_E`tdr0F=q~rKRa0J?kk}q`!uocmNQ-6k02LEziOq9xg36GJ%~X
zo0llMagC)ih{QK4(6TQEkOXd{K4-g{BFeJ^W{u=To>X@W3;XTuU38i!ob>$!HF1U<
z0{`rdB9fM51tT~k%I9)v)j635>8=tIoB>A$&~gXK*itW`;U=ns<Kz1Do3*?=+8Eh0
z8#=T5r-2%G&Hj9ZVuZJa<b<{6j%1G{=kSBqKGi-2+9|~&MGrYRNzbcdkH>MkrbKE<
zN=cOudw$H2407}VRBj_Xq06NWdo9U}H9#^p5Xz5$+fa!POd^Z@NOm64kZZ5*DUW>5
zpkFiSIF8f>jp-&7tNP$L&XuQd4Q=DSQeq6s9lgyk0$~PRPwnvBp$=JQ-%g4PyBWxY
z92_p+F3<dqo^>aA(i-a5_S+x6Z}zDG-1FjY{I22BiIaoj7hDG{a#FLuuP?=4MdR*Q
z|7aJ<)ka{1v`Z%b+us>2c~hnSX1v;n6R0Hlm^dRd*6ZSj)6gwS!*ti*K-&(>mwdi@
z)x&aIvUsFM!^4k5?BsD#5y7b}AP3N!`48xcows)?Z2N!|+%hkI*&Sd_=O$OZZSY-g
z$U9}d_?=Cp)=}Kge>XwJ&t%(2Toddw>O^Nn-K5|In$=zEi0A9GY^no3O2on^d{}cT
z%FDTl9GVcUFh~;|J!mGkdO|3;s@luzuEG1%Rg2YU9FSTsw($zPljwVAHDBAC%9j9&
z_o;?NJR#BR?R3w>WZIu@H#g>ma)^eY2Vx8MPAu-&u$>`cSbmAnZx_Xk-n=?rx;uv_
zqlHz^%mb*F$-_Nn*RZM}uRp$_ad|5||94;7U|L#Q#5ai-3xiMb#^NGc|7tMIy7B{)
z{o#F?kG1<Z#uX9s6Qz6xCo#`BMNW0Yhk?FS2<c5);)m{$m<T1|n-IN}M5Tf;uZX5`
zSH9^A>2RN4fTXFK{koZH1JOO>+`HS{FjU@!>iG#btaKOx*!6Wz9g&CIO$>D&LwUbA
z?HgTR_}{`gz66+8bWPcmV@Q90ZLZG*eymE?(aBHrpM^_HOOf`L(hveb5@a&^AfTz0
z)YjJKxNXx?aVzWTwy^6pqNf7xSnQYdl-25}v+<(8(uZeuH#eudHve`pnJhYmV$lvr
z);!SLQ^qhjGLiHba7Wf1(T`fGfpgkt?`u|WU-p3_PNRi6o2xn<hLXK3+6YAAP`0PU
zFl)}eh#9}>r`4Mui-O7qWHTi(r6g?mo0ht@EuAq3D+blJt#HkQHSc2#d(#!)Nj*+k
zeF-V~`^!(KDAo9IdpkhKLgIdGBK#k&e-<3gYU#Q;dm*7KL*bE?RXhfu3$&+Esszom
zcVwN1Z#^7QSaqH^PWYR;0W$E=2o}T1idk1tllk7EXTKO;^E7@@qCg#dXHPocg;qht
z#?kOuHjLqQPPawFpGT>2DePog?ebhe6B|(Q-h_SX3e7fzBzFDFQm9@_GYaaQfQ#N&
zZd6P(aDjiaI;&x9|HVK+lH)ypcxsI7O~Lq<w0#%0_ZPam<KXbG3hBZ`lfGe8StZpn
z(O-fm?}cDsLsS<`;@0(3Yb0M#)sy3^Fy1FNCyXB;_>JoG6SIY3&9QG*hZ^dE{^=t9
z0}0tc8D!8u##g2oxvv?<iC<j%`(~bZEx6ZEmXQfv>HSlFF#JcmdrmvM7k%wH`T*5W
zD*Y=0OU^dMdAHnbqOus*V{;Km^8JgKu0zK;%RrJX)KLeVy?1#p$;4zaW;^P_l;q6P
z<iTSMI!4|4$-KOWjoEbeo@y^-r>*VJ;hcJl9p;GX)E*tqswvr&k<B1GLplbOwoVHI
zjF~Sn=}FIX8i8$R)nfCauTiRzct(Qz@NPrfC)Sl$?2q`d`kxAqk$U?2LPL}K2LUZq
z0|`?U@t=VC;%YAvh>WU$yIG<ZQcS8%<9FVsehUHm1dGE`K8(hU<V-s(+IU^!eaB^q
zD9sec5L<;FKOJjtYsGx;3e!sQJZsjE;2#oyvpM=@cbeZ4Q(jntp6y1Av;krz9q{xm
zEDK1&(>Ww#(uGBM6j;ftsHyD4rSuAnYAGPUlYrTqG&A!}BGMrdQQ?vj=GQqm={QdX
z-p}{<SuEn@q?j=1Zb1g8r3v(z#*bP*-pfQOxVBAySJ#GW#WRJk_{sT|`YdFzK@9ur
z_Y`a?@Oi+Q2)T<?zXJ7FD2va=cKgGPOkw~(yiju_S9oT|3bl2>x2WT=p6T$K?Zb?k
zB#o$-Zf>sg{u9~cAbYI}K%V#PP_qtdkEyLWHb-=4g&g}L8Tb9*W>5V1Nv1=aZ}K8)
zeJLnBJQ`RM!)ld$$V${Xt&*G`(Y~%?1v77Y^O5D{HcqTH6=%w^Q8BCu4g2O~!8Pvc
z%HK_1eM6vK7qV_Ui{8@}`y2H>;AnS(#Llkqnt~Z6ON<vJ9cwBo8xo&2Z+jd-AhLGK
zM!)wvXdgM!*}OT(dcwZW;07T$#4upTCOAa8FS0=6?#|VNf|pB+CQ~nup9_`gh7G&q
zzQ9yA{;<fq>s`AzMGiTI6@J124^@4FH}fX#?^2q9j8$ryV1)}@_gDIi+Pnej-jEmD
zyT-xcjt8!Dbd;lI;dqR8x@}uEb)$YwP*8d>;>k+50HVx9*ZT@?VB|$lsM-#}VIbyQ
zLOO%sThfo{{rUCk#kcvn&p4|B)yKL0P*d(rinQy5<oCFcG4;}0Wz;eU%0NaeSmp*M
z5L9YezOnQ6<OKH#D72)ydUb+?U%q_j27lh9cXXnZ7`w<L)Jc15_O6Cpd`M(QQ(W>T
zf8#41>Ia*1L<dIo_U>Hpc_sKL8+WVJ$+6?u6~PW-knE3=PusfOZzTl=%Dy$=G`?QU
zp~&xRP5PM8R+{ZLmRncsB?GecbW*!q9GHVe9pqeD4rSf8G)P|VeGwpnit2&6rC`W+
zy{T>iQw6SqBO9Zgpf`6~a=g904t+YV2bp|1rWK=IZ6{OwesL)Gmi_41_7*UNR}!!m
zz@=r*+FgS=Hr55`<K=ed<LvlTwo_hv033)(X>5M_9s%Q(1KuZYCmXlMWHS)om$C(o
zqyq8-4*yI8&s}{>B(x9B=Yfss`y|7??caM_D|^fIxjKE)iHv=f!8N6!)OuIrE8P&G
zXDv(FS_bOz)387Z%e0CG=>`v9@2&8t7J|KPY;6MX&+FLvuI$5U=|e2EYbD@Y10ffR
zDa#B<(JXU;%g;xK)Q=NY+CBtf>>K8`e*d1EQ!S}9&A9{M^f1uArOiKYn0QA=@$0Xu
z2k!q4u&20t4voBa(}hu$Zxl<%N<LqWVw?7_9UJoT;9F&f?EqhWVIgq|9pAZXXns)4
zi&T9mk?0>95ur#^-)$Rt3i&L%$H0F`$+r0j>Wq)i4+Ps9auxf3q<O7meVj<;+R)Bl
z^*PlQQC>>QtA4GZ&Qty=|C+U-$Lyhm+OYh<04$=u%kd-rV**NLa@TpS_r?CDjx|>M
z8kIJ)4mb3>Z&R&ZDw?zUd3I-zE45(uICPfphgABxx%E)<S-b59-<S7>{uvC36SqlA
z<yhp+*N$N=_}1CfjOdmH>G7k=;a%x7>DoJ?>>UE>`~bWEwf@Jq8A*FL)27(F7vwi0
zn+6v2KW{V-mW^iq4iks8gyPr&$Ba^-c3y#r@Fb2M8)FxVK!QA@`Md%D0ID=aH1{38
z^!l*;a%-HXEO&3cM#yv}xMB37?`TGPCYFp3(az({mKBmkh#<OjjeBq-*9ZFB5*-2)
z{c=g0Rkza&@1$nPBDJD;r-&*STzvgC$8}|8zOOF!o-1kx#)Bqu+lzWlZ%`l+cg;WK
z_B@Pa>%YOeqM=C~)E%H}LETKNo`P4qbtKcAx3_~2tb8?|fStR^3g2b;FufZ~zz8CZ
ze?_c=nn>*Z6znv%A8vjsZY18<mG0_%n^Xd;9EoCQg%xVcUv-e>aW#ZM7U#PWbNu88
zY8m73VKPEECQ1BtXDs@Z2u8-m;dm}jeBDO#couzG9j&J;arA3uqv&Y!nNa@*tG?EQ
zYfw{=mr7m5c}1}tYn7QQb&hASPz>57S#5T6y(`fLj$2t#AE{4ogv+G!yhf0K;~;G~
z-Ns9SoIZ3-OsNq_RLpfmN9XyL5op_9&+Oxxx(-^d9Y=xD@36tc1qz`ip4dXmH0i=Z
zDQPLG?`l!I+s(Dy|JF3wF1{1wGDaJxWmr`#6&<cy(N+Npz-ooDusc*UZ6=hA5$P*b
z*z);tJFo`1e6ho^y^ZZyLmJ8dDv2NM!U_OS3#7wsBD-OE3tycLHp>70x|Y@KqcDa*
zw{0#~wzt%_c6ket(JDgE?C)%fze&}UsF<Gi*D_Dc7E!sde2YF&s&$5wOI1cRV|i|G
zX?P*ZH>ShBd!)c=s)YJ;pS~3kq&hs~pNyEz|Ja`AWfByf%kOuSbmN9C5tal`r}n)C
z96%f|^G1=rb-u=rMYzjN_)QArl0d`LASkk4$=(*zX@IoBv-<Hx@ITrY1VQ5t#yf34
zBQURk$MZkaiJdNv3>(vvGl`x3r5;D@>nGQ$)?yOK<YbV<JG4{+7txo_KR>Wt9mlga
z+fXRurxCWJTu-{*M2)yD{`;YRp_G*@qNfw}8{6HL4mDD-#RaFW8Z7?kN9D9MJ=Tsv
zy=hTe{u>_F_G!=nW>}&(cIi$DKA3Yrv(3Q?XE6Zi9ici4{(Ey9^k$~VA9~%S>8rLe
zCWYR*YN0b<oItm`{wSlfc;!Kct+}Y-+2U9`?z=qQH%aWv79SN#Krn2!Hj-<#YUb`_
z8pB6{%h|(_zdQD;abD2($ekhB1CWn`As40_@O0#<J>_w&Jdqw#A_FhsFu~dSi}1*H
zTiml2O}#r+>l{~3hQmqpNU5ntyG_kIhAJDPs*}3n37;L*ODt}Q4Omn+fej^8(45Z6
z(~;qSfElC@nBr?$hrfPB+IwVp)Ow7G@B;-*Uq-yBK8k4ojO6q6y<Z4pd08SeBBrEC
zyAO6Q9yZWMvukEZaUy%9Hn`h;+?5}I1B{K;wnT-So}WS|e`T?5HAJ=dQ-5+j=Dm{K
zg>`-(F}%FjP`lh-OPI8F>ByRE_;=W^=wc}8wpQw{TJMJEd(^=n>`zwn+N;+;%KOzy
z(X;8cjk_h}gWpRV1ge?mq!OKQ-rn}rPVr<&w`7}}MW^KFmJF54cEujRBXF*ip2R0B
zwbZ=Ct%kaR8pNB^o4YvW{k3V)g<)XX*`oHFm8c)oB%65$Bm<<NiEbemgNOAZDQpr5
zS<-Y-u|!O0JU>0rCFoWC4O>ClC<C${sde*JGtuh?PAFp^2wpA|b%V~}@0=p?nbyr(
z;5w<&2Y|i|@Fma%g8pQ<2^+6iV%U#No|ka7yx-(wFSMQs9o`Jm5IHzH2mn=yy)}Zo
zdzJ}~85*1Rdv64d53Hd*g(ad$P>)~RS=|H6ZLw$TBeunYLovC}l5)2HTxT-kbCO)i
zDYr{(PJ?oD)T~49Tpu3VCO9!xPRw-Yudem82XM5j-n6SCh7yU>ky?G3*!aOB{TjaU
z4flQSgrhnuW}Q}xa&j!zPh4==xNjJE(g3a$_Vd(HdHE=>d4rhLwJMTJaVqZKHx|X$
zw^$tV^S-Q*2u~dR+~wrh1PSal%lqOSW!VYk1ZDThMtQz+zRzAa=fm133ib5nFnJTF
zFUhY4pEsy1Qx^(J{$gr-*58$1h%Dy$ZHGy(zeh`|xkMeh^Pv0$7U@h#@VRUt6BC=0
zM{Yi5Fd12O)f-vTFA+d`sTX9!9b)@>U#C2fPiTBVH$H}3f~3*xF8APs&XxJ;S-!vP
zoLU%2&gitj58JigH1IGW`e|^!Y*7)Y-n}ITN;Z5B0*&j^Lp)V@=V7}$d{URX^Pgnr
zcz(*s?jPVtf;Dehj$Vde8@PPkTns%g8kqL#nLMCDBIXy;q?sg}OAnGp)k;HkihyF8
zndS?1lkJtJ&f)<IjcvS5+ybE4)X!dPgrsu`nTDmX@M{lpWrK&VD2u#hmtFM+b8*Ut
z$ZXA^xs1J&;dn)NCts$8xBCcYqfc7OO3?QG2lR!A765wXv_n|OmsB(~5GTwZfjY&6
za{S*TR8o!=4I(+$pEY>i>}9PqtZ|Q*8v)1j0~sE#hNs&b%0*5}W|-_6inS*AH0jge
zVApQW@BcUmxb6Do0<p`fBm3d4jjVLn&`_+du5RS+@3KrV2q8ZYXw5_a1G)T){ya$`
z?)<~u+?3Uv)tpNc6&8u_?weVkn=8xuF|l?>FQsce$S0!Z15Hg#NbICFuW@tBE_*-V
zw|4CTln}gpwrx{lNEGls{GIM2EnwyUoA)UYi8D6;fJ?&qrJxviyx2QG9lfJJGy0#`
zM0LI^S%75X;kNoB+xrOu<81r$)Y1ID%n}KLmH-rF#@1Nz8PR?rDDg$;SieO^M$561
z<NfiPo9lemHDY_R;^{g0AI7)qsjOS9az?+{R0KJ{s(QrER7fa3Rr)=t3k_s_tJ@N@
z{_6Y8?kBcrf0gKzSk0NZf=S&CE!kKPf>-=q5pz;%B-)1DcLbOfKMKZqpp`uKLZ5Tq
zR$#WQ#SvVqDB3d`%xJRh%B*odP!Yez3=8vLGBdjp_@#MeV{7|QkR?Vf{cs)|4}YHB
zlDC<#v|2%p;O4z@g-EiQqW*B7sLtl(2&FzECw<W3I$5a}qAsrinOr%j_JU-;(GDFL
z9|^x8rdoN^TVOA1^zGXml;2T{j|cpD#7SG4-gDlaPW0@r=_LspG^JYISS|opC8kiB
zixk<@b(-iDD318p?Ly#*F&#l@LUpNRc;(Srf5%GP5kX8>m84E65jI3?1PY>fkXPa}
z>CFvMH-rqPVA*l9u<1~^Y1Ps8(i)(~^!JngELCiDbtxB-+@YvMX2g@;|9}7YDbO#H
z-j1?o-Yn~Z^=}5Z0OAYSzkj{qMOS*DxUDzsXYIv#L_8>!Mk6|0Cgi8qc8a>DjQwa|
z+b6YGUi2tc5WNP0|HVpWb7HTGVf<z@RqLzp5MTx{(u<6WFq-H>m*zKd`K+TYx9oh}
z*^D6veHsHn)JPW5YOY*G%}%FV&bWb`IAuFGW_BZUT~mk@Voe=|MPU2dS0#Y&-3?|w
z|M!S*?V0r2c~gvRcPAWNC@uXvNCiqtPy6U!^W!tK;NEAZeH7<c!OFw5&;GemQlB&z
zfV%c0(65Ds1^K6(6{2Ysz)1}4O1D9NUlY+c=9lO$9hqY)>vyz8KgEBlaO01J_br8v
zGe+5#eKRh(u>TsWkqX?MpTCIWRdHuMO8ia^cRUB}p@s(~RS)!fUCLcdDmF-PRTQ~K
zqbe3vVPttV_a{w9N5>fIe9hF_+V&sv0x$zc?@5y?lG~lOR{3HI1i$E;QN=pRF9+>m
zmuQIHt}cX%$Tijt)ey8X4|?${?+-}<F3S`l{=FD~CJz5WGbGj4>IXa_)-|ZxeG)+2
zQ|zVoRE}%8%%p~^6R0aJ62q%U%GZq#got0&3r2re@>!ZKIPyKUI1hDphC;oEjpAm~
z3sd_%b@;>XxsDMxLvI*!lxS*Y*J#J&YG4AsCSS0>xq~d}&JWb#-^D^9KzWkjZu|BE
zb)>Oo#8flDTdq2m47wFmL}{d)<h*K>00-NO^YCB$*x*y6y`4!S6kk^uX;5Fc0Vc${
zI%Xri^@hL9eD@`L?=}k#SIpfK+No@Au96d7l5<0S-e0ZMUbRaF?%++y)&r(g%$3_b
zF?v}|d6$}#`csm~==jecW{uTmL-ig3yRKh!cKX!ZwigiHpx5TI?`;^S*7_(!>DT)7
zbHxult0t2ElO<)>JXDbpIs=tF3wFpml8!#OIBJ2)%g(h|{l^1Y-g$Sby!NAc_TwNP
zMOX7@RoAB4l}}pxS)d#icGo%%n<-;-4!>+)vl^}W3sjJnDuKl??$^2EG3yVsn1h^s
zx`FKYP2<a@BjsN6N=VS`*ewft0<Bs+5R#F8YES-wHA(b3{=7S@g_lnfQ|^M57}yP9
z&l5>)w}2alY~-Im{tkC=TQM}RZ#^#-`nE!)Z@+u5Gph#G4bTL-w9x`dmpEklK68gT
zwYMFvOh)f6cSPa0?Ye84K;s&E;~8g0c76k|WBr-7Iz5qv2^;Hx$pKms*!7EW^HCgF
z+}i0WNi%&1d8qsB<f>R(VyCp8p3pmrauJibGNq^<6&e{D8fqM$>1ukagBbw@EzZ+U
zd|6ppS?>5tIr^!h>XjS|+9udD#e@L(1Atng(xM!~A3lOL5GEVT2PSYVp#AYMN-5rQ
z?KUU`@vakY)URpLH@zNaYFlEAcua2E9yn808_}e6yzHle-CS@Bfu)p8`a7c!Kpu{)
zNQWCozans(WQ-MMTOWLxo8T0Phfxd+z$ZzH^jz07hBq!Hua%z(9jpoWosnQ$OPIb2
z^UdkX+(27O>lsu%<=i;!ki`wl(2D&4p*C&%IF3X76j!!XNQb`zv}R|rKP#eBHId=w
zo)UeT^_+}yZc+AX)vajW^RZ3InQgZwCaRxe=hyw<iHL>I)%PMC8ifs!ka9I;x}$ec
zosAdE)S;pNUYXiT=Dh%j<?2D&&)x!z8ON#)&Na|)w8ix4=l*aT*>S{)iR+TjYR(E}
zaHPM{xVV}MY8$Ji*=W4S*!uCj$D|lR-YP2>NtNsC>gcYX@k^BmW7(-Hmr>~+20Qml
zj~-GiE`i}B(i8A>qkC_borDmC|H1>&Bj2%!Qu|cd4?~$(aRk2|4Uo*S-t>%{xU&(~
z2S+6H5L~b^&wVbA*0=0|fgo6hW2NpDMzsi0>!+U)h&R-ywp?BzU@hcK$t8V@94}rc
zw2Q6bJFM}&(P$%;&3}~}>GiV=(@(-*`$vnP+L&^aY+WGl`d|C7nhC?xFgL>z#F>yE
z!3s!uS8i`A!^WF1&zS7<DTfYJ6KB*x57W^q|HV5BP@C}Hv`j|kF;3o-0-l-Zx$kug
zqn9Ye;~#o&KJOMSN&lC&O7`DU;*36DwPz$|lZ;7#8^?Ddcnn35$P#tRg7+caiL|@n
zL;|>j!%S8=?pbxn-c8@;2L`bp*x6Ng%KBsyJwuQDcKk@KEgjn91$-Yj?0!@KUHgL!
z{%O%S^)uv8V;x@s#4U5!U+GMTj=n%w`bKcmt&z;^K7ZgYO}Y>MX$VIK#Fx(nM?{71
z%~f<9vOJ*`UFRgO(;m`>VQ!FWA}i8?N!pmxd87(&`O?fFoVTE=z8}~=3kHFLm8FH6
zbDEHFp{tN#EYwyjKVdgIH`Pc2{D$`U14rL~Pn&BSH|kURWM#Z+n8X45NB)O^pK4k9
zXsM%(bhIuVZ(z{r-@bfyQ}kW>V>Hvw8x{5Gl#yJC%=grf=i>U6z($XMetmKFHfb-q
z+px2ZAL*yYeOW!^gsBnIaG9Xs0R@{BRJa>Zo9;W*F39{-eyR8Coo62N9?%3bG7ci?
z2iPAxj!E}<9{%0+CPX`^#K^+6R*LwyJ^OvtxI=<=`25a`H}zwBjl>t0)~79LIi92Q
zG1pF4$=a*|!R^xuG2nYmZMS?nNReTyW8b^Zk<na;O`w}6Of*ZcRxaqL*J0`_HCBbX
zg;~{d!$rFn_89L~9wSA?8-xL&TR5gVDXRIFZ&~$IMnIK))xg{E8~o{TE>Ql++mvPp
zfpilvTqu<3v-J@2CFGQiAb2HY{?Xc<RR*5ucMS3?t6<vFn5z<>^ly9)>6@g44nH6|
zC;z+gooXB>m)V<H?k{2UX)EAn`sYy)3UdER4x_e-OHL^0tcNI==qD?qc9!?Yp}H+?
zFU^*1VAa^!$){)O#*b1yUH?QCeqa2V8LaLl)6ckH)bLtJ;@C#db#+P!72#5JQ=kiN
z@!b0u62<dMG8Mz$r8Wp6y>axl812U^O2j?4R}E1OlH;<K<HKa$x+NgV9(^0r9DgcZ
z^O5Yc1)>8bFHbcv9b1Yv_B`{_gb>;^yM*mVSpQ0YoIYO5L_G=(4~f<?GP+x7FMHy4
z+-i><4j0qzWCVjEr#?ZHMBP+Mm8~cp0Mm$46b<OJbM=`ozKW8rzYWPh-gE9!`Y|B~
zGT2(Cuy+$c5_J4sN&^Ic7l|7*%5wNOFQc*j>WLS_7lW;T`xa0fSrI8JR-7eHN(%t;
z@aT|qxI<Z>bR2Fn^UXe$sk^U&zZ_f_1c5FHXN<dO7j@{64tmC_aSrz(d0wA(=jL|z
z*E&=Q%F1=siX;l&{-M9|%q;?)mRXH&(l_ym5#pG%{QD{1I3iN(ICq8%xnkcf_gyH&
zpRaHKKDt9szeA5C&^-x6+Fy^_+1GgdAj2-l<{L@~0@xm)i#&vclsd9Lt^0jx^Anp*
zZ3%9)Y3Uhc|NEvdip+&O`4~|aA{b(HVIkeAke)R_k|_4h*h%@^{w(LBDv|e&-|TA6
zKceL_mc^?k#U2xQR1d#5YGK1HxdI`X_~rfj0GgY;T5c{H*AFCvh5?`r>H%DOg5A68
zoXT`>x#d&+YrgWYQv?>O%OM2fe*<@X1t5lGyAZChH`*ywNk;3oa&4TD5IEMwR%0<#
z;yr1=SrH&TT=wAnJKWc2R0@Z`waLWv`l#64ua+_@m%KRi0X{Wf&FD{mV4Q3J<w_Y?
zMN6g!_BX)sQyqO{<M-S$ao~rs$(#G`F$SOqprmY23_rEvy}If@a#IEcJT)z3T7%<=
z$Ya<~K_>JGA7WG2!T$UijnU|HGa8F0z%nEwXK~xyK5HC1Ym6@@5pG$r6m33(eqApR
z%W2CWfR(6avNnDR+g93h?qoiG^Q!v%sb`#~Df{=$q0Wnvjh`DKew)|?&1hEcQ3Jzb
z_(TGhUzV>A#(qW<zlk+ojE__W+o%(3Y73?+Jy^}vHvx1LO{0VC<voY*yi`fEA$1aU
zB-d6`o%BfZ^ZpVy<8qKwE`>vYmu<P|Mo8dRz1Z?mig?mXlRNZl#pk9QDurKiE%xFX
zSNZ#KjYA+bQsjxOcDkn8#;prC9Pp3au>HeJe}EtGGaO9R-*J%jJ79o*LM}*%6U058
z>B|2sLMC{dyV>)!&AtAF*4qc0q{xHDj;P(upfHEu)Epw@An$tNDOZ?>GZ&Kl&el}0
zZ)WCz9kMnN_2<_m8~kWTA#;J)cCW0mgWB6RJ+}WKqVGhjncQ2&GV~x2auEUlJp>|b
zhb?RIyUn+<<`@sSjo)1`nFUQbd`_XMzp9_#nEU<ug0eF}`((41Dyan|<%O+=c$v1D
zd>fS~GfO^j!Lg~|%@aONo}6Glu(@R%93%oR$@^wHLf=gd3y{z31hl@s-9J7+p+m*K
zCJ(~~ez3kO_a1^S4YJCdNA>(`dXovX>`Lpk@2}73v)THlWRs(FVP}wS5j)Rw8M(*A
zn>4pjPCm5YQ(&;r6e$(OV-PQcee~>v!N{%VV868&J7KTGt+=g(W4lcZT%AwI^p{~l
zn6%;X<0~$3^Vl5wnrdoZP)lo_nNxUP5U^(_46{9R=l$W|)8)7V-)I*At)GR3jz*#C
z1V+T&mKTRq=EBjlY2L4mKOZh{1iZQumt7)JvFVrZRWl|kL>fps25G8l^eU@r!KvDI
z(mpN=57TE3Tp*sXOTXlV<76^4VV*@{7e~iNCge4s&O8NUzR&@+ZaLvvCeHg9mu<57
zGj;tU%5weFAEI&b2FTb`Wd8qlr9A2e%sf3H22ot@K4&zLYtMao(IEA}prD_(ihRzw
z&BNZI<vcLMB>1}q>W?;Gr3*tx3Fi<!CLmP2sF8PBW-9?^oS<gDp=0F<PdImr7JKu&
zdiS|^Mtvfq_EdovAQce0gk>o`KKb3lhzy2jz+xb|8Zv2^l#O2&!gSvVLvt}U=acx`
zQ0jHtGsoVQzN2oN3bac1j=a=Aao_`niOA$O*eH~dW-YuGG2~NuC=p>*p&K{@(PI|4
z8}6l^63|2=r3Vmi@39y6SgWSjPFEP2C&6r!+fG@Xa!8V0vjxQ${;~7&jsnkr2Fz)G
zHD%&W=YvU(FMv8v6WJ+0dnP&SMqB%M7NxBAjETX|D)0$6i6L|;Jo}V287%T#gNZ47
zw23Ee?lwn<YaRU#ioW@qdY3lgZ6we4?!q~K1gs?F`nSyQY?s47hyOG9PESYYSCsu&
zR@8T323Fatt1k+)Oy<^*d|!iqhMo%XNQrfu0#1X(#Kbgdpg1qBY!<*m*LoKr+L|QS
zSCo;)D&9@wS<C-h00c@W7K$S04=!wx8nTyM3LoU;RQRjyxa2Q%v4@$=e0z<baI7g)
zlD8WqGe!Q|tj#n2qUye?tg0X|Phc!O^(+%x!t#5GO)nM#+a1$pM9*X|sXgM9uK(5`
zHh{wdzx~y|w4=>F{l64Xy?0QrqknLlX=(bmC>zW48_Wcg3uL%Z=k;Ts1<Xm*z9%9w
zA_{TnN1UEUzr5Pg*r*%(&!^c$6vE=kHPG!d2-k~=ioO;&LwL)FJk9b&9-mU6LggmF
zO&dGq4<d^iI)C@fE%(flIq1Qgc~WB{Ao;-`%iE`wIfCy-hs<|GIU0b}6201+3z+?u
z6%fR4ot_q*;Aj<fUmk%FZIU^H7+?KaaM7E_w3qrsrxSG_x#Y^Bdu0qEoAMfq_VI0x
ze(V%XJTns#h-jWp=PCd7x^<qVsPZN0`dd-&jj<OBXjyR~vY*_v7v4*~?{M!!6Ef&^
zQ)Ls4fj<)fn&i0IUp!XyVWNNO%cb%C6(r`00ZiybHLZH3oXZ1QN`lYm3>d;6Jh&?6
z+nBC#%0|4;U!vkO>HalxHU2|ShF0PEc!T}~YPHJB2Qf+Q)Sn&dG;kvMLQMN4+-=(5
zv|{G3h1uchuD9%GxPhE><}L54&4eHbA{&MNH^k>ujG--+D)a@L+Wj3=U|2_0_u%3U
zu{G+H!IZt6t3TWeBhh<6pINyN87GXe63|9so6_;P1p?jM4kr)hr7HAW$2D1PKvtLQ
zLBUH^89_|$$1j2J4)Do~@ib}K*3s_&j2F&D>0G}U8IEVB$jMK9+zkh|^2cp>aglmZ
zjHemukx_1f=zdcQ>X^_Qk&kFmU;5nM@B*c%qqSAZzLq4J`XJ_O(qA{e{`XvlAZM-<
z@vHOeT?gx3T^NDyK25ZWjlE^}w%ha{%^SLq*a>_)X^r0Tr9`4LK{W1~H;o}~h%HP#
z7~H~*MRc!rtcXRW{Bx<)6UVA;1W3Y7nFOtV%>}!UTLBz9VY#V&Zpv@dZEb4?+9SiX
zI&M+iQKYh2ycxk#2LN2=9>>Nk>qRiUJjxk}KSs$LI)jPFIM0pQ1rS&ZMMuG5dtUUt
zLDJtR{4;j+gTx)t;jkvQPng6!P#CZ;C>{Ox*J0nzTC{<wZqz|u);j6Xu)A*`faPx5
z66_iY<b1FyLIjiXGU%<gh;4;6elg=pEVlv4+=X8D>h#;U6S-TTpH^HL4uTwu9k|wV
zo|{2St83vglk4B$;5YSYEE~b++WqrN3BNKY{e*mN)&+h)-fnrNLB<d@*H8;=zY{08
zPTTa`k7}q69);6i8U=2%PkaHL{#@RCF%Kz_amww3zm!|;)31#$d{R@8#&M}rqis#6
zNR?r=;cWotin6cC9$ZSdfta(8KQaLMGz>OZo=y+|jf=w}e44z;iEg<GRhbP+%PZlW
zJFRvgB>~Np&d~w>s><h=7H?&Fgw-UB<TA&9+Zh@_yD!8{=!e;IZcZun10?V0m>5H9
zZZ#sNdiy`FVpY}NDIfB`c-aY*?n^TgvS2SoH$6TSo8l3kUE!Jg9gW3%{@{bmdbrc;
z=qn8qxeCaaFT0%uVnXlq1pAY<vzt%HA?iiOSXZo4W<;$jcG)v*4Qp6C;kmTi!h0`z
z9?ob7ssKZBpOp`m2Z8xy$Pi;%gZoW#PWDLavbk{_Ir}1EEZ`U-2w|p}OQHwczgj!)
z04q~J+8tg1Du6X9HfHLhofjlJIa+CqfDh!L7RB;rg8iWMT^htRGdtnyf6X9gukLsd
zL2#|tUkl;Pp3I*8;iOlTm+(Bp+?-|MRA(nTDZk~c?aO%mjFsuir(&Gb*S$D)eKerT
zX%HB9!giOK70SvK^^=LDRa_5Tt{2U96;qtoA{+++V)JkG`PbGK{8Qd%S}501lqr9k
z8}5~U=1k}>aAw}wk>=XICOE0?6r<IR;W{zB28=t0tDo4ujqVOi{0_C@Ay7dnx8q~=
zs(#$jY0ycDS!dsFr|7Z07j7v$?<w!0s@#;@!gUpK$!hfTc#eB(#5T{d3=q++QC6$|
zNnUQQLM7sNdP0|fcA96l@w4bChr7n9F^qYCaVbP%jjUQDHq{<Qq}Nf-_--BShmigr
z6s1;#B<>cTGYAp<qwn>cw;R^GKfIr~FQ|X8A2?9+Xen`xa=`|S<mjMS*?{pJA(Uzx
z@Gq{*p4*C)LR+vb2b1#7790jype$Pz6_KJSM%Td?!AD;y5!b=zIk%RWl&b4sypCQu
z_Em~Q{idMT*x!cvobT-Q>4TXaz-_j_5>D6T#H#-H60Q1}<64N4r`Y#Nu=)^M0;}a`
z>pef8JVDXNfzNZaQ@n9t1yYEglMPhTcj;RH4H_;g;B0`aw6HW@bW7A&G8S$U;EB0c
zKP8}{&|>LI-`Wa~&{M7p+3<&41%|LVG|y`L&50J73%8?41|h60f_=cikKVA2A#?)g
zklpJE;4<vplZyy&4v?whyG$jNiNi$dVPG}THMyZ5)!Z>)nPVcPHO0E<p+`d5Jzv@`
z1zs-iykK5FwgyAScN#`=+CGQhKea|ipuNZNh(yD?7j^U|`rz5S;Wq;SPrcMaEdJP{
z!3ED3d~_Qhdm!%_cOS+nvvpeu3jG>h0@=EzCC?WN8xWPm{zjMjo6cbQyM9@m-#sm0
zJ5ioKk7_dh$GVF|z0MV=3O=3!x%FZx`cG*+fmRj&d7n@BHs`a3@^ZX4l_r(e2a8)h
zVtIM80M;uO{V=yG!kkkWF;`EBkLP!|?FmgCna-rIjJ?h|^*v7}cDvOM?0gK?di9b2
zS=Yq$etgKc=F=+v{8NjjqF2uJ)6z8^Z*&kG^Qw@?9#)br28qOVWu^|p;EA2K{HT}0
zFV5ehtn4!MbRD7(|Asp<%H_@DW#VHsxy6xI&n4y&b=Qgoj7&2U3=+EqLpIj20pdyv
zE032oXFNt?b{1Y*Ifi`G`7`lilG7&~_bF~T6TqrO?<DH~ofZV{sQRySl_`NA%IF=2
z5F=Amf(B_2g%G}W(%~+7@fzuHvyr|fHVz$5`EZHYq{I+^Xugrz`-P)HQJF$-sI{Ls
z1mN*(A0E9Xj4(&NCP+&dh7Ia^q#YtA{sr5&3pW~4dyGghSNCPT+kn+Io~|C4fm{&i
z^peyBrC%iN9E3<ukn;OP3>E{P7Cu_H@cv`3OybI|7;foXla}sy)i7EgdALWU_w*|-
zJRD^#54)J)0R5{{2)l=tn@7m6aJ*ctjW|dF%;*<lx%>Qeq|!#un>m-;(6butp8kvX
z_DFA3On<{_o~UJZ<R<DL40i}~v7Yg?1oDzimFnv1!dS+kh4LrbwF~s3|N3j?7dk1C
zOxCwrmAnDJ=;4RMnF?c|=j4apBFA0Gqhjt0_xO`5ky!a86N7%l1FYun=27Ox-0s}p
zkD)hpD1|TXHmY|xG8gD8Y6+s}jtLn$r#`ETOEp<xpIz+(tea&7?>40~j(mJt>6b2;
zE|j&l7k=Z;;P{esOz505=PP`2?c51ll6ZwRScCI|DSx?_7wrUhMX!OjO}l3eEkVX8
z8FNL>E6f6O73T7RWqKE(q6$UjL;G5(i&?G^&%0D5I!$9U!uu--Z}sHTrBW6_ky`M4
zoPjA3eCLTxQM=wne6>uQ!wVftZ*|L+<j^!myht9vWx$&>!Z?K<Yll9{*n3Fo?r)|a
z0C)bbD^~HOy+6?siLfE6nvqTMxqr<(FXsW>&}!U)uRR<a$2OqBIegQpyuT-$UdyOh
zh^SI*a(|%ZSjP~z|BtWt3~H(i-@QXX2#64hib@MANbgOm5CoAXReF;kz4x*K5kip;
zp-Gb(n)F_j>O+^_d+#6w2<7Db=RNa&Idk$Uvy<#hCTs1z)_q;~@3P;meK;<84kA8K
z>Eva$)6+MqQ||mIqga$?frQk|$~m%Ni<qjC8mAu8>W%BA{%jtg3kxWjuAF}j*z;sg
zf@oQ_9qN2FKH_OkJlhO{zgPq1A;7A!sxIF3VsmsouLKC9f*=_#4T(u~Lk%q)v%8h4
z=1@c@wWp$0Mm|o>Ts6(ehrQG-+gx{kgqTz;(_kcy$zI@E->qr2cdyc<wN=g9ekQ)^
z07ozCJ!%cD46uoD6j#CJ`_2QHMF3a*7oj7sc}wuof-aP|>!ZuIN|zHYoFD-7+NRNV
z-PtfBX*Pw1o`Jz^=~c5~xm#b90xx5ZwU7T5IDnQ3kAUK*fNFu>Nu$|e@}UcK&RAv!
zJ+YF{L`x(&PZx#3S%<9Pacb7G!aB-t{4qGcNG2|(9F_NJ$4RE48rUffdUkW`0bCt4
zMT-g}5C_nETck0-)K>ypk%86ij;3fqZq=c)b!Ryu*iJ>+@A<jttu-<y+7+HG4bm?{
zd#7<uNNa5yon#PYV_L(#Cln1rTEpoPfrhBAaSbFTWqk6bdz0KF0;c?#YFnJueOngn
za(MeLN$td4-(hkc*g(TNy(H;XdoT==&38xhzV#8?j{38A*L7`_yw`+Lb6Ubjk;mdm
z|3?I^-9C&AKQQ0P`Kx}`{OD~^z>g!5{C2&u2hic+|BhYn9ACEr^;mH!KNZ#3W9Qjg
z^)xT7rgd-KKD%NIbs}m6JX$IC74%p|cfLu^@9h8p$!{$?yKKg5dNN^PY+SpZZzJa{
zZ<9aK9sqfQ*3&v<Q`MvzmmHf51qJ)OsU~`Tz5zC2crqTo#Sx_B*TnJyHo>SwZ_e?O
zTgT*u$FA4GYRp(picyW%?moAYQ*yzvuTi(a+`NRZmv`w->1A!rPDaJ}m0LPLW9HCq
z<wid6R7P4ROF|(~4L&u7M~D0y6HsAwD!lWM1{5bYKk>WzdPC+~)o8K1@$&e_D#LQ)
z=<?FG_rd3Z8t@Y$nooUR1Nv59W7D$)_g}lu)wGt)Ad?}o<oXa;cOjxwu(;;&DuQsT
zjziW~I)sDzZ-%Y>cbPEs+k?pZ4w{90On*9{?n-1=I)3%2atV!gc*V-JlTpsKT&h<g
z5BcoE#3KARU%9e~)$U0epA#UWU&=ZRUG1>)Z_ZThaUx9s^T9;ca472Xaeu!Cf4>&{
z2BzOoSxU*M?E_a|U(bu%&_lA*7m>5I$W0^_?DIWMT1qi2;lB0$jH^}h;!bf3MH>)G
z-%M9qF;s<h--HaxbU?{*0ei8;%EezOO_9Imd#s+~B9dr>f*El=%fs?}gcH>UW!&7G
z+?%`XSp^BWI*a<h8R>6kJo|)}170qnCL8t;P}FaHSYho#rp#RSH;4aTx#=*I#umt4
z9jvbM#KA3L#6j>-ubm4bVs$N)9UvIM3P~;P+j*#w&Tz0OCEm>Nw|f1N%oHo*-Q_yA
z&4>4=8tvcZ+k>04t7`j6)Ff@NLh#20l8^-|0vVRKObBWoJ`E;@g0WJ&$#w0tvj??S
zeWI$PS=KexIZv#4C)E|(+hGJ`_8&{_>z;}J3_cLC(K9m6=w9AGin<9gXGM7$;g%r<
zCj+)bhGw&EsB9zf3%qz#w-NXQ-U5oo7|e?|k5V)S^W9rqL0#pe0*%2mgxqbuSO0Jf
z4`VRnJvT$pl_jdv80-L`aE3u|?+^T@^zpRa%g3%42OZdGsq3SrML(qvN$V$4{x>##
zHfD?j8|^f&q^a1)Tvuw}^nLu+j_23&_qB;CyW_Qw`MAefIsdXqcZXBzCd{F4vOLqK
z0q<js2?B;|qK@j3Wt|lvlU1Cf$VBcO*lYD!c=>tXo_PCf*Q5h3q$iF*UH(=+Yxm^l
zrUCj#0dI|-ODlyVXDyc&yWVq+@Ha7d3V8jb%JXSKPpI;OM`;fka0S9u@;uJEGXv{y
ztaJq(JY>7xHZ%S=SC<Z%fR~@wgAPYe-xpc4>-Y7W+px>Vx0l9WH5T-2x^2mZE%p}I
z$gdaclfN`|^9%x`+Xn~pN#%L=NP1Zs^Br7I;wlt^|JVKjzO;*nD`hJA4}~SD^PhpI
zpn9)`+P-_Cu<lNz|NNsKqnA{CU$WWj8?`wt^c~nEIz=etU2aa!P)*p-gpft%V?^zG
zsr<G|hiQf?Ip&a_VKGw$10(B8%`_ew;@&jS$COf3QUNzDEN%McSffSd59k4kpLj)=
z;y>xs)4QJsI6P4*s+}pZlI6tn%OLtGrZcWSR+0%x-AU>@J#!x@sHt5T*$#GsDFhbq
zs>n93*J{mQTDN$AF5Sqn(e5stI`MAe)VH_py|H9Q)Xh{2vLpA}zB_q1FD#79zJWW+
zyp4m#Gyv#xAUEEvU-N-<Y`_&Aqa*a&8<_;C8E2lY{z9!2ZP=97<m`$9qWgscR3u1N
zQTAy8x2o!ybC~GiAPD2u$NAWSjrLP12e_X)2<d9cgGWfb+nBN~5|gSU=;5#<gX~2R
zB5sz{Qj=_1zjnOhja;mbY%7%ql7lNMCdtWr00BCM6%50Hl;4%x{6xUEtAaHK7P2!;
zs@^X6*iD+aW}C#w=K0Ye!xlB0$538<5%iS43%o%HA^7-yKN4nFio-eVSL0I>!;o+$
zb-RrPIdcV(sMRKqLn9+2@7nZ+Bt6I7O~J!llzIf3Po9_pnJR2!4;ml%_26O413WMg
zv}wd<!tgyHjVJOQnMR{jWPz9Xpc5rI!N$)d_vx9*K3t_^bF!~g+hPOu)e6$XD!Guk
zAT@_`?FYFx8F6zDtqK0*hXvtsQD`(CiRk-?It<k{ZcORPj0%VG?!{{V^LQZ9ylj4D
zD+>7@b-H?kgIBK#d=uPMt+V;#rC<)AM91}-Oe?9#WNAQ!f-bfjrB=M<rS@A_2gli}
zZJ8OFLw#B{_XtS)xk|M(Y|t-t-hwvUiL)r2AoKke2ZV340!Z=)31s6i3{f4_J)%jq
zf}AxQxEdV~X^*1-u5vUsqs4%0k=RUWHxvx{cYA{YE^ef5mdWqHjEF%F5P&7fbf3GC
zVv~o0#`MS2`|SGOv+CPv4S4gYDE+BEgG|pnr6j|x>Tqpnct2u@^>rF-MSD86iUj8?
zxbm$_H?0UB!4nvdp~MY67d5r<pKAYZUe?jDWPM)nUJ?Q&d)esOu;-!mAjht+=NF}D
zRWPHjq*eq87G5|1woi?3@orFcpTMV&nYE7V&bEMyg9^<7xA$O;7w;lzGNtNJzbwE(
z5N}=J(i^woM!iRXg&9KcZgl`n9eGNUKAvFAA+B_TVUW5#3hXt`z!{vIQP~KvwhTX7
zxu$3$&c&v5+`WbC0BWu%>ug7G;&^w^(ZoQ1AQjosY&TBV^!j{1xazTE+q<{7K-=7I
zDtGxNZgRG@Hv2cAL@mKlcn+vuORz5KB41y7KxyomLfg$Z6z6BKsCZkDUn@LP3Fo{!
zkf!stohY8S55f+-ynleknF#93<+xF2r{&zY7@CB;!`(C?@2gGVa#dC81{bb0iOTs5
z)+9JVwSiDqpTUwOZOyl5qm#HCA}f@P6<82hl#yR9?SnKafxpwreaa}n{Y3J+p0?Xb
z;A%Cx&-s%7s;{p~P{P>3tM7Cfw|bFc*l5CHX*9yE<gt`-KLuZT-CS=M&@8*Yri&&*
z*z0MK;HtK#qoZfemq*JT0?cnd5OkVtK-9-k7W#F6$6`9iyfG00bF}@*2AwIkEkFzX
z;fhF}%1tD!e{^pHg|l(%zE(incHt>NKsHvVB7S!`A8hko@=CLTLqI=AHSp{w!`%X(
z#2qgx4Og=3<8HgXL>b@i?@nf-)*n)E{G4bUe-yOGJANaII54>Xh}X@7<&8VhKRy0{
zQAt8qnQ#~mI>ON1VfrF?wsJxp(ErTz<^|(@hbOc(`vP<M1L-c)S+YUDtfK9=Bj)9P
zo8q>Fk_+=v{h=Dj=T1BxFYsQtlne}ndFt>V%=||uHYw~iI%DT9SL*F?<lH|!xDdQF
z@8n5p_4IFTVlx6fJfUb*H%loK`KH@Rgnlz;roYa>+NNfEboXrAzqO>G26)9nnPU~n
zr4}2l|HK^X&3|v$f7ruL&4s^|v8&hcJ-dw&D<Nf>f}sP~Z^XH~af^pRjGCLnFHC<7
zq4F<yv135l*P0eur}4k3wG*ctDer>D^s0*{NF-!Q7$1B=N|!ZQlapzVPhc)<8D8mI
z^6rn*Fqu7-*A&K;LSel3zXI*BEF{eRqNMTM)pDGf<CVl%7YJ8U<^p&;cDRKo?5_F>
zhNSNHAJ;fAJ_M**{=Vnz<Y6el^7NlZ!pk=&Uomj^?fcG?jt3;a%zYB`Uc(J!N_Mi{
zXX#EZy{rA+Zh$mQJzOAawlM_SZ8c+DWvysC{Q7WZfcZg<;P@l;9UxDCL6PKF%t@-@
zH;{+Dt&RPH25k(_zlui{R)C)C<i+!^G&K_wZ1_P?JeNqAS<a(MLbreP*7KQ4RNb#!
zX}WQcbeIaO(uWX6rb=z8#j?BeI^(<RQxxvDhkgqW6n!V6w2FgPeBbCWC3~aCUd}x<
z+1D`6M-Uab$vQ8U0oXGSMciJyVtvc{zNw3(2Bl~fhpfSw{xVF~{XP>e3O;L0N{0XT
zSOo5FdZIlr<VydRZ!j-$W4vKaJI%0s5=VvlpzQIT-Q9-;yk+Ml(dKXEMrCGKQ*(aV
zx4G^vK}WvZ2$qJx@@O>12;cp6YQ`QDO>5trT%K)P`K_J<V9=N6uNYyC{8s{BdpuUR
zC7+CSh~Il;`iXSw&sY!~_KtVp^bLH%ZT>{NX59MJu^za%^=QMQnpQ)pS)?O*e-b<*
zWO|{gL7=5owQ)zY|E<CA>f{i(*#>Uc7w^h9%Ypkis)m=o4yT(<2*@}2tCtdPO*)wv
zh@%K6u~7qa(!~wg=<$hLd$|KQ!S|$Zl8FN$AA9H0b{g#z@;nw^20Nv8ds*k#x!DYa
z_4=&6{%ybK%Hxv6Lwa&_ba8a#`1*T@=_ld?O>143ea-@e;mRL6ANeJOgh%IGe~lGu
zscq@TM7W-t%PHB*?iplnrh|97Y2*sbsUMk^mJ|n|*5g7bL?dN!k|d|wydspj6J1*}
zi<=*(r|YIFGOMQ0x77S^pucHEU&cPu#tfFa5DAA)B&QFP_7cDB4J`OC3uNu1M%EJR
z-7j5JB<L6M_@E<OIPX=p_CWs>JeT|7BiPicm$H_Alfh$;-iGQKO|5w(ba=#%U29CM
z6actOC-jQ(DV-VkMjhz1G4ztsVrsp_Fz2~%W)7rpn82SrbaaGe@j<RKAig^vZHIC*
z<dgZ*q6tK4pu6>!aLJm(?ntqfQpe4g$W@O189h3t5v54|QEN3cx+XR-c|a2?n;;Nn
zk<lyeG*G8yLztUv2BC;mPE&v)G~D=kfC!Gkof%8$XLOBb=|Dezm{@X}py1e!(ff&f
zB!PX~OEDe{KeQ6c-{ey|W3912KmLlBIJV)3k0M13ZDV3z7wj|n?in4;M%3qC>KFa8
z;!4{*`MioGHx{Z5)1#N?IN{`a|Mc{-kur`)J{Hz(MXS<-TZM{{5zpE8?Vqg!R!y99
zR2t^UQH2fzZJ&Yt$<6&sbOYrFYddyM;aO6!+EPHUL?XlvCBK(M<_O1ULJ;4icw`Ag
z;=(*@G<BMdwCy(YahG;+3o{Z}NX+?P8XUJ9xWcs=4c$%IPtuUtyEo0_)<=Vzs5oJb
zXE?bk#4AGv*MD8ZJ+^c(1H|@4ZIp>+#zMR~#N^(E{-7nd%E-wkpl~&LVe)HgyxK_d
zrSw}8%Fi2!iGDmUNpVT>YeyX4A|a}+BG3B|={)N_YXmxP;)Hx>u6{c64wpBxKALHq
z0dQWrzZNHz9ZUbg^L#R{w5Duo2XAwOaL_(~fbLf+$}Bj?dS;JP2^gabfckj#G`BIh
ztL3CjtSl+FB<Shy(opZ5jv#81;#};;u|bFYBp96Xrc54<DHxp>pfjUY$az{==T^L{
z?J|C=yni4^<k}NuvIw*qoa&v-C~9-=Xep=Zevn`K+X<4)>=Vd^5q2X@Ar-;sf#xK2
z=cwmhuSNw*^jqNoG1Kq4F=xAS)E%nDgS?KjX$Rb6KG;BybsF>=FJ(isB@)Cj5`dAx
z0RarmCO|hN7&+{pgUs4`oZB^Md4H|S&!lhi)r@qb>?sNHkSh;FfMn$8vjR8>2ID1&
zla{cbpY+;mnbdQMlE34{h+-BL<#tG3I!?D~xDAQ!?g2Fu1}@(OC(_NMj<I%J#$w`a
zpEvCcT$Lfz%1q`awq|rZ75`~F+UaK&EAD*6AA1C43#<!l%AVx#<uAARGP~>Kq9K2j
ziqt5M4XkC~$PYoP<;5W(4P4UCU)STU8BB=S63gS&u8A2DU*uDNN3RN_)D?@azGUfj
zZEWQ$jR1De!%5qCCi><d51q!dOcBgP9!|vq_FBpWBXiY=2e8TYIz8XOjVMhz6T@t;
zlX~S;vrc~@b23)J4sVGBVyEJ>M`l@So&SyO{M^Mtrv#}e*|JV@?1*2TxbE+0x&MGZ
z@{^}kjHRK0!aL<kaH78*ep35<tW5}3B#Mco1|WPAL))(`SBm`HEe$W!vnY+BO=a$)
zc@>lTT3SmrBY)4@YQptKJTIZZ;BpbGyNP2AMttOAd(t&;NNcSqSr2>KP4i?n#L`e~
zH&w{l1e|N@U;Tuep@iOhZ)SbB_+5)#Q$UQ_XZ4*K0b^GmAAUl%RZm}6M_*5*iQS?T
z1gNlS*qzKMQ@z|#ET`Q#TOyl#G06A8%Ythpuy-yJ8WuE;M>s}GEM4x=&c55e&aN|L
zFms6YUl6XktfYP-bh$H*&(AaIgp;)OxG<^t%0+KhtWgKI`V5|nfY+p0QlERnq}avr
z$TFfvuY4lE(tzvy{5+L<wyfw!_?b07P&#SxH*VQe5Ut%ssK$>z(;)~J$`tIgn8b9x
zd7q*_3kpC}=6dfyZEN&U>NENJ<-&5+nP+Em<F!>SH%;f~hxRiq$usQ)CV`Ol@sb0w
zrH!1|@1A7l%su1Be#6|%{FGFq!*X#xK4tFuVl#>G%ajophYH7|oi$d2V|m^JOq!=w
zsVBLj`lGJu24gO<dfHW-6qZc+ubD>nJ0FhB&&gTPoQkM0{qE1<i|qKdoGo@warcaE
zb3{qjBIT`XXxaXwi1pjsTU$woS*~vQ80fVpeRt)8Qrb4W5pM~u`rn6Q6t{MMTxguI
zD0!Yzj@+Ac>V7p^9z>0q<F<(+P-K;c_yy}O``#d`q~RW_&eubuzSB5X%Nqi1m}m`e
zC>L5apCsen8{*m<3Va16s+FcmR$s!6kdHfTtqy%A(!b<r*)lEWrNh;J&`1-wWxbij
z6NsVGKn7X|_%zw}!9N|K^;pz3BREr>`p9b@O3J?=f>t@P2};Gqyys#Ax(3>(^=(rh
z1<W<C<)9TvKQqZBszq4eAd9Res*6_K&$W9jk&ym=*i%y_3W#`9yU@s=TrRpv0W~IE
zqU%?|baLsXE~UA#3`KczP>x=B0#i>Yk$fmo$4|nIlB83f^;jl@X3Xz-4?}5_SgR>p
zIE6t}6wX`>H29uZWrzYwFC#Fg(<19OljYGr7}eSXln#Ck*t~>?6QghIv53@g>JNuJ
z!O%HZ%2<wrw@R=VjB5RY65@Qy9UnDQneG07$8)I(B8#k=bgiE{4y~HxXTPHj3TR+e
zFr9>CKv%TY+vV=ffkkO}x?98isIG(tQ)6Tt17N!jzDu!~=0hNRx91Dh%MF@WjJfT~
zB)H-gv|5SpTYT6aCkp0>^Pl@}6iFrg6dl7z`u<x8Ei;mM3MU1H{E;IZp%{J_6qk&Y
z7kg4inLhSDn<_|)ot>%eY`6n!d7g3EZ|V3sPGfce>o1u$Fk2=jE}8B;m?vGTn9ccP
zRNz6h_{prOLlV%ZyT}6XU3K(zr8b{iT(r@6DwJk5)xS|Sm2saI;??l6%(hChQJZ9g
zNsgMf?rYQ%<n5Nyuz>ik#v*@=h_zzTZezaHFNx>|B*;@MVEeo&fig5e0XHP+g(KFd
zT{aI4IR-H|^b|0fpx4stM60RI2((39oL1j{>T!JV7a;AMU5B3@VYgMP&~9SNdfe_#
z+k{RbTUPOufEB%>qTTGE;)GN0XyNEqsC@D2>bJtc=M%)llX7kzL|P1?=><W(MK*M_
z6r9=N4Wbl5^~lJM&rycLiWL0CvLBcgi&6$2DThb2wD`A17RjoB-0}?N_Li2mX3l<$
zA2Y55bOJ1aemcA#;&UwG_}0V|nprJ0IX5;G4%QNxsSNPaw>*Q3ds0P=laGy*stENS
z!h3bCD_Hp&Eige@N;Jv%p(U>$it3RQ^`zlCUBbJi6aA&`m1%ykBk6*lXuHtT{-V7P
zl>z-1WzA?txIq9#f?-*rg%Bnlje#;JN8Y06%>YzpRG`SJ-?n91X;l*+b1<mHmPq~L
zxhax2KdG38{kWDZ^aC6>Q7A>ahdZ|r+VBF_CLddoZj(fnVLBXHfIdj^!}z|YIhIly
zB;5%W=?VyBp&&9Lv%4?z5qiQ%7^0wck!G$?R9Z6O#s@hFrZH=`8#@r<xI8Dx$6W~l
za`I^Hj{oK-)5Z}W$#(_2k_X<WAhV<GY<)<imn?@kPdSPT#W)beqVA{mWTrHI#|VOe
zpc#VXI=lc`6U|4OoPya$@K1EoEeQFq0n!SQYmlwa)(0W|Zb?)!NY*@RjU~1Q1Lwef
z<Xj#3J$^=Q(++jT#M@-e{xP_8@H{0TZAJn~+!{3eMNPrf!}Zl`ad2iVZ&)BKxhF@5
zT!+DP?W7~xP+$3X#d3S_aNWMy!r%t$TNy_xPTf<1cYl<ruF4#+U%TyJ`iftknO^O)
zubLCBA~?#z8R2F^4eWSKiPG^&B`{c}hO|sUyeR}^S_%#1zyUmY^P)roHAaDd^R~Qc
z+}P<oN4FO_>Yb#ZFk~1$rON8X(ep!w9xWxj;mO^^CVQc0FLqMPY@X>nDOg^*aI9$S
z3{jR)53hKqylE1kL2(bnL<(7=r<tWrim-3^JS!vu(VSLyP2>{|?p24&$&f;N`*D2k
z=S*SfC+=035*||K4o=yJ7*U`al-N!Fi;f!alm3GCi2x4zvBoFiwp+19nT*ca7P0R(
zeZ?ediHV1zXD3FA*9=II4@svIZC4qK$6pDR2+I!Yi86)p5SIP&a-m`WU`(+wEMPY(
zB)Cq<v#Y%|m{$9tu68!rbk;2~FYaffiF8ha*(?q+RcFUwmeZKNItx5qAGdb#t)8)E
z&V|;;4HQ|A#2I!fCh@I{uQmViqUNmO=?J6^5>0dBoUkN~KORw7R;WpKr?Cvg5j8i{
zZNiv%ajDon3U*XI9CBK8V{ZJ3#YFOxQRmtnMw0~1%UhyyQTn0=8ncw%V_@$Wk)7Qm
zfdqbiS`J!U55lJ4Y``h$J@gxj*qs9Q^|zl>2`S#k?rhb`Vs{j3-8|r(&75NvG8|0S
zAusZ%Dcnx=CR1Dn-8iBP1;_T(cJa3fmh!>8-B9otsX)7)(%VF}*F0)3`KaKK*gype
z^tCOLMxz}nl1~$-)YXsF)l$x8_Rjc3Hcx0jG?FRm;$E{r3Us_xe9}{4B&yx;U45qZ
zzMX2(yls>@WRpAHZsi{9`qt4_f5TT%)}xQoS~O=Do`UOfG^rNk(DsH;e-656W4h3u
z@tSBg(rhcex_C5(${jVsrS)<JDyzhsPm?@bo9Rm<B&!@URj-<4_<s~K%0@;|g#uj{
zG6gRiRp|p}0kY-GC4Dci#($jU3LMIqFvrU+?jH^$^-T#8)1Bs8uvw(S^M&&A-&Lg0
zp&Z?Q_U$@*X>l<mDu)z_e8hW&yvG>-OoV*{Y#`G?i${YG#s`Zn3q2;nYae_0cz#x@
z3fS-+Jy!^h<aN8Tl{DtqJe8}zN^UBi(Jriwdk0J60}NYFwh#tI#VYF;o@x|K*0Wrl
z`NUzH@1;He&EyT>gv1O2S;p|HOUlo`J8BBs-S>$hMPbd!b(lGqR=baO2BlFV(o{K3
z_7gq!KkI<*0SdIFEFUI;iRsZeFX}^<ac2{^WNBC*wU~SmP77}*RjZ}_jvf^tOB>*^
z&A0Lf!mq`_Jl=G^5!(r;$yMP9igv1FlGCS6=s!72(p>l(R%50d@;$qz9J*_8MryG6
zY;LWcdi<Pe%M;CrBF>E)=ZeA-?l%LV;%JJ*S}U1ZYC#W4pM&4GDOHBT#bFF=W@u15
zEI3(4sXsP`&5e|ia#SR0_e_NBWq#>O2r>!avwG4(JrJM3y6N<vY^<%85B?Kz5Oe5K
zRDe=4k3jf9p)b#cfD+%2s5CDNVuFW%VsH`f63`mEJ#4LYW9k>AvK+UuRI{(0m*;YS
zirnKAu*TU5Y%Eu|m}%|o(PWJhSdljUIgi*=l4PCKhdt80N{{+;^)!`zgV4Edm8ACp
zwRtZ5pcLZI<1VDUnvM1#&<o*qJv{P{+fJQh;~+-c$;)tb7!s}RUB;H(nCM4lnzp|!
ztwmRUh+_h>QTeR%%>Es2q6%8GMdLjakgaJ{2!rK9%_ct}L7Ls9OIq4p#BX-lxQ^_~
z6z4;J=!|D#@BIB~guR~PrN@N#uJ-|`G@P!AS0x^0wm>o1n9D>i_3ze)YoVp?;~w|H
z)9z;a6YKa;zPP>;+@Qb-KpiyZkrghR@1rG2?Cq!mgX(8;R}&tDY3wxGPpxr$ea;h@
z`ml&^r9Y5$EG3q|qL_h>;m42gCi4}z(hgfrfM&dqVn(+qzkYvUQGL*>FDyV#h$G@}
z0H2KVU%a2v(6CKIgZ>%aMb}u3cIg15nd83W8GGA#20KylN-r$n3DHg}HGa(l?xcCM
z3X|2dW#(v`D%PLEb}VDGc6%+TXfl5L$Y_%2NB2ICwX;1Se;)kbplMc)*MBxIoI^p<
zF)D>Dbp}aBvQKwNUTtWMC@MVxM9$}J8w|KGNy?*WHYO(s%UDD%v)o(hCV$8MiwnKT
zEi`L4Kh_dd_-VSN+w`L{A18Kw!&8Dj+9>_;hj1L&kDj(HN~eb<wMGTR>pa<Y+5CMV
zwKdwh`bim;IWlom0P0MI->4CZ%ZjO!GSe9Jn0H&w4yE_s`_B!p)$<sz@>o1;cXrAs
zlKWJsD$d6UhCnHX6AHv1bRb%c%5{pzc}P{~kqb)Hkh#-@_Y^p$RyU1IiKd&TJn2i-
z5c(7)R!fGEA7^S4yH+i%HImJIWsVTP$vw`^zLse-qwPB9+7**xC#~wv;b178W0h{X
zaPa^_VxuUI0R7vnLIm(Rzm~xGPTAHvwmZ>k@17mw9y=m=Tj+v^w;)hJ-rBe~zp|V2
zPvy=RXhH3lsz27>oxJj5Q+NK8Ln5s}N&|iF=^RN$Ez&tApv+(fCJTyzm34BLa&pP!
z1VGd)*^n|~s&=#IIcYVyifY?x1afDeUlPfu{W#-!c)y?55lHL_Qwk<Sv&nT9dk@!M
z&Gtt;r8a?Z{tzx>YZ&v@ej$+8C`p}OXT#91sFxQO>Qv%k$EN>?2SLW$38e!egTAC`
zZywm11XmA)p*0?AxqMNFuLu9Q7l53tE&ZW@U?QWw2;O`W_64o7`q8SMO+pZ#5P|r_
zgb#sz{!=Db2{e%;2kUj-Cu7pq>~i_QPKlH;F`ob@zEd6;`&T=WXWucBhQ)0?@zRIu
zs-mGXlVM*T;lVk(@YH5?7fql2*e!WqB`L^iYFJ@WCg4v%ANB$NGX%%LheuB02Z|EG
zWx#@*2$Dc#VdX2$vM2JwRI#kx?SWyuc?uQk)E~KC10_A}29}muyXeY@F)LdNR=NJj
zWFA7TBZGDcs5|n3sbk5f-rB^bFTK4cVMX6yyaIvdkqW+!KmE5Virdb%63<>Jz(U#~
zVO}GcQ{7Pv4u-1D5pAIty`*7spV!e=hmziKH}P9(BEN|cA4w-M&+B^89DY#aW0!lv
zc)!FafJ+gm4n5zeGIj;EFsI_H)S##J6zZRFJXW`P+l8Uc(`j2t<>!><T64sK_-UI=
zIDG2a_?0KWt>;N^e`nt3*-qw~zOJ04T&J}ktIuQO%zosg4d6^Oo}Sh81>3=O;N-2n
zX^Y8x<{Vu-61&;GOfhpVxLby@)(?D;riEKGK=}cV+rZOwGtCSpVHU7Uei+G?cl`fS
zP4hs4Mb*w)OpI&d57Gt@{0ZsrSw~%cWd5kqlc_0oCnJY{iSY0-Ui^8@rmy^!|FK|V
z<M8UJrchetlRlpR5^CbKCrqC43$xbP?abuwKaD~{VHcYDW?hvjUs6+KS#|RY6$dpa
z0Jp5<nM+)2-`-b#?53F#n-Hm+N>xUPa)x&-DPlJA9^8eYV0-Jxk(Nbp7C}4tdp-V1
zl_Kj)RGggRH~i=PGE9+Cc#L4e9H_dY^0T!3N5|}*c=U5+s{Z-rxky@RMOM+r`0kM8
z)T!DEJ2NWTkx0k}&!ZOiYVW<BUtbv90jCK;C1{Fz7PB*nmXG29ZP7^<kBDO_tGXK#
zJc_rTDVgato-;DhS6QxFeSV^nklDPfa?m`Msv*%1`4Y-s(Kt?<%=V|ydiD%Y3>JgV
zK4N>=I^gCcY>p`V59zk?D-qIXqx+I^EVVrSJ>7lHhD7O)+X2w;6dIrMA@)i}Kn+Pk
zHnVzMQ({gOQ=_5M^H1H4_T)=~#s6v@n<6S^KWjs5O=y0vTwe}YbBa<i)o1HE9-1c)
zyq;Oy#c~k}>A-E$hTRPfxt%x}X9%*mMAfL0Q?NoXr2|67rLNM>UYcac%g3%P;8>+k
z_9R^JKzx|jXlHU<(}PtJGV?!OKyU4OAM@;>EHjflO~Ho<#@OTvF)%v8WKfA%B8Mud
zaNB(wpGo5dDDl6+5x8wyj2SqJ+_?3;HI%-omHs|+6mKw;qLJntw-6}b-Q*1QbhS$q
zeonpZL&EE5oiJ;IFX3v7pBwZ-TBR4(J*L>-)vLv(oHq_<Q*(Cm$PMTP;G|2WD$uZD
zV%jNC|Az<8A1I_r0@DKQAxaZ&hEltj^CWqBLAMN9oNEV)PRDHgm)N%|z*p7ejD+Lb
z>Pj*>A4S}M%$3j_quy_85dHN(Um;XZ$4^nC(h$>xoSK~ws76i`cSdw<zLVi6%4Hnv
z*lo!0f4fb)sej>x;l3X)#MGda!CfY5d8CK_S^nJFX@tijW(j|&adQQqVEe^Ovoxb>
z<uCj?iB`129f&GjokY9X*+SX|iF(yhI6RDrNl4lGXL<oAN-R74A?vDuBA2=*-&$nH
zTkT?}-ng!svaIJ|*z=J1EpdK6bL3p^=dr2Sb>0X?87fXRA|F7inRj;w)2Q2?pATk;
zy0ji{ng}SbirD8n#rMEAaTcGrsx)oyrl(EG-N#2;pZk9_P*4%cfE~rwa~7FjOdS1e
z!lU5%AfpJ{>s9FI{kpSm-ud(8pAP|lmOoxAFa|?qnRti1X!33U_BV|c<bN|vYfK1X
zeGXzOTE*?dJE8$YI%nQch0frxhOoJOXot|dhqM&dy$|Vrf?}Q(zGgQmbkL>3Kkv;(
zLpQB7$C-Zr2U6*W1!cdZ;D0#J6~hRcj9p85-dmS<s%;aK9cS(HAQ$;uhOtcGg?*X5
z;7$C`aO9i3=cS!~0cxT*Y=+?t?2LT(asx6T`1s>|R4X)qXA^ayTrXb?yZJ+$RgCRG
zIv`e*Pv>~#^<!QSkM(%uMttYKh&RTRrkKN?RoAd8$tiu~blQyf*~Bw47%pYVgIK&!
z(D2id=@cz39d4%DDbYU9>8ln+koD=Nsn^l7kX5vPF|0zeLgNoPz3m$-K0msTr=eI@
zk<bqrtZoC^3bpvvUh<SAsCN&jIh!J6$Y@g&nDK~M;P>!l!cEiP^1m!d|JDNf2q!fu
z`Awjtw!2~d+MG)ES)s${kq~~LsunK~cZY^OEVY-W&mj@L*P(+1IyvUK6EYCnZb)VR
zJ3pf69mIG4#O;+AFzpnXp!cAv@40b`V_pK@*Qo#i6d;Q%iXBUd`}jd?CR|~%i&jpp
zT`}~%pjMi}cSe&wC4_dk$`4w)vQSg>pEJRgY?aQ6UF^DZN(535$(qmHja-wLC(<c5
ze5Y)%)NJ?LS(IDD{jAK+vPzWEsKxzW|A|0p)Fj*M%ptCHDHqTB*OF1nqh-Pxblr?*
z@hi@#zepT9eo+dz?Ko&Ue7<qh%5K5x8eh^Y#AKRUaC~vGbhLD3?Iu9_jAEK}6u~bk
zm;H>0rXV{(JwA6<2=-fqEQyj4T=AuNTEId^=xlY>scOQhl1ZD|X5_u$jO)L47@r)o
zqagbR1yaPBH2dAVeD&%=cC9WvSF$lTJL_kdtKK>+;Ip}LLzym@`x+yzEULOHvSAfh
z=_fn(@;yG3YNTJO#x>V^Gf>vVIf_Od8J@3Z^^VPN@mq1Z3v>p3|EpS6uMJz+bBNhn
zv0i${J(tjt&y;oBTUP?|d#I|h?)-1cq+y~&Leg@*JE6PFCp8^6UX%SZ3Yy~rY0w#(
z+1`eH;66yilM&pRD~%AW<X_k8-$jRO-UB7emdv@J^_*eduf2Zj$;l(%T9a)AYoe;s
z4g28(w8)?zX^IcNO#J2LTq)~R3d9dp{EzeiEa)biHQ|A#<R7-W5$C7Kf<a-LLxY8Q
z+21OZDZ=X5iFtWM0;#gtHi)0Uo*1^azW0&#VZ>F^#M8*(u$UBRItc%0Uu}La=?Cgt
zIQS{^-U8*o@VKVX%km!Fzu_rlJk~@7Nrn48HdA2o_Von8mDiG2Qcwf9_X2_(KhR1l
z_FOwmCWG00XFgoIn#dRtF#1c{eS57Z6-i4lZ=TO{&#FJliy;I_A19~MTy>su9So$c
zml*!Da#-l<?Md7ALvaI3BQ}%T&mBLlNGBwpZ&X!TZdeWuV0FuPb~`$pTDB=`48!e+
zDi~UPY*)U>Pz6vvQR76&n*n6JLZ#@m3GXn_u*le%3DM(Rut?(@f&2EPAZqA0q^3c5
zS_<{OP?}XCy^nuNMf3C122g_XHFh&J&nXoDgoaL*8??Imx_&&#GTfQX50Q4}N;Cbz
zrWAnG)6_Ius}19IhG@!4Ep2TmK*2rIN>H$n9h^itlQi+0HN~@>Xh1_q>n#auUfFuW
ztE0joCS?Vz<w2w8@@>TCVWt~4@|&zhtV@Znq`kfNiKnaU(UdtVimaTDfx%A5=0l&Z
zb5cR?$aCq)H{-bOo0+i}iIvkOc#OetLZ;WsUNtj??E+Ifg^|w-!BDN=M|R8i!9hg%
zWy%q9i|#OH_s1TDJV4HH?u(+aMv#Xa5sA3XQ)WsWJPc{qG-sT#i+w2)_8G_XVcld^
z)yU{(o!pz|V9NR3Yo_h`pM|p;0%FRJAYKZjk$GX0M`oj3n(v>6|Fp55f-!J3>D^X!
zW!X82_NWYgIw`aH{je{N{o`f%rs*NHGFy!ZA@&Y|hc2h!L4c(3X&<g+*W~H$04(4F
zKX=h=lr)WUGI)jY-%?4R{?Rv+%K#0ObQ>oV(<*Y&cDUz%CM0Vt)s!aC<o~3?n?>?S
zuqFVHHn*^7a-}3bhkurDHF<pf>}fkJD1bW5sqZ#gQnxf%{UPWDoY{4*HAy{-=E*lm
zATjv)7k=*!Q(&ada8#yr8y?B)XmL!smam~w6=JhcMWq_9n@#iR!$UWISC*5~ze4TA
z&;E<=+8?3L6mvQF`y)ECqn9FWID}ee?b3Hbuyn1N*T!thxW8wB`TZZ!@*XD%($NIV
z_h_Z~;Ds7fhiQ2g6qljTMfQ8k!2=IvTN$5&ez4NK7B7ID{TQz6K470$88YVOtC(#?
z6-ffI=6j$%Ude7D|9zD)J0+@PZHe_QNfGl}k<2&EAd0b6b&!w)NDs~oD%Oi1L<SZT
z&vg#xO?edAm@RhdKBL$+sqx<Q0e1iXvZh4ZF$PP=(4bCl=cZfuJ@)ZNmd}0^qsmg(
z1k&96<N>?PqG!1c8D4pX!GvonAP~as1yp@aU5x}3!2g<2O)J}FRGnJ9SPwr22_~Xm
zHrdPCK@-jjXhgA&=SHL^E7Ho?&a`GmAU}^*^Af<*PjE(I*PEFd8=W$h(MG%H4IYL=
zooSUR`9-_9dMM2kJdJpncIcF*wfh6uuRn1z0I3oI;vR(=2@cR=rfBy$7wJ>`E-ETo
zOQ4G;E&tFG)@?xY6>2uKDCTj{Z)kabs5dwyY(>%VW(bj(RMuMAuaxiP;_B&nd9uFx
zV;6T<>FHr7Dp68Z9oOTKnRHtf!IPHSon?OT<-sb{%zAD*8%ZW>R4TN)Do&QcB4iE+
zHa@15G$Vq>f~VTYY6hHncdxcy?3gjz<B8|uDmDYAsfw&%g;RgY%mEEt;E$cqw8WsL
z&*`Pn?0KtUEl1uOXX$R(YyF!Kxu95d*r(V(&7hCU&1Ek<l_F_*MUr(B3gu*YbRIjY
zVZZ@Pr0uPV3KjA6trj%fpq;{iC_GtUz>N|9s=;<bR_buJcHi=^Lq6_qFUaJ<P-WEU
zeB<dS%t)9Y?@Cf)PK&5~?J0yW437-vy!TS5lXO;EOvbf_PdQ3(1FZSwnv5-KYpKk-
z$jVHS>r)0EkJ@m~LlA-Mj`-eMqD~l9E&t=UqzbdSj|mWz6oosZi2m6ULLz?PXL~Y}
zaO?;S%4f=G5~MKI{WIOoZxS4UZj>dZg=nmKq2@R?=UB|52$Q=ebmarhL;{rw<r3Rz
z+96tZ`t^FpwIMQjN{LcMT;Iko$KC6l8+`+-Iq|!nt+#uAk)mtfkk5b`ufCJ4&(yS9
zm_t96pF`+<#|1)C31dI47I6ahz<k`lU+yxVHbm`0EjvkuDz#%y{oAii9+nh!f65mw
z0en*TrNMQ!0*P-;o7NM!XC2a4yBAncKBsP9q;53|SkQ|qskO>tgpN!k{r$eXc^Ppf
z7gl0*79_0?NigxxoZb(ONnT{pi?Twey({BuLAIgto}-&}I%=ZQ=nudc<+WXvgZ1Aq
z>wlrCu~(~(Gw;cU1?pI`FV|vE-%alqC2HG~1yC)c?3gm|RGv6^>V5Izzqh^3%|hPy
z$G!fYiwX8uynyNT(t^k(`}etSJJD9V2jld2ofq48!mad^3-N$G?NQPIOWSwxxq_Fa
zt@GRYx5sIb&+w470*-3H_y}5O@lm^^6aRJfc!6cP#I3Q;%l3Gd<v*9A&D$J5^Q0Tj
z%B}|Lw!S!AtnNRS>SR#Clor%C+&Q%Vr6Tt<AKjj|^t*e(wYBD|YG^b=CbsvpgA9Nk
zjghb);+~1^9bu!s|5!8)rE{<EsBroBKd*a;xJUWpj!Z2BWIcl;|MSMV<tfoC+#%CY
zF<LSCw*UQ6$58XlEf3rgx&y~O+MEA5H{Jj1Z2#AjUH_-kdV*<sAm$6*oAFsVKVPK1
z@)>k*`(c!9cm<G`Uv(ee%3buTdOEs)!MoiuygR_QVjehLC5A4q@E__glnES>iCts5
z4`(*kPq^ETYs&pr(}1%I&&AM1vCV?o24L#Q^CpPymeJ&_amdc;{GOD6WY|!W@oi%?
z@v^z`CA)KF7gx)2xu0i*xLDO9C8rdwuk&(?)pnOm47fI0@~}L;x;?yV_QB6@xo`P2
z*Cq2(`H(}^RR8YCR5+*Y-sQ5fXUDGplKjHIh&F!#;As1?9@r1FdspY+e<N#|c{lQS
z+8@hud)DSVFZS0Z(*p?g-nNtaH(w6S?$}}v4sVkVeeR|=fwhrmZOeBT*%)C<`rC>_
zuT^eIAoT!y#E)A1_x%n-ezS+&4&0n;GXa5&+b{#=cej&wO~#F5b9end2vU<n><DnI
zyokv-*#1{L=|9=#H*0>wpW!z-0{D&#`)@|Koras<RjYd5Rt+u@?cW3a^jb^<mIS3P
zM#%?~qMMhS{0XJ@`&#$-xwI{QFJ7IuUFHGh`Yr$54zIpCVB`3uPB|Bk?+mWYjc*@I
zZg|kM>>tTfcmxmZ`m~<)8Qpfbv|*|aZ!3Un*67xa*@dIB`<d9hOn;}xDUw}>n6wRY
z=$gdhNlq2~<`R2gM5r`$`=2mfL$t$1&b?Z0uYuU6&CHw3j6d7=7LRt>+AhCyw{QbY
zg32zpR^%Loet&GDx`l7o`4w>bYrY+XS8{(<>b$Jhj)vE_Q>WEy)gA?0vrD=Eioj-`
zMpnfhdXKbSjsKCj8`GJS_dTqVT2Y#0ma&yQw^_W}EeE<Xy`P$UURmDmANtCB<+S*o
z)|TH@y9Ccw53DTsV74|!zBQj}0l+rD@kL^J+oyxe0d62>w7`GWVdv1i=irFwkw^Hz
zY@+d%aJk<+a6Pc%e@$h44jMB0hhAJrxmL!_7LBcrMXUvIqG5DizL3#!y>fT*0Vclr
z(1BIe_Fr1&l|H$n=fvU_W-_0=dBXz3m;O{jYXAL$ryM;tV@OQab@{Xvz-BuiIo$2|
z(MsJiw8*Z}-EC2sc@8e#OxN^bTQ9%5yv9Cmy=z?0yb8x6xtiC;59!WECR_iBcS#+e
z9VR^+ono1L&+UiJG#uX3xc$Rj-q#^@fuc95w|MG9(sX{<wxSCY-z^60uMQV8#jbz#
z&>>FUFGJf-?vZnOUb5T)m$M#^fcx$)X&S>4+}O?Xt4!mgVPU`J`%=eUcUZA5soc57
z);mrIF9Q0DD&Tl@Fc)ilrE}Zg@T`SKa!141F$u#^4={98865>RbzBD$4aK8avB%u4
z>vTalr}51R_sz-O84H)Z$aP*_>-!xeV2|JUpNZ^dCYBddRSxLvrhW3wbSz&+0SBeK
z7uVcZeGH?aKDTqmUdIxuuFp=f{uhV76Yty3PFVZ~Jg&cN7A$r|%`8X#!u^sdiB*21
z^RJm_wl@vHl?N_*B&Q7}aTZF~<Ql0lg*CpZeC&HUG<UkvcH0H4^3z{W0-Liv<hT8}
zR#b=yBoY(=@Gi`SxytU*Z6oym#Z}49HDPgh4fFcLR+QAC%Ol{L=@%9RLZTV3iN20x
zdUE%RikZw4t1Ct?4$))(l>2QPUt6}hF5fr48C|^YVi3eSY(>%CLYAT`yxOj#gxhXT
z8aDlB-W%OUOJN&`7;bZP+Ro({aCe5CCryESkD7oe5jijH(Dk*c?`>b(`m@`EOfer%
zD>3@^vcnJlt8{Hb7aN8S{(a90LDx$?5{KzG8__MdrR**L%*!v2-!qs|yWT(QmzToe
zC0(SWT2^*j4~mUX+BYhS_owgQ)zI?vHHaS@?Ym@PV|nkcZQ3@!e=)rLb9fs|KKI&l
z1V@M)zq`1+`&Wu-y_WaC5_3254<-L5*1fSl*>>~Z=-+SDO?;;4zvY&@?e9d7&WAYM
zr-_7ry8J6ZGTdCN8mwhr|F<dWxOjSd%T04JlJ_3JGy1MX%5P-h_AJxe^l-(}Z?=O1
z4}{>pLhJZmozUG~k8&MI+(ZDrwiyX>UsbPx*}J3r2X{l|*9n%uzpBjJKfu2JA|JLe
z`dUpzWd79DW8~tlb>-#Vo|NAl)t=nY#qYMumHQp4cP*QSQ$9<li;iQ9&d=8A7EjT_
zhM4XByOpnxH6nQjC0ElHFrPB|=xz&_qbe>h!hoA%DeSe+?RsXTh!~dF_y%!R-*y9x
zXZrYWZaJ7dZWO&Iaj`CSRK;-KnCWSZog4B$mS{TxT5l(39LJA-4f&i6-QXtUaw%$)
z;cioE-ha<$k<aIsF7!Ic0UKLzv!=gz$Z%UQ*m5hEc{fADDvlj$TS;Vax;uRh@?WYl
zzIEG7%5097_VBwx0sE6z1q;g37y$U?(Xwyey0v&cR&EqV?*Cte@%1~55a|5h-Ewj3
z!GQ~4Mviw>rh0c$exrAj;PXoIkfHS$n|Zd{Hsb#`^ea~C_zNb;r*+HC;jS?AV&v{3
z>W9zWzxzXN>qG$Va(_V8*K^=yd?8qVo&M~Ns{c~_xyReK4x&}u2!#w?0hJ85<C}L|
z<aePS3}@t$cSFq=ml(me=$k7d@~wlm(*pPUhbOp*5)G*{hGuLE_f2u8+ks*ET@)bf
zKWMXfGSA)Ov#I1!Z(dP#DK&<>DBis#J-j1d#N0Ss)$}!DmU{foH`~@$x%W2Sis9B6
z)7K0}|NmNI7Vu<g!j6o5xJf}mh@o>2vhv^IftbKxN*IjsJ3{4hyUy<I!wLKhdQbg2
zx7AOcqmMyM_T&$?mS$H?C4QzgG<>JM5Awxcrjg^buoT{<SJ%-u-MY=b77_ie+?G*0
zF|V2m{3V$a>RR52pRc#Bzf=wvz6lowPDl4WM|-aGf&JNxru8{t@7?3f;GUK?)l{z^
zsM`6e8xFJig9BlP`FYjo$b!0y{4cmB7f#N`vtZ9YV|(kI#goxeOgoPGZR;Z(24z5n
zhZZ~r@z$t%hA#?=t9BHL&fDe*3wO^e<M4NpkT7Jjwd(itvp*Wn9S2Ly41Wpy{y%je
z(5#ZG(LU-iB%Cl8N%y4$LBk+;e%P(KPjh4iwSbGpc8I9Me*HQ9iu+<3j)s80EBw^y
z^u{;6!Y^k3*J?&=n$Ob4y#4eAeCQriy7;@>`WuzD`Mp>664#t%EB~ga&%IiN#P<mU
z;zN6XIoR#43STG>4Hemm-|n%H4-ObAYaEVQ$uEo*dG1;+ReuxqI4$BQa1=hYtYQ&A
zqgoN7y|J$GzGb-k1lpQ8&+s|G$&WIl-RwH_F3gd{X(1Zcsr6eO_#TIgKCNw@%qWI5
z@MA*H=2mytYLs{y_$APrCtpr)V7y5YE9Xj+f75cKyI26lCd^Q$L8~K2tEVH=iq=Zd
zWYikgO|_Y4RhJ?vyJv(cVE)$$R(gY$eXT5xe6k)|&sd)1S7tcdQg%TIRyL3oVX%81
zY57(Y1M0{vdnP?a)_Xe<y>xhxu$|QhAW&zd0um0&-^j^|0+#S9l;7kk;G9LQIcT!M
ze&yN0tA#pl;RqNXg$QF#N)a0<*_!p<95kT)*S9dUNEjiU4+Ns-7nD{Bu$7Jn_mHu!
zmk{B^M^kVXG(M$NIL(RnJu70Ow#Y$aQ5dJ53DQ1M4#znGtXg!_i23sp(EE8b<|MG6
zfnRB4!UF&Z@@(codw(Ump|~NHeD$MhCo&kWqC?+sJM~<L-e*lwYIktnZDRp@u=AKs
zPF)xz{kwfkAPn;B<}3GYD$Oj5L%%BJ%V?>qMjZ}R-=!}ZM82nZ+&bO`CkKuRDa$Iv
z+4+e#W4|BXe8V)~)kz)oevR0q_dHsN_8j~YQ3c#rF{myi20=9xV^d_7bQc9#13S)a
zh<4B3!W(rn)k*rZ=t0xc5j>&Hf_Ae}tFFy@baZ9?62Vggjo8fd>e@_0F%bju(GhDA
zk&o*us~@{z7(c09OT$Te2D$`wnXD>H3?`{c7p27P`i6V;=T4(Qrdyy1t<WeBPel>B
z5M}fo48v<e1$+ln1B(&X2$0d@;z1c=R+t_fit|`NV|3>W;syqOQ4IIb;Swa2-j8tj
zjrPwVCIm@A5Vv&ov2LbaCu$dIlZ~?7nOeEv@#E#=<7I)8c22^B*;<<`YW)>~$Wfkk
zj&S&_0KhZCw?T}QA!Y4!s){vG8q=E1H{?o+xa13l7=>zr5ePz|$PVwh$kmUUT_rRk
zS=5Gf4E&y-lwuKnqcyCAInsoTBp|D&$4iXHk}<pN5>K0vX-|_WKX!}>4=kAR+6zsV
z8{Tc`0Q!qk(`(za5tWH3%M9>KJZ99l46q~c$;eU9oD{i*$X5^5vrc;NG^-*A*^!&P
zqoGjdM7^uIvO&P*>zKgW!#nZPaKp(7yrvYG^q*-uXdne~U`4B#^uT`t>jwSQzfwe@
z>MgVgCT9<6OEOv}Wa`$RqiAyoNHg44XjQ<1^*5|B1)Zo4f7q_EqMNyKYA<Pn6$0`C
zwrZbFjEX3-k|wLir_v-3C}~1sp>$aa6duk*0p7$#-)Nb5o2Wksx)8Z)IFfe;!^z6Z
z<o-(z9RM9vwI&v|H=BQgYijX$2(VHnCudf2n8Fc+81+hYl9ltgG;X}%%I9zo+K}b{
zYH;r4Xq{bKmH$Q4xyLj4zkhs$Mm}Xup>mi*jB;4gJ~_AL5aqBrpAYLWXOZ)*qGpyu
zkxfDqF=ujEP9>QbOR~u^iG3vJQxbmr{_)%6vH$kZeeZr>_jPz(ujdjNdD54-i6YXP
z0qPs7q?e_3lDIGU-7rZSf^5j@<mza66$#JE5G#>|OI(R+Bdp|>`AIG!!M5Lu&kFjk
zHTX-Zg($u-=F*)K(7P0y;co5w)cLL-Rt`wD94T^ck~!mU3UB@4YR!l{=kD0s>O#-k
zhNPeD>cYIUwzS3goL_%C{xZuhEJRHQm=zrFM;DV0g%~UGX=QsG@S1=XKJ+l=N=mps
z2y+QPOY9dTL8bM|aCu|nkaH1^mkx*Vd4ZmU%2<C?La9JchjnS>UUASf10_QFRnF<D
zXh0nT2gZ@Q$L})U@-gt4_Yox5vyZO`uoo|OpCXHec8pIQWq&aR2aavJA4`J*P}J)L
zS~m1?@ejM#0~?8|N-zD{s7`U3^@|0&AwFmbU$`da8{?myyjOOY)lR7W#HKBM!$Qw1
zlx{UXAD!qH-#UM)O;xnzK7k};`YX2tr1}`5V<Hx6n40RG7x5xckldyH>@jOq<L*sc
z`w)2~=r++>Y_!Xr@_>F*H1lRJ0EL!-3bASD2<NbcZ#*h32f&jJ{T0j7gZ-qkz|r$H
zhyE3nFQaBbY3Po-NffBclIyjjT_3Ez)LT*R(-ak6ToltuHa?nP%y;{+d^iRJu}AqN
z1|}X7%oGtBXS@>(#XZ@aI2wc<9u%{E-bId<_e1iDLq0kq-&<7lS-<PDvdVtT;0{lk
z0#6{QW9O`kv$%6WxI!`rYkt+d1PhJsow<<c>V@T_yu~Z<VO3NjPD8oM%0e2MAWc(L
z1QABFBBZ+YCG33#%Byp(Qn&}}M9ysrF9INPJ)Ez}f-_7(D|VJ@s8snV!cA60Ae+Y+
za(wW9(eS&$7z4iuJL9*NMd_L3j~F)$=AXOcig1NDxmmnHx<8*3^IZ|-LCBLX%99mI
z<+&l>8iNzf3Aq<TN!_-{lc0C@FQr@_YF+K*8!82tKsQgxDZ=?7WBBGdDS5sgN{&Vm
zFBllc`8-#9bCE)jQJ}^~EIdAP?okR=GUADT$_+3C>}%ydwpA|;N*gML4Yz;#=cKX!
z37qL?JI3O{tHp^JWsAp())H+bnpxfx@`O?gqFzd4?%<$K`msBP%t1Dj=l$h0>OQ_3
z&~SSDv`AJgTl%=3qr?vKp2iyj-Y+zXE(_d~cCEu(!HWze6SwR^$pt@sFsw<RWMh<}
zN}L5hwN^;`bW}2P<;*Aa@Y?J&?kXfrSKS_;B7~WTIK86;@RE86E9Q3!a2SERXlaby
z7gR?(u-FC<j}kFKGF*^$natTGUt(!qcomJc)>EM&lDp#FAM{GFNyV&*^@ty^&Bu_3
z&y^}!S!vl<|DeN;r88jYtcOo_BitmE*LKsErnX-dsp=@k5fzAIgY;1!7+%56dLgog
za6iw>C+6o^fwfUVI?=DDTGz*b?N{jF*VQI?XJM~d@rT+>ZdcFse#CaB6ne<lABsDB
zu)6nfH?E@qHtrg}z2avYOb2@Tt}Hx!SluvQ65%^*%0Js1az}FdOqzr(_Htzd-ZBf`
ztaI}i;~?SHAuGy-tH!=jG0MHtuo4jwRrOK=kIm9QN4XYhGlf<lhL!DZq}^QQ;tw~|
z|LCic9rklF@2g+wd_fyy_zPVxj=q56D!q<U#?wzydc8mx*c&-eZ3&<ScbeTSD5FwK
z6Uiy^R*a!NME6XtGw)^FGaUp!{v9Ex){*;7Ry|h1!P~og7mgJRp={5b<+1hGNhMIs
zQE4;3*Y+JS$1j9;r1K5#v+BaC8me&pYS7&+FK-F*IrtN(nrC@+BC6R6o%OhizL7oS
z!W+bTv#F$)iJ(WE0lxD|9Jq(=9jcmiu|-|LBaz6Mj)~=_$#iYbq)FM0XyNT_=>`nE
zCIsI9Lstr2Dr%g}FU7?5BdQDrFWAy{S@TkHBr4Y?2;pFL^!81bl{GzNyGO$mqkQ+7
zr9J-@Ni^u+%c<HvX|M2;LEBADdKeEJW5G5lzU22)!KY$Ck+b@cTOXa+VEl#gC7|-d
z+s{r*zHZVAM9PMf?|`nEggLU%w6KC8o;^wA@tn?XyR+Tg-J|;~zO8NIZj7O9JC^Z5
zKue0kE9#Tovms0v=?uW4FQY~MmD=)TE(_YXyQPjFA56%SgdH*YMds5gf{y1@lQYpu
z3}I1f&_6&{>5oxMJ3szf$>BBKiPX}9)0n%Ws($1f_ALdR=t-~>j%FsQYztAofg&Ob
z*t60@+b}EM2X!^A1x=dUyE(g$_@NmpNoy-~jIpyX&cR~()8*8FmvL8R_E(IR+0593
zCQsnvpU{>GfUbXVvEPY+N9mpxD+V)O+tljU1D2m89uAg$n9WWtbp3YmPwn`r0VFvO
zsrXk*ah4)ed+T@a3dzF6jv|t*RL)g}J0au#{uY@FHDKjtnV`rt)|)DvenGxk1)s5v
zsFfcjrlw3w3?pKUcD876EbbUn_|S>u({cj4CHHD%80V%yNO)qGkZkR&l*o(m(<lg#
zsgqixeIk_rR?>mr!&m6qW1c#BWR{A~ZIlY}oOq9J(YlLl<`W(@3wUsF+Z{@cyV`3f
zAm|8%#-;I_`dcux*(B+~7+eXA0(+zU>`ZhjRR}4Bubo@t=sLZ*0m+Q5yTx@UTaSLE
zT~hs_v){{ene&2kbKCV;fw8u2*C+sOp^h)<+28DUu}>T~P?u)(YAIy;tYz;ClnO&6
zB?U;d?MQdWvrNXC_&w(_Q%Su>mSAfubY^DePm^cD-}k^?YME&BIkj+~LP84q$1!c6
z6ufy7J=ZsF{z{>lQTj|N6{pXU%FL>*S_pM#YCm{|domMXria;>jKQI8f%#(vt#ZEK
zXx&jB>6~J$MyE7Cp?CIZAyk~fcg`Oycv=DmdLTlO((lHD6HyHflNAyW^HQRT#1fp8
zY-8cfhjpm86g`{ylVdI76?u(^FMyS$+&><F<T7>CrHUrYR9DoAOG_m|Cpa%*618ob
z_n?1c48IVQO)wJeR99dIrtlN{utwrJwfJ9XGZ@1NqsCzE2(ToB@2hc3_dPJBsxZW~
zv~9LTt)t&`19T|$=T_u$ueAc~!|#@07qn(Iq6iunjmjqIzz_X-GYnNskvEYCCv`DK
z0#es@tNmdFEv$+Q4ig{qV_k;)0lKH3QUk3<6od;GjAe-d7+nbn=Az4PSWDX}&R~VA
zApbi89y)aq-5qZ2uY0p;OghVPg;POqm?IFCw4CIt>5tz7Z1(SM>kZk}H}A5oG1f(d
zq9k5Zvl&;8>%-7{6dhj)ZJ{oHm4IC3^6@r?yj!*+i#dFr`q)OpW)_CAAL_1{<__?c
zVAR421Hj3Du9p-}#34%vDKhwQ1d_yK+d9G*^E~QcU{{>S%g0<S+WH8T&(GHXg}jD}
zijjLL#17)?7llFYgjEZRgf0k3KHwFSoI5j$7BJ5)Ci5#mB-B8**i2qwske}<kC-<+
zsf@M&Ef5u)PDPwH7CwiezO?j|UE(S1W%mML0lCV8>w54{jX*61>yUg)d)Ff_75}y(
zQGB`~af@q}gn-jG#WN>=wVo=u@MbbNp2!HH=KAX3$taqx0dq1}w0PG=TQNTM%gd8i
z&a`oQp6^kA6^c{)T$Y`fgqDAa8HzA6m`e~+ub2F0;S8>3a{kWl(%p_EmJx*R1D%}Q
zC#k>Fd4jl>m(Zlrmm8Ln_q5812)4`g6O?-V;^}uX+5V51Lur?`|BII0`?(iG71<JK
z^sKIKm>Yj82ftYSkg3Zf6sEAmih)a*bT1UXf(aGMyY=)G3i0~Gh8sclv|PJ$ep*OU
zpRcH(SPD3_EJLVFR1tF{+NNX;z&(=B?7x7{re=x4cRiqRsHB*^gs}>RY8_E@rxrQ^
zf1kUk27{{?QImQ|2q-<nlK)<^puEpLs)C|=WrR=SbNYzW6?hR9KLT9(;h_2)OqI+^
zzPA?I@!UFzux;Ar*wVxBV&yD(Vs?1Fx2Uu;-(IdQ-o$V*Rmn6*O#q~EYcAYQ;C|fI
z$v1Z0c{?X(xR7g&I#wz_qc8rqi?l~}Hp6C^#bCGuH_gOdOdk>^XEtoRSLan5dD@bt
zwQhFqX;sHP+nk-;kL;Nd^hEP!fiIEO!_!t<B1fhoJ~<|p<8$-48_yG8s%k-a<E*8_
z0PX8;<tca4bS5MdfH+S>oU7f-jZLcWOocN?yBOKzu?FwgB~T5DhG{afVNu?nbPKMV
zNf_C_l5`!das6q*7xeJtwrJN&WasD?)$b81ev7At<lk)i+@s@cM+%y2KyH}bBMTUw
zp20IpY^fO<>t%dMz335zrXvgHo_!N=AGhO=lL{Gk-5^RpC)y2(>Z!=>GZ;6>!`D$8
zy^WO^$gF9|G!7(^iG351NEmrzh)MjMg<4Fw7~E99eZIRCocKDzmOtw9!2D1=3g0+3
z!3fI#aka9(R6E?^o`Naz3Mq}%KJJPeefIo%wbPIRJfCe(3U0FJgMvQ$qW#fTVP>Yu
z*@wi*CcrG$Cc^F>4F>G(yLQOZquV<=)awadVQ9h9GOCQcbClbRwm?}4hY*(*YVeTz
zPF$VcTa1iU;L=|^+VnpgnbUY6r4g5&<YP>b=i)q=xJMt|u;^@}Wx@3Q<XF!7m-UR^
zhB>G5Lw-~&ce$6Lt+SQx^CQB*@+HaOGfn`=E=|lREy)N{;l9Y^uA;~iuA0#YYYw*I
zRlSs6-R|`_`n-HThFo8bdrD%NxNg51f@-~t-mI05uNaZz%lMcTeKCD9%d_1Rt0nH`
z9fZB0;{(YAgFDAl<SR<PFYWG@V<k>rg=%3$P+hX6<Bo6OM$TC}{zlJIRm*iU)c}SW
zj(u`nX4(YfH;Pohry74hb_rrQCm&*~U4Pr3H@ButP*J_S*)oyi6drY4Q4%jIS#CF6
za`XIO>l7G4<FX_n$vOGO!&^vO2;oBcV<{(gA&kFVhE%}i?K=4NiQ=9a5@0?LQjR%>
zz#HGk;P(F8V|9H|qi<GN^5R1)0fdjTJtS%DMa))F8IQ;d?Co>o{{rV^iwI<lYjH}2
zG!ystENzV&ZBfNk?o-UMO{$Iam$Xb~UTSZf+TYmRpH0xBKsiomXENF~$r@_P&x0bd
zOR4j?zHcL=g2oy~p@X2ezo%^4$u#{6a_+?PHr}nTxa?fAg`kjlCV;!Da;vujCL(cV
z)gV8rQvP1<`HKTZ4M$g`>ZwUyR9^J(%qpH(VO0nUip&pwY}%u%i6?x!v3SG%twg45
zN-0&z4kGqXy3@jR!t1oXy#(i$KDxL_>PvBqc9c4suXpsx5&sy-mTf^MHs-cH<lN91
zJv-YZkce%4udwI94<KA5ld2OgS=nGH$JTY!0rrLp@MZ>%Mn@|cLa>j*a!X`kr<v{T
zdNr%EutywKJ>RZ`82?|_1nqiwk0()e)KYuX(p0x)YFj_6u}jkN$PotGFYT_CgvBu%
z!BQxCBQoE!qg8Bht@Xk7`#)b5e)JjWL!VyrguAiwF#cE$5dAFNZ_jpk26K*+NGPpY
z&0ET^J$vmmYMCCLN8Wf9>^V?>>R~=-_1eM4E2D$?RB%3XwofnhBh)M4K*Gu@fZ8;3
zgJS~o*xB33o3@ZHjd-LxN2k--O--)B+H7xY^2BqHB*ZU!dF6eHgyP@T9U4n?u;O(>
zUhpzsswol}py{xP$tC1m6v)v#I!)T}RQ6jh?0kPN235%zzZxABeh;RoqFiek{N`pI
zV`R=W>K;$Y%N}!x<b27S)31mxVZD}X8$?D0+8C)8&Rj^R`&LLVl?auD9IVdJOUi82
z7}nX&ThB$7PHfDx)EDM)#7_lghf`EjosXnf@B+=JUTMJx^FQgqOtP)~pkAs-C*z(2
zsq7Ig8>DDIM`B4lv}LLyf(NZs5Bt6oB@&7jy0Y;6KEW*sCGD0>+@M|d2WO~Cs!@7Y
zdZD6&)>*Q9T7x+TPp{JW`jnn!Y4cxF=pM!BT%MCtn*iOYoCxA3)k69Pt&fdavwgce
zz$?wB>g|~uqM0Xlj`Nz}YvR`_l(^~BhFFK_sKpx+^^XJYw3lXpR)kU#N#tgN&|e=)
zci!F&>|X1Xfe+CF`qIrRGt+j-kS-2=DML~yIisjUhiwxz@L;&r><xFMYB?oo=f5@>
zr&V^wrTsC%+T!ICst`&CLKPnWvP1{UG_q-7M!hxKo>)nUI7-8F`diMXb9KMLxS$eJ
zWRgb;A|U}Ty%o@Q@z67uk0G9=!plhIPOK?fdMO`Dp-`Vfa08X(y8kY3I(Dtgq;sMM
z_y?;+AYx}EO++%Dy-R0Qe8yr&H<Svp+N+$xuVQ$1T`(4CEJei&T4jnw0fBKZ36<J9
z;fB=^w4e;GN>|bIs4FHpR_oQnYUkSLSI#vQHO>6DBnv@Lw#YuLg{td2F75Tx^Zwc@
zL-|Ubv*>@~J^xHa`165yuNQOe=<8H5$Y+E)a(mlz&yKc!-Ki<A56wbBM|`YfHZ&J^
zHN8pCQk)MWAx?e)GYf75w#Rl|wPkVo@7W_$13PhsRm_LAHqCN}sWw@Z85y5ji4Iuh
zS*GPS^epmsuDY^5#rVuou;8<r3SC79|1$X}<xUvE*?e17@ZBm}Zduv6v2SNFL2v4k
z{OgU;3|A4+lu`zj5t1HN25C@UhC#ILu`jH9JGZyr->C{XG6e<72;-UxY+dO6Qbxd~
zY+vDM)O);1H`R3dd>+M)6pL#bTa3}KqQw!<_IimsFHmEQg4M_2<>pyU`8zwS6K2@v
z*boVVRgyEBs31KnF*Nf*u{Ftq%1t|3c2D}g8b<qRVsVT|FuCh`Jl7==@Y^SnN-4KJ
zF|=c=-Mnwv0FjFiz4tbWjem%=Erz3>{IqocC~p4vUfpY~;hxd+tQhx>g+ls5zX5YS
z(l;A3Ej{{Mf3)D>@BU(6=z?9Xl-J;%tv%+q>7a*WQzB}&<VBo;{=p@gm4*x0_AXfn
z77r71lz_8xPdFJ26v?$Y8>Q7JV#w&FK1Ghpv~4C}-NnN`C)sehwA9Pef_jQpK`}>%
z-i^Hd%7YST+wTUK<Z;h!4S2PNh$rR_XqeiB1km(-k9=2oCgv{col$bxZP$JrR#2V2
zbV4pwLsH`F`)G{v6DM3>+bw^U`cKS_nukVm$1t@j6wBAezwIQ4N^YfdKvcakm&j!^
zQ6>VAWlm!(%(*~RfDrIwKv4%;g%iy#on%&>@IQR;dlQjgAne9&;Bqdc#hjicrr3xO
zHgL^k|6v09ql6(ViebX*K(3w3ou5SJRQI69_^M^t4H6@h6}n=BA)dH5*KA5{;3QF7
zznCt9Q#(uQ_ut>=aP(+NRg<g0w{*c;;0C&X;my<2HM)ObmRYX=!x_X`2QFyrV0u3?
z7U<ZUEjXCoTbPKc|H>=j%+$q;K2pw30g0N_zjra3T7t)6zN74_cYGhw;d75!oRe0=
zNX5rx`Z{VLhioEWW?8M)6V-)5*XHX=ww1cT-`|g%)7KYU+iXz}uA?#K6|kSc-ngf8
zpnJ@v*{C>=aP3kmMtLw5=ND3+FADK`ciI23M_GRiB<nTT6V~`b-#bl_gzL{S*?4G}
zdu^>81$@UNRoY8{2P1S=KngV?ur>wE-<JE$7;mu6yngO!)o1I|^^S(@4g8)>D?DvI
zp=Dbm7grU6vqnSX+|qUPas7bA3x;(U<4f+%qOP!x@q%z!&R5rmT6~yqqP9FY4uH3A
zwz2>A#l5KKhmPz!VwYY$<b7QPm;V*_-Ua9S+UR-uC}wpwx?cnW!3<N;C9X5O?+TvF
zS%%o~e#7>U+3g%qHxB^M4zX_Z1i6VK<Ew-zJUTd2(^M~KElr!V!2ve@D;?V+zr<Q~
zZkna_ppn)l5FT&H&K)OM-%PO>KFj>x%cjZLU@Vj48|31KK{c^MD_3{+!jr1;6>AdE
zn5{z$i*qa(2Hv<Hskb;UYsij_7(JQ@`5@!V;19X0Kj*~cQx-F(ygAsq&794Ocs*IM
zS&3rMZc@)$WlQS<-4e2$!i%afCe|T){_NO;jOXiL)2toEC~HoA2645P4wCXzDD)q*
z;B+RF`lh2zLb+M!MrDeNvih@|_Bexd>PU${#QM&)+`jsr(RUA&1dy-ceJ<_RcZnDz
zqw7XC?v3Mbmz-S+g=hn@hwwCmv*)=E_l$-(Ye#oynhx;u$yraPwrqF}H8R!8;b#A2
z+PX+KMOQ$K1c5wLV?ed2YwN2~hH{Xz8cp@_yO9TLvs+Pdx!?82cE=X?_9j~ls?=ec
z9pDe$4=498{Ct41s)YBAAM3O1m1x(Ypx?26^ffgd>~<V@hF7qDwCZUBL=GO|i&C*D
zt0~hK1~7SJw|wCJL<DF3&D_a$Y}Btk0hrh5%Do1GOeMlNt>g9`$Og-$iFihHX0Oi`
z7dU{3kHKO{o0G4<UY11^`Bd5pe$rLxOMPvM+M&Q(j2|gJe*9RZ!AwNL{;a}7`|*xA
z?-|aFNuP77Lgw(lr#KeR>cLa@Z=$i{APHMtKt!jplJU(PCg7p+R8{H(-U?l@(R-@e
zKeIQFy{S{$Og5i$kl}n_p4oPRZ%g7Ze0cq)(3XxoxXv@!Jyl-x2(Q82WR!J90}G=O
zr&uy015efOWR&t+A#aag^-hT8eh$7v<O+;reSaZ@g%kMS;5}BowAS+{Nn||B1qgXX
z)y)wpx2%*u*2`<}4-vg3Z6S&T9}{Ss*lp^QWB%UP_6>v}j(82S#t!D_w0$Of`NDsm
z*G@wqWzcQB<$vXi)7LKvB{YV=Y{^;a)2b#?Pob}f+${_^O9fK?f7`vXcFiRfq-8d@
z^s|&=wN{F7Z$=Y#{15U0j@rGld(`y{)`182LBF)xum6(T^!P))!O<pw-=rLsk}>uj
zAdcUH&#EYN<{W<?DN7_+6IcJ#5x^HF?jUU{;2^Kc&6vc+!oCHI`p<{Hsb024&6N!3
zY#*x@Ox2S5%orUc-wCL8co0i$(DzMMFzwQ~C0hG+Qt@;7WB$cfz*dq!_=X44CHTag
z<lKmNiz%cl*GX^i9-60uZ!*^o`3ov8Q7R7~V&^fzhsBO>{KqW<%<g+V6Rs|`8(tgP
z#X$UL3A)_aWyon|`AGD4*BL<#)Akec9z}RCNwU+m+416ngpY8p_9m}Tr(V-@>EL(z
z)c#sdFq3l$vkYv{=*KRXSbvSK)XgbhIz4wSbn94iN%d}Y<BUe;H10QbW&VjyOW7&<
zi{~ijn6xZo_}Uc$V$Sl~S>9+wu@(}YftrW3vtT}QY<fg|TtdKl9CNo{9M-tbK!Q$V
z{3-%F4rL?%)?@wC+2Zs;tB5nPM`?3T>RAfBJe$}~Ks5i@c7JEoOP!?RJ4HlM?TLR=
z(^tn0_%$fEC<N;w@L~Up!IU}&>?5jO!`xMV_&P&qo5A+mGrc3;E8^gw6A)bw>h+s5
zP#59O+MXDvtgVnuc(tC{VNep(Nao=~DsL2WLp;L?miGOu0?0=Tw0eAnv_-hpox`@e
zicmiQ_6be?w>0;bJ?=A7#w7A(@OuOHI$~}lDo5~mVSgGcsOed0T0@n#!-Y#-L_z#`
z@&z(E`{0V%r>XaCm~08zaAH_7c<l$2r|zfn#IO2uwWq#K{#qE@Ik~xtUs<3%am(x)
z-(58e-KbtDZv2)^Iga{4+Mk(;x%eZ1wfm170MV{~K?U9DQ_VK9AoAi(>V{7pQV8p?
z=F_t0vEmn^mi8EnDyTgc?2WrPP_OHMekk5>sXDFne1?Pe?_a0y=RDl)g<;Poy30SV
z9heL<({u35&ji`eUDY$@8$0KRshg0hheuau*}qJ5ZJHeA;*&Y+^GlU_xCdUm<MI6#
zfURZ4x3DQW4?pk-gQQwUNCjbTwr$R=tT?@+iq`5>Av{bLH(lJ}m(b26sYERXBW>)W
z|12dZJxjT<!4VpQMG2HUFYu~3JibhU%*t|m`?+l0&Z4H0-dm$qiqJeAHxjbK`C|I|
zh1lgv)l{e;V&YYLZ6o_b#jQY~9Rb`aHPF$A<%L~u<U^COkw!tUzD5~Vhd$CpW@zwT
z{edStD{0|4d#w{enOvXIDUO@-=2n95B^$dpjtsYakIy>m5v)BPN6g;V%x-8$QxM$u
zHvvzZi!e}M|26R|VQc+@YTPkbi&9)aCqm->Gki~!hqeLP)*63nXT#z)1XyRp5myma
z(gUyw@z?Xm0;#wvqSLf<PetwA*Vu){gnSImkrH1MV1C(}aemoJnms9hskO`w6_YC?
zH6}I+m#!xW<><WOCjM>(as|-w*o(|bmqS|J=WyLt!s3pm4V4nej~bPm2s)i(hK`vF
z67!y`{Gv`mIW*r|sAK}O?N^SKlE|3zF=EfO;<Pg?G9f!X=WrC$XQ=|TU5A6puVIJb
z1X3ZiUxR{BR5#P;Am9eH)ym6k^1QS#j+nfpOK5A1lg&xnCB(R5s&nqwbf%o|)x@--
zh<cT^gvnvyx0|i`>71+jH*kBckdX;?E)8y7d+TCxBGpUoJMgXIR279VJ|A?IH#0?4
zh@nC4e|r)Iy$2I$n@U%8S784h)74CmsVeK*yRcuH(oUFjD`0ZC7t^G<9K5X85Z)zR
zot~a*tGXTWcRux$HOV>q10Rw$MpWI_Z#lzqO!!#zRBwnCD{fG|N^*?(Kxpe0GQnGS
zktDn*ldc%RgUN}AKYQm|N4Hdab3<<SbSc%6*;+L%%^HpF&^AF2R~So27zmVx?YLd3
zL=A2@&3{bvDt1)Ka96JXtO)PD<V5;UKe$p?pgW=r3@>kBtJov6>Y6S=zy3CiM18&t
z^)>UZ3ALlXbb<86SIxdRV6E-dQA&-^vBNxtPgVkDWI~M6TgIDH$0Xw=;PU*i5}8nU
z-vJ5R#cz>Ylj-*)lpaIl(q*93?-$RO!b+f4(jQ6<fb;cZ(F^^^m$1iwvyDI<Rh@1L
zkEXL($o{CMbdP^Wuimrq7ua!OM;MF^)!Yy4lE>CzJLTt&`?yFzEWEFf@Y(yTnp5CX
zYBex5=`utyhx$BL5%!i>^xcsQ9}v6S^=myQ4S2*&d2035P&*4b+uW0J71zxxkYaAD
z-35=U@a#Ac%3jx>nrxTd`v&zrk0pSA1huRL8j8YG(YiQJ=(*UtDQqr(-2<aM8e{K(
z2zp2soypDF^_f$$!|=u-bZiP=GU*&({f$FZ+wEzci#mPp4O&~DwwRF!WKhJ#V++<~
z8)@9u%!9o!`OGYd_s30>&%w%tP>j*h+8uSR>vQWdb7|!(h6%XXPIg|tnux^prxm02
zZ|8ndZ_&i=gCwht_rFDd;Hx1Cr6P;*_io(Zc(ka0wLZt_&2kamYmPNqqCmP0yt;D>
zf!B|t%Mgnf(Tz9#V#IAL@wI<l;PnK2@^VvbaDO}`ds&g+58JeR>Km)Yxn<60n3WET
z(Jx@*Nqym{e{J|Ai*tCHTD1n%*_k@_pJj@qo>`6-{YY4UYlVX4w$E*D_=)Y&jexze
zV0L)dD<<N)De=JgQG3bm%qsAAe}>zx8H>I+V6Q;JJ?iUlyr(l%o>slQ_+Mv~^ajSx
zmE!tDD%&D6w)C5Z@=}2NpAaU-SB%P~wh7kPox_afd0|jlAxTEE1xAJdQTtwR6Wr&i
zl{6n!LfwFqE8<W;qpxN+9QmmsF8ShpALjF4`4Q0-M4!Z(IFPdN17PX$S7NWFII;`v
zip!>9Dy)Tgy(k-{6{jjjuAE;^UtU&jX|T|>^t+r$!=G>XM)tjWjY+J4s+3H0tiuJW
zS_b*4fjO+dSy1(`#aluW6P1DdYNNw<x#?1xkGsbZyXxD>W^QcGXj`IiucJpC1_d!L
z?gaTqC$ZH@Nj)RL;yWl2T%DxpBZ@+bWD{=G<5#S}!3R6*t7Uv^7y5n9Y-7kRL{+@L
zkAEVKaBt0}D8-3mbg&cO)VmSHX)_U#kYlD7{u}hLMpSB(uBIoBe9vx}n@ZTz)j@6U
zM=o$zmVM`RUok(96dx9ui?<`)45;~Z$@wVG45v+3w*h!n-c{|0QMzl0EbdSBiYN$<
zqM>?yq-<H^(@s9Zn!yn~{_;Z@Sk}TsoLrl(f3B8#+tW)Mx7(*TH!I6<%R>Jn?<p%;
zoQrfdD|)_=UrPZIHngj2#~D&x_UJQ$&|XI7)ucaP7Z1ZEB`D+n3Jel@DzGQX>d-P!
zfV(Sz7Q|b{MAx>w)hoMkDb`Ktb@)WUElh)tgEE*)-jfYah)YO{MPF7qQm+bkP=S<6
zb!PW4X0O<&+PwU)goN-&&@uBzD-Q}!Z8wyadWBVjc_E5r6qq{Z_4d`IrP@pPakNC$
zPR_>4OFxq~CF)+7bHe2~DPhV=X>$B0Gi3mb`Nrt!no50nM_>I&4U3UW)v~ReO-|Z>
zC^>)7wHf=QcE7c1Dq(jum=m2(U%atBW?{>gVG<eiwRvsD;Hk;oXgs*72okRC9`0rs
zwXmJp8347^D0y^(cilH(|Ig}J^;AyvY&kWE{*{r&((`Dqzu(T%&F5M)1H1b>fNzCG
z=Vz<jP-q!J?S2*F^ALQAGZrn|7C|^StkO*}aYD_jmy7;rcVH1V)+|F*5NMLK$?HBl
zH4n`D7Kg;zookaXj(z1(r)Wz?%M<js^!XuQfEqm#Cd;F;KP88{F8W*l^LjFC$?5PR
z30w3D)q#qhf0-q>G(+C8-PUsfuFdMxXD+xl=hLEr4<%C~A-=Xgr^4Ns3+prCjD|Z8
zsXmGP&aFozt2w@I+%^<r?Z)S)iOu)ya!s3csuAx2=w-jBw#XsJXv6B%(u%QAS?5^@
zA9t~iYZyZ32IYnI^~DRdYFO{{<rR?7e)8Ms*Qc&@u(Vn>S-JDOdJ{l|3HHalQGWUX
z*DSXv>A>QkY>GMg5B=ct(SrjHW+U&uHR0#XDg?53A*jjM)h65>mqzfa*$-d^nk}2{
zat_wma&iUXf7Vx<3$7deVQYFO?63jnX;-?<Vy|VNEz7u9$*OMbIY?C3_UB=zFSpAd
z6iz%vw>v;G<xB5@Q9yfrjPvwlY|GdqOhR1TWpP`A4#9tQeTRg*qj6%-O05b}KrUR1
zeuQ(+3BDQIvXr@GXjLq?m|&z>)KJAjNBFKIKXg`ZTR&gSY_IA2RZ7tj)Zh=8cR@G)
zTTK<2w64u`UK)!jrLL8ZFG68dtUFp!@zwRgvl|zJ^=MTKz>Zt{XGI+uJ%539($Ew9
zh<Y8bw5XC58jmOwN8&!Db(jj<*<ijQ=u*eUCW1M*7%_qRMoCQiYS*#6JuO{(gR~;B
zEP1Ji(nBb;N=)*bYVV}DhJ8>!tpu%JC6<Y2H|Q8=RD?Z-`Fvr@#$eB^2)X$tFPnev
z{v+50xaj~-@i(IN^Vw4gJKVX_-#-bv>rHn&-^BIz)Z+6^^dqxAS}AS|@5UUSh_`=W
z`X9xbnA#hj*1+224*&ONo2Crq3TR*d9>)~1a%+rhI_bWs9o=dUS0G@M8N?6srUX5D
z>2o&SYdUthY0MjnI4vgT|LH+4YfYE!)ROoBMgV*ieX3=D{>X{guUA(bGPrjV8BWZp
zu44tUvMDZxRX7hvl#o;GRKaTZ+-F5X)(5n8<@c$fqnHL30pFvAd-xpcmSxD*5AC|@
zsfOoq&uC-jE8R}tO4^)T_yDcjDO`iX6Qgq^HE8M}@|7ytL!%LzQraL%+kMF1j1wIM
zLjT?`!}}1`<Ftns_a$Ysbmu@Ie^vMqZL{CH92a}i@W0w}*QbW{aE}m<PsRRA=s1NP
zyD`RBXPg@$xvr!DmhzIg3v)G5!n-UO8Fn1}Ufr8%k~`Sz*lQZwsXR6Dgo@{}{r=<P
z5U^8ypwFUEn9OvwQRG3PUjPL4nX?Y`3PneJSalofh`n5aJ2xD($kJ7usAihCF=Cn~
zjlt;)F~9RJ9^YEs(WZ|C#@Ky)oqEGEu}HUsLfnv_!)~!V_9H5tAQ&!JFrB_`lGUaM
z3TN=5$A~d4w6Yg#zfSAAgs;dN#phZWy_Rcg+IE$wU&-O@=B2(vNl!~ZCISnDS9B>y
z6p^#3c`yU)|5*qQXC!QmEnZ+o-d4uIh_#LHred6MuIj*|AH)y}wV@r+I*Pk>rj6wW
z34Qe9SL~`n*~Fiwa9LlEX%EPW)2E+iIr3?hT=zrCHibVsg06l%qRW|kHy(fS^_zDQ
zBKJ$j6%{@H{73Xb>tv`olL3BgwhC6cnok-TS@<I+t~Zgq#8EvY_~#^hCT9Q7Sa8q3
ze^zG~``DUOUx;-X4LD8vh?wY-7GIxjna%Epn<Z~N?%c)>4*bU&Fi~1vaLgt7wcKr3
z;3e#=5<L_24>s8gIa7w+PJA1z{j~PKoa(6)BnmYbxA*(^-~GK^IzW&8T@a<Wfj#r;
zWY&-E*zur^eXO$^N;Q|=P_;1p)%BCR($o3;^<3WyjxMmXKND*3>hKY@`qGu7%Iad%
zi0(Drocd5(0aY?I%s=D)TXaO@B`2$!{yaC@?%+(b`jWF@&_2$_?Q0z@f@8g|d@B>>
zEuh-TZzDN@A}-+u0Nvw|yEpJ?B}^say}Of7CEke*zN3OCqy)$sC~k%Gew&C4q}VQ1
z^QMiOd1D|&+f^lu6(<ixMfeP6vfQTyVHNSKKK^RK5L@Wjwoo+}Ki3^mxsZ#aQQO-l
z_NNQ>SDXK7h#U{{`UqTu@xh%yg>sRTZ@@ZqosX(;OeTN_UxVMTxgO$RJJ3+2%aLN-
zz<3Q*6m;<N!~1?b7Yd5iORrcj2|#DGsR{aaEcKQKbS`uLVMkQb)YI|3qp6(3Vv$N6
zs_e6|^rmXhd|$%?UpIECXvm)SX#-=QGee5ya%OiyX_S7Mo$!xp<ETnhQmlUd@kcM>
z^P0v;>wpY<X4UCm3{CF@PPFKzCbYUITeiUy>hFM}X?pTCvZkh{C_`bMxc$PZxPN%d
z)Dm(18b0aSAE|nYsrT;a7V*5EYp&vOl^HkagD!f;&i)nA2J~wJ)5-9M)!f9g<1Z7S
z$+9<h{%!_$#2>6qs}Jn%Y%c5;w6`r3OkI-#Qqi+V|D2ZKfm%_aBDUVerbx7Ss@4(V
zUdw-kJ$$<lE{(P5BEpoe<VHpu1CgDF`}3FT>vs<hh6jdz(R7^|y?i3mOWggb)bC&E
z_RuR})Ni{^ODNB=aNXf+{U?9r1!n$CQ->m(B_)o$xbP3sDp4{&$#;M{q;^>@zsp$@
zNGc%rBj}+~d`FIh8dyf8%;_XPqM%epz11_pZ~5lX+dlnb(yI8BQr(Fd*Y%Z<8l>8Q
z5(eAJBKY5|J@v2!-q+|ZqBUL?!ul#EKPQD7URTpknx#GyQWWxOD1Vu=+`8c=;O+mt
zbNG4a55D>fCGoq439(ba-d@!BD_0>PMQf){d+u8NV_>PUI>JPS=o=qrIN1Rl?5}2%
zgwWrP9{mQAi)}rs^$U`hRvV|k-S7Y}A}^-q-F$65l#$LYVSIsKzA;t8;y(TolG9JN
z(!O~w7c5U18PV%BJv-cjPBaM^7C9*6rEl(jIsS8u6}xR3hR=GNhb`q3YzzR1wVk{^
z$rqJHZyDv}p=aVsDlI^!Q1hpdx%Cl_n{q>6^r!opN4ELWT?Abu1hrGo1=?Zr!Hh1D
zt*DPOgks(UKWjUuV+I{&#TunF)Msr-ED^gbt%%4(aY4+@y5*K<$p|GaHOPapy>M=H
zTRzw+;3mhL?8z~s{r#o`U;$CB={PZ}+3~l|y}Bm=i%SDfHL`NKy`!<k+0AY~keF9L
z;$$l^)FNSbEjN*SW1eD<u8tS%N`)UybP9rKcOxl*EpFA^%%j~{Kq1oBv{x8<i=ERv
zsNH`?>xJ%JEzj_Q+1Ixhf;}fsPLpP$0n@p$g@wM3zrMH$cs@Xn{^MxEuM=t?n45z8
z6F-r@r&a%VP<iKI`-o=Fn9KY38COmrp2?G>Pc$`dMt*I)lI+D{cpui^=4eB?28?pC
zuM>z{<J;yVjeU!=5yWUfzhGBE<JtMkq+R&LP(MTeMMQ^(Cn}pvswq2<5&h}9-UI!P
zP2w5v6ut*KS^yAwtwEt_d%H8?EkP89DM-OGuqEdA>8XNrgQ7J<kaRegaj<-_aInt>
ziZmU>#px%zDjixSi40i%S^H!85_aaD{!`6#De9shn=5H!&EE<V7c>SlRH>B>Pi1|D
z;a85rF4dbSU%^QgAlnnAq{_IA@A?&AYvYQb43t?Zz-t;al1E>+=nNlV?(Xfx@ft6?
zlhAmwbV^3jJrw+1e>wd$9Q@YdEc1%!k6ut_mnrX8(U%MU_cvxn6irQ>U$aa<lG4;K
zB{_`J_bxS%$sy`Qm*HUL67tv4>kPzP?J=QW@{XUcQV8l3@_lOj=EfvYUBhII|9Q4*
zMtmG#IM=ki3naw00}ML6*NyG^zKgCcm(IENkDgHuvCX4V*jOGC_AuU)8DCm=-7iX<
zAOB<ByQ58x9e6l6==RM|{y0XW^Osb{$Tml6VSmJEDSg-QA>wAyJ%SdmiBGO(YQsIn
zPVK}@TdY%>?W7jOKyy=&uckI3I(m1uj}0K^LQm3btF_~!FF!tfL(wOY@KOELDUnms
zDb8FtjLSkx_3ITYA9L*GUoiI*3W|fQpS6Xgkeh9PVf1#U_SHhgD+cgeWL}V!l}og=
z4!mAB9YIfe^3b!w@=8zVW<O$S`&yFt4?jO8g?Gqj=R@l(-8TX%H<E@bVAC|Y{e58n
z@4oh2C(BdeZxhfC9Q+=s+FQ`{6j=qpo<U81Ki$jXUbcA#v#Wt<-|iYU7qgFl_I7s{
z#@Oxd$nYf0#hkCBoYnok_yKO*dJ7}J@_BE*x4_==z5aS`I-KWa4Z0WFw&C*nw(4Pn
z>9E*eyUW~oZe(a^Xq2Yy>NroOAaMinRS1E7@YQPnY_!ib@?5&_uc2#OT?HFZ(ZHVh
zl=<TKrb<y*wen_-{N0da_0aeTKP6i$oK1Y?5Av=ySS@qlSkAl8yeD}Yawb3X!l5%$
z2Ya3e1zbJV*qwsfK&}$g)Qb7uNN#<Vq^ZfumH$A?fe-PX;c?0F9^yUE!^|NK@oE90
zht5;3INDO<;wy9aD%?Kmz7i?l!uRlnH!pAPa(;7mSNGbtAVJ27?i-`KGyB{z>fg<#
zV1Q+k4%Td1IEi(j#ocE5R^ZL8;CDLtOz!QiY%>Qp@+Sj(7n0GG?xkEF`62EF{Eqvz
z9Z#G+s!Hk+BpQ=vFY><DS|bbg0Hu3wjc73*$el&P$tB?tC(pA@?1jF*MgNu3_S1T<
z*hVY-k-j4vho%c`ZHDT_Yb-m8wyqr|OL~9!>pK~9e2cGQMEv=P++w9JhFW55E~)Ux
zHdjgz+$;1Lq5vZ!=7hHq-7Z&9SHy&Skn(u?dKDBS{DNv-Y$IjGFmDe4lx`?NlTf2-
zKNLb#DzlldCgc`qyKdjoytUI1?eSqn^!5GO1@>6qW=u<?3I#o%!SY*xS}rvPT>e*-
z*Za&F-gknQ^L5a`KmP_qM{OQ<|JDf`Yt&wC+`O_}8R0#E-&=K>Nfc(g@WcK8O+s~t
zsrY4<w6ffS3etD?R(GNR{WNW;CK3-`m%<HbO}Uu-RD;rM;?qPoGVs*1*Y+z|5ah2m
zP1m_(1XBLqZPBX?rNhQ>xeVExvgW47QBYZn6qK%k$VdH{U?oCVV%qKRN^0=_RP`B1
zVnyIk``0?%>n{~~;asAslG{?<>xKbp7t&PLkk!>=<@FMOg97mLyUkgpAq~xVNn<ms
z-gD6L7Vwp*3U4N7Z*OmVJGf(+dEZ2lhZm$EIuSY$r_XUATVx^mGK2c!$b(R%s{_;`
zr$htOs>hMRnf%DDCy2fmtLOa$;tTp<MxZEag%27)u2o3s2p-XXQ~wM%#By1PkA4@?
z`d1KA8P(GXR4|Iy?_Ni}70e{rs@grROo6|D0u>XR4hRl?m?+$YFb4~s5cRA=eu*(+
zZS2v~Aks^7(~RrF|7or3b|<LzZPraxIqDx+`uPul4>gUjDfA~I!8ws%|3u!T*dfX^
z(%?U8jnvku2BL}|pU5c!9b--DJLA6ne24uVR7@@a`McZ8a$J$h!TQmdxY%eeIN-N2
z5;^b^aS8+q|4Oj*wjs{TrIT0ZaaDtVt=|aLD{H9!aO~1%g%_mzrgf?aJ^^|0C7l+L
z#HsZk*C?&th@4lF*&LR+@0=y59B@@2hBn#W+U~B&=F=D--i&ELaI}Bw{eX~oKoUA`
zSPMX{YueV;M?q-aNp?<z2M<q-Ii6X$jfz~dg^Fs!%X&x**I<$SAgT}^GT)~lnrZpS
z8JpP5n9B=tFd^AJg?D!1)iRd0cl#Du$gG{uqb%3rxW27+hN23Rx3Hf-lfi2a`R}f8
z_Y5LkddN%Ka67pc?0a%3-T!9C{(3<N@OQDggZs@1R5T5bU1ENptJ0mz3zrS*8|Src
zIs<uFIeQtIUn>-!W^4+PdKx5v&k2mBd1z19xx3hV-?MT(pC0|IWz0@O)H^VS=EYCV
zac`V#>SJxLMiHZb(c9B!q;?GjY;vo2Ptk2}9$y^A(K_@q^@l(b_>5Z5W(+X)U%`(z
zJk4%Zt8rOQR!CE;9^HrCIt4=vI8;>sE*haID4Ib^l2If%3xH}HtzSQPePo3UfK3Of
z^CUWDIYq=q$K&EJZtX=)8n7B87JFe9;)!j+o(F3Yr0l!wBVv(PPm5ro(5_CmGkO%!
zcIY7v*WI@YI=1i(EQ)RBMt9=Og{{2{m!obemfXsyZsj}tjr&*NRjs&*+D+enUt?Y|
zGd?jY7=7ic!vpp4N8|@z!>7s(pg&FU;<Dk^SGI+Rhc3F){sob65`Qg876?<8OWSeS
zhTT7aua!K^PFBJ1f|#x?Z_Ul~jyw_|n-ko{CNq`CJfb?HTD9O50Azx7B3WNddkH_g
z6B$c$)2!T{ngpf_m?zZ*si>*|)X|l^@JEQW=OAA9+#5ql=3pfyf2^?3wPx<Er3VhN
z$7ahr+|HhAu35XjFdk+25gP2-`R=~(sIPtp7kB?Yz+War>4<3Jncdc@sSco(o5zVA
z`%~>Xc`NgKB<y2O+gIL`fc3`vKpM(n2yDuea0YajB&Z{3piy$#G+9gex%t3WqPfbO
zjkLYbFWbgF&gNV}#wASlOKJz^^y#NY*|a>?$y<Nw_U{5l{ki9-I9Gd^Nw1xZeN2`+
zjbzXzCD`b|GoRr1Z2cH&g5e~^JLk4e5-Fr`{)`4pZTl~1WDs?9fwSINZ}W_j(b4(7
zrtFZ(*yh`EKBl+MO!$u0`O_0lGWQ}|QqDuK9zQmo&Y#$+k>rlJ^t07M&$XT9hSP91
z&{X~Zar+6-sT65Dz5;r<(USQ*ik_6$*X;%%y-LnVrZH+SLu39Gv?p+3Ex5|f5(ddQ
z2`BtM*<9N9Efmzs3c`-MC`#VBd<&NUva%jl2YL}bqLc2G{m2&WoRy(_K_8-aS8?#l
z4obl9SkGH%pM&?OYsGxPPs|C`0-Vuh_3IQNu_NKJ(a~{nX&mm-Z)3!;ybESc(7Dh|
zg?cUL>=^GUfU{j|1YZBVd014P{y%NFFNjhaFq)-o@U)=UDauJH#0yLKYnyuuYrgVl
zwzw=ZFhdpsdf{HhUW{xzzsL-2Rz4_fI>-Aiaiu{6WRaDugIyGx7!OgVZ^i^tlWYLQ
zZ|5loaA(fgW*CjMD>KyX^ZoNG$E9{|{Q}4Y<qF_jmm_+bAHFKGKg2(61k%AsyB6OK
zKgm-w{`9%~If>3D|M1HC^|XvnPxO1neaPA9?|lZ(WiVIy>$gz8Cm)^*TnuaTXzDB8
zuG~y$8SmCk3zTb@suQaNL!TX1$O!51kUuGX<{u5Z%n;TXguy*0eaPL<n{j&0^c{_g
zZQBGYd6X_W6wP5SK0e8e*7!X>vz@*<M*kZ(dhYg-jZ&8;ZTKNWLpdgAr2wrC1WAv%
z#M0d5xiT2zZkP27Y3(<0i$aKrMU8MlID@Yj)TDLpMpNOWVw}PSF>&^lXI^~p6jo)w
zq8Lv`qLjt06Run0>3n*(@xm*JOERG?ude|16OD`92GMS~k>T>4J2+U=!at(O*9{<w
zs@JXj4>9gBCh70@IxZYGKOON<NkOChEI;+h128QNG%ay($jUgRi8f~_AQ5r>Ot_dR
zWWcI??2su&ZX_?|wA5i~%#Xw1W4S`!E?=&NT`l{4AIF;+)M&I#ER{hRoBGLbN`!Mo
z*Jf|Dlfncr`*licw?-Q)abj3LFvz%<_2FI)on+Sb0dv}~PblE#bl)SAs8=#Zs?CA5
zGb60%gH^*%X=A5<AX^%_WeT=gvvmI{*e<9`z4fnvxm87v9Vs%<S};W_Ltan{EXrfx
zRdo>Nl6v0DRK@t^lj&=K!+Y`XtWn^}1p}6p)H}RYmqyz>7q+=;C?Z-I>^gQn)obRv
zcs4>Aijh%(i;CqqJWK*_e(e*ghSXZ(Dsc;#a&+-kWE6pr*cU+5?oI>;E`Dyx8GDpD
z9<Y4@hoR|zv~xLJuJrQZbrk_ZNK!ft{`~3Q<mTq$EEmuDr%w~u++6rKz&Wkx=5eSA
z-RY)R>Xxt3+|M>e6&&7GrNTdG5={n^Zeb5DJ>xewRpn&0XhU?BUp6q#T-8DkjtDB&
zFQsjkzxL<J|L;^R?@3>+D|rT8AZOAe;ieGB>_4hG&nJ101Cg&b|0}GYam1X>S=?Ow
zCjos|H?V0TFl}d~K;RwJ<UgLdQ0#T7_dH+?jn5h^XJd4;aaWhHCH-U&;^!4<y=7`x
zNtuoazlKqI=+ye&^ub@lgo6vREzd4RWyBhv?OCajpw}|QQX96om9*cgc~r1aQ)>V8
z($d2EesnCY6=)rA?AsdIj<;sw@5v|=e}4EVbZ6xZL@X&$97Rehp4U<T1k<L?)LDUS
z2U!?C=4X{Ys};GpJ_*A!cLi&j626`to8=y`4j_(OXE#kvO*n{=={Kl&Rcg86$$2Fu
zKH_-p<p+-}KR$licNi{WkO*FV7=y=OuU68AVg0axlFdX5jJKU76aK{lBeyWg4FiaU
zYWeD>6M8&v>748DLNYR(U+W%)>v_VZEJy?V>QK|mF)=DOYi<j)WhVU+QY3opEJLU(
zF^oa4l{;@X^45tkE%MZRdg+Dr)Z}D47ks_7RTn5E0^#AK-kw>gi?FOYo!IhiI|zH4
z{}aG;G4-x;U}{wuokHPwK_I;o%)QkporTTW#l;0rG|Ke^edzs_XZ~2b6n=1sUD)fC
z$dgY{{Mnl)CnrQRQw4A2XJEmSg66Ir8DU<T-><6fUyCOKKZ~JuV((m@-b_l0nr};u
z9JbVl56#wX_RUHaot@#SVVzq~8Xqmr1_9PiPSnX3@*`iblK$B_*N(#TpC!rOeF38H
zXw8RIZ<Ugnv|H1+7xy+}cGvzq5L}l+JoS|?q)K`=24+7<j`6y#1o=Q*;fFMhRpL%R
zz5HjkZ(-_SwdsHT;h7z7CvATrp=14wp7+&5=}tjMmI|hfmUGxmsf`0f@Xf~Y2|!1d
zIbq;36w<enRF8Th0;inComWw~X^$jWX(@RlGfUWG|L}AYG?$8}e~fT)1wS$}Y0E98
z=a+SFE1UK^>BibdQxwXR;fk0O7|mnw{EIZz7Zux*p)RIx>x8G|yVMMJH)Gm7g4sDa
z^qJapUvf&%W?1E&U|MbO$^{lL09G+vc=f`sRggS9_;g;D|N2`#B@3RES@jw1!}>!*
zjg=6Hyb1K~<sPk_y`h9%H|_)8K1EM(ivH6%p`#kETs7bKe>7ctJX8Pw-&l;4jfirY
zk&yeH+|4D8T+98EOSufW<nBxj6S*Z!Zsk_)cex9p4239jNtnyr%7=yU+xL&(fBW-1
z9y{;%dA*+34e_H}ySu<yiU)a+x+BI63SfA@6snODMLX@Jt>9LvvpO=1R-B>ykl>y=
z)!nVs$tiqik^`YG6I%#)3LJr;-l>!G$B*ZlZ(nFgt@}f~u#dTIVAu7~@p!-4A%lZ$
zHZlin)@?uE6-g6qX~o_Vg$aP@KQ$O$aWAwV^9Mmg>?%6spJazAicT}Wz3&8S-L#2|
zFbl3Qy&)l|oE#Hat%qtMNyw?^5)#i%MGU%=lbf->5B>yn&NZ7>(KdNNgAcET(|Ip0
zjg;G*{B$pSOl4R*<tH9ZB*&n#C<;ipVPjMDt)EeUX(;L7Dpe)FPB=S_aUl12uTFE8
zJWac<G|$F>LmGmbk_BC&L%`df`%Ryhk!ZQO{D`HYX(u8tOJ#{p9WaZud(+=|_~2YA
z@^s7TjB3V|K)ka0@*U_1)YDtlLfq`E7_z<5yQ)mKA9FS6*$RM@=j{|gj75xLC_Tcf
zhwb)I8=DIJMoQ*~CQeWx6c$D&{Oq9gh8l6I3F9|S%AzEWybN@~KDT+CQTjag2InfD
z8Ja=ILr3JRWTV1Y;Dtk_adpcuiFU}%grTpvuDUwgs;Y)=KRWx{0nCz}7M9-xV`b?o
zI1`kj-?PBzV8}7s=DMBW!H;+^Baf<OsrsR<m6erY(lBNCWbgFww>DpS`1&h+>Q9XW
z8$qa+ch#&i*-`jO+amif_x6ly!-7`Uda~wv$cfDLuy4BZaLKN~a(OdYuB)}JyQRzv
zu9m9A-JCtmvlGqjv9gGV-hMTHS;Gfe-T%gqn|F@)lr^X+S#Ou=ZUB)3$Aa_xr%_ZH
zb+wHb{N9pst@B(r#;vM13y%BGh^_*Ar@zD69qlTVYNHeU!|(q2xA8LH>}?`0sD{fE
z6|QgJvh}ig>Ia5Vy0B3PVxUv6S81rN!^x(8()x%^E+G}v6tb~7cSQkS*HrF6L?lky
zpJ_ojps@)suCc35o`9%lCgEE1-O*nLDn0L{r{=AeJNRYNBOnlA4<S0vjt2uo+ERef
zv6`#w)qUU^pvqQX>va>Z<o>pt%3Sf6rupf+fS{P6u*O%@P)}Y73JaL#<4nw&RtWps
zFKm)%`0~#2l1t#g6__>?3%JYY0z5pb_4Lm$?W)M?QsC49Bz3&rxjo>lOZ~p;V05iz
zwg3lulzn0y(D9$K?cRUPkvnAWRGY)gCP_oZpLN`?<zn?Xq<6Cei85*p%NsNPl?vWt
z^D~_5d@jCFA=Hp%+Ie{CP*~DQU22^|XQ%yO_VCEaYQ4%p_FmbGCubumYCVtCR;?Y~
z_;A7jbIsMyvhI-S7-4CzHUxXgIj(>0k+xrLO53tL$aIPTf9zzx+Q@)y^QOV0tNERw
zgv>%z|E40?;$?Sa*vZlU)*fkamhnEWMu81)7K{6z$&i1nlc`BxT;~(62i+9@94}ig
zC=7olCi9KjeLAl7StPZlKta{tvA?_oa!n71s|&CYh<r@1y~GK3HBmhnfAIMfWtg<N
zu9;z_Ujjgc3R%uvVgwMDsr9Z$0dx@Mh*akKyQdC9&p`a%+3G1p2TS=0{iGE(cyL{(
zX8udHxStKbDzF`B90YZ)eCV7f1EJ|R9)!NA<KmHqGZ$2s(F<bf)?7M{kJ|jL{3m-o
znhub$z?a;f@ZfY!YAR1zTlKBel>16QyA#6U3xta~BLi9+=U)=a!ICDI_P9O8pUZ;#
zp6xFC)%-pXdy|#4vCz~+ZsZ_{w5joaVOM_*9osLEre3)*ZBHv1Zn+`+<bkys{~aL|
zJg32_deMcPYS{m+`)NR&&&UOG(5r7Y`5&9-Gj_$Foz=wh={y$#HF;eZNXYf&;|fmE
z9RPl9zWR~<G%b8b+tRzb^=_x@N}+c-R(>$sH8MJS^yF=CjDNwlcXE;zglK6srH^7-
z=2Lv&4)qpS{<IxApEDbkwEun~<v4IlkT}b!|N14b4D@%vx}P7-7Dj3om7|M!O9$8U
zCv6;Sv+VEhy>Jsnb8UMk8S%62O$D)c`T&=;i0&F$yool}86{xz!}GU_IUDk$bC9T>
z@5lq!`t~+K*Bu8OM#4ky*Hk!)sE~4^C2m+l2a;<-enaw&F~Z-ixszlkLaus0U)|$!
z9o6)H!wBl3^AFcY6xRB<+t~&ZJ-a|ys320#ax?)sBBD)bFPGK+tDN5%-9P|<-MR|>
zH%(jNoz+cb9_nh8#Sh!|uC7bRL3-kWN|!lZ7MmqP^nRgfs9v8j;dKadrg<x1U3`;|
zwhQNyB{>#u@7iA0^GV}Hz{NYtb>`<C&;DddEp&GPAu0dKF!W>umiAd|94!5QwDX)G
zC2m`0H`mHu>uXI;<0iRNK0^WL=BeD_OsU&KtI!@i3hS?VzzbI|svMs!&np-ITH*ZQ
z`U%IQ>XoIpCcAXe<|n?RAl7<WxI5k6kJBOXvRiTQgtvMfG9AwtzpJtrx&>=dcMa{f
z^M}w#t&Hnwys;~74!&8#wsP?|>ZIuzKyupNVl(AYdbWd0gyZ~0F>wItTmANnHi`)$
zi1|>zi(C91N?x6t*PPZ^w03?lDkS{^A5NLZ2==IKr>q8!GQn(*htCcutHrNYqdPjg
zo*7X#7Loss4vvnF!aMER1mHy`SuYG2&30Uo=WNItzgqmpO564?Ccr(TBQ>^T5S<~I
z-c&O7J`KrWV8Oq+3m{BAB<+s(2p6Aou+K#%75#nNLmX3Q7xSBYh4s(hRKC><HK-(2
z#YP0%T20X-XVnCx(^GRAf8OW+F8ct<ajoY{o#4|1%kS#q#9phHXIi!4`)k&pW_!)8
z#DvuDCCO~_^P@x8PsZC3?kKJZh>&F+hVshm>M-%ZRZ`gVIq)(&0sLhajE{x*Zw7X2
zYA$FdM}^3VGx=@0RQTMQ_C6byqG3DJ-LsCqA3r4Hw0QVZWOk0VKhBrQS;|fYO;?4?
zx4k`>Z<(jm)7%BxX>IGJn2E>uXNo%c6D(|UtqPBn8pINUl*TfqS}08zlJ|G(K%$w^
zW%&NQ_kIo**xvT-iX6R8N5eU2BKUimQ?_lM6B{?#;O^1c>KdAfQ&7Oe`58g4L?PG?
z=j=^0wf^-V13{{%{HUvYE*`X|>9!_~6m1-Iv6(g=sz;5&vnsr9w4{oG3(KFC8)Scz
zVgx-Z=<Aator4qmkP@`hXqd__z<6jsc<R>M1bE!#xA2Z4`O+fYLTC9S9@T<qT^#gS
zLWG5<MBni~U2k7hr2<(#@N1knXI+X6C{|d=(YOBs?8|g`Wc2Bt=4dC8rkWPyLcNuS
z>;IB32o!cBDWcMH)>LMB#uH5SEQR2D9}C7@0lvV=>3Eo6Ky|B!=cD0Sw)FEHgDiL;
zO=dOSi{rU{Ajj+ryk0EU$99sB#0#$AXF^s49U3SoaA@Ev2(SdJR*`#Ym#mUwt}qdi
z&kc$HmiAnE<>tW7-X}CZaQ2_k=QNFJe<v$$<P6=HIh#}>+hDyYoBjP$Cb3Ed`OnP)
zAEKx`J4;J}>QrS0$-doRZVWnX_x*sam4K*LN6mT7e438x(U`yclY*P8Kmd0;F<FLe
zD<ySLTU0BM!b`}q0Wp}V3!HsV`R3Yy2!S0QW^YkD-}$Pi_Nw0nF&Bw6B_|Bg;wFi;
zRz;?{hT<;EL0q)6j}$0QviG*tPWVj3L(e_E4uP=g=wqLp-b86xZQ}0z1InYjc59os
z5`gUYy*^bd9=%VDG^MYgGS*FvWrn_^UfS9SGJiHR8_MCT6dK=<tmyiqAgGHa80uvl
zMiOCyy;C>g?XidU+rz!|8xQ_{4NoTv%)nICx@jxj@jeYrk0$MmStm0ss+5knvJMR)
z)OEAVF|8>IpCCHe1iLK2{o5D>1ouq~ws<?NO!sgiYdR~82mYad(Ib$VJsj%%z-+6Z
z4m6xs?)fKtLiw=9)db;{<xSy*4JZ%T?&oK#G*m#5mP@#2`rhbE!^k&^KcBZI;8U5V
zO&U$r?RILUq`Ec_BjZ|6!Wg5-Cx__7$_*p9P;FE4BkNE~RXvd7IuhJ{o<D2&Bbiyw
z>4uz~a&N)$MC1=!{2C`-z6_lH(3K4mKt1_BGL`+*3Q^~yzMWP_C&q0q!ytsV8vRd&
zQ7bdc=Lee^t-;5&QiZHaMl1i2@&1t;C*Ktr#{Lq*Jmi|JFra*(Qr1fU2x-$>$0A6`
zPPeU*GMTZy0|i>nWf*>MInUHKc*Q{8qZHTn$SBQr!1{c{K^5_dmf>Xhi>dO7OhSNr
z_J|4?OAzRlc&$@Ya)T`&B?97nH_0UImd{1NR~m~8u=T7L;M_FSm8{=om80-I>FMQp
zvw7CFpqz3Y7CvV%lMqYADCbR@Q0geB?k>j<!N7?DF(5`JQ!ljjmb%G0GoDQ#jcy1m
zZc>$<hSRf2l;{c1YaWO^J)ogS0Ve-BhY%YgH9+-5Xz*(h%c@r=Z+Va99F)jiDRumV
z6jGu{%RH(1HGYNF@0xi6f}`g6w>uNau-{N44mgXJ{*@(F)zo`A=&s}?2vUHcKGo=+
zEs*gNV7vu;f<XLq-=<*u?_YjUst0;4Oy}OZ)f*F=Zlv4?t+5n#N4Vu}i`k}n>py`y
zowxa%o<}o3$S}Bl9suND-8<)O_F=Z;_H^mEJYuA>whxtor)9+?4PpEF;T>YFS#<X=
z_hlH3UcN@t?#Zk^O7hId+Z5d_C}p+t<s5zkT))2jeX!>2-vI^qcW_EN&he!zKgeFu
z@Hu3LGBrUE3}pA@KytAql7|1B#M>$)FNxbHWpgS{U)dXvuxNFZ#61lBhSOM$;ZkJO
zmrpBYCT4!LXVCdk%?s~~B1pY5jd8Qm`c5tSPg*>0P>mfaK09bH)iOOWo*e=Tp*8_L
zGIEwB`V5@aN{^E3Fr?wF0qnZQ9qajm;;*aJlsnqv8+xYhl7074cYov~3L}!1v~a2@
zGMD3t0TiUN$`zQy4SyBNFzQ*JcdH_HkeH~|HsvPCqbeZx52}ATLvdI!^Ib1t(dC%s
zZZO`)nk$92MibQ&Ep-w;eo3mkivoPIx*SxIJbTuzNlUe^v3hi&Pp%P%>Uh{QOjEs(
z7xy^2e729-p?V}56CAz1*t#S5)Q+9=YpI_3vwj)<Y=`FPPM;Zhc97Yt`e&^#1kc`{
z;{OQ{$?L<Fj91Olw&3uXw=(p(RT6rK@#0dt*S?+iv=X-X5<RoAWX~Gc{talB{%5&7
z$6Gt|xT5>jYw1T&|7H{<$;=p}Xdw9%ikhvyd{?`_K43<+CyBT++Dt1ad>3whErEX2
z-EpJ3G(eXPs~?$n@ScYJ57wrlKlNxPhukq7LMO4G+G_BNeE0t3q`808`m`$tG9WuP
z+t^ff{ew|UkC`{w;>JkN#A6^R+Qm!#d^W<Bd`tyHr>`~}&kR)EwiqFz<9BzN7V#jn
zLOnyFcni_tA)|ZwoW5xdUQ=mNY(;NPDZNB=T0Rr@7E}wW!;M2A5a`%kP9dd=P#0!4
zGa>2pdXSs`sEYb%>>Vp3foTlOGqdYr#nwdz)xjR`hnJpP!$XeEfd9lMZ9fE><U|RZ
zTg@{3%zexwnA)gt7eiX-D?-boSX|6I35Q%`ehV7W$6M6ZRXO_cadG4<F}cnNr-NXW
zM@gL#BqGR0)AKeNkhwkDyWD+Rj&uhHQQb8LYh}v+l;{_-+}q-N+?}FuPhSMi4$1W;
z&KTIl)l09tCSd`$bk|MnP%pL?a~YYx1Tv&!q3$ZIhN=~~n22y|gLEddySq9&ySik)
zMsEftxACT^thV@>N9lTaUu@AaK3iry>pnD+4rD7AncyFcZ=O$Zlx&%il;%Y~i1-j>
z@MNkH^UC-6OUz%H|J5PKa>NLN)B0vd;BaW-vOPu|3Gno$nF$jr#N2SLV=}8f#43f}
z7hBJWisQYliu!tr3@Gw{jkQgm2QsjKXA3wJp>sZbeXHi?Z2Lolk&)YjOWAUT0_LJ_
zI!s|ajIv1f^E9Cc;w-KbI2&BB^IoJY6tBW;q~`uWGWkE3Ar4xXS7DxrhbcPWi)E+Y
zX2p2|cbgQsi$5Z32x|ss^2qvHMJqEMmOX{L-^Q$F>WCw)=?SvqSv3k#qf_COE%aC}
z!$+b1Hk5ug=}E)-+{E1g*+&#F)NvR6%POJOoW*7`C{V|8QRe#h%E5*IJg-4r-Iiq@
z3o-S3QBax94y}J4t3>h~*Qr^rW#XfgfarDrh*-$75OO#Fhh?4KQD;xQe*sDiYIhtQ
ze$=f=hg4g;F8D+<?8^i32G^0Oq-oktdU&*QbdZh>{GnE39D_YH$f-KbJBxh*1~tr1
z9Cp#IJu}qVSR<SHMJemnPY&It1Pvk~I`=(_Us6O+X0@=Ft8<)%L?)5mH&$1oy=(yZ
zMHqZ2lCQ!ZZ;JHAf$o-cMK@!m;m<NAMmjm~8ltpUEAGCLnVnO(G{z|%)GIEUU}L9?
z3LLXHPDJMQ7K_^k+I|!Rm9JL?Ic;5KqYhiy+HIG~2{;pvmo$<W|3s<fj{#>bK*ZU>
zaWoN%P9Z{Ynnn-fM>sTy1%Vg#fR{Z)uyf#cVB_#6Nx$EN_9?NyA?S58D)YMA#jE-i
z708&_E!V73klHF%yy4>a9Gicli?n7Pf_{RbvLM)n-+=DzW2!lyh<esY@SpT)ndFm0
z9OUI)*2hY69!gi2r0vF*Z2p4p%kC|Oxy9AXe@&`<GNH*7gT>KE;GG9s!1acgrJ$er
z$m|}%TE0R$XGHrI55v&bL$t~k)YKS#&0i$p*=br8J?@!;fL7CNf;4ECbs#)fiyI-J
zze%G8Avk&%owSu!4i>5>%Ps@1O7bOqjCw{?xJ(C~u=T#Mw^DqC+x7cNn6h)Xy`!UV
z-rsRs_D&k>sZePYAnXa`7>-4q?oCr6bIp9d<jdpHQJp&0pFWErD^o$-oMFKBa0m3K
znL-n)IH|Xd2rl&5<rD9<+#w}F?~8ey{D8Zdg&xhQ6m&C($Z1|UZ@t80h6HEC<?)p$
zWRw`_fn;%gg!9=r)8`uNW|yziyiZ)$^CZTJ5`FNVi4UJ2zyT(1Q%vDX!Jn15ex^pX
z@^2_RsY>x(G;`pr0WyvhzPU%DVLP5_=E_u6;ks!5;(EGkE(m?$#Q!kuo%ZUpBTRtg
zg;R?IBawo?^}&^&92`%NRwxV6VQSR<M^p>#+fP_}_F_V#+t|x-7une8{q-~o&ZowV
z8r3IB^Jkkt0f*KE*+9DNX%*{X2md?L@%{{KX;Tpk;VWN2>E&J$FjMBtK5?Dvl&`W-
z0Sanh{NbC=+cEU1^c|L7%5%2M9xqKR)u=ywPKyxY#MukP**v=+T?yF&bZpkBh#DUz
zL3_!obXZ0?IRy9;8H51!$=%L&^PGvbKF{xWoDW|FVEF~q{o|J7B^k9}drM0TrzcM5
z60p`koN3|4phoWz-@)HpgE;%N!Ft!~#};Vm#3;3_2shbbTa5v=t5V~ImA~+~Y+pWK
ze81b`AWuwc(nWzcI4FFO2Rc3z>vj_?T4s+GvKK3V{(Eq=RE^D7<3O-imL|ubX$6vT
zOt%25o>Vj5hLu*e0mLF{Yd}1<B0JE1m~KVMid-;p<Lp=pb#%ONCX?*C8UG_^SnY|S
zEJ$tok8O%n;%AHvw;^h1Q|7?>O9wp7@O*4(XVZLRg1n1%PitKJb*u<`5!73zAz{uz
zEDyHGP0r*VZ)|zoI)AWa?JB7RHtd;FiUHb2#Mtznv`*{yBa!qSr3tF$2;Ol49^QbF
z!w!DxF+Ir1$?>84gZXFL7P5kmrhbmwA%uROL?gN?6hYbK9!z^zo`#(Z=CXcFC2C`K
z5Q1Jla<*JV%F-)*!>=Tx2`UG=w(M!yv_Yl)?wL58F5mp84A|7tcIUW@WSiy!xTLNd
zq9cF7$}b>a7mmYYdS`Opg0$GK$%zQ&WxbK};~-=w^gO@%E<!;s%aISuomMflOu%dP
z-b&5Bcpd%X)3a+D5_mn;I%%XZwlO~@UxkzKv2Ki-a`;;%g#eCZo(s$W!=x~Uq+=|i
zmt+;rm};g36RKK&UjO#0KxOR==#G7$Grk<q{%?yPAn~yxW>fm~j{=6&ejkCZ&F2<5
zhB1FmJNN!BUj)o$e%dVe|9SV7jL53mr$afgKP$6%hRInrFw?EnOW2sC`u)e9H1Uh^
ztML4hKS4%_HE;LgcIteqf(KnHao7JzF_Aqvp#BOLWGTcRX;lX}3t<g52VHy(gF5FL
zn`}p?&&<Dlr}xjZPcY@OLR{>@BjBD!sBsd*TjNpdm<VyI*9eDBoT2G`JvX*~xeWU6
zrH3anv-f!)pGDFJgp<SXf?wxu&1jzf)*N<%5GY(OaiGdku`22`ewC}erlq9?^6zh`
z;Zm0<Oj{OGstaiSm32yF`}{c_{?d$1fY`q$#)&*SSUD@9h29FO)cbTs$pupJL&UcF
z_z6MY^|LcxCS@+(JkNmdKItFJKi|f+zxvU&<`Pj~(jp3G!Gg@rv2myIc^ubIwan-4
zY~!Wr&%;4hpBUIU;RNAKHvBXtIO#R+5L|u*NhnX}13Y12A?4+dS+CAS_(jQ!mQ3Ax
zmOENDn%g$s_&Ne|4whyJ?(w{vW{xsS+hu|~CuvH_WilZn*SM-8Aa&xyqQA7b-j$TS
zes71g-!@6iFxvk5ig56{%$-nb{#pLcp1j0fBOVG#k~!KcX;Mc@X6?!6k=@zgDU;7l
zL;uq_0HQF*r87*wwgU*67KBwhO2kIEO>D8hs6AP=;Ld~lcWIDGf*pmPUH1794-5Lk
z&m=xtUjJQhjQ8~F{>pE!s(YCM!1aFaFcZe#Pu?x_)$x4h5N?v@2Yt~{pYuIlTG~Hd
ztj{g=tj3q%hawrK&C9rnXmE<hwU$A+miS&j@Dc`{%L}wbWyrpMEy&n_e_K938CUMr
z{H;cG9xY_362328GO0Y7)2j7D>{{d-Lzl(*t2&niAfN0-b=(9zw8>8&tb--gCap&F
zaY$8Jy;9%zueedB5N2=@{W&AAH9`s>c>G}zuMO<=lxPOV+QQK<y!~kS_6{$CE8s62
z2s76K2_dhY&8fv11YXuM_=$ASL09b`7*Blmi^&12<(5qx22uT6bIJJcr0Sxv;eWf^
zE+i^tBJr3(P-I;mB&w|*U5M39dwQ|QS6Y9w{pFLUA=w9?pD-WViy^PmtN5A9A)xx%
z#l&FS(ze{yddq4$h%_DhzkGCR$46ZN-jkijU7XPiD=aT9-_GTk(t@(=5#{B2>?2W-
z?t%1M8ozq8Cw0in?`+;*614^+w`TavjKHk?eY4*j)Yq(5bahg_2Q;)U@W3h!Ieu*>
zCKM0hMEHuCTe42VJEIQ&Sha820pXE9nvLTHzPfxtWZDu`B<8@2<y!qu3|Rn><!u>`
z;^yr;PWfCeDr2&b5H8ZH^axd1X=dgGnBY_CdugpxPj%nv8pq%0|9-%^ZO30?QgW%<
z9amvTY1bhd2IqTMm0y9CWNaqN4h2o`JEQgSL-NMNOKXLmR-d+w=RO2Ma2doHfe+?t
zsoztO@$esWN45CT?8ea3y>Z|}-%Dwfh`qJZy9vuOnM&%s-A%lDFN@E!DE?zEr1|P0
zrb^dpP|M9jGB!aOF7t`mq_BzqcdNl?6OADJB4^YZSCm6dId3o5@3v~P3^D#{f@!va
zwQ`w~)tu$3dkl{QB<AUbh3P3<{&Fyuh*~CB?Kd~oQKROjNnMUg7(j3`n=6gJ;bvU%
zxk+=J%`+Y=JKP3Mee1u6nVYt2cWjvmv&np?&~86EEK9qK)pSm>gNro>Cgow$95}s;
zmddmv!kIR_JehwVwGbiM>Gfm`KPh36`X|9C-F)XnVP_p$TDn|Wi#V_2M&~=8`Sh?U
zaM1D#+pwvH9B-ckAi(k=K??^u&#vqI5N7TjvC|hXfS^-(v-&$Xk>wLkT<I#t%1wX`
zY92ek{&<h{ZR%Ci<I2qBkb+AEk;{%8iP_tbZv%MyuzXII;05Z`WY@aPsbcs_h@Vu}
zas2QrrQs9axCO`RJ1HrwJM)gCWlB5qNiyNy8uCE@biqv?048t;VAO)p`GaJi&)>hS
zS%80}Ar<IyjR*Y@+(#=A<e`JIKwm@cH~u4U2ejs{PHfUW)9_jLQDF=t82WwMdRF_|
zXmiuO{DY1Gb4=M!dV1*4R~s09#D`_em50k)7`|v;3?-w~7q568Jzh0ut5oeJ%J-&m
zA?&(L_g;SndJo+i3`75$s=PplK-^2z5_$Mzf==)ybiL2XvAJxptgPnF9j5~QX^i*S
znABoShPzpLiv^>2;YDKgTRs%jP&)=63nMgAjv{X#3<TJa-OJ0Eo_<h82yt%ILf9{f
zmJC@^>d|`2v(yVQUoL6whuc{cM>4-9rgLkyZy(12QK6t`(&s@rSNUYb6~{l#AG{V8
zC-C?8mX>P-+`c1;lAW}gmcCqEV)Cx62Op%|{xR+TWw-oM#u}t}f62eF)qh+Py*Tb0
zBg39K50KhE9*QvfG=f(ZQJx_f<G=yUIt*$N_lRz<>QvIHc9)%y31ZR3hUI%nY?EEM
zaT~7zRV<;{(qqG=h>BUDHXCqZLufzzD38~RrkJ>xzJKfe&=>W6x7n{ENA{k=`?-TC
zZw!eR|KQgWOv;(@Mk@Aj6>U}d9d|W-jk{Qe%>b#!(062(!a0{EK{XEkWTV#dl+yBx
zSPI5zx;d9T;W^*@o4ggecd&RGEgH-22CTzikFqW1=mc1L8vFZuK|fR5kX$*{F03dw
z239#;Y*2RD@~yewZz-UL_lmc77(%5>MW=7iAhO|~#LXT)13JuSb-6bjL0^kfzDDf0
zT-ip681@UnTmK6G)OB%jfxEz<4C{yI4vJ@G13<WttxkI8N9QUrMVLg2)%2?-;6#hz
zpG83^4}no{kt@W2T+!#RX&hWGn^&r#uQskRP!KCB+<Ug~q+!oDFG=+)G*NeUqS8E^
z60n_xjxg%!Rx{?fEqQ8fKe_jBQJkh%WSm<YFW?u~dE7kH_5EF@ks(yEt?Y$FNj)_7
zAzPOnIc0}$es(Tz?EJU!505FpmCpghv))|jjq_xxN4Dp@aKE8%ej0kQ&|``*T>6vR
zqh)kM6K26+rzFGW8P2@Sp7A4PLl8FpK_IdbTD8^wIPh|Me)08hX#IUCgLrVxLojx#
zIcmq*F&qA9$6+L16|7rm2m96BRFfh_tE!sVKWv+yI>B#Is)GBSU#kKf9#Ek{M27e|
z@s}akZ7%;L;gJz^Jf(}~BBhb_?vADF@=|3bqZXp+hG?B!FO@v<#9{vfhSArAmeQgY
z5+)djp1Q#3>?TQL2V&sHgG22)R_Zggo;ee_d1y%iB<QMclOM@(SP0p=5J-JZFY5j{
zQ|~_IDeShyMEx=fh$g*aWD_iR9aiTu%43TRn0w4wzJKRDBdfJv%RLa|?P^Hza`Z#T
zQm8Z8_69b^4DU>u!usFCi6DnNgnVp@lOO$;I4mpkqIJI6&O5?o=3vCW>XH^h(j4;X
zcj0Z*@h%zy4O9L`c>ngMjExBRJVGl&Y>m+8Q?=D%Ow(KdJEY(V#jePse}4kz>?nQp
z<YHZ3<wjQC4bIr}U}PaaF#~CeO@6P01)1`8u4L}9{+CDa+UXOA`3G%b{wqDMrc_l?
zs+Ueye8tXL1^RjR>0<?THpJ)D?@@_pp;o#^T{XN-sP_9fi#I0TOnB4SMG*)-t<g?L
zwEpL(Z_<Esjxg`c03IwP&S&VN8h4@R3)E1A6)AXIbV8o}HpRGohU>Lpl+e4i+imo6
zFU`wj;TA9JGdxLla##6E$v=mjVNY|vWWR%*TP!7mlYQVMGG)ZCVxP0r=G+6aDZ;t>
zeSQC!pk}AmkACyT;}GiV#C_L}jt;KBQwM(G>D@rpo3m3XS|_F-<<Lbbbfy`@KXymy
zU5fb1QV2gyzWZQZZYCl~Z){jpI}ijqEq{)Hg5sT7*9509_BH>VZk1TF`KYV)!kop<
z(c!`*_hk#VO;+RE_vATxOt+XlTx{Ih-5WfJveQfmA8TxT?5m63&wjx*><!<W8k)J?
z8w1qcQuxRbenaw7iBMK6Q1&X59;iVZD@Z4_v_=SesI(O=-4XL?c9JLvQAv2Z?d&ij
zAA2|x4~w}}CVl&IfeF&3ZvTX`x_=lJ72c+<?ig{k==pq5kkz-%!~p4+uS-CM`F{@Z
zlR<(d2SGfd6S~H7RbK?Af^Xjr%)2gntrPeQ`hefF+}_#Q=G&H}h75;kIAylC$9^Hk
zwF-NE>wO9$5N!Ge6vO>UUFBY?FTV95k<K}#>7$3za;!?&LY^olwx{tVf2+h=HLz$e
z8oah`ejDxr{uRYybrs%Z>A}=9HSzmuKT#GuIdveu)xG}uEbakc;iw$SYuD1i&+fUk
z0!HZ(?hQR0!3MEO+hv36arGjfK*?dkztq-7W*r2b*ZgTl)fVzo--n9Yxkh<J9#cDS
zU4zdOUuOTBHn0hvSFskJDp<WOYHqrnDW_c7vX$ke2}E_MXVo#%vxCoPePpDU2SR|h
zS&2P!tX5yVm@}-%U`D61#j(t+R17&7vWZA<Yqe9R)6&<*hD_VumIx%sc^RD`?g+>V
zzB%`ut0g#C&Vt^;4Yf#j?HhH2nXAW89F>&Bb?G8`s>Q>#Fk-RMgpMt|x4pqARx7T(
zU5=QTH?Y;#RY@9dZbpOswHT*?*H8SmcT$=Cb5|{Xy2;Eb`b2a)H0}64tEuH5ldmy^
z-qr{JtGv-Pokvf|&ULo})Wtk$^{A68FrxA|&F78@cz%tp0uxr#%(};Tk1Ng3SIfNf
zR%yLi3!1YL{lT48r<LW-kGPD$^sMswJUe3hZ_<#pnCYym2QlC$JYYMa_sD)(E8$A>
z(Y*kmtNmudswV?>QPrNKVJ;EH%{C4fdpa1Eb(M);Sk*@MX6W`~wqOt_!Y_hETC>N&
zBs!kwuUv(J1G@TNw^dQXJJ-2jGSf`FdFJYu1SczOTDF=sH65L@BADnEf%fC}<M`vf
zxs`wZT|nUJ?rC(++e_(4t%r*q(ZmNI+K<+v?4~hxBhRyM=cU@uO;*vtWZ-LGd~mc4
z!ac`G)@P8bYUQWGy>TuuQFbh_>e_4z#=G4219C!EE}{Ga?_uW+!s!P~Z%@qa5E^i#
zQ#lc!Or8u1k~1K0DzE}TF4SWDDNC<PmI$Pr7X%K?7c&Dd+m_+E=;;H@7YKw{<ESJ7
z<n}g0=l<GKb2TTob1$eD8JvSCoRT&E_`&Le8m?CQG7`==;@fi2w-9}LxH>jAwl(e4
z)!g)=uX&Eh0#myvm=66QV{{IxCB!a{<j6iJC=Ien-%Cb2-vaWPFWz|-mh}3asV)lc
z#)%GLl7%F_j&ADmqii)-Whw3Oy-$tt;*v9Te$UDq6%;6CQINS3*(x2F{20!JAPRDm
zwbxWaxt}xNDUnWW@z`FMy!GCi6RxPQ!P@WU>%b9l=U-H4_BEaB7zLcRi;J;(M~wuU
zsqm7(8?9acV0u7HShm4FlGv1y@cO8qh}$j-4!Vi^u@P^S3B!F3%jYdq{G9l_nR>#i
zdJFXaf!<kp-KVFr_1U@*D2tF-hrJ1Z*Rg-=%@z#Fx8hwUl9jsZmwpK#X#>>Vi|>8o
zDD!bD2zr0g{^9z;`o!wa+0jXS3V$_aYg(f*M0!wnht|mZ>u9MTaJa#U4Ar^^w9;be
zRs#R7j6`7^<`~kZ%`@NSlfwv^N>tiitOBy#(KQxG<@$i(uTszne?Kc4K#MPz5D0)+
z8*<U{h|s~#eR2S&rIuWR*19mzBnoV)W{J{)X?TzhT;C(`YANR;dF~1bv54ucA~kQ*
z)VzR;%N9LM&!}pGk}p!kD$Wy5DGJu9wB!{~Yt-E-?W>kKnrb0+%{7a^DzA<Be0FU=
zONJ=RJPdntUqB4?iAmjvUd?_@oXDrew+xQ@{zb-egyZ6j{atkG_)R<ld$ita(tT$*
zGEBMiWOw(3x}(|Aef)Rs)y_hc_{(_Ga5z8;T8Qv=y)5|?4%_hX)iW%gR82q!i-}&p
z6F1pKE9q|T+^WJvyogdt8J=_aQ)bKm$%fVg&hRTp6|U`P91POR+c*p$lyb+!iG_St
zKU|^pydd644Yz&XI{H&~4jpp^+5jEd>!`9L`NkMj;|!pQm$*Aiq&$-HpKfZ{;sBXp
z%get$L*TXIfmM3JNnZyZ@@b4TTzY&JD*f%9A=BP|xOdS`X9M(}KraX>wijU3GO2b?
zn3?KG7K??U_gfaC&;A^)l<@;6nrA6zzlWt#j<-ow@q+BnT|nA+FNTNmNlb7-5cEsB
zfEs&U55Hu;xsim2tIzY^?AnYcB669xbabiBt1D^!6E<8-hHo>7MIYo17pm=OG7AF-
zv3tS^o=6t`m>|uq&db4C8WLcLP1gJc`m11-tfLF`4#4{La{*&Fpdg-Jr10zEzB%gI
zQC3$=O+oj+({*a|R+(p8<UcJSGnvnMf~ylV*D@smGef=&>F)9=h(4B;815q-11Bxr
z+Z09H>F*$xNsHEF=FLKR>KX9!5m<MLS|eqSUAwdmbwn5~dX{tZjq6i6>hldV+y&lX
zYyA}fEhQ_uAT8da%$!!;Ow2h@I3gC2B;@1Kmjw<LbpIoS{A8jd){ny{C^6w>!hh8U
zme4~>3`2DuMv|0J0U@AD1^jteFEmVaMZUzaWC;7?#m!ApXxOc9=D5D^gWuaLJi^Xl
zaj;9oLe3Z$SRXn*=Z8@h9z7wXFkvo2vIq9zLwtBNQg4Jk_NKh;v<vf(?Vm9jqddi%
zF}8k<b~b~1!4Z*N8tNLrztEQXu2-c2T||*VRrRNw&xF=kY%ep5HIZ9})DJq>{W<c=
zHprr_o>2>Sfg$1?a9h$`vSVy3I=H$?*5!R1tJeJi_v&&-BlQ9kb1#tdn2(f!4q)Xj
z4Cjm5!QA5o(&ZG9Q&B_?GBVg!R$exQE6p0Ab%7sKkls2VD@6;Z#dOa*{t4vQy-h#~
zr+t!P6Ht+F*kTaKNIGTSjPNs9&QQ_1f=95c@qj~<JY3JU#2+8KJ7&_9S{D}V&S*qZ
zf#$zIJ7*{H1;FW_zm5CbMK|dH&a~XGU?D7>TF;A;Je|u3$T+CUG=)erhJABU>|xZ&
zkY&F;@g5VkFtg@hz=$=EJzw@Y8JZn4#2%8la0vby|IT69v4pvPtifJ#qOZ?0e-!1u
z#x8gH)*p6*ZH0s$%z_<1FTC7tXKn<Vme%|9?V!M-K8pxZ4?<U~CH4zQ8)9AJ(z3lK
zD&d6GOBMNN%4z`$Eo5AUKv{D`5jaM{YqmoCDu`y+FFHi!k^LCSA(@5G7^fB)wcPAp
z_0)v(U){hF3r^KlO+JYMRpj-P-%CYLrgtlcCT`;2`rDMUE<}Z%omRol{v59^XhzNZ
zMsVelSA6q#?F>IIplSfzAW35~q)Bns46FPj|5$^sTjS1fzh2iFrCY57Xs@D-%)W_X
zw1)Li{v&h!T$cJo50k-)|7trJTn#Q)zGIr5_6C3HRvXl(6XO~>RZPU;eHP(|W_n4A
zxNDd%J=$oU$ZWizG!G9HEJ!2@s6S=IG0;KlGIfY5vDbx#0m%|*1A+J6%lL`|nfhe?
zC;tLcs$;bJ;mKZlR)#_sAJIWzk3<z(eemPQp<(NHQoI)=T*E3WE*LY33XlsgB2#&s
zBpz&hoeZaxn)JJ08TQl|*|(fHUQuL7g#?h_l!V+8j61C5%2_z*ui$}${l$4jddv(c
zWF;W-by-;!uv!-AhjA?JjyyZvb^#|NlaysF^1Mddi!a5c$_fN<w(3SC&Zqv09Ow%z
zpz_L2t6JTkrw-Wb89|IE>?xKwg77!4s1O$v?AGvgHZYv$BM%DKuIEjzE*a{oDpF87
zo`6DYeR&0b54}iU<n&;05ESR-;&OI%mgv<%PILguC?%!Auaj}q>IwPGA97F?orKiv
z52jX3X}95QTSXWNvxOMLYk~K^!|!9gw6u1vsM-%&QtZn-^ATUdou*Goekq!#f7dln
z4~CZ}nlS`xzIg6^)@vVL>s;9D5%g4Yy`Zx6+tPTBN3HjpY8*QqJt+N3$L7{=_s%)R
ztvmt1$OUFXTOD%PRr~+C3PXC<xp_DAjJ=q?G~*-C6OpRZw(o{)OES&Z5m6|JGOi61
zAva;awnbuqTU0G-gx|BGI{$ii7dKfW3wGDP2Fh-~@VqiRIawVRL$G*J<N@m$N>Y&-
zL<o<f@BG*QNvFO_+N@EscS2?;7L60>7y_OomH6Ad+m4t>n}IlAbhDU!++MZRd{C|O
z{<Q4eLC;NkE!(DXD?^x}h~=HVm90Lf)0Ou2VRwg?v4U=ms3`6^W&iStxw$zSa&UlO
z=ytr&3+5F9&OrF=M^0O`9H&6hg_D(bCBOOPDFaG7;nm4+Z0Ec^6a-!O=!ZhJ00ZPt
z%=F;uRy*se`G)+5m@e15tbrE%oN1gF=`u8+Aa!PQ$is=Bp&aztFWB!Df&$N<Kaao%
zaXk<>+c})*iJAK8S#HnV-B$f#(~t-@MP>i^jt|X#E~C%CC+kx6;@#E{XA=(wY<Z~<
zni*V5FrxJ883+IEs+O8Q_|s9{a?L}qhfP!2{$)?h#ppn!DGV=qQJggpreYpQsiMr7
zU^-gGgMh4^OY~QOmEg(ysWzFq@nXovDbjq92>e@XzEmJwxj<69@@Le4t65q5kI%Vf
zCcF;^F|e6^|K4JJjF)+Ksb9og>awlLb)j1xclce?Jo7X6ZWTk}ajb9<ma_j?1jUPf
z{x&`$?=tJdDhk(j>*6x+UFg{Nk7&e^g(JAr+ogefxN(*pxFd_9v>kHtahbNlUcmBA
zRy}R7trvJzC*kiPwiXtY^~!v*4yxqm+({qTQh1avW<utRJ>AR{6~l-UEy>V9+KheC
zzGrps!Z~KT_><<jRIb!_m3C(*)GYq!zx#vy!0`!{q%m8Mz0U#~Ilr#3dHG^|gqP6x
zv!siKWVB)yBPTs8VCUylvqjm%kRskM6qMd#Ud5M9BK4qOM_#n%Tvp>yvZQC$YV;4P
z?KcGd>Qf#GbSX!^ObzE3=Y#Rh8!Q}jQe$u|%Y3E6=+z0&oYJoY`-is0pOLY*H=YvS
zPkEK!gb2hYo9gKdJdI-z9s86%gdQ|(MwD4G!JW*W3UyEcvag7x^_QxBjRBCBMxFFv
z%yL6<m8A$cuuwp^QsY1-PCzDgJL=@0``JHt3i;yl-=nYeZc^YiE--wJDVz}mVIU08
zk(3tBNKUj#u5RaK2RJ+4TdGpJU;q(fGpYLq9PW=~|8gBNsrmFhK8kC$5VN#S*;@dP
z4g*4>kFzdarq>2`ZfSygVXzz#(TSz(JyT<~i!=~^#9qka4N_<DAtlGnNHi8D+x&PT
zL?H@$?3tf+*l*68s&}*oOEn@{{t-4dF3v33p0949Owbsrt>fy_pI2V<kzE*ET64ic
zPm(BLEr#<+BI37=_2{;%{4m{HMXbRVyN{HHg?b<&%fN<*s{=LI+m7Pl!RNsT;@iCw
z#jHQT8~gO5_eEbtN6@i<_ha1U;(`SrGNe`WjN-IlJaOqKhGgJP+CTUcls6W_HFe}z
zX8V9prZPF+SdX!L$}2;5zyGG^`vGt2_yx5A=Z$?!*qkx-ByP>MF?h~!IY_npY<K%{
zrzgK}8{p_P`$6z2lUk3S5OHMSFfwP=W}{%BO*^~vmXW)pI6JcasJWV`_URH&Q>N;)
zs<C@MkN+HLzxifwNA9n-3*JTE@VM8tRba>49HOwNOr@S}xl*nReX{WoF9~j06Jk|}
z{-rlQ7gbC>O{<bVT(bHu;oH*Lv_I{Xe|pk=_8-ufQlJ4i#D8!>_gQN59gz{wTuom%
z%)uGso#FP<2H)3wqc$gc{NXydFD|5Jr;tv?h#yV+i;FljTxRQR!K%5Di%yH_ZZS~{
zhD7R#xN$75>3#r1X8q+OqHP_vPJLTv^D@IsL;#ixP>b^*4+5D9M!1zdS1DbP`J&<-
z;N^yQtr}LV5Tm_UO&RoedT}QGK^g_L@jKji8cp*sp^QAa&Xj3G%~KdIxmU93Q&QpU
zjSUATES-JTI^2!c04iaFaADBWL!t0OU%nOY90kqs7mzUgWCA9@y3f(L7K|OF<3wI>
zw&q$52;1A;-5xm%aBMkQAD&np187k&v+x-DH>^=ME&tos<R-;qAvNiz%Ko1WtTUXP
zrV4<gmF|<Bv*CK;jz$N6RfU(F!R@uF-wVs_H>DwiAb$EEJOe#ZA$joE(0i&+uxErC
z2w&QJyYq&I=BL-pT(nV)^}26jAg~&<2cVxnbFQ8H`ytg>F5>-70WA$HAYw<f_#<>{
z?wd82pjzuzxg8A>&$7K$Z1Q4K6sGW^R&6{h8~vEbkb=_kiLdLnFK7<l5w#5pymu~;
z#g#?pdQdNdr74$71;XYm(A%3Nur_^F0N1`hU(>?T!6x<ptT%M)aW%CpJpZ;pkUs7q
zW}&$=gkJg0MdU@`!2Hh;C}!`TL7JM<*k~f0ZltJH)9F>;!~*Ri+a5V-r?R@&<Ll;p
zF-c$J?6gd|a<1NWLqBI=YLJPKxGFgN49Y|}>{8++2oqM%rGb)$1R|577OKgLmegtv
zZhYCwGH0Ic+bT+_2a1PK!x@uNae^jUyXADHtxzx2%mQi10LN_62hXtdy~!fxl2q2A
z`u(p*ZB1aDyAEr;R$yg+%$ImNACN$AX&+B-l-Nu4NCchvwOs%F)s-NRwBUvJ8Rj<<
zv0*H_)OS;}qj9eu+to}jbk9crTb!UZ3TU1jrIgW3uQ{99=a(V{*uXBJ8HyIyZs9{0
z$eX}_i-3&!-F&olwm?V9BL`}kUArPl0aV9qHLzzt+F&mny)O~Ay(;*GDVyzd3k^yY
zxe?4tr;pm&ZhIQ89eqsRvLCY#<BvcIV~$*uB(H&RVdWh3xKwO&ROj4OFsaP*KG~Mp
zy*YHp0dgv>BlFU{)SMd38xd#$mBqmqeL>QIRggU6a*s4CqZ^yY{cY8M3X5to>kmKl
zu6?1B<K5KMTsq-gX=`lv;gY?jESUb_t9PWfnmzd!wHb|Wvi*wS**`Sy*SrV_<a@W?
z>^$7vK3$ASrRkUd{Uc?qmf87cz(4fOn7*$9GF_NKr^-Lhm?oLVw%M?~MVOPl9j9$z
zdyK!Ert#W(6-CnD=?vzIyYfSDrRRD;PT<1>>9aefU&Cj8!wHECGA`~6g`i76U2l}k
zV7P1zpU|gkEvCI<_c+kPL?yX=d1i>bi?&$i_1IX{>d{O6A__uW&zXt0I)^=J&Joi&
z()7#`=3MuUAk~oCX!$x{gM2Wn=Ad;TQi<_b*L<9Zj{;PN9{h+#dx_hh`^<6B<xLuR
zctwlW&iz$I2a0<V%a%2UTFjh$uC@jTxk~Hh4L>gTIDfkiYC+rS;_fDuFdO5oKAZFQ
zTE*wwBK+X~V6Ni~L#K~dZQ7)~0MpM2l;p;6!7h5e-`dqYTbh>rrnDo;f1C&47ksCo
z&gn{2zXWfXRZgj_K+i)7VmUt71P1r0qRy*|d2yknX$A17+e;xuR2kK@^|c}zyceGn
z4v`H|pb>ppwx0vqiVNGC7-FyndpCy6OD;K@Tu%Q}_Gt|aIu93Yo0*tMJm|!<>^E}?
zn|K}in+GEF6KbosW2R1iyZ&~6E67nTEg@I*O7T9@D67=h&Z(snFKtE;E~VdD-F~Og
z1{@2I%|RbhwpANj_97k0x43GgTK1>9B0En8Sqo187YSK#Mnjcm(GFba>3Pg<+Gz5n
z+_^upTiao-YP0(`E1!@4D&R=|LzZ<A<KS90>L22PK}_H*{ZQ0TC`9YxLmBi4u}+!}
z()#L&R|r&6cKRK|8_#=doTZC4Ss*Rx40EyD*}kOH|Ni{hr%Vr)t$LPIv)cV1VEy^4
zs=Q>a2<7eU{Gd_$#fLjYH0Uu;vdjQ9csW%^Q5*@;!^1oq;he2m%KY@w08n`g6hp^6
z?AZlS$$d_=9sKYp@?>lJPD(+6-2|EogvCBAkBWR%{Fij`XvtJxr%<io%(=>C*-=h^
zqNbq0D=qA3d#mr2sdvQH76<LG$>))d7PfnjZ)()yc&=_gYag#J{=gnc@JFJrRlY;2
zxPbIsIPdG5w78(&bDuK>dd$n&`CDtC24029IMb!ye6Te9i%fUMH4_C|FU1?_0<YNV
zZ@pXLC*e`iHhw17rQvM0qpOFF8Rj++d}-LWZu|ag+xHNGv}_kY%tG~;lT)=*4P|en
z?W2DpE@I!b3Y~tK4T8ojR)fhu!9T-}<<7}56w=QZtb*LO$Gm}wc`M9ZyEN3|&l&7~
z{)TJ{zEqK@)+5o&x_5=omj3L4?_CCv^U;6l)Mrp|CuzE-<$v>J5aEn;cJRCJc+XR`
z#I#C&>`QzFBU{00>*O<#27|do@1j|@cX%KOZ#X*9C_dcrUKFdnhMBV?$-OM{wyvIe
zNB4Q&SfSM11h7zFSQv|1`!YZ7#;)g<zdEtfVZSZ!><#23^C1W8d^dbuyvkz=(_npC
z$nUSzoQCBqZZ@DxuG~brfY>7@!}4+TuH`zgx}E3A-xs)aFJR#bqOiDRQri`XY`#>;
zwnq2@wR7i2Tmd=UmK>HdySWl}e7H*E4+pclPyQYQQr&8p7Hw=vsx;IL#~PUEk1_~A
z#p|->6c_(p^|}e2_{V*IcwK%WFUy@k2xX!D3H$}L$jWLvImQh0)z4Wa8usJ~A&Zyq
zOpRXqh?IKdBo83Qr?1Ai`1)?lfg2k#k-dfkLG4?#BR+xoZ@4eaR@Rqgq=WKMm565h
zEC8IFjkmel2QHr4Td7pdcgp{;DR2BzS(DQ{s&Q(>x_IhJQ}X=W^Mr{n=>N`6W1jNp
zr@jIsGRF_*W=YNTuJ7E-U&MQ3KZ4HV5sV0E$aEpH=ch?&(Y|kI0Xa-8&^z?%yzQn$
zT-s{V;O`V>jq!_DZR8nOJ{Uos)DuJe_)mfTg;V2Vdne!S(?9Dgo|yJF-UZSoQ*R1z
zN*Vjz!uy|!%V1w#1T+SeT0fc6&pxoVav=Y*eK^dqG(P^Ago$dbAzy?oa|>PFEsy+s
zLx%2&bZ(FTJ?n@{^_<0YGyYt8RHF1n2;$dr8sD0X#rcqlm^Fw<Ki2J`D+|ck0|G!N
zasznpKkX=jmD0@rSHPbX--@ScqmdqpXoG1=cvN_FxXtul---MfTd+z&se5U0I$UD7
zeoG=T9O)s64`|<>&N^BDeY7%h5A&46_yV!l(%|aUhvzZ$=X6EZtY5bBe-;X{cn-?w
zr&zp~Y6>kBY|?lP3H0mpy9(Xw5C?`?K>oHKd%IXlTl*`1bykRu<OXQBsZGV!03`&)
z4H5I8L5y+>&8O=V4gZesc)0#YT{v6WnwyiAc&X=0N$%s(!j~e6X#aRn@2p)PRKxc_
zDCiPgbg^saM(OYJmMOb?7YgrL3ag~5dMVL8Ep*RZk^fkXAtl>*`A!NL1HWVGQcwHS
zHK^-hQXc!)u5l~G7{(Nnr%6?B-7|t!^}tXR{>#&Z_t0;$Odq<cNzGEjw@{@7^P<J=
zO`Eq5ukuoxD)msr$OY>DW4R_v;aBTjAG`J!q9bSzTM+$kYymjgI$ZH&O}zD%?NVIt
za*RXKyXRU4{zI<JKYyk#1O0#Y?*+c?O5iS|T`Loo5D-<-*Dk__G&Ar;>rFzfwDHST
zQI2L1c3W{^tuxV(LpxoKqnBPr`kvN1v?8msf6?3ij|^LO{M~><M0;hHlHX3I#o`nE
zulc|Ll{!OF(Fy|(F>#B$KAZ2oH7%F_kEZjCr@H_D_&HX`J_pGj$3ePmva+3IlTr4j
zkdeK=GLI`|TypG@6=fuQhmd0=;|Rw{h?99TkMTdh8~?lK=6QU+pYeW;r}w=@wSE6W
z{Ubz9_K)9A<1qjS)IDXm>0cB=c8)s(c;)zlf&+|%z`E`RB>8=Y4<sP%uD`W0n>vNv
z?7deWnwt*kN|h95a@R)9ws@y@#cgQp`Ten4lhp1itc+$cNA*H1j84!kq!Iqb=_+Bf
z&OsHpJU{cW$@97}lRUc9uN+-*eTcVVJXhwYj`>e6g>xwR6~)jml|JIE1CN8M#Tlk2
z(stvlV^7tm1^q&h{3h53jTR&q1MZ4~-{olKPi&h01HZ3+9s-dnK!O}s?T94xeW$sl
z>*(_J@+junE`fAZwVz4e+&Bx&TCg@(+;7YmveI0gs_Th9-@#+1s{0j87}ToMIp;vU
zIYIue-FRLsFU*O8B*8CRG)*gw$uJ2YL{_hJ)sF=&EU-$+%9(7)EjI(8{K|0E0$m!(
zladH|#*=#6BH;;k(51RCB~p?P#>DqPQqzr0otBM_qkd`LSnYD0%ZH<a1Q86GQZ~G+
zWoW}&-St3>(9jt{paRWzM{#qrV*>Z@j@hXkN40p1-QQ^A$;e1KgHsJU>BhbUspxT0
z->4-}F9tq^7n3<<&jZbjjB$NI@bg*fU-Dhbqon+p9@8U>h&>*0Q4|ZPOa3l~fqAr#
zeg)De88Y**jpp&b>iN@!XvO6*3^fY7(ClsF+trjlgVF)VL@ftc*Jwp}m9xM}gc!W{
z4CxczN$R=0SWE4>{ChI7u;F8pV3csD)N!<lqE~gev}*7(@HRESkeRGge2HLn&xl7G
zj)Z?mDx1T~<t7T4<$S&emjpW%+7x^Ik>f0FX>N&96Qh;T-{ww@<FP++sQnrNk_=QA
zRu8O{3Y6(;x{B<=)#5|04|i8_ZiTn#cWR3V;ih+>8lhDQ7?ot5^zN3ZE^p#&WDO!g
zff$ifhJ*qzD78$h%%t*!)wWx?Cl91ov54aFjOn(Mez5QWp#cPAii^nDz!k_0qw4Yd
z^$3HHCOLOjQ}p9Gl)~OA)kS>h*b`$s*J?QokMDcmu4aD?dI+UiL-?w768CXKn6pDt
zZAj1MF{ToaaTkzGTPcbaxwu|uKO1xt<(XfIlXu<^H$c2r;737Mwu$pRY5=KlYHzta
zinwv^9~ZMQbyxJ2eqBdVBnG0vmKLXojB9#T=|wz52i+#O0o-ycs%A;1wWFyoi5ZjO
zIA7xh?1TiiGOFhl-DkHzX}1tEiOe~wUI7^?zLG&0kH@oQy1vB@xhA_;I!wFYW_)b$
zT^b5fU_ME&Z}TZ`d{VRt5!stu8zxR%;`8;k3FPa2bKKB45VRONp&1~eTgmndrvYxn
z@AarXQF;;%*npm=q<3F;@2$>8@Rkbv?B0{W;@P;tUIn9_sgq+3G&g$x^n#u$7ThYX
zDB*CHntsUw(=AN%d!FRICwG3-UK=698K2qF;I8_U7A<nn1I%Iy(kwGvF=@}i<-bZ&
zQ0}|4lPb>H4BRU4!+qU!OIUTedAbQ{5s=Ob(@lUqEhsxrz%iTO38v&Qo=I-gTUH}e
zf<g|Mbag#TO~pD~ACS71x}(qhadPwm8ln`M580ojht;yvb~7$(%QuDEU9Emj_<9k3
z10uv!!EzI8?J2UW3Z04Mjm8blv!q!eCT>~yl58ZLF76T?ct5jYGcL#0spniK=Tu|;
zeL#9KA3vX&$-5jcd8|{+uynhJJS-jVi^r`;$`_$wauWd}l<;;B_`{rcsNP@R{V2&3
zpHe(q#c)RK@2Ra!ssqrshaVo1Y&$M=W(goovbPXEB&_zn<uAyj!e7mm<Q=jbF-q$a
zY3UH95{4iFb?}{G(Q8KUz3;z<U%Ps21GF*ef$c;1r4O|^yXikhudg-C>`dR}$itN{
z?hAXkgZ@g<vI%?&xhWRly$R>v)!?A3Fj3jOrwRHzSuOvvZLMkW`$xsC{b~AEK{CE`
zdJrMiOqR;+E3pA+y}p#_L1$hQ)UM7C*Mhn~%*rKD;HH0<cC?Qv0&X|2{qaI_fM6PF
z2B$`)1a5qD1ad5CvU^MGGIF@@cfF1g4sqBJR91*dhCuuJ{Q~{D)RWAnzCmXzHoc<4
zJEZSq|H=3)UMX6A$K^j(o-3pHIW9E=0atZP<qNs!<#^&Q6(pZxG))JIC|4fg%3_mb
zhSPJekRhdCHi>Jv!nOzDZg)Nd*#XWKASzJELdMfY=!#FzUOpNVuIjK{I8moIzULcI
z&~dWA3*~snG{^=?XQLPf;kX?gTfwoU1Xs!(Hc|C4{JfLLu%>9t;uv<`TnwZQG64l8
zHQ8(zej?XEO8$JREyXUN<a6Wr;7dy3vfL6I!76^B6SHN~@F1{B$rkYD!L9k*!^l4C
ze;s9E8?~XMbpE?!akmfluqkB6{k4vrcO(>)(fH7Fu|1eR#-nA^JhGNSb=*5HEE!^8
zfU2I4m?HVaw|U|$8)t4lwE0pSLjf!qscF8;(f;r;wLSD6`_jc-163cJQ%lED130O?
zHd#svirm4C<tQ}IG6T3qS27*A@|(tG^lB#?H>-e5%SS;(PP=&JlyvYJz9|PxE*Wbg
z8PDfF2&WiO&n$y^#{0ZwU~9mDZ)mXf6sA!p+@{_{z<+XX_UDCLcx@xuI5lWSUA(tH
z9-3VJ-LAGe^_!z?pq#E%3)mwc?jKW0Jn`<8hk|VTZ}4gtkdpCqxQnARWq<NL)jua8
zH$?jSc<k{?r*rYp)zMMq8jt3s2Pw_%Rrp>_!3`%DuKyG@%{k`7xT!+PD9AWNSVdl@
z{Q0e}BLagl{n+uRaOElyjLQ?#B?D0<ar)(jg0vmwEtPC%TR(Iko5YpWkcGsG75M~o
z`^@*h5F-}}0|4c#;{|lT&;OLlmb%T8mvWRt3wdGr8O`^jBeX91?p!VZo@A`We)E2|
z@U*viQea7eetSDwwIIOg<Yb7XqH}|JdsG@qNl@rmdjE2Fx%gxYmzNouJAPL4shc7H
zQkM?!JYX1Xs3l&4WtgX+axnG9z7kmaeLa!t`KBJAyLSE_oX-@iN>Qj)psMtrh@fuR
z{7$=PjdwgmpbvN8gIEv?vhw6s>C9wy%PPFj*7EDrN#k?7C|jDKp_}n%@1x40mLWBJ
z`=X^9FdI}d?v<iy>8nU;GBQxoZ<^JoyBwcIL9S!D)z#tZX>8xmtZg2vk4(jjO7)(D
z!6+mEb_A6yiMA(EF-$xk_qfw0S7VqE<@d!;`8lfe2juyu5jaT0dFhRRi-h|;hkEGK
z&=>(ywMx-gE`+i*ZV6CW!tGR6A#~@PW|bW3TrB>Dz!--fftbF^zVVPZ?8wMoGWHF!
z8~d{xT7<Z7j(2D1leV?+p`z}qnr*Lr=_L$6Sc8|XjYY~E3vQa*5StR$>ug=RhX)kP
zXFcCBh}Av#yCm|2pCWy>J&8O9RbQsO6I1r5qGL@1+9*a?zr5UOoV{tu_2Y$qViC(|
z{`Y9Lo!~ZCDKmK&XEVFm@@75J(PTxhmzcJ0We4%i3hwytX&#e)j-ifeP`>}bLb`BZ
zgu_5<;Z_TMtP}p;(BQSesFf)bLN?^aZ&sDQ2gKSC$Som)$qjSgsQZu|Z@b#A=A<>;
z!76#$I}L1MMrJ;;L7>Vigcig<-ZK2)#(w(3Ef50|+3VyNu3Crj@6Nv_USHN-dtVcu
zhiv}xvTM4#GiFZxu%qFdPm9B5SbQ^(p~IKk5y829J<Q6zk@Ik9QO^6eoAXc9H?XRf
zo+yZI7b<1gk=c-2`~FJ}I|+nt^H(^Y`4b3K(vNDD<Mq3n;FUn|tZXZ~WtaEEu$|tA
z3M|=uwrWHX<9!cQ(NM&rx=6}Jp1pkEeH9uO9=ZMAnNED>_NMRo8@74QrT84+dL92b
z=&#)LItZemB)p(5k=C+0aLlIeT~k_QTsc&?RlmH6=!rVJxcZyud7pwR2^8ad;$g@7
znBFwj587|}&6p{6Fn#yfEbG@-{?3{57PbZ>b`^%ZJ1Shx*)-Y7KX6uT&iI`N1KDNX
z2TvOc20E;3YMK%$Y+n1Z_CBov$W~2U$^4o4^)y(ZQT1YG9PC9r{28D9x|#_dVHf9z
z7pGTOa#*X+x@l}LvKQeW-pdyktIvnuQGc<$B(I5}j3FJE76izrdPm`hexIIt-8clt
z{@%rPzPC~bNj~J&^qm4BP)-bJSXOZ>H`x#|ThbX+=ews9Kl!%6Keqw+>z??c)}>NF
z6C4}!Xzi~YFT>{A<!b14Z$|QuMa$D}F0&EOSp)s?-O1cqFWyro#D)!WYP#CvK@40s
zts>uoeDpsEd23LO*KLiFHUmdJh~g6=j$IP|K?I#|K`*e+T5@%Lee>feHeUvwl`pBD
z{ok+<J>STv&M!NF)fNRPhz6<CL9Hw+Mq}u1!@tr(J+Mn0-qDx8=NIq21qZfFN9@_<
z+~#Y1AMOg1iHwN8CX&RR%eAcw)fRWD=BI3*%lT7uEwhuZM7iRhkLZ$zFQyqIVG8-r
z<9kPw+YmYbX4-=4L-4GubM0dI_SyNy1oB1KHJ)q~C^G+-$QvMPY5j6q)On{AW*-yY
z*u*KInVlqX$>K}5R`@<Afx?-ijW5Q1vOVaNx@gB^U6D%B1iLY7mtLnnO24~ySwC?8
z<AUXWd-Fz{BdE}jkhwiQ-KJx=?$=q5DG*;#zjCv0_a$+XgQ(!%zj2F!EWBWy0mYK`
zK}noY34(J-m<-6#sClnF@NUTZq^Zy2btU%bsPlt2sGWQ8paABq9NsdCHyMNa8%^P^
z2xw!*lau**!W3p|P2VQ(XzuXLku$Kj#zuAS?e+BWwS--d!WWLT8;ija14S-W6m;%)
z3_3)pU*<f2*ItJ{OH$hn?#PsWXZ3#4a4=JAX4+E0xC6+S7WiqBS6>>zcFank%W)~H
z3PKeh_)?v5Ik`~m*Mw25x<C9!QeWNjaY~U~0?=B!KX}FFootP_#?}@<Rp=yV+rusK
z-pJLJxlQe0A)(G!lGN&`43?&twDv(I2?;f$R0oHp!U;^dXGo$A8Hq>w52_@FFG;-^
z77GIVK$dXRcH~?utY}qPa#64h+Dn$CINNS7;EzIqvy1z@`N+c~fA@-f1|qIDPqSvq
z$Di@750%|c<upzDv$eL~{ibq?BwS(nbgj<|n3nD<9>?u@$^4wd{keOZ6mV%6vP`P*
z`89#g45T$Nv#mLb5H%Z@wEelkW<#?p>4~Z*SBke9+JCf4_PW`Hjt<Hsh(#>{^D8NR
z#&0a1SqiEDO-HyGq0}=muT3`eFAq-7Pf3RPt;TWOUAZnQ3Qftr71w8f0ttS@n#7D)
zkC~^erSu9F4gEA)NZr;<**E$68t}V*pDMbWqYKdU)w%Rpv819wRINe0I;2k)0lVE0
zXp_f81`Bd@e}Y2ZPU%CamyQ5DNesM23M?_G?+_{WOcnW_pXZ0Cg{Dh;j7db?iliV9
zg1=DU!^!I9%0QK?_gG`o(!V-bQYf{Y)K4_7xIQ=i^%}x5y~O=2@*&SI1DRy&9hi?h
zLV6=|MgEF+F{MEMBNHFPJw>YKCWpsSMLTD>LgS;zs+#vXFW6gO(ftN&c({DNHbOla
z=-lp*6vvDGRsJV!ojpAlp^9CFjdv8hOTX20Y%kO@hu(|7o`+wRzf$$1c<$|aY7;@z
zJePpnGC7^<=l}R_{{Bi-w>>(j+dkY|(QQ9_S;DEifiBL}tH4S%`uy+e+6^bCZJ2o6
zmtFAdnzM#EQy1@t9i?qR@zgissj-}d0PNqC#IW~<QORR18!my(2PsI@Q#z8dH}r8!
zSdNYF>ak2%V{?<2Z`XTH=)bO9P>n^)nY9l?19bVf^TNF?BY=pui__!d)14sIiKYBl
zs<gTO-h0Rgk{mL42<|j2@8_66`}X&63whVOB!Y?c16ygow|z*UA`0QPz?$`^)TA^<
zGUS~$g{W&*Jg6TgS>BmftT^}d9s6m`v8yu~+4MkWPH<oH1A%F%_RFGSmcHkX4`cX>
z#lyq%PKZ;!Jr1f-R|wNuj6gi|&;EX`x6=kU-aU%9dK&*!hJpEwT^&Js0@EPA8KODr
z7ILxghp>9z-Bkbu`lQ~Lr<ZuvNI!Ho8rVpLDpEh=$qAC*Q7Gqpe?8G!pDAs?hJ9~T
z{;P|tf*NY$HJ{r&YmT+H@Q!~{)z%)N!2h%k6FJ%R$Dh@if+1-smsBvVU`s|pOKm3J
zX3*i$q6OoV-WlyfxV^=@sT3J6cdMwW6^L)n$nvE@BL`{$COVr_@6u(aA=9^Al2$)?
zrlUMmxrFngV);PXZNHS?h?LVocPGctO?K$@hS6$P)TqllirvZS=2k)7(b$Zbs(XJe
za4drofqHag-TKOJL`VA@-}%&!J&u)|{*)6eOYyfi-7O(%vIR}cgvPDDT;1h^DAG27
zVot0{vyV(5^ZQCBU*FuNLh_5Ex+kemYmR1}3AJRPKOsbs&4cn6p5tx4`6is}&LUyz
zp*o=owB|GY-=rjaZ$k>S$tlUzgP!)chIEWy9+?*{Z7*1De$oERm|b%3Wo95JCzRhA
z>VcFAh`n^WChj>@WT3;Y4laj?3svL)a4e;0=PiJN7TtriJAFdQmxNz$K*66Hzt<c4
zIy<XV<oYY^Rx&2(Mi3sAV1v85i^l8F`Xdq@xLwJeUX{fsGCT9uc*EpYQYg#nw0=2>
z6GjGuQu47ivZAddUUKg_4r%_1CFk#tM~IdJ8!o*Kuh=KQjZKR}$YEv~Qu<q6A)h{~
zH7@}XkJZx#M1-B@TV=Es=<?3<5d4I9zpdnuRPnANAP=jR^`Nt>)8m#NHHQg(8|%Le
zr0!JmHy+#7D+QAYaY@dUy_dl!eh5@Y#Jh8AVs%YU$k9HZzcd6Pv&Xvqjcb~!n`&qe
z-Zu&PQ-NuRli_eqdYXN3vOBDbVN<C31>o>AzxvIR(Q?Qi3dUsS#EvRW3ciO-Fxuc!
z&@xMaD+i|ODl{_UbboENzkg_gcW+OPOo>)aRjuk3Op45Tr|WpW@qVxP9`s{V^XSVe
zVWav5;jbj-rl<S7MMQR@2-D|Ki!}wgWXOY0C_`y{%!}Iw-PCWc%l<JS9^V=Zw*&rh
zs=F6ZgkJv}%+VcR;>@&)^O0tJ`OPoDx5g#3X2!E0WlhQk@6fZ^)KpI8{pwuU3GTp|
zOBM9f{=bHPh&%98=3ylExA;0tYz8oCyT(6l>glmB?C!ZZJ&%lX_;5xQ)~!LOo!(}F
zk+im|b<ad4ORrepZBPC12soKD_pi6ooBrTA@euT}qs9>3RO@y3E;93@3jGE*GDSR1
z!z46>-tgZ}kKdAfy&vYKsdt&4Z0ygqF2%<$n4q&MrDP5It!e$#**zVRl7zE%xg&h_
zi9?ZE<RWm<O8k3Jzf7yKxU*MLRljed(ta<#jLv`y<;7W&$0Fqp0(QYX^&jxrqpwy+
z*t|YM=st?5<GrmOsUzIC>0LfqfK&Pp`6d$w#~_KQRbA;y)8(wGu&6K_iCgbyY`I5!
zWy3=PzJ{soglg4O@x%GSVWDjc!0CQ{Q7{&Bc0>Xb9!+mhPP0+UUE(Br6k>eKN5}kw
z%myzW*Es|ZCVFoI_hVqZ?D7>7pZ@sgWTb`I)nU#qFM^|^q9dZC042NZDW8@Ig(X8l
znT))dM+?Of0zpT!f3H@fb4{6yT-4l}L9}MC!X^n8sUca6rcV`KMy?1_e+`PfR&VTz
z$i5F+ET5qviHC7lY?Gs&c-%Cuqq==FpH%IUmQL1Ra#9Pd3B621=#kpIJuF*BZTabx
z(#4O@!_O9$3$ISs&JGT4(ouet6r|FOkz_kfSo$EZ8KUtvUI*nU)!fipeh!9*Bk|#O
z5drn;2yI$(TECzWMn;pk)m3)tMA>klzD*$lOX-=FqMjW9rw%-sW8Iwn$>AbNN8y)=
zG<$1T(FZ^TqQ1%HP|=4S@wvXD;^$X5-oO$LF9O4_4vvXEqa9i!Il9HQcqkc+7T$20
zZN+Gm-%%a@{_dg(UwPkvk3HILr}@K7IIX<jLK)S9N0>~m0Ejsm!W`CGJ*aTfc)VHl
zPd~lK-v&vzLDS5<1=h`{2JY)DvtigNRj97{4Y2W=89Uc`FLOTzL1*CQg}1H>Ne>JW
z&hRGz+^EskUVBko_aaWiz14Uex=DAE!@uq0@2iqD7M-)n$wQy<&<S*X?-T8dgQh2B
zGn2oF+hRRdaOJEQQ503@ZcpmK&d*w7n~rLsy`$0@>@l%mslYP(S7zj)^1cF=|9PaK
z0momI9P6))I&jFlAisgnb0CQ1k$sI9=+1K^3LOod7safY#jNw=;P&qBKjS?(o79OS
zMxx3~SDHXc-Jd*KKkioeIBnFfEH$QHvam@)Roa7XI-9eQat$w%AfFJ)>L3{#ywadf
z=fgqAe7$5VOFH;sZ{%76AN`~&2+M#4zKCNV{Sz9}xB~H$Kk=K$(7Qv3+^I>h(d*0L
zs`vO$G|Fz-VcFin!APPGvre;-&{ifZ>szw`;a?8l0Lk|fCF)KbaKDqj{M_UMQ4f_0
zcVwKH=^}%=ZS9L3iX6Jtx}qb(Nb7icQ#-$+VcP976v)?Gl}g>7pBW9R32y_$2@TPq
ztw5`qYmT0tNcnDe@Nak0D>Fdl59Ya&vyjsRU#$Zcu&1ifp<n64$_ykEj6}j&$|Jht
z=PS_eb~B4$DRhRO&?+CD5CsKoO;g8E+G_;61U1v($DYuqEzMMBfLhb_nXkk3&h;gJ
z>qw99BqT(!GCo}rO25b#2@K3)tv@%>@|Pq*`-f?sGp^DPse?g|yBO=6Uld`c?-E{B
zm{agc9f8%`L^yN4^~0b!iClFvs)T>lbuy%@zRY)cU*<089Y9be?u?ourrF7h?HwJd
zYr!VHCTp*xG<y><!hYC|&8efUp@q~viJt2g<MLm&+M?+K-+cge0GU<B8Z+I&@V7hX
z=n*a-TxrW!-}1<&2s=NWMlH9)U%x+?W+te~uL*(`n(FC}K6_=EMWQ5HFBA#axq^fo
zSL3CB*)`R6oh;q^uJABD=hX<-+{=qYHSxlw8TfDbA^hv@0hajCf{vx1>|oqlc)_yq
z9%w1zKGNb-w{<F>E`oS|;tLQb)Rydx(BA6g&VY!Kh+E3eA9_tDbtjYe6NICS=%|RO
zJqwGT3w%#BQj^hl)?Y++L_-ik1NZU{WSds;hEssmoBW#G3Jt}Q4ayPe`wD+f%#1>1
zf^rmJ5uAJ9aomu8ao?0RNO)mb9R*=i(VS6<^lqKCe4htb@S^~;X+8-2nL)$cqueEq
zAxR3ARXQW;bK!ekHJFRTza-b&LVLp(ar9*-BQdUH3?DvAv%<>IAAFRm&2q7Af8Fy$
zWI(Jm5)mzJEP^wfDlsq<`pp4rYQif!)M4GMY-Yfj)3qzlmH&je_3=3+_x_$8JGWLD
z$Ip+WG5ya;yz2g>qzGM+FA5_=oqtcv^*gwp37cojxS8$DE|xa!`e@lC8{`WbmuJE=
zKu+EEUUo?akmM0_9ZOlOR#cfO$|KDuZA-3DgItnP{#GI))0AVIhag#5K99j19*3V_
z?ffl5i)oUpf$7oSn&hg0Th!IT;gCa5g#mKazWrp06GVC~X1pvjulP!iL!X?S1m!e^
z@N=beIP$nR2=RSKsWUM-!G!poNtJ#6iZ*){xNzZtVMTtLlA~_>I&9dkdyhN!S$V?i
z&=3VWEnZNr4K`ZwAL;ac)t$)VuNE2<VASo(fsgb?mKsugateU4t_YvC*CvJ?8Zsr0
zwi_vz=N7rUTRpEGK@X0=W7<@s)Ksnx*4BqG((({8$JEt2PI@Ew25kIwHO=ph(G2~K
zOh?BRYSW+6<#Hjf_^8}LTxX;Rc8Sf-%j4Rv)KPkl;Q*!z!<2iA_`KhK7mx+!lDIb;
zMo>EO6p^pyBf#(xCB+VIkxWtwOx^JJQAZ-Z;;iEXN95CYV5L$PCXv2^>SOHY%XwaG
zQ*i1fnD9&V#jY~KRg01_M&z@Ce-%6(@*nD+y255YsJ^NLyEImi++Nix`SLxL<A~yV
zet=O%d-J3mlEM!=t5}&?!JXMuyTLFl+SzXV1W*)1170DN(vAqD_|<>;@=?FS3yO;D
zqc1`|gD6p^rgy^PxyR88MG^rugd;sD?=C|UW9O|-iN>CLwM|j|p(+#ZRG@xu^7<<M
z=!b!X_e$q;AA({~3X8E}dRvljI;sqG8=x!Z6_<_oGW`ojtRK~E+jU_#JlTrvh~36+
zKM~~+bS07;3Y<JpcP*}tDqJ}0?B<<+Ek#M?Wf@?`_&Y9U1f`oHDP>F)DR<ov(2mJ$
zW~X{v7T*wbs*TDdH9fT~-UeyW36&Vcw7_!%az@A6tF>!Fl@?m|NI>1Atp#2QiPX9t
z;Jj1KfRSjYDeM<0+dt{ZPbP$1)mh99!q1Z>q$>^rzmmR<jZ#B!EsuQlQk1Gg_tjhG
zlxyV?bSv+^-Uubj85|yv4=RZ5XsKm&j2Up{!QuLx-h;sci&{E|tZQEk;$fdX=s<j2
z(5UHsBx*rHyl-%!k3OrdTcxPmXe+b_laucJi-rH^QTUD>_?sA%)R`0EC1y~0DgH^4
zp(_=r`U$P&*N9(M79{sT>bj;#gzrIKJ82Hb&vqR9%&*QqyZo<qU~#N5B=y^@;oZ?2
zF{<V9S_6pugin6D8D`RuAmiPvInlO=agQD~gYtL51QS}wC~2I144#X4{A#Qzt``9+
z9^1s6lOCl*%iU^fUE<gv(bT-ZVscEg?)?DxW$^*eX8LKa?i{rL6$K*wb}~o-ME*TX
zwRm?f=rp{KZV}|1&+_2qGnRFp{_i2B0^Wvau*6$1^$q2+kd;1<&RQ3R9?0TN5b7&<
zTEGB3-`FB)oS!6t_EYdeiJ+kDam|iNP0d3?Erd?ntaFL$^ZoUf{vqQ8c`j*~08rg_
zR=>djjwE9fQPF@XBkz6v@A#3EzRoQ^e>SHVV3b>KFfmGm66(zUcH61@?p`6t)Ci%R
zxc!>c!SND%aun!+@pMwqp)+{b<q;Ds(?U24S1PDQzwBUoo68!25!XEr%JEVukO;D4
za`}L#=q{c6q*a0poO^?xZ_jO`1b%LK#jpBY5W@e0S6|1k<>_VUjdV!w=r5uDvrTpL
z<3%tB{gv(?1MimcQIo9G`uN$q<D2^LNo3)nRNme1&*u#l>+_200xaul#=LB6u@!c>
zYe7F?PM{>nuFp7r2<HP|DK>gl;$%r-F7p7IX)vs7I_}o;66IDnbI!*EFfLhB_>BnM
z@ISRuiQLx}m3vZ69$S_n&1AQkV-;^^RNl<$KEHqAW2~0dbNT%IghX{c45SGXntzhR
zq@(acK<h?erT10bZQa|8jpANGBoZ%;me8n3lhoSY<k(Pic7~r>gOk`njQ}8{b0P71
zcC~meI}#M9+1O^)KX%>#JTrTGHz@UmAWtP2l__~E3)%W+FVnOyruoqksPn5wvMyQf
z>n&^=11Kh^$bK1oIklODE71}9IvlI>moz4R)&}=T(a&+c!>qBm%<a!aWs37VlMg?K
zTx+nUxI6dphd<v@;nRM^d}A2;qdnDxPj~B9)`5h*(~^yiO^uc`;JvY7cFF(-#owmA
zm6aKwTlGtMf6~jBhar9NRYqoTM?sk@?K`bKRt>8L9Uc!i9%~wOMpEgJm^eN%x+9Xf
zMg*>gmW}K7s^w+u2E-0SWxgxKCo&iFWuTy&`KGKZ4--2O&Xhjs(y3gX>D3#c)z#`U
zDiDR#_*P#&N?Qy7II|Yv_vBylko{D6@H$MechL38K&FC}7IN=ZeufE&XDD1nRnuJC
zEw2Ags888YQRXY^h)UByc{CUT>5D~q9$f7Fz1%w5TQ0gTIs@3Qc!_K2G69+z2BI``
z9o1)Igsg(6r^#UG{RRYC;HO7ZE9um<Q&8sn>U<B!&-{Q+j9^|>n>Ib0V4{_8^=@_U
ze|+!~HdJoW=F!%Acn(oVY*ge8f?{$u$Zmlc)@F|;5x_A=G%I}w%{jw4I*X1%aqJ0Z
zJ+h!lp1(c_W>+;)##rn1gSS%Iur2Ut-nV=3dExaJJDz9@d-e*@FrC?WNK(hzm0xzS
zS(eDxOu$Spk3tAK-E~YgfQUuzzR>6X;vt`hy}3N0F(A0iyEo=yg%7fy8SU@r!*5Bm
z?g*oBJ`Y|5)y@X$ZMFWxn@Uu6Yp`1Ati|ra2B@<8naP+(y@LoEz59NS97?<14EKk7
z+cKMG%4eM-*{I+I$rDi73!S;|Z%@MumzV3%fPT%*?r4?EAwczV;rinIBJi-5cXxL;
zSTG>8<8@&$h{3#!GVi?$giOk)gN{WX@6#?yE2Et*BK0ywi`Z#So6;-V@XwFQ>bSVP
z1(S0+x^a*p-{jMMANm~cY(`KF5@*02IlvVG`WZT*2zt@4EupS^n@mLDIfW8zMd&dj
z5ODVdmffszEQjwII_&{zAics>?RtPqsU82E@3>2bbpBB~R3G<ZBDFF@QfY;A*X(?S
z>Fbc8%ul&9^R<{aiS>qvk5K;m-)>VzV69bb{q;<5_wr{24d4f44f855jJKDNAG~Tv
z6aaB~vepiVT5Qv7zfVh1FwvtWPlyqCywg<+UX5{58otBBuM4d@@!@GYs;FQ^inj3;
zTh|47ct4kb2Pzn7R*CO~7Z&jp^;}&ZtPd?HhY<2;u>Jhe<3lQ%xVM~|5E+_sL`J+!
zYNLg)|0{k?V>knBgR_pwqyBM{X2c&Z8~O>ldT$}lQ0F-lIS^&*8T8$mUnl?<T21zF
z&i99FTFlJDX(thy!)3659Xp(8Z;k8Hy6wMo$WQGNP#G}XNisL#Z7HA8eSf{+0*N}3
zVRJ`tJ;O5k`f@ka;8EIf${H}{QKbL1s!1DVR0f0gq7ZbuPJAMhJ2J71>@YrNnY6To
zuu$=-eZJ$_@CDgX7N6?(z;Qa`64_GSRkY8h7v9^EH6bT_58tV*+SS=rFo6y-`ESNH
zUJfe!FqF=2YT+!hd)&)=^$yEZZ;6k_p*{L1LoUugb{M>HnkJ62TLsZ?{iC<ezsS@n
zK4jS-UhZNlQ+Z#X+%JP{%H}$*KzRo3{Pj9K!WDSn&RjwhXt$bnpn0hycPPkiMKin1
zV5%{%JMwD!=;P6BI4|ZF2qEaIU6vnUw_uB{tekqThBK@x>nx2&2(T%Uc@J-=lNC2t
zdm7$zESBWY{UJuj`6o+Z#=5RaDL(fuUR^f+4N{=-mY^Hzn-AV}qH;aZdSdS~wMUht
zPO5cBMQu0nR`>(J+>W#`0~P0*xxC+UIJ)bf*e88`!y<+h4E4cy_w5Bob7&+oG~_Pg
zfzxWnrxywBxp$O6=d->qCJouGvx}+|=^e-wsv}ZXi_e8UgFt+*=VDghyKf*ylaVOq
zkD%0Szt~^YEQzK)9hc9Dy(_o-2h%~IPw%`U!Xv-lH+4@3dNQ{xD_pe%e$84l{x7>`
z61!pVbpkTl@>9;WvBrW<o)^3Gb6v_+fF28;XjAN_439!rs|nxXTxrs8UTqz3+Je80
zsX%jNe!+yExi>uu*z5YIQfC}V7vG1R_T2HQo{gw$YFZlcx#P~0Cfw*`$uS3>XEK-x
zkC~#96WY-YS8*UMYCyT01bvXu-{IH|WPap(64r49itaarm5?mU83{~^$WVpB3Hg9$
zQ%!mM?v?^_CGOAoWo?Xhjzd<SBv7DAGxm$ne~ieLfRWDp`4+W8$rx}auv-)o5)%7D
zk@i!jbh4k`Tld|iv9Nse=bSC!p&|)B$oh|r6m+??)|Q`)2J@L|T$udKVk)ZbaI#V|
zGBightINPz-^#N~**bkeKE$qJ9U70vFRZ0*o)9`ejBzcd<mic*>GrwKlO}bT@SIJ2
zLv0gKod=bCo+YKmsw-HYfspzsu8a9G{TONT8`^?oCzijtct39GSY~I$n>s|&nzq(1
zwX$iZp8nJFIfB3@9f%xN#PObpV4Ql?3O1*Rk=^3h@^R9?w<PhHR!8yrY8~SXoE=7o
z6=8ooBCxg3@Nt-YzaXMjz$fjU%0<D7H-QpS-ro$&k|0TbNy^n;M|ubUy{`8aHP;Uw
zCEW>nIl5bOe|p}th9qB9kC|uC9fpehFzAr}lx=N?kF=U?8{0eDIyk%};-}W|#09qz
z56c@3#U>E`RFv9@p}|gc)y>6!p_%6K3gxbooo!WZHuQIwpCqHCYB>)5+`lF#u>y~W
z0gl+k&wObu{Gv?!IM>T-yuJJqxaIYg+`rnDuP#$joQ5x{c8d}arF_m5OdB!kzjtT5
z@~U&G7}&-odngj)QjNU#uFn183`VE>2d4)FlKB775NnNgv%?DJBmKX+LCnWAROr&o
zoOSYkDEj28sXG<=4??_N^);<mIOK6J3bg-d81+dUWmVm<1~|5<=s-F88Qg?=OwJGG
z39UB#uE62P%s$13Ki4DeXRAHkn?*4}-MexNRJ!ANh#2|gW!Ci=DZ_O?1+L==NVWul
zkczJR->r+}rxHcKct|csC6=S1L_!>NJtRjmT!zlfGM6~Ut6CO8_IJ^YrsRr1?fagB
zEG!m7#9PhykH1i!z)unICqujw62AI7akF|5c<1t-K1|*QdlcE-StNlD<5`AlW@e@)
z9j>-spB`MT_7ANk8BeX=&XDSa@anyuYb*qnxR54Hgx43}wNpnM@+#JFxTK`D{sX@`
zbokw=X4!j01z~!x2;S<}N5EsOs^GtUEFA;VZ%j8VYB51pv$IC%5Td8Z!F5;*@pP-(
z3SSXqA>)AcIaZAc5tuCx%G=CIMoJ2Q7fSj-(TNc;PQYX)7)I2Lz~gmDUTiY{PtVfH
zT!=5zyHv)#r8w$r28R4(k*y3L#5Ia@a_;+1=3-~IzZ-TT6nTP&fPQl(r&bTy-^m^V
z{VDPMegS)SHjHL@Ou|Af`w$tM4zM#bk0|x0xtRJY(8v2jwQQ<?{ZeFmy&0T1aqLNk
zRK84HozGt<E-mu;#ZW+DKF$$ktKT_hUGB0O83ma?G2)Y*!?{9Txswc8;d04>i{1V8
zF9rwsP_uu+t(FUo0SV(el7x7Ax)q12KZe*~1<CX&Rtj_uAw9O!Z{N?%CK<rtf#x*>
za54kXq_sE38Xfjn@2iJ*(65K2Alfx~3|?7SS8&s2ZPmCHl{6bLUo8!e@tCi{m}2#G
zdi&s2ndVGTl|?nkM4g21tVM_C<3Wv*G^v>^`hirJ>1@EX{oW#XDsefQ7&*t=4MgNi
zV787(6P)nqp4s`a=Zk*Ja6Z4%L+_T@fQ<gk*F+CAtv;Di;?r$RHD(??kG_xIt|1DI
z;tUcR4b5GU9pG1^%I@77d(rd$&C|KS#VH{YR4UPvQ2Od`h&w;{=f*6zdRKx)Q&V*t
z<6<ZXkbAk?RV9J*Y~j)96O<Y`CXzMBGJb3LR$JC@Ze9~|uLzBaC`i5BHJNR~6i;qa
zC{*t|otr(T4GH|_QASR-T0iL`&i?cp^`LE>b_{ObMpRv3T%kCKTY`s%?`LKuN7&AO
zu!ph4=H})Jh8P4WIDTtrL}RB?E3upC6O1j;p~MII>Z68R=H*pgY<VPs#o!mV{s%&$
zqI*j#_2qxFrm4%lBA#fHDsi&heX%Y9v~_goXY|LK{g>%KZZ;NBgH5QzFW$Jl`?%KO
z$8Q(|u_KAdSVR(V_xJ0U4Xr|<!^)9fp|5U;YnLfeOEoaZKD_-!{2@#7?f%hZ2tZJp
zz~TuQvm7RAZiWtP&OU1|`2gRm=x(NrTtW8`o2)PWC?|QrRROiKID@d<;0}WfW%xAp
zppRgA@76{4Y<2hFTAgguUH4Yq|HxnoU#_b-7Fh`^3=!`X$xc`|G2z{fm~wjwuet)*
z9%F|wV+il-BoW%x><y(a`?~z(00iU$jqwgsT*{<nW==`B$K%LMet>g8E~584*$vaE
z;N*VA5K*7kyOVr9Ggb>>(Uw^-?eDkLhts~Ma!+#KP>5X-4w|M@I=6lU?N`j_r#Mu{
zGesaieHEtMukeN#4Ly!$k=zaM-M4D5^NIlOsYPEN?2`;Mej%;Pg6TNG=Z>3?_per8
zLfMO7wCW+>mKNHlgR1Z}JmoCLBBPo?TuaO^QBU_~n`WA4ts{7??|^>KkJ(uO-ZR%D
zZ+Qbdolv*d#}c5|zgP^C%{DU0>TA6o<7>1A6!w;%l{-~qlv|cI!-6d;WPRc!35^ST
z3<g!$=;(VrUHMCe#=aB49>#V|uK#kUV}TWK9Pkel^>g<e|F7ksWU8l4^9w=tv-4gj
z)-@h~(?4{7Y>?YpZ2kq#mZTPRV$Y$Iw_TNtI-4-OZpNiLd|W-GrqqBW@zMAw3Ufff
z)hnK)Xfg93Vz6vu!xQG!;_d?;7(gXUGE4C(WanC07MId|jvfJcd3WUm@bg)1;sG|S
zV+p1e<|tdROIkp$%5)0fDF0jRXUU4C_Ed(hhw``!`S_G?7DM@uW~lNqpv<4762KJM
z{qt+zch07}RSJ95x(k=<eA3JwHjy4)duoM6_N3D|V2689cw~Ry@8aUDG^uD4kKeky
z44<j(l`YTsFB8ZQzOSmf7<J`$vei#w&*P$dCd$if^+w_?D;m7or<;A-8=H&%?d(5$
z>vcVKCtUO5@4?ijaS`e6=?^GxARw>VlxcLBc9a}lG_#qF1-1S(e##4uVR3gFhbE*8
zX(Q<9im>ZDM`JirG4xBwGLCz2#6D}A2h-k8>=acL1cc0In&fXR?_h8r7kNX5G6+5+
z2{^4L%6NoOFG5sgj1W0<Chfot-^`Z|U-b8q|BwKJr%Tb(p}u5A<Exu%$5K3xTChi#
zH9b9bXv|wX+NsN{T|)ms|MQlKzLGU+$xon_Ie&U>Yi6=Q!pF7_KWDC#grz1cg&Zm&
zC-zZ*fmZziRm?#}*}(;!5t#xK+N-vE@Q&o1`quIq(M$D|k<s6OIkYO2pv550q>T)u
z<&+bWyA{B=l3}G|0!OQBGWx%JWZ{iJXXXbfrtAjsy}fbljwW23?hIY_lg>r~47ww!
zBPnUI-ChF2CYGI-z|4Fwt7@@>X%Hm@bpy7a*RHO9yH~g;?;R6m4U^~0-&^^1o$PL)
z%qwu}1&y<PzMo#Jt!X;4ewvTqjqy+a_?>F(R&oNOq)dIgt7rn<-uxm{YAQxYkw4y}
z?&dSA+4(HD=Q3pMEl!<dQct@@Yi3)}QwyX`eLXO|W7@*Sa(T$=tSETlc`WCVi0?f~
z8+eAJ)vPME{R1Yb&_+JMoU-Y+yz<EJxs+(i0l7n1#b|!cX`<*~&O7^+u~$XM=ingW
z?i%CORzY-tHMuz8gYJp!OhvhQ*EZKibvK9kh-lr(rlETF1soC>IsX5GGjqwk_T}Gu
ztu@iTC^oOw_p?h8zA&i-sQ}-uU!o%XF*%wsX@B&DuspMQk6#AOVy$|vTCa{$H;0DM
zOHuMW&`_%l(w$_Pw`h3++tjqGz*6Bc!mEZ;iv$#|F7c$joV2Ig_!?$c+6D_Q?^Jg5
z2Lf}r<u@Idr0H1(^CYXUyyYVEq<QDB>FM?`?;8|^=HE;`jlTh>(g$y^50<a%I{L4!
zLbLUf_m_y0fb%JeN;pQ^ki2*l74<mWtETqPHLv&eJfh`!;7(syHQ6^6nKIY2AH?fx
zpjLh<idPh`?yrEGqA#vKtXD&X|4MZqko#jY>}1LnM4tR%w5fKTjBn2C^IQJ6lqGNj
z<l~<emzRe=)n*0d1``v;F`4h6FLmOy_Tr$poo;V$*{Z{{sl%3u&1(YYiKA2uh!QE|
zZF9091~@YE4OqXwuTxcS?tY9+agTKX;xi5I87{&oDWLHB8~5jSmX_)+!zYP`SmIoZ
z5QyM$wY7e+brx8Nl<eKg<`drk&M%Ux$|4=rOOAShK#}wH!MWh09~ix8%5`qC4#;Ba
z`5Pc+V#=~ra!?S|eQ1Q65i_uAbB6zzz%tBdjvzaTNxylf0oL=<)04SWe-6Oaf4%g+
zoqsEPdq7?E4eGce<zL6=zkjD{2E^#6GRRo{_tDp-qh+VN?qr8xT%>ya_Xl_U(n(jP
z?H;tn?K>yZf5~u!!>M}q)r&F%31bO@E&F8ohm#0XTgwsCY8J(Uaa}PUm&T<bwp6jf
z>yyXrwB}-H?XT2+A@ITq5g9EYQE7#()Ypbn<=TPCsf;G`>|uEJ7Bwxw)>9;QSDnq>
z^SC)=|67Su{nBv@*2}kl#w_e##+kLDc}1};!Bpmi`v|gW0m!n?!cp|H4W|nQ=<QSS
z`-^58M67`jd=W;L+R4DA36cy6iBHvee*L~`A{Akl0eKOyGH%IML1M?%Lw@K4l*u1C
zsVWbClPs^#`22*2=XcEOr);~4Jii#KRojKCVEr)u79Y?jNf_F+|BvQL*Q@T61@EG|
zo5QiAF7=io$`%@|DNtvnMN9M66@Dj$!h$~?iB=0I-V2^0m2Qe+m83fKCiwsvv!Y55
z@tKYFBmd=)Ih)8ZJKz$yJi;z_M*>}Z_3XC3*MzHR;`t77J$xbJ<in&{ShUjR<(?W^
zxAp|Dc6Ek7I;?d!tJx&|JSMR@HRYi+u`Y}85WjEopv81-+R$u&`FLlzOFfsi2u+!7
ze$tF<K04Ooi02jgivoxX%bu&I3Vnt%59GNOg9Af8#G)+bfLEoE&t(am%!v!i3G7N~
ziW$tG_#iK!uy5YWElbBi%PsbvQQ=pazMmEgZ4}tb9w91koe-~%78ZPg&Z~>F3(WF(
zd0F+zM*%W6AEoKhjWE)?UWDT_+*IFm1;^ubA4FZoDkPo}(garL6G(V9>A3uMufDFi
z4tLzl60SG~0M{3d=Fahw7$Z&?yl<y>q$4cbS%DnFZ}M}_%Z-lT3F*A}Adnjzm2GX1
zVQ>vihjYbj{2+p!mYELoPR{Xq-{MY(fX=L6#z2JBDV#(#U{Z%OWQ;FpoO!-}CjSPe
zmKA+V|9D95r|cPllo|dkwu$H){iW@6Lw{B4`gtIbu;N6shb596nTO#y=<tZJ!cEG!
ze_8(PU1X0*vRY_aFzmbAwF%Wgu}((Za+lw$5$HA7ud+z=-|5wi8L;%cnf^?L*L%JP
zxOKh%=9MT@g~u)MXlk7e5=E+So1G3#SqcaIV&R_=C|;Law=&yi%Z5NJLzCalNx`He
zxmo~yLQ;|N>nd7pJmb^|btFx%!Kssw;?qO7>M2=6s_BkqocPI3bC89K{VH27P0kHQ
z@u8O@20Eb+qrb50Z}V)Vh5uzKxj#R5c!N~5Z<sBzBR<fykC46jwNXg^6WFfneTx&0
z4t9qyaHve`*WU2>DIIjfOfZ1%3B|t<Gfv<gd&BclYBm^ItW?>)ElO3*qHhl!@ov|q
zXZeN{AGoAwLi%ye2oe;Zk{>tfdiSB;4m>j(l9Qe$vgLk*YfLL*8F%U8&eGQ6m^)Ga
zU+VL&SIgb@(U&_I-yWcCS&pK!b2*%Mp}TG&+B~cxthRD0P-(*S>^wTGXUPZqJ`Wfx
zS1v#kq}9d<Q?^D4?w^yLokwiZ0@9L;-fU#Tfl9HI$@P!2Jfg!$j>^M^JtE_goSNxx
zutVX5B9o+y3BB2jJrYID|F7$Cu1xE%$d`PJO1$<eMYcr06y*y6KC&Cqx+PtUvnNrc
zX7Eoe11y+g;>Sd#z-}<8xPNJB3AjERYB4U#>bbs5-VzyyKV4;DQ&gm%qAOEoy0iEq
zErD7@nw?9Nzc;XzpSmFv&QqH!ruf3;_E|9-7uCGMNn(XW<=LJDVcEfbw3XQcCD8e`
zUu<zK+w_plz=Vx^PKLZpM+|P#nPp0&KUA~azN^*r4_NXobb?(X)e^bE0f4hTMiVtc
zm5<YIUiGs^W&UyP61t|PE|JV3DyUZ;@ageG5yR3vLoXH|FfN?zE<7`huM!d}L9tMO
z6a}XVk}+!Dj5J^!{dq+qY}q^XbQKgCqp|<+-*{_d(oH4MTfjHpExaqnXu6TbW6-p!
zTJ`z_?>#y*v3V$MR%Lrf7bqDr|B+EcnRX#D!~0+E<+q#ZI!<4er9)d;vvW|OK7*5u
zXRZ|<q@L_{t@t>X4gM!w=}>ePy!d`LucAI%45*^HW&vGSw`||jv)mlI|NeH%a@v}@
zUfd%8M<;PpJrUdO*`wKW;!>R#UW<R`>5Eh_gsbdD%H^x#Aw(iH2T13wGho!wx~o2~
z$76D6p!C+(ghHJUB*0SAmgRQGY&9kAP(6TcTdV7R)tNN9e1w3IsN9k@EH{5Zb(h|e
z1B8I4q$dMeytka^on;;+1mJC4&jV|bCC&7Bn`LZz=nr_S=zsgNGwis-hk6$uZ_H`!
zyj+dZ9M%2y7h+K!1N2ewoGhqJ$p1jP@KwK%MC(!B-CeH5E?-4nZtgi;Ul8(MUy!+E
zS8;3JgI71@8&|rCz!&F_kAuUAR~yxpN(IIk!qk?saiukD3A+o~KWU0k!L$u)f@>;_
zt+2QAR}eFm0BZ%|?C%+26QOrWY&l>I`YU-$ak!XOhYLvs)YpZ`QySa^^T+ZO{nPt-
z0DQyTdEJv``+1BTb?ZJIovqa!8FV`jxO-*Dlc`@|44bU1Nnt`*@FpJa93B$4&<^%c
zT%@Y8&Tr(f4#G5`#Ew?_Nq?b|w>HxKLE~iqOw@FS$o?~*tvW4HcLk-!8JDzp7RdL{
zs`f$`Tehk|(XXmR2HvMey9_?FM9^@6PubsxF;9?)ekWXhuM}AytVGg949teqwo}PH
zE(y-386DR*w1j*#nT!!rvjC|*@{4@8mhmpU)innOi&NGogS#M3M;8g`GCE=Av`<cE
z6RBA-Wf3WIx%Uh7LJcTPx4Fuw`VTweTzz{rVw8Zs#KTMtWY-CP9Va%wwA}3g;t$C}
zkij(nZk_2`n1$?CRBXG$k(QsqF!M1xl3r<`)zN-~>6X9o>sXx$U;kyB4E3vkYEGuY
zc;!X0X<7KtZnol0Ewktxjp5L6CE8CpK6n}w=&27I+v{zYs2+aDnkfQC|4oK4n^1MR
z0nqeXZfSWBxVSpnIg0v|3pQn3gp{}6X<&fe%Ibsnfu<8f7HYkEb}pChj;UUl40Z5Y
z$OK66cNMQ*?SJN}?n0&HjEW&#^&q`*urEkD^{<1&BmMCft~r&JgsrX4)?m)kowe?x
zE-!h`ja?%OGZACVUukr}8+a_L%?(Zm{CnHL;wfu&pD11Y)~gb8*Yrw^4(QG|>Bs*z
z)ZLfBJ>ZN;B9N>HEF(^I+BZszl5Yoan|n*XWzj~|-uHA9VS62KAgV)3GyEZir-SZ%
zzMmaWiQ@jMGF!VWega$&Flqqtczt1kxbvD(KF&=PLd7(#n>E!bYSwBmfu6Jh_P9mu
zjW##k!j#4;ry8dl%&lmQDa=aTa<T|v_N*diB5s8nZiN-r{at$t$`Y9xeo%ISsN7A5
z=&P{Y4doY?dDzsg-@JX2jiR3m8%xPUvZJyp4~08;(!;7pk#0Y0)@ILlc29^)vt944
zL!k{2KgwKinpMpw9S}WzDr#0+`r;tl0HiK`Cw=kx-c>lxxbmJp)X?m+|2qqmrB2!O
zou;yXZ~_@05J-gPXi(uDbG<&#c!s+wjWXV;I~91l7MmW2W<1*D2{{*7814Op^rSIl
z@J{c=Wn`}0xS|4ohH;ab`b4BaO?h)*?p7)&rjWc0!y9lq(93TnzI%C9MgK@xK+~;|
zol0eaN?eTn&jxedQlvl>yEedOa9C8A0Rd@X%67mOL>><{B>dTp;g9c!_pPWa^L#3g
zF){l+9Pxf(DXE&BS@F}h$xMuT8DRP=TjLC<9uWZDWPk+5#vL0}ZvDx^OZ!_O^0Sh8
z?y$i=zt9<gd%OigEpdbYN7H%6v;F;TJdq%Fj4G`Wp^Bn0TBAnnQALdwrM8w@HESj`
zh*G<D?XA=(F>3pv#Hv{_+nTA-8nNB^Ke!*{dGh+6@5wpWb$zZkQb6dImT9lL*)w9v
zoL|Vr(FM*Lb#>HFJZ;aoq4c}%#7bi;<)`)aQ3yI@V>(dAy49_xMb61WO?>wPQ1<KX
zy^nEMYOyi#Q+@fxayyGme^Hj1RdV$siTsR>FS}PO_atXZ_q~+FCQN3=31e;Vt4nq=
zUpGNjP0|OcKZ~Eeya|2G@tTD%8zX3|&z(FVG42iWXRa=Wa{$>2y6L4ZFDB@hiGVSn
znym@%>=WKnY_hCiy6Hm|fdti+^e8v+hE$^xnskNwZYje_{4g4tfYPRgWkj7J>9u&P
zJ5a5cyW}9?887Pm<nZjGJ)`)|w|LcDm#2f(ET&q7nUK~fE#{20O@-0XA)n`4l0%kn
zL=+^3_4ZBH$W0^8=BWYC)r(Rkrz8znkpn7Rs_I~wiLkc`4w)s$NZyQv^D<iz2m;b0
z95I!uS(WnK!RLQCMhgKun7E_)nq-tHXeWrey^_@6cEd^7z!qEt`wJ+dHKpbMJ0`so
zwH>6OJH^MX$zrSir{}J}VjC_XalbirQ0|q36Vkzf-8RvPt_8Ok^a`>8ztu2miuo~*
zpP%fO;F_PMp9$)Cesnm!beBXVk;rCEfJm-&>eZRYdNk@|SEpuEi!b;4A{ZcFD^H_L
ztII~fK_EU{&sl}yAHY-!!YB9K3@fav^RiXzB^u!8z42CEzdPBazGn?n!&~(PTn|-2
zkWH*!YhMKb7lK-fx4kc3OEzi?8wFGnmi5~ZWc|&y`|74zzok$4hi}02A_x`C0ehBT
zR%$sUh>{gRTTJYF?4mCJ9{S`MtT&tR6<G?8NieQSi?nR!k1L1Pyb-!&j#eljn&HH3
zr~Bc5(=AGti?Q_nOvj_&?!-^=BM{epv^tH3YM%m3WZ;?WC$mt+xF@NyUI4pq{kb$B
zBCxN7Pd;0kxm4}clB-GFmpVeoRk$-G+_NQ3l$k;j)tCgS;P`P4Kg&6BO*V2zZKj%&
z{wClYa(mz)Jc^Y8;I+vZrMA(Ndo+9dO=nNkVJ9E#_i;`J^Q?mDc(!S1&en-EIS-}*
zvny-Jw;Bd;kMUE4CAtXeaiJFT<NT%IZ28f&6Jhj&(ejAm$}`(zNFPN$R3AzI1#+Qt
zuE0D_kYDs~XnWsn@$zoV=fzP&TI!}SU`5eeplbi$?;pss;h8C&U*4}Sh~kY`utt^7
z(o0Z%tv}sJsLPWhBJUE4oc1^jOe4GkDOz;sJ^CaBHxgS_E^Yw=Ao0j}Pn?9NhuZ>a
z+^^;_V<slT@hv+W&jLIax@Ppz>BZx`wKpE^cQWIU0>ULLHxjyd(V~{R=IqrCn#UXq
z(ggU(BL|oRd|_2S&C3-egCZ{<zmV32&oZ95G1;XS(Hsi%Ze?+{sX8^<=;RtLvx$%;
z4Kfq}o&OPAVj*pN>suVXp+BGoZE1aX;x{1ZDK?hA8d}X0MD!)BQ@SLU?ng%?mfe1|
z^bwkF4E_VL-S`XrS(K}&gs)As4+!L_sypJn;_JosUj6sB9FP%|CgcX<VjCrEF#Dwj
zWPZ2~kLADeNy}-gr8F})kPr%`?aK<-y1}d-EdLR17aK}&*U)w3vojhw(US+8-cx9F
zYqb=2ksqBHD!ophTXHh78Jy`94`aSZCzDA22jPB))r%!t788G?C^tX<<UpCHjsYhX
z>^6Fi=jgu++yQ5t`JbN-tSukwvpTM=a0-U+rR~xxg)%s78}JFTC_P)`5*`*1fY~a2
zESbobq9qWj?OkcQ?envL-P9TJD=~Vd`6f~xN(J~$6(^@wN&%dtVx~+WyZ#^?TM}L5
zEAN{E4iTndO`@#)qiBWGmgkF_9-FRBQ%@W&2x%7XAQ9B1@5TOEU|W1ieiS<pzN)hv
zCg-3j=pdj;-`hN0Jo&%hqrWdb9D(5%qDKSi$D`lJ)D4Vp!}%R_8JgTmj*&84AQpB_
z3aNcyFE!?9sXf%kCs{v=t4qt`-DwTNn<+nv3!L5k!<$7G@D?b5T^M74Lo)`$%_vmY
z%+b22LpK7>jkoaqin<{DuE9Fqspa|73dDdqxF)-ggV4UTLv;JN+q!W7?1yjzXNL9Q
zK+uwJ>~Y$VVq=JFtCF(vbwMA5y$>yV+|&Bk(c+Q^*-Z@zY9)ujTdfw`;?pK1r&O_p
zm0^*UjN26zn%a=Wb7Gr6{fltG=nrWT7!}GE!mHe99J}J2{qQ;1&V(=KC(tF8={^f}
zFQQ{<CSSw|po4Wu)OwmT5r9^&Wzje~h%({7q>TO7B(XB~y+2XHS5*5iKW_T1(W1%*
zBYO>#Khr&HORNGM|EKA&?m9jRqs<gq9TdyH>w()P2iujEl@&k6XlFQNycoIFu#T#R
z<4rJ$CuJ}7GJLzp^dE&=ygV;cK2KqmO)~d5P!A-hM7ovzVlVP5NpYAD(R9GPw_S|P
z?kgMVAl%t@iAx(9bC9lK+8WpEqQB4OF5ueSHPDgHP3QSA*XPqN@$fJ(q-|@AZCasS
znjRT5hE-q=72;18pJUL>0|H_EsJ&Y8VbzbXf=*}CdTd+dcjD1`)DiMQVTB_vzUr+7
zC_<)yK%~0!k{X34ZZR`XJ=dY`(Z5R8&iAX!I|vZjnFPeMDWs})r#}i-O68%_M%z}b
z=t9Ad)6-#=%nDrnhTtSmmXgkH>1#huU!H{qv^spCglH<{I%XC@XriI0%2zx)oPD0V
z#KuN$=F;+<)Ci~>y*Iyrjb4Ohd)U$aYT}!=oisvImHkr7ZE8*gQ(k%C(8nWZ^PG{=
zfQpiZD)_Qxn$hjVaC)j9)2Xh!f&s}~^4T*uGk5LZ7yQ17QCdQlcSPzxOb;cuAy2#}
z#l>IiFO63thlLS5ku6!Mz~6=+Ce;+@73Nih?9mcZG2;tt*^K-ZiMu^Vj|kNjVZmHu
z3FKauFK8v7rK}1H4C3S3$;AGT2AuCB+yBBIgRyu$4*P{h-<vqoySVnlH(qu}e<U+b
zW?d^7A6^Mi*bv)fZe>=0<<S8+$VAA^8IoVKD>+WNy?u0XczT$2`5GhzY)_N)0tup?
z41E?GE!)`>iaL3R1bd!uY*Kvdk~kdCXjnyXsn%Dd=K%#2;W7XHQc}wU)g7jH3;lV8
z>X@%y`h0fvBbVHOKUjxf$q94mRxn5wxhCBEGhZ7J^_xFVR20q!7kK>&(B>~F=tQB}
z$rW$7_Lb(UbTGJg;5vPM91xpvQ0$a8;{_52j;2w0lfTO)%L*^A$&thPnDe^B-3Xau
zX!i90WB~kw<p3sJ2m`HdbB>oVVQNkU81cXpSsp)%NRWWxr>G0=uow#~sMu1J{MuUv
zyrzH6+?inrHL6YB$;f`vP|fV1?^*<8lbtfM*sZb>=RZcuNjyG|w#n4(5;}>5sPj>U
zlM$?EW2mnFxn+pLSbox*AOL(#b!XOR-VbHH9k`1-Yu|0<75HF(_@7RwY8gED`SYBS
z+srpUns8@h>5>3ah_do$c|g-XNrTbck-G-DBm1SzBM}_Iv5Yy7>Vqm4T=?ML2&QOl
zq-o#*<JG^8kPEYX5`UL@{X3nN!d7~&5?jG^Xqt!*2!`VI8m3;LZh2bZSbe@<N;vbi
zIx~#QGr_4UZLPU4wzMzJ&3NpQa&t%PkqzBpVQfz@Ae(2nX)WdM#BU5*<S4?;Lc_fK
z40IE-q}LtsBtmHOVn;i<Z|7*648p4ro$$(hBR;^<#x#+idh+)jhei`iaUUv4RbWa5
z7?uc0LphF6<E7lWJ~ExF#pd^SZRB$~hU>Cu1EBq32@0YnJyyf%9Ds>piZ@{U01n=%
zw{Zi1dtgjI&iN=<f7qJR1b@M3l^-XJhNRa3IQZ`IFxn#dp$znjt#rZK_u?ilRn&XA
z&b8O#fMMZ&OmynH+AZZS)c(08EqHH?R7@mM@M0}Y?JyMex%j{lj0`_7OGx>-Gv7Q(
z=0xSGmkd$@O%5B~ujJC)eL?q7huas2Lu8%yY-Wf+$)nbQ)Z=u7xr8<%0V$&po6xQ@
z-~zIqDF3hU<oE)qondw3cy{@Fe$;;{p%S;}Bhxok40LNp9vewg;+P%|GqKV>=3~Qr
z@o>?v15pSkP>hpXDcFD`Fe$oww@r{*KHbqVP)YvMKT&|hF!0}mFUF{pYzI?Yxj$A8
zD-}C!BxZOQ*8odr@}9{_Nd3@RRaT%n`3}$mGlo$#hv{(*fIQ3mZW0=qwi|fxXA<&j
z`_7u8c!#wC^-ea0;3fG_L6)CM`0q8Ajsp0tm#$yg$)fXBZpW1o#$w4A!4*f%*b7T(
z4vYfWt!y5gA8eN9qEIKRjv3Dd19H;8o^c>7lumjgpMjZS(0aMDi1gYA0kHMj6VSsm
z4=ZKE4RRHfn;KJs$Bj#`&mKEUU*3PTq|3jCruc1R4*+a>x~h%06qci(gx@#GTEscc
zWz@3Urj9>qi^|AS+xS=?wmN!ul!J6Or+L3Ku|Cizemc;j1+{y6sh{==rSvFvV=YoY
zz1(#pk1#hi|2ER)*|XLL9uP$jvUPKBal3s77ZlX7n?~FwFT#atrM}>bQR56b4p9Y@
z1}1J7m+Cdl3bMB|PFu(bk_bAoy<zKaM9Zi7y_FA=&&J~f=%+o9zyRrOz{o*c;tG%U
zlzme`o;(62`CY{`0z#%Y<^f;>d)@%_q;T=%Ev9*N7c=7ydpzHR&Mw6;tweS8%VTRT
zDg{lj%a#|UepMG{8J7g~HuO7oR~E{2YFwPgd5~3M(rB*rc`|Vf%P(+`Guj;rh6qfk
zGcpb@kfs&5pr#)(M%P(?)&{?ceb>`nnKE%ggJPa`+nx(uPQA*ft4;a2AjKo#7-_1z
z(v4iv$G{zPUYNeBYM2h(<>|t+^U&_lGH=AKC`D#Tc$Z&`Ua)Ul=pbJga<9&aIOUFN
z(v)KPXXoEt$ODCFAl%ShcDvRqTCWuhfW*SyaUCnL!G4UO(hcg|WCO>e;x}d+XcKRv
z2880o7=TWxMWvZw2s5V&0@uq1T(hA{O|XU!84GY3y8zfAgIWK1E7HYDtoGk7-}53D
zCfe0v1v==wh2dZdU;<@i%bcNil)FGlM6HQKAN;FHCN2E<@Io=<)_s%KI3<glLrKgB
z*ocJ3R#6O{Wt2VLk&SLI)MXyjE%2k@-Qi@qr_<71@(HM;%8fu6dyR2j-Dl<8z6L+|
z7ks*!mb)gRX8hQhS=7j2ecvmSD4<;pgy4%|I!$jG{aDZxEB5h?fdq3}n|}}E?p*ik
z2|MkSta%Bg0lcv3{NFCwJ$ZPMTA|?J#WeF;&(AdEps_mVv2gtxjhoE!P*iv{cg=&a
zLa@oZ=H<EJh6XpQnFr{!?g?qf{u?$GYqVfPb^RgoY!%Va^nFRDqB8)XVe@K^JbOWK
z_ei^N7EZ?Z^&t=`P{0_s)<ajwTy;zN_7?4ZJ9^%Q>8hofi+}CUcM03=<mSA_C9AW5
zq~by6pQ^xWr<{Xd)z$2oD3&`E5~6x&Q8-!~*n>13hP&L@gHA!X02s&g`mqiZ4FSkN
z8=DPPh(c;I5{qQZVatZm_6FbLdh+_g0mn(x>dComWvNm;&BmBBW;IsuY3Xr(<<K`J
zxMIoO6MGe#ItFeo31z)*{sN)eO!E-#v6b1h63?C~+*#;}-^mfGA2*tts~i%#SOM*N
zYs<r4duWvhr*jZ@Jd`x$=6}3fwJp4)+r}VO8g2b(N-wWStd&0V{|)9IHVORI+el$u
z5pzrhEnRf>nuc~~;Wh(vL_h`W&zJE-Ky`4+ca5hQ&d%YuGVhyGquND<9oFt-spMb(
zXO!jA;u%VbWeST;&>=vgJP&FZxY=}MBiE>-b70Uv67FHY#Ihyq_R_fa6Fzb8q|Vnl
zCCBM;q#|{X6YN;VY=N46AQy_Hw^CqlS@y){c`YuNoY+^XmOs+wt0VcxZ7X_g`pQ!+
z<kS`=r{c|B%giy<*UP&<Fe^^tdnRjNec@Sy{Y3qAqEA=2htYW{ZW-aeOWN(}+go%a
z936&+24?C%{J0MP(1f6+l9$xJbtlYynbBP%ohhZXGD7YE>$^XLyF6{QswT8FPPbNs
z3~>C5$Pk+`DJ8qhxz;(|4djl<AYPT;h-Q4l2<vs15MF&odLj4fkeO{ByMb%bg-dN!
zWhe$`K|RN3gUNYJ-0W1drZLX|>2d{pmxgLrYTSywtEbN&W|W8sRNt7Rv!~{rwv`^s
zZ@Xta^&NbNGeUxr74B67{V=FN!^X&0ia<;)DvF-x`5`7YlqSy*=!80cRmbNpQ{}Ck
zW%1OxceqTgqKKX{O_tNf|Lv;}{f^+Wnn2;BR9K%1M8oLHY#Dc$7eZoZXa1gET#&^P
z;;LWUXR>XY&mn(Z@2h2!^F)PI-x&RKKn)Cyd~25<Mt6-F2#^_73;FJYmeN+5XM>yd
zf}%8Jh`-ESDb!&QLlPt=e>y6YTG>cw6WW<i?U^~ti~;-<Qj5m++w(9rlkWMWoNS6<
zHv6tWl|IQF&u%J;0O7w?Z=EksQ*d`+C=5Y>aH$RnG;c9T(o=8={P%jukO_{=#z37P
zoqGkI9AMatIQa1IvBQ5Zj-Ho9>u#z&{Am7r-$42OG`wIsP(Bx{jpeJ|-MK&6*~)Nl
z*fY;_>4^M>^FHXm?0m%A)<BJF6*N{XaWv;}{*w42ZgM1u9C4l&F^Y7)*`vt>5QOGb
zAXKKtE43*O9<^1Sk;tA&``7l6jw4(4@>z#t+3)|UiLEa4WD2hR`7i!cB9i1l$D+&A
z7`JrD;lzzS`9(toK+1aEduPKeuQwFW;s@4a)>I$-*d*D)@`msCFLqtjy}O3@gE>c}
zT3V%mT>?zyvdX`+Iny3<8ju6wnRJD5PEc}rHjMz&J!*Qkq~4R-Je=*yf(`GuG#mi|
zfYhk@jG-S?UyB6i46btT5E}1U;m9t>>ABbCL2zh*Mb3!#$0ide6^Deu2{@$=4J2gU
zEY+eCMnyrH^j128uz+f}o;#eIb~~tUdmsP%;Q;AQ3u45WoIE~R-NZw11ag@pVTz)-
zsd>Atp|*0<@IBJDq{eAn=Rkch1i3jPal_&T;-9TeKGbCRv73(+9r!1}Va>V1Z=;JL
z^moox{y4*d5-glX4TA4Z>Eev-*4#1~CX+`bAYq+gH*p^eREb4IbXYWyRzpSWl@0el
z8#*C@A%!#xc<|J?Pm?AfwZt^V37Q`uZhxR{%<lP4%Tm$4V8P}YiFm8s^mwy4TTo;m
z%6k&j4S@iE)KXDv`%knjh~qpwcCDY^wZ1!pBl#`_8B39L5}NcTOkRm;n8^;C?4-6$
zskI}iX2ss2S^}}_n;e*CaNu_~IIox=*5@7{2Eo1-EXdoXTf{ilGKcn*+TH?Gh?+XE
ze2aeA(g#IDWIKC?feD~a#jFQ5>L^!K{U4SBM4+b<(FyetDF<=@83(*)`lc_w@!Muo
z&JGxr$CPM|L{m#5tJ&*~pmdt(tPG&~?k{p?05wWBtwQC*)}1qR)O%n4IcuyYyPjg<
z_^2O<7wm#Me>`7p`^by!OAfR09WD~K`?{%P<nnMOA-}&~M*XH%_Mz<DKFTp<D8TIf
zX-$Fu+-W{vrsuf<@*eqR3v?G-t-xf5lfE)ji~}RL4!4g8%Ar?>C~{n3_3UCX5C=L*
z3rB-EWf!-b8RbL00G&N!|7I2$f$4@|4Wz7ZMU-}A_;S8wm?hQk-AGvMe4HH*oLf1i
ziE-8xCxPD&Nt-}}?xKPZOj^BCR52#+M2*F|+lk~p*5eRf+?ik7V(8`hIc{_`H)qr*
za|ll_r2i8?@*#r*dQV!D9u3JuQmqjxJZ6qgL#$-nhM(PR%i6SW117F#H4jWk=B8yp
zab(7A<F0)fDA9tnVA1iy{0Ln>Jt56UMllr>-uJkl;HN>4q9j@7zcM6_yT&5vD8$nh
z7VXJxMeJ+TyC=Luu~QOIOK6J3*UF>4uw{<W-y1FQ`E&F0l90dBItgf-5Qr3{Ex}mm
z6kXH9qt^c}l#GV#5v3Ssx*iFk{b^G4l+mXS0R#0GwUNdN96f<=4;j-~u8q2~UQ?yl
zW-bF_;QvrB<>$uk&&}=ZlIN`cZ4(IO(CcAib<k4ODpz&UUtLD<qWy7(MLdY+XS{JE
zC4H2;c+QPNs4cZL(;niUrj5evd-v1<q@Vx<WYgEoF6@|I>BtpT`Ww%+aE*fQq5F@x
zf><m-FQ5@SX%99War$?0l-{L2@*?ALTmz4~emJU5leesMkN?e%VKCR6getfnfcMy1
zv7|#%GsSDyA4teE!Sz4Kg2K-@$hI%;?q`G1QtQjl*81y?Ikz><c5%~Z_8F`0N?-UG
zSCTTA3rQAtSo9~<wVl>>xSU_zj^8N6wYT1sdMJEVZoykyjWa4Wy2DK5C<F$I?kAZn
z{zJ`#8JOO?=D#w21_0#Ii@YJD7>qe)=8lh656QjcGvt`_zau}wGx0THmJZ4vI={C+
zB!Z)UCPhOt&|VK%d9%Ut(kW4GR4}^NaWhG6AWxx#h;oF}2jxjcAlCdymU)JGB}~0H
z7A(jXj=U;}9Sr!YR;DH6(F2u22tvb|QG)psmUPMGndMphkqG|q*5VJD=@|nGdSasK
zWwh6s4`MCVeK*)2^@N3w#WPX@y771f%-r0l)WSuB%PQ)q-JkG`&^R0^__7ry2G?i;
zq%E@;A!V6;H+vZjl%n&ktGCZc+krnUytx;%wh#~4_ElEEk2BUpN&`gTF74Z>5UXc5
zU4TXp(HWbK4P;k|l_tj^21@<3#2krKF(u7v{^n<gAY8$2ij6y49KDjcw9Kk1Sn2+j
za_;AKyiJKHJzuFlMn10g@Gzzs_xtl~&;;_RY!CcM=}2jgBfMsKM<{X?Y$;OyAo4nG
zz70LaZ4AJxXE9qbl4z>~Q&kdYpN))$(Izb`JS=yO<M*5yLwmleB2zi>zNzw&pK+n~
z%vzbcB>XQa?o^Wx|9hPn+uA;-5sr@9pD!(?{f7a)2AZeIM1UkwyCm}KeDaTwJ7iQU
z0=IO@Sm!h^aRe6FRx%mi42mr`vF)Oj6Uo?ea5{LIU94SBd4haW<dcFLQ~0-Q4a~Dg
zy66G=Hf$Fg8OWitg9DS&7IGIKX>o^m5o#UubWO5_Y}*bw^L4JgQ^)^Yfh*+dPkjA)
z`S-M+cj@x)QU|In1hv#yeFLELVV!Bg=khP(@``hShVJCT%C|4tQc|IYxA{FlvVzC`
z=KqdKvcXdk&9oAreEUmaf6lEH!B}j+@>aqa)!T~?LP=^jnWX9=bQ~2fq4=ZZflYCY
z_lLpeafuoCIhx011!pfAI3C(+de)EolEL(Y3Z~_zr`0H{g|5hEc@6-AKmF*jHrIHU
z#4jeAq$m8vlg=fQqCRHnm91GK8uBthVQ(sm@xH3nLO)|^<-f!~N$c)*S<&XVnYkp(
zW2Y@um0JLr$;4B?Q8$`=DfNPR7h#_9xu(x&Ei%4Z1%t(Gv{;EwlOW?nh%qDUzi2!5
zbOETSOepy>Zk|{8((ow<|6l!kO=t+(7(?VZV@EnEO@_X-Jb61}=079RNh9i-?#KPe
zIB~g}XUhh8Lze!C*~yQ$(fLngE`81j2Zsv}a|9VeBPhtl?$YSa@W3qH#pTNYU!>DM
z9lbP-eneP8@nCSvt{?B>$69TvpryuupXC#ZnqjEn#?Ti<p3>gjg-j1CONGf~&aw|{
z%?iqQyR>hn<=4e?PYvJEa8wvQJozHpu<^2OyUpL*c(~l0#$St8Tc<4Q&bYJ$Ygkmo
zgD~dwjB+)O(%<jLG7@N9#J&gZ^3Qi)Q>A--jfEWtdGJpIDglV%T>pBlQxsQ_VGfDt
zy7$xv6@~kcsgj=32ooY(20$ZN(DVSc_%<P>wR<_nm%l0_O@<xqofFjxnY?IWrImb9
zC!W|K1ui~vfxoSLPw-MdfkZ%E&3q=KDU~}E%5S@IMRCpoZ~HPyeWo-kN;T4DUOO1(
z@vy={{_V*oaZ0|W+ooRspzLFtBR3EG@EDJpFj4%}@TB<9Sx(vMxV$DC`#rjSxR=t$
z1RIZ=>InnVrWi!j5Mlf3<mzJd>T*VTsl5Xik}eOXP3xST>iks|!)~*4gW3eZXl>lo
z?>VstHhA$O#MPN8!hcb&AGI<*bG-yE8jguqXzKa5pLi|kr6)-85Q>4<fA=!8=UGFg
zs_Vs!#rlr^c+B;gA(Fn$j)Kf!U|t~uhDzbx>gg4_7|4*%dOloL8&F+BWeb)6QvM7;
zbM^~cNyV{}my9KYn}j{f;urPqO?K1An`%PCPRwv;y?_1kRi}?u&%fY@deews^okSa
zARt|_>@oMti;$2G_p|AK^f)<eg(A33<wG8O)v+OMO3l}F%sjcuzveTN!AbAFDa;t@
znjKhC%Tu6lnx!qPS~js!@Cj%=oj-f|sI?{EEP)eNpv?!b342bNJudMp0ew2)eZ@52
zr~G6if*A<M)9d*j)b6&%RaeM=dd88*|A3uxw<PGPhu<qD3ax~~NgdHuK9s12875uA
znE7a$?|-(C1VZ}nzqs&(zcO%9RbP?Pb#JQz##uIMD74BcOHO8C_pHf7=3Tf_3>2L;
z$6-sgtd1}hj!1v#KIo9-T)yAZ@%d3YrfX7g#pjcv*uKeYH~%Z-$4F{~ghpMKUcAF<
zr$q!u1VZ6t*Kfc^QuRg`s$Lre05h+3Flu7TvuHv55Ej0$J8!tMWWZAOMZNIwAAhy&
zp;T%RN-#&li{G^mu1w=;;*P|X1lS_3l017m^b~nQxOhT}m17^XE!#lZ_?Vq@DA(AW
zZ}#$0sq0#Lwfj=Y!ti-atxKdOGTwUASx)hzLj_vf$m=MX!L`>41hwY@TGAquN<?Al
z!q*p^))eR3;{`gT0>n!$jeU8bfBFC9B)SL>Nva_|S}Qf1YLMuP0ySGkf5A+m^b8Fr
ze_T#R`XjftDi7=Ss~K`s`s#Rn=5TSo^WqIxnp_Nsk0O{Q%`yfg09;B7ivisShNLyc
zfD%v!z6TJ*qgz*zP*dNs&(A#Eb_-jn0rE7o`G%h4LH)_dfSWh6C<-TYFRkYmU3KF4
zzp+ma;}`fPSpi)YJ6m_YliV(^-tPE3efsqJQ$WUm**xASgZiF<kpoW&OwwlsM3a=5
z$lO@QqZlJ9`hk}9rbJ8v!Y{`2seiXaaDG1uOr-!#c+j<hkR42Fc0}5&Pdq7mlEywM
zI`>ir^LmMpX7FN<!L{pnZ)d^cp}uE&S&e9=Ym6uz>%<LDquq*l($QhdXY0=mdkqPd
zZW@c^IgAma(L>rD$^TyPf}qh&li&LgHw%+%V9C^rBp&E(aVXFh(-jeGb`9@VmcfW}
zlJMQG2*Z4tymj%JMGx$qZx%sP>ZBLeP7wN=DlOU*dkM2s4mK3HxChUcm>d?Pe8zd7
zIt-Ayct~>udExMi-aF<E>&EhKub+p*37cy1){~xZRr*@`Ui5TKAZjRmS5Je;nqwRq
zutrry9NJ$k<yBVv8WebS(Ld3;-)Mpwa+%~thOobrn6vraU93e7aGGm$Gf(aB@1H)J
ze%3)6+Md(GLG0eCPh2)7^VrZ2L%ZrePCNdXZrBdmZNrt=rx?Xx->YaxaBA2^dp^{@
zSXogSmb_E7dqE~jo)E512xN)7bmsHZDOa4VeFfuV<G{YIJr3p@i7s^Sz@w-IcqGBI
z-e>1Xj)q$MEO}`*MN5uJy2naK$oB^=6yGV}{fA=j=syivxTJ<{8GUTW%D)}N86=x8
zq}*9UJFyiGuQ$<Dfk;K*Z+Lb)hdIkjr)PJL4<|xsCcb@uz;54Q-Ws1GRl98%;(|Zc
z=$YIG1OCxWWU0Fn8*{c+t5M|R$hKyjfW~yOLU!7|1+U~64{~`^s)kAbMQH2dkwq#Q
zlPp?V^@*-d_OXAH&%({J0&0bh=k3qM!wg-@dn<d-_O~uVLmG@AG?V9zq#$o!h@FjJ
zeg1Td>cK^|JREmj4bpG){kUs*0i+kQba*nCV@)E4{Pt~f;LZdrqHK3o<@FvUHx+J<
zjdwURpp0c@bMy%*YJutUuprNyugSbdMPp(7H!jUFz<o9iLtL8Txg+ikBRL!Eq(X(k
zH%$9dvLS4iDYThuaEe|WzZg{_^JuF;_K5Hr%#YBZ;;A-uo0oew|LP7bT8O@})q(z2
z-7S3<X%<5kLoE;#D-);@Nry1S*-`<D_fEbzRX$eNB*yGD{yyi$1}J_cE5M!qW#0F{
z1Hgl!93vP^(ONfGCz{eG@K>>0c~DtJh8FA>eiD?mU6#e!J;3eAUb6hoed9*wGEK_u
zu73$%snXZx#e_E1<(NY3zo~$aiyraYK#kB*)J=1IHE@)H_1Ay^wNy@;4OJWRScRmK
zxY!W6dsga(s4-}Mp8q6N`WLmJZ0da%M!*VtAz^=grsHac$=B-YaOvuBv&Yy<#-BGt
zUD`HZmxlPFodzBmKp=^V!jy%kBq9w|O8C8cdDp!`fMs@?QyOD2N2{lTjAS3q7vZe|
z>sIPBM8!y$Rlvkejr^fuYl(U5#Qnnw1MAR>74KFiDZC~m?AHKaEd*Hypqzv}ii&j>
zkEF43#zgdIznf4GfW{47FX^Oxk<A7Wc%bXWkdbvWe7(|0a_twxJRRI-@Bo)S0s$od
zr>xNek$W9_Hsn2i!_Hz(gd-|hqmQACOkx6~xIx_dbgU42ZRzOO&*eCT_GKCZzOO<h
zyUjV`05^-WXC3H;sJedzTpV9A10o;9e0mlVIGdwS{N*g?L4JbWCh3#(iIZdWJCA@b
z)l1jdU-PzP$t*wfd~0Xv4_TbvRSt%E&~5a!(@5V4xh-tR7<9AB{3&ma_y|B-?N^{1
z>f={jz7wDe+%RUL0qL@uit5}5Y~3X*7`eF~pK&GZ4V~<I-~tKJtpnEj>9akKwYDs(
zKddb$JtC(Q&KeF7uAV;6qMgo#<;8Py#T`CgFi%oiGXeMjLpeXSeLu2UbcWRBXhlyq
zGs~lkEjRXDX#tOMoSu`z+5V=GfmD>jWZNACl`R{eE#-E$Gd<Vvp8ijQh$`&6?$jb-
zv2ts%YPvBcaB;dStwFNzGu`J(fgJi3ILiPhFz^W3ssHT;Mk$cD+oZ^y9la*XSQIeB
zZmTJv+rZZHVjaDc?xrMJ2ZbeN9*t`~X-QsN=An7u@YDqhec5AqZN7v{gdW8Onz(lD
zdxG*|H8Zr_KJzUiA!9L%QoRll_gYN<{MP(17Y!gIF&oC0{A2ak%?tUi%@o}c<)Jn1
z&SOJ|TT}OF;=U~=y3%ck<w-#);k9#3Ul;dE4eHpSv;|TAFSNa{6<Ki5Vxs&U)g%90
zG;FgwIe(ZsRqFH1O=X&UQ8$UoDj((j%cW{3fZuyX1(4z`BrU$O3Z|W8iHpqUoDor}
zSH5lMziY363OxBYv`-u|jRFZ+^BB#3+Q}@>0I$7d%(NfPZlr+cwVI<<BileFnwMkr
zthRk{AfTu_HbDr2qN~;23ZbfPt928akP@8CS?1Z3uFAG<*J>f~R-oj=zqSE_=jzjq
zxKj9yrS2DZBI*(QNU&7FGFGp~%U`C~tjzG#TJ8Q94rIU#@qAT+LUI@#V04*?HEUuk
zr%MaRGCz+Pd3NssgMYc>LTc&$k*5VuV;vj2ogi3%9!Vu$wnBEiW;fQVh)sXC-Sdrt
z%w6UFI!F7KH{^`oM)7=Aw|@Msl{p0tm7OHGhs+wzoV)S?zP;`9eDChWtPjSU(LMbp
zC;PzA_)XjUyLWL0*5adSZNBz`Fp*9b76Ha0MyqQU^`7j;Af`zGw^Icw#fBDYKpZpz
zzFg8GM%F#^eEKJn|5&MSaf&Ok8#FbZ!1V*BfSih|3=X^}isB{GyfN2*D7XIuVOTb~
zeZGt)h1_17`F!^*RLWT9p_F06O9<e3j6|5BjR$t%!(3Yj{ASqQ2GMIBXJd1VkssRb
z`MQHvp2V1xbQ?xAHe~BJu!(eQG(7<v_(VAhH^`?q$K5~OAHwc+_PbfB($5-ZKctkx
zm|<!WfFy_X+bI5mfG_SE03(PJ(O+um*-W(r8`g;FbH$oYgcAMSnc)<(t+2U?H?>Br
zCvfl&67`T$_gHCx1_*euX!e|$HC+o6%e8-?Bkc{++vAmO^NhbH6;TouC^#BIk9u%M
zL(V}g2rYgG@&OKD-h!_mY|3lB8?#;bHRtz!{m%DMA=FPSYNZ2V|Ay>^!19i|9B;gR
zTUpNdlzU-;3~geB;J>&(wE1&ZtrXfCafx0xW98e?pao}g@|C;n7B)OdYZr}q5i17Z
zd(Esj4yvat3%IuVh;Vqh*-zHq+6C9tW!0Vzt1a{k)T%c<vpLE(=UGdq^eUn7ka{ny
zt6#@vi(<LhoGl;FyLHDZH<W{!nGr!#+fO_Anhq&YDZ&un$&5yKc{3(Z7(oo|6y>=}
zgP#Md-3Oy)s<YnbTh(1s!6m)DweIorY`E0M18q%f#BV)%FNqrgO{y{AZ+DrbL$E4!
zc>^4xi3+Q)%%3E(gq5thV+`*+t?;#KobY>2L>bYLOJ5KlVnkQUZrq?3U9OG42qbur
z+73gX5<q^=uc^w3PaH4yz!l4V6s~dizijV3(QVsp++7UR{FEN4GWZgk4u)Tdd~as2
z$a7eP(hSjBW}T+ZT=b8UNkNvo!=Syd3C4Wp4vHTdl4>+x5+9kKnx)DnPOhfcqJHUK
zq8i^_{tXta2GY+qt`L<5y%>VsK-I19VH!V7LDCe%mc!f>G(F>CpuZ<dd%o*cpV#+W
zs*TbBsy}^v?vd(D{FOhZ5?6ngRPBa%RK157g6U_eA#zv==Gw%KW-n0axF4~Kd|JEz
z=1pu44R6>DblEcQerF_9jLu`Gj_f?4(nLz#=z1}xf-tcNfZkb^Z%9DfpAd6)+weVG
z+lj{VBGf@5Jrt5)<pOT<hC~8Z#rZWw!b(cio;|TmCUi!`|1ljpPS9H&;+O8Z;))N^
z04RN|<TqJR>{cmxbhm_XjrGlMBK3KrNQvFbaNectt#I<_4=g|9Q85$Vm_Hvyp`E#k
z*9G&JZZ0#RW$XDslj}4N35Cr*BHDs7>+b_Q4)1ofyRL8l%2k?fpz7s#D?+D=p|Mxl
zThy$$0UI$eK#yrjNiml@TBY8Dy9c)}x>p3A3!(;+Ri6Naj7+u<iFaVnpIeg}`R&Rn
zp}D!aJ+9^CE|JGBl=>jKJP1nDD}u0p7<EMk_l;UtfA;52r23pH9?Hyqc{Ek}v{+46
z&q9$R@`a9KZIWlMraDBu6aVd1r{3NzqyQXBMGq-JWHpet&*fM_J%1a(khANc;LEwc
zwk~&*E0>}wDP`F+I~ll!p9`wAC(5!q(F-v7803htwoscbSG^CC0izID5wmwtnuIqA
zEldx06E>{+4b#O3JGB#D-^8a87Vs3~iPxGqr6;CO5B&%P4)&CzbR$4{mI_G)TJl3d
z3mGVIVB3@T?2p(tf%_&mI}_}RmmBAdG&kC<n;X2-^O_GOcJ7~Y%tbEvwQes`n|W!j
znPjOC8qL5XXdY1?+n{P1kme#}URMZm=5^}=>iFPvX#YZf0L}3Y8EMS4qkqvN6$z2F
zt#Y)&Qc^(7Y3ozU);e;_E>KEKDV#Bc7M7>f)18jTE$>@C8g4;a{H1zVnqJ6SB4y2J
zc@cJV`D3PUchsu4N11j<$yG%sdfKwi-@p69^7s5pCjwiQPzTfQIO1wx^vo;Gzc-c<
ze<Hgb*c}8{;)hBGX7Zg1TWd@dKHBdFdMwWA$qs)}^rav~EuPMb+%=tsKLP^qAgp?k
zwk;e6^a8bQ|C*-y9%ALf90_4a*|gAx<S3emO!e>e{;a}s20`=q1VbXD!JS7YhHqe{
z9r8c}pYIeX(XsU}K^8hE4wH_GX|Qh&|CPz{JozB~LvsH&CE$&HCzoXTxr}j-(%mfk
zxX}OipOZRqH_={a25)=5rK_V0$@P4J>F<M|pwh2*vgxg=kMeVXx>%ps324ffp~@D-
z@6O!q_^N<9J-z4%7Jca4&JUETH&S7N&=0Ai+TC9X+qL1QO?<7c7bMp}t~>edXV70p
z4cjG`C1oh#i1)bPxRY8lKJttbHou5I^cV)xcXH+<p#MS8odB;lA<dVk6YIH`>tu-G
zbJDY?hn~{?qAOyL+WFv4pBlAh1teY{y$XwdQw03pl|uo0XQ^6jDF0Ba*EU|7Unw*S
z^_%D8lE*Du^fNBDGt7L6#A`-PTez57kHWe_3d}Ug(v89&shzB)!lo^87~6a*$Y@_;
zeEE<()L85pSyzJ4ZHwRFKdn)UjFJE{OHw%G$<U;EdGsjIzmyFm$GeAJJ+?zzfs#6W
z!UI2nRj+s#X*gAU5Ay9_KJ_4{1>Mq78cEMfRw^~<nXiXU(BP&rzU<EJH0(Zm-ZkBD
z&(y7lv_{&ovX=!@NPI9kO|6?eB$EM@KBijewZ3PZiwLgljbIpTh&sAF|93hxbg^1Z
zo8*Shk6pHUn>ja2)5b0We+HG3MHtzN`<AO4TOW=pFKK=jEk;}iiENQ!1B5aZAtO!9
z%#R_8V<)7pKG7yp*g(YN=L~X&`Hl1LmFU;(#wdRexi45K1K#aHm4QKg8#%OfNF+W@
z`aT89VrYP9Nai6C{*7jU!4^PT8sUP<)?F<4hrTDzX<$TK*bNHx$jJSz>ftX-{fKA7
zQJNFzZ{|^s=8gAKEjTlZlGIR{<J;TM--(iI8&%N)EJBua_Rzymx4Fj)<4ZygBy`2>
z#J6%7M2iBFXPdUu+8QL%n^Wtszw<S)0B0*?w2EW|!lmvT8|GWjLULt9D2A7EFBUks
z9ivXuU@&Oa`>96?DFWeR`uzj@#>Q)z^EPnFIfo9|NxuR}Jjp#FnWI*REt>;$^<H*y
zb@w1>fhyp>P7`Q7{<GY?O-Fnh8GLq0o@{GB+c{rZ32l9+U4Bhs;3L1R5UgmRKYod`
z`!JM1I^*~?93n!)Q@(dqa__u~39j+Ri`DxP`5L>y&w<<T>1i$XHD_LK-(`Qjd0}e~
zK@NWdFxd1OT<z~Kp{`Dqm9LJ64(muIB`d(Syl4%)Go7s7gMQ4~#0_Zx0|4M1n1Qhe
z+Z5?7KbYM?vrXKV-(QI1svRh2_#MHp{~usEy>b0#ue8Slvj@g$+ld)q<I*tm65b!l
zJDln{fS6@bD~fLbv4n)-;mOHQm52FO*I+KfOH**Ni@-CBni|jZB`%QMNg?r`%$6?K
z3Jys8EG#N44m3<^bNBk%mq=E==@v~MrBM-g(ry9fYVupN*aZdWNI|=W46hb;9+67g
z?tD-+3K!aJW3-9sZ29&px%w4o$pcs5=3T|+N?S<SuvaI(eznM;h2|G9$b4v<*dj{a
z345ZLc$L1QP-K7zseS3oLf5sPI-E;T>(|B5(^Vp|dYA@F%L08+N440p9SGvZ*S5HJ
zib`iE+!^Y2G{;rxWiId<k5eNFS!{r$uit@T??efWA9<&PS{HZS4O6dqhv&5avro?|
z{*nX4hSAM5xR<UT)MSsuqJ^^rD>U?p1@2Yxi&H3tlk3Ufm@zA|3YnUYf*Fir5}IX?
z2`USq6kWaoN3nQ=>8Px;jhxM?UL>2!&f+A?wOV9}s3Kn>&NM_=6fb%Uz}-Zl(rJhz
zGIGQ?>D7LU&~ed)(&UN14C^UuH7BiDP}B0WI@N_`IdK;8+(<o(oQi3xfhCT_f6ze!
zxKV6SsV7qk_-u)9L`=%J<+0)ULMZKMox=Jn*2@glGU?j~zbf^<rv78sE`#t%j%N#s
zq6~R3@RCSsg?Ujx_5ket7X%=TCcN8NlYGn9dYVwp{9{@$gL)qL=QQw)3}rjs&%N?d
z`q<cFl$Vu=;HR?#rkrhiwO#gGU$ta6qVoCS-$Cnstt5&0w}}R&`z?CrYswu-9V;rL
z8US>H`e4T;xdrHSKiA?Fv4=AwN9k+F0*P?wF0Wu*5+f~Ra<ht9MItnXE?k|?kXhzY
z)_2pU{iYI4<%;|SGNRGYu$_cTMXged9EvpQ&_sA$K_PKnFr!AHIIVA2ghY?xc}>y%
ztr?tc*!m(Uq0Df39Dy%d<i`ngD^98oPpR<VrEuuRivAqts)Gs&*ok6Qd%JtHOJG)3
z3rs!wH`iR0@Zfl)zm&1ta!48V!Agb1<Ww7E$|BVw1Hu>EG#T-bQe=D8tC<OaoK!kk
z^s{CEHn4lmd@>|`w+c7i{Q;GTeYx?{ioiiCsj!U`qRUJFg?Q_UG-UrVa(?`qILF*t
z;e55<v&rA9M)<2VkLQlGmD%W{h0+T&eHtRWy3TistBEb{<N`#cTmxd=r=bF(-LLF5
zQ?jN{-X^O$ERUYP=SiMr!?Gh3>|XjiPzZ{u|5krDB~jyPz7;s9k76PVuC&adHwk~T
zT;COva@4PzGHClK62BLVKg1#Li$^3N;FjL-(khLuw=TBk`b<pm2D|!9@xE5(v|+4Z
zH7o+nM&1{dCh=!uZHLp5#dTDCQ0W@;6lQEE;~KL(SAwQiJn~<GW`wRVfo3M_sH8B<
zSrVFjmjxm}g`q_uH4<o|JO7xye(oY%45^*e#THd$m|TPMyn92PS36y7_=pbzeq2Nx
zK->iWbfD@M0xu_q-LjpSoKq0<jupngl|w8*7<q5NlMWDJH8OzyE+E#aB1NsNMN1Y4
zc&(2?B^$4Dr=TzR^mSurUZCd>82i(|UE5&P|4{${B%q)G)-~IBS-K#ZN8NolQ<xyg
zogC~VyF)pLvGYw?ow)=meG+!m8b#(bm-=2Dm)>56_&caOX?uIGz_>!PiL5hqxJJ_A
zS<~~+DN3IBhvnkmdXCi?1HnSY007?yBmGiVc<9MLjp=uAf}d`NLJTyKe@Bf}z34kX
zw|#PoBJ)2^f-lJHHaGWfTP33mUA&x1tz1`*!14J$Kz{8Lg6<U#|DPw5fOcvc;_pEc
zJw|r=p5Ex)bm$V;nXw}H^r4zSr&@wt+0~5eDE~jhCv?8p$%SH(EzgF*X~x?xljHhR
z@SKlIA3bd4m-3gIIq9JMt>RIl_JH+<e4czPih0eJ2lyCko*qy~1pY|u$GUH~#0MDz
z(&Zarzv0EH7Ks|hPWL_?F7|tRt7~=I&YjNoZ}h++Ivlqm+FclzJO31a-vk=0j}+vu
zXxhG5(dr)VQ(EveQ;dwF;!uUw?QO1{ACu1v%GvtlOwmvacCSwKp*dtc#J3(o-rLFO
z<L%W`ubH%vc?Jq0`zI5b2=hix)u(jS%tFxod_$|jK#RtS#*gF=TY*f;p)x7Ba^^iH
z{cr78Yh3lWzT@Gg`!v{n9}JrwR1V3j3p5&ZCQMKCOtdobc91U>o5J-Sz!%n*>Hr|a
zYkF^w!lk1NuV$-i+@hx7tQ-XuiZSZdQ)B_ff7vF$Fe?dwwrcA&7FBb0i>FObBPt&m
zhDqMxg7SC5TjUE?YqApo-NRPvE+Zi$5gASPyl9;i&Ys-<kX^q;vhH$PBh~;_*DP!e
zc-crtvkB&)Dr?S+V_ws_J(4i&nc%sfhxwYI(1}kn>(N7R%J6QwNDEJ*0O&r>68dT-
zt6oep>EAOZr4=qrjq@!Vpu6#RpBlI^V}wr3h$hY2ohE1jlsTPR9)MZv6hJ3}xM_^@
z<F(n*Kz(DTT<=zhj3+%*fKLE#AV_hzi#h}HjS#PV11)KOz?!fsf8^Oz8V`F;&Ocuf
zQ2`jx{7%pkIkZBHMiE<bMqj`q9d2FP8VFNTV9TZ*%jv){Z@#rmtwg~0z&W;`jk|Y0
zUv~?cw+L!Oooo)JSzG6JSl;~{^h!-jkI6s&s|LuJ{FE65q9eH%bdvafakN{&Jz@oV
zRfQnmKD^#XCOCVfM^4oXqBmU;+!ude)_ng7u-U3kelFMi5XHe`vsSSf^}<2OAmn}#
z_u*d<1=sh6^xHUk?#{k};<L3Pb^z6nPwGec;0>2TCNPqZnhO|GbjNG&HzQx?&3c|{
z((7j23?a0QOlobriw(HbFf9UC|HD`nPY8KxTp#^J6B9woCn=qB<m6|mqkT5?hJiJ@
za*BH=e5qn?s5j$oudf@^h(KE-9>@=P-sH*TqS6%knxkJzJ~;LKn)vl=^XqhRx{H`^
zIURi6iU+NN8)~!8vdUZtd<hLNUA(3ISQC-kl8mH>k%TD29xmt;Pi!A3ej+wj5slKv
zJE8;k-KawZDO&qEwg%8$3N(Gh#t)Of)ZcXb73c8spn$^Y(+#7+6CZhHC9*H{CRp}*
zAL3q((NJAVgE;;xDYRn=b@A_HXn%-Eb`Nrlk;8#bPnc6~KgUq2TD_`LQ7!f54C_Kz
zikXC^@TE)c$ai*&QGjnf0^GjIZByoS|Gw?keHcv7P5q%%r^Ow*r)Bt06&Y|@lU+AG
zcJ)qYlvu-_$!j*=EuJ63c~O}n70fomLt75gK4nNraddP%g}6TsxA=~Y{TsIdR4W_r
za9<1DwC*&h*{jfEr(PgWGs@c|13$oWwsX)l+CO<AF+uq&ZwILVck*v)(owGZ<Lone
z1tXo`JT~9N9+8q94#Ai}2=|B8;QU@8>_of>N&%VM@%7!nuaonI+(pH&${(sYz%*Gw
zkheiHSLC_`C^NQkUq_vzEX9aeCmPcn(8Lvfsr-|#K&Otas}nN2Rk93R-YTAMNFg_7
zE_&4XgI#AXPG;6el|xTYPXaIIj?OrGhC4<7yqAAF@oLSc!VuZgF98_+;!HZYP<o<-
z(w3h>NFbXy&Endb>{eYb*OnK#p?GeE^yUIYDu~|;@Ka4rqC5>@-~r*1U+7gpLyg);
zq$BF7OVZ2(Ct!S~o@-7O8QqUk?kg-i$1rFMO|0k^O@?FFc5ZZr_ga373ZuTofq4mZ
z%6B5&e=bg5JPq#9PV)ffp})U9=ANMlLA-6eI}gCiTRDtPE?Oqn3vLh3I;C)@i3~L=
zDQSqkTUl9bwSIn9YE2bfN-S{M8e<@!3D|{h8g7MWt7jqCMf6sEl=zK=S66(hixIUI
zr(m%kVbg7ot=>L_m5NjJ0Gx`4@W_3;F~C7Wx8+Pr&Q@T>kcEq_&c={pyBNxKoxm8F
zUGTMcwsQfhf=MgR87*xizd9Y39?UBE1~zs%Wu*^}ZO@+r^QP<cjAEQUA`+Xb$;J7N
z7HJZ$yQizGhyA(awBO18b|78<#iAzf{eKTB{Hf`Yb{OC+cc!XO7HcU0;&Y=bTtHOY
zv_RKFocj<0pezDvY$+?!qg)+WSfQ_Pcr|5!njjMUb=qDEC2b^*9|!k*|Mbm+C~5>>
znVIR*_%ijv5Dccb;&tIhvteS0iyQ?2+L=Nt*>BUP<cD9W*1s;c{=&(ORURO~mN~~w
zUEI@4GO93pP~KgW%8`3*B3%@5+x`CMJ6JR<TVW``M^o>^KNG8Blrr?clr`QD_AI-`
z9~nIaxX@VS=rj8{w+QMI-Rl&1d#%8e&!_Rki_^=~gT0k62g#pqZIaBuau&N-p<)FE
z#UF+>W@*K#gZL_(9`Ulo@!W0H;YLQ*QK#CaDx<p-d6YQSk<cOn48YB-50&P&pa1Jw
zKP$#9_^7LE?nIoS@FzoQ!;<nuQs=j4(b={^Z(3HGHF84q@IrdAkVGFHz?-w`mm)Ap
ztv|RkLKg@+89l$qxb)Qx0KZ)mT{E0Uk3aa5Jptos6kv78{8DxBT-(Z)_QAZEPAp{x
zB!d|(Y7|4{>4IwG2jCIK;68}dzQQyxni80Ykd%UD{D<=Nu)2#o3M2;ITzed$QG~@p
zq+$?&OlQG*CR3TwXdm_T=88j!$zVwgjnva0(jH>~JBO9x?{)No4Eu&(_*RV4`MO=p
zd}GeAgz86DrB4(9U5ZC-F$ca=64L8J4bu(N=j3<>deiFb#}OujUGcQaqx?c&PhwBE
zj(SnHep>V=5JKg0&f@I+q`p~dsHK`$DOfgll(+qC?<DzPIvv{eD5%vsq6RS?wf&bo
z)I5szY3}$xn$9wy$?tFDgApTy4G;l|F%SU}Mkw7#m&8w`JEc4C(F19b5mHJBN+?~D
z3d#VLk{BV~F+%#;|HboW@3;G$-QROv*ZEv>K*#0fAco`mV6<GXbJp`2=KLf0DNz+h
zy9yi)tQ?;N-W(9pSc}ISOFOu;-lg;A@1{+otrMk*&Tde6>BlXW)rq6_Qf(gZkBa2E
z$7(W*t1w3<MkdNGG@2CFJp>_v<*1wR*3M54mScqE<$)_7HrW?3&m$N*BOjktYUFE9
z3&VvqxC@Yk4w1rl7xnH!vG@tSG+8@6I#Ix3)hLCBGVO(8>brN!`!RhmN_Aa_)#3u)
zYsHa~1qBqc&?9j4l6?&;!BZ{vWWulL;(N~BUMvIX9~0-YO@5`YI28n(M2w0-us_De
zNxu>$wMN!KZ<k?9*j7&;SG${^{CcEu#))W{@FDSE(Pf@Ky_&dCFCZq6dbC~heqn~H
z!7NHw{mHQ3K~8-QQx4CZ0+b4K{e9vmz8Q?q5<+=ZV}cPrDh!VA9kTsJ+2Xio8pAJl
z+X6bH-R4*5LMC@t1srg?sTjT*u#})#sF+i_yYzU7_jpAzE!Yt9AZe!c#g;*$SOca;
zs8Ajt?ERGh+)T&3=g_e$CcOr1L?n$_tpUdZp1oHM#}RGwgy`Si&C&JMDMuJ7-Sd$(
zlmDJI2>+I&N{2pCPxGpXhKyRA<xO14U%<yq5nW@ur##f9nigBN)l#h5{<EHw!be;9
z$wI=jp1;8p0Q`#t7p|j8X2s4_``e5H1AK)deivd(_h6nknZAM6Nrf@}BZ~Qc9uI(6
zosz0RRuu0;{3l83XdoW8SI||)Ug0SEn1lg+SV_Cbeco4zRqwLm!*;mOIbKCyKH#=n
zUP@lRqXr1oT?PjB70<=FI_`N9i~Jf=AL<^rk@$-|RNRi`ih9v3A`FAiwe7glpao*h
zp>BM^KeS*<N&Ceerv(JwM-2EaE6{836pf~4QZt7}aWWN?54-pn9W*5Uh*s{MjZ4y<
zCeld{029yyNeZM8RC}cC*Tv=7DO~7GVPV7ZR(L1u@bkyiHTsy-U*4Wo{%nk!Yoq5!
zL5&d{#(vvnY@g)hWaMOJW#u~Zbn}=6YkIZo($r!csu~jGcdMReZ*h?00Ft0i&f9Qr
zr@Pg%pBWj_OBq+itKNr#Y_;K4nTWRtX4zNN(yF`&t%uOLlZ(l%ozWV?#ya!qFz=T$
z5El7{^an*LEhwT;G8zzTC=z1!jOTRB&kICyTkn<iJ2)rw@o-skhFh3+a1@-JKS;n^
zoa&x+d4{NtS0IPotcN%NGRMTK|0c`lbL}RXqR*IX<81y=!C(8l&ub_Qq(sooe>gwR
zf2Js*ql4=m|40k1ujjEtD8Jmt&cbsSIf(izBC`|6Nw#S%u^-cMYz8e+@wT3t99+qR
zB4tX(ueQ%lK1<ZRIwM!;<|~bQLE<=*3Qh9MmTu<xZ;w6Od*dH)vB4JPQ!`iAHpRti
z?=dL)!`MIRms*n-_v~<ae1E^`<W3k*YCxOS$+v+RZIz8$k*eG92u$Yb4L%w*_*tn-
zVP;R)yo8a?qiW|$H#W-@YQ4R)f*-6Qj21^rm`O}%+m_7A(oQBhY8i7~?cOs#^q?1?
zmMngMJ3H>tlNi9)Pptg?!HKExXge!(ycgRa*}Vv(q6Veei9-5#vDo!23Q1qHo8g*}
zVOLPJw@5Svk0q6Qa8j%ElFH_#!hQyS`i7&&DvV56U=~c2{T{-F3Kn~`B`X3Y?};aV
zkLJyTX2e72g`ojd#}EY=kF_Gh;>RY)_ug;o?w)o@&gddBFio}$ht@T}J&=ae{oADR
z+}rW~0m9OOfh>-_teUWXb=2IdwPm7u-%Ils*|FR$v0+dy`aruQ-mXC1s$j8o6;R_K
z>rPuf+UsKW7{={dWl<A4pLhHX+JI!L8H|FKPV~q3W3=nRT!bL8S3UT_Ey1R(k<PI5
zjeZWxFJYMX^<Y(SK-&uDpdO4NE12uN_$P}=X#n2`!aF<W1k3ROQuz-CtBz|5t>kVl
zSGF)S5#zr`{oo@<3B`+dZzy=!d|1++qx4|mhpVFaSZL1k>^dbm?BlS<R`|@oy0g?x
zYkIrrl0kM3Lfsf-v~{3$3^8_+b*im9^MY)cgoyPSF(W7EkLibwEE?Juo7}{#prV|L
z%}+mpF07c+P*M-EFnvaBHuLNw5s9Cxp(^S$lLjvq)V)zmP%xe2UrJ606_-K40rOo3
zx%1}vy^Ak-PO>WOi@7}IQ!fy7!Ui!pq^pXw-JtFmUijA!CM!@eKi}HoPqgi)69Ho|
zbIRL&Qmh8mZ<I{E>=~(%4RX`YO&9s-B0(*nGu1y_BA<P7G#HoVZl<}|xYsmTg%fHT
ze9~Vq^?{(|Z23sZzbJoINmy(f5d_Vvqv>Q#KKzDIT*1M?-uUG3qxd*Y;)FFH8YTR)
z95_5H^V)AVx76Jt5S<eY9;$$o{rcvPd^|hERc0HH8^z(34O#pKG1=_w922Hpw*Yur
zcF7Nd0v{}$?^nMjWOa>~zuvn!AL+~?#!E#DYfK!!kV6oce+9>rBJ7y#ZwVh&Cw|I`
zp=15Y8h7@SH<#7U_WAJIx`mp)_FK*Auh{poVq$~1bP^{j(GjU!bjFIvx8}Xmpu7C5
z-h~>o?5>NeZ=CpSNKD^{1{GE2xB<3`F{8PU2l@KO+RBW6*Xe66*cG(Yf#oVUmr_N3
z`txiow{a`bzA@IuUJyHi^c_3tA|an@^=^L2QT=0(B_+dFcaPS*h|2E&Bt!JBg#}t}
z!UJF=XM98lCSIY+hy)GDwAWe=|I6&*-ci=9=efOhKD~meDZTWP%Fl98L=o2C^Sy!f
zMU*0U2kjP@Ey#4CegFFG^`%bg-PzNL)`<o><rraGx1GatN>J48zb9v-lbu)n^T5B0
zkyEX`hj#aAJ~~ckaKdsn#va_KSP6VZEkFEA(kgbtd+7O>V9KZXlUQ?#gi5`4wte4m
zm+rf=_ZT4E+O&l9kPP8vdj%ML)h_ImNi7y$(M#7JoMMk$+iqge_%F&88+_{Rzwk+f
zNWpjq;;5*S>ZNm9fF2OTFS~VX(+GPFekUa@${}=kw!NcQg}x1{y_-w=)84n#!nEwO
zz3(XDTz|0>QRnBbb@)Esv~};_CoQ<zh~GN(<+=%*HAoD?_>Dh>i5+lOT-{a}w(&@(
zc{kDOD4-Zbs9QEn%bQ|)cS`gPto1O)!LKc>BmC@gC2UTvUEb4;uScsYHUZv476+5C
zLDyy;?*CKAR-I06jqBuW92{^en`%B>BY-TV#tr(JJ`D6WQV>r=cINAsJ9U^ZTIbQl
zjL*^HhirCgos}mYIH0<1^S$wxo9m^ml}*%IeEg`3olRhkRuv%>#>6H9iW@9;X%yZ*
zJU>4_v?bfJ9sJBw?qH@xd4wnue67*yHhgAtVhe-XIT!FGEufYcH;ZU81QnFFq*O#g
z_)hh9V^I}!Qmy}?xjTa1OX`YXRkrARIsYP}`aH+PXs}Y#09~qr{c3_<=odD_TinVd
zJrXI08%n6UZ_>-!ggPLu7t8x}!5<1}-m`xm)U#sI2nG=mGuaF0B!~V6|JId5rx7l6
z*JZjgV*OfnLPufV*$CDX-tWC#nf2@j_a(B$k@D9uy(o&Pwde91h0jTcapF_c8#Z$O
z;`By8?)HH{1J6OpUuJgv3CKb`0btM0KAcMH#QC;|rL`$keQdPniILjY`F?Hfvf25&
z#E_RvuFCGu`z&VW9`#we|Gx`5b`;Q&T-MmPH92$Ddb6}Ae|0#zHaK~)UKFBY(n;|;
zrB~X;id>zT{M|261(_5fN%^Xe>j?%^>A_#W1NXZX)NyZmSs&kHVe~~1`|=W_nPN9^
zODAi4^EX?aVVD1Kp^(t<@Q!&wsn~!sU0CV_SA%dLx0vvkH+`T5Sc*t`+2uH0>-VOD
zswTA(LMuVcxcTBW&w~SyPSkT$4#i8TP86;1PfiCPC(hg%=;(+LRgEy=oZ(YF*Qgo=
zSi0gfQzf~4sQVZgh{@<kOqDQDN>Hi7Lw#;gFPnfq8v$~{5d8}tcOn;;dynUhHy?z-
zy7R6F7ZI2JUxU!A^TV^r!(DHOT709XRF&|(kyL0lghaDq@U!;ZeN+pVO58_LyN90z
z-7NU&Drk1tjS#BYgqrPcF4%#;r$Ytcw6C+U{GLYDi|E#totRQkNLWx%$m!*=F|LX5
zm#nKIW7K4-@(1dF^x{3S{=@THnF(Tj;~6>7y474bd7mri6XLP3Y%>df;!`=FrRB3%
zj;qSIU~UeGj9!eVPdO+(d&?G)k`0@Q(>==DmB;kVa`+y!@JwWs-yecd1U>`U_)J#%
zBzh^*!lKCe;6>Z&@ca!n-^>j@wtRYZ?WvHz+%8fgd4V$8&1+8V_%;6P6YUs>sCQ)A
zSiBQi_Yomv%$^pnWY4W$b+~v%(5a*mmY8#VxsX+XPzVS4xwL;~FmZNd%nJE!cFA0>
zfgX9pVMMs7OkA*@qvn%y4?s8M`-|ek@cWKZo*#RY1LoMs%6{#hCpPKHM}(a29tZ(v
znDSF6-(U{{-uU3_ylh9!K;`De7lE12lwt4o`o(2)pRqnE;0*T^vg(|x4rPiPU0FHV
z8Y#RvDZIfSpUH)&W}WY?zsK)o6c&u-on5eVuk;G~MKp}67C08hK>-);1sl=D!1HJN
zBxRVZ?WU9f&&Cj_N<AdM@lXf|xgj84=dVvYZ)QeEHzGRdrfWJ6uTor7@F7HUatheA
z*lqZg4j6*&%C0LWk))k6#_ne2(CV#>Ik8InU*9K!Ml!&%#OC+?4zvx0)T8)B>7Hx9
z79fWfA&0X=5GsB^A^4BXN>1>XcU>|oYZ5Hhf=@t6Q;piCUd?KRt0xQogq)-X$ugf_
zueHhh`^i}BN!EY-D5CZ68g|@?a7JP%h=)5pwgR43VfiKE36mBP*}GJZgi?ldgr_52
zDTF+$G!fk@>}yf!I#~>5^hWe0*?-Teqq!>_*Tj#qhMoiQisGOV@v6K+Hcb*LS;Xhj
z>gg~=!Qqw5f4d{VSyN5T<i(u8TUwmiC@Y6(&4=x#dR_F)@!@5ArfH0eNv4A`qT+Dh
z>w5p7x5A~uZ3|~E>xxQHC-Gt~h~l8lwJ<Sh&Q0fkTC5OP|8xZR$c7wFW}B){ODxSx
zs3m_@BmxA^_jNzLU-U!QbCWYrulNb{>FV0Kigk8TW(lpTV?jEUJYAGLlVYb7{wT#}
z#n13O`Gl$SKob(n<)gHJf@LFp<(;$534RucJu{ipQIo{n49URZP^B35LfxAaOe(=A
zBACmI2c5t_!p)p@dEiUnC|Fja%XVO6%iC!gC?$!E|MCOe1xN8dk>Q){X|kYJyR#6}
z>2~=ll>*I+@L(t3+P^8@8OmCg$Pjpn!_LYgd`O#(j*qYd;n^2A9DZu9A+3)yKt9P{
z;-f2a?}I_IBDAp?jt~t_MSFy3uer|g`g1y0Wb)h_iHNI}LBDJc_B|br^u?>4%ae0}
zEX6hdQ>N)^Z*hfk9ikIc^Blw!RDq>lsKyyxG<w(&b<ZvAoE;vW?OeHU<>jb~G}5c-
z#ZjZ~*P$X${$_g({rkp6On&t;A;AP4K)LhB)s%nYTN}Nw`YbX_g<?LkwEBmk?>&gt
z*Sx%ygM$;Ro8xleYLbIMtRw2KSL6el1@NtoA`cKHk&1L;YOIm>rmu9Q<s;zpMrT*4
z?9PM&Yw1o}=GsvaSoM}5YK(4KR72Z9Kl{~(!n+Bd$h)iwIidg}Ved^E1oq(p0tAIA
zXpcAL`uh{!tZ5cLhictyNXbCPf)lS{ZN#xxmnj3^XqMuwF_PLeBZVezXCQa4q$!Lf
zA?GDr4X#B{iu4is&0_{#J7<GyYeJNbI93d8YN~Jkk9zfd6IVV}b>Fvt2xgE{WK8uW
z3XOy*I8QcK5o<=#lb8E_n^hVz?b|b4=UniAsoWn}PFatnTIAgK5FCo6<ku4!S!l$*
zjx}pSH5{iv95<N=)UXf0bJRsh1n^Kq$FS<FEKU3<t^D;eySVr%VZ3c}G}@fGXgT)q
zV0JI56tI24oW8^!>PfL$z57=e*51#Fl7<q&Zl}BaHeltO#pYw_I4Bfoo3jSI6^OCm
zhrOcoh<<A8$HrW?sx;ak5>-~B!PZ5GL_`C9eVp4<w12B}C}IGt&0J)UC(SaP=kHt_
zs4Hd8obwl2K`g4+ofHKz@&)0A3s(!ae{6Yatuubg{46kPQ_*kN(J4PW%MLZP^wbmD
zTf;Y2%rrAq>xhI<XP|>6{=GK71zq1Ifii0ldUJi-rOi7{Q@zvzIW&LN$S(Z(wo}JI
zGonK~zOcJjGXr>bhMig6;DM$In*IucW%XYW0nFDR0oPnCBZEvuT<Z!24!xYk)(0Bi
z790$leH<Dp3!EQs;5|n$rxTBNEJ7z0jadI+gM1JM^MyjQu#=}BpZ`|p#waa`fl#eE
z$Dk2wVRQ|2=x42gM~fm9C8EJ!)?>I*nst%4veX|BZ~*6v2l4&19G<EDpQ|;A{WJ3N
zM$dlF3*G8*fiEy2eMo&c)BdYacByY#l@l1+&O4(EWM?6@UDl-FL&kC2VIQ4U(ei**
zhNrP_be*|-`bZ$(lNE_Ig2z)H+yiBDa!9>pQoX+nbN%(`SMbZl3ECi>YxlaG^~V$)
zP9>5YK-e>5k#A>tn25L!^yyu|m(EhI?OR>YtQHk`8a+H6T3g9M6Tjpg9C~$leNDXV
zLi5cak~EErtqqpY%%GvRLYUrMD$AF;L+TKNV&kSg;bua4T?F9VgLBqq(C4j*Ni8#=
zKjpgPm)W}}U4KWO0+?2p&TXo`s5>9c{$PZF!V+!N8<kJ?ox!fjoU6hAHB9J)&I@Aj
zXG1SiP)i3#LN_a&AqrLagH`UlTrkwLBkwutc#i8akK?M>Aa7oGwh}=p<+2k~qY<GM
z71PSNrAN9(QC8XW-N4UR^Sgwt($>Zg5@neckTv^qbz>Vqs?hu-&chVB$4ix9-((BG
zY^Emcpl=}`@}e)0?jxhL8pZ0ob~SG)pry`^KShzxhoy~k(1*f<H+_+TAmZkhvil~#
zX(@!qld{A9DK_V=h4}iuSwlpXF6=BiSC-|^W+@f&SrX_=)v#5heJU3M2K`B1up6TQ
zO;`+)Xnvw@dy8VI?VV}${%5hGwtl@wdynaeKqJW(yF6;yPvD?dC###ol}k_HTni)@
z(fpy#B(<@jX>zjHw$zAQ?9s9Lz_q22;XM*<j&F0gzqPEfxbrj2Nay9r7DlMhvJvC>
zl>o;qU-I)c_s|{ekIyRvt{3~|FR{LotdyVC{7T*%O5#{MWjbOXNEu|DLc;dXeg!VY
z7AKJXF|1ZC4m{=lx^_EsXX{ihmyYGC68n;Stjug};jYav63a|unOHKYlfAdBOUE+9
z#tQbN)=Vzz$NwzKX~_v>%hUuheQbWCaH0a-ArB1H>Gn>cs8Av7;>02djWs+D<@IRG
zb@TC<9SY5)uDA^GuS63=riSTjb8vIG0^LD8#>&*wKwa~-L1f7Xr*@6)kza(^;wJh7
zKNSa!mdqR#ecU{2YIQy=#H@o;@6&>xiaZ@#OrBZYNvq|T^5UCerOuwCvcq=R2#`5N
zYuG2E2o>FT%bcVM*_g;9eIM=G#L17~4&WAyH8C-e7K9e~+^l%<mU>v25h$LMRmTnf
z@N5io8a8ucd`#kYpf(m7<@iVX-<KmzB8gnZ`_|z9;9jaPi<CreXY&wq`}-*LgOIaL
zCLl!4sSdh@_*Nj2iiehEgT6iw(1<nj;e!g-*?||NlIaiUy5h=d$YcVXO~-qTcz4de
zg)5`@jHWr>Ti8UWi50jdjHqd;%-#}sao>d21cFMaBjo}qbk(XvX-VR6^TUHGtVH<(
zspaSR5e{A74*4PA!t88ibZ{%U<3)$dZz?Fn5486@3i14)cK$I43UvkY6|+hETNBcL
ze#Z(7=H+m$T6z>2u@@n?18irL62{`CEBP~yhUN+;Pcg?Qe!fz}$)@w$8+dv-VuU(h
zsfg&sUhD{1#d!<Lm{s&}OhgaAakdfOAp@Kp9iIDGPENE{n10s#Ce|>S>xq>7H*LQ<
zO}sxnL1s-^+EX+0b^zFgfa`L3$a$s=C1KwDnZGE^|6qQU_Z}@CEE3+qi;Ip0<rnd7
zy)LFzRXDcL(Kr(SN`KeyYv~Me@NY;%9sRjDD36TH=Q<u+ZZed*?(LmhUyE#GTUMR^
z;2#_$29%U{wpK`qD5^d*STzbEQH-mFDa9y8q)ISmi=BV`6Ufb?+$@f|K~fp(4F~|6
zVm}8-zI3r6csGbNpi#>n+J=$>4mp_i0Wce&PK<t<m}&ZNaq&9neN7Fmj*zSA=Io<_
zLMx(ud^w>$ZiFPx2J%Iqfn3Z`^n;-!8?AiUr2;fQI3g_Q-zji0IymZtwBVm(xUAjJ
z-~R;H^h*{WFCS#=G*qKS<*x;<jO;o7l;KU=ODyV&p+Qyu8WLKkBtoThF}Z}3V+Vi?
z7g<R6>Hs6!-NMYxWsXc?aV``9I@_)OGtQ&!U-J3rZ%QcY9_2B7{nMU+(x$wvq~BDN
z72^0#?DJFMpNgl}2zyGKMkVHs{2g90IJaeYfgiXCcZbusP!`u3oAW!&yVmP>QgG3|
z7;KWpb3-n~m$$;b<Y8WS$%MyxyM6-+{_9c!l#~j@TN-6N3hEV7u<nKvaEjhA5!q79
z?3AJ-__gL*Zj2S@TgCzcBISWd<#B@*^~d3(#U-|&J1L_Z`$v~2M~By!=Z*ebtIRC6
z8<uHd=?{1_t%hsW|L}Q)+7;ZRhT&k<nQ!d}yZ3ck1drhHDdCY>nGh!I(|^F_3|^g!
z*xu&!=bTO`j1gw9Z8$W)G2RphQ4$k+sad~^!Jll5jtUW^%)_%YpSk*(l)==-iyy?(
znA>QpZ~-VZHr6CfqSJ|%=?m%PJtt6-w?AC6WfhK1hfnFML$GiLCE-aQ-&~GePj>lP
z;GfqG;aPi0v3sQ8Z7SS>oe3NR84&x~*4UBwZ~CJ>m)~h)YD(G>49j(G{m&fu?){_0
zd<89$Gu~FbP<1Iv11V^iwfNJhOg|q6LD?zwovnCV^2*hr8ZkMat(WVhNO~_uP-G6)
z0||JwychR+>ZO>pa3v49jaoUdra{$W<Io?>9OJbQDi=+3%=6&@3Qq38^Y;(zJ~yA?
z?JNj)3Bn;~YiMV~D?8%yjMl$k_UL4nC|z*vWo}33myeiJO%ts$!Jm!CFMM;FCqiTd
zjDz^27W#X8B9vx!AIPvT+hpYs^ZkbojGkoFbGqt~3IXXqA6C7t&Wj@=gBJ#IGIoEr
z<0<^bBjRHKha;6Rqk~{pK-vm)5EL3H!?xxU&@?C5WAZXv-?@3p`GNjVg<CID|KlNr
zbKl#hj!j5s_8In5;pF^gL&;$S)dew84&>f{;RIDxkg``oI`a1KkGEaVD`Hv<mgvsB
zxk}B4@N$8F=Pu3fq&n_?9zlfmXIz=mR9La8TI(DBe&8oDpn@cjEAJs#Rq0n2J)qGF
zlv9!R9$MT9mWj~xSod%|sG^?L2^BB(eQD`@V-$bTBow=S94_lQ@lfKY&sPPe$mpKJ
z0m1EYDPg<2@@>ukxh!fkXmFi`=gS1Vp8wl8$>^@VE*E1?w13ugACYOm&hAkrrT4m<
zv(Q%c$I4mY;_I7#4*=qcLDOu9g7Q{|?m3~(%flx5n+vO(b3#c(fmKu0y$?df)$ZGW
z9OKtcc&Q%1O2N-S>U;(~FkyUEpgYcBzXPcD;@I2JKP5`d$Vrx=hY}&<AV!fD`$H}v
z&GNDl&7B>c!Gwjk*iYMZ8aOJ#g&!kS)Lxuj3W6pK->yah61R!;6rArYhcqV(F<B2(
z<n(v{Hx5#_g(7yp0qM0yCx@YH`D1@sqhZ2BPBLWjetQ((Vv!$tMh6})O}ebHCpXz#
zzHa>w>c}?pVBRzB*RT`aOWK9GwpdyVc58+6_~x|99>dZfS!%<qZMNduj}Z}MPsy{P
zk(8pJ;Gfqffw(?_20^^ceYz)Te>%aQQkk%SL7{*i1W!PMH&%5tHM&=`5lAE&F@Za;
z>8W<&I3aV5$5;LiiEX0>QS5yjj*>c23lA@GOGE1o-Gg6&CZ*JMgHQ!$$Ao<2+lUB>
z2}4fed}a^I%AT1C9C1F07oC!duY#sPMnVF`OY&+KH^ZHdoZxQ{4}YFte8%>V4sqQ4
z>tEU+S)9Sd41-?*@xATZgGQSNt6shU{`?Y4`sA&8*!j`rrH`Cv8O`SV81`i-k;g{g
z>r$_enZ7ej`~Hj2E}1z`m2!sP7hhhoE%+xD&eHVK<R>c+-^fB2F4=2F)|zGpQ{}I{
zLN98uzb~f7LT0eH0zSmgFWrXeSrwE?xx8GL(TN`{C156iJw4!J%Tup#et7RgS=_hN
zD<5;u4>b((G1f}3h_EngK8uH+rXkK9o>lHGJ#2CIE)fBbk`+`K5sO9~)XQ;1OelC(
zPrsdH{*R^Jp{u1KT<?&N)ST?xtUSSb;krWsDf(-PJ7>K6cVF!*r{TH$;8)&*V0Gv1
zr(nbmWpyYp8H2MinO$zuBg))5YM-DWPV4&A_91sis#E5+bn1YiVQ&`>J6<z`6BbK)
z3A{->xVLJmTXj4avI#}PosVQ~9CeOugXuX#tC{uFlzC$=s+Xe{$?5(iwZ7$lPxD)z
zSaeP8HjMjFLCwCq;?RgN$4iYg7(H^|TG_d*o|pjK30p&8&`2vvw?ll?l4|roKZ8;y
z+vk$d*kk57L;8_kgqTZ!o}+}7$m1tXEb*`A7oZQ*Q-3=KTQGZ4hx3!aq<>n6g|^qk
z{ZXCcd+LuA70dsXM9n1W?GuHtRaB6yrfDFenn3@)`(~ttMR*6M64T$ynWkCgQ5hZ$
z*?I`n$NNM4-KWK$-BP=6Z|ik8e`h^{O}kiGUxL(Gk&(tvc}F#iGjkEjSgfH(L!#s>
zqSpsqbZelk)hDqLvGW}fyoXqy4&^xwvnopI_w{R12T?yW9Vi-u`bv<kI+Cf@h`hP{
zmCkuPxI&jhM+j<L^0MmwR2LM5)Li|Iurvb(yHz!7Q6P{S{3+#Mik#<VJ_8W3^x*1I
zbIm8G?7w1_N!(szCI?&3ri<K&&DYT`dtl{gP>LNfbEqA63CLRAnmXa75jVn1l%Ouv
z#e4-tfN(BM#0&3}?ab~vR<jAM?QG!jn30kAULW(_4_JY2$=R6xJ;u8)A^3W22S89n
zai1Vh&)*zw&EK5Q;4zc<V4b!;8LnqE##tdv>LlstQZ!P}@M?+U27>Agzdk%XpJ>SI
zoeniLZn~<Y79$QA6?PQFewG#@US?q>X@_Sjoi+{Pa6X~Zz`ujVlf~5_-!uZYos9er
z|La#QKkvenvPHUOCWEV8;w5_k$LdGmz!J1VTnLHZWJyWnFLZ9JM~d*5f`;OQls6a~
zs2?XNKs0JGH!^yI7>%`Zd88rv_JgLk3@Zp+5=h%ltkWCoGs*dL(wD<}b*Md*?g+su
zk?;>q!rJ%j7rm4hNp3VtKx}nZ?A)zxW(_o@sPCLjR$jCVKY~Cs8w>+v@<ZH4sxfP)
zbB}k}Jal6g$`)1acpxg-W&BDitPt0f-qu?bkrdp|TA4`XL<=bAYFSZ~ue2n=GM*=A
zFRbQ{N`I;1uWK)rYxdqf+ePvdL*Fkx$(c~DGG10qbgi4y>Pt#dq3&Y3H-IMOx#72Q
zYhj4L24ya9i7Zw3WI2KO(T(g#gRPaMg_xq5ujgcC3<b*uhEJ#wbm^RjdLXfcw(6g<
zO2(ldV_i!P%zH!3@2H{{pe|{uF{sS739y7Djm!N-nzZGIwvP?&4u^c#-)DNPiOL!{
z8fIZv4<Ym2m7eGbgNu^?+Sk3NG>uV4mG?^gv>xxXy!|Q_-NFbkX;FYwc$Jt1MaE=n
zU;X+v6DK(cc?J27{u5>TL+mx-cZxz8K)M=tw&%(<G#4v0VG-eOVY`^|vG|t9<vQ*$
zTuepuj-{GHk4n1IWu(=aqKsVmdOXTBXYmFpezd8u;DQzKm*XXtg@)0_6T*#mTHuNh
zYB5<Ib*YR+zXKW`-5LOdCT{)R2+|s0YPX!8CnS1-8$v?jOM81~%}iOx*6qf-h-jvt
zqt0Iwv<L+TXblw)&%M^ls}mPD_bnV-BTfN6OAco$;K&!WMwkaQBjJYKN^t()=4W2-
z<-^*q_78SXB7*-d9(aYORgkGd&8MX{v>kq$W(#|w0YsAMWJP9%u_I_<@)iY6r7)5R
zo?Cuko^h29aVED2pj>LMJ<VsNw%B~;DKmh(KY*&OF859HX)%vdobTR`Ny5I+Pd~MI
zp~7;XTd>f<((rqx-c+-E&rrvVzGEbLDydpm5$z)6$1R#$BK>8*LQLPrSEf2qsZuzy
z3M$*g*uOd8JKHjRSibaW{P*<9;blpcU7t+>OJu;$x!(@&vqUBhv>Z4G;E(lf=|M^Y
zNlGTnpMProQ;M*cV&v~qs7J_qo@|-j90!Mo?s?|73e3N5#OFsGfu6t4sGNH=H+O*v
zksQP9je~W`+xe~0eTxGal?ku|Ac2i{u<fbUUC`zwuy=N=85tk5txC7bzoGH3EM?@W
zKF0ddNH?*|#zUBxbhr()U@D8xYZ#*GmJcCGv%F;1a9wS&k-@OQ``R51^K#*j_v})|
z8Qx90<NuYbWe2;-!U<dw&na<&#J_YSiG8&Z@Z|;O_;-lj>t5iTibw@m#m}q|a>fLO
z?0N+=A8~e!Kl|*i)0P{OCEs-vr18QvP3#-Xo1D~sLy$+u)zzdP=J;@{e8g>#RpZ~c
z%%DtN5^h!1kH(Z*xp26`mno8N6zPJ=*-8CuxWpCa<{Y02{0qA9lA2|E8ni_SgFCnN
zt=O*S*tTNp?4du0Tu3ZJJ)JO>li(x(e-NZ-A7wfyqZe~M@k~@1n!71Bf|n_6i_R-;
zk`tx0v9U&avhgM^cxM?##ukG}e%;%581Izp075JaEzNviaPnNX`5%F3KB<kyD;AMg
zc7Ca{(rl3~$fndI-}^9AR#ic&H)ynKtu)$&h?-Uo?KE#Rb*qXce7CzwmXp9E?8%46
z*|o@f1-IFu2~eh>>lF!^dL1rw_CJ;ysmwQ$)c=8yTkgjSFjz(OK4zjqU7Z-DWQ68^
zmbCzu(JHKcDyWu%{NSS3_JTdsq%QwJ6j9{7?QY|g@adD422Xw7b<d>SO5#CEpkQHi
z*yiGv6eFF0&L^WEZTk<%$(e!Q?AV7rY>`k_QGFJs*myE6N<Ck`SgnwF@W-vvh5$(`
zODij*+pyJ9Ma`U5Lc>(=LBoX8zC@=dMo0W+Q(9(MU6+c31X)+p&AgEQSQ3q9gDFU)
z4{aPQ%FWKt&4Iem^}Wz_bAS9Z#hm!4dryv6NYg-k5$6Z-rj#%`2td|yW%=>$>Hn?}
zU0)uZ6M78rn6>h?$;rvn3FD!0++o1udf8pB`Ku$SwM|9CaGiF8zqIGh2yk^VaWNCz
z@hyt>#4_tO?&uXP=Tj}x?i%56LjjMW$A6lLpXpudzu>r8Spo>3R|kYVP^n-;-saW}
zQA@+b^nY4UN*aL%4N4;I25ey9R%z_J{m~-Y$`7XJj(hIm_LFiGh0;SjTRuFN$-MXn
z&ad^%wj<ozwuNp+!?{PoE}}XnnPCetwRQY~r?KX0cH1KRtQ=$YZhMt{4x5{dY02{D
zciP#ZU8pDqCdRf4h$GjX#Yltt9Jx-bY0AnMJnYK_%Sjx<S;}(tbLQ>?Wyqd(MgpMN
zukC2(cjF_FQ67?m_9bRy8Vt9lVt&>9|CVt0szgx&IGeW?A?tZQS$=9&C@AFAA*Qbs
zEj)3vRI|7AYc1(9{1ynTkcyw~Y=48<*tixOF@IBE;ko7D?jHFbq74~{{I<T#Sk~eG
zJtY6WX%-lxJOfS#bjy3iZY#hM;#N(AjSYimJ}n>BjokPK!({S3JUj@J!Ny6IZe3t#
z&$FbLo;2W~Ths`J?wGWPox1bEPwZ^yiX4N3DVRZikN6spcPc>3yAM=6an=Gz7<s;z
z*S@cE&hU3A>UwKl&NyqyTQ$ve^-tlCn<Gr<<%S**92^XU2ZwaL5yA{MdOUn0Hq~J%
zw9}rOpq?EpFq_>BIw)&wJo(~Lvo?;%;Hu}6O=$zO(TnVI%k|(u4+2bBbAqZ@2>y?!
zr(9V3#Q`lx#JQ*Z&C&c#P^)w-v+&FcU8KKdHWlLGAQ4fc`%V?TQhwRYOrGg<h)2U1
zWL;<I*DU7OEWxywY_cjC9`SVFdEb?__-1}|xL2MOXJ6jpn@^nEKN!3NoL!s;oqE+_
zd)%6u@F5YQ6oJ(AF%<%euB{p36!bBJD=#WZB*vx2zdI&Bg=Oe{@K~)hi#aKC>^Swe
z7^$h*TCuH!44~z<-$Ei|-OBg^w|b%rj1x#vxO8;{Z##fAt8GQ-f1ZG(BbC+3NOELC
z;^0bpy5Wr3Obj#B<Xwto?HPWsY$CqbAK#?$EM;$vj*%YldmX!(2*b;$91xgvvfYHV
zWRjv!@*w*kANnYG1KzIHrk$_CbcjMKiF(+C8{?ozo#Wlzlk3GYqc3kpm~1-S40#IH
zN(1v(9e`XOZW%V(p0Eoq;PgPGFa^f>hCs_NyNSF@(W1vht$_bkwo3m>TZ0s&FKQ(I
z$RFj9E9NE13V-W<r{_bY5DAGC{Ha`_vIawu>&(QwrO?fm-i?O^aDGNuJ+-WQO@XIJ
z7Yyf+n?WnU-E3xJqH!){#QdkMf&U%GBvM2|);yRTcr!0}?z8KPqgF$RKPg%6A2m*g
zK%SyUlk;6BA-A$@YxVhL#(jWL{J~N$$Ibpyf1$k0^`+Noe7}euZV0Qe6E-D?xgD9~
zT)<dHw+IcAD_pB!Pa)6pel37kY;KnJed0e~S7f-n$bIr!n7KuoO`STjci%6{GvuXA
z_`8q5Z(T0c*-!?iG;tU$4a;h&iIxVrm4E+{TH;d+)kH~RC4S~^oEq{&Tj(5r=A#HE
zQcjeaZk-ZykNRUHX9*8F#$`?O#)*!=%?`*jo2d`&;3O>p+ti)oKC%Rb><`O%*x0l#
zK3M?T#rEoSkKIEo1LIs!E@F&d4>ueSu0#M1bl%%VPmvNRE=p%odJwSz6~96G=oSVu
zA~dtr5_pf?rvymRyzGdTyAWW7se#DQun)O8dV6<lV@3rAy&n%P9PjuLor*kYd+NQ;
zW<i?YV;(y5A17x<uqBqw%rFpwfi*)tI!+s%kjMVGkS!0ZXhLE^Bb%~Lg~bw#0A3pm
z8a*H=b#`&ZHgf+VGaG`kPj+JFKuj%fNMLBC<;+AyBzX@q-Xb>LXh2c?J6S}g6K!8Q
zaisJu(v2--Xn68}_G-D-O{ktvq)I$2496v}x$I%;CtXa1<d3P}eD%rvVgF=#%8?L#
zvdJlXZld|nKV!nQwI(h;Q|U=rCD3yZ6ly4sc+Ia;ZW{HIi%>k55`RvpUnsPipQzeM
zc}`RoPeKVs<-Stm;oa2<mvc=*xmIspZ>^cCgj&&PJv2~lR#QXXJ&|ZLrg8ZDG$0Q|
zS^t-yzN~tz9s8Vmh2D6Dh6W}Af46VIHISeXx!yFR+;A%7$|^YCTZRTN`GiOXT(5w4
zc1^%jtNq5QdHyVesWmAMWY-2wJed8ep+FMroNF&RTt2g;Cm(*fK`=HKQ{~w_Fqr;D
zLPd*J<4}*rANPG8uDC0*?&I79BhGtH@>GHp@$oc5T=Ckk?z1p7EVC|HM=bhtv&my?
z>|<oG+I$`?5d-WtG|^AlVK%*2qR)PzxddhB;}qtrgrKh!UO`3%bl^P#x!Mb`@2^2`
zS(`K5Nno$(d+kB$Gu<EIkTlG<T76W9e)iqxd##V&cxA6Q6VWN!Lz&7Vc@eM)gR)>t
z8gfkJUxG3{yM)J|ZLGArmkq~#dD(RkN_!c|c+R=-{gjsTlmT`o*I~&*rS)*oPeOz%
z<Cv=1)KfB9<ZZKNkRmSwV_Ep9$ZOM5UE^`Cw}A=-Cp4kfuoC>U2Y=`_FPzpc+=6D{
zCL5`us$-}y%;%w0gV<5-X*enL*dHz+Z+h}YYDa<Ylf5(p`t@wjMr9K?xvD*H>5}zY
zl8RCDgts0qTe6G>$nFV?)yR315UWu>iiq1WKKs*syPBL&@1L&+&r&K#r`|7ZxJaa?
z@D7yIvz&S_GN9cTEvvW;HFN8z7&9#)hA*<v5GVqzWfMH`k{^rLM?~f&+JqCZ%%RnF
zTp>Uxy~oVqUc8r&)JNXv8d~(S{nU}ub@hd-<1S`$ATiD^Y-xuNqnA6ANljH<Ebfzg
zL6B<1F9>?H>+aS>ffZq{5u#li2MQftsF(}tK%iG{6&&}np&F3b7Bop3n0cXK7|mGh
zIwWJeFA)6xfYsm+Q8w>AdLJe6=c0i%fZVrP*L|N{-P7{;)KkKI65n4a$I70n=>4mz
zqv6MGVxK;-M+h2PaJ&M?TLpblPhE^}g|Ay>rNwwp`am45mB*lOdat@wl?YKlUW3dx
zh+-cqF~?(7CByg^c^BQ8qvSS;W|wDb_%@WgY@D`C!->BY|0}LWYBs;+wZX>@pf6hT
zk9uW7ri1X);gv6jkA!V!`%vtW>{;7|DzZ3xq(I4-CU$j#ke=ldIhUA)F&N5rK9e=Z
zP58?eM(Z1mg8B00?e#oIto<BwH|quBP0Q!bCvs27VCz4MZ+Ab18FyhH&j^NhC{!hx
z^Uw&G#UJ}zA6{Nxf2oMdnDOcY)F9_9@RQVzFYkT=5AR8m@6f9#QKS8K6W5Yos&}-s
zNY2jARe9>Qeo>xhQtV<nGU*~78{*E*F%!j^WS96<h`i_f;`uK!N9*<U2P4kabn$2s
zg}zt<Y(OANAl(Cisq~b7X0Q+Nr!wAK&2`ICh;o}#-}dCRG~0wV6D2_aUbN~!Ov_YY
zm%=d917AHW7h6_t=K*)(-Rw{%hqqS0Bl_p(eLSnx06&D}S1@S1UQ|Upo=jY|3;U3K
zo+wl^ApTZCmlsyi)|Xc6H`n`DQqQ;_nVD&^C$(Bl3_5f)#|;qbjnhmo9)hqtJC|f6
zW;2=(OBoc3TIQ}UgPI99EdPzBhGnnMUlYpOfwnhI?#8rST@d1)o<5b`;yCAAxl8E^
z`L!ekIz85Jxz~pk48zHSn1sG5?Qu}+9{$k(zhgA^-Jg(h{8yr3GwGV0vt!)VjZgIT
zT01K%E7ymm@fe96`xfXOzS69Q(7i_g9T$2M<Yh^q8B@<wTVygYph{+c_&3_8hFw>Z
zg9t=O{OS2f==ArG;`-YNOK~=={cCLJM<EPea+YQ^EVQ-4U-Mt{e|!c#y{*+-=SN4}
zJt!u1Gq;%uBceC0cVEJbb1O?Hyi{yTR{h&`Bt5#mPz^15YyYknfAY@cqE=v7A%E8P
zn^*5e92t`|xeYnM+P|G;<ZO)J=R^;^+L_wjzBbsE6S(!Qj}`Qt^+Coq`p+Hm?~C2`
zm-gki(y=6JleI+<M?$#Sg?QDYnyA>uu@s%w%v>zB`Z!RgIou5c0=2`^Tb}K>x79*a
zmUX~TZFPbQG3di<&}(Ng>E97EWm8SZKi8L}Pc$|&Jp%l%%?E!GEXkhC&*40Cv=Vl9
z=@TJYBioa}YD0}|6a5?5@ulV+$2T_+1<f`QQ?zV3%#AQ<*F`sdnVo0-u_&f@+wb_x
zQ<BKxm>+XfimbhPmF|Bi_@CzR3v1jNFJO&(V@;)w#fyw!f*JX~RqBFb1k~x!I97Am
z(;;G1HbB%3ej$+D%W11_mk?-kdq+*sY}aN-EA!`mkXv1aMh47)Bv+aRzU-v@Lr`)S
z*I)*6DKjK`N-7ZRWkvp^<OldvXKe0I(hWJGYP;A2%OD-ro%176Qp?QzZn+ohWBqi=
zMlPN#ktb}jT>?|Z9v)`2+#g5B>O0%Nkey-HoB(P8k(U08iz!(n!hTC<eLOjsW!yVE
zH~F%{P`{&s<NDvHSe8G71SxRKq3x)u?TH|FI#jlk7_Qm~Vq5m2*3Nsij7@K=A-LO!
zZR12F5fLZF?UMZ@od3%|p`%w%#kVVG+rWf_a7WES{`2oQ|6U?HyvN1fwZ+ZG=aiem
zG=9Ov@uyW*xn-V`yT`I?9rkphWuLnOQ8e+_yrfQioD^#63Ab*iFL_HAtcFzf(~9?}
zHXVCJCNOA+(x)r=XM7GFwzhpe7zW3fY=9epHWTSncMBm*wQA|aO5CQ;)1Y9h!np|{
zPt1@K5t~gsDKck%erEELe((>Q&_C0|)1oqz&d~5&h-U}5=gOrTZ%`D(wfE47xhP7A
zHM__7R7Ys;k5>L>vbs%r!KXaA*83#y8wd>(TDKV;m=m>@73x8vcztf-fgN&i$vqb!
z@jWz0<PJ}$7aGQtVbxB+rF<#%;4UHueSgurrLbuYj@}<DUY6rv)ouglMCZ`cjF8Nl
zz1nI<dIcbizy50Fj&`^-@wWQ&t*9(np6j=4mWI#LD-NbK%=3bO-oDmRmActyu8t|I
zNihN+9X;3TOhe@h*IF~9Tpck~AYD9QUZ#Y7HPdP_iyOjyJPwd76gPI16F}mIPp=|I
zZ}D~b{nuJQVS(!mejmTVLX;Ly3bI9L^CgonSl4RTi-Jr2l$hmcvDldI#WKhBoSipK
z_v|YS8bIzi?BqPfJ8}^5&*f5+)lw)GDfgcTI2jFzJu|mWSXUp`z(-e60Rk-AbThF?
zSKJM-Sl!(&43{OLNB=ZB^O7NW5c#`$_4|tiPa<%+)Y`$Cq(3;9M=;Y>R+L+?RG@!p
z=zs*&;SV>NX*~9rll?s=j(rFcT^}2x^P)N3QgSwNUH#S1wyo>*lmt~maX;oZNIk4n
zDdHBV@C{1@!{qmy#K4_C9e}Bc4Z6#$xLPd>K?*=1QSo}h&8@is#tq2Sq_=3X^!bZ*
zH&I$J&Cp}HLfYoUwf<AatUEce_tp}KpU0RGx8-sLKP1hOBmrsH4e7ASk|FM-8>sxg
z^3DFbLaI7o@N!HOF=6<<Q`p3Y_dRKG^HbWrJDwjJ<5Wg{7NmsALv<E5+(!v)fH2FE
zs=FCPXRlMlJ!)={_pi8WfU=*;#J{KRq(*VbK^Th}&G6MY9I?mQu70XvmCb{Xo<rK*
z9SEQFC;>=qK;5HrmBDn(_%=8j=>0@o<a+A$%yU{ssW;EH`8FZ+EE5)HpWdZE-XuQP
zm<$c|T(mT4oD*_yb9SO@sh|b^L7_cH&trxOnKcg_?sD3HS%P>Vrt#Hhyv0Fx1KG4Y
zb*t&KTY--J+E0S$7;Tk`=bpl_B5y^VBE-cXfgZ=#C!Ve89?E-7h%&=&ov_h*^R2?v
zH@z%pf-z@1;Cq13agfv;(0sGF#-W$mWNP<Cdf$A^s7yqo0$ayGTwm?Jqb@U2mnkAN
z=p$u0jBz9YMH`wNaTdL|gjLY!u5Rpf$c8(WsNM?0r|pkwf0!S|fZ7xao!<!Vv)v-n
zMm-{8B1tnBYPRG4TxVswF~_2i6O|{rd#z8a8r_z+(hShEGt!VKr1AXNoijGutNL(`
z#LVa9k?98WoctE!y?NJih(;%ounjSk9;@>((e^$dsa+oC<a|21yNG@B(l2XM&ejq{
ztdSOMnYB`j#hWKWZ`~14Q6WQQ86<qgwLKnx{g&!0wi!~O$x70f7fV(=8Df$$df+IB
z6OUdyJq89Z1_w(mTA7PTC1REekXNr)j>2p99vszECC2GqoFDD(t#_peH+typIWfXN
zOW|g*PnJ<kif%79b9o{ZxKV$ISs6O0hRpdnk95)2s~+oV4|2`-3>Dr7ii$7P^MO@h
z=?DW68{y~rM%l3Md|gx7bnf;_eqt!pCxo&lHy}y&#JYab?mq`I1&XeM1Xt=Q-L*|#
zN$v(ln~GA8s(5MmrzH4-U(^qcc63fKi~PzD35^F+JIT9baiXd!%`KHFc8Ohm^`g_C
ztq!&gA&L9%HKxWE<i0_9yg6sAM<*JGW_@NyqMMHy6;lTO8>`BCuF`tqKhrq(N^hO0
zbOK;j>QVeNdn=m8Ay|TVQEUh=GP%|?iD`0ovv|mNoGH5MT<tR=JWOQ>@d3F_-#))K
zCbQpmGOR=Ru4rVdP$Hm}s?FU@S2!agzNqV*=o{Kj?Y4>*ayl-3XpW(w@mB3rSy@UN
zR(-e%c4t+If)5ncwpr_fFS5|=d{&~)1!&rQL^S-cDcB{%EjV^Dc}qh}xUf)kViU(+
zd2hR=;Dm8#Il3y45~XkqCFAdDaHKlBJSPMTX`lx8x*B>N%#5XN^r6u)ImkP!;qkfL
z;eCtnrM6)1J%M7yzdcxD)+-63BJvVA%eVW*Vp6SfXaCfXv?a6--|^nI>Dl-OcpsOr
z-hUiEfTE<TgB!S6(1#^jWHO;3Jz_Cb7X#bu(o{A#{Nzq{C2y%cd=N>f0dv=(%8Xt{
z*jPn9<WYDyKpSZD*Ffa=e-NU#n$OG5Y6bq+g+zDsLj#RADaouDOGM3FysWPIIs7_|
z(VUWllxAC0vH81Exx@4K`#ydEzYu0;rM;uwU1*JfGl$VEgL-w><=F@K8V)W9`7|J8
zBv2kny32%ING(|*sr(webr?~7Td&;5gobcnQyMVH9Z~%*Gt(+X$LA06b2vgI>Eqq=
z67Rk(o?@3xA|d+jDWa?hMcDH2&FenxPC^(xppP<bzSkO+y|_h`vJQAa`jXNu#X-`-
z!d~-_7EO?m+-sf-SQYgaFD+%mCi7V12I@qPXHQg`pR49@Urk^|<E4pn26_!((P+P+
zTT>V3+#aoxJUF+m6susxvl<FNm2EB5<!DZw6}#J<mt)^D=?5uPW#wkjV`i{ouh5L?
zsz#Z}>y{S}+_ze*b%0w4%1=lx5lCAAzd6k)#a>zPZv3&A$q%En#Vdl`F%|M`AIEYM
zqFHmvgWcmz%ce;3Uf+|2>Wq)w8v6zVLy2t#W74yV(}Sw-viwHhTvgj$S_nPzKLx1>
zLY8T8wr+D{tO9ADP1NUB1Y$Lu>(?MOj>WRG<jow}FfOjwKnjY<ZDq+@iVc&O*SisP
z{;!uXEQb>A)~>Yt?Q04Y9~HTijuFzsh3A&-e2!ClNbdy~MUvVET3wu+UpH27$>_P1
ztv1xV;|TWYH%{W&Hb<Tp%e5IVs+XDaa-{U3NIG+hzFIECnT^YPN>sFMn$+$C50MkB
zGL8+rexC)XFjJ2)CcDdK6Id3rX!Axbmv8UXe0xuDQj&};i9|&SmJ!bO+0&*!c;@1K
zF_U$7NslSjz%+>P_Ue(0L6E@`(RW6!ydspb%8%3csq2t0?SCzfsh0e6yF3>YQnz=n
z3y6yl{KOW6>8bk{!wF4dV09>PHf8<bEn9Owh`|7u)8p_RGcEayj!1T`IM0ime{)~1
zLyPm=@~2KSdpU)LE|-r>VaVpC*0ZhP@Q~Mec9QHV{i6n(bZ~*eQbN!Qzou8{mL*jV
zAs3}CbW=p@FNrXFPmjJfHG#lnL>^%k(n`iEe(JvmrChU9+C4P?05RONVatK<yIB^s
z`2$-F-#ez9GMFAj+8?f-W{rSEzl(o&MXO*VGue1664a{<`N?xzbNG0m^ROrpAsrKK
z?a@beUU$%_KqSKpKl#P%=6L#*<-E)~*>jQO#CN+E;hqBG|8Z3j_pg!fLl6USP2$?>
zvH%whzBNl^h)S;^tAcoG>p~i7m>%CBZ&}(fxboY|vehMqJg!wzrJ0%JWv0X5yt`%>
z0j>2*rP$Y$nomOFP6O3$`S0hyOI|7-afN33vn$82Np`-QjqaX!8_oP;6Bj=Eg`vPL
z$E~3qkN8dAGak(v+*uA3V!V#Q$fpJom%Z_XSna=qrTEbULJ+RMcV{DfMk)xpFVK+5
z&9lO|d9r)bG+})A&LTqH9`>Oz2q56>tfn`MBV%K-2>0>`Gi`(87wx*hz5j7+126~2
zU`OEQrm*qUHG@mz^P|@}k`!4aC9Q7g*|QtKnIph4G*A24|Iu{b|5X3~8$V`bk0je6
zQTART4%uYIk&q~kV{;H*Rz}H6wv4hnNDkREBMyplj+}FN8<|-hviY9R58r>_{BUls
z=XzY%b-yTvA+-KaJZi!Pr$uW2dGd<1I2mtH#OiW^3Fvo)Bz`V7*M%fwh^>+?!lH=P
z8cFa{_=4hWFmP3x*)O?w8V!H0HvnFeHF$42Uet`jjEqW!-FRCZBB79Zr}@bnAS?IF
zN2dwm*V|@a$R+3(2RL6Cs^UML&%$B2_L<9}lkILtv5OBlmc2D$`qspQy}C695eLrG
zZs(ui7;Bia%)gMl!w<{U$%k?lRI$aEIH1ty?O8WK{;hTQyk|5svdykGq(WXBF>l?u
zS8>gD1Zcz%_skOHD^1M6`|%8*b<}e$&^)CMnPq2Ji?QuxN_N9bfHSrh{nrMBti&bZ
zhkM2Iuc@P}OiiT}MTNoCmS3z|5<zbVngsA(Uau-R9j$K86q-xMdpy^nYb^IUY0sD}
zOTQBEbfBtA`5kHH;V;pI<{J-|#2U7OuNCQX-mkzjnH%abGnZ!tKgNauazyL0f#m9g
zNI;V>;e4E;pjzV-{O-K8xvET{Qu)B25iqN~$0vv3Cc(j}DYx3-F4KoSV?TqBLkZ|&
z?T(1^e+Q$u(O~FGquuN38UsbklmN;Dmlb4>hxoMvRex4tg$MAAMGwrF9TH>g$vR^%
zhF|39#iY-%4PC}FxT*qwMEuK?o|}juMYDeJ@#Trq-%4(>Hed}35s?MmoNt_VtCe6B
zrT%X*c39#8ZgfHRy#R}apnz6n)XC9a|I(6Pjy(Mx?o2sVS$d9KtnF%;7(_2L@|<i1
z`I!&?DVypLp9`xk&$b#>bk387ZCDx5>ToBx^51`0ToGJ^-Khswed|{@;hs__jNp$Y
z7x}jbhzXAJ5N}9d^i^hCN8(?-$$4i=w!>b&_qSANzzON!ed*)7gp*o|uV2_BU{ZyH
z*y$#vhp*Kt8*<~6Yun7mBd8MN*Rb=#aff<4D!0Z4JAcuR6(~;iQe@~;QV5QRv3*az
zNiXDPYy0Xj%bNsPQ?6nO)26)nJUlCeTLK|^+g9u&U}i&8un|P`rpHL031;?tuI#NF
z^83k?Z&*pP%hc6x_hi5Od@^74JjUZlIpks;EEgmtDJw9#y_jNq$^Qs_PIAMsr1~^_
z!$so}e+|b<sjl&KQ#<k|dVS?;H~qaO|K@I*o-WmuTT?((x5bjHt;(_TdTqVR6-#T+
zYj=}bb#z-cqP>vKHTsSeuoQCfyLfRDph1-^PJigo<^p$AfPePU4V@o{wD33ZTuHko
zPR-Z9lP9(iPC5U$CiL<y7Y$NFj5@BlXnd=S>Q%C<P_g&YQ(k3-xs!IYJ7F!)d?f|G
zUdi)Dv+DwuN@e<OKvfcG1KT%m5^-6wCEYQ*9Xh`ox)+t~Lytv0uAllHprI+C-~rjT
zQej^D%6sac#4@8(8GJj2!wTjzZ2Dxw&P48EucEOpI!}UuL0AzZpG%iQb+@nsHB`B@
z=XywPrX-sXBJWz@P|9&3w_0AJ*~wg&Yn<tMsz7#NQfgLwUUfhcr7fy?Hb2|yWwnWs
zx<L@NUXYR={IcVf0L38-3*=J*smX^y?$iHEomP5QhCO|yD}DA)!0K(KYQk(#P=m@y
zM*0U<xAE#tjVF}UTzo^@p|fPB^3l<mU%0wJ+UZc@lq;YU?fE|2vpQIbnmYCgP=oG8
z*+KkBqd^WP{s-<(fw*s%&#O_G#yW<tm05x2+sZPdR~dwsyL@e7U+|CS>PB(6&0r{6
z?~ZP_%&MHvgRiw5b3^X`t=M5~RH`qXsDb<FE5NP_l9`a0NMN0o2x%%c@eqT^e&HVV
z7O_+SB~j3UP!*`=P}EHJ=o(N#>#=5zSRS-<Nl5%A)x=c5<Dwz5)WrCMpCaS_Xo_Q%
zwLMyRkLw1^OMpDQ5~0w)51Z?mj?U02yJnbaH8{_|?*M^JGIh#5|4s-LkpJhzRVqrj
zB{rSqsrl#jo+hRRzaP*Q(LJs|UwpB(@_G>sQn{n-O~POHo?rCW6aYuu{XO(_|M4D`
zb;>jWD)<L+(8%{&0e2G5!yi4QzHgjNg10e~gF?Vr4(9l@%ufobUK6V{m1#9(lz&GN
zhFuA`v-uQ$4@Q}u2e<z3rIwmKe&*(S`!s?K`Lji_=`b$xjQ;aV$)k5QAF+{&O>mnl
zjpojOk>(sQxwc-W2zUhPJYQMCTIz+Ef)D}1Vr<~?Oc<W?PtsFno_8$alfjH4Vlnae
zl0eh@?}Z!hag?$CD)HWfu$7j^y|Di9^=_~3m&Thh3Eu`<h#=bNn@Qd)qgLZNo}qWs
zhRy~4g&Cf!GA>OX@YBbX#$fDL+-W|pWLu@aldGxu)ik!*76Rzu(&F;aQ(LvoM4$Wd
z{gXEoG(sdDekNL2XrG_0hZJ1wS89__m5cR<FJ6`lwL-tVvC=P)siduV_$HkuiJPAN
zCIvZ%@U;euef7d|6v`{?l?GOO01nc<<9=YcezKr(DcFvux3HM-u*5=NeYTgcOwWho
zRjw%d9rWORz5%If;*IHkNs(_YwLc~%y5VkeeUFTaP>;MgTVT2THxz|qdn^0qeDXJ+
z|JTuob}0__vUD^kTC{reMPLIqJ8V0EwuB23V(V=K^F+37L}?48C7wrJ{>G6fv}Dkc
zKj5<9-|f6=#a{<I{d?Xlsl@S(s&O3{>p2&&|8FnkGk~oA)}up<po`x5ZhB43gHrUV
zuHy5CH@e`~yyPSbib#$&R)e|zQ=Z$|Z=QSo*JG%56KqEcmMTZRh~i8b*%mlp8pazv
zf6#Y|RRB{eP-IYjn+wG^(k`Wk#RUz<8y8@_B;#{)@4jxWwv&>IlC<=eNq^!|?uxfU
z(b9nl!GIc;Mf6PiibKvfC`ZMB+H1sJ7PrgYUB8($#tL}I^J<HhQF6|mto=_}70jhg
zeEqzrkgNuuxv_YTGNQR0xz>4ql4cOcGiu+>@>-?N4l+%Zpi7PUtZaEJxVe2~dZTQP
zkN@p5G4$-{9?AUqq&u>u8teYU3$h8t*?m`k2dw{GDX3?FaXh3td=bs3M)%;!_v9hs
zWUv~2sYF|)q?)@@!D=h!r)jU-yg<xwfw&^=txyD_$K$_{E2sQ0AukSB`INla)>3l&
zozh5$l3M`SSc2jSY}QK&ai7@=Iy{=mEtCrev+~fIN<F1`?y+1@v)ly|mJNozn^$^A
z4Qw28-Cn7n>l(R5=^U}+mz*>iGwlt(2zOdL-CuX6(6_>sOQN^7!XG=E;116x)0TEm
z&QG_LEpRaAY`H*+=SXBZ<ln+pIFUaDOiAV{GW(c0${`d-`VyBG_SW|T+Luc;wBC1J
zoctxHt8l))WCL<z6O>6rSd&w6J82yFuj36!*wW|fwrK$!OLc}~jhwAGrZuNfjo#n<
z$L{4?<R#1eu3r}c>WOTEVxWQWHBb*iCN-wI-C+1;?xEX!a5DQ&h#L+eQC=9BV?95*
z-t_DqVv8IxCv2r@+;_^p!FEE(6Lt9kMKcnRSOP)~VaTcH^_;vGi=I8y2Nvr_$vHo6
zS`Ry0-7(3YWiJvOTS3y4*|frDS{n=n-=6O(y1Zz~={`Hm8Ew+FHcBydWZ=FcPNPMV
zX1zcQB^Mm{D6q|mv=T))CkanrOsua=mo;wS^$ZjKMAr#;-cI7~^pu0yI>P2^WGl0B
zXNZn)Saok~^Bb-|K7&!fm8rQ~bq&2)jE|^l|6B1xJpGo5unq}${<^Q@b+4fqvqUMo
z>y!O~3U-cANZJ+VctTPk9r<3xneXLk$7BYHf9v<OG#2H7bojXY(Fd_sEI`IeYi@^#
z#B@cTw(A8xA0<4Vz~RVryxQ{Eb^__^cH_&#yU(kt+^Td~y{ZKP498`Og7Sc9(%4#a
z<kP*cU87aH-i$kw+(RCZd9k$RVeGQj@X>>i3ZMU?Wb_b6yP_=9tLJ~u1C9?y@i3=R
zs2L1CYg|wn(m5OS&7BR-%I^Fz;j?qO<j>!PgKFNkEXv_p-Jj)}DRFrw=BVsn`S1SL
znROh#6V6VVv74U*DjwYF!CqTzo)i9$8#F5>htKDc_fx8>@y?`6au_RY?y{lh{Z%wX
zA?bagWf5<`F9OlwBr4JIs9$c~YE^-Q7!4a2{`4vv_z<op;7_j?CB<@a`0Tv^n|QVn
zxL$NdM0PGa-Rd;Fg~G_^khcCF9nBmR8A@W((?6kT(*v&q7D;)+nbpKP9MOlf3_H6O
zKpz7^GE|AY4pA#Vzi3z;7$EE<q>-EIPY>|*m*)qYWa;6!dF*906ut=snewW;gy!`Y
z^7V}L4!;9q^m3!A9VA|&DU#FUt;hAA@EAMb+%HP;M^~f|gB>=Zcqs8lnO*eP{vSls
zf)`YlT)=lz-HYAOh)_+Y)4lyro&+t%rm-GI3yWKj`r!389|BX@+Dh;PIB2rO%Z@`q
z$-op$Ee_$o#oo#ghI%(^Of@9x$(_)@XUKMJ)iSOwS~MH<xg5)*n=qTN)J~}AK0hBV
zCI>5Xy1OOIm*IEApAA>*zdO?c5ky}JS0nsD0De-m3I3@WD)?mgOU&F|5Qy`Ot&Ug5
zC3UJ{4{HmDP;S`Fa6wHiyN+cXXvC%f3g`Lpj`@(k%Qu_TBdQqcTgEwm`}{EaZB>jX
zo*Nu^_OKwGg0d_#VE(XlR##SR%xF=MVi#+=12A(?3RB$#g;0?zV$B}D=~Ve3%6{)5
zpn#l(uK7=;BhpyXii?Z2G}&g=J8f%yY>~w5laRcL9QS-jSm?aW(fQQ;zu;z?;Dz=b
z(V9*{ca8IzdzW*+Z<H$Bn;C6Bryx{@8!jV+s1kUa6rJsgauP)H)}h%l^HC2LrxGDW
zDNA)0Zsw<&jN8ifQ&u>r9TNS0W9k!-Q|s*8Bw17~#BMz!$9FI3guv9Fj&P8tX!CSi
zAOt6#yYk(?G91{Qg=%iZhFc7^7Jo6-B@?0BeV(aYj0=8U|LLY_5Y4vA1KB=e0DFh2
zW<y@m(dAOMMx!yD@50=2L|mJ|4=SRC)xWTUc?N8>J5nET|Duif<;6k!)MFQr5?l1?
zne{-#7p$hUVnkLyc1|?c{E~0MoL?T3=AHNk)ZWHOZ3YxnL~$nsV=9o-10DFSN_S5b
zn^n#iW-8(22CsG}>4TsRNXL3hy>o!^#6+6AQ^$=8zC6~7+eU4r8mwd-7HKF=P%H9m
z_we%KFZqd!8<jFbXgSYy6$5$Y@$-e2wKdIVtAFTbW9=!#B^d{#50R>>x5}~tv{Tx<
zJ}Ept1}4WY;vR;^^)vAFJWFzTyzc}h@MPRvKaoCgh&oxS)IJ*(e6OVi9PJhc+95%Y
zN3Ayg?vWJACP-4rj7z!gbto=Oxe{hpQljdak()N*r_uxC1y1TfBv4+hM1%0yPIM(N
zGoq^Rw9jsCgTl=botY+@-8&&4VaQed8IFwI+nkRII}R6ssE+cV^SV&bsf9{PyE5EN
zjr@@<z6u&hiQizZK^(6CQ`8q+?E1y@(lDM$JX?;dFs~%gr!6?Cl|P?7l-s#)e5P-7
z^Qf*eMeW}3i`b)jU^eOwnR?Yi8~GxdEm!$Xf3{&Ns0tBn#lC%V5vtk-t3^$}8kR8H
zZnx*27=;!n$y(DbKq#-PG$uelAyHE0QsBY|=`;_By!J}g`srD1<jc=~MCiq&U<jE=
zqVDv%t^$P(&&wvkTg~B8tQ^(t9lQkLqv?(33yW!eL1YlsB8Fs9yT+#chpl6C4efmj
z-ZnGP<n4M%d(>&?gEl#^>QrGj8vZi}bwt-Z^`x%Fcy`_97Wg-mR?PQ<&_ry?Qbu<)
z-^<8&LwZHK{aOM<sYd^Ree6%)HU1wN4-Ca`n#N}4w%#K1`h3<4Db{EH@fC0?u7vZ_
z4yfE8Cf)=Vqq`&?Mwu|TUdMcW(>O-7flXP?$tqB!7vQ*13Du5t-nxwp+a5vMN~|ge
ztZee}yF;u(3^%SS!Oi3#MUwyVY)Mc>#~B>y4qw~azMmFo$gBLi^bSY#tfI{2ab>qv
z0?{+{%uil6*u?2dG)h_6NP=qmSXoZN#LMh3k?J+TSC6wZad?xlx+%rBb$B`<s3o@0
zwdoT|8>{6`U%=$zI2uqOJq3U7K~3Nv)c(0jtSj~j%t;J;T%WTZ;<#Ea39Cjul4%Lc
zR}vG|m)MTbi`How@{zgPFU9vkS*rZD(CRj@%yo9z&F7N0Vwd0Cyl$VdX!gg$<r+#z
z5gY-iZWX^582qfQ#UEIl{)oLYwY0hC5#`PM4{;vG<D^E((eK0OGYq7l&bGE}ER-sG
zKt4o|Cp<$D9Vq0;7M)IEXl4Ry3G<PW@j(}4csbqyuMn!GP?#_$%AHtK^jSm^aQs&t
zy{XEa2xw>)Gk-*ywH$=hw#h}fe94C?CINCnA)+aUf-|@IY0?^4m`ftK#1Jq%r9w|y
zQ#Ef(+(J~ez9+kr7ih7E5lvk>kxUuL?=WFjaKosNDe8tlTd5131DdO&^<6hz!yX9G
zc$F(YenR#SOMIiCnIxnx9UU#KFYK??Z8kox%c)tRE^H+Uq|>=SX8822grba942Ywa
zMxJP5Hp_|sLhLzPK_%D(X>+}8;NtLd*AxxA)%;?nQkcpOzG+qA!9-tZ|HiBTXUT1E
z85=)>WPz)l4r%Rhr;_^R7IFL86w4!R`yZqP=$nc@^rIhcR4)!C2s!`Vb9*`IjP=^A
z*(^h?$`U>cF<ccv+>k>hm>`+>fwnA`M4!B$Ti+Gqs3cOd+I*hJ**%hRcY@J81W{76
zK7V@0>!m?ba<zzus0h^24Wsqj=g&X1F$OiQPCi2QmrIDrH$ZY)$4&}uDDfS!iX+q^
zSWMM%1@wH+HrK%#J{x}z!Y&t6l{Y9PZljvS`of8eCK~<q37(Bz9`Mr#gc`)lDDu*;
zUT^DmZjRb_G7nR+@wS~&E{s>*=q*Yjr4J?gdxgBu!OM3$`N-6-k`F1-QmR*maCp9$
ztd7n@=aurE7f-8DSUZYRUUdbgyK3BDdGZgdP5NT99MVfhiyER|+zBqqvE!+FSnGsi
zTAD;SjW^Z>uMN3NOf`)GCQ>&GE5p^CWmZRnS|fol4d*)LZ>Dx;{26S*W$Z1bwI8an
zR(+EH@gg>h{aR(Fp44nd-NHum#zP+tx7Tf+eikH6kSVAqyNcT<KEs6#&8X&^H;ntF
zSF`CRvx8F=!oS?f+ZmLrWU~Fel2Irsq1l}00BWdH;#P|0?B6Q~I88_LUU@z#+f|u(
z-5_ed9olZ@+!e`>aNCeq;GI>>*$n2&AR7r!c6UShPL9w1oxer}*Eu?JS0Jt+EK`10
zQmG%?EuB>i{Nn5!%i=&JvzhYiZShl&abLo(WVciK^aD>IyFzLOaW~>Ptih&6pwG~K
zW#Q_M(P5PiSP~X!F4bjZlN~(lW;1VLP&s3yOn6Gu{ETLV^?_(*hjtNjW#V1~!kY83
zBYd-Fn~7iH)*5}LdK0#RqD(<|Xj-^IO`3t=6Vg7bC=sA1Bzu4C+y@aJdUDL7y0Y>O
zXQCp?4#nbshjf`;Cne~hU6GG#aI{{$_JS(s;vJW~+g|{h#b`ER5mxUtqF`GF@TVll
zP+y0->*4<NA9J4WS}zJvk8Jc*stYr!$-R2YCNU=6uyM68L7K(kq-#g+!5((wBQ~jo
zV=T}5377C*mOH}5WhtUR5FWmFtxu`9gpvV^0oj~R`x|Gn*F3}owBL~beeqQgF<idh
z^a+`l^9?Ay*C7z+y{yyFyFvd0)a!P6ptbtTK$WY-v9J~)?v|y=JlBMwHGR*)!FJDz
z<bhCX%=dY5TX=ES4(?-D&^I{7NGJHj<zu|i>-Y<?*YO>nP-2bu?rc3q$B(Q2v&=AP
zgiq2%Te9!4v=?8V1GWRct7Lcc@i)G%s28V4=MjMVmX!y{8HbWAuS1&fb7!H1K=RjN
ztD||j{BC>9;ObF+ksSAH9s##<%+<yI1m@U(`lV^+?l9uNmf(Rk8gC6$fbbb?C4Me2
z*=~2T)^!~Ld%nk!@W-R|&KG|xyL*7R{9_*3&jwlP>5$k-E~DD3+Z7%Orsmoa^Otcs
z8h|8;(cgT5UwbRsXH53rdjHhI@$vqCMEJ#ihc<se4z7(1NjeVce4ppqB1nE0D=OW|
zA+>r^iNb4@^+_dH?;{Ldm~0A?$tgn}rTchpw!Ptka2WhCm*Jx6pFt<uFIG7B_ZlOr
z0!d#bf8(6HHGwnPHWT&*a}05B-|yrYAYWbZ6Tr0lv2KV4F#ofJbw@tvN)pI!zDlUz
zu@<)IYi|sZ?jGG2zgO(ZbdTI#+&@(7YORgQPrMp)^M1Z_$*V_i_>wFQA%^h<G?Gf=
zVKf6hQw3@iAhm?taH`n(p<j;2ycV0;(}VGdS8#EiUJ6bE@mfaP#e$!vme!j@;KP0q
zO3dRYq6&%eLdxIzo+)G`Ws`JURh=UU@&YW6`WXuSPkj?EKQicTntQSkN&qh6hxm6i
z$~<t{)D-)tcrutDw|Rh^h^H?9)zO5qhv9{^9y{QhUEN*`HM5EbsO0MY$65D7@FnV^
zBsI!fM}+kTyYKj!Jp(TRXqyKxl89`S3dUUmhws1#3aCGEz#WHZ9QI#?M$9pto~MU&
z7UhudFFV0Vlw^xzEBY}H)dA6uFC;*-v%D!>;|2Pa*1u+6f9+gW52MIz0@Y8bS0^z@
z%41u+|6(AF+iIQ>w*l&SJQq|nY3Jt>ApZ;`KjruOt3t*-x*55_Xdo$-v&@juVwnHK
zHIyRT%lIoGioC^Y;XzT(h>k;a29>;=QIDJshT@UZjAiPMXA3V$KHrYD&qIKc$`(QW
z$|i{Vet;r=P${(Xt8d%IjJgr~$l3nc{@!YFNF{OenB0!3NJFjA`M7R&e|=e#UBnZa
z(^h7=aq8|gR*F)t2VQ6EGQ%WN{`-DyGN2;EYFwAE*-}L%@Loo$;S8Z*Llf;)Q2~k$
zR+2T&%2oYYnpf~(JCvK$Px|rbX64Sj%qj_9<ePBSJ5;4Re&J^7)hi;doENu7?ks49
z`dPzhlLiW{xQzPKiv`IyZ(rbSNl;r$OCWImbHj%vzLl*(2w<dos~R;8Z9;iK^Amh>
z9IRG6A3ws_<cYuDS7YgKXl|Ph=yD1)Gm1}kgnNxkES;CuqbQ_*{S13N;{@7ldqq*h
z+)h^Hp3%Vk$+oP?L>6+ImWo60FtJ}%`nn*Zzk!$kfl4PKVXObmH)_}1<nQ*F{FizM
z0vnt{M{fkV0ZV0SYTdt5BR8VCHC5w{L>Ve)0>9p9Q&1I>%`~J_oTtfgw0MQkN57cF
zbT&|Yt$*Ssx6SOsMH9h#zi+$UQ;#j<MiSGn^bAp!rS<jxi~X#`q8K7p9vvw6TY9z7
zuE7(D`eff3798Rd?d4)jcXvd3bs)A~ZdHq-l}9-E90=8XFQ9d)NS1S~xLT2V2{FRA
zp1?Q51E06P&HAe3jTzm59n2EWq#Kjnw1#~ekazv{mf*2%WV3%k4OUQN1IP*vB*Xc=
z<TSsaD_Qa1PJIbMB|9^ZY@K24<D0JQ3E8x!HSqU!IW0L$7wd9@=1bKkiXQw^nUD6u
z->o+TU}QC^e1Bk~e=;v>YuRcwNBtmy*#AS&yo;AdO<qilHWr1(d7fNAak0bwepO~g
zqQ89-jfyDND1#Kf{G<cvDO~&Y$;ZFnQ+#r#*}gmRYr@Ba@vyW7&1g??M+Z?c^_p3p
zOx15@%zhJS=C3tWYLr)Bx5wVeAc}-&+EwZTd~Su47};pQ<*7ERHLQ!quV#@{9?7fb
zh_iDXh%QT&#Yuam-*bKkLD*Z5s6#TI>ln^TiuUUkTAPDQD$4DAs~|>YZ+yI95_Pv7
zcfrx6VMn-Y%C-p_v7U_FQKO6!<<gr_$3SYZH*ZE(V(BeTzbaI-T|U<pqM(5o&k*3x
zeNq@t@H&{a8GR)=j9|9&i?$h2PXD956+rteo>QQOe4=KFo=+Z}M)HJjHwNX!i;M{H
z<$g6GD=i8XUPxs2Bp&``HyzzJxAG0Dhj8^#QwtS1sZh@$Kb(?M1`J(F(1wK$ZFRf^
z&6%`lR8|jXJ2Xz+sa{xtug#|#_(r(#*`#rv1$-mAF-a(F^t^6o=zTn+rJD*5>Ax*L
zn)!k2PglGuc))hA4y<MuSw3WeL<1_1ppMq<EnPo)Up*DOC$!7O6x1QT9lC7Ec;!M>
z?2xxZ=<zGV&v)8yQTmYLkh$Fx-pZI?r^?yQ9yd8X*M@bHRV!Wj$XMyAAz<&QM{R<%
z)hwB_iE8dsE2U#qnJPo0TEE@`U#o{)U3<@pAo|U0xu}Gze0AeoHnQR|5V*3IDfKd%
zHrq`ggwnD=_LYmsCo{)#D{m-NVs;wMry}BebTz{IC%pI`gvh4wl7W|vP!weuwfjN+
zgsHO2DALTVz;T+Ir(2}!9{_)|E~gzFERL=yf4tgwCv5cw|95!+SMM-b7&1h(&N&tS
z(Z$OW8O7O^5_$Yr9?;0a=Zi<G{?VA!>dhhHoPA_8miPB!w4)|J`_^H3vK9lKDm^bQ
z&j0?M^)hL7Tq*yAW>jTz<1q>cDINqntu%r{pAN`_6B(}Z!$%)d06i-O(bsoK2CqN2
zeV{Vf^)?NnyP<DdP?NpLXzmh(7^LzU&PlJ$yA3XreH$%4|I4z%cwj(M60d=2A-X-W
zL-w})N_CL6NjG#73h4A@RHnGmV<2fT?m3gz@3WSu`R7D?TML+6ovrai23NcLfBC#~
zN)Tj8OtG0Qb5{z20CnYdhjdq-*SLsD2JMI}J(`kL%8wJa)>VP?l&0<3+hKk}tCdPD
z<E6p<vMFvB;^j^FEoCztKFA5%DwuE{b$Iq?sDHbG7ZFL8`Chyzn84kPzhchFi?c(&
z%lZ}tTfR(lj?EoW&@)ewFwstBsT@|ZUcs#=$W%Soo2O?0gr;ZHLvg<g<tcKTxxzJC
zWX<lgw|LTgSr5=xU{m%0JzGH8()j*3@qW#l_U1bGtB)CZ)_-S~+VSiupqn4D`@c``
zI8TT#j)@nr`>(aTs(d6U+4|{xi8Aabf5?vmDEn4?%wE!EC>!CNUXq*IYbsU}sPWm|
z6!ae$7R1^_fz-Fj`cSw)tRPsAmhOvqUX2p51l>Cis93amrk@vhiBF24e~tCfk5KU6
z(<{(L6t%SH8Ms?GdjB!F#Qy<NG*imzl*AK8{AP*-m@Os}E6M-6Vhfp~?=y8sA`1tC
zA@a=W3QYX+;&~iY<?-9W*WhLOd58UERg$rtm<c>J(7jJ93N|DR7vP-CE48%Kj&pT7
zJ}I*}<!4H6lRBESSM=*!f~BMyO`Fq{E<zk{ppq{Zcfh7Gj)q`R$p%Dwh5z_-To;DN
z*F-@9UW2x2iM!BJYVUqlR&|J$UYnWDK}pq0n>glFd`hv$p5z>Kv;LgqH)jEc<<mvC
z0ovI&5!^xi84Aim|Mjb;+d9rO13V`&EJh&~?fXH`CD5ctUO$41y1Rejulc@x9%$<{
zx<!^Fv~t|H5YiF|sx5O1R955oru_N_I_rkxSY$7MQQ1Al51-Xll@YeDt*v*mZtBQ`
z=NdThjko7$nVKZ7#>DE&-d!XXKT90|0UpSk2JBqsYv}0w3=|?o!?*a^pCa9FoXrE4
za%67f*O**S-mRC|YpfJ1GlCQfCg!a>ZJ$TuJ~H2UpxOS~Gyj=$M~nmr3R}&2I2nDT
zKH<MN*|U^^*;cRQ?8IoJp<bkSY#I<haa7wou-QkE9#mFKOUo=&vR;BdIo%MB0u-+a
zi^@+yYh;B&=YR7G$8!8wXy8rc7w2O+Wtp%b7hIhD-5hnk{P*qRq`aoK29r>5>!%Io
z&y|BK+5v2Rm5Fk150VGZqbMY%T~%zRR?N*@A9Hqo-ARmEYp*jg3gvD)_tA+Bb~15b
zy+`a(wq)UG##()M(oWK=Wg5%1v;}-g%kjiXMC+;uLxxaHSPn0Hr8EL;U{B;L@pR0&
zmdMs7EnKI0|DU2l?XSi<o!mg4!^}q64^Sz0Cba-u+T85c*35zH@LxZAZ$b@s0pUK_
zi>p@bz6fQ|Yq#(_pWIq`CN4Vhx)3%GG1n0Gj3$X3Av?!cb)W=l)8`EjSOG6sxfhJ<
zX|m<Ick=ZWFE0O2y=<j$;bb|;6H4LqDj-AT+Cb8iS>%SH(TP70t{!z16n^2ST9o9M
z=er5b3F~?<GZnPHg7;J3($Zvh`kn_X?!3YEt(80cR#4T}5-G9xa#sd8>YuRv>u`Yl
zG_yJ#+}pPn5I!rI+VA)0DNy_%qR5vN(b?76`Dt^}aF`fSyE>{VQ?5h<>Y7LMHr!D(
zKBQ55%izX7-YEQzFPdI~#6ld=4w=M3pRD=R@F3q@p&4oZBCzfFt7p563IZ03zK4uz
zmm+0tQCQ{BBe4PPSm&r?lX182OBX_U8r0Pi?<5)Wk`DP^4UoMA0LTb(&&EB=yiNK`
z^DMcUdVPQDOtYcd-f`UC3VH`wi@Uwezm(}<Besk3bj&8_o;!-#_j)qbe!Buy?oKP~
z;aV~^W)J;u1*N`YFX3fXQ&0ER{gFG{@r9qSkEhIMTUZu!rLkQX)W6y5@(1#VQ{ao4
zb?bT+s26RY{<8NGP*n7wpTV7r>2sclxPl10$2DcoR5RvtZ$D!BN{_*AaN4Wu{OA~Q
zL>NH0dsq8Ung|^d309}55UsGf?uBe^qr4p-TBT$=LhlenMfy!ROn(u;I*>|(36e6@
z(cwM*)k~;VQnZITU{TXtoz2UI?WygbR?L*jLC`SRHGWn=-1r)OovIH#H&t4aP0e<8
zTf6j9nEX)VC(_m{kG0eVU>_NH{OnQE(ca8fqGrn7p!WJN=C!q@30NI?SDo+F)+QcL
z#<7K}zfVmhPP$Wx`asc=<hv(*cs0m_+6DD^{Cd!zK-}*!h+Hlf+w@%FuH05zp}3@3
zYP4h~C=cLHAM2=7l7HJ)sGuMW`JFnYD8oRfUvfQ{!|I7E7=c7qXAh{|WF5>BznMIl
zGKo~BeF+zN^8_+BL|EBQzS#ou6`H&E4DT{z&mHSboxr(yeNog9G2F85v{TP^K_MWM
zyLP=4YgeFeq+jq*)Zj$8cAjg2u|S8JLVVJeIvIE{x%QsE&GV9tX>trBSJ^dR%xoc;
zY9V-7Ipo6!b%MdC$WvtBO)~C;(w5eOk?5Z@*tMyU(Lj|mTO#+%M~bZWouA17XE@+A
z?cGshHc4J@D1e)^<au~JwKSAA!IG9($ICr48bS8rbw&Zl@)kMR8Y;yD{3EgqQnLZ<
za7FTZFKS86aTOLg4AXD3xj=j_##hdUc9sQI*Z4ayf2H$(V!e51XU;#RS<>~`8-R&R
z-foyU^nc++m3pfQkt#yUQ_#+?uCt%hg4z+|#7UOEq>S0=4zIiit=Fqj*}-0Rg4&vA
z?PsHiqRZ2Wi;JzFvdJdTdR8=<v)zFke-H0OcB6e;7O`0T!PX|uA220OzLEItGb6}c
zIYe5p9fej__{fR7Dq2!m<En(WEkCBeHAtZ4P^WdgHtBK-o+M^Oe4*x5#9S3;3oyWs
z`4vp@4jAJ<(L<J*V={c&Y!&>TFt&r+u3S&puBtYJaYWw&DOvk<v52W0I0D=~Ka_c%
z=A<PONhH0N&6PGuR8xha=0~kq=mr!z6-^^&rKsejP4?exi}$Gow?6ZreY%_a&Dd%B
zrL2*0rae#o7elxi%n^qn&+V&GdAu(7`8Iv#Hk2CdFol+^pG~SqiR2?8=&`XM%j&H3
zXk-38b7EAEbqm1<Iaaj+c%WSzMK7j|!y!jO8{~G3w)2H@9j8jzopM{Si!!WB(pusR
zMm{pU&SSW&0y1-$*BellHKoeDvVp(8i2Ss=Q$N}vW)2*!R~BYgx3<i_%CA91@tr6i
zVMq1A24&(}V;%3pBirp@ttKpKELm3j2hl#%-=htmMvuA7A)J6>4P6^}xg_2Aa=Z_+
z$qub3o+<Zdkf|b(>usox3U0ngr$K_BO5le2lI7dHnoP?EKNqU+ZZ`XdO?7CL{_#SE
zOG0c+l!Cqz3|$irub}PuvmM?NUlnfa>&0x&C;Jz!HNhM&PM0jC;B98Wcj@ROq7t`J
z)kApGzBIM4z65OV&ulS|J;DY%Ws_HOK<h?hL(Md$N)@(IMR=Gf%%t{O^ox~PLvZ2Q
zF*%xOK{lY`iW!C<eTO><eJBhe4h{9M7XPcP^q+ADHCSc;{dcmpDRX(gg(r`Y2~L<*
zlvCw_a!!rhWx_oDZKkL@Vtdkd&yp-8%a+-;wg-+&fzA83!xOPr%k)1~K*#bgn-i2m
zbDsYiv|`3Q853NbkbM~jb#@+i&~%@~Lj0u2!<KsBcobkC9ZC}=louirts^5I70o_+
z=!N<;717#Mj!!FYOKVpTc9^T1f=*{mCdVXbDY#Aj3)cc7BTs+-j&S0e4L`lu`Ws*>
zafMw4_^l~mK9pG5TnDY)y!t%zsbNtljT5SlH^bf)*@Nk*%Fh=`Z=g%(@VDIGcAi>!
z`1&RA#|*<q16KJ?Y|iz2^+(qilU$g5Q((C#Bs!+_^C|I3YDH30FeO{C5o^OwbY`=W
z&ODnTs{>Q&V8-{CbS~v%Kko&_mSgRV*@Zuf0!H>dtYqL`5+6qhi$z-h*LJjmJg<%e
zgF2mmF}o0J2`j=WC-A<c5j(rl_Z%`AXG?RwLbAv(GhQd_R4e#cnaFqdH`Cd&56Xrn
z*wNN300<A2WSTw;C@`tsh$>jAm#N?MYWaE9Wj^ih=x5s>-!`Epc{O)L)3;mdRxa0G
z0!d_PJ>9iK@u-`eIL)CETUyDciHA$0P#Cn~Pvmq?%N>w=?C(9*pyT7c(3xPh1<iRV
zaszu1xfxJQ@^9g7*D`w27nYwLW}Fw8(^jVtZ-pgy0@UMb^ADU3kC!SNeyV}q?icZ~
zgcj7V*crR(BAD;_2%Km#tGPo04jKW;zUf1LIkixB;UJYF--lA_POEmW8V2B&Q0ptM
zeNDfVY_8EU>D={Lk#n)Ue?8SqR?)f>@4BV(W4PJL<q0*8Ebq-ukFlmqrpFG}Ld+p`
zB`ZpOZF%W4-e2N3opcS@Jcc3c{Z?)|!iH<fnj9Xf+*foxT>OoVtJP}s6v_Za#?!?9
zCDQi(<W2~9q)EC3|NG$d;PkMKFoRDI8<P+Dd2?jj>dwTSxIi#19wC(IW4OQT7aBUl
z&+<W?MgN5HCDWCWNFHmRTn!zAc65#}`L39weYrY1I$v?D5na>QIIGzb=4tZ4w!60E
z{6Z`eK}A66(+Y1rM?sCZ*@14n4GJqbYjXZyO4qwx(>OhtT$N39yQ^y;{MhMjf?P>=
z?ZZh58&`*o<E`fT)!xZc;n@7VP4aY?ursL1n0U@1cd+Ca8qO2I0y7gQ1tEWF{P^hf
z@I>>Lu<FMSbC{<k)7a;oKt>~$_9*RtdbrL<-^&(fzHe>)!)Ca2;{gtmEOTOFV(|O#
z^6TEw#-{!0;p^XCAip>UA`p*xOkeN8V_KiX^zqp=p$Ky}f@{E`nG)!0OsPK5PTaw0
zPUZ^9eU%MFVkZ|Rna5I~)(SA!hZK-mzwD&3fmc#K3_&SHF&b5yr_(Dw`gXxjZ<R?s
z?2T-5;=ALnD+?B-0<Jt;>DoD58fQzs@sdt6^*`x9$|%F{C{G8UltT4&QqNif&lxLK
z0WR19m!<H!ezZ|K5cx71l<sgsz+in^@W%n$3ETUAPeH<*E1x$}zuK_wQv5fmM2$u}
z=2m_ObUOK1pHZd1YHO6FjhAdcHyV6eizK;AU%f%vwEznV0brK)+Qs?$QCu}qQVINK
z_hFE$?zj$P9cgy5ezxfu>pp&}-&314M9&;vL}Vhq9iBNn^;cn1fq$1oz6&mb-8q+C
z9?Se#ZBS+ZhuwNSY>mDoTYVneJl#V@9XV#Y8YbLFjB3GtH|<wV5)DvPqI^~{UCt#T
zA*uU5X(sL~0E?e-Uk@C2fGVZsoUd@qFmL64h&QGsvg#xSJjrH#SxUuIR*K^i@DS1E
z@^ICi_9$0IO7T31_akzDm7`Li;;3?jjKrtCq@Q<R#P{$>)ci_f3t|uk#Q4|`^VDNR
zLHD|I7=PNXVj4JAbTSEBsfme0{rzq#DFdINqe1OJN7tjrU6}R!M~FD`cHSR_;6t>#
zK97DgOep1&z4g5E9&>x8$j5$9Uze73Nl)sB@d=ED4Wrd25(;vy&E$TbP3MafXXow<
zxWgR5ahx7xvw6^FYAvB3+^=%hmEL_n!rQK)<R_kMm%B_90-yClqRu9Z&h)Nyw%F-S
zq>=qEYaXK-%87=zyDeY@KUi7Ri>;%I&>x#*_#;1uNX-XzkbL2Ub5|$dV87gN!s>%Y
zJ^{hKP;j8)#FSZGyXMwLgp;rDkrPdXj3;{Z5xhnnGsxKYmA9?*ZK+A*cp@qMG29B<
z{0aBa+;I8)tA*ND+K_vjRLdsecIdr<tiOP&-vo314_->4O51hyJDqbW(p$<ns2eni
zmxp>M?CQ=lDxt;~Gt@uQ)061YxAS8p*xAzW+t_`%YAi01VVEgeo?XM_-M;Q7`5ICF
zDj)WxnaKC2bR|eHkov#V*xRC|uAjC7zDuwzY7(OrUl?Qz2>+B{QDi({Xjm=0iD+N~
zY&J9}hNs8!t3lCmcKQaC-!t#MrtwJr{Je8KP@){hAB}6I%$2RSb;>ei9_s%ca!fv8
z6(??K^Rs9euvSQqs4|u-DkY-o%CkL8s;Anz%t)j7V{$d)>ZT2*O_NegzxC6*9*yyH
zVL^Rs#}Hh>y|RweTf*!7E`Rq&gD*A#e4FKJA*;yvBjN)q+A*w`Lr%F35><CFQ|YcE
zIode(R-kPl(wzM(pV+qi7GeqSSVIR1a^ymNk8QQcq-Gn-lKWlQ?y6H(x0y)frfM%I
zLz=2>ev6*_7=oFjS|9Tjve&z#)ee(tu+tsNetO`ik8oF`!IZ*fH|4RSnH&H@?I3dF
z^eI|7)O3s;oGY;K_*uOEK%U&^Wl^y17>K*+?aF`1m^;FZk+&p_*g=z6FVC8SYy2wb
z>9mVRna|JYzkcILz%p)s%+R6pS)JoYv2XT>YVNNquAB=wM3=4qj{m8~Vtt_gFpgGK
zBCUO@e__EBp=o4Xhm;;Y*!m5eoWNT<KBGRbPSu^BMVuV%?}ln_CG2f!60?)otZ}bq
zw6<6T)6yoEjD#lO$N&BXsHzLu)f*RPF4vi4CJlF*k?`d9w7;v}XY=mIsgqV^g&GRh
zNZhFA1PmEoj3=yr?`G<Ww00#tvw6^;h+yr3G09pS&o_03`HC84ZvhAug}_7mVM0yY
zvoHmgxEoBn`{&)SqBA_gHr6*bK7ALxk+S>kkMv2uru~7a&PIf#8XdTodZZ+={TTIP
z;<(7yYc9f3-<My2J~fH$l)W3_eWFi!7!cJ0+8>nr116&|E#Cwm;BfEM(l|KoUH?O|
zx%F><YE?VSX6|lO{vaUo`!6l4Ktk5|QuSTwdW(AJj~JzCaiLXH$JfENfzPMhoqADJ
zsI;l!FjWa@gW3l)O$Rk!6#XRJB-Vqpu*bcdluC^MtCYU_W+|5&l6$koaA_r7Odw&Q
znSh1E;k9P92ew@%Ya#bMTHw}@cnX`B5aG2mk5;BN|D5O<1prT&<b0WtXDqR47hJmn
zw`lsqgD$FNWMkzj@etU?2D7ciY){`Gne24RU??}8%4JD(%}<!Cy(?(3<R9*2ci)dZ
zTnQ^Fw7Tut@|q*#Bgo$Qy`LC$L}Vl!O($MS9kba+ZYZ@j&!R4GU<_+@Dt_P9_M$ef
zT0K<yMyAZDIv~^1mX`=#T<Y&vBo|HuaVL6mv+BYkJuNZ=B#t9*=WL!F{9aGfJ_~i+
z-0aZmeq3Bnf)l4BF*;+>zY2wP`V{VKk?YG6OGQDH-{-4tUznuP9Zk7kM8IT>72mU4
zk#`~$78Vt7Z;$MK*!IHmy+nWRivFOU-DV^{n_i+Eh&Q;q$<~wiqbSLfTOlBqSI!rO
zg|<Kit<;yP+k-W*NcplFpI`<*kEgVKk@RaDnoq2(otPencf5<aA;&D2@LwC8O(<*T
z3CV?}kK8-tBf6SpZ?$^I<xeWMVYlnM0s;_|ec&<Pm#m*lF+y3RZ)=V56Y+D+%Ca_&
z|A;mPFlOz4Qz;H#$(Bv9-z~|17{~RTtzAooa{CVO<Z8nEHCApz@swmyg}VZkX)Ki%
zI}?eqe71Nliqyqiua7)TJyhv|5<^Q%r!19;fzl(L@KJbFSk&?Q7i4{9%(&o0C54tI
zWB29$HnCrCq`VsLSZ^Tzp<I#4O<llZXvvcYu-NoMimF=t?x&J_t0be@1ppDDS(40c
zpIWrf_MCyEg?Skn?RR_>Se_qtGWB@U)TX8{f>o)*Q{!Ko*1u)59u)`c>o$(@`KSt)
zx>*}XrE@7q6{m#7uA!8z+T(&A!ILxYf1fZj{3CmJ(C*~yYxA*Yi@$NAo520?mujB@
zXp=g*VBO7JM3|v6OW~*14|mt^5@8fUn6^jt*!+Sxn6Bx<&wR4I|C22jGbC_Wtz437
zc%_NQo7v?x98{W_=xR7bdVsU;q}iC2#3FBgl*oDXQ#DK8R4QKOp=nQFE2AS5<ls3O
zXGZ+?3CWS)hMMSGzYZj}^Q>>*-nk3#?>K}vo!*^96T%}R#K^Xp=FOZM8oA2%-U(Sr
zy40YW{Ob9<K>-13W{}f%a<r+q>6m@va;HRUuxACS#tXva2D8nj=y}wMs=B71UhQf9
zWwxf;N%}e+U!vU2A09|d&1(;}_Z6hQ(v&z|zKLhnTNSb6vQG!CtZL0L&2Zz9@+@$l
zU+!SPYE~hkNiDvB_0=vk(g~N7Gn!LRhLm(l<AZBWY?&B2sSI>f=BN8IK)Y>MdrODI
zxHSxpTh<BD_s<kJYh6E3=g3M^qp2@L!!L6vWT@pdwuATrukHprWjx{jnd|N%sh{xq
zqkp0K>Qp<+vm1Z?_$m|E&xR$xp<n)75OSc5tR;~~Ews;$c4TrUa^N!59)N<YG1*(Y
zKDLuyj#s9nNRF0!lelNk!qU?GDHsX8+vaJ?&028$6UySPkmlc4ST6eB((6Ajxmw#I
zDMk8nNaE7-1I-WX-du2u7%yKXM6_S_z4;JY+)$;zhHZFh*_}^+fKc2;-YQzXNyxex
zsDjeLTC8;_>4iOa+rnK{KmkmcY=6c@yek<yGFDyCYXlnZAbMKnzkQlP9ZkWQ8(*5{
z*!$`5N5^a5e!lR3$4;h#bm^HNB?mYuT5$SsS~fSEzZ-9oiEyNC`z!-}?g@3(&NW!=
z%2kkr(FCVF3>7wzYvP*i%15r1w8WvCJYh9F72{cu!D{M(&IB_6^2k`gn9bug-IJ$l
z=GsVn<5){rUe531-{E;UG$A?~A}XA$QPr0Jg7vdzz0zI?x2U$co^c*<$J8Hm*3o3}
zOPnU;qnHpI#|sE-#U<&Uc&0eVa=aQ#e3KNcyh%1hU-w30zc5uvyW`Zji}lpD`u433
zB_y(Ss8Lc!Gi3@$5|?OoAT|d(i9E?qaOI=aagyR=o4@-$nVw7mj#cPUp7<}Dx*xoO
z6|zCsRvQ?L7lx2u#CzgPpm=0;zJukJ^y5x=O1s9u3vqk)UrQRFVO$^b>TO%F;KCjW
zdUiGju){<5{CmaXJ=+Q~#G~2z1%Y26|6B1vlge)UUE%;k;P>jR1=)B0hH9m}B3v0H
z@|ZE-d?>&$rVGwn4}+OZnc=+9l>hUHk@n7x{4~u>e>B4!eYzfh!;*TgWIQm4QKw^0
zrYprDtI?z}2`0V#gaFAfaf#pY<WFIua;HlJIFn|%^)bszcv&liZc#^e{&Br?)XA}S
zcf`3X0ILTtxE9Gj`r*^E72*U7R!ET^jCxCOv4~MPy2;a=hCd0DSCqy7IH^H5ZNBYV
zIP%yD1?qRvz1RGF>->6nZIE}&SShJIWp)vf+g-x8uV>4YqFHr+<yF+7nn!<cE^Mw&
zEiA|u<ps3p#U3WoTwQfleJ`jbnS<ZlJU9slnEV)?jI6*gmQ)7Mp{rp&HLD8diAE5<
zR@w^GINGr({*|fSa;VmmX&RmzF)^i5wrSe~ns2DQLmZWnCSNN=r&ReX{J9geM%QKU
zo}c~xy}qeDvDxuy6IxJu#h?lvtwY#)vFrr@!h7DAcjG}g-)9Ha`wz`d-oNo&;d(Jz
z3t*dt1K97ah0M|!vSU$7bT6W0LLB%B#cg$|v>C>?_008m-}8NBgV)yDUWX7&tJgiu
zHtkm3`B|<QH-0iP_T(EP1Dg?jV2EM5iE*E+Lje}reRgJR{3^NvZPTvo;y|7kB*~_V
z=&;Z&6b7<~#Jzvxc6TYZd^N%W4#%?tKWq}L3=TKj)@zt03m>R*1gNaZJ;>|}(-Gu7
zwW!T;6cH5w?qJQR&yL+I_zE+sO{m`WIZ&t1HY;1li#sH}q|Vz>5tqxX>V^MG%24g)
ztb^i$TZtQu<D(tkcmfaos_U>C;!#PByaHsO<O%Lwu2~pQkq8|B8U8rfzU3pp?h4|o
zjH`_ycN|~AaB*X1+Z63_g~njl!Lawdq&->wia=v<d1@t2ld*^JY4*J+JCU8@Mb=IB
zK4Qa{XF8NYMJc|hV4SywnM<Sj`<<pGqAWb>`g9u<HwW5xUS6r2gx7a{`An^olxfK>
zv7~CAxE`U^{eo|hf`S}T!UA-QLF$@?6^P0O#dOrx!8P0#6tC`!pY5^yX`isCdbf~e
zDEUT*6f~!5bT~V$y5#9A>Qv?hcy;h&p1F4Pn5bI0aHW5GR=O%vVuU8sYM@pTDHLMr
z^o_oO#67kj{^I!TSQ|K-0&=DLg>0B-^Oc_~y?0*OMET%T-*>lY11|Gzf7mtFh^(Nm
z^?y4X3`jytkD;&$;t2rKxSUV(g$NyW{z~<=_;A5m)*^j&ZZV_+;$aWDd)4w#xK-a9
zon1cXgLku4p{H$esmOE)?nILTuDHZRmVmKLauvh#T&j@bPDU~|;qLONez%qq6UK72
zB;K1ko^!haj!FY!^VWmZf4g#4VEMp>RVhaIGW|RtzNPT?8sR%V9zU9#E4b`FAmL%b
zG#@4T?YYLOm+bc==xP}#L}Ya(BGs>YZB#70@tem42@11S7vz1G_%W_Vr(jJD7N#xP
zR+~XTORAU~zRfOce8op(e>`Q!?F&)AmY5c&^B?n|k)bLeIJg(2T4*t5(?VW1ZMJXb
zBIDXHQkqSrSDrjwkSsKozG*gP6F>`st*o$0C#%v^&ec6cHA|8P?mvB%R59C(tRqJ)
zs=*NB8_9fS#W5H8r`L%5Y0>tu+Q+tVM=Y74Bv~n$nM5-NQj10Re5N{ydFhb(kq<;0
z5_sO2?+1Kv`gL%5$;OI9i485M-^^?Fy9QxhfAey3SJsv_la9G$PEkh!(^IPuJ}w^`
zv<|??C1#GBf_A(*Ed2l8?`4Yqc;D?B?9}t7PXv6^Lm<#tl3En)Xg}rUL3>!ilp9Fh
z#uokWEuox)vHGKx)y-%>8l?u&oA)wYKR$}q=N|(oCqK9q&uzOx*2{)8^i}v6OC4-b
zH9Txd#DX``1BL0Td^YImz7P0$&?Ds3KRxy9Yug`hr4SIJ!={G(8X4}~L$3K1WK;q9
zMOAEVUNg(#fcnP*dDi!bPQVhQe6Yej22oq}*DM-ba^<%2Kp?r_{ThT=FQ({}&TX_c
z2g~Q#zB_HZniS9fU`RqOQ`q`;|9V73sP=mS>pVl_w7hVR_F8OM)cHC|@bYX4(26`C
z%Q;Y?mZkWXBc@8NL{XpfNR^r!Xq_#i5HNgXMLV6aW{9E6cHbJwHr&sBDk+~HsB0>p
zG9sQ+SZ|7j*5aF`*|rU7OQSbRJnGPRFFfNoWO)wC&^N4`BA4}Ximr}(ygi1C>W67d
zuv1QwYy<;?kZ7s&$U{t4QqsqkzZj}F|3}k#hO_y;eLM(aC#9;kP^Go^mKv>H4Wafb
zEo$$1qe_%oMa&vMdsBN?)LyOC8dbGNgT|=+<azPWi@d&%`#A6GI>-0?e0gm4r|o=>
z;*IcV2ceC2*8Kf^fD`-`qF8$R?_~O^TAw>4P9uzuKVj8fL9c|{R^pR`i*u@)rxs<?
zY-7_qD8)1MH+|2i^}_eo`+6bY6{ZDy<@M7wn|yxWpQ|+kH#D}*RJr2f(Q1lu4D$SJ
zTznC*rOG=cJl~ets>;X8cL-f<7MS+1Y%1dpAIz|Viwx`y^w5g#6D&Uke>qlAhhMW)
z;Xy<TMoUE0eC<s;u3X<}9;s=>;kGU|7!~x{vQ`}?r{Dd0&Xb>xZ9;)Qu>)RDskii=
z5!M<1JPPbx%=m9?(bUD2H`{ak9ue|^X5`%#lPbz`l=QwR+geiMyTKf)rzUE|qg-`7
z8|KN7bpGoCz;gvbYx|@L;k7R+f5@8~#z`IEg*py^B0;D|f*K=CLqk#HfhNB_&kMiF
zVw;~uc52oePUWt>Y9X1doi=Y&(0ygpa1{-vM-bVfLlK$*tT&-_h$igyZMni<tF--)
zs!6S28>Gt2YIAd%`JDLX!<wjW^eWa-fzZl~iem>>Y>C~O{fAXnOl>R|6qc)-v)Xl6
z^J^BjRuEx(8e<W+Qf(Bnm)R!v`?qz>Yt63OH6fh6J53grUDsz5H)rdc9Gho<-PxnK
zFba^2rkSPLnYo!PT&U&0d$@z^x;t;K3Kk}AvcBCq)oE*wZkllGUjLWdd);{z-&}D!
zFN+7rqeEHGrQX#4-A{;=UscR2=^#AAh*t*h-QnpEA|`!yK7aOqKK<WXdqwjCes_I$
z^^AmdWsb~HoXU%m(D_;&c)IMiFjrEtdrUBY<dZ=#aM4YyH;p#?{D6n4_j~x?;J%`=
zNJZ{b4yHsl`;4os%FJ)?IosAGzX3-20<4ELSZo;{Nx#bbDFsJpY3gU$`GM{oCPPjc
zq!7q;J*RK8a>HNchZD-0N7{?b)xc_~DvafZY$_JqI6phR*pAW~Yo3M5q%-|?MUp)3
z*gQVrzm8AZl`bbIrSx#nL^QaL%uA_qJhjuhi{A9vo}2wx0)ye&s!W?1fzO1BiNEN8
z6J%o>#lA02ZGZo_S(V_d&x{L|G#9fL*`Vf&^#0=VIRf<G-bOCXhMDZKR<ag9=@8p&
z(QG*mhCox)5)Pdb1%?KPmNK_SQI9mkDE>rN{TXYzn?h|L3B2XO4cQaKtEyLe8(Xf$
z&_ok=KcdHaF)f8WIYOHNRPw<7=kE!tlF?-#_;M+FA7VPQPB2P-I`Ek!jMM<`=XpA>
zGn0l!m!(Lg)G-jy)$oI!ly5RLvM+gznv)G4{J>@vm6tgZdi(Yk*x+H+yU|lClg~)>
z!m`m!NtGF`H4E^V0jvzI>i8QL!4co~MjufZi>(B=MrK@*nHzxk`Cspn0v{)JNBL<P
z?cOJGol!yzD||%PQf*Q^OIL5XW5)g{vQJ0Hf@o2_hbJ4!ULsmFD!<QiuIlX<9!BpE
zo^EqAXzsTueaNRT!8^@KV#ulLYfCd+#!9HA&VprhrV{%jAIygp%)|-&D*of!uB5Eo
z-9bok9`m$Dx&v@VURs?QbU9UioPy3KcAHRwsgw;|Abx#xih#~;oDrCvgvtx!(r)|e
zh$4yzT=Bc<OIvFoP>+;m{Uh|#-)wz9YT5r`=Jq)7rsqIp?`T1qVxyULxqh;H=C9lC
z*@Dr3R?HqH!%J4A#2_4G7lgJ3?_JP(xd}EkTGWj29XF=vpekg@BOMK#Mew`F+sEe;
z@dE<%V%#UyPZN3{jXKKcB;+a?HIpiVDT#o`w-t%??wxj!yK}2Cj?2c1gptFkH1*2X
zJ}N8{sNY6kxMa$K&e2ijmmU?(Z9ltij^fj29kge3tB9g_M7$6VL$Xh?{;cGZ&X4&b
z(e`}%%iYom1XP)QEX474#%L!ww_ns_a#|UiH{ZnxQhV*Y{n>x(D@O<v#PlL+tT!Md
zmb%cm<l4MhLm*r(7#Y8X@PPAj@--mf!BfV1)t3L>u;i&PZyvW%1{VL4NvTT_*b4zs
zP5uy&U!w@JzArwNgZ`mfh&DlLDelYnOqnsi)WYg}YR0_bALTZ;kZur6DpGTFv%du;
zUjG#l^VYr4ym(bVs?U8em^~ID{uxN4cwWp!|9k!HY@P7)Y#pP5uM?UKJ5G{GJOL?&
z8ZxP3IqV!iLs<K{pNaW!{@Ib*Iu-xdG(NYBU;kS)TLGcA-#aQfcr`W`&~AaTFVj7(
zGI~h{I%oH6ei#P<;UpFIO6@rhT!W%Ji#YrmSKXLtL*0h<au1bIXJ1B1feRU%o8IeG
z2*jOy?=CwC8ebi6HWgsv*6fp=9399c-zqs(>d|i5-PLiakb=d&;Thtj=7RVpyx_V!
ze0aFhBw$uG|N3+lCYDraK0Li@>ZR3NUv<1YR}vjPB@%~~0eebT=>B-n&P#na7PBX(
zX2u!wOiGJ0!(d3(%y8}i5TX4~A7U3F(8sy;hSN1}V`HJIx<VsDQM1)gq<8e<GAZM{
zaJ|cv{1#3_!p~M~6stdNK4b1PH;ur|gxQGnn|WK7u=zX}Sfm+(F|n7|y7-1v%{ebf
z$tcI!^V#zeGSK=;hcq6JVYS|QshOSfwa*$3k4>wZ984b`ZmW?1F&f=hZA1e1G<$s(
zq)sRMdv$o!BEOlM&HCib&6_%Fg$2b6>}3JcHI3!I3i?hzUA5JRW@?$e>nzA+kCt4d
z3j@@ggHm3d5J8=tciS>2n?ERD|D|H5otk{<jSK7JFG70Gc|R%{6E)D%`CUqAkUo|w
z{p@h!ozd(gOHJK`xvc^=0fd#UtSr;)Q7F#7U!8DONZ~kJ*m(a5gl-x{CGwsBBs3aJ
zWiZ5lpBE%OfDXB<cA#h+?04;xm3r~B2M)L8+Cb~OsHG$C%)b~4$kyUdz9r^!^dZ9d
zC~c5@BO8on8v^Cr8s=1GfNI^NQ-vkhs)TBw8HJu}QNGD;k+hfvd_H1)s>+q9Td4_E
z$p!?G?8)A*ke0IUucX{@;U0Rblqw1`GuYAi-8Dtja1hYi)fL)lncKUoU&v7B0DHx}
z(6LlKa_nO{7m|N!5dlHScY^m{lDBPcVP0S^raxtkZgrXLOTBUH_}$&;^J$>|CPDfQ
z$x-yTu9gAZ`uA?Y{uCf0y9GYx72<l;SKRj6#o38Wv0y}%B4%i5o?rrkjF;u$c|YYn
zE=yk$1|k*#Q<HJH2M6Db6XtEWhe>|i-dd|NC$q|kN@ugQ)VmpunY4ENJnKG|oz-~{
zP1O?+w_rja4m^JSdh|j;EH*Huk=ebLms6(~Tu#HC9!3vP(cBwKevX>+;a2%1zf5yC
zCP-s^>F?tZ-*qP1cs9d1H@#*Na9<x!WSo8wgi8(FMoPrz4`%@Lf(WQ2d9AU^hes5|
zV|7NkZs(UB=2m9TXhS41=V-uE(-;Uj9*upZ=GrVW1%ebXKGU5uMWJ-DVHuu7@S#y-
zvr3btSs?o__AG7ytN*u?frMD#-rR$guhULb_9k2rY6zqTWg0@_bPh8z+d)0B9V}L+
z9DKIw(`7Ly_mEG_(>5-W@pb8A4ba!1AC0rcv!CX7o0^O$hRM1udKvU&Guq|B^sySl
zql-FYB!089oSsL7^<By5vnLF5TsALmZR{oSX=1k3z{-Gkz%x|s;2^<F1*;zwV3KA}
zV03h_(M8wAxfV)RWB!ahtx>-IcXYh#V#yG+H>lSUZ?#w!k_i1uv^4V{_7e1!=sz#t
zSHd#TUQ~Nwy}a3lMQsI{Tz1Vq^rVh*2*>+=mN_g|1eTUu?8v7&bg!gO0KBp?{EIkh
zi-L>d69mvlChpY9*?iR6fM?vJmwCg;S*O*;ODV+9(<GT@pR@Tk5h+x}1Wpx86Vqa5
zs-=6%QzRA_FuJc45`6j3Lxw&4JA6^`9uF@Mku6n)OGu|uXP1)TZUxT+_4>hiBje?c
zE`?<w;kf|ItW47RYFThkPY!=qEXDrGx6a~r5+eXdrL?I14_8Tf*=<#FV>KWxa0IE{
zJw2e=Wjejnq8xN!ITnPFM*l>mU$H>+LG(x7zLcs~PG+;FGx5>ixGz05Vb<Xqw=Cz5
zJ=+!c7g{eaZxKgaXfW&1(K+~bTF%)Upep7!!$%d~8>pr4?5@|luoQGt&tX@*m3Gq~
z+W@!v1teZPCR3;0NQs_0yY$ayPr$Z2zQVti5q5SyEEr+MF>TE6yt9cy$K2EW-3zKs
zQI90bn;pY>x9REB;w0XT>mb1H(r-bOq4*COfBqYF^@A5EW-y+k8?E-COkfaVdLm@6
zndij9(s)_LpnL3iGgmm5(E2)G`s1%~#Ol4VYGb;wQ02;Rj~YvkpYkwC+wJhClKT|=
zZwZ5V0+s@vKCNjk(kpRDbdBw*HqCiZXtL{IJ?N<MD@xa<!X`YjFPq0FAf|EtK0O!_
zsj=L@ah=A!n3^a<p}F(21#e$>hX7TJ2zSLX>cBe%ahA^Rd3pEs648hqQzGiS6=X%o
zD7Yht`V>((ULFm3k;a3SDNKReE=YON+ax)%k^?ZyrX3r!J~F;AE4tir^R7BUzcsgD
zi=y|`G8q%X%=u&x1&?Ng@0Qopj6BUH&?A(uFWqj=2b8b2OLkp6AiN-2PWPq#!;4@2
z+&!Z+|H(cNQ<st1<KhWp(y3)B^F7V=f!zU<wVbotG&zqJ1S!^$WQ{5`%`@K$hhUOD
zf$=Gck530|nBcwGMk!7&@qPqXVG49uMrU&XAHDx~g_`+^Qr|bDV+W#@Us2Vsh#puw
zvfd|i&FVG^3FzwRB0wd>qiq<5T2ss^WBb^c2<ENN*?%f$vvUlr=F88$0cUVhAXONs
z%m|v4bNtB2#N6&wn`H*8dG=J3YvfQ0U}?3v(BaZpma!a?(zza<5Amx9{k-!v<k{8R
z@(BMM-hcK<r2kTXT?@-!mzYXJ-B60XKW%n|LCbl2e%kvBdfddIyS1wz9>|+sQ28I{
zt8EFiNCBNaq5$M>4R>>~OdFYeBmRQ(1%iSSvTwKVVe1tHT<AA5<3?P3|CGH1(s>>s
zNF-usC3>u^2=jtt|4dmLP+~ImJ$X=RJ_#h%BlvV>)sG-b^jF~cREfFIW70IM>EE<C
zDlzI*Rs48xRX_51x|d#23Q}vlY_`~8{&h{04c%V<fgTG4B2MDBX+59x7f0j}#;3EC
zuE&z63b=RaTI9>?5rc8eF&97?9YjSfLeKnXdFWN(8DY1JV|*axAGQR0$Rm5l#<$tU
z?890gp@#(5uVRi%{=maOH7#F)o4+59;8YwEP+gLeu4tsBe~N8lYjUaDEiTPT$?;;g
zU1Vh-%`kJu&qA;fV~!T{1Oce9&ofJw@vP-?`Qn`xiDdy0KZa;A0w~Wkd(H{B5t0Rs
zR?2^Qo0v#19x|nV`99jwerS|W^W@cQW3`B!oCyfSh7B!yXQvK&B|krl<Sk4k`d^b1
z<dII8p+9SsW;z!`4>1U$$<$lbJ<+#gW9Fi(E2x>AokZ9as*cSN`^tAWtMz=gp~{(k
zVD&JPGHO^ux3SST3verz6iqd6nD1B&y*ymEx=G-;{<9qVZ$kOfyVD#N%tTMLIMqSY
zCbY(mUHY25lOM+TDk#f$fSKNQVEl!Riu)3`O}o^8^82GXrMVBcq#hoJz@CZMa;m_F
zW!JN#^W#=UXt9a-`f8Vjrdjk|&Pfy`!ys9LRc~P6DP8V$Il#(6_C!YDubh=@^O=;H
z8BB$fDXte{H^@(I9J}BW8ro_uWW14r7l|i_a3-ZtJxkaC-_n!S9l%~Ba3pLRu#-U_
zGcph{CWATI^_@GHgb2(nLfKt8Qen1Kfe>>c3$N-uZW7ovnh7MAvUQUu0Y^TP@s;bV
zVf=2{)vb#FPJttP>A-gv(ir@~pOs&ybvrd(-^u$=F$@scmS4%9YgxocTx_I3-(6Xg
zgpaD!Yx8@@zVbPgy0Z5@{T?IW1m*wf{QS3hhuIz~7}2}qY*gV#9ohIPzh!_9q;o`S
z8`^n{E(-N+xVR2O`~QFDA-P$jk*qz$|9+~})fuR;G40<Dmg@*Q#a#{TFLwrI*HdHB
z0?dfAyux!6%<gWJ3#`;(^cF5I#f!JkhHn!u3g=m$+YN3X_uZGsihe@^CB-R?qvPAR
ze`JPA^Lp>?C4MRwqRQyB7gAf4589Dudjq``d=qj<oJ7>uy-L>7vgYQ*4e;+=`k32&
z4Xi3(9WUQp`O(p-%e-%$HgaozC`#CGOe4&VIKTR*!+%FBU&Sbh;r0JsDU$Py%F?w?
zyO?3r8#^tnJ`fDo{eEIUPZNo*dk)@1=I6UQ?XDj;2~xwpexMMHK+3+-M_FYFqcF1@
zy3I~b2o(e|s5Uz5G!30<Nfg;U%bD#<`V11j?S^sJ1-}HY&eaJX2I=(TWCTGGa2{}d
z3fx=@-~8nuNdGrCOLk?}2BNVxT2)oMV!{<m7d>GObE{(oBbc0ns_RHaN}5}%`rmAr
z-xiYRsFuj@#eRr(0v*-4f=kt(PVTm8gSQ$+JeX;xBy^)aSd$@Yl?#Lm)D7R^YBsLP
z1mlKj19y=bQLC#*xT#e?Kv6MNe{H4d>t^m=$wZY)%yUlrWS!?)<M7B9sgRIPMMVoC
z%z}cO?5fD11bd<gXP}B~3prejmN1XR9bOGQz4vk`<GBV{mWawG0x=^sff?DwjI0v|
z+a7MR_qLJ2(JAkwDyzm8nj~D<Y1vlyR%i*_0C04t+Wp1$T6e2Ep^bf~^Vi1bkNZG3
z8~M}H!bjcbq>v@P+gDykkG+gP1a;gKdvxCjM4|uqZkV`!uU;Fi`)P4k)t^8!(4ttg
zQ9F(g&hs8SxyV=IIA}NY_L=V^Bwg_RD0i4FYeGWftV<T58Z!Yb*2oYT#h@_KO0y84
zwL>d;b=xp10%XDSfj*aZioiNN-T%opL5KC$(oB!!DG_7WOTeytcRL;*eynEeP)wT*
z+i+xxG!WwcHdox`BAH1FCeF+VlJMe5;RK7|^&$@-Z7=m(S(qY4=u@*x^7jJjgJ20E
zA<Ob*zZ!DMh!&|r#)D<zr@Rqbx+}mg-p2&P5&AD_s2;%x4npg58Hv%E&8~0OSUEYX
zC)<SuB1E+HB#I?yuqz^5_@*kKh4c8zl<=fQ3F6H1GA>ULYDSGZ=b7n3M`g492Za=|
zjNejjyBk$f(swlb7N_oEP;=#Dur`ZXIB}KZ5d*aYC77K2l>?Nk%=tiI5vUCkzg+k$
z-n|WVxW#3|8D??!&XpS-h=Bqj-A%$SdTuFCEXverGmS1xkJL(rndnbX8oM;UeaNo$
zI~%E0hORAxd%+*rLm-5TY$2>?bhmzVq^faPjc}KUu#ss>7b;z5{1u5?^-yT2a$RVv
zG@Ek3<=pNlbJ5P)9HCQ|?6_RHxx75XW2#&SrVgY5{L(4mJUI2YF0Qfax;!H|r`-H&
ztiBz@d-QP@Lc|c}r)hm7k0E~s<_DHmv1h>wI;yvD$|IwJ$Mn@?XvnI!0zovfvbxW%
z9DIIowa?ytQ?YWC_gSQ};;cz<!P7*z20AP-B|&?S`<H^S@B~32r)1?8>jX+9pcNnR
z)Uyq<wx)=sVYmM+?RN#8USG+?_vr`>67y)p@qu5vRuMLQmzpXTlIag^05yTgJY6#s
z_(#*9Ca(wBr2J@5)FY-AV*lmf{aYhe{y88EgRkwiZi<J^q^H;2%kM7#^}kgLP5S#q
zQ6f3dinP0<z$*1B<o0?VI3=9Qok;Hy^$OGXsa~Cs3YRV+FphULWQLNCe%o;<d+PZ&
z8ES*5m~>w%BaSI`Q2mSScpw^&#r`=qb)JfUgJzFxlGY0Y#f;gR%#N)4NaMulpXq51
zAFM24OX&95IS3h-CcB0jPmy`QBXh$2aeK?z`SOfL7ot}>Jc^G5S;Hb_bOV^^M(Qox
zytIaF+^k_A)YF4deq^|S23X`*^^=hETPqF-C%zQ5Vn@Qt|7@)lka{Xz5L>zUM)aQ&
z_7V8*+L8a|(b5GS$8zkN&(z=5sod)$LO14kfWwA$N~Bgli<itBe1x9gnZnbPrliCq
z-7-_7e#1#C%NDhwz($01nlXJTDLug-qakTF+eqbzE|s!yUVWC%^m#eLR76HJHIhbk
zCNrb;lW4uv`-OsCpgb!uk24X<(ysp8;05s$l@s=UrqbCTUsK+iKgv3Of?H5OHDlG8
zzO61FE$ORxR{eG!g`z#p`h90bfan?Yu_S%}9xI}+hYU2%q|+5fufo_{2K4~Ra%wZG
zFg`>1&5cV?IuqP*nO)1M4TkN8gPu;l0rh8d&<>o^B{VPm0f@ie`_S&xtjZ%D6qrdE
z&=P>cB`y`KhDyQymnFcp@G@|7y|R*M5J({AU%HR-hn!tDEN*<KL58GIemCa1{3196
z)_ldPt;c^t?dWsL*4UPS34U7T2Z@!hg`h8>?}-p<p8V0+m8EvVO?4ml2X1hxhzETZ
zO-xK$KR5ucr>w3p-B;~^OgBM160DrVA)6rBQ>ZZOlO?=xT=5e$O;J!Gg(6hA-N7ND
zp<M#by95k)Hg4Y-BQK-&VEgjl!M5LW$iJiQ%Y#HuN*)C64W}9x1K9hM5ymHk5a!#%
z)V(pMa-byyO>=p({ZgiL9yh6&K1Z(4-7eb>K({0Z3QKY{p6?M}6M<~?<=@y?3|IxE
zT=%Y;Jq+A(!)KxCm$%5fjf?}OkNdkA{{!90BQ}f><0Rn>BeMpnx0p2OZU)?=49f<`
z;P@-5iwSld-5m@oV)0LS#%178ILYzr$ga2Ol)u}-P0MClvL#0beV$onQ&qE#MyE`G
z*3dmFaZt6{hXn$a{4gB}*QDG33EW(rawvzS7e9+k;?qb}Lu$$Xx3+O`7?_0#FuUWn
z@GbKLeGa^BjZ{^TPJyl1ye-3uwLtPy@gnI9oD)0fQPAJqGo_&|A84D|bLaPu5*?YN
z15Er$r*LqdmUaPRR@@=dh782Nid_X(+^%s1*iZs+v$7iGV43!3)o=gaP=^3!Zh;W`
ztuJ&cIh0uYRcVR&?-DsJk!eoJd`uWnNO3yH&k0wdAkZ{Xd)xYa3rk^wHX9!-mjiXr
zz?}I~MP^gM2y*SpU45<LdsOQi`}h~1CkSdg*XqO<A{ALKy4a=}cxIpr@a#6V%w%A{
zI~@1B-I5w-DXRuYy0Cwa6NuY#oy=@=?sID#K6UPjwghphS%>{tTKwiiZ=t|KG)a;|
zLSe%6pR*-fezOA`f9(U1eQ{Y~c70a0w*5Qo^QJ=BYg#o5icR||qp?g8KkLF)nq&x4
zg*Wz9asE6tab!!O;qyr>U^dle;+%hVw=O}6h;wv|VRC|QYPB7|e7&@PeMSKIVX=K`
z^GELgu2-Y|_BocXMs7~JZ#IFORnHv15K<iIH3-LiVfk9jfImkgPIso^Ez8e!zo2q|
ztMw!SxwLV3-otxvl`;Pk)52^#AzRFi$L~^kXA>ibe(2VidNBiDcZXMwt}nXJc5luT
zZVo1Hc174*;y(Wf8?2Pco+y!0ZrCle9B?`8DtcAkHnX|8i6JaVx)ra4hpTmvn4a8|
zgu{7?uxI}WA3^T;rdS+kdf**V=gKf!dtsI9u8WD#uIG=YtRM6HJqP;74obp*d@MH_
zne#}?PGhs4p?d*y8XY2P5GQL=pmRY&NT3%u^y_f<-B&g>IlxTg%}9cq<(A*^s8CV#
z(lOx1@%gHZtlbq2CFLi-okyG<4wk#<<u)JFQGJj2GWR1=g$Y6mN>57k9AA}4K|*)_
zoDjMS<Z8dkZ5UM!+XoACsci4kk!yM?u$N%bk;>PH<N!)htSlqLFI3<$v;<UIK2rZa
z+D=QhE=BC9I^_4#udXNaAYyIie*6U;Ky_2-GDT4MH;oj{v<ZG9=YcWz3(-n}B4t=<
z`y4-F2-krRO+pj{dli4dlP8e=k3G)qZy$h&6y*9a5lMV8DcPvGkJ1q>r!LDpMx7#e
z2RuV>0Kyw5IJahXUtF~(Bn<Yb(77vQmk@xSApxO+F^!Gxryp%2*LI+klW%ARMeo3P
znZ06W&2`PFylj*TktWnxnQxgXFxj)IiLaRMgWbJc=ADsn-emjgxwDhM!4`8L{hscW
ze$LvSHAOk^Y?SITD%Mrw4c##d+<|U?%5blN7}v&6udEfPswe2$(~qU)DkUYq7>!6!
z#E_dFnG^M(+k_xJ8aXAgk(~>_n5YS$@mFMimTm+<Ju6jlO4PU6fFDMUa%{W^fqNbV
zgT{1|sxi~IeJod24V^!dS)XJUoe7saWwj4bf*B(+Do4v1w=?di$0h~bJlSI0-Z2#e
zG$RlKUFjL+*CXboq|d!D8W|1Ra5Yey*_P+ESJx*9^5a|`M_YC5sFD4wwp7&vwG`VO
z0}SzhEcxWcLZm3vLY^IUib}9o&~4SW*3L4ADVq);$jeLF$HH6uiB>jJ$89!CxZ6@O
zBw%#Jsi2D{%yNvQCd-L>(Wg$k?!6A6Go#DWnO8<tHQjy|6W(I;a4b8+F3v)WK?Y2~
z5^lA7?6ouK?#NmF4JHD4O%!FCSz=6g8<cgTcXyi!gOKy>la*-KHcE@=Y-*C0{{pO;
z4H~etKRfEgIC9yY$HP?w*MsFI@DFSUS=-08M9Ye1A0XAaZP59<7ue(DKUct?<FPRb
zJ+8-Up21bw2-nUnGTx$5%<9xC;X_;2E2$YdH+OawjB;>|DtB*9u^%cbfAG&ilKTBT
zq$A+hSI8r)ujCrPre#ny&aXd>1LCOAtT5u&n4A5;_uYQ^Zp@;uU$TJwJ1t+YRPn}g
zyZrj-P`nnIk`1?h&B1z0gftA;q2!fZ{i6N932^z^@N*)k7uykJuQ>~$Dj4K)(X>aE
z{m0y~(*1!hj%bhNgD3H`yf0kFm;Vwnz^FkdF{|-OhoyKFnAP&Hsjlh^@tT5k?Q#1!
zQ6Dx~a|0w?JW#cc%UWE^Km+TIE;Ac1S@)FdR!~${7Lwickp%pxM4?%fWu?q5Cl-{V
zl8+%BvsOr#*G%pMQLgIxlJ`Y;0^h@1GQ(vuVP4j_c0=`!@~$>B7TO~gAqI-*I^g02
z_Y}B3TpEa;#3#R`Olg~b&p+}eTq#7hbubit;i$nCMpBv;`>>x0lD$Zwz<KKs`XRCu
zUT&oQ9=$gS1E@qxO{yZH#BX~xwg?jvuD)u&LSE<6^ZK>bk$|qX2iAIImVNKeA2b<`
zHNGQwmSi=uXSq>Py1u}jq)9{I5`(%6lr*1B&4&)M7EU?-wv>f`2;ih5qRG#5@sZy^
zzRN0t=0MXaS|z*h${?-dMtN)<c#Fqg#j3HkR7e^!0-|N6Uxc-N3Jt4vZK<O`#YG!?
zygK@`#18gD20PrOk4IY>rqsVc!uftZe&>dM&ijrPDdEmDSwKp}jrZOvSvuOt3Hl{b
z9J<plN)+}$A3-~)VUdf(sKjZ2zYkl~p-iPT8HphLMc<1URP>+u5E(3L?NDUCvO5V3
zzW=1&7~s3Z)emiBAONerHLyhI@3%7$YYxSxbX1S!wpGTq&X_JL(yY!bE%@bz2L4+~
zm^-CgJns1%vrzbPq3Dg{p4`&@##QR97zwBCkp0I^)1sAQm}nmaO=kbF&Xc3$a07>1
z!L}1ZeP!ptaaP=kocRm2$DDp5GxoAMdI`e9K(`=Xdwer_Gx<|HcjDdj{~+e$yKM}P
zq}-2bsTri!v!o}v`NC14AC~yJ4|rdH*Cr~kQ&YQf&gJpT>x%dbX+?m1xA@B8#<kN?
zy43~VK2{;MPAUQ-&4Vx`_HNwqx!<FtPFeo^MMh(}*xmfHvT_ZoxX9{Vq)yxmqY4dv
z0N*z2IKN+fB@*{N@a^SkC1pAZ5j|4pVxb}-ZgVOl=C3^k*YtzdwN;oQ3RMB&<QzWq
zkK@;E&vPhBGygluY41`Rx41Dzh#)ciRXbfkr+K{a(*&!Xw#+6|5i?^k176Q5V;8Cn
zzAx7NX-F%In)eS@87PRBdJK<zI6G(nyG6t4N2af8cjbHz?$uNUed^4Q!kPV!ko8t=
zR@M<Er3_n?xL*hgf$S?S>e)40QfB$-SQI|EHz$AsWs^iR&^J4okRJ3Xc;}SQnnJC|
zD&~f9M5U{x>h@#Bw=QTGwT1B6?OiWvxqNZ@<rQfGL->yB)?3jU=C?QQ)u^WdrgH_(
zjc}oznkS4BUdC>@Y8N6+Ssb<(z6-7OU&mu$IpzqpuTp)}%3JvwqMkdeljOX~I}hKu
zKJk_=D;A<f=Zz9YG%35&B*n=3SO}pkSr&)#WjaRjnV8K!gE{g(h8D<9QCkhaSX;!y
zHnsjPs$b8wKQzBjL~E#R_?Q#|0=$k)#Om%ZT??b!xD{F3LbtwC$4>}97}gk_rsI#l
z@^Ac)fkAN&sf{wQbr*q+6|KQhuf;W8lexI+=nL)=6+_RfV?{X|aIe%4G^SMyKtHDi
z#NE8_lZ0(_N9$r}!woP<jA@^oNK%l=%cT>NDrDBxGKjDL>Ub%OE>{V-I-1y54xP7H
z0ws~S=}Ukg=QTmX!%1%Wm7PASN--bKqyn`C-5zQ^A^>PqjfY*EccZJdTDv;0w+V=T
zLV~7E62hHC(wKA~u{uXeOa^K`|NHlNdwXtV9mW~`OQ#2I`xEEUxYIU0vYmWqW@k39
zFCGVG2*%fRjNeIJJOFM~`4*zM4gzhT1kF4<IF`6_`EpaeDPI9zK~2v0hyUR?9k$>G
zeFq)Ie_s>ieC$v-$;67DJUw($?rN0{t1c;7W(UkPj6vS2b4CUlk<?+WYfQxCWK;Fk
zt$d%Xgc1g(j%EI|v&XQgNXlena3;2A5sA8^S8Z-^(Fx(bcwn=pWU6d!bL8wradGn8
z7ad)L-K>N%Dp8+RZ>5iD+c_GBz_s|Wd{TXi<{QN@M?e3IYjVMgl;h1VdF2qnd+ny<
zcYEFs>cmDE#ECvyZZ#47BADd{vvu=)BgMxl1yTe>k_jyByj)zOnV*cZh45!^X>g<4
zWc5eV+Es6xeFDtWJTo>`f6|maQSqxl+CeU`Ez9rg4NdB+e;A@fp#xC6I|lDY8AwDu
z7G9%8M5O5(T3R(8@1Wu2k7-ADJ7!H*6z;$r4K0zj;5?(}rWNKwyCVUzIhg9@XXr5f
zkerIq*6N}Tx!CYbJtvlS)<tPfS35Q7F#78*)dCWt#z)t*f?rT~6ICyRD!%{=*rao=
z_Xf)Gim+IasIw2hWh`q&<-OR2$1lM5;eOB$l6Oe-y<Ge$K|TTnO)pf;K5vp@bZx;b
z0Osr~(dbohE1Crq1x0Fc@rpe6g2Coyw{M+Sldl!Ki+NVvW|K(_V;obY)=7doSIfQ>
z-!;J49#L+Nta~d>jFivs-TGx#B@Ut&CoNX$>xslIQx-PamlpjpJM47T_X6p=RB{8y
zx3kLO{_pblFM7bmI}yah#q-(yRWB5PBAd)=SEZ|?O@iaM_vwKW4fn0)IK|iGPrT@x
z=i7od?VsKD6aTt;C^4b!_175YbuXgp>6GgxZdHcbZX#wGIQPG~<~kXA-5%r?O@nys
zs&-qX#wTY&xclk^vx%`9+nDCEZOCkR?U}=zf}FcoGh+IwLVv70EjwFAh72SDyhOa-
zrF{Kn?srWJdQROTXo*7Vg$|3=6?@@ZH=()i<Fl@}*Jk%CyOAi+e3k9Ol_~DiQlQk^
z*p?#gd6WON56_)W4>nG&uof9l^@U6B8>x#M6Tv<>6(a^mAED;5mvTrpJJL)Io{XG(
z%7y&+L#vkEMiCDAF(1}E)V^gM59s9Y*6i<pjR}{H%)}qAm&N8w6@6uK(q0Pb8nsw1
z>8c6H5iU6k{yhmR3DXH1T;|n*pvyR_UH89M35fr9MIEZ7v}q4ZW!&%;S1CYn6)_Qk
zoCKZ6mq+VdtE(Tpt6F57TzqEsXY}#uGMuVv<$YL;tu6Pfj~)i(AxpNfX+Ysl>*$ER
zrBjtW`;WSEEk>1{w(3|-e~U+-W*?&yMBhTs^Aq^f8}&ux5x4W3HsHkJWi>^HB&#3_
zbYe`SQ&p<1JNfO3@<yrH`>}1MGv&^+jsA?a<<+fd3r~}B0n}W9tpi!)U{Xx&E7Waz
z)hBT8Cn%SS?sV5mdN7Z5AHblGQqvlHC@&^9rTjqbEu<`=eaXFLUU&6Ze0diOYl?H-
zl%okIz~aucm)@SlAY0ETV1rM{orekjQ!OlaBPLmz${62ATHLYg*0Z}q(l)DP3)lB}
z1`Q8L67o%okVTu8`3w<uC6~WwOww_SX*c@T+@s|k*5UG|8GnNiEe6Qo??MLag$TNh
zLQ1rcprkQ8&JHbV7`pVD#LN1Q*ITy&a(L5_J7st#_5M4YZo`i*5^}#+jBgn8rST3a
z65>u*pU{0NZ4z-<3F=6kX=l@Rd#qHyV&5RPp}P&ZIKUE?bB-q5l!NE3mLsPnRYk4e
z6x}UtF{-*Jj1O)#2ZS~|Haa%!KPHsm+XMt<jGR%o5u6bWt^3g)%q3V?zfJ(ZvUGhy
znV$J}Nt!z->$5nVKJU2&22U`5+r&)ZFIiO{Z4p1CPEu7z85`79IaHyIL`3P;^+d_u
zJ&SW9X+$DO){IXJ0<vZ=cq6*Ig0du*!Mc~eX*cZ-b-)CmvYa^cuWi1$IUU-4#s(`r
zh9uLw7)yfT=p@&g5<j|T!Wh!u%kTK;OX$U2NM{?0$o*HNXWlgFe6$cvkr2Uh0^hOx
z$7!hGn-B8uw)eGM*Pg+S-By@w)>A2|8{e<5L~V<0=IJAaPF6+4!ZZVD8`e;2EL!z-
z(rQj7UN3-mae2qV!7GU>duVPr7px(K2^5wnay&)I+)kCRUmzPQ1(hL1A}$UcTt3->
zlsQDOq|6QEoVksliwU_OY-yvsmZsKJUQ7mx0M%Bpw2)SDl99pbq0bCN+>~1bJk-|}
zCh(aw9s6>PnXzd|64LkOJb1aydZsNI9c&K>V6xNU=d`Q(u>`f~D<8$N?D7=xz*`NU
z@o#{cR1mh}zY<TPD+CHQM-II+@n6FEd3+=|{gA4g?8F~8f=$!3$0;lIW;!yRS1IW|
z(WBFQwez)7IP1v=C)1TKxw^PsbU1c;AYiN_4Vl)*FCu#VA7dbhV6e0k2mpTKy8i)&
zUkiv-^9<{+hZ6?u6mqk^(w80jm{HK%wF{&nRpQj9w@~H0a0oObyLRpNQk$eNb_*S1
z>#cqtJ;*c@U~+6ehmfFwsg5?v`J-By2OPT%y!AARj7rN0B=?5gVgkcS9TgpFzJcn~
zu{RNi>$c<ELmIW3M(WVEjKl&MV9#mM&teZvTt|DceU97l?#P=D91pnI2?q^k)8D6!
zgUpWqo&B2}iLUve7Ig1pLq>i^SwoWh)T6Q-k&|fb%nWMTXKP30HuI87x<(<f3gqw{
zhr=cSgrL8<S}Mt{-seCncWH5q#qPQ8m=<kx2x0Ks-odgQ2_t90fwBO5Zm#gA*N{p1
zw5eL3PDWW8YG=~Qf}^_HZTaU?h<}h5aOa}SCYCw;rMQ_5f*JzGAqg$+v%^D7b^=_y
z#mLnZ9?kT;{3T{X$lV+303|(DdMiQG?>LTRu_T6|KXd=vUEcmq7NZ+fpv!dAsY+>;
za|Z%!R$S*$DMqD@1HN-}nw)SEWSCKGn<oDIWl8*E%4Sn6&Tp0AJn(O26}xc3LTu6I
z|9!trD7B-M;=vl+-@^eN4t?*hJzH*qIsaI;D@Q)Id+2{DcGM0<?wo|h*W52Iru!`=
z3y*jF?m-=sv^Z^XA)52F#K&S07Q-g8>EtB4xjN#(C+TeglsO>~Fg{-d%xr&5I-Q!+
zW2^sa>?(`yQ_5|aZQLg2WtsZSD2+gv(G!XLA}(4XpK?jk9IEdaNUhBZ?Jl4wETx^{
zeY&4ds*>24d*fm=_|QbMYQ@!bQ=u&ZLHL91l|L63Hq1-m{9s4{5`#p;Cd1oo-nEPL
z)d>Vx)NcU95Vj)dEzw&9cV{d@2NNL4kN#d%t=fcqR<#B%UlNhwgwvPc@xd9SVe0SX
z#QvG0a#&xD9UmSN7V09oDjS91VXqNw76#G|RzjqU!uW5@+UX?e6^{IOE&uDaoBJa4
zZYJdwldTpaJH5Y0-wRK7{hwN(T1%%=M{9R`cRQdSc`Q~DII;52GdNgINtuJFfg{LD
zn}t4>f{DvTkZtn$jmtL>_*bM^38=xY#`Rvd`fEYkV&v@GPP)bWQeN-FdnWM6;Y>_M
zVzDFIqI-4+f2&;$mZ6GaN)p0*D&&lp&~p(@c2q)`RG6c__RzrWtWWsZjFOOze0Rz8
zVt{s2^Ea}eubs{h6F3H{Z+y;=@xoT!p^7;P4-M&lmDza`li&F=Mnd-9=xGhf;l}6>
zumf0C0$=`MKHI-%RJ5pi*VbAEWI*h7)Wb*nYjBkRZS}t2ob*0$xqZcPeRQ(Q;cMK?
zb-dtstLsWPs@efLyB-{@#FkrBnUn`@aRjOw@5({K2Mu`C;<zj=mKVDNE>4!BD=rqW
zOJmpP4+^V^%ycSK8y}Il+~N+)m^IbgELlAqQ2@ryp8qZS9TiZD(&Bf`O8q%jDCI@O
ziCm7Z$lBe#h!)xFR?6v8ViWetS|o7KM2Jb;cQsu)xBq~z9-uP2<=C2eEVT!?gMt10
zYkA2)%1I^47t1HgYHL#e@Wz(L&fNY!`M2O=Uu8~CPn&Z?3A2J#>`bUaw*nnM+sTDv
z=HIpy6dPhjtIp?hhi`Jb&0Mpqr~@q*)ZiEAEJMyzM~C;727_$m2}_YnW=l>yj4gwy
zoym^eDIjNa<e8-SqvZjqAw!?AFMuhcc^Vo)GU_bkWKvmbWwF~Z@UfIAx|!-WC)98C
z+WpkpRF@RQ8mz>@Ra0DSN<=hP9&Si7Bh+uR<yAJ4S)qYtiahCr1?(mtZxk<g+*?Ro
zfc4DG%n&7JPYz{IhoxWeFD>o;3~d1}4#%(q(Yx&x-_=P#FQcGbZRwRN!1F}e{^=4|
z^qY4V$uJK)@s=WJIOirSIW<=iZ6ag|Nqt*`&n&fL!?c{Jx)t=cC8Z{z#Qff|CPM{t
z`ltyjiN4IoxtfWRff9@>6F9GJ*Y_Fcx$lUdaHG^2qs>KHqmej@UgSwnPw@Bm?N~_O
zAS2wDE#oZ{1#Dc&*~jM$sGj@h20%UL-?uyNT~rG^lozob02`^-XX@uos!KWR0n0}I
zZnP(Vl-t@H>v-Wv>sQZsT|O}Ks1BnX$5KB#H6aL*4ONwbM7EZrwP%W^6OnrtjqeZy
zmsF=U2VBvHB*D_m1jS8WhV=Y{Ts~D@+|N3z8|${!vgVwb3<mbd=l=;ts?}*<U|rEm
z$AmKRqIx$+Id{YuM)FG|3G{4Chsi^m$jxnUf2~STd~w__e#dBRCL9t|y8F$Q`CKT!
zeT~%|N@TGFHc#S={7x8DpKsun1_<f=R2H2w&JDLJVuP;9&-p|7y0~$zUc#<#b86(j
zjhV)@WbN@r07CyoD4ZO{8wpierfc?d`>D_Y;IFZ`lR%lhsqf-bk3YSVU8GUaaFMu=
zoI4Iy2$jw0syy2=0yX3}Ep@zmy)0fZi>%*he3xF{Zl`0@&9S_=yjSgd(Zx;Nhe4{L
zmNF3|zdE?k3qRPairz_o>AP&)UOg^s<zHWZ^5|bR@Rt4ktp>KNUyaSRG!}0RKCIkO
zp=dnlThS_U(FCi%3&J?51Mm-$=1KFxwYf^`Mq`ib1s<e!!~|8_x<s>Hi1ND10aOzR
zyVg7s4MB{(?Au8{88LCujISHLuFrh1X+7Ni-%qoVEVrpcq>m`VraKHhdQ9K9q^^KL
z3Tz3$M9JAMep{kuUp+|`EsA)BB?F20!6Ye{a62D0onWbwHSf2#KB~u--->S3C@Url
z(NfAUw-VCkM~%%$*N5hfl^s;kF<I8!uPM)d6P_hQuu6oMk`yahEL+_C8#pB_Dh@=`
zxzYMX?+`Eg9}O~HM3#S18_B%5>Jzn%OR)w1JEXCA8d*?Z`DqoQBV6)<7<cmhG$bym
zhkMF)qpB}Pb!g=9@84hr`EL)dM>cB)535<eysARgkEzSR2mCE}0tCIvBEoGY3eCY#
zwCp{y-r09%p1z+_ZV`(-i~DLgAFjWk%U85evAwcV&G9>Vf=nx>aJHv%C!nf$onz-e
zAVHMaw(8hZjuaoMh6`EVKWV2n9H%NL@}kaPnPYo|a<0;t7TKNkw0hORb<rA8UY0=w
z5-0kLEH71y5K?)Pq)O>n*fTkvf{2Xg;;#o$T37WV9zBXhz@7~SWDwjb))xawgfJNk
zu`H_%S=4!{;{5)<AeCH?3nz;#S^Jg%62FiFL>z-}sj+diBdcGK*3}51;v&sCpR6+$
z@%fPE<p}SP?=n5C&B!21grJ{KS4eFNNYxy;GFR8o^4HEPU`Pj2h`FEhKjd_e%h$dj
z9KNR<N|%Qd1gV#I3-H7d^zQoN!SV|y>9UbvB}D=>;S4jOe4{?Hhtys3mIEvZV5Xw|
z_O4jbM2+`J(OtR4S4&yZ7twUh`^$dq7F%siDH-LB*^SNo0xl$E(q+2}-PwBG7D5Sd
z?YO>lHMMd>80ybA5W=>8++JxjK&VLTwqInQg7;x9%g>Hhy|UEFA0vkymE~h*VUS4I
z=`^X=wI7Tzy4toku9dGb=YUJ<jS|hK{GZg6>@LX#SNUeUk=O%Xw@W2DS6)-z-yTeE
zZm!B*fS;GomVuhFib!fU6|pEOk5nonuLOx2Ef<Z&#EMx~M}}Dy3Kdf>pMe~25<1>F
z56$!yB~$U_@xjjJd*DOlZuzjCD*9+HhQHb}pnlYsCP`u~R6tgY0wg4hwmm;`Bw$v}
z>-}NfIdgTgq0vO3MIAM@(Hzz8<>gK*?VLGt)C>bCZA9J4<4rl)plIRr6WY(83B3XL
zT;tmYUZT4O#47Z24r5s3r14H7S*fWgd0Ea2s2wDK7dyc5^LnKF@*i-`0q802(N3+L
z>?F@I<~m~5Ky04@KHlG24aX;IrnVT|{(e+fx4F>71R+Bm_OUMx_~<o+7hC@MQ<6iT
zdlZA0o3OXALq}Z(xA8%vUknl^DUCZ*hruQz$<B;pwua#a6GetTsz{KV-6J^ScT~`#
zrM+8G3;f&cl(}b_x*=;gw;$J$YDadi5D9yf{L5a3wz}J!(2al<sNpfBR%rdga<%)+
zBf12G`RN4~1RjQTm7Nq>dvMp***%W0wBFT1<hlAVe+y-%jaKxtK`CFI@D!KrlFF-j
z-zj?-Hk(;th(L?ds&er4^*mVjtSd^xx9L->Y5Xdj_1HS^<C3)%l@}hFS~W3D=O?8l
zJpNVgIYHuI{BQ2&-|=Ozjk_Vq+#l-#v*cllN{X>8fQ+J#l~_L65GKh(O8^PA%|9i4
zBUo44BLWfI0;8QYIQpvkO6COHw<PnkqciZI_vtm=)Ujf}4Zm0-%)|hlsXnr#SF%co
z_I2T#!)^}X!d_N+KdNE(!bS&gugSmS%1o%6Eie&TP#C?uOcXRaKi9X8BkV`8^W%WG
z!T?;~E0ik?Q$LO$IlDX-Q#Q`90~v^+BhB;WK*XA<>&aG@8yJk&r@gFj5>C8Ke7ocP
zd#vl8StB)^>$>2<55wjKs5A+ioZK_<6yhz1E|M^sr%%CaW`QSW_{rt-;7yf@rHa<|
zMir)Ck_u6JlMz=nwPQ`=Ueu0+uv=OL7MFkG`bDGiP@l8(4Mr<7oEN;mQma+2wNa49
zZ+tt@aZ%WG)hM7G5g3?KV12sLufuOC>Bafl8wkE9nk3KlWinx@A>k_T2P}Wr7D8qn
z$%OVyih2Oojw~!6Nv=>nwdz7@;OlWES1ax<ot-pXAQz4ZB9^y@?tT@#C?|ek-WOzE
zUT#Db)&m`+kPtbs=;)e{b-L{%T*Jr)8GXl@EM2rQ^QbZDu@p}<H&bRY6F2G0P=Bur
zg64++)|@)~i+)igAbCez*!(=jHV&5b8O5V*qF!i-FvqZl_kSn-+3-TGw5YOrqz#0)
zy#not;iJA1d8s(DD`(|9)kO?E5Fxu&;yHwfi)<_<sn8UvPvQNlUHu!~_c<$oq4QYB
zWp>E;ZL`G%Ds>F{)q*ITSRO>*&(N;a7CRHi8<T`4{w+hz1%{c-H5$Phy9Z1!r4_pb
ze764%mfiO}{{5RNn-Du&;a%*ysXczH=}0*%A-K~o7t0x;)jz$k#4%uIagwV@qmM<q
zIOho(n%z@@;P1*~D2<D^F21)}u;IbPfNCJ1Ndm@Le~(nE#piALNFQ5JPPwY|RPXjX
zYO$K4pFUFm@_A9WqTI<zLRdwuq~|N_j<9S^KunTg<e;eM#Apbn+867%_1dx$?Xt#F
z`zM`T?bXjV4XJE|*DlaMSoA2cC!=+L%2oQO`F=^hoj0+a$YTT8=^r^Z<nn&a`Su3x
zs^sDnP+07wTN>vqMLEL+@KVbkH)S^o6%v7eeV%)By?^5y8t?+cWl8dSK7=wqcG}O8
z)nNHPui=cT{dfeIH{p;cf5!%DF<NH*mjysu|3q0jw`N-j0hrC~i3hP}j#;c<Pe|1E
zK0AN?;c22hj#3-Tm*go<_ibO2h__qbkK#JPHY_w_*E<B@GgUV3TiyN-*-8lp+?q|3
zUw2jL7OwGQZ}4mrem?#1x6{23YY)AFoBsqWHGd-XzMqSuZP8aEixz}xd(m?LG(;<o
z(+p>&KA*M>{O)>hZr3`z9lV_@L$()4_SZDF)M@u6r`k8u!s-9c@B9vzrNuTZvq6Ru
zlT7JN+cz6J?#o6J^4zQof-+B8%TjoiU@|M@;oR>r)tKUgO3|uI;EJ?{v<MIwISIl8
zBa^3d#&eCcOQF)#b~{0m(gOub)W*`Rs*uJs<nS2~{5YPrBC{W=!byfvU~%W{YMnPu
z3NX>>L-nM7ZA(`;XmgkmWFKzOJf<0nikmDo6@ta7^rW0j)n-Li@9Vvk&#@fFV@x}Z
ze4>?awo7c72HDhLgK?I6cla_q{giU=iuftM(Y7G@d|hicTO?4VrLM+zc6RoIx;=F6
z>6{6$(lKj49+i}M;b`?G4gdGZ^Y<hd;Q;@f{;q7ZQRA!AN1ECq`^F^$sKK4g<O-I6
zABR4XFCE~-xKRcJ%GbSGk)m(E7fp*kk(y!pF}Bn3Pox;?+wSOh7b7`W4j7}?-27P^
zlBvR<LAZpqt8S5b#h2w)2jN1`RKmg!H=LG%Q&st1JN~HBCVS9g6H?OH@q3L>qe8h6
z4AAu?Y{a8A;9bw_Ol>)N-34Vaa7r%U5W}j8fz`PZJJ^zL0zcx8<JK7=_dAWgxL8=Q
zx3r6!YS>i)%^)9g`l!zBKM^+*#&ifgE{tsCd<!kT?<%-R=&gL3Z3`qC!M7@0p8%er
zp(mLn=?E$^6N5XfP5Jk%W2IHG`QUyAS@r5<SPDeuK*rnnVtb?ZX`Fp0lrN&c$6lM-
z0L_0i<=@f{oMH!*FY)b9m1!;(%sb0PN!^WM&ng<(6G|q!LqY>XgF^{_=jUg5H`6ri
zcTsZjENw=9(O;8v7I)f(jzXA$=unf{%|KJ<&nZz?;RXqPP4AU$v|ocJcWDW77cpgJ
zz+~h8DXIBq9jB`h``9{F@bcfp|H%R1FElNU>1@byuOu7BxsI-CvuLp*J{HonS2*|Y
z@~Gp@)$OzO7J-T~segox9=ne&esFS$=3ji)^=y}kl|=O`J|e(h#r;b^#sp|S_xJt`
zCoNt2Bnm~aR+4QGROU;1n~*!x`Jofwe{*oMavjR-7fzplOY)II6R~DrvE@72MuYLT
z*(t}5#a$fLZ9)y-Ttun3PU=>nqm^18F~}XftDj{2luTTRauXZ^0^8=1@i=*rXc1c9
zU0hcyRNpbVbHG%<$0%1;xw2kR9F*9h&Z!_-HSbE|i-+<aT^YyXp?_EQIA)-v#Aa(*
zq2p>3IJ+ZyThic81Vj761$nA%@kpD*T5KZ`gE&lUd<T*p71}ZXw1@%Bhkt8MZ2H2z
zhe0-oDy~74ohYK!2xaX22>gFEeS0+1{~!N`B)QC`TDfkdrovouzvNOra?7nEw~0dL
zI-2mN#K^VW3CUff<hr>Pq0A+73k$hhCe0=H-~0Rfo!^dsony!A^?pB|&&N%U^^6}&
z?V@br(XRGUkOSbSjQRaqA`9wo^uY|p7xAK0&=Sq3XfQ3~Lu$T$oc*$r#AlTR!3<d_
zhf6x;a{Ax-_4Sai3;TJk8lqrYWQ$?%XtWxSgug~!+TtR0cJ>7v!p?aj=7k5!q5KpY
zC-Bj>g22v$LL(c=u{EFHEo5{C8n8biRM=K<1|ZN15THD!;jGRQX=oe$1##Zr-lIi0
zF}7hu8bc6t+3(Z&7u59E?e}S`Kd!IM7F{Z9FRci6h~LIhv*`zNTvY5M+GxG9vH|WZ
z{_gEtw5V0Aw6({rTafNr@%&gF|2G}*j7*MRPZ@KAm=g&7>r&QNYFBcW=R^Lye(PwX
z-Ur5}GJ*5kC#rbc6={(f1YKgVnj?l_zXA_t^RI!q2$KeiG-SJ6!*>79d(?YoD#rS1
z#~sMMGuYM23@Weg^9ybaIS2{HYvkr8q^G;s$5zywtX0c6`>1`G{o-V0ZhXK<$yk?8
z99!p7Xuh-ZJ;&8bUdvd5gNd{f@;(QfX~zelPN2Pa^I$Lff2&lg_LOFb;en~()yBLJ
zLw;G{1idcb54NMXhVF2~tS<Ry6u&-i1ns?_`Ef4w6G^aBW<ziGb^&?O_S6!sON>`}
z8?JWz>1vx1PmOgW5-D$~Q~0tUWXMKa_veA$^cxFmj=7!=lkezzoujS=f|eWzxo3I+
z%EG|FSa>)*Ip$eud93#~<hN>bbMs!vlkJJ&St|2CeY*@EWwyEEtIB67=Co6W%9<zm
zAdg(msE15l-cmoaEBK7H+}R7(F#5dZ)Zf+OuvYpu72KnC`WmB-C4{b&_>#1QvEG%@
zEf0!2pF8mgv-sz`)e$}@DK|p-Z9d=u)FM-!OT_6yEkf*CuRamVDl*i(U?G|p_PQ91
znF@WvtP_(en2sbsW3sDSKm5@(oUJjNdznsU_RKCi78;##5&s49VY=e?<gMD_;poSP
z?Q-#!=K|SFyIfwlh<{r5ds?jG|C{`2mnRB)E-HI3s3-;RKzU=Ga0j(eEIkfj#35`4
zV_o!iQ-jm;Nm&*vqy(6M8S#@|DBYP&2O@tP&E4iTzF3KU`m|*6jM9^^mL@z9%5)=<
zmi4A#m~`8dB15(}9gOuWE;EIVUD5-|myq(eUf<T`AX-8BY(LdJv9=%myAh&cU%i`-
zcQ8<rk6m|=GXPZkK*q0H5bzH$zj2Lo#k}s{QQGFITYTs5yGa&E7kHg%A4!`jO)w@H
z8ZJaZ;)a6kxvp9FGr#mEmkM=a;o&<<?+k0dl?Q)yz5C*#_@RhoV(YT+A6(@PhkI5%
zes^W8Q(wFYK6UDQTY$>%tvwOskSzpJ%vSD+qO9`sPzo{S1xN@yqZbgO-+|`O1S>lw
zlFMSCa`Lyzz*gAOtkViASBCjG68uq|*7+&JaaBCj;o)8t25owI8K1lOX3aqVDO)G3
zQEU$@tfJrl#lW*qE8e60{LMZ=^|^=Th8*GPH^cC@tzlXR`(lpZca`sVP%yMK?QKE-
z%PJv%nWoiSDcs`J0J!aqzC%f-aSa!Kt_IyE`>sg5&9oXVNDyU#SsEj_+YeV3R>{g@
zM}HVpGjHv~H)6cuh(%J0g>@V5_2MiO)Y&@v_4Q<pD)GAX3tJdb6>Ya5EK&^JQhTu|
znRvaG-AFc%mUlmdyJ$1C67Z4)30f#L)TVx_U-o=a3sV77nwpw44tINi{Y}{;Qe_uS
z(E3G1!5--yfnzCjrJFDQuXG7@K7^D)`mi@qbwwri5!)?&@j`gJb1S*Iv9YmbXMR)`
z`HNBhFE}n}BYpwH{xjT+=s7by_ta@Bpm>ec5oY$_@-Js?Xntu`(QEln!YnR%zrvI!
zryQC?{r3RI%;5n!=yo)#g2GjOC|?v@&xmiLGtp2u!-WHsLLfW_Z8=r4pnNgGeST_H
z2+*;faJ>V5r?67>m6g7D@z8VgZ|nayNBa*phwVq>(cB1}4xxj<)oLkfSM3bvPj|)d
z@BY(1>^jO>EV}8g9q2NZ$SRSXCGqurUf++?pMZ>CiV3raoG!XZg(r9inUbE_9%fY@
zu)4RtyRtB|N^Zwqh}QhvvQ`W}q*l!@Bg1{df>UrdW$|N15E)h9$u)L`RSC~4_MEdS
zYy}qOf@RiLw#d4ee1d-Q2+Jw9QV&8ZL1wWBkIHM!1>3&nv^sonnuQ>C(tlX2y66hz
zniy2%B+9<o_;7QXaW(L@P3BFgKkF-7e;NE0K|%%}tM6s&yj(NKJW!_C(Ud1;TA|BP
zDBFe4y=fg+=HF#YOp|8k^lBRa?L-u%)prOKG%B^JMH9laV>}doIXiLchZ;p6RKB=n
z?eW#aNO0FVj%H4K!}&}Ars`3~#5V(=hQ@lnwUltws>Se8v;SMIb&Tyct2N|xPxYMb
zd3nvI=84r}UttG_KIY3|NKKk_(S67RU8M(7$|)WMxXa4!PuteG8PDl}3K{p4W_Fqe
z_U|&v46kp%x)(ye{%Cu_0ePNbbXxyh)bi>?P;|T1NKaMk{yCT5<VvU<L2n8up<xXV
zAPLV^V8?k-=Ow>cCoJBLk8L2;I#xP{h3)T7{M}^EGad8nIWym9*6(N8RWSINZB+v4
zX?7lOr9Nw|V}&tB!^V(jtj`;=i`|zj#IL-#dG(Y(OHf=&l&{J?dzz-k{-3Ty;2)JN
zcG%EL5@^Q}7kffL(u9e>y9o3wZ|v`H{M$?>?1fFPZj)(-)2`P_pS^tPsvFOSE-It%
z$`r$Oxe7=&s=YV0?P*#~!1lUjRwd!BIyUUIq&~Y<xZD1DYfaHSU~g3x?80t6+@%8<
zi*W32<!r{}?+|yI?3G=!+OgTu*~NaxJV#T^@3Ez(A>+{Aw2$Xs08mz6!}W!gUB*J7
z^6C&5%(J)m$6#@)7ymY=`1?*?&sQ_?5s=-in%#s&_0zj;`$zjzN24K`W{b=n4Y7Xa
zeD!7exr`w4`s%^v#NLGRwsKJY=(h5nyD-Ney9+1VF1<eqg@jNW;W1gMb_U5DnKau~
z!l`xKDJ3Lh^7&bA;8EGfG_LTDfQIqsa2e*kZ~Ge?6RihbM@F<1(mF{@nZfc0cr*Wu
zHCEy*s`?g%rE8W-N{imB*yi>A-)qfRdO=OLy!J(zSXUU(Ol~onb$Go}^UBu5leqaf
z1dk84_})SXTF9-<jm_h0L9>6BdGmPoK;hxc0WZ+8Kmq)765rw3^{?D5bjlTZt6JQx
zl7Kv64~-#2Rrc7UiVR<2-yAT!NqLRWp!`|N-EvQ#2^oH0Ecr4`cfG5%?{lyKN_4sN
z%pR2C8gq(L1TPkX9b~GVG96$|2pj6yM$<dISL2Yar@gFdr@k;2%g8m`q4!wK6FLlU
zMf6vc0=<zMdFf53%g)8eJXMdewLYbA?2pZ;2}Sr^V$)EpG(!g2t7Jye=OTD<+!*7y
zXD{Z$m4hr%OeIy)SKNAR6DY~hqB~zJ{TdowWA)2YnDZ7zhwZ;p8T9If>%P#UZunU8
zka$!;MklR?S9lSxnG5Eo;%;2<gJ2+q*U1t<aikC(-3iw<uf^Gq;*^wh9x|oJ#@V}V
z_b3$vS!sm1zR#^oD?c^%#;!x`wTrQ^&9<0e5D7JwFK3-euH`tjN@}@@!hPVKY?#tp
zd2;`B>4GP91p0e@eP`-ubrCoiok~+!_|b8h?-=JrnFgGkSLg%C`yyqLCZ$T0!X|(Q
zqp&b+Nae+$;~1WZ0c$@cYV}hgeqUo4W~qy|xjIa1&jxMUhyTn#+M;VrJ1o?81Y{h6
zz1@F1J&UvJ89n1Fx8gKb18<KiD=XLQrQhp0Cq7nitvlSF#TKLSyI7DlysA+v4}7wE
z2v$zn(^`*IR;w=Zu=l>?owzFsWL%bmX1p-ELQj`%-3M<@Sm?gNBW|t*P9fsMp6j5v
zGWOX7hy!JR%mls3P<a%62g@_sAsoYkP#*{zZ+^lA%hvnNWXa&GaI~cB&D!JvXz6BG
zx?>Y@z1IB)>2zixvAtm3u2uEwVJCQjf-$>mM}N&^!Bb2H4Ht)h0WkUUHZW-XP{h4U
zE>J^K>h|cbbK!82$NoxC2*g%B=J+B^k1(0=wYs;|0f9Aiv#D@cz}Z)vHKh)1mz^4r
zrQ2G++w71lNzwmH$^~|Irq;a0u-Ko%mDrfNCefOgTF1xQzKtQPa)JrV4MVd#Jr>mx
zE{0U>xz5+%Og#q(Mmv+n6Zb@l(6?8~*mLZuG}2{ohdvL=W=yyIR7HWkokf`<!S<dW
zYILGlC5Y7n9&fjc1(HROq31~&dR{-D@RfhLC6a;0a3a_y{#u@9*+!+wU`KjQ$nM9m
z4PRJ=d)tz-ylV_B(~4sIZ}xdun+WN2b&S+DC|No!0nu-kU1Rj_V@2a^K~<<lU0m>$
zRMlb`oJ;G(>*$$oM5c;(1=Y2WE**C#b_iK}an$&o@u#deg_+@R4tlO0bEto(QCllE
zX?@C$kdS%KkMI3|!|f^=R>xn&cvtc+o^+JynWci#D+?XJV<?AXpE(dQ<_=dLcrh1&
zfseh;RseTgs_NCE7g5d)WU?75=CMqzl~-8A^@o#DcNahqq|n(W=z`%P-0|d&n7IEL
zIzCqSq3-O&;&pvV#<1<g)dpxcC_BEP5#-r5m_r|}GRx;5tGJ`yQ-z1#b~9GZT3(+Q
zoIScZk-cg7^8Vf6pkxE<UJA{n_g&@6bE@!D2=M9XNP?a1U)wd_YXO@9)(`IM|8vWh
zLHeE1BH($%E<uV}L=D-H@*%fQol@P-Su3m_A_{VTPip(XM2%MOoS<w>j4nEc;qjp;
zoW!g8>7<AilMiU}!I>@j(@TOZkZ_$8OW|k@?q{r}jnFTGdrO~c8!}$~S7F#8jB<;^
zj{#Fh>tcsgazD@j9O2cP1fF?Mi5(qG9fEgEc9h;!S;BOuT|F)q<y=3t25wQYL1Sy%
z<O7xaHCbQpq<`zRtJnOT@k=GaBlvZC$s%~Bn@%_G{Yhq~l(0VMQZE1GT_K5(PiB4l
zkXE-E5VF6u3J%kb>6?OZKOUYItmih6m?^F5RAP41tG9BLm{M8yGct?^(IJRgQDx^z
zR@v~<VlkcFpq7{+<z;(tjiUV*c!oSeRl#3v?xU|aJu=89ZC-q*o{vBt=bF5R+E>TK
z@~u<%_BvqQ#T47oO`|jKq>vTw&PZ?T-9;L5FWKHSLKPu8x%j59D@hbMr~~ILO%d#p
zg4WAD6z?Kkj-LYmjZ}`#A0GYzh3V1JnrQKv9lyIHMP;RM2oGNIIgVHba?@0%MxMu&
z7?$r7ocIa+G%*N9(EB!vot!$W`8r{dXoItw19nSL^S<eRp4_T)zG}i-?)CnC9!k#@
z=dKlbk!f|5o1_uqJ6BDhl5R;va#Lg$XSdgm{sn<Gz>EE2Yg54C#<G8ziGkgkc<Jz)
zE&Hms2!y31Pt=&iZ9X)ZQI0e*61Wxgaq2Oy{>rIvKKOvpH=Yy!4Y|HDDZHyR3P;H)
zSO}bNOK9go%?Rn)*m;dF&>PdQmpM}dJ<sVNIU}BM|H~K@AE5x>xzcJI)FsHT2&vL;
zp9L|75jzX18+}3~>s4Hdmp71reo7{!RO6D|tHzcR!01;=aZ{75(oUnG?LF91h?cgd
zCM)KfukmE4e&-^rtK9sBQRtJ_{jl5p6^-Q5|B+V%a$G^%+pd@FP#!;-Ge6NznmmIy
zRtyo5-3WFb00GJ0QjI%%m>4Xfz<TDGH47Ia3eNg*`>O8R5(mEoOCTh9{B51_o$MSL
zaMYNYpLaj^S#bfLERkYs#fgmMH!b;CM!sh1GTnx-x^?{FiebZL=8m65tv8rbm78Xx
z=jkCG$FV?5U7BW>Y0f*5q;{G3ikFfh_eF&pugr}H42#Ob^&Dq8+N^7#ncGz#A9xPJ
zJ887N{-ce&qtQLyJk>b6)Y+Z&QP3(ab~N9rp{12R9_1deBp+skD1>cM!<0?*z9r+k
zRn3K;Cf~a~cFx9J43pCt;nI}bHzk?~?EGDQocX%_kSZp3zkQ5rr?=`adyFL84bG)|
zUfCf2tM_gV**ol6|98=XCmOXl#~9ri&uY+1DD#h~`D9UtN6|<0KM8XO-xd!%W66bX
zxw1W0aZ?zIUfiEK*rO?Ky-j*(SHCIWU&36bk9s?{YNN%>(p`W2yeh}T0nfCU%ehkv
zoiy?~+_L+3qxcSkIUk@ZB3mU(CxM})e>VZ`!(c6u-nh(#o{x34mziDLV)eLH*3N?&
zIg>r-3=9QboPlJQ1eF5!rQ8xw{Lxx1K5Na7IHwIC@+|fkw}yEVOAq${_;d+JIYT~!
zq4YOjwfQPrObeDeR8{qxfl|n$<u-j|n(3NyzK%9sTqP=lcN=6*X7fC}IzqHw1Bbh-
zJJmzAJBxW*tw2y~E1=D#cW%pf+)^|?sx-yGIxCh7J)aSk;`uFKio(tJ$@4A<aOc%P
z$=!Sv$h+qqZfpa)WUveK--iE#=fJ_Gn(_RrtjCm<)9X}e_l<8I`)-O6j(+5(;$30w
zyEIs$FlkO>cgxrJ$)1IYH@*TcvAh>-$(g9Hk_BUX_*K9=-d0?>KpgJTXcczk9=dm;
zS;-x_kRMgFOu`mriB)+fr&B{Ta{p&J2vWQdbVFZHZd<A)hK#@O!jN>hP3HTjcK3_%
zOs}<kLc2-=M4XtKRU5R16L-~-viVN;bH^#TN7hv`wf7IEK#6r05J(S{F0w(V8ZvAr
z*7v!SMH*uR@+rKNo!SA24tQ#kaPq)p(#82DMml|S>hM68puz>}CDxSr9&!uY3!iW?
z+B*<Iq*_eY<P9e^)Y^DvrWc(I(6lw^fJ{%1+!%Cw809^3nxQ6H(EQz$vg`2_k<VIA
zMt};74AyWisvNKy<!i*OZM?)^oKm<_H!Mv&j;BOqS{C?rm*ebQ?d&~u;P<{1X4bUH
zNXra<u+#iVfV}a462szr`~zD7x}-hqn`+rzg)_>>5nkgRE1mawN+G6kITz#G4-a{d
z{+I##T|n6OI&bgvM17a`(N1e?i-u<C(IL1L`@)5xGy@J2a)~IgTTBSl;noY4Ab5?@
z^XOx8Hl<`jk%9Rzq9|n$Tx&WW%}zKT&i2gGZ*eh1f5z{Tor5?~+sDp?hqWsQ5aQIs
zU(~4eFM3Y_s=2vF_AHjM=%}%;Unq#KnyEE^VL0CYn@yTUz@_nygW}EYZOttOEHB4K
zXHF&Hzo}tkB>b374r0x_hZ5atZ>MEQaO;8E=<Tl2^x1heZ$r2qJX(dv9d2RtVY_hy
zSo3a82AtjmZG#=}RipHs$N$^0?{CfT?~ZvugA5)`#8nbc6XW%kp0@@C1?{!d!T@*|
zO~r(&sKx`P>*U^Q`{UM(z&sFllMhq|*mRJ$D~R?B&(D_7!k2hf7{+rE216nlnF10?
zit3_ymWOoE`#D)}#`m4|^U?80HV!b}IC=(byR$xD_-8&{wrg~Bt?I&2Pde2?%Q0R1
zaAwggFACL>P_JK{^QxjDCcAQ0F`lYuD*cUMQh1DU>LYUV-)=}_PMTNqJ(&_<bTn~o
zjr#t{(fT5lPWNk0dtY1K5StCHzRGt)DZroQK4-+m4!L+e?;GpIBg6$`Hks)`>UH}s
zP(AbCU4Mi=Yd$O@1>Zdn`_hKK{>v$x)4a%wUS8>@8%f2MT6lUHB07X~VmZ!c&f>15
z<1<HzCqH?J7oipS_&&lfinG@MQaZ+nJN-!&8?~aKLT?g#th~Opiz4VUtwWsPqfkr=
zodf)pYPD{2M~(&D?a7D^;uQn)-Wy}bNlItj?a(-VH=edg#QuQD(X%t1g15fM>D|AD
z+hB-$)L!LVfOJHSJ*maSF-~JNCTS*AQ|lsYe?mkj;>BIh%HWE!@G0PB;_I)YiDW_t
zlu+A#n-bG_J99%QGfB@-Ux$mb#9;y!Byfjko<Qk~Gs141@~@g6c(CVI-p4*L#8tS6
z(=&{CmpXnm2{T^)AI^(sx**U$v2l}eTVE>H_%-teMEG;ktm0e_KfjU^t_o=p>+Wjq
z_eQ5y%tto?UE&LDGl1C=P{Oj!Xs-vwZL>S0Rf)hMGcUcz-R?}#O%!jt92YAx3f|=_
z%#>SjUg6krrnD#hO?s-%T{B^j!oTGAda*z@oqEIsn+NC99UT)MB)F^n{u#QzaY$XH
z&B$~F2zVVHx0$Z1#wmQSzGvE1LS6rF<s94+Y-#trfRfS$&PC50GPKO!8R++rYUxeb
zqF9p4`)tbKSIWrLuY=1%nCSt8yS)R1vwq}+Tu`;)=@Z3>pKHg7mJ>Fw&JAGVsMsd;
z-~X!}1J8F94BAbJ+3{{o1n-CPM_I=(cOt0BZQb7kfZ%<bkxNqrpJjR-96hdNE3F28
zg;TMufVwb)A@*~18@wvf!9inRv2cIFeJKJA`5dmxA<P0%yCCOZRsZNpa$~(zQ5|(z
zm#rNJ!6KryCJb~IAka+uvR~lt=HWpXji!|-;cIpl-6BH$!(AtcdAxn7$;UI>hJd$i
zdruwB(zi=ycfbT=8zI<pMV<^lVb!@D7T__%ZrH1SVkGqL?I4%Wwx0Bq935A?oA!%%
znun%{V{7QaX5JB1JM{36)*`w%hDvZw#-$cJ@k^9HL{%%Dk5zSTe+Y@o@PC=TU94Ek
za9!ReY~X$h4{+xUl-7%em%jQh&^3wc9EVzEY;u`0Gv+Rp9b(v#tQ%*CGCALs0z^Z9
z46FNJHh(W|o3(LxTgj718#=+!tczTD*n)A2{g66%zCr2~L4{q?DoEj^aIjj<f3r!J
zT>5>VO$uwZgygImH$SOGnTExKk~z|M8V$@L|KTxAtQ)sQ;V#7qv+y@&e`xnksX3aG
zW#d3YB0yPWQdx99mk!P93dF`8=&4J!9S)8A&&0)uitX1QEYE~ow2VtGB9M{6qt$JO
zPqsOc$VgZ?6tdR|3NVK?be1$z+Tx%pPH{VF#*#im(!sAsd#XD-e2o?^VubDpq)UNj
z6OI%6SH=?W80ZLMEhFHt`xUOqq(`<_2V!NU$13P#8D4J}Ln$80zw5wX5x+^3b`|c9
zcRVIEVg38|XMQ*tYRpaZm>7#h)yygu`)aR>4ayPi4q6ZCtszfBg2V72v8g{Q*iNI_
z8|j(qiC!0vLzq$(<7dId9~BcY71Z)vX9`AK;be;ZkB8EuoBuZUcQ^L`${wvcCUyxt
z|C6YFFf+Rd8ua#=UFZ6>SGSHYu(gF^x23jlyAh2qJn*5RAtC#?^oew~TY#p9Fc>xh
z0QLV9Qn!+|g&Cw?#6I?9GkD={R&0i7fiE+hotux#g3yBt`J{BO(BNIr-}8j30amCl
z9|nL&8;loW95QBT2Qyf0(`TiMWr7cLI^DB8zW3J2{L1!CthMzvmej@k3Lo?2YigCv
zzR2&(LE$n)_sqfg0mn2dPHSzA;eQSrvYqYW|F)G6<J708Ho2OCXhQ6??C*M!`>cEU
z>c;!hcc>-(Vp{YW%`eW{l2V3F)!M)9o_N+{tqb{_qpMidnoNM3cO7%>Cemd0XjtKq
z9V93%LzodD<M8j@Fk`!j>4i^}>@w2z;uLD7!q2+FjWi&h914W*-s2GIg$`YW!350h
z7cS|#9qws^@Rx7d_N1b3cz1nKkMZ|>B|M&HOTY5C89}uaO|tX~BvD3PD+=Jc_oWzP
z1stZucPdmY79u-V%x&pGmC&CfS9Fn9{<uxIL0jVRCrm!#xtyeQZEG{*Z$w6&7j2~0
zo&LxT;92(BelJ5`5BOnZU)Kw@i2YJ?sjk%|xcy4}I&08AuRTwG@yfz9W~E-fT}Zp1
zBH~=Jxieoh@^oykFeEqKYtL_NjbqXpBW|hbC_^Wm(=q&>mYTs#DvJ_Yd_Q{ac^jhq
z;nZZnt&D*(v90_UdmRx_?yoU?wBASdJv}{q7p*?6sr@zj;(6+n!B7gulbEI!|Af-f
zQRKc>DQb~6+*h)>Ji(lQoRV&q#ij`C9_%r9q!*@IwEKE&Y`MA6b14>mdf)WoVlwo&
z+U39p<^TJ*D}!0s&gU>j$+V&xF9%RUxdZ;y^GEwtM}JrMm>N}GOr^Cz@Ptx<-TtHf
z_1P})#Ye?*qJQdp9j(r$aYB-$K#}6%&eY)=aKHpC?=pC(Dw1a~SR%{gSoD`MM{jzQ
z%uReqNNBK@s9E~iY&H>KFG0Qi|5Nw1F*P<fQ9bhgbg=W?mI#8&`NtvP2MosR^Y^Tt
z5*wRn*Lua9ERLRopaPZFjx*-OUSfZP$dj1n+A`wBTCdanFI#8)N-fx{>VG~V<VCBe
z*_Bq)u7e)N%s-zapRuV`y+Wd(q;Mw~A7xuVUM`Vdmd$n=5X(Sd@}0G>aT&fDa=qkV
zk$v~c#PKH=3?Sm+2o}B)K`6T$M?@M<)3FcSlJ&T1*}LJYtQY7Zs*autO4=G67Cy}G
zB>$X17Ze$14<3_mBPLaZ6b$3s^UMYz_#-;9_FjgGP*%8FSLC#eG4^*-LD7J2tRBO<
zN5*6_G6v9%RlN|@9vezT9G8mvnLKgyokvezoK4c_itD<<&icG_j%c8&jOV5)L{vaQ
z;8&9j`Z-(=1;S}PFAq;ImhdL8ReA3cW^F7B|M9DO5Q<K@i0T9E^b%4EBZ7hLvYlD)
z?c23`J(q2n(`Mg$YQVx6JN`@(9hKt80o@g20|NZTkvt3UD<@oY>#NH>@@Bp`jN|P5
zGW8_gcHev$;&s@vIqqC+NHDD&Klo1h#AkqWI1PDFJkqgR_r~1a5Y8fS?UmGo2joB?
z0rk|M0}=SS7T;jqA`ekawy3yv+h})Ln@v}@%GI4HHScf$VUwvW77q;8^z<sWO#|hs
z01r9R(D1igC?vx&A!myHS*#)CXNPGCecQVq{Pbv$ZCKtj{u~jt=hCEfuix*~5zLgZ
zTq@A#NDwdIbyN0LHiKS~*ELSoTY7M_gad9w4Q|}6Mu>3xOvGki?}SIeF=7jd%{Glr
z5VTB+Z3S&dA2@3#;NicLRP=rBjsoDIyANI&a9LF1ZPtHuDE2?la%u*yeY8(K+Hc)w
z<{cg;f?!})u_y-{p5+kfbLr*Eopb0*^x#7<?CLT8=Smh!Bn(3_yokhdGLXIcu&)x8
z6IW&FO^=GMD%@}bz0Oin36u}y=or;bm<u5$`O`i7nzXvgi4pBIJka7$HOeGa5|wU0
z(QIjI0)Ier9d}yLa2Ct&62)S<!VK0MQaKt$`+D{3iS8R<M{zBdSQ(mnflaDJsEu;s
zKM{SfuH`KFl0=PVsX+WGdSRNPYhpY>yg}r4K^F283OZ4hJ0@*7xVaD>A%B+IJ45zz
zZL#1|1Rn*GoVp$!VRRX+R5ya>L<U8twvJR=mpQDg|4d!JrBN9gk%@_e=-%)6_s0he
zMkABS2c$IokA%gdN4S<gdi7PN)NWb&UM`yxdzW+9&L7`&=RCHv0O;c0e5~nMS)fCK
z{ATFG{TxPD78rdU?D@Tgo2Q)@HGG^fZoF=8e{5Jqbm!V4DIF2Q#?}x4#0JkJxnqC8
zh|{l>fou9toSHM!Mk<-}TPMW%!5$e&YGpo5p;cr&tiFOS!nqJN^6cxlt^YPm0<v-4
z2<Zo@;^xC}%q9Ub;b&Pt8VV#6Q+(@r(NK)3p7L7(qyp<Tu81T`lk%~;5zIGXc!VxT
zoPL036q@+Kf&wM8%_Oj6<-9#FS(o37+BkGh6kFpBVtw^O$ue)N9XnnI(ecsIHA2g;
zhuG)kIffnVg5R-5YHon}0Wxja?Eh3P<OEvo`sa=u7XS{nn0CTNn&{rbmiPNq8^Nh~
zs7nDFGkhXB`7ONLrsJfze;hbX%Yj5g$KiNF<fqBu%i+)AoxWpYTc9N)$B&CL$My@^
z6B35s-vT?r|2Lk_k9Ku+rBkVN_e-fK#C~e;FRvXj@{V?Z;3EcNo`Rr>4?>i=E>B_{
z03MHltVc}0!>xmxsV}2c*kVz>9QG^n3dtilsID$Yt-BY|EK@i<X7Xz?hk&mnCxzxK
z;KGIALzl-`zmjG)IaGlz)jlQ&CLH<}>6;ct=v$D-VLM?OEtJtq8>;7>?vBT3u%E+V
z^aZ|;)YQ_#rWT>V?54|)b=#Y=^GLu)*NFw$s!HbyuIJwruQyb-iGC>jv>ciOb_-af
z;a;cDAJS-w@d$ZHWH+h{c9w5wwb!i-GGHQHTkHC^rZdzNbpx}~H7fJ)UX{OfQCquQ
zM+V266j!?ksVn+pLfLydNu}F^h+BkD8l@u)-t5kq60^K-Benu)SS{VQ1<=q^dWQeG
zfPY<#`a&>VnHI!|GF3WKgop0Yif$kAZAqV4x_Up56D|Gmxi}Hrzoyn*8LvY`(K5Q6
zjSzJs#uHdTJm3|HVr^3F9P&75%ELZ=RjvQdxvST7VaIU4qm){c-*}^;a!Qv5TeL&_
zpO}*0lk+f~#v(kic=*wu^zZ^UJ?LS2M*$o1xN_f(qnv4C`f35&b*NCf;#)QeAf%5q
zW^CCHU6>Jsc2h_yUg#9mS>4=YZp%)AVXfU#&=2)%yC%=*jqWZiY%k(m64-KG#^OPR
z!PIK8Ge|u<_K$)TSDuBbZ+98?sWB>!@}|tIh={|*<|q~-=9~><KCj~Eu!H)#3Hl#O
zC_k@Ge2e&-QF71`ajz&9j|J$%H6yE!Lk<pCN5Q=nJ-uow4|F0orP`iw79K1~6axWf
zkjn2)RCE%-<>-ydnAhUyhi~}{zkb@<*RR^HJbgH;I7wS&u`qjEZb(c8^^KuiC%8OP
zy{;ewXLSa$e-sOR7ZFKf<(+*$VTLyjZy7%I;3blYQ*FsLU1V<h0rm?!0U8c6i@BTu
zmI|7Aj`dzp>cCfzaO(iO1+asH(J9gZ2BckP+SEml9L|_&zi-_^JUL%uYh85QCpr8C
zrVt(j>%G?DsbIMrFIo9>(mGYo${&LH-txjq!N1fb3Xv^@jBG<h%KY2Q#@F<$FvO7)
z3;y}8nHVlZ@LvxIO}9mcW1R+yBvXU6nz`_kW*{WiOup&Q^JeM}!tXXmF!xjbptu#q
zcyN#R$OJO|xaI11FkSO@Xt~Lk5N#7O+DuQT7TUEuqjyE<=@*^3-f_1A2R_<8?~7<<
zJ<c|+j!GMFVk=7ZvRkwn(Qg3%b=oGgZP4V4UEPtDUVuFdOJw&RjX*AoJ74lAxMiB1
zFWkthcaA%-|G;^t_cHrKrykcZd~oI#syL5TKs=UH!ra){*+Hg<q;#ubE@~7X7NXYL
z<NyX}7AYyQaq62+)@zkZ`8`YELmX%Om1|=u;ji#fnuUO3|GmJ%M!$KA7W>R4j>0i*
zhYIMq383?*{-p6hN(&ewW2n%F#|#<eZsPnZkc|WVHPrdvt%qTT3T#NW-`Qt%^aprK
z5yu7bCt;iaUN&Yw?+$gNzIac*$r<x2evk{g{C+BRaD-@8wzYM}#MsK<3k|m1>$qm(
zovfRDJ+6CB1nJi7k`dF*5j@qTm`sF?EyEdWK(V^89zmGjPx_Z9tuSV2!EJ7=E-)<2
z-t_hlY1ZSUE3L4z;l`mK4D69EvCf9h!t$he)&D#0v<h|cWv1a!MDbtPbG&;r71JB3
z3mns<MfcKcd7oS?PC3u5rpAv(XA;V<DoaYb@%(rO@l9_bxH!c{WSp2Bp%?X~pS%Vn
zVjj6Z0cEI%2kU#Y+oMcZhnoV<1vEu)t~#1I02S8!o7ZpiApAJ&H8!%p6uUebx+Yv$
zS35FX^Tp5&3X8RpOY8uXz1aRZEMg!Nb-g_#?1Wt@pRozI?QONMtj@;1IZ^J)Uxe*E
zyGCc%H*4;VGM9rjaw`|<)t8BBUckg|7t!WkMa_8hLP&**pOhe&e^ON?>v0vt0{EYG
z)x>$=XlmbE%=CZ|!1~knlY=p%lYshLa}(|7#3t$9QrT;jy)pf&B4G4&mIRW_3W28Z
z+&ME`_X=h>yf^(c)7xOm`=Tma5M2`hVTMM(YDp4{!Ae}9E(IaS0y6<r5zO*=KUMT|
z^qR#vzt1DVtf7wnx_KAT$^K592LVkJ@q~@*1J8S!QjkS44ySJg<sYa_u63HN@$^)~
z2f2mK)ASQ80?2j5@XU&b;0Q93RqETF!(YP21NMS3C%@GZJ!v*)8W(CRH)p%HQC|?n
zDHcy5Sl5Zh<s_=y9Dq1KW!4UZUWRcviBFaeS0)M+2~K?$SYc>tir3nkrKfj24pQd3
zOW}JSeD3T}y|kx$c^?D0x;ekGN(cLQ3<Tf5?beG!s94-mQsN_8XP3rGK!kh<-o|S3
z?`@Sq#RvqkMZ$8@a(-hef-0bv{JxK!9o_eh@JH4HQr2-`;qW?eb7-5!BL02#gCd=A
z?iSQvmott{gm9qFpnD%=7#a?}z3crj_RxG%-c7mT?O9#7UvLRX?F5o|!26#-KsSmJ
zM23}T{rXkz_oO|OOlo%C<=g#!hsb-~+d=bJTX)>phvbI|7$<gX;|^mQ#0odDQvI$S
zZkdbe>jv$=)s=Qv3_PILRW_BI`@w0oFLh&uGmw@>i1`TBa+POk)-HeC<7Shh5$LwC
zEG^*BCn%^Z;wG^@siW9A@*u1g)c^ac<yKjHR{Y)f(*lW6*DsY5329m7&oZM0?e*=H
z-1XtQhWP1>iyae5>0O4bZ!$-`Uc1`e^H^h-HUa7^TD3Kso4`5D-yO3+(OCF5bI>jp
zb_g~PhxRq98)d3?n0{R~(9*0bc4L3Y!2GPkIiz`3*eQd<*|)j~^+I7xTbtAMfS#2r
zAkgipt`w<aBn;9NIE7gZ^gd_kiTaBQpsQRbjP5ROhLCa-!9-{h#j~Q8_oPX!{~H6q
z!_$sSIHOHzBkEF+l+O(u*`W`a#gy*{cR#KUGyoFBeD?YSa?+QXk1B3=-u*M9CHdQ|
z8g-d@8g90TyD~`R<8i8#9*NeD9Fitv5o&DgJn@EvC3X?62z`iCv)YgF6EWTI^!Vn2
z?-tXr?@`Fi>@q`IREEk0Rv<{l2QZ?)7cpb*1nh5bRqfGe<RECc$+4Ud*$RBS_{_2!
zsg`|s;TyURynCB8b_mtA<sK;1q~u$LhBe=->LGDKX(o`Ss`*;MGEcjNO}~3pA2*+V
zu&DRCr1@)^O;Du>#v+~(8^EgXZZTmqZn1*%B)2nWqj>>%(Bw%qn6(EW$rVwIDuD%3
zVVMafRNN20LZ!cT!{;LOa=~OTH;AmOT|LnT!Dh8xPRHZ%AwO66yEx*6Eh%_7zhCYa
zud981QI1*9`euNp<7g55c{ubetGsUjq^i6CY$PSS1nN`nrsyse&rReGHRIbk84(C^
z++=DD9BFBY_Ay!pq=L78Ex=s}CSI2wSX(FCmhT9K+SO+BsJ>T^*bX!GZ}qAos7@bc
zI3HqfNBb|W<~AKjVYG)E$Y%wH-3Ilojc-8o=cEoM9o1g5f1y00WU8|HD|y+Mm6>?c
z6Q8wv{SAHrQ`?A5v`aKiyy9<+0^RdLck5Tc#fsR;H+lpQkU)z@cTxn6X9BWNBN-n{
zQ*6Ig=a&<0WpG-u7UyB?pSx8gwSJe7+nwR;d{exNj&JYAMEDg5<FymkL`A`pa%Xr{
z@wlxCic$6;nd2hiJJU5ZB>3*ad~o9<JHA@jTa|Kh>F~H|SNi*RY=QuTha5o{uFNO(
zR$pPuwG7tmF9Exn=K~rkblY->PMkA*FnXI;`)FgmBu{H<YAxE?>6hp#hNt=TfR#b;
zy>_efa%$Zq``o;i(4(#S&G}>wj!*jcHq#7m=(72n-w?1*1J7=CWs%iK2M@Ltw?H8#
z5_ue~7e*m(c@ni5z~X4nPC#RB)dZ<_oKhq9$9SUdGK9U7&D1F}d^)siBKix(P-mi!
z@X*t%!u8U;U(1H{x7PLD56=&y60h+Z2SqPayMjPI@<O1F6DHtZ7j#nJTK(6z_aBu{
z&}dfSF};&^p2S#1cdyBK)cF;O10i~BaGdBT%R-$<LpPaTppKaN&(2-4!474oZOVi4
zxWX`sL_PQNDxvnw&W^S%A2Qy0tlcTW_Ope{jNzWT#i63^d|*)%;8Kn9(N35R`U7?f
zwG59(8mGdL_*mzx>t0`0q$Zx)el4%JHYzLH^!PPc7^7rxjO7<h?`S5bJT_qE8R3Pk
zLc=2*ikoj6uv~ez3HRsJ*GK<RDReiK8h%%H6(U7YmG`57CMFTgw3K{*xQvtXQs#?P
zS8`6}^)yXA9h>-2i5Uh`X4+Kq5}`xqK2HI|<cpo6zSIp}&W`Jh4jn1Ch2j;g(?R4i
zNB3ie;!+k;BSL@==FAZiG$t0Pu7HCmNUTV(p1B1q40QRt8z5~{CZRFkgJ<j11-9Qs
z#RTy4HYFP^*>LWwuWUr<)Oa6z@{MGc$p`j)w4a<NUXqSvRI(~~4jQ2*>X3+N7)xWj
zP1aIormbfWDTP=w65LZ>#)^qUMYde!BV~IiFm;y3;$>;~=^P@pzedn;eZugwG=4ow
zGN3ZNut>k&V`C6?5Wu0#9+mk9gOYcwueP?-g9M!~6vTg9m-}*nB_Nt6qVZ=Jc09&m
z;_O<#qxQ(l@AC-Vu}~e3|2|pGgft7PDm`!A|GOE|STnv-SGHmF<XPV>IaYD&bgAji
zr3ejqBc~qEt^yy9CH0=diCTF+74W(53w%`SSnFS;Hl@1Rl_6sZAWNTs7anYHpE*4A
z0b17*`iJ7<>vd20D7SKFsbLLH<&Vto%~qKuPA$+?7v<*kA#_qwX0}3L02*1)kZu8j
z6o(P73TgH>5nt!PON%=CoPw4Fgm^@`Oplbwh^xAo>}XHZ70(K5&~eW%-T;)94moIC
zn~oFBWKjU+8`2jWw{2yWBCc;P1Vkyv?0GlG153qg;U}f~V;GDeF|El~0C<qUFyfns
zuFiZk;a1L+{z>7UOg=%;A(GgSd#xl4>}7a(%_`2P^kTI*krp~2;p_IBa>>VX(^mgW
zTborfU&auzj-R7AXQl2j;L@imhs2lBsnk*5F`9_zB!n<i-#L+hgz9r43gP_GP7CD)
zQWTX{&|)xUtND7-mH7G@_%wY957Y~k9Q7bJy5IkneCZj9WR>bBh4t_#F0EwzlI{#W
z3=`luNV6ZRJMGwmcc^6kWym#?!Ro|@yePp86K)8YAm>Np28y`oez@80ZbaNHHY;$p
zL4k_9<WxIk?s;jL#9E8>YOKu3lxIykW<V`=7$sizTpL^6Y(mT@8E0L}FDKWP_aVdu
z#Gki?hc^7W@8rQF(iA_257rN3MiSJ^IeiyXp10Lxcaq8N|J$CXj{i2H{xCWwxd3yq
z!m$lV+5bpy{J@q2Jm&0aA{frSpNa=@^THUG7&e$iw6piv+xoW=oLS|mo-2!tg;b6!
z-D&+nB>9Vw{K1sS##eeeipVno*G_VwbQ2!tBVDf$I2vR;y-Z%`CXC`eM|eZnGTWnh
zdCBfh#qJc%tQ5;{WiIRrmO5=(KUbksEqONr+&BGH!w!H=ZPStmwbfJcE~eA$gHOq^
z<=YnO>^!!$DS^#0^y*saF!HG-o56UF1VgxPmtb=Cs0BHr<zD>`DFx?Q6jRfpC9c-c
zh-!#Xf1}Wz^cb#d1Z^84RhW1(O$B@{`L4O0I+Z3yiU(>weBx_X{<GStXu|hC8pQ>%
z68X9ny$*mPj|^}nvbsqc!g(Uk`Ok%mCzO_td-&%o^Sw<W#?5vJTkQtQ!m2S-kf&#1
zFIma%Wh5-lzJ59<2tBVKkc3+`k+I+6b1y14#DR!rb8~b3kbx+%H)2)We);JTU|8n$
zWx8-G-_<>K!h#b$-?~T6(|sJ3&TJCI2{Hk^=cV1IT58kvzo_UvT4gruT1T%z&DBsL
zUbS#EKl-_rjjkkQ+1$v_@_oHPRNYp9hW+ABK+9$3?)xd)+6~RPXxiL#1Rg%9SnS>%
z`U;&eJL3b7icozwQOuHzHa{+uO>TS>n_gRlyqhDy&(Bu`tC|Nh99TZWkSb?=2!Nib
z&aKyD;pGoKSFA-ed0hx7`Qc)n*f+A|fhR%9=9Hu-u5U+4jd#xUX8yPwsY9<T${BXP
z<dys2K56|&Q*5wlqrasw&i*#W6csHKyYD7uz!rp5{OCpdw~)FS#_~ctO=KtHB(xKr
z8E1ae5@&uxDb-Z@WHOXsNQu(ekOj(83d%)(3_U1iYxbDin)iuQrrBGWoREuxsc_x}
zQVTAM7eu&M^;3DG-iDhRMi9!Z%d9IdLZL1<F6hY8j`2Tt6}Zq1re$<=qot%Uk>L@D
zN3Io~71;&Q&9^Mh9Kv9g&RjRzN#|s;W%uuB!?COmaMT2?mmCp2BWc7IAg8Z24A8V6
zHQ89+1eyC0E9)oGnEqe#n;Krg`x7u0NMojZf17yA{ILwbgy@3}9{_m#7`_$cpGt2c
zo{|}*R}UAzN0mYg5RZ-48Q9%rW)}=^@0M)inJlF*!jq#HN4W338vRfiady%o798gi
zr-0D?!{^U<_3n4tC<Es498m+tX<oA_df$XuSjugGcvNV@O~<UCe3Ys*4WHaD6S*~z
zH6%qGas8QwCweLjGj7mNYVygkSOz8EbrO#L=L3wbMdv(wDDd58Gx&;}x3x6Oc5Zc;
zU-(p_!wE=a8gg}ehuXDw;LP#|_gd9F5BJD5Rf?vrZn8ros{_Bzwiz5V_J(>WLRim0
z6b%$sLfNIyxyUo?y^(Lx;pDftx0c(lGVH5vhJe%{?5`H@tRnZx&sX~$#aemWVbKv5
z;~tlYF<(i&={-I4^>v!&7Z|jSq?))y<mOY+iBsxe>(&=``KGVe*DeRDfn&O+L2;6l
z)WY=7wz_c^Ah_(&!f#I4ug4iVuV1eJrn}Xsp0=Pj{$x;>w)`%zU07>(Uz-#Fa|Qj+
zgzE<UsH=XbS4KX-lGKgKzVPTr(al4o%h0wica7g&DKFI<o=f=(TbLr`6bcaM0^X+7
zm=p!WRm=^APk>Dcx&p}_#dX;7D8poUgx&@JnDWvr;F7ryCSl0iP4}st)X%c*AO)pc
z))BCkg#(1L!Y!p!Zo&A_*XgM-29%tETl4HXOOU;sA8$&={5(x;nUgL)L#q~JJF3@z
z3LQVype`_ZR{StbJ-7o^Qrh`WzaXKvNs;e+i}J=qAXm(fNh4xtF0aoed#H|r_n;GN
z)9`<|GXVb-Ea|a-W<Io_3l7Pzm40e(%8=#IV~Oq#9h;>!d5_$kzKEgQoE0(F(6>Cz
z#K_}Y68Ct6)`G|}D0ana%rJd&Ww55{3pp5*BgejS`YXCC$HhtBz~owVVX}YUp1UzR
zcD(6PRSA8T395Vpa_bAievFS7#vDYwL=<CzA5Fx+1g<tVLwFf$2<(I&7otRQ&Y42R
zj?=_0nhhJq-b?fHa=T*$7HV;;_q;d}l{3|<zQ4~biQaow{_s+HuT^h+h6@Gl+A%;L
z$FlP-?WrrgvISYX;IJK=o|hK`pcHHKc)pt95$SyY`^c+Gy6j(YDCG;#b)s!y85~f<
z$T~T3h8GspKaIxOjS?$|_4hvfR}_A-fR=xG097c*0}i1x)RLfRGOAcyx_NPabLX#d
ziS#&IVMXzvbbk4P!yq>c>{HS#%^6&I%Pfu+rnw3gqZi}!C+bp(9ZC9E*n%IAZKpeU
z5z96cbRgH{ba6LWSDt$m$!?$|0KXfO{K*21&h^PO&KP0gX|<~E@Zas&F_DKkfrBd~
znMK#D;ZQVW$Tqw5NXjxVw`4{xL0!kMVY$ycLH|J^EE*o&hVIN^!d$vliv%sc9psOC
zY*OFZ3hweRI*l~3rYZ4BM#8&l%tWuhkSILXp8EN>#n(3zdL8*^1M;sV0gVFUpR<#w
zC@UxP4%3^cBCOr5WU}b#vwjKA%JCLKnkTin+|OCW-5RD~1cca^El%(c_6i-dC3+3o
ze3f7$WD2kpDH@Z>uN>-$X(n~nw*E?1$2Wc+8}pSB|1~7o{y$B(lB%ENk1W#I0t13X
zzN*Fb$4wp`C4wou-a)qkW$)y#lsk|Ha^|u;??QKZ3$Mg$-pD@@DBbJq5Pt=wVgD8h
z!S=T|xTX%IJkH5=!t^&kI8LjK>4e64*m@y@$=;}WxCO^25rcQNKeNT<Z%O<@i1w>y
zOE-{FBE{yS-JN<+O$&M(zDDJqge6aF`U*SLR8_earYNt>lr9IPbS3tje{Y5vKNCr4
zc(6Tk4w)r{R0}V|EosR8dN!(gh+qj-{F_8TM5#p81m8&#Iqou%laqmBi$XLB@1KSH
zT9r$4^)20h7d6LY|N3#JYOz3cfyZ3_@C0#nwAq{XMQqDeoJEQ=;%376t{h8KG)!98
zCC9>t7oVj|C?Eq4pluQ&4*5_2nw8Bq*xOPM+Gw}vquZm2dtPg_53w;3joRbmkGb@c
zqmE(P;#seWcXVS^N+VI}7GNe!8a^vsRmT6u&kEZqtR0OMD<B?!CATe8?m`$|?O+EK
zZ!onF&VsGWPyWM+#qx5wh>9>F|9ws-DPK?56CbGa0-l8K{XIOa^)$G}=(~S07KyR3
zQ+ltMn(|R-nE1+~BGir?;4jIx_1(cnS?si%=&Mtf6WilYD)w54cAh5Wr%&$d28<>=
zRD4f@5MXDG63iBtnTtJ({ps^y3)M$oesi7E$rGVU0t&a~>RI3X(Yv&BLpnGz=dr^#
z#WS@ntL%W&L`y<LDNK)EJ!+FJ4!cD;-|C}N*gW!PJn+nI)*vadcvJKlk!m?hg#~nv
zF4C{?6YVw=2#F;xwQS=&sWyw_!_W+>M(cj0s=hA2l7i4<v+Ii9+p8|#K_Ak6q3m2d
zaf*Hr$fPc%k>Z-=FN0b`eSyQ<7~7W+8uq5rKtXw;r;x^HNAtG2a&Ar70?hAnQ;KTe
zy*fpc<<k=ftk!UIk`~KiSG!j6F&!N?Y%qLtz+RZNJN;jT-dD$r@b5=|mUCDZ1m7K3
z;EqI@$6fm|blD)mxu!a??d3Y;iSm!&=HV3nT(sZm5?4GzEp+l|)&;clt>xd3MUu8S
zfs0X>k9(bjb^W*4-I2`wuD&ulVW7TFiT>d_pu(L@;B(>F@G&ieet}uWmO;Wxk>CFf
zMsK~Z=i#ZX`$=Y<v*?WZq!{@(UmBuCv0yXgN=6xcix_B<v^Cm)9T=>yVp3G$`aLbc
ztJgm0y`fGV;UYVuvbaG0viS|6{Xfow8Ne9@W;oe^2PIWO7JZ_UnH{6%9*2&6tk2$%
zy`tl-DR<-F!zqce>6;<ju(-=-7GZwU(r!`rmlhgzr)D;58NHTIjRT+9O#yoi%?`LT
zY?7Q>yB{K`)4Y*-56(&)|HbE(&cgj|_vdaicIqegXiE%iC!*7<szz(S$;eD6(t+N8
zxS972+#7t{Cw|{KU{#HUjwLl3<F(9sz@zP&i;1O$5Q!d2Ng!FhmCuh<d6S-!E7OyA
z_-}`~nC||q*?QaO>3O*ls|*a+`A&Y!S*C?bp>uvd4+oaGI^}S*wRMQ2Rn~gZ{(JZ4
zwAJ!nSSwAXWZADJWOwpQJ$Q)0rT72p4{q5hi%Cez3(3jc5HFPbg`qU9TAOT#R_u*<
zC@A@`T7E-cM^=AOEDS=%QJ5)O%{khkYN5fZ>HIO!ftW7d`FX3f^NStCiYzh$M!387
zch7p`v0CpURo+Y>AIIl^HW*d#S}0Q$#try#?gyyGTP_BZp}P)ykAXVpoSlGw1(Uz?
z-v>g1^5s}`JB25BeMH4!zaZ7*^3BR-cTWpDWy!b`ugb|>J>4y+$A+jc1BE|Fz(+(C
zUM+nmW;P)F#f@v4q6j0D2{j~lf$k6#w2E{a_mIFKgTq=TN4+YDUumM##t^@@x4qc>
zUm|fa5&ZWJ35e=m;8;<FA#H0Q@=JC@X+`6w@Xod+LO8UPd+Y^46_SD2aBK<7z@Ndl
zHiv{|nW=WbqsP?Z7TA;DUoRHC*p=cRTlU1#UsC7zX!&-%koX*a#;@Get;Ks>dSt~;
ziOmmp3gUE2Kp7`_A0j7o_KTZ`@f<~j6p@IbI^U1yL~!o|&!Qj<w=l?JqQfzX!Hmev
z6~i(S)(}fCuYX_3W!wNK-`?YBY^75%Aw0r`E3Vr~)rT{M72T=&{AK&vadJ!?p<CFk
zuk4tV;&R0f78#9)@p!%x5@qSuF?LJ%KQx_tIMe_C|Hs6LnL|a*VMM0F5at|nNC<N%
zlCyWn`IO_FV&v^m$}wja8Y+jFp(t`@p*J?<7&+uLXW!T7cU`~h+P}MYZF@amugBqj
zzul|t@lJnIbTP|?#S?mbuXwn=DVcETKUpU#a`Auq+s?ohXo4G@8h&W)FBc$hp;gKk
z&-xrfeWg90)psNlmE%iYTZ<^Evq;E5?0SZgJ;#k2Qc|KtMVhV@Wx|&#`oxwP%vm$O
z<?*YYvlI!eWVE9Jwmb6a;Vxk2qtR&NN7eN<!9(xOVPAav5D0eQ3o^U9I)B5!_|4|c
zWaDQLj*@w}dkE>Qt@S#^)nZ%QC#5BV->D_^m>E>ucF5s&0S@^4n94oL|HfV|UxvEm
zl)E7*aO~|QbxE`j{d0jL#*8wtE;R04Pi46E`sr2wRHq~uqrdTcLzhSM&3odL&24Sd
z>y2x*F<(U6!fG>xsK{XN`a2u8)*khZE27qO!gp}VV8=dt4qL1N0UHvjY&$?vErZqW
z+TQm24px^Cz=!8=_ijH}bEgqWq#~8sjo19=Aq2|*SnF<<Rm6?rc#Qj)byo)=G^!7z
zmT4X}y7SxZcEF;w&9ui+<hrN`?DB!r)f92)lX-bwh#tqxbT)Juw5a}6Y2=*+wb6AM
z$L!o`AI}7j$qrmwiO=d{U)-JZ$d}$e;ORp8iwNUn8bN6iacP7PeL#q5qbDg>KvJK@
z<rnPNLCcdc+wuOcb+AmOtDd}!|D{0Re`FnXlQK{`#V@$;gC;jTsP3K-#1G@GiYOo9
z4yQ(zAU@J;QTgSk(k4egbCxHLx~#`>8m-5vZQ)X#>Asx<C`RrsO_T|n>VkachjsTc
z-TRE6vJJ>q!UnwpTV}{z+UnSbV#O$loe;w>obV$|j#46}l1GGqq?9jnawavX@FhEP
z%bV~iD=f<Vk#JuJ#ThC&3J5`kv9}^rfgRUHPTl#<VIAkoTM9z*C0)#HNE4hR6yh@9
zBdfCmNB-S-+stoyg*h1s<M{NXvAn;p8N|h-)^ky0k?&U#oRa=*9qPrr+?`0ZJv2Sa
zMMIA-u!t)Nc?R>}Jmu5V;;A1qXJSsf*H2FE-~?r{`h@p<=1uPgWq?2sSYC5`L}uNs
z^BKThufA=1a?J2wBn$;B!Fgy(AP2;5Vx>^?-N4y;M^0r6y)>Ca-dL+?Ydz_bo^EZ4
zoboz)q{Gyy8&xrvhTz(IZ1P#S-a#`d`fid~0ykQ00J-3zP@3Y50N-oBBlC9UF`dKc
zrRe2I&ihwFHL?ghI{}9~zmZ=bi?2@j8-xGAvH9gL4~&~`z-8^9;xw1q^wP2!ezZ=H
zrB1F9zcc@dD;1F>i!y83L$R}FN@7<p4mZ;SiL1@y;!e|{Pj<TS^r!lGL*j$Lk5U$;
zBsK(0IR&Sg;4t-&;Rcgx|H1*_6c6ahPP=Vr?P;pr#{e(AJLxf##a$rX+dv;9)5TpO
zeEVJQxk9VMOK@Wx0iZ0IP44+xpb}9D46pqlh_uP_2_1KDqY2rYz?ld^%VCozo7X_C
zq+l%QoUnLI6kdXDZrBoaEp+)+m))we?$a5cov(Glo<=JZ>u&U!6lM@A6Zw@J>ra`~
zyHs((0wEUO)>^jZl2d^0(??>puI{UFJkQelzODMk0N3O-smvAvYMEZAhl;X^{m<Xi
zf_u38=qqo^M*&^S>FD(U`LH^Hv<xj_RcVUQz&~jdW1eJcP2=w5WY_A^h4$~}ed+BW
zDk_bW{@Wzut@edwangUbe`w}#sU^yiSm}#${l4~P#0yASJ36ju-etvOR2<|UHP51`
z@0_`cJ0d>|&Xp$5g%YP*sSQR`|7Edr@o53GUDK!CUt!wwC7!aEhf0PRg(4E><oQ-)
zW&G#bC4?8FAjWe?jQ)&02sYuYbB+ROe5$<NAWi+nauMx%t+-gp4#ehAT%q0;_xrds
z?QX<X@c`DxkTxqNCC2i&BlYCeZV=X4>E<dff`#YhMSoV58H7B1o0TIvYU&;7ibPg6
z-wVpD*6jnY5<aG0n67H*1~vZe;J?`fw;Th9UxBtB)czZBFcB5a@kK8d{9n}Ix%TSH
z)@&8gVmUCEUh<u(Q0n`j{ZA{+O>h~jK#+y*Ki@Q!%AV#J+<9M`<)m-u2>GUJM7KG?
z{an|;kx!3mYBbLFPJL_Y_tL+}!e)~3aeaKE0y~#?P*EJ6SYu#J-Ag%;&s?aR4~frc
zrH+pud1*4=e)CXhly*D&nd#5ZbEnA=)DU3*?WJJZi^ix^cui(zL1bA-_F(T;2Bk9(
z{`fH*1Oc(|2XE}2?Cq_sjnj`j#&Ik5Wcpfb^V>yLBdk7hp4%*_)U$>@8A<Y-v_5LL
zRaNZ;lN5CS!*s_%?6tS6Z0kv^F+8UBR~d!M3MD@wp4v-_o$ed}l4e5Qo%(uO^OzLz
zDOx4#sC8qbriO69)Q)`SlD$`XM6<pGXueJ!WjSiXo1~B|w`9l1RD?Tme0s-hg^e^l
zhY#b2r+=$OAQ}FuZ^T`bkb+Kgv?+$6;Wv23WO8KRd)>UnY1=h*w$Bb}etL=)kL!ta
zPm^jD`M>LhSfFw2RRufXzYFbLSKKh8H+$`hPaMk>!1iN1*`hj;nQjjTj+%DKW5h;D
z$)i@gn(?tipf^F39pert{+Zq#sq_f+m*^*(PHxZ<z8J<FO1aM^jk1FmON8F-?PjTM
z*;hae+hjuxanD#AbP>XT=X<{oB<<VU&<`RjZEt67ZfE<_$s7;Gt%o&*`oo_+=bhLC
zgXVATGlnI%oqgEP4UHZ)hkc6n-5xUM;Fm}fY=nS87XpPkzTli5`}o}EZO11>F+kna
z{2{mdLs!=omd}B$1kEt&q0pH^$C#1C&u>5$qY3lBxcZXyT`L*OE6ast-d==6c_p~X
zZMh6Ag;@I`f(3Uoxt_9|z5o0JU?nKMF<}4c&eGUywU*;_(6K}?>j|VH#WgF*vDsPN
zTv`7n@-(qv&f&JYia}9DHwE`|i%bQ6ec@w_s!B`_y2hwJmFJpp!{f0ur<zE*(2A#t
z;-z4X)2Aldnu#XEIinHSCs6S^PxaU9LrVzZr}j6{d5hsWBYAT#<}#e1f5b(sNl&P4
zE!4B^UVsy^w%fb9+b}9#8F{?dC+|q>`j4AYL}|8YHXG-!{_3~<V1Zq}dPHb`Ijh#6
zs79$=Q-S*te6~67MQ0FZd<9=hIE`kI@jylLd;6}zY^Gf#;!7UoLthC>zmp<A*jr&J
z=gTNuD{AW;a1P@<IZjUnT%~{q;jDR__+5%TIyOZ%qyToSHzrh4E@L2yB2Kj(=J3qZ
z#(B%(;1g-@|B`b5({!q+{B%ABrRBU-6sqQ|pcGY({z@^+&o>?%T;lsZw`b{bMewoz
zPY=0|5#IHI<p#V0g&8e3K8%hvXI1Q+YM(#ok~}oymdpA5)&Ro<*~Uv__<ZH(G%P5d
zlQ5ZhO(qn|><&6HC|$JT2lIciX=za~c)*Z44nB)Q3faACdKo)6EN}9D?6AQL964FU
z&n6336KF&*BIfh9x}RV+Y4oGRRxhFLWBE+&B^3oZCQNbK$o=c9L2Klo^gvs0`s7LE
zhVA<5NkH)uVY_QL(r$I`X#x&b7io9DWW+BR7HeZgyDUOz%3GAVhXq)aIW>*pq%9mr
zBYBsh!--ARxmCjbByBu8hBVRKShLaC%?B}?L3Sa9dBhuefBHW$a!*Gla3jxg2fpgc
zj$LN9CQ&Uxl^K;WT%h#pIXNk>=iZvlm_#xhc8o6+3#R>DZv2-~@|BzH$2Se0(=1BL
z@q9{9b%vy6P*WwNy|AkYll(0n0S+8LlrJms=fBQb=OUaPg?`CdIsE!E`*6fXGSkJ{
zZ<sdhS$NUHVxInmS5(s=!|^?`eymR=J%PK|{IyPRVBjyQq~Hx=3bS-2>GE4{F41SG
zhgW*{Yaevq_W^-Yyf%MscAbxZcXvTEIE8s=SXM>%Ba6_iJ~Ie@pHm~l83OUbmDETt
ziW8G^D>ZpcR3aJKtJDn9c}vIPImZ6^{#pKcj(EfR-HF4l3EJ3RTV&YTpq(A=bn7s3
z(qnuUE~+ao*fCiI7LJL2==5cFIa2*i{@~K~@0~3G7IVymsJ0y^k~jP|tP{w*@tRM<
ze=(&#)p?qw(FJqVgq9=e4QIH$$r?03W>z)&%E0KK9P+@p*KZ)il?^wTrUn0U?G_a%
z4R8@{ktC`{jy$>naUPN3mSapH*ONyl%C7jy1rcq8pxvM0TWuTqo1Z+{X#|0yX4jdd
zLPODq<E#;wE!b`Q9`qvb60+dlHihy#AO|F{@zzzpn-}CWsQ{q)Z9x@@%;M!rgQh>m
zfvitU(u(l4vqn3Ov62FU^O$#Iq>VR`KFs}?%%J`ymWuZ2aZ~E*Bi>cCxWI{!+IE8w
z(q|6euiIRDU-U@WzVL7TkY?X!T!c`9^Md`*gG&`*=c9Wwlse>J0fy$bX~7P%xb4G}
z;{(;w{}$hl&>pqnN<YoFJRBH|<AQnu#8H@kcUgv;qpRFx!x~_d9#7@8P%Dt4PZAsV
zx-?Eg9+Xc;Tq1Twp4w>uSguM|KJtT)^@+cwsD>N^+h_vi0JH3$h~(mFnT;^46mZkO
z!cbdvZg{N1L+DuumP}K`CK@L5C92;{<huJIkFiWb#{;*zgx0Wo?J8yY71O1NQl-es
zZ#VnikFJV}L1yig&+F=KbHRr)5~5IPNM$8dyiXjXx}lW$7FXR!sMTnZy;CvYmk#($
z$q0+yzY5I`9f|9%;HO+^S`f{QcD<JufmNr7ZP++13Jcw->&rymiw0-8vG|mJ{={;0
zyW&6G8EyeYuDQ2F2+47$N@q4D1}~Ade>P_YTKN3?8JsgGd|0ef!7sufz;rE8$RK)5
z#~j^>U<$av5-jU!TgEAYj;dO2t~N9jbrNteIn!ca7iI+e7rA4iwUbd5RV0R=wd?kv
zfAq#QbEabW0?zp}F^M#4i(nnYl@|)~jivAP;#>yqLhmM`WkGKbBRuS;H*jG9sdZs@
zX`1#(W8H#TIfJ=Ku1x;%W6_nQ_C$EJV<ECDv-f~PSvc^^&!p7lp%!gk)ikb6dXo3l
zH&d13A9kJ0;8No{wdKIkf5UUmww5%Wte))dP69G*`3oY+Xj2G9zQCBUy`U;hcvy}D
zJ19WEvPAO^jZRv6hmq)yx{B_!qn~B)5S~)Uj+%)_k4UiiD3oI}nd&Xn=s7ttOe*^O
zSZEgpHF-wr9l9*6TqK8(V^SZIj@?x$ty-d}8ihRQp)3RdkHkClHCkEK19ns=9<=Hy
z5b}#6|D2C|$;`1E6yCNWG;Z@!2={WTJ=HPEW>P=|#h{5ti268d{#<RPS!pJtq<Z1M
z<dfFz75+?!p5}oEu%>dQzKxR&^&FpNivQ(R7J<J=;nGz-{e_@ic_#!4(Btxys7OqF
z#>Adt->b^yWZ*l!*9_1{OiFJ<)(fjAPfqO4cF&k+2-|nRU#ytfw@a|+z}fSan@M6+
zWimjueZ3kd6gzeG>PetN5i*n+44JgClDXm}y`P{<q1#;%NRfZ*RU1@NpWW1lEEWN*
z7ncfHe;Dx<1zCVmd?DN4i=H#D6OTZ}DabS0VPoo11|>suAIaQ<X5T%|2r-_OFU#yz
z{X*y;Cqm1lVcbfeBm<cKY3#n5DLf3MpV5QbsG#?|<v{`8IenGVllIf)dgio$+d>4Z
zCwy1CoVb69v~wC<Mew|WS75A0X~x*ryq=8vleBjZ>>@YIm2Qe<PKm{x=tabwQ7?()
zWq#=M&E9r1UZt(uki5xbDKxC6;y*)}$Oj)TgP2xnx10Y}`;*q2s`hS=OL$zn8y79g
z7$9f0>Lt*++6zkmX1gM>n+@zW)66iXEM1Wj8wo+VMz$7I_WtxC2#{#V_*Bz5kXz^p
zWiHz9pBUbQ-|`UvMQ<lOG-ul_3Hyt;o9do9NmCCBv7MBzX{-@ijqq3;h;uL$WCbT=
zXd5TIv(nKO#Sh<QTWUJq32)BBExH-#*9CI@_h@|YeSKJ?cy&A!Ja>B1yDjQ8!_W=_
zVc6LTWf^dH>NIWId{-KG(7PK--BY7^l^H@|PB$od{(@hbK!O>jaIJ<&GCA^KZE1X5
zdSs7}k#QjdbGREMu<$(KKM(A1`wgSYx6w6?fdV_v5TYuP6?5ebASc{?+0L&e|9n+t
zXG)O?3K5`|@KW56K=E;VF}wUn*{%(_2FM)7TrGutnHBW!lrp2+mL|Xq?w~Jl872Cs
z*p~LD_LoKW@lTe4(b`c(Z|>bU@FK3))5bWs{KsG+l5#8!<dnmcAxzHO0H}QbxkTMf
znwn>gcjH)K9VCq}9S0NOZ2Ic^L}0>4wGta(RS%u0Y?#m%e4&m2%z?(k7PkF;q26)Y
zb9&+vt9wCVZwT8306Cz*4lkYpPw7lS%wA<fl*x1bmB`V8^tv7F{Kn#EfppxUHn6=h
zf?#j4=KJi*7E#YT@CIdLfwdBQD@wA`z5H&RR=$&xG*%dQ$<0s+cDZ;RkGytOA$B5^
z==r$s$Y)Gts&Q}aL_ITuF_U*O&+kW3hD72Z_VQAta<amIZ&y<9jj(s}-2kB+5)fvB
zUvhl&QW9blGP;eH?b}p9!U#FmqKd@s22VYB!pc)|A<Ifr*%YBm-%P0_8E$4>jH{sP
zhbT<cDV+<aARKBhfiy(RMc>gFbjcpc&_BBP-K_J4e@{hUqNp#vo3c?y5&hL0A|<wf
z5iO%In3Ed>yYDhtSuvkCcl5QFv);%(uW1jK9q)J;F9R^c_4~l*K$jAV6a#(O;tXdn
z^Bwc*LqvT{GCUj2?p(LyaPNv=EhLJZSk=tWCyC=>5LU@4wAOTHF!#~H$^Ye2WwJI1
zZ|W;37%V8r#4d_G%gf>;En6QQO+=EsfZ7@u_C-$7t9$ygC1lEAx)S$+K?vS{B+Q$m
zFAF7mLyVlaG<ffBKAXwg^&{H)$aG_Cq6L^=klQDL;jZ#j(Am>}h0c1I(0}E{Tpx~K
z!GZDy2=}B5LN{U;rHb*q*a763w#<!`C-I+e{WrPW>kU}Hw$ECoTeA&P7`PbySJ&p|
zI)+lTpUE&m5!IfnwmDxK){ZR0HeA;uLAu9Bd{9o8#k5#a5gzt8|IOXuJ7rEr-RKPn
zQC8C^VEHEpVm+zjLbWvgo6v&KlF$VZ<C4T+(>q?B@|&~BXCXbsZ1rbe9rrhfPX&`O
zV7+!T1J5v`ZL`X!Wt#QLF)P*(uOusya<{x9+t&~-AuYDFb$oTzvud}da&@M?L`9@y
z?%OS<ptIV^*zCVPC6LPdM8CFC`S({UyJ@*Qu&0lc+$0x5$zCnCQki|(?S>>e?&R=v
za{6?y2K&e$2W7~4GWK&d5nY;rXD$Q*=uFwA7aCvh;dgw=Ip}C3v7V<k)BpV4Ppy9A
zRR8bAO3_M6wN$0+n4GGMVRmnxzt-rtIMk0}cZN>cLF5JJ*VRDpxEj<}6Gp7}u)+f7
zc3cG62;C3Ny(H0?Z<^-Z6?!=PFYUifZ0zV078ZXt<T!6JIk7k@MoY~Le(q-f!O@+b
z&4=gbO!;EJ#D1pEm0lSbY_&n=ls*l<6;o-jDy7biZU<#^4!(8Y5D>aiYZ{E=5`}so
zgnz#G|CQ0Ppt``zdi=o<yQt56S+v7^i?YolTpdWN3HI+r9vdGTFzw@Z0t7h&UdDsa
zI-yBU3T&?SaRbYU&Nw~`-!|%JF~U`5M<<1Ai=vfT*giw#RunhGW&GQw`0mbV5q>4v
zt>`f|VIHVL<+ri2u`!S5D1UbOKUS4Sm-iq0!Gt-lJd#Ou|D4AD*+-x9Z8>)M?)<iy
zy4oAy1<YEJjv{=bo4AR!26|)t>G9-|9qs_olUnOQ+zjN|o`>WEU`)RpLIuppV>j~T
z?NPM|Q$C~G8v<}|>~--?s@jjpOkEjg&W%!<;0je0Y}OV(OFc?HqOZ5GEF!yjyA9Vp
z&H%sS&hbu$S!l$Oc+ik|mi7Qo!VYJDfv*$OaP!*uIK2&bvRiYu(RQ+W=Ksj-4&s|q
z<-fD;`2e6Az3|ZH5*O1C@)dWFk=!~d@_A51L108%y$T3kk*yi0dOvsWW>#)hRVPM_
z08D8P4vx11fLS(Bt4ypH!XN(;>Qs%4$tu)@m|EC)gjLx~N!o^Hq%gS{5Vphd@W}aH
z?a!ixlDIBGRb~nM$p&gQJ;EC)jd>zYr0$VX&q8uIy9?1gJlj6K{81IA!psI(K-3G5
zm^{SW2MD<qZD0pB><;%k_6r$cMr6eY<I0}MK~#o`$K;;pG(C1g;AQSD0G)85-QPg^
z-X*TU$!S1=6LzM}b`BwPsngVMXgFxD9S+9nK^fA`->rw?-y1F*9E41-wl;n*Quu9a
z2)#Jy<k9ZXyf^pxN~3<TESR^`h#{xxN>m<*#~SHS{muZ1g84&$DT}pVzLtI%@L@p5
zzZ)U^mp*2pv%41Bn=Ue!p>+qr^howw`*}xwF6}3dK()6@`-&l{Reu4G7($0{;v{NI
zK$qFGIigfz7=;r+h8E}I-*y-}2`fJw();cxRs3eN$VnUiJJhy|Rj)I>gcMAtsQ_+4
z?PU4OKCUJ8Kt6UHj2{eQ5*NO-Mtb=*FYAV^5xNiAkIf_ei5j0=i;TFA^}UuZgcUv>
zs%jKhFf`VM>(N+aqJxJnvr72m3X^&JD`LyK%ah~AMvcr#d`esio$zyg7Oboh?b6WQ
z9{ZEnFr_g8IeTv-i`yyJi{)3143m}S&CIeHS*)&%SD4AdI4c~4_3@_2+;bS2T0MT*
z<ch?n*<0Ig%CF>?WUl3laD5_e*wV%idnXUa?WRx1#~a5}ZFn3y)0#SQJ`WZNy$1`w
zw<to9)~~Y@)q(J|<@;KRI%On%$1J(7FB}2hLBze`O`4k(ch?;B_;Eyd!qD+0bZ-&3
z7KRj)6iW$h%J5S!D0kz2<)FaKR1DKHs9<M|1_N>mmVng=(sbm}g7?|s-@U3`jsX4R
z_w{388zXnWqp!JY`MktQkww$0vY!H8z;rKH%|3lmTPH|fIG(vnh0h1ZDCG3ta1~<b
z;YiE;wo4@N<Y)i?vH@Vz|7Ovxr_V{@&tg(MIMT|XNHZ$PZey-An@}ROz>NP9DvGhV
zj#jsQjS_DcYWjM5fAyPQS}sIg!nVF)Oc_*QJcxWfhNF*<kC768W-u0Us&o%?+iP8p
z0Wm=NP)Y>_m%w@x?gD=BXC+sHFrBkbJp9Jy{u)v)Gmh3a)AUE~l1ZqG!6+;!;r5J+
z*(lALO4rQrA3$dHt^xbkQ<okB-;!AtQk_X2uWtxfV!!g7l(-WyeQB{G0|+Fg2rEym
z=U&FW&Dmqy9I}p(HvDFNPD$dt&wYcw_f4cvvD|3pPNoc*O9~wEs%CZBAqMh;zXO`z
zPK4$xb)ui~;b9G9fN*E;e=`%+vJ5A_HZS2GLx!QK2Bht4zeih^Hfxel_Y#5l!j$#T
z2sA|4CyinHHhdAPD3;v$`Lji{flykxZzE1j3&NfvDSP#ja{S*F{SN0U`>4)jIKJN&
zB`hoaFP3wPWhE2hWFKH3$NLy<6vbHWb#YiE1S$#wWg_2{Kh2@YE=`mL=W|PAEk})T
zUH`U@T8DW^M&-$(P*Esr;nY9LZ!w0+yqM1zaH_d*pcP1GM;n?La6@Ku_L<z86OJg$
z_4oG7d^?H>8tbd;C28~*R2f@&*j;!J{S|`;f4Cshl@|rUd`_muJTUtQtO^>`emNHh
z%P!}Yz8k4*L_ps}8>ag^T2OPtpBzbh)6b5#PQB0m-?IH^42CLG%KS9H|NG$hfYN)f
zK%=b%cqE0@Od5gpQ?kHeOu!~;$DB{Z@<;ArSbc7s?R#8}7H4ArvaiNfO;OMR%pqCw
zJN8$|0D-~_&U~>LGSCFl@?INuDQz$LDBuQMsC-J@pUcKQkDi|GZ|$xQ5+7mPd9y*7
z!v;_7#DcpDl!$3i;h3vC#<bI>-jhhIOwv%k>tm9bB7S1pE3`I|fFlR(AhJ<}o!<v_
zd4vP=*V|@h!n*0z;(9}>DJ}efJfdy`kiA-8iSnN@>uKrp+=x<Q3zzdB3XudtN!v;s
zlCoKTq=8?*NZ*WXW7stBS-OYIFQ>HD95uu~54bRFW}wfS<<(eK!ObFq2~>R6-pPl*
zC;gE}csR&=($c{j2NGAFuwh8q=^Ng11qI2_p|K3C!s0eWm%Go~0YI0N)YM*m<Dx4@
z{?$m>rF83HkQMeS)5nN5+{3nkZI_$@lyj0O4h(uMfAQ(31}=-bu?Ejcd86KzyunK>
zQ+p?qlPV(5?^lt7W3@+BFCx{{QQ_URZ#F*KGGJ|=bCw*qlxSfL-lee8L|~Lkbkn|v
z7)CCtE@kQNyv}vg(~4@aAl2+z?en2PD3io+Ng$QL(KQ#i*g~y)8tTbCQXLDkT^)sj
z>^#LNU;HcIsZxBqMHms1<|u)!2%H`dtEyiUfVUh+^y4K^_x#-|M|4QC=fI=Y#b2O6
z=f~}cn~^c<#ebUe5DV7T?=wW}=xV;wZI{mc7<Q>b#50@n&@3rK)EJ$251W!ELg+^S
z&3Kf5?`386D)R~5o%$)vGq9>Y_m$h}X^Y(LJBImm+fYT3WMn{nGi&o9gkr8;%L`_>
zTr0}XoKI@{t~t>B9sgCZ0<j6n)e%+{6V~#=kvRfc;1jjJbL^Y=YK4#HY=jxmV@63~
zMiJ-uMOn??%QNvzaT%g&Hx~|6HQto>KGZcWV60*GAhv~*PIoiM0jle2+wqcgn3dKI
zBoK)EsOQBTArrXP^UwAvcOh3jr>i$S$)VGt%EG|rNKY^Ji`so1Qmoym;yHe$v-i0D
zV^aHeoBqqcn|-Nk(|G&*Ey}{qad<Yt%m67S=TBW-3-YLMexQ$x<I{#OsDe1WnL%0K
zA9)K2wUJv-*XB;nYqjMqZM(S<iH&QKZP{12ntty_POqwq_&Ro@ztX~|=Yyy=tLrJ+
zHy1Y}#DficxdaP{oCP;htn(!Z`?%wP!?x2djm#!{O_26Zb;PvI2DL8_cDpaae4pHy
zn!4Jm$fb?(PemHLguJo})?-5buCJfSFHj%c{?U9Kc|pz35WxeWE};0FB!bf5_V&^s
zuIXWZ8ND5712ZE?t*y8|+pULM1|CF7HPoP3?oO+xtJ$L4BiGAR>(UHd*eB+zas5=C
z(e{93&*=6y0wRt+@NN)>*XP1V%gJWXgNs_7A$OwF&>~=EiR&WeSy}v`l}F>&0y5$)
zKu@j_d3}MGf`Kz$Q2H>T#AJ*FG8K_WGBChpX-sg*{|8zWh@?2LkXJ+P`f~$<<up%8
z<6!7}IZ(JALu#`i2It@j_1IWkit05t=6N4VB=_KNnRrf~%mtz~zGg@-HPu&+P;Ysu
zpb{TnB75$7+fDmj_0wrQn@b%;euc-V)KyRBZ4eGx8^b(5>?vs7oB1Td(IArDyrJn3
zu;&D^SG9kOQ6l8|`+l^%4ywGY5Ft_UCc8@&Y5{|gkBXRY<3#-E4lyE*y1f1=2;Cs;
z%XN1qW?{o<E=IpQaBwuX5X2xW8e8~)t)etdgkjmT(Q*TeG|YmKo;v)hI36lKi@uP0
z-+)VQRjYHK?ksdIF5}Ct6M6c&Fv}wR7p0=k8uY~5T}Hxis#DKNnYp1|QVZm6Jdc1{
zfWN|(nAqoW>l5`JawtWxzlf`daCn~<Uc2^lFTaxJW_dA(eVk7sh7yy{3g+IInMOv*
z*|*;pd9$=rXL#@K1HbNcNX7MsjeDz;Yj&sG<G7=RHDGo-_6mk&_rJ5`68gUaVe2>m
zs7zjHZG=(1S9jk$EbqkVGob<DC^V@4w3Tk-fup~n3wB9u%<M<}3VQ32Vz7MV<kx;n
zGut=k7f|Y-?PO{If<oM<hXbe&do-J{OqwqCyib6qqXCLJ^hp=K94MDJRvYQ7-Wx)8
zHq%`>(||R3!dk2oCL$+?{VrD4iC|1)`(aBbdxsIp&pC!*B#wVI+EGor9VU#f>hZWK
zV@zK++>fodA}?P&*}fD4S;pYi2fTDaTnn0g(LD6A#t577c)l!CBieM0D|Ar)p^jsM
z0>FL4f=YFIHWZW31#))${WcX;QBSWfd#E09ND(^S1UwO6sduyS%b9k&b92|@nbXxa
zvkbUC*<ga1wJ$Jg49sOX4y@J-;;b*Jl-DZAD*HysL*HE!Vfa?E>p3$)dR#^J3Y3oK
zuPx<eL_l}O3bRF_1p}s`xsws(h1Etk<$^(Jw2^js(32h1Vjkgzzrn~A9?tKZ^6p8n
z#IT89s+klw<u~GTKU<zhZcK+~T16_BhqtK-Z5TpTUIm)A8s0^s3Ex}lZ~#<IR&chJ
zS~Gk9Ipsk0c~Y}4yK5`aVzha=E8^^+_sS=(eYfn_6^4ms^T>qlj0Q@84Sh9Y(qq~|
z=0|F;IO7Lx`3*@3ln>OMY6;HK_qNrE|B~dS+P=(kYoh-~JS3`Iw-W)c_PLC3?m!T!
zhRd6W$-miPT<l*|`98(q&Wp%|IUBNr?UG<wdBa?{WSN*nK>q2ijkaY=e4?l<@x@M!
zoPup|jjWO5;(ZqPOEY+BFQdmYA${L?gtAn(KKZ}xpsAplucB_>3YGZk+r2T>W$lk~
zOo~RQi8e$ptU9iUa?fXNr}{Z8-%EcY@7wnc?#&caqLMAv$ez!ND;7nkFsaGu1h8fB
zEx`xwHs$q%4EQIaYSB|3M_WNht2F@F(|bBHgd8z-U4(g&drtOMZ%8B{T6%rpXYH4?
z{=u8R8$>+H5h>$yVng}qCpvk#e^_hme^gT+6u=S)+C)tVz1{56)iBrW;0=@4fh9w*
z&|bID$ma0F<1O3)K<jvv52l3LjUNGKW=>Wno-!qT4>GP>J}H#RXS88_wQIMLo+>>)
z>9Ik|{RACJ<}yq2t$Uo9h!;S_m9FIOS+6#Xl7mpYqH53Jd{JEAZ5b{k#O*4Tu@5K8
zV2n|v@N(oE>FwWci<tIUwgnl-=rqCXcA1ePs?iD!m&^UltyS316JR{96YJGmBf0#s
zGEP?ohOG~}@I(#y<Xz&}j36k(bFv|jx;LQ_5qUT#bOs#B!zH&?*x&7Kg19Gfa&p%;
zowt2T@7Kf(R#bNJ350=x`kE4=PVsw94-ZVBv21)de^R<5NTJ`3ys{j#5vKO3s<H{6
zorSC7@Fxl}IbN=PXmusyrdCdmycXJhvZgUL<j``J_=?9gxhUDeDcKEBq2+&hJemM=
zGC(Hqfo_+sO$N0^=+_{JwkaV`Hdn`XNkSWK@0`Mw$^~?{b^H4+)G{Uwxqaupc%z9)
z>A~<ApyQgJY-j`zbg3Ev>w?8<FU*gehd7F4v1r$6@1#X(>~G+qEky{COiVQRRTkH4
zR(&_5?OUpf@N(_GFiNhgU3KN059)#s&tq_r@neuwgw00h=P4!3=qc<@T2oUD&(4I+
zs*6;>d_!8gBx6)C&r7d~sgR*JRLs{ziPCP(KS_m^|H8AQzcbnxBiOSD=RYYw7RFtZ
z_u(h5jD-<zCCFrQr#nLa79O2Kr8^&%BmTgAwV0iUft~J2^EbY9rB|#VG)9y6)-C_2
zVSl0VkWp9n#d#^HemC+TO{jtL)sLXQDEkTGnw|h_6fYx;gXvp|lO}#1DjBnmh%<SC
zsG|)he-_n3WWm`O!3Tv|q?A9uar*LN>i2&CHXq<zWRO!F2HF|i*`)W`wnpU1L1SYo
zF*ufYu6_Q^;%`6e`~qX(uF~@i(y)7a3XtvWj6x;0<;y~(C=Zk8Y%8p)w3|Ox1UZ(^
z@+n^y7Ud66{h_CEYRZac7u7Rx0JbwOO?gW$Qe~wtu<Q=181z4IcAj!v9u|qtgk#}R
zO2Gj>_12TX?0CG9KE1xGmpSG^T)(%3O$LzK8N;G~XJGj3#`2@Fv7^@N`i8}s?qtm<
z1{nks3uYOl?%2ryfAZ-Oib+b$NGuab<4{)XCv5Zj^D*37Hg3nOzH>!%R`6a}BLYKF
z5~7px8`#ydI5typG%qVnp9>uDuKxXQH%*&yl?428TW;^^nVrVF5-X-_?5L;q$w)_4
z+ei72)C&d%QASaGj{AVVy&(Rw-zO@K1e}@-vm63kTPxr<JFS#j1B{jEz@{x^HtXwG
zeywu1?|_5rN`rOSr9c^`;==BrsT`1Ce(vMNz|6*R4;+!cCL3|Ma}~Xi<&@Onf{}sP
zE{mdlBvF~=%<VR0a_bp0Y&UL63?$nQ`FNcof3x?UWL$=sA0QK6jqoJXHt60pH5dRH
zn=R_kbiBMZyVdTYCJ)uaMdCvRrl|2$7w0j1%V()qxC1Uhq$z|DDXief{xRh&s<)L&
zCm*h2exI^*zPRQ{J#z%eGw;)JCZ3#mY}n?`e@c-zYI}1R^Qe*y#A?Pn>Pgx@!}a|k
z@K3H;R_PdY5=|_!`F*9<)rE|&f(+C?#=bB5uln88V_iOzSN#0^v%V=utAj_qCv&)i
zaSfjPY#MNagouBV$H%$66w-0PBdNQ#L!&qj_VAxJawD(Kd4f(1E^8It;QtXu<Yp<j
z_h;q10#ljxKWOeml{jIMKFmY~=_OB$Cf}2rAlJ*E9GU3ianW&6F=>#iMd+?M^^Ffv
zHa5M#1AzVgQf)3Ui2TM`18(}GO<{xJo%Z*xe|5FvDnt5aU2A<e8XZcn-|;GKCtt>;
z`Cu&M(p)NlIk)QrgSGM~VzvsmT)RT73^#w_b4VJPpVdAX%bA>85w=mw<*6W19HGz_
z;K*1!MhX--#l^=iO?-U_5|3g<A-w7yV->~ZFD8E|+b4B|f9&l-8X@Ch9^;&;y+XLq
zqpkwl>G8UQTYP6C60#(Wg>K*2#@)ca&g&%CWcs1i-V_MdH}Z<br<Lc?RSbTQ%yf*L
z9@a)G)o(v*mt!|p<BJ9(-nU!cIpekBiUq@D%(&5l1kz+j@1-u`hp&4{ifx4e-^42g
zkBH;NvPWkDKB$w(9M?tttDwi_L}?H7%~c>K!n?qjv+}YNTjyhD=COZq>ix}6I^>pD
zDs?6ocWe@3sL5h5md=?05EInirpT2&IqEtOo4<iWTW)Yy2#w%&iQeUw^OE#NNT}E4
zJEGDl;WXfZ{#A^Br=|OYu@4tZpDGEid)ZCXTFtRc?^eIoePUqI)e^CMvK|j&Tdan(
zxfG`&lA|x(@-NIH<o@2oZRL(DHcC6@<3&W2x7VoGg*lNeVMmz+!jYMDoFL;8+$Dpr
z?q!hGTD>2nJK5F3hVJ^-4H{<*fjBeUcKLmS^WJV`N&=}jynj+n<bSH&p4VCDQ3A!!
zxV`~p^TV>Tb<$jVu(XZA%(8|@tB0GL`4aC-^Ma#T8Pb&E<@<4@5Zb8q=i3--&(y>F
zV`aORJ<SgXro*>G+f*2P9L`?vOE3AGE;pEd_K~RFloZv*Y^KkWJsPVq!Z%#QTcZG{
zJ8~)SnG=9*U3~AJ+!6BO;lPq=bg9xpn%unyi)e+lb(2GV`k{~H+1~afvS-bqYnrxZ
zrjJ^GvAz`WF&CE~)ayd@aC5C*UtWpVH;ns!TR@-)#49H-w{|1NwQP^prK!LU9k0Kt
zgHcz#(C1z;^NkNL!p%t0G<Z@gTFM}K+3Rk{Mh?VfuB;G}$}jH_5bHkX{yWAs2Jd?M
zUD5LjuYj&N>P4zik2f>P*UsRQIE#7g#G&;er72^P+JvirZ>PiZFxU3=>3&^hj<SU2
zYKMmZPb@4QRK$_i&kUQGy215^HDc~`)r3jd%LoeQnI*E`c50fi<w)xT{g~qT(R~wi
zH4ZPO{j#OVDtupA?DGk4iof#uh_8V^6b0ed$d}e2A8giWv>p6jJL#f~L?<dimLy1d
zP*Pp46K;Sz4p!J`**oEm(7NN6<Co`Z{PJ1LW!(J<>D1l+u2Ac`pa7naUSk5uC}EHY
zIHH*4h_Jv30x9!|&2~t&o$f6FrT_YRN}Sik#)dS$HkV#vzRZj)%lL@@6SOTqh!S)B
z`#yF!v{f}vT1c>fCcVl!yc7aWw)T)-SmBt#*_g5qb2RUbS-b7-1fbQVR%PLc&(Z1U
z(4V$7Cq=VjV^W(LKzy^hwb@H<>V^rq)YG7RI`SS0QXeSh<Hv(x^j;(fi(GAVZ5V5Q
zZJpa1{&ywU8@Io{tFcD!vOcf$uJ<7P(8Pn&F<Kk{3*x1#^z(=Fejs~XjVV=_2Is`-
zzJ=BYYUa=rgd*vKHEn0q*~OsM97e`3@@o%#@(FXZK-NVrrw{<ZYEk^iV0+uiUaOe*
zg+8A0BYh0mj(ONrRv)RdvbYMr=udk5b37~Nmb2zOLnTQ2>el*Vp};f8j2k!N>?fWF
zQU>}GWfAZ7dmacV;qmNSx480+g|u;5<-cn;TzK)HO|<~vO3NhCWwcu4!;QPhg>dD+
zd_avwBA?7AXuwl~6%{oqDAL~Q=C*yKk9jlIIZQltEQ15&Zh4d4yyKFTxSsV^deo2I
zToOm|$HT4-<SQ$sWfU_9+H&#3MBdjfC6jF7u`CVd7<Z0LOWKr1Kbj^$$N3xo{favG
z-E<EP4?b|!upRr5G1gxRl}%_ya4~nf<&Gq*mov*;RAuV^5emL~dE!7i3kxpn2K)S1
z-^BG(W+L6eQcZI3#>`vWUTfPxP`P=c&yy7vT9DjJzirb_cHdl%>iQ~hq9ISR*Lk@v
z_)-Qzwcc#fdTj5?2QR6++5lgt%62d_wE4$&nfxt_Cv0Wra-4{us@@THK70bfYxU9S
z(ET^*S!hO)JqGVJXgMw2#B;6%)V^%Y;3bn*<m|`vrSnFVjB08>d!Im2u)PV$Ne*@X
zak?;j`G2DqLDXk5GME_jvlCuA_UA&Hw*4=!)AhNzw&Q~u;D$g7#-;X77&+RPOSJSJ
zJk3h!YMJih;{%1RxjYE-_>#rR!ixCThaE{X7SwojC@nNOxY2upbrawUd3cF<>1u$u
zA6$G^b6y>@9BT(;XfPS35q7q|t_ejRpKJiv;G!#i2e59Rora$|lM4t~11#iyqo8A6
zIUGA7l!`R!<BG!dJv6eh%QvP2kzIm<w4C}7#eS@79FWUX#@RqwF5}@AEv~$o+517N
z^6+m*S64#4<w`$`BkTd_cO|oc7F#{S)6ZEK3fImmZ<!nfG80lLr(QUvzBn&p^PUoH
z3B~xK;{B70%_&0+&9Nm@<m2@&p)(47`ph3+@meRY0B1~J4Sn=l@3khP6E?(~7$>1?
zN*}o#-mkE?%6;<uQg9i)UV(`wwEn2&=?QgJ$lKelT%KVlFk1F)RY-mm#0kK&e5hU`
zq^6SBg7`nb8p20G`lOehKL<T;SQ)z`Y!tI7qVEVzNhVNqJU)6YwzqOz6IMzCgOy_)
zy~&DCa9_vo-c3*Ztf>uiyliq`Z`rHY`tsA@V0uWtSzrC*2lwvX0}`9;o@K?wTtU2#
zQ|dp8@E}`=H0MbZ2!sW{?8T6_X>v<gNSJ@)v`|XdT(9@a^G2aX>bhCo1A%Ay(}H1>
z-8UGXdOOJxD~+x{wXpKC;1D=AtKDfeS81D?cE%;;i*2W1AxqZdW9~IEcnIqoKLbTY
z<@|S*4GID+1C@J=r@z}y{}N8}oYK5EpkMN0|GfC)pH<@g`PpZV!&em?@tDq;&o`qI
zql?4Fy=Iszheo3fznP#Pij}(d2dI9}3enHB<>-sAy4+sdChp3pO;|N&2R-p3$4QjP
z7|?L<s)FRYIvk`?gDfyDJq>BK%LY5;QeO}z_W9*kG1TLcP}9|@w&5gNKfu(E<b4;_
zok#i@^9}%Wc-9B5LEaXc^!gz)>k~07E3Wk&BR^J>xbgPvOfmd1Bt;6X?#_(Er||&)
z&DwjOcj&{+inYV(yH%>sswK{07=8Ty`G{kE!kF42>z;g_#EEV1%wh^oOg##pD(lFD
z`VXM_N9+Jxex}CRmbXUR(Nn2@&;tlcc0mF-PwIlHz<~j?5X&Rn0vSB|^0#le1r7XD
zvY>To)NWncd%b{$Q_Y+R-ws-F?SIE}?w3zlqa4i6MJe|D<C!StJ|Oa;%DeRhcajRE
zGIVVa;ITg5yT>~}Y;Y2(5Fn5q=BSTs-e9p*?PTw+TpY?7E6bB!om<%Qd0?z9UQ&r7
zz_bmys9{yHEt7|ii`*igP%p9{Bv_I!0vH>uU)O)Ry(?;5DbI}ti<a@x6Wk`Axf%Pu
zSSfp8M%iAd%fGjjkb3d~zI1?-*<e7<XI$;h$hCGm`TZAjwNbhi4PQUq96bBIOV?OW
z4UWqj;VnKsAZWZeF5{QbWMNq)xG;?JG9gqxodjh$g3^S`IMt%2sIAARBq1+vJEM>;
z_>>|d0SwrZHTRI5+HT?d`$Gn51sd_Nw3fYod%q~Ia@DWY@CO%}IB{AC<^b&g;a|0!
z=!#&y>lO*=Gct^rJ(MXPv(D60{#M{!$;ka01umNSqwE~_CIFfS7nLJm(D%?r%N`pG
zI{`SfMrYbp6nFA}6L|yxx@~I#$nrnDHY2!YQg(iQRXjQ|O8!s4qMt!K`No%bwGQlh
zUO<}ka(KrBsDBI%_rz!7ee+8+gP*D1gXV<Hqq<8|zCTgEA0T5#WsR~o4wF?4l>S_k
z`=Q?b!@@%4V|2mYv)-cwjgtkcSC4;+e+KgF2`?vh;+~tv*9qlfi(S`}e+o-HZ!QnZ
zD8I#81?tJQNCH1ctN9lWBaY8+KeMw?{<22ow!u5u>(-TdBXKh9ix39hp?=c@0Ys><
zgjbY6Q=I&@d9Sw$sy{)>?<@v0lQhx3^X>2-e_CFr-dn0>>&QJ)gan`M`_TBb4dEZ0
zBm}A)u@6a&<*wrWpDSKf>sHTzjHr1(3UgV#<wnR?7$K$`W7gi}Y%oiPzKve$v#kF1
zsh^37iQgY~@!__zKG5Xxo2(L?^8;QCYQi#4{BFrz`{`Oz9_9aH4(>(YUUO(w={rI{
z?*`g|RUzDArbb)Hw$G!{==*r^EN_RyK$kXcgWzUh_|78Cn?|y^5X~q0%v?=X%EwYE
zFwFh5H^J}he*Qp}o1Tma`0-2<)X=`z%WB)I(-W(H5x3icW7Ujv!iV%wRPSwjDH0rN
zBO;a6R8?=Rmem{g$qk_5fq6+w!Qz{^Iv3}<vGLxmr)Z732H0)m@3vk2lXX`Zd7@v$
zR(VHSDws#Ekh23n43H1HxAX(l7UoXZs{3uZ(sklOXbT<v8P}j_lR}HTqa&fW0*Qgi
z<YF)i2R6f}JDQ8wvWfl)@>LY__OfdYJvdn%|6j#{G6pS6-Prs!3e3MY3~b)K%`JK`
z4y}xaQLU?%wpm-tpKg!#{{$8#H@gE=r-1h<FQ{Q#s)xS`q%y}1M15IC{@K?DQsj@U
zdpbJ0#(rpKeh`VOe6aCFbz?D?L<$yKe};o(JMk*s@`!#M{Btd}_Vs!0A+vFd{(h2(
zot}$>2_sAKi$2JWtjbBvNxR-=EsqMqwvLvP>>$auN-dsOt8Shwl?gy%H>TNx-$*z>
znEi945U&G?#8*V^In0ag^`|YlOKLA3zTWLP=>qaSWc0-@DuP4=6ktAYHupS>TySKY
zjdF5W@Dpgi?bKJ4LK;2QmR$V|n&>6c(p|UIjxf02Ywba-ZIHhPeId8I4)!fuA=_B3
z;@wYt-EfDiN7M|?qk}H*4YQP^*D0ickAs<kD=VE5Zb_cP8ys8-!oo4v5<b6`+2r`9
zD9KgcZkJ(69p#1n><T(pBmZ&O0LOGIICzEDl6v!Jln+;Wh@zEMpoJe;e6T}__4SB!
zS#ZL_hcC=G3?6?n-=?fU*SY*lUF#dV2)n1VrRxZnv*x6QvnNXJA0A)X#zDC`&<f`Q
zO<=pNKcba9*%_JjM@Yd}Ca#D$3k%hjnC@RTvskOIYIN04VPBt*d;tT2cySf~V-P&n
z)bB><4_T73d6b{SN|e8}2c*6!ism2sY47<(>Kf=aAH8!79xQYr3)v&}*CC}bAv`SW
z5q$Ad0ySf9VwT!PIoq{6-QVz5&(Gy&{m4|_{i$H9-y!(OktVPx{BXRwENp}6&3{?g
z{GfA`69WykAEW!lPu(MND>#DAxU0QD5m!)@?hJdfac;~&UjlA*%<pG&PVOw^yln8t
ziO`L2o!jz4P{qtSKeV*#l9^4#txg~S1-0PT6T*E_L`&DdDiEy(D7p0-)fbYT>tENu
zyGI~As>)?fpY2)KR&X0K@a-Bayixh4{i}I&qZBLq)j^n!N8=bQ5BJOrf_z3rt$Dw|
z%xD@HX40&Q^-OW_FQ<z^vobW(7!!NvpsER@tF4hI9sT{ilO!u?yYNg3LZ6wJ56!zj
zxzTDjQL*FlAphH?o;b)=Omb;`p5S`hO3qm`#+#X}!Z=N9QT^kM&4V$0t8b}rsG+mW
z$QW1!u5GR<wp}m7wM1-}(?0+h+1~J<$o;v65swP{(S9Id#Q<fDRaaA!>V2jiU=9uU
zS6KqAeTTCrP7lac(Ic<^l^JQ`!eVV-hbrGZaxy@RB_hjUU={~1Qx|bX6+0>K=~keY
z&ka`WvSIY21l~LYQ~|%<FkWkF(!q!)B9~M%&Bg-PY;i2k54HGkLcgveqOB_;loCi_
zE)`*6bm452DJS8_n9UmTttG|P-6>u5EtA58KU3AOCIs=50vRuT&aOXz{mw{fD8LoG
zx%<8lLJ+<kbmRu(NI+=~e^wq-6uNh?)QEi6k+BX9(YROEaJS;Fn|q46uP10_i0rL(
z#hLfZGHz(ft;+IPGT!SaH8Coi!30lru-Rl3E<`B500*2~TXCPbrO$hZ7|bF7SC?y~
zksCWZ(O_oqczOLz$~vgqf>`w-rwF5`t;>jXw|wm)rg>33aeKYIZtPFv2W}?K>>%r^
z%n6S$`rc{4qukJ(#kAsV+p$OPV9ndtG&igWxTkW(?-c{Vb+hN7lh)nQAinhe!Yb`m
zw)5vji%Mre6TGnakeXQ&sxV{G?4E2Lml$XY##`GI9JP7SI}wjzHGSM?5oZhaPr5z-
zVef27TIh894A7HrNLMj_fcV1li{XB%f404Z2)0S6ossF21zYZ(bnS8&n85mFy<Ze}
z;`f2`ynMT3-CL`dUO!13JAXER21I#AcMuyZi3m1EhRrdWP#0@_GG@ve=p;IR2dum6
zBg#X1TypmMhS@{MX@_qOp24s=j(sB0iLjO%1VR@*RcLxFl$8N%K(+4a{t|auTixq&
z2eoPcUMV>QPh-E`Qi9BWp)y>aHH}-o?zQezKE*qnjS~BBqrAg+LTIby66-vKSOw%A
zd+?T4*VldPYcIy@*Ns(s;FNFb&n_;01vYHBKS7FUAwQa&*w5G1GbvCXC0Zo?Fo8DS
z=zZ3F;~wZ_aZy~DnJK8iKQb_kZe#RKS+jz8%Vl6Xg5Ke!-fY{LkzDZc824$w{^gtG
zheA%x=+=EIT_ZC5>7;x8psw>z>Ok)oP9Yml1VoOjI5*IO8*Jq`As|~!w{+^f{Q)Ic
zM}(Xv<`u2iubLeS1_2O+8{CPEB2Ey72K7SH5rsne@NOa_ouHa3HE^(V8Zoi4sLhtH
zgMXmKj*xxcR8^q@+s-P-CHY=V?Bn7ucN+*196_;wuQC;{q2}!sRNsf*B)+DtcK1MM
z^w*!XFS2(>{puEYdQ1LVAQmL-m*&(B%lJ{-Xz6YTv&#0~L+WUJs_@Rsug`$8)yhX~
z|IGuU-3}LwGA=ceJ{62O7?6<p;|-SA`{Ujm@)-zCGz{9+bu_%j-PRKF(qU$22d#G$
z^JpNo<LB7VLm6HnVwxp)ye8^1jj?dr$Bs)69pyJ{siV(K1V|}&WHmdt&2k?QzfR%5
z{>KZ4D1oX^{Hn=P(a*B)Fv0F*p-)$S9Zg@>iEs;*@_ew`LD_oR{G+<*K~Hw}cqLJV
zo1ZTy)d3#7p>1>zcLQQ%6jo#OOcbY(rv;J?HphN3<+*r~H3lRw9N00$ak6?%Ur<z;
z3-LaU$xKde86t4!@5|+YBSFuPBJ#nSpmI+*vF`Nm@8iXexw4=@KJyWvmHxf%it{Js
za-!o*#v4@MjBld_#aW}B;?md_>*E@$A|!lJ>zaV<PlW@g^KaG_4scSs@aNA7P1^q#
zSq`T03ANzxiw8p9e2F()4B$Zh5lc)trAZ{YK}=K5lZ1%KeZV!$kT}MaVwMnz83|4M
zo4@|_%P)U=e0&oGLWt&exm?t`yOvT~Yh8PA5g|qf5ad2Iud2|Rl~sEm1F#6RLGp*L
z%+<k-5kiodXuqG1H-G!%%^VXEn?xR`J;o4)19Oy6Td!Tia6(qIuBbzCNXNt7x39i`
zaraG(JCV$k5U<PS>nuTv`R3-X^&i0P{d+$>-KQx|F$BVLIt$Z&o_{k0g8RA_?LJS_
zci;Vx^IWTnD3M=<g#M`!e8rKCiUri#Q`-I0KmDKb6SZ2_^?!`!Of*e-<cZViWa<DA
zq^opy+mL2pL@>>%tQUZqGom9S)B*#W@hMLYsfe1_ZjKz|wA)Pv-ARd2VhAzDm}>1y
zDa%q?tt-$;!sx>|SnKP<<NLR7_tU&P9B-IeB(MY`amOL>lMqtQDNP|};z%5r=;rpt
z&Fx*vlRqt^ZGSh4)@Z{W(X990YOAHzwfCylM-$Y60Rngc6vW69Ndhn{awh;V>xLDa
zm^ddkjvma{wbrGUlS_)ze%>8sLLy`^7~o_#H-yvDet!4x=QsB+j?;@nPC-(TAS`2B
z`f1>D!}I_Q1uK9$=r$PzBmiw#8`GBc{hFq(J6Zs6aDU>Eat5L?d!h!apxss7orr?~
zt5<U)Ay9Mx8m1YJr5U&mw&m%v!s@DpSYjgTpl*Olq!q0!r_0mF{WQCKt1Zpb&F$^Y
z?fey&@z;cdKU~)H<#IY-mbEQayJI(4*4orV46Su{5GK2dBZXniJOcpGHr#4vwHAoP
zEZcbWcz*-imW<?EUs9NdWoj#RJo%Wp_u(Gp+xNyyL=sp6dov|eL_;)k!|^i&=xW~7
zN^5~(<hik}7^{p=d@k6CEfAQd6mO3E+q+|&W2^1-@Ob|C0Hz^u%!$*~IiD?mJi+^d
zXTwgYzyrF^EW%MF^@`x#jll=gR{^c}MpSBBkbfj>HUx+Wn?Ma55Y##l9dB-LZ}0Z|
zqeyh~5OPdYjG36b_Aw+#%pv3$vq&UjH(!^_zyIl9-~93?v*qQ>8#gX%fB)fqZxs>D
zd+Ut|Ly!=7EQ+FnE2HT;cvfp2)9u&1<!EE(7(fltLcn>Jc^8kj``f$YypQIs)e6YM
zB7Y+07GsP-BuFjP-Bpzk$*~y-M&>w8hi||6{=3(|pYnkyA_f=}&@fodHh6AGlJoB6
zi*K~mT9(I$4*(z{De!uF0)!C5f9IP0kK=1r^bVKPvYT)3ZeN*VZ-%Id#^O8*<yZ9F
zC%!QFKuEHgF~#HIuDn`JJD`>NRJsxpvw!riy=stv4!yNPh(I}KVHRc~Vxj$RX26?+
zUe5je)Xq<R2o?G$2|J8ZLpSJMPp5?$YFlIGG=&@!kpP0JUM`nf8#?Ug9U}Ky&reSP
zsOrMl+Vb%D?!)~}OzE&cvIOQpBqNhDa|j{leD$&scTxpFCW$D)0Ci)eZfZ&&(|@6@
zYAda;z1P}%AIv&rM*#>15it-)kt7m`1&EALfdK5J4%LC^g-p9hQFv#dT~GI?^W%v`
zjxTNq$hxoRDzt35V62bKg0PG!dG~JKx|P~W?X~v0xsh(A<AIOsfDN&_M_~~Gb87_@
znF9_8tB>M6F`;$^z!-x%#VJY-Wq&Q!)w}ChNg5CbAhyrUCok&?!(A4U5dg6DJ|xKo
z;0E2blx3bLl$24<OL=@cliPkyDR9V#kSBB?Zs4$-tJ~9WRik%!_i!q;Ue<QGtO(38
z39`8lLzmVXA%+l%324AlwfA0Y8$unVF$_6o!ogg-LU#plVghtx5+(&#Re$VAgyy;t
zSi#LViP$)vF}?uI+cmb8ahO8%G1A~VM)L{5%xmvpiq<0gK8G<#q2{|Oe{*+yeRuft
zFYk0|Zf3|sd_Q<dSBI14rZI3#kyD0{7tJ4AcvEOG=rkC7Lw{k!%o13#w&A7?2r_j3
zh~}uMRa#{(gIN9v<~d{wK7U}4EYofm(!>%S29PG^Jm+Z&AzV#aflx$(gcu|+^Xc^P
z^UweO^z`1Wgh0&Hdt1x8oX@RSW}4=F*zbiX&k5YLt2!R`({abW`kP<wAD>!j+HG_B
zasxyJB2WM`bYr$4lrrx2ao)v{NUfD}S(ft+Q;bnWT5sUCESJZ}hkwiYTw6OFZ`cg2
z_o69EPSdxqe)r9*-yQaM+5iMtt{}Fn6np#6ktyW3+s*r&cfzr(jle8NWOBq-%U4~=
zUrDc|aC7tG)yr?00&J^k2QvUeWEd=qE71kPhnk#_92}4kWIrGKi`Rf~|L}{S@0)fO
zVG?GUa>gO&=mvm6qJIbn8SOh4mK38>2SWl9mSxeks5gT_fZlrJ(AC#c<H8~&ff1Nc
z+|f*X*VY>%PAL;f@77y8*EULUs(`f{(DCWTix)2e4<aHVN{B2`B!r+b#*|aeDW()8
zA_V{tNyNcNPr_h~M~}*U*QUK_TU%ROE49_yx|%yOf&d7h0Dm%)umlN#MUco5Z7YdG
zb-$={^Bv1f#-t$zX0qPx>TafHz3UZ?IBtJqf+_MhcQ?n~?CxvrYi(<3Pv_;dl%@7k
zwc9r1I+Qc&q>O`B0EomSjFnKFQjFbA{i^YE2S9flOUo#Mqrf($IoxK(P2+eiRycr~
z^<DuuiU^Bb+kaeFXFNxDa8$RZSejF5hL`1XX{EUKa@g&5dCnnb3d{tgU0ZiwzRoE5
ze?Hv(X(_F1X|{F~LImtKh)G~wd+z|qrHBM31jKIEw2v7sa%Q9uM2E#Ffuna*b3;a9
zaaTkzH#IdNANP)$n-SACTjjn%&S6_N3DM994ph*nd4Dyt>K<Y2yv9_7sd;sihP7Mm
zR$5<cudT;~hdIBxJKXMbX|{AjqNX|o3o#3jGm{823wgh8{iqvO!&R{Z7?6Kn(*Y6D
z5WqIQ8rTY$nH$=Ge>CC3EJ*A~JQR#Xqm*S1%q$W^oJ10$APSK)hm>;8Q%Z>kgp9YF
zMI^9@yMH}Cef;_7Kd$8jpzcu0#a3Hu0r$W{jCENd#<-jEl(nl{QLxs`ySL`0^iq54
z+7<1}!v_FzG&cf}NYhU8L@CoW$2^4)*{yAr9e{DGo0^$c_fksNPC_ia;r}S^=HrXI
zZ(hE9J>@-us5*_}6%l@tY+T!$BYh%rPP^TFn182zsi)eMxo!OGue_4K${ZYXo^OtK
zcegKv1HhOq8lz@5^857wHtV`=KnjtFOjXT7$cMunF{$^`>t$V-Rgz3;+D-E&gsKH#
z_!$LeAtdX=I}O~G8FJ<zfvCEB>we8k8+bt7d)JXo37JU<+094A*xZxE-88F$^`c#C
zEq|>K&;lYxbG)pl)>Z&#ksu+2!BYup8sew`#qkVB3WO|@n1VZ+feivUz%bV8T2+ho
z(t2&J)!v%wRtDjK7yuczITB$BNIYoAj%Z_H4%Xd@Vt3kbWVBhD5i=30YFW#&tV^k*
zQDyF6DuE9<eE;h1=CJd&thFwsEu}u4E`R4sTe_EKHb^=|fM|e<oBaPecIU&N8jw;-
zR#tVp)@%^QmyW<fY05y@HWa}X-3IPHtfO$qpSt$iDq;i~tU&-%1GkaPhG!FqBO-@x
za9QfIwsl>4Ev`z02Ci;tnvn%b3}Mvkeyhn?fYPk1HFb3yKBuj9X2uPkWp2Lsntzgu
z2qJXTtLg>-NeDv-UCrFahGu+DVG2a5U|`y{yOKM&H#2lVL<tOo!KLcz+v_2)FaeIn
zrfN5)!GWl1;@Z#y!KcdH)*%8i2Q_cqhTW*Srx<PzyPMs7e=3(+)x32bmq$!2kq$Te
zkYWqsgL<k8W_D%CUzIZB5x(U{B7Y(#6hb3cau7f@W7zP7AiYU%%oUkI5s8eMd@w;b
z3lWxxB-cc5!ZL*OF{K#e*yKwHEW@9L)%xl5@b>**-O3=mpD(AOofMW31kvBWf7i9A
z97M(nwYlNtye#L|8=THdslBVKfnCjl(1C#5l?ig<n`2HBB5BSNqlmEf&VPj4N_6Y%
z4r^KKsuCon7z6j_?wCTpy?OEK)$8N&1yN{Skx=})n%iP5+kESixtn>2X`XiTwC{a&
zYfXLAK76I57{6-I+gce-xG#ENpN{nJI|J8;lz;y7_7`8oCf3<B!JmHCujGJFuK;5$
z>7QQ(0RH2{->Mhew9=Oc_J4Vj{)=D!)r(yFjD$q6y~PI9yuLDCL581SbacJ8+woWD
zVB=Q8e|)s~#hJrTjg7zkm+jKKKKDN*(ed%d)7IfSXX9V|3X1R`dyJCo@PT#z`xWr^
zR{rUPe=>x)e(+zN{a>B!Prv>*U-h*F{G}S}*Wc_HKk`3}@w#{|OMm@c8!G4N|LrCI
z7vp@n{OXFlUX6eAi9R3EXXY+k&&_qn;D7Wv|Mp)$^-sR|-xsI)%S#EjANb<7{MGj~
zj)BG^VVnGiu|V6*K0m+d_4Fd()&z#JlNc9-Ut3~-`!=8M^)GJKzdw>sH~QyF#jj7v
z0EH8ei}v$tZY}>I4}be2#;VGFv)TIFxBKF3;q~VIlwq!S|EK%uZ-4dIC-8d3;&pyu
z%gSHgsh>anr`zZAN837MpYz7`qWKT6iq9w1Kk4#*{ZAp}|Lzh0{YUtXikr{z9MAC_
z&+-3}_&NmIb3Dg$JjZkV|1O>r{c}9Wb3DgCcswWi=Xj3icxR4(@OV!2&+#13@f`o)
z@to+N<2jz=IsU=pInh7Ib3Dg${Da4HqJNI(c#h}z2ao4O{~XWp9MAC&9?yyXIiBM=
op5q@po)i6ZJjZi9$3J-dH*{<dJX?lfn*aa+07*qoM6N<$g2G35i2wiq

delta 320177
zcmW(+c|26#|Gt(OQD*QFvW~G8vK7h}BRgfwz8fKX)}cEz7?Ca6my)GuRQ7G`Wg8?J
zS!QfmN`~wae!jo+=e@7jJ?Fg7^M1~IdF}+BBkLtcswRkj=dNe1_;04d3tzXdT#k+?
z^T?h*XQ+mR!xG^1Nm*Qm4s5u?L+)a>Q&=?8^+#DT$Z`*N9>=~YlFM~hR@_VB0mJb>
ztqgGl&nJ`3BRca)Zg*dDlr(z7!eV6U$hCm#`JqW+vGshR<Ja=g4Ks_%3evX9u3e=p
ze2&FK3GC(`-i&2F94YMJ9`=uu&!7MJDd(m7*;D_xYySC6hi6Y%JPv3XXQTGXr;|Yk
z?-S9~rj^)hk@K_vhGu61dihKu7uZLN)K7nVs2|Di1@5n?pZ*m%`rg}iSlNCUcj4sl
z{pn#<3zet$kJ)vlZf>>NKio}o5s{lu1X_>xz-ic@wKnS7$+6$a!P2PKNsZX1BadI?
z;^qClJ%{68_h%#4@S$gudk8Pl2*N4H)X<_^p;WW$joZVwHrkG;@)fGxmt@q0FGcPt
zEoSMY$WIKF_m~wOemkPJYzUnG5pdkjo%lD{{%?>RI(yu<PuNi12xo<#9Hg8cC`NXJ
zivPB*oyq|-XWI)U2Tw#0PIBM>&22xPZa;@9S6%l6Hj_hJ7niPqe=q2qOn}D|EoZY5
zHlj8JPUZ#R3yBUp#kZ)9i^;dB2a`p|llKX~EG|jg9V}csS!h2<YzHv7upw`U?Z#eJ
z-`ukIhZXPtXO&8Ktol)nI{KFxSagULQ<<-OA0-8EJ6`u=#5A<+;!VN_%=XUqf6#Tw
zp)K<7DN(v$zMaEYSEl{+s2zN!OF2E+7hcGl{fBCV{TFp7275VevWr7~lTLl}orkKo
zpWlXU+mfvpzNqJ%B4T*<m^>S`^}OicL9hDBeQ@`_`cU!k!IC-<5N85LpV`R|1W%hr
zqNr!D|LmWc2-uWY-<Agt%fa&i8jW^5S#o^5LD+9QB5yEYHV%?E4w{ONnx_5Qj}6*>
z7{C`^3ml6Bt#jwDg&oYt`zdW{g6e<R)&DSpn~Z>eFJI`tP1e)R^@iDtfa9P49RCd5
zZ({@6PS)Dh;1uxha42m@19;Rh8#JQ2wtINC4qv3**`^mhC7MKheg61&@8jQHvCP2c
zTKmZw^DA&$5!_bYpQ&kYKQ)@&F*+67|HX2u?XMN{moR%oh}DHJL~Wo2fYzTze$7Jv
zGft~ej+9}5zyIdhyn!DhW)DdX4kPCjfA7w-f>t21ZO5Q}XZJmL{3-Dg&D<Gi+UEaf
zozX>N@U}$2N<k>ws8Vvn?Bk<&^?&ng;O@UOeDiIDF#`XFH~w7&X28?qOV=DCCt4`u
zHguR4YBMunSvfA%FtatOvU+rOVBl)}<JHa0%@y!u1)RV1cs^$&a=C@FTXnW+5TwI+
zs4Q}CJl=0<UrlUV)e<;_!p@Ky1^Ay?Ty3+1uKj=OOV4V><~SZrP0)4}vmf`R{OjWj
zo%!>>Fa8gyQK!fI*NFacr)rHtoik9ylHzkkae{ujD1-M|Z9Apvf1j)W%>w^sjf4*e
zscogDM6QHB7PDK5xx5zv9!<|yRrrMu+7|A&u3YH;(6=_i;PG!udCu?)C>=H&R9F)k
zcCuY`x^SuepkO0%rRZOOQAx)CJX3wYX@oF2%D>;(OQSK&4NzORhnH$9Xf-633ePxV
zpMu_KUXw=JO`VRZAF~50$3IONF&=8$IfGe=w7V}H|Gt=rI|lo7)V9tUHhMG)oxxkN
z=fc|uo#jG(m(wS==9x^wXaUIqqevI-<qPQJZS_-m;LKJd2m9;?J7>;5LgS%!IC75I
z8u>rmoq-M9jArCNz|S7y%T<;n{KC2~SN**EN8n!IzpL$ASO2#lG-bOmCF-~9#DYy?
z!O<BiK~`Whnl;kt?^gY^@72;bUda6X{K$Wc0wAfTaCcSAKW^mOOsW%STIkZie?Zt~
zCu7z%KV9u2n4$rS`O?gXi3M$I{^Nb{biZgUP@VhCa&P@bR|Wo~oS{r@-P7b~RN#1&
z?3)$DALk5$MYuB0!!Pk_JDseLH}55%{z(BrUVuv2r~L1n1ddmA&>O&}`svE4WNGXC
z1;2y0MTc({2|tYt2K|mne)zh_8~cwpZc+XuUpuP4hTf|wTBoLL{q#6!?Y_XYoL~|O
z2WHOP=slXe=J4m~eMzbRUuuzRj{W}w{@LuoGhp-S{nmwNXAHaMczUP<wuhWq(&U-^
z?bA2^bC9i;_M?{n_c3(a3VJ#Gaqlh4$A*S@-1~^}O0~U@DTH+&KXk@o`@(<k!w$!a
zP7=>puHvxH@u=>MKxh9)HD@ep-~Q)v`zRV5w54nnUm`s4{2Ozrbx!IGW-YVF)U4JY
zA|{Wwo}Zp^FsM|z^JXG6Zbbgf%7rI8<?X*L-Z!6~9D(Z-kJmq4d;I6?R-E|`_n*n<
zle>2Q-3}vXN9eHMJYnAyxKw^dQSIx~ZR;OxPq^EGe0Opr{{o%BLC2uOUUkFSO0%vx
zp0sTosR3cLN0j{{)wlAIh4byK{~hmJ|19scmf4d9dyZyLPqGNV-HUdX&KTfyHRWV=
zl;!AKxtayi;bbcdoCuAYID0YsZ&?v^Zn-qFX*Jg__+P=FL08748t<p-ejg_vo;f_Y
zCpo<N_t~kM`Od6I=_$WQ%*omH+Rfe85x`!Rhf`8(O~T=Hkysq^evkzWDo6`t&eG8W
zphK1`*P!1V`^XDGqXV;MQ*bd36DAeo??YSO^l?zJQH7p{cE0&N?9U_YQl5@zG!6#a
z?4!#HFkac~Q!%+EJ(MY1RW$zf*HmM9RF(sb=A+?-ste)k+rlHa3O4ILPoQ>n%OM!|
ze%t|*Su$W3vfup44lS<s^>`}oOI^Sscz6#0x>TX>Z?rUz`yUHV)UO;iY;HhdVZofx
z@>@-<u8sA}-ir(XJ^UOd2dxoir{37o($u@#1Pu!=JYc#Ce4m{?+VNa0x(s&-3S|3e
zxDGQ#AQ5(NK%x5B<*F(wtm~Q5@$VFW#>ZX!H1Mu26XxEYpUJ9Q@la18)mE@DnSVYu
z>_@D9URT}8{qoQ$ruWC#u?hd17XX~M@F!tm`>1k*5pT10s+3xnR@i_`QQy*p6nB+y
zWo7MdU(b8BMRxj5Vd6o*9rJ`u=k8$p$>_xO_Qz899mjrVr933o`6S?C*WOR#*YYB}
zYTmagwC=U8iHv-es;b@=OUMOHzW=Qp>@Zts_APo38@6~l;Qp^<GLmNodj|}RJlJ&%
zJUTh~MY6r<+p_y_aieu{;S=+iLP@$36Le^sMWd2H2wL&u?qy0?nQG3}`8I_p9rtqg
znokCL+)(yXS@*`5oBmp$T;|6kRPmyr=+(~ni>$1@o-@8T@FZtms32%awlP+?Sj}@z
zlS!3{T7Q{-CU38N+>0wNW6Zaq5?@DInNlis=4xmOzrRA_VUwyp2f1H4u3!W}`H@N&
zT{j4%X7lFUN#6$cWoNE3PcH(RUatzDckBF$^ISgdD^hp^d3c(9H_$jin@H?Zp*2zF
zHXD%af@?J9ga%cPfeKC5?+N2~FbN2n26EMUz0C??x6$Um{$dQn9B!OQ2lXrr^(nOV
zAzT%@Ai}Q6DWOM<@Uy7X(?4Qd*<&3R_Fy&-+SiDeS=(lTP5tU-RTJ-8J0?u~wi{X5
zjCH?Wbf=(3hm`scoz6X2wns~q%CG|^8AL*ZF%CsSUrPsLvuS^!@=T?>=e&A|dIK^*
z4dVkv@1VTUZI$=KWLflI@U;E(xI;_qzO<i7@(ryk3Fk0s|5DDl^0aMZmI`*z&_+WH
zj2C6zmH(OP%+0azQOx6p_aB`soK8;Lf|(irCS6-5Bacpx+mn-%uQ{qk9(<keN4>6)
z`18qj#1U-u_n&Seq`3~XDqdLIy_uDSB-IUXht#M}k)B(ppzUr_)FQ(oe-5O*#mCM6
z>+JNpPNR80ct3xeLZK|4_S&P5(1^&T%@0kN@A<=*)5BE1_^fPw3i?_9B*Os5^)d_5
z#R3ris_d_aR2j9zG0<b0-g+A6W1sga7@Kj&%YlL##^K^xS0t9`f!<!7x*>jPWk%%2
z`oRx3Tkrel05hGQ41<1=Abq4(CXvV@CKlfkZicR`;K!wZ9;bNufj0P*<Yezk>Tu6|
z^8C@5mu@;l7EUzusfx0#%L^J7XQStdGfYRkC_zC}wb__y;%qDH8tTopZ}$ZLfTcsg
ztPE)Xx8?8~p_*5kWK#3?!sF*Pdyw<VxJ8+#x<D-{VuQ*6V6%~*ojF|KnjPyQ=)l0h
z$jBs!a=*ee!)&DfuyQ%@pdva|UoTuqdD3Uavc(|x+sj=b!+w3*9v>vUkYT_bx45`C
z^@|cQGh3wpin5aKekJ_+MZ<5)f49KWK|-Csa2NBX;gm5PRIFF5C7e3t&hyo4?&CKb
zMKQ6w=g<8jMlYN4s~hOIF+D3xrAB;R+-N+m?<GH8>x@tVeNUah>s9*v`d6%xUOrTz
z^z!izB7V&-6feORYOxIn+87&zIzDn+sN%UTaVf3DK6E?sQSIaqp4X3vCxJYDb#Hm-
zs~O?+f;9R-_H8p0!~7DC+lbpX&%BWe7MKV4SX*NkFREo|%gqV@5ecb{ThKUX4rdmb
zAPgXQKUm(ZAXJ_f*;p^yjiD*K(!>99d2sYHf7vqhA`B%`+S<$Y;ls1s@GyWL)56$6
zX}cS?aRCBPf=2iOm?Y0Wa23iQ<j470d1C_~Pn3Yk%e|!GS*Y^)rs^)lnSuS4vt-_j
z)gB8?(nY4rvB|PA<l3xHeX}xU6G~r2V0&Z?wIr72oKENFW<nH4e1fVcs}Y>e3xc9T
zfI`sqb^A2k7^bxBH1Vt?-IAxc%JG}THQ(f(r!-St`9c`~LJv1E0CRK1IxOEZ|FeA^
zBkOx!e!tM5eT*}&r8`SvclY9Yh^@DMZGQgOTAtpIf=GPJkX}hX6Fcn%E$+`hWxp)b
z002!aot|pu{h#wx4SMDNyQVNn?E$410laaV$U4jGbZ`XnjvNl67%H!5YsjfdPY*{F
z=a>HIF-$D$g_nT85*e1Am+{@cR54qw^TyDH<nQAdz)W%Qd;G>SWn4-Sqa>1^@dSG>
zTN{Z~l9hFK0v?&+>nm$=_=|Xld~WP(Z$q$xEM@*~9V!Tf!?)YlgXwxSH0KeeZjYR*
zZN^<b{)ufWQF$gTR8Pu*h(iQLU?@!lL{M|t@T-$A9=xh#mYVt!fI}P6SjQU8DsFCM
z4>9q-Bp?wPhnczw7fQ>3f8-*s!T|B~91Wacm!ZAStqo48@7+08rZx|ui1R?anLH0v
zq|})QhJjqF?47aQ;7`WUK(O=kC|n0pPll-+;ciu1w@fp29Z=z=#eL_i;w`A9^vafE
zmBIf?eqa)~aW>wB>kyCSrW-CUsOyRTE^hSR_`P2MSE|S?htO=&OOJ@Jr@JLB)Yiq}
zTfTV=usGANC!NhKB3Y8~3a?+(R0CO^4I_*33>t4d)CpzH2Gd6RNI`Qx(WuJAl#TNz
zD(;52wN6sA`-Dsf($jTP5eNoE5Hq7F$bi(5(LvF^1f;_aC6%)*y=#&Qo@+fBRyX%a
zXvK9rTYz-8Oj4r#4ZNtlcU`B8S(tc5OMLv>I5al?R#>BTp0@-YyuRu4+xK*2Vnp6_
zK2qP<ovYbv&B(&%ZI|GqQ2ATCFOZG5kTM!FITCZ4MKnqLqgr&rFKR<iG-S9yx~%`;
zH`zLehg4`e|9dlc_q^tV%DLtFia2UgpQNs2AM{N{3t6SX@-q`Op^piZpnRdylP7N6
z%l@-xP2LJA1aDk^IC$4Kbe}S0XFEZfT9|is4!i`{OTcC0a>~cQ*-Vdfv!OLu8U2BB
z|8-+1&G!fcxmRwA;KAk(#|?Gxrpna#B-H<Uvd9t{6&e;678xivQUBXY`e)`gqoLA9
zeI7YZgVFItI#)PP?{UykO6)A^HnhTao#3e$iqXJIs&H_}WpIlm{4k9R;9?Zr((vNT
z*3Fjclk6J$#_4(mfM#<uVK6ymo_6^1d^Y&OLX##f>j#`a<0{v5fd`mi+Cfvpt>I3K
zm3cN*8{$pol}Sy-aNWfva6L)IC17z(OcwY${$QBHtEuE0IzjkfpW)(?5=*a?grG1?
z2~w9(nLD*`6aE8nE+I&oCBG*vmq(DP1}c_G?4QI4iQEQaoDlmse-tfJtQh_yd5(wX
zhXw}(dlo-1l=HWO?sS>$WHKxuz=mzXw19sZAF83Rypo^5MfWcFJ{@4Jhr<eX(RC5M
z@QH2O$Zi<~%-NxE0f15m<zlz2lm>q?t;Yd#{M2PpNjMu#73sB>5<Bfn831;db~xIN
zzwU+wBcNebTRhZXABva4m&tK#aDL1GnOpfFKyEdp0c75^Gaak)MfM@dwZpfDa}wfA
zeV;D-X?_v@dK%Vck7j&sm|w2qXVSXQycjOQ!bRp`3xq!s6!M48%em<#D61cQ(ITuh
z9Y5(wHyu#!?e7n1Ag_m<#bt}n$^EEX#+9#<7G++jJZ#7-Q+-jrnfl{LV#y;iufT5A
zSgM{C8)%@(NWPA*Ggfx}tc+SG8xBb>_4F{~1py&8D7$;+xDs1UQ+yMtj35RJwy!A8
zQ7+F2^4<Q;ua};;QQrb_Y>g7riSKUI@9rRSs`=Hks{WgcsEZ*4!{LCYW=~HSJfmlZ
zmNvnZ4e54Av=(w>?d30$%`7!UIV(fX{f-cb@N|}pX>H8mrA=)tDdzO?r4j7uu9yiN
zl$sA&+A1Y4H`bpo)6BNcHcdAaclNbkvr}xbSA37JmJy;=qGMUiHVxcg?=Kq;3M<JN
zkomp7u~F2rH1Oi8PW=7kloZ@c&uQOKJ^f)rUD(|G@@Jf$l9=_3Sw;^Gib=gxb^!?@
zT%3_zT)^h4+mRnD|LlH8zgE`r4Kx>Gc~@zi92xm2^6$Wlp^%9dS^&->QluX0|8t(v
zB$>5g+NeJ#HHZFQpmFzIHMGZ)kr3@N>BhEi&9w9sFMB3&F||x+CBG{|@qUnoi*c9X
z48VjNyDL|16|5+48*?FX#4=qsodrw_nwp)yrKH5w+X=Vvu2@DT4n(tHRypfFp^}|N
z1y`rY^&;}}=(%@O=6q?{3EVl+G<|8DBSG|Gn9E@`k0S%Cl_zT{6KJ%%IC7NC(ZzJ;
zlegdG<Q^4MLe}T+*`@$4(#n@H&G*e{Gt38MpavViE=nfQXx5F?1RM3NfdO#MakY(l
zf@SDc5k?n5eQrT{*N`!dRg!n5$WlH%z?}f?7@C1%aNY&zHDaV!Bb;q?;eAM`8$Hqm
z<wEa29?Oekg7U`XD1A>dDFt3?eGd#PsqLAToLSs~=#_G4vb?J!g;$Lm??wSE8gSS>
z`E{~a;u20faE(xTGYI65{;G5#2){ie3{5~Pt(fb}HrQOJ#aWHV6Y*kqib&o*__~`T
z+aZ+TwbF7*1U6AdKAw>VA<EV?X)}?>w&U;b>t~ZCCkpLbGu0oie3hXI5G-4H>>v6}
zT({Yf>`8!}Aq_iG$;3pi3+0M4OsZ=L5GTrIysE1p+pOb3Z)sO<Il+SF8k+2!oIa(j
zG@MnVCu7pMSJ9WQygF;u++Pa}3Koi&d!th$)~A|Vghw^?rs{_O1Bm7bm{VF$7xoKH
zg!df|8V=cuuKJ$hPjc!kL${^75*}Q?NaGt^j>qrqeeG<6cvg}QTggkNPfIe=bA(_k
z8G*UpWNHW)^ljHG_w#7&K1KLlUVO!_FJS@_%C5JlbfGD%W8Rm!!n(&?=x$)Y@{%bK
zP;~(k=gei`>EmE^OOTe<T9?auK#9qt#-G($Lb>u6D#)d5Oo3KOJIVC3t+e6Vu1u0{
zMqp0a^@Jb!SP5C#;s*Yt)imYUCztQ0_q=5Z7muX_BfRmJ+O(n0Hx%KV&Y>f3Gcvjh
zpybxvkvak2L#_MuI$}OmC<c0Sq@~s*FQ0$_pJQ8~xwoytOhz^nI?tjdr#U=R?uU@r
zZ3GJVa{11}2_2>*!5YVu1E^GgIZ@z!S5DYf$S5x0f?X0a-~cK}<A+<69H?{Q%BQF#
zU4D>aM0Yz~W~?@sbc7!B((WR7E8d61e`?XfVrk|_$3trb_->hHCOl^`?7$k9prYq!
zB<ZpI%f1Xa4PrI@Jl%}lg?C=w&QE1+S+!`*A?M^3<pezd7>YJfW`o?_MFSm>x!A4r
zjE)upgJ}V`A%9Z!;Q@>&XM{Ils4iFpEF;Ow=k7L*lPX(oJr@=nE%b*om9_mOTt6GP
zSu`0qg`8pV-Z7aH9JY7-!L`Pg3OQMJlf_2lmF_5}Lg`m(t+_1F0G@5f;(THk@k1DM
zV0u1rW(dxkih;zA^C>_zhQ8SkMc;#_=?1#Y<7Bo!-4R~+w0c%!;ex^qTfm`l;xh<M
zLbj<f9k!`~WN0)V+QNYGWdh{1=mQ3FzR(bZ1A?J7j5Zy1mznWp7TgAcx#XGH3yLDE
z0lWqh3G~+^c)osU&|jQ`C8l&4b<@$P-W~GF>kTN_)<86nBZOuq#+vJvsaV;nvd+Po
zD?vPQ&bcl%IcRb%ikfKv4!f_~4`mL)IZ-H>gu9jbJ*0=Rd-hDOr_5!xJIMQ55V%do
zAVhKegGdRZrW<VPsDCsK4Rk~hq?OZjHGnjl2i`P*F4tZC!q1-9@x<VUMM_>Ce1M|h
z8yYlm@j<Kvt9#$i&8;QMt?Vbt<#wi1<2kVg5<Pa_W)=!H>7yVYag<KAPc!v=`S1|g
z$Rc|%wKt%)Os6;AhhpUZAV?e28AWJt9O&qobDj_Ub>|!{6s60AjdPYQGnIa=TuaL1
zfiLG1EEU)AmUlwCAy>G#Bn=JEa<>xG&wvK}eAcHONBB6VYr%`KhwS3uRndh3zkJnR
z--JPeyrIx&a1Id9h-2Y*qTnlgQKYNAyITQdBI&;wA(*JB=!5OZ$g^~>`lpvc@$F&L
zY8hb+9X-2;p0z3JjwbzDRGUO`J5w%WpLN;C-|uz4ki6Yq9a|_{nV>7jWmpQ?gruts
z;O7Bm8Ak;tQKsTgW4<sN2){E`7hxSZyr^(tQCtMf%Zu0n&auE*5LRvqNDTKkRPW`2
z&$aDW#*_pVkUFB*jFXm)^=Mt@wf6mA3eszKe31EYG5X@LH537DxlBth)oIJBVBx+j
z3GW%ci%zln;gDMvz;l053H!GT`Ej+8lLgnP<RCs%Nusn=IN$Yo!SyX-oMbN?BM437
z(&GYiR9L#mEJS))q4R=_=bhkeX}G)FRRfY%xHS08kiE~x(O6v05Fyxm84mB6PeH8C
z1=my9q=ZHFXn}O3sY@EL>kqMO?X<7kr7mMvy}X*UzYg{f!gXsK-C+J70k~innJA0-
zLTEq*Q`<_602(Y#vOlP61|1p-$UOsX_+DKVe)S?<DM*M!f!+>`fs9ko-u61&mX#7&
zZu9v4-hp7c@^DSx<)G;*w^&2F0XhM2?#Jo_RhI1#cHc3D^>sdAVQFw%>XNT=DAjIw
z5r0X#)@EhBmHuIlCl^ET^BkKET!&%r{g7>6>MwZl8Mb=s>PZ!7ma)_zelV`u;l0_w
zyxPcg5aZc6;~=Hhz_<HcX8eIMAHo}Q;_|UfT4>h&PmWJufK%Dvt@cMh*?{E5MjnPM
z*4bV@^E_qSzRbFCiG-Iel}!~gA}~l%==GJz4-We_Gyz<3ES(ko?{JcB64#k6*aF7$
z#}tyXq#Y@~ioT-oyv0Ru{ZD|gUN-C=&T<h+OmO!d?_v*d<`0xxcJhM8zHN@>hM3I3
z94d!fnn=x^i__OApJ6OY7ql`A=_@^nP0;i7>-F1u1>W~H;q%U1pI;%r`%;wYVIsyu
zzr%t|Z7;z@2lV0skR{}hENE{=ncSl>-wOWV!n(@JKgq9-08=o|`GtP&7eN*}niO=y
zMHFhjCqp~Ve5RY1w<NdlTVwU8{e|m@Ahw>(brkwhVD;of+qDlpv$Wa|>I$VPkJnyw
z8w{R8J8&Fnkb8pfd=%GZuQGP_1kTSnX||?&q61l|Xjf~KF|WESr7|^8k>z}5w)k7J
z<@xaewDin|>6YN4%#YdvTjW7?1tOg5v*%p;?e9=j-6VldNtWIvUzCp#bB(dAGo;*}
zrs>_jB}ow#gn5ylp4u#xtFQQlT1F*lAuzP>yd8ftvE_b_7FiD&@9OyHRiWUFA;tYV
zr?isPbDJe8-&8_N-x^}CDl6vVZ0<i%H8k@mflH|AoB}8x#MmWRQ9(gZcsa*i#3IUv
zAECBlb%68CdIJ(m8C?D2;uu=4R3!Xx)VH@sBrAX`!f&eil_otq%q4r5v`l`-Uz*{W
z-4QjZA>$#03TW_;hbSgi-XC!o<cJp;PlEt98+p&!gSr7B2UGq8oHnQ9Xu5hu1zZm*
zd>2UsRcl|z+A0SB_`px)<VV5gh}DvIraZm3Ti!|tX=3-<M&nkyqSNE-;_POUxDE7C
z&C|Z#7C@$XT!2Wz5Z}*b<i3$Zw8k6i%Qo!yZx0nm?8j-F=}L*%w=DMK#QSS$IWt4=
znry=q{a76%I=+>ACRtVf><^i4Da#0wUhGH$Cuz<2QxvyHNAI+prI36fZ@f_TT;5dK
zjb9-@Ew>c##fYst98+XEr6oCpsEYE<9UP+C0YCF@pP%1A*GF6#fXpAP_L0jx6~;)m
z_@L~MFQL-?(6Zx$Re!E{z@!U*$Iqu_Qu=l5<lX@emmp%J7`ic_%<dEhNe@+3z6nwu
z8bh7@>hzVHa=hY!ARUbV1j*V(chX{|EUHQ@UdP!oE`X4m{e{dj)5*6u@aF|WWZS=b
zGW(YTr+=h^h%EpJ0(!{X<{n*dGXNcWIB(qTp@!`w$su#{IANcMs$Bl+$ESq=j9`ys
z{PfW1G7{HmXGN;qZF;gTa%qhZoMHNykq#fH6n`$0sv68mJ*={}P!E;)LC+<LFf>$I
zlJgFg$9?A)So3W=?(zY!;u&SA=C@yNh6FbNq2m-N@R3yYOPTehv76AN=~(8K5kA4M
zDib!-spo}7<A{`Z`_zalY^RISkh2iN)P-k{k;02oR8>-HT>94(#JB>MkJCfBqG1MY
z67Wnp`j<bxM%-r=ML;w?Sll>-+#sL-I7O{}vZd=IHV`I6w61n#wMuBH4=J?gS_p96
zo-XhW<w;HBc>;HC@P%P+)BmE_3Ku)`1`QiV+8D6CB4w4>If+7Bd2Z-=I%y=E1>tqo
ztk&`62~d+{lW!MrXZ24843CnFAT&Gr+AI3&w}K;_u%Y{2D6T$3y}Q{o;`48>oqL;+
zj*xv`Uk|;f;NI0Erwe~6@>+BT?M>JR?$%hvFf1fN5C|t4Y%Rpshg@r>0qd7!apIr<
zg{lj8>zAmr93l>NbT`)n8iK`={X>~V+*0f{^~~>YnzZPJ^gw-yODr3N5P4^Zk{duS
z4<KoIXE=$WuEk3}_5*~q!5>nWd?!YuuY-e1g}{7$QOczm+2>LNJ6wc?V6q-a{yQ$i
zr6r!%Gv)MLm_4E-uOF5!KgoXXhLB6CD!`fzGYCpBvnLR@dHK7Q9<5K4^+pW=!v0Oa
zkZE>{*j1nmL`|>P?-rUx2k0Bhm1$h*UiA$wGnISj+E7_xZ($^*$20f=vH4KZ7VcIy
zo*ipC!Aaxwit4qTRt+(T5oy+Zh@(0{hYb7E$~I-T*F%`0SPTbYf^-E5xyK1}hD`XD
zJ%ywzm7B+s?a*QfHvngOS!gL=Ysg&6<~GN?j4OyxEDCM8yMoYWgErj~tk%Wm<yJJe
zG_K^UdW3FN5J=p5N*9#n7eX40l$+Q^s1+2L#Ps;DT#^mvK+m!UR)UG9IW|{b2>P+4
ztJpWFVCX*~)6AvXJ1oKv@Ow=$O~L>DXd>aPj4C0Iu4@i7(JRSRT(<Z{2(26YX3D&i
z9S5e%zYtnU&z4EdZfRyR+gQeXt^seal8nV}sZ$=v+kEFDM=0*#1nVGfX>P9)X@l2d
zZAq2mqoaR9Zsp~2=qKd})#UB_dU2+tCGrFDD(?&i=FqYAu7P9Yl#`|>Pj({+4)29$
zjO>r#Ca?f5hW@7P^^lqhrh9O4O*6gxK~N^sSu(tDEeL&y+W3b5ENQ>d()2c0Ip6X}
zK?K1#hg#bOp?jdX)~`1`{F(VvTpi`xbwK7jvBgF@Uq8nZ{ek0Eldbv%UO>)Mr$gi%
zE+Nhwk$Ry^IPqQC_}}B8N>5ElZMLn2;@w1lK~C6Rd6`gFnP7`GYJL8l<nQE`CXiiD
z76vfLvddwwvDx9cVSMH-!RNM-8HiFuQf*1I_JPYt_H-adY~Cu;p@IXVYmlCj*W#Uh
zPXpC2l8Dfhxi53B!9R$O?n_lUJWmNe!@@Kw46pT;)aQV+ogclcuN5dyyYp$)n&T7$
z@ej4i?~{-^{Z@saD3i*HhGsT`=fki%2X7e`NN8p=Jqc}We2(V<o9Q2gr2rm8z{KRT
z+QrEsVa#MFO|;n+ct2l5%Z)Bfl5V8gqsX8b%1{u0If0G$qg#-y$(X*<&@>+qp<y=t
zFqB@2W%17+n9y(h{Y9Ay()!*5-t;TE^^1#C{;6Qa4W41jp%;NouTLC&c!YN6^}D+G
z>x7ZN?KC<*G-%DpC|F#?z@-b}Ng2YATds9dt}yS`^Tj7Q;U`~o3y%-4uns81p&zMU
z_)Ejm4Gs8ZDpMC;C5FNe2Nm0yX2%|-i`WNhme_c2@WG5On+}K=spZ-X>$%Zl94`9<
zm=>)t$%}29fTOI`TM3X$_JXkB4yi$G`Ya&2B>j&3@O7SjLCymt@8E|*%TksgJgS~P
z7$+gi>aHtsZR%ROpeQZ(K8ci@%U8t9s;8e`pJ$DO*{)wy$V>|q(vu9&F6(~X9pBX;
z)=RZfVRV7s(HM_2v@<0SWqhZr;&o)gHPz)a2;GS>H`P$O{S{Qhi|X0sk*4x`_3*o|
z7~>I(sTWCOUf0XgJ0yj0B`S!n|L7kykmkb}S^4%DrG2FNhKz3Wb+8y+#y)Qm78KLN
zLFQq`Ktq1Zr_s}&MW^mHljpytZfA$-TCpj;RNl#BF(h+fx0RCDh?Pmf#Q7V^XYE0P
zP!@toFke@w3yf6u)?e1-z;xdo`n|Pepk;hCmTIQoZBAMn@9gZ9Iyzcmy(pc}+%a`j
zYLKP+PeGOio=Pj5nUV9!D&dpQK;7Z|pMP^&I+df^%~NH2_85cVrzOinTB*@&fwYhA
zIAIxTnm)<`))NkO!E`D4fi@c=xBE48&s0q-7cZVktiT*yv_M=ZF~U2zN;7>xiREpb
zC3=JK`en=wpG0%{&_tOI06FImFtZZH5exiPocybRTcTW{l^nD1QwUINW4ng>Y<N~|
zPLPojLD00U50C$<5_K@WK|>tPr>!m#d4bBmL;o8;gpbLS(d!3ReUo%0kr`Ia74a+i
zX8I8DCH`-qE<4Ow|E#?H$sry`Ib^$0L&>5=-P{d^vBl^hzunweHd-GSqV2oDog*VC
zse;RfbO-y6tD0URve4k`WS&5PD_WH4Sp(&$o!40r)8vTxNN;&|PemL39{W<!aZ0Uy
z<L*BhPE@Oe-;xm|A|>qLl>wsRhb#>?_AMBG@Z!_zpmNLG<|bQx>udnN!T}?M%HvyX
z)><1V?}J(*|EBBeOQbYGYC^^+CdtXxZuw<r(yurCVtMv!*Ts}T-{#nhAuX1n_&V$0
zx%B4<QO>hS9R49>eMp~%-kh}nJ)Gbk@J$DIN4m<ZERjvJOL>L#0@vOB&Q(b`1I)+v
z)0wep3Tm$7hp#U^M2%|NlZ8dNdplJPpW&1|00@wmNB1K?L7BLg?%io9iC5GB?%?s2
zS9T2^<s=AZP6?dHEDrUjJAKF%4Kk~_abIv}lXMGUGaG2yqf&8YPUktRc#G$LFGD%z
zIDrgAxn1fxK3cLx>;qLJ`l>0guChd>q)U(Y-9DvugaE@pv+~`=RsVG!sb8dutQ;Q`
z^4ZyX7ETZulEXiec^V5=77McY3VzogZ8tU)X3Iind|Esq*bZu7^^^GI9Y=w}_q;=d
z-i2DG8q2GVCX0{!{ZCF9r5i{@bF5`0A&3!84Ou!^U0v;=jY}Vt5d`VdR<J!e+NLKN
z4RR_G4_9uwakH%ie3G0mG^KH}2;I&<hwCzzg2sph=;MCSRH^-llzDc^Z}PIdw|^l?
zDJdfnnpd%>_Rg|fAtS*5Ix`(B3uMzwh_!s)N=b@MGxW{SH36#18;%tJ3rNn?r1Tv4
zxoQqKYdOOOuW&f+`5@5v6CK*j`%aJRfVL^WDCJh4^0jJ{aZ~|0DqA$Da=VrMu-c8G
z!t<gAd?DlJgN6t5Jy=95N3^7w1MSNyBoU%Xugmkr=u*mcNM?S5_>3)nTjj`86iwfi
z4*rB1q1T?JK1HxNA^llDjLmJNAACh;%fb`Fd~iSoZ#0r<0Qx3rBlVhCp6K2fJ1aun
zL_Y9#j>8ySfC$AfVuVMn;^6&;-F8_$zU%P(*jUeiciZDfnpT{2-<U>e-{&}$-_Oo>
zX#eo7__jVh6swSop0BcBX}XhLX=;457QDuk24vdI-pq{Ar8Z9fu5THdS$5`y({}h>
zkq(v*C5PEXfU8?F%K@tXpMAZk!J)tEpw2<Izmf*;LSF5jwQ%Q?QB%uPb@FEVM5Lxl
zrd+aFe|S*!v~R<dib^Y&5TdNd%6o|E;@)Sw&+GMteZGqKk)dM{u5!=o&(;XRenFMD
zH1;%{jhh~@GXD00bpkAsdKyTyU)>!pxKs|<)tdVmr1aIsN-&pa(1^SI`QObi!FX-S
zr%l`!2bVj`$RrVxI5+-u8_7CdD3*)rTc6==d%pwF+0o{-C9h}B5lu1rIJ_mGQEYmi
zZI@zfe8acQm$0Al)il=3*5W#*3iXS#YKZ3gOt6GvG$nz}40Fy!@|6_+1z@?@$I>!X
z1~z2oBqH2yQi%4p8w791cjm+WP#Zs;Y4)`!N$scp%b2@s{0xt54%jc2jdqeFY935A
z@1vr4*_i1x%(R@6zM<PAf3w>CWOXPssjuLfzi5Dygnul_NSGv=--;>E*qf+4%h_*E
zL@aH+2(Q8-e%9X2#$k1{arb`6>c5W1fe-8(f{a+qP(4_dIMbn5^&(66v$K09VRub!
zynxqNk!O7fujbv%ZS>l8|FnSddni`Ahjr`qEzUMe`SG4_S_&#Vf6DS0DD8RS!F<v!
zx@K4JfJjpjk(Yv^KY9|BFQq5Jm>>yRzqM{I**JdP%26ssifhL_;R=^r6m;i`15LM-
zjJ!xT1&Vp3E3yUZp~=?SOfG3%lv;LXDBB;l4LPJ+Muw^J{KuJK4l0UF%)X1ma;ro(
zab2YauwlG9n%^iZ;g%LDPA~*gP{YnN-V0wPiZstGk+b{tC2PRt{(9ZMjciOdbaV#R
zN&Z=T5d7BBAE^(wPREJI<S2vI6ICc<1iZ^S71A%9iQ$H^y8vtsguu^yPqM@A4I9?c
zNFQ?7l-Ol1DC~>WsT&_#PuRb!;g~#chQw{2oZN)l;0^8;H{@^|7<cJHKL7rZkx)Ph
zsSBGm$B!ptU?FA!Z}VEd{T?Xo9;Y-;3EV%@^2YBn=qiihzRSj5W(Qw5i)#gpFD||&
z*VBGgN=}Sd1(t8Je>sULN+pd&5E^`qx}6%P^)mAf*qZ)rif{%1kinSjCJvhWNs<0C
zJ`Fw%<a*S~FrQuHF3I*9B_!<c*FU5GVF7xVn@WwP`%*D{`~?^^oF|AsUJ|MMd!Dcl
ziN{q<*}A6p#8mkdg4<SsFvLR(TeEco`mluhCpcZ>ZIx?<MOEaXvD82b<U`I%45dY<
zmXao0+9Q$ElEW6QgN<;8xx}BvvqB5L`i2I3FY;h8m{H}6qwikc1ZO}fGv91@^K@)a
z+9(jAS6`?$G@}Y<PdU?~<-AITw<4C(|DB#b2~^|={DcPM<;KP>&I!s3pOxu$I%C6&
zn|LL!lyCw_w*UHe?wXmYuG~sZ()~#uQWb`G2y{;YG~&dhZ^TU6c~7#10q`<5#dx5O
zWMjp4@nxyGv!`q8<G<%sq{uXe7zpwGeteD@Oal_<%4=IySTDeepwF*4lK^#mv2dMh
zS|_LYx`J=*DxW?*2hZ@2MDw+`fA{kXY4H&FnFdYfrq`<|Qh#hWu(h@Ib&3YsWo{uo
zz4dQr$MJO6u|4XD0ap6vDS|fPOw;rA-%f`BX@rh`ZapCBv7~kx?|TR&e8aaJ>STKf
zbw?7@SIo59oxI3Jpqp#9SM#99WL$S^QcUb^n+BEk_6`q2jyH*h!T6qx?6daT%2?bi
z`gWMul(W+JzSonZx>uztY%0w&izROsG|Y#jt7osY$%(3CA#7fF1M8~6bL=m|B<kKg
zt$0JlN1>yR_ttIWCTtQpAw{?=X<Zc;)1bI~h$`{xJ@JH+Z{_9X{zZOc@Al*9BAoBR
zgq~~ONCDN0Nqe#vmnGn0z=d1uJKM@`FaE1H)eozE9VJ$F%;>1D@OtEEdi^Bu7Nu~a
z4kD+ohwW!@D!$SD>^8(e>iLFZoHl}O+}*nVS;Ci@WLe_V)fpwWvG^2|Y4_k1S)wpZ
z7A_Zr4?cv|4_aa;DmZhT)jEhn(-VH}r+Z8Hok6y)W=V*&W}VmPH^E2${w|&#Ii5}r
z9{b!2f84BM-%yw7)542ARs!UR<j6@9LX+M!^9thke4o`swf*kn7IaUqNf9vG6@M8T
zu}rx*;^Z#j-gtE@q{(n-gG*IjP^@mK%snq#)&y(USlXfkEjFj<zc7XVS*4?#3A|YD
zg61CHmR5pt-P+nX2KXIwi@O73SBkdlXQf<N5W%=*!s?qk%d}C}Te%IwLd;vci{(X+
z5BuTm1VY=l_hc!6WU6g$kmJSZa`JOAzIUBr-ez~TrNujqzV8E<YlPlRbb_92Y6?O4
zE}Smla>#6ovsET*&+fD<7`}7FPwwCYJLu$@?$w&7t?e?YIJO;sP39BwhB^OUKiJ)9
zpK)+F8vXhtHL<UFK*^h*RN*~3x7GX1SH-z`E}nbT3GOV>llaT(iGg;y&pm`8#`)vM
z@q^w`9zm4pK(ie;ByCr=xOY`p(n<YrdTmHbf7Hje3Bu3iA@P^7U6uyQHZLg)j)_mB
zdE>LrmzvVv)@&T{K-IeL+=0CM6<J&g!KJB2J!1d$v`eY|O7iRQ5)aXIk7`Z{cV3^J
z0Cbd@D9&!;LU!u899TevurBV&a}34$oLDtcg(z!=dtsF-^HOy$O8p_1tmCz;Qt2C?
zTSJJ8(M(UT2u-zSpycw!MMR(e0j2-4rYISiLzSJ~a&l@o*=a=KoD0|_e*|K|<~;2V
zr8}}(ko#VCSNVhT%P`?|r!Xr4AN7b0z*@g^tRYV6(tD~P?dI^c;xX*?nfPAaO$_oj
z0}EqHpPA07T1uC5G!8gko5=bpps>xPS@6zKQv(on`>*Hu??l|f*H|AG5TzDdsJ_K@
zYNd19sG8g>YtyR%gKMVXc9OE6?n?Z1a;AM_Lqj=S{YZDoqUxQ9v;S>P{X}Ov8|k%T
z3j?X;whf&%VIdvwsPEO%X1^+uj2~n`n#4a}>3R8-di-(2K<cG-P8#Ymw&y8=Uq(f$
zgN<LuRTPGTqiz@bo9M#sffl+nnk+Wbl~wNG3=JK|>n*?Ua<L1QpHkkhEWeKs6})L7
z(0BDr-VxKAA+}u`2++#8q7WKLW7XRmruu~|z)Nv?rT~n|Co53wxNh+QUdR~<jrmqA
zDY}5NU{0j{YObb%+E5KbmbeS1L$y(1g3bG*Q&jH3X34ccf*n{?HPJuc2>$yQ(#}UX
zcP7PsS9*a?165_f7bQR@HvdKNFbbBn*d`j%o8LM6399$eBHT1n-_TcCa;TrR;lV6r
zYTcQj#YLEY>QXUdj)kfU&p>Wxr^8@fKJDuK`)G9@H{&l|rsfZdU8-d~VF$uvY9`|E
zOmO7iG}gGhT+p3ZSoa403e6kUN~Z4jKHf<t)^u`cYU+<1$?l<+Q!zIdzN2qVE&V;*
z`DZ>c#UH<zE$l#4p~0*yfsZTbqUD{Feyc|41aTT5851YlT)DKIXt4$E7w+m}pn%Xj
zR}UxmSj;9yEiSM0mJ=K>gz(?);i3yV&+|!^j$KF;EXctPm?|&bvTN9{3L;-WZOb-4
z_mZB7lYO_ZW@z;;_t|JN=ltus9YMGVXHOq;#9xPXtB1ZIkRi<D@16Xe%PL~=n!NMy
zV(pda=Q)TZ9Mb|f1OtJoU5XA;7efmEpc95a#3sm_xvV`;zh34jkzHnDl{y!dBtHwl
zv^6!s+*^V<<;wJeF9prAwF^Taz@1!&neAsDs14tcBWgqX6?hVW3p(V_Q4aq4m$mx3
zr>L78JNwD1>YmQcL+o+=v}{KZrA~5)2w{Q?G}z*osvj1qJ$_eLqVuro%KmY{%tjwa
zPTPUaujYs6{x0obi{ia;^QtI4AOgre4UCxu`5PK08U~jtD(a-CtAD3VFaCkva}R5l
z9zjIY>P`$!^?s1$CrD{$m+aq0;GdoSg9YxF435S7NToDnHB9pfWtj?Ma1k52t@fPw
zb8P8f&<{rhiqxVGw$|SFw&~DAGi_*&rl%{LLbp+_-}{Kec3!(gQhYV-m#EqJ>4!Za
zvgDho*MDjK3u}bM9}6NqGTyGX4Vw>-NB8)IuT88Kf9hS0Qt=H{G$9@xpPXICVJg5}
z1OXCtx0NC2sY|SBWo86rQiy$SdW~7+!vh{&iLB=6@W%C&-jC|O`?gqhRr`tHt9A9h
zHUrz+@@^-`8|{ye4@<3Er1QR&MjeZSM8ogX&Nm|bj$AW}kzBs-7MxubT%>OvUrR5H
zvv$8)DqR=htzkD9ThrS5?AokDTU#xuzaYRee1G${p|VHaQF6|7{kyfV=(X1-+y!h6
zBr%Q@rVP(}60}_i!{<HB{}`XGwyXbT%GqAb!o<mbX_@vHM%}Z|jD^^4T=UzX0H=>X
z3g~QbrUANwN$ljXX`Px62thrK?Z_7Ldx0YLsFQz#vtU%K>gPqJ(idY=*DH@obP9Qc
z2VN3u(_DY7c08b<-Pm6hn_5_1(!k-=;*K7&i`l5swC)#P)aGOF>K^jc4UiW^g`{jE
z^1m?-_w;?h!Km4~hzk8Hnpw?8;A_8)p*X%2zFMcUXU?t~qU@rKcMZ1$=D*pQ-r1q7
zNV)A4@>i68IMWCC;^TF4qZyHcG|WanW_RenaNH_<R&Ur{`|Af^>9l!d9H)EEv}E2P
zw_=0rOhJJ2f2cZ&4{>;Armvg2j}|Da(2j-1;$ZamvJ;HaFIt4zE8*p6FbQI4=!S5z
zMC>BK)+P^x=vMxmw*)Z>!~Twx;<}TW1CJos>02)!3A}DHnXeXh`|>x}bsSsb`5ji+
z_U!avP3L9=hQzB!3bxJqk6YC$!PZ-ig)7-#l=nDrW`nR>ugL0AHm~-OrQzJ4RiE17
z1c^7ha7{ffC{E^RR>2iaGbvJ2QMDHqjpucAbT}J(QH~SA+0O@hN7QKo5Y6S!T*2S2
z*_&oY0V7M*G?E%}Z!59ZSw`I2iM`*5a%?q_tG3xstK8{sZ~oqYq6d?55qx^3`Di`l
zGBv@<+j(9ByYIgityjC6KLQk27&mdr%ClWF5O7sbmKlMFN`iILrP+*Z)Z^7m&j1zv
zpF&qiad{Vht!PUZd-aRRY39V<wN5_%pw{({u~GYNCmb_Jy>Z7|O8AS%)iNbjd#Gvc
z=XB^=y_%}JfrMjwq*UoQ_qV}+w~N3whoWTH{WGTtYbCR%zi@qIBCKgD;XeHAhLe8Z
z;P&?MV*x@PsSvDkPdE2ypz*~vR^2bz-KC<B_I88)o`~)UjeO!^*K)C;7X~z?md46W
z2>aMp@`iQ>W~QC-Ov<NQlX@0qm!3r$UdMDgish{5XVDXr7VtYR*iP@}=?}w5^6&LF
z_2k!ttnb!UV$n)zlCqfSFfY-=&54GimDlP~VQp&aAR!7J-rDB=d8bPg#Wr+kT;8>1
z@5A(^CFM?P7dA;vGCNIPM$??;W4z8C4R|1}jw!+|&U!<+Iphul2G{XuaFrjYWE=P&
z=RKF(pOv4N$>0QIlok4t42in(d2fL6Vo4|shc0(HFY!r1wt)oGDnI-t;kL3YLm|6_
z(zE*nP*G6~$pFW~;x<BJBv3jX`;(%Yg{HeUYt`RO2R<YxKz_Ksoc{a%$@3uPQu!Nq
z%nO3jl_m9c-Q$uZ=0q{2mJ8cT%2y*}IJJxW`Ju(`_9}A^RFkVJXz2@8@9I{P5`~-+
zNiVFC+RR3_w`qmmikmwv$THNtv2^hzfKh+DK&*nF<xk`rNi2I{Ar;1L_Ao;b2bRc_
zp75jFRe3(XF-3UdQs#x(G<4|`0RE$ZcJlu5%eO8HiAm0jtBgiI4qesnOua2WK1PbV
z+BB96i<)x6q~w}oJs1p=Ha`kL9yoJ+oua!d`mIPd2WrvvdOxk7+SyfM@d5_->5Gd2
zG4#EQeKLijSew#}kXpAAxEpVpT1;x9w4s~?eC)H_>zCO%%PRC+QQOOy2UMS@v#^}-
zizExAIqZYL&~mx(ZCrx2JYA<;^AUpOkn2h{e$6vR6WL?qg_L=oCsitLba83Z7pB&(
zj8*WA#4CzAWag&H(#8fZm@wV7{_qZ@hX_idNO4hr_bf{7q_fs`BbpqpO)rJCtoO>s
zT@5^|^W8<}M2|*YDKlSlfoZ(K;ZsOQT;v(7_}I#YEF7Jc+c`?|UI_U8aTr2p&iVwJ
zAOo<g$ncHIM?bL68XPmb+&z=n#}(3|D#<$x(-*`*UUX?|WHgeReb$=)1%PYc`%~iN
zRUW{Z05%m_T7h0(+Ypz)l-_T{5n(@CwwYK00%xLwsa0%(@<ZE?XJ>u)*Lus>#?I37
z)_*7G7-D+$v{Qf{q#;(`^mo+NwL%nka*~b0xrlpg&YYb{R{s%yTnSUjMGTDQaUq=l
zRWZ_ynuO`kzNjWK;h<;z0>maPvs~zmo9_FN<d4B-5Z~sDc1SS1W(uL_fSVZaC)~Jz
zaMsi7&Yu1)VaOpRc=JLf3gwDbUUq9?rV*EyH(}#7PF9tbDNuAR&hKzWxiBC{n3Uo$
zGV$E}GGcGos{`47ue{+i#pVQ@EX`k>ztSZ+U7b;ZTlG~)InYJj0pUnvq@_^ie76eG
z?9q9{k0q&=h}+eD%QQAfB+Rp`12rO7WX}1lFZ^M$ESrj4*`|26*wdT`Y(HD4AK+oG
zQX;e~#?*NJMFFJX%g5nO$J5u^k4H|YPaARKcWHT#Tm=O?>LpFGd`1HYf8)u14JqKE
z;g#N1+;^RZJSq1aP@VG=LP&ZlSw|Gw_4s#=9+GC|hq}BfJCIdcA*Ycn3v>>(Jw8@t
z&)VI}YX28(e_sC5TR%U)l-@S!U1PNM(+aP*Npr*j<~ZwHdr_g6wU-w60tgR_BX`tP
zce<0bmobm*@53X1@AJdb2CJRQ`{XYtLiy&#dWEB&Zk_e-!T)dIj}Q+V)gx~GEY^PX
z6{)jf=Y4P7RIbMRkF&5z%`R^dDfsOm8zf2Q(~Y%i6?OH>UxU+mIsjZ&_MUKvK%FJ*
zC;pjQzY|S1gaeiQeTaIrc=)gOn?ACa0>$V>vKWfG@xC00;ea_lJ`F_wzV`mTc%aV(
zYtp(Ya%vQO;7<PwPoT`QZCCri-PEX|S{>ox0_3s+O0>#HG!-w8%9Z_7|8TmQOV{Nm
zA9%XgNRGcnU}Oba8~QO^Slk>$wYo7;*FI}KP}xQc)7u9AJq*sKF|yNwmdWA0?O*&K
zP3IoZ^!xttIixvnln)UzhazWk$eB4N<x~!lbI#}U>Af^e<eYPe9E)-slS2r#QNqZK
z<dha8=ksshKYsu0uRV6W-}n2zUf1jOymE}l*tbMC;@SH#R`})sI4<wY%Mo>F+JfDr
z{WeV?<2p!E?i`3hXD2h$_eP#MkNU1EE2a~A;-c-P$km#$SPgr(T8+`5R2~a&Fo<z}
zAqA;?KJ}AEBdgKRs<s>~BLq=To5QlQ%Y#8I`E=Pbbj@PFN}&A6pBboL9h`aJcw&JF
zV+zyV97RsBZjy-r*C89rn^sz6x}dFVB0xl#4zt|7eli=auRp*Stu;D-jiYYZ{7s$C
z%KBbd&wM`=Reh<+*vy?)q1mY-ylC3!Wh;;4mr(ad3FwjnMLN0|@rSdLl5|?!Hlw>D
zA27@=A-|0DM8Y>UU^-~8`bvVE4Ixx}c&NLF&grXa0(#h-^P3R=@z!*w%CJw{JwS>$
znU$uPRv{2gCjDtL6(cQfarX~$ckP;o)VAg_feHfoqFqaqnNbi8TUA#%#Y|D5k0X8<
zyIQQoWyVYzz$a+?sr-?<{TqjO-YCwMrU_gdbY*IXda7Ow{7*-esX1~ha+Y#kxNnHt
z&!~^=CcmGTN5m0vZ4!B*t&7)zw^q2tGuR5&lI!#4<~=j0=r8zkwI(pv7P{l)l-<zk
zJ4N0#w~VPpFeKL*|CXWSL=U~UwM@wkwX3UYy&CkRfkzZi>YL-|**eafA9<OO5+u;I
z!+^A{17BPU>v&6y>LIE9Q%-ZbC<iy==-?6J{JtEtR^QXrts_eOP}2|mUR+*YZnBrl
zhUVGwj8##njq3({OM}TzcVxTon_(8AkBh%!S)6CmW8*)p4owbjI)UF)TS$hegk%w#
zXI}ZL1rt%E`|{rPMXXBf;l}q6prG_f0mrf3pbic7+hPWku2xz72#((ET~DH-V+OS(
zSy?f$KqyR`%4yTW7y+Y6Cd=I7Mq$ix-1MB*!LPUvr2YB0=;*&CWt2(ruO4?$$|7HW
zGLl^Od%+iB%ez<`B}Z+pQWZ&%fohaLzE>wXU?;FeBk|PifDS~hl&THoXMN6oL|<td
zddnw&!WDVTcGY`>d-MfQ$@7!_zUKu{PocC5O7wBgdkN~z2Iy*CSovA4#{>#8$5a*b
zi@HELdezfhwEX(s=D2t2m3<@Q>e*eyG%mac*BCdCv0kr~Ae(zREp)JIeKC~xhui9X
zvc{>jd^SpCY91pE{_S!Knya=XJr%!f<+kOtLL+}KFA{5TjqBVp%gNmpsq&ghu#eB{
zvegZ@I^I=;KM+(-`<?VScr?}cVXI8AyS0ww%@w&)F*&s+c6rK~!xa+U-JV+$MrXBd
zC9)EioSF|fo$wF2xSSL5W4unLwHml&RTvy*s_&?{-z~5|6c{4(-HFGAg*?;53aF(W
zy9o3rb8{MCDq4xB8!P;N{ig075%pgroFr4<_hKLR0R|kG?Nh`Ndm3^&%*{Ua6x$G@
zxRz<c7<*Tohkj}QMMpWI_{ZGB%-ht1p2Yez<JibvY`Te)l(?;>vG(O@fh5)>pNf6L
zBfod?9$k_C`N@~9LI`Dx-12)!@Kk7lr;}kvgO%RkdyS*9wByB(;~wAGoEToj`lBG=
z*y%t%z=v-<va9}X-gETTX*zVP-HrynpEUUUc#(gtJfy=18rh<2Iq5qmcU#OvdGOn!
z2U<}g7JmD1@96%U+T7abg;UnOsNwQI?)9xaTiGBB(XUdiT)_hF8yZ^-`1zvV)BuFA
zC^7&8F8))N-Iymo%$rb&9=%NZ0G1zf9ZA;!@(mFydVyy@;qsZ>VqbqylK}7KZ#2_x
z#VJr?|0WgbR=xKad)*Quu(E>qD|^H}y@&XRNZAB)i1nF!PY*k>b=fTWiEm9)1<ljF
z_fGTC-%<Nw*3PE?364H#6@kMzC6~=LCU8e&>7LMLC&g0@R7XyeRAo&K(@1Y=F#jHK
zksAAtyCO4Bl3$JD9OVjSIziH+opO6s?lF|yP&+uA4==n!E1c-&f4Vh2#9f&~-X1>-
z+g3!<R?AR2%lPkt*h<Rz(8C&db(;K+_E2^@=KTO>&_x)7D>A`D;|mg|hcg%b);GJ7
zo&c6(&D3kTbeXF>ywuV?;`Q?oW+fn8s2qNoQ<Nhs6AXr*H`*fQFCb7=$@GwXaK50N
z%V@T{YR*C?56aq-!}OxIsos5Bgz4RP&n{)c;|*6H|GOk85^rGQfJv}(MIiKDMT>RT
zo#<dM5|Z;0Z1TZE=z;<n2^j;&>v<;N;k9WO{odCdpIsP1=`wfAxHo+B?*hBv+KVhH
z8dblhtE~l~$ts*t6sibBfYU)OJ?%rqjKM^4@d@GpR;02(Y53AYd2y!SReEOXUMOhR
z@ldhZi9k9MaLcC!u?GtjJQX7us0K|L&`#LCo4TO|*QVH`twQZ{*-JGheOdLfWb$I*
z{66utd!BlwoSOgVa_bgw-^`h#-tE(($8p?_KB7aSPOqwIQi9QjS^BoGJ@i#;zdkTK
zuTS|Zpus%OElE2<74=A18Ts)o5ZNBhJr)J1VR=7vU8UAu+xRi8&7)lNXJvA<yK_Px
zBya%_r1RE)L1o-F%jKKo)S%FW?$|@D#L!pZ@HS=LBdN1e91w2jslgt*%wKC{1U=*8
zg1+!2r7iyMN+;OZ<T%;#^rwe`;)~(mw%b0--(-_>O4KSt)5170bFvnQ?@zh^tAAIY
z6prvPD7p~9aX*uu8hxbJ*vcHWod!$XD<!^~d&eW`irL-Lck_%^RF4ArMrL25AFur8
zPm|#KfxQ4q0lezVz3{&r)cbpj2WjRVX7Z#vRhBJTO3uL%)v|q>OmS5<!U(3*owaIT
zyWSznqwe*tt6h=GjvRff=~G_?O4@9nap^2|<i0e<Y5u*LbXn#0H%$#2M#wfq`8LO3
zDpxGKs(SPs3b+(?ix_zz6+eoa8(i!bFyMSaUCDWm06@1~?Q&>D)78xh4O6pIHD9%l
zv&WlGHZL-{@xIn$@l;D6TE`Mm>eca<jHhSUsTLkwc*^j;cfVa@*frjqd7)DgXn5E!
zb9H=Nl2qVEY?EC)QUgQaozFklpisk(`dn64++F*lYS)&1TtFa>QJGI*3N_lV_7+X+
z=xPe+pFWfvdH!GXv*HI+XmBvEqLMnAdv)LzC+P9@XRF12oNz>bL6Msnxd}hO-&+7+
zz$v?F-<(lFvhte?Tis7N&8(caP>TBGis#T^Z`xk-Ot6vKbvYVvhIP9XUouZ}$anUk
z-x=EUbxrll{oCW_H{beuTfLIfrZ@aXtiK8v=t$u>#L?({lY9SlIK0z5U6z->wiGuz
zKQ|cIA`q1d<T$7pe+(4^{g~NmC$5L}XsBnSt2NW25pQu#lHOKgM<)x1Hn}rCsKZ80
z{$pLLrks!olatf2(z_``z0E(DMJ+FGbL!9{yy<Av#`Tj7O*ker1{dRi$jBdUw*k>v
zzjXSC1gadIZqTXe*K_g@r+GTAIb{p^c~>)5s*XwkqKdd_F7qF^drmda_kh#V^ZoVi
zcjTkP?drRwrJ6uo_rnL1ZsKvi`34+WbtvkSJ*UOYJ^GXTFJ)qgzJ%X-Q&V%%wUw><
zNWG9bHGRr_cWhC&KGepzK6DM}DzfpW$^<Q=c=E1tT7{CkLvBJoY0&ceU8=v3?n#mS
zc34NMMwPy;tKS!`+ccuVPr=R&mEB!kFF8%LoZahjSXpy6?_+KB+pK*b1KyQ%Y7wAl
zn<|TGtyfCn)=1pviJcqrs3>AbLd0YRP8V=(lHSh)5H+nq-?ug^v;j(7bn3B7#OD4N
z^y3Etk1d{6yBE`BttjWm9pUa$9VPSKC*A7)dDXj3lkewTw3MeEhkfn6Yi%v{>T*Nv
zSEM$QN|w$KgxFPrqQ3u&y{fAD`|+XXeg#D>o6_^NwuKt_)Xq_{e%bs-*=-#Ht$2Rg
zxKF^~n%>^FpW5<pVgtgsnu^2e4&n@pDSbk=>6qj03E{F$HWg%$WK$hSR@0?1z5l&=
zJ-JEBGW0ctK=gnl)3dOcP!@&rnabtD(bTvBC@?o#94}s8DV|ax2op@O0qM}g8RZ2F
z>|AHKEkZ{PH6G>{C}(iYug0N4GJ_Fn{O*HO)kg6x?g+rL#fRvYy0CzfH<Diw^+}99
zQlY)@)zMv2uS%z+ww2_Ae;J17n?_|H<K!ZBOz<_Se2}_~oYd4*kAg*(Sl3@5neF6{
z)9G<Lfv1}f$vSwg{;Tpl5;WQ=Z;7D{H{uRY$va=a@*U#tTPlquZ{TsFM03=Cq9MHM
zBS}Se6o3>1;jH>9Ln;V}PR6ypU@<9K0UY&nHBB5CLE@WR{kyBBnp6?jmPk?5_y5k$
z#h_oy1d;D=xMWsanS;nTzt1a#*D8P|90vVBf$>Q;%{14!O3-g5#5XE1XaBB{v0h@b
zdH?#aCw4qua<jwWa*_TRjES`P?poG?Ie$5zw=_GuCwt31jqr9~n%a}Z&ByT^AMEB)
zp;NoWvHF{!>l@b}wyksU(#p_^6vxw0xOghk=WfwhO6u(#j}$RhOE%U(UcGNogJtC>
z{s>?;#F$D1c{MlLUA*VC@PbZl)n(#%$GRYzauuBPWA^rAQF+dP`Sp-*vrIJYU0=Td
z*Hm{En(!<0F29uJve6d;&3zf6`LDR>x%=ljh|s_<Go7sZsV2Zc$4t3V*@gX0#9c#n
zQEHl)rJ&fU%SIS3=PVkjMGlN6rzds9`p?q|&Lw+TSkrT>bOyTW3^wJgrr#CNRDUpe
ze9D$_H*fu&=J}d!Xirz`X}8r6uMQu8IKG`^i!<rEe<<rfPVbf*{7d{QW>)f;FiX}h
zId&_4GmhHt2L6b0kAC8zfs~LAgv0NsvL+g69-W3gn!-wvn;N3K(komTmQ<M+1_VF!
zi1dnm<KsFWqo`iW#0xGNw-Y{ZkJq(&)!(|Sce(QQ0KGHfVwC@zugNp1wJ&*qV*GWu
zIA<A?M&!P=<%%Ld6s7W5BE7T!^?~VQ8<d_7Edw2#mX4Kog^|Sqta(AQOUQLkwknK(
zpCui=cM=y-NqU~V$4tpg2ARyJ(Qz1yIq_%oD&rw$C?|(P)4!ew6IWe$`7;kQsGhTk
zHeAOwE7CR_Eayru2$ebPumg;=*k$%B*qOm)h&E6Gv?~?)GtE%m;EJ8Up}YkQL?`QR
z3#HO95MWt#eJhD)x5onZ-7BtYWXF>-kg)#xt0mj!2^q3@wm=h^Qu_bEhl@<=a1C`|
z&GVDP((^-U>TloEWkyZN?+4z4Fsa*@Cdl}JIphg&p1H&>l7BtWx(U<LHgUH+JXoMZ
z^>lrh6G*53!W=vnZ=Di#G)8R{Bk1dLaF;VHj^ER;K?lER0w=e{-2B&<pL@T+jZ>Rq
zf-?Hn&JEtr(Q`Ansdu+)`)2Kqm;HNA^3E4~KEy^p{b}A7dWsbG4=3+cwK_E9h1O@k
zmdnhOjp@-``K|^?wtJXeP)^pb-*5h)6BiMemP)uKWNM`&AS7p&k0jvsc1}_myFz(&
zE#Hx=^ayQgdqZP0W4yA`ZlU|9(@8{Bki0yNlXHH0g|{TiBNqR(IbHq+4~GxSm;5%q
zaDlEnhyIlMoGfk6{uvqxoe#cm3W`+47T4D6Vv&dMcjy5*X3aefeH||-hpEm*lPnt_
zbY%^CnA?40{fv+ZP4;U2*!3}R3Foc4hY}v92P4bm9S&cv<!M2|NE8y;hjTkVe5d)2
z+%!+=Ir~TUSX>J_Y>_xl!mShk?ZNYWW4ryPaJzSbqgsMov$dSWyx0w3_{LzZgQoXs
z^ovq)07F6P;G6k5NA1*pEtQr${$aDY-8r&7pynQ5K>a)7P7Y#7B?RBD7vEiz4AI#M
zGD`IGHoE~jqu<o+1csg%T%a|0o)BX9#Wt_pt=>-3N^dpxwT%+9L4b*HPOxZ%_{mCG
zIQDXPba0mGfVHW$o>}qPOYmjE{5#sp?J#Y9Ait=ffQFW1C|X60Cu3<qSQw=|nBSr9
z!UBhX(ug{0xenrVA3Tj?r`DQXBfZw5O&XT3YZ8xBOGCFa^8XbDU@G<(Wq6yv$!9$y
zxYgq7Y%>hKp#;t=P(Sx_Ce+FxD)sMZ1|)fnhNnL`0`6_Y{9<RAuv)7(-xHeYIVS*M
zFe>Un0$|as+ai+XU-Tkpypob2_ZWl_?Zn^pKV-D-U6$;NZ$^TUb?@cmTiTi(uk6X<
z)%zZA90Di1bO|NNXWQpTnz2XYNq$3^sduDV*v5L}H@T;__Z{3}^Rl+M#kBD6O+~R(
zcM6fxe<xp4P2Fk|{9hGukt$`6JOo-|?&-nxT~QU3kwRV2LV$mT;62x%O!lv(ei+Og
zJFI<uAAdjv=#j%Z>R{!-$$_xAxY|*`{=cmbf_W?Kr!;W9|LANFGY=fsYbFk7*_=ce
z+T>ZOyJ`MfCuqd#zxXVKABxi)GzHljn{~x~_@*lJS;DM^aBG(K71-_o_+7Ey`TmlX
zyAkNm-vZ+9zpd@nRd#{6^L30<Vj`tqU#D=5F+QV?3H|QfyZ*NlR)`fsJ*I?WzuyG8
zwtJc%)bDHW7WYg}NKV4^j%k&QKe-S(w8-z*o66(VFHgIv90u2~+HJx+p@wAjc?oL8
zFL*waVB;jUV#Gb+$SMoKkzthm<@<y83Mpgh$86UtT?f9i-AdzA@23F?(!n6`e>o_+
z-eTN%^ouP+5gTc;=IO4q|2u!+oG5)9n>6hEL&F|(vbhuC@F;H8*XHbGyt<lfWo60y
zfUoNlazG&ZH?Z^~V_Rx>^o1J0-|UNLY`f4~`(98B9}c+)(-sBj-D4=*)sdP%LniKK
z*%WHd@sHK#26g!OQ!gq0g^U1(f|9GJ9n6XQdW#eHo?L@?TUp=md1K16;p(ZfgsM7D
zUC5_g3F0>k4pT6Yf@gRC$FE{vJ04jG9gJoBz$5dO*Y^8EVbIcqFwIKK)BJ?P&($v#
zo5AjT7KR|1WqnxyGsw@c_d?Hy`ztgx7k+^wZBbeO;y{zcwxe>IJ|k+{fE~R3eqq4X
z33KdgpBu-wPvqr@%J1YX2fz8!Btcv*udCnRmbv0GlVPj+ep^krL~5!o!DeIMz+aLh
zZ@mxxLOFxe`pL;lx?5|9D|5E{AYrTzGh927871T_@gfBXs>L=i7mMjC_=b12jzzam
znOA8!+mNx62<50!R)pBI)XZ+m{($fbGMJNNUj6cU;}rX)DCeWERJi2P&LM3Mm!H#<
zk5-1C1=GU7<K=RiSob-Sqjh>JNQaBTy`he5Ud@pUWvspARPI!<pSuBi0`A(EWq=G3
zz2)ulZc-VrpT%mP1>9LaUcVss?&|~iCU?ESIBlz+$hcljZXKK_a#rBoJ8I|IVFPoK
zLGJF^rynWMK+0K|Kh)$cp3RXHy0Wry%#tmBYc0#&FEi~)w$owzJdLbcz|lxS%9rk^
zMWOpQl|Y6_vA$0xTw!3c3w6Iu>=-#@Nf}N({R@y1B`eM2(`6r+;IqXUQx)W>LY|%1
zWH4{RhEsomho}m^p_<DLE_g}QNmw7zH~d-NCZa&jZNJ1(qUCL?p`b8Sa3zdy{WnjV
z5$cK)Kckkr%ysUUnYreu>k=GH<Ynb{C8EGe*cWqmgHlN5C5VNBGP8Qg<x8{+|3VIT
z4uQqxrqh4^=V#=3fIQNhO|5}Szl26Rgj&39)c_8LC7ftPQts>jW$ZdNTlfhA%}2hc
z1Q$^j8#a+^0y@PZ(gQs7>PZKT+!<`3@l^HWe_HtMsSq|KA2l-h&ObeUiv8=|`R*88
zbN$P_GL?-2GvVw+x$`*s*Wgr`&i%nUz&eFX((rF)(sP>6e!=flpVp7hM)CINqtR90
z9~oH=j2KuC7^eI#*_}0V?M3($M~R`q27~feH#yY!2G!EQ$zof=v#F8(f&;=#;o#vz
zi35c1ui1Ba%5@0Ss~R*|_?u%0yc|2h7>n&8iL&<fC8HVTwX8DlDhDO%x=&^SpI6uz
z#l_#EVx%yfN52*EL-|+ylrU&C9!x(+xXb|pX9&?Iw@Mky(Q<$@j^C@?xS1j=y5)or
zmHm2?2L8C$gP%)*PU5<U*oMXvRd)uE-eiArK&KB_Ma-%0v0;kLHT)NTi}jrPu@l{L
z`K>?fe|{(b);yIyrz>w1X54QBl)OXVz<>2e39v^v*5|p=q$wMi6c~nfl+{!Vn#gj=
z3Bnq1|3YHLTbsNnLpY3omlE>B!#0h<u`7`I^a`~->el~SqVHeWY;r>G74XOiRpvH?
zsCN!w6FZaGt1gXI<BSo<RPD-t=<Ij>^93DaggC-b=Xy@OOjOUj00|JfD{A&W=UEP@
zt2I`0_JrC5ajr0kdR<h++PsOGnlfh+ltr-z>(r#A%SKULX6OA7OB)+;wmXo=a_gsY
zMYi*`lA!WXqK^hR)>0}MiR**Q#ageKA$(H_>Bo!EDRNsw<$ZdV<~cjF7v*oCfy85M
zrc6rlA(!39JcU@00D7Poow)`bMSZyLERwaBU7H?s%_uqr0e;>4dw~a`-1|`|(O%kH
z@&X*?t#bjJs_}Mye+XPUW=r$aqJsk>`{@Z%Ik3o3I|K1>`7S`+fA5x#FD?_wxGAOm
z!%(y^8S7{D4CXC{MLpKUzJD7kX50UF`dsUYTS-NAEn6WlzBO*Y#u*WfWkP2<$IAqn
zhOI~kjZ$P!TRLb(MPRV58y@htM9jPSw)!(1a6Vs8eufz4qfwDorzD)(r)%7lqaS~|
zLa&_=0d0T}aDap=h+4eLHdwfhxuv?Z+$S$F+yAtjUG<&IYL`AD!Ql#&#E4v;ZAM-?
zT4Be5VnBza<BD8ROicbQR)NQx6qRTw2k)c7zfOm8kmx=Zv^*&qn=C(;i8nD^In=`A
zKiEs7C-z7qu`IM8dNX}^g6o1sVU9NF_81vE`uu@4<Q3#`{PV~M$zd&Fe-}1Zc#I6w
zPd$8Hk-Qtke1XTmbbh|AoUr;66tWG{QcRKseE9Ra(w%al1)2G%&&4RDiLA4Ab4%*`
zL8}3{d-*8$RcHbu%(^4zN5%pRSE#^tk)oSk<{Bg0H~e=Ap=VzPZZD);9tjkV^)K66
z-Y{>IjXPbZJaTw+b~06}p>gg>Y<Tvs#n(#M_wUNEO$&Xph{ATgmt@nWI~c}cuB`O0
zfB<y?f5wl*%nPI+kBsEDHp#dl4CIAO6MNyD6}MoHA)tFJ>qjh5l28!Hx-T=MtF`C+
z2t!`3y=3%00D*TCqrO19!#ni!OwFf+>*|*(9G9lmYu2|gJ*Ur?SE**vc8A-ow~o36
z(#`W2llNFCBp(w}qt|nK*8g5?g5?1va=$)ij1^Q`T55DaJ;NR;^HW`0u~9l8UKdk#
z-xaFnlwS-|)VIlP^?Bfo=kw(|UVtXNd8Y5~T~9-8!`#5Rk(ii;7`MP*!QoTpmbY#G
zG-|?v?dynOP_S+x+|2#5070KppTLTekRRfZgWJ5XA(&ga*%3-LPJkAm%O+jeAh$A7
z32J5~zb(R*Rp5I6fs)MWIdI!WCcmJ%R_8v~=jj+(d&H8|Hf#kS`IZ~`EP^e+GbMR$
zxtOBHcieb2wEl7wJ4GLoI<iwaReO>EBE}gON`8+!Xxp=uboS7L!yt*atS5@W?XUhi
zxn86V6x$b=KRv-ohlc}Hd~Z7N@R3qD;u1`)@bqYnGd3o?;Cs+{x7~FahO0v(%HbWw
z=%pA9VCvQR`F5(!$@IWrz*KbmP;UY!NmS)8FzhG8_Y%SwPdhZ3*6(zDKYQEvVBu+a
z%cFNwbHufl57Flvhd2!JJ|}1n@nUNzDNqIhG9PLftNVP~ga?*eUQeyhm^?iYy)STE
zB;5Qt(c_anD<`-a)aj*a``&<F8C+(Akbgt!o4iEZ{}I3Ij&fbd$;gbrznFD)OiJP!
zo^y!Pb(Z<EJdhQXHg@xN;rGAX?lWGy?UJh_Z_`>0kSvqCc3=J;bQCZa7kEhNp@cpZ
z!3Q=oZT(#oepdm><H9Njin^3~y0VBX4D<I@Q9>|5A*RHFnMRCn^cPRnN2V%91z$dx
zAmfF_?)(`k+-tDk-ODaaKd5snp`nvm5u3#J_`Q*7b@3ls$clP^FchmS*5hVGCxfl^
zgWMrW2JCbo&8T&1HoqVPjPjd{8;|1Kt(l3&i!2YUeFgBTlH<M|<qOr@wMjZ!{?CyX
z#;dpjkjD95$luP3OC}04HrtJ*#sluRrtPl8+L~5J1~J|gK_zvn29H=sjkiojO%xhe
z3By`2bn1|RXP>dyXm_62mawkkyd3LETcXv?#|3%C%my{lkp4q8;Najaujlk+XLnii
zY<((>517ikwvcMFBv&|&U0a<8iXN-H{PIw0m}Nl1b3hIhUzDiY*m7UjUTzd;luvUq
zy|38xZML9&iLU~#t3_|6o~E8>GQc0ld?76($FJNQj)av{b7ned+hfBbwbHMkvZ|H2
z=-_Z!Xl^FEW{pV~0s7z(;=3UtC7}7qhj{Ko!0)L8Ozc0ckZ>}oB}tq%b;j%Wj!<8J
z#;=%?<-X4W^J%+Sjdu@HyYKTZ|NTZuzfml<{J}b>uH8@icr11kGk5~`tI4HVb$9=N
z7(lykJ1xn7+Z_&(!tI<b4#~Y(tfA^khgg1hq>RaKH~;!LX0=sV@SBsUZr>&V|2~o#
zQ4*A;3AKpP{WbUJx00r=GU^kO^<z77s52E*44Vt#eFCIL2mIW>SosO^ga$3_S0y}t
zKRs5ntt;nbGXjW>voeozsbh^>DT~*lhd5<ijLe|1lQD;nBJf$L>-lr&^osL)%t*if
z6`Ypi0*N~^Rr;ALpUXH=Gy(YL<YSP4Hdidy@Ol!l-Ht8t;>Dz^G8+s8(5Bhc3-h=z
zaCfsiDRC#sFC$FQW%d`FPZ`dhKiSY2V9C%=TWO@sv`*dv4M(InJrB6p{ECH@MLuCu
zShrp2TegBkX#cos-_V`iJC|;QeA4Pawe790`XZU~w@y2g(luia7Nh~qvg<**eywf3
z)Nd0uxuxXam2{D_asZEDQArVVw)UT_PrrPDRT^hkF_bFz6BnX-$?zv8lk4DlAGnyl
zAoUw`qTYKxX_$E0f>&xk+E<%@Qz4+>t5wA=FfW}qj`a)GjG7!*yBOc0-lfrbtFer?
zQ>85FNy1)@lkz|Rv<e_BrOfk|Y_M^>^6BQ*(*znP$VXgrp=s0WtV~a$(t`@S79IIk
zOsb>nuW`>SNpA^s5CMJxV`G%lT)(d2m&hnedi~i*p+ji{)!p%LXE&m~GYS5i1G4Z<
zx3uTuvYO`X%Z*&dZ*;PGo<DbK$_%t%(6#O<Le}A7kkZ}CseVTUm6{ZdDaI`a$)c;G
zT5OHA*CBJ!k2|PBNJeIc$Xt%&(=Z$y@%ztETc-g>!O9x8VJI-w_A^QQvZEF}Pz(CU
zjOGa<NNOOS-<G$XG%)*g^$r}~Vd7T!$IO0mH!=I3#}(lI+d3;=u#vHgiddFKVJ~x8
zeKgT~KSN32FQ|FeauIlVCc`k%XZ}4%<#cD9-D_`W_mFCukM;MG;ked!ht#9i6GwLq
zAuK2?`eh*a^A&CQlZdLd&a|z4b(cz}^AF3ie0@KC1yht{J|3*nqaY#8ilnx@#jhIQ
zT*2CqWGB`d7^$N6{`y?B>YuiT>^%?jOPz7}hChK5r{Hj>#8;}J`!t+!<Jk{A;eY^f
z#p{IO2@CI<*A+h+ABbN~J+y3=CCMd2l4VnLUSV4UQLK<r*1Iu*edi6tX>`YQQ_YV(
z-Q9nQM>uI<?N4Oks{5TT-jK=lnyJuMjDr-upd&zs9%`J-qstxqEB2j<aJFNrvqM#f
z^|ZAl%dhKNptQ7gcd<QC4q_M#F!gD3gGnuIQGbuy>Kk&c5yF)caDb4i(FdIFK0qu9
zkIUUQ-BxI&8Y1e4qGDq=x^jyRmx98%K6}NH_8ORaSKazH$>cY!w)>09Y`IxDD!+DN
z&Z-EUj^gHsjEP`*y`gwA?p8QC8#Rydwd5&Nlq0EB&PCVL8@?lpX91%E(xq{ylCv8W
zDfllWSBu1W5_|e_=<(*(#?i?K4S%PXW7%5uBpN7(3Lp7wXPMaVrt)&N7LO>as-M4?
z(`)AKZJ$mk^ooomVW>`4T%<~=6j}y_MZ)lG+Rv)T@z0Cuk~(gIOjJaSP!SVR6EjxT
zRJatp<Jciw-IfI0fEj-I%mNX4%rUA_;N`u$d~{_+WqIz3P{9v$_6VG2+D<b{h@tm4
zfv%SpQjwz)$YJblS05%LEu6(HtB0_-87|6`CMoRIrB4<!gjlSoY8dtHS`%W4f5uEk
zuJ*N1>-n}R-_zZFa^Mbe;Z*3%F9IPe<1xTF_S*#|TMYzhsK-T|Q^%IfQy;devo?U^
z<6Ue44V>;9w|}}VkI?U27Xy?fG<`08&i6x%0oPvXcB(Op1T3ZuAkE$(@%_>#^W3<N
zk}Xd^|H|j-;o}D{E;`Bbkop%89@*Xr!39oCg+~9F!2fCUpSoFzN9S#eU5F$rRMF2w
zx654-jpsb_rKRV3#ZMZ0VixqJxP!Z$O|d=9_lnnIcVf}6$4J41jQaU(K_6Z~*=w?)
zi*TPBMtmJgL?<U?ySnoK!E5myC(8PD^L*_~Uo^}V+u}|vD5#Tblo9-O=MxjO%H3hV
z>*__$mw5?L+45FhyOaP?l<-e_u;TY3+O(R!K?yh*v@#dPoAv8HdDBw<8oS>gaw`x$
z(-ULUqrTqCpi>~%t>J&ZcYXrIk}v{sM_c_v>?@g6NZ%v$J2KYyIM7n5$XLu~P<8Ik
zzAt56hHOJcfCRZ;dCO;NGttT$mH#B325OsMR1|H>Siq{2Ri?GA++uTJ7_%{xv=t}p
zMbqo6xpS@7&-v0sU2JPaq?m<ZCLu(#EtV+gY~v31srxTe?Q+Y{hA?2U5f+tN^JK>M
zf$DrIt|kBQ=){MBhAn3XhT#^kUcJWnICETq$Ef<xpJKOSoLfn?=NFN@tjv3pMyULe
zr@{#p7bGfQ=k&euemvOOIX-{3H-#kt)Baj!cAswekfq!Tzw+??iBpR`JYpXl+TYJR
zzFmL|xbC4|m4z7S-?jvhKS-U0VfC%7t{>Mzj5~@99_Xec4aEXY&4bos@7*^D%|`!I
zcx4E2@-CK72EAGvk&z=b8%~cPGwvB)$E#B@kfQ`ixPJT6^f7H;%`1k~I*gwan^qs%
zB(;i==0*C0C69f4DSwTNiW~DHOrn2tUAp)a8B1>r{S4+7s|6<dbUKFT{+N*+-gqHU
z+CWr8sfH%Tk8KHqN1$jb`Im9ul%Ge^I?0&-Iu1`yPp9myN5n-%Ms~MV{WOtjBI;jN
zXPeJY87g|jf_1{25hgcWDErsc@{=b!u^nqQ1|x6=e$af?Pz7w8*v<2F?#rRln@e(O
zt=EE0g9>q3K&2x}X-n#r$4raTdZ}?k^ySPjlFkc3j$oLwG5oI=>21-PB>0p1J@vw<
z=HFNBbvK`5e`@jnhCM+I=%W76KvBX2aHElLG|o%IbhOXarQ_KjUT-f5H5bk?6J<}H
zI=DmMd$^%v=<@28j;?`>E(|6p!}=*aBLCtwo*8<j2Xt25X4xwKynu9TB2O7hlWen$
z5!D15y(6=|sRdGK7EIvlq|>5xe2p4iIi&`G--QyReI1HCl>+?NmH$MM9Owj>TZ-=r
zvDLTw)m7}k5_8&3Jb&6B$F*SJ?14GqRaq7sT+TX)AYED-%cnC*JJ}OH!-rO^;K`T4
zO~?vG-hx1?7#qC}MBo-UyjuNE+>H;VHRqdCsmn^CHT2DtFZF=BJ%F#r_+6z_o9**{
zu&cFFtUSp8O;x_$41ypL#n(E37$xaz>!;+^uM4p%ckur~6MN^Vgf=fKbvH<SoD}rz
z?~8G4L;o5dlzo4H6=&0g&?0xOQE4CLGf^dvEqqssg8=a#6K$auQxE?=0?u%|Zqc!k
z>jw51YB$?g-85X`B}8q;q4e!kEOBpLmqO9(7RZbsMxks9<dX3?bC4yg`rB}=y87lF
z&nKXu_YDk7!on&-7V+2p{%H3NoE#lJ>#MyK$-A7sJz_XhSSzSi$Rv9)J}sq$k<oYQ
zJ@{*NHW0qvN}5((=z4>_@Vv%FxVNIHgY7%3ynLgg`U%y2dsV2>|1DSyVgPml39;7a
zzeaSx#;uj~U{GOGp7Iux)B*0F<Zh*1@H@NMI8x~&rRQpfA81@C#!5H-j%)_Iv?~h<
zuZRH$3yY_xr<!Mrnmw_x4F=)7t`4v5UaDRC11uhV^iAX80{!+N*!j0^NKfT$YF1Vm
zAXTbK<n?laH-myr?b>`v=CeYBTLxvaO-E0i2Qh`Y!m_J5Rbum!;{!i^y<7PW-WaI1
zonog}9K55uKIE9UDC&>)-@n?6R0<1tpw!y<O_a&trax(K2J^HzAkcLXgtq&}$iXZm
zt0D`^ud8J<bafW&Z43%+<;IoM(Cc#dsf05J>m~?Z*rEc^c_maXY0Q&ER<Hucf$syv
zJ~@8-w8)hW4&n!{sYsz4UuTKPx^Nlcxf}h9XoZdN62<1GNLQg!Bi3Fqm=;aIj^pA`
zR!(B+6_pNM^4UMEf3z=2U28I$4C%teQV$-xhe1GG%zjzb*sCIHR|tWyHm2fG<)o__
z1>1I~_b2Pw^ZM>x5`w`Zrtf@KPu*cy{6wd{qPs`u1-LKVQI+vVtiMcxU7<pV3VuPP
zWreiV+1f=iVulLNOL)}#6iU6=p?6e*3WRT?MsdZrzxzek`=&aBGAeWaE#n>pcXsw{
zHxN4MUVCcTZ>%ndbO5Qy*!hyu9=DI3IJsq%MA8pTHPyev2s6T(;1&=6oo&ovyCXit
zoMJZs)#jxftIh<*%3bV6^i6r!G-QJ<6_c<_P+mRm#Hn3&<zKkAy{0CaLZkMM_g2nB
zunKMc4Qs-TT`Sc-xwC<!XrAP6OQ?B<x>F%ytC==7qT*uFQ&V}>jXXyuyG0$i8si|i
zC{;&s=O`%F&ouQc`9>o&`(d|x$~K}T_%i_S3U$oN8LbWmDXfRu6>wbBN3aaRB2w5B
zk*lk#GS{kR{C+B3)LIDmFSj`Oi`a`)!-cUotgE+F(%vf5hU|KU{*IKc-u@s_+8s03
zR-Y5(U|N~;<nQm9wy4zFgaQtvVPREd3F`%kf0WYwsa=pOn5oR{QkvN(O&b{ulpzhc
zx2ll0-G7>{c@}W%<@x^kE}#}09Y#lgMaulZ<sy^P6YNDu{7}so5XgXb*kM=s{&_GY
z7kv<*@}#xRA0wy+>$Z17(9_5c5UDLU2z@F@P2=XbfLz8Dv}>;0m%Q<R0V;ZSQ+fV(
zsloARQc1(mxb368sC?>Y?hH>4&jaL13Truf5Hss9YOCfT4-)3yRnl`#uc=T~2M0If
zLGA9PVgrwPEzM`Cu6FM7WQ9L_PuRmkw1O$8zyF9aHiz$S4Fv~nf_WjCb5~OGZQJJ~
zoDdsxEp)W23hDWs!b-+3?O-gA?tf0GZsbyY&<qaH7v=ao?lT0nPgZ_WTn)q{$9P3M
zR7+cKT6_{?&SY_GgU}`#j8-nm-OWy7-yK)DOQlgOuG<Fjzy(}6pr+rwl$wjHbC}KR
zu8>1dk9MfBcuS0s31hKrJQUfYOCh!R6T|onQ0y5c!ivox*n(Th-Rj7aXPy=$w<LuT
z6NMeIjkv!TMfm;*yTXM70GZ~o7j1qUrY4?WR%SXCeY<Ss)v;BzOXl?E;~RCf0|QlD
zj@&dVvG0cesz&Y@$kS<U!RTCJT1FynAOV?6OKt_X3!nz~xR|#UzTONo@+^Fe$g=#d
zFyBWrTU84|6}DE{1ev-O=5F;tdPP<{mqWzI{W@W$V&PVBah2*~K;O`s5;alhUTHEA
z4^O5F`hhkpj@S+2?li{sjX2eJYX0m(p^ZZ2=0nR?%CUYbwE&m?TJ*_3x;b3{$X|_o
zgz+>D-(OvB(#kggl`rvF)aW5QR6F(QIY9E&EMOA0@B$5wOKzrrv%rvt;vEtk6!<%*
zV7TulUVYEOVic%dI?lSPGot@0_DujZfm_7e+q7F^J4gdFmma27ammxLBI|WzA!H11
zFFEG>&mO_RottcMYsO*!A(LQhMd6HL+;@WD&zjH-_ttqnS?+sT0cH7@Kl4vW)biJH
z!$2X-dTB~UQ0LihN=13n$KHI?>YlTc2jGK$hq&>nY8T**6J10o^I5lE2#iLqjP8hd
zGPLkq1hT-zRkdkPhdclGRF|cyp3pA=V##FjPdt`=d8ITGvJLf1gbCjr=r9>?5jNF5
zzlO3fy=_!b=F&bCYTpxcy0=E$CAa#GumsGjMj$YeUzCd3KnNKsUPAL^Q|NJEs|?+M
z8MtUB5NIJ|OPi##Zo{vXGG?d*>gebol6p0158)^5K?1Ty=93>kbw?(F`bPSE0zhz(
z<Lq}OFSTkK-Nh$W2o9*LvwJAJd^~dD)SVw_LvokLgjxkKyhdw;L~Lm3ZahWiuZjsR
z6}O5|o}IO2aZ@5OGOJs5cMg@)2-qn4d%o*{2WiBH7V4iFPHomOX8LT3(goOIq)U5Z
zj!wsSmu=p}CnAHbwIJi}gw9)RGcvRYV+obztUZ@>WX|@oUx6{d(y@Yws<i(<3%SQS
z4Ei^(xw}>g*JB)t>gJps$FMw!*M_y|t}T#Sb_v*9PlQ|poW(s$EKIdtRduDC{oEI6
zdp|!^@o1`|mJU+QqP3y?vkQ(ucZDGwR+<fE_oIBolHd2<$?l4f0bEqQeCuu@$6Wb6
zDwlGCle1PozwP<Z?RQLQQPqJYfUEuv=n7oKm<8<2efAt6*L=cM=Y4%d6_)phl7{Ad
z9fB-|aS31=k|08DHK)3{llGQbs;;+_p{rl(&Fz2yik6~W#-Z|`_a%iqFpk>~e7m5p
zw7C)Ctx*#6IBG+d?|amK6e>(hOW4`*wb{>>NCx!y+JKlnCiL{f^K5jaC~$SRrqwf@
z81?CQVk*-U$AN*`*T1hs8+66;&<sZ08U)e#!}c}ng6Uy8f(8E$H~wKtfw<U?vqj2z
zhksINirw`!fjTWvFNi-2uO(3C7(@Q;-}Qz#5_@z_4SjI1=~z`C+PO9VayD`L`z&NI
z>ILL#-_T<NEp1!lj?RZ08-M_6&iOYa>mCOO?c+?#i9pR69C??xeV>)KGW@YH+fowl
z+ZDEwi-Iqjq^h?Qr08-LQ9@5)V|G5oMy~cEF;fu<35CkOKDr}CeW}G$1X`&-Po-Kd
zCPvpHgqRYMU!+?@hGnMO51>y{Ks*yab3kV1SdlYXD&iu_|L1uOlU^iZiVZ}kS?izO
z8YKmIFhi3T)Do34(1?s_b<3{Zvz^_wZAxiR>{-y;2Q9rI8U*>(2wjUeJsua+QEXam
zL$7!JF(Y`p%6&k#$pl{*orf4VFDVJr8MQIYSB5sXF(I}2@V-;N^CKbRoXxQ3R`+Xo
zeoETVBwh(uNb0(9@UY6QgmSl<{%tEikpZC5&DgWQc6cm))IgZb{N10lf2Xr>%&t#0
z1tQ9limc!dJm`Kz+1Q}_iMd-U^K*KCDsO5(okH0z4sM)dp@Ct=bqq%(ny4EQ15zib
z<g0Bs(|bZrr9kJdqPcajSa^8&o!vtsbrY!uWfkXb*S=Bj>GmP+E#B39Im*nb4rppl
z+LF6FudL?Wwk;()wY51+GRm$7h2MI_FeXuHCU>{mZRdqs3F2Rz0z5MAN690?`N{6~
zV&_n#*a}vo?fCe3`GesXRGOn#XF`F&`b?DlzKe<^bi$6c8T8ORDk9>)g_+A`Nl2T&
z29jPfPnYOT7MtdxCnq~uCl&D94*_r1n}Yme_^#H-0vcI#+Wm9_t?VibXehAQlvxlv
zWB8gW7*FHm1aSu`ixz2Q3C!C0x^vOcN|-R#_!+RLY<Za$<V9}kPguPh%BnuzereOy
zK*ViF{Uuyh^Eg-%^DFFoh3Fvdp+L&NBJbvHAMaDB%EcLu>XD8+qt#Ug&clGNUD{29
zhele|i3v2{pBq8WU)KOr$>$9hq4Vk*b5SfNQ=z1Ex+o=qKifoG>b#NI)KIicbx7fA
zI)z^|?qGqmsA};vAU^Q?r=ipx^^3nw7jc-W`VT%$ggW=Cq@fqrWlL8I812`U*(@#&
z5Z1@5r8fc{Mh!^#9PV}Vk-YnFHVod}L5W;zNFq7&8Vt9>Xr2`4Wh8Vk98637tXGex
z(E}r8-=zew)6vH<D>QBLV`M(onnV`Hq?XVm8Kx$~`@B6U_4^G>;DMarD^uBxu`q!3
zDc=(+u{>;Y)Qd(UnO;PMA>*Mw`E6Czt@}-$;&e_d3!$1NrBl*9K(|Iu+bMePW9&WA
z50s%c-%!jZ?vy$W^OQV)Mx@>d4i;i;5Fo82(I#@E0C~+4-*NS(4v9LnVCP6~<n}E|
zA^P~t7<%bW&+UqRFLk_Ow9WuZ49GIDMyR(D*n)8?g4tL>PDBgAm0QC?WBGG17h+P7
z>7ZL`?>J<qPEq=H$*N$~&0yz$f{Np9Cc&nWy*f1SO)Ru_2;SZzdE?6)8)r@>T$vM5
zM8b_kqtU!$RfNR&CA612zflwTRY0p>z3)`sd%uQ(2}`Kqop)1Y2WqJg*ntvj{m6vL
z#73X_{Y=zkG;zwevmv-7JEPOr;6gcdKu_D`r4Ycj>Y?#xmGl-Fc%~&&GK{k^KYEHw
zgZ7Oiysg#Lf(EtfPh8sxxmCI$K-rMmZd#gq<xTHAPaaD<RW_uDwAeK}c+XF%zsNFo
zUKt~E21|XKb!eDcRQ?B)6{%|q{j~ftz5U#t0|awq&lp!(m7pG@(_wvnUl1<$<h7pq
z5o~Wx-e=b_F`5l&cXAH+^EBJgd!&=`nh}-#{2t>~=lCQp_>?7qeGcKtQi+%{p&dqk
zhKE*oO}q^0?Md3CpYOXl_&k!C^%^8(U?`lr8mX43UCg`E?zh0ZqS~kPQGbgpcDTra
zouBD?lQ6=YRgCdw<LST7I!(92)s^p1f7)M2;bk(l>{wkThqgM9CMSj@l{uL5{c4ik
z3IsMNRAb#1X>pNsIN=rADQSq5flc^Og%ey*IB@Q#y0^ha4)t~q<J&z7Iu#AF<>KPU
z`BEKlafc^>B&|kVjDcZRFvl8AqL9w5S*R?{KS&GJ=mNHI4}$ZsDaeDbAvk!nuP;h`
z+P$*jvKB5+(qqAJ&FizGkveZ3pIV3-G5nw%WEz~fJnJ+4C>R^-{w57fGiFOn)6KG0
zhThyTQPT=gJGp~7Xy(1tH0j=z^jwgg;Ui>7h`AcjGNr55R=Z^{+QGw-Br|MTU!X#N
zPx<DyL;5<_{t>nCA&+D7uvh0sY%eOI9O2D`<1yQ===PJ{Re>|i`Jraq*>W{i&T=z_
zY%HKH<vLT6S{=+9TW|dv?2Ih=PwEqhPpD7f;aESuWqpX?H90qU?y{hpp>HDGqE0yu
z(*yc$cbJ&tS<x5e1o886mDk}8#=37w233o|>0x)uTz{UM%KW_j)Mc7tIGk<03xiaa
zUNsk`hlL<BgzMX8W~P6A#jPzYP?-eewt8YWbilL3G!GYW?$gQ^EDL=GQArHQ-yl{I
zR74zWqTM)xHTM>4?u-}-gX-;k)j=eHN^4&PZ{|cuw5NA*mFT~o*u{4AF9+>~*TWvI
zT+o;7|ILp;WEpZ?6V$nx(s;8z>UNz~eKd7k#3s3A`Cp;&*+KvDz5sREOV?diR^D<T
zk72LuR3RNnw!Ze!pe0Z2Ea~Ow>edDQ*PbP5VSkBXe~*K1LC0|9HxTWE0CAvY#>BU<
zdcfJeaiBNX-sOTk*Q>c;F_l#Z14ToYX1L(qoVl^5PZT5f4=e{K`e{L1RXt;{k;*Cy
zL}Or$D^-V^=64lz!6#B&Oa!XK{60XJj_~u(Y%zBHPT8XDpMB!+Zu$*QzfL{zwXH3u
zf>q!g1DyX`yR8}ZTkEX%+k0RZnTKJfr%ROPcR?MJd>k<3pLe@|KS{3gjM5s7bT&m!
zOm;@^@Y8@0!bZlQYnLpI3vYK9-=$tc)IERprKuhVTjMQ|Y%g+6JJr#`Xe~yAdMog+
zpYk)XX#LheZj%?Jm8#YgzL0;@w_c;?1B0XKT8uc(ysG#|g+UE_Fd(8bEID_bufP9;
zxAH3?6*-}N>>G}Mu<Nzr%-1qyqiI2+j(wR4GK!-oxpxcAnZSR%D(LqxN?h>}_QtXf
zAu1bspuyIAlF;SqNROPmsB<|{*~)q)k|#SYlgRW$mfl1t-Z{U>8b6ZEpu<OFWDXz9
z%!MqL<Y_64_a=h)fj|7Mq4HV`9%Eiz-5=skPnLHNLpwL>Z8Rq)qb#PF7Q~W04N$d9
z>;eM7A@T0{zxlJHjm6GowUm2PD=Q#wutbiIt!kn+3k{a~hX-kw&lC+w^1SVgitt@0
zjw@Mw0uO2-nAzEX<pgnEk|%eiUpR-P(ES{n>9yONNsE1?EiCe3D|U_jRUdFjUGF9N
z7Fk*5x5d<22)^@Ep*Ney*AO8100On(S4Xg=*eAEQh1oyIJUQ5GFMcK*7j=Yvmz|0F
z53t9{ss6&`t&V@dM-u!9p-U$!$crGIFMa$5H+}7?bn>UJ1$X7;E-U?>r1mR%{N{Gl
zq#2W~^_%KyC+BP=g_f5G;ArT~pn5!{ZoP&ykr>du6rTe*t+DezCFV)viuw5|lzGyl
z@8zdx!^)h6U(E7kTko5}nG(e-o{HnZVhxk1%Kmzky;Ar-Ih4iHVsu>Ln$c7qx$$bH
zNpOkrDB;+bzEqz7B7(*ANAY*_FH{SVt$JE4Z(b)Fk`+pt<OU1Jza|t0hc%LMWSB>l
zCx@^ozwWe#1Fsm37TG@I$vqHfAiu{)l#>~I@Ie1&sK7>A5&d_ak1Pc=Do9|9)rSm0
zD35F>m1@k{OU=vxC#M4M?!I{=&%w}lr*H+5C=1zsdUzRgc@-oMA|Dt0e}9{PPG&zX
zE=BybTld`lF6AXU&r{sZnag=YXxZj6DC@QP%6JgAvsh|%{E({dZvRi>#XSsyvy5Z>
zpZ;-jQS{i@RrVqQ{gT(Zi}Q|8Kdi6U#A;&e_UACC`Bg%B=DCbN@MEu98>T{!Psjh8
zY~DOQT^2U<>E@rj+WB$pA*(hL5^>@%OQIRegsVLq8KDzugk;K0El&IR?sA7g<b_tu
zjmfNj%-~zLLRH&`#W$x2d$AYyu@q|Xenk|2yTA|cJRk_jU77sDD=(Kg%?-s%%pm=B
zjk_gQpIj5LrLiKmk*RHMRfg9&LB=^oog7FJL$$JK;hcQVKCPZ7Q6au{4B9H!ZqBwX
zuC0OD-31$nSdG}@qczMtIm%i|r*VVucG4l8)YTjSDjwZlKBEq~KlRr<|Cfg&sr)C{
zr^UWG^9KmzW0MJLOS2t{`Hc6jm!7%|LPC<`9XB6MDu8ZT%Ui9}XC~al7g30&18#*-
zrxH<}Z*xtrJ90b=XXzk>(DhdPk<Y~=*Nhfc8jzW|(jlyNah04_h>dG(EU95?u5;fM
zQQ#)UDb%e2xHvgclbNQ)EIQ`mm56~rzKynZfcKghj%C4|32sd%e8&i8WL#c9N%(RA
zpgtakgc}$UPGLVmd@u*<)U0i&_o8zA5*-q9RXa078J=3=;Lt{ni?!*AI}3|-Al~PJ
zphps2wg?=9!%zN9b_waX(>W{aD9bNP>~T4Nj`3E!z{p2@MX)1>*{3hZtiNK@zqAbq
z<FfJ&`^+;Mh~3K%^CK@kJy7KEt?#_p49hMm_>qz3N%3PtNEF{L2{IeN`o>y+g`)4>
zu`pFxE|!O~8KgoOw`)G>2cm=-bgR;2E?ufKiE-e#T@Laene`Bs7yeucfi{M$|5+h@
zC~Cf7zD=F)=R|3||0tb$Vg>64ymA73FI<^r<Kn*2AK)w%d`R`An1D)w^O(0qZ49?e
zniYh<LUof)+q{|D?=1P#uxng8>PO!KlOboTq+PK}?NBxl6~S#kMaJ4=Vt32T|Bt3~
z@n`z||Nk6f%4tZ*aYiXQ7cz$==R=_bBUBV}$k`lTNSTtul5<Wu<S?h&hUF06shnH3
zGMglp)*N#_{a&Bn?e_f>%yqr4>v}vNkNe~E9(}fVM<_#RBX?Eng`-w(Q59C=y}SF-
z9_T~id%>#kcE7s*O{S04=5)w40Yo%foK9iivn!lOOGg*ykdCy3g|9X>eG__qD)kh=
z)s!v5xfy)GDPM+AXn!%G_;AvY=d~i!&02Fq_0!9irJ^q;lcNIU8)t9PiH#0ZHyf#P
zz;2X!q>9p<88nuB>$S^ksSE)*ZBdJ4)YlGzYd!|T_5G8bJJ;a(JF>BS$|3!XaBdC%
zMLfm$bsLV*4pv6IKI&>fSPKUi57pG2^;>@YjH^E$|Ek6>6qBt~Qq?cCC_80btX2(E
zOR~j&gIn^OL1C%L?jXC_QlqHkx>}(Daj#>tPtxSy1`q$#eSY+b93W?*{fFDATYY9{
z#l!f+l%!JLmnf@T$7IPM<@2RkbxcM&xPNFmFKTS{9lUy*F<@6(2)A_q71jgyMwO=(
zVu5<%dJq+6k_pDNmGEk<fQ+^Fnv4IdSQhV^^Sis-RO^MxY?`GaF*0uxf0|?C)#)3w
z7YQgfm9@RLN~Bf;xvYWkI6C@j>zH%i$5a<?7xN!-aMQXJO-%(+MR+So%3R}u21DzU
zr}G!Cn#3OzXMeCRAO?+o;g$q1FBw+)i5+lW`9i93#Yzj}2pq>k!4iA!rV0h3;0D9>
z(<2uGtDT*aZ?~#biaYrw<JI^d9MfB{ugV?qs1Ne{VlvQ#u<~v4;<#lgBCo3Rk@w_n
z&;R;$#Z#T}ZBpFubTJRvpfH4d^7|p;QlMbiR|@*8nmTp6ef&c3GiY#cu!7jFWMNsi
z9d0wh%XZ$@)nxECvQG{IwS&Fq6!DH%ULvx#A2AnVNqoMX`w~Nwr=q1_=RVNyP_UR}
zJG3)l3V>Dnr)TnaRG`cNASybJK{zeKm~uA)pQzN5_}Es`V5pZ5^aMY*#`vi01n=Bq
zm<;@J%>8eT&UOZ^eAB+4;{kIi!m0R<1#lVQ*?Q8ZW{$&L9d$mqNQguJ;)1$r>KBQw
zhc01{D0KIbVu_;)35h&DT7fL~^f{S|chhY^Z?N)8u&$FfTw1IEhULQOGanpeWH0&_
z<hu;rFQ%V?%_7c93&XsoGA^RDNei@aln0fy_rN9Rk<%%%&X=R5M6WW^y#AEw_(8W<
z`z5Q?EAp>(7kew8nF!m%{5oWOf~EJ+bQTyE2kEMrJ8@HX;hr3+8r>TK2J3hk3qZmw
zb1L(`QG5x#zBZiwRG1mNvDxHiUqo)==f0qEI)Rf%6UNmeY%a8*qg4qW$`ctKkuoLk
zdpO>w@C$g>3kxSmI9WT{zY={}<y*|8ZHzhNpG6+!60^!Be^-5s89;VF>$du^`)8+k
zk@S1W4sX%76?>~sWflaAx%fYP3~aJlcSGG98*TFt`QO*pCJN?|F6T}NwSVX3x;3={
zy_oDwjv{pd2Sz7SgIRVcJ|Z%gOh1$b;_19uXJ+ehBU?I5m&*%_3supFG1b+ebPu8V
zPtms9JKni%ysC}Hb&N;+4Ucor2ndS7;k844y5U5zmba`#9Zw<EwY7oAtmX!Os|v8D
zDeR3a>blIS+X%A{iEZMNd=A9-uYBJ!HZD4yD_sJa3i$6=)<W_*G^+Ws{QT_<U?7XZ
zM~$4^1JQ7Bx^?rjTgU?V_+2vwcI>W(T7e0B5d~z4_#zm*2sc1ap)U22LsnL2zY&5<
zs#Uu3`9D<L%+%<43mh7ojkQ;q>b?xCJq5?%UyW<tV5(|0=SN)hvDSZpO&s39Fm@v7
zgg+>UuJ}%)j{0+0efp>XUH?-yIXF@9jnEUlY1a($%pN#A$5E;q9Y#F5)+XL<*seqs
zlL!;1S|<W*50n=rh`VxTQ5an*5{bmxJi7049^I;W+*2MHta4k~J0yJbrbjyZEoFb3
zEY!8TrfwX!o0-g~0zuU^P|z2tO|rFCB8qy+h45R$zCW%%(r-?dY?X|HwGWbmzwng0
zU!)UUi8HMR9@Z*k9d{}Y<OsE1LfbRd9aLs#-)bx`&9+})c|Eor6h<P3;9xW;0!0)N
zxLV@U1t>k4Z28UyLvkkV=La3vd@F+ah@IquJV!`PHXY)*vO=lJTW<%p*4g2-+6e2^
zH${u;f|pUZts-+tCV#8MH9tpraY+2olKFYZ)F;Qc98~5cYa&sNsN*lsn8M#K#nGr#
z^;e>d4GKzkJVIL0ix7E8TEkoR-lye)LC48@fNjxubW#9U2cI^PRDm$HUXE0g46MI*
z*R6?|PuFh8(%1JiHfqFtWJlq5;8nhA)#NWDprHxmi`}#1{UduJ`Dl4zfr@5jc1}&v
zTw8?>v!_r0MVpwJ7P*0vE){5H7GfWo&m`Zp*bIa6JSsb`2^S!*MIxsn$Fx^LlN#_~
zkjyWunAJgCSME6*ZukeLe%XH@y7Ok^0*8syqB(MX<Ko5gDmZfFc?Hbu(<*X<ans|b
zD;=k64;nY!LKf#NeHPsLJ}LMGv6|O!R!4tsd&?Lqziv!fXOmZf{XHsp`hs^!boa^P
zx3YL*m7U#7i~%oqRhZ~xi{<AUV55Ql;jnDhU~?_yRRFEIf%@LX+v}O-bP>eFJ+A&g
zn6TAbA)2(njxS_goP{s=uX7<XcXod|<mRybSp90kQ4?;Fr1q48Czkf}3n9csT7-L{
zZ?V>KDkJ`2Z-crURy@}~@DdY6trYW0gqnQTVsKR7m;>_)lxrD#mT+4D+4tL46|VBh
z^6aZZ83hmOi~K4_PRJuYD-qe~vfjC0vVu1<40}SY-Q(yJRMzspw+*P{{1S^Y=!e0d
zqVta>K{v({QfeQ2G+x&hAcMrQ#jd^e?SE;1gsNv3!qBaxU)iqzI}UbZ=^1{yA7(vK
zbZic1@4OHs{UCAwofUxHujr?zC}V~5;hIw(_HxGWj%$V-tgL#(Z*NYwsQO7=w9IUQ
zy9)VSW8-AR(vzVY4&EPay6HNkk+kN;*uNzFhd}95Kb|ikU(~kC^k2zP_vjj4eK)wO
zQ+<$2*tnOW`e|4vL+ecpUB(16Sbr)a^qb+xQvUuy*R4K3v135?f48psm~(R`JUu=5
z7vqx0OUg~V#0XFm;b_$`VD2y?znk(p2*4X@TkS~XLXo4q4%_x}9mU-rp93#D^tk3i
zK-HX|Vz0KWk8(vrRfolm>E(0pj_vb=`1NrsrAEh+4D#FB3<fMecDFvsFHEE|0%;+r
z+>WbUdKE5J{1_l*D|6Kh>!?QmS}+xJ5ySQ9AxB`uMka^1nmFZu&kY<y&4wkIXzQy_
z)bQVFup8>(M9CP4HpVoDFS=Wh4Q7)wyB-`QVl?hLJ#Q9}yCaKvF2nI2HsSNc(+uf=
zylXj0r(9DQC~_<a5f)HeHvD?K>RkNMnOj2Bfqo5A3xJB%DJyGV3N?-zu~EouM&3?e
zz9Pdwow()n|DAckZ;D!Yl_L?l$35O*ZeN~HzTA@SyN&z(0xEA6NTm_j`jx#pml6o#
zFxx)cKCwCcI0cQ{i^4RZyA0d>7a8U@Js^`-MQMbb{K~`PF9(fKqJEeCg0I@KU6N4s
zCmJXoItzi1PSR0zB;UG#lDXRf5A=gdj641k*(@d#*qP}Hzz2Xn(7QR){$!)5r-9<O
zMTg8pj-)^Rm3jOt;o(`1ys9M2t9xjFA`K)>_~UKc<i|>3!hDKRx<XXJht-Db4T;d7
z?pZ2O`PWye1j=bdw$S3rgLoaS`v!@TK)_7E?Te9evM@8}Dqp$Cb!>*viL00Th^sWE
z@?zO1A0HHOC_cqtE~c?vR1V3>eJHv8V~Fltm8AMuEYP;d@xxBVH{s0u8xrz}%H(f$
z7H@#je$Fv-uoj-vO+fx+D76*Go(<?agTE?tMB}0pRd&HvCwD%gBarn4Wz&;b0R*x-
z*RB!0Yie^>QFKCzmf<W?`_5>J?+S99eVL{qn6?B0%g}dXLqm8OY^4LmWBbJI(hSJi
z7qC(>X`9{|uWb<NjSyMrI+*XTP7zh?3p|`CjFg7hi*fv*pi(4yOR`rJv_(wqlQc4f
z54Em6lQK&>ndX%9@P!ZM>V_*YH5>%aI+vr)+oq@<I}G-NV8ZKiSs^Hj){E>)D$g{{
z(baJ6oOJddPmMeBb1<VTlEtow&x4p`+#{Oa9%iWvW%S{T1cnoq$>luWE|_yfr(U1g
zyl|aLK~WbVcp{GCLs!tNbZf~r1^48UBUzcyRw1$|x89+&pniirK)}4o{B_*|lj_|V
zN1JH>TPv7uN_Xz4%hwWta|rGVnsgK8Tb>PDOSC`GOTR>eSeeT=!?P5@_Lg~&AH-Re
zer3+3iZA(sULNv3ZNu|dk!7oJ+r`GAK_t)T_jFu#8rCzpd(z^Cp0-?dZ&=X=rMqLi
zgQlzI@x$_XVF4cKSiogVat^*551hEK^KmlqzrSj?dzY^!O^fp-hyv7qb2~S5vDlQL
z9l|2YzNVsLg6Uf@r!pcb`O1{#_eVAgX_W2g__=>3E7Olg4A9oL3eg)B%Bappbo=9=
zgGCI+FqIYUqU$A(o!Ie+-v^rzjigH$9W2Irdk|GX6%bj`$-UCj7y}qA9_*?ETgITd
z<A~H(@2;XLq^aQ`SJsE`X7~%-oP<CXB|VL5q`hj}9h27qQyX?YjyY|%q_J>Z*E_cy
z?@+S7Ri^xrRQ{{{my!u%UB_TOORswc#RJ<HFW9=6i+56erdc>j#}S7%;ihDKL&w<w
z2JspRr~~zbtZc3_X&>t}?%h-2w3Z6i?Bx1OK9Ax5Y?*HD$D5@EQ?eb0E5ScA5JzcW
zjFHjK@6{2-))xFp$oj{CWu^Mzs4S)Nv}BdLkl%#_#%0Y5g3>)?&1oT1ub)L<mV-w}
zm^4RqD%gg1B;-UZlW_)~1swnb%%J^-0{B^n+{G(vzF}>xudS`I+4m;^^ufuFpYe5}
z6At}InK{RF^4GI|EXDXg4IGgxvvJ5(f#l$Rg}xwCTDWeC$p=*ziAPz<!1m_;{Ib|W
zf3yfZK%kdb$8#5B`>S?ll6QJ0wtfP1OgKLi4zY&?PN1xvh8beW8}~kbowRNQ@Wl3p
zfwc&Iqrp|LQ=xN@{Ky(Ke*=#|a1S75g;p=n@Eh?%HP^H+YIC#1pGc&C9iJ4-zrhW2
zsjzjjwp!<g;V7^dUNEy-oZu*wLnGl*9cau$2bozW!|eI7`4Xr`Vt5UX^FgFG^o~8;
zw;)6%x&TpGr1k9nJZ2dq4`B0&_r!F(t*x+kl*{BtzrA7)uvx6Oy2v#~tVS&=&zCpS
z6d_n)vZC3v0EP2A4G+8Tyq*?~=Xz?+y?w8~<~yUX?l7#m!Zbsl8Qr5|yuY+R-)r!Y
zBrE`-#Q9Usyt5SzT}fD7^Q)~COMq?!Huzb6Y05kj%4<?F7;<tJn2!2f(~hG1zix}6
zZ}edE`FKoyX@u>~g<hnbP5!ZWgoX}Xp_2*GJ_#4T5LHfiA+i~D1vP*i*EGpoH_u6B
zHW+zBL`_Nb2XbWYCJJ?PBU>9DySwyY3T3?&j8s<9pGAD4qk?FJyzg0p`Y*fBfYl(s
zJ|Ue0=%5nX{|NvuhzvBhte~IpxGPb1x-vUC<Rh6x?(H0Bsy80&jvQ>xuWsd*c-$-r
zDjW4jQSEJI%jO{G7U&#PUx2>EuV<RV=dz#1B)jr<lgpY6IP3B5*Y?mZ9>ci-bYW9_
zvXc=~mg##cbrc^NT;|FD*fP!X-duK{4E@!^IuvdQFy%`Hy)LBj5@3I2HdfzA#vvo6
zw^_b&E4YskNn{Md^-_Pi`L6|w!^>;3v&tn6L!bVx;!#i{Hb`Ho;*&=Q)k-bY*8W+I
zwZx+9{tdmGyb4RmoQ&&Rs*aWldy>8SjgRO`myxV-5{@#9e7Y0O+1(ERIyYDUp(FaF
zqX{P<pC%kp5K=HX1MdCVE!Myq`U+dcYFMS3Q#Rmx!@&01S}Y1&wC;hlt?=l`UHXpb
zc7M^A88H9AZ<0%MjY+uzJ_zTh8?qkBLsK5azk&8IW*j|8DY?x4ct7o;kuGLcC}m#a
zd44+IQ3Z{%^OKEtYbOc`I5zl8T8(OU12}=-hfsYGzL^^*X}{$ve|gP?nuZG6OjK(e
z0fFmS5s+ojoULP5dF|{f43G3^aLCAzGuBYh4Ir}6qyQTEZ?%feI%OjuK-L7Ia>ZKJ
zb}*$^p&G0(tHHbaE5S`#pybZ+3G@k{sdKA4XP$S4qYYIjN$cNk*MXnFop|8Up}a57
zhq+U2$%FHP4vV8MrjB#j5{c&(VFFLg)vUCVUMD?D$=4;hud||2>Lf)0?`6H##E1Dw
zV3h_xnFB6xJdu|_KUB^lAJd#E@q6|+<$Tzi-?f6mmgoB95~1=chSCnH|B)70j6X$M
ziWh_CWNX4QOkUSE2hhHaJy`-g--7c2LT~JO**RaS{%DNlpjZ~>;ul0J5@cgJo>KAl
zGS|lZ#urcBtlo;!GqPQcMH)hq*Pl#%y_NjY_G6y)a8I&lT_lm#zIzRchM0Jo`q;yN
z3MM?zlOYCpY_Qt{E`=9V&CF|7nj_%)xjhKk+b(fJ(hzQudqI#yeZc1Py)p_?TKKVO
zmG9L0v&>MHKYmyJ!7|FIBGEU{)ZoSS-QD%bgDK<K{n)rT_mRfa3Z!V_a<0Htxh_?N
zL`yW(^C#rbABgf7C!8fr!R?N<b{=6OH!iwuA&A<6p9}2yn3mF04Nvu~IPv(R;27*B
z{IL!{Od710&a1up24Jv8uX8%#>^#(1%(!^HcLz)VHp5~!rkAIeyR8y3v8AI(xsXp)
zbbMPOCQZT?bgnUjbBki5?xOv7a-52VL^XY>0kk}&$)R`IT9^>q*X~|TX0cI%(lVb?
zGtD1xVa(%=CYO!}2K{q<Uj&s*)I_(U!umTdu*(j6$|tA-s+NJ1<Bw)T+(9E~XXi^)
zop7KJ^dw>eti_CvGw5;;jG^1@#J_oc+U2Ewaj`6)3V4p!*EW1e#ydIbSVVSM==7fl
z5Pt5rRh@tDh&xTKwfaU{V9tq*LpU-;4mTfas}V4k?Ta%z-cW@jwzC5ot3D?uK&LX*
z2l>rROp(AncCuMF-=!-8ijL(MF@>lS+gITMJNxVF=U#AffcC<9!qu}hlv~39i#9vI
z+zbZr&>q9g+3kfU<JqDyMc1GqX+$Da<BCX@v!FIdy2-C^C$Z49nJdZY*U@>A#{+^L
zRC0@wB{h29zaM{#3hIbXZg<g3(6$m;hTh3hjRbtpl;<mmAUvlB!R|J5j#Q_)TSF-S
zj<$$5fsDT^A?^(*8hX^E<UK^egu9hfFRBkJa#7fdzy56dqww)6DcI!Lmy>j*(jBdS
zIA?;AzF=tvN#B&4Ta+{bGG$hAwvw<fh-s*;*O>cIhweCoqcl>cLr{0z@8Q3&laKO4
z0DIxvNyj|xWu-Otn?GWI#kWcBTXxZElb}k~^HKauab3(};ValQW3$itOz!DsoS)yB
z&N}0jIJtlLKb;%lxEC?jIu5h*)0&tFr`Iuy=J!_z$ziQKSb`yP@bB`n0yFmfOIMmw
zmdj$GAH7#l<#|)9Z3>@$#f;vIh^tUbz@-W+bAQS`MmHe6HE4g;d+#4>z!)6z94svE
z`?}nB`4{p%BSru21-K00wDf*Kg1dg_0e^J8VSXzfy-c_ozY&GZ=3OH7{b7XzZETxW
z6}uE{RkA}$h-%T5IFBznR@NLzfx^Ob69&AZM-pzEy!K<^@a|`Aa*&Eb1z1TSOS03*
z!=v-vEc{^LVB14ZPIk&rR;d5tDmdIy${>>mIvWPy@Qz5Y^M*x~*~I)RFdj=M&PM#V
zRGrcvxj6MgWrM+B+|1|~g^_kT7b~;1M!1EW>;CjkkH$=G#Qh53X^RQ_xEFE!3C8h>
zh85Pg_$#i)l)pgmuoL@lb^=f}P)n`MH`TlIb0#~v=GwS-ObjU=v?P<OM>>e_6Q$xz
z;env8HM=(Z?mn%_DSv}fQ+JjNo2rDW6N$IvOo_IR_7A{?`{OM(f7@)Cw72KeUkT{>
z@OO=iE)3!Qd7(r%ENpG~83u9?gGGxCId_#!frZ=nvx|=*=9bgI!efZw`)u!u{zHWY
zI)}J$6U_z+ir#&n${j`pBF1Ti2IAVWoweZDk8Ma~@O7uFLak?5)gvRd5y(MI*~Co`
z^+pGBN`Fy$w=Cp+&*3Xq3sFCNl(I1EoorN|K~!<eFH%Pi1aViOF-8DlZ~9RE19#en
zRz<JG&{U36E@uLuC+P9;%lu#Ajz}vNW;SBY$V}zKpws6RU!$CSP8G{KFf?am+;xT;
zfZDirT+$?Ran6u`TPDND9+qbE8#ByRd$iZS;x<Mb{ha;^{Be4aHd;rZwk``dpqL_2
z`Hd<(fni57=N~SxNNJ$c`!{xacT(mrx7^{>Bu?XG9syoU_a^ufVbC^e`~uU+ZYp*@
z0Xj`t$JO}-B-!daeO47obRTkUbmv%Jp0CM9I!@Mzb(@%LAKzNQhJGcC8_pv1H?H|F
z%HI5?YeP5dYiZ~M4+!q4fS4|vc;aBjBBN_(vpJmY?$0LcC_eC)Vq~3EQm6{31DS*t
zH|$%0z^72TIj+!<rwsuvxf6w?L%|8vj_HspSL@;`YztD_;rC_7l|WZI|Nd^Z!@;ft
z;BuW-7s+eyo9F6P8enlR;iSCzeoz>T-j7SZq1PR=Xoii7n!QAd=*O*3jH?Ro^gci%
z?HBF6hgFV*7ASVR26QmKzfnz2PMp4Y?_4eLc9>HXc3b1#jqryH^z*s`5Fana-QCrv
zwT7J~VCMv=#Ov8>3sc&`8`xX_;pl>@_C>$F2vRFbWp+Bu4^N1rUK4V9u(UdhTVtRu
zCQ~9JHQu<g>Qj2BO`tPDZ;9rZN->ChIrQ7c`J$^T%nJj|AUN^baS{7ABxooI1dfdl
zBgomg!yY#qi1_h2KZTDe(dhFsAY-4!+9K!Xde<;nU$&Q=Er%cRN&31|Mm&nu2ih}W
zT4wgU3)BC=wbuo{fTwnuUE6!WHY-xjNF!wZq&xEq>C&5fRjf{rT-NN$L|sOGuW*7V
z1orG4IkMw-*9_l1T=?0@MH~>MGTW!Mq(1nN`dio?m%(igxgyn|tE(HDp>ag&h@OeE
zC+Az9dr&oRBWa&Ynrelw&bkw8Cx+-;sY=K63Q&D5XQ-_IxT>}gSPKnhc|U*VK;_!G
z-05Z}bzpWaL~lw^()ZTITMAf2QMKdYH&J~pYPGc!?sjy}^7Zq>rbU3`2Q4jg#qf(g
zK0LxbW)eO>I!NE?-)+nW^_e>fLKhsR6yCK5n7RpCaOw#0oA64-It4ISTZ(ey=7Sok
z1GE-gxSOFUtoC8L|HFvu+M?yQeTo$0!%k3G*~JCuukPDgR;mepi5)u(Kka*hLe=`3
z@(FeAU|}7-(c<5-u_RC&nM{dc+u8ZksR7{2>)%osKQE%YI~mS7bCwy{+m8GR@fmB&
zdbZc1=)?E!%rW*v`(f?MG!;pc>v*U+945kdy!^g&{ROcLAPw**t|(Oa96O4r50xiU
z3J9CHK1B!$*Hct&nk;S<e~gTjsha<{H0tV+vliQV_^!_aFyyMMJfc+SLQO1QsKu?J
z&%dk^nI2sSt?jeuBi!Q3%0h7aw#-}p_%lQnINuXmExG}gfULC#FbB4COG-eckQ%U7
zJDrF<B>b2&iI<(D0bPt$HJg^4RsKIOK!(DQripl80w%a8GiH$w?isY#5=g|2SBYmL
zfU3Ugi~2Ik%4txMKz!g1p@HR9>|1bB399k5$>gU2_DB&Q&yzw;VIBCO+cUeYe2hN-
z6u4Bqqhk$r8&sZ3&*gKw(b0GNvOl|7H<H!^WrVXtMKgHxI5`eS5{UyYLmH6LfNmSu
zhnMZXC%R$#wL3;Hv>cqCY%cDG!L337J?YpmXSn%7IrZE`q-~$&1=XUH6?HEgciMdy
z6&;3p{w^>F*2qj|39$F&s2|Jj*-1wgAt9lv4mNm0ii?ZJ)pBSkPo$_E{OjGBI#}8V
zo8O`ire~I?y$XbMv~F+&GTLj}c*3aEP#(H^`{{07a|HK)-%$JvN@NH2;342(0$L~6
z8VHML_=GNuVJ?emALcc=eyBo|!q0Vm{ATzleJDmxkF$5Qg}<Jchf|MPq%@=*l-_nK
z5bsXySgbNf%-BQ~Hx<_Sfk3aApoC<u<C3KdF?F@=K}FbX?9Hk4pxhL1|BO&ItB`^a
zar;z9@p7o{vv|4V5}cXA!oZe{Oj=(y=J6rnlv35oX;p!$?%@8N@{bQ4W$#AckyU?}
zD1Qi(RVMy(kWaM6mHScIwZ)-fbx=OYr7~9|a61z;Mhy7@C&ui~<AvWaiW@E1?;)a7
z@!aC62%-L_pxq&ey!jwJ$FZ)XaPcZJaGWdO;f)v%SUaC+?$%kk36$4mU~Ll=4=t3|
zg7JR;9R}6gxOlGF!?UdU!R!N7sk5u3I9rA8NzU7&_A=$;etxo<$^DC4_F5_OUY{gA
zN3l^l&NsNNGM|&8x!dZF!*go#3~G7gA8^)h(=H6b%kkp+q1I$CbLH<*?zu*#K7sm5
zr^!7rD!e{{Lxcc1;9-3@O7Z7j$_9~DNpQCdZP)b*%4Kb#JJg6X^B<9lSOW<E=3ykT
zHSTKnkLVyIG#{KN#2XeL_x)fu2o#jSi-EWMwQy6h;U1XmLq2<M%f!-$`x1J)WA^77
z_+ewNLwq(ly<R@$6RuJ3-nj#NyDiRaLAbBA)(hkdw-A7E^E!LZm$vThlEV_na#S;^
zKkq!&tuM5ibx8h}xTMmbAWmnK?kjIS5xU<9KGy^6=wkwTEAtQFbL7~N>$g<(@z;pq
zhO&Yxm6#hFFJMlexD)#D8*%QbD$ROPBA;qBs%b5BN;nXw)ZII+;XqysiAKj3F)yAX
zH=#h;0?;Y3cW`i^wb!%CtJ?0EPg4FsaN~?qu&QjE;t>=|m$v?Td?+c)`jg-K)&P6N
z7y$G0B}PE>3oVO;e31{x5zwGJvzn_;Tw~9!;no9TW;a?iW|lSfHkZM)RBY77zXa}D
zKOgU5eJgtvo0~a!ylg^raY_1{g{KrdR3u=bETMQ^T+*{<5j_S_Z@b5^@Eg!?9DG-V
z58{j-rt*@ie(AhWO`dp*rx1qFy2o*mXOk>-mS1P_ww8&Oy*hg9PPXGpn4WD+^}Bz#
zb7#He;Uf0$SC<H0+}PUM1}>#@_e?k>U^stP9Brf5*T|z}y~;jCr;^MqR4ti5-U0Yh
z{WSkN!Z@;cuNs<?+NWs6W+%<`E}y|5q-E?D{h146lm5Dhg1LFM4T;Or9}?*j{jc>M
zTn=;1K>Q7&m^ca;9?0#AMQ^P{M3dMyx5bn%++C#ONbSiFlI4>kxs{rr(iH~$;Kz`6
zqq&x&dk1)A`LXCZSSYWiOJb7%u0TI*MMZBmw}LH+$5J_?d7%R0{FE{7&Lu6(<KpGM
zzDf%(sNT|$$so^%Ma7;kmd!9#O?}S)LRv9loq=LcgeBaYL%<<r4VaUC-O}9WZk(A;
z5dIil%<P^1EttSniyX2*m}Yg7@?FiwjsCE3XRnHW!iEs|x<v1<8|nk-KJXH4w~DlT
zgqf?cRpLuRe0J%<vv?q&l4Uf@+3n!q_W4(!YHNUQ=tx`a<`Rd-W5pVFI4O{8Nz4bD
z=QQjzUpv62xE6+3Bb$E^>RoB)>65z=EMR|Y;D5<{@+$Se3zm<wmF3M<55pHW?r;Sk
z;WqgRk)Am&aDqz-kWbF8>hDhxS}d+Af-Dj63wc!!%~l&AKfGv!o1-QMNh!2(Dx;vr
z@$-erp~eAo?m=t_-!Dy9c@v*uIP&tN>8c4C8SGMD8n=Y3*d6<TH&pfOZ{)3w_xJmI
zef^GB6=`d}E~&Z*|E!duTiup{j<o)MRl(Hyicp0-46Mop#>JCs#N@Tc1EqEz!|m2$
zNiVRXF!?;%{L;ULy&Xp_PrWGmMngvdE<86o>-)H0qu_BInm$kXz4fl1h-)z5FfS&$
z8plQN@9v`8a8f^Tinm1KW05aq?GrXbjDWZp(*4s~BUV)^#?s74p_C1WemP_0kke!6
zRTg-869sw*pd%qnZqO2TUPj0`cK3BH>~&Dk3~KTt$Aho6!09HJKEfH?yxyDYY*<ZU
z+15#!{}`ONM*q1-Ee)6&m?y0L9&#Ng=mv{^NOgpr(kV6Z`W<lnU*OwCBeC-zMUQ!l
z`3PD#MVcV&qt9!?v!w3h>4}$q2qz2k=pFVEY@z~$z?~L$!lGmTiWz+}=Bm@3JK`ed
zeO0b_HeT@Gx65C&HMtqQKW@*l10VViNehFtjqW57j?u3RUr*#*5W|nPo=fST34;&P
z1a}&8S!mR$0sP`#L>uV%Y*CMC{3DaMI4OQ1jFaase_3Ws!9m0QcdVy8rBXA6Nq;+^
zUs?ie|1z#_q`66mJl8mP<Z1^!$HlZ1(T9D*9TvIq{+o%=<qV%|{@T}l?!R;*x}G`n
zf^3&3n6eUl9@$~Aw;*Kfbs5W<*5ep5Y&BR<c}K;q*OqbN<zSAVH^YJgczrw-Oy!o1
z`%J<TuH>2gLfX~*Dhkb4{?x)o`Hv$9%@s-jNcUu(9@DgE>d}SGPND5A`4grviiY_Q
zubjxb@Bm!%$N6QQSeCvL{7(m-P}r8Qh)#YQ{7F)P64!Av2<vEum@YG|G*J*2x#<0)
zH>|=JUK-pO6SIZh-<@3_IanF!+Fj1265Y=}egcu_kvRmD5f+vi?V)o$Sc*_Im!$y~
zfn182!BK6ZusX1}yMx4otytz@!#Y^)x4DH)^C)08Mw?8kwvfM)L8(K)<g4$E4Sh&4
zE#=ZrQ_4trM3ebwsZw9q8yu!l8txHRhpU^WQrW=;*ggkiWG8T5W5zihK8Y7#bMi*q
z(v|zE%meAW^6`wy1M7rebKKdNfE679mQR$ela<PQJs|}+>_iqqRa)4)IP+7F;^d@G
zkE`}sxmTC1C(0t+bmQXMCe?_Hf0ySd+q3P+u6G^9jRV(6`se{{NPheVNzTz8q>icw
z1rg&yG0_F~{aed=#xDgI<v?H9+)U9JZv50>2a(}=TPje6|D%eg63<;Ax$RemoKg3Z
zn~k3xal#MBAsnc9a}9;$C}kI;lsb=};(h5|G(?JsR5ZOE;=S~%<I%aWL+r7-@eb;l
z$?Sx3D3mMlxI$J8<m^5COy^F>_dsHY%6Vn6Y;4ct!B!6N@HLBdAdTsDO)Uq)Aiax+
zrkmMpG)kR_|Gv3@cLTU^IUKW`#9_q;nS8fy2!WoI$5tRLgB|dBI}%eH9A<K!hjlG}
zxV3!L2oz1V#Ae_BaU1DOFL)%vL77ksV$HU{RA;p~bBEwN7g^yPM-H3G&52p?^H>B2
z9|}t@z3|A~w0q~s{=T`;@%)K05CCv?qtsD?Lq)m_QD9SGlMYnPb=LcPId}?%S}Q3y
z(n%m)!W|zTKrcXF)z&`u_m>^++Qo+6$aAw{`q=}t@$>)YaNoh+WUFu58w<6R7`f6O
zsB`GqR|zV|FbRwf-FK{4j}{2(J<C2iPlAqyArL-hB35QYik?O6*sujC#0hHx*&DL|
z2P>O!_Q$O(0h>)h10fv?JNC98C8lcwP8}-WpiHXFa&fkOZ&#apHPY_Uz*ODcpP$bZ
z!Vw0xcK<EJ<ttm6VkjsJP_9}YTl2IdBXg5&6$U5v269)8_j}b3mNq~!KbC&|PzqL)
z&onuLt6bjn7Vkd^zw4!tjjX}rn`a*eLlGj8V1686^mf!O`VBBB9X6kfgeO0It495H
zguQ=@8(^c`?7Yz%`?pJ9-}XOc@=HAYJ@l~Lw!h$I-Z!$s-3gFn;Xq2=<4YPLvWEYl
zL3UJJabvhY!1-ZnZ((n0pFPzF<d9Wl1hq@Sg=(LItp<a;H^$L*MJuCfe_6HX=JK$P
z3V_h>msZxp6X?3}c6+kU3#?f^ak+P}|KjEQJEnsUiwwK|=m1m{x5yIVjLpB|N3wDy
zpTp9^lKfZ^G4VzoBSb`W{P}v?`gTNgTnB>^2cnI$H0~A)Z3!LBZ>X!QAMExUcNUK<
z7!8qom;X)APw(|k>o9Lm4|f09xyhmN6lksSB|lbTW+yMOI1tIu0hPGe10&{q6gp~S
zKSg~+-J`gRQfnzzdhyD?<A$)_uN?ZJpChZTrbVbJghe`Ck1V_V;3{kAZFywVT!$;I
zBXXS5lMm@Y36xn5Vs*aOu2LS*cF~<LM`B*J@xqTVQz2QnwFJidRHbLb=WHhY0NVO_
zZ0C%dkjDD)CN?|7eM*fQ-Wap^`~$+QM}qP;oIQvXfOk%?3Fp|DyionY1WOIdM?X7K
zibe9X|KwpB9Nd~reZ}%x_D*rgCnXzeyzbM>855;}qmdptd#u%2rZEs7txpa37kvRL
zs*rfsM~dkf8p#y~J+$a^bkqglDV61u=%wjWu5hRS(h(8%g8YUQ-LpuGE)Fd&^n*p~
z*$%;*KoQozBS~r0G$kAJsHcO{zJNPR^P|&7oo_?Jk7!tgb3TmmRKHfvijIzHqtHik
z*Y;Smz~e6E$B=P?(4||&C#sty4qf-|n^`_Nh;YL{xBTp4{Xpoft~MZ|jNpLGJ$NML
ziQf3;Ul&j`mGqr%SZzB9`N%1#;aGgI%-nHK8kjZh>2afaN1aIhol+8?jQ~>x>NJof
zu(se+>=YfFl(ml1-JPaq%x!OPgQ}l4u6<2u(!P3zOrBYeLd+2d@?mo67N5M^KqYN4
z{B(bGu@Ah}J*N@y80p&IUatoB{tc+BPbg|uCts51Q;4emEh`1%;L&(;1j*gV5gA3M
zUr$SwgxOf9M2AG4;PH`{t@t@HP8XKGQebiN+qVG{&(i!n=!>A2PZ5I2l)01>B2A1#
zub_4=7>P`2`2-LstfcAkcFx^yd@$>V(%oSdO%bJYHHU%J*LC|_4q%WuerpHBEwZQf
zhk$*S+)6U_Qq7$wql~<~ycHe1E2*|t3N9@9g%Pu=kpb1S#3FFlZplV%L%yMKG;=+Z
z;jy<dkFr<2+nP1BbaDPT*3A_61o;~+58ZOsS-Grg2am=eUFfN3>zX29Z)N^}<L5@&
zwSF2<aVI(!*}1dS${tu9Sj}x5!4snkGWvVx^&C>Ui*T=`o<QgN$yNxt^pv(=ZP_!^
zmp3}r<MWthxgjZppduNiN<{z66-WJISui;hx4k<%%O0oKpN!QVUNrtEZ}K_yQi43Q
z$VS=fAwkc?!0SwX#@-gs)3qV|V^nhCd@CR$FngaxSDZ+XdD5eC4c6^`VetD|;UZyI
zI(30(6m2-vk5I{VM7^QZOEnL#{@dT%VX+Bn15JV%4qH1yY2B(Oll;R2>k&g~ZJP2?
z$u>VJ6D^?AQgz;F#>{Na?QnAx_pXrn%7tf6x~+=`S^3;!TzPs*URIcIwE_N%b3pA_
zfrr}m+Sc~$>K1#JK##y#cR%z{HZzsy?&*d~z$S+HC!sG#`JTa|z6zpG{A&It^&e|t
z__oODTa*q|)YrgQEsqd_+1ju5<+F>b>RA-?<?_O_0=zt8W<KQ%T8nB&3{@k8A|}l7
zoxX7UiNun-NrGn+hH`v4{QDYf3;5g<xYsWyXX7LOczQA9y2hP5VhTLBtgWD)l?{Hv
zkD6?GKZTKQlKjlRP9Se_E4E$>l|YnZ2VV0EGB&>zb3trm6U!FSN<fM9JF0S)DV|MR
z)^k`bs@7+!2xCG$*&XgV&)}1W0k*$jnWD3U9I?n^xao$>kZ>8lIa5al5DYD@f{DGH
zljOL*aoM{E>!czHr{Lqa|80Y<g8OT^xvc2_y5Tx}%1=~|h(^qK2Q^F~>$p<Gsnhf}
zn&&=yPs1!xwMTZ8er6P+YT7T-C!;V);!>6!w6oxzF0!60T3-nCn;`v3FN8kyg&{1J
zQb`8o9MNgqrcD-lt(O40>-CLNErGB@5P1$IguF<PdA;4#&71*mLl44;I!L0NN!6?3
zg&pH{;4(0;_Y)j~F=tXUk!~Emzl1OeAMhV8-BD9oxIY!cBbb_eNH~BSww^hCbZ=cl
zUjMO003!x)nW$C|DghJt9qv_H)QPV6xV@F#m6edrGlgwj<$%UWN(fTM+UkEhFg`vo
zTwSGP4y><64$j{AqA+H5SweXl6n(ceUesi3IgpE19~47-u7XwfoZV;vOtWbUrU+3T
zf|RauGy68_+Gs}$x|(*Wi21$OqS8odt`8Df-WWb&s!;y}>P6ATdr6ulj9tC<hJpiu
zSmAg(&ly0^KDH=S-m!FQkykNEGLW)P2&Z5BT@CKAV8_rmOG{tRKFRItk9Mt^^t)tc
zCbc%qQC6{2nbSBBe{;plG}n{gRM<Lw^leJNnW{chtL&r*)6tw_{8tbo`;faF@|5E>
z-;o}jP6!6V2eT}sOl*K7(Vc~XS`Uvv@VmD?0H}LRbr$O{>5Mq^lMyutFX?lZiot0K
zV^V+US7El?PDy=cVmewOzY`=9ZvQ+l7HcLpT`LZ<^3JqxF@`7|(PH6{Fq1dUZzRY<
z{)U8!oQcoQT4TgUO}3x77>BqfJajH`6ZdRy(IW=)8_L0?5xy!DQTbRG^38+oMqmS2
z3O0EUN+^Pm#O~DmFANmfr|J*I?)_Izr~Kz>p7LN9l&kKBr|H#;&ZbhV)Zo@2`pfWl
zbCW{aiM;X5qw70AU;MZriy2$rn+GSh%c+hGKf0^UY+fPhd7A(ooT9o={nciJ_3#Av
zC*+8JC7~hQg{SI@qvKULvN*l#0O*x-4tL3!jqneD+SKbdr}8wo5lsp#iI3ZwkpuR&
z*C-_ADwQ!rUbn*~M-hCzcueEO*Q~`bmc%<vYb^a;%%di#)Gu|@f6%n1h9T3Tg0+dl
zosPp!g@lHOo0b^UBo*GL94<tL^H6(lh>Hjrk$mY1hS%&m)C{{Y{p0rqKyZ4Lgd3oo
zz41lon3>Sx#4}6GK+}-HidVr@AQALZ?*|?9AM9luY%>q$yY|O{ouGr+fA*i<97ABk
z`rcMMH%a#(ojaSFlj^zJ8VX0a()X^h+=r||F54f1f<h>q2O|%+SLxx^ADBc&535v$
z<=r*BM(Zs&J{tW!h~5b>mdLBL|9O$9t7OqYREOFCHJ7*=KJNpO@A3XzO@2_^;(I_Y
z*$<01xchS>Wrj4f{GeDcG(K-d<qD@kb{KL{sDEbppT^CjJ#H<w`gT^fc*?m5Hx!*a
z0W#j3c*9XGDcQ<WA~)QHKx;oUwV|~yWDE_~lkTU<^;o=p0X!NqRpc~OO0pc(Lsq(n
zPt~o9|0rwl8+sQWVCSz(Kb~|~@J)|u8-J=g|LfjT_N$nnaQAEDow$MJ;Jm8a0q#Q|
zVjWST0sZ>m`+i^xc}?wMGA7I-ozu~?y<v|)*;YF;18L!ei70A{Jh1jIxX?=CxHYUa
zi4T`ilZ8NR0hQ)Aq>ST<?E;I08aUP<f1>k$J+innHpU)*s8nza3H}opH*cyntS>|r
z$qvkT8QJS=PwzdMnK*4C&#l+G&Twsg-)HrOo;#zc!3)v2=I&F)c(nmr;=A685;u{l
z+olk{z7>0(&Dn(dI>FO3K}WqgG!r|E71hc=erpMkNf%M9XxPa17UsMHfqdy+i;0w@
zB?T;AIo@Cb>_$&f85_pC)BR*nJIhvYRKH1|Je!@G_!~`P(`qLMviq%HMHJ2{+O(6R
zVOD01x=u2OMRh}iF%lQSK%M457?HA0asA1ixM`?>-n@zShJS{uRAX1?r&IODt8!@+
zV9;L3fki0{*tupS{-EB@Mm)z1GflwltP!-bgkSiI_#hmKt84H6`i~nd#<iXCDCshd
zvVIrn?{C0)hPdZ(05(5?kBu#meaFkmr*IqEgV0e@sf5^selu_*UeY>LzVljAOFqMw
z=P5!&!Luy7_>$xG#(+qKKr4>cuGRrim+#7hrG${xr9j)i!ebpSC(O*#61v4)-s%Rp
z*MaclBQVWSv%j=!sF^;pmwogEkwCOoeUxDKLCZUvm8{SAR9;x`u|~}t%Q<;~#gf{T
z)MTR$>Wts*Y(D7pEO~-N%JTgR)^mx1M68ZImsY)nyNY>QL7<LJ=)B3sA^_Rh8$QKQ
zlt1pQ`5jo=4Yk}v#Lm*b;UCZh)~KG_?HIX<?u_4FS()9MWyf}?73K%T(EaN=xV&0N
zQs(>RjD$D0*e%rXHb(O1Qt<BP=4O;?f2xqE9iK?=ASf||6$bB<txis&>Bsn6<yOK_
z4X<R4N@Nko4xOrLf*!E|qKa4CT1cF)qx+|P!niQiDJj9>tiYEpH$=2f^Ry=ld)cex
zKZ?N77^EOPHSVD5bwv2@#4}BPf3MQdXf|0EBae%h#mG{J-mVcAV-5~{W8S#yr{AGb
z&Tn8o`kcXHt#^Lx9z?o(QX5WL-M{W*J`+ffTdMd2Io5iT(}M2;pq|W)J+0N#dw*CH
zZSWQ!829)2q#<XIXpcaRu3!I^Jp0wRU;kKLtcLb}PRxYJGSIpuU(1fGie6EezFQfu
z|1!_TPqLHt#KA<9`-0g>v#i}!_s=!i<+YixJpFfn13S)s`QQv~(>08sUEkS`K45PU
zf&KL@kg^qxTwjVh07CYbLbjGdTn{_HO|mRX^wg*inBkVmcbH982~)-@7&utU|ALi%
zw0wHc&ri}4A@VSghtDi#BYe?<1#c-T&*8%$fBvpAnoxk-R*L_sO{6Dbs(p(=iobv4
zI0pLkA8cCkZotQ}QOi=y+o(Thv_-+XJ&r4sq=UVw1H|J<K(0aRyq<&w4eviqAqCRV
zZ!boi!as2~i@J|ehF{^Qs|{?0{i(<K8W)U-{<!Z>8%w{f?Oh&5WfT@1K4wY?T*^IK
z)s$0X?LHT6RPv`Ks;8NrQ6=RvZV0nXs2FC}T#iD`xmwTHCL_=POuv&L^Pu6<6G%mS
z%v^he8Gk01m2{UX7;avmjHkyPZ0=COLNgE+x2LCLTKw1z8q0INMZ@|<TTV@m<DIWJ
z__28$DkA1NequhC?iE<$L~SnI$#|RQ6A#qm!k?GCmj>>n`a-3@qFEg8yZK({*s<@)
zAgxtGU%+i=UfxdQ;Y(Ij(9jggO`?)lLfT?G=%7lAiwvSYKoSU@C3p2aPTjFzb@r1?
z&&^8I*0wq89nc}{{ouNViRousgze|5+N#Xxe{&1An=9K5O?S-XGd2UWT>o&NJW+|t
z-J`894*=|vE?|3Rh8gD%`3>9N+kBX8sZ&ax>WsfhMFsIHyquc_xp&3(%uDv#>Ti}b
zpv{n6M8umb5BQ0}k(SV>&;$e`Qf@?}d#uLOM_I~5`)-zZ1K1el6$I8<g~GOg9qlwV
zNht~m{=7*Ham(_vXEJv>_v$3^l`cM^p;;I3j{`dQcwNg+_^(Wlp4E`aukgzkXcexW
zg>=;UvvB^4BgW|T@rN_y|9;h$*-nteYA)Fiyw#;PmzgA#1t>+KFNw(LWqB9aT#zM8
zV9zUPht{Y{r!V!r;BdT!&`va_N6DrqJDtG@Pk(<7HFUzqu#(}(DbUD=3#tS5+JU(3
zu$`HWAdjv|_|JEPi;dMn>eVdAQ<oHutL`oazh`O|K=ViW274yI>nE0=gYlZv5=USg
zfB(ej+FAb8=;b#Tn_ifg&+(BWtT`s+i(W`EKkfCkn(H#x*Is0@i&+H}W}e(mGdXzy
zK9=xm|M(-!)QS6<8taF$cFKr5MgY)xu-@oErIKS$IbV^m5zIQ(E&6?WPO97Oun1>)
zTlkyk0$zm-*pbkt$d3_!^W`1=a1j0B-MP-i7=Ka$rzuhGWZ?cmD19HK0*N&y*Vx0|
z{Pvbs)w@@|r8fjE+*{b#@e_xyhRo$gHV!y{q0Q=$)`C1b?a67D)w6eGcL2g7E#BzN
z0P;nT<DX3zh8%I3p+B+~W;8VLcbbNtTx752(wOQS{jU^0LT}6K?MB~uV(lVXE<AKq
zt@=C1SuI%Vf1;C$yziyoE2e%%s;l-m-r(lM`TMsUMTFm;?%BQLek3}<EANsy4|i}u
z0ZCNk3B>CT^Itypgi0lpTnfykYu$EN7QeO{Nb8idGkE0if@xJDp7IrgX(q_HI*(xf
z)wjGO*>=4HzOV1i<bnjTsW9X1h~H!JduxZZ^Ucv>hsz2d3ElK>N4*w`%TByyG}6uo
zVmN%~V+>sVGc{hUENri!<32$Hp5=f&g_QPl17xeuyDQ|a{Rc6Br+|2{lo0&3?QU+J
z<`U|dTrH^<lFZlF>g<T4%L*y)&eKQKb6MUu=i0KZ-BO?4sZc)r;`7;7^AEus_ku<$
zF|~!7tMhM#kK_>>eAM#$HshR&9(sI)f<}jf?b*9>sjs?zJcKaGDi_`Pw2Dul5rgsx
zuBWhOj!wFE-HMz;KY)9k3x?*xBKesD|6S3ErZUE*Gb>(9SXrz5&F{VK)i9+}q0;As
z8_;=Sb^o<{k80EDwZK>8u$$s<=#xu)UR(EI(El?HB=<EJOMaKM6w5xZUi~?r!U5U`
zf`9vYfk34xl~it%{7SKb$N7lYReaF}C%m+DZBAf=BY}aqAlbg`5+g&|NH%Z$PCLC5
z$GZI#)0VTL&McX7zd1s7%p^-DK@{aV4Zm<HB*OAui`k!Wk_dp8CJ8T$wYT@(xRf5#
zQ6RVH67k5=T)4;au)O0-rRKT?I%z|p?Cg=_6KJ@@?;LtL>IPzY)!`Xezv05#h>ZAL
zq{?3F0Fc3ZW;D|Nz;a*jZ^~3qf4@Ue7lUwc^fD&d>hSX)zn9w#j#y=R?Ml+_nMs|_
zgdc)wsh0RW(wkVA!w<s;FWE;~L&qsssR64?=Z|*Zt^ASF@Jr71L(*f}7=y_swQAN8
z`4?J4p_t>aV<Mt&UUQv$mRji?z+SB$>Dt{L7$5<Wi$J@yMNlpSMWqZ4x{u@zv<zN~
zq7mA_Mg~TDfH2Z8-%lwyu7k)PLm!!2aX+6WoR}0|%e52`ZL;Z$b!@Dgm>!iOJwidu
zKK#D^BZ_!-BOqw7$NSu3{}_R72~qYzJig0LOBJ~ltn<OuVo0%DbcpLqVkV!4H1v`L
z1o(QN<_^_-MQIsyT774HEj(D;3wdTh>pJ4{b%9atQqJPnhx03gVWyn@Mch+eMV)l}
zjBr{;X9GLlc-?R!tD{$*%Vn`PrBKTh#uI6M5-xZW?o^YfUzB8q#0-+(<)sw3P?v4G
z_E?eRz(A<Hn_tc9+o(ICOy7NU%<d`?$mlv)u8g#}1Nm1LjW)_Q_j{3k8`u#y4Qnf^
z`%0xoGJW%{r>&(XWcvdKYKEC)mtiWIL<JWI)hW(JSs|)bpnz2KOn?^e39tQ+ao$L_
z;3V0NrBmEfNr9UQS7w(mx|%fhN?q8-`|_ennOv6;A|h!!3h*&4JJsihx`DEotG~X;
z5a&><y`>w8@qMmL=(iYI9+H>tVDG>Lv(2vB+x&A{)hmgTe2y%Q%?$5`a8u54f?UeE
z<BM7DIOs)FFYymo!lQ~0vNEG~^PKne!lu~wZi?SRr+1bdG3UNhUOJ=ajL&hs^kJ2g
zPvt>Qk7Is5%&<o>`Sg_sk(NLJ&R8~JYCT{r4j(Wc<Pp>#?>x7f0`lFX4?>LWR2{;f
z2`IfPth=E5)eIWEl$v@npSN2P)gNwLEAOLqu{FACP*j<-{B9r)7x!A^k;o*u_^)Z$
zDXSl4vJkq+Af0~Oq>T3^2&z-cQtc^jsb_vT3y0j}4_I-zoG`unV;Z<9@kLl7$X)_g
z+qtlP?y#_5Zp+tHE>q`p1}aFSo~84d=S8dVVSz#eFV19Ne-fz>UB$%>7k-8L>e~~Y
zbE+S7g}YkMrnwG84aiT+wWkBBH}&tGr9Wo`rW~@m&;v)_zh!EM@Ek=6eIPp@&h;*6
zWIn}YJ2tATTOLO|J^^qiU^Lttnb8eqKc)wCMpk#f$+FrD`Ww4CNgZ)JEA#8Kt8pk2
zEiVLF7lY|G>p285<Uy<7nGBX}zH9a`A?j58K|lk>_Nhs-k08#NaGQ?uH&pbEx)*%!
z3kIXnS!h#on81|ye>9zYJk$OA$2W$N*pQoYn3>~E=1@5w=af^Sl2gt(<dE}ul*2@Z
zgdvBVqMXVh<Ph$NAvwj&AxtbAIm{`3`~LC!*I%>G9-sI1zOL8xd`@y9pT=WuWjqUj
zeodI}NV+<qbjO+hxwJo8b@M%}_kU~$?wwmvO1?$APvYS5LiYnF_1MAuURMt(?Z3D5
zHAGw&(+PywPJ9Ub!qO10cPNZx{me=5uM{)tXLWW&etKDA@>4@`Ivm!hdg3?b_NJH0
zrOXhzE(OuMA9)cFBJqLZyivV3u<n0;{CBuJ_9VC|U6esS?-SZ$(J);t3HguI`Yyjh
z>ZI+%t1sDDD(RpHuCMmwjkP~#efyDJjyW?{FRiIjo0?tDWu&|@?z?Hb_sG&OUNVVJ
z5Mf?9(rIZr{OfI_Ru$d!D?x@{ho=X{_*C8FqXYwnC-X_lBbfVZ%NOMlenc9Hm=uZd
z2@tQ*37T|cVntt=^=%i<&4uY8TC^=pd(e!BK+ztVK00W3QUM4R{j9ED2z_=-#O-o-
z@oDVyKZDC}45$m&I@)s(qw=a>B}T6)_7S6~oy`q{wuvbf>?n6l|A=v4p3=Bm^Tr)y
zpzqVIOj?Ivfj@TTv0aG&%K-WXK(1;yqD`jxJ4C`z8;x*+yO*=%4U9W9bMxVw;SB-R
zKZ*DlLYSOAzX^2fSWs3){QGLA_X1E72T^%3|9H<`M7D=vB?@po+TJ{GsqpJgldEli
znEL4MdFs=_19feVzRy9HdaOH2xhslzP7Cb_@z9a_MS2X9;@`TpMoeK+$6I54qaNAa
zuz*XU6&0YY0QV==B2Lc6x?x0`!yU6?4mmi~S(~p@UE7UUD#XzoYHnKV!;*Not9pdh
z)KV(5h|)ZF`5)**6!@H<uhPNf)tBgFWmWLhZZB$B{hXe+$xlJK4qoQ(r1dJij3^>2
z%vd57pMmA0`u|t%0Juy#JAKc0wYt%C*=7q>V<qb1affYcQnsLH&($xIPctQ;ii>zl
z4Z{L)mdC7s9>vL5`W(A`Ehp{!(|7>@pVv3ZVu_O-$XJ?$E=$1b7*H?^2rfmA$yo8<
zIDX{N>5I3oK~BsaFAx<(t$uC5EnN2P3u0p)b<A4ex#c7kLbaje9dNxTx4w|k`#eUN
zkG@e_bFKcZ!|n`2?`^B!7tTJiBS|M@+UFrNq1@Fsp;CT60GRfS<_^4qgSzB&a<$oV
zH5BccHyKFZD;FYkwN{dhlb1WUx?||&%o;s*q(Lm^$f)Y_haAO2ylZoJkBDRwAFiAX
zhQWBg1h(oFX3Msmqxbd?>d|sk^HfvamzFclwXL2VlhXEC-~JrMz9Ttk#2bVL<%@qn
z-k{lw8^Cng$rYpP?q_d*z6AJH&P7o(7zdN|?%&`Th&vPrG!dzm)PH&)gzaPK5s0FZ
z*o^o~bU4&kld+`0MJ<85>lF|+Png8_u;_~@ZE6l!h)%c`GG#IbpuaN0v-lH?`ItVR
zX?VZ{`a2}L#irhe&14no^s40O0{SQ_;`%jyk*HpEvkG&NyY|;e|GBA21=h^Wa@vo3
zi^2<G-xzDh&y=TIeLLz4cPQ$`7Q6d_`Bm_=>StaQoecS3ohg8g4SKYFr6)kA#F%Yd
z$t<VDfzDccSU41j&i~GGbx{4YXa+b_w3eVeAo`g`iy8#}x;`Agex=?J1I~uhot#h0
zsor2vhLO^AsD;Ef%*$<+7SxAf0wTb$@$eE55dL0N=e1h&QTy4>4q;$yaiLSII<q5m
zqLuLXARJ-*$|A%A?@WG;vG$pM$!(Gu;9I+OGHrKn*A!j&&}Dhhv$J^`CnbV&cQj|5
zj}$OG47WIrA3hw&Y3x@o7SH^N>#X*=@b7eb?C<4FxTV{&SB<-LbX(!^xKOJ<`t*zt
z-Nt;2;bN~e;P=47+=A-ckqK+aF1fW8F9Q!LLTflEY#g-29;sWZXb2`=fFzSMW+mbb
zH=UH8TcKO@;n_EXaDb99O5~f63l9x=xhhRtF{E^%k^D9+vxYpkB{X8;S0-d2wKrTh
zjbA*idA(#%0RT0F2Q9Zf%VyBSXjj-ZL_+G5eMj<D7`Lqf)zv;9I%mtgXiho0N|%aE
zoVOe2#uB!Twr?I%KH(>nE}&2MLe%C~wn96K(E4s|gZ|ybtN*lh@n~8MN7La|qQWn-
zhmSr9cZNho&5M>jo0LW<jqEB-NeDT*y}71gmE*;WcwI-zOj&}+{_&lnOQ6Dk-zM!^
zXmz6JDRg5d?r9(<djrRcq5t8WvTys=%~V4pjZ2b?@mH|QhtZnrV;6E~t+D&DpIL9A
z<1nWgzcYt8GB|hv|8dWacSIs!yEbFI&g7WiB?l{zC-4si1^%_o!X2CED;5mD2bL>w
zNV6cRUQ-=@{oc+nrY!BqCoi}Ke?aVdrv<?yrT4#0h+h!e0B}H3Iv7f0?CfwkOyM)?
zkxKie@<3Fwr#J6%HKUT5OdbzM4appxc$+W5QrR7%R>U$E&&LZ^8VSwwu3@gIBG|k`
zGziNAWf`(E$^|t(gD;r<%*-3Ku0d9`PJ)q77nU10w-$t2{aCaIl;3FxwbWnwz2E-H
zkxoqggXio2_?er7uNr8;)zD+~;bQ)yceSFc_TkxT5HBQf#C2&7Sh}y7H$_5kBR0cP
z(rKR}(q0G;{#x%4UETgSjpae_t#jL%9|tkF{u2xW(CxArth;>kKD;S=BHRKO-;3Ob
z58fuZz!BHSDAnHB<nx`KE#dRsoyb?3pR`^w@TRA{tU`=?bk@^z-xa)4)BB!t^PlOK
z6Au>`;Tqp-@T-n2{(jqEh=MF=iTyU51xa^s@gKMIjrD=T#)@fGvkIP>(qAFei@kh6
z{?rB|V0mutu!2S6?#_Pd+vn3yLOU1z)`u@a*CX0#9?Jqa<WEa`((Fizm!mm=4kcLh
zTkKj8rpZ3~q{pc1fdGL##DN?)#Ujf!LSC-ar;9c47FX3kH&vIwM)Wj!(YF!qkWU>h
z4B`+MDR^Jb^{&;Y=}%Gppv=Wr(H53W08Qv63ksLw1rsiV@YXnrd4tzGViURF4eu?)
z+HoAYN8$%FK^4CVM)M37^IsJQA@-cURAl4o6qjcR$8?uFK@Vl)2+WeKWZOuS$}>d(
zmW1G3fV1-Uw)?r3XtK4>tM)6pmRX_-T*l0J#3{&8u?(XxaGB=Ny@1`?u5+$)eVUD|
zOl7QDlWD^5Cex^-wu6>}haeH-SET=8;KZdpwKsl!9T>m+txUJi_(QP7H;vlcm$5b}
z5n)>F`TAKU)?YB-nYw`@zm<^Ft?4LoC*fPm-;Sw2=)W(}UX>f^xYJ1vWgrsjmZ#<!
zw&XWGGDTE)fiRXCmz~lR+}@ZN8He2LA2jXZJ@oaRcn7lz3gpSMtuf<={nEiGQCvky
zf;dPDq-Kf2-w&!-%CfxOLeeO*r+^S{pzn3qlR30)6W2%8|90<SbVoXvU}9MjEu`gV
zUpL0;e9@N@Rs+8Ibq<U_THN1JreIw=sV8%D3nvTW+g_Qvr<8X4u?g!4)Aq?%26j@4
zr`5TJL#0bMKcsydtd#<~C2ZS#udFeI03fAKNe{_$vNGfG#ZQ6hRVqj>QPx1~B|{oJ
zkTsrJ8<?znkv`|duH-%AAt9Vl05_?AgH`x1G;4~*zkcFQ%;~kPPA+f*=e(Ozc%#Y5
z0nRDe#f53~>B__jO`<e9#^K|(f}ANGf}sE*I|(rAxEXxhgaBUl4&E|Ji4ebz%7I&i
zmKxr3CG{dK6)z((VA(49+~eV2F9BAv?vk9SPZ*DIf6y&dk!TpnG&&-pt>~Ybm5-_D
z?DoQ|y+28vVNw;-UXGQ%Wf-?iRkPs@YqA@#OkE8Ax4~DVE2gtErn8pz_kQiRytlc4
zg{=<E;58~3yYed}-y4_51jw*t8!_2%P9}Yu<rKFBAq?2Xtx29_=o6k&d9gDZfHMOF
zFnAySXgygG!jc1qsVCjdz}&3WnMB+9*HxAl`d~Y&57PB(8frXy7JTW5n6b%u?L9sA
zLucLBJW^?>rMnXen@0){)+2FMziRynPgk4lpG{g*qBenZ-=$#<`LipHh2|A(tectB
z9d>ff>33q2^brgS^!FqNc-`n2B_qc8S+=9K_P>Q7wMPWPakv{mvs(?hik9Y=&)uk^
z)z{CMlrJvW`s94B_!De<mA*fwv^Zlp<9Qh>`~HK0T|<!n*JP-E@-LNXu=ZSxMtbe+
zxUZ_iP#Oj-vT$-1-eAQ1X0F|jvob`BO@zelu~+=ytJWTCn5l%3sY$9f1L_GFfc@og
z(NwnGD{KY~Gmz?eUBe48LL=erbVn+q9XMl<ZLeh`+EOLnd71U0%({&F!0H>0z6t>i
zm+hloDef%x?=tTh9#0QdJTuK5YrSN9=n2tE_I}g3=pXre#u4#VzyX}c3QI2&9=XK^
ziP9$El{=qKL*OETKqIB+A<e!sAwIN$E?-|e>{esFf0)lJNL^JVy2w1}^RE3{q^>}T
z5c9H%crE`jhkTMrU6Diae;hQ=Oi%_tchC(g$?%@rA5-|7gbn(3gdv~;$(5RI`CWFT
zw|n9*<eTW1@uD;JD3NsJ6;bA_Dq*vAZ!1<FgWl4|?zc0+0?H8pfPi1``Y7VDdIg{K
zuzTGd5I_>B0C8?b|Cl})pguS|Tnkj0+V!$EAGS93QEzcPoXvEuj2|g61qDns)8Nrv
zbj#FFc*Pa?MdsY7NH+vq%AnGe)I9%3bA>T056|;P${_Z+-!shH_6-HOpxSJt(zI*M
z1Xk46qUZdG8ml2&pKIl;BDWs9SJ_!`k?S0mGGnWM7@UuOu-F+D5h3>a{On{A>ssGD
zI`K|csC8c$eZD@hs8x%dy40(vF>(=Us0O{y)!2r0w?JfD(m%^kGCvHTy0>Tf_<qgP
zhnn3!{r&i(P97+fuDW$y(?dos0svzd<p5k2*kF0}7etUQD$dWq2Im!Q+N+VF)OLlE
zjBcaGpckdG?WJ91^~>Mi!XY)j2lG76eqWUhhZX_!yOQgAdtOMTt{z!mm|W?Kj<#E|
z!4e1^!;8}9B|G6{D_lL@#T9m80%KTRF&U^@{3|diVVbC6z5f{fziqR=WX=T7TCW>}
zRW}46L)PSoqZXn}fvP{Di+Tk@QgnX0SgnHwK<abX(C0I98!?ONq3lOauRHQ%&W<b2
z$0^t;4gRC|*%6h-d}W3-j*mWUs^xF*!|c8}H*Ix>3@7}SO#x`<;N<nxL=qmVwd1hu
zUP+1otu{{dSC&4cz;~iyd6+3<Qb+ovX>o~V14geoSpq)emBR&vdeuyKcm>h$AoObh
z1B!7@P&hwzd1FtNRSsA6&laz1a}#yzEIj(J#|AGnBhxs$5J3^*ux{k3nK-5NS=hW)
zKb{uNQvcK(S|lmhIk2+TvfWl_k|4o|Xg4gaE*d38EFKKBH|wR8V<_{G4AxXIB7bfC
zk=9<LkUwod7X64E`cCbQg{fQ>`lv5Nb4=Y>LA?CADF?KnB=id5R@eX@fFJZ*C^(ot
z<@YWH{OKuLXm74FyY!U1TKQvC`&)?&m^>uaiS-?8e$tsKS_}r?&R}`Iw#ml`_26<6
zPb<*WDaVEF!<jO^yyf};>DZ_2oSJdm1as8#h%p3u$Ofv&%B|lA6Cd%R_2YU(6kQy!
zi|F&fPScs(wSA&VHIsJVrCqP3lx<h7`FGQ*FJIRZ{cngywmHhq%S2KwrJXAEYn7(f
znm!>vV363pT6s~CHxbmWWUd@7<#rJUHh?62BYmt>QpDw_WSf|RKe>!b-r*BU@*~cT
z%g8`AV@_Rp_!hd)8w%09Qq9+5qN0XP!>&9(rN;>Tuz%@U(4Do38CPGxAB{`b9m|r`
zxRM;W@)F%I3zq9z+8|+>ajBNF<&KlzdYvT=GCkF?J4YTWU^&sNY8>G0`Om$@C8=6+
zQrdKnWmc6Tw_AAyEQ(6;1CD$Mo0|OlD^DdiP-SF&!VCA0j1@ZavXh(M@{^<WHI<J)
zSdIDXVi5UxPrk8PioH%er^(GsSkCa)46F>#Os+(~QY;;k{~D^AvTfRHdaK+tL?O%x
zDy1{6(MpI}42>(NDrDntO-bwL5d~AYX9ZgOgbFAlQyO1x^As6Y>DY+l*FM?Htxu?H
zP{=KQe{ZGURv1PT{ubO{q&rtHTdBxi7(rkXODwr;A_YA1c$aJH;`8V25B_2XtEaye
znM&fl_;{yFjSSjx)%kYfNcrc&Q=Ik#S_e90NX{bTEJo6vi{I$hrd3IPb)`ddK{~_1
zM({IgZe|M|cHQ#j<Bf8|?||wo@#ZCtK!>ZlO3(=xO>s0=(YDRKtMko_C^dL=u-li9
zFno`zO$7>6KNFObi1o~ojNT*3i-5%iebg!gxpXYJQ0903S>9o@)1AfeD3--j#)~Ju
zLQi(A$6~0okPaf1cG9`UlGj-@R^@J6smglK;B&NU1-|KqiD~xs=T{6D9wK7D_j1E<
z8Qg%(Zggvo!M(aLBDSdia*rj5K~bT!0hWu>{eVUr4lJDE-dtInpoWY;99fAH*X&^d
zU;FS@urt>ZE~40;tj{s~>+l9-wMRzX%+DCWG~^|{9%1V(f}y~`V}k$28wweo|2aB4
zA|6#0JEG6pIIRdv*PKxzy^Z&-T`8>;Y!vM1K+~qTs$1oAR6TlDYzwNq=Iw#eHIF>J
zS#C3dmxYIgPnUifumL%qZvT?$h^h@SO?VvVo_E9dWIWxF=XF|Borf&c=x}Rzrt_Lh
zn|<9%_eUbdV{E4qlkkw-NqvmA)B3x!an^F+2hY<!3VyaUcs-&rX#mo26X)YD;<hxD
zh~{R175UllwipEfFw_*maOQ@<J*2an-2i$1W5S<BMuBLGTk%%kxJ!ku4PRmge`01z
z)3s}@ag~iBYLnnU*5rMAtxxu&!<OD5*|%oE6OIWFmwP4ggX~I|#Z*qq@J*&qAOM4u
z>;~}p@$xBU|9ETsY=4LDl#O`scJ>5}?m@I3?$6M2gP@sk8xvVeTRTQTV)alYdY&2<
zCzGhC6U9i%G=jau|6Y#!74P%Gzm@*B6Hac|I+rdQzYB;%fJMSP9p%=!-2_e^|Bj+w
zHX6)^Vh2J61|AF4XwJCvRu=N>NwVpQ!~uhDY}K5Uj+C%?VGgdEHfNVU6b(RwFOu25
z+QVU{W(E1~@UdY}{Zse6pCAnd*>^Xl65P1*Jz&P>ZleMz0Dp%mCB>BL>HL1zzCM(u
z8Q#sq>{{nZ-e0I_?rOeeBYr>ll|^crFpIWFWYT07xKpDMIcImQOlLW~sg#mQMP}YF
zu3WPOdANmy`TM!v=7{%g{fizC<6<^3cxc%yWAp+!lds@vHt6>_SS`KurovVoWpe{K
zFh1=Y^^O)}CvvJjHDO;@<$l!_c-efpu55RT0M#DMbmmt{+8FG|XCS%o&DwE3y3pzh
zYux}o6MIGKM!u0tq9u&U*eFj72=oPu1*VY=x-=BIJP~Bfm$1_@^hq=q0>RCskE%Hy
zw?^s16*-!hPezviyIg(60Oi8$;bsaf<QucPUt|4+9R_~FH<(iDes_f6OAmUT?UT2*
ze3?K}GQDR_9QzS;A65kz`-%G_;b`8@YO9EGCJ9e36>46Djs-gpDTRg1*edH{U75Dy
zo6p~R>;Vn~RqD}%uC0^5^e1`}D^T0~zwN}pHc%XP(5IY@d(ybS*~dpn)uMUS%E80s
zKCKT9`K8=$=SJ7Q(*%)FdOCY!i`@D@z&DMsskw>i3xw#%56NfBvk!r!Os!-s6Penc
znSSRC!`}WDW-%2Yd<hzPYz!Fi5AA12Fai&JK8<)0yZjH@H0U{qP8=Fx{#W_W+S(tA
zQYSw8vQUn|l_Q8zyd-|udNjx#gjlr&_jVLW>B6BOLp21Pf4<mv%cQ$aV(L!T+(H+z
z2OAUrqzXJz*g80V|H+;n4^zZCn;1z_oH-)8so3Vut_~w-?WQ{8qU8!6I3@t_%b_5G
zcGS4(x`h@!@YyuVlIKOJd}K=8peH+qU5TyJSPzFgm6p%4jpu`G+Du1gduz7}W>UCV
zpJ4LxJa4<q9ENri?Sc*A&hWVJlXSJlz}toKgCDd`FN7emsTIUijmMPr?dAUWOXtRi
z*rt^(KcV|5I#St76B-w`hq(5m9oz0xI}ykbf)4-fe-aBDSCF%+-3|2X^dE>PCEOzS
zyZ^LYcW<L7o&PMalTV6sB3$|(s(-8XDz1v`Ec{!=iMPJ6MVPkZ+&66iUqkjiUKnzH
znxbPvvh~*IWcSCB$8VRpLBkf)jT;nLMa4ZsYCq2tpRyy`s`qObsRUOo3SlSo0k*CW
z9i^#^bs!)P1B=cNme+_8nihyWkIl`rOhzw(9996d?g;=y+NkVLLuucx$uRm&6jjvC
z8={F^7lCDp&()a;wv@4r^aeQ}7$N(qkuRx>4?zGv!aO^}m&Y8UmkB?pr7FGhwK2@Q
zlhGQsQcnPzrBL1@^g9xgG43{I{R#TR62b=HVGyY~okSz<@;3--;FVJMA4eYro~&Rk
zyy*f>4A^KRfF6VCIf}L=ZTtueV;mel`aS%*rs#IcD1123Hpe}`GKP@n>F5+AaMB{Q
zC8<9d>0{`|hLt>ckW<D&Yw@SG7U;7B=o?+`FrUb=$3sG-x7(FTtP={|=&mSz2JJL<
z#e%AheWHct_l6oc23*18A#M462aAq5TR%TNSUewF6s8e1;`$Z;c6>UV^&9D5@0;g-
zk@qp6I`Kl3Q>%KuknFS>P0y_FT%ebou>SF}D4+y<X?aMglK<$p;pgr}QW3gL`9OUo
zoQXf})`hV<-cZmlx>Fg8t{i!Uw8njz6@H5u^c1fwa3vwE%|9UTI=_BM9#v?sxvypR
zk~I<b>0vm0uTIJa0F?YL;#{{Doqyu`_h6qeelmRaXSq+!fd6xflKS;EUEb@D=6Sn>
ziPvf+63FAN1XtXq`b$%P4OoYSZJ<@S9iiQp%AdX>Z#)QOb!}!s-%cD2lI?|#-)ZgK
zs=X2NR69P!t!NDLW+ZACAAE1BAMyFwhL9(KzwO$2%Vw@&x3FLQo#?ttDb)p&85jyL
zAxQ?K1O$E!mlds4{pwY1^g9!sDIfsTYn@I%Y3U@=FmEb&Zhn!$#d}DN1W<uC-mlj0
z(<qsAFACO`Z)g-<^#WsGSAR0jh=gCwHNlY-MrLxuw9dEug=5c-g4?W}huC!!()c{O
z1jN2Z)Ks~55xa<fSXaUX944!yG4<ptkf;30O}*dNPA;YGi5s07$FZiLWY<Ef*WIPl
z5>n*jzJLLyX+|||l)o*45B$l#8VggE^uP}Q1~5kHSfi)^ZZ!&Sk|<D+NVaZ-svK2r
zIC(f)S`xA`+mUod+s&p@1hjjYB?7H92-`)*au^vHGI%e?!HbkBRGg{urId)+#{u5h
zO2ZpQ9{sv*Brr^OG0Maka4Hciurzig$o2e4wKvT2V$Y!~LrF4IoW8-mR1`zlM5Z(R
z8T5t_TK*HjxG*pG<_my6Qrr#IW8?nZD9i(Z6cKk(6&Mks|EDMa{Y0soF=fKPPtG-p
z0=VcG2!W_@_KV9-YRoY8wpTdkjLOZ{)-K_J$4zuEOtF3ge?)$6wYgUC$T?r7O%y?x
z-tH~4EYG$C0aFHvwrUwF0DQDH=1x@YelS{t{#AG9Z+7{&61IL|ch}acIO$O4gBHqL
zTkcY?EO(p_{Y2CFeHMED_rd?#?E|5EAe|@3bH(4_Lmu1oULM9t%bT);rWNAOR)m7u
zsKrpbdm2f_fHS&|L5utcqcoUls(0fTP6(`=<+$U2$%~;!QB3hiD7eSes&-}=-3y|`
zHl6*L8@JZL$E-#a$uW(I55MSso_5DZT!<>StSbJ5G~L`5nqQI5oqh$V?Ly{_7m_r{
z+0aQm3UmymYjzujP8L*QA_2^{G*`0Kn&s@o?TkG2`3dN5c#|dHXPWQxUG`j@B>X!g
zm*i3K-s$O4BXyG*3gK7=SjeDJT%b~m-t~$81d?mA2eB6(sVM!i-=4f`#p{JX{(Wq~
zrR>~ZaGAL@G^#_mc;NsWrlAD+`qZWb^t^))_a89j)%}~eat~*o<v2G|L{<b>Ru-BV
zBjgkpP9+>GM;hc4ES;~)U;uuj4m-$ITN}50iBDSAC$YaBI&i|ZQRx1TFzQ0gVZa^-
zz&o<7uxd?dg#HB1|4fW&(V5(-+i%X1>cXdSU(?{~1_AIVAd)jFUjwEwA%Oq>H3fo~
zApm2umm*(B*itoSUi$g7^vTb}I?S!vCrUE*2KHb1?o+G(%-e6o&+g<;OROE1FzshY
z)-*S|&1C*kE+|5SJ_8#Y4v}V3bhy(QcISi97&5_WD3NW&DE0;Jp{i!pzZ@BiTJ{N<
zH(9qcOH)yBgLm-tiTU`*rG>kqN){__ocuAvP&NG}9q7&N+d(nrzl%AaP{R%zYeyb7
zzT(Fl4o)S-<sp0LH54*2T#Ab7theIWzRM`E-c9E-W?wqgLbo@K!FRQ_MGeZYFNRGO
zVd(jZkxPKyzDd)NRHH%P#pW=O8&rhha^5C>Fn+sVfG}OwOv&^pr2{_`EZOK&dwP!g
zR8QwvSP!;W)^<hyRecracP*`K%F=v&*IF6^TvjhVDv8cR7R%-C?thQ+uQ9aB$u}mu
zUWwcAjd_D^o|zfm*xOaNfb9KfrDq^c+8?fd&{&qg`IzyZytKdGod<92bB`lCM?>BV
zwC(&zKilf6wIj2sVuBgju!ERII{Tcz?}}X>b57)&YqAhjbOMAsDH@mwN6HIwap3}_
z=~JKmz@wve8hPa)Tz$G0XV|l=;q-_H6^@8z_=Yyrd78qM>XgJWKDFgrOW!FJgF{}x
zObK4nL9%{0Af<8Zxn$nc*#SxaqP)E0h~sU_F`-JXSPLEbj!Z}<B!^KQAJjE}Q_ppm
zM#ZgE;N&NOdb9%yad>z?I0nrSO!e@}++ZzA5Svc^OD^vAFA#;espoQA7~KFPp8KDn
zty3DKf<so}m-;3r+h-r|^=|#Ra}C1ZBjqa1Es{qb&&q%tH6Cbn)zOd2^E2OvITO}7
z3RW6BN;7TD%hQuBzHyDYIE#m{9LsWDu2Q6)r0CbZd)HjK0~Mo!W0ZU}d+&zlRg<&s
zW`32q)E7R}0Tgw1d$ZX^P2tO!yCQ#%wjNqMBX_q;Gc>tAAklYMym0gtQui82m;ZpZ
zTb+I$>Ptf2wsvZo5-klSrI`C9@w2kxK>l~xS7j*e@VJ8pL4Xe{7#6p)?U^LR@5Nlz
zHkv3E4@vxJEv5fOq_402z^~JXT3IA7o)H(<zia)Vex1(iXu_jBD{H0;_QB}Xw86*Q
za!<H=QZhs@STq}PiSC9QK5oV2J*<TN*+XGZ7ySN95aA6T<XcD3ZlzZ9O6fmqVBx-$
zxwnTO-+R6uydP19?#k4Xv+vL3=D#S}%je;S!tzxg#J9K$+@To74K<mrDb3K431T80
zhQhL6sUy!HwgKbM3SYY=g!-U6yg32R)wNQJmonWQ@i7Q63Iu?^e=BUpa(|VM*%7(S
z9Y*ygKSsHLJ~Be4EXmOf(%y|*==+=EBEnb}u2*vF)zd4U5FNmsLnRBKBA1HWy;qet
z$7*j(ds*K3u$?JoAf#h+A+@1w3x%9BSZ-XH(TY7;S@9R9wkFRGE~8xlD2~8+h@$!{
zQ-HQn8XhI{WH9+SwCSXMK`81Ti_?W|z*zWM&*MCdA^;xPvwP?z`I(RZp)MWCa2Wv;
z7Z>4e4G7TH8H7jYR}HTPtI9T?b~km}VFiq{Dx_}QxwdNp!sXU&F3j>f#h;yPc}D${
z>S%iJuGm*hKv>6{8npd^&d7a@yxgBI-?g_BaZcaN9Pa3^7E;R#B7ElNZzsgrUu0FR
zlB$xD5LarQJ_+xD_e&~R7lNQgLDE*E_r1wt54rLa5;>~75^tn-s+4>Wa<c=N@P1Z)
zc)X@6XHw^T6yCj0q6!bIo9z~kS*a8ck@utXhv+t6F=v-Az37P7S&b!jRHA(l^hVHy
zlBRhY&7XX1)6{syyEXIPU;;QVNHv$WDMv<onNoYwvJZ3-@nF0D)ZMo(wWU=1g08wt
z=6yh-n?lCI@8N-i8QRuxr>x+G4CT9ttRGb;2X2ZYWux;UUir5faAG67rltT6KYU)I
zRhIAv*0U8KP4v>jU&pN$JAWMz$L%y#loK*B&IFK1o1PXLadDWQw$sjwKs9V11r}AY
z0wCH6ES3g&3E!Ic+b`&S4J;b=Dl&9Q%cqlm7kqr|wp!^Ac7>kua4h(b+nV(kz$nhu
z0hoh$IGSo0a-p@5t8x-pLi@EuBEAmkC?X9PT2*GD1EdkUfOr|UAENnQLH^4lMEe0J
z@yu+*Zgy2+kc&sn2(zm5Z`5Y#b!D?(@}p4Q1R1Fe9b)pcM6ci<8+>nJ$_c`U)!+tK
z5c620XI@8sbKN4g2ay%@r;?cMA|O?EmowZYqj;{MluOU$01Przg(<{EpZ;ScItS`u
zWn@L)Z{WdIT0bjC^+La}t8W5$(nW*Gp0WbwkITWg_CtvNtrgc8mqO6U&tCa;>lZpQ
z;gE_#J2JVwKp#}8KK?BY2{j1+9p%2PP#lN4=zYWga-0#a;?2qezVq*Ga02Edk?LMF
zWDm}SDg@!gQo+!Q{$Ap|H78sR1i>Axpc@&om|$NUltn}|;-Bt%b=lLqW3BY8Ipx9G
z!TuKcq~*6CRC}kKaA*`QOmhRM8oUkfsR0`Q`SY<eMuz5I-DwDr_lh&H=?QSa26y(k
z^Yv|~jlFre-IH0mVIfQTc>NjJ%6q{XrUkx9XG<Ri=j(r*Zl9U6W;90d|KdAB=Vpyn
z-A{ncE7f3&74sCGk<W+k0)8F^W@Tr9p_N(O^eLO+MCP^EQjdthOgt)MoV8Q}{<`<j
zYgBlB!nL;b<Y2+yU-$wSJ^2walxuagjy^uB%Ki71MOAiH_RTCADLq}LJ3h@(o}{0|
z5dgQA9}P?E2)m#+Wz*W(vPnUUNdVO?{!9#i(((hO3y`Msr#izJ_4vB<!|Kgysmsn3
zOb^HugNo;o{A|VfiMmz4NO?fnzgx<v7{EDqQJm4IMV8Y8qT|f*#~yhr17(RLS&cF{
z&xQ3uyQI<cG6&O3cb%(CnG=@zUN|!<AYg-Oi5k@rGrnG}wYB7*9%%UMS`lZd2{PFS
zS2>;E0d*No;m2ts-`$}mI=4HT%gvD0%^c|X>sn&YT5Pu*eSQ!bB_k}{8e&&Kxt^r}
zW&G?C&D!gL^BkI8pU8E;0>1T?)#~m~B#UC0KZ!UGG|o#@d>)Jfr6mOGZttXigHHPX
zXzD8ikBGgxDeX0nRwFu$OcDitFih$(f-&(1{RdzV81oX+`hs>ZkWYRB3<JP0an50W
z*8^Oa1MCAfA|~G&l7|v{`=u%h#aWtRled*YvHruc#Z^MIV;~3Y^2OKG=dX0c1?U6l
zzVQqZ6t7!}SC&^+HBvDR+W6s3n|Ia93*qj7=vpiz(>L@KdD^eleetA0iSSX+89=wm
zW{GCBwUVHh<Fc~cjEw_x9DRtbAp}RQ&QbRqr4YvXj{hh^{j6~vkss$2FYOCQ#|D(4
z)k)j@dZ_p1CODJPs9LJOe{QaiwEH&=sxUUhp(Vt`>g$Ahu{qtv+vrLv#P9z-$gz}m
zKR!LGLXT-VoS)K{XX%Tk=rbz1K8kEc{OPo)xh|+@v?gRE-%jk6!Rw;>NeffTpHEy*
zBBPoX24n^WCS8MRhG$}2wpf5f+U+2A23B2=KI}Q05*VtJV3=x<VDMTLplHS0RY<a^
zujlLXCFBmo)K4*34#h7$erM(3qI-oNPtZe)l_T&<<@!R)z1e|RC#>xr&{REQtju5B
z5=c+mUZS+zl@Q64zu2rH^J%Oq#U|k>vZ$VX7HT&wB-->kUyghb4$hl#_m#^vMee)|
zdR7_rF6JTpTOFy(z}JJAnN{0rn|oD>Tc1qG^<Qj{D5^AGQrKBHIa&)j!=8tYQ^dro
zPFy<g({6=nY_33~hnL-)nMZk006x-!%=N_i-jOb1%lRSU1<PJ6nywz)S~+m^5xZG=
zIc~W9zY_0w?)Oa79(NG{c-nnQiQ$wM)%FWt@D)qJf82dG$7^ks_B&fnR(2w#itB<V
zCam)|8sbNT3@Y_H=GEFI*jvMK6IE6bDrPA?%{P!!nxJ32PU)LKWtV5hf&i7ZC*ZPu
zHg8Hg5RdQmo431x*5dIMnx4WlKn!?WF{?kn{{Arv9so7hI95iC_FCpgZVi=}@e1>;
zGA7Hn98%NsM8$CoPfq#EAo0KS9ONl0^yA%FQbGzBRH-iC&Y;T`FW~2Brx%I});9dZ
zrzP@V5Eleo+34>Z@u+r|o)G6wXm2L0EK~XmgW%<btAXgYrt4fj8Wq8+BkQ{xXFDzR
z!(Cf+wimB45x{yM_79jvTp>vD;_IE<vz0OIMc20nvEBYB*Ucu{Ox4G%D-o<<oabVz
zYCw3c&qRoW96VvLe}$H{E7=vOs`~>xwtl4BLw4<;r(676PX<>0Zgu(l+8|k|>BTP{
zQ7|YMH~O*5SCwM(s!TC@KWAXNCb2Kv*HefRkI4dg<rQ_&CYZYc^H_F<!=he{#RZim
z8+ByePZr$eL@?a8cFvA|cKedz@q^#*02}e#2_gymY<@{`C=bS*zIuf!L_(=YQV-Bb
zosa4fk4wk{C_fZioSBntcg^Sz+mJ9wXq`q^Nd<`;DVT+-j#TYNbl<SD3Y+YKeF^U-
z_NdvKtP=ic<|@5<=6!=Uy!O*isHh4$#lRd)IBuDZX%=)3L+^z~#^3^Pjz3^4M9fD@
zvBH|{e5O0vF6xyteUU84b5k9|0G!$Lzg3~!^W%fUY7A(EKQsO`!{p0+2~U0*X#}8s
z?oL68f@~L5|Emkiyzo9-#iGiRR5+|YmbRc)JFF0HsMVBqz9Ye)dEMSr+M*v+11?ej
z*W%TCd3jw{b+Hco(SOiC2(R~XAyj(t=OFtsP<6yOA!BsiiZ^u2wI;_(b!7L5j{mJJ
z6oa?Sb;U%T{2eCHz0;=$xr>^9{<Y|-!b)&mbHV+zenrmL0U*5}Q(52avqbYYyK3*!
zme`@*=Jdeczg@y59M5!AV%3FOlcUuu2SVElhwTxp#y>x2;wl4G?Cx=s7F2<B{MFyJ
z-Tn2Mx3rOCN4q;y)4~3{2;OpzEXA}(m97V0%l<t}xC;lQ^Qdq(uLB;rbrsz3nut#U
zG4i=0p+aFwy!_E(cZ6e`dX>#8LhkG`TBa{PFJNJ05dA$%)Q-!9{END&G|C6+KR<nO
z?Qk4OeZg`>Vd_>BJ-_$j_{f;~!TG_-V%G=bV=)GZf49Gm&$ogvK1FM5YrE$MFJe#O
zOpE8_Zu<`vr&0{Sun%Kr%0Q+w>iJpjOLkz=%OW5a___5e5Lf~P?YDMAa=$YG#ere1
zM+`tgAgH+e{6Xtw5KtEg{G02*1gvH6@UbHOf4?c0jTuNrA7K|47X9XA=AmAz_AIYO
zdS6`NG!o(yT2SM%b7DE@+j^w@ZcuxLyGjR=Fdoj|E&p6smVuKWAx`4Uys8NZG%dUB
zH$G*|CDOK!N<JJI_ys7|lPH<4t>zZf#(>GMtc;(earpU-)OHjE;BNAp1eks^6dOsS
zD_@+=u3>(l;ojsZavX|pb2>`#o^uw0OmWJo)RoCHxyg_F9FGL+f=zO~h9>M0>+!J{
zTBn<Qe2Fv<+DT}K@9=`aYzFyb2y?JUo<(?9ztT*DLx=bI(f*nRXrkp!+ZEWFb|AYU
zCp~6Vs!$w=gWyzBR<kA|G|k#pvzvWjKt<$O6?%8QIQDF!_#Br(Y2}>hs>j}ym9hCA
z-F13Ou6NJRG$NXedU$toe@Nz(fA9TvSoFcz6TanfyBRI-RJ5k1_n)#y+6@4<&CS+|
zt0RtVC5$eJv=2_A7p$llc`l3q93#~TQ>4GNM#^eh+i7=&v4&LN`*da^&9}o*i_TU+
zlcMJe7)le2T&^YdWGjH$<~{&fH8=G{E&l{C`LN5H&d$$U@|CixevH=%|ELNmDO2P6
ze5)<|%~4*W!MC1QL1cJg9)wlJ%YS%bQfqX#keeH2wM*YD-~s|lp;uKbqn(@{8Ez^h
zh%zgJ(wr|IE?>KRYIDW#%|TJ%Xj`@dy4BCb_s+<Ob?0@_*D{t_pzj|Wa7-__0ixg+
zqPjnnL-Pg=SzujbBqzWZPI{y5?G^idZ5LSiPEC@@3wV$2i03R29KW8Gke8%!FFO+O
z%ZUByiuXF{Y0o9Vna#BSgN3e`n5e(eSJ#;M@fvXE2xmN&D3HnAt$1-U|9*W4THm{w
z@dNM}$rS-pJ{)h*5g3irN(*7s?xaoIO1V3=(;)*wqLWu}d{)x#)Q%2O7pfp2(0KZ|
z%MVL<no?Q4idZ&Jv>c+BVYcM<cf&DTc7}|qh`l&WdY{Cakm!L((u2rgfA$7WL-HF)
z3Tzvx3*c(R{kOXtek`62RYQxM#en<`!+;r?hy0=uY2I{a9vH{+{n4tOj=;o~pevU0
zE(zcZ<S!^)BWJ(Mb#WrPcvH;Slglj5eZZ^+aLO%#-6#gwHNm(G@ShmJ7$<f>PZGDB
zn3ITcgeL#e0-@dCK}{P0HD=A6{M2d=fi(Jt5XtTIQ-eZ_xxcx9zEg<{^19Bx_og>c
zb>iqdroATlLa)2@Jp->O+HD{A=OQWqzAE0=`7<@9zMA(z4clwMSEVkRb~7m(%#8Nu
z4j&btpFRxWW|%pEUY>sCsrlflT>Zob1Gtvv?qG6Lsv<(VBF#qecjvXA&T5zOg_y?%
zC&FQ6TB>bhVjv6BD>VP_JGtJ3-|V1D&%56w0u}u^c#v*F2_RD`2`-%~``2%){{?&`
z@I|U<U!7PxrWf>MT7Ulp-P|3B@$GE8awWtUlJRHqEpBd^LJPnwsEM+PH~-=ze)fDr
zu<zs78NV3K)0g~DfU=2p(xE!X_xBSzhfIq8qCK8Ny$OZA@AlCxQ_Ws@v6l$%cq5ex
z8Fpy02qnPeGv+Q+Nnhu%QrB3z(DK|#{HL9#Lun8{w$HuDmk9|YX^>ur<QA5JY@bK|
zJlck8kf&h&I3PY9qzzvM_woJZIXzj68I1q(AhH2EFmtxidfBunz_0{>b@veh^%x`E
znQ!QNxPjQTY3M%QsKJAImbic5n+p`$z}8A7(Q+uu3$i|vG?M}_Qhrsbg0Hl^Y0kH_
z2_5(L?PMtxH6CL0#fmy4O5&k$I|GO2T9akTyKC?3-kxawZU-zox1J<{|95?>hi93j
z3|zeLIJVp3i@9aFkTWvH6~z1a!@%F1z&1xYm0Vv3bfVYQ-9i$)&6u-n<swM=H78s=
z5KzPaGviAJV<^^JbAAjQ`f0GwN`HVe#@c_yGr19~@B+y;fw}WkVo$ONuc#PSlFbcA
zCJcU4JJopA^G?m_MtzMQ@VZA2WjP5XWAwz42ve80s0Lfrb~puaF)l=1scIW^>Q7Mv
zIl;Km;y(dRb$0`K#kYBN;%@nfxh(f*#e<YfvwaDRnshnTnOKI@%WMVsmf1|_ehX*N
z{>gwZk>D6Id%W;T&2<R)J<|FPdsweQAZ{$Y{ap<u2Vye%hIlcN4sVi#ZQY?!v{#oP
zP9%SSDqZE1>qEzhMW{aa<Cf=M%WmITI6=3+gX@3at3n?)ZxzszoI3qFxB7SKHD6#%
z#H?M17KKWsKIo49yHPcUroX1?+os)!FLWHo+Q579Li&&WKHrH@jNxZ+(3X}`-tqWD
zo<sTmFvhx{et^o4zU`)WJ;r11`3l@dW`83_#R1&-`*hvRpI(nkh+{*DW(dJB=#9vj
zqx$o$^TVTDtqOa`2U?;-gp;iun<%QA5k3xSR#l)mdCGS#Vt}fD&PYP|IRkhBID^(%
zyq!%`Qe8q%+S1dv^0uYsm>#+f@WU+(Qyd2VDJ19~yx$U&sI1dE*P(YO=#@4mHQ?%f
z@VDmxz*j(Td`5;OgGQW-_ERbJJoKI68}G0I?PMSq-`X|EMWHA0*z5XB8NE1c+fE6?
znA8q1|GyPKwfXBT(k!Ab^I!8sT|mhk!l#WY{WRB;9&z>NJq+si5(^v%NqL*-HzMFz
zx6bg}!=il_D!2@do-{DL#R5yiJ;-aV#kDt>iYdP6&}tcBho~4ww0%PVQt~BdS<(wh
zneW~+b(GI4z|6?<{ysj{y4IzIvh#DBKCS-t_)bdJYOp5b>zW!{a!kI-Rek2m(MrB1
zDv{3pq%Pl+;r-*lE(%$-*g|{K_TPyF45bKVPv@O<$8^kuByT^Cy!ibCCnM9ivaV=)
ziM#G(-16HmV7B!>VgkAk5wf=qD3V^_Ydv!AzNvd(n~yW0SDx$w=da{Y^SFU=2Isgn
zo%S${;C)d!0)I8dRdbU6Rk*nDuDZp$hL;z7NW?u?j-D1EIrP0=N5HczH;MRh0a<3&
zONkK;>5Xn+33%pdY)8lk5A*caT69eG{Tbi;GqoE^|Ip8kMs!?fi!Y(Mb;pblN_bn_
zT;KebJF8)}GzZL?47j-#xTe4~QSxZ>k4@|ewita>{i<@+BW~pJ+uAm)$$v&a5Ir&e
z#^T1+l#|w0|C6rC&a>lPx`Xo=j}f{~<AoBYg~)b3#N$!WT=VMb@o|3D1|7vO*MI7N
zen2-H;byDQ7R6bE-$;&*bu;AQDvNqQUjiZFYRtb*DR^7E){xoj=ZsN%1E$nlU_rB!
zZNSu<5nHUX7)vM%_@6jaQ9f2+GE>K4<~<3<`h>yp$a@`&{s`6-6RDZpd&U)76SP0-
zuQDR3QSLh+h;lhG*mKOF6bhg$<O_Lo@ue1GR(sI3Y&2x(714|8rYweaJ>ZSF$HBO^
zi>xS6LiK?yE?~(0o8Ab#*T?GGKL+Y!^%KTDeUT`^68?T<m^B%1NVyDCRAI<_enl3*
zb{iCa=;;Y_=28*A>bIg8%Gt_SlN=YX#fW}w(Y5#xYPbf{dmr!m?b*{BwZfK_?>CGj
z=mwUQvhU}VqR-(!+rA7}@C^WLv_FKudkI%ykpBRFF^;L(Ro*<&8{kH3J?&V~^b@9c
z5l%#NEdWcD;XlZy%V@1*APEYSa;THT+@c-tJSI<@0W)U{rNZCnCPJO)W1QH-xbx+8
z7Wx0?CbEa^dl|icXWp$;C^nxCK|P(c^+Ed*TEcuMlqw1#Wv07M$!mpE4kHZAthT_0
zx|7!K7(3x%&;7K(z$V)y9dPCs{Ospswp3{PB?i7y%qTu??A?3$xaZW%^P)%u9DJMa
z1;mP-<8*n25e9eBEzfL-{SgY|d%~KmbyDJG3;XGuCm7Ms2W1v@lsA~X66$j28v4#X
z!GBbAvigO-3DzXx6!b<@@>2MZk=cg#s6_n0FJ*{!9ymkavM8L`wC}M@oSV}WO07v!
z%gS(bdmK0COPqU8(HwH@uWPwO$iKU*;|Wmz@)pXADKjxw*>{$<X4+c4Tu1SsTyEMA
zf8cG%_F9y800Pq`;^((w(cQ#XXuj45p`>Wl2V=Xz(i8j^@k<9S^TPDYcBgz&dqqEr
zu0S1LQ?NX+r9xs=N1$~zemPmIg}|?-1@{xLXNs|k5L(w_JG!-|5BshkpI3879;&J9
z%6f0vZTU9XW%OO@V?v6mC~_$@1Spg!_ikh$nDx5%&%)J{(n_+u8g4bATL#(<01CMg
z=I(IkUB5Z<f<XljYKJEp5N8-XbvqEQfh~8ILurGwViy1?RX>>(p)Jg5dAIf<@}`<d
z8X=haD&^*oDLd<Y8O^WKL5+Jgd*rd3U}G8y4T`+yYp8mJyu_~P1ahKIw*!~5RRHsx
z*S9iBDx3bT0emHbSJ1v~RgvQ>M`;?zhu|(@+h2OXAk4$`oUO3`)Z>IVAN8;Qcczl}
zS9oTubWqto<Mx!M(3K>I)E<$<&6TsgzlO&a=ZfkMv~h33bco&cli=lk<9_0u=eLp?
z)Qa}mgYqpN8pG+3bmC^bU#yTf5U95sQ@(N}{G}ZTU{M~i8bw#nd_lS5)klJ)+_!3K
zYMZA|a^)UqYA&?KoX?>D(%WPfD2S=q6zox+4@y1yZ4|Y}!C#oRcy{7Hrz%7a`u&Or
zRi=D-h7Y_4j*6Xu12n)r4w-S1&lfsRBsbIRflECSE1qQRt}?4AyOi!ykUDzXKs0~T
zNnXJKZnP+dej5*T^7a!^5fykmE=m;a>l^87iVs2R|06W>`tH9OANkr9%`ub_zu|PJ
zD&q{54E%T;zRXAIdFOuJb5<Dh-6<nb3ZwDRWPH?8d&_`p%t?6hI=#8&m0w{Bw6<G3
zXs0%ZF(!|<zavgh(<<ZpO|?rFLJ|a?%2&t)7MZ_9|JN%h^DY-rf4=tMY6`ShFu(8n
z-d2~X+zjANXMq}^?O#S8Dwbg8ySQ0X8A3KnbUU7?*LroeDwmv+SUVf}z;CXn!%32X
zQ?)&U>JWIgmRoFBKm6wh?c)4;R?4lcH~apJUA%w#g`kqm-fS`V=_TIE*^l$V=R#*)
zx6?i^pc`MrF+PgwkR!y5en2*cFV6EY<TW7WU&GvGm|w?}M7^LX@AfIsldk4EVBT;-
zf6ZgGOzm5ifT0R}uhg17xvMsHlLKhFYP$SOQ~F1#U2vA`*URk9PoUiJI2CS&(xk66
zv<dr*v88JY96XuHZ8#Cd=N@kkrXlm|h>f;?(HB!HZ#296(}NNF>X)32j~WVUC{fh8
zFTuIDFg6UpLB@`wyB+)>Bsa%Ica5M3FUPW<!^rrEL9x=KA3_5pR+66|h(~GuPcruM
ziWFAh`P|O?pNA|HR=0$;QSq)_XY<9xQ@~1^CRaLOUK<pK#FAp4zU&2A=yQNB$O~_w
z3zQ?H4h2p%D+u_k`I6>^4)dzO<=wTwSoOPz3`8O5P6l5_W}=n3@q8zqo3+ER=VPW$
zLIJ-nc&$(o@B<1vrLO|#@T)yA_nw=%D$3i}C%Uk(02yyzpwj9z)?GYyH};O!i`RGW
z6A3NuP_plX8F~*E|EQACqK|q@&scXyA8mD?uhE;LoneYM7;{W}<H|XpaXJuX-AeT*
z_c3|#^A3H?H<zX2wfOqO15RTy;+*Xz0C`CHb%gPt8Fu=@PTo8`OL!(<-mC(Z5pnCc
zp2i+-Ec)ZSbh!@BsgC`tV0;3kmfY1^I%nEuG^DzHJm|C4NP|AIRN!6*qZJIb-)uG?
zIff7m^11ZCib;qArW}U~hfopT!o$9^%w_X3mGqmkk+h6ToGp2{eR_ICFHa!-)<k8^
zQ+s98B+d5<A(CEkFGQ~gU%UpmAO>cY@FiI@M($EDS;HG36sfKTfRIVqJ+PU3Kd;sG
z+=<tC<FQ@W+1WLd&)?|ksmQibz8vmXq{2is5nZISc`tjv0t0r)y0Ut*J!H#j)U?T;
zu(iV2bZJ@YY01N8D`VY8TGv4fF6^+(<^}+8_B%Pcdlu`=p1}03IY!-S=uhNLftaXw
zwG{2_$`?Gkt6)(MUBG~d-Q`Ux<2$Cz_vO}sIG+i+6v=i(C$P&NT|W`5+aGd&5)4H?
zL4Gk^h%Zw38Ni~q&KK<DlH(+VX3;@vL#{0SC|&uMFWA?){Dwi`e8~CDdnY#p?6O<x
zLIF~-D*h?}VEt^t=YyR8jT=!a7P(SPAj<}D)%?3^?=98v(0&ZW8#>5qeJ5r%krj$J
zox=0XOa1`x%}zF<Lj>)$ZuGz|FIxhVlmC9RQU(Y&aS2r99EtcF0^mYB?}a`m|3J^L
zZ{J!~ACTMW_v~MM@aKbXG62v_aP>(GkY)ui7sK?m<KkogS^C)n2Yl4`Z?6xrYl)fN
zT&R<qsb0;^y;teR0c_l&qsnr#Zi;x^xn^}k<CXa#Pvl1yGywU!I=*AStNv|U<2GmQ
zPvN$4QOCO6lfNyro2Eia=G_Z(kB^_vWq9RFc+jtOMjvNl2-9C4bxwtt8CNPaf)#Oz
zZ)1+g)Wz8I_WA-r<qi?&hlWZMJMl>qB~dH?2N^-;zQG(~iZRX!4<kDG!|{B_M>F+l
zej8p91Yjf)e$MW{mv?_SU_(P3>EQ%G1WX_VJgNaZd~abvKyWD2j+!yT-ONlyc&r_>
z=;lqG37DA)U0pr2kI{xhz*G`(;grmyc7<+R)xFwYb*+^Mrj&BdNr(W6g(PcG-}|z)
zb?Y8MNJL!9R7y!H6R~z3Dh@SABxd%3jqKXHMZjUZ0S|Zg?yi5~PLu*+Nbx#{f>|hN
zx7u3WHi7^`qh=JunApS33>=7vkqOAUwz>i+F_vjcQtI9wxBFZEg4j=|lsMRZK>Mb7
zs{6*wDJ2mJB$6~bZ__-@r+KeiU2UG{QgUnU@wjuSoJz{+L~vp#431F2A5E8L`yZ;X
z{4T8gPjx;3jF5liYPMFmty@l5C^TT*cZfJ4Od>)YNWstn)ZJ7qg82aGkEQwD%rx$o
zghZL=^ZE7Dr<}8?wYongxj+bvfP#*MKqKe!`9pngO%K~09`bJY(avy>gYr1WRyf>%
z!^%YXRPnjvxH1jEIzaT^OD-uPkqnH?&{?V3veuy)EBSx0t@%Oxj%J3fkK6|UANO}a
z2BfaO*UnG7VniSUF(IJ)NLSL{217+h7{zyK{P@?#P$nbN;8GGWfO!md?D$NE5y$a~
zL&Af9@Pw3%wl2_Np8CFHe1F-8eh&eVKt1-R;{utOkJ;^Di$4#|h^PJrV08LbRlD}y
zd(+;Om`i^uK7*PQLR0ndoO8*USkz2gGwp7AbooY72aymPsrBCNyX5W1E;m2sj6Gj3
zO>6DDnuZ4hrm>m<i0FJikI|63BN3<cSnhk>t+fW>z4unXe*5~<r_cZNPk)-vC+((t
zfB*9KxZMNs2+JDkjsbc$PbsC8%}_xR{y*B@v`K%G<hT-is%qvQM`j&pkYEGkU}koA
zWd8r3Sy}9G1|;t4Ix;g3cQ?~xKd5<Dk^6<UvK`$5k#1yHMufYosXl!N3Lq>p7&X~g
zIdKUh^Kf$n4V~1ke13I0P7z(|C`~nq_RN$yR-Q{K4@6`rM5PegosTkTD#*qFfj~MM
z1#^GyZFj#Mh6e|+gDuIg2y&=lZMtmR+Pk?Ik$Il-({=OhzN$xS-qo9WH$P6(beQWj
zWu*O7=e67Adi(rxeY)J6dJzG!NSUT-p5{^{5X-tO%d&qu0C4xNy7k^oLFw<7c$fi3
zp?QltV1@iRY2{84?rmB1<z~y>DK;sS936k6!QBQGf<aX(gSB<QQxy=w!ey!l@6p!n
zdcADh#rkc%e)*?g|N3-(^XA>#)8~g}Syio;$|9v!nPv!E#PM`Iolk9B$^G5iH>bnF
ztuOccMD}*FGe5jNOm8RQ63wxJF5NC`-+pf7W;F4@DDs2c=YWHn$G!FDm};$ulYW2s
ztlKJ#X_}`}SqcQ<;ihd{El{ROL_8AsgsE!Rl{>N2QjW*tbT~e~efz@?KLD}Z?}fq0
z@FwmIa4=DjDBOcJ{MD)@5hWfC2mwLvp&FW!C6ByLxz8M)_D;iI7dx??%w2*+*qs3M
z-pvgJVJW4I14%l}HGFA0ul+FTTp5?a9sv}8FUvwiQ<1|o)gn`sxyX=EbzQf%ZB579
z29TX{HYt%#wC-+1lIFt{!a?SlBO9X=Psw`#LJ?Wrv-)$c4|Z<E&gp-pVP<)W+=!%c
zin})pY+VRGq2v{GAkyU`m$^Cr1rOr!{*E1{cDvkr(?DVp5EeI!B87RHrXu3z%l&?T
zy<V2v6~JLWJU%`ikH=C=Eg}eYbE2t8x9_4>0^i5FasTUI|9rpQ)^!!)sZ0!R>-P5T
z+vDkkh^@8T?fUuiXJ$@6%dKth%D}s~Zz%kJy%bK)VzhPr{ORfU|NMs!fBST~EZV(y
z^AMJ7Rl>tZXp6ZyMS#BgN0ORpn8I^^2#b9uGHjYhAOC7;79w1T3kwC(l!M@<Fx4`)
zwMd>sNZw2_`XBdP!b8=&!W|+)0BHCK05DT?Pa2e@kqAhG5)Sw9*80}9sdbB0Eq{yl
zzAbB^Acg_mV%w}+sE4^1HSOBFrWnM{K0jZ-yxeo~SW2C0NjJP{FC40ETUYIWod!t`
z*0pz2H4G2c(aaNED2NfY0_0qEE>`H)x2>ZwtaWw>l_ImNc4rQ#_13z(hdVjoU{8G_
z*IJ0C>{g3R;UrbU{eHbXKYzSmJ{O9A{LMd=>0o4s^PBo|!+kN;QcJD1Om*o^BaX+z
zfBEfi@t^(^BYynV-&6Qs|MF*lihh=Ocj6x&=C_Ah$Q$AwbT_$ev2;>=-x&K{uUB?M
zGAk*UF|XteB8|9h?cVi#et7rhSMGLy{;X~7%UzhKQq3AHDOPD!r0IlvI-T8o+nVJH
zlxr>0#j^Uh<z{Zn{oc35+G>3ZYX~J6BX_70C~1x{fW)z+3i!aRnW>b25=1E!NiCv>
zTMQzK;C27>bqh6kMSGK*TiM<lPyEVmsHF@c38<;xw?;I{D4s%OV^Ag<B2zO4%wk>E
z)>~IgF}8ya5djGhJj@em49z;4gdlSkPWuuXJ6^JU>K#FmoOt08OoZLDLK3YZM`UUd
zkZ!NeGE5ABolT$7v-`7uvrA|&s!0SKWT56iH&siCWniQx?!KHmjS1uy>$<I5YrVIv
z-TA(?O(dO6**$vOdhY>f+q4#{B`d=q9hqtC8!;kKtM_iIaC<yIsGIuFhS5jFmrtL~
z9YC!j#0Z!h5lyv<NHV38I%#J2%asVs$u}WpDYd48qMJq6j~_lf{m=jLFHcXGt!X!p
z5toxRN>6|12z0d!ii+VL2zOVv?kSi7d&Ewcr>}F^*ZB;|uubVovJ2b^?oJ-NmzN>|
zIDc&`3}QI5fl{2F<dZBiHZU<?n!eoE%d)Op&k{e?DM^{NRuFpMwrx}G5uVB{^-wca
z^<D9slRPED6)uDz4q|6EhH(UkBf`N>0-_+dFe3p4(YCeLnwzEaCWq;quc}umAfhTW
zK@fC29uanZxx73-9p<uaJ-`-FbG6(QNPkx?$9bxfKoSV^G(DWp$9bNsJf2V5x3;}!
zTTay9*Lav@stiygUY$4K-emELQscW@od4*HXiR#8u$vt{zLIe~(9~~hJ8k{pG(Wt3
zR~VNs9{~?-=qpIPb6=aAo4R_;(>zVnVLoolrrq3p+}q4FxC5cRwQXH4FWdd1+kfJH
z>)Yae3$x7Y5{3v{c9k?rGI=0DX?F@v0wpn+6FDB39qpy?>(iZoebe5)PD$>rs=fD8
zYW7=1RHQg4%t4&6Mq#c+fZPceE<wUzcz9R0aPly<ZA%+@*9hw%(rlhHxFmx5Foqm{
z-22x)f<C`4JQFGMa88qrgz3pGhkqG(gm6;$b}8}Bq!gjzk(sgkVtO!f_KVJWFl3>R
zWPX4&&>a{`ALPmdNc9zVGooy#!aZ8kb=}r=Gjmmq@F-EaGE=yvGSu8ndv9B~np$Qj
z5+TgpEZpHc^n_u6l#1c+x{bfD(BCfC=`bHp=QpRb5SyB*<%K^S4o&+!&wr1PkEe%+
zruzKhr{nP;B7|6$l^Ka#b@QdQ(D?Av=l}K3|GF$qEzH1D6T%z9Vt|+}(A`!u4+3)`
zc6ST!uIg^#6r4n(7}MF{RAKOKF}O()M8T513YduKm5!NLpJLeGLl~K3=4N?nIMB_T
zIl^mYAy^u*=T0)=aF#@t(tows0!?k}-OSZg%?;22Bl7U*Z@n`IgGxc2>Qv!~Ze17V
zB2ob|$ENyxz27(00GE=)p=lb0ewrq8U01u`?@22fc5LRR=H@AEmm-mR4>Q`o;UG9j
z5iQ&ib*>~L4#}K1^R_ldMgT@zx2^Y`C(H<D=2Ao|2Wmiy)Tx|$Gk-IGeEa5heY(6n
zN5nMG)A3|M%eq}Jx96uP(^dtG_gZ+W{Cr=-qtseNrc&mqPL)4@{%E}&<~oHvGQBHN
zQYSIh=GYt?%VN_8{fUTD001BWNkl<Zk@@(u)%n3j%mq$1&4bUC4KLsb4ZPeJ<`D9~
zd-w3BN?1qVC`{KIbAO4Bwyr>R_ufo<pUPCE4lN9k>|;I<!c>JRye;?T^0HiB+|+g3
z?pIYs1cGa=c>?WRNkmvUBW;Q;%Y9Dx&Z(FeArE*e?<v=@G3ev3?DBu}%I7Ik<<xbW
zjP^55gewzCdZ(eOt*v{62s3k;#`_~!1XPX9VXj@ZtC^=z#(yFcchn;ST&sXcN1><J
z{b)YZj1bPIF(+@{x5tB&BTV722;oq3^WfdhGW^xT5`?tvAm!0frKwix;Bu59CV<_Y
zGJQM@C>f}t9z;Q5X@D>Lw?A;9{qzhE3z+({Z1?*D4<Qy73KWteqV998LbPpL;(FG}
z&3bRzJ2N|bSbvoCNGJs&M%)9-vaI*zw%mU<9(^uGOPOmu9FFkNe!pI?rIhpeTx(71
z)zj(3;II(lwl)*0wVcmKVQy`krup>na5_Ew@%MlF^wZP*wpx$0x2vUsB)RGY4@WoG
zpdMi!9!@dLN|V7Iz_2CuA#Y0{otfPNl)@;H1R}HJF@Iz5xGH%~hHt$R(a0^x@q<9g
zVTy2%)uKCyAi|(9S2G<^Pwq_2OqEE8L6mXEIXRkX3=SMQ;KX&BCT3wdPIa0mF1+^5
zLy2fQ)(8;u<Kx3TRS+^w;=cC#+6^pa0yJ}9cYOf4+qSLiy7u0)$`g^QsjAxE&J!mo
zfdhp&kbmZd2EY)_jcTd$oca(qx2?(6ig~S54I;y`ZR$IBI|i?LQ<dZN_;4aNk0?BG
z#=y<<um0ih@3({5`o};1OKaM@{q)nP`Eb17?mvC_o4_Zb$A|NLn1j%Jb7GFtd%xZ;
ztu2Qt+FM~d&E>Hk-corJ{Ww)Q%&pE}o^C(Atbd=M*9B9PpeX+@w)_82F9K)2UBqjl
zj|fly-Lbt7kGL-z!>iE4`S3VTZ{EMhws_ynoXBUUV<l4)iYk=aB#|;tv#O2-CnDUK
zq6nA5OpyTtQ(+Boc$ivmp3&Oj%o35juOZ-qo4AoD11Auu6cf3b3Q(kkdR}|Vw9Guc
zvVTejrGdM9MpHQxl|r<qcbRGLO}Johs(8N@BA^z_)>~^pu#_y~ph%@_bge{%rBo^4
z5-g$W76A+Nfx>7vFq7p%fO4rE9O)91yX4W-gD`dxS*Ay<9uYwh=~bT2-U%i#3x>&!
zW#=tLiidgF-XMA|O_GV{qhLo6C>XF@3V(wvz{rPa4E2IM_&udU3__%tk6Y*a<@Wsa
z)O#a3l~RCU%9XcIb(*JIgzHoRZ0nW~LainHGuk~Fh)ASRpl_{r?LDI~zT@?MJo-9K
z0JL}QjX?RHO5T+by{oAL<R03st=lwD;n>#I9?o?thf+$BAi3SvfBj#7`s<%Rwtr?}
zgdk2k*5t#a18{_T^i1uF2xMxMTZDz%=t-H06QKu%2p3^yG6JQrK}aYe2pv#9MK=00
zgyIOdBobuZ%%(m+UCLe6eAAS$q_#1J9bf>Spx%hH<Nc0ltufPF>sZ_~2guFUhQ^V=
zTq>6uOp4ac(KV{NQ2<<~Zt9DQaDO*nyLAtcl){y%mLgwu$mqS_@5{Drz#!EGV9CuI
z-%M^D4B?z3GUa{=<J1I|Qpz-qzSRTf6kwvtl|4Lc>lyM%yE=i1Ig-s+>)~_`529Mj
zR7yFYjz9d}FV7DL^G-rP{P4p(AGg+-#fQKhwMdyxQtIwXh)7-vBN$YfB7X#4pbwQF
zq#P_z>V|Sre0W)Y`m$bc)@ZT<ko<hRSY`@;rEz^FUt`B^lWf`D?_1P6U6#{aPv`R~
z3T%zOZ6L3NLy2bIlYi%$f)}Z^)GEF2l|HJZ5u`Qky=|r)rrx{v=Bl1R4-o+(>DU1#
z@<^oG;YMzpPGKGr!RonG6@MvC$-tqZFDDoW4^WeJ5GmY^jNO?Bw}b~$uWD3fhkb?+
zDS7Lma0&_wH(~{+rC1=G;Os%nA_WAfXB+?Ok>DWp@Yd6BC)q!-|LzP+!F~ZS&bE;o
zjtF)daw%j_6%0>p4896ri87xXL_;ymA|jUVs0iAfi@)8v(<}3aD1X-;L>}Y>+usZ!
zF9LiY_VUgkLT0A8x^8Of$jpf;A77cL>gKB5wWA6Usx4_erAVYE0%)xXbFH-=rwD+H
z1&oN~`%3rs@#t%<ZQCweBad1uGrxWN=JD}dYu&bOS?<enS9Rvvw|=`^AI`rL;pdm9
z*4puWI-O7Me!s0>K7V}v-T(PN{_>~4g^?Kmkt!o&mk^vPIC66h*GR<_G04meVf*rx
zB815RBuXw+D-(fz+@OWiw`q40jM&?NabZU@L;+^m`OR4?^QKbsX5K9`Oi-8*26qP1
zOREqeWxL<pj=i^1M2?4=%0uGK!nO6L9xhR>l(cCJW@G89-ha)OEje4QlWon-O5q^Z
z;9#L3A*!|T>oCh45v^_a`>nOc>A8_^JU-fuT||(AV#i+S2?&^F78gm99fO&IWH`P_
zl_KUaGxM%$0ETpph!|41)?7#&j%R?}`Qdb~Q#~E$$G4A%;{>K(|I<HBwaoKWiiq%G
zp5DB9AlE8V3V%}(56tsnZQa8?T$m11Js;*ZRJ^e*GaY7Trqa1w)IY54FQ2YYmu^0F
zo;>MqKif+R!Pvq=Q<1oncE4V3AmD^<c3sxz>-9AA!|BW@@Z#1Zlo1pA%;@9_Ak^K0
zKrAehNpc`CQJo4AdfT>j9dVV*vTf_OEUw)xRetDb9)E>Jz=dgCPeQ;I#$=V#zZf3P
z*n+@9B{&Jl5#}D0YzJXfCMgAEy{C@E#eH{x!j9U<FpwSW#Nj*`5oVztg$O}zaPR2`
zZxPHWOod69jW|<w%sf{$jy>-Q(M`$x8^S4wncaoi=#@N|>$v1$Mt1t7%t7fko+Rkp
zOyt16*MDscsrzn|z=NpG_WF$A^>YqUak7yHg9#20gAuT?IB<%Pu>~XygO;(&9^}X(
zRjN=yz}<;(Iv!I?bvhireED*}UYGlwD5QwHmO6phjil5f1;Dm!iE18?XBMVXofuro
z#5+v=A9{V4*G*h67j4Z=+p@fU^Y(mxC@c)p-hXe`Yi~_hq)P8D%<ta3efRd=JWr?d
z5lpRh(CY3V|N8Owzx(~m^Hsa6xf^BKa8NhxZXod>b##R9(E&^(DQ^DCB#U%jWeP4_
zt4vc7=5S+SW(Z|iaYUriP2Q<#8>tdJ2#7Fd`L<NV7@E-4ntD$y*0^&VzPEK5Yl7fB
zJXt`@aJSZbgbSx4WW3?E>DDwnqzDnI`O>>Fv6oGuV0FK~+=MfCW6SU=bB2&7H8_F@
zg*o<xd+*(>FU#VdHaNx8lh~2L=rP>&fW0&Ih)j!%h#Ub8K;2cB0x$t3e;+;|+*IGb
zeRG&AGi~d#E_c^X5k*R!N|`FLggJ?^xDXCgEgaUD`^&wOpC@{Is0Rik%t9ymd|m$e
z4<9bq-j$VLfPdHh{{QK<H<DQl=<8CJ5=+PypVGTx*|tqv$f+JGMb)-$Ycq`kJ{GnF
zV03kzRSzRdwAM6G1xN@Hf06l{+qN$EWqEm7m;27AitOLr0~tCDPoQz|L)^Tz4v@PX
zk1XWs-i*7GBh!=+uzkTv<PtPR5FpG^iVnZxy}i$*cEHp6BaDbkErAHL*NtfJP9+H?
z3I1h-fD%)wQkf%64Pn+>Z%JVAFhCqET!kPU0W)7)<K00v_tO&We|d~Wgj*JJPdbD<
zJB*mgv;U3&4|xlAKnLk-5&^g2Pette`t)lD|IG#(uLE12>w{VoPT4q8B$R_Zn!wpc
zF}`Z@Uy)&Wz@vy94~I8z9<{aJ+Vyhj+ZLwUTU)m})xz9+v#!1CVLsMUS-7o=&S@{|
zsVh#U!uKIJ-v!+Se@we-5BDmiYv0z@r2t@I?b>^TN8v&uK2<xPGgtdCAC_(Ht%KOr
zpFh9+<qv<|mX2UTNkKw_!`-}_5hHWu&Eb(|m5G@n$nt7u!A<xr8|_-C5Hg8iq?P1V
z043W!92g$dZsvLHG7Z56?W?c|m?xs+I>?Rzga~D1uZ?{Gf0K_}Zulzaf-78;KgUdP
zL`0AVRLwmhi^5D6v8lCa;BFDA`)yk<5-u=!ky1q9WDX|>Q{ilXkgGIiHq+Ky>n+Pg
zI&UOEWM*0P_gj`IdEXH}f)suHNd9T1NGXb-a1N))89$X(=IPJ{B0xY>EyQ9m)MQLl
zr+Gdch~#`ee_bxmUp{|sZJTPDr+Pl0)-boeZ5ttkAVl+Fp63b70TS|3D*&cARH<Yo
zyl@=Yk4y(h^xi~3<=*4z-oCu7U5f=P!Rg;Tnm^AgzWE#bV3>tJU70ZKB=aJwZQ9iJ
zIE!G`ZDRwwviU4z1zN&;hj)UPddNSsw>Hd{iQHW^e_r`@%d)J?qOExZz4nhul@F#!
zK21_*fk=*vYwx97)KbE|IW*7lwBQQcnIue+w_i!Heo&2LpJ=~9tuWG`2GwNjqMg2F
zV-k@FoRdG&)dSJhwTOxd5eH1oJct<<DgH#2u{BL^HPvK&rbQJ;$g8$1*_1)U>W9Js
zHD~0bf4sl;6}pYPdsOq-Em8oaOl&aj+;?M@Z?1X(2;W_f6Q-hYMv_Q5Mrc(elC<m7
zC^8ER`@vyu-TKxjP-_upWAgCUHbP9b9_B+~VWI$sTV&J`Q%aGf2$<A;SynS|y(gpD
z{rgmc-v{&%oCqTF@Nf=~<$l|UueYm+%+qYWe^CUND0OC<!)&?Vx3--g&s`P7rAAx&
z?dAUb@kM(O7b&2f5~gX36#)-*SI-3|gP5cwN|^mc`r08alLQPBfiUILb=)#R9?lpr
z1yK-VCnq=ps?JQzT!bja75mK{u!u18tnxWVysmaV!2)3q2RY-FGLey$5pJc<BAn>I
ze+G3AcV|GDOA$`#U?w6*fY~g}JwZevFf(%@E>6_CS$M5OBS(9cXJ)mos%G~3Z=h!8
zy_=eiGC#nmeUX?0W;$?=+yx?*X`bp~>P(clPuCQ!xvKkc1mb{z3(H}yAl|x)pcJZ=
zOW{QM-~aH7bz7caF8AB*bUeI$d{|ERe{fxwdu!V~)${RCkH^F5bUYr`ty=^W&o!#B
zzz>zgV`e-Y=E;`>loe|ynGW?dKi%5Pt#{`kNy`h3S8tP_<uy9xA$QD1IFhE$yQ^J9
zr0-q?b@i=l>y1z6a+vLW3Z9(S?kf`=f~`t72e~ljc`CKCFabLe763$8L=a)Be_gwF
zH8rFo*&vfKlL$?D-+CCb8WNFt^wt;mZc!^lT(gxh3}I^qsU><z>7y5JiCr^eKjOnZ
zWOp9e4IpEv3W3C>X{T@I=|j1^A<jFxtGV}iu8qNnDWjeMW+8Hf>#p)or3|N?teK4?
zC)chqkfPW@6L;*naM|KAJY}T$e=G*=VdN<%+wYSdeDmGQmV`JMl>E9eo4q#k`AE{f
zbRGz8dItrvP7Ew>43>GO5CFr@hwocku6OGSp!a5`S<KR}vz7`bi!g0ORO(bq#c&s?
z^WgwM)o!<&duUf-5#iQ;wvIl4bzLbi9S-Nyskgo?>#{B)d^{dc=i~FIfBSm7H|E29
zlsX;fS|;JCp3V>VbuFd7eSH7BfBoI(pPtm<Lyk?L3}6h~g=T44GvwVwAPGvoFagxv
zy7ow&o=7bs>6|ni@{@9#f^L{<HXyunMT(5n&Ru9id(27RKs$$>QsbV@FbM;0+K1V3
z^6tSbpu)Q_J0jgG#$@2$e-*vA?yOXZFx4VLfZINUs=Eb<%eHk>J01?1NWda_cDcjh
zK$Ma;4Z%VzC7<nV+jopH@8Jv#GF7+C<1IpznFv`AcM!`WEOIM^(@cdeES$q4RHG@n
zLM>$^%p@Qou3V;4CSi3C>(g=07z)>33%z^)_U+p@bt=<5{pAmTf5^>Yt#zIb$HVmW
z^d)I<$J1$=XOaRo4hYeqRtn|l5?X>M@*1teQsQhur5p|q@1OtrFJHb~t>waT-^mf-
zXY;xtpa^?-lGlSNV)r*6@u*)rUIg6H)wiwP*G1nP=cyha&!@-v-0XT?UY=hJ>*R5C
zHUWh)0nDY;!|~W!f9t)OIWdXIG)+<@fN^4zfQU$lIhnzw)OkJxpw?AQf<;(JrXo;s
z>#=nT;yR9c9}yXD?iwTVn2B>ImbZ`Z6hj0lqA&@gl=RBZw$t5=m~((FrG#4g%tVgZ
zENLyDpoi~ntdhB{9I`|3B2+LwU{|dfG6XM0%skwft2>BNe*-!6lHZIo!-dE^D58UF
zi6qXAmFu|Ivxgr9iu^x@iG8B*?lEK%!)5bz1rx7-?QX{Dq&Z*&ijZK3Fc*e!L}YO%
z7EWg_=Fr}^<$ig&+^)CferExI<e^f4geExz1FBHXs|=D_>om<$%6vFvYrFQgbAf2b
zlfL8iUC>P|f4Aj$JWO?(=Q+QeZCjY7)WX7t<1xCf_xoX<MRGkND8<9laUB+)KYV%m
z^la*;k%3}1`u9wE%HNX;x3CD3*W4G+;G9>}n{*p4l9xZiY(KuUFV8V;hqAs#evf%=
zhn=(h4PDZ>J3SI|9dts<X)P%#qjQS1DUNaDW|xtbe-{{IqlI;Kjqr4?NUolZAIOf$
zB3&Z}k0tNol+S@+VVTy)dqKg#u)@RLatitiQ}471k95KjuC*ZCH8Cg}@;?s_DkLZs
zby6m#5ClhrIccyuC2_}+`kxpsmBO_O;0O~6hEo`YhV{&ga91HZ9u5x=XV-q1C+ph6
z+&eQRe_S*j<{)ZayIK(`g%3qSjVvmn1SE{K0Ujhl!eus;=j-}%S=O}&D3LoLr{!0l
zil60GMAV$fQpXhetFy52t8ypXeFDhTTeo%HT5pHSR2Y!Qhj+jD@$`q^|K2`*w%BGw
zhXhbXoEXRx#Ld(*123y8_I8;|Qo&gYv*enuf0QC+>e}29V3tyfRFSN7t{NZ-lz<Z(
z@o;DL-O~}^)UDtZr^b%agd^C=JuN?wP-M2y1N^3(r`ZDR!6R2YK$+OhB$+tn?vg81
zjx;RUKNmp~YB$@$f<5KjOt2j*qhVK&Z<u!8Ga<igfC(C*9t<Y}OT0o@L}_gsV{#p!
zf77cnD#DRi77-X6gv37GLE~}59zl}Is$~2GFm&b#8Kl(#Q+QbJZzH8A$rChhTf4qo
z(o>+$g+^7OyxhzJFht5cxBIfK+trqb^8*n}DZHEdtC@vH8l<P^?swhB|M4FUU|rVZ
z;W$mx015)@x-xUEMYYd|IZT)PUFNACf9D<!7Du#gJJczhpFcf+{M)COr;DklgSrQG
zwLTb*!zDCtq_kZu49fMX>ByUjKo%Y5$}9<%xQAIO=|I3)z(>K02!yE;Q9431k_iK3
zZD(LugT&604vVab0D%}|3>bcaG?ofx6o}kRzLGIxh~+Gk-aPX}$70_-;FfEFfBl!i
z!;^Z!B8ho(5zc48HA$kh3j)VVc|fK%v~%f;ZJzGl?nxb%TBQ~-Q`P91;SiK0BOzfC
zF4aXeT;P=;9^?TJ&Q8qCtS7q~IkXalqN}=@FkHK})utL#)wXWSb-9*p>)ZWyILuX!
zQ!VSdx_7{M2wXx+Rr76Yy>~gxe}!c(q-qrXP=ZZEx4z!M%sA9TDbsQ3{_*37%XR6N
z$yOK|mB8!n>t}k2h_IzmBIo5At3C{W)Saq@(O^Q=x2{`jTNPn8?~e};zxnO|c3Zbi
zz1d3*^C-bQ6$wrwywD<|lAuIn=ASz}7%WnV!UIw!770)(RZ0n<tE-2Ue_EzG)jEmf
ze#KdWNQi{sY0<a4`UY4$l6F#w%V?AL+D7|$I3C6fZ_Xxm6oNo7INg-QS&1+<M@afI
z3ri6Pm28AFMS=!3<@Kz{ZGh!n3$JD8ZVo`qkx2dc&j!N!`aX7>v+#&+V2<MBPahv8
zpK-H&tupoqnNF4ZVT^ofe?~I>9DohPbGNo1uDB6Ixmg&P=)i<tPoja3(zqvy^OyAA
zZ<p)+dhcuV2g_1JpshDGV-YTeq||xFeOcD^@^Ud9eQ!;-lUmHu=u=ZzS4#P=oBsRg
zl|>4HOl?~>1c<4%Mno*~;isS0+uge!4~KG?>*+{{Zcb|KUTgjMVBz_9|JU!oe0te7
zZQZoPEG%N{>V(nE0uf~HNl)Ju%#kACEZ#scb4j*iWQ1l!1UyJ~MtgF|Mn^x^L;%cf
z+14sjL`ITKfJf8d2^U5FE6X;(nW+kwA_5>n&Gy)`ua{;%0Y400<!MBV0B)wL&1_Yd
zu|5GMf15DQ=Tl#oRiu_uMbfd=S}XG;L_LB$h^R2nr_<?lc2AkNuIpMEG|Mcta1i@Z
z$h6;=>+`3NQm2PE@9O#e?Y@0}S^oUt(_O>T<b?M6P82^IZj&zt73M19=7vF?{$_B_
zdx2j6nTEDt3L*{DZe>0?Ve9?z{V#8;UN?BZe}yu!^xo^l^E8=iVVO#$$kZWiTU(c<
zEh{{-nFRor0Bx;<S$nkB%$<ejd7kHKp6Vo$(%T~B5gf*z^)~~2Ai{+yPzKygRCY~N
zqytAbx#6B?%h=iuUQc!=1m0`ffGv@-BPM3Hecee+RG7d>!*QhF>NrX~zD^GmPyiN8
ze=+{^2s6t^2{iIl@@vR9GM278huwqBp&o38*9Gd=i!|(sGL|HAjUEF-$a&iVXG#Ws
zlFri3EF#iR2ZRxbGGNhs3Jv=%VE=*wlDZl)a~IZbx9e?NHvp&Ok(k=HK0iM%>uT<0
znmO6c)1l7u!^4BAfpEXyw{1HfkEhe|e{eW>c$)rx9l(AjdL@V+9v<gucJt?#mvFx?
z_tyJ-n2+c4uYUaT(}xfD%PoqmX5?0<$@*sMY_Tov(}yp=|DS)jyxdx|*4z!|ut5bG
zBUXSS*kA9g{OWRT=y@_b6BVIJ3J<)|^Cwx-wqP8-c{K)r4FLOoNjuyKWvXRAf02iy
zUn-11B!E#cT*!-*sg$Wmn5ns{#fZ%rhtRh``H-`Vq?1%`mcLHBIYfA1EY&i=kb@He
zC6YZYPH1d~(rtjLltfB%wJ@lq38$cldYE}v>$!x*SU4mytM1K0M~~#`DJ~wA&riTM
zlMXV9-lHqlEw&Zf$(45b*timEf3DG5Po6?46C(<PDah4(69Sm0Dyt0AU#`NLAS}nj
z>3ll7$FeNzvce6fq>YSbc4m*ZZMT=(<>hj{z5T_3>#QV~r9I#7mt|8f3fZ@2{FTG@
zb7D8S=uVb46Kpuy1;(yz=UR^+&TXWylwi&@jgqb!BK3T(4{zjnjZ1e6e*uLV5W-YM
zTnb7s&`Gs7-|t@=t8+49n1e{&(OU9zSftb<A|ygABy5|RX@rJ*q$L(G+WOZEvLoHQ
z&*IJ*vlL-CMUaL`#y{<X#T_30cVG9IO7kRP-r;`V*Z*K4j@{pJP#(V4-0_I=$Wq#A
zGDy-e_A}kToduHa3B>L`f0pDI^HB3qbia9*DJzNn*g}A4xb4&KT*QC~0J5ZQAcKI?
z+$&3e_LJd}7GDTAb4#iR3Ge#s{UZ^B*aK?5EbaOE<@w9=uYUYxN+q&B0mRIi4B9Pf
zsq^9Z_AQ8LU=nVvwX|knPAyeB*Zph>Qv{qJ9;%cue|mY!chGx3e?LCHdH3$efBX#-
zedLc@Z};AO+s?JN-gRx%<K^@1FMs~<;cuT>v({ZT%y-iw<`Dus$WUY7kxxxT?orHC
z-E#)zDqKZUUKnn~!X%{>;#BsBxvTF>mK{Q+9q(`ioad=9!%cT&o`#`LZf@bZ`J4*Z
zDn+E5nOW~jQe=OAe_q@9Q4R(I!%~U|;JUY-`%OpeZb%d`Eq!S`Ai2v5@BrY_bT51;
zwxwXuhW4&uciIC%y_;$8s;ZXHb=tbd0Fl(wQ#atj5fKUVp;&}^H<jkJxo(@S3$`X*
z*_>neJtY<%xf4(|)kLVKQiVtWVF(4niZGXADm9nX;prd9f4MxYb1l=>`*L5_<$gZQ
z<UKm5d!-I<%k}wwdD+$lET_kJhlh79=(@J++7+Y@qi`CuhC!*=;eem%Mf=7#XPUh^
zM%IFHkb`1qDF(hklv(O9gpZoaG*vkq<aFkF<1ixG8GBJhqLhTAQtNM~DdSh`y}Jd4
z)F8|3r*2Bbe<PKs);&rj%)Q%?e;}Mf_R{7;D9J$_GiiE(g)lf>&p<@k9jD|wJoB7v
zcYw)ecP!GzS1|r5!o!Iuou?uq8FVq8BPLHWBz$jr21q+j$Yh#vzPHz<*WP_o?8`P!
zO%=v!G#p7dj&a_=b|;@#aRC4iN*Y>L1Ok{Xt9+W>e}kNW!n?t9l7PUf<axyG?M_w@
z3(fI5cXC85K5X#;jG2>JLI78b*7eKh=bwK1_>cErpZVxXF%kiDou;-doAyohd^}I3
z9*@TzJIe^R*tU&`N-d?7pfH%4e%C(sJEH{6s8k1CF4tO00BT<5S;K$#yZ^hbtGe+t
z(Nq<2f4SXt>-U%I(}&N0`opKc{ORMm_TF7JlTn9>fsG`{aSA&?EJTu1s%K&8TMEBx
znQCQW9+G1u@RA7}k|>!2>Y+IR(AUWzuU?L;(^P)-t6x&s^?F&BEh1a>9a_V9vWIYE
zY)zBx9x^Uw_APE^dzuGc%e^FTpHNB>PKwbWe>t*9gg4cu%B51R<CA!}sZ0}RzOs9!
zl)NfO@>xR8p|4Ng*Y#QC=z~CVGxALh8-$%V9}e;W7a|Hb>xgiwlWo?YF1Oo`b=8Xi
z03ZNKL_t(FwpQGWjOCjT01Qadn~tq{mOt&itxH?)#U>ET)bID}bT|^+w2=s*2m>TV
zf8=~Vms(YQTUKq2!U~7?)p(QBxv!UYd%j;U7I=L3{x|Rc?&19}jp|QN*XR3s+nST?
z>bafcf`0?Z{#>t=4yn2JFpV&L(@f=Y5BK50MS(r`(#%z%X7_EqF1OSD+JpIUJpbZ{
zU(oUXa_{$*5kk%qWfFsH7>IQ1>$co)f0yNcTer1aSI>eb4wnJ+<z1F;srx89i1MR}
zO6nGKm^p>heq0gA!x08V+J1{<vTKGyyh8FxOvtT|0~k&fxoX-y0rv5mk&D<!66k2r
zUJru2d4khsTrdD`+VSl1ONEJvUYVx89~<NA9-Vt02Cp9nyziJToP2O%0C19tf74Jt
zU^YNpq8KsbY4=8hE>8fHr*lpO1(6#gLM`?_Jq9p<Ftb4Lh!_bCCnsVmJeZ+r4457+
z?zmp=pFce>>q><!T+M@tg;PR2JQD3nfs|1k4~KW}-ql(ev<JE&y%YvR{Ci&C>2*`9
zh)i|*^^d=HcXc~FoZtQOhdRxd9m_>nWS(W7w%%`-t9EUd<@u+lzy9HGpFUikpYE6C
zM*%l~<6-*Y{hJ^E;g|pAKmXJHcKOpEbzN8Y>=Xx6kH@lKM#yg}w$(jA%#<?*_9gIG
z0qj;e$q~yLIPYK{jNcqfMOeG5>X5BQCXZ82rMz3*Wv$QgU`Tksy|fX7Y?Jo-=AKxU
zB}5-A;|oOu1q75fOdh`TEttR#fOwh?-rd$UyWQ@UXcDd@2{`7Tz(Gu6t~nPBbf3x1
zmizsQ`FK7bS$IC&mwNz3<S<W3=F4xU)@eSRq?gc00TzG!=KbIQcv<!7a{K&pyKU(Y
zP2cXjUTZ!5oSRqxrd@mMZBy;;if;;i($)uvgRzX<(O^zCItvH9n{K*!Imr2NF7x5N
ztWVe5$Iq9~A5ELV46LRat?9O`xBKmRy<N4f;R*x<M8s7j9TS9WCZ!Ga21-d=Cm<CM
zl4m`HNoap(xC27~w1d23YNi2z(ZBP0CeFsocNB?8(piIpeawiXl^!(Yq<*?$9DKv^
zJ823YVHSBu?|4X#q#58d7@Qs<V<Yx&n)=-R#`uQ42osSoJKW*0)GQA^+kX4=SO<JV
zl%RCK+-anU-TN5hEny_lMu3<hGQw-I|NK`dHc@{g<QW6b87(tM#^l5))HRQ-_1kTI
z`f{C)$2w13i<<>;?$xq+>#l`)p6A|rTHGAx0}=J!w{`V!cj<11WN3fKZTx*0{WMKQ
z%3(hI`o|xyx7)I=sB$<R&F%JbndW&q9Z~CL`Et45w%hXj;mgOreEQoTKQ1>->7RMn
zzyW_>PwB5``fgb#j&Mslqr#mj*A)q@1Cjn(sn+BGc|=#u6^aFQ4@>8;d>}n<bqM3#
zo5%nB+u!`R|Kqp6`uiV#`su^V^Ow(Gp41Xcq>O4~rWhtkNrH5>42MYB|0^vaUhkek
zLmJP2xY=j|fCwH-d`cXj0%7K0L4Z7pNML^i%SjIY!x*0O$)-eOf!A8U&p;e7Gt-=V
zcBw0bfnnhko{p6?vrvRaWc6WT<Y~ZVu2bvQw<fJ0j)&tk7v{|COg5&FOm)_-1k0sV
z0$knNx~!%a#5^CWS#PbDDolstA)hX${o>V$SX|X@6S68O)DM-<6AtW+qaNnd!=rz4
zrTJV>51)Vf<6nPzx~#f(wSE1Ri#%kl2ax)*@3@J-`_+4@kDFPKw1C5qbnF+zU#mPu
z*ul=U%6x)aq%a5&;yTTzd9IKBe4D__<!QUE+BXMxx7ORT-fq{Y+vU04FWxr}AGv>$
zTIt->5DR#S1R+9Gk&)u;E`?1w?FfG)n`c5YLr2w&*ZGtF9hgnK!Fm9>;UB1U7V6g<
zALhH#Wsuw=lJ`lJ9@vo{@kE?V>HI@2yaFlV#%s6-WkSg?Vd9Zh%da%H)E3b|%CaW9
zWCz5O3qH@Fg+YYJS0jP}<=T$g@J^1(ZI<r^2>Is6V(HZ)KRKYhe<(O2|Ac=S*#6$4
zIpBa3`REs8?4vX6$U?g>_wD0PPj7#Ce|R{Td5+%ALiP1?A^@r7;o))H)~VL1)M=g&
zNL{445tH`b_o3}Sej9%uMlYohNh$SwI&WLM*v-PqR7K=)Iz7I7SLXTNHjz@ebnvI2
zK7aVj$NROftC~gx`Jjz?ByfMrV|_(L=8@#n1R0WZ6=r~^Dzz5jF(;du4N5E}4G|WR
zx8{&??qhYX&zUe6dORI}``aJ?`+xth|LuSO?fw2zE4_XD@b2B4>%FOB7%&MH7IRPK
zRM=~p40gzZgr(HdT7RwbU(q2NFig^R2YbvlJ=4I{nsrr=fcqW@S;l{^DLjpWNr<Iz
zgj&yPqN+)gpgiUPF2%<VHj^o}cU4D_P()!N=G~((3K10&sTMT`)i=y~DO0Vr<ke+@
z$f-_;I;k5=>E3SlWu6YxVLBX+tu<2*n7gK<xH|}L?%mYV;Gve}{<^2|SRzQIa;X-u
zzP0UY?N*?Z@I$R9@d<xg(Tu}|d7h@}tUNtk?;k!t|McZqLwvxtxyIPh`7E_LBz-5R
z_&@#%g!WB)b5pn@47(`io4<LqcmYCcSvA$_Qg|x$?VGpn-oJl%cvv7?P;+0`ySw^a
z<#34Q-j>BZ)UCH|dHJ$DecoQ4^nOEY>@Gwo1*LEm-j9>mJGy^T%2dldm9)Us?n)!a
zF4reKns-8410|UqiS4~+2<K>wJXHME$ZT)eBf`yl?_-1=SVuB82^njXUl<LY0RcSL
z)Wpe^3P<kG%wjlQXL}s*Acni;#_IJD`WCF;FG(s$%!DOG<dwvudo#`Znw)keoyWkb
z2qSLD)m>H1J<ET4B2~<^^YF*HQ)C<ruvA^;RLmjC`{6XX$?dX5s3WHZCluz&lDuel
zZ<}6TZh!vMUmt(*rW~g-&ESI8)@92?*@&p6rqub6CL!}wO3Bn&W|kuJJaZ9h+uFD9
zx|6>Tx|wPfDYcgHs4|`Aa|=zvg?pUN4~K`dUoNM$-=BZ3Pd`09f4*E_*0zQ=OM~SU
zNnxKEM=$`VA}C#o)RQ?Aj<I;7Z}b|Gt-ia_aAS7|8DJO!zp-1z&_D}9h)$EdeRKT#
zU%vm%uYdWA_vcc<?9;^OhvN@Fyfc4(e!fu<Q6?~tLgN0~pdllMhvbSWXUe=ocxZC-
zm%mPsqoIEud*(z8Q@9Ocmf_q7!oC^9STY511~x=;G(E|bqu)o2-EUFb#!;V{W*EC9
z+M$&K=pLO+0E1=s*z*X$OXgQvCo_Q`YME=9=V@kXreNs?cc^-R>NJBW+%mN}-1e~*
z9#M*%PRFTMDMCz4L%Abmnx^A)I8M`af4OhBXV!n^bSQ6TK2;QGSZtd?FdYxacMtbY
z&zJjOKD|6$Z<{93mR?=a139RV#EpTQ{CqHdylU3Rjs`It<%WCX4i4ny3rc0Nwcf!q
z6#=1JXQnp~kCSusST}p#HUX9MX+A$p%Wd5j+t$9V+x@zJ`n-MqqR-E?uH023iZF?c
zM2~;rqX%YG;dv_aJWZtzvavhCDU{%asmQ>ev$@S14+GuZccA<K><`Vb=fmKT^1iHu
zhi7Ewt1H;P!%ac#Ncr(RN?!ZqaZn5d{&gD_gE9?7r2pV}f}HjwrPQCp%(B*#S;m#;
zN_8Kqhm@}h59!l>5*kqViop~qf-rY=v&4Uq!UJIehGXC0j<vyfm@P*D5puAHr=MKr
zLK~-iE7v2LkcZ@5axMS>86g+B=fk%3r!Oy`K7KjAJ->PXCI&MpM5Icn8Z2Dh)YEV%
z_pz2jm<VC1Qza*LV{PAcBYz)sQ%VtLBDB^KG?!UW1Kz!xrx&V}QVVfgw!i-AZ<l|k
zWm`>Kj7Y5UNTwkezM;e>%P?h_K5mRij|(F3zIht-oZSw7h*=Hn;ItkYFx+5oS>>oq
zha$gtGylV{-v8r2{rKar-k(m@eFd~uK0h4Z>$}T!dHVdsBpElK@Ll+zP7*-GP7Xvq
zKLK>rK$==-H{fZ-31mQuyjo`OMpA#VQ)&{+7;w*iv(oDWmIp+PeHtSn&RirhVKmDu
zL}r%RRfCYqLW2X6yAB9ab2*q{xqiqtE%#Ew#HlDTrSK{&Od?W`C+Zq32EJeKt@o{W
zmQsmSdnHlry=@u8o(4>Hnog(FG}RHfo&W>$G|ltzG#?Ix2-|$S9oXK?)0=-o5i)j1
zM6-T491oA@`SJMl>Gt{Z@`u0vZP{MEONU-&Jmj7%i*YdSG>M<<rP}=eq3u0-Wjm5}
zu`rprOu222Bimi^3eQa5K>Gi{12orE0<P)XhUWNfs+P&jh|nN%)#e2RsE+^^2YtvU
z_pVx%CgO|F=yt@n6voZO78!rwI+$&5Rn4l(S}Nc`%z(Bmb1kKhDN$#^R$ZU10CDI;
z9Qu$tCYD<3yws<2eR#BoCpu55Ry9)uaSb3bwQ>gtAw*6w_9=BKG83qoY1L{{Oe&fX
zLEvV+WLFa!06>Dx>~1@WH!u?5YFlQok>57Z+BuYtGJto=j)XWwwUK{Nx5<Tj4C}Aq
zO|SZqT$5*w1EbnF#ReP6?0Vk36!!je7@L9#p#2aLb$~rET~BMZMldq~uWCX-Bv!@C
zyw+Npq@=l`H2Evh_OUl9zKAqoG6EAQp`iQ1s?x?Z^$q1bCNmGm05l;a1*PU0*t&wM
zm-GDia6Ud>x^X}-sfvGqT}+H&wblJK+#{u7vSG6u$2Nc7aKitDx&42>zRdzbrIuC7
zngfREdCf(}<Ddj+$bo<U`KR#t^Ux2U-+%n&??1o)@%?;J6_VDey?`6uZaZRAst~wf
z3gOD9H-I)2^%@!j5g}x)s^UQ*<{D=AOK9^XAfyloq@o&`rHFsKF5c|N@8913=}*7=
z<3If&j;SAGNG#_{0eAZ$SL?bE7+MiTSgSa}#$y7R$pDxTjhNAQN|owl1#b2TDE^F`
z4y5D~RFUf-%aGiElGtr>e9R>hLi-jxPQ>R)M8J%O+qLCzxq@unu34_l0bz56LTp?D
z0(5Qy7&3a$gsOkowga8MU;w72njoM`_2{)CnIw@J5D`ZXip(XK$EPQcfM1riOC5(;
zYb9g?Bf>7G6g*-q0Fh6XyDs%z@5UVjuujh;>%PN7>W^dU0xa`7bgA!Ch_<^q^@n{`
zy)5&2o^lqEmN~JlZSUBwH6j2Q0x?mH{F@#1=60=3%yEB(ZtTiOr}mCa(WjT@qI{KR
zM1fMvI?u~A9dD1J>nYSC;j*lGS+k4a5v5irHB{-cXi*Xn4nzP3#XOV<195UaMZj_B
zQi_B?-s8F<V69TJlx!l5=yMeHcvEctmeeTZjh+kka#LDkk<}*urYlC@%4dMhyvz)d
zjK~__(FlK$m;y7RuXz#oKfTu7h(rNk>%v?{`0}Iuy#SaA)GIXSMG=5=5jg~|joRzZ
zdtzpxX_L{#L<sKbO<*8ZOqH3sE~c&*#d%qM-r&|z?K^IyPxSO@hZxM0L{YsEZL)Ae
z&vX%010g^%1Z<xa!PJ^>wlx_SxKW&^WtrhJ=dyoRDN=Lw6Fjj0<v!9DIM7xkmXc#&
z1nj%cMGlwCWp#{J?YF|{0YFt8h5Yj60Wrp|3n@NNPfML+3aJ~If@XR8^7#1qDKF|V
zf!m^TMBa#q7^}EPz&888mJA}$9<^KV`p^Xt@ue$7A9`Gup!#S80T>~IA_|iRvb&r8
z{q2A8_5IDwX}r5Xj)#8A%hPkdELf!?QHx!Tv5V=&s~EO%3$%)d2m=M@eE1gL*j;X^
zfeHrZJ|67-v3n^wx2R2pbo;w)0`{N5eQX2g6hjEob{JGuFEGH<Ul9O+iDKN$d0HeC
zINjL|6x>l6?CN3pGIsMPrq&X30E{UZgByRyR25dYxv>@mpP~WOQl@$KLL;d9b|CUD
z!L({ErR0+Oo)MwVUt>&>kh_$aFl$W=p-)3JB%m1Xk8k(8u}ciL>`r@(;o&?#U#4l!
zqQ0}xv^}p4EZ)xgR&KT--f#C(tzZV5w$I;)pPh*ED->jt%Y%6@SVg5u1v8W9r-y$J
z?|<3t?uXq0`o5g#Gd_Ksre!+M=kt1=WX@0|0pIq!VJMT>rIb}_EAJ3FaEfW@yVOOF
z*!PJEt!Ar=pd2`0$ON^tA!LKzJ&ks&jkPKM!9N0Q=}2wX;ej(ux>f0BCT66ff)=ce
zY1<<9rq_p2NKD(f9O#9A_A>MJhq-^v7TW2r?P-~+nS&LLHr>VtKI(1X1Oj+VdUZ(H
zWT_7A0y=C7SND4Fuw$gVyVK!#GN4i_e0{pit29Ll0=RRInJNJ)ZCezPnOh5^>Oc|`
ziHp<SZq3w;=*qmp1}VCAg7du5!>`;1NKLBEmu0!EkyEa#nFbCfYNEl9Bc^{q&b2Oi
zZKtDiLGrpT%d*z2wiz-1!>@0oSB5}HwdljwX@Bgxu_vZ^nWpt}OsApi5cuKg>FcM5
z$1l%XN!<<iIy!S8hLD&*tJ`}u3Zrg*+2*k`euHnLb@#)nRlyaJ&3yo_6pW3h7|;SJ
zqox46F21`ze)sOxo45DlZa9D4-wgY{&bpSR<eFkcuv#rJr5L&xO*YiG?Ob2$0APfa
z0uk6+d`4`<o^cP@I9uQtRJ*i5wNWhXddO-!yYV`1Xd;pA#YE05VrHrWZmcTc3Pv?W
zb?nbcZ`jm_K)Qu6_?^U!t!$!AlhRsgZx-5p<5}Cy9?pOiOx(`}O`3nFp8{Hl!Lip`
zs;e2Qh_7)KLo{G)nl`D}{6U-N789qKQVeK9rW`nhE(VK{#^e4EfBGjzQ(3fx(4|%C
z)A{mzndhaNF&KjZm^jnTU_<{53|pm-0L?r2w|lv!vX%4gtatFj3(p3Dw=M1q)1=C#
zb}*^+a(@2!;r-jUf9ijSn_V1U?}Ldh&rf~e6`>j=AfOl%^e37Xg6LWwRkcbLB4p$k
zLKjlsbwk1k6d6pk8WF@0V@yQME3fO~i_IHmFDJbv#ceWATP%3P4BM5706<J^tl%tN
zfCkNju&S!#D}d(Jm~FnZZt3CW!$yI6DPP)CY`e8hcCY2AZSH@;szB%_MlU&5h;Dj`
zwsC%3RpRVsFf$`RbFr08Uh>&y`C!B}4)Ol(bpQGlQOvohS*bc_xt3sU=3pl1et*X8
zxv`mnI)nyD6o`d3?vk2$CeS98bDD!1G_QJWkUhd?&4z$tm-Bpio+3w)3TPpOA}fLh
z#x77mG%Yk;^+S{I*B&DTBsN0nHb9BdNH@}rbR!+o2t0tq=nxrQ(lA1#HcCKHY3Z22
z2LWM(h~xy`eSdiWh37f<eeQFeD=dh!A%LRu35tFk`%Lg8%S<7l`aSUPl)r&_Xr;hX
z$;Pb9suYV`jI&r64G4WE&naR)MScX_{5?rO!Us-dfZIg#j^=kz(JHhz>*7AHoIT4a
zo8i4<oK~!h1ZCM4?q?&{4s1=uf5CD!DY4PnilkV-Uq5>BD(b6%+dbdcK*%WuSlJw}
zWyc+jtm%#Zz4#pk+NnNyCM#^=L>_JB^-H3=WHNe{ZHXbZNm)@{-Qpu@)Yta<7iI7A
zHQ=c$*p8f;2TZIByuTyYZ9?;?jJjm%OwYZQp`S=5z+&Cg|MEnm!<99=zA8gW%gU5~
zIMBkv{=mHF-p1FPLt>yzigz8A8R5Lc__iPz<yH1ap+^PTmSbf)`LX_8gBoJmihu)<
zjl4y#U3c6b%SZkj?Th$#DbKa^QCAOfIWstOndJ|gz8^k3t`gD65r4>jHE<j6gSEc}
zy4xB1JVn-#&)xf$ZAb<dhlRP66s)x)(q}{wABrc_@n(8#@dcaXktxTg_9gq{y$!Zk
z8(Lwvd4|?G`p%-Sg{-QQWu8x4C`JQ>s#Qaf=Unq3It6!<TBlbG3D-ht*QF!@+jw8$
zZW4EYCU7f53_OYnO`efKR+(hH0udq%ey>8|QeNk+l(?9ZNI&-i`Fuf@uZFkgF;f(j
zw!Ds)=|wW1NE<eL;Z9w1AXz)><APM;kWDauU?}$|#~Cb&_l^8Z_KxpORbK(!&jIyR
z3u#6x{kjl%hXif~J$Q|Q1->De<(ugixNJxg#CvEH({q6=v2@I@zc${N?lg~MN7sSj
z@b3)vX)*=>nJ(?MW9aCg66sAVKa(#m>63fo!12-;W+BXFs?7Uf$JzSxQ>qAw0DRVo
zg{RS01>Gz2rmzkmoD#59^#Mq<1Uz#5BzzU<?H|?(>uHhRH@BC=kGdyo?0v;r-o|?B
zX5zg$`MiuYGjj<v6g{-yzPDtFvZc+=DHmdX&I4J|xOhc+A`Yrw-n%N%9Ku1-hU;b>
zIQej3jr|JKB9CGBInurMR(VLMs{b%o(8%T`7a9n}{R_GIeOJrP1phqiu;kW-CniGX
zh;>qCeeVj*@e#`g%`1o#Y|Ib1o($Pb<Z?Q6-ir_v)0HkKuz4zB(tba$tPCp3M8gN;
zw)1;oYDr?stb-CuFkluJd-%qi9!?-SYW}c0@`Vds^%Gz0dnG6F2GTnH!uy}f4Bz9?
zgi@0o`<fO7{;upcO<+TzMJr|OM?zIUk)etQF$ZiGGEz{0J6gl^zBbaSpRY4<GWggw
z`(UBXwbo8e*Y=Oj&*oy0d?xCQjLT=I7nId4sXHy7)VvlXE(UM!toYuOcnuOm8orv4
zYL*Vy$lKW&84;{5kQkZ4M&*J?28`G*JJ)(LS41@$d1ySIn*x(Fj^Q^nQ(%R?J1k*X
z2Qtu8dS1@TFTBF_mG&BgQ>oG>Pq?lnW#oz9hZU`Yt=^w57Z5JF{0%8m_wGRW6Mvbs
z;7|jLuHgN&VMHLK9VVtE;B5WOc?jP7xR8PbUn@RBvYukS_S_E{zMSd#p;6Zf;%E}r
z#sU&M_y2^#0lU9O>;$r6jhPQgp{e~tM9`~V>7ujX;Ogs0hWN$v-MQ;y?8408;EZU~
z44?U5cYn-p0uy71$2iI?TrLj;QXm~WjM+%6@XCscRdHo|xhMw{my_dihx!q)8b4MQ
z*ttaX-M;egU=L!gp?P1D_~K3VMd0QlwX6ste<xHNn3)Z<UW<clhc;F&b`Wbvh^SG|
zd}#DIz!50;+kixyx?HBgh`^hx%j4?4zLH6E;`}AE72d0Q%yI@L=yz8cv+`?arre$<
z$C+30Bl%xW9fZ_Qeo`yI_rI}Iz+;ymt)|<vT#OeK1g~L~)p0YDZ1IbJ+9}am{GfY;
z@(T@sq4H0^g>-c9kx_93%F)ut#hrXRzek4vuHv*lpI=8#Ugf<)37R-L!(!UZNWI_v
zr!T?*>Anc^)UPbLC}@Lcf!iTwwI<U|_XDbY?RdZ|@qbKMcwAYs<BJA{Kb#r8l&<#0
zGOs2KAJKsw9vVJie-9BhLX<OkQvGpoKO6=Q+ci@c-0>146L*17^Mx*p&rzI2;@YqZ
zI*C#Fz(`|qnL_XM^QPeH0<Jk^y24$dyQSd7QVnj`jU3{ZQzeopS54QH^`)f4h`;<q
zu|6SUR#WE5HjSPXF9JGN#`%c*M}DQU8+!0pHjE}k9R`)t7Z>;0RP9)pxi}e#+rI^1
zhWLoN&8T?Ss?m?deQE3?k^vE=u-K0)TKD6ge{xr_vwN5Z<02vcX*b5eMOYE}YXlZl
zG`APR+5V|^hNCQ6D|N{w1S!^@yzX2Cx(H!}j*~X;BHJxNCMI-r`Jrjld_9eb94Fo=
z!SZS>_o+nH3jN`q!~3w^_+$c*S03O)w2=IbA_FQlWvbGSTacI=<`)MO4*2b-RbTJA
z$fxiI7Cu`(!r{<^gV)F5y}3H5Y<Ii(WukZz@=Crk57XtGB_5OUPb7;S)q|BvIxA`L
zpNkD8%yKe9!Ls)rsh5`Jl;a2DnYBnZl>B~hPojf@!$WSafvmYPw;v{0lyT!ID$0Bz
zm_IdkX})sSM?~?W>J3|FObxQFS%~>(clgb}t;<~&q2i#9ogKF#=P(;BQ~JkkVTD2;
zAAGN&Z8q>Iv?YxdNUtlmR_7$)3Hq*7;M$J(<VZjdyk^oHxfx$UqdS8A76LaFpw)ro
z%FkG-LaotXA}E}T8Jp>f1YUwx*n=jiX>()Bv-gxGOpYTOz+gU06&Kw0bS{W6IVJ|`
z;EM}a;(Enyv^r74f3@vA?&!$a1zx5wkPv?)Qm;W$FZHID#=&WZDOYKaz^c{Q|7%wr
zgE=j*t4PW+C_cY<Mxv0M{~?t`-7-7cQ9t@kO(J6y%#C6y_&H^6-3jTumF74bl!~-p
zK>?~P=3;Ul^7g|7VrOTU_W&iSBJ`hJmE7L5xAY%TIZBzKtDpT6tna{rSf98ydZc{e
zj30tgDX3}>F;Q%d=_mZUE`G$L^gxu0qgkOC(6}im`um?cSG3=gtAA)82$)e4cpqn-
zhobmi`{+X>n4`|F261^QEIWa9W?@iJaJSw;S#@SxGbMOCQH?U<V6T)8{Z@}Id}lY^
zIA7=#$@PDMKJMf&asJ8(HxjhoQ(CCDUr3j(H}NoV-H-+it9)@?Ry#x5df9s>?K#l{
zt6}mc1{^f>dSU^Ft5r2o+26aLy2QsuO9^dMZF?vin$#*0GzE?YetP3LZ<E*gaftiM
zLV}2EVAjs-{^QQQn&U$#97a+H=CWurFZ0BZl_!4&@lHj`v!tW1?KN2aK`t7J-n!_c
zAFB<B*5`8;+gx)~@0Dhmtb>$Aqa{25J-{c@nTzAalnZ!EYhYP*i}&Xq8e{Mpfwet2
zva=q&lmV_T@W?)T#!a1-lSa6jLG1nu1o`iae`BL8=zu%{sh8(yUsId)Quk5Om5vVV
zz2*Pk?-%}E3z)JrptNTB%D_AUa^cXgpP0GY7(|@+;bWl+@k9syN&sFHEC;$VUwp4g
zz+&j(0veJzRFl{cdBepVIj#FOKH!sD0Qp#L9EzYki-0ggaqVQ>@BHfYU}F)}i<0W?
zMOK;b{kx!3df{Pyk_3;?<|J}J_4F_fV|LQ1C(~(Z=Pb0KVhtu^YsGo_np#jlSleq9
zPJLio)`xUXrLK+(Fj$P4hl{_}LIijj(k39rX9bBna~hDp;-6S3j_2rkh!Iux^H!pq
zA#g?xl81i09nsi?ye{ZaX-mpD-Rzbdk0F#0oMY&a!b!3y%;JtT)G}Wno*au#JV$>@
zg^nM;K!g);2pcvJ5O8qW^XJ?*uMP|=uIBvA0U@Oim<t3r<5p&N`)+-1(JK0191~XK
zxGrb#?YKHYN|5Dk;!}^JYO8vwj-zi7;YE1J`lE(VYNPOIN7AuA)}BA#zMtcTrsOQc
z^J{NxKO+B#5Sl<*3@hk3g~c9Ctc%JQRgZ+qj?8j9>>KW75#6&?_26wa9=qhq;r-<M
z%dg{Y0rL{T{&KqOXQ5fWOS$RZjB}QAn;|Z;TTu>Vu*V%I+EB&}d%CACoEt4pMfQDO
zy(51~e*8?&vz5W$N6^d%QVz{*V_~8-<)UCAWmE*&3{*E>jKsh0)Aw!&EI7Eo^w2bT
zb2CV^RsGmAiC@H@_7%|p>ArPEr%MXFleKj_KPw)%E5$#Ok?W9zZ??*syfj90sBBfD
zVOURuzh%ynasS^989~)#cjBmf{aBdFVw$@vJrlTwSNdHxuaMTJa~-d^{Ef?oLMJ0?
zDZU98uhL>@SG<tWq<6FiB(bs|o{^#8Ab6<tyhT?sOf3z%MAqO%br}7M>%W5=ACqZj
z1;DgSf%OrDx2uMfAa%3%-l%zRc-Y0?W9&18Va2RP(h0_rj{^1Bb_hn$$Ix@4$(yz*
zP-X01l+)IEtthhio9L4lmGT#jXdw<MA6iASr=Y*u)z;^GbGx{;K3SH80*Q`l+{(@<
z&uF=rpvS{v?mM7Cdy37)K({`57Cni;cfhsx!ny26wG5UhqOc{J32XqyaB<&zxlEf#
zP4hpG87(M=obd?LD8QMfclou<oh84uN37Vao+e@vVnJd28H(F`Bnk7#bRpX@ZvSFY
zz#Ou<;(PZ;^!#fk&J0>kdyi>oc+xiKYzY5IVY;<&?l1#)LhoHr5wK|Snc_1@2beyY
z><2R}kYo8~oI}H2Bs5ky(HYd=17}+INn_S-kNa-!@BZ$V`Eq$ztx*iP`eSb)KtiHn
z(%|&9vX;lw5jOCy`$NQixzFkY{n@L-#}Zp+?Sw?_p7!@omP}T**DM_yIY6a#&x!k5
zBvrf-V8k=MlNskorg-K1xZMAm*MLtS?kNiMRpvO`<UHb7DvFnul^tmgdi)6zO~>s}
zD>dm+Jin7r?VNA?cY8asLmsTED|-h!{5=&Hn7@!uY0<oOij<=6j4lJoUAwV2jx$fQ
z$}t@om=IGM&_S|GNgfjQh84K4tS^=pZ#!(m+Xz?E{FcsqEfBkA<X+vn%fQeZayhRA
zuVk;gi%3c)PR6q9dnF~_@?kg>gdh(sa2Vf<b%4{M;VBMD+MA8(D$d0fbRyg_BPYUJ
zn(j%Z>-OBGE5`4SyJsP(v$9h%C3h`H4k1m-f0*OnEPCd(T$7$S@e(L}=+4>M8O5EK
zNIIx44VK-huj<-%*W|5=1V~|>?=tF4-Ug1RC3J3@l5P~_FZvArs;5^XC5jFy`X)SC
zcZLaGqYc7Wbg>}z^7upJ=dxU6ILIhaN8by1kBj)u&Zns81p8HXs>)fz1+ks~e)aNw
zEORnxVOa8#<!be(Eb}u?)p=btlX2)nebftFkFei_;I3R`4wXu>0m)NBu$-)H+Mw?<
zjEt;IcA-Z~b%BxQb9d)$jCb&}B16r(IgW<5dr)88YnO*-ndyX_0@oz|IgQXP+LwfH
z2?as?9n2slQul|IO3V|YqUPi7NZ1gB%T?<K+5Y9d-rA@{!xR_OiWtqLh?H7!;tsvL
zFRTp=i9cB_ppHAkfqK4N8ZdDZFQWoI_vE=K^Oanz?*kzS$fshSjxg74e2yo!lNsm7
zQd~TF(w+Tuf>*um3yZe7J5hfiZ@Ty+z6k0~f|$EvQhxjC)i*acTE#HcgEvmGr1{vB
z6YeZZH0RJ`unUXwogCFyNSc32qW5{M9+e*aAjeZc`YA~+056-17_v*bUM`<);uhQ=
zR9FrDko$PxJ*fcPpuhu=p+VQv58I<Bc<FoCJBOc9_%;YHDmJyAADJvE5wI1Q5urZD
zocsKS+#GBsj5M(a6cacaRs_CFsJ;rYnZPF(TM%V5kb~oLl~^FPYrCEDTh7z@2KOM!
z&Dh0wGMNd$N`h=(&F9mTuV1g9K=hHQOFMe6Dk67XZ|eOjNq$*m+Zx}`ymlSj_wFCd
z`ZFbro%hQmD8H;cJ_&mvK~%*{oXJE<39)npr*`assa)y(nd7W&hgEn#+iUO&iyWT4
z=d80xN=zj`x%mjZyk|r)=31h&msCYMSL$BS_K{2t*bL>pd`w8p^q%b9>cUcfG5z?L
z>wL_{yLOVQxwF()<z{bM1r^9Y@h!87nI}X+7Av+ELm!pewat!HR|k&6)D-qn#Zv1X
z_+B40^+t);qwxVgk20_G=)b9>s@i{B@;7IHL(ZhfO#U1C_vbWUEU|`9++PFL)76FU
zDOm`#0p@ph9*O-n>K(Uc?FvHJwy!@AEK5{xIgqhymirUA?{k#4c72Rtzqw5B!wKFF
zMHdbuuXr4IhXg2L0d_+y3xVU<1&p7z^}?y_$n)`}*$l1*LruSMvgzEeP&vukx#aTH
zl+{JFwqeCj;if$0?nw7?q>sUa#ewkgs4sC7fX*F{1KTuh-XVru(#H0qw4icEe>f*i
zoJCubsUH0!;nu+MHe=uBthbYHP`@W|i}WAE4^rz~LF)&PXzLlHuN2i(tk>3l{Fu+z
zny9=7s)R@ezGw7*nmp!VbCT5tv8Rv`q5?UQpJ?hKgynSI_R~kaJJMkXXZTTj;eWvY
z$SAED11*a(i-U$|lMN;!ZtLI2;>r81y>t;FArn&E3ci{FZ@>D`MywsXxcsc+AytEy
zNx5G?6da;#>v3uH&h;nU=(bjJf7e^U-hzFiq#dzFt4@pbm(xS+=&=K#f#=>Dvi=ki
zcgChpO*QRDU&ML5fZzXS+&g-gSsQEu6o`4TE0_FdR!Fa{ljLfxH6HVY&XBZLVzs2I
z`zcm*m#;Kttm2ysQ9&y1EOg!ZPnUrMi-Q=E!b8GV7a3xFA!s_24XIFcM#n@B-z5dd
z&-c?pp|*ol%Y9h0Z{ky>sR^s$G`^c=(OoYqb|Rk9tJ|s#=?18Z_|`|wKPChLr3M9v
zGi6j4rZ3FDfTq@@_Q7dH2X+L*M*l5KN?%j4O<=UU@$KxN5OZP+fs4~SXv)ntKYwr0
z)~*SYul@c0dlKPLPyhZ0rB#nkMn0atuQOG(o7;I=8bznbz^5U&DM<kviZWLHXiCKy
zWNqti-fGHkoXE^S(Km>PIFWaNp6B#mKSKydgaRu6`|wkbz3<=O%}K=0^NLg(f~bt}
z2!+?~Rj3E0!o=G+G@3o~V&h2uTD0%hH_wHKukDgj?C;ua3smX1jA4fhUS;b;BR#Mg
z86E+gV2*5n+rcB-+l_;e*IzsY>e^;QZ9Hmc{aZIShUo6+Ap%L_8?nlOh68yl3j?Dy
zQ(YCwjcE_DstC8rj!_umFpK>D6&F0O&J~`fZceKyv~h3FC3b%lG_8D?2<bNy;<EQ;
zwouYPJ&u3&-kc8PpnY?YW2jYIsd-wVFlR^(7ncxmEz~FWPnTQ>1A#mn9xRh+?bDGf
ziP9-4=r=D`V(bA_PbOfI<1AM)`B_^#^XcEj$2_KlbVw733K>5arOMZo{&E982ag;p
z-&h88<m%_^;`qqW8)4zp2o(0by12|CX*@!I);%+nAMtgUNCU1T;TBYBc7Y?|5t_E(
zBPKI58(90f)k)xedPp+u_@2nRrTxk3_AN9zT|nm({auyqu^7O)n5{Rmw~#$})1EX@
zNrj4mzJhs46cp4fqZaw0N82YoVPbuwTYx@ET7~`0pQ491I)68mo?lhd>qEk9Al>k$
zjIF><Lc)W|-n)#M<Ey5w%(MjEbV!gHPu+Fg&#8aO3||ubD7D4><>bVEyc=!6j1)*T
zhoMRO>$%)P&&q*2Jfd08LWRB!8+K0AG)fSNYqK|Izw9Z#@aB#yjv$4*#;7rXDV2jI
z4~JVE+wpzNkOGYnIwaArhYj_&HJpU!2@DD!xlc&#tTu`%`O%`)hTVL7<ayOqpzy~+
z98sPP?Zd1*DiWjWqJGlEZ^unnxGmjJpKm2O^=Ab|FagqsZz&lJli*m<P~>VlH|O&)
zQVs2~Pr!P+%<c4o#P=ul!a@nLE(zDL*qh7HO9XH|+IKSw$b?7a@8I2I*A6ZCBmo;@
zXf8)A#4!m*Gkq|QIRb8R^0%i=Cm$ps)+PsGmtJfk+6=i?0;q$h)lT*fOjj3I2iNSk
z7uW2-?dAD75Qaue<G<sBoqYVQHMq$Qk9I%?N*tb{uVr*>!ixmoGHIWnCLRRs2E2Zy
zwAnr=;mF`QsJ=6OA2~z^Ax-HlyWn3oi-ylB)aO`%`pb!Rhn)IVi3svXOB0|Md!{dU
zLkuBxNwwL9va$6^Ng5pch6x&#-##=-EdX3xAPpY(Hh0inc=Vtts7wlN!k6EYPDa`3
zmR4J2E@<*@N8ZkaH3*G|D6V(=Rj??B6k9fSL>2-6kicJFB8xU6dzuE-dOBjwvcBh-
zGOVk_FeWj<T3jKDAqMucUUA+!#fJ*Fw#H@~Xs>bg&$u`8Qgr+!4@)p^vIH>-IRB^+
zUs%~QYb0!brviFth>WtQc*+YJ^063R`|D0GD>=LW*X#C8Noo?DguEdBh;ItPXHnTD
z=Wab;n`5P#bNL5vCAPRmdEM(wBIK|x$F`XfG;)I#tPrmD@5CICl;T-T?)CT#EXSs*
zYUOnK@+g<QY2yMva>$5AEju-EeV7q%|Li@ZiVF#;-Ri&Sab7}S*{vP}_b40b*=099
zHJvMTsu4zt&B~{Ku1CK8ZPSK=cT7;+%{Ve_RrvR_ncqfipI~1OvL$|7U>^A&XnM&i
zNYhhH6>YUM7HeTK?j#C~dCKwuo8FHf_`^3PgRCX~S|ey;I34#H_$hFJw@kpG!0L|_
zi*KUi{_$;Gj820Na^|k#`l}zaEl=&oQ8WZY;?=5sP&egrZJle!<#yq*;Hu8U6}xZo
zr-$Qp4zQ|R!-aFlg)V08g}_)I>vWUgL06a24^Kmph8+|8>=6;8Li78(8}`=Q7T=}<
z+xn-T7@S|uEU@3+%!K3n{{w{8oOrQhFWnzI1)^x^CluXz!i=aO-ayIz+wIO;<ki-$
z4E;3!sEf#mpR!lE0$r#`A|70e>L)<co#Bpe=EkXSPfLK?>Z7!#G$y<JYK87bEZ!ia
z8{!9#CPWyPluXJGPCh{Nx3CY}{<xJHKWUkdf=)^LN(5Z^3IL%dyf%B@ojO>3XsN1|
zSd}(&cc(~-7sops(S=^r{GusTaa*Ip1qvG~{mEuPQVQp%fkI|!m-=sw90N{t@3mNc
zL;XZuQh3mtIQMJN<tbCxcuR@Cfoo7agcFTJ#}rbP6<n0RNF^s(jwCY1HXkKmj%QP)
z)1DjifrJ7b0UsKC?35V7)4U!@L*W=TH#m}Pns)8G$QqABuLdHdl3(QI?CzT925m*|
ze?L03?O9F@c4RW2oMe{+{z>ZGH+oRP*WTc4p>-B64yq2ckj64kttN$uad;}-<!$>p
z(6=F!bBxMefi~$nbtz5!aC_`Io^|z{W3Hgdfftky@JcpODC9VQ3l$S{e{M%Y`X$=1
zV%b2Ofh(TjE!X-d$aYFsc8v9X2MV+`n25{px{yC=_VU#P_Is*ylJzf?Hv`%)p^(m0
zjM)3&j&KXKRI7cZYcH-_kkKPSDyGb&S`F9zZmhxddD%O^ANjOVb;gPuJacF41M_Kb
zU<3r5tR(4~1L9cs5NQH<61IPdkhvX73htyi3>%bKnL{Xs5YPfE1)D1!O*qc0+Y&=3
zH4|8~#oh<)c=5v#oJ4T+i_Zcx{{jRM=FqVo`8VY;5_i`JTT{s%n`)(er3Uwd=JeqA
zogD`2e~0N`Bz+p^?H{$RS3M!0OyAiOPPgu34j!2~?pzT>xD^pwUo?IBve`+S88|*P
zHYHPdd%V!mW!4CsbdR;d1InJ5E53U7?{su<bMOQ;Yv>Uufe~+a7hKec(<RoimN*PT
zIA8z19L4F`7FXAXu*}5wZMf7h=W?~)sj*%P>J2kox7U^kPB^;xeK^kkWB-u4b?cv1
zuIs`nA77>BoIMM!F`aWL*Z606o;EQfW=Kz43t7ljO5h4mq_`dqsnJ<b6wDN;<iRt_
zN8C9)Z5K^P+CGPRtw%vHFSO?x2|RUq{W-0Wr3G%(-cRLT`tnH@=qP^nS^#~qep_Et
zy%>T`6(Dn=ihgs4?UtNc<q?VJnyryAM>#sF^hrBC=n3f6GON$oh*95)<No|^xH+Pd
zE~y1;s8$2mC7YHxD1L(167Ig?-5S$IV=k`ny9)d$DIOI%kUebQv37u(**FXdiIC}$
z#S|Ite}!x}mB{056xZf29ZU&I>8$*%TeIujXQ{3buYS~W@mqG7#0JU!>ViblR$mm@
z1Z*u`ZvT~NzLuL&Q@A4vQs2Wg6RT8YcMEu=r~#kgHCl;v1TT<X*(NFsZ{E=AZY%yV
z*S>G4$7wFA_G|W@h$+;BNGJL#vgE`XiE@dsV9fm{vlYsRj=0a206__y5{~H3yKO>p
zTOC=2M8$-ixRzFTdzj*a*rcD#s3XVaN5_yI`9Cuuj4%#5F;(aTgH4^O>234g*)LEp
z1K_QA-PxN+%YOxp^j|5(J1+O<wcoUfK2g>Wmr{skAaFJMA>Lu>u$K_ox5k>DP*C8w
zpf2{c3zF$bQaK-Jy)&v>r~h+s96GC6JL61}t8JF4$);V#v3-0vZ)f1~0$S7TCbFT<
z^RTs^EUlK;+eOiGWVTrIeYz5Nc5N<QZZUuuovGa3Q%}6peNjRNCk6YilSZ*fDwIB)
zz{NZN^O~XrS7d|ZW;VoqC9LP=Z!dk)ZHwD1YMG}iFyftzZG2opoFXuPP3r~f+N1iV
z%}@(}HQgvrjumB?M5F!CmqM2Omxd;^4~C*LxG0rn3Ww1h9Y;5R8>K1roUd>=cA)HA
zFZ1sklQnAVJ<`jwvvziphlgnI;_W8eUOaM;BSpA8j^P&(ainNzEVt;Gt^~uSSvg7*
zm=^dLxH~3dSnPZ`L833c!NP)tbCY2{q`BdyGkn2DYAMmVgJ2!36Q4~>QeED?cjY-+
z?+g{4G-|cnp3wE5+=hz~fsMK600!_|5pL3c^V4=O+ux%0M5HW*?;h06<-bo@8>5Fi
z;LA2>WF;~fG<5Rs5`Q{nY^kapwTnU)14f$HpAlA0DT0Qg7NLhhoAnY+x4Q|qTjM2>
zAx{I(PdSHuPp;0+f}4W-Ucbg+&Lz%$)b`5Bz7&rBnMwcOL+e%x?ob7k>x<>Mo5*w}
zHKa1pHQkZzEdI2-wY<nyZs;+N6u<Veq4538N!Jm5M@dW8>*5XpIR|0F6Nj=YDSRYk
z%+o;zM{*6Kn)3Bjc(#%s&rr@@T6|Qrl4iu-aK;sv!^2!~<35#m<xfopR7fi`7W*1@
zb7B0D%=F}$?R+SCt+Ei{vb{JmYo>dnO>9BRRZhc1tCC*pRzE?})UNf5*Wp#$eCM~A
zQuFVGASlOsSfVj>YRXiqz0+^`BV{ZJuc(=QpkH?q0i%>&3xUMhlbWAQN1K#ww+@D?
zGiPiPX39i=1G>cVZZ(-B*N48(_S4Y}jw#TL#K#EXNv+yIo19PpzX4T?Wl-v4H?H{k
z!79gs|D|<KQsgGoL_}C()+;X-87)%Z7V-=a`*S9M)^oF7<9=-ytk>LbPdf|7-`cF_
z{U^Lt-86}}B+-n>ve5c!s->dTrjiK<XWUIhWrGwKYi0W=)$~;GzLtFgqbU63yfv7B
zCxqSnMfCl<*ac1&9w4~Hl53d2?V$TG|AOrPPm{Tj<WD;zJAylx8^@tYFbgk{32b^E
z;3j$bjaF4jeQ>&;|KnHB{Y~$+(}crA%=PX8f9kQH6e9oA{^54e!l^eGb=2W9ox|$E
z86F+ZTF7rjN%zXi;F%*2pB?B~a0stqirM;PWaVw9fPR&6XYz51xl2J!4fhXa1+MYr
zIUdOGO4RwFp+T;2VaHgNukYoi!7XMQ(TUNN&*_blJ~^|t&FR>EE437Kbb0JvxtFh{
zL6&$up}xlAYi)!aZ)p3KQ=;3)m)X7R`cK@2L(QRV@jRkK{?O+TH>ksMrdn5HU3Ga<
zt%ujD0J@u<X!Jq4BU1v+^YqtXylxi#CgRG8-94pKo&tXhpP%854x^83t--XED$3f4
zugfQdayj=t+T8p*x?QMFmrwrLn{F*%6IWd=tcQ;fBb*ZV@%Tj-=3Gn8=Os?=TFUmx
z#3Qs84iwzC282+9C~t3u0yI$+0byvWR<8{Z$Q(y#Xd_}rZ-0(7X|-r|^(u`;<$VN`
zmS@?oGbq)mY0Vj`Fmckj)8jL<p6$>}K~6B^nOQO~3saC|&Y?LiJL~3x=vhiBo_8^>
z+7O0WVl$4$Gdmtd*M-`=q$s|ShRZ;4a!>(wDNx#&H=2u<@s65X&a|b62Pkdvb`Kz6
ziY6YY`?eq9D}wc334;1PoFw@+s5b7of8X;M=>=)-=^%l7cD`j~60VY~68M}QkBRae
z&8l|e!}DAnAAy&ZwOMI~A*k%DDb+N)r<zh&VWht)Jtq-~G7$-#$IpT%`u27eDK(a3
zEDm<>=BIpgNEAHsEFy=s9(eaQeF6|c<Im(L_1mahlDj8dxVpAgicE_<V010029fY5
zGSr|o`1Ih8AT?2U<LMv{ADH<U@7ll~dAYTDDTClMv0Tize0wpdem;F@NUF_}p)}!)
zU+PpE6FBG+Br?#YuF8R;xC~rXWixpBBu7Bg4*w0>%dDnsK+c=eFypzkfLhSZiFe0J
zAj`H}F@jM_e|@$xk|n|K;?L=*b6y5-1w4i?&}QCZ{>#Qk=}dxLHJ5Y=zbYN&X>m1q
z(OYuYNQ$cd#r}sT2C7a7WTLyBQhBQNHugtoFau%PNoKy_+ph1Er-Nq9k`T_BiOU?g
z2{P$kX5C)->zD68((C=A14K`52AWtB%q+Fg`~m@<GtM?nU*ZWmna?Cfa0}IDJ_&=A
z!Pa??44!!86$@5w2Xt{Px{4iMW#?YuNxA8@PaL;U4SLW07Ap%12FG_nbmtLuS!Ume
zAJC}8k}9~CUtaz@UAz5<7uZLT5rDEY#(b?$KK#tO3-YOc+#j_18u<8DihvH3&(FbC
zu8CZEe#_j`-s<GYy=<D+hW!j>F8NCIAIL7MuEwM;#*b9R-Jz0>ux4*++(NT4%Y;6Y
z$!~~Gn<0hQ_ZOWj{fp!nb7+l27hf5DCT~+MR#^kxG6IKLez9nA&%-{08QATivQYv<
z#-mD}F#Qov&n%=9pePoQM;^02oa-uCzo-@&O2A%7w#nA8G6jZ8>bqvglI+~>2oAuW
zbMF|J)<r`>8QwuGr9q!1OX;6drmnGd=C{>wHII|5GU>oAL${dUQD!p)<678X^E4P%
z&X0=foa|SJlRuA1tbY!Y{N2fp#2W1>{d$!XYEfPN+&QxfDEY80@SmIjXi7tkDpqmk
zx^>i<k%{@sZp;e95YCAH&iH!LqNRt9N1L<2WcuW&bEXklqDozMU1`e^sAc|{&gar3
zlV#!j-rZBmI}ylJ;AZl6^AsD*OLcpE1Vmh1g+%sRZA&fuDLFO<sn-zF_-og)-%&AD
zFtDuG^)UAb$Y8?KN`zx7ED<$2py=rSFS7L}1VpNd97I2kP&1ef-@XL`^d?BQWc1%P
z#w`USB1=N$u5U6)O>Mo_7k2Fe$Q)9=rrOv<Okz>1tE+aU`C2w+qN0ja-D?L6Cl>zZ
zRkb-_Q05!GukO5t9y3y&J8emL249d+bKT0=k_tedq=BSyNRM?OZwp<j1;a?HvcX_8
z0>GPUQMy?`kihbw^vLmNe-bHQvE`AB)tl`M!e%c9CFJ+^JjR}%UeEj!?OKzzrW&+B
zElhmEe)4SIJ~ln*(9_1X1CVi`Apz38&4HIMQ2g!A2<Bj88&jYAA0)KzDD1B4@a_lf
zHjsZc6<8PpBglx5lAp+Z68xU`{mYHrZf#q3xtpz~+xgF7k`)$zWCW$~UgbtMCrv4I
zWnSBAn`f)VI<geoat1E0Rgd)ntzX-5ul|7Jy?B!D$t!3?E2*#AYm|P5L_O6)R<$HY
zC`DN~!Cd){4Ce!ja-5m2xZ|kuW(mFA0|BreoQ&5ruCk^U`T12G*DwLJEW0jSx}rB3
zo!hBw<xHbL9@sSmnc`5<wI+x0!CETQf>__ie^?iO&f8`z@&?7*axLJMC=%bvXpTOX
zFX!a&8Lm#Nxd?*8O6#974MN0gP?X=JQ>2=4MYT03@2LV2^?#Z0juxRS=&Ha85cIm{
z{&;h!a;>AJw!+wpiab33D?y=pWwV*RZmvl$9Z%X3)|z9uqLYdp;vtH|`<lAupwi;Y
zm`666OLk8Gh+6l`Alh;)w&JXkqBxMBeIuy50Pzj+j9}`A>F?-$tbDn*>jeKxRZ=0U
z80M$@fBJyu`aY*|YZev=FlXzPk&TRC7#7_}?Fv6fTK@_n)aT&NN8HUE9=bW{4P*og
z*sn1dbZ8Df!Xy0V>f=P0vDeZpSKGH?C4#ez_mB**uvuGZQGE?~zD}bsY?9921<v0~
z2x|Wo1{PI4%AST~PdSipMs@;tzua>l>Lxw0rHyefmhaA}pMaQUrp2gpac<#lp5#O2
za2a{C8OWP*PRBW=l<F175>#VyiClQMOc6e8Hh<^eg^aA6-id__S@+T|L5M1Qj7a>)
z9247TX9r0>*Y3&TuOKYMq^Z29X{TQ_Xmoqum~V2Gr-5tKloD0SxYD``r;6G`1M3qD
z4-?wOhXw`oOh9R74LAeGpO=)s5@W8MMoJpZEdsqm#Y>(Aj#7dem2ciO_jO@Pyl0v>
zGb&u@dt_uwN{T;blZrzQ71Uyj_JsI7Nke;NBITNPxM!$=kn?AUwnc~XvXLl_gM<h;
zrh<N0k}VxKqqCP)qwy85qgk=NJ!=|s&_{IT7$$H6iR)W3TW#V$?FvS3>r6;M&D(ar
z3dqL-UVAR={nrt8d1kCY?A}MQl+nMFNf*T+ML^79YC<H&L*;5WBg1+oq2j4^Fw&be
zp~X9~UMzsKCM$D%s35LD>zj}vN+EH6ZtgBNrbR;Y^P2&r^5@p^v6Rr!Ij)!8up$Fo
z&5*1Mm>{-z0~3`NPOTRzv$J9$i-r`N5NV3<28WNELca<(#U#6$)>?weN<;TG=#5AZ
zB`DY9TAA|vdva$J|NeT&e$yi=C?emOXI^1`M}pzyd%hScjKY|LIH6FjcZ|+Wl*Ic!
zWWY3{TiE~#^%e1D7a+WF<yT`0j@aWOrh8ir>|6?4yH9jToO7xOQfbznbo)}A2y;8i
z|E<*;D?HoWJ;l_IoNR4$|3qnycR?<2ghX~j3^BYlc<9K)iKJg`z|fBBdg^GHbH~a9
zb`5mUv~QR1>qsz_GG$wp6bs;sXG;OM>0LZIw#q8a{?ea<`^a&qk-9uJ?ncMsVg*=z
zLb-`(qGtaA6jv{ukcyLiT(F{2cBCj&(&lxJ<1*FETd-2K4C12Y9v)BmM#<fMc_Jn4
zm0Nn7N9OSJ+Fk}zboc{<>F&Z{SStq65>2SGl4>_Oy~=EieczZCQmm3LkM~s$>StSi
z-Cp5Keg3uz1}R?+{8Yc?tg0;b7;w;fr<aa6F&x8GGZ%adl{4FU7*%fu{pehsDxj#*
zuuX2lF2~K7OJ2j(j*LlV3QA?*Q6lF=y^piys~L`BQF7pU*yJu=J~A(d8)6`X(wD+(
zcJF02E5f66FR3ZC`|+4K^l|X45v+VFb!78!+eMPaQG(+If1sRgj$2kBKvtR{P&j;e
z(=1;^6(P?cl55VZ|HM7CZRPgxw0;4%l7`zs%|0_)q?SNmx4hx3EWmT4Sza_WT>jY|
zon)84hW~8N$WxjKH1{;u6yUr15H=UZq*`|TVLFc_mHT@?-6PrBoA^6D-a4^*iE~dR
z?G#9o6s9MGIY=tRbJW2?fP-+e*h>TFDh7wOQwNe;YA>I8C-o23y9Zw}-2Z7yIXs@-
z+M0Ees7w&gbIQuQxh-+^NF2ZSbyTS=YD{KFC6@ZNk8B8wl^hK<y{Tyq@e`2SBW{`1
z1skm8WBC5Orp+sJWtJB-9Az-yH{OLQ_QB66PsApY`8e+LKojB17JvtSrLOVO76DO8
zg7JJ>4fhgz>2sGuSc%Dj$-xrbO2bKu|6%2i>e_O){l(>)f^6#F{b5}giLi8canXR!
z7u@&$lg5bRV+_s((-08GyvIE4j_s+@T&3*x8g;T&T(1(s<;1rkT?b|ZH^15e$e<YQ
zHMym6fN*Wj_~`Nq(0OoZs>@=ydhm#>M|hy1HZM>;J8xlmJ`mZKN0?WDHtDfyeA6yk
zp)u%N$R;P#h!B(Dce&i!uN%Me0u^=njP=lTJkheQ+?A#j)523ml_{{oJ9FMbWAefI
zv@^k}4Yy=<Us#iT`4>)vZC|Cp1jUH3pjq!Gk^%nTN2{E`?FBb@BK`7LU$52e0ivno
z_CQOM@cl~3+EZqlQo)}vjTBggsO>z~V%|{ta~4_0ENuu~+JI7N<$LH=X2@nCCsURP
z0j)}zpLY7uA-lXxNio~-3|){L10Km6+&y&C!3+6uKISVk=pEpC2N*|Ea7Fwq(MsFY
zKfxP3PXY*kI<^ray7>={p|G~=V<BC1zK7RhlzgAfiQ%wrAt(5b5pnQp|KKdIbcane
z|Jug323^LB82tH;`pi%carIb+JQX?(_gqV_FUZfAhQ?($Xxp*sEUXq1kivwt)c4V7
z!eqNSo0gar-0;<HSykIykjTMVE9q*L7w_MB0NJY$E#>YZQ1jU2Q|5y-^vrdDWF<yd
z$@RfbXz2A7l0P}>c}2qJ#qsXp&iN$%*%kY$YiyiSk??C|uuKuvdG)4!pBbXzF?IE|
z-Vxi{ZS26XYYGeTmszay%~jaV=`(=)2glWbvcSQsS}F8Af8?=siCF)S2uwBP5c6jN
zSbGbKQdmPIulX_5uWbI6C7P9GkD)6pe-Ao~>(vX)hLcfzg1o+;DlnP^QE_@g$f>Bw
zWxRrFJQ3RePY9CK8m6xx<kL?V?8rhh^4KIXL@TH@w>CCop<Mur#abr$F~s|2xLC~9
zKXVmVAk=+6?B=TF)O5J)UYR|d4Zb4*h-EZr*b2<`j#q|yVGh;$f083NpRoERUu;rt
zyDvIMV0g6`k0f}iZ0`SHB#k0E+;f+GmNX^7Q1CD41mQy<Ahz<?{bc+@uWA<>G5u@O
zbo)r8wYWIurX&_ZsAu$S|546i%N;Tq-M8XTbeQP|ycPfCa={)nfB&}i>R|R=6%Zdy
ztrsjW<~r`)hRyflKV*eBoAVOBX^9ACc;?x~d8eO5hY&<HxL$>vM+$`(0Fqq7$=;MH
zR6jyjIle77Y|kf|6ArJFC(XOpi;2)c7wRgRpt1cOBO7-b*<bt`KVZK-F0t+F$tfR*
z-*>6giP)qJGg~oA8Sso+Xe&m@06Xs6V>FK1$_{F3b}G98(h|SOYF8fLA%IBhP$>8N
zF{LH!?(QD@+{~=?-u^9%QC3Ky8LnyT4ZjZSv6;(`TbtOi1#n*0^M=^tkmZ9QSfvGi
zJ9BbMqsqp1xF#!KGP#4tW&qsP>E%XV;vb7f#OaA()9pVWTLG=<x)ccTIyKFWx>GpQ
zE6*GpmHO^Tf$ge%RE27>T{kTIai$bG0rx#4G4<+$nq*`x0g5}bc?jZ=v1p+zV!wMY
zRMdHoP0p*X&5gIMZ957!+`B}>1u~ndn%7#vDWdCpx_9k{Nr~cnI$6kKg<|6!;`he<
zSb`Glmq#QaS?>I!fpS1!Vc(5QD27;$EeQDc_b?sbFt2`7Mdv)&iIIMuX~%(A0t|X*
zLxwl)Z!f%2uanT5AEp0Do}Q`j5YgPRM4ZT9r{6|KUJIvuU_P4uH0u%Fu~4XGjt@>C
zfFw$Y`FkW|b1n}b_wYJLk?!j@MubpbNSAwihyUp%Q+&mtP5@Bo5(t6fbVC%CK*R@)
z(KjuGFDADeNOg$Dv*QvU3@Km~!1I)cD;05HUN5Rh4oMzX*)#2o&Q0ef3UM*OwBQ~S
zTjDN-LFk@+BGJr}s>f(+0x3cKfRbm}rDoVM|4KlPcw2}fku=j0X2UN)nd4I5!*Alj
z8_6x@^(gOV6A)Ti=KkpMvyMD|Y{jJ-jb9=VOvU3k9fA+wSa0FZu1(f<({nkz!NL01
zp@EOio*n6cbVWY%DQ+W&+*f(fGWsDG?Pm^0&6!-HvY*}A*lqz34~<IUSI1JJVLujS
z(mK?3pgE$uaa~^viLA@3Pd~>|1OH}0lc8}<O+rmT+p^c4w4T=^V;0roSKA6=Dq{6W
zq(@`r5Xa=CUaW7ImB7N2PUgpdrPx5>O_PMd=ff6HrjrEMepFqD$gsM*)nzPLX+d37
zc-vw*;9sNN>qoNyjh@R))}G6fo4@s@+zOo!y07n1?HNefr@`ju=MCq#J-oK39630L
z=?G~7QUdM}k5Y2)*-XnQk}&D?^NyB6wjg#nyp7An;qJ!9=41&~@W?2N&SQchmQ>rr
z5+Oy@QY9!Th!hL9T?k|tZq(lnoSQckR6J!H^zfKKU;|l(VW}j)gSu``N+K^ufGdRa
z`1g^g*FTH{O?a;4ge4?YW|72Dw0UG}P!}L^Af4R%uJ>!crPEjGhE{Br#w%tXvetlG
zioZ{T*=6n3u_1j$9|Jr42+#87T;#DU8FeJ<W1Xii;8H?v(l7;Tq7+@yr(>M<4B0W_
zLc|K>ZuxloY)k|TNw7y~Pj_TbZ)99fx)DoI@TB@pa{7^+th@|{)z)UBDmCsq<P3P8
zW)Or+bPyu5&p%5~4`v7clZ2}GBfA)&tajE2gpcjbU%}e|IRI~iyGao3-cCGcZFxHX
zZaFQQ2qfCyi&)d3^qukWUQyf===Vy!lzkAEaASQsOC%+n0V1hbXMimF>^vH<Y4Bc8
z^0sqD&RPB8gWsbwwrqNlLOwQ_0emn&(a~C&?@#;vFempvNu5E-Grd<L`$0+-LkwAY
zG|#EMIKt>&Weo8i#_v1mPxqt@KBH`Zylfhn7&|DmA0(~bb#Kk=Sp$j=N!5+V8M5vw
z?6Cj+{XIv2&%VN2YUQa?yL0Wctw~(Z_<lzRm^Lri!{SJD&0TR9O`TdB%m~ynNd^$!
zRO$TqL)shB#qVBlv4pw1t-(y{KMMZd{>k^G&dXXD#{GeG;LKe$(T_o0MC|vnyCVsa
z7xH7ihhG$3yst*}4mx4&Q1b~=Y{sD{k-X-vq`-GHQCqT5#P9VV#{S3@Nvhotnf+ii
z{wfq4btnj(`-pKtNmx$1tN<Ps(21SPFmsI1=gdw#ucRQ8;m$+*NBfOAXK!9@QEN)r
zPT~(2Wcrh_rI)v*<l9_3<@PTtfl`8@2cq{fIXH<D2b35{^9uVlpHxv2kTSMscAp1n
z-lHkAzb_vic7tC?9Nq3U-M;Eu;kKWPcO}In#IJvEJIeizYUm5c<p6kYdHeiFH~zp%
zv8D$ka@x3)Ew4NTe)}Tsr9zir&ns0$_g!e`pCEAcPyS}7k0{GcgO|AKN3z96m4L-s
zEhCo!_Q(A!dBl4Ky-%L|M_yx7>5?kCvy<<uOg-rRo^w5=`tmfkq0J5ZM#I<3j`9n7
zaK-{_6ZhlZ;o<ULl@XBN;Zo^l^0r@lb)|=_FcBQeqX&Mj&ye?+%8})qp;~L+WNuYz
zNK_X+czn1Kt>oI4ja}JVX~JGQOZ<)u4wm@u1oJy!nRPhYWC!bPE|RQKVK2dw_WGeX
zPj>&{;~5?_>*%k3mv#3JLX@?DPl^2P?kFBrS-uQmGSvLFUk427P<r^QgOxykycLwd
zds0P|kGJ|es;^z>F~*?3-j=r*0q(9#-CsDRbe<@sFGU&2q>364f-X~a54&wj%eL+i
zUDl?=I^>a)mfJe#mom22+8;UEtL}9fFNmP=ZcNpQo%eZZ3HTVel4Jdr%U-7_zF4)k
z)KXO9cn>jz(*e<hs4W=fX7l_znT*$ZUn}iHLP8!>E{C#`tR1Y>GA5;9!;108P;goN
z*Hv)VanO$p0d(+ng<!a~DkqWQT=Ax;i4zUE8tX5ex<+K)63LEwDb*Q-q(W^LzS_aV
zkFmPE#{B5JIAlHAW+uL%`o1<sIhFX$Vr5&Dd!22a9w6gX%TXAZGq#QviWhCd-n-so
z#1`&23sgNF6*zFZe^`7ZxPV19v>BVAKm8)RK&xxK&N~$s1^ZOh$m+(!s;WouA)vtZ
zpSXreHmf%1anRlKqsRgwX7Lw0Vok)+dLz;fBBQ^k?8YRAV~2W92cXX*_Ze`GgNqDr
z;Q_Db0C9S(I4xIVt+=&T!O2RYwA8!22k;6=0z+(uf=}SvIS*nH2O7>+HjL;!ZNd`W
z8C?2g$<6PZYW%|I-`|a!UHoSF{Om03GN{*PQuPUhQYifLv^AeKJveRo)z=dl3M0Xy
z>)q<~L80E>-oxQ(lO{xRMl3_F<PKQFjiKKQ_<H$j+_txSzN#8Mh#MKvJ1S0h{P(Zv
zR;I5PPa_R`^fX$8V|pC*15ndmUwm#*5ZEs*b{}71=5`NN#O`v9dk`upJ#^xvIkYJ*
zZ%b!wGu`8SCe*sUjdU|<((y3KAl0yzKn418iqPJ@<HG5(-x^7h&dE|0vKq(fU`JyQ
zB)+A1U1|JRlL)T+6n%Z;E5p$_*CwzK9ueKnq81i*UKn09q?_Q!JNgGZ7CUM|U;y?p
znS#|k(_}bHxZbrdq*7`@RJ`W|EBjc%dTea^BCkL9&_uVLUHuUsnodP7JwsgPqq#?$
z&6ZC;4e2DS`>yH!=_r~SPv~B{iF|q&xOtB#TfLF{^$Q=P>cB(7^Xg@lAB%2+wOtMN
z^Z;vxAa2Nu5fgtfQ}xZ$`c}=wA!h#1=@n`W@#=+fddiCL9@<bvo43kMf>lo)>2<a=
zH<qYc<+q@;?*q43`S6tYn|Fs6Oe(4C!#oJSd=&daRe_OHGDb>tF3>t($)mCjps?eN
zA3ScBE@I37QF3TwYqK-DuhZt$wL}^CtsxcDgE;jALfVw7{{=`@BCt`kiYfc<ReL=t
z<W_J>kSErDc--N~8e?beVal4<zbcZd>$2a>K&1b{S_%!ERApOk;Ta;QU|j0H&mL;!
zfU0`)Cd604-^bl!yf%kSMrRYy|I<CTFsf=YZLP!*{@3EeyoWV~rz8~)jX^b7&~^8d
zoUFpzK@x?>saU&hB(4EExgZta0U1E$JZ>4`(K;Wxq)a)L9b811ae<z8^>laRbqz8U
z;a!pbBEimT+!Fdnk}uD4FA}_Q>>tbqyjj4CJo~?$kIoEKcnH?5R)Bnu1LFOsoI@Nm
z${I9=b|i%UilA1Pq(<u_D1iz&43glN6da6BFDbtLjVO`3`Df37_XB4z<Wll&)urZr
za)<3dSi)33C1;M+d?*1S5d~)oC)G?W2#eZJcs`B@-4)4O&gmHKYS$bBRZuNBU!Qi|
zC%m|rLF0kAzEBx9pmhEO>w`T-iD?il<4y0PMGHJK;W&n>+Orlpj@&2UQvZ&xf8Q^;
zEGv3~4j-zF@x;<y4c4rzd|z#Va+aY@I-!99%%nn4!ZLA8P^<UU9W(7i?<&gGHj)9@
zXB=)5oiMoRrZO6<tI_`)CO|2@nL^~ufEVp=bUym%EVlz-pUgY6!zc$g{pg6}Nl5)C
zkU4b!W!<GQ61_)|V}dkc2B)Jpy#Y-(@Q9@>HKc_*e-|}@7+ndMr43DTV{!cY;G06!
zNubSk<xdA@f^`zjed$YN7vwhl)=LzVdI@KK1CE@GV=Edsm^}KA=DbnV!KsXM7$aJw
zi^I(jSOd1b7G2FR*KL%@hlV;*h&Z!^M(AdQXvm@(v0_dB^;?rvLyg`XoG&!|I@VbA
zMgncL1U@I_arH~4)s_8}v2D7JML0;ZP;Cl95-Jsu^hx$lC?Jlr_5m?=52zRtV6k>C
z*R|%`s&mSuZiW9x({=w-`M>{TZw@IV;t*x;J+qRL%yMjUijz&Y8(AT%?2*hITaLXI
z%6=P1wv2;gc0!!+y+1#E|AXT^j@NZv&*`YA>kBAe=EQjXUhwYBm_g#WfhbWs_x!k(
zv%euuH*@6{5m@rL)tH+6vU%F9`Wg4whvB0MfK-#4`<7u%*SU*3NBd)HDdJs6TQ%*K
zf1>K0EX(!TwaEMg3kos=*jI5{r}=&VetBs$YP8zCx;4EN0p~*^5i@31no&c2eW^ST
zUb+tG6_@mj+mtt%$$1I^7K9Kw#=3>Dyx_VYbR+q)DE?Ky?{OZk2t!SSC8wpr3p(e9
zpL=pFw%D>3*<t;#^NWVC0QQ)r6mo<=-Xxw^QKBK-tvH{nEg5+Au$${f`+u|eWrI!v
z5pAoWXyq8vPb;auy~~mEfjpGvG$LCn{0xvHhR_UzyBN4o!j6EY%eBivZ1+`e_unIY
z?B(TJh^kaVJGR3WNy=@o%nrR%zQk(crZa-UmsTn0au-4Su6(QJy~n_3EeqY~_npT$
zw}uZakr85g-2PU3u}<bAM)L7Gh|FDO0)LgJPfzezX}bH1vWKF6#g%@vSMLVDj?7e{
zDt^w2l1{`H0qRRP`za^fp3Tv@A`bKn42Eb0^xSL>y)@%qK73A;WyQN!(r?&#4{_^w
ztf)kUlX5HR@#<(!wo*hPgFwOKLH|RLU0hpRx)inRcBw=>VHT_0%0nY{?yoz5gnPLQ
z<Yi#sFeX#8@HmUVH&p$}<h&^Uq(PWc+(5$4qB7kH(4?pLwTwp%<|h)?r8$WoRxXuJ
z2J5iJP0w%smbF`fTWR+6JIovQpP&3!Ea_33>1+!}9-OM|FHF^Khb91c*^^_NOxIP5
zUw^TK8)qHz%cSO{JzWMcV<`K;(Z9b({{pT(0p-2&juF2=x=6B!R953S&2jmY+>ZiL
z;w6p1f}#2(txszW4m~@<B+GGXVd;Fps*zRZ0?2oCcG--*N;nyJ)qQlkdEQ%K9@YMM
z3KAz!nxzA#KBX-*s1FloX!I;CM?J#JAIsDAi}M5t1h(4F1{%lZfxG;abTn(yATO`v
z9L?Nb(yC##!e`6MQfvpotl{us_ICqh<y~Nsk(oJM={}5H;vvp3W7{=mxpi<sWRd7j
zqoLy~{Snlu+>)Ek%B=4khj8xH7d7Mnr9q7C=+=BqDZ=NT-jw`4`K<?KD7rR(imE^>
zWUDY9u0M)yL>+XFRiN0_DIGwTCA(6HmgBXH;^UOjX|$!=H;)3>2cD6K(f8}n0MM><
zzXuMV8t$kfxNw->UDXhBZ!*~)v%m0nYjZD)M>hr9+IDq@5iTT?K*9-<n_DydOP3?<
zE;(-S@>Emj3zrbZy{FQCNJDg!D`Z`~!nE?W4N6f@_bW$CO$H5bVMVTqJ+U8hzV}Z=
z94}+^eM0EP(ee2b0qd84$OU>RRJ)WG_c3jv-8F%h3-S&9_;ph{B=zyyfvpvC9$w+@
zUY+h|$aLCn<K~J$3N#Aq3&cT9{_bxgtr9AKihC!L{yEPazx$7L{D42rg#8K!R@J_V
z0x>zYq#XQJOYqs%`M;p6(|yBU!<qbxp=rfrj(1HHx3gML%WgM}8S?|4+jpom%qy#T
zOzbV(RMws5dw8Wt!ck4|V3W9syU=GoQ>=Ug_PW!ug6@?)U!&;BM%Wce%UD0(iy#y_
z0$Kd2{q7xVcNof@*mW|*A1k?)W>{o2)trY|-lj|YrJ7h?KKMaH<coQE(_ZAtexK_1
z`>c!b^=5SF)d~JYF%}9qaVXq<KyIM_`XOH#HxsB|{Tn`1jdH}&c*maID1I}JUuY<p
z`5|4#N-WRx-=-w?XH90QZw*@MkN@-L>h?&vIpSO4&62Iv9@{IUn@yKjhqn^QAd9yy
z4Bx9e^3cQfKzRecf8}!D4)}MZtmV#kZNcJ=wtx%ia`uz04t)haJ*}2{t4qts2;$)A
zTZ@KgrBxD!29cq=3k<^Kog(+j?!q9BZhphFyfwI>&M}3jq_v~i(#EZDVOZm9QJi%~
zrG+LpqXWD7kdT)aHI>54;rSdFZBHEv@frA!-NGnk+O>eWz){-A+xnt42M1QXBTXcJ
zex~1SGq*V3wCw}CXuOpH;}TRI{n|9FdBEQa#t4f5E8kD<a1ujynFzAQZI><8YQe94
zBu`I<W~SqcNj_W21gqFxpJduG_>*XhPs@Kj$*Y`=9gNLNBXy}B!_R=Ra-kKajWHvM
z_9$efR@}Cq%|&f;Z0DAa4hf%}I2x{xtX{J2BIULr2ee6X;Rtd6yAt6=HE`{_apFXh
zU+P>Q4)~c~SfpcQ9R2S_I$L&!h6bLWUi=LR36fWsE<;n;Qc~rd$d|a4ps%-2oc`z$
z`-}S_Ytqu9Sf|WTH$CoV-G@LV^zG?|wN|NTN2ZMn&tA9CO?fJFYJw{9y)&=?sN?$R
z&%ZyOKx^0jDVVKs>6^9HVT-MVTFq>>3HLi1Mr-MoIiRW~y~l>zs$A!%v{qA)+EUD3
zQPAg!pLiNAe)y?8CdemA>=kf*)UR}p2~U)#J^mZ}J_@R(Vb2ey^>^GU?-!+aakpGr
zO!S;`fXLp^sAb<aX^nh;qbN6b9ieid?C19o@O|CgYq8Y(_EDBqb+xn`nj(vmj>y0Y
zAc3U?%kh5F@Hm=OrQn`(7h&W<+{-zgYRoNjdEP8B&=11joM11OT(pD%KIHsjztr^r
zM;Z-P_%6Sn-2OppXh7YVidyG$u_x1ii#3?xVH|GL+xP_fR-Te4P1oQqi5~5vq}YDo
z_{wW;pKc;(+cM>mB3nmOgK#G9UsDYR{?yZsrB(~q5UFq5`2lW)yv;bVG*j&`{v*Kd
z>{!;PQ-SrG_HGyY3*oS}^5No<%tX7+XQiQ}*h_-{IH+m^{^7d+f!`|;{PJRfg*dt4
zdyeZw4rX4j;ypS4BZ<?NWK{EPjw>L#OOFfs9=N>!m@-oLU?aR)_UXMMum1&FhP2v_
zvrNsFYoXC5l?vi%Ib!-_tr`DmOy3r}74?btDMT|d^43opJ<>D;Qp(3)N8A^`wG1JS
zcq7NJYdipR5wT&6npdyR=@VI`PC}H#$iav`i@X0|WLqMIP9bJoOTV^?=YTenc26&w
zzb9;Br}ILKPbe<)HRut`pPcQRd+4~KFq(2$Sqk~RfKIa+yY8-#4%I~*5hd0yJG0eA
zi6hyCHIfpBhG#k-C{>Sr2iJm1Tmy#@1-*Bdt$(a$j?OKIH;^KSiFwy*1+n1{c>NtJ
zZuN&B!XOXi4malL3pD(N02Cg+s6zcl!7F9r`y#3deRv|mt6sk3^4+g%41?<!X*D;U
zGQp?4U~-@mc2s%uQ!4X%4fjww=*8Fn0{>r303ohDm@#Bb8tiXW%X)7sB}B)4Aa^Fx
zKPvaBkgkeNkUD1Q{xNu}mK$-EyC~ja@b`~h5?xy2qK{nyHXH5@xUe%2-n3?-_sS|Z
zO)^5lLP@r@OgW}lHJ+!X4NH#CsqCR!g9T<99cXzpDCs_BidN&R_=+vrW7Mwx#bSof
zLptM*Qhvz~^sj|$<&WI4jCu_pUTZX(^6bnA!<>%js(GA3!ket3&ogHC9x9S)I4&~#
zP(1XFb^Q?qs1VIrkz~Kj#=vtbJ}H?V=KXhtWlUo;`{KI%YF*U@N%|NvqM}GF;K^R@
zAQ$>~r$4Nl_p5Wkq)L={nK-={`>%eN*loM6&aHXo6cca8M*{o)Q(?NAti_9mCdJkc
zd_u&j5HXNQI>BLT?R;v$O<%msL2w3YsS)1~+MWREvca>%!-9gld=%0?PA@cQBbhlI
zY_QGEwHO?73}ewCGXtlr+q(5!a?l!S>~`>YJAReiqSxwP?P7`RUPfMZ5_^+s_n|@k
zA!**qPFh`R>?lWy)^pbXrlzLsu8zg7*3_;qr}`zRYVbEltkD?q(?^e`gve&ec2#-o
zDJ6hRbpOOC8l8i5DuZX{z3|@`9TmNK;nx3f1PuH?K7VAT(%t2@?=<fq(^(s{D3Pf+
zj#C!<N*o8%P5ksgiWj0mFL02Sg4H(|N&V6Ghg#p^Ra5nkJ}7b1f0T)Anj0S?(NU|q
zT4;@G`hnYbvw9R$D%u^JUaZLWBOi+K9+3m;M}{A_Yld!fQLpkRocoVIFCSnpbWNuC
zWwEzu|2Y49n@lkowPqeYVJG01L!VvP<;7`G>li*~CL#)4iKqYj=ze5-b<Y(&qbN3<
zl17OkM$b4~*5FV6c(OQF)z-%LuN|!2$)1PnPWC<V5#hu7SyZE3M+pOOnmH+%A{2om
z9a}z`SL6qZwV?c=i1tCG?jaL7IE<9|h!%-g5X*3(=#0R8vEj`Q$Shho^IQWwGgviE
z`5;%l1sh!TPl-R2jgH1?a3nD#6Z((6ZM?lw-HBTbDgVVB8$17DrD_D-s2{KY;@guC
zKdz{$sj9S8QtI~vs=i;ns<Zk{229LcE(|B+a-GHYU)5RHI$OSp%KK80v~GJBQ-qd*
zY}1mZtq!Hpc)QQn;FU`nNAOsj9}ClTOn9wWEv~^k#qq%4K?%A~eh)ORJRc3T#4<nU
zC|^4M^Uw48_X6hnJo);gi;pR(x9P)%QChNtI+<%?NnYLv%h7B|R<Mvfz{ALGU;a2_
zceckc%h6zDF}Pfj`xBQL+ozMqHqU(;MW|-hT!ysV=|)&#(9^x+#ucu^L#~34%Rqn_
ztZF#T<OkmR_rLvS4nMiI<J@ZOFj`i`b#yX9|I3o<L2_)rJ200_fbD(7@S~|n4$7x<
zKp3i^)@RI|PD>;4O;jDgx!sffC^8CmkVI&?{UMw*HCvlo$DU{v4CA#1;hd1mi$I%N
z9_f;4f$FK9Rnb;auz}m1t}VyV-f%J}H$u*L>G|Jf(TJhj8nhlgL34MmR!mzTe4X+#
z(mi<M%DV-l8|+^byZ!}w!B~KR>%UOa#+uS9^aflHiB6E3rv_kc!J^LD80WYB%%h8q
zFRR~_Sr6yn*HYkCm=lv9>>4}{a}8@<J1_`Ha37~qjeB6zFu0R>*Wh;R3_oWClLNnl
zyj{)CX7n3y_REjpc57BeVbKn4?d@Qhum1BJm5nDfe>;Sw#i~SCA^h7B65@)2QgKkm
zXm0T`*5~s;pbIG2!<5btTZK?T)Cn1bom21R;KcG5`!pJNce6IP&|(P&8*>$BWt@?9
zm-JFnMepJp_#DuKp4YGH8x9h<wbqWsm(<ku4h~!ljxm3{!)&|*JFUx*s4@<X6#Fe)
z+iCq+7P<P5(&rRf)|t))na(y=NW3!1FX?x<wme|vWe)f`@?v%aLgZgxCtI4D4p%?i
z83|5@`I42DlfF`_N|CZcn$uNMCq&(8XDKPlms~rjb~Cp3f2gimadF{&wLd(3>?vPR
z#v1$G(2!9*x5<Kv?b5&TsfGnkCtE&ow@x@P`>7k<V$VE+7`F2}-;&JZE;RD1#G={~
zZ~^oW62L$ff1~7N1_d#Q&c^P~o|<8_tJD{n=8h_m;V54U$+xONbk+PcBdP~M(k@Yb
zqb%{`6Uo?(r2E2}de0;KH1(>de{CNaa5%wzWSgVb;YgfRDCW}r-iWJArUDk1mDD%k
z)h!XT<j$N*J&!`1Fl#`=moG_ot!7IBhMg*wym-$PZl2El;l-Ky92ZlJ4k^>!vN&C|
zg`B3;Zq@$&m5iLvq9DD&?W{Cj1(X62<pV8)h(XY2-tN_{w#0dGkR%b`o{iH{YO=_%
zPD8#E-?NPT8sWreYVw$4N`3&Uj#FP(6-PVU8?37rmRWeiN?51F@0+}HNwEbClk69S
z$M|BVmRuC~`cqQgYN?Yx%OXoPbhD|&gNx8pp4+ie#e3xrV9d(Ado!R^i*Yvg)HEqU
zh2Ousm;bZV*h7}NFQZuNYeALonN`rP(&2Cawyg8(e=FC`vyC5`s(#WrQpR3hfy{1a
za1e*7uTUwvqiVQeWLSAzs0FxDzcwpDy*hnzk7Bvam=hAI`*AyGM#j#kz%+*^)_vgU
zHhH&^l5v3I-kkz2fbbg`a)Vz_AX|~<Y&w<F&jO<;mUIcCkGJlYxXiDpDXVq^VL|f@
zuxAySna!A|Nh`WMm1Mpe2w2&jzN7mk4;+~>;j8abh)Qp;8GgxiT>@UFHIw}qb+3Qj
zcxd`a3bOmm#;4Zxr5>$;fk7Dl{%KuQu7_5(;YdUQYA490+{t&=N$s8#Y*3u@v^=(*
zP~LvI#93-j%pLyKyxMK1#r{Bn#g{q5ZQ>2<vnV!3r^I|mKAvjx8AU^0#y{molupbR
z$;stQ@_9)V6q>yl(4v~7^lF;!qJ!;-{~NdBYfUB%vZrKzL)F>2P->VL<z!Hkz0wxs
zZ*}PBih}V>JK)ld_5J$PF&N&ltk11wHsxn&?uNr&FxaWJ6w7ID^+du%kp;Cl#aern
zUQ?pFvyVuhsWK5>0n@B9@D2j2GUQb`?((GhlQT0$tE;!q13y5D(F`qHS-R;gL!uKV
z!p2GIWr=~6X8lSVcD5BA-dlBI&kD$9ky>Sr{h<DgF-0NfN$C@F7?T#qKhU^G%`FgA
zt$>AFQ7*d4RA<0%P4N&PxY^7HjY3|}9S*|MTW=v-@4t_G6;3BpiBQ?ym!rOYLpzl<
zvJP6WAlqC7z)>iHEV80I+$7<%^9ZijSh}k9PO0YL=^zYwxMU>cv=kPvRmAKp_m0r-
zzIa&X64y7cH)ZkL)phmV_qxIJ#sTPUT@G_{n!>HVe(A*bkN{iVScqeFVuV7{{0|)N
zJfz@l00+O;+tO19ZV3+q3F3!wnn5991*ZfkFVzh|>d1oijk>FwjVLb2*d~^_$Y7-I
zHia*#mBJAGw{t;i3rSlpG0m9)RDGQ5sptMHp`Pmb8Va}Lrc8>x?paLw$|lp+HR1X~
zZB}L+-zMC;5feToucES?Pan*W+O0-|9<jhKTP*lw!w>W4Na=W}B?h2Ql$@=!t2}-G
z(E#Oh9a|xz=kxQhucWyP&`}R}6Rmrz%tpzechO7L&Y+X|Ce^{RD+$EWE8GM`^dC<f
zjM`}ZlIJ`;JXq-t4mn#}TRRu0o1RsOa=6ydN`l$spo*E8BqDQNHJf`jBpP9-4v&y@
zi%=DEmj{jYxu=hA`ufpXk3JthTb?Hh0jvrfc()&z`jC?QKJPVri@TMVWaHQJiP2u)
z*{xQ4WkOWWqlidRYS2YqHSeI{z=)xIrzKN@O+oT?WHlSNWAl?XMMX!gr=Bl$Ih>fO
zdcKa}m9fIKuUyL%OI{*s=Jm);bqa@Ez!tbD>_haA2!*h)^Viu}$8SCE74>6VJHRwO
z`OccyQT*ybYm-SH;{c4`fhvN^+zRQbuX#E?lKa!o76gCZk?<CiD%?r#XMJmI9_?hS
z|9wA}d6;Wq@=Y8u+{f-WGcU=>e{^!V4i?DE>L4Losm7TnQ(pi6>|8GOyEJnxi`LeY
zs`LJaZu|Oov@6-vV(q&Bfh+OTw?G9$M_E5(DtK#zbo~hrs5~i8<{G$cY?>c2ADyOB
zpw^*^aMtY)P#!_EMuO`PP0rL6G`ie@!vyZ?-uYF2NOygMHImj#orGUcf~T!gbkhXW
zVJj8Ml}xyFs6vv(_KfY3&(ip`eYrF8_R|;E&rX$EtQ9PAis~d;7!xaP0OHF6yDMNG
z-E0^#81oexH(_Rev(A*ES)7cKTAkW~)n8(On<sAKdz^dY%y9L3%e@k|-p0b2p@OXF
zM|7wNA`%L+v=6`PlGmeFH$-5;1%A=%(zFph^TW92=4O1R*ov!9YSu4t`Wuqv^5O}a
znA{IhuVXLQ31@vHOmUh3fQ3o8ll{ExnW9W*>*=k^AIHKgTf4gou+A>1*^i9eRNo)b
zQi2tq_aRT~epz(tXD>8`N|}Y(9Z>biN)8|Tz!vu@a?=Xv=L=>0AEt`aN{~O0_BTU=
zKh>JyhAP~=>%RH1A%RGU-evCaG<K=C@H{5ur-|Ibs)dz}o+TU*T4wvnd9kq&+#94S
z5aj|E%!Jh-0|}==$_JV$&X#Zy1U+}>HsZU2;f;h62j!Au&o(ucPBoR#xfL~#Phuu9
zdxTz80iI=)W|+W5@MssBB4-;RrifDcy@6>NJlxyuB&f4(ea&-kHgLOG^Y<NZKT;WU
z<c4tm@w_vO79iz`vCWwsq3dYPaiZ<5KP-9hEUIeF-=6UCdM7Km{augZ`oMQab;mi*
z*ZxNS<+1Ar@})K#FwbA>m!Hs%wwZxmbsaU;O_$OV;v_RB6+Q?ejtGGi&$tI#df#hY
z(UvZ_HdGUh0~E&n#Gi*gZD7aa>Y$ds{BrH!==w_S8Yt-w5%OF5*FBq12l4Qw=T?V7
z$bDt%=5w01$XnY2sfA&><Hi-I&GWK&pQ`x`8Gvw#I54-8QL(dyykZCa&0QNH{d&yx
ziTmG#_$>*oD0hnR*|smTD~nH}I`>h|6HJsz<xB2z9oK*V4R-(C2>{s3C^ta;3{En$
zA$I`ay)g?KX`s4aM|`Tw?j*a|V$MU{TYXJ)azzwk11-WGShLxBqsge~32#o^>F!Do
zDRg(*j*5KqJo{t)n@57S{n!(5oiAGoQ@lf&1+fyDlGaN>TBbNNKPY6q3$tp1?S!c<
zt#HEK?F;HHJ$~vueUqemz@rnJY`iiiCB+C3Kl_5Pt#mbYBj4VJ-=0CKuubN;B3C*6
z!V$A@Hxy6ei27W)G$nJCW8##2$%m*4eWK}9fr=a1BH|E8D}1;xx;H8{ZkkAb8i9P%
zq9mzJOVaRc_%qpuT~|K!HZZ*NgI_lG#5;Q&i>wGzUN&sMT=VQ0%PdH8Ducc$IRgG&
zErCkW6cdKClgwMhjMT99Y_!R6gY`JJIWsV<!q3Z8!_{10#vh*fpB_PqP=dYr1k#hu
zL0ZQ{t)X)tvb7)>el1G7!t2uA96(YN{YZn^pf^vJ6vpdz^ZD$wYc*dcd>8Sc_jZMD
z=PUws8PKJo4Owm3q^pmWR7X$e09HbB?~~o!y&!wS97Xbp<?V!%h7YA5tZy~X+DLTD
zErsI!h{)fzr<@M1FI`S8P2qhNOI@g8JQNxn;0AK`(QL*H>ufuVt{S$mr($)YybJrD
z;yAW8(EL-pTZQ*$<*woA_f<s2;lkgf%WZWIbIsT4L#10LNYjTDexB1-K=~a7ad$?+
z{q#HG^vlfF>D?H*KOX96rL;I0{o2~jK{9X>ODHFZ)5omRwu{rYKg)2o#O77Y@cm+b
zY2d_6gl<6iXnzvFOXO^dF!?g7_k6jyKlSI^ixt#-m8*Ku2T+UUl*XFJjoDR)tZX62
zEQx3%WCw0SUOQ4VQA-R0h$P*cbzsE^k+jz85D!lmTHO9=WN3TNZdyXOjXdRB$bH6*
zz2R3F+{N$xnXr49cf9fmnn0Phl~CzM-rFcPvLTMe5x8^Rl61($&haH-5G<J>{N1^%
zyR8`?9;VP;AuDbuQAh{@F+4_3));GZjESBbs&v|Kv__Aam<b?n8}pfDA9QJ6Fl^rm
zYUNJlK0IxA{%-GX{-9HI6W8nki}~v*xBnV_OUA>!-<|mDQkWP=7AxJ>uoXLxWHq`@
zmv++CnO>rl$x2@}X(wQQJX2x3RR%x%k>|<w7Au9Pq?t#AaUAg%aVc1&+=&S~qfH>O
z^KpqK)^mHGQymB$Rev&rGF&2UjlI0kx<iz$`LW4=U6mO0<WhpC^skh=lu)9I8|-h0
zb0CybUOg#&EiX*q{|>}0iWeKv12467>}MAn8z}3gPck#X*=sVNGb}H|qsl?AZ>v{X
zIs<O{^n<ZegCE0!0z#hi_#EY)5HdWq%tZq8Ydx@dpc#W%$v7{~m>DrK0Da{#tpCgG
zXtsFk-fyq*{Gs=YL-RCF%0yDz;JJ?ew{qj9D<N{L-Rl~MVPV0BE*voeUy~HD3$C3%
zk?j;^ALi%%ACR;kPA!3eA2!XW^4q6Oc?CxQVi6`u2FeLnQD+gEv$=COiUpb70wKq@
zE`TzyN$?A8AH!7U&ZdpBbE9O<XRqk8zYK{d#`f<7v?*~Kd9e$S{^}|!GKg)z=XnOl
zd8~AGbal>e8e-7XTj8swh9WwdVQM~NZ#4X^GMxM0`tJ$9dcH%yxX2$L$Fpdbn~h3U
z;+{qMAsP1J(-~){s#X-tj%$xKc5YiyG$R0h6@Q9wdRN;@#@meC|H+rd4z?Sq0UWR2
z6paV^A%!J%nWWE(-Yv#yGN)cjYtzrdEi-oGBdqI+oRD$PkP>oxRIk;=xg7Txwn91^
z;b>U~#$L@7Xfdya8``zJA@uy;uddIt%L}u~0`}DQUq1oTufm5QpSMsl%`v1rUo`-p
z4z%g^=#O8s4{6X9qipBI!pTPd8hM!)3m$gMwqw|4Jj0Cmzc}CZx~47qKHAO{4_nts
z!+`B4f&Pp+1Ah<JrvOl7_GhZxpTX#?{j#PB4j*lWK-Oi;AU@U9W!sN+1sW=Niz*6|
zVnKP!FxHO5hw?wy%io*&=!?I9q=7^MIwkdyYMI9BDpDxXV?sOYJ&GK>vK<z*x)#mG
zRlqw%@$!bSQj5loZ%8t6bkj!Z2aVCfRuR9NrzSV}NhviRJTv$Cj_8_U>Dj;M(lmJn
z)kxvi$R&)^sdG+Op@=6eO6+wwz`ke8Qd#X-Xe4Q?@ns!udg6_lA-bX$h-q=$YHpU3
zQsH>r4|SsMmrv5vjmTay8<iG6PI>aEC+q&cQTZUl*`aEFU%<sri*MuIp@e^%azyW#
zn2evZI&uuuXG9{f!zW-pq_;5@iQF@ccdNCJEkS%gsFV2DSArKi*Bvf~!QxgD_vqW<
zv`J-U%b@bu`|9_mQEjaq&?rKw6SaD+=5{!F*D~$ZP6eu7KaX^@YJ{gYSX>K|8UcB>
zdVL)e7)ca!>ihI5Uoz#r%&crC;XR!bVLOo@rCP4<1(?{vmaO-;6|lVS?A2?a{Mt`&
zcvn2zQ>03L8Kx+vquzAxV`pwJE{)vz>E>{zm*UR$P3@@g=}a=nKXN3=4HloMFMGVh
zi6pNKSRlq)r{UVsDbRd2%7o&r`%%u{aXIPH=%&e8idyO-jU=rhW1|3ihmZe_5hlWg
zMz!Nb#O&$c=K1%B)tcYMbb$P}x7U3z%?0dyh-6IS^H(J2%3DDsTcpL`R-#~py><e*
zNjd-Z<FGkkco;C#=l2fZ$d{Xe)1)rEx~qk5vbFvfTj^2xf<nuL#5Zyp#e3VD<Ov-+
z6?3>X#MjjzMprA`TIfcA&z)Yzz{3u~FOgK-GD><vG;R&n#eI4tiSii#ovp39+k|(Y
zAM2Eqj@@z8Q|0d-;kJ%rzr~gqZHXMipX?H{96@{(Q3@b#3Js&sis8(TEH|=0<@m66
zSWQ0>Ifr)m$MN%wXvrt`6S!6eKU_>lAJI(8eL)HkLk>>(3u!5>ZVNyJKgs`IM)&>b
zYnY8+E%x6SeUPC3v#EE5hXsymcwDsQl-MhVzxcD?YH1%fu70f~6J+tyG6o(${iDcL
zu&<FZ1_qQIc=~8QufYla8dmQADKwe$bW0E=LDycdw#N`S%4Q+%W~d?}RxtUaA>{I}
zKy)K+khwTp23C)LHwzcV;+4n2q6e7?FV^L*K%802j}GCJhK(P6OqzF-g~7mz7hH==
zmz<SV3LhW8s1`=Mis+`qsM-sn^+tbwE?#XUa{_v_R}N#>FC{|%5>uK#OQhAM%VF7*
zqYFjejmgMo=9Mn&gXd7b(;fSyAjv^Eu2YluwvNu>{G@$sztF?@>P4Gwr3J*dU?}^6
zKq;BSh;UiU3OK)P;vX2!c5vsi+ppV%vlZ!wa?rZ~&E40c`yNTF<31KFLiiIizkEwq
zE`VLWQ%vEb-DB|27;TlTWKUrTzgw0r#H5Hd8yyr+>_6V(FN@N3oOR8D)Wf(ZC(N)r
z|Ni``_#mKd;Fb5n=7o^S%rvu>bgfV(PrBs;RH3on*h7!8N2wy|I`orH>s2mz;{-Or
zS6{358$IaXsnIGsBD$|G94w{2+_5BtB?1+ASfe=)cP}+h$CjUs&)BB<Vh_Sdid&+>
z-1R17^@u{I%!`o?%{W-r16%y}0#QuSt_Fw6ZoYrZqm$a~dbes^TZDReW{85y_*Cn3
zU|XBD9#VA^{xr(jaw<h>3oew9#Rh8uAitWcY_?<8Jil~za&<2mlw%u2<F-4Ifar%?
z4-;Ut{Q>eHeiTvT;LofYzW0RFnpZn3=9IF^5{hB!xeQR~cR!nMgs8Sx|7g>6gkUOq
z&Ij4UrxkI!-JIFu@s3NK-n0l>!L`2j+CMlr@%uQQjKj&d`Z3E}cuh(+7DpSNWbM0r
z`@WB*ln**yRt)3!jWs;ZayQ2qC}Cin_pUH=e(*(fkK@Vj`olI>6Ka7Nvz9HtQ3a*L
zB)M4rN;%wyw6iGBt3F56%W6`yi6IzSETrpE@QAtLte|}iE2ZZ;`_Ea&`K(qR_)kHr
zo0?}E%IPw8V(;TyR$4PJLMc=()DeYnZ`q7$)bzZ)+XfgJ=AX=<Z=1CZ1m@H_tBT=G
zgfQ`xsxdWOgG;KVS0O=?>CBxpV@eQlZmpRH(G^v|OLsGXaCvr!hi~n~Kd(qom6ymZ
ztNdIbhP|a4Z==;02p<!hI+rw`P3fz<9rDL9B)>-W@A^``<d@H=bZ4?#q?!~ODI9gW
z-%BN1AREeozb9k2H8Z_|wd1<t)k;KYb>91!QVLRW;y?R@dD$OItOE8(2KT8gp%)YH
zF2(6P8>DkC`6nCS?xQJZ85z^&${ZBE5l@6LaH8qp*f@!A4%=t#jH6qnk1^joVLT6q
z_-;7rUe<Llfi!^o%HF|pqTHFmq)j<$z%M$ev!)#0K=)ngu)_}Ye}X%nUAHRHlPUY5
zPp}4E(HSuCcH`e=|K)n_;6l67VQ8f^X>h4QA+N7}kF{tfw$@H9PS^&k!TR9mvuG=9
z8ReI{&slG&zPu&Pd`*D83nseu&emg`%!u=fi(iHN|3LD*#`8;?+Ab_#*Tk(J2373F
zGH}hxNF~JGNtOhPQZ<vcYCe@EX8VjG-Kh-&-!*s3va!~^bQUnz+x*d2^y60uB`jVL
z(^fT(3?rv(U)tXv%|8?ICbN2Hn;;m;`}+g&=MylskiPx$87`Ib@|QC~II?cT#I-^M
zZm!Sh-n;v5<3H8GaAp&^z2w*h!m%f9#SL?5b*@F+H|8F|BtGVARjFRJ$zQb0OwKAf
zswusGQZ{xTHyffk24gag7x>n3;pp~$nLbEna%FL~iLN`2`y%(%%&r7{l#lW6YMbgF
z19+MoZy{Dndt@+rNs()}MobtbrDx&KW;htN{3!iK`CfRBP0z}Rp<lQ6mYX4+5zW>5
z6VuOHzsm#iMZDPJ1v({--(yOJpjX9KZ)_uU_gDYA%6(~ee27g=CMeF&uyo69c|}qj
z>He-fH^>=HOj9cu#j3(o=o2PH9M~yR^trD7M;g-MQ(55K0Iee+tCNZGjPNFd2Z!SG
z;CsdY_A|pqr%!)aOY5Wd%e#_NYDP2)owJfRA5Q`dq0KZ7y5FJWQ;iupMYx!&)0T0W
zkD@Gy2UF^Q0sw3t&Lk7Fd43+)W}7tVFc&?|XRKWW{nvhW=&0Ae1f~qU*4=pgQ8o1D
z=4@O8+f75RgG0?f5>d_t9Slx#Ar}`jFb?`)?@!(pPbG?(dBSbPGNV1g=<wL)_WsY*
zRr0_xXTSUqUf30Kp@xo+q*U@%y)cY5fojTBmYtm73lVX<LuWRB3ZmE+o~p3A<vX|4
zA53x>TlL-NPV+eQ7yV}8;C#3w?>m-U%ha{$2bwjpc!f-+j@z*xFn-Nyces|j2*=~e
zhd|KZgUbLsG)BdE&e+WSzsPg2H8Wh>7qEQ##0u+NkO5cl>rBoN8pk(my(+}GP){Dx
z69$_-t}m~z4!OF6X`;qrG1_oECQvIj+*Q&Do0%y#KK?P!&W2<%>V7AC;Rt3R^=vGd
zI_axY{2hUl!)4V)l`g5@-63F-$&PaYem19QY$PHz^-I~CIgg2##IMrExLA{9fG-T)
zVOeE-SQE=<b3Kv%s`iy1pI4M_fdvA{YHOiNZ`1E~M6q*y78n;}C-p=Dws4wd=#cT6
z<ml8CUgqoC@o^`rkxa(jui#}%gksJb<U1Rz64DnYhti|$rPcamA&|~Ep|IQjSi9x<
z7LV{_D<QgG+$Hx{={<Tm|75~C;I>C@EQNSkxcmQN0>-Zz8%O9Xxs!an0Lv*>Na3MY
zIO*pG*!eU}%iHSky@25TUuFFseO5je6};3iaSui+nH$FZUUesz5o|KXiJwp2R1h(-
zbu#%;=D(?-iS0i+^7N5UTz8*)7mD%M;2sh1i=*!~5dK@obyhkHAe=~ev>`=a-A|)X
z@v6Q{>b<#w=c^FYIDxMlp$i^z#ce*S(35(nTVo;V!+vcRsjC_er<3mp^*@Os{UHbV
za*ytRm%I=j1mb6-@C5%~o+7exD5M9{`s`+xC|{snZz(jf*GxA3%CE@*Y!`8{<TO4~
z4Sdl>r2>pd-hu$iJh{X#<0fNQ4N{gwe}x%<(x3;zqWQCMS|S(0j*LfL=21orXM===
zg`(HlWw}d7p1>aA)1k)Em2a3pY21)Q_t^%=DvOl*pQH>wi_Pi5B?lG_`x?-O1<2F)
zPfUcA{@w|J_3S8C?27xUl44cJh>0bG0$TLJPA=_d+Q1v1=Ge}PfQRYGG5wqGR5=Ix
z-PJxoTb`4v8or?Y&;}--Yp_E)-8^YRI$bo@8Jj1{DrB3gSnC{&Z)i$PRQsYs5x1=#
zmwE@khf|p?sLMS0#2XvV(`JD{1C4)p0uC7nOKWv%cTO)Sl(4S0`T~fpP>9sSI7mIt
zU(Vlu1-Lviy50hU0|;SrD_@s-tw}WxojHQq+K>OKrmBY0INRj=pn??a^qlXSZ2h1`
zv@d1Jm$06-=d&`bAFYEJ;nrHs?H~(AvKul*Wktl&WHbinL7~~$8+hRyDo*$lbA-#x
zL@x#Zh>4Yfx~xMl$NbW${9dv^^e3a00=uzH;O*^K(G!X4J2&smbjXhNBz8o2{JuO4
zO1$0_lN#bP!c_V^CUUJ#|EZq*db^E&!gUmcJ@B(Xow)yt6gTLK`0+t|K~4Ja%g;rA
zmKoMy<rC?6pQ|H35+sYT!<!p<XR3|yVs}IMl9fnZixq-?$Y65XndxI~?6m1J(AG6R
zKv#zQlbf4xW>1^Uz=HPrfSV?64xd)Wg1Gvw0(@FJZ@&(fLjEcCbGsb>+$XA~SzG%F
z;!Uq*=U|nqt-bGe@KRIWwbCUhJ^SX9d4;69bYu2#o;uf?T2g0iLCZmTO?E2bKWCrM
zsCtm^|CnTO>E?2zrhN8=of-f2rQ2`l%r@{(oYwwx-MOkE8D-n2?#|%PeMnJb8Q<^G
zP|cLS&+`f%t_my_=tCu~yv$O`_7v!&XvPKncH4Gn=9cs6Snwf1bZME>iB{q?w%?0&
zz{Ad{eCi}9-?c&^6E3K3)y;@fBt5htg?vF-w0{0vL55{qs+_bJ+AVt;ke!p`a05_f
zKif&3+J5(;lS+{z#*0!lTwRrmqFZe*b|Lno(PqcCVl}xAX+sqz5RS>#7ZEhCHk-}C
zE@;Mw1+?s&BociT^AY_?m?&knelj_Heix<n94y}Zt$LClZ-D>z`!|!LUcdcpUoism
zS=Pu@Ah4S&)=RxeF#TPxJx3dqmj^f)w6XoN;m!K6rc_s^ZY$Gsi&1=FRV6)=&f4Z#
z9qO|b9x=vz`;f9aTX1xGq8D*3DPu}>pIW;LB_kH!#3{_A=;rB3dIKEK>d3#WTd4o4
zp3%vDZfzijDm)~82fG8poh{wn?VT$0p}&(_EX-j=pAIG2!xgaYVam(hoj~x}dM?+M
zZ}9cDh6zh0EKc#E4XuM5d8o8#nYU_Z0JiaWhcHTS<C|V}dfaI&Z&oRo;L0hqvJBdh
zcL?s7H=5GYrY_@)7JnwUyRCnCcdFK&S7qRW$mm|?*ERZxUwlqdM^{<LizA(Hq9N=Y
ziJ>7aN(cVyVQNWSZP&-mfL&RjfQay$8gz{NYT?s}<!M?guyF_Zn_x2@@+M)PN>2d&
z0gk+_!w{QK<8BFhnnbuD)l~?Z-wA9ow>Z9tSF)9TNT~wl^c5L!q@m42`tPX>Ups6?
z!G2Lw4Bh&+-6ICdG1vvagPSVy)y<<Vj(z`Pb~B5fKpUjh5HTz8^wt;Wg7NaHgMg1w
zy`=1cS2KLtW1y5~bg6&zQd5rhTo7TBK+eL!l%?36gYPrYk{HNx6)~S_vi?9^naYNy
zKXq$$bi<V(3SaKCAJ6MQcN>l`q^6BWsBhO#W<q3x{(9Vm+S~8nTDoEx`lgx`>)^lm
zjTusStEOn{k%|C7#KdlSBw99+@~jBnvf+2qtK_}zU?<d`f|OF^9+-Rc)C><Xp=Pl_
zuygW^;Hyvt9gS%k;yl{i-l`YRY-Gd{5BkEkP4)Ucx7&JS16o>x!xUAz^U83bxHCk-
zMxE54r$-Ai1d9$YBzKICwxoXMOw)LCkK|RVeJnKZeK_E%7dM&eQM}0&%R~yfuTb>P
zDaZ$C)}Pe3H5y)!V*vNMDzBg)VCKzaFa^Xs5K*DtwNQ!%S!1!xsc#Xp9Xdvny>qP#
zm&-}9An$@#u8q^3wielbv%vX!O8IF`>xbdZ5CuDL0i*G8ZJ82x)`2V&Ju0oT%B~JD
zN@w{Duqo#^w^$QI;Y(y4oU_X;y}eKZfBXEZOV#eQIY>h&?huV-e!;W8UtwsEB~aY-
z?*wDqKVmDg&f&)2<~8(3M@noM-cqVFsuO8_`1Y(?klT&+-F`T6LT}(<6N;YXB{wJO
z4U${pBs?Tgsiv~jH~22D7!%1X@OepWzdX1E2xle<)9L9mt#KP~K8oHGzdZovvw>^V
zvT~P?%M+HAjxJeI*e8X%4`oQXObnLXo3VL5pfKsFbZ|K1n*3{tuL$b4ZXo2koC|dC
zj<xQuzoJWj8CSC->uT~_*~V_`-0Zjc^VU4O5-!T^5CszZI2Vpn@Zg#Pwaj=E@0)-G
zGo#krp-wE6M5Qb6?C;-zIj*h1^FQag@h?r^jP~TjO=i5G4vU{nB`X80YuPZau6+%{
zNDxW2gsWI}?!-tNX6CU-5QWF|^J5=@eH(f{F6OD$dSjbcSMrIE66y9rKz6>cK1Ahw
z)pFOb&AJx5?zMlxS)}QD5`U+c1W=S>ju!mk>c!gT03BPGPw^Q}6B@p<ui4KG;V#Pa
zX|=7K9PRh-Pil@QCQuBM<?Hu*5{|Ym8Pr0r*2-(UuaxtR!nsZKiAdj^yKY1~vtA&$
zT5W4k)T<?H<<BBgdZIRG(vRSZm{U_a(hf%ECY|*aXeC`kRa8L}L?kO4(CcSMf1Dio
z3VFb(nAx{>8gRCCbvo)s*_B?H0HMKDX>3I%NTl7IIDG{T#6;hLni0d|1r#tRp-*ya
zrW7+-H9k7(!5}q`dvu*Ny}1v@enL9mk5rNW=%`|Y8kaQGd2hnAP2;sTB9m>7Dw9J)
znr|mJ4-)VvA*}-+Nj{4Jm{#wb)j!C$C(qUIXs(YsRo@($t*T+@4a7$4Gcwlxnl$!>
z=}2grzSxC+-SaOkT|H}hnj}D*nxq*d|HtW1@8_8I_5IP3KjYUY9v`1}P+N=&*dIYP
zilFcu;5+(O&2s8&d%r`o61Bc<Mj9_qAQzmrj*i|&uk_a4qcIX<Afb+0fyI$oM*KZI
zAJ132FO&p{EMCZUI-;MTtjDEAie^gDymb{tSMhP5RNN+X2~vn&hGDI_vHq9tmKnGc
z9ziXMQqic#6O56lkNn_&8iHH%Px!myp9K!;Qc%WNdsp)hbl>WhESnKF>anaHxHjev
zir{bKeOsyS3}L2#vT^!PYMKOSD$z7NcmF@Rr(e!s1DipD!%=YxnSzdjd@r~njV_ng
z{SU+bP7%1Sr<0we4x21dXb5#=xI{u+mcyU5p#MENW+K=;Oh*lk6v@zL2-pF#|GItY
z>S$}Jy<N^Dg`MiPzJFnaCE*O#1UI!6!+?T_N51>*alqj1htr=2b?*w7XX1y4cf4C`
zhEZUos;k6+7whYFQTfBS?x)3^9e9V)nN_vpT4gb$((fMMFOzVk?dyA?g@H*|Sx?7)
zD)9e?+&7FecCOGbU>FP?Yo&hAi&ar*=5*WU2>d$RNGsS-bMD<KI^0z+ZwmH_kT~AJ
zHAdWR6$OTU6}uM~4b~6%OXzJ9rUGBKz_rD>y(Rnzm-Uaan|vf-YJCH9Udc=9@FaLM
z?Jy_`5$#`lTKrenft6Ds_BxL326#0dZ(SYl6Ap19)wsM|nQyYJq9(;*;k+yz$r4$p
z6v~A7IM-rm&+KWlO>+~vN+28d-|W5iP5vf1pgNuHRRZNpKca3CRySNrfZwDn2C?1N
z-@ml9FxWmAtR;->%ByV^&!x(dKYFQqpHHZ=p$%%CxM5vU?ah*1YMv4IQs5>thoaFk
zQFc7<yh&VdLIne3O5T|;*jAR|T)_W(Iq*=E`;a~3jbW!xuToyEtK^>Ip(zF&!X2YP
zbLh&u{=b{+SP#38!_NG7NyyB#PTn=sGKY{p(KzZo*X<?ZMoCseVQezd%nt3!R6f=J
z+V8Bd$W8j+8Qe&AYFjzW*L+d3GaC}5Dgs>ovz{{6e9@DhonjI{_C9XIPe422t^5IH
zN=W8KPG(@v7VK@En`E@6x;TE!i>$^Rpq<UruhyMD3FRFQ=4xU#6y?`|o9-!3US7Hx
zL{Mrbj&TUx`1DzLIH=23xg{kCn}1mJCj3{v5A3Vx$zVStr0g8-;|11ibqb=OQZ}_G
zo6ZU!GX<fP5IQQkS9>_V_vUoQGpI)T6P+HCaB|K!tDz+I(4vMo$*iWIKW&Ww0XL_l
zmp7}65ISQD4^cTLdu`Ka0#T4MFgP|)Fz^H>j59%!y@w|lE!8Qhb`aJ^30z>#P{fce
zjmI$;+L)gQ9C->fU%c~crp`koX7cbkEHz|MXhjA`hZ;xHna8qnM_{|!@-v#igr5j8
zzJDTHF8px3qD`$4{<ot=JD^I2)iQggkr<`5oq^s!f3VJC14~vuf6cmZpL1t|Otz*d
zVk0Ah0~PFQ@M%J5{FJXyiAQegJDKJ{;p;-5<&i9!FPhJ_Ff~U-j`l?(ec?TQ+i@M5
zMNcn@m&n82t7t1nhTF#Qd1*#Ao*Qr#6={~#NM1p6e4D|P-)KS{pz~X+=0P62bcc%R
z*m)NhxJCypjhDYL_M+}JX33A&<CB$2Z+e`S=`8y6Y`As<Jd;Y@J(o3=;upqaSslBT
zmQ_1K{X?%5^FNkaxV7+VIOs@9QEYz>4OO@|CbKPMPZ)GlnzqoN9osw|MqVb`Z#4<S
z7@KS}5fjs=Wi(lU=BxNA+&_GqqX$3S<yR-Zet|QYu#g+pyq<&UuyDTRP9ti*=1T4=
z41ZS1YO<Y{T<`62s3Y(VMEe;hK)p=(5Ol`uE39?Zgym)rt?BOkV4!VA+ZG6GPE9kz
z4mEd`H0{%Kz9~6+72VQU+<&N5PF*2|aFb-o62&QsR04vitn6^62u&WAA!X_C93$p5
zToG?FU#&@e4wH0%&Vs4*`uq9AkX4}E*YqjnR(<zE^4K$0w)XCBYar0_-&)2m`cUaj
z<@{Dj3{(SxHT%Nq$|#|ry!aRFlZ^>^TtfH+1cg1eX>08?l-TTejCs_O(J_V($p@)$
zW(_?XAhXDf-d9HptHKCc4@$@Gb#()a`)7np(XzR-rGPmQ{~9_DP#(htDdu;FT?GYq
z9FKj(d^0&3tC$bzO5&w;S%tSsN{Qh)R=O1SlCVCal?4UZ>*Gv2wXloV`LhGkGQ`E?
zHzhQ*h>HhWC5<boLCvwOo8Lm|z_$OodN|P6heqilpFR{{qJ3_as_uN#m^4nJ?1KiS
zF6a)pbnu@$?Cb|&AEWL_apKP}ZD)Td1UOcxe#?~udD!dY0DlI<3L;6~NKATUeEf9Y
z#{$r`p+AZ{-#ia$-RhV%F`T<`p~|SCcN`$!6^K6t_Xg!IU}b5~W)DIFY5o=m2+pY;
z)s$J_!8oIjVwiIHi$lH(>DrmpgwQ|Byij7l2N`dT#H6@+8($mI!=SBGJ0bEMW~H;g
zLbjd$9k(}g0e{!dH-GP+C+dEFUe-8iMH|lUR+o+<xiYTFJ!PtKT@BUPhy<HSr<hIR
z1TI#y0)zV*&C<2iKS1NGh3Wy|k_v|P&M*719QbEhJC2X#hI$X{dtC?7k<#zAq;!MZ
zw)r^_i0<g+0;pbcnG<*-FvCl0^RP~D^Opf?I>*5a{UZmrqev5`cX@Ss;MJ_=&U}6=
zTwGietBE;uKe*#@z*Q?|WKI3l-Cc7jO@Gx1QUd)a)E=0yD*MsPr^f^Mces+#Feg4_
z63wZIwKi}VMZY7zV5c;NNZcShZ0b&kLzETGm6&KQjaZ0ywb0a87Kve}q-_$+;#FJ1
z#~xX_inJn<DT_w@3Np=N-8p<LH7O~N_7_YuLE9SF#e+|QA=-KJvGy_asVSphnZKWe
z{s0p_+YH{n6%<OX95DgrJUJOdDC6xHKi{Q<fO_XM0^Hd~zXI)PybfJDXj$N*$0*=B
z8o*G$6m`#cEfWp*VKH;x!VEqo9|W~d3qoGb&NF{&JaGvxhV~r>#Q@j`<7DZ&Kjoj!
zjCAmXwAHLwyl&?;CcgXPIiD^X&RUU^eZS1z*9L27A{W=QDG6{A3b$%7*72Xyp<7;y
zG<d~v8J3Dy`$N3M;ln%9ut>J-lS`=MXC8sS+f>|3iyUCmQe^%Z7N`7fxTFxZ+i@6s
z1sW_E&)P%P5>=z7d<BN9cq}%r4n`N`Rzkt9M{flm5hu)zmO%L9alGBt?;W*EK<ye2
z`~`o%%CMin@oQ{8cf<YL5%F$29wZqecRsnf>x9^}vu81xP>K!|B7O*4e;PF{L$xf~
zeR`)g#*B(b1s!(y+kmJ<iDL6Nck_thwi=%5NOO)NBIS<rgIKj|oZ7Y6^#M332A0;)
zQ1XQr<YoqfnH{22$~L$*9V7Sck6j_gGpnI2yjeg_1BzC&djUjSb7SR_o0F5}l9RKD
z<s;pQgPoiH6Y5<yFe}Gtx9knLXA?`+BAeb#MMklPv};&-idAQ%r{ARPd8;k}+VA~r
z^aVy<QEvzYU7sEwmmiLoXX;mg)N$KbyO&uy&hIzOVm44fxX$Hy_o2cSr`RixrW|ob
z1;Bh=5C^`MX`-gPIl)C0V*}RNm9viUi!nx*;x)%&+F%a^(iVpEFr*@li{ea^Y_J~{
z@>$kH7Owqw4SU(4L-iygetWZRI{c-quRy?xzp=TSfM0uD7Eut_&7+XktAA&9TA)M~
z6kR<MKr=PYw}`<GJPtj$mTirUoNcxM7gQ;?I%TRggsV5Y@5hKw7NjuRWOXL`QC#&3
zc}_~q{(GW{6cg)(S=GNQN3CG<*j{>{MXxRdYWHR=x1PcZwW$8?neMlAM>KJo+VWxy
z>sa}FSt`Z}E$^{(eW5_Qq>B$GiIWuO2abPKk5W0$p~^(@R8kljB%NmF1?(Ky7Q%q3
zKn=QyI|1?;X_>9gDf#6VZVVd}gR7xWAP5~&>`>rgCj}|v198)ez|K4|d`I1ftbhBL
z5!stl7&yWyTI7wU$Qvifb9%Gfl!Og&twdwXpR~^j4W<uaB`GGfJN_ys-A2V+iru(h
zP?gVSTNOt%c>Is|Onz_JBiI18bLeD`Iv#E%6-p~`&in)tTG3E-zP4pM*=>z5Hl3Ei
zR2OkJd7PZ~dZMactLAeWQ#+{9Yi4TD;@ir$t$GLt2Tiz(sZA4P1rLmzWxjTTqj%yg
zAecqtkA&k<OOW<0ve=FMd)Tr-TTC?MqDHEI3V$+;&AdLfySm`I20mW@#&m}rX~_*9
zU$(b@IwSXPwS8=qxznbXx<xy8MKb598$QpYjIiHQIiE%JiwJ#<nvImL{#%i%ZS7Qb
zee0&g%fSjC<_?-=1*}1QTpe7bRMxRXK!H|5HSD_I=z4c0<a(-Gad|nm#v`~|RtJtN
zHnFG}_EqUr#&iS_125gOdE(^hJAEk2gg5I}z;M`Uy7e%*J^!V!+PEUr$mr7N+rJo^
zB>s=4^A4x_fB*O~lX-+tWQ0$Yy+^i^RUHmFHraa~gg0f+$ety~IJWE~Bcb9L*<@uL
zvK{*vzxVf#-*uhK-|x$N+^_q-pU>wb(b+r$+`aW5JZkf*+^j=8?Wt(e3uMwmvzF1Q
z)ZFs=ZvmvLPIXVH(kPp1=x@m>Bf-&D35S-kb8>4{f7@WxrJj1#WL&g4Zq+A$tsxy<
z2pCvWLI8u7=_`cj`!X?n-D_l?4JG$u*;^w<Z(#&fo#fD0arkBBl?wrO!v$YM{6Kk+
z>(r9GN4xUq>`(&BoNPVJ^1GUws?BKu;?zm0y@cye{h{N(1<SBr@P%|by+FMs%BE{C
zVL1=$X@9H~NAv|UyyUSb%A#ybB<b^Mf&D}p2@rA!{TmTkcfI;#54$i>vb?bGEV=kl
zp7W&i#>ulh=&+QV<_F2y&n1-9I+knEe;U&+1$W9P1O(tO<DoPrCz}tEH=Fh?s9)8J
ztyDbHWF#o2benn!jgvwK-f&FCd6m~ceWysi`xV@v`<Q8!d8_}rOPHV+KqMn#BbEEE
z0I2?059OpO=&o73YZ%)iKq2QNB!kNV(^OxIXgF8a(WntQeQSUdJS*`MqS_RSZ13(r
z)eJC0)g1`%(MTnP*K~P~7&LoT(%z)L84NAsMsv>kteUMyRdK-1muUmP?Q-lUVXpj?
zc#^auwAU<z18#*669q7lFd6bPo9<Eo`#y@eTq429#hVZptnJVC*q*D-UqeGBpOV07
zb9P1#!`3k&z>HKm*Bb(Ij>X3Uy+L8`+%^_`3Bn(ID|=qH+jauRP?4W5jXuTw5lKvP
zpMRiQbtb0UU$>Tvn%Izo>+daXBNy0@`xFERKN?;UggbOb!zuWC*Mp4vTY(jekgJQc
z(Wd@0v;Zb&pHOh`damPg|MJBB?MX+->8S-+E3S`lbfb=jOm4}mf;_)f-Dx4lQ}JKl
z3Gwa*m%)w0r&H-6LC!M7a0WxK!4^n#52=*V%~Tlu$9L<O?Yo#U)S1GhubsicJ$emw
zlux@n8|^M|T5_UvdA5TEFjF*|J83#}vofTymZVZ<!WQj8Lj420xZ`nz6F=`Cw#ZNT
z{P6za6RuN9d!p{epKoSaiYb&M6Us@(UORNx4KB2YoF8uB`p>+<s(5r4M-1tf$wX`F
ziZcBwp4>pR2fxWKHXci+()=WNp&nu4dE@t{ZWv^}S?RpsH+COLXS4R^alPqDICED5
zKpXW2IC}!=s{N~8dDqR!=BSkW2L|PtbzV$RGD(Tuyy#w*t#OMf?gYpa+8*l0F0xEn
z9LN!=N+QhVx0sgnE11OuO+|p&QpAgOLIv%QJAzO2tNI<_*bj22;PlQpk%aSh-sHJY
zt132NBPw+s=m=(ugivUg#jcA|Jn&y2C??+c*l7)tF$_nJX&%Ra#^;wqAYpjN=w;FE
z#7`V^KBzW>GJ!|0h+gn#vdm1(T2;rh&E8^iYNnB%Nm1M;KRq`gs@SG;8Mi+U{Oztw
z$7jd4#%^6s<UV0cN+zkQ_Coo#iJ{#SkzF%YMpc{|@M>bfdj6~%oJU_G4AT&6s7C=h
z!LzN!ioRu=1h@58U*moR#Sa^b@UZt;$sSD$EsJK<3t8YA7eP5G0AufRWBY}8g%%Qt
zzaN)7gfsn;1Rj}DRuo<GO%4AZiU&Bo1pCvB$nK^0*3!?t3sNSuAg_fp%4(wdQ;^Y>
zH#L9=Rtirtj|bmk!E}sx?C?vQ2N6{#qC}U26k{D1Nw@x<EIcMvjRL6&{igYA2d-rn
z_F0vF?{0{%{^ET@`g}bPKVK`emNFUtkFe4eK#oM5c3fR>1yCl=&$qJ%4*8z-TLAaY
zWoXSLO<%1w7>LI6%T^aEs(a~a^k?UNhCT*lz0`vjaFR`#&nZ4_g3qd>nNuaY?ryJQ
z(bi<<0iq>4&6iirlH3EYV3zSrrN;%UA>eJMCww+(p$I;AM?USu-#X4Gl`aad`@nmB
znRgNnxef$*Z41E3(k{pKGH?ojryY~x+rqV#c(NCkA8br1%nr;ZxvcrqoK^IJsOo2~
z&hH%@`zspLZS9zXzt$JhS+Vq8Ue(NQw|(}Xa^`4Q!`6RPXE~AXlime_@R!e#?(@DF
z61fiB78AdF*dOV#7K6`+EI=<dSjf&#Yw09>%=Y%Z^8GOE-$bY{mqd=u)Ft$>5hb=n
zxZr<G=ASg5Ohjs{8mRk_hEo9ja>3vkr2~Z7rUl&#-2E9fpVOrvJitUm9MP=pu(J%}
zsRaeEKUo*63Q2KpyJNc{Q*N5k%CrQ<0HoR=bX<qA$d_a#%EqJdF04S6V@}!~m{M`a
zUd2bxf65YJHiRBU`|Xq3{YDPTrcJW*xHQaFF%8~jVToIT&D>^&0t5A|1Mf|pwe%&K
zRi%&DK7Z2dU+uj|qppkSwh{ku-zDt4H^#1E({^z4-&gQ3Ub|^M85O{!#=F4fAH5b8
z^5>H_{Ln}4ae*GUS6<0(`y1r7$AZAm?+<!Bk!?!z9hV0iYPnUQ^1_bZt-TKI%v9<^
zvDm2I6yt;`DRmt&1m0-azgRYRx9u^p`i~PFuP{JeJAO|dE;ar(QiCwe2C9UH24SZ3
z%Pb+2j%OEV8(16xL%6I_Q;roarG$LX^uy#mAf&0wq%d%Npc|;0NI}JV{LDjEammG)
zf`TH(HjH^y%L;x~b+F-|DNccpG%cmaK%ZK2wpFO#bao9;My>fuz8IaCk57cEXuOIu
ztmF~66SOC^D3v=J(v&1bj>C*a?ImAd1(;{S5xX*wY;>kFnV&-}$kzyaQWh*pTnI6|
zJ<eqQP~|G3dci1cZ|y4G+dO@>ZpB^yN-k{_jccih4V~u_z_$*J$WnB1(Wjiy4u9}o
zVLSV{UI+Y@jmuBxIyd8}e)z@1yc&$PQ`kDkU~*daQUeg$ps>5xO?fng&3Zu|hme-i
z;IKZ-L88Z6ZRKEEg(;Jg)#V;vSx=_A!f9hdV;Q#9sMpC=NBRC=OVIhB6{u%{sp&+1
zhc$R$rJmzAR#B;ceD^&tgwcWy0=@$X0FX`26@#@4N4pDi_vmmqTw7IuAlhV{Q)RZc
zF_DU*>%<I~M<Tu?FhEN*HzD);-`CgW$y23pn)d0xb)oC;cF)_j`2&+aDn9Xk@e!rN
z6+T7RJ(sgPf3^2KX*Yx|3Hc=C;f+q+LAbbn*4k!N635kf#`VE!xA3M<EK(ZSrqvDa
z9!4#mEBf(MtNnw?{coYk<txN{sM2^L454`O?n4uno=6$Hxq5N2x>*fIu2UQ4Vps|D
zgpcS^G12n-6^~4NsEZv1Bd`=BF`1vv)^qWd5+m7<Tyu&d2BD(<l#29HU&5X(Kcti&
zW^~XTy#K*2Wi@F-f&`-MpaD2(sylHP4ut+9pOt(%4$~U=ZFlB3^HpFE%U8tx4LdOt
z6yv1HsH%>m7WI}SiEZzrC;8ls_~InV0C8mIKNbAoYH1QFbhNzV)t;ZRn(=%AT-<E*
z?R!k2Y?&U#l6Ga?HI?%r+&wg7!lDL7!P=JLm=!AG!~dQ3J=?9`VPMxS5$S#`O}FV|
zTpafe9#JU<sm*+=mH|5yjhs-to>4fSVRo&~3Yz`)a{H@<q!na=O)X!6A@1Mt5m^V>
zSyP>y`Zs9sp^&d98v*)OAWv_zZEREF`h010$xde8#~s|3kyhrBQ?pr7t1z_$>eDEO
z=Tp?gir{~tJj!_c02cO?`#3X-dFcbbTcuk3WYOwHB283rVz{<KW6FFsVni7zjCiLO
z5Y_e4ODKTZ<YpQjU+*Ig0m(r3El8Q6$^$bkQIux~rlsTSz*+xji_-+RVDOZ6w?HA^
z`f!`G=o_9W&EX%U>tpXVT4LRf0<M2g?#i$49xpSw<Y(X!K)ZRmFoQ*=*U~SsoTFOt
z@wP|MLK2Gn&Ve@>=Tj$U-pP}u#f2=ENS5iYQQH8}&MzxRqt_o~?(KMS3{sC*9nBJ(
z9NGNT4*EQ&kRv%+HDScC$)qxER4O<gX9hDw-+1kKDOw2Qa>d_n!zhIyn{8DN>1`l6
z8HouoNC!%AB_*SAkJ$3`l@0oB_veqyBr5yVfD;e=>CPIVV5c{D{iF;rVf-CUC9`uL
z)kgp+&iT*v&Bff_38nn!!DGn@P1%G=!1ak<K3Aq;`}3Zc)<UpvyXO6qS8fT8Jg*@u
zrqo28sSiKZ)oIc5^A4?C)L>-i7aq4?UtxfAJaCCz=nIh+ltnSl<c__+$;}Ne?Jt@-
zF1M8~@Jf@ut*mdn%+<979oR@{g10cyoW;Iui54Bf4<9XDo|d@%DNh#QhO$89iXfJ~
zzl)gQ@&yH81mx+_X4K{pFR2Y{XrLPO3(E)OJw~sp%>&ao5L6*mSsO;A7EfbOZv&eJ
zMC6^&mSZetSbV(GFH>!P=fO>dA>oB~y-@{W{K9n~aJ{$GedKkvjDSKR$~43D5481X
zl}+O)AfW=KuPWiM?>f>TfV+i+r#l^CrCwDiMhFRy*WlUvtp3(Wp$L{J77b<`bXXlN
zAn#>jZP~wI{7$h{<zMUWd`Edh$R%)bw%jyo0b+Bvl9G1YcfH}+R9`@PP+!RPR048s
z^cM__Wl~Toc%JbgiR*ao%3~D8)l=jodeoOgh-%ucDgp1Gu=NWB>)nSbRAsvaM(c`-
z!);cR_G+@X*EoWKcfWhgy;;p^SJVHz2LpD0#(Xo>Zw(NzXn3OTo&S3{0qEP`0r!2g
zQDaA&h>)wz67W*a_ec8|u8zem6JD}kZXAb*Fi3!zN#!EtWzmoDZ!PC#Z-X+7=9Ram
zaDi7GYO!5NGq07jah(_l%~x+_n&mRyYDs&3aQ+BY-4hBIi6{9|!{{|d_w(t$)~lL@
zBmT(6Ja6f!9A80UVO#?+2F=5l-Su-8nv|UN{nvVr-a-%e=r4>qc#@xJ&DIz~n#`E&
zOh&HkF_q=59KPOY)+WH&XvtUeg9G<j`b+<rk6AzJN6S>%%=coUQb8D1PUR!rO4|88
zXfTDKC`IA`eJ_jYg9?q;*?1#4WS2(WAZD<y{iLru-&xVjE4mR_L!ft%O_+V<_4c4(
zL<#?HWcP4F@PFUH6R@#1d+$qwtb&2~_YeHV{R7W?O<P$w+;&ecr(WsV3i3_#C{q@Z
zJ5A!La6PO+gaa~aHd&jST?_2U>yzuFj@ud(Rx*Ak?um{R5pi5H&!gY>*RAwG*BS=3
zc{PZMl_v}>)CPeTmSD;**|A@xTC+bEwq$Zxd#^kb>%V-@yas0Gm!H+~Jj3f-uh1ju
z(ksyA80Ww#YrER_k5%>-NZ#V_)u?47_bUdd|1Bl4zJ#+4e6~+A;WpvR;tq&bizK|U
zb3!ki5Y(xoW1`Qi?w%(?CjwIH{leJ7w4OOqnc-}1C`JNv<Nim-f7kNjY~ZHeYkYb4
z#UdQD?}6d<96KO#$KwssqTITFl++|$wpu)lv@iJM++yB+?m$IFqmj9#zi}4Vfye-j
z!JNixPA>7#WLaxo4e<vk1$HncL`hL`?pSU8&2F2#uD&2FE)~`aB~+2va6heRaj2Y7
zrvgiWHbDP{SowR|7^MeoEoB#%Ra{l60Jmg@t4Yc7zM~x2L>8sq<KURTKE3KD?51z{
zG~T|ZrXo_altuoR-7=P`L!<NYEy`D;h%ckMq4DwaF`aTK0}+Gu=L!BXMBG^6lf?HU
zcXPtD9?vpEa_KO0!zXz*a%%SsW<vk|DN|}XBLzCGnHPeOt0uC&+5!KIoo-j@t=|5A
zp=K*&?S0b3_Ks-6LoYS7ouq6b^zDh2aXNS{BEtPWtK;@yzlJ}j>2{k0CB6G=Un~^<
z1h-)>J3oC9<Y$rYxp`lc%V)l&3e3M}5dSf)sHtwRe>HuJO8B@W<E_Ow<YqE^i!xuQ
z9Jx!Vp#2tViK*4!H~4IIA^2>XdG|QuO%eXhQk!%kE&%?;)|`W*fP*7(-Z{QklM<#S
z>hM7~f3h+|V3&voXiUrX(h4xq<`0*tcBXK$Q#@xWMCP+(bz3Q!o0<EFuB2o*{)U-&
zy)^hSo*rPVZ?6ASwnT8cUUS?r+dDm=e@(R`X)8!C?%7Kcm0R@6k5H;2WDnu{s7gmj
ze$0EPUob7~(SMKeO36=!VT^`7B}F;4fqcgUf5bkikiBQ~2ADC>?iQ4DDfvub)fQ9D
zORgqUc)i%imxqsf*K=PRH)6+O<-}(~JJ4K%bqo=`7M#a!m&v7nx;lr-b1GqUJ5^h8
z3t1U1$S;cEOk9~-%=J-Ygnq_XMYLr>Gi~^+#Y9NN2H+#c3sG~z=AY@RlDbzl^S}Gd
zO(0^!YSnlEXk~vojK5Gz)3V48ephO;^cjC<SH&m7pkR_zSkLv+z7o~?NWG`<!29T!
ztUk}?wyCGP+K8sMS6sUk%H?B<Tgx@wBBpKUTmQD1G38CYwd=Cp&zzWQQ-Ak8FKEAB
zB9ttT)(u)qCCZool&xrZ^-Q28&TT7*Bba$A>b=(`u)i#b>^|C*u3l*W>_!*VYp-@!
zIGZcm213dokO#U+pP$fi;Lc>9u${MW68Z=w^}I~@L_!tlZsB#-@->^3Ts1}Cxy3~h
zEf53YOc_fjrq+rV^t(ZCU{Jpz=<1k<8{IhDSh~78J8ORQuyk%d_~JOHTCc>OVP9Bd
z==vOp7^V}&WoyIJ`0m<2dGTQ)S$9k#H1q-796kEQ9i;Zr6kTxZU6vOWT{4P<-aq{F
z`eCf{M>Gtsj%u{ye5pOo2XUxSP>(ZfZg2Sox@1Z?l!9aiH=C3~wm^ov(wob*wLrjA
zd9r1vvs;C0&6Wh-$7Z6_GP8Wy4gO?yfX0AY#1i)Ib7r%{sB)u>$?BX7KGz&A+wYx;
z6!knd-tmMvnd%t*s<G1%%>j?}iSD@O-`ORJH1#hLOe3P=vb(88k}G%0X+x>0!&Kv?
ztE<~Du)@I6Dswwt8-X^kvm0_A+y&a?^VWQ4^Y4KvF-d0imtE4b`f1r-cHHB@?GK$#
zD#GKp!0FnJ{~0JYXUs-*dX(YOHA%IxE3U3YNVoy&ZFG!;T?tV-f++bbm8cA#hyzhU
z@FYDE6GS}u*B09sB5=Q*n~rmQrqO7iH#+v0Qk^kgThv#v?!Us2SYH$8O6-}7GIuOI
z{^i{xoNeOLuujDqOg|j~Gy?Y?D^JqIjjJ9skN>lojDY#BTPyEvA^61T>6SYaaRFlL
zBo(M98v3Ph1M1{9y2P(pMX8m<4)Ejvl~IAW{3xg@M__M_9>4m@^XplqHP<RzFc)q9
zL*{XJXV=N{&Sb}xm+|d@D#GQU{F70x)Ag@U%@a$we0l~9`aDI%p94bvQ4mWnnJpV3
z6y=8xXxTIgcLCiyE?j@F>VT7;Szc7%vZq;7N}Y}0@^D}N+{Nid1>c}JKUx|7(8)!U
zoUiu>X_4Tok2eiVTSOOKsMVc%ccdD{T?>WXFbd*kY{%6guXO25RL~?_-#2keRP6qn
zNgAuRAVN>ZZu_a10`P||<B_$f&w#0O)^jk&aM=jVNWnHOgtK52%}vdb^Qf~28>x-s
zpAScNqh$PQLEnn?n*RLN0SK1gP`b`&D!BFnni(nKdH*crbp+WzO-l+^f}G*GC~#A$
z@Mj`B*Lx6i>2~c8NW6nyV8_AC2zMbr=1qiRP;(wGDgZFozs(axDw$y#|D&t_^i)I1
zRr5(@I9XaY&K4|Ej}`tbEG#H0D&*>w_!=u89q%mn(e1uF8Nn=GMLF(V|Bi@`x|G3c
zZ@vv<<qkNgyvtRHvYUcX$NVvVGdeS{VqUy7Q3vVjOg`C)G1fJR9lw<qH!n%k#+5Fx
zDK~_7Gy~RPx03BuonnO|rTj%1NVwiY|6J8P<Si(cu6Ey53^_eq+Nu4);azE_6m)!s
ztD3c%%HjATlERsS6bj_Xv;S6KRWiCUq3DW-H70bj5t(n<NSn6@zQM13DTyMqResbe
zb{$K_7juX292|J$t_93#x!UmA*bRSKY}S1}^Z?M3qvDz6p^VS`5%8a-HLR;MHQQ=<
z3@UqCzS-Yj-d(d9YiAwTl>k1A1<(6ekKE=ObHQXd@J`7bhDt!f=(&DNJlCN{6FhGu
zhVtXaE@HZ6&KK3FyWU7tG&Gu$icoE-?Pa<uR^k;V1%ybSsP{otzjqF71#3NnXAQYO
z0N(nFlJ8<GUUPk)lljg<<o~d>a$|qvbm@3$Nd_Kau)4FfBq7Kxmxx)ER~25t21M0#
zOxgc#a}+=@p|e;gU|4A8QBUX^#Gu}QpWm6^FGT#>e)4(QI{l?cx@$g~*T;DzSNi8P
zL-;y8l}*w_b6h}!5s^>LRbx-*B<r3JWYM##428P<K;;(|x$@d$1L&s$4kF#MClI5p
z>l`5`CGFk>GkfxuF&Izypy#W}ZDHW5UjYkwq0RNB-lZf8H#gg)wxx4;deJs6=Uy&K
z%qh#m?Z@*rya=u(WVZgd{lAqte-XHyU6je3+wLx<Q}oxHQ+%Ag16^d2(Q4N~()Gnj
zw@k?S((&2x{$ZPr_h63YE&DMIDWaF6{Iv8pDMqqY*skSzU=iMf5?@AB$|-II>>@Es
z*z$fBQ^bD;Dj!%{4D^}w4@Cu9_Wz!PSZrKVLGTC-fg9@(9m<JTx;njT0zDw364kR$
z2HAy#Y^Xn?s#e&MjKX~oUp0U~nE8ZY79TR(_`Kj=2$(2}_>GW=K{<8mJ`3>sb@x@j
zrz9+mn;TWs*Eee`+q2$h1?k2utAAe>@e^&lE5r$k2tbFY(%fSO!<W|T-(az<{<yrP
z@3b>fHcp(O+z78;#Ok5K<HzhAJD`vE-p;O@ySow=pW0ZN4)dCs09ukpBr_L=q+&Hf
z<Kn)CrTzX5gAB0<aY~4YhTUet1=+~`?bnzcJPzj3XRJXaMBn_W6Z7mlnvtAlFbZlD
zzhd@MKg(+-v9DyZTDsW5ekG+e5NA4SO2R_#KNKC(#6cHz=?2QD-(=#BmIBUwuKkZT
zcK-c6_?vx0T8{h1I`B9Os&%ZR4v8Q8Q`?f7#1R5sYw-2B5rEHGxVTihm~+26?hlzV
zGwqnSm|Jm816iWRm6ihVANR#?^qKADA$e`1)^4(~%94LUXy+^~8p}^h7-0W%7I2ET
z<RNa@XX{~u+y@hIiB@$Y1q>Ml#NMm<K;_y)8VDHRv_}T`zA7P&h41Obesq_R9h4Ql
zMN($?LhK1YlL1i$iljl?C%(&@sv=+eSLejZ-0;(6Eed{|l)p6hn75O-eAA=;Mpp!x
z*9N94E)4NvIABXJoPx4t&Gy-By58u)>1hzbUuQgaCL07IcgwCm`W|jq!N<OI_4gZH
zRRw-uCW08qspEWDD~f?4xQY;!j1N#_2OPH&3+h4;n|#lLb2DafvSl&ZvDpTyoFbz8
zT&>V#v$ApODNpBiuFP0NGX5S0ajwi~mBobuq=*W)2g?80o^nv<F?=uNi-30S?O*XH
zB}3OF>BA^aWn^&bTmC_V{P)+zuy-wWjkS1r_=EuX8W><VUmw|y90<+ys%l8hg$&Db
zgsH7tZ_z*K<3<m*xWzXfq)a-R-{37rOsWph=2B*tQSFRla@u$2+giaR)@T&n35Vii
z)T};}+Q-ghMrw79_%hGhE@}#dXNmB?dX^ySdxMNrynMR>>&w_96!KLgQ-cio!F*>d
zClJ69b?MRE8iGHo2Jm!7)$?)-m1QfY6ZNQiP3%JO$+po#kSk^FuCS1haCJ3iv4JPm
zpxHWt<!22y2$?in$eF%e-q_q#YP$ws9oTLC`y-J<?t>YA|8=FW|2AJRZ5-VYbrmlJ
ztu)AYtU~?&rpR_N(3l4$srtbnsU#4P6Y6)ei;fQ$=>72{><9f<&O(ZDo{V}oJmscb
z31$|9nI(XKakyptzJH*i+bqH>9nt7Y^nryqTZHJ^g*#APVJiyO-|A+fHMPFU8DB(7
zT3lQVb9SUvI^~d(ksrld{PSo?_5SsfnNm^YgIJ7##}>=YRQgU&W;MaHLm<dBqsNn(
z?dt<Q6I1#;$*TB)*H03HrhSvyjK&?co9Lcn_D5w7WAp#IO*RbEC7L}7@E&Z)im^Z5
zB6UWjpPl&=mM$9yA!6}g?Ljiw=H%`{<63>X>v&+uQWvQT`{;1QCpuYIm%l&R!ib(d
zX1Ih5{h4^;O<WTx52x24R|c3$$BP66YSa!^9=|IIne;x~%3_Cy@hw!q%ocNuqJYyV
z;VN3pnT+GJ$qd4xanY>+9?4tczFTS@pLfH`G94-@|3zb<*345Fa^d~?H#jGul(?*j
z#dk<vdsY0y4AV9|I<~lsv8*9B-PB3Q(AjhAX4>>17b69EL#*b}AmDD2Uo44ga}7ZJ
z2FtOwAXfe6Yn`4K<y^6>j4$@;IllD_F0m*eGCY}nElMR*ma1>HacN3c9#MZsdEdi6
zPhq?cPcX;s6y$@YdgHRBkLEt7EHP&dBF;c^uM4RbwKdWAxs*Lp5m#79k>2#j|7@Lg
zT-^j`s@~(-J(Cp!kgDC+5Y=ok@C+xIP2Om79c{HMEZel+daD#y^1}eC@!jc}VR^b1
z`3RW=bR@OE&P7`{!pegHl_yom*TcTQ6TBWZr|I%Upb;gZ8;5Vd+Z(sJ`J%T*haZ>X
zH0L$<ON8_Wm*8JH(%vbIkx{|*Idj3|>jUP6AiNE@<bhoRf%#uPi-hOfk~yXXGRiXl
zp;RtTxF{v({%b1jBf6^oG5!ho0a}Q-lyR7L0?o-Sw9{~L+r7(XOwiH!7mMKI{)INe
z-~K5M*aR32>3ivOvN5~cA$t$(1aP*-w1n;}kLG~OwevMNN89+A?6jWY?a?;Qm+>Ji
z$nRID3YWm@><MsjemUn!8T?=9kNMES`OuD?gUI2r5j454S-!i{SLsR{r7h%R@eG9a
z;IhbfDBFiJ@L(qBEvM`4R{8TdUP_1g#^2$?p7|&%Hd7C@n6Q2s%1%~M@p|tvi9;!*
z4f9o)x12(i!FCeiB{(fg&cS1(>b+w>`;Wf2b^(B8%K-|K4<Dk5MP-%$l2;mja2hyH
zVHi!%V^zZe=#32tjAqs68mg|Re6xBbw}rBsXy8Ahrp|m!pO-8v<P~x>pV<H8gU~~~
zJ~TGgTfqfUrG~g%`>5!aF+z!bpQhy&MrJE&j+cx!Uo!HR+TLI-KJnd&D5=Alu&nOr
z1`x&3e*bB0T*v9fmruJ3UkTj8c)7gBfZxTp86a>8-tbSmerg%%SqAUAJ70}NLq5&H
z!#Wum%@xD!a$#>`MqgB+QqlL4+ctYe;?4dXWmF5beu`6LZhfUcQUPP406T#5wZC2+
zgo`6xW2sWqSlz0tE~_RrwT7yX=w8T{1u)n4WDRWp^<bQV9UPHXY|&?qfwy0Z$8oI)
zP2}pPWm!`%Pu<VHZ%SkxT^wm=qSaql46CQi5))^Ag+EQR=B6lh0@WG#KghYp)zNxa
zChuf}*Sfxgs5^z;pj+?`2~&|4lZKl9%u~wln%Rg%AFUdLnF00;-=DlWg)ywk0E3T+
zC^;u5SzUpy?`Je{b=6%$SZ;N@(+pM>gp?o|y{dFk8xxb)J>{FVMPsGpf|N`U`0Iyc
zI`B<d>4;}n+1a)Zc7-TL?Ll9#L&#SxTDLVR8O7JTUS5T2KM9AnX{8~GTy#WymONzM
ztC&yXO8yhA1Vz91|4z}d{22z0lnu&TrndcTI4~HD$!lgzh#ZzX^81`fv{Q1d;Na{T
zq=xNzN<S!qkItjN*(H;>5)BpZWj7dUyDYe#1bdZ*%Q@^OK){1f{p0*xboND>PNDE4
zQn|@50>yTV>~EetMU&Ewe{NZKuM#phvj~iYsn<_r8;bZ;-3gxy0os~(=rRf}|GCd!
z{+(2$9+gmHFTr34?j6@xv+f<B9BmhZm7P54o=UP=xM64Dt0S72T*WuVmlZG1w`H?M
zX9PCf=Xi_u%YRYN`y^YcFd>&~2WxPposU27Sja74v2NU<<D9~?8+}O&*PHIw@+yj3
z=j(-GtR&<jU+P_f9CTTa2zB?A??X%_D>&%;k@e{g`oSkTRc6ykB-azNH;Q=&#~y)-
zbHRbl9d8sA*-J)W7Z0rG%}$_J!Yy?|KMG`NepNeRh|#^qQxVx5g454sk->gqlR>Y>
z4(=O$aOFE&wRR$IkiuRvnIv9%3lEmc%~#`N0ZP>L_1gVF8;ZuD)rx0Xpn;O*K_*)g
zUDRCK?8`EzV=q>To{Z635HO0!RzzXz1T*K6oi0#j5TGc?;l3**q_}m=ynJP?@NpTc
zKsF>A-oD!^<R>}-yOSP@c=Y+_TnXLvd37r6{HPiT@b>lg#jl+n`Uka5!Wdy`l%b?F
zuWza?{9FZyP)~}vB2%>mwGz_ZEW=4F_N<f0PRoCGbCD%kE2{t5ZJm^6u6Co1Dxa(2
z=<S~y$7)FXptU+V6GAM**V*q{9)EO!-7Ox{iGLi%RZ4DcAn15spV(SFY3#yrBudmz
z^VGmuhzU#l<^Df{BIun5A=EF`{>ty(=?sg>qf!QBS2Z-)_1E9V<>B+eUX{K!OR&GO
zf*(cs!fV3R)lB;rcO>P5Nb4WNfxlHW;M!#ZPaytU4MDj=%h;Vd@Y8~!eC8mIV^_p(
zI`$WicPcosS^Ka>OH@}(tDp*QhFRJz1UGj)t|~!spE?P^RM1Yg+-t>6QY5zA;_qQP
zE8w@!Z8f3(-vmU*Ao}OtfIOM<NV-rl%8ABOJyT3~G}{yli|_I`M{Ucf$0)<K=F_{Y
zlENl*g-_%q7=s_du?Kao8Ae8)>m%N5Pl<F|BB$svg{>7Pds<s10eoJ;(xzr+Ol6TT
z)FC4`lF-5NE8lu{REZwGBGaBlI0?!M0uH5iMNx5&Ljm7l&!)}S=8v?AnDmt^3O`ap
zwwlvk3;I$dw|!QO{Im=GBs?<ZyX^1D(#Q2SX8d(Z4m%rBSBn)8aP)6`JZA{3Lp?hl
zX<nDx9oNSgcQ9I0E~FLn71IF(25z1EH#e`Vu@OSn&(|LliABkL?^LbxAK)_v)Zt6}
z^_OQDj_b?k7S|qxW_dPqFrjD%KO|IL6+|I?e#MOMXH|k@V^w5z>7JJmY;*bs>BVKE
z!n0v$)NOC|L0KNDFML-1VMzto>D)8XQkt5JZB%BQ#V}BY<0F@U4-^qA&4c8%)(Tl(
z()+#t9-xX~Z1>;NKgdo{iw2VOx(oc0E*-V=a8~QbgqJ-<ZVmNkzhsvu2kIOCeji?s
zxI8%6Xjs1YX(R#dP=-eT7)T6#Fo`vgA?2=UsA@^)i5Gi5B<qg#2p5*|dO>2~qLLk|
zJG=gMb{9Eml+j<e%FW&Cvaz&r2!2_ukSx8yf(>_(17X8K9<@}Te9a!_>ceSNbg};$
zEAc_+R~UCS3a(SdtGc^LBk5|6UAGe0uSRsk>lN;sP}4m(wx{L1w~j_~!j^TSnd-}u
z;)fYmFSeyg??EBl$6W~+w<%-+aP<$n8P{Jv*Swh#OGf<gR*(Bus=)n#zd!P`<Zw9O
zxqn+3ePqPEfUxr~S*B+_Uf595mC1UyGDFeVy9!~Sh*;kX7!o<ks%?elYMD{$7^if8
z;wZF|{MG5oT!DrYbrIv;*85nvzqqW3Jc&|Q8hTP~2S$qT;=BZ)4*H~OzR`(!HrCwC
zkIx{TOd7G#>t_R^;`9&^xRW=sFjc=YEi~7ipC)q#=+M#o|3!aXsQ{+|u}`nHV|IaF
zU#CBdpN&$u(BQM1FPXHaBNtgkct^s1^OY>?bKL05pDM`}R~BdiqgJe{Ddt2dKc1vb
z>EZ(w6z?VVDf*#)b&rCKj$=%Bxsi0CT#2i>5pyn5&*keM4swnxP5OuzH?L3olopf#
zg+sF-ZeN{w5c36QSlcSz<K>kiO0l-K3pLOZ73ZebH98^I;CkCBL3?Qhi12VR(_pOF
z2!qLv(@xU{u~h75p<H!w%#+pgAy?%{i=cq!AYSqUH}~s>YuxqKByct<h=Se+;}7yp
zIsUq_fTK<8crK_YW^sA2Ca_JG8nlido*o9hIoShuBf!}uNGcsi!jFuAlauQW&>25o
zZWPrY_r-Tr>9{Pi!H7h(6ur9P-+@~7XrZH(^}io43hI_@MaSpd*dn_;GjZGd8!|)M
z3`&`A-^Pyl>MC~ykXrNXwUeDYXgSU{_JJWum!r$e-ZL<MV`s@14_bk&u@rj77)bbV
z+j7okT-8bFf%rC_Jl#t*JY-$ySNrRqW#9?KJTo;qJkr7PVDj+8ZN;~`H>-v1eANrr
zl-}n$SB|p@eBumsC>tM0`R6!S{|eXYo2QR(!rH}K9&&75`<-c!@^Ru!<2^f?jH=zz
zC9!xp;WEaoiNoc-Srz~zbc~m<Fk9H7&ES4Q$`uJE_P)PH!Yzjjif%~v#rETuPZofu
zI`bpg@ujt+*}mD?wt<Ivh;JEAm8iG7S!R<*k{|BxbrE2h8v|1EK%o6iTi0h<>J+;7
zNkk8hFh&+ZXJ<+v_}LtEipQ|Tza06_n4qpx_lfhSaYNl|FTkt@$H_!4em%C+TxTnL
zRyHl#3+gWK6N_t*4!I19a}nJN{ZGy(7%|~!rtI1T3Ruu%f+|d0>bSO~si|^T>diZ}
zm1A~fKLlPx<Q=VtI?Ta-QKc{}KWeTpnWP&$aSkzs{T$ETm5b7_(rww`FEV7h|9AMP
zbKan9w9x0*Ip8E6-YcqMK3RiGEsuU{bBEhvyVs1FM1PJ)<{zN|<av?n|8}u|pJ1c@
zf^mPHogq4u;zfiv+{Sw55L(=*$G{?E$0xzd%SbOuq`^Zy5&?zr(|A#mn<P8k)YYYW
zmqrR6zHJJJ6yL~Ge?^uSl8!S#Zm4zev|Rcy6Z_`p1BoB*u+d`THggCiC4C!UfY^CL
zt~3v~f}+yg+#K5BnPYx&y{&3)F(@Z~G<&`ml<V~Gd{7)qmn600z%nf=^4WX9vtu?5
zR+J89<9hZ9IJ>?&U5c80zN=JK@<Ge(&?+M1g|50EE#&A?T*E|=!-rRXwC_zpd5akb
zs(2#+NOX~P<tC;jesjPczSG}v*$4iv|Lj&PhO~lBEl77gGU^~~cRc3cP}^#=!9Bj!
z17(wcVe@w}INEITgv;NKm_y}V>|}1o`Ek_sm3xPLesR)CQu`)wbY^dHg#|Y{Z!oh7
z3Em0UR^MHy86=^hXhZF1GY4;Ugxt-MW<cshvf3YZTEp^_`LQ1i5W{3gW0=tPS0lKn
zL)~8EU6GL%G#YM)ZBgLB**-(9_xGV>|0AWD58xwJgL2RAo;SHyb)5b++sbotCH+)j
zvx=aOVMD5QEEtsujC+mTemm9ZP7ZU;)CkS6(4Ae1p75#1&g}-Tw}wqI*P_2E0zM>i
z;%XIWr?s{lVthi9m7uaidto~UAto#!@xC(Ood#eE$&~F|9=5PB<JO%`NB^jI0aFUW
zabJ_~<HsCPpGt-?!wjFP5?RKJB9^k8C{%cE^T$dJmF#w0HWqJpTyb2l{uA-*rk&Py
zZl1Py40*er>s?A*jG`cx;|>`C^rh6piC9uH>-A%ykocE3Q+3%8)LRO~Xa_PLOXeU>
z&Pqyn*)A%bTSFtuOerxhiS|dGPzlO$C1}rXyU0k6kdRmE<Shh+T2<QSWN2k(tb|2E
zvu+wV|1{BRbyH^ydeAVItF33FRD+Z>B2vexDy?uknoI`VZB5k7)D8w@(oLHTs@><q
z-K3lMt<Lu@Wa6;Ty|X_QIl(K@OtrZW3uWy`-f*i~<JwX{Pu!V7MxP|$_5+<L>K}#R
z{qI>0F|;{!ZoFy%nXKHCuoxW_Uf}roMkvl8f=%{60at^k)`Q9aRlyqrxOcmU-{5Z6
zymqYhwtV45F79Ct$&3LMnj%A^Rex#ziS7N8M_Oa>69r)jlNAx16&nhNMzPDeFE-lt
z$LEK2>1kW5dkMkGL@bfHMwCkj<D1L)m#~fas>nEQ5n+a8-O<!WU4u4!@T7&A#S@N9
zPFHQA=@(z^?MRt^8mp%N(@1Wk$_K-b5^S`}QkiZxHYf%KUX21A?H!jMMvY|Jkpeh>
zWfDiLVnM%uE?F<(N8w?PTk@&{C`cV0*sKnud}yTNgm6B*)zXY<ZV4nzmp66n&F%Rs
zTyAz;5(=(P`(`;h3QBNK##qniqoUQ-7MfJs<7zk0BfnC2i8`f${o%&`)&9nQIoQ_x
z!<;z(;&;W&0ef`Urzaif2N}}ULZBg3hC4`9Mj*6tsn$t@QD3yu@>{$a^&m#TS5DVq
zLRR{)yRosR9xt8gWjxuiGQ;Bs(X{ViMBMEW<4P%YGl7lxPA5Ju*{Dx7RUTJ&`Y8}G
zXjS{fE_^MEmT-mL>@P3>N@Xf}9aMSuwNqN)`nRhZKs+K9t>c`w=5FJHbZ~jKl%bx$
z+@IS1!|h|2&X~a?93{1;ljhcAqA}CPlrI#4yX^9Op$I3mq(0R#=ecwHuPi_6gfU|7
z-!EDRd}Ow$_gFo%Ghjz&^+BgTMzB@|9gGzON4P}oBAu*17Q(=}fXLop1av%{{n@y0
zzTyS&s-tD<xlsnbt^ThR2`cwTYF@c4$ve6Va>E`S_P5BDEDGi8^UaP2_Br0vkins}
z^ftl1K9GUr<~(<DpCm>~UhqKXu<M4>VX2DN^nqRc_buyn>m<rD7u<!4TQ99O-?$bj
z+-FWo)cJM^dGtbKf;x#Ps|*j3#LG0NXu%91^ZhizcNRA2pghP``%177j`)nqs%^}5
z4WrDyU6bql3${P~v`lNrdCm(9PjogeZ?GJJ@n#4(>5zx2|Djb1?waJ($F>>BS4Aku
zsJg3ud`%1+4rSq<W@)CYu-5cM%VE!O)^jYKsl10_c@KFRSV%r{5q@a46!TUy${YiK
zdP-?0Kmjf_+X3s=>yO-9W{aT%4t5Q0{7r|i?c698l`9(35%31@L@;VSmY7~gd&YuA
zUPkZXJJP5;q_qew*o-CM@3FQf<2GcO1Y>(@q7fq?w_~=E5cQNBF8JNRNu|*9Pw;-6
zg-mA(DA%=|VfU+|D$Xuu9qiI+01UbC?VC|nTh?2JCq(J892_O??N^6am)$<OpQZmJ
zFKoycAI}ahHakCL^Qiv#Yas7rdANMKzT(OJo1wu^;BZFbDd$$&i=P-75E~4xUkN}$
z_fB^S287Pc0R;E!?Wo;$E^N2|`nhPN8w*mA)b3tn$X1jmTa*<`Wcx0Ffxbg(8XkaQ
zMVEWsO%}l?SKUSn*QYXq4Th>yyC)rIv%*T(7g!lu?mG(NpR%}rWqtDPjW<#kEr)p7
zHSLg>+0Lc+_5DX))ZJsn*5nFTSDS$uH#Wj5f33{!v4@#d-$vjfDImY7NbR!HL7|uH
zv`!~|=~#-9ou%q9B^uCNW24Pe+9JfojoeU|I?vjIK_H9ohTxDf4Qa9aixR>n0xe^=
z8^$wlNBcd2H@+N?yDrD_QW!94I1L7vNgbv}{3Jn3owsLJZZ;OVPJxwgT2eA9TFZun
z*o#Uza9VZNB?r#63U-mLLY89G?6=Bh>fFeiUCwrvUOMwt0)ss8Y1j^}EXu&ykRRP+
zU?SnVmtT64h*+Cx9RN-*cl$o!!#V${H9;y(_&&`22rx+y_}Jf48ZEs)BEPn1+E3Xi
ze$YE8mBjqb3;#ZICMX+i3_o36XbZla=JMflYC!P+a<`xZYjv^#im~Y>xoF?u4KKYf
z2JqK>#MUrC{P~v6dxW}kWp@AXkDmo`m17mMYGvRoSv9OegE1j9^k-)-9{h?~O@Yg*
z-#dLyr4W>q&kY&e*QiA;Y(S+3C8QJ`(|NHW`cwZlDXn#8Y`&B(v6c&|bweztYQH`y
zk2r%Q|0=bLgu&m}UP_gohwPud_DM&qIA1hq4vPY0s>BZk$o^cWbN~t)L-ivhkbtBr
zjjs>sFQ1YV-Ix1=v|;P+`Glm^*!x>pZOm%M+)Bqgq%XuB{lh^W)amglv%E)-rNY)*
z?PToMq}t(7ha~|`%F;p;{-i#fE50z6RI~l=vD!k^qE`cM`LlgnduwxZ!17}tP-cF9
zp%UQ#Kt-po>-eVmdXe4KY-gdZHQ;309&2x|9>WTqXlwbjNs!r!lo0L<xLPak0RO89
zH@7>5Jo)Kp1z~Uf{>#<!U2yEQbk?OMiVNBcB}YRtw<BgJb?~RZEj8XLM%E@RPtjX3
z{e-8{vyuF0%w>zV$&{v7;~nbsoHFYO1PEu$9Ux_vu<Z_9_wBZWN=c8Wq@sRYjpGdB
z+_yfQg#4i~n#{Nyj#_N}A$m*o=cZRs^ChTW0ZuRr!31za9Ka6;b$=Oz^8PqK7EuQ7
zI^@=A1+|28;)`(vN~}y1(KHIuAnxtBYy9dYQmruS(=_wYWI5NPsWayH%?BnoB<|PC
zyik7)7qo_thY~@kp{&Wy>4GTR{O3bpyDn)2BYE;gwUkRn<e_Z9Q?BMqp^&ScwY9ZH
zh00mg_=Il!hy0a-dW17m(DKNub<=nudm&Wg#UF1W_b(@^aVoVF5k>eHjk=$l2YFqM
zUwDZ7tg)MI^@gw{I||<Xl@7q1G8lZD(*(y0DQGx;{}y|XP^vE+J5C0}^L#5C#Pqkw
z{S;BR3e+kMH6|-_{WsVG$Vtl4n~jFab$X?&&Z?q3jfK&RozWux|EafV>G8Y9jo)FJ
zMkjq8qlO{$g<WUcUXxJH_X(A<iYI_koszm5hXa$3jq>e^pS&(EgaF@Qr~ya9&V&X>
z;Pv_3aL__94!Jy~L5I*KQ+bEWt&5@eleH(8V>@p9ugrID!qsdnGq^ybwd!u4Xj+?F
zc@g%Fy8uGdnIA#<9;x@eXsmM0fJy{q`o4%!u*XT#t8&L40$G6uzH~?Y=Hq>6^@5=c
z=6uv8%lsCZqM@GR00^NBv>OJ?49OOu3WaoR+CN(&++#J%VB^$3`QQJri<M;Ej1JtB
z@&EUDJ5p~+>^tqi`J%92p3Oi*KJ0P&x#urr&IM1~=<}KNk%%9I4c@HSJ9`W!FCY#J
zzRO@(-yRq73uXxq3bw3oUV(bEt*$rOKU-?uWWWF6q`k4G=nGK3r{Lk#WfNHhbFSRJ
z5&f7`*m<T=*Xd3na;3bpZq0U@r^Hp3>^Xe<d~N1oQQO?(ynFfi&z<M5{i7@qlifmz
zHbaANIAGb}!xHkRdszP4R3F$1*CvS~1oB(r%;HD-D8BVO@{hLW*pIZL4)>Qqe+Pze
z>Y4H<#){#6Ebumys4%;XGV`D%noQMUopMOR$CE8kzDc8IoIX*vzc-9*@vOetZ40?v
zBM)gQS;li*?e#5O?;!stk=Z{hGE_V4B2;a*(G6=+iTJTuZ$*W$L_yg<7(esQ)}ARI
z9hC;o@RJKcgzmoCsO9=qO7boY_^NDfJ}l!cF+~A^9GR>)O(fo<*Gv)rf+|}$RJHo)
zl!A~!tww2aV2ocPFR$jb{FCniS%;>LpIJ4>Q_A_3yUXrRpi!j`@2bT<h&SYV>4^yA
z6?FS0CSke)rGB9NbGMA_OI9q>jjgpjZcaqq?h&tRz>#PoX1_$CSIL$B71rce0QEig
zvq9j<#&p&LTM=6pdxtn*(%xf!YKHatEx%(xqL&_Px4=MnIBG1_%lr2SZ?0a%-J8Y(
z2ic8+?i#$1J8}-DJm39_p|U*xS;OFCZ%+nWs&+9|3?riw1$xdx+2uOg)};EZt?T8T
zjpK;lVi^HiCC}99qo+r-?Mq`0*$Y3$WTF8>5;gCYp!JGdAGzLYO;3DlclKd^dH*{!
zXfIccX~gahZ&1r}z2a>T`6HTFq`y<{ap30W#%Sv04}&?_`!3y-4m}kg8&_67d;z<q
zI_c}E?I6?a&w{ixa5kTapd#vO6*{W0u*Q8Z6aB2lBPm%g(2_dX3UmI|vv-lOVvwu?
zgvF4(&zc_2Ua>XluS@rR*l(!iWC(?=!r;2Oh2CitlcpJuoF}X^d%Q+WP4ov}b~{Sm
zjBMQJ&Fxq)%t?h66~g`Yww40A^WNo^2$Anz%Ea4!i|%>}&-$WJwd(hwPxtm}L4xl)
zs5PmSq+_<n)0Ld^q3<t(7u%DThv_iv02hk)Yisj9TRJbt@f?1oDVMrlv+`Mj=y4uC
zMEe;w7q9+$_}k3ZqdH`2QM!QZ*7rf8mrgUi^;2%tY-_TUZE3k$dLPH-vnO+z&TE+~
zpW~00LE~i<LGk<(U6g2|+S3&paAaRnvb=Ftzp$NY2}&jJ_Fhh%JpRIf>$M63PLAh#
zR(sq!h6+bv`pK!c-skI;C|z7_KIbTQ6qBf(P`d5t&To0Qk1{jx$<swBYptzgU~xND
z@IqyafEcvc72)jZqUICrLOxozeX*!IxqECPWD$IXEdU$xt9?gfqo<mRayT~7p{bs$
zElDb6S6r23ws5xC#pC{^cmrI0=N5wlscw#tYXXO2kiSADoEq}A+iamD0RL|d)OEYe
z8XiC~OQ<kDLAHie7c>l&s&LpgP*Qz$P-2_ax(+s5$oc=cH>CR{s{8Q@XI9jokm@?r
zkuY7<&lfZ1A2rky*tr<Q<;z+JoGWd!i+!`S^|1Wzh}J22JB{B6z(C*w&DRbY>Wqvx
zkh$Lo0Uf;KpWQys6aD<Nf*NS(-4uRTLfktl+N{XnCRY@rq=xphp{)CV-a^K<s^Jw}
z60bkAmI_ccJmMv$%@00oDKh;fF{JJQ>3TXEu8ihGG}c!+{zUmWk&k`rpW#qdc6m-E
z<-gK=_ua|-4eJuSAz(-!q}dver5nI`N_@OqICw-<^qjLxB=RPfy`O1}X9m6Ludzj+
z|KO7I8~gxusuY0f#IVVVSK5;&X+5+C*#I)6b~rY9J~RXp-bq2LvZzoXX{knR&27J;
zvna7TJ8Qq-xPGy;_!e)*%ns_B)_!+RIA%r2?)3VaG-#rL20HWdBG=hQCJOX~ui2XG
zvj%fBGjUJqJmgC9E0v$ZCHxtM<SnVDT1ttrubc!mRK6=uvC2n5U`|4ku4n$a1tq?7
zT4&deZ$8|6mg1tlODk52*4??6Y@7wJMrT9rk`k`3KiDw@xj?Z328O+hOFL5QoZQIu
zh|E6cl$;CzG762G3XjTaL9eq9p6PkRl_|fgP?xJR9HB#L@<4@SHRYfRO(2Yq><R6U
z6~4$te`(qy=fbGQ!>Sfds(7YT!x`RhTz?&-`*M$EAHRS){zjC!3T^wQReKZKereQN
zhc1ZKUbud~%OfIAypS+DmBSW})>Tnk@zexUE(?I`i@sDZ<0jnErqKSj*VV!`9w>wQ
z5s1{LoJhW9S{FziDw8Yo_-&}N!S^sMpWS()6uz^h+h)1rTdTlDIE(%~QSXe4xz?<d
z`ocD?0K6y<rz$-8?p%ZhJ}ru}@;Zdsa{2q8pPn8j^M{;Yl#j_J+Qe5(tThOdy?0@;
z2JVq>U);A7HPPCZPp?b`i%P`($3QUQeD$v!d2fKR7;7`r#u#hp8+CFK7&Vs{r#kaC
zdD~@3EPw2oVNUYhTg_#R5=j^xM%8%Gh+`p}ptm%dfm76_7Uk-+QtD-SFE1e>`gBwe
z`f{A_x3!+*kDcSyQDK<ju%Sr~ZgqAO*jxb4j*h&Hon3Wps7tzcl9aAbN1rPx<|dB1
zY4X#Hk%)_N`EAHIhaKt7!v5k{w?{$jSIA|MT*9aX2(aG%TPVBd+AcJUF4m6^2NOn~
z_y}lV-)DCHZ2K|?39T(4fWZPYPW-ctLJf=}8X~Ii@&#;n1ngoVTp--GdlVS$mMNZ@
zn8`L0)V2OFgO3U+$uNF>55{?zYxAq#b+&nk`^gXT;)$>6!D*U%ZtP*mC;YeOa4mn=
zI@Q&otlm2mK0MzMPUYAE>-gQz+RUSTm0Ap6@<p!(3n4dDS2}A?o`9l839&EPk;-a%
zQQbaY|7ht~uF+}?46xzm40}bs6EPXI!57x#gd1Y-W~xk<g-aB9-a$qwBn<pPl+W_H
zdOId7++fqUOzd9#&W2adShz>b4Fa#SXafKin;)RbsBElh5*N)vG2%HlA(ANDd+H%3
zS<T&$>7-``WKOtZuuTXm7w1aJp48D$$b^kGTi|=7V`cq>I=jORdq!nFAG&6seCiJV
zjy|8YJ{6bC?D5grIAii(ku}@OhB7T@YmEvCM{cu8n?dreljxi3oIAGxE2Q|Ypizk`
z1O~>MmzBmMT&)FQjx1xI`meGpvDvcp3H+#Btxq`!$IX7z_D4sX+zW<zOk~%Bzqc0f
z=Lt8yKFkg!{2xc>{>bF}|M5AGG;^qygdw#`EaZHMk>rqLnaDXK=fiT`M2sBYi8)k~
zL#ENeHq5CU3mqJ$FsCuGOmZrU@BR7y2m4|7eO>qK^?W`a30^E^(wvDfws+~}w1K`1
zZfB`_ViSF<#7kZ)vR_BR+*dW*-`oO;vLidMxOs}=|25j%c|~g-lEe*5DWrWqP`(ZJ
z1MZeu?LpSHcU>}nK#QqEFQ}<wmBd9#X?-mW!m4|C@cp{lf<B}(7%-IDiyfNypo|9P
zLrMAlo6@GYoyKAgy&+|kLp)*T>Lp_bZr^bBEd}2NUewm<u^-c#O5V~9D%P{7j(Pp;
z(}%<qD;uM)h;veHaZ}c&7cUmVylHmwmo8oCgd{Ed74^yR_8R|_f`hJxX-1l10ex}t
z`{v1(hhwVy{51m`u}zOc`Z++OiCl>Q9W>ScQB~c`VzI*7r?XhO4%zU{8E^g{+$Y&L
z%=xv<?!p&UZ|<+{{L1t{e);*hqjZU+$peuT`ai>kiot#=PNih9pESNb1jbJt3AX@a
zo61O|_5<X6^5~%B^ZT8v5*vQ_Da?f~owY<QLw%X-UgrAxBE>PnHYPNUv#?WgZfANU
z{3kF!4+spF4%at)4C3<wJ)xlk$OdOcCtp>E%x+)J!v3(PmDXH&;~e$XOab*1dMdSJ
zAXu3qK9q*57UQ}HNNR#Y#z*);+{>mv9~Cm3w36tT)cK7ahwU?pBxhNjYb%@Q>nktV
z32|N&`vwBF{Na3pwvbJxz<?{pn$-Bb2LAIM#)fsCTLn|Z)0``vF(=A%bO3y{Ot>=p
zusyIp_qQK?VcHREPc8ZrC!#jz<`FXWO1uZ#i94&Y<{K1WKr5VJbtY=ra7dI9G#XaU
zXIXn%g6ZI4bJ;AT-ukpI8}9AFck=@yaGtTe5*_H`O#n2n0WQvJ9PhOd)sV3Xg@ASR
z$FT+7`mRPNvh{;Cd(l^Ce#bM%qw2blx3ZC(DzgbR+97Z};KAuR_O^h!HXb>xa|neS
zX`LpFTq>Ntrir&mGVxH?)IjlyB6@Nq_(_nGi&0V$c27mf6bdyWFXT#s`*`3hS4_#>
zoz+x?ASjYgGc@9Yfadm)h*|MaC=hB$!_Z2icmHwLY1F#iQe(~T-|L@n@h;R@mq-M=
z!wSJ4bA*usOFud_8_kv_Kqnntabzo*j@?!iDv54zfAWT_rY1ZC{@)jSAGg3DC#Qrn
zwS*^RN1pQhUi2{`>7G{b2#KvNN*^D0rcRufEOt4NIW{pmFczf<1|+hISG?At!>APh
z{vE#${+P__(Ic2{A9aYqpYB)}pHvGAi&aq`UgIj5sSPlssZ{ZOmTQU=MO0c8e7$j^
zO0<izM7@yRgQ<WzSWN<KuCKkXAjYGO&Cg=%>3ONpqsy^+uC0yCwuH-8oob1Dzcv@n
zPczV6AT{$u?c=t+%(a(S5UJeTS;!Qawwh3AB(m`UC8>jX@Hlbp?^bwgOO-_cnGqZx
z=~P3$kf)KAR(%p%%)vFE<8T1{XY1z4b*x|qGa*XrQ(4K8n$u*87g{`%x4R>A_1kes
zN^bZ7a$cqi@g#e<T1+gY@YckDIF|mX=HvQby*jVA-e8FH2WLxm=~gcvsKm?d#HX+W
zPj_*x$)n<N5>Vss^S0_sW^GM0WmimG-;BpNC=UTaG?O%Va2;uBmkt#bQZz5(8J3um
zWTd*u_iq!SA+X$BfYE|da9+V$Au8D#KxJTfqx9UE?~1i@BmHpRva)^!G2PpWr9BeG
z?1;1dy4XKg^44;;AQdi#=(%ItTqiJSX~X|BbaD|Uoab%q<rD%Yi6lOG-P1u@X4u_N
z_-79B&tAqE|9gE^-O;hY+;^Sh+uFjEwY?OYuMFyfBRy-g2#RS^CLb>I29k%Q6)N+S
z5L78IwH&_yNXX(9%P3fM;!F^k)<5VNPB5PBUj&}eIpYw(V*5Hc3P&uyBUxr#`8+9=
z`0_*ZdwT<ZtqVscMhdEEKpQ;zQzsJJA*<!FjH|o&h?qwN#U1v%cG*)5g%mR8JNXX!
zE$QuhresR4gGD;8X8xQ~$K2=Q=M9z1zJc|%Ew{t#E*!Y+MaJB+{JX0<Hw`<H{nn1B
z=Ki0_N1sjK5lzM<$I$KZd+~Zt+<eAY(lq2LxnuI|m%$uCL5IhN`i9&9gPS3X6IoMl
zQde<Gc5kL4D1co6<ugNe)>^HI4!!n@Yj6oQSH|sk_I0sT@%zAI2eJy70Oca^-hs=f
z7db0Wp0o|nu0XW2-@i!Wta*+(TDtHK53`1~t)3X+R_-_02TEVMBsFXd^Rhs8$4}M)
zp@zbR3vm-w7IpTH6wD=^qKlo+ga9v>JPvdjb<xq-E8P6MLb*l0L!XW$3r(%^;_9Tr
z7omTsj8c2cUFcAh%XHnkue+%<OYL61STBq4>=`90Xu_T*@YK{BR-~&4KbYQX(G%g~
zC7O4%;ZQdR4S3B%0wt5DyyvC%7Ho~~85=O~xD|wdwH-zcbn)bkOb))Pm1n-iJ?wB*
z`NWyn+=?wx4Otw!+NsU%RC_tTva&KkTU?6#yT(fEtrnxjw7QL<CswEhLx1BatVFRR
zODCrO9ra(Wj_13E%1Oy&s24$#gmq>-jSiZMX7;jziLaB>m!0pExZAs!;%L@}o&pp+
zNFJq*T0=?SUJj3sywwT+fLO$R4OscCsnsoe^>!v~huZ13ddQ%$xu)nSq1g#zKUg8Q
z<&fd`^Vf=NW*)&yG@yB4EacM@Ifzkdwp{x6e_vY;=<ulZNO}Lj>LYt|*)g6XyQZ)-
z8SjqOgP=0KjBn@#{d!>);vy3{(2a@qO4cKng@B9wB@rjOU|d0{lcX@BpAleMk;Y0a
z{Bx4`6z=T$*2;tyztX(aGd{RIcrDV3J{$&1&41JMYBzO+<cJvqDlb_`ZuD@=SwNWA
zzi>}`HM(~g$OQVjMR;J<DU~92Ww~Pk)~!d+mM{dJE?NdIFIy;3-cy5VmEt8TW{=yC
z=1Oba1EPJw-?!sPdiai~)Qm$sc_u$V`VmKPPFkE|T&QX#e-)AU!Ii*-7fLhA3i%?&
zQNMD%WPd@0<JYKn;@lV5Ps2diw7IIX_LEgGXpth+1Pt@O<|Qs|=P?g7nd^^4nruo*
zXgk~S5ITS7Mn?ed+st4(a{o80cQK@%tDX6b_>d70I54Tsasm)tqs8`J*^*184!90I
zg%8ci1t2#^al66V?~JZV7_Y`H!L%8_)31Z9M}=Q$B+zf9m*z-9=?5ep%gR}E7H(4t
zeBJ1COy<FcXO9EOuFOfP$<(z*_rE*QuEE#H$6~yBThBbwZ8#1S|B|i-6{wQAGk|F?
zsbWew#Ot*OO)TSkB^SF;GC;^i2)DBpoe*R2BqYMgHu=P>++xBF9fH40A(f2QG7+;v
z0HF#bT=AQYL@*7uj$)mgpN%dlB&&?pT_Mk9gMwMPPh`bxZBIHB-KvQ?Q`x`(Apmh=
zbN%j~wTTx9g*qMH)-qWaVolZViUa?5x;ag1yZTx0cT~p>*9q$DI;RR8?be=w(VD@P
zoy0DBx-`*Q>!Jp~m?(U!`!QS>p3=w4%>iOMMr<js`5;$n(78&&^XD2EooX)opMg81
z3}hg8zA~k&s`+7%+XVUoa@Inpdi>TbNm*nbpRjAQI^ZR!#?MiZ(R7G17D*a(LYFKS
zAsU~ROxKM8PS#9CxETY@_Di|B_p`Io;4t`Ui2AN8NADZd^h1n{(5P}bdkXF31nSdS
zsajjuyo<dj#QR_J{fQPJj1HB>?k>F*OHFLcxLI2`bDH-EE>!n#e_EQ%p9?n&b^I0i
zr&r8Tr**^?UU;}!u7=uH{ur&pl5NMHr<atUNF^9Yok3LQMttN`h>E~iCe`pEN3G2)
zhZ&COVc$n%LF!uUGC^02>Y^+8$Xvf#qN$dk-b(MkhWqfIFjZ?B8(j<2O^ta#v8XBk
zT&jB{4bb%9Vp|8-CF~=+NOxsG$$ZYpL+YXCJ_-u<E5CSzHX#Y>+u41_?MCuEAfmXj
zwFzOss*f~k_B3aehiN5F`g21-y9=)9;qqf>hP<<^1i~LKLYRB(wyJf9-W)-R%%h+R
zLhRfd$3Uc^P3kB~xN|u8NIKg$$_w6;EBJA^aUKUlj3*iylx8ok`_f7tr>@1*-|Ex#
zkR?SHoa>fCa@B~GuQ@{EOCB$5!iCfY`LRyEe#ZO?0#cW-BoT2P!NaxWhxVLYN%ay|
zkV+cM(0W%}={Eipt0BVLcY;d3QBU)Ap__d+`=zE}92nj-P&q0!Czd1kDFh%s{6RC$
z;3BDNQ?ts(&;7t1sQJTS2N7}4zS);@a&oq5tR&u>A-Axm<?vpN4B^h1N==u-U!QD0
zO}l^MID5&C*bSHlt*<DTN0IICV-&}X3wW~-RZ?CVh)a(ZiinOU3Ucwo>k}*Su3BsC
zA9()M8L{flCwm*ip6!7_lHY>K1dT%ixC*F!X0o?2SFZ1EL#<-2qc!_aOVBmrZ%?)+
zsLd|5S;SPPhU|6-M*&QsydHf`VT{Zke;6@sH;Ghpv*1)MGnzWVmj9YyYHG#Ojq=*^
zJ$=AOr|t@13(5XbM*9OCnFc<G>`(!<?X`q9k}l}{Vh8jtd=y>i>%tghYU8|Qy=t4B
z9{4HTz1=D*utAp};wxi%gLPTXhwxq>EaZb~D|^6TpJfVFYB*O2brn<TgddABEDWlG
zBL(@x$5wBiRIijugMW&sr91@YI@YKHUjrGN;owXpVJSf-(TMYkpsD3M?wzsQ%cuO-
z@gRw`YwpiY=|cS=#P)J=)7S3LRWWAb$&iD;RdMf!6QMyYwMJ5jY|bvs>)UB!4Mx)L
zdxetv$z90R<<Vp6jrffF0vYK-q5TN!Vd5{a$524fp!hqgTYce;XU`{tY3SmNnzzm5
z{^p{icHifizBToWF)DKwh?IXsSG}4vpwXNO1V>`*l7p?k2Dstwp9Qk?rROinr_wmF
zhSO}{_nmkdei>oB3{U&w2?^|dFx}BfQ&U6U=hC+F_Gb|PHA-;#R`5(SO&$C35ye@+
z6l9|1ltO+V<4%%21cpvq#xcj8T0j|kM0STw!hwyrnqajQeJehp$74?2Q0@;AOk$gq
zLS~m$y3V=&zqsx8e?hVFbDk2_F~B?f+^*Z+@3p1bS<f@I$(w(h?C)E6n+S1wVll)B
z8zh89v=2|4SYfEZ*;5rF#z&A(`6Q2a0k+u85^!VA@W-wZeB3z+7qDTbL!k|s(pp-N
zy!Ufh%s}fmux%5tLNfypDUT)8SSzm^hN6ez-9KPBljm+jz}xxLr<*|b$(|#@2ba`-
z`>1x-Kh+}>@2^wUsnM9vBpM~&`TKhfvEXQCQ_WJ#mSJ@+%~n52gkXLz{J{Y|gZRX_
zxHyB5%4plBfdfG&y2_S^97z(q&)`RyPP;Z>#7zEh=i1NTvz^?fw8b$_|5%|@n~H0o
zAOz<#gMz{o0RHNq#r=);JylM4ZTzB-n$T!l-yqzF&)myc>rmqrv>myIMmc9Atq!Oz
z^xALJPQV{Qc8Th0-JU22&DT^Ek&OuE_Qx6IM2|g);yBhX(!Wugj*0Vg)!?C=ca?f)
z@1A;Mv&-_doev37-R;cwR{IHM5oW&QCIjF5ww-j?htyl6bHp@9L126(^)<OlWpVyC
z|Lx+yJaqT*fzrO*=`dMP-;bWU@_-73(-bGWi?CFEH28ED0ceVEy4(r$g_$cWU47}?
z5BOBR;ygs%2>*<(=Du=$RRFNfKaFvG+XjOludCV6q3<-ZC@SSUd)g?WA6dAI$ltMU
z5n!p*+240UO&bIbj(ir!50ou}wg1~RZrp8@D<B^RPTKx3t-ot&)$pOcPzyvrEgd71
z_BNW4ro*9nfg$HencG_Uk@o^vLh@yVdigtOD%|Cxsg*F8X?#Nq6Zq|8;kxRR#PmEJ
zsCcuYf?{Z(H9D}o(;vzHhueEQ+uy=i_dVR{Y2l}4Z4%*fKd3*NKE<Hmy=AM3LQ2tK
zh(eY9=(wMAU4QchCqQlno-4D4KIGdrMi}=72S!~RWfa#SB_wpq666lt>AkaqZTArs
z^{^hPXi6^{vDdEJ%S65l3RG=NNEobh8rYwx|MX;vN;+lGXQ8BkgS`uRbLC%8@>H!q
z@Qls*v&jg>q&UgeDF(b3o@LS5vb@sVHpO@t9IbK%v))o;k@0(D!@^?!?*?bH859O|
z7Z1KKN+ywb<f}@(GIc6T)fH+^i%$vZ`Av@ooW9FODqrKXzm<x8dZk9-I9O8qUaq}m
z*T|Z=g>%5})~_BMp_&yeJ#1b;G=)`YhA!1UV=aMgkX4V%)#kh4Bjm(Co1FEr#fgOs
zh-e=kVr-Ev%=a+r;<i-iKxgm=I#RttOiT>-d#$sx^NUQpODeqDGczW%Lv`9cIB>Up
zf9(^E)=zCcqXhb_t0+{OD`OHe;x-G(Sk_mIt{|oj591`OpPv3|KOM}8L6{jQf#|Qg
zkA_(Q*ZH*JFfo4xO>;=tHA`V+7=fjhnhVgk^%NrifHzJlPM~AxtqC!EQMD!hba&MW
zDk;!nX^kb8{X4uBRR%0AN{#@Z-Bh~r=a#r7uDI*e2^1Ebv3={1WcbY+*v&RXd39G>
z=j<A8Y4*<7)#zlB{N3Wa<%-2FfFM-RsmQgYa0PpM`FKZq0NGxIXe65**0;VDpfGb-
z(p%m<vNwucnL$`KNcyr`GqY-1u=`|{nbkxRgdi*W-ZG1bxw|f>$rnG4oI#8^j|G*u
zs(hTJtglig6LJv>oMynsGO@|#rp$9IZ{xUz*?+=apE6HwV&7FgQ%veqE}Kb$npzD9
z-^#k!>J&1La*Z$(y42F&kw+Bm=qM=y20yz14wMn}FLd0f6w;{{I|J`!68T<Tu8pcC
zM@Ky>a^ksmP3XsCRS?(MW|KsnWC!UzGL)q$a~32_Ec0R+nWW+xm-}HMT+VAEj>k?E
zuEJrUicB-Vy?Su_u8yXuko{snqwvg)j{f*Zfi>e&d9}d}7>g!LS4;Xpts<X*ah(?%
zWbw#qaWz@B<-l1P8I+R~vz5ZQ|FI5sp?&Nx#Nc$Z(csUv=;ilcf-S(SzZD_`4kgbK
zG82P#mN=aAs!l8Y{hR0;Y9x9suF;~sB~da{Hj(q)e8lY0N}R}DJ?Y{rZ=%S-FYCff
zN^bYB#6GGe7giSE4CYW7@HA#xS{Sjw_s_3|+UGhFKPQKG2`_7Ow8h!^_M*rhSTcdi
z&{Y<iLTniu+&n0XkE50}QLEUiTUX4|q#l*=H9iXv;LXT9M}A-FL=JMrL2PZs##1}b
zi6N@=Nw?W(SH0tMJ)UD3>GTpcZ;>0;z^8I+OUoz)1xeDD=Xpm|a8Q~%+SLFN7u(f8
z7}FP4xZW!!CTFDoqbiIXrIK%NVk0#~D;!&-0l4Dz>BK(@3fsPqr`;-!`*ssD3E~PS
z*kNQ8J^<V)6(Iu=115`d48M>oo3SzUx<H4dx>TVbj@oRD)&4J4qnyrAgC}EA%#|uY
zKcv%BI;1A5$BxIoo{T!*`aTj_GWKWlFEC#^Q2yMqiRq}>qx}P$mX=0l{Q5OiR_qic
zo4?38|0&`@Rn>7$l7B)I??3PlSkx+TbxX9(jaGLEUHwSU!2v(r;8FB2l$U-0xtT0W
z7&sd)(=)rWLK$JqZ`)r}FwXhz8gN40%M!r{l9UMlC~PGxj_+}+a|)z6Zbd5*r)MLv
z{je$Z6HLcSry8VM6gwqD;=JPaVp-7zOmOPg2gj-Q1@_KXd4E0%#cBC5#dd;RX+c2t
zU>hLM9m~hrGvSqek!i~9IOWhEN;gC0uU*~7qYyc}2@15=l%RVt4-I46Bu8>%BSnpk
zW+r(i+v4LMqAP<Dl`M@4qL-*?wkda#<vYQkXkgWc_f832QeM7ga2LDyhBz4v3_*Si
z#iFUM7|qJ~$%83wEhKJg3fx@i45i65vkUKvz?G|IH(ihnWFC@FWlPzci&%Ri>`9mh
z3sk(0qCn4UP3-J*UMeHnwN??12!m3%(HDf&E-9FsD4V3l5i)9@18n#>F*+MEtFh$p
zUiRKc8O2Me$Q?)r$LIRHsk!jRICJAJ=9erx!-TlF)==>{7oR8wh`mfIo8$Hk!1r)f
zb8K2xaQO1$xaPrr?My_avq+=ldT_hY8{o}}OQaV|ffUUh#f9t$NLeuegt-5?Sj9EU
zc<%Ep97uV*m_ZqV4Qbn$W+dgMo0Qqpd>@6-Tzi#j0=tZ@z&t$6QtgUTp0HZV$awp6
zm;w*+`Nab=;0H)?J_$x;8OHbyu1#e<%COy?m&L;Dr^x}=^Ioy6ndrEZ!tTr&#G>cW
z__A{)gSNTwD;z=VuI2BA3o2XXrIrGa4mHtdq}J9{wBN78eEdWB3-d2aekE#E<QI!5
zRA@NfTi@KO-{0Kd`4vqspfbj5?*$99A=|4aDiz=kFzc%|mNWye^u*z`_2civ$7?^b
z38U@qtVFqpymXGJ{(MgkOz<GcD#}6V?SKE|P=bLa*ito0!WW{G!_w{sW{S2CW3{4Q
zH#z0`wzfWQZ*{v_J5u`fIC>N_c@UV~;{I&4yx=T^Bb=;~;M}=j7lyBkQ-i5T?9vyz
z1gu+3+L!{=(?`JsNBj6)GxT`G*VWYF&fsxt<Vhc5p{t7PgXMft=m7pI>fB;~we3wR
znP)Fp+yR@Ibf5Y<NVRZ%YS0QGNQlB09jSwkrsXWwN6?~gn5L6Z9)agutc*SgBmttk
zsQlX6F;~*D^T_DJk6iu_{IWzj6q@JjUXsBhD3Lcb9>BoubahO7#(-wH3e7>j6B$KW
ztlyntM*I+345Bg;nFt=R3fMBTiduWT1Mw#io`XB!qAU%i=dL9o9#gk#Py7G03FZj%
zVjp1l%?eJB`QEoOO<YUs?DSNsfN7h_rfDC<NdH$vdL0)jWOD76&HKs0hm$;3Z<$f<
z!S#z{{VO<Ou!+UUCB5%fj{SaRI@PrPT2jq7jg_rkZ&_@6VvGxQToe%GkToVqfG#Pd
zpV8$F3DB-L8-)prjyuD2x>4{*UTJ?Pyq|IDsM|=zSp^YLQgV}`$;@Nj?5%(PD^(^S
zKjkiC8G?1g<&z(=st4`BGgS&Gz+}dg*K*kbhmlVW;1ND0au+U*cho$j<*8<JqubAH
zmLSL#QTERuB(2c~7I_i_&7MRkE2-Ne*O|{e!AvK`@h+z?lNH5h<h$T}>YyrEcG_@n
zj*+2}kt!!z*T8MD#6ta=`iiiioZ+5-0L7VB(2FUqsv1fYgP%WIL!aW48f5CE8pClZ
z{#-I;h_VHPBDRRCKYdkP5mLE|3gtD`xqZh6@pfHPQ^CH>;6>Ln$CBogfGr^e_KUIc
zx<#e`waS%=3yxpU`L=yzXiQcglr7rwHq2bR0}Hn_G8CO?zCB@$j&ksmwugoH)l9_O
zbtV4W0GtC&4IztvnhPj3-{L;6>fz?Hr(P+~K)y3AiWJUVRi8G}=i10su@tjd%brca
zSJnf-T)~9I(f++S?e4XgyT4~&<p;U289ua<W}(63;j&ej=|@@22C->hEj+=$&icN;
z8Nb%ac~2Pg#qH4%(;$kbiM}iZu!B7scXC3_?=?Zc#JY*Or88X2AKe`~gNt!E6w0}@
zp9TyC3?E1L0`wKW8rNDgM}iCLOR4DqMr|xgMd9&K!-T=qAZj1iE@K>E_7i$H^i(CN
ztl@&SgKW5b3kB^e9QL*Qt?1O$e~mG%T0c?IpW5C*RL#{jki)W1mdA_L)a^v`@~j*K
z_>9#bC@*s}1JL8t;Puz;IPXbh^{!q7Ho_@r2pu$(0`2wGmd!?<fXwz6s*-rO<G&>)
z&F*uCH)}s~?|aEZN_#N0Xa789s{Y0r-dvm*50HFds2L$|+`B!ZcD48Pg^(2vP3!7G
z{;jup1Yy35qWKEl#-<RNQ`Hx<@b*@w!_<ebg90T590dF+1GI@1TzJCY-^AYAu_63L
zjT2<h(*g!+5@CME`cYRp5+(0k5i{ZbX}yJ^Tqj7ed~&E5Jww1#jy50SNdgJ_1uHBo
zgA&#&mC5OZ^AUY-cRpV}G->^XyH5OwK!s{;LwOYdzYGI?=+HH@vix^p?WX2@5gVnZ
znf$>+#nib$-t&AoQZoH0p7{`Sbuh$<r|gu52&jYkCcaQP!UK|rtDg}^4d|p76|qax
z-%rZqf}gssI=ZFsi~=MDM;>n*<hJ&)058@0TV9Dfzrz503h}zNX~r&VEiK3B-1^_0
zm0A*i6}&o}8Jp|+8Dz`_leeY{2woNll*m&dr{_9eCvia==fx-8iW`)T)jYz5?5D*{
zjT@>6Pe<*k3$ZQ~XY=FgxW9l1WBgvJXlDzB6V%2$H=!G+^2Nb*Jlxef3uqgP7lrk~
z&u0oh$>v*goCtp03w)Ra2~0*iJWkCwf%0j(UfEq31F)4%uu0MdzNnv&%NLE6AyUN8
zEuZ*W3kM8&MR3`+ww-I~zuI`nPSn>+j}Dn-kL<>_&?a)dv>!ooa;ocZwinog<h#0L
zpa_(7PgrTycJf}^3%b$i&x4ea%8~9DmAOQDrK#0zRaI4r<94Fax(0w@conpAx^T!5
z<#+>i9J<71V^{m%uQqP!HGCsK*%*qjaF|c4W&wvDZjRVhOpmewOCV2NJ2BURthyeE
zaRPTH?r*Si_D$LyeiG;!Jb0*=)jzAR++!C)u_5FJ=LlJcjMp@#XDEosp$1@|O?f8F
z5V9>if-^EFt)Xrn$^%bxt`X4W&vIYz1M^mF>cc3P5DGb9@CHv?T3->#5{Ng6U;O?%
zNgpR_ioI1k9_RmlpJm>LYosR+@(j#AUJ8<85x0eOxnjeO{nPGi3-NBwQCi=fj!TqH
zy$_nf2fRGY(~A5ZY?Y924o-@*4Dwg$`i!j?*AYL-ZVaASxX|9XQF-E7lfN5nVqxd!
z9V?_p?jBn^1@??cFWE?4;#QMg1%%Aaqj2^!l;uBEXQUACPxLWy$p=y3b5}9k0IOhV
z+r+NZC{!&B`?I*IX7l=KzpHN;(i<wZ$;s*tGkYP-0X6gOb5#Ke=3?9=U~t&ULRNU3
za|$F_%#-ZTI9}B>Q{IScYe_n0yq%xkIV1{t_}nyHCKfG@yt+SKgqsl9pEw=shYJ`-
zok;NrX3N@(qB6mVLSM{8l>)v4t|o9v{#}@~WzGKYjsFd3JlRHnH-^_DcQ*TPv2(+}
zGybsF8iF{lCmqqEZ>{RbI1>|RpuIBn>*qyvs*x1ues+MjtO<DLx~pq}*+>s_>JZT+
z)6SJi4Vv&9?a6e!$1B2<yxNOP*TFw<6z98SqoY_tx(CG$2loM!%VWO|r@|KoPK4ec
zZmb;(SLtug=kC-{AT@Nm($8PaqOQQj9CS0qc;baKLAJoU%kzl$i;}Z}2?jUo`=i~i
zXi9mQzM;!@@6`X4-C1iZmr^$o&4kaa&E{i;SfH4CnK2S1f%#rAp%pi(h5v_NHcHu|
zGnXO<boMzwjSRH6u`yB+K9fHOP~}&*JJl-k3BPx{#Kif8kcg<gzng`$v0F6@cfR&^
z3~O_@cB5P-S6)ur(vK@JoB>cF2&P+X5?Hm|WhKS++QCX08~N@uc~?#MY7c0LE``c?
zD+`2eDr>4OXe{L8D&9NyFGXg^bd=?nmse#&b9SqZ_BM?E*&Iq}UuGTx>B*Jj3eacp
z;+{`v@^=GFe+^?Y2KYy)-2=x(^LRvcw~}Hn@g+%m^Cug6Oh)DbMu|tPkq5e|o@KD(
zVN=wY$06&QVhKe8k6wmdYAvQRs8K-_uN)du01q|6%IV=?VMl0@zdr8ML?a`F)YOF1
zDwpy-j4FEn0a*uA*%D&(af7Ktc6k462zwb)#<tFsVJNJl3&@01$VHiI_-Zc?4*{_V
zpB~$5DcFh5LnAH)XQ4Eq;`sZ)v_C5g-Nr+?@YIg{R35(k*XKBM*#^GHPCXQ%kgZ!B
z-58sHaGSrP+k%{A2s&Y(8R!rP{B_Y{{~$p>o}}m9p_iIA)}*&22Vd0pPPP_EcsahY
z56l!V;wj7V<@)6YE2w)bD2T!%&+CXE)R7fq0jFqo^>K-o8#f0BjRU_qz9}KQ_k9_)
zu3YUM%__pb)N!P6RipVMqU;2f%DbMjP-TIDtAklTUDb^6J>(D($nLZ1l7bza9C0cV
zcKGjOnOP=6<B-VY_JV>)9ip;-WeYIa)9mAPKoA;dl9a7;<qZC#6{q=vxOz_B(HNOj
zr~-(1EX-KY=dhhPf>Kl51W=coE?gUSb|sT~WK5KNQrj+jN(nq<Sh~=-NhX&vF)&<J
ztIiWuPS7{^Pe(G>euLx;6K@ohkYYQ~l-pj6byjSP*m+MM*>LGJ*(Mn;`OvQi(=YPK
z_kI?Lx!K;j_fM{UZ$<72n|x?3;2dZ>OAPBHyxFRc%;QsEkP<`6VLQW&j@$0ehTev?
z4c@oAYOJ_idtT_5dsq9(jd!i(+tA;*&fZHVzk!GP_1WVsl~vVg!%G`@FrqvEd}95!
z4CwP<ph`LLXs_@8xw(J8Kg-TRl>a55(>Y7f&Nj6~3I>pa`X^m)r@XA2pk4^tsvA#z
zjNAx`fJerj<ZC=2F4zO}`iP;)$v%$86rQppD0$T-MAY;<pIf&==;b;!xz``zG^fvf
zGTH~I8UJ!{JHxo&t*2wVh}gnlGhQ=XlxujgxWmjW{mx~lN$zV6{O*fPu!AK)*&!eY
z*Z=c6VRG$a#%k4xJD+M}Ls&9QI~mJ26O8j|ZXY`vcYpF&l4Xd~-}OJ?2_wQ*=Kg~e
z?E0KJ{tBjkaiXwZH4U-an-ZK9HX#1R+mfYSm4E!@1(mYz(Y6b}^sWmJI91%nE&Sbb
zCU+rohBY9jKm*H#*8Ji`3JPkqbm5ch7qVqdkV=$$e@nqogTw3haP6k;b^v40b5@{n
z-7$Uet<&mhClhtf<*d-O>vtTYQtBR~-V!I>ft~*Uuk=xZ#&c6&8})cM0_Y4&rhKo}
z+oSTmY7jWTQG2}Z%HEyWc-&XZ85x~l&FB2wEnV58c&^ap{d7MVh!8s`tL80QcW|KQ
zets@*UNIGoc1e7|J4^HJDR7n%2aBG)CuM4qJY`)JAaqmh=_$CUw1nlGRf|+v&O-E~
zMFsmS!g+%?4puPG$B2TdFk%tof0KdnRfi828uLGFm^Nlzs23gzI5IF_CA#8d=|#E(
zO-@fB2$ICBD4F)P_9;a}5<|&2C25{5tHDNg)i@g6bQXFvrs8)DTWs<HDImglq33m@
zwI4ryg_P*fUFu+o?1u6tnP#LMCJ+b=`<o#PtH8-Z!^Q7xuG4&nt3HqD{d*zC^s4Ba
z4;~nKGvpw;i@^bU<JfrZQdyhb;a6UFm;o-N&2&|--l<5@kN=MC?bVu`e3$=j?PV>c
zrL|48^XiA_R`=0-VsF@3?Z_U#K;r>VLZ3K_^Sbn=yWziIIs4I*-K<1W87Inp%6hJS
zCMjql-fgUZ^L&<`S~Q@npKvxKtM?6ZvHT#=WNSR?JI;@t;Q>^iN_)bHzxK3$$W|?>
zRlx6-jF<q)4tu~zQ5YjDCS>f0p+%{<Ucqb^w*UQv`@_Kj`-;VB;Uqks#U0VxkYVLU
zcmxlj$qAo|ycEh|clAztgZqBKJ916Kq&UJwi6O6`KivhzwSg&Uv7XQeQ)Ls5_i4Xu
zo?M)+Z9mX_d|OAmA}Uk`&#YNB85|gt5fwKD6F&3YLY(k|tSyF(C;ZzXr%$taXTYg8
z+qQz+K?B(0yL!!WeAh|Jb@cop!yNk3(ptr7zi~$yXmPHt!sfTOqm55_PjxuYuQbQU
zCa#Rn{ewY02}z%UrXbdP=JVLYlx(-bX<S%=-*!-zKRJl$?Ac9qQV-LDp&t%8c8T9`
z8l%L=HBA+akB}^Cf@-a0Uf6bn(%~Td16`sWt(HgP9@w0JisJguEaf})sJ$5P%JhKV
z^ruPd3Y|Ly@XyDOW^GXX&&LTsq}vZ0`cu2YGgHU&%$qCL{5n@<BOdpJ^;I@-a8c$a
z*Z3-6u+o8ssKFER_wd}V%p~LElVI@R`=n;vpd<NTBI(%orNc0nx4jrEWMPbHbhx1R
zRmBZxvR30*C~82;+Yy__SnqbEjF1#8zG9s_)vSiUqrJg8W^eRlykV&r8ehjIl|U-G
zTr2bJSqYJTR9E6ILZVwLu%-@|n$2DIFm`(<+E?0}_Zjf^t<SLsWK5;QW$bgMbyAe?
zv=;pDAK|b;j3|3K8SEgt_IvhMIM2o0eVd7rD4qY5Tw-6>kW@hN>4eFErRB%77(>zh
zebrgG*!!i&b)**qeT1I^__tO^#AlO>Z$%ls@`eoCv~;n#ppg-j85j5O+%kh8P*fK8
z%E{?a+3FZHB^lhPe)m8(QFLj2c;S-~;DqW0#5##hXHT5DAmH@#_@W8Y66#}uI#O;d
zMW{x$(%W}caXYL|mN%@|X=S0=ojRVo)kf0dG-SEu{|#KM9NUt8HZY9;UfBO<VPT>D
zA5bG*__I)Nw5PfFA1(r8T|_sIG#cdtf$pP}uwOqe0^7PoDmB1qO#WR}aYSNF%-+^S
zh}8tZZVl+KWH7vBzGz>Y2v?~jIdirCJLZAvph>o?!@tNr`Dk*{2R65~N32qlH7R|!
zwkJPO|LIfxr<=TM|FakdQAQc0B3%WPoqRR4w^MF!qx~=LC*4Rqr+w5s=Z<b`1QyGw
zW0FUL*CMnIbHn5mE{@*wu~=euGNJuC^M#AAeSywAdfq^>A;iCG(2Du?s7zBFTl^k^
zAo*NR;&~Cx*%H}_{_yATe{&d%UQdnicy%kDaymltp8UFo?Qo>Ftb--ICk;yrGbFO2
zzMU&9IgM-4Ilb>K*8B9NJTIQ-NM|79(_)~HHcFHioA!#*Sv<D))=5nPAXVO87CfHA
z*L3v7LbYmr-)%f@XKxLP5Hx=&<(v7cBy*&U)SHV*%ZfZN*w{qURQG}Bv7H<2YVes5
z3nYJgY2iBO*WZ@-y>PcTy5bUCkE8NXW4N(`G7cnf%0Yw=4u}hDU#)(-e9ntZKC+pj
z#J@X8>8vmTC)kcLEZ<vG<Uag9*nAJv2PO98v%?&nXq8mn8F=b2rde)STa<j;bf!Zp
zSq1_qCbsljx}=sF9R`(3+!xE1=HKxWtVG2Je6ZMFl?@eM8G~%k?3*-pdtKCU!#y|-
z(n)xSL}C5O=7%BTy71{tyrVUf6nIjb^o?2<aM(-1Z)H%md>q|$I27n>NQD9^OFEsk
zX#0c^BXg<1qY8Y7SgnywUfKz%r8mWX%h(KgX0C8{=-YXzQ<jY?B>uBJ&0(WtJ-&!h
zvIZ!#uWP;!;sY7fKhp11=*)saZ~)J!jou#Te~90#bZnj4a%26qfUsK^$oQgmCCv1@
zEa6c?Qt)`H)Uh{kRTm?}oBwmF8yUK}7sdzr9XHumet0VtmU0x91TO!XKh(MQ`wnh<
zkd6h)H@W%Fo2rYOF>b1^h~<#lu#m>B#M!A=0;>|2Ljf=5`tQGgQ$X#w{Z0Ta!9M2M
zcCOI3Daw;4Pwjw&4(QGi4ey}A7cc(AH|eYNAGqGy9*4#11_Unl$F!=zEmpOw^9Sbd
zG#~3pzR=iGM{^Ax$K&sU9z2j2>`1|X_oGT|hPQb?E)_oUQ>~&i%&@H_XM1=lF*}JO
zWe$MDh0g==UZ!Maj3p9RV9(IZSzn{CPXGJ)&uEfqWJ$1Sf7l|CH`Tg((K>LDKWnmd
zCtsC@ri=0<fkS(x_~cEXr9D><$OoKr(`&>scJ<M%)pt*(8+$3qBh?N-+4eGcVlRe!
zZ%a+Is$+ocG|I4UcAFZjRBg4^-2u34OI!`>b#2iw3yHfrlekAVSS;4owjSP6e1T)o
z;O4x~;H=}ue)9cr-Rj<qX>w7#Xl0k}5UXuNaSF;Hu)FwECP!Q;LnBO55n=|t5S6~P
zu(qd`qCt?qBO!b-w@-%9@JYWUASfW@!7(AVWSbSOG(TxC3E?pyR?UAGVOI6I_{pdK
zsBe~0A{R~Mad{o0sMN3VVsFOSl^W{&Wrlap)BZb_xBZ322g4WZ=;}t-*=pq@Bfy?8
zi8n-ysON<9=9Jcv!%H{emKI0}V-a}wXEfqO$T9vr_`zZQPY%+laFJ_49t_Ka$c&$#
z*HD|zfqZB1qM*$0)M26R>c(%-a*(+azx-YH!9H(M97{j`Vb?DX7wTZP-Jt|8GCecu
zc=@hcO=xYaJ>O~?Qu7DR45!6I5QbX2A|BfRs-hIYx-tjzh0}`&wWsAXQmRO`=Z7>P
zX`S#wlN5?(kTW$1d*XT4FuV#yKy*)KR%h0A<!vKCCbscsC=^4U2!D3v>Bdp}tjR-k
zBUJ~+1<0ie@z6$cAgaIC{bq1~zx|kmlKhq<&wV5W1G5($$UKEgO8-*AAr6_xs_ewO
z85uspCF-`H{o~w#UZub#1w0Q+r9#19^hmcY*SKo9{leL+yhm|2OKNc;4ZJ_@9L4%&
zCV9^k78U~gtJU`X*--t0|HRTdo9hx%H`vi#G*uMAhY{YSG6qB$F%##K!es2BK&pTf
zgdFbFKcKE|%kt-TYJW<j18+VZTJTq7Xnvg@gur9mA^s`~E86JdAau(d01;?>^e!HX
zFNI*&wX-<5aF<Vu@vXAG2urnLjI@xU&zqn##kHaF@!Nx9)BBxr`#*11{Q>Je|3@Qz
zmy>I!pIaO&RINLhcO8mUz}HBW9V4Q1PM$PYcI5Yz78**O8I6^rJJfBtcG5ZaCdJB6
zL4B4cUAG~wQ^lkdwde<05msLgPFhe%&TrUtf(`1pIS>?)($O`nVLg2I;!ZN9U<l2#
z4b@)Bw@&E1$@}B;U+w@dt`!gRhm6KY>D3oKvJp3#kIBHs{P^n=FeNI@jV78bgPoqr
zeI5&_r|Fj#7II>xJ2ftQWL`{xezy+(OasRq8(r0FH7!(--iqvmAGDgq#0=70eaa<k
z?jmqNDaysk!(2lFI&1N6v|%J5``_Aw2($105q1%QWQyfHsEzM6@gimVL2J?Qv*gC@
z2{W%uyu89$$3UkEYvBQu1+pim_@0``fd{9jg5e?OD>SlM+BQ%31)f5kDF5A8jMBdb
zP9h$3b@Cm5a7H&;<{K_3Kma1{0!qMnXWqm0A|7S0UsL&0FDishbJo0N;vPY+96B}x
zPw}Yg@L)Fif<&w-RC5$dTjD-nq)023K@^vnv_8m_+?D^Xpns>oTIC3zaY*1fw5uop
zx}0QeHlGDQrm57sI2D0q8c1rFgxQ9JijDRCyk9G;&4=8MSiB=KNL$Roq=3=HOx#ar
z$zF*^&o<Vse*3wEaZ~?~c<0O_XCdnSm8}4bRUa>?N~KZ*f|0iWv3R~p2{mJLs`F}#
zN+Bobi966N`_+8ADO&q5uh+?6IS4gHNiYvy{5rI`q*rVk*a-J(5a<hx#~(SWg`5AW
z)N(#<%k#$u{@H6)-8a7<4mxHJHUU0t(1GCDJGxuAv1^!Bar%DqwGE`KKwgzo5Pj;g
z>e!+^bxCcp)D<z-FEF`?(Ur|<<u|plCSTYEly-l^-3lu6{|i*1=&V66RrsHq0oCpP
zaH%aSLZ{7$AwI6_%PEVN`MbWS1U*UInOIo5ufj3fTeCf%>8i3VH|1cph+BxJFN6_s
z`@11V@jx_bS8l3(7lPQ_*sMqQ?Eg6+E(e@De^@yb7bX=RE;5i})rERDB(Q2_^~fCa
z^pdo|VG#13O5w*~exbNPhUvA(udRmRO8gg0SL&p$j~TXcrO)Q6eG*bT)x<0S(9G!L
zgU$YwLY$R?NnV<#s012ErcK-hSLRPNyVv!{x<YSTAVZv{2lV&XR5~Gp->|MIzT3tn
z%j35*G*Et-CwgNto1DF-UpMQV&wvW+-o3WX_A6sIEpHYWwcg+^MjBO{TEf+$R5f^c
z8owEG%7bASK%s#i@~7ryPRV9tq^s=5n_tj42PfCjJdGOjYP`5>TIv<$*@90aT#6US
z()NMw5&f>hJUY-fpjh03^})}{bTvskq2D~bB9s79Du^`RD}nzw086IBy)apKsy8~$
zrAIESZy8I>&8ihIc|NYfWOUB>SJs@7&n<bV(teIM5gSM%bi9*D)&$KxIyBJ@=Xxv$
zgH!u#{6mfzAaU6a`j6Py9PCZAi@%$*5;Gqu)~bX!A7tu1Hzr7x62BIKz^8!64LUkr
z;0yQjvR|)L7Z+CW8gIb#pGf(9?<~dt+(gh|n`-36e8%7EoF@QUa`A?x%BUmXn^2I@
zPX@aGgyPT&mAFg|GQgFMDx9+OlJ@Fd=Ec{DOY$}t;98FLE~Vh3a9;n_;`SngrL)b?
zBZIUOTG4j#B_qizEiS{U!44USKB%v(<g-8rr~b-B9YAsQXQB7zW5f0vj2kk|*A98-
z!aAW!-S}XxC0Id*cyxA8t)=}*FH*tbLTih=;mu#1@z;4KVzBeryKYaIIOtbXgZQy{
zo@Hn+4thPM{ofXV`qsm?GYg{J0-3b&bMHD69!$~=20w(yB^cn^ZnWS2gQgI_aX31c
ziPc5K!It>|u7uJ(egh}dh+Dt7-;ODvct^6B+28wYf&&6bWvP9p(vV(H!zxdSdpiqJ
zPsMe3(Ivf@J1H`C*`Fupw1X>0RNCDOT1K}|0T-IJZJh4ospIUpauCyC2Ds#l5#X8I
z)!P7DJ@IdU#9(dK5<R5w;Wb_b9;3p>YrJTlTZNL>d4KYoW41I4g<N<Qd3p=axbVh+
ztN-_2c#MlFeBr`-ffqMjIPb!HhHqi1uuuZci`H+m4f8un_gT^k82{VME}CPm&HiK2
zo*3>eM7J)K@{|@`zfL^*9`szVut8+4K1mt_gUB3jtkH~!Fk3Brj^;hX(^5Es<~@yD
zooTC((VZH4N~{m4-I=$U?=n;}Y&f-yAn5WUbLWS=zWTg97VH@PsH=bjT(#rf-B;Nx
z=zpOgFMdza$%Hp8*o)WucJ}aGs>%d)zIZfion7fEDpW!CZ(XpbF)}nw=R4+&?GI(3
zCoDDPm3`p+{-838iczNnAOT|uW!#jj`@f1~ifZ?t#U-;dtdd-aR+M&A0+jc_+Ci%<
zFHyf1#ng@$n8157B2T~l{C!8x+k?&VeNTBY5nC!0aH5RwRmH2*XRkB`=@eFpKukpO
zBM+LiwLLiL3L~aV#8dnx8V->&eVw^I2mq-o+IfvUU$0>mQ({tFJuh+G>>!S2FFr5`
zKC7k>`l@Po_Q410XqAM?>Y0t)gDK<1#Q|grBUWG56x_T|9UY=lpps!8hcz9xiGoTd
zAABdk!bh9?hGb-FIuV5v6pXrhm)1qzl^8OAhlKo{2huE?E(pAc!hId3<ueEC`10^n
zrX?{D9??>lIMH&!!A#<?9WLm*GJhK4cr{!Nl2i?|_c(F=TI0Eib#g*{^WTl7&6XF|
z)<p0{Fkgg-2I{f7&VZ@=Q{!W)`3eV<An9^kHY;pi<HpB{6JrrQgU@+Zw>_V<a9_v#
zY;|cDn%p3j3REMQMw9=zK%@?+=-+9<`D@-hxA@<cLfk!QDiClA!98{`Q7Svv{2BV?
z>E6a#NFnD;<MFDa50aO|-Ck%OE@C6Hp7Gswbw5`a<VI`xxA_T2XNty$dn@t!M1Ufd
z<U_xw9>a1^Ym|p&uOm|{FiNJPMFCUPdy8=Q5G}`&Uh(VF_)*<(kbKY{cOt~FWj|5Y
z7WA~=DfU9oHS%nID-P=a*k)C>Jt6*Y&$x5{uciJ%l_ToyrK--Z^|onnrxRH3a`Tz*
zow>v`VuMgeGo8DsUVIe#ivO@o@7f)G=lZirMV2bAKtsR_LoT{FMKzua=ETz7_XA}2
zHpKR=ptsYce|$VHf4Zrqt&JTdIU7@a{WmvqXKz=os&nbj0`9KV%}<=y4DxkO{g3H(
z%Py_j7u;l7eClT~y5n6>8R7I@@Baw=T{8BB2a5Hsz%J(T&5!k&h20TL?=zxZ*EvxM
zZId?M7D@j>;rwja&gt)jk*S2a#21bf=W!~7V?S3}R>pU<t31;E1BX-O20aRWM1&^s
zUQny8{EiMr;wr<WPlT#>ly`ZMs+<GuLIXq_6c<XUz)60-0_Ke>FBp@0z%O3VLQXlX
zpw7yDw-h=skr9kY%}r%2zrGP?+D=P7ShlHgR0`G1WDJe#d<Z(OxFcVh+G$~`c|(it
z?pk*!5A<HI>YI43snk2t*Bgh@^Wq+8LE%lwW3U<{oW;Nk`sDul*tOdmsNaCGIEMu_
zX@2NuAtEIZh#>n9pZ364d1NyNIL>iXZ_hcvC)bRpcJCiXRbjl?_LdIEQ+@Q{DFZ=F
zYWaAtto&PMx^_>6bznb-nhv{Sfrxq|D*iN@4(9q8Ir%W`kI?U$)}pUS7SNg$#d9@X
zsjj%*gY0!_ykhtAN<oOo8T^M>){R_3icHyv)ePbn#zqXAMrXn9rFPgs-$tCH4ZUXR
z&<cIy+?*ps-(?bcZ4raRk^7-o{efYP3Z?$dmKz`6{XoBu5>etgKQd7n-^3Uj3@*q>
zUJ=$vNWS&=p~Yrk|GVw0heHoaWcMzWgtP&<o(UjvDpt*U${>JhU$E4wV1ehDHQL$9
z1Q68tmUSWdMqU4KxSnJui)AV*bC1TM1b-7p2%;Q=KG{Cu$u(8N%kvhnmmIISU;A-<
zJppoe*VMvw6g}yf!s@7|cdjMMh#|8qBrJozvq^})RY0}T-<vbm?~5pkExmdUM~<zX
zF>l{zK56^I-Yq?*u8t+}Mhf<YM8kQ6Y}W7W{N(Ny7gjIsgg2EZLpQHqn+@h(iVRQF
zhJ?XBBT_FO+lqOwjnb^8oUsB4GY2Q{W)g0coc90N-BH&+w~Rmur(Yh3r!y4}@QOPG
zp{bj(tsC&9ChWe_l~XZvG7b1>i5P{>>C5F*vv$Sgrf|Rg3@5uI8_6p-_crTMqV0DT
zv7-lDf_C(tmG2UppEo!f)ZfdKpE;$Ujd*!DijQBTAzYsc^wL)4xQ2taPlg<PS|+Ig
zP}O}jxXGOnJ`qqf!gKHF7exmp5mbwV^LDV^6Fc?0HT8iBh9MonshYU4vBIFHC3N28
zm*N^tHF=X%lJhA2gXAcRBBr?w2hF-eSCX<@n?PSLnNt*W@5?((bA>MxG{FygVOQSw
zAA-GS#y$o*Dir^!Ye7@7y9XR(*ckmMg~)?gtk{z$?zORjZ9(MbTem5RM;Ar=7T23!
zG}WX+bD{oX_VQBV7jRL=LggAt9=FH;X2l8!;^%E>6g8FhX6XKdv9VXdiFa_ti$yx|
zv4+RO`Opz$;~pNtZ{hCi&S+1ilm6E9?BiH+eeDGW<#(Su51SB>B$vR!fR*mXBAzXk
ze(14W_<?&3CnxNxua=YY^c1;jsk9{n^=c{RMb|(|+r6`mB;4{b%M3#EV(pbXwa!1p
zdB;op-e)yn9^4MFbnU6Knhg2q94MmE^*gj!Zgp`uOop!#Zv}=lF;0vHQf|7}-E^;a
zts|&)<)-9#Hv>siIoY;<+>%axVHQ<oPG=4UM)Mp>>-yMsk&dQcaFb@!b^=1WR-X!m
zqy_z0|Ix<O!QFlHS*@#6KikyDWOx}cjC+M{hZXmPeb7mhhg)Rko!AhkCuss7DgFI_
z*8l8?d`DCQJ3*<|3s;}PtmtoDBBmzc#HM4s_-8+j!6fo_H5BX5Dmib-d?;)>zm(hd
zXD<wPx%^Eo|Hbl7{iwgm{57P0s&t1G{~*J6GXwDvC$_yRCp#sE84o`I{~?vT|M%Cw
zC#|sSryxtAxb3~z1T>A)9|$o~GC3O}V7e;I`&3d<TwMfKgviTrSY}Xxdc)w3;6%eh
zX*K351&#h9N+AX!PI$~fd4;rIIMXJwWUiZFOEp#MRc%>6qwn<YrFtVaslhsn9q{hi
zJ1vFV-n@r|>AwbV3%Pks7!@A!;Z@@KuTa;A*N*3LVWtl+l4qi@&4>5A#Z6Y>4k<%+
ztuyS-_TI&zGL>VNnlE45JgCqZNbCG$1nfutR0SQx_=te6maNFf&=}|%6$^sG|8e%+
zZ%uU3)*-Zjlu(o+H583XiS#Bd6oE(w=~a;4r3nLy5Q-F~gn*zZ0fK-^?;u^8NR!@s
zRX_v;?!5PT?)L|L`7M)~$;>(X>~qfEYp<pJ6Z1R@zI5bUwA%c{3Uca}A^iyI1+0NM
z;HM$_Vs>1wL3r#h|G%rCv}W_!Zea7-JvFoRUf^=2OsFh25{@HvcniUN)EkbSq7ZVB
zfKncEYHvj-l(hWXducp3+B^`o#7f`gk`vP)WYhGNxI$s8{4DbYya3$Z(dWA_Nz)*G
zG@c@e_4UR=87oH{CkI<6W~08q0IIv~d<LJFMPRkb&Gjov%;_9)?LSd(sr*@IRPy+x
z4w6Edf`?pGOgzAg+DRvs@qu3hAph?oX)dj5dd#m#g5VEg#ic)PmAIU(^cFOSot_-6
zNYH=?_>>6Gj6wHb7rXV<LLy85mly<2yMSvCg565k%4L)?J_A6S;y$owPq)k7Canu<
zH3V8TrPuL>(AXsuDZ0$OROO-3te2pH^v?|4e|#?i5vkfOAms*`Uk#oTX5oltW#T&x
zu;2(y)9I#(E}h>ruT1r5EV;wDgy!f!I>^Y~+vPT34}a61HIZJ;l`lsd6MRo$8YHuS
z3dqjHNZ)Y_2f`H;o3s5qW~$Y@9~>U%{PTRd6B;I^RwSv>Lh)hspyi3C4|@`%>5*vu
zq-!~Wwvy~Hn5$*i(pX!+JRI?uXdyf}J|&SmHW{K(z|{GIhdifKnsy_7$|PW|_-ON8
zB|g5?jIG+fcWbB0y|#$=I5!hF<2pR{k>|^0NasvXFc4}=PtMRv#dcpsX_YwHeE-H4
zu82JXL2?ur(V{F)L8Y&PtxWym5j@rD!*6eD<rUAD$@ZE@c^*cGq;<)@r>HR6E3n!i
zM>;(&RBhQ%<egmk1tw|;L;QHtO2*3nam$IYq~vNnI^9@wg%^><hC6z$>SF#SpPBA`
zyJ-06UErdQ>)ej9v(s8po@3abCw4Bp=JM~ai~tK#!Z2^qObjbgRZNp)+;W6OV1#fx
zsN`jf9Prk#a9SZ_k)YBS%6tWPod<3!R4Eo+Q3+}gR1VW=@<;J7#%fk(hO)fmU64-P
zn{5mY4&IpXT{U-YEy*2kcx|mEMEu4ffaY<X1wPi27)en|868TJ!C*O|MG;~CP|vTT
zf-*WA7os_<zofhF%foSP=97=yY=r;Y3E6-F;k3*cW&$rq0T}WQ0VjO`Pc^v{(%2yX
zdc0*I3igCG+D^x8-hF4KL1Ata@8Cv-nd9&Lq+vcZCvmtRowQND!s9PHkaT>UL=S-3
z@3Cyf*;0Wu-j`OZMs7RNJrAU5>>zK)`aWk?6FM5I>fUYDcg5U04-W~yCt_N16)q}T
zUtxkNDN46wBgrPc5ux7B>o1m3d!02BdJ8AZM00-@kft-_SbI1*RxyGfEE8!H<!Bqy
zsjM<^GY!*;amjIp29C1+I6I!ZSOAVM_hS_QF-72H$|2iKPbYnjn~f8BAcfULnT)CQ
z41=mkBL2uW8~4g(i{VS91V89=9pHBSP<39Ws#4zx-^AEr1GhQ{E1gmE!C}}P>`moG
zUPj-fUv3^Zdff8|54=zhl!h9PT09`42V)guCR)ao@Wl=Px@JH+d$(YNmh}hhX7BG)
zkWNqWawUV?6?(vO%lOy$(R%aQ>Sv?=0Bxut|NR8kT*b-^q#N`Z%-0OK@ICpJkx}jj
zmEx|qxg6pi9ZiY)dY5M5{Oq{-@*HG&KaT9S$!WH{*PKD+4mBZ>%CL2ZUKiZs+Eq+w
zYw#H5te5?>6IIUQ-K?d40W?o<;YG^UbO%M68ydo!GTymE6G_T4L_MHpG&W+rEfFOt
zN1Dy&u^DL|P#JmPSy&S~L1#5Wb61Ho#C@!I#`D?PsKAw7Vj`rmrc4ctiVcg1TpY8e
zUhZubONM~n=5#4Z@xuT7BqM{`6KX{EZ%vUj!`>5mkNRHo#nDm*2)FReAbLV2pf;)*
z3!YFWU`r8{qWQZ--|KrKb(oRi>=x2A`#XKDZR27g{9a$+@iB_PJe>GPfZxtd@Av7(
zQWEOxhCk?K`=%=h%wQ}R)NtQg`MuB<<D&B{4~FgNWB5Q6>8<k2FXD|YI@)C+sg=Q@
zYO!JyA)yMh9X?ubPT5p(1rz-u8Cxq~{F4Eri8#_^IArX{s?X`4DaEHjmtk=SzKtYz
z%fDZ^w1oc)?LGYGAuKhRv9*>9oK%An5#u5@;pXY4=ua)?TP7B;FI`5*XSiBFp@;1w
zGjGTyIKR8~Lmr&5ey2i4OVhPIWu=cvfqV9vYkRzlB}<98)9}c9q-xkZOVV%P(Q`V0
zilRWFE??vc3suZX(B<Etz)Ml?2fT^xt<x!F$$g(CrEV-*_1C!F4&1h!TgJFmPh73s
zoN4VvIm?AIJqzqwKz(r01j*j;R7w`NgJ(RoVJHf=u7#lNcvrDG-=dOHf^OM(!P&`|
zoUy!YO)LRbr=)hz^HGBGdOx$$nerS^sodXR@r;=-P^jc%a6YSM4I-ei7M3+Jb+*~K
z^el=PdZo!ZTKqA!p!o|@f1!^%E7;=nv5LwZ=N=u#4`X?lN|SXq1BeSNnha!G4)uOZ
ztZEvpK3AHMjdv;Qgfk3F=hYQej=^HhZj#ZFsL_0Ga~`cMkSmt7Q{H$dOo0WKqytW#
z*BSz!VjmCZDu$nKcqtZ#kr(|fUQe@j<&8d(o{6!`apn(EXjVVD_%b~(-mgcCHB-Tp
z)QTHZ!dH2?6#@HWNQEa@>C%^cGhcHF%y7W$vI*y?VnRDb$${{!`wu+}Ll543>$;Va
z4{#b}*~BQ!d#$r-bycB*1fWvoc4Nr;Q?C~IQqcgb@8(f*mj}im3W{_Hy7Ic|Y3#il
zJ0B3YOHhfilh<)!UrhbV{0N3XxE&qroR$G+D|28Fd1J+L6nj(j233p;h|(??ek$F^
ztLe6Kv~m(6Q*9pIg4&nQKkFLqUk)D1&GF0~2&mPk)-V(;cO6XzvNh4CJEtZ!7q`+r
zS&uWm*8ICV<<)$CaaMmuDR{bnG4-nV<j4|SZ<UU_X`<Q-3H1Y4>l>L!4yzh+Y!s{O
zs6JOsnx7q>4ExsSB*p|-k?=%;cod}TbN&tK+4KE3?#1C3ho^@XO+KL{5v>f&$?zGr
zF#Ol0`d|LJG@Eun4m4i$F2$iE$Tyw*0k?X>1D3XV1{Y7dRt8H?D?UA|pF1C(`K21i
z#M8<85PF4_fsu%OMkJaY%&CVucyfb&3{3F%=3f2=6oC^9Sdi?%RGn*)g&5g<+nqj#
zxVGfFQW{G)WTnx&c?$D>TH}(cKY_P>%ZknHVPAD%XrM)9b=GLyL-khUfiQ2vu=s$X
zSYwHW+XM70-#Eh)hl1ys<xzIYF)>|WQfYWYoQ6-^yTG6q@8%^VvBjvn`%?}owN=Zv
z>Qsu+c6sC`QEf6p%*qP9i5gQ!UpyUNqH$oRk9AO=o~}r3xRW8C&~3-@Lp;f?Xv^f8
zD6dK7BVZ=Sx-2(Ggx-&lo&LFK9**i6yt5k_JN>CH(DRL4jZ8)7a+H(OgPY29^2QY8
zdQ;DR*I$k}P!KUC8x{@T+;ojdwRe@SPMv~?u%b_}R<1!e;^gE?U5)BM{wRUCsU8)P
zxNA|ASL%B1=TbRl&Y7HimYf;B^QrE5pJXyp1$Y2US@L8Kb$~H4@YL7W)v>Wd-ZDXx
zPv*))La(a!t$SiBCEaK%3{{)$1puogYm5W?F1I6xv)$O(COp>3ej0D`+`zcxB~80K
z;Tdb*2cJ6!=+#)arV?{8o(^VJEY^Jz{Y*5Am=$T}=wmW|@41p;NxrLx6PM<6>-5JW
z;GM~_`4>qwe(y7hiOAaA$mAEei5a(1lPuT0QuWyT*gnn~!?=j$?<JPPbjqO*%l2AC
z1JSAk9y_i@JFOVhOMQcN`p|FvIRn*-mxm`_xjBR5X~|JCaWtq0OWv|z(znY;a~>q;
z0CxgY9t$}*cGYZk_SG+wZl2BZUW2AUe;U@n&BCR2!U-M$rR3i(LO-sTx3Bh(;A4lQ
zwAYUtNb~I)nt_k`b&E$|T;-&Mrs`IZOolL2R7FN71hT)eqzhbdUxw9NIY`$`>^uQi
zKQg&CTC+MH99=%FtHIRDck9NR^q$wdX$Z^V>vx~3Bi27}GYfDg`pSO@jsv0@(cdFe
zBFG&~heMEQ5H_Ud+srI8G0NqW{lEMBGv{rAhO!sm(l-Uf{0o>1GmFyAd%Z?#eaeBe
z-phl7E+91cB7b>mV_%d;-!JiNf~9uVSe<<_Wp>te>H18`Kr%P$IEFZ)OL?`Zejr3;
zdxi7h;^|N2SSR!N=Tb)ou7F30gI?jFM%Jjg)<C84-j$}oo6`;n2jU9ypzu=Qus_y^
zC)V}t>{v;uMZfXcjmml!&hMOZ&%!QGeEZ#c+-{py7Me<UIH_?)YBQ1k^%ovaj%H=$
zi{@=-BuXBbYxT$^GQZau{^WVWD1e25)Yp}b)}FaJ)^pr3ZSX_~M&O;TkSMhSb)&7l
zN9g?I?x3+iz&5QbQxx64q_@;z=~t-(&+F4RNm*I4&jNWL1oDyYF!>p(qSy8BqobxN
zzhmh<A^9WZ<DAmepV|T&NdNlMJP2Eo@~Rx0JfKvrjOib0HjLX{eEa4Q@+p+&%z0Yu
zF)p*GwXrrJ!9tfwSsaksA@VNY3*TD3A|CjmiR2GeKX7rXrGr`&AM9gv00|yS*n9><
zUi0u*nkm@Wbh$9vjjL!DMttkC<9XY1veC=(gfC@DoKd?(2u7;n)^AZh`CXot>dv!z
zzQ7k}RURcsfBL&TGPo?s%1Qh+W`xQpm)ncwWCI#yO-&5v+XCRmnx&=CfJ-(Vs`Hih
zTC?uKx{k52Te(5yRrlT9{&(8HTWN<xqgtf-K@|$FVfaHFqzISmHT7SjpjkdFq{->X
zj+LC3;yy?kE6)K!Rx%?utE>h@GR95Gt`kz!zDn;)3vuQ8o%oIpmOV0?i8NgP@!Py~
zRY)41nZ<xe0UnBoz6A@b8v*ES5ozu{Aoy>PB4*f5>tMmi&BAM?H-kk?bX;gH`(VRA
z;|JV9WwA+k<{T7q=Dr-iJZa8-(6-^}yAMtgPa1{CJ1X8h03V^AxlNQ{ZG30k!a5ya
zd9B7$YBc93#8rI;P*ZvwBlG6Lk~l;B7=9@kqGSwE*<X$!z>6!#cwhGV7Gny6H|dFX
zxV_!x+WJdj+CW~quVl6=Wb>|LmvRwVQlrzGf3CZD^=;c+bX|TH79mcFcMktbwhsyp
z!51L4g2G%p$3nhg_LwNX!m)b&eggt-$n+mSiJ+}pRF7_7^AWV)yc)YpX^IZ?^O_th
z0o*M-OHh8pRi0bDTg7rz9gk|$zhoBqe-qD>OK1y{&)MDETWhmp?Sv!rq5k_I+gI(1
zMR>d?B&heb??_XxX>o0OWx55@t)$k)`0)d3r=J|6%$2VDzXZB)(W!}dcoW*TSCjpG
zV$83^@SqYMKx8!<8o1{scD&_rwz6>qw4MK3xPYKl?zpSm)X|sjfJa&6E}nTlCn5Lh
ze|K=_j5`@|C`<uY77>3e@-{LX#}7NAIKEf)2yr*ic!v_X>7?y6eWDk-FN2U#5}Wef
zIn`7t&37&Kq!_qot3?W_4HhysPGS9S?lux>K61`SAf9Vy-P$dvw$9~6h>B?gqF>>p
z@StB5_1}BUxj6&bZZa4QwtSr(pyb(i9h)8r?|sPy0TUu6V^4y^WSX1)HE%b%eDp9g
zR+Podthqk%^EB))$-Aw;ugj0P^~{Qf`+G8sf&7_zGY+oe{yoqB`K2?&usu4@Kshq`
zn5OWrXkqJ3lZ_jnly&|g7_jz8;O6M8#c__)UgH?vPkgU!dTrX`(MEB6>bLW~x!PlK
zw=$>F+D>-AfuVn$q8}q=idW=yf<M5dW}lwdUN6<KO>z0@j%Ki=qo?OQ8Sy<C9lZEv
zcgw6Hcd!#_)8OyIK^_UEaP+13v=-wv(9zpUs`nxkX)+wqslr+@HyC;r0nc`36)3dB
zvaOA8SUf>3-{nX89X*VeatjV$m6cB*IhLn$V#l}h!f;v7n4i3yr{eY0rmiWBu;8I2
z?ffbA*-K@v(~QXgMknfcUH}Bom*>{+{TP}@$I4Jp!;s#up_IIgWJ0fXmK+hGy^mni
z0akLkpo#Ol(;d#9(eD9`dnp=plnkU1cD^af{$&2I-tr4B4kfOYkae9HtVh`ze~asG
zo(;S#8V<ENNgG`L&R<ZKb$tF*GOOAu(A!^hC0jcZpQd7=Uf{yc6*PZE#w)$oZ0-iS
zd&%2Bh}A(pz_knmwh_TjENjp{RMMMgdBwi=OaJM<JrWnstPglk3(7v@V|^#4JiPx=
zH*8R8u@ISGx`ii@AIq;Ip!1bboj{^C(g%jEc$hP}<n&qI@6UFBd`Cc7vj!R#{gzJe
zq_u{kEYmgfgX5@grTss$g?T53Dev6$NFO|*Kje{9?egaNxNH;=S{H|Tz54FMNY&0q
zM^2EcRrY1LH&7!hJBekNwzo?54j|Uc6`o2j%U?x5u-cl(KBhA6>yL%Et2uJ0`)$|h
z$KlN0#NE}`>UXbtY+$*)b9C@Xhtn5YnDgZ7=kxQn_zXHzjptZWM(*o>tUAz`xb*wq
zk`MHL6H@gnaanw|uTAkf-sIkS?-{scezc#Qvh8(<y8&;Fm7}ur1+x(EEuM(LzC}gK
zcy|*GQ`(L`0TnXrgrdhZMjOY5^IBI&33@vGUT%rc`EU*-Dlnx9eOh#1OEv_m&qF3o
zc~zOri8v%dCoC8c&7$%cFDUUcvqUP(C7ZY~JCk`o|J5@`evnvT#8;?OtD(Qwj(O-+
zd>mOK3J^!tWnLX}`L@e-JrVUVpUI!Ho%MapTLYtj`jMY>P#6Tu3N2t2-REC~^u6;?
zhFwwNYHxuTUAtIiHE_LM9AHKBOpZ62)!VHSl(THAiz;etG^C`ADYjeL-RgH0`xtYC
zx9?lc*J=OWC0V6F&a*G(oDrOEG{SkUJc^?}7?6q)O;~v>231n_fc8y6z-p;mSBUBC
zdO?<soJx)A8ZRQ0G!vF#>yKS`qx%BVm{hRHd}hOWbKlx!=oDM6mkqU((#IFH+Q`bo
zt`GJEK`3H79l~g+REQ4Tr6yqEGEb0wXHUim!Y>H-+vTr?E!!T6&}W0_toY*l(vlni
z4t`o?)i*UE-=PWf%oA|S5k%IysLUOCG=KkD5|t#_EUsfyBYLkv(X(@r2vgyAdR@=R
zV?ZVD(!<V(kR#^@Wg?>?@G%;OZp3N}hv6ocNm>Fk*MF`2<V$Wpok<MhmRcVq8H<zV
zSrAV=R@^`FvFu=mgf&{%ZaI}J=K-Y~byYrTKJ0&z=&lu{|NK;>p@otx+Tft^zLf`2
ziXtJICUwaA>N_OF;xR=*8EIa{Y?ZG=$NPm@KLt(-z3Uf;>cnd1xkk^phY!~}?+TUf
zof&eNM`$g@%+tj|zF@p=2eWs%u={4_$OCuk$Oz^`{v1+EqRc7T<!7I%MgfoFbz@&)
z@^)&9ZLKPw<WX1V>tk8G&LtcvXjW3wr3RUVw8Ao&-(!50OBTB?DK5ZbeITn!zq)D)
zXEzm)mjCJ#(N-xh)qX0#YP{Nm+dgk|;Fm|@Z`Vp0SxXN|&@ox|tQu85?h&>Z(9ZPb
z@||6lk4wkn7L@2(v-@iVfi1D<p1<B7ybH~B<Pyr2c%wmLGzqm7oDTVr!-uL^hx9xm
z-DX|pH@>qN7W!k8HtTV+hL7tqYKc?6HH0dG!$PV<P1aa2T}*YFUOemiPKw{2si;qk
zxsY8AP0T4xz5dRH9eqcQm^b%erx-D@I>W$lAgP~0^%+~p_A1o?K#%mhSvwENjMvKN
znHZL<_S1iGh{&RDXH|vi@qLH5-DNXY*=SKp>(+tuqA3t+g)7dg?Kmj7sqIL1yyc|c
zM|7o@_<kVmyMD_%AnxLDL#+77iB;0zn{?vQz~}TSWWGoJ_dltV{i`>ly=RV`@j;EL
zzCqqWr`un%nS<`Ay}gwX;$MOHy-(Bsg%UPz`Ni?rp03-zGU&AtQIF=Gi+T0Z$76>l
z8I!`s0*fG{3W~sro2trv$||W!8~u~9jtBBt^G$|CB=io@M9E!y`7d9CZv6F+hAWYo
zI@iJiUA?L-JFD)m<8BBUB8U0!Rw5-ZasAtcQ5qC-&BAM@EC1{!9C2~auSPt32)SDu
zS#B@fA|T$!5P{+m<@v<q>EHv0=ves<H5wM^zG*6T3FTMm#MOmM=8B{roFL?Xf2uE@
ztkUxkc&%~=cq9ufwD#8+n7;WoH1rmM9;vdzVwe~QGa~lqJ7Of;A|3=Y(nQ`ISk-<@
zML(7YwU}dq8!!c<yu9kbL10(uQz^R)!C-gm%(!b?XN!hC6ZN*l`2BQ|2zz}_KQ{7H
zHTl3LF?@~AXFvX*!Va)L$fMq&B3aDRrwN)iC%i4;#{EG)$oui0YTN^vB$?vJ*GS}r
zt?jHmOeG#Nb96sy0bPE=%Kl@81c7)C7TAlHpOQr$StI;`54|Wx?Oo%8n=S9$+S1k&
zYau$I^F8q2CDr=h3}!c8jBJQmNfk$CG(LlJ03IE*3__27+zmnEZy*f^mQrI76Rtb7
zj{6WC$1^$TUAwQ}Z%V%ZeK9{Q+LIvbxtb;bH{w@y;+H&D8maocH)TFN%7J#j6>{;T
z^(4Oc`Ic_3wSNjy)L4jHg%)3F9@TJfakH2IZF{Q>wEX?Mh~RuPADXB>$5tI8dbH(6
zA&^#)IP(c&ymOZaml-h@lE@paHBJkOSf+R=HcxU}ArV){4v8wsEi1%h3tjWqhyq5{
zk8)Lis)x$zl_ZV_3L+3`v5mes50~{BDrH~E(Fj`nJM#bodl<;k7r=~wq@hTIgo+oV
z-j2p!d75TjRIGPMJpI*t+Aiw(TO2A4@Lo+hSR0h`=*5sa1#@&X@MQ<wSKDpgJCK?Q
z^7Q&FU0GwtdJxiGjN$sGDO@hdpF`g!P2-S&>Ku5(9kQ?N?Un^&ZXXroEO6J%HW&|R
z6Ldy$L{g1>)D+?y;J~R^vB@yipDqkB$P#Ph1VL?@QUtZ8#N+qNE*z5O9UM3y?l$&r
z5W*GaEJ%BkZ-j>va)m4<Xttx{<C`N4iwrpp8TKXx3%v<&mN{ts)8?uM9Z8?G-uMUB
zLXV)&+t=dBub0tF23uxfA{$-g-mQ8&9Oo#?ch{iq-H|tEw$&#s7MD+e-;#@oj9W#P
zrF3&bA<D9ZoM~UP5DAyU+i!7zuG($XqxqkK<tx6m(-xALVj{Mkm%~ux6`q-mWar04
z$NTs3sJvnVt8T6=t&%{mlLTFx>noQD5h;zWUylzL8?Xl3^YgZswgDBgvZZEQWN^Mf
ztL(w4iH{p+h3Z)gjHu{`!4F7zrBH9>I)ZkmWq=k!wpY!<>Dg}m<-!#pgY0kT^<1Cb
z=mHUzaa1Li=IEsc1sXzvfZ~6r7TEp#IKFk{TFD;0(33Tk6$_0F&GAJ-A*c>kRcsnD
zaWmoFa?f|Z*T|LaHt!m)Zy^dLQu(Vfrkz*m-tFNVSDw~NS@hjLHyJSDaF-~JOJ~Ky
z^GH1<w~m}DCq8z`8@~dSj33^)<~(2-IE>NF?7s$<Z?(BL-jkJ;HRsG{wmRZ(<Ay*!
zr8by@3teHlsz@AKqKmlfQXzkzOF&mmOW4ea1Qts|YEPz{I1~wu!QlxTsyZ#o@Vo74
ze^uS$<CAalJWKsZo%emfaH^_t-ASm3(MXnASzp1j=VD>iTLIANfcnu`_Aa3XRVa<q
z5Q~)|-zCqc`r_G{ZDB(1rAk5Jvmg_G@%&A0>H&*6%JNrZKib=*#|k9Vgs!lP#jx6A
z=^*IVR!p&PprokEWG8wwKaN-n@j>4FzpSA$GS9bf2&&ZXF*Rn!FUCMg{j~2Op<UUy
zz~`elV40SKCm|IRyPHB11+Q=%M@(3e>!@HB$tf9b6Bla>uTr6TKcU<rzkMJ$xrP{r
z|2KeE!!N$_K_0A6jcUN&ZM7@ML-NuWA%#01_ddNG3K+{b%f+XD3USZ8{gOLU!+r$9
zt^&hcXIRQtvK=5vNmjzdXR7cZh<B*d+&Y1wezap8%G=gn2Wjw;nLIt7ZVM$kK5AAH
zU_7C=dAACiif~fbVSRopNO+zLCf6P<Ds6JtaVow3d3U#MY|%?zlz5~=f@VW=bJN~}
z#>G+Vm96{9Dm0M{j0bspYvqK!EP~6&hg`iCT&)Ey*X-wimh#Kjj9s~Iaof4Y-H{o<
zGTHij`=Om7{vxYjj^kPI$1NY8QhCR2O9f56SRc0|333T<z7O3H=C@mQ6W~s0t$gAs
zUY)<w+~)=voqJAv0%c5w$=3V+PYXSKVoVu%9!yyy!YuLlSDh*JNpxyjaWCUMC6`*k
z&q7Ex!C{u#Hc!ks-6{)@bt1o60eJXrZoByL61Uum60+b9-bCXzgW3lde6-!2ydC%8
zgZ^syqt&v<KWG}&A0m}#Z<Sc4!C*4FZBFGAGy-b|-^8t~-Ywstqr55Lpo3nw<zQul
z|NEKr%|vyfPYshV8}bzY&B;*GflWR0i^-2~uI>)2g85Y#tEmdBHU<U+8Ib$U=;K34
zKRq$On*znm#y!Ee^5~ci<#I&^b80URi^@@gSFcCc5zKVk1a%uVDat+Cp?S4Q!~Q|u
zqKU8Z9(;*zi@$%xnSrQh?H)w_WMm2(FN1m)1WrP#4@2e?USbSi|GnA(q4&%XRGdK>
zpy_z>Q?uhq<j8a|?o&S{fzPq;9;=i2vBq1=TPpnPz7W!_^%;}UfoB@L74)U9{F928
zF<4Qy^Pa~We%M4~IUKyzMAiJE%kbML__)XUbtNveI-QoivPo)*y(R9LJCV59(UQ)V
zkyi=Csw4WJs7ND;7vlyCQx91J#hg~n**=%b*D%@zV~WY5@CRvrK#^ahxu*c!M3x8b
z+|Hbde0KfyxH<VZJyK-PjF!Fz!;I`OEc5f}4oQQ0n)TJvF}C_jrxz_0(Y$v`>pjcO
zwZWXJ=i_VBXRJ=;i4S&;!dOVNZs%2E9@m$iYFl2xN|F?6(~Xchtu|p`uj@%*K|Z%p
zcPRu_ddO#DmH3%5fvjg&^yif{XT?#3WzROy`#khTeepQ`(o=Pi9NW#%=`w=Q<agRm
zjJPF$6EjpvT94v!tqTZKa*QK;YcL^$$9gas=(PO0IABo>oNLP%b7*Zxf#v2ZEN!z5
zHoF1m4J&YZk<#BzQHU^IUWx}VeM5cTo+Y|R7$uW+Q^~XgIP=(A)>^Inq6&6RaxLNA
zL}qv<mL$8C3c(;7NqRj(hNvuj3#!TM&Gel%vZt<e%lUkE_V@m$=m9?%Z#%1s&{|`y
z_rgEvmwnr%90&&8Veu{J_>RTXtum2d-hA9d4cC>l6L(>?7(2fTbK16>YHnk+V2#U~
zXM?ahzB2$wxgP~?o7yl^2_Ulcx*-`QKNLa98tqfrG)E;9O$p*HalYe&Uh>By7C9H&
zS)Na2m1XH-e@3%-w9WS!KVt9d|1{w~R{5=}+t@MDl5@Ji1(AnlV8{}bc{nhf-up(c
zBQ6z02eKe<vH5n|@fD9sJN7J^E39Lcy0HsCWr4wjK30Sjl3n7-<KVu$SKra`V1W|C
zDC;wdH<~%Cu~7^pni@a$6U7sJIHZPadX|z8#9un}Oki#|T2)%<EP-oZ>D8fWY%D1+
zNzEJYAum$K;^3b%5is0MUIEAxNB&iu%II&D{m)m+H(Qpy`NMC*Rj|A%xHWkiLrIra
zz-rTQxU&>vrPY?o07)77^<!c4B8HpI7Zsz~9u->~J57MMNoro@U?BEf<@^LmzFsJM
z+qUri$5@ikWxZnfnv2mW;_ElagFfD8@9r~*U7eBDghy=Y`1Ud^PmKf@%XI7AuWV#D
zXlLMso3b-XnO3u3+<zF4RnF%tX%CCg%cx9%G63RCi>yIg)FTK-w(hjg*SH+++}c%&
zNRhcazW6>s$bNVCTi3Pz8`xs~!`(K~u`f0GjsgAVoj&+Mu&do4dPzs6{;-}C*nd0b
zpVAs*R}KWhx_PTmJ_%Ui;Z>~69cK@%nUHMhc3FygCk8_v!8wb=X@7^IJJ+CuR7(eB
zCcu!rF;b+~kB>BmwrL?BP16XlI0-Jdyu0th=<TgyB+B!FwbMo1(D&QK&nppX?U2q}
zIxS=pUZv4ulBkNv)97m+(v4D6S%J=f(tDe9Q#2|=FIOT%FOQTvqBvqzX{IT6@vKae
zRyr7!`S>*v<(`6c&MwWOQBgU?-!G_4wE#9fEG-No9ff1R{{|U*)i~dSVPmsTpN#yY
ze~po9WzQ%Tw5re1DSuu3+d^vyD}mk8Mu=+@%~9WEBb-r-EqsBqcePK@3=gWg?C!Wy
z6niDR8uL(i?a{Z=*~%SvNy>H&VQWaaJf`@4W!%78`Bs@+nD-}H**MH_sg&WH&J2}A
zs1TqSFcY8~ylI##v8P*Wkgj5ydUe8_U*|xai*tv2+s3?vF2;@zk!3gO-g%?Pe+jZA
zo@(Hl>^~)lX6<S<fTDSxcuQmvbh-)Q{^FivvJXt!<LogZMWq-GdJcr4Bh<9qo+}c|
z*9FI?I^I%?uQMo-ju?onE5&#$S=*fa0Y+vgf}6>l8H&Cs=o$_TOD6{7Ke%gPH@}EO
zPz?QjMur-$JM!KzodE__YW*SRTSE8q_v)V>2hgfaZL08&-2Bb1wFrBL8~RgD2`i{+
zdn_J1<samN=lj!Z=ISuqDHz?JO?v$;!tv<&k@v-7Yf<oc-=wBdv5O>c1*$5=51?g%
zc<(EbVed!w5g$2!lgJ!5zFKq*VoUegcd*E{=LPH5B9+@UPG0sR861;AfU`_JE*38X
zC)bclbEPOLaxibVip_*l(xs)cO_RH>M=SO383{POef=-k38AFQqf&@}qfe@G-60FX
z=%qk~5|xf*dV6b?ici*L*wvN%;Wdu<)ATX+gzvG*nec1~>HetRjAdG021zoMU9fMK
zvH$z1k)%Lb*RS>U$C5T91xzqJI*i3$6A|M7i5`tv#w~7SD1+ycjAmFSL$3qHckMB)
zs)x4UYos239(dA6WD^#ZKpn=eSxsk2^_tbM56?ElrzA-ZuMOm0W>AV)TfS^bi6zyy
zpUfH@$j-^VC%5O^NGzkL>pUW8k32zEakhsa?k>b0U%^1M9F-(8Ib&toe^wAZ%?|{i
zN=tg@t#rbg<lg?z!P$rvaueWTXDF&+mJ&l!gSn8B%1^H>{B$r<St4!;u2U5#bwuhb
zdOBe%>C5j!3(7supA>mzxk7tt-J@R;MM)5@`uedcQ-Yn{ym}Jn5%073lwlmCC}r5s
z3I?Qj*j*z@;(d4Qy$2IUY{(C(b?JtAE$^x0#ChMyhPAyMSKR5`$^j&|9-t)Y($oe%
z|30ayHS1YgSh#4TR|FRhU<-Peo5h<CyegN!AIC*vPeUZET8$F~i&FcYXd%4(ci~_J
zuXt3QoX~0#^N2(FYD6Jpv=@^N94;>!;9-8}wCW1#bA63i1j7qKf&2NZ<Tb@|dh=$(
zqA7*3HLC7YD@7|4K$Fe0*Vxr;zZ~D3-{LDS*}gAx`Voj*33r^-3M>UFA4S8fI_{9j
z!&R}4B_l2+MWf%n{FGE#q@euXlXt1!IF5~4x85>;9Pzr<(Vf=^o>=>BR6JogSrukA
z-skiPN3i#_ZZT6fb*KenZ&EoEauR#0^WCGxab(e+DSD7i!2UhObM@cp4^wU^i5350
zae~IYy3+N@3W<+=?kRQ^PBKc9X-HG}ObsJs3!y~1Zc4~(NpJL3^+`L8Hd@n4qv6R2
zl;_0d(2e3uJt8zBsyixi$__~$mokKf<9I~xO23!lxE3}gAvBWxpdE~DT^m&LlwA>b
ztLapb`}F|_T#d71FxHn+4FpxMCMl$yRD<(sWbupT1%vkv=Si1DecJ^qcwfMS@8t$y
zZp(P};aMLDtu8se&R@<RU#j7XcKKE=U`)8LvN!pudYi8xhN}+5($K%T=SOt2XPPW8
zV6ojSjZ>cPH0b-{_T+Y&#D~%)LuyQxt&x)6YnNnmK&al}$%q6UBYs0#hc3tdL$a$<
z&MhNt0aoRHB)RPY#alH;=qGcxA8aVNL}V6g9rKxR0A@Q&+KmsfQ)ZIebZo;tX@h}&
zrj3Hb7?*r!a}~-1PoouhWNZBYcB>}gmc_FI!ZbW`i>`mDO%2X3jG|b60D(9K2*M4P
z4fLo1NQ9Uyk6ww&8X^B?kVz#44+A;f&9ALLl}C+ftLz9IhD2o+3DGp^sGk*KiYyF6
z1Q!PwR(fS)eSQB&M?8y&{9Vje)yw7C00kzmZ%%kZsasvCog=txr6y6eH7pvEh$8C}
ze!+@^^-&|*nWYi%_SDJmnheM8uQ|9liyi}UC`ruRph`k5Gc|9!zw6%i#IRKMy?{2~
zaI4qXQm!Amo49OjCO>kz`!cKT`ExJ^!OdwcD?hC$Yc6B^C?w>|)6bx#y<8%LoPYGn
z5~q<&P{kJcIL9DlX@V;!KPLT_4$AonQXV-x`tqBd%7E0K)Nov~x8vj1``3l>nKk6V
zezM9)l|+oot$Y|y|Jlh?NnO;-rHTp1W9i(T{<Tn#sk1Q;Gp!yA^YNuk_aoC3D4Hnv
zN=@(pYAdl{_i7%6@@=2Db<)qfM%8L*)vi2q1@cS`a#n?o<2>GfPe`GAn@GLRDmwCz
zQYGqDuam;zL6gT+xZt@Hd%TlApYJh%+EQf^<}4W`vA30_D`=7JvJ)(R?JVZ4+OGN2
zK+j&a?~_CYAKLWx@bW_sVWrXTy|;^4x?N&AUXZY?vdRV_B^NR$QBu0`tt~O8K`Z-{
zRjsJK(j-MAkLYWxQpr7!&W4w+-lC*e5|F-3Y@ASDCJLw^Hp+vh*K9mc2vP_T<=R^!
zMb&Zwd-{C`!Vs&EBh1|21Tf!>X8P%<XGE1ilAw1Kzu?+YjHgb!Ykcuxlf8JWc*lRG
zJuI>UL)zK9(GrxY$l1`T!Np{=<Rf3h?d)PbL3GuaRyXI=W?x*AItyBgz2kW~{pzsw
ziSpldx(JV;(f4#Rvj%IZ8{sX0`6mMP>rCtPi=9VROjOffhK$%>?4&a_xi>$3*{i`l
zx0`lMd-1dJfv^$4A?0!X3C%p38JBr7R2`!7W%0(&Gy(ayO0R}n5vy^ev-2Dn=If$D
zSYVz^UyIl|gq8gSLMhWoKTxY-TIVFs;siNOUtS!We7IoEK72(g-ancKe1E#weImW{
z2d#Bmin7ILA;W_w_wrJ$Vc|aAY8vy08vsZ1ovx{cI6UbTXHr|#uem>+zY6^Su~u*V
zt-5Lyx_y1=#-#|LKxpim6VBb&&q}(K4X@QTFb@m;e)W9G6eb0Qd$`QMHPI+G56gm_
z{q5g>ZIWx*_~$PeQh+4Q0Z}&yYx{A<E$PseA>Glkr|8Q+k%}hazd+g3<)0z3x6dxW
z$tjFQWJtZS3>!PvosUiwIbQC_-}%%=C%3(?NTTa$x5D}4r(F!^#nflRL%D76hjHmA
zap5uG(FOLd2Sa=Rooc!C6M6c-=ecpnsWtz1vG+@df^{J9ze`JfwwcetL!a6t65r&_
zfk&RybpC(b?efOgKI_vT85fHp7mG|2Cx*GF%!sBHA+K8BrLv37vRU6=q286>^-Wv#
z7Yp1M3q3BtDxKWZp8ed6lfb5fH-lljei^+hfxvO#T;N=o?gO!AOYku+nHF}Br~L-!
zE_UX^ePS+GY%W%k&VMG&_zz<ge=#Zk`jrIyGcF4m&8P|n8%3ZCFj&3(n{@fN3;5U7
ze6iJhF=pX)-mP)(ceGT)g7#E{NFeyWk7~|z&w&x$|Lzht2MkMv4bT1SpM$h;|7+#u
zS){zTPXhT1Uh9T#SW1+G%nx2{59S<XugRZp=mNu8vqph`!COAVmIe2_4ZM*H+?}rd
z?;WaVo=i=YJLjHnI|KigxXUtv=FVo1M{UkW!TsKf2yl|GodKD5;67tTTr7hqp5-ex
zf4_Fv|7x5&YaA9=IIbNekDw3Ve-83_ZE+vpg#109yF7PR*fP@%+cX50q{5d_Qoo0i
z!kHoAJLBOyC0=1$x2=E0>4yJ)%_401S7hTWXH%bdfA>G<dk2|ir^K?5@%q`LRjK9^
z7v0cF#q(jsbC#p|j4NCgDchH}z|(bAZfFGe-$?G7t<w6{gOlp;t?Ifn5HM^9`CBq~
zDinCxe&Zeq#0xl^60xfDZ=;o(279r2|4Os>pY4gdHtzEwZjzQ-zqRl`&cL6}>t19L
zUVo##f|jn&!5|{xzv5>a<1YSt!us7hw_3~a!Q-b%^``j^)z!zHQcb`wHIWboNW)&(
z-Wdp+rH6=wPV{mGFQ;@VY;s@z;#P&(97x#so$XZ5dAHF*_Wyg2W%b?0<JqL}!}s~j
zfxu|7b)}`S+H!NgcHm#`L^t)-+q?C9P1TJ%9+3V2Y8;yDZeKbt-HLl|a}acKkP8ed
z96#+aecDl=c<zD#PU-uef~Iy1uDSlr9yc9VmW2t4HE&?NWKY&<y@J{X|Md-WaBwti
zS?z^@GWO@sSQ_T(FZb!WgXT#fg~Q3YXSKQi^?mN;N$%zD-o>sey@F9v$jp}Yv3B#J
zHgIGL90|?+7CN2{-J`#lr-!z<OUn7J_APN8MbbZ20?@k+2Ll-m%K~87(R7P!@ad{p
z?n$3BwB>B>@&Yu8xw91R69%V?AIIZ#HWxKcj!la>p}WCb7IU&${W<>`+5NMPUM@K`
zfDJP|9WZY7-^&>0{7wa$(Grjr#mfUl(w2jqV|hqW>lJJ76%DBwe<sL>RIt7w%d4~*
z02%g1WO_lXlFHeBy4QF-W|MPqSU}p6Dds#M$NueVQs~Iqy<?EG1#G%Yuj_j6i2s>T
zSQlqt!)`!XXqVjVuXb76n9C(^p&oh2zkRT6{_AX(=D!_t$6<4U3twN%{97fHyMF%Y
z-)R{rTOT-g{y8aZ`#-BMGAsZ#J8zwzot;6vFaDj>eJ<rtJjfh`v@iv7TLI@sm|>&a
z-&`w0_FE~z(^)Pk&rCOf>GzwHwbrL#Q}1#4z0GX=omK`mq1@b2X<*};!XA~KRQBHf
zU$CVv66JRP{_(o}bA?4X`(D`TR@3g*pW|a)^CnVo_@@PdcO?PlL1PL3ueVMHfu>X_
zWH6{x3^HF@zZ=qPG3&ckQaz?%19~(`%s~}sQW?NW23SU3t=M|d4E}F6VM7Br)KL7}
zO8>uuG2Z$RAr?BV5L>u+rVEns<(`fda2*%!{TtXLi?Hx&JTeVmG4%us$GbQ#ZUDb;
za9;rXF}X({ovj*zet|Sl|CN*{B_=i%3QvHpTY~REYVPTQ&A$$thSSlqrk$#(dx!7n
z71m@1Wk8E#85I~8d=htY*R%0x41Dq6z2L!6^s-#jlH7j-;6H~j7&?34S-(>i!xgeF
z1sq5L>oMVlyC<my@+;NAS+zDY5SR^KbF)6c-w3+MXxx4j2!2&Ko<5iyZ~9|47&4Z7
zv60)!${nbSZUFziY5#>t!{zxN*ag<VUu86aj(uWo4*1UoHcgvPQ8B=_12<_z^W|Cd
z<y@Dvy!=1=&_^+DY~u-iz~wP;X)$*?HW>b=931SS%{x;~C8w`}KxvYDow8MJOszK*
zPkO-NI`f|s{`VWCf?ESF*5*zI0>c-Xa{nfce*j(kqnB^`05b=v_m}qjpeMc&wSPL^
z45nX)n-Aog{XnDK<`0i?yDpb``K8lS{WAA!NMlaF)8$MFILlqiZaZG|GW@^UZ|N3v
zZyGpnFVH|ok%3Ir5s;(n2q=`1mzU>F`^srgQk~oRd=S>>*yGvmO8mlA8wj>WLE`Om
zufdq3cF)z-Cr^bT`#Tq=ITVQ^dd-~G)j9Rybsn4?xeIZMy5?KEUHyul<epiJ-YXr#
za}t&^uNJQYi)o7g9ObM(XW4uXoAOJl`va`ZhHU5T<9?mhWM+iAOesu-F>E~vZ2rBy
z8}92QAKY%5-KX2cHJYbk1Lr>Kxr#Z|IRCTmY`ePaR27Ct|NW`tC0|Q|Eee~~{#URN
z%>JN#LkrpX1-0wxN_28UYognYLN`B@CC$C+^`$h$p2t=^Q(Tq&l+(5lqzh33HC)IJ
zcZ&U9c*|vXU!-C3dlwthmL@~7Qx0X!#l0dQZ9Lo41`kkQ|9Cmla~&jFPLP^%{V9Bz
zbZpJ@^;gz%lFaUMS6bZSPyvuilnxH83bWp$)vY6$|103TuTg#U`TD=O<2AG-SLo3a
z%i6cnBHhgod}ZrKPY>^T|LDA3e}Xn+gVtmnmrnrMr``gkDnn%^#VjC_l2mED3UpMv
zo!85}CW98fwhcs~?1sIyFib3AL8rBbQa0mrL9$g;Dx#DqDPytEtST9J$w~us9PuGC
z#Ji#?D5xd_kp~q3T2$%peF0qFXrQkR+2J1SyulxZQGeDg!hEJu?v=)5_19!-4G=e#
zQ2AAEXJ(37NlO*hUdc$&_o|tMm8VtPq0k#*{-%{auk3DZA-e?i6O?#76IruZ4ZN#e
z@J=&#USM|x^YTY*C8UaVGm;CNW+4!zK5UdGnXRiZ^(GA9f=@A{<5CT@QV_(=d9Zu~
z%>IPQ)tJ$w#I}$y^y?Utv>0h4=c6HHbe=lQ4m2p9S0s?PxFQl?s3R$!S^9}4(s^c1
zahiq+@&ecA-Z+u4^iT^`UWEzEI&I1v#Fs8~KQ{QFBcr?3KdFO5<{Lc{>0mX4Mbh4B
zhZ~Tg0i}<}o)D*W<9@6D(Wk6kA-wAR{JwSjE%`CrHT(OdET?9Hf1-z_6py0AhxGXa
zNFn?*I+)j3__WRMFPFDF|HNK7U_$pSabGqomRkz19mw!-aFv)8V+I5|Qw^rXX?gEg
z9S!XtgZt3jduJyi|9VNmj`_1uJ^k<FWkB!ZqrPWMEJChONPDV*BLN;X7KqWB_ug2%
z2x(+U@f<5ig^CqYDoK7v^u4=}nW|cn<}5LZp}g%!&-<i&*JUtBeQHp6m-}PPhPTA_
zUYYMNpF2y*XHB_B<)v;(bjR9A-tU=W1%EPbet^mYwSxTo?s$R={tl6j0*3C6>QkD3
z*rg^R1}R&rt}mW=(ylrWmyUu%?-rjKeW{g^hT=1?N7To=Ao2NLc(bX#x@Z!}y8*+v
z$%l>#iS~4`J`grHRfPp3kkYcUxag}j9W7x~PcXUm+S$=Q?1XH!6l~$99+QwmPQ=aj
zWET|<sp~x*Xsb$(Nf7LY6H&#^2i&rCa<T<W7rXkU6V;6S;vz}4hDL`No@;b6yNpR$
zf$G`Rj*vV^GJ)av_~0b0w&F)yYkW;nBr;M$Cv13I=-QH#-^OBoAh-t9om14mJ~^8I
zBQ--OA1Xk?o8g-e^#+2hyj28Au<K3)g93BcbnRbDg?7(EhL5=U@`_gO7MX<gC#<iV
zpj)WT!<evzTa&VOQetE`W$H;0q6l8mvCpI<A9V}|I&>JaGf8<zr|bj~ft3EMExgYP
zW4UxMrK2HA`Q1)p4-C^09C=wq=14Uo5_y4~BbR%7tL0T*h5$9zv)E4I<@?mH5lD%m
z*vN@K^b0uE5J?=f&y7H^#w>ZVo0qX%o55HtP)UmXSV>278LXlFkP!TnX8zumMJ9Eh
zoVq&NnA}g}H6k+h<bFI-$<fJMM4;F;Q_z$a?J->8!KC8fWydOPIo%wz@wjqebvaR~
z9o6}e(!Q_57#KCKgc!zw+Q_P~(#}3C4-u?8<ONBH^`~|EBdeJL=p*Z^j8Vp_i||G1
z1Xm}7EhE)x*n4%%D<v}H>vdw}um@2@$wW3I%nd>3F9@yL_U;jWv^|{;Gfp9hb7W>{
z8lyIf{fTZ>@P#v(l1m}gjAJMINc~<LX!W$_cc%h3yrm4&?CDxaQs6B{(MK>c-Y8~1
zv8*gSKC?FKEEfNua*E__X?kA-grs$7#EO=)n<0Wk2Mp5-yk^eCW_L4`KNdEd;UIy;
zfvob=xKG%JX)=CHqn2cMNts0X)ah1jJ!>_E*(|Fa66EgQSPTz#m9qks-zz-E0~Ms>
zumJPW>5565X8J4bp}HTWdCAG?RxDJ8%aSijwJ}WC7KXk?>@zW`WxpIPwtV*wg;_Fn
z&$D8?diUl2FYo;`>h0vz6PLiN+CPhp4)Zj(>2BHIf_-+wL*Htucave-Pwogi`Ha1&
z5^!Yfa@R6nD<>m>+v!L+$#*bx13YF5jAiF4xQ`|v<FPm>NGkW^<_}!tW=Q2N8NQ%#
z$?_&=w3Gv(%4MpBZe26t1JNLE+2DwRF)j{Q<0T;jXA#pMyc<imniv(etJBR4l?h5`
z{LUHCXKbvdb2XNvih_<Gt)k>DMtP&XL}%<X=X4)2??zr7R_p-GOu=rNB3_kaqa7?V
zvYy3e-ipkORCV;6lJqqKy|Wflz*)*!g_@)<IXhXWAN@PE%Lh{3(g#CFg8Z+2Iy##V
z&>T1iB|ovNV!j&c5avqcmS{roEkY-aNleDgB5;hkWjrW?$^cLQec!&6hx5FE%;%>X
z#9dR~XCD5zHCaj;fe_9+izz;DI|}LE{16dgp~lT8_PV%+NMwiBYB;I*OCZUoiK92P
zsnQvtzDlBjN_RCxO(hmuJmw%N%o2f%qA%ZLr6JD>G*K0FcG*!>1bqe_wM|y=F60bT
z*qT*V9<uJdYZ2F=5=bBHys?|T6|zn)FOHp;7;2osK?H$G#!cU8Ig}o`xvBgx=%dY(
zjLe+KHd5Q-$9=-z8Tr^bnZ<?}(T8v*mBjC^23+lpnCCH6M{dXE+!Teg8R+-i8dX}_
zRf{W3&Ca~9{X~oc>g_M3-;viIiQrRT&Aa7_c%|)Qs054r{)UsCE;PQ8;DXguBT3G5
z&;b?0Ud3Ypo#78)f%_|MWS)~}Vg!gZK>o3E)QX)FBnUFTufx@67QCJRn8N2!CCcs*
z7axbOeP0|;jZN5D-|RCjs4MgI3eXcX&LqTAomFLONe~e5yTR16)yxm!WTL(_R=C_)
zI|X;0m&qSe@AvB6<qgp{d?P>*l*GuXu0$4~NZ$dpIew+L^jpgxqf`$pm)G|-`9++p
zY+Vf)8LslR2v&B`#2OCAX(83@6EPo@5=(76jL_IjqE8=5pH5>cbe^L-b(&@~>g0D;
zqGk)p6dh`vU(j=1g;kouk&MibJWvJwsqUF-J6vVB$~GE{Cfhs;phk~YIn;{U4p$va
z&;Whk#CxR4KsY;JjT}M}Rar#am(MI<04Y|bi^yU`)KSs}K)%1q#3n<%E;~RGVGh<<
z7$h^|W<1Ss1z4ZX_wG3h8)wZ;O??Qzy$aDu1%VrME;pO1B(Wrw4dQhs^A<zUaeU_$
zt@fz5?3)Qg-z`3eb-&BJ;n$m=>T^d15c3ODz#>JRyPYUpZKv`@_)xDk=r~$zryY@!
zFJGh>=wRGw`0uXCzjPqEOIE=2dh2zBpa7-mRWfj9^2CQrZe<bP_?EiqdEd-?t%))$
zGZEt_Fh6HYi-mK{u%Sf3ij)L_HJ6WK>?Ga{MNec^TpeJ%>@lwj@-msPKR|T=@HX*S
zEMDeq{dJ!F79<(JWLFFJ`-A%h_YcHHW14dN*AC!?smP@3lpkf)h?#F;Z~{)A;<g&$
zV?K{k9IE}H7Fe@Wfz=s)Ow?Jo&3p8%cDaW$G7}}rgR5t!^fcf{BX4<PhOz6?7!%*P
z(DSF!>upYa0&c?03{+GyAQ2iVxrrExeAE>uVJKxR1u=YHQV{(;v!%-VN-?BRwdj4;
zduS{h597PyIA-L;NK0faEV4BV2dA0oE*7}ofAywxP|7Pcj<*}K-WsB4je~C97t&i`
z`kv%-Dk5*y7D8tWQw#VO%&LLvH^}(fqREPlWwC0`?oThzVa-foA_eLS-PJV?aW@I;
z6t`FznX#{b$+q!d*Z1dB3_qs-LqAE_Ih!!9{ZRUi?F!Mn#ZM680P4KGcViF>>=d-U
z{_e^HVNY^VBT}2G3N^GItR;cQ*1*LfN{Q~C%dh)!MUN%*Y}X<<l3%Jgu6A=62Lx0s
zUM5uqy7=-<REZrAD?owF`(FkQv}V3Yo?p(5;Hx&|iJDH(h267JQp&-`yTkK8bYA$F
z36PM<|FPb(xQ$&6!U=UpJnIuLM|Cy`4>~d`p*t6OAdga>{|`;)9?tat|NmKNL)o07
z12d9_9F|kg8kQ)^oO3p5&MP@(Xf$W59EOl{b3Pk6B$*Gxn~Io25|WVfDg5^P<GbtH
zpS!Nt^}3$V=N^y8{r<e)EHXfOKnNQ`=Isu3OJ1Xp1O`CBv25?SWa=Qow5u{Owg@2Z
zt&x{HCtLO8ITQ;iqWFsMDxpRJ8Q_Lx2Ry_f;N8v+Wb03;C3C@qLLy>>0*Iu<ks&+K
ztXUoVbP~dmyOs#aKq$h)OmBAf{jkQ5+lvf`;1FH1UA339<bvx+3QFf(OzalMGP>Us
z0XnsA6|yc=1t+R61|VG?#U1R=YyQ;So*(^q&IPJV^d>(b`pOej1$|8I_)&hl<&u(o
z`tDBGfK8J+5q(ZlQc@`R@3ut65lv+OD>e*MsJ4{ccow}NC?X~{m)p^HvfgoQU;ub0
zDUP4*MgSoTGoMHtEM+9KaUqkH)1MSBN;k*2I^YJ;z151MuJ-slchh_5<4Kk9Z=uT-
zdIm*0VZ-f-1aY~n&GX4P5<>C5p*X6DCgEZL;$w5+*XJxIc5$KfIb~sm>K-T<Uwr*&
zRwGc?N%}I8YgZ7(V-2>6KuJ<lJ1rPrhc39n$e~JbiFYdEcfNWQC`XX$(;ENyCnog%
zRXz<twN!Cny2?WxQKY~ATBy_v2o*6&#T6hpe-WssUkPQyFDpru0?J5x0_zN-8`N!s
zM1QjK+=2KOhPXP6%ty$$RP~x*$x9qhLZ2<^o2Ep^|8;El)YEEPNQm>>j6*n!!}VV;
z-j!401E>SoL`1AHx1Pjr91igbproN}88%<1B&rmV@fnA~z&wMbhDzm<D(^m)ms+8Y
zKoYHZ)-vUgy!p^JB_cXO2+Gm^v^1eVEk&9|DB|gbK6cy|Zj@ExOGy#l54H=~%56y?
z_~h@B7J2Q_2B^za={WI}*@uspA1{}_2ANZwIf&7^(d^BhY2^XC_Y8`u+0Hta%u?e{
zxFM9vXKwC^;uB3y*3i}Ca9jC+K*|wZE7a);CeH5W?8bOEDSfvOSF!*@{A)A8xj>X8
z>}SuvY@rM`MQ#K)H^tg!|AovTPggln#6i&9XEikv@TCQ=_!h>+$L~E{tOP-kvK|4X
z3=Ndvy4Xk<Uq1XEyW71~uTD5#{sNI?1{KBN<WR6~DPXGE(!#7L$$B135Ya;$EiaB^
z1f7{3%@{179vQyTwkHIDkzoOB+UNe8HaHJ)U$}S<xm&!rV)mh+zrSLLB61dkyT&s2
z2|Y3$tZR}2Ecpn@j5MLo=<B7VE@+Qs;<%syBCX=n<-RU-3Pn`#f8|EFt}aQ`JyK3e
z^z4(4QQsMzG!Zxz#7*GW->YpXjR+U}&Iy1-h$tY$+g=K%#1a>6u|n;nq7m;Iu$N_;
z%8k2PXU7W`gSUncn~7|@oQ*S8S>v|M4GH**t-tQvP+)dsOXIXfhT>fCe>Vl$UBvG@
z8**9z^ow=9xfP<RVAMw*eQf@QHNXVG;IZg}@xXO6x|~E)2v@i4B_e{6XEB$%F6*M&
z(|Z#@60lNzc6TANRPAtUHF}rvUvX$$AEU1n!D$AM>s6=SgFMdwKqCMK0?68XcBl0v
zOVm|IM=ahV;5&|_Nv@((c_dCZ=ePf@2WVVC3amtjR0WbLdXjgiI;b1qu0I&OT_<>+
z`I*WNbaNr(JG4Q_))vp?dSUF2@r{+QYb+h`?Hx09x_0s|l1jTvSP_TRRFpKl;vKbd
z+K0X$B3i;S@Q+SEZ41ykIb=PU?tIS|-tE3IOs>O3)oGn;kDt4AyGRM-_u1o*znne1
zfBKn0vqJnv@J0GC=CR356xJ;={kitk*2-woAd2vd3?)NtW}!Ai56U3GoZR-soQ6me
z0@;MS`7Kj{zYc;Fz<z6dPBb&fj8n4eK5jz=8Ahu^Px6YXJMHNpc3`KZsQ^63Lw~fr
z09NbOCj8$VP5O^-b}aNAb%$da6WMDt_Pt(x(Y?`0Y*JE}CxaRN#6%Upn~>q^1U$t>
zFn6+LSKt>A)Mdk3%HesR@}7<;r+Nzm6rcdNAXKv{%59;z_DMwCw-#*+SRq%<ul|+y
zQ-jmY>^e9^&(;HqvL)m81DZf-fuZJKv#vDXT<*?Cy-r2<@&J0+jc~aREbTcf$D^ux
zK1XD%IpbHL28ApF{p@P#ZedGv%_k^;IrQB)UzcQ65mEWne+Nf)V##xwGt<*VbjHVv
zWx{V`Nz!nSSyu)<8}>3sD0o1<a%pujI4mTz?adpu_VA<aCGzo|zB52redV`i4-5cc
z;y8C+<(&CDeGH%F0atGK7mYV1*{7{CD?CS`dIlD$5DAw73KWA$b=8cc5lEUNI1T}B
z2`3k2y;Z8LN{@Cnmy%4Q!^X~1L;h%OUz9QHeEui_fsm4<DjEQT-DtZS(5;MU^y^GB
zsnVB9gUgCz@ZR@tjC}R)&d4J{9OR7BcR$EbP5=Yci@F4$prWUqId<gR>dkg0Aff=!
zn$w7<x1Z+6Tf<)Mw$TWHS_D6m^6Q~@Ijx6>w-lJ#a}Nqwx-f}P{!5dB2?{{{&jGu8
z3iC}eQZW{vNMrTNZFNA!TEO{O-LJ)-D*w*JVjwm|8|zL~7cqc@5)g1=bH@P1Co;?~
zqv3I=v26UEcrTlB@6Sf&9H`PipE=%LU3%Nr5WW?TEB}##C)i`aBwaPHfoe%<-Qqw}
z_R+(9QZJ9`?8Bhtt>K4*2soc-YM+-(r6R$&@*gd>JrDI$Eon}Z%P-Ro>kR%J{5QRc
zDWB`8-FA^^PR5<Lu$1~D<jfDK212;~5^XU3-u>U38~9PM<H&eUIN!0a^bF%VuQ!6q
zf|Wr%7bXeBth@k4A$`<@E&G2&G?5&wJiZ(B!<{W6e`HH7dGrQ+fJTjD!w{imzt@9M
zhL%z;;`csphRwTm)+?t(izJUGjf%X3k4Pv|5o|;on=T24=&I3jbJbc&15W##1Oy+T
zCn)Yce;!i1^AkvMh&U`ALSxihxR5Mox2nD#fn5S#S89oER*$R#1K|278z<~8UAX_{
zuTA+5z(7m6*lAlVhk!@S-CYy?FNVIp9{d)O##v*sR+kf=?e=|};R|pUx+i`UBju*I
zG_2UyhRv|LQRsaDilA6vGRCqqt}>P)%@WL@T>%;_c+oVVQ<J_G{!m9lshRe|h`3J(
zu$%S5VN$6%i{2HKL%{JR2Tb)S%2`ONs9{p;V@%0&r!{(THsx=B`rXCxq??1Adn^fR
z|D$|v#(E25KsEjr1KoUNB?*HG(k>Zk@IrR_*2WsT_?4=?OtO~e=H@n+SsH|O;6r}a
zTn6QWWguo|Uzo}m?#+d(BC-E7Calo+4>(KBg~>RZm<k22msf2f(3pog_-J{$arZ6M
zWR^-rTrFerPOuLih>#GHO!4!r^{i!lThI@Ji}OjM43+rtR%9)8cS#F9%cv3$mVFPV
z#O<~ke6${9VmDg}M?+Ybky-ZeE@P=jaMXJOc(z?Hx3aDFT9f+C@QZ{?owD6jIn7!w
z@`S$<l<g1rSplEfEkIuaF^bU*$~aL()Hvsn*saGvPf5*fll4yw1r``o((*bWuhi|w
zs0N3C8nOm*PY_!upOyN76l=iXfQUHQ02sBgBL!l88};-N@x<ZRb^cfNAY8RCAI~OG
z_Ct+TN#AT+rv^9Q69h*#{u=B4H<&v;=%_uV-Eq-qjx^b;!@48~@R@m^2g}`y9ZJQB
zI%wD$NOm&c$V;1`x;@bBk)#M}UP_M17yB|X(}rAmc~QkhAQpm%>g|<&>WEha6if4o
zm(?fqW?Qa`*z__`rCmnL3?3W7Pbi}`3~e-s;p_u8Ky(r&^-#iaFrZA}U2OZs5{Omt
z3ed7Uai0&0bk!r$q9wE{hC+N+XU~FoB1|*@A7y0|63mpFrr$(n+5XQ@wVY=S@X3VA
zV<xhi!;qDQXF?lCM@?eg3l4<*6Ze}J8hC89C6F?=p-4{H0`{*2vaIkO^}70pbA?Qh
z9WGYkB_RuQS@&J7v;>=(AlC9rj}aAp!5LeLV?-DeJ^Q^3v@bPMxpra3yiJ&zH@Cg=
zjeP0DQ(Ex6_fOT&Uj9)2w|??4YpCH@K^%q6+@e79&*HU%+z@wmIPjk3K_Is{_q~iF
zF!wuFoap|3|8nj2&dvr)Z(-prRiW1++;PG03GMnW;;twT6qfXN*lz7;XUCSvn`PBJ
z@a#$Da!A7A%~<it<WptUFq?&p@_f=gN@~hqw>Iulg)VQKfFP83y;@4-Ey_1918aA%
zs}zPz0;Gi+I*Xs`vL3atXkOxc`ZnNm>Zk53jwtv<6a<fQw{(vQ?Y^H_Di>^}SVx5r
z6us^VJ#!Rm$Gh9@Vedd0M?xb#2|N+^zMjDhs_%=OIi+vL=L{NKf~1;!v1^_3nOhg`
z4;viRM_cDr#P{|G*ZB>V1BE6+7LMjxhZuglc?elEEL_M^e^9#0TJ)~y?A0y(m9~kg
zsj+O03ModYVjWF|8kYS)<Z4}KtXU&LkAz8CwM;#P_;!<iHNx@u2oO4A2+GUPjd%k)
zZ*}e$5J3XlOLteilRvq8r6{K=_r|aF{n?zw;8t=qCMf3mjpUsdI`5zDKc2i{B`nU@
ze@uOgn5=fHDzod1Ix2_^o7OzJdHyd=4Xgc06bpBQ0;vf2b-N4JT7v~*ZJq?3q&i52
z(%Z=aiFBJF3pdw5RJw)BFsHS-UP`u^Mt}_m+1enwx<F%FV4nL?QH6r;pUFK>jnt}n
z{Ad6itE(3$m+s-EXCI8u?}5fmq9Pv=;_c?~b&qHEtlgm8o1%kd2(Yysau?BW0*?_}
za^~$2<%N#+$O5a2_V`><2~zpVIJl*6OT(Y(WOiGecI6TVoZ&FjK0{FrScHl|U1cL&
z$mw<chU4Vv4u{PamWA52?x!7+Q_?emJtCTitX*98$kL|J&vvhV-<n_eR#3E@-O{u*
zLvF+=>pKRYP|>+lTyeFI*g78I2omQpHRk}pKTF7VO?usVPnaZNnEpA@DWJ0*V<=ug
zHy;L<;G9FUE2xU<-q<=69J?sFmELi@KF?aYfrtGaN9!w_@jJz39F}emRM_|~_;azr
zfJhas`t8ka_h`Gz+P5Crt*MAv$<B?AWwL*@hQG|yR1$~WA|m<NEM8}TtkFr+U0M=s
zeO_%H1c{e(VFWGTDl_t|2<9uIj4MQE_p(d{-q%`PY_}!>Lbb&JHED+Wk0xG80Au|N
zl)yt@uv%c_Z>PAR@cU^JT6%kLo=hjV$D%LP)QClZo+?SOvL|)R`3m8vF5o2RMu(_p
z8*do=_efqR^1T1KRYZKRjJ%y_jc=Shk-i@gx@1#L@Um%c?CFAP(?2e_H_Gah&eu`n
zR8l!(7qlm<8j?2uR7mhES&9qRDZu}n@bY24<P0k1{_+wB{64nAuC~5_btMhLz7MKq
zj5j_csEIYiU+%}3vqnQuGF2eTt|0yS>KjquFHkJ$OMNhM|9>w$jwBPV)*^j5low4(
zJ8xJMl5RQ<g<lspvAbXY79@TY%KJh4gA9M*Pmo210f#s|C6;n>DXniK=r=p#Q#NSQ
zZmmuA>vBlBX?#4Wf9FQbz~Sy8PZOy;fE%lR7>NQa3>c$Qx8ApvZawC8Y<%#h@stz5
znUxtvvH*$VCQ?DSB0#dNT*H<Ub+=0us@~5CI4I6LlZw=HHylv!dLC<bL+G`~{ffTF
zi{3Z?4(@x=i{q7agEMF^E`UAU#cM%G{KtYO`jG=hzN(5N)*||wt?RHc-@R0+i4ajt
zjcGs7dNrq-QK!q)ivtsTW#5nn!m@xOmH?FQH-tCMs(IW`Rk|1CZ0U?eAv*iEjtFho
zVhS?=+Hc%ns=?EQ8>DAS!XHSyBHDc(o71#z?p%=6g(r{(1h8y;mh(sJ&6P$(<ob%2
zWBdECgU6wotB9bSl@-?N+!<-r?jiNfo~lMKr7+iwrc$co0^tsaSR1hCY#Q`(RAHjd
zv*07b@!?5BaYN7I?adAPdu%yDP>9lb*#u!+YHdKPkh+)3u*N-bEmxkh4El4#8BKQj
z$USjh>5A^p!K0#9q^obmlYws=O6|`=yeldi(&UHAxxP+kU7zZeaJR@{y?S9hROgmV
zxwNWU9Ww47N?#vnxB^94N5_2_7I)Bo1}|d$Z{uIWfA&7Ql(b14uptuGK3qp2xuDTD
zD5!!fvyUhMHXz;QkLLiZLMYbGOQ{+Tw;8Flo}lv<UtX)@WfF1!HOxDB|DNtDheJB?
z7n%-#*ptUm3II5St4Ws^Z6ez@VY}RQz8(5meW4GSY=$Y-1p?H3uJE_Dvp&x@NXdtX
zu7g@H#KpBvJpc?5f)x0MbEZikR$Dstwl7UF1(ePaq$|W{A>@?*&=7bU5OZ4mefyUD
zM=`6g6KiKX)S$j(+USLU42u)~(?^@D-3$iiyQq2WhF9p`a7vWUg!;%-{E}yemEqU3
z)XP9vUP_~(K9-Y&7NE426>uqiC>RYW=DU_(auk|#1$m9&Tz(^&1-H=wL7xiGAOjOZ
z<ey=xH-$MOYGcn@c##yw`g$u=Zu;=SC*OmfEPlvC+kkV=$si?)4wFE&f%+&EGIY8#
z@W1)slA0~uLg}ukIELvPcOWY!MUkH`;J&xfm5G|RTAtkuY);K52q(<NJn5U)8gvPB
zD`Dlmo)CfE`(X~n&iBFsHvgM*YZuVXyZnI-7mbURqt{q00Br>s2uCX-KYLfmbde*t
z=6})sFB;wK@mA6-Fvcv@?3Mw0jk&Yhdu+#SKPC__Usykm`=xfk^jCSQTxl&jTa)@s
z)=3sPWc`tD3o4@KYPbDcy6bqnkuE>B4I!te&k;03*AS+?|6`=ldfQ!(As&PmV*#P+
zLxI_!G^dK~qU9Y^#xYxJXP3GIrU(DRsNs+F%W1M&?*JeA#VIK^!gVmMu7UE1u%u)~
z?Nu?R=FDdJeLd%?U@QP-T-sah-63HK{4Y5(Rm<V#MWg#rWSdm7@_;^wLWYLimpVK2
zf^dj+*1QLzG(U|_aIs|N?+pd7<N#fS%!NLf7ASDIY52-|rP<U`QT4Q&Z>b^J=+Fgx
zZFs&$lzy+IWM1?9fIiU#K2ZnMb#rk8>bnH*^|A0p_adori1G<7F=DCgg?#W+yZGoV
zbYr-I<Yir(gL*+Q?%GqMh(}z2jH@JqDKSsNg5NVg@dEzT&r3^8zoF(z+Coq3>vxc%
zrQwN{Kv%M-%0j)LJ{AM)?tMUdMX}lMc+7qzGiishnXOrnJx2oEML2brhZw6K%9>7|
z_48%L+rGy>(}aiFYy6LOue439<DAh+nATUJu>4;fQ1$c<Lc-Ydq2jig6a<_wngNd@
zRlC3Li(v24XV?E;u__HTTulZ;V)GziF7dp4_$ht>KOAgE@}UPeN};H=L-c1$)##fk
z2>ly}x6li-J~jrZX!KxjwzG$CQ-MrumWL!<Z{dzA4C(VZ1*QAcl`;5J>;4^b!k3nv
z(be6<qn63G?P60!5M*j<ln^D&##b0H<LKa~@RG`>t1l=Bi}PEq5Mu?`am4x&-4m2^
z^}ItXD=8n~EVvcPq?8&mISrFJlcRl4XnE9uEk#L232kBLAqivi>EttCb;c*w^kmEq
z+wqH=dm^CFvm%4+jEVX;tR|{$*D{5J96_5*AM#`_!}*3AJ7x&>8ec?3L<P^%kah`X
z6zW01<ejg<4_uQxUx~aDPT|A%^vOC|c(cf2tYGrxrkQqD9M$$gpv0#L;^Qi7&m4zI
zrOn{cqq9}igPr4}pkU7jRy~j7N$$9WGNqwujN@Rq=xO4bvkV*VTT73${v!rs^L&kF
zZ1wq5ulF}N6e0)@-L)6q57l(tYA%C$9o)<$$@PP{hq@S<jf2y6zVeGUsiBuAPWQ*S
zp=_RPCZ#IwaEkmfWk-zMHz50k>W4~6@Cg<3b**k@Ci|Z>Bk!-RXolj>Ko{+)rm+!_
zc!YT&g@vy0$33gq_dg*S4^4W*QT3@IdT=cFInTCjcfxAx79fs<4k}Qs0e`%5iUJWK
zC!2ij>8hqZWPdvIyF*3!sbm)rD6mj`emGk}#v>y+-xhbR_fZ$*3Ll@vIjANZFAKIS
z)ss&O7+MmwK}}lBtgYeYP-;n|m<yZCeK?h*R_U)*`6ei3^dR_@<Wp5U8q(Cc;mti1
zwC5R#AmP5^Cus5W&eQuqpftCWIJ5?RUosAe?Lp`Vg{tRwN74U1DW5e@mt*@)Uk^3i
z!c?bCRk0e|<N43bWv1y?c1_pvoGU~qt@Tnz*hqmQG!`Nwp}VRg(W^D&6TeMy+>qWZ
z9tlZ~!ZGOQmjs1E6Ww(=?_n?Vp@6^IF!O}wO2yTM)g{-#CBurZR318&y2Vf|_ISvG
z$t418P!`<c1wF2Ci%fasun<T8$B|bM5xF)cXdSnQWr&+$u!166_F_eYAqmgJvOWy&
zOtLbyU+xBiIxo!7ACT+Gz2D4*C)KW6=-Uu!^2pp(=Y>OW13mFhCA_M>rY%2);Jl7T
z!C$TfLHg@5*j|}i8^C)qa5kuh=M&?hI$RAY65?IEr5|Lw+UKI+<y|Hz=ER1@q2dq?
z-?pnrD}yMV&C}>1t63#6d`OXs5XSbfY%$6Eie8@5sY$N{|AyQCr<oBDl5Q5O2-Sjq
zrjpPgWhnsP-e<>lfD7i?@djZcQvd}7u&GM9=`(s2PM!@01K2Iwl5v*4g0}xB@gGZQ
zukL#0q{y-YGwtp_Q@`VFoIvbka2WLF%zkbUKZ)yF3~p}^*FMnWFfK$vpc$<zT}}HS
zRJZ+C*2Wv&0!t;)pFAj_THE^SchOhO1Ys9(%@?4TBS-*3ewHhzkbc*Ze@9RC-;m#3
zMIml!FG7QYDXm7O)d-2v70r58@y@>LugMoyo1R|;+bB*c5cTqFsnro^TMZRLG>Q_O
zDD%`ldS~5^$uEivmXhGsfO>q*0jhXdVHVC3BhV(>g(4s-2oj6Q$S8Pe195u3mAIvO
z=WB^h3xQw+)k9S@nBIi+-dTIC4h-5$QkM2-zVk}$-I9)|C3$Llxm&V9KKHK_H7)o6
z<`2<&W?z~8-0+5J4Aa^WULl70%UVJbX~DLI5<;+=&ij#9EJn4eQllVxQLOqV%Xp&T
z=kebZ5Et?C)M&HjEt})57egK47se7=s~Rlf1ev3x^UvgV>2n9i$H(jIA=62l0&ezL
z{NAoHk+DUc+qBcT!LNPh08uG7JPz#YBlc=pn56-pj0`?&Dfv1{{B4cxLFrGT9^<1h
zg61aB<Dkj^zX1m$HdYc!rC7P>ucM0=Yokr_L$a@+;(ncNNwaI_*Z)I%HCY$)!d-wM
z4$QpU?V_(go>(iAnrhw4K#bUyF)CcV??Iqom=*$eT4v{^>EZIB)&G3F@15a8N~Dwm
zR2sfh@N##D63T(VsHu)k3C?S+!yE<^*DChK4J->+9&tI3S9nL;SC^e4Tia^Hb-zzz
zR(}7sof(=w(H22W8G7H0F<^tgRuCuXrx^ulJ-;O^(yEm~QRtExJ_E8=fCG#(iab$4
zQcB(@`^U}Pu%K+J^MX$*t)B$!l#o3YONGH+4D^)0IsW^iX$tYVHAnu8NWqV}15oK)
zpaK}?d=%byhQ0Igg~srs17#(fpYz+JO#dxIRAX9|LGx|il-QkHZxJ;g^mz&5+Vz(z
zK4lK+g839yGj%v05;KQeM2*&{X4Ru1-EB&^Me-xuRd$ehy9XOK?U!^^geaH|La9mC
z^eV?J`Jl@uc0Q?&`SHRL=+YRFkLdZ!F9CK&&)+ONIGd9oOZY_3N-9w|-rzbLC7J}k
z1Xwm|64QrrpZYnf2#|%>5i;+&*`oBte~dQRurfXPa~%F$`)exw?LSdLg+KPH?;c`o
z%BiT%(eu!O23CPHE1F8aWzwxrOan)l5lvDvC<{H-)*!NMmP>Ka!T$tO%~Js~VJDRs
zYvFUL1^^&L%EH3EsCsLftvdA2e|Z!n5VdD3nv>FPeK+<|$AceY1gMz4e#BqtkI()y
zx(u`wcQ=$eE<=aV>uMnl%3r0r*v+sG8PX?j4GwhH560cf9bLl?caDcPHHYQ`>(iNF
z^N$Ou<b4H4QJIaM`HI}r$om@SA%Y*Q@q2|8Ge$~eC2mOB^+&?Z+M`Fy^rotrfq_Zt
z@4#%+?T63i<~1}}glAWWlYiT}tTp>tNWuTUk8vy*!!DvQYI1idNZoO0^H;^q!no^}
zE?`|5bW^!8`_nqv_^l1&LK8m$DY#z5gEX%7^7GRTKw{QO;iZG`m#TpopF2+#bh-}R
z?@ibi`*Q~F9O;3>|G57(m@7<lX&;4qTK*g&lb@$zmTg0RLdY>i<$jN1)+&$M<0!Pa
z#?h=mKw2Q0?Y4)d3)@TDLUYg2oOb%k)LSZuEL`{SoNUsvA|11oV<VcFr3l362JpL4
z88w!ByxkxRB$pE{by1m9*OS$|)%HmMmRL@Jmg)5UjJa@v)}2WKB>y|h(F9@K<sZTI
zmu`!xV0;N}4aFNf27wP!5J3lr+k4EJ6A#5Bo8w8x!D+$dlcGz<CqK)++rE0u{d;a5
z3O!QuX>zqQ>yN3hff6NkGM_Mq4~LIeI{s}R&9$dFCe!<acM1qBLvs`1-&9vqlK4n8
zJENvv9K^=|*XOA|yO4WiIIO1B>np^$>2hk8Ab_C2Qt6x`*UB9%n|G%jwhX)0#CAzQ
zv``mZrKOTyJ5^KuSEA@4?g0ffk7GEy?xi(0EsHL#-j6;Uc5oQ%eOM8%0v-RRsSCZo
z{OrC*XrcV1d!U&)C89gJxd<mT{rf!$_Sq*ckr6O6dJ~gsH1Bf`DrCMQ%O^t9<PLmG
zCj_^y3(vJM<h0-g=KOHUOk6M!5Nu%S#CckQO`yltEDC_FnXBrJvzLaG#xn6|fkCu4
zLraO^*usX=5C-?m`kxnZ)L#FK9KXm*Nj4B0Xi6-Ihc#9-k+lo><Z*{Y$(<9?16!uK
z!(bk%r>xkVkAA(q@>0LHQ&BF==%or_lozyhc-fT=z?kz-bR-<In(2T40uMi?tEwtC
z396s@&Xq?iFro6%JiaYn`L&v*<kmSjV*M^pbL@KE`m!(U7dtD*&5PL0T#LkqIFp0~
zA#E|CdDleTnJQ2@dQ4whDx&gCvkS!t^7-~!6Z0*DCd9_())|w@m`pQeauBYYnL{o>
zLwC%sXjN>AJrJR+LcfEm7s3t@*jAf#?!K_3nx0a$;mhgWr|)ZMTO^>kB~-9tZ@2y?
zhGK(4#XQp3ax0G7Ql9eKZSpZ&P-AV)ClYs6K7RBXw{0A?RVk_P1n|$oU0lZ7PgXec
zgMy)D20&e$T)-|bv-ui;pq8XVRp1^}?@9#u$XCs>+2Bo4C~*O}3oPz$P1?8%eE2Mi
zKK^t&aooPY_9Dc=!RN(ql?jzdMz5(xO$hLm5+59lj7t$xI<1R4Iyyc&Iy$5`?Xs?(
z95Yw`jU=stF}fsK-COePC-WW0%<Ipi4v$=GuBXjh6|n=5{W=JSmVfJHlG5Z(Rz;j$
zRmhbqqn9P7=vwz;9<5DW0LIXTc^z?!U96)IH#k*HPi1c<6$`IFOw5BxXhoF1V}XAN
z%G~j#)U}xzv9VLIP+ykm4S^yz<~#US=358v@Sp#dSx>*~NS)+eqOWYPSQ>P1Err<i
zgWka;G@rlnBP29yE?g8;tXkha*>9?H?5Lm}jQcjtX)=d7P<eiLn4E;=w-w$M|7foa
z_7X^)eGel1AM$%Ix%b0e7*~IWuCrB69W>!Z@riy_q{LUJ{5Oz5;eKQ-Jef%w;3)OD
zRbJ5$tnhT^Y=xKKrASELq=ZDy!LK-Yxn#G=V4ticfzPMh_Mb_L_~zNy3G_EtsD16p
zd@k_G#h`@c<tE>ZCtGu$mrI6n+ervF=h4bG6Y~5#Wi^TWX_dL}<A=WRoqq-%LVpr`
zRxYA|TDc>rPUe~N2j|RC$ui-gthwcre>*F#cTROrPjvj3e)4bG<J5gKO#h#wxnmXy
zAvw9}FTwGl>K2xA^A7rNdV2HJ_Z3WCDOa^qqO+@}Mku%Yxmfp>#q+p3tvv@TthtGO
z|8rp?DcbYJ%V@yg%+Hf)>Q@EbV3K+$TTV3~pT(OMCl|qLS(I1XQ*WM>1rwkkc^((s
zQo>soq+%zbw1OwcLd3g7%t`L&6kC|KjXVGd!AZ$V(qi0pcCg!72&#f@=7(|1Y%UBV
zU1qL8JewZK+<tvn7g4ZdZjI6X7J5z)zNJt_Ayd|6*91qWE=nowj<=^(&0}Y09cCM_
zNs9Dp#WYHM@AC3WGU52Me^&lV^Hhr3iPk#^3Y|du#;fJ~Mg&=M)#Ei4ghvresBpq7
z4>L=>Yjw`Jfm_51u2`jI`u#jJ4AQeCia9+XvHTu<RbXTKb%tNZT3@M`iSObWx^QFQ
zJXF^+bmJd7LgS@lN@M@@djI0xG5ay)Md7+Q^6y>sC1_-Wdx+cioHRQsE^6TXzI&L`
z6CrQttY|W>Gkb(JyB}!ixtL7<=thV&^S$BC=KRV5;6P3jZJ2?(SsKu?F2FpOeSNEj
zKTHmcYJ}mlB47xTnwL9Uu(ed|B;+0Jp%oGV|CZ@$&vIlGTr$EvfJPHbD*Ok;c)IvK
z&Nk8M^qG^LUDnAO{_JR*xx35Tq~<s{lsgl?zad2k(lC#uA)fxH^A<oRP;`kRl;~S|
z3UEYRMy#-4H2ZnJmji;UY{%K&<eelLIO?HcE4TCRb%-Kch#_>aVQ21fpplsN1rJxN
zJ=&oyE(uDJea~&ybo4wmRS(^19RY4oskZ?@qb2o+Y&Z2eja}f6Op~X*5_<viucwf=
zG?24cPlp`3F!X~g-Jbw=GW@TyuF?TOTe!~LElCha%MwK-NRtki5Q146Ga=WWS1YYo
zmWB3^xhL_mC)><_i(|`T9f^l@E_rciu|h8X=u7H9{OO}cZmQGS!{hLigB51enj^{c
zoFGaUz}~qlGb5$8V^Y~!pOM_}6$fHdBJ*LpH!@tG5jt4?f7PL~)oV;XM5e47q{Ezs
z!jF~@$9!4U$iL0>p~e0W6?)~9mS7k6{f4d71OKgQSE~S$S0ruH9LzNtHt4wo_7u^J
zCQjr^U0LyO@AHO6q<~&p71Pg;*)|ph28(V%7DUV;G&#K)UOpEg&k47WniT73aK2=>
zF&t%MHbiLdj_C)+TCk+AaqHH`mp&pN(B<9TXm+$swU}qX1FPne(*g=tq1Zk8;Q?JI
z{E#^leni(f9u7TRc0Ar9=!~vx7aJY2E?5Lj`t<%;ZNy?P8sSV)dn=#&mzmuB09k+?
zj<cI{F7C(trXAby&dOK-Z!dkjdBGkBER6ObEM#-XCe6K{^2ya9#2D-U$tPN*qQ3Z4
z?d!V+2f66sfJ@FeIc_Db1J)q~bS~fD3)1~s($q9YI9whMvC!_>+?wGn)s^z0nwI@%
z8TO6d9Md|M>~ZTjH+(j@J!1fEH2u4G^A?1WkZ7k7qLDl|DgLeD(Ln@#TNo*`m_0^l
zuH9B-&b8EjR~t+pYf>Mhvu4~e)fXp(ogXB}mHItRjvAc{4CTjQUAl@VEN+#G$<+o*
z__*Kbyxnq>uai_%I>`<hd*41algt$FlVwlI|B__q_I{H-7mitd&#OD1xw$;%dop)&
zBt|$nISxEIhWOa<bmR3nkxR?W3uFSd9++HT-J4Ba8Gmq+?dq3XDSF2SK<rG&pRmOV
zsxLm?{ZV-4di)>1CDER9=E5wWE{jKi%Te@V`w^}}8b)JuOVN%vO;O4BqaiF1iKZyv
zztD41f-YD?xa3>6VFW1o=wdEa63|pzS_kSw=oR`|0cqfo70o=C1R10t;k7}NoJE4*
z!&JE2{ESX}nab01KCOn{I^kv2lT$0Uv{C!H%-7%ad<x&71S`7Us$Atj>skW>>ZcC|
zy(T6NcJ{*^JHlI!cb4atcb^b`#w>^6tb@ds!wLVs_-*3<_l+5B0R+aHvD>F>Rle&p
zw-KS)MWCnGj)$t|0_`DkgzQ1i2T46V)~(z5h^l49l4p}utO_t0XUaQb*lIz9g1GuW
zY7V?aA{8RZ=9VD?v~dOKvuoQN?0(hZ;YVBaA;(dBR!-x!EE!ouJY4Sr6n9^?k&8!{
z$4wIZb=0D%Ld<zy1k-KCNce_nQPDTdFIZtlxqKKuIJwJgz5YLfzvOu4!Szj2uSvwG
z7R=L-;}e~%CAu6RyOKoL6k|(~d)sZOwua@R?d2d+{`#q^YJ?0v?W+p-b9ix%{YLE<
zje?k~fX?p1)O2wq(rz;YZ=x2w_la6hl(nD9o(n(GMeE%$H00wmsrTf+Quak}Xw>+w
z+5nX==WuqWLx(_kpG>RXwgR0TInEH?cleGSZ;#27_Bx32O?FD#l@{6~(^XB&V~*1q
zie7P}_iB&#){nMVa5LBP)w`%~`n^^*_m1Zdvn4))BQZ*MKgEy_#;R8E*!qQ8JX>M4
zMIXFv%eMZp^l7cxfY3jOVmj>v*5@Qhr^o9-QCVE2y5lZwL3mFj+o2!vuDdJtp(y=Z
zN$#K6j$bz=CsIE1f4==98vU1X&(hyUb9Yu0YjE(p7oDFmns;FPKjf|}<R1Kus{I<7
z!=#%h7eB5xnvIp3Fi=r7DkPfhGPk>;l+#njfk-$D7qmHZGDA37Kjh7=@dgpQ^-YF8
z>WFTxXUNNsX2L$Hr_YZA<=E`f`6qxXY-{Pp6F>z5<djPOZ7XwX*t5jUqjLNIDVIRX
zc1XSK_~x}8I(=_!caHvSMPf1n(E6)km$h%#kJp)NGso0uQK*B@m^!08VE<PR-5N^H
zz^)R(QS7WYu4DK)G^FWj)wz4I6M<u*0HuK+wB%%Z_zc~8H6kiDyA$-<rd>jpTOyYd
zK}eVwa=xt$h4_zE_RbDQByP_KgtoQ6pLF0%{v8@n=Z@_sHdnvb%>|t#L}lg_MPLpt
zDH8&E>p?OLVw<C_)mNN8760l!e&}2A?w&G-el2L&!_!3)CM|<DB$m1@WFRLOg>iC%
za;6F}{z#bUP-~opBI`0ws=l*=qQrhJft*JsL4-U@QeJdY>MgZ%0}17etN2Kbm``d8
zdpjQzMTZX4s~mmh$Cg(nGD-J}ZrZ-8h?k}@YekmOAfj?Ti2aImQGZ<h(sK3u>{qeJ
zk=$io%${!MsC-+~XtcV-4&62Jl#)G8fbn_>A|%pJNioRWu*K7!X|C?l>2rj2wI0%Y
zh&Y_#TCV;v49D@Vgmm%aj=J%GH{sM4&HXZ4BS_6|yBFo$0wfYNa!XC|)kVNSm~Kd6
zFbvE#8yh5fS?KGQuH@#gHnKF|sT(7)uL>F0<tr160JV*#huHqnY;qEa@wm<Yr5~+-
zWb{TEA3r#F&zfHX8A*oc#}b4D{3u9uzWs|}C!b_`D<$4<4K6!d6Qh;8yEb#Ocf68(
z6s+)2QWpS2<!cO5k{9h|kg+dGun}*HJkpl<u-{m6Y2<ZL-0p(Cs~pK5Z)xrl<YEap
zAJJK{hc<uAj=k1$<wvaGO(|z_!>56GZ`Og^CNJNdl&&M5FUV^{$VjC<*8W=#QWNJx
zavAj1D*I!D?zy?ZvsdI+At|#er@`XRH-Ge}l5E%Lne94bitvw9n%VmaY?rX@J>PiU
zX%3Cd?YPBcl;uK#f?xbIxe!h8dZ1@|g>SkPBK1|@54D@3hp&{XCL|PJYVIp;B)2{b
z%^lJjHSsgUy7LhUb2Jua?)b22jWDI3>?<!Xw$CF*BK1<Pmz)APU9&U4$`hw}bfk#X
z4jWL;?6@`N)JWur{uig#6SBJL&oKWc>C6X6i1^Fi<sw(KAH}ha_D#Q&Z=O6V%*Z#9
zcE&6H*T20pzr7MdlX9A^tQ;#FP)|0WNilNb1X%R?Q7#jTATNweW22=RuE7~M*<pi#
z-w#lY)91$3jbhQ=j6%4pl5d+{P|G_6%&<`Y1sa7=#!}dRC}rVWY&e{aZ3aQy62-bL
zGi$7)jNg3WHVcfAx*^9rR7%c(fFo4Sp_I5`26?BGM3NUZu_n*m_a@Kvz!UsUbkd5B
zc39~VgoE9VBi8ppV<Ko^g{Bjq)jFDi9}I4m!*ipi8c`P;(=I@Rdk|adK(K+6b&Ckh
zp>btC<P)(!(9<*;tmHkexlps_a~p#uHZ09Y36?qpj)%-#vm2w&DK6BA9cm}J{y7xZ
z^=w~#deNxpK|QW&_ILxeVRFg>rr#4jYf11fAdr_izX#BIHqBiWp<qSvG41PV``OqW
zDET|MuutSxN<_Eh7vT>KB><_xJAWnnksI9QK1*){1G@9Osm%UmyJcq08hvxMrm+HN
zWqB(f@aNg#LI0SuC8vYD$v2utqyj%wTxj7D4VwB_P(+Qz+gn4_wzbE?oMR3NnRK-z
z*O_EjVu^|im2<!EwjV${c|MZV#m?|dQyJCdlvMCXDZGu-ZTRWt-rO~n;hCYe+cqJQ
z??6P)Dvs4Z9x~Hv`<z;KDuQPDp^8X(E#X%2HT12%{MWy#J?#1Ld8mF2(wzS?APb8Z
zG-CraK+7O_H@FB)T!M{1K7{~+xWz<|*UzDE%)Rt&O7Gtu2{}1B6g%9cxL($4=uNHY
zWn*1lQl!`L>*n3&_s)cXBZ-BK1ie}CuK&fkX77WL*0AmL{-RUoERVh3To}4y^K<34
zT2P8!Xu@K0$w^Q~g~_QTH$7mryV&mGuGrd4$H^Y|xq~Xl=|?F)0ukxn-+`A1E5BYa
z1LDhzp+w+6TL4gz-;FQC+T<>g%h~t_IeUeTZ1Q;N(qYl9P+FV)wRF<nUUQb5%fz+x
zV4?NAz{~DxyLJIf<yIAs!}T`3P2aI!z3pkTWF9#0vIB9I{ASBhc*sy$%#CZI)F`ks
zXbXzsx&}k*4dd-={i8N85%)&~`z}UT3e2>j;d8pL%&g|F-;#DW@r!K>z23b)=}0Z-
za3zr68<l-cv{{TB5N!i$4T$*?^2(<%tb#Ubtpd2ghb#nXCohIil1Ji`!0PFVwBv+M
z6Fb?96xT%5)?>b6fk=V3qqfVY6var;m~ad?St-H*5L7Y6sBbWoUA0R1Vxf7<&I9Ad
zroUP3F8Tp+p`f%@S{&yW1iuF~%1_^j=jjzqi8M&7=*r7A1|NSZkuuXB3s0~dL<vd>
zE$nP;1f2*El~5YIm#KR7)Svvdy2&OMPt%uYSudBKT%!Zp9NFfNO?4fdQ%h0=m|LiL
z_q`PDzw0imDa)s(fHXLkHC0AR1|4mTShT93lR**xT^m~tY<0-mSyY>Rsk0=PsH*@A
zljGLqbAOv)vx!nwwf~=gn?n{t{0r}Con<DM-d;ef!?kmGN7o5!6Z;#-_A;YkqNib{
z4{-$V<u)Oa-M^0?KgxssyP{i+a4oy;mm~p5W7gKi7whjedR0CNoQCp+ygW^@ldXGc
z97Zd-U(ol?Y4rWWgt?UBiU*afXQ6GiH%$AcGu3)(Rz9A)8K^I$=e7xB=UW})CQ1|(
zMb2BhPVj*89#!zyS%1#OUyR=gHo4$i)znT<j&SRJ^yU2{qnC5)$sA&#|JK{X3Ati-
zvtoQAMKQ}NSZwDZeQy8a@vBGkLuXv=aPm#uZrY-}z1BEU`mm~Z#RKt@);I3goMR}u
zBHx8#I9G(egc9Z1J)(Yuc=9JWNe$m<2zG-5_#hIuOkMkPevrS%oF%hqd!QwJ(gQsS
zRA%D?U)86CRXY7>)yH%`uuU(D)F`AYz+o~_XI*t${+ECAiRvO?sc<z#lIZ@mNcYJI
zMBM5u?EMcy1Q@icGB!!i%U>;b_tA%UQ9Pf7=$8IIm4#eqpMw>|g+=|%dj}~tIFeet
zCa6P`f-GSV@k1mk`IIDz^PZ#PSSA5qc;TkBhuOUR?U(t*-pm#iXTt~ppwR1qY*27Q
z@R!S#H^l8Moo;ipViaBjWXijKDc24TDJPI9MX+#sM)WKGm5Iic-;~+*yaPKpng*@<
zyKw=gSOdT9Jkg+hS<*!}O4132X5WgS9Wy=UCL+KT0QpXg&JpI!eUcsSF5$<PwW|}m
z1WV4`DrRB6R}OX25$)dIOxdW!`1=+MuSpi3L<ZW#-u4n7{7YN9po4@<>ZtA5Xqn|Z
z<2R(g0)lMm<HD{9Y|x2`B47EY3lg;PRsT_HK=8|=`m>g|(7Zz?JOZ;IF}%kOxx=>Q
zWoCa>$MMRT<H^Eu)01x(2}Yl<p`wy%MmmL7=;4g7p++X7kR}uB{>k_6l%NV<bB^wM
z3&Fr|v+u9=puW6L8C7ah`(S3^dL@BzLu1s+vmdBAtg*wc*5j0>7D{5PRk$GHGh2yx
zB4fERnI@hmaM1ejBiScS#3YPnZ!fDw!<KCS&a0cV!dy#djRMbT6UsU>StWuBJ?zij
z=T2nU7GRU*TS}M(unP%!3N6dR5cV(^7tSKrvrQMb&3Xn}so9=X9u!-Dvg_pTzvI7z
zBVodOvE-=%am`*W*g3bLZ*g<h3$dSU55)-KwLy~aRil}o7W|&>ZnL!cHhb3Wl<Cgu
z0@{l`P<n}chTDW6ZqXWhX(X_|ZXDdCYveJ7)dyxw`rAlFls;TWM0t3=a+lh>G7}Je
z_qG%8PlaP+e<01iYD{cxZFiAXey#bG#WGy4V^9KLF{(y4z+g6sred;l!M)nT`Hc@e
zC|;QUKMP*y`kz!u!%qo}1a>Ok%Aj-1)-}oY-S5WLHo^n>^txlNjgVG<en41Qm;ldd
zFvY-bP65BE{G3c&x>yZ|yA_sr051<U<QyJ{X`<N4O)FjHx;)bKu#1Pzc;cCaBH^>n
zB0{U86I&}QE6Zb5zP|FzVgMAXn`f-re$B3gJrvl?9lnMQyvIt|yDYHqXq)Hb!DsV&
zDFDEaP`5rQmVBko!Et)J{Wu>nF`<?t>M$XTd}S=F8}aI8foc6*hAg7Ad~sn)4ob_2
zSO9y*IFTdjx4EyL7P`7h{BiQd``>*rLGlE&+?`^GCZ$ieKD4UnMj{_vv+J-=Qv9tT
z3c#5n@Ag(aum%4|j4t}!mNl-%iJh?bY2W51Jd*Gxi!?H%pxMtgBJyK0_p{45?~F}5
zRyE02`DQ32lU>z0fz(Iq?o3Nh>(X<9hJ5A>kR!yub*d%M#j}uQ!aJVJC7iI<bnZ~{
zqVMj^&fyzh8W?{*s5`*Uk5xVTcf4<ZQ2&G#q*6D(nzkOax;mU0V;6HJTP~BB0VO=K
zu2Tfx%0f-~{P;hu9e&R(N*pW^cfD=e2L!PU^Orw!g|0njg2pB>p3N9E1Xr0_!M})>
z?KoKEJnVfm%RqlM{M!F9VbSE8vGj7AL3SFXTK1zV$lrJ<yv?C%jJZ8OzaI2w&Ke63
zXtQ7bEvITJB1*RfeCYN#B3yr+pekN>V?t@F;5$!7gydCNO_;mpLWD=iM58hsD2bN*
z4<sT*2>Z4sw(O|WlYID%SLUu`vmK{<Ww-J(n+gLRu;_8_$uJ}AWc{WAo9-as>BXVN
znzfyP(AKtVyYsoOg8(-Y>ba&;$6|W|V{7^L+M5<&;gw>@bvcw6fZ*&c*|6LGplRh^
zmQAm9@IuT%V^M*e<Mpbe5azW$F0haGKt!qIHDAYeZSCo|T_pk}NJQjM8_7f^vq}#&
z0I{_K<8WS&p_%o1nUnhMobI*Iv{?s>Da04!ex4-P@7Xx$@Je9639rY3OXT*nKZRBx
z=k&hqy}fUf_p;aw!9-Jx7X{N7w-*v7iv<4uDZ^OIp3gJ#LrGybb@ThT&yzut6Sf_a
zf3^Rybd94q`o$q@en#&p*TJLxZ(=^1%Pf>>)5#&>WRq}GmEN)hgh6j4`i^lc$~9^H
z^yjH-SPQL4UtR)g5oFutSc$X$%7FKwC+lk`L+0RE#>Ak^eYP#Oh-f=s%6O<1xxtb+
z<j<hx(32vV^KeF6(<8uHL*5NioRaErAPrJmYYF|DEB72H2hfGWHrP!BR!pr}H5`sP
zUD+f*b}mb1u?uja?vJ$T1#T<ZL9P4@76623mjB=(E?@mht96#WZ+cVH@_fDtL`6}h
zr?dX?<G+)0$H)GJj)9K0{~6%e+}SPV_<X}!cpyLse8`rC9eL7|{{1&l06TZQetbui
z8nTo{O8Kt6V`B%27`3JOGulkppLAX6?hm|rJI#n%9p_dUhbor)yjl7JZXdR2;h7Ip
zqgU7WmAXhN`q-VDRbD*VA3AB#$z6S0(bm@f;2KlD<KM(#<J#e-n)c3ay6rMhh7i=V
zYZOM<OP-={{Vmnd8MVg-v?*E^mf4T%Jf)v7aN?V+XKr<{E916yglFfZYKpauAi189
zQ&t5f6467X<;FHvFWb5T;CyXK`i@37k_4}!d-d~6bX4!SNJ07XcJ91svn!B-H(EBt
zi@1?9m&HhVz_$ue;9T$z89Bn`0j0X`(LS57v{Tt!t^*GY1~;J+G$UOn7elzyhfpko
zp_X(CZt-;hq4`)vszgGZyR4Dl|9%X$#nSo{fd!5C0oYW*fnF0NNxsR8ir~l%$;WY2
zxuNp1yI94YIscBe<QD@XI9>M33haBe%BWx4^Q;tiM~4}QgHdY`jR1OA^zw~Ox4uiR
z$JNxy00JMd+j&R*XV6ks{lb#q-t+(xR<|LaElj?Z8r^5C#ukro`mVT`;TnYf-&f^X
zYuspdTZ7z)R@Wp1O-gwOKL<A<Zweo}io4=MLTO2g=~-Dr?)U`7#k*HHPF^%=cKT~B
zT06Lj5~TvB8Y|<7LL9$K7iJGqvwll-Eo8dx*~4vzZHjNCi>v@u2t)dhTyn6ka&|j5
zlQACh4#k(Hp^B}hofh-*jt(8+&kpx?w09$gXK{(_E*9dm$~{4=biA*A1vH+Oy`XSE
z(Lo|V;Wz0PG2hr^``;MvnZg9%{(()wsYShVTVu0(D73JQs$iFgD&)EkRB@ruRsLD=
zpS#5F-kjri<w}Ul1Q?t9UI|A`gqdDn(o#Pd^8Jq<qSto+DOnmAR_?r`nUt9vD#hoU
zejD}W!_IM#<%V$7P+S#oLkY?U5iEtDD->}X?Un#4sDq*@27JbeTipV~ivNxi#MXA&
z>2tlV#fRX3pG?Hlu9;@pBq)wjwexc|J{>cXM8Ai3l=*!o+-Kt__ZMwfGAz#c`R8X0
zp<O*+gM}@LR|uhrO;*r0H;a?X3gwcDUexgXQbf)#ut8AeNi=P&F<_vfq5ny%3p^<1
z)8;x#U*N6!;rw|~?Ach9{gH;WNAKvQ)14-pH}f`L%ckI={Jj&=U*ccV*z*imBqhn8
z2Tm0X`uAEBvY#O@;1%wzX^|c=l*gztmS<avxzP9o^&XbuGntII2-;n0U+4v)I!!R{
z-l*5QEIXuSOm!p2mGL)2_g}Rp*NE;O`V%Uj5h=_r>H1m;PH~58M=L7_JAq6UcS9Hi
z1Iq{b*lCG%n<eMwO5S$v@^f^xI7rE8D$e@l-bw(tT-&7mv1WAT%brQ;b40_HGq$%r
z3y1@IglfcIO?|y261oQW!YRt|HH}KRR0jV!?V=YV2*`NHN*G@c(q=_}$TQO@?M!Q&
zW-eVAe~i%x0k9N`5)XDDSHERrkup7dfg3<lfGgdcpENHNxh##h!bpL{=yhYNS^HNA
zVy%xPZJDf|!xlH_PncFst!P+YzFl+Is$1zP-We+7!wj&J9PIYXMNYmJ3AvK*&AgUA
z*tM#d<j?S2y|>=RUgdAzEi}%?_o1r{@slO>r#rU)+uJyL5n7ncy0VoW$|60v>WHrG
zPG{_Q<u|l^rD(7%6Wv8AH4hpq695D(Zh+>sUyBMz0wkc>LA$Oq&|zdBVxVH;T_0Z?
zS76|XBXrSLJLxyO{P!zgpW-=#Qc`23`7?%gjIipaN+u1TuT8qM@R?B+NH$~3(~o(R
z{~t})9Z&W5|KIjC;$Gx)WnGtqC^IuIF3HX)TV_T^Mz(v&xJFju8X1*YX10tFa<7C?
zX2LbDJ>wd``~Bnhc-;T*I`^K}>pAO}x8m`Fl*nM#-w({Irk9_$e3bQTW@VIq3`nrm
zm%Oc<dh|X(IlpC_v#J}8`b$V);6Y{pQllJVjD}F&lqV50o234!8{CW=_-twy)(76_
zt)A5KWH2`Dy8?x}N5qNh#aW*J{!B{K<f8{}Hg~O~?dEOm?Ch$ubm4*yKi7TtR$=Rk
zLdL1iTLKZ6M^7uPwo{>L1Ty@{3mOfb6k>8?7A8neI(4Ci-Dyyi&V;%K)q=Gj=8_{r
z{qhMZ4EVDd>7u=fc6!jtx8zrXNj|-bvzE<|O3j?_!n%jpC<W-t?=#*yiWK(~GJZh$
zQT(}+L>fa|T#Q9?#a;pEVKpz$&1@_43Z0nf6>=xt`S~0A+wFEC|GiHFIi{eaMH@ZC
zO*|d5e@EcI3Rc#lSfwR_93`MO6-W->I$r5NCVvC7OuF4%p}$0}Mh~Y@Cx^t_<a+TV
z-<!<O1*!k;1SJK}vxpw4yeUerEot*=-=EMV4;+WKah}e!ED1mL<vep6HB9RvdwDLY
z*l(5bZP{k`-?s&q)OXladrsT-^yN4v8@FQ0j&0vfQv0I<1NWExR3JOPEe8`Bt#$Pi
zqx!iI=hHC4`#l@GyASHM{@tW9!HbDjmQTZ}t%wnw{cWMNFrS8qHq`<aX^+Z)@yf$s
zCO8t*6ma=(Va+eO+1kviCB~E<Ff%i~+x@YF8|V#ZW@&r+A5wxPu^xnE=T|d`=3esS
zA-$p}?~4}UdX<NsZS6B+f1jZ?zJ8%Z639Mlmo5F`2Bz&vMi9$!?F4oVugO(s-mB6F
zu;-)(wLs($2s`5L?r!y^;)+z$q=)Ahn0idS1Z(L@o;3e-BY(or<ouo?{x={qq$cYF
z)3X!hqZP?}cU7XHX?{bpx`|eV%v9?#c^5<yDoG*B^n>eC&WAtwNHI}rr?Iz$Ut1va
zb8WCiF|J_zr|Ca%m5aLI!{xWun#QIIwC;iui864>{o|D50K7jPy7JJjBdq1*b%sEA
z)j~&w6tbSCu<a8p`+BmoiC}Z@8JnrOqicqzMigmFwIk&?<#1*Z+RW6G_=*#BAJg7|
zB4c37A>o%77hx&|DWS19+0JGZFJJZia(<p9IT;e%!k|NuW3@f-8BZyJS-*0!686zH
zm!{>#KuZ5p)P{ZlF_L|&meL5SSstK=d>&Zf7+X2T@HM`dRm&hw@`<7ttNP$TL&?>Q
zGJF!oNFqob0poX~7GCYtpf>$#5cIVgwSLUVVwIToZ>;yKi-))z8;9v>uPgXLiH7V~
zm?N8+yYow7;d`3=9{rE6r#%d`IDB2r$w`SvioV6Kc*H^>jdSt8#PKV>b{xD$or_P1
z_jBT0HI{bHcF)hJy-{JL&FKhKpRXQhSkjX>6rn}C#@TJ^i3|UDUkFU8H`6MH`Vb)m
zV0;okzEbDugCKG{K6l_3F(4skxm*+M4-G&m(ftMoV7_Il|7tR~?;DE8kMcpAwbeFS
zqK|x;c{mX8Rd9^aOE%GKm68v{5-~YwIf4*eUC3H`ME#KLNbC6guWn_ZlW;UCevo~<
zA}wB5aZ@w3YeKQlexb5y&>fX5J9&O9%=#92*u;lb;_-h#s4c^NXAl0jR<O}LTJ5(l
z?}G<sj~(T#9GM?R_SPLX-|J&2bJG3_O_eZH7+c{6ENttXb&<2-^C`6iED<Zc;MD)o
z+3(uuzrAIBQ&wSEioyhR@z+Wofw?!{iN0Sd$;m;-y`49&zBV81g<=#=tMUGi)jvfx
z*%hxxAMiH=-jGzw;WA{EO%s_~b|_S7Rw^#NJSdUtM+r!4L|duH*y=tKRE{?S#Ei??
z^ahkAr63ls%+&?aW2*Gu!a}@ijBlI8gCnnBH&bd+s(nl$V-^Ed?xJMXhCZwmf1XrJ
zIFNO$6s@~QD0yi{iE<Q6-0^)Ck~ex4d|e$*lS*ZtXDKN_pIFCXg+5gDe0g$qPKFwn
z@>aiWVV|6Y0Z!SFb;N@*OF7a<9@0%4IYLKDEnh!zI#IZ3Zo07uvs~K5T8>Ojm<(4D
zqX6Qe-}-m>RGGET_KF)G{)_rP7!!pM6#Uem)5s(pUAG5A(W`^0sXybxc~Z?c{^ILF
zu%F06u_hb&8Y5%+C(wo0gYTK5KaSgIGWrf_gBLvbbuH?~9A7H*q4j?M8+O^R0K`PG
zxq)j_ra83lv#&nKVpt_b644OO)PI~`KBR|=z@AI$l`n-ofX{^R`SJE+3Gvv5fR(WW
zd)$zN#laFPnw#`*({4V3R87W;FT(Yk0`{cwzI=^gG||zPlwhiTseT1NW6`w6=W7?`
z4__|IGyW$~5<}Lyd_I^nt;%BLpR#VyDA-3Z<x&acXq(Wu;2dsI6+N9y#w>K;v-Hal
zb)v8h(t{9a=6A<@dI-msV^!9)hoc)_3Ef7rOh#cjD~zpZIjR!n`+t$ShUbo5{slM0
zwr`KF?)Q`8U~Y%-GN~+kN%)7ShTtD}`<}tn#o>NT)fWQP=sfb2G?awVSEgE#8UOx%
zyxWnW3A6E#`j7cdDyPbe5t`zmx|hxsJ}yDOR9=|zIB;<EE$#-1`3=Kg8#zeIIDWhe
znMM_RXC;IBmB~j`e#6{84ak2?FYPPVPko*J{S+kbgu(94@i~3;TQa!h(Rr{lbDZ|2
zl0MOO3@)Pi7b<`!LlU-kJag#LW%?FhXTGJWeo%$0#{5<e&pvbz5Usllhl6EZ(4QKL
zg)dW=oy*>b@K+8xF-#d+h374P@oCE8K*bg$d~Tz9IPG;9D9tX!Oz}MvgO;qI?;;F1
zgCj9y!dDby{?NdR7Ms>G^P8h4L)px`LhxBv3to42+|)~(@&~4RjQ5xf{qu38i*Wv1
z+1sM0&yIiAZHHT-q6rm0A(pDBHvM??s>_^^o!i)WC}_UMA@7b4b1&u$no9nTqATh9
zgr=8riMW&m7D9R&?@u=exK(6luYq}0__-l4hI&x^l8>k}r^VXfShu+E!y&~wLkIS&
z?l#zLkx^+133>{Un(zzqA&xFj9gA9Wxq4eq4W2-)Wz2!b4=9xOE9{i(h%70-RFhms
z#;|)3(>&gAZtVpLJnWrSNwh`9Sl$f!e*e=jvLay`$Qw&`q)u<_tMF8sb<Bv5#qjG@
z$Vu}po@`E&wmUC(&w^YKmN}ZyxxXbJ#v6f_(6Ot_s-K*Pt=FcM64MJqdC6zFY2Ng`
z7<#J%zXH9ZW!CYNL%7GREpB$(ap*Nfr<a<dl<O1WL1vMd-YHZ7mW^71D#7X`U{9xG
zFe~yvI?!`BY)FB3_z9LxM+=Mw%M!@&S9P_wUOgWv@0~%v<M)%zT+{i~-#I;%Mq{eM
zi<c~_^ezVKSDjD&PJT=%jo_9aj?t88EIqxA-O~P%)95)?SzC8SwGZlud|w0i5rp|-
z+Z5oG_Gx(^UTQ{bpyZ_n1^vuok|d552;A6~2Cbkvo3!6rmhzLmWdnWafxLW<#)rSL
ztl$FDm;<CJcZj^d>FB)WJ;A1=oCk6H(aIo2WAudiQx*|gP7X!Kd@M`zX@`85XZJkB
zt$~c<27R7%X)8gBD#_RrsBS}pe(LpM+KmdIzY%MZ5<?ZJ385GD=1UDXPB#j)kQ@^w
zqB<D+j=qZ179FL`7KKMr542AA8cK$LTY9&@S<4<P#73LP4S@WteD1V|PU#p3dD*;l
zoewOmd%a{WJu|`m*Z?D{OrXv^Jk3?9q_$5=QtnF@j5POrGcekC^U>U`xD-d#OghF{
z){&ImYVTuj)co->`~7V8f}MraIq6#A33QiU6EWa3_c2KT*8L(v3jXgEV|bsG(!@-E
z(D^bBH+g%qDT_J}M+kn(K#f-&dwRHi=AFj`V3+g1A74d0Lb4j)lS^qmSYY~QY-MX%
zs>Hz$COaG|t?%(CM8%XT4>tnN1_%g8<>-zHF2M(d#vllO7R&0<tj`3fYeWwZKhZ?T
z8@<o-+LEnNhchUu%Cf#(*s3ZDYD5>Yx!k{xpSlLm2_?8&vahiVIgd=0nth}!%$d_U
z9`E3P)@fbbupFVtM%njSb!!oNOSNzD<ZSrzUkX`6@hgQhaxE<zz4L^Av!<mx?!>Rk
zzJnEoRRErT4^Muc9X^fuC3duk^>9Y&mk#em<L!G~1Iz?#YQi^0IngWJsm>D=R?>;8
z*R7uyi;W3G=*j-LjYVx>e9Lx_itGnzF|gfJ?5kBRNG%*)|GMSUx8dT*?LrJzW>iJ!
zxkM<Z;Eghca?;FvPXFTvr|wL4UM_w-cS1FevI8?F)NJx>v!5E@f1Rs}<z%(>QZX@5
zqBLbQq370l`xfOw1d1H=K}O<%a3<jFn|TVGFqL30=G-?PgZ2T`0rJAi1}@TefeRJB
z9ufc4Y?#(UPzUTSJu2-x@=OIBnc7rY=r*!&<e)1yJ4=tZS;XRoShM@z1@+lbYc0d5
z^P{_JO_g!ECo%QpH3s~!t}X#|QxFlWIaffEx=k5msxJ-^c~^7G4-n@1OqVDz9*_%B
z)BE~SRS7P}U}z`3vj4$HZhx!>-S9?0M-V+)Gxn}Nw(g(|goyQ#_oH|Pt{Ay-cl{4i
zZ<KoR{U^<w7H4_NAVu>NF*rYps^UuhnMub?s(fVTy@niDV$5rr{QLIFB<)@wtCa8v
zk|Ro4+?Zj-7AH7yJK6MMp&(Q5fKytBXH(XPxHg4_;o>i-L0kG|HoL}T*%lmbz9a0M
zRGqTYF}EEd>UuwwBB%pA+@@5dvkK#v1RgEzw*Z{09|dus{m8ZGAFoll!qPX;!Idv$
zXK0S8{XkD(5V!W@%16vsVC5APloZ8}74UHXI$rS?;JSxSVlRa%HhAR$g)ZIV*s)4Y
zuZQ$s2mK0$2n96uWHSvp0(@rhNWYf}lG+8I3U$_nF8KZi1IpNJ3i?6`I)DP2Z_zEN
zJ9ewr@?X&OaS|wQ2X)q5t=cimm$CpMQxKQC1(&z9se1=$8Tq4uLK~0m=iDjJ^aid5
zizyM@<-SsX`ql5(a!=^f$7M(Sr(O7!LD6Jw3_18LR$GrF0uKRwTLr^)FyLqSNLZzf
zb|YJ=?;wqKck$aPDC+yuL4gH=qbgji=k7#af5e71Hr12{#l$%-X_b$2WMb`@^}-uB
zCn!8>6f=r2alTM!-$Pxfy(c$j6fDa3MPMsi@B$b?sxCkA>fL{JWiy$tBazW~UdI%g
zAudRKy^-sb6;VG|n~j4y&!ah%5+fy<P{Sde`)js%lpd0A@0HE3AL@}s^i)ljCHS(T
zsh@_km5rf3!XF4?L-mnjaH%d2xd*xo?==*o-)U4S6daQs9>5<Wj<IqBZzbuf<0ANd
zs`-EeL?6`2hnY9y0`&a;G}_D?6jatUGQLARga>RnQj#Gzs97@xkI<KOXb~`RNj^`V
zKne==<q@jJRdu7$rR+g^0Q>6+Qpa}VLnm{nvy&n#TY-GE`wMO1GJj-YO{0(~kc#}q
zG^o8&w2HUV6mo(4ENZRW>W>s>(LL5$5|s)Tf~Weh3YPTHk2%#j#2Yd?q#1}MVd!C%
zPYRzTYq~t!*e8*~F0uW1KqY2N`O!)AR2DM@6$i><_3|?2iTdc0Z`;{<0cxI?1vZn?
zqjl;#puBquT%XIcYTske5McDFR6c@r4~m>v+S8@U@-N{OAXm!5b*Tjf)r<jJslkcq
z!{_bRa-QqKcW`4Ok%iT-)3)Y9S0=FZ2&tYP`gAEn`N41Q0Hsly-j3PVs5F4LmM^Y_
zA2o9Ld~Cv9-VRk<QA_v7vW2xsq5{E<hN!_uGcfO)t{CO!)lwdtZ;j{NjAlwJ>`fRD
z#5@tEtI^*lIW~V)+_>01J0cygIHn*2Z@68at#rQb1W8Rwb~90!zcI7iu7(UtKDYDs
z_NM=Lago{=ZUT#P0_nw}cbq#ou4lZyjuH@kC#a5%UwI&gJlModP7`lJ9@_I$7IL1b
zt6xo<vyg#oACoL`AR=Xs0=y+V?Tsh+a7#uyH?8C8cLl;VvU>Et*ThZ5Y}_MD$%BFS
z)Ns}Zz(#I-SwfkP);(%-Vl^?zISOyyIDt*OFP4Rq-A6*RAC;&Hd<NgDo=kK_y~E`7
z%o6H|%cE@-N0r{~c8w7bh_<=MP=7!4o(ow87t-mKbqk)h%vF?>=f~6U$KCc@a9}bk
zE23q+@ot~;;3Q5+aIoT|)wI_nmt^iq2VbuvK87Y+%gKPY#0k`=4N6%v<a6XZ{vV9e
zmh#*DVABlqVK2|%FU_k7?guEt6^2*AAz}dZ@^m+dd%kmizT>lXRlE0<2WH4du4p$R
zS92Rhe%$LE-Y1l`WOWHYSc^K#6lr>=$oWIG1ybe{lz_v3OeQlHb9#c>*W0yP9|Ut-
zey}Z66ulazu4xhUJu+1^8a94Y6C>6uHW0QaCSdRMJ|>$qDAS85`G}f$2?~tU8$|Ly
zW-3IAu-x4b)<QE({X%jeT`Nm#Z}asH$KBo9-IZLS){bRWt$V1WjaGmE6BW9Yzzv~<
zf<C@&ZzJ2TN`DR(MCMS0wHy*RPfwSZms_0)lFU|qZBOkkj+|AKNhv89o`=RGZoptv
z$b~w6sQ?`ny2rj!8O8kQYl-jg3g<7;4r8$cUlCPuWlCT{0soQ8LtV1gp(LwM9}PTo
zh@4NPH}>9@q4ps>rhQ-*OD*juo%rW(jgY!Sh-qmy)x;kG>61^Hdg~nj^}Xeg`p<Ak
z8mU=J;q<jZz{HGSYGMzRSfhRFf07FRiG6lUA`YN_<fd-D;(&rbSi`X1ciH(TPZk3<
zd82+EI2MrAqZ=eIT-qv=D_GY9Z=JRk%_I$TP!psD+$+4IWb*q{yj976<J0f`;Pz6q
zR4`g82?mCEu)qrmxwOiLR~l}(nTOD&sgyJb$$F}_f*g%xzu?GA+rr=A=~0<&p<dzs
zCb44Rpl0P83X5k?n^Tg$PHV1dwtnCO`P0a1Tp#jS+enr*esN|=0)FqGb-n^;BF8Yw
zMky#3QUY=YCAt&a4;#qHW50a--EO9h*nRVUOZaHxX4XVzc4f=}{^KCh8iRG9`R|9~
z_tTkA&ZS<8y-dR`Pb*Kt=7xmcOaDDTIT%<bo8hBTqXJ{n&2{A*UX3=lj7BwCssAM`
zD)eD-V&w2>Ot)i_!clE<uyV@EV$fV5*Uh19Xc`;QpGlL^wW3>0bfrps?5XDa(AfS9
z<@^*f<uW<F?7Qap4~N_IT@!sHA|-B6`EF0r%PSte3^P@Duz+d!aEQMqly%H5y&~zc
zI?jL!({frEk1`E3P0JwU6@(i;k(9mqB!<QUTzAbUu`v%{a3fHRPP|sk&#$Qg&NK>%
z0wF<0*-+N8^wd6if^{u71w<J9E_3CYleVl=4C$Jo1_wM+P#XJ+HzjE1{A5S+`^e}?
z)9nPqS>J+W&5!<3zaJyLJ|8lcXma;M396_7u1oIrJ-*J*pFf8OpPXKf9JvgobKlR*
z_j1|deU>%z;lqb-oEMj!8VN}<abq+x3^cH;HqGX??6q#XL&j9^R{~MsFO0@g%Uaef
zva;iT9s4dVZC`hYxaBpnP<M7ZcYkfDP0lfCE{A1^Z})J8nGQLyz71dy05hUw#0s_X
zcZE&z`-=0ekI^sa#UD2}R%UAekre#wk-3WPE$uBW3d5zpY1qSm4gB_H-jhn3OP{`K
zukrfVZ=C6im}GWJ!!LBS;M8}YVBp-}(5ZfTvZE8;U91hk$A@&wjvj1&FBUWph|YG4
z7LY{Po~RGSi88Ld8iJF98$R6uq0uO7amu2?<u@GAa-=rj#a`XryDE3@Dw7{CtEMON
z3+%e7M5f=GdY3YrBY3bj5AI^XsZo>=Mt&)`$-|SJjhD+V*;(cJLQJ$obgFq-538Qh
z`wX~AB~I$<g3r7hwYmu~jtkIF+Y>32(seo8c$Dwj@c$Hi>o(UBYJozyTMTgZ3hJ%-
zrF|oGIxeTiDFV!xTX8xdezyhu&FGO$k9BIMVsSpD!05sT$Grrb)L=!v&j~f)ti6@-
z6G_Wxdg+HbSz|K8$f#fVr!ZQ$^tffV8`Jt!t0?%^{@Qa3{8@-isCTqA&Z)$46&efd
zE00^w4E0EX#x(*^fb{)x{Ldb(?dgtf&lCw@duzJGZgR6E#q!2u!N-pn=~#~Kd&m_y
z?t|9je?n%DD;qpf5n&y+lb#;#)$zXo$E3!lj!n{p7MEfkr7`6yHS^9?j+S|98}NB9
zO8U{DS-6Z@0AO8BYHeJWDKOY?d@1MMpWAdC7^o|f2h&Bxp)oyq>l`3^pyav`2Q3iy
z^Xiu${#Qa_aOof{C*}Jd<?Dv)$@lE&utN<NA#z<WB=uWtM`^z#X}iR$o1C4${>wXP
zbWDli4Nsa2Wf}9ip{@oOiXZqYN5RLy$e0LvfsoK1zFl);IWvt4_Ss{b#|U%tu_u?O
zmG!Mw+saK8ozx$e@V^hFSQ5G=x03OX;ua!u)=ixi^}~~b-iIftGMeZ`lY4@~np;q9
zGwqRzhv@KGGV<Sq_#C=Vm?bq$H*OLYE#FN_Gb|qSUaanNrmsCeYvDZFnsz5Slv;}N
z^Z#1SH)zZlgtNcB|L=LWYDXK&FEmu;wy7@NbfpJM7$<qj=m-BOhMmu~FA2^83$z4l
zNyWU#vEW<%BDmk1COtIK7|z$8WgIKQs`JK|hrRkk3PPJMZs(b%tLJ<}Zd>Ys;N5Z;
zsfv3NkITo9j6i+kW+>f$cM>La+g;@+J(O}b3fxqASGnVi+)A0Xf(x-Z7HF()nt;5B
zSvN&t`BS-H7}Z^^F`c7mb?lPNk&%<;sUmgQRg>@dtl;dDvCny#iiXA7a7p@S70doP
zhAV6Zobqulb2-U#Rg>coDGNDnbXp6U2%y!H;Sm^SFgtYZDRDFKdoI-yjRBrAjdys4
zioe1vMGVizF~86FZE+{Ql0{vtK@XSCc7pQs=G9IrOBO$VEF7S-^UZnMqHID{oLl*E
zyR97sw^$2nYC5SXI2(@6_NW9@wD_Yut*E+fAa5~Z`{^~#X^MT%?5s26v+;+mHdfP&
zV~ySf_hNwZSsGkBna(CE{U|&*II^q!d#~nf{5N%abAaNlsoF5Ni-%{Y*C|f56ZJGp
zfBN6i2@!{L_O4{X<-cMM`13=>zs!pL7Bz<4n_!%Ppiulf2T&pc4l=F2&yNJURJ5~v
zLUfHZG>znLwN*_vR<$J`%QcyZl;wo#a^0cih8RCt+iGcj;sXD6wUFnp8orOQTQ-k1
zz)}|xEy{F{ACdHamQ^Q>#;_`u0yQSH_T!Eg6U=@W!7wKI6o}PAs0V{NEp)O=vdK9X
zQ{#Ogo;88{ns%c+*zyd;*YVhjjAFg%xt!lS60&snXkS2=qWkn>gh6t$EIv+&zwX#o
zN7z(0NetbRmyV;P@ca5&JR<Y#fO7eICc}Jtn^FM>*W&W>*RG~U)a0wPa1_tE-R0hb
z|8ZmfcHoOr&*`JxGym781I8*5r~fX_m2?Y6=HA^5;WePop#pXuc~YwZ^3<USzFMzE
z-=br0Z=LI3CVB*zC`ycB2<(cnH(_jZEesW!UqT73u=ye;mix5n^GD84_y9@}7sns3
zg_53}w?WY?|FPa;H)j7m?uTOe{Vwigj!S9?2k+Ln3J`+TAzhzgT@Z@&X9s$U^|{Yg
zp>thMUhV`)R9>%IU`>@vPIuU*1_qH}f-O!oypdM(>-1tLAUwjhAiv5mC2OwF^_?7#
zjhXDWrUG{GhcRGe7#xapZ%s1}RZPxUD=Nu;xC$n=78kdCf0dDLiwZ+O?hO_V3?$e5
zHOU1w;eu7>#hRKJi?Xly>Hn;Vp`(zJ{uKMt^I)zq+-%%m=tG<)vP(w)`kQ@w7hW$w
zP2rSPkd~!sp@PHSf<mNQk7D(xBop3E*-L^}+=087<5|&_rHgWyY-h@{%FW+Xs7Z}W
zo?&Ign<to`f69~R#IDu4PkH?XY0|+Ff}lG1!gY9pM6@O~1i!D>Yq2^sV<Yf*)rTBP
z+C+F!m8HN;#Uqq!{JwDJ))OD|YO<^)B0T6b8A~ldoqC4<Ry$wYt!=Ow|H-4kl@-QJ
zC4&kLY;Cb26a$g5EM{FWt(5v0bMJP8cY0>X<n3Q-f&qR_KGf)>C}6v)A=8o|_2g&9
zsan3elNOS(MzV^y?ovWqGJxr((W%%Tzbb)FGsi$VC55^1mEJ>rh~sXzH`Ia$(Sc90
z7$TXPnx0_)Q+i2s4!3B^^7PE6{DWq$LTS*<&&<NF{*fOp?fT}>>h<d=qTO|Vmfg-?
zHQ&uP4V)RaMCpq}6@Jbxac&LuJKpM1vJVLf0)bre)(f<%k-aQ@^M<9N!Z>p!r1X{P
zzWuS{@B3}X+uK{)%75!@AE6$a)yOjyqa*3A2_y%;4jIvg{AwiPu9CU{wPduopf$uH
zTa@jg$VzaF()?Wh@V*RHmk&=>Lhx)Nl9P25BKo2~IBS%JN=Wgylv2>vbJhR=2=xw~
zo?f+%28(}wnZL;ertt0PDQc}GhP13h!IG2oJa3HevJ3qDV9~VMs;X*hn<YgyFrlW~
zrrStu`<p{27h%VQvxW0nnFk_*n&=<hLuc6GsKCP!Qe)M0V-A;OmLh*kYq=E-%mMzR
zC>U~w3zF$3`mKh?u|D&~r1gDAY6+b3a6_igd}C#KD@~#v4lh`7$A9U#%ITEOq_Vg{
zeD*bJkg?VPw}3Sn8yVux-meP&hgw8p(|Mi1-&>IUMw<YrYu_}U0XC_q*$aiHzB>Jm
z7kj$y$OL&7D7RX9k8d)dVE|yG1#^^-QByRKpPsqO{SSY1Oba-E$Mt!8Bj3fv{;(=4
zJi@E<V(-U<0rT{8S!u+Me3CYtx?lJz3`o<f!I(bgNa_Nd?tfR3ioA{r7L`}_BuK`V
zs9%9-rdmkMULOEU#GZaV?Wmk2HqLzx4jhhQGVQ^tQTBwduD;6+%|5LcTgl~e`95+B
zP7xg}*rN=(OLzO-gN1T-m)!~c-P!B(|KWtX`vz7bvG~dl4%_jaEG(aptM4mgC4|L3
zmd718#E4L2z>(}A-YPE?Fo1N)OiXFXl#(<pz-IpluTS=hldw8B`o<b%q%izx@uyaB
z?Z{EbrNIq);CN%zPWFz4ut2T?65WM(+l?Dq>nUZCYg%daoDPs3H(t8fI<ratyM6PZ
z>3ADWbZOr5#8OECO*RDuRb&I}IdT5tbYU?L7joXfvdM$r{edF&d}zKX`&aF;jG$&&
zUB8`mUeFnKr8C&?k=3InwWWp(=Zex`+mhv>+1V3+X7mKf$OI&)<uiTH<7MiWx7Sq#
zd=oJftUxIy$V9=+PJxM9szPfI7&B1+lAw}5QlK9w)5NxDBY`52z^9fgCW$`#URQwB
zomTQ7W_7$tKoMeN8dwnQA`J%Bp>LB^YJ65JQzqzZjRT^0elWM^GPm`vTFqCHp3l~K
zR&;NPTxR{W>yy9<i)mmIV#>hRZydN@+3_j*BoPV(r_vqRYxEn*K9YC!`z@)avki*^
z)>l>(!?QC_2p0!OJ>)Cn^rD9}ol-D&b~xK_9NgTF`Z%e7+}?EO`2r;`MUfZbVf1^Q
z%2xxXrRlzVPTHHK4o{Ei>A9&!x!-~N%`NTv7p6o3g~Z~F8AP7J4ftwhIWtgZa(8}4
zZ!8I+@mgs)|BOI(4yz&UG6P-{mzND0v6Ql47f8-ZuvJr|4ONOYeF~UEL@Gqr4iCLc
z_d6U_4|2>bzt#Ji9{$*ZV41iMefM%iIV|Ui$i^M;;|h~BRswN(N^3s!38YFrngVG`
zSkCY&53bT<-)hI3G4Ny8y*tBCjt-Yg`B3Mb7sHnqxZ5eDt?7n10k)U+fEvb~)>S5n
z9wGpMex&m&v(<KO-|98i5qEKX`m!5alaXc1PDQW1kk&h90j1Y2%fP?;BBO10+sY$+
z`ksyGBcXUVIFnYSzv^cDe*p*;aSH0+T#|mUvXl;4K)9id7qUZ9R03ZvG>@T%jy{vm
zs%EFMLG8SvY*~R&1I+f2S3ic%H|;JDZ-;xc*pXl8S+OU{EyqhGqEJ=$9!rf|(p=+~
z6_EoY4Efn+XWKa38J?TDe6cx8j>4Ef3gD{rcnZ+2m(PULJU7ZAU7~iH(M7w<3z`hz
zH{Cp67?C2%`m#M5Q>z{7M{MzRVn}u{M-?q<wL}NaT;}-GSUPZE&yoVotIZD*pwA6n
zzyY7rq=H$~oNn-6bx`l^&86;z-hK9~Wt1Y?SWuZ?rel)}-l7CdT<eXIloi5c_~sbD
zH+JgU_{MYO`T6sEcad)U4abdHvrp}m2z3yHi&0{i1jfEZW9Vt|Lu83tdXAe&#-ugs
z5Hi4v(SvIavBfCKSWenA1kZk<5_zMb9ZcumKH}xfGn}=#d$M=%?`(H*Gvr)Ge>qJ0
z9XIiS$x=AIP}<c-xa?=nJpB0AaY>Jve||-7=~BsIxcAU(ttrygGf-(u6z^Q&SNWR0
zM>Y&#mA>U?c4)kQ;-79z1%$eUDzKs|M5W_*Rztnp*8jcAtA*5(Sz?L*%E!`+(pq{j
zy_6L0|6EBKc-hS0YzljC((-9~0SRh3C@2-Lmz)OrR+Q~22P$$pc=UKaR0&>+rx-<0
zXa89Mbvep#-E5wnpB3M)`<V@R{+m1PX~Ll{c5t1SO?(j-C)0~#6KyG5Ae1@MSBowc
zVAuf+GoLfqpuyTyKh^I_rBKkD#{CBPKo17>#Pk<i55qm)1@iC!I#+Myi7LS#)4~mB
zEFxCukbEW2MM7VT4yX+K`bp2u7ZBF;Ok-6f{n7+gpDo$G0?iz?AC5qZS`X)DlfaxJ
z%9s#tR<B-kw>1(5FusSQc;n9`k}2MsrH}4iH>Qs7W;v9gLDPl!7@&8$;_1|bB#hr`
z?xj&VDA9`dCXHEduhA+HJX*GhcVgq%Oq%JWp7H01J>DkaiW@Q~-&%hN4ndv$EBhm>
ze6ucu4s!Jt?qFdoj_mMc*u8Dn5pi-B(6(jolan<a+`Y|z74=^`ij}vTWHdiM<#UhX
zuT$t2DJU!`KQlh@?9ltt_wr)1vFeduA}#M@HK2Md*Ae}yQ;-GG3)Eq7Q+x%J<$smc
z?6nf$#)50`n#0DI3%_4UvBB$N?k(NBDmewttXHF=lce9ZhbMRW_`s->B&6dGn0Ue0
zK(Jm*2FVrnd@L5#oB_wXJm)s!t(xi|P`O6Ghr8QKO9{CrUvbrpKTW=u9*pQM{<&1g
zud}_lT&#MxD!eIaxwh(#%=P!t*cXZzgl||wUim!}Vr8x(B+E^#^h9RTM1MPL8j4^Y
z$WF|hyZ#bN=y7n3m}aZ?MN)Uw`WFKCcq()4&Ua(-^KDW21>f5ng0nC7CXbtr8GYF8
zd*id$P5tJs7fVEE(5HR+#xrGg5X-(I&D%#(-cD_qC}-X8kRym1!(DR-mZfmKOnKjG
z8>2v!wxvoM2=m940E5qnTS^CtYaV&>RJQys&_bXCh@kj;`Tk5z64I;J?AN-w04^#I
z2nTC60ri!?wM8Mh<j&{7Hzl?Be6<0HPINC*Y=sWSIw94S&(kCRlo=_S-PWB9rZT0X
z44M3~pjcwLqNu-59XUa2J(`ao8?E|>FL#gacncamalL!;__EHM^vqY?v2?um+c-T2
zI8#d9*9m_hRTGkVU?`CK^N+K4%5CCdtJ^jc>H)(GANNNRV|hRL`?jU|0fXUHqucJr
zy68sD8OI#LSj&Wt=7eZNh7R4D0KAL;r{{ETHbLEc{tgD~e&u+it)i8V&9DLfKH@dK
zu(T*e*NmsDQ@{VrFDeLDK~}+?mLds{W%FZl_=`V48s>=!PBXA+65+}`FRTZUnC?WB
z*nl}h*6Z~SIi7NTfPlF3)w7AHSmm*u^Ma@<Rzo+(35-^}u2PRes27Fgc+bs|``%S%
z=Vm?<j}D1FV?DUlzDScdUf$l9VtTdr!h)Nj(Nv1s7Z(&@$;FN-@^!Opta83QmMiCd
zkNECz7C#h$>asPwMGZvzOFK|g7or3pJB`G}{iC&>mJCIGpTceIc+bqwIL-GlFMwhe
zN<>N&JH?cYyzN(ju+)21xND}nn-T=)>EqP>8G0e(Qz#phBbwD+T}2dBV)X;qAuf0B
zz<FI9lScYK>zNas+(T_8Zb(xh%uMbG#Q<Y^#Q_U%OBBQGCbM)ygKZ~0P@f==<N3I&
zbp9adIu#(opk8*m%s2%McX`X<M1W`Mow*q?|G*q6(|lpjy=DvfSPg72UkL{@wP1N-
z=<DUBg@6NxREtcFI8k{0#PpoMFl4+X-<tsUQJJl&i8>xBu(O>Y(xh4l{avSkY9s9(
z=kII^^a&e--)XnE6fOIs)YF}g4kS6U32q7n$ON$cKQIN*Y+V(n{`1REOi1%&@Gg3y
zHF+VQ%V5qvW+%x8PyJSP`Rb20jyt)vdN`h4!Li1wT<?WQir0*i^b2eL`I$7>=^sTJ
zJZU<Gva${N-`Cx4Ax#<P=4MkQZ)L4a+~J`bx?boRKK~p+F2WVLC76EB#N%%kFm6-`
z^Z8XhO|E&&9&9}%OVA_!Ell#E&NgvbM3tGkNCzfhh$`y?Hx&%!;mwkDv=cUa#Jd$(
z>XzQE@K>Zk8=~&N?^H#gF#LLD98C!p=jIWUWJL$58thwRE3~h_pQB5pi;g=UX$s#C
zhw)Rj#Ff*Jo%oY0V(Xs;cd(b&X)mOoJ{GEK$XRDr(=HM)O-tP^EK(cvtGg~WCYdfi
z;AW=Y`{*Z(e!Csj!o_fv!{4KwEHh3u6vID*ort;JysFd1?K%MdkP@C`ug&y6l$_)-
zO@6etP$_go900*tfPWGibU@us1yH`gFCT5@{%>wj2|xRFe33$?I!|_|a$^N3S?NoE
zt)os4XR&`B2Mnse{fY~R&J=nst}UGKsh)@Vg+`FFzVT40p@#3P0^MM6j<CqsJA81N
z5HB6lDaYkjy2t}akz;*;zEp9BI-Jw0ZM!XJu4h(e7#HSH&O87Q_XlYgdx{3XDiJcq
z83%f25tr8%d@V>FA4Hn9%Ho2tYUa0|0B>N6AwcREN&yC(u}K;_lxQ_wO<L8MWe=PA
zE=4av!O((g{cS!OK07thnw=e&O^2pCD>D2t(EQG)CTQr(P~v}GDl@lv7HHO1j``+i
zXWm>MdT*DwG3+#MzH8X@4G(U`uUIj8gjbtXe!`D&ySrKbZmj&04BeNKeDL@wiCKG%
zme*0@FhmryRYHo;Jdm7f^%C?|d^3*<3T(G-V3D&y-M+o`C+!P0?;kiT!DTT_WoTdQ
z(9m%frZ*;7EcIU|eVR?U)gzaDCx&E8-8UX@N~8>#o3}Q7XL;;x4W9juZ5U|)^%(t{
z0IG`<*fC$I$7E*ny!$%hoSEr=T1q+PVXd~4gpt7tONe5%1}Rz-Q~Rc7+vAA|J5=!L
z@#WDNibXZz;<SeOMGrQ%JWBasNR7R}P{=#r9yfQCrq35{`*&@JwT4tZnjnaJoac1J
z@59jz=gEn<I5sVaCBZi^UsoiCepda<D=8{942HcZ)h7BK3k@CEr)Om2#4VtqZF-q-
zRFC&lKXG(LU@ZX&oQxrQK(BklCFZ0)&sra4CZNIXVMDMmqet|Y6>|r%3i-oJ_D?=K
zCQbOYxm1MD3mEJ4{0pVu0a(<*ir3nH3Q0%BR(uG&4lwvMl-x_@N5blzBBMEhs}Kl(
zI^xlb4&s<ID$Hv#9Q&m4oBfqg*)OyJFQYjo@0Wvp9&RX?L<;o&F1HdeUC?nioE#|7
zyxqeCk%yiB$O<6cy1yH!F=VX0Bf6GCl@%7Pd9v($eo|_8fg?BTiQ~?;dPpO!6?gpV
zqrG@gH|A#}`Q>T<K?evDv|DY&B#>tFD1`d#UI%Ls@~@XTjQ{>g95RLm!ZW}ePs@t<
zto4Qh{lcS>{STeSouQ3|t!UuH6t&r2n#v(iANa`8K51vE@q<_UCdt<QdYYoXaVg@b
zUiY51orBbG$=gAKlnzQs=gWKbm7#_sZ*$v3v5rtjuwoG#t$o>Y?*M-)zzE@RVN&a~
zrstojJdSA5a&7W!R~p~9W+;+|BRxoso9$d(kO1(rnymDDw4l6B#e}66RPfwqDE|-T
z9MXQVg7W!!oSpalOnWHGPACyNeK^$z^=zi0jwx!f*X;Z0nkvEN*W4u8U5QCe&E5_r
zr+}vi+k)x4Fx=h!zm7K@yfgx~Ez=@{EDgLy%45i9-Ce6U?Zh3T-7+;?|5CGLP01o+
zes_>Z1iliP?OOtphi#xR@MXdluSKJZrCrk<bdBDs8i#B0YkBag5&8-Y`O6=x(QH^;
z7srA`R7A6p%W?9)i`8v$Wf{9WLnKn{S5?Myrhog4c@dZQmZuN<`BW}Rm*->Emtn2S
z+tuHaTO_{!2Zw-Z3)`8aDOeIXv|x(;+q`3r%ip47IJ-&EEK~rhwNdwz4Hfu<k+bsf
zvNCR{x4Pr{@Oc<gg2JS`^9?<~>Rh~~)`cmXwFxx3J>Wh6`3*7q=C)(Xz<YhwjlN=P
zw1i=U0~*Zwqp^CD*y=@+(Z^fs7U4Oz+Y`ICwS-HI2Tj*IsvNz&7}R1V;4@c%l7wUe
znj1=3K<jm*^zb%uokDpxgm}cwp**K^XQ@eXH$KWy3b5L!i*;-M$|!o-{B&`x*hX$s
zKzSA6c27Ml#kgP|dDi}qan~BRxi%CR7<hafeXu#bI62>OSJlSvT@r|QBDOo={1cN2
zuoLQhNma&{-=qsLq%<B@+jY#&y-|$_xg`H{hfiyvw%Y@svDbp{4MrL+$*la?u)okP
z840nj4w3$^)46+)2|*#;)&EeZVOpX-m6}d4vRgpz27V)Q>J-{jcr^|A5nD2#MCODd
z(TT7k698c#)4=k90U!m)nnfR=MrHKHMrV&2?HU8NNziv|zoxzy1EVL;nR~<AnTLM4
zgB?(`lz+)Z@)peu8H^j@lZ`J&WYDaXj_>@U!5|vO>$eKxyT!}7ELI_aqlD%g-17<-
zM%`MFNoZI{^uwfL27Qd#pfOLkSzONj?BASUjMX25_0mTxKY!wVO+huwUz`^S1v9%h
zZ|<mGEKW*|%WZq999cm+ZEr)m1iiQBSSL}i3K@N8b@YfO)(oQX1_sAll0OmCpAwkI
znfoVaYm22LB<Py~|AUdH_Dv1_yOf+QIb(}sy5i0Xv+oOA8)zfZ{mYM1i*-ak=Xq%l
z76`qg1H_O?tL}z|x2KL9ZOFDDtbF-$&;6lD==JW_Rys>^j80ZjvT1X7JwC~6>JcVh
z*e0jdGko?4-i>t5!7_g?Xe*sq-P+9k*arDAF&p6Oy>jvy|4PEvwu;yg9@>1-&r)!?
zh&nqhxI8&Iozb&_GE#9rQ2MRPn_}oNv{UohSkYo<HOuZg6&Gp_zdSVLZ&Ka`dj!`a
zpkY1<k6~Ic%6l-t7m;Z1?vfk>fep;Eo3?5Aq_r5xg7w?fn<TCbcTfXSMH$8x-Tu=@
zrCC|z*!%v{6}O5&xu%8hz=DdU)wolIlv&^n6(Fyrdt@#<8`uy|mh1xsETyX75rb1}
zV!&?o5RR6tlj~FxsLty9KqT2q<<=2;Oz5#w(H{#7?l?OluJl(ASGox?0SJW5xJrki
z@mYAxh47#@C_|PMIB`vDOV&{){NJf*B7Us2>6q-;+w`;!q|sDj#sulR4UKL7>tgkN
z+I(|Mr!Ae<5bxX_FY&sjBw;C<0vcn1;Z&H7g)F$RBPdY1<pyC(9dPdQNJUolm;Ip#
zB;3F<lc!K=oC*C7(;ZfE)%FqNByM?O@2t-DO^x26V13^3zMDvK9SE$yKNI(ysL|j<
zuMU^&nVagc$>+1Z#gcV+;y&xptl8ZjQua{jTbLd>oXuwv@Y?})l+UxHIH`V6Y=8Fy
zq>fh_+)&#UjaxR*Xps{XNYXGM`ySGS%Z2w3$3k526_(1#%e36r6aZGGeW@L;j@Ulh
z@i||Jvon<2N1asjseV>1$TVgc*V$(ZZw`U&2IuQmg8n$d#^Kz&F)bx>bZE?yvOca3
zpO{MtjkcX^Z@6os&>Fj3@kX?8W7gDBbXjH`wLd#M6UV$Mu7BqNk`Syio0pYqb>-Jo
zLCL$gfoC6cTdlA;?l4!uUWruDk<f6K)X@5OF_ZGXWNm0yFK<BpjfF031sJ&z`cB|T
zuFr_-G17q{x@fVriZ(K7CD$sDuYH>|cYL+S(C@DJzOug8(cb6tv;8L2#lHD8doci@
zV8DH(n(6dB>TC~p+i*Gn{8&VgNu7P6X9mtalEyu<@Z*J$h{%PxC!Kwcs-wNa8^ZK{
zotQLRO40Cb@=8N)#dP){ppDn9Ph856<@Vju%Lh$`i;eEEn@H>Te}#=saz)yEd9BY<
z)PEwgDcXJh?OWt%vHgI9T+)8M(+mol6FfQMkIfd$&|qk0`E)0iO1p9;_tSb%L@@2A
zF&VSJXXYOdM=a$W(=0uHDux8g2^nZek6a$pG5bQ*ylBzpI@5uE-qXxU#hK*>yj2t&
zRAR6wZZvGh=0yVVLMc0WqU|suXQdopxeyTKQQHL7%9rNtpI7>)t@PAsj2o2^^BP3$
zob5e7cS-C#&(&`EY>hf!-p7qiJr7XoXbQdoQoSG*pprYQf2>6Fumo=LQ&S=aAF8q%
zLMO9`W2Ro--}j|v=jUyziSO=Wl@O^<QlEBx*b_<V1%bMKD0^(|H*1evV|BH{eMZ6J
z-SS{G=GDEB!hSvR8GX99Y^XvoReI4&NdTP<f$8}T24Jr2C8=ZG^?Tf?1*Z$^khGHV
z*xwR7Hn+(m^vRmD&G2#Hn6mEA<*=O0>~tNidNEvM)6?~MYzu7gu+Gl*sI(u~G|VzZ
zdo$-6Hl{rJ&c_#ARCElnMn7sr{ZupVZ<So55RRWL^|5o`Mci`_M{R%zbFp@EU_!7(
zCMWQx9#2@H$O;IQQ{7SqgIda{x+b>>&5Cfi8Vu>Og)CIUo-`f~7ZfbdqolarbzTz5
zK@4-Z#*8hNTDRo!6w$7t-FWII5DecgoAhT9nL>7+r?GC4IvlQ5hj_FHi~}K>&goR;
z@16@tfGk}s3{WO|<UkT^F+q!Gl-*7kSTlUO=2wuv;_%$gIQ;<|hF#k|qQy4RQp=*$
zgdL5D2HnMC5{<-Pfj5Y)o}{`w@;V|Sdu{7`{>y~Mk!N-|OJpzpINYxKk%_KzUG|pM
zhSlX}Q$OnL;H3ZZY>((YeA_V8Tu=RVp#-}S%f>Z*lqa=flt!iq`h{*f8*b8WsOVJ4
zCC^=<z|aH&%uF|XZ902;aA{D*3w3wPnB|w6{-Ok0r=jnTZ^6x|@7IY&(>Xk@LK|m4
z<?Z^T^&md+%2bm|hw{hv*^hE$9N|`JHx1-#`^u89Lz=Pi!xX%!d5L1&r<xeMs@qe~
zo7+~FQOa5C_oXC1jr8PE$indCQpKx~`>r)gf|u96P}0j#V!-ea4+8T-lqQBTzV)KY
z@5cM?JyCR^&aR9AB!ti2KoFk<uJoSf4W%1W-}c$CSd`FepD<CqCW?F<En3=^+q69&
ze)eyHI2q?{$H&{D;_2b{QkB}IS_;4T<g_3{DcrkgnBhK<S8uN@dWV(!)iG$>3Kbl7
zV!eF(a_?F8ieZ6@*JNdb0k|r;dVWwoBjR~d>H}Ya=77$pSq5TM;uLqY!o;YcsD{nS
z&DN&&!}H8}Jnre%zSLxS>xslw!ij$-(b805r#?@}LVaPsx-;Tpb79$S+@4!Fj&FS6
z%2RX-UQt16{_uD)1tUS(depdGVd?&cz#0E4NM+oY`4^#e(=!ozjZ4@sx7rH$ZgRKa
zyC4mg^rJmDF4Vkg_|*y%I->9uiV_Y@1<|JpK#Qa(3Tc>_u9#PEmKJnK#O21W&0ie3
z<&>7Yl(~^Fi{mn9+sr`kIRG);6gQGqt4C)yHZ$%xA2>Z>?37RB7x{+0?eLg8Q?MEI
z<K5RZpFfHPUHWlCl%n9+a!=BZgs}CEx7%+J5tFzcZn95V(B7=!cxwUG9i>h2Bh=*6
zXZt9NW(saKD4|MG^w9#t+&w<eY47h%`@WQa4_iH_C#<|Zo#f!gz~lMjyobaebCx18
zAo6G$V$(ocb|x)st6u(F^j1AzBsR8inHHH84GKg3#3^c6_FOBT)vOAXu0#XXiN|3R
z$%8%G<oeUYRCdyo))>miNmM%M9<>k*-%69y=k-Z4#3OY2c&u)x?)}`Z@WD>&rLX=R
z6}pH@k>ah=%c9)1hQk%yVTgt5Ij)TqD7XLF)isbtmLttL64b3LZoQ?AOgL!YBnC`3
z5j?Jo^agik-PfUpP^3w`P+}X}&#Sb3UO|Wi^7ilhn28tjHlmyY)s=5Eg83zzZ@A6P
zD!fPnwK22kcC$0SE@%DTj4Ug96<4v}+TTi=tNX#%`g)R;p?duy=$sD|WdywuOj?+Y
zGvp(0ewV|jOOkhISjXq5cPOdoFTi>Wm+4QIl(K4rT)kNmzIbG7b3aoD^LLFOH{6W~
z2X~XtcSkshz_aHYgC>q$7kjB&-Bu8W1F1n%Unn47G`uX;B^_mn-=?xc4aw-6RPxQW
z6HB?cbhL=hohqIzc6PT6NfA7pQMgs7!7XZA;o2+cdoB(L^_K$5qrX@=qgw(!-P^9E
z3C<LKuxXkIec|IkZd9KrY~aKp95@(`C59F7Mcwk&_4X|cbb~RqN-$aq7=k}PhG3j-
zAUV+Xg3@*Jwq3sT8|ijIakPA&_B`2=;zMjt8G$^TKhDBACncjB3|!{=S6zO`P(7Z@
zx+y_*4YgcX6)MdYozYh3qVkt3OJuhbHGzARYM$<dOMOOB?DRkjDIG{$<(1TnOXh`G
zgc%DyjCL0>6Z~n2Q%B<Jeg@Mz@el|gsEf9F^mn-@gOTEoM<=q4kZ^}GTf4^oeq8nC
z*%9tG%Y27TqkfMzm@(4Gw9mpR-=i;TyQgUcr7v!qg$nYqm+G9Oh>DLE(fmpl6(qE=
zAZ6#{?2c;h&^U2?JAJ!e`4Z>sh~IDy<ndy#_(~D0#IuYZm2>6gNT#Ntr^t>{AF^(1
zVWvW6j_9|=W7vbOy#Mz@2<EikSI#nqz0DP3eI1_w(wn=J1oGRy)0%9bASQxQd1HWb
z1G4};U)^bdamhXjGZ0J=mH?1v63b}KWbjLk+tm5LZuyW|#tTZ$_(HaIBRNX;!YDkP
z8qD1u++<E?YxR_>?Nv$Xt%XbwdbJwBs(Ot;O^fqO#?EI;JHv8x&UfR;@BhTUePe<p
zhPu6C#IH4EL!om;<YH??pIZlp>Yw}_cFUh~)s;xmrEb;P3+^*szPQ-xi3{B7sI&_5
z0oWk0=s0@=Uv9)$xtl8}bTZ3>v=`)TDq)fE#aQ>M*=w=U?zL?srf0V&gIq28ccGei
ztkJyfh3?YqVv0v1b;^GeYLMM}7#W!_G#Dq2(CMfF_)qc@9?Z~Z$%vAskpP;f(~yg?
zfL!U=pH`crtSsW<WLDhp?co#;tGJY%1@N^p);=ic0hD?GnB{><&H!q06~c9lw2Vgk
zjV!+u>%fdKmi_hsGe$HMz)-kQ!*a97BLId0Y2oeJUrD}Qd}cq)BJ?DpKSbd86DTIQ
zd|BBn1?nzt<H_5DTHLD>ppG&$aW3h4tl}UHc1$X&fFxro99Jd$kk{N#v;gfvwen|N
ziZTFUxz~BGK_O^vIz2cZq^F0eP}G{7vm~zb-WO9mO>k{rFhWGBR*yab$(r<C3qAVU
zKeRJ$*Ryou!8lBplzl(i98)gAK^x`j*sQ?TFB$Uok$5%=NCyp@D7-20w_S?<I)E3~
zRZk0EJvyj&e%)!?#wh^^fM_&84$gVeXecJZU;X8e|D)+F<C=WDHoh@pAT>Zl=~3bj
zaCG<Rlu!hTNhly)k~iRhp$LM=Xha31bBvHikX9*Wq)5mFq+`Uh|BL6v_F|vCxVO(8
z=bY<0-%EP}EZS<U0M)wKLmS^j>NK&#VCo9I^OHHY_e8oCpcYO%;rCMVEKj7f%kM7j
zh0?lQd8`eLwK4HA$sg%&+jYw`gC6v}W_myRg(KdZEa4pp1wee}b4}$RWufC#KjBnQ
zexR9`hKE(QU2&39x^{CDvbE-Mie5?0SpvwSiBZoRi?qy8F`!=GTCn~-q%d6zsqa#K
zTTi2pUfjUDk&PCl&nUlGy8}w<MAa}f{q_49$y-QWT1gQSD^RfI+s-r7)0+`vd=MB;
z+F1q?_WW-|p6!(Y?JAB9|0%TR4~H5l@$62s{!-5Wbs2X((!b3D@0b29`u$gdcK;g_
zljkdcil+QSLP$H~B9v#F*o7Z=Kq9bMh<}k=Mt5(qAJU+ZvFXYA-t?pA*PWlUx-s@O
zZkLQ-|Buy@E(glz&{GIY)hTwjm@HggLQP8o%;41cWVWZpN~$8Si`Ybi95M=LY{!^U
zNV4ER;L=*@W#_5b^B>OK^}02BM?cx@3SFaDPxLL<d{egBkjq^H`2hxmAMZZl?u?GU
zE_QUBlySmBNtVAQ*}A{Y%EsOYeQ61@wmyjFll*}7Bx-|8N2TkKu?G+e=0u84kQUHp
zN6*AyJmy#Fn(slJX`u`E|BUGH!zeyMKgA8gL7FIs;y<f0S?=4QrNSj}@<fj6iN}=A
z%}(z9t&z=<5%>HOxK08&*O(|<U<l*?75IdTFLiiN=Jwtv<2Z5^UAx(uH8(qdRTMT7
zpY^QfL;2vyTMY%=)L~=*H)U7%7SO+5nr@NY8tbvWAYXHqe{btrR#V3J@GFxOyJ5W?
z&pfVbG5?8tssO@3jY{_@1jnToX7i~uNsSKq1-u1^jgS`t%I1~-l&R`7{FO$AEJ|*<
zZMYdfDI<`nhnTo##Jx?wQ1^v)2Z`JEG|=;ohPL~!1@UP`rJSIEqkotO`oNUbxT@6c
zy-VP$6*h3rNLX50Zef@{T4E2h_i57SjetqG5x)?5SVe95F@G;l#>G~Ki-U(lkn)R5
zv_TrCj(Ys-;pfO_PCrgZYj-Z80U7C(v?Fui7M+xT-`a`X-Ejjd6W~cZN0?T)9oDXT
z^Utww#3@M7<14O2P)H>8FQD%+HAkU=*;C#929EfBNEHo%!F)M#9VCc#52~=2hXA=(
z`U@y;@#IESmNyo+W!>WUxW_)Tc}P@A`FfXy78wxYvrOXs`L0JrdLxrm%ur-(%mq{G
ze59X91?u637rSFE?v4lHn&f*VU%9`yK>I#q2k`}Rz@}FrRmHwQx$pf=nWLe1p3<#G
z_^uH?P@lZ|Y$B+}i>A}+aC|(WA2?Y9{3r3-)ng0!q%w|YMzVt^tx-bphw<{LRe($g
z;%;LIQUp;Xj}FLvz4F^jp;ndJ+>W@89sfhX2L9L%NOJ^EBBG-7aBzxe+Mv!OQiH^c
za%xdc1q@1T9T1__T-IkIi5l(xW}e^pZbyH=B#9>>`;doG(wJ(v_wSY%eeJ=az(eEj
zjZbBMn*Oq{oSL%LjuqEaxZ>1sH^%p-6j(8lxrmPrqxK_1^ktG+3iE(d)BMazLj-B}
zbayi%s(@IrD5#&LVeAc?n&dP$=krp%`c@-oo0C6{LIap<$SDg&-?yoGeY5F)0}gTm
z(hNNEo}U!s17&_>D#W{kSb7l&Lr^30S;m;<NMjDk=;<SJ>fYmTriv!_!=!sOrfbJx
zg&<9ha=)+wri7uL9p8wEhXcJ&yv5FJl0{u*J&T1N9X&Rg#xX_Evh1n8bN85R5YR{!
z)VCZh0|ToDXz*;ylx#ZR$^&795IR?ymEi~kD^52WEsKXeO))2v!F9y;#MhEALL&6c
ztAcd~5R8;GOXxH&)ipSy5jQgg$(|u9%?Ej|ze|Cf!Qpjt9MxOoj*=gP=EM)?!|w^x
zK{v~I?)TC!IF6FR0Ov!zW7xgf?9V^mp(wFHJVF?S;yJDNW$qIdVOaOmQ-kR&`u4#n
zq(xTKTy9zTgFELd?QEQgjTL0D&p11yFZN})6P)6oUs{RPVJP;<vL1s=(R~Wu_%`L8
z1V!;A?`waOObV}ZV`ss*|BeM^>ky|#C~E)C3l`<pu?5R|z62-GUhA9;yzdq=qzGWs
zyW2dK?QaSE>ghps{3kRygFTV6b%4DR6iD-Y_zueJO%LM0G!%vejNgf%<b$oDH*WL<
z=AhG_a1}sZsBu*{VDEW6=cegO6!RCR@I!Tt;PqgR{t#Wv&ib5QwYIGG7ex~$(pX1C
zSXjiwltaE_K}TC7?tY4R0aK$U4KUlf@=W}+VCLeoh$vEG1KEuxc&8jY4@XCL8OOQH
z3!x;X8Lf{Flz=0j);WnHdXFiOz}J#1=CZQvqvH}K!x)7L;lJxV@ok?~v{!#T^5BL)
zK2v`dkz6B5aXhaQ>s$ZvYNilh#eaQFIo;|H?mc}3BsC%{gDO3zO))&w9O9n#suc%v
z?|gm#1_&s#vd-&eQt}}Up4)2Z+dJoBr?*Rd$n-Hsi6NXvNPIYUQ3=8U=E0sQLBs%o
zF1H3AIU7mTyxf(CT6#`N-<}tJ<Km~&sBdy0k_el}PtVU@s3|7?CZh+86}b;v?d9Wa
zOXhy_Am`{?W|&H^!hiM!#4LFl(K5EG8kO45Q3%AR-B9hEz@ZeRSC(o+&`_ic9HZ~7
zvV1Xpy%-dqUp^U^X<$>Fi)!i}kpvpqW>FK&D4IZMeezOe_f;;hehv;kLBlFo8?Nhm
z9Ph32A2r6&>jDOSxy&&)vD}5j)2HVI0$$zGvA~QtYwtYnsli7nKyTXDXDZK>Al7hO
zH1qCDH_4f3=Ud7Z2xL0i$iQ--nN83FBJa83NyLYiR+jyA6171FzE~eb0#G!cu{KK#
z+z5{o<!8P_;}#sRkM6!{Od(VD*KhsbPNFs#bMqzhYt5eldz@9(3!|bd=NG<DXBikR
ze`&J(QtS@noo6!5HJq9<wj)+z0yj-^Y&O~?h`E-@CV6Km8k(DPU47S9mbUmn&&pYr
z+;rj2qcl-#mtAu|La+S?27cANc6PEOc}?y;x&0_yadHUPSmj3TwkRj{c01I!NoZW_
zqaq@`G~)DZrzfE#cO^`%gB&#0me+7g!R!6k%GM1+73))V^yMu4qV%PjFQuQC$8mT#
zUE_~9<RA0n@p1mzk2BVbDhu92N()tOjk!`!$)K9dYO$WT6Sj{sQh)-r$m3&8o6Iou
zkW+1Vw6kZsSmT6^-Ne>oiFKKMwTovXL;7g=d(^y#rQ^fB$`g(}FBJh8{Km8GEynpt
zFNf7ByQXt}o=UIH8N2NbYBc;h4U$>@?tubPQW7FESnR+#HNW6ky>J<EdALT1$ZAuh
zV&JPyhzY?YJIT0B0kD>xJ-N?0dV!Ng2W8%0JgVC8xa206MPyu}F-d=_QC#e&UZE}+
zwPpSiYV$>m7920&WEd*18Bp`hT>4-5=0Gnc<;O%(=}k0iitb1LBzE;wQS@@>JMyj6
zet%zW`>4bbXpK5OGx#u@;lpf-F*7quw2<b0YYTHLguNR96pXEcTStaeZV$CQ7)56e
zZIPi(`gM0B0d;<~Gtx*vd(zhRU2iUCgn{3m<%fzC(R;?vFy!u~qaRAtH51`2pJzA_
zu4v54*ErQ6>s47=g)5tZj#&f*9r<c~b-<pKdH!4x>xFVLw#h>D-uX3ALfKi0ANXel
zk9#M;HC%uhax%{#5>K``=U$xct*o2`wpA2^W6+BN!Ec;?bZ>K>XDDQ_Y{m_+9KGE#
z+FMfi`InUv+&L_?R%$Hj8tt3n-CnWTC^qeS7pLLEqb=l{Qr!4i+~6U5I;HVkVf98a
zBI>Q>L*{?>ze-(DJjFUe&%D}M8Vd7A8s77h$?$;Ke}7-JcQ~Xr{blUn*kX)*RnK<?
z^t2j>ziVh9%3w+hPnt`4&2hXEyn~st^%0lbz6I)x!Wa1SINtcUnoH`V#odv}NF#eF
z&<XV#)~fW}5J*z&S9ck61js5uE-4&Om|ZYEIB9El40L#cyo|oyQm=1tR11c|sZ(`j
zqydp1yoVj<heKAoQez7pQR*E98=u}S%{^{t=Jw5DV=9F|N&C@Z>OkuhV)P#aX<t~F
z8u?bVWI5{8H?u+B=W3W2o^TQv+Uj5#HMbd@;7+r(q?Y48O+Hk^-bzp6V0Xf~+AM|{
z6MYlZ^%mDO8fO3Z7^KV&zs5zTo(iI_1LxAbkPNS3OuToon<fV)GcD=9;dWcj%k$cU
z`BsWpK~qKA@xHw4Gj2a776fUt^mNAR_@(#+jllXt6xBXM<QyU3@_Z%u{>=AxA&5ty
z8y}4=?@?_hNpPvK{Ri4QYE2N(oSoC2oSwPoHS(!EfhP5}CSO)4-aKLTt=vCZ1t5wY
zmaOX}Z$EVvw!Q?mWMN^4nb&?9$k9d5VVM}gWl&cM9Vd~e`niZ%ldEtLIefvUhLh^J
zI8$KU#`}PTKPq8bC{R<0JUb)vzXaaa|HES+;UwX@8WnBR4mpj`A1mDO#;J@@$%9_n
zV3J@KjOxyaajOT@I6?=`QSAY6^$Ww69<$dPxN#$x;|AmlcO3)c+!vzW->>!*`Wy6U
zYp*VM-mC|Uurh{!BABeAU^l$CDud{&akgF@cRb0tX>DNV@`J!IY;H*FoHstlFiL?Z
zxqlf29Pf?Y#tsS53P@=+%#8U=uT_(gLZ4<wU&t=-jfG&kZc5=jt^>ON4NWGoNZz4-
zP_BEFI=R7<W5_6Hl$dS3sYu5cjK}?{n+mEdj>kwfmUMl3hWSP%1*O}rDl|xyXW*u$
z%(S)>?y~<i%*Dewo%70<bAzRxFFxMg#uLnVQ=1_P>ZZ0ZRUBn^uroB2kU~n~4JlCL
zISl8=B%JriTxI7e(*r^hUZuw*Lj1f00$;=X=sQ(AAsIp3ubU+wTz3{{kKvc2dfX;H
zM2G1$ZbLn3ycO+d!y`EJwTZKfSyIJd)oO#}-%?S%O+4_xWtTs@o~gU-Eupww!tDc{
z;zJ4w@-f|JA~EsiAC7MYw~?{5`3t6A4jM1-IE^nb?7e^--vtI|#qWi<4g&|L29KWn
z2zH?4*h%$!bSgC@ffc0<ojz!MFp51IA`Fp+2q!0juBNSXTYrubZ-(`~@7j4i+j==M
zX*b?q((i6)pz!iM{U?Q(vFxhOZFyBS8#VpHC9&NYtf1iZ&N<}U;TDP=?2Eno9|%Oj
zc<PLlF0Tx+3?a*D8~Fhga5ha{I{Jk9*5n*9t+bm(G2+FjIjMj+#5%I&2Z3;~4FDI*
zDOj@T(rM0o#quE^CG=U0<N8RLK;Sh&Lkt?q?)=}2@G@S$cX|y~c$_@V739MP4IJoR
z$k9KLEC)}1s+0cvBbuBS^&$3e-3Fj&MUga(kQ6UJz+1dZlDbC&q-P)_G~;?J3GtoA
z>$UsWSNE8-rXDl1#5(&#wTj&yE5UAWcbxBiQvQ&eMY++!$j@BM2Y->0b+FjmTwaE^
z?a8CCxMlozHR%47_mFKR-WP&5@u<nmkbkd8fk?$5QWMHOk^;;;fxH%}*;l%^uLyKo
z*;U#H0rsm?w|0f)oacKvmuUql+wgbiDpDGhEGhN7I~y7stKNiNGhkwZi|RRf=J}T5
z4(I!8>l3Q+hvx)h0T2-$8D+ONInl$nB{>`eR*_?hPjzrdu`g+2zGw|6#&mWCUJ2)_
zWB8HI#r~FYu4m4Zyo61h3A+b%a=Si2uhA6^a41<+_mzS_<=|lz12TO^h+T@8ySgQF
z-tTUJ6n;590a0d#RZq!+-YAGreBk?_VgAC^+lc&hnwxhB!8-C*_OFd4oagIEQ7L5#
z`lS^9y<v$*1l;rWndy(Bqvycz*td#1*crh&e|Bo4@24U8qO&T4xSa*(Z#pAPi+aTX
zF>08%Q~QqAtN6!K(Vhu}htqcDS3fD+`cgLKgf~4h5PB2ZelXQ7E`%~*{Lx^L(O$Ji
zUf04;j}r<0Z84!J2~ma-JEEM{7ks{AC-=XLapL(S`BPNwu_TQ~_s;r2)$emFmOFXX
zWJ(VnJ6%2I1GMKCnBn(kG2>PlEfoMSZ;GX;%IY2a*6ijg>uS8K!xWQQidjm6n5BVe
z(Zt*|<J=R6&A`%xNJUe<VZp!aUN`gM8YBHko-LUTt2BN1JF>4PJYeC8Jz@WQMR~R|
z->QYhfFKvT3mxrLhzmaA_?{}bETq}3aCnVrb85<IW6hGpdwICBaIw&lHxGDI?jHYp
z6=29g3;EW`(5?3*vhwPKR%VtksAN(F^AYpBFFhLwwAx-_eJ7WdKhE~YV?!MfizP3J
zJnK!RRpkPwx_{ZR-IT6|q6I}?+g7v@rmCB(*;1l8JP!HN;c;1CDqCAg3f?qS^s@6D
z0+dghA}B(<q7>)yJ34^NQ{a@)5phdsP3mDHS86Z`0<ROqfj-O!y*(PKz;CS5E38n&
zwBcq>KTQ?K8`roq#l*?S=iY+R!6-2Q$WF<JZl1k$6`uCfGZ#vsWOo4giby*Dc3IG7
z+tGY#U*hxOCsysnwszzHB!BbPpSL2DSqyafi}`v9vj;q5&~+X>Cm}*D5B%xqIi}1l
z632rNAnlUH)5{d~c}Fv>j2`46aLZKGHEHuNgBFVh%DjARZ_~fMts^hodeZUYz)>Au
zm>}=j%ywhtEei`(Do##`ej{HF77Kc7+cYCJR-Az>wn)=|%5_K*75S9YGMw8ND(PhP
z5|Y&k7k$z4+RKFxNtwh`xWUsd_rSAa(gen(5DTI&6j1Urh)Z-VSjbobsF%jA1YP-_
zkblvXYgA-Z`vVCt`=1Z7Ls5eX!=|U4|G(0G-Zi%x)b^Ol_euh)^;A!QE-JYC^74^r
zulV~2!6+#+QB~vdLSswe^&2o8jzBJm@visnCc)_wRf~Qqe?wfm`A~_S>r*4AwyF}u
z8Hs7)xZmZ;-ndgjiJpPzXNAH3Py2f~d!|O#Tf<p(rNN$C-)uwhg_TI*(X<i^LzqD{
zV!<0%r7es`y515L%NBcpv=Hr;(U<=c$UF8HoSBewMVrM?lp!c3G;`cy^B^b`P>wv`
z+vmlu>qFGROxT+S5H;GM$dilH{)P4rSRVt(RbcPrlSKWY1X{oXGTdW4MTxut)7OHF
zbqgYmKUCURUn9;n*lnXZ6h)By4fE54O}VkTziSl&>bS2kUd9B}Iz%IYI8%L1N0?>4
zB#%j^`4)x`1Y1d0XzZf>?<P4ia7|Q{3hBA&-v;{SJ+o40)VyCTP}FB{O?X_*@8lB+
zG*7HmmyofQ!1+bXFheZphx0K}_t|SqD7yZA8AXd?{qHnIAr6kGvFT05r6cwNn3(Gz
zz4iq8koCKycWs{A0s+sccXQG1tjml?r@~5Hm*~yJuU&=#B2emUed#W@Tz{WZ#2$Kk
zgEKRuDes4s+xNtCdDaLZ;4%}SygPJ2BHNOwHHk1}y_rI=ZSB8^4)xE_2;pI=D~zS-
zOn2gHwo27U8}xGX<p!97eoV}HZm*6a=d2!Nezt}nR&xDE6-5l}36DA96iuGwj?8eJ
zanSZ&0F4W~gS~fsM}7vevX$aGf<>g;SBKdY>EAkXdyHJ8sN-8GkfNe*RE)bB*E}Cc
zOgB%R1Pfr8m=bzu=C7!+&!}^eueaW(t;25ApdK$sO{r0oMc`h&(&>+iplFQ3Y;36E
zAH<u4Qe%TU=0&UG^9JQ~|17MD5kv?`(bI5wTNas5SgFfr^6Jl|cqUuXTuQUAso~ly
zXpos%a-MD>MW&H!7!v3qcPzAiDn$q&4eMk!?Y!Bd`Ke!<3wMUYMw*RR-R!O3-&L{y
zJ64VFINv-OQ+IIeXgfdrvv+||-3+14>-_eOUhi2PG?O__z|V84((?+N?kWfHnpaC+
zT7)lf!M06Qa^vH#F38-&y{7q8MD%-M-bGrM8v?8g6G3!a^DVZ~V3yaRfgx8CDyb=x
ze3|$M?2@iY!*nT2$leEEjr}`4to+`~O;ApsEAlq0rx*UGQbo(l5|4DBr&^*hctT|e
zl?Gk|39{O(#3AwyTZg+9_SJym=h{v2l(49ybovi4p;XEC^)>;8cv^l|+wGZF_k{5J
zgWD}E0rKPXs#<pPlL%hWX*vrd?6oiZXt$7o=>Xq^YO~$L`rliMCFd82H}7Ufw8i-l
zYwmagjnlThtlTss#cNQsTtPK)2!Eu0adEJ6;?-Q~96DR%Q=3{N#|bdnP`*4;Na>w6
z?vCg17UWAr8c=qEIbJpHgM<33dZt<37O9z+KPcF`yuW$Gk<qTq?e004g7}01Nv-j>
zL9UrD2XXhD=qy%&G>jSF@}OEKireBsf2l+?qkHaa8*8G!a5PXUSKuN^hf!HrqQ&Kj
zV?or(!M`2LC58m!UI63atya;j4JrWX8k#XvFnu%K{m^FpgXKFFhD+DR3G^_lcrMD(
z5RA5o{}`6(Gtw#&oz6;|QrIFu(>L>L^HYA6*L`t%n#34doIJx<@73VuS<{m7&9I<*
zT2|u^y?+WBbchF3mJH#lj2<Qh<#mRzg$LhE$1t|~+u7Zd#seC&5ib$5F9G`$HNM$Y
z-v~<s3xm*m#gPLf3r5RZiLVi^ic!ZCVi{HOY=WB8LqYCyEqyEgUbv~C^W&W!Z6(uI
zQJUbrAumQE4j1~8xH;8OPEx1YVhR^*s4_ZE{MSS<q2!t<m6V}>5$&-GjZLFfG#G6D
z8w{%*uD9hy0>8TKUVuzPe`_$RmnSf(C4OEt4$^pKWug0ypd$Ukcd&F~s@s^;Vf64U
z@`S`&0$d(dSC5sLKYACLAsT~bRgenQ2$@+`$E2aVO^9`O<#|$y5H8)h7M#)-JaZ9G
zUWn4l3vo`dEGX@UJnOyu^_ZLS`O5(s#7$&Qj_hp3ZD2?)?ylt#hs`g)8`1i-0Vo5s
z-HEBF5dC9MYo^ZD+>A`Yw|_FD82*4jnJ0fX>DLR?)HfM*qzrV~k?;0SB7gYneCNPQ
z4Oj`sK8;ZLElB^NWodVNe;;=748bwdi`L%%H50CpR3XgxioY(1w@@SPyP4Q>!Kdy|
z_{K)t1aiASJW}$Hn*s#A!!98vx|U`!5W4t&dR~PK{*qHjbTe4lV1V5`VO~78&Lvgl
z`=Ck#n}iUPUW(m=>66{4^SzYn%N=z<y)Dmr+pPr?{(HymgCCSZL_<>$#y_f9B-Z(!
zKE$AxZWiv!qxe+7dIa3vV{O0yf<DY0k5h!bv;{8I#G3Vjz1dYMd;OBR_6(z64vb<%
z{T{|+<~bFG3Ng$uQws)LFZ)RQ40987Xtuvw8jQ$*&U~7=$;^zVf=&Kf85^Ho!!LBS
zpQ-Nw@@MLY{Q{3x(Xp8}Y&9X)d*Q3%%HRI6LinD$7*MXeRTg&%u(B8P3qOz6c$d4k
z1RTFIlCunWc>`fcUq5d(XDutDiPVy0mhBiO?vYtqN2#`%mbVO`?<$EcjeD-ma~VfR
zdk4V~Ula@fU*RY5W%g#%gp;tgS+`9S=o5S7y$$I<uZgyY{i0@;683F}t<`?HWT63@
zQ1Ip+IW$zNCw;GHX{kw_t9UNm)lg~$NdGPzLrWP{-3FFS<Vy`-w9xaL8CebbyeEWd
z`rW_$K2*%JZKOOq^y$5)%J%*5&M9jNn>TAlC$mjOxkFZ#{R&Eq<veqy52jROe}Dp>
zYB|L=S9+^KId~Y2wfoyf3Zd5(uyu_s6G}-;J>+@JDo?ZwGp(I1+k(X2NkppyIXvRu
z%FrL+qJOXBJmvQ7F_G=lrKKg*J*fmdVcnf(QA5YJrdcVTk9oOaNTar4t9Ymm;mnrB
zQvXGvVay2E2_1&Frg2v)DKs@R6peK*(to97c}cxJ_3CH*2?(S=F*4DmH(X1@mwc29
zVODsuPmzv>zmj6nd&>&;_CG)X_X2oh+&Nu*>Mz2RI;fy<+wq}4hX+GP?%1Jp9-a^k
zJ<3gkZEI1^c+pRROZF-eHzoJRKSH56cd}wC*v?x9?FCPagNnxLivDPw+E_}nOw$50
z)>a5Oh!nfesH^~Yj)WiDdu|3*1~s#G&Yg#~R^oVqThAK^X03BFp-q6tb`S%UhJIZ0
z?he!-R=+J|?_;IBEROHt02O#LtkTHyML4Xm2gPC-<c=RYJKJ0t+Fo$z=okx-C#N&0
za{cO$m=%~1Z{bHH-kTf34HWkD?SkOb9PoX?U(BfXom;8Y;XyYW`bTRg{|$o(Zgnvt
z@eHmI^XMF=m~5s55#S+9ex@*4qkvmkSBlp$o50b=HzcR$Cr2n|4Oi_)i@s>cl(*7I
zvhuM|zrTM|iSCz`V2KO<<TUu4wBTqdk+_0iAWt`3sNBaCE<8+04-8E!Wwip~FQY-a
zf+$}mh(Y330W9JR29%W<gBBa^3CN14sl`tbic!VBmi9OxCE_2uZ~Mu~cM0-`onmy-
z2lLaRCx>UJ7g5a*rZr9O2Eb-Ca*<1tf#hw)p%uAmTwJ;NTU=YO+S0OwDo;wHD_*%G
z=!#~{w+|twF@q<iM3J#itH?W<yhp8$<h3E+2r;;hLdiEuNWwqFR+{CE$@usVGi(_V
z+AQHYi31MiCsn+O6!*FXU9*el_Q|2<_8qu++im9)is%pMVN7~Jm1hp|^ejSYt0Y16
zO?pWBD;G*_y|)baRV_dhcdaYkY3SL6k<zURsAqGmCALJ|-30S`H*QRPP1<Ah;kys?
zuo*A=jg^L6Q{u|(<uPGs8z5b*Vb>f;s9Pxsz??+DFQ*zw(;&!EWysG~_)}rAq}=bw
zMUW@nySSY}m#Rw-9db|GpN(BYk<;d5dFeaITc2}wj=A*0xRXMUAOB>d-?#L0FqT~+
zD;Yt(Fiu3o<Zv!pF%{|uXJ3S|n!Zwr={Da8!Sj0Ya&Wo+P+$Q$u?KmLEjR?Err-Y!
z6buH6S=JT8{5>lZAn(#GZfXWp<HA8DfeJh$;+3A!Bn9(&=Hw_&%V;tzTYw@&*ToIv
zI~nQ@W7N+Y><c7$R)N-6c`(D*ccH9u^|s~BNAqti1s}nwwNP*mT(G<~v2A+B*2h);
zsx;FV))|Bt!(WgWZLx=Hr#po6b+AF3y*K$-+JW#$K>4=O=8dhuYj=H)LR$|&QT-;c
zLX7@sS{#ca@~J{!ihRJ6aa`*^HyZ)DLboUUGIpk$hvAEQag><%Y5#b!CDsrnpzl-e
zOaY+c41zqpoK@fE;h}^Cu)B%6vMi+|M`qh{_7qBg$$20Av!I(xbaLi}a>!h3sBz0n
zQWRn96I?6?nOL{BxIc7rW7~k$pZ>8}FxxwksWuYg6WlWKnI;ej<Yd&_YeSzZM!05^
zvJ6En4Z2-su(+zBu)H-wT@?=NbeC*?IznD2mA*NvAGLA%U~kHs(hppao&L*~nRMfi
z|HF5M?{P>4B8L@Kd5%PfP#x;_tK*^Vvl#@7HS!;|MjWBA80|RUB|Dt<hfc!sr+?nv
z<7k=WyJ7zQjBq=gVk;;#QgPG&mezf>uMlA;kdu}ZU3W~^jrQMUQPRZjGNl4wwBBIP
znzMWH5jSb;IX!QcQ!04<_l`wfR<Ns^B(7!Zq@^MtXYAmFa6p{Kule|_9TC&VYZzTL
z3z@ZeV8B*R=FGY6e@$2UZ<dlXwFcgW23Q8*Z+NtivA~=k$o^g$PGe;fuX=Tfplp7o
z$>*bOBdFCfbb2}6=KTq>#eEKVvAs0qB=N%Jv)LJD)vp-nYW6|}hruyl=PJYHtz{Xw
z5V@!*T6!AZyqWp)%>yJ??!Q0FJIiZK{R!{0t8fxSEmcMeiNDi^m_KGe3*+D`>3t<v
zA22HmM0cyKB*CU`#CZ;x$SXJL1=hP}&CzFxM-1aGq$U_RKvGyi<y9RNR=PekKf!-`
zPVcv~Xk5U%-l3Jc3hT+?C-*Z55d^PBhO}IH1CS3(hxj1aIeT!)SK;pwGs1BVyIt)V
zb$NPv?$s`VRk4A{QQyweR7woTKDB`$83AB?inY)Qt#-cv`fmw4Xai9MyvsWZGRP5A
zC!6m)6xTHJ*A7lrHi_huq<Y(E^VyUk6#i*p+XyCaH+|sWoPT?oY%lPE@f!~IYoGYe
zl6TOL_m`G;dP@HIvJG%JpqYg9p4gg2f32M+jfLt(dA!$i`)M-1v$WJRzVqj)!e67d
zEO)zX`-t4v@whf7>j&jPxn}{34(hr6L-NVEcMHGLDCX%|Lj2ioV|WIr{Z)$!dotDH
zZa+CZjmycS`RCo+(%`+PBAp6dMb7+w^(Hz*ai!EZtFFeq+WJ14nbz0-9-m8wxP{Xb
zK(Y?s-z=`X9Q5LtBRL*A9WyBem65Jr7q{JOv9)h0#R~TKY&``s4dL|iFR4`}y<9R=
z5tQCrZJtDwhvl!{-|RdnH&1zuC*9A4Q0$z{a##5p*MD5?N#IFNrEVJGfAU^=xn^L|
zKK5D2=(n#qoRo~OK=5w+Cn+)yY<t2gsCx!VlQ)KG(bTOu^syD>uWI!Q*P=-w9CBZO
za8Q2Bo~t5~$aTZP$>#VP6R&zCGi-_nMQlMjC4wTD4S1HO%WVV6W1?3ENOW@98yjRa
zEuU?<*L|YYwxOtEWo5ETS<u`*0P@XQ6C~6BeN@h<uI+pcGqtl81|dvZgYIwG($d=4
z2);=(q<B}-FS2y9c`(=Fac^@_;U=3@@l0LJv0u^+OTdsRAz|6?l2~29MCzGV?--e0
z*$)nljB3sG!376!$<N}U(Akd>gv!O---x2+sF{`SwCLCT*{5|Ib%#ZVuHTBni0!li
z@|OHwtdktyyfp-*Q?j<dw(>USisM$vlyX0{=x^AC?r*xM%Jr{6egQ<rxAWa)Gq#hn
zG~Oy8AV=qDegf-_vrFMyO5<=Pg@Wz6Xdd5xeD6EzR@qdmy~NPLgEePBAw6hQ>hZoJ
zIQdpY=EcNhH#c|SjX%*E<bo*Ei26ogtmO1tm=pex{mNI2G~;9Y{uc=p7=@KmT?ge)
zj)Xlj^<*VNb>T1pLM?6-9;qA&M1_Z{I<CzClVDB#U=kUxNIsEkaba!t3-e}pf>))i
z6SF-=G(VX%R=@+0_V<Y;Ddb?~A@%cdbyXD=!2wiPsez&5TIp)@6rmd*v;|>1lAa!W
zvJWK-5#!5bC#`zq)3BNCVlfq{3s1Z3r#{Fv0gF5|$4}a>o9Pd+H=iTZQL9Ka5K6x9
zZmt|O8`EYnCfHXppek`W&GPDw-slf*i9hEuTz3!;f0!5?e=+;RXPI9-C-GV^1sW7v
zYiA(p+mbtNJBgjYN^3bdM=$SA_H))817+0s&nitlJRZrNZJ$?K9I7_MD)Gb_&quxF
z^L>20Ct(>isHITcY$_RH@dX4>baGHS**PT3uXOiPI)9iB!-K4XHH14kj9}ZktRG^&
zfVgf_yl8zNF7stOXev0LH^UX3x=+O{<+AIn7uXL@<~!EpmVK7oepjg1m+=uaM&)Zu
zFVDb}T?6nZch1=HovZv*Nv%AHh>R`b%;+|j91Ex(si3B&%Ec36;DQHo472Pw6}Jc~
z?tCW{U%4BO8YP14H(Fx#L_b^7)thrC9QDHCGp#{hxJumAd_o2ts3kby-eiH@fqMNq
z4Yv=+FP+c;{$9YJuDDk;zz}RRO{RiQ@juKx+HU-gm~_pz+ul=Yy=8d=ubCM-vl&8D
zJmJ~#$&ow^-M(FNga!7b?$wX@0+(BjpHF*cpCh$$?~}z`QH^(C+seb>94zi|rtPlh
zjYz?ApNa|-CA@ejNZ44*z(!ZNhx?v_vEzmFc!$r?vL1N~tfk4^&5wx<A5EmzioVrk
z>lJy7mQ0i^ZEm*hovF;S14&9ggbyR>p^U|D?Ck3wwQ34>f&P+&kuNQcu2n;`89D)l
z>7rNR_c}R#bNMpK3P1xe>p_U5QvB9?Z69*((Io@FK;Wcgi8ZdPSg^PEi7^TUUadK7
z3JzDqsdw17M}?EGTJKkucW~q$YkHeJ`r2M2J_t|PNt9Ai<yHqS)|ap+hrvn(KJuG%
z#+J7f4F&-Bd%u^*d)Ank$daR)n%dtZas@nC^uZY8daJeA$ct2L<Vh<1a$Vw{J<0Lz
zH`ZSzDFtf4DY?va<VoqzrR;(|(*qm4v!{7t3=JR)(wMy9;Bm%R-D1Pn-yy#DV$d;5
zdo7<j+FA+Z5LtLPdE!=Ehb?O9udCWpyKQ8e2+&#wZH{z>kam5=Qu^m=e}BsHM;)1a
zf%y|@5wu2zdnyZo?W6Ju<eZ5lo}<|r22cKIeHUX0<Y8^mLQ7etr}TT<^K&oNj-$QM
zHiv7QwwXLfOa%qWyvePR?2O!I(OFcW{tQpV_iX03IVmG^ajeTuV*6Di-2*ZR+99W4
zzy_ow1xk_^db#$<m!*?G7Ifj4qKDJziEa-P1j}iJKYjI@)Mxa$mDk@l6#;)|_oGI#
zwmCB#gdNt3Wof2l345iDEAz0ME`Wl+paQEQFx49QR)<v|APR_3ia37ia<a1YXX%e9
zz6wv2(!H|kw$zLKgfp`y)rk#7<WvLl|DHyuv}Bdu|3%idO)TMZlO9uc+im0oi8t=@
z<|382f24}#>L%vQ&lQLBG;bfcR5WPr={)2&xGDN7jNDM*s=WAIhh_LIOrfY5Pb#|%
zT%+s#*(8lVFAkpo`s`|>@`Lk({fn@$aMir|>Yz!EsUyF;7iO)S3SBQx)8&;j#bE|D
zVx9$__ERfzW0g6IWwEyV$~>-tPjovd9&(;QE!{?qBlOr$xvw5;YhD%{o`Qo!+N;f$
znnbV?D2N1r!U8A99I&C82VVB12+zG)i+C%<Rr(*#A;MZAfgR%ues<^4Fv%(3dzIvY
zKl~w`^Xh!qz<moXkaQe~|K}UyD%A@4YIb6UPkq!$6p6rld!fDUaEm-T>>QFJcN)6~
zgrO7itcx>Q8<kyrwnJVnO<3M0r?`G<Z_k?#2Rz9oRo#ClD0Nk?V)7ksr%>f@Nl#UJ
zE~fEmYuB%QOz|(FlH#cmPi4MLybkubROgbFQ<deMyxR|`avi4h_gERCkLD9jSbOwj
z6u{~3H$bkse_OxXc?P+LfSI2tqP{bKL%)L2eamX{Xw8)<w%rvh<w{|SmNDmNm!xL|
zDuc8F@r1ew2ce;alDRdegoUURkQg=d+vvx|LGDp)$?88KP|39?urz<etgJ{rUA<UG
z^oP`7C0{)@W(`ay4;yOQt}T~v@&R8yif@;&1{I(0@9g{woaKEErB?9b{tf;V$dRtM
zO(zi93Bkk&Vjz6oa4z_8MNm1Nr~$yv-iyjWX}iFRbDM~7XEZ<>nO*`Rvtq0{J}bN_
z;`A}}TDov%rgdk%cFX8^Syz4ONNqp3M$$FAdkTNw{_fTg%fd{RB|Uel{j-jL=?1}i
zUtM)F>1)|vxr-$zEo3n97N|v>Ei!6hgS@n>@QDQ7@0`7hpK?AZc3V|mhX4k*QkYW4
z)31LM8o?9ZHr0HvA53%UP^b6s+iq)|3hywJyf|{~v488~{FwazZkhHLrKev1J_!!r
z_PX!kv)qve6>xo<%3G9^m5-1#6D<*3XGEil-L^d;Oq(~T_ktn;Rs4*7<V~}ks<1Q3
zV4jS9nPk>P9YKl-0q8>Y5O5M~=egNN1A4d4@w|zNYEwN-3bvHY+_Vz(ikTw1xmDl{
z!*&F)9W4KMUDN>G`F?3Mlf&>=t_`|^Dd5As1Mhj&uxUh5jZEyAipp0o_cz2WgX$M<
zt2TM=54!YvXvrW*COT1}h|5i+r4on7k(UeuM>W`8c=A45{7R#`4<NKK?g7qsexY;F
zsPJIjxyg}@7KtHppS~STnv1&kytL87+ik;A)N)X6)4>cM6o>0Onm##FcLXkX$-M)*
z)6KX2qJPCdGAdL#uiDd7?QlIUnTg~GoTop}^;sil;*d@P@6@O5p|W`n$Cpu;HMbK)
zH${G$iPC!bWYhs!Eq|ELrh|U^-+ooxn=r?iX*J#Q0DE$3c>`VWa~(#$>*unT^uoN$
z=5yS6z4Su*#W0z1a|&t9Y$gw~5VOTBZtEJW`liFNe3V*9jp-nJTl+Qg+u+I4vCO=i
zbC~y%6QTkBKHRUha_lzwx@#vJb&gl|_=^fR+Qu*-pd}|*w%&cFDE+3;ODEWXyZPeW
z7D@*M(HA$)b*b=xL~H2M5!<?3kanm%ue_YfAndxI`uXV+3Y!S)equF49#DF7-kl%m
zVe7fB0ga(fR(;gR%%UsUhtnJ9`Px#dliIb<1T*LcKUVYL1-&d}+YOgFOsRX(s81Qa
zWBUS#h6-y?qKq|R-}=j`wkWRl`U%o7Y0wCMt-;LLb9<&l?kuBjbuY@D3U&|w)<tm|
zEhStym(z9iut}^DNW_!<u<YQI)2F$`+|)t_!ZFTnARf0M`>K1i(@UxeP^|{cR5qeD
zw44EKXe#>X&tNn73Yf)t@IF%aPw4IO3nEZ{TvC1K3w#N&36DMrZq-3!W}PdoUxxeO
zaLOV#sWT~e^a!7+Tn9tHI)nc8fBSBQ%gMpbjFk+L2qzION#<3p0Ti^%sT96~XNcEN
zWKOL7MT<DhJ#FtP&t(w5TLp#3Ij@_4nGN-Q-==`3g*eD~?_h+2m;yM$_W8A5pa5r1
z0J^yPq&xNdmR2t7wZb%)53<c!L60{~+JZgZFt4(%4Ed!+idu^14qoaJTE&}R3fj#?
zJqKyH6;I4fN|QE#%Y+-GN84(XnijXR1i?r&?`n5fMf($bHX~untUs>|C>WWJaP5=a
z#&3otb742(#$IpTzVu0oI3c0XKHyb%0RQQk6>W#(R$+_Jc@%GNG$lGpZRmmPEmknI
zls}`?>q(?2?uGe7xQnh<@rN94Um2#&$dIFytUJHwW8KxTrfPj8ud}m*-LZm>v$G*A
z$xFL)=wOQU<B+`c|9((<BtSK<csw*?aC>^_;5;}S2&r#QFWK{VSS<;S0<wonzltf`
zF*eR3`Xj01TN$($(WZJee6XoODYE)!R^Z@p7<pMZzp6*4x;V6_s7K*OU4|-~c>$H=
zN2b!>51_{8`oPe3Ydz&_uRVE_QxgD6<f4YKXakl&Ms!E|D2EY7G5@bg8)+KX(zoz>
ze$-wg>(okrH-XL5YBEIdwBhWa+RWI@k6rUhNavRU_e?jwl~nU3Mzol*e(Jy?Gab0N
zhr!@g5!;UM<dXqfF6{xD&x?q!ZY}aMk%}WgWY*=WlyhI+M=h=;6CUp+@4oc3ey7;(
z_0Ih3d^#Smm|wVb3mYZx^Qj=-GJD!+SlrXmr=h)YDPS7fY3wXV-@V+JQ1Oqqas9oE
z2vPRLQ6CyNFxwASdP#GKE{0DtV@HAUw_4bb_K*d)*KlK5Nm`u``WRQ*XffDR=CVcS
zIw$rc;^E=x#p&keSwzdHc2<AB7OE9z1jQ`vyUp-l^SgYm-haBRIJaJJG_^1EJvfh$
zxGnu=un0(F3p0Kr{?isTaA|QaS^ATy4bCDemZe9{G)*DR3e(8^GoSl%SY~r7&?*Gu
z>pQISJ4t}$my|VS19VT|5iNwdbD}Ybw0m$sARZCDUeI(r_a`wZGJcTx!p*|D3NK3U
zF7mebsdPNLXm3<wvv|pV%n{wGdE?8E1!$P@Lnx4X{bc9Mezc4g<U{3eu+Z2Cfk7%A
z5G4gd3zqG9zgX((C{&b5CZSMmqBT4FcEJ5I6|-IcY}0Wo!yp&)aZoW>gQANg>-f=0
zFn_VQoU^q8zFnu-Y|W6(gHY3c5Xqg06^|W~qBXP3`zG#F)XN(ltvb=p!FaPbm(nD~
z+!~;}NuI2C%<V<IHj$xzcLN$wYGuBBv3Y*Jw%~AiFf>gL-YXVR@8odvNcftJ@ep4d
zcfbD}if?Rdv!8G8Xx=C_T^rv$3sf2#FU`<-+?Ux&=fa4j{m?w6^3$0zUIcMFbPnBx
ziTe|w7I|@|m`K(iUC*?+RflyLr9M)z&;Y66zr;Bcu%(U$=50TT&KKd=34UYKd?PN5
z%jrIagbD<5FJo}Hs7f#T_Hk`???3&$Ab1Khc;YOW&>qqH>CFH1Psuf;e8CtiixNzV
zC2``x<LTJppEV_>Ik429Ft2^zAYPmcol1_~DUtcb^<+k`0}ercs@TAq6fEs!ARtj-
zwri!Gz$7CQLL*O%1pQW&w>Z7oDGg6ss_`~m<Nz;W%rk2<K=knURXipnb7fO1x3wZ+
zF|^Ic$DJA5x1xFfQ&5}mGq6&9P&bUhXyyxnQzcm<+_p~aEaG3&e)(Bbx@Kpj@m*Dy
zbL5?=kcu#cXb_{tGt|jX%{uu-K0t#Tr5!rR_imm<TG?Rt+V)wv9sNq#h$RNff4YD6
zp{!In#L_ZI8V0XejE0u2R)IAwiVN-MAc<F+Mf5a3tcf&liyQC(xf}6l5q_qSE3z-Y
z)GE7~e*T*7=*BFSpFi6Dq?ph`0S9GsQpdkPwW*e;tOyE=3N5wRrR+iCfb`O-Z+!=0
zmj~EtpX%vaPilQtUxD<EvM)-=2R3vspb&G;d9QFslGpZ8mHXjRg3|o?$Ly?d+CNkb
zM+r=a=9@Y$!!F+Z{FYiEo}`<?4?%`<x_5bmE8s&as(EH+YoBoh4lmxL%5D4BpNNnU
zc&1^fZ85evx10@z92RE<Q07QfnO42|RW}r$u@8L_{Dq3WE`@naMzckFq<GjLwV5Bv
zxsMREMdpmo2qlhlV`LWbt;aK15d!WfZt<T{m#?79L(N}MPBJnsq}-BLqFYc*<FEc!
zZSZ+;^X0N5*%Y*gzr8@tt5;T=rO&A4<Ip!3xo^5xXS{S8)tm&(hiyCU_$#U39vXL7
z%PTI*C>peQ3eThyANn69q)z{Tb3O9fm(1jit;mJd&_epyXJ0t}jH_Jkuc@1j=Ze%^
z{yT71I!!>(iyPI5Y_4~5yA`aIo()v#x#ZuHzLWKo*hG%vC7rk9gSN}qdC+1M4`SRv
zE_aDyEE?K90{ZW8^ni*mE9g~{p<u{tYezeIt1eFnQ(Wq#IB^RpCTQXD#`gQ0EttVw
z7B~z4s*>5&TS6#Vt%PiznXJbC;PHbhLgm730trj&4dDVSV}-rRO$|JuQGe0f!0;ag
zcow~gVFZifRdu1IDZ|kE3La!)tI0qy1V$aLv;GxNblX@PG`fm(VTV;Sismu)_aJg=
z-L5;+gJMWijve+6j<?5EV6WM9RdnY-{K#>YR%IvXQVj;7z)ViK73G^?Q`(*1Xgn(P
z=tGWHC|Sw!F_!{tif-l?de)(oZ@a-UroLbE|H$Oc`P=J~v);$0zG+hlz!jJiZ3IpB
znV8B692m-?-yff6LOng_tO(?dhW5*|F~|1v(-sGL%btV>a_Z&!c@}4+U)RYy>~pr;
ztDbl-8~J`?&o3)&yr7_nn&R;t0OPk)JEPDb(?3b#&j*a^Wgkr``uBFRn>0mSbyz(O
zbQmph&E8N7<7~lyo1CzhX1u0_p|d<$oOI>&g23qnOyTI)Uj>Itommux!@L(Ix^Y&N
z0?}GlnW(;wC~}w&H%Q6pX3_?Pu(8bM$I?$(?;(nzw7AK&S2I#HWicoNe*Ng~Q|jm0
z&DWVJSO5dC;GQopW@)|W0Q)~b_6hOw?0@v8sE!5sdbZmqch)U84S!m{+jvZB@j3c;
z6TT;vw{{YmHELw<mUYD}PNX=?J%3PTY)D3YNNVf_X2PP_cGaB%s)*6XU>4n!e3{dX
z$JxI86R6MOl{KaUQ==4$YG%HFeDnQ9QZix$Fnoce?=2D=da(M+wKH<NGvuJ1yrnI+
zBAOJ<Cdd<QRcMG3<jL+kuZc|2D^>c?qYqEQL>{kHm3Zb{Q!MZ-&8^pUqekkw!PrG{
zZ^Mag`1oiE3k%W%oy?JuVH0t#p01Z8zHLWGL+$gY2mZiiR3s3oY({3D>^cI9W`6;C
zYPs&oO5$TdDlk-)LR?Fgj#_|UAe7C;@CwT-nvdC<(1D#!En}FW7C}`DLRpie%`527
zlK&x_TTTRh{ul?(jDyDBk=)~ps$K>!Mk{8X{F9BE?N+qXQ=$K7&{eA~pT$?wROLm_
z9q_H_t}u;L8k{5g75`SkyL^A~oJG}FAP<v{4ts!fK>VXpFd9&*rQdAv)|qSA-70Wg
z`#lhW+A%*9U<pM|*LvG^SET6tlEfKz8Nse^@%-z@Ciy}50YU-on>dW8=!y8X4BPVX
z3Swg=;mBv9<81ffG?++$zR>#)Qd)IjQy&aK#sYI({a+*3XJntDAkn-_lvxh{eBXgJ
zB<7ZUp^?YqJS0k{r^hus3Y2KB1!-xDoRVAF^@C$%5XGu%6%LA|$H_I6Uw;b;L-HXk
zM$n1N3(CO6j<+rvE^`N^^h#6l4oz`jkUBkYmkl3VH!UWHy-;1<47uN!Oss0e-?6!}
zsdF?*x^B1xl1FH~2}EeDEAX^fd@$B|&63$z`zbpv^cL40kDx2#`(sb<ecHUJUYNf)
zAs!WwU33>w+a}=kLYa}*{=pi0^uF8So}_i>{8enZKSab<-Pok!Cfq`NCB<S61VZb*
z)oUV?OYn_bq{qrXZqk@^ch_lNp-b_G>5}UP)B5_G<^iAN7p|J<@ujhi`!al4WYS=l
zC=$CeZ7UFw<elvWTN9RImP*~)Rl|w}K;!}9L1Z}4I=A4nsCG@!496n*w25HO-6Qte
zDHSAwN~tL(wZA~5Q-HLUVJ5i8-G*u0*WYe|oWRZP4#flD!z$+M*B|gdP(7R?HjZ7H
zfRp1$t_a*Q%WhZZN0dN+z`DOK^Y<F4heu>O3-s^qBWzta)h`5kF};_cpOu2L!m+;s
zAY5$t={4+LiP_lNo&nE;-O`B%Q*Dk|mO#i&fcHZSEhy+E6iy}zz$A4$yzua%I%Y}H
zwFqJE*s>cu1aeN)HQBtiKP%%`;?2g-{sgL-)a$*%WXI)vQMvtsQ3om9_tyN&pW~j`
z{cyje1#zQwkOz<CJ}SbC+CA-R?C`mA6yOrz8&RAnwfB#nRr7c*go?L=rtYwWnU$8;
z&MZ0IjfLeEqQG?Aee`<Vrq{G&Bjr-9`RL9|!AHuLrLqDl+T2hG$RE<5TWA>r@_2@!
z&=2=)i?O2{pz6UjGqBWkmuZkAQBogi@I>0*_IZyFl$-VVukK<6%N7K=1?&c$Z<0=<
zculmK#3_)2uR9?DUy_W!=qGVt0xFG4OU6%R@>bomuh8n5bATWm?^rpnxu(7+=;(e)
zzu`+T-ih&6pcWKMqI6q@V3?qrWN6v9Zzf9I{*=7#q8mNk!>VW=e`a60M`H$yAbRRo
ze8JlC7!8+OS02(rvmSwO0h4&Ig~rGvxp>wr%qNvR>6}03=X82cl@_0^$n^a<JSV5+
z5hE5h8-Et$S>&Xi6eWE+vhT+L;m!SxDPGfSYir$TMz<tWFyh9R=WNw^7-2zzNFZz;
zkYD}>Y7VADR#*6wtEg{ODK(Ur-Qed(?p7boOXkv2s5~0~7|j8o`6D8Jno7o(_(_Dq
zFn8j_p3{81KmS9Yi@E14$}yo#A1ugUro&?&YPtqy`E1DcbDY=W<*__saJd_;ZN#hB
zH&M&?Ic3(C$uN|rP}znI6IGpv@Z;abAnKRD4>=%=Sm3H0EnAYd4;H-12Bo&BsHiZd
zYpGHLQe(MB9^@1=oy=0Gz}}s2nNq~q`x`Q$hH5HbZ}OJYc9qI3NnQlIlv$&JzZ@j(
z>Um|ByTTHb(Tvo}$fmgW&Wy8L_?)tX`9?2icI!<gmU6|BfTq-%6>0%MN|}8zL+y`E
zHIYOlk%+B~Ca95R?H*7A^yL@Ur35LdKkw$8C~q)>qyP$D;&-u$=EjngNJXyV^5s(~
zG|c>C{WXP+h`u40{OM=r+{_?k`k|^ID1=z~M}OZym>}4p1fI2F++vj0Kodd#^t+rz
z1(SaEu6g}IMN@JV->9P0WR6Ru?Y`R8<43Q<M=OG&6En6<txRd$#^R0c-v2+E&ij$7
z$N%H^x)+y>dnJ)|U9vK+88WVGR`$rwmXVRY@5QxBM#;TKB3qG?y)W<VkX5)wHW``O
z-~0LD`!}5HIIrjH^?W?Yj+K3j;``52Qz4kojVlYe_H@UsULMo1XIUF=7Lbu|$h&BY
zZzes<*Im6ABRD_poGr_GZT_s=?p3840J)aW`j$jkW|Zu^<(9hzeSpYK7LDC6gvWIF
zm<AwA1-}feu7@kUTPbN!B-6^?jNM*ZvCc12T<}q)PVcTrLt)PEl0{E@rz2=aH4;Ww
z$+z^uMn0Dp*ET#|qMeZW-<lV#UOa5P=h*(N*wbtrrOe>kqwySKiQ;bD(KQ9433~hI
z>nDGvrW=c^ezyx!sJrkdefFAed^uf?x0(x8l$Qt#%p3@RDr*{Fgkb_!9nNMm>6%h-
zwl`bh)>~g5do)?s9&er=1)g1=UgB$mLDE2F{AMtSpU$K|MRcfidsi_2BmvJE0^-zF
zP|E{-IV2DT|9G@L_S%`7JHDz>QF&XK?wr}F1RBSM<no&e8zuawmIh$x)#9))S0^JQ
zeop{GcXV_Hz)4tnuKa<J>z4(6;axq{`1laeU0@x)Ed<1miG04Ba#uJ^GIH&!-xoZQ
zJsk4=g<_s!+5<5sv??<iGCBLTo3%WCMjIf+n5LarT8!>tg|2LXMs6gQG3VwXfuqlY
zeXw$I^^kcTK}ebT!($!250;Y<Ndeg(0NNaPp^D=PF+40hFNYv_u8!0#vP%y~gWp3z
z9puZNh&7LSM^7Ix*P6loW+ed3k0<D<f>)ERSy<cgbi~{-iSbq6=hNSODm*4>`r_jH
ztFIw1^~xFax+w2s*;vQMM2E4$j+rH^`m6X7{ne9E>s2(JvL^LV@h_!uuNr8YJRx44
zFHcx0G#+!J5{VxQCA)9DeFxb|)Qx8Eoh#2*_3JjtMDlzF)iSTx=J7#Nb3n`N)TH<B
zq>9SBCsFRPAdt{U2z7g{S4B!Bj`uZt^2*MfcM>{kN?Jt6ajyG9xw7tJ?|)|sr_0lI
z{mZ`PUv&E2emB4!50rJ{EmyklI500fNWXP;pEk{eni8yuu*PJ_dn1SHpZP2AGD^g8
zQ5VGfrqc-DhXYX@)abn`1NtBtM+bloo^~FVv}_-4!l;1gW{gpE+K-3mIK`XZ-W_4$
zD*COad~8pou)~c7%i(}8QPRkRKVrK)<8pdu)G(OeE^p>NDlqMok_c3YXT7+bb0#8d
zLvjfAUF23J|8X`o!}!^S<-_%U2o`)|$RqS0sll&|60(m<)OmA_qIAQPbSZ}y2nWev
z&ZZ9!`!`iCSKs05t|`X3M-e3ZP-EfVQ2@amSo=BF5Y!muG8wR>wf}cDDpj&cdiQfj
zks9L*p$iASdU}b`nA~)Sw*`exv|ij9S={{0@6mz~x4E$#Cu*<}obo+2+)#)jN7aSJ
zU9S|IyoS4elgCL@?-wHw0iyZM<5q$Zl<zGk`u=R{A)1oHc6imJ<x_8C1s2No0;<We
zny%7&vZKZ(RR|h2gW_jc0?L^*z4V-pKeA+8xnSkozm(=>agp?MLfFxAwR&?-D9z68
zWER;s^l1~uM}7v2lCz3SayVTEscGG`ljC6INKh5@J7fTM5l6{fkI{FRnDbG6@%{tJ
z-u<p6Qa5&F?>o2b3?cy#MHo~iyA4RaAo-L*Zu6<DwL?&>GGcpL_5g5FydZvjy0Wb{
z{j;|QY6%oXG00K@QW?5snJeyP(YoGtRjXN#@F;%E<^gkIBaMX-8su@a<a@Hq^Sz{J
zZJ#`yKpgTPw2oaXj*Bs8m6LaDX3zHB9MtdL^(X2$t5K6CDp$CqO{S-&Dh^qptW;CF
z@yA}r0d1t|!&ci_Z*P46*3q`(K*_1%o6MLFqMGU1aDMTMyB;fpI^CZH93mDfZ^vB>
zTj4K)I`FlyM?)Vh*i-p@L~*OJwgWT)g&eh9+fgP;aoN=6APg@rC)g|jiNpeE0nv0S
z)By7I^}uGYot~9rVczR6>lYBm%=KXA`|3+6SNA06B@hP2cr}8k8O4HcIvpYAOJ~rT
z>WFmpVQp(fll2@~k<<21U@{-3rR^@sM1n)0+Qo?+99c~}6(MqPvR32WkU*E?$u8@S
z_*gTclKtx##K-BWnv3?^S8C#2M$dzWf>YDk9tp)>1I6BhJx>qGH6?G-EjL@jz*1Jh
zg7tUEM8UW-R<CaZMUVIbQw{^KPB$kp7Z>ly!Byj|*>PcOaT+yfWS@^J0oAbZwY%@O
z<Q5YEj*J%co!>XWI^DanSMIG{SkvHEe(yLbte%U37s%C0g9#R|a^0EW>Q3N}875;e
zA%@Ky9hoSipXQ@e1~h;Rin9zIipYIS_kLGM><1dCC;W$abuKnglmLZ*Q7o(*?rMXa
zjcEjyZ$@csFLsUGgcIg>COu9cJS`36jQA>>ahrFIY>q5wWMIb{UW<#R_avvZUV%<c
zS{QX4<<?5wV5{v>U4Nw2bZuS!BIH-!nSkWf9rd;4lY>!xOx6cFU5z4*Ds82=L;Y@*
zui*_;a3nsOZ-GJK=Y0SzE=6f-!}Ey{wpU#KaYHIQ!x(i&ltzp;Kn?^Vb8=H|>$KQ<
zm-VUfANz22JDH7Zd2ep8y^(xj_7#>7!tfA&0-LVOJwO!^u2qt(t)?deT0p^MEsSyC
zM_{pgB?_Ri^i@ZHwVJ_n+_n7IYbt;Nm`rf-*s0w*3jFmhL$dejQ#z0u4G=(j_O&Z6
zZ>v0es<P6K_uKA>C+ReTBtCg~(MsB$-m3LFBh|bUuqe;r7h|4xU%9Wcz2TmZ5&DdP
zb`rro2#N$|B%R5)$opa-qv6YTAj&*W{@R6=sd_Ya<{{8!xZ6>lh75C9xZgqZ9yOpv
zw7*fOmXcOyl{=JD9lEhkz%snmH2?1E8}OXFu`%Zl?+?e*SGFu*{^{Wk*#01&BH}sI
zhs^OwAZK!f8|_`|h$ixR2i_ag7I}7lax%KQT0sj_WF_~Dt>!3&zo1r(9*G-|OE4YI
zPUY@Zj7=l`vi^;m41ymh=)gM98Krd$je5yPyyn_V(&A_eE>G_S!2#p9P_3|L4wO1d
z$i`oT0G#=gpin2VpQNrp9-Gfj`|-BqXy9zRW{21CWdbDg>q{Sq7KU+pSTzmj=o*>2
zMnloTwa6luOJw7vYU`I^?44Th>J{~nUGKr!(-h(uh4_J-EOt(cdHisBIW=UwBP_E&
zqAP!;x89}qO+@dOyBrf|u>u_#M_KkpYDv`ig79BlmPW8ony@w7h}xi{4ri0*rac2f
z%<CJr*I12!TOpca)&g5F+_G`UUemRoj`v4L{Vg{XZ<ck%8X||&RD~^Q{6RrJcZlJ6
ziZ-GH`C1>d+gcAN>$dh<D?B_}K1{_Y@o*D<`RWFoo|BQ8%R<)Evqvq*Fm-D}zxAf#
zv;2wUg{5}v@0F<_(l$?=b(-8MY5Mt8steJiwn&lYbl4takF}|(r^IIR-ulfbbD%Od
zqGmn&sV3d69R0gIzXch%YtN|(BI{11Oht98UKz3)BY6<i*9^-BUZWPEzO*3YEm9|u
z(VbfA<+fT2BM2gN4`H%W^yqD4cGPL`ZUfP9JfNN-D5l?8$y8DQtIp$|5t+eoR_SLN
zuOMHrw^lIoTqC6J1oD^X+$USxxyRyl$7i8))2z?){W`k(Z)eI@C^*s$&Qk)eQ9z^x
zS#H%Ez+eEMeZ?^WU23_QrA}R*XbnXlbN_1P<p0R4y=Xu{N)@YnL}|z_Z|o}uAV(;m
zB#~1U0#BAqPY70A`I{PM>-U68la@kuh=02?GzOFMcD8^CjutZ^GJ1&)|FseX$N=DS
z`|O~9h6-@dlfC=Uw$VDBfTlUjN%zo@cnMx#Uz;ZuMx06bIZ#>Wk;yMy#3&lzk~V}n
z9d#8&xL1_yCnn%nna;Mo)aaY14MypxX^%bU$OW3uet#IP4@RuQsyrme9W66r(=^#b
zFe}<y&06V3?iC1ULaoHp1AraHLK*FK6V8_Be5kpx=|*x(^mMC#bNcvLc>r-En~qzn
zF_?_@pHUm_%4zwRCd0Ir`&1R(+d1*Kgsbx-NCGp{0RTX>Dm7su=y<miLt-*kh?t<=
zaNSJPj6xh2ZjCNmq3nwj?+nol3gXAW2}?XTR^xM|7A1uj*U5L=e)eM{bm#}pLQzZk
zE7&OXD>)UG16h$~_CLH?nqVrRo!uJTiT1sbpkC~daf9_jtEbgyfQL7}WJr9KJl5k&
zme&<06CXOWs8f|GiW;S+V~2y1oy%%y9pDCVd2Qo*($-l~OIuw4dy9wnB8K;bY47B@
zSkk_UGKL(KD@YN^dZEqPz#5d0msgm#=O1?k6J7&?&_@e@CZ#lAZi7a@q__#Tq;OVq
zL9nTx2|H<kB1#UB8Io1b8R`?m-LQgj^#`>uCfE_76xB#)AF6_?GzHUYvfzUb4M(bG
zUwfNT_Q>AB&H;-y93H{kGLKjMNkL1V5sNY)wgJ+dl`w3P+l?YNb=TZ}R;3^5lsUg)
zW<vzmyFVu`E|}(o&)17`_=S7o{l-8<2n6`+GZnRV&}a)utj^neQv4JC6aKhvvvIqn
zEc1*{+q&4Tye0!7lH(#A)IYnyV+#lUVbQQr>e~0GBU)1^xnXG2>R9fnH)cW6tk?Wz
zU#HmJHllX5@kG^U1_l63GlV4*m%47-I6_chj`@WZKo4*}caRhLUZu9pB9O+ARY`6q
zpn6U}F9-M~s;v22ypzC9f8VMDj?Y-N;rJeLcsqy<ZuY`LD1nl<eQ*Hq{`l|F#oqK#
zn^Id&#=f+!z6ZiV`Y6I`@~m}-MRld~DM9MmYreqxnm~oOWV_Z%BKd4+E6{Nv=0A>M
zt5Id&$5(25G2Ee)?OTDmrBsIK;3Jy{uxH2Ckv>$6FQepF|5gF&GJH+ZdL3bpAyI1p
zwZn_8qkWaglhwJ%OT1D%2qBld)xe<8>I-P^9Ns&dWdITkz#1sIfbw|lxaj$O>nLlg
zrFU`gq?u#~E(E7adOgEW(>jTP{SY!a9VCuvJ~V0aO-JKQo{JcZ5?eQ0>mvR|aMzt3
zT}GbmudcR*M_wL;7mr%IxS*afJqfHF=)#ufOE>`30hE%+xknLU-k;_s14bvBYet$f
zl_%JtA+4D&$ir($Ce5>HvSZ;)IZgpu7BE+F7?$D8NN`uuz3yA_Mp(gp-v#7m)`c;6
zam}jW|M$B-LpK_!YPbf!LB7K0qk9*Xj#i(!+DGnYU40%f=LB_7s<L&GSsUDt<OQwd
zO0t)9a8NV`s{a~H1DS|N#tdJRkA2Swx%WrH|KnQx*|5X_0(#+NpO6RF(+FKg30wn7
zqI9OhxLaOIe7tME<0EAF7c(Gh$kT;oj=F^sM5GiK(0dkCY(=)%iv5qx&}v(`Qx`xw
zQx;kEx&6}27rx8YIW!i^p$*04d=a&i^Y{BZ#u^K`mluMKFO#5H?C|ohR;QBWQDp^4
zeXr3L>a-SJlvBL@WM!=qYHse^AIR*i4z}y7)+prS#rn#=9wLL8>G*w>`;&Y>9>RT@
z(q&FjlyTf0e(}E-hm&ClIV!tIBgR!rfI@JvdR9@&NC@;<s51w2rn}0M)EW^I5vqJ~
z(1uZXix0_VwQrfT+(~iHXgV2=lfyl9(M+24sL+<vz0Q-F+2mJP@mBGD%IxX3*MLaT
zxnYg6gN=8QUeO~AVnRX_bYQ-6yR3Qn81$~(U3i;Va0~>X*i3fG283c{WzApm+j5S~
z09Uy>U*6iPt+O65=8L@c<#kX09bOuS|88IBkpwOW@+-^H_`3=+hS8%~esg@f9Y#HG
zuy)PZen1#g8b$%F{__k`$x8$!Ca}@y2!f+7b=ltR2ghR!QoLs4cwFps$j~VrbV&mX
zzp;$*T`f9$O)!~IBO@V|7DBKzM;hJ`L#(kiBy0?eX2pRh^SL=`0l3Ued(@XNg?cq(
z&*d5!wjy?F>uNo->2-$#RV~$@`vAo{ZWm&G5F0-d*>l60)Y_WKtlYZe{)w#3-C44C
zE}y_iixEiZlQxqdG=>@Od+uybO&MHEwWJu(sjE0tHhHVP8*Yh{tCt+4pr9>p?3P7%
ze@0>Ld0jU@2bgukGfqqrptR^h`3c>v-B%awvFDEE*%E<}h20x;;q`&l3&mLM|4OYa
zpN~Huk9>pTJOi~PL<HuI?vo^%4Iq?`P&u9l;8gabl|K^Jj1MVbzMtD=b-;$H;8lwU
z&VC$GO>tB>D-TZ+`98IJOd`DHko|?+50l0xT<_Ma3d_GnaH|EZZ8bjjnw}1bXnTy0
zcma9_e2m|D0V3ZeoZ)TT!cR`>oXd=Fz{F|>_LD?LzW_$6#^cLgDt$94ud<8fc%u^_
zDHSUH-KcOZIvjq-NTE-~mFr~47MSOVa@PTTTeF|7rB}XMUtcfN3&<3JZMTpn$*aJl
z$^mEp7TcYEH=g!|=jsE$7Z8f%jJLV@^7M4`)7<4>6%|ZdM4OkD89lbJZ$I8k;VOiB
zclT>E;`YY&u9Ib9CYws>>oh!&<NEocNJuz${HI2w8IKFxe*iQ8vzl)P2mAZ)0|SjG
z^5;$>LjLPtq|}QlY<~Gl_fwU`oNct6Sy>J{r{7lbTWG9sW6A{fGqVf$F1t-P5?nf9
z{Km3qhkRJh`IPV>1_+RfcH@rif^l+4U8cg_WQ-oXbrJ4F$Hz7_HkBC_`c>adD6#Cb
z7Fc?Y_+FZ)cIFx&B=V6ZS(+knYtDAg*0wIdlyg2#!Dw!~L6pZoPCMO7^gmup_TvEC
z<Kk^?lZJFOP=bo!?k&t@-)E0jHXNXM4US&^CXP{2yynkLzd=hTgOpL-gaUn@`*1qW
zQS=PNPcgo{gBfPq+*Y8Hi;ATmtt@a8W||pS<D}CD#djzZ2evz9Uy^oTcZ{-u)CPIq
zBOm4Wb>TYEgA*0@)Ypu4U$V$|4L^;HisJ7mLqflpWa=N2wmX}oSlnDDJ+lRJ6-bWE
zKQT`?gEVQiq$VX!tG<S28L=RqTe{`;@9#4ueSO9zeD>3j)NEw(ZQuHxe!tO+<caCo
zu)BB1bF$v}1QlY_aiR50y4R&(!RGbXS)(#D6>!fU%tGGMWadpET8@>8qSicLp=Lbx
z&t6pV+Ja#qAHuWmR=@$@J6V~r2CUY88bO#~tMjTp1GvV|_!wFMgcsNYKpy;j@!Rq7
zqWrRovb*-UwQL59y!0u&1mN3v%a_0yU*FvcV@hM<T9GC5>zgCmE`~5y#^pW>t9XIK
z!k3Fk(7pM>ZTpoPz>6`m(TteW`fIUllSP$_OG{GC#V7w_zIiYHG2scrU2SLPeHCd_
zMvn?pp|e(j^2o2ruvm>jVsk={z5w6PX(||>r0xx#frY)mv$m^MZ_L#WrX`}SHiDo4
zrp339jkhnB;jL20MXbm2-HgEphp}al7w`pM#_g)}aHs#cZp~nIl)(8tAHQ-;5aJtd
z!$uL(jwHW8=5EIr9<@(|b|T4NUG6%|N?8IDR$R`+LzE<ilqNu#upi1)S#b=W8!RZ|
zrQGjwpr+g!-*2y$AX<)%g1g4_HONlKH!yWO<%uojXU#UHnADmGc@@!p$-m1Hmc>Te
zo;Y9;i=OM<t4NzqJJYR)5_bM`#}|QC0fOlC91%+BPq&N^CKe7YVeisTWPFu_<WNmI
z6HCNoR5D<mYT|Z1^rl!W3er1r<5<~pw9-qE#fd6G&?o`&G|vq9#D74JFHd=I<_d)W
zn{p%6UL6a;NP&zwBmSIK7JMjY*kqwv>q*w=Xj&&Tg-KZ{?XV@m0H%i_kcTi99+v?V
zTuVUDa2X0kN2lA|TmW}Ee+hYE@V0OL#2Q};lKl?|R{!wJ<-6|jnez7Z%lCOh2=IKd
ztR?{{2h%Zqz_aoyTQbwha-(^BvWDh0k}$=WlU#!-an2R4+}h?Y9o+@+7i$N;UQheT
zVg3PH0F4v4A~CS(XlnBYv$z4UaN>C&5}`vfVP#b%L-4IT7JoML5HfGyW`PYPTiqge
z$<c!FcB~?A3hCrHRLAFQ&|jW>!Sf~@@1^I)-?hcvwL2H~N$&yFF?o<g?R^nsiS{&{
z0V4q?ADQl#U2g~4Q$_r|ATRbm&FSg)udnkNQ>wpuRsHU%9WNJ_bssWs1b8Lw4D8Tp
zSd*37899pJJ9*pdCvm7X9)jvhsS499CvUUqfGt~<w-$&L!1Ct|vGOySu9c(K0!~SC
zp)L-xHUG<-PA=!J$Y9Zn$;rdnnf42cdQSX)8i<FY<vO0_6^IY`4u9_zNER4^_k9IQ
z#S|!nf(ga1+TiM@oEw`HR?}x|+d!;dG+L0EgQ|e7#(p;LTXDPKp}yp26JedFT_R}W
zPf?-`iN^SEELXQ(I|$ftJ&7TvKqu3R3LK~`8tWe5raJikWi2Afle{T@+i|msW|3?^
z7_3>Mich5Nm+qJBm6nhg1PUBho;KIqAwSBtIgf=f9y6=w;gQO1IjdzE(lgG_hm@({
zOwY=SUfRoQ3fQaiGtM00w_k(UFw)~fCuZ_GV847q?6Gn?osf2_;HoH%NN3U|+yUcb
zz;|*B_MLTOB_pnaE;|MVco-PL5QdZbA`{o35zgwrJ2w!>?QkNEF4}z5*;~P-uIEd4
zjS)l*rR$)s5l)Vmr=TFPf33vVk;@L@VW$UsCnaq;%5U*ymVpJ^fDJ870c!!hrkHi8
zKftco*`nAjG&0i5)4S!cVs0Xgn}kRZ^36KxHVwr}NUMSSeX~m3GFNMN$XD}O$%YR^
z5wcV~H(Zz#g@~eB8)WFAdxnKu#Exy`<zC$tStoOGF}ZnWBa>p4BVo#DY%IBG7n}rn
zsa;2@!7xNZ#7+BD^I?L{KGRTZpZD&#_i5c@^v5;5GC2Km15Sv#pImdTy_@>xLP8I7
zY`!G&QW+l<zsM1#rEME(3Hz~Bnb0Hh#t=%>;i;y|*+to9G0^i1`1t$rp^1DAOu!{g
zUbIr<>1+)>=SBub-<-evItAcFqWEs2`HHjh_;m5)bYta`Him3>=o<}D7&;VEJ6lm0
za?Vl@csd<wdrW?}xC3-biUzKKkuwac4<NCZzNc=s@!qMkf~Q*w{}uUrq8K)}ZFVO%
z6*V~7;ypOdlMwZ5FwSz5cH~$$Kz;Tx9S?``{oR=CCThZI>*h51xbFB&cJG;-Gj|rK
znR=!_fDp+445KxgUvM=nbFQ{9AFLo6&-tTs4EQT%Iw63&;!vjuDd5Md3i@fltV$9K
z`{!I2e66x%)IpUBH&8PXLdA@<o2Sr*Rn;l^vMn;G#^upgfE<z0Rk&ZRkKF>rVXU_l
z3C%h!8QY=U1s}XBjm;+sn<gHa9+?96zwO&uT!F<K7})_p9<%bh6-9UEw2@Q3W}SvB
zYKL6?lG!F|pU=gEBwBzJ5F97P!&BCPq(<5S<$eYyiUu&6%Irg&6eSPalNU>~*@!*&
z`b2a(Zjgp$upNE|xfK2J%QEfK$6l&J0(}u{@eddf?To*qDX4fTgM+ajl;|EU_5OE{
zYc|yrqsidUgM`a)unFONBtf9AB&z-~0GYnt!p->?)a@f8;m!b9U@4fM-aO>K+V^|A
zdUbkrxw)(_9`JZ3;s7>+ixplrSJ~HvN(@Z5K5+z|XqF{QoFknS5L)NYa>xg<bv^m-
zkmE)xAuhjnS=CIbrOd#HKaO-YtT9F8;yvbe1U1Z+Oa7p_l|(MnU2G10@?>`*Q1gJ3
zwHOjL0Iw(o^E6$VBpK>iSJocziYO*m$Z;BT;kr9-GQ`~AdS(Pj-4N{c=$V;;DqcN$
z@Lza1-ElO8AbBCmAW6}l|64_;5z_GN^4~(p$Snti=Z&P!Uc@5R7zhfE=DNnIhr`g|
zB$lKT{{}DF<#EiFalR45r^3DycOg}(eJ^#tVIAO#@BTM$QenxL(%GU~zI?x;8#5ci
zz}T6d^2+UX@=x&UYS`b*%<Sf*4uD1-mz^oLpSJ?nSg{w@h!YEkzjlH{?O$w+RMQ2q
zSk#^R)G6gmoQ?X<X4GWFmLbb9Y>2XawwWx7;RL-w(ciNEcfH#u=c)dy--b3<%tib|
z&9V%?jJWP9eX3{F#c?ESSzW2DXp;j>jr+FzLm|XlZ4eMm<DOsh92(tJ&KNav7&`r~
ztu->FvDG@;a*V<X2}}4^RXVCqR<w0G5{|uYRq3y-Fn(CmFS(J%<3q=e=}c<`1;tqH
zljnX(l`L%EFsmxnA1I5Daor~C$`|WdYJ7l%h_z$W9iC#nj@V5ml;hgy{(kG62@l+A
z&)~SSofT(@!`=AmS1jQp-N~Nl{Q(lCeaAG+JM2?=Se@mx(FWNl|9GM=sa34G@>Q@T
z?u9n|33%-n2zc+d2lN!P76CC%YqS7lRm-KW*jVA%(geqi)^gv18Xap-dV0`UZTBfH
zZi{ujte7yqKdCllJm@@nKU1G<-%_LxosOA_w%X-blP0P#*^PmSatZZjf%X#RzSYS6
z#rxd#;{!Np`bOh)T<|I02b<s$a>4CNBs&POsFwD|WRg#g9}}kaMKrGt`yIuc(B-~(
zxcpotBBKav0425DQ_K1QA14~X6~Er$m2}$q<XMM&_~GNku|(Y2*YIKG*G02B5tj+8
zP$pjcI9n+)ONO=O`0?g%<Gc(X>7{}CM>JU%^01~b?#1=}{Y7zo@u8cX45{HE+V<6r
z9>p$9gaUcjWQ>yE6m0@_+XYC=cEmuqb)qJvgJXLE2BgwTq2I&LMZ1K{j@i17zSZXI
z3=s^%qQlxpzxqa=aIJe`N;e>TM-;nmLk<s~86@ir{!FLn93T#WU;_!!ypkP#Bofpy
zAH9yD|26CYSho4JkMD-`NP|}HpJ*W*7K{qTtZ5M#&EOjf-s69yXNoccthl~UliV7n
z`;5#JGNGwR09$=tT*|_VeL;3z<zj&M-D_{A`W^Zc^&ftGSaBGx{CEEL>GKD(FY}nA
z+Vf`vH#_4RSDsu0YI8})dH9n-KM1>Tmh}(5cl?UVBK8L;D%+~4v?{l3SFlR~4S@`=
z;LqRzn6>iZtk`P8t>8SWHKg%H0oWKDY-yH3i#k)byqivnZgRNG(NQOHRJoGC%vG8k
zx70(m04!IwUfzFUe`2EF-tTMKmpK#{g)Z`1y+_@=bx)!#JYx3vLuR8zR#dr$8erEl
zi3Udzrlwvw=5;~&3ld)DT=ws;nnxJ>qoav2Oand6L)0jfR_(^rgs;N5Q7Po+f+cWx
zlry7fY{PCvVQNi(h@23RBNrM8JGEzGHwK^%ux~_w_AC3`q=c1`#v4Fwo_`<b2E)B=
zn!XvpA1F1KX9?UmMY`^&nQGz4R+vFfW(gvtU^VDM^oHQVnZE!SkQ3q<qM>8Q1(U@1
z4vY$n4A3JxF!KjQrAGtISzH^LipLV?TYm)FeINk^!4Ii%fZ>>d%&x7%O%5S+d_aoF
zv`ns;wn{+)?`7;2F=>GZfsV#bQRCg{7%GbutONXmc3Q&Z;K<QN(WuWmO{f31SWZE?
zj=H$|n?g`m7hGX^OUi_$&FL4BT#Q650N15~)L<N7$y+fkyz7adUgd66jJV7qTwkEG
zkzowre%-~i;1%`D&n1N5$kLNZ+dL>;OkOA(;qD)lIF=x!-hA|rIJH+-muu(i;<_W)
zEec5cST3zdp#`9z8j?rRYV#;29<KT1?1fO)Hwtk-Zt{|i3+z2*Pv{vK#Bh!$1?PSE
zPnim5+R>OUEz>unz-YiIgM0@zwnxCJdZQ7JRqA``pPVf;9)Pp5nqcS(>#~V}UwYFa
z6%TJ?e_C~#h<(8oxVs{&u-(c>{Q+&WM-ykdo{T?>{$neww0d*c2`Y_dzg8tbHC)?d
zC3ILj-5R#Z_SxYUl~4hX1BNN#;Dh$6Woh)Q5H&uLJ3gKpDdJ}-wvRC(%`;il^C3JX
z=r28?!XCHnUk*=Xmmho<d4?$KmvFrL_U7F4#>K^njx~t2%CmDaB7K?Z-w!J5Kz7yV
zUp`Q|g0mX3?tz)&3P9AqM?L5JzgyW3ky}(nVnAHi`kMQ;bJ=#Jf+z+*X?E8$f_x(A
zsg6b1wJW1~RFe`DUaAkhVc)H;_3$Q>BWgQH5eGW|AY%Y^%C=gdEY8&;xG6ngLAcpO
z-qQ-s|4|a?CIq&d`c$#G-@iGFcZTspxLIU9vmcylarXkE0KbYfvG?NbAO_o#4uwpv
z?h0R>Z^$;#TmS2MjHyXt>@{gsL`$6y{rzbMo*!*cK#!AnIh>{r92#vTqO@8g+r2+H
z2(!%6ZKN0@AjKziOj*B7PTLh_n1r!~7T-UrBYm?NW&sIO?2@R6Tojdj*29u$CtttW
z(mD^!`C-UC?{}?NCX=_Ij&$WB5Ey?FWk(o$nLAcaMUgBO;T{}NFU%klT@&i}4t-n7
zH~way^XSuB0Q0mBYw(}MvHwo@zSMET9*ScB9UL&bJgi_mJ~q1JbE^lafqZqx`kk1{
z`LX^xOk{KOjKBJ=X3NHZ8s`6|&E7qJ!?1vKJA|<2-ahPlK#aHr<ADOD1U><!I`DwB
z1QaV0o<Mwxnb>n1ik8YklA?ejC7*?|^W3`Q%UX0QBe^SC=FMFx?)pfFWr9yMf9nt=
zQWb}lLTYu^xa^`PWxYtyc6hm+5SkD5TXjQOq($w05T}}bqacV^^_wTM?XkBDOG5I;
z_QAZ&+6giwWA<BB-q-GQ3T-k{pD`LaxFS#SgEt?;$WBFa@V?LgIe!We$9#vfZahO0
zeSATzc-sjctVnrQJ<>fa`=@a_fIOVMl9DW$;xo!<WD(MFRcNP<si5pLin56D0yQRY
z9v*eB;%#%i@Tyc@+EkJ`p*}$y__q@3sWcv<n`jXzs>g^>q%L!Hp}uvse^NYPJ~Q#A
zUI9}@CDu9VIktr}(i*PF=3(eo>Rg~8b{PG`!bTZl#cuht`?_Ho!!XK9Ip3e4iT7V;
zRDx(?RdV$<4}B-CnjVvPA(O!kPRyTx44qzrIxea%Op$F*pNF%#%Qjwn_}{4fQDxmq
z#UO4}XWCexDq^iBrDD&>S@Z`ppj{EGt&Vxl%um4tzdO9COl~ccCW=_>6NXWlm9e<_
zUus205p?<@v(E{ise!i~emo&)1D@(r#!~y9ez?A?E+E=Y5uw*q@+Mth(`Let=9SK)
zq^w1I)BCSQrSu;Iuy?5jcwYBmq-6f=BWS{#fP!^b(sDUT_rY=-+H#fgzj~e=#)n{l
zIvpl@2a3@WP(}OOnhu3Ar}0x;hbMdg`u9gyKh0g77+XBp=cFNIzM*}EvNLDd3$S_!
z+ia;RF5hf~zQt+OBdfEAoOV^7vSh2j88~QoEVkb6*FqZj<ByruD{;06^<c0Ae22tF
z32FksRH>c)`^%H>TF(#2ab!gJuTL18@y-#BYFwT4Nn8Tq^Awl;>#nX}pD6i%r%`EB
zhtXmMeQDmH*F`vC@k&3E{{7>873Lq=_67&^dS>wFx^MQkE_v42#IJPp`#Jjt(I)qO
zYiMSGJi1(zhVcK!q*gYv<&rY>PQ(pGi!z98eOW}oL*@4=Yn`X8T;IUzJ&;sP)GVKM
zUj_X>GH+pxR2gx%x}|3NkhX^}wh;b;`fFuGuz(YId?Y?O-=;;zv#Hje<$h>p+$R<Z
zb!)Jmz|5m3KUzJS##_BZD6J+ZOls)ovgaCuUM@f(T_#5BR((d3WRvQNzoomxmaKy7
z&Akeyo3es<uBo%p{eJPGy47rquy!plPBJ=(-re0%;J~sp;i>Ate7kmR!&niMZp66C
zRXBUgmvo{vU1=$q-v6|~>&fF<yI(RK4)w>q-!-GziWSS66Kz2@n)4oER{{&mUr8-t
z{rMU=S39q1w}{~Sor<<r>(<ur+d?)<Rlf{g81R8;!G2F(Q09HS<`dG2nXQDbOc3!~
z;MIKbn?Ko0c~DNSa-X)Rbk02Ks>j1!rRfPd0p=ObR*q$F((Vh(7qXFNy^4B_HXFsj
z+K4D(4U)=mDD($R>C~0{omw!4Sb4wws~w$nH}U1>Wn-NzmDHpWOPb<W2n=3vO;AF+
z7^<l&n;DG^sSoq58BW3M765W3Pz@1jg84`H8(t6sf~mAAd6uFS*<p-242p|<^Y;e*
zcmM<*4!Go%)Rppw=FB^H^%H4C32MQtnmq7?c2)wdL&k|rJo}%zzvs)#-ECLre;2#=
z<C~@r5s0EkUGSjf`8;eYf-HfM#`SMC?$k+q%RVz<QZ!n)(#3fEv*K2NzPrS~lBM{z
zc0>#5wc6Aa#JKqd(e61X`n%@1$(aWrBLI@d!c;DX$nht+c5%Kaj-PC~D?`tSq}CO2
z(6ef{pOWX7em9W^bR{ztNBj9Wcmcf56muPc)P)4)k<6XIr34K<Pq_F~f;bhR`>WDT
zf#Q(qj;YK>%+r(sK7s=MfaFC_uKy5x4{Q#8vd*MU<r;;1>Q!9qVpGAmRSqzGcw+t{
z9Z?i&XQmZ1Q4uuPL^@I;YWS3yCj6L7(bfjvPra>3=QhcD7C9X>ZtuR(`gWV8(mDg+
z+8MX)Wl12o&Uv?_$bJlHk|AXuRhrP=r+Hs^tbi58P;BxaGd5Pzhqpa`Q9D>;R3((;
zRgeO!=&lL5%K=PIRdf@J)!6uwqemw4D2M;sz1;r#a`r58JP>3?*FRgnwRO0BIx22!
z8|bNrGD*e$Oq1E2@|?W$&n|h~8RU|`fU~w>A;X_MoqznfBs^CJOwASoQkUK^Kunx|
z)oz7j>d5|~K);wT9|_kF4&HEg&WIEp(xK?A>nj89iIy}i`@4X3en}gO`dZ!$@Q665
zT<L8i=p@7Gx)(;Ig{6A^T@xkHFQ*tbxbjSBVWJ)lm@4I&8LeZTPyAHxA->1O8Knu7
zlFsWiY%LWV5AE$;gn(=9Hy{*ojW&Vp=F*Os=gy)!M#j<FfB@BN99*QqM#f161J+FB
zYx)`Ke)!>V)y&<GkGo<i6a6DU>QGRu<zWe*%UA4IlI}kiDAb{f_gdkKH3-gFmbcB`
z9b!fo$>($1I+L5Jbttq3BrS18+WsVF56&zhqd33d0-RT;ax07lLUf?axsta402NGQ
zwExSVTG?L3c?LVOPEY0XsPo<B!NrOH)W(A-FHK^SNCKsS*Cg3Bt#Wxfx%`!$6PCs$
zH@R7nvcQ!TuJCr!xli)y<jpxfqk|9y0LN!^xh#%3xzwMlo0w>ll2i&0#<Xlr_7D08
z7wg<~^D33Kob=e4e0I`!b@=IOe-y*m*4kQI`yu&bA5RT$uMQXo@sGFOl2^DraMD!5
z-$_a~zLC&??8Pk$wMa*rH9Xl(xDM6(etRA_r8FI*9!rrDXU~7gMGoAz^I6<08E%EZ
z;MX;sVBRaKu}B`6qbf6kg?~kwa(!bXk~MG&X|`|A9}I5H&U6Rus43+!V$5FVS8i*R
zPJ9{bQ-|1h6^U638@LU9`cwx5BuvH6PKga5*iKpyFB-|(Kz)qY^8at84<{_^C7EFK
zoE0AIixVa?12AVGvmb{GlmXI*n9khc4!61IU?#JLX}lK^UGH&kHZHrN6ZwaVWDbmu
z0^<XI)dt-2EE>ZE*i3rfdq*y6xY>}u!u)NX=8QI%@0>+8gtr{6PF`pfbo0qlzNe&1
zVV&a4#L`@Y$)`C4(1}=QH0IWhl1{$9LF4AW$)53N|6Xem7|}2VG1&|?`j_?p5KVpg
zkcTe`GMUrk=8<-qXK9`>N+Ml!HjR=ZY|Jvy^WY9Bf(`@!&)K_zQD$G+-@B#8zU-qa
z1Wf;w#TksKiV49nP#2Dj@TNAG8s;1I1riU-Th`ZHvgo>&UVx+ZWb!EtU`}WR^4_By
zyyXKSv`7sTqRgG&loA(@m<70g3=lFghES+7uPaCnXkz6uqmZ<~U?;o#Kc#DA_4_#-
zY4`0IFgkamJLBUL_>^?uf9Ttv`2gN$3NJ&;w_$n!RL=E<xh0bj>f3ghgx5@qhive>
zEu9iBzLbqI#<hX~03Wy=7xI($J8@R`b=6<0We`PblYql(^9TkuX^ev2E7`)RQPJQ1
zliXLAqi!}T5mx|?h2*3jIyWyd4qQQsU&Op6=4j$*HjEqJANzwkIYemub_gq`M>Pdd
z#0-EJ8uhLFqr1AhwJh#De>yGXyeEQIPp)SqHHA(-^PDW{?;l-0>0d74j*JKm-LeU&
zw+J4fc|J&U#MiO<j(S+)Nvwr=le1V{7+m#BGcK_N?*#AqTxlN~*@64VukB~BJZgIj
zg8*_xGpk*+xrA}0#ecCZRV&;UGc}C=U|Jg}r<(1rSnTxORl(De{O}vH8@^wO!Wl!Q
zIHxj)hk@hvNl3<hNCKLiIlQ)E>`7MC^!LC=kh84zvCot|H-*R>aFHy@5B3$u%5mAe
z3KR^Z6Wz+1+$^fvL_whgQ$GN;i#X<uUQn<?5DPv1UD|us=B|6m9<9f9Zb&Ffw?m^-
zWA%ep&zhR{PqzK1CIW0GG;nxli#~M1_J?eZf2}(kwx5RZtJ%ssi!M+UTXoh1uJ_9z
zw9!&70Tl(IkKHhu82$3LcI)fg+9GI+M8#FZs#S}Zh#g$ZQfieTn&!eH{}!W<$|>P3
zw2STmo*K}m8tF~f=5BM%M*KS<1u-%*W?s<Co&82oKwJ|X31RFPjXR${k>5o%b!jw#
z$=ImQg&#LkOHmR6rYJizI>4)8uUMQxM(o55FVh8w>n3>vwA5XYs)Wj9S9y^qd{`g9
z$%ZVycYgN#+drx8n4G`umSCPuH@&S}UTB{6d?*S|8(H(>%GyGs6@YO-_sU{NT*!|^
zDJLloz8OV99&tWKWHsLe^19vWpjP88U0NAH-m^HNm=F3lB^H)i&L?8@<*0d{TGE(f
zr_-f5KlY@n+5dQ7Z5Vw*D~1zP2d_ji0P^LLG<HiUEuzU9?<-0>EM|19b9F|HNY4QR
z--pobzfMGvA*hjL%it+S@AwW6JNTwVf=xW4r#4dV-4)rVFv!a>A3A;3+Sd9Q;A1nn
zf1=Zs`msypk5CZ`02^-<35WtXiFH};3o@ZB^|R5|#}_BdrdHgO6OX-8RFsr2PxlUj
zLn3}zu`4)))`u-&L~jwI^%;ZU_oe@=>Aj+3>)st_3UTD)VL_{ES0-gsN9%-jN`K6i
z0qQ)7Szk-hih09A4KNORd(hWwL)-;OnH?%C@~VYwRP$_=rf@^&Q!D>dY;ax!$ZCp_
z?h4cZkUcK9zcAiLnXyN}0krb-kTqVLv>1UP2ym8g=+l6CIcXKvu#NB6H61;RCe#^u
zR3Ok@+ry{zvw<=)*{;s&=E2;dn*}KgY>oH9+VS7VH~=5gM>*f^IQ;q%cay{tB=IJX
z;<HtKK@45pLVmBkT_JIE-CtpjL?Wro%)R^WVe`Cf!U*38_nln^vie7~96qd`beZJ3
zg_|!6>ea`j!v;#W`JkX4r3b3>Ms=g1t<!ozWQ+P{BeTna_xsuKY1Ma$?b{M@3X*n6
z{pi1qp0^L>V7XN`WvfkLDB3*b#&lohx8IM0cDS~;cfQ9;=+J*0`y%38aZPHcWzMV4
zmK^3NE44LSe_NBM9Q*jh<%Vj=Zw+(!Sk1&OK7UIL9>z-7IgY-e%@!@RSP2rDL7)tN
zfPKD<0OQ<U9IqF9fhdS$nqQ?sA8Wq&#P$u8X9-2C9Zr06QU&Ks7+ur9q4hU^JMAq+
zXldE|TfmGjAN4fV;%WI7DG2`r=Y501HY>F4(`NpHpLO->xcI4K!@P>0st*+0(c~+C
zy?ZW!3Fk$p5Efpf+1+yqV3D#xlD}l;ne{onh1{<0Gym7{sgw7O`Zh3Q=7+!d{YUz!
z()eLRi|;odBt^?zloCM)>wh)Ww})0!=GvOVFv3<65GL7UDAlpo<*0ZY90B!QeVu?b
z5GtB*wqWtw_mKx*V7k=QUpom!N5QJbbvr<q_VI@m9zvTZ;SDzSJCL7ZXQQ9=Z6nF(
zk^beqnyosCxa=s4pX>otC6fakCiz(Z6h}a=+vxWCS;WWiRR3~&2ds}&6-NYjSKqJm
z!HyO(QtDnX(oe%)gj#b6&=v3NU?gszI}4TbZ%hmU@Dzr>b)$X4s3Tr|P^^m9TNuC$
znn=ZF3wj2(kFv0*8WeZF(JrHe4U%NAl!*<m7(t8=kdP6I0ra4kr()Y<Tknd17{K?w
z7v{+Vn9<p(*|W6^pE(|Y*|C4S>Am-jP$jo-vIT$YO)P2;FL~Sz=@^e^FAj?^m(J~4
zHaAU;<Gp@TS!?}a(OaAUB|PWVWMGdGF4<>Gdmt24;wKuLXgV(7+RN~D`{l+hYYO=Q
z7%LU>j*Sad*3B5Iym;o7#)(6dNn;!GV}FRY-Tv!oe{U{uc#(Pe_BX2SI5HJfgq#U)
zMOb^EZ7$z}by+{OPussxvzX}LyapJv%(zb;0xOVAlK~YaqmMQv<F;(L-`Qd!&(~MQ
zZCkCyCw*qTJrC@8+xUEf+5vo;tXNXMkb0f`Gg>UF1SX<Qj5fJlUby_jFa=<Mx@Ay!
zIojE-$u9wfmGH!8%VS?arNm$GbvsUlrKIWF8h3jQD8T_9U*ckm(hZJLbQGforhBD)
zh*-{_6r(%6AOL~iN7@a~fN{z43<~Ar)_JD*;H||r3;<<W_@%i-At2x<*H`v&(Z$JN
z!e04hFNNQ4m;J_Q2!z;Moul=n%S1?j-1HP=sN;5ND9`^_d3=3WlL$<J95l-mlO6@z
zmHYFwD+U&>UX@?f&xq}aGX`<%Jyhwnd=;6Zky>z=oM6;P>WrbM#JmWwa4#E}3GM&7
z?QQvO?ntLNt-*~^aY=k>-sM|Sy0b-X#RsO;ps<k8i}TS%ah33fmZRG2<b<A{7N2+W
zn3FFSnolmc<U;PGNj+uMzpGf+Q=O8?7Mlp;c_l^z@Fo|BDb-gJE&?(4f&tMAY17jS
z*>&`&C9o{4_T*JmCMGw-?^ai5hEuf8>)JQ7*UIkb*vB{BPW%kN&Lc&Us^R{O#LG&N
z$YxPkmEO*3?~>cC0RIdJFq6CNeATSm6`PwiE!JnVaUmx5Fd+njIDnwTx@7hKQ5zrh
zLvX@eD%~UKWZu!)69-IX2ic{utQNs#Uqxc>cPcV_YS%NnAQO=6?xf$czZ*S_C(CXy
zqI^;!b`TLZ0d#}%4=MsGWH97w47<>q-X;5_&Zf!97kKGvkdo@@l;zpp@{0y3GFF#7
zmdU*v;#Q9Qw?|G}bQ0!^$!EG~bJk5#Acp*Ae1ZyXX;F-bxH!Rgb6@@)T^(whsl18*
zQVlW$*5JWkK{1%k{==Qi#s1~x&i>Al&8GeMv%|>OHC`nF6%TKWOu$hOEGA_JD%L|D
zX6pPlk71DBObI3q@(v(Hkjz*Z>>46E8wNp0rd1i(wbiZ;4W+24L_{<vamf-zRG!bF
zvcty<^=@ymrV{v=^a<R>@k1jKQ?+<v$fiI~?q`9-OpK(Tu-0`-?b^KVkrAF?^lu0y
zUy0<tfj;w99vMLhM-lYvn%p8&{9HJ$n`I$96oxOB2432`&;r7fpnFw%<AiKV98eR8
zLc<u@T{_?PW>O#t7GxBI%MXTqD|Wo><Kq#T`zJhznh9TUz_K3O)bYjCnRYG(M`4tX
zd-e!Ewgz+yO)7Z_`w{%4tLWEe<@Y~0Wtq(-hKiR(<qbP6B+D|J^Cfck!k{0y>7qb)
zzt;2Y;ICs7sl8@r?^fvO_KJuzN$(iHPy<pBvuwKiB+c{Q_~}NuF*`R^G4v5@UBIu2
zB4#)DA+k_4{ABN7`ETeiGGU`~@X!H&zZPTyRKVxff?_e1ofk!ik!0TLoYzhp=IXME
zv6Uk6<Eo#?hYwS%Ztr5X6A*mb^<%2I^~!VI&C7QYP<K(7oqL_t^*E!V$^dgSPKIIy
zugQMa)5G;m{i}cLJ69)jZI`=b;o^i<?w*7w|Dq?HHgylD76g_^eB+(|EpaMd3h_ew
zQG3#B-Rl_G{?oT6gfW9)T%$*{1vU8Y=SYa(iv`vW8lt*WDa7b;-SzbQ|4kV7tF;C)
zU8MA?Ufy^mvG-SV??piy-r}<9-fshMHJR!lIbi0ROX14=29o0$>k0gb2ve1{&Llvf
z-lRYoYV{1XSVB#@;&Ks$NAutCY?wZYLy8DAOvc!k39748CGm<uI-p6zpjB!4@_&|^
z%N_alUTT^GTi9alrYJHBKr@d5jGArF{q3_y{IlJ%rLRAAx?F2Z6JjxOwogMmTdS{|
zXk@r{xbGHGuVU5qQdl8yG}(M-XPcXgYnyvkw`7nGKWSfV(ZfspGGY0j+VO<oz+p;^
z7lxwQHjOs5zM6iL%i`Me9;0jf*^kDQ022JEj~>7b(~@%U$8qv9rxme4ywA2amrqY7
zZQpTmlM~?2DJDrt5;~mNJZLZ!4%cU*U<K0=C*Jfb$#8N2RR0_0`zfdm!M&_@D_FY4
zZJZ15iK69tI5UW*`B}E4bA5)?%{sYx0|S_J=6Us1^~rNfr0Nl_@YoBo?Osm%ltsc?
ziW>{1i6BU3>8_eW^^A>=ePlL4V9|<RLeDTlQR;4j%RjC5XYTM~2N~d5%uOi|r=WO~
z%K7}C*+06q7r!Kx8I1FVwYG?S-*mbJ8u&aYgR@8dI^qTMa;`8`CXP2zyI3n+Vj@GS
zb7wx(t$Q6os57Vcl6*}$9S=rCIcVzk*S9ZQCj#!h>n*VH{uJiugyw(mZ=N|p3vGMY
zcXoB6+U{(@W0$veS3u0@8cp5AQDg!7p(q4-bacLaGRaLQ?*08s{t<d9=${1NTL!{U
zirPG~cZ2r(b`_l#2WjpV+v~y>FdUqNLhj8ij|Qa0snrjW`_Jrrg1aa{fTf>w6cV&j
zQjN~&?=ur^`yX<vZ?)Uy4A?~{PEr_g6>NW(OH)vRHWxl{YRt?I!Ku<+J4ONcg^3V5
zF5SV<6uPdc-F$#BqK_Z9#6wi4!;#QR6b3jNbEt;grYN~AD-k+9&B}{G2o7RF{6pQZ
z+Gu}qJJr4Z=V|mP=hyw_1T(hSsQmb{5650%qa#^&uBXf+H<D@B2;ZpOSpS2rZK$2y
zu;rm@LECI@g??~W@(d~&CAnG2!uZ@PY{bImvJbyOYmls;`lF#m6K73l0XE}Lxs~n`
zYbzg$PS=V&9OT>5Y1kdXvp0e0faZ9?CeR)F$Di=;Fem4uDJqKO!m(WwND~y=vVt#b
z0^zB+X4}pVIPsdzAOVagIo`F^U)}OoFw5}Rv~5$woSgjKTnuPiTXR!LeRA40y!-MB
z{F*_YzDK0JBtbc3Kl22gFd#Bu!NcvsuAKcTq_64MxmG|`;%C(#&MAy!uNThE#>vJV
z<{s7`#$Vzo3wja_NWKGnkqUL*Vv%~uR~!+*1Y`gV511U9yJGBIZsogZ+D8#+deH-P
z+qc*_T=$g_eDNMavOa}!D7pk7jnuPPC@vNz)buZjo|Zde#;Y!3Zm^mQk>aEKdi1F*
zERj`+I6$$V7jK8PLo=uTHvQ-exz#BDbNDCqY^cY*4TxAQtpg{6keMEjsGQ_ADuN}q
z=hw|GO?3l&5oYdN<Lq0#<IVMEul`!mBA@AF^1o!#<4>PHDI5(vw!u5w;73oq?!7fR
z++Q84n?58TkJXTq2-5L8WzrnU^Tm>dcR43>MTZd-=*XgfwY0ChF#q@TG~9BssPbXD
zJI^iGLf)g|PjwSvYF(aIpTa=zEj`K*8VTe-2(Jpt84Q3T@qW3;=9jm1qxZGKLc`C8
zuKu-MZK_<k&<<?UedDv3tUgJA<ghSuup}her^<8<XA6I`WP0F(@?%|5D{%EqN_3Wg
zR}jRhDv>Dcc2lZ=<;So{kQP&p=z!v;E8;;$KkH)c+;7iPLU#oJv#`?GY`@$h9LzO;
zdm*aar4o}!3DfhK5-EWi?_?R>A3_U7rG%_}XZTlIo}$u+n#?SGz4Hp7<nu^EV&@tG
zh|0Wp07i5Xl`pi;MCOeTm|2lcRQhdSXdJk@_rOUj?x-Ax%yXNSr~QL)D}9PAs(NiU
z<lEIgD?{4XUm1l)vP^RNAL4CWGxj>y{Vg;8kEXK@Yx4d7_DDx>fOL$}e&~>p7>INU
z7?gBNcisbr3`I(0Agz*0sC1W7LmHIPDGg4#o_(L=`2DkgckI};d)IZn-mmi{BLa@E
z+*^4+5Fp~I|5QFFqlsLj@m&`iPyWg&d->ojw=Jf1si8m66}8hCW)S2vWLiAG{hB~b
zx$a&72u#~~si5T!_hM02`qV7LR=+gkD7pTF6^>l|qfD;poVDx{h^)6w-`kYZbD2Au
zIkQ6x>+`1S(GYAv{bc;bcuDd;fVty*{_j-g#gBB=2U4Og;iP%r7EJieDas``=6WZi
zdk08Ck7{`{$)L(zvUA~Z_&D}W6`8a`Plu==Gb7SD38cT1WXPh1UhedAWkS3edK=`H
zHIO!BuphZVkV8z4)r$6i;Bwz=Tj>>x=1?9lls@?25QroifjwpUByS%A6p&?oBT<NN
ze4v<UvHk7~=~6U8U0==1*&%WM77Lw7EhJ-f_h_5oW?eaJi3|ber%X^73n~UR6b&`6
zPZz{acx~@GFVk{g9P=t$YW(WT`bAB;#m|lLvw-rVB+19A&7KUTHPD%Mw|*d7>e#8B
z{yxFz;=OwBM^e<o-SB4MS<vLePZ1|`C79_~BzgOLh9$4#=32#SDKA(Y0{TEKV)4bl
ze^gInFqjP&>8H$=8L|qZI17^kW}?6{{V{9*pLE}tUm;oU<7MnfZ}HJRxfPfaKtyjb
z&675&*9iEDlj^@68Ld_4+x&ii$T!VSz#^Z87Rn6ei~3OPYdbXxWQ35yrHvwH@VlAr
zFjhF^hi|$OGjWdWJdeu&5tze{Hy+bxc@%&+O0kYQAJZNn0TW{5SN+dt-*BY_{Z&I@
zBUJK>%)Z$cPj9!$5mG+kF1zG3C2D=0ZJxlG*iaJuijh2ZCM+cH=*@N3qs7_CHx;>D
z>!!idYDFz-N-!XsY~)X~wau}(s%q=a!P(wk1Q5&wzw@~;yg5!oS=34+u%<FNm`bm$
zNVUAQ9$k2MBIChq?XH1A6=vJjU9Tmgbzba(upY<7RU<>M!q_>@lEK?@g=4-Ug1g%w
zokd%jFmiP7A^%3n+uO<y@7$BBl0V~P0heX_XLc8BbOZK=8n(0#K+q2ot<4M9|K=`$
zGXnUBd*S}C!NI}Z^Kw&H%s%6xtFj6EeSLL?l|db95Z96rgWG_OXkm85_V2K8jVA9d
z{ij2>_q@N-(~a<Dl4WL9gGc#JAtan80>Mo08)+o3F25eIeW-5}+}QH;;QLc_*XS%$
z8s3Z0R|9x#VQK~Z0ThKGW6BBQ2gDMyoy3!B$`kh`S!nEC_Z>>j?YmepF$kjL=B2yz
z66ykCgsFmIn<&x!y(i@Nygz*$XEXhK364%c6r?wa46a5JXWmnH>{FDUf^UDFDuSnO
zjqIuM+oObkaWg5=Lv2nTXM~Y2kQyd!*^*Wj*a4A=nqaz<;9$D_ytcN|iy*}+Tmm|N
z>+vfv`}?Fknn$4W2izIG7NBx>+%3@6*_;;aF_kcV!Fu=XmI<`sB$ZuoIA2DZ+a3M%
zKZaC}wum?P$k07j%nZIgiNEd!yk7{oX=Gf#&U`g2`z1q3*-cZsp?OQY^N!#TRzu_}
zgA$-kMC7DJys5vR`9a@k^G^=Keqf40P-i^#cgu=hm$E88Q}xMVy$f7~fkW2TK2KE|
zT+D6gb%a!6g!8Jozoq-%ws0PL*ff!^s;Z*Wy5-67XMf`1I)yDceFnE#$f|1+jOqs6
zcX2dyz_PBKC476cSz(j^Tin1&Zsumo7l2l7CN!8OQ?q%K>Y3`z>DgTKMf>I2Fx~#}
z`rnD;>A_Fp*LzEm4<0{SIC1)&@o14fQnhV!v}IZ7FZE`4`1Scw*zazMoM-t27X0O!
z_syBAO}^I$7ss-jq$I@~v9nE`)~xn@H4~K`^^GU@JXhV0-N31u#9Vh7qDla=2ev@i
z@8PIz*AwSYx70LVjkx^>Twh<^T-)57x8F>+t0?7oVW<kP&i`BRDqWYaVqRP!>7h37
zqa9n2K_bD43hcTm_!R5!J6>652WEzmKfai50Q4ZJHq=fBPh-S9EOQjYmnp@Y`rWMG
z4@5+{l2|6Fu5qG(95T(z+PFuFH3g!QJ`k;&kalEx4w_NMhxdRaNS`rR>ZNEP^8U(D
zzlaMP>%%_xb}J%gLFVPiXub#A^MHtK)#=SMDT%YTwx_GW4#FL_|27%Eq{B^ChxFew
zE_Hl0gFcjVuGAwyh#0%zhN#R`5e5G#J3`z4F&Ttf{YK&s`Qm1IApYC^1i-d1Tq9`a
zQMP#B_r{$4g{w{RoAaYJKV0^Me}&;5?^Ybv1K(b~o1*Y-ZBbDcC5%_FF**HC(Jr5z
z%T<jy_i?w_nhBG8lFVicrzf&{QL@+@>&(o=RO$Rt*Gwpd$8*9z*x(ltI{FvyL~IA`
z05Pb8uM|5c=j$^;QTlZ}I}GTXGj)Y(q@OjaKv@KY_%~xyvt0Ok)K;-XPU_*p+OHj+
zD$J!veNg`4-9oRvgwBDqDEn)5I$8;Y_jOEi`m1CFk3amm21wI^!6-=KMf`aC_1gb(
zqTq$dhMRrv>*EF0^I;pWxqk>c+G-ZR^RQXgP+_&j!m7U@2{23-fEo?d9(Xc$kCMqR
z?X%siIuOHOqO*HD74=IyWi*4APuDhj*J`3pPv`&fN?8&1MdDS9rFU8M%AAyQn-l~7
z`>S>Kjtm)%8Nh<}6!imZTI0w_!{etnaO0TiK@avJeO^{v!Uw_o<Q;F!I}p7~nLb08
z-+74zuU;i2JVnz2l)E|Z7I&uB>lq|<W+mb^C>ElNDhEvc`N5XoFD`ahdgFnc6V)3+
zsb|yXdW|H;E#AZjHx)VT_*g@mB$m(hr#!J<%|-Cq<s~92cz|0q?ER5Q+S=u)V~5(H
zMhJT_(Z>wf37@2nnw?=~lp)_b2Pl>`%k(!{PfyuBv_w77^9qE}07*VtaKAfobZW)(
zX>~O@=ANzJo`}-!Qcjk})+VY}$fQ^_maUY+mZuEqr86R<s_orKmXP%drUlb82WJ_>
zjT<IRI(E+v2k{?bYA^7|n=8GwVotw5iiy-R@_v+kxMQLxT8L2ho<<VdLx?{k^Fc)X
zopG_IiVlERp|kRGvceyp#iH+Si{HzLJ+bN`lo9nAdkC_(%;rxBA9kQC@@dC>FM?D$
zkeXnFo)qaI6a$~V6d5$}@J$}y+`CY{`mMaTzW`kBIc#edwLPnhr1l?l<$dG?br?i*
zoXETDNdyNn2ftk8r6rn;2BS;SBCj%*wU_AV4}ix13q+mw95YU}L17&@r~<P#GPJ-c
z#94hclL%Yvu7T}~c6*ds8sP)dH&cNTv~BVz<!v=38R_U01?(;3(`GV-goKa@F_2Gz
zoOg<=28^w9qK+6rTCRXR+cEMtJpQrmqxXmDOZr#OLAOsE{*gD=0XIJC`k}3Wa%8Jj
zRYe)Dg=PMw<G)NkDBkBl?y)SQpqT;T$R`}*b9}gWaO|br#W@#wM;Q1-v>~(ir^qMF
zE3A~9uDzw^e>b32H>dma7n>~syPo@TxZR*n--#Yl2FoJ){^IGvW@W|wq{k)oA2mGR
zZj=p=Y>c{I+Z?PKuGiZ)=p6wh1L$uH-jxnE)Y!VfcaqrU$II%mC9E2|RuUK@a^Ikf
ztRm!x>>ARtpVLjfAfyHQX)lWoOhP}cU0of|1#@5j+t^o)x;Y#VrrYrU9n2ry)OU3?
zcXU^*o_0wF>O-B`NUQ+S{xnf0qp0hjpW%Gmd`(Zs#YSeos!bs*QOpN89Lt`4G7=!s
zIG(}g7`pNilriA`*J{K%p_9@7@q^NNxD`LJIP<SU-6ds4Ltkx9(VgC%j4_Yw;NZ+w
zM}egr=K01=XZmwVNYO;gh~3LmU6)_%NS@~&2?&@V0?uRmNHy1R1Alz}H>5Fw?61r%
zhq4ahzSN{Sw1(Boj~@U(m=yl&L^Hl$vmQ=RE?9K6?$!%b+d3`3Y$NwV?#%p%+P>g^
zI%=cX8WhH9_QZMl_7~FGCtnaEA>A`wT->U9<{*Ux`~E2W1l#@3+vV<&nMtEw_1j02
z&tD@6x#Plh)y*1FZRN(rJvVn+#Ox1=Zy<ID4H*IFb)P|@?0z6q9SxS|b&%1bm7<fD
z(MzKdCz~eADDUbw3GmnLm$z5n-W(t`Rayhta;;VdH2`FYh6gXaEF`#vs1h7&?V)z&
zQx1^#;7{4H7c!p1lXN<T+CnSx4;Wy&qu7rwwqJAT%B7%dcL#x<gkTVHLu0Yv(=j><
z$;rgbh*O;gOJF(-0G>_5V~Q5LH=mam$IjNN+X%Hivyhtx`)7XbWlaTx|1cN@tHaPe
zK1v+3aK0}uQ0sFKMYTdgWeS>mE$2P?h(UP_%{c}?C0-#w0mDt#+bc)jz?Bv#>hfP@
zrTxlH;qK78Rrxd-wN~uxN*}}G!^!{deZxy!+yEn$#uLW@)_g>(tb1O?y(*=lpw_Q+
z8amRbk0~&z)GWKo;`7a54crWFd;8F~$c#@$_Q<O5`k}ZfKdC<VkNLUf5P6E`qd{@P
zQGa(@+(xDKIffHUfLLyf2ly+_=Fo9-dqrKJ#dF{MlOA3R5$WR^O?_{;7U54w5t%Cg
zXzR@L9&~PKExcxaA;4y9L?B$x&%I&Hf3j0swQ%98>-z7Llz5tPwpVVfH%g&f!bLtr
zI898>uI<e|9oOf<KGZFeLbC6;hVREZuRG5N8s8zZtzPyc{>=bbbIi`klJzPGe5Ah(
zf(ymK<&n`W#9fo0Kb7H(lZoK=FtSwMi844P%<-O4hjUPPaJUM0Ui;-(*c)||@^nf*
zYvmOKdmyRUr0lVl)g$xM5T%}#RtdY)q0QA9*24Np_P>m`at!|nzW?Ck?=kgkdM%vo
zd|%|>fBN2chS?7R{j4jA2OI)md-AB)e`nM%UajUx#nK@5_i%N4tz{E*9DnT!@sjX~
zDo1kBM1=>z<@mR$wylnq6mLh)CtVaMD{J_jlR+9)KqwWoiMN`7lAK2<ej#szb4KB%
zDf(o^2_T(ZjuA-^^@%7GQhPm;!M+mF&54cakJUUU-36RYmNKj;HLyFsKn;c~s1~{E
zrLtI}WX^YwYb9>$hPU}HKx`@B9^Z0*@(afIdc$HFKS~b`CZWPnOp!1V+b%!v1cM+F
zBB_feKZ8%|@pkYeq~U6b|JB(B-A3bl{swUQ;qve1+R;}rOYe98h)D@BzwnF#?Z-+x
z^)w(-6ky>3aqj%$1?LNpk{<l)ypC?funWDtB}Cnv%Od9hKUzvEf5{j5t$yR^%BfX(
zA;-x7!<1b3`6VhD2{zF!!(y+smhd3xj3O$I-<7HsJ03xxMA~3B#Pgp$vmuz@!{i?6
z9rtGzA6wvQ1a8@6W34Pf&u}v!{X(<TKubp8H>aKL<MGuFzVJJOaKkt$#b~OF!oO$o
zks%Qw5rlJWL)OOT_0nEQ-eV7s;NJNf9<^NeT#LV<d~rNzNvs(|dOZ({f-ZUIa{@~O
z5iHI0LHK>i(x-NnXRODUx-IWHH#e)i%9Hd{9==Lfc*~`h!t)#ju}fLtO(GL`t@7V7
zkoyC&N3*DkRi9*#8q8Ou`6m@E?h_Bz4MWhQV2$kKW=ZjWnfFRe2K#cb<609#;ZOWu
z<@%(9Zl@XxJLRFX8aZ@`(GOpyTbwQ}T?c%ws-iPsmK6=$V;df>0-X{-Ow$+_;%*62
zRStJL*J9_#C4MTL!0Bl;ED3*kdT;;`CRQ*6){AXeX^81Tcl18Wi0t8ekCrkOQOip;
zPGwL~-ojhe_G~qL0ix}{7-9xRP?M&d&B$$H5NEXH*g0)xD*t#GBxWd86oPB2n;L7O
zL3XvCTwk}fh<_>{>K)(oDG#*3@XyxtoAJy12JxWO;~*2epNjH+3%!1BMhygvC})DR
z)H5rLHN9Q#y(9uBJgF9(1G%0&(4KTLs`s)hma*Xw$&qp3a;4yrb}*TmK%=W%u?Ixl
zvoF-Bk0h!C>^Yl~a$|^7I>OS*C?F=~h8jK<SICt&{>aV{vEZPT--x2IMuDX#=+d~2
z(0A)E6Wt?sUqw%5sYJAp>;qlCD%|;b3wh7x^KBODyyled-xryqN)z`Zi~Bps%0zMR
zyQeqy8;a-sy{@+=>CY8d+JvX$xK>BLh>xVv6U8b}<^$neZW>6D9WH|LT)Yb&ljv4{
zcB}P#3ZjNy?^fftI9dA|dH(Mo;nCL?5%o+gI|%1{HX3AOj=1<rd<c9>CL?}d4WXjn
zIN2fUu3v0grAA6t(((yTbD$LH`EV2z(X1$Sp>+$dpZM;pp_TXxI_^DgRV9w$l9eGh
z6)tzn=0Lf)PMgBdPv2%CLmY>~-^uVDr9$3tlb!AC9X>MGrD*GNlI%E*e(+$V?ID{G
z9|-cs%`2BAH7G+GgaYD;8E2Sv_tu5!HR$b%MOe;0({>(QOBblIviL1ujtW>NNL0}N
zFp+bkqIkP^>l;(ve+IHBzaLt%J1^bkA2U1=1jG8&vJ&KjdMI=T%;NZoDay1Jut)31
z5lV)t*Ow)VMqXjzCrIr<r-$E8rZIt`Rj69WEn-j0;k%dwj|%CMl|o9R2awAErRx0A
zx(mu=t<X%TV#Lea%X5ph-^U-^lqPY3%xIZ;foX$o5J}2#1Pl{4M)DufEP@PaYQ2|h
zgqLZy^nN8V1s7ms^SV1j$tYt?g8T4RhX|}gaqqsJz2l1<Y0ih*Xy42(Efy35=mqiD
zbqX9S3YtQj1xaKbkc>F0ya^t0s+?%hh?><Feu8fq&TWuWo&NqIt9j>J-~1QH_Kf>m
zPvW606Y2jbIccS!`enXA0yutDBea3N2bQ891k1wF>%7OnH5~0M7nBfpI1Qyc?&q!^
z`d<CpJo|^ks2`uVJ-fTj#Npa#4@>M&TQYKFK6cN|{PH^A+Uiq{kHE{1(2F)huI2%6
zU!2FQy8;2+>Xo3=OB@4KSD21Y&NR1VU@^r=l6URzW%I$n!p%zn8E8)R4U0FUdr*C{
zhOxOmoW41|ShzVJ9K>(e)jA$M=%{_oiQ74t<#Mv)e;=?&2@-xVzz273JF2X(q-;AG
zzP-1i%(<nq_)zLwstZAJ@B<QkN8)yyf)U-xMlan-{fQi5rKw(?E*)QcDPsGSj?S+B
zMm4rjA&=5a8)1OqgANAB3|S+W*9sLmi=X?tZ}lTtP#Oyg-F#@nC-S!Saz0eBR=(F(
zLm$tY+(|bV9^=bH{i$EZZ}2p5rcrCflLaWMyp1yNJsvn=cdWr@^(>P#oOmtW$J}B`
zx#kgDtvxU+`?{7Y-COyKruj8GUu}X&K%<~bUj)5#RtxY*>Q%&t<}oC|dlV^{mrS?_
zzvtF=K?2tm?TVY4x5V9@rX_Gk__~^p2NL3fx*??Axby2Ksqb@V#xYzy+|CL{F^i&?
zb1OrCYC8v)fv1I90cFZCnQuK}<&;p>oUKMk^iOIzvPVmv+g{Oz#PSq8?-gMDO#MAQ
zsP81xY_(;_^H`-7rcl?bE1yiGjcnDMvth!BMOQsWagyW>(7O_mNo!hGlD#r?HMEVF
zW@b!)A8ryFl?0n>mSX9L|H<P>8;Tg7WORKo`RVD<F&%dwgb!(l=9*~lUx)=ErzNBO
z?or<hxJ#>WbAmt)hPSBL%zI(Dfy<@i<0T2beP+{bPieVd6<pWoCmKqTq3^anioVCo
zaAJzC7H)uP!jd>y$Wv*zo_}IM^O^txZkT^;BOC?pM8OH>)((><3;SB=NZ5h}&M)pB
zb^qAqjyqd4T12C(OogS7hC~9S#zC1+*n){A)5)DA%LUWP#S_A@!}aUKV;>dYR3zxm
z-eiUM7am%`j6<sE#=DV)Uxkk1wX!?nqjhH>u>Y?59`&p2%VS}rUvA#Cq8f-^J}CIn
zH){e?%zgtJ>h8|ckllD5*nIEB3*W<JQ^~d-tHH9`c}pWsVsg^gZGza(Wnvj}(61Zs
zh{zOB>J9t4Vk}Xx=%Ld8%44=*SG*GbQD#szI9-4SQ_dZuT7A<Byp`Ug@j5YYOmJ--
zKrqElb}vaxrWKg(57kU}1t?~GRA(sWPz{z|pTgB41JyNcQQbXS!h%Tl8izuJj2L@I
zdC-@TeZS4M=A_G$z<;zdEDK!wpLpt0gLTa~vD5}GGQ1?3hB-L}=%*0lq)+!fJ7Q$t
z|0BofRch*olR}p^L#RKRe9CrTfOUldRJ(SZ>ekuELyx*gDY95*HC^sC+1xVy17S2I
zvPg*|nKN!qbpkudqaL`9y&;Jn$-y#6YMff^l87*>WnzELe$z_;Hy!8oD=#pYt{igA
ziPD3*-!e%F$@Xw0MK@+ekA9*S1OgnjA3Q4`*BYlnv25CJ>)e$fg(apn)}W_>DZlDe
zdJ!0?1FD@K6Qb^%QN|%nY~Yx;ErT%`ZPf}`f^=G}tp%FP*{}x~AgMSdcOz|GAsxQ@
zFbD5M6eEb5B9X^8^`!=>m-Eb{@QiAf4!VMSH1|OqwZ6|iy0Nt@e!ufnR0+Xdgc6rw
z{KXw<Js(=RrNj8|Kj2+ToW+0?a~ACM?CdI^;1S3NZg$TGx7WTFTifTQ{uh5ouho(-
zM!UyEBeSIJ`fL~=?^sX)LPI0BXNK%{Tg%FE*XHDGM9rHl(#+1kbD4?!XlRL_L39f*
zaNTorOUFUi2P@sX-9ww}LjD!;Y#dglW`^zvZ^Bsh04d&Tqp_HC9RO!HPEpUM!iib8
zpd;3A(#SP=-M5+)ew>C)z4&p@@7W8rkvPz=A@B<ia@t}lm=sVn$+w2Cyz~9g2SlAR
zuY5o$L&H2^9gnaXjznh3;rDb#UimEYE%Va*Kx`fu{po<Z(z^e;-DJTOCP(VR$8!gZ
z4UISVqAe~%8w>l%1K1LLWBu_?lT!Oho!h>TzW}KmjU1b_8mvp^5fr|RF!tw|^3(?F
z!y3oF!w#|ztqxwSfz2|l2FMmezYQ~A{4ExLefQl-{915pTk^BU?ou$K;KeJcie8IP
zeXkiOom(+zR?Y+#q@->NlvUeD?M~D1nMO~M{<sE7OX7SupiaAi7bYp1T8mjk72xb6
zNlYn?{L0GU<j{mZ$eAGMU61vXKU{_7%2VpKCc>3=%9T{S2p_bJHm^Ejz<--4ha<k`
zjn6IZzf6+sWi&3WjzS^S?EDBl$1oQfIzHuCg%i&0XDTUKv92M40GVhPPsg|&hikRF
zebvi;tsX0YDzS(Z1ZOOyW@g>u`ql5c;@O?!gwjDdYCKmHb$T}-a~rCLAe0$#LB9Lr
zd_@OlMd0=}*|vX9vPX86LvE(*OJMRpDD$#0JsM#jzF}JW7qGV$FRQc5=ucYGdFQ$v
zH1b@|klt4f<RQB?6<32B8!HPyMJ)#%{!D37y$2Ybc}KnqFm96{Ml|s|Qa{5S4+Bx>
zHaFMPCF0`y``d&sB`j|DMIVg(pNxg?GJ#G(*TV;|uMf7KjJl|)tvlMI#d*OXdn6F*
z;bcQ*38>gTE0&ZpTOP=5x5@~mDQ=bdsMC<(a3JJ#=`bWby^LHFBI21FP%Pytip`V*
zG@w7Sw#{_=??{6VwqKCmnq<iYL59Xd#QLwbNo!)o?#209{kkxudLef7NseoEkHHb~
zbb4Gtv(ZYdvtZvgL|xPQZfW0z*j@N*271h}R8^VdK*{r2?xpp4PoZxevAr?8wj=RA
zq-*5y<XH~muF9gL_a}S5k$oVQro|LW0O~(a(=U(8GVq%d?rxspp;2Q~lFRZ&uMziL
zy*H)I^s+q42s#Ew2)oc13J?#ICeLST^FREAyGeV++jlw;HE>n$<88~EqtM9keec{R
zj<)|iH||~iA^LiU|52+)OY4bvpu*Jo=}P4O);uJ(lYf>6ts%+rL5q}^DY{sv0@F52
zB6I8O9b0tIVGrhPmc$8A`q|svgR;x*$TCK4nax^@iab~R85{s8ac`UHiRxQz$8FLF
zxhaRyLM0^LX@AUsqr1L66DH^HTZ6JRKBVqI<Yc<$(c$NEC61<|F8=*)NmiaWPJC8B
zw35$dw_b6CJYC~%cUrgvu7_`Ya|FkCzdj#qs+l)kJFvM{2c-gMf9sS~BhPj*!vqfr
zPP{#;O$0Yp^Qp|`j6^>f(^E{Y<M6pN>nuR=*5VXS6}YPTJXn>lQYGrox5RF`{7?~^
z!ovFHv9O3+3=(wa%s@)Qgfv~l;g|Nh&wz^s!r+Gyw>S6C-ev|$?^f}__nrw(l6ZYd
zW$ZUc$jDHrHA%QPGsbqFBoP?gtbs|XRXF7dssr1fvUtD-f$ug#^Ty^nhsG@PS>viD
zBr@x186O<T3^sN|7lKGf=}*6IE#%LuDpwg*RVdU!7_Pe&!$9%BK5TZwMo1=Y7N*3D
zW^;e)$CZ3Ine-f=nusasHe0t<AMb$2%-j#R$*7pJ@UUAS-+YlW9SG2TyDFy<5YbO?
ziMyo%Cne7V6lm!@$H8!cwjZBp`CSOS4G=cWQrTETV+(d!6nKX|4TOZ;hWEeaKLrhs
z2|iCBOy$M8&y0&@_i{}=cezXi3$cY;{xY|JzVjrFVz;>cpw_6-zGWqMa08YfcG{E_
zN2+Hk^7gOT<v^#}3!vtf|0^FY7-=tcI=SXjVP{&`xRvfw+WZn|yQyX>WAN|!`TKIT
zM1C?s4KzG1!);Vms}kG07HN5R?()<AK?ARXXauAK+eYq!)S6j^!GknFOB#1edgCvA
zfbeh<5bqf6{R02VAsaF$F3JqXNenZB_~YqjFSom}hBk}^SUuMI;x)Tc1zc_5$F~WT
z(_x#u{5%^1uh#O3?33U@-0zX&hUAEo@%zZQqAA5;smZC`ueI&hSNWFv-k2jcDwSX1
z=8uEMjWiv9baf@3y-L3lL_!R)B^KOX$;sUNz-4{%%`R|#(#y^5@iwe<&dg>ZsgD5O
zuCurUudM(KOu)9~PHoMW7^b8yqga*FO83p|*7~DV>x-wro=o=V!K@J3_t>g}M=G9Q
zE^2qbwqF{Zq$=Tesg9qyXdoH7eiA|*W<So?Uj{q;BaFu@TR#=2mwmEC+`aa(J_tjA
zNGMHgpYH7OX1=yU-O?R(KEG_Mxo!4aU#U;IpVfg&BtF8mwQ(lJzz!8Enz+=uS@+M=
z4aWv5H^?;Vjm5lJB7vDjuFJnqAjuS}hAVrYpOl=~<Y!=BiP5Bw*d`ve-yB6Mo#%{_
ze<ZR6{qZ%wz+eXL!kScvrI`giw{T6LD>z|yh#83e3M3U9<J`P0zdqYYppECnkS~*@
z0zWraWNss1a5<ZLC)|vmd4|)2_ZE5zg#XL4_4VTr#a!=&Et&9o!HDlqBm7x9JZvFE
zlme*dXMyKG|6`+i1Ckva)awHke9|e4Yt$x#1>5k1ZIe4hC&>Rn#m45d4+I##0R6MM
zIV>Yo`J`heq<XK-t}*SiD}=HbZg6lL7#!dJ-<w+2XTaa>5QdL}o(*a6VU?J?^sSH~
zrk0lBHuckk?$CbdOe$CJTyzozYcM_ck{9y!J&`1RrD;EvpUL%dgs1=2X$hU`Ree5i
zz7MD@<aQ@!dYZ$lf8JV;(T?wi64iO`RxqgN7UOW)7BgXPwG-e!21BE>2AGT$k}vlg
zYi#nwhP^#^OFJ@_au&vacDIwxIhHi~=Mul`9dz0SE;i0XPJH|`<EFg`OYZM6{T_<|
z*i?bEY*TlhPal%8W{-KfN`S<PS5<9)&AdY_^`5R-NZoeno!h`%(?NmDBYPs-V7@dk
zr8t)9b0s+av#dHF7LrD!CXk%RgC2QfnolM`{{3z@kxz<n6`wV!GeVdDKf#av^6Td^
zfod(3R#$9~snMJ2S$$3)qr>5jyhN@$bfB!icYKLo=KfI-`Pde!VT#NH+>t(7i4{Dg
z`lhYT&8<L#@LS9v8Sm#{&w(C%#2@RDGYJC*`aB~oySB)N_KODWhN%YF#RnC;2`p=>
z?=g^*j6~Fwo%QgUnh=2|s3#2!I2R=h*h1vFc;0(bElqYX6BBc$2;~nK@=+SsdTx7H
z4c_1}H`SG@SC@wo;Yu_)KkI`}{`E4RgKuyPn5s1r4!~-jDY7x3*6RgtYBt>D@5*Mm
zHYrM|XY<F;8V?k?p5)U~#yt1N_SUo0!zBXd`EX^Jn+u3)xELpRG4`CD3xdoE7Hw!|
z!un>18JvSy95bjixw6?ORhiZu!6>%()9T^eiO3(uQ2B&NCw<L5!&~(C7bTQ(7qjQ#
zMhilpfSPeM^n{CyeAS>QHugoR@RZ3|=aV_ft5kb4OsahnGDTO@=LCw~+RLk_bb8IL
zsRmxm82@biLL!akHd|6QimZ_yQxUkI8z$cM=;`sDeb5U1t6vV`%3svrko7xf0{%RF
zKDxu)q&~f?SIzP4uUGlsZItOxDupeW1?H7&16djz_gopNX&2GN7M-L><V4KNiOtLN
zHG*O6IBe6pKbm9H<^w(8T{x^lrPTvP6502ZJFkyls3-d5t@*q&kPeBoWbzUJ_Lcp1
z{QT&n5n*g?z2|Luc-BU2ql3p5?UEERb63U5{4sTLTcxG<+^$D8D-REasa=M4w+G@Q
zVB((T?$mMUhLw=P_ENt1%0xHbmOH`@;>Emy5kExdm`H-x5iIM~pCr;K75)hP?nwT-
z6D@B7^C>BWMRGb<RgJ#f(Pr3TGM2AH^l8Mt&#Unxi2wSzSXS^IFQUIaO)|;)j8?*7
zMbJVb<1$owWkrT#7J4U>2Pr0{eP)k<22xw^Ey0f_aeu7zAYna9!|~%^uJHt#@N|8k
zN0Y;2<R(va7pP!;HI?FJxM_Ye8*jn8Pfa)HjZCBSbiMaQO*9*{v?SP%<(nVTIxRm8
z$_7!^EQ_~au6tnu8^aq2i@de<UT?e4G0^)TabBLd1m9cd1!Dcivkp6s<nD+E4?&|K
z(GjJ07RyTAsR1XW>@{vZqn;2c0g=2T%3xo5h#Jp3XfXxn|K*ti3OVce-OZAGmFs^c
zgaBkCKdC+*1N<R@0pd94CBB`zA#b4FQL&3*LmK5N1nwUh2GxB(9+P^%xSxUMrG+1@
zk3{={Vp<aS`V%R2F6UxKIT}^RWQq7rp=5OR(skA+*B&S)&-;GC{uDD>r$7hgvMK<v
zV=@ppK(Fe89@1L%-;F^JB-lxTKJMB+hLHB@IXo5WcqW8p{~SEWW+<;8Fbhh6Nr9=y
z5sax|58-ho?j$XY!X^nPcx(laU)x;!ig~t4co>Pjjw{OO4p5ThO*<m9s;JM=WPNDG
z-vVQ#q#AE?%hIhdDJkEAj^2Zz6F+MF5}qV-T?!3t*k{q6&$QB@q>tU0iZg#XQD(>x
z$BUATJNYOLvy%$-$pn@RS$}QXUIc5GQ3PP+a6BR@B)n4TW4_h?n(#YT+YN&0Nt$_l
zNZ<?nb(PB%j+wUHumeQ@?LPL`n=jMYrxPOol)T6aa(my5%|t$UfoiyyFIe^5jMY(K
zxib&uN0`~0H**&?AxiO~iJw3JT3J?00WR}vQ3XKE!sCY_^0}wpHyy*n`<S9&Irc3h
z7$dfQ+-%zCR=}I&H!~bylauQp9%3HF8h}>&EJN+@!VhB64>gr<NaRT~d5P|Zghk1o
z1bPftd3)~<b~_!ZI<vAv2ggvL(e{k<hcOhXDC5jm+8~7#qP@*TS7`YZmX~}uEuL+d
zCyo=DU?RmM81P2c|76PnTS3nKbNuZ3^5F1#rTyyQbUc1#MOcVwZpQMH6%R~HA28)z
zK9KMM|Kp(~mBp{uSn1|knm*bsXG+qz8*%DqnfHoA^W6O{MK$j8nxrNePHu(`kQPk)
zOem{o^vMm{2^teeh3`1T!z3%BBn%WyM{|?!Mocd4VJp(@clo)noth>FA4Ogz_=%n5
zmM1`01jzq^|5ZEQ@NI1h#M<!`0gvtP))zzFxl%`amy!xHx6WF(%C-pj_ceOd`*&R0
zo0mM{Vd3N-q{q!lx|g1APgpKkjL*Y}trcqE@^{`-X8v}@jZKreJFQCj+WuDjJh(Q9
zzqq(4rquQ7CgSG#<AMCOA0DYycw?dKM>``(!16;j$8L+WGpC06`}?;v0<D${c>$5}
zZIUra$v<UCveIlF8R>fGrY{mYYbH*{rV$GVB+{5WwNJ{_vAo{o6i76cUaE;XsX2V^
z8jeQkISKdeY%!J356Mo8M3YXpr2bk^Z>gK$^&w$_8be;EQ22k_M@HvJK3>mj8|#tR
z9!bHG&Ptw03~01}yh>^Z+Slf-lCS@0_`SRZ`sL<=&ZMRBN<?N^(zcNtr|;0RD@8A5
z#`dgpWqi~!77VP^cCV|;P+|$RA(Ymbry*IziX<i;?PSPCf*lxX>tR_|luocDOgHCq
z|8K_uoTnx;TwT}zXB*ywPPoTM62yrxn25gqDwjLDc(c3x4RG%N6U=wmWA}MeJYRQd
z_K#ndcFvF)O~HkVxQ)xB<UXaB6~Rp+v_8LQoIHoC_V;d7<^BLegPom{UlpGjpvMg}
zsC-jR+|8R$yey~l?y7;hjl$~VcsTgU$I;In5dz{WLMK~G>3Apzp|6hT>%|9zDJ?{{
zgkNrU5A|9q0SA~Pr$(!;UHKsb@WY}6mF)2r&ad=&yQz|Ztb|jeqY39~PlMYXWzAo}
z^yOt1?lfAAn(ohfap^1MJIx~;T)lh-;ifV#Yv~4;sVDkHovGzoyj_uB{2@UdzF04v
zxW}?|+%dnDe&_A}+!J5h>mIY!#}cJrzrg#G(2>4C3I~_|52K(NNWZ>M01eek+hPlX
z(ZP0aoejd~{P-m}Asbrm;N{;Ni>t3A1uectV_V(5ab0josvHeMg6Cmb5eIEATvmzD
zK%VBuCi%ZLJ``|9Q5MWkMTJ_<FNp19x;Guo&a`pG;uyle>Dt=YW3ZqfO0r|;Cwk%N
z9B505VyPnHldryIOMRT9`Pn*uUYvIReCc}qY0^e$dBLZX*KlxvfZ6d++GUn`GtX{A
z<AedT1f8bR<5h`OQcHwA;e6YI{iZGXIFK0TsFsG<SzmmKc4q?hP=mQ_9o#r*q4(MJ
zTBts`s`~x*qw}zNk9yH-nQdT@mY4z`0cJ-*JgV#_LH$d6m2<?}(z%rAJGR283q>LL
zBd!MT_s2h+FIADYiPV7(#lM!QewsbTn$IlH$j7ZzN58~=;gtN6F1+=J$hn_3&>t5D
zv&$_mzFIqy?Q2(3b+AqCC)l9G_X^#uG+wS&H;P$4rtuHAVKyx(2s2f^JaeFu1*$Rs
zg9JHQo=?renbWk<c{W54UP~mcpFDZ-kdEJa-`sR%g$MNb&Vx)OG6h9x67plO>A_Cx
z=5;B1(4)OMBEm^_e){z1=hcN9xW2cQ?9iV|o)f$Evd0<$Jwum~+j`^wM0;D?Hd-+H
z_CL&+PH9J$M!1U>Mfmxaz`|en3*ml>oQ4~XdE0~qxw+!mdapx+vbZ%tyHHW*x3Y`G
zmx1mGJ`DW<NiAUc+4{V$a&T<2bRw*wFzhMgv-#<g`lCVIS975P-koZY0RAHU;vHbe
z+Pp$$qj$|U!B}1P`@7|SMKGd4bvRIkQE0v|oazyI>EbO~{iX7Y;#YFVCM99KCQm=Q
z>=Y;OQ4qO{j>}A%#62*7u}rT35$(V#F(!cgE;>POI`07Z9gSjaTt63wJ?V>8W^YMQ
zgLfG_#UCPs6z)JVMe$b$Oes*ParoA)@jb-4?K-Mkn&@O|RZ)yOCs+HCZ;e>LEvxF|
zS>`&)Qn!_1bU?CG+U)7-E%H^T48f>bj?98`2IsbSSsxU5d8j`9#;v`dnit13a_Oqc
zesdz~aLWS*&CsEKmi6D~=4OX1ya|~^nH2sIG`<_LurD{4GN$KJR8+gSbv<-nS$`Ug
zR1yI}80>W2ln9Jsr}w$6LPZP%2cfpk0@So}viFCdZ?uT==)El`bKQ1syH;aS20@Sl
z;@_pj<DLq^I-hN>>vV-KSePb*1e`O{{3(gm!2liAkC9(GbJg2Y!7evD+dP>FRPjLX
z_~G>-VXnEDgj;o%W6v-`=qW2pr*B?y0#%X4Fm-~SaQyh@dH^|cUJ>%y6Bj18&*iqo
zzRrYX4RCj-kdMCz<dG4P#s68M4J!F2M2r|iF?3Oa_}WZ(+Fzt(0yL^Gw^xDh%|Xio
zKq8J7e3A(BpXMTEVIn2%>@H!kvNmWgx5(7ee5moRZ$I*1ZB{%{yfb_L@GYzF_?2*~
z!B^O>G`vQ%h`LC$^-yX{MhupN7*zawy*P=BB=wcHFnse`0lWN{nAfrFfjd8RyKHrV
z<+B9iM2zulY5d7|TzLi0Xn?$5F5O1B!Q!Upc*)>-gt>SHdm7Q%+ax)!sv8->tGwxU
zlRrE}n>(+xI_bDcUm3ru2%X;ZBC3xF{#;TvRegJ^-nn@38MIMbjfwPW^KZ7u_uUDY
zWIRbGuu?``;C|xs#Ct|by*t~z;dWVk_tMe|?<>+5<!!Bpb)KkS;Z!GGUys^qJV$@>
z`I!=C!gye)vaeukF^uP2Q}bEvcAqb43WcYj(?;`htOrdEjwPVBl5G!bmih94PoxK>
z-ueNa>7?QE^U}-99rT>m?KZwDcXh#cN}3f)bIeKSE4&WX-tsZWIbpv2BjE?Qw>Bf1
zZhUUKhgI{qy=#9&1g|!hmgzgCnPj;Lle${tW>Q1)T!fueIA}M<{mkEOI$2K7zY7ee
zk?Vo*z=>~3du)&4n4I$lgzbNuCs#|mgNcHoRYv8l7w@qVXM#M436r6X&wP5d9bMXc
z##&{OQe<*2vsZ;&iR8g@gT8>Ek_ZPF2{TBaYyDI=Z~5fNPOM#8jwa`?{!Wn_L9*??
z28FUxD83Wc-A>h);Ck@{Y}=oEU;TaYTz8fN^QOpTWKRo-G!2H2kmpDuq>Xye_e(Kx
z=srY`$(=6^;nT%UCIN++taLFO^jbuR&V_1!YPsl2HUp&d_p>|k7r^=2)tZx=o*YRH
zDBfq5p}75PPTT%bWL#d5+USUzP5%59C+9E8A5*)VS6##!k4v(%8_^t4Uf0(K*5f^N
zXi|tJGWLCNVU`|0YMug$ou%RQd+=sEk`jF6wN9RM*9l4R3v1{KhzL-<q?Cp$kjb-~
zzJINcP9ld$Ja*Jh0Qh<fz=?Mwu;o~g5YaE`hp%~Xn^v6s@gJw)BDE70M?SasGN7b6
zhQ<mB%#zYZD<xK}f^PiHt)p`tK0nZ4xhIpp`q0eGAWh5T|MDU!)u3Fm!r5%Iy)0dP
zVvg+3XJ+cMER!xmA@%d{KIre4O_Nfuw5Qv=a8u!bDU=H)0LA;-^5E=dyOA#(T!sjF
z1AIj0HXgS<h^wou%t#L#VD*=I(zwC05hqNlfRx>(jJ&x+B6a+&%sjR6FP}b~3<`e4
zJ8PSG?NyFsRJ$4IyqOap_j2Eh?g2AutWTItDn73(E92#pPV@KZiEA`gr$jDRDn6f4
za>b18-xeMN&?YROv@85_N-p9raPb%aON<V^+LSfF<C0IteV<{zH@WXKemOEEf|PP~
zAs#eLJ?F`{v6)|3AW|<UZ3uc=_o$YEoJ>+XlfvI_KDbn8OpAk7sx4dV?@xL*J~6CF
z$7nJGC=5aQ-Y$Y(_rnK|%;aQ^VbnK1)t4qvfgCS@<B8FLul;yDb!oK(6`_?Y+gbAK
zK>`sPrhmVoIvKn7<s0O<%FpMn=AD5>0l_XX8e25=`HsSKs9LV>E93F$lM^3i0Wp4F
z&o~IIJiS|nn=Zav31dP3Z$Jl)e)=)JY~m#nck!ySYDDa91d|e@3I)e+ml&TnWrp$H
z(UjbPq2Zd-?`LATLEN#Q`7{Clj-Zv;L0#j&Ts~uOCrqJTl$=DIkeDte5s_SLTNoBt
zSk)0WITnWzV!hqIiuy&6uj%XG9!d?@RVHcZauQp!wt{&lB5=jaFI-*gwCH%%5P!3=
z#t;-bQ)i&w)J)<7cya($z~q%s33EOTLjt@_?8K4|AE#+L4E3pJgAL@F^kMyH?DEo3
z6S3K`kqOZ!c8^43mwO*Up<Wpe1$5eF@z|v#guvfwvmO}S+W-D>n@U2qSCY;NTlLD5
zXU5ka#GKFzde}|C`VOY|om>v9WN2qOZ9#dtVxGYd#K?h@<)Bi(S+Nwxv&hqW8mKK(
z)8Jup)*!c=HjEqeRa}{q8NO=ZN{*Mh?HB??T4q!PlPy%(Rh9f3@b0=yB2LKaV2Qt;
zCmA&}m_vASNP>7^>*_3StZWGzO@%zKPZ#zTe=HoCm<H1!R$~x*BcgTd21A|1!mr|1
zt$&(iqBO&k7Hyekx+9Sn(X5hEfXOBp{pGPrZZUq-$KRhIQ|AUwLry~I9;_ONR>~jl
z&J6`*zUmvql<cV(jMsf$dN@eR(MPQx{cCQzt+lPaVq0<8^j%q~sHbIp;BK*RBIu+t
zi-KL@Ghg<93O}5SX&FfqNuLl)?x3}!p0bq_FBY(TOuBfPov4gA4(<h=0|}a2O!Rhd
zJ2ka_<qF${uxz9{4_qiSUo;N}CPfqR5~UiW9BtxvEPwgVAGbxx^T%JDGW__E&NL6N
zP4l5GM603T%isM~a~2@@6azk`ZV@cpJV3q{Dtyh>t}=9s?`xfTGon5HS5jAe*SAv?
z4<wP1!zs^IQU38tnE4R~5GO=!+ZQwS8Bu)UB+41o3&+xjt&vwJ>T=VH-}Z&$@7q5+
z?cdhpwe{%@#7_^N|GOhs;-{ErxN*NM9%QE=VvwJgrwW`~eVb5tAfIV>YDJ}vV#D;K
z$MY&Dy42B&sbzNLF?>aaaWB-#eEyvnvv@37Jje{A+wLQcw!sqva}|~)+IvGPZ+f0m
z@V$|1ua6+l)M?Xse4o#r_Tvsh$3Vk9*F<4tVn|;y`ZNiW9_rR*Z~v~6HMSW++7er6
z`{9D3w?{cgNO0j}&h99t><LHU{#Lj{;}<r!+wv-^NG*sO`9peSplofCC?=2)YtkuA
zS`<?zR_*V1=s)1H<>IET5FH$|dDi5Hy7i(-L8F0{^i_IykDOfH@or1axXHRMSrJ^h
z`=a^73&qo)qdyKuTx7umipe}(NjVD-_d0DoMmwqR9__WcsWRpffbz91c=*_@uE|)A
z(Dss>OpMSaiNwRg%l}<P|9==D9idOx-J9Ht!(Y?|9>5mO-Gz1Z2Sg`!=XSj-KHU}T
zdXUlwf9!d9eO%Wf?s+uQp{^cJU?GPQkX+7Dna=z3mNw!TPfXb*i8BB&LG6|jbv2~@
zyU;>!2(`=4)^m{_G*-ez<gs+`ytpM0>~JOpuB0qM({0Blw|0$i@zEsyJ;Nd?6Bc)W
ze%|#vYclwN*+j8dkzr96+v53kd+rcKBTwUB>q}Lgg_Ufl<^9M=47W04Av#@UFR_g=
z&FS0CZW<Fsj9H}l6>t`)z+J>76k(`D1HRt9Ipq%CXFfZ*Il8>`BHVeibX0#aD72U;
zAIUO~Q@2Ex)%gsau#6u!K$2i7v*Tn^MztL47E}(Ib;X|e`EKg2RQdW`c-n8V)CYpm
z+J)&gy9YKq2E>RU0k1rcG(ISCX886BPX|0<>Vub;*B{MBfL}JIFu<NCK<k!zN@^49
z6P`jBUOL7;^zVl>8TgS;)GrI^c-*;&xu+i)QvT*>d`)7gjUXbbsP#0jcM7~`TG$n*
zvk}_X+Nh9YL{#bpt=>HTd(DO#YzmOOgMFw(!o%QDq3(rW3OP9~XeFGR<T?hV433VE
zYcBsC;=S{u2y+wl)AfV|vm*R`dB!M$V(sH(Y`AMIkI?N`pXJfUS*0Gu4WYp&Lt!UV
zjAM5*`n4k9D21`f<x{#49uBXcS@MOHO?S)%3tw_U`uH9+Xcy;7#n}4DN0YF8?=u;$
zD3(ZpAhf}7jQY5UoLp~ne0Fd0iR$&g({92rHrHMWV50M@BQ=Kysgv1J;aw*}#iQ7T
z5jV<L1s=<4C~_n}LS;4|Pg!p7er=JotCf2Dk{anF&r!D`DMHvyHa|DAd8QH>8X9i%
z$p0ryU3c0>E>_7?K8?qZH;vc~{HHi_HQv@rJUa2h)^NzQfe3ovWKE0Mj+$!uaHB}&
zZr3AIz-`CftK2S6zN9(?Sr*LRXHLD6!Z-J0($Pme%;v2`_lYwI8~ctGg6xC45#<kr
zLd)Zdz%U0n>49)v@+e)%XEv|B;y>$uo^D&6`Y2A6F{a+3NA~cUXlUr#Wmq%@ZqFo<
zd8FQtYlOUDuvg}AkCy5`To)>FpbKmK3|#C2YtzHtm<kU?d&iGfpq()CUrWBH<=1Cz
z)4}R;yq4?~OIu25@&Oic>*Ad(D<d5B>s6TJ%P@tX5lV~!V6b2I@P=Q@ZxJgM>m9x4
z`$V>0?>Rr9A&AZcF68=5t);Z*;sA?q>ELm~sRs?;d$`h*tT7MVq=Ek&%4hE9YE3TS
zEK8L7pRu-o;xB0flOs||u}8B&vdew;XtGW<t-z_sTj^*!3zf5j{vfkpF8ti0?%v;>
zrSBx+Dg7)Nkg-R+S|6|CUxEpd(D}3Ft6RJG4&8H3S&l<O+gEEyaZ?NPVpUc9a&+~#
z|AQMjDRL$Cj{B-`{k^qgi`g9;Gwib>2m`WuL|f|D?PAu1zIjNP-eyua{ubl|mFN0%
zEoR2eMYzu-M5k5x|7+s^^$W!{DgN1;&5+VVpnRky5YEe1KY7Sapw7iVbtq-NRa#G4
zY<J1Lg){Jy$MC*r{IJPAJ<7XycAbYF@@;(nQJb9Xsr+xNOg;Uc4xfK8K6N92WJgVb
zo|4i2bFDsJLm$}@)4LK79(;Ye(%l_@(cFu};RqA4b3<q3L}G(Zt)E-HlaPp!S$&eO
zFfG(UNNbB%Mcnr$&%k$osXPg4N?e_~hH&i^{nB8QS2eMEuFLM!<O1OftGhlXXyv|!
z@;yPdd2UE13+?Qn{<9)qS_0-~pBrBV(_ebY=3mF!G%8ld#(xGgrQe%b<(FoFH-vXd
zkq`&o0B+%L1R7sIw+s&6qlGnL`Gm`KOTL8P+;E4*hX4HUzel`Zbesb~v%z@sqyqQj
zm%Uz{u52Sh#gfE6JI)o}L<nvqD5bo%_-b<5Qp}-2p}~PvIfdOa`Zu5~vSf{`E+eXw
z{+gd*9iW9^Y<2n}I#Fgh#B<2Ry1hP{O%Br}-r}r?kI5$CjsEu-rlZSxvqX>^!DE(K
z25g1#9idqHbP;u(5N&-|;qic9vdd@l^Ij+78-vr^R=c?I_PL1G5dV&$mo&J`W5A~!
zsNX6E66ZQ7z!3IRK)qeG=Eknhfwtni(#6_ix{ITRN^o>K{RZwzkt=WE=5S$ZvFcT4
ztvi(r$&>nbCQE#5<Xc;c)0Wozl~hfX2<Kyi=;@>W+}hXt{l;U)Um^CHS@a0*9^t`b
z_Z?}5uAH>&P&&3fJS|C2Wn=QSCDBxP(=#?{a#(46{C^f0fRmvk)(WkILVsH=&W_<s
zRq6;^iyhrmOi~tzMya7e*xgi~uWAfogqPpv3=dFAL{N9bt7FprcW*^0YAp5baWuBK
zssrHJy<aErYbtGdB`Z;M7fQu;Y`tzG_j!1H0uG7;pFKv59i;o&uztvDT)>1{PfKOk
zvCT!c5$4gd_JZ!gIeM3|98aHq{dl6<a`_KWSC6|g0?zTk&2)Yg{>jrjCYhhS)9=hg
zC{#Fv4N;gQ(W$V9^ay9KfuYZ3Q()`unRJLJA~i<{N?thKq-nHyzIX2?zZpkiB)&ge
zSXuGd<#1Ka+P%(mfp~;RL~%aozaXWKlsHuLFFAmM#vv?BF(d7%5Z&3?+P{_L{hI@B
zfFmd_6XnMCBh5d#+j<@6*~rU7c}7jJ6bwiZgfC)vD?f7Bu11iwnR`Wfodj;bWaY4W
zVV^)-J@jjv2vpvqBh&>?wyi8X0w<Ggya&gm9s-EFSIPG@l!&R`w_8VexPNhg-3taI
z0qy~P1_5VcK1cNc<6oLQ%8gkGXd)s9h9<~yOT4#7Y1?xRxp<J~fH3q+=x2|L^5jdQ
z9f4-82|1XzVBrb9N?jflnn{5sF5Q1!FY4jKA?2O&Q>9SqXeuWT!r<?C==1+*I`4lr
z-?xp2*fkPbwP&o>-h1y^loqv1jmF;VM$J;IK4KLes1XzqVznq*qZFYcRPC*1?I+I<
z-@hQq^~!bM=Xspx@jjZ|tu8W2H*vQ<s5UkF;3NY{NuZ*9r^jGQ;Ps1IVCYYEPp<s4
zRLOmL8iIbDeAV~6A~3ds8!t*oWZfA$3veu)UHZ2zXMJOQUfhCXKELpOO?91i|9oBJ
z?o2nfEB)0#L)ennaZ1GJI*7)t=TwYOzg&wKMA!*Si;Ek~n+(wHiR>ll+F?W5(JjK*
zVuBn|jjB71=w6RX3wE{yC<9Z%NCwSgA$%3ZVY9EwmthG$Oxcagxg*gf&)`;guJf)=
zavtAzW2*9BlH0+u;82o(Y7Jx&a314~=!)NqOxh59;qu!}V-0N~KJ|w(@DX0(w%0!~
z9m1*Lcvc?bx30-m!8Ra_($|W6;lk<^b3LfPawdFnU*9Kg23jTnGs_I@j+g6xsVU;I
z!_iY=ouWs_*S+1~mme%F7-@j^lIdyNBxj~%PZr1Sm`?oa#c`q15W&XxansiOhiZp?
zeS5E|HV8Dw!}Y|ZMz73>QI(DEhu9AUjxTfNU(`8aE&WGnS657oQa3(N$1ZT2a?xIc
zA&!aq3yuP!_<aQ6iuYt#<Q7V`*s@HVjVJ$B@EWo<$$Gulxp%X{$}x&4vYx6&7StpO
zk%9Jm)0TcSI;$+xD<^ph@|pj^``ptq5LziVGVSH6oG2k|1W>yHb32*s>UMd|(4;3%
z=o=n;@6tl&Z8;1_NmI|WSSGhRY7g1m^`983vSlvGOEv)oP#1Ti3h`3krmB&{j#?vo
zZ&POQbmZ0L+2v_>%-M=Br`YM8NtlErT*OIDy=qB{lX!7~Ca2YAPnKbkH%IARGI2!M
zf-N{mhyUP3I=j7o5?ieE_{2lVr_+il`=Tl@4~ErTdIr4&S$pp-VUGomFVI~!I`JLX
zKUA#7-?)KV*E>C6j*Qx_Wqj_NIg}2*BkfJL@mzU=ZbGUAeY-*FXI@?4x4K}lU`i}Q
zXncI3)$!^&4^MGIB;&o->Y50%MFY1`CWjKby#=+s5QezU=<CQ>vUSO=t+VrpE_+|3
zY)Y0M$Fx?<KAdjWYBKtxSi_gb{|$xJ$ABq|g&Kit(`V1`zUGp2{dbamJaAl;UY+)8
zkyV=ifl8r^<wq8S&v;|R){e2;va^`0LI;p7z;Y0C7SVQj@^=S?42Jt4t30R9>ZWTJ
z>RW|fsO<lweV93|$<;9&b<=kf77FI7r!+7KFV;QHDz>d@s+m4FDf+j5zx{FJKnu`+
zBD;ekes0h0v&Bxl3szr%x!YN4(K^$Pq}$UINu))NVG%5EbJW7vKKF1etfr3Ly`oUz
zdda~;vZZ*lBcSt92n5b~E+4u7DjVlZI<Rw<rlI|YaRUD*Q#-G^WsVW(jEd-7w$Il1
zpg&=~MmUj@($b*^+UFKDl;5oGZ~<#SOCh0pFu88RKN%0xu8wGbf>D2mA{Ea1YG~@&
zt+uToAwk#vw>MKQ>U!tU7q~duSR&vZ?N$xdXM^nT#}>JOLd2xP9)L(l<rB@#jv@S9
z8nTA?)mUx)9-%b}iC3t)EkaZ&{F41qAEp)#$Fxu!Iowz8?v?1lOKTDV>4}oBA1JF3
z>_KKrF^Vds{r>A@w(d?z2n2&mb4#v81wMa3vOcYVI99Xix#;d{dcHQrgA^(O`f!SH
z6Yq&?NIvY<5@D02O0TOK*|eKuYH1hci45k}vJ5^R+PR+Y1}+c9IDxARL^-wtsmDpJ
zaOl7Xu!4N(zlv7TQ2?NYF<3Law#An7{P#0(rye=zK`D|r=j2j=LYp`#Mu{jsGKjUU
zFp-%k$P2>NR!}I1;;?7lpIIDmH*j9{wcIkvrMD};5&8XL`g1d!I=k)0u_%j3QJZhY
znF-@{>q~*Szy>Fk`bUh)H3vV5CYd3Kjkd13X4WfuZ$vOv8xSAH7gmL+_^=%Ao^6~B
zgo^DsQbc!l%lIyiJn=P(Y#&}53MjR@!+DVy7Ilv68bz0hUal%_OwFG*zUkU4yTPs~
zIE!nb5%3FenDKFzAcs2Hw$E7gr&?LDmkez<TSR6lAoFM9LR+W8Q|^OMWi-bCMd|d#
zPptJL@6F&>z{PWf_Y>b{;jbo;)GC|=u|`VLlYal<cHsh0uZfL3B4CEI!jmdina;?*
z2#Z4LDUL1`yV3Kk>qN|ige<B31IyV*aArcCG6Q$CcdzNRXAb<7Qz@Yk@!(}*uY6XP
z!+ue{p81zYlB8@B`H7J6w`Rxf)1U6@b`d$Y%q9>41<WpKsbNeFEx*_oR~2B|X(fU0
z7ToPdq6>mw{R3_;PqRRjqszaCFmt=7&TIn&D<)DzVlU+4qS3=r^PU1uo<4`7Xk#b9
zv(geuiTY8hx|%!R8CkbQznLHKgHR(ISo8;Q{adW$REr9`=}kTbZg2x;-O*hRJ_g^<
zqQZfT2=~0%tJMVTu(8~mK(71{7t*o@aY(Xq%O6NF0V~-nzmtR+4O_X<kh(B)z_ZSE
zma+e%j?SCD5Up7>TsHXr;AhrKrP}p=GH}s5lpMOZ&^@G3BopfqTm?VdUsCGJU$diU
zt`tJNtM0@E$#`#m`^;lhnDjTVF4bbwR0G&=u%3EC!;>G*G~n6I8amNo-p+lV@J7gh
zOjBm(bA1=Dy`9p+YWL;UgyZ%37Q$@JJfTT>teagJWKTPP7txSNe;PfWCwLdsm!7~#
z<W6_GaUy-J3Ug^<rOKvIdf?a}wl*B3xK?48tcuo>i(l_Y_&l53U7@+GsN4>W7X|h<
z`T(h`&W^dd8J43RB3-9hX<9oT1>!OlkfA{<b?;xlba6PcC23H|fWy#{>BrFQ<efPh
zj31ANzI@v|t7+7RMKE+RJzhA4@$K&sJhg_!y^E{s75pOpVzHr}KqB3n_NuY-M%wQV
zowYpK!&t8~R3&$?@{}D`iHx<U0o`{i>=-5-Rc;zzaya*Le34`CZ1;8Sd;e|yhwk&Y
zM1Md2XZ612qX?NmcET{U8iSGh_Ku_meGw(&r@WoX8AG8>C)a6%j3^DC8qdmpbNG@L
zn#KwuCZ#l|A7<;J0ZU|=>kIN3$LHjSbE&GT_-+)l|NOM0AS>iix&>#-ngQ6j=@9y$
z{)#7ODhkWy50_-WA5h47d)bHg4wI-ajyF%U_$=BxP4Klt7ELoP!Ty$XP;RZe*Yd<q
zx%IeZta?Q^pT;%Jmq>z09@9Cl2j~F*#xE3}zTAvn|4!l*Q;N7bpKuiH-IZ4DR#RwA
zth4|%$lS@vl~gbC4Rv=ns0Lcp*d8y3b$ZnV2kVy9qqUnC_JF7xZc4iHxqWI9t?xhK
zvauNKN=Wyd?_sp=XT0+{?^P>~<`nnjK>i0IG^QJm4X+xqwX2yPf_MGE3c#j5E}Uh6
z*zxnBk)7gmJraes$PthN;wIZ}|K^4&c16|O$`T(`tAAy}w)>j~K=KGeRo|S5iJ47E
zvpHsj>+5r;S~`2oN+8sQ#D$IVf?*DMZZ^!}`DD&rj`hyYR#taF^g|E(rD=IKRDZDl
zdI?IPR$sNY@;$6|>`L(d_`0Bu1{x^3YBV8RZ#2-(^*W-s{;;0Yaqm#kWNA^2cF&Yt
za_R!!i7$jxbI#NPt60`PRW4fCcD8DRl_J@AuWBoJrAb_;)#uhQZOjJ~d@DP3Cb$;C
z-)XcHEAOWls=*ZSE~9PN=7-m&X^!6u$CK8SI;Zu<LXoh*`vBBwiRe1!wuHIbmQ`>~
z?l}=@+2Q~p1oY-0C?caF*hXyHNQ#bl^oa`yrdcpe<sA$>P`}-?oW-SWH+&k>xenh;
z{+e|kWNS%H$#S@V2wd+Cei(oGZ`ZT5pE})Sx_o}>4rTmIaiNQq2N~|x9b&m5ce(pz
zdV8mwl_bHqcD2!pz1FU#cWX4L*`k+zbiM1WGs$L)pGfwkt=G?ouu=AK`&ybgno!}v
z_?|&NUVba!3E{H@g*a*5+A4AXc_x7*^9BEt-1MNCHnN|_a#6>Eh)8mybiB(F^7E-A
zcooL0zq@OFce@vPJo_Ap#QXYessxO2_EE>Gs-7jg2YH)!7381qMc;Q(d^Q9Ibpm&G
z>)gHWQS;=aQ1;Cs>Eba|<j`}%NCxC2RrfC>AqXf?BM8c2V(*9wvhF>K=x}U_IVvhX
zaQ-+`-2j)Ty!x0-!87VwQc}@=9`4{XbQ#hX9j*fW`>R7wR+b_?b#X;YOTolA)Ir@F
zON~TT&_KRFo<nYZ`@%gu?yAp6dCS6GBR(gsK|KN6GMZ*hnzo9}l{8&?U^v=X2WP8r
zs{&3hB*l=}U1ikJS%iY)E&?|#=DRjDwHM9r?GrbcGWMPx{nopz2xZ%D!zH_m?f(7p
z@orj?PTz~i;Ki{ijHbCEbCPHN!4Co~>EPOJmd0(kxM5>9$RH!9VYxOH6m7E>qw5>O
zmx-5NX?|PX^Ikx}V3UKe)(D!xVwq)Pfyv}0BV@jXxD~oj8^53|){`Tz^Lr<RfL@wt
zF0%MGyuE(f=1||1a=AZ2w^v${xg=h?&=nJ8zt)0E#2;}$h?1hum}50QP04rP_m|ZX
zlUnKHlZdx-dK4K-?9O=uC%4uH`4G5X+9{?1C!o+It_tn^9FWW5JTAy!MN1dxTC{KX
zfV18^;wC=%d$pG^vC~7N_9E84kr>w$Pk_YcSk9N4CQlG_g=AK$dP2!1qA4LB7$3)F
z6I=a`*hO5^`5wh~+z>IkXrV5qo3)V#62`Uaj8;<Vyu2hYZq$8sY|*c+ITX=8H;TUa
zss+^tV;NC8mf{Pa&s#V276BZvC}GF7E6($jB%Y@8!d~v+l>O+ZB@`0Jc}Jy6O@E2b
zN(2Ow5BVL$6{cQ;5#EXMUDDTwDFm3hesj%pW+F3wsD;9h{73Q9X|$NWiNfH6a61^R
zw8mOeG|OkbUPnpB@VEg<>m>W3N65g?V|;jXXLT?)c`0U=`w{A*5vY!s<Wqm!&7t&5
zFxKhC6vI2P=F!zOTH%Lm&)=B#<G^Dgw{1QGsNB=i>1A<#3Pn!%N43ep)s~&Agp{JX
zipBX!Cs+3n3Qm^+eM$|+jq@VmpDS+F_Zl_sUzv{tz42erI+o6Xzt9CG<b(k;SC^$X
zHQX_(j@JiU^&I;fhX5|Qer>`NAqnoo)Dd3UQ#n-flPDgL<fG}s731N!=)bf)Rm(kt
z!8VBnJc1weoqM<ueSbC&uN)Bn5f4aKM`jlXQif`?39$B84a?BERq6@~+rUN+BVv?*
z%N@cx{5k10*(!&3a;lKIrY8MHLQ6@r7i4jDckuQR{`5)>2-o>6`D5&dP}J{Auy`?Z
z1$QT)&=DCL5gr<G+S`7$k!9q7v#@)!hC7aMPr}C(hP=16s7D;ym=Jvnl94%Df4Df$
zPTv||fYIlP2T9W%y@HN&hNB-$h1+l$i2KVUsCQAwA?3T@<k?eNMUT-7s$CcRF)y=V
zJzs@+_K$*q^Rr5|%Yh$2v)h?I;XC7hmO(frk3j20wV%v2WILlAL;sMG(+ZGq5BF3d
z^tB;abm&q3Qe=WczTPA5hUJpgX>%QCQzC6a7!RyxlRa<tM>x1IqcW6^7(})0vPl>!
zFg^2K@da$rA7uHX9z1U+zD@j@i>C)tMQxV6^cOgDQ)F`!th=`>d0To4`-C;{oiJri
zv*CXqHJrpXxMvx%iyp095KsT|2}NuVk0y9BWe!|rs{QsKxX}C%GX<k?a$M-@=!(HO
zEF>B0KbTt6ncl3%admY!w!nM)Pj+#dWhL!jCP_{4GUuHQ*uBYdUn&n09=$&!1q;(X
zw3$FD2c68&3paMJ7UAQFqR_}uSMI#vH@@gY@!iaU(02cB&IzCW!=Iw3E8cc?HA6er
zzJU+j%HF0A_zMh7*cp1SE7;1Wd+y^CF2)nZjaV9&iL$V>vpeqI!I|pJrvjr(`DDQa
z8peJ^c++u4mnO(EiyWNQ9*ou}VNeJnQUXpSxLmk`&0!&CgW&g5-kv<0^c*0Pr<-Op
z68vvR2QFF^sMPuSH~gr@Z^eplJ^%RG*AvZp*eHkOO4uAHC3A;TOj?dZiG{|ZS*L%$
z_|vs<et|W}(O*S`igCtVdxp+-mi!ouE^&|jO&~c>O*)u<VUWl{g?QelDtCU2{R)5|
z20PrX`HvWD%gmk5TDE4)P+pwaiqq`vRUy%#%5Kp6_$#V(1V0A?;WmHg3QG8!B@_4-
zPr`~1CfeU88<;wFJB+U36c^ZNX8t<IY$PFK9+JpYzTfZGcc9GtqSe}_$VDG$XajrY
z8gtrlL!g;1ao+q}JUh!z?&*n(3k2kR?=$c%7Jop^+R6}K*<SYOAz`7zXosHRk$k~4
zYK6?DRxdkSm4&*tasmu#c2Zg1ElT+aDfBx><;_|mFG&NkM!xJ<YmsUP@a*^O9+;dy
z;^rE^xZl?|aJ49IGQ?Mbm*1^WZ#EG_J74&@N@&OpAAO%0o@&=RhHg$0>Tg!H2CZKZ
z2!&7!>>0=<Wt7{P$435GMUNm6Yk`P)Q-k7G<-V+7U;oiEKNMNtG@k(}TYrj{;2>I@
zzxB_QZ6jpCg!N_0u=@63Mmj9K*i>xqv=X7b%XZm<y-<E6C#ta49KSG)-Fzt`ZR-aj
z*!AQzne;;x1Q_Q`6zC`c!C}fOYQp|=#QERKjbu8<O&)acu(;&1_H#@f8Z@F~;bJw(
z82k#QGqE3*^88$4;hRqbQpI0u2_BCrCV@^3&z`a$FAQPANn9YFoF!+gmg~5$1Dta~
zWeW{$?w@2kN4K1X-d8|h7mS{6t?Uf>-y9SZmS-N)oBXP1{vZx?IIP6jWn&+@RDu7t
z`wRP&zI-x!S&k<kI0#CwwA|J$SMTrZQ*YRe>RjmVYC-WF$a56hW_@@Pk}u2@E2=IR
zT5kxNk+D=MlId9R!`@k^vwN<r`K0~B@t5$u-f$4Q?zvmG>bQiI6^wFIPT@1JTzjt-
zSP=5yF%NZ5$|V%|#1I{PuP-`%B3Oo+YwpmK7xdh_;YAb<ddgCJw;1{_(-F+|;E!^i
z^Y`S5n3=uS{f#Ie+W}`7%yKWsnR`@704k22wH0JB&zaY)IkWZ7|Hjxs@q2yzYKyGs
z9xw41Bpp-C59%tE$kyyba2!Q2Z_a@{Z2~b7Qf8{!8vGmp2w`T075;1i^i;kgg@sdh
zY_d4nYDbAAc^A|hZyN72#X~JlTirN9Uc;FCCAUvK*;=0p6(mIFsE~;J8wlOHB}HWG
zJ?k}%gS+gD7SXhiq}8}2h<lq7We;eK5F#G1knC(GN;cy+u>A5|$t1mJYQ!zJN)__M
zcV@35b|it`Pr2n*nNEE<PrEsFBtdRa4XTk|AEt`91h%nrpXMJg(1EvI7#y=A^V@Ci
zezI^BD;K$LK+wl(Lnphftq%3Ja7vG}hhY>XZ#(05>(j_N+U_SQUALO;%7tkFLOjRz
zzj&d3MV!ySUb?zBRgHGRT|K?6F+4P2W<lthv?(yL7!+|P&R!}{n4;qIRsuR@k;M!z
zL7y;I6XG<f`r7>CDb(UT+`HeyW_Fz9beMS??k*cxun3gs3VuTvs`Cy>WGWGP+)L;A
zm|$YUI<<MK<1K^oqnk_bz(0Z%<@j7sM@+IyScW?DYHE*#ey8htq}RPTDOXirRFrgD
z_!&4aU3a`_=}%Wxqq+5Q^R(Q44O0_-ml?b%382huDvCn}@J8Qy^F16Zdb3Is#U>u8
z#T2L!dN-5YdCn8k=lc79FE^%f2=;*U{XWiloU})NjV4k8eH>FtkZYsDrz(UHZVsE(
zyPrLae`-sD%Is>lB&SKaB6Nj$141>NfimB-d!Ub>D8v3r_#UZR(@}rsP{s);CZzND
zc)6p<v>@^-#eZX256Xw8=BT0O`3CCgoX-X2cwJl=^LE~_;C3Y>icq7NUl}%j(pe=K
z_1h;iOwWlWV_!W^FP>;E?Z?ZfhtJUfb65Z7=Vb^=A*y_eZ?4jri6E1(Nm@(+UpF_I
zQs%t8p(<?ZBcJUbuaE1U1dd>;#=&d_^hrgwbkW<HZc;VXC=_iX4>go|z`u}8v&zbs
zFd`wy(e*=$bW5!Zd-I)XOU|J)Chc!KI`Q-KtB1iGJukSDUU_?S==T5VmOHqa7SkBM
zUBE5SpX-)iQEB-@PHJ`ff~FfNdz+~RWmNYt$qsuIU|Im0d)PiN+EG*yNsqPzpI)rI
z+~wIG27T$Nk$o{XfBrnJXq{Ee@3sMhRORjHOwL84XCzGARkMit`Y4t6`r_}O8}|7N
zF8LU@ePkvnyAZGNm!qEGQXqP(+gb?W{>i@)@iFVI0IIMknfFV}$|EjkfSP+w8_{?5
z)CCG>vESEP9t7cq4SiO<Sex4pPrC~$dyucMrfTnxMHV)-Ye6hPWbwgpDwUvl=q~nt
z<gAOU-1&k#`MT465IKqApiY19?om)!cxc#}_%1(PJt5{U<R(JZyQbG!V1hzn;&IlO
z!52(|eCZxF`7dqU=QAgOp+lvaojjNnJ((mIRKulw%%i>7R4Ybq#g#Y}bFQ2hWY(xX
zd{|w$`n3)njGS#UZplR5Y_&*wF<c!3Y2v|in)F1VCa!bUy+K0ZJZlTsVcLrdtBT;{
zSFT<sVLy)EsmiuHmDvi{h}MXt)upkp@?ud}jn{ITEJKS&M?kVLufZS5pFV$jJ+=~x
z&EsO}CiLf{qLnT^*>qK5zdE;n4ry2xuC=vzZ7BlLhv@tByDlv5Em*=}_Bg6;NO4=!
zLX>yMl+Z(AMz;dcGLx^EQ9q)&iC;v#n0eLD2>A$gpEom!e2@qFmKxS;`h2MRTK0js
z%xz-u-j|~kIshmkrw6yz@axhunlf#wZ!VwHEREvUew~`yf1QM<+JB{vW&cXkXN>cx
z_qHd3`n|Y<iZ?CiCR<eG=oyeQoAc<U*cGll*8D;PvOm6Dxl`Ml{z#9ayxeo3xIFiq
z={^Hky~f1VtuLN}2!-0W)zP6MszY^kcFIh3byUD$fF_TLofQK&EC6j?B+k|M;NTU6
zJ7wK(#$<_>IwSe&B2-N6`d{zBLf7R=J?E7eXHjB7xaFL3wvYSnLY3OTk~Bz6SfS)u
zmdEVg-7G`Wp_w`9_(>-R>i4?a1I}!QLPC_8uD|wlVUMR#SW>7kTNL<M{B3nb?vS`)
ztSo6Lz%Si|i=yk1q~)WFddf`?+ZVLSmjCaGc5y=@d2C_b(90!~a#-JgYb7cp!>e7j
z_aUpr6`dc1hdt)K?AzJes4YLv&u6%;C)KWt#=Wwe8sN{&PH8)}vMggDN;-9oxL<n_
zfKt22^tlDq$uCGy(AWN1@g%-$R+1X*2XoR)242OFJ@l5c+IyN>_wyIMnDTJvz^9N5
z3ivI9Z7mo-gm~jvEWnJ^|DEcN?5z*5-e$E<wuKwNh|_C7Md+gB0v0857^UL4dumkR
zyR^|9d83?{D~0AxR<!PIQ`ChjN?+~oUi|?6q17%I*Z9w#pIAR)3%FO_ByMObcG2t4
z24w1-F!N#0WPeO=;c$pxdt%BqVRT5>@S%kj4d|c0obTtlI`atO!2(UD*M504RkuOL
z`32ourq^9Bx=^#pyFTuns8mypgPq#s0qYjmPzg{34}Ywdw|~3Hp~?lrhaAdIR787=
zBXGsHCbZ%SWDS#Q*c!?VX^Q?~F4q?V3QX*ne;y8PRG#4neg0sIKUFjnG9^7sc&L_g
zY>amyHHD%>QEgA=N8J<~o|=ZwaoXA=&SF$b?@FPpcqm?67*dL}r4WM!W#4*~&PKT^
zkpy|CB<cml5s_r6-6o#t<e(na=+nTvws_u?W52_dpmeKIJnXTbxb3NC+iB7apaMxR
z$d>&re)88CAif|lM7h;Kvhlv!^<ffY;FN%G7fZV<_=#)zxUp&)2?}vO|JN;9)k%p&
zEJ1&>stcAAgj?dy0yxM9!K4%}&KiMs_vs;Hx-qo-DXTAOy=#Ur*ow~Syy8HiCyn7B
zZ4B8x?o`u&As-R<ezyH-<#S30l({Y7Eq1fJ7~{2ea;)1M-$y|M6x-JWGl~e{Yhns#
zhMU@|KTFxR>PU!VFB{UEW_WGuJfu51A~I5ey~K>L%-_Ij6r%2eTHQ?*Lco|%M6C?5
z=nw@*algUx<zek;ad@@03EtGWJ%=IK$dZdI4W`ad20Qie-YPKQTLXrTDLMlNE}|V6
zA)9dyzcutz6H|F=s|u(36!;+hxda69TV{JS{fy9aR?pj!eNm-2W#vz|=fAL~(2f0)
zo$+T(8Wkhf+uXYQ2c}Oqh{kE{`Knxk-*r1hbW<QbE%*5Dz#Xzxf9^G&!=a-Ml*^D<
zMa)Ys2`WVuNe#IjHQ*f>lT5lQO?@%3xN`M(rT6;iNrgFXws2l-`r^1{e(yMjTq;SQ
zdZpE9;z5!zjO{Z@NbvO@_Wb&4=*P{;0C1Ce(@F1$jc{-m6iyaWD-9t?@t2Fe?Uid<
zLcCzTH@?0I^mM(W>gZ{j*DVmLbz~>P(M#$gS!aQ)D9ktwK$EVyGg8ItkuaHuMY#rE
zABG-;vK<k<MaD3g6jb|au>!@%Dz}Ln#CWqg{9B@@UTpexj-F~!4!liI*~Hp5Pm6U|
ze?7e53}j9=XZlnr>nU9$#X`q}L}BKq2>XGRH@U!t*WtqGT79Y5O9Rc9;K+UHh&;Dp
z$NHmV*={`G)mn$%!d040T<3p(T#*~Ve45<5aTXPZ!nCy8hH>VKE9SqjdtA@W@!*4~
zzz#EuG1JVi>R4Lb-8g96?IOMzjS-fzqb#Xe_at2L{M-6wyj7XrE=2xUg<WtmSHI7r
zzLe3#x8z=46SF~|8DiU5Nz}+*7o8(nKab0?JpDER^zP!@D;w_z+T~eW(WDMYAC={`
zSX*s1aOrNEm@xM%Q?tbJ6kC1+OG$BQ-Da$Ze}eSDgbZBFkDRhAyiBHpkM&-g+gZ|)
z3-?9iN9;Pu>PD0)NTEiCd<a$%6S#xd8po)K*ia)z%hauVs7e@Jgk>J-v1f15`O{td
zK`#N|G+o&^tJ<b>yw`jjGi8f{5;e8=H_my#>72Tt+bdLR%x8ttYPs(h3$93;yifjm
zl9+%!zB~>tnC{ykkO6Wn0}d(Rwj(1?l0Dgi(vgT{?R4$Ve|~hxo%0ZQUE7DsH37q_
z)8p7hlKxoz<=Elh61kGgnknwKTk#NM4JM$@3e`qfUp;Y_^AT4$QcWeNrn^XHppY$h
zP1Xx}<NxS&TS4}NXN#Z7at9p7dB-tc`6mawbD!BsmCi?KM(n4GbS&k_oiZpQA8QW3
zv-udV4hdyClPz5XY2HrI&SGxif6sYi=fhj0bxYbwbYk~kK-gJ>k_rMJS3ofA4Fv$v
zk#?ew>=<QVTQ8E0$Rtx~h$c&~3g^sucfP2fP?_&kx;`R=w%lC&xbw8`xJ5L}+_e2*
zkuX>?yO<b~-jxJ}A&sI5)B26+?&#}FZ+k?Fcv|?>R&?hG%BB?PRin7a9wWy8m)rW;
zdg<fL9KEAthPC#(y{dKa!~5?}bw_|&jQ3?%Qt&_i@q<_8b5?25*JC%da`}5(TZq6s
zeP&}Tl4(pw*Y(Me*o}@br)rj;fx;eb<;i6S*~XmpPwVN6sJ3;BlL@+&-a5Qd9(pSp
zxvI<5r@oX`U~c00W6iAIw>w6`5dpIIbyRcgVzd0to>~-8LR+~skyclOVgy=<F)(J!
zY!9azR4^|+Yb{M{_(Lf<Iud$DA}XHU+noc`&1X~RJ>7dIBVf`9pN9#)5{r{fx3uga
zqzpztmQo9tjqMFD2V-E6)(0fRx;cV`-5O)_Q<idFj)eZGHc5T0<a;8;Z`7zoS!g^;
z1}T#YvB-f5|E8c&*-as40P-pkYyPwNRptWrvbsCqgfM+d2w}i%B&2ZH#ULk3-=eE{
zLDYBcEaZoUb`Ff>oicURGVY3zNyztiC512KZF126mI+YIkN<hC<`A^nKr8XgA$T3i
zrf)|TI6uJD#njjmB|Eb`&Y11X^YE@Yp6}s`%Ux(Myr}{1@({|KNao74o}0rpM>SEu
zQ0%@i*LEPh+Pn4j2Rpdqd>ts4GH-3^@j?6AlJld0p71VDZ$<)H%ZMXDbvMM4IS0mH
zym#u>RXN=D!Qv(FBqrwiU?ZZl@Z!OD!hmeu8Uu35c{}(w7Jat4h_dES|1CN=vXoUR
zQJJpqYz`oF@{`^|EY)Gq9t!kX2+Heps}@CN)J;=eO-jK~F;eI3uD%XpjCP+Vr)Fg8
ziDsbQbf;6Qy%A;k)pE-?9a2GH-gxb1{7%T2^z6{Rt$mwpbv)v+90Ps3Z=PSI{@}-C
zX36X$=PIc(2rOMYn^NkOiCoj&iPBKl04$>Z1`vF;biJ{E=@S~k5~7s7jLpOC9GoAY
zAA8&RuqnEiG#cv|=CE-i|8_`{2)OJ=6Dj0Mj89A~Zg_?&8#5|fGfO@UVdJCoH1vEb
z8k-(dBvz_+b&OEEnMQ~Koj)9jhN<(W)@;^v5MJ(oA7~6LnGkO6{urY|kY{qss*sL8
za5LfKj*Omm*i?FUWZV9Xh4|q`*-sEhQsM~e4xI_h3DK2SXFQP;+0f5XU-Xq~W(Ltq
ziq&1`?&!<Ih9ZI(aw66zZM?L&;x-Jf6?r>+>OExMm-C-Rva$fT{D+SD>FO~yWk03>
z&K!=A-WZmCi>@b|F7zLdPeLN^!{oP5fIel@CBm<HvlrZk+m6*WCg_FCFf#AJR$Xlp
zF{;zAXDy$fn4k)MY+~uwZhryupUyl4tINwr?D?WEI>Upc`Sm%Uv?j|uI*MTo9Zqj6
zZ1+ajfOz2YM%wN!T03Y`vXeI&B#&p%RU^GGpEC|>Xxr6k_FaWgKF_86lA6T^5XToJ
zJI|Q4tHa!H^Q6R(AbilCRu(>9!EF)&_av9ICD@RMT5M4c({A>bqr#sE+3UkdD`H}G
zcjf@kYBX*qN&)eyxFmy`E1^gEzb}*1C=qXwT(&Hj$YDegV(2)e4SyjP`%om^&7bS(
zuN^Z<KF8=cVu3P)`2zryhzJUV&)L)5#MrCik5>&+Ym>WOC>R3BTquejeQhnA{mjLL
zMpRD3z=T~*F{987sKHZ~q^sukOgT$e#wko);Dlnj>!`57WNq@xV|$nH3YlP{=Gj$T
zwb52stSM(8I@}?}550gyN=(SV;z=dO7jxwc{r5qIFHuL39;zp>s|axC32Tiiq{(Jp
zH!egG<KE)e1nv1=EK+c3D@YjKO>_n+zH^n{;cyqwWOR4dls=xFEIiS@{kGkQt*0Qa
zAdXVFL+@J<=Pibq*Qvf>?+*?mCrn@VMVEghxc=d`gz6KIZJg$47~XPV_og9-eRB&{
zChWSDlnlv3n`~-MaVr2J7<FhZq`RU4FIZ_GKD(DO;Ukwp^--IH2pLZvSE8%1?U0d0
zTiN+yVfUE<Ljt#;q+d^trGb!?M4~4lg>eVbeR*1G2H5AJDMl2CVr|+T5Eq?Tua-sD
z>rYvLcGPq02GjCWk>8)vav&xqgh71F^qWh2^5~B(b!&RDa3H8@g*{T;>>gn3-Q(~5
zE(X}?s|fqF1uxlC;@53fzXf_KU#b>!9gt5ukb9UV@GT%Lrhh3_d5m|#HvqqKQ8zMu
ze3Im7!gZ2^?HC@}8`(7}&|nI6Wll@yqfy`--3mW$|FqHQ;HbcDPCcQz(yYdTxPZ6W
zKZ8T=KrO#d0E1xi$NdjT)J<!~gJmN+73_~>(b)55h@3Qq>cQ=<H#D(>P&|K1MuN8S
z`urICTvkf#J3Y55IdM&cRKf$f3Mp&3zf-a2Y7-@BHQYNH+^!&xDU@PUfy8p&surc(
z>7UK9G);x9UC;~$JiTo&r(-)>-0v^Xc2eB5OpH?p%0~{<jW_Fjh(SbHYGz77BeI%#
z38VC>+29gyq|Pn%W(;wwc!8W<x<2hd?@Gwo(b+_FiJ8A5M@TxCG2b`o3{-Xa=UMD*
z^|=Ku+J0g85P?3dFYPL>KXrFFAjzj*X_So@EKb%^eaUAZGF_zguH)h^_XhYk$Hw&5
zB?1iuzojpRL>9N{lzpJix3{xLl*@V}kvsAq+~h=u8izxba{IJQfBGvJKat!Coxs_7
zQD_I*F&_5EhwJB%K<8dbtP~H)^Oy_uwica#_NiJMT5%{M1S`q$XUDca5hKz_oyM#d
zlWR*oI`p7g)K5s?ngu=3E=%)BEfz^#?&AY0o%MUOo<zSGII_{CtI!LZWXAK*59dD)
zpoen1gSML)Xeg{GcoX~ub8=urAWOK_i0pkupL6c56afR5E8ou|B@NyqzcpAkD8B5>
zMV48gD~eNQYAKq8s~Iwu%^wA53Vm7X)F~=@L2U4S;{U~&h+phxITraXZf#xBZCwBX
z*6h2jNE0>tALqgw)ege#Fl+aZP+IUc%7ivGBz;udrsK@`u~+e0y(IYT@?^!2TK)NW
zx;)>va+rbk?YE}{(0yk2<r(6}5>Zc4NWb!A4qt08cBovB{N?I%Sa$zL6_ja`_&4AD
zn58R{oah~jcC~6jrsJ&Z<-DRYJsSY!ZQ+e6nmH;W;aItJCt82uI-@NQA}eDV`hyti
zDIQ2oExUCRdmIvZ^$%BSbnk~5XNA96TNIdrPRnrM?^wwyfw<Y!j7eZ2T`u;7f2uv^
zm<@7Z^xIM2xsndxpncgi?(&cb-nMDa(wh4zH%NYW>uNbTiG|F1${3p1?++ML^9<+M
zeee&~XQ1o#EAk~+QChGSQ|Cepd_hHAO}G{zo*q!O<~vBxwR;=Qx<4XzMhZ7Kb~`b<
zz))z&HW;}IH)`Hy8F^Jh4TUh4xpVAdgQ2T&yOWQ;jdj{*-y*Oo^F!O63zJl92y4kp
zF__^rUimML!paX!h5l>)1$<4_w*U$I)`tm@bUYD7G83)ja{1ZhEyP^pGw^8$32T-3
zPESvZ?apsHRnwLjB`qIz%&pnkk7z;vZUhzfJEKcOmc(hmnjT2s9V-h_zU~hJU80Wj
z4mStW-3xgUZDNjM)AiFAaGR;<arNQYU&UPNJyLqA1WKNCE{VE2aHeGS^=A7H5umgH
zw4>zsSm-E)=4kLm$;qtQPEPsnWlsGL%JA~k6F#%fVRne%k%U|e&G#k86u*bpP5Y$N
znH7d#<xkssBfXK8xgiQ)-0>N1$n=lZ-uB>P=7KmsC1Q+XX8z>EyD!ijzHUys>00UB
zt+n1H6nrg@Rc2lQ;yrQYtlt#)@+Qxyn2O%Z2zQho4C6~FjdKf{{9K^!>W#5p=2H)4
z<|v*p@vY?2b9+iz22a#=xPC_Rq_4MY(|uf>DkHV+tN4)?JMgVEKmM(NA_*Bil`GFT
zW3!8mtBw7|mDc-%<|riew_rw<ZuK}wnXr!2xS-bAp;Vj!w82hW2$Ol-G&-~USAYU_
zP!ym{y}v6>=wS~JHy#FfdTq_tWDk1?ne^tLT=d2O`Tf0(kMC1vScZr4xHqmQ4j>R$
z8wQQ?>m75~mx!Cg>F#U7rEvWQbwWas$5A$fWcl6GW1@FSzOxy~x=fL9hW(=2s$10}
znj^NB;=m(#HL~#fzMh<Y(x0xc>F~-kxj40>D~abeHJ=o6mN=ym(2tB;o2QYimCd*{
z|K%t73^h76r(&fyB@_CCJwhw#&F~<b`~X7-kUeU2x(2nn`nwQf+1gxDjgs;HFA+S>
z7GNqg1Zq?iA*%t;Jio90aoL&a=XLA!$^+jx3?QB62B-A=&WDhuEvL35z{u_}H28`3
z+JenhW@oW(ffC+Z2^Bs&O@8P2g~d8uOxyPh<AKOL4tNY*kQADOvouu!R0uyW_tF@y
z5-8&4dj}zkt#;p}tN4^m9tT!OMDlPkP%%j535T=3bbV8_5p6E>F6mRty<w13-MsB2
z@RDb6vbo9zJ~cVICPk9bo1f#+g!P)TY1*cn@CWWKl|e3+#vis<l}S+-e9E2lvP1>H
z2^1<%5#J_g+q8TFa&X#3{sf+mhmLs!hs)*J(DY$++vrv>GSG^K`ZicsUCwc;CmXbn
zq4-ia=+C62d*&$BMwIz4Y&y;P{n3rV_{nN#HtZK_4L>}LcjYcHvO-Np<dg}#ij`Zq
zsi7__5ZHJXQC0Unky`hkO9<kuSNZX=du)NJ^Q;Dh(p{4I$WN|2WKW?RjM$JS86i0h
zi#SDKWJn~Gf)q0pNdIB#^d+c0hb6`HczPfIlsBzB+}OCXNm6DT*l<=<lQi-Vd>bUM
zpQpaCd3kwo$Qg5qMhuxD_+?4lfmHEwY3kmfRe`4)Ew9wod7O-woRzV?y+{#n`x@&^
z9~K6(KKHI>EEqTJS}gudC{-YrrPX%W%j(%J6HTG<MR{A5A0CQDI(DdUWCKFbZ{q2d
z6aG?EKSvt8NDkz~nf0I&rvzV9N7s{<GwVmskp+bjL0^PA&Qye{@%tRb05jBQ9WRx(
z%fa~N>FVUjZy4z5{I0a%h(V)1@n;a3anINQ(g}OvG5;0qOsjHnDOy2CqHkcq=t6L7
zZ@rU!ks@&~{J6hww*^~mzmrH}98ElV`VwUoNE(~_;Tcr{YF620xLkZlJY#RZyMD`k
zwpqrC#`2jV&(cHUKMuY80FNjiT{%UQ4vB<hP-Q6=ZnKK(7#=#f4Cox*a`zz|NKRHC
z`$PK?m6XRN(}iJ5)H$4$+~vc7kg-1~%GUP%^IT#j3x=$eRF!u!f#Q8!w+(VB1A2<d
zHD~is7Pd_7Ja5x`7qf`Q@JP>R4OO0@?57O#g;?yIs|_O;N53baotryX>x0A8K_yIj
z#c7Gk>KHRp%rc+35ck01e^WP%NVW2PX-L7yP*@#$RcF&vRlX77f1><d9^~v@?anw!
z^fbnob!nIOF9FY~ee%zAyYQ>Ws`nHUMIqp47_Z$IuaC7Va32$(YPc|VlIuG0Zq<`j
zpQGevZysUl6#EYJ6bIaMlQ(u_voKxnPtjS9QdBOfnE1qIQRa*5-<qBM88m^iZ36vK
zUj7@UtftCl#$egJGn-db|LD`P^1jB?T=O#8cr6}Hk6>@FL*+oj-mHqZY(^Z0h6;<~
z>1&4@?R%q~T|>Q;Lgf!B+A`VW<srl9>8&+F14k&L)Sd%7RxssS%cT3WJ1zJ+Z=_Gs
z_}3ZV9M)|iKeBG^-V=X2HM1l#cO_d#R6$7ePdT=QZT7uA0$RV7cz5IuZP<(I;Brb$
zC$1gR>$)F!xes3EcFklEO+lrJNfb7uElMHkY$GU4-uPL?+)mc-E9bbLaHfp4<kO**
zlk<qqE(ZWS7yr2N%vNSH7Fvi6FDcqQOP_EibPaWJp2)yMMv=EXukR?7WHJTE>FPQ;
zc^FN-SCTMW0u|x&Ot$FjguY3e5+td9?v(&VV@I{<n6q=YqP1WLY%p?U49q(ztIoB1
zqH<8z^6OR1;&8g4O2&+E@%tM7=pX&kfoYv9R)9ASqL9iJqy#5{QEtt$B&c-XZLg)>
zHM4)D_d=iVKkiDNFf-bpDJ@e|)wg(jDkg)PDuQbc8#9u@d^T9Va_NjE-<ZP}4HQuN
z<(wS9$Jh!KmS9EEH-L^c7kEvR32#|U-+}h~ldKbQ7YCa5_>49A*FJmFa8xW-gM{1h
z0EOQld|!{Tm$RRWOO&E|rnyU|!f$9HPq*&sdJ}#e;GNL7<oGg}fe45w5F&r}bn~Zr
zxgR6ZhZf|>ZEizG7jBK!3mU_8n3tcOtP9LkJh6rVeDXnKY7ZsFaI$D!!f^+Tvfuq!
zJxmR~iYn9>Z}V!Ea{XT1)Lin29V=1=XfuHZHhz2kCeUg$P^7x37Ev0kOl6LKzJJMD
zf0VLu@<wrve$~>33>{cpuBJsGRU57SnZqqL!B1{`n>h<)FR)|>$IIX>8S-miuPN<>
z$r%?dkfQ3aC^^s-d0X54AY9Gfb73OLJ6NvvclnG7XT{PVW|_V6T%R$*vn+r{v1xMh
z#8D87DcA<*;OLL9Lm>}akTleMuiTh2CwkBPiu1qV-UX*((x0E@QyEt1|6Yg8=h?26
zFLvUFjdLtn0$9imC{ri=1_vhmRM#dHGWGf<t%^ZPnO}_S(TLKRb9^Rn8QSE=l<-}{
zZh{Qie>fD?iG}0F4^L0Xfo(g4y$g9KD>3NQs_QP+gX8I(k@rrU-3^`7+Req^R}>`O
za=5=co|BxKN7%&fo6Ub@>RLT@dPJM5Fmwv?SK}lO9$CA||K&I1@I%R(I`r8}z}dn6
zzC(tM1tDdNCh+ra6T(^9v@)3>;UVfxv}7g~1i4v>xD%D0+wHt3L@_JAF%4%*tddEl
z5V+@Rkp(Me9=G$r)p-q4HP}e+AgU6lJiP53?1_Yo$nz4VzQ{AGaQ19!ex^u&U|5<F
zUx@Jx<DCnY!RzU3Z@RlbH*5~1?HRrur=LRfF_JdqkojGeT;ofU^q+A!icYDei<#%4
z<6fsrT#(im{`B;tK(fCwJM9{S)D&{$5aVO@K6IhXC-@c}d_9iIM*dc1j*-EH^nb-f
zp9|X*5{!6E<73fVRs;@!@?U5H&%3G?c+Hlrw|6x~wuLTF^7^x~?$sEb&;ed9LaX-3
z3YQmbCnq%|lm>Gll2XrblJeV*jd60$g9(wx#^Il+)$`8*9*;t<ZUB<LVB|Lmg{KA;
zB2P%6+7YhY(#pnn5@7M*k8(5*3}yB=(@UDDIqc6DY>~ke*z@EmvN5gX&SxDkgXT5g
z$M$#q-tVXArxpwp)88U8IAp{&7Z!Z!^aQW6#rcqwUBN;(NVM)%<vyh3dZIxiY+&%<
zDU;A$Cv);90CcoIQ;CjbGRLg{H65oR*YaSrLRRaNxSYE4a!CllXvwXlQ2XmAa-c_B
z%~x5Yq!^_QO5Rg;N=XKb0n?|<{r>qC8$rq{${at|a$i@UXj>z%z~cqtO7Y|mQ6`-V
zWK!m%;cRA=f5VuPa(~(1*L7==7|)gGeee;Dse{`~x?8Pu)e-YJ>rMI_W;`9t{qCp8
z$vOY3a^N9<&QI5h`M%xi;1WmRdgGLUcMWyyzU)KPsgj5j{X#6LlE?sJ(_2H*Yy*<V
zy4H_L9&?5nexLXoC9h(5_7FY}qJxPFfxWbXK5P;VHJ&^*f<+ihGBOV;aPu?wKg~57
zeZ=pLoPygSC_-T`9;gRDkeXhFqu^xEcmjNpq|Q&d`<V=XdXhY%Uc;-W#d%Z^UXn)I
zOkIhjkemmI<r*(TZEfJE-PvZGZcan!<|oAny_#@ssk?{v8DpmpAn_gv9Fvhzk(&ZM
zF7hP0F!kKr0e^Xp$_c1Mp-oc@%8*BakCc!#ZKsoxAgJvr*e37`krUwJO!Ay#lW~z<
za2SSD+<WakBSJiDi{G@C3)Rzp$D#Z3O>PF{_LmeNY?>wg#vYJiJ>_l{Z1)KHDur*Q
z*zi_xE=Bbe1Wc^xl$;+LYcR&RA6!EncYW7(jfl7fi2*Y`2!T)GH9xS%r(`kV$4^H@
z!pan72S&=O?tJ;AuPKoXuK~HMT(6cejlFq?ZJwHaag@*|rB~F3nw>M2Czt?<KRnE#
zkn(G?z4NffXC81>9OUJ`ldW{Oq%nmgU(3Z{xD5ssx4hI@>4{_O%Xl8QI`10t$3&G#
zgXv(kdh?L$10ht`%4CEnJ1ZjVVUERj{lVQblim)aft1g~QxEY8sprxP5xr!IkW@=%
z?Sw4xcu0WZk~2$2GT~2FLQ8P`jpMav5`voG4)-4^WdV1k<$@O`t6y-F3w(1eq#a0}
z2+L36reRj;`AYIzdOtU@2i4Qj5>Bw_$%Pj6YdN-!_r`&KWfzk2-RljLS7ZNSwiLuK
zI)l8%!$=^`!8lPdM*_KZA^Q4gN!gKOv{!sNhn#z&1%tz!V<IN-KW=sRvCpfWxHLEh
zU7P`yfbE|WgV58lld~N(+l|on0`5mxyDbsMfcX3YX^8!q*!`Pp{Z^@YaK!hEorrd2
zglEoMV-?ERI?C4Qh<253hc4xkg<fb3TwSI~yz;ZNvN9p`^r|KLj1aNgi84{5qJ`)T
z_6D|z9K!8eD4eqW+2&PLFkVfdRc)ezoc$JH)W(BNPw7px$+lI1Ux7TGFtevtI?p;8
z%_L5jL|=%~D{z~(4CX+;SkgmA$d5HJnYGnZh|oKeI1UL%=cc8^eyl$JIiy?}kY8qv
zD=dse8&;iRS%}FQwB}5_KBt^^%&UG~kO;WaJuXw{9u=lcAylY1_w^-DV<GV04pHE0
zcJ@%3r^g;`wT0!8ugv8uzs*6yWjs8#6rJ>TTRW--^%YZxfiLoFYkebRHd0T_JUk|x
zL$|Q;fTVHnXF??{RL<4bi<z{YQX!_0AhNdN{gay<kEND7M#}P!5}eSaIpAuR?Zwid
zstot}Ti4y6p%B~qg|8JbMM5M`0meD|*lbH4Q6YoMAq8!kr~>rA353w$DFFi=cEhew
zgSjL$OB?cGpav(yWz$`UqN-VXsKkFcNs@WQATM1Kj(iK2O76AvzD?@@L%ztcriU^z
zaP(PYt#`gzxZskXBAcSuXUZ%FC~uC8)M>CCjPM*Aa2^czd++wk2i*dQRv>R({Y6^$
zb_Q+H3`Ra|Eg~mPh<kLbIu#ckl9YNDmY*@e01a7G=B8?*ICUA~P07knt-2rEPLZEr
z`F%;cA~$?3S_??o6@~Z_XU5vf!8ft;)WM{rFZ&Z%p2ku=9W<@gQ24KRBTBK9j$jY|
zng2kC^+9uCHtAe5-e($ETU#Tfq{DG3Fa*T3uct4rip4ybJ2yLdgYXp0m@p7L4&$Vl
zvowM9%qkPC#aBm{2jRFnCYI=UXV91GR`@5My7|)DNgBQOG`V@}v-U-~`IqjP!v!vJ
zWoo6X6O;HekpZ;ukuT~J7jfM(wT{r8s|q-M6bFh`>QMyo$S@IIboI{JrNQCTTLIA`
ze3n!9Mv0tfCs~FsBd*RPt`su{gc&4xPwLDvdo+eU<&85r)gGe1Ym-nB0g3k_NQw=2
z!|8tOrU(}YGCG6kv>9`XExxIQw@zW0YhP;lJ-i7hC{T0ruR&wxyKeqz8+$jEZ&qMI
zTJ-q}TusZxmCni;FA8ONZ)^5R=7#Rg7YtzFxJI3X=9k8+^DnZMwBCcsiyJhynIWKp
za%xx8a2u?fy{-4KMUt~3z*g{mXVwReJ4BovH^OUbG)d@*$3`E(NQjpsZ~fw)8~;Kx
z_HOSm!k$#c(vR)?V?)Lh)mYN?=XvZ02`UY*3pPyqwz)0(goVmWPT7BNh`f?2cNT2>
z(D&u`DzvVSbvAF%;$1UZl9D1W!0$_+ro4sdgukGci~Lj7Or5d<Ct!#~cp{7P?*XAF
zJ~#+yn_DH#4*1q99*io~&rvqK2P-x#FwLYh<4pRipKHvn-;SX<zoPC!9v+`ZwMEq(
z-y_yyR=N^Wwml4<_nCIH@tTOPnJOTP?Mj~8q2eWujijMd@?xgwz|A}RZVcG({JTp-
z-i>~G@*}LY`qE`J9*8b(0Gw|ONl043C)9Kx3cW`I$c>I#G)~d<o0dxYo4G>I$yWP%
zM+XO0nw9;1E(1vlEsA(NuZ@0EKvYz#$dGl7Ebb~S!5AilkUVb17UwiP_r5w<+3M}}
zWa@WjBW^34s(AjNJwsCZSECU2cOqGRf~<U5!&S@0BGuku08_HLV1k4OqYLCn10tsz
zcnG(fwt{^r`7o!%<h&YvJ!l;LGr$oErVND@DoOCIItzN$dDUUOy~>8XK8rIgi73`j
zWGD(V?j|i$P}5n>$hLQ3w@`4_V$OcaT*+L-keiLQlu-_y8;>)*$Sc_DzzHVBi?p_r
zLjLzJ=v^G3{b3;n56a7dEn?tBkshwj{PaAUo=A}`%&0dUDZ}?!MUU}hq`PR^)=G&(
zP{tQs{%+apF%4#>?giWVZVM|z`uWoT#<D+^-=S0tr7u7!C{(b$$vW;ECvZ(lxUuU}
zilOR%axH|~yfI}HtNB8F{Qi?IEmpj&q?1lk7ZHeo|D)-=|Ec`{KYolv8OO-VItSG$
zdy{c8$_f$Q_EyN=^Flbs$VzrZaX3O`o?~`I;-KttjO=w}W%E6sAHILUx!umW&g*(T
z*W-DAXj0k&gN0S1B>e*VKG^=RbsB`yaFrhZ!XY%~x@+V$_AQpscHBa2`BDZmb7Y=F
zR~s$#xZpwC!z&E_XSk;=BV!HNSG`OIs#|{3y>>}auY^U5CP(2^tMfsK&Ju;M9(h-?
zW27F9Q!3;JbxAs+h+6AEh>C#vO*SFH*hhLZxgwi@!eAf9#F#hEGR~Xpt}i8@mOO(4
zkIqjY#X{bK*}}rv5(z3>H*IdMnLvd|R%Oxu^=ijA$Nws@z%)LaQRa5bcs7z=|M%nm
zHf!NoeB?uVqMc`xQAGJu!O3Oi!>v1#7|$^*e*Sj#v8Ov>t7hwJA3qPeD5?`B)`ZIc
z^8_X@Y$Bs5AGOBoP;=Ili&&6C(W8{LJp)yHpT>XZ=gk=o1J|_{rQZ7vbiw!{L9}GA
zJFj!`&MD=rD-MPW?s>%Mxp4K&o?b`#yZ6;f@Jl`&wiAl|)G|;pe#=jzVvSJ~`pt<+
znVw4#ne@c@SxT)51^pOhj8-<h-j0&l%N{5y5-?+`*uZ$GzYJK#LioKIdYdg~%Lc2s
zaGf=;H%{^EtZ|Xq*)ge%t_+YEB)pLibWJ}dx`Q$@1{C(BJjXv)`OydYv5ozwg@sw0
zZf<mG+S;HP`ReHg4*1U#FFE9oZB}Pa(4!G3dfmJNjdo!dGZHZq$T@LMl&+Z01&ku(
z3Q3VgyDiHe9$B(vE=WSoyJLhNsHakE-q7k~Latn%`b_3#Ey%sszjMe?2mtRr$Lj}f
zBqZgt^A}yajb>zmA6N%e)|VfC{)efoG>)$Av=Onpfy(bEa7V{$2sk-=l&(ZQ+~fkw
zhh+FhatgVfagIam9GO<O((Y+x73)x{0Ec_UXxp^3t5aD-5$%4hP+zS`V(>#AdIt`I
zrkkbc^lG6EaK5{lCE6JLY^#R<;jTi2H|pObRrTwBrYZFM6zbQL1}2MI=i$_*+vz`v
zueJj_t2o!Z+S=46dhQEg^Q6q(KqBAN|I^MbEut07BJFX3V-OEs-()Yo8zayU1>jnb
z&^;S5!S{0#9W!<3Y0V?M?yu+GYx<NrG4@}ke*v4Sc5{Y3`1#NOe%^Dj?NC-s%lIj_
zBCZmC{`YtR=cPzqqtO(qt*NU?-o<J%O$t7zr>FjAHiFM!=^A4VnA4TR<&%Km=7IgG
z5uu>~JOo5V)ZMkx4Qk9HN)x!h06sQ<7p%VpzA&D@x=^^dQfRX)dJ<{Xj_~}ir>VW;
zV)FHo>nsK4(LvR>tvPrkdG-g?S0i@PK(2NS9iW)Ovyi2e0J^E3{ufh~HN71?H$Yg;
z0)uCs{Pqe768pM&h!M#{n~y3pIIg(lL%%)$ADnF(eBso@Lyy`w>yf$vycKgUf=Ei-
zvll}t<`1m)EFJ`L6_gg?+5WVAuKb=T;@jkq?N;$1jstopQy~5(MTa3PBNZuSzz(k~
zh<9R!X%!^N7&{0o<oAy8`pyk05?De!@}Hym8vj^CtC}qs1bU`&R5O=85g2S0e)H*t
z%(#A`h|0ym-Y#N#H)xG(y}t)n*WYMCF#L(m734-O)(VXWue16&2Vdc`PM%WA05<Jy
zvzZF-b01-ZPcg3k=v3#*yeJ3?5_MZ9wleM-6O(WH$g<#=YoRU}#0U+)op)xD>}9I@
zM?j#waA6?pyY&YNaT+R2vxQjvK|@n$WoJ}$f!re~dAZv)F~`rG8QL5PpUA{!DAE)v
zc%Z3f=_E>+VFJ>9y8mg625EZ&P!F0!BOt0;^64Ht=#cj>Sb~GXEIkvApZ<V3Nn<TZ
z@&T&9;xDMGD%#paRIdCkNRGHD;9{r!YLORcn`K4VlFx(4$1ix=89Wrp`z@7nUtjw#
z<~5E-l^oegiA{42yfS0+eq<Y2vBT($?H0m}Ge7*&+wV|O31LKf!C=6u;?u<wKdv|k
zRMFCnLPFMomPJ*YhSXv8i%@L865$>A;Ju_K<ZkTA3a0m1OJMy#p4shfKi_cTHk!5C
zx4HQg&ltr1I4xhNx8#fFOjK09T*~qS0k^*Et;mD%RnknQN#_$(uleGSwW=y4g<Vqc
z9Y6fJH2){R!Jl8Z6gW6lm|JnLc#1zKe%umfpw(!4d+l!<l9;$5`utYcnQW(MqEJUQ
z<q<HuTj)xFw-_H%ha$?`L1C|J=Y(55AHJamId&vdr1H_;peSpoK12=keM2f|kj0aC
zM8&t%2b+ngF$|<erIHEW<7-K2?TN>eV1wZV*Cx@%$$C@)V3$rwMmHp6J`4iFt*k}H
z7hP7QKyYn|j4jZIIC{S4-Mi}-h#k_B)K!qOvze$(n&YJT-GxaX$zidI`dm&3X$ipI
zZb1H>p*Bn&WuBMo^0jp;PU*{S+(UOhzJA)7_pe=^{KQ_*CMy>6cTjMTq`&Aqk@1fu
z%F%$E+ysaKN|}$Ekw(EgxZ-rVZ~3VGYTZk@v};F&C;4t~Gxb-j`*g0rT|H8X9pyS|
zn~^V|l~Xz^??0|gLrK1_<;=0s&}>#|oE^m3u>Hl|%}Q2d`_oqbFnDq0!@VCf=6Q{e
z*;!xdZT@bvuG%+bTdrSdIjO`rpp$j>@#LvK938?y8M-<48kA`zouZ>3V`NE9BQHKI
zYt=Oulx>pPpOo&H(jA~7Ffn<O1h?Z^ng2E27sJlOiIF#8hxj2qE8@Uil$x<UcG)o;
zs$@bS5;Yw%b9$!xnWtfNY-A)|pqC@w+-(z+4P$<jUXxSd`E(kyLOK}wWwvEWBhM||
z8i5mMb7rGnjS76<b`P)3<=NFP0=O7gIVY+_kiImva#B2q=rT)Y#paW;KzylnXaK)i
z*ZLIe@P?+w;YF<&q=}gJ#VYUD`@j{wk-=MjdGAi!kYybIfs3e@^L++z2HoE52vTul
zHs0WF;nqb9Ub`RD57Xi#>F5@F5Gep)oXqf=pNlp4y=dcfSYBl%H3Uri*e*tVi6yS$
z`?icqyQN6h1K*z{$^->vs^-twvM0XeG^qU2#RUR_p=Te)KtaualXOXHk%Jk^>PZf$
z)iyv|OPF<*X(Pro%5jR8+#+_<-E6E9eMtYRYc_z2n)<cZmLJka_)BXTkOxE(fwex5
z1C}FCk=2E1T<4C%s7-hcL(+~Y3=AFG(VHX~UQ;5l(U>76eQJ4>5bQtdCpWhP6mI4E
z=q(YGBGpwj9f|ezCH45vQEX6hmQDEzO{E+kF4Ruj6%QzTO|t7eWRp?nV(aD61sUTm
zrhawRZNMf~AdA#yMLi<`V<xz|q~Q~iydWgxXyFWn>Vfr%CH*KS%ebM9R$2E6Co~9?
zk0zOYd!Obr5F<ys9H#s@=F@oyDOgVu-#hPnd4Y;M2t6FXyk(UvhIhJSlHYUtn2b6>
zzBcy9B12#3hoc<8(d|xOcI=(`y6LN=d-^sHZLSVgtUVi7S4DxVV}0Nv5I7>>)EM;~
zMCwPUx3&9IQc?u7(ywbu@suKualBWWJUPMlNH9JMrj>kz<wiqgU-6W6b+y)~jNcD;
z7Hl^nvwh^3n97F?Wa|(2LZ(^ckZZj$tdGjMq{`6M_$;)J+kUBqv@{=tj|?<{!-wj0
zX;*{3S-64qsId29Iu1m$?mee`LE;3O5x7WmT1hI>Yz?@Sx|(jkI1s(s+qv51RZ#+3
zlhq<fr0vp?DU8T|?NloRt!AjwtPmz_l4fp0f8yg1pE$iBcj+_DkRf?L46aQsxQq(U
zyZJcM{D88x3acKehvh<%Rx5}7*n1%H@GcCgz&w;0+-z9Rk6{#NssS<}?_A`R_4Ijp
z+s`)FcUz$hjc+1OCnGLNTX>8&Kk25^==7S!fX7R#s~f9H!M4x6pfTT@-I0?)lO3#7
z#5F^~$bleT+#z8{3l6<zK6hA|9_e+cX7d@;<M)^MPEUMJ!t!ipEcCQ6EOU^OVmdds
z{RZKuc?h*0xzHvxz=o&OPV^Vek%-XDc&v@^ZfnEUyEaDgj`rYthxnTp)e#_7S#6Cg
zuGU{Oqf0!V@w!9tx8UR%&#2~UGrqj7qL^td_jJR~vVt1>D@UsyCQxfFib=`-bm83l
z&&MvqDs_ZnE*@g3PT#ZWQ0YHSqQhB@*|0T}4LJ?+*ph9U05&4k=}SQh?~3@~Y`sfe
zDBJ;YcY!cGyn8ODi^_=6()78%xNtyXonM&je14@a;OxB7!C_Yr_egG80vhhJF>UhE
z49@o<r}>HG6IB{jkmQa!b;^fOs(09^N7RzfdV@KC7Tm?M?M{pzwZ6Tfcjf}_XdPPZ
zM=Z)cz5Z#01!z;W$?`wJ@kTh-3Z-7_<bgg%zU#+j`kx#u`?^<Z`p_CMP&@a=cW0E^
z$#LCroNcK0u$W7GCn1)e_ZV&@Ev-&MJ?A2eB(>f0vJM+wr6UaQ7|pD>6Iw=BR-p0O
zp_rxG4fA?=)y6(^^$*f-3Zgr8qvq#9cv`4{KV3mAa67;=4zP$=k>&oT3l*TH{PJM2
z_=-pt2_;)w2_)gX=G>m}QGJw6;!I<CYoH|oy_vvyimq(%`mUyxZ=?;o#VBL!sHa{+
zOA_I```-p<2U}$ZH`klwtoh0Dl;Zf-<3<oFMm%xeiphwE|22vI-KP3mzedZ}ySwa%
zo#146tuGg^&i`li<mvygAa3sRA9klbG$ce(Y?uUNyE<4LRJk}NxnhUjIVc#TCS!>V
z)Uv%&%4##EgaPZj3t`GUr4X8<tG@n_L*^<fB=w@7V*A<gC5|CwN2TrR64A4B<tnH2
zQjB)fwavY+!E&tcL5xJ7)El*4YPms@Gx+|h=-~KtF(>HL%IwD*@vjgPPee+Rmc0}r
zj&OncJF{;VHT%1wh2YU$NY##&v8jftad#|6Se}vhG0>$-JymC{k?KC}@en1!#Subn
z%N`T;Ao2)*BvNx8J5DnseXI56*%&im;k5&tRsKlvv|L+cgW1fe@zb0d|LuElYCJh-
z`ix#1&d~&7n`h|)Cr2p>_m4DAt!%2!dwiQQt^6yNR*=~*|3>f)PRuDA>C$L>C~if`
zr_;S21jwl<&RAF1%WTUb*B-qLU8JN=olwklYPGZuTojRHT+aAJyHm)c+y;b%$>|RZ
z1oJVbk#-hPE{X+mS;1y*70zP~*FK1+9Z=SLftdbCE!!yy^#!!zVS<R(C=Fd!JEAdx
zqiOZmWAJ0P$)k0hm^`xbA_JzJ(xL_v3tWUTFgu47sr6y#o<F0#&ghsrytunaY90YW
zchkq8_NCZd9rp~9Xs9Ze)Qj11QL$4};ve`wJQ*=*R|ZNV(<Q_W?Gur9;4TOXCzg-m
zxBaiueDm|lio76WxpN;T07Az&TI%Mmc!Qb$*uvl@)!owcFLq_-l9GWkkH<>AE&Qwi
ze<KDKN+x@UMryYN-FeL)vp9B&QH6)OwW$wYe!ct=j$ONCy3Xvc@oT{Djg@|zTTy`V
zJKRO-v4)%uoF(QNK1TfEb+GssvPem)x1#egsj*)7)w$sGKhk4UHJ*<W>LuTNn=+rq
z#97?OW$b0|%`zYGMT)N?aJvND;&uQQfcy0>RZVS!d+}`%?@3VZJ(8fry<+kCEQw==
z0xm`8fGgs{LaNc}=?URvA=~Kme0Y3WrpF_7<Ky#=Ht2$=ht?fU)y4mOv};~yR$zC^
zM#Ec{LS0+c{+o%?fX^Ua(t@DrI`UlfB-SGX!(HMw)aXcQAJi$Hnwm!HBncB2Z7#Cf
zPo@F6_A{@LFxmIK@+ra{pGamsd7{&YiL;;)n&Ju>WB)cQ61pOb-hz@9*hP^3p!1-J
zAWiveOngU=z{TJj<QOy3?i0zQVARivYN0EXJJRDroW<yLtp{FuD2<$>EM*WwYa>2e
zI;@98{-}H4KF{XEPU-h&=&G)2v070ZSlO+-rF77E^0X`7^We2j#e>$o5C@yYXOTAz
z$nHPlNF@WqDa0b7GeT29>3$2-HMm~1`ca0l>9CVofsJ*77e^02(+e&}>a?g38Pz#e
zb?p;B<!Mi$l6<36dHZI3<+{nHc||B~$x?bO2WY&q*ZofEdXO$G?GpkONP5N>h?E_a
z3yYn=4JLuK48wtY&GKPQvnx)sW8bG6>Q<K2QAm;CPJ=!QeFOl@@>pR*yu^Qo1(M}p
zbFagP3eI1&=}Uxg#}HY^%pJgvB$=Ewt954`mYRkrk_bYCwmi=!ga17K8+5!qce%oQ
zx%<B(-qYV#8-7fDFOZY9yccptDva)PFmIk6A$C(=!?j7_xAD<$spQo~afXuIh||61
z`Ws~5BuaGwU75PRO(#<Ov_K!lOfIr7G3ODL>I?bfZ<guLt*zDUSTu8PRUmYHvTQPr
zzm-nw?EkgWWun1c-Kbg^9Zh!Ke2`tQ%PMR<v6Jusp{b6>^Jk~UG7{*_jRM~s>s0Pt
z<n9!)!KoS9Z3AQ*{Rrx8gO5Q6pk=r!Yxi2N=@VeB((K8N&o|&XF85)mt<p`yoIhfj
z!Y-6~6wg$@1y9x)f4$Ss&F<pyS|B<jwHKP}QYP~n1QCJVVm<gfR_8Hh{t|e+^-&*0
z?JIfhU>|pI?01>hez8zUvYJ#<)Ybj;lmW4$(gK7W9g~un>8nfZ)$ZA{!p`Z98L|n`
zj5P4Cb;7aKP$7*y5AAYOwxw_9-}BwJpZsQiLhm);rrPQ;xx0Gxn(`!OyPq=m>Nwk)
z+_g{XU-PW%SEN=XGH~mJqxOr<_NyM?c=gJkox!qu{uSKOglk~cD35pQ@OkxOL(rU|
zlrdpre}BJk->E=$_-_XsI!h*emd?q@m=6422s=N<9g&>KY575iHTH1NCZ8S=mx}2S
zi_AWZWy7{}!@q}OiqBhf26aN;X9ZVMy?nHB@yD5^Sj?2d2t+|@?|z!iU2`a$<%+VY
z+w-I{pKx2X@!!5WU(91WwNm=3vu+o7F*p6h3qW^P-G-_^`uXTu8R!LH)Ii%`YEwXR
z*nr<)*I%XxOr=T&au)cMzb{8u!V(jL4L><>feS={ogpmkiMprFe3`pmsRBKV-U>;Y
z$BR2xAJ9+kNK5Vyq<p9T!vZq)H*47S(uLj4G>n_A*KfH!bEJ72ZmWe0ia5IToG+|%
zHy_9J2Q=52H#+@`6i~%YeD9=+1Op@!3_1LMZGRu8YT;Mvw~x{q2W!-w6U;&GjRx2F
zBA+0R4aJxQ)6&0gOzl4gAGK}gb)nhrEdDpJ!*hAkej(a^F+Hak9;8mi8WOp<peXHz
z-{{+S)B0Q)i#=lD7cfN_MDqB(jBiGY{mRy<TXCH-&%U`2dH2(PVnk>}vk6$LtBO&_
z$w4CX+=i{+GKr@N`L+%;X02{w{<VY}KB1e{Z5%_J8_7QLuU(7xe~*gB&hY#Sb<_8M
zq2pKaUV~zjBp-^H`2HCv!kHTxy{gxJn;P^%WAGLzscbVElr$2I(VW22C3nKRkVt2*
z)x$9no}L<8ZI^rQ@3Y*ED}Z4;@-9QBKz+ujoYASz#+og^iZ!QO^l<-aYxU3iHB7IJ
z>BiqfDUxm)@qDqEaz}B=eQt$rxY_TXKn5)mMFq7IA*avV78-Kubx$xhl+S5EemP}x
z7^R{Vc5&cjVPcvX4)S9IW_ZzBO&QO9Hu#xkZ#L?Qt`TB+HVB7Z0DPwA9kH_?R+27B
zllWIGdC{QM<(_<VgMXX*J{?HWz5<VH^mnNwS$5%0c6YJurwg8!m!v~4jJOWFCjan;
z`UiyAQ~8q<YF?F)_o6oKb?(gRCYy(c3WF7Be&74w-Td9_>r1Cbn2H2oIIoB>QdUk_
z944t{G_{c|K`B5!oY=fZUsXTQ*f%b!(yDlMetH^gR%kIDl28=d^pRu6gLJT_`f2U6
zIJHqS<U5=}eyU#{o`Pti6;K;(-2v<8d1ZU<<Tt)}Ube4GT#m)I9U<+i-9hthtrMh;
zLgD1P*P)o~bV);c?a1;%IRRUyNLH#UXO`{>kh#6EJrK<uBLEd#En2c{s#h2sBudc^
zi|H=<Z5)4u?=aUYukJ)-;_TA9Rutx{hF%b&p57w2Zm4h8vOjw1+BE83VWi#60wsDd
zM%yMO+pBemWZdapT*&t^tE%qXK0Mt!JUQvBb)UwB#ON^D%L*DvBEA$kYBfu$&hr3T
zm0#`n0P=^JTb=!{JJQxXG^l;A)A`+Xw&T(EK>z9|Zj2>UgyUkN3vzUG;YXLemz%)J
zLUnsqRF?alB1@l~z0$dd6XW{TMIoHS>tZC959lTX9WEn=?_M+?Wg4<4!#KS>H@HF=
z;sb}6*oz&FQ9p`NpP=9;PyJ8e9V9ix0w<ISIAL2XSEt8V6n{X_A@%=_W;t89p??V<
z5|w&Z+#hwm?<Z96(LaizZ(dJPee&1C)F;g7>{27y%a?i&?8E36lhD?)9rUttp+N8&
zM2y)-Kf;$<z)rXS(+081!|G+Cx{#&$+ir9fKDXQ5B`PB?D?MF~ylJ$-ec>mif91S2
zkhPtPQ2&OUHmR7l--v+=G`;=v5Skg!q@zb4MQKtAwY68yCnY$4cM;SJipvWIr7(&}
zj#R#z5{4W;A|xpiw1bD3p!B=8(UCDS)kjb81*5q1&1<qVY8x~m!`&CwZAN@2+&?vt
zTA!A>v<wFji&DurgIa1A0&C?oCLkQxs1@OvHGhLt)MMCbKl|U|KAXEFEFyNAX*-y@
z!VL$*hdZU-*`pT2LcW;10I7c_(X^VSw}Zk-0!z@ZA|k|uao8?|L=8G4y&C7oq(tTi
zZuT*jfsG@L@gDiiD>G{woA#EL^Ok<m>43^<%SK@+_y@P5!7Yz*e=GDqAd(RUqDpVw
z$3I+SOzOBDNz6KF^_fjv6~V$kh+wfty*pPc<KZ(Zo5%WJJWk(VBp#IREd7h$Qm&qQ
zC>lvF0H@JRbsSH844zMAj^{A3a&~jp)mvL}E}f#|H`MNhNv4Ooq3Z`Wr|1#GyI!0<
zG@(@n=YNdJ82D^ewNDNJ;zBh?b+zf8DUU>CU8w~@giTCR(hVIMS(eo&xEv;z&^&I^
zh<U?pIEMK7BYaku8v3K3H{!U0#9$<8D%!(bH#f!{I>W4b3EB5%h^@ko#ra*nr|mC8
ztDSVxC3SY#h&=X*{6<-?q{QkVg^_GhZCC6zw%xD92y3b(3idz+((B0P@3%#P+(1os
zMpLJEE}1!@J>E5!^=-!5;gA*?bdr$(r5}aSudJxQ3Q68Z&#t=^byjVTbgs745%K@x
zI5E}nH&=$}2p;Y6Fayw1sXXM`np)7=Wq@QrU&)u?o)HbSp%32hT8F%srnq-SrZI@2
zUv0eHwelKSW(yEhxy2KykZk@u^VR*DosQNeiKoq*p(|kfRr(T)YtD+|)4ucv-pDRv
zyWD+4$0~mq=J|s^#g{&gk0b*R|7q1RIJBNX8ax}{Ke)Y}&;G0TnmUWShVI2%o&*+t
zlrnv|EwsE%Y%N4R@U82{pGfMTTzoIju91vHvyH}=4VO!R&i7miIJ?|TMU}g=e&KN4
zsKgJ}=(BX_#|KTtI*wadP;$Lg@HBNiQr*#f#q8&+^f-tCW#GzV*JGLAxS~x&GJgIz
z%;4E$`_c$a@2#E+;gSz)&6(!tT&CBr<6nvhvX$FcX!*ROrHOwJ##4*M4h32zkrnwr
zc@X`im^GIi_*?3n6a4WZ2T1Bl=`+TUIv<mP8G(*P#n1FftYdgzzMI;*qJoPPt9Gng
zuXcM#H53t-GF@?-&u9KtLR<&+2rER)3;svsP`T_~x{jF-89cBWDxVd>9_PKLO_EYv
z3>m}58%b*g3**3rDB$|RQ^4ng%pT>-a}2x#G!K|CAP^{&epVkpoW3(BlJR>BVbOKe
zsVZw&a1cFO`6!;7Ky&YS!zMUK^nUZg6~dr_UTuE4EBRmX$N4^&)9HjOtHD4Mns138
z>XRv^ZHTeh$Dl9dYuq)x7VxBg%_4C*aM8}<s7ILp*Xve#`Zhf6{hw}vFhy<^FacGn
z4#3ifHj3i4i&>^+#-wOg&*FN>%oc_x{<f;LTuu&ODmITy+zYuFar6(t5cO?;m1J)C
zd~bg_Ov1>hx9?SHY55_(XCyWo<L#?lfV3Z{V+mPGjx4%nKr;^TbU0&y3*vR*#Ej*m
zwudT88R=HStF-fv9ZNMjVcppuc4j_j0)H~`<J2{>&5v%gfNw|gfK(|ms}3cJL@M`}
zYT_{&{oqfG>IAMbr#?bag0Pv0Z?o>OvL|AMzX)qwE*?XOm~nrSQ_*jh^$Lw<M>-&D
z5@8z72~pWi?lS7^*?BbPGnNAU{_1HI{MT2U%r`!H6(k^~?u)|2$qgO|RT_%`S<7Ap
zE1TbI{(s=DzuLUn$O7eiZc}>Ss7ePWI$-~y^zGD7k?TjdkkjnH=;mX9X~?grNOJ&@
z`3Ft=19Hhf_NRSkLGvQD`>9H@5@RAr+|3_moyb)oB$zm@?MQhS8VG(X$VOq{>mGCi
z<S$e#^Fr041I6KsiZ2Ir^5cG&z#W@41#G1%gSU*beq{8BJW9nKtbtY{+AO>1Kx)*m
z5->BnP%{$gw&_O`K2G6dB*)P_mO#&?2~S+%p(@oxN{{D%%{eUB5f6#oCrPFYQNhpL
z18Dx%k8woC=a3um+S{rgd0sQzlo&DmTX}44S|$%FlNn2EKL!$>8Z#<>A36%RB{%Ya
zL`G)huD56+QSkfAq$T0}c(c9jYJD6lyYpKN>fJYi3-OYY^{%MI2Ijsn_LgN<;vsQw
zW$yyF#7&`VXLXYvgm5l|-In|=mSm3E9?E>?*f85Vj-%q^mnB&zkCP}JxGd3jC0^y6
zr{!v0FQr|~fMo^gdnF>i)+;5>rbW5N!t(OoUbLW>1CbMBxBk0c{oUpsqiACPDTev;
zk8GK?Q;SsAwr0z#+0`@FQ_uHbE>(@mUN$E9AGtLT25g$$68V!U;_~8+$oI{P?L5=Q
zo!Oap>y(E+A@;^nzPG^Q<<2)^`njI__f8Vzy#{p3rTqaZQPPp#48Y#~GvW?LrdHAg
z<+UWzul!)aT%KO?yl2l=E0gDC(6^$khrO9Ku-G6DJ8hgQABN-m{N$gmG9!}Zt~rvc
zU&`32!gVja{6kYboC?^t1OM*t$`5`ms}pO?V`RE^1NNO7WZ-7zbLpJy@9S^;?A3BZ
z<Vy0C>kGgHbH!=)lFpM-mxcNEqy3~jg|$A`#1AXYKb&$!hb`R*dq0wmjcdV%q;RSG
zt&cpnUPd)62^KTN5U$(#IW~Hd3{71x&BjQr8xAYsWWTNnkJ0^X>{iM0`eE6B#+CJE
zie@7kl_plTBHzD88vEllw$l47ovJV_w@6G1^<+RdMxz3q&91;y^bqZ8aq->H={+AG
z8GmUc8fxfjJ&~4{b7__aQ`Zci@jzUR21wEq%QmJvA2qW+e81vUkm<zg|M>1*5Z|3Z
zbusA&C2fC^rf;xm5b9RDJl9{E6aM1=`kc4y;p0<JbSdwP+<=PVe}5M-6ubWHLH&R{
zs%;0rveTT}sXLdC5$n4;)pKAmRh_}$d)*^&^%@4V%w`eechAXcL`qh)W4vj_<9>?f
zx!clj*;oC2dI0Z|;n0H^yj`&R)fz9e88|;rkGCI!9yre?r8_^a1X$Cu4YT6BCuk{(
zuqM^+r^Mnn-VEY<UnZ>x?GgL9YL)ZTo^fCYn6a9R8Ie{Kw|!m0QPyNqDL6Rn*K8@W
z+OpKL6r7uk+Lss<mNRiPz6IxtW@+9*cuRWxo8CSoj-EAQrytn5vIf|^*>a~EMCwM`
z`{xYm4;B_O6r$GUW1@$y6|+MT3FxX)dpEo}W2LSAorHXQT9%LOqg1XXkyJlKiGaqw
zOy!X=B@ldM;?hQwd%P`l#elcJN{elSRO~v{dwPmPBK99b4K^7C?wVJV_E*9D1Z&UP
zw&sh4i#*dnQ+=E5z8(15XyuJa3ek5IB_cy2elGX?3rbxIN(liLI;`)i-lprL_HjWI
zB_lmUWZz{9U%k3KEdFKRrHgZZ0)R0SDYze?JInGz`D~2t_j<c$!R1c9X)uZCvHjf+
zNbCgT6LnSS%QG<$2x7M(YZ_QRcX3dm^{ck-oG;^BUG8q&i1!A}k4jzLHCM;VC$^-{
ziF-!F?|sELg$~#4AM<b!S;v^iJZoJvJ3CSruKu*M<ziqo@xlF7m)y)(z<l_4`KV=G
z2Isw(&joWbVOf*#9y>`|+DP&4404IFTms!Mjn6-q+G_O-wMo6VTkq`kA|aFlG8FVK
zjqN|7A#EfLHq3@ATh;Ihh0k4SKO$t_<tHk*ZRvBz#&di{vrLs31o1-`PhUajwn$G^
zX|Io;`Zu{GlEg0e>Qqh`2>xaGGhgo7oo--El_7F*!pdrgxZ@#{_YD2=S0u#XzM0DI
z;x7B(s`G)mYZ$@gTi1k0Vo3&=pHVvG9Xg(xT>Z{nq2Nc29`OV*1^YsFt6wxwiN_>O
zhtjwp*gXyjSuy8sL0i6AbZ#{E)6*hjEi313$isVlJvrX}h^D^)<n$AfIX_z*)u&!z
zFAhhclOjFP1T*R{mbDe$bN)5_H@XzIWX1zcQH$3`Gkpg(O&KWREBSKXw(4I3G8;3q
z1k}TF1g388e!|n@bxYpF<k4$y$J9*kR997<R;ih?p@-I3zo|oSfmM;?fq^?DyVo9-
z%kf5)%aHJ^JBN!v;=!+$AH>osW7mPcF!@5g9^6S#{^08HLQq&lTmGQni*iFZMn*}e
zQghaFHNd~M9n%&b9(Z(lh`+GfI*Su_r7ReI6-y>gT0GD(6Ukcb=}D32)wttTIa8-j
zFaG8Epa|Y-ek~5qcY_#y!$p!E1=Dd4qT%DCN|_f0N*#dP0y5k2CU)nZ+=j=0THwdm
zZ4P%V*Tn1m8nA7;<cv{i<aOpF#<@K!D*j8Qij7u&joiTH?gazyY)e}^WT-jujL@)Y
z*!wm2gQ`e0+lx4M>R-rl$I4v)@|_$K;$WzSE2XDet(^{1;&FZL*vH=Wl&|V7|DWtQ
z;ht3X-lqVeo|n|xGS8gkPMaX}tt!p*Rpv=nD=*sAHl`p?fY(qYOoVv0bvlCL)ntht
zM4Q&aPeMqrgCcK139GTB>=%$Z_pkhpI@lpTg8skUW^3KaR156qYE!|W$<Og%(`(b?
zj{{y(ONuxb^HD8<b1=%U0&w`vJXw{t;0t2J1dv4RU?Hb`R_+VyqF#d&zQj+9b=C9-
z=X_lur{*pJGl?@{aszO1-7FVz7;>VdO0B{0S>+NGO%RjggEFJ`_x*AxBQ+8fB#sS@
z-*dg}J>1>>8zyH*tyB4xn(uaM)%eJ}$t)eRqduv|Zl)E`imSPHTvplN3ICczduc&V
zATFB;e796wY-HyA^_5T>52Hu*Q8n>@?*OA+vo3npSB@t+MZaftl|S;I!;X%NgS|Kz
z|EKyA3eL^;r<{?Xe6MYm$bhZU`k1dsVU7OhKPA2%MV)mySBZm<gL8Gv_7oX^J4<#|
z*|;9MBwH0D`-^mvG_zck3-nfp<5NMv%Um`}S_mEP<$N4cTr4z$zCyFyJcE`3)nz4I
z)tPS_Sbwsf#o#awOog4E?=CDX_$iARC`F=@dug~VSxgz&rOcIDX}j_yU);)Or2W-j
zfqv=xhfK0315=e4MMI`b)BlU&qg*6EiMWVqo3oy4Ia?nORJmF>L=-lfHbQ{IO&6xN
z3$3^+XT!{surOur!V;mh58x)3%6bk3>Zl(^0_sT+Ypc1Lc9QOD`>f(@n&jt3JiY&^
zip)ay^Sk;-6z)v?kvU`LdaEQGt4%4Tr7)orL+|g3Blk;K8b`HvoDK7uLE!Qo+$jF=
z@FGuai>E-3LBFSO<W3_}4+JE2fM!ArEj&XWK5Ex|w<4<7BOO2fdzX4HwCqanO#yjb
z3i;&t1p{L<-6d&&H9AFYZCMtB)RjuUxVYF(QE9o_#pP8Hf|L~*VB3LMyz|?vW}yub
z+9W|(a;4<$4Bd2_f1<cAbyPO_^_tSP{L+)5L;hz=;3@r#2A4k}3czDONBb@NB<`|h
zr(BUIqy3cOywd_kqh!u*WCTMsYI-r@<}#<6;t2oBJnnz$baE}$WuU00+kt0T+0xPH
zppH(|KI5LfEBY3A=rgCLw86GLSDbQ|f@pJsRMMlDbGOqPMa(=sPDd?j>6^L9EUm1#
zLA^BZ`?cth&2yiydBFDO(NVQZOUV8L{*Yk2RZN=ws3E~tx~vePA6yO_MzURxzK0JW
zn`${`43Y3u-(-roRKx3<F&?nE8v+@X+SWe8qo6FyNY(s4+#<|WE1c;&n(B@bYY&Je
zd3AZ=gwzmUi+cn@W2@M%6?-;@9T~d4^<*6LG;@<Sk|y&7C~5o8wwnkkpoCWt`|&h0
zWEyy3>#d1t)=tQqtIO+*veVvJj<dE0``_r3Cd8wR(>~u}x#yRL&mEo{zcripI{c!L
z=;8i39?8OPo}o4U=~{29kFmLr)0bRfFgZyXlS)T!S9Sl7p}&aqCzi8_a%gG~xQMek
zC$v1cmsCdqCZj(dp))M8BS#nDK&*4btNrS1CU#9}Dc-XvlZcy;XIsBL+PUomTL~U#
zZJQyhgR6Ud_8=}86CXN<hQA6nfB5F&cCZ_swsHUcM~tzrKc6ae15-z!4bK{~@}U<&
zMv~|pRhnSctvEX%-1q1o?yT@~dGxF>K=#Oc!WKY^7QH{b@v5|bi6r~#BHjDX_Wzjc
z^a`$;$Hd!bb-cdMtV&(9gP6veR?oJC9uhD9{R^4c`8^>yto`++{LC#yO?RoghFQyD
ziV;fh+eA**Oh9!-4TP42&vTN+V}`XWB?;MRtb?vy<82$l+^cHRuiz)|QiBzvJ)VAW
z@&c?GyS87Yj2V8fu(!YbeX#5E+w$Gs&PTqman+Q^G*ih>|Ngd^h^y)>@BMofkB)0f
zuvCY1Uw9l{60juGs@C9eS%FL^lE4)^xpg$YuUPSX?+&p!p`P1kt0%i@{ajd);>~h2
zJzu@$$_b0=+`^CkF2~0eH?bO{bQa_(><+-3ueSShu>-=;U6A3v;de4O=|3iM%%H!1
zipqth2ol`hd^R4606Rnu#nFLmiBMM-%=W0#+Y#t}9c#|vUrLFr#)9-ZX@*bfPBX$@
zqa&L8OP))c+Vq}AnV`NkD%%D%7qpc5&{KK*f#X5&5A@lme&K-^!=gJ8Cpd)c4q&sr
z^KbN8)9zZD_&*xs#Q03XBP?AHQrs}%O}DI{5@W*87TshDac;f4*Jg-ccFx%Oqn?yb
z_FKd=BhptFm;Z(-KKu>6+})?}Y1w=%ZtnW-l@Vj|5AlN45bK}M(Y7Dh>0oSRl=gb@
zQD5r+WabluN@75|EDlod0{BJ(n6-iRf3Ev0cq^4I`9iu3CauaSv(bISr+q4;Pk0?D
zyUaJ2{cfz##Lt|y%@G?o%JN@4r&wWE#Z^CyMvs}NmD!C|F1HQ6)cEIn&Dcni0_04-
z)cpKeJf&4kG8Dn`TQjQYJuHfvTg}jsjHSaoHYz%rH40jszOQICkk!|GNM#9#FgLQv
zcH;vaK$ZDgofJ=O>BOP2B8PwYh&)Z3%-8HcebH#eOnG*DV?U1Y4)poK)<zJ?t<mLq
zb@@8X3xwV82?%^YZKOK9tJu0;{zXl2Y-8TSscRv*8{R0#TNfr4{jL^7-ouWm!?v{n
zN0%3(SEnmIDqA~qhMbYbSAhFFPl@`xkzj?w0^YWW%j1)yxjJ2uQFW5qZdqc~{IqQo
z%T5tL!XCVKhIbA;+Q47r9g*rv|7rI%OhZNvx<0?Y5hhegK_y$>C&jIY63y!S_=AgI
z{0?K<ePeZFtQ06UTgWJ9AUIdM-0(X$M4Z8D#uH&$U1j7d=ocUd6s&602j2uo4ZWHA
z9Y+_i_ovoHYpb2(!E{N4)pEN&?nEzCP!Q#bZ`d5zp4fx7{Jah1U6*ARX5N)cYLso{
zoo#Jxo;rdRKY(PDT<S>IJX4<bP3BxOC?YkkYV~jB)7k#YF*I6z^t~?Zg~dvUe|Y=k
zD+k(`CC!aEQME;2<;+iDLxC8&k={h{YY4?2jl}Cvby%v?=~ldS)gE3}RN@isDR<n3
zMm!XD@I9xe05#<xk=YLs5slc;#ta4FW;7ZG>d#6X`9oz)m+GKdp28LVT7UoZyt459
z(iQtL6lKgEQu`Yr3Vk0+;q$qaF3<=557G@*9Y-*b29uY=q)lFx%Ol{a(yODqi|tWC
zQDNm_kzO)|=x4fJ?_O?i_ttlb5a#CS!hvSXf<n~rddB8A-C?QabW|{=wbe~I;BmId
zs#EUI+Ac8N$%)@HKKkC&qw#T_QPko|ZHD}~Q116Ka*aam@{PDppxN$d^~f%|qo%f>
zv)bMopTsIr%}a)fWpd<Ch^lLdAbAxlsu_mBff$&d!yrU=xF1rs!DyS)chuSq94*r8
zBeLFxPznP#!`sULmDqaof2nbJoc!6gPZdsDua*2naGux+oQc)AI9S%wH5$JefLs$X
zWCK-7o0dm{9Y~SU02yWcbJ^Zkc-5(t9huspRk=8H7_a8jWgk$mit71VCT4Vli}dxe
zDhGLZXU_!SqK}m0#dbcUvmAdNU!`^b?w31K)c~ge*q3@s(@<k1RW3&CVBNK`88fxh
zc>A%<ig~6}8%unD(#lkdgUYL>+dX;Xrk)Wkp-1~Gr#y>dY#cw3pJD3w;$;ui@b1wZ
z$wgT7YIC=*=j6PF`**_*Li}l#d`ZkSeyGH53&S!KgUp`j6enr^E}U<lk!9n@oio^u
z0VuijDjRtajCAA&RuO#%mQD#|y1jf%KCv4hY60UF9c8Xk+`^2|ne#m8hhp+#z}LxS
zT)>t^JuMUx@%-H@^rky9FN&V~eHn)~nV<9=Sw{luF8Q9c(phEj+D>?zZ_z68`^=sp
zg!V5JkJ{arotJDA&K9fkadp$%L=O*u=;_(G$gVboo_|^fa~|ujcgMV>f|~eP#gaeq
zbqOk!De#pQcorL+zUac4eAm~Zl%+aX9&Cb9X-m!i@jlglt;geQ;>Cwa#-dGoYAV|p
zcBD^Q+<mA{%qR4Ieuc`B4ez~AF4h5ggoyU`)>b+QsFF9D>~^d-5Lj`?*cAp$1%UD^
z&?aP&9pE1{QeywCTH2=KU2pk|in4N3_b2eBo0hD6e^j6T*hB!q&z@V#&VTJi#Qw%V
z1n=C{`DV&$AAkBgiE+#R7O!=q1o6k}RHkV{PK;@WWZpTxl$;K1L8Ix6MVYlt1AVk0
z;}|v2-D0r=T*jfIaG&3#N5Cz>AWA${?wt;my})M;?*E?pq%}*$Tl?I4B0xXe|8cGI
zg=@WD+r5_UscQ7AI3)1~HOmB9wpi)J;V-Mt&Qe-Q>MY<MJ8B5>N3BqOLfOUt8kn2A
z!BxIlPxO@1D%jY1=DTWmA2S?yNg0#RB|B)ZhczS1q8fvZu}O_LpTK1RQ(rYi29&W{
zC?|jiqUuFnO<r=GWcWcsAvz~BKQXt*&fxj45Cv|Opf3`k70>nP*Ji+QAnaAcZOSc-
z$IL4x1nVCogJ<4ifx~EYE+;+_{0rvg*EOj#pi|)BcoX82G$ZLdS5mR<#i@|z{J{Rf
z3Dqm56*9)$Mv?SecZX&GjJS_WDvy^l@vM%JEIW?H4~Z*8iL-w!N1lber!U_LG`(0^
zArexKb|NO{T5l4@_{xmk7=`v_<_bz-tINB)n>JUcBo+P6Re-W$Z6COr%w$_Rc1U9b
zE0tWn!_16!t(0);1Y@>k4k*6UgDfkm>CZ=1kTU^l)%BJG2%v9Se)a3mEt9D`PwFnI
zp#SFU*BqOL1Xu4DuMBZA8R<@M2S_G6%y}Xp|8UM&H$<OHfd<+FH$MdInJ5@YW})AH
zF8Y`i%QJSdns<HmVZqKulE#hlSX0~O^~>$7u`hRmDdg=o2l*06pZ5w34-NDUZJj=Z
ziAR5K25z}n01=62GCpsX@1yohfIsYCHaUK{2^XU)3!}hVZfS-o8#eLRSLA5(2m$(<
zaY-Y&LsLqqWu;z7m&eTPjQ;B4$+JBl4uptUK)i)ya-K+uP-5}(gm792rCLXDLA$ad
zPjXh#PIWbj8!}A@VDBC|a`i~o2xqVIkc$Jm3YTO6`zJ|U+^o`XATe-Skzk2=c$jZ}
z1$0<34_OiU;G@5I@Vrwm+D$RT<=w{ZbUKcrXZ4m)_Si_+whS&nl9T``Ot1Uh8?<pd
zG#}-@JY;C6$H9qw32<jWIjawjKM5yc&K)y{v?BgAl-P>^>lKNt`?na2mR46^NN%0F
zw>JX=H&#mqjAW8xW;_FoUnvjjBbX8N%{Pk}x}WPRiE1R6?7xZ>7OvPXh|b0gWmsDE
z_9ct*D@rmmdc7#kHne3nfhNJaL`st7Ys<TTc$lCj5#q652p$bt-qt>qrwp;0m&rYk
z9yEj5W5Iih<M-{vR<lp>{)2>`YD#u?&nh4wT&0d*G1DID>PFS!E^O@kF79H(u<xVm
zt=%kv>xd2{QsB?mTNQ#-_T=N~zOZ7j{rHp7(=bJ4MefA4fc<Q8zqGgl(k5?6V>{`B
zd5;q8Zkl|B1;gt{ZfFvQ4za)bGy4PwzcRw?gf$_Q8}Sx5WuveopX)`eJNEoUQUDuz
z(zB?RkEE%yO_lcVwBt4>ZwQi<;?lO&O*2}$&+#Zek|ISduzq@GD=@xBh>cuhh`s4^
zyxk4)*a%5dbw79ccVShf?P7P_E-<gcRJs&IybpD)A6wi#IO$tOt&-|})G7)wSlIqD
z=g6UltN+z~vrk=-YASwKTbQ9Ez=exsXjZW4Qhv>dGayt?iD_ElGDf<F3@-Zc1A1Wp
z;QZXbpkU^O*EPxVL(c~mN!&~jojpttHS_V3XpuiXsi4?)R-w(S+UzcxBOX0Op)bQo
z?7HOrmTcG8`XG?h%4Eo0AGq($Wb!+koeaI4e7CKs_~-TqfET;ej0({}Wqx_-h!kVa
zklF_p7!_@lQBwIPIiiTSI^ZnO)4IL6eJ*LbBi<lBmygFHO2FPo^n3mk?1kCJx|F+2
zxQeB})ytA;dsfvghaO0_**isi$;j&VKUXf<MX`Su6A64EFfxiWwG2;p&mVVr8h3!G
zwzUCl^j2LY{T7F`ZczcYF<`MoW9D~U<l8cNa>O`-^5N@;JcC)oCws>SQnVKpvej3|
zD3$o~vK$Qf(*$9*{Ta+zLDW~R?Z4Y_apt0>&&eGwYdsZ&(u`=36DCy$o=8id4SnGm
zl_t~84;LFpZ@3C0LC*84dLRCsAMOrbEdtB(f&AUG^2HotD1hZGgNd;TsV#ns`H>MN
zksLMc=4absC<#qk)Qmb|iWCdz?e~G%@bV63y&AaM-Q5jR92nMcZJQnUX3tqXky249
z<l$wo2eIE15D-C^1eKd~jXdplt%P#Qed<Ck5%2%L@io_hbNUv?7f6+B<^7@9u6|J2
zoj=Z?qD{lO9)Jar(eGb%;-g5aY%7ou!=&AZ6bo~^Kl5rV3_5WV?+1D;-AqSd$VXKh
zrKIcGj$LND#@}Xcc7|nCel^%k-JiPwTmHOULxYTOJUa3|IX;%j!PXfnRox7~C3q(W
z%#riFegth)iVNPcc+xHukiXF)w5mg7w%$SDSMGEH>MX9K!zWB?TZTPD{MjF4*%yam
zMV4KZ2Zo2ot4*if`#wEO!rE;i>|no-dRvU%Ii(LT?ENJzI4BkrBxm7j2EH3l#hP>X
z(~d2;Qm|B+4n9`cG}>|?dDS}{C=Y$%uBvVrKYN;hw0?SD8tO<*sn=C=yZZuvc!8DW
zjX2o_##Lrpg9}e9?|t@BaVRITwNbcTiKqJdgEnmuWR#;nxn%Mx1+~fWjs>+#v7H!}
zZjl4EMbl#Js`;(8ZA9Te*&}v^>(PfD+VpiM%}}U>5&Tm+>9gk~i?Pd9zhK9z;pOI)
zNj#>_GtBCaws<U<Tv9FrYW%#b>lqz;jvRnQqmM%vItqq2r?G>Lfi_h<BnM#V)HlB-
zD4%J4bFXdOMim^hp@jt5KEO^aoH^IbHItHiE(1EZ*Bz}w2bHOw*!95ah(?Yh?6zDo
ztfZqx|H663a_6W~c3f>v(r8gt>^gmvyn&=e*F<jK42Y&;LMm%Nm?w_@%iF0`AmxK|
z@mORO{f4zKEJeRihLk9gqSQWyS4O>)hLM6v)&_7`h)wOri=^ok=no=#26;=g*&O)(
zKuXwE`SK;7<Jcp`gEVnpIAmQWI6v^D2q#BT_Vm%@{^_sTq1t@Nj=_t~%(M%W_=~UT
zD}B1O#bu3uu~Lxnt@S8d?=%AUCEiAjWRPZJG1)rqbUYhw4E!MS)aC>ikjJocpg%JU
z={6bJIyu=RX}s_i_|d1l<Onc!#C+|;nEBo-sJT-?sVyhAj$qB07QaykR()A=137e~
zlu)rzn}J$gTu&cu*BgeS)1D;-M#D}Q<hb>QokUjFJ|su8j1+y|1+Jtw9~48ifAnjr
z8>>%?#8nF8o^j`Z(^C8C$k@wwFznIj0#sri|1;Snf*5FD=f5=9cN&N3eJw1{;OEyh
zd4Tc7`M&2VirBKu6A2sieKxkk`}d2ZT-xV|0T$|I^&nMUn&(-D+0%kt*;?P9YiNMP
z#PVdvYRKWZ!>SG*Akc`J!q&0a(CVq%GihOGuiOfLFc->@1r3KKa@cIknwrWkHu=6r
zp-4lPxHdLM8YXtNrHC!8_dHmMjjdAt%=x*Iu$fQk1mua4hAODVa&N&Um}IzqA!;*-
zFv_`8%|S9dcAjGo!w(be{}SEMIrpy3(pqW%BE#q<HQfbfi4CUH-osBGOO{t@a4S*r
zm1e7Q#t}yGn~l}V=jZ==Y;1tzX5OPn$f`n?p>c<QPU+(QK5o34^!E4mr6X!y(8Egb
z&UIU4l)5vq23TcUD^A47Aq~w)n02ju^7O?tFEBMX^Gp*zKW0K@gQ^xDQ=K>13zJif
zI6cm42QL4uI>~#;KC5rKj}hIEMn=w}lDU4}6ZKKB8bIlkKftzMOhkYDp$*ocR80`y
z3Y?zK%gfu$ODSwiKw>NY^oh^@GR;lo2%=D~uZm@RdowzGIHzrs=}hT;4f2dY1=0wI
zZD-J|fZ5Mj0;d-)|7{$89}?aCe~mx|Fx}`$oNEmat)G*WbWmdf+abk2vDCPqBfFyH
z<G>UM_ToF0C6OOFz-eWxLa2SFZ{-<(UaBWWX>6oRYH3i~Yg<v5aHTL5Dj$uT23pB>
zOT~<kL74a@&fn}!q|@P2oBH<`af`s!&Xq&%ibx4z;!U^;eUiFsvH;G*cR&WX+07!m
zC`YUkyCWd+6}KNIZ^gqva&HDU&UQ~XR!INC*|y2QhcQ-j?!>vUmk4`CvY7qhk3w&?
z-MdW4y?WypJsxJe@S3@;8<Mt=!0!x8G%JgLGu1q$)DV!UXD#D{8>@X*_mj67@I5*A
z!U*Z4oq=(qhpC3TFmO@MN<OgaB5|wIN0~=;-688Y!F@}iQsN!?3e}zZ70W%tdth;X
zsz3jsm6Zy43I>OvT`kkE<b-cPQhAa|%qcXx(Q0zNxv()U%T!SPWT&|jNx*bh-PbMB
zpPrG-lzG&(3)jJdJk9urb!YpC9!Vd~xOaeK#};AKZc*Lk^BFNj?E^l|8G;o*0Yk<s
zAbe)Q53>{;e1p1*3Jp?t7d%j}?HICu>VYLBEqcl2qvR{dUJ4s)_kspuTnBVL?u~-Q
zyGIUN>mxXDSsm-<R=-&b8bT5<|A`x=C9OeHpPr<DjbdX2-4K5#gp{Jq8?Pp1D0#J(
z%HiiM4qMnNGBKnYFo)U>(7XU8$I5}P)58P>ei|pQYQ0Hl!qoFtrL?Okhvo1p+bCZh
zPL`z7<OTOtH`j0Qh%=l^c)K*0$hbncm_2(SrO)w{VsZw^X}q(fA1pO<QZO$;KI-9P
zZJgXzbro^8<zyj?fi$VuW@z84B6fM1ZJbDOlP0--1zEtlkAjklijuM-llNT8G{FPQ
zP1Dj(?QdXMt6P@FS6a7@X-kB~fnS%280!tscvgALwS=DiA5G`|mgM`laU8jFYek6z
zwL!x~O~sx2<EmUG?nREkm68)p6Erg|3tU;2JG4w(xXM}1a;D<UiDE@g9MApz;n6<;
zj^n<$uJ?7G=k+>gJFDtsQyH~;4r#)tZ?a|mA=+*}h&uecP}6+43A{ZA?OL6IgILMV
zU#KN>dW`RPTxq+MbZVZiu9mcw$J*F(!;nRE@5@i;9^=?Jf=vG`uI@1w!>B(?Ps@yC
zWq&HLyfh)tIHu?JM0(_0>m?4he`kF+CHt|L|HbVd9X22KuK;tw5;I)mRZ+M1E8<y2
z;0RA=(MPvyDi{F_OvOe(Say(1OF-^r(A5?NPS9_=B@cN&d&~(VS+6jEiky8L(P)TJ
z8*&Pzg8NWsg;WBG4j%eRc+^_hzy=2gi8xmGkp=Bs=zOSd`Q)*gBg1ZPZgCSOYp=JY
zW2v1*-mm~h1+Z&}MuupoWP6kkzDH6?dqHd1(FuB)3S+1QQ*B)axIR)(@vVc;lee6Y
zShR!n-t}QjjU*m`UIaBu1ON|(HGAq_A5N=tr7P>a-q156zQwQbEYIYW`3uhmPNWfh
zmg^^9SSPEjoE+~zEM;YCf@R)>$Megn*T#uQ$7BgeDKPgB@$}mj6#<x?C=GKmbK!Eq
zJ3VN@dRQks3~D?FJ!=Yk4J?x8-322^dhl2Hxw*iQknN~%LioCGB1!4_?0Khd;j($c
zHTuJ0*4NI69m9%JN-v`LX2|8)B!r4oc*zO7NoT_>|5v0Wu<(R%>Iae9dYulix{;Ny
z&1&W4$JxE+2qhD)43D$aZ-<L7VWI`reMIvYdpADa2G-JN679z_WIf$XtmDByVw5pz
z)Rv&Em-D+(nJ2OLWw~Xz^!%zcaUN#1l~SH5P)<8Xpx^qW4M&1%#X|}@_C2M)fv}S=
z%5KDYijvRh{qz*gIk4%u%5_U9rx@Rh@-O$R|K;EX(>DTYG_1{>*ErM|+i81Jna4_Z
ze^iBnBBP=JDUbzpE*&c6!Ec;HC0)3w7xbAD@4K$a5qqt1WriGNnOTB+3Pc28#xlVn
zR?#wx!aUArZ!2FpUJDiJ<b)>1Ou9_F&>F)xw_5Rl)xj91f?1vTqHtK};4a`10aY?j
zJ3ST5=MCY1QPjrzrn<U1$@*W&m!Qum0e!AD(E-u<80vp;(9SbDb0xU`vhnQR`tSi&
zD4TYo{qt*wzwX-IIF9GX&VfR1m3RX{tNEfEM}L0r+`7lhzsPrq?9*$ka{gU%{KV{o
z%*D0eYHvvD@9}pqf}?GaBzCCg0OnIPYjyL6^s79<iKTgF>)`wVmJy>NDH4CC=ds?~
z@v3{yk1c+TyMNyyPeUYdOH9e*+Hl}9^g%aZ$wx%!<o3HZ<8-vOcAlg3>86^AALpU)
zy84r&ptj|fvNB!}Xkh$5Ep<Ft|5upB%TVJ@_8<eLPsmNBE8W(;fuDZ<i}nNViB#I2
z@Yja-z|N)vbM@2aBi6w>utDBtl{w~LtKcTT>u0n7pj6#;sYuF0qi^%Up$2iS3;Y8`
zJ>5aJ5w`gS%>V8o6bY=OnP>SpmWBe`R(EUl5AJ3F;WNa~*<h`h-Ggm~0`b<_-~SF<
zS#<irho8Wul%@5s$EHGt#Jk14kyrI8>ZKgY#um^7H~*M#`ZGH1KiY@u>-^NnEdrwt
zfV~rJ&3sbr^G)maAKyL)a;Y!JGHI+*yavU`e}Gd|X}lan=;@8W<?r&>>08#ViUOt1
zp#sbrKs{1~Amqiky+Z3P!W|Ms%+rzYPA+66y@1_O&QK9ktZM#Q+CtBIQB=JZPc8L0
z{vB>(punzL5&`CZzUl`5)4O>XHQVhS6Q$u)h&U#HJ8jNRCPnxe&q90fh$`;zyfz(T
zJdA6(lj&(w?*mSxQdt?#2us%H!;7bU5;{Uj-i&3aq9?xv{M=$JfifaoYk3ax`UzUX
zKl@`e#?Wxig-k)mOT@m%FNL**)P!B`t<%xciZgSo@L0dx{Ohz+CVauCu(K7`kx^9n
z%K~~vR9#hPA?K8CY$h=I{GJGSdgE!`Wt`VvQHat_=Mxg2Hk&`Z_^9_5fM9UuE407a
z=xD-%uBp8yc}Qs2qE3Y<l^jbV$+Zi5js^gO*}to571PeNCMm3%2T<P(ESbz^U1=cV
zw5!yQp9ij4ac{6<KG87c9qXu16PTLp^#QC;ft+V9-|dRtEgpm9R>k6#U8QJTw7M1~
zHRR<1O&jZ}#<+0*gS9{t`B_4jp<qhBx5LX;Ch<mI4D0XawyzFnV)=1FovR_tr<oTK
zZf|dxmkjqUon0!dF4xz^3w!XvDG(04DP$Zj&DQB2#aw+sa4KBU%FTPqEof^V9u<RJ
z3~uV_@AUnO93~sN)7S(=5N@~`F0)yjY#~_0mF99bQBO7Oz4xpB`(Hd{mAa$yqlTAY
z4>MxR&lY7^uX?V@u)nue2E>iG@K9J=Y(oA;bmp%8#e|tF>0X;rugc5AuvoRxQAZEG
zFY_xZy*75Y;Gi`sz`Cl_(vC(zq!fq-FvKmH2nXosbMErScs+=<8tz`=`qVOgcDFVo
zKQ(#3`>N+^qUEX|x$BKY2Mv?0DzRimewVgae!HX_^~r(=%k6%)q@~->9JMyQNXkv^
z&(l5HKWxR1R%}#YT$>{{HHfA8lu+S}xtrr9qu<lgV4~ax_tMJ<0K_da{bF5mG4-dK
z3oYUuyCOSB#Py$095ku^awWm`(&8DpV<_X&e9p1JG3ePS_b$IHcJzg)dj+@r?oi4t
zIZO?qfn{vc)~oHMh^)^lcd>W9u%aKG+AK)={4m7|m;VA9-cbB15NJnMNQ%g$k^E9Y
zoP}e%ameMzBNy3%-dh`8J2~!(Hh4Y%XJ?bnAycRzn3Vmr3%3KFyD4sF54+0h*4^+8
zzlJ>(8rua3Rje3X=nG=QulI{rq!|JnBXct7cd6p~1c=9sFu<kDZY9l5uRiJFcM`Yd
zRZIYT@b^rO^EAGZGfm+cV>2{+Jg($k60Dwkv?n4Z-JBW?#w!^A;;Hj`k-`JTx6b;k
zVg?GJfiuFm<YJ(CSw5$*`_KKG6yU17bh8yD0Xb|77U{{p1n|wAkh)?r*X|(Yq)Bl1
z<e)Wb?-(}vo82+U_!$#WdFhF+h67E<>T?3_V`~AfM}Iqy)>j$V0JY$6)c#*q0PX;H
zIDupRU4RnDpT-+J|6{`kee^-HCOc@&=2RxD{5YV$xXCY_ewlo1hz0I^vjhfyJw;dh
zj}Av~^cOu28EcBbzfn#ZPIKMAcs0#bKk!{_U;E4Mg?}r90eLz{0QkvBek06t^ek4_
zwU3glc^`gFA<drj7r4^WV0%l^B!1)sp|@S!AmwVecJ->LiQZa0gRuo<FV`>1BbU5&
zoN)^H`Puo-8`8Qs?--q&Lf>#ii@J=PmsvWst@MQ370NElb#^1uqt4X|KL}(X{;31+
z-c#Jcjx-J-^{O`|Q=y2P9P&zGyJovvtA=eptVW^}J^d6O1}`MZa>--1BD~I_QemQa
z)uzCJ#(R7-rapq5gO6g4Z2j#cBcmH}cLk%TIo!#^y#;17wWlux^4azQa$!aD^kmaG
zT*T(nAd64j2P?_hDNLmeZ&HPHGpqq-p!)v8V5{%Nkj>Emm+2`d^#8&DM>yZeigV`A
z?r8=bKP3_bBtaRkG(*^g4gNmrI66|>(HlGVB#ayUeb~9=DbFm!@%QH&s=smS-W`Hv
z@9~wQ4w)iS4F8HvvdY?gRgu;ifm^W0r79;*dwhcnz(zB!TaqMXAC6WY?sbpSqjq<f
znvee6!0in-AFjB@?9C1m^E58)k2V{W3HT}B1m-N?>_N*8aXaC4`HrA2R3ft)c>=Ex
zt<fqljCSKP<E%jlvOUwRlf{9|XS?g7S(|5-o*1xqLZ5Pt4GlL?A~KhVHMcpg-PRAV
z%pxF)-F@rJ9ZMDgqE%P7Z%m|SMZwEbhL05^m5}Jhu#6JJGP^n^|7$1@llvW7nVk8B
zTIL182SM2Mke9CcZ%onkzQtO+&mUd*v5NY^gLYmz`dx9PVM+BJ$B`#WGLY@IvUUFG
zfTaM!T9@5B4)3OU)x+}R(2nASUxb9kL_?Sfw%MuTXuSew(B=--iP^7B((|8Lo-+m8
zjBh+mur&)BjIlSE8$pu2U0x~-Z2FuK;fRfQl>RUK&FLmb$69(CVarMrPA}t4j1??q
zB&tXXdJ2L7yk!+mq65}Dx>#;XvVK+yTwSV?`UU4kStWHq%SO;I(4k~D_rMd+Q9szJ
z#kHoLn9Uwp_JqFmvwR7XZK8yZ96=Ur=^d)qX%do9<ttVu@f`7*T^$~ruOi5z>h9x@
z{&gPhED7Zu4e&QbNA7n6Kp1-W;9zU(^wE}@G6B9FfqT-TRj4I6kI50F(l9xRT8ix%
zo#HKb-Y{fep1f{{&#*ST@$@O~sI#W|V4glt&jv#_bW>RKA2bs(SPzfuHP1?D7<Ndu
z7$I|)m^JvkX{#Fr3URP@u)*3dkHs|4@r`YSU`m66_a$>!{R#A3%-|l2CB-^j!_67t
zX*F7hi;%i>R#1x_99xP%5Eo+29;z<_MKEE5JHL1K{)W}{Nr-j`cR*7O&Ln17uSF{y
zOtY2(er?=YrX8%z&PFu8nx1-<mnYR6t=+6sjcjoinNWS}UuNV#FB~zu@#n#$bSmq>
z9fF-#u^e1hH{_9wFlK9|ZYC0P<49k}Xu=>|k`uW)Eb|m9CyXOsvzV%!&VBM0)(vHq
zqW5EMeI2pTJRQB$6U`DA(U#f2AtSE`l3&!~ca7P<Zl}-ZiaXrf+S~gb@fY~~D+_q~
z-%j2Ae+%7;kH-GI?8*oUKA{#g7@AWUl-2Q{(dmy#RV|Gk_V1b+bzBvSyd8mj!^$RZ
zli5jCS<{I*T=}<G?^JR29{F<l94hv`C7@5zazL5-GJJ50`GJXB4JDyv2(h^1wg0Ma
z-41FI7H7ft69a{iCFL&myoT_=lyr6QYj#`TiTcvAOySuoiP9mYgf$PGyGT~wc=3UA
zx3|k;C|&=<8B^aL3&Xa{TzmvJGA~ZphG#_i*TrZux4w*0jbq&EdOuvb9jtzn9>%F0
zBIXOGIa!kX?Kq&S&Lc{Tv#ke3p=itU{Ax_E(YU>ZWO^otqC3xYXeVyz5!}dDWnS39
zR<co(fFxTcAV43XTP|eqTk0dQQ0KhLx~(1Z8XoKesU%SrOKY<DcCPH=#xkQ0Y9E_I
zV((~1wdar`U;Njc&WlJe<ZNq6NgWs(K@NjRnQ8i*a}2C4e49@&x;mh6^&ZHwS1-av
zEkV2sS%T((6_T@TM;Ac~ernq!$Bh^xul%kd1dg)>%8#%z=z~FT+1PAflr13HbDcfK
zXUk$c{|g*)X1h};51<+B6wTvrOiSgtO!YPI|4$Oz_pSqAT0u1#e--&#6=ggKL6A;3
z06Q#9O_K(CC}Bdd+Ce?4vmq=_VKiJ}FBoyqd~{fohdW$YSnq*b;%WGPSDlz$cdU9M
z36tM@)WSDuUyV73eA&ISzVdGiV3VZ41T7{h&&9={PjG-K*>GIkU*vlwW?iGgu_T`X
zUOV^czS5`7ng_bwiD32nrbyzc*e~Qz_MY6!&x{`W60of6^r7nEcU)EA`6`L0_=6Z-
zMxk$3hp;Rt69GSBZvO$40@T(fy`M~BuKWWuiS44!0Pn`XS1zvoSlcWZsdN~`5L4jm
zThvzE>;vqPd$TTZR*%qH^HNJe7%GVw@}%nn7N#H@C|8VBH|kjZvTM@uxi#Q4LLa*G
zfNIhV^44W#LSkckHt!=e4}!P<Fd4P|^QMzF#wi<}DpB|=kH9)n`=$x4>VZ65zkjdK
zN&{ACV{Wr{Gls%;%XB^#)|n1Gm)NByoE<HUilz}mL)Ff`r&L>g<+ZJ*428ZLC65eF
z&}&?AyG-2S4c(|c5)(Yw=G*q6tKQ5;F-gR(O8jS*XnRhjy9&s-2l2Hx(k2VJUHLE4
za8DyZaGA@>XUI`VZ|JK$Rmqe~fK&2zd)H9*Zrgib4v_quDYS!}QTWXBTWc(CZ8aBh
z^1_`HS1cor%CaaUm{raM7pO>xiYr{Peavh{)af~&A%sJ;l$v==P+M~RT~HT{OZM*U
zxgW?p0dVQ$o|~k>`R0z<2nX8og9hE)tB&OQVSC3*1(-+a270ns<*=P6&$f<Ok3di^
z8O2Y+97>zQ8EnPbrN>$>qJLK7tjp@3!b^NDe8<-_O_A5cK;FslO7WoEh8xI}VBZaw
z>xK!!?1~__>79Q&Rax9Tjd)QK9K@XhpMI9anTj;3U`9BMkkqZdLX7Tl_|N^|EKTNt
z#x--ZM9O#wU(gqYIUe4lT(%}!c5@-W8y$-bjUHW?jpqN_7I%93za|4*D%BTU6h|6<
zQ<5D|cu$v=;7`IsCf_8L0E=)=hYF+UAo>i@=Ww)*+pPhPO#N3YfNp~2EDv?5r!WD8
zKsY!~F;U%e5NLT&x0+XZ0<eF4!8XZpJ(q8O^Lkt2G+I-%G+ML-oR2Dw4!6Scg%9VS
z;&%31iPE^e6{e6bn6bsZ$k`H&*;d#J-HWSx>=zy#9Tm0jO?>u>^A?UaEv*y1_W(O7
zy^)ITq_)nE0W~*R$K?ZhzpEn|DDRk^9Ytq;ev%oGlN-D0QoXsd;>c67VaZfn3oD`y
zhSy9>3*zT#e-2w|v$fJi*KM<G`BZ@KX-Qc|IiC8{Tib~wtDO^t!h`hIC_GcsL`(4`
zbxsu5JCb)WPtsLf_lic|b|*C_QDXX_aC~w(AOTEbrU`aRDkos&r$=3z(4%#wh%fdN
zyGj7=s<lY^-|zIf57MGy@`^vDx5$k(jx`s55I(ftW|N#MYA$$xwV#ac9zDRQ3DToP
zfLvMHc|#C7#m9@MkgONTbX)ATb(XXF<~w>iz|l9Ny8}mS($R}nNJw1-`HAxoc#9R7
zC!oqlbm``IX@vXaV_E8Yzb1Zo7y1ljH8fzzn<pEl(pF$XT_viT&WrR)yIcHFB{A&1
zTb$=@-rGT7A^IMn7s#8Mqo4r#NvKF^U44YIFL?wwtPiB=gGFDAPrGzR$rPd;YeLcE
z8rhi!I1R|gsCTrcbG0=Q9TJ=20bfsE4|qNmXzm%+DS6~f<R~twN7Rz`$UVP1`jtaW
z#{8mV!b>PJTRO13CGUwQsVz-*><e(0Z9fsfGH?l7S9zaZ)bc8w?Le%B-h>7cE{?0I
zIg6jWzlzshp3@0((=!P6FvKc5jsn1AuF~@}ov97he}J}jk!uBG$g%>fc`0RhkncFP
z44>gmVfsezZAFDgJ|}X&;Z(tG&d%;eonQtCJ!MwDu;(z~M-%{6fq#WPWfz5875P~Q
zhm8OIVa6?iQHPXBvi8El(AlHiyTLKQx87|&>~4;ZeDG>%b3Lk;N_z^xnevWmba99K
zyN9g@RoVI}YPW#1Ol<)>=-pS1{tDb~z)^qm(L!_S<<3?W!K70p$1>0>4-5F#=!8Z^
zx3!eCYaTG3SP{R!;gs~EBvsj)!@89KKG}#hl1!v;0yRy5oxHodpwUnM8iR8S*k)Pi
z7(Q%E<9IhV+^|vmyxF||BOZ^p?^bf9SI0TZc<o!ttiMD$k7m}1nl!afQ;J@?7S$Yd
zGY;1e8HXQq8AO+S&a=1x64c|hyY6*<(>FujmqN<29U`Z>BhMRwWMZN0B1PhCTqvHn
zm!aCOeK)QK5WA!5dsbb_dPaAHn?_K6F}0`O%`B!vm(<rgti$N__a^FJkh5JxVC|FF
z_>_ax6vsx4{OT05RaMwRQ+Yo-lP1TYM&@rGiQ^~BUOH#h@`~XeiHBG*5>K#@27(s+
zjdC?dZzU7G^mt5Xkl*I+{;eUhg6gLiYTplQvOTcqu^=$(+r&Ni3&ce@_$ei13NPqy
zHKn^Dw{P?!pDARFpG}dYB<xpCUEJHpBtj_R`m&f)8&;DZ5c+JxDu^9>?Y5pdyv|qX
zSsU2;8<caoSp+xG8%Izlwbo;Z;ZxsAOQRD{Mf#4Szj_O%lUAr5*3;oXBPA0uWuY2%
ze%mI>+$Q3~;SQjjOr%)NM52BH)SbtI(=zK`8QZ@b>V-Ug(1bDG(ZyN&-EW`)Lo+g$
zGi;^kNSKGW@nr!eD<g7~L%6m!>*3hS=6gz}pF>A02B*FzDhvIXy~*Zgg$Rz3=D#S*
z_CQQCw_Io%m{jwK;bCM>H`?rxBd+U;WkO0<O56n{lfnU`Bw@&P+f5o~4)cl~>6Ji|
z#Siqqv=ULwDBKL*^fri4Zb4tdoH^|Y_8>%)@IxF-uqUK52sb9ZAqdp5L70kd&>+H0
zWjT&mQ?uln2iz@+h}qlvw?B&82L@?-k<>r<yqJC0qix-z71yH&T2h_9W(XSSS|iV_
zz5MV;6%Tu$&-(Yj&Go+x!`F`SK-);^Hw`DnCyla_WS*Hg)u3HQ!hf2fQK>zk<wOe$
zNTw%NNGv1;$e)VI#n;vQTtPe6ka3rpxTr>#iltCzNB{Dzy{*50_kdq=Lyy{>DLa3@
z4|zXGU(!4TA(?@HGyY;}9n(eMo0}R&>M8YGcCUm<IT(=gdHGObG^zsydTmkQ=06AO
zfeY<!uCFr=0~>T1f;K#B40xy3qKZ5cV1qpK;QZ!gp@!e6ez9u?EMhih=@;~iOLocB
z+DsWdfap--7#?388AsVqiwH2fb{JEfoTy|tJ5RdE{K3JW`A%RgjdAX7xEHWga4)o}
zk6x+DJhxXsY%TzZ+a2zmyT8NB*k=%7`hi}2KhALL*Y`dKHP||LTW1kV1B&SMiOs0G
zWOg{m2dWqV#dI#qrfP~w#Hv0YaZJc%0>IQvZBpjvNP+u2=^SE8V%))sc5D_21DYQ`
zi?;xPxwFa!+#>67n<ylb+DDu?A8QIBq^Da@AL{q)$N1n_wDNN71(F0(=$yZQFU=qY
zG#~*>>SAk9H6%b&49>Bm_<Lk64Xdp?XgY121cU%B#fs3w&k5($S7J_yk0Bq!Lfkm{
zpo*g3$)`K2I?UkQYoZC7XKFt;kjb_s!KZ(8vb1BGG$GvQPWcz=tmA1_Hf*sou?ZO-
zUUR$kI33;Q{l9d;UI`dUfJEq=1VKu$RyXlip<ErNBvJ^!ffb@1{5BqY@WeB@2XS7_
z-6Co4!8w_a8c?NRagLwK57@j?;R#?4;}pGwG1rN7MG`gZ*azWO_5Z!ExUnZTsd1S(
zdQM)M{a2q($L#J_-~iH6On6jO6hgc46)+iP{yv;P$UEG0J=|Y9TA0VPGL-T6<4rjx
z6+Z+%4o4JU_pj{j-koS2y?9n2h)3LJPPl%^8zoDovnM1714Z-5JS7B3tnRs-$hkRF
zaT0a@TnAWR*27i|NS-c+H@YnC2Inr(clW~(c@?APuy#YXncaoi^>yzimpA<|99#Df
zt6%T|;PuWTR2P!hw^%wtJLfAlx>HAy5wKHnV(enVyS)x~2dFQKx$BJTKWZ4x3)>33
zem_zDqVhex;WAblaA^~}K*F*`5Cbp6n>E<%0(L#MprMhMPkx-)CKo`GN~B$>uY(%<
z>|U@7wE`Peg7r&uu3f2^d%!&$gX4PslsVoI%Ef$*r0U)~gO_fOf;sv-@xizL-#@o(
z(;F)H-`(38;$I90YQ3@w#pPs6;%x}wJ@djg!8b!fi7p2wV`cKMC(?ex+KPKlKCHS}
zkYPVw?K6=G4)c`4!yb(xmC@9LxB<&mrQ6ps6M<RDxh&MV8cNsl^c~~my)HiVQDS7@
zII*dnvt{VLY2<^5u1G8D2WmD^p@h?tZI2jL!YRv*?K3&bnQ0^zlyDleX%g2;IOTBr
zY22b82Nyg{;Ts-3C>hJ%C$QLYdMl7#2)HVAweml`wRLLD;7kN&oCNwTJ5Q2}TLOp8
z2Li!wU?{~CM@N~P)@HP#3>!#rX|{|FpInHKpQEq8bFS=p0bibQuyuZ5q*PtDN_v{y
ztu@%E?cVC(*IJ9dx<N9xeZ5~z;3BS(Ln8r|ZD*^yU-#&MULn*UP{eF_f#h|E`~VP)
zz-Sq$zSaT5rx84l3f0HZOQRpZy<VJ~LOR#1BUc~#%&S12Qe%e(i!WOHpG}})qz%~P
zk=~Av2{d)#f0*hO!}})Ozghs7dCNvh>~%nBd*-{YDCa2`z+2vpsKY5Rqaq@yPJP-Z
zmJ3l4I?b2~_QE&Mhe2h_)jdK&&5RDoS^(*PnQ33m8aZsuYx2Y~owv+<D5YjXGqP-A
zZKiNhXjVL76jk^`;y<Uxy0VM-;zdwfe)Opl)Z)!-egjoCL!LMT9_QcNSf&qgn0DQ;
zx_3#hTd5tKCOUw|skQ+3_Gm4y&HuZ&85ME*Z*86+zSTsQxCqg2{;2<}T=?G^&a({g
zu<se%&TNQ3J1AhuUySA#5;9ygZrigxSS7(!`DHB;n8VN=u7uv<H=KHpKDT<}d+%nV
zXo&JVP><f4A8=T6LH?``#dupLOaOJpgDQd$;iTXci{)|L$XOXZLc{cv+U#$4g83?z
zj$?$`iibulrrftDrC(0>c*Js5J1r86h`xCqdoV?|{9;A+_BQdwh06DiVrR`T4|10}
zi<sT?(4VXJNE?IKdcWGD>2LjH)t12`!9Z`F@3m`6YRYA@u_TWcjDp^WCfo01b@h&}
z6j0L({h9`xILDqqzzQh#;_6gf?hm|u?5}b^E?J<BbNa@=SJqz+nHzW&0~!66rc5?N
z&h^+c!FuW2pj%I`W>OsplXq1cd8pKPX&0t0^a`g5vYcUPn%N7#4ctJRp`GodK>F0d
z=G4EFFmc~Dg2fcnV?;AClaTUyyVqRoCtqCd)E62qNMjB#57W|YL|p0TEZ?cB$#qs!
zm`r+b{Wts5>I3NAM_nMV)yG#>MNixC8}QAieMk4~w$lDE80x8VN>}LC$m?x16{#(1
zP|lQXmRzhWG1z%TXYCEi`lZV9vT0kJmu%_yWLg>t;U&c0=Da7H8VBd{I2Y(xLDX<n
z;KXSse|)jNvI`uN{s8}*QJJGZ3`ss}0(m-9(R(}YYEKLszrQdbK=e-u9iTq0j7fzG
zbC;mWq1VvxXAYnQ*%Li-aFGHqTf=A_xTPA+tSq0{lJeuUv56mfVf9xJYy&BBAGd^e
zj*L&xCno5=b>&Ngz*c=<+m-0;OSy-e7}3R0=8y%$Ix0(!?di&INH0U9qoZ$q`V-OE
z$E<#VzH+p5G<&okxElFR;`a#OmPfFfu|;~3b22~~k@=4qiY^>~$DQX?KxGU;M1D$w
zcwXOcsV}b1P0%Zy;y%?Ab!jig<SVq+abx#WC~Yyk5AjI+g7w`hF%Q8wkyszxH{Q(K
zSBTw7u#|A(k2JCB5s771me`(QQc@0w&2rH^4P*EY`%uAYu|OPV`LB{t&#~S)i&2Z&
z4ZQQRJq<GxTPYu$kv~B}AM0^4a`dNEjTk;LSj<pA8#q=Snz>|OjdrG2*LPhj3MJ?1
z;w)f-@PZ6c&cI6(eU|u~L?})Owi|OMk;g4g55zHROb(0Axlphp_39>eV}-3`J)*ur
zt7AS$wvr^rx2d939AG!{!#0;@G*0hcnW*URzk2FcORB68JaNX<Xn4r5${wA6F%@*L
z1=^&%($(Tedw)%hBVmkQ54#^yNrl_YfDL%kKcms)&PTyLf`nB+DIAF<aiD!`8hbp}
zu%`3Jd#b<Tu^wKSqUe2@t&Q4>=1oso=&h%y#<#FNgP?8_Z6+1PlbPh10vf_RuG*TG
zw;X^L)lRSL7yBhYF8j7FaF?;VmAYcGY~o4L0RyQ30*9QkN2hT<B027~EnGx7l;bAA
zG3%}r^t1?1M#vT6DBtQa6!DT{wvEWT5hmSX#EcIX?3Qc+dz8VwSUks%w14us2%c&n
zmDmPi`R>i)Z*<P5a}beuXp&_s59=VNd0JrpUs#hpnf#5}y?ejOiFksF(M0-{3ZkT}
z?~9;^--!U1FBalF^MwNpT%_a#5>UvR(3j|-l17PQ72Hp=Xn?~V4xC^63X(FFnWXum
zr`x!lm2SKc7O}**;JqlUdA8=%?!ShttkbXS9*0kiggcazF|_v+$V(GkjI>|^8%O-<
z?^nud?oBjY2FR@q^_phb{tJOT;sU&$g5f>k7HthL(U-h)N0}=TF;pjoQEwMVa`YXa
z(%j=$aMq3!9y2YR<!{uS>XhR&OD8siPkzrKhPzagFEdvUC2O-c)>2N)re<?n!X*U<
zi*8_{?zm@`?wK_$4>T(L(=ks1K_PY@b@rj}?5__QzN3sIyAAXW)t23O<k>;=;jdg4
zriLiVWSP-!rS-6QOOdaW7DlV8ZEc9tI{*z4XM@DAudC$U2*9C_nx-Xnqj*5~F3zOT
za*_u6yDhX!0iyf7y6E4Cu?lHb5J`c#P~XG!;3`;U(ib9}N4$BKG3+zZS1JljohSa<
zUk#v#R=X(LuU4ZAOtf-i*^9*WL7X<}nAi)JeiLSnW{ns$b_}RRTM?R1jl9mCZ-#r4
z03A?S5dVTXBRdF)htAay-zN=dit$WZk*{vb=@#>!Bj_?a9=74e)d&|Csl1x{HaNRn
zr+A1|VnPleJUNV_VM~L5w<79aRnV<d^jl(6*fP~MVsG1h2l`2Aa8238%VyB~Hh<AK
zR;L==k8QUIr)e<08S<yY-htH%aQDTuBZERx;BAG~FFuu)nqC>+&ut=)TjC*Oxh;{p
zMRn8Fx7#4UaXpH-nzB~JzpxuqF(M!H{3dBBNq)+{`4<c2K(fKlV;EO+t>QgUs*634
zhjm%i<wHY-HJhEEXH~-+jQ~rby_0NG)cF!oD9*0@kow*+6cAOOJ{o}H+DO4Ytzb_S
zUtyp|ueg7YgRR*WZ-<-1_E(4+d8bN5c{tp?AfnP+08R~@c)e|#qj%KD-s5%71(T_#
zz_`oGi`h}(H?TpI3&ppGKACcgw=M5h>gY6HW(GGpRg(!w!Y#<K^N=69sOEWT)&-Ih
z(#-;%5>7ybB%6xk*%VXpw3bl8+c#;bpC{S-VU4er1{++6!Szmyp;zocG98NiD}$ZD
zr|j>~Kh>S2trN>Wah}0xN3r(y{+e<oxXG4Bn7@;QK13A1O;Rw=T=I4pwJvqIdW}6)
znffGLB%K5385U13ri@5$1FuX}y;J`vnubK9#)TdLBGO3^$8GlWiu!uLo95SBJy7zm
z(T;{yl<MKi7r_YEh!YFC;nCCjxYo|s+=cV*0AUyq1$9I<aWaZ<jq-UACcjiN$b0N(
zb^-%)v84^gAXDs(vZ_=W9T}pV>4#53jkJxQ&cLi?^yQU_sEHAa{YDkhNpc2R`aE8o
zauR#?OU9S#>cJJ@iqdGy3S-BCf_-mObC+H3)SAY1<qmO!_jlNGc@rT9z{S2lY*9d;
zPyj_0YI7EsVEW2~RCvqrWBu3a*Y=gG=oi+0zKvP6D!v|;x8h>e+XkA$^GjuO_hjbs
zNB9*Soy`o|4l?%Kv^#RxMQ-tH3HqLS_TM%CQh``rkJJ?4V7?xP2iGR`D8D{FCJ?dc
z;r&ygs4U&PUpURqbK10)N;qC?)6XXh7tV|?j<orMJ1LO<H9d{}W4b0hEh)3Eq+a5V
zuDb@i(o_npG!}#DGUPM_vr7xHS}}JE1>I1F(OxsM-ynP7)Dy_<SDBhJaAd$_*xk8T
z!;0xvb8%Q}UYc1P&Lmeyn6O25yWy}6Q)ZXp2z$WRfHitFKnrz4St4QH6ZZ1Ig!N^x
zV`expZDkvOv;(p7#RxLM;aaF(3jEQ)=sA6?4H_lIBT4;JkbO_`&0G6Pjl}Pn9CB|R
zJqQTVqVfGGn$OSDt`B*FvmGsp%C?t{7U}lYettYN<p~+GHxN>VwbKId1R<8<)mAN6
zO9}S27CbEw3s{&*g!f=jLYlb6iM~<ubexUN&P0FntE)tNN1Eyj7vF#Z1Ro0YZuJM}
zBv?Dm#F+P9`1&1jLAL0-nfIyo=h)8=)?dL?C+_@?(&=^dt$uMapHMq(IZx_@0_rh?
zA&AAY7&{iUu?LrI*|5xs*Cq4zwrqaRQc}3^h<#PpegqzRp8d|~qn^o}k8dA~;6&OR
zGyXT9RAL<M;2;L!_$NmMx2*MRAn3uyYa8mI>RYKWvAYh`Z%<TKBaO1MW!WBPC!4Le
zi*fcXr^6J*<Z8<vK8vi_v=;NR4Y0cVfLMF<b7F^z*%PdF`<t`;poaBtg)!gxHZIGt
zGrFS-Rf*0n=_`AntQKf;PLTsgSv>=(D$&&6@GPY~8O*RzrS4uJn6d#TFX^LlMT_OI
zzsLy=4jUt5H|v^lGZ|8F>BQU<#8M04S3&uW8_lV$j0d$?4w=O1GXa{(oWk&gG!IkA
z!_2beEXDb%>@9q>|0WQ8rGyDKl_r~cj0yZqcPgVS7N)~Y%K2AUt#Ni)nYAS)nfdkg
zkITT@yDUqYe~w|dCXi8Amcr}$y78_5W-&**hnO?gZmQQn1n9f2P_+!D_ua!!<mB**
zfS)k_WWLX0lpH&3Z2$-jbitu|%0hiG>Knh_NpZ{U2<Mv0_l=%4qs+t%10~<`*iy!L
zsRlmQB37K8V{10~?DM!1#Qoo@9!o1CyS-wvjN*%Jxi%-+2Qg=!cDDXcKLtR7&P0z|
z&NAG(_)B0J>T++E*ro>ce54@{-d%I7sUO-}|Fg2G<T|<))<n3#Uh6mNc7Uo`?F(J*
z6Vd1SWhKb=o!~TZ?26($9z=dzat1|4cBcM4<yvS&1hvocQ_PH97eptze%g{l+0v8o
z{auc=m83E$lbwC*4@PKb>p&~tyo}gAQGsjTpT8ere~Cp?U*yEdT&XQ9j<`zkoOx|$
z?mJ4aRB$Yt0~}S%SsG!6kN~}JAyOB9h4Of%J2bn(+@Z!k%al&%$dAv?ekf5ZrlAZN
zFVJ`5APX{eyh?RLMMUj{h?o5>_ucY|ok$4@$oEK&X^h37I9nv?3rKG`{T>dCJNk|9
zhb;w6yHuDv*PP3Tn+WpE0T)`xs@xH$;ATI+say42F9_IXhhEwekRh=_5g)Q&meTD<
zLyz}`pR%zs<aQXH`t`$GI^DiB*%n*p1Q136qh%YhZhdB;j>x~&Tz4Z{<8CT!(u?<m
zyeX_L<7mM(hJW_7OCQsE=vEJIiA-K*j#x3zd@X`7!l)R1P7$?J-^c$&Yb}F1Oxwiw
z?(n9A-WkBYT7I6o0UjR*rT~rpBBpodDC>ZM(Oq@dT06%r)B`lK!z~RM*(pH~Av~be
z_qfwkPY_b#G{(jzdj;$o-d{QKbuK3@jd0=7gDXVhk{Icc@aM0w1~|xHQ1WUUm#vYb
zO~vYRIM~1<y+rkxHGlNaM^n_t2_A`)mn0fTdfV%L`g%}MPGrrMh4+GvzDzB-!6*HT
zc0(cGa<e^3>Gh%^W{CT@eyOo8$kvLz$wbpT)GFR@i{068TI<3(@hEn5c_kpTO5-(3
zOYYY@5cu^BJyd_aIl>XmV5Q^fmeGl317*gTBgxc1h1AETFt!e1b`vPQ!dm>8p^D=A
zX88E$K=A=ym@u4jk>Ypatc=^*GUsvuRWJX;y*bWs9(W=c(a~m@BCM#GK)x4>4z*Ev
z9b+rQALq%BcmzRk@%uo7`7#fLc^#jFTWLWx*@?on_s~CA0efn{yOq9DkIBDLQPhpB
zE-W3_6puVfELrkp#aqS-gSvVxBDuK!XkTr~2+>+AVo|BB2L~T?`|Qe@)%LYc72WpP
zUkYDy5fW!$?ejr<fAIa#zx6$`<Plj*f(8W0zKVREnh^Sq(O0T15O;NTm68Sox}9&I
z@lYktG6`I<F{gpgx$f$qZ%}VRHa0u*J<6~Y|M|7jMhCIm=k*w@q0|dLmVVFe=}H0J
z4I_X;zWsLd(JLNsTnV-7vSO-1L$l`Ewn5+ZswAJXh{ltonu{mmB5ye8B3d>~DmX#<
z=IL9k{RJlNP?2q?z7JB|uTfoT_MDNMTev^lnvQ$G6?ltn&@uLmKugOpkmdK3h`hMu
z4pPtl!Q(C;{uzc3ZF%TG9yLlAb$x(UOK9kCRkv(?_@S3s@l>c)vCaK;M)kOvzegno
zsB`(dx3x2y>H*yIqUD(Un@Z$?$aI|XVE2+ug?r~im)yZczN>AXzeg#iVqzhI0d9_H
zHe92vYDm~Io**rYKC7<}wnCN1A;(KLodULf??={&J$$}BTZLut*2X8-9oK`_-*a>r
zc{4<dQIixJ#6UWr6Yf_j3g)2SL1%<m@Pc20m^c4H@EJBdJt;l}F4~2&Xqyr#E<tfo
z6Lxv8cLRp+vP#I=rBG{m<C%f4PWcy07jb;zIV8Ew?98;8mr2;w1Db`Ut({CleK-zy
z?zWbXGF@C;m%P0_M3m3px2<Cn)@aeW8@3Fqo<u$$(b@(>LaE|mF{$a>9oLUB`Z%#%
zgy!i~si1GLQuw9xVo&9(mWEXu0teNS1|F(4*8X`*9Fu0ZK?6P6Dr2C4P;1&FT$>uF
z{FKu<pV(zM7s!X4>_&XfI{qcC1sW{9^+0VSJM^S_@>so{F}Ik|_l))ilEtfyxx>Lf
z^B6UFd#thj`-oge=d7384{qN`5$n*S$s!g3UD=|r7`B|I65JQA`vhiuE1S5!c|STz
zN4qJfk7+-2)zL9q<>T~CrM96fxa=A`sdP3Hh*vH&R!G*J0R5GbsFe~97<^QYg_1M>
zrFU17d-=^v$R#E2cf;YI)0mw?fD5u0jKkhMfi|LKmez@cq#z1Yz!A46dS}?8ad%PH
zWettm+RagA-v5c$gBC9qckqn+`JWg-D2ZgBfyU3(XJkK*-Lb#&!<HW>`SFyZ;xRqz
z?~b2a>&d{v>X&~o_RhNH*mum$lgr}PVEpethrbtqJu@g7Mljn5507rr8Z1dxWm_pN
zsp2d4YbM^02N^2<dZ1+?tqTcBwe)kSc0Uu+XIh1J{AtGj-;n-m7AqRJGCY!n+~DC9
z=9II=s=Rh7w%^mwQ<BAD4P`|T=@MZQKgnrbpV`Wp$^2b9Yl*bD4&U?Sa(b`}iUTan
zp+}}9?FA@pu7sZvoUi4O$7eJ48K(D*`GN>RJP^+w1fVehg*VWGNuda`y>4%*(VMG5
zf~a0iHwUrn`h*}0)-U%eb$z6w6Rbz!1*BfvkU#4>_c(hdPLoZx;W#2!{}`9iS``ox
zu0C-BvGt)VErg%yn^8}9*l{-<O!G)7MV-tZQ#2)-rKMq}E)4jgpI4hnLG^lagfA4h
zk4xT&pj9JPf7<)@4&Rg<h&<)iSr<#NvEkQ;H~6>xcAPS_<mdoDEN-eeHRgKn^_0mq
zimgavpT8Lbmng0uXd+TZp<E<b*e2CxcZT#8@-Z-GH8(F{z2<w_XpZ!jI&(;QIt)ME
z9U2c}<&+_L&SaK_i`a?1nK~F4jcB%vm&AnjGbr|kDr$!JK{oyawVrL771Sdc_UKc8
z(+JSSF??0TV;Rk`j%ow~Zo|V*d+>8td4OJQYlj2peE)h!;HzYo#vjZaX$!(-C&zuo
zr6QDi4~<qhlyc>gb;}oi&=u4LU&^rIPe{Ibx>J18K&n$XzxV|jNMy4%|NIZE74fRV
z(LdG5g1_g$d8A`PQM|-NMq^T4m{ULHa$>F-2?!zq(F&?>PMFQ4r9k*4KXhZcH~OL#
zE_R6cfc@^5oSjzMVUe`%(I3|s_8RMI)pwPlo!p5!PgCpS<!49gM@@({nVX}4_7<s?
z`zrgS_`;$^V*UkLRKZCO^ZB&ku(y?})};kx^XI74gqEoV?@*?c`=X8moPxeY9<D^w
zD@TwNSgpx9LoAwLELnCumM7i4>kk8`7_)bO|KCdO#0|*T_szT)WFB{gy7aX5oak9k
zytz}Zxf>bwj|cImN}|K}rav78s>YaCJm!NBopUTH;A0nQ*ZL&{5%#@QK&_^=54Yrg
zr4GOr^JynhejSr{9ZU^Vhw4~<hq}7$miH&dD@TH^*!Zr3dVI5VqB<b5(4;I}W@y8C
zxvbG}?SRmetO22x*w1n45kjC?M*>1<P|*q@1PcJ7h~SpRU40&GXJc~lD(j=bRJZHR
zmXZ7u+QN&%!H@@P7D03Fz;U7cW8Ym{cRk-d7xiiBEuAYMcG<u|$D2a5_BSQiSK{jY
zyXm~(gYHfPK9bFclRYZJg?czRm$%I)aYUkm_cWjR9>-d*7kU9{ZT6t0%ONL>DrCrF
z?muNJ<y}?2keN?VAcK;j-A}5I5e_Vuy4unO-VYb5Nx!KGC69O{9llaAlr>I)4nzi<
z30wfO2^Jh6!bQ=|fQW6W>hYF08RN*1SS|j8C!dF&{C#~9G}UU&X^I1c4x?{z15u_`
z5_20>Kjgl|K(df0$^Q|Nv`~)6oIp{8#|*>o;ze!t(C4265EiRU4-_M?rVHBc<vloY
z#%d9jZ*|6#A3AlBW73lHwq^=$vZF99cW23b|0Q(US)mX5OQo1}`q`W8C3bZD=xC+g
z4PK?6n`ne~Zbii@8jh1K=^<8IHE?x5E|BTa(G^#Fd!j%ux3Xzil?db;_C|4lmhc-0
zNIE3nn`1wWX;{fqn3TmxMjO)DCpHdKA?=`hUBH5^kwR@%Uk@7u+uU0;FEM%?XA2Li
z!d`?v6x&PtK?~ZR7y6%9qnKJZJSVmF;dl>Bzyh8K?Z83X-t?Q#q69dOh3W;Cg+_)S
zY_9)ncu%w*U6i-K#SMfZRb4A`1|MiabYuMH=ndrV`kWdfF{Gro;b~ge`_@`WT$+Bx
zV>q+^#5Gzq<(~hQ;h`1*3s@&OscVl6C5yOwz3TY)rxi$u9{@XtJE0A<+;(tvffh5i
z9&F`=%Py^xV2>kF&i!xW+-9wO0M4B}N7CqkAHPhWdHMc5_qdc@QF;50QW+|7@95}l
zp;YO&)wPswzk5%DN-rK)?!P{y^=VIWD2C@I<V%ulk3eiO&v9ntw1IzP_Vyd1?0DU?
zOpforPnpaD+>;!Q(FuWp?H?VV1m=5+T2C#YaNLh5Y_%L;d3a8T+|z9dkJTe^bTLXt
zvZ_ldTh`f7cLLas#`=)5mi^E?1#Y^&P~D`{9Bny>=of+%ayV)6t^VP;l@gS-$W(FN
zuB7_YX%_w^h8BiTpAuR1=z@`$8f@S>5Mt&#N^O84kdhQ9D^DPZjxDA%>Th^Y6olaz
zE>vvjH8t=67%VBG2%@k>uFxQ$9^vr~*Xf>y&BKVuy&c88X9>cIV@TtW_~K7og%>jP
zEntab4W$Ut*YW<}S=|D{KN<pcPMlYsRwExX|3e0|*VTuCX39rsL+NR3E%&^{vrFLV
z%4$AbI?0y!^lRh1zBAIcoRhu#AY7kxBcJg*IIr?>y0s?PP9c*1`?;Ep#G$zC)*nHd
z(`O42wnqlS#$F%fA&;I3dy?M%lE*>T&MC^ue(K5bIB~9O%AQG!(Cvv2ypuioTUd_|
zo{6Y`2u=N{c}C<-eAuw+m*O0PVnST9o<~Rpd25Or<8(*(7WZb4WN#!HlZZ<#1gb~8
zNFN9T&JOH&pQ%K)p_FdS)e4DrvU(fmNAv^}O~pj}FJFKNf2Hwg7b)Hd?}G`vPQD2V
zLH)k)-f-1cBPu4E^$#dm+1wl?5{aXmP5!<kko%WUa3WF>@(G{I+zj1O#;6p|I4D;i
z;tj#Pg@nkJzXBd)RNeN+IcG=Iq^Cy6W1mhPf5@HA4BrD7U=dLbJcHF!9oI!Y0`Xdw
zgH^wL1>thXq+dX4Su_0-vL(JH+coqW>kYv0Qh!$02{+>TlH<{3^NJYb?6^Kr4>b5P
zYSuqI{1O?n?RK>pTQu4Da@^6saRiCHqLY73@L8L>Ob7qSz_kwC1C51FJU}+<wGsb3
z9`@)}@R{GAr((O75Zc<WhRYXwv#1jFVu$XTA7cYGESH!KIVg(t#ME$_X|Esp!H%s^
z+UeIto<_ny|KHArD1XS((Yj3Nb239mmQ43Y8*=K);iQw^hH|S0pW#b*%k_K1g8M!w
zL|QAyg?ti6tMOar@@jKS<h<;GvdpfB7Q*AhV^~>>-BE`v?q(3)uIWrikg2?^@3(HP
ztMrJ8YU<B75AWYRcanFBS-I}*tR_trYAMc={NB>jScb)(lS94J<MovJ27kyq7USA&
zZ-(|?iS&(J=MwaKu(s7A=&7bPWXrK}%rC%ky}lZ^RsC|4Sah_t^fXglF?9@KtT!hk
zo{;}LV=jkTU*;g`&O3t|?`{O$a$=;t)c;7auMTV8v~Jj(wN>QsN+R1SYp+(e{!9kV
zIGfDR=H{;pa=j&So_O}g)AOaf$*mA>8LuxQ^c@CB$!>7_Jg0HX+-Cwrr#{xgU}`ZQ
z^{)vRr!St|^Dxb%{IZSSjX2q(xYBHQihGOjBcB%UY&$Y)L_-+<J@Ti18$=%R4KC@h
z%$PyNrb2$wA~pSR%AIk|9_}8@?_nEn+oivl`s5?awgfr4iWa9jX1PH<oa#E>aNrg@
z=qw1t&RH}v+q7M4CuT;gVNn{V==PRR{=*T#gzt93!Tceq-5J4Y;_AF6?r5~_Kx7Sh
zobGb8x3YjcI9z8kC)TSs3bqAl#<3;yX;6{BJ~uzhP~ONKn}rxyJ>%%?f^}5&Bo3un
zi2mk;XeO7YhD{HsAo6FoV<~U8<HM7Mgo=cur9600(aw%$L9w|87K2rnyI~8DaA(=R
zY2H7K<jg#+=gUGzQ)w=K)3Lc}RoHbh6bQa_XCUsTiX=sjR@Q`Gq1O|~JDWC}zDvtZ
z%|oF|6u<Uf<kd#CQ7fK=)OqOvcd*)Rkl5$J9BFB*r$R*;Jg4h7k$yLM|7&?u<FYLa
z@%Y8`{<&qJU1Lu@&CYjp2hH@vY42tHx}PO2eYF4+aC?2kz*MQlT2NLWlkG=utU^XQ
z7cSiYcH_of6ra0ejbsaU8^!k;)Mh9{AUL~gjgR*xO>KUguLzZt(+X7I^BB(>8Sh4J
z1Yv_rIAYWOiW>-itY29tZ@Tc6>Zj~PL?`VsRdg$ROyRc-<#)IR9+Tv8Md7ck5Uw?7
z$wykCsetXsI+6dZR*Ih9s0g^-NW}1U{uGoe+t9;mWHr`ErezrN^mqSb?(#Pdg*pao
z)6oZ8f4BA{js+^MFPgf=GA{gTjn)u9yLwYYg--q1IML{bt<AY1eJTlI6-SEue)i^u
z_@n1tIfF_RiqAU8FZLHVPOW&m(baXs&IR36l}M*VvBD;*8GRjIgccb%Pproh^prtH
zi^+tS%4?H1$8`1tJAD<0yxkT%zKp%e=8#o?vT?kJ_rib6v2N$uG-TipE~K%=ruA4x
zN|&CHh1i}xIs1n9y3j9#+sXfA|C3jGgS~FB^Y+VU5k=$4ZTBccE)tHt)z@J|^G=P6
z!8cL%q$)G+v$hGblke*!E|475<cnG#Jv^X3@esajWULk8-<w$6|0J`>H!B!57wT;O
zv)95<{sV^!nGqHXo?fa0joA*Xh1@l~kCK^-g@)DGhg}w!DzC>(36aV)VlTc`ek*JR
z&ne9D?1I0ONiUIj2)}_FJg}dtW(Bxb95Dx(0j@67jDieR>pNWRIU?z4rwv}myTt2o
zebCn`*m<C%t(|it#Kp=y>n`EXM09l12A7x2>aRVQb;%3|@FU5F={ylAkG8h9mbMmW
zx^-BH+(RA3-==VYCGPSPFMfV;S<he|qz6*(s*Su<Q9}&4n&*Ps=(l_vPZ_CZtvX*T
z@XeAp{?EZ9_#OJ{_CyTObd4Nuac+ow$sLAHsZ2`a6=_Layq)bALg%qE{9?+fV|Lvx
z&>+zL8)E<8pFgd5e1<7y#6J7wi2qhpnBsQyv}<184JL1_XNGs?RzAzcpWc;Yz5iCR
zJyt!~*+cZ2(&#wuj7);C@jl<{(Gg?!N+O`3ls3xUG@HGjMS`^jraW;B<<>L7%8G&`
z4@jO*)O&HY<I&;|TV)nfF{L-noi`?4JE;}Zvg55_3W%#r{shtd=D~`|hRsT1=%t$S
zVId}OT=VlgY-#F+0In}wceI*VNkP_cMNe0Z`jgj-@^CS!@aX;DV|z8TovbFvte)!5
zM-^^g{_AYLzBaX3-S*vQi}>#O`Ffy?aN)l5=tOk|rOMa<bS@#~-7l0nr|Eo{&#JxV
z)Yi`W`oD?02Mcj4Dg`{9m{l0Txa{LqgiE}l`k;^SX@~*aeM&V$K~1q-!EQM1|7be%
zXei(R|BrphG9!9h!=Q-l+aNp18lvoLvSjT0auYEmTe4><OVOY*_CYAh62?--*bR~`
z%aHARe}3ou4u3nhGjm_p>v~<!*W(eFu-~l90r9vGkfp^{6KU!G)1gQWgL2GgOv*k(
z7htDnd`2$O*4nvfs6{Z(mlBl^^%%FP-&U&${kQ#j5bahu{rTPmn=)fH4P`08d73dQ
zA0J+vll|@0?X!o}vt7VrsRKdpt0q-E)ZU)+R)?>RE&A1gsD7m5)GICO@-eQwU<(B0
zCux81{I7YBA8=%YNHu(+0%YZ?we2``Aa)q?=$1}bsAAPGfdQ`R0qu1C<)jGi@u9Xh
zK0x)YK&j4KkBRRcqE>Vp4L)i)TO+Fi{LNA%p(^C7N~Nm$a;b`p1Y*Rtjm=yE5Yuw8
zxi<iuo*nP)9e|B@%d?}+e9zwY^iR6f@FHMmWM@$H)!(3k)pS!9J~dSnJT)uVJmz5N
zE#rH&REeRV@Ew?Fs_6ob#du!2|5){Ge^*ow9B7|Qma1qL-}7K(-IW{K3;2#V;-LT`
zod0p!$IYHJf|0|y4mFTW^DkJ^b~Dl!JN|nsyg)7HXf5G4$h~?SEQ`Ulh_#kI2Ljjl
zd=j;}WfWpUWw|}XEjPD#KXVAw@7={+mEq>mz_ayuegk)pzP>={en_Nf1v5(n@=ry<
zuZ}5mZVLeh=8Qm_oQ{_^wu<=KX_lMId^Kx<X@ee>Z*HvPsGPZWpF#!G+)t~#T79@F
zY2gTHW@doV2k(tD{To=#Lw+G4fS;O<w}It$SV?wz-S?SqpLadDJeq6>jVP)ZLMc~>
zn^J@_N}F1%U$ipw7kVDENj?ZB&XSIuhm(dwAG<@*I;=|%#5O{FFb%U)!8jtA00&_z
zj@wn2Z>bf{)sjzaX46BDu4`_PpUU638JdLh^ZstDyR5W|7vGbkfy04<x~8DEzf4WI
zMv5L+)~hm(f4nv`rsAHoO}9>gQtK`eYO^o{XI8iFwJT3=J++Da7HyD4rCU~8dv-Rc
z8ANq+2qgw#PtpAhjqPVQHoiUJe^1m;q!V&RoLaHB&)8jSdQN$R<SFy%Rin=woszOX
zj5*#k7PkRusirK2$-)83f`VXPCeABt?ud{#YJ*SIWInL$=6gR{91IxVm01~(xhd-H
zy{2&YS@rY@-&0Y&SPs#|#}@Cg1&WAVp1;|-x_iOFi<9izvtFD`4);Rk?+PRIvZU>}
z2af8k_dm=-AUwqR=pH^vj*k+XbghAMv1vKKqzi5xL2t`I=DQ+*qcS0F4L|PaGZYG!
zvsF1=7IIf5hLW-7meHNzzz9|2D&)`|5?dvI+`DlruSnaC0h<djPBy1@>*DbE_@FVj
zvInV`YNVRjq)n$mP1W)>p=b0gO!-zZXl+<u-vvWc6hiNfr!XB`XR{5Ak)n)oKaYCp
zXP4L)h^PO969EBI#_uN1269bBD+X3njek)jAQKwIs~#e&<Yd38%ap^CcJ3IHTXDn^
z$GSNbxoO&4y;0zDCRmiOH%zroncMg#+C8@)m`tP!kV-n`Q(<kNQ=r|(r`@%e_0zpu
z39sEqP(y^HHS)|EL=(deR`I7he#)<|Y4oX7bF#|N0}vE}M8X7XRNFsI799*+`wiZ>
z28$U7XS}#bcT}HzQ&QzIiE0Y5^~GXGS>IwPJCf+yKKU8>EA<Ux4&1A!?S;hmJfT&Y
z^Zy#RY&VFB5pl6`U#GOFr0FgWqIt4j#%n&pkU!ps4_ho#<Fi31-&sr~-`2!zYfXuI
zagrQB;K5GiuKth~7rmLgC+HljZL30}=oJ@TRW%-b>`rKm20{P-4lAoy@hdTHSR3H@
zz;r-{xD`1S@MrbnK#(%FtRQx@P~O&Ckr*j!5x*B@vKyoZ|5v1)+Dk1N#`nVOHrJRQ
z<gwQnR>5mp&)LDUjy%*rcTJB`hcBlejRb5`w50dnS2XSwU6nMH-?nNE=AG-^F(Q}c
zPd$G~Bf-mN^f{u&02L(1_nE>t9Ln+3X!LWk*)o(SO0ngx$?XvXA7jg;C(o0g^tU0q
zZdb6RaXk*|NxV!PqTSKT(4j^njy%RFHDK#FPw(j)enwg>15x>7LgG%9OkJ6@OILwE
zbQSIwtCDmcKG90*gZO*jL1;NPGH{l=f3nruf@X?e!Xqty?b%kjJDNHORx}<hLgTez
z|5BJypwv6m?>c_Ep@~A6&Tq4|R_rpq|G{mJadrAdW?WB~-CM5A`UDtw2g@DHw86}v
zr1Fu$Gw20sgl!NGL(ecL8!;~|WCJk~O1Ex_=OS<3$>A|~g4{hgu>wzT7w4_#2Ydgk
zM~cCMge6FXN$XLn|Hz2mID6b>BqdFm=+ml!<DQ<C-GJg;d_$3Rx$U~6FQJJL;o~#U
z=g-)ZcJle+Dt&U2J8i;lRKdk*RY}vq`RVEA=B7`93_OwTj#~i<zaG>96s@jg{|(|F
z{91No!cP-)(E5_#u#I`?@4@Pq!fLV1QA>z$Yzm$;+(<xhP<|y0oE?G%BIYWNn=Ad$
zjAa30*&NAdL#1;)=$aet0@az!gHG+%2(=-6c8;*Rok?zz@JiU+7DiG25~~n3^#k9H
z{>vqmEuq?r*?}3<UE~wsJOnp!?wZAYWM<@#Y18Jv2Fc*3kE5dMb~rrfpmc#cTf5Wa
zMOwc+{)xSPio7&zVI^Dc-$ZD{GJlqO{hUo1!IPB+*Hb-T6qx0H&j$(Wo5i}MV{H|a
zyXK*dgO{bhavb_d+JgIrE*H8vz7<8D>O5WPq}u5@7QKEc+aEZs)2^qLVziKUM_lDF
zfIJ|rIG(Tk@b#VY0ZSCcYk8{;n^@7j=O6w%wRp3BQumzsHrGAb2Zf_HL+wALOA$+Q
zU#^$Xk7H!DJgaGmPRm2LlO`>jCrDIi<;v4YpSjsBixdln`6~YGEXIS=?69<ef<^qp
zs|$EAS&=yQ9~1z^9qa|4Y@L4V=rJ==tIozgp<>#WO7q-Tdg)n2k+kR}!q$L|=p;fu
zG**WMePi)g$~@e<JuH&WQh_zu*{bGkic!5LFsrhoBAk_x9KetrU*lC%C2DUHp7K7p
z&PrN;TfSH7CURxGwg3E$+UE=We~6;o_Ce=-yl2{m!!}@x^TlK*&b}@!l=Tqm6u-j+
z=`B!P8~S=Rq<^LFs?{_!!y9o&iC;rD5KEezZ|h)$dCpX%nU^GU2<xX(WX>WFwZqjL
z2#<eP|91CCOVC5FE!%g|WKO(0aVosZlC5&Nw~M7}eQ9aEst?~&WMYwrxGfMWhM7D3
z*GOK)o&)D#d$Do0$!p6<#PzSfh`98fl%)6fBl({5>-{hr4m@Aq4MGm#*)l!6apcW#
zjIS+Dv4%J?;_Nij353VEF}72IE++WXg@)y#^6ayf16|9D(^YlHI`H%B+05Bnh6p66
z4vAKn8A0YXIll<Fc~24cRzdBWI<OVYf`9_qE`M<LzEfr$`;%mIY>gCo+VDP=qwj|D
z+9hqhkW=$MV=BnkdN=g5+dopmIiNhwfBq`B-j?pgXMsG?_(56Y4+qnzm5$k_=nb49
zTV~HK5*l1@w{eQ_Z^+8L$SDqF#YlwjSs%$<os{Z>qd>-YjSSx&L{8Wonxi%n0kr5l
z>f*nH-Ho|soqS-NV4C}qaiHF7czb9%VQ&DVz<ZNYNKZ(oPaEaWW1FPFlm45HXo-oN
ze2ta6qdhj2EP^iQu$!0|e2+Cyl8bS(xBcaS`mq;uj{kA-cQ@H%u`56*gpvMwrhckF
zV$n=4xj(>0E$WD0v$(_&MOJ|Tl_H`k>{zncOolIxY}F@o=1^R!%uXd@Z#fi1pYhEi
zV^4-YeETImoz|Ng6&E9T@=)oxl>sl;o<h+9GZ_;TiTi=nTi|(UB2B|<kj!(e`o+Jt
zy2s<i^Xl^>p`lXlczW1n1rEfDz<bWV^)RI{{p1;C-anyq5Fxip6Av!{NA@A4DKSs%
z8g<yd)BC0rQ<VCq6+5gil8PQw4RZql_2=nu1^!m4Wf2PlBPZOe)pjHQlyaX-ZqiOB
zqLq28{c#T{|L)Um?xT-_iUZR`$Mn~OAeNF?`f%p?-`}0IkZyDeZw@-b>0k={fwu-9
zLYKsLLX%rEAYq{#5f%X5bpbcZ>j=JU<T#Wr!ezN@P$pv(#}ql85l;hojBTqfj4P9Y
zLq5I<zIRz?E0&>lZQD4C`oA*aq^g0_-QCk&f26R(>+=5IpRKhOxY~n(;FGhJ!}a5T
zA#;T{WUnKfQxJk|;uRbg8uha_HSB|V(o*6`wO6a@EQSR#8EgPKx{f^S&;;)Jw9#Su
zUA(m7>wbi@J^7eZvCHNCc(#0hdtZp3a*3L1v<HT%AaXlJU!al=uTImm*ze9^_X6jj
z4jgZSq|~#R6ZdHgm=26I{Tn1amhmPrltG^*V8NNv>Uy7(j+c`K`E7i$@C&uX9_@VO
z@;thqTKf@jcrL^UP$lvY$k<;sNl#z>oPX_!4Y@X&I9W@aThZ;+5y-Q}JJ@bazu}f<
ziV91WF@0B~<q7<hx)RmUnS>baf=jL1{dO?e7~*Z;aPU4pqHsl{m1GcUhL!K~{n6st
zIYs6%QEMAMVk>pk8}nUnR1iHU&%L34&OMvxWS4*iz4f<5gLN|K@MtSL>}Z-Ugt8b*
z8zaFt_<7U7N|q8l&vUO4)aIU^Z3FpZSxc!hxQ(W-4yJ>wu0mX~l=`AK2Z{pz!PNN2
z2nSvyYK4`NKEgk>XMI4(CQpq|!(CppS0M+rUhX$PYj@`<{Sei?hcV(PPqn(lPYO?y
z1_T3rTYXNzec_}v*StVMUcM8xg5UEo@E`WicQIk*cq($+oTV$Fn^-p9xK;7as|grX
z9fzgzqKV24CR-G#gBrwGp8Vvvyxn;b#*pEi?aT0YHi4lBlhZqkz<t5DUUfrl={n3(
z8l)cC>S>#v&?`URw&kJWNmv-{&rc-aZxDbq6{V%@6vt;PWyB?RNv-Vf96`|DzqU4r
zGv?Xq`i{vEzN^DaakV&vMoQe(?!4zuI62B>TlLXsO?EPYx4nv^O|wb5aDyaRo#(1=
zB)t>MO2M^XOS%!4{FIVYN#<|d<SqWo%nH?Crz1qMFS)c9`Q`rHug2cr8C?eWHLfbC
zO0#&m=qEZUm3zAGe_liGgn#3^Y6AYI=N_|3v8KYqJVOYC0UHov^>1pfGdiXtXO`YG
zMULTFc(KIlKtMs2=w(;TN=&raNiFVj7+VA=AqQd01B+~vTVz}(d#BY;<)GUtTIqe-
zq5aDj;U)eTv#&y(i_OtmfcsfTqx#vujiny#)ZOD1+opJh&*GxS;=3gU>Pugt_5+QM
zC5|9VsPi18)#P6IJGi8=YH>__^^-S64h`^F+`t~p!tavZ-(nA~B7JaJW{tT=_H(|Q
z%#hn&*I)uHlccaqB+~C`?ee0<mH+2tj%Zd@D{2I1Btq|A;Q~(Rq|-}AFs^i&rPbI;
z+!d+qX<OeiMWKER)mN+9%R7Q~d<<79CSv}4c8#TM^dW3U%#pe9TD|VLNkYf{Yxz~j
zv5W<>($XwVMHG2l^wM{Wx_j`8YlKNX>Nh7+m!eAoVlKw|_`cFH#d`e490uRSU%~ST
z8HxrInxf}eyns+N$_hJ+x(_jxR=Z^valh#dyha$<_+cjVMCwtmnd8W4h^sSmcbqQl
zPTR}qT-e`i5b#9C{fT_VW3Y({-k^K78tvWX+<Lh5nY&3pREFvgx{&f-pkizcWvT~q
z!0y3I7dL+Q%)GyHH${yR{*nhFZJny>oy<b#4uxL#L5;|)d1;rD?l&`?usa+H|LG)C
z2#=!(R;Z2H)?lB*81}0f?g3EL>#j835vTsgkQq1|#F{!UC4}QOvn-QVP;+T|)|d8B
zpM-5?h2G5Csh}c19Ui6T$Ut8VnA6kLDjkiL<;E~+purXr4J3iSAG|02rj=rJC9hH_
zFm;a&Ac-R9!Dn)G{F_aDN9MEOKOL!=nD)slN>xg2zf4W#TPh5@>r$4$Hux^?OIKxT
z!sIxnrV%x-Bc6#2lDaP}v`^8gG%gl@z_JzaUd6r3s-{{?y%vqLsUfz)4ZkBvcOKaJ
z*i{9XXsDJaYQf;~<{%tHs-7(4#$h42D+5&Y;e)>Wo5<WQri3@QZDTrXSID^sbtMkx
zdm96L1kB+uI!lSJMSQPk<#gZgczv}b=b`)Sri~3=F86N&-oCSuZh4oJ2HAt^LT(;f
z?aq4K|DD#ydxSgK<<;7ATozAa?h?A2njOe-5HNfkrDlvD$Fh6^1%b)>snRTzaRAlT
zE1Yye-ac(}9jE8y&UReX$HqA3Wc5iKTW?GgnCZyOhs{K6U<5L1tjQnARuR7Lw7scv
zWAcez-;zB1+T$tc1a_ozcyU}>rpRGKWk*t9RT3WbUG}QF_n>M$l;%T`k60-iP9pfO
z52^B=i70B6Pu~!zwlP^)_?RjUP@MPQJnXYkRC_WL(HClPUlcn@+f{{kWo&6YJ8o%o
z#}U|`jwJCed(cY>FllPgU11a%UhVtRmDy;X$a8yPEs+=Qf7Fr8c)hxXFIQ#<W#dI4
z=NJyA2)L0^l@R9mu7UDu%;d*yEf7Ow2HM3A|Ni?IZcqgy73GVzU-AH8nRyQ$BtHK_
zjPO0_kQ}raR=?QCgVBktLaw(3jg&jJ*nbXrsaOWTyI1Dxv-4bo$y>|@%OYsFh|BlK
zHxgCN8}h?b<bLc1wZ5H>C|K<mHGm-!FpbkeVw)QqH>FIaEBTkE*kAVw#nLty?|xj~
z%W5<?#BY@<Oglq=uSo#Rqh(Z-hI~7`z0O;H7Tp@+%(XaAU#c;=twtSa_xxKN%eSX#
z&h5%$Qon9+@JkC``p_cAfS?*vI_QcnjY*b*zM&G-DO-`F2=NR46nmxrMJfbhR~c5&
z?U^=EWUztRWed;Gsz&{zL+xX;$Gj$(`=#NRsW{%yhdVve5q1Gwgw`1*EyKzw-?dSS
zyygqgFL#ZLW05TK<o14<jAo&{eu)mA^I1tyano`aQXOCYufH^1G*KT-^-(xZJ4J)q
z{&Om^SG@WY#nsr=vTy!cP$LV*T;W?qm3W?9A&Yr7YfJL<T;mxz_he43^Slp;IoprI
z`Le@Z83UOO8Tf#!Kls%D*VJ3W<F!-Dupf!{N~>b;{G^IyebYAxt1q1$cYH={{gbj8
z$sVEqSTFs=o>Go7g9ZNeZI-HD@oJE=v8K|h(vLs`PM8g5+g^|L>Z?r+?fK=NE=}fH
z+h2`MQ%xYF^W17cp5bd3^++V4(MM$0!o<CGn;VjCZVWsZ+26wcY|OE_*VY-DGhfQe
zYy9s`j`}oq`*3<R$?YrKQc@p2pqOOgFl?Bnm5Niu#!wF}^%wdx8yTHAOa_ZpP%2$o
zEn0n88%fn`HeZ+@4;~S*vnPBWc`hUhdAH9+msw}G6(Pj|=SK^s_@HTW@Lb5wY~f)o
z@=-E84?s25ykortf8AR6L`s@rcil6DQfPWBn85Idy{k3tFsCcC7-=)h*u_kmb&}$z
zXg$DWw|}YVG3`9fl9`J3!M1Y3jgd{05!)4iAhnhHP<N2}*B0|?bwKV_ybyyRKn3xT
zyAsyi2*-MZYkniBc@0OFNHRBXo;2m&*VzJyUY;0Eo@&wu{q?iUJ)yq{VV6))cPD6V
zSM{O@H7oyeT(ws4z0js=_FD$!7W$qh0%>bE74eA)kYKS8$ARl%+|m$D5jwSppyep<
zXJ!8Y8(&VlPq1Y4cF)Q}Ur(t-B78LOftE9mqA0>|!is^imp}Y_#h;Uix{7gJTs*J}
zyw9I~nOB9p5B7u<9<owRRNAYkK<TK3be!`mAL`g`Rb;3L>bO_3Q3;kF8g!ZFeu~_{
z?K|Wmhml`3MCY@YqYBJpIh1HkGOLdE!wT*N+beni_cp!~4R6|t4vCxYkyl^Q2e|!h
z*v82VcEc*(y~Mn!NtZH~r+%gh9RXr_ouzhWhA=G4?``=${1lX@p6zjm_Z!orGrViG
zi4;i+k7tx<X5Z8diIWD+cp#YgkTu?j>#}m&RM~0FD18lm(UG1kz8`x%o~?u7YzJmx
z@(&D;&j<%~o$40^kY~8*+qrE_FWu)?fCoXs41C2B>d|4RD^=Iwzk}y$fx2K^1c6GK
zAFV$>?~&1x%-*1KWEpL-MzyMbr3Cmd?EF7ZnXAReAI`YpgBB6ZMOXPG_p$?rk=NZN
zB3p>yY@Ff?4qdGbsqf(su31Ki)J9|r4i_CBbACeDc$5%=1pAQY-uAwPvlv-LL;!x{
ztRpvCXOa4jXrfj0+`mnrb!vKsH_2vj;lyYwFGsK)XjF^(sN@y?nyqK4XZ->MloYI{
zK7k~z8526Dz!DxQpruGQTWGSHo2@-@h*Z>(Bg1;Kb63y*Y@tUmV8-)ZIt$-wcs1t2
zpZ!(<Ity*B(;fTPlgvR1k2yWs1*<YwU*iYGCBR(M#@+@2n29L()Z3i%8mCJ_Uf;m@
z6k7k-larKvj=E`XWL*0sp}r;ZzPuE#t1Z9Zqoi1oG%C(mG_PHo_03>nW6))KB}p@A
zK9x*q{6~ZfV`e@@>7R!;zffrR62PPQrr~fE?79p5yPq5O1GJALzy!G(%>nviQ>*YY
zKV+YcPwvS9`4Qf&v-kI=3)?vPFM2j8l5|l=1OGov@RJkLQ5Q>Yz1rGs!#g=S_~Xey
zw&~>$5u%uVxOb^d(+OcOAO<RJKOBhBjSQTs-|QW4YpiX;oStnMGU(K<9Rg>6Ib8J-
zTXzdlfvqK^6Nl}JL+k`Rg0hd@v!4Cicr-wlHTsJM;iT&HR!6BX+dSlx4FjZ_%PRW5
znu=_CIFHwD{(Vvs7j<$!F^N6}&H14)Fa!3}7^8S!P%3s<D3!k-833caE<h32+dncY
zZZK%(!Ve*Sw;G844jL()H;VJazN<H^0Zv~8Nkwh>Sd-qPbT_{12^=e6*JGO|19*zO
z6Q;CcBNMx6$wd1!*_%9?$J~AGJ??UB!R3~OGOyB%1l_y+M64&?i#WqLn4Fg!yAypy
zOWY+sSJp3f-m=Xq7T!G(DK)f~d^e*j>DYv#h=N9&IjDH4N9LAz_?;|DxVc3$3alwN
z=gtq;)nkC;{Y{%fRu@K@)i8%j;D5!!a5~HCp67Qn8^S=0ZC%}Rbc#`ePKK7g=%&+s
z@4CpgiHNyyU$F7PoHlnImDEalo6w<tjn0lLo3lzZ>gec55;#9F?L3caDbjZG!A$&{
zw&gXx*)@u#mrms0@8A_Iu&FzUaqPr7bhdXyV|@6$V+fPx2AHQXT}nV!*1`n&wVxHD
zuaa#~#`#2wVuHC#1wy#FfLPanjifT_!6?R-SLnP<YxC@tY*UuYv{L5O)IkUxzDJ6J
z8r$Es2D>wDeFz;GZ(TO`dm!xMzHHKMuG-hABn5;%)C}Ueew@zz^-1f1U%T`aE_ciZ
zUnI$=cKUDc2hd?x49whGO-9FDl>7QRn%$rdS&)8-@N0_QVoaQ-KUGS1Wfa9s1$fy1
z$?Kl(U+bk9Q>rgUQ*)$EHpO{++k1OH6pOs#-o^6<D!{xyS(-1%TrzAHQAmi!RcthQ
zDwq5)6*bYW40*gSP@Y{`WS)w%ymmEf!E>CA6>U4ytqKY?0q`zB_Gf$W*=*;rMBORM
z=r|@)ZPr@-&%MdkzYlE!OhipcEji|18%uB$;p|4AP0{dbi}!s+W)~8{<{l<Ex@{Xg
zk5MR6s^7%RtcaR3stO}b{Z)*cf;(rMa1Fe)+Gov=5&SCDdzHMODQIF(J>*LdX1(yU
zN!=;hN?mIJo&5BWxIRR~LeM3%huuUl;jvOVHa}RoUCp(XQinQTiiNgzo)?~XOG+;>
zPEr?k?NN{p6JxgqwbN`QVWSpT0i%@8(<Ak(7nJN#)u|d22R|^5OLV~<?Yyf#FH3Wq
zwo8T2`>)<j!pR4)m7{nk*6c;ov}0D)CJO{vEntAV!0q=`15>T<()BZ3ht<D7EmPP1
zINB@C9oP<srfs=hA+m-cplQoWtows(qjyAPhVKj;b`<>$Hwk3c{K>Ob5fjsLc7SiX
zI41jcs)PMj8u(LWQoYe}KrxsC$$i*Sl*1MuIoEU&(UueaJkA9wU;q`D4$!y6_@4ND
zS_QmCnTVRakf_u{x*Mm_Z}@JMI;Y$!KtS~pq`w6eT;-GCQv+oO>pe+Tx1DXaHzGRc
zw#fo<POmv=&5$Y7m1g6Du{D!8ySZ90I-ut0`vbFyA_ix;r)Ic7-{=||>HL8r%qA3l
zDzbgVB0Y)Y>HBN}P7pnzP?~!@tdmRGfQ-&{Eqb2#ULic&kKE&ZsX(E5&8GnzA2js0
z^Sh+R%7~a;oTV@$K1y@hd}?;`<CG2I=wGCwB8{xx7##C5CiUk(?Vbm72JlddyWsJ3
zGMHd^va`F8>v(=J<#-WXAgObDbX0cOnzOm`uUenMX-m-GXXhQ9(?W(?Q=@4G2rB;3
zP;}zm+N$jxPj;L3Dc`?}Cu_!VuRYn0ywzM&9lq^OwDZaHKI4bsV^><IOCQ6>va8_V
z&3?z=a6|YebCi4uS}Q{}@wq^+XF+9n)@b#A^a5Y=grNNU4rK6HGaDghwzj*Vs&EwS
z{STDTIZjG<s8sUm?dc5-veyHULz61uU~8!=aleo6u@1iAF6UE6SCk$O{;h&Rdm{Sn
zWXyB#ZRLPJ>Ya9xzE1z}N&geb1-QND3I7&g{Wthq-PBG99X0P2J|TA!|KJ-3$aW>~
zjPiKg-Rp9YZ6Dk1UKf<OCKmThsDH-pGrf%`wo&zwa)52WK08HMArQPV)X=(Ds?rv`
zX`;hQMfE}KNwrmg8|}5rjcmHKgLsu@<+mA^xx)A`zo%_Pg_8%7bmjc4%&pseJ6cBJ
zT-|B2uQDiFyxZzlV(hh{`aFBovEmf4wt@m=*IG;ZM!T21d|<#fvwpv{u*MZ(UJ%OQ
zt|8@PMShb;t-(t18i?x(+vb2FTH+7Wwh#y;mzEFH5Qi$i+mmksf>IeQDJ>iQ#mqVu
z%X@5mCda44s!pWogN=<%;QVZ$^P<kQGdet+U?YnbU^JTbES9KR*<A<%Ljc*yPPW^1
zKjs=wCL&&_npzaCst)~9nt5sSL8e@#&`Z#rE3NDeg-n(v(5D@}zceQ%kmS6KHv8Y5
zP0m2L1F0x*P?R=v7gmw&CuNjpo`M7kE@)@S;>_sh34?c2tGT9<b&gvbwj0=&<}-ZY
zHm{ZHFY@O0;%5PvY@H<;Q%jDv2ifMIGdbfRG*SpO6+#ZeC%p(^R<NrR=|=X6R;l)d
zCs!pMgWo@`Eco5{E$Y<%9<OVxl|@+Znul}bNk)Mplia?(lH~P}$QFBw&)6D24#g$G
zDheO_ZE=cYKr#31qBG{f?K-H5h^8HfHY0~gZdFoMdexpQ`ZB*k2B-+Q@OhiH91`i%
zI%VUU>R)W0fnLK&hix9j;M6)hD)FL2my*T-oJzb<K@Rm=LWA?o3T!QNqJ~h)<30@i
zJTdU{VEdo5m1CRl(QQUxGGd?2LSG<pK?cE*AhDmxqqG9Lt*PcQGsTXb5b9eC;cyLF
z4(AZ?d;2ioQbphk2Qhl?dxWF682RLkuqO+5lz2qy;(^TB0DI?YQxQ{2JZHxOzky)C
zQ0|c2OzZeN?6WBSk@k31V?Gkn)N@e5+s7%S(|!}?+FR6Z$|s5L*g7c+v<Qwx%|<#J
zo}3+Tc3zxZAnMu94`g)V`4%&U_M{-|zXu%{TjIzB6Ju1bt?Zb}rD@1N6;J5Xk8XFj
zu?9R*Y9Gi(jRGRP$^#!Vo3)_`pxdbYRjUe80GjZ5lG3R%PE60wSBpa*DgsT`U@Zu2
zFZ!EBGc)agYC0+EQB8U{EIm@s*TEg|E2)wl21`0Vr2FEozH`Gkxkq)k^*-HiB6+MW
zlalPE(G9m8(rCn-9}=p3#IS{wjqY3egODUT=f`fb5(o`uH=$Uf@hvIfYS|X9N3#K0
ziat!=F}!|_-wS6um3i?<DXi9hkj9BJ4|aJ`^O-`hg_i4Qk-jSzLB)JI;0<oJ3-m+C
z*<s_^if_rq-=-hndq6A@@ONS8M)qg-s1B$OWD8&S-#)RKBpk#<<7dsYIkXA_)NW>v
zP0tf1&(B+dHtdPPJSRX~bv3msF5|)cS~&pVo^kNR#%5;qugHM$SM``?>V=OON42vw
z)ShY^5|IcqH<G4y8Ag_vfEl!W!AwYLeq#ri$=(EMNF{Ym_bs1G)40e{QElkW;Utp6
zY3=Q?9x7Qe8+YpqKTt2ma?EhXc2WHC7-#H(Hp`SrRk4`>1MQ%UJG)Z2SOPD1J+!S8
z_cC}upxQMl#$J(?ew63KOPfaA9~#kONl=IjD(uv&MJXqv<!_?8;xaSP?1gOtu}@mW
zy7^j0`(TL+x<fo7q;+Ry;PiMZsgFh4R0j@!b-cTNou5TC>iD!|Y9sKSSXojR@;2v9
z+RbvJWh8*%p_F-JJu!*n^eUh#yG%)$uQ_EloEMK+`w)kSP4^Zr^!BEmg-KnP-kg3*
zg=n_EuHjBg?d}eD%m15gT6NMoG3iYN+w-i#IoW&Qrv*uN)yKCF?wNbj(7`>ixOLlK
zO_R2wCylef`SBD!fFgTuU3M3k-PSE5XwvimWnJB&oScX!>AbUTAsYtN-7a6r?Dq=u
zO_8hOJn=@i<kBr&dA_}Ph@0D^&RWdY9#(yrzxuPGO?%z74HOj|r_ZUGnqB>I>Edrg
zd4`K!RxxS6-e~HfEx9$<o%B1c7KQHB$>SLxg)?YK<DGpU`OAr6$!d(hq?GKoZ;B#-
zznSlxtPE)sziO~$!t!X%7h`z^nC_11+j50Z>4?xt(&&bTjU<dRRa@6oTq>YP6daRh
z8hS*OpBT9a@#SW#Glnd6yp%9veoKoz!0m6~vPr}itiPi!-Ap0I7HHe5u4HBy`Pcu0
zXO~S!NW+85EfA8BCTFPgNK3u>LYWPqK=)<zbm<&%XpM;6Z>yom;*2e5i{omciA{iw
zKJ>_9*-uHgs}cemN)JtPh3zZlR$=k{VEXWkQ004ZPN)K_znB&u-aK@?U)smzAcrxT
zz9M0$fQBz72REUWkToW^qBK-=?NYenc_Vgi?srvg1fd}$=<+AjByXxTKL9~cMOV?C
zkpI#_Ag_|Mz}A1~`QP0s$Kv8*i3HSiNy&Yk2PC`)d#k2_=!!UEK*n^B#{zMMPmu#d
zJ~-%ziEN7@9&(=po6+kpc>6LP9W3pM+XbAl)S@ZVOpzBC>R->7nQSM$bPamt`&3#t
z`i+3MF;sWQYX4)F?RGZ;Xc4hpgBw$~Eki6x+ZrX6qEcR>H&<&EW9xLKbre0Oy)Z2~
z_}zerg4`ArttV81mje7&C4ADIm>Fg_XdzAsHBW}*p5sw7yb6Ru&uM1QRL!>%m*QR&
z^0S7?gFR@07Ryf+*}}Qz2NT4Nx{mX$LdA=E;+EmJ`T9YSJ?jdH-bB+d3#14(b2?8|
zUh9#>YaFqTGc7uI%4!9RU@68Lh|%OiAZl;ZHzcVL>&*a#E;T>=?8!xLBle~Hs>htJ
z9|ifk+XvA+kE2gcltL#cr7?MaL*}-Qx2%0Me&gpvi?GI^Tce+gKG3Dr+0tTnLhH=V
z@HCX&P0dPp3ImL68{Mb#!b^4Z#bwINnR#?{TK~$!i%nPrQkR+ujWd^e=S2jbFNI7$
z&T03sfak9wSEJNkHRBtJzP_lAldaIUv(1H_c;&PGL)Xz?cF!aH*Yh>7SFDMnL*wm@
z)fm<rAUIFrnZy^bPubPIpQRF%0{)i8QQaB#x|>6jNdXA3UWznm{?%)uTHnHNj!~LA
zJv>*c2y2@&`5&cPKtJW9v2*90q|GqvzeV1%V$&~?<4nm?{Apr_d<BJE8jd?KehVM=
zjJK|<>iDAHlqoYb5`|WZlpA^~V?))EEBQ1$SWkzPnvghAOT?eTe-gfJ1$SD)C0!Dm
z1L6%me*+qbB`A*2Jt+OfQ{R=#TO#<I;Ta7UB@w6#2^BN`0isT__k{1uTfS#N{wK>K
zpdhwq$Hakdt3vrWjE_8cm6?a6L^MjwPV*Sy1j?qX_b4Ga6`JOhxFDeyf-vSmN+@yf
z=U~?i+=hylMoQr_*rKJ^nv!dqrYvoP(I)BxQcTV&UpF}vM9X1$&H0@`*gue0_*YWB
z1Gtcr91@~+Pw6fM?Q8y>rUakVIEvDjIja>DH&Q*DUcg+8RGGOFU-Tszo~)Oh9|_Fy
z1^S@-qwU6XGDq_JKNKcVWk%AKg`R2ADe^l#kz=mgLdB5Jd_gLtVp0z&EULZp!Vl<B
zxA9)xm1A^;eweu8!qW%Jgxcm<dir@N`UfW(8jr(TJHZL8qod*XCWZc7mj&%WW#=UJ
zRfOcL3=X)ZPlp}pq18lEI*Vz#q3C#ZXtJgoMY$MP1C{8wa>KVwPNW~BGY{Tf<$0n1
zOge$EiBg2OPApEl<SM1ZGmM`<1h6B)3A7_TJS&G2igUn)>0Hc7x&X8A4@8tIQF0)>
zxa0xlQ<%1fg%Im&4qaj(W=ELtQ_^rhG_<J&TXBzib_=z=H5;`@IsuJIJyMh-!Es}j
z_wjO+@UI+69DhF+flr1(DbRm@If9Ob4oa>*dj@$nxj3os#6dHpo6!sK@Nk+|J<IVL
z8jOn;zRs_dK}Dq;?^1)YeE%!gkQZ?|J}!~JcU;$*x2$Ezuy%fkNUZYYak*i=7odDq
z{VK{KNeN1$sO3WMU(`sjvAy{}PoGKOQdCPSzJZ8xAa=AOXHZRTXE7JQYXZnyOAWv5
zK3a+4VCx&+z7>P4u|U(bBEH%N8~b1Iip}`QXGzXL@4ydxHJbuv`$EZzstJA)58q)M
z5NNbisRdVJkLF3C<wQ=kech+GIGp<V9$(B)6H!|5w*UH|Xuo@g4}(Rvkdkb(W}BB+
zyZ~G}%<qzuOX=Ed17~yG4Y^?c7bZ87#1h8e7aS^b%7(6etO4SzLoKP+4ySsJoYsOK
zet~rTo-{;{U3#-TQQ##TIJ4Qvif^D22-@G9d0nPc_Zz`ONkj7INqC#@=r!QQwDX_M
zXxp)=I+6T@9hM&$Xq8bx3KzwWS29tsvl$di9)G--xu!AnUM{c3#`-QLJV=W_PoOzc
z-p8_rdMh!~@;*Sv5{WWwB`Y+?9i+W?To4U<X|5pzxgM&({aJ74=gl|D_P=n2l+3B|
z^X~I!3Y&^OD6S9-15c@P^-kdE3>?JH<(pG43Y6ElqLpeSYQWVU!c9JDm@|%xzk*ko
zl8Lu;H}%YN4lsH2=2+{$iIOcz-|ckpI=khbBkom!wLIXh6Q^W1LEuE!Nfp&z@|cgM
z9aKtN`rlH$NuImdI`Qpei}$NpiBfHxzPfk0+p~@Elp|4~)wk_^7@9uAa;VJeW}up5
zPSwE5u1txtk)EY1mlmTxdTrKg{d*gB$X!C<3AbQ=zWm!Sh6?i3{6ZZDst5A>HpC!V
zKc{kV{%{k;qv%|esARBf=W`dox6)8#y4sf%k`JO23maqE#-#K!q%E;_gxz%=A;nO~
z&5)+flRh9~Z<9<oyD7lZ!}0obJxM*|w1rDVgNn|(Dx7vg0IC(>ce3J^LZR_wd2$yL
z5V`l8;M>vG5`L?aDbUX`j)yt6Ux(sX-#AkR07g_lkOI`y6JQzb4HHfuFA1i4rWUTt
z{D59f&|+LmT+`@v<{m4Z<+=Iarrr}G6!uHy4J>(vKHJ!g`+;^vhXV%DF5{42?bTst
z;LRf_%R*HlSLv+dXfsf?bhdKJ(Zf$AsJS{>DMwP3+^Q}JECzK`GHHc!siC2cK21pg
z#|xF}0;)s~>jZ1(m(qV)o;y{H9^ZKMYlz+kTZ>Qo@@kMi<CF6D8P)b_X4}t9Nlcen
zu6(v>Yba>bEy=u1-?r(IINd4}Y&P-r@%-w{`RT%LIuO&ziO*hSd&sjGlG^6~Gp#2i
zwL(Eh?Ds^(4~<BI&Bz<)pu1I-zth~qbb&Nvb97q5MjqD_hxm2o$}@)O6~~Y-j18n|
zc+3bB2};2>GGx}qD-glokBVhNCOLCu6qebLov1wO*={*9P*%hV(veEtSB2Giei*a2
zDv~8(Tjx7r4)Te4N<w`Z+*iI-H_O#o4jEO)TMQXLxBE;{-HLptDSO!gIfmh+1fXW#
zR^Bz30xG&}eec`pPxFSlW_~6oCx5e3VQTsG{yrl@_0o5$mQS`CEKo)<yP-jPE7UlL
zAD5f@)56W5Plc#Ze24+!uhJB7-ow&)fhRPjuj=xmK^JD4^q<OxInZUu(-)VF_3)ms
z4HQGbT+5}{SWIP$*iDAgCV+oH-ge2~-S#{MYC)O*g^RkVul){@2XyTeSE9}4&T=6-
zpn~G7oC#jZM*%;=)cB(hzqGW}Zs+CGtGdAEst+Rj3I5ai9FwW3R)m235{cD5n@0%!
zp84A7zg9kpT{@-9^S^6KUIu5EMAW!HfA3#2`$?76-Z#9Ya|qiih5$aIrA!MOR8l>c
zGbm|n<o+ANvRh1<D{r#%qB%3PQrX_Ht`~7Se@ZU&1e(JS{!KLb&PFv6=DsR{8)$tO
zSW|76%&xX7naOH@s${xHCD1xyGrW2^>dr3dS>XrdYLRz_e`j0x+3%r@Mx)(Uy$#u`
zOwd;pPe+sL7j8wL3<17|!|CZP*y_09a7~Eir^$ryX`>_pfy}r46q@hKrYK(%=kbZ~
zm|VYniFL#-M|Wtb5aIal<A^0?n~rck{DG8ZlY|d&bA#b#OVRLu^v*w2eH5?_9Eq^k
zOtaskkZP(kots<AW?FQNmCo%6dASwi1yjt=FW)gJJUb65Xaw{F@^jP4ET(-Hiu%G=
z6d{OB1(@H%u#?u2kCX4^@ssUHtEAr&TZ<fG{Yh1o4mn0`am?^7Rv)Avnl-0A+nhl>
zIFFV&ixo?<Uo@Do{Zu&#o0!JiT^!}I6iRP4;>nJ4%%Z4``Z>lVpH^{Ki|eRB%lGzw
zXJH?xC=o~Vz=-?}#whlsjh9f1`5y)~TV4#%KhbSZ?#q9gHK3Lc4fS%s9o&5I-vR=$
z=U!X;yLy?{sJPXib+E-gz?*sRVVq^Xhel6JUdfeEzw|tWrC5a;Z|+Ioi`FE8<6KMe
zi#H39z4zF?3mze^fcFf)OZk8MNZ2nn$Ur&x?|9F?cLjXW8FB5;cn`YoCslii8RM+9
z1Kj*>Q)y{2Yd)DA8!Q^JPT-h;w;F}cadPER9578X{0X3hC8%AWBsE9Yb)}jqP;FO+
zeKJ&R+eIl7Hx*n{4CBcxRTux}&Ufm3`GzM_6RasM+-B0&>j^V6#)U+-gn=<RMc?P8
z(%gF)UjgH<3o&JUR|j08%v3Eq)v@^ZhVu1^cj>t`g#Bpss@OFjk%*naKFIg>Llb@p
zH2H<r?=bSNiy%m$t>+<V4Eyv8nOk58-Za3pc*c|}<rpr(1>Us0lqpFg@2z~pJQo$P
z77hOfs?i=-G<(9<YyZqTzo3<W6#Ab1@B3+}9bm*Bwwa}oyS!fN;bSHC<kJYOTPk5F
z=I2$~<gU0P4M_}mQ^5=sP1sy;CyjKPJ-O@T79uunB-9bUM_kzXO*%ZNY4aVczmx@`
ziD0HV|5xa{8u8tyc-0jsH9~7>_3GRFUEe)__VqMZI6~HA2!2G`@oK_xE~5EI`u?UQ
z0=Qw8E>w2y@9FuCx|kP%G2}xZZ}54GRy~`{cB%#{UfOk~6t;oyf3H&7dQeos>ZPQ$
zw>(FFvK4F+tABTUEF9Y3fh8ArSyiB5Wwr8;m7s#A6KDM%et+;CprUhzn?1aysNyKG
zJ&P;MZO>OoI%A?}S^u~UoaN3%zUczuc~G#-JBu0S1p&hY9~&y|-R!j1F8CHjoFP~{
zn^5LZIz)(uMu0|-gdXvgxpz&)#Csc~ns*1f0r!RLXNv_Nu>D~7{9N%{H){AO;&n8K
z8;7N&7k08m^&#7WfPThM$e)x%B8PzYl=5T~OV$2Tn=`b{znNIEDMvDwY#9VV_lEiB
z3d*<IdySMn8D`Ko+QC&=o83=q&Ca3MDcxx`1VWO6KVOQCO<dDw9${;ESL2E2H*<T7
zL3&Ir$Od{3+4ANo33oZo+hE7TXZ}E1-pX5A$!4#4^y~9m5B*h@r0(w)@}K4jz)Zx5
zifKZ!{ohoCNnpmw+gqlq71aRC5?i8&dv3*wImO)(`Vqo}=xplSZDS$?6B*ajdL@7#
zqUR>-@%wIJ_h7fah_)XD!vH0=Pu0Ic!Qw{zZZTp~SRp7V!vi9>#?gMgRhEbVL$HJH
z&vGdyX|!F=q5AphZs#eP?>qpoX3Nk;$M0}F%v}Vnhy*26B8T<>uz6N8$D$<~pwH$q
z@0?H{exjrP>dz7*yA=1(?=R(?jqdgljggO)yIQ*zroBdO;52cNFrRJm9<{a}q81x{
zw6hzetSAqTsdyTN^E3^PGTPv+Z0ii3!jha*rIV?zY3-<%q=ACftFtcy=qZw%J*<)v
z#EsU{K_dl8fOZS85odnIGFmNwR+JOA2)J$o->dnTQW&|KU5HSYWsZNWxP($ig?dh>
zPzNMT-)n_=W7x{Yg{Qs`cD+XHH)K~Dsua7I`zg(wZG~QVenOTv7<vxKb+rc3ga*YD
z&q;-fbJ)iANOEqt8f8Pz0t`k)*=8z=>u5Qv$os&YVF+L>^YMyvQBC=WsHvO6OI6}?
zeVp&>Uleb+@_+)gxu}z~o~1UMlxP`54=JShQB2g^ibVcF9W-k4-W^7}WY15A=vl91
zcw5%<{4$L0&Z8B^U`gZEnZ|m~ME@ckgqemLqd?a*T~H|KN7`%mu++PXA8!N4uaa1L
zLS#1+0ho0{o^US=@d-rRw^B)mDp;nz(05DgmrAoRut-Wna_|elFm(cSpSF@G3~Gg&
ztRk#JAh!b^Q7Jsqpk=#_>Uvg&d{;;b$&CM_{iL)!5i@Bwo!Z<V^pjb8|5f3?YW%0=
z84U_jHMkV6_mO=acwf&DA1{XRge=u_E``4X3_sWz|9ls09?{nr9WlbZVM9xgYh&k>
zl7cjcJ~N5ZOAPAcyz&QfLz_C-(qpK>bKv>cpSVYDlFhvyvjwd+m&%kCHjQL7diU(A
ztwn#$puhZL<AEwcBsK#iFoSVV^Y2i@?&KlQAMczwK7X%v|JJIq<jSs+`YVYF&uWG8
zUSQC0bebobk#R_cvyW(d8-MwdmL-1Dz;dC#gs}=1JIFhcT|brZRQCzLhIWws!!_me
z18bpl_$zeA1AWSO`J8eIiRDL(%JwC1Fjfg9W@uN#*l2(cOWz(L-86S>ysRcmZ^4)g
z#X{&Aa-Xh%X<7e{Qk7f{oO#Yl*#~v*2l8@wW^W!!_k3xlu&kop8i&ifpvIX*BA!sw
zZIhUh*{MXytI+TRVTTnLNZUByhh71u9ub*lO`fmWdi3y2+yADdv==#&1g_C%=4FmF
z;LZkE)X(<&s;Zha`4c|owy#4VOzU@aH1Bz!atCxVE#AW|lhX|}|G{0{Aety%V6U11
zYva&>HHp<NwG`@H`fkTN!y!POOy$$GbY4&7t5kx0xU(EXaAh=8lDbV_(Ksd-blpwf
zJ7pT4)G$YkIr~RwBh<OfpoJ<ac3bT_0_YR-n3D+89aK#$e5XO}d|#Em!k~NW;zk6|
ze_w=PGr7Xvd!9U<$E3FGO7nYwPVy;((Eh64j%C{W`{Je93XFxoQPPFD`rVu1*25md
zI!k#G#Y(SCDw!dg1qOjlmSc>m(665T12m=O32XgOVeQ<Zuwe_{8?Q+f`GS2)D1|qg
zf#xn?yE|2yxuOBN3M$~UY1V|w$j&_P8&`JMd)h+5it*1Oo%<Df@i;(#<^$EKP=Y{(
zGl<^n%~X;aT_A1w1*(*frU&{G4qnUKgx;3(#!Q$SN<8$$yv5E^2Rp?=u0m*}Kk8&b
z22*CBkii0HitE=*sJL&lkOs`Z+x`9a5Dhl0{#?^SxJj*be?y_5RW0X7=ewr9+uMgr
zL<uI%M1IW&{~Gu6{eYJ5`;Z`Nu~_A`@2QlmvD296o<sj)j4h^kZ$gxm{5i}iUk~3h
zqp}<!Rd1Q}YQ|1HYsrL(C_{qrjD2=Yc5RI->mR!otMeZ$YK&_TAushW*%X`RF7<>I
zfy@hKqooBB$xk-=v($N(-%j5^K!=^l>OlK=i$~yN+CB5?2moC+Nn-AHyYE*n6*Mdh
zfnb)cWW|KFh1{*zN6#)jwwKAOi@xK$mGe$M_nhQPnrw~Ct^;S#(L(jqhh7G6(TN_*
ziD#kEk%{)|6&ot;AA!-fqxU&~{IjdS(Wv3Xl@#_9?9-9`Q?TRsCH(OM+(>&qkM`=<
zuj`N37pVDUfMO=@GRn}=Xm@xTLR$<vJHyfL{qII8)+#yRoorUZL?Q(4O!GF5>WK~2
z>of450Uh(iXF7^30$;%O9Uh3Eyz``&M=M1-SqBla_psx1gYROMP;%~}&>0czx76Oi
z!l&L~1N+=QVU;XJO&@piTI5EvOo4>`4=>SI#F!z#Hf<8w2k}B*p)`46RM_14Yj$R<
z%15pa%cgPV3S7F0!<7-CX;4Y3txnKSo6wMw>GPk7KAUKoYd_x13(<i~Ydq4>p7Ls}
zT}>1c^>H9ZS75VM)o;a6L*Sg0l2wBI2^{CorBBF9hF1H`x%l>c6w243(4{8vE&A)?
z%q3vu{(8%!Xy+XXfp6|#*~o2#(+Fvw*qi}rMPVU#Ph;CrehOY2Og=2m+@M>0P^(vq
zFw`5H^vYp)GwS`S>xRbDNyBbt-1rE%@8Lb&6U}q;!Z@^3s4^i>-Z@C*n%_`<T*ZWF
zG9T^}gjBamO)9Ia;V!}ebWc5D%bGHfAIK`claEZuFmH|cg;se-$^Swp+bOoytGw33
zVEfOA4R{XCnKbak1P>{@-KK!j!lhxiGc|=^mtdS_loUVj@TO3WzDh0BHw$wuH8v$z
z?ctMk4nyfEKavJophMmgYa4=}_{CodZ_f^Tf3owEz+_ulDUM--W>%KCoOH#v0J8@=
zT{B*^x};e9go()A;=H?sluoIduNFCkpSHVs{~p5D2{ECmV1(;K*MAI9678>K!XfpI
zlLFsp=V^!JSHlbRvIWYDR!@UeUl2mL61xO8wYWy(nxFJUNuc^1tM#^b&Udp7Tf`3n
ztYLXe{Qaa^CVMYT0*98+wNi;3pc;cU_oDmFA1vdxnh@Wj@RY<b-%GvLD`dn_Tm3Tw
zIuvZ_M5E{^;BpqF98HloCEb|gpQp8N;_0_J9;GF<2hOs;p6Kr|F1su`9wl2@ha&r;
zLP0ScZg^sdu_b=H?dsUXLV2qvBTrJl_LB^EV;1@LAvtcj-&|3=fHS!Nus+C|+^S@P
zj`^RXNpIikdN~i<e>k-s{M2)${h8*+)Iwtpy_~fX8c!Dk3$8v`_e3^3R98qJtw||c
zVv??YepJ`DME8k@Qrf~h_}f5~+LTqEXVDm{js3M8(|o1zgntt+nH%+G?mXxs=zWXV
zr->DiR|H3EPz;Vd|MdhI$?*I19Dgk_Zrp1aEK>OLTv8F{wlq!Q9jdo(ISj9wZ=e7t
z*@nrF%w}U!;A*evh&vxbMb@xu#ID@v4SDKP2k_);r4$Y|7<&rx?3vRKn(JJ<{azRt
z)j!udFkndPA#U7vl%;uaAJ6O#N$elNir1z{Kxapn6P7_y55Q0GoZFg(be}Dm#&pz8
zKS;Qf_t1!5rRWVOgf<z$Q{J>~_aKquB^L*4M*ZL?32Piz5X~A^k=Z!<oA9u>7Ar>_
z{TJS$26+qyv8V)6a)xX$tLBmPChYZcRMULhTU!nkTRJ*g*)Z$Z`cO^RRBbMEu0<*8
zWPJ*V4(;MRfV=KIvObEFnQQPGu*n5~0^e1PSCZ0F`5}}dv8h7hnp#e9p#U9Q!^OKZ
zlE3Yudt@@DJ<)Ob@8#QC3*T9LN4(VggP+TDN<E#}bneWbi5B${4gPizR%{wrY&f7>
zh~b$^D>A$O@Q)WwtiDnA1Lzk|^oN#rj?yt(kTnz#?a3+Z$M(Rwfap_c%GKxnp_^j-
zJ(jY|Oy|M!|J8OaOi3mRRAMXKHY3c}sw?=K^|ATZw9-V)M{7oe6e75;V#t+zB&cOy
zwuGeV-g49ojn)iT4X~@^4NNQ3vQ*qi35-fp$h2BA#TRhdA92r|Gjo2znK?7(C{&`5
zcfq3HZiq1b_=eno?xETP!*bz_lGG84{?-ST;tJ2gvDenq_YWF0+1hvY6e&K7?8<<=
zyyf9q5|G6M!PTCyD^MugSa1g8r-xz?ELu5gL-Yx*ih1$j6rYs>P%|_4y;fIVWp9^0
z<>eBiMt(^SG(|mz&|0n&7FtRDdy#Jn^Ai~@g*jDaH;<8PC`oZ1!hR4PT~sU<%enY0
zDl-xz1Y4%U;2EylNtRz|%2kw~C$l^i><u?`a>0&`W@mXt0qqvvenFi(^{DfH9q`*u
zInmy*#ltnhfOkjCcR4$3pc_%F`rz*(;>dQ)5;4R^P!;9e&JypncQ~`I*>)fl^ylvB
z)3^I;E<f1Y4*j*xiseHc@ORDnP-Af`aX}?48u_U<ksTN%{xAtP3BGCzifPJmA~zxa
z+STvRgvxpRYWuUHWAVd3m!BHBOLjt75@rlPQn@-M@DVE@+_RBdh97At(v(3s2wT|_
zDS^YPXBzWb{JX$fbjwpbRyXHmH%Lq~t8|Mm)qVUp>@jIFdN-;eleD*&n2;jmCz&2*
zMW(f8S|Sx6sTX4J%_Q!8HgwDP$*zm+99+pE7x2lj8}~gKX}nX-C1T%*I6G&k1zj*K
zC<MqBKOEAIcV?iGBp8ymK=29FhEI{|k37JQb8k*7M57xV_vdwcHsh>JGA_Xp41P20
zh5sW0s7z;HxdU=6E#=!!$2W~z+y{wnCrTTy3TG5?JYgpeBlCr9eq{2b{YwnP1)hW5
zO^ck{lGm=Pv0zKxqxZMnwDB?jqgCYsNEUq6&uuX(GcK&4_F<Rc#5Qq-tk5leB!6V4
zKveIQPrcT54Er3hArgE9KWV@!8t6dgzVDe2zBE#0SGOX8_YhD2i>8n{_o(b@DwzIx
z9QATCAwH>PqtCgMyl@jVxa-=GD8roP<XB3|t}|wj^tz-$ejH;HpW3$z0c^Q!jTFdq
zhlh@y#q(M)`T6sq&B8#^0A|m#SErZKt?`l2Kk5jmAd?uRHS&0-YCOna7joVW+H$Uh
zELFvQdP^7{1#NfnKDl)#221?C^!~$dUfQx-hp&jtB2nhDvL7JVJn|@|kSBI8x&!BE
zH&mgIvO6a10;M`PVjOA#P;_qleOG&=!v4-LtN<2$42|*bXscPbaFY@zDO6|2R3ySb
z*JU&4h#%8<@79DNSKo#E{pfhfr-~jwYAl#t90vpbVz-}sd)kg^{p*|Q0}xLo=lqyr
z_UrPeKH{rfY(W_Dp#5$)PXwe9SpK_*Y<n0w>+#|Pvvy$*htdXQhe+;ra`~HR0b7up
zH{@bnW55neWb5s*=*!ko3}JvP9@|JEb|pQ-p{wD>f33HPO8e4(Q_biCVz2!3XBIN6
zj4>q}d-Z*>Tt%rJQVvDfadZ4CCdMthaAVl{J}YFIrhKVesCvy<yQN(`%>^cG{@c^E
zF71^WmOT@J_K`BvGy3FS_sx48K&!atVfGtM^!i!BSa~(?>&k`V)aSs`B*%VQk>Tuo
zZqXw#e0c7X#yxF$_WHd!;lalo<!_iSm|icR%AlKZY;6HuRsQ?rH*1pEP9g!A2*DQy
z>gCag7~=m#dkxYYj%s_8K?358Bxqd@ohq6P1H@}#;t7d^p)I=jClF9Mu)0#Eo5$*v
z?;rP5j(s?xIOJAg5J#*#46Nj?$XVT$1HGcggp9ATc~*R_D$aLV%U=)xtFOXV<^>Y!
zqB_D<MqK_EPCp;Kn0LJ==4hQ^)arPfT;1$-?=|5YJp2AA9ag+=S?IDPBQETbg&Fq`
zfm5y$a~=BTl)9^F%XxEc_}c6h%Cp|SzR>cX)?8xN9sTNDs5q&hp<8033R&Z*r9|MX
zI_U3^RLxW3(T*=_zeVX-SOO=Z_6}_zbf#kN`il6*lyvK22L81K*nwOPk}RL_nzVAK
z_B=Y!{SOUJnsnH4FAdB~bH*q`a<}tpZNUX2iG}|!yY6*`qFt)!sZg325gT<dl8;O|
F_dmUcqF?|3

diff --git a/vignettes/writing_a_custom_training_loop_in_tensorflow.Rmd b/vignettes/writing_a_custom_training_loop_in_tensorflow.Rmd
index 4627c323a..23d4570fa 100644
--- a/vignettes/writing_a_custom_training_loop_in_tensorflow.Rmd
+++ b/vignettes/writing_a_custom_training_loop_in_tensorflow.Rmd
@@ -156,21 +156,21 @@ for (epoch in seq_len(epochs)) {
 ## Seen so far: 12800 samples
 ## Training loss (for one batch) at step 400: 1.6450
 ## Seen so far: 25600 samples
-## Training loss (for one batch) at step 600: 0.8552
+## Training loss (for one batch) at step 600: 0.7421
 ## Seen so far: 38400 samples
 ## Start of epoch  2
-## Training loss (for one batch) at step 200: 0.6606
+## Training loss (for one batch) at step 200: 0.7254
 ## Seen so far: 12800 samples
-## Training loss (for one batch) at step 400: 0.2213
+## Training loss (for one batch) at step 400: 0.2476
 ## Seen so far: 25600 samples
-## Training loss (for one batch) at step 600: 0.3709
+## Training loss (for one batch) at step 600: 0.3578
 ## Seen so far: 38400 samples
 ## Start of epoch  3
-## Training loss (for one batch) at step 200: 0.2968
+## Training loss (for one batch) at step 200: 0.2974
 ## Seen so far: 12800 samples
-## Training loss (for one batch) at step 400: 0.4781
+## Training loss (for one batch) at step 400: 0.4441
 ## Seen so far: 25600 samples
-## Training loss (for one batch) at step 600: 0.2455
+## Training loss (for one batch) at step 600: 0.1461
 ## Seen so far: 38400 samples
 ```
 
@@ -281,7 +281,7 @@ Sys.time() - time
 ```
 
 ```
-## Time difference of 51.62403 secs
+## Time difference of 39.32149 secs
 ```
 
 ## Speeding-up your training step with `tf_function()`
@@ -380,7 +380,7 @@ Sys.time() - time
 ```
 
 ```
-## Time difference of 0.5661066 secs
+## Time difference of 0.4121809 secs
 ```
 
 Much faster, isn't it?

From 39d3f56f0b8452588b23ea3ff2fe00b02b8ce59d Mon Sep 17 00:00:00 2001
From: Tomasz Kalinowski <kalinowskit@gmail.com>
Date: Sat, 27 Sep 2025 12:07:38 -0400
Subject: [PATCH 77/81] rebuild site

---
 docs/dev/LICENSE-text.html                    |   4 +-
 .../custom_train_step_in_tensorflow.html      |  16 +-
 .../distributed_training_with_tensorflow.html |  18 +-
 docs/dev/articles/distribution.html           |  24 +-
 docs/dev/articles/examples/index.html         |   4 +-
 ..._machine_translation_with_transformer.html |   6 +-
 .../nlp/text_classification_from_scratch.html |  22 +-
 .../imbalanced_classification.html            |  77 +--
 ...ata_classification_with_feature_space.html |  72 +-
 .../timeseries_anomaly_detection.html         | 189 +++--
 .../unnamed-chunk-10-1.png                    | Bin 17824 -> 19143 bytes
 .../unnamed-chunk-11-1.png                    | Bin 7110 -> 15315 bytes
 .../unnamed-chunk-12-1.png                    | Bin 31199 -> 33046 bytes
 .../unnamed-chunk-13-1.png                    | Bin 24995 -> 36351 bytes
 .../unnamed-chunk-13-2.png                    | Bin 5748 -> 13529 bytes
 .../unnamed-chunk-14-1.png                    | Bin 30523 -> 32775 bytes
 .../unnamed-chunk-4-1.png                     | Bin 33191 -> 42503 bytes
 .../unnamed-chunk-5-1.png                     | Bin 24266 -> 35821 bytes
 ...imeseries_classification_from_scratch.html | 648 +++++++++++-------
 .../unnamed-chunk-12-1.png                    | Bin 53506 -> 65538 bytes
 .../unnamed-chunk-13-1.png                    | Bin 53428 -> 58558 bytes
 .../unnamed-chunk-3-1.png                     | Bin 53086 -> 54970 bytes
 .../unnamed-chunk-9-1.png                     | Bin 225954 -> 321147 bytes
 .../articles/examples/vision/autoencoder.html | 302 ++++----
 .../vision/autoencoder/unnamed-chunk-5-1.png  | Bin 54370 -> 54089 bytes
 .../vision/autoencoder/unnamed-chunk-7-1.png  | Bin 103099 -> 102613 bytes
 .../examples/vision/mnist_convnet.html        |  38 +-
 .../examples/vision/mnist_siamese_graph.html  |   4 +-
 .../oxford_pets_image_segmentation.html       |   4 +-
 docs/dev/articles/functional_api.html         |  20 +-
 .../functional_api/unnamed-chunk-10-1.png     | Bin 17031 -> 20488 bytes
 .../functional_api/unnamed-chunk-11-1.png     | Bin 66248 -> 92394 bytes
 .../functional_api/unnamed-chunk-20-1.png     | Bin 121898 -> 197539 bytes
 .../functional_api/unnamed-chunk-25-1.png     | Bin 270993 -> 400451 bytes
 docs/dev/articles/getting_started.html        |  12 +-
 .../getting_started/unnamed-chunk-12-1.png    | Bin 32193 -> 32138 bytes
 .../getting_started/unnamed-chunk-9-1.png     | Bin 19900 -> 24497 bytes
 docs/dev/articles/index.html                  |   4 +-
 .../intro_to_keras_for_engineers.html         |  26 +-
 .../intro_to_keras_for_researchers.html       |   4 +-
 ...new_layers_and_models_via_subclassing.html |  10 +-
 docs/dev/articles/sequential_model.html       |   4 +-
 .../articles/serialization_and_saving.html    |   4 +-
 .../training_with_built_in_methods.html       |  38 +-
 .../unnamed-chunk-26-1.png                    | Bin 118726 -> 206413 bytes
 docs/dev/articles/transfer_learning.html      |  18 +-
 .../transfer_learning/unnamed-chunk-13-1.png  | Bin 373610 -> 375562 bytes
 .../transfer_learning/unnamed-chunk-17-1.png  | Bin 331799 -> 333527 bytes
 .../understanding_masking_and_padding.html    |   4 +-
 ..._a_custom_training_loop_in_tensorflow.html |  22 +-
 .../articles/writing_your_own_callbacks.html  |   4 +-
 docs/dev/authors.html                         |   8 +-
 docs/dev/index.html                           |   4 +-
 docs/dev/news/index.html                      |  25 +-
 docs/dev/pkgdown.yml                          |   4 +-
 docs/dev/reference/Callback.html              |   4 +-
 docs/dev/reference/Constraint.html            |   4 +-
 docs/dev/reference/Layer.html                 |  11 +-
 docs/dev/reference/LearningRateSchedule.html  |   4 +-
 docs/dev/reference/Loss.html                  |   6 +-
 docs/dev/reference/Metric.html                |   4 +-
 docs/dev/reference/Model.html                 |   4 +-
 docs/dev/reference/activation_celu.html       |   6 +-
 docs/dev/reference/activation_elu.html        |   8 +-
 .../dev/reference/activation_exponential.html |   6 +-
 docs/dev/reference/activation_gelu.html       |   6 +-
 docs/dev/reference/activation_glu.html        |   6 +-
 .../dev/reference/activation_hard_shrink.html |   6 +-
 .../reference/activation_hard_sigmoid.html    |   6 +-
 docs/dev/reference/activation_hard_silu.html  |   4 +-
 docs/dev/reference/activation_hard_tanh.html  |   6 +-
 docs/dev/reference/activation_leaky_relu.html |   6 +-
 docs/dev/reference/activation_linear.html     |   6 +-
 .../dev/reference/activation_log_sigmoid.html |   6 +-
 .../dev/reference/activation_log_softmax.html |   6 +-
 docs/dev/reference/activation_mish.html       |   6 +-
 docs/dev/reference/activation_relu.html       |   6 +-
 docs/dev/reference/activation_relu6.html      |   6 +-
 docs/dev/reference/activation_selu.html       |   6 +-
 docs/dev/reference/activation_sigmoid.html    |   6 +-
 docs/dev/reference/activation_silu.html       |   6 +-
 .../dev/reference/activation_soft_shrink.html |   6 +-
 docs/dev/reference/activation_softmax.html    |   6 +-
 docs/dev/reference/activation_softplus.html   |   6 +-
 docs/dev/reference/activation_softsign.html   |   6 +-
 .../dev/reference/activation_sparse_plus.html |   6 +-
 .../reference/activation_sparse_sigmoid.html  | 115 ++++
 docs/dev/reference/activation_sparsemax.html  |   6 +-
 docs/dev/reference/activation_squareplus.html |   6 +-
 docs/dev/reference/activation_tanh.html       |   6 +-
 .../dev/reference/activation_tanh_shrink.html |   6 +-
 docs/dev/reference/activation_threshold.html  |   6 +-
 docs/dev/reference/active_property.html       |   4 +-
 docs/dev/reference/adapt.html                 |   4 +-
 .../reference/application_convnext_base.html  |   4 +-
 .../reference/application_convnext_large.html |   4 +-
 .../reference/application_convnext_small.html |   4 +-
 .../reference/application_convnext_tiny.html  |   4 +-
 .../application_convnext_xlarge.html          |   4 +-
 .../reference/application_densenet121.html    |   4 +-
 .../reference/application_densenet169.html    |   4 +-
 .../reference/application_densenet201.html    |   4 +-
 .../application_efficientnet_b0.html          |   4 +-
 .../application_efficientnet_b1.html          |   4 +-
 .../application_efficientnet_b2.html          |   4 +-
 .../application_efficientnet_b3.html          |   4 +-
 .../application_efficientnet_b4.html          |   4 +-
 .../application_efficientnet_b5.html          |   4 +-
 .../application_efficientnet_b6.html          |   4 +-
 .../application_efficientnet_b7.html          |   4 +-
 .../application_efficientnet_v2b0.html        |   4 +-
 .../application_efficientnet_v2b1.html        |   4 +-
 .../application_efficientnet_v2b2.html        |   4 +-
 .../application_efficientnet_v2b3.html        |   4 +-
 .../application_efficientnet_v2l.html         |   4 +-
 .../application_efficientnet_v2m.html         |   4 +-
 .../application_efficientnet_v2s.html         |   4 +-
 .../application_inception_resnet_v2.html      |   4 +-
 .../reference/application_inception_v3.html   |   4 +-
 docs/dev/reference/application_mobilenet.html |   4 +-
 .../reference/application_mobilenet_v2.html   |   4 +-
 .../application_mobilenet_v3_large.html       |   4 +-
 .../application_mobilenet_v3_small.html       |   4 +-
 .../reference/application_nasnet_large.html   |   4 +-
 .../reference/application_nasnet_mobile.html  |   4 +-
 docs/dev/reference/application_resnet101.html |   4 +-
 .../reference/application_resnet101_v2.html   |   4 +-
 docs/dev/reference/application_resnet152.html |   4 +-
 .../reference/application_resnet152_v2.html   |   4 +-
 docs/dev/reference/application_resnet50.html  |   4 +-
 .../reference/application_resnet50_v2.html    |   4 +-
 docs/dev/reference/application_vgg16.html     |   4 +-
 docs/dev/reference/application_vgg19.html     |   4 +-
 docs/dev/reference/application_xception.html  |   4 +-
 .../audio_dataset_from_directory.html         |   4 +-
 docs/dev/reference/back-compat.html           |   4 +-
 docs/dev/reference/bidirectional.html         |   4 +-
 .../callback_backup_and_restore.html          |   4 +-
 docs/dev/reference/callback_csv_logger.html   |   4 +-
 .../reference/callback_early_stopping.html    |   4 +-
 docs/dev/reference/callback_lambda.html       |  18 +-
 .../callback_learning_rate_scheduler.html     |   4 +-
 .../reference/callback_model_checkpoint.html  |   9 +-
 .../callback_reduce_lr_on_plateau.html        |   4 +-
 .../reference/callback_remote_monitor.html    |   4 +-
 .../reference/callback_swap_ema_weights.html  |   4 +-
 docs/dev/reference/callback_tensorboard.html  |   4 +-
 .../reference/callback_terminate_on_nan.html  |   4 +-
 docs/dev/reference/clear_session.html         |   4 +-
 docs/dev/reference/clone_model.html           |   4 +-
 .../compile.keras.src.models.model.Model.html |   4 +-
 docs/dev/reference/config_backend.html        |   6 +-
 .../config_disable_flash_attention.html       |   6 +-
 .../config_disable_interactive_logging.html   |   6 +-
 .../config_disable_traceback_filtering.html   |   6 +-
 docs/dev/reference/config_dtype_policy.html   |   6 +-
 .../config_enable_flash_attention.html        |   6 +-
 .../config_enable_interactive_logging.html    |   6 +-
 .../config_enable_traceback_filtering.html    |   6 +-
 .../config_enable_unsafe_deserialization.html |   6 +-
 docs/dev/reference/config_epsilon.html        |   6 +-
 docs/dev/reference/config_floatx.html         |   6 +-
 .../reference/config_image_data_format.html   |   6 +-
 .../config_is_flash_attention_enabled.html    |   4 +-
 ...config_is_interactive_logging_enabled.html |   6 +-
 docs/dev/reference/config_is_nnx_enabled.html |  93 +++
 ...config_is_traceback_filtering_enabled.html |   6 +-
 docs/dev/reference/config_max_epochs.html     | 124 ++++
 .../reference/config_max_steps_per_epoch.html |   8 +
 docs/dev/reference/config_set_backend.html    |   6 +-
 .../reference/config_set_dtype_policy.html    |   6 +-
 docs/dev/reference/config_set_epsilon.html    |   6 +-
 docs/dev/reference/config_set_floatx.html     |   6 +-
 .../config_set_image_data_format.html         |   6 +-
 docs/dev/reference/config_set_max_epochs.html |   8 +
 .../config_set_max_steps_per_epoch.html       |   8 +
 docs/dev/reference/constraint_maxnorm.html    |   4 +-
 docs/dev/reference/constraint_minmaxnorm.html |   4 +-
 docs/dev/reference/constraint_nonneg.html     |   4 +-
 docs/dev/reference/constraint_unitnorm.html   |   4 +-
 docs/dev/reference/count_params.html          |   4 +-
 docs/dev/reference/custom_metric.html         |   4 +-
 .../dev/reference/dataset_boston_housing.html |   4 +-
 .../reference/dataset_california_housing.html |   4 +-
 docs/dev/reference/dataset_cifar10.html       |   4 +-
 docs/dev/reference/dataset_cifar100.html      |   4 +-
 docs/dev/reference/dataset_fashion_mnist.html |   4 +-
 docs/dev/reference/dataset_imdb.html          |   4 +-
 docs/dev/reference/dataset_mnist.html         |   4 +-
 docs/dev/reference/dataset_reuters.html       |   4 +-
 .../reference/deserialize_keras_object.html   |   4 +-
 ...evaluate.keras.src.models.model.Model.html |   4 +-
 ...vedmodel.keras.src.models.model.Model.html |   4 +-
 docs/dev/reference/figures/op-gelu-plot-1.png | Bin 6996 -> 16638 bytes
 .../figures/op-hard-sigmoid-plot-1.png        | Bin 7293 -> 17921 bytes
 .../figures/op-leaky-relu-plot-1.png          | Bin 7242 -> 17394 bytes
 docs/dev/reference/figures/op-relu-plot-1.png | Bin 8250 -> 18241 bytes
 .../dev/reference/figures/op-relu6-plot-1.png | Bin 8073 -> 17089 bytes
 .../reference/figures/op-softplus-plot-1.png  | Bin 9302 -> 21578 bytes
 .../reference/figures/op-softsign-plot-1.png  | Bin 9079 -> 20855 bytes
 .../fit.keras.src.models.model.Model.html     |   4 +-
 docs/dev/reference/freeze_weights.html        |  11 +-
 docs/dev/reference/generate.html              |   8 +
 docs/dev/reference/get_config.html            |   4 +-
 docs/dev/reference/get_custom_objects.html    |   4 +-
 docs/dev/reference/get_file.html              |   4 +-
 docs/dev/reference/get_layer.html             |   4 +-
 docs/dev/reference/get_registered_name.html   |   4 +-
 docs/dev/reference/get_registered_object.html |   4 +-
 docs/dev/reference/get_source_inputs.html     |   4 +-
 docs/dev/reference/get_state_tree.html        |   4 +-
 docs/dev/reference/get_weights.html           |   4 +-
 .../dev/reference/grapes-py_class-grapes.html |   4 +-
 .../reference/grapes-set-active-grapes.html   |   4 +-
 docs/dev/reference/image_array_save.html      |   4 +-
 .../image_dataset_from_directory.html         |   4 +-
 docs/dev/reference/image_from_array.html      |   4 +-
 docs/dev/reference/image_load.html            |   4 +-
 docs/dev/reference/image_smart_resize.html    |   4 +-
 docs/dev/reference/image_to_array.html        |   4 +-
 .../imagenet_decode_predictions.html          |   4 +-
 .../reference/imagenet_preprocess_input.html  |   4 +-
 docs/dev/reference/index.html                 | 136 +++-
 docs/dev/reference/initializer_constant.html  |   4 +-
 .../reference/initializer_glorot_normal.html  |   4 +-
 .../reference/initializer_glorot_uniform.html |   4 +-
 docs/dev/reference/initializer_he_normal.html |   4 +-
 .../dev/reference/initializer_he_uniform.html |   4 +-
 docs/dev/reference/initializer_identity.html  |   4 +-
 .../reference/initializer_lecun_normal.html   |   4 +-
 .../reference/initializer_lecun_uniform.html  |   4 +-
 docs/dev/reference/initializer_ones.html      |   4 +-
 .../dev/reference/initializer_orthogonal.html |   4 +-
 .../reference/initializer_random_normal.html  |   4 +-
 .../reference/initializer_random_uniform.html |   4 +-
 docs/dev/reference/initializer_stft.html      |   4 +-
 .../initializer_truncated_normal.html         |   4 +-
 .../initializer_variance_scaling.html         |   4 +-
 docs/dev/reference/initializer_zeros.html     |   4 +-
 docs/dev/reference/install_keras.html         |   4 +-
 docs/dev/reference/keras.html                 |   4 +-
 docs/dev/reference/keras3-package.html        |   4 +-
 docs/dev/reference/keras_input.html           |   4 +-
 docs/dev/reference/keras_model.html           |   4 +-
 .../dev/reference/keras_model_sequential.html |   4 +-
 docs/dev/reference/keras_variable.html        |  18 +-
 docs/dev/reference/layer_activation.html      |   6 +-
 docs/dev/reference/layer_activation_elu.html  |   6 +-
 .../layer_activation_leaky_relu.html          |   6 +-
 .../layer_activation_parametric_relu.html     |   6 +-
 docs/dev/reference/layer_activation_relu.html |   6 +-
 .../reference/layer_activation_softmax.html   |   6 +-
 .../layer_activity_regularization.html        |   6 +-
 docs/dev/reference/layer_add.html             |   6 +-
 .../reference/layer_additive_attention.html   |   6 +-
 docs/dev/reference/layer_alpha_dropout.html   |   6 +-
 docs/dev/reference/layer_attention.html       |   6 +-
 docs/dev/reference/layer_aug_mix.html         |  10 +-
 docs/dev/reference/layer_auto_contrast.html   |  10 +-
 docs/dev/reference/layer_average.html         |   6 +-
 .../reference/layer_average_pooling_1d.html   |   6 +-
 .../reference/layer_average_pooling_2d.html   |  12 +-
 .../reference/layer_average_pooling_3d.html   |   6 +-
 .../reference/layer_batch_normalization.html  |   6 +-
 docs/dev/reference/layer_bidirectional.html   |   6 +-
 .../reference/layer_category_encoding.html    |   8 +-
 docs/dev/reference/layer_center_crop.html     |  10 +-
 docs/dev/reference/layer_concatenate.html     |   6 +-
 docs/dev/reference/layer_conv_1d.html         |   6 +-
 .../reference/layer_conv_1d_transpose.html    |  17 +-
 docs/dev/reference/layer_conv_2d.html         |  12 +-
 .../reference/layer_conv_2d_transpose.html    |  12 +-
 docs/dev/reference/layer_conv_3d.html         |   8 +-
 .../reference/layer_conv_3d_transpose.html    |  12 +-
 docs/dev/reference/layer_conv_lstm_1d.html    |   6 +-
 docs/dev/reference/layer_conv_lstm_2d.html    |   6 +-
 docs/dev/reference/layer_conv_lstm_3d.html    |   6 +-
 docs/dev/reference/layer_cropping_1d.html     |   6 +-
 docs/dev/reference/layer_cropping_2d.html     |   6 +-
 docs/dev/reference/layer_cropping_3d.html     |   6 +-
 docs/dev/reference/layer_cut_mix.html         |  10 +-
 docs/dev/reference/layer_dense.html           |  14 +-
 .../reference/layer_depthwise_conv_1d.html    |   6 +-
 .../reference/layer_depthwise_conv_2d.html    |   6 +-
 docs/dev/reference/layer_discretization.html  |   8 +-
 docs/dev/reference/layer_dot.html             |   6 +-
 docs/dev/reference/layer_dropout.html         |   6 +-
 docs/dev/reference/layer_einsum_dense.html    |  13 +-
 docs/dev/reference/layer_embedding.html       |  16 +-
 docs/dev/reference/layer_equalization.html    |  10 +-
 docs/dev/reference/layer_feature_space.html   |   8 +-
 docs/dev/reference/layer_flatten.html         |   6 +-
 .../reference/layer_flax_module_wrapper.html  |   6 +-
 .../dev/reference/layer_gaussian_dropout.html |   6 +-
 docs/dev/reference/layer_gaussian_noise.html  |   6 +-
 .../layer_global_average_pooling_1d.html      |   6 +-
 .../layer_global_average_pooling_2d.html      |   6 +-
 .../layer_global_average_pooling_3d.html      |   6 +-
 .../layer_global_max_pooling_1d.html          |   6 +-
 .../layer_global_max_pooling_2d.html          |   6 +-
 .../layer_global_max_pooling_3d.html          |   6 +-
 .../reference/layer_group_normalization.html  |   6 +-
 .../layer_group_query_attention.html          |   6 +-
 docs/dev/reference/layer_gru.html             |   6 +-
 docs/dev/reference/layer_hashed_crossing.html |   8 +-
 docs/dev/reference/layer_hashing.html         |   8 +-
 docs/dev/reference/layer_identity.html        |   6 +-
 docs/dev/reference/layer_input.html           |   4 +-
 docs/dev/reference/layer_integer_lookup.html  |   8 +-
 .../reference/layer_jax_model_wrapper.html    |   6 +-
 docs/dev/reference/layer_lambda.html          |   6 +-
 .../reference/layer_layer_normalization.html  |  16 +-
 docs/dev/reference/layer_lstm.html            |   6 +-
 docs/dev/reference/layer_masking.html         |   6 +-
 .../layer_max_num_bounding_boxes.html         |  10 +-
 docs/dev/reference/layer_max_pooling_1d.html  |   6 +-
 docs/dev/reference/layer_max_pooling_2d.html  |   6 +-
 docs/dev/reference/layer_max_pooling_3d.html  |   6 +-
 docs/dev/reference/layer_maximum.html         |   6 +-
 docs/dev/reference/layer_mel_spectrogram.html |   8 +-
 docs/dev/reference/layer_minimum.html         |   6 +-
 docs/dev/reference/layer_mix_up.html          |  10 +-
 .../reference/layer_multi_head_attention.html |   6 +-
 docs/dev/reference/layer_multiply.html        |   6 +-
 docs/dev/reference/layer_normalization.html   |   8 +-
 docs/dev/reference/layer_permute.html         |   6 +-
 docs/dev/reference/layer_pipeline.html        |   4 +-
 docs/dev/reference/layer_rand_augment.html    |  10 +-
 .../reference/layer_random_brightness.html    |   8 +-
 .../layer_random_color_degeneration.html      |  10 +-
 .../reference/layer_random_color_jitter.html  |  10 +-
 docs/dev/reference/layer_random_contrast.html |   8 +-
 docs/dev/reference/layer_random_crop.html     |   8 +-
 .../layer_random_elastic_transform.html       | 197 ++++++
 docs/dev/reference/layer_random_erasing.html  |  10 +-
 docs/dev/reference/layer_random_flip.html     |   8 +-
 .../reference/layer_random_gaussian_blur.html |  10 +-
 .../dev/reference/layer_random_grayscale.html |  10 +-
 docs/dev/reference/layer_random_hue.html      |  10 +-
 docs/dev/reference/layer_random_invert.html   |  10 +-
 .../reference/layer_random_perspective.html   |  10 +-
 .../reference/layer_random_posterization.html |  10 +-
 docs/dev/reference/layer_random_rotation.html |   8 +-
 .../reference/layer_random_saturation.html    |  10 +-
 .../dev/reference/layer_random_sharpness.html |  10 +-
 docs/dev/reference/layer_random_shear.html    |  10 +-
 .../reference/layer_random_translation.html   |   8 +-
 docs/dev/reference/layer_random_zoom.html     |   8 +-
 docs/dev/reference/layer_repeat_vector.html   |   6 +-
 docs/dev/reference/layer_rescaling.html       |  18 +-
 docs/dev/reference/layer_reshape.html         |   6 +-
 docs/dev/reference/layer_resizing.html        |  10 +-
 .../reference/layer_rms_normalization.html    |   6 +-
 docs/dev/reference/layer_rnn.html             |   7 +-
 .../reference/layer_separable_conv_1d.html    |   6 +-
 .../reference/layer_separable_conv_2d.html    |   6 +-
 docs/dev/reference/layer_simple_rnn.html      |   6 +-
 docs/dev/reference/layer_solarization.html    |  10 +-
 .../reference/layer_spatial_dropout_1d.html   |   6 +-
 .../reference/layer_spatial_dropout_2d.html   |   6 +-
 .../reference/layer_spatial_dropout_3d.html   |   6 +-
 .../layer_spectral_normalization.html         |   6 +-
 .../dev/reference/layer_stft_spectrogram.html |   8 +-
 docs/dev/reference/layer_string_lookup.html   |   8 +-
 docs/dev/reference/layer_subtract.html        |   6 +-
 .../reference/layer_text_vectorization.html   |   8 +-
 docs/dev/reference/layer_tfsm.html            |  10 +-
 .../dev/reference/layer_time_distributed.html |   6 +-
 .../reference/layer_torch_module_wrapper.html |  18 +-
 .../reference/layer_unit_normalization.html   |   6 +-
 docs/dev/reference/layer_upsampling_1d.html   |   6 +-
 docs/dev/reference/layer_upsampling_2d.html   |   6 +-
 docs/dev/reference/layer_upsampling_3d.html   |   6 +-
 docs/dev/reference/layer_zero_padding_1d.html |   6 +-
 docs/dev/reference/layer_zero_padding_2d.html |   6 +-
 docs/dev/reference/layer_zero_padding_3d.html |   6 +-
 .../learning_rate_schedule_cosine_decay.html  |   4 +-
 ...g_rate_schedule_cosine_decay_restarts.html |   4 +-
 ...rning_rate_schedule_exponential_decay.html |   4 +-
 ...ning_rate_schedule_inverse_time_decay.html |   4 +-
 ...ate_schedule_piecewise_constant_decay.html |   4 +-
 ...arning_rate_schedule_polynomial_decay.html |   4 +-
 docs/dev/reference/load_model.html            |   4 +-
 docs/dev/reference/load_model_weights.html    |  31 +-
 .../reference/loss_binary_crossentropy.html   |   6 +-
 .../loss_binary_focal_crossentropy.html       |   6 +-
 .../loss_categorical_crossentropy.html        |   6 +-
 .../loss_categorical_focal_crossentropy.html  |   6 +-
 ...categorical_generalized_cross_entropy.html | 173 +++++
 .../dev/reference/loss_categorical_hinge.html |   6 +-
 docs/dev/reference/loss_circle.html           |   6 +-
 .../dev/reference/loss_cosine_similarity.html |   6 +-
 docs/dev/reference/loss_ctc.html              |   6 +-
 docs/dev/reference/loss_dice.html             |  12 +-
 docs/dev/reference/loss_hinge.html            |   6 +-
 docs/dev/reference/loss_huber.html            |   6 +-
 docs/dev/reference/loss_kl_divergence.html    |   6 +-
 docs/dev/reference/loss_log_cosh.html         |   6 +-
 .../reference/loss_mean_absolute_error.html   |   6 +-
 .../loss_mean_absolute_percentage_error.html  |   6 +-
 .../reference/loss_mean_squared_error.html    |   6 +-
 .../loss_mean_squared_logarithmic_error.html  |   6 +-
 docs/dev/reference/loss_poisson.html          |   6 +-
 .../loss_sparse_categorical_crossentropy.html |  27 +-
 docs/dev/reference/loss_squared_hinge.html    |   6 +-
 docs/dev/reference/loss_tversky.html          |   6 +-
 docs/dev/reference/mark_active.html           |   4 +-
 docs/dev/reference/metric_auc.html            |   4 +-
 .../dev/reference/metric_binary_accuracy.html |   4 +-
 .../reference/metric_binary_crossentropy.html |   6 +-
 .../metric_binary_focal_crossentropy.html     |  13 +-
 docs/dev/reference/metric_binary_iou.html     |   4 +-
 .../metric_categorical_accuracy.html          |   4 +-
 .../metric_categorical_crossentropy.html      |   6 +-
 ...metric_categorical_focal_crossentropy.html |   6 +-
 .../reference/metric_categorical_hinge.html   |   6 +-
 .../metric_concordance_correlation.html       |   4 +-
 .../reference/metric_cosine_similarity.html   |   4 +-
 docs/dev/reference/metric_f1_score.html       |   4 +-
 .../dev/reference/metric_false_negatives.html |   4 +-
 .../dev/reference/metric_false_positives.html |   4 +-
 docs/dev/reference/metric_fbeta_score.html    |   4 +-
 docs/dev/reference/metric_hinge.html          |   6 +-
 docs/dev/reference/metric_huber.html          |   6 +-
 docs/dev/reference/metric_iou.html            |   4 +-
 docs/dev/reference/metric_kl_divergence.html  |   6 +-
 docs/dev/reference/metric_log_cosh.html       |   6 +-
 docs/dev/reference/metric_log_cosh_error.html |   4 +-
 docs/dev/reference/metric_mean.html           |   4 +-
 .../reference/metric_mean_absolute_error.html |   6 +-
 ...metric_mean_absolute_percentage_error.html |  10 +-
 docs/dev/reference/metric_mean_iou.html       |   4 +-
 .../reference/metric_mean_squared_error.html  |   6 +-
 ...metric_mean_squared_logarithmic_error.html |   6 +-
 docs/dev/reference/metric_mean_wrapper.html   |   4 +-
 docs/dev/reference/metric_one_hot_iou.html    |   4 +-
 .../reference/metric_one_hot_mean_iou.html    |   4 +-
 .../reference/metric_pearson_correlation.html |   4 +-
 docs/dev/reference/metric_poisson.html        |   6 +-
 docs/dev/reference/metric_precision.html      |   4 +-
 .../reference/metric_precision_at_recall.html |   4 +-
 docs/dev/reference/metric_r2_score.html       |   4 +-
 docs/dev/reference/metric_recall.html         |   4 +-
 .../reference/metric_recall_at_precision.html |   4 +-
 .../metric_root_mean_squared_error.html       |   4 +-
 .../metric_sensitivity_at_specificity.html    |   4 +-
 .../metric_sparse_categorical_accuracy.html   |   4 +-
 ...etric_sparse_categorical_crossentropy.html |   6 +-
 ...ric_sparse_top_k_categorical_accuracy.html |   4 +-
 .../metric_specificity_at_sensitivity.html    |   4 +-
 docs/dev/reference/metric_squared_hinge.html  |   6 +-
 docs/dev/reference/metric_sum.html            |   4 +-
 .../metric_top_k_categorical_accuracy.html    |   4 +-
 docs/dev/reference/metric_true_negatives.html |   4 +-
 docs/dev/reference/metric_true_positives.html |   4 +-
 docs/dev/reference/multi-assign.html          |   6 +-
 docs/dev/reference/new_callback_class.html    |   4 +-
 docs/dev/reference/new_layer_class.html       |   4 +-
 .../new_learning_rate_schedule_class.html     |   4 +-
 docs/dev/reference/new_loss_class.html        |   4 +-
 docs/dev/reference/new_metric_class.html      |   4 +-
 docs/dev/reference/new_model_class.html       |   4 +-
 docs/dev/reference/newaxis.html               |   4 +-
 docs/dev/reference/normalize.html             |   4 +-
 docs/dev/reference/op_abs.html                |   8 +-
 docs/dev/reference/op_add.html                |   8 +-
 docs/dev/reference/op_all.html                |   8 +-
 docs/dev/reference/op_angle.html              | 111 +++
 docs/dev/reference/op_any.html                |   8 +-
 docs/dev/reference/op_append.html             |   8 +-
 docs/dev/reference/op_arange.html             |   8 +-
 docs/dev/reference/op_arccos.html             |   8 +-
 docs/dev/reference/op_arccosh.html            |   8 +-
 docs/dev/reference/op_arcsin.html             |   8 +-
 docs/dev/reference/op_arcsinh.html            |   8 +-
 docs/dev/reference/op_arctan.html             |   8 +-
 docs/dev/reference/op_arctan2.html            |   8 +-
 docs/dev/reference/op_arctanh.html            |   8 +-
 docs/dev/reference/op_argmax.html             |   8 +-
 docs/dev/reference/op_argmin.html             |   8 +-
 docs/dev/reference/op_argpartition.html       |   8 +-
 docs/dev/reference/op_argsort.html            |   8 +-
 docs/dev/reference/op_array.html              |   8 +-
 docs/dev/reference/op_associative_scan.html   |   6 +-
 docs/dev/reference/op_average.html            |   8 +-
 docs/dev/reference/op_average_pool.html       |   8 +-
 docs/dev/reference/op_bartlett.html           | 108 +++
 .../dev/reference/op_batch_normalization.html |   8 +-
 .../dev/reference/op_binary_crossentropy.html |   8 +-
 docs/dev/reference/op_bincount.html           |   8 +-
 docs/dev/reference/op_bitwise_and.html        |   8 +-
 docs/dev/reference/op_bitwise_invert.html     |   8 +-
 docs/dev/reference/op_bitwise_left_shift.html |   8 +-
 docs/dev/reference/op_bitwise_not.html        |   8 +-
 docs/dev/reference/op_bitwise_or.html         |   8 +-
 .../dev/reference/op_bitwise_right_shift.html |   8 +-
 docs/dev/reference/op_bitwise_xor.html        |   8 +-
 docs/dev/reference/op_blackman.html           | 116 ++++
 docs/dev/reference/op_broadcast_to.html       |   8 +-
 docs/dev/reference/op_cast.html               |   6 +-
 .../op_categorical_crossentropy.html          |   8 +-
 docs/dev/reference/op_cbrt.html               | 108 +++
 docs/dev/reference/op_ceil.html               |   8 +-
 docs/dev/reference/op_celu.html               |   8 +-
 docs/dev/reference/op_cholesky.html           |   6 +-
 docs/dev/reference/op_clip.html               |   8 +-
 docs/dev/reference/op_concatenate.html        |   8 +-
 docs/dev/reference/op_cond.html               |   6 +-
 docs/dev/reference/op_conj.html               |   8 +-
 docs/dev/reference/op_conv.html               |  10 +-
 docs/dev/reference/op_conv_transpose.html     |  10 +-
 docs/dev/reference/op_convert_to_numpy.html   |   6 +-
 docs/dev/reference/op_convert_to_tensor.html  |   6 +-
 docs/dev/reference/op_copy.html               |   8 +-
 docs/dev/reference/op_corrcoef.html           | 110 +++
 docs/dev/reference/op_correlate.html          |   8 +-
 docs/dev/reference/op_cos.html                |   8 +-
 docs/dev/reference/op_cosh.html               |   8 +-
 docs/dev/reference/op_count_nonzero.html      |   8 +-
 docs/dev/reference/op_cross.html              |   8 +-
 docs/dev/reference/op_ctc_decode.html         |   8 +-
 docs/dev/reference/op_ctc_loss.html           |   8 +-
 docs/dev/reference/op_cumprod.html            |   8 +-
 docs/dev/reference/op_cumsum.html             |   8 +-
 docs/dev/reference/op_custom_gradient.html    |   6 +-
 docs/dev/reference/op_deg2rad.html            | 111 +++
 docs/dev/reference/op_depthwise_conv.html     |   8 +-
 docs/dev/reference/op_det.html                |   6 +-
 docs/dev/reference/op_diag.html               |   8 +-
 docs/dev/reference/op_diagflat.html           |   8 +-
 docs/dev/reference/op_diagonal.html           |   8 +-
 docs/dev/reference/op_diff.html               |   8 +-
 docs/dev/reference/op_digitize.html           |   8 +-
 docs/dev/reference/op_divide.html             |   8 +-
 docs/dev/reference/op_divide_no_nan.html      |   8 +-
 docs/dev/reference/op_dot.html                |   8 +-
 .../reference/op_dot_product_attention.html   |  16 +-
 docs/dev/reference/op_dtype.html              |   6 +-
 docs/dev/reference/op_eig.html                |   6 +-
 docs/dev/reference/op_eigh.html               |   6 +-
 docs/dev/reference/op_einsum.html             |  11 +-
 docs/dev/reference/op_elu.html                |   8 +-
 docs/dev/reference/op_empty.html              |   8 +-
 docs/dev/reference/op_equal.html              |   8 +-
 docs/dev/reference/op_erf.html                |   8 +-
 docs/dev/reference/op_erfinv.html             |   6 +-
 docs/dev/reference/op_exp.html                |   8 +-
 docs/dev/reference/op_exp2.html               |   8 +-
 docs/dev/reference/op_expand_dims.html        |   8 +-
 docs/dev/reference/op_expm1.html              |   8 +-
 docs/dev/reference/op_extract_sequences.html  |   6 +-
 docs/dev/reference/op_eye.html                |   8 +-
 docs/dev/reference/op_fft.html                |   6 +-
 docs/dev/reference/op_fft2.html               |   6 +-
 docs/dev/reference/op_flip.html               |   8 +-
 docs/dev/reference/op_floor.html              |   8 +-
 docs/dev/reference/op_floor_divide.html       |   8 +-
 docs/dev/reference/op_fori_loop.html          |   6 +-
 docs/dev/reference/op_full.html               |   8 +-
 docs/dev/reference/op_full_like.html          |   8 +-
 docs/dev/reference/op_gelu.html               |  10 +-
 docs/dev/reference/op_get_item.html           |   8 +-
 docs/dev/reference/op_glu.html                |   8 +-
 docs/dev/reference/op_greater.html            |   8 +-
 docs/dev/reference/op_greater_equal.html      |   8 +-
 docs/dev/reference/op_hamming.html            | 111 +++
 docs/dev/reference/op_hanning.html            | 111 +++
 docs/dev/reference/op_hard_shrink.html        |   8 +-
 docs/dev/reference/op_hard_sigmoid.html       |   8 +-
 docs/dev/reference/op_hard_silu.html          |   8 +-
 docs/dev/reference/op_hard_tanh.html          |   8 +-
 docs/dev/reference/op_heaviside.html          | 122 ++++
 docs/dev/reference/op_histogram.html          |   8 +-
 docs/dev/reference/op_hstack.html             |   8 +-
 docs/dev/reference/op_identity.html           |   8 +-
 docs/dev/reference/op_ifft2.html              |   6 +-
 docs/dev/reference/op_imag.html               |   8 +-
 .../reference/op_image_affine_transform.html  |   6 +-
 docs/dev/reference/op_image_crop.html         |   6 +-
 .../reference/op_image_elastic_transform.html | 178 +++++
 .../reference/op_image_extract_patches.html   |   6 +-
 .../dev/reference/op_image_gaussian_blur.html |   6 +-
 docs/dev/reference/op_image_hsv_to_rgb.html   |   6 +-
 .../reference/op_image_map_coordinates.html   |   6 +-
 docs/dev/reference/op_image_pad.html          |   6 +-
 .../op_image_perspective_transform.html       |   6 +-
 docs/dev/reference/op_image_resize.html       |   6 +-
 .../reference/op_image_rgb_to_grayscale.html  |   6 +-
 docs/dev/reference/op_image_rgb_to_hsv.html   |   6 +-
 docs/dev/reference/op_in_top_k.html           |   6 +-
 docs/dev/reference/op_inner.html              |   8 +-
 docs/dev/reference/op_inv.html                |   6 +-
 docs/dev/reference/op_irfft.html              |   6 +-
 docs/dev/reference/op_is_tensor.html          |   6 +-
 docs/dev/reference/op_isclose.html            |   8 +-
 docs/dev/reference/op_isfinite.html           |   8 +-
 docs/dev/reference/op_isinf.html              |   8 +-
 docs/dev/reference/op_isnan.html              |   8 +-
 docs/dev/reference/op_istft.html              |   6 +-
 docs/dev/reference/op_kaiser.html             | 118 ++++
 .../dev/reference/op_layer_normalization.html | 139 ++++
 docs/dev/reference/op_leaky_relu.html         |   8 +-
 docs/dev/reference/op_left_shift.html         |   8 +-
 docs/dev/reference/op_less.html               |   8 +-
 docs/dev/reference/op_less_equal.html         |   8 +-
 docs/dev/reference/op_linspace.html           |   8 +-
 docs/dev/reference/op_log.html                |   8 +-
 docs/dev/reference/op_log10.html              |   8 +-
 docs/dev/reference/op_log1p.html              |   8 +-
 docs/dev/reference/op_log2.html               |   8 +-
 docs/dev/reference/op_log_sigmoid.html        |   8 +-
 docs/dev/reference/op_log_softmax.html        |   8 +-
 docs/dev/reference/op_logaddexp.html          |   8 +-
 docs/dev/reference/op_logdet.html             |   8 +-
 docs/dev/reference/op_logical_and.html        |   8 +-
 docs/dev/reference/op_logical_not.html        |   8 +-
 docs/dev/reference/op_logical_or.html         |   8 +-
 docs/dev/reference/op_logical_xor.html        |   8 +-
 docs/dev/reference/op_logspace.html           |   8 +-
 docs/dev/reference/op_logsumexp.html          |   6 +-
 docs/dev/reference/op_lstsq.html              |   8 +-
 docs/dev/reference/op_lu_factor.html          |   6 +-
 docs/dev/reference/op_map.html                |   6 +-
 docs/dev/reference/op_matmul.html             |   8 +-
 docs/dev/reference/op_max.html                |   8 +-
 docs/dev/reference/op_max_pool.html           |   8 +-
 docs/dev/reference/op_maximum.html            |   8 +-
 docs/dev/reference/op_mean.html               |   8 +-
 docs/dev/reference/op_median.html             |   8 +-
 docs/dev/reference/op_meshgrid.html           |   8 +-
 docs/dev/reference/op_min.html                |   8 +-
 docs/dev/reference/op_minimum.html            |   8 +-
 docs/dev/reference/op_mod.html                |   8 +-
 docs/dev/reference/op_moments.html            |   8 +-
 docs/dev/reference/op_moveaxis.html           |   8 +-
 docs/dev/reference/op_multi_hot.html          |   8 +-
 docs/dev/reference/op_multiply.html           |   8 +-
 docs/dev/reference/op_nan_to_num.html         |   8 +-
 docs/dev/reference/op_ndim.html               |   8 +-
 docs/dev/reference/op_negative.html           |   8 +-
 docs/dev/reference/op_nonzero.html            |   8 +-
 docs/dev/reference/op_norm.html               |   6 +-
 docs/dev/reference/op_normalize.html          |   8 +-
 docs/dev/reference/op_not_equal.html          |   8 +-
 docs/dev/reference/op_one_hot.html            |   8 +-
 docs/dev/reference/op_ones.html               |   8 +-
 docs/dev/reference/op_ones_like.html          |   8 +-
 docs/dev/reference/op_outer.html              |   8 +-
 docs/dev/reference/op_pad.html                |   8 +-
 docs/dev/reference/op_polar.html              |   8 +-
 docs/dev/reference/op_power.html              |   8 +-
 docs/dev/reference/op_prod.html               |   8 +-
 docs/dev/reference/op_psnr.html               |   8 +-
 docs/dev/reference/op_qr.html                 |   6 +-
 docs/dev/reference/op_quantile.html           |   8 +-
 docs/dev/reference/op_ravel.html              |   8 +-
 docs/dev/reference/op_real.html               |   8 +-
 docs/dev/reference/op_rearrange.html          |   6 +-
 docs/dev/reference/op_reciprocal.html         |   8 +-
 docs/dev/reference/op_relu.html               |   8 +-
 docs/dev/reference/op_relu6.html              |   8 +-
 docs/dev/reference/op_repeat.html             |   8 +-
 docs/dev/reference/op_reshape.html            |   8 +-
 docs/dev/reference/op_rfft.html               |   6 +-
 docs/dev/reference/op_right_shift.html        |   8 +-
 docs/dev/reference/op_rms_normalization.html  |  40 +-
 docs/dev/reference/op_roll.html               |   8 +-
 docs/dev/reference/op_rot90.html              |   8 +-
 docs/dev/reference/op_round.html              |   8 +-
 docs/dev/reference/op_rsqrt.html              |   6 +-
 docs/dev/reference/op_saturate_cast.html      |   8 +-
 docs/dev/reference/op_scan.html               |   6 +-
 docs/dev/reference/op_scatter.html            |   6 +-
 docs/dev/reference/op_scatter_update.html     |   6 +-
 docs/dev/reference/op_searchsorted.html       |  34 +-
 docs/dev/reference/op_segment_max.html        |   6 +-
 docs/dev/reference/op_segment_sum.html        |   6 +-
 docs/dev/reference/op_select.html             |   8 +-
 docs/dev/reference/op_selu.html               |   8 +-
 docs/dev/reference/op_separable_conv.html     |   8 +-
 docs/dev/reference/op_shape.html              |   6 +-
 docs/dev/reference/op_sigmoid.html            |   8 +-
 docs/dev/reference/op_sign.html               |   8 +-
 docs/dev/reference/op_signbit.html            |   8 +-
 docs/dev/reference/op_silu.html               |   8 +-
 docs/dev/reference/op_sin.html                |   8 +-
 docs/dev/reference/op_sinh.html               |   8 +-
 docs/dev/reference/op_size.html               |   8 +-
 docs/dev/reference/op_slice.html              |   6 +-
 docs/dev/reference/op_slice_update.html       |   6 +-
 docs/dev/reference/op_slogdet.html            |   6 +-
 docs/dev/reference/op_soft_shrink.html        |   8 +-
 docs/dev/reference/op_softmax.html            |   8 +-
 docs/dev/reference/op_softplus.html           |   8 +-
 docs/dev/reference/op_softsign.html           |   8 +-
 docs/dev/reference/op_solve.html              |   6 +-
 docs/dev/reference/op_solve_triangular.html   |   6 +-
 docs/dev/reference/op_sort.html               |   8 +-
 .../op_sparse_categorical_crossentropy.html   |   8 +-
 docs/dev/reference/op_sparse_plus.html        |   8 +-
 docs/dev/reference/op_sparse_sigmoid.html     | 118 ++++
 docs/dev/reference/op_sparsemax.html          |   8 +-
 docs/dev/reference/op_split.html              |   8 +-
 docs/dev/reference/op_sqrt.html               |   8 +-
 docs/dev/reference/op_square.html             |   8 +-
 docs/dev/reference/op_squareplus.html         |   8 +-
 docs/dev/reference/op_squeeze.html            |   8 +-
 docs/dev/reference/op_stack.html              |   8 +-
 docs/dev/reference/op_std.html                |   8 +-
 docs/dev/reference/op_stft.html               |   6 +-
 docs/dev/reference/op_stop_gradient.html      |   6 +-
 docs/dev/reference/op_subset.html             |  10 +-
 docs/dev/reference/op_subtract.html           |   8 +-
 docs/dev/reference/op_sum.html                |   8 +-
 docs/dev/reference/op_svd.html                |   6 +-
 docs/dev/reference/op_swapaxes.html           |   8 +-
 docs/dev/reference/op_switch.html             |   6 +-
 docs/dev/reference/op_take.html               |   8 +-
 docs/dev/reference/op_take_along_axis.html    |   8 +-
 docs/dev/reference/op_tan.html                |   8 +-
 docs/dev/reference/op_tanh.html               |   8 +-
 docs/dev/reference/op_tanh_shrink.html        |   8 +-
 docs/dev/reference/op_tensordot.html          |   8 +-
 docs/dev/reference/op_threshold.html          |   8 +-
 docs/dev/reference/op_tile.html               |   8 +-
 docs/dev/reference/op_top_k.html              |   6 +-
 docs/dev/reference/op_trace.html              |   8 +-
 docs/dev/reference/op_transpose.html          |   8 +-
 docs/dev/reference/op_tri.html                |   8 +-
 docs/dev/reference/op_tril.html               |   8 +-
 docs/dev/reference/op_triu.html               |   8 +-
 docs/dev/reference/op_trunc.html              |   8 +-
 docs/dev/reference/op_unravel_index.html      |   8 +-
 docs/dev/reference/op_unstack.html            |   6 +-
 docs/dev/reference/op_var.html                |   8 +-
 docs/dev/reference/op_vdot.html               |   8 +-
 docs/dev/reference/op_vectorize.html          |   8 +-
 docs/dev/reference/op_vectorized_map.html     |   6 +-
 docs/dev/reference/op_view_as_complex.html    | 115 ++++
 docs/dev/reference/op_view_as_real.html       | 115 ++++
 docs/dev/reference/op_vstack.html             |   8 +-
 docs/dev/reference/op_where.html              |   8 +-
 docs/dev/reference/op_while_loop.html         |   6 +-
 docs/dev/reference/op_zeros.html              |   8 +-
 docs/dev/reference/op_zeros_like.html         |   8 +-
 docs/dev/reference/optimizer_adadelta.html    |   6 +-
 docs/dev/reference/optimizer_adafactor.html   |   6 +-
 docs/dev/reference/optimizer_adagrad.html     |   6 +-
 docs/dev/reference/optimizer_adam.html        |   6 +-
 docs/dev/reference/optimizer_adam_w.html      |   6 +-
 docs/dev/reference/optimizer_adamax.html      |   6 +-
 docs/dev/reference/optimizer_ftrl.html        |   6 +-
 docs/dev/reference/optimizer_lamb.html        |   6 +-
 docs/dev/reference/optimizer_lion.html        |  18 +-
 docs/dev/reference/optimizer_loss_scale.html  |   6 +-
 docs/dev/reference/optimizer_muon.html        | 265 +++++++
 docs/dev/reference/optimizer_nadam.html       |   6 +-
 docs/dev/reference/optimizer_rmsprop.html     |   6 +-
 docs/dev/reference/optimizer_sgd.html         |   6 +-
 docs/dev/reference/pad_sequences.html         |   4 +-
 docs/dev/reference/pipe.html                  |   4 +-
 .../plot.keras.src.models.model.Model.html    |  16 +-
 .../plot.keras_training_history.html          |   4 +-
 docs/dev/reference/pop_layer.html             |   4 +-
 .../predict.keras.src.models.model.Model.html |   4 +-
 docs/dev/reference/predict_on_batch.html      |   4 +-
 docs/dev/reference/process_utils.html         |   4 +-
 docs/dev/reference/quantize_weights.html      |   4 +-
 docs/dev/reference/random_beta.html           |   4 +-
 docs/dev/reference/random_binomial.html       |   4 +-
 docs/dev/reference/random_categorical.html    |   4 +-
 docs/dev/reference/random_dropout.html        |   4 +-
 docs/dev/reference/random_gamma.html          |   4 +-
 docs/dev/reference/random_integer.html        |   4 +-
 docs/dev/reference/random_normal.html         |   4 +-
 docs/dev/reference/random_seed_generator.html |   4 +-
 docs/dev/reference/random_shuffle.html        |   4 +-
 .../reference/random_truncated_normal.html    |   4 +-
 docs/dev/reference/random_uniform.html        |   4 +-
 docs/dev/reference/reexports.html             |  10 +-
 .../register_keras_serializable.html          |   4 +-
 docs/dev/reference/regularizer_l1.html        |   4 +-
 docs/dev/reference/regularizer_l1_l2.html     |   4 +-
 docs/dev/reference/regularizer_l2.html        |   4 +-
 .../dev/reference/regularizer_orthogonal.html |   4 +-
 docs/dev/reference/reset_state.html           |   4 +-
 docs/dev/reference/rnn_cell_gru.html          |   6 +-
 docs/dev/reference/rnn_cell_lstm.html         |   8 +-
 docs/dev/reference/rnn_cell_simple.html       |   6 +-
 docs/dev/reference/rnn_cells_stack.html       |   6 +-
 docs/dev/reference/save_model.html            |   4 +-
 docs/dev/reference/save_model_config.html     |   4 +-
 docs/dev/reference/save_model_weights.html    |  87 ++-
 .../dev/reference/serialize_keras_object.html |   4 +-
 docs/dev/reference/set_random_seed.html       |   4 +-
 docs/dev/reference/set_state_tree.html        |   4 +-
 docs/dev/reference/shape.html                 |   4 +-
 docs/dev/reference/split_dataset.html         |   4 +-
 .../summary.keras.src.models.model.Model.html |   4 +-
 docs/dev/reference/test_on_batch.html         |   4 +-
 .../text_dataset_from_directory.html          |   4 +-
 docs/dev/reference/time_distributed.html      |   4 +-
 .../timeseries_dataset_from_array.html        |   4 +-
 docs/dev/reference/to_categorical.html        |   4 +-
 docs/dev/reference/train_on_batch.html        |   4 +-
 docs/dev/reference/use_backend.html           |   4 +-
 .../reference/with_custom_object_scope.html   |   4 +-
 docs/dev/reference/zip_lists.html             |   6 +-
 docs/dev/search.json                          |   2 +-
 docs/dev/sitemap.xml                          |  21 +
 tools/make-website.R                          |   2 +-
 811 files changed, 6336 insertions(+), 3071 deletions(-)
 create mode 100644 docs/dev/reference/activation_sparse_sigmoid.html
 create mode 100644 docs/dev/reference/config_is_nnx_enabled.html
 create mode 100644 docs/dev/reference/config_max_epochs.html
 create mode 100644 docs/dev/reference/config_max_steps_per_epoch.html
 create mode 100644 docs/dev/reference/config_set_max_epochs.html
 create mode 100644 docs/dev/reference/config_set_max_steps_per_epoch.html
 create mode 100644 docs/dev/reference/generate.html
 create mode 100644 docs/dev/reference/layer_random_elastic_transform.html
 create mode 100644 docs/dev/reference/loss_categorical_generalized_cross_entropy.html
 create mode 100644 docs/dev/reference/op_angle.html
 create mode 100644 docs/dev/reference/op_bartlett.html
 create mode 100644 docs/dev/reference/op_blackman.html
 create mode 100644 docs/dev/reference/op_cbrt.html
 create mode 100644 docs/dev/reference/op_corrcoef.html
 create mode 100644 docs/dev/reference/op_deg2rad.html
 create mode 100644 docs/dev/reference/op_hamming.html
 create mode 100644 docs/dev/reference/op_hanning.html
 create mode 100644 docs/dev/reference/op_heaviside.html
 create mode 100644 docs/dev/reference/op_image_elastic_transform.html
 create mode 100644 docs/dev/reference/op_kaiser.html
 create mode 100644 docs/dev/reference/op_layer_normalization.html
 create mode 100644 docs/dev/reference/op_sparse_sigmoid.html
 create mode 100644 docs/dev/reference/op_view_as_complex.html
 create mode 100644 docs/dev/reference/op_view_as_real.html
 create mode 100644 docs/dev/reference/optimizer_muon.html

diff --git a/docs/dev/LICENSE-text.html b/docs/dev/LICENSE-text.html
index 6df9ffd5d..173dbc624 100644
--- a/docs/dev/LICENSE-text.html
+++ b/docs/dev/LICENSE-text.html
@@ -8,7 +8,7 @@
 
     <a class="navbar-brand me-2" href="index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -64,7 +64,7 @@ <h1>License</h1>
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/articles/custom_train_step_in_tensorflow.html b/docs/dev/articles/custom_train_step_in_tensorflow.html
index 6e50a2967..8e940a86e 100644
--- a/docs/dev/articles/custom_train_step_in_tensorflow.html
+++ b/docs/dev/articles/custom_train_step_in_tensorflow.html
@@ -31,7 +31,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -204,9 +204,9 @@ <h2 id="a-first-simple-example">A first simple example<a class="anchor" aria-lab
 <span><span class="va">y</span> <span class="op">&lt;-</span> <span class="fu"><a href="../reference/random_normal.html">random_normal</a></span><span class="op">(</span><span class="fu"><a href="https://rdrr.io/r/base/c.html" class="external-link">c</a></span><span class="op">(</span><span class="fl">1000</span>, <span class="fl">1</span><span class="op">)</span><span class="op">)</span></span>
 <span><span class="va">model</span> <span class="op">|&gt;</span> <span class="fu"><a href="https://generics.r-lib.org/reference/fit.html" class="external-link">fit</a></span><span class="op">(</span><span class="va">x</span>, <span class="va">y</span>, epochs <span class="op">=</span> <span class="fl">3</span><span class="op">)</span></span></code></pre></div>
 <pre><code><span><span class="co">## Epoch 1/3</span></span>
-<span><span class="co">## 32/32 - 1s - 23ms/step - mae: 1.4339 - loss: 3.2271</span></span>
+<span><span class="co">## 32/32 - 1s - 20ms/step - mae: 1.4339 - loss: 3.2271</span></span>
 <span><span class="co">## Epoch 2/3</span></span>
-<span><span class="co">## 32/32 - 0s - 2ms/step - mae: 1.3605 - loss: 2.9034</span></span>
+<span><span class="co">## 32/32 - 0s - 3ms/step - mae: 1.3605 - loss: 2.9034</span></span>
 <span><span class="co">## Epoch 3/3</span></span>
 <span><span class="co">## 32/32 - 0s - 2ms/step - mae: 1.2960 - loss: 2.6272</span></span></code></pre>
 </div>
@@ -291,7 +291,7 @@ <h2 id="going-lower-level">Going lower-level<a class="anchor" aria-label="anchor
 <span><span class="va">y</span> <span class="op">&lt;-</span> <span class="fu"><a href="../reference/random_normal.html">random_normal</a></span><span class="op">(</span><span class="fu"><a href="https://rdrr.io/r/base/c.html" class="external-link">c</a></span><span class="op">(</span><span class="fl">1000</span>, <span class="fl">1</span><span class="op">)</span><span class="op">)</span></span>
 <span><span class="va">model</span> <span class="op">|&gt;</span> <span class="fu"><a href="https://generics.r-lib.org/reference/fit.html" class="external-link">fit</a></span><span class="op">(</span><span class="va">x</span>, <span class="va">y</span>, epochs <span class="op">=</span> <span class="fl">3</span><span class="op">)</span></span></code></pre></div>
 <pre><code><span><span class="co">## Epoch 1/3</span></span>
-<span><span class="co">## 32/32 - 1s - 20ms/step - loss: 2.5170 - mae: 1.2923</span></span>
+<span><span class="co">## 32/32 - 1s - 18ms/step - loss: 2.5170 - mae: 1.2923</span></span>
 <span><span class="co">## Epoch 2/3</span></span>
 <span><span class="co">## 32/32 - 0s - 2ms/step - loss: 2.2689 - mae: 1.2241</span></span>
 <span><span class="co">## Epoch 3/3</span></span>
@@ -361,11 +361,11 @@ <h2 id="supporting-sample_weight-class_weight">Supporting <code>sample_weight</c
 <span><span class="va">sw</span> <span class="op">&lt;-</span> <span class="fu"><a href="../reference/random_normal.html">random_normal</a></span><span class="op">(</span><span class="fu"><a href="https://rdrr.io/r/base/c.html" class="external-link">c</a></span><span class="op">(</span><span class="fl">1000</span>, <span class="fl">1</span><span class="op">)</span><span class="op">)</span></span>
 <span><span class="va">model</span> <span class="op">|&gt;</span> <span class="fu"><a href="https://generics.r-lib.org/reference/fit.html" class="external-link">fit</a></span><span class="op">(</span><span class="va">x</span>, <span class="va">y</span>, sample_weight <span class="op">=</span> <span class="va">sw</span>, epochs <span class="op">=</span> <span class="fl">3</span><span class="op">)</span></span></code></pre></div>
 <pre><code><span><span class="co">## Epoch 1/3</span></span>
-<span><span class="co">## 32/32 - 1s - 23ms/step - mae: 1.3434 - loss: 0.1681</span></span>
+<span><span class="co">## 32/32 - 1s - 20ms/step - mae: 1.3434 - loss: 0.1681</span></span>
 <span><span class="co">## Epoch 2/3</span></span>
 <span><span class="co">## 32/32 - 0s - 2ms/step - mae: 1.3364 - loss: 0.1394</span></span>
 <span><span class="co">## Epoch 3/3</span></span>
-<span><span class="co">## 32/32 - 0s - 4ms/step - mae: 1.3286 - loss: 0.1148</span></span></code></pre>
+<span><span class="co">## 32/32 - 0s - 2ms/step - mae: 1.3286 - loss: 0.1148</span></span></code></pre>
 </div>
 <div class="section level2">
 <h2 id="providing-your-own-evaluation-step">Providing your own evaluation step<a class="anchor" aria-label="anchor" href="#providing-your-own-evaluation-step"></a>
@@ -570,7 +570,7 @@ <h2 id="wrapping-up-an-end-to-end-gan-example">Wrapping up: an end-to-end GAN ex
 <span>  <span class="fu">tfdatasets</span><span class="fu">::</span><span class="fu"><a href="https://rdrr.io/pkg/tfdatasets/man/dataset_take.html" class="external-link">dataset_take</a></span><span class="op">(</span><span class="va">dataset</span>, <span class="fl">100</span><span class="op">)</span>,</span>
 <span>  epochs <span class="op">=</span> <span class="fl">1</span></span>
 <span><span class="op">)</span></span></code></pre></div>
-<pre><code><span><span class="co">## 100/100 - 5s - 53ms/step - d_loss: 0.0000e+00 - g_loss: 0.0000e+00</span></span></code></pre>
+<pre><code><span><span class="co">## 100/100 - 5s - 51ms/step - d_loss: 0.0000e+00 - g_loss: 0.0000e+00</span></span></code></pre>
 <p>The ideas behind deep learning are simple, so why should their
 implementation be painful?</p>
 </div>
@@ -585,7 +585,7 @@ <h2 id="wrapping-up-an-end-to-end-gan-example">Wrapping up: an end-to-end GAN ex
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer>
diff --git a/docs/dev/articles/distributed_training_with_tensorflow.html b/docs/dev/articles/distributed_training_with_tensorflow.html
index 9b7517625..15ade0268 100644
--- a/docs/dev/articles/distributed_training_with_tensorflow.html
+++ b/docs/dev/articles/distributed_training_with_tensorflow.html
@@ -31,7 +31,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -264,15 +264,15 @@ <h2 id="single-host-multi-device-synchronous-training">Single-host, multi-device
 <span></span>
 <span><span class="op">}</span><span class="op">)</span></span></code></pre></div>
 <pre><code><span><span class="co">## Epoch 1/2</span></span>
-<span><span class="co">## 782/782 - 7s - 9ms/step - loss: nan - sparse_categorical_accuracy: nan - val_loss: nan - val_sparse_categorical_accuracy: nan</span></span>
+<span><span class="co">## 782/782 - 6s - 7ms/step - loss: 0.2616 - sparse_categorical_accuracy: 0.9232 - val_loss: 0.1510 - val_sparse_categorical_accuracy: 0.9524</span></span>
 <span><span class="co">## Epoch 2/2</span></span>
-<span><span class="co">## 782/782 - 5s - 7ms/step - loss: nan - sparse_categorical_accuracy: nan - val_loss: nan - val_sparse_categorical_accuracy: nan</span></span>
-<span><span class="co">## 157/157 - 1s - 5ms/step - loss: nan - sparse_categorical_accuracy: nan</span></span></code></pre>
+<span><span class="co">## 782/782 - 4s - 5ms/step - loss: 0.1044 - sparse_categorical_accuracy: 0.9682 - val_loss: 0.1004 - val_sparse_categorical_accuracy: 0.9675</span></span>
+<span><span class="co">## 157/157 - 0s - 3ms/step - loss: 0.1004 - sparse_categorical_accuracy: 0.9683</span></span></code></pre>
 <pre><code><span><span class="co">## $loss</span></span>
-<span><span class="co">## [1] NaN</span></span>
+<span><span class="co">## [1] 0.1004203</span></span>
 <span><span class="co">##</span></span>
 <span><span class="co">## $sparse_categorical_accuracy</span></span>
-<span><span class="co">## [1] NaN</span></span></code></pre>
+<span><span class="co">## [1] 0.9683</span></span></code></pre>
 </div>
 <div class="section level2">
 <h2 id="using-callbacks-to-ensure-fault-tolerance">Using callbacks to ensure fault tolerance<a class="anchor" aria-label="anchor" href="#using-callbacks-to-ensure-fault-tolerance"></a>
@@ -337,11 +337,11 @@ <h2 id="using-callbacks-to-ensure-fault-tolerance">Using callbacks to ensure fau
 <span></span>
 <span><span class="co"># Running the first time creates the model</span></span>
 <span><span class="fu">run_training</span><span class="op">(</span>epochs <span class="op">=</span> <span class="fl">1</span><span class="op">)</span></span></code></pre></div>
-<pre><code><span><span class="co">## 782/782 - 5s - 7ms/step - loss: nan - sparse_categorical_accuracy: nan - val_loss: nan - val_sparse_categorical_accuracy: nan</span></span></code></pre>
+<pre><code><span><span class="co">## 782/782 - 5s - 6ms/step - loss: 0.2575 - sparse_categorical_accuracy: 0.9237 - val_loss: 0.1428 - val_sparse_categorical_accuracy: 0.9537</span></span></code></pre>
 <div class="sourceCode" id="cb10"><pre class="downlit sourceCode r">
 <code class="sourceCode R"><span><span class="co"># Calling the same function again will resume from where we left off</span></span>
 <span><span class="fu">run_training</span><span class="op">(</span>epochs <span class="op">=</span> <span class="fl">1</span><span class="op">)</span></span></code></pre></div>
-<pre><code><span><span class="co">## 782/782 - 6s - 7ms/step - loss: nan - sparse_categorical_accuracy: nan - val_loss: nan - val_sparse_categorical_accuracy: nan</span></span></code></pre>
+<pre><code><span><span class="co">## 782/782 - 5s - 6ms/step - loss: 0.1011 - sparse_categorical_accuracy: 0.9693 - val_loss: 0.1185 - val_sparse_categorical_accuracy: 0.9602</span></span></code></pre>
 </div>
 <div class="section level2">
 <h2 id="tfdata-performance-tips">
@@ -388,7 +388,7 @@ <h2 id="tfdata-performance-tips">
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer>
diff --git a/docs/dev/articles/distribution.html b/docs/dev/articles/distribution.html
index 4e312a1fb..3f1a8c610 100644
--- a/docs/dev/articles/distribution.html
+++ b/docs/dev/articles/distribution.html
@@ -31,7 +31,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -254,16 +254,16 @@ <h2 id="dataparallel">DataParallel<a class="anchor" aria-label="anchor" href="#d
 <span><span class="va">model</span> <span class="op">|&gt;</span> <span class="fu"><a href="https://generics.r-lib.org/reference/compile.html" class="external-link">compile</a></span><span class="op">(</span>loss <span class="op">=</span> <span class="st">"mse"</span><span class="op">)</span></span>
 <span><span class="va">model</span> <span class="op">|&gt;</span> <span class="fu"><a href="https://generics.r-lib.org/reference/fit.html" class="external-link">fit</a></span><span class="op">(</span><span class="va">dataset</span>, epochs <span class="op">=</span> <span class="fl">3</span><span class="op">)</span></span></code></pre></div>
 <pre><code><span><span class="co">## Epoch 1/3</span></span>
-<span><span class="co">## 8/8 - 0s - 40ms/step - loss: 1.1536</span></span>
+<span><span class="co">## 8/8 - 0s - 42ms/step - loss: 1.1575</span></span>
 <span><span class="co">## Epoch 2/3</span></span>
-<span><span class="co">## 8/8 - 0s - 5ms/step - loss: 1.0540</span></span>
+<span><span class="co">## 8/8 - 0s - 4ms/step - loss: 1.0557</span></span>
 <span><span class="co">## Epoch 3/3</span></span>
-<span><span class="co">## 8/8 - 0s - 6ms/step - loss: 1.0072</span></span></code></pre>
+<span><span class="co">## 8/8 - 0s - 4ms/step - loss: 1.0179</span></span></code></pre>
 <div class="sourceCode" id="cb8"><pre class="downlit sourceCode r">
 <code class="sourceCode R"><span><span class="va">model</span> <span class="op">|&gt;</span> <span class="fu"><a href="https://rdrr.io/pkg/tensorflow/man/evaluate.html" class="external-link">evaluate</a></span><span class="op">(</span><span class="va">dataset</span><span class="op">)</span></span></code></pre></div>
-<pre><code><span><span class="co">## 8/8 - 0s - 9ms/step - loss: 0.9620</span></span></code></pre>
+<pre><code><span><span class="co">## 8/8 - 0s - 9ms/step - loss: 0.9707</span></span></code></pre>
 <pre><code><span><span class="co">## $loss</span></span>
-<span><span class="co">## [1] 0.9620273</span></span></code></pre>
+<span><span class="co">## [1] 0.9707004</span></span></code></pre>
 </div>
 <div class="section level2">
 <h2 id="modelparallel-and-layoutmap">
@@ -389,16 +389,16 @@ <h2 id="modelparallel-and-layoutmap">
 <span><span class="va">model</span> <span class="op">|&gt;</span> <span class="fu"><a href="https://generics.r-lib.org/reference/compile.html" class="external-link">compile</a></span><span class="op">(</span>loss <span class="op">=</span> <span class="st">"mse"</span><span class="op">)</span></span>
 <span><span class="va">model</span> <span class="op">|&gt;</span> <span class="fu"><a href="https://generics.r-lib.org/reference/fit.html" class="external-link">fit</a></span><span class="op">(</span><span class="va">dataset</span>, epochs <span class="op">=</span> <span class="fl">3</span><span class="op">)</span></span></code></pre></div>
 <pre><code><span><span class="co">## Epoch 1/3</span></span>
-<span><span class="co">## 8/8 - 0s - 46ms/step - loss: 1.1676</span></span>
+<span><span class="co">## 8/8 - 0s - 48ms/step - loss: 1.1454</span></span>
 <span><span class="co">## Epoch 2/3</span></span>
-<span><span class="co">## 8/8 - 0s - 4ms/step - loss: 1.1134</span></span>
+<span><span class="co">## 8/8 - 0s - 2ms/step - loss: 1.0527</span></span>
 <span><span class="co">## Epoch 3/3</span></span>
-<span><span class="co">## 8/8 - 0s - 5ms/step - loss: 1.1034</span></span></code></pre>
+<span><span class="co">## 8/8 - 0s - 2ms/step - loss: 1.0115</span></span></code></pre>
 <div class="sourceCode" id="cb22"><pre class="downlit sourceCode r">
 <code class="sourceCode R"><span><span class="va">model</span> <span class="op">|&gt;</span> <span class="fu"><a href="https://rdrr.io/pkg/tensorflow/man/evaluate.html" class="external-link">evaluate</a></span><span class="op">(</span><span class="va">dataset</span><span class="op">)</span></span></code></pre></div>
-<pre><code><span><span class="co">## 8/8 - 0s - 8ms/step - loss: 1.0676</span></span></code></pre>
+<pre><code><span><span class="co">## 8/8 - 0s - 10ms/step - loss: 0.9674</span></span></code></pre>
 <pre><code><span><span class="co">## $loss</span></span>
-<span><span class="co">## [1] 1.067567</span></span></code></pre>
+<span><span class="co">## [1] 0.9673891</span></span></code></pre>
 <p>It is also easy to change the mesh structure to tune the computation
 between more data parallel or model parallel. You can do this by
 adjusting the shape of the mesh. And no changes are needed for any other
@@ -452,7 +452,7 @@ <h3 id="further-reading">Further reading<a class="anchor" aria-label="anchor" hr
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer>
diff --git a/docs/dev/articles/examples/index.html b/docs/dev/articles/examples/index.html
index c5504c81b..3a4112bfa 100644
--- a/docs/dev/articles/examples/index.html
+++ b/docs/dev/articles/examples/index.html
@@ -29,7 +29,7 @@
 
     <a class="navbar-brand me-2" href="../../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -210,7 +210,7 @@ <h5 class="card-header bg-transparent">Image segmentation with a U-Net-like arch
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer>
diff --git a/docs/dev/articles/examples/nlp/neural_machine_translation_with_transformer.html b/docs/dev/articles/examples/nlp/neural_machine_translation_with_transformer.html
index c418ac647..a9c5254a2 100644
--- a/docs/dev/articles/examples/nlp/neural_machine_translation_with_transformer.html
+++ b/docs/dev/articles/examples/nlp/neural_machine_translation_with_transformer.html
@@ -31,7 +31,7 @@
 
     <a class="navbar-brand me-2" href="../../../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -587,7 +587,7 @@ <h2 id="training-our-model">Training our model<a class="anchor" aria-label="anch
 <span></span>
 <span><span class="va">transformer</span> <span class="op">|&gt;</span> <span class="fu"><a href="https://generics.r-lib.org/reference/fit.html" class="external-link">fit</a></span><span class="op">(</span><span class="va">train_ds</span>, epochs <span class="op">=</span> <span class="va">epochs</span>,</span>
 <span>                   validation_data <span class="op">=</span> <span class="va">val_ds</span><span class="op">)</span></span></code></pre></div>
-<pre><code><span><span class="co">## 1297/1297 - 61s - 47ms/step - accuracy: 0.1590 - loss: 3.8930 - val_accuracy: 0.1533 - val_loss: 3.6121</span></span></code></pre>
+<pre><code><span><span class="co">## 1297/1297 - 55s - 42ms/step - accuracy: 0.1592 - loss: 3.8833 - val_accuracy: 0.1691 - val_loss: 3.1841</span></span></code></pre>
 </div>
 <div class="section level2">
 <h2 id="decoding-test-sentences">Decoding test sentences<a class="anchor" aria-label="anchor" href="#decoding-test-sentences"></a>
@@ -660,7 +660,7 @@ <h2 id="decoding-test-sentences">Decoding test sentences<a class="anchor" aria-l
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer>
diff --git a/docs/dev/articles/examples/nlp/text_classification_from_scratch.html b/docs/dev/articles/examples/nlp/text_classification_from_scratch.html
index a806d6a75..38dccefd3 100644
--- a/docs/dev/articles/examples/nlp/text_classification_from_scratch.html
+++ b/docs/dev/articles/examples/nlp/text_classification_from_scratch.html
@@ -31,7 +31,7 @@
 
     <a class="navbar-brand me-2" href="../../../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -396,23 +396,23 @@ <h2 id="train-the-model">Train the model<a class="anchor" aria-label="anchor" hr
 <span><span class="co"># Fit the model using the train and test datasets.</span></span>
 <span><span class="va">model</span> <span class="op">|&gt;</span> <span class="fu"><a href="https://generics.r-lib.org/reference/fit.html" class="external-link">fit</a></span><span class="op">(</span><span class="va">train_ds</span>, validation_data <span class="op">=</span> <span class="va">val_ds</span>, epochs <span class="op">=</span> <span class="va">epochs</span><span class="op">)</span></span></code></pre></div>
 <pre><code><span><span class="co">## Epoch 1/3</span></span>
-<span><span class="co">## 625/625 - 6s - 10ms/step - accuracy: 0.6931 - loss: 0.5276 - val_accuracy: 0.8588 - val_loss: 0.3292</span></span>
+<span><span class="co">## 625/625 - 5s - 8ms/step - accuracy: 0.6964 - loss: 0.5261 - val_accuracy: 0.8668 - val_loss: 0.3185</span></span>
 <span><span class="co">## Epoch 2/3</span></span>
-<span><span class="co">## 625/625 - 4s - 6ms/step - accuracy: 0.9050 - loss: 0.2397 - val_accuracy: 0.8762 - val_loss: 0.3251</span></span>
+<span><span class="co">## 625/625 - 2s - 3ms/step - accuracy: 0.9074 - loss: 0.2365 - val_accuracy: 0.8754 - val_loss: 0.3245</span></span>
 <span><span class="co">## Epoch 3/3</span></span>
-<span><span class="co">## 625/625 - 3s - 4ms/step - accuracy: 0.9579 - loss: 0.1206 - val_accuracy: 0.8586 - val_loss: 0.4119</span></span></code></pre>
+<span><span class="co">## 625/625 - 2s - 3ms/step - accuracy: 0.9551 - loss: 0.1220 - val_accuracy: 0.8284 - val_loss: 0.4862</span></span></code></pre>
 </div>
 <div class="section level2">
 <h2 id="evaluate-the-model-on-the-test-set">Evaluate the model on the test set<a class="anchor" aria-label="anchor" href="#evaluate-the-model-on-the-test-set"></a>
 </h2>
 <div class="sourceCode" id="cb34"><pre class="downlit sourceCode r">
 <code class="sourceCode R"><span><span class="va">model</span> <span class="op">|&gt;</span> <span class="fu"><a href="https://rdrr.io/pkg/tensorflow/man/evaluate.html" class="external-link">evaluate</a></span><span class="op">(</span><span class="va">test_ds</span><span class="op">)</span></span></code></pre></div>
-<pre><code><span><span class="co">## 782/782 - 1s - 2ms/step - accuracy: 0.8437 - loss: 0.4384</span></span></code></pre>
+<pre><code><span><span class="co">## 782/782 - 1s - 2ms/step - accuracy: 0.8115 - loss: 0.5299</span></span></code></pre>
 <pre><code><span><span class="co">## $accuracy</span></span>
-<span><span class="co">## [1] 0.84368</span></span>
+<span><span class="co">## [1] 0.81152</span></span>
 <span><span class="co">##</span></span>
 <span><span class="co">## $loss</span></span>
-<span><span class="co">## [1] 0.4383631</span></span></code></pre>
+<span><span class="co">## [1] 0.5299004</span></span></code></pre>
 </div>
 <div class="section level2">
 <h2 id="make-an-end-to-end-model">Make an end-to-end model<a class="anchor" aria-label="anchor" href="#make-an-end-to-end-model"></a>
@@ -437,12 +437,12 @@ <h2 id="make-an-end-to-end-model">Make an end-to-end model<a class="anchor" aria
 <span></span>
 <span><span class="co"># Test it with `raw_test_ds`, which yields raw strings</span></span>
 <span><span class="va">end_to_end_model</span> <span class="op">|&gt;</span> <span class="fu"><a href="https://rdrr.io/pkg/tensorflow/man/evaluate.html" class="external-link">evaluate</a></span><span class="op">(</span><span class="va">raw_test_ds</span><span class="op">)</span></span></code></pre></div>
-<pre><code><span><span class="co">## 782/782 - 3s - 4ms/step - accuracy: 0.8437 - loss: 0.4384</span></span></code></pre>
+<pre><code><span><span class="co">## 782/782 - 3s - 4ms/step - accuracy: 0.8115 - loss: 0.5299</span></span></code></pre>
 <pre><code><span><span class="co">## $accuracy</span></span>
-<span><span class="co">## [1] 0.84368</span></span>
+<span><span class="co">## [1] 0.81152</span></span>
 <span><span class="co">##</span></span>
 <span><span class="co">## $loss</span></span>
-<span><span class="co">## [1] 0.4383632</span></span></code></pre>
+<span><span class="co">## [1] 0.5299006</span></span></code></pre>
 </div>
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
     </nav></aside>
@@ -455,7 +455,7 @@ <h2 id="make-an-end-to-end-model">Make an end-to-end model<a class="anchor" aria
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer>
diff --git a/docs/dev/articles/examples/structured_data/imbalanced_classification.html b/docs/dev/articles/examples/structured_data/imbalanced_classification.html
index 16cb6a244..537794558 100644
--- a/docs/dev/articles/examples/structured_data/imbalanced_classification.html
+++ b/docs/dev/articles/examples/structured_data/imbalanced_classification.html
@@ -31,7 +31,7 @@
 
     <a class="navbar-brand me-2" href="../../../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -175,11 +175,11 @@ <h2 id="analyze-class-imbalance-in-the-targets">Analyze class imbalance in the t
 <span><span class="va">counts</span></span></code></pre></div>
 <pre><code><span><span class="co">##</span></span>
 <span><span class="co">##      0      1</span></span>
-<span><span class="co">## 227451    395</span></span></code></pre>
+<span><span class="co">## 227464    382</span></span></code></pre>
 <div class="sourceCode" id="cb11"><pre class="downlit sourceCode r">
 <code class="sourceCode R"><span><span class="fu"><a href="https://rdrr.io/r/base/cat.html" class="external-link">cat</a></span><span class="op">(</span><span class="fu"><a href="https://rdrr.io/r/base/sprintf.html" class="external-link">sprintf</a></span><span class="op">(</span><span class="st">"Number of positive samples in training data: %i (%.2f%% of total)"</span>,</span>
 <span>            <span class="va">counts</span><span class="op">[</span><span class="st">"1"</span><span class="op">]</span>, <span class="fl">100</span> <span class="op">*</span> <span class="va">counts</span><span class="op">[</span><span class="st">"1"</span><span class="op">]</span> <span class="op">/</span> <span class="fu"><a href="https://rdrr.io/r/base/sum.html" class="external-link">sum</a></span><span class="op">(</span><span class="va">counts</span><span class="op">)</span><span class="op">)</span><span class="op">)</span></span></code></pre></div>
-<pre><code><span><span class="co">## Number of positive samples in training data: 395 (0.17% of total)</span></span></code></pre>
+<pre><code><span><span class="co">## Number of positive samples in training data: 382 (0.17% of total)</span></span></code></pre>
 <div class="sourceCode" id="cb13"><pre class="downlit sourceCode r">
 <code class="sourceCode R"><span><span class="va">weight_for_0</span> <span class="op">=</span> <span class="fl">1</span> <span class="op">/</span> <span class="va">counts</span><span class="op">[</span><span class="st">"0"</span><span class="op">]</span></span>
 <span><span class="va">weight_for_1</span> <span class="op">=</span> <span class="fl">1</span> <span class="op">/</span> <span class="va">counts</span><span class="op">[</span><span class="st">"1"</span><span class="op">]</span></span></code></pre></div>
@@ -268,74 +268,74 @@ <h2 id="train-the-model-with-class_weight-argument">Train the model with <code>c
 <span>  verbose <span class="op">=</span> <span class="fl">2</span></span>
 <span><span class="op">)</span></span></code></pre></div>
 <pre><code><span><span class="co">## Epoch 1/30</span></span>
-<span><span class="co">## 112/112 - 5s - 41ms/step - fn: 36.0000 - fp: 30458.0000 - loss: 2.3118e-06 - precision: 0.0116 - recall: 0.9089 - tn: 196993.0000 - tp: 359.0000 - val_fn: 14.0000 - val_fp: 714.0000 - val_loss: 0.0906 - val_precision: 0.1041 - val_recall: 0.8557 - val_tn: 56150.0000 - val_tp: 83.0000</span></span>
+<span><span class="co">## 112/112 - 4s - 37ms/step - fn: 36.0000 - fp: 29717.0000 - loss: 2.3629e-06 - precision: 0.0115 - recall: 0.9058 - tn: 197747.0000 - tp: 346.0000 - val_fn: 12.0000 - val_fp: 1039.0000 - val_loss: 0.0984 - val_precision: 0.0862 - val_recall: 0.8909 - val_tn: 55812.0000 - val_tp: 98.0000</span></span>
 <span><span class="co">## Epoch 2/30</span></span>
-<span><span class="co">## 112/112 - 1s - 11ms/step - fn: 34.0000 - fp: 5698.0000 - loss: 1.4544e-06 - precision: 0.0596 - recall: 0.9139 - tn: 221753.0000 - tp: 361.0000 - val_fn: 6.0000 - val_fp: 2982.0000 - val_loss: 0.1611 - val_precision: 0.0296 - val_recall: 0.9381 - val_tn: 53882.0000 - val_tp: 91.0000</span></span>
+<span><span class="co">## 112/112 - 1s - 9ms/step - fn: 30.0000 - fp: 6694.0000 - loss: 1.3508e-06 - precision: 0.0500 - recall: 0.9215 - tn: 220770.0000 - tp: 352.0000 - val_fn: 8.0000 - val_fp: 2418.0000 - val_loss: 0.1390 - val_precision: 0.0405 - val_recall: 0.9273 - val_tn: 54433.0000 - val_tp: 102.0000</span></span>
 <span><span class="co">## Epoch 3/30</span></span>
-<span><span class="co">## 112/112 - 0s - 2ms/step - fn: 30.0000 - fp: 7494.0000 - loss: 1.1267e-06 - precision: 0.0464 - recall: 0.9241 - tn: 219957.0000 - tp: 365.0000 - val_fn: 9.0000 - val_fp: 2262.0000 - val_loss: 0.1346 - val_precision: 0.0374 - val_recall: 0.9072 - val_tn: 54602.0000 - val_tp: 88.0000</span></span>
+<span><span class="co">## 112/112 - 0s - 2ms/step - fn: 27.0000 - fp: 9781.0000 - loss: 1.3090e-06 - precision: 0.0350 - recall: 0.9293 - tn: 217683.0000 - tp: 355.0000 - val_fn: 10.0000 - val_fp: 984.0000 - val_loss: 0.1063 - val_precision: 0.0923 - val_recall: 0.9091 - val_tn: 55867.0000 - val_tp: 100.0000</span></span>
 <span><span class="co">## Epoch 4/30</span></span>
-<span><span class="co">## 112/112 - 0s - 2ms/step - fn: 25.0000 - fp: 5976.0000 - loss: 9.9649e-07 - precision: 0.0583 - recall: 0.9367 - tn: 221475.0000 - tp: 370.0000 - val_fn: 9.0000 - val_fp: 1593.0000 - val_loss: 0.0781 - val_precision: 0.0523 - val_recall: 0.9072 - val_tn: 55271.0000 - val_tp: 88.0000</span></span>
+<span><span class="co">## 112/112 - 0s - 2ms/step - fn: 32.0000 - fp: 6894.0000 - loss: 1.1099e-06 - precision: 0.0483 - recall: 0.9162 - tn: 220570.0000 - tp: 350.0000 - val_fn: 9.0000 - val_fp: 1557.0000 - val_loss: 0.0857 - val_precision: 0.0609 - val_recall: 0.9182 - val_tn: 55294.0000 - val_tp: 101.0000</span></span>
 <span><span class="co">## Epoch 5/30</span></span>
-<span><span class="co">## 112/112 - 0s - 2ms/step - fn: 22.0000 - fp: 7713.0000 - loss: 9.5664e-07 - precision: 0.0461 - recall: 0.9443 - tn: 219738.0000 - tp: 373.0000 - val_fn: 10.0000 - val_fp: 613.0000 - val_loss: 0.0444 - val_precision: 0.1243 - val_recall: 0.8969 - val_tn: 56251.0000 - val_tp: 87.0000</span></span>
+<span><span class="co">## 112/112 - 0s - 2ms/step - fn: 23.0000 - fp: 8892.0000 - loss: 9.2992e-07 - precision: 0.0388 - recall: 0.9398 - tn: 218572.0000 - tp: 359.0000 - val_fn: 7.0000 - val_fp: 1492.0000 - val_loss: 0.0665 - val_precision: 0.0646 - val_recall: 0.9364 - val_tn: 55359.0000 - val_tp: 103.0000</span></span>
 <span><span class="co">## Epoch 6/30</span></span>
-<span><span class="co">## 112/112 - 0s - 2ms/step - fn: 18.0000 - fp: 8517.0000 - loss: 9.6051e-07 - precision: 0.0424 - recall: 0.9544 - tn: 218934.0000 - tp: 377.0000 - val_fn: 12.0000 - val_fp: 1033.0000 - val_loss: 0.0643 - val_precision: 0.0760 - val_recall: 0.8763 - val_tn: 55831.0000 - val_tp: 85.0000</span></span>
+<span><span class="co">## 112/112 - 0s - 2ms/step - fn: 18.0000 - fp: 8209.0000 - loss: 7.7549e-07 - precision: 0.0425 - recall: 0.9529 - tn: 219255.0000 - tp: 364.0000 - val_fn: 7.0000 - val_fp: 2749.0000 - val_loss: 0.1010 - val_precision: 0.0361 - val_recall: 0.9364 - val_tn: 54102.0000 - val_tp: 103.0000</span></span>
 <span><span class="co">## Epoch 7/30</span></span>
-<span><span class="co">## 112/112 - 0s - 2ms/step - fn: 17.0000 - fp: 6409.0000 - loss: 7.3375e-07 - precision: 0.0557 - recall: 0.9570 - tn: 221042.0000 - tp: 378.0000 - val_fn: 10.0000 - val_fp: 1549.0000 - val_loss: 0.0697 - val_precision: 0.0532 - val_recall: 0.8969 - val_tn: 55315.0000 - val_tp: 87.0000</span></span>
+<span><span class="co">## 112/112 - 0s - 2ms/step - fn: 15.0000 - fp: 7189.0000 - loss: 6.9032e-07 - precision: 0.0486 - recall: 0.9607 - tn: 220275.0000 - tp: 367.0000 - val_fn: 12.0000 - val_fp: 934.0000 - val_loss: 0.0539 - val_precision: 0.0950 - val_recall: 0.8909 - val_tn: 55917.0000 - val_tp: 98.0000</span></span>
 <span><span class="co">## Epoch 8/30</span></span>
-<span><span class="co">## 112/112 - 0s - 2ms/step - fn: 14.0000 - fp: 6080.0000 - loss: 6.5853e-07 - precision: 0.0590 - recall: 0.9646 - tn: 221371.0000 - tp: 381.0000 - val_fn: 10.0000 - val_fp: 2677.0000 - val_loss: 0.1178 - val_precision: 0.0315 - val_recall: 0.8969 - val_tn: 54187.0000 - val_tp: 87.0000</span></span>
+<span><span class="co">## 112/112 - 0s - 2ms/step - fn: 18.0000 - fp: 7759.0000 - loss: 8.2432e-07 - precision: 0.0448 - recall: 0.9529 - tn: 219705.0000 - tp: 364.0000 - val_fn: 10.0000 - val_fp: 1481.0000 - val_loss: 0.0777 - val_precision: 0.0633 - val_recall: 0.9091 - val_tn: 55370.0000 - val_tp: 100.0000</span></span>
 <span><span class="co">## Epoch 9/30</span></span>
-<span><span class="co">## 112/112 - 0s - 2ms/step - fn: 11.0000 - fp: 6592.0000 - loss: 5.9467e-07 - precision: 0.0550 - recall: 0.9722 - tn: 220859.0000 - tp: 384.0000 - val_fn: 9.0000 - val_fp: 2094.0000 - val_loss: 0.0806 - val_precision: 0.0403 - val_recall: 0.9072 - val_tn: 54770.0000 - val_tp: 88.0000</span></span>
+<span><span class="co">## 112/112 - 0s - 2ms/step - fn: 16.0000 - fp: 7565.0000 - loss: 7.2956e-07 - precision: 0.0461 - recall: 0.9581 - tn: 219899.0000 - tp: 366.0000 - val_fn: 8.0000 - val_fp: 2075.0000 - val_loss: 0.0718 - val_precision: 0.0469 - val_recall: 0.9273 - val_tn: 54776.0000 - val_tp: 102.0000</span></span>
 <span><span class="co">## Epoch 10/30</span></span>
-<span><span class="co">## 112/112 - 0s - 2ms/step - fn: 7.0000 - fp: 6334.0000 - loss: 5.3129e-07 - precision: 0.0577 - recall: 0.9823 - tn: 221117.0000 - tp: 388.0000 - val_fn: 10.0000 - val_fp: 850.0000 - val_loss: 0.0396 - val_precision: 0.0928 - val_recall: 0.8969 - val_tn: 56014.0000 - val_tp: 87.0000</span></span>
+<span><span class="co">## 112/112 - 0s - 2ms/step - fn: 10.0000 - fp: 7374.0000 - loss: 5.9522e-07 - precision: 0.0480 - recall: 0.9738 - tn: 220090.0000 - tp: 372.0000 - val_fn: 11.0000 - val_fp: 995.0000 - val_loss: 0.0425 - val_precision: 0.0905 - val_recall: 0.9000 - val_tn: 55856.0000 - val_tp: 99.0000</span></span>
 <span><span class="co">## Epoch 11/30</span></span>
-<span><span class="co">## 112/112 - 0s - 2ms/step - fn: 8.0000 - fp: 5922.0000 - loss: 5.0009e-07 - precision: 0.0613 - recall: 0.9797 - tn: 221529.0000 - tp: 387.0000 - val_fn: 12.0000 - val_fp: 1488.0000 - val_loss: 0.0506 - val_precision: 0.0540 - val_recall: 0.8763 - val_tn: 55376.0000 - val_tp: 85.0000</span></span>
+<span><span class="co">## 112/112 - 0s - 2ms/step - fn: 10.0000 - fp: 6586.0000 - loss: 6.2194e-07 - precision: 0.0535 - recall: 0.9738 - tn: 220878.0000 - tp: 372.0000 - val_fn: 9.0000 - val_fp: 5097.0000 - val_loss: 0.2103 - val_precision: 0.0194 - val_recall: 0.9182 - val_tn: 51754.0000 - val_tp: 101.0000</span></span>
 <span><span class="co">## Epoch 12/30</span></span>
-<span><span class="co">## 112/112 - 0s - 2ms/step - fn: 8.0000 - fp: 7243.0000 - loss: 5.7125e-07 - precision: 0.0507 - recall: 0.9797 - tn: 220208.0000 - tp: 387.0000 - val_fn: 11.0000 - val_fp: 1014.0000 - val_loss: 0.0414 - val_precision: 0.0782 - val_recall: 0.8866 - val_tn: 55850.0000 - val_tp: 86.0000</span></span>
+<span><span class="co">## 112/112 - 0s - 2ms/step - fn: 6.0000 - fp: 8285.0000 - loss: 6.0991e-07 - precision: 0.0434 - recall: 0.9843 - tn: 219179.0000 - tp: 376.0000 - val_fn: 9.0000 - val_fp: 1926.0000 - val_loss: 0.0781 - val_precision: 0.0498 - val_recall: 0.9182 - val_tn: 54925.0000 - val_tp: 101.0000</span></span>
 <span><span class="co">## Epoch 13/30</span></span>
-<span><span class="co">## 112/112 - 0s - 2ms/step - fn: 6.0000 - fp: 4141.0000 - loss: 3.7800e-07 - precision: 0.0859 - recall: 0.9848 - tn: 223310.0000 - tp: 389.0000 - val_fn: 8.0000 - val_fp: 2491.0000 - val_loss: 0.0986 - val_precision: 0.0345 - val_recall: 0.9175 - val_tn: 54373.0000 - val_tp: 89.0000</span></span>
+<span><span class="co">## 112/112 - 0s - 2ms/step - fn: 7.0000 - fp: 6864.0000 - loss: 5.7087e-07 - precision: 0.0518 - recall: 0.9817 - tn: 220600.0000 - tp: 375.0000 - val_fn: 8.0000 - val_fp: 1900.0000 - val_loss: 0.0837 - val_precision: 0.0509 - val_recall: 0.9273 - val_tn: 54951.0000 - val_tp: 102.0000</span></span>
 <span><span class="co">## Epoch 14/30</span></span>
-<span><span class="co">## 112/112 - 0s - 2ms/step - fn: 6.0000 - fp: 6118.0000 - loss: 5.4537e-07 - precision: 0.0598 - recall: 0.9848 - tn: 221333.0000 - tp: 389.0000 - val_fn: 9.0000 - val_fp: 2488.0000 - val_loss: 0.1019 - val_precision: 0.0342 - val_recall: 0.9072 - val_tn: 54376.0000 - val_tp: 88.0000</span></span>
+<span><span class="co">## 112/112 - 0s - 2ms/step - fn: 7.0000 - fp: 7663.0000 - loss: 5.7664e-07 - precision: 0.0467 - recall: 0.9817 - tn: 219801.0000 - tp: 375.0000 - val_fn: 8.0000 - val_fp: 1540.0000 - val_loss: 0.0656 - val_precision: 0.0621 - val_recall: 0.9273 - val_tn: 55311.0000 - val_tp: 102.0000</span></span>
 <span><span class="co">## Epoch 15/30</span></span>
-<span><span class="co">## 112/112 - 0s - 2ms/step - fn: 4.0000 - fp: 4646.0000 - loss: 4.0941e-07 - precision: 0.0776 - recall: 0.9899 - tn: 222805.0000 - tp: 391.0000 - val_fn: 8.0000 - val_fp: 2949.0000 - val_loss: 0.1263 - val_precision: 0.0293 - val_recall: 0.9175 - val_tn: 53915.0000 - val_tp: 89.0000</span></span>
+<span><span class="co">## 112/112 - 0s - 2ms/step - fn: 6.0000 - fp: 4607.0000 - loss: 3.9004e-07 - precision: 0.0755 - recall: 0.9843 - tn: 222857.0000 - tp: 376.0000 - val_fn: 11.0000 - val_fp: 2546.0000 - val_loss: 0.0994 - val_precision: 0.0374 - val_recall: 0.9000 - val_tn: 54305.0000 - val_tp: 99.0000</span></span>
 <span><span class="co">## Epoch 16/30</span></span>
-<span><span class="co">## 112/112 - 0s - 2ms/step - fn: 10.0000 - fp: 7978.0000 - loss: 8.1907e-07 - precision: 0.0460 - recall: 0.9747 - tn: 219473.0000 - tp: 385.0000 - val_fn: 8.0000 - val_fp: 3430.0000 - val_loss: 0.2915 - val_precision: 0.0253 - val_recall: 0.9175 - val_tn: 53434.0000 - val_tp: 89.0000</span></span>
+<span><span class="co">## 112/112 - 0s - 2ms/step - fn: 6.0000 - fp: 6195.0000 - loss: 5.1274e-07 - precision: 0.0572 - recall: 0.9843 - tn: 221269.0000 - tp: 376.0000 - val_fn: 9.0000 - val_fp: 1652.0000 - val_loss: 0.0693 - val_precision: 0.0576 - val_recall: 0.9182 - val_tn: 55199.0000 - val_tp: 101.0000</span></span>
 <span><span class="co">## Epoch 17/30</span></span>
-<span><span class="co">## 112/112 - 0s - 2ms/step - fn: 23.0000 - fp: 12673.0000 - loss: 3.1681e-06 - precision: 0.0285 - recall: 0.9418 - tn: 214778.0000 - tp: 372.0000 - val_fn: 12.0000 - val_fp: 3416.0000 - val_loss: 0.7861 - val_precision: 0.0243 - val_recall: 0.8763 - val_tn: 53448.0000 - val_tp: 85.0000</span></span>
+<span><span class="co">## 112/112 - 0s - 2ms/step - fn: 6.0000 - fp: 5367.0000 - loss: 4.6364e-07 - precision: 0.0655 - recall: 0.9843 - tn: 222097.0000 - tp: 376.0000 - val_fn: 10.0000 - val_fp: 1378.0000 - val_loss: 0.0575 - val_precision: 0.0677 - val_recall: 0.9091 - val_tn: 55473.0000 - val_tp: 100.0000</span></span>
 <span><span class="co">## Epoch 18/30</span></span>
-<span><span class="co">## 112/112 - 0s - 2ms/step - fn: 23.0000 - fp: 11244.0000 - loss: 2.3250e-06 - precision: 0.0320 - recall: 0.9418 - tn: 216207.0000 - tp: 372.0000 - val_fn: 11.0000 - val_fp: 1717.0000 - val_loss: 0.1895 - val_precision: 0.0477 - val_recall: 0.8866 - val_tn: 55147.0000 - val_tp: 86.0000</span></span>
+<span><span class="co">## 112/112 - 0s - 2ms/step - fn: 2.0000 - fp: 3696.0000 - loss: 2.9266e-07 - precision: 0.0932 - recall: 0.9948 - tn: 223768.0000 - tp: 380.0000 - val_fn: 11.0000 - val_fp: 824.0000 - val_loss: 0.0375 - val_precision: 0.1073 - val_recall: 0.9000 - val_tn: 56027.0000 - val_tp: 99.0000</span></span>
 <span><span class="co">## Epoch 19/30</span></span>
-<span><span class="co">## 112/112 - 0s - 2ms/step - fn: 16.0000 - fp: 8824.0000 - loss: 1.3502e-06 - precision: 0.0412 - recall: 0.9595 - tn: 218627.0000 - tp: 379.0000 - val_fn: 6.0000 - val_fp: 5161.0000 - val_loss: 0.1947 - val_precision: 0.0173 - val_recall: 0.9381 - val_tn: 51703.0000 - val_tp: 91.0000</span></span>
+<span><span class="co">## 112/112 - 0s - 2ms/step - fn: 3.0000 - fp: 3638.0000 - loss: 3.6945e-07 - precision: 0.0943 - recall: 0.9921 - tn: 223826.0000 - tp: 379.0000 - val_fn: 4.0000 - val_fp: 4639.0000 - val_loss: 0.5475 - val_precision: 0.0223 - val_recall: 0.9636 - val_tn: 52212.0000 - val_tp: 106.0000</span></span>
 <span><span class="co">## Epoch 20/30</span></span>
-<span><span class="co">## 112/112 - 0s - 2ms/step - fn: 10.0000 - fp: 8621.0000 - loss: 1.0916e-06 - precision: 0.0427 - recall: 0.9747 - tn: 218830.0000 - tp: 385.0000 - val_fn: 12.0000 - val_fp: 1107.0000 - val_loss: 0.1877 - val_precision: 0.0713 - val_recall: 0.8763 - val_tn: 55757.0000 - val_tp: 85.0000</span></span>
+<span><span class="co">## 112/112 - 0s - 2ms/step - fn: 26.0000 - fp: 11743.0000 - loss: 4.2168e-06 - precision: 0.0294 - recall: 0.9319 - tn: 215721.0000 - tp: 356.0000 - val_fn: 104.0000 - val_fp: 395.0000 - val_loss: 0.0507 - val_precision: 0.0150 - val_recall: 0.0545 - val_tn: 56456.0000 - val_tp: 6.0000</span></span>
 <span><span class="co">## Epoch 21/30</span></span>
-<span><span class="co">## 112/112 - 0s - 2ms/step - fn: 10.0000 - fp: 7096.0000 - loss: 1.0528e-06 - precision: 0.0515 - recall: 0.9747 - tn: 220355.0000 - tp: 385.0000 - val_fn: 11.0000 - val_fp: 1099.0000 - val_loss: 0.0539 - val_precision: 0.0726 - val_recall: 0.8866 - val_tn: 55765.0000 - val_tp: 86.0000</span></span>
+<span><span class="co">## 112/112 - 0s - 2ms/step - fn: 41.0000 - fp: 8860.0000 - loss: 5.8042e-06 - precision: 0.0371 - recall: 0.8927 - tn: 218604.0000 - tp: 341.0000 - val_fn: 10.0000 - val_fp: 2791.0000 - val_loss: 0.2183 - val_precision: 0.0346 - val_recall: 0.9091 - val_tn: 54060.0000 - val_tp: 100.0000</span></span>
 <span><span class="co">## Epoch 22/30</span></span>
-<span><span class="co">## 112/112 - 0s - 3ms/step - fn: 11.0000 - fp: 5263.0000 - loss: 5.9445e-07 - precision: 0.0680 - recall: 0.9722 - tn: 222188.0000 - tp: 384.0000 - val_fn: 10.0000 - val_fp: 1399.0000 - val_loss: 0.0632 - val_precision: 0.0585 - val_recall: 0.8969 - val_tn: 55465.0000 - val_tp: 87.0000</span></span>
+<span><span class="co">## 112/112 - 0s - 2ms/step - fn: 15.0000 - fp: 9031.0000 - loss: 1.0670e-06 - precision: 0.0391 - recall: 0.9607 - tn: 218433.0000 - tp: 367.0000 - val_fn: 10.0000 - val_fp: 1275.0000 - val_loss: 0.0764 - val_precision: 0.0727 - val_recall: 0.9091 - val_tn: 55576.0000 - val_tp: 100.0000</span></span>
 <span><span class="co">## Epoch 23/30</span></span>
-<span><span class="co">## 112/112 - 0s - 4ms/step - fn: 8.0000 - fp: 5412.0000 - loss: 6.1044e-07 - precision: 0.0667 - recall: 0.9797 - tn: 222039.0000 - tp: 387.0000 - val_fn: 12.0000 - val_fp: 873.0000 - val_loss: 0.0727 - val_precision: 0.0887 - val_recall: 0.8763 - val_tn: 55991.0000 - val_tp: 85.0000</span></span>
+<span><span class="co">## 112/112 - 0s - 2ms/step - fn: 10.0000 - fp: 5141.0000 - loss: 6.9193e-07 - precision: 0.0675 - recall: 0.9738 - tn: 222323.0000 - tp: 372.0000 - val_fn: 11.0000 - val_fp: 1872.0000 - val_loss: 0.0824 - val_precision: 0.0502 - val_recall: 0.9000 - val_tn: 54979.0000 - val_tp: 99.0000</span></span>
 <span><span class="co">## Epoch 24/30</span></span>
-<span><span class="co">## 112/112 - 0s - 3ms/step - fn: 5.0000 - fp: 3469.0000 - loss: 4.1501e-07 - precision: 0.1011 - recall: 0.9873 - tn: 223982.0000 - tp: 390.0000 - val_fn: 11.0000 - val_fp: 1369.0000 - val_loss: 0.0758 - val_precision: 0.0591 - val_recall: 0.8866 - val_tn: 55495.0000 - val_tp: 86.0000</span></span>
+<span><span class="co">## 112/112 - 0s - 2ms/step - fn: 2.0000 - fp: 4232.0000 - loss: 4.2135e-07 - precision: 0.0824 - recall: 0.9948 - tn: 223232.0000 - tp: 380.0000 - val_fn: 11.0000 - val_fp: 976.0000 - val_loss: 0.0488 - val_precision: 0.0921 - val_recall: 0.9000 - val_tn: 55875.0000 - val_tp: 99.0000</span></span>
 <span><span class="co">## Epoch 25/30</span></span>
-<span><span class="co">## 112/112 - 0s - 2ms/step - fn: 7.0000 - fp: 4489.0000 - loss: 6.9404e-07 - precision: 0.0796 - recall: 0.9823 - tn: 222962.0000 - tp: 388.0000 - val_fn: 11.0000 - val_fp: 1580.0000 - val_loss: 0.1499 - val_precision: 0.0516 - val_recall: 0.8866 - val_tn: 55284.0000 - val_tp: 86.0000</span></span>
+<span><span class="co">## 112/112 - 0s - 2ms/step - fn: 3.0000 - fp: 4149.0000 - loss: 3.7232e-07 - precision: 0.0837 - recall: 0.9921 - tn: 223315.0000 - tp: 379.0000 - val_fn: 13.0000 - val_fp: 359.0000 - val_loss: 0.0229 - val_precision: 0.2127 - val_recall: 0.8818 - val_tn: 56492.0000 - val_tp: 97.0000</span></span>
 <span><span class="co">## Epoch 26/30</span></span>
-<span><span class="co">## 112/112 - 0s - 2ms/step - fn: 4.0000 - fp: 3469.0000 - loss: 4.4795e-07 - precision: 0.1013 - recall: 0.9899 - tn: 223982.0000 - tp: 391.0000 - val_fn: 12.0000 - val_fp: 951.0000 - val_loss: 0.0502 - val_precision: 0.0820 - val_recall: 0.8763 - val_tn: 55913.0000 - val_tp: 85.0000</span></span>
+<span><span class="co">## 112/112 - 0s - 2ms/step - fn: 3.0000 - fp: 3757.0000 - loss: 3.5368e-07 - precision: 0.0916 - recall: 0.9921 - tn: 223707.0000 - tp: 379.0000 - val_fn: 12.0000 - val_fp: 1111.0000 - val_loss: 0.0569 - val_precision: 0.0811 - val_recall: 0.8909 - val_tn: 55740.0000 - val_tp: 98.0000</span></span>
 <span><span class="co">## Epoch 27/30</span></span>
-<span><span class="co">## 112/112 - 0s - 2ms/step - fn: 5.0000 - fp: 4854.0000 - loss: 7.1714e-07 - precision: 0.0744 - recall: 0.9873 - tn: 222597.0000 - tp: 390.0000 - val_fn: 13.0000 - val_fp: 1354.0000 - val_loss: 0.0818 - val_precision: 0.0584 - val_recall: 0.8660 - val_tn: 55510.0000 - val_tp: 84.0000</span></span>
+<span><span class="co">## 112/112 - 0s - 2ms/step - fn: 2.0000 - fp: 3220.0000 - loss: 2.5512e-07 - precision: 0.1056 - recall: 0.9948 - tn: 224244.0000 - tp: 380.0000 - val_fn: 11.0000 - val_fp: 839.0000 - val_loss: 0.0396 - val_precision: 0.1055 - val_recall: 0.9000 - val_tn: 56012.0000 - val_tp: 99.0000</span></span>
 <span><span class="co">## Epoch 28/30</span></span>
-<span><span class="co">## 112/112 - 0s - 3ms/step - fn: 6.0000 - fp: 4904.0000 - loss: 4.4107e-07 - precision: 0.0735 - recall: 0.9848 - tn: 222547.0000 - tp: 389.0000 - val_fn: 12.0000 - val_fp: 651.0000 - val_loss: 0.0420 - val_precision: 0.1155 - val_recall: 0.8763 - val_tn: 56213.0000 - val_tp: 85.0000</span></span>
+<span><span class="co">## 112/112 - 0s - 2ms/step - fn: 0.0000e+00 - fp: 2287.0000 - loss: 1.8121e-07 - precision: 0.1431 - recall: 1.0000 - tn: 225177.0000 - tp: 382.0000 - val_fn: 12.0000 - val_fp: 425.0000 - val_loss: 0.0272 - val_precision: 0.1874 - val_recall: 0.8909 - val_tn: 56426.0000 - val_tp: 98.0000</span></span>
 <span><span class="co">## Epoch 29/30</span></span>
-<span><span class="co">## 112/112 - 0s - 3ms/step - fn: 4.0000 - fp: 3909.0000 - loss: 4.2965e-07 - precision: 0.0909 - recall: 0.9899 - tn: 223542.0000 - tp: 391.0000 - val_fn: 12.0000 - val_fp: 817.0000 - val_loss: 0.0405 - val_precision: 0.0942 - val_recall: 0.8763 - val_tn: 56047.0000 - val_tp: 85.0000</span></span>
+<span><span class="co">## 112/112 - 0s - 2ms/step - fn: 0.0000e+00 - fp: 1435.0000 - loss: 1.3034e-07 - precision: 0.2102 - recall: 1.0000 - tn: 226029.0000 - tp: 382.0000 - val_fn: 13.0000 - val_fp: 321.0000 - val_loss: 0.0250 - val_precision: 0.2321 - val_recall: 0.8818 - val_tn: 56530.0000 - val_tp: 97.0000</span></span>
 <span><span class="co">## Epoch 30/30</span></span>
-<span><span class="co">## 112/112 - 0s - 2ms/step - fn: 2.0000 - fp: 2985.0000 - loss: 2.8492e-07 - precision: 0.1163 - recall: 0.9949 - tn: 224466.0000 - tp: 393.0000 - val_fn: 12.0000 - val_fp: 1040.0000 - val_loss: 0.0593 - val_precision: 0.0756 - val_recall: 0.8763 - val_tn: 55824.0000 - val_tp: 85.0000</span></span></code></pre>
+<span><span class="co">## 112/112 - 0s - 2ms/step - fn: 0.0000e+00 - fp: 1365.0000 - loss: 1.2132e-07 - precision: 0.2187 - recall: 1.0000 - tn: 226099.0000 - tp: 382.0000 - val_fn: 12.0000 - val_fp: 241.0000 - val_loss: 0.0183 - val_precision: 0.2891 - val_recall: 0.8909 - val_tn: 56610.0000 - val_tp: 98.0000</span></span></code></pre>
 <div class="sourceCode" id="cb19"><pre class="downlit sourceCode r">
 <code class="sourceCode R"><span><span class="va">val_pred</span> <span class="op">&lt;-</span> <span class="va">model</span> <span class="op"><a href="../../../reference/pipe.html">%&gt;%</a></span></span>
 <span>  <span class="fu"><a href="https://rdrr.io/r/stats/predict.html" class="external-link">predict</a></span><span class="op">(</span><span class="va">val_features</span><span class="op">)</span> <span class="op"><a href="../../../reference/pipe.html">%&gt;%</a></span></span>
 <span>  <span class="op">{</span> <span class="fu"><a href="https://rdrr.io/r/base/integer.html" class="external-link">as.integer</a></span><span class="op">(</span><span class="va">.</span> <span class="op">&gt;</span> <span class="fl">0.5</span><span class="op">)</span> <span class="op">}</span></span></code></pre></div>
-<pre><code><span><span class="co">## 1781/1781 - 1s - 436us/step</span></span></code></pre>
+<pre><code><span><span class="co">## 1781/1781 - 1s - 427us/step</span></span></code></pre>
 <div class="sourceCode" id="cb21"><pre class="downlit sourceCode r">
 <code class="sourceCode R"><span><span class="va">pred_correct</span> <span class="op">&lt;-</span> <span class="va">val_df</span><span class="op">$</span><span class="va">Class</span> <span class="op">==</span> <span class="va">val_pred</span></span>
 <span><span class="fu"><a href="https://rdrr.io/r/base/cat.html" class="external-link">cat</a></span><span class="op">(</span><span class="fu"><a href="https://rdrr.io/r/base/sprintf.html" class="external-link">sprintf</a></span><span class="op">(</span><span class="st">"Validation accuracy: %.2f"</span>, <span class="fu"><a href="https://rdrr.io/r/base/mean.html" class="external-link">mean</a></span><span class="op">(</span><span class="va">pred_correct</span><span class="op">)</span><span class="op">)</span><span class="op">)</span></span></code></pre></div>
-<pre><code><span><span class="co">## Validation accuracy: 0.98</span></span></code></pre>
+<pre><code><span><span class="co">## Validation accuracy: 1.00</span></span></code></pre>
 <div class="sourceCode" id="cb23"><pre class="downlit sourceCode r">
 <code class="sourceCode R"><span><span class="va">fraudulent</span> <span class="op">&lt;-</span> <span class="va">val_df</span><span class="op">$</span><span class="va">Class</span> <span class="op">==</span> <span class="fl">1</span></span>
 <span></span>
@@ -349,10 +349,9 @@ <h2 id="conclusions">Conclusions<a class="anchor" aria-label="anchor" href="#con
 <p>At the end of training, out of 56,961 validation transactions, we
 are:</p>
 <ul>
-<li>Correctly identifying 85 of them as fraudulent</li>
+<li>Correctly identifying 98 of them as fraudulent</li>
 <li>Missing 12 fraudulent transactions</li>
-<li>At the cost of incorrectly flagging 1,040 legitimate
-transactions</li>
+<li>At the cost of incorrectly flagging 241 legitimate transactions</li>
 </ul>
 <p>In the real world, one would put an even higher weight on class 1, so
 as to reflect that False Negatives are more costly than False
@@ -375,7 +374,7 @@ <h2 id="conclusions">Conclusions<a class="anchor" aria-label="anchor" href="#con
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer>
diff --git a/docs/dev/articles/examples/structured_data/structured_data_classification_with_feature_space.html b/docs/dev/articles/examples/structured_data/structured_data_classification_with_feature_space.html
index 2ff0d0ee3..4de2a39f2 100644
--- a/docs/dev/articles/examples/structured_data/structured_data_classification_with_feature_space.html
+++ b/docs/dev/articles/examples/structured_data/structured_data_classification_with_feature_space.html
@@ -31,7 +31,7 @@
 
     <a class="navbar-brand me-2" href="../../../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -312,20 +312,20 @@ <h2 id="preparing-the-data">Preparing the data<a class="anchor" aria-label="anch
 <span><span class="fu"><a href="https://rdrr.io/r/base/cat.html" class="external-link">cat</a></span><span class="op">(</span><span class="st">"Input: "</span><span class="op">)</span>; <span class="fu"><a href="https://rdrr.io/r/utils/str.html" class="external-link">str</a></span><span class="op">(</span><span class="va">x</span><span class="op">)</span></span>
 <span><span class="fu"><a href="https://rdrr.io/r/base/cat.html" class="external-link">cat</a></span><span class="op">(</span><span class="st">"Target: "</span><span class="op">)</span>; <span class="fu"><a href="https://rdrr.io/r/utils/str.html" class="external-link">str</a></span><span class="op">(</span><span class="va">y</span><span class="op">)</span></span></code></pre></div>
 <pre><code><span><span class="co">## Input: List of 13</span></span>
-<span><span class="co">##  $ age     :&lt;tf.Tensor: shape=(), dtype=int32, numpy=63&gt;</span></span>
-<span><span class="co">##  $ sex     :&lt;tf.Tensor: shape=(), dtype=int32, numpy=0&gt;</span></span>
+<span><span class="co">##  $ age     :&lt;tf.Tensor: shape=(), dtype=int32, numpy=68&gt;</span></span>
+<span><span class="co">##  $ sex     :&lt;tf.Tensor: shape=(), dtype=int32, numpy=1&gt;</span></span>
 <span><span class="co">##  $ cp      :&lt;tf.Tensor: shape=(), dtype=int32, numpy=4&gt;</span></span>
-<span><span class="co">##  $ trestbps:&lt;tf.Tensor: shape=(), dtype=int32, numpy=124&gt;</span></span>
-<span><span class="co">##  $ chol    :&lt;tf.Tensor: shape=(), dtype=int32, numpy=197&gt;</span></span>
-<span><span class="co">##  $ fbs     :&lt;tf.Tensor: shape=(), dtype=int32, numpy=0&gt;</span></span>
+<span><span class="co">##  $ trestbps:&lt;tf.Tensor: shape=(), dtype=int32, numpy=144&gt;</span></span>
+<span><span class="co">##  $ chol    :&lt;tf.Tensor: shape=(), dtype=int32, numpy=193&gt;</span></span>
+<span><span class="co">##  $ fbs     :&lt;tf.Tensor: shape=(), dtype=int32, numpy=1&gt;</span></span>
 <span><span class="co">##  $ restecg :&lt;tf.Tensor: shape=(), dtype=int32, numpy=0&gt;</span></span>
-<span><span class="co">##  $ thalach :&lt;tf.Tensor: shape=(), dtype=int32, numpy=136&gt;</span></span>
-<span><span class="co">##  $ exang   :&lt;tf.Tensor: shape=(), dtype=int32, numpy=1&gt;</span></span>
-<span><span class="co">##  $ oldpeak :&lt;tf.Tensor: shape=(), dtype=float32, numpy=0.0&gt;</span></span>
+<span><span class="co">##  $ thalach :&lt;tf.Tensor: shape=(), dtype=int32, numpy=141&gt;</span></span>
+<span><span class="co">##  $ exang   :&lt;tf.Tensor: shape=(), dtype=int32, numpy=0&gt;</span></span>
+<span><span class="co">##  $ oldpeak :&lt;tf.Tensor: shape=(), dtype=float32, numpy=3.4000000953674316&gt;</span></span>
 <span><span class="co">##  $ slope   :&lt;tf.Tensor: shape=(), dtype=int32, numpy=2&gt;</span></span>
-<span><span class="co">##  $ ca      :&lt;tf.Tensor: shape=(), dtype=int32, numpy=0&gt;</span></span>
-<span><span class="co">##  $ thal    :&lt;tf.Tensor: shape=(), dtype=string, numpy=b'normal'&gt;</span></span>
-<span><span class="co">## Target: &lt;tf.Tensor: shape=(), dtype=int32, numpy=0&gt;</span></span></code></pre>
+<span><span class="co">##  $ ca      :&lt;tf.Tensor: shape=(), dtype=int32, numpy=2&gt;</span></span>
+<span><span class="co">##  $ thal    :&lt;tf.Tensor: shape=(), dtype=string, numpy=b'reversible'&gt;</span></span>
+<span><span class="co">## Target: &lt;tf.Tensor: shape=(), dtype=int32, numpy=1&gt;</span></span></code></pre>
 <p>Let’s batch the datasets:</p>
 <div class="sourceCode" id="cb12"><pre class="downlit sourceCode r">
 <code class="sourceCode R"><span><span class="va">train_ds</span> <span class="op">&lt;-</span> <span class="va">train_ds</span> <span class="op">|&gt;</span> <span class="fu"><a href="https://rdrr.io/pkg/tfdatasets/man/dataset_batch.html" class="external-link">dataset_batch</a></span><span class="op">(</span><span class="fl">32</span><span class="op">)</span></span>
@@ -471,7 +471,7 @@ <h2 id="adapt-the-featurespace-to-the-training-data">Adapt the <code>FeatureSpac
 <span><span class="co">##  ...</span></span>
 <span><span class="co">##  [0. 0. 0. ... 0. 0. 0.]</span></span>
 <span><span class="co">##  [0. 0. 0. ... 0. 0. 0.]</span></span>
-<span><span class="co">##  [0. 0. 0. ... 1. 0. 0.]], shape=(32, 136), dtype=float32)</span></span></code></pre>
+<span><span class="co">##  [0. 0. 0. ... 0. 0. 0.]], shape=(32, 136), dtype=float32)</span></span></code></pre>
 </div>
 <div class="section level2">
 <h2 id="two-ways-to-manage-preprocessing-as-part-of-the-tf-data-pipeline-or-in-the-model-itself">Two ways to manage preprocessing: as part of the
@@ -560,45 +560,45 @@ <h2 id="train-the-model">Train the model<a class="anchor" aria-label="anchor" hr
 <span>  verbose <span class="op">=</span> <span class="fl">2</span></span>
 <span><span class="op">)</span></span></code></pre></div>
 <pre><code><span><span class="co">## Epoch 1/20</span></span>
-<span><span class="co">## 8/8 - 3s - 325ms/step - accuracy: 0.4481 - loss: 0.7302 - val_accuracy: 0.6000 - val_loss: 0.6799</span></span>
+<span><span class="co">## 8/8 - 2s - 262ms/step - accuracy: 0.4108 - loss: 0.7398 - val_accuracy: 0.3667 - val_loss: 0.7218</span></span>
 <span><span class="co">## Epoch 2/20</span></span>
-<span><span class="co">## 8/8 - 0s - 28ms/step - accuracy: 0.5851 - loss: 0.6797 - val_accuracy: 0.7000 - val_loss: 0.6329</span></span>
+<span><span class="co">## 8/8 - 0s - 26ms/step - accuracy: 0.5934 - loss: 0.6798 - val_accuracy: 0.5333 - val_loss: 0.6670</span></span>
 <span><span class="co">## Epoch 3/20</span></span>
-<span><span class="co">## 8/8 - 0s - 31ms/step - accuracy: 0.6141 - loss: 0.6458 - val_accuracy: 0.7333 - val_loss: 0.5951</span></span>
+<span><span class="co">## 8/8 - 0s - 26ms/step - accuracy: 0.6473 - loss: 0.6384 - val_accuracy: 0.6333 - val_loss: 0.6207</span></span>
 <span><span class="co">## Epoch 4/20</span></span>
-<span><span class="co">## 8/8 - 0s - 32ms/step - accuracy: 0.7220 - loss: 0.5843 - val_accuracy: 0.7667 - val_loss: 0.5644</span></span>
+<span><span class="co">## 8/8 - 0s - 26ms/step - accuracy: 0.7012 - loss: 0.5829 - val_accuracy: 0.7167 - val_loss: 0.5800</span></span>
 <span><span class="co">## Epoch 5/20</span></span>
-<span><span class="co">## 8/8 - 0s - 32ms/step - accuracy: 0.7552 - loss: 0.5539 - val_accuracy: 0.7667 - val_loss: 0.5367</span></span>
+<span><span class="co">## 8/8 - 0s - 27ms/step - accuracy: 0.6846 - loss: 0.5780 - val_accuracy: 0.7500 - val_loss: 0.5445</span></span>
 <span><span class="co">## Epoch 6/20</span></span>
-<span><span class="co">## 8/8 - 0s - 29ms/step - accuracy: 0.7884 - loss: 0.5248 - val_accuracy: 0.7667 - val_loss: 0.5146</span></span>
+<span><span class="co">## 8/8 - 0s - 26ms/step - accuracy: 0.7718 - loss: 0.5595 - val_accuracy: 0.7667 - val_loss: 0.5134</span></span>
 <span><span class="co">## Epoch 7/20</span></span>
-<span><span class="co">## 8/8 - 0s - 32ms/step - accuracy: 0.7469 - loss: 0.5084 - val_accuracy: 0.7667 - val_loss: 0.4960</span></span>
+<span><span class="co">## 8/8 - 0s - 26ms/step - accuracy: 0.7801 - loss: 0.5166 - val_accuracy: 0.8167 - val_loss: 0.4868</span></span>
 <span><span class="co">## Epoch 8/20</span></span>
-<span><span class="co">## 8/8 - 0s - 30ms/step - accuracy: 0.7718 - loss: 0.4921 - val_accuracy: 0.7833 - val_loss: 0.4804</span></span>
+<span><span class="co">## 8/8 - 0s - 26ms/step - accuracy: 0.7718 - loss: 0.5053 - val_accuracy: 0.8333 - val_loss: 0.4626</span></span>
 <span><span class="co">## Epoch 9/20</span></span>
-<span><span class="co">## 8/8 - 0s - 30ms/step - accuracy: 0.8216 - loss: 0.4617 - val_accuracy: 0.8000 - val_loss: 0.4668</span></span>
+<span><span class="co">## 8/8 - 0s - 28ms/step - accuracy: 0.8091 - loss: 0.4774 - val_accuracy: 0.8333 - val_loss: 0.4401</span></span>
 <span><span class="co">## Epoch 10/20</span></span>
-<span><span class="co">## 8/8 - 0s - 30ms/step - accuracy: 0.8008 - loss: 0.4550 - val_accuracy: 0.8333 - val_loss: 0.4551</span></span>
+<span><span class="co">## 8/8 - 0s - 27ms/step - accuracy: 0.8050 - loss: 0.4682 - val_accuracy: 0.8500 - val_loss: 0.4197</span></span>
 <span><span class="co">## Epoch 11/20</span></span>
-<span><span class="co">## 8/8 - 0s - 30ms/step - accuracy: 0.8133 - loss: 0.4299 - val_accuracy: 0.8500 - val_loss: 0.4447</span></span>
+<span><span class="co">## 8/8 - 0s - 26ms/step - accuracy: 0.8133 - loss: 0.4541 - val_accuracy: 0.8667 - val_loss: 0.4011</span></span>
 <span><span class="co">## Epoch 12/20</span></span>
-<span><span class="co">## 8/8 - 0s - 29ms/step - accuracy: 0.8506 - loss: 0.3980 - val_accuracy: 0.8500 - val_loss: 0.4372</span></span>
+<span><span class="co">## 8/8 - 0s - 27ms/step - accuracy: 0.8216 - loss: 0.4360 - val_accuracy: 0.8667 - val_loss: 0.3843</span></span>
 <span><span class="co">## Epoch 13/20</span></span>
-<span><span class="co">## 8/8 - 0s - 29ms/step - accuracy: 0.8008 - loss: 0.4105 - val_accuracy: 0.8500 - val_loss: 0.4308</span></span>
+<span><span class="co">## 8/8 - 0s - 26ms/step - accuracy: 0.8174 - loss: 0.4081 - val_accuracy: 0.8667 - val_loss: 0.3684</span></span>
 <span><span class="co">## Epoch 14/20</span></span>
-<span><span class="co">## 8/8 - 0s - 29ms/step - accuracy: 0.8548 - loss: 0.3820 - val_accuracy: 0.8500 - val_loss: 0.4254</span></span>
+<span><span class="co">## 8/8 - 0s - 27ms/step - accuracy: 0.8423 - loss: 0.4034 - val_accuracy: 0.8667 - val_loss: 0.3531</span></span>
 <span><span class="co">## Epoch 15/20</span></span>
-<span><span class="co">## 8/8 - 0s - 31ms/step - accuracy: 0.8589 - loss: 0.3669 - val_accuracy: 0.8500 - val_loss: 0.4205</span></span>
+<span><span class="co">## 8/8 - 0s - 26ms/step - accuracy: 0.8174 - loss: 0.3956 - val_accuracy: 0.8833 - val_loss: 0.3397</span></span>
 <span><span class="co">## Epoch 16/20</span></span>
-<span><span class="co">## 8/8 - 0s - 31ms/step - accuracy: 0.8257 - loss: 0.3786 - val_accuracy: 0.8333 - val_loss: 0.4142</span></span>
+<span><span class="co">## 8/8 - 0s - 27ms/step - accuracy: 0.8423 - loss: 0.3952 - val_accuracy: 0.8833 - val_loss: 0.3274</span></span>
 <span><span class="co">## Epoch 17/20</span></span>
-<span><span class="co">## 8/8 - 0s - 30ms/step - accuracy: 0.8548 - loss: 0.3524 - val_accuracy: 0.8333 - val_loss: 0.4105</span></span>
+<span><span class="co">## 8/8 - 0s - 27ms/step - accuracy: 0.8548 - loss: 0.3445 - val_accuracy: 0.8833 - val_loss: 0.3169</span></span>
 <span><span class="co">## Epoch 18/20</span></span>
-<span><span class="co">## 8/8 - 0s - 31ms/step - accuracy: 0.8631 - loss: 0.3530 - val_accuracy: 0.8333 - val_loss: 0.4078</span></span>
+<span><span class="co">## 8/8 - 0s - 27ms/step - accuracy: 0.8382 - loss: 0.3609 - val_accuracy: 0.8833 - val_loss: 0.3080</span></span>
 <span><span class="co">## Epoch 19/20</span></span>
-<span><span class="co">## 8/8 - 0s - 30ms/step - accuracy: 0.8548 - loss: 0.3440 - val_accuracy: 0.8333 - val_loss: 0.4044</span></span>
+<span><span class="co">## 8/8 - 0s - 26ms/step - accuracy: 0.8714 - loss: 0.3412 - val_accuracy: 0.8833 - val_loss: 0.2986</span></span>
 <span><span class="co">## Epoch 20/20</span></span>
-<span><span class="co">## 8/8 - 0s - 30ms/step - accuracy: 0.8838 - loss: 0.3303 - val_accuracy: 0.8333 - val_loss: 0.4021</span></span></code></pre>
+<span><span class="co">## 8/8 - 0s - 26ms/step - accuracy: 0.8506 - loss: 0.3541 - val_accuracy: 0.9000 - val_loss: 0.2905</span></span></code></pre>
 <p>We quickly get to 80% validation accuracy.</p>
 </div>
 <div class="section level2">
@@ -626,13 +626,13 @@ <h2 id="inference-on-new-data-with-the-end-to-end-model">Inference on new data w
 <span></span>
 <span><span class="va">input_dict</span> <span class="op">&lt;-</span> <span class="fu"><a href="https://rdrr.io/r/base/lapply.html" class="external-link">lapply</a></span><span class="op">(</span><span class="va">sample</span>, \<span class="op">(</span><span class="va">x</span><span class="op">)</span> <span class="fu"><a href="../../../reference/op_convert_to_tensor.html">op_convert_to_tensor</a></span><span class="op">(</span><span class="fu"><a href="https://rdrr.io/r/base/array.html" class="external-link">array</a></span><span class="op">(</span><span class="va">x</span><span class="op">)</span><span class="op">)</span><span class="op">)</span></span>
 <span><span class="va">predictions</span> <span class="op">&lt;-</span> <span class="va">inference_model</span> <span class="op">|&gt;</span> <span class="fu"><a href="https://rdrr.io/r/stats/predict.html" class="external-link">predict</a></span><span class="op">(</span><span class="va">input_dict</span><span class="op">)</span></span></code></pre></div>
-<pre><code><span><span class="co">## 1/1 - 0s - 368ms/step</span></span></code></pre>
+<pre><code><span><span class="co">## 1/1 - 0s - 298ms/step</span></span></code></pre>
 <div class="sourceCode" id="cb24"><pre class="downlit sourceCode r">
 <code class="sourceCode R"><span><span class="fu">glue</span><span class="fu">::</span><span class="fu"><a href="https://glue.tidyverse.org/reference/glue.html" class="external-link">glue</a></span><span class="op">(</span><span class="st">r"---(</span></span>
 <span><span class="st">  This particular patient had a {(100 * predictions) |&gt; signif(3)}% probability</span></span>
 <span><span class="st">  of having a heart disease, as evaluated by our model.</span></span>
 <span><span class="st">)---"</span><span class="op">)</span></span></code></pre></div>
-<pre><code><span><span class="co">## This particular patient had a 48.8% probability</span></span>
+<pre><code><span><span class="co">## This particular patient had a 48.6% probability</span></span>
 <span><span class="co">## of having a heart disease, as evaluated by our model.</span></span></code></pre>
 </div>
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -646,7 +646,7 @@ <h2 id="inference-on-new-data-with-the-end-to-end-model">Inference on new data w
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer>
diff --git a/docs/dev/articles/examples/timeseries/timeseries_anomaly_detection.html b/docs/dev/articles/examples/timeseries/timeseries_anomaly_detection.html
index 6eba1e9ed..b5f49d178 100644
--- a/docs/dev/articles/examples/timeseries/timeseries_anomaly_detection.html
+++ b/docs/dev/articles/examples/timeseries/timeseries_anomaly_detection.html
@@ -31,7 +31,7 @@
 
     <a class="navbar-brand me-2" href="../../../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -104,7 +104,7 @@ <h2 id="setup">Setup<a class="anchor" aria-label="anchor" href="#setup"></a>
 <div class="sourceCode" id="cb1"><pre class="downlit sourceCode r">
 <code class="sourceCode R"><span><span class="kw"><a href="https://rdrr.io/r/base/library.html" class="external-link">library</a></span><span class="op">(</span><span class="va"><a href="https://dplyr.tidyverse.org" class="external-link">dplyr</a></span>, warn.conflicts <span class="op">=</span> <span class="cn">FALSE</span><span class="op">)</span></span>
 <span><span class="kw"><a href="https://rdrr.io/r/base/library.html" class="external-link">library</a></span><span class="op">(</span><span class="va"><a href="https://ggplot2.tidyverse.org" class="external-link">ggplot2</a></span><span class="op">)</span></span>
-<span><span class="fu"><a href="https://ggplot2.tidyverse.org/reference/theme_get.html" class="external-link">theme_set</a></span><span class="op">(</span><span class="fu"><a href="https://ggplot2.tidyverse.org/reference/ggtheme.html" class="external-link">theme_minimal</a></span><span class="op">(</span><span class="op">)</span><span class="op">)</span></span>
+<span><span class="fu"><a href="https://ggplot2.tidyverse.org/reference/get_theme.html" class="external-link">theme_set</a></span><span class="op">(</span><span class="fu"><a href="https://ggplot2.tidyverse.org/reference/ggtheme.html" class="external-link">theme_minimal</a></span><span class="op">(</span><span class="op">)</span><span class="op">)</span></span>
 <span></span>
 <span><span class="kw"><a href="https://rdrr.io/r/base/library.html" class="external-link">library</a></span><span class="op">(</span><span class="va"><a href="https://github.com/t-kalinowski/listarrays" class="external-link">listarrays</a></span><span class="op">)</span></span>
 <span><span class="kw"><a href="https://rdrr.io/r/base/library.html" class="external-link">library</a></span><span class="op">(</span><span class="va"><a href="https://github.com/rstudio/tfdatasets" class="external-link">tfdatasets</a></span>, exclude <span class="op">=</span> <span class="fu"><a href="https://rdrr.io/r/base/c.html" class="external-link">c</a></span><span class="op">(</span><span class="st">"shape"</span><span class="op">)</span><span class="op">)</span></span>
@@ -130,24 +130,13 @@ <h2 id="load-the-data">Load the data<a class="anchor" aria-label="anchor" href="
 <span>  <span class="fu">readr</span><span class="fu">::</span><span class="fu"><a href="https://readr.tidyverse.org/reference/read_delim.html" class="external-link">read_csv</a></span><span class="op">(</span><span class="va">file</span>, col_types <span class="op">=</span> <span class="st">"Td"</span><span class="op">)</span></span>
 <span><span class="op">}</span></span>
 <span></span>
-<span><span class="va">df_small_noise</span>   <span class="op">&lt;-</span> <span class="fu">get_data</span><span class="op">(</span><span class="st">"artificialNoAnomaly/art_daily_small_noise.csv"</span><span class="op">)</span></span></code></pre></div>
-<pre><code>## Downloading data from https://raw.githubusercontent.com/numenta/NAB/master/data/artificialNoAnomaly/art_daily_small_noise.csv
-##
-
[1m     0/140187
[0m 
[37m━━━━━━━━━━━━━━━━━━━━
[0m 
[1m0s
[0m 0s/step
-
[1m122880/140187
[0m 
[32m━━━━━━━━━━━━━━━━━
[0m
[37m━━━
[0m 
[1m0s
[0m 0us/step
-
[1m140187/140187
[0m 
[32m━━━━━━━━━━━━━━━━━━━━
[0m
[37m
[0m 
[1m0s
[0m 0us/step</code></pre>
-<div class="sourceCode" id="cb4"><pre class="downlit sourceCode r">
-<code class="sourceCode R"><span><span class="va">df_daily_jumpsup</span> <span class="op">&lt;-</span> <span class="fu">get_data</span><span class="op">(</span><span class="st">"artificialWithAnomaly/art_daily_jumpsup.csv"</span><span class="op">)</span></span></code></pre></div>
-<pre><code>## Downloading data from https://raw.githubusercontent.com/numenta/NAB/master/data/artificialWithAnomaly/art_daily_jumpsup.csv
-##
-
[1m     0/140548
[0m 
[37m━━━━━━━━━━━━━━━━━━━━
[0m 
[1m0s
[0m 0s/step
-
[1m114688/140548
[0m 
[32m━━━━━━━━━━━━━━━━
[0m
[37m━━━━
[0m 
[1m0s
[0m 0us/step
-
[1m140548/140548
[0m 
[32m━━━━━━━━━━━━━━━━━━━━
[0m
[37m
[0m 
[1m0s
[0m 0us/step</code></pre>
+<span><span class="va">df_small_noise</span>   <span class="op">&lt;-</span> <span class="fu">get_data</span><span class="op">(</span><span class="st">"artificialNoAnomaly/art_daily_small_noise.csv"</span><span class="op">)</span></span>
+<span><span class="va">df_daily_jumpsup</span> <span class="op">&lt;-</span> <span class="fu">get_data</span><span class="op">(</span><span class="st">"artificialWithAnomaly/art_daily_jumpsup.csv"</span><span class="op">)</span></span></code></pre></div>
 </div>
 <div class="section level2">
 <h2 id="quick-look-at-the-data">Quick look at the data<a class="anchor" aria-label="anchor" href="#quick-look-at-the-data"></a>
 </h2>
-<div class="sourceCode" id="cb6"><pre class="downlit sourceCode r">
+<div class="sourceCode" id="cb3"><pre class="downlit sourceCode r">
 <code class="sourceCode R"><span><span class="va">df_small_noise</span></span></code></pre></div>
 <pre><code><span><span class="co">## <span style="color: #949494;"># A tibble: 4,032 × 2</span></span></span>
 <span><span class="co">##    timestamp           value</span></span>
@@ -163,7 +152,7 @@ <h2 id="quick-look-at-the-data">Quick look at the data<a class="anchor" aria-lab
 <span><span class="co">## <span style="color: #BCBCBC;"> 9</span> 2014-04-01 <span style="color: #949494;">00:40:00</span>  21.5</span></span>
 <span><span class="co">## <span style="color: #BCBCBC;">10</span> 2014-04-01 <span style="color: #949494;">00:45:00</span>  19.2</span></span>
 <span><span class="co">## <span style="color: #949494;"># ℹ 4,022 more rows</span></span></span></code></pre>
-<div class="sourceCode" id="cb8"><pre class="downlit sourceCode r">
+<div class="sourceCode" id="cb5"><pre class="downlit sourceCode r">
 <code class="sourceCode R"><span><span class="va">df_daily_jumpsup</span></span></code></pre></div>
 <pre><code><span><span class="co">## <span style="color: #949494;"># A tibble: 4,032 × 2</span></span></span>
 <span><span class="co">##    timestamp           value</span></span>
@@ -187,7 +176,7 @@ <h2 id="visualize-the-data">Visualize the data<a class="anchor" aria-label="anch
 <h3 id="timeseries-data-without-anomalies">Timeseries data without anomalies<a class="anchor" aria-label="anchor" href="#timeseries-data-without-anomalies"></a>
 </h3>
 <p>We will use the following data for training.</p>
-<div class="sourceCode" id="cb10"><pre class="downlit sourceCode r">
+<div class="sourceCode" id="cb7"><pre class="downlit sourceCode r">
 <code class="sourceCode R"><span><span class="va">plot_ts</span> <span class="op">&lt;-</span> <span class="kw">function</span><span class="op">(</span><span class="va">df</span><span class="op">)</span> <span class="op">{</span></span>
 <span>  <span class="fu"><a href="https://ggplot2.tidyverse.org/reference/ggplot.html" class="external-link">ggplot</a></span><span class="op">(</span><span class="va">df</span>, <span class="fu"><a href="https://ggplot2.tidyverse.org/reference/aes.html" class="external-link">aes</a></span><span class="op">(</span>x <span class="op">=</span> <span class="va">timestamp</span>, y <span class="op">=</span> <span class="va">value</span><span class="op">)</span><span class="op">)</span> <span class="op">+</span> <span class="fu"><a href="https://ggplot2.tidyverse.org/reference/geom_path.html" class="external-link">geom_line</a></span><span class="op">(</span><span class="op">)</span> <span class="op">+</span></span>
 <span>    <span class="fu"><a href="https://ggplot2.tidyverse.org/reference/scale_date.html" class="external-link">scale_x_datetime</a></span><span class="op">(</span>date_breaks <span class="op">=</span> <span class="st">"1 day"</span>, date_labels <span class="op">=</span> <span class="st">"%b-%d"</span><span class="op">)</span></span>
@@ -203,7 +192,7 @@ <h3 id="timeseries-data-with-anomalies">Timeseries data with anomalies<a class="
 </h3>
 <p>We will use the following data for testing and see if the sudden jump
 up in the data is detected as an anomaly.</p>
-<div class="sourceCode" id="cb11"><pre class="downlit sourceCode r">
+<div class="sourceCode" id="cb8"><pre class="downlit sourceCode r">
 <code class="sourceCode R"><span><span class="fu">plot_ts</span><span class="op">(</span><span class="va">df_daily_jumpsup</span><span class="op">)</span> <span class="op">+</span> <span class="fu"><a href="https://ggplot2.tidyverse.org/reference/labs.html" class="external-link">ggtitle</a></span><span class="op">(</span><span class="st">"With Anomaly"</span><span class="op">)</span></span></code></pre></div>
 <div class="float">
 <img src="timeseries_anomaly_detection/unnamed-chunk-5-1.png" alt="plot of chunk unnamed-chunk-5"><div class="figcaption">plot of chunk unnamed-chunk-5</div>
@@ -221,7 +210,7 @@ <h2 id="prepare-training-data">Prepare training data<a class="anchor" aria-label
 </li>
 <li>288 * 14 = <strong>4032 data points</strong> in total</li>
 </ul>
-<div class="sourceCode" id="cb12"><pre class="downlit sourceCode r">
+<div class="sourceCode" id="cb9"><pre class="downlit sourceCode r">
 <code class="sourceCode R"><span><span class="va">df_train</span> <span class="op">&lt;-</span> <span class="va">df_small_noise</span> <span class="op">|&gt;</span></span>
 <span>  <span class="fu"><a href="https://dplyr.tidyverse.org/reference/mutate.html" class="external-link">mutate</a></span><span class="op">(</span>value <span class="op">=</span> <span class="op">(</span><span class="va">value</span> <span class="op">-</span> <span class="fu"><a href="https://rdrr.io/r/base/mean.html" class="external-link">mean</a></span><span class="op">(</span><span class="va">value</span><span class="op">)</span><span class="op">)</span> <span class="op">/</span> <span class="fu"><a href="https://rdrr.io/r/stats/sd.html" class="external-link">sd</a></span><span class="op">(</span><span class="va">value</span><span class="op">)</span><span class="op">)</span></span>
 <span></span>
@@ -232,7 +221,7 @@ <h3 id="create-sequences">Create sequences<a class="anchor" aria-label="anchor"
 </h3>
 <p>Create sequences combining <code>TIME_STEPS</code> contiguous data
 values from the training data.</p>
-<div class="sourceCode" id="cb14"><pre class="downlit sourceCode r">
+<div class="sourceCode" id="cb11"><pre class="downlit sourceCode r">
 <code class="sourceCode R"><span><span class="va">TIME_STEPS</span> <span class="op">&lt;-</span> <span class="fl">288</span></span>
 <span></span>
 <span><span class="va">as_dataset</span> <span class="op">&lt;-</span> <span class="kw">function</span><span class="op">(</span><span class="va">df</span><span class="op">)</span> <span class="op">{</span></span>
@@ -255,7 +244,7 @@ <h2 id="build-a-model">Build a model<a class="anchor" aria-label="anchor" href="
 <code>(batch_size, sequence_length, num_features)</code> and return
 output of the same shape. In this case, <code>sequence_length</code> is
 288 and <code>num_features</code> is 1.</p>
-<div class="sourceCode" id="cb16"><pre class="downlit sourceCode r">
+<div class="sourceCode" id="cb13"><pre class="downlit sourceCode r">
 <code class="sourceCode R"><span><span class="va">model</span> <span class="op">&lt;-</span> <span class="fu"><a href="../../../reference/keras_model_sequential.html">keras_model_sequential</a></span><span class="op">(</span>input_shape <span class="op">=</span> <span class="fu"><a href="https://rdrr.io/r/base/c.html" class="external-link">c</a></span><span class="op">(</span><span class="va">TIME_STEPS</span>, <span class="fl">1</span><span class="op">)</span><span class="op">)</span> <span class="op">|&gt;</span></span>
 <span>  <span class="fu"><a href="../../../reference/layer_conv_1d.html">layer_conv_1d</a></span><span class="op">(</span></span>
 <span>    filters <span class="op">=</span> <span class="fl">32</span>, kernel_size <span class="op">=</span> <span class="fl">7</span>, padding <span class="op">=</span> <span class="st">"same"</span>,</span>
@@ -309,7 +298,7 @@ <h2 id="train-the-model">Train the model<a class="anchor" aria-label="anchor" hr
 </h2>
 <p>Please note that we are using <code>x_train</code> as both the input
 and the target since this is a reconstruction model.</p>
-<div class="sourceCode" id="cb18"><pre class="downlit sourceCode r">
+<div class="sourceCode" id="cb15"><pre class="downlit sourceCode r">
 <code class="sourceCode R"><span><span class="va">history</span> <span class="op">=</span> <span class="va">model</span> <span class="op">|&gt;</span> <span class="fu"><a href="https://generics.r-lib.org/reference/fit.html" class="external-link">fit</a></span><span class="op">(</span></span>
 <span>  <span class="va">x_train</span>, <span class="va">x_train</span>,</span>
 <span>  epochs <span class="op">=</span> <span class="fl">50</span>,</span>
@@ -321,61 +310,61 @@ <h2 id="train-the-model">Train the model<a class="anchor" aria-label="anchor" hr
 <span>  <span class="op">)</span></span>
 <span><span class="op">)</span></span></code></pre></div>
 <pre><code><span><span class="co">## Epoch 1/50</span></span>
-<span><span class="co">## 106/106 - 5s - 48ms/step - loss: 0.1863 - val_loss: 0.0316</span></span>
+<span><span class="co">## 106/106 - 4s - 39ms/step - loss: 0.1863 - val_loss: 0.0316</span></span>
 <span><span class="co">## Epoch 2/50</span></span>
 <span><span class="co">## 106/106 - 0s - 2ms/step - loss: 0.0333 - val_loss: 0.0233</span></span>
 <span><span class="co">## Epoch 3/50</span></span>
 <span><span class="co">## 106/106 - 0s - 2ms/step - loss: 0.0248 - val_loss: 0.0201</span></span>
 <span><span class="co">## Epoch 4/50</span></span>
-<span><span class="co">## 106/106 - 0s - 3ms/step - loss: 0.0209 - val_loss: 0.0187</span></span>
+<span><span class="co">## 106/106 - 0s - 2ms/step - loss: 0.0209 - val_loss: 0.0186</span></span>
 <span><span class="co">## Epoch 5/50</span></span>
-<span><span class="co">## 106/106 - 0s - 3ms/step - loss: 0.0179 - val_loss: 0.0146</span></span>
+<span><span class="co">## 106/106 - 0s - 2ms/step - loss: 0.0179 - val_loss: 0.0146</span></span>
 <span><span class="co">## Epoch 6/50</span></span>
-<span><span class="co">## 106/106 - 0s - 3ms/step - loss: 0.0150 - val_loss: 0.0113</span></span>
+<span><span class="co">## 106/106 - 0s - 2ms/step - loss: 0.0150 - val_loss: 0.0112</span></span>
 <span><span class="co">## Epoch 7/50</span></span>
-<span><span class="co">## 106/106 - 0s - 3ms/step - loss: 0.0127 - val_loss: 0.0094</span></span>
+<span><span class="co">## 106/106 - 0s - 2ms/step - loss: 0.0127 - val_loss: 0.0094</span></span>
 <span><span class="co">## Epoch 8/50</span></span>
-<span><span class="co">## 106/106 - 0s - 3ms/step - loss: 0.0109 - val_loss: 0.0088</span></span>
+<span><span class="co">## 106/106 - 0s - 2ms/step - loss: 0.0109 - val_loss: 0.0088</span></span>
 <span><span class="co">## Epoch 9/50</span></span>
-<span><span class="co">## 106/106 - 0s - 3ms/step - loss: 0.0096 - val_loss: 0.0084</span></span>
+<span><span class="co">## 106/106 - 0s - 2ms/step - loss: 0.0096 - val_loss: 0.0083</span></span>
 <span><span class="co">## Epoch 10/50</span></span>
-<span><span class="co">## 106/106 - 0s - 3ms/step - loss: 0.0086 - val_loss: 0.0070</span></span>
+<span><span class="co">## 106/106 - 0s - 2ms/step - loss: 0.0086 - val_loss: 0.0069</span></span>
 <span><span class="co">## Epoch 11/50</span></span>
-<span><span class="co">## 106/106 - 0s - 3ms/step - loss: 0.0078 - val_loss: 0.0061</span></span>
+<span><span class="co">## 106/106 - 0s - 2ms/step - loss: 0.0078 - val_loss: 0.0061</span></span>
 <span><span class="co">## Epoch 12/50</span></span>
-<span><span class="co">## 106/106 - 0s - 2ms/step - loss: 0.0073 - val_loss: 0.0058</span></span>
+<span><span class="co">## 106/106 - 0s - 2ms/step - loss: 0.0073 - val_loss: 0.0057</span></span>
 <span><span class="co">## Epoch 13/50</span></span>
-<span><span class="co">## 106/106 - 0s - 3ms/step - loss: 0.0068 - val_loss: 0.0053</span></span>
+<span><span class="co">## 106/106 - 0s - 2ms/step - loss: 0.0068 - val_loss: 0.0053</span></span>
 <span><span class="co">## Epoch 14/50</span></span>
-<span><span class="co">## 106/106 - 0s - 3ms/step - loss: 0.0064 - val_loss: 0.0050</span></span>
+<span><span class="co">## 106/106 - 0s - 2ms/step - loss: 0.0064 - val_loss: 0.0050</span></span>
 <span><span class="co">## Epoch 15/50</span></span>
-<span><span class="co">## 106/106 - 0s - 3ms/step - loss: 0.0061 - val_loss: 0.0046</span></span>
+<span><span class="co">## 106/106 - 0s - 2ms/step - loss: 0.0061 - val_loss: 0.0046</span></span>
 <span><span class="co">## Epoch 16/50</span></span>
-<span><span class="co">## 106/106 - 0s - 4ms/step - loss: 0.0058 - val_loss: 0.0040</span></span>
+<span><span class="co">## 106/106 - 0s - 2ms/step - loss: 0.0058 - val_loss: 0.0040</span></span>
 <span><span class="co">## Epoch 17/50</span></span>
-<span><span class="co">## 106/106 - 0s - 3ms/step - loss: 0.0056 - val_loss: 0.0041</span></span>
+<span><span class="co">## 106/106 - 0s - 2ms/step - loss: 0.0056 - val_loss: 0.0041</span></span>
 <span><span class="co">## Epoch 18/50</span></span>
-<span><span class="co">## 106/106 - 0s - 3ms/step - loss: 0.0053 - val_loss: 0.0036</span></span>
+<span><span class="co">## 106/106 - 0s - 2ms/step - loss: 0.0053 - val_loss: 0.0035</span></span>
 <span><span class="co">## Epoch 19/50</span></span>
-<span><span class="co">## 106/106 - 0s - 3ms/step - loss: 0.0050 - val_loss: 0.0038</span></span>
+<span><span class="co">## 106/106 - 0s - 2ms/step - loss: 0.0050 - val_loss: 0.0037</span></span>
 <span><span class="co">## Epoch 20/50</span></span>
-<span><span class="co">## 106/106 - 0s - 3ms/step - loss: 0.0048 - val_loss: 0.0032</span></span>
+<span><span class="co">## 106/106 - 0s - 2ms/step - loss: 0.0048 - val_loss: 0.0032</span></span>
 <span><span class="co">## Epoch 21/50</span></span>
-<span><span class="co">## 106/106 - 0s - 3ms/step - loss: 0.0046 - val_loss: 0.0032</span></span>
+<span><span class="co">## 106/106 - 0s - 2ms/step - loss: 0.0046 - val_loss: 0.0031</span></span>
 <span><span class="co">## Epoch 22/50</span></span>
-<span><span class="co">## 106/106 - 0s - 3ms/step - loss: 0.0044 - val_loss: 0.0032</span></span>
+<span><span class="co">## 106/106 - 0s - 2ms/step - loss: 0.0044 - val_loss: 0.0032</span></span>
 <span><span class="co">## Epoch 23/50</span></span>
-<span><span class="co">## 106/106 - 0s - 3ms/step - loss: 0.0042 - val_loss: 0.0033</span></span>
+<span><span class="co">## 106/106 - 0s - 2ms/step - loss: 0.0042 - val_loss: 0.0033</span></span>
 <span><span class="co">## Epoch 24/50</span></span>
-<span><span class="co">## 106/106 - 0s - 3ms/step - loss: 0.0041 - val_loss: 0.0031</span></span>
+<span><span class="co">## 106/106 - 0s - 2ms/step - loss: 0.0041 - val_loss: 0.0031</span></span>
 <span><span class="co">## Epoch 25/50</span></span>
-<span><span class="co">## 106/106 - 0s - 3ms/step - loss: 0.0039 - val_loss: 0.0033</span></span>
+<span><span class="co">## 106/106 - 0s - 2ms/step - loss: 0.0039 - val_loss: 0.0033</span></span>
 <span><span class="co">## Epoch 26/50</span></span>
-<span><span class="co">## 106/106 - 0s - 3ms/step - loss: 0.0037 - val_loss: 0.0031</span></span>
+<span><span class="co">## 106/106 - 0s - 2ms/step - loss: 0.0037 - val_loss: 0.0030</span></span>
 <span><span class="co">## Epoch 27/50</span></span>
-<span><span class="co">## 106/106 - 0s - 3ms/step - loss: 0.0035 - val_loss: 0.0032</span></span>
+<span><span class="co">## 106/106 - 0s - 2ms/step - loss: 0.0035 - val_loss: 0.0032</span></span>
 <span><span class="co">## Epoch 28/50</span></span>
-<span><span class="co">## 106/106 - 0s - 2ms/step - loss: 0.0034 - val_loss: 0.0026</span></span>
+<span><span class="co">## 106/106 - 0s - 2ms/step - loss: 0.0034 - val_loss: 0.0025</span></span>
 <span><span class="co">## Epoch 29/50</span></span>
 <span><span class="co">## 106/106 - 0s - 2ms/step - loss: 0.0033 - val_loss: 0.0028</span></span>
 <span><span class="co">## Epoch 30/50</span></span>
@@ -387,20 +376,20 @@ <h2 id="train-the-model">Train the model<a class="anchor" aria-label="anchor" hr
 <span><span class="co">## Epoch 33/50</span></span>
 <span><span class="co">## 106/106 - 0s - 2ms/step - loss: 0.0029 - val_loss: 0.0027</span></span>
 <span><span class="co">## Epoch 34/50</span></span>
-<span><span class="co">## 106/106 - 0s - 3ms/step - loss: 0.0028 - val_loss: 0.0023</span></span>
+<span><span class="co">## 106/106 - 0s - 2ms/step - loss: 0.0028 - val_loss: 0.0022</span></span>
 <span><span class="co">## Epoch 35/50</span></span>
 <span><span class="co">## 106/106 - 0s - 2ms/step - loss: 0.0027 - val_loss: 0.0024</span></span>
 <span><span class="co">## Epoch 36/50</span></span>
-<span><span class="co">## 106/106 - 0s - 2ms/step - loss: 0.0027 - val_loss: 0.0025</span></span>
+<span><span class="co">## 106/106 - 0s - 2ms/step - loss: 0.0026 - val_loss: 0.0025</span></span>
 <span><span class="co">## Epoch 37/50</span></span>
-<span><span class="co">## 106/106 - 0s - 2ms/step - loss: 0.0026 - val_loss: 0.0026</span></span>
+<span><span class="co">## 106/106 - 0s - 2ms/step - loss: 0.0026 - val_loss: 0.0027</span></span>
 <span><span class="co">## Epoch 38/50</span></span>
-<span><span class="co">## 106/106 - 0s - 2ms/step - loss: 0.0026 - val_loss: 0.0024</span></span>
+<span><span class="co">## 106/106 - 0s - 2ms/step - loss: 0.0025 - val_loss: 0.0024</span></span>
 <span><span class="co">## Epoch 39/50</span></span>
-<span><span class="co">## 106/106 - 0s - 3ms/step - loss: 0.0025 - val_loss: 0.0023</span></span></code></pre>
+<span><span class="co">## 106/106 - 0s - 2ms/step - loss: 0.0025 - val_loss: 0.0024</span></span></code></pre>
 <p>Let’s plot training and validation loss to see how the training
 went.</p>
-<div class="sourceCode" id="cb20"><pre class="downlit sourceCode r">
+<div class="sourceCode" id="cb17"><pre class="downlit sourceCode r">
 <code class="sourceCode R"><span><span class="fu"><a href="https://rdrr.io/r/graphics/plot.default.html" class="external-link">plot</a></span><span class="op">(</span><span class="va">history</span><span class="op">)</span></span></code></pre></div>
 <div class="float">
 <img src="timeseries_anomaly_detection/unnamed-chunk-10-1.png" alt="plot of chunk unnamed-chunk-10"><div class="figcaption">plot of chunk unnamed-chunk-10</div>
@@ -421,29 +410,29 @@ <h2 id="detecting-anomalies">Detecting anomalies<a class="anchor" aria-label="an
 a pattern that it isn’t familiar with. We will label this sample as an
 <code>anomaly</code>.</li>
 </ol>
-<div class="sourceCode" id="cb21"><pre class="downlit sourceCode r">
+<div class="sourceCode" id="cb18"><pre class="downlit sourceCode r">
 <code class="sourceCode R"><span><span class="co"># Get train MAE loss.</span></span>
 <span><span class="va">x_train_pred</span> <span class="op">&lt;-</span> <span class="va">model</span> <span class="op">|&gt;</span> <span class="fu"><a href="https://rdrr.io/r/stats/predict.html" class="external-link">predict</a></span><span class="op">(</span><span class="va">x_train</span><span class="op">)</span></span></code></pre></div>
-<pre><code><span><span class="co">## 118/118 - 1s - 5ms/step</span></span></code></pre>
-<div class="sourceCode" id="cb23"><pre class="downlit sourceCode r">
+<pre><code><span><span class="co">## 118/118 - 0s - 4ms/step</span></span></code></pre>
+<div class="sourceCode" id="cb20"><pre class="downlit sourceCode r">
 <code class="sourceCode R"><span><span class="va">train_mae_loss</span> <span class="op">&lt;-</span> <span class="fu"><a href="https://rdrr.io/r/base/apply.html" class="external-link">apply</a></span><span class="op">(</span><span class="fu"><a href="https://rdrr.io/r/base/MathFun.html" class="external-link">abs</a></span><span class="op">(</span><span class="va">x_train_pred</span> <span class="op">-</span> <span class="va">x_train</span><span class="op">)</span>, <span class="fl">1</span>, <span class="va">mean</span><span class="op">)</span></span>
 <span></span>
 <span><span class="fu"><a href="https://rdrr.io/r/graphics/hist.html" class="external-link">hist</a></span><span class="op">(</span><span class="va">train_mae_loss</span>, breaks <span class="op">=</span> <span class="fl">50</span><span class="op">)</span></span></code></pre></div>
 <div class="float">
 <img src="timeseries_anomaly_detection/unnamed-chunk-11-1.png" alt="plot of chunk unnamed-chunk-11"><div class="figcaption">plot of chunk unnamed-chunk-11</div>
 </div>
-<div class="sourceCode" id="cb24"><pre class="downlit sourceCode r">
+<div class="sourceCode" id="cb21"><pre class="downlit sourceCode r">
 <code class="sourceCode R"><span><span class="co"># Get reconstruction loss threshold.</span></span>
 <span><span class="va">threshold</span> <span class="op">&lt;-</span> <span class="fu"><a href="https://rdrr.io/r/base/Extremes.html" class="external-link">max</a></span><span class="op">(</span><span class="va">train_mae_loss</span><span class="op">)</span></span>
 <span><span class="fu"><a href="https://rdrr.io/r/base/cat.html" class="external-link">cat</a></span><span class="op">(</span><span class="st">"Reconstruction error threshold: "</span>, <span class="va">threshold</span>, <span class="st">"\n"</span><span class="op">)</span></span></code></pre></div>
-<pre><code><span><span class="co">## Reconstruction error threshold:  0.03910884</span></span></code></pre>
+<pre><code><span><span class="co">## Reconstruction error threshold:  0.03994837</span></span></code></pre>
 <div class="section level3">
 <h3 id="compare-recontruction">Compare recontruction<a class="anchor" aria-label="anchor" href="#compare-recontruction"></a>
 </h3>
 <p>Just for fun, let’s see how our model has recontructed the first
 sample. This is the 288 timesteps from day 1 of our training
 dataset.</p>
-<div class="sourceCode" id="cb26"><pre class="downlit sourceCode r">
+<div class="sourceCode" id="cb23"><pre class="downlit sourceCode r">
 <code class="sourceCode R"><span><span class="co"># Checking how the first sequence is learnt</span></span>
 <span><span class="fu"><a href="https://rdrr.io/r/graphics/plot.default.html" class="external-link">plot</a></span><span class="op">(</span><span class="cn">NULL</span>, <span class="cn">NULL</span>, ylab <span class="op">=</span> <span class="st">'Value'</span>,</span>
 <span>     xlim <span class="op">=</span> <span class="fu"><a href="https://rdrr.io/r/base/c.html" class="external-link">c</a></span><span class="op">(</span><span class="fl">0</span>, <span class="va">TIME_STEPS</span><span class="op">)</span>,</span>
@@ -460,7 +449,7 @@ <h3 id="compare-recontruction">Compare recontruction<a class="anchor" aria-label
 <div class="section level3">
 <h3 id="prepare-test-data">Prepare test data<a class="anchor" aria-label="anchor" href="#prepare-test-data"></a>
 </h3>
-<div class="sourceCode" id="cb27"><pre class="downlit sourceCode r">
+<div class="sourceCode" id="cb24"><pre class="downlit sourceCode r">
 <code class="sourceCode R"><span><span class="va">df_test</span> <span class="op">&lt;-</span> <span class="va">df_daily_jumpsup</span> <span class="op">|&gt;</span></span>
 <span>  <span class="fu"><a href="https://dplyr.tidyverse.org/reference/mutate.html" class="external-link">mutate</a></span><span class="op">(</span>value <span class="op">=</span></span>
 <span>           <span class="op">(</span><span class="va">value</span> <span class="op">-</span> <span class="fu"><a href="https://rdrr.io/r/base/mean.html" class="external-link">mean</a></span><span class="op">(</span><span class="va">df_small_noise</span><span class="op">$</span><span class="va">value</span><span class="op">)</span><span class="op">)</span> <span class="op">/</span></span>
@@ -471,7 +460,7 @@ <h3 id="prepare-test-data">Prepare test data<a class="anchor" aria-label="anchor
 <div class="float">
 <img src="timeseries_anomaly_detection/unnamed-chunk-13-1.png" alt="plot of chunk unnamed-chunk-13"><div class="figcaption">plot of chunk unnamed-chunk-13</div>
 </div>
-<div class="sourceCode" id="cb28"><pre class="downlit sourceCode r">
+<div class="sourceCode" id="cb25"><pre class="downlit sourceCode r">
 <code class="sourceCode R"><span><span class="co"># Create sequences from test values.</span></span>
 <span><span class="va">x_test</span> <span class="op">&lt;-</span> <span class="fu">as_dataset</span><span class="op">(</span><span class="va">df_test</span><span class="op">)</span></span>
 <span></span>
@@ -483,7 +472,7 @@ <h3 id="prepare-test-data">Prepare test data<a class="anchor" aria-label="anchor
 <div class="float">
 <img src="timeseries_anomaly_detection/unnamed-chunk-13-2.png" alt="plot of chunk unnamed-chunk-13"><div class="figcaption">plot of chunk unnamed-chunk-13</div>
 </div>
-<div class="sourceCode" id="cb29"><pre class="downlit sourceCode r">
+<div class="sourceCode" id="cb26"><pre class="downlit sourceCode r">
 <code class="sourceCode R"><span><span class="co"># Detect all the samples which are anomalies.</span></span>
 <span><span class="va">anomalies</span> <span class="op">&lt;-</span> <span class="va">test_mae_loss</span> <span class="op">&gt;</span> <span class="va">threshold</span></span>
 <span><span class="fu"><a href="https://rdrr.io/r/base/cat.html" class="external-link">cat</a></span><span class="op">(</span><span class="st">"Number of anomaly samples:"</span>, <span class="fu"><a href="https://rdrr.io/r/base/sum.html" class="external-link">sum</a></span><span class="op">(</span><span class="va">anomalies</span><span class="op">)</span>, <span class="st">"\n"</span><span class="op">)</span></span>
@@ -497,41 +486,41 @@ <h3 id="prepare-test-data">Prepare test data<a class="anchor" aria-label="anchor
 <span><span class="co">## <span style="color: #BCBCBC;">4</span> 2014-04-01 <span style="color: #949494;">00:15:00</span> -<span style="color: #BB0000;">0.746</span></span></span>
 <span><span class="co">## <span style="color: #BCBCBC;">5</span> 2014-04-01 <span style="color: #949494;">00:20:00</span> -<span style="color: #BB0000;">0.792</span></span></span>
 <span><span class="co">## <span style="color: #BCBCBC;">6</span> 2014-04-01 <span style="color: #949494;">00:25:00</span> -<span style="color: #BB0000;">0.802</span></span></span>
-<span><span class="co">## 118/118 - 0s - 1ms/step</span></span>
+<span><span class="co">## 118/118 - 0s - 808us/step</span></span>
 <span><span class="co">## Number of anomaly samples: 483</span></span>
-<span><span class="co">## Indices of anomaly samples: 216 217 218 219 220 221 396 398 507 793 794 795</span></span>
-<span><span class="co">## 797 798 799 974 1658 1659 1944 1945 1946 1947 2042 2046 2050 2052 2054 2056</span></span>
-<span><span class="co">## 2058 2060 2061 2062 2064 2065 2066 2068 2069 2070 2072 2073 2074 2076 2077</span></span>
-<span><span class="co">## 2078 2080 2081 2082 2084 2085 2086 2088 2089 2090 2092 2093 2094 2096 2097</span></span>
-<span><span class="co">## 2098 2100 2101 2102 2104 2108 2110 2112 2114 2116 2118 2120 2121 2122 2124</span></span>
-<span><span class="co">## 2126 2141 2521 2522 2523 2525 2546 2698 2700 2702 2703 2704 2705 2706 2707</span></span>
-<span><span class="co">## 2708 2709 2710 2711 2712 2713 2714 2715 2716 2717 2718 2719 2720 2721 2722</span></span>
-<span><span class="co">## 2723 2724 2725 2726 2727 2728 2729 2730 2731 2732 2733 2734 2735 2736 2737</span></span>
-<span><span class="co">## 2738 2739 2740 2741 2742 2743 2744 2745 2746 2747 2748 2749 2750 2751 2752</span></span>
-<span><span class="co">## 2753 2754 2755 2756 2757 2758 2759 2760 2761 2762 2763 2764 2765 2766 2767</span></span>
-<span><span class="co">## 2768 2769 2770 2771 2772 2773 2774 2775 2776 2777 2778 2779 2780 2781 2782</span></span>
-<span><span class="co">## 2783 2784 2785 2786 2787 2788 2789 2790 2791 2792 2793 2794 2795 2796 2797</span></span>
-<span><span class="co">## 2798 2799 2800 2801 2802 2803 2804 2805 2806 2807 2808 2809 2810 2811 2812</span></span>
-<span><span class="co">## 2813 2814 2815 2816 2817 2818 2819 2820 2821 2822 2823 2824 2825 2826 2827</span></span>
-<span><span class="co">## 2828 2829 2830 2831 2832 2833 2834 2835 2836 2837 2838 2839 2840 2841 2842</span></span>
-<span><span class="co">## 2843 2844 2845 2846 2847 2848 2849 2850 2851 2852 2853 2854 2855 2856 2857</span></span>
-<span><span class="co">## 2858 2859 2860 2861 2862 2863 2864 2865 2866 2867 2868 2869 2870 2871 2872</span></span>
-<span><span class="co">## 2873 2874 2875 2876 2877 2878 2879 2880 2881 2882 2883 2884 2885 2886 2887</span></span>
-<span><span class="co">## 2888 2889 2890 2891 2892 2893 2894 2895 2896 2897 2898 2899 2900 2901 2902</span></span>
-<span><span class="co">## 2903 2904 2905 2906 2907 2908 2909 2910 2911 2912 2913 2914 2915 2916 2917</span></span>
-<span><span class="co">## 2918 2919 2920 2921 2922 2923 2924 2925 2926 2927 2928 2929 2930 2931 2932</span></span>
-<span><span class="co">## 2933 2934 2935 2936 2937 2938 2939 2940 2941 2942 2943 2944 2945 2946 2947</span></span>
-<span><span class="co">## 2948 2949 2950 2951 2952 2953 2954 2955 2956 2957 2958 2959 2960 2961 2962</span></span>
-<span><span class="co">## 2963 2964 2965 2966 2967 2968 2969 2970 2971 2972 2973 2974 2975 2976 2977</span></span>
-<span><span class="co">## 2978 2979 2980 2981 2982 2983 2984 2985 2986 2987 2988 2989 2990 2991 2992</span></span>
-<span><span class="co">## 2993 2994 2995 2996 2997 2998 2999 3000 3001 3002 3003 3004 3005 3006 3007</span></span>
-<span><span class="co">## 3008 3009 3010 3011 3012 3013 3014 3015 3016 3017 3018 3019 3020 3021 3022</span></span>
-<span><span class="co">## 3023 3024 3025 3026 3027 3028 3029 3030 3031 3032 3033 3034 3035 3036 3037</span></span>
-<span><span class="co">## 3038 3039 3040 3041 3042 3043 3044 3045 3046 3047 3048 3049 3050 3051 3052</span></span>
-<span><span class="co">## 3053 3054 3055 3056 3057 3058 3059 3060 3061 3062 3063 3064 3065 3066 3067</span></span>
-<span><span class="co">## 3068 3069 3070 3071 3072 3073 3074 3075 3076 3077 3078 3079 3080 3081 3082</span></span>
-<span><span class="co">## 3083 3084 3085 3086 3087 3088 3089 3090 3091 3092 3093 3094 3095 3096 3097</span></span>
-<span><span class="co">## 3098 3099 3100 3101 3102</span></span></code></pre>
+<span><span class="co">## Indices of anomaly samples: 216 217 218 219 220 221 396 507 793 794 795 797</span></span>
+<span><span class="co">## 799 861 974 1659 1944 1945 1946 1947 2016 2020 2036 2044 2046 2052 2053</span></span>
+<span><span class="co">## 2054 2056 2057 2058 2060 2061 2062 2064 2065 2066 2068 2069 2070 2072 2073</span></span>
+<span><span class="co">## 2074 2076 2077 2078 2080 2081 2082 2084 2085 2086 2088 2089 2090 2092 2093</span></span>
+<span><span class="co">## 2094 2096 2097 2098 2100 2101 2102 2104 2105 2108 2112 2113 2116 2117 2120</span></span>
+<span><span class="co">## 2121 2124 2126 2141 2521 2522 2523 2525 2546 2698 2700 2702 2703 2704 2705</span></span>
+<span><span class="co">## 2706 2707 2708 2709 2710 2711 2712 2713 2714 2715 2716 2717 2718 2719 2720</span></span>
+<span><span class="co">## 2721 2722 2723 2724 2725 2726 2727 2728 2729 2730 2731 2732 2733 2734 2735</span></span>
+<span><span class="co">## 2736 2737 2738 2739 2740 2741 2742 2743 2744 2745 2746 2747 2748 2749 2750</span></span>
+<span><span class="co">## 2751 2752 2753 2754 2755 2756 2757 2758 2759 2760 2761 2762 2763 2764 2765</span></span>
+<span><span class="co">## 2766 2767 2768 2769 2770 2771 2772 2773 2774 2775 2776 2777 2778 2779 2780</span></span>
+<span><span class="co">## 2781 2782 2783 2784 2785 2786 2787 2788 2789 2790 2791 2792 2793 2794 2795</span></span>
+<span><span class="co">## 2796 2797 2798 2799 2800 2801 2802 2803 2804 2805 2806 2807 2808 2809 2810</span></span>
+<span><span class="co">## 2811 2812 2813 2814 2815 2816 2817 2818 2819 2820 2821 2822 2823 2824 2825</span></span>
+<span><span class="co">## 2826 2827 2828 2829 2830 2831 2832 2833 2834 2835 2836 2837 2838 2839 2840</span></span>
+<span><span class="co">## 2841 2842 2843 2844 2845 2846 2847 2848 2849 2850 2851 2852 2853 2854 2855</span></span>
+<span><span class="co">## 2856 2857 2858 2859 2860 2861 2862 2863 2864 2865 2866 2867 2868 2869 2870</span></span>
+<span><span class="co">## 2871 2872 2873 2874 2875 2876 2877 2878 2879 2880 2881 2882 2883 2884 2885</span></span>
+<span><span class="co">## 2886 2887 2888 2889 2890 2891 2892 2893 2894 2895 2896 2897 2898 2899 2900</span></span>
+<span><span class="co">## 2901 2902 2903 2904 2905 2906 2907 2908 2909 2910 2911 2912 2913 2914 2915</span></span>
+<span><span class="co">## 2916 2917 2918 2919 2920 2921 2922 2923 2924 2925 2926 2927 2928 2929 2930</span></span>
+<span><span class="co">## 2931 2932 2933 2934 2935 2936 2937 2938 2939 2940 2941 2942 2943 2944 2945</span></span>
+<span><span class="co">## 2946 2947 2948 2949 2950 2951 2952 2953 2954 2955 2956 2957 2958 2959 2960</span></span>
+<span><span class="co">## 2961 2962 2963 2964 2965 2966 2967 2968 2969 2970 2971 2972 2973 2974 2975</span></span>
+<span><span class="co">## 2976 2977 2978 2979 2980 2981 2982 2983 2984 2985 2986 2987 2988 2989 2990</span></span>
+<span><span class="co">## 2991 2992 2993 2994 2995 2996 2997 2998 2999 3000 3001 3002 3003 3004 3005</span></span>
+<span><span class="co">## 3006 3007 3008 3009 3010 3011 3012 3013 3014 3015 3016 3017 3018 3019 3020</span></span>
+<span><span class="co">## 3021 3022 3023 3024 3025 3026 3027 3028 3029 3030 3031 3032 3033 3034 3035</span></span>
+<span><span class="co">## 3036 3037 3038 3039 3040 3041 3042 3043 3044 3045 3046 3047 3048 3049 3050</span></span>
+<span><span class="co">## 3051 3052 3053 3054 3055 3056 3057 3058 3059 3060 3061 3062 3063 3064 3065</span></span>
+<span><span class="co">## 3066 3067 3068 3069 3070 3071 3072 3073 3074 3075 3076 3077 3078 3079 3080</span></span>
+<span><span class="co">## 3081 3082 3083 3084 3085 3086 3087 3088 3089 3090 3091 3092 3093 3094 3095</span></span>
+<span><span class="co">## 3096 3097 3098 3099 3100 3101</span></span></code></pre>
 </div>
 </div>
 <div class="section level2">
@@ -557,7 +546,7 @@ <h2 id="plot-anomalies">Plot anomalies<a class="anchor" aria-label="anchor" href
 the samples [(3, 4, 5), (4, 5, 6), (5, 6, 7)] are anomalies, we can say
 that the data point 5 is an anomaly.</p>
 <p>Let’s overlay the anomalies on the original test data plot.</p>
-<div class="sourceCode" id="cb31"><pre class="downlit sourceCode r">
+<div class="sourceCode" id="cb28"><pre class="downlit sourceCode r">
 <code class="sourceCode R"><span><span class="va">is_anomaly</span> <span class="op">&lt;-</span> <span class="va">test_mae_loss</span> <span class="op">&gt;</span> <span class="va">threshold</span></span>
 <span><span class="va">is_anomaly</span> <span class="op">&lt;-</span> <span class="va">is_anomaly</span> <span class="op">&amp;</span></span>
 <span>  <span class="fu">zoo</span><span class="fu">::</span><span class="fu"><a href="https://rdrr.io/pkg/zoo/man/rollmean.html" class="external-link">rollsum</a></span><span class="op">(</span><span class="va">is_anomaly</span>, <span class="va">TIME_STEPS</span>,</span>
@@ -587,7 +576,7 @@ <h2 id="plot-anomalies">Plot anomalies<a class="anchor" aria-label="anchor" href
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer>
diff --git a/docs/dev/articles/examples/timeseries/timeseries_anomaly_detection/unnamed-chunk-10-1.png b/docs/dev/articles/examples/timeseries/timeseries_anomaly_detection/unnamed-chunk-10-1.png
index f79260b1cec5f7cc9b463d87aca320ee654e1a65..383ce9364e0550ccdea02ab0d955640182a06423 100644
GIT binary patch
literal 19143
zcmd?RWl$YmyDdr*2n2#_Ah-v22o{0|cXwF0yC>+vgF6HWZUKU`AOv@pg}b{ur{Vqf
zclNG*>(;&J$N6#2s!E`m*|U4jeq@aCJiS8XWyO&Z@eyHQV2~vxL=<6QUV>p@Ui?CM
z2^`7YdG!nU@y1R<!x07s3FGPCi_v+XI~W)e7)cQ!Ww*4wc~>{Ai5vJMS5Qr!nuNsX
zH(3EFS-7HT^abjlu&_Haby{6IEa61ql$FnB%E$^N6krv`=|n|fe+J-_WCf#1e6A!l
z*^N8BHF_0*w(_zuId<*V{mi|)f3H6^HvRe_nV*fExf21niz|{H9pDE{+Bf~zFfdC9
ze(zvlzLFz=fJ2ev`}!^)(^K-cbwuH0#*|ei8bcAoSsLfdT(M1QLjMji3#+K;FeWaj
zq%9x2oiQ%I)kvr>3NAL(W=zbc@ldl5qVMxav<qrmUfML%o|*t=tJl73!e*CQHO)5(
z=Si1^6m!37)qYimQeH`asv5OG`7}}8&C!6$YMnNt&&6>=nGHM}49q8f|Nnk>BuOt|
zV7UGM7ylq}|8~QqF9DhGqm!9K6|GX<mJSjO%pbUd;MTokw>-Wh-$hr8^Rez_Y9IAF
zk8=wM9q{OmRQTU++}+)&Mx#?nl4Uo)Io8n7@KA#VHm=wMuRkOtB-n6;lW=H=kv6^B
zXzGWM|5+KZ$$!+&nUmmzu6ByceL`&?YCs;>sn@i$wEvt*=f404E8^vK_uG8ztPHry
zlHWud(bKBu1OFEn_g|jk|EGNd|G^t;3@XAiLJ4k5q@1^@9WlcaWfaggOX<XmrI0-Q
z3<l#FF(&p|RPX3Wz^SN+Dqz71Jsnk6QNil4;@ky}Y9vI6<t(Tu%VPzybjtB-b3}Hn
zMpJ}jdTT3S`QUf0J0uTmNX>PLDy(f5<dOrcRu)DJSX$E8OE`1mQ!Sp*Us|b^#}ZXh
zkwv*o$*b39Bu=fs0qh{Qr@cp<CaP=Ij{C~|edGL~91BXY|GzHb3cqPgYzi3D;Ex|9
z5TwE-m&Q_6I_gZ#osWztaex&s{qNhv|Nq<mzwwNqKP3tAdwPv7QFK{dCh4ov6HF^J
zEjwK@?jNG@7JRli(#Zn<C<Epvp9uT2t7M%qnT49SGNkEbUhaX>D1M{P#p`s!10Frv
z!|ODYhqQDCFxLjra2_-jV3WuBziqKtc5zuDwcIRvk4np2JXRY=%{AtSfYqTFfyCbR
zPEzej1bERuu(981ottN1;b0J&x4v-(&Lu;XW@ghu58Jk)z{oKY{t`+UF8wkZZN_~q
zAhAvNak^33Txg+rVv(UYUl_j4*8QOQz`#`2gspSWrIAT%lzm{?4orsIT4s)KB%NRa
zK_VqH3`bXRcduUvRF+p!NiY#wT0ufm0Y2;b_E@6JMR+UQ>1|r8M^#an(b+WrgtyM5
zw&l8ZZ*8m&z>Iw3m0jU>csJM{iypkm15n-W=7mPkaX#Fw$3Nbw#JN~Yuean=NH$xE
zp}|VVaK8Zwp|Q)HR-Ak~ea%MQfrf4*9YZ1Omkjs@aK*Y0owwI9nI3f(?}gAr6xMzc
zb>azjxP@f9g@_yZprGF%iZCN7j2J3Je`bKwiYBjgmvCk%P46FY=8T9WA*J98M~tu1
zWkLWPB#{AdiYllTGwjxCN1(!!t5hY2G4(?>F}$ogHJ2RfrKvEfiL>~KqOxi3*dz1I
z@37kNjKYtVP}L&H`zm{kSN`!8S->5)oiyoj$^>Y}b-r=ShaN*3;!Cv~yoWNS*f4IL
znJZTMIc84>54e$fzfltVKtQe=H@0frqYYetyC{H%hB#bc#j<?UCB`{8Uw-CkY{f+v
zy+Tvn7Y>G%>Ro>oGXOk6*Zr}mpY^E5@imQ?+q)QbTM|IP-eUYj{Vnh)?qMn~?S`~w
z_LushAOi#ZnIH#8o%toHOd~w#7MR}|v!Mw1@m`4e6%33WZ0lPfilV~;K@mnAt^fbN
zKYU}!b46BR>3ok8%p}z?Z<8}Z6RPdiG|V2sLM0iq5Qv7((g|eU35$uw?}$Z3q1iMD
zeosdWi;8GMp;nW?(Mo!P2r)=;!4Dclqwr^QjOrqn#1Oo#d$yInb=K5@#JZ@k?Cv22
zDe>13q=Fz4gx1#9X8P@8w-VLj);}G#Su~+VMQQI3IAk2_wFNkh80_k`sbWK7qvVOA
zx|Z#Z#<*P?(=!d<qrm^`5{D>*JpzMxjvv#7goHeB6<haRv`STF1**rLwcpbK=I{sU
z|GY;#4!=_t)g`OB&FA`>Za}+w;}ZuP_fr=7dh>~n=NC)Pr&3U7^x?z(PI;i_exj58
zTMt~or0qu3dnTFkn=V&%EqI$U=XpD7%qO)n?<(0cQ_k>D_o}95Kztm$Y&s3QxtxM6
z2{nx;B?Ou&PG!xSTP&|&KIwFvG}m8lM9sHl_?|D<SPI^K)FMAA$#C1uKas&SZ8|mH
z>Fjzuw>H1bU#QQwp1CTgMrBb=HZk&;(qlLSz|@aOL+XS<8Dc*3!P5ECIL`BA(J&nJ
zcb+XZHpzcNt1e9%lQp%KU8uZ^ld&nc)F0fRkOgqk1*^DQjygGdG}`D-$vO$+#fKgp
zR0huTrY27db%}0`{6Nd34on*)z>^D-r+-SpZ)Vb6_K2)4RBaq$;=6PDTo-cNY_G`W
zL)?51&2mqAcJM*=aQpxSEhz3!6lAMkuWq}!pXbfms>6skCN&BseT52f+q~a?L+!Y~
zz5$sP)Z_QCQXkH+)lGp~p72y1c%6ru9ms;npYCrEBQ@g2p)dZHElN>w7FK-2z|k&e
zI+?%|w8w)!SE=J5(UV~f7ror`3g=<LRIpmwOvhQ3#<fsd<fYNKt%0hTd&{W<d&;xA
z634k~v7Fn=enA>WMigXZ4!pW6w`Y$4RcSmyA$A!cHnXs>JlQz_cBH8k)>eu!pK%Kr
z8QB@*PrrE^R1}m+07d^BzsQ=K7ecKz{tJZBuY&$<YonALZTx;iLIVGV!L=ux2O2H2
z7Juk-;;ON}<hp;gQG6bmk_b4ES=~$Y{+VOs*}vt@xpFfqwN8DDb>bJYCML5jsTX%=
z#HhT=F3l!@ec{)AN`nQ}kl~C$B`CzYg5t1v@seT5NPnMpzhMf7JlG@mDK}-XB=)V3
zCEz(W&R6z}mRaU+=^&s-UlJhZ$0nw>eF)7qPa&~8c8)A6O3AaI;MT@o<rgn<CB+7^
zn~DlQ5vt%Vw_3z1zm49?l!_mTIz7&FF5Q}ivgPMN-51+A7;=gx<_><Rs4(XhS?D)Y
zwdkkhx$llNqGc<!_DiyetnnutWA@+O<!+77svmy%;P$9^H{PbQ%G?RihJ2#3&@82{
zKT;a6sN3Y7aKZXG+1VZ0WPzH9HDuz(r{8x_Iyo4~5k5;ufSI&nOn7K$5d2#PbIns`
z_b)2mS-uGNZvh}=Stpmlks31dRq<=0DC*}icYilu<r*E1M&TZ+u6a8W6S%A&<!^w}
zAGNFS7K^o@y}7fOclGS#6?*WPbc5Pe{Bp(i$zhMkz`pLF6Z>mG*{>&YCmSO!9w98|
zdXLmLw8;Vnx(n`d8sIlmf}EB12R^xr<VKAeAz0#tIv^1Wbwx(v?^ZZP8UX8p@#2pk
z_`>QEyzsEmz3~pS#)<HD!I|s6`pRDr<pY)}-w;y3f6tiZE&x@7A<^^f|EyjdcY^KM
za=!)9iwV-EGI_M!-uNEi=9LHxSTYsJ=;gfdJh<OI!7VjEE9W`@Yo4}xQo6R-%Q-T>
zJ^{vzQ#S_{qp>Z8$Te$SEk8O}YL>ZsPR!-sh}hY$KhYve!x?vYnF>rP-J}}T3xn)M
zSVLt_fGFBT67hdD>5Y5G28xS|yQ*w5y{=1LSn$k{`Wpy-xQse5I|2q@)#;jyt!e3X
z-(UV6dGs_IZGJeMtMFY3aQFAYBM8<7()yP4Y<-06{Or$?lD(E8*}6h!sl=i|_1n$U
zAwk~_6hxe`>5u1giA*;h=NI0+)IoKlN9}k0->G%?j;7*jedl(;TlYt6`S78>-ke;Z
z72#alW=Nm>FCGZ{a$bn-9FfjI`-t`SpX@2!bYB`yh7L5BQ%h9WhVC^K0LTCf(IoR_
z+jU~r0_$!~jolvFi0m{m$+-pwb{^;D7rog?J#6q=sIljIEy#0pS<^Dqz~HqWMldUP
zvvEFrc!<x-<oO3Y7cM=z_V?VecA@GbwaZ1&eP4XnCPRL{o!V89jq7;dmMYC{J+v_r
zHN$J9CDOK)KA`#w?IY-*c5=RjLg)OqtFlWfF3Sf37??9^3W{`;jgym;SaC;;&%)p>
zeS68O8f&>lLPc=dv1Fa}j>80pq?qBo7;4HjbPSbvw()-L)8#f(ri}TNE;cnBR7<4S
zrpPASbNBVopQ+ST;n*^QxZg(@K7_=)Go3y|5a0Q*L$&Z;pEo&`FJQjn=Gi}fyg#JL
z=6bw8v|b!YRN0^0)O9;*h4<Er8ZoAbnUeEyvfmsMe4luBiAi3goS`+ppGhYZ_gQ#e
zlrtj6SW96$caD8-if!q0|JSZXa~`wP3;(`{a8I}6DF*UkJ;2COWYRyXWbkQVi4q^q
zGbBa8S>lxL<|)9UhbSBii9cwwpAJ`?r#kTMqJ$qgSLMC9M7de2w=ch7wwh(EZV0gT
zfFmY;y#l>I1{=hp=AP}orz4)JQ-Q3X<|+lZZ}DFg*VS>evtyxFa3|)ObEYy-<#nEN
zR)gDu)kQ=^wg7rsmNE7<SbZD@wW8qL00;>cwY(U$V)zS+4gZfI{E6-hYa>yz&ExQS
z>c3jGNCS%*#83ftrrEi&+$PVKxEW4Wr}cy0Hx^U<^|$g9A%!EiIOQhRDWY)u9;?vU
zaz;1qZL%mg0kM?T+FB5vF7&szw4a1X<N%ZQLp$#6YZ)^guE!@=&ktGdF`xf%ZxwO)
z9Y;<CD+NbLWW*W_o~flIBa`}kt15y8eV7aRLVwR;K4YlB?0P@-yFKhM_=@|(tbI1k
zPvQJd?AbS8_9thWo8`%$5ivUYJIy)P{XQ*5Z+?LVI$noAKZffushAdiA0D-at7Jy`
zdP|j)655wy0fwG=?e;;-g2SeDDq(8TW8*@SGOy0%6}8vnUzMzHr2x$S@#3aSzs8jQ
z+T&c|@ElqSA7#a#1O|sbs9%|0=_FFxu_V9)qZY36^+73Z37Luc%IS<jT&LQ-CYr$d
z<Rg3$L8QMor$hKHbAb!O{HZqs{*U}`D#Q+Ur@DuS2MDy0TTXj8wrtewns|6+e%(4I
z7xi<6NqnK=kAjJU$ynoKMyis>MGS6mSfexmA+7@H`)U_Fc2Otw9k_|4zFaN&qdpu{
z7MtD=UzHKe;>VRx>JcBhOWoBR`t;C5?eiKcw@`_Qv?t9TFk@C!&I=CdWMIxhOBMLg
z>4t6={X!{k&FD7huSkmvX7Bg=3BldoS5eo3Hjb32_1@~Y%$qeGbYIA=Q@LifbRDOV
z_uC$}SuoFQyGqP8H*8!g)ek=cLHO5C!AZpY2e#7rtG3n0Y<QwP>WqgWhT1prAm{1d
z)ia5noI^?JW#5~&!%%AyA6O2WiTlm(wGNWmANp1oiv2Wt^4tZDJt_rTb{{A^T^NvB
z9=xi93(kVYhv<_R3&bDC^XAKKQmdANrJ)zAIYt7twcA)V4s(RW2Ws!NK>Q!=-Q6<T
zH#bpb=e-8kyEY4I`9asa$ukG(qf5oe!sQc<I6x*)dn8a?yy0l0d`KH{ZeLT-X{^?$
z^_iVaTlqVTvETo><myrn%AXldI4WW3bJ>&_@;Atry&SqtiEM!0I(d|2hl%d~pqt`2
zaGI$*A-fZM*+xirBL1%9z$l}kLT;^<E_;D-bWp2{_M9`Rqh`N)p10aM)x4X7V(Z@P
z3k(cGk~d^j_HlOKMeuA`*q75<)}<>#I$6N`W?Ligl+S(g<I+t6G5x(Rbl#*r41STA
z=ix|3F1*1){qh`Y%0P**@F>Zg9R9K}`7lagC?-FF3PU8?9G{$&ru#!$#*N&LHM3ND
z1C1b~v1s5-bemxnjz=P3FyCUVs?)ApEi{hL)mk1OJmux(A0O`F^eUORDtA{iJ)Q24
z4yh$O785&N_P#I894#zD(cRZg5A0|~wCy9@rrK>a=Zm~AWrA*)O0TA&Qi#ePPQE@$
zWUS{6AANRyGkMJ8UiWS03JZkdPH9h<lgp&AZ~xV8#Z2ro;Y{tY5QKdfv8Rqg*NyvD
ztit+MlSXWey)y}5Q>KIqMA>CDp%jt=`=sCUi;C#`aKhFzaZ9l)=~Jbt0IEjP$BJ&R
zf-2ZiH*e$L`lm-gRwOhVFQSQWedvk3C+tLZR?Eix1TrsZAm!StpM{b$jM=&F;D&QV
zlHU`o<+E6LQFr(tdo9i|Kf{G+F<PNhVl2BgZrS}ym)?8-0lfU^Y83X<#Xso0==ab1
z+jX5blsy``gAvARrKuNFC~s#W9#R^K;?xn6$QZD#Sc@#x(EzdJs1V*Q@HM0SR`JoR
zZ+B&7Wq5e_bf0==(E22eiH4^8Ze^DO?rvlxG`pb4f;06MgjCNg1ylt!G&Fny!2r`t
zmQLII{Y)AE6`LF9iQd02amB{|H+4)K9aXG?_5<v3HrO_G0EdW(5$-`^7R}E7$6GX(
z>Q@@2Yj$Rym(Qy0SGgTW=Gaye=e^G&JCg$*iBBSGtOsGgh~d<M$j|*5NyhwzY9I>T
zRoxKvGbjdh(<mA{Psu_o0RVkvdd`<5iPk<i50+?y0$88`X~&ZpQ*zz|Bf3T~EI#fZ
zVOX%>v6Y@Z5_1I;%wpYz^Uta9=YmNg5<4^L;Bf|mbH*nfn3x>^rHwAs3dUUy&|~fA
zaRrk82q02=ZE15E5sQx>2Y{eh{TjVrgd+Ex;&9pu(A5F&{Ca9Z(@I}+D6L9I$24|f
zvh%D*yK}_pQ{^b^3<uhA0?6QMQr|CG;w)oeq1vBk&tvLIu7d<&^$)}bZs94aSa+%Z
zb3q~|xx2D(mJNq`FYVEN7f<5-1Eee?+~p|j-n<t6^fAV)KRf36SKb4-t+z?m0Jax*
zrFp{kL#Pa>^eQSGGOI<;!qpEwKvMmqvRY8IbL@5n?K08W$hU9m+L!X$Xdpqr3G2|K
zoY3KWT}ARNPf-HJ0H_rwp4UBt`r$*y1l{%)x|Dk%+UymEMM+RLO;M4<37WIa7@%TA
z4X7glf<=#ULwow9-v6s$b?I{jTr^EKzr!&4Nr~Qh1qi#<V&HI$RBFuV29wz5N2{D;
z0LO+NVh#aING(cur9}Z-Q&7?hS2bQp_fDo{?Wc===05``mi&_YL4`)9WQtsmrLMGq
zU8rz0=~H1eUQ&@U7{T)pKJTkt!s~baJ+z;l=xHk(7<3MwlH1@Yntw8^(9S{_cwB}%
z6+r)9=HFZQ2vZqC>A8Xld5c^bfKjIG!}>UX+d5RO-=qU1c1ijbGKeh3<BW>Td6(+s
zIx+(o^KH`3U&Crj0pSg0zS3;9@bmN^*yr@ABryQ%(H;geF`xOLR%wd!9VwBS2S`bQ
z{PP$vz8f=AMMOv50GNeu3|J)h$PfQya66GVSYq~$E5jTxX+E>9R(fYy@~Y%?`pHU{
z-6u=><c|tsP0eLA7x`(Up8vdR1Q1!_5DeL9rAw7z>eJqQ5)Rg<A{7vlu=<jcXS49*
zX*I*cq)8#*pSt(J^uKU~{{W-(fD_3IrSLWAOG=dc=5n8h{dgMIZ5-3!L`~-0rU{V(
zVC8G%R&Al3`AqQd5`Mgw0Q&j(nOO~OB~^J2&T7o#$x^=tA+>r1Z4th#vF9%=dcKEO
zf%*c%J97~ws;a6*`wNB7vwtE@qGL(<X~Pee&}k7n1nOx&zI|rS6y5RUdE@7zPu$0w
z_dE`bG--&ey*-rH8W{MP_|KvO^vk~N@O}k+G_`wu{K?nM?9lpELJA233d~OG?027>
z-Wy@V5y9_r61vmC(Vnz00j#slProc^yqF0vpRIrzCZ(tS|6w4+tk#83;d~&rAp2CK
zVE~pSx8wQ_-8{K5sZ`>5%nM+QoLLMFr=?JMq3>z~U}6RQ2pgwGbv71b9x?pS_Js`R
zciq$;sS`NN?7%&K7V+<ccNCI4p6}2LEmfDF&m=_HNRfct0#y)1IkV)(vobh<%frGt
zM)>&w8`3jt4s0v}AMB~%_t$Sr+!n?wa-OF;C-%E0Mqr6(=iCSO;xzbqFYa**^g(6W
z<iSyiw40d)IHrfa-N0wMy5&#D`S1txKiGhI$eB`7pFdq?hew!1wf76B&)tS?r)q0(
zRqEpH-2g+xgtS%k<86O=Bt>l$PT*0bJp+NpO9ikmn6@;K>dl^d+0CV1P;uE<<9lG?
zLoY{+KsC;Ny{{u$4l|BjDXg_C#0y=rRh7-Q9cPnRj5_Bynk;{#ePz9q-0_7&FWh}3
zL$%=qEc*o(n5atidw%;R61s%6G-5S2_q#wu73z<BalTjOP}>6sI)ztV3xdp9ilpHz
zQMT8cI*KY2Fn2hnQv07svy^o=7V08B&I(=7pCTHP8)dUWIC}K#{*xLSnvE6>2pMmr
z6|iK+07*barT#+@OQ#C|9afHc%4K^0tIninUG&n7J~VM0iV53l%z9n?vw;q+wSUfL
z!i<EhRiC|X-zIz5k@`!anjmwtlBUw}pR~jljns?&^{n_PC{QlTXJFYlcC*;Dvz6F2
zQHS{5p9eR3n-(1mbOEybL}xxMv70}%3seHPK#ot{S^|_X!ZSOML0Lc*?Z~H2WK~&e
zj6JHWO^(xXf?H$_sCU>CCsMp=mHzhTDWA|Y7>SfLjiX~=lPPnJjMG#Le-|aONWf4_
z(dyljW=8TuBO*HJO9%}M6aM`9ZO~UQCEbZ1qzA~BfreUJ(ZYtCHk*TL&s7AdHbKW7
zrp%ZXXX<~7%yMH7E5~j?N*0VTs-5-s_dhSo_G??#|EHj!q^SvHcOGchJ0sbAHX0g<
z!_Ob4t!jpc<t1Z)JTHLI{!Q`A&N#j6s!ec8i+OOh5-*$kmy@vd_9`m;DqP6JLeWbV
z-g*W|foQ5aBg!;Km3pqOcZh^W^?TPXSeUOcZzM?YXCoTe?Av@z>FEhQ4wq<z_%EiK
z7B50fW!QhtIp|%cT^~wkLg>yHW@o1&9C1g#(sk<_cR5x=lK=66*gq3$Shz*c)bMwL
ztO+K1^vS)K^Xf+yo{p=ftD>qa-grVx%p5?$;^vDdL`Tnh(y~5kg(*Qf{uJ)vyS4DP
z6<W0{t$fl5mQZ_3sxXdnFja}V{S;Bhg%LK8$VEiM2tQ&*-*h29$1@LvR<G8#{y{M$
zW}d9-3--?kZ(aij1Fm2zRJ07Gbs&n*L84iRS?}$3C|%F3I+9lOvv)%L^qcVv9q3Lk
z>j?j&7ob7|Nm1qe(&V4@Ao^XCNvi1zi|iysY_Y$B`ASTh#1u2#P&%DWSy<G7_>^ys
z2w+b;Vqfcs{tIg?aUpDk1wRF=D?-2Pm>*+6YIz9MZ>D&jz!#WcVq2CCd-gf7>;Qoz
z7elKSJC!=N$f7O}keC?X`9V>Cl1^cu&(<U(p8PY66Sj&NaCacbNsgIyJpTKjFJ2Hb
zwa|}#otake=UaBh^MeWkYX{H=nWFD~G*D)KiVDSO{XXO<gkmv23410ljGr;qxC#R3
zF(f?ux30*~yAyKrl;<d%9-x0wEF|1x1)@HgctTlEfr$dPN)HZB>QxyDne%n|&u$Mj
z0O&iT`ay@@CL-hb2~YPJo)4+~a|6K9#eR_aRy8ndb2yL@8}We~Fhc8ppg!LLl$Pi{
z1z)Y_%q`!=UZ)QVzrSpGB;;{e`zaaQ+t>H15?P3O&JO8~UxRa%%a#5oP9VdcEBhmD
zK&wnQy*N&*@+#8VYWVKY#$L>XPV;H(O$XvB`P>m*5&#1iIqLLvt1TwT&C^R?w^cKw
z5^JFU*%W#gG%q2B**Bn2YlS$c3Z#{c!rIm*=iNz_V$~jp3Co0;THBTrI}~(68pb!f
z3v709fC@Wm;t!U%H+fU9{cRTQNwhm>OX`XuUbOE0Ep*yKz-g5<Z&S{<8$z9-o|)t|
zb}0za-0`{HuYIK%WUv(f*$tMCNlA5C0f3)Nm_YUYN>4#y1itZNSSl85cKzt1QEA+>
zHIgGY#F08c>~m$n5O2u(t_&iSaT(tLiXSfx9_DUNr~wS$S(Sf}1<B72)P!y1N3x|J
zKFuQoKwE&oq=hNy0|pqiW56sfD#|}m%?5d79JoFy+Sdv~B`!1N@YAm|$9fk&6K&-@
z=D|+~0Dy)cL>awQ+1q3JQ3KuF?kC2F3lQYj>AvEmRI2!HxeY67>zkB@vlpM$>zdC2
z;54*j5Ctt34o11Zy`18YhXB5Y7=a(#S9sSX^5j=vEdlZgI7v{dcr=OGET`pD#so6t
zRiJ(urC}HrZ7D7~t)P$GrnU|M-+<<TGXBwD%b)EK=sqYX+?PC|h(ENjj{E3-e#l&N
z0G~)vajOGNPPteB{*>#Q`hicQ2xxynZ~PgO-hg4$>sQ{hI{9SMOR^B;jrX;;%Mke2
zHnTjSp=L}z?|jR;V^va$IA~~`p6}>02?vIOsz`Kk2|K{egG^?%_PcvcFNtj%RWvm<
z)z$Ip{st`Rw=vYBM>qIH)Nrxc5xfL8sbJx?BYCX7P2*(W$6C1xO+*uStBvMwvn~g}
zc`bY<Qf-^v#WfSV7<XM4y)P@i&I>64*m7?)9Zv~opvylWFfdk9yJN=@poe2|rFYvY
zWdNj5vb`E8HMYQ)@HlwP7uEUJ@5pX-8W&@w?d@p26b=rK0E_YP-0~FK$oP5cC{+`)
z+mZfa-z2-t7(218IT?5Q-{7&rrf}!oRvWzKTUD=iJV);LK)ujM@3aFbO~+7pA8Pre
z6cqt9@1LfTs4gSc<TD_H(D#M2`0Okf*#iWiRex!%Ml_s7BjMWoeQ=vWg}Q|R{v){%
z8a+-}%a@|YI=5hz<=e&1r+P|W$|WY!2mvj_fei+i&tukl2en`~-Br!7OZn{ar!Njw
zc{0T%3<t_FlmS80`xi(aVF0HoDuP;Fn45=*64*Dg-nRTrkm(#pG3;V&w(Z6RDi4>K
zk7XCx?vILzbz?VSU8@RMftj5n>=DI=U~sX=6|Go;NzTaDt^kpKI15m>k67KOl2qjI
zCaRO?d(GyaiMD<k(@a0d$!(^wd9j#I^S;*9wmDlmT3YEr{Z{N~V^eI-!6O+U=xG6Y
zRW5soY^g1S(^YFl#f$~P)M+cLCyEyElcryRM_p;wOk0?Y_HO;Q#)~$H@#ni&`=1G@
zf4Et`Un#^BLQ7g0GA{P3HM}YW`m4Y#9y^ST9?aDsW2^piq)jFFcjyXLASMwmAG5=6
zf0c}C;Dprc><p%L9@TvIBf@GAz+f<f6MxXI4MqyZ_c2;uwlHOm;Zz`IK5&%iQ<3Fl
z%NJg*fI4B9$Bf)dxvbnDx9pQy$N-+6BOc|m^bI~c`wr^M&*b0g)&()<tV**o=_s$R
z5CyDW3kZZ=M4sUb8DdB+xgkt4G>le+#tf=)U6;grb>_QKfi*Mb;@}VtZC^I@!r}zW
zTf1{CfN4xE`PJ(*y41wesm?hGUrg!AHkw}z9_TDuK>7aacRO$go9?}9;y3cFzxOKF
zY4*AfV`{#dcU`=8rD+_`iJGR20BpP8U}0<z;3)Dyl%>iBfNZNjJeT!M99TO0sQhPu
zH>80-2F(^NU@=Up!Fqv{(@`(7ap#`3<^Ez8ZivrW_~0A=qr2<LTF(PRkJ}gx>Bpe>
z_z#Yzv3b|7>j1M3W^dOQPzxWqGALdn<Zp1X=jF(Ty2JHYatU}I*;Qnx>4KL32HGOy
z57#T~RY4kUr+>8$VXM=3@^;3w77$M7-&M8`(ZK`$`>NGT3=>*qz0&uSg-lW8c!0c~
zmSA*ou*cZ~ADzQ|)L1S$0xZj&_N~rEUP6L3ye<Mf&QT1o0v|m=Akx1<*kA*y0R`r4
z7;t3S;pq{|9aMBf@43?|N$*#U+%YqJA0o0(iZ_{vQr)krr*x4-UMNlxDQ3P9`AM()
z$HDPs#T(#Jz4yBg2TYMJ49R>KXQnn0G&C1Cb91lS6Cu-SWHUtWiF&r+d9+Ct;(NAJ
zn|)=xl;L1d_W-I8m5MO|#KeA|1p`3JmDd9})^`f?gSq<hgiNoCPpTVSjwzfW#EEFW
zKh$OjdIrdlgWikgj<}AHubTC0jV*W-_gapPjfqDQ@mq`+02xll5Iu;u6EVx6WIB)j
z3h`AdKxpuZhz*w;{}Gck>+<g;MYkVYgBu%uMw6f^fJ?W=e7watLZe~7fqQ#<gIuj1
z2awfLDYg%CTfBdtMCF+-2Ht))hX$|7EmEt*-{^wyd|0CDhS>)|RUPf^+F#!Jw9i5+
z2~oru=AZ-OPhARK5!~-l0GB?uo8ZpP%>`8@XL_3O{zn_xKQ&nU@@MSV(}Li?d_PFI
z9nljPD@;_+_u(dz-*eV><|A+UClEayU9p~_SKI`KA**iUC>cNhJq8ywr=>yNH7A+k
z)P3Q@c`&gLfCUQ3(fXV5+0Eo9gNePexYfjh9~#XK!4xJuX+~_9)id88#6`w$KBwv=
z@E0}Q4Mz{A^XlDf{5*jCY*j;#u5_xXvFyq%2QvQ-Y+7%#glHw}>Q=?$8quZh?t9Vk
zt@4RC+o?p2w+95aO(TM{`qCQA;`0q5;bN<=Rsp%d%a>NxWC|v~((>!S!0nQ94Y9rb
zkCL$Lh#)mBcK*4>+wnC}gp3%00_d@ii|>4onD8U9iweXN4?R9HQSXQowv~B>v%;OV
zr}AXNbkDegcn8R=%>aUwOcL^_F<;4Am>0B)PooG~FZu}Fn<~Mm?P7X$*6)?FDJT>w
z9Q5|OCBs}z*L^*M-;>em)$1IC{(C@Rp`(izd;~+8_~wCny{r)P?Kc%HS7{&0O^B-*
z)K=lz9nt6%3DMIQI^x+E=?+ItC7Z#QBN|&?*0K{;Ge9ky=9Wc=>*pVFDL@erP>Fec
z(jV2cj@Vufz=gC=%(t!0-nRk+bU*CnhKGx-=n@7<))S`9B+)qI5+W8LuB|fUn=Xfr
zb5_<^@7L_O84d40OP3r!i1*;PJ2<W}2~b6K9ku{;)b3^L9(*Nz?1=iY|H$~{s_u02
z{x@CoAqq+n{I_LG->0-(tvHSvWZ#s=+P-tQH`))Vqw4e|CX1Eq6QCGMmnqt5IRO9<
z&|51gAtM7BAQ@GwGHY$JtU^V3&;1>oqS0?#W0f(Me@k7eRYQ>nUTK!q5j2aAX|V4*
zOPj;2-1FW)UhSfm%ed>MD_`5#m~6tXFn3%{zl@<U_hel=Y|d)nOCthAD)>aHYj`88
zi6RxoM9Zz=tcl60vY&gaMqMqu3Tu<>WUL{7kw+9<#YCHCYp24z8>IY>di4l=dZ!8T
z@ik9_e(!6ox$9vVDb{8LMdxhu;gT#n6yh)amw&W^F;p@sKV6Y=ho0IxK}0B*ek2r>
zlpMATbH=PHjh|vVyd160gk@G%2J+t5QV)?Ix^QpeV&kIXzki22)o1nni_`=TZBI|n
z<FBkbH&VXOvsGb)LzG(4(`7lAj9Ds2HDez_$s}XsbqcHHR8%w)ULEfZhLQK3TrY4|
z%O#kQtrQXYgxWXGzxMwYobeUtH|`wM%ibP%?rI~7>^z6TXIV;<-Bgm}cfLQLiXz||
zTyqjgo;FsCD%)*vZs2OR|4tGYB-#|8He>P!xb6EtnF=iVa`v9}+GQmr^tHDoNIAJg
z-r=<oqzcNlHZhytA@D$>^f)!b##hh%>Zyf@aGjg8mdCNm0mmW!&Gcz%dM8k7T+;*O
zfG>rx-Zb%td7zAzG-EZFM(y?tVPnf66OGbDsaCD-Sj|MR>w=4OYTW*8t(tEBH040*
zz)WOmHwb7D3P^NhJ)e#w<lQPX4FNYe@i*N~^;UhR+}`CC#~d-px0G!S;kl-+d_SmX
z-O(}=TDtOW3!=fqe=^*?LMHAGRkCfW&~NqYEyw0{Wiu-}+2;aguFQL9+mc$x(*}4e
zE;9edS1Bo}oJBPZbnB}|ARMb~uy%4<P1||RPU5CefiCtEGA6ILe+GzD^%R<mD{1mi
zJL<1|UyeE_boym}T3N*u{n6uW%Tn;+q>3j%M4@k*(f6BY6tFk&c=7vOPpZGi0a&JX
z_I}2)JU2hT)^^bcXbg%EPO?aG=^}LP=Sq^qvZ+X9Q2+IZH&4OncivCq-W%j%lL>5`
z4V`JSP=E3^+whQP(vQ7+PHzv#4GbT3EAvVnieqPTztbbVkzDW$Ko>KaLdXF^nF7Gm
z3-<8!pF$bB`FSeTDud76p&BTb&vNR0-+DZt&OS8ttuSZ{0)hZWG|lij`f`L8JLEQ_
zf`He>YA&0Sz)s^^(Y?0vouk6Bu;2Hh!hwTkag{W}Ua3jzzV01Q-&q+;4V%cjW!}zP
zq7kxepwx1Hx#+9sz8;?=7O`SX&km4_m#zKflCf0QUK>22k%NK?cDK~Tk&joa#E8~m
zNMRQr)86?dv25*RhHS0%NLX4qI26Y*>b%wS`fDY6LVf7FoORNoDrXjd*Ac!=NlA$+
zaHSv5(WuQR8FN)X41^xp)s=BW|A}hLsoA72Y=Su&nNhzIFAnA#PqZ87Wuv1-L`GQF
zO>!Y^pYy4>`gWk!fR#rjQ+)JE1JjUYU!@qhZen-t4b?BFpg)=~awwV9@`iRfACA)V
z?d3i|e6H6MKM(*_xO{OJE3OimUe7~_nzah<X#h~PJHk9mI;TMRJ6G?RIbEkP_Im;1
zsm8PEbvnSslzxnY+z^xS$muZNJvKHA2S2~{`_Xj3qT!sAEZZNRT(gF+(7Yuc>T5ZZ
zl2RjU$5Tf0!K}j9_Us^iAcdp2X%Ha1A(m=PP1hgE8dhUbZ5#J%>gwvo9BXanZ`LBX
z+Fxt*rPikLd%CQ4MigUNRYB=-kOcJr-FRFO0hIp6Z@5^k+`=@1khcbC<tN~>*}Fe{
z1MtFfzGB<OW;NOL21DqhghE-*^g~+=aV?L|V-@zJ1>J(OYKG;Tl-S`kjFho;;nZ#-
z&&h{}WdrZF&$F-^NQj6r<bNS_=0xbK>gwt|Nz_0=@3GU8oMEH%nmSK9S+QQK)O_X@
zDxZAAg<K=O-u2cMb*5M4mui54clY<d=eE<2f2uy%2{C_|wH@U8!24;><u+S1?Ct9|
z8LoNd>w_3-?Dbw#s$bvydy#`6t)}ztyq#uyRvfmO{JeKt2D`1oLZH(I=iLuc+3x|z
zblr~)doLR{!Yoj3ywK#HZwkOyTY^P5lh20-B2K$2nWpy;+7BOg&gNVf+)WSDps!*u
znrG@wPws3zFZMLt8<B7s(>N?u-Q2FEm>_W}DVpoJfV|PY+!j1GI%;8k!~l2`%)h<K
zN^pg2bcqZW9)8hzN@r6xiWx5kh4@B@eNJ7s((W{+6g}EkyuD1$tYOTIF>XD+QkpA@
z=8SR2<x+Zafrb9Jj=a3Zuj=@Tl;I}#qfqB@ZbqAd9WH^ZA14;poLYtw-II5i22@J1
z!{1|KVtQMf_S!BJ!|T3*Jvc%<Z^M{8F7_mePzw<y6>kQ^J7RS;?FN>}f=>^MgCBfo
z8c#%(^t_Hc%57!#N4YO@g`s+755?}M8}VEXK-dF*5fu@UPUEfsIu})Z#(-Bfnwpx%
z$CZ5!a^J$(Z4Vn=q-q|pOzHO}FqmJ?%Sp=EI`axZjx~3)UWq6Z5!G8CP3UFPJXH3}
z9?O8y(;czn|KO2}`AGo2<q*l|a4I8-<CqPNGvmzcA3B_7t>5a@Dkv#g&(OKO*e|{k
zn_C6JU5`DSJE2Olg(!MS*vYG_|L&0>8mbQ@{6K3xnyV$R-GqkmS*Z21{IDac;N6ai
zH`r`5B%3k<gIX@##bX)pi2KW|TS2^!N9y(W6|d_15}C@IiZrS~49O=R7SiKlmRon%
zr$$CbbVKP}whO@TtE;Pe9^1JA0RbOO?#`$50HoCxguME5tY0M`jei6GN0;`iv!r{3
z);+E}ST^+f`GfxceiXuw5-~qb4ippUjcdrrMoWZ<iTfWmZqT))Ra5S-^Vt>zf6z4U
zMM?*=&918sK7tTH{kpnhSq7v<7_~;T(>0~3+`ahNYh0D-VEaZR<Ju9X%Uv8%RxQ}r
z5Q@Sdzx%D2P}x$!H2h30r+<W~!c>@3!C+YXhY}5PMXfg?!LkWN5mI$4p>z^4-Nsbe
z{Us%Wn?C$`MvbfM7s;X~<#O)Aztka?R6We@J8>(EiM1i2NsV-t3pdE|Mp|RP^D})s
zE@L`2P|~U7GJw#u*_#U>RNo7zDI6Re5SDjFbEV?{a>(=z3`9gkK)}JO_VGXk2A?F!
z@~i#@Ic(LU9d^IU;T>gPZ_)Zr!^E1C2lZNSXs%LiCn0Z0#A<+;y)kP~IOHr7HJ`92
zS2lgCJG#&FQt>Z2N~5-WQQplF)f-jpu%a(TGX+N?EhvQea2iI^dFqn%i!Fx}^{4*9
zsB+xqVX2n`d!`vUy?xy_rn0;|dmZ&%u)+M}Rz>%oKWI`wTzKU8TsYxktXkbx0&Wyt
z?smQw(!Z4yS{Q!aFZ!d#jEGRhl1Lg?%X+QHzes76wCkQFsNs9;D4E=!jWs5T4FOwm
z&hFPPF8p4a&Wfd!d)QnB5}nGzaGp%6uJ65bVjqXs)e%5G>b|^dblJbT-pu56J@A@P
zQA^iat1#*U!cx7f1M+)VM`W~q4ge{m`$28%P(VzglP;zOTAbz@jYA$J@gxWa%M_9#
zYj*X0d>Il|JzFl(K=p(S0S+Wp){$abxuIDC2kJA3&Hb&g`nif@QJ-Gl_!3fXRa+Yp
zE`fj&%85rF5rp9X+iN7dv?795V4=ZbJq$7$<Z?(<QX~e&9GqS_JG)F|LV_4F^Q0TJ
z+}}-J9sexqU%hzn*nL!whj8_gfW;*7*`3TJBoJs3dx*$-1vAv_wtn!^_ZJmrr6j~9
zb!#&!s+`n4W;KP8OEN~hsVrfY!DuJXHL-4RBCSe>ji$%xoE{RSZ?1G#%AvUTxspgN
z4(6~O@uSDV5NS1p#wPP@-n%O0WJkp(itD;`yjty4OJhTWW7VR9+*nMyp8g%!$WE}{
z7)VvS_C7dv0~q&pI<G4yK;3~CVIuH3inK=nbS`+aMV{E_42W>m=3{wZAn>R}+ATib
zQ)N1{)t0UYwpG55KB5XUHP!^+yaoiIBq+o@WrP4~DFg%pM4;^NVq+v8W#-<cA><<y
zV#18Ly!=_3oRm>FA{6RHA#U}4YV+h<Jg$AgOif#`p3fm9em$~ve-`;2%Su2IcJIj>
zjls)q@%;-gjV!4-9i6hPVYK4bPa;_{#%KXRUQ84bjUEwkcPAtxpP`~cpg@ZjLW?vu
zDRb^ygS@qx=zU|<De?|k#28(auA&??IH#s<g@P0Xt>5p+&-j9fOsRqri#@VGxY**`
zCMU%=GBVIBg)c6mJDHD4%UZ7^J<+}0&F|1|QGqW?DpCSDfJmue#nYyMBEblVNaQ(^
z-zivAFAoLnj)q%fj=EWc*+{z`&Yh=XM<ghK_JN`z;(;`Gm4XS15WT~>(<=g{HUbVR
z1qw-w{*4F%E}yFtse!+34qv2#?{}fm<l-?SPB-<&wTnKtGkQKa;E!j-znr#5dvPam
zv9TkF1)H}Hs}&$OgGnqoy@@xeoYo!(^_wCJX{;t+0HJkVBP9=jAiGh54|O*?#eg)S
zvu@}e+;+SgC72?ew_D!wFsC&FwW<N0ab9k2U_e0HEVqe?2~}+248N>G+=s-MH*kLS
zuK_OinfX?R5OJf5xl<O0O-n(PSaYm<pR8B~0SWi--=3Y|`3JxOubc+CI>jQy1W&+}
zjEsxqvsk(6s<Iz?M<<G-YRx3Y$m0`5lLz|omGg=?ye4#R$=<%UjloKckDmA<vgXkF
zMbXTRh6RKT8}_NoAu9egT*&Jj32FKdUu;Gx(%8SD19d}sUamwk@`&(cDXEo+&ZX~Y
ziYjDOw4oMs7z{W<XwoS-2fwtCF<8j5<wX<JJKPb%@-$S|o%jtx$wreeLcW`1jYi?9
zRHxlzRU(Mue5lU<R`LCdB{^A6o_l_Qm{DPysA`|2f3z$PS7bP1WGER>{UnK{Wn)wN
zz);;+2_=b#4Zf(~V<jU~AghAf)DGRP#jY0ahg5vupUp_(W}#pY2TR1j#+q;RC%5d+
zt1T2nr+My0>hd4fZxXdyPzsTd0dDhn)ipaUXMw-o-s#Qid6Dc5wYs*xJ~2ILVCyz>
zqB)t1FaGGXW!C`eoCc6dt;soY_*Nu5af*Pq$NHifAV|43@u#n>tk&d}YgAEDQs$n_
zQD?Xr0verIjy>R}K--|@x<*$dVdMFv7AmoT*8s<q7mzbHo8SG<__t%hJQaG%uVb;;
z>^&rKySK7>Eq8WsJ*j1z{-Bhhj1^enF6DD~oS#`zQ33D(UlHvztub(S>fBhq+~RO9
zZ%5~7W-;;pSh0EIMK~ThZMZ{NnBapCI%O?IRAWxZT)teUp#~{gejbmC^Lj#QT$;4V
z+Hn9I)oXZ9F129SSH7`$NJ?uTF#WMRBvc4`vWrw@avjyNJG?TOxc_2DxjocyvNEWb
zHhdP7Ht07isiR?(7=zVt=_X>;0^`+dXBjNVIBk@w$lLvW_fFLBQ>Yaqla$T)v^2it
zYm5K@agmFPVqppwQ4#W{%ryI}7p4Nl$VJU+@7u5L@Ia)l4i1mU`^#cwp!ea0A2bRm
z)A{R{eOn~HRhO4P0tlUnsp%pB{((A!YkYlRU?8BSYAGs?nk+=>P(E!}s(7Sg)yCOu
zIgj9dQc{we$7HDB<NK2fUiJ;5k4{%XuQU#uZy8m3_nxroL64%vdvUF?meh|Oo`vyB
zW7<!BVnit)BQ$^iXj+XzCh~%sB*xPexmXN&96~NHoTPGoplRMtDUsXq+<X7(rD^<i
z)FMcq6#)Bz{{48hRBxe>Zf<C{nbmS={5WTP`0HMQIWCR~PXtB<tcZf2KaG8TUWkf(
z-wR<+CWBlmQ#+Vz){8Sb*LU9v#F@03m9%-*5#3<nmQ585APs>)Ak}1`K=iZ^FR}ak
zRpMedL!`q|InrDWtNXOkU{9~dZ}pq801_hLv^5Nri{!6~XmtT>srB1T;dEnw*mJMK
zv1LlT)<#E2sCB8;UszbUxZj~(d$!ircCrLozi7F!z7DxPXcQF{HR_7opDNdjuLs1)
z-O1A3;Vj|xC<I;L1!LQ$<DhvPU!WU1`0YDIlY<~+0w3TR$pZb<?_dvH8X3HAMpoq%
z<<Vq(Pq!|S==3mZ_{nqngrnpeUE<=;SLrZsFAte4OYW4Zo$qhXmGWgx{Ok_v>=X5T
z+}9_X?>gi38k}-E?UmhDdItzHye~A!94z~@3d1V^O1mEtPA0Z5CD(j+(#IH4FMFP(
zR;tP2dn;=(5z2ZwH`}6Tae0sre@)D7w^U_4TWE=oJ~G4ch|kK+_n~3#_Oqx?lbeFi
zW|?sXxTC}RBvQv`hb=)^=@c<+5(HBCLN}^xU^}V8@2Psgl5Z>Ps!orE@ulW(-ZFYb
zDu*NMiI2D4iLs=t<Qvg0G^04zCgHxd-p%XF!^Lv;NXHau_{m4_(*wdq-^crUUwYty
zkR;vHkhJ#yKlQ9gyC_%n2LfF~s6gO8e@I9JlC#%lx+|a(wV?{mK|C)Vxf0{!<6~kl
zaBy4>EQeg{$AOei>~++dDT}3pi+O(BMVxudVmMo4?MI?qe>H0fpnk0ytL^zl*LZ@;
zq9PBV*y>CDu4mPqglcC@POy~E6$m1iK-D6XXnhC#i^p+8L|j}ggV%I0txzc+=->q2
zn^3;417xj$H<+V<L>_^EzLgWDN(-8+vwytyG^)a+|I`Y|<A?@N-wSXJJUYmJ&FqIE
zO^C;NI^aR<q!q<o5SM+;Vpj<v%4dLkPXlzQg!*{jJC4=8@%O*<I$lXRDCeK<u^Vjf
zA$Q16uj+>r{?{f5Br%CH=ymnu%{Dj}L#=FpyRAk7^zi?fC^`>&jrREdGb{#X>}P8@
zS+T6o6CoFO770AU3UQ}p%^00Y9)t4ekb)xM6BQ4il4ri=$Ts;>W(;`GmLf?-#U%WA
zlG{T~L1^r^rcD}Pu_M=u#lVZ<mtWf?`NLVt@PlRdH4-ad3!$OyV<qW>Y-%mo(W3R4
z@#kE)6t;LUbtV`bch19sr*i$95_sX)GoFt8e8pS5Vg6fvSO?q*&Z-e`xlTN4U#gGJ
zB0$sP6;G4LB9Hb&n;c6@USU!1Q_n0=T`c?Pqsz3Ab4*z#g4J<!WY;pTln+Oi6!^=o
z;&(WD#fhT0gfduupndpg)G?9g+(U?)zxOwYJKkGhK?(ZwqT^FDh@Eo{PVHLB3qVHy
z1F8aIUslU5F7Wc~7s>RLb(^_;`BSd_UoIA=RC1C&6=`mh?LUh8015gN(6Sl6FE*cC
zx7@;plYjzowTt^fbZfa_d)r<`t*SwVW&(^F=c#K2FWR|%AjqjK4*a)5?}7}72vK$1
z@_?Z5756`XIm3TzsQ-^|ZTRO8e>+Nf7*l9M@A&Y)fv;$TYJ-0tjj@VQAiSim{dl}m
zgje>U*sig9<nm{^)c%Yy>!D!Q^#Qm@f$7mZpuhjWPNtaXui=bkY>LaR4B1^uR&?33
ziyR2FhZPYiS{1oyCGW6t*%5E@m5r5(BINt-II46^?ubpu<JBZRQcWexGSrhMEH1_O
z*klapdhhP(YGMq0_IhazM7Wo(!r0A6DIk`0H1xjfXSF^(Owb0mz$)*>gsk_-@#DCT
zd%EdmoHB3GCHU{!2#8Dg-3<q_&+}~fV@r@5BKR~M^)i|pw0j=h7xKPm@TbJxMP3fk
zo;=u6x3mm0600|mkmuPZibcv0%O(33&|7ZHb){ZPMXZ=Pwb+dR?dPXlj`&PDRaQ<t
z)c=cEr`gO^`fxBJ0{J$g%6TI$^Q6ap#qHZ<fzIhKa^O}ab@o{-#Z@28A=}3lmjd=V
z>rsN0BfF$zWT07y)gUjiS8aX(P`Eek*Z{dkG~PZOzmiHQrL&kOHk^R?+F06OD!DEm
zALV$Tsvg=c-8%-QRPD>BHWIg3o(b1d9nN~GSC?w3a+=+FUxv4ko3DoQZ*gGKLIq`g
ztqp==qmBA1h0Rd~*f*Bp8}=R<<_ld7ZVV}@_h_dU*g*YBEUs^@4F0_6@&yn~@wNMB
z!<gjmxbHOBI2ImLZEvKU`VQumn&t&Tf*HgyaWv6rDRDnC%$BSv^1_qsJd2{clM|_f
z1CiVw(?}_!+%6*HR!XX+)HWz;<bL?^Q3}wCkkGQwj;k(N9l@&)J#-*Nao8}TM}v5E
zF=EFhC@Sp7#B!-a2`&i4>H3+vwNEbM)Tkln)?$T0uOc7Eq53Y3Jx@JzX8Me3Z3o9`
zRInU2(2umETlYQt3CL8h(4$*_G|VI4UVTs}Gy1%?r$XG){He~|SRDN@^8@>44<<Fg
zpOq_Xt=4%>M5~!y<GhDj9@|vLq0f=Fw`IQmH*iu{DSo(!h#xQw{*5g5ywMuDoUGeH
zZkOl#l+!UGcoc3QB{hF28ctq(f`y2t2K2Gkv0P9_*Z%z<Ux++aE>@U!yA$}OXk2!*
zkbI8u$|rryFe8+A_IRBe<7$KLBCYQ3EIK#}hEA#WgBou?SmDDl8JjuDe&MwrQ^eC`
zjBFM8;jL+w#qVj5&`<9PKiYSBikrrFoxXU<v*|u)@Z!aR_cv(!lH^ar-S5fhJOl3q
z|1`E_V^n12m0zYa4?eo7Nmhmj-(2I0%T?8c^~5?NN+sBFIlsc7T{H%XgtKYf=RWA_
za2o{`DWz8#rFYuIUWa6*#!7hi=P8g?ruLJWe{H-NI&P&vN6I#Sxgu#ggmEE=nX?>D
z;L1R6@LLfy>@>44CS&=Fb*7PE_wgt~DfGbn=6wQuN`^%4T%+&uJIlhdcYX60K40pX
z(04XP^YjMOC*E==*O$hp3XmVEn6$@s2CIsfZ#-a`q#YVZ*(e_087XZXHOvs7J7_Pi
z67`L{JJQR{2z7mm9pqG_J>5Anm`f+mEYb=}E-mgMQ56XtvFnVu)$Qqdooe}EHv+4k
zOpsc4X99B#4rO&cHt6bn+d&{Y=g>xFVqbT$uEcqJ#Ao3yJgNS;{TuO2rmY5&(Ug_1
z?E}KjCUL)2fr_3P5f&_6)nX4g`g?A*tha_Ol1@xwB9`B4DG~CI=kBJmKH0=W3(y~B
zBPf!XD%X8$**!#4YYt}|6KO5iU{A^q47L+~<}caPj#rwW-a=pcHs?$SG}-?lCw=;A
z8SUpU=lBx{#M3#jX#PV{7OpE7j2kN~+t*wDw2QMuWM8tR7ktTy#7nSl=2q#omW=mz
z!OS308`Y-pk7e&4Ioz`D2cyK6`*V4UF@ys}jISZh4F?bM`ky;@rgGXv+WUhnC)Cbs
z-qAv#lFLJKo6KeFqa&L9VUoU=rB|Qce|d4luFmqR2{xz2JNkTsXaL+&usiEFj>`1F
z0cvV$F`E!qkrVanS}f7hbXcaX;@H=Su2(ZUt)8EUuv(W-icRw7nirkGOOHQ#OInQ7
z3y6tCmh}>5S_ZOi?vM7bu*uoap#NFb?|936ELKKLPu$0o_az2R6n6(PQg`nBjU#L+
zzgs1JM`Gy8Ys?Gp%i%(PqRO0#Cgy-H<!?~Qb?03++4RVn(%DP`i3?FSfw4<oNwIdh
zIj;g`>?N-@59jLUS)noJkQaPN8|;4qjM}^*ZU1A3`{u8U7Z(uRyDd;udqMS~7dyEG
z1ussqNKNTqxODMxdpE_$f`a-QVGqM^CT6_n>I(Jf=y2g|wHEG((C0j~IPIsP;Q8b3
z9l1FS{w-^B8aR)Co-b>vYp}8+9oSI5d9CL{VBv%*?#XX;a~Hd~YzwP(-(V9ZcWmoc
zJ_Dts^HH~T5<h8FX4L;q))t%?Tb^=xp_{azV1K_2+lLHbNc-5#pVAxsL;JeO>~5i*
zx4OK$l3AJ9H{Lsyd-T$uCl9sTUr+jer|W!7SjmEfyVnkGSC*T*^2L9Ly+%%kp`{hY
ziJ9eL4bNT_{{F_MzkinG0q+OvavG)|zE-*V*Q2{~6P0ZBi`QK#Og_D*tJQS&kFQpG
zC53%jnY$)*6y|=tT6)&`QcKUBEg6P_8xOK?y?Xu2wGFPT?o26=-}K^2*5yyUD;L&0
z+Ok~x>}m5QZHc$J4{QdeOSQu<dUZc*iM>3xPSpRK2g8A*4^pKQj6OJvrRnw^SgykG
vgMSSqDTkc_wlVa;=V&}|0-vCP%6XDMD>gi??(!i6pgIOmS3j3^P6<r_W^Qa)

literal 17824
zcmd^nbySq^+b4cOK#`J^mTp9l4uPSh98wxd3F%HH1cs9CkPvC<1_1%-E@24i?ykKD
zfA7A#=Y7xFvwPk@_Uv-bFfjAX^W4vMU-`K{*95;-l*W8W@(>9L2~$=EqKt&}PY@E)
zz3=G%fLC%hkiUbc2evYr4oFBCj}ec1!*K5_BqS;%S%{dbYsxm<#S`D;25oP**Gh@u
zBhF*x4>JBZFWtWnLd2k|uS+x5)|p(?h3YedY#$4GyUqvSXZ<j$0>P3*iP1xOL`Ul%
zMJz5eHM-4pcy%+0V}$-RN?}Fzz^&Qy+&6V9bwFgR-^|R}S;`<37c68?Y*YX|Azfj$
zKSDzKfs4-t{*iv&3x1QC&$e&b@~Ntg2}shZsS!vHf#;5dKGP9__FZGmmDOhwlF$^z
z+;JxFRp<K3@IAqZ&NelQ;#R&?Ma{TM#eML9mXrV7)V8ztct@Y^A-ierNY{Xzn3|gR
zY(*$%VA~fb*xZ+d%^zbMV~hKHanhUU%?|Z&9j~($3)>V$2GQO9m0~v{u%ee3euk(>
zNb~5hCrC&iXc$ew^1?F`?ja%Z`TifiA=AUUf=S8~O2%(*oSTryWoD6jA6!A{-N*JV
z=Xx`di@lBPv&uyqwlnuJCDH3_Cu(rz3yY8Ko;x$OXP*uHXmO$T#t?V+y76GJnvH_{
zgbu2zqj}#?JO<KwnAx;r+>Ix{ztv&@yZ)_&@J+8TmFLa*swIt~S;41?c-GC~oRGht
zl*@JcAcc#IyWcOk`z{q6KCfi?Q5&%XeMJ9t_xryRoj?)ts4wS=qrfJAZqM?8t56tu
zp2+B71(rDECov5*VzWNKDq}6kXGZrO7Yho)mxQXt7PKrdKQam}5W4b*;<s-*A0ASi
zx(Z!A3U_H_gJJ%j#6xK@etyoZeAsTpiQ#L&lf225JF?DA5+*=SG$0;lA_)zZ%0iq?
zRjqP*`<SeQe%?_46IB9vuA=FWbG?b?ikYpd+A;&U24{y7MKNmU;y&>sRK?lFS?oD*
znYM*ziSkjj=H%IQ^M;ZSfXMXzD|F+3C9D7YuaRjPil&Bd(x1Z5lguU{;69u!JI0VI
z<o2ANUMecPwY8N=<!goIM0ik|+e#u9xd68m@I)V0a-dZSF5ABeNJ#~qHb%i=DJf*T
zVWFj=p`ladW*$Nm)Xa>G#KP{5uMj^EwSYDLD1gHFxwyDEIU|D}<Kcn78hHg$`y~Rg
zm|%adH)|bM?HwFe`;x0odg3~MfgKtod(mCI1%o{dX%~l>me*!xwt8euSv4(&3t~|S
ztz8$i8_K?VwJ}w`I#cT)AtB*$v<7u@LHM150pB}4$^6yMyPTi@yrba~nY)viCNsZf
zA|;cNB0?GmrN|ElmkE~1_)yf3rsuiF+n`(xEE~VJ0gsd*V@-86wN3n3Vk77J#*e#p
zQ2a*tQ1^>hoR{b4z(-Nr^O1Rbdsp_4H2e7MQhvfDjY~_@YfK21iORlUmBB%gjpAc6
z6~j?gD>6IiSYb0&Y{kJf0@|ktO9a-iU}jtMaPiu+3!$c9Ay9n&t~TODhR}FM7^V*C
zVL5D2JqzE#)6-K+OG{i_TubYBs>`hArjT45E8>2KvZU43)xq1xvc6OaeJdBNzt$Z|
z!lY7*p=s^ubUkI6j8a+9O)ALhDcz_8aq<UXN5qd(JpJwK;bp9cTzvWY2Y8=2JHREe
zHT)L$T_2$$Gwj-RB(yEeo7s|K1$G(#+3*uT9AjkZ{)mibYG#(3o$VhGASo$nVr<--
z%v;hS)E0=XUu(Z?S<<MkrInPDGDFM>HY|y}rGB71k(r$x3myFj;b8#S1g)aCZ{KR6
z<^7x{EQecTV_}U8$j~MCIhWdaCZ)lx%|iF10}EQ1SXj77!crPK8?}3{JZUT|wfua2
zzqfDU=@c>o-~9vQKgJv}PGlq`Ta^C~-@qj{(lYxN%xPLZ4ZLK}^6=&{Q1ng=HMJ*3
z2p8uc6ohckeWvB(V$4i1YFH)}x)$0~YGr)MIFqcQZ@@hR3n-beOwEj9?Fr@dKCpek
zXI-rwMe9Agc&8zXxg+nmdG>_aVbr3QY?$j6Sg$d}aIUVdHn?)9w|SEL{<~M^(}Wk~
zZy;hpsKO7RhQQ^Dh}Xic>*4(a^OIIQ-mR0{&K<;}ZZ#xiyQC@6r+h}e3$|}q35)zx
z?Q3;R%M+uO*!swcSY84B*%SYNZ%^4;1a}*_Ty|zKY^tZMMsk&GE`;JS5n9WZII^x&
zW*j`ubc2FOrnoiASZ}X)eImlb$e$U!w6I{t<?KlFzF4oRsR;@SlBV>E;pA~USlZg!
zYQEmKAyOO`@ZNURLIZ-Uk;fon7xFx_m6DROv$Kn!5WYCwO;1X4zdGAr?2O>*K=|9|
z&u4vZuSZ8m&kt8`FE&eF&^vAaR@*dtjq&&D$-Fjce0G^F2I$z>l>GLzxW`*l4mVfl
za`EiK_DeF>*2N?1g-NyHvN|orR!B(iiXkA9cT;0Ld`O*Hq+LDhdDzDuzqwPtyE9#F
zbDX%D@RQ*0i6uTgI{y>GAw;2MXSbj9!ZPSxPaIp8bcFX(q;#+8VT&Ic+?p5oV*65A
zvd1;oC-ow*kPzg5{&5swcmCB=R>%k&dN(S7jvJYuuU_zil~sO|Di&muEe78%3<}t)
zr{7mk`}z4*Vkvi-mY<)Wn*+7}kB;Su5msQEVd*>a?5Hsd%!d!7X!C;OS$Ynll|;_3
zPv@%NPxJ8e^V8F>jBDF?ZIyN&931>vT%@6<#v<eU4kFS$WYnXzA)}4aychKE;dPb4
z$P;L=DYMtTPA@qpJ{IhVaZcQ8*Fh_S)k^;PSAr2%R+Pusox<;6#iQ-$c$6(0ov$R|
zc3><oA4)6h)C=6#pv8-B!oy<SdP5_lT*XwaH*XT6qZb<^IpZQDIXzCSB03#P-N628
z)k;kn_V@OZJu^VUrK6`eH8quzl*El3Sr>9#`+PcB6VkWXPV7*tbF7JD)VuB6Q?_-v
zy0@B_pP$dnFgN94w0FoN403K_JQOhy7$Au3tb1VoS6Qzb@t^hBwz2td`5RtzmUphB
zV{G1aVj5E@9D;+Q*wi|K2u6~U0jHl)%BOUo(1pe!E1o&KPI5GCibA6_=GmeRxq~Kd
zeoc5eA_sJAN<jF6mr#6f*wp2DaC*yr@lK^8^nsrWnwnZ;i7_zzKNhw@-xe&5p!m>U
zN4Iv*t`~*r>FIE529P}3mC`V8CP?G>fB5j>f|iM$z3L4sAM>U7Y*rcKydT=K%FdsV
z6=Ml?K&PC3fX8>OA`p!En5P*w>N965%tsCfL<fjTLF)1P^&pIZWg~%KL_~y*Ei7nw
zco=LchX(~nV8*{TwF`mF1q9REv&9I)kma?ZtlRSepN#~!`5zyrrly#fm_X!$LZMHd
zJULwHEAU9`3B?9mRyKNyLIbaJ5_CUWa~KewC^sAM*lpzH;o%V!tS@cTeXRJ*o|CUd
zhTPUSQoumNnfyCTJr-X}5cZgh<Wz}%Ga7}f!TG^5LjM=Tcm<rdCcl*;wmA2K@DOn#
z+xZ{In-hVbK0Wl^TaMS~v0z0kH?K+XBM6BUK38_T>2t1&n2O$~pa2+rsa{ao&>%!l
z|8+c%((4FnY`g?+gu^YXgu5F$g*b0lS63px!wN`|<mBWwxo$y7k7Ltbxpo-S{GKm#
z@Qa3s(6)&8;+0o*0#LyVp#Ey)v0k^pCfrt2!(u?HoN+TMfwq42v3GfSnX+l`>b-`e
zGjsRMwrKO!K~FQKv&HjZ4IQ@yI1#&!D{wd<8T|9dP%{o8O~n*Gpe=!cfj(yofrqOD
z5il4tE2~bqY1qhm46S)z(hIQYp#a{vxVTwxIansT?=z#0P~Z~uXGijEIEe@dFrPeM
zs$EU1KOWZ>bX>Fj^E0?d8?l2nZgH&n%%*rlEPj_Bx1I!%2g_z^$v?zh{THURVCP=~
zDSlVm=-b-v-CSRa-kdE0$5%b;Zu#TGJ>U@sy1Vm}*nqXq-QJw`a5Oi(pRUrfL1+)~
z2mD}9mX?-238h{TmY;sxQb^`$JQ|kcV<tsHI(rInE@WPT!C+A-bJWOJ!E45L=JMj=
zV@ehQgK}5CM;iqt^qD|+pKzJJVT}Xn@WzB8G50$h+<;G?)J5?uj!8#GMi4tChI4VU
zJ^eXXWzNuhxA8QNUH2lw=L%Wc1o#DwM@m8($E!S$ay`%e`EC;m$;QJRjhl;!=J8eB
zhTNlP#E&3HvmUz&X+r0}K@zrIHJv}8vRP75Fh@*8<N_3UZEX!14Qs~h6dr%7v!+BI
zOHM>o?YeKMrWS}rPC`!ZH5Gyae7Ml<)vBnlO(Px5P*}<JF#*BxK@W$%L*I+*i({J1
zX`AM2kb1g-<*^d>fB>X=EBsUCtNasD@2kt*W}D}UB=$S;7is64eY|rzjqb<qtKsyr
z!})3j9bu%WAjIgOD0|&EyQw2iIWNHCJJcYzT`3Wm@6~tuk=T3}1wc-DeR=8w0)cw5
zj@y(*H`sHR^<lafFUBsZzfW&BdYmdMDK$4pUf{sa4gWe@gqy{8Ab;HpS#qvN>?{zn
z$4+go&+Se8HJ?xBaK0k5Z+QsN{ZWHw#PmvOGkcxjv~w<VUceT>=9`5U={35y`ai(b
z=Rq9x8tectMbt;c-`x&Psgp#!+!s^#fp{`r)H`xUP>MEnupzv2#k-;ISQb(SPyoH$
zbXqs`24NzJ3K4%_d%@h!_7)UAgS|ugcB2swO!EW%hyM|9&`IIK0Ew&yvC-w}?&;y7
zoQ%x7c?q^R->Fzxdo3zK;8wp@JS!g%;YyTufjem&^B1XGy1AW$>LP!@%?*LG4BER2
zrGaeX(6XFzYTIX*_<YyrwtgYv{1;7xj(r-~@2%<T-2fuJcV~aJ>^Tdjasxi|h`-Dj
z7X!C~E2~{qFm`*Rx~xXEirY6`*0`~<;~m#_?XK^7`b8qJSg)}TC_2I|U2a$FyDr=>
zpCY56xLF9KbT>;7_3>6!RaI6VZo?KOBPL$6<)d(4`_j<R5RAv5-dAh8&<f<by1WdG
zm6o614ain8w_GJxF&(5YnUW#ng_?_w0(7!k!0=E1L^-`Ho#XU~V?F4-jjuDyMXc|G
zrBP?NZnL+Sxj7TAY}65OLq(e9Uu$akSXc@!tbih(0Ka~FJfZKwK^mK$uJ6?!2$m2o
zxU)8#gMr(ZD(C_{!*tbqj7N{IPNr>w89|h6`-BM;XtwJ8G~HpvB%3K`Re6&8OaAXl
zB;J1Gnh2qgQx4{C41lzyw3FhJUm(uvq*UJf6z)<$|AgKCV?DwE&R5eJGC>xFhJw<Y
zzzI?aV`JlAWjbC}cnpe<D1;J%8g5HG`q3yo2L;VQTn0fWL(DITk(+yTabM?P9dX^#
z2{l^T!g(^m!{oo0b$Xthe^dXQ@7*F%x(~1CuK|~c%iSxtF$T#emUDqKrAD2ZnVBWc
zHy7Yee1P=<C2jyRR!kKbODF>Q*z7hO+}Pux00(tlDSmc#>(d=L1XStlyQ88`V!edg
zy}s$q#B;GhG02w{{XiKNK)OF?+lD!z<DhWOq3n5mJi*oB>+5R`<f;iag_O77u+;Nt
zNP2A5?R+^t7ifaj<g;CMn=t;Y%*>{XjRLU0ctaxq-|AvWfAr+}Pm9X%CtNE7D@6Gu
zP5gf=zQJ{ixu+!*GppwZ4lSZT_PVtlBZY}YXb}}@iVj3qO|4*zgplwns6Rk~Vel*{
z?6%}JrFHeRbu}6qnkC50`b-ZG4@Ura2+D`Q<*ACgx+ymazZ=M5L1o%*I0ULka29<0
zqueTt1;d&*P@c!79*>`d6=v%6^$rRE8++MUv1CZ!D7KLXfiRz#A|t)z8p;ewl-YvE
z2gS6Zb#0F$j{1>Pd1M%_gyZY`Yg;{nV)$RL?HSUG;dmU;lO<yUR1G4>VM?9nqED^N
zHT*`h7O{Cq>|d6N%5ezT{G~tV=jWl_iP!c4TQeyCH^u^kMv*-#v4F&F9~5OKw9k6U
z;*4&ws@p1Vf-sKcWKPh|>R*AoD1aAv&_hvF4qhA0QN!oL4feXgi>&j(&fB0f1Ni0O
zD@!I*raay!p@kKy!ge}$pC%g9f+arc=pJ#{>6`@__`F5|)5@Vod2CyAR8Ni*A+Vy~
zm0t~a>;rJf#I39C*D~%FY@u-(8(Im{LxI%el#yk7w-QZVO7RH2*73fBS>2r-Sxodd
zg2zJXq6a2>dZyD0wul^=wI@G+{5Eqr>=hJWO|6!Mthm*c<EQ{^oepMrCZlzAa{6ga
z=OVx|8aB64^&E6@P@=n*3S|0-BHCHeeW$F72SADrs>lt;V=qFhd4tm`wiu`3rSBN;
z#YsPx5|**Dz*@T`N7eo-;Jhhg|L@L*o$m}EncV{(SfsMpke<490Et@JdrB_U90Av6
zd{x)zRUTH$9x2WI=f!5sU&&ZFwaXKROctTx=6Sf7liwXTuF(rop8E*N&fe+7PP&Km
zfw!nMp`GP1v(I2ko&McM_zaSSB%*vdNFtm>s7Z=>Z2J~GnGM}yoXMG8>F$tb%ba%B
ziU$OO?Rb%8*ra_x;Ud+wdLgtYq%x3=o5pNLEi(XYd>eHS(cBn{?j=-*7Jnrly3^mB
zaqZsh5xK?-9EA2FDQVep;>pf7POVF8%9h?OJV?up6C;v504p*(f8(Rq-KnT7mfPjQ
z%EchrTkrr&EudQ1T3(#`WAR7s>Z%!Oe8qPfqORX-TqPsy4wmInKMn;)fk67_M++pI
zLe^y21aYbl<5=zThVm(cO~_2Jiymu)n!dCQ1Tu|6rA2%W4GSUW_g@l%gR%=QB4>XT
zN^a~s-W^u(2^>Zpt+^2_@dJFygK_26pAU$W3;*nXRf%v#MMSXLC}MIJ#~^AowbKQh
zXF%H@!OmTjZ^f42$wm2hD^P#fekGXaBKn*d+;WngJVK^}uf8M{a3*H!Y2ZedITRq)
zo)Ig~dHXTOlg*6-MKhrUaf9!|UibQd5(ieA8pLUGiuHH#{)WX}^+SbQa9QhWgak9j
zK)pswe}WIpf8jOO&Im&A6#|Sa7&1|l9Em|O??o3zLHa!S^L20KXH+bDS;N)-{9L7X
z2wl=5!gpZv|L~HtlmVAXaIc;TTveHI<+s|6^~PRFVFl!FwXlo?ONI{T$JByR`OF@F
zj{+A6i9*#?fNboL)Asz8^*`;NP~{R4-}Z>uc&6r^iMy2$f&)iLe}41}mN`F{zOn`_
zq9jyUE&K5NrcADpo7os%=U-v|177s?`4HnsfXk%KYk9Z4`}l`+G833IVy{Oc%B~#l
zE)$2{?)gCc63Z}`{bxeiHN%})dIAlbxYv`uK(epOlBcZh4kT2ZAjZ!BpUYN@#?tTK
zZM$QbwtrXBWSXV9eFdOxSmaNJpZ+`I^|o<7itt8V5+3DEJ|MBE|D>c<*$lgX|31hq
z6lD=V3$?MqK`3G*p5(Yp(>~NwSxzKCgk2;iApjqC02w+YCLtjqG4Tljfzv`5Bqljo
z&}O#b=g*%{B|eS1qD18CU3U^kn<Tzyi3p8$$CxWIB3RzJzVx5<vP_jk_^ZvthSI$J
zSH~0`^cjpW)*fG~cc8TWb(sYG)tTEA4nPu!Q$pT`GQnD9DLRNX{3Jk0>@gt}uYiLB
zU{tL<H}jaM-@bkO{)QD|%#+N*%nU@JTkj0kF6?oVBIsiM?%g|pvgGgd6PJ(ePU|4R
z(|1W@jl~+2*Wu~0={}C_46onI9eIn`*25KWR-zi-ZXl}Hf)_z9$K7M|jIc~6*sECr
z77e$v2aDHxeO&-{R#l6w<$wF#<^9YLv`AVeDykn0Zb_9=Os1luqV;e^2mmk!21>?B
zK?1qk=mlCv>+9=dA(}Y;^Yil;tlvPcL`9WhF7mc?fw$&#2Or}A`LI7ogI<r%kF^Ju
zNgUQekM1CwpRZLk?v7=IXDg73>L-R@%Qrc=?K}-SZ>XK7<!X%z+-B9&PmHdX7A$PL
zySqcBXaF{BgX>=H>}pV{1yC7U*<k@0xj_I9pPijmRB%(jl9LOGh$uPkpp%Ou6L1U)
zDl03arluyL`DF9p=H8tvwYt!|vk60hp<|QrJrfiZq@$x_Wo2b%W~QMrnr(1<VNE9D
zc?QyJeUI%*P^g3Yt0QQ^VYsqQ27hde^?3BbUyY^ZZjV$6`^w~E;^EeGr6XLUB)|4G
z{pd}YEZI&s76!%$$Pc!<i;*=vEK5P%JsGQuc9$;r>rRQZw8P4t3}KRs`DRcEcH2AL
zg>E)bSPbC)cEPE4rlCI1RrqTJ=<QdzvJz4_kx6yw1O$Jro&^6eoxa0~&`2;cpILNK
z>obVJMvN;x3-EX9C)=j~@|}eD+dcUab^)R7qA=-^x?6ULuVE9bZw9Fy4K7oCRr#=T
z(FkykJs%C5<PA@=yTew}eMax>o(!BXAPh7u-=ToN0ug`h@#8xn8)Q8}Lh{6HpP`z3
z2hxK%aZcVxJRmsz$bH?5N$0HeSO5~W^;Wxd=n-fo2N+sE6B`U8mV{UhC^A=Bq=HW*
z@#2?b2f?F%Ch#fapDeoE-R25r`xfc&H|-!9KF80g=L&aw6#6B>OCofW)p1)Yv}?|%
z;LfRbyzUjDnYG1x;G2~8)ztJ(AW1Cv<-7jUuZPtjv?aHnh(`oKEaaABG)*V*MqYPh
zWzZJB`s<}N#W^okZG*V?EU2TJZ{t+eHo1YLxXMWIk|%LSPLwaC(x0mQd-b~)_YOxe
zNo64ipKdvm74Do$FF01I&ixzCcsQA}$)fw6RJle-Le7(In%w>bGl-ncBS&<OFk*bB
zg)3A)jx4hz6-=IH3;{V2Vg4AwA*e~mo6mpG&6re}E`v&9PDT#5{urlTPd1;q<l2UY
z4MZlbY>*>ZP855+x4o^I&ybacM|d4NEwahD9B$LfZUjF%wzXJfLj-^XkYNL~(A?a7
z%NNh<XM6kh01H*715Z6up2p&RoZkT&H;xu?4O-rQ5eKlBc3P&KO36G$i`jwo3SNg*
z)7GGQJ;oq>I^!Vc6jxd>F5<nh)tksg`r^~h1=Kg4vYe1fyPA8$7X+&IL38pSr@+Q?
zt*;`#_EBZNnDcBj?MvEUEFjMS(fcnKpi@KP`1Dl2&e0rCl2bc7lDA90=(Eq2-qx5j
zy7>sC=RVR0(}Mffob@{F&$XFv|E#Xs@^L?+!6ast>ipcv=2NNDMhvz@r{ag9^+Yio
z8F!)Lj7zIJZoH^*IZnGmo(8>cdY2xmIx;dkimXKsa)gzP`-EJ$Y+4m-f%UsOL(NL(
z1ETjq7EdY_6ciIxLPvd`JeIw_A_x??<i2;OJj9nOl$cX|6Tf)`Zd>k78%v<QyxTk$
zM((o?j(KT0_Pjcw*67(<$E+-*yZAjwr4TlP)cm?v0V2Of@2o`Kv2@GM0H_66Ah~|n
zKI7<E9Zr)KxxCCQdi@Qdr1SW&+I4iXmbxLZ{|$D+?A9Q`s<h~8zj_~e{JF2#2`2g(
z^$}r7O%#56`4z_2AS%)ajLJzv_Jd#1;U;>OyYniE3@m4PFM;jU2lkpepM*B7d)%$U
zMaznFS}mu#IlXt%gWt?!OZa>7x0;(%kiG8vJ`fYDOYQ(FXO7&{W!-T99@0yPQNxhh
ztf87OW~X!-dbrku>CSc5B92EVZJlMQ#dW*KQ)cOyVcUEHO(Ot&nsJ@V_V)YfZ=Ic=
zwyZzjsseJdRNI9B+!vhy(`6sZ&~vVb8$S;=)TQKjWqf`-XD`aHk?~lQcE9~sQ10(Q
zQplG$ez{i^P40smf*t6PlA8*n(NIkQb{t0(0SGu>K=-<TN?Bd%aE>4ai^ml`j7;}7
zXB5w<7vSZWmzP(O91)$1L+g&duY11p3Twu}t*hG&YmW*QT9@sggh>^ts2=Er^%*<^
zg~4OcF?1Ag643RUTMP^3T-kR$JC84$vI6bdoyvW>2fjZBM5{m5xCV?-q5JBUl&q!*
zq$);b3#^zIOx&)*`lG2^Z`0f?d;+$Qx47gU7O6=5V-vCSW)ai0Qmft{#iJn0<INiA
zAa6ctlY;{0z=lWs;OpbK=`jL1e{xY-4QD)o_6Z-c;{`bBZSDat^R9FpYLmIwFTe`n
z;sd=aJSiytzuHalJ#?~<8Qh4Cd<+9d&Nj+<iU<78n3c*qj5zV2*SAiR(87UUKVQF-
zqxD)U-rZ%q+>z~Yk1ec=oj`&{kI?3ejeNH{=j|lI;gic~&8PH>tAQg;WSoDL8h;TR
zg_d5-Np5ich>k3#n8{_o6u&*f5PxZK9q%MNJK=&nM=P>)ZT2TSfj|-$<zsrw0^6hg
zqwa<Gwi~sr=Kp|}TdS*V)K9gvm5>kulabFTZ|#kLHqQCT<<LPO!cjZU%q3~vWvQa3
z+ELldJ@L7x&F)6IETU6aDz2`sM$*G`cJVmlacVI=Eb%x#b0GhGIbtc<zO~%jLz8om
zVX2nuA;`k;cHkyw^<hI1@#Dt_09jlF7+q*ch$W00HOspdnO4@qJ*n)yoC+&0KsCoV
zVCOfNySe;K6*kyjXDv0oN9cY}JC~by*^CBnM#*!Q71E`HVB>IGJlw9$W@veNk_fO)
z+Z^CpKR=bgx?a8J7gz`kq<C{4A2^Jrb1d!{Ba816AFJ<0bzGdao}1Di0zdr~Nki$m
z*V3OV=tU}9=Uj8@XmC(Akmntk1&VYzy2kwuQZ2I@&%3(uW=aC~yV4Q?hXdFHHxi03
z?T<1?G|N7)En-+sSvp0ul*SSp+jQ^7bR8SJ{%zYQP3+lp9cY7l*Pl83aR9WFU;_$@
z(U!gUtEcFH2p)hssSxZ3A*g+ze%K1eq5Nc6-rl$XouHLXA+yh8gmo@*CdwmP)ngHQ
zLKN8hM-EJYN3v|cf9SyUXX~>QXijudmWT_##R(2FROm2)*I93-;&XVU<O^I@b3sMl
zr^kAAK)3M?LI~Nko>}}<(EYQbCu$6O1x=V#XJ~6XX)>ALC<MM;RSZ>Q125Kx4$7Y3
ze*|wf<g%Yr+pjrtqzJ*CPH!aL&L0{<@QH~NY|8PV0n>0|(FZv7I$!GdO+oy;R96t)
zIP63N9{pbXRsGD24lJB*SGFZQ=YD>-4f}*dXE?)-@OSFwz4mxL_X9}3-zHD!<!&AL
zCy(<VnF%bSXm>oLG!9WS$$+p@bl^A1z%q2}9G_UTWzYmNZgP7a45gjb<du>m)2=7L
z?Iwnfk_S?zXTB_!`Dds9u8KFQg+502^}Z4ozeKm66xR2Yr*O$)1#yw@li^LOVy<?B
z>pqK_Lz+m<?p`iNz4O+Yl+{t!N3XArq1+*Vpb_#uZZS7E_<ad4EXNB;PW!sM-4_Cg
z(!5TlvZP6?h<+Q4&fO9{PoA368Q3tvbY}z+^b;uR2AjMt*GF;(61j3cPIqRtTaPZL
zQjNIKj@Ipup5f9^KPRqIsJRu~+?|MY<Rin@b2(0xto3n<6?Pj518qUfE-L@Zu{kHv
z2fpqG=?65$Q;(1a>oO8HbNq(BQ$0<aRrpeFCJU7lfP}vi7I^2g@tc~3M!@~(9pG-2
z>{aPXNLcUw&RGz3BrK61h4Xz1TB5wkrMPM*8UTin1Q2O2KoZ-U+|o?v>u!3$E&3R@
zn8;t(WVxE`)h3879AZzK3ANNZg5D|Z)UxZ=aeLYgf5sjDQ4q{I0Bhd}Uc~fy32u}S
z+^G0Ou`VTmS2?dQyeBJr_V0PoOG_m{I4fIECQ{0qdvBNWuMVL4MVqG+U263$q%tl&
z<c6OAFu@vJpR!mD?%09zu)pq&XX-Fcp`2SE{ULI@M*M_I)=dCjibl!b@GkFoMf;&J
z#T=_d)X}NJ<5rpk_=@Jy)ADK0LqIwF!HMtSSHVkc>J%<4{C&5gW{oHF^?anhXe@cT
z<=k9&x_q@@!y!Ha0o-Xqm)9wnl-GI&u;OlJ{*gDm&Tf8q?_=RJ{hK2R`$aTW(B&qB
zO`-i>b8;)cF;BhZ_+7@7kv&=bZ8x3YV>#_QOC`eFhAr`;gH(xRz1bx+k;Jemlq(wj
z>H8yWkylb#Nlx8|=ber>8$%#Fp$3`hJZNa|t}YGZ4GeyMyKFh3$YAo$7KcX5hOLS7
z_?MYP$sCt~ZC<~vWb$VKq4v0XK4sRr{gIVDQqUOz8b+BGW_uv)6QZ_nAp;C6bd^KR
z`1y&0An)<3t4f=aK$=#C%r`^yM^igqe?Emh`Q4<_`L>P22G?-a^|}!|iZNeWc{x@t
zm@_>G!2W3<Db3KXwqBI|LZ?5NPr#cd<PBtk3yo@?>(s5k;`z2aZ8!9#tdo7+`DCX$
zW8j0>r3%sV@^t@hWYv@v0vm4)zyQVNh8}FPDA0OFm=?YJM)=@%ocp32<fClph5R-i
z1Vlstdfu~jQ&m#~l3sHZkPahpS|CtXle#U}x{Xdg)rFS2@H%EF*467y2Y-1#n_@d5
zj4FoX&p~zoJZPV(0~|xS>V4$IO?*)H(}_)~usoUw&;!T7u$(_UAjlmte@4t#t6ne-
zx=7nCEiJ$U8kbK2vPXp?w9C4htK+N-G>BtjV?io_V2|KX-xSwc-!XJZ7P~!)J|A_E
z+Pq6$IN(oJ%o_UDwxNgP56X80SnUpV5zAm<UGZp}R<xR9e>t3LD}W(<z_Cn#afs1f
zPfsf)4c{^=OBIn!EeQ$<Dew?8bS$KT0c+Sgl0B$tKl9fzTijM)LHOvuHM|TIKcds<
zc_O>8W%Qn_P-T-9_rktwP!NJxXjiKvkQyGyc8^0dj$`K0AncCxCsY`wjr7PcE82vg
zj&2#z-QO&#rVg5gh4hn5o=#K1J3a+$1U>v+=~HVtz%J40X?u!Xe~tW}h|KYQ`PkKY
z8TS<`d`%kwG3(Oj(<Mo3<{;cgzvSeti;0O@J$wktkEsch6VUC#i=;*84fy04oA5b$
z#>e$yC?my65~r-**`sUa(Ol^%pt9P86jF3N=i-tMCEhofo28Ks7XpNU<Ib33(&Bn^
zRs@FAqhIu+)Bx;0D#xX<)^~D2=RvN4XvX2e1!-kd)pUI8L7S^h8c;q5ZiknU^>Yjm
zq&d~cr?I?oFQ!s0!QhX%xOjoQYwIYgqQwIm+#F43nHOF;mVvlNK(QF>t0Jk^tjVQ4
zw-|vPeELD?kOewX8Q0A?xa4vnp`i`7sPR2xni(#Fo!3BrJJR5kFm&wkb(oO7sd#np
zL>P>M$9W~=OQ)Z&u+I!DI~&_I?e))_DsmZ|m{e_HeeX{W5dcX;0{wV1K%>K7Zl&P?
z8va#2@nGRg8@6+!;mOw2moHz&az!q;${n!CpF+<r3HCu3wz-kFPSvTg(siXDj)p09
zdaf@MxaJ7t8H$O~SgP=S-aZU89G?34NvjQ<-xnb#0+pL{llbiq8e{Iq;oZe};DP+_
zkCUhP<pVl}37fWnwy{&vAf+TM2RXWF(AT@_RMXH^d+~yWWi$K%o_FnYPHxVjv0R$~
zt#b|&!uGT>2K#GXn<tFJ%Q92PJzlr2BF+QOQNrS?6*pfxQABT@HpfA?`2FZN=!Bj}
zky84yiTMA>00DKj9n`!?-OY_(ZA?N&MpIMsIIje>P&kx?<Zy8%`T0-1F7}1X=dSAp
z>%M+pOPMa0i&^_Rt<EK7=0G!Zk+!elyfLeLePt9Ey?t)5v)XHGHCCqmN6_cu=h95P
zUgPCX{R>xXkbG&@O!}Hv7T0a$lOEce!?m=Ut(VsE{7^CTMJj7*5`I@RP;U%=$<8oP
zP5yje!?B-y)JJ=lB3d4mnVD8B11r+ZhEh~yj>N8b>7(}Ud8SJ1)bOx&f13AYT??{6
z#clIuiFT=>WLQ8y;Ow|s9qmo7<;cbJOmTr?(p{X>QnqBm=Xgv*|75+gLL&lntLi$~
z%Sf$jK8l<NDW+ye5Qb6+|MtjY9GuVi@nD&*!cAq_)2yF37$$UZV1G%Y>(PN#?=<;2
z`K!*9N1VYK58xF}^a5;4wAO0>c8BD0JGGiU^}-@7#3F^y&Jgt8vIII65}tkOKophS
zm|NZB30=l@lf^3xX|B}F8wK!s7Y3!8!#Hj0JP{_)gk0-QOm=}pQAeD)9p+bBx=f|c
zX@)eH6H=B7hNGa6KS0l%WPzxK9oFQg@HsiXHn2A^s1ftII9f0GT}?v`#i<JnSXo+9
z6(!zoC(=)uE`4uY%_(ZK+6xFhE4_(hL>Hh{zN=mJGb=x$Kjp>SR(}$6rF0Q5(1sN8
zy4X8}vw=F$->=zo<vG@k1$x0)9caI&DuOQb>3vjGb@)wkk5donh$AxEjcHj5*Uds<
zv~KU$9e4OF_N-0vq4rDXxak3(&GDkK0T;e}hZ&8<ju<{&R9VlfP$$42QH%BlByk&k
z*OMa4=qE#yYIB7`u8aTxwAdih_uTIbt-f%1*UWx6iZ*xq`SV^NUVt^c?szdm6o6#O
z&rMzd8i0jRB;YZCE+8Na&<k3jp!tm5`}fB=4pA&>{1?y(1PrOAfNB?R5~B$6>x&Ib
z%7M%6mN~_p2M?x)zHS$vkSje9|Js(gPE}`SSnM($6$K59I7WZ8X3D9*ZWhd&{9D(n
z=sa*6f1JntG~B1*I?&;Zv6&e|6=n}B`~JixDEra{U5ec~IW)_b3OIR#d~J<#697p~
z<9a6qLE14Dy_!$Kok(R7I{bl*Z8iD*%f*v&kRyS5<%WXtaTHj@aE=@>No}t$+ErGX
z4X$>VC)<edis5sy>|8JR>J^BCF0~TqNIz(-I4|Y#Y&a+0?#?z|Q+lKrF1iye-wK&#
zKfNWxU3OZl@tMSw^A@cCZJJwE8K1bc&lm;Cl%b=WFiH1bNqayZ_9>j+NV`!4kbUN2
zHCFU}kkwQjMZBKV&}3~?D-{+V1;}F08pNm?Lhs$+<4o3id8q5X@|A(G$A{Fy=~E#P
zv_p)?#mz&aLMi|Jb3cSq^NOPO{niZWr;Yn>JrsB1`r%&{lGO=T=3D$mc1U<V&MGx?
zOv>fRfFcx70K#6SL)w|P@dqDSt5sl)>bEg3As_+MY4UUf)E^ZkVS>kxl^4w^1f92z
z)<*zLzq~*h!Z*$MlhA?uJ0MPg+s6U4wKFDDYI}jyKy1pLsq$RCToq@V_~y$vGQe#P
z^lyrEcbpQ&UVinM?Pe=|<ZG(xXcmu05?pU7lH=r^6!ekul4n7zcpL54PrH*Re=_co
zlUIHBby#pSYkeihwjsHz?c*%v{q(6Zca$Orv4SQ)f5&4>!H~qDD&t${<C^+VLJZso
zs5G*}(!0H$8PdE^9IRy`0_P^fr3B7cGupAu2$Qrou?PKajdf<RaV7TUF00K-fa)wy
z`^uR6Yj$?FcN1fi!X9B)g<(5vgY-{mDInWyO_tJ7Q5gX91*J?Lg;Lxy0gF05K7J^<
zAP?aAv|d?PUmg_u{Zu;zDgOt<Vr1A<AOJPn)Ho;j4v7jFHO76iw&vvWUqP6M-qv%~
z_}H(Cb&0!rJf->^1VER{ThC`4&49PgbSgiWop$o$hgD#?xEe<-UpX~eH$`;*hDojN
zTQU37odrS%ZCQ%~JY9aX+4H674?{++a<B`ltDRt3?6QOB`QHqjNvGcL3`fy7S)|HA
zL!%&&W$S7<2Ng9>TqynU=6%0~$@N0t>&qw_npZl-IRPe0!nZ=}j@i7QTd39?RrI`b
z%<R1V(@l0}kq^BmRj>a^<w@TDh(@uOO~L3?JEvD?wKv}~7M}npgztZIC-?E#{rNd<
zU0)=X#%0!j1;}ooivhSg`E*4+%l5VTloErJ|3<9Yk^t!KLTy6h8@EJIB$fGK3npI%
z*0K?n>Us=s27mkh{d;C6H7nyUrASS;CtdcXv$o^!`d=g-hJMwFK2I5n$>k+Psb+eX
zcu1!^l<t0xuYKvXhO@p(ls^%$=GY<tzaZ#2P3m3QJMk8^N%OwmTk<b9jo8X{mABS#
z*81I=Uj5K7%8$9b&Be;YwiQL0fZC9qC~pH4epajfiN{fyWfWc)XR^x6bXCT+Gg_*K
zY|j&}kA6DllNh{9*&vd9o`eFW%^K2De&w7@`!I^im=ErSs;>a?osx$bmwDE1^^i`Z
zoUb42f6c-6nE$fIu8Sf|?e~#Zw_|-|SL9a5Q)bU*nLv2gW1b^I<P|DSs(xi(y6A1P
zi^>(|#rMrmUVANQW6jALYEEEY0`O%O<>ft>V>K+cx=Ox%I|Ce7GOW(V(^*mq3O$;d
zF$-bGwb$a}tpU+NpgHpfFd`LKV1`Y^_zoDK3{Rfigw#G1OfDe13P|lTbu-gE&+#rP
z9KqQB{osAuGSAz~va+$Ng^3vfb)nmn!)NMi`&+LBc^q1N8<~%TgDLDf`B;qZQ`G)g
z>~it2C&Q9PPD6ET?qX*O$Hc}KZqDL+!;o{6q&3K%B2O>iWbryM2=ue4QfVLlgPTJx
z(lEA?5l$W)baT4h+pG|W@-!$<UhP`#WH}du>tV<h>&jkes9CyaW@B^n)G|K4(&x_9
z)X2ytFVgc%7svY0o14bshA87iA#3~k#LQ=bjqAm;IlRwg0S_cjZLElyA|OCY8Z9s|
znkuRb0?7jm%I2G-ZAxM-oh8aKyDNVW8LH=+K|$+ITyu#|MPh>(e_%$Pxn3udFAb<<
z1)dWJtr`3-)nJ$<!Xt~(GxpjM#gGakQS&PGxtyN6xtucdec9GItG;Mq3N$|_UQfX8
z4@5_21|aB`p7?y;S7)PnDq!G@=e-u7ZCMB<gTmYjgomF&xI}4G>T%2IJ~vYa&j5Mn
zP0jqlJHY-~>Wbdq-v{FdPz0U#&O;_aL1($GIIYJh;J!}{U#ubn3JUti7JOpHHzak9
z2k8MA@kO_a=$MUO!xCbgoB<R{U!>XNmJKy$E6kMMb`nR?+Sm3*Esd_by77102jv}f
znxaC#2mQ&L(Xqf|evD^T^-k6T@3FW%oz%}!z~HR!vieLL%=uJQ^l2Rpu3PmkW|=2D
zgEY~w6poBp`Tg}{(=1Mz{h<6;`5P6}lClqDg@@0p-+u2@BY65U&Z+Rg10|C2^@*y&
za6x|a!R6`2W21X%bGeTro2fSoywISFJI$iLEia?Og8++n+u3ZHD5mngwC{XZUn4aS
zJvVoZqx)`gjJ~qURBbs7u!g7FrXv7c&*!6GKYe{sP!e~#t`wRlSlw7XFA>tNT)E!7
zRz9Hab#CHq7*Q374L>|vJQY3KJ4bPk$OoiPPHu*t9L5uCSunGbH`@oIk*nhto0P2b
zw9A`Dcc-u>1Ru0nyHp3q-|TA>r%89rMzr(J3_sXAE+}~kD+|2I%Ofo77Bo644O;#w
zpArHMV9@<lO<%I*1I2_rVA}!C-zuPOWntVo^t*hYd&1EAF;RU`J0Q6N=43l2UCyaX
z;~zj;<4%-kSF0?@*M;il*ZrBcnXs8Oz<O9`Hk=)VMgD53o0&N>*6na5=Y;&_)0PUG
zxke4=ZNLrAD|8L{R#S1*w%e4O@Ft%z;<Wr|V@&(0Iz0<SmA=PgcqC7P7B`YLIrx`>
z?dYQ1(fE&S<`q#re(skiqi=dpe^J2iTZS4`U$#nPM8W^uFCau#VSUAxRb-Zbd+ww2
z7{z9;eukYM275^uDAD=G0MNT56_sfDQm`!D?w?;qcV}n%o`sXMX#C^*Qow8Di>i{U
z^zde0CUtgUh0>|D8U|p~3b`KOOG@$+NXiQeY<x{F5P695{C9<&-`Jn(P6Ls$mE<Xx
zu#%xuK!L#f8}s}Oi~R;l*ke;uuR+ziy1E)c#(%J$n_kTMLR8cTFh3(BBWa{Uzj2x$
zEOp&RK56e5&W}xGi3dXri9xDTIj{zY)&At!REg7BB~4A<bdmjzbG-E0C*@4t%{Rv-
z$G(uaZ!0WU(Ny$pM!qT>w_kgm)W;-1+{|{j9n_A=tE@X_FZbFO;U1sc#iy&4p161b
z<Ltqgd<%i77aV8Vj4NBQ#^Pgd|9&AN(sXuJx}Tq4Jlm97_cNMyXk7)lPVcFPpsu`_
zon2L*l=sb&J?aB-+hm?uB??G3<Ez9GS<eQn<SeoF073F%Ynx`brC*UG(RUH<&alB>
z^Q&bMp``$QOaYz|FqQ$y`^8u5-xcf4x7RlE^1Z+vSduyXB)$af;P~`(pTj=h2|nD1
zxUsRZ+f(HX?Cky1M`M5teLZ)3tq+9`0cL+xbhM>B8yg!!Zw(C%9UUFRWC4F~GxQ19
zmh&Z0Kp{XXCQ$)ukG8uo1Bw8b<Bv{!WN`&0U*h#?qJvH4MH?OM<mDgH9PHTe0nmGI
zH7%>)VkojLUFZ-{sA?6}uDyHWfQshjU+lGEN07Y8mxsn2ZUzGS-t}NEWk6+C)Ss41
zXVK$P6*U*ulV4n>y}v9d#X{Q&mYVlkF?^|Fm|@*-O!^(pNAtMI2Z#yg=<;qqbl2x~
z+h}UuDhy=!TR&E5KQiJy-g-hz{4&(7S#7GqTt!x#7eKlKWM9TJci{DSkuJ)h>-l@C
z9DnE|u?hx8os@J)W;Hf(o`th>)%09?s1$R_UOdF;4RNXuAy#14=r(=1BsIONsy`~4
z+Ivoo&u^>>4Co!cZ}s<HykG0);C=ra;saC@k&cn%fOZK$62tpUL&C$;xXqP|c*g|>
z(gW^dCb`xX6F)Yk$kl+)&B(U1Qkqb0$HX^?>YF!bn<dT8o8xTM)PGtp^|n@(&H7W$
zH}!ApiffD=Z3h8US${EpCbZP+@+3HOGFK@bgVJHKBTV%6>JacrefnNZs5a|DV;<nq
z7sebdIg65vo3{AgAItrklOudK@8{w8=_xAwV7bRK<<EPNUzQYUdPr(^f%LjFf>Pg0
z87Khg<VL4~eCBup;n!Y-Ga)e1jY2Esv07~TLc5)Y`xt~mC;Q^Cu<h?g)1d#sarN!r
zZ@SuM&Ksa2c8eW3+1c_5&syqrl@~h-3kxw;&&yx;5&?Ua2;o@y^zG867nXVE+}A@a
zXgmL5=w0CqRntq5&8ghTsi~?it*ihST05Q_{BPnM21payBjT5k;NalI#KhU*B0D2>
z_3?m1)x6uz`yIp#9*a|MU@}ZV^h1jOe(U_`$Mk1T*x`&Yk%-^*F1rG8-zGMb?QH=R
z2<9d>fBD7Rk$XPpBf7CDCa{PxEdKUw&pVtexsQN=pmO%jrjXujQKLJ1K}E%A{19nj
zhr^J1SPz?olMbe$g!dMlxTd>JjktQcZ_H!Hy4sR_!3AJgn4Vx*l3l3Qi)ECteG9lE
zuUbB`@O&@B7uJPN6dX|e&01{2`Z5mG`m)Urx?tdgiU03&u@$9741vkrkR%CiPEL;7
zDTc1AQPCj)<16gLlzRaiy0^a{@Qof|V1&S6-M@Y%@Y?8wky>6=avFkzxCPJ1%Lo2#
zK0{}_sJ(dWU*j6nfgo7}`0I!NouTU1_B?*}WJZ{dF3dmunJ<;?v7C>$)F(i5nqN|7
z(9;Q6*wBaZk&?}-^EF_GkK<~EuXwF+Am|CG=>V|*Psh`_C$h@^XN!m$x8t)1UO6Fw
zST^bzWNd2t1kBCW!K<LGM@UHb1rXm0u#=II(Mjo}lRNU(HXome2sBIp`xhwL4TsNp
z2!i>95gQ*5Fy;C8YPxUVF2pdYfbv)xG9Yv)@9N5j+R2c#`5%UovYg1+pLINc{@ipR
zZF}DpUE)Yu^lDE6Fv3Bu)z;QlZ4HOD<XHp8HSdgV8)7b0N5}Ke&j*kC19)s^uZE;4
zac9+iRA0Y-2F7Lpa0-6FFtLHHXHG4l1E5VC8dCpju*rfv5JNHdaPW%+EYtNC_B5aU
z>C>nD{E0zV*Vo7H!l=1*ku=gLn-e8K&rHj~RE43f1s@1afSSB?%>~{D!D)GAr5~i{
zV9d<FWj}RAs~4y!VMstwuPI`ZY22;`44CiMEMo0i)3xGULjjf`AubN4F{;65!MKQ#
zZ9c|>2X8}xIimX-l)ib!HwL^eh>(Z@y4hw!+x+r3tibLNSWblKjqBRr(?{VY+!MPA
z*H>3t&Vsn0wFo#{MT>jr5~^*8y?{dZ!C**IpXnn-r+|O}0Dyu?K>aGKDd1B%`NoR0
zDlI^mx_9s1-Pt$inww9tfh^^elsf0nUOB)G!y_V;<m6~FD=I3^&RqYV84gxVC+iLt
z)|kGAIE!B6#oj_2WXVzBzSxg>RY2qd^zP|XFen0~2_)7zuUc)~6$MZMTqqbJhConI
zJ2jFvnLzH42!{Wrx3RFX)4>QJP_hHWSjj;*3w`th2~{{?w|`Dc+j+kX&62_-;YRmU
zq2u7N0nWy~QG|+W9<Ukcs2?d}f-YMK5r_B5l-Yh&K061;E*K&*3kCNp3<}}=d_!B$
z%vwS&0H249>bM~0Tq<P#(*gO6sYy`sflg1Q;yt9W|33Ts67PJX`DQFnCD-5IAN+ni
z<a-}ur*5+tIMmLcKZQ;vjb`N{8qQae0R#{n5s}XGKIv2sK(7|SO?Y}<MbXIxeap@^
z2Ga?kCtJT+To1mnVs9PkC5?G;?h?v7J)o%#R#RpA-sh4f6A`4mJ75^-X=WHHF9jHf
z<mN{7IS}=;k&zJ~2NaBss#24HQ2pT^vK7xXn5haU<5vdse_(0tAw=L*=im-XOH2LG
zuxG(wQ4lUog|{h!GG_YWD`L>3V9b3r)#dRJm@%jVdgpz8q3d&f43v-|6ND@uutYG*
z3267HX=z)4GBn-b=3s8VUSqfT2vq`LJzyjOL8W^)m;s1?)EhvbpwP%FrH4o)$gn3F
zOknt^E1JPW$<-V|=L6e9*xtPxb>!a2+5@3Q_dO+)jRJf_6Ho`a+Ay)QI@apUQlot!
zd^E?2%pHJBBgLv!@eNqmy5p?tV%V(1wGDvXGBRGqErTfx&tIVD|LKutrA6$>dd2j1
zfqD^%fa4laGL2##rXG7eJ-xXmug3VDdmrL3!S_=re6g8piV!-G0k7(^PxxVrOadVR
zF)st7Xaa*>hX)6ru*feajUt|ebpVFt(biOS&<?1ifxdBefMNg1Mh}<xo);b`TZ4nI
zi_+1MxCL~6{_~O)gOK%CSC<>;=u`mGz31IvDpI&X$di|(Ai9*X1_lI}lcOM={p-6w
ioDoX*pUsS)``i>o)e;r9A-;D2Nmf!3QY>!Z``-ZhBi4-o

diff --git a/docs/dev/articles/examples/timeseries/timeseries_anomaly_detection/unnamed-chunk-11-1.png b/docs/dev/articles/examples/timeseries/timeseries_anomaly_detection/unnamed-chunk-11-1.png
index 4c29cf710f3c08f4580446c8722460fb3b301bf1..c9c9d6216b743ce4bf0dcdad542e45a375cb576d 100644
GIT binary patch
literal 15315
zcmd732UJt-+ASJH1q1{YL5f7ADIi4zq$^6V(u))Y0#ZNeHBm%S5m35R=~4rP4j~lj
zy_e8?3jsn1HMxs#|6`wh_Fu-i<BV~~U1I=?6<F&n&pV$v=ku;FI$FxKS6HrqKp<LG
zmFK!3(1p(+5Lq$h1>j2726-{?gX%986L%1Z`uh17+2Azv6a=~hQhokZ-zNz(?QL}f
zPiouk<JKf(mf1!qFxz6ooNTY?$I_57-Tl>0&tSu8mh|f$UsD8!y3Ku78?0s;gXu5-
z{=pc?6qc@Eh|Br?AlKNz(INHB*bGwWXvqS}tQu=e1_DJtwcWf70yU86yFs9Pl#Mi?
zcR~MOpWYMR=jAmVQQ^aKpdXSRtcQOqwCBwYy*m9PS6W=KL95jD5ut`7#MdU9i|Y@s
zOnY=;ziW!<%!F{=+xrs!A$#DhF<Hs`ahrH2p#<F}xXXK|Yu)#_xK^(lIF1z=bo4##
zB!-kpe~RHZEwdq&tByJf^xX=Aot(BUUUOS6A~%ci+lxv@<NdOkXrIiKB0S08d5({;
zA(w~k&AeZ!w*Cn|QF>*Rh26g|%a@0<iM>QpBxm?63t_mf&AzHpTe^|a>)CZuB~`;v
zIKG89k;bllJcuBZlQoq6uGpITAz-t+%7$KHt;xGP0I7P()lS*YmDC!T?L;*~lztXC
zT;JzE#*Z<tQ>uK!pHu1e{ZV;#${6*J>Q&@*hmJIS5DLaTP;a_95l5p?4|7<rb<R7k
zmb7O>b`Eb6J68HNID);pp3ucCjjv?<d`ed2;7=&koIS%d#gua<#J%-C5HLVt7ZwSK
z$NNd1>XY)9eIzV>hVFRe@EW*dkQOH>OJOs(9d>7z2{Su;9Dqx=7zs3g8ER<1+ba||
z<mT6t>UjxaQbYf#9qCT-2rsm-rJL!5+`cht-t+ct-@d{Zr<t>)jdw8}Psq}wMbf?|
zf8Y9gc{rb>g5-l<jnwg_n<(m$k*Y{HwP)$28tq~x1u$6g=GNtTvp=r#4t48|TpHfN
zssHqjC}KuWsa{X;YlhAE9@P)rLM={TVNor`6^&M?pKQo^McNp7wS!BPOk#($wF*}S
zRJ~>l62+)?Jc`Funu=DU-K9)Mx7iI}au0KN##F=wXCLf`9%ni~Q_tgyF|s=OWs$te
zIHVG(e!@L}S#mLgY&()ZvU3{sYRI9UMODS?;2<12k~k`~ns+@idopvVhIdQ^9ZLhW
zD|u!@n|qiT$W$a7(6)n`6^SnzLwFw^h92xxa`D=E&o*TD<Q+|_E4VpS+%Zl)livHr
zACK8|cTh9uRW;A_il_|1I~opJ=$>3EtaEJldK=-L*z6yNuT7o)?NcH`wRE}h+l9qf
z!gh_1;%OzA#1Tw+caxb?!$vx59P@dvg2fl|VUf0y*6uI~)yH1uKkjvv&L0>RvT=8m
zkEA=An#78_22A&wN3tad)x7zRg=5X0b4xyohsv494)tm|&e>n@j<hbkmI!ajaqzqx
zxZ?`0oAGWK#)v4H^wtR%hAaK<`Zkq%bgy#jf)ZFGX8GF(@Ns@ihRQ>wv;M}X*TZFr
z*v1?A@p|fREAt{ZhZXc-2R9J~0k4rM8P9N5hBsk>f!|)l;>T7V4N`WjmzU>Gpg8*~
zf+ilwE=N37G>Ek8C$%u!n)E)FD*BpSoG*7&AXwWXOJuYoDcBOW?ezM3R4!W3gp|H$
zP*J?>3M`6x&o=oYr<9Sr$R6}XsV*wCT!C4EeKln9U~=RN%NayeA__cYE){6O@5pkn
zaob8LxL<#5q_zZ^<U7Bwm`t4fBHE!;n`iE?@6OPatAw&B*JQLdswMpvj0t^=M&}O4
zT<%UeR=P2rO|WobGtS;EbC=?<IYNCH-cS?Em!x`S^auTnTffC|L@@h$WRCYWX^!B5
zd-9IzeP;@?{vq0@hRZy(NN$ed<qMr?_0obNS^Sl?oI)#cb&09^zwUS3o1NY9dlAE>
z8m_rwIYiyu!{0cW^FU|m@E`q!^%~u5C%-6qg(KwC5S^uDl>>2YM)NGP*Tr0XaXOB5
z-vrtgkJnZo%utUsh~x_X9;Re6y6#a*?|^T1EZ~w3OYX4kii|=GWkk-mN|>{HupgZ%
zVh^x=mMHfX^#celu9T=<QEI=wj<S)8u)_*>DayA+x>+=C#TMzRD*RUWN<}*KzzpGs
zzkaM7F`n5&Nd9T1{scskJht<KP`H)Se3akDYKLh3Z2tBm@190h5pP0xVV1~@yw~9%
z%ILWLGV&>-f?lDECWWfE+u1RFGQEQLCSf+hRxM$}Gv9IRLY~3f@B2nxPSgWbKYRm+
zvJn&8rJZk2NHR%Wbu}8tZc`1tV(e2i9cShhl4bies55({z~nrx%Vm9|R4m{B_qY44
zk<R{v^`RUiYPi3hc5*fT$(cZ9<kHUh5c1lZSFY8Nn=E+UXOO#kK&TK(aW^bDb$k}W
zvaP!6Z84HxXIfU2aV;`5F`p}4n|Qdp@K99Veh4=8t|bIEDUt2_A-PBGIr~hYoQZ*7
z|KZb{Bav%z1%zP~O!l-xowmYg!2Ge>16ef6je9IIt4_altsLvELYyG(rJP+;;&@nI
z?f2<Ro>RpGsvkX9QS?nVNK@nD=G8dNU^1E(R(U)*VWwQR5%4R^sJ*Bsq$z1qkll34
z<bcP`Cyh-Ia<*MNNs2d^Uf86Ie-**}F~YoOmvu}Vqt_>5bMh4on`U2_S_@M&i^U7^
zP@O<rVN>N_S3g9dT;)=^)Mt`W6Nq($g=caz!Fk#ro$J+~N$VfyMTiOhURN7fc``;o
z>oZ7b&+IerP5)V$z?Rn3JzQLkuq;}N{A*pqXF$RbdX`XY^KH8H7#sbvEajQHq_5jQ
zM&D~aoHoa0`f--slTP`3DB2f|z{Fb;T-aXVC&F$1v0NQXi_CnM-wUWryj&o=L&01b
zu_XoOznOpHmNCeZCM{V48InxXtf!nIy^<u3U)oW498J5VbXY?dRDZ?Y`EkI^IOF6r
zF3v)ts+@A;WrF_Yo%Hc=Q1#rn+vw;`uqPkPyQLjU&f}t%E9x^);mJw9xAHXp=<EF8
zH>m?%!!6bTs8}ETE@E#1WHzTG4gy_iViyIzD+rT=e%-nF|Jzfq3KvjubaMz@PEJnZ
zSiV8NVV%$0O#(TnPa)EFVpJC7E5ymkdFRfrXCYU4@P~(oS8nnNysYZyF(+%#D{1={
zq#C>Iw%i%UE<2eC#czyQw}nq+a)W#rldLa;o;;<5fItkvR{wTd#m7po@cFj1R~=!-
zUr<nx2Ok|BU1(N0`ujyQaN~Qyl#nNs91Dw!AvDa_85nGINynS;sHiCC!RzmuzC4@w
z?+KxiExBa1H+H7Z7x(ohKSbBt!{cSK)qACk5>+G672zP}cGc(4pGUCE@);F5=~`C(
zHUEKv?kmIn*HJ|v2CIvrI}FR6G1N3PkM7sh)}k|+L6_K9<H<le^at@Ts@hB0SXup#
z_uoVv1C4bWqyxp@0WJrB{0MYKKvs4-6MC>U@t&Mw@_Yo?8oxCu#xC<{CVXR+=*^UK
zU(rb1CkOF7Zu@^hW$A8_xh)>!H<P72Be^;<1-tf2tlOM`VF-F}b=eJ=8S8;`#g$nR
zF^@GJ;O@T~{uz6FNI(3>1R9;lKzsQEaA+Qry<>VR%sH!<KhhQZK=eKjFE8wLw@rMV
z0%UP9r1ZR#LChe~C62%Q`!A<Ool5m3P7|eiIa(7n)SzO4Am*dim7c^|=!vtkvU1e%
z1j5c47!2bJjo(f)^>cMs0J46G>L26wFZcPEQ%Ca{XN|BbuU%80E%L&pH4~X%N}8m*
zEc_2xK<`eQuQV2*5T1kmKW1-_F~W^U$26@qP+(?1{w$A(^)6IkWq6{^5g-oQB?Q>J
z-S#)=l-M1QheisW)rf?0d@IG6#J5dZAH!rlcb1Yx8c+jF02*!rcH1_cYq(ZOitALM
zj4{hf!4f>`+$2>i_V>s@?b^VdybKaYt7~i~C#m25B#kWX`zu4^k)y`&B3@-+cznLK
z(Alah`Uw!ZJ(fZ+>U13fx>IeMoOWD_$~h<HK8~Y#5stX$I&%}|)o#nFFro-NUq4S`
z#^+had>y?Gy)g%~f!u?(LP=4c2M1pEK2Be?me&lLW>J~aX!<!x=ka2WyLa#2xzmtY
z?=X_5l`Iu=??n@^8rsba(Y6MXCx=Fa`-$`chdW1{gx8K@Z(in7o^&<1=UJkl8A@Rq
zU;PiZN>oKb0*wo+@tzcI`mU9a|J(~s^00Tv&kLi_Bta*F2<TyZafpBPz@Rz>=nkyp
zGzvvCS@tMdfO-*Q$wOKb!|mgv{E0K(#K}(;6s^7T2}k?06rf-9B{Ksy&-SOQCgEN}
zu0p}|P30D2ql4X!s&1mGPJ-Pt#jlzNv(z=emXwsh>-IX6B;5tj{7xG~+VW?(K6&$(
zFEuvodXs*NJ11zGUI68nC<(J#YrCGfYiFb{=TD(w!Vs13scN6gy8Od!VmH@P&M>-9
z|2}wHJU2&MT3Sph%*PW9^OGOF(Y3PrrFxbXXHlvu53<r^q=R^VEXoL~NHVLovbJCA
zG4pl}kXRXc^FBl_Z^YWVu*KiP&F7c~fQ;wB6z4q@f<0PYhLP-s*MpPk4Q5}uzfzih
zC@5ob5%iAN_$Zuvq*?h6L%v~OP1hx0h~`fIUJYGX5mrt*I1#6-#yD%3COENx7kuB#
zhy{V(1vKw93QXkc_7=w1Z6(=t3E~65v(GB?x52}b{XcK?G(`0#OH244IN^?C_>Ghz
z*qzUyAmgZ;K@&M0wX+4HpUP_XIy&KIyTML#xe5vjYH@<j(Av4*FEY*n#<8@w#WHW<
zw)`!#8Ufz2_jXHbYirxvF1lri#iP9^jdodm)p;*hy}L~8UVuPL)?0rr{2!;ArVqhZ
zZ<z#Nt_@^_z~OM!Gve8a`)V&{v@lV#<^ss}Zv3ma_wV0#MpuZzdU35`tWlAXPJ@y7
zz@z0jW@hFUe70s{h5bO&@fk=*;HoV-BX8GUB)589kA3;1nB(YpjWnneNP<?h?Ck7e
ztP<k)+2sO!)(5kIND#>l`S_8R`QabaLNXo3Am-wq-QC@Wg{Faf9lSjFOVo_xnP8CG
z6X9maHzjKa2Zz_M(^bv$M<z2<WnMjE=^0arwM~3Pen%c?qA1XV@v66rKw~sE+QM&l
zp~jw>*LoGY&em6KUw9&?1l)w?a`Pq}HIt>z6y;ZcRXj$m{I7Y1sPqB*fD=q50A23_
zLh}C|0DDEqk1UjwT0c@V-R0!e+yK&)L7skzsyfIm%<A$U5>W7k#wAT19UYnVz_a@M
zbrQ=i-Wj8W{JP@<+%+&DK;tWukVP4=BrPufCoJq0;p}TK1+o+bY2m6%clBxo5Y6Ag
z*Sgz93D{ZEQk3@b{<^iUeBiOi?m`QY_drY17e)7UJ0iI){V`%u=>lCVmE#vcrO$=Q
zKlbJs<m>9`^=y_qj=78%r{Pb?b{-27FI-8^)yv6_r=q4lnGXRwO|yUw;@>y5b#>*R
zC-W9*?`@~?;zph`AXoYSmnO?!yG%M(%wes6{yaDHCYw8=?|$<BCL7ySrK{yXf9F;m
zZ2CyuvkBa?94ZP*x%Qv>`Zw;2P(v`IbP1lUa+`6}MvI;m_LQJ^P2B?V7_;zrqUrkY
z2T>MO_n60_JDJ?DlrP=iod;MzppOwI!s^>(&J>!bev2h&LhqS)UWvyRu80V|nLuDh
z*iZl1d2}84L-o!THiY=WUJ7l#VWCAWRJk}CvKfaSb*tHLX2m<8klde~E`u6gce4wp
z%i}y}=cx%VjSQTptJUME98o^G*{K--Bl!Y89j@+;6|HfBkDDPi0s2OPu!?=jPth|~
zk!dRi9y*MvJ2~^Jh?C9*a(GMp?M_1@zg*~okZbwvlKZNZ&C4InS-eR5eP}6;U>!;S
zlPbr8Sx%rEpG<^VUC(y5<MqVaBAga^f@?a>EH&Wnr_HbAKQcQH+y?&Un~CtIU<!@z
zdb4(|;eien$1pn?T;taCUhA&<)G+Y8<)4=j$*^~`9X-<9>)^9l`tr>PYj#`p+`_3_
z!2kc4XtCUI$f);QEVmdrvm41bZxs_=SlBgdQ7%Z<i^7Inyiyh%94za9@aWCppUjlV
z$cHXdRZl-$p2+0?MgOqRtRti256YEB%71Mwru~uT0F}C_xB&Lxb;R)KjBJ_1L|$?m
zJh)A+UoAYPg?yu8TXMh0DBo8_=#eGJS1EsFw9w3Zvg{2Ho}A+9pFe-bGr52L)d^8H
zUbZNLZkE!ea!3BWl{z<P_(v>`pvvqbfSG|5Rp1f>XV&z4|0BV>9b+eM!zNIt4NRE2
zc7Qq(t=YTRbC^6ZQb+cmLTZ~~Ki9`SC(6M!kCz~1$JG4wDEWYsv@Zr=OZ~p5t`)wh
z`N8B4%(3qhqzK@h8j!{$-?b3eho8WqKlQt~wHfcc*}^;=8r<gzjr3$TeP!2HsdWLg
z^B7?zF5`2@=r{9siClqT9g3^Od%tW~TjTr*0l7+&V2QA#9}LWHkTEPe>g23``~jSi
z9Q5S96=QceW)^%r-PT<H_SR6YpGcf?6t}XrWyH(d9gGSFHM6AT+<MQE8kE+l=R<MD
z9JhVGYdZdOd)CN=n9gMZGSI!>-E_8C(oq|ABQw+GEuXCgjW#3Lm7@}~B5VC+gADX%
zmT9+>O`t0*kjAZkOq+Cx{0c^)%KQrGom~_jqx<QEqwQ*Hz4Py>5`$fh7;TX;rYe<C
zbQTEY16gSlXv6*-ubz#dBDM^9BFRIUXSV7y&b~UkPu}b?4_Z0)$^~d_5ddPDKC8XS
zj1Nqg@~(kEdt|u$QX!bcEBCNvqu2`R$HjO9P2_$5F|Kl!S{tj)juCf!)z-^d!I27Q
zbVI|Fs0H9B*Ox#iU`T8kyN~->e|XkZ;tk}#%6A}M)JnbGRKVZ$ZcN%lXC);iot~X(
zeuYAzZ{NPHUMC00UV}iqm>7#bJAR066^-N!rdJr9d3o0LJ~B+c10QGnSFJu&+<CGL
zMm%b(W50Fla5>J>6a@y|+v(mjG$1B(BLw#yExn3NGslzR2hP#<BG%Kz@0gAzm0jXG
zE3dN4Ja(RZ6w|q%)cohmjdPaD;A65yDXqVO+S2duc3g!kWls7U6&PfEW3_C3`MJ8b
zHjx71dkOTNx0Di+F>t3VAw6cJ^7nqRmGCWYG1k{0kU*b+OF1irZ!NKF;zdTu&|_aC
zwrW|?O7;F!%l=-}^Id^x4Bj#N^fiVYq*EL(f0V!4`DZ_dMAS|Va3=&+%F7sV5?13e
z4q?;vmE#SJHy$BK2~#_}%c_Dc0abV^(7QI7<=2kVge!23g#L$f`LJyi>s&ILp<HOp
zafd(DE4oO(x|Aml0RsKfY2RkUp-4Z<Uxaw2jzn*+5*1D=uEGUmXPzqZk%PXABCJMT
z)-21g$KOaqp*J%8-X{U(I3^kc)x!Bf7Z50=^_0!Me}kfCq{99Z2o!$_krY6fOy;s?
zl;@l9jmgz-is>o`fr@Y8>})MhosCKBAx&W}EnU%nsYcL*s*FbeW|^}V3N~LR7{dK>
z?uKNNfr=RrRu$AcLi2+Gt;E^Zash}?u;6j(CD(5?a$DHtY4;;!1PC<u^QWv87w+&N
z1FsbR#-!VVk|Vekex%Il2%ZWM7z1Pv2-F8;hY@bIYVdTiY}3vf=~H_$D{<hEVfjlE
z2xOdcl(tm^7TlV^<2T-0mb>WBaN`elkvStz<w?`u{l5b(k-<@XQ<1~0ve)mf{P>Vl
z^s$X~a{tK20PV-k=G=dtUbMX;mh^Ln9abMckgDKI$t|sPPzF}p>;Gx(ynO+vdqz7i
z&A9nJ?;GM{#?a*r;{69(AG2qok&X#Od`D#0rvfM|K*#O_!yhv9xnQWI5Pm#xwRCCX
zw|R2acO7K8G~f!*Kct3ehFy+{AnrFGErn<=N0&<&k=%aS^yFzDK~+GY?-c(-jDy(!
z0HJc86F469<Znvk1VJpMVPbbjdTU=7ve-_k*n&U>O#<<rCvC&#gL_l4jRveo2IY!>
z_Yblfe;W<ktDbG8Qa^ZkH6lC|#>Meb*_Iq+bqV6%J&8fIIzNssneRW=CK-8b5xX#{
zmY&5BYhY}YB7h)a@Z#<e-5E*cyaB5)e+FEIL|k%TDcUU*KR}+qWZ8oIQ;E>2pPz8E
zMa|~AADb6iLMxpn(hU<WR=iibnE|D`Oi4{`hJv11!p>}U``X)$%j{o#xJ>IbcvI({
z?TxxmX8uOoxa4QTOpF{H9Du&bGcHM2MVcTWfL`s~oooD%BI^&V?(s|_ATZj1bd%2c
z>)oPb>_26!?}AQw0rkvrPu#qh!djoDHi<B+Kw0~)w^_<yNd!pB#^ctF1$-k_2-1Kx
z=@bB-^w-LyB^c?!7H!zfxd<+<Y6~(MI?OOYp^I(+voUQj@gyAJf&f-J>S}0fR|3>d
z=R19OJ8hWKA_n8E`z<sS@y~0#yg2}6PGsP`d)Es);~WJA0pi^2J}_4QYKORYgWMLk
zr)y^ea6PdP%rrJ4xq1>M`lhF+0p)iI)JStj0(c_2($$rS(X|ZNRSXD_X{1?V0X~TP
z3Lti#{{5(Q1OlO%Kz4-&psfGL<p2NBmkI!jzN_;fo*w=Bb%%n2BFfLl=Md>n033+#
zD4-g*7ujfac#7*&j}>^M3zqWu4Y&+@w*f8izqezQ24a#*ymS9Pce7QC?FG<xxG`sZ
zd77EsRlnS?@X(CVTEKilMEU>pn^q!u@V%&LYWiD}gS&H2+PsN#_D2M$jE5+Px$7gF
z?C0i_QaSlh-_&~bXJL$%G9b{=mloH!LG$sN+O4W-y%WYbcE;ge#m-?T(gD-VtFUZ#
z6}A=Z1Cts6;r3`lu_F2E(fm+%QVM#Kx9h&q0lPN;i_VYwmqBwPjI-8;F}ELL&rp7&
z59Co=85_$1Q;*z~bZy!vZg8*%!p6r(2L?j{@oZVr6oq=Z0X?+a*<>O~{<Y`u4?#bY
zH@4<#CVp=-c5(YSS=u`Ya9+-JfQP#~!U}-w@#q-#j`caS3}-lt^GEtpLCJ%9{Xe{`
z`iu&IF_k12IGK5ig~hbwwIa0`u)cdtWa9BmaaYaer2}31uJbh~JEvY{KUjW$E5j6n
z_84EVbr0g3v^xn6V-a&0{{2Js?R1SNkXvIh03}419JW}xO=TfNT(dhB!p$U(F_@c7
z$?6(I*)tQvoHkS6;X%Hmlps(9VkQc;-<Tt41as|ljF}#8>a8XEFV?iMJ$6gA`vwB-
z+;U8^jx8rxM~R+w8RbX&@vHHwjZ<?hW!4fr*o7qBY8iAM2Y>n`4lv6)iwdU+K>rz|
zz)wm5$sOq(bDY2pH#+8<S=;!&JNLo$GKTc~bOcPXf3m-q?>Jd)?6(Vw<`5NipD49m
z!T;b?HbpT%={PrJ#FS5=PQmnU5I22ed;25=T#a%9xtoJVts6wJrUEP{D8?o3x<aY9
z5W7yT&pUv^vmoUQUKL&fa$m$w;J=BQ&m+Ae&#nP-JZ`!0zA~W1$BL2xt#t0<QO5xm
zs2v0%?lLu=S?`B;IJfGAp9(WcWvRtnIrjsl9zPxj{J?iq|M#n`DbAQd13(e~&BDsM
zx;l*w374r`Vh+*g-iQo!P_>nsii-VI)h^(wXlww6YB`_{-`V|1b2<kPSqP|t;r8w6
zOw}7v=@xL;(Siv8T>*t?Wd^v+W^SII<`qsqRYiDsAV+)409qDH{>yr7O4~Zs%pd8E
z7kLIBOB`-)YN|sb&*kNuB7OIfnSw$SU;^~GYKDLQTzvjirsu+|{(yde|NcEsOBAnm
zc1{i&kme1cb8Mn^{uhoxAXeY=y#W7})BW-^#`PivMNVdBJj3F`g3}=2#3@j&1Y`hR
z0*I&heI72ZzlZcA@MtB#$2^}BU7D*`SAH}B!1`DC|8gUN9_S<u4-W&vzT9<oCbMX0
zeI}4F4Lv&s#7lz|B_y#r;Aju<SGqR=oer3#rBL~IZkahb688O8y-AWhc;K1Sb-qJ9
z56BFk38TB2ECc4Zwq9ncT#Gt(MV%=LlRrsQ6h`X;27)?Z!01lZ2i5|Z3c5$e1S9+3
z!R+6?{3eX<*BeYs60e#r0-_jHOdDkKc41**W3<rGVz$;Bj!z_u=8=_^6%iT8{B4*I
zKs!J~zpqk4xPiDT^Q3-v-5do4N>Q3QTFDP{LqbCOQsq%VBgEIQw3}T-wA_|-TN<;3
ziNV3Levj_mYX|I>qc{sVcD!U!8e3<rYgz7K0D)u!!L!HK==2sCMEw4Kfe}1KHE?sZ
z5D0*KGVC%w^sKCo7T#~)?&5F=sC)zBPJCgS$NJ!kat;PaU4t|_J{n1nMSyJ#Jv}{{
z(6#=w-@oqxhK9_|Mw(v@4_U&1<v}12By#C)OYOJVu`@D?Lr;B7Vl_ATVjP-+sQ~8z
zx~ur>ue<SNOa4E9{*-W?kqJC5Q<asIn{lJ2rK2;-9|6+)!`$;`cgFBv6g_uQzGKbn
ze0EQdaAQU0qPf_~@_ZPH7;9EkTkD^zo2Bv9!onif?CurvAm%6ayj^Z_g65Ouj&{28
zK0CjFL}PIg@R_yW?#w;Kp-TiU>g=N+S8q#YjyPz3^+U>WbGHl)4P^?d>H^R8IaVVA
z8SjbK_3PJXd=|q1MG-#?w?LXAAe#IVQ3(l?C>ZGggH9ZNrl4^B#tqg-_Er{P8K0f+
z9M1|y?(BgYlk5Q8AdH&1{v5Nmwr&e!wF7LtB`T9x@(UNvO+8&pq+hwBkYHC4GYkg9
zL5PQo5s4D631a~7c@3R@%n)wXD|+qPwFhPuk5IJMOYJ{^nR%B!1vpnz)$S{vOiXcr
z<!gch-(v)NHM%qHJ=@aSdVGAGFno><Y;0R|b5>DBzn=V^OH(2f?n_qHNaYn)jn^?U
zF^$!DI^win0QMna8wjuK?^fMT4lqF7AaU&M>};&ij1VLWkgemr<u0ZNrZ&2C0QoAl
zQH|j<(AaPsEqF6omdmBegRk|*Cd`J~3g42Ek+CRlX8Xo03Lz2CR(cMPk6mW#r`@_*
z77MO*Pxcq$a(XSRaW*P1Yo0PjwifD#89r^6KXI}T6dLlLE!9eT{9`$d%heStfYN@A
znJiaTRpn{<NAajTNpdK&9tgd<jT`S2ZEb8$_qv2u@G_FS`v9xE#;Ft!P6h~(cdUGB
zN=m)c#6z6jJ}|8w0GR!Ki~D-nGXA1~?bg`X=!cxiWWRA^aeI4vZOvYH7MR4!S?K1*
zhQo|TEY7mp&DPX3LG?w>5a1x+6tbAgB+fOG1BM$k)!WIbAU{7Js41yV4-Aw5VdsFS
z#&o<Xp9DaJh2}C&6?o)5qoS$M?kNR<|KK5aG6@A>y24zO|K(S}ZS_v}d#1`4WnnQb
z(ub*V6mhbxt?jh;{Sj`}j=4J(dc6LQj9laE{QP|E1|4W`7@&8eqM`sR)Yy<e+aKx@
zLOInd15<;yORctmr#`gpY<N$uxnTrEXW)q!fJd4efKj)9a3GHIIE}M}z8T4j@l`6a
zL=gcc3~hw7p4)rdwb&XjjKjH+sHLeXhI*+*#iedpoFgq&aZvWP)|-Z(xYS!h?9Nou
z<3BxXqsKaVB>+)zzDlEBtEqMNr#)+n0H@YMvNeA0t){~A^s`|!&y|(!j%Jh<6>stz
zc4x-_{K<UdhPB);zy&TdEj*&i93MCK_xHzTXag~a+uao*iAu`I)M@;DxY7fJbPh1Z
zJc`L$OgC@ZT4)0S0JO_8x2o!Kxv4XeX%irkK<4Qo1@spE+`3u{OnBevs-B)6FeAXl
z@Q8>wu3VKLcX{yb?JC;Z+Bo^t=H_Pg7(QF#f$hY%0sA_CLXofN`uh5pFJHLTqJ?vL
zxwvkHwWceECJbk!r@u5e@A1B~Jyorj{+tSjT1Gf2QC?$YG^}tkiF&Q1q!bVk09dJ-
z8^C`9K?KZLCj0H%n5``*VhjyXKE&g2L-vnWG!r6dfApY_>&T^fcv^vg3$U6+`i}ue
zfR)`?w>$0D($doYz6b98ZC2L&0sC(3LfGfe_jEJc5p-KXS8*a%k63`#bhl}!ogZJ1
zc<#!{S!C({gmnV79w7IZvW;h{$3a`4O-#^kZOPeEQBjo&0Hc~X4Al03Q~{Jux)bl4
zlx7|^gn#)W0w$+3>O7561x7?grQOYn6^K-$$5K+0K!Bg>G2XnnytrsR(+$)#yX7w_
zd|q0z!JP>La{var62;d&YC58LQ3sQbE4UKLPzr>|R^n9*k@=JWC-0Pp>WVKn1+fPz
zw5FO*&g|a+94-#Xt`*~5eD^}VY)<mQ@1fs*|MbERo_|)J1&a%A8+Im~7sqs@_%9;p
z73{4AiSBbtBRk^^$(;lF&!1pc9<WAxaz+E?v&fipBUpyo%k-Zb?LU<o|F(+f9l_4o
z+Np9cUc6Y*78LLq930%9MY$5E>MyJVl8TIstYW)~B|Re}IVq_-^_}s~Et*fo#lisW
znn}E-O-oAyO1Pr19p}cV=;+~U_tn*7CHj1hc<0uh9%rCFdy`MU*prjd8-QS-w9%6w
z)-@Vt{0vBy-k3?-O~Ha!K=F7a-$;UVH7kf$PL2p5F_55<g{&o?09-n+L^Ga2R6hdv
zF0b(Hne(*b=oMnBBB1s;|D759hsHwm{MOHF`b#_Lo`xe|a4~{kz+39~hy$<=fTyMD
z(9Sg>O4ZmK0x%gpK*nhR67Z<G@dtsfH*ekmWTwL956hK4bV5psi>oVqT+?iZPsY`8
z-b4>+0Uufi95ttD6~@2I?3p@jrNG`y)SZ7X;QmYD_3zUc+R0KQqp}Rblm5IF1rx<s
zzfYAg6CwX~=IqXnMi|_p`fcAtty{^vhwKq4Y%yN$*hvz)46B=73sVFu4;|eLO7d_f
zU>z%xzk1y>NQ)0$2m6|6X`;$Wgb`M+<QnS$Cj<ACh`VdneXf=#C$mi0+8De9OrkV_
zkJs|d5uYG`kQ`AVbX#Wug+~QiO}lOqrF{7_R4$S39*mlQv5WTGoNE<>u~9{x@kP5y
z_=lSq>WN}+qpOYy$!#v#iSU@Dlw0tT$7udt=bn5rfrlA=XXQi!CTC%a@8TgJuKD;B
z&HZeux-fr?=)1NNSh;tTXqF#PZ0<)SJN?{Y+~X_9nDs|n9SWg5s`qb(stYNF5asGj
zYA5W<v{R4g8V*+;D7n_fBm}JIkm1ex<dQjpRix#Xvluf-b%`wA_I@WySgeN5jJ4I$
z5)v%tEc<G8dhG_TR*b!rLk+qy+)eF{uDn%ER8mvh`PxV;Em)uDzY)ODc&i+$(k$p6
zp*S;*e}i7JKH%xnPEK&MdWpE2im6FV@L0)b#~&BfSu-fuq$Hy22l`Wmcym4#fhWp#
zmHnflD~l#Y5tCxf6$g;+W$=l_#_qGv^SOFqm9Q%g57%<2zr2};eRG*{8!RbngFXe9
zmMM=2&VG`UWL0cGA0{E#ba_VuScv=2?Vm?(p(PZNF@z#Hl9y`u^eJR<n>yl8W*d9_
zX#M;ShN`rk37xeHZXo%&6|BgBL!*9#lu^k4)_J>BuWu6T8D}T9;~!`}2q;NQoaq3B
z+GWia*zXq}MkhdA38#^?M$@Av_4+n#OvHm>)13(P*NYhsMtLb({bx~I(=HviKM1hi
zkT?)}0b8*>)60gi9}E9#%mAxT?;5VNMD?zvJOnHDTc((xjAv(Wd@FUTIyvi7ln#LP
z9Y*KANu)URJd_&*wi}AfR~jKLj`_OpIC`UuYJk)$YGV?MmspY5-Kw~2@y6A0uyyQg
zLlgD=ewW+idpbUY!?|Av_x12b5Q&2Q#`iQPvCG|2H}YfryvJaubYxj^7(B{7L$XoL
zUmN!qNEz-^?sI}|qA3+%y<6FHxRXQuQ`|JmNtWc*aSOQ>a~<xg5o-%{5jrp>YVTQ^
z1g=o;lYa*mVa}KAYg-t%?p1HyS>)AAsBm2p2rkYauJHUJs$8_@mcr+V+(1bd==qid
zpn3Y$-!40g^`nhlj`$kBlwRQ$jZ#c<a%P#e($CfK6gvCa>?$?N4unBSjYN1#mC&k;
z0cIAqJNDtiumYu|sI2+Hj)#$~t(@8&-eeiC44076lt}u#1!uI86l_UjMK4>eeCEWn
zH@`lTULiBb&R#|)i95}^Xwv$*jxz;NsA75m>|D#-EH%k4T!XXAE)L|G-wdN~%6%0A
zdBYq2k=ZRJZ*N^@Gv^1M4Y21^Q;NI^`#xm`9gjt^XFXv`ggwK_@tX=u^B}qERi754
z1P3}-({hJR%!YOSEz@V#wdI|<f)(mbB`v|Ta?Z<q!St>BWg1M{bYblj=&YACRvf`c
z=W8-nS$LKE*-MYQxYWL8#5F!J>hA79No!J$TK#Rg6wzOKrj%*OYOlS@SIH-JOZT?S
z!jcVahkhB&vP$F`MR04SLI-<>7&oKHCB=Bed(eHW^B}j~UC;2)aETQe1A}yEj#5?@
zVs_tk_Sfi1<xC1G9L2W^fSHIKPj6$}n89@gN`n`(>FXjT_uYE+Zg+_VR)?n^%x$qd
zJB+2K9dF@g6uIZ8Y?k7hSpe%}|C+R%*<a~zutot473QLUTDg2qZU!E@j2|8P$w~$x
z68LTwP89iGYD?q~@pdrH7)v<_Z5^(>q3}7Hb4YpH_h|DAKKW)vINkCDNbAcN&8uh7
z_68xZRO0FhN0^zGRP^gau}%w@3ia(6ONSe4%qOeCE)N{bj;a@e`lKFz1<<Us1;l6)
zHj=bz?R(KZB~p`U9k9T!P@0+-L0s)3qfO9MWi8*`bW!Ab^;yPyExYDr%wATPVcwb>
zc4xhqSNqESAOZH-h0ao^fb3ick`@?xJkgds2unKM^la8!%>b1GAB%kd==4Z#NI%2@
z9+vv(D5JHKv3Hzp_xHV7`Px(D3O>Yqvt(m6%{H2F6xBj9RAwx7dUU(oMBCt||9bM5
zC^_G0>#j^!ZCgsoH`q)%?TdlZ*|8t)TANV*@Fq8I|MV%U()G}5AX1)`uvFr4+*$jT
zwULM9JBo>18nBvci{hc9ek|@hlXBu#KQ1Us0QEx7!AM<i<`aj!-pQfwY(yB$r@i?W
zIBE@JrjIYnb=&K>WX^&tL7aYjW+(BjU{#PRO}geVs=2TpU2K2)TO<{3a&(;B;Z9WR
za{D6AS$%jc)z&6A*`^X2stF4)M<nraQ-G!Z!q@NYxc29vcZB4fy}YKwrQO1DFD}Mt
zm!vblef#Eoi7vjr!4?=WoynH-wJOv<JUBbAj$iBzZZcj-{&%r8rk;T#_+WCcp4Reg
z;?<vCoV2`SR@t;INAP4$y!UI3sb=qHdayt0^Xv`I$Dz6Fl@S;@FlTLa(RBJyf3>J(
zm?dgG@}QNFT6CTJ28<Li*?lu7a(nv-a?mN7D{hjpwKNS9+`t%Sy^_^drA*_q$cOy!
zue{5^*j+s?$b46e6u4K^n`T@Nxowdu<64Z%o{iYtYxG(_q9{&U{(G_I6lhvObNJ9z
zX`3a1c4Z2b&4uI?H#Q{li+nqWyB86ch4^=u<<XEjmYqW;MpoKSAsGtdYc~%hnG`5r
zS>~8vz6VlR+j9K*06R#bwV!NP@AP%(Br+p;h|DD6rt9hHwA<{0y1lp;7dntnht%u{
z>}uB(+Dj-1=J4sdl)zrSuosE)+Y%>p&DIIWYwgC$baY|D7h(9(Z{1WU2ZHQ6LwY|`
zWojFE@Y`S+dIta|`?jPvrm9qZJ$v<Nb5L6ZVkcUOBltV`xng#WukUYU0*auTnvhZi
z99k@@&e*Mu%on4z3Et#YUQwo%+DWRBAL(1$ajQ2v4fryO;lF!ZygGaq$6h_Fmp$1r
zz2fYlpPru+!WFzwc@E-YCgJBx>(Xf5U;%x9NWg4~ez`^JgZN#gKvG3$uU&ho059e=
zv+Fw;zK%#p@b!4f4L?3>)Aq4hh!%^(mu^o+_S&AIX;bA~N7q~o)G7~8^%Dxmy1$9M
z!LDuCnuf=Vx=!)O<RKB<oj@id?hQUQ=fP_AxaEoX4b5HQPP}^Nw4L<jY$G~P7NVUl
zl6Z74RK)G<3F*r1w!T!+I@^z*QrXoWc9tgjdT0^G7?b*KB}eFwt;ckiCz0~jGX)Vf
z2KPq-4stzAVAjwZae1<>#1lj9__EXIUki<TG+KX}XsV5XW!xK=er=FFMa>L%0cjQi
zc^*eMs(h<7DIvhhspob1iDr^00g<NJrO#4FOqnk^dgD#UKV<yH$aZxaq;<M^#>dS<
zn<C?`fKA9^f%*xn`f?stSE`pDoF+k9_`FZ-59*kk(MNHZy))0`B%#0aBMJA%xM!F5
zfmWn#KD(6A#fRH$=!!+3GXLaobF*0QYYLzD-ARO&SExp8`Gd5Fi4K%Ak6{=^49qoX
zNhhPPDkWy5Dc<MJD%c-B<y|zM?il9>9YrF~n3-=BZB5ol7#po*xg<Lt{RZZ{76z-m
zn>G?(FUD3H&lk+GeLAXH3+Jwlb{;SB4=oL!=uv*_F<|HIB+XEiJ6L3P!p9y--}F}?
zuIWKuN=B`LzMY7{=kCwei4x$NEIf_<Ou6HS*DSASNSm9M5!0kumrv#`E*U3dZz~2^
zY}KKE;TkGXeY_{)+f(H)RQ)-E*<gn`Shx{buBs%*M$Ujza(&X0)jK6(DEjwwy>hKu
zQ3RHi)y4V555C%v))VdH>}5e~_3ZTM0g4njq?7k*b^Ae=<b=ydx19hqH$X#c%VCSD
zymy1O%SYk$x*6@wOn1OeaVap6{OgB??uV+OoPV=_+mv-+BiMJqj{9%7b^g05_^-v-
z|9rKWN|Lzdiee9sNhL;gNb_9ok<3E`sQ29q?6Z@F+&0vbfAaL1FuC}vZ!;&Dv0BbB
zZIY@qs^L@2*L$lO4xg(%4jraMA6>m;6pedBapfAOejO`_@;i(3{v;qxAViae(y^I~
z-?bmKWWKxHi5;4`x^bvBR?oRYx?Kdokfh+@AN<7(`kyB`bjD?#xXfP_&Qf^hv*wOJ
zPTrNG1Z`kzMW-#+(`lq)GtrVQyavBI{M1~Uz*ceC-)z1w5PLUR1zGAcS&OXo1n=)g
z@GhjjGP0nwo*KZm(OvZ`?<-zMS!((li$32X?KO{CVx|K7+vHQVq7en4cCtpA@psP<
z1Gr<ueXJ}gl(@VobcV-|`!;&ctQ%>21}OC_^YYWA$T({-y}>lR$X~SK@EBl<co`~j
z<<&Rj{%YBa*?fxL6w)gUzuru|zglX~R#*v>gm&uNUk7c}yhsMIN^AU>cns9BGX$?7
z2PQ6?Ap7mg8J@LArrcgt8uoYJ+F>N1HfM05`4c^-^31$)j<Yx>=loY06#DfKbDdL{
z=apX3m`mCB^Q@*=^U}h>QW;~>AAUk&WT5xoUuf$cr3HPAL?(tHTD7qA>jjK%pVV{i
zyU!Fj^FDO-Oepbt9<h=&l5PRPF@rxUvMLU=LDj`$Ss!`35fZ%8kNY9i;vZR`f7p_1
z^3s$~B537b*G;YugRDuz<jS&Ae@|9!SB7wgrPsiZbY=QTrAuCY*owNLsi&Vmd9CJ$
znYjB7mzhdB-+4$!I!&DK38TI|HuTx6yo~n(6A9|L^ZR$=yp}<0nZszNJt9k*MbTgF
z6Md8#d;QlZ?}1Hk;&w(QYKp-7FU3MP=McEAO2(9WH=K);NA#EqBDFwgVpfZiW9v~u
z5}kK$-B%wq1xn!c6l7&z2F&5PmwE?HH6x4n`l3<q1U1L=<K*DahCh^E@P}M#SFLoF
z|2<Wcr$ouYMc5GQnnE&XkCpjg5fCKBah%{tm1dg-Hk>7WfI%czhO|k}AjsBv*-5Lk
zatzqPFu##LW?`AJvDZ<W%q!&LQyDaQs%((QMC^uCGn9YG(8IivKdV1C`!>g1rJ=Zu
z1@Qw(3C!FUSB{KIpii%{mhbnKxuttlnho!k?Xc5R<v#X>?cB08z18-@g?09%Cr_1y
z$t&)5DII-LPot>}ay{~f<tTsHv&gIE)tVM*kFC<>>&mj<DF5y4jGld4G&3{9uIhL)
z7U*i$smHiIS*#eQmp?W&OvS+qHb-*TKgBsDM2C@gd6}{--X71xu2CURvC}n)Hl`&F
zd}Nod0NcV7zfXnwO<4$g1<VE2kVYPbo$RZz4Q4G4@ZIG@R%v(ilM<0NSXc8spAHit
z+XmLqY%H~dp0#<{V8__DVamR8#NBfiVBll7yA+7D$U}(qKiE=JI)RIO;^_0lKlt};
zp{uq(N!9Phxx~g7m&6#EobdAd)=52i&)LCvl*raFbSp}s@1?4sVTR4#$H%ig2BC7@
zjo;$u_`R2}8}QvkXj96$x8<5*B6E$jxENhuR)79n>&DfZqxvgCjZ$)ZW576-1nmkw
zMWsG!wU9ipKR90N%GqXe)vx{>-J;uBrJ}GF*;F*G;vr~R9#r}4Ah69sk{vr%^|cas
z#i^XCvS_;Z9*a(HCK?4>*^Fv}TSN@{$G}Yo+C&nsyaV3od_s)zvYxFF72)YS);Sv8
z9IQ!?Dz&?M=e!;Vyav)@`QN{;@PG4G>#01~+Etz#RAPYX0|KckYCSJfunPKr04}`Z
AJpcdz

literal 7110
zcmb_hc{tQx+n=##9U^2KN(m8SP__&rOR{AzW6c&~Xe^T~UlEBKvW1X6gk)<O#MotD
z#?siA>^tvJzn=HGe$Vq>&viZTA9J1S+~=InxzByypZnb3GhsKi)##{CQ-eStI`!-D
zTObe#00M#EQIQfQT!z-&M8W<|O??$&7YqiIkdTm)l9G{;k&}~CP*6}(Qc_V-oj7rV
znwpx1hK81wmX3~&o}Qk8fdK-6FfuYSF)^Jyd6Jo#nT3Vr)TvXftgLKoZ0zjpr%#{e
z;Nalo<mBSw;^yY&;o*Top}f4je0+Ro&YU@W_AEa?|G9JL&YwRoARutz!UaJ=K_MX_
zVPWBm7cYv4h=_`cUb=KiOiWB%TwFpz;_~Iol9G~AQc^G&Oj=r6Mn*<fR`$x3D_5^x
zm6MZ`mzP&iP*7A<R8ms9cI}$7vN9YFS5Z+>RaI3}Q@eisy1Kf$hK9zC8#gpHHMO*~
zw6(Q$baZasys4|Jd+XLMJv}{reSHH114BbYBO{~Rw{II88=IJzn3|fJnVFfJn_E~|
zSXx@%xpU|4-Mdy+RtN;b+S=O2#>Upx*3Qn(-rnB9!QtM$dybBdPEJnF&d#o`uJ`ZX
zcXM-dcX#*j@Obdxfv2aZmzNh3iS+jN_VMv~`0ybLh4S_F_4D)d_xBG72tcFJj~+b=
z3=Dky_;FBBP;hW?NJvO%Xea;xVPRoUo;(Q;508k5c>45dWMpJiRMfL)&!VHFpFe*d
z6B82~8~ftLi<d87VlbGvxVZTE_*bu9B_t#yCMG5&B_$^(r=+C3e*HQ%H8m|QEj>N`
z&6_tF85x<GnORv`Z{NPn&d$!s$;r*l&CAQn&(AL?C@3r}eE05MQBhHGadAmW$@};3
zOG``3%F4>i%RhYhP*G9w@#DwJ%F3#$s_N?Mnwpy0+S*T_K7IcD`OB9tU%!6EVzG5~
zb@lc24Gj(7zI|(KY;0<3!r^eu&CM+>Ev>DsZEbDs?d=^M9i5$>U0q$@zkl!U?(XU7
z>Fw?9>+9?9?;jW#7#tiN8X6iN9v&GP!Q=6xqoZSEW8>rF6B83Ze*7R12$PeOQ&UsZ
z)6+9EGqba^b8~a^^YaS}3yX`3OG`^XfBsxvUS3&QSzTRSTU%RSU*Fi+*xcOQ+S=OQ
z-rm{S+1=gU+uPgU-#<7wI6OSG=n8cs65z!B>&6}+5DnAO58OHIzXt-bxvImJ^t}@o
zhEI6w+0-;FlNu`77umxcO-){B#J+S9W_f~sFF;PlB0Mgo$npZ*kJZ(ORJ31t%_kWC
z!Hi3lk(rIlCX;03k=0lrhk!KcnJ;LXo9`4I9LD?>8-4IGQV!R&pTimz8kcsju3fhC
z*_}*e2%_i|pyegGP?HS0mI-2vrh+hr{r@z9v{p~kIQ3a!LNu)c7dZ@5u3AjBzP%zF
zB3vpFS%+~G?2TH(1)j3<G>i)AHg3{Js#fdGRCc#YpR%ma28lg$MB5h)XaO)G$BYOB
zBb373kfDVD3sM;=R$EUOaJ;>~7TfM3)SYj~Q*+L>LOgIk5AK9B|9X(j>gD+v<5c@x
zdM3v!yo?aEq415ei{TQNJnsE9j)oP1E{tw(-U5@<8tbaCa3~Y1m?uO9K8$Rtijh)U
zO~h)p;Lt)_P}>J{J^6H@-g#Q9rn^E#Q&R8sCb<ng>ls`!gzwJ#^NZ}d(^=V96;F=7
zz_9+fpq|5>VZf?Vk){{tPRXOGa7Q^)+@K`&DZ+<2l&zCb2!27V?TeLwUt3Nb&b7se
z!n_^X37xgPmXckww&JjQ?K5R`meQW)%umvU)luFBUe9o*Lk`nIcrj^<i9s>%X*i)M
zWwtEh;Ufc!g2bW1%eyd~(W0%5Mvm0WZO{vxtbCONCoWwYo{?Cc|3LEBR#UNI5$m_)
zzaz!D91fUvG+QL-KLj$Q&A53|h2G}M7D(dhZv$7qx!Qc&*^w~4tHFzKwTGA{>LN$v
zvvoeCapDWlUXs%AdAgt{dk+!M@GhipOMt%UA_7&Xn|WlogUCbr_`)S?+0H%wwfI+d
zocGBk;a*k_x;HZFT<F@ZYyv;>8_KMeiN_ToxA2VJk||PAmV0%Q=YX(ta;~p<U;Tny
zM#637n_}y3?j-X!1tGg@juh_yNrapwdg3jw|4QHadCk<7V(Pbn`^xIlEmA_g@60uB
zvv4?sNUQtYGPfa3+eX(zgz_TYX?5|dJ_OzRms-0T_{Ilswr@;JNVZ9jQr`o3gi^~R
zt5#Nqo#Jk-(Fy544=I;2pODwsO>Sw1q!<m*-s{$s=_IrI3bq9@-Grl#vC6+qB8Sr+
zNT)Bo*^kZ-IN_H$$xdmeL>FB>ypYq*9bhwGGSsfq*MFfgwZb*mG-{YTc|VPPyrSc7
z623`)QZm?H{u6D-<w$~VB=0P$IDq~hhu;$8s`da`%dziiKLpaIMQi8n1{DQ2ovBGy
zyH5^i?Ua_TY<Y;al<z#`{>%cEavpr@JZN^9I<cSV{Pf`Lq%0Ci{ptUQr2i=jd#>Pw
zbI$s@W}vj9^}NnDqEsB`Cl>>yhp?kiJudl_Kf9jzu|i4sh{*$@P6nW_^C^<SnZSI|
z=Ts05DE+*`t-03ImTdx5q*w}||3*{kPV&$7h`>vVPGrj$YlK1YCCYzZ1dfI)omei{
zNZxX_L|=b>C+PM`$(us1vbA&o8MP|Vz0Em<(%W)YJPY}2S7v4iKK?W65&k=GP0!;(
zZW5qIQ~-Lax>Q3bhA4YUKbcDd?${)!#2!w3`M8>33v8~~h(UMc6(6u$1$%!Apgwfe
zyoR0@$J%99<#ewClv&)D!D(|4qo2d!Al++dD&Q;$T=6X9(T0K4R-L~2GU)4tuLI+1
zZqh(boYIF>uBymDJ82*b%4W3@Mz$z@#aMReKxkY8Pb2GExtt)hiIzAv-{9&`3RqA<
zh(|oe#AZTH?l9YzlEasX-se?)v1-ExeJJ{u0l%?fbHcfkL+nW5J2D}+<zN}XOwX?l
zM&VzcfXqy>jmM3d3e5?YvqY_aak|P3?Q9dn0{5JCQ%1+)ug}MMdsK@r7oSsnlY#*H
z9ZTynIdybYEQDN=^Inw1;3<wF8B=u4#07<1Fs!y`vW=$P_%_vpz2B>eEFt_kAU3w6
zm6JQgEEzah56WMpbA=Wwhg=z?VO<?W3rIX}JCiIu(eLn$4ze%W00~V*>{be4=yv*P
zN}`#eJC>2Wf_8mqbp`Gu`7c$Iy-6mf^gixZi)6q7h8Gy~oW(PzH#;6+4x2G|CFEqp
zkR<R!90*G)JOx6aV)9{m5BtwZJck?ujX+pWTVTuJ)%Mk32w2xbwZ>7t9C>Be*Eyz}
zXvY=ziP}fRfNP>TGHfI-cmMg~5fFeDAgaY09D=nsb_Kx|V=<3HCyO5p&ON5EDkL@G
z0ChZON8d(9Ev$bgS)Q9_G^r*^#u8W>AgG^&aP;)&e-Y1r`R%V*0ChM66(_fDoY(L}
zhLS=RA8=UBcN&<m*HnH5L!ciS++x^<kw|IKpWq=mZDp&PY)F$0S<?BD9zvRiB}y;`
zzCNY;z@}U)Mh>iN^-t0EpD6Kn-u+$AE=cWkJ$|!h7c(Mv!rxF)DR29|rXw{!z@tq_
z$nyO$95ZG|&Wv@IJLi}IPQ$a@gy5E@diPE#u2ud+g;)x)O!Q_D;SH$cCR%{b>b6W_
z;qMLupXL0boy-Nr%go2D)Ro1~ZCB26OE;b;2SP~TGY|x^OD{MULjqR@r#%41bA7Ky
z>QF=I$p2#h2_UB2;{MJ}V)Xl*03^B)+`_pYi^tt~o>zyZP=)cD^{GT@=A#g>1?+mF
zqBL6=Ebs39UeU%H!XhI3sgJd&o`L+x-)QN{eL-jA^43WU16JoevSXRIi;|3K^}C)P
z?rsk~69L}Fk_4w;BU_3i<xX}7w@pzI>@}su1@5NH(sOh{(By*e10cXZ0nN4^zB6_^
zC|`~*r&Pr+aGgRHlFPSMIZ;XpuhmkVf2XtmuJ&=hTygKca4nf0j%md_&QFEwda+Rd
zXN~wie=_K4bwx9F3D1s*zFy}A-9Ox0r%<dqi|yT6g_YHEL7u)*h}v4{;@9!Li^BY7
zAIz8mz=U{nSjtlYs^S|d+<<}~DR7X2{vf6>9Gfq<H#)<ixOPZ}P4=i6&a!o6Q1ITN
zfb1)P#X5TsqJ=YEu)!(Er~XYbCq2#`&%45!R93FmmE}smoKhiK^1Vb)f*ZOMKw{-e
zc%n_L*2^|Nb-~cn!DLvus|l6I=)RL#t~rsiC;=!*S```gq;f(QW>aDMd+HGbnILM~
zSDssJ!c+6*BIi&3CLEg0ysgt%%hVJlOidTo@tc(3qPbHnRz9y<%$uAJ9@7xvn#r&&
zMsyJ$6n2yoD!&gSc8@(egO}t%IrltQZsnpTlDQDQ6@3OheRu-czM4h<%fYvGb(gm}
z{J7m`k5@?g+spUXXDfplA%t$W@l@$7dVY}GD$2wal!ot}UIMBM5*NRxl0h?nZjNx!
z0RBV+c7KPpa%q6lK1>FX!fV@VQ)<9iw7OzM)Tj!Vl>hI;D$o{eIW#=;$<j`O{gSOI
zFt+9J<isS(9*3hW*0q~qH*xDi5-l-0ZAr#E<FfFHo(spCW<lxO#+TAidKranryJLn
zUI}rtFSk=n5f70C-SAOC@0A}7?CsX+q1i5NZ-!@q#O}E7jbErdN@}>}@T_@?qijYh
znp{l0Y167`tGt^mj?BKY#Hs_f`h?}}`60XA`i5A>LkRo_aNoVCjeuKO)yG}vhU2hQ
zOKV$u8OJZr6?$tQ8&s5TTynPXg*MNTbw%l{fGvXA5$K;}bs4~9<Fm{{Tn#}Tp$?EU
zYrhZcZ7y+ix&J-M3_?%}O?aCve#?}rCg8Mk-gC_;gWQrr9qxQk#~cI(4Ey@6*#9B?
zuNvWRF#l&diw?Nv8&r$b216+{4^AxC7hWuoAqMWRoN1{n+(dY>Gw(#y&W}!azGhh@
zB}wbJ79n1xf~E?yT#(^FHpF(40#r8Vp^F9$|5+ced?NoI(4=)QqNm@AeJwZaF>n#$
zw)=I{MDnWYJBII&&CwlM#+_CSm0#8ui}-8M5NpSbiD-H<X!A=#QR7%3YO4Fu7vcBy
z$@-bnmCT1OCiEm&G;VYiINy5tqg&0bx%E2k_~>Ym*r+?L+c%lA+DiBS*D3k1(NuN_
zfxwpL9zqwlHqkM2CJzQWaz6z7Q+D0;iEa<akvnZb<KUMb=%VVzUws1GIGtPSn2OVl
z@8Q2^{HUDtK$VxY^-3Aj9eH*ScIkKNvPWZeF+i}oSQJWOazO5OYw3}R5y<K_!4h&l
zk6@HfM-=}u<nQ_a$ozjQx_|4fL-Ysni#oYPjC%**<&5CHd;FwUuZeN{zw5z&y&4XE
zIsB7*>0NKDU^Xb43Tu!ZQ|c1X<7oFnAAOyC{7#LpXXNqNzl@B_yoSSPM@IbUi1i_u
zM+;C3-!d?k_0Q@_N3%_Lgh*es%QIwxQar@+YLDU)Kw|Bre{Sz%JS<cs!5xhpLD*re
z(bs9mW7il>jJ5E=60grtEN`AE4WSUlHlEyGDcU5Ym^U##j$04w!J(xWSPrQ6AMeFz
zid#YI42($u(!wc*4z3zZz{?910G#*KRnc>sNqZ%lVt-Y0RFE*`=GqBxzQc-+6!D<a
z*hm^9hm1yr{x>A>Sx9gA@Q)OyTa1skc4)bw1CQg*wm#bo2&_T!krasr>=}~-8_<w2
ziSMQ(>gQP)fqmgZYN%AOfY;Lbr8hCuf7poq6cX2jzmVOquu|~CZiNwBIJK(LwlavG
zK8LLy4{B9#sIZ8rLA;ymZMw2eH6Dns&3SPsO04#VgFfU1rL=z$749V+&oF%#m-2cL
z+#v}V^uBT6f4MS<)7YMj*zonU=CWh}%B~1<D^iF~o~MGC2YFXL!|$cU@0HOPt=_Ad
zH~-oj-XYYv<rDCb#8O_?e6Ra$=lYx|OjLPqj_QEwc|}|bRB??S+xBr~Ro)hAIdY%J
zl73)vXC0Fu1{__~4^E6*yUDVy+Gx-p7A$Jua!CL$UfZ81UtWZ^&U|uhxUCd1-`G0`
z*<B%&Ahz7G628WK(8#&Z$hIFa(f1RuuIE_yj8)7)Bd2m+azf{|6Y9z4j@6&fVOMOf
z=EnC5PvxFBQPSM&n=}B!D<O4@JSTvts3{z;A>=F4cz%&;{9Rj}(FDdpox65>&%VYC
zDzfh4=4mH9hC}AP1#+vJgE27a5NX@3o~DmY&UF)cu)ZMg%@~?eS>u9()GL;L#Pk@_
z_sTMJ+Y2hc55aTGhl}c6EL_?2@m=qmifAus{#~cvu_zHV%GJw$TAk7~$uY6T2MZPz
zFPy+dF8ju}M~foFx5RJFbJww&*A>UGWtyXkP^`G&^liQyL)uRXIPBS)2(6QqcBb>m
z7rk^UUy8;IEE7DFuxE>fz1p6RhKq>#_RA{`skq_ODovWrQ_NJ~6U-x7kutJh-I2MK
ziNcHS*C=bMHad|NG15h0<Fw|vcmihv(pW)RJ#0J|=$N?J_RJsCg*Dec6{9{YJSn_b
z?EfOB&JPkZ^KsxLq-0yD8b**-kiMqsL0iNppZ>vG-*8Aow3&k3`5%d}dTJ^#P#Tu3
z2ET5B2CkC<0Q{euVqW?sQ)p3@h`5@N{dv##B0l13bG^Eg_T*ut){#9eJ++ffT2k0b
z1{|mJgj0OFSVbWkHK`Y$zCR@rpI(%3&pL8rQAk{^gdO^k^XNOvoE~CB0f<ln;-qj4
zDLfvWh6ATHTPs&>0?i%rD8otnJjGdi4;YSfVOmaH4WXiCVf8~h7t_euVq45>sLr+U
z#W?$GW71GR&8^UwD>4=7NebsZrA@~y4I%&bh3uOjFzCToZwNb>pJ<k<MoM^D-yF1{
zB+!+M5OPfY-2&blF?|W$8}hD>=)f}D&^7e+tFNfWcyeLy7noBvRL~X4A3UDD>%rmY
z^<f=Pd^<+4N*MOY&+OHv3~(%Ny$Fzzv5FW>UkE!`{NX%tIv*qUDlRQ=>^ryp1?S7g
zy-KuG1D=<j4MK3@Y=f^lr%R9&;$AuZjV%=s-CWP5%v5L*j*0#V?78&(l%_t^{vQA~
zQ~aP0u<=I|(h52^6*0t%i>Lj(Cdm*9gCzuI)$??_x&!yK(hnbaPt%sG=_xoZ*1qp}
z$etc{q3Lwwe(b^1Rmj5Ab91u5m{}z~g*%6pL)gw0?d;+$C%<M{!<@W=SN_%S$NF#@
zxNj3os)KPXmAOaDXY^t(57i|o?DYEpUQ0Re=RPgv-st_T=%Q>Uud!dv<X+Tam!+EF
zV&mi>g-}b7*;k%hpeyI!s2uR=UyRD!aB-h7Wl}w<7b{tTsh41d4sd7<c#eLp@ZzHV
zA%)fHHg>+@vev(y%#dQ_>iq4Nj-p4_A?$heSn>fE!k{H-KfllVJvokTce(gu$h!D3
z(9m#dYJTiZyMUqE>JDpjO0h&oqC$%Sy}rYlJLCNkFH)FO@>}zD5mV_fv$zh+f%_bM
zwO6kU&bW%cirDZUSj3;(5G%w;{PK?3*2#_$`{IJfY^|kpcc7aoYD4i!m&#riz2v~!
zMP10$E_#v&@UdFJ?zn!y41_ZD&28A{z5m*+LfJA$h{)8@?fGFAXpg+C-mAnzL>QIc
z!%_1wLiva#&41UVbv!wZe>*)n^x4v;F;N3b0jSc5`L$bK-Wl|oqNjkXfzv*R2+myD
zk&DKi;tyFLYVh(xpdLm=HVwXU9f^?}owW#P3*Q$#w{a)Mrhdrn8r~`JgDn4e86wZ$
z7mke?KXZ=?^7Pb~ntS{BDTcBk^VLyry0n!qb#^Q3`KZ{^?%sx^t;}xYqK4(47uUN&
zW)lp6-ND7Tes?aLCYf(5CoFBN;fEMsfzvwl0cX*71D1VZVd<9vqeNSL=A7v!*33E}
z4tsk~si?s*IAXtHFI`?Aps5$zxDg(naPrXYk=OyFBpINp$D1i;2$QTq1<)6rMC_i1
zW{OL#8rL&1Jdv|id6Ub<Lfp4YQ(^oKVL805%=Jpq%Yd~Ou2gpDX|3S5#k(G@ZX2gU
z<L+^!^&uMfa|Yb~N@jGhO!bkLZe#5_POd$uy>!<_suoa(JumTbGu-uRfZ$=-4JBq2
z=aZc(ugvXGo@~gW&pO7FVfgdGZSOb5g9(3De4X%mJFm?GpDK*l9%X{manFBk9V4IQ
zNpsmL@G;Q%kzZS!V>$qr({E7NnT%|wJ$j$-sNRr&E7@Jz@{RMqh=P3uEonC}?`X(8
R@&5>rx{5Zu;M$$Q{{cISw#xtj

diff --git a/docs/dev/articles/examples/timeseries/timeseries_anomaly_detection/unnamed-chunk-12-1.png b/docs/dev/articles/examples/timeseries/timeseries_anomaly_detection/unnamed-chunk-12-1.png
index 23e6cc392c1c6ba3e07b2abbc422dc6eecae759e..8be126c92787d1720d1c54e0c595c8c7a7063f12 100644
GIT binary patch
literal 33046
zcmeFYRa9JEur<1Ig1cLQ;7;RC2<`!bI|K>tPSD`)7J??Y2ZsP5xVw9BX<YxE@0@?!
z=lgaaZjZrecFXRyYL(2YniZj{EQ5|hf&u^lx}2<(Ism|a0RWh4WLWT(q64^U@adJK
ztd0u+pyK@dgPC0Me*^$3Ku$_R!!!G6#lv{|>PhTu?&Hd;M|0~@4i;4)1F0HzQRMp9
z$b!hoSvvzwp>o}nMgJ$AibY-9zC{}YE9Rf7PE>EDzm~(pzJ9fxIJEMnY479X^X+!5
z0$W;p;M`W;uBgY&oiEP5$mvyX2o-W5mLzh1&1bS>QgP(hA>&R&!~dR$g1V!@H)3A@
z|Cj$Cl*Xmin)nNU&S+KVvv>U|EVx3(7}|k13aK1u++_4(-9cVY_gAF6pVE13XRX%$
zyjl+>G}P|!6_7J6K$b*Cui;zuG#gB2?Cb06ak{`|NrzdlpHgnUKPZ!VU{eOlv?^Nm
zi?gimdmmIlB`{l_t<VcU-PTrq9Rgk<BXUb3zlMYvbI|<1xx^&xe%hzud-md6iffXZ
zdxD()GHHDsA~=71J<$<r58>g7xn`w;&>3@J=KZNTUOVx4Rp3YZcY4Q>|M{vaz)fI&
zcDB)cAW^M3FL%YMpBy9H4BUDCM0mF_&~`j!#F`oZeU6yL0($x~riKBCpPl&Lg)Q`H
zx!CXAiTRYwcJ2%e1+<>eFY%28T$XuMjcVV!g9b=#TUqyfdo13PL<LFC%*^b#-Hc*G
z;<H<X@Iu+~P(ph)COV=_rdL-xK91#1bfCJzb{U_Dxi3L)I?W84r#9?Ypf8bg!uNyr
z1YJ!hdkQbeH~yMyZ4W6KdI8r^%*cSNRfpYe{kwzv{yq1mC)p-?2`kx9$=6kLOFQeu
zN?F1bW2MU~W0Lu*40t5RtwhqAnwsCyiI9s(&dD)C(&skl(1;H<nlBD!%8dN)cDl1S
zE~iui?)OW4yb(n1Hm>w62;hjF8)X77z1m=<c-voU?tWLXF~_TMKs!1+?#Y*iKIEP#
z2aS)Tgi4C$RSk7l=rzTX^05&_p-;uS!4Y7D%X~GZ7-siBZkT(9=Dm=e6}1~WP+<|B
zElqqkn-9)dfnHQNqz61a4Ash>{O#{-d7j<twA*1E6n>c9d%nJ$#2LUcvfrO9Qb=PX
z-%N6UErl6y*5$|UZ&pxl(I-Wuj1*|B5G7{okAVP)|D`jihZq0r<g4TX%E77#(FH=X
z!_P)fUs6#qVp<bD8{=kc&a{g}>3jAIm0kF>4T^ZS`s2rsZvDl>E5zXL?rfMG>acc3
zJ;xJ+&T~g`p!@Emoi9NB;;8c4nAY*KM*;KcF-(k{A6rPP8IFCWX0qLP<oT@+Z-w7;
zq$CO;ZV&mRo2BOo0Qr7HKg9eNyXZURu;2lp`_%bei0l8ZNmFxX`ulry#NeSFl0{oD
z;oM*(OpIMLn)jTPLhhfg4rbtds`|Ggq^RLKAX>wvanhA>86wqwYrRVpPfpF5u#29*
z?Zk{G?~mzjUd{V6OZ~!*cs<2{F#Ej1e0qZNTN|d(g-xVt<Mic5?!DrfX8R>nW>&}?
zJ`gKL3V~``ci~w{%oD!5?l`p5pH$8VCnQYEh7x+S79jR<l&oL^v;*Gyx%<Bbafl01
z5U$|PY~{U^l~%9Z)2AXvU(q$=lkq*MNniB6DQ%F>m!IUT3_ASzEXKAGZe&n0+em9S
zE>rkQvEILfdC%9I!5&HsKhph=eVXU=zx$ldhIhp7v->}{<FEGbZl@1bJzFlK2DB|o
z+Vtm{JX_se8a;AAM`cvj91XtrrMH&b{90HrZ%LApdHC>_KEsiJxyW?2-a3hB_^lzQ
z!}8A>qrFeIwt6+D?=BJ?`GsMX7KF#faz%Wmij|)H3aF6%I|Denqw%>93QSJT=%Tkq
zjh~&}7V`Al<Vmidj`|KenyaesXY%x~%KQjB+r<Tz(<m{_uM@@mo3`W5!<$ywo~ulD
z{T%E%+4tQpv*|tN^WcF%?zLq%T*~us*%U6(qr!)zwn@+D>Qy1Zq7cc~oonpKuoAAm
z2=2}L1W|p)9B5vBZMJi4Ndwb2bCd$kg*7#tcYoHIcdZ~y374YwJCh3RPwT;$yQ4Wv
zRY^#&Fu8k3#KYR01sz1!kr}zJdp>&CGeT{ZG@4m_{?Fl2!?b~4Gw#Y%I(&A?t{jd0
zSY8+PzbDR5z(NUSh?K9G#KEsyJEcq{Q6Bc`TwA+of<Pdy&6iZnyKqPt%if0-BAvzU
z?FLnIW0^d*L#eD!BAxTsf%x1cFf_CMin5T5qW8PQOj9`G9}Lgy`tBF2nkdO{`B6fY
zZG5hetQ{OEx8*=bA}uZbElURrK9LrNN{zGg`Th{Jqfx3VSb&tzKAZcaa_8#gn{=9l
zKU9YkPY*XGqv(g<nwtf<xszz+>K2yQ#7@_P7xthr4U9HcpGYbYq;G%UoOE6!sQ74C
zJA>N4S9IIQ?$24RMv9!HO%d>TPU(MZ5>NjsRMN6e0Q-Y|lU~V`gnHld`1rbeE0PyJ
zF#prZEI==el$10nSO5>@DUx;}c}MmBr0X+WJHfIk2b{QsgoK};U+1V8ZFjl)cf?0`
z*8`$+1ZhE4d`vrM(djAXmYf9N)Ak#TEv45GUPrPf|EK#Qw$@Mad;2)U6%`f4#Kb^+
zI-91EQI!Tpw`|<;Qq#+IQ)dxFq=ioA$yAB_+qc1z`Le=j;XQk*)6WtRoJ;0(C=^Oc
zLZU2mT;Gwmwg=6TNc`UYI2Q2i(jA1bY3=hH+BR?cS#|8pXuxWnbggYGbq>=7Yw5Ce
z$p22~X2hZUrTu1(o#bI$xQnmDtDp6)Z@86)-BP1cg%+m&dOgu_MDeHao(b&Xm>AQy
z_HW<ztf%d{HtR!_6pGY?yT@;?HyRsxgS+E>%SbYJAEp#m-ud~73|-e&SBzMa5~2mo
z2~|hl&71^zt+cqYx1C8gIo$251T2n^$NSQ9qG0Az7<0rSz8=?ZXkTRqtXZ-+S10X0
zzrJnkxDJ1zIO)M*k^EX?^Psv$O`>Gvd)`-TF@#7t%$0U{Q6Phh?;^wDu%^7?Nz?DN
znb3^{J@vn76*_Ih1)`-3ruN|j@9*zBTm{~gXz-J1CD4j}sZk^AzasG&85!B?d?uMK
zmxFhq#ra-Yc`|h8Cz=;Q^^MVpJg9qpvB9>LbvBMd#QXB^#5N6A#?kg_DI9tPHuA2j
z?MgS?ih$>^3w7n=oBFnZHS&PFvz^v7R-M|T>k_fbI9Q;1>18b?F5tSeiL&K>)k7El
z%WS|!xKI@sl*>-<ruZ6T9nL>N@ApnR?k9&Fytng&RyIorPDR`NI&%;JEMd);+npG(
zj>`j~50#Y|)tX}~k4Lk+FUv1mFH0nchYRBan@bLz?MJo$MB&`jI=!_|G92*CeftDp
zp&=$6?td0=Kbo&JXeD4bM9WuAkfL6Nirg&MxgE|@tl}fXC(Bzn%Y(rip(~JicXn0_
zw3akut%io~`zzD;yuhc$`u&^E7e6Dm4d@*t2utCP0v8Rjs`aGB-lN2^S>K?sLyG~A
zpM&(u<~BG(=z0NnXjic`ox|;VAt%S<G=PkL-T8y$?fiAXvd5gw@uZKxdxZVU$ll6L
z3T2>!Q1!-DcmUMyWM*!3YH6+A|M^Ksp|a)l<Xz%TRr7kn$m_r}%uXH8nI*?nyC&Ae
zis!2+pFW6v*+PBE8#<u%G$V@5U%+v;;>Iwgb>|uN)X2ya`tsx&c_$Os;rxL2!DuW@
z8O5+F?OSPETU$30H!+$GKbt^$ojF%&RrsT?Iu;2jDZO$we|ltdP0e}7!->!3gbWNY
zTQ1-+mfM?N3xd&>a6VoLWP5LqIb~2vKD5Uql&VX-<c@;h@hf=GxwMAHMtR_Cv2@j3
zRK<yq#|d89qlSIc!w`F?qVT0xyL{}HI--Ag&xW3TbJ3EqyI>A0E9>Ui<}R^(5l#tL
zfg$?KI{S~y7EL|6-O+oehdLxfexLb06yFC{-~0WqqpJF7d~3^41^9IDwV^m8HN{^G
zp{#3Tw?m8cW9?|->mOIHj&e}9eV_84A6K&1jypfrAQ~(G&Sxilae;0n!3FrowD_&m
z@hN`WXmhGO?e{oHj<hql`f+%`_v~<o_G)-6>edMJen&7RB+(kpdCBPBhu2kdsor`j
z<72J>?U2s<{CF+`g%dM2Tiak#`@ja<3Je)HH#a%CfsLMUB=-wY70MbK+|D=qKIVGS
z0bQxAE!sx`Y=i(7e2qy@n9<`V;r{B<{JeI03Xj)Vt~ix3gan2P4`skJm*}tXzbFw6
z`tYsaF#||>-Ur{qDgD#G#a2m;xsXYy#phw3uX*o{=IlR@B4LnuKHu#TvS?F@MmIJj
z1jUmK2T_&I*d)@v?+0UFr4+}5uV{mA3>bT(5{Oe+bLzJbTzIXdP3L<;UUB?+^-9p7
ziv6`x65U=R0GOXlGiC<NP;3E#Og|V>U;&?>_wfN+gHdB%6hL4DpP;;*gp6J=$Nq&g
z2@uGkBEVWT)rjBSgLG^$#6$~zzdJ6^^XPV01CF0mFrSB)mw<p^(7d+$D2*!dx9Mb|
z9}5{q`1cNqTyK;{2lmT{vL8Q~6cpYt=5&+ww}av9<@u(Q(sTVY<<no8edP|@xk_|W
zUb28Tzg9sUr)GV_Dn5Gp{==*5YK#w{n}yFP=dfmSVZcjt&gK}msVy$v9gJeno%A}G
z`=ME8-+J7*{o1m{3hx5CvFx%@R#mTC(Z6MFWhLv8WLV_^$}b|kd<X&1&rA(yH#~K3
zMMvy;etMXz(DN7*xsjsAZa2(Bwyc^f(=2mC3kKykgRGSHZOn81^kej&6h*qfV=aMj
zH1KV!u_sQWr>OE!a{z#+Nx%ZcWi28%%kl#DR%>H_($X?1Z7aQ--g*Tm3nPvk-XnPt
z4#S$HmM<;~pu=F<ts3kM<y0Pfx~z@zE_ZT~aDs;5vOE0w<w*rM3jIO@B^S);w~Z}j
zam=}@wVg>wX0<d|u+S*1?T#-hN3!?=?sn4BjQqXq?Ce^MMct3Im6b=&`xB5kqK8jC
z1anNe8}!-M8twF2G-W`m9-H3BV>y(CScFTW4JE>n6S=K-eU212c5-nseVVX>=o%TV
zt*+)GTjq^@kW!0JLxS%01Y;85;c0yMFtl^wY82RJ=<xo1(uzIdP5<`dBDmM@9;iVQ
zU!?oKZy*NeS615h$C3l<u7Z2XDGk3L(m}y$bOu~(^n7NnB$_|{qD;=u<U<9&mgZ2|
zcn;`KJ(P)l7|UJxZfe0amc80?GKClcE~DCLh36?Upgz!lcR4ZU<#sQhA#nWpDvQ!B
z7wJu&|7^0sX{Qk(9G1RqK*#NKT(&T9Cd`hu{^@VXYyv!cJH*Mcrxp+&bxIyfvl@(*
zq9(Vi(6XhII9hI=B18M}zL<!IhX<5h73iwM8(p2!er&~+{Qxs#qD;_auCA^S5D@mV
zU6e6bhSAZ{btX!KEWkiIefS36q^tYr+%wtDO+HX+aCT<Ki7c(nH*&<1Ic?}QAK}Ht
z`rnDi$HxZSIV~-%NeJs`-93mYaS+%5&-T^Y+FGdzmY*xI5``Eltv_n8R|CX2saG}}
zA$)vrfkQhjo%s?;)5$q7Urlw&)77dt(vgB7c%Fu$4)@K_lHBvI8e1c8N3FeLj^jQR
zQ9iekH%XS+0nharm_q(77x%u2KYnhh+(1>vGMs0FbX@##`M2to+mFqkV*-2?@0NAm
z@3CidZ{;qQ<!P%)GDmyAEXp`M=skUV&PC4|*VWZ6*J9TApvTWfLOz{p<{S$bXeFGj
z!Ll^*-mJ#_tZisVU64#y;;%D))<|kjjvZG<qRhK<;g_SjF**u&g9F<kf^?p!1X+Ey
zG;PkY9HE<s(@qr+GZ05CHuCB~GLbDu4|l9H_wA^i1!?7et+t~`$SUdydY*k%E>XbP
z1s|}9A&<U5V0N~X;0oyP_$>yLdj+k^K2#|izFj)J%JsWm=+~=jUrpCSWyG%w+A^MD
zN*bWe?cH#k|HKP83g-N_(x7}<@@_W_reYAlcmPExZM##2OPZenohOXjBNyGCN=yA%
z#sC->i=3^#yeI9mcl}|!KeR-+h(Tuc+J$M5)fpB*J}wWRn_1Ov?fg9SorUJ$I^=@U
z*iyJ&@I%RKU^~hF!T7Y6hDRpo_hcj)Is%aPT=B!I$v}^RF@5=ALHfMvb6D!o4i`y`
z<GOh4!t(5_WrHq~FOxvOgnQ%oDh3clMP$lZtqlMy1T;gtJ3H9u=-!E!p2C~IXsByU
zPilDz$M$?~Ry<BytIS^NS6V$jXlYS&PVSC4_TeY%pE%gsT3MxYnuY0BoS2YTYO_dB
z&-EutThtjkXcQ!%TXnC<(o(0$=NC&$V&$vKR?WeQ+u0q1aVwEf*xKGc_fIWo)PSn4
z?lAP>=2Yyseu~#_k=Fi~N!`MavNHDOZJhHf!>XI|y3zZy9?ZUQSE{az5rf|`s6k)X
z`BlqM`IR4C1ag#}B9o`~vq?~h@la~qRt~QSoU?@qdLoIhP=dX_TDhI{7G6Cf`z@gP
zT|kqWY#;Y^a@O7sj_j0V!2rf>w}x-;hI{|uN!l=E`ub+0U=1!a);T}8@Vatg!GCsa
zrlRQ?gtDdC#&;{?;mHO`0B4@U%{X#09C{UtstjUclm`c{cr$O(?RfP-Gy7(zwSDf1
zqqIp8CsU;UCe9r(3%$1PRg&D&IF;5gAf9M`5t}qnGHKJW?)<J{(YVbb{pP_;w<1}n
zqy94$GFiV{YZv_(RCtU&*T&YCn*?oqS}NY$xlh>d{&G8V!M-UEADK2tU}-C1XlJLW
z$O|R(Hmk9*k&c~r^U7O6K>-&xA%$f~kHsKJfDA(c8P>kp;-aj|AL2sSOCN<4EKxG`
z{nERRm7pdMRAL)bP8~>OceHn?(!!H;DLremb0hwXv(@eNbl2k@ag8=+dmE+SWdhYD
z*I$)Ur}(6T$UdNJ!*N@oi0XISLOB}zU|rp7Aw6bfgJQ0wWB0lRmIPT-xq>m!EpJ+6
z@j2x3*lP9T#>6-dyrI&{ASPbaW~n(V0hPtQwFE4BIc@!h0je~lw>qizpf!R9&V7&9
zOAbaZBRnh5hG4jT8c^|1^aKkJTEiUiX)<aW)ai!@2gS-cnfjbf_4TEuoLK!qDXBj)
zvm83^EWIDD`$+jByFYOdB{qGz_HHMMktS<H2jo-nSqN%fxWze$XytH5g~oK>Cs_^V
zLaY-<ETIs-BtQU-=2{QLC7QI<=;1Ytq*o|gMQg}gj}H_NPMpfBlEGLJQY{7xZ?lh4
zcz8JMGS#tqM~GXmrr)bX1V?-S3{l6T;``87OI67xsYZ@pC-p5=KJVu~U1_<-g>RbB
zhF9)421h|nIgt(!F{p8V03TXfzXnqrXLTI=$<O}KoOQtg09_I<!IoF9=Poh7D?`l}
z8A8w?@)Gkfz1zQP)twuuo;`R4OJw2=gE>{`(y?KyyWIh?KHD5vB7J!}4T!>AT3Q;S
z#!lx?pdFABA&HP;Ay7z*)JdFotEtDcvCRv3_L{GGOBxkX*XAmiLxmhsC1lE>lEL-v
z$WqIhEFovb!c&)nsE`3gHgS5t&0Pq%_3Z#d1n3B@%*+bQhta4I>*fwqkb>cr%=$ou
z>5d*}n14>|EI+j;&x%bjLc-|!=>$=_QRNgOi=~l?y>yIFWhlgh|1+T0pVr5VDk;MO
zhXrqP=AB<1Pd9M#VMo8i*Kn`+NSunQvU;V)Z@Zter)lk1CXeI&J~=yA1{bk3TUrl&
zUf0u&2eD6zs)2!sTt(O~Ny)OEEp{wem~JaY`@zcZ-@QQR^be9w*Dh_Hm+LoQ6r~y3
z*x4Zjf<__PNy)@K;Q8L23QP67hCv7MPdi@eB~DaqF0QT(008Adj3JbxTmvJ<$H!OG
z($Lbpa5?=xB_+k@<^%zFGnBZnpt0AiNsiAAq$JTmcp})ddZD~oN>=s+bV^*c<?}=*
zSKdW6M5P*FWya8ig9d`~ch}O*L&4q+-bD#LIc>_hS;!*MNzzGSD50f=f(;uZ+yXFe
z__2jw$4L+@701&fZpVs;Hx9`G^|_&8|H}LJG+<+%lUw{-8`Qf!i0TXRzI*G4zZf2Y
zdwbzP@&L!(-CY+{RRaQ|`K8F_mX?-uT2P?l`#9Zb-Bbw)308xZ6%`<6o3Uw#B49nJ
z%_Rgpg(;%?YQYLNSR;tI$)1@-Oi3|sV{8QlR9IMe){O)wr-bpNk(cQ%_Z2KW%)x<G
zgcRB?H$X>-|E47)g#r_L(rVGy7mh)0s-ramb+u<(d6NzYllTn~*VUG7YJ}~Auk!ef
zDF%kU+6GNQui8iuo^ZcY8{iR<+y{&G5R?$rFDI3m^-ITEW%$Ar<2DFoJekuw87z1x
z712`5a);_}4<w1SplgDG50(h;$wv<dZ;(Dx?PhDkvGl1CHVVQ2#mDnRfLV_2cYv2L
zMVLwNhn5w4=xRGhfd8CArlMsc%g88VC@h7~r`6et6~wy!Ew%Y$Rc+aQ*La56V2R03
zsm|H+iH|cMjB<QVngL_34D}&<5KGV4uo6URgPBT$wFt$Z5&Xi*(tbFgp`pQ|sHpa|
zGrZ7&H&wRKY&(rXh}6bbf{pN_*TSd9y6fYt9%zKIR-b8Ct%dtT{@d_J|4xFy(UBq{
z6;{-yCMXv%fF&UCTb7m#V-F>$d;gQY{VZazt9>Qy_qdF24C>#bF}#J?T_<~trLFLg
z^W)1~KdXVKrh^Fg#~cvf*^R6y_)dl)gVDp`_sX(;RUAt+Z_Ki8z>IU;N-(!kdp(@;
z>(@Y2A_w&g5IW{`Lmf|=Rn<*#LZtLs<_<?jM+r#VSL~ZCAFl0#7vMPz>4LvekJ~Ut
z(DJ;0n;);w7EUAcwK|@cWZ;(aGjeEYG#YW>&$a3MtHVCylSUDJvbh}O-Ro<2U22%1
z^&MxY$?NW5BzkK|NIy+X3Dp>M*9xz;E-tUA9&a^(l1{~v7AI}Wx^BTWK~|1G8`+X)
z=-12#X3>Y?J?NdUXbKfpMuq{M>pKHBEjA?J&mfmHHAXm192FjK)`%syRDp4WJS`4q
z`V$2vYYL7oE1LHT(f=cuZ>@AAsi~>3siBkVKs*6~dJZ4Dq9Z?<ZV<=Sd@^V@5565>
z)>fqM^=Ms&f8_q>4^{o5mAaAh`}glf0Xzap>9EhBI)bhXgnszJ3#U*{UQ#?fSJ0X?
zGBTRKmLf++0zk5Sz@((^w^+Nc2Z#Az#Z9>q&iB<d_=o?x;>m^enB?w(vHkjY4Ct*R
zBqj#_q!O6f*-@v%xeFD|zy^L*t){>Pz;@E(wR^Sp<@x6wwlwB&lUP0?K-`+K?TI(r
zU(kXB*}iGS@_ndeacb(@aEyqtG97|ZG0H}V62>3x?KyOEs$~}XCG9q5Y=rF3nsv2`
z6Qw;He|Gt~=;<Ag(uszn4Nn^ufWY%R*`$ccLrxG(Dz|wJE0@$&_K(rf9+-F`LGC4A
zgRTbx#zk}D7!H47;7?T4&I%6#oVd^1U=hA1RkDucYn0bC?>;M+?A)DpzW8s3yz1=^
z?$2w~?&YbL4St>KN|80<-uQDWSdotYi&Wnc)SEpL6L;w9&Ui4{p3RzzlLC`W2(hmZ
z0Hn<zZ_>-k%9xp$jOW-?ncmRh;#0nQ7uFN4#T>hO@$dxZEN7!a<CBR34yzuF_%3Q@
z1=p3WgS(@4s9>$!!Jdwr8wQlSQ_`!r{sLR8K<Ma{!&2&PkEJ?Z`bhtnx-N>(b=Q7L
zPm5dTu4UGUbeV>^nVD&1dZdL!Rk-!ksHm^!T8AGu0}39B^`5ci(h`ExTy{AN^B5)g
zjxI!=MZfgn(FJ!XXhz<`V&EF|=SnPy(liySSV$O!PYz*NK=zmJ$3@$!D`6u12oO>%
zTE9Yuq+%s}iM4n?w`OH+&8dpF?2HOvEiC+oJTTv4QpW05)K3wkh7#2<^1S0GqGdpF
z6~zCqP|(?YSh<byp<1^uF^!j`xiI&rxR@?@jNX#Bha)0VVAD}wN3V5Y3noy$ig0)2
za5oX>l~jS_ZT0-1l7)FSWAGdZb_tcU5OrArSw%YQN5bT4mWqjZT4IcF5ek)5E!2^b
zvpq3Fk;`}?i{uP0GP+!U-3~98*-8TdsHjNvf&zhGkm^AH`cuF9Q-=Q(-8rj!>&hV1
zfZ4KwATp}kYP$AQOwa3tfj}w<mkt#c6|y8I;4PBvcU}HAeD-)*Ko4zqRB0i53Eiea
zdt71a|8fCtZy?Y|4;~&DMwH)iW6x)AStSCc;RvF@tXh?U7Ll7VMINbzf?v=r+p}BP
z?O?hS?3fn#5hA0bU`PCcX|uD_E#SQSIkuMs;>kie0MOAXhIErzf1)kdDw(o=IVj7H
zVmr9znYeaWrh|+8_R17&(tJ!!mAs%fy}MJ@c?ICq5xbu%lQk}UV(O%Hc7X}BB>LEv
zPN2`mGZcwm$3}usBdPQTS@i=wuLZYO*;(g{o0tE~^V3aQTJTTN2<d_^Dk-VXWDF^$
zeMipCmOu3!yAe3NxC%6W93Hl27t(CgsmxaCzyUwR14jp80m-6GKhg8CKh-SJq#rlE
zYCg0U6%|nv9BObLUY!kdi(`RRG0)sr@KoHmTwKMT!XOl=(DHzH@$xcZ(bCj>HwD3x
zN*fhh-P%XC%uH3E{(N8~%@kT&`wDHU=A>!Cccyty^x_YVZYn|nwyZkm_*K`$cesai
zY!dX1>fquf8{1a0R2Q;8=2UFF85!2m(n74KJC0ow1wa=M52t`YM<mMH{ndVDSIbJ9
zBn$viV%M$zVr_e<!LFszHKhUv*g?4DxNX}F*+-nHz-pN_N64maI@a-Vq~rYIl`~n%
z4!`W;_D|+cqh(x!)D#3p)WZ2=_v6)esuRT&ykV|fq)^H7^c>~A?VfOxeRbfg<=AYI
zYkgEVnvM?f1Oq@j3((Klbe|#t77!F~5p5Prz{&jQihY_I9!E}-PxC$<02FX4FwMvm
z7}s?5M&QDH;?0^noVgJGd!?$My>U)9-~9e>rG?~~_rUG4RhWss&kc$eb8<?GGa0Kr
z^i$0+GrNP8l`8mkNRV{F(K1o2DifMzh%^jwObZ^0!`=i=5LNWeo$>GJ*nX}IExIK<
zhiC*~V@l<&^MxGfvQOLDOX9Y~Iv1VR0E2VcjK``M{rvIxaqW^|4ih62GVrCTa!N;s
zGfU}F2OC2nNFs0U@X9*|9>&}6AOiq2n7#xg#amj)3C0OyA_k*`!ZzB6?oYu0^@~#S
z1N>~*!}hHPq{v^Og>M!XYLjyagn|TwmWFF5tTgB?nSi2+FRfzu4^N<irH}l+9gTBu
zHK0xqZ;F;iRZ6ph81ufhu#f|-rcNJAJRkV3A)%(PmlaPmo;$o((2L#*z&GgX>M0Nq
z1kGfgLY>ZI+Pdrm&ie>ICMSRSuUCl=@*fHUqosUxD>_|h?DX`OTvJ0A>TqQ)E-w6^
z4vpv|cURVY-$T#)0r6J-gG8L3a-GSrFxRy8cEcKHO>4;17(bk@uAt^HSY$QbAGyB1
z)(?~0ZUXB)6RZ^xeaNZkLOEa)jUL&W*)r+7nbA9dg9wuI)(Z5qN&PW!xkxqlicFM6
zWJoBXT^uHP?C#shSwt6hT>FWsM~Hz|kRMz8oauG&`9(#2SPli2u7XZU5sBgeoeE-x
zq6C&HhfrqLl-uv%iFAoX1e8#1&W{aEQQg_|w8687^dx9lRnd2a_Ljd+OSab3Jj>kV
z`B|7*5rX2a8c>|9S{BeMDhQ%>vIpfkNF>TM!8WsZ@V^v<s$imYfoxw`skx%UE@%vi
z`y*YfEH`(WV(#Oto<juY!mL(tJ`AvJSXEnETK$tBFG4&M(|dI5b?=Ev1%WU{^-?RQ
zWd$$xFqk0Sh2qlDCZ^QtYMu)ikfh=ls;}<YI5^A!;!Qg=K)NEiOA*jzJnlL=4c0;O
zinJ(=e!XY1z`^Gp3sz_RW6l{zZ&NoBu_;RPcIl25O#hV=sZxS;D_(!e%p^e#ye*s1
zWyMCj5YEHLh@$DS&LqgHeH#;_sFn7#uo*2aGPWmIWLy*2_w5y8R#uh>#n^dFg}lPp
z)V^#x1t6OkP3!iU^clc?iIp_}I7Wxu_tOdZn!`K7grVQ0E&eIeqq!PPT(SStk;uN=
zqfd=cHayWdnLa3%0|^4)`8U5+SJ%*M2mc&<Vztt(6{ZU-3R}z8($Z2yPSQic2bS#g
z5FrO+*?zwMyP2X6p<qttUf;WFv8T%^ue-Z`1;aWWLZiR(<Wc7c2w?T?+|w#DD0zfT
z$lTm@81yv-H41%g9;d4V{sA&b+yDT^t{7<;@n2A?lnXAM;Vo=33^{5kZZ22G?(8p$
z5mM^!-xut8age|SPS^~n{yDrNf$62e>-o8~1S3w2thyg8@Sd6xKdTmosHk>lp3^D6
zGL|qoZLE-C(HH=NXnW*vvA-L0U=J%wbSnl?^>!NS==9(6C%x)r{CIAxb?#dZbqG-<
z1UY0euuKGID5!U%brIjfF&`9*m8-G@l8Q^u&vSJrkL@?SD%%bEYQIil!%uwGKzX=!
z3qbG8!#mW?!I-TIpPY<6?=kr6>K%0E-7Y*vDd@`ZKD(ENE2~d6y3bTFmtI~Txe2U$
zc(lqmhm+d>orlE%&=AS{Lz66#z*aDOU599MMees!Fv&#@nr~wLvr|4M^HV>vcQJ!5
zDR=9e9IA=8aITd3r1M?3-vcck0v-Z@)1HeG4uW?HtS)m{z_1LRhX*E5l;jvfE0h76
zj@mn>6m55f$jJHMzvHJ~MO<kQ4(`g*f{+4OC!$NNmqf<J7~0g+VV0MN2(Q%~?5X&}
zq7uRa@#Zo~VN2-VB36*hLi|722L%74D$*%CJLY>5LkYW<yfsg1U|C{li+mt@%%j)*
z$IAh8B2{lDkFNUWVs-w5?d{X1w>CL6sAXvwq_mXHz<LVjg4<Nd;zcJRiA8de4&z^u
zEteRm<$A}ztiYki(y!OdL!#6c^#LR_F{Cj!zcW~#eoclE<pcRF*?s35X&f)7*q<o7
zO`GA1lQ+ePL4pKyn_925XGO#{r2f59gdDiH?f~Q~^OoGUbxvd-%@<DeE?@#ps0iR%
z4HWIE#n$l9r1kVddjp~_oNJ52#`l516j^Gymzi`&ZGtf$O&uLs>ixfwY^<T#*|CI-
zDXK=8!kDzLbSreR9D^CBx&K8MxG_>Dru5M$C22j~7GjN*ysF4KZF5>`#dOENbz^@$
z@rZXvuS$s1l@MX;=b0*}K1)q<VhN@wr<G}~xi8uMBdx6mp?=NZxK+lqtjnOvRxX)4
z<Uxi_PA=8q9bZHlQ1H19q)PWBWC!Z&>yr;%4JH%5?FWt#VuZiqOy6-HJ)M&-z+!(3
z6Ii^jyzhUcy?Kb5l13=x82BabfLep;)ev2iTT(`buR;`<5+upCP$uV&&(5CrkBu})
zQD^r^&=#kE{1~jy=FaJ;ZauyA9skA|F<zQFA~IuiC@GF~ZX_emp9E|GnpL-`e_!_k
zi}_<5Am~5<8*`|rDLH9MJK4(LTCHVid+LR9fUeqdi<9N9-!YrQ3;=`Z9@%Rz<Lk9&
z#Pg!u0oBq1Q#Pg`pHn3tW{4dp2_;Cv5b?PNapc1K4W+|5Z^kgZnvv1zxhI}5W^?lo
zZI;PN$xDui=ELT=j|U(DXV<o&U&bE>5LaZw6w1k0{X#7VP-!xCpPp(*(Zn9Ky=c}q
zXT-gvZ1?ru9UuE{_Q@Az?<LuK(+Ooc@@rKYiipzC2pqM(MkXiUYi_=S->~~sh`WLU
z5A)`YsRBTs)VF-HxVQ-RQs4rUw`6N*C;=|6^^NdW0cfEdm=b+iIfj==par>8AOCX=
zxqHMh#Y6Y8l6s^w2A)#WHfsqqo2`*!Vr$;Fjw}>0NhONsNU&I9W8Gkz^|~q`LJVxQ
zO8G9lBkaDKyYUs9Dp7fPemE^Mu)LqFS7#Cdi|qZ&uY)B<YRh{h+zl(rk?@q7`bqt7
zj+gjCw_4!2%T%Rw^>CFILez<!TG7dn<ou4a)%W96N)Ulrd-~0>AA>_Buw+%gQsjyI
zy$slxR%cm9M&9V^y?{(arCRMtUJJNDeT)bg9WSpt+I>;y5V?~iHg>|LQm+h_U3b2v
zPnBVae}#`pkzp%aemp?M?irT-Jo?u{zEzt_i~LWqAt=6O6=B0&p#?Rr)R3~Y40mrV
zvqYuJT<3Kz*zLX`>Jj(7@+=x5Soo>-n1K=g#*fWT=xCYoBroBsDnqYvF$2n_H|tpc
zk$Z7?$f2U@ci8iUy_lGo1y)vrpp5rS1<D^8h-ev60NJ9+rO9syvJ{!42d*0+_XE=_
zJUBYQFaP3V^eY3?#On%;L4LMRb?4_z%&4daGc!YJ>?PTwxj_q7!mB*FX=wwspUwIU
z2b9+0L=bU7Xvv=BBud%tc8KnCbA28M0AE$X;>%{-ztiOV-_v}9BbT-rap_sOauw`M
z_r9<Od)Z!039jeXoQuo)`@!|_aO}BJ_qrM&=X;HZ8zK9kzvzDZmqxpm_AXGrNKo70
z*s0Yz45r->6U)Zt;F_-W4H4RgXU)>|bRlx+S19W%8@J(|$e%0OQm{b1G_E5*En1Q=
zc@!T>fgC>f#O!Zcdn)F#?7gB0Wv2|qL1_6(?oSl!1}$9{koQUxA{lT085`griR*9F
ztw2<jDwy1+qU6J%#liAOz4ZQ0BO_0Z-Qf}}L#c+hb7;?OAA<A0X3BxqGh|+A){lck
z+-bCWtgly=VrvuB$jhz-7uatuNbwGsoKm)Z+9d@7;gGK3ZVS|q!(qfpqOKZDwZ8YJ
zuxuPwZDhM7=H;^zI5z*PsA!W%Kn#xE(+Pi@X^#|z7cxN<b^7>+kB3vPi4|wsrm|gR
z_uNsP5m2>p`?F&z7<W&{N+4?8{Autw72F|r61*y1D&#dXY@n@O5l+Q8!B8SG_axiz
zjCitP04J<QmUhLXh#9YKjNZ<M1Zh08)<eta4_ep(xe(%+wCh3Fzna&VD`UknH~M3<
z*XIxT5dyq&rR6O7MQ;`-agqmsuiHVWgNg12!~8e;`9DyAN4+LQRFrFOw8zr`fXYlq
zHd^5zLkrcOk4yxGm^dOsmfYFJ#Z%Prl7R?~DrFE^k^<$*>@2M>asIAm_%M`v^zQ|%
zddb$xN$Jwa&<-GO3#qww{{a;l*D8z0`&{2oqw<(yMZ!YhW$I4|0LaRF=ZU&iNUiN=
zKh!a!e{buI)*Bj0O|4WRRjQu(cXoi3QFEJ>jvR%iae42><>e>6#>*RnzsVqTef&!l
zyT6iNg_0a8&0vp-roH{a(3|Gw<SAld@8lUPAkgbV?qZl5cT6igK-9VpX6WAk_H#A}
zrX^vP8YO%?i$zydbcow7P<$`3Sqzu2zRM+9{@Fnn@`!W^x60(@R#eebaBFrc(_A)=
z+`Cinq%2lG{<k+=VT#!;>>GPi2#E`XiMGlQk!oO6Iroh9>OoUZPSO0MM+VWz;_;+Z
z51d37=)Hc0wl71Mn+KE-6rIkcVh?%KU!}++!z(7p4i(CMzB=qNXe>K``)#$oa-y6S
z_IjEa1Pe55I1vf7fcaRqm(@y3uSw+I02=}`0Q^BkNpEb7ANd_ImB+=E*u*~k=g<6`
zw4PMf#}2<l#`hRDkY9~<;}%_xWByQ5)v}Ebok8!ST10ugE@&mHj~8e#2um0S6{q|#
zEp^;^iPzj}f-Fond)e4I@Ck)5DJ|O#In&<=8c+ZeUREtE^x<-u{pFETv;%hga|UT-
z!+7to?d&eL!@`Z8akMH!n5&NVrRi0aDN^XznVDnyTF@=ki-N=h!QR6!c;}@?E033U
z@BEq6kCqgXV>WLY{$*7(Y-F+Vv_N_)>(KY_`)+6n#q}B|SJIz&sYi?}{qFAaXUY)5
zX>2EHYYS<0-%oWuyhiq1hXKh+Y^Xu7UAO1WqLlRw5jwIHu84BQD^Si*CiXbjRv57R
zUO*rrYp_Z4@kHNEA!^AIH_Au4Yy{z}(2YqM3aW)%K~XZ<oVDFV>#6c{retCvHQu+D
z@(uwz5`BL!%2Q(C%cE}?Rm(9l7$QOA0Q<`-Vg><c#)@=@57t6fkoQqh0k<3B{+-x4
zT>o5!(}ttQ(M2}I8s#6ipl&biuw088$3#zW3Zc&Y`Eh<P56QG9C@ARHw{N^<_|Lcf
z{%e|hh}#(I-xE!ufz3d6iHgz4zI#JXT)L!G3$q$+7eA31yH;<D9kk}hG9KGrV*+MN
z>pg78$VcCIX%U?w_3-2MhRYRR92XbjC5P8c$<hUtRX3T6#X5Dh`fQbPIS2zRsd9*L
zh};J-|H&Ee0NFCg_;=hI6Wgm_MJpJp`Kpd&{X#kWbXY1GycHS6m<i?}C)31aqeX%3
zd~W8-Zu!G$VNp@ZRK2heDFENTiWMVS6`YNW7Dg7jx!xU|pCeQdQ8P7w{ENfK-CY5d
zk)YeTnOJ<)>x?&lg5kbRlbSwd7Y2<t-Nj|90fj@iVrJvZ_O`Z*iyb^?ys=?DL)8x(
zu;>)r^43-TUwrI+!(<s4ZCkk^`h*93!;oR{TR(@FXwgrEK$HJ5Hrb~S+MSyv^BGuJ
z%FC)=^wcI|OY$fZv)<6Dbg0tdM!Eem{_0J!y8KRBnzsuCJjQFHwM$C$Fn{}^sy>ui
zJNaRT1_AL!7)o)NDUt7}CF98V$~A0`Yu$DM^N(*cMX0S~ev?Irt6?imHK6s}I4zuw
z($hcsql;g3=pn!Px57W4R@*X;+u)pTHvNVHDxLbj(bw1FR9O7(EQuYD7aY-pmWTeS
z+~OedW(odW$C3`Lf)h{(IB$1?eZtT=eeU?R*E8D#1f53Z5*7IPKa!jOh=9FhV&W^k
zMuvj}4Q@-<mrMjy=amzu%|N?z_V~pa^1X_(sLN_)WPm-586iJyK&}0owJb?;TJ*#|
zGGMvP4_I*$h1h#c9?kR-vYYCJ_i(2BNYQhG=sFB>4BD`l7hrxkyK0oF!5Q3V`mHqV
zZFo?B?7JxyA^S902wR%q(;x*?3KKUsE>K%n$7jEso{~~@utmFTsFaHan09hma7~Cf
zAS6h-kMMv^3hpC!ze&6*Wmmy;fg~&im1Z(zYMyj6ygptCFeX&hR{J!m;PW&4t7`B4
z>f~xLgg)V~t8k$ABrPy6;NR8`MJe`caNWNwkeRwM`1+n4RLr+qWmO$aufoY&-LIY}
zr6|D!))N&)pOh~wE33KJd4%ziNi&(sbzje_L^(%SRTTj6-X+j_X|}B3A$FO5US5v<
zJnrqcCgUHG{K3F&aiZXdc6o`EQAe9um0TkHcL+~<8C&bH|E$i&yQq%KhcTb$I|?>>
z2E2Tj%ZKX>sS;b*>g_!^K){|XD@P>0&fG#yN2@DTK#c^%HRv&#Wm+lwt|y|3nmCgA
zMDN~ET;S44H95pB9!uUY#}K0(iy%UOg`$o{rxiA>`9FaFKDeB=d+p`KQeNJ7jHMiW
zojS*nW7xkr{vC<mKS#ky^8UuB<1XbxZl>7F>cK%oRlD?+CzB-Dlq&aXb|}<6_5r|7
z-M_QIhkL?VS7Sm*vz2u~&k_E+Y7~TZpycD@V{2Q+MuPsE&-(!8@Ww^%y+hGdF=v@p
zjk6bKMt_E41&a`)A4i_stWHGU>9NVp&gTrCvaRrJvcK+*ZgB^o8=JVfg2>CO$6>>t
zZ<bY6>eSKQ%^l^c4@3@CqhV$!Uh7p{i#C%1Z0(oo`v3tEc0%EJ+6MC9<Pb2(V%>r?
zrb^wG<mcTi?a5-ieaFv0+H|P3@WZE*^ZvUw;o5M6ES6NQ)RY_k6Kh(<On7U=4>wfz
zTa>5EHD+?B?C^jr-$5kpKoon&&p8i5O)YnW6Cc?D^V(rgl>i-`g)S8-HDN?<c{8aX
z0q0LWSph!KI?8K4L0@di;vtqC?@<fB%AcM=jcce~G!Qzl1;<)i8n|g%jCq_?zBWdW
zP}fk{ON@qceC&EZjEjQO$b`qptdmDH+PWSTIXKwS;qCag$#r3YBRlv9QCM^B-i*kD
zG?knv0qLXHxSTb}(Jk}4OT6ohW6`bobQg+X$<q_QJQD6+uIJT#xYl-kOiB2cGiySw
z>-5I<mq}^)bc7zw%@^oHXyK1oS~*N5Jv&~x%&WJ6cxu*zzw#*VTbDW-R&H+Yx-p@5
z7_-sD9MuVve%Ih22gWlkEU!huWtd3%!ivM=jNb!2slQr+@AsifJ<b<!A|qhez;TAd
z9qBbh9rkmeAYRa*{1)Gu6m^2KL0q0clU2h|uFuJ7z)LF#Xo1&wID&jlv$l|uxtZq4
zlC*B)@j}YFnApS0C~uZ;3}D_V*dxl6WlZY|Av(IBQT-B1BR<8?K$b%W-h1*C)Q&ej
zWFp@6Q!YjU$F19m2MxAMn?8ijpMf=jbP+GF3su<+|C1QQs#6nLUcqzY_q_5B^URhu
zk5@M-qk0gRHT<+3O`Z6B%aH1lJJlJ&Q@_pjgZ8~JCtWy6nm?yE%#Z#ItSJNL4e01^
z<lOYC^EY0VU5Qmxe7<)7Glm2%KWMZS#|fA2w&OuLN7VbuyFEV-_sfJfU3_66YV=n~
zSZ7?uaQi{Uj|kf}-y3K}o`cJ+S)pQ{rr^%vT%*1y{YN4s6>3N#xSQT%#5$dC8oU&u
zXnVO?o(CYed#^%&-}^#t;4>VD1qyKXSv;-y(^WF#m2h5CM@>s<>EDyqQ;LoU8Jd`H
zKdoHr9bjYBXt<cNra!PX*VRHlz4{#WPRB4PX5nukIMiX4hRQ-PXyc~!!NYrWkGZtG
z-okwRoN-i=LCSq}efjYb3KLs|X_=wO%gmWUlmpLzf)tSNDs4-_{o(1deE;wR+2ciH
zIi#odLzUsv#SpuPfalqIcOtHZBa-0E{IPM3EEMXOLpb$Cni@ezN2XlsTTRWDDnD8d
zlJVZ&-UA*==5IFyW#akdm`TzKOxPc1VRM~2Hlwm)IyV0^k#+2Rw(4_gXKzojs`u8)
zSn_rJkY-=sCr!-@lhiE@Cj7Ldx2H=Uk1071>ClS{$4_(L5tuELS4$gchwA8ZIY);t
zpCv<-vEX3<bIW2nJVAdBoQQ}A&Fwt}oHfQClW$OxA&}(1Txf6YdeO75&LDEM`OXR%
zYaDynxP~ZE)Nvt|AWs7Y(6Fy$5j8;hWx{0v4gNQLD0PAtgZ^*X|3WEpsSZJJqgB}q
zR@9ruMNpXiu7ZH7bwdP7M6~q#VB#EIFm2vP51{0J==|vKx?lGoDLKfU*sSt$cH+!j
zIS`@jn6LUrK9PK@nS~&V%Cp@N%p8G}MO`NklLlS#Ew3+qfhO>=W??P`2;lslD7e8h
zByN$r&nmNeRU%Lm%bmSAMU@VU{Q7H-Thx5qcT`w^3K=*+R@$&7Tt7&3;7s?Ei5=Hs
z21+%4SZ0(DTLwS+=I*@-Uitx(bH-q!&9@{@Uylvas2P7g-cyP^SljD(3B+DY(o=S>
zIS_=z8e0+(%YjUZ^Y3R91+~lifNI%{CeEwrxwrluRLDfbCH$~iwQAlp5Ci%9g0uks
zyj~4y`XX_vco=|~c=K&$u?R)nEw^nG>D~y)iqQ2-rp*V7>G0+0*h*%g^?fWv48X>F
zYf{G}C~)`WFET_IlwnMyV*0@{{H>8L3ILuE(YuK5VY5EpL7M6~;-c_^T5OVXajLHh
zD3;1qn0!1DV%+vkHQyv37ub^A;JLMizkCbalX;*UC;CS)G6rwz20@W7xQ*w_Fa_(n
zVLg%AWal4SHBe;YU<0X;(Spf^N+AF3;a{ZG0(m~~s3VN&hAq482?OX5U>M{d1*Eao
zuOvCXC@mnDDEu2<p@C@ocN$S6W94Nk>QsdEzyK3g$+SW*Mn`*M;jO2!3ABh%#@Iq@
z{X5j6e?10c1lZ5pK<e&pBFLAY&&*6D#4#bFYxLv{=}|+olu`rdq~zg(iSuC=Iz-OR
zsIb0n9?t+&Lqkd4!oaxq8C8iEeJn^c>0HAEqfK);TJRm7%s7ONBWSW4)Isr7cBD?D
zN6yW2EN8DxVokM#;l1F(EB<#j4dCcukaCj<JrYtt&L(o^ZjGt5^}pp)B_MxVSv;?s
z${_1<)k>u!Lj*kRSd(&c_$$@FKl+OSQUz+|=_*d}Yy%Ph={c2WYN%3HZ{EPQJ2hjv
zM9KR)7NFYAZE|z7*Ca-I3wj4o5k#4A<ZU8{flK<2vx5~*hB06J?lHr;mC-5x6ShPk
z6%csz2m9)BKca^zMIOx(RD#OKQ13ucLgMwPlJyOm`fYNX=d!i6iFx>Z0z3_k2bN>X
zWJg4g;8FH=8~wZ>{2wtnr;n%_paU1;fdj*e{*fj7sy?lvByTEadnLdS0UJ9gE%L8K
zKm{*RI=kO%Z|MeE{2oV4RjTe~V>3Tm76ncdpjpyVgB%F&e}_JPBfN$^T%&C&qO2xW
zX7{j-8wbudfvo|u{x@Sj(j`=6;3_#JA2?}T6BA@<vvQP?LjiF@(Be~Z`dkH{538QM
zyateP{f3dTf41VuV=`NYAO_J6fZooA$Fp5GHFUI(^CKkl`gjXGQ%pK0FucSOUjhG4
z9At5Byu$tZL;e>uzWxH7rT}OAj#HI$vwb-@Z&6?4{4=h9YgB#r?p^T0u-$?r*dQ)p
zDC1GYx@UZ!h-umvjpOLSI1L9PWkjpxpH8i`IQB!S(+T>C+1Vr#f0!9$MZfqBGkpGP
z4GR*8|AT|zz4}IXEv>Au)ZGi?3Gmyc)ONRbx&U)79ez*~B(S7U*e23Llom>n{`aT*
z!q7>7A-<LUesqycN3%7QO_q3c|H96h_&%#+;yWU<<^TM?Y{^uGcFkr^ON;De@ep-|
zR#uV#*|y04<pLxn-4n!aRqIQG`hM`Q&-By}g_QuhV)#~xGDTOGyhM}%j;s;BHD8y5
zhu?9OonSX8;pbaOW5T~_1$iVwB8k$MpId#S1`eGg#h-9#Q{>awjdJfhLQx6*!I_aB
zUzfk|;+u2?3?iUv9dvs3p#ks`9E`!8lRDML-QZ9`#Gw;4s3DkOs(Zlq+2RfceSOjj
zQlnjdDiN!!HQyioPZ?D^ohC}-xi(wkMg{iBmUU4Q_?bwH!vDQ-&ccl5Iyi+us1JUQ
zAp#zhs30h!D^kiR^m0a003?CjFT0GPA3uG?#TYgFPYp}gK->AuYk>=&kdq@)e-o_C
zpDZQ08SS2o_)i49>7te)fQ&@}#YQ-)<P)9aav<2xhs_4p0s>XlPOwjj_zKhl%(EVZ
ze1Fp6(oZ1(cp3NK7DWG1l+h92Bp7oPxc>J6Ty272@Rfh*vj0!2X$O54-t~QTWEem~
z?G_WcDENgJ>wr&`0B-pUEIQ)5jdk{9jDM>Ln<^SNR>S+;9X&~OZM+-je|6B=)PM>A
z8#I(+=Sg=Zad{ke-3@UsCA|F0bJA9x#^ACrI~Vd@LEzW?ma@`!gOnZ<Yw>IdRm=CA
z(e+a2^Kr&fiN=fbhRO$_Gg2Bs><w4Oknyv8nIrk56P~=oQ=O+})-47gT|}wHVe<!j
zyF%4u>^wOYL<)?U$mn?qsFso+Aq+bgo*5GIMR;SN9F(PC`!~S-k8=<|V-iMgdOSVL
ze9UhCyW#NEns?WCU`gTmP)z919-wsht&XS1-zy}H0(6pCB%(croZQe{v7^VU36SvE
z($e#h{FOC7;QrEbRf~UG-trnrY-v0~mZ-Ne4*B(T9ms(!#v?|+<wl_!O{Wxd9yO!P
z9_cd`j?1(EE7e=aa?ME-5;pMXiqqq_rkIHB{XITXXB#W?$og!Y*z+yr%dJdWaI5?A
z;Mf@Db?e*GkDy{<k?^h{xRhRt1$b^2H7t>E+Z#Lyt>{r6+P9r<;w+3CrZKf`Tsfp>
za!;xV2!OL1OiWBIWWE;z^l+cM##d1!pF;7uC9Q4?5$ilw{0LXyHyt&_rM6v#q&1CJ
zAGP^Uj1}q`Jtw%!?%553@F!tr3Kb9Kotm0jTCi>fFHvu(+&DuIsA=P8a<6v(?%s%!
zzud;sSH7=!KKgDm34wkLcsbtXey-!WE`ZX)waaeLhw;5!?|u|Ja&Wz+4pB_g5xmMC
zve$Py56K_#4v|LI8b;3FNWvKok&^dX5rxqpM}qHM>cRXbxNc+AVg7td6Q#wX=Wp^^
zXXNa)2Q;7jP|NV_p0CR9p&!9}*wS^g``==!<z87@UMunSv+4S3F0rN-*x&Fuh&_}L
z<~FeV1YC;`S-9Vj2Dq&~4}=+M&>9`o9%>DDn%xiOc@eG;8vX6<+42XMVUl#eL7j6k
z{?SO-$$KeGcTnRc@8(*^;bn|aWjGo05*D5&zu^CL_LgB)H__WDDxp#W(!D`IrMpqs
zba#V*bR*q@;HIU!Q&PG?8tLwk?(RN=&-tJC!?~_=o%i*V>|f2SS+nMz75B7|&941=
zg)Hiwb5xF+41<q(1ESsJth{xE(b9SE=v@x|BuR)OkmXiNnoH~#GU9KFYjz8Jd{EdZ
zzVY>c&u?-*z&cPq;L9b6FehUzgAe^DOh(Xt$3>p^iwlH1?myz4*BD}@@urfDnJ@CI
z^QnqGPUZ)zJ)8QM!orz7RJNkbb!Y9al8Cmv2Vg@qST={af+ka0X!jY064CjYYw^fc
zDkWYQnd@8(Is$P_8d*(tf0Cmm5p5B_bz&yb`vE<E+aeXM>;52wOHF*)Pwb_^coCWW
z)jY_y>%})Huk27}r5TCfCgtQ!wP%lp0wvOKN1kuX|4pZb>o`By-S>+JP3MOaBwkv=
z^hmV9GPOrV7B?eh=@2t4TRAIjSd9=UtUoRjq(y%v@s|^P6jAmSyw7j87&+CS9(^nH
zZkwu^w$#s)8Exub6Y%<wIj7_vl442>zm0o1`rBBqKi?P_Z+F+T7FoK5MvkG<XEOBM
z=PxIkFMjs+>ni*!1O)BaO-$YB^!9a}Z_DtZ&;EFGIDC8#_krv=Jx-gK7>#o7r>4Wu
zRELWM*iM}2*TwC}W`N}vQJhR1<Z5NxRanD^aLs$v!0$9^-dIYzDZL`~zqsGwTJ&Wr
zVd!+Id-W7f>pdYVLJHZXTD8abO{G}OKEa*Y$w9Xo!v!s~QR?Eiv4cuVB{zX;N5i}p
zRv&4efMvy(>IcY0+g9nJ&6^5ks2Xr}?bm#8UJLEn^$WW{D=Oi?kGEHK_<$EuFJvOs
ziXbxd*MZfLA_{WW=12BC1u5;ADu#V;0T&_1Vxs2R;_hz-$LDVzEo6}a7~{7JKJoEQ
zmxaXWy2n$l52h!9{BGNmC5=~IvHY<(>w4b_1YdA!zeFVf`)GS^gcB#y1CvlZ`upC=
zl10XkLYl>{_kQJjq)nA{8!gTAEH8SWeZvHU>gx}T_15W9+4=^uj>$k=<K^-D7;W45
zq)!#NeCM6QnGVL!L)4%4Mu5!nmPb%#pC9*N?IpQVJ*ca_t9DZ0*Ab7sF@L#Pkb3ZC
zshQ|o-CvgoJrRJM15m+ujb?|bo2A&dNQX5Y9~I~1Se6vV{k>TZc(Ya;jXQVV>1k!Z
zkDgxo{3dSrZ#X5|z2LkA$C~fR>wXRAVLx2jBmr@qi~cUnM6gat{N`VOSX}TxgxVH)
zz<E^-$1J>lwG~JMdG{_u?Uh?l22*K=qi%fwmHY}Dfu=rdEkR~!*A!1FS)P2mU};(g
zStIUNBdzB8-LD>Dcn1dB`m%?`T*LL<-SV~p-Y-2#UqysM@ClC3{|IYgdTq3Bjphy}
zp-T<#uo;c4HiiwL+OYmsf1ZYzv-+)b#RgBl5uF-N`J(gB+VzAmk8Dc)xzDJ}W&2=L
zbAywb6AP*^0%BS_B_=y23oDvg-_Q>Z8T#r2<$5}<`(8X*9_>Ce_#v_wsLuwIkO4W9
z(6a@xz8#MXns;d4VK2b#&P1W`($jLJW5G{8f9E$cVfF79A!Qju4#5B5n8zfvV#{gL
z<L^q&^sHNkwiBLWNDS{50rjIk;tzZPpIe@@Vj_F=4Z%?w!A+-kqJ;Q^H$2B=@$OTL
zdrwG!jurgj*^fBz;uhT@XR-h8i!WtVWoIQo&MZrRBGw027Af}lg-*RG3z6X5XCL_<
z@AoHQVI>Lb6;@>b*AOFG|Bpam|MR7I@Fi)u0DJ=Pb`&z6QzoIz(jRgEUdC*|f&%)a
z7I%gG?;=YKv{hbF(f;+Jt$}UrH+j&{$Cm2(V3_c-8VdbuTc;78)Snk2-)6JZ$ouWa
z8N`~uSS*%$|H>PiQ+OgNwAh!v<88}g9I}<yWfmHSWMT%Sl3B`0)Y_*jM|zd2)oxqg
zg*Hcv$FwuL)m9|!;2hDDQFtjj$&Sinzw>Fx3lX1;qfPXC$W2`O5hqbjzQpiHZ{`R$
zeH>w;1x40}p5$m^Gq154({e=m%~GDU&qKzp$Dwd(KbcV4I#1%#ZNWB~0CK@^gd-cG
zsVy&GP3?!F@&eW?jLYGk#R+7}xoP0<-`&{RXM3&x({}@=6OPq&_N9t_c!^25L;YxJ
za!~8-)ovd<v5Ob4HxY=d8m;7G5+JY{nZRyu;-+I#iO~=2uco#NE=u^j>9ES!Q2L1L
zr1<c-TQWoH<^+g8pyB<K{tFhbk3tECpWn{@b~zQ?7`CQ0Xm2$OhoXi|AKL1dGFo5|
z_;svBKr=RJKH$IIjzK>PN}Y3?OTxYx<UK(gum&-X`}=zboP|cW$?k5k{oyBY-S`P*
z>EwYDeME`)->pJ8*i36ChD!}ZLA@L|GRS5JyHIzbXK+6R*u%c(SmedFQi%_-9*oW3
zY9j=L_<Kmm4z0y_zOsr+3{l%tI1?N&GD3_exNzQ2e{mAwT*syJ14V%Q@MMaY#Fv0&
z42i|-2^<aF<2V1G-!Xxz|MxRcw@1DI?+?O9&3(=O_^x+Cq%5rP$qjIDFTMZM?#K7L
z@B+PS1iA|M8qT*ljT~q)+{@ofM_{zJ{n}CB(%>G4g^cML_#kM=f7k!l^EPq@{)qow
zB?ON@P6Y6{ydtm*kB^g41bow5Vp<+j`=8ks4sWbz^3;MPte-J3^l20Zm}9~1`i~RX
zB<>SDfzRVJV{z$k)7}#PuRm<Y#)hI$&ji724N;?M|I1*)3+{PrOIi4HKNLbFG*%o$
zFse45Pkw|zDFs9D6W|CZo#X^=8R0$yiLs)xC;B}KZUYx?nv6;4q~7U3M?xYnBh%Z+
z=LFf?0=!*1ucE@fBM|$5uFC7-ezw}0Lx?vG<#7^I6c>wtH%dUE?c)Wii<-NjwENGW
zKgZ}uQBMaL;L7yc^#%3)sW34xjuzOr%gUI;#iF93#(2rUwl3X1B_p(TV57qOngg!u
znV6_)Xk3RfFD)$@bcISwOIudueE^gG$-h>n$Hn0>FqAk^&x880i~V^`N=!CQr8G6p
zX1Hz<Ep2I{P|#o;A3Z#fKk7uij)YE9Q(Fs)+0*5wEWUny`e@tyF@l)h5hi^5K=!M&
zw6v_u)rsSakx@=<t;=kc<#0j~<X?v{UJio}`{>2d)6>gDN1LKjulx8sH83!Ml$DmY
z{I<YuQMQMJBgGF8d$cCvztm`DOQ&pa&H-i0SP3aA@wH?8;h_zvh5MiTzdi~Pr~=ih
z{SMAWU%Yiv(q{GIHr*>i!kwH<A!HoR*jNgtg6B!PaG%>?9YDV$e@%=l8<bC@q509`
z1VB&Er4_@&J*q5s)(&KSk$6b>0^18A|BWZjXFVl$RzDS2S5_x_4A9?zSX}Mq`$Y$G
z^3@d~^NC#+4&ziVA%o7mi*3W3+uBxrKcjCzj|hQ|{Bb`*K*bk?;^5#g)H~(&#%2cz
ze@8<IBN9(!AD%A9K*`L_6@^ylcVvwzMTot|V_-#pY@=_k<i|+$+6XmT2q>hJkW<Wz
z!^3>J6(&&TxKAXn9Gy{2A1-!(GiF4QX+n{yud6GJh|>S`QM*2^Gnm+cs+YHK-;Oo%
z)Yi&--dkAib?AEC9WJ$1RIo})_CDZfHgfB4DM0dzi$lZ0RJ9t&bRUP<A0q%vdR97f
z^PSbz&(}*xbxsFML+L7LXbTGq=%oC?VPVzP)g^ByPJ$R{)e6Uha4&Wy4*)rd@~7Wq
zs*jUF4+;T&Hg(t<Sp(b=StYXW_m7ENgQ*AG+t-&zE8GF`u)<h|gYYPOvmKS4U=(3t
z<YTIf6J*7Nb6zS<vb#)f3X_LA!xLQh@}&=-5<V=DqLQIwf_MImkBeJfUzgBrc0S4g
ztX=N>&4q=u?CcXQE$NQp=NqYEfiRPsy9@q+fI!&m1m8KV^3A37VHUL9z<~A_Ze5NK
z_Oh(AeS$n2tv(An0{*IhZbqLZ5SjInp*uP{zIgHC=;#O(wT-+%Lqh|8?IJ}3Dn7oK
z=(#(SY)8L-#uQxTtiBc|nU7>WT5a3AyMng&Z87((RPRWlg^ZY0J%>wpdL8$XWNv<5
zNKo+I`}cs$f<}9!(d_~>6(0|8jLz&cbkpBrIN>PuePPTLH&0_v8!eWoXwmK5%bY8I
z(ll{kZ&q98=lKA;5+ma;U;#$nfK1^MO>M;SA))~FC>x{xvT;pc_)dAG2WL46=yGfB
z_o}<1g~mkwloTXV;yHQ*yT7};L>L(15fS>9mUP<7K;~zYx*SR*(Fr(G`y04c7umUJ
z9xPKj^^nG~F=1+cULh$T<RJgMcRDpU1q04C*zWdtrFdXLvjf}~T`QH=i%muWpwu=j
zOjc-3{-<fnY6`wOch!EI#r&cc=e686Cw&C7dlWF$t69KfQ;Bs_FFs|VCyXGAoN2E?
z){&%Jqvbe~j{gidc~IyJin|3~X`ElL#1CIkKy-gHDDd#R^{K-daLd+qf(nj2WUm;j
zn|lgA3)Uf`XQYH8l&z!LA4ot6;*Bq%K|UKRe2KM(KaPkz?maEV#g^p!qVh7(G7$20
zc3UbyDuSbV4#of<#>!S8XJD0{K+O~sv|@FmiykD=Plx@B^|YU;?wpjpCNciHe!<}L
zXQeNh9;n8~#uOkeP~_zye@z&E)YrJdl7U`;bI5RyW&5Ot_t+=|<`t)GKcx;#<kWFT
z^xSs$COQ?})=lCx3yto&pf(P6KK{$u@5G}*Zx)uLVKww+{&9FUesu+wmsP?20QCr)
zPvh}uH>n5sdJ6RiS=_b<mrDXVi8i0U<up*hnP^~w#V5@sI5adb?ZdFVw^p~uVJxLU
z?oexXxqLFZg89yNVSK7;fc*4fDFAO#Pb5`U6_m07IG|tEF<7q)CQTH$A4P~L5!l(V
zu*_&P$PX-EZYv?a-HPhx_z2c#etygsR<Te047Rq!1B0-M5xV_u2N?+XCazO|UAoMA
zty_A8CumZAHph$OvERD*<L}KNTHV5mdN|wyPmzW)b$u<Rk&uxIS8EcA>}9pXf8x2v
zQi9L_A`L&S6Gg(?+(axp`?0SXwG6j!OmC>A`I1744KGd7fjplK4*gKnDSm8h!d<Jb
zR!m|pz16-ff(dsXlP{(m3%iKE)z5^+g5ZwPvA=rVxb6*j;m-r`U(3w5e^ZoC|M%R@
zZEKZ=&2)9V09KlKK!-^WqTeVhbHsD_$B#bsm3;E~TaJa?Q+Iw{h)Bt&-bc#|R`8E6
z4zJt;{(&g^$2I300SnC&{!|9RkXpaTbsQBe%QR2uV<l0ncrD`}?`Q+dc9gGoe@>4C
zeh)JIV5&fDQ<M9Z9az5Hmwh@P9}0)Iis&Q0ksDZ&Gc{FEP}uq|{iMofsa3t)NL;r>
zvzi_P5lk6wc|Zf8Uj#ICO|E54!Pa)__UdEv-T6Z6&H3Kw@UST;im<o0myqZwHy#Ap
zXNlxum!yANN&n}O$CqFH`UNXv)~MW`_X7!_xI{$mfW$14T6TMT`;iF|kPo^n!>a@Q
z{;#iGN;b8xiRmz2IxH_QXftzgRD*imIa~(+WFDuxYX<>=W`?HAr));q|2HJ+UwO;@
z91b(DKr)o%Znia}@`IeR@*8a$oS@8%jI0t+n1|<2p3&Y8)zs9SX-p)>+ni}?Oi4)r
z<-wTcBmX3!77T=r2%)2!2q7FkKR3I+#vmsj@dL<|LL?Z}yhB+5=>TB7z?9e5*RN-{
zq=ERX_uteRbd@w+_blJKSSdt6GQa!(!wemac+sdm-oq30zbe_bhrZg{iZ|JOtUK%u
zr|?BY<@UI?|NQy0jm;qt7mS99sZ`bMoph4eN|&a9>)7OgjKSUY#Psw_R8oR}V_%9V
zJ)MRU7!q>8%{?<Ug@;a}WqU!7fsKo+y0&2rO8fw_0fa=SC+GAHPGK-iEUZ02A=4gE
zmYNxiqEYGt>d;h5G@CyHp!Tf$`2~}&U3YaUSLcd2R=KGL0Y);fiycLq$=%H&79B0^
z=x9AWJf5&H#&B;IB`SX^3JMI?(D?J$j~r<seHD#&332b=zX!~mlpe&dUNKtER6HhR
zL`T1+j03z#@bK{NpO(O=_OQjRtFmAZcRgkC`Wi7bGU9rF3yM<crl+}%ok7Vb>Oxs|
zV+siZLVrluqMZE4?`YF1N@A3lK^YL0T$z%EDjV{>MOhUU9Ef~2E!KCB;E2e`%*;$|
zV!u8mRwim{>QrtAiR+ogMX!-8aZr{oaC;~eY2kjb&`4f_q~9<J5@W)Mxa<J`D*9V3
zHMPqQT;12NU$a8$s7_Y@M6wtUu-Z}l{P`1%?2je@mVD4Xn|1{XRGX`%JUu;ibaaya
zTD<NYT!+ZweF~A;jO5VV`31<yTY7o|)o3+wV<pYZ+&GCANWFfPGC3{sgEH%}W-m8?
zm!oB2H;@_N<ax&K;bAr&czSxOUPwh0mz<n@w~(t1<V2@kclwNj6R-%SrID$(#c_Qx
z^7L#vbr2Kl{-xK3M#4+8Bl22C80(9KgoM6+#;F5PHZZ#Yht%P62lCd300Kb%b@^L;
z2dv3}+mYPE1MpBKY+u%cph|~;vrVP0z?s_xz=@?EH<77c#aAqE-)=pyvS!7;2ABX~
zDAMD%A;5co2h#CeRr&7rs(bnq32W~FNdX`XL-rGdaD&oCBpt%F_D;v?nxjVaOl=Tw
z0E#<0I%>|~Ug}^zOD+2uO$n0TDhSDNyO0VB#G{~yuZ9~M10)Fn0RjI0h#151EDg=~
zBe38sl_D6d&OZJAPFzSxR(pPMumm~rZn6Cq<o!mnF{Sc2(Z7FRu%OJ(&%ZI4T5=Vs
z7Y7QsxDAW7YV82~5xTpEp`c(ZV2sM5MuK3z^gEjn-*%UflatfZBEnc(Tk|*@SM_?h
zb8>gT{qCNmAV1!o6g5^_$IZi&2m=xXMhRvDa5^AuHS9m|hC!OHsFq8b`2if2P1UKk
zN<KXUiPjGG_9+5hBse%m#>Uh|2!Qxy{4?wY7L}5c(kppawW*k{uyHmvw{e+L{Zg9&
zqOIQLNa|j*y9pK&fdv%>OvlS#Vw4H2rX#wfII}xNfszsu+qzEoyq5DC4;LCWjOU5g
z+uGT{<jmFR^0qMo1>31}l9iOC$cW0z>+K5l=>C|vxnthWcywU#fR2g^%87tkO=Q#)
z?8lY@!@qIs-iBTr`!IR5GFHuiN+f1JxsS*Etz)CN0l_<`bK}I)!b0aRJzE$N@l-_<
zt!sq&a#x+_D<JFHE{xdReBJHbrF80#nqD~|#s;CFh|SKLySuwccmr_8<?YGf&TBvc
zH`iUPZEjWW$r0w`qn;`Hup#geeqRfARMqc!pWNRYa61sN7zI~|7B)2Sa&d8ylYfO}
zWMninHiB)PuHx5wSe}VKdyU;1v%G><5)`6WuDb{C=5{m4e`5h?)<y(;59T$MwEC`p
zRYhfrhDnU3Kq2vwNw<-WcY)aNL84wSAU_eGbyF==lON5{C&G(+g)0ihCMU1GyBV!s
zoOuC1q9hIV`HsHz`$>L21mY>^<5-lWeZ48)6x17>lyr5n0jB$L%iRs>eY7tcDL<2y
z7T^1_vSZa^sm11pJvLB)J#RJl%l%03dILZ#^YtFw=kRF4!o2MDWf*(g<mBXv($cu!
z{BtO&vBIzV0Xc00kV?ZvNQcutw8|`6eFr`~DoUvyKS?jn>)~b*usYi~I`Z56{cW7;
z5fdK%5nAu{FyAs&qoB$2o_y(<@9Urrpkg^#)SK(l*EiP37Ys{BW}Yl?FG2pMp}F}@
z3NxSUa}vJgFSS|4I>ymMa}DpPM6T<8{jveG-Bd9Lw!pP_o`%ctjzQbi?P_7Z{=2er
zt+CMrZ|U}ae<Ih?l6WPim$0zkT@Y@4W`>}U(BjRwgrp?3p`nNEn#A9=UN95K$H%Ed
zNo74*GLijSyMl0a3-UF1oQhe!M1_U9cz7fWy1|ZiZ>Ew0f)n&6*>DZ8r!3F(TL_W|
zKo3Ho<Vwf43fg-Fn3$7|ZqX&y|2Dl>1#4Vf<p+zT$IBfH$_22^fgI`E{rQjui|5(f
z+tt;$#KiN`+8&^|6V%daXe5-Y&7E(Lqq({QYE<=7_bWi1fQ5yX%IktjNGJ&f_OSQ(
zgvHl4KnxSB&B@^ndt!Dj?#17)_U7uw1D~0thKn?z?M21wV+6_Mo<|w3%k7~eA_Ofh
z=1-qqi;7Bm*_clh{;YitNFm>=7SGjwF-|568chD;lH7e)=OEeSaWfuE!eLdE0xUX7
zqjDH;rDIWcX6C{UoPy^i`Q}+;#L{Fj*oZY)OwzA8sCxKL6lDV6>fKIBb52(71{de5
zfb$2a$j_g51^F_KnfRY>99@rmU^U_4K;z;yh_4c3=+Bz%veyT}OaXAF*KCnrqpbbU
zA5*X*KhnOOpVOnFjFGa6eX>jjcxN_|frf_jmzUQ-XNdi+bAKzK90@_jVI)F}PvDo9
z2!sVj0Fn&gbCmcXD!S1apkD2CFl=U)1UB*N#)fZXq$o204YPQv=eFamvy{7g&-Iy7
z!T7JL!l>7pXrdSW_xzCjyu7>y56$_muAnxM<_fbX?;1RKnc58agA()~v}%5(<-xd}
zWbV96i^avo-5nez$_#Uz8xT&`i2+?GE=&25wdZLnpspjwDGwuc;&YaBbqom(e#}Ix
z=+ugg0|+5&a!oRZHoVGC(Usf9b>wf$Yogl)vDbocumA_i6Mz3@mlKVv8)`~Q&MHeu
zTieq?0aiy^kL#_?+Am0?+^7AAOT}R%TYFP_<gX`A^|B6TqE!|T+(4naE-tnM9i0cz
z6e+1-@>oe=Ti()a-mZj+kdSCgOXo_(+tqnJ$b^w#$d8<CR6(=#^q^Vc31-VnAEDR-
z16L&{frk@CEbHs|8-uv~9vWE%nyo8LC8CT!bF=Vj>j|x=%d|BUCkl@~#eMtsZEkK3
za5B+r)!00;v;+8be!f4HF7)~HYG(+cInp1u<8}~UEvtxv)$w^Gxc-I1h+b%ZU?kTY
z4IvCa%CG}PH~*ax0QkBR&wxL`tb8<=uF3+#j<a<{!C#MGQnJ83`G6&qB>TxTli%Ji
z1wazOZbN_KNelz^W4u5EDGLf2uJzE&)ot#%zk9Q^1pTZopWbAXZfavAB)xCQ<D^2w
zB?$HR^9y~if;Zusu-pCncaq0VR>>DTJ4b+Zt|MhP|6l<(%24`1GSByK-vIu)Wzu_!
zN|d+0tf02|#AWaDh{Swy0hFa(Rxfti@6x=Zr^?ve+{6$T{zF0ntpH|sy4+Y43I#i#
zo=+g_4Y*a#++K3=$VQ2=x}Cebp-v8jMgvb>xl#X&zQtU&Z0g7KbaXm8gNc$7CrNQo
zeCx1dq^v`quRM(M@?`?A%kdn$rO6y;+dwjD5`RCV6kt*xoYV5HvMhM|)NJPIG$Ff9
z4E=Zt$6&)(z4y?l_;`FA91*BGI%%MkwCNblyk1)jOJA{a0k56i?z{Kqbp*g(SUT-8
z^4WMJqNK4~`YWs8U}J-f@&Ov)YLlVLsF_j|Wn+VsRab|Fg)uNNsD=f*W8jNJD=Y~C
zq<xJimXdNX?e;PwVk0&wDagnOhlHdxA|mj;>Us*Fq-OO<N09yYYZsSNYUwhx7wXYb
zV?_?;+hgL2qSVVvLXdpdt1{<~Krwob>;4`S7f@+7UeLKWTMqc!T~2AXjtq)Rna;L_
z%Z*2!7Ka0{3RqZfOiZT{5ggvrkYzNarSbCd1qMPOk<0p?8w1cR{!efPIrp(CR;A9R
zoSd9@%N_Y>#C95$J<Sf1b)AJ3O7qk4*JlQP{v$)_6Hq-pJzNHzx%2`z%M*Z29tp-%
z`CaTZY!2zssHxwZgIxU)4nQg2TUOQ{O1v_VB3<u!mtDoaac&8bn5qoIXO|KcrCuCn
zIyh)H_bQ?0u;MHzDCgr1)Ts7!*-<f4Qi=|j|K2SYAqEp;;jGyU#$V^>zs;3htSner
z_8A{5qu0*Z-8~kMJl+^6Qpds}YDlYyiecEe*e}AsQdHOI(J$W$Vpu=ED(?z=D3DHc
zdxoSkx!>P2C`&`z$L~@`#u3}QQaUz9$*c3HHx{a-rL_zFd{g&F?@z>=L=H;@-KHeK
z`LZ{=b!l5?Ck)u3+m?p*W>>dNcemcU+>p@un_ip@A|`T#x!q1tGaVHN`~e)bl#lw0
zFCt=$;1d`aI5(#yARzFVbWB&|<YDe@)UtAPc|g4X$=tkqZ6YzQlG|R+>x!A4z8H9I
zH@3|*szYIjNu2Vv9&QdE(!SA?B?VyY&(3rk_|@m@?Y9=n48hvIsi7j0lCoQBV2nqg
zQ)mB{ma;v(EuGIThB>&TgaLy|zf@U6qky6LGb+hkK}E$vkhQ-aj&!Q$-`rGK9xtsF
zWbd_ZRbgt5QCh$Y;~PgI5?O8&Rhri%#C&y$vAD*+;UF2~1=QF$U~0b+hN_YBZ70Wz
zL-$rz&fMKi7I<+z?k;O;8cLeD2UfQB=OsHj>>v1D+yFYyZq$^*#wrK&idFvp+1YEt
z9(VNL(W$Lxpl;_w*3IGBYXz8=v)ieB=BKmP&~{F4DkS(<@|WIEu7YqJcwNda`@`FA
zF7~zQFo$O<&8(~poepCB{A{(gV6e_GYGdQ=8Rfy=osD+r@mXb*n4O(nKG=~Qo=-^2
z8smv1>xn5w!ZgD0GfBbX9hH>S&PcH_xbgXO)M8`7cy3ttaOvRyF$6QH;~Ii7PJmu$
zWls3zWPNWmmj?&iR<(rW=5Xm%Y0X(6_UbA$G<24k<$##)`Vh>rz~{)LW1|Ka2fEH%
z3Y|-B!dO(B>l;`YJDnkpCmUGd3%~Sye*d0XEHLW|%?Q8>$g9(?Z{=|+qt~L@oMIYZ
z;CGoUUS8HO7*Qxn;hU|kiHeRLUF(e|=8@PYgy`%ZJd6zNJBE?GDJ~Yx$Y|~9!P}ao
z?9pjX!RLyQk`8HXkYZ+Gv6`b{XUPN@NK+$){(a*4&4C}EYkOacG+Q|$BI3ft!~pHP
zKtB7tnVAGw4>7SpC^1C_Y)CGqw3I6-nuL%L2a}nceZt_%E`d9GcNf-`d9ne}sz!Yb
z4g(b`>c}|T#qmggcOK!+Kyp~Xj8uvg)XdC``(#~S?D-pN9ul5TQi1#H`Fh}QX#0-N
z!Nl|e4UL_NDSQbDDeZ#=dx2tZ_4m(M;Y~q7j-CMHB5iy&98UGZ3=)Rs?9CRSleiDI
zF1PH+tLo*->^azyLz0aLX{n_#9d}}E_mk|Em#Qp-f&#foc*SyZvT{V)zPklM!Ny3E
z(=%yn*w?}W#nNJVFy&3g7&5NiOa(b)WMYDu+4cAwHDi1{+~dY7EQFYpND8`}M<dO_
z0Re!~;ji~V64zJ(ZtoD|Tizs?r)JabLw-JrnMfxdftk8S=k=9?q=Y`m^H!FVQ&QWE
zj23n}7>Q#_Q7uN%sb{BP8MR&!S=ksg9fl<E)k~1!33%RmsQYiRQY%aKfvq|4_pI5~
zPKE?G?JX_>G6mO{4!bm#YeOfeh0oE@_#LInGg`l_i1hM%)QE@(gb`g$w+A#eH9h_M
zVP{fTvp+#NCg%DX(y@Vo3lr0|$8|&L&x5AS0@a_=gd>x(FVQzdL<?>%<5ZLyq$KpS
zF1TLBPtDln=Zn!6H@UEMf+a;X*Sn3$^R}kHJ3PV4DuY@UMnl0CeY1XuI-f4&_wnQW
z$cXco^`YqZW98I*ZX%f?!^<5NpPrbv`0)ArudXTI9vimSevyi$x1^FDaXuB>+t=>h
z8by|t9;MgDjE><x-4%z@KSO+8R{qlnS|K;}j&;m+ni-<7GoyUE8UW@|=yxRJmZ$xs
zUQ$)n8)5iDv^cT7jn<!j)Vp0=ym+BriTHVf2K6}^de{*XnniK`w(Y&U%49LV+XW`Z
z+Q#hnZxwfEs*vUQgyVt&N$bU8;Dv&j9O%@!M6x;VcBgtCTTd46o=vB>Pjb{NQh$87
zA17qbtF6VrSnMsh+*XE2G~4;vI@0e=GI-8b0XN09oAXdI5AcB%kX2M}_2HM5W&DZ!
zGZ_n;D$>|O_OnKLI6o@_ehsT;Pc~$&dPlyoCO%-YPL8Fjt&Q>4fp44$jIf)hCY;1(
z`-=#w4i3}VsZAvnZerfN#xK-cn`5*GHtSNcLAa))IhW6nU}C@AT^8Hhrzk{OdwN4=
zsyT?b6mDI=<*a3jnxT{YsN?oH$HxdyNl6IlM?noKi=>t|>LY{9A2jXY0YJe;Ez3MI
zLl<sx_&0dIeg|IhdZ(nCg+-H^QZ1l^0Z?x!C`jqVQX3cm)*u+AjEwNbK1TRSI7JS%
zj9iib`{a(GGY^mNIM~C%_zXowqcOBUnb`qL7Xfz^m#6#B`WqR(#?qCplJbhIZ}3Z(
zN5Fr<V$`>(y7NPozE5-w99#mnj~1&U#k$Q+o_DVPdj2T*Bjv`j0np>n*5#q06(b{2
zTG~L_<OuqoFjzM}dttt63DWwy8!_>UmKNR?&hm0`g_XxdU$Dz<&4RRF7%|~5i)kT>
z%%0m@V}E}qr@fbZ6GcXQ_>z)GbG28EZV%jFsM;8t_*}E>YB2*=?;G8YVq#F*zL5x!
z>@Od8;<UBO#3xcaZ2n%|;CTHI^hOfRanH4Tob4b?0M9>QXxt;e;C;B*Q{w@5GxME|
zbt*cl*DoK~IgC^Iqwn8x+3JT?R-T?kYF4MrPCX(kw+Bx^r~ZlmA*?^qyh_0B+yWOj
zwz2s(V)^gkqAcq3Pv?6EcvSs8JqIJ%2_2JzNde{6#&|i-b2a_XkS1uG&Gj+2)`4kH
z-GP6&sj{p^LkpJcW|WetcjWPUXuLc!cp)EJ!qCQuy|tBoeGNJ!%TYZ?%%gCy;E|qv
z);^bn5kxpc$d=L9hnbP7>(zG<95$F*???<N?sL4MkfYPn&&Ap&yR)%+angzG4tq0a
zjfn`CHPC_rr0MDL`gi*ajWj}SM4(ouNVB=dmO@yQC(Tk5P$_nF=piDaw5=i_-j7yU
z&R4&Hk6u{7cjs5s?q@`y2!|5l8|LKjTaHpuu_|dmlOnN3b3zuqpJ1jRj@S3}Fwa!+
zIb1TfF{1PG`FdV&1*9y@BH4Au&~;s(26&#+<|r4`rM|~(92$a|Oor;kHG|@sKBEJ`
z{gn_6ywSkHtcRDQ+w2Jm?*^_-kifvapI>LxpMXtDy4W4U3dJ_>199?+i3up+e^XRZ
z>C`V>kdK?MEmt&LlarR#CMDg%lr}bwe6KntBv`jlJNjpMxOGu?1|AAxg3zk*@zqVW
zgPgIOQ#}FqJSt36JOY*Yg!NSJ4o=Ro!bT>GDFe|iyl2mP$%SBI<z5f)e0+dr=<h0J
zULKkMThGHZM;8c0b@HQDF@Wv(1njSySJwM|%gUu!mV^CKDKdz(=j|00ae@rlK$VAn
z#(i$8M38OmoHx5A<(G9iHN5#6`AR*Zz8!OMLRD2qTPAGW*_yiT%}w`>4UUUhYoJI5
zoxv+zVPVPKYz2kES(&9Hqo(uR+>XKfi-Xt!+)i5xiZz4X;lfu}71p};X8C&!E^NJD
z`uYk}c$r*xwjI2yFNpd0!2Ot*BX~Kn_4QV<^w{j|R607)EK|L7gU|LGefW_TtSc?P
zbbP!JXWsry(XD49MlXtru(F?(BO@1t5*x^*+-`u4ESu`)BW<j-G+m{K8<wIX<oM|5
zs%oMd(FFe~D(ZW}xbdzgr-K_AnQenkC9uqK+WdVh+Lgrb0fQA46m+ZzAI!x!8cJ?W
zl{zfu<mA{c`CJWWQhE#_2PyBa$J7{n&#D+p7ezP4s0t9TU;WI+exn{n(#{BNx0pT8
zO_+%TFh|T_j<eLhp;r^B?2q@8+tpR$ddGl8tMyeI2#?bM30oixWqzKT5D)KsPqSQs
z7H4N?2P+&<KPf1HT@7y6xZ6SAh+Ic|5{RD~8VWz^z(XTnGkHP+v9wh1M!(yQYHL-c
z!^PP+D~sD1M^JWpf%)UH5QKR<jKoZ>$J6Ed4h|5^Ds$tbzh84II1*s6iQB`r=4VfH
zx;h=de4&7JaedL>=p%=Cc-*0&eb+YtS-J!c+=0%#LWG28FR&J{o9pU2MzVFx%szwf
zGw|`f7YU_=B=UO{pIm8!utb6nsETiEen|1W@_YL9K817m?5rUgY`fn;O$i79T{&!<
z_5O5!eimqbd;4l#DB72OqUflhzIf63d5&sRA%1>zmj?FAnIRBm&Y-2Wn=G8xv%LF5
zojOKaVlqju4k`8A@yP)JhZ{B8I)|Y3KCYtlVc;i%EQ><NSnK2a5-%+J6TrmyY%L4*
zjf|Wfmnc3Mave&$T<HoB15xXltSk_Y9v&Wc@I*=GcHjn<4sgH1x?5V7MkjHZ;OP7}
z>kqD*i|rbU7MI$3Lqcx)Q)Qt59PZQ_h1l1MfBJNbF0eUYJ8FA>>$;Hh6S(cF2Igw3
z>Yb<);u_=;mTek}b-m<NR2VG@=yZ|@U&%2W1=rPaV@%7WTH%uDDCrvdPX1I(EMs>3
z@F7uyx|l%bJDP)fsSj*-cUNY8L-HUxqO77~ccGC712|^EzL?+hc4~b5ESX1ScvT52
zHj>J;CxzKi!p!Vl8zVSZK@*VQ-Yov|xZ#-zH1AF1W&_>_r`KXjb_^iPn^Ed=rV^5I
zs#$GTf*k-A4I}Zm(f%?wN7?B1T}{2Q!LxCnTR?z{;ebysI+msR7UH3<rlzK%Lgk?i
zNb2jAKtj(Ca-nv8rZz?p1d@@E$X9o0kWS>Q)Brr>XGvVjmNSebBn;>zM2w$)gM)`o
zCJN&>_H}Znj*@wDon2qT`*7e`O&^oe6}fWoxE!wnH%(yROOV0is)CCxOn~3fY~>L9
zQCvJKMdzc0L`>sd)7~sHGP00vb@@1_`=w2_)%^Bd7+l_SJ!$ikM8Ni+WML`r8Ck!S
zGpU!4%b;_%V0n4j)5+$!dZR0Fb}%t9vGrQx;a!ezZ|BpE3?0U&kWgSYu3&)>mkX~B
za9#9AfKz9*8(ozN&Ox@vV-$uwyg5-Tu}s;P{ysrm7#vcdWj3x(c<M_3>w@|CJjY4K
z7B{oteIwHEJZoJh`5m1`vl4K49*~CTjkdn=Ms|4k&gf|6-26^4A^+{2o~h~ZPQXg`
z9+v{3>Zz0jZdo@sw+G;s_xJZ#S63gI@cj8hr%_{ey75NH)!8kToi(Y<b_MKAK|nMn
zKAsEM*t+KNWmeOLhJc3)F9)hkMn~J}{oPwbkHoX%<2suqpRZp(KFUSCDA#znw)~e-
zU>tgGW+azG{(joW^tH7}qk*xBiT$gIJ4)?(M~~CZaD!*WSaA8?v%gO?{Lx7l>KF9N
ztpNx94KFW$D)-r*GT_R4!Q=Q1D*Sj%gChb*>u9-j5&f^3{E(2^goM)pd`~bN!Kr#V
zI_7!1+mnM0wsv;M;N*sji-#ZG*y1uTbY&8<I^CDs1E^l)kc^K@xm@u~lo0UWF6(+R
zp->d&yXODA`l<<-dk6mf(F+VpBx`j7n@WJOb|y<o6f?C!cYaP9ED}T{C-3d=KYHOo
z%f(|s?iMk>`=t;+2S2~(_Sny};-2nqJb=r3qTfeGM(%i>RGRlynu7?KLsU#m%J9U(
z<+L#>Hi+$l<Qtj#r*Pe%>R$2oi4e0nAegU6U*R)2TAmZhf-6<C(9@IOxIJ$UHxIcQ
zr_3M?0%S)UYis=_)dnjDe-920mX{$#CC1q~3F9CS_c)v#mCb%n%K*gY9=XP)W%o`t
z1_9?TDj`XnN}?F0>%n>A95_snS~lf2l=}#zzkzcN=aa~vcxxFN8&^m0dl%Rm4W=~O
ztx42hu)Vpr2H|{gf(ylDdtqQ^8TfA$4kjqe_r?V2ws3HAkjErJI0^Uwd^ord0guUp
za5NYoofQu54M_h3<Ur7W&qsBOp1(ro<7)(v=<53V-rCyh`t&Dg&Be1gWe^ka16fuX
znQ}P5Ee~>J0z(Dgf{1LooZ1aOT)q^|YfKE;YkR$D>Q~P}h%oEqy0##nmX1!^(>OTH
zLSjH90sIPTM$~RzwB%d-_I~p9>(|Fuz&e)6h5Ug+3k0G;-+uf49h}~`3+k8T?a3sT
z0B403z0*vE>Ektba9Ro&tjo0T#~&#`_R#-#Utnf$j2=0Y3no-Ok?1cd>eXdD)o5|f
zMNo;TDs4GIzE9>RO%#GBVZTtNRqQ)Dx(;GzwqG1(Z3}rVj8ZNxwIDev<C*+RP@mkP
z5B8gHxnE@iGYrCMOxWKVeyaQvter^|5aR!D`NFpQ94WhN4mNPF6}k*JO=LBjw&zJ1
z#Ou#*(tilPz`!UC;WHW_{M<s7ahx?9`!ppg?i7zdUzO0IqjQQkADKCbtzX)V#CG7d
z=);$zRNQfuluUmIrnM#mJ;~Gcg;!nIR($yKZ<1%d4Y-_JP@JB{NbI88dVX8H+8NH~
zM!V|_-9EqOnQ*hOxE1n$8&K=G@L<|hr(6<lz(+eC0Q=MI^f%0lxgA*P5xYD0EfY7i
zrG|IZ?-UGhx(@JXSIXxH%diIfF7PVeYNc&@C#3#uNk`442<*-3DR={g`W%M%&TXVf
zbO+1{)NW9_t#ky2D%{5v8t9@-+vr|i4@M+Px}mFK{7CWe>fO{~8EE`WpfLp*pC_#S
zTL18ct9W(1BujyYJbH>ys_LMS#CLMRSzZ3sdG98D=0$tJ&##$62x##mY1TKVciZj?
zE`Ls*Db}G+o#q9?po6lO+(a=K*SGR-AgwOSp?Jz<HI3~9p7Ni#+8A5AL#&GU1#}BH
ze<UPwT1k>_U|;xI)6()kxD;6JK=SeZe?He38L;|e`Bgu|oW!SFvEYHNA{niEFGF*D
zL^vXtPbe7iE}Y5;Pkia>&-y$EG;mvYYI&@j15a<WnwZZ^`t>Az?o&6Rq%s5VKby^E
zZ&2D8J7SI1xQl8S3LiG*z6#}RW`&h@O>kng0gv})wTw~uZ2_Kmi*Dvj-?7(8?{dRC
zj(73~F<Jh6G&_6qib1fNy#_=)e<aZti#NqF!ch@FtO`>+&AJ$xiI2mgs!W1(zJ1#{
zBVFyLfb^62^1a*to>TGr<p<QNP{YX<mW_~K*9R0Y@>9R@IvkRibW-VWU69VP(XV=9
zC>{tRbkdQY?V{#O$n<eUK}i<QcLp*!R(b>SzG`Qs#jSnAsN65NS%)*g!3b9|XKBn2
zl*n0l6g|Gkry-Atxg~ZY2t?;SOXhZ<#ESq$%n{A4)w{->Nz0oDv_TvUpfR%V!s9wz
zrmql-q_XBkEFx9>ucm7;5xU0u;+9ueW64D?{MVy%Z@f5u&5lKGVPbBqzs9R%j=4vy
zp~^v|b8ohxo9_xG-IyOv%b^Z_JNT!F(aaDskKm4_Sk$oVZtH=px!~#IX8nVckE!-=
zXDU)Ec5G*|&QSeiL;-?SV|||A9%^XOVBJBNT3J9`8EoD>6ebY=x3l2laU}c=xt*_M
z6}fF-Vn1xm&IY0|QqM=p<R$b+%2{rTjbqjpZd{bA<?AzqIZJxvgNVA0aB>DnF{9n=
zkb(qE=Fe<q7l?zW+WYpHTdul12Hl$|{PWhim{6F4q2uSs7($gug<vCSYNFTM#eQ-{
zXCl~@a$ZLpAyrO1<#2Lj=*zDf-AFF>;?iX}Bl)dIb<*OHXe9g@De*hIZdlpDz83pK
z*9}_1zysIJ4`IP-L)`s`-F^K|3XK`-;*;G)tHFh0^&8@ME;`YK(RoZ?q%`oc*VZKy
zlPIsa{6um(-yL<B50}~PjJCj}q3z7uGxOZkbc(G^rj|-^k<@(sfhS*rP&9u|dFmZW
zOIIF$(H)0-e{A<YL@I>qj7}-wu65*#&4FP_F9f1ry)`kuYIrRv^5z^d(ru;wt4EB;
z>D5Zhov$J>8!Qi~#2YmHoYj_zaK2omslw)y%ef4zI$GdTayqhSTEPIGqVfJxPQJ{i
z8_HMm0e9V1+v7q!tc+S8pvzflvk3Ei+()m8l}_>->hey8ESVq>NY14T)ZY@x4Q=vI
zqvr5l8g%RV*zZ?GI{WPJ&Y{k1s$uj;U#q|1L*6yIos<e4q#zc$;=bb^G5WnqoVv@p
zC59sen65m}8>X6k*+j9=MgBx%e<^k)Lg&BRBkP^K5Eyr}&ub-DB(6FN@>riexGo}H
z`R!FHW$z_z9X6h>423G;V~2m*pOPVbY;s7BMaoU9wZ-*=`8zL>ji^X3Gmjk}LDgY$
zHkSUT$gX?0-5S~QoU^~eS0rI6HE!SezN@vx&|kZpMonjk$AqEZ?VWmff6wPbO&NDx
zu2!t_I@(&E3s0y%7r<*eqo$U*Lso|CHR&F^-!TqOz~LDdKm0p3o!>$2E~QSJ({X^>
zl3X5}U(t1GdB6V>tPQBV>+XCbbxoN<MIEHft%oOG##?m=4^{7?U@*09?VQ5<v9nR}
z1YfE!x@5K?GPb|nmKV~Kb2aTMS10*%zwnf|aPlOO&rZ;LCme{`dq({q?vLqYg(VKh
z3DF}Ea<spPZW(93(e93-g1^#s)6kzWZ_i7?q1&TWM5nqKxW`)a#YL-ho!6jz3#;?I
zG7A1>**r|4CB=E5Hg()k$q^S(ljfZm#-eUHFB-aSdcTb{+<GC>v10I0i^q_El~}ZB
zew{&XISq83?)C>pSJI;0c6=HY-LJ7^fjc8DH#B8qzB)%b&MIhg<b{c~XrpQo6_|p4
z+mArSJIq_+^bxnX)BXolK&jCGKEC-sedt54t;POk$i=q<y3W^EMVIcg2J$RWcsOPu
zeMA#31=4x9aB3H?+U-Kv&{Qir&2zYZ^Z@*{Yz%6mgq!}-9j`tT_D_dX>O1sr#CL|I
zK29}R=Y)|8+$9R^ufxn=Hltj7Y*omLNXK17>Iz0v?u`kiR&oU4L#Q^EIj~d^R^RWM
z!{xoBH1dw|e=CVLf3enc!xv+Hr}uQDFStZzoryAZn%!~Cx|Yv~E@^6g;Udvm6$^vu
z&4W9eO-}Pt+?!r1^*QywjYFvd_7*#%l3p(hXzq>odS;W|5%Jz$&%)&q6=wj^a{WcU
zJr{V`E|JZ+O1)*KzRHH;)gOcf2h^9)<K>BuNnajz0=p1M|6to=wl=D~qFLTOO3pL1
zR^sK!SLqljdGMxu8DW^hasG?4>M8CMIX(b8>0Zswwdf?mU>%Apm_n3d;(fM6YZ)J*
z&>8Kz%8gFU07d=fcPZ4;1kS<2P^pjh*)k#|2E49!5sTdJ)*dtq+LGamPd+GvtjUTG
zwCp$iR_{o3HH#{0qR1(5Fcg*5J6>T(Y733pw_i|moG&-J62_Kv;ZSSW{jkW3x0!iZ
zyBzEvnnf6zvl)Am_f+SiX+9&A)Kx(Sx(1v`A+Innx+dC^iS4<3m{^&aSNX5fQ!h$}
zmn$fGtIP}QY#k+T6Aqre#1}?FOLLVoK$JtUIQgl`ejP@#d>j^}uJ_Sg)kzx%<L8BT
z?G9FIe6jW{muU`Evh$#M?7B4~MEDDXwDn}uM+W9$fD+@xp&ylQyS~mhFF2PhPVl=6
zBcZ2hsG;~u>VvC>Uvn(}UwHigRA2vhX2v~fwl<=RLS{25_&l7rune?V@RQH~1wiQg
AVgLXD

literal 31199
zcmeFZWmJ@3*glGapmcXRbVzq94Bg$`-3UksNSE}`(hUOA0wN+cfOH9nl(e+;+4Fn<
z@A-DtI$zGW!&)pB@XRy&+53*`zVGXr7<E-SEOc^oBqSs(1$n3@64K*nB&0_lpFReE
zDcwc>2!1?smpAZ2Lc+jDd_S684!%V~qD4}GzS8!~-CurfZgk!y@b`#paVcd}oWM@#
z8U6F{cWds?)W6y@SO4HqR4R<`j=yN>%A6!wlu~+Q^rtOtg&2MLM;giF%(D2j_pk72
z-{TI%G=1JEv`}%%_gaxmEFEkqna`kO+qk-Api$VS%=sX<x@t|Y00FCE#@_c{1wWA}
z!_85_uaKv<NJwuXVgLUx|6lIGSG7wl<@k}!#Ws(1@6{(rNV${x8C|z0BSh%^F+x`g
z#BjH+^I6fO&rO#7u{+(#sb9uJ?oV@EJ7R1mm>(l?_2<7iY>&Vo8G~`JLvwHGF>E1*
zpX?{j=gcIRUwh#DZDrYYpOtbR-ws_CkRp8;JeY4-|CUF&Nr;5>67LlkG7=IY9t!XO
zyI)S9c@HsUD*rjB2|Al7l8ICOF^W`jc6q6G`NI0g)D-V>E0Q`PE}ZFy<ipkYPWKa`
zQsKNY$=fxYV(b91{1MKcK*K|(R`4uSJKn07CP>PXD5uu9`*xB}_V!%l33mDZTlaT2
zo8ci>BWBsz*~23vZasGw^7Nipo0&}bF;6gvoqx6+=X#ALn>7X9TuiUWGd+SyiHnQZ
z_uOAsgd4TGz8Jg3i0Vt0x9&LljY3XI`LoGx9PDZ?3ND;rc4Fc<!7b40?)qG{+zYH<
z@@_{h#(uRs=<=u6`1SeT>!X#s2|1d(lgyq#@Q(iZt4(%Enw6B>3P@$TLX;Obcg5JQ
z%N;Y@rz#_N8^qXAeQSTcSNorx^`g<lyL>?+bt3l_`s*jgnyy$rl^2~_Z`69I*K@5S
zVJEa6&l~(VoMyxCm5hvxoLtnh>!qI<dV~~*M1IKq)p*dMm)92jdC}_qMXPDO;H3^v
zK8m?xMBmy=TCUw+iBzIn@GOF)0kt2AKB>epP;_<f?bEnb_dcwihr1)DAcvk+8uSRp
zJ6|CmLI{krbH9<q)bCZTE{%xypN_q{rFJi-0T1w}HQsSCEnQuK=DCiOVYd3uC&QKE
z+sy1E5*PDU<B|`7{{HQz^_d|Nhlht?uLn8pU0ug6Ochc*bHQp7u(tHs8J|k|_%si!
zv$TZxL{!2a?p7ayMzi@zjNN<bS8l*QY<PjiNb>Qejg$Q%)QB$I*E98(!9{^|9JDOy
z))_BMOiawo@Q2*rb{=+ABk2`*a(@2%^?L%v>x0E+tARw$z%`7M-o{3rsJ`ct!4G$r
z4;QUHr9Co)GFPvRjEoXfk(6s2dT+rSv73HwD=GPQ;H!4f94N6Gii~z5EN#KM1=jAf
zH>+~kB~idho}hX?Spl=iMPD4ay}vnn_%|R~S6f?KQQ>zsAt&gvs6AM%M%23GB9N0H
zTRgsfb6y`p71tmAhqvGgeiZU>BOQg^zI+HAth}j7E7*gNoS!_Q)AwL^cQ+Oa85cWM
z{+|T}!E*%NB=%~WYILgy*0UC!5dQ0b0U@&en+ZI->&@%k&Xad2k3$0wxp1F;u#7xr
zR_7WyWu%~>(2AK^S#jmbLiaka^y$MA{K#&krRB9#ni;&wX137bvlqwN-R5`v`zZLb
z6S2#=!d}2LHom`A;_UJUR?%cT3Qo8aY^_7<()MUh<~se<wNo#XIz2v_4!7HKhYt1@
zOFgXBd4Z?U3pgQX@NwLZ93)x(QFfz@hH<;+=4`DI)3vNZN?5PCL~nd7+4oA^AozCv
z8&@TzxVgPe|NFzvqu}!yJwdmXz|C|$iN9ZvpR`@4@n$LSXsOrr+y`9szG(gFIr5^l
zq=zTHIY}HCQ<Atqe(L^W^B(jkZ*KP2HFqx3H2N3|(yQpD0}eJeHWrqwlh~_19LY}$
z&g&s}$#im>W>Yp-dhiB)_I3Jbjk>nqu?W+)D+s!Hr%H_;0o8!>!Zf)1{0NIIgKrL6
zJ3BgZRkvGQmby>#j&SkDPu0ZFr<Y%^n!K$>&Tag<#x>w!RsBi3pAr+byd*O9RDu*U
z@!<s(6;&}uo?ZXvkcayvw;u6JCsZ`F7sMm6L~JWTm&-BvX0C5u_G3O#Z<DZXHWarT
z%e5KFz?;aj%Wrp_sluWVAl%<YV#*YMbAgGCz2%?cE@W(In7wP8K_6pd08!L|_uMQW
z-Cxi3Z21#G78`9w4D9mw3tQvSx9s_{?Zg)s7e#zs=b5{@x`>E~*hdx*E@JNePjR+h
z(yBXaO@9zl=@YXxaB|vPY<BVrd^ILSclCFzZ}ob%RrPP}Bjuhb>VQ8=X4jMD66ckH
zs`%JZ2flAu>pq*-FF2Qdifu10E-os9JajIDH!cUU4dI+r4@M%QqEq3*l-LiVBaDfe
zHNyL=Jr8Y9S}EaBZ8Niwvp*Vf{XF-lY6@I?=hq57RH{@{#BesfI{00P#&@!z%au1r
zF`rr97STTaY6LP-)_5S}_$hdz(9qDyNll2NN~TD_*`GA_^HXr8`0d|5$^i!aXZbX9
z<%%HA=~k=Uk;XtdezgVpp9QKh!}fM}MMOjxKiMhM;&L_a&zs-h9fkA+obMh7_Tb5A
zxHcdSVm&C9g^q8hO478tt#*6wOgeK?Y;S2-HuW-nAIlTPLY2NpF5fg)<+l;d8(TA{
zrc9&o6crIUG0UGi@MR-JPgm5@)a1Fput*xv($q`|u4c?9VVqPFAD$?dBZz?>&IwT_
zPF-JD)T`COn7r>_(tZqG|BiZ!$XNvsx7%c^C!3rhU^nUR<Q)mAUt&MqVi71jN81|9
zyBRUN=UHl+&)rQszuv1idvnlyh1TR?{;Q?yjTX_VJD1z)n8};k&s#}TJ8(%DB1?i~
zml_Si!pJ71s8}v3(XPlnKv4#QCqAO+AD@7avewoxx3skMDM+-|tN>vr9pIUdb0u+~
zYI7@l!$0m878ZgK%3#&7GK2U^4~Ew2hBON9zGcyfHun{xylG<oGE@GO2E?}Nx(>&o
z6b8oB9grd&+}xOok|eJGd_Em}^D7i><2e^1nSPDHU}I&i>$+H+{+u>I6Y{yNtgNu`
z`ONE;l@*SVB8`jDO3Q|Mp>fm#QpDwU5~o?W;aQhAN>41jxgLI@5hmbI{P>|BATBPv
zyu3^k{BO-B(?s^}F0M%9D*vc`JD*C_cf(JQnQ(H)dxwPxJE|G_+nNnn%cIU^iG1@}
z9PG&)w_%k9=qC`RY3~0<t+zvgB`9bH!Y^UqSt;B;bFUu~y0*TVo}DwSB}Q&qYeUC_
zhukc!X0RLZfW)C=Mfhu%=_?40-RHG#?Z$N&9Bscq3{1s2{MkNb&EvAvGJ4=!Z_;@|
z((_N|_3PIzE(fE6tIE&^;F!Vx`e<C|KU$3Eiz8rF1;N&QE-UU+<uhk@&C{DZ%7K~L
z*)1Blv5+$<89X6aa>rZSbWBmx&Tjkm>h!b{CMqg=)OjS~k4@ioGTbN~rD-)R;QAA`
zo&Z9N56Fz{If|9?^hFFsL#mmJJs}VK3r$Hkjo|5@jBDv-_B&@7ErU$Qpm$~E-?bvR
z;zxIGMT`4`0skv(%^_KSqh-l;gcQz^SzTYR7R>IJUpTeHl}?sejT_$6eLYhzASk%{
z=BO)TKJM(Gb*FjmM5HgC%zLl4-K-JBcYpK6klTR}vHO4Po;fhh%Fhq?HwGC&AeUrt
zn6Pr1xsQuqPS4JwJ`M$2zwEo@a&k|MBk}JSh&EGdFe|Srg<LFEM*yP4w{yM9tc9I?
z?d2)rN-w#`7aAji&*-iXU$dd2UK}h50;~cuqsrmN>z^J`O9$QijU$n*q^lh>x$^s-
zoiWiCu-7a8OamyfV`F1ivg^akozjKxXB-)d+=Xh0WwM>^4_13Z#&U&?TAURm|A7GT
zkh6NrIN(u9rwQq0PM`j%-1&M$&i7z3{on`@tC{lsEYEjmPLec|qEhKCl##>f6;Q)N
zZ5A4=I)A#pW`!Y#zU?&!k$YX^gT_L&Zf*7tq`{#a!K9Pi_vQjLmki)&PLT7rot&KB
z1O%J~bcso(M0a#_Al{*SL7R(oe{avI(K;ruq_EIp`k9xrDt}B<uR?88Q&V9fz){;x
zW?{;C@)c7T4ela&hDlLdqs1pDkuQIR<46p)5g3Y5ha9$V-`{Rah8#GVjc}Ve0sA@v
zj@M9NWOP1Jv2*~iBXHi}9?1a_V_+&x>55&nC7P_1*wy$k3TR$NMoV@eV$0CA8U-N$
z`$>9z2@W?~LVrbf{+MLJN+7W7tHAEW#6)Mooy(gkz29-2#!$W%%M1f{;`%uwP0h*0
z#cM=i0IGo)*<#{21|oQHPlyJ4i}$6gHr(C#3A%0mb^$;#QpQ8J#W6J)Lc*1$C{JGf
z+1X*?<itS@ujCxDv9ZxJ_UNw<xpRUj(&LNz?pOTxC7<w^s?+~x#b{+BrLgCWj-JXj
zaF@NkJy8I8{xr`GpHMTD;Gsy6D&uUeJ>eK0sZOFac<4nFWKOqBau{6h4!Qwu?Ex@E
zYip|=1_Gi11Z5@pM$EdYvZK?4BlD+n?S#GW;m<6WnPgxyA#{_B_=(zDzCtzf^z%da
z!#9^l1r+Eei%U6%A$-JfRN?{N?(SSf@X(hZLTxJR630Z7T)W%58$1A*Cr@BpBBAYB
zU7ebmT3zKlN@Bsn)+{sqI~6E_9<fom#=yW3g+qhDmNA2yo+gGyOH)fZB1Wb32ERQJ
zJ4Ko<PPWwQ76K{&8I8dGh5=SKHUQM73c5B@luQG5N?)%A2@=PNpZ$<NN%;k0^QM-X
zxX-UIq4M-6(^_f?<fSS9>q~tKe+MMThOG|YE7LtSrp1+S>5ff)TCGcfm!7G|1+O#d
zWX&Br;cl3on(_h$G?p(eDreO3S!t%Bpw*?Ah>cLObm>4M;LJWK=+1vTFS9rYk>uGa
zXG==8;YW7#F<H)BH-Qq%h0-0L9Pq$X5u-fs(i>cJhig_>wq(@ko?PuZIf50Y133Rz
zjB~u{`eyYZpVd8P@aS?po1aaGPsx-MFM&!h*ujlc{L<JF+A+1We{jG6X<fQwi0Vt5
zO>?m!&7%d7bzO=}?Yj(#$(iKCsibdb5U8R;?)J>&HZ4_DM%_SsHgat*t*HTRi`Dej
z0A~k>tHrr)-}|fW@jTJX6eSL_<|^sL(aPcBVa5Ofp~~~~^YqNjsqUZ)0Hp6;@`t@-
z)e#eJoyU*UFd)Oz@NYLETy|{@HvdAeP+a%hU8t(6%2~kZx`J7QFyth)pd|PwKsnJW
zv@I^h>1;$|N4}k#u~+*|<13&2y;>;t4i5O+jGY&MaX!bz^*8nZ95cA^kzJ|twG#si
zzst^-P-NRCVR6`-H*faHA~%97A|q|P?)yp1jN9BT9H|hgskFZE;POVV*3hShe`#BZ
z6w@l)fbHp7=jwv9z!kk}xt8mlXS?yNehodfTHYumb`koRh?kNi%pbI6l;-Nqjy&62
zT$fps=0)z#rlz7=On~8l$|F2H9Mq}-RnphpH-`u;<he0;(7GDDF^C5d0YxJ40T#(O
z083L!1kUexOMn-9md}A2V18z%RGW3cU*Mm$d*`xlTm&{%)mod_<=e@pk7uR_T|TDJ
zr+-_oH0OxvcXV>fDwcb}F6wCRtEQ%=qcc-c`y4s+%jvje!L`Tb;c}%14iqZt`(v^+
zcY;&_wW-&9umM|7jM^?BAd)>pTan@R9@hIIe_CK@!d|;nQ{zo%iiIYLi<_HShquFZ
z7!v}&9{+Ra8qc0#ehMBcj`)S*T^V*1U65+i>XDm+gMtsyaiRKGSDG0N4uuq5T>FWl
z_VDl%1!Im(F`3~_&#bR&DSpN&^jFShP$`0>;{U7*nUF*w1(t|y`0;}>;VC7o77wMY
zOqQPy3Y93Rmo5Bo;t44_@zfGB*3*-_yCaSyBZ6Z`6@yZ7Mu6?~DzI{5S)xF#evXFX
zAn*@vzh%i3w{IsmacX#0w0=f9uLEJhxE)WHJT85$KuV2$|4Uj|p32EEgOn0yHvgC+
z*nQf-dV?W5eJtiknI3#3<AqW^?B?dCwh(t_79VdkTZatOkuQ5;W(LEG`$~|Cl$4YR
zuA-t+88dhdmHN1pmE1rc10}?i28nO6#ig=re{ymX;Hu1#QVf(TKht~9Uzayn4KNXM
z8LmE$=P;&yV5_RN=#Sl{KDA6<ZD)R87Y+jv$83!B%?@AK%bA&LXjpWvs$GJnurYUL
z#xAf>Jsl012n|BHCte{TJ<G0otm((ryzF>aclXmHpb~1a>_=bh&I`Wa?KqO?I1sq*
z=Zuba7SIyQH~cqc&Sm}cu(Gz+C0L7~QoG2MM(sXKiuOHMj!xM7fCaPrlT1Vpf-IyS
z5zkHw9s+mek_LF=wksQw*W0wYXe0tb4cfm$<Yi`7Ae$mz{2b!wIC*~dGhnaww#8-p
zaOH)aqk@9MHSt7n&#FD&aCmEse5Sno<|cg5_AS_l3>$~u$4yIlj>8fMpQb3W<vw&8
zP1{!qsUEgsqde}gPfMrgv-xjhYU=6^;k@Iyhn;Uzjhv5GepJ^digS$nmd@%LHbSM-
zidk+z3=4Mi{N^SL%#%q$TPqcP3q6vo4koNxKM{DDVPwB1)kPbRZmU5koLKyD`LI<w
zx^RATR~hRM(gOxD`-g{rIINuIBOJwZX609>F<_UxAkSkVO;%_TeIRwNPc8tfR#H@|
zY|-T*eYb&8qAt(QqTn_he0M(AWqJfUCp;IJ0HO{gQh}6Put|f4hGr*I?7ir~K2o+j
zVb5m>nk-e8x+S)<{QMKW;c8;XC|#HN+|-$N{#m283MraquOA2Z#H<r@7&E`-bs~6H
zR^>R|b__PEee-R@QXe)N9@f<zMs+oaQ0pcp6T)7@(HrIfMC{5#`AF5d5j*v$tT(^s
zcC&AQn&M5M--ezO8^7SGSV6I+uSj0|0P(v%i#&P_pXTRe@X{h7V{QU&Os(JNlkhC`
zjM0y*cFfS%%lvG$E9X~%h3fLA7E62QhBJ=8g*vv#MUKQ2v4TG>u=<%6PyBee>+ycL
zEB<a;x(~8qzOE7$R{T&JH9Tb|+myStxmo<-_H^~({&FA#_}@{o9Q75bo`ku$UP7sp
z%Ae!D4>zcT0rY2DPjfy>`*WsL`QTCMRI+h<_d~Y;_3L(HPvl_WM~*&DMz32B_RP#a
zn)iaTL>U_J%DlRI0JTX>Ev3FXH6S4R4`0F2$>oZd5asle5(Y`ew{=gew$-C;;ly=+
zMBaXC`__RrQcAh(2Nm{Kv0P<${*+@Eot~lUWk^S-C=TVIM+`vb)}*X2RCM42s3CaC
zx=L>@HLd8zPj`r#fcNXOpB2+g^pj-bmkx{eta2)(<R+Wex1RjbFl_fcV{mCfk0`3E
zON@+UCO27FEoiV@X>~ILdn}FYB%m~;q*RD&)l_;71JztkcD6F(6ho#lkE1vz<m4(4
zAgfHX;C3<J$@OFH4r7F&j7LR9ZnTc)3O60^nc<Z33zMRuwV!1rC^E<tZ+LpJrc=T(
zw5y(g5aVhqL;D3ZIQJJ9!_jaU@t=j${^jTYIk;dWU5Gp40WER1R~N~HYEvgH-UUbX
z^$O?hZrQpR_QO3Wi**N1=Do;O@&4PKh`wRvIPJ^J%goI5WO}?2O&J-n+7Vt8AxhI1
zPuE<QJdGig*iGz8#BmY!_V$B2-sBKG#RFVJQ`Myw4e7oLM7in3P8@LK(y^WInW2*K
z03HoAF@sCz_L_rlxtl-5cUwC+u=MfGPQRCWC3#~o(s^WyX<IN}zOz>xi8^LkXx%V|
ziPhlSX-K%4u_3q`czSy}&IHLB6=>N9*3Fn&SyQ9TAuCRuB!T}tM@75R62Oz@*4A-H
zM@^R%Wn|>a*Xc5lKDr)jUfUnDwc4h2>dZe~Tl?19%;)NwWVngo*M7OmWd>AKXgJiE
zsNsx|?lRfr*&|h`Mv|O*bxd?^WrHkeew8TVf}4Rn_FX-?zEXpk(31-d`BCCJX{rKz
zIu_|Yb}uFuR*AcQs>9v;?(S~X#Nupn$Zto-joFC?5R{1qDm^(QBq`yxO%WE4L*MqT
zsUM4#MrtyJXRZ628ET)Mo#k>Uc&|lh>)I+&5FcLNR997P$hy_fRWCI~eNX)Kalsin
zN{%%%m#Qj9p72_kE2_^TK~`EWQC&LxtD{ZICLIB`mv3jb{EFLdZR?6*mou3#CD~Hn
zrpI!n`2{xi3HlSf7;=@LdkjU$XxO>7TwK9DB3F6RgfdQIs+DEVkZk_A`!gmGFbj2Y
zMDjqV8b6NzN%%mRaPkKVw`u2j+^3)$`PMWAcH-${XeEl*|I_tVH3Ouos!^s`wY(`Z
zdi^T2t}aDCet0C^(I>=>6R%&Kdd$*d)OOfSlv9X^7K9&Zp_*~N(boF<dI`cJ|I=uB
z9kcOrqQD*Fpr9gItD`@I{t;`j3+xJ^|J6x9hVdY5oJ^6=i<7y@tsD||LEGLZ1$ut2
zt!at>PP>2q^2P~luz8|?SFRMd*k|Fq(OKtKGd?R+ettYWvIJ=d42Pk}*3jJ}rcM)!
zX3rsPJpp_T*eGQm6oRKgNQEu8DS7|O?4OjQx_8HiIj1d<w&Vid7*wZ6gOtB^il-40
z5|TF{ot~ay9lGb+S?!m<e_>^p#UD3YTg*H5OHs+7es0bEPh_O?@-ncVv;skTeD8~e
z>De;#h2Gv*iR<#^S_zn2Z3A?}BRASQ>EA8EMgh>IS8JI>-Td8BTcePflV<mj5n@C8
zB|W1|5Fg1}>qB#6`TLPgk-&e5Nr3|GzzH_j9RzP`P*%gh%*-reNoCGd37SpDJDrss
z7Bl$zuX#vN(86)p!mx+6Kn}gb>DDO5Hr%44GfCXUR22&g`Yj{vTi)2TMO5Fh+XVB>
z4BiOqlx)G1t`)SK>kV_)i-4?nitO|>=EskA_e-*PW1<@jt*pK$94)oh`(HR;|Dm%a
z)p20S-0#~e(uf$MPg+4o4F_2KY|+V@Z`Rw#g<OkUkVZ`3b1ecjE`sF+1EjDuGgZkV
zPj%&PX!|rPiMkYn^Ao+%&CT^StNw*Pi6|aFfAXA?UG6L0>W1RJHFxH-fR<LlbF+Th
zr`7^f3wHG@O?F2sT_eJYpB6QgKG370MTNIFHX7q%Q;B@awON=_<#k)KB|u^2+-7BK
zUS5V;#}5oS>iqF`USX^K%Cs^>0#}o~<0I}MPwI~&DR=WjlZl9dP8Tuvug8C`hH3&p
z!a|s06q8lO>XdOu6p2kktf)jMF7A;em)!D3vqPSX`B;oBLFCoHWza+aOVT4>-xIvQ
zbK>dfOb=PHB}iF+A5&3YUjFT<{5k-l$xK)oYv3-Db9b*IL4(I1DV65YB$UCRQF7ti
zev{R-L}0~w4@OTWx@KylqtKF13)(|w>Uc?~j07W%$@1T}PiM$6H+-k8l4e!|n_k0-
z^&{P3{(qjY?C?Eo2|~Y#<PA@$F+pVONm3FH7DFF(9LtU@xsAB(939mwP^76sr|0Fp
zwp040bgQ|kkGw_FS&1C@1T-)sq%iwn3^DA)B9}jHI$u9n&DEu6vD)_TF@%Rk58Ctb
zC@*yf*|8B$Pis?QI~inXv$C@B%0brL=V9{iXNIe0sy<&bhN^u|<%E%Nn%;ssRRzr7
zHX>qV-;jL|*ogL+tvOrwC`i((R#<T?YUDBEk`onOh>d-eInr(+z3DS&6Q~mCuKvS5
z+CnH}84T58BO})o;Ehi({=}KlLrRxb9fD@CqonsPN9s4GcYCcH6<u2;s59AYtst)q
zy0LKas6^=X;ISv3EbYIY6+w@CPD#=J$`b-9*2ppbqVyuRWM-JCY3aap%Xg}uL`xc3
zPs2jdwIst5Sc9#V4(rGdT)Z9W=iODY(fZ-9&{1QUai22dKg$;H)K1wjJpe)4KkOy(
zNJA4dA$`(i%L02*#nXt2qOdKvL5EvG-x?PwCNZQ@-~M7hQr~$FjmBL}4L!ngqlhDH
zZiF;p=#+)B^R<!y#b~Z@&U$2zdwROb)a1#5TgVb&QtFH%WqICyz%TdRmo&Uz_{@tZ
zf=efJ_Q#YRE^Zw9W7lX{S!){Rqc^*Ly6)~ghfJ)g&(31=3r3|7t?56j*PC{XiQc|a
z;%JAi9miLYC@KC4!-$S<H6(ISGW~ZKoNp6j1Kf0}cxcwm*Ik_;2K?>St-a%y|K$SU
z#)rM<hml1-el)ogEpK8ll;{w9baW&L3Wr7;C=|7?!JHUH7)F;IJ+he{@{h~JmkiHC
zDU*BvPDX*<Ud3v}E(_h-*rKfIrA2)VwT_k@vZd=>=1@2p{Cj^NTq^ad_`)Jp(5%&v
zO@~#9U3PQxll$7^SMEZT0Ls%VSPN)b6~Dk0YIDkjITXJhI<iw?yy7m`jgq|Auc|F2
z<F)<1-bdk|o?e^F4Vn{uxN*rdtb*p&_fwDP5g_#uK|%ml<gY<hGKCc4>%f7s=at%w
z%>jF}wF9PqcdLIt+~VMIBO@<Hq6YV)bKJQ81Um+vbI_dYr_Y?p^r|ETrUkbRGQ=v!
zmC>w>q|wu1NXh%wArHmFMhlYYSv8a5D>_rQYbZetu4ZK;qTAlb#=>%FHgt3<7WRr}
z7BAIy=oJ&<GRKbC;BdFmNe@siuQBG<vH=4dywun3jY~1nyyTlU(zdAKpWL4>9L#Qv
zCEeZK4No_k73<g9h^<=lX&R8dFy!ky$_zJnQNKFmK^`G>=2v2ElbiSqRakgA=C?r3
z-5{Qf6<y`btUkL%Gd^B=x<<$YV${Zfg{7^TvLcW49xIH@ocv?M{QDUZKF_m^fpr5?
zY?^9V45-#={C<UMPkg>HdsyrV7RO@|qwa%e*~F1FSw)JMM+f&%!UN)L7Jm0n&j}W>
zh%&r@fpTLnHMJQSgH~!bnsfAkLdtK|^0_^JCC%B`1>4l6hL~tYMP)i^dk}f!Y)2sp
zvUP()_S1~v2Cl=55fJ>R-;yX}m|knp<&27XrWvynH$8o90N|7<qe4(nP-TRcU#h~2
zwWSa&U6FrexL0g`d6Y?MX`)7#qL0s>Pg_@3MWxiVo)Ru5`EzPOPnUx~C(9Zs&SAZ0
zGTfl5#uz`&VH^G&0?KzmE7S>huFUlGyyc(Dx@=j>6|l87TsYFy+qnLB!^3TT$-(X)
zTlM-+(*!CNbWDSHzD9&xPZey?5FjJ(j;TT>pFF<K;-w%b4?7mJ*@+^)3zntMwB`XB
z$HgVB`%p*GkjriU<J#|bugzpL=r<khk}dBn{AYzFQ1^B)+iS*5wWv#D7l+mA_6;g8
zgF$SZshVAG@-3UYGtY*0*h`VT3f+Obdsts@dAD?0Mk0Cl;pvq(rnCqZ3!!7ao0C(M
zQVJy;y(tSTN=Bn5x_=`qN6qz+@L*SyvBy0g3<;qgeo;7yS}f^7U5JS@5uDF%?1?O$
zQv)vENaMsONg)b!md#-p0G<ZDTpjIRF@qg}{-w808R^MC36Fe>St!0gLx9aTzud}d
z0@;gvSsd!KgevG!&%;BTanJpE{bm8I_f_h+95ll(mxZmmqC)6f<Mc}Y6?y0vcP=CL
z>17ph@rAGj=GYU+d_9=!3|Fx#>QtAOZs{ur+eZ%o52a7CgIXrc{CHWOCvZFO$e#$G
zPLjO$dnF|HvsW1kWgq!&$qmtrfmF^E8B$$deG7dAdPaZQoLd@tb2}cDYA(#q+NRO-
zWHFP0p7JWIf;F)4YW06&)Zlo+jjwkhJ2kT>+FAm5W;#Sd-XJhj=1g<hjqxb_-Xv&r
zrgzcj7vxJLC#rX}uSlksy*K22BQVq><*0FJ%|!v_&UYxIe2I?m4)^1>fGuT4mp&^<
z?8W)B;%uXF;ix2Nc1pFoyQy0zGdeaJQ*4Pk7v@yqjN+S1+t2R^F&jNwD;AX1Bvtf`
zldp8)witaNd}<KlP>bmkW0$kqHZE;G5&3=GW`H|i#`t~jJfN#R>}uA?VKjSffPsNn
zLpiCT>&>OEw=Xu%P(}z{B1VK%3OzUU8&`^9-rpO8R-@D^SC<z2SJ4bTG`Pv*+e$Kw
zR(#Y8T}Hj9Clwr}{a<AuphkqI4BFB`8cOIVmsOy2v(IQwNj@G2hU>ERyyUlvoL;8#
zZQpqA4261avB%4fY#g6bk|-=igyBTZ4C$!)G}+4N>vnT22ML;ZjsA^8356zqwV1RA
zX<Z&)02cNW;V-Xn;Z1gIvK)DMdwZ!VR2c(eCxav@CKItQe|=nDx!TG(NtA`A**khB
zEv&!#juM_rd1}jhcbt@Fqe}!@4v>Oz#3c8HNk^K9NTLqhw^N6(Z|$SluedyZC5nUq
zP<vAL=BifD@F-u0Wv{uz_KwudNVHZexbb5!A7l+{ai7{pMG{udRcL2hOE$3o^5w0x
zFiuZDT<#<qYX7AeCxaK~-(^bF_g+&0nzEe#X9o|ZH(TBs%Qk~SK@ls_%E_sA&Aq1Q
zVLl{op^2<q>(EM-IlbXij)rXU?uMh~_ja$Y%14%BYC{jQpKP)2E?WDNuu#QYtRvR}
zM@bo3BV~VdbT~OYV;vcp3{{7pKMSf+*8sCmGim#H8uE~tGGq;=R)v?V<9^S4?V~e`
z`z;HJ*ko&Dv&O8e1V|AB6|qw|FNs+o_xYK66<PbIY1IwS_S@hd144)Im_cIHhezeg
z>#4j502su6eSN(}kuacH5FJe#VL5n2g9q>Zpg`@N_wDpH+WTlnfqseTUx!Fbmx-dP
zwIwF14kUSHc2q}O{_YMG1&W1(<Zzw>HCLSS-Q7uhzBNzo%H(ZSWF4KM@!{T|D=YWk
z2-MzF!iQqB>gzaqw(~LVx*viAw#ZU)>J2i~miWkN=}-XBcuzlM-ONLW_Cm>3Fq2rG
z1=K}<d<irl$rZ-UvCAF&*~(>Q8N|<n(5h(eq9yM2gZ<cH<3izwJ(7F1e*?2-u7615
z94s{1%~a{oKOGS0;L2iWe**^opdPn2V12<)kKfSUZCo7}FLBwRi~x0xwE{tKVZ46+
z?7v^W{B2x;)|hf8KC^Y*xYy&3l2%@M+tJX_yGNwOIlIGG1Y4P}tEjWh8w1^Pexq80
zo@`vaK6S@#KOY~gJ`@qrt7%m*qOwaA+cD6t-ZWonaiPGgL<v=C_Vc0)+7=~Kt<#*4
z#wm>)X<(0wCG(X2@0baBm8+>JH+UoDzL`Izlwj|v2A&#u==z*Yro13Hz}!AM1vv;u
z;#_;2=7Cvqiw-Zs`1FODDwvJj1ZTW3YVkj{)nU(&V9X_zt)2i(hY-KR_x+ag^uK}G
z^oU!3V<lYKa9nzGc}T@i79WjLE3oTb#%e2Do#5X^{7N3jR-+oLq_&;vXs>wVrG4N2
zLR@rW`d2XW!+ng5tbEd{J~uP*<qKkL?Jr<d7rS6zGM!nnyGtHPCgurEriMGs4mJ&<
zG>aUb`Y$b#`yMeu+MRCyg+ya;eCWAT2_E#IWVIc{0`Z=TUZGN3aVLTtZ|{@^JT2Vv
z8I^=`7%gu4ctL$CJ<AnOp6Xn&y1M6t;`x`CmZ!E?+%z#2DUSq9-MSU3bxUb}I5vB7
zc~Qy{$7$f=BO}NDbouH4O?6;y_tm$vvokQG$qNp+#z4vAIP?XLCjCS@h;BXT{NU6_
zWTkQC&^c85E2%Q&++Kg3o5|~K46HgZDc8b~g0JDGxXtnN`CT6M?3Cr`Cq;P%7JM{J
z5;xM&$d-TC(W$dD)qT{Z=ekBRv-@s!Jm^eAX^aZWBY&a)f13Ep$gc>~6>aoO(aTkA
zw(f#eSgj__j19^6_DgzuKS;fTcA+8{Tr~EccwUME;?7l|9BHJIEet|V<;GRsO>?_W
zb5^NVDp}0JBK-Na+mB7lX~#x=cBn;^bh<}jigHOxiu?(Qw~sHYj|h3YLzFewx8=i%
zaVBO$=r@Xl-q1%j*MlsO3CD!eyL5d}0d|BA53_1$&&*;!44!_bXSlw}vL0?EB04>|
z0P`vvD8Ze+?p{Mc_BKCy1PAfRKPN?Um9Iy3BBB(<8Xg;}TI;2{QDXT+pYc@5{%fHr
zq(~J;lP82bG`&wRd@rEIrDS6hbLi^Xc_*d@4UI+jS;S>MFOoq%i3KfWnz*7ki@_aw
zh>lntrJdbJQ?--E%r_T5!k+Am|4mP?S)$t8BX~i~lQWt^-(kEvJnV6w9Y2~OKQ_Z5
z9H<YdS%36M6Tbd3hwl;1I6{7(U4LAvB_&%?cBq62tmRNqYL~fwaZG5csVjVbcmDee
zQl_Fo=iTrAxF(M8S(eRZoN;B^tRrcPjO&fXhP9+mnW+*%oBYp?@R$Ry#?ryTQI}wh
z;MYpzUQBiOB4%z(yk`syp@7m5mA>{-LW}EEM`wDK=I98gat8Y8*YQA(Tif$<=bfIr
zoPWJ&P{N-Lbd&RKN|hzAR~aBakXKn3CWBv`84HpY7eSFVAzlWUY0(2w9UT=V@BB=A
z8+YAdWS74<Wel?3|J?VlTjg}K%Sz<Vde__AHqKNrgE+q9-dC>Hi6=xvnv01Oialmp
zCV7F9eKXVca;Qxko-A5TwPyjlyG0il(~o~;CG3Z8?OzyBX8Mb}*Xm-Tj+(mJxzxTH
z$@u0c0+VZ;o;jdt+97@^g2MZ<H=>U_gP~|X?8zG7flv|SKa(!Yk{@0)9NWS4@nMLU
zg-izTYB?Lm!mp;w+w!Sn2c$)PKVlFEmP`!x)})oJmuL*{luspvD1*j-6;)0&m}hM&
z+RVw(3?=l5jW78J-xfEf6A^9r5+pOnVo6IE4|_l?M{Q4T2!eY`CUBF#TC*nf%c^YY
z>gw_g;aNjV&{5AT8PucacX!W}Z47+BeYJp|?KgQ(sSISX`ON5e-<=QC&Ck!X$MX5A
z3gZ<|#q@p<CO&1?$6fiv!qeDBgHt-n7|=C2ctH1P5)@hJ9QZMf?0BOcr>gg64~ij?
zHoUi-m6i2+*w~SsopSXUvQTJgY3Y$e^VyiFrc-smL=iIM)V#_2q;)5=;&!i{H>;_F
zVU!kYU$^vYdYW~pImSGUS;p;_4ITIM^mH1AOh}wiLMI>l{M)jFyHiQz{=M2T*UcVn
zhALmX+e1MA{#ud!S{|C!-L^%broOl*oM;8W_s)nVry(>)U?M3vZBKS|>&%TJyR3|?
zLI1m1>cn2DdNh4ZcFw}(N?gKewVFt=hpzusFSls$;}Xza1do)=wnc_Pg4$<s>?a)>
z4`u`foU?NS)?|90%L|{M%=IvSUde9vs%>s=c==Y=HOV8eby=2*ima%jYM45SH$2>v
zxKXh*&*D1`{JKMAY@4pAZcMdVQO2s<SG(X%U3=^6kL7atWu;u?PSJ#PB+aukbJdf&
zmJ`vxqZzNKGf1J(Zy#$UD7R#12E+>d?P}R!+2l|9tc~R&cuC*g2Wym0zA(`A{G$$8
zvZWSIOJlFATmVFecdJj!G5^QrCaj_7V+A4FE5(q7Zt;u-*K&P!ZmuuZWHz;c752}_
z&7YwX^eq}1j@LPZV(K~+g(tZC418s@mz;f|CvQ0?>}!TK6>Y!GJKMgB9$3#`?zAGC
zyMzh4#mJs+U7yFR$_>RD2z$C`wY2C89>(;9X*wkJ+ZUaFldY^{Ol4U|vC*xc^Y%tj
zR8$-n>f1V1VT2hO61OiuwH<tf@Oc0~qS?Wy?{5T2bTlg3ox$vfS;yuu+Gq~FmS^AM
z#)Q^C&Mba!1l>U_EUY&|V^@Q8a;KQ(^yep=+fDm_{n+CA)8sd@U-Ky{D3mH97!dPO
zG>jW6wBJAG`sn(Y(a@3f8oH1r+iu*)$M+Xev_lQC!wgN7(%iW8bZl-;3c0eJk&zo>
zV`HnT=rHL@>IZmK%L`l&cCP%!^FJTrN=yLY`y^ABlCyIS_3+1Y{QiGmzNlYdD}~w!
zmuscltbdh>BNlfPco=a8>jrC-OQgK&t1~%_qRCImu_1n8DEloV10PRI6QW|@mbz@p
zpzy}`$aZOwu=0Rb?)O|x`;h&{rms~(B!gx-stL~Oqdxj5#r~+sZ|n!?Jg;h7qCrJO
z&^FY7=!%QF@y))I(l#4Ey?q`bMZVp3nGCZT*}}pqJ?eb&-@L%)y1osP+~Q}gJJf3i
zEtzqAb9ViAfCk6swVYNpN79-GbG*2Cwnj|4wu#A|FIU@~H3T9IGXuPUW~VvTt=FVZ
z6A)Fx%?1sk(b4b8o$8cmebE>s*xmOXbxl)3wsq4VXR4i+4~64M2qyl9eOSo+gTB8=
zG<>fIfoK$444D0}E7qRu@F6SEfGImFXbSM-O}w2EX*Vweyn8NukFwE6$nsI<s4(&(
zX6m4W=ELJNs>FR&?CH5q0ar{>N~sHz98^F{T6LooHWtn^t*@`Bpe+Kl3xKM$b#-@{
zGRdq)a}^j<#a>4@E%4NB<%O?p<p6$`l}D!ws6QP`dB@)0z@lCUkP^TlBaFAf=%`T=
z4_7OkK`aBC&;-UbyLR@G6DoNN0fJ=L6*0)Q`O4=NZh|l0H-CC<V`DxYA{u)nivarx
zU@kP3BHwm#D1NU*m#lB7lL-?QY_yWAspnjGW;Pf!f7(zZJOwudEs;3DLQ=@*{MI;U
zvlQfyub%-3eru0rKqi+99>N3zxHsl+iDV??Oiexd|IVsW$~m6Tn@hOggW)RiH^7Ip
z3%IkpKb7<{Q7eyAo)Jt@PP6q&8x(P)#$&M?#b@G~7d>Df{$Qw50eg-CAo(;xPEnOU
ze3t&XF*!1FlFfD|h2dv+&~a!eWfln=3z;>HvNv!-h9u8xO!#D7^ssHcD2^k0u#0Bg
zr4>u^qyG1vjzO-(J%KE-c9Q<vEy10drbLsk=A`Gh{D6S};|WP1BW~LUNzX6O7H!^u
zAZPee`TI%dLweTw8hUUTtOxX9aEYe@txWrv{=_=jpO6_zdGOb-51l<`1PQX=8+P8~
zqCazZWn=d7t&hk$2T4wso6g6D>)*$`jc8s<N=m!OU6(7OhZF@L%F3*siMYz?)@cJW
zo?~u3b$fSNyamkc?s{%-u3qhdpV#s%z#@$Q?p;N~cP9Sq0((BFcR!I{1!KdP==k+-
z9yt6_S}KL%!njZWoCnzc#(+O2Z0Z4=Z9&t(2EV@_xDwOF@q0hOh~a>Imk*Vin;RI<
zN<Da?EEA!H%a;n{7i}gd=Rvh|U5w*g;lF2GoiiQrgV`*of&R}x&}ICTYO1PrKVZ(9
zRaf6U!^0aL=6U2SMxQ~neq7nX_U|1F6wt#cgEx}pY{&C!b*lk^U$Zio-)R=G$|>Xe
z(cOLCJ_h8rESrXp<#}bwMq)e^7S>#P+S-)r%r!R`oAcZHKrwU4#N(z~fdNDaUIjNl
z{mk-$P=5O9XS`q|>-_j<8Vm*t?)L3YC2_*YJei?ftAK<h`uZ^NdZ!#Qm_G*b=u6YM
z#M0tozyQ^2Kubk>f{TfYwZ_$C9_qh8H#$I5E=QvxNsEo8;6PWXpCMi|lZ}cxZBWKS
zklgFKD%kEJEAyeK$Z}`0^hw8jED=!GFrA^VUX8sj`^8SlRLDZn$PCrG0MZn|JJqUn
zCyuk}k0k=+6enkA_AfJ!&}j+LccLOI>Qw0zChWbTbL+`;pS-rmK~UM6GZQ4mgwjV!
zO%8Sssiv~@o_6d9!2$X0`>mn^GY2r}PAP3TKR-W(fHSF&`&5}q%RY@XV7IF?mtaH;
zy6RPeIAubY?<Mj8mZkLi@Qhj*LPM&<O`uUzIW%%ATr_9dutA!F#Y7cj!~GAVkp<d+
zyOFA98gxDbSf%D(UVvBh6gAw%#f2ktuQbz?^}hv^VpbtPYFyGGAMGuc0|fi50M^B+
zQ*XVwx$W`Rqh+P?JXz@Ej~^BMbR*;xG*f#=gJf_9g1mB+eoU2x_a@K7{I5<nL7-n~
z^RNQ^@}+|oefHLs04HGEhgZ=^OvzyRl$7+Hc`PhD%bl6J^{V1Ca|`~foFQcQ3*+xD
ze6OMjv6IM!0gLE$_0rwwe(%ru`F}I@!RIK;V#>%FL_gB73<&Kz2nRf>rzEPP2lvih
z0c#{S^A!?|5%fj8I)_~KwDoWfR}1d7`N;Exy|&&E5h*va<b1WT-9Q2}dPAU>;j^ip
z*(GAr&3#J|<q#H2)|k`sMAp#AXs*tL0z0ay%Y3=Rrw0&vWvRs6f4vN#d;Svd_-fF8
zT7VD5$Y=^j&Z;4Dc-YUS7E$<t_43l$(a+7yfUdU0NK4Y+A-V?kFtWz17M!qjHoYx*
z8_564c{I?DfPb$X@2?=M`(a9BTEV!b{nit#(1N;=0GFYB(a^?h^)Ch;9yuAjxw)Bi
z$;a_kKjwl(t@GQ;Z@Su~Q<n=q?VrJLc}JO%g}+}I-$Jgg0r((@O}Af{8I%Q4l{}?<
zK_$fU0l$9@GVj`2Qp=t4=it0r8(K2ASIn-o#)YR4^Ue~ks|2E}@tP}185Jmb>u^;T
z06z@(sdb>^#s0#RwnuLXYU!zIR@y$+)X*p6qP)uY{^jD9S)&W#A;nbf>R$V!ku9p1
zLLVW;-6e=Qh1rX_Km>sy#6Wl3A@_Db+|pU}R3v_lfR++_iHwm5a9sXe-n<~jz`Mes
zvK$#k%nv9h9Cv;^2e-kOu{m_XT_eC@3wh-m@pKCucJ=7!D54>jFa8GIUHI)|(DvJ8
zS1K)qLK#yvQUWCkDB;q1c~!fPkjcICWkm)m5hyeT%rn${DCLTTmIWWlRJ{y>1Ew+1
z+G$wSJyAFBHUF4qF9eGr6gyHm<R&^gn3<cwOYSun8uswyihFAN%l{b>ueHVp^DTdB
zE2Cejtc2(b^bTz-Inmqx1>;PT8M&%c@GPh`7b&(PKIjENA=!;;{+QRW=c4TPNbl>-
zx$N;!6tN3&!E_sFqqfATM}D?LOLVKdVo)Na=wt1NQNMb#3}R1}BApPSM<_SwSFwm%
zH1(%aD0>vX5BeT&#o6^&>UzWs7&zuYGOhBZ$;)J^eoVxj^=S}8n+N8M490+fje#)Q
zKi(o&4fG0t_~X6e?~&2(U8zMi4tmv%z*^qd0~d=KU}=ByL`$Rh?9wi{viD`djdA?C
zI(NbE9|M@e0KT6PVWJ8r(nThRY*bh2m#0QshQ74i?<O?68s3tL!-ba%MQ%@|VJ-Z3
zDSCwd#r%Wf?sY)vfHV~O&E<FHvT2V9DZHcsg<99@ikDmkyF#i`T++nA@)Q(u-jQRY
z%zyeYGd))wBmwa}TIDokuUtnBe@=>tufa|%%~kMQom(0kaWUW`+JM?ofq|jY4il6q
z=I^7!&~QKlU?7nKGq|T9+OmNZH6-shY5qV;?te2H&0N)hhoGzHd!D6Rig?Li9d!oQ
zq4a^b+=6bVyKScRa{ZV*8QTr?Ox8dp3&Z42{t$Af6b#40ZXB}diB!PHtHhz2WH(3=
zvE?J*bea8qEEIM{7**zsz}#G92dYdii}aFPK@#0H&flgmrBh0djyT0n;nUn*rkrTJ
zthVM{No0+PW>};iVKFsMza_Nf<R)E5XJ`CcI1v?%paIs^$m0V90kjdlhnnxHK;t~+
z+R<sJ>oQuN0G(y(F)aGl(8YNq<54M35IGU6t^6x3fsMC_wJN0eW~L1;!@ujtD^T3Z
zKrM>7FOovM|D5ljVWA?2ky-ywWx%CCLGUbKn+n47eZ8`_)Vd)Be<wlkGu=|FRxkV0
z#{(V_pLLi8YNSHVBb@<4Pc77Dl|0-pQIAIH&6NkVyI$L=)>aI!tr8mLb8cin9i~KF
z)!XPyk6-u%!+xB)lo>SF?`!N^nS@kx1Y)u#S#6Op@5BDP%$g@l3WH=a<HbqMrnj-p
zrbT*HRg1u#x#M3Fr0mA$XaI!Kj{-*V+fhc>m0XE4!|yY5l8V~trAcH+z5kgWqXSSD
z&|C#hb_d09OV=dm?ll#v9}LEl1;O8c{VNs@j^Z&m3I(t2F&8JNxS8p9-W?ihS;~wN
zQWiArs^x~YiBJfXKEdHSQVVP3bp8Ub5rIlIy}V9A*|zTDni!v3uMD+q0xSkkXM$GB
zA<RC)Kg5Olh5zSC!6S74>*ri)^;lmkD{D~G(bG!`ilcqXoEeJnj^n8yWF)5N8fm1F
z(XQ>A89+^TmOWaW-@0@F64~DTCBL9x$nEh!M4L3TE7)a_JrM7o>$lH)&LnxIU<ns^
z#hSibmg7o((;dgrVc!>t#q@9M7ya8i*mE>K8*nLY^iZ&6qy-^*ztCzc&{hGV3G$xD
zV-ScDMrff4d9LD!lsRD^Kd`=z<N2>diiY<;3*mZBO~NGFcny>beY2WxL09%AcxM=C
zl9E-KR05dGo5h=>s#Zd|qy>O;JtJURN>hQxk#Gis<(6Lr|0)SrVW&kL(YsA8ZeC1Y
zM=TeiCmxCnabySYp8!Ms`pz$Kl86_nFp11l3UTK&dF2jER9Y0?jdwUC;D*0FfR|Hq
z{loD55{2>QU5vkS=3`n&#x~g*_kYg+Y{U-O0odr=ziW@VEK+Dcuuz-;JEZ*YS1I<$
zShADezhY%duLS<NKZ0>_Cj(K4zr``m)5h$`5DMxfCPH+g`wk?nCsU=gW|UFAX75#=
zOBizi#jHHN6f_E(%CN;bYsQg3OWA3*29Pp10ca^hG;L`j&!q(-{<#A#y&lRd+M*@b
zPM6K$bfDz`DCu;2ya$Ld5tx%DN@D@Z{rBI9&+|86hL;p_8@}xi2>^0aAbZ5{_jy|h
z`3}JIR1gW$-y!V}Zg-y5K2}y%c6Q}~C0$+1);vU8yGU9XAd3I`#_}y{1SqN=yDtJ+
z`2Tz^z|!QTCCCtfk9)cNnqXpUG{?x$aAN?8i3%M2ml!P|xCLzToM=e$AW*yA1xw-`
zi#-6gN3BU`3z8lh%(DXa2;A71rpd52pVuz&_8;Kd0znJ_fo*?^;?RAO!QKSH!4p_x
z0a+qIvp_AT`Xd5pZ~qs&{F>1-3%XdalcNG+KODeeSN-wgQ9J?gVIdD56hNeG2h%7s
zfC9I+w#xGZc0<vq5mi%}7`9%WaTZ;A3__8X17LJT`0CBd<wv1s%AzcYb3$U0k94~O
z%fA7tDIoBSwqIzp0ZPQr=d(uTy8)puUy`o|2pZP5cwnXueKYSAhq>YtlL92cDBb_r
zVX7=YpE_|p8c;0|Y6O1)dmzs)xIZf&m$-43lam9{*uT55a1D@~Zv}g<R7y3o#<L!k
zMMTGvA*sSJSFcBghXIQ^<NEjxOp$nD2>A{G`v?X;;0Yf41)7`xTCghsaJFkbQS?Ok
z#g|X7Xn2uwxyV?zfIjK>q!On|cc8zS*=i<W<pQ?n72;ha=Jb1lgTU#^7AJ>%(r0JW
zt<i6BHUK9sQT_zR0TF0G0_*RK8mZE$#-<kYJ7}I;0kj9;Q!DLW+W<9ZkMlb^Itn<?
z108soN)FIK{P#kjlLDrXq#6&ODs>iE0s>qfojJdgA*TB7>s>UOz>n?i3+7xX;k_*_
z0+}P{U~wul3`pn-_$X)G$X}kNJco@7Z|63kAu1XLGpqk6OpOthBp@VQXo>okfJvGQ
zA)n5mkxv04vWv4RwebQ;$?*|)N<@SaQ+upVPyY(=V~}!b$^HZBl>-Ppzn$dGLStj&
z!~M|%$1jsdgo3Zk|Ao0`p(B+iu7~Ws7Ri$>Za`{%?B4k0F&(fv%JL)w!hVxZ-(|21
zPd+oewPa$%N7yXN(BHL_aGG1%JGtVCHuK&vsb!omLoabbtT<s(z@H_JY~VpT1$tYZ
ze>X@%G3FhF8%Y+V8id@!Prj4m!GT0!_@V>cH}7B0BOwKlls<D~(mj0~Muzw=LbhAB
zBY>pJm_mztVHSETk4g7!{a+M{Rl?J)DK$y)^*GbO<9=tgN4c?FCg>-O4K;w~qAIv7
z_*jMs93J@40)!|9OrEa>^x0d$K#Z-ft>~j5s`}@TCnOtc)}-9j1%S|QnP$PUl_0n>
z=#e{WyLWVC272V6gbcZx3n986LyFC0F0zaiQ(%uXkBsLDZ2Q#G;=1^Wg%q>SplK_T
z!nF<P>=2=9ufC`IuOOOw?#C}rH{9Bi$jsw0X*hJN&wurX>sHUKxwkF5g7MMHis?U5
zWM)%P>bdm+pSH68=iYwQeFJoqlM@pjz|_De0z}w0kzvFm#lPLrd@smnBP~D|c7^yv
ziHk)iaOolF9sP6`0J={}a4E@8*GnwHpIj!FXzK9*VXff8-;YM)WSVpN0qKHM^P*E-
z`&PD7ok5594mdtRlh>%~pGu%%GF;NZ|6MAvwF-J3fMJ5D4aBH>cB`t%DJXh?jMHM2
z^Q{J{2FNVk`nt;WWjBl;NE4`U`#p@6lt$K*Bv-&^H+;q5bf>)BQ40x}tudJL^g`-?
z=42y%4=kW@NMy5b$GaOK`lS|^Q$I1%+b3LzbVmUoL(T*Z2X)dmNM5&lW)B@8P-J9e
z06p3<_zNF-0^&w8=%B2*2M|+Bz?$cGpAGm#1Gnz0trP|&pl`oMe1HWIJaxEYe5e4a
z=QH?rmIbHt>66&iMk?rTO!){RF+GtboEX5o`udA=&=SfOvQDuIs{@vp^%SuVk^%6d
z_>U``U%pv{;tl3}Yy4vs84q9)!S`rLwK~A!Ulg!Gc@PR<8rL&W=t}-?!m|Q!{YttY
z(*Pi4MgZ<<-M}jmu!|s4!WQ;e-yEe8k)shQiXz-RlKubw&wt4c50P|9H8#?Qce*iz
z+Ckp^A1uMlqn|=>()U*^jEo17hye!fB?Qj<KMRmeL3q_WkQ@=DrLO?AXdwZ-iD-bm
z<4$=e0iFlE9RfsI0NV<|98dkPn&AFFcY1XXUa>$Zf>34BD?I6D{xS`A1@;&;G%O>P
zftYvA{U6W?WQo6NF{~9JPf#rN1P6*EdF#$j%=Ulh+bwR8Z=Y@f$OV8}9-x4FyScfE
zQPYr;`hnyF^3hS-dg5baD)V9PlyA>r&($cSz``nUM?FCM>S%v|+Me$u>HjM3t%Irz
zzjon=E)hfp0g(_8kZuVj1nF)>x|EU@q)QqEq)S@5RZtqF1f-><L!?XMEPUVJ%s1be
zIWzD1=lmsX_r7`d9c!)oTGzVP{~s{PO`(O;wM`mD0`aC5?)IFDs%oWSOQ13qh(VZC
zj$eB4>pm@Tk0J<#IPpE<&D|Kqhp(V=@zK$&AKV%LT{3E_tD)>T{+V0Bf4kJnIWXQL
zF;RPC@a*gBV~{g|qTsQCfjQ?oP!|9K_s<_ce3Q@=G~W7qeTibbTw9E!ou-$QYr}<u
zk*;5+e^1ovI}S($d_wUqXk+hjoIy>QyqrY2iKjgyBO?G12e!;vvAKfs!DZ1oJg)7u
zRke;H$O7EPGW#WdetEpAFm}7*?mJ|2Uh7TfgG#7KfwA#%(r`(|1rHSD1}dhPC~v@3
zaOxSO>m7qvEGoy_uTFp%1Jue1bJSuYO8l^p3?R!Q!qMENk065x>_dEwb5tY&^IxzZ
z2DS_V`XgAtipGEM@K$GlrzAC8WJA0yy<z$>?|Y#ja%dWv24EiO2Wu`rP;dn{_U^?*
z-L5x%NioXoJ#t-acdghq1SAC*-e<gd^M-slM5FlcK~adh?%%YpABt6~4+=;0UKZEQ
zex>>q$Phv!Bf)Em&rl&QMP-uuCsOK;m`J{ek%Om~yqS%e0&7dkCzs8hS-+!eVp2HY
zTq(JpW73=X6RM!r>*4ZXUFQ<RW<+^rEZN)C8b&B4;~h-;pBJ4?vz82uB(bUK8k_X<
zA0#AP4j`q*l-Q{0eiwtC<F0=ckrFp|*?n?G|6o#T|5yaeEAziUk<o8KyU`^X0-w(h
z&sOn$tl~_#QW?ZzHgsFA2fI2No$bnd=l}dIYuae3^0u&tq!Qw2*SS=%;hC)c^~Ups
z4MP9xVXXb(gLmw<NS<GXNB=2FLu43A>LFmo@$Uiit+Q#a)g%n$-5s<zT42j2{P!Lv
zVl>i=m;Lv$lo-0Mq7={zeWe=W9W#>r?I+j%qDPAQ@-gsh5|o$SkG80jdo`SI<j2fz
zV`SDe5pz9%@|4vD!dFTb=dIZ$sIGMp6Oy`l`CY7O#i(c7M28gK#>4=f^N|iSH8nLG
zTj_j@C=<%%bBjySy4Kxz&eqx4N%=c#Y1`GPoS%=c)<H~6CMcLRvnCol03BxgH|>K5
z<KMmkCACIr>3DZ}zCFCVukTM)ew6*(<wH=@65T!x_}N{kS$cYQrmOH72L}fv2#m@<
z4!#%b3g{s5|B0#FlTUFC-SEYW340^YlRe7@&$E2B4^p&iRgF<Z|3+zg(UboLN*Ums
z8ygSqm@YLpH?!(}*3#8AwyjBOYz<SIkdmN*b^7f~g>5TXcTKleBQKj<Gxz=b=GInp
zM2ngDYLCWs*Vm%?*4V-5brs5WRMApj>F8oT&%RzE2+C8u`R>*E>8H9nL8MHJ8Hy6=
zw~l=O4N5fK5%yEMHk<Y&5dVeyo_9F;O;E?ln(@+iJ>lN_Gqc=FF4yzi_rnH4c+w-)
zW&~^L-1K8OCo$2-r%90MrmIkX38wIu3zm`(z89x?{@rv6DF&br5_vn({6%5Cf%0<G
zJf4{K-J8Me@JYkx21v|3G$c+YE`khWwB(C^<6L8FFuNtu(eZB%-kRE4EOBvsFRzIu
zHi;}RqJMkg@+9zSXt29+BaWc1)8CE9`V>KC?XrzKSbM*DuUI4qCAQ41tgoXX%=!;A
zGeipJ-YfY2XXd9SKcVU9AO^<9G53i?P3ioDTE0cQe5t>>v;fsBKwf$G$BLdF5_k$G
z*_>#Ssi=``u<alF`ZXy1g@Dhfa-7|PUdZdJW~q&glvcTsgprYvsHjg|TwHy^@!GI7
zH8le0b!A0GED7717M#83<d;)4TC7`JUoVuvfyQgA(9>I8S&4^@EhQ({)74c`TN_WZ
zn<6+W$)pE{5K^a5cGeJgi+xQ=hX1RYn%dCNkmI@YwdW-0u<hL)92{ij<YMK60s>fc
zs=iH42{SSlC>voS$?And4i1VG0@-@XnPF1O<||>Xw{P4~)6~R5#>dCg($d=7+kpX=
zB^l1d%DU-ae0S(;wh_nINYZLH`WS5;9j0UjMMW0}hpfR>p*w%}*GH%7Tx_-8p6-tt
ztEg)4E_uWxefaNjU;qBe6Z{wzfvUr;g9Ee5aSRL$C~$`}m0ITH|C5B8cs}21Xmev@
zqXNVRDk`kjZ8bKhzqmU(I%W}bt*nZw5KT5H#!JN#(yiS{J@Jk<?@OEd_|i`%2elEd
z0}yEaNYs6juSW<6pQ5q5`~Jg+X?@EC6as3z9W5>M;9mj*1H(htV2uN)1RjVDSqiMS
z>W{pb+WKvln3(v-_Ph8>rLW?-7)4We_xkpH2L}g-c;J=j-tDa|i^-=p$l%T+E%V#@
zm;`q~Bca=X-QM0F;zv~E<;$0_02P-054n!KuMMF*efktE#<-Ppt<98)!{pSIW|3Cr
zP)3x!qok%YWGj%N;o-%_MZiKXH!(gQ9yy=Q(|cAFnFkVY+Gv>`n%t@8f>N@q0W#ib
z9&T=WxrB_?sHh6Z>1_#F*+Tf%6{{#{m}=_kpq_<=iRr9^U{~Xb(OepUyRPo;hbseF
z)j6Fe%6b&?vW0i=igjgHUB7@3?d5#?Hq_r?c=ylvWmPC+Y1cWk<P0yrM#T^R`OC<I
zJozd<4r9QN2sSW232_6<Z=$5Y-aDa_g-hD(KS)=Sl~wXu=Q;^m-I&LV?*<JAx@aj%
zmc)bvSBOe-xn45!sjymc+*PL<87R2Vs0>C{+)>hKY4`XN!htLJxK#~!T$KW5;uk-w
zEa5gg)ZcHhsS7)-Mt&N1^VcsUE33(-ZJHsPf*ZO$QmeF%pUaBm#+5U!4v==lL&b}x
zhQ=MJ1@G((ZZfd@G#M5a2EhWGPSpwo`@J5aZARr<MUq@Nz6>oM*BEcsBt$bs{a9_@
zu{V9<Bk<Uqb_opZ((<M9MfXE8GBUSUuU@&i@mdq!%|6*8<5Iyu_oHq<)H)S+Y7cIR
zi&MYccZr98X81TbR{w4Mt>_dfx|LPw!Vj?&=xc<jFMl<FOj_{udmvt*F%s81zs<?X
z2}S*^`d=q;U?--|y`YT7rCMD#HagE+3u~PO5tPZv$%KSkyvcgcp4nJf@Oz!VhFkK;
zOnFe$S|e5%mAVKoI~6W&ECmLP?^qBfmU)GoSB}@?U_<+OU6t_UREpN5zCA29{J918
z(~^oT5ehLJKiLbY0(L%^y!M<=g%Q5>kAbjoMjoE72vjHR`gOES-ezJh6%;ah`h<jw
zg4n18z#hhr$tCi?%>Gu9x@p0gI0U&6m?D27qTR_)4uW1;*wy6Vk<`3v$~m;Dlr9(2
zK$G{MpO(VtUv(@{`>?^F6!g!h{2K#a3x53mKd+HX=fDvJ7kEs0<)3@^Yly;ybr&K0
z_elSoaWK{8Y~CD-?(Bc(Gmh#*MMo#W>j9=2yjDaoo(I$AyRJRFPhJp6hKuo`aF>(o
z)9`RHBP9_LO#`DDC@z50r68p^`3tehkfaKuv*}%RCD|O1WNXTn4qhMesM-P@v|Cps
zBku^qvW!_BbjPznW=xP;YLD;I_wzPS<U~D>s!pJI*5El3La_r%2=Ly|?sLtFT;@==
z*d9r#4`L!XqVF3}E)6<_9nsJ8b5!OAvNS<q<kKgXcxD!s>YAF|_EvN#2v(r#|M3IL
zV;2D`5xA!?q%js3u0$?PYeFVJE=q+i#=`QeHQ0iSOSt@FNr?ilDou2DPEKboY)2el
z#5*PBHU#qgj&Ia7mMqQlN=t#l6|OhVgm*cEv~j;?*|Kq41q8NsOiNNJ*b@s5H#^NF
zB_zV{2<tEtqc!#<azTiKLUr$6UVH(gntGEj#>@kLX4r|(7iH$duFzpd>Ocqv!N$ME
z`5+>bM<V#Z$;pmte^u4)xWZ})3|<gEIq|h?(Md@sU_#>A^if3=6ux`+$R#|aXpg9N
zJD+@_pH6(;SV=i^XiK^B-^5?WU}Q2Z%r<o&At<PfzPtWw9tDa(*>`sj?fCzKLQa`s
z3oxELQw5Qzm>A=jSoEeQ152E^xUKoftwF`y{Vy|8!~dc!DgGloSZ&$v`_<LV#G$0O
zEDjD>kjE)PpFS6b<(4)iD7f@=mNT)Lo0y`AXgk-?$KZfdMp}aWYIwu$SVyqssi;3s
zN;rOf4jOu<gM6Sx#9`bH?EY`_+p$DFnc_h<lOG4VnjzD=+HW9`eyaR99gpa8QTqo5
zpkTkex*Eg5fn2^f+Q;YD{`!8yF{blx+>)ZFj5nIM=ck@O&sHf=SHD@{^5;-rxBiy%
zAIujo@S@-JE#KcKar{_cUmw=DrSdt6ehGfu+OivW1TVbM8B^!DT2XS7Z?(1^ppp6c
zd0lOOZtlIK?RoGH&*kU9UT9aEspq%HWy1Y>9pJw7k8e*;Pr+Vx=YmDc7|z>@BxXCe
zNZi@kfnY+CYd-S!r`hGvTq^Q`;sWY{)$|HX|DPDC3Vq+d*OZr|AQQ@2B|EOyPtU63
z*|6m0*EXlB<dYj)TJBH@3x5d`L0DP;H00)C$<bR_4lu?#-%ikb`t)>QRB6Yx9$=1=
zxcezw<_cC;R=O<-@B`=|S_wRS2&x>>AgW@sE)%ODf1NA2_j_k&wEXhI!qL$Y$Pe-h
z3+wSJiEZw!4i>+DPLbXhKXU`T0>vwbdQ(eF%hkU0nOg#C>vQ4(7%F@=a>vK!#~ok_
zZL6wG@@_z&VZ~d=qkDX#t*s5zf>b}Qbe^9{2s*$1brMPW=-IPp!0AJ%goNtjmL=&J
z8Na(9%C!A@7s~Ri-nA!z!*;sflAp`Co4A|cqN`X%9RmjkLHKN9*Vkmi!zb!ta>!E^
zK<`4Q;_BMj-vR<kPQ^5IPEsVpRv-HcI&MHo*7aAvb!>@n7Fd?txbee(@IEwN0EsK8
z7gGIbu{l;w(w83ZMT`9&<iV5^;YJNolKF9^m*PC<+Z~jqSXc-l=ul43s;jHxKLT%N
zW@d&eqOCo?v)IiBMFw$k!%4jbH+bZGj<vPL(S5<sA^9aGX6(0+GLWhOS-s9fG=xF7
zW@GrPN3pgcknMowU^)oum{&0G($Ot;#@usucCOGc(AG{A5|>NiFR!T~MsU-FCr(a3
zz0(rPFbugF+$t(s(fOe}{*$uu4xMex+jC$-@Vluo6Dy_(9m8=hzUU+h&i?!fYb}i*
zrBH+MslD}deJwRLF3GCIlS5f~YY#;J8u~R-(obLY8+?6H5!aZv_PedYBd3E)dY|5l
z6ONCTR4ZLt32w6AU@EWQQ;1D(T3A@vE6Ha+B99-rN*QQ5Xz=HmR>QeigXayna26l}
zso#ytby&sC%G1&WN9C!!?+BBz5Ru+eGc~KR3}qzeYab=Juwmfg*>4yh@6@gBS_c2K
z449sc+6Jt^{M6(^t#)k@NxYy#v;U*>Bk_Zm5NgPi3k@hM4?fic&58RjM2Cll78Vyx
z&CSt7baZq?L`1-fFW|5Ay-A`s8+<VufwGX3C!3?fdF(+qHZh@?$hio5hjAWZk)g@W
zi!!>Sj2^#QaUKStAu0uqVupqm7v~U~Jgu=Bhcjw=_62^4J<K1Q`V<qBh=Wu|-ci4D
z8~f3P)qTRDkp?=+t(9wxoGJ%DZl$K&cJoW=^*AUhQcQo))&?*uE*THgc4y4U#ll$D
zlfE#@QUg(lIV@HtF!Mg!gY4lZ!5#MXI%3c?ZC|vX!D1xjSJxnu$~heo7S_*%gS9HR
z^y|pg`Pm&rtF_fCcK|@~tpPTB_-CMcy-qfM!NDF3mb2e=^m4$R0cMBuv)G?0u9_?N
z>Mz*Njfv{+h|Z$nVUN+Z`RK<q!EfGp-`Av~Mp02UT>bH!N&RadRrX`hAgrmWVTm93
zJ)^&O`Y9q1llxPLqKTpM;YqdCXGrW_Y);Yn_`L1!Cm!Jy;)yF|;GmqY%wr4qrL9Km
zWjs8X&oFgQ(VkrBZBb3l2qT@Tx!cKnN93cE!H-Wj8*p)#rs`h_a&q3g_k`g_+Jjeb
z&kQ9=f*&(dQ)~J+?>j65rGx?jg&2sMP2?9qE=Z7{A3=g2zZEuJgy^5LzA350kKWqt
zX6CyN5}(Eg9NWjUO>f`4`LFA3eLCe$o<Cc2Y;R)@i*&INMKSSC6cl>f8P6>}3E@Yi
zq@?HQ<#|O#Wo2a<83qP=sZ+t%uLcGh8X6V?qzjUg4<0=50x1(X)+pY6%U-J5+P6rs
zkhgn|Ae0f+Y2tC#)B1S)XGmYFy98lM!YwN73aQ7>?PUFEh4r{SUlTYu%wrLZJgZwe
zT#K^``n#H!zHl!S)yDwpOdzdeb%Pef@}?`kYk5zUlx&=IT&q7jS?~S)x$Wjn`58(V
z%V8#Q0)nX%|9E0Dw?Av@<9n65wN+tZA$2Z&;ct$lti3LTZ`@#bYQM2EU}-xYd$6K>
zzQ>5%T3#j~3U0nQtsGZBI~S~fje}I}ITq8&W;DI(Tw3h5eJt7aC{5^<eWN!DZ=FjE
z-wgkQ2M{XTg0)0JK}ngY#YaU&uc_HsS;;K~ZX<uS(|y|`J8T@E6R%k?%hsX9*NBM?
z47QBUPfeaYDNqYxG0Voib*25u^~$u`Lw>UGqI^U7wGsKZF=z;YzFhPQSH^1DFsW>p
z6)bXPS0T7{;nXhejLQHD(9Re4B;aml3`9^0(kB;Ozge9wH(lpmTR%4S=FKaiw4uRW
zdB10kP?fv1W@!;+_FX71r`mO{84S#+p&9RXi~PkW%fTDSPW6itD0KVkiH{T%6zIR+
z4k&74R_uX1MqFw<*jcraZv5L<(Xm>_%C{^%JSm>AT;KIU8x0R*z#~bTsZUHz6_#WU
zJl>^dC#0d7T^nw+ZJ@YLTqDHrjfb(*qMMj2__1eB_SVIDY3o4wYIi4;&3yzy+I`Xb
z*T*}<g_?4>7@%UMqdee#=9bmuC!Ej&@VUg%@xI%^6~tb@z6@nn_!TOWWL4AI=z}QD
zZ#?xp`)p<Xc5KYy-8);hXGL-{ABILthx(~PLQF!r$E}Rn*aksB&~Y7eqUM!s4tPP}
zrNe~+AK)K+{HZcSaS;*ktSn<UH-J^h$uJ9!e+7gqeOtEop>oU1U7{2~M}`?0#}(5{
zKDn?3DBOE!<ptE8bvzO)Fq9%Wz$C`@P3Scn-Tb-#l}RB5@$y=YqON8+xQ8Yh8yFB*
zs)d3yE;diq?a-+d46RHHw||Rv^23r%e}v^-QeB-S`!1oagKu>48C>Xpcg7$FO6rSy
zXNM>F)AWM$^tl_2DC3UO(w)Hb9too)u(Zs`M0paiw>c$9O+5zzm|+aPPkMS_;fvv{
zC-S2g27PJg5Z_>h*YGegO-<=(siGlpIJ5XuQqIp%3?Oe0@(!{J3hi)J!DtC0(5L>p
z{M@H<*R?<`&+8&r7Gl8#a7qb$E}N`A-qmKk*()peMn+24OS^d9x*u3Y1U=!qqc~~c
z+5shuxeBQQp`HF2RA~A$RfBmrzWIfP>zW#plsvfa-!FLhoMdK}y}WCDe#+W=MHj^z
zEXh5^G(^Nu<t+eV^zuI-1578nak1S2#h*2=zn-cXD=0*qmFSs0i4>dpIy=u5-BR%3
zgN(eqUP>%&PZz59``8<!gL$H~0L%1sb$uu(=v0t(pYeX+i0$h;`<-HI?^NnHW3bL0
z{hi&?q~wXcRaG&<_GI2yi;GQNadvArZa+Y6Z(op(DKa+p>grmvw!}F&NIE%TyH12;
zl;r1ga_>c#Y_3~Rkd+kW<_0>ReV*&jG%$@jTaf-9O2U2j?$)-?c;#HY{_N@TFb0NS
zb(PhhnLnD!D9B=Yd4Qy3_MRicURaV_2blU&6gAa4SG<e_9o>QKJE>puGf(U2?HrkT
zxbBzQPU+ZvN@8W5U6qaqL8YW5Ct8Kr<MO(gmE+`a=0-w7LO>AthRfWn-nBfaiQ&~D
zD>Jj)fdpubcg`;!ti`?BeA-m3d+g>`g++i+z38}l{^r^h%&dCzz8wPh{I?(l_aX0z
z^UDHuKDLlx#-49cw@4r4Wq(V0yB9=Iywt;y%<cmSV-fKUO+b<ec(`|rjl6qH68X#P
z&-KcXt%W8(uWi>dBaEpD-Sy#tt*mwgA+^5j5pdtri6>ahXLayGB2O`yn{jCRp;;@o
z^4T~a(l}9TDfH~6tE+@GPNL*p<pLu4-nC{i0RQ-frL%4J^{&dsvz|E(()X$t8mw!K
zRbVKj`LB;!azkmm&)-ACErZ<B(yG<O3bnN|cRo6y*YYMNQf_X(Up>bsKOtL8!X6>f
z1F!Csf)A~#;^yR35f>MK@+6N(kW7psL^5mOk(-})HIKNes;;t8uAsnkViI#Rvj}PV
z>g~gz(yyq<g=*eWoik2AbM7KWt<3-`Dg!3w>ORTQomAoETDv*6_0dBj+@`72oyGXU
z{r%EsUrzR#&>XC-kB*lE5D>6HN8Xnx08Y3*RFLiMjUtrR9AoQC^-eL})2am%$$z!;
zcx?e4DKly_(3_eF42%WUU<>Z?483S6M|y^DP76|98KMXx8y`_z;<>XR4ToU0()=zS
z9*Tj1WT3`N4!M-EmY)Lm@5AJb+$NH?JLZ(ivD>^Ymq?>BSJyVg!GQ-Z_vcTj=eZXi
z#;>k!>-iUdV<TgZe`0eA|9d*$bU7N59T9P?tc*mlTWPiZYV-J>QgCjEG-sBT{b@5g
z>ht2px|09uDC+Qtqiy&mV$h7fGJf6daD-g&V6<N7qhqHE4I?AtH6afj9wYzhWIl9J
z3+bGUEHq@vxP$B)>p^yQc66sG9?4rgycd)Yhe-LAHdeIg)(_uFU1>D}-ixiy;Pm-^
zsvtpgi#s;@(2OTJ2|2meGn=okA82X&g!k5LZ0xlJU5CR)_2A(%asBj-b=_C{&*!%1
zCua@ZBrpik83lgYY#Sl$w|Qig($7sF3%Y11rs5*M?g)SQ^=le@&vc<?i65H#vclCM
z<CeuWS2uyZ-}h&?XZI}2b8^%Met%0L=L`A0vH9xN@ZN8C+Pil%KPTlk<f@l@ezExV
zn4O({`_P=2m>9HSR)1jh^s+Fsl&Yy0U&Sd{AHHuo&KboQ3i9PZ{(jfe3DuR~j8H_v
zv*y=T^AM`zfx894!5E3BOZ)5oM@I#gmVW2Elt;ht?CcOSeAZ|Fk1qUg-ZwaZ-9Lhf
znblopcoJTsgfylJb0j5Aot`2huwC6d->%gQ@xSIb=(zgx^tkrVOk_w18PD?@F`cjX
zR<6Z#TJ3KJMyW(g(UQdHidUdCHV(%rD_^Vg_)AEJ?|CB4JJ{1CdY64mI-XT}U(cow
zDkg`E$&r;WuPu2@L&|LRkE3pEmgpHL^gNS|_pRiLl45tvFV(7t+mG*jM{tv0(~lox
z?eu5ABoJ4J%}GIWas;8EkiQ<&ko54AD3knc4kKB!rTFb_L_|h<U}wvf@&aT&yW_=Y
zzWSHHd}c(0Y@E&oeG<!5_Gm#{aSX{ao^A_8vv-hA=G|Ym`yeCpx~m&lv<W;fdeVeZ
z9$hfkzntG$8$K)14^quD7%LO-#lCt~A-~~s-gDT`%gf7v#HbbwnNXxo)zmP%yJG~T
zq!{#5A?#cG#o>{Wf)4H8E!*Q&oj<$r<m76?$ast=KLSk!7acW=MOW?}D+c{we_v#D
zoJ99G_Ndg>&SEVTmp{G%7=|XX$q7>K`!a(337yL%i%Va<2nmqE=`<Z(Je7j2gWrzy
zF--<)N(}AN(i+C?D|>&~&ez!xS6wBM;UTM!Mqi@uN|cwM;o@>H^#ygu&xNvicL$7D
z8jf>nRlTXQ9K_Vr{M_&;FN5mv^woPs%I?mX>NMf3wl?j)6m2t`XLj2(cKiouSg78~
zrECpVYMzES(NSBnY5&s6A|jQx`kkR9@BbclqG73siLE5_-YnM|Dp2)#{9dI%`s5_%
zo<dJ?;e8uB4(SMT3i+LbG$P^y;ot?2IkC*34&8Bv!?#3*t&c}a8&Hwnp;*=%8XfIB
z?g!||ST)}JEa~od;gc1MW|4R5PdF0m@km1S^f1H&-|*TDZEYdQNq=TaG9OCD)`^py
z)8g@w$>!Mh^J$^dxt6&mw5XYt9s?0qepmOF+L`7YcvfCsvo~#&S1uw#=P4hZ4|P0w
z*fKTc@g(w&-S3|$s)MYLF60PXA@l6(=l7_>mO9%Pl0C;#NtlksMOEu-lM1Pvr@M+&
z(F?m2cWF{=r#?)5;&Si1gG80+=7Apraka%=`HkP<zP_}Ff4XIyUu}!AaT2{_@x#DF
zkg6(+^Ye?6!ooVcCZlN~#Nc$x&*8NY5}Ph8dxux<fMxs5wKn-p=Q*1h$e!4wrCQ@C
z?H#cA_3}nPvTf`qT<XI@n#U8b!zIJ!o?=p(@cl=gsNPK#rjnW^M=$x=4aG|Q#M`PY
zKQR+qSTufo=}XS{m6F}Sq1L9w{JW5W873a_8|ZDoZrBV^5zt2_GTj8d1?bIqlUeHM
z^mTOw0Raqsucs05z$!sOL*skM&7v8v_}-U7FO38eZ-cbv>)qZ5=F6VM#6xYNDBeSP
zk;0MY+S=0V6Ni54N(}vQZnyNFez2VxVPoF-^+)l&lD+L@{_mgrUdkM|ek5{97#g;M
z!%#9b&(qenoL@3~Xh#*^i|O6m5s6`DMp-o0kw=hDiH?YhiqQC^Q!;BIcO~ofH<l!4
z8uZDC>^*vJd0G{%CUCq<OA~;q+7R3cA2ggtk2YOqzBt%L#W^`Ud*UF{(iM@diCivM
z@WT`35Oz`_EiOS_K9*i?1`8#-t+Jm#lV6Mq4CT!|X$i!r_gMe(s&a3s_d>#Xht_Hw
z6^Tl-Atf}hvH4nBRi&8zaAbEGuk<Mj0+Q&_XMa`KFG@dhCz(A>3EXgtjJvtA+PSo@
zkoMGNmFP_yRoo62=SCp@`r)7=T45mzDjG`Ts{VzMkcSMR|JJs#xR+Nboos^F`Kpe#
zW&JBIZn6{!nx>=e3-?A;BtW4r<`Dq{4eCg-AfIW7q90Zy;0<|8i%kJ|Y%7jTanOOJ
zAL4rt#|(V258o#1Khgbs?P)k#lJv!mC&_kY)5du*#I-l6LN=bC_U-om`c^j^hNL87
zKziHiT|5@D=(1VQ#olkgJ{!s=;@Pj;Fx@fj$t}8abc7(5-=ZEmIsF04aQ(XWU~Xr~
z-qX4obDK>)TU)V*U12S-8%wGU(c&cMFQc^6kCsk6dL#Q|$QLg!ng_?jp(J_*YHyP#
zutEa~TuO>=BqgJzp4%))tqZ#BJc)>9G2%>wyUFqKF)uH#qMTfId3o*N54b3Ph29=D
znx^-ZJ6coX;~`(khZvM=3EhbtnpG=jv@d-Y@%miK+Pb|pxZ=UyR}cA8Jy1EG*j|(+
z*VJ4;5!E#?HI=6s#|fBS$+uHadiZc5Pmz%*_}y2};!m#@R|l2SNl<3n^ObSV&qtp`
zY&<l<MlAlGEUm4_NM&ydyN_%iQ^jdK@{n=-bKXP-m4pRqs<N`|>=vx&&oVR7y1QAp
zxWb=ApnIzbf2qyQr6nqu*cgwnsyUBktT0kz;`z}f^VtMvV-ly%1&`OdtK?C<$^It(
z@NnZ?+2j=ImG*-dqwkn?>Gdy_>lIT4cs#petuhHs_;wz-V?K#!3@6{(vrJSPt82Q5
z>`UOt0Z4njl&!0~?qXek=a^3K^H{yB;pD=`FdNluX4;+G!nmN_r=x?0d8OaMso1nU
zK|Bc6`&|IW)EGZ2jjy&+<CQB_JNq8uG?r(__o|Fc-hCV4)f`FGP^UXObNRJ4ywdu(
z0W&!9F^#2##lrG32y3D?`r}-`Y#QO^SdH!R?0Zt8@@R?r$8XbvT+^M>yM?9Mg&O-i
zdz0&%=OZ(e$hRwy0BtX%5Dd9E$}-y+YuCRUz2bFl=?vodUgAUrKL%sIm;w@5EPj^h
zZDg|WHd*(DN$hm#%-`~=7b8}qKgj+LYa$uKp7jemrc`g*^G1=}^Ybgbg4TyNFa5r=
z96MhZ77FQzz(7<9BFo#`m+&#2rD;Kt0SzguuI}Yv|7gG1<r;7p@!^V-T+a@1Woqiw
z6Wcy4`Q9{Q;vPELq&u6Pxw(GdQwMo@;`#<4D#{p`ZMkaxU~nMo_uH86s%jLGjah@U
zjYJJU)#hQ4&B7zN*gkP0ARs_iIJ`WUHpUSH1A{2V95I;?1|IdofdCq&;WBrZ;?5Y|
zkaU^q(_?d60lkrU^@Vv=L7)PRUA;wEQu^ni*Vp=trVLS>hBHiBqLuZazQjTE6{Ydl
zfS1d7zqYn!io-?11-$CNd`V47NswET$Moa(x|1gC7Zr8(@L_Uqui>v&bM7QKBK2kO
z-i-iFueq%3%e!~@ikin)uf9^T<kQx4MgztAAch-nLj4vNfK33`R7}Lpo|FC7pF(b0
zTBqMSveDR9)+h!dqCPEKb1cJ>!lBAp(*V{YzF*wE+;+{P8oVSWC9_hS#>bBVCWYaU
zei%hMh90iN*dCI5UPnzAp=?RI0t;j17wn0ic*IEKq#xFq=aDGy3^On9dc6J>BjZ#O
z7L_d?PExWxGEOnE+XsL8xVc^op_6&+KNr)}8$Uf}f1%;29L>!jrK)-zu`=%|^Egj2
zwit74<Gyzf?~da9l6+wRUNY<cSVIG}2SE(X&0W4!4P}Z}c64>gI6rl=w}+dtyo`*!
zi;HchPHf~+k$+GSfa!m)YjRrutU5h??MwAN{ZRwt>m$A9whVn+K-WapypNSspeg`W
zoBL;*=NY!ID=`@p2E#|MhB=jTt=Et_%biIYZ)h0mURSQQkr)^VkDvM?tES;e%bXd-
z@1eu1t*zZTzil%3l(p8d0AT6HMv>shVuY;_!T|eWF*HI~XLEnc;?>=55@tfq#8yrH
z#hHdb%gcOVltXf4;}8DK<d1Ei{p}#2fF>AZ--Y%kY$u!?Z<IK%4HW=310`bT=jXV@
z4xdORB}=u7-WjWkP)YH!=gY}38I>RWnS=*82E<rW0^r;cP{R%Iy5>LRFGZ`u|A>?G
z#}}sy7ouR=2uZ=O2BL!!iKY7K<u#u5Zu<a1>VI*gOZVz%oc)}h-dvq?i!^6#^_q$$
zp6<f#F*jK;Bnuy!L&6i<MOBo6Xgb{aCX>GD^15tJ76I0*A1StW^!E0#xCD@g^pc;p
zD;XSYXq1ebv|{IbDZCRSJvsUD><iN9x@Cak-R6ygAFsj?TcDc9dq$k2S$eaepkZP8
zg_}Fyvyr4jj(!WJnGd`c{j0@=g%f45p`o%OA^?<~nVFbeLOa^F4L#S{*c<}H7s$4D
z4j-AAR?jwE=<T(U0N3<BGxO=rsJ<{rf}kY`09yILflN4~rM1_(F0{KgL<2~^UdKD8
z;^Hk|!%HSdFEF_X2$Xl@`;&4qqbnT?oJ1O(C{h_K%~sde4s7ioNvA*bfReP4N@>A6
z)cc~=OA^n4g|hQ*is_icg!5Ht`iw7q<T5()cXNvC@V9$xbU84g1Mw%hDJetEbgEuE
zI8;Pjt#Un6mX&?c9!8pk@5VY>q<uHp&ffmB>$Z{F{WQDxCKRz-f>L!=35uh9loRP{
zGH=>gm@f{x+1Sz^$`kraG1RvhRM{5;tPaJZj|~k4ACJ)lKjt)S_V4fSmt^?~C`C|E
zQ0?gy<aB@-1}VnP@yaK>dHM~WkehhX>kKrOTmb0o1^|48SL&d3(#D2Wn|?&--Cw_S
zD&r&pBFDWRqwcWGt0}@vJh-|id4`LfS;NH49P89p{R3_lkoLR5+J%SDaz0j`j{!NH
z7o#QN5w>26K{W|8j~?$iT!SKL4UHP9xq$FKQ8l&L!otF$FXymYczAfGPllK(N;08e
z&m!>Vx1bYVcj-`Wb~c>7wr}6^d6Ef3+bH$C>U4p94JSp{;yR^}s~t2BNF^l_%NM^%
zLXzy^>fSxW1xYsbLJc?v9w#4?nlIUd*wyku^kz=C<HF(R0%pk9Q~O8Z{v#G*7ZCnW
zevRtepd$v75zt`jL0Wz0?+JYER->2QYz7*XaPA;M(G?-4zI6<y*xK|-CJ`iJrD%Ux
zE&a|}lcXg=LBb_fovD~asR;Q61-aH{>OOqO#l;oae6yV>{C4-cJZvN=>BXNREcy<K
zRp4I5QUu6ENzp>vCg`f90Uw<MRvtir&_^M(0!<kQ5}mm!1zcu5O7MS%P<a7*P&UJ|
z$Nl#jORW3}#iRoMQ6>G=urJA#|NF1`nnv_7ohB7-&>2suAbq^oo+9WmjiL3oxdv(R
z8$j*JWDvwv_xUTitW-LY^YJO`w9mbu>H)%$z}upPP=*ZcSh}u0rUNn`EH^X)+7MAg
zjqCu5)e*y+h}>tF{7XWWmt%}RmoxuAnrcEP=^lWWLABMCGZDzdi@V2NSD!q1?cpK4
z@9XPp!ZloG=Dabctfj?pkza;6l&3>B9O$-5tE{YSV`Bqr<cEn55*N|EY7-RVhSvGZ
z5t-cnWUz?<{vly~nqOQDJ&z!OCXc0vw%5|pF*7#@3i_SMj^$mP{X$Z}oc<nI>*-B}
zg<ToH9l(=8Km4PL7rS~=8WwED(aq73JHGpDE{HNEB?X&jn&ZX@X68G=K%xQ(qpwcu
z??OXEtpr$Ezk>wvr)>s|^gbvAw&qrydTw^Iw+iW4H7516kQ*c3Noul#Kb87Y7a$!C
zSaQ6)rn))^U4wh7o`Nyq3*C~*Cf0+j^F(K-H1F?)g;ul5IPypp$RjEjs6jjdM#|g!
zl9Qp-kn{j11h^yea&ij}|DVBW85mYz{EF4=Y+B%48W?2M2nII8pQR8F9-`rhkB{%3
zQEY_56I2k6Rkx8!E&crY=fs2{q_Lq_UuUP~&O)a*&!8NBzIAo}vVvINV0c&<G7Ge2
zNZ&!XyZiau5t(JYLk=BEG}W%dsUV0S*%2e}@aZX2+=oEh@7sHOK!A#s_k!*%5SBwa
zQHu-rg@|>bgtZIwL)(QuPr$;{hkPkHza3TnVZ=Zc?~rqc5^=Q!2UO{Qb=#NWO^uFj
zGAbY1sy2l@dqikxjQpSlH%Pk|6a@2R!SdJW3_0Ap7W^1a3WOfOxTz3Ae_n6QP=vof
z$q_5XLbMx2Ee@J0aHeR9I)UtwC!d_V@g_DSKc9i0pa1S%ai%z%i5eCvDsP26<6ed$
zl<H#Q{YhzHNCCqG2oTU75(CvogbMk$*p~~_mAx@wo=VHXF#tWaKuiD%7~u`$a||6Y
z{O{ktA+!J=7z#O*%e~su+e<!8mHnTcO14Q0=~^pCwW1FnXoj3REJ}EX>|p+gL<v)!
zu6o09$&6GA{da~}9DM%k98q6#ow2EJ`u@NBHPh>=i_0PGF|VpypnlGom`X?>wqgtX
zs{b7Dv%&1#+^b~V77&a8y#Oc?Ku)YU2CoYq254_Uw22{!2oHAxse!Vx1a3>vvu^_g
z1neCm!W0Kn#|Oytg9a38o<Lv$d1Zi!b8~YezIihoa6X`2A$Aa<0aqTX!=T~k4rm!x
z7y(BSsLB(-xQCXGMa9MDPz&;6gc-syIBsA<fNcjXmHYcHRM`>vT~`t6p>u9-Ztyj-
zG|>RP+@+yGAR(~}d{6MXsNOJ61UcVT<dPm&s^1{$hl$b%jcZ`wS3=W}=>_~oo+cUy
ze)*I7K2dJqWcnY0W%)5YjDg?=&?Kv>sV&USQR_1$GcYhnL45i0WweK*z5TCkS2+8?
zJG~_8K?AQmtO&w?ABK(loV%mBnTCx`k@xKM*kRm1F>(B|`GD7jCjvu*B?QnFD1K~=
zjGf=UNhv76_jb7_8BU1W_@W29Ab56m781vhlLkW*5D?JGdC9G0)Bk#Uv~8TJ^zpI*
zYM~}x05p;#0mr`#Lu*53q#{dFio(!HQi|p1<fKHaym@A328xXKRt6jeC=iI*K<*o)
zQEzi`Rdz&DQa-R>fI`0*J*Tt>4!KGULe3ir6=+YSen(;ZDmh;PuS)1fp!<LCUH@nP
g;s5Bjqb{&^$l4xHD9mYI_C}VLP<UJ-X5jn30AEJCM*si-

diff --git a/docs/dev/articles/examples/timeseries/timeseries_anomaly_detection/unnamed-chunk-13-1.png b/docs/dev/articles/examples/timeseries/timeseries_anomaly_detection/unnamed-chunk-13-1.png
index d3ec5b90fe82057cb391423175ada4ce7abe973f..d40fd4d5ac288bd565fe6782d920495b8647cabe 100644
GIT binary patch
literal 36351
zcmb5W1yq!6*EWobbayw1bazRINXUo?L#H6!DN+KGN`nXr2q;L$P!iH110o&5(B1u?
zgZKTs-+JEf|JV9i%Z1Eb*PQ3I&mG4;_TCez{YV8DiwX+`1qD}4RZ$lO<@Re7lw030
zZi83y*6w@*zc8Iu4ct&ru!)fWZVgWdUZ9{bpr|Q6(DTmR$nt6+8J)f!>~m<wt#!Yp
zq(rcD@9B^&Ntucgt2$8(mH$Rm9Iqi=Oj*3`i0Qqxk7_%SRKiqmpMJ;kysL-vN}cq+
ze$P_%iwbyPkbj%YltABmX+ay$<MXROO-EY;ncJCX;>&(Ls>xP_s1FEGAG`|YI=cma
z6QG{rkys;tV-m3UVZ36%AY#x?MP9=q_nOceF605Vf$2jan(*IS{Y$P6qD1n#g(Um(
z;?}VG5}OKxYCqgIKAe!v83&Qx%m;l5Ra~4&>Sfc0kx=Lu1Y*@RCZt*IS251zh6YLO
zT6Is~G?vG38-)vfO~f%T@lAUeE*_Ff|68d4AzPbgsg*%erBjk6fg4i+-0qrqoDT<~
zh}VRINsLfWXF&*AJ<<j@Gk$I_F{4PT#7i0>L*NwEEB^9mO#Brro&bRy$|fk;DTlRk
zeL8-u#w2)Lz};K)DWLch*|%v`jBe|SzFgz~ysUS^Ah*(^3`7u*65RTR0X+Jx+hLSQ
z+1y6O5(I*Hg^Katl94y2X0Ml{5lQ>T{Vw9`lfL68<!LH#A+^MdN_z^leYzy|BiR<u
z+u6jFo<@4*0k6apIQtl?8=1cG|5zrvdmfuA__|I{Pd5>_EuJOvMW-GsEG|@U607(<
zi(21c-&9iqR&^!+J%J5JawewGDzv1iFI>G*x&GVK;X^mawu>>S(l_s1yQ86&h||V?
z2FD8?xI5WcXB4}s%s_)zamb#w*@`JlDkIniO=LOE?&90BjOD6;Qt%7f#OJc-5NJpz
zvfL-ZI?e2IF1hW7#-F0B2~P!rexuF)PV|#=WV~WBi&`QJf~aFk1mOw}IliKg*&2xX
zlcsg718bEW1VTTdOvBF%#HIRD&g}n>{2+2RHp|J}hKjOx7tn8X?G{toL*t;!!?0O=
zY7vK#&pH|@Jcri_1lbReAB;(mA?nn0eRV#+A3+~@vhp#mb@?G191j0jR8*^?;D2*k
zEONiKOiXH>=QG9K<^*Z5*tY5ju5b2*Np>`r``e_|ES+tV`f>c)KbOBeaGa>boY7SZ
zqij+!D?z_e_S@|fRrcJy5}TE9#6VVbt5LT1mFIn61nts67k}t68PoIa($QKS(h)_F
z;yHa1gfEH;(pVlmfIw1JtC2d-{j9Q!?@bss`J?wzlsqSA>!o2b4F5c5@LU+V@yV)I
z5LV*FxJ@?q({FR7L6^lP>uC>_;X<|6hijwe^hCS2km?pM7q6ZNOpLKzsNYh+E}e~0
zam)h(arvkRy-`yll8ubi%~qW31c_VZL5x64Z}*mqI8LRivdozfShVqJr>^7v{UB3=
zU-~8iO4oP!MKlFddUf<gN_zfzVAKl7X$gbmeMR!n@p3LrxO2Oa{s%sAsu&3Tpc4gl
z5-Mrm%1aFrME%{HMKFHwy=is(w%Pg+I-baa=`GW4awE9*$Lgv2rN){yj|ej@O7L%9
zL$xq?a^GxgZEbC3<pcT~hC>vjIlcDAifD}{=MBflm)en&$U3@=Bruy>-~~edP7>fY
zvYZ`+N3Wgqckcw>#hQnI2GZRd;2*nlqbdtN2R~IbHY4TvOGHol`3+Z{l1nyP_d*yj
zblH-vs%mPy+}&RX60i{L5Lp-7&u2TnKm4UUg&K!^UyMz8W5>nHkak;kx1`E`mPsJH
zcL}SPW+UIG;x3%KsHl9p*ONKEcj>pIWi|G_D?XKEhglaLo_xEt`T=L=LWXkfN5FE&
zOG?)JB604?errq|6w(A9@tGx}OWwzHfWaG1*w$EIVweo+ZY=K;i2A(vn+u^Rg7a~z
ztDaXL&%l%-99Vedmm-eIRn}yyC0FrkoISyW`}uYP&B_%-uskMetS}O#(-~0*{A209
z{(ZV@@3?H*iurkbte{Hz=10N$8T_I)C~b}L`8cAt#~qS4m91glsuKEno>~^$IXlW^
zh>Ss)GV<@M>9Y2x4MyTX2Co&vdj>MQ>PrOgl@}HDrOz8O!FbW?OH!~XL(}Sbjk^o8
zZ~S3Ke)p7!05-uJj#DgcA*v;wz9;7w*M7{NErZo122&Lt{YBict)INl&+E#{%690W
zW4LXvl|P2bZpOcV|6ZG&B(9TkOEXUw7Z(=?M2Bj}>4~YSz9%QCn?>N~xtp6{Tcb16
z12G!w+8P*4k}9U$iong%Ed6U<UfBGM`i6CdfhIgXJ)L3;0)gnf(|(;t=i%Ye@n=_+
zrPZ0&%@Vh5Wn;rq!?L$rAPpzrzL~)%xE)+5x*AQAC8IyQZN4G7mz9>5HYO&<q3Wf0
zf(LYab@e`H!nCXz7a1q1D@RfyOC)!QK-xntG7J0J>KBv0Mn>9`AAr-ad&ZGu&e<Jq
zwo*UvtBLiknvzZf83UeMU(SPb&6TmT=Syd??#pK$lX{t)$hbi8goB=6gWHcTaQJ;_
z>iWWTcASKRr665dEYLE%+7s-uHYfTDse@EIg}ShC@S12Xx%ImpUzEZBGtMwk(4aGW
z{<V~l);GUzeBHl6h$&Sa?^y0R{(lFae=a{yC6v4|hgT2kls0Y*_Fp$lh!guiZ<;Cp
zeG>yMjIs<Y@;}i{kx$J0KW`rM|FaPg?EhLnTIcuIv^R@?2QU2H<^Q~?()YjKME%br
z6jEuouHU#Yg6}V<ddj)f5?5><wBM{wgrWR~;^)*9*V%o7-G5e6?n{AsQi4stLhg@=
zhQL}a5j^J2Uql4Xv6Qz!wC%DK$hSiT*rn%N830X2{k&dpmmX$T{{UX8$I;=;9RH|b
z92l)VMYtJ-!?8ROuq+W6Fla07u*g3D(s*El3W+QR)~j<=;R*fL!kQMfDdOeYhrhSN
zBzmhkp2CHYIA*<mApdE;owWdE^|Lt}CIx=>vrM2*-Ij@08NEyDVPEXhk>=a?>obu&
zS8tD{fCHtT8nCu*qeM-(B^g5--wBqs7)`7q(~*-DIl^1Y_=L5cn<_>C)i&LlorHsG
z15fQg1_CsZDZ!M}zt0osJ0xk-`19JlbLzZt7i~3;c`TQn69UJ)8hfoP^FdT@s65x9
z>e%5^u4H{iXe*6D8=q^P5i8|8RThdZ%>p=ZA2wj!CnUg-jCjc7I>(>ia${1^%KhTU
z=vnsEVfd5Upv?J~T=sCY7Oh6V<7#X-7NYzA%O><7F`a)LDBCl;((@_9zZ+q@Lv8G5
z6T3g<vKwO))RzP3aXXkr4YpO!$=E&v?cML3lK~MB36>Q3e4?8-Lc|HIs~M?u^Zp)f
zI33^Z>xqEv9oD4hMzl}xZ(c<WYakSaK2yz_TmeBy%H$#pD5cork)(i?oCwG9^1Eax
z72HIMMI<-i0wWwtka00y0Z;t$0FKG{?V*HI`<s!!q2N_nbGvz8D!=U2ZruF&pE%jT
zlbw0w9|e!Hda9vxg~~%oF+igzcJwtwm=d~z(doo2e!LO3|HVQSd31DCU0tn0AE0;e
zmMQS%jp|XrIH^PHw6bLct@{{4ap+|-2?R`AUP^A&S}gS>)t(g7$<2ab|Jqc}aejx(
z*?`<JdTp$9w&{E;dNObAugWK|yzR!9y@)4Z>J@k>n4)W<mIz`h{s&HKAS9Ma;CAXg
zI2jdn-U$wf-Fc+v=-@zdY4jQdYam?3+#(MDcq^aw*MvKdUJuG%8*Myhy!}951gxoA
zKF;O;t@#6{D9@PSjp;L$#ie%0U=k60qX80pjT%PTeW&*p^3u<IlhAQh=^Kk<A_8*s
zF%_31KqbVCO-+#cyHaBB(%c<jb0~lBn8Fr(-)V5~=73(e^WmJlEO9{ni=I#cI7Z<@
z2hYtEe5{-4HREH^8B2P}rOI&gJ|QN?{$lsr-40TWze1!F#fI%v#d+Mm3HGXT_R|;@
zRtB|vFuK3jNBWs5uc6+!mV}plFr&nnk_}&mc19eFKN-ieC(1UeKye!mHZU8@W+NCd
z7NSIF{r7t~OWl3=V~f^N@tCdu^os9OX~A~GsjrqJGK{_Li9MHrS2h}+nVFfM*3@<<
zq^42s+uM1|OZzL=vL>LBXS~v_VGjjJcnu(;(AEd3d5ZQy#g`!}TS1756T*m&eJp(y
z6?AE5p1Ak#=|rN7it=^uSNjR3;Y`Ma#vdJdBK$bZU>@sF5PoUhORjyCEF9%Yzw?|<
zc8LLO-X={#W;CbB)W@G!5DJC%9iPQ_RVI6QdBF|UtDFG3(;-=oPXnM|hc(pQ-JNfi
zvxr5o#_7|mN)D=VMWzH*mW^3w0&C{FD#?+vkn4K0)}GFi%dv%zJQ)OKmlMVMrLBhU
zP*EIa;7d|9*xp9vw*TQ$*msx6A`d#Qtf@I(g-3kV?{(0Jj2iKbfw<<(D|CkRCNm&3
zgZK%;9LUJ*f%<*@`W5H+iMeP-KfpAe)geVif4$3R5R%EFm6#G}O`f6L1Y3aH;05??
zjmcJ46$XeR8-phb3U}HXi;BYK8LQnsNlHpmQ&Ig!I-2$R_Yc{kMFPTxgjPq3_3iD8
z?b7Rvc*vqj8^!42M1_Rj6)~^6KTBk(#_6%H_-Xs@Yry)-imjugbuR-#@s|*H)}xxr
zO1>%Jy-U^|HVr|jwi>s;Rb-U<^`KRDdRqF$54AhEkhdRyHW;kuigUk+=(h|zE}v#^
z=b2i%Zd9~oI`?-9i!_3=*0t*DL{6(8&HZv_g7;b)<j1Y~JFzxP%swrSH1c+8VK%Je
z(}+QT-*7Bl;b=7RIW-_w){867V+N@A;obKaVcOn({GBRUuAAHidUOOJw}F;-()qoi
zlYVyanBl=Kd<qc%3HxNuHv~;vP=0C!qu#yIWV1P+OjX8zBJ#gM_x~1=U7A-ZYcxjx
zabPa@`;useU+EFv+z9~&`ETHS=NEWSck-_ry#vB?*gyYerMwRe9k>z-5WcZ)AOj%|
zmY_NMUnk7|Z(#m6i+|-JYyKmk?;q&R*YoTmjBZ=v*59i?*Zvd6|8w<s!vC+2YUZH+
z&)!?v?O`NEu37(h&Z*A-1~3#fSpN&%+-&vUVpZ@%Hj8xYnUs9HXUTc%5(+QtOEU5e
zDDtU=G&S;c0kB091OV8AU>^S6lb1z@C&i+oxQN3L4U*c`cZ{WG*~SObQ>$b=zyORp
zJb?>^`=v2sL*hUXt#H5&0pY#6cpb!ZhheL9Yqqy$#doh*0QmJZOtzv7*C`5s3uz{P
z(q)vqUc;#@MxxdKTQrZran0AWY@2sxAu0>{Q^(i&XSbD?(Gln`qa0jFAxHbCsh288
z(sNr|)7k`r#BNKXNL0glurd<w0uT~Sj~79?6*grCw7g6H(0YKu8ulT!kMa|{I(xfr
zoI5nG!`z2-n6)oK7>7Bf7n1lOQt{RNjJw2(bT&7%ur@O#9SS7Zvg@7sm5lG6zyXWV
z40aV4QDn<uzFkq#v!Vd(2(Y_ay{<w%^&8`|6exT8=4(I*`ag!%4~UazAJ!!d&yydr
zkt*#zd-*paz9~QAC!XptqW`N^3W=W>6~bv;+6j4hkIsDJ8U7wBPH*|qgneoE+k4R<
zq52C9|JCV#aq54QC9nCodRwc0snY$U(OVY(B$`EcXbi_NZ(jXBn5CQb|KC1Sk)k&4
zA9eqii%}hYiT-c+AH5~fxc)bg{HwqJ{mB@2YBU)Cn$CYOqAUC#?D9#XWqjqYP5iqK
z$fjQt#L10#Y+qbv)BArsqVpOajGg|7nZ*q}1x71I1YyOv$nt#MJn_|eksY(lHC)hy
zDk<VM0Q>ZFGS;66@LC<Lh%MqI9lW`wy!t+$Bx^ojVq^k0t2VZNPx^`@dsfhf+5N(Q
zJ@Pmm@51WhycvWXxX`|18CP%l7bBid(?@WjHa>vBYW2qsTO(Yw1kx;F3F9Sp>EUJv
z2M6D#H82valABUME=s9q##EtHO$6*EHNQKvE0Luv#v7*%VCJHp;Up`7C8d!`XdB<t
zOml6s@eKfhL5@o_F4XTX(Qkwec%Ohs;F#-6;L-Clx|@bPDQ*}BV;*XWzXfh0*3;x9
z^}OE#s;ZMuBT-f-&0saQmdpvNLR3{jZC0^qmSas<o!4395^K4u#OI4vj{wr0^0q-z
zB{>Aq9&LP<26z!|ILGPv-JZ;hiVz5hqMNVe%BQgpAM21*`;jW$Sje+vR+fZY2%JQ#
zw7ThI*#u4ah)@h*1vM;KkBQ_-l*}|NHR19(1cuz7SwME@<-@BOW9(p%+}2pmCCS3<
zPMk(8PJ&eyY0r~_9VUMRV~12MfzjO((0mI+Ax+i-XW|jCv(sj>;dlBb<eVInyObGa
zrQVT3`DeB##z^H;t{={&Ab$Y<09dmSl@%Rz**F9;ili=uHJQKozU^IVo#!(|<5wX&
zn+9T6LlDUWLcqWX0*OIJ@z8l=L{}?cmUswPuQ`l7T^@mh718Ffklt;{lz`y|A3SQQ
zFDWis+naF~<LVV?q)^t&+ro`#OJqqp8|A}^Fv$wew6dQaUUf$<euuBdDLtQm>^5|4
zd_r_Z%a2=*CRDuQ<g1nIcA%5gJ}zTVJpDs)OY$Q#Y$Q!NA`0G?I=Yuqo&VjCTlm;|
zRT$7h3>fkl^UQr;VX<~!r6>TeRDWX>@6t=Ocmoca<0ichT#mU;PX@2u&U;NNxJ(p~
z8SbZtQP>q{^G*Z4d_tQSchutQdPDg+OC&vh$iKT`iFY$|TOGaxoL}C82l@=J$e5|E
zRgI<9XwqNI7|xzt+9DU$4I&}6uJdqij>+cLFbE|uPcxhia?)IHnYogo0PiQ_H^tS_
zvDcqyJeOWNb9*noKBSRIy9l?)HYjtfvA+)FDA|@r;#MBwVq*0#J+X1HgF1XI=w5f?
zVOl=)c-+g9vzHMAxzww7YC2{jlg6;)8GX^K<@QlglS_0RH2Q$?B^BJg{qCD*!n-fg
z$<=YTEo05a_k20Le?L30ikpPjlNEeGE8docGS+>W&MCj`ZksL5YiYlpmli=n$HlW+
z5-l~IU>^_S&)WOX<qoz=9JT{NmXD+YqzYDEkDu(#bids`U6AF?9C-^(Xxk2aQJWoy
z%xf49!M2_#8(+i?4hUaWKp$Z#5$niozd+KZ94N<t)H;0yb3mA>Tm%*xJCCQDH1S`r
z{gja=K}rt=`Ph45J_isd$J*Qv8p)fsqPvSoq6~F6^EP=EOD`59OXJ?5jWP6BArLQT
z)y_}-qpL|n{kWa$Q?SBwiYcBj5?%i-u}cK6R)g!l3%bME*fR4#<~u*1-L77z+5Y8w
z$m5Qe_;fOv_lx)9xFqGu3^AQ@>D(1eK7G+;0rxdF`2FYjBT|u6FLf?9b^9ALM7~~0
zP}7zfcAnFJTD%pqV8T1Q_&=8{mfn&*U;5d3vC&FSZW7QpCM&&`8qQ7IDN|N48wcnR
zAQhF}l&AhM??1J$*(OJMw^tYIry;WA63(-jZ65<=sle0k1vt~YNo!xQl0{V`==C4b
zh##fyJNeB|X0bh_n9<WHM#~O38b9A|HaU1aA_H=zi%F+u+-wu?`7n?5U#Hd*8T~T>
z$I~{ME<fx`8xa{lxecq0{5GqOoSM$oIH9~^BroZ^EnUaUt&9zGKEYRcn@*UvHQM+l
zCMNv9@q|8Pa~lv}zPQ{DGO2n_%DA{JdwmMml9jgq_2tE6HMc=|`}Oj!@8watustBI
zJ><K7#85jm>?)t-6(~pVwG(SmgwHoaHR0vsU#>m(*?Iad<4#83ZhyUS-O^O)Ae(lZ
z2;Ei5^4s>?43tf`@jqQk0gpp@c?{^{ppCLlA<g}f6;09UgCEpREz1DIgPhxbAJPB*
z87f!r+3fY@tOjqRP};*Y(}97hI=AJ)Y<$h%Sza@zTd?gj>$8t>GW82loL6UK&4E{^
z%hJ-)fL7m`$hAxFr1hF2){-*2-Yby3Tm^J`*ECto0N~7d>Q{2r-m+bSTf&1!<|F90
zvP`G^_xq6o1YG^x>{r7YV<22Xn}JLTk0KSayXAK9e6X~qr>D%(Z;PO-{p@VtwZi6W
zAhYZ9+1bGJO)Z(z#biTA@$-!eC0S;A5r^L>v@4~}S7&>@yg_rp=opL|$y`2r-S;pV
z!4{d7-|XlyOG-%i`tFgFBTpB6##b+8Y=f%jqnkCefn!iFC@2yHPH;!HX>fh=-u|u-
zzFEG$)09>!bF#SpE(m|=7P3;XR#ZF(QqHu8*6Y8%?re9`2l@Xpu#rAn`8Z>64XOu$
z`x&$ThokzT@%P?)FQi(;)`2^olo%DRpMl*w@pADPH=Hyc&-j5;DQ!HOhz7EW1N48s
zW)sh>At*7-ULA?{mcRGgs#`B@Zi2xkg*4+F4i63%t>{S$?q&P$eMhI4TK`hf@C|0w
zn>>4cbzC#;NzDkD0j_72?7Xq{C2meXbqhy$U>oH?o0wz8%vXo=GDKcHx3`B~9!;9~
zuI2&Bg8D?^$B!YUN*@XemeXwBp8~Pj^rzpQ<uYly$cdoy{SzY!NF<J=y{?Y4lRt4b
z0%^L<fji&Pdy~2MTMr&<NejQ2s61RB1E@D0fllwgGo*k|8NR#tRx3Tf&T+aPpp5Cy
z1jar~Nr2)AA=m@TBXd0ErX4;gzD(oXmI)p(+{|0T@pSw8YWsW(7Iy_RYYX2Wkko=#
zYG#N4cGky|<k_1)v63Yf6~s*j_3o<`gEAT2mNY#XPQdjSUrrm4YtDG3B&e7iUm)gX
zVaHjd8rOUKX28bve|<GPo(sWm-RUU*Qt;k)?P7m0BKrg5{>aZWU~S{&U+!X)xqEqC
zUo6YcmRm(z86;bQ+KTN|t&6eu?QF2Og6u%w)4e77pbLIbB+)E|-fqd)%y{zU=>rU;
z%GJD>OzyUHY&?7d9rN?`Z6HPf8#kY*v;$Na*JXJ12sSy7gT1{*p493So@bRS;na>O
zPU0tb_7-f@*}VE~?d{<n0fW-VMm3JpV6TP5=cD>2mt)OWcbB3{l}rKGL@nyb5;{aE
z$PNd>{a`l%c#uJ<{fE38#sP<;cqJny7YpQoa5AWNp!5RR44i*;wMZM^uP+AsN0Uwp
zqR4#wmuD@t#~m+NvQ-{@b*VSS#DBLVlwd{O)5}ZYr|<RU@oEZPnN=?ty9Rj9U9f_v
z!|$D84PJv}Verw0j&&W_(P<A6KzKpRSzgA#!+j4yzut`tVhQCktnTB(2^P06sO7VP
zC%7zM+I>$io$)_N#@6}`3p(Vz5YQCNdPMKHQD#3@a=GagRJ?u$*p@r}?Ci+}74CDN
zPkT+S$^$Qb<-2;4I58Q+&89pjUnE<R*1K>lXE-&LacRJXAP`K(d^@_pllhSh>7XnB
zEwW&w34oK8IUIT1e6gn?ud1&<7DdV_bFtF~=mEcbed+lI$x2X$#q$_)J&`R?dJLTk
z^=>>E)-d$~!l|$sy!sgzSP3dqytx;v1=&RMz>QiNgqsbueq5LM6r$3v&XM#g0rl(s
zp;UmHHQ~dnNNryVgHF_~uC4-~uHol*9*RjkR{+fG0hx13On-*xtj}^*C;@bgyZ2d{
zxssL^J}nc78vDSj{;~hbYTU<fn0z=OxxLYReKjN_m(y=u5&tv(gibgPI4SXsZ*A^S
z^pA!$j?iic1Bt9h1@KS;O}Nb23iFmSz3&R6Eh#u37wDK^S|9M+iWsNhgmV~x!oG(6
zbiLmc=Tq>Jw|1_5_fD}%-2M37`Vv_njSua}n+y#NC9=YUq&HcAzVcvW{kIq37cmy<
z(CFyll-uA^KzB4y-d3KzZ}Q7uYi3@V0E%%3#Dr*=j{NJA^;P#XV0a-+cpsMmSp^qr
zB(}B{cC4Kb2XT?6X}tOc_2p=^87Jf7<?$?f0X8aLnV{x=P4nd@C<|73?=FCwcsJx*
zslv_D^Gnv<^KPF|11SN$$Gn%>GR@O3<NXparIn4Bvg?n^$2UMckm@mk3z?PAfk5$I
z$Zi-M+R8~SZ~|voQ*a;P%;xA;wzl$3hL7rti=%&JVVlP-R&sHI+G8A5t(lN!h3FgO
zlgSeB1Zjdcue4qzuK~pf0+As8tNSXuNlEPjmb0#IiZBV(8^_tw*Q|z%29auyHSv6i
zQs87SinHXUf$_HGpG>y_gGOIr|LQTDXTKT=thJe;W1E~fT4M94so0fV*b}r{a(6$%
zMcf;ycZaFb?x(5e@ibq!I#2T&)y7|~>14k|JbTaQ<>@J}v77)+e-m!@TM!gNh57+l
z_&p}F_v**p`Uj%Z?r@+w=zxao7X?h*(Pw-6bc7r53~R-u&(o{jSBEJDtv~7Bf2uGC
zwMkb#%8lm9HSrZH<eG9xKSseO<9Z77<Q{+jbB7a$@ub{m9|0I8uC#~pvU6{>WP;1T
z+EID@@E)lOLrZ&-4WjJ8dueJ8F0T6o@9%ru>pO}JczPEt8=S@2y2<qfHhh11uz*8#
zQa~QVd;WDwt5NNN$(Y9qjdxkm0y5G7?NG-k^Gn){YoBJ4gNZpcogzSZ?>e5;jw_;!
z!Ar6%pU%b+JOh5_E*IOJ{TNSQ(B-<x>hVa@ricIht%vYWg2z5;-xGjq$@}<G0bZHJ
zsXxJ7<$vnBkoIYOS%$`Q0K*C9qeDTs!!ikK=ew<r5?LZR7P|`_NH0E|rxDq?KTdE3
zD&49MfP6RuJ9?hZ*6+Jb#hF496nR%A>P)VMuoFQSyvTWMJfS=37I1ZO>TZGnE&|b)
zrckp*sstB8D<Jk5A&_snI<r5M7S^0zRocEAz&*y!rKbZC)?zjRL2LGM%WFUxlaVjT
zpkz9RT2x}?vcYFR(<_xvvfp;_hfqH$=M&p>Wl8DKcu-UhY<YPL@4LXeA3)h*I-2kg
zq#rB3)fsKkU!SPx>yxio{gi9T(|Ho93rUi2-A4=vYlhs-V;IbPGcg9KT^@O?m8Urh
zngfJ2IVXIJim0$iopgVz@9i!1GRAtE0Mi0KS$emfcvV#Lq2)1|32?;dblxb*55)fv
zWR){z=wD<O)4f@v2~1)s{Gxc(DSM?;n?UqLV7P?N<)PWj0b&?>Iva*~O7DxQTEf^{
z-|+Ql5()FKDbuk5DPS}nN`T}73MSe0ARu4NbG-MZ=~CC{UT=9W`%S6orCbJ|pW#u)
z>;>8YE<@-SC@k5J{KP5feDXCRv-;Q0oTcB<r_^vE0I6Q6CGLp~elc#OVl%IE!mPd6
z>&@~S8m4sI`VHHU9}zOdWPI`B#pT)IAA6V9P|Or#-@<98MYhWOL&g{T+1HEq?+65(
zW*Pw8rj=Rd79R7sJRCC_TyizW5!>(^`hNMik`y5`@WFF`;<BWq1Vlz>P@~!5?M$Bp
zEgNw7W5JK}w5~|N>praJWAe&ze2mDq<_#eS_8X=-Kfhau_k>3Q8AsvdMw%W%<~$SI
z@hshZZ*-wVHXqz{wi?Tg+U=ZpZ<zofIQKmyD+EZxD#pneL`_g5`lL5reoRW^YXeGz
zEbpugz8y)%t(U;?dZvK;S|s?wqPE25YGYHdZyuw^aroOJ?s6>XGNn!MEz#W{!9Yi@
zfI}$@BF2l@UxRQYxnhuob6Jg#3dJ0GGn{`gl_QamkXX4BDxagvsFAq1V_#^vVsl>4
zQXYVbZ^W#oTWFfomy*$SbP$U9AvbS=QDSbMw|A>F@Q`!uln<xPI1IjeV3h(@rsbm{
zru?#qYob1HI>Ly(r%j@iY~{<}rBLW+Wl(LBsRH;HCvx`mxDQXm%|Ma6w&goJ0aln1
z8&wBS{LyXe>kXPLS7r@|Ky_U{A@mpF9EhQ_!lE7qIXQ5nkHvI9|77>1+(rl5g9uDn
ziG_I?^R1PYF8iwcGq`ZC-<=y<TLX`^c^dY5IFBM*L}$G395GLn0+3BQEAFv|%REcZ
ze<x1*h~l!ZMq4~onYQ{HL-xm?QZRrKXhj{3etw<z@1#U#D_Oy>UU2)6d%Fne$^E(#
zYdIc|;?em;3U~nHJ8m<ZybPg>uHxtJAzqUD`v~C;zcPkdyI<JixFe4mG2G@Un?IS3
zl6oEAh5pe$(63KREmh8kuL7HCkx7HDjf+vSmPD76M><@4{rM5|EldKxN#i+pZ?2@l
z<w1b>(;Ppk2C=o#86LyrD(n9ATxr>+NYhgCa5EA*zpV}dJA26~1;p*+7>E)zd-}JU
z8DCD8umHciyc?WOrh<3pviPp&@q<1r%OcFu5dbur&ZMKZBtgkrFMy#Z_25hT+yBAl
zZQuiVCPMuYwRTghKU`Dv`yP)?trsSg%xddi2pJo%!~1~{7++g5(cUFw-1v{S;_6S2
zZ{}&dJQkf^&!&xe-J!fjYg&fRvBD|!LOyp_e&n_U^eL(IRl0m&R^^?sFq?g86NCoi
z+kf#ER$m|WXopb#SE-4t*VdF%^Cg+c*AIN>4-5@6GRVVsxKV*`Z$?Ie1GrX7kHxdA
zlOqCf)6~Wb0@l9flMdd7YV3lY<?qnceHm<WYphNF6NeN6d3^8rSI|_DADX1L8smJO
zf|cKqEaHjIZ6rwlCUjqFIw8B$Z@aN`nyjj}?LN1d3t8AWzEKpxaW}Pm*Q>kOo)lbO
z`qde|ikO1OA>Xmz)_g_-@QqWyBvZ^~0WammrN;2o>Q;@tY*BS_aj%(aXTnQZd7VW%
zqWS8uw1L;Ib@<bcd#rdI$OJ=S+%1|~ct^S)`Obq0%<|8#KOd^H=pLvyZEEm^iIYX2
zTG!aBM3nMP#^mc30(XvP({3c_xhdnb989ai)jLscg}t#@hz<hO@_`N(bbKoM*uW-h
zDL-$)aA!(L=GzYf@}Se;7DB-?HF*Vvko_bhXP;P#4Z;Q6G4V=8rg?+_8kXDF&p_Jk
z`)|myKJVYwq4i<(+>A653X@K^Fa|)PVe|tXd%AlVv-Q?p)2*qx_f4Xo*&Rb)(m5<&
z{t*jqy`ZJTB!%?YR-PC>xY#Bok_RVTubV}=Jo3?xCV+yU{OgYYCc-MsN^)}aap)T(
zSpW}HuF$qGj*xzLe&d4*_+)*wwA|6;l8XI-q|Kj(ZCNglO-((DoXtw2-!|VCF!z-C
z6GOg5`q~MGpuzUlwa|=B9V<04c|QIew~rrqy)k*kF2~BDX>~<xw-?$%P3}bBnf0q7
z2=T%O=UEXfa%&p?pop;->@)=sZ`N7)xu*8>)t)2AcB5MHB3rDT07GYE%SJK1Dt3^F
z*UN@)yng=4^C^8}DhA06(>6ie4}4GC){<vcSdWGg_9Fq*tUV<==Gy12nf=yK5gX$a
z@1Rfvz0!DJs*Ln>te`x?W#a|elev%vrzixoiGIarmr+1YT%S7Wc{4+>;t;8ICUtGj
zQ`;6t3ojE|fkI5^2fGNZ8LEijAHQJcx!TJFiDbaZe1v1k!Lw9s6Gq*&9(Cu(pODp^
zv{I;q>$F`#)=COUcG>-RR8yxT*xK_pv;^0LO)E#<ziQ<?o$&auQue`^`YV5?N*{?1
zvwqXuYjL?+UR`5*<wC-W>o6Ij$c_5-693I%VMlixJ$90b7d6u>HYY7LoK)<*_VdxY
zObM^?V>{&=_|(Tn&wt=q%pcBwJe2!+LbkQB<Jj&GhzeJnamw@gwSg=kT))`QR>_Wt
z)SUarHti0)uZ`cNu8u!H->fkaqoEVB)n*~e**&yvr}-Sv_|w)MfQ}=ls3!Ou7_>B)
z%Pr6=o;*SwtFZ2ntxsv)V*a-GHX1%)0ixTI{5L0NrD)?iv3Wm`CT~-E;(Xw%Fj_?u
zE!hBk4FDEe2R6GYms#yoVMu&ksX!3dNgUbUeJ^_oU{ZNAGGX(h@4Zx!isAV5GX+}o
zXSgPHL`!rph3kE?aMsAnt~o86Uh?88@4HrruuNHTog_<(#Zj611u;<{JtfIX&%uA}
z%?BI&&!UP<{LXW7LCSqc;iNNbJwHnVvN;>lAc>y}9~TUt_*O!D#DzrRxip(U%*&=d
zCpSNY;N{M~I}d+mqn<Q~NzCS4sf)3((HGr&vik><G2r|tENI`o-u3np_AOh=63UH;
zII?62>^#EB^Eb375UWC(jQHLot_B7%1EYf0%(2r;kOAtgsasEB2x73z3P(AHrk%eY
zfA7At;MHQUU+rd__sG34Q@6xHEpbChLx&_btVqwYGnRH8v6~t^WNF~|{CTb#Xh3_|
zKw_;U#*&pyw19uq_1<&B&Zl{Tnsd8o6Yw{y&Us!ksespmCVomzPDXM)fofp}wUVFc
z!hF_73ITZec9{WqHE|j|zs=X2?p7gkds&0*q}UW4!}oUeQokoRGA|@>PKx(#v~Gl?
z2$MxWXieOlW&m=VvMJyH;MR9}e(Ypt7tA|<Cx(FwmsoggJq{WdTK8pBH`LzJ4x6D@
z09v3{mcqG>#vRDRQQ65vf)&;_AAAQrmQK1JbX4wGIiDFHviF7Eri|`*_Uze(NfPeb
z(jbFwvK7S=fr)ZoUg-EBq(^DY%lZ3*)SQI7yb>m6o?Qm9Pq1>Bt^H;l0V|Mfl@`l~
zC$-Z8dQL)MI@c~_)&|Lbhii8+dHl5@;zC$}Nf-xVfSQ_mv)NbFTm*-iEIOEl{3|mc
z$Z`WTad<Zj*m?ylKXdg8Ev)n$#p$+ksSPtm(A=JVLnH25kmb4eJxSWE74^LR9<8a*
z^OTADxmVa2i~(l{;@b`TENXjQz3M7`M|cG9yqlCGE&N;nw>mUWJFE2cC{0}Qr!C2I
z$#I#d%j-R!J7GHl^QI{S+AS%7@jI_|`D3d<7NeY%GFDD`m*G5C=FB#B9#)po2~MwW
z`MuvG5Qq*|e#%Hi;t#C}x-7I?#EM(bH255uNYnDi(aQz`2IU@W<<4>Qb-?yaqxiE=
zo~t%WI`W`+YaNC9VGMHKYJ{;E%GC=*5bVUJnSKBm*z<23=YVg^aU?DeSnT=9ej1oi
zKAC9cGaGQs8p>r<>kQHqdF4LiWBJ^l?3)*$kBID-D9=zAqTj6TwGUu0`uqC>V^f26
zLg;a5#J;T%y9URm0{-IEA-!41EJt!dhZptN>IQhghTC3tqH!R1KX1k6)gxggiu7ir
zzEaR2S!9Fq%VFq{Sb^i?X-G_gt{L_eGb=>kv4(C`!XNDvdkJbs<r)Fl3xr%I+In*;
z=aHNJSc@@ZtvE?_&a!1*o?7BdBxmAOF|RnOXJl5h_(VmJKWEH+BU>Ydk1-iRhsYCd
zzkolIx=1EuzO5^Pd2{x7d$t)U-W@cY*NRNW7=pBQh=69W!#7mIQ6F=3J*^A^7ZX%g
z1b^ETaQXjnB>5)E=;`eeCYH&D7K4W22D(3X9zd3cz9t==l7DcTZ30vU-(`v)wNfLm
zcBpNYCJPZzH7}-wIm1kVcCKgqwwEyXw*!x1ptY!T3te8nTFc$jldK0n3%g;fe%%yR
zjvT(n5d?CQgBIK#{!SVN4Dp56%qybCu=1Q+cjGPr35$3B+!Qe?X+|gQdyuOXz4ydQ
zA#!;VyKbiydp77g;KYC2DuovCT%OTG;dqn+q0XBGpNo(*BPB>-BHqgcob{y%Vlw7v
zuK*tPo>3T*>)%{mWem*(Neb9k!R0;G=e)5eC=9)L5-SW!I=+W%havPAFL|6l;-(c8
zmorjsE{`r=wkJS}pCp}OBJTg~H?yVJe8PJW%F8ar&CUH$Ip$Tx*5*0(Bbif6UgbWp
zAM|mV{n3dDb&S)|(m-MNm7z8~$v1vJ?*Yp`?KKPf0C1C&X(TQc9iYVXjU`m%i2IN8
z8_w2>L*uV5&)ZI{hdw;;nsPyR4@AeK!3a^6s9%6w4S=(39ixQ0^$S~`|KgHY)zIh#
zt#2!ZoEg=&i7dEOLKs(v@~PQ@XLe@>$xGf+fi1Vtk*rr#?nA&qFLot>kR}V-?Z{31
z31v8=hDSzFlW3ZPrKC?6-(rr5Tl)oZ4(=*h6tH8tH-ma?DHRak0icKR-wfBep_ofJ
z{N|JI0;#9rQ?(Yt*RA&pGH>@U#WO0HMnq0Cw+7gxU&iMXGApwXVg9DqY7R~B1(=ni
z1LUbIz%mCU*1sT48*tGxHrW9u6QsAL15r3snG#MGqIpf3o2KW5%-b>A_{8X(aa(b+
ztR==YFeqgM8(7q|zn;Xt<LZ4@ZW&orG{g|UjV%ae^=-oy#i0^<9o>Od=wddo<m$FO
zz#1wfD%zHS@og!jgf}nfG&v+THn!Cjht1Yr=K08HQ2DXxif5#&D>v(BEPL_pxt*QQ
zX4Tl-Eio)NiL55zb>`eGESM>>Bs|vM-G2-Mu8i-&iZ**CkO8s$+41&Z2a{<!TV}w6
z>amhGp#HD0d0x(lcA<IH@2Oj&o8`B4YO|wei}!>V6(I`)Wl5t=0q3n>Q2hmc_AKSQ
zewSIqNvyGN3A(*y(**V7O0T<+ZW(hlGlVTDN^Iw4#{t^0%%ZdP&QMIO?8Q8O_c8wW
zx!a12R59`vjbIb-RZD|pi!zFb?Y_&|J28OFs;sK|zT9vsT;8bt02<q2UQ{$3%^5Mk
z--hr`Wo4Yq-~vfVYxx(zKRZoU$7~IhW4m#Y-EjluwQI8o+8;l81ciC}lC2s+&KOFd
zq+@Du3+jTx+5F7VkX0?0vaQ(azCStRLl}{K{=>ArVw1}#0@qf|dj$B^yE=8WA`TQn
zDW1(k2`Y4$^kTkA-?MR;eojn0rzgvH`9TRN$U`s-Asuk$F5e|QX4K{tpsla3&qYSf
zXNodX0u%+%2Y2u~-m$DA<(PomnNS}9KLxHTM;kPU=O{$PQj6j>MLa;6=jc27j;pu~
zx<!M3t0fA#EGRl-jC%>~f+w4Ai=cNOdf+wfLC(1NMsT2ZqrA6;hZY83=*Om-1y+UZ
ze)(xDM1vP0p92aScunYWoO-OrVQYZrfL7WPGFr~H1oOEjTGf5Kf6oZ!Yqe>Cu=N6g
z0C;gp+~h#~JdMOpAcuG$`Whw~inR}DYZ?EO9aR>2KNR=0*W^%o=or4(+)H39p>1Ys
z*uiNZY=wN6lJxOtB}Uez@pyoC=4;DRkRm`?jSwkES5V$@H9<`krI5Zl`Lj9&VIiUv
zu<Qcerr6IqXsN<L9meVGU=?1V#VcXf=5j5B<M=_M!@R8ZE{Q&<=>T>FlM&gfOCJ5Q
zWF53Hdo~_VI>}OW|BSkf29=4OHhg;F=Oij-)%0-BDVNURS?`^>?*ORfaeLd_bK+D;
z5h0{RwS=e|kxqC&lB;(}f35?R#m+&5Y%|uP`1Ibn6?3Q%LbB8+I2bC1@EPNbAR7n|
zM^j4#%?@u;i6a#)EdBs8zz>mT13=LOV}C&KB8h|XCLF7alU>m5bSA!e7lx+X2U<^(
z9FHKE?%l{JJ?p#v)xM+-4NI_-r^<|CC<G^>4TQc+P*dOmRRZK$Is>0Y>prwI--gLJ
zWWJ*`b_fgtk4{RADrQ)4a}~88q7Nm-3JU~7J8_hP$|p2#+GIICxQEYMXMg>A;_UrO
zin!!4l}~U>A|;HRoQ(CmF<TRMc;r_l*h+5n!tTi<$<0c$HZ-l8TQ8Lent?p;5{bk)
zi%EI!m^$0KnSl}m@<DKHTkHeQRtppaY4IXJ<xHz<2sBbP07fn7q>Uc1MYX@Hj;$>X
z!c(*cQ~BGP2pqn@x@#Kyhei-u0H2rup7I;0)+mJRfx~E%XL()9)$1?32kgYo$%%KO
zqHeqC9Q4cQut%9ufRkJY#pZ_Jlv?CCA$*)s<&%yuK!7$I0&;ntyU!G;?kuRzG+!+7
zf)ed1&_6sXq0bc+`}br|BQM>Ts0gfk%K;foee#gJi9kM9_UZ&YR+#4%sF3{{udvw{
zenEleerTH0@>%DjCxALY55YXrI_Llkv)=;ejzq^!lb_cmDoKy_qyjHa@-$M0$GWxB
z1eO3;nzwxyBSD17&u3FhG_^apgD*9g#r-T557a5NKrjL9Xuo8KH5lyzH7`({1cj{m
zq!$1jN05g9>5QZ2Gi?DiguRl*-;5uS;1U#KjNIGQSct~InJE|=QqR^I??ftn)PdD(
zzI*&AQjtmTegr;!6a10nuKG?x2yfut+iWi528w~{DwAJRQ~r}LYA%svrTM$x%6))9
z$^b=*jqwU3%yE~{G_jc+84<D_`jEh^LKXhKOgsc2QQTb;=V>67KTbd(572235E)K;
z0_s?yr|I@3!9GzTcoG=lULk0>=^`&CuFr$6<NlmF#yf&4<QY(F|9;^y)kODEh(2i}
z5g89eINxsAhZ3eu>$8|Z4s}lm;O+qDN*SZ<Lt3Ol=d%bpnx#PA%v-le)W-LrlM*Wo
zi@F<mcwVysqU0ZUc0fBJB4DHRWd@(wt2SMK9cN<e7l#-4NeX!5;skiUSlkZ*S^=mr
zQ^-!Y&;m;=Hq|a;cz77uunx)*i*}>P3NdJ@pV-`609XRdkcGyB@PY4tXNYi@q<(O8
z&<wS|&|9j>^!uy8%d?+~fNEb4_@+^vt>*^11W1&PblzQ`W?#><_Z_@(aSpuLiIDB<
z9hU}b-WHCV$nvvERt0v{(e3jvOM+A+VLuO4e`Mq)==^uSN9A%s>xB|jTg1f^R8d7B
z__Px5Qmn!8q&j#R3}%W~>})c;134nbvF8(2W3~Kj@XTK4J(BWW-^|*?nm)e(03J;8
z48`Gx$&Zlgy}oQA=Aj5@R2v3Fv(vF3;D%yL@NU)m@l($6A6p-O@jrJup4ngN%;;#~
z{e>HqTUlB8hEg#6{OFUsZT_$lE>CGG&9k`%F}q*mpN*Vv*U5q{j}CT`(UGp+n|G%0
zR3-0-UcP1KNhzQbjCx!cAj-Xa6^33j<%-RS`p7;_%16LN)@O(Sl9;1dXGhUUplXnO
zeza-mUeB%VOvxYNhdX)xQkp#A*@h?tf=vu2UExCGM6^;KJQ_r{%J%l_AgaCXxAuVY
zN*&*Mxvg=DCFZ`hp<QY*+Y}IiN4w^rkL>hQzbrz|v<zrjp)Z^R6sq!3$NP=0lUq5m
zY?g#IjvO>IvL;Af0TTxxBVDB#H{t<3*@Y(GGkd#+6+vUGqsupJ7~gm~1AZ-N5|tAV
z_0xE<TgjyXfkY?*;KP$L_uVQ-n>|5QFR_at_MGT$^mI$6l>we7Xr|-o#?wkt@$wS&
zf8F^fD%Un0&5cZn*ORAvr?}TE{Q)LUn7kyH4zje7(9ao2%2v*z#<MA(^TN?_z`7zm
zKlrEaCI{+zd{<FXJ<Yycf8Oe!%r3mI*=7`5lgjnhkKvULBI{goVEn`?FVG{q&gWee
z*R<*Bv?xx_GIPh-R(U5+fJJolB}NC&wuUz@$|N;PFOf&>mz<pWZ*6#wAxV$yNCF<o
zzb-1mOWf8?UA1CXiEUU)Hq6|l3ZLKMY#+&kmN$;mN_ui_;wKH`#2cf-lmtuIP303m
zfnYu$UcS@l=H9b;*;S5HBlbHC@kWD5K^&mQ3*ZyZkGD(hp3Qt7z+?9+j4v#Km0_zU
zR<?D$vKFJTV%HhTY(uSfQwz7Puf&G)hsJ||i>%Sd2MmzH8Jp^i2J)=sMm7>WY<V8S
z7ZbaN?&OoVHRjdZ53o||7Jf;bs&SeH6(<7g2pJ}pj|yGjh9%R;=bthZ5s*YMzWL%~
z2ufOh;d;FSQwUmJnW48+p{X>1waT6`E1KM{`6S&K98L_#|B{mr_Z(VuNA)-$J<@mI
zvV#;ePkyG}!`{uUBBNC$BPl#QoOS}t?}XWi>wM_QDYc(s94ZHv0Dj~QxT@%=e;DU&
zfUbh&ub*8}q<#yL;U$E<yKgm8{sn9=B|54QQ#6t$x&w4E7RH;kqnw+2VUXZ^$ADQC
z2&CS=4xql3KcvH>qoi@2(1Dqs3QC?722U5C_Ix&@$jZv%Ax`t$$@vUAMY}E4PeG^t
zX9eW!_RV}=m^Lxemfo1G?y`looXzNZf+im-YU<vJB)Ro41*WzK5_xad>Sy5aKLQE+
z&rpWWb-!v$(P^>0&69tWr<=%9=YFD)4$3Y-^lfXtzL^S)i7djx!b#<@7n34l_u6Bi
zwqV5-JmR{XoM4gZFx&KG^}MNj0{$<bzDz0BJ12`&oKtF13erRlPJH7QcYp0b=TXKn
zxm+hF@oC3Q2KL!5d)ev3{t8<{W2%rV9N`+rb$qA56GKXNV_+`^3=F;0A7AdvQ=q)I
zk7xgQbtjEgZMa`__Mt;tWUlj|QLtN||D6L2W3Bs2=eV<l_dmFJj(;JTIZwSD<2a`3
zCVbX*DWtx88pgK;lW7l0?pQ|PRL>C33Z`IH#LGkO_`RDpl?`msVsKNU{RZJq>n+xK
zdq#h>(70gnZ!ds}EP{m0Yt6D9XW#g0lJqw<LsPSS6_*(z7L0gqSM1F?8npHuyFX-d
z<XaZZRWCm$6jVXQOa6_qLXko)kLKnMeK{EB1WL5u@Om|!-ubzaS?Y_ZD8>k)WN$_h
zsWL*P)vIudI6i%;)_Ki2iCx6%W^a@lEa74m!U4uNPNeQ|R8xBV=|Rk3@6HnH>oXsg
zi)gtOJe%0UBGk=wtph!ijmyJ41N_WVgiAf)h&N4BBceT2{BAW{`>!dh?IP1H$QCA8
z7nIjz9w&27o_Fq&F*ea7`!*`CJ@2pd;1PHTbQ(R;BwpxY%k(EYa&C~p<%na|NRF>}
zv3eZ(rRQk2SN)E=8)auAmm(spwp^(yMyPz$HQg`<8+3Ys_`W}$l;w4Gg}1p^G{3`P
zaNYRDyZ~w@TPA4SP!^(Ly9KsY-OM#A>Kt1~BLAe`CzwG02*mhGP>OfP8IpW%l#0m+
z7?00y2`kv?6MlD?$Uh@XX{2YI%o8&=_yxj}@9`E?->q*{0M~%g>M*dVk>!qz)7LMd
zMLycr{(<X+guX2pggKPvU_fG&lH}?*5adRe<kn)913_kFlN|??jLZmMd+M~p%ch*(
z4(%y@tKM=cyI~el&*EuO)}xn%j51{mk|=e5n$ULDEq5+!6-Q{?s)+r;F@4!Ay|)Xu
zWgV&IGfXDMiwNtBzaOQ0aJXm8{5;lRM67N3<kzU@{ii95LAbZg@Lcsv(Fm&IC<%tM
z$vmbg?o^qe0&^tbEj}Op>HjtQSrT{A5c+r$lscr&R%mViy--=J(LHACGJBCQA#%uK
zPf2=?wBW9{5m1DH?qbj2P%559+lNFBW@P3meieJ5YI?XmJF5lTdEW4p3xP0?mEc%Z
zeN+Y};GW0UMH!7l_CGj0P3|}fEWwING_9HT4?a1&NWK?!-M<3lh$p^Cz>||?m4d;c
z7lMv|G|t3z<-6NK4mBRE(T4Os8N&)zbgRG%_tthh9=D;Fjfpy3<JUefaY$jdjvW%W
z7A22PSDA-#P+K$Ag+AfAvL6L~K5UONS1NQ`f@(tJztdn;^y+WA+N=!a$ZymKpIb+L
zjFpJeu(f(%!nq~Lqtf>vwfFNbp&+9r=p+N9Y}4~r#$S3$q&^pY!&YAJOxkRkaozTS
zx^}bDDEz_h>`M07x(~`vfR?C0PD9WdQ-<1gVG*wM(1|&l#@A>N&d5-pJvT``KU`mS
zE7krYS1*;DxZRhjVV2)WAN%Xl%L?15O470`p4RyBo%CDG){LAR)&rT{oS$Dl-lBrU
zg2bvkzMo{&b%VUnN57iD#K>Tu(qH2O`Aq=;fz~&qIw9?Xj?YvsTPDC?WbHbWldzKT
z@fS|&F?64(mfbei6eFf~Zo);iD5<qPk}ZO*l8s-#e)WWKkSpncdFqQAPPJh`q+Zm(
zKt6FsKZ`(EXLH{M{VPWChGbj3d|%&Qw%@J;d5?cdwYK~KKzIzW@{l>V*4#n*-g3Y>
zgMJ@GGn5gYoooddN|1iR_|;Di0|CVaIv5ua=3BsiT5H1NXeC83dUgF<%D(_y-fx9)
z6;f}ft(})mw%P>rQi5uGe}^!E$-?<Ww|Wa<#XSrodX`X|_dY+Cny_P$HJanupyuA_
zV(EnG<ND_x6_GukTVq2{Jp7gV$wx?&t&mvyUY2{}3c8C%OkAo`KmXWhjuM~w(Y3d*
zPQur0>q-uyv~2}*gzt-Y80w1;LeL;T@eZlTt&c~v!W4@V>emR~-7O__gW$u8v9AMU
zA^HKp@O%5?ib^tb>6*BBi^uRooE1lk?zW?cS~^1B#WTEH1h+F{I1Ci!QZ7N-l)k#R
zZH0}Y1_@wjkSyG1@4LRWj@C-c*zZMfc0P*1Mny!Uo!YDd(SPJ_#q3`g)nVc>N=PWI
zOP<)}K{_r4Rto_`RM7F!yApI7grNu=OtGf{M!J7m-=JbOp$F(G@uC*;ID>CGxy{};
ztFxKJeNy7!+vtT)%<deUh8t|?upTkz^}_xGi(p7=AHBQP7@x;dG<M)5_YhAjO35?l
z$Ym8*Fum-}Ttf>N;_iL*ajjQr%mXfj-NprS$eXdo6M#0)>eyG^CCH-1FV=q8vb_Z(
zCyrg;OTc8G*}K1U2>U}q{f-!k7{)IQg)IGgNG?aM(9-hLx-F+zT-n|-hJtRV2nBnn
za`hJ^vczjn&H%tB*4P30^#*y57H7?vz*oX}za7Ra<KoPE1XHtQ&&vx8D_RyPA9wim
zUO?MMqspMp=+W4rE+gorngL&^a0I@-1q%Mg1+#<T8pfuyhxvNkU4rTuY6b8E2>Jl)
zIHQCWtE(}#sX?+CMc#zya^lT6HqR!4^pUN;1{92fG{%x&j>hbfl0Ax8v8*tF3(*pd
z&yPv%VHH>z<l7}bpWTZxnJ>koTzpvAh5W_<g<k3Z+myBgV>wr+IlnIvm_&3HO9G=}
zV3K?7)kC&Fdvo31Qaoyj(lqF>vZoJ|mvozv!(K(bZx>tVoyBPK$rYka)Nhk2MJj*H
zlMkPANqdmU!b$|770wG^rUZ6hcNsmWlmzaFqUbQdHj!W@oso|G3}8}WBJ?doj$*!(
z7#&A`cL5j;<C$fW!?3fpb$;8mBkvfI2`0g|hz)c|eEBW43&{h)EfI>K4Kd%YzA_`l
zaoj<K1~0`|SlgI(5Db%b{BaR=;(ckI4uGy!tCi*caKDfy1fo&p1kBL5b6+)2SDSMB
zcfUr>LpI{*4ltxT5Q&4oftc1D5?&q_d7f@ffo8}=9iFr9TX@7xf;GPGP1kLyt3#4<
z`&~o@$~YKG4nnQBpE99A_0sHryea4gSw;cTF!1Dg4u`rPfw};en<nATDRpqC_`JAg
zl{!KC&-8)`V$f>+A_m)SN78oGc!6@~TJ_PV`p?D8;=Q+FsUGqmniGHtHI(0Po)A8z
z^=CL?&U^W?)#&jD&Pn@Myzz%3okoNGHBKWCCUnMfz({^~GI*tp;h$eL%M*4Yx8o-(
zEC*$E9aXCS5oUk^=9X(8dvI$cy($e*e1!t`6I|l(|J3!?QB}QB*XSWcML-1Uke2Qg
zNoi>~AaN+^RJs&Y5b2U`5TvD%5=7~i1_1@>ZcyS|{NC@qW86E&J%2bH2iW_Z{lr>p
zt~uxPsh}v@yGd5(fF6^?D9M0`k@W^6S>UPYVBF^I#Lje0jOdb(rG}}i-}>(|A}60-
z+?A~svO-38f^Y!`V(>W8SHA{rTm>6sYAe@w>uQff|G$$R?B5-c2Or<4$7m!r=U^F5
ze&ss&kj}LKTK!wm9pBBI+nC1m>vexKx%cm%Fs5t!WqtbgHTCftYsH6)`A}|kB8Q4J
zK=*$wU!QwHV%Tb07FufmLC_t4bikd37G+Q@egzM5hLW#%>&j+*I+7ul-dRp8qGFGi
z%9*-9!iE-@B!L`Cy_U6eDknDn46}fey-WNJ&=myAPS}3AXEnbT_^Gv4C=*=KGi0qN
z^|cKG0SL(qixVuy)k3qwu&eAqK)QQ7A|jyh$r(`qxzWD14_(5#2}q40L)!RzxZdQ$
zAR|mWbQ_<huu1m`QIUc|Tx8-K2{jFjrl?={D<pwQe^Xjvk$$OOzgpdmv9Mjf*A021
zAoA#k?LDgzibI_k**iNJaxxmaySUQkkX_!n$L!Idjcj0y($B3|Qg9RFpg1&i_`grj
zu(bfY(2QGmfa5Wvb{jf+xi+=M!B@;Act#c7XrH9Dh)V4vMvxYSn%O5<M9*>-rxu?$
z1ivHKCJHv()J9!hzbAXDVHSD5v#3ebr&s;Y1a`=TLP7-%7=n+ia=p@HgJx;wd8Kl_
zU*$m<Y_;CHU_kqITWwimp_TG1p7BOti`AMHesGxv5z3%kdzI2fZ^Wg#<hiM(FtTHK
z`Q|$f4fplUKV0k@NUL!R^=!InXu6-M*B$<<4bQ&u-#2`-jVL}kl)`Z@@(x7pbdcSk
zu;92LK;m3AF^#0IH9HhrU4@@)tac&k`j$m3<Yq9XZ-aD8=U@M9=AA6+rNs3@nZ>UQ
z3aV`9ZIzYjCl4RZKUK?k*`1C%LCcEqC6Mdan@(>R3tds<@A&+pC4=`cH>H?0K>V)r
z&*1V&`M@HuyCyFUjz%a_Fpx%T-6;A(?ku*3x0_fQ^-~?nfs<AZNQWxJ2B~~I_@nlz
z(W!TYK8gfnVn4jS-bi6hU~m)?kGW5<$GbVrw&4jW#h54OeY?gP-fH1Rl2Bi}(#Gl6
zNAL7-&G%DEXMJr7$$J-9rAZF93JXTw=V7XP?)$H&OchK2=q`u8cxOM`7WaNsiJ)%B
zShP`z_l0x@Vbb0A%t1aDrxK4?f|x^XDSDLtv~u;}*0jcKJPLO9jLDYUymZF56pM?W
ze;$la5NXi${^DE`gy*>5n7yBZ9#Sw__2Z18C8-=V419b<&zWMo-q%l56hk|wy}ezN
z#D2P_FMO!9lyjJWQ;uqbDOOQsc#pIH+o`V7gR~m<p2fERLSj&;OI>OY4l7&qP1?=P
zO}y>)=ci|&NU1MvQ2S%Zxl9StVMp+%nRI=yxkoFYvc9_d{4-dF2>nI?OjU(q%B8=R
z`~Gn-tG@7KWD|(;`z+HQ0WRXWQCPSpex+4Z{j;^@DD+zra9yX{LEjE~G=qOmPS$;^
zU)np>D$q9lt(>bV%LP_RuW!Q@b6Y(o1&N9KHRTk`AEe6zVUZ#S-VR*VVT}gP7tXg|
zF`W<|%br0*gMi8u`y6l5f{x@xOe+_F?sU;*e&3`L9|Z?gGuqve<|T`nVBz`>tqG#p
z?Dcw6QqV7dHRMf#ZZmacL){aN_rQDrO#DPBO?A0#E`q(M$@?&yWO-<I_KHU$#2FDE
zPplWe`CE9x!PXW#P&zlPurRr25i(16>-R7!KtsU<$GaCkY)|i#hYNz*zXF(^@IeeM
z)s5;!?(@=q080(m@gS<63b4P>S&QCdzKVZftP-v8LAw+5+v9FQcu8qb*e%t}kw_xh
zn}H$YISLFvCMKSAJ9>C{uoH*M&^d1C$B)7XIu17MX09re9d3?-UV~4)UVL1iB|LV(
zKdkxnyEzZY!It3X1n4o3S_=woL5sTB1|(6us|P+gIH+Do{D+J&psJ?kM<^9RT|<Kb
z2*KstP=GebrhJhm2o4aNQ6BDHbS{6=`<wPA?U7}8jOs^5MdwMux6)aVI+H61ri4Ru
zpQ>#P-Cz07#=w2SQY}~0Yjr8A6WY8G#CZ#C^dF=-+Sp8HX-F%;$M0C+XTBova*ck0
z$o#J@P$S9$hHSa~o9*M@G`-!N4RW(jW}j}Fmp&n!-h9{C%aT0&<9`=kztDrAjakRv
zak-8G+mkQdKIiU5U1O?v92y9VP&eM9=%7(UFBsAv_MRh|(ht3!(Vdj^20C=xWsM>a
zkwfeIbn9w#(|if<6wH@>Qc_axD<_tjd_@mZxb`r5rZzmCslth|a3yvFG|TMOst?`1
zU-th!mnNV-jSbbusFutvu93W<dDx69{E3y&<ANn&wT}CRk&LhQF0d4jpaUa|JbEDI
zxwkjL$;fzgvP;ywgs&b>Q>a&f-7q@4q$5kGTdncRUGbN$1)=nWC4#Niph@o9lto_q
zZPJ4Jv;3%E^2GAJ3fy<Z752!ArE+gbS3s0axF&CGxbwABCt;?Ze2%97#5jGuu<i!r
z#b<YTmsz9gL!5T1S${<?W4&VRJ~yL#)%Mw&Ejm9H_hofoPJ3Y8+FlMpVO42`hOS7D
z|7Bp7_b7V)T96gOE()Xd)+0zk^t63?q`QBzXlzEyHeyY0nY=EwSGy^lByYN;hN^Q{
z6!@<DUQPL3LN!)?5dqR|j18^9n^VfPlnSm&C;Fi{mFV+OK9-Wnd2v}$(X3isFnc4z
zIAjON-$mFBjV-0Me8{vq>HO@wMMvEf&t4Cf!I!#~HOr6TBXQyXMg%iId@oG+N@#CE
zvvxMrjj@|tzjr?3&yRJ<fO-|o{?O$r6i(eTMouhV!z)yJ)unk88SO1F$gz{;JT0eo
zuN@Z>pb6jWxb|9SS%m{)M5CkolXQGff^E~GWWl5z-KIvAg$6%kI01i^Ja{6-(lm;i
z9v-zGx3%tiF6R%(-R!^dm*72vBJgZVOi2xu8>((u%w7VL3|MnUIDs^SqJFh)T%OHK
za^zjc*se41t=k)B4t%VcU=ZVm`ZQ=(cY|G{m5Km9$D$)54LU5M+l>n|`*-PjDojad
zW@bF!t0C35gcqQHa>dnQ6;eW%FpIbF>WX2MRyKa#Mg-cM{b4iaySys<AS5{&W;X7o
zY82n=q@NX2298eDWfNLA<Q}p~OaffJi$7*&&Y_QlBvsI3q{f~r_?UhrB>ZQSu*2rH
zXghjGo-$hXBOq-zR@c@F2kH2_%zyahT#1RGj6SQ%XYPKBOYs@AtxUUP`Hl<oUAlms
zy}<XOQATcyZZ(FO*!v*qsD-BLzk3$g*e>F1I>kG~soBf`Zi3`uz_Rk?6l1_okd!7E
zItLHPF)HfU*iV9N<wZ>cbWos;_fG;Tbz`^gIa<~g`{_z6iT#H4F^z>DM1mkH_r$0y
zxkd2-=YPY7VHLE0itRbuJ{QY?Fuw7x!J_+NJSWOFIj?2cJNy>kqw(qGZer{}aP!Nz
zNgi-#ROD5087)M<Bxjs9;Lm8tj;aJU3}}%`xS8LMveg-uO6S;{^O8yQef`jD;M7zN
zdNGMjbbzS?GnPQG{#(cl@8XAtUi`Uk!874hWhS70dTUV8Xt*o(-QvTD^N{Uf-3`gU
zQGV+gSsovbz^vcpdG{vouSz-Tiii8Xd_-@*33hB$a)+{ZJzK1ITW96hC#IC|#pznS
ze5dzL7gYCp@opfF8OV^Bvl{kqK?*bXo}L)b2Toi&JI;9I3+g*ooFIp}zyZE!|I1*U
zla2VO#Y=uBX(iqSPTjim#~LE*&D=rXOfT<`eC3exHO`d^MZ<QX`L)t*Ny_tYyPGKh
z2n?q-@9mw20)S;e4H;UJ!e7t0`_6T<s@}SR?mcGESquo`PH1>Te;2z%zq&9dCr7#=
zo7tmYx08|&SZ>wM!#L-jcP&4ekm^2v{=KgGNimxb%s(P>j&aWKx+8_y8K_B6@b9o`
zELxThP@!uudI5P;=DW-VfE?18rzQbmR}P0_R{fL`;q7KTCp6uB0Q{m!zh^U0pxv2c
z{yx4OZO2a@cw1a3r!gQ27m!tQo=3rZ#)<Ng)GFU^Ru23I4Esu)+!zOsY&V(L#3x52
zA`t_vgFd$V(0#9BotpOXU@(oYI{-bu-VDs5R`B1gdHj3moZGtL;tK$iyr-vVkDKtT
z9OQ^ezf}jd%-E$0F4%)Sv*jF&L(M#!#9zHIWA<8{ZM$_j&`d<v3zQ@Dj+}vHTJ!Jh
zh0O!czZ>cVU(L=4v7?FZ1j6`R2ld1p`hO}>bgglHMaNp4zxqmNOF6FC9@0yRJ<tqj
zxt^h>@XMmgFgYb9c)RtgodH6NfD`Ju*(M*tx#dLJn|vK2OBGQJEYS)leial4^)j1W
zzh3(^fNqNL&X6Gvg`n~~6k<6MBVgeCg$BLlTJ7=V7W99Ew>d4OJ2`59@_HXwXZT1!
zuKE<38JDOg3dhf{PG*a5{N!Zpskj7Fu_bNR#Y$AG4Sf6?z))bM!@5mk_zaD!VT-F>
zWXXi~PxV7+hRRF)y|ddy7Ff!>+eGgSjGZgC=f!p~<pWo1Ev?u~{pu_HYi|1sw9+|2
z3aa7Kd$G->ez(o94`E8|oLX^Nx*<bX4pg@j;~3MQ^ZUeO7<j@a!21Zv&g8!S8bI#j
z!lKf>=AFKDu?Cn6;zCxyUJ(%1KbLsz3bBxIo}OTjxDML8-!C^cqZQ17o+U2C2JO<&
zHZ$Nz?KX>0h&X?^6ls8a$k5$+)xn~xP)8LF5|#eTRI!~Fj5^`^?)gsoNZ<M3`)1{d
zlBRfG4m7b{cbNbciMJ?^vq6ok47(;F!K1EHZ_V)sf8W<b*R$P@_4=zeuIBjX4Js>M
zvYv{KSHjbfttAUT&psDnxw6su{R5S^?#-EyL-L3~#Gc*@mMgT~=Z1$dUR*u|r@Z=J
znk!?)upSvs-~NY&a>U(n|74f~67y{QRF&$k94nh9xN*SB&t$02Woi>{FZ9o$Ma@9R
z928N>o$7~%aym$)Pq$<m$EZFo5LG~#3#Fs|5coUwHO({<K<-((j>qZ@T<@zUyDUbK
zvPWeSwr%Y{nTsy9mlg2&vq%32PpU7SzViHCKH~V)K6&p+3_}8liR7$xsY)Z|a*meS
zeW`_bjUDN3lk-~xy^|}hGuKuJoLMJvfzGxQfvp(Bq@0Kt&?Eg)x-QyF6V$={w*|nN
zPj`+C6IUq<soqh{p!A9JY#^V@{z^=zsQHlE$=cfAPpbf|*xo;3XX<etvir}S@0r#K
zHpXNYsxz_GP8&@6Mo*1GvY8=)Hp1Z$7Zdv!1?(hz$Qfl!<s`R0Up+x`$UW~n^FzeU
z-67AP<)o(wZQDaPRW(<aKfY?Otw6=W8X|q%NQACSjB9vh>8CTn#){eMWN-iB@wFKS
zXGmBZ!DJQ64&DmDYN&6R<G*+-4cFbnre#c11+zh&n950ez?wpPuwRAf@$dZS_(x@n
z#t1s#Yx8+^%*maDvCqgR0gu89T5Y@X3R?U5rK`f8E8yt*+uUc48%x#UNy<!s|5Z8R
z+Yc#x#Sk^ehiU-COJ0Fd;1Xa;%6a9bGP(Z6l(fE7wAQJDeR?=3=E_6pIg6y%26x4r
zp4E+}5aw~zH#?+kT@QPpr=Dr*ee;5dD8?1(jla#RK{+U_fDszn(I9)jtoTd!Kyyb_
zb-0WnF8Wo?n_HN|hm|AqJs$C@&-z?ON1KoJ14F2ixvWOOjY|B&`Y9caabii+3u+|l
zgfxucPAhw%2w_ZTcA-#0C-1|?C>dg8Hs(~*<;A~qA^oP&c+-!cnK@1(-8A0532w2}
z?i7CW>)qtj!l~<8y7ZQBJ41dH!bq_{DKi0CTO6jQ?EW?2?=r*Yt&N+&(B;?AxRh9w
z`&PHwGK$yd9GTWRs+s+HBf~PYuI?0c&UL3tDd+t9qpix__xN+zrXOKad*~8NC-WJ(
z^tx5Isw$Yv(%ODE<OuGy*d>8}X7?<#R3$dCi)*Cg)k1P1_cAQ3*L~vH7tIjdqtIm7
zfE3e1aw5e#dM#`BPU2InkxeI|Spoclf&|nZt_guHwmpo|WE<IRqRcl^lZM^SgYT#B
z%sHM5Q?Dx}UYjD~WdVf^NF7<dZqiSiIKoQ1XpSKZ5CK038|}^qq-E1XLmH+u_Ab3#
zO$RYi#}0&dmmZqyU4ULy{Nm(k#->sXh^y;?;B#uP8@~wRobP@)*#TbQ1F`e^x+N+9
zw~AuW<@ToteIm&D&}-Dppa?2fPcYt13a<bfMg`;-Tt*uoHJCLZt=j@Cia#GwvKm(o
z&RorBKh$!}8@T`Zk~8B_gR6pGuyV!W%PU$4cp~#sTqDKe&_%gHHdDk})q{n-e;mb7
zP*k)IoV+J>&bb*tFHRS*i@#jn>aKVQq--z&II$o3{qSKz+33!dupDSDh2I_PS{{K;
z)(0}lms4>Tr#)QF(32-&yBW}TeILjZbHFHxo8TbP&-SI3X?<m;m8ZPx^+DY6r?9j_
zAfC}%&Lv1IugH{e#R`IkMRYsu0nzP;65%nM0=gr5D-i$e*R#F=kxL1y4@aP+pdi>@
zf@xf>p@JPc8UKhP_(*L%<>fp4(J;U;gtgp&QyNbjEeCSOXD)_(9t{`G{}f^vkjTJ|
zKd&k~*x={f&M2_4faK|{O4NPh$>rQM8hVoIN4})640oVv(*6Mw`1;1#<z||Pq?Yc#
zj*OH5{8nt!V0)Q(bo3D6JMuF=bg!$|42nu>kVRv9@($HX4u_);465gAbc#mn<2dL#
z38bz~!e8-jZ%1lvbK$<L6C$>m+do}rsylutGUwCp)i3VmU_+AI0>%ebxf&sUbx-qd
zxsgjz+~#;<V2~3km~Qkc&!gtt1PR%tii(PyoSebb;nC4sL0!^Z=Bh^m3zCU$78os%
zQcq3j-ocDcOGCk)7%#`tQ9Mx4i?deU$@8d4zU_?a2IlLT29Mh3O70NQYiq9~#I)`{
zjXw0PbG;{Zxet#%EPM^az1Q!zgNP;TZuRql;G`rn#QV6oG%=r=zOafX>?1Luk&y$v
zz2?L>-k4QseN@f#2u;$vhlA)`+ylnrx+iJG1x<sVo}Po@qP+Vwd-9D!wuUcXzHn@M
z`}pL1DHZcMs(RjUQ8^h2lyKnC6A%z221*Js?w|TvSy>_QgJr^e%t(b5)<&+2*-7Ad
z6QW5n5Uf6Z&vWV7VW_Pg5HfXk=G&)t^YHL6Ffh>d(JgH3z5RU*|HQ;Z5O6mgz0B2|
z{E3Z`A(e)#00I5D`%Zj90)`|wB#yqy(<y=%8GDB08zf{2=Npv+`CM>;{O->R4B4!m
z0DM?$-gcYPS5d*`elP|u$EBzSD4Z8o9^c)wTL9lgvSl^b#IPsq>AV&p5k$Y|=MUzX
zHvg9kU}jmv%xnPxVQC;EBRzfg?|Jee){D2f1*}WJw7W}4NJv33eMX(Dsi&?!s6<n8
zuV#X7<)!u~2YdU*XTL!j_Xyl(b@o*m2sF}_wajA}hJXARlv1mwkfl-uRt3;unx8Mc
z9s9&wqQnqG?DN*y*%>+<z}FxfX~RHA7e7DH5=?0V8g+NKQm~A)Y++_*<{<*fdgWs4
z<t0*y#atNr=g%LSj-QPwwQcrv1i`>rL7}Mla`F?WY8+C6f`Swnx;s0Wxwyt%M5-g?
z&?M=iKjh_Q-_7-UWMq`CV_v}b2>g4!Z{e_F0>9ko!4f=m$f1cyNNQ&qUH2R^)tLc(
zN)%%Q8`rIwe*i(0m2t-~oS&U-Y;HQu)Gva<)v}V1fMC!86v$xaT2@)v)7hDs?qF^0
z>2Pv+I^Y+@yS23i*AxSeE?5tg6c+<DwKPBPe7a}hY5>kA&dz&K+&z+(R)`4;4b@`4
z3oTVeMa6GHx3|YiQY;fcRr?&QXqb<D|9)kvkO3q-6B83v)qWUCLljE4L_aPj#(C?f
z$lVA)Twp2#nV!d@`@LSJWjLr(z)rSv-F~tn5L=5gp=WV#Y*X`ARccz=kPzI%SKO~J
zS%8p9S-ogfI*Xf*PMU$B#JHK7O8m3+P&Vl0M1+M?#a~Pf4ede>j)jGlo10rNuj8WN
z13tfI<(|J=1KEk~V27o|yPnpz{^yS?L<lg(c^eqmwYUfNzZbh5_ZSrWflfMJiegYq
zf9T~5nQE1cxYr(o#dv=|Cooq5Wo2dtcSBG??gRNVXG>;z75)#b)xSlbzDxnz7SiE_
z`1l&z(ZaH_GGKCQd1TYm8}KMqQ~f(w#X#tm8fLvLEGmlaGG|aUYH;5KTKnkeDEy+w
zvuC6Nwj)=Ma;8BZA6UXqU0n~3HmA`f`Fa`!D}0VNq}+#0eXpKCfi{@=f?r1)Fg52F
z7J_idoR@mZC@4}+PCVBNfCYA!hzQEPHDEjE$VW2WFq{7#oLr&65Hq$3fFL?T3LlOV
zfL%qf@!{BDTN@)Q>l$>KBZGr)-n_{Jvt2;oig*rk(cs{&|6vd(rl$J->AZ7LJCH8+
zIFKbyX-H(SL=qRPFHO_~{Ddkk`yRNisLRT>K@^1P1}lBi-d#pU=E;)@z!zFuTY35T
z9xExqZZY%o=ZP=a1N}QY-4727<Nr32`<X_O0J+!!3`WPFpJ9tkNJ!{tZ};~0&aR*n
z74-q`GZgKzIyz&)g!ic*dP>Hao0(<9?16ht*b9cXC5w7+5+y6;+S=RumJEIVER5h5
z5;{IO=<Vu~Iz5K|ehQyegi<am7gud{b({jjebrRS<AnF`Z(yQ}XUxII&GWSV3KXB4
zy9C23ZT}U+{;}`~Ql8hWd7w8VCnW{fn|}}%85tQ_;<~Tu)(mk-nbsj>O@Ieac(@7_
zQu-xVOJ7B06PSn)&|dysz3=1Wqfe*hb{qi|JeVc0Iam#(r-``aDKk-sy4!>0?Xzb)
zOMNM@IRO@|5@T*@nHLn)0jvbr8_dnkq2Q&Xr>CN%1Q<jqut1yDc-?=HoD5+77>3P_
z4e-5IXNJwIbtp%^va-^^?+Ce)n5__Vwk%?)eY`!C6BHdi1nC8g3<80wca`FDf~pP9
zil9x_5(7;IR)iN2ubG+o`1o9Pwm=}1k3Q4h&IB)bhnYDa;tMPzkVe7OGcA7t6lDy5
zSXvkeRL9xMP!5EUOo9^F$~(t^Z$A_m9esz@1CLS|bYNsIE-r{?@Q<@IjQjeJ+@30%
z;lVUls{PIB+MzenSvw$9NZ>?4B}Ruh*qjz(X0{&AQ&G??+2Yk6<1+s8v%9<d+cyk^
zwa5~zMw_=a&WlPUBqW}b6NR5XeJU=Nxfe|<FE5Wc0p0haa~<%x!HXz1HdYsGj>dT|
z0RT&Ohn@w@wF&zg1%~-&!mv@nVt)zJ02qzlzCPeQyA%$w;NJ)f3xkT_uLNj?GFyF1
zp6V26tE#HLAG)Gl)Ym^?W`-U4w#X93b-aL}AfV~)M@n2|IxaZZS<5_>knjVo11&8r
z!oTrzMD#hpb^HSzRIs<gvQTd)g9!q;kmo~cH);pS8<do~VFsU_ox!mkK7C`XsE&u=
zm+N^AwqyM{Unttv$3I<1+`b+8?%g{OooF2oy7j*AjE;`hp7921?TWdpb|u*ZS7wX9
zu(|*K{Tmowk49CreN2`T8$7mEul)n|U9n*u0SflgMj(j-!I|;K>w#rJn$TN-uF~Dp
z0|^WS+JZN+WWtv38HZ9rLP9DkD)tT!ac|r}_=6wN@}EBdhQSBXGK%_vnMT}fJIX2{
z(G1T{6K!oi5m0h}bSN9)JC9s$kAJ#1S_;qKWvN#xx+Fhe_~qH>tSnk`UJQTOnH~rW
z4-)9HO*;!y9&`h_=OzMrDtLH!BcDu7O#wZW9|ujplp7N!sv|7o=ITWG0(hO|r&jnJ
zZ$T294(sI5%DAn0lIIe%74OI3k-m+Wd{hI(c&lHD8q{L}&OSGwb5xQhseUxb-j9xs
z4x35a7HGhRDmx#)l5=>I4MMv<dbYdR1s@d0dF(jfOUHSv<t!b7@5RW^Ey!}|@NcLW
z_jGr=MsC2qBk)bCG|dd)Fa6hAmuM<+w{dVbrmEQxIMfdl<2w_$j2)@jLvGZUfwT0^
z&S02l=%8~igQBJY$9G`qBU)Np&0u%rO|l2Y2%=?q-(aJW=mzG>>pn|6a&HIo<E<Hr
zyLV+yaUj>Y%k@ffu-VT~NJvP94d5sUI>h9hQ~-l4se!yHyLF_Er<NBHur`?dL{Cp+
z95VGGWy|IMw4AG-(5v~kX4zK3%0#r7l}E|1!Z;9JWJ<lMx<DYfU3&95BQmjMG^}Cf
zjATe%l>j{nT;o4py?Q!kIBsWVcJ;0AV<LnfJaC*SLqnLEnX$98pPirIrhZ6DN}6x;
zd?;HMhtkE<)AQDL_2f=TP7Xc65yTRGkGBB<@a`bNDRQz#b{Jt|YWj+kf9@K>_h`~e
zJF#BaGEmyo&aP*9zW{+|0)lMlH-gK8)A^)sR>RY!DuU%Bh~zij&H?Bm-j;fU4|5U6
z(p)*2-v)j;`H`7^PAA|7c3OBb46lZ`-!d_4p|!wFF{rwQ$Y&{#h18II{n}qb%0W%c
zfHCIGB@?Sq1LmLgZnxrSbaZsyUk+Xyxh)l)l2UZ~ye))?L)>k5URFCO_xt!B=cl&u
z@q~e8p^RE-hx4q<cg~^KAvg$Nnxp$eu>wsScUWT9*)2tu!|G3iAZgCe%B0WB%Y%mB
z$jLo{oa)k2Jb^TUA;YcW<*7Pc3PJUz6@7kV%N!A}e;|`gR@|x{rRRWP2RZ9g&&l7B
zUyAed5r}I2aNdX|R)jX=3;N0qf#J%fVxa<UeH2PXpa!3W%jm-z{#_iwu#pWsAhnoO
zPBJLcp&2)NmU|!C9$10%9lS``pJXy#fZ<S-aFyf>x4t3Pzxj1s*6fdBtE%|>yu&y%
zD*}5KYlLlQ8%J_2$yY@%qQCwATvylD(P6jL2VP$VqqZu$mrNI4vCXdFYyoK&M5_AW
zSxA-Oe1NetEZk-FyTrsV0sR`(yd`;$Ni99BKAI(2Z7I7ZYxnqAo<!S;fG>aYSIMv$
zwW^AWtX><G(nSY0R9Wf}8bkH@>U|C%wArZS*Hl-J&M+e;rl)IH+*Ta5WQAnINqNj5
z?jSN3^&lQNySTU+5)~rP{WxI97j<2U(g`iY{k4DkNhua)!-&GO{eX{}1^l=4syu?W
z>}TqUy4Yg7lJDzkCk>}kPR%PTT3(;EM)C{Pxvs|OsM$L^SEP$>(^au@vc?sjI%E7v
z&&p-hfE^s>%FK!5bd8FO;dV9uak6HCb|qTjE!en3J+@?WL>zh{eW>~(MK@eX$G4EN
zoCv|<0R;3jAKR!$en!X&>)Zt$N{ly@s`vr9xfbik>Qo5W&m~E$Z{k0NWVRqvpO%B8
zB+j^6Kj2f_%uLgV59W|#BFbL;bJtfz-D^PjWBqoOs;rWAiY;Useo+U8Y2XFOB83mz
zcFtku(mm^PpJq2<%6`z(`?E?qU}X&WU7Xb|iRS*%WWH;iD93!aKanTr@R|OuJBOKx
ziNE^?$SZUp^%?u*x4u_0c4_Ofd>|)?B~?)FHmq+)nY>I=&yqB-Y!}|c?XR;<ZU)ia
zIy+dLn>lt!xqZo`<0m@gI@IF6p8MOzqkXLi)@wdqUQjJ+Eb<Dl+!<<Ilg|~Y4;R54
zO~K`;CI#8I3`z?ZuY~E&o9`sV#Fkc8$A!izx*o|^PFQ#`ZDuMUaY0-OJ3lXc?dS-4
zg(`6}8ogI4l3HWgM@~;q&(EbugP*Bkst<+A6McQ&qSs|lib2@t3t%Ts`sQd62{QY6
zogtP}h*U^LO-*T0QO<^KKdq!K1u1EjX4UGvvp%Mt{TJM}IaHY7Wf2swm=BtE^{<6;
zl0uzFMh0t8IUz+pf|87qGVJXr3nseMvuE6>2FP1-nZ6EQBZ_=*^uy-s!tRQDw)IyU
z#^%EEAS#!hsze)s`@devm745`w}+7tl9Am+0O!`I(UZsd)LzhlO)>E8Te^?ZGm<Lf
zw&HJ;N#HB0p*|Tk>9al6;+uht)7F3sEEkV9CW(oOp<oJ&OglO}JO{JE-$VkFOGKkf
zwLHvQzb)p++6>f%tXBszIVOCcS%a=6xk3vEbk@#i5Nza!B{L`4)v2zeiu(G=Tc=5F
zW*VOWql#WbvqUI$A!zPmzj^&Se|zY$ACUqxm<8@GbntO^FYFGU(IX%RePrzeb5JPa
z2ybBOYbGWqYbJsIdK9)?9LoJ7>gOIFR^?AVtzlKdlHBh+0x_H2k=yC%sh5`*n*Y+?
zLk|8}PAU9P$=|22A+)~S`_t8*CW=PG&aS+!i-CdBm*i+rW3Q;JEHin%@8~)L>Jdwk
zgpLIf5fPg)-XS`4W8=B8>&6hwpr^Tt*G*IJzPa}2kBk6qadGjWpg98n&K)#_z2HA2
z?ex@?zMfuIJc1*iB^I99NSJF&#3>R+5oolK9gKXJVX+6^vAtjlYb(eVJ2PbiSEj1h
zA!WV-#bF17+Ah=V8b|(MWB9;wPSpiSaAA_*P(5(noT^5c*xGg%Gi=T_iH=U&SX#bK
zCb1W!V3fag>lXj+9#l}AoSa|3euaeT&Ye52UcGX3bhH*MwoEbbP+u=IZhmRVd1a;q
zVIPt@kL}sM{{B22J@xbgFN}n(%sZ|9{rwXY6W_mM#^FI!NN%>}EC)<(5%zpIH9{bv
z0da1sfcOR0)mMZD4I(_K5p8*)ZwJRgNx?iNe|bFPp`oFH;C}Y2<)d0g_8MQM)Qi_N
zu-7P@o2sb|ETLW}kM2XpBPTBp5H_TLgM)*B7{<KP_zq9t{QTVOVS077z?!Y|vuF4>
zFtwkC_F2*on8U{FYXG%6r|}m`YHDh7^1p?_pFdeZ%MViu-1pihCK5G2!}o!TG6Td~
zLEnZBu+d*37@hCm{|LoB1sNHSXB`~)0u>@;5`scPA5*K~>6Dl6w!Fds)L@XwIytux
zD!?m~L$fyQkz%W*Ecr0f-@krIp2FD=-mt7;LHqk7h(i~bmgMR-IZ^eQdoRFvVkHHn
z+r&gfJZuF}4EFW*`uh4_dDbD|<HLek`3Orl6DN=<hlU4AKrCGH09IQ}DQ;?MY+T$X
zkljJY{3&F1PoEN!lD?b5g7JgR^HsvS(pBPIg?aa@Svk0>@`AAi@pU@m(2NXY6O&)?
zorXI8XM-d4>I0bs<T7WlD8Xw3ukm0Q+8H3?UkL<L)BMIpN{2id5-qJ^V9DFhHcrmX
z&2@I(NV3N0I2$nF1Uk2-me#`jJWM{w#c#ZlEQ2ZpcCLqji+~408qd?W1qB76p`l8-
zR~EBg4~<MsX%CCHDo0sBYzX_jnXN4x??y^R1?{E>=AHNm$?0=XAD_x6>~QXot*tl2
z!#{t_ITL(Ve=wFvy4D)N?KU<jNl8CL#tJJK;D}PgI%i<(!~eecU0wiyl24>Q()K-=
zP(GZz8v3Ph0D&D=ZjUod2#9MprfWY{RKzfJ0X`?5q1KcM;6n+t!r?4Gi$kdRegn(^
z)%C&QAru513ou-~p~VUebcz))|6%4&S6Zn-EeiN>$D%W2C8^2D<U~Y7R8%VXnApuQ
zC7{AX?974N7FheioUyU7!P&S1Mnpv^9Ljbei-=)JN=*&348g+0+?Z{m=)2bN>^Hh%
z3`BBaVNwoy0M4PQ?*;MA(z>2KA^TlK1m5l2$9sGA9@|C$5P=!v&!0bGCR71#2v{N@
zJ6}NDoRyUY9YPi^E{Kd?P!2=zN{os5@sL@ch0Fox814%u1J<CLh8+5V&yl^GTXjy(
z0+{g-5E8PnvI4#dh{JMU$~d?It*zO{FhETzkU$<&&ynzhfrX_2oM-9j7oZrmmcXqu
z`~tN!Y&L+&EC8AVMG*iYZZ0k<;AjBWGr&7o+1yl#`2+3-N=i!k&s_C5Wrs#apamB9
z{{5A1790qy0T=kcSQ3mAQeFMm!a{6x^m8Z^6B9}4>2t0i42hSi3|&`CkenQtV||&{
z7SmTwSnxHZvz|fF20Lx3>6(qXW<Nmkr)!;R;Z;_q0I9Q@8VHYkp}v55@U69#e>{)g
z)sQpy@Er^hRQhrelp<!|-$8{7#I{4I(P7fVqlMi829rV2aqbJ%rpx^NJTG81_wL>6
zlZ3C3s|Y{~>`#yf*x0N>1qQN_&+eO3Kr=Him<HT4rzGe~NwTo;6gEyLr)EIgmZBML
z^2`ggVawNlO#`($D=Vv9IQb~pv{lKgKy|05mjnpU)=UH1FDS>ZUAqQz8d}t&-IX64
zdvft*Wo1{g%gcH0-(Los5<woGuAZLZa|ydWC!>xqcU#$UKw6;D%g_aTQ#RMAt-#^m
zl#hH^5oENnU2wM%TQ__A;wS8|Gjp=B!SMrKU0o}E5L4AkbROsaSTag_g@PkLLjFPe
z{`bF^4t5??l{1ZAd~i$nSjlt4&b{DpRj*f|4a3oRdJ8V=z4s?L?;Z@T(&NX&Kaw2H
z*VorIZrE1WXJ^y#@M!Fh!%%3AYZe?pAYOSOAb@;K*v_TP8W$py4aSRcAqw{P_Mw6F
zagSL5wphxtu(Z^?r}0BJHrDvoLMLmSQs0b8mFLlh9$N$XkXvuE0t3N&WUgjGo*5Y#
z85s2DW6{JZL6l9Ac-+II#V;TLSQC`zP(J|_3jNVOV?dL=y&oKwL&4q_N|G-p{Z-OD
zNjV;%RM-Kvv>v_C)7K}zcP}F&Ba+~Wjt=qPFKruLEj+Odu%J}x48btnNJmF*@h3E~
zGEofGodW|T!7^ZMGkz9R`gH6=ef?kfW`OcCC}v}Z2?_%AkzZ5<ATMj&8geq=RcHP?
zI0dr0x_X78TPvHf0)Q}Rd~$I)@nFtSRT2{fGw|?4f4wEWLaq1OzOqk3K%F%Xo{pYz
z2q>bS3HL6UJZ2!Uq^Crv!_l`kHeJP)q2(y(xW;zntD^KXRm_t55}^eF2xYGk{sNrW
zp@%~Cc?RBo+R#}CEwZRCb4U>M#&ur2XacP1%~$O~`U>l!KjZzEPIqD$ay3~h`+c>y
z_)<=WM@Gy|Oj_)BtsNXn0Mx8BY<Xf1KPL`N!rjgMOSZ6wyUE-B3&tL@sl+%ZSmsMh
zOK{Vt(cq3CLubm}L4nX_jngbJ*ZGo#T=1*K_$wd5^OTzu2W#Kyny4iX)|;;;Q+Fo*
z=dIxgPnQTimbf3{19#zlZqi<3Ppt^Oqk~@E>-WqC!mi#5Gitoqp{&D1Is5#V4$tPB
z)T?W~`Fa*7RBfnqK0h*LY&hmS^2kT;%A{}1>-CPaE@A!s23mFN-n^wEZPxbt?r<<|
zdrCELO{mTSj}LK^p5^*IZB`;F{4nFe%ZN_f>SUUCeo>taimY+nH6(f&_C=AMO1Y4x
zS2w}r3z?*QAl4(-&Z1PJ!{>vKc(24*?MI2a_$c!dX%1&oR@cZ|e5p<>JR6$nRVv%r
zpfZ{3R#N6nnErX3^Qb#94o|b-eV-+t!8uE87parXM|I{z2acjhRG@U0Hfz%4=tt)F
zO1UK&{DatG^xy(IO`w;n$x~UR&isU(m@gVEwCW5UDME{Nipb>Q2wjl-_Hs@g1(kPR
zji?5QuW@j3lbUGS>58-q!7QPj4ltYcMK~O;H$ARr5m3n9R6FP*RhRIi4({kN#qto6
zEuvfE<wFOu&8$i9`bZ<b$yVac^tx(XG^Vq$Il0g~gH~_uz<t2XNvy|HLj>^-v?B>U
zpQrjKO7%-DH=O)evFLe|iZ^BDL>IV;mF_h#DDu|yc)qz)<%Cs9*)LpS{gkC$nnyJ}
zpR<30a{*TldOa!I&Y$SRWOZ2M{*I)Qew54A?D8bEKX4~?%KX0Q{NES2`rxB7F9b=Y
zM!z>V*qd)9>Mg0FW+$XV{~K@<#KH4kzd#E<lOxR8Ilm>WtC_@oX*-+9CQ#*G?x$sH
z`rj;T#-2UBb@ih}cTD%*|NgX!zU*l`OeLH=s$KUyJDyoi1cKCw#6R8Qh=v=5U;Gi(
z&Ng`%&}O!}|C)6Ayd${x38Uv@Izr5>$WA9l<cgkQBT@8W@F#J1GI^bW_h*qlI`^`s
zXk?oT(whHgHa4O?(-r*h18v^eNaVS?tBt%#$A1%8Srq;MnfDuzel%@f?&yxleoW}o
zyJscNTopUCzVs#fP_aRX()WaJapP1oL3~zP5Rw1b%!y?B^OK=eCr*{H^!={i`g5je
zW1mWj^V2UYc2njumQiA49k@8|>f$G+iyq~RjJq9b56jmVoE~*l@c#St`gBhGIrUi2
z@9OD|*v0YY(>-m+uT8PKxhU_3^5NUXlSIZE&f62KX5(MPrOCA%x~7WOtvD`my>B1!
z_-vlF=JkGCcW)!MKncfL5Y0Q;Ng@#Wh0Yc0eGba2ecKkC74)KB9WfY@tk}H6ytaL3
zhV0f)M(Jhs`17cz8YNa`x3)XB30{e%8UJgvR1cFqT6p;Pu=k!KWI(q|DIc6nc;(|c
zpG$auFhBf;VSE;4sNuOVvG}(1#f*xpP+?@}5@W)bS?>!Qjrjr>M$Q-3jON~S7>Fo=
z+l~@Tq8`8QaNM__$&7VCLrCr(=%jWI>9yGI=9CWJArq4%dRs_BGVb^MVKeLT{BOH%
z#}_}tE=%9zWU^G$|Jhp#|F~b|{8G$0a=7UuVH(bdPo0v2#Fj+<)lu?xc-7Q&MWOx0
zGp@Or2a4W*U8>)Fq%w{S-U`b&OHof?%1g$*H<6f1M~9|hBddXYF}mOQTB(2>p<T_T
z^!-O62#yI5oynhz=3m`fZfxLg<yqCQAxJi2NS|qXF*jp&ZKhyJC6Yktk=$oX#-$>J
zB+6LZ(7-q*!`Sc>0#Q-EAzQ7jZK$t?j>xb0G~*F*X1(+?5M%9AJF<t@QDXbzt9-P^
zEi13ZpD&6(Jk?1`h#K~ii6jXoQ$a`yzQ2ncPmt(iD~^js&=h%V&_=%}8fQjBj0Dzh
zlZG4fuQx?@>^jb{MlDO^<9S`aS>+=RBmR&VKU}Ufmdns{{kD**3M&%Bzp*~Qxb#SN
zcDMqI2oaS%PvWLh-DZKyfRJ2xQ#9~;HiP8wt6%*hN-;XS8MU)s6aJC!13z-?kvZ~3
z?)Mkp8z3dpxPxJ6qX(0ri*4Jm7zcHb(OLIcoYKB1j|e3u<5+=1NkImQl}KySwuswn
zl9H)NYC5((f_<@%12-ejOt_Oyq9Qb2$T)VNE~Fn;$@1k8_aFJXiWx|CuyaI9Auo1X
zUu|AC9cW8S_uMp)UokhJY4h?>-8eaVrW9Bo80nfx&wEzQtxeL_mZW0%K-8@NK7D6{
z;;3uy8ZED0TDqyo>afQ*s{tI=$NlHi-R<$(jO($-PG=YHH&v1j@ySL;k&Q;f+3=$4
zz<R=a+&y?fSy|7&`$}VH1r22zqxGF3KeW`h)_M4gU9hglr)l>bX%3|a73b4$mWPXo
z5Qw!9-rTgDn`rje<%g58S-J#QCj`C^L|pLp<sAf&7@c=qGswL!jzr_nrAF2Moz{B(
z{NooR5B?&>^MiT69FI%iB1vg<1Pcp?&~sN44AV91Pld&1r_KK+N#%*7<|~}z>l~5Y
z5*mw)=58BeSsowHzAKaYvmXCkLK@tjQ#|AF-b^=-E<9U$wJGX%cCOfjjgUkd$q5US
zqd(iEnm{7i=k$8o^^xX3PZvvXY`SqF5V)@tWu&w_ho?i{A}hohH3)Y4%Mc2Rr8KXK
zgJ+tTLQy1y+h<kJhd*6owO9&XS;4qZ*prMvj0of&>O=^#n@m2Ke8*}xir8y(`S{o#
zWnEMp+1XvV7m~U^XK&Q(xZZ_xhpOoB2WCatBn98gmhqHvClTFlzMNRS10pg+GVjZ#
zx#!wWXeBtMk(~i=B2_M`*A>f2V(+GJxfZX*Ipk{ocsGD<l9ZU=t;ERR`-6bYzpZw6
zjcO$z*_gwwl(4FT!~a!FSM$f(#h#IzLk$n>V4R(V)+h_1jp0eLFvQ(B3p2u3cW^8J
zzEgg3+u-j@kJaJtYzY;CDAdXU<Jx`RsOsk(ThWd>t`=kR<`v}?XQxf~YVGxFpC6Z)
z=~qWtR`GQdb3Rt)!NIO@y|u@w;yoyd9%XyXb8A_q7scEE>nP1kpy1D`inr6}1Qs;J
z2jx88UkJoko?mck8h6>Vjh|ids$96)EcE3qd=rm;-J?-Z@y$a={3uo%fo9_^iIw;l
zhgmoZbu<2xcW9~f5D4C+*WVTqh-bpANnvdy*x?Ly&9Y8k#-6WPKe$keONcTh*|1uo
zT1kzKV!V#vUJ<kyYPhp6ezWKcfw5X)?rX68!+R!xiBRt{_cya^p-oBFMo6+!*cQsL
zMkyw{)+&W)#(cjJqkMqruElU)Hwy8g)p4iM>(2H|a+?g#^Kxv&gC;&JX(2XrghHKi
z8xOEWgej4+VfNkPhoQa`cO|A+*BN<DC!g+}sfinuduv2m$#V^s6&i=c^sFg-+R0@T
z&#-yH$D~@^peH7ag{b&Di+tW`a@|`*n=4vT;H$;sp3o3Gq+t!rw<Z}Kcf2YUQ8!x7
zt@-jIaK`7I>H9%Q=Fh!3QCB&Waaa`3KaT4?Nqi$cE;J}~wC6&*B(naIn)x6rPTi8u
zl6w5faDv2p+~nDwCmxmF!C}%-qsXyNNk)ec(Rl}idZ_37)*&5LSyZcznYkIA_O+&$
zdM8~KaT#iS=jY2$$lGLHrpBeh&Al|8y_$9}I9AuNkVc1(6r1rdjhTO;`DYEHQwv2g
z1xQ-!dpgXW_A(Vk6?|P{k50s-p^yn@k|u0_ah+^8No5gty0+{tod(iO>crFJCOalO
zz&0x66y^*)hSr>lVok@YJVRc88kX9_56uX0M(45OlUT9}KcBxi5NjxSJnot!P*KvY
zra%-&;@7-bTBu``7{zgXt+iLW1&v0T{aQEr%`V=WGLuE4pNBr;tHRtr`jk-&ns12S
zwS7nm&Hi;kXq4%6dDGExEe=#tpCcnIWm#^jQ4sMM)Jn$|%cL6$WQ&sC2(ZLNOiNo2
z4rXG-xBO`2-Rdm(qyOddWM~P?EE?UPo+P+#>FoafV-ok|mE(t63_22=1a2N9)uEY+
zT+v@gSX)}da7|{r!edJt+W+Q458TB^SJzIkY%|qCM)nYq=RjLQcFVswOwHWmGC=2g
zSUMd=!mRitzxCm3@rTITr)*rU3<H99CPx~_Uk_m3yxkMlisgCXSX$ItC7aG|=Wodo
zWHF}fGLxC)U=@CxJEv)lmP{8EbX@g=|FNKD#?QYl$hk?o977x27i)~XdT~jb+js8@
zvt}3dMO5FEr8ji5|GWb4zgX{BZ%iL9Oc%SWE7tSr;HYejr@}kL-eh4WZh#(<IwSfn
zM7ExUq|e4UqV6G25}*5I_mChaGxd4diBhsTyWT%;@uIIC)ihh>N&kdvFcA~|Z<aYJ
zaS)l@&p9J{Y&UFUQyu>Z%|(^Z1bn=3`Eo(yf6II{*UgxbD10-s_pvhbJ$}-80t&;A
zWQ5lc0eUSTZet)2owgEN>3>2@&mYG=M*Vxxy}kX#Qq1R2+k;aqr|>L5o{E0#C9$oN
zKZE;vNI*vjbz%0u-GPn?gAeUQ(}DWs&S?JYBDQF(v=kHp%?u>nR3lw=V=oDDSdT8x
zq8e;fNo*X&dpOlS8g3KVeItHBH581yw#<`3rBr$oUC-Uu+2Pr7{by~RkF%8TFia+o
zBF^%K5dQTo2F`-Hr5*_f*DE+<1M_m58Xp_|k>GH$rpbsC#9UoVv`G5<Zi6O#?qV_Y
z`Fge?+m?jj*r=)h7nbhMS55{|QH)3Ky~?Z}H!E#G@<?N+sj2t!-tek3<_lT(isH&*
z?^LeUO_rCyX!_WocP_Iu*0Ci@OC?&{JpW_-E!w_~Oj2H_Um(8w=nq37YpZD?OjPt`
zNd3uPaO)uBW2`|6QZzAHCW}a}?{|gS?>Qc^f0SM+bh&Iwr*qk;zaDD4zt3t%u^wVH
z;@=;krxtyy-|@lI%hey;Ti@Q>j;o`>YZU$^$EBquZj+7;v?U@;ZmE2@jqO?5{c+L$
zG>1*-)<(Nb(}KOafx7A1_~_AqEy?uMw=eU@3ncf*0{Tcd9b3^6>SM(5yA6LQR|TVQ
zNB=b8?Hy0Xl_b@BhCSS-`a1hQZ)`<F!WNrPe<A@*R9us~@ZW}u$rn;PC@~x@)wFu8
zS+5@lMmqENSrmgkoMwQf(jFe*-639&cMmD1gztm9X*$4&z$p&1rG9TPIWAVPJbE7$
zQ38sNCa3H9?05D4d0}7T477ahp*v0>)ZH(Um&bGGUR^OgKQrq+oz@$4Y5(Ae9#JZ?
z^|HQV+(ZrSmOnRPKJUGvpEwyrg>`C9XX#z{9=e1jC_WmxrDNsSA<+<cL&{2Ma&uuh
z=EKZKwLL=p2_J80Z`4tY2BX?>_W!I|Ba!cuTW^}MSA-W-2FYIJ@-@%AV4@@zXpQ(}
zhd0hA`&`u1jlIG3EP+8|d}p<(*e$#yo$l5|CTDM_z|rkh8K>^ou<}K{r-@JzB7UMl
z!$P)=u}P)X?qafnYbMj4TuUP*%fyeY!dA?SedfeeqVLRSqi1XR^JQ4WVPb>*H4t%%
ze2uKgew{maYOUvR4nm&5kNs#g^q`?@*TV_A9t-Uy55GCiq`8(LRvCGFdPh}={O13R
z{^9*s9{niQ)uCveg?C_Tv<F_pavFWN$SAV^{qZ=LvAj8hGT&Y9M?H=^2~@jPlg&5K
zOlIC7m*fMd^LBLC2(s0kj_L}L0~5DE;dJ}0vC^o6Jrd>p@>}V@5W$+iH8L+BvtWn*
z#A;-Wo>Q%9qpQ8<q-Y*vZ_Ke|iO*j1O7Z%AiU&P&vzY7{rGsC;rk=7q?4_B&TB4%j
zaC1s*Vkwf-iE!@?S|Quhk$GuwbtPi=xrk0DmT9x->6SGE0<sLwz$@%S*(W=VBcb)G
zbY1eJ_<t3aXM9m<H@6SIx{2hcQyYIdj5nGlX~e;P^R-HT>hup6z%BQK1WTI}H7tG<
z1ZCJDOIn&A%?`~e(QJ?BEE9J3J{0XD#M_o2GHCTUuHd8dxA$~jkm;>U(Ie<lb&P+D
znQzDU^55z5zK1hI{8npYMzhE>o}BIvTF-6^IX*w*y03caE}AnnF*-B$F2B)dh3KpE
zy$sPEb*9I{Zu8!9B?WfNH5q_3okjW^^pf^UVEEmmFp_Vkc5Js0dMSLAnKj?PbsAmd
zNGSO8pIy*foU@b0k!~8BnU%n0S(>_9=gtvSq@orboA=Mmf$fmE+d{Z0_05ConGwcf
z_D5c&a~V=eBWZqqzg`;`g$JglQV}xwUdrT_`dXBA7i$%@-sfzf{ax^*CLGV;WkP?=
z7%7gN$JBVgF+4})iI2g4$oXqS6~4=}*P8@&UOm1N*?zyvS5-2cQ1~)9?BcsGUwSPj
zIlZh&dT#BRNsIYFr>(j6a%?`DtPSs+NjYWIp+7%M&)MqJKR+x+o5U~M=FRFJ^j@aJ
zPUk<;3tE07zv|@AD0qE5zJ0cnbUh_??78!=+Gh535?bG)7{Dx4ui%J_Ec_*O7h*&g
z&oBMtCz)(`{`Gl{F?SbdzA#y$A*^<SzNIR+J6<^TPpL_ZHvjq?_N2@vcDx7wv%0tC
zZvG)}dQxDc`sS2=8A6iO=dW-jnm_)DYc(e)Ge%1U*&DC@EA-*4!DQOaoo@sGXg9fw
z7*o%U!@Ie+;=;7p{(A(49s370zhttSjfs{z1VS>tgMFGgJ;Bo<EDHZ##vhLb0ovG@
z#tzvvjkG5pgoQr-uoMpO`^G~Ww)NkYADL8ARF+;wuO8R$Gb+F3N<MWv36O0Y79~lW
zNa8;@2x@FC>PblLzVOrzs<<`T>aT{G+7PZce%ZNK<v~Mz?mll$AAPj0)`k1Cv>-0l
z;8v1n<MNx^!HTntA3e-fe>Oz+v;2Z#v)U=P$63li>lo@m>8o3AQLLERE=|phN7!vp
zu57o^-+28?^qRYm7tvr}uGXc^H>$P$TjW&tU+h|ZsJ%1rbj8z;xtGhPf~tO_==cqL
zs)o9h)@;hWPQs~-{lETJx9w|!ah)^Xo;Oqs@WyI??@4Xlmr$i~VPl`w`*N6rKXso7
zP^B~okY?PZsHn8GEf+VPqjXyS0iiY%ZvL-5xd2ddmQngdZVjO&kdpS!;W_><wApfK
j{QpHeuSVwo7n$;_m%TSYZshYC0WU>aRhd$0lh^+Tyd2Ys

literal 24995
zcmZU5by!qg_x8{YhX$z`x)DV}8io=OknZjVC8YaE3Md^CB1lMwgf*f_36ewUP)ZL<
zDD@qDpXYsl*Y*9uTyyq0XYaMwy6=0fv(FwL=-<|$AiY8ggTW}YG?8~;FkBD@!)YYO
zgId^3oF-u~Jea<Yks5S}gM)*Mi;IVchmVg>KtMo9NJvCPL`+OfLPA1HN=imXMovyn
zK|w)DNl8UT1pojwH8l+l4J|D#9UUD#Jv{>h10y3N6B83NGxL=zS6EnBSXo)w*x1<F
z**Q2k;BYtsf#Bri<l^Gu=H}+%;o;@w<>TYy=jRs?5D*j;6cQ2=78Vu}5fK#?6%!M?
zdiAQfxcIee*REf`E+HXt<Hik1Nl7UwDQRhG85x<IH*d<y%F4;f$;-<tC@3f@Dk>={
zDJv_hsHmu_sv?m{H8nMLb#)C54NXl=EiJ8Ew{B@`YwPIf=<4d;zI|IyPfuT8-@w4&
z&Ye4khK5E)M#jd*CMG7Prlw|QX6EMR78VwkmX=mlR@T<mHa0f4wzhV5cJ}u64h{~E
zj*d=FPR`EGE-o&vuC8uwZtm{x9v&W^o}OM_Uf$l`K0ZFazP^5be*XUc0RaJlfq{4L
z-o1D4UQkd_aBy%)NJwaC=>7Zm!@|PC!^0yYA|fLrqoSgsqoZSDVq#-s<Kp7t<Kq(&
z5)u;=lai7^5Pb09L2`0(N=iy<YHC_qT6%hVMn=ZNhYvF|GqbX?va_=vJ$m%`@#CDF
zoZQ^pyu7^p{QQD~g2KYWqN1YW;^LB$lG4)BCr_S~m6es3mseC&R903#efqSjs_NOZ
zXVul!H8nNQpFgjyt*xu8tFNzbXlQux;zeU)V^dR8b8~Y`OG|5OYg=1edwY9FM@MI8
zXIEF(%a<>^ySsaOdU|_%`}+D`y?XWf_3Jlp-n@PL_T9U8C=}}b`}h6*{U1Jj`1tYT
zz`($#PoD+{2Zx4+hKGkoMn*<QM?Zi5JT^8qK0f~C%a@6XiOI>ysi~>y>FJr7nc3M{
zG#WiOH#a{&zp${dxVX5qwDk4s*KgmxVKA8G<>i%?mDSbNwY9bN_4V)HzyJ90V`F1u
zb8~ZRYioOZduL~7cXxMhZx4&b?(gsa{Q2|OuU`iT2Zx7;M@L7;$H%{a|Nism&&kQj
z>FMd&+1dH|`NhQrrqMbA;sA-Crg<O?Mn-e_hcmtqc?yHUVOmIKqmZKQ1)?Il8*{L|
z^&RQ^yP6s|amSTk#&W6^B>xg*Q@+1TS{@zC&#8u==|J#4Hkm-^4hlDU#>Qyn{yOE(
zzV5Z`A)TyZ<{amvbg@yr3@=I9{NmrfReaG#)gK9Q%1UtP1^xIz!DhApeS{?Bg_;j_
zKSHcM)mA7hTCJI1l@Q<kL@zF;drc^7z?nzeMa^WJN6V(F(qf?Q<8$3iBO!Br=4SIu
z;cBCSI!%Rwe8GJB**}@X$(3k2-$|QHqfHL_7l3oEkij*^<~9zV-{n~WxiqLA@&QK8
zheOFu0p<o-=9!PQwI4wT=Kb1JaZUlI<vcB%Z}PLa^XVTPm<`Z8Gzyb38Or{qlQVSd
zxzia+(?#u$w|Y@tuv&9umQ@WuNLn82MnzzteE0Ih2-q%z3xbFG06mgm>|OvVCXGN`
zHKX+rIRcZ7>OqPNaS1LMc0CFa7rSA6r{2kgJ6k&=Ybd$Ws<{&VNJ3m}#Ih~-(gK0p
zJb3<NZ3mlFClXEGTn4ThllDmNP_3*3TCRM0t+tvbhee)&*js*20@Reb%iwF|pnA1#
z3tFN!0vUE`lsp~s(x~F1*lKvdrA?&eo-$BNYZBD(GATAM4$*Hdq4*sBGRF|WqrhUq
z<qfKjskSQqUMKz;h~iKDoI6MTi9R$n!bth9$TK+M70i(uwZ#!k<=Zz>PD%oCBOt{M
z*qwyG?fBnqdh}g>t2H*n32zJz4G6~Ciej1EKqAuM@Xdi8^sPkS84Y^mbO5Qt6U3rn
z=kTUyx6?W9-i?#$*zXp%mC3K7CUz1mc3EeB*IK~B{=L=?v9miS((_gIGuH1$H2uGW
zH3GJMy)3VhZYMs@1O`+D;u3&t5CIca{bF4;e4`AWh>vNoNQ3pO{Onz&N6OQo9x8>C
zv@%iX2ImljJo<N1G#K0UU<a_`$w2!!X*oNo12KH0NE@CW-CuZ15WLS1BBG_e4=Kt0
znIzEqI@?^A9phq#7{yTt{go!p=V}wG%H%kxnlBxWU>wY4-4r;`2UKzVuf&9qDcS4)
zJ%VAs)rdv2chMoa=um)C(f*z!o<rW@1=qh*c>-wD;av~7hCKOW00c6j24GTzs0@=i
z8$nQp9HS~lj3P@R2Y=*hR&ra@aEclXX#*{yYFhi;t63OS&m%gmwt*7f+_g4zh*n7@
z+RS&dN*!X$9Z_Wy@8@}p5Ml;a?l^R&l1PHuGa5Xb!s5C^DLF&Qb!anjF=G{;#h7GA
zjpBTvs2gzJ_t{`d8Y3sKdKRC{dyUH>9C$+g{j)MYx$rbj_W!Z<F8bi%Or~ez-TJ|!
zqAOKg5+sBA%+7XtuQ8Yto#e3`q3?gp;C>r8V!hDwF4)d|Cpk>(o2vV>{prs{wpJGi
zoepl4CPpbA)hOenzp2N3Oo+{U2T$)%wi;3R4<6}}NBWm;@Fy(?E(4No3?%%AQf>1T
zr2eh3{P7Tq1;K_@2<2l?%-{$Gx46##0+j~CPhdd&mygzi3r_$IK~UvyAXSv<R4$|S
z?Xu(YOD5!WD<dckFS}QT`Bzh5?)-z=|B&PzoYLT*bB+HS)JGqTqeCJtL&;hVCxrk3
z!@w>F6kh>y8#Dgeg;G-4qWWfkNkj;Db;ig`<}u-dRIjhA(O*&&5j{gyB2ooML+~L%
zttT((1BTgJ+`Qam)GT6a?;n^C7-Y$5b;Uyw5;wS%3S2Ai*h(~$?<DW7=ek*sQnHO$
zO1an6!s;Xw2fyQO-UA7+rJ-9DS@(DcZly!P&Gl&0%o=XOdwkayNYfW_p!B5Pjaxra
ziFawlP>ge4fkZ<ZEOb>;7`AgJ_ZR<{g50Pu`$3)8ABK&^H3+Ex+8kj&r^EkLxraH>
zsL+KKiXMC<mK0`x<4LOr^(EM0O;kNsEj=mbCTkL81VQOb$81r|xpr8?f~V^7Ch0E8
z2Q?AMuBFQ31dn`2yP9;yYW|HxnBDrkKVP62@|E?mKlo20``-9GcIh$CkCYA(2M%Mv
zdQbq2iJxH$8ZADymN|RRJ;{mD2jZy2Yq+<H6@NU!!Ir80OQmSw0zc$ns44KqFH=%Z
z5(4pKAPOf${yIJL&pLS4<y~01*XKIzk#tBJQinYRJ3)xH@(N^&0*VVpM2{4uLw!{}
z1-{A?WRhLpOohvO2k82H-@WJzy8oP?$J=X)g>cIh4$8t5ZXj;kp#Fg_9x;|6dh~BM
z50y^u{L{1do?X^I*IU{v{_Rf{2OCX*d6e)!yE|T6K37|T;GlpS-fNHj2>Cu79DOHi
z0?Gt^VePj2!8^Al0LqYJk`lauz#Rq54%hdlmML-qaSi2pvb`M&9C1nc$g=HP=pqVC
zm=r$puKOZq#Qi1%!2laCoEg_p85fIc7nEwHwDQE2rd&ltjj;T+@#c6<9qON1@9XNS
zb@W!pq5SykuDF=;-<+$ib*UmZ46X?=JJ_7^=w)}aia<Fw_EwfrnE6o3V__VvHs+@G
z+-@97XKSseV$Xz~I3E9>M0p=dl;%*PEUK1Po)01QKA9*elN^@u_|il4^&N}Dm($%X
zf?BVZa`>7A$YBmw=!uSyNV??ckUijOqYwPPL>CsP624ysCuGC;e1GwO>2{fK|MHvY
zUw(fi_1_5pH}7bIamFc-QZNO8&(%<WSqikdGeDufiORLFzT`7gEQ9}lzvTUG!3E#Z
z|HJLF%fbF;&Mp=@j3x^TqSt@YqaP6bhF=Kf{x=Qc0~|*Gzdk`+cf1sdziC(`I8KGt
zGeJ5+^k15>qH^tu>eY+@|G&ay({`y)kR*M#*da`iR{lq)9{eMKw_lFIt9Y|Dbu<|`
zC2fXMXj`%>xI?o7NSYjM9Lg{2A^c7>zE+#~;&a_cMupr0jC?hc3R*)chi2ZX+p8u_
zwZ?-}qbtn+M;^l3xU-xBx0HAmSdd&sDS7WKP)R<m3}cXrL2x3;_6FAfl`>d6NPlRF
z7}38p_Kh#~uKNGv(-5k3{uUg6KQ2q5F*ZzkgXZ4}$e?Eb@fh}{a#0N&BJE{e1Qi9<
zm%3*JEL?v4(*dLrUS!z!K7Vh$Y@H6c5QC5W%l~N#jPRu%{a@XO+y#k$dTeK}?4AEc
z;GfQWA_3Rtf3$d$(7!zv{LfgI#Ya$L-u`z?Tl)_IC>Qa=`>+h?dmjrRLF5O;eknnz
z_j;4t#RrI42Fm2a3YamOKd#xdTHGi;QfgYhwfD*95B_TNF_TtForTuo4I>u5TFE!N
z9<)xApBNr_nQ%|4L$ZuM3|L-}&M=m}M|x5Xm3rn6#-MVuzQs*kPhHJKq<Ef30~~#E
zBl2eMIF#3)l2GTu*WRa?yZ`OGQd(S0#{8cW@xaWxyb@hWCoVQ+*(Pw=i*PmmY@O!w
zv`k?f^EUDJ+|l57lf2`(UpQPY`R4#7eTb(}RUqzAl6ot=oO|^6sF@bUW!XjE<~}k|
zxF&ObMk8>h>i4Wb;CiIX@MT=?u|t{kZ!s%Ar!pTz^{EDs%DAN<$l!8G0_B*_uMHLg
z-i*ml-7hn^K78udM*O(oe-BL_Y{z3f!OOEUS1@-E$F?Z34_u@a{ReM0Nl*}o^QjgU
z2Exm!80(Hrgojk1xX0mQMQ9w{uRwz`(Sh$0DK?E8Tg(a)WAiHDKHCoh_>ruPwg2@0
z8SY}jsQm@96J_UJkaUAZFsJITS+ZuotuDJ88Uvm+Ww`i)M^_M+r`DXnp8=#Os#M@V
zcRlecxIS7W_gk0cuQd?F$H`Xp!3qzC?`FRcmu*n2Nm0$Yu0_)Ihtxmn_`$MRv)}t<
zh5D-hJ)89M%*I!p@xRBVx6Ac%&PD%@=LQ<iYP86rt_(||K<D8-O6=ojzk9wah5w|D
zKfG{u`5m9v?(+2{N7xn+%Ni>fxtbLKYAKd9RXkSKneuJy-p}CKEr3_0&m0L|X>_$7
z5-3<D_wOS^Em4-eV&wc)n7@vX<@qVE0W!M_?+P~D;3&0zHd+X|ZxJCHrGMgjo|3c5
zm(ckg&l|V)qQz+d;UX`aY3Zy0Pq(BAe10Z+eUJ~8fiuNj;%cq{caf}*H>8;TvOoi_
zXOArgt-3gurIW|gVYbkWZBMVdsCjBH-cUN-d>qBuExeQ)_O`J`*RA9N7>&yMV>!C`
zec=kf_}e;LQgIVoqpsnEbr^Q~vCwS7``VX35I?^hW?q^FKbI0=7GbF*>12^C?79YW
zBzh}}wUM5j?gtBN<6=?Yq+FCsY`60Rz$H4{o;KHtVwW6z2TaiE>NK82*j7-EKun1L
zk86*}>HV77H<NG0=M`<fm}8v8mEOC(pl4PET|3+eD;q5{Ty_f7(uD7TPdzVEILwtv
zP<&*O*_$Ii#hl4hT`v@de{x`Q*lWa8^o;rhUvdob>X7<>4%puPZD9Wa<DNr@;l10D
zbLP^bMr*ccY~{&r+&4I72F2JHcr|>1V^L5mr6OXVnJ*M44NK%zx+lHw%rs}C1s7%S
zcutEfX?MK^w~>z7cvE*};rT1KH7$4~Tf6m1Hzlm9)h?d~Br0RO&*<}Z;AWD06){$r
zu3G$?OSG2bMxOjLEqKmHB1h_*<OCbRtRJuO0E6rmp05Z`m9Fyurtb>vZ;;-DF@^7n
zNQ@F--bwJN%!hv%i2ck){~Q#Ic2_0EHY9wX+p*1!3!7;%y9R3Qxi^FteopQu`%oBK
zf>`XD^=BO8Lixg{IWj+U)82>)b!i}Xpo-?W0hIoVFLlqs$AC>4tqEWN`lhUx%Pn}r
z@RAg6zsiF>#vwTp6j!6g$;l&=-O4p#!}z=|67&--eV>IK(WprW`V`<u1$of)6sJF}
z+Iyq~-@zu@fVh`PzgrD|@R5@6+4f^p%Gj_3!S85qek6FYfd9h7_M2ohCy$(p)mIqy
z_k1SI+Q3Q1h}}$^FD~Ty54N9U-IJ7zG2L@x(8y_HSFg0%VpkH$h7(w~O4$2w^J+yZ
zehN_7H9$eAJ}b%a<T=7?To58bnK&{?P2cKG(4X;u;BwqK)%ljqLQ8wefCl7HjoGEP
zsC-=bIn_BCP{eaf<5kMPbA7O+Iy7%B8L%a+mb_Is^ZiZRM@>C$ZXghkTw^X6$Eb#(
z9}0x_V%X|~`zfUGyOdwKhv^WJ_>H=?#)X2>dLY_lfD5G?a&uTrRqHNoUW!h<bg<uV
z%D~TajW?y}{9^Mi!kEbQ8FYJ`m8A#CB~fB9Qg_?U7viv%VJi|8^<x;NJ5O7w<fg3(
zZqxhkzbW<tZjImu@#+xz^E|?*87&N@0WH#s`_!!40+wfN`tUJC7J+g7$~7OE-)9&r
z>IABRvN(cN<GhmlF;{AZKI@R6tkCLo#v(9)at)KCB+MS_NAT5l?L|7!HakeBejM95
zD~nqH7$2EXUR~t;!6sgsG`TfD(o3;!;#S8~2%w>h@h7kVZ)V$Ig+_V~FHoaZ=JVRm
zQ3SgG<GiFc+~iBL@z^bo!T~M`J=~_*xE2r0dpeE;DIawZcqOD_Iuz(qg^y7letT^~
zOGv6pe2@;&NA;N|Eym$${pS%*jxj50Oi(nbJsOpGSbrW8S|AWp{hK9PaPxYe)^#;-
z@SWX4g+-=mYq1^?Xi6EweBjn_0Z3($6vxI^3+Ii}rT`i$PPMm<O<!|~0)da?b}TYC
z96g8eBr{=3L!Y@scvQh-ZUv1wkFPgcdYP3@_JySct&iPSuSO`R6NHqmJw=hdzmk@B
zhm~>e3qY7RN!qecX(>*Xm;b}y?TXFxE%Jft`ZJMWp+TLs+OwnR_L@&jH8*6lfBq78
z6-$31>7FCWX7wYwVU?rz!J19-nrE+u(D==YScSM62`TXyv1e!ZOU+`}MV9%ZecgR@
zp7&17EiASx)oHhUaqxOtvRQMc<+|y*=wu@DY9vTenfDj@$}m5kr&-VtPvqf!6|*lv
zp)dxIr*of&Cd~TG)I0mKjqXhhx5S7v<HsaR3iZD-ZNNiO9lWn#(^7d{Xx0Gp|6uw!
zm7r#Os!unN2VgOIp^Rz(qZeJ;PV02sKA7mpTsOtV^3<cgrfOO>;tbNkO{@q$3p;im
zX5Z(R8^S4Xk0MLJw(g8l49;}WBQT|RCQ}Q4G3!)yaiTn6LfCW9^JlGTy9Nb=s`Cpw
z_0cNAQaL(qILOM}L_V{2VgpNO3QJF(^~|Nm)SM`{ZSNk1=c(1#$jqE^qTS*<bbs1M
zV85m{L>|V)H@fc}-Boi`R?tjG{(9(Dz{2$YZGASa<S^bay<rI1AidPEFUGp|4S(#Z
ziCIIVyQnPD6M-u+gB}+rixM199(!~n2iYWt)vz>L{LgZoe~Nk)th;G&WC$Y?67h<j
z%$`<-`pUMPO~c2%UP%o*-ve84Ft^@Z$r14>Xyfhm5vpHdo#!0a@&8iC_CuL$`{L6(
zp<zNvyf=#)5*8@AN`HqgT-3=lGm9V-Z13hUJ)+-4oH0}&M@VVQ*Zlh05j7&6?^OoJ
zLBN(y0<7uR)o5O~Jn>#C_jZ`SVtupJYo$Y3Y6OODrtx{gL9<sT34j%!$eOMEyg&&Y
zwaR=aLJ3IlC<)6K=(v%-Ud*y9DF6^2uPQgtqKrd+@p;azYWnVO1odi_vu_fQTb-F7
z+e7Arm0puHdUDpCyG9Ba=2-HY#yXP$yB#V{8hD(@vyO}BHancd^zfYVbfGI1hz4Nk
ziLu9Um@5&OP~Y+)rw`WDa=7!BxAHR^B@o=vzRfl6<qDETs0!am*3mr4loZ6VHg6|x
zY7K%v_s;)nuK{irFa?^FY0u`ro^pGL=qs9TPQBIQ#f~}8(!0VU7@m4YGt2-V`RJQ(
zWsoDH)HD6EYBu2tHHPVq-BX)%YcYB_=DKo6xpD(w+c9A=i{qfYxwZWSng=DiJMh49
z^APsIU6KmBI!9^7yH1Ar8Szj9<WYY};yzeq?_d3{Sq<zbxt522EKy61i*i;Q%4Z?j
zT7ausIa;r*wLePgWc3kp%$d90LI76OXA7a43BZWDRkIEOP85+A&a(7_1$Ry*;j-R2
zfWj}~xS-8~2qdUmoq0WgiwuYR7544Uqj4||IVU37jsd5MX5o1;gd2=;7Q&YCW<%A@
zO<bt=Ir-g0Gm|gjs^Rl5KWaBaGr#lE7%lwab~!0PWX_jc3b{Ybe6Y!|hZ7mWWn>~2
zoCrk*Yv2<TboWEH{4Bpx_@sr%92Wz^V(y-usuBc(81vr!N_}l2P^>NF`V%09froZn
zdhrgON3$1o`8}$4QmEfKcZ`RGZYU>qj=%gkTzr$gJjl)LZ6}f-C-5^a(o#D&=`!f{
z3Huf%X~YIg#E(CxGz+ERBAZ>F9|a~HCu@Medz4Y#H|`x+=Tt#~HnC*Y2NPesu~=1m
zy%ne)EiIk<6g}q#DMk9z30=P!7b0d1*OZz~k@HW0F_SE&wI(XZg$(G1=9j{jKn!#j
z-iz(JM~$H5b57+Mooa<;Zmk-1kJ4joGJu~UabU{iRIs@ZfwX!{C@5QJNCrcx{E2Us
z&%BvXc4d$Op3`iy$%4_*uDP~G8TJq%+BQ-bdiRa0l~RGKBQ443XD=KF#<^>*QtBi?
zck?y$xPzhxHAeqV|ID3c?Ri`zXPe;UHVcLwDz->lavY`fAs;f7S4Xx6r{JL&y~&<R
z4j;0VHxY$BldM;<3nJoEYLx^jfoEu!3^r;)W(?=14xRn194d4r;{NSId2_hZme<Lh
zt&LjP7=Eqy>`aD5e06E&gNc+#tv~`WLHH?b#_kO+YGJBNNE>Qscc28ghE>|w;67AW
zk;}p61W#O)tB0CHLGyNx2-PAlp${eV#Mh9UC4zJ=&@f#qXSb;D%LJ8KN6(BMKAmYd
ziVoouGQ=5j7Etk0`@}>s;Zz``)XKJcQ;Btz0wbCMi0Iy?AOh!3V?XxQ!^WhSt8S2?
z|5T%(5t!ROESn(A;aheox&`?ms@hsd;zRoqlB#VLs=Zk2<+LE=Z+qD$AMA-Jd6g}N
zCqAnYF(?pHY1Q?SDD-46-{zmjDff0<Wx<TpCYep{dxwIS-r0^}C{~niL?%6ka|uY3
zH#_5hurS1{q=O~)udAPzDPXe#qh?&n;e+dl4cS)Xv<j`D2*c<Z^tHCeJzuwWSd;W@
z-Z$;-N!W~<EsjH#_aS@bEG}}+z$ecaq+W#Y*sT-!HLgFv2Xh7<2Ss=Cyau?;AKRIS
z;G*y|0PjK)Y3-IcGHmdiP9EQR2>4~-WHMEVa_dhOtSQ0SBlLNcE;K8GkNz&A+fO>H
z)Z$^-W(}~vU(o7Bu6(E<<huAqvI1Pxw*CYy%5QG#cYLSJYalJr>=oQ$k>j;0N@_BX
zMQ#aF+mbH3nE~`ymDAz&)cvM#c8tXS`LNor$unt4v?Om-#lJFJEZF+45?$TKQ^7g7
zIvtsMPuP@!06bh^g&dP+tPmi`Q|b%05ByH-Oe^W)UP4JB3q5~6DM>mGk4B(RTWp(P
z*x17zygDoG0uiMx*j5ixy6?j@_*F&D)lxmwT!9K}zaL2vFzERrjJ;wO7iF0y@NHPV
z4_&8d$&R5y&t)g4cDs|y?7kVPAOpl^T{0H1B1#hNm1-6k=Am8M@(Ar+77Q!=*;h^3
z!qISya<p4vkuO-NGhxxFjfhlj{Vlai^%a2$%OMPl29Gw=M4}Jl7Bw&3YdTjhA(pTQ
z9H^caIGvF*^~OS!-efU+>&i+t;jH)aCxBHoqhjPx43(QzVZ5}>NYb;54RM8~_8K16
zrRt#`Xqx)qMsSR@ilQT$RMEt6ND6uKn|cW-t>EFr!=*N@-7UsYM2AMfsHm5|O1;HH
zwWGV`4^#+L2(fjm&+bdDxcrtwX;qquG)EGqS>J+!dN1uo+AgzD2w7NSaZ-p*SlIo6
z2UMQf-VQPm0!lhAgG~>)kjvF&<4@R+r_b1j*AU|(y?yu1EIE}}DW&MguOxNv;TU&$
zCUs`!eB=PYPHPTSPIIVvP7=uRNZgmlPA`EBm~-3RArXq5d&rqn>+&Yym@I$bWUuG-
zjAqLmF*^o%r-8ETwPMKWnS^Du*@t5^{S9;i3u_fbT;Dd%mQOkbHxpWKZ`oN#=b>`A
z`YPPHA*fsl)4h@OYa%G{)^U8EkB-+PGE|v9*2Q}r6Iq>9T4rZ0x(94q-nUyP4gp8U
zY1E3~3Q65BTa}!w8~1@Wu1>$RoA`aEvpvFts^NNJ!Er?3vCfCuF3$s@TR8zB{9MPS
z**5Gn_?FgB^8j|vF2gGuStm`2I(EHrq<|;0ZFh6~P|EzLoevlU`(*Nv$kFFk_mC4J
z0*HxYa(D(~)a6an6Lw}|-X9uT^+5N2lzO)kEAC@YkHTK&ZxlF$l=~!D(wVn>DA$=6
z0zGchjaa9%gzC1<16!_jvAz0-YtH8Em^VGY-?*jj3arAglJ&1Q2{wMcP&%wu!N-#4
zkDhlVxyzD{N=lSAtRXh|<Tn>mf+rJo_bHkZn?BZ##ACj&G^O;(p<c9ck_qaRh|@k1
ziEnCA$3;oMB-QZvAYre7825H+p+0i2zugt=u0WN!6ObMlBhB=DQk3Z7q}iM=llT%J
zGCV+ld`6A46fkd@Xtx=)Y+x%7!j}?X<v0Ii*L&%^TH~;*Bdo<WKVa0Qm|5G(o!6f#
zA2=UT$KzsNk5Cu1UN>-gMkt5urKrj9RY>}l9iduB`G<ODIP5E{$J_KPA)u|h8iQ%U
zwLnTGB5;pFoN&58<K$p;0$rt93S3Wf^+Bh<)bFG-+9QDs?9)8FH|x^}^9PJx%rU<;
z$Yy{!4&8PO0Yh?I@P_Usd@SjodD*A8>@ZDka3Z67>pfpI*Ujhy4<;)As7=bsC=zea
zEF&^ATTZCf*g4vk@g_9pX;#I@N|-Cfy<8a&!i`E7e>S8#ar7LG2rrVveCxYJB*n+#
zclt+cozVs2)N&}XHcq{Xh>)r9(?*R-UvTU=v^(~FE88b^bj(lUf??GO4{j=|!{YQ?
zIZ2-s5WLJ9g^G9m5~Tr;1qeq5hTca%$_Nw?80(Rd#`tTF&vrw>j4_(I(!LkY_kvzz
zz5;05@^7747Fsd)B>CD<EF=2H4ufW5GA2B44ApJZVlL9Z1`?w=Spu$iIj~;={r~**
zuXw3`9!7Rb<9R45)<*pAUI3{#%BLENx5DD*(-b`ld_iN_ZcJN3lmi^6O<EHw3T)lp
zQ&!zBY$4_-ShHUr3k4ky@mPP{F`G*$S!l&wIUt}e_2$kXVO6i*a;sh=!(gMHjCxq@
zJezyRilB`CMv-l~syy5lv7SEybblr4^hX{iek>`S!Op)H2)4S8;*TDv8E7L@o^65k
z4%OPrp{CIp3D(rk_CC)B(8`r!VBsH03a;Gsq)EbXzxJ3uY9o+sO67-zzG$C@k9Y2Y
zSI_JcK2cM5WVtBq(z3bJPx2)|C|kZ!_N~60(-V`e->&Q!x_z$s6ZE)!+e;ONQlQYY
zc5m<`*?oAgXpI(8Dh6WF6Ccx3O5sM4R1@Jzz4l2Vb6(#yf(mX2KPMmTKTe$CSoC-l
zU~4#m3Gz0bl0x<@Ulq7m4tdpeb>)yE)9d!$;?IUweu?D*gh$%9M-#$#86U|$W7A)?
z?_JslY=3mJ9<B%wo!r#ZxEFKJ{=IT|gUi|CioqT=THRHz3sCOa7Og&qNg#Xo-H5QS
zcwr}ZSdq&%iNfK}Y{-5YaTJItO;Gx9@x3la8)O`^c=ndM1OnIy1Ob+ECEfUdJr+Wb
z<aUksUg9rnGu*ZI0`lKUU&1Ll`*PGw4%g9Q-;Ar5$;yMiV_!4yo5M!#mJq*q^|s_Q
zHC*8{wzo>)j2&6uikG>%hSx8LoX`Fcv!=K9CFi!Ox5f}3D$}BZD%ew&sk@Ew`U`r@
zdtV&(DR@v!VI)}+!5T%p9I7{1jwbSIRA|U@gG_aa9>`L5s<cnbJ}F2Zxyiu^*0-{O
z%zi=}b4&}gK?>9}G%N6p^&hPZnZLRkbzTLg$HSU>f{L!uzx=))gKmy|ECVJzff-Fv
z)s!9+lsKC*go3&^S40Bu{_@&33tuFgD|;daa=86G3{m@NYGCO&xfpL5u3Sg;If~_%
z51FEvE!Bzb0Y%diSgB`7c^d80P4#skK>;yB_`Mh%q*|r{`}3w{kA8R^em(~Xn#TE{
z%V(eU7QDY%_Z9BbLuHN-EkPQxqL|S-x|8srg#9LO`z~LwTUPPr_q$`sdp1W2-x-KN
z=PCM(Q;oS8x8h**s&ZMJnXjaT%FzqeIR?*@eXsZaz=L-_okv@)1t=Ni9um7%U)t_9
zPBC;7J3<;<Xye+nP|v6z*n@mUQMe`+=Jkc$``tB;vll`3wt#;UyP{}PA&dJDmr@l~
zTfo!3J4;NdG-xj=Xy$xMN&%r@?(j(AxzgMjbtAM6osYJ+L9-PmH;72Di-e5vp*q@Z
z!w%GsjHkms=0(&496O|qx8i2x>u^66^x~XX`tAX?8_f<O`#+40m)RE*ckV`;VJa+1
zJvLpv3O6pwr#G6qcjG9U-pIKn_h)LXWe@w<YyktIFNw#Kb_3<p6i76Qfk9kur@HU;
zzFrY3Tcq#5Q90IEE!`bqL~$T`?7FhixxQwuI4ZAnAz3)8@Vn2l#jVGSQlSk>MJZ3D
zSMo;O{nEQF^^G>4DC&@;36VO5DA63DD9!7*^%Ho8M}E6I%<{oKiZPD8=XxL!eE;AI
zhurz+haG$C>sCT2feu*s3*~T0nkD_7_fjav?uOE$RMG5n@!RVjmlcW{rIc`8ed)U9
zhkIwLgiT|P_N|skC^_Kxh_(C4-;Qj4OX6@8u76Wxd7=Ift@O5S#cx)WsQ$H~(>oW9
zVItDtht{}g!Dnp&+NPLyC6n*Iv#=uPr>F8C4g3+dHJ^RUh^+5gj9rXy_#qG<DHy!N
z81y(NDdhT4<#OFtv0rGW9%zV@SQ)Xe_t^KS{FlA!&)>0~9Es%?Tby4KJrj-rpNkuy
zE63a5x{Lgzg}G7Xd5gNRDd+Zww)+HFbMW}MEFJwvM!Cd<oEQ~&D1(JH@hG(%zxZKT
zD&WF+5CrD76_=scth}%!P5iIT1B>WOf!)|^-`L?EVXMxf1*_pID7O2j4wT2swB<p+
z&Nz#$7@#aqD(B0*_dyCpjec_<P4DzbVz+MvE}D?(rpQ9K!XIWdqz4|KXMb6XR!j(2
zW+gu|Tj^P6;S8ra;ILH0oBHx-322FA&vhBxBW!vqa}V-nYcKXC1K8`9oMgrXU&KR=
z=!=fCpdNh0L(Z3dFxLaP9EYHi+l*S$64Ct5loCS{-DtliEu?^0dm}~D;_-r+v+rUh
zl?cRp0zJqcnSU7NM2!k_=@GyjHiRlDH2(Xx%r0z}^;S<NgP!esC01Sa*<>D=PUoO-
zM-!1Q&->X=PwC@}-3JDK8Fw$#S)A0(Q&I%wXP<CM={`l*p5cBilW3WLt}OA`?M0B-
zO8PO6Kte>#w7cMxjkmx@*Zg^^_wIr`=yQ=zZ~I$~+MlfHKB<Whv24<V?PusXO->!h
zS2<*PuLiuc&~7U=b)M{>Nv_|mIMdRt6kE@DRugh^lZ7lHFsd}~mMw#Vs^OSGd%%&3
zo_3q>jq)0&u(nN@LbLvljzr_aJYM|bd1pS5cv$LxdvE#=o0*Nq_rKSY`Ni;7c)dTz
zGjkQ_C==3skLF`6<+$Dd;r*DYcTl6pcFQ>B6${>@K%c|JUUUXdmDwkG^pZxKm~slo
zvZHz1@Zq28cDj`got+Tzg5(xc7TYV5W}k-H2YFpz99l=`N99dUg_Ks6Z%?YfrOg?L
zu*x@P@fK>Y3hudAJj}klY?&JGp_X1~o^ZR#W7%u<{^b;2ENxVvCt<th&wG6zXLuaQ
zEiXwxJ52T(-F;$YN$s}1pU?6=jrdZOe6&9tU>P<}_sV^Q=#^zZW^+$a*S~Nr9Ub#h
zrYmaptaHBVH~-f%TeGeks_w~i60Zp5>*G|~!waiAbpzc`fvbh{=0SWM9qxTrOzokw
z5>cdnv)Aw~>c!lki#0Ht9d&45^J#LZ7xVDkXI<_#*D|3yf0|Xxhz-n3FrF7*_T$^R
zG+6bUIa*A(a4lvl|G-n)3`^v2Lkuufvb<pfUrSIF0c4wr{H4KdthtC}7ui-Mv|#O^
z-PNY*mK=9X8i|?3L)|$xkPHT2!Fuds+;-~x0jg!sk`lNNIX6oD7%N2UKs{y7L4;)w
zd3|RaA;}v@jIG^*VO<6%zC=ej%c7imJca`1cFF|U9ej=l$xuPZG(CLaV6W1AQj*Yh
zL_uyR0a!1Reezgd@SU<;vTXD(7#1&l_zNv+9hU^FJfD_L1TNS;(Ex3Uu5dDzz)g;D
z#Ysw!_F&>%j73MdO%#2)U)V5Gu#C`{PR9PQ_?dn(RK{M%0SqfYf6_rR7VeT6mq(W2
zj(RR{l2CODy#UCVz(A`qOj`sP!f>vqB>tZbCFqU7VeHx49iXjZt=I<c5pdq7)Dyu`
z7#(3^vF!S3X61ZP&5#oXxooH3HB?U7Yn=dtn73>q8KZln98UCBPGz1|6Xf#$P{oTX
zugd@^9bB0<s4-((u%;^KjohKg!EIn!t7CR+nDKt2-V7oVhoHGf$;)}pJTSI%<j&5(
z8wsfJVGbT$!Ew;~{*@YYTMSr=^~(S{@M$o_ZeIx;?ng5D)EvgCLodSQavb6PI6aX4
zuE!*#RDdPs_V~<-=uM^YDKgadUL!rmo5<}Jv_f;OD{!lcixP7J>+Vqs|H6eLp37Xs
zLK`v2eGYKN^X0tONHG3ns5f<S^>Ed?RRRa$*X&||;+xS=g}|=eJ3BwXawh$XXmb*g
z3}~C}v%G+~{UL)rEH)%b0~E7_i#j{H2LiTn#6qZ>Uhyh8<p-L(A<hLDxWX5z&{%1@
zE9M9WN5|GV1aGFO#XLfR!d@|n-J!+O)3U0>C;iYNiAgun#D29s_?ZAD#$%qXhX@?r
z0}@K$>}`+2JOEvP;2EZ5Sx!kf=7Sg%Zrq@-m}AnFePaJ^4q(_r22_3WKm&9Xo{S6;
z(j@`f)Dgdc-RH{$4s<^XKd@o0oq|<Fpsu$Dg~6792{lIC7jU&i6i6jP%ZvNC*xu*w
z4?|))i8$p|=Bc5*GJBISS_l8Bj3StSkrcSyf`@9j*Q?Y+yjLIvuGRsv+hqcj4vwd*
z3y4TH^}u~x{{{`vS~&Y8O`dv4)e9Z+f)SMUXB1(FC(hlLDYCQf`9tClS0LG>6sNX}
zl>$2~5tQcIhHnx=^yCL8u1D9~$)YacawB&}SWk-MI_ZL0e&9H8&3&t-MePc&PEk|}
z;9$ZtfbRQoGvTEKm}V@5k9nsgw;&lFrEr0V0MXY#+Y<O~$PW4Wd<wYac327{c!5cK
z$Ok@#;OQnCp5`)okOD;Q;bQI0R!AKDzwZ!Wx`!=ZVF)nUoktRSfGs6A>il><m!t_L
z1})YXdDR?NBtl<$WM!3D^_R#{QJP8jpv18R)S4lJ7ImT0kMJ|}_OG-@*~%edCQ-hH
zM``OD1!mj#e#|ldj@~#eY604&M#W(r5-UB)g4!_wRn0pl-Xusj$3PqyE)&=WE_jQt
z6auqg9xICHc)X1W?D%S1CW{i&(s*eD7yU`x^vpTnRf{+sJ{V=&u4JW<UL_5t_z`LH
zqVD&>6K5gMI$2K#0#b6(a>_j%r2&3Pc}jn*4sn}IC8KAY%K5-`tq%+~@Toogrjl*Z
zpq6K})EKdxfK8&$bVUBQo@U4^3lsMCDx+F3l>VZuP0FW_s_;QyA{tCY<7d*Ql~HK=
z^UgJges~ay8{+zAaWJWgHt=(`KSd6!moC6Xp-sTxJ{`yxH|-<jRZb^qQDK=H;1L8(
z%AMkOcqrz*UJdZsX|&r8wI6Wu)^!n{6F$d+iZlV=K?@E%n_W{vYfRW6Fe+;?VF4l9
zA4eS)!-S1y1Z^K?`pYY!^uqFhxjtoRB)!w<<Bw*LA-5Ab<dyNmTL^0pWhJjBmjWkW
zDMjWH{n_b+JxZ*0aUa!FiA7OInT%l6L&^YoC30Vngpgb@u#Kxem_I*4@d0qr_$ch9
zB19;@<%MuPVAP!Vh-mZkupYSlt(Crf*5Eo7wc;B_X|NZo@5hU}<0VJR?htp>s}xSU
z6wVp|rcoipfGECVi{=2adytT0j*%=VYl-)&L)YK3psqbAfn$8KAf6_3s0Sy707vdW
za4>OuEmB~J@Xl-HtpwRwEH1VyN@Oz9suc3KWiuN}$L^<F(lHD3+$0_{oT`Qq44n>*
zo1qVl?IaG3J9f$8065rGAuoRmc~k}MGtJ#0aA0-f^aX4iRanpQb)7Z9_2ikgw|vO`
zqD+#eL2Za0b;hUB&H;Zo-$4)Taz`PaZ7ku+;cnE;DiUEgzadIgW8$UjE}X@H7WkCK
zwt%Edekl;<qh<m=)%)`36=3^G8Aw5|IZRtitcsziTI7`LylZpn1ZW~vxPvEK2F?cl
zSR62NE+*##n^8K3=gQm^k$VV*SpgXqCGLg#Zq*52mqQr!(a1L(&L4Dd*f5zlp(S*c
zqE;pX8qAzXCLg%+s4$m7oDw>epL@s%{-Ml(*2+Xjy$HoO36nJ0ar%6Kh&C|4EEcSr
zj0_-yqBW4Mgh<j<S7}mmO6gYw$>de)u#fKrU}woOmLYIGS`4YA&fUZ48CBWNMkV2&
zw7Rp1{$nD4e3rfIvMLj$YObp`ec4qZV5gooEoMR1{SyJFbu24tbOZgICz#;jSNLtK
zSQVVi`DzS-Lr)ue*SB1Nl2gX^1$@J>4*@SIe&`QqeK7Y5h-$)l(IAErBE?Lqrj`j@
zz(x0nGgJ9Tcyk{P!`VA2LSr>|o=YJLxTpy6%R?U&#?3^?HmeAqaOujZoVKz<GttUW
z;YDTEx(QYTERpd&N)z2QPOCw|RRZjseJFdCt(S3CR`5Hjkt#*e8rr=+fN3-KQy?Ak
zZ0#nr=}iNC&<Ecj%XrL1b)NOn9-yt8Jl9XJqC;Mn2JOAlpc(gRfKpL$Uce7NT)eR@
zii|k3J-yA9^AG)A`?UXw#wtv8>yQ&N>Ie@qYNA6_8rp8&ix+9)qsLc(L11F!bU~-g
z%+5aGgBXOA&Q~{@Rl-x#On1mLg|%sBam&v&3uKYwh*lE+2}cQf7+5pFO+%ZN0H$^H
z_N=mP>#>G0A2RBy%_bKMXL^YoIcOBz0^1%J-XJk8)U*u>Eiv{ief+%CjPHn}v9cF^
z1ce~tc}Io^hgI}Xvo`PGF{9L8H14zd>=PM9T^@Ap!rYycY}6+a7Ij1aLl2G+s}tH%
zr0tZVrr}t-$MQE@&o+(0W)DZN?67i<hMfA+gbK5^!+YvE&-0BB`D-NyEVf4XoIKl-
zJWi^E-ovd#3YYB%!;K2PimhHsf$AY{)B|fMC-xBw?hg$9yKEAxt|?~Q)%OC{q*bj_
zs!IdxRyBRsxlg^Q$xah+XF0PP@OxXq2U8wV!h?&A9pFkPE0qZ6>^E^sRAJw<nyqQh
zE7a1U>#bRck*Lrh(5a&UsgEdFxOP)TZZr|L0`Hq8G)FWzyhEN4iE~K+2kBO3ZGVgf
zuPLSj99CL}(3Uv~C`V2D)@N%ULeMJEjE$4_k@-Xj_&PiO!KZilZ78wIf#TZ>v4IHD
zK$vHk(Wz;e4je7qHpGe5Tv~;11gAP#f=-$=Vn8+o52M7YqRU(#2n;yKUOWTza@&U!
z^uWMvVBkGJjdFrp(BMbqp$kPiT<}L(Om%S7r{|Nq+bHY>F6J&|$YaN;fv3<DbyEgh
z5q>`0Uk=S6Mngjap)k@f6r4Eo9j|~?5JkE)W~2m9w^24!PSEsOlXzB=lE$=7jD4Ig
zK@SJ-O5jtLpRX_rb<~LII#Yp8r_j!7ZmDSkO3vT+WeCdq_T~{IO{&}-epw&cfuF@1
zh?qRDdXo!-<*)E8n<Hp{_B6Ra;XZ!TzW?UR9cXnaPa_l3YQ^d(z}7dZ4k)#|^d@%v
zAi)gF4KaTsF0s+$RkfPo2r$shO??Z@WOVyPjj3gG6I23z=*Xys_fhE*;FrtrKQ9G1
zW}@qWfjBqr1~KIRLmy~F$EJs>+zU>fsuQCT);Ey*J<KE*@C|u0Xpbl-#S6eSEZ*!#
z{JjA8Y0}KC>5%BWBi|JQjOR9Fl8jF4E5NXBa%0`eUs_e7bT%MSGlz#n78y}JT$`wQ
z*r63q<FFIu{)w8?+>kp2yq;+b{K%ptN48X;H_4iou74y$wZc@ntu3u@pu{}KjczAa
zFTK1w350oHYR$%(LQ6nlSc@J$Zn?}&UpsgfWxeJKKe$qoerfTx%?)HTtr~=vK`vyo
z?K|G!9~OwLTAXPL$S?^a?73^E6j(FmaUh%T=Lae34m&(+M-L)<RAD?@`Qq1geK7RA
zOh6Gaq7tcGIj~8d87?0TIx$N70fsFEs23ieA6-WVbUi`_#aAot1wvN&R@Fo+&5x&C
zB21q5A}U%_l=5L`3_Q!5kwgPr8OI}W1tUtH321;2n;@uB(r%NFu@+A=JA65UB2s>S
zN=3OIfugTayngewU!tZD9Py19ySX8;N}f4Cr-ixTK#|)52QlR4UG?(rpP0*M9Ru_c
zl3}@3r*R+KKp{f9V?(_&#dyt;<1D{;&7Ew{H_Amns5!e7JYqDhf0Ja+FGfxR!}jIH
zh=REC0AP0xY~VviFCEYkWp3d@iRy*jOI2<vfkakLX2lW62mTL<z?MvUhb3K!$MW#3
zY}_4S;OQ?V6`dB7+M|LPuB*U+BO^C1Zl+v;Hs%5k8E~Diut?^4;&qhw(PyLd*gT9B
zYUvk5g~-6`x~ur$N}@bXI&_Z@w={Simkg`#XxV0iD6kH;wFQq9Am|ppQZ<Z;bxYM9
zRpvImx`5b_da_Ojwkc}ZSU{Q_Kn1MB^`KC>#>8pf=iB?tnWg;$GKu0l6uKAdi<}#J
zDxIp_L~79CoCqFK<}dNb3-CW4)V(T2>}KGuRFm;0r24mpRa4xOsuiu{-!Wfe^g7a3
z!~ukFWOy&x0SEToa!N%h5C_^*@QE_-X20PL<}m{TXVn+6B$;v|VGHn$-dzl|v7%NS
zM+2-pVW5FBKHrEgSec>->4(`S*$NSIT$}m<Ou2HI93od4(Iv@OwjMEpKRzmg{SUvg
zBKKRjj8)|<(yE$ykPT|wQ1*i2_0ovv8+j(H3H05K(gk1wU8Wb(fdd5@KZraU>-Q38
zFmPR&cJPlrLb?c{ppodpL+j^7t;HC18G>ILw>ck)l#79X?51}_Fs-2<X_4d4N(ss{
z7@*ba%Zh_r{+p{MsCC{3zB-A8@B*dI1;sUekl>;cf~0~a%KeZb(+ZF4n{CQR<@v?p
z|G=2&QW!jx;P<_$FL1B+v)xP7{KLfu;d1V|#JvTCO=vjQ#IpZhoF*YB;@dpupLjQ4
z>ftx!&5#q$Du}?9HBMTG!47vpIB-EIUGXfIG&2?RbhP6Ykg=s&d+s{C2OCq^PuBxK
z2+dvQh#Gc8R>__S5YY-c@rP0Y17=Q8G}DC}SfD7CFH+UWA_KbfNwBeEVC6K#$3NfM
zAOxsd;CNHpN~ck>^Zxg@8%@1+zY`Am@_U7U$?B>(Tcohn7CDO?nP?Z>(r(oCLCbCH
z-M<+z>x-w$GvoSni)w1XS%v&EPxpQqGTA=#k$7C-CM2~*A}7TBYjr`^P4GOc-U<~p
zHK3L-HnHkv*|bo+SJz0;<^$8^;hX(*5Xo*Dmfz>~n?d(|m1PQ#(#tk<pqiT?*OVl*
zj?=1WdgsSr3QUoC5<P96WZJ4&<8}7XSd(+Dxow7npam-IEEB56m%Xwr6--V7^Xp5K
z7{-mVdeOpjE6|J8GrpVQ(UlGrJ^jPGGd2uzqGE>zYIB1Ny5TxL^G0)nHnri8gbup4
zc}_XS+B-70R})7Rc#mG+nH#)4dd$!0-~VT0{VK<M$&QX*tJx14T@8A(Hj)wMHe12R
zAO7r{*>hxD9!77knn#a5UA6FS>0L2Fq|g$0i7D3ci!t2r&D&o6&aCihmh(LcMzUV*
z^gG%D_4d`+Bgd5EAhH6C1*+y}i(DJ%9NB72)~ZITlQMIlcG<j_Vs2xB%9V9Bw?rMB
zxhH#4hk1`Qw}bVRpC3huU^dFkrx?U>z7kbfoZR%h@W^RXt2>a<{XNYo%988sP0c4(
zDe%e?#r2|Wa=J4^v!9sDen~^lgur|-)89#J?*#mrd5<Gg?pXV+KgSrW@w-qCe+|<D
z%pvu_laMuqa+y4wjttE|2<Pk+`@zgP4emJJyc2LOWOZJFGnD(oyoT2SgYN7R^?|*U
zS=%edDuML~<OFMKj8JC;^4k5Rt_;m|sRJ>mIgN>=3YW=^?%&eTiE%_Vbb`>H@7smJ
z&+OXvxodJttr==-^cdM;cW7a(Q3G^B>b6)&=HU<-yx^MvhMV|hz?G)$O(HA}n;5(k
zU>rBnALY6qU(FHFrCQVs3yn3~Am1P7yq};+`2Hs>GddWm>_eb0cej@bF!U<~oS~6u
zT<YJ^n*BX@ld7keDKk@lGX{W%h$jtfoaYRtRAZ(o)w{SWcR?&p;>;n%T{gt+0R|xD
zRNW9KM)Qw17c~IN?p6u7f@{@RSdruVkjTF?aK!`f_P|`hl^~omvV~*dfb4NyQ6v!a
zP7maXYBwGvyuL@v$xOHV9MNwh1Wmhv1lx}1^@$aw<JtD7kA+eSYz_9hfS;S-o7YZ?
zrGNVh7cGu%#zCH#IYH$~fE{AP)USa9`G@UqviU<KF8Ugn>v_<?ha!-mzXhs|jy);`
z7!gp{b21ND39wH^`7M<yi)mz4xii=@iHT83EjN&7hc+g+Uoqn(l>*wQ4z)QHE?>zr
zVTd=%+!fagfZ=b}Az%~#LqJQO7d3S<`=}VcF&*vSsa$F8_lL;fRn^%A9cR})t~r&V
zX4TZ6#AB><leVDwM>6F2^?D+(k{wD0Tru%n$g@QkR%qRUHj@Gq4%KhSlP4k7GcP1o
zi49%_UP-!+yr#}{S@Rsp+bDwSOV^-;UH1xbA5BtDkiKE5I?+u+%!<OedQUdLAumE?
z#Szn-tE8&hz_TI@AOY+9elK2Aw#X1LcM7~5vs`B6A)7;06fmr8P>jfmkb+X;0gH+M
zF-ayXT|S`o>=pMGaG*g0)xorHq^;m16*c-Q!jU;uQ0vHnnatUnU{+RSfaz0`%vK{P
zk>dBrBF|>?HVMQP+yL*Tx3Wm|&u<3_ogp;MDn$%ez{4?SRf0?k7r+U;3%z6;c#4FK
zdWsAN#m`kL&%y;jVMwb3DDt3e$65?6Ib`_2z?D1YB$=tRm8MQ{)kpl9hH#S>RpB=;
z-`tBow&yB<%DCGS*?x&1|2zzas+3VqNcR_8vKpvAFc`d&CxaXoxN^f4^q&4KwnL2F
z^+S7BvLOF_?+HvXD@_~OBsSQ^&_EaMyWM=C-wyJSeF<8m#(X0r>7N?`28M|*OV0+i
zEfPH@0a=vLA;XvuYPe1@k`Yuo5e<qz=J>pMCsJj_0`;ekixeuQqD_5(AH$P~1mc#|
zM{MN^lDI-g<_jbJ#k>&KZ4(GE63|LX7VWxATj<7>lmci%O^zx~QfClM^vb)G!bd{f
z#+WgZ?INr2tY-we;CA$4<T9~1;rH~kzgCkWCCRh6kS)7>jNrlBv<(%q%t#?<(RrRD
z2|EOXv5Q~+UjVw_$t3|XObUydGO9yl{S<nlKv|JNF$?m(=n%C7XSBZ|ArbT&F6ei7
zVaTwGOusm2b3GcNo;2I1hCA<3IS(HvQY%JC+F2Vy^}02;wy1v~z$hPI2e`tvX@shQ
zZ9Kk;J4&)U!STlw(c<&P*HL9k8G*$J2K?oYU2=b_&FZu`&7J)wl_GbjGX<ZXZ^ZsY
z(f%r<6c|VKPt*>A6H<5Nka9K<>{pZ{X)*O%LG~J8RMN22h)I`<#3Gdl60O??d9NHU
z6ioh_$uKES^X7DmikEMtG$aw%2YGz12Ju=WN~WNvh%lUpym?B4-J~(nvF~J`nK7cU
zWv28v%|A2Zi1wrhm*WvmRQz;II+mu&l$%YQhf6xI)374WiR@S>J5?lHgjxZ2*<E6+
zEX<LgRto&(ujmT$)DTmaFPe7Hm-{D`u%H6eKT%-3<e}gCNJaazzo)>salKb2?}$hd
zej-o*Qn``}M_SeE9uf4&o=R}_wwR1+5y>x-a)>Att)^6z4&HNCr!c#_%HN^ir+*I1
zC_?1?zc$V?tjXwq_<+)}frL0>qZJqeA{}EOB_YxbPy~^7fCGuqGLV)~DM19JyF+EP
z(w#C|N|4Uy=J$V||Mh$EygAp|eO)`}d-@yad`K6npV;AiOo#9Yp1$_?;O$VcD?u5)
zdC0dRcq~%hgR2~(bWGD=(Zn*|+3eN{>|EWdDQt?c_+sscxXsC-tGV?faewMo!(%yo
z7cTgPP}C>o%9!6ph>hpj8(_s6q#W&&y|Z=%Pq2pxY*_-hjIbM<Wa|9-u8t+p#rKVR
z!Q@2phEK!tK%bvlNIc$85XT!6NB$E4rYS6s5)JFuaZ806kuOB#0Mll>N*phHXHdbD
z4Q5Gp7l@dKzL=(FGQ)ZFjnNZKi53vjk~oiASvy?f(>LkAp?|nhWp1t{H*=-aV071H
z7>XeHu&WBj0y(s#ktnLeTfp{z7%*Vl48hANKndyg+Bi?*rriaKM~=8Cj$hI^>P@g6
z@p&3xk+FXATyO`kk1rEE57yn@NU^=>IPV3!=5VFLuw<V+zy3h-T;opxcM!2pchM3U
zuey&t=p%T}hPF%`pBsqbaR|%vjB~<HHCMVi8ROcU#@O)77~D+e=eTm@a7XC<0^#d@
ztCSpEmK67K?WAM<%GoRC8{n96m?hC&mt*%}^@gv@Q09|XBWL0F+5D-n2Zn*5J)kif
zO1iT$$`8rj7v`qIIEe0IMx}5zH)(HJ8zs{{=w>Ald1tSX#m(9SgD%y7;gbeS%Cgh8
zfZZCa3!)lONN6OBOQ(JrlE?1qGW3{DMXWua$@+bq+naHk_BFsVdvW(B_kau%MN#Sb
z1Taw<4er-5283PJ8%)%Ge~^{y4^rg2ERv0^z3n+2AwRAYg7*Y2lV&LS6nMNDzaH6u
z1ghc=e#6Ywl<F9EWNI;mdLi5VV6L(ktVZM~gl8q_d+F)E1n;9Wf4TXO_!$Fvjj&IA
z_@6&2MINf*Zka6LGsN+p#F5U~)OeruTYyZTzt^(hX}OxdzYQM$NheCWny3qG_JnfN
zO+GVWa&#a&!Rb>&X=aECvkWVs!>_uG7XtkSC-dIE4Ol-99d)+cF6q60kaDdji~tQ#
zmwo*5&3grG5HMrncTf*;+t+St-yn~RG<FOeFR6|zMGmXgiWQ_$#QdQImL-{p9WD?3
zn0uZEO<p)~Lr$g9-$mKC$teVl5KjOfK5Ur*cDu8hys}sG&}hHwrm!5w&yuxL$l;7^
zl`XJ+2?rcJ@ybvZuNCW*F)g;BUn`CmfBhDs{f-(Rx_6)54rlHT$=yB=Q703qQSwbx
zsGX<Tak5xQ=dr_Sa1Lbhk`ReXd{D7gE(e$GR3|ev4`1Lvlvp{~@nku*&6d1#5*c%E
zx^PtnZtv!))=sJ%oLs-T-7#tR<s0kiA_oLogfWH`Pcoloj~u*aSotuL4u8U9?KIN<
z&HtcFnC;7D3J~IcCb_8_Z@N`dzr7YN<x|CO1=PV`HpXcz&H%X6{QKJc-j=yXUAfcD
z6&Q^jO%?ZNCGx9ED%xEM4!_1+C9+5OOy9B#{xX#HsH>cuxPh)pl7_!DGjlF~<+e5p
zy)~byS_$HHjr<VRI?ILCgvXe54~TxLWk%D?uD%z}5*?*wFjEoxk!b7YF|Asx>RTjl
zEZju-(UHE(3!cOrH*@`04es_qR~+_d9ANFDGb@CC>7B>y``0mF4Vpdugut@{<{f38
zL*W;PVyL(3Z{lmJag}Dps!l;M<3hp&HQK0<Ce^dAG{&33f>(b-D<~BBx0Pc%>Ex((
zb0!pmBrW!FVMkrU)2RL(5)|~$v*+B#IFE<yfQzc4G=baV8W%Cwv8f64PcR(DVSNh}
z7r7BI4Z9C`%zF?#3nOOG^5|0w;P_es&M*fHD2K}vKz#wcfrhf};}U>bc*;j}Tg9)0
zrcQaoE|u+7U|xy$B30Z0-3HlJNhsKbc8O5E66)#nmv4yapzToL%z2Tj<{8-OjRh`C
zw$_euDS`JBgHf<}ZVeu#)DC2}JUWz-3z9NR%C))DLz;z6jX=Q$?Z#)I@lRJ{1@Zp7
zHz`?gHY41J%w=(Cju}!eAB8Sbg404!4g~+8ofips*flHx8R!DmN4W;WrjS%|v;pfR
z=uZO$QNUr(P(h7fiB#PpD|?5A04vddjn{<*w<yI_R|CX5)-LOnZ2I_tQ`-I6S4=v(
zBou-_6<p)j2eu*?A@IJ>UaX*sOhSzLTazC*N$zHYCj@*jIL@(^<sb!tZUkEs$po}~
zo<kaU+&jg8*tioC1i|;Vw~4agoSeL6@doNuy3y0AF+dNC!dHO};?(JXdJLV34xgeW
z2=pI5wSfN7t%ez2>E9e&*wRj_vTV`oPxN5%&Cv4zbxuY=kGUoeG^Cq@Ty%Txpv!s7
zIiUvW`-R(Bm_7VMu=<0))k-$pqNn6>KckqeA2004(E*BnAIT2C_f>n_+7n;V5j1^Y
zcTu6s(Xt@GHzWWI|M_<)aCdvD*1<l^oef_%!N3JUw2>0@7KEC#-gbI-VC?DF#`geK
zOuz?N_E$+)07ECcwfjPMmeyOZH?w}rBys6#o{sb@tTmr%m2Ak@;ow|WNx%9k+?S{o
zQDv&QS3@pgc^R<M1W2`Ji}>glRU5zW^_9Rerv?}sA|@=jfjvuVU})v^7U|+FO+D=C
z{crEYVXM389w~a1r#Ek8w``N4NmqR~;NTczQ3g2J?xtr_Ent{t^xKJKgXaXT<<jR#
z&__IO0;^P_$z_k+1J%ikUj~nVZfSl!4~-9eCk%e}G!grwjZQQ)b)+&~t)yt~*D_hz
z(hYCjP&n9M*ht&Idz6%*xiA2c#H~lEVNuZZyq889vFuscRhzGvB=l!#5L%`jDP$*?
z#jKCpMmjg0LRfIzQIN)I=yJ1mOWg!1LCeBJl@0G#<34h5kAvi8;7#(X{ovayxb?`C
z6u?;Q@1iOjDLUx|24gP5V9;j<9awwIg-<wlN`YYaX|DGhqLE24$}aGTeQ~U#0v#wt
zzNdB91?(!8E(W@zB9;(l`zLJmAK-02G2~e{ywmS<mjI3ap?hEkXeAtP3DD)^4u!g8
zbH)zoVz3WFW4DAF?}h8nFu?@Y&U&eI&=Q<q?KLa_df#3>DS<Pz+m#1cP<_O);yf<k
z1&1|1<yxgPj!3u>3JMoZH^Bs(%+|^ZWP=s6bwAfC`CjiV-6#fY-R$G9Q<{GKxMgXC
zmcX`<WXysKCyI#$T&vinDQ^V%pZJ_ET5HFC*VnHCkBkG$y^&791CN!m>PsTU-~Kbu
z!GcaW_;GhI_E$)-NqPaMY6nb2j-~^Zk*tF9-Y^oLdiNAi<8zf-yjIA{?$AJumr&5H
z0wS%y_4ClJrr-jMc&sSI>&l5ljpqqaJ^kB(g1X%0DDwy&C)<NG0^`<auHWJ{PqO0$
zdylglWg_s;r**BCVQ`B<pGmpuX`jb4U~su!I0$hkt)-lbZ(0XcS^R*O0T_BAa^pOL
zMfog44i5yrKcuQkKrKfbZlC-$a)I9`j%<5RgJaQ)KVTRCe7XbnL2CGFb#s;_>jjM-
zWv42A=n{CaebM7xLQ1q%vaB+CyQeVwE!b1($j{3-4KeJ>w+!RE_E0c1LtGR}XmHon
zfGSY;JCD(`6P6TzI!%Mpn5@n7BJ>aVRe7y(l*#)J6*~9ZT0_jBIZL6M*^;5-QVCA)
z<=GQ=$wBnO5dr@3BA<KsfAYTK{qw_`m|&-e7LMQsM_C#CNUXg_Okf8;QkXxb9)N>u
zuB9q8D^eomwD^+KBL~IzsH2T=HOj)3>%+%|>on-0uIN0(F>-{vd2q4-VKa?bA6C0&
zFn3Hi<ux{bt#%sDZqd)+hproz^!x|*Bw=K8cU=&N<o7Bbc#Gv5d3K$*`t<c&D*zTs
zHB)|MzvMB?LDwCHyJ&m(`!n@W<2xMTKq#dM(I~l=`S-Z4*v7zn<z+{tc3A&kS}uH(
zS)H4suB3*~t+~Op&~$j!?pOM`U)!CChr}<mPjEv}Y^h+5Md%iPOvsmFi&o_DB#vVB
zU9)Fzw1{kr*)2&0+1|dnN$$w7agHC)BSShn?;vOW>TENWOL$nOtM*svi0HWPkpBI>
z&{|`WYN)_x{iMT|-T!O_RdqJfv9i!is!aRIxWw-Av~g5+buWikaZ@_FxmZ?^3l(rg
z%J}tFj+mL`oqsK2w|y#`eq|k-Pgj@)j2oJ>%0iM^cYY;qllvYG=MN^AV`T;~I~Wdv
zOF)Xlu$S`-@4l*E#=hrNCuPq|eg+ni8~=vU4Ph}$Kd3p!U*DnxLT4Gs3SaVWW%ht6
z{tK!@N-#eWV=wKyDkjU5CC-#-HdF6sM-H{LUbs2Q;8G)i;<%d_BwWHA=#WS?=0kKH
zmqElhhb|SVv@d6KxztR_uU@uGE~I3`gZ|3?%U8O|qk=HGZV>X7HIY+MCM@iT3~t<P
zrea;JFNpQY@0@-akag^)O2qMfibS*`-pJ|U!*^B-Lp6Jx`RoCREt!`tFSN?79OWHw
z{j>EeH(rcZAZO(!qo>N`X-C8CP37=(PgD$dJ<s*?q+enT>9dSBpKK;K@-=?DRjVa<
zWPNE_cdvyDpEH@F`*Vnn=3Q9mQVD6ng`fdcF@R`C&^d5V#I%LJQ0m=!>uz>pCqyA*
zI^kL9I62f<YxAG;KYJ?=1L4|QS>2~aMI}&kuR6EydIFCSTCdnZf9<ace@9LeN&Xs0
zYOfHsR8_uCN5B>>-)grOrf4F?Id;l|ek;Dh%(pxZFxKYA!tY-*AkI_MtQp`pZVaN>
zfjrb*AUV3!@U9Dc`zEgGdC#Ik>#@X61`}vCl#e~Qy=GOn-=wbw;(k%uUiMt7^fxri
zNXlJ`aq9gn<!;!TZG?ln+k`@$w)Q>x*FQ2}G&t0%9L|S+g6RAfbTq;}vzNE+nv0*j
zJX{b7-j1$RF8s3O*z#)-JWD?5usw%}03^6CTtkGl`%8p+(&Z#(K}>km7$3oW=Q>mJ
z_nx*}F3yo!+RV@c$>nI@fe7BGg0Dx)`HY1vydEF~a{8{V$^4?)ar<!AR@z76$^|1K
zM{}Dj6zj|3TV8S#dm#{C)}AAX2^{ejxU(YLq*goaLa?r><GRyBLCJ*0>Nu-woQ9MY
z1E{K%f9u<`-HFRE%)MR5v7dIr?KSHnpnCJXo8%zTh~`HjOyS=IZ<O2?RB=YS>4J7K
z=8Yet+PDM919fs7UC(MJpr7VO$#$f^YFvJ|Z}B~E7=~kSPQc3N!s+WPe9x8_wK|2*
ztL0SA?FuE>U13*(GDxy0?mo_V?pGtEO#!Yr`S`_+_}+Es120^QY3?Z4S9{Cp!(eS2
zgEg_$=F_z-9<>uAT+CdwohEMdY!MDh(uodxZ+#TjsqZEGgT(5%LiuMWsb>{*o3T^}
zu4pA(|E?+b^FR?LSl5hCzxN@5pP~@_nD6wAx!NJag<qIo$+y*O!GcHSH?+7(qa>u7
zkpxdV%8OsU$GmJ^x!L1OqsMvKgT$}A<{5~{KrOqCe^e7Xq_Bd|9$;L=GU!hhE{t%i
zX{A1GUqg#nn%h{u*j98PRan*<C4vb_2VrdA%t3~Sfs0z|P)stMiI{Nc3rR{oYXcpc
z!U(NkjYx)kb^+lqpMq*cWQDIhoWyHbP9G5C6$xIgvb`m{8q?LQPp<#Kv0#gNnWQE)
z?!;l};?ij9xXs=-VvcNw@sxL^LP+!Cme|(V+E_P1%sWjaS-re~)fF^yVU<uK!`LPn
zI|;=aDWg{By-UImY$sF}Qj<l(Mb0LmS(Ond3Qmdw;&Dk4Pz&V=DAu;o)-_rc=ckUP
zrl?75y%sbDy@=oDSU;Wl_%2bOF(ae~p<;+vH8$8D?=afzR4+@N)ySqL@GeBf#DR%J
zq;yI`oY*Rt{@zgB-zIOxGpWKn6;V$fnL5`9WK|73E-0E#nViz<3Q7fDG0#2S?X@0h
zRl&WGpIMH7M!!G~t-oNuDBo~cmZ%QM__w6rImLNv>V4WPVmbr+Plvwt!SDAE^`?H%
zjnVbrSqHIBBrm<26tbTKEthA}>oLvkzX-!w64hEvA*U^iM;TFhv7-jU@PG#9YJ{B2
z!_OG0nfJ+*?-0HXAqPgmnSZH@7$tu`t$kFAP+7Sg^q&KV$M|7Y&i}5_xtX=om0@_N
z8CHrPD+c*RrH`yojubbUIx=e}b+qJC)c@1V;c3onC8G~Tik>}|4ho@w;&1c1F$Iyl
z&Is?vcv|KODywAZXv{rgrHVaY>Bw(Qj<c8xQy%z%I-8;X-Sr3jh|Y-Vj>mP0OR5?4
zipjSc%_-OeWXyC=$MVI--i^uRURSo~G+;i@uS!lxzx8Ncc_B`-h>c)uYqA!oub@g>
zadYTa+F<CDIdUYFw;*)(b`>pGa}%q>n$Jw_ZI^u`m0z%v-kS5NoO?gb{iO{APHszI
z2)u)QY?{lE+w4O`(iMz7Q}{MP|1WoetkT*xFTs?-6XFnjWEDd~Dv=eZEmI|@^sz0d
zXJs^m#JY-Bf}Wc{B!jvnC$4aeymHe3GPo4U&J|H|O3PS8SL)VSow7%L3)Ar#{D2Hy
zgJg5BRy!Vy>e*WrrOy9$^86@S(sPt7a}u<D%JF*$)H<)SxHWqhyo;d@R@1Oz!{Ag;
zoKW%GI!9EL^h!AXW~qCnt8dErbLs9F@VaZxwuF3j+?m-^W*TGuLqo4SMyY$Ja_PlZ
zaC{xEX!_`QSo_~r{pJ+yN4d?G<{3_Y$%=7UR&SqLru%V~>Dti|_x{tN;-&r8-3u-q
zIDHj+L)mft^pFmXDdd1!(~mU$b?%v`a$QBPv7|3CdlT&~c^?M4PnFB2FAU&rk3xcL
ztNk~*?H(95JewT<VmnoHu^ltYzt)L4*#5}$@5f<NLksg`^xPOri;QHkA<aoBieed>
z<x0SgKk_tcM;2t5nYQRvVa=1q5QhtnTK932C>1~Vj_fZrP5tM6^@@!rZq3aN13_{J
zXuiGBL(Dee1~WGNzOL2d=JB$}ucHq)U!blNMC@#srTUB3wYRr_%H}VtRzC02HQeHA
zav6s?G$^fzoXO^HXZ7{?n(wI3)vI>v>Lg>QDpjAdz%i~VsJ|uDBN?HtmqPsyT0Yr?
zd0D1heCVLMhzOzB;BEGg@9#B8NQ^GgR)5Z{b`o0H_fE&EQ_7ozy-QP8;9Or|O{_j4
z_L~ttuGLPRQ<M#_vbmHvb5eNb!R&av+#ocu$WX2~s_Hg&a~7g|%9?q@*!QQU83F6e
zkO2|YZt3kn49|TxZzu#jQ|SOwb~OOG>WPsE48w8rQhu2?!%FR6(-ilnSb!2(>)du?
zXv!(O@Ww))FIc#X?w=>SABnBESf%$UE5=!g9pheFY`$#&N<BH~GqU=59mQ4mECDEJ
zdBrOs9R5jhLBwt_6lQQLQ8!x1s?}63PCYOnRf03nZ>C$;2E~6}r~eQ3c3K+*giv@G
zcq+I)uz3yz_qhRIP72kbldj%>rKPuW$o-5K!EL7W&|OLSqr#+ywZIUHN3}r8m<6EA
zM%*OjV`G(Qwq_)OBCo((|BaK?B&7Q30H8Zj#{c^i<O<WJr}WUzw(<%HCwebs`Iro!
z?ZtpD+fl~j)pzlQM*;w-6jjJ00@vAnG*NiHdeY>wAnKn<k<)xveiT5htK`u=c%+fe
zBdB0jeVxQyI1Tkt7Z;X()*bR58?K@`$O|Jdv^8EAUDr?G(bb3F*7>QW5oeIHVee*d
zSrC=HsF5GFZ{+My2?)Q$U^9{v@mdtXragr7KF}|6I#=Ii7{WKcxWWiWG5yo$H6XC(
zp>{?ADCYfN4nu5{+7<Lyc+g(%|4{1w!}O)fT3-a(2tXnK2Rry55+l=uL+r4?S8$tV
zb6z&8-e#l|$_F45#p;B*^5YU4cp8$8Nkjp9B7iT}G}2DWMtxMle|ab>VtMt+B0L%3
z4xeXtJ_lIB1RgLi=daj%9Qt?ct<(M7EMfV2XbS*Y7bgs`(*8Rk{DHwuoc&%c?X=a7
zWw{d&K(XDQ-;V97xy4(TQ75@D&@E<<aLo_D%{Cd1T^qVQ(1WiZtBt<?ikaJ6dsb7=
zqGej)RFT}pEBvu6T%<jbyC)fIn{3T7N_v~ioTc2YJP}r8gaR^yiGd6GFuvZKeL4a*
zTpl)@X8VfsSQpNvSug5MnMF&dL;AgOQ_wr)>gN%Ti1W&+shT5ePMO`3FgG)|O~tLY
zSwGsM<B8TqOjDTA#XKrn$jYonOBA#o#28mdZ7<!`9n19YDl^dyMUQ|%n?5HuC!1ex
zaocYfn!oL*&K>yP`r~|Bp*t;H^;8if!kog<601llAQ%Q0BD6FXj?YW1gM2<T6rH9M
z#v&Wd_Wo6HM6mmaVP^d~x{HpPn%c*DfYy!WO%=TldxeNl9}TTcwXyAt(74Jt*~Zp7
z;NJH6H*q-6g&wf`uGnub4f$ual82<U`zFcC@Q)zNwEi*gnA4ip{Y-4xwX!tYSO;Y-
zXYA^^oOfR)rS9zs8|^6GFfGG}SFr5zf3{sZP(0io9!u+qTXG!UH2T?Ee>Z|+UD;9x
z#aeD1{|Y9|Z_dh7%xt~eB$!OoJ&k2JFFj6+p_4xzxxeXo61NCY!{uA<w;{{&g!Xp%
zIl4EcFXkLTZIx#+2d@QcmTt;yd|}f>Eid&b_%k-Sbt3BoebT2F^?`mpu|P$<HevCm
ze`?X$6E27}dx&Cbey(f6>Wvb{gFLw^cTYDP2AZ#J|AwDOH@-2t)ZJbBvJ_3>Gc{-p
z(am(u@D`|XUTb_xKJ#{Sn9md6x~MP{RqeaHG{rHyy=Lwkb<nhsIjx&JFsb;T)(P3y
z023xuLHfVy+jWai%<LFZMjgm1XUNz1bxo_<+x#(3M^&+FG&fdGS3F+sJ^$G>-crF*
zSWkHBMzPbPe7wl~(_gFySr^IV&e6B0M<;)goGb#7aJ`7|04?boDRfKH`%0JlMwM(_
zbthC6{|Y9;UrfU}LM5E#>qhAAZ*BR9n>r4jbQDZ(;?K>eeNc6QkmHeOTJ)^jke4=e
zzMF4oY-C{bRu63WzsY&jqV4JoLSa@9J(3AMfechV+Ex7DUxzE4m)}2ev;RrXQ25%O
zmfZg*zlu34B=glIHm@m8qE2&yyka~BvRz_iAGsLE!UPbFFW!v)wFgsQx{L1C6~a#y
z9hg%*#+L2gd}<Z1l+?jAr9S#c<KzBp$6wsVpKn+^wF!yXlk~vJgY#o?IW48Xmvzz{
zqRyIs?19Sv<{Texic_cCxVmhF^;i2K<y$9Y8CCp|VdGjA7>$Z*YYFAsT>^jF6g(YW
zea&2lcIfygQXr??l@-Q$c2YiJL94DRW2MS+-det9Q=h3>Y%WUl3}`Do$Yd<`AUCYT
z-zIPSiv2$sitn*L55%SCTcU|iDhxvp8mc6*Ed}#U?<U`aeX8?ily@r|sKgoZP<;Dw
zW_s`jCzUl=`HFQeN<q)R4$=%2W>jIC;QIe17ofaAjFLQ6?lXWiPo$x$qf)H=B<TMD
D&Tp8S

diff --git a/docs/dev/articles/examples/timeseries/timeseries_anomaly_detection/unnamed-chunk-13-2.png b/docs/dev/articles/examples/timeseries/timeseries_anomaly_detection/unnamed-chunk-13-2.png
index a1a3c90e06af8e8c2af7dfea244d053e2570e08d..f5bcc02546f19dec19bf5b5c3889a0977e6e5aa7 100644
GIT binary patch
literal 13529
zcmc(`2UJsC*De}g8x{l$BK<`vBE3sj^abfi?;yRG&}&qhf`EoDRgvCX=!u9@0t6wn
z&_Zt^gceF5;cooCbH;ba`R};njB&~y5Ed&Vd#^p$GoLx<Gk4x=YbsM;WxNUkfv8np
zJl6q%{(1)jovWhy3z*4YKVJp>x#Ic4*c$}8cI)i-+|YE`2?+EzNagucy?{*owEq+R
zrG>8T6|>Ey8at|agFBLsJ=GqekjyVoeZwiD{t{w_=>{o87svE{etrk%dE}=DhMJj0
z@6OXd``Et}zkA$qQ$*mxOz+c~O2polXZDJ$f5_v3z-(_0<CKdakoF~gFG*k#C$UzS
zKu?}h*@HlLBK{8#JJDU9i_^`FDoesXy2~9Jr1Oc+C7c;G9WdK<8jXE@H5mW0x!7ln
zvb}1^$@Jl(I_4@qbYA_Xi!!Wi(Y3T(-mihO%|^G~m^99RG5F~lXt<S<(*r5JqRFC{
z9VSb4Ix~?zrcs8xrm%T6-rMD4RotBcbEygRY;ex8O$zu}H39N7Ho%Ot+4MD3(R_*;
zRHlwspd{oi#J13-*s8umINVy7COc@*R5<BQm7m*U^XmLCw7mUkpw7FO+s8#`8r_4c
zcI}rquQ0)AN83ytE*qZ8-Fx3=!K$*ecHmwlEeU5T&pcid6&;=x9<9|IhCY2aEI0LK
z4rR%9KwY%)a;CDY&#v}u9kgKXgHSkax{R`3)u+0rIpibsDvB7soGE<uXl5|d!>`B`
zWn5bAU3%*`u8sI9U6E><O)&MPiXRJC$h~=-tqo46o|u8ilx_g~Skcj69#hpg7JM8c
zm|DFsmGNy-sQZB@#V+qRTZRd?n7}xVKz!`=V(=h*mMp4IXbH~?9(ip+D3i|dLpVNd
z;WWG`mnj?cfM7tkLG0g6M&fh(+%Ct&UYB*-J8jif=I8J=#=b6~xiHLAm>hw$S+zG`
zZdVa^izqmnx^_uwfp7)xtg<wkJK6h;B(K~56uOZC3E|r9Y^O{73(xuG%DpBZ9443x
zJ<abi`uhcRA}_3qURrWXPYv3<{+qg*%Qyv{&`0I@k)vLDTB1C}ydK7=QbaDy8Sbrd
z@7mgV^h)=OiJhs6BV)g`+lwi&vahN-O(Qtwp+eK!9&sE}ew5ATm#A@}?tgA6>NFjS
zd6;9>9$Zmzb8#`Oc#10+$<$LE8qNt)KRpy)#&9geh4VW2@)q8bkSDKHPO;gY9Qc30
zB=9b*?s4ry<AU>M$D(N*bq5V<c<5_3qN;PJp@ygQTb?@`ak3`Ksbzvi>4uLE!jU*R
z?csyJwJ%%{&6Yn1m!ijnV*&!hm`uNb39#AcWtyYWr~8}Tb9o4oj7o~=C}NTCO3S7K
zcXy3zD@i|CMg{DA8rZ0rvec(`I@a?KwC+ugr3CU|bkE;VWx~Gp$B>;A{%4=PALgxb
zjmzVp(8FoHtQp6~t={oB<jeiz%r?-aK3!2Jmk5ny;TBOgj`RutlNmi@)VK_)OG%cB
z2c1P|apZhOf8w%Z6YW@`0$!%P7k7;)Sj+T|+A!Z8_sVy@XJ)|J6m`Pr0JW1^m|c{1
z{^Bl#kEpPkh8DhM(X?p$WLUZvsGmx_*xzEh6I8xCPQhY*zu)fAmdG}M9)HxH?vQMz
zY$Z)d)tcxFS3Q%vEER@s<eIT7txs=Nn-;P?DRs-Nr9<aP-7QEI??2un*3&6EI`6Gb
z4jG_ieK+c6Xek5R9Er{D2eTwU#qne6RrEts@pUD^Hlm`N|7Fv9WWi{zl&FPMzNO?>
zm);<K6O?Q*y<!R*wa(=5PNj!n#Kd&o2h}2^r0Zh&Eof<UiwV^$k)DfgI&WmU1Bk=v
z&l5vD@ZW15sX#YOLK)p@J9}3IR<DinA9WKvcUI$MP4jyjlxnA;V=h@;`=@?o-*5j~
z6il+8SgzgA!PSHj36PU&ugNdJ*(jL91xY5z<EB?>hVJ>CYwfb`$sAYMZ4*^eh(T?9
z5pn%wDFdiU9!g@&w8p;*Qq>c%hTH7#+#D5y;PEeWG%Nj_oeK~4?IsaP$B>;QwZxs%
z{Aj&LDpVOUZT65M?8+dsaK=@Bu?gqXeAB+V{;9de5vD-cIzpd3OM8k>k+sB`HDIS7
zZy=a?w>SV!(%~bn4ms7*<7)!LW@6e>t=%_14{O^r`#YYdM|(qT^V%O?`%;YkY00~m
zP-cAl3%%lMWtm0jL){@U$Dt>A`{kC(ybLHe%$+av_jl);eYtmJ?UEP#Jf?CUPFbXp
zmKIZ6jyNsjAD%RuifTOi?txVh$S9#uBa75|3SwA}_i<yD1ETJdTFi1`oERm;DtDgG
ztw9e44Cod)eyBGr5>2?5;cZ3Hm+zc&=zI994kjrnm=>(aHwlBq=BQbZkM3Sdo)mo=
zC$0$g@q<3!`}9lquHKQSYji3XjY-vM7lP}O#N<T54?J1s2UIWlGD5DgAdVx5njo_<
zaEgain%DOG?iJpMC|o&OsaVvjhGrn|w1M?fzvaUm{5Rotx9;m27G?Mrl>iRUt{!+g
zv(Po-w3tC{fV#pm4WG=0x!;@|Bed9Kz1oF5f<xPz^Yi>~QmO`OC7E{|^`EYbp!p}z
zO>9Yt`1vUqJ>~oBue|Wy$=8vxX~|i;jNb9an3_%J)jo<_NE@@bgmZKaeBioql>&;R
zw<goi_3420rXOP}Y`25DhR@--d*)0-ptz(=+-SXXEq!5m!V5?vp>E4APiB;VWl^tq
z`WxGfhs*KdqTpn1C0qXoiJ7S#_pu4PtmzVQoFw<&B;1bU%Ogo!n;kMnXM#Q~Ce;q(
zY42wfSGAoL@<Ol|+&MvshgRX5_;|Wr0XHm~h^S#prI3~x1SN_W%j*3rhZ>K+S^j{k
z*PO6517t6o`Dd?|=Bnw3fz9&O4rJxiR_8D_80sg!zQwh${s#Z58tR`|!T-o5g3{eW
zDPK{7k>-DFdW|${UdR=?hrATCNGhpjw>0;5mfuNK8CdlTY2Nxec$EIpJz2#L6!~-P
zvi2Q6H-)R@z#qFK9w)N)-*f%P^2JzpjtcZ`LUKq<uNVjV++BH>I$@Zum7GH~N0}Lm
zb{TMHATxPv^M5Tr`R^Bg?v`xw;eG19J0|RKia2hYWxHZB)1Y#^(tIIl%p@NS0;%5U
z%@OWB*bn<%p*g$y>$u|EiyHzclZENpFS{g(9mv`WRogF@%b?unpKOg`ju#8w9pRg!
z5w-p=B?rnN7i^Wn*mh3l=D5Q)>^%n8&w(Dk<O}O`ZDvZ;uKt2hhQ+v+itX1Tm{Q#_
z2q33?_=+zq&Ujkr+t~ESusrp|OLe=Y%WGwyL~nU0GoO0*OeK$S6TFi7%oRW&&iB8f
zYa(f*WFV(oJ#Q@=$ljD7*BA@C6@&JMmo2k5bvub$zEwWcAW+i7QL1Gqr?@zAwmtml
zZmCYbYPm^$*>6_Rlj0xX;tiPF<o;S=&S<IOKuC3|Vddsj6Ixp1FHjO|3XT2I%*>kL
zo@tUc;#n?4T#(*DmvnaT@KAr9UM)$oSYtuwJoRCMt?`YkR~a9B=BA5qB|4oPZX3AV
z0_E;i1qfvboE}ovn_~wSVP@+K3ddDuS3$4&j07%%UcWsXgDzH|h}*X;i^5I}wi7Q}
zemLByUeaN6XBo@8HSinAoOxaPkOsk*;}2y$e?m^m>s`hK%o>FzI}Z;JXJ==NbqnQ-
zVdp@(50T~qbhpk9sM|t`e(C)DJl{BwpU2Q?rTzur?LV*hs_700boB?A2S|>eH}&QN
zmX9o}HG70N7$5XLkY$8b5bm}RAPR0RkM!N?PKLaFGw=0W0+zxnYrc+Z9-D+SI_vg}
za>8{g_l21RN7pbvwi8sI#!@V5Jv5n=V}S(xHMC>C-L<vVmGVZQRFGn+ywWHxD9(%{
zhaL{mXGSuUy*+#Xktu+!ty!(z1=4n~QCO4SUU|z1BS}h4amR7AI@GE+C|;3@y=km$
zjg~F7^K7*>fI&-fbz2H;p8LoLpSDRQm6YZk4Yp+3f}~g%I1L?<DQWJ$C<tV1B%8*W
zAa6<8vp(^!q&`|-B+GaruORkz+)3@zhIcicDyn^mXV@@EOD^BpJlx!gC&-f!ixVqj
ztLHT1N7HF<fS#!6Ft0ttND0$3fnHB};`smf)Dpu6C0$8DyeB?~|IG!*2C!Y`Hu)uh
zzettQ_C*!_F7D_I2{tl>+v^KD#PN`4@ZG#T$>9$CN6Y=4DCiOs2=wIrt8J^4JP4D_
zvNn!wAlk2}D70}fDZ~rE2!AH;9OC1CO*8WxDB$P(SQ_<B?hag=tn?~Y@2+A|rVEjE
zWO8!~^BgloJP*=_0=66RHI<Y<o$gC6oEKqpOGJCH9)_V*b)VceRH``#>i#kR3<C|`
zdYf~<>jTa#(6JymP|G1M^V2<+EQ`(bF?kdD{7?1E+2=soB1Y^C`q^&Wm5=3x`S+a{
ziQ{+Lkf`#v30nWqSMZ+)eGDH-QOn21#=U<Zr!lJHC!HKyn0oCDlHG5X0Qf$nIvZ=#
zrl<iJ|C{eWp#0FkX?UXEg<ilc&CSNg$EQ#;i*Mu%@EIeT#`-t1<*8*$J1xf%3LD@1
z$D_(kM!}M%fiM0Xu((o#@|f7zS}Re&#0GZ}on=^=>ZgSlZ*s9O{tOkJ7;}>f82!p7
z&ot6r+5fZOUp=LQ1%ULg`F6e!&2?gJeAmz;IR*TF$6MpS9>|=Y@bZPYb;%{l%cZ5C
z5&#p%7|Exxi6~*;xeV7D`kJyw&}yEW4*$00d3GBglf~a+?gYD{JF}FjM{rY9ZIa8B
zE#C=D)@6}Weg8ksX>1g-ld2K7=@EJ21;#ial$9*t;Dw=v+rg-^1`oxKSuB&lI7Tz$
z^afy}fB0DYN*<YW_u)xf5Uw_hqR67-^%&jS3_Y*K5C+ieE~6F3+D%lkIg=zFk?A<(
zsHa%eio2|*N<HZ}>l~$D0`T{@st}>xYSYdJ*WZDCtqC)Z`t_QKeEQ9SpAyWcLF1QA
z=hIIvf}Y&HOJk3&U)miVI~Lg3!(K)yZHDmMPs|2#8*>0}ZvrdnG<_-T8$OY&dxbso
zuNe!G{?P581c6n4bHIH#p-VXe`e>}A>zZMUhr?C&h!Rq%C89D*!4o-}A+1ymU<(Bp
zIgt8rBQa=_bQ~ad`mpcsh~`e5g5Wjz1Emg1@9lR$P0#8O8ErI9g70;Fg6|T3-kTZ*
zm?GIt#uv9VHy52ST|SF?`-d+^(fMKmGjwdn5RCL2^W8_+K~J3Q+Jf!QgI<T+>^s9(
z`*nbcY5&&@-%Y3z({wx{b8aA5Gs+S-%zBSWN5_pWWqRVk%}3L{#n948`;Mo1(BP0`
zW$R02!ui-1mO##Y@KzZowiPpOToZC0Sd=heNARA9O&7B>dEw~y4Z3<7$<I*1imWOB
zZsoWHlw13IfaDU4P8e(1=WPSpQk5^0Gmwf@$=>BNc|ns|G8qp|o5O(ceJl2IZL>SB
za|Mynw4-OXoaY86^`FNM7&~2coJYQpIAh9?ho)c}i!I&7G_NdPntmTa!k>_vmxu0^
z+nrGEEflL6odPbWdWn}K4_S*}r0e1?_+=5-%9`=j5Ywc;wOzdWquve#0#{9di}!-7
zg@voDd!&0b3&Goa;6HGd`d6g;&$tuupWeMBkV@JeP#CK)&v3J`wY5bTXz>5bm%*uO
z;vO?o9UTuQJMohZlNgL~Su()GqQFPMzP{8{RNb)couytcFR#1qLLktus6Xd@;a*##
zE9g$f!qk*rH8#ImSCb1yNcc}uKl{RKbr$_DrankplIgwVf3I!FI!4&<SH##D#%FbX
zd!_r60#<~s=C3nX)%KK%sv&myvq0=O*M#Fl4MJNHgQnawLFnm~Tx^t3N7YqWWbx~>
zOWrYyw@pGePW+mRL3eFfHZ|CfNzFic){(x61{srr^w-bAaBeR6LT$k>oQG4(rKHop
zJBp7_;l)8tuPOy<Z(nkAG%i9fMf><`;uUV?m(!1beVTq*{Ij*DdxEFiSTkd64hR&$
zv0_xyy!r0;HTE(0^wyGK8PzLK3FR_#*l*BhUtB)gG^nrc>{O=vI=_ou@p)?LO0siw
zGiATtg<#z`{{9L7;E`uKzzEUcHbWS4bTnYaXw?Ifsb1w(ZcwT?r)y00Upu$x?kNCZ
z)%?tMKJ4_7W`+rWG>Wp7mCjTd2{{NUH_-L_Nf%_CxNV@PKAr2gq^C$#__`rg-P=-j
zyX6|`YaMXOLW6iKiEQ$I7BNf>m4%ffGnjt4k_LDA2j@lz^&d6rzB(*U4(&(@t3UQD
zpj9j?L6C^Y6G!I@UE7Lnt^n}EEK%7%Tg;z~%fYSM?amHWE>3*aU1)sQ{bR=Z`3~b2
zhMNc@9&9^OUZy>z@{JpVSlkWSjEf?ta?|#MK(YeLLM|ttN8}F|4qmAo5FXyg37qbu
zvtuZ}*AVmLqX+qH@l)^Mb?(nu>S+~U*iC9q2TASxjQ6c=ZWi6W2Bd7@#qtso6E@Vx
z`N>MjuIuv{=`xC>|4-3cL+XQi9uql6dO}PB8R44sz2cuqP7p{d$4oP#zTSc~=hz)R
z_iDu>zeLViVy|@G?5Ex(0Dy_&Y1B?C#iZ!5Q2e7|C$pzi?CL!4-$OeGf4!M>_zPs^
zWISZh;G~CV$BBY@?Z$A0rE0-<=U|y2AQ0AEsD-TzO<qxptpsT9?}P2Ny`pea7_%%}
z;+k1|gV*ocHzO?Ra@3&Lbc?OKscWPX_PCR`3RmSqWj5`{I|Ip#;*72HCa>7Ku7g07
z@XP@l`Msq&h;)VYu?9Vj3Kjd}p}Eau{NcJ)_lbmT(M>?g?e*#`LwLy*tE4)}vc{XM
zZI-u)iE~aZ2$+LFMi%i_FqiF+-SFLl3Pcj$HTK1%cM9@{i$1B8Oo8(ttBZgiQKzEJ
zHLgwGGdt$JEGgOYAtFL7uRy)YkW8Ufbj&&3-}}0`|LeCpfU{`kc2}mu&6K<aFk55F
zhTIE)T=*WYfO%vZM4G)PX_UF_kO}jc0+Zu6_>$$!c4`nPy9oG9%t?Xr_p(NVEXGu{
zyh&@BlL12EjNX4OnF;tV=kN-Ku<fe}uFN|EQ8V8ZwIY1nP~pEYezZ962$&(?9GEd&
zldp0`?~6k`rgGnQbd1B?B;Clvu(GtjK;M6WZ`ctsB#AH3{`4Olmxu&$`wXye_V)+r
zcF%%ToeRGCf@>De4S+z4Tn<9YSYba#al7xeFs-s1=C-SaqVnc2;)~~I%`fjlJDZ@E
z#~(kjg$R`uGtqPbI~dD6mWH3C4%*+Kua(Gb^^+leN_%ep#pT5`*{`Aca%qB5D*+tR
z+-}>a-sH{Y<L=)94I|Sl?YII9uR(fsW(FQ6^xy0fwZOGL>{V*XR#4*G{KBt(k)-=Y
zrbKf_I1h4knY&Bk#-{#>41yBbRex7>pX2($_UyMmbO6AqYrbgwmsZ7J8COS;mNRX;
z>%X6l)LsfkHkn7GAJr{L0rzE_yo<7wx+f@*Q(Z{e8)Kr&b=52>d8WWnU+$l{o+IHt
znESb~Rtfc@mBnj5=bZ={aoUz={>twP=({ej!}#Z~qgwZNW}0`eq|<D67i+2Phq<im
zDH_g=+fH>}JFj)xw$gostL6zx&N4lgUqCobmX(M$+&T}M^^(dF@ZFenlEL+ncUy*O
z!pD3k8gobTJvze-U%fypW(g>KCGOzH3!loI(YBRAe4@cZdC||;E6t-7#E>!RvC4v^
z(h=YFhX*-#S%>N2XFC!dJ~^@tuJiWs^7Zwr@-M#i+vX%e`8F65MF)Vv5I}VC`hu63
z&t$#WakDQm*w0)XxmH}PqQVn-<K>^|U?rtgG(vf>B8xpOpHn-Lyv}HSdwA%;bdp0O
zhpzl;u!82lgNPtYX=~20?Exo{3EIFiLj^W!Dip&>OZaQ1=l+ymjZ@1$`4&^n%c^lt
z%}bTtNFgMs_kQ_Irf{)Kz9NcZ{qLCN{wHRoeUi5$zUJcPYK5cL$PKa*ADGwbJ;+s8
zBh9Aux|eiby?PZ*!-j%wuMXxVKhl3>CIh6<@JyQO$8g-CsJB*vQ)c!00y%C*dV4fu
zQ~okpB7z}G|KSfC{yK{=mg7U;JBr<;Z{xwx76*828VfV?)YMdsn-3T~<QQ<m0h%pN
zevp<gQsF1uFux~!Y`9sD@r`e=jcEo?f8!qx1*-BNo<Vb4Ic@g+IP<4=A13xFALlGL
z-wR{L1L|4fv(ZiL;=4D3Ezb_DE0djZd~7VQt0}O?X6BXh&vz>k`DDJ_kaEd*U)EEd
z5w&Z<^;ucJPxDhE<;+tSP#Y5vDgC?PsgmtVkBXzCo(M5>=i%nrcRkM#%~pKYyFiiC
zyqGzZUm<gUPqDk+l&f-mCSFJJcQ8)#P_lSYd-3yi`uJ=&A?uac`YcX%+0XH5Qwted
zH>NU-GR`8oEEJb)k=$lTWSohA#Lm$$!5^E6tl6}1)hQ?mBAXO9Y|1u<r)35#6u>`}
zvJ?Xu<=wlG_=v@qg~j&d?<TAjVFZH0(Xa7JL59e=zWH-c7f|~l^@(};rDr-xk4TbR
zbEvI@4>7mqW=rC_Uzy?Zm8{+pvT~)om<g4X=o+fNyZiBF!DQMRwcZ0s$Tau&qn_b&
zpzrAWYC=pu?zRb!;W`X1+ba`~e4f;otqe&loo^P~78t_R+v0R)8eSe9qsz#o_+4Hk
z3P>1K?XK8Q?%%d|%c!>wey)yO|K~l*cp%N~Ja*0KC(sxJ9ra4{PFBqHT9)1OTpx=R
zW>NAB0t;ud-eyS79avh^gXa0isoDjIoh6)$J3hs4@*=w%y-8bx*9I9(w6F#I<lag8
zxwp03oue2s*U2qEE!^_j&bYJV<oBj%!?yYUpRPc@j%<oIhBrV^v7sy3JPQQF5uSz4
z;;I&6L7By|1=E<~rJ^{(Y<@JOo;%U{#|k~AQmz4g0rdR?!0NIG#70iLmF!j$m|BKC
z;<x#cuj9%#c2D{+jQoQ}+Jzi^{rbepk0xOdXnCm->iVvcmgP_f%Sgl4T~tn)&xKXb
z-Hf`gpXV;U*0rs1^=R=05Mtt(+spRL4Z{dp2(uT#Tc>-my{Vf5h6AbE7eU{5fLO0F
z{n9dy@ppr9a$r5q(C<a8v_E?-rcOU^T_lhZ1nTV#ywBHEmsYnn^zoW<g5`%1C%MoJ
zv2tA)m6r8ccu(!3s&~Uol(!7S+;xG5r2p=j3iGEQ$&v77gF9eF5CFVY3SyTYNce?t
zQ%Bc;B=TMICQxR9uD+FloYlL3CNBbgwX;ISpK^AF8Z(>Ga+cY$o~+&1_097x4*;%G
zN)fep8({P3<WH(S`5kbu3O$(lT-$|Tl0U~-DhVV3L)<uR*hG7NUk9XD#UU(99*p!@
z8|a@3TCQJa?IQ!R?JBsXGc#!WAVEpXD1>{f4XYU+)|n9rxPy@zY`DL%`6#iWGc0rv
zazfL^<yUbMVQ+(T-!P?g?370jm?xZ-?iEX+85fAG`_2O-yT*QQptJy>XLC68w5^Ns
zY0z(IEH-$!DnTlgZ+R|3&t07H=+2HkrW5dP-?T#9zB9rwyK?#86_MA*fYboFF1ROK
zP`c^US^5SQPpiz57BnpT+Sck!!j-ZL=7;*VD_HvQ4<FUe4G+X_#!a;nrkjnmP{A}|
z^&Lga*BZEd8MV0vEH8p`hkKwYFmuG?KuRpP`I~Rm`VIy83^j3|dr8-`v`Yh-#el^c
zhuEiTw6Wg~ne=lBgn1DB!bW7X#KB7Q_;(AFh8~)XRY7lR|K-EgCrK{)OXm=*9<<FN
zBQOYz<-NvW>BiKfLy4fko-gulkDT}|0M6x%(H~VTo;nHou#bBcR|0I}VA-v{GZ7=#
zijVDIgYQ78_h)974{ouBUD8STQzbS1sZy?fom$dqaG#d;-!y~8IiNCmbfelmfM}`)
zsCM@sJn-Kb|Kw&95YU*DgT3u?4**Uc@PRjf4va*FN&O=M0o2zxzykNG|7eZ>V+HmO
zsIY)S=PHl?S%}G}r4I!DqkYTpN0`)xv2WBmgexdy2&^2wcblE+D7DzOX9uW&##laW
zgqquJW{JEbN^o3u%@~z+35Di0aQ`7hgY*+S)AE6j4hmSECIkEgqs#DR(38lS78?NM
zkh5n2RP2(i?kXN{{OB;(`z8Xot}#zqq-|bxwT?a@%bE545y}5gweO$1<p-%3H~i3w
zw)}5UCZB{(i5rwKo++7E+g;yX&60NQYl!w;axox+N7IG~)A6lP9Y7|6dd#?P@EOC;
z0n5Gz$RkpuhH9G|^_1-p$LN@eXeNyw`Z3K-nZ>+L<rUcMPaxe=22^egt(QpaOmAK|
zMZ=Cg0@-9hc-e+WWoabw|6AM+H#TqZUD(C;n*}}+*d3IRA}_xotdxme6!Ws<0Q5*c
zKKUg6rptndJm9`YRC2U^bLxT5Xg9<A!NIkq>_B|pk}jjM%O<dooZ)oPmrEGGotTB&
zXzNNPi-doJ^-#}eZUQj9l#yGL9;t0SEp7a)2T-Wn6cE((N2g}~mud`1ZAliDS}H0%
zFkx}=5nB=Cih5ul1!FJol9zO1!KMO4%(27mrMq^!j~!-e?0V0B+&;v_T)lq1PR|t3
z*xUib_)d?;4M1eQ?P?j?$h{LYSgt~>>>)Zcx#0DZeTo_pNLG{+SpPe0!Q}vK1O5}X
z*7aTTCCsv+j2Fbpmr_{i{H(BZvD=qiMt~691E`;)gskyg^joS#TJSua0gnU9A(ug#
zoVfMbG5h&tnZC+PpjqJ*1sjR=p`pDRK6OLLXV^Ht!-hU#<7Bt8YO9{*K%KN_WlurS
zfS&MJ*MD(+^#Oi!tjM8q{lLLq=AsV`^hZZvGvFr1#t+gos=FlIrUHkx+fe<}!8V&T
zmW5f<fq&PU_|m)oODfArcGJx}Hf4uon(bo<pFX8lPBre~e`br~@EN~_4iBQP`N7c3
z2te>ry$R?vC8}dryj$c?a#Skmo!Qsn&GW+>T>8vXmS@X0fOkZz=G|1JB9P9FmeT{M
zY!6CA0J1Oqf2og@oyaa*@6QAi@Z;4Tot>TP=`#MC*Y62^cLo|Ff0qu?^}T%gk_~d=
z)dXJZOG;OOz(*#|gPus_^QN5xX|;9#IYMH^UB`c;(deuAqe@_T-*xTwH!zQqOVrav
z?2%^h5UGa`9j6+-3|#cv0S)eTp#~z1{7A3Js^GgNgls2tuWU5?G{P8j7WDpWb*{GN
z+8QcL1`9cDI}&lO59<EN{)i4j2KX<=O5l28Pf~h%IuvRRABS?PDJm-VRpy@ux)*Po
zKFZ3<Hv6nu!_^%fi+mZ+g94m|kU&4?U;5uW@o-~_Z`|Bm1-AFfjer2g86*DIn4W<&
zH|yuy%8X|=H#5&-*)T7;uy^ltVekUxYVsesJZ#5seY2q@E^hUBa6Xjxq%=^<MSbnS
zU72}uzvDv~S+v1}VQ%*hO+kiwLGZz5x;mGr-3=CLF$g6@1=>eIc6AkM<;XAhZeVJa
zE5|Yf6SK2fX=$w}Gn>F|eEIT4KtN#G-<M1xMo}|;7QD^IW?5^mVQ6S5YUd+8{r)z0
zYinyC;gOu2oRHA(kaV?FfnQx+r6bETBTi0EhLsllmhjdi^XW#fj)8%Jko56!BM1Z%
z92_ia2Te@u=;)A>mF**_R*p5eVunJ}UB)V;q@>a<L;a8U@z0`fNXf`#V3T`nk^bw}
zIyz7U0&%o10G!cnQeu}yAVRxXk44%?$Z~nOST{aCep!JExF{SBpP8AlN3Ie{mSIPn
zSXMxF?>;zB$G{MX8|B}vtmNxOp-@L~KPCy+?!G=TOD<7S(Z`P)t$)QV_nI|&whs;t
zo>WKx#OmVW0#~=Tw$|4E*wWIHE+U;JATKYErRm+AZrR0ewfGZYPUy~%-bzaZ-1^tm
zT}yh!cNvl%%g2#{_=);+=P#uR=Gj8|O!r&$A)Y#U>Eiyb3>K8_Y%%7X32UV|1<C8g
z?!0KjauZ??caH638WbwxI$qU)4Qug?OP3ENb;dFEL&i8bIPP7K3Sk>MJ1Njx12%z<
zE}7u35fs#%m6dgr5Zz<j97Ya;_g37xeS4QbF(suyr+8yyqt7rkGc%J=0fj<$s0qWv
zdi5sdBhLLv!i;gn#l=UPk2pDh_4M#yMNG;K@&?m*Y=rI&q0zf!R-PBWJv}}9`}@l_
zJb-OPMQi_VK{luW2Vi6g2?@ZqKGNas9|`pSPiGzh(?=ARkDorh5=m!fW{$Bix3>No
z8><FqyUYF@u9BXVWZO@prmCt6C(OKUg#$a}Tt5+#%XmGiu1>5p)Mzg#R7ojHT!ByV
z9e4?uj@8%G(_2C&_op^_EkvcJrXHb&aYR6q4c`7m-%ltp0vbdbWAM$PH3B<?;-O6}
z!=iW!<*#khZ|}d4%<6B$0gP`4<TZ-lSS^1e4*G2D7#ke5`_aY^95`m;A8%_tPp4o{
z;knBBg#l-=?6mJ{J6{BCa~c{#kOKclAJ=NqUO(#!p#HEr<e$|l7u!{M0Vh`!Iu9hr
z7l1V6e;lSD3?)=Qudc2FZC!(6o%}{7ofxxbpPsR?hDOtuF~8>w{1!Un#9c;*0n7kz
ziqQ6Tqa4K~+$MaEc9wokf7Ias*a^V?ujv?_h#L(x!Zw|r10UYnJOr9-7uf-t`*(Kq
z{~CVEQq9*X4zMvwHt8o(m459g|KO2)8qp<xpB~sp0PO!hn~3xjQb$Jz07Xm-9p|M#
z>?Q;Q+NJ;h9f<M1G~v9)$=CB7+))IhHf!{tqgy^uqBc>!q%c5uEgs7_mrd6dCb3Jn
z?jlDZ+3*#!kuJ0*av!}{8f-wWai09d%djaPGUXPVAF0Jq%Ob#?h25Wwp@u~MGik72
z{_*%S==9U(vI3;c=y>ecb()+KcNhjcPjhWLRzp|H*{y7{?$+CbSH)c1_|LzkSF_3`
zzn&h|DffPHAuNI$7>@RL*v%DAls(9!a&hYZqu1RiKU$wPc(_6kHnTs?RT78+%RPZv
zrq@k_BzVNZ{b{58+#9Nk`wIWq!aH>L)!orc;dn`%+X)al=HC$Gq~gYjqdg%dzV(ri
z??2FkpGuWsYho*n0u71<lcmuTuA;bymb`%<Uk0VS3#_jXuC+YO<IA`e1}RoV(+n`y
zP|W=Omg*$4%a_V3MC|?Ty~KBltsfu1(j`zz7c_te)T)bIFaojepDE;G_pWzHVj(l?
zqn)vY8L2MWB_W6Kd1r#|>W`4<!g4mfir%l2zyo|5+lW0MmxB>;n`&_)&zDIVm{Pwm
zBp+E*rf~eS7JMF})%UV901s)`XJ&EMoEjKY-F$2#&@ZCJdVLo^^`%vns;53Ql`HiI
z#aj?}B@)>LP88@+@+n1nq7&J<S*3gcZS`+M6)OFs_$-%5mdDvggaqmvzTwKzVnaZy
zOarsC^zDNW>R}8=&W%*Y_(SYewgr?{@mryTBQnXW+h&UHB0xijFH~_Te&z_;brY8G
zsRXvisdx?7cZXq%k`{SUql+$_S{J-|xC%Re76c~kww`i%?(Ko!#lkau`LxWN^eaEk
zWZ_!2-H5hno1H(M<k~Vh*dw%)iH|Z^Z)fZfaujcrcZNu{rk4X6k4x|695T9@P@C-&
z`f?FH=Qm>F?`Edi;IHW!t)$^lDu3AMVN7aq{NOd{KaC8Wf%v5&Ycr3hSI4_5Xl%2|
z;GbIc3hC%{4?0n*<spr`#tJMoeBN>WUu}(F7zdf%t>Fny=snQnW4Jl)6A-YqliyLY
zb&mh@Up|e$M<LtzVl-D>s|aMDWEujBd4bwzL)#>%ARo(n{G712>~mgG<jd$-f!|4F
zZfj-U?3913q8S^jaj^BMSLVlCz3dq~(fRoyTlNT{U%xsh1B0X-$WcR)f@|2;6&?I)
z`72We$ALBQ%XEUL)aD-Q{HNka2`1&zt&g{RtHohuC%pNGjK_La+k7itg8LOOsn{dd
z+|?CF33sHH8))wSG!!ErVfrw!`$|~3Mq3okv32@v+HK2@{B*YkiMace2`1ia?9kI0
z-yaRW*E&=ARKQN7&7<$lk98tw!5cFO0g0>EyKC+v1LjJ{>>c2{V-**}Pt1=T<<Bk4
zO!+vFlg@eX&kR^U+rt~Rh}aq3F%l~&;m?p^A>kS|sb00#ICzAhC8$}6#f!fWlbxQK
z21x@!U2439Bd#s6jo)L>tSGkS+(RX(Kepj=sMI&CX|>=u!6IbwXkdA?_U4ZO_rWS2
z$;*|QBZFTWnGT1mi$#4@qa|*WG@8Z*>=tN*Jx+!Z0G`QT0z+3yv8JQm1I>y1Gb{9v
ztxfXKx=9<BVWO`(Aj7OC9lwR`q+uMa!gtROR{9lx+&T~}&Frcm*ji5U3#^HiD`{2C
zP~uWqqDhvf)x&dm-)XZ5T9wHeh*SQ6gF@?sgV_T>G>uAp-z%9<9pIvUjdBNhQSu(o
z2!|UHMQ$-rZwnkKM=vIHH4t^Kutx|`>|1ro64c1zNLocGM^ClMlitr2nd<aei+Y4e
z%0Oc$R=si}cs;sOoaez6_U9&Mdd?WG@<+}?S!U+q>(Q=qIM|Nh0KMVMH2jvd_+N^o
z2A|eo(WfQDuQnXKNh3pd{e#uylY^hU$l-zwHfRtn)8*PAEWrSd@}Fac^qdF#R<PHu
zK#@%*waem-<C#n<DGYN!?aOly3%vese;_@nJ-}<hkV)X)HA@S4?Q*#Q%jGcdiI4fe
zn<54M*JXp2G|=JH1unfa8&hjbkt?j^uX8`&%Cm&mAWbZ&7Jji=GS#1u5KJJWl;%bz
z&=UaHa~mGDhI?VP4sIEWDXF6Fnd1#B-FW@PnLQWqX`-7f!f$YWdQKx-1ESq|r^$Y9
zV#{U7pH~whQToXQXlY%cr?D&-u~}nE@cRa?*xV-MY=Du;d0ou9OWdCC0HsUR{>jah
z8#zt9UI!oO3*5EnI%R_4*+P#SgL7_rOPZr-fzCd6j=9uA{YU!d0~TSA-Vt@1=c79b
zF9ga>WLzyyw=yvSuR6H&OGhh)I&zpEgjoi+b)6i^&5xB!5ee_>7qy#S;v^F4b}Fq>
zgCB_AA01#AL>Lkbxx#8CC(=wiz|()tBK}T0_s^<=hjp%6>3anPdbH6dB>I)<RB(LO
zIAR|DGE3P*_2WMi75+he?;J_X#~z5)D)-e4HFy0`J$RN)O56D1-o|E?6PY2LR`H&o
z+c<~C>e{iRSzmw`eqwbqyczME(Dc*SMUxRM_Ja>yz7+nJi=|ZFYNH!96hmRbVp6%l
zVtmLX@7~*MY`J?@51F(t9E)hepre>+VK=-#Zz-H<KkQut#{@~~bYu%MsQ{5*T~7|C
zllBiN(C0snDs_owLKflc4|-)V@-U_hM!A+r8gq!BNnX^@l7a+-<zdM_k&`COUKX)`
zusADt9?$Y2v?U*2zkJwN&@ZkCkOpj{U-LMNHPcAVF8eruILR-DVRUTXKAvyux6jU)
zsx9M~2(rYC*sX11-t^Z<s(-V`-oFAmz1V*JVa)bKq?(gM#U7!`?3QBp_S_mua$&Ki
zUAMD5DBtOq2V)&>$Yv{TkOJ+t`7mCfRLMH=1~;JLHoK+lXy4$mK7N7?&$)-HF!z=i
z@uH}O!hx?Ckb1AU#P_?;=sbwf4umlY;|9g;Rdg#11T3$2;yy|8k|Y}y1l$*Jme{Ie
z)yxyK{itGQZwb2Zkf%cDW9~cQH`<y*tLF1kXf%yE>ung4ZPD%%uwI)=7HkXopt8$X
zmOsaQls|!`$YpQCUB+=b{hYv;wE>8+Aq&~6Wp3{?HZdG~d=e)up-5%(ar&d6A=^=v
zoy+zXqxbDy>?LS;xVf_qGF-)~Iq6Q@=!_<G%3G|=f1kfMxJB+w{5e(X3gPoeARl)P
znW)pjUy<W%2@Ys1)k3+B9qoEz0!VqIVGH{ur4qq&^#@T6$ra{tsccCDO}8^Xw*@&;
z4#6|ciz`8+kUOxb#8QaXhR9y>4WRe{sx$Lw_DUa3Ox1GE+kt+au^X<Rs-2EDFBnsG
z)F&fYix|d-`jhfrRTqNAh*x~BYk&6ay#6p8*;a;z2MP{UZXx^;)L~vm3taW<7|Uyt
zk$gAYieSR~EbHiMj}DuD=l8vZy1|fLC*S1t&hQ-tt?NDku38dsxt3B!jf9+#ZY_RR
zjt0u%y6+X1<%Te^=?V6&$lP!5OKC!<%%~QKbT$@`mv<cdFcMqS`h<oNdzW0B)HO?g
zYSy7|c5Tj1D-{Nx>#TjuGcu`5jTNMVb@|5}zU!($?MAr)RrJnv;K7{#ugw0x&iDVP
dXLnA?*Pf}Q-&6p;xC8>J{G<82Lec8&{{yrSTRi{(

literal 5748
zcmbtYcUaR))~2ZlB0)u@2m}yRnsg~4l7y=CB0UHwf<i#RPy&dwh$1BdN)Z92*Gun+
z5mb7W-b<u}5(1%bK;66d`F5XY_kMrm$((a$&dj{?o^xh?&vdj@8R*#PC@3fx)YV{m
z6ckiw3JS^sT57W9qLFPcS+S{e%RredQc_Y<QBhG-Q`6AU96NUG`0?W>PMn~nr9FA_
zBpn?cJw5%YQ>Pdh7*3x)&B(|I005Ynn9iIzbN1}nbLY-6Gc%t*f1ZVf<-&yvtgNhT
zY;5f8>>L~%7cXAq<mBYy;^OA!1_FURJUqO-ynK9omo8o6=jRs?5V(B#@|7!Bu3o(=
zC@3f-BqS^>EFvNzDk^&I+BGpTF>!Hm2?+^FNl7UwDQRhG85x=D*RRXU%F4;f$;-<t
zC@6qHAVoz*Fc^H}#tkJUB?trpg+gI4n6k36ii(P=s;Zirn!38WhK9z?n>TOWx}~Y9
zsimc*t*x!2qob><tEZ=@udi=lU|?uyXk=t``}S>PV`CE&6H`-DGcz-Db8`y|3rkB&
zI2>+eWp(Gy9cycA8yg#2Tid&L@7}w2&(6;7{{8#*_Vx}A4i6qYc=+(4qobpflM@1g
zaCUZfadCO{=#i_dtDBpfySw}2$B&Unq=$!xr>CcvmzTG<HwuOF@$vEX_4V`fd-CLo
zzrX*}r%wX{0s;dAgMxyBgM&jtLeOaRvuDpjLqnfGfBxddi?Fb;moHz2hlfW*L_|hL
zMny$oFql`bUPVVo$Hc_M#>T#W{rb(DH*s-s2?+^_iHUFDzD-I>diU;KN=iy<YU=y<
z?>~I_kd~H~o}QkOk&&61nU$55ot>SNlarg9o0pfDpPye)P*7M{SX5M0TwGjIQc_x4
zT2@w8US3{NQBhf0Syfe4U0q#MQ&U@8TUS?CUtj<6<Hv@EhEJb9VX@fHpFcM?Ha0ai
zefjd`>({T}zI|(MZf<F5X>Dz7Yinz7Z|~^n=<Mw5>gww5?(XU7>Fw?9>+Ade{d<3Z
z|G>b&;NalU(9rPk@W{x>=;-L!*x2~^_{7A-<mBYk)YSC!^vul6?Ck8^+#C*v`|;z)
z{QUgF!ouR>;?mL*9*<vMUS3&QSzTRSTU#R#2<z+XL?UrxV`Fo3lSCqIZEbCDZ}05v
z?C$RF?d|RF?;ji-gm$#1lBs{vS<S?if`b0c;g7Ow#CMm1;w-N^Ov%6_VQIw2>fB6X
z{ptx?(I6?lGg0W1R_CsAYIrbRsWss9czU_^$@>^fStTKE7WljS%8bIsvbx$m+g>G>
z?JNaPa}oD-m3p$Gsq@|ityz0AoxSnmhT8bn)I9+S`KtJ^B20^x?^j>2A76Z;=UY}x
z{z=cMmdgyZ@8Y&;o?L?j(gxBp(K7J>cwpQx^bIt?9{{8RLTLU618@(bkkex{)vN4q
z=fnKCLQQ^krC8Da)~gpy^GQvFX|eQ}^IfvofX@Mm>Cbzc!rl;7=R>lOOo9DtvcN`#
zkuLyIJyD+~g+ScZM5rWqa5)ia{{7~6_RPyAFs_7Nc}X0yu@8~%&bk)JmW|9#=%tRu
z$pmvCQD4)&F!LsT6-@!zCjx^Va}!%Q0i235yRan%v&QM^eK)`1>6m`?oHx9@LbI95
zZH;mKYn7KYngQBelJ|A)IilfrOKVZD=O5!rI34y`R1$+&kOtunKDk$MV|}N@3^HOK
z?pY2+0q6NrfxW9quN*e_rg{uZFay+<3+YbM^kQ?e5n~%`u03B;3o&Na{DpR<s{QhB
zfP<K!?*9A$-|%BR!qA+%!K6|-&*%Fy?WOT>(Cw^-e4==-A;PKGC7;tVIKyecu;!0-
zFmbfhVT`|betDZiN=fu~dYr5B@e7?oxrvGj?RyTvebC-!G0f2KIfAr>#VbIXwv{oL
zg`ubXEG}BGPkzkPAXePbcf7%vZEt$pT~8ivz+1c@#%jHzVKc|r+}5g>g~H6|xo3K5
zAPmx9ll^0ZuNRiCdt39Deb%yWeHdPVq)RfjedHbY#Z6eZG9*Kglf74|1a4Ql77^$4
z@JCuHI8~%?dtlIym#w{epmHU-&h2DKzr=7ycO-z9+8UaWlLodUN#1*8*21^Z63G;^
zk|}!xg;|Ol<7%&&Wp6ZZ53h<K7Y=E-iCCxWRrX7HDgUlX86u=##;$#dR!p2*4$HVs
zJwIm(R&>>DnJ#^;zW8Q|X+vv5|M@K^?Xz$4E@^B6;JV4pO+NWcav+s3(`?lYjL5>d
zeYK?Q&^8C=ldF``vVYhJmZWN-+n%VQCp>HMgWFB9Sr|bZfimnZ>-NKaii1st$SZ@T
zJ$Tjyg6*}uYfJv<*R#FkpM0BGCPSoSKsEMn_ePuSu_!|n+=wndzC+r3Sx)Fk%1A(~
zry|#z#7$B8N&X{$9ISW5j*cU2k4&yQ07kd42REze4%JUmWdmYi8*X}a<={GER+?hC
zBm{jjx0YL;a6+_dIkx)J#@b9*BfdwR>F)%fnh2J<Ne{)QhQtFo6UM4)zTw>M3@Pja
zif1?{>nuwh5X8~mDS>0o(rRXS;4FtEyw^>en8}@35z=3`LHEj!En;3@NzRpVHK3up
zNn1o%gv9rxNNG@GvT8PNMq9*VGB>U#iN5PXvkbt0h8<nM#sIV0I?hvoqGmF#^`{jY
zyAc>HG$wRKjG?^wuEEUO<s_-~Q<3M{3Q>gWp_wH($D^>Rr@?Q5T@c<FggMmw(LyEq
z;1UcKt0FW8;Mwg)&#*%R{m9*+=oM{p7h0x$^^JT|)`FrkdL`T&T_uosKoY1KOsJCb
zr3N;D@jy$|XZlZ-pfbls9El;2_MklEb%}QV<-Y9&kyUo`a2Lix=YTwqn+qlBSOhnW
z)C9t+t~p6tDvsJI?H(sj7DdYQ+!LSr+`zJ+pnW*2NCOPL!uY@Cyg##KSa~}gd10B=
zIs!l*d-II5JkH5l==vZzAQj7eAnl>zU(lk7L2y(30<$AxCZ4%S)5gi2RSm6X8dZ1P
zNSBqUvelFZADI(J10p3i0jgYG@off<D)^UU^|ANJn}Hc-hQvA!$v$}ab2D^e*@L85
zYp0(QtxFMSfyn|x&7=$&-6_sV&tu6sx?XqjDMwiG5iAJF3|#Y@@GOEPdC^}3%iB>E
z;C*{CEW=Q9Z`16Q52jpsMFK}ipH3|CQ{KM{foW8Pyxt1yYu@Gn!rn)DqwT8iR9^rm
z0wOi=z}l`K4Rke#l^YNk-KA9v0!zD*gOBa%B5Cd<)BXg{t-n!E4zVpOdqO)rO2y;s
zO^>c$66XF>fz<nVEIh;&0*ygz7|R5i<HREXk;P<QR@;p5Z%W7N2BO~={>j*Xp1~NI
z!NZB?c>009Y1x(@m2#b&WRbF9Ee?~JFUlz?{9B&?W9I%wW9Tr<+Zm}xQ-9iJOkUYb
zb?%q=ZJ}}>v^*+heJcf2E`No}YR0h0<8$HCrZ2{s1+Nv_^Bnb7zMy+Shm3L#Kv2Yf
zWV7au8HnK$nVfVk!z~wmvJvbP_Ls?^8=*a=D)SW+mVQZgIg$|6vZOYY3}zw3ymPA|
z;Q_&NL-G}1f@`O0Hveth4(>7H&^p032>5uauz&Vw<URC*hi`+=@a&%5>e}5jj6Y-H
zjh81g!X}6`@MO_fc2C-iO@ka@%##C1!-cpl=&#j?N$lRHl9D|dY7=yQMS1FT*G+|=
zD_zx)@)))vwx8v%PTKvsmxH0%37L=Npeu|;)Q+(IoZASEL<%k1XCOqK5dP<iAPRRJ
zYrLI(U^Mk~t2v=wo8#MP+9%AD3&-vfPCeWBB{2q%$~8sYXUq#nQy2s9ab+QYH7yH@
z_u(}Y+h{eAAa?)ifdSOVyATyX_G=Co?Bo<0<E9YvD_Yxuc^9Ap$U~hH^h`~?lzme`
z#<fZmIe-(|HNO3b_AQIw3TR04i3OiIzQKSQcJh)6o4hP^eECEtM98j)IRLK3m6=v*
zk;|FLve0iCGBK}Sx0Mgvb%MaU7MJ773=9w2q3%u}&PZ=ZjJl;cNIXLB9#`Zg=cP>(
z-Doe5fzrgH#>S(xv!^Jif;Le}mU74YD3PJw@iFST2eK;aWB2;}yTcrF;2$bKqAQ;N
z3|HSI=3GWcl1|-$USc}=Ow}h91&KfyDTLk~LbC}#J}my3(N3)jT38-%G6svUX&&V~
zF{n@uPTyXmym_tmSqS;qZGt^i=lX#94Hj8`Q67iP>Z@aFlCe5$MIww|MspFbA96?#
z3j#ZgC65?-|32ikh1#AO{rr`m6tioc0G|hS^>+U}Z*L#lYSK-_FFK-t-d%~9aKF(b
zD;b`HQs>!*_L@VwguuO?{JL4LV3^VM+D6<|U=x4ZUE5<ke^KP{LTDtgXX4Y7$+`SB
zP(-SA*}#0va@g+JmGfI&s1=rh&E6V?!dXQ@+Q5sG*~&Jh0;2&tIouQ_Ks*P$yto!b
z&c9Sv!6dcV_%rZn;EO}sJ#?04#=aoPIf<F3{*+6Z@Pr82VP@>`Q|V?IHL*>A37H$i
zP}{BJl+KKihdcnmVh{j|p9XS)sAyYjUAvMNWoB{w3{O-aIZ=m(Yi=rxi-(f4^1r0;
z-}O+^cB>IWE+T$fbcdH|=Xi7UI44ipS7nMQQdf)Pqew6Hz%Pe?s3NNW=WY8}<#F;q
z&Qu;O4I3xNz5j#zP@24ChCda*O7;COYprcFX|nVdD!;aYw2K#spAwJy>oy{$^ZgAC
z`g~rJDe;$@bK|H8XX!DD4wbt0_5Um)%+^LFtD>KWe@aY~e_=HUOQ3_8q^^G(Rz9Mp
z`fSxKlwGlBxz1qju*{GI761Cm@C`ARc2<K?0(ne)L#|I~3e-LeMBDzCx1k0WKQqn`
zZJ#d;Gx4CrR^!-tMSSXaJRpIh{Iz&g0n5m;j{>lP<v%`gTKTcb@fU~Ke06A<Z=t`A
zB^K8Lz^&{HxX1N7&mg5eA-E;{dt$Ih48dchU~g3W+}dL3_Ei~?8DL+HeCCG<;J)+1
zC>nfP)Wv+O+-Ck^Qx9CO9jqYc@ClniNmxd5g59P-FNacnm3~Bewh_u0xaKafOLhD0
zeq>!4F3$L*`E1Z0&)YV23>+>uOv<4Apc@ju18>Vgw7g`yHYH}4=2<4&J_eF?Nas>*
z-7fNMdjcP0a@UekW_azHPuNXiPmz8_Zo9@3cI&jm2Y(D^G+~_FPJ=vkdP1_tA*xNU
zLX0=@@7ozAef*T3)-2<|*HSX72p(2iJyXM&P#&%8Up;HfpA%5J>r=Z<ZBuVM*&n5F
z&L!d$@Juy8nQfP3Lb>;MugCL)BcA4fBH(rhD4|JUR@o6PTlcnY%4v(Z*mG%n#N{c0
z^xk&#qS(QQ&bipz;|ln<Emf)4GLfoE>rG<wE~7u%rX<d%^I*M}wdhFcOh|!a<f~s5
zhHBp)E5|d=^?x6m$y-6?RY-NrkI0x1MRy+xuD)58$(Pyj`2M-{ndE#pOT7|pCq3jA
zD`fpCKtC8z!;O6fgo)CM-91vjA@|NZsfk>SDppvp+k;b_u=fSd^Y5OmUsqU{QQ~;d
z8jOxX|J#>tM*?Km`9_=6iQFbe{QVSYU6&|fk=u7Vdc30!|5mD!5?Mi*>M->)iy5?$
zN(DTi^Ye3VK@si{<CDWQU2&pvHT$hO-Eq?$3xxg3<xsirw)E2-!B;Noc^BL}H@Vhb
zV%a-{B4KJ-{ey8F+XM6^qhk(meGadIR?|Ab^s*wX<`eOcqLZB)(;m5-rDdaE7$sqI
z9ns9D`IjA{pzQg#PkHCqpRUPXHiZ-mzkZ0)H0onK2w)JK?Py+$mF`R)yNW+QEj5q;
z<*LFlR+0U{GFQR(gWaW;t)MUD*7v*gZ%cJ1Gqe-Q-oU6tek@vM=RwWj{tX)T6jvOt
zou>Sm>B!X6H7hQ;WhI$Tp-^qwAi1oLc>~TI)*xfo$s~uyu}q@(1cLb~L#OQeq<;v5
z<|7=7w%I1}g=_qDM<Y8*#SJxFw<Bd}iuBMtagprR%)mGTsqDO;&60rH&B&$Xl%2h(
zg;t~gk8%0m7!WF#ymq)8G)TRY%=_JG7=pu#SXoE$Qxj$1&z<lNZaoL|!bIYh{E9)H
z=w;*0$0{CN%<HITkE?cJH79`lK>$z%jF{_H1>ZMhuAHq*NEm7@3j;t?UyzI}Uf&J$
z=oT)@?_C9o;1gXx-|mYTj73AyqqexUoM<Tt(az^P%Og=+Z-}g7>+zYUW28W)$-<{~
zc|RWUiC{Gc{PN-+l}KUP(8hr{&0t@LKwZlPYC*4BZ-XIt9Mqfs<$Q{07+{PwhnSgf
ze4w((F?v6++&QWbUJ5pn|2^z&eE0VPLVtD2BCFwJ^ma-aSPsq()LM$bbmP-ql9=p9
z?W^%RmjCSU+d~qtB6@r6h}0yLg4EF3T6Hsbc(L<&>?M}*p4X+ZL6{N|1hV)sS_)7|
z+Zq4j(~#>}ew@UuF?s8zQ6+75!~+TJTEnRP3jka~v|!N2{vkSv>%gAalbr+XyMg6_
zAw&c-yehPfz8d+AceM;>_D+<w0*L73L0Xr|oE0qz4R@`o)4?kOwJXTh_D#Ik7UzSb
fx&Hp=rIiCYk>t{-6u2$<8ybbWvKA~4V)gVNn}DEr

diff --git a/docs/dev/articles/examples/timeseries/timeseries_anomaly_detection/unnamed-chunk-14-1.png b/docs/dev/articles/examples/timeseries/timeseries_anomaly_detection/unnamed-chunk-14-1.png
index c8572f66c853a17a005406c9229427b07e24b0cd..f68feea6841841d40212f585ff4de12803ea069a 100644
GIT binary patch
literal 32775
zcmeFZWmuGN^ffww5>gV1v>+voNT;A6T~g9DNcYg8w3Kv%gd*J~U800^3|&JG-F<HS
z{onVTFXvp>`Eb6S=TbqLdFGC__gZ`H`!Q5mQ5yFF*#ig!f-Cd(wJHRH8Ulf!lwqQR
zfBCkFRtCOc*}v6xf<Ukd?|!2U&-z?JAWtDOuV1}$OGD1O8WO*`727-2@BhY!Nk6Jg
zjM?=0sroD8p-BSlz27;LWnV>4f^ZvX51ulAzJCu}Avg4hIg1Drize{Pm${Oi-6;3T
z-Rt2xd#6gz#DT5SWk;m>3CvZLyRy<LhPC%;z(a`;-t|#-2o3n|2A>`(fl2dPl8K85
zd?k|zO@1hWIrLDr@9t$m%Ek=f>&pN_Zt&u4nz$q^@D(?!Qy!Bh-~;CW|JMJHmnr1c
zHtI`xAD^3F5#(d+_h|yYNGarvL{N*(eRz3cyzG;Z#AiLZH*MFV=Q5^lNdmU&?Q*sK
zVta;&Cj%Yb*}qqy;7|#l-CXW!HV}iKla=Z<z|L)#oumF=?lZy8x=Q0A-kdiHqteZ8
zdMu;EtjmWf;-R5+IFZ<Gk>z<v%GIs_z|y{5Z^B|P$Nr$Cy`aNx6DIB(GkR6khvZ=j
zse={t3<~Wt^_ckGW159}8ri6u0Cgc+yDd@Q<hWq09wz}Kp;unQ$HW)Mk`^1<>FkA7
z#FggH*zxAQ^I)$+1@Ax_B91_WguhlEEKo00HSoYv5{vl3)O~*V0Z!<T&i3<o{7)=n
zBIHNWYwhGIGy2W1S=JJm1Wa`I0t2;bOQwBqh!Rsin{?Z+{O%15EDa9IF5~=OF}!I1
zq~(?h9W4l7v!UoU<){gO{l_%L$?dvBMH4eKi;?gAUQc$LJ*{LP2s(?)_SobpURRlg
zowwY+iDgLQRoJc|bom!6AEAE(6fyi!{8(j53Yh!S;NOaHe6Kzhma_AZDO4cWryk`g
z<$M+Y->+Ah^CH@sFj<w?DptwQSIJXKd`sBVtCTALVF(ot=k`3!pr*DX7o!>w4I||q
zWoK8xxC=Uh_KnV-__{SJ>#2KTW_fMOj~|~n8E3OF7gYBLF=n#z!&pa*1+2@pVq>}7
zI4def*|}6O6mDn(9?~2f3>G|+3b#MnEsf2M#FPG@GBkU60*4#SyFJTp^y-9Ht_}}7
zhtZ?^u@=79%_jS|Ll4W#!Ti;}Q#|1xG~zYia2C`S9j)5PR!k4Cz^Sa)&1*Ov&A)VX
ze1U=gYb`U!*SA_ue)QE{D)F+r^}(UW)BY7ow9lv{3-e^j<0h>W7P~E3?#8CWYHRaF
z#~{~ZZ=l2Vys$8ZE>%yAvcLRqlKWiRza;UE{t(J%grCm!t`8bz&SvC{xIm5+%|rhj
z1r0H^oX!#EIG9yJD^qzzc%V3QZfFd#QwWC5uXhPbG$HyU9_3&WJo@+V9+xd%J<npv
zgwn|HD^w<))Z+@Tz>(*^du><k?KVS0vn<TAOUq+mnTY-^6E!xj*_IW%f}H13>&K0j
zTU-13zpHDBs973TS$~{!M)EIjE+1;&EM@vWIKJqo6e`d&&Sv<xR8d`>ts9-xzPk19
zPAavwc>Y<$lg2O6(coEGS$M%z_EN7`Hl5jJV*J`ln+sr}p(+0`z00QMhv+SxY!<nx
zolisWAoOp;vqpWNkWdkvojW}IP@8>p%U`!7$)vl?rLgBd!{)yf+)VN)@a7s9FL=fN
ziA_yD%#)<S`{@GY6GbFp>jUwG4obXA)(%8_%WX0EAj?ttzHv$H+rvK^7e?F?Zx@^d
z_m{%rU%iU&Phm^r-F`Yq!HK$Ns*T!?65IO-rw^UNPvj->#>Qlg!nOpF%a6n)7Fd(R
zK#Q1Gy79Lz8%!D$4sIedv%u)+lTV-iP<stheDQnw^xp5$Ej>T|gyU9uaSUs(Lj}m$
zwQDuD!f2L;G5hu9w{Bi^m#;F5AdzKdqtesE`&0Dh8%|$p<DoM9If*lTjYZ)4;k-ln
z)_N&k>V7!bEK*Q#2Z&(iP;GYI!#*`>M(-0t#wM~wESlF<cFk>*<=!W|)ba5ysAVV&
z3-1jde<_&2hgkbzPolp=4Q5n)4db<K_-_J(X7K~pCudIZ5ucm7j=#spUe+7K-<;O|
z7N4wwoFU}8bG6`)e^KEV@UVRvUU{|>n_ge9gq-3bDg)zLYo%`G_0Q_G^(9ZjcE0x>
zC9&n;0hI@)KYG62hitsXP<VXhS@{Gr)$R3M%|$`_u{Y=QyDNVxl^fdk`_G?!qkT-8
zP7-vy&I1ICbY!FBM2TXn;#W*gR3*sIf`ar`h1Pq9oQ5bRmwO!!w-<?1^b7?A8A<8S
zg-+6RtFJ0TEc|i`X38sU4*Tab_3fLqonHaoPRmzwSWB1bj*Nhm+=H;9rilx4zKt0+
ztLH02(?V;+4*wjfQD601y?vL9y!7Jfx4Npk>*eVddAsz0%s^I=JSLViI$HB3J3EIR
zhBBThq!x0DR{T*}yP4=WT3V}vA<f`XD@t{k+~{lUa3BdDPS;S{P1w>aqgbQFKuAN-
z!^H*d$(1pMA6T1KRv7qbZ!Zy!{Wxz_-LnxR87h*Lc>B!Z#M(h?l2RO<b3+0#O~u)B
z05cRA#F@nMx-Y*zNOROjikt~WL3_u}t%4@<7DTQ@k-qoxR5ir(D{&Gg<cr^f1<O`E
zFRYpS!suzcGxf)3`;#>hgigz^CE)=Nmpr+lhVU18`nwR*r#SYg>VA*PAg}!HC*j;f
zn?T8vijvZJytMQ$s~iLQ(s}xr?UQ~c32D8USPR-0$P);uJu2)g1t-d4tIAil!L=|B
z!&(!B>>XxbiQMs-NGh(8kw$Dj3Fkegdj${-!+XxHq~hqRe!8tVufKg?YQrF)DuHGF
z4?FGhSfDms8%jv9?mc~orRB#gj(XsS*GetU#QB8j{;+hoxMRiL*8Ug3r@b7%2dHaZ
z3aQ-oCHk%EIa6#<gCJjWX%}R^BL1d-rN<-Zy3-g-o?#>=!?{->A-Z32>>=JCts&y~
zl(9Nd(6vR;iqK7E+LO7NNs@j9-wCy|x$g>Rn}vLg3S_9c5JFD!vxk!y))=OSJchA!
z*n9Ww7yb&30V*Nt?*TMNT?v0ke6RI>+PxD9x;QGUIK?2*mycF{c{oAL4c>Hvb{neB
zJ4IlAmOr}hvMTh&tAyy}=e;ldBE;1eo*6ziM4@ZNN<*2eoZdBQ5mZ<Cc(_*Xkn_LO
zaP#TSW!?3OMQ>sz`YP($M`{QuCI&Mkk4{GX8(lTZ3qRb_#!s0(?Vh{;0_nfm&i*vI
za89K2MxlW8pl~ar7eNR=61;k417X~IHi-6Q?++u80Pgiuf&Y{1C0KP%G*5TXK9;xk
z`(3w-K%d$Bvo=65zMyzvzx&m>0C+l;*D^8bhd+Za0RXYa79u&h6oQX7frj3S{_=he
z3drmN_FARBbBk^Sn9Qu5({?6@(0?>`2Lghn6e~efGq!u*(FDKZ45HbfkUzY)?)PxP
zR)!hX84pAB0+a3i4?qHVcjKq+q;&1f6q`pcL)H6;D^B_pExuKzsTuQZ%J|RlhA~Q0
ze~?wFdPCi7o;yMPIz*aEV5-85Ss8}{^%sO5q#E9@%6U4t!YV5JS*#=wolG}W!Gbh$
z4#AVVd+#s16A*|&6j&1!T47HK>&OSur~?C7(G#>;^-RxI3|sRuX<6j%!f<ef0}P#(
zwWvRtJMUxNmoItDEb;n7#jSqB!-wLjJv|ftpFXj$1nar#ZF%mrO;vxP<H44HC-L8m
zgUCNiNbaJBcXgeMKmUTSg~DQQXKjryBg_3nzp!BQ0dcieFt5#*gztb{{!?ZTql5{y
z-@nHZPtZhRN#&izuk+(v*p7*LsvQ}Jc&gp}VEy)djWgA#`i?IDTd4rL_|*Pb$lKx1
zNl6L3I==oW;-vOwX5X+w90a3M<vaWy-(LA&Uv9XvlCk}~!=D5(aa^i=I;69+(Xy70
zg99H{&y0foR8F;7kx@o9gtY}_y|94O1YM!ewraf<$i-bY{qWq}PlJPbDB3ZCB#))0
zx1|z$QIWB9>Aho}2-%2;raZj{Mo8jN$o@YCfP1H~_RBw^_9gpQTP2?Cc#}TeaPH2|
z!H)QK-}@pBZEC~0f{xU*UdL#$G*59G^+3qw6Z~I<{NwVd>qC602vkmfdW4m@;#)t~
zL~<P0+rucz4@-{f$?|wl`=DX=g3FV>`9<$4TVJc+VW6`+<>f0|cH8<c<H};xmGyNP
zJw3tVBK_i`fZvEIGY%g6@D_{wcmKu37L+!3_qrR8I$g`e)F1x;;$3T^Wb)^u=g%Fw
zy2|~0mVY%a7|K}}=_h!ic*uN?HB-|{$}%5_RY<9VDB@Fve|d<UEriwo&z?o43k7G>
zYO&g2KgfVQm5J6UT%0W5iy@3{d#dLUNJC4jTZ(Tv_41D)Q)DQXfG?Ha&xJnH>Ryl>
z|H(UKtkT}7+WPeA;VXY$nHXUg%lfv3&mmfTmha4Z6TqmFlbM)We?_SFCeY+aCCkb=
ztYCVaYBbEeru=W-s%p`3bVBLRG)hG|WK^Z3ipJiTd$^1=3Ol_*3Bs2Sd+dhf3J6$0
z!t1JD$|P`^Vouel|DW&v>C(}#`%^IkxnMuT)caP6FX{=wR7;D)s{CaHpYgFJ9o@q9
zX>@!bzE~e}%H@{Qb>R==7|;{{v&XRQGJB)3!7Oe_FdYs3{rmT|Dsp@CGT&$TQu*GT
zw~X-}dxVp@H2U6FtoHvChyOvNE6ibC+05FSZ((sz`3M<0s4TpHFNLprL}@vUWACq!
zB}w!L%;I<jkS#S-KCZ8SvMO=M`0$t_JWG<xkuC`DjfY@+`$MAT8*P~wQ*DjH+}GIu
zA?J8E%J;gktIePMV=Mxp)o|il`^Cb8Bp(;o3?MlwFu-t{*Gj&R%J7}~BYB2^z<fQ1
zY`y>1alE)=DCZ5OU^t3mQa`^v#A)@%|Nh}=Ol-%=?k=wITn*c=h>p#%$?C#!{9zF0
zr~k0`zPf#@6a%qVdcTj9l=I2nVz5^AgX_yey%*VU{*^-1o@Yiy`r9wb@KN-1Y6A45
ziddigFnDb#{SO1yBMgnVqJDvXKHYf2=X5}Y6ETb)JIKV`dmQ-zj|UE9PogQQ)%e1U
zfTTA3!PE5=D`o9a{ieZUo!MLe9WFjg%cG>??v_l^MzzZIq{oQ{*Js0Oc#l4dW72%;
zOAjmKoV2dl+Y^BxVMps9F{s})XuWReWQ0y^+{YJ7^|)4&9!MABFFl>d7rn5-<H5tC
zLD^c%j85J9xmf@%VR6^>hka08hc|lEc@EaJqFh4Ue|shr8fwksUQu)J1sSR2MVbo+
zl6-n7aZkV%uXYxOjFy&4OdLShUH;YP9xWIZR6#DV9}4GRiLcI_eWIQa-RN>|PF=!l
zAXI`qyUq}BuV}nDItF#&gN6p)({Fi&dQY=mED^8GdnzU}b?UZW@0|_2Pqw`79R4P^
z_BUN;yFV^24v$jstl14{%|Brr#!H<wIyGr1?U3gj_yh#UxA<iJwkxBueL<p8Z0B3$
z1uL2*6e3RWvP|2sA_po4`nyZr)9F_@sepjMr=GW%G;`#fklK0Mo0V9DkC=wTgIRAC
zx*{o`@Rya`zLJDTr0r!{PvqY}4fplLPyI|eTEAxI2pe}7!~%;|Bpr2lI@ds!bbp~?
zcW=t-Y!Rcn_jQ$=oAEf09SO{BPFgHahbEEoe6>)aTvK=GbAkX(z`e2qM^D`u=C3T|
z_i?Z#E%K+#Xs75!>E8!69UNb)+Zn)V8Fngi>5HLk?hc0Ae92febeuSpWc%IJ=!U<)
z&_FnSMN8_XwqFozt1ePi5_7L?(BBHPiE`2e{Get3OF){8QYU(TujMxK=GtM}#xgPV
z0|_oIhysrIc%#0gidX)>P?FOH+Ys$6rFTfT_!kC=&FAR17n&P`S&OZI^qV++gcI0{
z*Orfph_k9i?r^A>@p|NYvDozy^Vf+W+`%m_<JZ@>{#1#UEC19@Gck1(M~idkuSf{(
z;h~Cd-(%Egx$OVl8WjGDk<XYn)ff8A65BR9o4t>{(t;|m{^bPvCMuihgwsu)Crsf{
z-PFEcb8|DIM)-OElX;`%&pv<FRZ%HL5x~N7TOoL9r$5k5FOxkp^LyxyqxZilrUO3%
z)bze6HYDtw0a)vk<rz`E74kXo2#SBQHIWYMa4q!(_W-J39?`~z+wqu(n4)%ldCtF5
zcf5EypQFX=p4R2AugRi>7VEJu7#g|2KQ+=W&`L_23ws_y{a;<&E!g%_UP1W31a95-
zDZq308oO?FIVw7>2xbv7u5xGk8N4q?P36l>5%YagcAVk4Hz>)(tlA*|dgNcG;$YMM
zq^Tc`iw=n&8miluvYyi&=(ea!{pW!gR7@MaxSkhNV3@Q9P)>=T{r%r9f1bSeI>7R@
zsDhU{KQC!ioh2lQ_&5G$=Lx;REux)<UFn5Qc;)_#Q%IQF|NgM3ri;z|+Z)#)y6gRR
z>?b@0Hv@OM>tF6sgz*c$RgE2e_p?o&{qsLIocxG}HS#?I?fvd_Fu^nXtGcQCv?)^m
zvDJRJQp@FzoYeBov|V@Yyz#Ga4^cT#Mf`iC>YMw0P|sKfsT`HOkt{u-$!&Z7&2#FP
zDgSHy4<1Ok3;2Z+F@gX@@#{zTe}zyb!2NFqxg%M=w+z^i1Csg~ARJ8r(a*-%iH2VO
zld~q3&_4_#RR+FYE)!YqPQtg1|Ay4@90+?8bji8S(0P6eYOlSIQ*UT|{5<|I)+=sS
z_qTsG>(FBl-Gb=<Hk+~cg-c*)@ZCoI-#E6oSvo`i=JWx(zsGjv-<Ayw6<II+1H}Kc
zuS+K3)EOchs$C-N={!)fqIO@yWeJ4cTild*g(x{5M`k)`z1vn{2jR~M-?I&i9QLgj
ziz6$t`(}5>a+QpiG3=Y~b*57Pz^>p=@8!r2-?{YbVJERw)vYhUSw{IPWekI^3QErk
zG#8`zecMt;Yp)z05DZKN*%xz&PO^AcFidf0@+aRc+#OlA%BvR(&7wH0b@*MxCOIiL
zQ7=YP{OVST$dUuNp$~`GN998g&W{Ho8%%s&U9QHT=rp`^QcN!MIBJi#i};KXQY&M&
zLFnL%9L%>*H|vQCR@LCM>iu-ynW-O}lb4YJi2`ln?vrWARG)DP;Sad4PITUvUX8r)
z9Nx?MCCXuQR_-<#o2f*0Bf%$9`c=KV*?X_X+|}UbH;E7x<yepIWpK>DtDs51<X3oF
z8ihUrECiQu=G|&VlZC<^HeV~2Vm(XUymhsG8;vG)++*O?gQ6c=@3q{ARjMJ{;4>kT
z&gNAEayUxBT_Uoa5$0t$*=oG*b^k-p4)L>AxL({+U~4e=yUlA4PdOJOla36PyY=i(
z;jtPo(r$QGoHkK&=gW&}KAYd1n!P<I$Jlu?yJ{oSgY!pEOel<|l<me<T7P-6gjUBi
zYVw&u%k_95p>xKC*)5Ew>K-Y0tLj#fwNqZ`?QHDPMfpk&91)a+Cqk$b?9kAFv%X|^
zYrVaFRU-{d|9dg-2`hUMab_mgdCii7<$P(1iMbOhi3wm0_v^_v#(yGrk_NFq`G-sS
z4-t+mE~@kL$bvVbWWBO7i<K&oGUlB>(lbKbAMeQ~D$H!DfSZNR?A#$apB!Q57>!~n
z>DHdKmPzX2)|Eh;(yW9@Lp!*v8llS;3P(8JP)olVUv;^CGF^k4Cq)4UzU0Ki;H#ae
z7~XWf#*CPc;i@=1<0CP!Ab!sZHZ|9wAUdgEMl5d1S2>3X;1kn9sW0+#jH739EiawP
zTQ|<@d&GBlv9g7a|8(?N^Ykq)*j9T*)U0^B86@aq1EBHtdR>||WNn79@;s>2@j;7f
z>D`UQD=WxF@W&4wzq5KQwp>y+b6|t}46l#L_myzBi1Jp*r9STG2-)J^PINAOVD{|J
zdQS&vD0lsfy~W|VHPA9g{YYe`xY(1=TV%UO1%GBb@w*QJ7<02G0Al3;@hPdN!kXfQ
zzLU0;hqi8>RYoS&_3_L$?B8+*ynMk!SlO&_nIyt8o>o$(>r?MN66BnxtF5oEsw(9~
z;McJ8Q_fvaS=q4q!rrf?u5QwmyfmzzVW(Dhcz9U!_R8txqOi2NXJCNj;Zdqxv-{a%
zuz}|`4@&tB=oPphu0FpN_B=t37wdQvZvj5L+A3{{+vTLE?;=q-*_rh`SovdVcLT`0
zIpkY{G`089>AdfJlP8~FzFHo;R+-_*cLe}W0hi6VvKMb-R&xFr_}*Thg1!WVRlm`#
zJCYJ|f`mKm%m^5C7W{!$f&x`qo)rRXxJJS&^YioDX&v7^<0Ahl#=kXXqIBi?3bv$o
zx;alZS?;uSBxzztKB}d8)6>_-%FJwBI?M4Fe7f&cl|_AheKV%V=4gIr=WB>!9ybfi
zRvQNO)zOHOH^JKk#M06dh}D&q6`5EDJ>To&<`C6F^+Jtrxw$K=tIfAJmte^IZ;WB}
zy5GKi+u7|)<~Hy8{HOu8E$F&40~X+~h_R8;VwPkuM5}n5;F**u6w2%?=yP?J>3j2s
zYuv#1y4r5OnTS!5LI{Se91!)qIbV|wf1;wHp`ohk2<GpElUuj(J=M158kX4GbtB-g
z)Je(>ZM@hlYz|RZRb2|}A2(?az#=6j1<qkG9wl%EdPkJRn!GPhz^_LnBrR75z0<XJ
zB)U}kf769d+p#$ylXVWuM;jwi)MEO;FBp=hHZwQR6!X2Ab)6S++xyC~l)>*{{3nLK
zN36G}2ev!c<Z*0i(CH|c{^$`X&XO%GERdAi38erpHq$l5y7j{(1m4A~D;)V7Nv=17
z<ri5s-+QU_z1=Qj-uP0@94`I{`E_AByVModa=w~cQSk!8o2rn1aBz5-$gbzH(28Pk
z&-(;!?&Rd8rltmps5KG;ul<JI=E9<)dz_}fzaWv;kskpBGrSJ|bOhsbysu)Sr+;Jd
z=?t&IF@eKC)Z`aVV(;<!S|+ui^N+H!!3}4}sA|vC9Z~O#I()Ra#Kh_ClSeEX){|vz
zB$~*DHvf6IWn9SZ)giu)wf=c*FJfsLcdF6-@L81b0V2`B7GBY1TdVo*9fhdZS?#>f
z<wjnn*S^`D>%8~cK<0ddYw2(sGb`&%gR313c9Q)jY|ed+vU!op6ImV}9{&Ee%4KW3
z%^!n9zcI<K6SU53z~>?y(*<2N^D=#UdV5*e*+s!B_aY9uPM^!WyPubvbbtQ*Ss$RQ
zuTMU)H$E}3)@GVFxT72(qbrO=h(gqT1v-!+0_^5{%F4=|=6}<kNk@8JY~)q2&bv(L
zjivO?O_v#Ufbmxtbp!#V-YqA4A`2T^aaGm31T!)3y`qG9lZ+XQ1vVwGI%`+hH|vm{
z!90VB)yF)t-w71+4gvHow`=FAe6Ji*`)hozFWTEBz%~^Z74=D*w3rWMRKu<9n$N$F
zS?ZM;q6l?>bZ|J^;Ci^NO(g{D+i;$;=G9Jsb9gJJybQg5E+pg$qM5>k3lv7vVI0l;
zW<A6WvpbIUV7oHX)2}BCe4k%)Miyyo0cGUq=osf}stpYWw{eWB1D{n;z+3}2@HxNl
z*NgbM94Qn@Eq3#Z+|C~VbU^Gj_xtzNwYA=gwA9qp%*+ARkbbC%4eEH(GU9T#g@c0w
z!?5<&t~qqfQUr$7($)s%(iFf_FV|y8aEFSz`u(SDY;25-i~<7ZO{a4vzEfcJ;g#m+
zb6d$gRvJD&%^i#O^}2Eitnhj#bM?C`z8~tjnV&a#f87kT+9ygo0Ra*~wKe=;x8;_d
zg~cRY4uV7tm_|V$X6l{z=G=0|EX}4ln3&FhHl@JB2=iI$486WQJ^XA|{l4Y;L<elF
zVH;}U*iQgaX)5>JA3vC$J@cnjFnDK-I(j9tT3+gVdZfPUE80@)nyT2b>+UnRDR~~@
zzLeODsF>a+=d;x<GyK@G`1|i)fbgR%dbb~j|LZJ(WN;*fK>JOP$P#EzXon4NI5SU#
zv-axL+E%|WE~;5VfCR;=3AYC5Qc+QHI9ea#Xt^R3cXT`evER@2*raw^DN{5(F%iQM
zXaxYY=j`mz`c9H^$c)=^L{X}~hDJhkbX&8y0p04xNS+V{o`!#Agt?j7_1SV1NTo%2
z?R;q+i$98rObNJOy!e|Y&@pFfuzGR4bv}2a+_P)Xf47(GIZ=-vKh{uJ*DfAU7jku+
z7Xqk{i+iW4T46a>z*17Tlod=R2tz)(BqBFhoF<Z0RIHiamX4xQP*5<-S=Xtt7Vy3C
zdh@0WOcL&w%;|Ml%i0Ta6S(k#!TwoK*F;oLB4I+8cvgSO_oUR-$0o~7Vq;^mA3T_R
zv+XQcYd3FOI}MWCc!^$v)dYLwqO7c}&-Jk;FGUxbDu^6L#$X;*Gwsy|*Y;709=u3u
z`(??>W^GS?ZnLr8iPuy&$ko65-uja>TCqUdIPz2EDP^*QrBdKb?5#HG#)Emp$H#9^
z*GAcOK3Dr*MsxvyYs$x}&84N3ZpRmbU@RPaudTh|SkKJN{5Mmq#dcN@g5>Gef@l<J
zMW?0~;z)oLOv}Nc^e4^>6lca@`h2s_v168O;8OtRBB*xIdLa+6vEihjur0z}xZtjy
zo1@R03DOk-XyZE;&p;xZsr)`{vU(ZMtmDK_aejUdD0x5nnKUyCiw*BI_~wjZ_2}pb
zWT<k_CNO6a5$~<>Vn89rW?#&bFX8PyEnp@H26}RBqtk-UK6&Tv{df7W4`2uX!W3ls
z0DKXTu>D3h<P{Ya<>X94-2t5z`R~~C%gV{cKl{x!GgWP+S)}z5JCWQEFrZ+%;yCnp
zWcr~?(G`MKq{-y`6r04Sc71Kz%;pk*lX$s}&<G`z((v17*Lqz>#k9!^Gi;$+;CUR4
zX{IG|sk0{mB0z}#VC}VzM4zYP?p5MVW#ix|ZES3;sv2Fi=THB8ez?Y=T@j$C(g@fV
zq+!I$%Dd_NCrSq4{N)dRBOJ|4O;-?z`uR%T3RA3ZUDbE*-hnUxe_9!r0Nbai=xVP6
z-#rlROH7QazCMVVZP?{vM{xWpC7!g5jEsfF^ZA7CUXGR`Yc>$F+mz&HMvOS~n^6ju
z{?qn_ti2#ZgV;oYH1m0b^o1D7Q>ttd^0}%zXugElfqluGD*!pd{MD<EmqHqaJOTow
zw#PX0QtGz0>_UP7F(47|Ep|-kxl&?x*WPahGXk_%@3hulJOak^_pi*YLLbjKt^>uD
z;}j8f@k|EkoGUT_D|W51&P_aZ@K;K_L)K$M1*pb$iGI@s&}c@!fY1#KHHzK3fMO`U
zINdFMCk2`O!U54vh@bkR$qyvYA;`wFExx|a&M#9XC<qB1fy^2!)FdG${#}t4A3s)M
zCdXbLghNUQLiW%*Aqg%_5IOmS65Ii<>2YynfhffQ7I9?c>LaVg_CSy}DyO%5{MpAm
z`hZaAdCN+Y2`c!8<e0QHq^NuMqP-U&Cyl~*J6i2(%lHF9KEB#FVUOzuca2CeyL*}{
zDjc_`s*%-~)lYj5fJ%I)t^KFAroX>m-Ik^SpFq9W+{Y61U;%1fe1ogoQ2@O+@X@*(
z<<UcU=a<%pa*(M<L&TLwN6v_9RlbU;7<#C|j1BQuXphT#!Fd5s?<sg(1fcAr)mIUM
zEX(c6yjyW)8#p_$p?@<(ddtyGo!sLRSRrO9RWF*p?8@Vg&j|yvz7$8S_NV%uBS?6^
zckkmV0Mr!Y`$NslP_(Ko$I$SFTUj3a0fBG|IP_PbHtNph?>S+IrF*3sks}+XXs%!}
zipI}?sD#Tyo0nse^85f5mp%d%c(8%M*|UW%Z{!M~FHL5HUn1TfcGnljJ%{sSHC|_X
zi(q%s>7n`t3=I{QY%Or)sn`!qr-?WLKyBJcp%tItz@{Gs1&-9R4yAScQqI1zZ`^Y;
z%*<M-dP;^&c#j&Lph`Tl*(zW!WObs{m3E@8%TI(#wXDy|i;9X?mxsEesLk`9c2haM
zde!DMAd(jV<(NEv*85`1z)`q(i)_!70OU!npw%~|oaQ2{fGX#~Ja+<(xQ*o*Vo%FW
zwdcx=(0U&uz3c_k1=AyOCC-fX2zbLXnqa$OW%>Ac!OwVj;!X5H6~&@iQgo7=JWR1D
z@rMF3*(i6~k`NzXuge~(i6W<0t=}~?G$hM1qhxM2L2+U}k^}KN8j`ji`~U<fd!k(4
z@(7w*JEedl9XGdHJ&yN~$qgXbndy)oQ@e3*V-X55g}#UeAuleVWQs1Ge2SQz4fewN
zw-yPlzzxD4vl&w!rBsT;({%Ra`T_$3Yv())9pQDSNwZ?V_8lGTbzO?NI+=?`N8d%_
ztFrJsw|bZ+>OJLMj}IGln=MN;OIA>D+~vZ!f8PP_rbiHeIC%Z!`ucPp*#%M?KQHN)
z*cKAj&<_RVM0%>#)zvjdHDg{(9f{xmll{jzqkg^Y8y?sGYswS#Okww!W&SkPSWpXq
zi`o47X>bJu$eI#^mgdz<_C%|TqYdj{wyij3v9}$oI4qA0`WqT4Jo^B@0-0Y4?Jqc%
zSGA-Cen4(+r*~E5I056{4%b9=h{d~i#C|{;_GOA?YVsqjOka%*56=V(PWtS>e;1F-
z%5r#*&u$LsSkIcUIdFUVp}?lvvkw{Q$<pb0oR}6t%4{ReQJ=X8$}ZRH-dFXR{jN+6
z<He*nn^(^G-I3M3h=Q83L{DRl65B|)oB&$kG5Z`iu7ThwHCY3;7a0`Lj8qO68(V}&
zA4*8N>6k~vPPIVyMhbF@iG@XtH|filuUG1%aw}tjK!dp$94ou8_vVSL{=CUW=5w?^
z*JD2=%d|6w<{*l=6Jui!K~*`Ck|f|Lwl&k89RFO-i@K?}Wl6)<*FF@{Jm}g#O+7;`
zZq7|%HTzw^)b;jB_g9jxa*g~G{)S6E^PSN1tMkJi;a7-7dMWpV*k`=&)Fup;mzTM@
zxu0`ke0`t!(`i?Yna{d0&$&Au7q&I7a9;SrAa254!={+<w7#?_l_55NY77zK(adDF
z#MfQ;7E9A+;jhTU(nX#VR{qP)N!m#E#KNgrtLE9#pUv_JY-hVg_uXs}sQ$;j+j)Z@
zE?<)E_!E11jM@WItoXN%Jx)83d6qt&qT5`%)3v)ml$w-!bHVg*JuO>GTF7vMK7W27
zM%nW{Q@do~O(CQvi9S&adCxKPd}$7$nvuHXk|97585RlChr!J#s1NcPH<)F5NkX+3
z_0`HI8r%O=AbnhCpp}#&9#vB_EbNtiGtb6vQ6fijYV79qZ=<UBwiSNbfr8D3$BwbG
zFlJd9Ry+~cW{5f3StLx6m3G_MERR1SgW!+T8*sOMNj6cevo?}X)hsEpdZGK0t3hmP
z<y(pfG^acv^Jpn4iQ|GaXXtl%(;r@SE3>t*PTfigSLy?^*%$gORCT)Sr#^E)ZUAi^
zX_-7}&FfSYWy#*7vHgp#`=;m3M6wav+sI^9_hMF@wMaY7q7q~B2&raCctKxJiMBPM
zrdVlh$KVMw_OFS^o$S#1?CeiL6=?l{2_vaQ-rt(@i?)^=CB<7k|3X*^<iFFDkjtj&
zi)5|%yiv90w|zTS#+<2qb|oi@)oTtM+QrqdYAWvT;KY;d=~Hm>QQfmrE|Gobc7>m9
z2;n&9!OT2WW7A9j&}qoniMprc_}VfQt~53yBoh-?$O?E~&!2+B>Ex5b+7VC9X4<9D
zPH=d2IKWT}m|Z;wOpByhWcAZ0)Nh5F#k}DW5rLDP6huTFEV_z{cnx~q=ZHk<H*d=K
z=})jj_Y{mzOx3)i^mrIz|1<@6YV<B;kLwjCoSiG;tq*2N%FD}FpEfvew0}BCUxf|%
zw@Vp&5500F`5_69p?hve#c%JY=P~2iS5#4eU23A~^s=kzuW2b53K{+YRdc+jyL-K3
zjk;W)br5msVV?-s#Bn1umDk2CP)(x)l*Rk|``}0!WNCRBo@%_q#Zp(g{%%X)Au+LY
z*CS1)TkGN6(9&rXU0!{wy02fK!hS!gsVY#+GhvHbSTH;jRbz%BC)?<iuNQ&N%NZ}S
zZjO>u_slCOkWD@XYT7$4iQ4DPh@HGB^O7|g*fyw`4<5wOh3a7!-Be9)&-mVGLuC70
z3aZOD{&c$q7HW*WXH8hk^qnISfzopTgJ`<RBzI!VJx~sad=6(p!0_9m*GFni`lYPt
zY%jRvVsbl=Exx}@?MFk$Q6@v7&@BAHquW0O)%Q${kN>t}2yOL4fjA3NPCsuGfpSU0
zX=rGePm=00lY0aV)YZGbAC#l@u#4=(wrsYXwy~sny0pDNUtC-SoR$8i=*F}~wTC|K
zZCujUyF+9?tfqWlYhHb4tX(GNrSrP70UeZH`R&^fPb<G=5vUm#5g66T#-~r8N}5jI
zs}mlAy{rGXn1<zAQW0ch1?nsbWBlU>)jKJkyJ@9B2+GB*WF%?t<^ruF8<lLrsC51U
z{wW%$rNUd2#&-1%sKk%zUzi*{`7f*|IZq@6@Ikc=)vY^n(_57#(I}n3*0Ruj4%5}8
zuws>xl472Ce;t7aIw87@ur6jEMK`xIo7!3Q+7x3@awQu0)R%Sk2{O49Tf&uho`XX_
zoAbclI>D@NGhA@#VxA|Q#^$Y*%76BH&LlD<WMPh|;8^DY9v-y#3#k0dua7O*?_BQ0
zh=~=)`bm`rP=Zeiv6@K%7&Al;eRxS0F>@N3p3v+w)73P&Q$9(v1t!=ZrRVrmeY^+C
zp*4HI)+$2pz86tt1*J^PPnRu?;`_)3bO)eAf4c7CEr99{z*$(B!gefWq06Z5>)kZ?
z<Ec#FzQwdMVJ<X!67dVf>ZL8y^oN%}2VCRnWb#JpZ)+Zv+V@m5%cueqB$h#8GP&P@
zNe7O%B}#^ap01eJX~x0&7<)diGF-}}-%^B$>23bgv(*?&+~;9k?FY$=sO;s+(md}g
ztJ@rQwZe))d4NnWE8`qT8YWAXPTUzOy2fqAlSuMImzM5p9Y^X|a}lVR_XCZ#lM)c=
zz`!v^d)MQ>6xNA!7KH2G8@DMQx40|QJcB*We;IO?eHIiI)z2F#UnM?HLL(CLoY^I)
zP;&<o4w>G|-p~h_Zhd{d+52*-y{p60-5|hemn%3RV8&ysn1FyFM>=X}BoDuvo_`XX
zUbFnDZ_bnh9p_1_E04uDV|5Q^khnols}SyBJ=Z7z3PoC4+L?LNP`GjV`msmzz?@rB
zl6$ln{Q5_^R&OGQ`7#Ow2fZVLZ|*l2g=ZU8eLe6HH*b>?Et?)foRO1UG&F#gu)L_z
ziNd#M*=lEo3pF`T&=|*p^6MB^1F=b#By94JEc>8NAPQIiruVvfO*i3o8<(rarGJxk
z-sHlFvkcX)U+nxP520gV0C74p!-9z*XsxTO1NP)XJL&f-oU<S_PLHDxC!gq7cV-(v
z61)Z;9q6PiipKR@I^_(lHwz)d5UKCAjCV%!iTrfzn&O*CfiXZy*^jDziQ;Zo)+8q{
z@9Y~(zY46DjC28X|Bt)*s>RrIU=hh}>$0Xs`WoBI9U{Rlo{S`;`_nKng&Zfv3r1fm
z^OAr4x%8REG7{qCb<zh=YOwJ;`dKhOHDERqV`EUxEag>JXm_MgGVqFgPQQ946>j*<
z7wNI$2Cdi~Z0atW*s^On9tWcTFg*-XD=S|*ak@3JTRou<3|~@m?R#L0)%8Tc(g>aI
z&Qp^J`p#f@!aEpa=s-OH+`;hK){ej|YXfKE$tSHzTqe$fvy%8Cxg#4A5)zOz<6ZFg
zJIAFY)3Ua~O|J@g8=#am4|Tt}Ixl`-{UTXDIy$;e*`Vw@eksQxkeRoK17axQWIVF&
zYI!5GK9}2}p`nk%x(qq1tl=@2$|s=Kga34~LdaEWu>wn_<h-i<bW)i&^~u?!NmMzP
z{`kl74QD_)(v=l(KIn^NOCOg+Js4@JJ<_$C^t2#tJ4e4XDm2t$s_J>EJ=Ky+a%HB4
z5o2hVaWq}%_HqCgQOZjT)UUuB0#ukc@@Mr;WoabtSp+$sVhr7Qt_l6V&n`&emmT<G
zOtss%=#pAmlVHTufCo2aV*k7j6@ZAGuf!se^>-K+0&a+KFExlLG9c&K<>Qwev93Yo
z4cu@Z&;AQ);2o0l+Z!4iFa1!-0ab=#9{B6mD5`p+%dt`eF_{<!#kBQ|q<S1P;Owrf
zm_m$<j69lrMt`U%q`cgiD9zL_y>ptoBgpfPlIp#Z57!26k0%URn3?4n)QpXdK_2RZ
zayN+euuowTF}$>yiqjPOT346E_I{&zDq6o3)C->S@;Ewq#(?P>2@lWqm*MRjYTxB8
zhI3^=K*T{K04**bI<5943%W3Ovll7mr3yHGSIh%=4F{_@UBg!B4(HA(`nC4xuXprk
zVDTj+b#!#}Lo3@14b?aV@WkrGR#K9Zeo(w_(ib!C3<0eIAJ8~hVzk+BC*&^V128!_
zH~`%T`$g-}$jA&%qfbL0R;nw0_9A2f4@E|5)ZiO}vm7yW<qXeC$yhl!>}G$uJPtDh
zwZqx1c@;P%RIc^O-yb0UFBSFIEM#jrXa|6thw?<wnW+$4DvGK&e7FodzG-225>&Vg
z0LbvIg@uKX5KIJ4WK<MB9v*LMKPWanM@6m8nSkzRWef@2=XiZ+Ig~L2(n95AX}QUm
z=|xCP+?lclX>V`8v*)4xDYL*R<l^E2=J3SU@mSC&j^tiM&+p$MAt8qTE`XCyXPgG&
zo~uF1e5iWq8#hC;vuRIE!epO`u(K<YelYT%w086BEuRKj4^;mF*l4^!vP|fEva!Fb
zUha;<u<xdHJ=z@Wz0?A}CP<oi4;~ovA#<{`FHW|5e^G&YPEJ-<3!8_50Xq2g6?kLu
zA5+Wenr)Ed2q`Ei9zEg#9TX7O6D?)${L8qndRR*DaoR7m(ntm|oenK8ExoU?ws=Wa
z|5E5)9tJS;d~bbf-bW9Yg@lG0<%u-#4dqDV^UjZWPqvaP=EX)Bc6WEPoF-|vd}>-9
z>x?#Y2iS#Z`RmB^S+}r0mV5j5?d{F=ZryUEM)7-xP^KsJt5X8$e~Wc$@6?g%l!r%C
zhKTocwbcaZ7&xUxxSZ`7on|3Hn=NIm1_FjjLPd2mWmWodOBPSITso41l!S!OxD%6}
zD^;EW0V7WA6t<fahJbZ|;vX~b^4Uxs{T15vIv&#m-5J5KE=S-cIh?=HW7h=s)#=Wx
zHZ2x5b_24a7eEJ1JSG>ITl1jYv$o|;v<Lm8BJ63v)}S)RZhfSw%jK~2`b2Zw1eW<}
zhrV#E=OT`k2m=G7GA{{aFJPWegANFUB9CZoZH?sd<C+gc##d+iYI%zHK`#QGS2urR
z%ky~C>_sx5DmV4Ip3?SkN}(Fy8bT}KA6^Cp1>sQ%Pa2mWZA}bV^F~pL{45DhCAS>P
zegljRh*bEKwnzf<(fpGcfzl=-_#rzJlN%ylQ_C3EvaUt05|hgXdgo+3%pm124;1pJ
zE2b+>fU$!KMNtS;0DT`r2a51FQc@M)nY*j0Ck^KS=s-YV;8FaZGj(yax8DO+91$U*
zgruZE&Ejj--Tm{PW2Kp+iT;)r-nWH(sFQ>gvQg48bd~TKu;=NDt|JP35l^^2Ei2b}
z-e6;4CBL*7292~hSAcyjl7wgUOQ4BDPEBo~tLp_Ab_bvAg*EP*CjHDWUs`ui>#Rn;
z<N5(ve9%J=T74m5VQP+xng-sFsHlLc&I|T5wI2#14tfY&(?m9%goK1uZ+=87&VvVJ
zyw*t|qupF=#lvNQM{@UVjm@;1f=yvTf$>=8&<AU<YM^yNCv!UI$x#@jl`^z-4BTf;
zAEDpmc8&YLK$A*KEM^;(9NT8(dp{U7P_Bw;{B6T@-BXWA97EyO(B+37O*hHr8FyV>
z(1J-%O|>f5stW*x)jKmYGXNbh=(ABZB|Y!sG0^G+8vKcfhqHvlrvol(JrYs+-im%G
z3kwU{*Qx^)pa?2lSnl4Cd3%M?8mlDwhlMG1vJLu2_^^)~Sj`|m(9kT_uVt9Hc2jvD
z4S}wyV`!J+Bi=+c(ytG2aU&ulZ%$ioXB*v}WMZhi_S(43|DJ($2i<U&o8LiVzPG!D
z1{9Pc`hj~Hs)SC!)dduLceVj|mAo@##KgH(@OlcDnrM)q9D&9Lj+JY57i}D)lv_0#
z4CFG{_WRgB#N_0?8A3?V8UQx{Y`lHLrN1~Sb>!h~YD!1WU1R}lIAELUwx{zuNW|oV
zA}B#(k{b><CRy11MPVMK%5uzdvdpLlf5fBXTBZ!J#y~1xRKJtaq%3v?>s_djZ|nc6
zfLjA3TEB;b&aCL!f<MT#wScrKyFs?KDya`zmV6ExE#>cvd3;%!czMBr#UTYyv`v)E
z;_}TXA0{@iM_fL51o<wXJPhh-po{UTgp<@efzM+-`;%ur?;JePSLV7q_xty6{U(nw
z4{X0>Haeq8Wf%;`3*kMCPJUt51BNzhzB^&fVvv%k?_qJgIVPCC1L_2$-F0U{UTNv%
z*t1R|CEX+Q1rgwecS<b)OSh;9&4f1<G&XkTejKP-(m|*5*$QjS$Y>>bgmWYk5Fl|w
zq;eb(J@G~L!DL&Shh(Y7k6!EQ>Xzy^%@n!>dAgpD=Bt5z!sp0H(7@+v_yAWm5N$t+
zjCe72oNU?BwSudiFpws|4a!Pj+_TsP1q3j4^y18$1Y=(pfY#2$mAtWW_L$|dC|83n
zyLJL_ZsINeAU_JedG~G<xC<ygp4};Vz%4F3E*0?r%a()DZQmo0UGqv~5h$eE9t^D=
z-5TrU(_>F<Z8+#bDKMx3U81h32UJAMPf_4y4ust>@kr}7%LC5=`s31NqSnrUSNr`?
zxrsD;f23W@bxmXAj5qF83aCc`7uL;`aj>zOsR@EU=8jKh$TGz5^XS{YSJykfEt!XV
zTSDE3+Iqv%!y_Xg%{-x|E;TiqH!^3|3D3@ma-a|p1A_!Sq^cR%D?;H1+n#bkP7Ylb
zyVHVxDIhCj<8ql^;B>Z9zDC%~Cvmn-ls`#G5+MZ*rh>8!OcHO%$KK>*&>aS6Edm1W
zjZc-BC%w8Vvfe%Yi*PJ|nO~Ilo<!fXJLu)>n}gtrW(ASV#j{4gw^MrA-v8^VC8(_I
z=iTvpB^xKxcA5!9`H$vD$AFjyEo1i<9;J3*FabeiulpnG#vXE4eLxMz0(a?G{-}jF
zM(VwN`}?~>Dmd7HydD>(Lzd)KNCf`2zdLBK@Dam{GK|t2zeT2{aK=m<CYWAcUT*FE
zDa>c`G6Xpp+DHu^*eutJgAqBY{=2IA%=?nswoX*93$8x}CIRm_a`xddkD0Z$IO~ri
zGyS8}PyswMKC5wLb`;~j4XkW>4Yj?TTA_a3waazoDnBnnbsTZR6Rr92_v>uk<yc0g
z$Ccs_gkU7wTz7(>ni1jtwjop-J3#L(6;!sF(z!SgZNewC%*EGKyF`Gm$F+PCyT&>e
z0x0#PISFdFeqtHSFF5s}J0sa~=(*M9MBs#3Nk3I(CY~9Fb(lZ45;JC7ei*iTduWl(
zqIOjBL^+YWK?H#l610shMKQP3GNRHB3vAHbn8=Msa)VEk1oT%&O;(j2Sd5q(zRs=D
z$zBSs*t&gEs$1cCYU?(bw=#`<cxCn($>kZ7bY3&lg(JpPX!3xn=dn`(W^)H^(prR_
zer)SDtigE^gO2^eXt#sLK8o_4eBaNS*j08_mb0$%{F;wbCeTN=*J>34gfmD(_9>Iq
z!0WxrB(0qR`H5Q@{p#YO52lYw*31rr&a$kH@_U^cBPs7iVpz{NHx(7h?+r!=cmLg$
z5)zk$A1~AmF1&H~dE`)Z6QK9zlV@LeSr!s>QT8P*+i5Gh<1CnT>=Azv4$m>IVQx2C
zl(=S|w1%#IatyH+AO*HFV-A>`xKm@d0)0edc|%9$7?r#%VRC0hoTeVUHZscM4X&s+
z#QuyW1Va&yV{_A}S)i({_?h|JzIMhjEOB~>Rkp91J6WDm2u94C%i#7M{gQ{yNCc{$
zVY5(uaPr38Eh<@)^+cl28$~;zqrDw)J;Ry6suZXt0~Z=h!WXivPby=MbFz@)9X<^b
z@~p7TgY1D;dxOUJ^8D&S6>sK-@kX`btY3aVwgeB01nXcUv#fXB=g3ywXfSONdly~2
zxY@vmP$J~+mM0}AXK_^F#*}9S1`#qCb?Y{10g53F1|EUklIJ-T<p6IO9#*}#u3TeS
zz)DN!#*|9vWYXzFp&-wSpBaeIsmGSn14VsXTN_zg`JY0{rxh0_EgUS?Ogb^S>+2sP
z#WdAh4h|d#m3i?q@YVLG8ZLal{UQ;T>3bPR`GNIu*zX=a(>X%nI9&0(5v){(eoUQp
zao_=YzQl&;n8};s@|*vqP||=P2zxE7@yL0@`}qqq7p5Q5W`@Yr{jakCEUuAZ8Xj&?
zM=U<WR)~@Y@#<s`!AVlK8;TI|LK9#!BQ{rcjVMA=l^8lBcv!LH)1pM8M~@IU4FSgs
zs`Zs|MQs>w8s~-bMmFsEsoWM&xQI@*K6Vv7!;K&BIa%R*=pGLW-M$VEwLRq;v6H0d
zToV=`i?v7oSQyLcJWUdj@6&B3#X*mD(x;O-FwISzC?F6_FY&#-0n()91HZl`T;I0S
zi8mFOYQAiZP^EOD<!D&0y*;!uvN`zlX`L=678VxhdCb{$J5k}8VKZU50(UMhDambU
z;B3RAmmzi|h}^bJi?yBBgFmjHC1D)u<3J~)3Qmeu^M-UICd)fn9aD~g?DIH5K4(pP
z%bwWVk=opt5{~ZQ7CRDS@*O!<%~Gmh5wVKiNygYFvI_L}?Zs9p?QVFA{C7|Yah21_
z{M`Au)!8W_slkhIL}q7a2jP;hUc>4ReRj#>lA0LZgvB|v6F}d02Gp$hVo2}K$;nBO
zpRusJX+PsI64FEM90WE&eF{#L?E46LIZ*W2PL$|@c2s8oHXa@@K1SY%{bETA4h_(3
zp`)cW1}D)n0uCFptP^|3zR4xp&i=#-Tc`2uo3CJvBIhdsM}pdEWn<`8*H(tUc=p}6
zA;m-Ca*dHchdvycV#$c5U*H>_A0XO|!Lc$QaMZKConX%j=Da&6LPOIQYwuI*R=wsn
z#!O`!yle->#KtfLB`mOob$P$Nic1Q`Y^MsqwO*Rg8MhAZxctQPVfS^=GEbXYYTkp=
zCaa<BK&X=9E;yL&LO8>*Kfez_RUBw_(;q{(kJ1+QrNGCu21_DQ-|4GTxtv`!V{(!K
zJ&l~0toWUUM7If^F!@Z|fh83R&pb0hDDqn>+ibbcKHabHlw3NjosYmc)BDI1Nm!|g
zDx{{G8RN!YRtRm8(_2!>`TR__?<PQ%M0yli$JDR*^q|}R6Bd=vhrxOUNjS-y(B#oj
z)x`08C(68mYGLzwu0EOzLD`+iN3wc1z0`>dgD+ahb;cd}cP(HCBs>NG4zUwll~X?T
z;U85hRPkjz!xO|cGa#mimfnMrB>Y0V(lqo<_7TKNyNGLS4>kB8&j~)CuFOiO?W!D`
zd=l03BSD+D?7jgMLM^n<#1$*35~XcbB^;;F{Z}b8Grs?gAz#RIu1lhVn=2B^hOl8I
zn~iTDO3F6Zh;j`hc)b*JS#SkfGo^S(Rtl}pK_axn0UWF!4GpTO4gwDcxK3GX>&In2
zo|cSDy2euXeJ`0Gmg7g=GY-vp@G`DgBrZuN9krpGVlw$f8OG*fFt|%Uz0Xg&ynpwB
zXWoOc(~x|F%aiVHL)xzP9iF$LUcLhl6;!oTr&7L*+ipSi)TmjzayTdM#}nI=0Q)g$
z$RS<jO7aa+4{c_b=kakZz}x|Y<uw3LG3=A4P$D1A_w091B|O9@1^)-gP)X~~SL_`%
zbQs1gwS}|BNhD>H*bP6I7&SWDcLo2U-}F_r-<E1A#c&y)94B4N7@o>8xj3vu6;tdl
z3eSYp*>6oOS5GJ^CY1sC16l?sv5cP7N7+9#hMsPM#XPq#8Ikl!k$e5%%(E{gwf3qF
z<HLOOWurR+4libAG2(p6g3YF*-=6T&GfH3~^Y79paQa<Wa9<DCCNd2JRvMNVM%@t0
zI8?PU&No+A-{_?p_V}$*_eZqN#Ls-b!mv+w_<eBJlOEPerTI|C3uCrJ>GRf#cJY&!
zn0~;Z>3I<@Mf~`JbB5l#u7vG_lOC~HlduI->KC~@c$@x%H7xv&P%Dm;?XbMe9U?M8
zlDER1kC(f_Lmn}F-}^aYu?M0;#A=SgbHKwNhTO!T<#p*ybp$W>ktY5zMW;<yq!0{X
zDQ&FMN@=PD2ZBYw#(xMDD+wi-tn!l6(PEVvJo*vOvX88!5fIq+tMm{FmkJ2fb9~+Z
zy`)utAOO3Xu(I-~FAFXIK5u+Gk<yt3qtYHXC)ER_{h&Qbzb8<#QM{@uA4$1k;J_eG
z(oEHJL02i;l0>lE;a2=c<qNg(NuY{~{<#;~D->}yfwtV>LUCy1`0c@)rx|-Ax`eH@
zZ_paB3*k-6h$T<PUVDDv6y!{7p*&^8;NnVrEa#P}>NJpX&i#nc=}%8$rhwxJ$7QYE
z?Cpe!x}8V^(=>ha!`z|6k)HGrMq7=OSyz+IccxH=O=oUwZ15x=j%a2k%T1ec^wrU?
zgtJcT>C3fMhGn}ga?~e6GagnG>Atr45A`t#f4rOy{APbcuBbbnEKI7RvOa22IO`*=
zO)4SD;VzeQ{o<M{&25Ru$!s;bebr{^81gmDW=u6R*&*R)U&>0JBz%f^lF$3vVWsia
z;=S$Jo!PSstE)4U(z3GK*KZb=qsqPNOHI(P%yy!xKkuz2nKjI!IDJXh(~1%JZCVSr
zyqrs5*SkBxYiajAH`jT6ux%}nju{L2K!VPfB6Ae^PRC)}51jPyJ~g^J6ZT&m2*I3z
zk&9m9fWsOV<T<I$d(&Cgi=o;NLc=)|h#xgPEJ35_dV^SuCaCsy0Zt6K#^%zvOr#VQ
zBSRbsZ;|L0un74x^J7Lg-UMi{<bBG9CF9Ots2|`J-#Qk>y@Yw~nyc7C#o#YJ3(o$J
z+P*R@s<4X|l|~vt8YGocT2e{`5fGGa>Fy3iN)Ql`l2SoHQo2D(T3UK==pK;nyZOHR
z^Tv<+-1!3xGjq;)&wh8Tz1E(2H#UFvsgVkc(0QNOe6*iVj{VK!TQey`!y`fTc9k(0
z{kjCU*G^g9Q@$;3Y&qTq<&t{-xR&qP1N-{Hu?OenQk=Z7c!>)m9`$~T{TzmVTg$8-
zIvSEy!scrZH+(N5GP4FhE3ngG93*~kcZ`+@Hw=)QZPb^RZj|(bfhoqnZqkR$E4mnF
zOR5_e+Ulo#$4h*I>hn2#=97|z@;~?R=<Y$cM2}Ogrx;f`!SuEBal*WDaY4O4#>BEX
z3g?5U9!!a}+O?p^*d69mjHfU8V1lLW=koW3fOH(@c1L%bBiZ;ST0c(VGz(7VW8;zB
zw@(v1%P4i~b&$~(tTQt#@YBW&Pg#E?$FvPRwNOQ&6!OezXYFSr(Um*K@D&O>P*%TQ
zH}Pd+_4&n78rA{g4ajv6*hhNOG85ZV-7_wC>VAML#8K>(<v@>CIhohvv&PE3^_!l<
zquDS!T0$=ot&!*vIAx|Eby95sgUMdc--wS>kNaT3J4DY)Do1T8{?Lcgm!FyC=?6Xi
z()Zo&zaN*S{3h+L6?<I_-iG!>;OR&Hk9`!N_6p0`8D`C#l*^5qHB+SOW~Hd|Kp9j-
z{3X2@&lz#9239jrQt|NEoS`w01{jbt6aKyJtL{0S_?nq4A?3}G;>MG;ws)imTiLQm
z9EI4@($f0bs8FArv2}fJC&|71iQ)!-h`x)iA`Z2y*^rzX6KfXg#HXZiuqMpdQ;=f6
zW>x@(8KBoB<byu=R0K0n{Q~N|HT<tS&TOKmb@>>vx_`o@Tj@y}v2!MPB|N;M0QkwO
zKsRn=WaRz(n><yS6Bv&*iu6{^IQvy@We&2JYFzt}mZrZ;5p)&0tDx`&)+aEy=^r3%
z1*d?R-%!jX#;7m^aMe`?c?}>X=uW_;<Rfgeq>}#ul;ypYY*CLhrrN^$jadQmqBTI8
zg^*%!VIa{w{G;pe4hQQ|f?3Y+UDkv_OZ2~H!dyJ}@9?UBQes76X;LjfGf}7Bv!pW&
zk=eFUc*QPn!L(3L6RB{Te{77TA?S#fykKWdXg12t&Fuv30;zIr7Et<m8jqy4gwa62
zYtnKJ=qVhWoO|YY>6%;K4HBZFr{59{{_*O)J^2U#Iyq_7Qxo$3-9aFiGIFL4TKb|I
z4(r9-(?eg$Ox(fhuld~88~S7EN3#(TCBY*(x!|Nnp!15&f(|y$lqfJgU%Xg&Rvoew
z+m|AQSGSIIt^pQ!!*tkkyeN8lLct4w+isEmHsx)eB`I6(2<#azT}CRg4udSh!2j6n
z0g`)jHxI6jOG6&pCI*sPR+=e+ls;C@#M_aNc~I{ST5iahVE)ls3-stH0^ufTUh9%v
zzYeed-M3mjMhMtVl05?MA*^YKcW=wmCmd-1RPQB;6JmWXEmhN!#V*M3`T|z~_1NEC
zrCu&Oy>Zb89`zq-j4E_^yqMDIJpch6xNHed+(aXkyj$Io`r!i_ihzb{h`<v1ApoQT
zdeeZzmbC0+!J7>AFhJdlV6?3zc+ID6)1S%j?I%$)#So%}(R6Dl`x*Om)Uuk8xHvlA
z_d%LP*GTOX2MOiXi@l`c|Hx%K#;m}5`=$Y_dB2>zypYvU+iA=#I|+3g-ehQpToLUZ
z1m}*e$Vo`97n8ck^johz5E6-HhP#LZOY?@Ic{!P$#k^k>mptMdh{@5~-p@O&J`i$c
zr~Vq1#;eM+5pu4sBHq5FA5ILc2Vl%PR5COUFL}icA{{*+stmXkNO2y8h{FkI4h*Oo
zf9K%mCwr5UG=aeJ@tg<H0T&mS#ASHQlE1Vuu=DsDQT*K+<|Rx@A^X%s^=ax6g8M0b
ztE;PsAkyC58-u_y&M&G}TsT8(FhKd49MlKjQy_tlhMqV;;^;ZtF~f#(>3^=Ui{lrI
zO&9ycupbS^5q;_6aJu?{oEiA;v~O=aKJe*a4Uifc5k&KDFc@FpK{GGYUbrKD2lgv~
ziFkFmUo$hITd9GXx<=<eWXVrJ$s6R|8>tMSI;akqH-3VEu?`xFEk$3UPxw$+SWBy#
z!obrQ?jE8%oBBl!K$W_j<i<Wi=DkV4DGyW72Y4`E(|{5L6-j)<mi%k`CvYCd6wrhs
zdq`tlqJ^%)e4zH0I|z)l)W;ODEv5%sH)zJb>0r_Z3b6d?C;uqG9aTX60+2jjel#@x
z{XxzWv9S<IE2^>Z=tGw&i8gZ0Rm3>09=t=iZR@pd=knIwhKR5*tt$JOpKoroD_zGL
zsPjO<XXra~@^6p(!3RLGf(RNe1)c(6``%^2n6!0(<AsKX`j8UPRo&33u-@<G>;Lhv
z%pcT@Sb->G*~gD@^QVCUKLeni@prt}$~mMpbD@dvYQEI#sTpo3PB65F;O{n4FC3FI
zE09d!TThg-q}*}rLccW4IWoi^T`GQhd{?oTCamCsK`arQpu@~0Y+3dosey1zHE`-$
zV5`^nFUT`)N_6+^k)<+=dR3Dph97)pMr`Jn)>xL3sR4z$6AXcTZ+aF*TvtAHhRKSu
zKX762WS?phS;6PBHp=1Nk-Sd()56>cDy()NRD(wse%en7qBjj5;8=nK%oKN!R3hUX
zwi=c9kY_6)elqXqy9xf{)&gn=9n*cNT1UCV<lAP42%!FYNHonuJc|$50qsxZS$>Bv
z&SBE|%D9!-_cE^@FE6jedW`Om7ZI>Fb3VPLoEGNhxOi;Uaq>cb4?{=;%nkkpGzHbs
z0b3lA%7t&|9(G2HmvGfm<KAq-I<PHxy7C$CG)9&C>Jk)&Vp*OB>Ejeg)Pj8l2ZPEe
zBT4J;AilMws1eAAAn}Mxa*So(1oRmY=av?~opy@E`rDo|Dpm@y&@SdEhy!o|0R#w8
z*=)wP3WOOH2OxtfxQ}@qR!YDSUVHQm3$lZtK-llZ)!u&VTf6njSZsX!KBx^MY{jSH
zT`eYF(Qq*$X&~D%gBW5^oFU;?n`B6g`FVKNA2c%H=;wwmPFYO5Q&Ur+kD#Yh9Rj)t
zKy!6UOaOU|W#$4-0<?ThT2@_auK=s+>Z91Ifq_?m(w9QVRNuK8(>UMR;2%POpChSn
z37Xeuz*}l8cf@ytj#Ntoy^rMz0%GevG{9q1-s79yzkP-C-@qaW!%0j}H_*|^+ex9=
zf+hQTuNYwaFdPaB3X98t%^8nJ>Y5a2+bKaoM1sa|IVWz(sXqFkC?$s>RiZ=TrfUNd
z4W0tK=g&FREZU#TVcfzpa{3|L-2CE^>30U7PNGi#S^BE%;{lz(>x1dDhK$Up1!h!(
z3(D`Xp=0r2vZ2q4Uk8}VIyyS2piEVECnqN$_^+oqg6JTZ0_7&d8i$&x8?!m&XtP<u
zLTpGCI6@m?^FS)Wz8on`^E*`UK^?XJID{uR(Nrx?PEH^aB5Z83i;C=Dy_yfzbAM65
zLu$C&^!f8=kCq+J`rh7NpnuqhuA8YQNnZE0Ni$e)nTuH7pG>2J;ROaw{xgt9h@Ok_
zCZ91P+h%GAX+_BVw3=k6OBN3f+yzt40l~K~xIe?LkNpMmkTB{ex0MUV;+|x|jmyr?
z-U=qANLGU8VS{^93$$w)f}CAkpr`N6NRzBo`QwM@*^!fxk<noS!=JplwY6{}@ADJ4
zLn@KEqI*Z!-c`D}Gf7deiSmkzQ3lnBLqpD98iJh6%(v|(SDLqqii)pa=fIn`G$=?(
zNdaD|g$LpYeqLT{E32KoJu*c3K7=`6i6#he!x6jx{soCA)z*sCTZdPqBcI-Iu9=1=
zZe?Yqvv6u<rC^Y%kB?7I<8$c2sH+bH{{ZB5(B}JtK!7!T`t<3ZhC7HR_^CD*dlJDC
zNhB!g>1Y33J${*&#l^;MF9+<*wS)M}u*tt~eg~v-7}urT-7Bjrf)u|&6H!o5ker_C
z?~W(jQ#(65XwS9s{4}(3aVcBb^n6I8Y+#VC<z>vv%ZvR5ZprJ{Xkz||fq~agm!@82
z*$Rp>G2Q6cG-wS|Vx8x)gM9;|AujiL2}0F%bh=GSwyU3*nB0-hrGDjfloAb^w)gMF
zK#b$BRy_{4VQp;<XbB#TRSgXdkWx`^qScdtP;MLMO+w?=GYC!I7uH=_`8UM1Z6QvP
zn{mm>$J<Dmk`_uTDyXJ{f`Vk_<*O<xaHvHhBO_tq;Q^%<2srll_i1TqTZrLRSwK#Q
zRg2IyJU=^YJU^I%?p9k{8!RodKQSTUAYD)cT63_r&`Pg$8he5ptfs5$14=8f*7$^k
zxmzl?^K)~b?6m_G_2T>#BB4py1ZWq*7^?Anpgrs1aR@mykl$zPz0_|<Fkqiy2nq>#
zoFHmFw5U*RpHfnI3>$q9xAc*TZ-w&Vgp|L7nuH-pM_0F}r^ibyz1riXW$d8?oQ19J
zbeV+;+-gvIhSFfIT4juXYmiiRbF0Y6m<866Hs>GUv19{z#`nmPJmMBLH8l;5nDt05
zoEVr1k52iMf_5;gwUw3Wrhq1p-2MK|4NM7P;c@8pj*T5cqnbmb;7SjZ*wf&MgywAv
zh$*2B2*R;eqx+y?GIwyeVhBYDa2J04+G2G7>hulj#}NpB==?x;)!o&VnT-v)^mLN`
zVt_J2(=IA13OKT0g}~o``|;xlIO(_W2n@V^GenDmQV@Eye0*@Jpvjzfu6c1!0Nw^z
z0OZ5JUcG~0pO>2}3<LpSIM~@0R#jCcCu>84=zsR#9g)%{jY`neIlK#wkxeE2g}VCg
z_I47aBOontI&JOivxDpdX!!h8b_<;`Im&5QG{MJ@e?k8M(EYaX;8w;Ua9kuLB%sSU
z1#JZEAjH&^?z3lhKwN`szt7Ka>);R_7M8?o7M_@>2`TRT_q`zFysDX2SD{bF_Q%f4
z)XWUFgS2!DR6CBk+pyvX2M0$@(4zb#?sE{&p?Rg@1xcNymDR+=#6wY09^J|U*w^>(
zt4>b|e&|||dW#A7ofA|=%|^XYyFgqM5fQ;=1l=a&6c28{2YdeT;RC3=_=SYVMn?1+
zeYF{gKrjrx4wfA5LiIY*QCl3A3f5X*UtdAtU3j>nj7%#u<w5odn+rPeU%-dv=H8~H
z>{@ibas4{5h=uH@K?+F+hk?Y)&~UaWR9#ICq+2rB-~J<j>}YR?vjrE*FD7;+69MfT
z)OOH2Z*X4h2BHihjj)lmwGL!@APEJ(1P&9{qA%$|<~lO>MP5_WC3H|9K;s#xn^WWC
zN5{u@GqwBBBn8kN2&O<2XNv2BNihZn20%3f02`i%H8g;O5K5jg@bEM_&HpM2mBD__
z{t+%Ck;f3|q2F?HfP;+mVg2>%7XiJ*8IY|T{Lj6iM1T~JNE#d~$Rog}lO_DTL2>u!
z(QB<jnrZ36+*}CO5UIeALDb*g-`8*S^|H0yfVvneGZ`6~Ltt)0bjZqL1}z^XJr%XJ
zCs5b@C@uZBy-l><NG3cDmIq!rTRAQ7%a<4!sz6IC4l-+~p&!f3*VWVvuDVi1f^bv_
zXd~cT;Fg1cduzNTv!o=Z-Go-m^T#@}x33SZeZtlqs7>AR9BFB3;E}F)7Zf}OFan}{
zy=PUtjjerl_7c))XraPj+z_In!9i{4ZpU@O4VT46Qb!M1^6LozMZ#la962|4=TEGk
zukZHZJ5l`8LzKLOPVmXe3AlVfmuMQR9!Oh)^_%ie^z}u`-m$W@WC*ggwXKc8j#q4-
zQPkAb#KXfI%y8>T;KIbjJZim7O${xwAk{~Y(17kTIN0oQV?)Hk!Xh=Qu&@yHuwb~-
zAPcpJVL9F1-JV@7e$%h7-EjPdgO6WPUQPo%FIdDnI;d)1B((#p?R0;g?Tk-I=n&N4
z!cJ#kc9Bp(OA!Hm2AuK?<a5En?LhvkrBn7Cm8s`7Iz8BQad*#FWe1ujG`ECl{rvop
zc-zf*w!oCrCu9>j*EKMZ5EQ)H$^HGs+1VisVUVhS&&}mDY}DgV`2PKFu(Woadm+$U
zAl?RI;MLOE10Tb)IFbnZ!k-{@OHWU)syc-K{DLwfFE9KuxGue@2xh-VXYg5ET_?<<
zlsIWL?TRf5J<ai(F1H$%z5JMvkPgxopkc##6H9}m2LZjOKJ@ILjkRu0*O2l!f+MsU
zE0V(o><vZ-bPNtYU}wj6@LL}%?hs%AvM@N7?vHnmxWMUn@~wIc2QtVm$6O6U2EAWf
zbdFN8^X^g~*u1u`F5CkMx?q`tDUqVqu&s~GpaVq4F%1R&4mF(O;^HDKjJTNCHYz?X
z&Cb;HV|@JL))w9E+gb(&QH0&9-(SH0U0ik{GYAa{aTnLfQLt<I^%bmUy9@ZUo}Qiq
zt9uY;uid~5Lwv9Xj>8oX!@C+volGG<v7qguBjAXU_&EdXp#|$}v82y|?O5yR$Oy3f
z4Gj(J*BN4-AP|T*Z{C2>p-!Yll7$=$AXliGz3{&*Z1Dxr{aNN*<iRE@-IgLII=b+~
zhwmyDj*dJr-!kg`zS`ivN|}QY(nB+vgC~Cegu&lE)CSMnxO$<y^<2RjlLJpS7#Z|9
z^J-3@x-}Ut1`0-f!hWUTNUf`@b9`vZOPR<rcu)-uJjcYGf`a$DfaOJ+@Fs(^h@jxv
z@-6`4nSQmFNE&xRGd3d=Q>L=Q_xyaQHNl<-d1hy4o7RyKY(*=qh@^p3*ZOeoV(;|b
z%cM)IlSA<G+Y%HFt4gk4!NlmpDmW&fxXYR61sei6IjR@$^7D&(w;X7z)xuvULHrd(
z(ShR5bMKPlCSq^f=xs%XfI%{`^i?|ST@5AX+%v1@K#Z2>+&~n87=eY3{#P>Ct-Y-c
zoiS*td8MzjusZlyT%NuXWZ$EjhR8u|-sJc47~4&l1T<mvVY*2`B$Jg1tE45Cb`=q4
zVZo4Gotu+pjoVQPA&N~*q~|k#uN8a@=_Q~TEy2`ORE>PkSmPA##DL2Y6?XswiH#lF
zcZ7iK@57j6nZGCAcCQy*N=rv4++G{<MaUE(kuc$sB_msE@-s5t;^z+aXmlWyQ&dn$
zPDzQV+XgR>JWtEcW=;S34zZyUJc{ArI2*-oZ_J8)HOk>Dy_TyL_z+LFC#-S7IS;)&
zJ+bo>tAhayI67xhghFHcZ*z;JoDdqxyK>kkMIGQkU`|Cf#8lu`IZiE#BOYpjCX`vj
z-J$=6C!5&evK@pDly||>(o%J`Fy`9_kkJ&O`{~Nc25C(Yqrb`5JPdo8gmJwYNF&?Z
zH$ybt{QS;g1jfqR+R*AA^hn)s8BdttoWkQL0S8q4rq)njZ);~Kmw2qwnn#8!$Day|
zeop$k%d~VQo69y6I|817kVM(wL6Df7Ov(KV5?sieyuH1@UY>*UvJIC_L|#X0UbMOH
zlZSs6q#dr}<e{xz%chA%giRTu9?At9h=(kv!1#itd=X8)bopn2hmepE9IaTR?46ix
zJv|i_6=t<52$gjntT0mfzMvqmrXgqWRKAw1l+JHOLXylLv|jG07+g~Fz<(g})H6cR
zLCzx>o6*C=Bandz?!W$cjDV1kgM$Nw5H2|p(Y8vT78Kt1;wPu4K~Ec8cK`l;BJtHg
z2JFGDji%Dd%By@{;xRW(h#bng?{h#Bf2a?aeupWq9AT}lu5ObH3W#E_Nk7KL={9&*
zR8<N2Uu+T_`mnRJSJ%~X+`Wr4dr>SaIVCjf%XB7H2hP;x=o;P6mF}Q9)f;GN^zm2U
z5ukhfCgw#;dzwol%q0i5=p0Y?z<?uA->|<xY_qhshRgy!3OVc|hch;+99+n8c76`L
zKgg3Ntq3I3qh5E!$ztgr`8KYEFcnp?S&UCjsegZQR)TT;%?jszq^}zCV{gD3(*-GR
z@Ewp$6~Es(a3q#qff)hU-#mi24;f5Xmjb_f&<F6%TF9F>Z*FXCI4Y21qcf9yAHGg3
z?KwMcRoCi;bf}Af^aSFBv?Y)UzV_>CYva&2<?sze(y}5po6aA!=_{j-9BwUvE4H$>
zK3E@9QB=e|`~@4yDW9~M9=1Jw-f9SxPU}#qtgb?eX=P(GznL|hy|TC%Of&=~7MHhP
zC9yQjsv#Ci{h_SN{;|R5@J^7asp<BE+sMu`&~m+P{;bLlu9IHWoikq1&quW+0($>R
zO4&ak-ywr?$<56TK(qQcUxem^zWPy!d2(wQ{{k-$9^C2ejQ}0vOBo0x)sLp-uI_f!
z-kqI8d2o{S8*P~QRAi*2Zvz;_F5W#$=523oz831@;Xy$clp^f(wR(bum37FnJYeQ-
zZhk&2vEI}+7LoYhF*iu5j^v%(+|nci8n?#?@$g6qHr-7i_l9!Z+&rZJ>#E4Tco9%_
zKuO6I<m%=IrIeE}+*VOhQP#MLyEgLj9pKuSf}oIx3Zm^(LKqU6tlV0EM6?7+0%WWr
z6>_lsc4{GQW!nmYadiAJSp_r~hjfjNjgSXI%m!nTb)bFwAwK?mE`rCg>=t7d4C@eM
zosokS6P6lkzF(UUnM)UWZ@<UDBmesuvOJY^u^QiF-pI#gWo7EmpEuXh4&E8N1;TbH
zCIOEDzG2p&B@pvpzI227jSP1Yk}KjyC78V7>cSNB1WFAxO-%}$e-zSBAa+CIJTs#Y
zxuU-Q47jM_VRgt>4GjE2a)r$(FE4+lcrQM@Nw>b1ikcd#J{`}6#)$1}R6zqk+<<Fz
z)KxUW+YPYM5R`r*CnpDGYD>!;aIYcJgUJez*^{y=K`k&iI{IzU5*&zLwIgHt=3R1n
zdU}#~3#+RN-_-W^_P{+rQIPfRn*wueV4yUV-H?>fjQ|5Rp|SBI{1yi+F!UHECnwtq
zS+Ku=(*w$tADrTgd!z7xo$#LrNTBqtW>buhPfU2hlt+FlE-;p8q>YVDv36N@L4j>s
z81Y;Pe0X_T8MT1*3y5t-Mt`sN@BF-sv^1Q?)qI8G+Q0Mjq*PSlnJ&VqkiUU#5`+#%
zZUEgK76H5`aIJwv|M4SPvXZ^MJq+-0a&hVD?}yR6ZEVj0<Z6eM5e^Uc`ac}rh`N%P
zt_9#%!*Kz)`K!$}|0@#k`f9J;Y_J3f;F+28U`ng5Cs3_HSpuaBa%~OJ&4K;(v6Q4F
zSE`~FT&T_A{gA-GS%P?J69}D9LcrA@CDp?AmiW^7^XJcw4hukwkBAy!I7hXQ`+Mk-
zV`0Ij11Jf8hwwjNIRf4qV5w(cRKYl>rlw$9PtVLirMiRcf>QxWRng&w9FTlDp#J*#
z(+C7`P&-4_31ABNZEI_5SoG#*Mwt7Xu`>j74ydV9z)ZJ*HVRvsIx*y`DhJs{ef=2-
zr<j<S@bU5C-H_h^4g;lG31}MNqdtA2x)=WmpbQwSQKVC$2pAFE&zbRGFbWEWc6`aq
zbV0TRPfqG+YK}n~?&EVcI6_xP=j7-pC^WRmbEjo|+!OK`sB{6g!o$ZeEG{;LI^c?P
zWaWF50f`$}H0&aoP=XSz0)RW9IJHOAQ!?(&M@w2;TAqO%+TY(_v)G`$I59C%(rs-7
zT>Yx6H+WY~^Czync9Z2`XfO)^d^uFZciGwZP_vTU+*>fisLt=SG9_gS^g7pX+<>fl
z+mig9EEUw4&{911y%1e~T#uDLw1PqbJIjwe!VRD_adB~og{*Nk-ur6+f51A!r@Okk
z0x|VVc{!P(#RatYAbSq*A?Ry@ybk0s$%1yD;2UEM3=j)lalr3nED1hta(k~m0vikx
zBrw{{%*<n7e?7fvxHu^7u>-3?4*+sA2~#BS6(ZifgQHp7*=@oQv9i+A1WxV1D(--Q
zOF({LG?K5^CM4J%r;jiad!8vP|B7RT&jJADzax~iIANg7=u|OO3CdPhtFQ@=eEo%5
zV4sTlpWT)$`u?3*nG3KQ--|OZ*nLo-z_)?K#>L?TTH4saks*WZfJj`MNGnAI`8%i~
zzzRTb51BsHym>`Mtk@9QGAk<&U_?iYCfKQksVNb<6HLH3`()3~#q~@{sV$lw)Yp(c
z0q6$Y$v}n#E?o_6Z3QB9nBB!@L5faHP7WgvMjNzXA3)Pevq%rQ<B2W>fEXJqEBM!h
z#6%$h0XbP&y;@h8q6GEX7`hbr<6Td&H;)Dyw4`kX>tORiScQpaQ0mYSuz}`q55rd?
z0OUV_b=|n}4g@<eeTV`3y5W6j@Rb6Z4S6TjCucCn3MRAE=xxE<0nvx>2tcp|^B+KD
zv^h0EapdH*1OI~G!OzIcV_l$NI8ZFKcQrNTiP`}D1YpK~wsEH65uh5_cz91$R6sJ;
z+`q8A3<*>Qpg{m10HVN%JqKfkj>hT%i3nv5UR+vA2Spl`dq4+=m51wwIX+MWC_i~Z
zR8$Vn0ICYh3N`~BgpT|_p<rv44h|imzo3~6TO5w5uceign(7P8Hqe*A^3;3o%-zTY
z#03EC#C8*6^yjLoA7C(%dD+h3EEL$FHo~Le1_w=y?zA%}XIx5;-5>gsL*{UA<-w9c
z1I|pSE+7m58Eq#~!j~%zoG$feTn*vzWmw8Qb)bEP2~RTX&E8syI*NZK^lg7=Km8gS
z+G`aVLo#zInTx`*DT`BD|DoxpRetbFGzntA=SdI0c@%wX{HPX7dYVcLPyXc_qsVzW
z7|K1(_c7ye*spl<us*zhxyf%!ws@wK9A2?P+Mu5FF6vj_FM{hamf>sG&iWBv0L@jQ
z=>MG^yMUvLzPtv<Me7<64XzP4J52rjf*_XHkf<`C)*$g$KL86sRkkF6A)N);yW_(P
z!}2|F%QjXrilevUpTOTg+3<h=_hO!8=yRkn=R2%_&ShR|vxVVFR}*TW$v=a4vNFiT
zoj4V#E9|NQw>?c6)#o^8P;?+uURC8WzePqw(Bfx~^meIZ!VYXV@hd{2Uw;Fxz4{Qb
z!T~!){HSkl;F|=?Fx+&&3w^j!qoY*-vndXyhO%e7^k>7*_V#aMCQ`!KY{^J1PV&Yc
zYm}v}Z7K|zD-H&fmT295o*tlNOAtRckSoVFs|v!YGe~{D5BVz^2|6y{|LV^R-QZ-<
z26i<hI`f1=cca1Gnq2ycVxDGy#al-r<?F36PndCohis4%C~+;G&^S}(K!%eOzP<E6
ze>5N+$Hqc3&;(t*(&W7AzehkCO~ZlEJ}{^zT`fA!_24Vp_`m)65yDvG#m$*!FPf6`
z?fnG*gY0gHNpjisR`uo%HQ`?}@5$O~E?#z&W$jw<Ae4A0EI+N>4L9Hm8`hc#TQK`h
zzvB`SaV~7K9FtOb;?Al#V<|)2hf(!ABjQ<c%%9ETyQgE<g<leWRnCZf<~PadFm9NJ
z@JYn=!E0jn?ss?|Z+(zpJyYYZHdQwDe7PvAk%9kY{odJSMS572Ccf>xw|^2hf2Vgw
z%^oo$)fPrY7hl{^x0~qa6#e>8a%&?w-y>Wrk2U0;U(F~^9tpt~e!v}3&(D>vLRExw
zZf-9_-;3!VQI~Mg>4|$hDEZLNKctzxkJK0=ogn5?aYz-Xy_K;ayHZbzF*7_S6ttC{
zFpO|z-f#bPvuOK!_U<fu96M|NmecHR<L~i@Q)#Z4N_&LUwYI4xOPWnfE<u+hFX*Tx
zjjTwP#Zg;M#l?NYCY_!5)2Up_YDu3TB4*Ww%}+k<<E5Qa{;pSLVZaVFbRJRH?Qr@t
z`BPvRGoSh-y?7zuOl3wc70ZTe!SwcdW78<(MT$11<U>m>C9N9c?{zT;<E?UHF_%se
z1l2a<4guY#9Ug~)(HXRt$EyxXN;Btz4#&+en=Y%Kj|N7cC)aIdTCiCj?fUT1hE7li
z=<%6-Ui#2?I3;mnvBm4Z!_9_4Nx4|uFW`R2VXuOJTX;!zG^J`coFH;gkZQ5Ye^K`7
zv(OA)E&BCL_tuX7B9^;975hxihCs&`&8td(;NmUx2WR_I0@=c8VK4C0TYV`deAL$|
zWRSw}cG7$cr=|D5<r2fUm8j!$MR(dn25jj8SJXmG*1trnoPavLZS%xFg&iTv6Lq?1
z2JEk!{m178g23+%G+LIk)e-*n!a&hMA%lGB&&otCLCQhzP>J{Ob>uBW{CC~l?<FgR
zJW)I2di7Qfd;(JK#vV(ZZx?+7iR!~YeGy;C5u<DFLx??bu(!pNA`1Gv)aH3BXNJY^
zW%5I-{NBFAi#3NYhcl7xP_+gcQU<V}Ne#|y59s?u$$rzE|2PrNbl6TlW-7NOp3>#N
zyR?+MN>wQ-F-L?xm|ovoepx!z5>VpeQJ2?q|D^hUz}bd@VM>k=x}-w+G-95@`OuAi
zA=qzk10A2OgQI3Epo<MVFr~ul^U}2B1PXVW+Ec}YU^uPDXZOB91|PoX4Iej@sn);Y
zJv=OTlmwAe)y=V&1`2zK-5Puq6N2>mU606lO_nX)2cyby_rw*oYR9V*_k|2%#~%}+
z3!#?3Pz`5(@{ufT@;%%5^xLm+tsGB?PWw<p*ShbtjE9;oBUI(+HeY+^)KkF&XKq5q
zAXS@|Mc2A_Mb6AH_SN)kWv%qU8((UK&CZSf_!TB0<JVQh&fmg_r3=qGIYr?>nFpUr
zg!Xhvaizoly1EjDE8?045otGO^sUs)`m7uxMptLQ2}tlFOi!jUhJOj@*Cw^fU^Dvf
zjM7TJKh(OVIdQ^kUbw!!D3lWBk@Krdx#t1pVxHXNu&f2%$!zvbDbhEcm`sVkX;?#v
zq?sX<V%EJKl}|KX6cY}R^ZvJiGFST0MDA5b3zK(mRY_}}R0nuemVe`B=^QrJFbzgy
z{5|g?>VJ?@#+seLOd~QcuF&(smC{OC4!dhS(%m}15rshyb7?tTmI)I#6IIL)7LANm
zOgsEiJvEINuqO(V%5pMG%a=BF*Vs>wO+%lM;lAsbhc`4YlahUxnA$s%drpxrZo!o*
zFmdL-P=R@!QI`K(aAAss_fm7mv*J=hCAV#fqmQ9PO-E}*wAb(w&Yri#UHEJ#Fc(@2
zo~<u7x(~JFpqm}iH@ef9^*_Q^NUrt!7MuL3fAhSJNLoCGwpoLY*I`gt23H#Uxj=bR
z>H&dN@$&w}>Ux5dP(|v222H@(1pyniSof5-Tf)S97o7D%PK}v)+%<<yJvZLOr$p$@
zQE8cr{#iPYwlh#}<~PTacB1I+3-P(0Jkb^B>L|403W(m?cCNyIu+~H8GTgrE{VJfC
zh-g!YVpX@Grk@czkXFc0#!62MBZC3kUoI&$<mr>Um-^pvsG@w>JyVbJ3ws&`HP<ie
zKAeTrz4<*!AQjzC&vePI)zzc45zw_vJlz|qQR`CNz=|Ds>X?Sn>VHbgEkzWlscF7n
zHks7SqSLQry5i^2uz!i56@ThC{#!bg@UWTc>2SQqLv*Qfg#Pl%zo+g!eRAFeeiN^@
z-fU<M&gNMrY2j0r8P&`(+i;ZyYucPzv)9S{*Nv^#Jx<Y9(INR^xj|v5^zfp*Sfsap
z+D@i7A?FA7#jYQ>H6maH7d!BBV+_e_e$7J^JCH$S3-j-r>j(Q7pL!LLs543}iky3R
zryo<x(fD>>T6yj+ByPB??KVux`*qWGD(@Z7@$p6OH#STUgkot+kY+eWQ!*+BIPiFH
zd=sOye%k0eYaTdd7}xkjsmiQ}D?(*!g6fCS(5I)x=+~z?nD5@C6@1<8m*C4SCMI@f
ztLKmBpKikcJD_z*U|uB3UPQYSRpxp@Kj_NzFT33K<;a`!4!WHeJv{oIKMQTYU$f`_
zbhW=+vg*9HT08Xiwnng-CvBLq=bnjHZmu4WW?}~T%^VoX{(Rk-oZGAzSwB=X;=4X@
z(R#P=9_`p2QbQQD{Pni~qBUnQ(Ve?~0_K)p-hVj2YH3$2+1PUHJ&;N;T_u?#*$w>;
z5eZ^0_7OTG1JPjWkHO34taA=5HxK2Wjv>+s#_Pni=ZSjZ;-zeRXN3)aG;a_t?)TMB
zGEI~inr6N1N={li<l`S^qL}R}T1#0tI;P-B$aCh4+LH9FcpQ~ZMyV?gQ@|Cvy4<i~
z>#oQ=4p1;PtN+Usl+{bKk(A5GabkOQ9-SkJsI7Y>X%|F$qiJuwnMGktbp0n$M!rxH
zJ1aw~TWjBjI;LhV{>E|lh7HY5Jr!AU_V3y3`Tmym{fjdjzrPgp|4LJw_U^V@2BH2`
zkn$r6SWQDX-v*%<zGMNJYC8*$p_5K}HnXIabogT#{d)3HTgUG3voby25aTdSrx*nt
zLG!>6NU=5S>KKcQ^*lS9FG-6IOnJ3`E|ad%e%~=29i_x~!xv6DIH@JbuDwUe=&wC;
zN51OwNikEY!{yqyG5K9KZIAvARl5RHs`d4C!@CnA_lq(vZBQ3NB=3|+wlUyK%AiV;
zlJ(gvgC59SEX6)&=_l6Rmv-XHQM0OJSzo^>lF=fFqFfEq2q#a5=~w%B<^xU_UCnhJ
zEJ9TKP32knUXI=jxjD$p6!b*#M&bUwjQD+p&9#_~$$9^Y7ZT!UwgcVY@&h78w#~YK
zeeh{WyA7kr=`y0Ne^i<aj*g89sbPt%AjIj_-`Y?pn3nm((I{vrC&>Pw>JE>9$5We0
zgtp^dYpUMtTFf&W94c!=<g=2){U`3&#9V|PUX%UdeA5ZBZcgtCwU*mndz`khRX(Nk
ziOxArf2ii}@i!<fJj((mqlYbt4jyh+CZJO9g@%O;H@WL%WML`K+$WM`zDMYga+c$8
z!D9H5goIyXZ2K0O8Oy5*>odN82NZb20cT(6VF<g(2FJ4mJ<shPq9D&2oGP2M$nqz$
zb}fBkamEP&8CrK@sqW;Z7DVvO6zN(OC|EbdeK4E5ouzaqPx<&*GR<l8_u8BHjh$<e
zQ^Y0R!^N0(l+3|kO|Ht*N1eKbuiS_)y`0r>0&*!aDzc`1b484fXCocLOUCm0U|8{>
zNN=T;kMrHx=JzeXm6Toy_wr`lBfRt%|EBz!Xs$<9fs9H<Ib@Pe(p<$<{)@H9I6tjQ
zG5O6uETqq(OS|_msrTER#t-PaqC5k!uVsW*y$I}}Ao44<m|IM38WEE)_4u66FO7|p
zeiu>skX~KfhWU=c%(G6@Xr~X4H#T&uCdivs)CZoPv6lYsq!*0z?N{qMA#is#ewS=Z
zC~tL9wlEbba~*T9eZojQj&^SP((3><$><T~ZR0n2E~fDB9fi5<gB>~*s>=Pz&emo{
z2lpGy7&fA<`;6D>sJL{t3?q72s4NGgbAG?{J)L1Ld>0`ecwWCu;gjNA#-!Zv$;0_;
zo+&8mI;U9a>Rb^8tGun^P1f2@NW`C#FxeiQ^gX+sYX5eMmUK_rhqj7RW+*K*1()u%
zWX7^4ZJp)2ajtvFYW*p$yR!3<c<<C)-<D1NNs|M^A9ORU9G5E`Q;(`<6<r#laL|zW
zE=9LhDzWTP$tiD|{Evx`ZOlh^)KJvBXDLV4=iZv<*!6{uBCXP4n<Zbwh`%-)x2}DZ
z%hwq8s9A0gS@;H%cRYxUSNA<unTI8krANS1$x=6&{8r|Pw)&>to0qAw;Ef^_Sbw`Q
z84K>Dd&TUc&W}~aI%zd)@843FB-bCkDpY&H1z1(UYW!5Xv4go_%>FOWB-@Ad`itRD
z_7d=BRg6w=&a=_i<yNk#csQ9Oo`)V6^15oTaEMy`jd1Cp%;)i9-)rl*q!9l|E5x_v
z9N#i9;zzxG8%tr*dnwz2^VA)&!4@sMH;N?KHqdkr?p8WCE-(6lKl$mR!yy^7(K{z3
zcyi`Mn~CL^va8eWYiK+!C2`;3rSQbcoRCJ7dc@0tqyRov<XugNpu`Ra0o*%lC>n7x
z!{bA2cDX6L!82sTEcr3l(Z7{NQl;QNY5R-d86i_P(d;A6I7uq=6c3J!!p((6@fQ){
zL~KIN4sS-gi@8MX)?{rIoHrs<$=$G`cMe4O2LpC8+Cx~`kAJ(ZrPNG_C(sU_@N@IB
zRz0)(aN73V{GZ>4)W3^;H!P;SSw-#Lyp?*>V%=GuYN8(8_;~vHa5Ra^{ivq!($ywe
zs2@As5bxc}?M?NI)}zvAgQR&5ru_Lgcal~LROG`9sYpNN2?S&pPES3ZqtKg91DWGO
zK*2wsk^n^U^V?$XXNe(xycvz5rSp=JxKW<H<O#&F<)cFRz9(M?j~>2hIkQ6Q84d(w
z$Y}J6;jE4ayr|#XuY5{&XRAQqlBUYKb1RO;vvJH#X`(;+P<VfFztg*WIc<h*O`X2`
zqL{JEnri)?U&yHhe0SX7=*_*wlRf8PaqQAl*XJ+y-fQt@SUG2Sxo5DhJos;co;`R~
zFaPff_5X7j|Nr=324q-BxjGx_4{`2dzgt{oj+5kp$u+cC3@GVIlvU$?NbYIGqbXQv
zO(y607T9T2YoWb1jageKL__nWcb|)%rS(x`*AkXL{aH>yh#QH4y%Xo38fU(sMowqI
zK052N*-uJEX!r2ncE#`9#i@jc`7uP+J<-#*oEbAi?2)Tg-s!Y)K}1cwt5s?8B;k{{
zusnHmMSUJsI+n0StIW%9UhGddI7fZmNp#WVp)X_$7DnNHQS)>ie2^D9IWycPFmygc
zUU1$l7dJF!NQRa{cpZ(QYrI%(Z0e?tUtMG21tCwL_8@2AO4!*aDWNnc_6OpOo~a>X
z!{)yS7ea;noz>}oFur}d9ZvB6Pnou-F!%9EOz_)^#_6<vmLry?ntke$t+UOMbw)#K
zDsBan%Lc7K4Wmo7bW)lq#8a_Fi~Ox~`Lw@=(vI{+Wgm=ht4uqwAWsI{Wk&3-^NstR
z3tZbieoe=R78}W0-1%8?hb;c~*EF@e@|S6N`sE65b0!_;z0mv4)oy4jRZIj_DgKmU
zBak?J=;(%Yv<lkDa&FY};&xd5HO6AbyTqZN8uDF)_GGy6@w>f46NR<;%C!9y;d*+E
z3Fi<8-JP$GU%!~8zHY@3+uPX}s_FGTX23CCjXHy=s)~kcLUWb~ErXK17tJ@kv-_SR
zbI=E}m*Kx&oUSw8Vhj>)$N08iWAn>&m`FM|v%X~Lo8lw4Q~$DIGQ&*&T7sfiYynlS
z1p6qtLDy#^bJ9zw-<>>O+M4Qx^$UhF3_<MPZwo#ufbTtTaes8$Y!U=Ex$X4G81wJ0
zbWC=S77uR3_Dd!xaqh%qEn0?I^X_W>S)e_(+8I;cNEL2mMPAf4iTpid^Jz@?w`*(E
zZ$eS#o}O8pqH4+H1}2BX*<49!xeCRslIglIUSBdN#pAzTV8XsOIJ)~OxAJ!-Mv2hc
zBD5AwT-t7?pk5i07H9hIchKgU>C`IeH@cTTZ!b$c6N*@r^f~FZ1bMU!etZ`XeOBUb
z9ht&5uZS4()*GU_?P&GG-n30g22r%1mv7*fkwj&E&O7qe{At~@d86*1bLm!rxFoBx
zb47Uh+L+q*knMc@yL-%e>l=3^pSk_05!Py8tdjMy`-m4g@|wbCr)MnAu<wv2#XKrc
zw0y)>vH#53eeHYkB@3>cZocWfV~5dbb_YvsKDWQa0Wk>%8l<jwi1qEmZlIlne6Hc=
zYm#$$?e4P0!G*<l>qSO}PQGBTTI73lx8uXL(NOK&SfXN9o(6>;gi<in)Phay?B4D7
z%UAk8UfA5NUn*q_9u+f_!Fwf8+i8NOy&h|mvDCy|-uFzvDpG~7$E@+k*4WNMhRhz%
zj8UB|;~k{v>Q@RHd_Gcq%G=&I{?5N?xrs(Z%H!9tNfI0;YSV|@Fhi4&xZRD37WS^d
zUefChtqT0^Ydi2?XcU`<>N@x@W^6KZ`0*$Z?)?AXKRYkc(0_{>aA->|z=6;d<dkKL
Iq>Tdq2QCAnFaQ7m

literal 30523
zcmeGEWmMH&)IJJtL_#S=0Yyp)l}5S|6eI)$NhxXR?vxe@=@bzKr9-;AJGXR;bT^#2
zy`Sg&$9Tv2@_snq4r6%Zy<zX)T64{L&1+tBE`#J{CGf73UPmAhc#<!kDIgG60uczb
zeC#Xm%J;wM`S9(U)eALS1OkT;^#^TW(&G$)ph8GKd-~cbX>HO`=l1Xk_tvJFG*|Ug
zFJHn;s%o6~US2P5HC=h2fkA{reEU^iwJp)MjnW5_S@k{!><<ICuYSV!Csn7qDk_S7
zZN+@G$}@>_cjKjAq@b&z!>-$fI#;KFk%M86ZBIPA5v^LxQ|$Y`H_ZEfY9d78Z)cuS
zao6AlywoNc_+MjeT0?mGnb;$?TkxG!G${UtD0cr1sV?6eqJjSMx^(cJ!<Ud9-tzVS
z|M%Aax6OoEj3dv_%gZ}nVy;+u8*KlVGg+3~VNEry|4&HBsjBW&h5hR9VpGfVMdD%{
zII(XRCx5f@Dje2-XDTsRnoYD4vU^>ep9=4^QJciRzajeKyx8;7W79lo<Nx0SZb%jg
zA`q$zpqyJnUb-%?sePmf@yEMK<d>_t-AIrb6ZWJ1@UtHH$8?DyUd_=JcB6*bI&=am
znyUr|FJ#4Ka7!#DG|19r>gSrVaT>Q~O=6>ec)5UIJs3GVR;?y2oes^_oVB8Qh}A0`
zrd2Kf+D<K_^AUYRm2+T7GxzilP5H0*C^vi{FL~|3gBy3s-ZuJ2jcT{zuuwf2@YtuH
zo3k2x$*@sw^!wL4KSW1|yhgTzS`eS{qr@AcSFhuXYf+S#tsEWJwx)lZ8G4+=vn{SQ
z#ee^PTW2scL(%wpi)F1}y;r@@Ff+p|tbK@yXH)^`T=zCAdsUT{hw|@-Qnm(APDLht
z9Lf2V<hEIEJH0t^cWcu#?J0U!Tq^2GC|?K^v0W0$@To_L#LyTy^DcF*$ZLFjnaJwq
z^r_!C!r(N7d#(XvVJTBn`F}Rkak@EwIL&5+&luAd71f4Aa-H@XEiH)*9ml)((y>c+
zn>8FlQ+y7LS}xa$41a2B{LkBb8@=~ew@Yi^o;t*JpW^z@_4QHm+1rSVa}}wxxjJn8
zR9rFKwJt*1EMT7;N{IXarltG)rCl4W_soGSfvY+E{=rnWeZ)|U1wngnip$oQkn{aX
z$3|%>e~lV4F#VKSC?-{BCUNW!|9?mHOa>PWu(u*{dQgmYRsWA2pPH=6nAP*~5^*(g
z5s@I-!@6=?0z<>}!_Bbg!F`pb{~m~flqts}R!jna#`&)Jqaq{P#GIdv)>n;FB5f~M
z2FoLSzkHc~|Nf(F#>)uI>a8mSgWCt=CL{?zd{7AE6wj!)B$KV~_ik{jU*cWP8_HkE
zC-xO1UQ@G=IO3-`Nf*O})D1XtEr*P?8vU2WKh{-lHs~u<PCO$=AyZT5z(8LGCn+c2
z&6}@E3_1@!QLh&Y2t3feCV%_Jm)J!Q{D8CVR_Wax2Z`tQ%uGxl!&EApp1tHa4!$8;
z#Yp3ulUL5ixxbPok#clw<TL2Z`!E9=yFhhECVd+}U}N0U$H2KgRm|DtJe13Ph?(ux
zEtJ<Zb$(5owbB_(Pjp;s+Z^k+V7-mTAJRfKqwwNI;AAcq+3c(uU$SBZ&6mqJmOnLh
zztW+6dd|HwW;Ry)WF2{{)*+`+N7aCJ%xhgl<JC>qM^g0k*JBn1s)hPiJ_^}iuXOmS
zq4SmEzjuB5^k;Hhr&QXZvfsqL&2UwxXdh!u<jbTMwko2{CT@~~s?&;Z`%julY}^m=
z?j{X&^^-34-XE#}{WqIt>rC^aF)(8|P4<RA-fi`2@w!>|pp)<hW?WJal@5gtmf8LL
zJf0Mlj#-+<U-3C_?PjYL#W1RU)kT8@5x!@4n`HU<voRan4EHl@J&`L{ymhEF^Bk|Z
zsEwf8BFOXC&kxaNmQ#kAIbH>r1RXD@cnW!(m8thbWc?UWxJ|5OGGr`sxWTT^d*cyC
zE@H(?Hm3B-6M6jGXuIAPDW0J*Os}+Z9a4k%h9tXUS+qTl7h_47)W1pD!P?h5S*%aj
z1@S2<5<XdD=3k-Hq1mDc|2lezmhE+G3@-|;^`TtLcj>Z2tzw*)L#Yk6U4@LoM>9UQ
z%RS~}`ET(k-&DKo&ou`#R5>54AptA}FOKG?7^&|2ZaEx35jxh_RvxiEJ?urx7P*;^
zJ$7}C>&|ua5>G_1H<yUwm-9qR-6fd`yU5>ubbsaVv$1HD<vZ^#rHO}hbaZ6S>fNG~
ziDP{$>c?%=oe&)#ztTN$Q#6odZ#`JsL(OwFhkJ8+ns^Lr0bP8CS42gthM?@$9jra9
zuDU;-7q&x<1X<t}q(uIOhZ~cXy3HT@`ubv+-o(Vl2Kf7LAd4j2*ueZr0Tc}>p0cW?
z7VK76v9Bs4OuaX~Y$NO_{@|}+_@lde@BDaF94tNbO*HIyHLuo?rLeG2{g?5p*RN@6
zZ~+CP>b)#$g3PDfsQd8Bu210kqDhJbGEQS<W02Lon_-~3{PXHHDGSX$785HZ716aR
zcfR{?1?8`=U8O@jxq)%Vd$+Dygkp;$Kcu_EV;&D4^*YT^H#8)KiGe|}00B?vS?t6}
zO^x*8(7Dcvh3F&1j$YKOV-caAVbS3YAFtqK)@rJ8rzeT~;p?IK;fAR1Sef<NQa3{X
z3w9hXBHT+sherp!2C;$l87;TObkuB3b!{p{$bE?E$tw&z0|45??gT$hQ!<xo@^EkY
zI<zQs;yT=`R}6JXZ(RL|*@Jh_|08ZwOD4*5|DtgDFh>=wT8RI_oRzGs=@%j$gnk{`
zm=|H)RbRA+-ozqdw9-07Z_iy&?($R+DA2Z07Ex*{apcc?tcW)B4m5t9t9EtIMKB}0
z8L1F~GLrT4VbqK=>!P>p|8b&C(>?^(c!6?`u2-u!b{%dMI$=28FHA19`y!rSuz7^_
zu|vsT0lIMh1Ey1tvzxrUX%6cH=y{r1&fTNpMK9|Rh=?eHF3h=ub#x|c)v#A6s<rv1
zoKP`+!8gDNt=WtG&ozxJ2w!vq5k{Tx#o#(|$T1pKKWw~%T>hn=L-Xm-#m?!JCkBXc
z+!u8MGbwdXbqJ`?@1bSU#Cy}$^`CAiR+a(4F+fJF>N0t*Lxp%~MSO(AiXha%)b-Bw
z!kxeruw;!IH!HnTt(+%b>41Xbe+gjDfbN$5-IbL>M)WaEj;q9oO*BK1D`a)gXQ&g`
zpB`i>e*a$PLNs3TQ`6$V3^nJzyR0|Ybhq}*22B-L1EU6Ap9=R823Z6RnnA=6*$ecN
znq#_}OA1Q&Z>SjmZv*ygfe$}hNevLrJd(e1713PB_vIEV4QoE{4B`UE4(-Hk2Qt6B
z!q@-%9XCuz&ic!3wPze=nCl$YDT;2EiNz#)k-tE%!R+{&whigo*|u0(X6R{TEkq4=
z@OO8Ylf=5nx;u3URd4?~3#umF$7rFe`*~yQd1&6w&PT1n`Fa8Wk{1PJAZ+V#^Bndk
zHCH|(>_phUV5uXNsc!V4)z`h2?7*^f;3sAjiKIh;>%V-3&Gt^=Y&feM4abT)pXd>$
z{uf-HtkGrkOSVLIyNWv^mc_aYFHoT(hGNyePJe$!T3bIBK?}e75pmu7Ijlx5V;oyR
zMg2rhtp~$F@Q}>zHSegl-5Nk<;9y2kV`D8kX<YZXx1h*XR-!(1ocBH|S4_K>@C(m4
z`j!XfK7C4cunsaY-XVdN-z5EgHrI?JVrI3qYhWOi_lM$(+^c@2FV1R6esAxm1_nOj
zVHO9#Aa)I3L?mF)(Wm~m;a2kF`}YsgTi(AX`-0tu#^{CNL`9b#O!J)-*D^MC8hha(
z%@=3eZ|RwUQvdO>CPlgDw~2@l;fr{6C;oT@=(1F|k0jo{{lX~GX7hz{`pTZxFKjfl
zd{tDE#6r~<H$@|)u3Sk+gnzT|@3&Sqwe&#~VY9MTi@z1<z$2R|<KsmafIqwS_Jlmx
zB^X6i18<a<)6yxvLG<@qBF#dirQRp?JgLKNi-<}al#IB|*tvP-WXGl};rXqiRVKbI
zR6=D?(QWXZ2oEPgs73r<Tzrl+6VDjLC5xg-9z@oy$OHu)zY%S)^%^v88Y@7Zp#ML{
zp=QBM%-popomJ^rixzHIq4?^Rv-o@=g<vk?8ecM_B&OO%T<qasx$UWUQi=J^8(+W`
zP5wD=EIFTVsu<dyoa_YwRjCPv8Z+CSbP7IwbW)z=MY>3G>|*ss@=(<S?T(HiM%GtE
zC_8KyV&a8I`&Djoy~(+U;H9Te&1V&0%w}a|yk!+9<NWmLv5Wi9b~{E!Go~9+$sufd
zGj^H(cYqmiPX%gLh9{A|k-vfdr_{U?|N4{GrauPd_rAn_Q&r>g^Czuv5FG#V^wx~%
z<pnmrxbD~Ai@*X=q)f|U@%qm0TdvZ!Zg+eyn&+yF{)!n5*KuBA`V8fz+!qj#$s;fu
zvwB}Ie}kOdZJ(acuC6)QBsTGXcp36x%FOX{f6?;SM2p|dk~?Aq(RXlrwzinW0%Dq*
z-;OeW%b?~a884BvwyqZoNcix<YVK-<qsspwW(o?dZ{M$3;UdenGifB>y<;PmUTF!z
zD=I>_v1M~&P2i|!@1^3o6U)+?o-S4R5e>cXtKa`w{73<{&{?1JfthKgdllpS*|FDs
zF;SO8uAX?kd%_g=xW6hfbftuS#68<hS>c{?avGrN`keEB4vLm$CM&N43qJ{CjP_c4
zds}yNrV_c(`OsMh!_j)_<r{a!yw<+d4A(<uiwV_r!dsmGpqK@qSW?n}jg8#<p_pi2
zF@vb+)`YF`#2uC@FIwrXdMp!?PsZ3e(J#Te%IY6}{FoTGI7JxWzU}LcZoQgwhv&v}
zFEz*Q{{GyQXd1~FgJLPEUta(D+~$<i(JWnSs5yF1Oq9;<(q*Ne%wXjUrGKSM6(i4`
zO2;=}(=M<0$HX@3iyHs<ZSdgta}<PBnROB;3B*rV>|@W!h#*2Knj8E~VuOrH{yA3-
zwS*`E<p#dM_*V?^kcO6!vD(@Wf&s{GhyREwlmYykSC<ydFs}>@=&;VDTsHaR{h55T
z9&wZDD+&ypkdbn_^Wp*haaSkqkYH6)+`Y>xUmN&3<lnN92kUXdlyxdz9-CTSpF<p{
zo6D?+-8CX;y#1&X#9nfNd()}4<kik~5Q$lD^HFV;jLR$gTTJ#9=<Q$nY_N@*{@sfM
zMmJRDyxZ2+#-`VzZ(#6K0TbZi?Y6G3e}hS}zu!4VouODPv(uS%aq#nB(tb^DXGh!I
z({8qd=`Vl&{7DzNvKXsPtc`lpLI7oW^kUUw%+Ah^nMiDuuNPu(Rb{gE9a-X!@Hj;U
zg^k+Fi-YxXiD%Dbl6cr?X`j2@MG;vB6|2qajL$A6s*>#jb-X6G$wCt@vG5E=ZRu&2
z3iMig61iDeST@6iFI0aS_csOIYL4}Wit^nEZhJdI)=*#n%E0li2}z)FvPonuJ~`js
zgl(^8xy^T#JS`;;o&CSV!6YoXTZT3^Hmf5!)!KZb*s}}8qpZ$0!O8qAS5mzou(!70
zi6Mrgwwbg`_r?6l`0W0iZq}yoSnI;O770*nvcdNn(Nq$rsWQ}9;a3Pnu?gHybmp48
z<b;l|VPmD^k_ATHUYiWAanCxX<cQMj{mB}2-Y-GvesR3Kw*QiQ(r#mdw{nuX<_=J^
z7qQGLlukO-TNFS7>&GqUD>sLjZ}l@15!#B1HoV8ciHL}}j)Sw@oj97OEsUPyC5FAN
zM@w5WHilOhVs@LA<M#9CC^w6}es{Zve^bEqV6Du09?X7nuztEfsvXRNe&08c-Ad;O
zo#2ia8agJTZs2>_I#+S$vtQE5xqp**ZI`;@)1-{y)R?$NC~khj<Pko;Dt!5Is@i(9
z`kx2?3#6&n8Xs>`f3U20cRPHlgegh3>Np*oKSBJT^RINWY;QF_N5}F<Gpe+294k<*
zuJ8{VEK=|_)C9uKoh%sakC*6vvce=%n)4=gRIe=@%q+;P)cYz76+DlHD74+)h6UZd
z@@!$@%NK-F&V)$d`+vTzxH{6-^-Ds$21BE27L$jJ$0;8^S)S{kNH+CtH0^nCgvI0u
zy7Tt-*sp7Tew=PeEd|&9MO`|Vd3eg^$B%CJ+RTQp%L)3mgqR+c(MU!muJR_FJ%@sL
zsDvqj3s1ZDg5XQDiuT(Ho?q`28(R(3()-t3LQ&QB+{*~x@6s<<R$|eT8@+dzw@2P^
zdL$LvxJ&%oceY-ut+w0t?fm0@iS1z()<n|_9+bPHsO2+X1}|z>`E`XGlp_CrIn@6B
zySRk1|LSREp+OFX(+t5(iRU?vMCcDq(y#Xa2x;O-PGczfatt&1)wPrHX;xdho!wz@
z366hd=F2SQ*=}wFNYZMrZ?}ZChLXd6{VV_5FI^%Y-QeYAM*G#Q;-$-{MTM)p|GkM?
zVlGl{Oge4&!*G`MZuj4qKx4B1EOfdVpf&xs9_tbZH-(#krS{{PD!2y6|G>%{%jDC;
zS|iDUbQSI9zEn=n7+?VZKGME$a!Ngd^xL;SJVP7J|4RH&3Uf}qmPcQZ^*(o=T+NWn
z(xm$O?!O;>c0Xtcpscarp!m#lQ`<8{t;pnv5o(A3miPTTHg-Ozu?`2RJs&wCp-xe~
zMc%AupRt4Be^!3XdV`i0K_~ZF;BCWIM85i4)*l-5U*G;`E<3)xjY%D3;fxpJmKSJH
zIi;DJbpQL@DOWkJc*C>3TjF8j^DRz`M8!C;Ie$fL5C?Gm*V-d%e~k-Y9O)GYi@G6D
zn!*37ucV@3(l=47`^vOe_WFO;Asb`x_xdgFe>v#?S$QQDE&clMEc%HPqBVQ}{bpX8
z8nVe9wb=jt@c(CNm&11{@ZTfn?28kkUxrB!+y13l3c1G(H8H1vz}ZrwrL2^c!gr_{
zMFTN^hQ|#K4sO<*ZV(6^QBhN;Y1eujPPyhCJpqyU{_k(1x>M!0Mwb`o0R)s3goHag
zoh&H=&QFAewNXVYtmMh-!+@_}zs}9gwS?TWCEzveikoc=C>jmGCzk?!BrtG>!Vf-b
z`?(U$(!|6B7KM|}`N=*xpPg!<emg^6RiWGQ&Qf>c?)AH5s4ra57d$^h=73T@>ul`r
za2CVmV_x2J`_+*uDOi!O@k^p0h!RTk(cDS+9m~q~lBsIf!`0Ev*#OFm_0nk`ai&L)
zoc8)92q=Zl8VQ8=Hm7SldjEd_m##Gec8}QT?MQF^{sKO+SNO84t*vi$AAyV6n3Y%4
z;V3e0STyM{ZZ7Du$K$g1qu+R;%mDU*jygw`=7qemu`xl{gGZM#GBOK^aih7KrlsS{
zy(yb^d`!f_&APK?*09{B45fZOt$dyOKmBQ^dn>lbJBtz&&I_<dtqQxxt>HoczP!D?
zJNS=}k7s9R$;4207s~uVvHRtzPtRvT*Ujp~o+REP*TW4ge6l-&E*!1lo3)oUXSI(Q
zP<p=1aF}#+Q<F?0*GsEcs1s3mu1mqHBeKZtxHdL>-T>S5%p6une*2vDfUP%?JFGg~
z;CSe#VrLA~>R3VAn8Hob7lO}^6JLLdg9WBlZH$+0LPpfn(<5Qlc$EJdKG3s&USWS?
zf3%~grw1Pg=a&3!)In%0QxpxZ2-yAU>xyT`AP$0c|E$e?(FKBx{kf5<s%nO8^0Vjw
z)XSsr@-UMY@F?FKmZvFA6FF~6{|Y(&>ASy4;(tT*?ACc8hV}F%&C=!BvM@c7&kfPG
z%TzVw{sg?ltXc8;IF-`#te`^pChECq^c!p`LhddXXUBeierqYpS@s~S%(sS-ykUi(
ze-p}_<6ZixnDKZ2BrYy4J3IRmwS-o=&0<=Z5TW)H)NfE^er<J|YYGy+JS!?Ib6RNo
zq?j&qmmNHpO>d)eQ)t_d8YRmo(z|r^_xBt1ehvu@t+87f7|v4h@$u<R;-#+9C$+mg
zKV-<WwY7!QKGY?;ecQ=$ez-tAr$P67dOEZEufS@p8aHQI*-)=9D4{D7viiZeB~IWz
zoV^F6@D5=X?V2jnQD&CrlTFXd(^g8)N7HX@Yzn!*($LYtxtBB~CUTkYh%*)oobL_7
zT6ewCu@C}#ePVEQ<2CN5d9@cVKmpzwgwwQ3N7i1P?e^AQoZ!@TL^FUvaAx4=6*h|>
zxOJ5bi`AbqubUMinYC+@A|f6zK9GQDG)i$Vfz7GpYLr97LwpPAHU*wk)xB_Waq;jF
za$R?^?nx2$Z25RsO-*eI8}+P%WPY;Hj~_q6V$TNUg!jd%)M9Y1Uk7FpM#kG9E>U&!
zyVN}{vlp_mJslkh!k!nfhQA?AYj%9WA3kVItCd;JLUZF_bJ}yUBieDwIl#54s;cVS
zHyQ+Sut8h+{gsuKjuhDxq4ROeTI-Md@E9@yr!DX=@THTrqFycM#b{ttw@FAyh>3-u
z0YZlUc%y2!C+OBgzGT^h-uB3cy$PJFIi*uzH-pe%)-;JQS!ro$Sy_O*-{rQ;I4o89
zuxB4E0^-}ZtF~M20nxdsL7@dU*nuX?m9Y}@@mq2N)+c)_0|NsXIK=rLqXCp2PHSUq
zmQ8}daWa*%qCAO7NGN%%bmm(^rUDnOc%-DI0YeoP6_NafjA0alE@1xE(Y!<PXO<9i
z!uvyt59L083=E9pcidP*7PW>^UIH!!vc!vL!(%&HY8jcCF?EX7qv=^5J~RS)DG^4o
zFWwic<?@(;fq|BG5w(K#;?bJprNqL*CkS**+%mfr!d4NW%SA@Lpurw}AYzD!jEtrW
zJj`cUj@9;v=QNEOT+sk|P(eXKOY3vv><q<foi{pwnH4$aqsNa6kY>PL$1Ci!xO2y#
zn^b$cQHcPmyI{qm;n=Zax3iA9K2|XQGvo8gTri8n#>BfKB<wScQn&{42msL3J?$kQ
z0wE_@Zd22<?>^D7v9WP+PAdc7i5QgKpm=g?6x{7%fA{X4(#1PK$Me%e?Q$F1=|C(3
ziV1iuK!@YT#No7ON^-KMyu1}yIAd@nWK%H?sH$Ry?Beor?QRc0c!X>QJ3l|b-=st3
z`AoeJgG%nN@%5>|(@|~D_3=^y5etir20t7*X#cEaq;wRzL13yFGrL;A{^SJrhemSV
zWXQztAdOOZlY54ShQ5A%pcdo4S6@U=xcf~qQniNd4sImf%LHCqLno)Qt@rQz8fTBd
z*G6(RPvM(-&d9Xqr63Cnf!DAJ6EpL{z}V(uXKar0D)4#B0#y)lTYB=J2NRR?Sb16Y
z?f|sn2Q-=#cNA(UqrSKdI&L3iNXIgpPuC>tF)mHHZo0!+rpu*R6d{vjmsvng1ET?Z
z5L*eb)-O;6u#b$7|5-)d`Gj8|KfuSe38Ffysw&C+0OH~5j2+*@mkD;@ppMh-v94ue
z0r=m)eOp*qXe&&Lj*gCuY*YvwHZd^2G7>kx2w}wZ_%WJJ5ps;5ICv4(*4*3-UPI||
zbiY*<vF70F?9A&0M(OJ6qLv7Kq$m$&c0JxPs5x|jg;pmekyTD^*zu|4s25vK*FZh6
zvtyGrd;(BYgwxk}!{nTzh_MqQJ}@{~k3B$jnGjKln6(jc2E6S!+MZ`3#u@j3^MUeC
zuXJ2(kKL$;Ovo)xg)z8!2dawk^#g;>7&XQibS!)U2op$w;4f>&E@_0fzI<q!C^j7h
z2dTB5ZyC&xn=Uj6mMbL|T(;vg=Yr<xI9T4Wbeu)A!fv*~@7uR;w;syo*V=?R9jvL1
ze6`~t>)@pB+{x9fWF*GP%5f@p+J4Zfq?|QuQ5qW)^I5=oCzz!MVFln1xx(gS^O#u)
z4hgfqkrDb4tc*pgN>yII&L)Uk7wwL)hrnt&Ku*T@?*%M<_kxat9)8XrU(d!gy4Ty*
zb_D317#-d0LhsgyTt9#SC1zt{S_f{j_VR+_-Rb|%0vs4zT%2yYS$aNNEFE9ZkW2A6
zS;^Sl+3D!)WFqbaJTKc!9$ZO+F2mcmv;KJZPWD&10dOD`3WN)`iE0E>;e=q7kq_n4
zzkSPpwU^BAs8#J!{Oi{qfQnh2%$=I!H!+I<TLS|Mm6H~w<BpqC-Rfpsi6QO*(aU&F
zU#qIlAO#8AEWEI=V6OZ@j^7Gt#*FKYwe>bIu*1#iUN#MQww&;JlYg437@&Zmp<$tT
zCS#}Dp==yIA$w)H4b<weR1fH~F}*2q+i8E;94kYeB%3j5QEN3T`t&K9>3aw>AY_Gu
zS!HEM5U+030O-ldeYNKY_bT7Ef2eo)>sI-cQ5cRLY?U#nrmpS+AOatU(8=R%xsq=&
zS&^O1XmDc(a36_8N~JH{!x<$k>NczdJYXW0e){yOwKbdRo)x7kW_WCDYtIc0X}#>=
z_L^~^9@`;*ta!kr>kg+--v3G9jHY0p&U=kan6&6^Z$BQby|9PO4-d{7cA0CsRugo5
zcD#FfxM_6gk}SJ?yu0kU{yRD<s;Psj^T}b6U$p1uRCUKoDXzkX7X0Qe8N24~*FUX{
zVMpB#FD1Kxy;bR=S@Dobgpx_Vd}$Em4B0Z<2W$`dE{eL1Z2}BCn)4>WH1?hsq;Ebl
z{1L<b{!oMB+Ey|mqUiAOz7q2{h(vBnVpSURw^*t_=U^2u)B%9o-jf4P4N$Wi@<W3)
zmRVC_!>6OLYsq+Povr9hzgWGfs7P8$>cfW*YYvr-v#J$#E0CPVh?;?XLRBaYXXmG>
zbGg?Ncq_{*N!CbRUERT<WL2<JDNEV9ixy(-`}gngUnROFh-zpid0y<MR5hSM@@9to
zpvuUVxDFxQaq3W6ou8lo<42%n=X?nFlzd4{Oiar`Xh>oHEq}}XR=VPPgEFO<MfcMK
zI73{-bYhJqWL7^MlCh~|A$N7>X?w5Is;UVIOC{izEZp3q`{cpRHVbVx%ZAw9?kNGe
zSo;$d9nDJ?3M3BFXM*4S_!S*8e2G`D`XP#D2i`q?{P^wiw@tbw`p<g6s*X{iVL3Gc
zvAsRtYWe%uPk8p6bo$L~Vx5<U%Ig*rW!sQd^*se)ssdCc=Vo=FVsMbr9oUF_I~Y`*
zg$nm}u4C2BTAl3;$^r8U{qR9oF%ZZdijx7{OiXB~#jrnl0#!{}6DhE1V`JmRLp5E)
z-Df<yX$}Z|HAC$c*kM#H1-}C$J^crj?7W(j%vWjRT!}B8`@V~Y@p5u1-3?lVdRgc#
zw_+$6?;M|Vt3|2j0n&&#*w69-GXJg54j%(UmSiMd*~sQw671HF>g;Tpm1|{MIE9QX
zs`+2Cvp-sz1G0Rd5OpAbkQv_VHU#Am*j9YhQI&yAL{c)ylsbaMLH&x(E|ctMf%Y0n
zmX~b&o)=X+>Ob$PnYR-bac^*4mwz6140(VYh!oHy;9i#l(k#Qv*&j{CklBh2>Fi~)
zbQNyLPg+UCyBA-n7b2qG4Rq{@*06J`y6%n@=;fkE^E<2^ZMRY)<0xABIZ!Smgq!f`
z6Piw=KVF#N!JRx+gwWYmqvzQ?Wf_-rojoNo(WobBcDTcuhs<myS2Qesm=w_NabkOy
zqgzDsHNZZHNtVjVo{jSSdz4jWp~-Hi>!nhQon`5Az2^Xu9lO&+B`8@CMG<KYwY2Ff
z3o{u#MMMy>H|R5~OgoaYKYqkLAh)S)Y|x5|jjhE0UZ$);MqMHF{i}dNZ(*Svfq_cD
zUT!b?Xf&{KdUHW#!E&L4*58{|x&^bXaH<_)5>P`VB)=zf?J6%Xha_A`m{6@V8pha}
z?b*KlE7zhf@;T5l!@CJ226ajMaiE+X&iD{i5c_urvc=1MZjCT-27<x(CV#hn)(VCv
zxUgqK?4OS@x*Ys2W@$(l8e2bzin<-@+|0?Ar(Ijq5NRl7>x)Bci_qC+*1S~L&;U}*
ztA)X6ujEW5@%;JY94pG?Xiev_l$T*wpkBPfWgHb32cK(SVPa`I@_}#ziZT>EHuw`l
zu?zWh{<rvRI@h9&TdT82#=1+ktY^d0EY%AqNt9XFHA&8NcU{Vwq-10?3Uo16zo>l1
z=vbO~udUnD*;%>#*)^)OrMPwNge^2Sa&RTvBILgETObiuZsF=bCDnXpfKKIhy6300
ztom0ocAA+)n1=DPvW9oIWCj073dn8!1J$xV3-B}^uDhl}tB_)bO*D%P@eWlnFffQH
zCX8|@LkSxA@_wE@)GC*H9_`q@Pkx}3BIp`3&y%DO`7HzW#bEUKFrW|t7<UFo9zE^I
zmn<i+eKoOQRagx9tbPfXfk=C?vI-6e5CUic)WKTOMiC0PwxznD@XJk0Rtn|ub_^@8
zcrHUtO^ucW+A0vKs)onf)Hn$V34i&9>JNfPzh;{0sKq}l(`&`g7<|W-$l%r`t!U0L
zp=ZdV@#g;Ltnvl3F8O!-E^Yg`m%F{f3>x?y=VmoxNF|GIogca)uMnZ+rjVrGzGWG0
z%GQ`9D~h$jRQ{t`b$M~xVXD>IV^ONY-32@Z53><jWc0IjEnrMiId$}pdZKMoZhLOX
z+?=j~56~xti_DL-JAWI9iHOQ`U497cEe=!I+IuP-4BeZJVvaQp>E&iC>bDt*@9gSI
z687|n4vME9E$XpPQj0HaI_w`$V6a#`z=~DP#c@}uIj=uQv*T(k=rI4g?=l)@jq9RG
zW*U7xtf+Fze(7^uo>p~c*R9tSTDnmWg?V+wIET-R-wme6$Hzy-oy$kdO_m=#j=snB
zS+Qci0~3P#$C*+*OUJ4}R=#r3>V(ST%gFZEvwwG<;b})b*-o5S`sSO(YPC!nIJ!mZ
zN}S|wZ+24xOftK{IlZA`ubsYwQBkrb;5bRao+olNeTt)m|6~^rXj*H?Jr!p>qiU_q
zggE!ti<Z4vf=f_&dF*y`uN8B*<TB{Ut*acW_|(g*tN%?CL;n3ZeDq;e8O@AMnz>-?
z+5+Ehb8+W)0q*8i0hs0*C10<a^}oj>w4f+9pcE0qO<2@WoUW7S>=$W(+}|i0HW75*
z(f2D?(J+iJ7tcOMC(DpnM;2u)QJT(<C)!Q^hS>{-X<0)R+Sb%H`n<<J^)q=vYR){O
zyqC`I@fNv}9oCO3{dXsfIV_Nc%~^_qRRJztoaGj}4(`9&#iCui#M25Y!kl}OEdoFb
z8k^Ej%q%S271zzxNUg*_F0{CtM??Tb$W+)8QVx_WIpIO*SFZ4kNJNtuWLfD&=y@(y
ztajxS)3nO`tWqZSw{0$EP}VF65vQI`HB&M)P+=}%?CfYyV=Ql6eL~VF?O;ENn(mPj
zLtZRFNe?At_AxsdrH+bAS4Z)JToBTvxR7_<e7uA+UM9Ok@$X|x8pBaewm5Qrhu2OS
z4ZCIb$_6=_sjXexv){q$kSsw_0Nu^%{9)_6AlZzf4#ijfU!UVJd&jL3C!WeXL~p9R
z=n~4@nUak7reT00;-m{s@kJzNd>N$$O3ND4)`I%k1~Ng&=_Ee;4*P<}zraM6mzL-?
ztFITom65@1qFeg&$5?Cj2|GJ-{V%F7BQ9RJO?!Y7v?Z(SV5I04rinop*<Kvbq})08
zwL>#2{ODK~-*~dZ-p<a>?~vc_Pu=^S<Rynmp9XQ=>i(yWi+N(m2%5*1P*O)}x=grs
zWfx}R=jo|9b2JzC=;u9VJH5x<OHmW;3aQYhMCnDBng8v)-7ZcxmBT8--KB14KSkxn
zz7@Oe?d|*b?<3~s=aGrJ8*B_^d4FCn^4_>}XAGJliz#@d99s_jUI1^nBzNw#;oohH
zQ665g%U-;B`}WmMBdKN(7|X9_@@l|G4ktu+A6oysz6oiEVRy=$D>0U^C~jKNCYg4y
z*i;1pQeFg2fvNX!3h?gJUSS-OpmKF3F0RSrJ9$<_nd(JI=t<@|rL|@+G=oD{I(*!7
z5zMm!hO_1;n<4$;#St))eV6J&yONov?52H_bX;ugKZ6-z1K2ZV>d&KR>BG2oe^)~K
zqs3x(KiNb3GPp7(du@X)_-+S2KE856Yu9<#>z-q%)3~C;$mJbYtW(aEcB=`rW;L>K
zMmh*^a0XZG)_nE%AD%`lFwZ{w+N6=wVI=2Q6cW<vl4p~5LpI_YFjW_7HU{taK0can
z(8)I07=q*f>;pjt^YB3VphUO}T0;3hE9!71g47EGnu;;QJR>_w8wsYHNqJC8y@*&y
zh*R5y5}QscEMD)S-0>6S`%fXHoSDaZRSimYhN3@;Td-L=em*hoqdE}fCD^Q3W!9+N
zbJARrB2ms8#Zi5y@z}x4jM0=){MoZdqw3uu=pkuM8cYt&^qjJ)s$<oMQ#i2+x={Rq
z3_ubSr-t>tLW#+=lM{p9TlIYWr{co{?BaTz8T)4XX~J<Ft|IZD)X%2$LP3(Ax|O-}
z26QXkoNy?i3eNXUnBPrSaK?tHEZOlPUnO$9BXkY;4ubHDktR0M5b?B6r&9%k1LI+p
zWyIs0#Qgf8qx?84;KNcd$T2&3Kdfn0J~F0M$IZqt{)y(NUTY|kdfk-i4Q}(_zn~^T
z81^JRmQ+wxy`z!ISs&bJ^%Q(!&5B;IZDvMiq~*18BP(eLaOXYX*8RqI$(!_&5waP!
z=jIGrvr(o>5#Q?S>M-$0^Vt*U<x2|5N8CB?;DX#0)TtHMZCE2`bT_1>W7tJU=ZnJA
zt!`Q(QU$}J_(k?cP7U4E1e0Gjratpi1;lTP47<npQ&Us3k@ZhPwDO>HdRpG}Ub+Cu
zy|nHTwpzNtN7w7pC3JE-`~A8mD38BO%sVlEMC~@+3qsCX<(Fe;kE^E+Lfx!?fPdKY
zv0g`q$llGZN;X5U`|)44(}xZRWmDAp89Pw&A^Eu{tb9~lyI#F|1sZu%ad5!!&J_0Y
zZ#{6)8sHxdjg0|#_eRnr2tuW;@=2s}bywo6+qL#q1_L%t2BexZ_g=$j8=Kpal>^y5
zZi_B9Vy~?&t5I$>U9QA#m+H@(Mcfv2hFbb+4uU3(Bf#DoTF?=9aQ&eiu*GfF%9pS}
zse%c@OR4lMR}pXu?}jsnp`m#sgiCj4ys~n$rt(dx#Tu3@J+7j)T$i@{K2h#)j8q;S
z6&2O}`){}Bn%!DUHOH+OiTR%+)iR%DP!wrHrzHBGtXbDp5fCG_oMtgLh<tajb<E7x
zV2ILc9am8u)2$Kh4wu;-*Lp%Y^<*F=p1^BCRI{S&Ff1;Iqn#u_Xx@3r#K=}^K2BUG
zcsS8Gg^Pz50@a5%{h3$Q!?7v*zeDZ&FJt1MT7KiObO#r+CPT`P*$E`y&9eE>>vHV7
z*IQe7Ma9IBaz>=M{&jUCBYMhzpKXXC&Ac+9cM~n)#Vh2xq;58vw>R>}5?FtJzwyV=
zP-qEwh4pk8;+Z<cjqN>h*de(4#5d%uVD-~{e0z=9C`A#t1e^>hSlf~LiUvJfwbpc&
zN?z%-M<tXsX8JdW^bS;p_v6Mvl&N&I)L1>Urd%^1NO=)4^y}SVU!S}KI>D%aVRip2
z9X$lK_(y<UP<w_}?CR_55mg1i?kxTe{aicwX;m3K`rLJAp*<pkFeVP_S`>Cfw%@*a
z)7x|(!}Mh9<O#G<E}>*WyTfPqV<%lUBN3<_7`@A|s9pz@rg@<#RF%MOIwEsC&}C>_
z>99^OY`(_J71d3LDFmXU*~eJ>FZO)N?fN?d*=7eJ@iOTh+LL(P6Lkz^r<2w##q>s}
zKSo#6($YY>YGX)B1k718QXi89_o9{5J>`K45&{kZO&hG%&I#~-=l-=UJ#-J42aTW!
z;#^Uc6qv4TYHEVMnQXbaCF>nH5N*(=A^j=KCCI=4u!k{Bp{}jn6>{ToU5-}KdV8D(
zsk(H^g(3H4^qkE#_rpoY%i~_pG1HMtXgi2c*SPcB|K+gzBOlrRZEywJ?yKJ=EXF)Q
z8Ig|o2D)}7oguRr5|YZfzq7IN53M{(K09L{Ogzv|zK^Z~cILt{Jr*=%)?OO;F6{~0
z639T2?I8JOq`yqWym~eNG;QrGIk$xdRBg`1=zxA>U!!xg$XE7HdXo7iBqWOL>7fNu
z&%hQ&&L6{+Bm~(7j6g#}^QfCNS+=!h85%-IePRUGYYl(xDxl&2$F4#)neVxTM5%HA
z`{Am!_*csFyVsX1UOd;=r;+Y@`}Qpw8XBgL_si-9p0k;ga8o$LC?7C<0Pfu<@F?i%
z1*QCI#*yyIp2Eh*hwc(gn>H60(bChKOq5vz>W+5&1)>6i1prCM0a{lAr*s6(i2j0-
zkAabqCD1f8F5&H_TSwD(%_C^yyO+Lv`BLwLRn&@Q{Ak=0R52rn8w4gkS=-<DaJArZ
z8poQb0ZSPN7woXaa$4ws@6n@20s<PmKJ)YQKJVXOxpD=idLA7)Z3r%b%>CH;)4B?H
z=x5gr-mO)(jBnrEz^;(uvz_s&I5_O!h#{Ym3VXN#fq{tMcX9Vio#C_PO9u7?%{;Sw
zW*(j+OV87BKsK~|4~kCOs=t9}XKn|2!BW(MTwKa`jo<o=nK^lN7L1o#PQ#prm3bW<
z4UIm?l4Z)nq2&C;b&x=zXR_qXC?2y2o4~-tEWCH{>C-0wR((AJAVp9NW&4qmle-VT
z<PIU@9ngPLY}`KwjV%1@*FSvxm?YqA1Dy&}E!9lrvAZbpjs6FEd@#fjb^52EpupN1
zWohXmF)J4<zGM(X46}$H;{}5&JNzzky#?~>+g(A;06c+NE1ofI2Y`H=u=st3F;lIV
zWNI<6YJ4xwvgoF+gHAIpEG#UPfIRR`@?|2|BH=4msV*g39x@2fE5NpX{MrYy<9c$?
za?f&Yx>mRqsS3RiNAeB|h>|RGG&g9Afzw#lo^zGuE_}It3IG-o5<>n&KQJQhC0(Qq
z{|9MZRGSM8_a4_r!J(|G>HN2i(74!yh7YNr%P*u^4Jxa-#e>s;WCbA1_{rDbf6A&6
z4-g4KB{=}eiPCxDCKq?2blJj!{)nIXNH%h!+!k$bb<`3tX)Q0ib&U^WQVc|z{l{u_
zU_xCvwiUYy=v#jQQjg$~ec039-rm`1I`FL>)DcRTKlc%kOAH#LU8-x~FLaTQ1q6~*
z{IbsIyPz8l-ScHHFGO1Z9YVr^T2Id+<OHCqmzNhDrP865t$B`W!2@b)^t$l-Vq<K&
zK72%<`In0wAGl6|EMW-)wbEU=B;=;lQV|@`ol6x9z{JF~IddioXiObkN#Hc)a@v}C
zV_97cZGZ?GPLrWW4<5YdRN~D&ufyDh9s|Yg+e4+6+6Q!5k>KCNELszw5r&7~2sP*V
zH00)Xp?!HWs_8h{Y)yfsZFUX+dn>i1f{2I+_`Dt8WU}KFuYTK&P%&igF$)imQ%BTe
z06>MQYj}7DnyR5ufh^T~3T|$0PoLHU%9IHkx8-x*h8W0F&Kc80sB|ZCWAy<i)+*Nr
z?F~Flidh46IOq(6;0)W%RyKu4iHOvpG}3IfNGREzc@`6KadF8wjKcxo*bP70D1$aq
z4Z3fo<K_})aSpUWlsusj15ycBs3K?tx9ZTdNp;_e=Z%({Ar+3-spP7_AsCdib<SUu
z;FvM~`frN=Ixen&?GmnZB$??OjZRv?B{MS7ywTdDxnO9_F2KAR2%uet?6Fo5#?W+I
z1)l&8^x5;`7^WFWh=}%A294mF0jiJ}TKc})kfpCyqMifY6hxG=kQcyUfjz_cuXnqL
zwU^{|fIQ9^2DkXf7~@H_-TGJT93b^Urvr$}lS=*rVEDc9>;ck@4w?DoNgW8K7cXAC
zemw*Y>fdG7dXf<oo@d)IVRkYnU&0rJw15z+Ift3Jn)Q<L*N?sUqngNW-ed!B5Dy`R
z1r*GV01Nou-Lv?kD5UGNAgV8x%x7=n*Zd&rp}12qIuv51&bx+?VsG7Q{LF6#<>|9$
zO^_Hgg<BZN#sjtT!U6+z^!%eyNfnwy8V$+3w&D@rzI>sI5glVkLU|O&Wloxvf#5Lz
zZP)kZX$QfeOT)foKGZK@pJlVpIE~8c;Rd{q9{p+iME$w7P(259ATxC&_@@xu!T=;i
zt#kg)?b~+a=B3tdj}}4KB@X_+6NUk!3lPlN5DnL^U2~kUmRQu@$h<d7M30b35vm3h
z0bY#4imX7(>irSJ?u2dk1Wyp&KzYA;^CpyzSv$mt);BL2@>AEK$%dxGP8q*4nDLN~
zuA*luZ$ddY5zOHZ_{W#L0qryWO_Wq8ArT*WgqrC)G@6-=N=fMzpMe%YUN-uz8#lI^
zZpoE#Wx7T21vvvVf!ZeXY@F)%{Z1R`i9;Nhd7Qh9Oo)Bo&vFiVS#bRU7gq^HeeN({
zV|At5#zZ;jSwDXKh-OeJ431&iU!*nL9DAf_pYn6-jm&45ynu}Zr8<Qa9nB4TbM&l%
z|7JmBvBxlwD0<Iy==Q+y$S$g5ZlESBC#RB5TZHxZHo!f2<CWTKz3xF@Utjvm-``=g
zniW>K8C}10?&%Y4z%Fy#ycT;nO-G!xYN1&LK)4B5#KEx_^TuKYMi=Z;)_xbS+IvPN
zP%~?%Gj>X`5A9d1B8wg-GUgZ$rsKsLw!(*KDk#iVV0swvC8PL+hQ%28e!*k%+OwVD
zRrd#>=m>g+hipXweZ09_GkB<R7sy+X5+b6j1CoE5v3(#FYO!88TWF7j?xdqV5%ZxH
zf{>693^{*1GGPt+T%tvVqzoh1JLHPp^)jCyWT7CpJM$Aa5*G36m21cnagonf5Kvk1
zvO{|VCRqC6&g01M+_!O0&$7*W4|`VZ3ORmeC#jQbxwiUqJiFKlD6Dyjmoj@)<@MsX
z=2v&%OaD8ZS$H^dHP;NfJsma=l;#pE)Jx2Gp&8p-x+Kd}IVof@!NbOVgq(uN4(P7S
z#GjUKOU}6COkO!NGXoJ*YC77pn^P`1rYUFOcm30>%fZ_5m|hqxmH))TBT`7Eyt8o4
zibq#hckJZq=z0F~O=Q6>XW=PFO1(<vb-JZ%|E>(BHY7sbHD~Yo(~!ucUQ~t$h?GU3
zOX8CfMk?ahCr|QDPn*aB3h!~6*iE)Z9&z!Ht=po_j)kRb99GWp+n|yi!>WyIVq2Do
z{)1P!+o0kFsZsfnSFRq!p?H39<j%ym$);?K24)&=cG_wDijf9y)7oca6~cv#DxdEy
zR;#XIyeX0Ng63#lHIBDlp0p}|UH8jl-CmjUKT*Z!Ss*rz8ipYy0t(Mdn(O`eiF{{<
zn*Hc!pN*7Q<+XOLjs;cawFL4yEG{laQ~K`HA4>E%50`idMBx>@8|ZfJB4e=?-e5aj
z54FDU@u+esY`cRN=9Ikdxl7e_<JW7aw^0=F2gO@<YIbe;6!N9dq?)m1e3;>9d|y6&
zw8VLB!mvnk)I(&hg7QFs?Cfyt=_LGuWALi~%i{e}3id23)6WszpJoiPY(ZXRSqn&M
zZnAMVtXPuU$utM)EX!(=CGlYsfdP)R=El7J8kLoRPVZ+yfSdOLp%VfQLfpU6{<4jF
z&5QbsE?y`#6`b61AOf|7y}`e*H$|Am<HbwWX9sal@C<NrC&~qwF}%C+4=J<KQq3HX
z)<d1!^q&Cwy->xNy)a|!)kEWP-l2Kco$}0lgUwdMI2+ndpDjDcot>S3lXWb;XL>}p
zHg>!%UoxYHeDB)T@z>))<0KG$3JZy%M{FwHK+N6^ijRdS%*2zfjfVT~$0<3Nc#n3B
zk1j3=8fq1c;jkTI=&Rx5;lb!d+RSp?>ajc*$@|Svw)X*r$X=<DkN$<{c&F>v*$T$o
zJ646z8{M}}*k)hf?I3oR9@jQ9C=z+_=7)iSfrUkPZ9t*mvWIEhydPCrbmJ<?I7c9%
zZSjO;7a6;OjLgTj=iOAS$-aJm1(esbkM-Usl9R#0xpL1xZ`cdB7;HjUaVA3y`Tl2N
zU^pC^RM-B^Q)>>kf(bX`7|TUwMti@204xlQcUCQQc@AD}f5Xmevh!2TBqLxHkmo`}
zsar})N<<7PP$n$XY`ST35lw1o^^k2)zND5dvHJmo<0oh7EqN|Pvg|{%1P*GmRP<^s
zmbHFHgjqD(Cr(uVVNwk(N*cxlP~s5Re750xOR>Q030`LPm8JN0S`BF4OsKtz(_@er
z)B7-Yye{vx{QmuW-{J`+H9;x;XOa=V1}r5V-uYUx`ug8iHwj0)q!)V>3Na~fZ@X=h
z;Rj4wc&-(+HpxurVGtLEa=cb&7*C!SSy0M^LYBK$rd|wLGp3gg(ijS^#%>Z-xkMdN
z6eR}5OY`{DQ`z(c(@1W7+}ly-6wMgihdc<$eH5?N^wD@@NLjB*al`t(1C0kTfrdg<
zRh~~{{f@gF-75$GI}1Q@H5Mlo${Ofwe70<Lx4<{Rd34P&DlTgnh}5b_i^S*~u6P+p
zhXF`4V&eYpR5PuDUtGJv)RGYZrD?@a1(;hyX$?OOeuG}x6mYD!v(Idx!k&Vzf<CB2
z8%Z%_?2FK>Z8KF=Pb)_&IFYmpD+i<%KHrOM<4^lXQZ-~4B_qVcD46JoO^OO1$Lf*X
zy~_*c?k#OhUUpT<YFgv1pGSM&UWjV60!OPne%njk($s{a=?x79fe#vko&_NlH9}pC
zjsF0x6qv@>1}wWpNSMI!?g5>Kyu3V&Y2=-<Y0XZ^&{Un!LPbO%xEIbp>t^w297xfh
zzEqIP8Xi=%lkAo<b^^Ej3wnKiAe7k0yv{>Q6h$l){PMa7H}cXfI7lbeL<b}?F3^vF
z3i&phAPX0vRC~xshE{7R0i^9svB{z=Ps*FB-|kf=MBJr~+uZEFe(0>wiG+HNIdOnW
zs_WU+w*(C*+}o>6rX`KD&h;ihv>Ue*W`1@n`i)L;z8iQ}^e|)j5nH6mwT;-L4JQOz
zhLO%hp2>MxE8}ODP^=?vsV<6Cjk(GnNNY{?4K`QVBG*<^GqnYIhn$keRkr7!X0N6-
ztx3BT*gPZ~&Met4w+KJdn4{Z19$#)jcgw9Kl?XyoUnD>NDJMCxJgrpJ`<;3F$E)U>
zw1(QEKZB`F95iALAKp0D<*C0)AUyPUB0Y6&^=p$%=JTJ@rf05?5*vOtFX!^7Fj$$t
z(mlZb&OE&sENx{R-y6bY%A2&z!G9b~s<Z*ufda8>#_(?lll@H_{*iaf4J+BdFfW*}
zNaCy#M|vKurn9zIhBEo8th<!fY!eFy-5q%T+;nk{PR*F#zU4)AUEqw+_am0Uu&`vO
zs5r^uAFbJ(>J%O6ddq1|y&X#pwfCB$;@qz(dDaAnPTIdah$&%NLUbR*y?Znc*R&-C
z7pKLL9@F?=X60Yr6h4+Nnk|DhWj+@}s)KIj%Wq!7nAa&0ybdx+jc=xKPU=wer)dHM
z)1p@xq#Yx05Bi(>#iZWBrp(GT`u<k^tbF$pA*E)7;=j*XlRo2*4PPBFOW!+n@Y1O{
zDvQ#&aL<G*o9eT{-T0&(2bvdx4CC%KWo{n~11OzR;XI7<W2t{vt`6*Iuv?O(UrFIt
zx_StUO=N+|>?7Q6{DTjci$}WV(t(M%CL~_Qv7U~77>AF~D}6SCIrtb{H>Wl(PgF9J
z$nN0Av7w7o$M$`W`7rI^#*w50q8RPPBnxT>EG0Hg0k*c(a}DPg{8JvMaDk%IqV^oL
z#>QXoK4VogXDV?7y1?FqevjEjhAk0(I=i4KT3TZAJY!QYSvxykrj2FhWMc5C_kw$_
z&BJe{6RKO+4&{ujQMXtB%DSnDB~mY6Os>u?p0sYPsd1Q<txh^_(m#$L_;#%)rL{lp
zZ|nzdZ2^b(FfKm4bJ~%Nb4j6cL(&avvRm%M>Vg_$*$N#32B02vvXXv^U2e9rKe4iZ
z)3;RdNXW%`K%VQqeFm~#>-sW2cVmz#%fVU@iPoQ+goSth*d0G<2?b3jnLE9kB@*W_
zERdyS9~Pxx+3zGGn9=;^w-46*-1Cg~+lXg8WG|!|y17T0VhlGK`?Ad(|2r5{-6L+M
z=>cNm;F}zRCg(0ydkR*DhF7fA@&ft;`}+7KhRpu%Z0^habCkujWP$qLy0Z4##;1*%
z9KF`ZPbPoAc=y<_n~&9Uib_Yl(cgF!UrmjoWl@nech<N0eXp`E%(%md;UP$NPkk_9
zT#N;7Eq}*dsBq(mJ=!*Zb-B@C;;`vdelgYlWbNQ4wK()M5N%!0?Cd6WgEkdSHjwPv
z>7u$0JY<fO_M<SD9uX7c0Avz%M+hzzZTEL12R^h=J{bDx^ymJ!499_$rG>?}b%;Av
zLc%I%KFtq~|Aa6R5$?iV+JgrVTEZwTxe`;LI)ai<!R>Bf?SL6cOvAASkF}c7Z;of;
zU(LrSxyHZE>H7K(IZG#;KXGSEa)QfA&^LjBYVYcezSKsTFKLSPE^7p8R@`1|d2%w>
z=A6@T?RO<QR=(Y1^<QHatP6kkoenk6&WcL)PR<QBf{Q;~?`6}jcH!Ln{kjoSS#Yz$
z>({r{-|fu8<Om6imd0#1ObWt8Wbh-}S3aI9Hy#|%9ep#*`2E2Yt7;d&p|QZI*ZtZx
zC^sH&Zy%jbvpJYea*1=h&lZUxVRdk)3)<NS6+eEo>9G2~=G0|JU1GXL_u9#JBwG#I
z%UUcC&}I@Rsk(DNU7OCnxZ2w?ux9n+i=N~51B<Ggv$LQzIaHGCwa%R%u9nblKTI@~
zBhs$gczve>dajF5w;@-d0-~j%IWkW3H;&|zdsx34bDlWuan1b%=9#Vb|H3@rGzT4K
zFv;jT38P7~#Jwg+Lm8sF6dcBVP;`O$m#lb*8C9)d3KT9aZL4rx^P|4Jcv5`FVnbtW
z{m<{>M!`cj*BSWNM7DN1wQyUbmuh5CCqJ^fZ#tUmzEQpmbpqT)ID@g*4=m{YRJAF-
zbG(-x1EZU!<EuCXt+|?kVFAPH#vF5-o~?q1L(Y1?&yW)JZhIdUmL(Wn_WoFJ%{)~S
zy>)o33khjw)4}W1S9yb+U$>V6F^EMlb#OFqKZVPrm9y2%9vQPh1F&UkfFr9pl%e0`
z885_LoTEJJX1#54S#txGiH_a50}o^s&nl6r$dy+$M{tv7${TJpecn3N5EVmiAv-4B
zA5`XZM6JWs-DeC}!{kopG}ukBeh~GLFgOAMj}_7&7p}PhcZY`RYAc^MIrhgp14NFk
zPP<~kJNVrSN`}oyMo0oEf1Md*YWi06Ee&yzb~wPbUInJho+#ASv*W{TvwC+a(vGEI
zcyv?ik#Xd!^#`vzm!Ht$8APs1%gDU0#MYV(c9&ub;D>Q!7`gZLyHyUNe~(GAOW7xN
zq9f27poP81`ZW<QSnMoz!WjL0yEw;qTsX~RYU;+_^|EqDm+Gk%I~R(cmkH;;O~rv*
z;0MgZZ8R9LA|$p3*&67Qt}fDyi_IJHBI32OatZvqH*aqWs$|Xf!Hn!G=vCV$&Km87
zYqPVqo|}6v=OAng1~r#?-qVFpep6z$E%259qqrWBQ+4%(fbssdE7Yh#ZPu96j1T?B
zx7D%`Fj%dysC{FEN2=>HpXY^}y83Sr-I{!h)N`QwyIU1%6+j~iVw4Ba$E!`cvmmY<
zf=uaL_(|QM12x3ErLf-c<U5TdNKnS3xycV_Co)D?^EAqzIBsaL%_7=RMPpOb6Vecb
z0B*)2hWIl>smS();~A-(F1SPoWAP9EuiCyl9_#mSTi;YhaiJ)CWF}-}kB~@4MwFG2
zvX#9XBqUoTq>>PlWECOVsf3J#L{|1*_i=rH&p*%odOfe_pZmW4s0i2RbAHbAJ&yNr
z9JqY%(H3QWchRI|dMFK!H1>3G2nZ0{o6|vcA4CuC1#b`FmsO2Mtbws{sZArT<sixM
z$VfQ(o=2snQo_P*aYsun$|e@qfDydIy0G5PHB6ram(Qx^sz%FDnPQJ+NpDDq7-J+`
zz(P!D{ijV=M?RC`0cnQ#q*X#E^YMdwIcs<D9X)!04(D7w_BC;t{Vu~NWqw|s+^E;|
zm32LprI$QNg&4etGF`2@7W%$DHeno()v6}>4A061E6;6X*Fz?riVquYWN;5mE^j$_
zOn)d8$Y5QqJc3`+AJS~Gswp(Xy4G9^`*LMr=F=4xRz6y>mml>Q<YYaI8Jl*4w*28K
zU8?q;C>_n^<$Pd7Bw%*RlaDHNTGJozIwKO-_#v}?GEj;HJx>j%^2N;-N^V;P<n~?Z
zm#v)be)aT+mHo;bCP5e+jVul710TbF-J*2nCCk7F@2=$})Qr2@sF>~BsjdLl($)P8
zm_(kOH)-APvS>2iO49NBfTgNrZ6Rg#{0GxqdV$5H39cAwk~_>z=d=SX{p!nktZ3bI
zbqRihmQiKr7HNVdbL8X<?Iq%mI(8k_z@uX2v0XlS@}$I-&+q4B4mc_3JBTK~Dk_3;
z$xfZ)LtA$AP&`G^xjD7jqf3>IJq&c8!Xc*c27C2rGBvmDvMhydk+hHy)%B#SazgSW
zJye%ez7^{49uN8P5_GELDPH_)7cUL~W0vc_xuTu2uB8UP_MqH}<-*;C158Wbtn;Ik
zm6d}t^MU!UxLDcx@lFm#5`9}lM08uUES`L6Sfk(^-J@jiY&bT_;nKpNFRa>E3k+p3
z2OR)0HZ7R0Wpn2jhzv}e%T5k&kbukMW93;DBA`KwwP4YR#UyuEmQYN)#mTfz0E=|1
z2EU<gkh5($Xz2%zrraGFPl$a2!FaE!srfL;lCrVutbQ*Evae-a6$vRR=^ay5^~0Do
z=9|e{8l3<PimB)?A7B(ZjFBH9G}VnN>j!M4iB=+nWrbRCLxhu4`MaSzU~7Pk<dn;h
zZD=Qp*2Q(Y=LswWvn=WsbevYZd+-aU%g3@cv)bQ5HUWb9em7f8^52=W(bMxxDr4f#
z8LHB1TQm^%fi<TiWCnw+LO+`{S4C5Z0fK)xvq@hY27aI4r#^QeK1se99UewYl(|vZ
z&KnI}QFkVXe*~^iaqhW!h9|Cp!YHeqd}0`b7iAkJs_UUoSpH6%d3&#bqmEJD5nEcR
z>*x}JXBB=b)NQt2#!voOU++h&-HiF&0UDaYMtXaXNi<o|wTx&;mHwc*((`%@4E2VQ
zmNi(lF_Tq>0`KH_am@vp7Fb|#rVt4@0K?&ItUMz<bFLZz)fTpSdl|(W=AEP0O`;EL
z*humT3Ek78c`jD1*v-Gipj^B<p{Ec|E4GI-SCOoZ`)2Lc<by*BG4i%;wH=8Ap4SIl
zS#zAE(lvLZK`CCHJR}^SBt1hc!o#{tdUo_skD5%J>kJ<_0|i1{Y^<u<;hqy=$vcUy
za>t1Rr?Z=~9%0)wR}>Y*Hcd=T`yf;{7`=Tjh>dK!A5GjyO^Imcifb{;<bmTyOcGob
z+-ob}0JmAcx{HO5Za#f`<1$33_6#464!PK^qt9X{`+NV+^loz2rlnu5l^xdkweq%X
zGM<b&<Bt8;JUu;AOO%qodY!czQSy&EpZI%W!QFC@)NAg~GvRvIp|%#8?PZz1#fHYl
z*<r_Qry`Z%Y2%k4Ofh>X+5Y>psBrxA35F%E3{zZ~QtVT7;^Y;y#sT6U;|7+4RIGA*
zJoMh~x9oeVjEn6%P;v<KJUpxcSPPCePhPreCJJWHTMZP;XL4(RV!6Mw5tJGXo6IUa
zol<K%!p#eU7CVe4{FA$uz5^$zE&6TeZ~#O>i&9q(Fn*m2?bnNKY;5EpFX1$y1~Rvw
zz8O+qX^hMrI$zKv@-O;sLWIOIViddsGggRg0vltwJ=Ui^8sbr8p^kfDQrh!%%pH_S
zqFLOB;JCwfNN~W!t!gl<f<eL)OhP}IJfA8g(mt>W$C>lbeb0I5(O=PH8-nQ*^eUS(
zwbUMzMke5ripi}MhW7k5LwQD&F7?^GIC`sk28>@ed0&GwrSqe0K!NQxMnaC@xx8$e
z02S4$rtag%9>?RBRZh;L$wYzT<u=x1gUGp<Hk<p7aUUB56vRGL;*3EO_){U9<{YwO
z@*9Pr7eYWGFiLfK^(|xH<>CwS=Z{rTnSF-_MJ6;i0qL|YRo3Wx@*y_1^puo|##c{@
z*F?J?Y-4mxONTXNlk;{(objOIMbk?4U1x%9qUKfIeDLO16;)ML+(2MpWl~_8cGmo1
zQ>P=#VE0rP1!_3eCqe}VnFWHvglo+=4v!t)n~PlzOzPxDxow)U?wGf3Y;2T=<Q`4c
zHad5XXL6YW$}n<v@UpnLRFnYQc0u+4wBj8!V;;TAdD=uB8VOEVzKKfKn=FcoTjrzm
z_AJl1NU-O!O{8YWt*tO1LSh1Nis6E=fIw42!}l_`EWH%a=L8^WJ;%hnx!qIIUdoeS
zykmfu{05<u@DEu74~6gG(eRrT2H})^eBqt@vN<n!MT95V|0ut*tKdLhu~6xbgu0`T
z-j%)#%Y1BQ99+&BF?n<M;b2nH>CpL!K2z0{m%-}k+gX-LEn+&NZfVJ4yjV=^yN@5$
z*x1<SGV-{&xu>S43=Iug=!FCY4;Ne<A0MB|=kE%<9mJ?BNDou0;}Q~Ks*guTY=fhf
z*#EdX#iyjKgLDtZFJ2H&+)&arHHF-W+sIQ$NGK;Khp3mq?v@tg=FY_660hQ733+)I
zmmoX4F{mnjzV&JE>^ujrsu=ZaZf?KxxZPzoF(!lfu)1V-PfrijYbhxym~=%lym#SG
zZ@VAJT%dkfKu|FLi0~ORGZv0`Xpv}XZ?3mZOiWyD^sk6|jd$SX-BUl+6GoG8y3VBs
zqqbT6NUx(jv|-^95h2PPe;G71HC4zwGK8tB=0O6)i&y*Md$Goj%D~O?^71uw9I2J?
z{{gG%Ddp%6Z{D|<@q*yuEywzo<2?<oI`01cW#r2tFXM8i5$NQgsYy8UFgdv!Sq+UI
zmg7N6N+VtB?c29;7kFV9gr$SK*_+p|qvg+lvMtA2@bsx_|5`aheVf9%pn`(J(W7}~
zWyu>W7>-3PKeRr;z)%&qxds8XnYaDnNVVL&Jb4g-aPA@^B4+QM`sh119JWwnRVXDZ
zE32qjRb0G?OPTvCutK|pvzV5ahJA)?2yX{oYZ}rUKa~{He+~vr^f=k|-I-n8KpjY&
z?G2~gqRw4dSfI+rp5zE-WMXP=Ztm^r84mc$u$4Oh@#9C_1u*U3!^3j$b(38_;oG}s
zslES)fteZ1)yC23k%iPiL;zXn&yJ4bvNDvR(sFV{EiITRbBl{+mHuU>xgA|ylE;oU
ze)=SI<cQm~Yp1oe_EA%_$$C0s*bBN8HWEgfLi7<h=~;Ps?;sVc4W-V_%@s>-!Dy_t
zyL%PVy8F2n7j$%@K;c>4z(P=x-o#u-2(pOGOl~f&UtoF>MU@c|y`TVK4i7?AVoXfo
zrDpJ}695E+hK6G3oYH91w8&Kp`s<s*LViI(s2i;cleDwy>*`k5*OB5$m${*deGC0M
z`V%DB;^N}U%F5%%kK^Ca^hl$f!_yb)N@=_(;+mp&{=CoHf*rb&sVR9>wGck~ZyiIw
z1H}$hn8t>N9_WX`RgrY=Ll}?_-0%kcY+C3l6|4(EwY@#?&C5h~O?33Y#6<I;D<<@T
zqn}t1n{Fr(7#JiGb1>n-$;{5p1-#Cp7}!@|pY`&kZhrk7Bq(=OA0tj;>h{a6dUDe7
z^5x4LKjBGo6l&w<-}%OcL$D2nIMfvuBW`Z%Xkc@4b5V0}Oj|QOHyB4o25Gg#wv~}5
z4r;g|=rog3Qn-R&zkUtAQ?-pK6s%mFoVIOgt)ue}_V$=B;<Q2=WN&XTDk^#|-<01^
z%JiI0stmNAMn*=<ligzS^6%Ex)*w<+^qKq6)Pym&KF)xQj0}z%w(H@;A9{M$;8xdP
z;*y@2cn(^5C8g_=gWnyGE8K^(o`8VBX>Dy!H#ZEx3^10+&2>0=l1R6uW|eFBeGSrL
zqEs4@7W<LNfopATg$QnV)_ZGf>s9EJ_A!vl0PvFU-AkmnVI~fk1JA_=$?)4Z1H?RV
zNfJ+Ij{N>joFwCg`;d_E0PPt>NYE4@m0QE*U7`*pJpOX!z<~pZW6&rodwMS2JtQ_G
zMu*02XncGP;!&IcqI&{pX(!31aE0qjQ#?F8*qLoDEqei(ObG5t&p?0^Id~T!P9Nx8
zh>Yy)7crtHDy;pMoskMqK~ghIMW?5yr={f=6m0*BQaop5lrHV|vA@5+x7Q?pRByl)
z`wm<6p{!SaMuw%Dnwp;8*Z(xn598y-u@1X$W#r@_nAise1+hj!uU3rF2k0)hZ|}~~
z^<J5=?5^$Z>9M_Zsrbzs{0Vo&YOGsdpFb)`w2nYI5(4DIRs3~9?;UMPZvTCM8?rz$
z4m!Fx;@G;|5yZt{Lg37uoE(At2YC3{l~=vI?mM&w1_rKdeqh*#p}2BT>H$PP_#}!P
zKTfaMdyqpMrWyhXGPppB5G(9e0^>0&RR&tnR90Snt*x~N+W(bp0tx=8py13$V-u62
zZpUC<b#<~^)gs4^RqO1GPfbZ-(yoVn7?eMtxsY-2@@4@^cx5d3*7j8-=E1cIBOveq
z2w+n1UjZ-5Cdva?i)aLfCRA9~jd?^xDE6hU%3c0&henV-g8dGHId&r})Y^D|z=HSX
znOAc|y)VDVHw=u7s6F{|n~H^zmGxeG_2S-INnxVvV<W4!?@CExVkeg4TSo^@n3=x5
z3gH7d&x;ESK+kwt93WN~XFLiH*!$}{@CJym3G#gy?qAJ~Bz*1YNL37~M5$I#FpbuQ
zNZl_eh}s6H9-kW<96W~L4$cNRcM=j3{>uTa4<ufN`WZ118>TTuY+XE1kUK6pI^I=0
z=jZ45w!pdPwZ@Vw$Wh)Qcz*l#?K?@9{roxU?%kk>d7@@p`P2K>M=XB_ktX9Rg<9;R
z@F9_lBnoG2s811s7o=jrVgak9-Wy}=Zlnn~3WndhwKJs^q3%sdiHnm{9fLhKtastd
zmm*~~W>tYKRiCS?PiSf$VP}8e+shPWj+_oTvm9$%d%GJ-b)=A$XM5pXEGb!1T&$(8
zF6BM*XtGdduRWIzJ;=RkpB9#vmzS1oMGoquD-;(M#Rak(B8IuUzn&bVKYAf}i*fKo
zjn*u0x<9A98yI85!3%Sc<|oLTB46C9CL<$zueB#cd3JH}2P~Ta!q`}7?1A<ZKW%ER
z`R8=HVG{7w%NrR(x0a^#<^1^%(b7WM#o6@$#!yLTc*@dH*2lD?IDxxEwVSKuo0l(@
z$Oh+bd+qn3oQFcywTmu4;@PujU-yLN<k(-lcro<b6>&!Ay9DL@{QPSn^9u`yyHqmu
zuAYxoV8bnPr^ta6)<(;t{jkBd2$F?=10tMz?#1L@5;<tIPmK&xR!{XNJXu4-j@DMK
zwN`2>sziBH#MMwY#^BFj8M)qD@m<>-7HFAf++u-|Sg+|qSDoq}W~x0p4IS~QUb1Qf
z4j~R{X|*m7pUi5X{c__-rfJA|zWttS3-7VdA8-G#oXH54WrA1A%%{5qqBXP)^)eFX
zFBLoV$u`3GZ{Pk1>kzd4GwC*FhvdBBijltONtLhj@1rf#W$Q1I;lX(VdzG}xqtu9G
zdt!TQdV41LZ3t1bC=<L;#-)6x`l5}k?NIXHzkdw`4g}edaUgcWGOLVY9(4~AMnghX
zzza(Vzmzk+_w;xsxm>ygTU36(QjlEVy`w#|_F_REcS(9;FN=)9S_8tao}nR;G;ew&
zW_QSHRr&cO*@fo8kIPf}>bPAK6B8dJ1%L*9{f7=68r<>lPr*|B3!#;NmZ$E_(c9i+
zN9f{uSN+uiw7&O|><Op2t44;WrnJrp`%t=}&0g1xzjyD`%wTP!VkR{SMaV0!Y4dLD
zS?{+HOCtwD<sQ@er^3DGd<o)k*(*OkW~Ho|@98D2>53Lh#{LS5($dL|HrfxR&L%!`
zLCqC+MD`{-&}POmiXv6c?sAVKf77aG@|~TXiL?<s&)v764f)0qybwwhmX+zPqgfhr
zUjvDUnS~{@Yc=25kTH^nMZ3RdV{6B#fq>!KXP&mR4V=6#mY*6Ll%`o~FIijPiYP=8
zih2W4VIc^VUEZ=9ji?wN2md{#*<<c;V1MXzTceuP-c!-gzsbtPbi~Lrm8wzU>QFG&
zu`@ShHC<w*xwp5swsz;T@L9h7a`avG^?Nirj)PR5A3zt@Xmi2DL_gaAm4e)}>!&l(
zr_pKb+h61HfLYJd$tjYze1_R&tlebn8G&PXY)pedN__Ou46ok2dGpT=rG7qRb@he{
zZEJ|k!JA5N8=du*qQq-V*o;8R5g@%@f^A4V%+oPJAeIhqXuJHTO^tM|Hahysx~TiB
z<PR2RW*H|IOqnC;P`Yfu?IIu`V0NQKxb&xriAgVPGftff^<VuN+y4fP;%K(#lUCN&
zv!s9DNB`zN!$Vg!ICoF_Cwg4ed$UXkpkeD5U_%jk_pY&EVt&5RmT$iJHH#_T?>02d
z%r7!BVv>@+Tkch4y?1>*?y0O<Pj7D~HM7q<d8yj&Zm>dswI-?`{`%y7!|c=H=oT+C
zu1d;CHGLza`l`E@mo6<X{R8=NtomV;Fe4))2!~&jIfES?9fv6{Ll-|h9IW|Dl`}(^
zkLBDQLXDGk=r5|#ke^P6zKkuPL}3qBSo^~@wz{;mG`!#^w7dNC#Ub16sn%#O`S95`
zffhueW1^yrQm&7}!x`?4eI_BHZzTSX0E)jFZ;FfSE~k2LL(8*rPKyDqAYcFWWDpb;
z{rdg;K{j3%syft#9Km-ZBWDk7LT@m$fm7VbbngY7IHMEXnaDUiJUslVZ%_Jkc6Qo|
zCW|vpea=uJS_QO^xp)aQ{Q;!{r&UlHThYj<>;3zSfcjQQRDu@iPFqwAd~z$J5c_CV
zh<BYK-_w6#ZcRl?E7FvsQWbRtyPt2&#uQDYTCJpPVH9o>Fz?FFn$7g@h*z*GiLCYU
z^^KJr1f!e7#S;w=LT*qm-@$_iKaVYdxgN*F9#E^}i3?j%nVp$|?-Tjz7^Gp9xjHqd
z-ym-J_;Ghogqo+j3@GPgAor<|B|mt8qcP-)#e{c@<M9fWP6S#O@hb%7D@f!>bBPYd
zoVUGhQ~h|mLjNo^HTBsu+fX+eq|~><cJ}s7e(Zlxx2>;>C7<c@1<VM4B%)0;YUfif
zKW@Q*c?DJgAvz{B%BWw5<?Rkh%g7LoRN!h;H|nN7-f&4Y8A-kF1lASSJ8k%Pc216r
zv^4H-(Dqb$L!+{mmJ_M~N_xT1U%s3*G`v*QjUNRUp+{e?z_)>ep}ni?zCs`0{wA!~
zzP)>g88<?RrGbRF_=nMK)Hk>{w1Vipq0EO4L5VC9bZ=n|1bx2<FCzzs36NZ4wi9S7
zPuwE(!2IIbGaiNedObB8ZHx&hGGICL?Hkcokmv(Z=|A@Oud|hvC{_@CJ|YxS5Xv1Q
zKNG#-Tz+?3Ta<Ff%E}5+mmwyG3PMXu3)Kfj$m6^`)cYtpT(df{OHeAJKB$5*(MAWQ
zj{~{#6;SaNA@4?a3Tz2|72+cjWPZLg3U&l9v`i@}CdS4|IXUB~MqJq{3kzpqL31A7
zCkU24wfmTkw=_1|p-Dt+huXBEtxX4WgFBe}em1TOl;7&ug`O6b@ii%C2M323bxd|_
zFJ5G)4J(iSfq5<yBja;}92D#*Uc<vF5>Ee~U&X4yEC8U2JxU4(2Vx&v8Bh@yrw@@;
zT%)6t(<MwRSRD-nI+~kvaXsj}ybYS;`SlTr0yft?;M;`O1$70hj5{J8Gb}#D#OBdH
zv9hv)hth*ojY?D2-KfC|&lr_cXlg1;R6UVv4r-L1zCPO3S@}gA`vKwo=Z%e5R#%}l
z-Se}?>+02*l1ug99X*kj&?+N(z_SCTmcso9KpCpGH(eMbA(P}Ikl`NW<mGE?YiFk#
z@k8N))Lk%{dDOIY25ka8Jw0>==tAM?0?`z-5nwH8XCXe7zkEpu`gI8d3vk+{@VwEC
z91##GpK1RLmS!`0G3+pbBS#=<K+=ZGMri0MU*9#<4$ySrfE%Ee&1#1!+BR~QOJC_(
z6B9r1W6h*K8X6dM7Ff90+dqho)-^Gi7#uunWHdTCIXONq<M>?!4+(}^P?cb7q6$zu
zc`}ksaSO`?(}X@`KFq25I}|sBFyH(3?R8n%DmDQPMF=`Uh%W()oFofCjsV1fgAB8F
z4=XE9b8{4!c!&^O?cMzgHR`Lo(SI)XzxBZd;rL+WgkBiY80Z4}#4Y&yIQPFrTfKrJ
zn8=aK$)TVYc&f{X*qRIx4NB!KYp(vC)OLIST{3!#5Y*n{c(SvzmPOszLa4CMoIVYa
zlc}a=Gej9gd}X=?qZCg}_gtbfR!tm_quuGoN7y1q1O<r(5u2Nv32#KUi~b(dl!YV>
z9T5hI@KwUwUArcFzZoZqn`{U$u_9)CUguy;Hs5?-7!+iLJTQXe>VM^(FWSv!hF}L>
z2z*xR6${>>P!jt2oDflmL*XrA<L6hQ#Lg!ytccvM%l9HD$JcKakr4&$;zU<TNr@MS
zskym1{8iHOv$Huu6UD{F<KyGu(<uncD6AHK3h4=PNl6PZ>_LD;cRF#wlk@}{{>!$u
zsM|+|hw;DDKPpPf%O9}cdHC=lmKD8pwC3&5(9z*x0x1m(OXU5rn|R09uf^q3)6=Vf
zb%8AaM2v=XwvNF9$C8$mgFa&Rei8Z2J9qD*4L6H0G&9?r8B`Iv-`m-#Mo`77AesS&
z_3>H7WcAn3kdXOXpbYCHtxPxBQ3m2&57TONS;e3*F}ZMosO(KhFf=haLK}wfeadjf
z#bp~M?FXx?k~=jtH8gxyW;jj|R`%?~jt1$^7uf+R3A(HN(o$nVdc=E0zoi$5MnIco
z3u)=-LX<6mvmo?A-GzUnf7aBb+?oc34VcKp_&C{3u<dd6IN`Qk|0E<P;w$=*PY|TN
zwE^9Dcr4<m<eR@E!WOiov-opJC$9OoColgb&Lw0Fs8{iI$OCXp^M{uq#My|qjj6{G
zQSf^6W(gY$5p!pMaru+X%#-Yd0h3<<%HTU$hVBw^3(g_{S75LW;LivX=2kv`g&<^c
z<&0?dJ5b=@VZDCyW~ibHCkqf4Y{U9sxORf@0Z3}2O$B*W=zCd}u7z=V+Yp7|{B?As
zMuvtb`^9$Rml5rE!K{ENXso1Z|Fv~~IcWzz$=!87WEJfzf=NIr@e|iyjiuD|b#^8K
z&IM-X6EwFUpZ)W5&DLNEU969sxmQ}(ZhUG%GeAEHv-(?GYq1j*uS-g*Hh++Z$T!dv
zznrdV9fGs_>r3-0I1$8|8*eZ18u+f?q;Rvb8OF5kIsrkoD=R%vpji_Ld&8;YF_2*^
zJ_<D>e&L*+t|cTkzU%c9#H;$e!$0Z}{?_Ge@{20dM_&!)P`UT4RhM~Mdb%)uM8ynQ
zXg<!6rO3hXy}Z+H9iE}Ur)%2z*_neUE)VYa>7ya#*iW(UAcd+K`4T$OwbP*I3RjA~
zThLE$va7zi3ymuU<>=_Bk%dLUOIIAK;kJK{T(798sIfXBMV9q&Fj$!*uqnq1Ck%#w
zDwcx)sfK;~yd{bD0$5uL%B+@0N0|P-Ox^#i?mr>_x$6J+;r5G~GnsnX26dS)23*A%
zwJa@KVTSmeZ=u4IKvlzUDMb;Y{N&4i`uw1hixgB@EqtUW78Z~5Y3u99B_`4b-*UxQ
z5ijZW*7^Uu!L`pWcv})a8on0%b)EJ`&EEh2hcVQ}B1Q%Vx_t2%VQRFQ7TX^PQRY8-
zG{M0TTkW`{B%&-xH+k9FF<V_XcJ9GJ^SkI)QH_(@(C{1^ni9XBYV_WdCtRVm<P;S4
zc6QZb+ZfnMNlE?r^QW}DTmg;$Damur@@XtQ1Tuh1epi?(^6QIoU%Yq-m;vZCqU_ss
zpSJPYZ>6UBP;}wA78C$;aEfw%(N`YF(wQ;l-By|7EuWUEC&9m|YN1Fdk1&7heIlz}
z>&zKk$P9CssugUKb8;@|>CJ6*Vd1W=;WQ$dV7d;w3osJ^7S_zyS<S-fR9r#=NAtTT
zf5Q-_u4wU*O)<%VtxhV!PA?7ua4U8KnSVe))%KR(O?J%sku-7Wksx1#CJU;;V&}8w
zNA*QqM7!m^yPt$g8foM5R5=SvN}|HT?3|p4+LNlPAQz{#pJTJ%P=RIV=cgqltpI{N
zUu26x{u1#g$ad4DO0Z1-G&=C2zDPfZC-djZb;_b=&qh#-H2gp`Gr_nSRbgF-G$$uU
z<jTlP-Q9<#RRFi3`vzy_s=NE)uG-(%Kt{Npg{1!D$B$Weps1*M!^2Piq#r?oz{my&
z5wZXG)^>MafN>TJ3kwP}Gcz+b#lSOfxrz~t(I@^ILmu7-<uCgkU_{yv#L3$b!6ZVI
zftV&J>&Jfm)op1i)KyoP*+%IWbs0u!YC>G}5zS3a`yWKbv+7I%zC*n99s%P309EE^
z|1Bf0S6ng<GW|sb0_IVdvkbIOSIt^7jG*-ZApK9<`4Z<J7`k1xgr^ivErt}z8Ln_a
z1(?f4#wjUjq@uzlD0t>Dt%tFRii!$?ApjwwC!%_rq^Rh4ZH-`z`WY=P*l8hrgZk4y
z31MRJXS}~E=NHg>W-Z(8?4cq|7rRWA=6vOfE--AeAxO^_7Z<&4!SX@Kdt9`p>0)z=
zplWEC0JA}$m{_&`hQFvSQGf%&1){OAxQJ?LDx$@gyBL8(MWq&-8`w8;0ai9!Sd{CI
z^W{^q&NBVQxKS6IEIF;yL36lKpsxD*?Mq9ZH`(DeirT={(%H!gr-_5?ab;y7pgl~6
zxrw)WAEvM&vg#TdS6y8JnnFexllHSDL>la;#zt}-3}Uu`!;+p*J#{Khtj_9KL>)X%
zeX_DqSYAfl8IC2(Kxsc$g$LBo@CUeJc*9e%cQM$LvvYHBaJq(8<=-1~Mgc&U0i6&g
zIyepY4vDGVAm20+o=PMFw`A};qzav%`>k2`d%mhXnMI>~%zHcLx%#^pmh85a_^n|T
zmW>ejf62;Lz!nhcAMnJXg9isz4Fn_v1@Wnt=4MS_ks2sSpBvaC%%cQ^spG@w=<B0?
z-rh6bZ_%5!rplI3fAtQ9OD^=Ta6}w*#Y_(Mz>Gf%E4hrvk8z$M>%AeP8Chrb&S!qN
zaQC-wPm#s5v%?5O^Q)-9pr^yHWf;XNUFQ@F_T{Yrp|Tqu!o9r}I-;?iOx<ID{_Mp;
z2RMYH181oMZwz(i+9`YNx;0F^ux(cSw=i(R*zftu+n3I^U1KsoJq{C}U}*d^jWMIO
zwU$Q_GY7;@5U`0XitPbW{(%9WDuJ}pY70hyr|urS(N=*3sA7Jsx4YZJxDDkt%ua7+
zVQ6C~{=is}XS+Als0CxC+QH~A^YimKcBnf5a~2herFD09J_F8=f<Rlp0)Yh()vWuu
zbDdZ+@|%P(Rn+zHvz&Y#dfL*LzXE=vSFnRHZb?299F30l+_}WZ*VYQmn}npLyQ`~r
zv|)=q)yodAni<dh1_g8Fl`oufMTLbir!$utPr5A^acM}hq&|D${VqN=R-AvcB{ZP!
zF5v;{CM&BU@JZ34pZEkUx~rq3qqB2leiiNefiTr+-?GfiOArO(IbtdZJQ;18@h-)0
zn~;O)0cnYh+_%-<-=9`z#m&ph6$}Ow9UUDvtp>V26htWCvK(B!ybAL2v@|sxTwI_h
zG=6z;hcc#8fHoke(&3CjL4%ioAeY^{cryb7bRonSjQnm@<DMCz^{k2IT-5VK1BRL(
zn0tcyVYELeZS^<e9Q^o;;R$UD15tzwcESecrYNrAU1+HhQ=V>?C8KqI)|X$F+fHvS
znYvO0qXm?gK$<XhyJidDu!PgF=X?|u)dr(fC_Pb>ejjVz+}NNOGDQpJj-U(Z8L5Qd
zzIIy;JzJ=D1^^dO95{o~qhVx+hBO*}UJMMUGIV9kb^;8s{*?FP1&Rbn|7nzogHUPd
zz5M<P7CXk=KWldLhnY9sgNa93A85FUCP4IIhrq{vpURhCLPw_!i-Jwn{^7$gNI#}R
zMmx&h=B;;ooYq+CFmKY<cn}|WTbZD0;j%MY1w%8~0?$2ES`cjbyR#!oUnQDFx0HG-
zoLF@H(TN`l6o^jE&7A<~yXJeum(phW#&~@;M*|ffv->4tm35l5{MElz;Qyme`EO;*
z|Lljlo)~SBWuS|Mo^Wx+eg{d@|G6`gfdH^uj0K$?9sO5kGSIxY<-f`5+)eU_-?eV<
zbv+<9$kr+E-UR@BgT92I45d4IUi|8We^OdKuU;*~%pBecSRc>n{z!}1n3%zV0g;RE
zuLDWMgJBN_{D(pOCkWS>KcPE)YJ@HWbDxWLcG;M)W44R!zv*-bT`e*;r1N^GPSv-z
zP64wDSo%xyS(t`|<oC13kCh1@FfLGo#Up`uLI}!W;Or+Owf=bbn1n=LK|xzv8~B0%
zPEf96Uewys;^^Q&Nlu;&p2rCSbgdX~yIi^Ay|qz><^Yz&4IPQ`@fZ}saUest6b9(1
z%y1q5cV<HdU!VK2<I&HvW54GAA-TQ*-(`eC{{7*oRh8I31OI7n2gr)3kJ<*H1>*ln
zvR!-jq$DQ;*cqFcsKU%1p&2+PwxzEjCd1%#N{EO6G6B+v?+ar3hXWm+xZB8?;Lq5H
zXiY8j^sFo`GyZt(xE@G>ZL=~198Hf8zo#-%?U$>RH&3LR3K{0V^WJi3fAJC)E<m+}
z0B#GM2+XMSudl>ZR&Jr{Kgmf`CLoZ&NKX$dN{nDbgyXl5lPWj#^)c%}BAb+)48UQ*
z_Y=v6wUXuTM;6D##UG-H$A|}QZdAQ2EFzI&PzC~OCt92$_O-N}5xTD_lpY-&4G+p+
zv*@nr4j$yQjt3`2?n{5a`Ps9UHa6eplSyWDc7+_ePOaqNkb76t*Ut|J5lYoxOz$m%
zHYJq7+xc76rhp0+`^4bvSsF6ZVS5luuqFuKtNvRkSu%y6{p14Q;@GjXoH0<!9b#fK
zIe*^cs4&U<-bW%kA5B8}{vt2$l3+5(g?KimZ@mjMG6sp!E=?|s02X;N7$!hrQRkkD
zWc)ont&{Hw0Ny~rW8nP_6dqs*H#9Y&y&3wcHoLGO=Q){#-{ORTYii2;cVRcl0@(_y
z7>Sa?F>omw8_{wGyyGBQ(81aGx9DFf-;e)9DGhLBd?JqozjZO+5&xqUret&+e?L;i
kRsMVWDa8JlKipzVde!J3`DyGr-iSowq_%3l^2M9~1?GINTL1t6

diff --git a/docs/dev/articles/examples/timeseries/timeseries_anomaly_detection/unnamed-chunk-4-1.png b/docs/dev/articles/examples/timeseries/timeseries_anomaly_detection/unnamed-chunk-4-1.png
index f788185b05a7a9953a0bd45f401c82f0e08167c8..ded93aff97dbad28f68050f443ac41a89ef41929 100644
GIT binary patch
literal 42503
zcmZ@<byQSexK&W3K~h>mTDnsyLAo2HrAxX*N=h1r7LZ1|L5388p-Z}lZikL{_^tQ;
ze5~af9PZq?_na^GxA(r`DoXFL&`HrBJ$i&CCo84)=n>M_M~@y?q9TEJ@^_Fc!3&y`
ztgh>$M;Ley{~nJm_}@Hw^zxCM)LRYDjJ-^c48lo7`(bOfUonPk0==|~RN7|(!l!0-
zjHthf<P`|8a;a^iP<nJ|ObKY|nN{=tKI-`LIYJ|;U~HFd;JxwRHb>*slk2tqTP$uz
zORJ?8i^cTQ+o0jK{Z_&AdG`as2D!8XYH$^g9?b_Te|osAM~^N7+BuNH>*vTvU6>DV
zupwb`KfE@gDup&#uHE0=#f11I8sC*t-t25J+&6hbuLkaua)+&MZU+zRSyL}h#*lZI
zpcCJI-V9@86Ec0p+R3T7=e3`Wyw4^$D+)^)^mbO|6b}Tae1qQ6$6*y3kapW(MRa3o
zT$h)9Z`N;yBZhEBt5@2|kshZ?tt&6F*$<;C7cHUrkzD3^Uh{L3R6<s?eszgkj78xu
z$F33KVTJUe=vn&&RMBfaKv*Z$zXld$e7hSWecjmia4TwPN}+;(tuA6S?3Wyi-l9@#
z{JPqWE_8FK+1@3zAeBH33D9W~G#fd3yD468vypd;gxcfpecM);LY`L@H*lu;ex=uy
zAJKMuCMLW&ylIvC7@V{n&&U-f7$k$K!tZ;oj!8|RdyY8YFTdZ#%AE46qYuW7Z>U(%
znYOxEo`1NQjr?cpx5luW53XCA?8aHgmvT_G(%IvgfP(wAY)AjPhO-oY863x(z5Uej
zLg>}#Y|!&yT(xs{|Lvp-Y*PjU7DL2rl=3-2s&T&LjBvm~vKamy&t;ksxS%C}wW+G4
zieJrTKQ4B)vofwG#Fws`N7QKiU51mrh7~ifU}ee;+SXN)OKsPy;#v3Sh59T?*iR*{
zwc5@rHnURiPUhr2w02gf^$?*A+5WDE8|=d&{=VniGfPRT{SM$}Uuv#cK=QwwhMfzv
zIT%)bS${pCbscKN<^J!fS1DAFN?LC_)a_arTNL_QsPf-Ns=R#ShrBQJGi^~fUy`G_
z=z?R3uI>(}m#*)!BJyl4ql6Eer{y&W?yRr27K!gKgR&#s*}PY)7iqI+aV6d#=LWR-
zj&YRcv)rHFF!=a|YG;0__F0^__v`&8n0K9|jhN>RxZ@+a|MxdSJ3E8xuDMrByIaXu
z*lf>h&1jCwIM>^*r!;2Kf5W~eGLKjM?i!B4_6uHu7GUgJ`{}kUJxtN_r<8P&!@|V#
zTgNoS?Qrk=slhhak&$=PGIHO{pGAs6xcoTlmy;q3#g80JWN6t!9=5zO<q`YaJMK^{
z`^!C#-BQA(2*UiJ!^7;k3QL`fxQd3pO$xo1rqNh&1p0SHRO(<~`yr_Rr)ef}M5@zT
zWxZHFxpf_{AEoZ;>eD7!%9QED+zoM=iW2{uV&t{W+oKcdOVr6lmjEbBp}*7h%@BP~
zVO#&vJB!U$FX8KZcZFx!#w|M~2g*a4(EYf@%_q1Z4U7bjGLzViVj~mLqOWoE>p5NA
z7{7HtMUEvE+OPKzPaPYiYEjsC?(0A0_q~^`c(eEZ)6}GGL!U>zupar&b$4qD{n@Y-
zBaf*>%cgcx(XD?R)s`DV)80t;$ZhhXDw;Xj|H2}Dx9y<orvO-iM=70-EgDDjD?!L)
z(EH<~2n*qBBOV^*ezE)2Q*7T;#g(%<GFs$NoIkZU`HAA5>)#E6aZh-yr=&Cqysl@d
zT-hnPd8nv^RSv$F%J44MI@p$(Y^&A}r#JAvTR6#vie7C^oV#BK3B!k%nD1#g9Y;cR
zawlrN)D`QK8F_cS+~0{y{F(k?z4rUp@IL7q)Jsx++cM^eP+y7)pOf%YdGR#f+hg&2
z|IQ_%Faqh@`<a7wU9)nb{q(0x8nzpe;s#UO&AZ-LWCahkF4o^^zBfbIQ);Q&Z3%Yi
zV%C5WMl369!?n8VTJI+q&^No9)fL9WDV>I<rsZ%}ygh*{(Y*PpudpQ4P^+rwS59)n
zP<lhnS}e6aH&Wo6mw5wEyXN+@!~+J{lLcv626Ck2&VLSD{TVvCvhKQ8fYdR~=bpc6
zxby3`h*s9*a+-pa>aksW&fG~;m2ZiyG&_7qd5E%aRqDkXmt8GlNBxmKgAQr)GcmX)
zbxls@DbTim)%Uea3j53NWZka{-c$to*<|0W&!xo?uTvA~-EV7KY|aO0jX?is8yZp5
z61>?NF4ZRPga74wyU69SxY*_~<g*jVUCEimZ!Q09FdceX<vQ%l_T0}gUCCmY_|zl%
z;+lWhFg-d}P2|)m(|u<E*(#=~`f{T}+|y})2LEp6IkwM#iI8=(amZ&UTQf#GHWKOm
z%>QyV7X$r$%=@6wbL^~>ABjm6e`}n6!}5Ea&pG5?|GwML8HXDPukVN}H1DQSYY#aH
zZ5*Th46-tEYUA5nJU$n{+v+wpzU#~)&qkE4E4pqHBPfImxJSiDY+;jD1Vh<S!Goih
zZBn)j_rxK>I8>u9+Y{;cBKyZo{LyXOxA?l!*|*miw<kw$4mq{?3~xryZ_A+tqJJ$O
zVwU=TqFDVB&32~QLaQZ`A!>J;9ywsNWY9YG=ewvD_Yc>0Xw-5iygR;&UWTMkg)-)#
zFV&bD4!A4j&9W1xbMBWN6#Y=C<>ISnGlgrdcKJ4!oW}oPlROuMp8pNY3NV$3v7_7g
z*SOcb@|w06n?ijk_1`d)q<-zbXroW0QnPku?GHUoIk_L_kaV_#4CScO3hA(M*B>x<
zzpL7TvkkW`{ze~e*3p!i`r#2H``O=dU+GTPnX+H?rTBP5Pkl0`E?3b~YSB_cU0RIA
zyk^b#ajZ$P_(gYH9R^7~NI8mZ@zJP_G46+EFw&I0FTaU3A!fY5ZMP%KI#JcAtPjTE
zkT)TQTf{O(F!#6~79>UI>koHCWP>s9O3x0Ohat8W$;0uVd;hoJi&#D!<@Edm<iAEI
z{n-TN;SJ))p%2GBGLR}Eg_}>6g<t)|%_%Oj>vU0P%X_0cB_VxFlxXo<W5V&0k!6fy
zpqe~dW;cM1Qd3hiPL{4}DO4h-+n;g3MOs?A6RIfCF*!LYk@%XOU3sq|B9|^>IW3L!
zuAe!A=KYM5K*nFk1l9U7){b<jrY1PHuCDGTNmG;8#DZ5*US_7bxw+dvy|Oy3AXC7Z
zW@U*!;e%>Fy&Y|WoKvTDd1<LDWLX38-_^f-{rG=j42@tHt2PzeqA{tE*IF)~P<_CC
zmCAf1!^15-D$1rS_lw%QTHf}E$r=@`z5Tf4QxIiic!h`}<|oL=*=O57?`2A<MKLCS
zIGj{r70!LNMw~v}dYeTS;B$Xbuv9Tm#iCoq<pcYn@7Qv<Q?@a19yI;milW6f_L5^v
zK6P=W3%>CZ)C~J36+;;}i#04IYdbdtJ!p&&x^%;NNAvvDh6h21CabN<OBDI4o=!P=
zqnk*-q$KS~Lb1%+6zwdrIwJn!yHa8`TA{8gf#NtFQ=G6uFk`8;UZ+b{b@(lxNlqgk
zx+>oa;T{@B&yluPvV~l{R1)hZoOXcgJx@ZFit86-P^p|EUMtOWY-UWdn_c-trf2Kd
z`EHo}G^E3O+wn+379FM5N#`x2r_C1_ijJ@x(wL@Zl7%3Na7sIlvrWRKeJ9%QOg&Sm
zB5OzEHN2v%z#46?$^1PLb>fjDKA8bc@DUC^+xJ9uzLmk(5{c4snD`-ytn^@y#<gYR
z>E6$DW4dzn=hD3=nsY;rr&HyIsST*&C^#T)>m&Gxu@;)ERF(2%b^@6R3BT{y{8W9{
z<>Tt=I_TIwB}7Y0`*=GkDM?pXH#IdC?WVrIp3lL1M>=_9XJ_Z;=BC@fl<~bOT6sxH
zKMb~BVFv=w^dX=IT)>;<<=7}e;FD-Kd?prz(#2(E3OX}7tjRiGN=r+@qPV-eBS95q
z9^Qgfmz2oR#C78tsH?A?UkkF7me_cCd71W`-iw)7_=y=SDJg+tb#xWSY2su%`==%+
zUuAp%E6tP+F~JUzKx25JtHZ+}QBzZsNhfeDXo)zApC+EnbnW;IGYols<vzSO+cHy*
zcs2U*_-f&`w}+ys>VKK#nbM6_bXaSWH+OSbw+XfNo^or*NMVr+@)w35`LIKygrcwa
zze~x`JjCE5x|Nytf`$QzYw`P?`}?K!4BG>tFp0E>KegAiJD5HE^8Y9ClNXjRv6x=J
zR)jy{Y`<mIt1YUmRQbgGC^RuK(Qc{D@Dp?QqtH~Uxdz9jMrYc9CKj!-7o4U~0-BIW
z#QksLU&+%eKq{?<DMzhMWa;p9q5h&0TOH4+%8H7Xvp@6NPTd@J;oNLYO-#Vjcx^9F
zHptCyFV<qne>4UUrt_yHC({Je#Qn*O5o6}#bcMsbw)0aO!&%tb+1c3O^h1;!up5Z5
z$HAN<bkO*qNHwn;TZH(vUa-Ef<#>VO=*Pt3?vFe|&a2%P1IZMk-UP3lWpf5}8ypt)
z8<xJ|dhh;hz@`#|i9e?DzgX!E!RQNE@;hH}gx)3x&5Pe1$7BcG35uQZrkdqvI^A7v
zuIWiNo-g{au&~(L+S10Ms*m2J^4ZU}dY$4*Feqhs?i3eV4rPSn&NjQ1w%v|x6c=Xc
zRh!Gri@$P3_d{yEK4|IerFk+;0`Kqde>nZc>A^xvp+aisy92Wr(X-q0B?ezDCnOCG
z4H3kIIx4laDkeghG&(sY#Z|s<Tv_~f!DA`z{u{ZV%l>TLM83$$A6m!O<KJbvHOZ{H
zF8{VhS65f@8I+bWslW!Dv0fe>2HgAfL=cAIP+5vns%g4SmmAht4Nvjd-`!mPFm>&6
z`g8_1XzHhauhRUVI9lZlfk0gJu&zdPVpg4_^rf57kh0whsQ+lL^p^RlZkw;*9lOL<
z!2R`7fnqwP?@@=gkyl8_NA8EcueQK=_BJ9{Lt8s9;Z<aqZjAWd#h0fTFXZN{%-*T0
z#@(eeWc%XbV2|_#6<^|pOU-lNZwPkzg7yA&e|LRxdUtyby@ZpAxqxeI@w;{pGO^qd
zNKQ-yQGt=!%8wFo3F2fJgvd~aAh<_!hS|?65|N-45Ji9>Ihd`pUF{A}pjVu2@n9cY
zNK;XN_>nZsB?R$!v0a$G7;t~PndS4XuIwX^o`%K`>)9O1Fl-8;A0<016|E18x6&0>
zgWv41&`crh{(b$R_9LS{2$V)=GYgAXV~A;9Ufvv)uOW>mCnw0yu%+@;4MN($1+V^H
zQBMQ0u-lGf0xr$5?UMf4+me#in(tV~T(2!8D;j;y58~tTx=Y<DKt3$`97?0cRl6f+
zal^pr*4r*oe$?4cmm4(2hZrc1gT2kYw**_3V&u()p({l*r4(SS4r&Vc;UQfB>)dkK
zHX!>?5=ZpY(em5i5fDJm$|b{_bcYF5Ukt3lsn~8)7>(X?KV2G4g;~7E9)(gZmFel~
z3W6j6pCuzxY6c-g7D@Da_Lt9jx?`JQz@6_+e%r43k3;TZks8OPw#8pQ33FyuDg~)!
z*0absBA@e~do20!r!xI*cC%$Anya<OZ@G-i&CRtQFCf9|%0a5ot1GFjtOS*Fn-tr0
zuRiT(MM=pLK>4bwqaZo{TK_Wd!!s!8>FH_Y>NVxe!fqA?x4$)U(4}(pkYd};I;fM8
zr9Xf2;wX`zO@@7RqFt&q>53xZe$jXPZ(R8pa-I`L*$m<&2CR=Z`I4x#_9{qMaZo!N
zwIgW`LIqba<T6jYaKsanl2$H6rm%e+b=dNRZa~7^Zi?SGt%Q>OGZWL))c&a#vX&hM
z;kFtAVfHVO%3wsjP|1S$c#H>Yn8r5ktdAl4=H_%*XM=>>a$<5)F_p^#>_(K7<3ck6
zWZ<hl&!diKgwkkXT`eAmmUPcg&*mJ#-zMipx$MQ@DWxUNX1(9`=Nj%o4clYgkZ1o@
z>y5Ap9x2f*{j>00zus;JtR8JRR;O#It})I(!razYUl6zoyV;Zc4f8G-822*lasNK{
z#!$udm|_sKX%OU>l|?nH(2zDTYHC^yXBq0~d^~~_lfmQT<4a1IUFq?sBZ~7yy#E!6
zm4qL1Px9<XvbSwb6yrzsbj65WZKW9dU+&cm$2&T`{D%1ak~Eq*LIQ>&@e$-SNTcqA
zxG20gM31f0&Q6EL&-Z5V<pn=&;Z~)v2Is1W)rVHOIXiRGj)K@<9~}MokLv#7oxS~j
zwZ(u;G%2su5IGTEwii6r=kVA0NRA|97DmUc86TOY-?L6VHfKH4s=wbTgxfAwF=X(_
zf;Ta_BZz&|MSac+v;D6|-%(ZV&aP^mgHlLr<Y@v5T2%4`Df1Et=*_`&Z7r>{9%ADU
zVV^@(exxUI#RMLmPhiyu`d}?_*Xe+0Y*>`*$+1%$2Op1!j3kL}mL+JYuI3_*CdK5&
z@T>AuBiLB(2%<9s8SD)b#hqdcdWWdU6!us}&nvN!mVTP`zBj<4L9bWERdzYeTKWo9
z7f|=l_?LXgRlb3?o;P}XeZJ(>g~ckdr7UvXZCml{VTso6jofD}%#+VEJr+$Cd|^3u
zG;wZrUUT*K0#97l|2!9|ckjDm2PM8Sb9%KWa=OCUak6PV-qy3Jz&a!(<a{YW9DBSf
zxrh<nC|t3PiG_t$F|PSz>%nY4hx(zx9u~Lhe5Zx(Z*JFhXjw}VxFfX$W1syxu+w6c
zHLL^OUaySfUr(_9pc3`gP)!8w^LqkqkduG{m-}EEuL2qCWxI(t5_PS|;iAW)*CuF-
zynnnmGd=#8KaRAB5}<#JG7e&}qjbsE#>Sy}n_gyM7uDMWhf1u~evVIZ{D_Emit@B0
z=oxs&Kvk8~>^J5-Q<HhxOiSoxI73!ZdHEEM?DK&YCxi62%#T}!LRxi`H^8+K3~SWH
zMz-M8nEYN4a*LUvf3@|)tif$pi-^X3?{AgQxx=kWP;m8(W??iLKWT;_wojdNG;f*q
zo94;Gy_d6n)lpsB>re$OZazHkw+g(k{J!L@XlQDJf=M3GWjaV--!Gu;{P~^u4ULKK
zPGWL05i#+u{R${oK^_&pCw&ZGwH;`KvwRK&MK+#L)^IqPK>v-&dv0(DSwrd)D)DWO
zAaZpdcofEY_G^a4y0ZJ?StLnc^zlwO25r9obvEY%ODnkDRt{+1s~*@wmqs1T77zFx
zMFQ#-N*An_jd}s^(<hseq`JU~YfWc$X~SH>qr56{jbPcEt-P3{fXf$_pLXWm4RkPV
zjz`cN`ayg6Sg3SdO(p{cXd#{7af&4}g*`S6Hetrb%sd9A;y2-dnZA@}8#n<yQG%50
z7^3^`_o{za`8!%sd4t1@xN38+6q9A>$59%QbS1-b&BEeCxDL<B{q13ZalkFFglzJL
z4r{3G^Mh`Xek6XgTc&(n<d`+?F5gH}xyS=PI6Gj5CQ_7ZUs?MUU?oR2obAu`N<`=u
zO|D`}c5dZTu^rCTgpu<)ekD?Qmc3cx2nppiW}V#w1@70UZMEQ$v3z-icbtM#*W19y
zqiDE~4xr|`VxC7Iw6utXCe19VkJ$q*2Lh3w#Uk|Hxo2X;sp;$MAAZQlHRw`U7d@X>
z*U`D?#9-$Tb?vfjA)pNh*r<JRZf~XrAn2}@qXcEKa1UHgp;qs{_*Wpy-W9y1Gji%}
zcCyeFbKldM7x%lrJ-<FbEMyzyDC$I@S5Qrfzlj3}Qi=KgJ<@ZDAP<8FU65KzF0M}E
zkxR@xoZi;pA3KA~@pBhYC0GYQ&;K~~1C7hxfZptjNmuA#hM;Tv=f|I<`BB-uVHd`&
zQPShJPcew>=IWy}dmkd*aZ#kTe>4LRGcqveZ5)6CYjOnGlr=LPRG?{tzA5Z~BakN>
zmqb%0JPwu~>$J+OS0ZQRWMjar3X;pX=O6~*vnMqoH$Q`j-`}|2_DxQ7^pxP(?d<o0
zb^5C)Ln>lf(0U0^2gNQYN!pwufy1}bE0pMmT-r~9W5^-+3j4YG(WGn8PuTVAF?)8=
zcon<W^mZuL0Ez<e?#rB#n_FF*Fb$=vt}e;A-=)$~hRN^5l#~QE_*#rO8!IcFzQNYi
zby`{)z;g*qoGf{NI?O;TZ@5#SDsWjl(3l$5=0vwhPj{xnBjL3nh*asAZ*n~rMQ?O+
z)~e90fnOZ2m6VVwdF7!N=cC-e2)MhuJ|yYu*^0SnfBO<?yiZWA+Ow{bl8EvL>wq&q
zKB;*X_ZZ)Rgsu<lYS`09=*;BkxjzB5zJwE!Ow<c!Pz$>{kSDOXvDMd79A~O~eBXc1
zWz12Ly-=tLlb8{|{f9pj&;)?qDkrz$GH$d?*9ymWXKjM|&;0y+fr#?^EH=0^N|?$S
zxP_Ka+g}Vz0YXgw#UjH?Y0U6J%N{fg(Ag{ige#IasIgQuGRpGjGYn9ob8~*8R^S}{
zq=_#`HJYZU!F_+$D%+1wSeWeoybVC`?Sk|od8q>G6hXZq)7V1mw96+&hu{wxu0tLA
zWG|?@&A!<}j#J>3e9~VrzAZfz#@l(myo)DZukXH}N69e?6QH9l@oZ3c-Arm5$6;|d
za`z;Cz8UB=<BPcl^Cg%7X6o(dKo^Gi_I*0eS*f}ENVbEx0}0VB&rBfA#Ktx>Jk0;c
zG{>5isEc?4GI_Zig!2MTn!&!ZgVMHua2xbqPgfW!bwP?Swc(PXmR6`TZ^!*>R#tP)
zj#h8P4ru8l+wWt&-|(_IUk8E?2;gPGnw}pLCe`2P9{EXa*a_o?Tm8IYBm9nwQI2DD
zdNsC97eW4!e8~s^kw9TZJBNr8^wX#7#5!1rY{QQ_p8A~anWUj8jz>g95V0F6d|P8!
z8YirE-5dh_T0=`~$1GhZLJUdHu$*<^T@2-<5~XcffcSx$?t~}T9&c%oqnH(cNX6mX
z3s0ET7Jz*J3e%jnI(QXnOUq$_ihiaaoiC0|_HyEFYHTK%T}!sm>rN&G%6`%wvNJQ!
zLIVQ4yqYe3?XoCq?tI)`<U*CRL`aA|0G(js#rGQXAq^ep8{_uS2)>4iShk`l3pZsv
zP8ESW3A8$`bS~ZBofCxUDL#+Gu=q(SA3ts6gJm3)CytvgX%8xX7?po}7~zX~Y?1u#
zPL71_H!M#{5QVeFk&9J>%gdbL#Lpl4rBQ(<Bbzy<`J!5<sbf`wJvS?+tAVqK@UASK
z%gx1czI;mBA#F~%<9v^?WlU<FiA?jX^xiAw7~R7fG=5mtDdgD7g|);|=yj$vqyD`y
z`EjHa4V*WYfcjbyp|ek)i<*k9;>Iw_o%EG_osu?2x8j2U`C_FBvcwi>N!%HQGQVVg
zfo}2V2LV+~sxO~aF=LkkMQO%*M%Qd#CT&@QYDF)MCJ~@N*|UW>m4qhrsxt^;O<vPO
zukuKx9xRI=BU#vjlW4m|`$v+WO@u2<i?VcIa0)8DS>;wk7czVk=(;&}pNWPBk0yP^
zr4h77xVW!5xS}k&4vUr%2lP-()Ie5wBk8jTg{o04B2OUu_+8Hw$F<50nEE-=#mAGb
zcS<Yb)lJLb*tkKwP3AZum$wR{`Jt=SF(i&(JF~-0=zk+uP)gM9X0Nor97`IgR=|el
zvk?gL!6rB_`hCZaHd2gdp&1M0+}Uwqk8h<g12fnMp0UKFs<wwKe_u2r%07y@+XzKt
zmyy_wo~uE%n8P{2%(Yh4(D(!Z42^ya3GKYuS**#F4-76CfV6OR4j@~vU(*UB3el+y
z+`|RKqBqlQsxYv8-n)?6E_A)`120J*0r-NLt;TH5s39jl-%4dd!YAG^lu>7G>SInA
z*vRrLcr$@Edvl>#s01}@K0xD#i@&xe-tRh<wLfztv5a4!bNb0?RANRL&P6UR|BJGy
z#AeZ{&->pkNPfU6u0ro`**68@-nY!|t}qc%L7M44o)^2psWk+c+z9j7{JYS@ZS<09
z4m18x!)P_l$(;}6GQIK-GlswiG|C#!e`51Hlcjy<BE+MyPb$wKnwmY&y2;vpu=G4K
z(r5z6iK`dJkodZrgZ3>Yw`qOFL{<m}@vL%Z)z*M)V6(|F@h(r9VM~2&twuJcSg7Cu
z2Es+xAiOj{pOuJ=%UpI|onxESk|wc?D@f4tc&pG*yG$og93pC25SmIJ;&=qn4Xz$E
zuI5YdzDJ=VXtzSIH|l~p@Vw+m9Guam;&;fnyF#z|L=>v~wb#JW^u=*EzA~#(tIj=E
zwu<{aaoTGd;V!#66&?q!BGjML96`OXn8vDiQ%W&!=ZTGdJ6<R5ugeofV_BU{8wXIJ
zhx}_P(v{_D9cLyZdD<-%QoY7@xB7D0Mi?tJ14-t&a!57S9p^!s067O9@u}*{-b8~J
z_ky5hyxTg&;r7~)w|6gRU+u41xyO&;+QkG^zA%;gOL>RWZzTtdZVYJh@!q1Aw|e3c
z5)|Kdiemhw1b5G>+8P_gM&h?ZRO)j)#&}+nM!PiD(jH&{_zUTG`}mXutGJIRJE<2I
z&~D$ZPw8rq5_d|@P1GxtxIe<dZDW^#9D^XWl=+%w_WS4@lKH~YJ%N@FEst>&9zu?!
zboDgd!omV>K{FAe_iIEXna~t6^ZIG;XA`0raLQRcvNmdG*@8HHheezD^Xa*<zq}{w
zRJO4y)H!xHZ_7;hScVu`1I6uTs`D4EiZFQFb1s+VzjES;9DSa#mu@Y|NN`2A&Jnli
zd$jwllR5GAJckcsqP<1ZV9k(1LZGcLD4I)$R*=UHXaDB~V9@y5%W6~ntx&`=&=BQb
zb4KUAj7+Ft?f`Zvxn<^Q=(#?HbHCNgTxby_#Nc)gcDnTh)Z8{C41So=Ib?Td2lITr
zk}}WlY{m*TXKYN(g@!_63!uV$&mX3&M8duom6twZk<e9iQ@YV610sn9JBaS5r5o+u
z4@_6oWh4%i+`G6TZ_Dq91k!tp272rkTT_b-!VZ{hk&BIAm_X~}xGA|Fur?Tuy<NZJ
z3V9yA%qBBJyZT6I0vIXVsj{3`rVGt(@VR5$r5<CgVOXYcO0$7#*IXkfgeMd#AIK6<
zCr57n++dYcpO<E*&l8YxgT^NYN%qLLF{!M><>zuko{Wh<OG+kRYQ!AT$WfN&i1NqG
zptydG$~73GJOx-xXYYB5Hk(zr6kAgi$#K6;S=q4F=0j7I&Kr%2!fckv3>em9+cg)+
zpqEW|qrTao|Ew(nciWx%bvh)hltrLPD1h2StS{)r#VOSj@Rf*liVC!+MUgDm>=%K&
z@5RMCxfAUjAszsrxNN+~4mXL$)qgcm?cds?ST4Dx#hu<~hx`E|Xd)&s8&v6j7(5OC
zso&s|kR*}kNYu!jsOa&AzQmk@l)700bZ>=kZ)<y<RE^?MNkRm&*$5Ubf0;rSmZH@X
zcTciHnhiiCsW#G~bXlQH$T5;p*E+As>_}YFN|m4Div&mc>kM-;+|$vvnnE%Ye&_6&
zXRY7YKUtsUU*l2NVl6*`7<;ZfjgT-js@~?jUB?|JwN#3P0;YQYtw$%Bl7?n<kLgg&
zuor>0DnThlut0}aK;2b%7Xv`N(eo%wk?gTf5<H~&MP;r{eEsy5Od&Vjm8XNBqlM5I
zcSSPK`fbvtzH9alA!kV>CRxlF&7k!4Y>H2$i}O|V{497YktrN^)(<Fnngsh2{D=RG
z+vSz94#d-q_OG|LinvrqC{o79yDj==q?9T4el;F@X@P)7r!T*GiHn+H8YxHu|G@FF
zo~7{sC8*<x(F^RbTYu>3(yU|11g%{Bq%EOz)unCx-y-~QsTWMPr?XEIfAX(yNjdP4
z+Q(K8<>*%rwIDb53#G6~>1GdGA>cZ!mZ#ceO-eg8leWhKGGD9ruFy-TSH+T5)9X};
z$Sn0z%kC2jtP#v}-%nibeneLn1*R&f5WI5uGKjWj4q1bv4(V=cNyX5~iDeI8I%X1w
zE5_r=It4LDPo|TVl)2^r4MSr}kRZt(_pHJ#pwA}Fl8SIdQg!C!#3f95?6q!2pOT0r
zSq|f`l4F2*Luz%?+a%)Y(xG~{-8;%xZL<aD;nl45O>-=f@f`Zt_?WJ+N4&v~Derx9
zfav8s{K+hOP<3*zsHn&j&@n2=?<V_bZTv?~PIG)#;xXpR79G6Yzs<0O&ZzI^h~$^O
zX}USqzy>q|L3aOUd;&DyD}Hru%<MQNng|87%#q-1^I&bv8rQM!ho^h)Oj)RX=I7c<
zME%|!W?GtcWo59wgl!Xu{RlC?C(Qa=x=$?(rf?^ryUYvDYl-c81_@D7Q&ZE?*;?=W
z#{t8ZoLT1d!7iN|g31cH?MR)69GpY8Y3NY4T=+$?juLzK>iBGtJz;+aLQ`FB8v>gp
zms_<)f2nn*-;pb~rgj-|z=q^-x}xg-{Q2{7p}B5Fd6|>;@P(zTjDDT_o#xwR+!a6<
z{kY(xjus(}CL$y}PTAFi+Wd*7#?Jo1k|HEkl(4VHx_%;+I4HaKJ!+jgJ(HBz#{2>w
ze5G*c3?^0A-jhH|SJJ397G~x$HcdtGj6~ud3|un06pCfe-JEn0qUlPZXNPBe0ditK
z+)}~q9QwD~AKu6JZjf<qOL*uPx*F_ilMJ1$(Wg+9^TMo~5^@1fCRlQOIz>}%!e^KA
zK1f07QN;_mx8w}qK<d5Y%MG{%URkT$2iv7|_A>cAVYp0AO6sA0Mean|Z%LQ3Tw<|O
zCd}w+m->`1bS8GrdV~`cBFq{R$&i=t%8uFbjvxsgi)Ie<jhvAqY%W#CJ)Inp`=h}=
zFmk+Qk)^GAuG2YBIG@;n`n@8c#jpNtcgbcHfootVwKO1HC55vYHjgB2{9s1SFe`|R
zowYmU<hLH96BbU_H`Q6A3+o}ZB$5d6f#t%R^lCqj)>^A~rIh6vh-mku#r!w`Y~OH}
zC=)stM*j7G^#wtanVlU2FmpPe{aCvhQ3!^pW%ls9X1AEKg~7hQ=QJr4ouZap`~41S
zFCDP0%*`Q=oTi4+m+O3-Fk-_`ad$qaFEJ%%j8<UW!%_uTVKGbbbWB*7HAYXf7DK_g
zewipCJN4B&E-k`SA0wl1#1D{cK8vb!r*UlAuOC+u#;PLu_ZPQeC*lM`qN<|uA$F*|
z<BN$q#Z}bZc=nyS(WlSp8L`_&1vz%5#h-{8d(G%|`^z&iYBh-soB{{A8^4B}vl4W?
zf~2JMzRPGL1+?MS$pHI1hSOx}=m>oyIFpW)&t_Fol0&x{oQVRnrnEy#u-rsRJZ9#|
zMTWn@Ohc2mtaj&b6|J!I3u#rDF(7*J0zP~)EM;`+o7H~dU>U><(ki`yl)qWnWnDHN
zPbPuZauu6p+o(fMhWdwExZw@2-CvFqMexmA5y)M8?l=C>oV%4*626$OFL6h--xbW|
z?}=LOm~eK{&mOP!s{2BdR8CXOZ{m_EYy77Ba?J?mlvnV~tHg!b;H1)1j4J_68yv$b
z?r+~dUgYZc`Zn`@4RDrnV)s|mP<U*7<7X3iu*vM?GH15ykl^3R(@S6SWYlPKLAAps
zF1Kw=b}Nd0zJFQdTAsgoVN!#&x;kOh&q!uPXs669IorBJ7)Lk~4my`PYv>&q)7j_}
zfI9W>8I><?I3MiFaY*=3gh*mnT6TWojh>tS5T_^1D$|X~`i5L7o=fB1RL-85-&L$6
zvJRPL*R|j$6K5>$gKp=5e#R%vBU)aAW^g;mno`RukJBk6o!2&$6_cAbCbc`+GfUHM
zN&IUw&h+7VkO|S}88dn;c3jl%%o+tq<5Rn@71c&b3k8Z~)hQ|Iy-|eb%-Hx1txjBk
z%3C|Vyh7}TK^4;X)owN+As_zsmhr<vS;7TaHNZeY9*XLjPxdTk#ttILeph`Ak){a$
zvn890vO?HpSE&hhZUSW|bmmd<)z!`)6HUJ79kt4%WR?zZ#pi2YJXtZH_dOjH%dQBx
zzf-{fAcpjXBCTL0{RA^z0f<GiA70uM1w3~$WM*XCns4I13vFi4XE-3QS)@zIp_IlH
zgbm8<Y+``DC30CKwVaaJwE$w&-q<p3#S2ShK1%L!l`{&D`5fYRc#_%?CIq<^yWYf8
zzW6rrTaq(-U;Hg|EZ<5Ah4mN%Mt&9k!zL{`J$)`$8l%hE(y;lH;BHyXk(FG05e{_z
z0|cS{#O0t(E~rOd@}VHiTGL|-DaVjt^lY+ozQPN?0a}am{g1;6h&QViCY4x5mr=a;
zajwa4%9=$fx%f{&E~N?n((y7<g|CTvErO-|O~IK=LmZHF!YEIZw2kCMn<JN&!1xb{
z<JeT6f})byV)a6~1o~|#olRQGlKQ&{4_t9t4b2|Wl(w@gC5gNA)YLw1hv7e-KsVf7
zs&M3$ngM)`YBpOHnuS@H?|$91k?(QOr;y(WNKnUjF~{q(eR*xhh)quM%k_javW<iv
z<>d}bZGOV;dv6R(5#AX*5k3b^&)|?tlRqCpTjzj5uO=&SCH$e;xxHnK@8#ss#;_#T
zLaTx5%JpeWMNiWN2H}h(W~aaB#YFly&M=Yw<QM`d^2;fGM_%i(;GiBgf_%+Vtq0of
zZlDbGJs1o)2ySkCWvoYrKXWZmIN(O~KE9gux!Nw=#o~_4a@ZWo)C$;{YhV|9A0u==
z?}~HxJ$s2)zA)<&5nA0>8Do0>3(yTgQo$oYp+FeN%Zj}Q{Omaxk5`9(pg%==ENM$B
z=(66lnI$j31+>W%Fz2ofpSi-wMc|D|{aB?u0M8*lgh?I>$Irf*Hv1-%x3RGyt33u}
zCm==D<p>cG5U6AnmK7IMm!kj*JX~NZG?g_>1{IuJY%U+Bmd0nVr>4eb+WnQi?cz(<
za(h1?DDhv)U)n*~8<)S#IkaG9iZp}a^8lqw;g|X5aJD1jVH@KZg>tp4L0@~#!U9-y
z2?<wT%jbK%MW0Kn7R~qssx>@sj(ZXJ(PaGvMhQSn(cf#zO$Y@f-`Q*ZdY+H`-v2&E
zh<#!u0@TO_cX6q?U9W4r>B#rDg_-Vpphx{PZZ;3fXTj@uEct1A^OOXW`|gd$x75F$
z27gINNHzgM_d_lL#yww4M-$AWIF56Pm19?MT^QbLP+?NQGBGcF?JMScLEOT#C2CpP
z_tkPRZ4|54v}^vSKEDWq%fRb~?+L%s#S#Z6wBpn!9U50qZUUvPgn+345cbQ%mN8x@
z>wTV*&m$8SFezPozfttfKXwqi`1~?r?x^kNWFR+}=9VA6C2H#QcjrM^KOq^|8Ox8<
zWzRs-b3srLjDEAzox)+!J*PDL9o08HJYA9J3kE6nrggjm35r$1t}=!fm?%hrjg&P@
zhlN|5WII%pX(qZC3ck0-);ZQ`CIp?Y6MtHh9tb<M9?Jx)aJ*QzgkTx(XeQ8Ym(>iB
z@_fY2d-6hJ#(PvTji>4EYIpnK{ACTbtW}jxST%w<wp6t@n(SS-DshQpsQEFmb<O4F
zrGVb&QvM!t|Fvj-bPSB=08g$+zq#7$d77D?9$AzYJz1))t)a06rCzhHF)2VjHW)2t
zd1c=@pwUiY_7eBy8gqIqSIfSd=VxA5xWb7d$h1Br$tuqYj{D=7fC=N?pXU%JQh46G
zn4P5nkgT~0r<$km!Ms;JpG!)vsG&Le7Ut%ayB%OQxpwsxu>_rwr4=)B5y)k8Egp^$
zuWz<S^T_xe_C|7$xk%tZd%s=533z#LvBF8KUPVe2z=?qLY2}0Txb|(+`l<IkATt-=
z`mgS1+>=XRZ)ikDoSf<mVt4w%wei=>e!6@C7zDlwm_nXmz8C@9-dvCVZW|P=aGB+q
z$FAgG7sNPkGl+LRxemS$3GJ2vQV(zY0vR;Pf#laM?)yRy9XOiIGKDjX0|tsh+irQw
zp8#$yE_&rG-yTBkbvS(W_<0_w7tklxa2oH0KujJ5<f~5~tDQqq*o~F$Y_|$9(9!>q
z8IfcrDX#j@i@<@)U}t!P?}eq5Fl#*!`a3%%@Q_By;H|B#!Uv5(s?V&>Ag(Shm4S4$
zj|wPVjf{*cjD8($f~5dSjYTein+Y@W9Y-dFjdP+<^x2?9Tv3Yq1B{E817PIAL~+8R
z+0dUq9OOBYPxmlu?Pgpiic~Ms!CK{bQl0|egOWUB1b>Is2In|o1P1>L%=pN1#B(H?
zZ`&7m)Suv~R{lUfyK<Cee_UGa<+L@DL;LPq?ER|4#NjzH0m1Tjmb{WT<><+D*0Z(N
z(A#~-xZ)Gl6yPJ%)FixQcd*M8a_dw>!J&*ocK$keXWfc}70EF_8Mr-JO1vO%R%rs5
zZiY2Vb1oQ|>VB<Kp0jo<ZE1_k#ZOciLz`STX_7;mz?y)w*i00S0t7E&Gr@Se$}TgP
z|2vv2Z0!jub@s3axGH+Zv|8I~?8ERvojjQS9j5bht|#W}8?y0<2?;fTp}%PYGJfEj
z&vsNPN()uVbbrmJl#5oF7-~!`H^;U`EkzE;i&Sa5&<XWGr{+h^0qPCS0xo_P+Syj|
zc>yl=W8^STKn|;NplEx~4>^C7J>=vT3Nd!PmgD#CrSMo^UmwYQ&&g`o@#X05*?dZs
zFb%p73<x$o+*_b{bVt4x21AX6(`&uj?~MqgOiYWlVUTL&<G8W0baDZ<iYHs_39)qB
z@N`G9QyBO0)-%a(8b25bwlQ5fUd#h*9YZO4^aah&DHjO0IHLb>-I^|t`DB4X<9GHk
z>K<qX00t4|inh^ZhhdU%Im`=4($a+LLcfI5grYnoQ&d#c-k#mft?~Vpc56pGx_Jka
z|0P&LX9b{o<VLb}2Jo_yDpIn<G#Nwv{h<MXoB-XTka6fGful4)GOM*wq`W1dvMY#M
zjvenf2g&{R9OBqqUtdP0<AYsn;mA<dtW+8!p;<K8&BR}Q!9M+F&$S0Mws1N*3ZYLg
zK>uwIL}p@cabEiktSjN<9ACVa1D_q?^**R?X}pmT&gMKS-Q+42xp=*LJ3+szPXxne
z1wIjx9UQd5AZZIK<fB3tOSLQL5C2{IuJy(QY$R#>BT9{bA*S@DHsFKl35c;;j(;Bw
zTim4!ZNhNoKdf_Z!r^BJ3v<ittH%~EaUmjJT(5JySB2EN&BP8{PZ<wFx_~_Y+))P*
ztkV`LhV*iHMpc2h!1HG}Swfh{Sh~o;?=bHP2w%D(*vIs?w&)D!-z3upu2{$9zGA@5
zwsy2C9Pu$t29A^kRA%lW0puQ*kNnGmU9G^+CPLa1+5}eY>C>m(pW=FmHAcuRf9+2T
zZ%yR!^R1)-+2I0^yxD*ven;gu0roh;<zcEg?xiw|5p;-g2ViP7P#jUBjLAJlp>LlB
zrmzkUv;3phmjzy@a8Gd@LvYiD+6=U`dBD8Z@u3ogn?m6!g7@+_nWHaaQadhWJGZg;
z_mPXAJlYc^TA8RfwK#)(7}CuqpR&dLz|hu_d&uVr(-{Y(my!2&{#E%}dCS4oaKN3U
zYI&QEUm>OE=jdEIIbf9QDx|4q)Rlnx--7Q9!yf_mE)Y=+S)RGzO9?h;7Dj_nbCzxo
zJUjzc-QCarR+*7J-N*lp4?>XTbCj2}G&xX7fhS3-t(#LwFP1qeE-r`N#ZNW@5%@(%
zkt>AT{15K!1h#ZCyRrXr`{N+95-I|Z-Kp37{3%yk`iGo2q&$E2N~-nqqXrJDdMRxe
zTfIOz{x=tPb+#|;b7r@*^4OQoYF>PWh7xqag_%#mBf^KR_~;==R^<I`jEuknSiZ=p
z7K8hur0Pfdqh_GzRrW}rT3A|+4VwItx-r?AD1HzDB|ai{ChccyO{^f{hCHmX4pU_$
zL%<R0+ngsF?+8lYgb(bAT)a#SCFqX>61KOg&!n6|1zK>M9?TL=Zx#rSEwH{C#8|%u
z)|#(nNS4(6k3~!%2&>T8GKO9Y<usmc;Q2Fs?$ez)00O{skas#3O;)lrukCvmeWJaC
z8(7O=(<+Ep$8<0n{t-X_!l!W{AuoBZVKqyHoBXc%c_Hu0n4a71BoL(>*^Sr)nhT|f
z=h5{cls(g1#}atswvE(Xm)iq@&+PXi2fIOI?A0xLz};zfWMpLA!5@!LE{EA~h4NJR
z%G-bBgE=AyhNzwslREL5Z*-P&u6BxNvy2gBU<fOFx;<OBJ7bmoS`JUxVp#Ot0J<dZ
zdvWxP((}#zukK}XPZz&-AaO+FHlOsf$dRuCmC6=SvoCQ!F1899=2Yp}fsWA4k)Ct~
zJP_VzdvfrJkcxizr}IXgRhc5NZ`UBu@Tr4w$VPeg!93V)>%P@*4_s6rt;~VJMGkle
zH@?3O@)a}Yc+q+Q?Ed_YN**C0GaPy9+B&g^RZ|4gED@&l=s@>>0Qh4R2a?35ki+KP
zazL21-go$xkx4_D3D7|)!$?)q92giN<F<qWG0@{?DGR8Uz{?c&xY!B6zOAqu6co%C
zl^;5gh$Q_p9DygeTO$4V(ihZMzIm^tRu!ar+=fN(T_9Nm5<4zB_fmifz{JDNVLk&k
zFkr@M@dgf&6Ta=Dy89f4SOCCfv9*a#4gg2%(cB&|i(8{jZXQV!8xo+0=ruSfD#diP
zf<-g;AUCY8sX6Xe2N@QIMHae=J7buos-*=iqqTrb9`5Hb2E&c#Q!$sYr9H#0jD!S9
zcdXUJ2bV{{^?pNb^PyropC9-Qeh4hCK&;;^DJfCSlgR}9yukb4G$?<8tBlOCwMsgY
z2n<HM;E1)~(Fl`M;9UbInW0qfun3h5Yq+3g@CZoR<G_`um!X0{2+v0I?t7K;U>{u%
zd>~Ww26oP`pJ9^&czWE{2tdy_)(<8fphw2KX?k6(gn=q4>tP;V)hsmK4obtn!lx?F
zBjV_azkmM@Mk5dv={(kBpC2RRN`S?I0+%J`GYWOO^6C@=mvH}mQA8#I|Dxv_@PFJg
zyhZf`gA!_vzC)7*=<6Eh&uwgN`!)d`P-`<;QeHlEgc+G0F6TzGR4HsP$Ie3(`P1lE
zGw{9r;eBR0w`|x2O9Q$@7vF19Fz#LfW81$a6=VI;Hoxob@xtk;snv@)(1;ND8sJKB
z!n!o*@Q#LFNn!&G5xD1@KLo51O9A(Ox7ctre0tFG-m_O}1H9#p;F@1dz@!U94*X1A
z-*=T}WIk88JUIS!upEFD5X|zo*p4iKafMzEIy!0X(&7RQUQj|dK<cq{nTVx}=}SaM
zEu9?jO;L*ZHhL}eb%*1JtLVLIG3i1%NtOJ?+djjy<5A9MGx4Xu${h`$OYe&?CINXm
z%}Ow2T^(Y*WhRPEfbQk|ZEC&RDczl}IGnDax<oNX$Zl+CZKc3K(+@6Y<wjbw16W&2
zQxlpPPa%Oa?(K>GDlZmHrNHI7<)`bu<qEmNoABO<*dg>2d$3SymFaY-dm}Y2<Q0C2
zm0`2Fd=%#c@Qf1?1%KioL0MVRs0RqnnT2SRS1GPBT84(vv@A;NYUD100XVcdH&`|8
z)F}zaJPX)^CT_WGU?{6|T=KuYLJ%zVA#uQ9AubM%aF)r&=N`wz>cM}RB@>glGu}}H
zG(Dbu)Q`0?U>f2H%Pne{Q5(hz%&`l`6%_8N^Bmx~-**hCfj)ZoCpMdH=-HDA{?CB7
zlMd+-5RMxPQIndu7z@g=3l!YTRH;wcFOG8m&5x@5!t#qh37>5@DSRkXxJKW_q6e_#
zmlZV5s6F4_B>u!>OoSTz!k0UDa$^9-(ci+9r(m`qp=c%{rVdSA4*jX3B-=_BeRBk#
zIIr(RrOrQ*T^_T_ZmT_6OLeUWj8u)k>u{YFyjeDYTC|uEiq3xmGW(PVCU9KM3L1Aa
zAa^o0gjlt9s&KRXQQl4{;rKqG=a;3AA>j@^@bXmx(JX0VV)83y)V?Oy?spK5*q^0$
z1cd#T-jc-9c>J(bzkusc#=)`fcH6MaOIl?X<Krb!KZg5#Q@7*VRAZcUZwr?5kL*(%
zt}=U{YvoheVnj|*#W(_yJGDJe5+k4I@6}q5UknTm4`=h*YURm9U$AP>FDHoYXPY>m
zz4=vH`CY;$vB<_3;3l=D=R;r2eTZJy9<TkhiXxfNSuiHpoy`lU<K3Mx6!wf&@7=>(
z!AB8!URoAwtv%GZWTupU09@B+SEX+o<G_;>fNgj(J@iM#AW6|^tWY^Skpm_-gj({h
z?G6Pi`|o>+mD$+D%-7Vx)irf#dY2R0DJd!H=eed;I>4ipz;GPSCnzY`@c}p6+DdE<
zMO{@jhcV)|$pc%tr?jk$j_Qwc86J4%2Z*qCi{O=lIZ`>Nq_mXQu|^fdHgi9C7$6o0
z!Auq|wU^X`BM=v<i4oR?VAeM>Dz_#jw0N+oiw8?kpWv513F}&!oz?TPplA-dkh1W*
ztyQZBe%WN;y032|)qf>Gj=g(I$A+EMQzEFEY}$usas0Ts@%v9a18wci)62aj1-ARj
zXO37NMXV*tnW63~;8Bou1}!x3Xi$PhYDBewI6~SopY`qzBs5=GaKi78%D{bfr($Th
zFE!&WEK|xeEpL!>l{d=SM`D@pc^RV^W{?(yi@v`8G2aA>T0=X`WtMc2?0j8BHyJ#i
zv{yR^N;ptT!U~($uH8eiDP|iSje}Jf399wd7k<d$1Io%RxS4eT7;dqZAJc`=@&SS$
zz`@31a4+Do7JmNe_YQ7uZa~$Qw5WO~y=a+d2WtL)($Dg8dSa$oN$mFi;o*n=l*?Gn
z3O>WlJuWi?JYjEME+0kEvC=Rw6dG9oje}0E8n1-$J+RnFT7+Z$eq~!>2h4qPM)TU2
zxBg-86wIE9zQCL+wG$w(*RHRw9$OKWUaw6CV;GO)qJie*M;;qZp_!Q(@SKd63wRj9
zPb^!~!ntAj{CZcDDCXj4a7l@ts_Kf0?wIBEzir^St+5=WiTfH|&G!me1zF_cZz*w+
zjJV|394l3cgyXa%coo4@6|SDhPui5xGRuCa6@+w|l#ccF^;I?L(-5G`H4l{RQxjWH
zm$KHIJj@gNPjqGg!6WG07M}1qNg5SABc}UGKv-B`m4M75oCqr)v=S2zdZw<jJ>!-<
zX{c=wD)su|U%|EAg-Jbqx8b%jHd-vswnn))rFh7vO(^d$r~YlMfoE*H@z0m%!ZQp4
zsn8wg^0ML0ah~Xi>UGX(X89^}#5<SxA1*3Hc~I8KplWYjktANKkwNttP#&Twgs1xK
z7zr%~t#Ih_luRiS6PDEC(%<x{*@<2C7>jW0-b*q-6@PQsZV0<uxc|aruT4pdc`=8W
zOug5M`La;aQrhIB<xl-~ygau+)ub(g7c0#yS<RQKhOsF_+lOSiwQM-ZUTdPnPBnF2
zC7u(e)=^P-cmGOEO{u~5;NJFx$uzX6zc^?9|GWSPb7&JSj;cVnV~}8iSN_$U$*q@n
zW}5+Ku{#MgZ0^*1cc$%iZ^j70wOPIar1K?~5V>d#`tWOF=E7zb)%y1rOcEj+rXm~K
z;higE9(t?YlpL$Y*vy(GoG;+&A6B%z&^Pn$CVJFENxliLeJ+%k!HRI9o}J5YH*=?O
zz_w6N**%d_^7yyL6f>iK4ar<)ve0dB8Z<Q!>t&t)c@k_w)N(OUhiy9h?;|BA?10Lp
z<9$OS{*DS;Pi<4cw;3$a`Y=rHs4wF>Y|4M4z7v02CtyoqU?M(JKv7myYg|&~ioHYN
z1opz70}snuoaPKt$Vm!|`RwG8l7r`w@AVVeXpI6I(+2ssF~9W~@pu+?j~)r^v|o%x
z9i_@Lqoy7Z^7gWpT=Ecr#zNg<C>m09^!YF1OagRs8G5UB4u&I4w;Wa{d)8oS&(GwP
zdoCj85NR|w`(MgrO%7*};<ck9cfH1dVky!x6N1RXTmx*Z8mDtiOjkWL>D(^ljJy`g
z3^Pi82l9ZNKRs2rM4l$Ta$A3^@)CO0yvwl|+5ofzlOsZGFVrNJ0h1eL>ak0^dY&Wo
zsku_<jrE#TGkBJHc9{-E(yt{kmiz$VH`5f7lU{tpusS98Z}l9zA<vz4_Nb*p53YvH
z9eExR-rMy%`(uLVrFKZ4O_W%2I*%oMPM5{tz^Pvc$X%#`aw-;siZM=)c-T{!CN)Of
zcrG+WlqwJ0jySo!qh<0MT(zQid0&WubwuKF;AsrW5f76F0$*lN%#vDguUK7?yk8N~
zDEfPlk{faK20fZxoj#`zKn#@i)guB7k?Ju;S=Qj->R%kadv`BhmVJ^J%=v>j??R_?
zX=?bT`lGQC0v$n_VUFO>Zn=;0j1bUHHpS-u!=cWuOU)O5k1HkH$Hzz%zhggjdo9Rh
zDK9zq?dVT717=OE0g(?~A!AN1{YeO_n`t>?KjB|p&u%k%L0A(h#=j(r2qG?8NzTND
zmt=>ba_Hyv{5y?a$HZ>)B7!M963%^#{@44!VF72?{`0R;vMYlUf9GbSv&2rU(aYER
zr3GFZW~E^SS$U>e0V=iyc)%vBPzCAV$30wO$k5OboJlD$kiauM<mgeItH|i$JWeBN
zIJNn|AS_pOVCFd<`ppp)+eyLk;ZZ<N)()fEN8ePis+2LT8bpj5i3W!}Ct#q?cAvGr
z*?@0s91}+v7)OV|h(~$`fpNnXRvSzo1ftx^X%u)Ez(xVpI+zEdLfmNh{*OewJX@2L
zEb1*7p4QnXsF9{QPKe)Aq~Q2%t@;)}*-f0lcwwovJhWOr=iut;g<&{9zZhKBU+}eE
zg>O@sYP63We#q%#Gfnjy>#QB6*`5dv**}goAu{D|<r7Q+QnPW9hr|q!h*sDMyeddr
zro0I2nA1%|AFOGgjJKXZ%-m^_v@LvgSi8P_K^jVyTrRPrPH+VeO>M9e;G;gt5Rnm8
zmcl~)ADX^9p349I|JZvZBqU^SC7Yzk-kXH%J<CY;k&r#gUX`t5MA^s4F|v2rloHB{
z@Vn0Y^ZWYa^Z2~wocq3C_v>|y=k>gvmxau=J6FyWf!ZDQiH=iD6!KwF5#@)2+fNT=
zhD5cgD4Vhsf@de1!iss`LvIYWL8|ioO=96pigzb5z*MlzrTMa8Pf@I)s81U$XU6E7
zk<1ZuJ{X$&JCk5}_~C1HdwXo&8=$tgs71;amIoAG55PG0<w?p>I7a>)^wHFtla+I;
zeJN68-Nd2;ykf^%xh!&lJC|njziYFtn0-)esa1p6>2)@z^Q-m4lPfF}8%WXj(bAVB
z^6XSDu+5bu4&(K;T*RYQgjo2q<O=zjH^F*M;+CFPw#ue|m*zcUzwe`V>jN)=3xveJ
zuC6M~qdz1TrB7p`+JguFjfhnOs`l^B4DGhN8bUkj4s-J6gk-()sLuKHat{43?{g$O
z3%MQHZG-p^2eU2_9jqZuEHgq%r*2vZ^iO3mO(sdt{KW9xXT5QaS345IbRIEY<BNv=
zq9M~l1&48&f}i7^7m=4^99=WBvzghv|Bf~JIGiaG<o@Hd-4E?w*(m*GY%S}35IQ|q
zV)0_<ZeX_vS&Zg=uWmYMH)%`Z9<%zGs(?M3j~M7fkmQ#;O>Hn+k2F~%f(Cq74u)>}
z-P!n_#>YU6e^iYb!eD5wvpNDMo5=a?U+LSo3hWl~kX>1ZobJm;1tvmFEE5oSTvzUc
z7TBb{^xWT;Hre}14!0UX%uc3~pv16PB|w1x)<_HL>QdWuG{+#V|COV0z&J!jP{_0Q
zcO!2j#KQ!@9eM$ic2gejX!T8RC}&*>9QYo#u)VQ;_c`V(cvBLo3{tD-?TnmVF_1QL
z1CtGM=F4J^iRfrv=^UsA)mf%h`4`?B6RU)%m4#%MHBuTnO&Z+HM@nlV-HA^DGX1#I
zNhU?+r&ad(Qt`Ze3I{e>hl(5gDuioH(-N5DJgtI)yOGw)#P|Is3>CG6mlIw)eU|+*
z@X{fdSDGzd=O<E3Q)a7f?DwBPQe<w!1s@exaY^!>%phaTj{B3CzntHgI;%e+xi<U4
zI9uhc*uoHjI9xniUs>ND5K#Hvu8mnDLH|5c&fOH_iT=*g_x*QeODSdD6-qI&l)qE9
zZ}GC>>V$~r&lsZ+3XF(5a@?WT@&w<Mai)@K<QT=ShSJlzd^~wbLntYD2Jta^gY&zt
zq2X8^v>vyU(J95;Wk(6f1elD~L7O{7!T%Oa1eE2o#!8)DO+0`=7RaX6goPt-1^QR5
zD=G}b?Mz?XWjFG$FUCf7x<uU51t`gYRcp;L1H*b<amIMF7SnV)0QQx<*o_ShqcOmU
zjZB<)a8uJI5%s+AoBAYIhgRxMiTK80y@mqLa$}=0F(us!8i9V4W|^&N!{17&q$SU3
z>nw=ICMSEoJPHe%?n+hSCgb^qZd^eH`Uueb$YUN?37oYyG)%sh@O)d*R2|!GV?^?z
z+eW;!6C<5k&hs3?$gHx;%J`49*{Ky5V_kmz*gH)N3rtgbabpHirL~Sv6mc6(e;tH+
zeo+ufrp0VIrPxW`!2XQ&p_4i692@`@UOu0koJ{__?yHt;yKf9nW!i9D9tM!60;>W`
zPfN04xL1cs%jNULHq<+lfdUjDLV?ih-9bQmxTL^pRa8+{)?p>W&)*+ok@tB+$MlI~
zlP2Z8BHazl@#)hR>}U8oAU`3-Xa0J=WHF<|Q|!b@21^$Y>q`Rd;~iGRn}ZO;KW$rR
zLP3nImD<h4$IZ>{ei(l-I}_9hAi8C;OY(~!6&u|6BPGiRf0j8vKi?QnMuoY;mz9>b
z_UG>&wqG2AS_X2y=QjKHu2-(`{Q&F57fay-edY5+p5LJ!B`rEYXm>%eK<AUPX_j-G
zCJL<~PE8u+S#7?ZLXEMHYENmKzq*5uxU_s0$}Q%Wut<U?Z{WKhmaJ(!{{Uq<*VGll
z9hHIIh89X?)in{h1o^QiE~)(~8Bbv^Mj5=nT&NP0OQ0r>1AmUt5$1j0eF0>iM7k*A
znGt#ygcFE6M1n5WTrO2oldP^`;WJc;*^IQPDKR0o$}~;#?zo6hp{cHnUe;F2)ME_*
z-=J^jszRV6E*K?F%&@V0+koqh@6w01zl}H$hixb%5F*^$Pwxm-n_l|H5P*x-kx=0H
z^_cyBZgZe&Q;d?x0D?fT>MI_lSKA<Blup=KR%GC7w?5&hJ_c7|c5%?23^_v=I-_FD
zlkSSWq5J>c(5h*>NG-0=yT#Cx4B`ONX`${{VsHJ2pJc;o{A}h6-5@Q-(TV?ym@7dj
zLm3DUk!aR<i(XMQfOqNc>pe$SPHz(`2CUXa`-^duGtlWnFEZF~vpe5UX4;ek;j3Q#
zA{^j!;vj0ePDK?mxT#cVJSdkt;F{Sgl*Fc!8&AzSsZFT5Q^hi;k?Sk?{OAfpx2(ei
zqF={z;mg70^iP6VJ8$aN?Lu___OPl`9}hQ!Ti8uyU;g>6Umh}q%S`ru3ocU$C&sF8
zJ@?bS*QQ5Lgj&xkY0`W!a98u>3@V`)yeMkTe)ka-$XMo`Wql1uZX2r?4mpa(Jf>$^
z%n?H%XmD|FF{?zp^^Q>4XIx{J5X(Da@nKy|h`h#~@_O7LI`PL33Rvyy#gSpOxVB^t
zxVx=;^-TvuRYmSiZb3#~Y3HRuPoFr<y!<TS(ktVHczBnB{EG;yCgHbMoKULLuf`l2
z7U;MqZ{kn4f8U@V`0I-P>=!1KvVQAtxuvG;$i)EXQc5MOD2ETrXUvZRet%0Q#C&K|
zZVGEenwEd94cz}#HlVeUyFkJg`Fu)%_lLNz$<&ipry}G@dAI`+az2z<%R|<%(IC|?
zq*^5jR3logtLD!pnh7{b6^Z*Mx?S34GyTxnoTV<Aa!MG(j=dfiQ3mbHDhR8)|D@I*
z8Oxm=8?LtbF|k~Jy1nl08{p#Jhl^lc^E_`16Ut@u4j_`aJC&XNFKwd9T!w>xU<`zR
zLaEgV%-4Re>fURt{7vg6XTRUx+56^<2BWMZzdMF1RF{{Ha6lm_8?CJ}pN9^@2YBHn
zj5ls{PZ5?qmHMGh`t|HvDBBlQ0GB$gW!42)*6UYUeHcks8Hg5XaJU&LeERO`pZ&9b
z$LlH=wB&d0?>_%Mn}(p6ak!|2|Cg|726uu`u3kF(uj(GcIAoLQxmTNqk(>0Txr~LT
z%Wr5wj-h0ECX5ZC)1f~|)SzA>r;N^?a$1bap;qKEvur-xz+@t+N1cs^{NL4#eSEyW
zt9^A?C%>he04e_@RD)3Oio_n>uJ)PUqkleoCDoCykOu}0Qt#c?UKc*13LZ^{&BaQw
zGamV&Kjq$)?NovG%dvAM^dan;md9ID+V`(R^Ln}XP>cL~ix`j2F>;BxG(-m87apSd
zygngqmize+SL4>??ejQP<3B!@+C`_Ed<}^BJteTvd&3k#G@AdesqKUip^%VhIKw-o
z+XedsZMbqn`Toh>^O8%xmb*t0A=|qv+E=*pdSsk<wJ?SR?!_IU?Q8yJ%K@8~ws6ey
zmHySDT<|WnH2ZprRjVF4Sl-&j^m0CKJ^jef_(zLTZhqk_B2+rJ>>!tMGeFJMgJwbG
z>iO@Eq=c=XPycF4EO?_QsI&%m)qc<`mkD3^ol3DotAj$c=Qa8av}(Qoyzp^lEIL}%
zCAaBd{%gj(=vK%AOWckYBchR}#t<C##5AeIK)tEZ*>8!>>&rdS_RR;!ds*g{IcFbL
zJudy$NavZ?;Xsg$=G*EQa%+BXPvB87ZCCiGsk4N)Ab7M`lyK?Ehhv?7>xWfA|N0$5
zZsu3+vQ>J|C3k1_8%916Jh|_s`<{9K=pb!jJ7tJLtm2-}tv{)FeYg$J@thL-hP(ny
zb$azVvqs?{6^T$!iSVCX8a!LI`t0>t@MJq+OOQJ=?bKy5`<wTE{_AAnYA1!{$1NwH
zqy8Gl8*FHX{MhXE_>tBjGEu4kYl}c+Un}+ts#I6q$GP3vb#vhC*W*_4H?0qx{Yuv2
zgZ6$6%NOb1ULjYV`THez$XN5ruzvsS-t-6d5@V`E$w8}YIIQ<~t;P;s!%>OJ+bGGE
zeB?W|E_HRAgpTFs-dvl7PvO8qVS6++AJ2|)DnIde3LhR?J;Rj=hNCP;u~(tKO1hm(
zo)97&J#lk$)6}Fm6X?JG*Oir))ajM|zVUxS3ixmpj>1dDWo56Oj7ZY$q-bJHurW*4
z!mEpyNu#oP09$e?643Ir!7RtAnV1?H+L=CLe}5ap9_gwku4-3F!yMZU0Z}}JGd$0~
zOH0Jwl{OHkC1s0rc6NT!DjBQ7h>eZJd0bK!9mO{P{H^#3AE;^7S_L=@0K>Z=I21<>
zRf-oU_O57VK67JZ2ggNm5@29KNT6)0nmmwWX_3tzdr{xq{KuiWwEU2|kuGfn%7~=F
znBMqlG>FVB>zKy1{tyRoiWOU1`uDaM_30nG#Kgx}MB3ZJPd3d%P~170vpC4us1|fg
zEud{X%xGc?xoL7JF~~Ew@#keNRWz@wb0$SH$KNON8*Te1r>oWwhCO{IbG^I`da+fL
z?-ZEW{T6XA5%+fcI?rn7l`gG2xqUcC26TSPC`P?!{CR^}ZmSs7K%wpKp*W|@ALARm
z1r@+jFRm!CeW&}rv`2iC=lOE;Uxrm_-kC<KMauDjtchC{POsvn58Oa2XaiHIPu8z&
z8kdGzt)}GM)OW0|2)^m=*e6}llNqQ!8DlN}_~${3C~8PqR$OhswY1FHj+A17!#%U8
zSN+9NAg!2UtFoqdTIP83%MWiflmvB6UrzqixKmb(LJfGdqlwTe9i@AYVgPtTShw^C
zA}GKbk7~=xR%>zowM;yKNLIPXDC8zJIpOKkTLBDIEtNDs4^Mz1IFSY22|jk5s@`t}
zga8V>mZb7OX&;m!C>n*vqfi1w5gm<mJ#ieW5cT!(L&)i42NWN2OplFdIsR0+D=W1p
z&r}mik~A$R=gRw&_L0|JsSrzZ72$@3e%`3<vpp^U=~X50YaEJuEApW^BVDNzVuSzv
zv*rx%bDI=8dphO@Q&+k@svnRVg7(lm@p7kEn#u$bx0&qT^(wa5ii$&UT>y*?ZY0F6
zV>-;QUUOCmpnj?T>`(+HAd-Y8_a7fU+!=k)O$BQU>$wtA=W*g+7S4b6yP-ZDZR8r~
zF6ENgv&MV12C`BZB^7t+kxUB_G$y8&Yo(uL{^(v<`w;E^U9X{B<+AtZmP=wOxL>&D
zeVo=?Po>(<R*tH28=qrX4vnM4Y79fmJp-cNym<Ty#3-^=d%};I@<fkDZ}^@Ie+yVz
zlS|Q}K@?MA{Slxfc)4!!F+l^oktaFE5Ky2aH0QD!F!~TO$@JFZiRBgta-k(ZM-}_V
z<61;9N%M#dRIyHYWZ+s57=(6(HlT4n*+Pax$irIp?)J6!VHh1M?@T;TLw*-MEmLe9
zM{$7(@k|}FUyM+=ATm%9n%iwy+`+@a>1<Hu;2p78UGbKbwd`;4?OIK8ZHD#KZzG`w
zIAtkxH4)FKzSbG$LXtcqm3t*0CvGQyh8kDQoHYleNp}d9Zy^><=i5dHDX!2m*c(Hb
zM}vlVu_9~}zY!m<OQ4MRowOSjn!FZzlA{uO10q+PeHQB29Z!qv3V&kAR%7n%L|Pw=
zuZ1F97RBh8n#Z2tAJ8c_ICVf@1U6m_bugFIw)#km(7}P&xQ2ibpE*=Ri*Ei_!7t=(
zS%!6A^eK+EwxG06z;94mVzbJ;AI8+<h(Hi`oUW+<cfF;v9>OA=e9QR<@w~VgTp=9t
zKfIxpYUJ8Cy$|_+{9cvnIDJX(*<O1E1z4e&cY6f!_mSqY_&{f(BnOIO9*C=PSSdWu
zk7ydyarSY`Q8@_YWfkkiZwv^$D6-;<R+(e*sXOJe6Y+s+rU*nb5%~@N`Kz+KufmFz
z?x?3b*@TL<pGG)gZ$%r&l@8%g50R5LZ<aSa-5okxiq)EOl?a!s1<IyNyc=QZC|=_n
z>?PYH{EJWMCzis)G{_I<R3uUbKr1TBE9BZ66*Zn8Ia$;=p;sU*)?P&Fdd{w#dEp_7
z@!<+=6d$(XZuA&GStAgcxX71uHTN-Y-bzcnc47FT1uRwwY=(g^ZPA`n+!YvPl9n@6
zY*C5`Rac4mB~ub5PFqco-5di8(axytxA3c1KLB^GL=u7VCWOxw6H;7>^52}jw<O1l
z(7}gwY_zVxZ)Elys|<C;LGXoI+ee)geavNyf{4w6jC>yfk^B9yDOV1BS)BH)c9c9f
zJoFp5-go>s$j$v>iQmZ5%R$Xn0YcZI?zKK2L5R&*?|Y#g5!kOUP|>^aO?iy&FvFE`
z-s}5Qh<nH(TX1CfwcUC@SC-y?<F0u-8C+LM&39qAwwSsm;W9#2F*G6b!v|?YLy8Ab
z5^^#VpjTgvdR};!{eo%w75S^AQBW7(-S{t|zd9y1xCx3nWo6F$#&{!+2I*kljVAdz
zPy&#XmP<&efWR1Zo-^~sQloSyx%kN$z#?fH8kXY&HJi!K&J?O@y0qNh7l9T6EzyG3
zc3zb=YT#`DoSYtDUS%nR&jMc-O^r>wPu2<31giXxlMQLRv6q-DR>OK&M$ak%7Sc$?
z0_c$|Z<96#zjlgl3|*Fo0-l>5U2JzY)sz2paUj?I7IfnV2|rV8cLj~>+dqUgD9W|v
zuL1#fs(sRzD{b&>c})46_W5F&9Z)zHIRn^)89Cbz0>2IpS#4dVadX*NsWu>5L!peD
z$^}bkho9JU@b2pJOrQGtk~sP{t5JYExWak>7zD&W*|;Fia=->4ZFk(a|92-Ee7?80
z_y3ke98~gNR}}IzhE}jNK#6g}N+c6f?7LMQs^C}Ph6i~iBqZE#7}0Vv#&cD;3eeZ8
z)dcr}8Uiu?as%a!p|t(KPlyY!3A%@S;{VWs|DWtbb?o))p6x?@WBSL#$I{h+sx+a1
zZ7y5kWCkb;e9g^CY>vOgWh4xJVFjp9sW%7<pn^w&U2%srt2;Rop_N5GcHsKpW-HX`
zGhP_RiYKO}l^D~r0ilG(h~D&Z)&VR$)?JVwb;J`pi|_hgLWFkhj0ZOcwu|J2Gd;)m
zloDZFREVIgd6<(~*L50e)r{SPRw`wO!KI_YK^p@UY@xlKm$CVC{yn^AppMY8uVF28
zMX0&};sf>@oegk{vu9KY;oOVBsb!VBfO1CG-bJXW!&(2beK!yjK<ABQpC)IO+cA5v
zbj5BTYx@v{>BKk)s{|(H|1x0DD<a3%ni1a96yO-Z)*~8<eG}4Y2x$fgt+dq)9?F4n
zT{<gMcqil2hAIqoRb^|BSw-y80TP<+uFyghI*vOlVAT$7YHQ`F7DE3`l0+Ky;ex+4
zgvUI>iDyVDPQG_<7|qMFMhz}gw9DJGPK3ezv*zGr;UE;s!0PQ#cM^4m-qGU}Nc!OK
zTw;zf<c5b-X(~e?c-6|XD70~|6kTiw%;MY7QR3)J@-J`chMyoF`FD#1IT$u1=8m5l
zJH^nD8VN?J#5iXY2Gi;YfA`2Npuzio8vupjdBpSfGFJr-_qU|7yQ!#uUTvp@h}`0?
zd?W2+CVLAON<kd)#JnfJRX5EjH^eO$uVfTPzTW%xv7lP*i}bt>RMQ|2F)Ow>9$OGu
z`R(aWPOH=ejE{Z{#)RuR8?S`;YTrg)$k2mtAi@KcQ}lyhSe66XDbNQ0k|_}bZp7w|
z^V%wcum5^Tfh^}M)e!W&4lv4IJ3TMlq1{B5wO@pDUsiV1pL5-gG^7AfKp^(0oXJsK
z@s0;n{P5S8zJ9+K9I8MV!fKjm7i0q53f0ryjE^AAa3(0>_XUaX52T~>o#<N}f0Oa&
z(*SNP9=1iGnXf{VEH|555*A8T0xW+`*i7XebZ|Cp_XH0ET?iDGLUvPQsii#^+kt~2
z`QHsMIf`LSzX4N8sA)AB7&6}J=4NIXK(+zV1O|X4+5PwKK<eBA<q64t&QR+Y0JXf1
zgy(#BPZw!w;73D)ETHk%u+nC907=CKAb)_`+Ic7TQ@_tF)P7Rj3E$j-MM}6LV`OgL
z|3Hk8+~-NMhpWhxO9T%$4Y|(;Ql)t^W2+w^@fRI(v_zCkLoieigz?9@uh!96z;$%`
z=&%^z_NPD%4p8fguPiJKf1^V#h@KxFp-L;kqoBit(6W=}eU9!@$~p=fP+VR9(0j{W
z`1y9*xn-(j*r50+h0h&hyN5@Qg_y)$vF(>{o6BxvEIm623?jveb+tJB2S}Pb0{lqv
zaSbB5kUR-5eYhTW3283J-Rr3j99<CMCOpqI(52;5WHO5Z?IjH*$B4jmsMNB>$A`RA
zORjsgWu_}f=FkL|rA-Id0go5+u*dW3Z4OeiKLjyXipW@Me=OcQHlkHhk)x);^aPC_
z_0ugk85tV~*$T3W4-}^E4>8}MSrYY_YY_B{w%{dVSU)1(&5RmvgSQjVsy5DML0kZD
z0Qv#`WT^AgkLo;J<b}J;+JDvCj3xa1nBqRqr8T32E1Bmv9`T3ysX~W@x}KxW&H~pn
z{J(}k2n5=XYHQaip`Jl+^hPEK4J7^3`FuK8(nToy_~&%eKXpbk?St)wyyjv2o%}bH
zNGbV$gUX<S@xhN@JEqu_RO-6wr!rR#_tQ%u1uBWX&_`sit9}~r{2EIwHFPB>dpPQR
zX|Ei(c5%)f0q+=I+VuT{&H)!oOBNc?+Ld>uTS;QOB|h+P%|?<a`)<Gh?&DQ5Cz)q@
zl-$(HxCt-WC}xEWb#y?BjXxD5q=l=+v4Q6<{rFd0*IG^D5U1az5YC$Wq$DWA>b5g!
z?E>13OnuHB^uoi}jeJ>c!?r4L)L22k`RP-6H9PcAX1(*|h0eTo`HQ+my1GSA*>0+(
z_6}TTBg%TV)hcI`Dat}3`5g!C+9ea8?OT-XMwiaN$!vCZh+Lwwoc^7Zk`i%?SSqp~
zO5^$xsat3DDV=NVA_MlGLF%ofTn+r84u-7I3wy2C8;)$&KiqVDYCqBs)J@qz>7dw=
z@kTN{(a*#t+CmZc7Qb;76bQ3It;NHcn%-XA_80da&G8y8ZjJvxE`a;04fh6X>L8`=
z8w1jp%ojfLnXCAPZ|Th=(60FJ-^20SrJXIhNA<5mhULcZ<nl25N|wdAx8JI$(b=2j
zsEgCQLOYCI9I6HSnj&BgnF3=7+*h-OZ*iyi$X<}?jgz&{lv@+^QGWJkvjcbr<hA#Z
z7T9eqLCYqwO6;^MpQrsPSB33qT05DXsAVIgsX+qQ<U{)KL}!h)pI^IvrBEwk7}r()
zeh!MIG|scQQyuwYYav8{qvu*dRN|kfO5KR4d1(ZT4`~w37ss-o0g7d8n`n4LJhQa#
zH=ua|PocS{s6%0y&Wy}0FnBXgdouXstG@PtU>w`lx2YSs|F%)p4%%5~YfGs=BScmc
z@pGS<sf4?6Y;b-9{^$>o%TqIjO3nU3VdQg^Vcif$E2bX@nn*$$nbsi+iPRCG2cx>o
zYfKERrVtg?1XJFBOI`U33;S72+?FG&y;l=3ocu;0DB}CW?eKn1{gc0E<NL3i0tR^B
zaqB(n#Wn(kfO6n753ZBbGX>URl%4vSc!W*<#2T52EFFgaW+v8n#y7q<)s-2KBHn_c
zm$sLe>bwkjcJPk~hv)>?`%N~6hKK!Rbw&4zAE9qt7@ckSQ!P_fsC9fNbor~(Tb}1-
zWidHiMNY2@d9<F!<|Wy^lywzOl}uW9sG+C42qg{6MEDmh1oFv0e$m2I<4zhN0$xyA
z`Ft$O1CJLR?)9F=xazeNMHwAs8UPFTUNFCu4-Vv@5xo^LmQsTf_ck<yUze3}RaX=+
zPX9~1OR$`-cqeUQDj>)%#v-`j+XGJ3WFI?<m#%;Ac=J%-x7$*W&Q#{N%js8olb)Kw
zXKk31O?Gq7{s04D$0C)1T0;S*XUi`JrKMcnMfz>bb<8;N0p|l0a-m1F=JUplxIfUj
z?(D5Ov&%<NFk8bMlW&D!;~>B@ldNDZmJkNF<-2j<sVmh)Pa<qG{J^N%>D3?~p~crE
zm12<2`}JVusRtKL3}|(t(NLF(Mjzd=vnvT8tM{=J1;=>o7#ELf6o=4?BaI>&J(Q-z
zo$=VLMV3BR%D`x>ZgbQ5H%kx{S4JGg{w7zJmqV@UIQph$t(qqXO~*?MQ&T7R#70!o
z==<I#RB`1?vZ#NPO(<lB<170#E0ptlz+!%#02*-z3PWx>ee7zOSLzV7ubwnwlHI#v
zEDR07{Xxv<^2+e0^5h^!frmEPZYCcJbus34hA1FQnN%6}E$>NQm}Rx=J3MA|)JneC
zw7$?o_bMmHZ8|fh#(Hw|th(h-T<r2D<57;=UW4y<r4Hot=bSyy-f<bxk|~QRU(_6#
zE*xRlkLd|#bQ$MRz7zMn#cG>^w=h4PJ>NGSmHNf*<Y)6$_w2{Dbvc*k37XES&U&mJ
zhO*~|nl>8dR|UovhfdK?f^YUe9DR0s@)d3pp(Ns6X3fE5Ky_uM+m`X$3Fmm_JI6Mu
zAPy*g^Jdhws}y+piJv#ncfk5U35=eee*g6AzyLW~1Te3lAQ1P52EeVKKllDBLeZ|t
zLaX&|rO}F{<pR;%^mIF!eBOhTxx4I1F)=anm}FtLHSr2pe}8{P#qgLI)f~+8yX??t
zn{XGTp0{WtSiJDAlJ`gC<lNutqK)mg5SkhvH@MowfP}FAd#S)bt#DRn{@4|NU?t{a
z%s@Hw4a18SzO=Gds)mLJN^)`yO-(4Y`SF3GAtW1nNwL<Mm6ernU3teGDfk?0lC2ZJ
z!lDZ0B3&NZ{a}%q_~A1x*ioJKwfhF?-8OI${~N-^T6Q-XHDPO=VSf9|{_Y*-n>y?O
zA?ck$c{@{q^xS^d{m-U=bhF5XM?^GPxVshL-zJVyluN6xs~guf;!2q~><|(X+Ue}+
zd3|sKyF!F~Oy5*ZiY&)$Ynvks1I=#1fdfWHM$)`oX=*=Cte}+WubYQLgd*o&9w(}m
z47Z9MtIwi27>pPXKhHSO;9hDSuy6YCs*YmWfn$v~O-fbxO2jOqHLl^s4*OacqLc;k
z9<z&op{*bqiDZ1*S@Uk_UAxFYJ1<r+70aBid(k1aGK3MYK^0qz*H{P(S3@H&<JnuI
zhdDa9<hh-2FPb-V62<p6gfkB^rC)|POdV^U)9xM14n(_s4LUaYRgxo9or%=mc8lD)
z{)!aN-jXMIHm5AhXP*8#1VVtme+M~Zvev3feuHyha8OWCkV5#9jg@d0ECKXDX|iX~
zD2oor%gYZ(cu;gtcXV`&jg7&--K?<i_VH0yRUNspvy*N$T-*KgL)^)S*hz{xA#0xx
z<0pU86;)JJtgQ>eHikjocMpi0(2gnk+*<tKq>_lQhksMg^+dw-SZju8wCF#!U@k5$
zXrKbdbzaySqg{AieZ3#Bcoh^B$SP}7cXoICK7HEx{o9#B=n{gJhv(DGOq0?<=7J$~
zvYdg6HG>IHG#aQS@nKQP46XkHf`XVhuTKV&|A7vKP*HDj4vwzQ&d!~|L+E-S-`<dt
zmhN1cqaa6GBNM#6nk>-9c_3&3{BzL|s`wV7xCtYLMC&Rhw1aAf(ET{}sb_1|hP?gE
zYVrNRU!fN?+cRo=BTs%S@f?&$iHL}9q^eLArM@*<et?SrIbI+}&w2k4aQhmJ*3gmd
zDkNFcqnS03c~PJG%{|46*3QX)0SWKHKQOR?#tz-t3A}L~(Wm8#S<mTp3*&}9dSp6<
zN+%gO@g@~cj!h^q5E{gpM-HctDkRm>qctYjf`cB6Rz%_&^qIeyPp2VsYQi+64rS{a
z=^-A@l^J!z=f9_M6it&q>~m`xd>l*de7C7G+9<77&}_zN?ZN2!7l{Ee$L|HM2EP`{
z;A^X$4W(QJ;`9n%75C0B7k(rpyu5CIFGcL5*!8pkU6JQHfpr$>dA(J3y9Gm?q1HdF
zXbTDCb`xVP{rx^&6LM<{v$RU*rtZe>C?Z(22qVK^NvQ3zd83;-^@eM%<*)utc(>qi
z!Ka8Sox^?gySH@ze3@6kw3gTL33r*~v(&QXJByncHhIWl3h4(C`IjGU7Z1Nqjz#9}
zEy1k{I?AW&IvTFED&2dm8+*_Y)sUL)W?1lZvJbmZ!=ra*-u`trQjF)hw5v5**g{=a
zHzXZP?(;=Vwdu_<v6+w(-FS?vyJ9@&dqdvXzyVm+Yu7kFiA;s*23GYH8rC}szPBsD
z!OT216w*7ml1_FT*+gk(s0%lU_P(&qA^1VEK6Qxy`T#Q2x&6;puqxU+snYpO(R^aj
zqFk?`<P-K@=(nPVvCF!5y?)=oyRLEVO8T7l?OLK&SvuD+H%Z_2&i_<c+rFPRSD2>M
zu<7<t;GAa7)#swKFLuFR&yvR?E$S?u(+HL*-5?c|b}*0BdDy4)MqRh0n-}{i)c6H@
zrJ2oPNI&Q?9vK;d*4~Hgmvr_>M9B4dXvsN8^&5(di-B)lV4;TTq<4c(I>IV%2$ev-
z=jP3{KJ~YJbA@hcx~V0*H1x4027D!YrC`uwtY5#WSn6#MC0AEh2PPT$14Tnq+sv#h
z3`ja!+Q!C45djLZ#z$L>cg&UO$kC88TxDQ*wj#NeF-%4UYFW?@Cs-FXiU?!_F?0IZ
zZo+h?D_2fI*^oi+?K?-qT6zQml)LZOSmB^EbSVDW+S#qGu8PpA(LL-|#Ps*4q@=uT
zi{9K$vUaz$Vn~1W3Yf|rT^C+El!KDMV3)!uB7F)#F60H=vi2&%I|X8jGa>;Ld7s+O
zj`{Bt2=cfKXY^BSAHSX0goZJZ01^t1&_Vm&o*vAaH`1H;G`HJ^?$RabrAjQ&VQg>)
zS$Uuf*|2A@lp9sr{PvHHZNCN^85|S>f%IZxT(c}15xGFxf3>CFnvhb@t7#rOLN0$*
zD4TG$@<fx1dj7EL(m8-!C|%Ip-l~9;lM_g4H;Q!-p;1u;BK}{j-<co~>(_J)Iz)Z~
zISYEe3F%1@JbZlDi*<Ogs?6l~g=Vpbpt$n-6LnH@%f5KWOeOZDrL>zy36~jUVPFyT
z;)>VYRk5)t)Zn^Nqr==wk%u+#?JCHb9bVT88&}!_2IMK2oSw4V*bV%S{6_IUgS{0w
zUf;aPSS$|i!&O(Xnb34d&P&^@JIN#Ea<o1NE}x-K1XN-^Ha1QtdS?|gj4>ox3#E!f
zmzK!$`Ns<)W6ypiGjuC6mfRpBuM$Y;#p}_bN221BYmtY4`1twrWCLuh3Uo{}hmf?;
zZ(K6sNY4BxQhbyLdAZjs=B|xOld@AjtMYeIJOq^Q$OSI_{>5a^$WY!pu;i)0|3j3i
zRViX@(Cp3%IBOh)T1j%By5r{;;xzzDt*))%AYcHGyGp^ZGUL<zE6pOFqT<gl*`&Hl
z$LKPWhLZc_h`~19{kgM*zTGc2AchG}iA0rR>}qGgEh&@q5)cyNBF3Hl)XD!1llkNJ
zydXBrKYVCadz;Z7Z=<5Et!*hvF>d0axDoM{%j-yyoeI~B7cZ_WmfBRX)K*psiXi(v
zQcY)+OIL@1S3=%QO=)w)a-*x=CoP@<Lr5sx)5G}mdb1AP$BizN0@wAl>^^T^ylvyf
zZ{LXHtTPMJY#QlPzOD)*yJH@;VcK(49>$BJ$+$41#QzwkM3^Ko6C4lS1nquj=n~Sa
z!ePgI=*J!}^Vg*#oB#qV!W1$z9#p@`-`^<4&l*HNOyt(iEk320J*Y1(j%?6jOdNa<
zKQOLHKo{Hk@@VT-SQIV-1YteyOtNNvGOn_K845Yr-!ZuVR-?N%%a^7G)iO7gk=L(Z
zBjQE%33|Oy!Q<TTzs3J!D(TIB^(qk3EC4XksU+9!el4`M*$*<}sYJGmt{)v8fm#TI
z>$(Vp*C>lJz+Q4KGCz4vF{(gw!jOlWnwqcLb@X8Er}dAZz#uPgc}@g67Jrh3>0?3d
zqI-u^>{)z#NsgviXpBIb8pMSZ0<?}>5w@AZlb-_ENTUje1WzG4OR26x`rRuvc(ncX
z6a>hfXEP)aVgn%G2RjrG!QdaXp6;9p4<=>PMT`j5EA{pD%@(m%1nvL7Q=mnB=Ybvv
z%?NSn1f?C^XS{N)ZC`R-otU?sxtbbjQ+2&s$p`;k8?ZFcX8H)aO#!S28&c+N9SmKn
z7cN|Y^O3@>EhH>VgeWd4fw+=4O|8iyH$5HllTY_JX!>Sq9g2#IAoRI%;lgc0_M_^4
z+>Lr?ICSU4QJt5BEuY;j1UdLn39c<*6oWDxj8*6bUgl)gef2v+6|0C)=$(e7s;jp*
zWN+ji%pf~7wXm=_KRbmWyKf;nI=Z*J8#?t|;5YEZbM-D6ZEd$E1}0wMz)@Gedsn4o
zl<f|({pbMXC9!H3$GL_)wD8hE&j~9x#JhQXz9g9jgcwgMs)W82Zal&Fpm&}G?P$=f
z>jPv;tz22&jQ*UQ978@j0W$+mTtPb=L_tBpY=aRib^E(%&`be+l3T&Yv|ksXg%VJW
zmgZ&}8W?DZe|!89>z)BU!t;2~>GwJaFF<iI#uhqOe@s?J$Q}Jpg7)cd_azu)G3hAA
zp2Wk-8k%dKe?QLd`b$tbRlWC}v%Wv1C^6;<bYXq_=7@5wFn|ARqzKmGoeB9vMaKAb
zkhEU?%>??xhDL4$kT@bj;ntJQb35L~DHr{PaUifOt4|F;?!KZEjOfzRl2~A#ZGtH)
zH3*nTG>2>qU1?N<ZfG~?*(7K#9dT@KX~`PSH8eJ!I=Ho%1I~f0K@tD<Hv!LMAi2gq
zf9}||RbZX2!U3JANoplHI5?`Rs*qRl3J8cPWR#XL+Q;1ez4v_#PTms$Ch6$t;QvK(
zUKbK_pC~gt2NE{4ZG+wwC}K}%W-^osPf^-|B10q+3BT9{O~x<a2YauEWCLkwweKQx
zFN`+u`!z3h)DM}=&dv_JHd;Em9NB>1|Ni~~h3tI3j0z9f?ldv0&`SUnbtfk$Xv&d>
z=??}5<1np*ftFSbgav*l!CV(uxzCxRE6`91Hw~{6AuA`>=aPf`2|yqO0ds&}fUmzK
zWN`$~dJg(DFirv&!7LZ(3yR**swx$@_YE|(of`W**-31{1u7QQR8$B^NF2h9lgE!A
z=fMVsGy^bKtuw6dw63l$b{0p#Z(BXRQ8)=O5XWf%gS;g90vgL6G<x0^pnuR3NXOU@
z!8GVabkO^NT;y!kM1O{GM0mLEbyHtoF;`FC+fP^FH3OFXDl?NONo9F?876dO!M2`n
z4L0XXgH*Jqrw6Rq_UF7?51RdV0qukhq*pqC!Mp;r+%RUb4QX05;wU4EM@TN|w`l<P
z^+qd)l0}w*iHQvHX<%R;25i8-NmVPk%YGT;RW&pQKwRKE#oGu8^Z<w;gswF}AYN9s
z33PObQs$s%_x-}tlR=3E?1TmgxC;wyc9*-s1^WN}T>=2OfdMT=u&V9#8Bi#<AIRJV
zoF6R%g94!Q3Et68we%UtLyUps2RVYZy!I?H29jqgNNcdl9+@Ixtj{1^g8nc!I4rPj
z+W!3}L$J!o1Rd=x>D=}7^t7^COy$-8Mj@K02u)$C)Lno7Z2bBC`?Xe1;P=sz#l=NX
zfdIMjpIEi(8xAiNSXH#O7v4RLwh(Fq=~=kB+tjVCtuT#3n1=^;kl29DJBC)Guka_m
zuo5tW0%+U{FEnmn9q;c~uQY;4LLP?WfPq0En0xEh?TZ{Q-EzFZNkj588zytWxe~N!
z^n@`fO-)S@Phz{YCo8iKR-kWoZ+BN!UA?HRB08F=c^bYP#93%bl`JggfdGDuo!wV9
z_dU*r8=1d4u!X?7PESw6r$Qx0=n(CIPG6*<f$|&-^C8bUg%;_N^yKX9<|ltaf6JRb
zmY~I{F&2!~HMp^IqjFVs2>SR^`otJDK|-~@{-#`p+DBqVKTqZbkfJs-n}r<+!OScO
z$iyqBq@~@29ep+A-$-XC?(rMR7Yi*xm$cp#K97&52&gnFPgCPe=*@zGB5)X+L4UJA
zrxBWBn0vvRr>U)hiH0*ngcx=dgQG+f6Zv`LO-TvFqXXGeDajTn7^U{<^XE&k-I6at
z{;p-z91>EcgvE$}6cK>Xj0n&5fy*8q+kni6h9Vf{r1B>U(^69_EL+TqM%uHb!2*LX
zD>L8&BaerRTUb;CZyE0Xb7*KN^fiN{*oHn-KxSTV@uo<9gxM}oCWZ+Dbcn=^j6=}m
z#X&rR#&Vc5FulRL@w(%6F*Gs&5&`y6tRZ>W>5To_H6`|uXBV!)s|W}P09OGZ2JC*b
z${w%g;$qwB8tc_;$(I`G+1Uf0&$A8q49h1#w+2x15<T#6*Kw5Cb=}>o%qoTW`StnI
zz(-X<1E>rJ_I6!{It~Kn-grZ&eqv(c-kz^x1nlfO7!^Yw>jxVcbb;#J<|ZNIhVxWj
z-HMB_XbHS2BXa^~NlZFvHbx+$zu;Dj_ZkMa0~jr+0X<}(XJTTydUc`JK^uV(bK>RY
z#lH<0vEv#hjV7k}!-rR2>sm)gv7XN2<D<cu87ZI`iHRAM8mzCbwyR`W2qnbDJvlx0
zg9n8InHRIRsw&%<Ehu_jeX@hYX;)wk=|)W+A2o3+45I<l3*zfodo+G+j5^)KUuW<+
z9lwT#2G(rbeoBP488xIS)6s9}&!3r|u7I@#)iqci5Y)=(vxO7@&K4Z>_4RdtD6rGA
zU>pQ4qTS1))L_))%hytU?%&dKa`#IOKm$UjK!+zy4R2uDu>)yZYEWfX|GZ*zcX#*c
z)2A?LX$2Y^U?a|Wj=goY<(z2?xh1g77>{Hwf#`!Y$_Bn2u7*ZZ;2~f&c&=ZkAuW6R
zcIV&#IVEQApUj=xDbMp_Wmp2zjMv~#lm5V?!4xEB`6rpZD@iBJsvI;yK|zu`cT|Nc
zToWO?gFOlT_E%;7vHj~i>Q<y^6c07=i>tS^hB^U{61l7TFob!F<%a*tZsI;s!-$R=
zbeGNtCbAMY`fY(@1Ls+)hHoY~yVei0EWt4Ad;}$drp88i#dHWD{XvVW7~>$!_yJsc
z{>zt#4<A}TyqC^dCM5$t%D<gcgD1z-3hbnMx<J?zObpmrhOh_`ij8B7ixG&rqGN-!
zDA*|X_|ldoElo^_5hm|!lyr43oFT{FXuD)tJL=aHb?-+D=z>v#!z=5(N*uSlDEai~
zwCE9cfsX&lfg2*!LdcjeP5NHWXeQO$oNG`>O;q6kz5#9+C?8ls9tJ2P{%!aW3_GBV
z_JObuRs$A#e|zDUs3=rMH(_Q0lst#}`ao@ipR;K;Mx#Lr)D%LSUXM76kVk0V3k;M-
zSV4Ua9;KM<Iyhq>L6jXQW1v#Vz+Y8bs>_!KkrEyPtY=5jC0bGffl)*#STzuY0LkK(
zCx2<a7KqYNxlI%*#fFE6BM{eLy295X5Zc<GkB<B$3$BZb>atQp!gKp0F+$S!TcTN|
zmbP|A|5}Z~*e_8usR>NFTN$YFLOun#APK7!_B6H6;ETY^r>4?Qmm&}_v<(`3)iBNy
zI@k>OHZ{~I#ctey(@dBdF&R#A<H+ONJ}7jFu#(*8_$?oANoh)3x1h~(12-ODV3LG{
z<m=b3v%$PkinkBYZ#!+?`OC7<;x9bV*T>VmAoh9ojzEo$GYkwN{%Dgr{HqDG_8Z*x
z>Z)zb-O=_TG|bP5<{sdoC7rE}#JWcqXp@w6S_P62Wo2sIJ<nE-AX;?JqXc_Ga&PLb
zA`-<mi}MwX0ld<77n4JL+}qaHY(K}13=Bjro0xs@{SH<F0M>lToKNQf`t=n^q8!M5
zvIoMKmu+VIOITqmIoBV9CW|M56v7IY-qZRJ_Nd<@R9s9<K?|!SJA1fn0OM;G-+MqW
zfm{8Q5SF3{NlM*@hJwn<Z7Ru0#AyO(fkE9q>nG!~q!1V>obu9A5=u%gvUZ3a9ECH&
z6Mc%U2bh~h2X+{b<@8GTVCd26H7_qOF`HZU6j^uKJ4DEblq5kNn>^QP?gYYSv<j*r
z`ktYwv9U2CRE1;Iv=t{yV49Yep1#Vo)*f_TY}wl{GHoxk38(Z~ktz3y$Vf_>n3%M)
z(=ssuxD5spPIP<LLE^yeE&B%?G;W~G{qve?i5?tlH4KO-Wi@kU@xIhJ_`RlXoB5;U
z(I}|SEJE^|pKfiB5<v=>?|UtE5|E><5=n7a{Pw(|W#?sPrsGvHIJ>XUn;z~Ny>LFv
zWUJ|RUHr!F{gf!BGEh!V4q*OIG!HiDd+~3RGl;Yg4QV+XWgjeM)+l-U`bORtgUSIV
z16p1zlKeyGTVX5N+p`cAbah3>@qc4B)!&e1%}H50++NV$$Cu7oPY__4O-Fp-@|0D{
z0<|yQ<dQ*f>42{vyi3s8t*H?iy~d_K1W~cG!y4?0w?lJfpL;a&<gYY~jv^5CW}55a
zS_;Xzv8He0*%n~tOM$-kxO4p@<Fq~D(hnBU><Nw!wg7QBYcQ`#RQ-yHCu7(F0mY}=
zb~(jkyGsB=c{n)Sd(Azt2lhKFGqZi^ks1;!k_dsGdFvA?1S>bU17s|&;xhdF>O8ca
zdUvJk>#q$!5u}D43KNcp3VQ{Q(8O%Sod{%8iPl-!{n$^ZqQ2xeES}%-c&=0Q1qI%@
zxR_&08J{rqTL2yd($vh%c|o+@DZjp-zb+g^X`Jf*iO%0?Yb~s%AouG$9$*H4wB-5e
zfS?zq8UMRz>B6KAIhvm^KYxUM++WLs+4!WPp9|9Qi9>^MrpgAxat9xzP-Q~rF;<3U
z_OaC%c<DZ8_4I7%G@96M1NAeIarrb|D(*QRx>w`H>@gcW>}w-JJ^Dq~YdNw3O@sjJ
zK8t**L9J#V@p*%YkZ`g}-nQ^Iw}uQwU#aDuq9!#FQMB~-56P=~rL@Yw{m#mCc&z%~
zNcU9y31aSrAYN%mit|hOEa&B+JOlz%YBu9?bm_|e;G|kAML-75<XF7CvI6nB_F0B<
z3ghL=rDLdZqPTkJIc|t3Xi5E?i*jpB1g;M_ifq=tfUL<Q=zDHcQ$SH$*q;%`BOJqw
zSoBgNU-^V{W9VI?>t=LUu25v@uFGF<%8z0Zd;TtWaA9s5t=DVOk(NIz_Se}MN-4pw
zuP;e{tus%ac=>5HQ)TNc!ho;RM>kzwi!TPWRH1noruNC@oJfDw$Pq<wML%i3A>^X}
z@sZ<UzsuAEMx<%pJ?D=WkXwm}8|mpO*7sVx=_e3vKcnByx3;(3{kgSuNziQd=g*3b
zZAhn0NWZT!XAL&g))wuuUi|s<Cny0r&#9E?nFcX@bSnLfGxgf(qa=9WxM$BC>z>1?
zD~W(QK3Uo9zLnKtkyF6?T*ZTS7CR`;D@J^1oLq;--o7k&Usr9t=kyV@2NyXixM-yv
zhgsLR3!KH<v#xR0jl3Ceijn<4F2IbnUY3afeYRhBBB85gjunGZo_4;b8&y6EmE5d)
zGYaO+;Ns!^a4q_pBWk}D|Dn!fQ=oeMs+#J`%E}#<E;8rJ?-eN^(kdk>`K>9s@IWQu
zQECJ$u^x5y$!yI$L<JBnUmnu14mYcud|(Q@0kR%Q7p`A^GI|03P20^^pCGr6C!Lrq
z8M!J}R?*UO3MLu)uD|ZQHh_4IV*5Nlr`EImqpn><L<B$t48N$9R87szvF{=IPvvIO
zh})E}x6xI*?{0NdHS@|*gaicO!*9x_G=ui}((a|L?5Ls`ivCj&2f_!SXz*UCxQ-^S
zA)pbU)@Fx7&?(+@fUgE2z0q&YMBeyecV&9qAxIT`tJ?(CAZ-Z!_rB=@XzP1ctXJxo
zC0K6;1c`A<5{5Amw3?Q%rS^ATh%&DPm4041mJq7n7#gR?Z%IV11qKFA<`$^ox!u;R
za>Yvw7h{Z<_Xeb8>YI!oSN}K?iFzX6wsckOgNBNE;#qfQfJa0!is3(byixmXzd}J`
z@r8Ik^`L?>@!^chQ;Pi&NFS}yNI+rA!xqFoh^XGvDZ(5dAB*?xT!a!*b*-4UQPAhB
zTsnFU5L>b*z12lqd@LAte+hQi(IiZIj&#<ZjI=VxtA3jSfLo~@O6}(*(p*#pwEXM4
z{yU%W^{Ng+e%v|lDUt7ox}fT9Pe#Qe@BPb<+-L!<bZa{1QNNb_I-&tp^Ksq)lN>-H
zVKnx){L%VmseljOWm?+e$_llT(R)?$Z6nA*`yb;Wj8f{=PrpAQ<R$)K5s$Isp+$>?
zvpTN27Iim{C<s|x#;Go-bT+60UVYOk!c9O>u$0fm#-?-4+o>7{=X}sqE^UIlc%BD9
zM1!=FDG_HX1iW|`ve7p;2|cvlR8mZQs1P)`ny`6=`C%1>T>02rYkQW_()fgggaacB
z!idecRQ|PZN30Z~hzpWYgF5UKO@7zy2N)PZwZ*i9OfI+;gr=}sN=R0>P3B)P-0(8A
z?xQ6n5J>BqQP33ufFXQyrB3b*3@+TPaLFf)cYj-2>I4Zu-??~V6NNxYiP&GV<&6z2
zkdbm?FCsi~oG%#F?5?mTO~DW!N*ce1bAI#UUWGyqCC;m~`@;1UJ~`hh)e$Y14gHLC
zQ38QML6AYc>TzM}86Qs-AWeNP+*He-(O;tbj47Rf{~Tty{+CiWCbqCcy&Sq(;f{6a
zW<B-cqHj(gup~^uS(&|6G?GloEIIt;OU|bL?63>IYKFS@ci6qqCSAa$!eqkWYkwW&
zMt@<U&g4*^;P&GqEc600SevzVE6NZaxr$BBas;;$ew}GQ0r16aA<wakfsm595$08*
zcqX0pUuImrQa6|W>>2UVOz|}W{%d)kxwpJ>p3=mJzj<SNMatVYxKd7T%-!Bz93=X3
z=5PPV{6XMeP_?xm0_3{-5W@E_BF^<nh6)wdbpS<N{B=W<($v{STN+5;Sy@w;@3Tor
zy%e#=ji<baPAB}O-v~1)A!AMXB9Hn7uRS_4QUef<vf<-0(w`5VJmzo_i{4F|6WlfL
z-%s_)WUk8v?jbyy6OjdnoNR1tJK+PY{`p41-^e?Liy$XsKz?tAV&@(Nd8$2{n|Nxr
zwKMTm?(HHPGHPA+wJ-7G83!3*jPTa^Q^>mk2`dz&ER@@Cd9?KdRNR~23JYPDJRLLK
zP)%=ZCnbL3ViP`t?xY2$hYwe_x2d*-Og}6#+XJyFklU|E{i|g{6G%3J8Sy)du}rv(
zt^s)K2dxUkQZQDe83sG;EXsa@JlgrB%#)uaP%Z+7Z;l^$_OXAmdB?c*Y%D9M!{!~R
zb%}yT@}*0c$k8fyh_iIXvnusF7pq2U%s|4Y%H&-*QU_#myrWFE#KZg7UcG#IJ6LrJ
zKi3{-dpK4OHRCFtb${+WH0TJb6YyWxDYAw|BF;DA5>^kdT)y15vJ3dxW)P;ko4|7T
zPphg&Xx%dkDxz-;_75IBV5I*6X!G`$j}m`bco?7FvLFVFE3viC#3k=zV`G!wBK7!}
z1)6U7>sOXvp9wBYd2>vBxmeF;mqbTo<K{*|A);~Te}j?9(NL)_zCZXq2E@=LpZMgo
z(l)IsZ=|SXlK0ilxIzef@nQ<(fO^3`TLi|(&+rf+?dlGJ6VPT!G~7^2QOj07IMqMm
zyhj`p_kG!W<1Xtz_)=`_z-HUF0AX5RAIatQKp^EXQ#t^s1Nwl`k3oP?3dWJ<<aN;8
zslE*e;{%j!j1~`4;F|#b516Y(MogT)yN+D)ZUWk>9Ei;o6|F$B3(sX$|GvIHieKOQ
z-ND{oWbG^Kt?<{oK#X{EsEQTcgOS$09uV(USNd65<)@_?!ckcyY_<@353|w0pKvn4
zDb){Lx(z245UUq3RRY8*X1Vyp#KxfqeQN3ugcGmvb0qYR4c!By3RQc^%#Im9BC#|0
zo|vnsd;+HIqB_)AurpDiC^PZC0H>6zg@%rvUfgq00d>z2Ca$io-fdU`xNQ6dqGQtN
zqjipy4_6D$IIJQeAz`Nr!(3!Mt?bDB#DpKrDqBtPqjyJ{2n3#CXP{11niKVtzqFWT
zLA>mgWubZzvG;Rk4CE5Ky6~T}$TdN60@}&J<Q6ZGm;CXUWgrVX5A^rPCm;Z+z>}jR
z$Blg$Wa>mrK}?LD|NH0>L@qZfVn8^AEmr8eYRb%;oCL^&HI~_3#pmYdpTM}>w(~O?
zUuJ0I>npw{SaNV}RkeCAuqr7j34CjgeDH_rYT+k%e@}Io6MAcFYC<_HVIk|~JQ%&L
zDS)+%fug1AdE<c#RsjQo+e4Sbsh0yDoJ=o)RN)Q;R!PasgYvy%0;Y9frZ#=}@Cd9L
zddbAh44yfsrVpO^<sLL+L+t@DFyQw9EeQ5KSayN%nuaOLphrI@<ClI`LXPn0`zV1L
zOqB$*2@Q(y;nvfT1Y_Y82v|Dx?S=1uSpD^j@^j1Lu0Lo5E_+JCV(mqGOEQzAO#~((
zxrf4x#tRb=B-;hZ9BMqXT3*n^=kG6B!2(mV&p{&d&UTlokIx^N)C#FcFr?M6D{5L`
z5I`+J+b>_50?Y(iCYkwrn8)FGm>IOc=?7rd|5za-F5^I2Ci>RV{!D1#9%XbbFik+W
z(8Syvc+AmIH$<xxz`Lsn{d;sYOol5XBU26EBO)@~*=Yml1Xx8-Ru0cCX=!Nzuu?XP
zgbS-<79g7&5pf5uyVN>XXPyps<~m)2V+j=9kuP7aa50XD_pWpdM5Pftx@>(|4w;q)
z;R*z>_~y=;NYEO)Ml}NB_S*o6;#4<T2#Fa$X#<i#c=)|+!B0@p*oJ3=Dm_GX=;2*J
z$D#P}Y0cYIzz9Mr5D*)GqkZ+;(+;XJlDuJR%*G#>_Uqu`5dtBXNdVjwsL(w+GXBjW
z77`MYnLGQsdto83z`Ei6`^P>$E1r_@a*E!(*?_VbGeq{icJRNV<XO^wWkXZo?kHa)
zC}KdV<kENH&f*Bc<dPE3Iuqa=&CfH_b3==Vw@dJ|0}SqI*vK=0WvU*uwYHXK?37C~
z4*`0#bY+W#NF#6{_;@euY!Xx?f&QrK;hIV#xKADX_AQS|BRm6?Ljf_3v1X8pM6Y?@
zWE=#HX{jJmLopu=J`}h1pfr|~ojp1>2H@D@>Z*a68EL;*-skP<cPp^rpf<xnLz2*I
z4n+#6%YbN(gxl<RyF~&atMQ@MD=`vFz+?8mk2zofwtuY=_V@*hoU^@s12)S9P@&1Y
zdv^8%SRPO=wQuyA2G9qzYTX<h-UA8>0{sx>L`6nszI^E=$?S$E=U0Gw74Xeq;b9if
z;%>%XXztCdK^9S)uKyL6+@T2M<fMAw0cv^{Jb+BDZEru&(WwE31pwDjECEsW4Npnf
zwNTK4A4m~lv#xkI0qO{8EdZDBa(Q@o7{na0Gh<Kx{p(4%8UjWc8-3--90F=vc)#zC
zq9UAJP_hFxx>`t6#|8#|!oyn#U3w(ndUosr#{-HcFci8Jo(wEtZf-6d+1_3v5f@X?
z$IHvigl?Ui5XP10ksM0B1Vv?l@Idd@ngGHFMVN20_YQb<z$N*dBQwP@XZCK!6|y>(
zgpH*oYq0P@4sdZH`%#^jw*aLCYjukA)#YWB@ApyQO%JC;S(_**Aee_=io3n4tlWb5
z3wDEW+dmpcMv#NAfpr4cn=9#6h(ggED*)sF4CJ=8^5nm?1_uQSagj(lTe-Va>ibPr
zm_pGg)KUe+YN5I%>oi&nB{<-o5a8p3NdP3f0hDU$IAQk$fSi%1fIZ;e=x8eJw)Sqi
z$fZ$JSH}#?2w6^k5u{pwj`v+Cqd_=15vnv4OyVMWr$h@oQ1^p(49LqKOpJiD3}zb~
zT77LT2yYW+4ZZ-0lNT?L=EsuK(q|CwiQ&3ZQc?n9i3o*p=Rj|1fqU4Q_TReNesJ~J
z76z_eAdP}lv)N~Z5)q1C64`wUclFTO`2*bU@UYITZ~>vUtN|aLtVALq6s^6RvRR)2
z90rDw_G@aNEmUB9y}iN8js4m9OcD--MWWd5S*R@^)IzBvcNlg&0M%Fy3~U>CW7mia
zD=IWf^u9pG2lQBJ>ErQI{r_P@{rzOQWA_U6N*9U7@&T`czdIQM9q^^+dSMFG5E3yk
zWCB%ZqjoPf80Iu!E2kQnH9i|a@r+c2S|im`o@KMYE*!UNH&<6Em|2{70p?>k1A7H3
zh-|X57Eookz3#t=bDJ9SMJQF}idc8TaNWfms6C{qDN)}AX=?}<>mMkp&H!41Ix3xT
z2A~_t{Nb#y1_Ku1tPr6=Qx-`$)ZAVx7BoVE3Gk*xcd1zn#`d}z#<K0H96x_7D{+#F
zfxdn%97+&o)p#Wq^Y6C3Jy$aj6B$IT+M%3o{X@arG=~cK0MS!G$bfa62X6<q1TuKy
z@WlQ#3n2+<X=%v%S)al72aWl}&&!fp_VH%|7L7ZgVk;^t3LFI}O+&*8$VcLaSwUG5
z#vojI;s!?@wmck*7b|d>pym$~(8$x|V1UE_tL&?zqWYq?FANAuh@><IF?0(ufFhy9
z&@s|6gmeo?2_j+9B`G2x(vCDp2n+~Fh~&rsBAtT341CAmTHjjlpYMDBymRKB+9&q2
z_kOm81xIQ53uYiOEx=g;_R>|FjSY(gcXM$5fudkkz{Qz~k_<o@KtXV*HUk;31ylcl
z*x-Q90pbjV6GRZTxJ^pt+!!82Jy(A-tM_ASs`cT{RasdKz{Ej=fOZdr?8&<VI<iZg
zoR$U#g+w<H&TkwCHuu`<>cqR(w<pKO#;U5S0GZX_4j>{OTX6{qd)wRMfZft1{`@&C
zGb7{v77hoF?AcjGad9uuHPB6-Gr{zJpxKF81LxLSfG%F-LX;$@8l~#oyN9axvIXlx
z^=bj}AMcL%4m<g^xzn(jKc5|<5`iQ4hVkkYFrVz^0K^E`qJeZzDFel>pc+FT2H+%*
zI@7G=<OMvwP>Ne~h(>)?^*X%_$5|=x;S~!TRGkF(h=!)78*~_*yLTzipD$HpE?GWz
z&HpNMLW4yGaupD!uB>0w<F8BY_D1mYYdA%woelx3T<|Ea%dR>=x=_i+s(r_FmyZDr
z%o%JMODVazRya0(Fn)Rhzjffis^BII5_V2cO&NUnQ<n<pHNhwMi~TN<T}kgR6*T|n
zYVJ{CPzoZk0%qX?0ws!~uK+>0W3B(@s7V?RC2(sPqzAM=MR>4<gD%pnf7GCh*5-R7
zp$>bg)gxX$?*t;ufTKGwtRqkQ^iW{4j8ip0p-M-4L2?<N2CvSusYtp~BTdg+7r6NE
zY-IyaJs=TE$^7>p2sDw@ts(M;JpInP`JQqZNM81I(WM8#PE5Z?4prmaIcnImW1GJV
zclLq4^}9CI#qhTMQuS%`TeX<46lSj;cuJToJGTr@eJEC&=JfXR8eRqp`SwFZDD4-3
z0zYE^c$qeQtdvcf?40w>xpN}^yDY$b$V0aUoF%W(QIe5<dv+@Zgn07M#X6En4uBVt
zx1_JVvQ-CZ0LcnjBdMkSEB=+zA>-1&ADz{CIswlU9i^%iDXZhZ(mYw0{*Rx!#(h#U
z)R*E9%oQ03t*zU9Lw1+H*Q9idKl|!98p15zi;6j~_V2YM!Dh<md1~z!O@ZKxl8-Rx
zC1EJvcu6?>tK-PvaL6lL?-4^$bviP4iT~Qolf}%4{a-ZvaSMpWvk$++NtbGRdbl#O
zm%o)y`~5l6F8Q5W<&=LL4{@g2@9CW7>xRpaYWJII)?Zel#bb_hrFmyZ<LTy7Eir$*
zmt~LRcqeWUAFZP|Ti%#DaGo4zHY4^skEx!Zb&(v}b=2p*rN*rVVQ9{N(GTXnh$ce!
zWZRZ^@cyp!n|v`C`hz%t42yzR8rl<e;fbsQYuB4vK|sSKZ)I+g<6X|_5HtNmA+5jY
zUxeS-!Oli`+0|!w`F9g;jVa0gOU@i(a$QlT58;BV+9qN6>d^)(*v%dI-El?^YstNE
zZ&~iJpA*jC>Bi%l!jGkRK{}^QyRAdg#kd~itTs|dM+b2dqK#Vq%;~nzqGgQ=)WNon
zha_w6d2loetc(re48jYjL7GiNKifNFnWp6Ddyq^)Z4;5YXh#Vqfb>PYD#jtdD9(yh
zpRLaIcRqV;B#7;xh~LDC`t1h%96PcOEvOW}f}Cr5bV5ujq|mUFv{YWltKJ5qVT{r+
z)vy#52=yZjOFeD1%xPd06p4CGJXH4&2reysL&H>{oifh&+N|Ng$HV(qGQT|AA!O>C
z#n5)!cSC}~xbGhJcBcpzKeH<YY1&5Smq{wGrQUiBgDCZF!%~}T^mwiSnx*64*tJdT
zffuj$f+ouQ`y?aMi(XHETAHA$d{8oY!_=v}&G|%sh!E}M13_2df^;V7%~<SvBRU9D
z3kq}5d7fXT0@#|enx_K?z6_rDefeas?Kz7zZf#D+OSylpgZs9w3e^+4ZecZ_=;lto
zv=$HDbsBA?O0p`WDH5iIAXWbnZr6Yr(Gk5%SOF)QC3OB;Mk&zH5*d!`3QT;KbQcmr
z!WWZsUp<FG(&OY$QpeRMC!dT9g4QrCQlSew@DBh(()RKMSFjGzajLulR#w?{K6Pji
z=gM%jvG20S96ItXI>|`KyTg8M9`^QL-&Imb#pJ%8m1S;nsyW{%_%oJpwYt2y3gje&
zV&m(D@R7bw$UKGaYdcfEI@wAL!1v(VyXc#!gN)t8r;#bfXkG^j)q&Rt|3RLsSGx7E
zSZj-xWjU9*y55F7azx9i#UD%xxvJW6YLs~ZM~Icq*a-0hoGJpLJioXp(uNZ;y5vuo
z5XTDU%?_?eEN7P{`YO{r@}AJc;*a2!u;Q=TY!V@lJ%4oqcYAUedfqAB3b!+Q+I8VU
z+`x4wnGT;(rR4vb^kUOd7bh6y_#h*0bul{2<4%k}tnc1iG9JOrvz>)i%j)xir}t|Q
z+(avo2>@CO)nFh4lQ?X|XsYl~t{Uo}9KB~`Qfiy2Kz=}T;8w?CMuiYkYSF<}wT4Ze
zcewy<r@#A7q+<%@pC^b39bS*yPRc&PtRuYOwQLj+)T|oO5Rw74`}#Vir^C*6a5^4C
z<hS_fO3LAW9+%<fXl1t@7Db0jdQt0#gIkL_-6xKP<>#P6$<1NKj)_RSBioePvfZTS
z9trkv5O^?s<R;JCu*<)56iMufmP6co_yooi{>W=Q{zDHQ|H9qOU)-`Hi0Ncatm&*F
z)%*^asw_Oc*2O^C_Hx9`c8yPBzG_y8OaHOV`*1BIbN$3JIP#$bv!u_hp7<gM2@+}Y
z^iN{eCnq1Hg_=y=TlLE>@t(Gz*YcclDFInxKAfq<Qc+l<J>amaOd$0VSQxSRq1H`=
zWbcd~)^3yYbxozEZZmgw$%XA388wouwvL0O5onh<s(9DAq#6h^3~A$1B%D3}OE5kB
zR4fd+^~>f2Q#7>3!G04ac(9}%=_|PByEY^$!UDGXF8P<?{Rf<F4)NjWFT@{fN1L3p
zCDJU~Ddf;kwe*Y1pLXBU-*G4tz9J&j*AQs4FlFo-fD<el(<xK)l^%!|kk_Ai`>sjL
z%hcGw01og4A5Q6zgOnAj;znMJp>`4|?2coMzkfRfMH@u=)&n{)Z@GGl+cqZdQyBk`
z#pSHIOlJc!*Lxd7t0ofu`tbrPkS1itjZp0RQ9(!l#j^9y2QU+LF0e=1fYZ%X0B~tR
z?e!cgq7xZY@@8c|o|{(eQO{|%+{D5D7F}3Yu0rCK$#~;cO?~nHug{AC$uXFM%dz&w
z!p@`NQy_|0^km*QudA&xP0uTOX^7qjrZP1du;O?kH!njA@s#5d2*Imk4Ze*^ZApbt
z^oW1ElVNbDN4-h`ep?PO-Oc~Jkg56h)h_Pj!OHFvBAG-rW8F9<eUeUN;HSflxmLNI
z#AqmD`0vXs1z&8rntlDJEyf3wA6<t3vUbA$9UTRFKgFNRTf9&y0Nv5~{IRK-gjjMk
zr=4O<$M&ozEtQec1dot?vfHJN@}IU@9<yHtCish%LA=RAb7?h@RN#T@vh88z=&AKA
zE~^Ie+CWSl%bPd%spp)rsiV^5^E^%Vb_rr@6(2L@M!c}3kaSI1jUaDvSt6?@Dt}Nh
zuqc0MvGstlh1(HZ6Z$k%DeyFDsh<yJdA!`3k@fq`ZZW5ARW*g<Otb7J2NU^X{8hBU
zk%l5Pk>>fFq#d*1%S%sVGdVS#(oGB%E*)v^|B)zbQGJ&j(ty+Nxs|XlhJUVI;7-uI
z5a`yNU@EwW&l6IB4NMKD!5~K$&-a5Lu9z_g1eB*&H;JMdxM1&&@4H@5W%~(r*>y~1
z-!W1qToX*&+cA9)_FMWbhobVg;u{4&YD2bT;W@7{AD)T*?d>sdd)#{$Rk><TQIUO+
zWpsK|&EoWCQ+cR@#vK9IuXqJNnd%P6Yj@9p*W^J&ZSF_p>H6)gX4M+>H7#EHRQ^RI
z7nK<%03!@Vh~Nwa_34D!TXnuJEVRwSyQabrvtLy1i||FnsSDeprp@k8)}I__0P385
z8{V~2I20jpPg}4Q^VfcX;W^n5eeqcH*5k}QcO4Rl?~%f*D_4Xi%v+7;KH`Ro`Z$Ai
zsjF?{y2zkt)$1P@&F!F{@_~6b_qp!bHj)~mMdCfVQ^V`=?+>+iDxRkGJl-KWr~e&W
zJY<TS{PK=+K!t+LqUM{j@NNCcN$w+8Cf0#f<;JzVF2>XxL_<34O_ige02O2v9MU6p
zXu&UfS}l~C=<bNccbCFf+c#UjsGhrDPF&LOZhyH>lV4uU6A&K1r`Oiee4(00h@$?4
zO%>Jd{{YuF!w75d$Eg_!kVY6pY$!`>bP@y#B>m=M;9+k0ee#6a)1vszTRVi?2I#vl
z%j$cC+gds3D4_m^tufDIQYeD&Q|Q;e#j;k;JcNNC_3~&r>B9iRsH~w$*kL5oxd)pX
zq1<@|t8pxdN3W!?UpKui3f3rS%hMsqsy(*S#-2BITbF+cJ}dJt6tAN_A;WOhyWh$G
z4Wpk!dAXp=#Jt7f+fKkEEU<L)V{I)(+%TZL)LkvML{i~>xxszoSJou^vIEajJ~CLL
zTCy|#bY2b%@ToAgsh{J@xGHjLoSw=4%Wr|CkkC=~CyxA$<rwPx4lg$?srPyMynD=M
zqELqaryc7+7zC+Y?wU?Ex>+9XZBVVNe32>)s#dnzp1*~*%R)_ig}!~96n%a0H*dej
zP1{?4{I)v!OJ&}~VN)npl4zA0yjj1W+W(M&Ak0KQ%9jFMw6G)Z#YHWr`rs>ZsNr-z
z$oWE2?sOohuDf<jW=-ya@xc=7q)aHRE3?7El-Mn4m}VM!c6fR6H;QP?BAelNVz713
z%qVsIb5Rs#(^Ge{II^Io=BY-7d>F>AQ2Aw+SqwIwi(6UjfOw!7@JlPsRk?+RNeV5I
z8(3K(&pSD(Ha>!Fz6$Nj3=E`<FXmc=bhv-U>se{m3@^h%&AV^9yr`|=R?<IYWwhxh
zg=#U~qmo|M$rC|rbMM*HM+PW#*qNN+@Ls2n8~B|qH=6;>hf4fGqG4PO*QJFhw}OT1
zqP^XO#-J5tkqUxKvz(W+?hUSU(8T9*ynbAixv}F%7k_;B2~#A))#HYU8bqSK9&~0M
z)$V_>E+pvYFHOusZ)w!b_QXr>>$7d406hgc`xtlbH2t0~kudl^IM~AyPUpL?nbhvS
zKlVBIxU63r709lU*A5<tKXth7@b?@hIQy=nyzAQ9kdC4+-oUQZ^ap=|$1(C;4;2@`
zZzIjxdha<%s&-*f@lwbXvxA^8d7v1==<wJ<<%ana9_yuV-OIRbM#s(3!Rtpqsug(Y
z^^isoL>_ui^|q0FkH)_DBI=03i;GLqN2W8lx9*Y$^YMcd#Fgo5`=Opx)Q~LB;6Pwv
zOoLb!{ozsYkz>VLN$(Y_8SIR~`+TJ5G109jz~gQDi!P<5m@vr^ToU<Wr6Cn(u>U*h
z@!WzMBs15BC6y?blJ5!mJL^R=UDr^~z!+>+gPhhDr&54WkH>F*?xU6Lu9Q^3KzQqJ
zf-lqNEWy3%Y}KRkaK^F>Z0nGo>96wx3gitwYlJOxQ)VLvK7PKL7XSYS6aQX$9XH2=
ztyw1`$~-ufycN|r<h*x7kbK8)CGMb6%%biv@JvsBr{q-TSb2L`LUIRvbUGJL@eXMj
zV!<FCWDpB`+xN4P?kl&>PG(+!q7m!;a>1jCJS2-NwJ;Oh6?))NVI6+Q&bf3oJ^YL@
zr$DyOtZ3bw0)BGfVvMC%^<9^kAhq}X%7L&D$&};S@0zOHsaj?=5jOY^Rn;R}o0@2d
zQ4fDdRR6HhV$tO<V^GU{oDw?CgQmIl-heqr^vQp!s&ApZXu8UO8twSw|EbZ&GDqRN
z9NOe3QlW5v2bVDsb%o2#K}<l+4XBIlvj$ov1RYc!S+CdnXtm|uHd#u=SX$K3=8BP`
zoG#|Zv6xZ(g)h80lM86;olzbhr-nse-%}r$H@q?Gx=wtwp(CDO`Y7NOFwV>b>DzA&
z4IMj;w*GKhN(;N97#j5CBC>}=Is7DZxT6O*NiEdW)RsH3D~a6D(igzL?6wFuaakJu
zzMX7G7yBJrmAFPH5+lm_T6>!8&;CccnI9p)6N~l-H^ttada?WB6)oiJ4%<D4dO7Oj
z-m4EPUL%z`OuLv{FBDbPF%r(YH(%3a=Z~GTWjO1Vsvbv}NvzjOHRp!|u0Q`@36a)0
zO!v3t#V}8vZ1i@?p#QX*<b<t1?ORBjDUs+lyz!FSGB_QJ`A$iuqMNla2*#xl*;p4S
zEkeDzX`gzqMT0pU^Jj0P?l@lNjDIDTa9pM!TP6N0VLhpaA^X((*x_v3hJkKf|J3FS
z_b28sLisR}SamNiAV=QLxp72jd%mr*EK!vZwoklB_?i;5ySMeO^-E+(Xr;*bUwX|p
zv4ag_1!s1~7$4%{+6wQMZh;RiA8F(;3}6zmX;hB(uAQV2KfQtu(gL>QVG+fJMbGHa
zv#orNB#>NGE7va<`QX_~F_Xcwe0QbFP7Ui1mh69j(IE65+0DnE!L>hZqy(|$#aK~6
z?ZTutA6An<8SYP!FP=1zLjQ1(tjqtZ38clinp-oZ4||4Ff9j0gb!{UQR8=AyD*rl9
zKj+?Xs65SLZd@F02mEhPY*MGz_gE)>3-<1wE{!h~v(AfORdr0%BORl;6X@C=>Hm_R
zS=kVZx>a}cUj_GR-7wO`<MF!uFBrQ|97>*3F@3gs7p_e)cdcu41QT3Gn}xJkI73c1
zC|N!?EE7ET+~ot`u1r)GQ`9%u>kjpBq}r7_ooOa*-&7QJ-Oi^KwjX+3S-rE7@WNAz
z{cw7VR;Ppmuq)=HCMi`DA{Q?G7VBsF;YWLbsX18rHks-)&ot9L3!F&?W~A!Zvfq1H
zwh)hX**2|ma)ka|ZaZEs-T10KO~W}&zKCY&%Xun_I#UZ&5c^xU$>6BH{Xw&zQQ=}{
z(*yOY=DRD?wVb%J&53|C3qtaZvs;&UYQ{WzM@l?{rwjkpJ{pBtROO)yE|g>aHjjNb
zbVyzNFaWT=<dfe2w^a9|m%ar$=(3DC^U!&+jM@KxEa$cL{{?8;3Ab|W*FlGp)PlKy
N?x`VFD^zSA|37*xrvCr{

literal 33191
zcmZs?cRZEvA3ttnuUljt;~W{uO4cd!93!%&jEszo%wzBExU)i$5JFT$h^%<WZB+J_
zl|x46Asi*Xcc0(y```EB;g56f`&`#+Jg?{Lb-k|F_0ZVx1`|CGJsBApldg`YDH$0B
zNJd8fj+PSshVPEkI2jownX$f^7W|Q%oScG!f|8Q*+_`g9R8-W|)HF0Sw6wHzbaeFe
z^b8CPjEsy-Oiawo%q%P{006MEvYtPGo{f!-ot^!{g$o=U9Gsk-TwGk-+}u1oJO~7W
zmzS51kB^_9UqC<riA17MC_zENix)2n2?+@c3yX+|h>D6{x^ziQOziUI%i`kV5)u-U
zl9EzVQqt1WSFT)<k&(H2^{T9_tel*jyu7@Ef`X!=A{vdpcJ12r>(`Z(l$4c~Ra8_|
zRaMp0)YR40F&GRMi`CH3(A3n_($dn_*1mD$hK`PouCDIQn>Y3J^z`-h4Gatn4GoQq
zjEs$qO-xKoO-*mzx@Bf&cKi12J9qAwo10r$SXf$GT3K1$y?fW%+S<m(#@5!>&d%=M
zy?gfd_6`mXj*gB_PEPml-*<L)c5!iWb#--fb9?aMfxElAhlhu!r>B>fm$$dKkB<)y
zhx7II_4D)d_xBG72nY-e3<?Sg4h{|p2?-4i4GRkk4-b!sh=`1gjEahij*gCriHVJk
zjf;zmkB?7CNWkOqiHV6I2tIuHFexc1IXO8cB_%aAH7zae(W6J{>FF668JU@xj~_pN
z^5jWYR#tX)_S2_Nb8>Q?J$sg$o12%Hm!F^i{Q2{Of`Y=r!lI(07cX8E7Z;b5l$4g1
zzI^$ztgNiOy!_RxSFc~ce)Hx{MMXtrW#!wqZ>y@Rs;jHty?a+vQ&U@8TUS^2{{8#<
z`uc{3h7TV;G&VLiH8nLiH@CF3w6?Z>{P?l0t*yPi{lEYI`}FBkM@L6zXD0+fU0q$>
z-Q7JsJ-xlXpFe;8^5x6duV4H6`uh9(2L=WP2M33ShK7fSM@B|QM@PrT#>U6TCnhE)
zCnu+-rlzN-zkT~QGcz+gJ3BWwH$OkWu(0s``}f7gMFN4aw6wImy!_+GkCl~`)z#It
zwYBy2^`Ad~ZftC9Zf<UEZEbIF@9gaC?(Pza#J#<}{r&x4zkVGY92_1V9vvN#NTlQA
z<KMr3|M~Oh<mBY<-@m7)r)Otpgm*R(@EV}=(XsR+BV#!K@1J~RF7kwo3`wS|i7^Y#
z-<+e#VV6gcecxI8cr+ij^YNk1j!p*i%^T{W#Ps>(OP2+IaAt^THPeDcQ(7iTu|=_|
z1AKf^XF{@{g=Cd<Ee^eQ?(Hi})cHO0b_rT|-sfaWmSB3REB4%{`50q*;e0HhA$F$}
zfQ*wah@F#p;-EDWJynh;UJ*IH5YOB(FXS|?)hIpZ#NPUn_-z&29&22z+u89`D2Vk9
z`x#~Y%5rN1?+bdhV0F3EdoNo<3P*ys;ISOwcS1<?HU?8~rLmq$ORLw~h|#VmpY-j7
zzHy%k1bqMBJ3ZK#1XsTVzGK!e$A{BGuL4_t<w<vR7mHn=*GjA?z0}71T*uz&O*{SB
zr`db1$**{mQz?}D3HoB!hpjnpJ-V6Yx!#eVu%PW~-sCi0ZV&!toaa1WkU9{=brUDJ
zuD&@=&cqvVaroK3&q%Z@KWpK~-;N_poVN!5c;DKqmIOCJuB-@o9<!ckuIm0Q0n@4)
zUz_+gMQ8j$QvUO0jONJX>D;sL=9GlCK9(UweepgM-gH*yI<U~6niLuTQ(`)YTLvr#
z9-OQwWu>;Oo)xnZVoJ@oqj-0uWzCZ(E3G*fd4D4eOBjHVqjRCVHha>H;E^mJwmw^P
zVTR#lW2Qgyl-y3V?O|C1;<%D&xG~c0ctxC)kGmsIK@>wcME&QMk+G4Dvpvqu+7O)8
zCN10sQ*v*Wn$3)H3I3Gbw9foGd~}Evs}g#<k<coc8hcpbFK=6in(qOg-jO+-3i|%A
z)P(ke3$Wv2<)(aCftlh4-dgiQp3kBbI{+|jug{GyVn^x;L4PE6O2y+UmSrS?Pad^h
z0!WwI`iHnDNuR}K1<l9aCb6Kmbi>DkXAyw|v93P|$f_@qrz_S|$*B3S>28kxZSBTM
z7j+p3^rYEfkEc>e>b%*$B~+|@Rr#$XtCHD%pxw0d-ylKN*(73Q_c4d2I42Epz)x&{
zkUMwV{DkmvbT+Lmi1Q{+gzDpE@3s9;^Kr&WOk%x&H?Kj^mBujOX3a$^Zo>R~jo22-
z7wmebL&+=uPwdiT7++ds@H07#U93!w@loAV>ClVs&~iaXfJfVnlQsu9bD@!*GSyu9
z+5MVz`F@|Lvm4ugrhfgFeCu-*kpO>pkQcfr)jX-PTjLgQ+eo9@p><3jq#<rSMkTNj
z?)e5Wka)!|@AuRp{+%1G1N>Jkn}Ga1C3kK(171i#M0~O#$hvqjyXa4ys1(}-9*IZ(
zrsaGNnD1!I4)9w&a{l|dnhC>*`Nb-A9W|hops;O;<K3DvDwuE68c7B3R{2l&G*@ij
zR+KunSalo~&s~&yfN$B3LJQ>CHkEO1$A--r*o75;2vUsLd_L$C8W!|*a{l<6ne=Qw
zgW{Zcgy&Dk8TW~@BV|-1X=0&M;p3swxAyU|@aLuO=G84DUE<pTPv<JCY<|@Y+-THO
zoj9fqI*7Gx{QBr!kgCGRQtp9&qm7+?Y}t0dB{1{bB4Dm({+O}+L(rtO^(e!t<K5h?
z22tpbJEz2S-7H#gdubA%v|>EGdC}URqrWGu<hWhn8@PfWU`_Brfw1vJA%tw%+Z#um
zeO6+X|G(m9k1mw~5TLp0{y)QglsYGKZn(ycw)l~O1|Z)7+mn_vb}igMAP<jho=iZh
zt^;R&j8}O!kX~wC-aHJb)6+UWscnsuZN%jDLO>12R7e7W4KMGr<ag^O>aBk-@Pug1
zjxi8kcqarD02gzmekdN$5tSyJZ^V*1qt*_CaknfR*CRH43IOTCMa9%Epocl|?LM81
z3H4-|E%4#YMVZ(B&pLTcXh90iu#MN~jTn-f?vvx3#@C$CCusyrg=)!$GU|{vOzBHx
z$O7;<Gvft~ZK_k@)J2TPAL{D@JGcFRsfFDVSnS@e<tRIB=nTPHk3Cy>oRb(_FNzgS
zR8lw~$LW8>R0Vi}=4IO5M#-GpJ}70>#FT?_>DiO2eRAApM^_8IdaIGic=e`%3aah+
zvz^+^MmY9FXjti48SM_lMnH!5tzRvy+Gnbn*BWD;pP%|NKgL(p*^u{eU}quC_$MXh
zN)4=H)M>`u9|(sE%Qaqqb<Zu-tq6=qJ(rr{e0162dswpk(TsnRO{s8W1+s!Xeo{Jo
zl%OO=VT~0b0bk8-y@x|<U0<6%{GhA3&flp-?D2oj0uCOrh%=JW80-E~ZB~TR7*+y4
zOdjKDm#!Xq?|(mtdQfW+fj!yd;z4tWq4*6Vc<K{BoE@cne62Jq>uqZge0sk`UH8Cb
zhg%|F6vA~j-flxR(24?g+ugi>G7vUMHD8tG5+W|!0@93{qt|!C&hG<-21!HPzQOCs
zlpePPfDH;fN!KRwIf9U}nPHFLRK2zEruNG`n}9FAP+PS6&VrMK@T~6Sar9J+0C)c(
zo0)_3-BqTVt&?kWn*pk*ZAIkag9vV`IyPDX;Hj6&h&MeC`kTe3L>&6Bt3~FmS9}td
z!_jZ|$!josA#_&CYFvB<*IV+~36CD{E=|3dy<7+cWpO<`jXw-7!W?#o$GbPts;N$H
z`cHcWq6GlO(}cDGylqJK-3yGME+crKI1<k}U8ahhh#~J@I=p#uT>z-l!a7hAo+z?C
zsXKxa^E>{SR1_Q&Nh!pJ5f~JCw+>LiQ3+wQQ7mT4w}681A*wI7hP(NF+c{+1RD~t6
z?eTDQMxpIf4T&c5^Z8g$<XuHEsN6pE<Q+cFm?c`-6s!>EG1E@_88l-1gUvyd%U(U4
zqdqw9?LJjaR|ToYG3M=17xOBm!}9=F+dFb?Yph|n0c3;h?B-RZ?Y5dNeT_m)<J<E;
z9~5r1pd!81Qyv-T?P}YHY^o4Z<wLe(6yGTw=@f=xQL8%`&Jk?QJXr3ovk@#$yz&N~
zJr^5{<QR}FCn~Kofm?TN%+$lnbOI?I9%i--`5o#$SF$G715r`K6J!Sk&Ku$C;q+x)
ziRCoK+ucgf20D(@>q;rGAMEcf&hFIQRz(^L7}MZO)EL43Wgt%)dbG9g9kaUW16$KJ
z+1W*_WHW++8H%TduC6`L;_j*W>R<=BW1^ObogMr1GYO?EpyfrfAziL>>BIl<LBEOV
zCs*LI<E<>;m<2dH_RkS|f^*O1hGRWW+K8mSO-s6aG7PKPqib9KSCF?WvN_2{>I~+5
zLnowZ9~(#Me9)jL){}NHQ)Oe4<PY>v6LGJJMFZ+%2Fy6m!3W(<za`fSIiqi%Cq?JT
znM88UuL1AFK8S@?x1lmiLR!M}VxGSXQ$G=N&uvfaVE>=Z4WfqzuD+W4k8x?Lt5h14
ze^E$m6ehoed5j<gSyofdo>5R!Kxc6Y`zX;f_A$nHpTgedR{3dIlYOLvR9>w~HH?Tt
z)QziCx9Xg`9%Cz8g(QVyMI=vesaHN`v6Vq~&%|8UMH)t*c6*D}5aVtd7ry;PUL&Q?
z6vsnp29%4$`f@qEVI<pR*d>?2owV+nrb!Ob{uDzJBF-u-=0Bv2>T(RwKvwR7lqp=a
zk2h~)qNhYuL+{(TjjbdwI=%$l0WfdL{({_bbj}Z+|F$08izYs+dv+R*=~Lv9r?N;)
zE<PI$m<QCbS}VLS#Un7vxio_hEW^6{u82YBykEv#aV?6SD!JoDS&V!B-iCY_ZCg#}
z?yy;;lJ$ZG3@Bn#&pLfo!Ap2@?q0s&-1`-tnoJguuEQ6z#EW==8jE}mYeQ@|^-Jmb
zlwMBvZUa~V78cxEo#<#km{U<YC_jmebQ^mU_FIPX-vYdo*w(!y;`k`2E<hFO9Le&%
zn>qXyS>W~tgTp*47s2Y}lk`7WtZGQ_ecaSeK#AI{Qxa%YzRR~h8g+?gcQ{eJg%~dI
zI<2N~ILN;;96x13fm1$x)=Fda8aaWFC%f?bg%mwVEMiRPmC6WLPobH9VRpzgMD|7x
zB^5v2?P%LO$Mx;oVz_$cwSX34W<pGhIomn#ZKv_i=9k7_4cBk5+{^nNZIP-z>$+KE
zqtlFv%$r6#CQXqyPW%Y_=Z9I144jZ<TtgxC1m6r3j{ld?!dF8x+S<`9Rb)dS6~p#C
zCae4<`8~Td7LYivb6>M!VN>85w0*mlJ2A11Y{pnsOw)1a4p!l!lS;z~>{pUxsW~>E
z(fAvw`7^?~`X{BJI)_fKtE4hhmYt(`g!(46Y--PJdY#O>*SK)?{Dv0iEzfQ}j(C~F
zt%`f<@voK>ZoXLY98Jx{@IGePW%|mbNo4?=_Ub^+=`?l3ZNQI7J2h2r@-=t%S5%<r
z!dGT9WBM9rkxTL+yl<vxoq!;lln-d(JEP0qR1iI#1emy_hfM1X`;f9(Wb^T7Z1ACv
zLFP;8pOx_zUy$lo2VaWM0Rmi;G1_V4#sZh$ou8$Fk`KrxFZ_t+<t}>3H$X?6mrZuR
z^Jj=QM?m|w^M~qKvYHFZI{EF(Wheckshfie0hm@(T3YBJ`m1>EEYSW5U|^!L{@dgA
z>rZTsA6dtP+zwN2sEGsGdIkGC%PjrA{z1!+S{>A#vn!hi>=sO|Vu@34CAfgpqjv!x
z4LUeAmnyv<(i0#dd7W(4wQJWPT{2-Y=$ija1i{8}A(gix<yP!fVpOsF9I*OKI-99h
zG<qcvEeaj%hNfncQpBYhrF$=H#!?U+VrPY0uqE#={;n~24qVJ!rU?FIVL{W_TKt&?
zO6h)9h<hDdab~e}!49Z9LIDVb$oz|cSp8o8O$SjpKr_8|lM_0axhxvnSMt63Dl);s
zlR@I9*Yc~l5It1(xtv{5e4{q=FuL17RLLUotv@FG8pe{%fgO)dO@O09V@-imfq{85
zYM%Ovp*k#dY4-t@d*&h@&3^Cln}Z;ZUJilWZ;SbWd8fPlf9QI=@)bFyii^oOaKYQ1
zt`M7c{`sOCH>7vZT2k#%L$!tDrc4VPfs!!Ey({meX8Kn&hU9M6F#kgj5yJCn<0^Pu
z#Bgk@jq##`P)p+7arzg(hl^c<-sCObW~@;#qtNN;XhH=_QV$_Ba5j&liWvzO;qtRj
zH}82ij8NgQC&5B`uQQu|0Cwb;v{3`aFf39lxWglBW=ZDk%)|9wAQM?x(JwXU_)22Y
z!+*F?u`uK7=KkdSDcNm$Jt{>bf9|dDW?ZH4b?isE&Of!V$)yP_tH*gvVIRFu+7sR$
zQq{<)^2PFYUvzI@!{;nnOz^MobA|A92-0bmu>aI=d#ap%iyCJgveddRj5fDsuKA*E
zk<-_U8JI2w7!c-7%L!pRNPc}PzRG*1AhAOydv;mv8dThETY==cuXrjiDLlzp`6Cli
zjFVBLduew+|2-6^9f($h%q!w>zvS9#bBWWI=CGji^;Da9{jM2}Eg?yEAnfeib92O*
zw{wLJ@Tl4_hxsXXx$`79!4K0{tETt#7l|x`jY9`%OA*Z6_r!*R*oFI%QXF{c#L|qM
zNck#3e#{uNnR!0yGddt$UC!{<H_I@ZK*xKW5OIZkIJklyWFtRc`0<44v%=#le+{`(
z5%T?6Ie~Y9m%YFMa$C%L$_CHUhaMi9CR9~y!gw!dc7Ja#eG6A2`n8|m6!UuWhpq)?
z_nwW|;23pfdd~&fQ<o$B9FQoF6j<873_e?R{`o|T^Rv7;1(78>Z9?(60X41?=bKYX
zablvwT(f6kI!-a31$@SEm1my^yWMUg=riOm`)Tm@MBaP=?)GQmg%IwDMSL(7G9zP&
z$uu_-JK2nypw0nDgi(uAd^&ajr`bhObxYI>o2f1|^t8~OIS*-2Ghy7{$|Iu~fIVSp
z&$FE=nr^w*V0Ps(10mTc<H&OYTh~mc2<iJ>4b;htTh6^06Fh;>(W1cZa{2nwCgbX<
z-!;o~LN`Om?d~jnxSTlkEc`QC{)xKW&BQxAKu>L9CWQm~lqeOM>*UBe=>=-bunL^b
z>*!3}9Hk;YF;Q+!8Og2g|2Z=UEF>J?CW=uzDV)}Dr#tczrVRV<$qR{1bfEk6kf(aV
zOwrVHG4IXZWik+oLYE`*>WA!VeWXBcxa9Y=`^4E<7uiNRH2<4N_A@d~N_Q{kevu-S
zbS>q|uPLE8aFC|2P}h?Vx=ki|3F7g1%Rt~dyfqR{x}6y6pr8mbI5(O<<ZEf`;`VAJ
z9?@M&eb&_QZ)JGqm|yr7YkBQ++c(&c#g4L9+O~`PceiWld~ASrWb1DY73*uAR$m)Y
z%&+w3gIF#=jIyI2bRTCe=ZeL+o){O_w{<g~2jjFkX_R!4&M0{wXH$$^lt|VVj|!(|
ziN9+VIZp16*rHrk!ag1GYj!}FhZlI3yU0OSW#q(UYz^35JylJ^R(d7+sVnnQ`&Fb1
zu%Eh4)cVLU;an5Tx&y@7yRxMIYo2OEW*U{>h<cT}|JD&k|DOreP75OBBo@ZLzS>34
zuVX7HhwJZDVS5dJdrG;E61UmFqu;#ZfaupzZ#f{tXrpI7m!)GG&rkNmE#(Dj@ODMs
zqz9!P!Ttx2Ju0=b5rhTj#-`{Ewyh*Jmqz>Ug`VOuXXlChX)5BS|459{?RUbnj#U{r
zM*$j@8Fq8Q3vQPZb+)q)*J#FZKl*qYzEKeqeE;XPL`!sezKSv0VBZXuDfe9Qi+67i
z<jb<udC|2X2ekJZdEskzcEUcBjO^#zRus5vRgXTH$~R0<;e@iPDT&D<&H+PC%_B_2
zR&Fnj@5vk!WL=`raH1qwS>GxRcH5Vb=hQD(hTTZf*j%W3h~qz})jRGb-SFB7ocn{#
z(momNW*YW*fB*Jrh{i0f{p-BaKG#Uu^PuO}#>m)J@FM*>MxR>DAL-S-y1g==m9B1$
zuf9J9+_115xrB_2K_u~dX-chcmBo00t9?GQy(Ko3Ta1Yn)OPfBZ_;i~FY*FWLgkAj
zi2Y34GP#v2aq5Gl*b|1w6*d#=LyC}dL=na5vZclyMa})?p*JUg@)P{NOq1D*ZIsIm
zPKxLudFb!nwR5~o2a#W-cg09NO*f+wM~OfS$(ETMS?bmg_;q#Oe3=Nv>c5PhLQCVP
ziY|XXq=$V;D(Cj15BVQt61k%IutC*1d;{b}8sDyjcDoaV%|J(&9Hi>0vu^sTidRrf
z`g|+d{Q~lB(iYyefY&NOYE$(U10i(DGqFq28?B$I?vX0K*Y`x98W$WX3EuBSK4gZ@
zk(B!-bocd;Qsl3UN^aY`+W?7oEln>W6`%_zg_DX=1#Ih9M&L?UV}v*a7X$o{=)b|{
zc*OOifv7~WAIQ|=m4X8YKX_>(b46o3ZWRKZ1&Ns?U{~tyR;=qhx5&8ZRnT2Gp#7Q1
zO=JV%vW^h&r~S#(xZYgtCS<luMeI00oRQZXkvI=}=W6BB6ZAIAg%U}ORZS<A+0U-!
zR|ROCboR-B%C*#&q(Et&C_bGa>1w_KreU26sZqMAQJWPY4*YocJV^D~MtXv2Mp1K(
zN(&btu9V@U@!qTVSyt30zLkN%^Nz(}@jNKmM1P(evTq7Nbx+1{2^5w;6L!1V!k5B8
zaF|W&xtTR>#R=Kp;n?S^`)bb#QRkfA*$_Is-#o!Y40_kbn7<k3JK#fdohsb6^-{{G
zBpfa8G)5RZEG^7PSnF$SPRHhZb5=?vx;DJppic0yd>{KZ5F^KuaOZilA~Zhx!Y%gi
zwheH17pNPmmjeIde~iRh<bD;7<Ep-#BWZob+8eBnvypTQIts+_KDe)28m*HQ5;`s)
zQ*{tO^<4Se(Q-IOLO)3Q=|2!C?H!$;IIThHJMcO@S~=%H?mJn5m>E-q2GUv0rzBlq
zzIgl7rEcH{l412p-52y?Y2k1%ou-DpEFY-h2cz&ML(NsPmP#VN5fon&2G%cVV5u~H
zKk7u8RQ>@G1WVb(5R*=xqDCa4SSe-OmieE)A7FidGjMWv*T7s=b5}O}n0)&4+lN&i
zw?sAT3_rao^Nz@_3dFYF9sPpgw`3FMgo4($&=*r5tIM$~3`b7|=_p8o!XcoW1)39j
zck^Ds)z2s?r;wD#5-=4IwZ!$5si4q6k#HmErWfea$GgIkAg_+YIISiw|Ecdhj?WqW
z(sO;Y2MtwSYcw^W$`Datj$tF9#5KgJpe*Zm9p^y<c~o1D{5#-lT*UbeCR|%YmvEep
zv`r^Fu(21Wm<_{;J-&wek}tS<0<#xlG4H+#(vjD}B~o#UR_qk3eDD4h>3}Vn=lQ;O
zXct>i3ZD!b-S+TP<p!Zr`HIl_lZu<j%$czFQwMnTqY8fMK2;A+2-A^wfVW5-RKUZ-
zZ}zU%-4#Q3GJUVB862P@y4&wqur=r4NiW|aJW91Wp$Ax1Be2kdyH7f!KFbedEfT+v
zpW;%C!bRkGfmpOR7(Pl%2VL3=3jL=5b5Q3Hf53m<VEwU2$@BmkHP5#9Klz_+`Y?k#
zbYVR|?>X`i`0W)bsE76Efd$q*UDUUp>)xO=?V|1dMSeQLw`)W^n!$8taq33}L$R^;
zH^mGaz<qC8yXHBZSKguRX=ZOk9<J8a;6Q62!Q=1|x(pR5v`LMSy-A)iXNAMgN;}OA
zc3k_>D!J?!KrIEPM!QSxuK~%#=~k2~GOFA>f3qSGlmB@u--fsvdx<w!2yowppCu)5
zwNNg_PCXjzZkA|9(ob&=N;LvX|FZoG`eS9(Hd}4YM5h#}7au+kASi-{KI>05BUN|)
zboyhK5y@hxk!2=aa{wZ}^ofVN3!0d-#AJeI5zvKS;^*ja!#V}rmI5OZd>AYn)zwSc
z5^prGMg|)4wi!&N9{Sh26L!B~+kW`h`w@1ZV;gKF^8KW?w6JZHmgv=!Z$|smS3~d{
zV)1!jJFbaBQhhFZ{T<iJSJK!!3`#0&*b<Eh0|+FQ%|<TTTG=+S&S&6bMTZOG%N?a7
zLy3UyFuY^?0Q*T^fC2CPU0TJwgnf5{&y7%@!cwIV+bttmc?3TbLz?E_1NVCZ1);Oy
zadW;!h;oykV>|QW>;|??as8*eBJc4z@R>^I!TAo!dy9g6$D}*S>A49lYrTQ+ka1OQ
z+l-3S5BxnN=lr*9VD(LZUf_`G$S{i6nVfI=-W#Y+dvw)?Z*f*>MPNbj=KaK$)r{2A
zCRoj#6m~aHo|g>6H?M2$d)X^axz1M=TawaSlbwm*66GC(F7yj(?i5?X12rc(6V10Q
z<!%t`Q{lP&vR`t>`^;<YvsVx$bU&|KnGZ`~e@W#g3^sDlxVgX|7|Sbcv}D69!+kpq
zd9AhTXfv38v5l^RnAl4k{qw-a{y)Fp&@=*Bu?pTG;aQa^B;FPCc#Fy8<jqIa9YP&W
zg}^rr6~oLPh;fq6!!IatY7$R!57?94z?S&AdrL{p$Rw>gQOz5}eg80bd3hgCVlUjv
zK-qHE!qHuPc|SflHt?O`XGAgSc3s(|6nNp-gK6E0Qxmt-vHpNl^RSkbbom*~ktPxb
z7I`1OW1W_gE+cEH9~y!L^6Y?39L5^C>=^XT2H5zA;3Q7*%`by=#N70L1fd73s<0Bv
zM!<dh@k9ug-+YUXD78?QDUkb#74|zkzBkzTI^;Y|hQ*|d&RJzLOeU=sM0_tN;^9YB
zQq@P)0Owwa2D?|g>Q$)Xe<m=OjLR7vIP3)mX{V?BxurnDBbEP8`DCR9p@z<D$fUyO
zHfsFr;QQ?cz8N#>8c<PlSboEPB^GVo3m0*j3sdLHd@-A-cvy3-UU;NO<UU(Cc)A{5
zBe0!rx={wpy9f=U^Q#W=VD+o>{{nQ^VA{v&@9L#Nb-$1t2lz&fVDn9<ekxpA`XHj%
z=k5V)P7c_d?1TRIfu@l$L~(H6aD1<`m=9)FYMB9cG7yFl+_$0+5-_?4r>UD#@s1kX
zkeuc%L#(^vRQiGxc=FI6X9FJpOZ+VVQCg7i-al&!l!-!m)@2XUli(zXl|ju2R9hNu
zv~YMz32QzGSMl9Y2t|w3i?cM?HlOBiR5hYP^5VP13%8YFn>r|>coDv$Xn^f*%P`*a
za-2m6>}B`<9vBIIz+MRU#QdNB(!S4r>L8O`ny-S^(sy7t&BJ66*V%$JWZO#^9alw)
z$a=dGG9o)ReM~`*-sj)dg_u&8*0B;lU~f6D!C1Vaf-3k2Cw4V`BQRVmAc9VlD4|cC
zu`#*BWHPHXVLwere3G7;`Wa4Q4qeI=AO1rY#GgLv;i;K);XyA|O~r0)<s*B24LKJ3
z((w?hccK^{Oyz%D9^5U6LN>ykMVlJc6e;jle7HxFfhbgmZMcCa2|RthUZQdGa3fzR
z&NXuUSFACcAmZ^174GV>l8PJ4KdZNB(AQ~QcZ)rXPJpd0*ec;J$;}Ivc3M%BDg3Et
zcev&a3efL-K@fycg`%2&T6CZ5P@A0GmcF;h>r2XPdwZ@pxXj&1^e4VoewB#--y}n9
zgzAopT2@yZzz!8v{-GJxMF6;!IpE-ytB0aD__f8Bcu$>CHqYul{$w8b>Zf#f9zozR
z0ND<Z{4SQh<;-MIdz1ciVW3Oa#t6JY-bkG`5A;{(+5&Ho&T=#8+88w(p!4&B8>Ptv
z;|Ns^qExVs`w72ff=(hnaMMO#0}>5*-FNtWx=19CLz}*aP@)lT$wDwuGUKVrs$nR$
zXbr}mB!B%C3kzRh_tEbE_5zfN$Mpup{!a_xx7!x*$&SpDlj7Z9?}!{WK>uJ`L-7HA
z>VoNG8Idmlcw~vDfebtH>hlz-$sO+3S=!jX8jt-4p=kuc`TOsYo5-}uI1=C2Of^q3
zWWnloe5+*4rGoWP><O#ghngo(U_O#lQtBUcP$;>L>9fZ!Yft#c%c`|lT?an1U-{9k
zgQB;oY0Ol`qX>iH5nS9cZ_#5!yhYl~^yPVgcusrS=~5dn!D4YP414nBqy^J*?>e_8
zQMreWaP3`}RNY-|6$)r1P)*evTp@|OhiYh_7%HO*<d&A@^0Kr3xxT+<(d~A-A|#cL
zSC^1ZoWcax1vVnpXMw`jqa@>}22X?&Ecy|G1y?ik8HF=6cnzA5f0$t2TXi8?%!MMz
zkf(HqKMs6hAZLDMBd`Sba!trhAd0X3?rJ_tJz{+%exi(AQRo>q&@^vBrlI6Le@V46
zI#`+7C^<<~Iq$_!@sl9!>vZ&*`*(~bU-eiyhGO*Z4Vczu!U-Kze_9%tZ$|sWelg!y
zJF{TjU*8T+VxC0}+FM_6O2@DpHL9Ra$qx<Evd;e_eMX0CEvOttO}~?@gN>9{Iw8O)
z2e+34H05=s`E-|Oi^&gB`_2PO8qLqpJu;L8V;j(;^l=Uzj5-kz`V_G9azi6R!7-1a
z_*x#Xa@->)`5(aNPgk;d2|jy4*Jo7`s~vR2qT{34?%Nn7@{~DHBj-#P>BZ48U5CEj
z!S|`|Rt@fID=K%XI{e)y5=Pieqc86^Ij)pgI~v_!bA%`HmWl?MK0@3-=J5sxeZ4Pq
zOzDVXVsdy3JJlx1cxoqdYykwJ)5g;q3_d^Sbl=qSx%x*G^Tnugm!tyV?H%GV%J#MO
zRgjP+kmol;N0gXKeQ6F0FoL)E+%~hxG|U@ye@T0ZCd+~%W9;_JxbPL3Z1<PnX%%OK
zr`LB=4r&7}f8ujuFtf`{HGF5bZ0|A*W65!jH)nHS`)Rm-`NscmYZUZeMs;@l01t`|
z8_c8Y<hQF8Arv=JxcA%z&5Hg=d)HgtBn8=_ahO)K3PmB=dq>a1cZN}skHmpp165x}
zM*|}mQ$%wGZ+t5$MT=R6b9t~T8r|1uo&Sk?=MX<dcXgoqnjzQI>AT<g8CIjGvZJa$
z$5OKvGJD>PyDJ;uc208odEf%jSwz<+ZZil2PgC*!lElQa!aP9Z(#1LPUSXJx(Bo`l
zjGHp`dP5;NIX=fm@P25EQcGw5y!(D;9#DLsQAt0(XUS;;^F_=-LpUuJ?q#wl<cK!+
zx>OX49)o51Tqe7VpJKh5F%c;G1DM&}slV-mD(k#x9OOyiIUOOq02C&ckpxTI6ruX=
z!dvp69U4fzKbdg%n3YurG%rAp`W_USz8go2XR2?im!Ce-P<|dL^IAv3hVJ{{bMrtK
z{?1i!{Dm`j#qDAkk5btzA22dp`Re2{((GSeM3ahfP3Sy>-XC<tKobUnwDza{7mCot
z?A^EWa<2eI7wCuH2d1Hn@#1D=cJ~KzwfL3BZ<Q}Zp_X8URXKy&T!W{7mYo-^OR866
zer@RuDGd|=K|h2$PqX9RX1QEKuGI7h9mbQ~n-zCIiQ7(JCUbvWeh*$k3h#{VuY!dH
z`uK16iPt2S!5yYy<zUYl@|TP18s2ckv_38hHrz{tL8tIvUyR*GFKS-y=I=f~d1USM
zxzzVwXhgh38H&IF^*<wi{IjiCB#WT^Mn{aavQDhMVD){4HO&28O)?*l=dXC0U*&ek
zFKxQ5$r3#YR3E=s+v0i8lqL#o-PW1&t%nN73WeibdHMIgjZ(opEd2K_X{xY1Tz*OR
zsk)PA+?NZFF%k(Ctv_>tJW-g73kEzFsD3Qg;SWDDh-#X;JrM|X1U)~ZPMyOo16Dq9
zj~p((uC)bL-^&d=T3J+XZ%b*v2G<WNnN~<e8`|eC5VAcQn~Wdj4V6&LjFA&X+{Oci
z-1V%t_B3Rr!FnRy4ZQR8t{b0v`>aI1J8x5m@kt=xx62u+%<J)u^*k|C%Xa?WY4i>^
z#VxeGK~dTKe0tsQ&ywWHCdzm5y$^QG3C&Dl95-ZU)vozaVgLG;G8uj?od-TsdBr{#
zz+51`$xzhXdqcHIWsUk(sR-TfF5Hmp4T)nHt;2_6^5=Wein))1ww~|`Ha&ThnaW7r
ziV6mZRD=oMM&GhkgWxtsSZEmv&W49WDHFj_QNQ9X7z9>j<-exQNKpJ5?C)sOHn>bJ
z>^)h&^<D+#^&0a{;Ctv{$W;*kH2l|tmp<V}py<_Nffp=tl}^^O;prm?!Bh0rMa$$n
z4Nujf58r-bDLV%<{f9)JDQcSjFuylx`-*Ns-jHLU<DsvrAUi$x58Ce`4+x_NhTyV9
zggsxK@Z9E$K&JOxUb&ytX2O*Tf7X4Ld@7^SbyvLHRrjrYP}^VPX6%ldsUq}h4fufS
zMZHdaS7*fR71AqFD3^pTd>Im0H@6b~&)O;JnOF7$ODr#gc@J1Y(!kQm`P-$y{!*AH
z=}DEZi-HH9jX>02yYwrr`}WW8*Tua-lDXM9`hLt=J<C2GMAR!&rS2veDY!n(518`D
zTIaHK?5V$abf}?%o<?h-8AQLxQ~*ip?aI-J*CAMFDN(_~Y|=wO$t6b=Do0@ksDj%Z
z7;!YS_u>3i*S)Jr%g6D%%B4o&XrMEqCTt-##-mt56tderuvU|+m>V<&$B8XpzRg~$
zd%z{iN9kARQ<c_xe{pKd+0P7692535!)?4z7;OEoo8*T*5pBP{VbWMlUXus&&$D5p
zApsuq3YSZIKbTktW9=>eD$H*)6H(7NHN)d-Rpq$Py%Q1j*IA?@R%mv+^@m(~zZ#7*
zjLt1(bXnBA=7_Qv3HO~6(Adnips!hUKj>q#^agJg#$N5={C-!r{WB`_%gdm=Mb;rz
zW`5{-W7`h$_mz;#!KVt0aGV`OFFQrD^@nmyHlHrj)$ph$^v!ofPmMIT@<Sp=KA25g
zR>OHO6(cI#+^tF6A{cf>BDAbp6r$`taK)Vb%{midd=sp3axEF>cKwQFA&<6wlNQBr
zBfYK+8B%aOi<#~Ptc7xyEv2gh)JQSN@Z2;M1F5~7%?L@}RH5yEQ!Mhj!xC*WaH)5G
z`jXZ?vMR=(CPQq3ReY|kBr&-x-p|S?!`BQKhAD!{I|lU6O|3I!m`O#%U4bH8&mm;4
z;Zd2W_;&RfzoSG>ug-f$U^m^37~v1Yn};O)no<8pWagr>UqdCGy{s%{>l$1o$|9|L
zskx^e8Mmg3Lh?qBz3wzV-PmAC493o~{3t@6s*4w~6unV|3Tum7LV6U=in57;s`~xc
z1zV6jlU^z*qj*GyYp|5-A{DDq%_LyOEQ{obmwN34>}$?R)b3cvx>OXGxpn?iNiV&G
z0TQ66BpP(iCzJwtZ6*R+31-B51&W%p)6S@02_%;vKwb{uO~$_%<n#{N7IQ>UdN00X
zyyiz^EAd4kK;7eBzIJ;AqJN#l@XPSfZ~uq?M@m6%lxWb`r>I5!r63gu;qa8FX?vld
zj$uzkf&SBB0kGf>mL&)aUwk3q<=@9fP<=(4Fma*ZU452mycNaY?_BxEE>F^k+x&@3
zd6<TTy>)ZHBvtUn*3bxppx3GLYa92a_5iIHe`vX(RujeK$k@z)3Vhod`@c1B8_#RA
z*z3>rWDt!DegTSVS`Uw$ye%IdC-wO0Aq~47?|yuc4+@d~W>ISehG;<b`JdZ_0W~^)
zC?QxnQX_27D84P|9{jr!3GozE;B_j;TcH|4HedG|=o3@we!hPz2e*wWGAK>TAd6{+
z0xZK{&BXM-{vO|0E0$Yu9B|cYJ})(j7~jbGMtw10{3}<&@D1ZR1OYLf%0e1w4;tjY
z5d*5VqBs<hZNt>N1G6y`wgArN&+O-|6w0OMy`ewS+)y|*M0qJ>;Q{`G4~O9sS->|i
zq1~B%;n~8aOVr~lp-{3?BB~+0g&gZ*>}4SoM^dmlw{InS91++_Ckj{KzEEA+;i)Q$
zqx=*H?j#jH()>eT^?7JVN`F7zVxh6%ZYZKZVp}dnFQSjlL5q!`nR7H9`m67XG${Vl
z|4rdcG#AwzBjKPG)i!$GKzi>w_qVWza0`qvUs7fDX$m=U#{WEc;Kw6y#nx3M(Sota
z9HRPfXG`jRO*z|z&PB1q6$3Bt{L5av)Rf97LV?9`i5G_w_{Qj!RLm5oZ(v-M^PxHw
z2N|!@I^A$%8jJhKCyX4~F22iGTIRC3yfW#CWH*2Jg!P-8GiHzinNn}X&@P=d5WkOp
z$3RbQpI#gmO&H3z53+|zshsTQM@RKdU+V6c`*@wuhGlP+hjMftYfP*U<RO67{|05I
z+7c7-ZG)@!QI}I~+)IO2)OL*3yHLA7FN3kI{qoIqpG2Mm8$W814bhj6V1?bSkAAg1
z>@~Z7RP+{Rgu3ZgF}@q0((|N0y=%984DSIMH0P{?#?gY4>FS$IP0gKAb)m(vXL^iL
zw!86Q_4tg`MW(~O$OPL=_`Jkc(L-1*Xa_!*mZeImvu{oBZhXMRqq(HaQKeZNr#1X>
z=ICCFDAfGUi_kXX8aY{aMRf@W2mbyTg~z%AM|kNFjmmFwEcmGxG`|TOe)E9cPV~y5
zt-c6kH8I9Uc-vvj={1ktfhrzYa~r`qlZ%BHgM&DUGeo10KE3%v{W@jv1cP9~w8!b#
zK*^%PQuF;3c85=wYS+)4;7xNCzYnoK=GFH_diyh7RGMp4Qc6tj4gxNytN;AEHVAh$
zM$Try)6D5I(PNZr*9#n6hIJGwv>7#^2&Zc8Vni@m(<@HqU~Hx1iPiR;b~(#g#%|AX
zMggF^Ki%P3mLB5B^XSXxj%%np4|I_O>kkV<HLRz&=akQ=pACkGjM|M3k28Z$eY*DE
z-gWiNV<Wv5g+d$In2y8C6W~sO9IEq15d~gWZ}MP2ycj!ivoXrH#TzQKNhMhH#y8ev
zpL5{sonw<)Xiv^PH6=zY)3F+9zu%|y8@2i11k`bDSQWmv{mMX4qjP7EIp;BI@+lUq
z*srd6dQwR{V-_DQltDL~J_!Z8*5sdlBCnBCUwP{M%knvo9Zey6jS6?o%r%cwL-~(N
zB%C;V1ud~z?)?&TB$j8v{h$<bulyIL##&S1e#dkCDLvtJo?bVR&Um3wF3qt?*y6?y
zK4r4HPMgFx@IS#ycwkem%x@=GdbU#%{L;e&tS=^4|6ONdeMEU0#6AOCSW}tBsq|wO
z4b)CVl0S@PVf~QNR)ue*xultCB~hZs48HWaUBJ4{K79fzLbY4_b1r$ISY@l455$Mo
zA4Gl}#0QfX;pYB$+@gahW`p)cH&}P#bLcX+t*Y?gD$(-44qX(-Lx((OV(b*_-t3#M
z4x45C&iUHK7qil?n7%4XqIC56*O?GA$Z7?!tgA~rbTM&U6aw_rdN7;4&Vv5xykrqg
zETe3MC^A15cZ|d<238eg4M<i=axBimjM5=&w#bHw^{Ak+l3a#&@In<nc)pR|GgFWR
z5MPmJJOBmUuo5xus{zRD&1^eLhp099_|h&s<`sDjntdcDSmlDy67sO_A9);HK%RCs
z&s{?joDp4e$OcE3cW7Od?JDw{!JFrRpYNhzOyoB&@cwUp7FEsJC+m0~f>-pQKa!)f
ze*Zd<d@vF-#X`NNesUM#z{&^aVOSglBET+rWW#V`@F*TVi*WGb16Rqu9B6$?L~}d}
z?6BS2@9?<AN{})DwQyU?3`ZKw90PPUI)xNGz@H9vKCnAvz?a#9MZp|oa4>;~+zp=M
z0dSLJ<_E~^q3EeEZBMZ0!QkvHB_6<z<7TK47)4>W>7v2gzfznM0v5Wd&%UOsp<Pz0
zppy9%otPN>g`wD_wQ?xVLZ8ino)2uSJH<$XZ*M+O_jtiXe0GTmWQ(V)WOR5*hP74;
z4_>%`<YHtkQaK)trvC271S<I`ARN91tz}X=pt$1XkW!R6BN$2t0WL_x>DI^V$bwE;
zzzrkug02QWr?kbtbGHH#pAe9>4S$)z{@tk~E8qwv2_6T)mkNyQqtqXP`)AkR^Y{OL
zZ1)mAfB(;(37nBfam3f;0t+`_g`z)Q2U6`0!a@PPpX#&V%tbf#O**kNj1_Prs{zG)
z5Id!r?)NaF-wXT`fK223`9>La%6#STJtV3ioboL?EpT&n(oLQBf(1zR9#yr747AdG
zdJ;#Xp)04Xk>lLvMO=iKh+%x7N_?g*wM=UYHKZNO%MBU*Lhc3JR*a5RW(Hlz1B=zM
z`2pPj>r&5BZVJaJGU1&TrUnDqcY7PYP1$U_EhX?D`*s>{Pse!hPuo&id{Oa~ox5nq
z85|RK7{K$$V*z2_KRcRM9lR{gDBS3xCPM6O=pemmDp%<aUEUx|XmaUc3wDxKZNUj|
zl}Fg^^_sKn`3^>sKl11vT-MlLk4Ety<0;PWpOUYJyZm9k$9MFr3&Xz{@85|a9DME;
zj>c)uy1Y>wotrVK@;}Jl^%s%@lafjYRi!nbZo@U>2v=U#lI>c9yHYRpCLx~?du7gT
z9cK7kBYlf6y58hys_eNf{KT97IJ6Uxnyq$QA5oE+<&I`)^igrH^HGcs?<$hu+#W2I
znAJkf(2(6I-mX(wS9>U>1uiyYWu#`+=+;IeWb%pAZw2<F6A;&95|Y{X<fCJ)8<W+a
z@>>J75sW^_pX4)-Ssm9*OB;ihifB2d&aW{mH||Cc9wPqpTKbl#%BK?RiQ4AVRa`5Z
z2ack^p%<yH+nog$)BbzRQX99si1oNt<9Z4li4I-L-TzKJuxJ#hgZB@`As-EgXBqHo
z?ET$|O<LuyQusvs@yoeU-;-8O56^vdXp}O}XE>7H05>n?ejxbpX!%9FkT{UYRqGDA
zyH)&bV{3Uc`L<cP>uGA=n|=j8+1P~rYS)6liNp}8oO)A>*W|cHen>JG@LOcP@*e3m
z6o)pWA-F|*@<Ih3Z_GI%`dlDtEN03IL4eoiLHr-pt6(Brz@Ln94G#>Xq#Je_i6x~l
zL;Zr#9liC;1wd^mW{M4uTQn?x0aI%^S>PI`vXqYaiW#EEdZsfH29-4@8WzZFnAul4
zY*xyW&eukyz?Wu_X{?8a)VL^*Xi}yEBz=HvtE3!8-<U0>tf6Twr6igwqpCTeFR(E5
z63`MaWN!<&yJ?(QvpDGTf#the!4lDE{b-UFrr|X$?$&=OuN9=~5eP(mTD@ikRKT1?
zCeZO7nLQRCjHPj4S~P5AB-rzTB`?@Hw?&5deAMM4p~7xvkSPG^H5Q-4#v)Tng=1jJ
z=-=3kp5nuAs4G8A=9C1-8+}%%@5y_CJ^sjSPULCN@2F-}!B~ui4T8X_tl9GTA)5m?
zAK3ogU4WV3!y)C2Vjhok4Og^NQP!MIouXfjCB0yVysltr2y`bE6VVpzgT*COL^KU9
zTBZ6en}e7E*uI95ctc0jV<zZGgC#|kBJeQ^=om+Zqm>(pX~ZB1ng-xqN?IaUTl!e6
zg$5hpAsO}r_-;mq+WushzybMQ2aE`}1}w;kS}NcMc^k6?<ar%v>CoYWi#Veql3=|Y
zEODHL6@2c&2ad-p;msQ@c@sRscFkYF1xaJKpLE1?QtM01P=Fvb;XMkUn48O?ye4C&
zsI=GBPw1ZBQGjggEa3LZa(UG0c?+<e2w2WUwxc2|C~@b^#$tkb7c3r80W=zpUP0{=
z|0$&20rhmd3V9q{$O+!B*E1;zCdyzs2Q&ful1T!R<%OQY>H1_m={dh<+sH4Dq4s8Y
zTm+JNB(9MoYTODCb;HiSrop{uh6IQ~B~&%#%n;Kz;CREd<*Xb*nEnrYVoO8tc`-7^
zR*Z{`C`croXW*;@Xnta>OXy^1^jA9mKFNanlV<xdo2}_m*O02B@9PfYNtA7Q6;!x#
z5$M)+;42Iz8x{vT1JFDe3>0M(g)UzQqVOI;XmBZ(<{#T15g+QRqatBAJ|1d?tBkDS
zMuwCQtkywaJEluSnotE3aTa&j2!$~TAQ~)bMYai<<=#Z{OgoT6<e`AMyGDL79nOdh
zE{3`n)g$@G;<(P#m<j4(vG?1NUgYR<nwnB(=n)bQTLG}6GPu9QvDCElLq`<ML`Oc5
z&pw|S^2T))azGlPz^I@K_yhLhWQ@h_=-{lpE|cCvR%kR9t^Z06&Q{nT6i<=W2*~{i
zJgPYP7kw{>U%RK_Q5P@OgftwFZ#1WIpm`BeA_{r$j%y)NWh_^q_Fu^CkvJ|pHU|TE
zc0%*GC^w-E#uZQD*fbC}=mhtis*T3wn6Np77=TfJz?PxNDOSIT3irx5EDEt!%;3Ps
zjzHdl9<RSbwN)_Ta+#qIy5dyw&>?oeTlO^^Pzg^PV>KBFtJi@^4QDmvx4&w0z}YYi
zLly_D0T=+^!y^k|Wx;X%bDiu@%!xo%>j?TrUz%vuXl$@L4WSNOadq$_!9LvPjfos*
zcTk-_(s?x2;*RU?R*2>-<8Gnhu(j4JroH&ayKDqNLf=MCAg0{d11;zb!^kv#Eku(Y
zb()_5<KW~;+|+$?8((z~QT1QSf{|&19q-dXm;g!SCV`r*MM0Wlu_RU@0?PNV8oYT~
z=VfhA{0+vTRV{8n`oW<xhnaS(j(JIdIT`P892#DL7!`2wI-nM3t3%^pPs(I)d313#
zHYYg|-CHm@L03~O0wrGux-Jsrz4nw*GdI#L;Fd>{F&v}GlbHVLYR&HhCi){AF7WIJ
zW0k==H_XXR7KZ|LtX$HHUMNrj=LV@RmnNgp7Osnn@Fl8QO|=26#~%(GUrhdMIwIUX
z<M|88|DS;*1)WS(G|F}&4sF5WU}^v|`6DBf^X$6lK}mn4RA-Q1J#t_qp47w-nZr+F
zy(NCC>z^g^LnFC>xEr>$L&NF0zBQ1zhS4vh!le%()|sE(>|zvk{Y|Z?@Eyr~f6+l0
z3PLWz1UG#S?73R!Q=BhcDqO+i;Qm5paQcb=XL6iCTcM*ycqH||>OJ{7CVY+s47}l_
z2vzVxtn)zPPmF#QUCj%As0=nJDqF*vJ|dkTn$88<zas<dU4C<*8b;zsGF93z-Egj}
zThoZwBePj~tUe=cC*#n40$|_=C7JxGsxNGWNvAu9FwBKNv@9&&huyKz9cM|yPGD65
z9|s_}KV@lO2XF!ud|(9yqYRcsa3SdMx?)H{x~F34cT^gjs$*TXzS9xK!-bNl0NFK+
zb!?*7cr5w`ivymP50p$$V2A7GOYQ+oU;^_+70^8xjDwp6M&r3&Q`XcD1>K>LdC38V
zcJ$w+(^M7@^N60hNkh0xN|H(kQoS`O9T-%=mfdIT8XE6u<oq$#(&~)mdWK*~<Zyhj
z1`7e6p#_=64+XabAR~31??2K268C_akGfB-fz<~X#=LT>nx-Lq2OEJo91G}xd2qmm
z^-(0B9g9PovN$}D1OpO@VL_8<qnC^44D8eqkOpr@N0%wsDUTW$jRo7WlHe%ZT-bR+
zNk@d^yI?rF<zg*@AjvPd;Y5gSgz>=x8!Cet>LjK<mO>D|ev0c2^7x~`GBZYpE9HPH
zDWIii7HAsHlM8U}EHc}C7~goI{Q()O>?2H7mqy})^WxAW2!faj=yrfKgj3efjdyX>
zggjEIAp;?t5A=QrG(Bt60u89Z_VOY)6S%uu@eQ!cfiD+P;UvANi^yyIW#GKPOAgCx
z2xqOX%z6BY$TWsan&A?|v1kn%f{3z;KY--85elT%U6BOKn|w}|L*TBN^t43u1RKGu
zlyW%VLxKd1nm;I^A>KCt8Q}P?VFl#TeyD;A-SOse6gB5Y;?Q>(9E5Gaby!Lb9K8dl
z0VqpBUn2?Mx3yp)PzM_@IDC=@I0yKLjF~g?)@R6Tgy7%U!pBC}fgmh3Y-l*67G&V;
z1j2q|JOZC4k~R_kpQ&*6_fP|HJZ7G!rH52?Xb9TYx#!M1Xsdvu#~)i<40N~g;9#Z)
zxQK}(Vei6$@N56SQXZT+tpsN=1PNYbSQUUCbns~yf$m>znCPL1|9Oakph+(1KC=@`
zYT<|aV&a1@HK_9zQ4v#JGB_ZKE=u~EB7V4FSz^J2yUK!lA__H`SOI>#@QSDG5(A|h
z7%oAk$^OVp%1UMkyYDCinf}S(hhB#QL3@j&&-?KnO5!N;;<*`krSd1~0i*rgJjP(Z
z;&1Z*sgC;^AT=xSc|DWYq0mht-wx1iB*XSK8ZkI@+4Qo2OYn`>XaYNAXM#jItKum_
zG$nS0N3IFBJ^FJ7r{G;aGE_v$**KdeUX%F$)7X2*Q~iej<BpLW>t%1wLH0;wk8?P(
zLiUc3k&$DUb?jFvIx2f)M6&nRaLkaAO>&ZwkO-OIm-pxQef++U?_b|PJRXU2-}iOh
z_jO&*Yx-d$xAp2#L(=LT7xc(tTP)Q#QT7YG@d#yP+A;Y*n8^|NifU{fWbZ?isKv*S
z;JKZyCaoSjB6ACs-a`9kBwF867#Gdr0n{F@0)&jj7S4U}lcG*|A!2GSjwnUg3aB;Y
z(*d0oeL(4nmx?soU2Fv7ZQ1}jq7$nW0NlNn8hBz`EbhQK{Zi?Us(ryn^^bb#RgTPc
zk{$waUY$582iES}hiUX7f}K_^Q@vW`^9?(JtedZ&NN>T^7aptxifYm{cc@2FHF;J)
zuw5PzbP5YW_s={nc>KR>0h)-S$Bf<+Fqa?WEHd`n4&g0tzO-$1+Bq<3)EOqyx`c&j
z_gnc=^KCk|VAy&(k6xGx=HFd$O6e>=RnOnxJl5Xd)FjW}C@Osa`SC{I`v+k<;}%{M
zDs4a|vyTp8@8MeNE*N@+jGB!)1>BP(`6ECCIt2>VtPOX08Jexokvt65P<?hNror?J
zqANL!+zT}n$PF<V;rsS$7Z$M5aTO(S8M71QN=+mZGPb;!A0pEfamP4CTQ>d(4LG>R
zj&yFx$~L)J_&gA~m!Wm;I^NXUWk*SzG5jGT9`rF1@j2cT9`U+HN_?Ee{DB{)yPATO
zGr|<aXvUSv9}^E^ty=($%0J55Ept{Wa{T556YX~Tu7rIBP9y;e2SiIjv4w|7aQ-#_
z8^q09%fu^$X=Q$1=fvr?aV8S%D=Ccf9a9wIbNG8i$tf)bqjN8l$P-{EGByxV_1?u$
zT@Dkr`!^%LUzuMVF)n!=XYE3r)DA@Tzmum$7aOY~S<jdu6X8g|4fK(-pEKrSVXh{C
zUjjaCpd(Q^zd=CbodbgpSWV-)oJS6>%=IPA1jsuiu>b?2-(4=q{Ctpb;H7;2T0UNS
zX@CUj>G6>TDJ+EK>amwI7i*W8q1dp%IU#xHgrp!ryY1saV+n$GE`(lbgdr(d{l~h7
z3TLB##ODJfu4xt*2~m&}D|O<nPqdiR5hcp8EsG~ID)YPBi5?$y80rD2XwM(qtww_O
za5P~6Ry%_(Cd2aN1&t*rPEkA0rr<Sz4U?zGF)eSHBWcirqT~$1XNWFtYw+oX8ov|<
zf|)0jQlM+yt3~?n%q1XfMc)F@^>Ii^p#|BN>1k#7)3l2DY!(s@eg=SvGoB0%NjJgP
zH($GRjWyG`fwO6HZDCAU$z3JPp(iCd(MKBOv($mSBOh%U3asZ=fAwN!YcWjbn@wa$
ziknOJA*IZTw8#N)Wb6iIn20b-HlhJKMwhLKjo3AQu#HC;g_it&oF&~JFb|Ebkk@qW
z%;jYR;{|>e*(oiyZZOA)#Cm)j{VGf-Onkxz^xZ>C)U*8~8{^GF0)X(o`*H(vwbh=_
zLGs6lPbol{A;Hr$v-<$uvGiD@p5p-hj-Sx~J3-0wg!reLD{hB-)WpFO5poll%%e^7
zFUZ19rjn?yBSS9}6AsKQ&%M4|?##MivAiysbJdm!fJmKtk(j{FVGV}xvN%urh$J9<
z1qO2GjW_|b<K$!90k;a!CrEh1IPr3Z3g4Aaf^L99WOY7e(WF$^yBx;{x<Gr#*5@ld
z6PpXcN%ufvOFlYA(TNet88R5$_W@njOfMF?$L~em!wbeEE|3y`lW0L4fa=`z;4$P_
zeGS&l0))EAm;Z?Kh__ymU7;jw!xB7F7#JnMYhw$Y&I%#EDimChQ)JWP5#LhC0ZH9A
z0K*J`2d3X-L)@g)<OuooHHO+Cb=wP%xPHSL3C*E<=XZEmdrca1MLv8;tD-KKK9M*k
z5}hzJrGI|VMnJF8wIW{kXCEZ9kKbx`aA^S=QU<|!>JgzJpWXj3;OSZPF>dA_(gBVx
z{`l+N9a8eKJR|e|rdFXO3#h9ujM$H_zS9C->nbvdZ1e&<ap}Pi(;-c6t@Jz+(Cnzr
zZw?4!)4!=O>;zGuRsc|Ty@xVG1TQ`ZP=J&M|2NnlCDF*&V+h)8K0Y`-D9RH^0#a_f
zW%phJKV%m<p(Ku#AX?+zYOY);r(z{6>|08pU!Ff1Z_wSe*6$QOv`q#_?=m@m;|W(}
zKGR?|8f_@$p1pVYi4Tm;&Jk5FEKq6AC8fkhS}{KcT9rOH1H|VLt!or`nt_v1V5^4*
z>`Q(@_H+ayLXXV3PGjsTHQtDY1p!GPEXKo_<&x{e2!Pu1{z<BebbcYNb2<ftUQM`h
zQ%wA=CQM$)RFwoIc(|YShRdPhFY7s#qCk2h%9xH|*my%uGiNF>J`B7@Pp@8?I|{ui
zdfM$!p(am>xC8pI`bbS?>GzGu<Pr$=#M2d_G?|6#?z@7|gA3VXS}H^K5<W;sv$fIj
zMV_Aidhj7`BZhX{Tc34A1DuV*M54Im@bT2-1UxLEzip9%oFA%8bObcSQuz3&oMT_^
zO=G<uZ~J+uW&IV-m-Pd_2WcU}=;(twUcwP9A>)f9tJEm)uOaGl;fY5Nw!uZ?jU}1$
zcz_7G3<9`I!C9!LmGVR~SnK%DSL&}(;Ts_hQRS^@a=qy)?UwPE<?knog&!x1@6I-J
ztd0MDsM)mkaoNv%B3V}L{Y&-1#3LmNhc=;v>`B-g$XtHMv<2GGq;F8GFRaD!hV5Tt
zLfS9YN-VG-9g{!)^ZOnhsX8rvarW~aaJn{JE8$+8cufj@2<-H4XUK=2I=}y2qyK%;
zDqJ{!LoCM~AMd?o#rFR3x>kA>xIB1wrCwX`;BkEWi2J;$<wb>B;raFt3{{wVz3*P1
zn<HDHfS^v=-9<ssE5s6Uhb}2q!~27wE&BNp`FJVqn~ZhCzaCB38s=wzW00DJN|qix
z`!BNh%Sf~3X7%Vxapx9q$DQR7q1}u7*JP3QZ?F;zSD`~o|MNQ|ISJ*54xJ_^&+A(Q
zS@BW>s~$fJjpc5K3vFf!gibpX$BPQVi23$%%`Rg5KR<@w^bXAXzCN_bg@Z2MFsoaT
zy?2XQTKnwK^3HwUmKA~m!hc<t37rCKS&eH@tzvm3<^w$sG@wKMmr|j%Tz-GQ{Bzb}
zhwl@YbjBo-<9^x5Jyl5v9vd1a{;(Q5HZoYz8S^&I`!15D#|76&a<lDHN^-kmmvC)N
zVuWwSM(%FKV+dp=wLjt0-q~fdR(zAR`bPMXBM=nzuHyOoIYoMzc@y=8H&j1T^MRf@
zgb!GB9^8H>ePbc9rJXdAh8?pc3caFWok-NN8U;05A~G3%88fvZg#cPjo@xRc%;E?-
zI`?H~B*9M&q=$T|Wev!{$7QUjLW5o|66T{5{*$IKeuTelM4uqW_qE6ygp^ObGa06h
zJsmj`niv0&8fYIr&$AG9`)j#~jqEYHy$LDzK$hSl03&`+Q?R+}D#GoA3Q}y!?wMr#
zo))Y-K+nHe2yWh>q+ldgB^_+bBp&Gavu<MJtiKpA$GmzD$-WUib2xa`Q)&Ro+j<`)
zZ^Z`D(U8pOA+LFALK&D;&Fi}g)ND{&(<yaN1bDak+So;G11~`ZI`YR+0oVWU$hSc0
zgnB<2I6j$yG3Lj0ECs)UmNA5hcsZQ^r97b{<Uao`Fdyi4msLs%^-Db+iXSCu`D<Ju
zNN5zb(BmJ`@<AVxq1yiFDv=|S*idj$p0Ius&{LVuNECr?hzhIauuBegx-zLH0BaJ-
zk(xtiEJ^C%52{$*y_8Ntg?Ik;F)hWk<1r$iUqLV+Mv8*GaB;&3HkMbQ)4z)Ae=Hyz
zubZss1|hAq<JeFnuS1dv=(3CWOpD&VZY+k8H**`(ERLOo(o5g=5ImrY*jJsVwThb(
z_Bjb6$#E6b`0N6mxQu`7kivZeV0PnBzhn}n;t}KSDKsWIO_#{W=FH=C_qGa<c?N^;
zdoF>+xjtVIsd{jT!x3t`2U;tujZg%ySjHGanUk#lU@}A#<mwS8q+yGL9&$)oY6@1y
z=9z=LUrkWlBTW~{G41oY+z<~_o(#kATsl(ZV@9=5E%&-D8_`+~+;`?6)bR=Mfb->F
zbecq4*)Mm;Bj!XWu<8pvm)n?CkmKjCF8AE<X!);et_WSDx=5gsu)oM}m_mgyUG~-b
zrIZc9u{pC59chBXJd&3<+Cl{p@@*iYf7pvFMmJww?A$bdK~#mo)sle$U1k|Ne1jOZ
zJOPTQ7E?n>kdCEuYdQ=)fgL7(j`40N!ZBybqLF||WtIQ~DM?AUyM!Pr98d^<0k5Rx
z!C%^uD8Z#i9OtBP7{7q@`TPwJt?&m@qKNT#cc9a*$;5^Z<oT-$A(N@m#iG3k|9Ztk
zN)v7_^%6wawgem<eg%<wGvPdL`XiOOP;H_aVvvkJ3X$x-Ok~7hPhS8SuL(sO?2HR7
z)+eHv!Xzhg;n;1+FQ%k&^0&)xWa>R9O{fp1SUT+jL3N3r2Xt+El$iJ*3DjsB1jf|v
zB}}GUk2wOkOU_CpGlaq<?Kl)vE=)s?@h^Wz!xrOFN}|QgwKF3d0m*Zb0UQkcR19F;
z?|Xm?63JL9B5_ZI6PZj8o&EqAn4iRnB-7sg=*0zq>3mQzdxInroGLK|JKe`$LhG&;
zJ<%%8*Sc89>0EXK`M!yh+vkVWkLjZ&DzZw^X+hV<P*-`6^B)<0Vdn~T!*#{f5SD~|
zNa-Gcu&`9nN5{YB=$Sw*gz+7zgFFn*@IG6xIOM06)}uc$x<aC$Jh;%Gg#?4)bcAUY
zga+a>jU46>KY)bdhFU;tF%|yPk_O}-D38s+OuRpP7%Nrd6rVxQT3v#WjTdxy!@8!Q
zP6*nLRN$=|b?-ABeYwsS@68lOwcKW0VvyeBrx*FH`y_2tVBwkXf4<+A^?$zKB7JyE
z0B}nZ%KWT0_ej5!=TRLOu<zYE@MHTzy4X&w*`zJ`+n~l!hrZaris09({<T~Ct<qB1
zzNO1?zkEci!hE~9X4mo2KR-OxNDU85U1bH^!1wG=-7Z^6Ta527X`uge`h0vWZf&`#
z{%@zxY>lw2p4!WR-+dJNLx8v=`y=%$C^C{(5QbSPfjxm7J@W6MA~GQvBp#!W2_l;E
zivrFA#!p5Vum-fp?8k*l!CKe%#TgOmHEEgvA1D7B);HlPTCgZPdpU**pIt8@g$}P2
zU_>O}UM^&T(D6S%2n}~Fkw9vo6w=7Qb>egn*N<TM&1Yo%nZ<}Xguhlt0%A^&iT~ej
z-VHGL?4vad8(?_M?hdv88j}?&>P4$Pboqr7X{zshNsYaD?q?9aInFgokOKJ_98$XD
zDUNj85Sq;@WPlc#G4MCj)<bq(&V}?FzXC1pqRfUUR7)%jhXf(%$*?i#nh_xjNx8wP
zrmNY=!)4YW&2wDS@`9TB{PE7v%6w;FCI`#&(W)3<a2j$ud+0zH$lf7RqXl?*`MJ8p
zjnE6`)Ed1UVDX>z0_0pkni3+_cE%%Ckv|wMSbuv4*f`A}qc+L;wW}8^Vq^-$fKR0`
z*rrFqbo>|kZl-1z2wy|a#ol}fOypdPjuY8aZK<T*)^1}i0$LxK#CfL(6GYAsdzshL
zNlCH&0-E7}l~&j=Z6b;6r`D+JL6DPzwx7w2r$#qCy|`_Akspr6)WZ0){naf|=V+Fc
zu+PcQ%daq1;|M5B^`>Ai@ULw%#fZfBCu}A006b4~)yII1vO;H8q6h1UH0Ual>k;d*
ze?|kXH%plcSqiO`D4=KdiYhWy8VWG}TzW!>`Li|)==4fK>VSs!`~r@{nKhfB^8Q$1
zcp;Ro;UYS<F{dL_Q^sk8V|C$dvi9~+!=so3oCXs%y3QUKDoj46%lGMB3OezJqK<05
z>%+gMkTLzpu<=i1<vdsmd0lCm+*gx)km6ZeC~{gh0QCMOsC1s^f;8BV#{X12@vn?B
zf1a`KI)c<6L!(GAoPt{MzmEq}D-pIl6_p5CfqF1}@XVv<5M#S^x(5)c+2(F7FEq((
zdfA?6{ZfdRqPRd*RQ7yI>J0PJgVo-<-xlBDF%mcq7minyp?v-Vmq9qJhZ9Surps^#
zMU4OgT<-Kxv>EGo#UvRNLp#<s$y)>@K988QHAM}%ONt`o4RxBIpgX9YSlZwqYW)%m
zu!fj}gW8OCS}bCS%|=#Ia;$Y2rxCmsRWf1Og3xL(!n?3Ko7f(TLiN#U%mCK}Rf9P|
zX|v1>HDp7;Wj1Q1yMR1U@v;)ZhQxB8fr79$RSU(?ECh-v#$-+cvI_14unGZe4bt|F
zB_-yHu+}bG!oJ%!Rr3+f{+u0J%0oqPB1$}ml<3z+1_!5Z4j@OS;x9U8Q2U>2I)Bh2
z@8vQP>S~!uEXK_QwKfXIhd!X?c!jgBX=_};>E5}e1kP4oqQr~`KZu>dCFd{#;8R?(
zL1~OfJYFk@+NrTWRqOhN{48o%%*8e!s^)!<2qqwnlb6WhWu`=MrmeUnjJd^MZh(?@
zWC3=p8X+p{#~0JIDvam%Un6V-wiFSmSPdU=0=HLVg>LwHcLn%f`w`NvIsJn;k6ghE
zaAy@xDPL5Bkp&AtJ>8tf4KOJp+zP|^2ayV=N+4CaHDpyX*L)3-$wqp2Gs_XSvCV_X
z5%M{V9mUvdtRG5H_Vp~lMq+JoUkq#?j(>}~z7B)X(YfBF{Q|N`^#f6j^3OGYIerm^
zo&lbYqa1hfdF3Sv!lS_EZ$R`&5+?u#{OD<9EVTHFNNl;mMU)Rf!)p0^3Swr`5*~5^
zm9MW-5IX)=T%y3#?4jtkFTHYw9)pmo$3_z=|KBZ~2p#t=R70;;s#cD%bk%d9l9N)2
zpk49F1ij!bC4zXLK9x#V=rZLH0a{*-x`8~U^;>|j(E(h&@;hQx?3QF1!q(Xw5I?80
zW@gSuhMOY~m#9k_m8#iMN`+;3R;R8*N$}Qzlq=GDUvmX`Tz&YA1wwK^*;nFb*dGAX
zHU?s#kB1@!VXMLqop4a!SjO!v)W&2ArbkCAh=Je!yWg*gi3eGY^Kh^MT?H?k?6Ic9
z%Cs*Q$9OD$eX3;-&8XzGS%7UjFc4L(eI(V~fE3&8K_n@Uwl&4rq3;UNU62F49d@dk
z(K#Oq%-l~Mim^Kql8FbbxU0OWaH7I$3^$SMm#RITjFR=^ujar~W12Z&H{r*R@lWm1
zI-Y{F6#-lCu11Dz-7aC<(g&wPbs^yX=uY}Z>P<NI(z9CV*S}{%gzO6Zq37qInoT&M
zYu^%0%m8_vz5%q;Z>-N_MHtHq7%w3O7Q(kFD448F53Xy2)qQl}D-tye3(W@QPf$P&
zb|?yI=vxZj;8xw?H5S6`YAR#lUs<SsDC@!?vT8IgBQ61=xDcZFy$;2;kCz@rEemZa
z&Vy$<mbw-=QR_$2jmC!ZIF+#&!Le7pEQ?}19)A!?R7}`_pj#DLO<PJoQi!ZPIvp}<
zCc_CW&nB<GLfCFV6Tg-kpy$g|zgZHSCI^Y+v?ZjDBuE`?q5z~3;VL^RraTCmG)IfO
zWT_}hXhaD=g(RX)1d<RX4sy^k3`U8;k~5JNUecl4wlV}ocM+VIPN*Q^E}Wdwq`)^J
z!YdH0u^%fmpzRzRnTe7(g$k(cZ#{bC+A2;rLFpUPNIv};Q07H}@$yD15zH#|Ag639
z|G-a#g;qGR#h#03uo?sB?=-HCnTbg-Lp9K8zl*mfO|yx5qG?aK^ZF&QMcUyDg~b*w
z1}b_%%{SMysPJuR=#FZtZ3RoJONkj5DoRTILGc-u9Bya${fG0g?I=Tl9;@Cyx<gM?
z_D_-c0c^bwx)owD=90`!RJjlEk#yWSUyw9VRS;UxIGU0FoYw<*EmbIsp36%(gK^iN
z2YLR*(twr<mMnyrsZ_Xg04G$mAv3CS6*?cCkWIEeQp!C(A3_Lyg78?1bM{rDRG{1D
zEX|q9G1Uhlb{9^P8PRP^cojpBCK|foa4W9DAa@fAnPy1Q{m5{+(kYKA=Wz8yTE<yb
z7`_Co>fRgE$-pWPo@y<?8nklZg+a)&xX?xTcQy#<cR+Ix48hD@L|5M?q*N`7AAeeb
zv`lU!ilUZ-78?60ZJf~{`18L=OCvpf?Q32_$W^3J2}=|m=z3k4i<ojRlbZswt~n0F
z%f&MyY;$YPfl9YMe=w@<NXsA0^n{XNSBkP|u~*+3pVM=?Yt#aXP=)Ok1h8~L(K~*J
znwX{Kv09xtRb&(O$BU2GnfAWY7o6_l-W;&xV#VeU4!q5pv7uv#=ve`QfoE9~3mfE_
z3E2kzpv%`W-|I%;`y0w_z{gpiqnAK^%%t26I_zU1!p3e_3O7;TQnD;&<{5<Mj(;&A
zZ*~Rz{q{jTxrpazVB%<Bw=H5PPu!vPoklxV`s2R!U-x`6oP6%<&D?p9>PYsvgX(aT
za$tGC_;9}&kh*kzo5R<X@9T1M=Ng;en0<>-g28Dvuu1$;u=<`yX5m+rX3H4Watw=<
zX@B}EoBDzeuFA3niUC=vd<pV4EAmj^khM?ai|_of$*?&*)!XgRREEW<I>$c*{(L*N
zz1YliKZ`o#Ja(ecyFOF0#dWE(HPC^^>h}`>3S<^6Z~mFJ)QIPsEpa5_Zg}tT=DR+_
zR0+K*vT*+E`@3(>XQVz?ZTBCOb!d63aYP;2w?4CDdpUKLsfqJn%<oSvcTT&lnjUT&
z!PiZ@t@D?A_W*@?>e7f!?>at$VdCF?h&G*Di=za;!9k%_gA>X84M!ZEdiSumWQEFw
z-<=;jKkR1Kcm}8V?POD}r@_L125k9nvdJ%!eEdhVf1m7D?($&`^h<9HX5@HOH)I#a
z=iaxGk*I`k=GLsyLLCLmdnaTc9t#lFBP-iH<P#Tb=-+>gvoTa((cZZb>>wtc6_#FH
zn)9ddpPh0~MT5sRj}D7QA?d7cgL*$_8?j{neUca1EJn6+XD%UE7`ku&vzcjMkjv@t
z_&jhp0EKk=j?wNPY%X;cdE9*P(7N5@bDz|E9_@OU<N=z|xCExb>Biy-%K2TzQ9I64
zmf^dnU|Q6D4}q+sC*MIM@G+AodZAw`fCuO)JNad^#|*qFrS*VN&+#KVu=s8@?e;Zt
zYDk!e_b(+LR0|P0<Z`%)3{z@)bok+GLWu_!;SS(qPel$wz%9RiNUR8Nbc{2C&Q4=&
z#N*Fp=H8X^LC6f*h2PIC$LaR{GexOthB=EN#$l)5poB%71yG6H<pBl`q~!_o7S_T{
zqJ-n~EgIt*tUywJo-dFq(A-g0IrjpG!e|sn34&b{-C8-OpYM==j{_-YfGwYr^j0|7
z?Ez&sI22=ILoZa&&%<P9^PLryFQ>!#d5}hM8L?<y%<gR{u}$$8QFu0PYq-G-I5zeU
zJ=4O3)Z4NxL$dmUfmS%)gU%ubLWOAp@gCF7kOpqnWCl#a)j*~-b|mEY#wm(o8kj4p
zZDC^zbY%!uUl*vYS<vhV3m9T`sql0|FL567q?%%2xA;*)O9G7w!N+#t=QN);Xbx#g
z9$T|gs6uVB#f%C+h}Q~_ryKslcWw;=@3sgL8XBHHV3*pFK`v05T}7@o-H#VRq?%b2
zCuHO%Gw?$v`yA&ijtMxpc@=YomYs4t8Zso;u7*Wk(?F)mkhc0g{V_`8ta#<l&u%T(
zB2#5V>Liqtir-->beUt_NmKAg+edFb@jwM$-(Z_Y)M;hF3$MO@e9qt{7#>GN3e_AA
zh$4(i)QBvAO^Y>^Ov)tuCJZaX)n(AX8j}$Vogq<I2(Ae^42SN#{F#%?J*osQ;wIg6
zEc;6+?SD0?1~NKW$d!%{##o|UAxrM8;CUo*okE6ZBUm`$L*PQlxi<89$OkOBy2M?a
zwFd&m3pk*}o}CofAn{xa(=Z5mil1~LTgacAWCp{J1XaZJpJUPcrlm0%(?9c|2{OsW
zy0rM4M~VpL5}#T^8I%wvz}f7<3JD8zgNPy$?$RwFH+M&>C`Q-4*#uQwFAu?le?N|K
zy5C+QN=g?uNy%jLC&dY{F&UD|*>IXxMApuefb4Oz5fZdS<I?$30ufp{U*dGjA8CUo
zMxaZ@>J>3CP>PZQqkJ_E&L1F|s4Gor4#<xBhWLYW_Be#p?Guf7Z-|6qE0q0?94(sy
zV*IHxB_{moG(G52Nr!?yelda)LwriDqqAa#cCB)`L^!sr1}F$2*3Bx&Y-)bVwUbzn
zJZ1mvw5BX^at?l#GBZ^1z7#XiXA@ib8X;7_i5EwV6nEw_1DRe*nEMVtak|$uSCMYT
zi<>Kery>i1^uP7KL3Qj;hAHs8PAtImROvF2oG_LJcpB!C5<`~&y&gGwnbye*wlKVr
z5jO&KedNr1`9Bh>oD5%|ldwy|g#d3?MBjl%uoz-PW+&rG`76uJ&<%4wWr&iRITVyj
z{nup<%);a^1F~bMkcbf{=s}9gn~-pbLp>Ph!~|{Nez2#`-$|-Wu-<bKw|@ksR^s}r
zq)5r|P4lMc>z<GAR^fEN`$M4qk~}@0$4Og>_?98qFc<FZHn(9!hfjb+O8p)fU(tI&
z2PR`RhVyozJ8z5ZKqOC|-%BnKuLu7XYzOg`(F(d2rv^B`%G=)CV7aw81u|RdM7CKJ
zf*&niybvC;_DwnW4%Tj0UBq@ULDVoI)%`S!H=4|WwP*2=hT}}2);WC4e-#sqd#+V%
z(JKbpc$^t4VK|K_@PQFJ`ruwAa84HOcRtJxYv8Zg|8(W{9Sy|z`4Jilo|E$XG(hY$
zs6d-r0G<yuD9%4y3r?~-!p5qiS?gL6<9={(b+qOV_)VkkJp~(odt2esivbc6UeWWv
z5pV+qZ3{v6#WXm-kL3%TE|+9EqApckN~=NqG|syX;2|4@I**-(Uo%VqcOSH?pdmW0
zSfc)ih(&t}LVN?PfIazBzz$%eC!MH!w@?YexT%Vn+ck&b6U@*OcdqtYp&F{C;-EV|
z2T@A(#@_Lv-4ytntB@kjQiia;?*|zE*-RwLb)$(HT7|C?=vm9(2=j+1l<6WZj4%~;
z4M4m49FdXL5~alL{g8DZ7T*5P5j^WjDKJy^*I@YPyM^44I<rfL2)X~0598mh4LL;%
z##It55EtswD*VA!nq}+-=a#$R*Brp}c`Z}nWxeP{UjqvK=iwlfU^=EN>jU`({?6?3
z7~SkV=x}~wkm@8Yl+@{WGurrG4l8g3hV}V<C-hY+AvZ{Ih8u1fgt;y=S&?DF%}{D{
zb#x>EBTTrSpVdsmqC@D}SQ6xmhdUoi5A_x%72eCd8vOG7Y_An{ikHo`0|rutw;W95
z?luY4TJkX11aK%33^`7l-VdaJ&qJ)2W2mvt>w&IO&w=$UN@2)xzmf5%yh@qh&DT?2
z!weN(^l~=r28E1P)XttY4pHYD@<*Uho>1aeuP==0gW_vbvMW(6m??cU9k%{zb3Rky
z4rqP5odud2<T!A5gt=5K%dA)SZiN_Pq^7sp3_UbAQ=x0cJbYJP)-ItO;ntwMEtcdq
z*QHIhU89tPBPIS3+Io6sp&h{C<;|!)*1<CM-*=gHt1mMdohig2M35{ARNlkq0JTiK
z&SGB4Xa-`$cO~L=FXme;1pBg~@*oOAGiRpS_9QSY<psUDh=*|JO?|hZSER%w_^RtO
zF>tan?*I<Y-(vaaNKq@1@%V@ZXhzA4L$1K&_aXW#{eZB|SM}G3k|v=mrl_3Tw5dfg
z8B1A7mN2XyS%CZ%7Q%I>&(Ck5C2nmrK^8{23GO^g4sn5y;54Nvir8!@`D22Xpmdvs
z(m*|MNIr)-4TPaZ?Vv@GS?M4gtgn1$X8}C4{<<rM=sJ0(G!aL}zlMIRHALg~RRx-h
z<yoH5UWv=-<BTzbVQ-5D(4klemQHz&;)u<JgO$DKaOZoM=RZIOLP!+FXItM=j?CiI
zAJTjz7+d!c(EF;wLXex%ZGnRUZ-0fnj<Fuh>NA32J$(WoUx*;_-`U9nUWc&|I;Lg>
z;o#Ya+o#vffqC5QGdmbI#X_+RVOdksW~DiV8-nUFCJTx=pzAEP?%pUvj4wmz>AS70
zd^sYm*-gL<Wp7EV8*t$&rv3_K*jO@dmqGa{TcoC3SkWs)%sp>Yv^V8yJTx8r=6c49
z7;CDbEDZ`m<rBJ0s4UN|TnFYXBMS+dQDbjM2#R|>Zl<SA$^w-`xjGpSt;5r+RpFc)
z51BX1lw)K?;XT9Zm<*OX*~=@yn_9ONbMz3Un}HT3#$N@b-j^aNi#0?4a&q{ggiuac
zxpq`ptw~YKlri0^4|+b~Fruryz%0{E!NcEL`!i0F`vDLZJ8KKErR(Y%bb(sy64znd
zi*;_EFJm%jhX$RXAVlSw^Cw_9JLC@3c5@~M=2lRvQ4q+cM(f1D#bvEOZ%B#qx)07d
zbhY5ge;l&ex=FE3ahXf>0gx;Ex`1K|b4?6%sNkffAnfatv@-pS(=CZwl^>?NI^<W#
z{k<h$!<wn?6=Jp72xI}QU==-QDo`GzM3n%Az95#NFYD|EqU78czsYaVW)GY}mO>9c
z8myaqzGqNnnR6@+w(akx1#pC$mB+xaNK}x72}P}fIWj+<l$dp}+FDFTl!Hbn+`LRd
zC?Z+=ty3IhO=6nr_5luV!Q^`O@#a9g=9McSUXv2tNnA;sSO{G;_~Yuhj31xKiW6oo
z{>fS{<RD)9dUXG7o_4CzUktbsnFe>((0%j$Q^X~L6|8>@Ls853cB18WG$~Qp>rfR&
zyre+j#)+F&0pQt_5QF4!BT$1YEYXg}4>(KyLp7EYW=PBMTQY!!u<2w}P>2g<|8jfx
zOB*8*VI9R4m3IMC@Mi*sr9|CPeE^LN5ZA>rpYWUjl<M)D#a@Y-JjyXu4ndkZFsF{_
zb@~LJU$=6!IQZ5VdD6ptJH~sduq3En(+#N?AJ19?)`csioj(lr@HhZ$980u$xu>6B
zTCk-!&L?jcMI7C9Q0}3?*5L0hE?@ic^YO<!#Ome^_oH;42+61nGU5dsZ@S9o(8H;h
z8z(>7=GWTO0yKxH{PrdE$Cle_eR@g)k|O=Y*B0AV?12=nYR-=b-dksZTqoNPu=;=8
zS}}<x<}FOlXdF>7j1W~>x`@7OA9?F<tXKSLvcIOiPu$VT++sT&Sd;^Vkrvyxem)4|
z!%Wkvsyqflsf7vBu$!2zW=#`~0?NDi-s&hHou~!}Vw@uLR35GD>C(1%mV3?@)qWp&
z-`_C&^$xC}Wlmce3T?cGp_=h3JrNlkch(#t^jnSlgzsz>Xca^QR=R|UX%pK?xl0#)
z7Lr<QLqzEdf(Q|%BO+LgK11=Dx}>D7mBDL^zSt(H=ff9iVa51nW&h;k&}s5w=7W;~
z@s!oS_xFzOe{0Up{TS))_Pw)SwqB+E_o++1W7qW)E(&p7;3!>vlGJ{Lm|~Xr_$Fsl
z5+zpsZEbl0%TbJRr<oF-s(9+R|6Q&|hEPB!)tjQ*_|2qI#?_I`WL`ef^k5_@=*Ymq
zLwngrTxl%C>7QN^*5hf_a%0~P;u@1fPc@fgm0gF7mzh={nrA9rT#e^1lORdN$D_|N
z0sEF#oezuFCZx+IRiDGJ(K;jEvcC}kW{#e3@ukQ91Uh(lb48_a8T!sI`SM1#joUV5
z{pj+lscTAu7~5GYF4S_Og6vt%RtE3k;*0F}gKgDIsNaTjfZy+HH;#0MZOq&Yf3IHo
zd7pfWi00M}^yz5N)N;(yaAVC&6!iYFByLBtErL=DFfV$Q>uaD9*QOBVnAUpr(=WJL
zDA0!<uio{a^Zh108sFtPIxM&UUccgPr@c4ZgCIHe<N5I3s7HM-{yw(;`m$s`Q5F39
zVDl_Vipryb=|7IP4F?x&yyfs&+99*%hhd*<C?=AAqEekz&*OG*_gIMNB;l`oKd3cm
z4bp6qKG!MHD5LnoBZxat!&`jlEl}Z%rEWo^tj0f-{SylvE9IR;7`@OHy!wd5k5nOC
zm?9@8yicA5C1~Hg=I~4TdV3wWexz9x)6JiNUCFD9&X@c-Vj@4ePv@^ukp&c>?Y=|j
zV|!H?KuGGjyb}|4me~vfP3f=r$!ZHd2N5l(*P~@T$g&7gD}Nfw+P*=H&D8$Q4;|g4
zbz4$?F~eGHzV=41Tdid8{6^3TwN1k!y-2#3?oQoOu8>!O74z;NJFIUiqEO-@Uvo*$
zev(<(CBDrYj(mf+l%yw;TSu^~s^2fQNjLV`I+c3MG_Czy8n$oF0TtA8U-`W7EGP3;
zz&qJLvRU=b_W%W8@p@@*&wAv87#2a$<VGAc1QxK}M}E*&1*u80@Ee;Xw+%X0xDvDb
z8R$2&`n!5xUBEWTDhdAXM*NO;#%wtzr+L0yUMM}e9DgXL@-A%H;KfW&^o;>393@ta
z*3Lo$l(L7$6*;7N3*uU0efJNa7Hcqmsc(bjhf@UEgDQKqW-Oo!$mU_tGpzR@j4X~!
zZwu=0H!wFDsW#-n-O@h1{;>4ScN+$Sg39POg`Hm(XmAWY$l9wP^zTB=X0BfiR(3n&
z9_3i&wm0L5rPKFO0wKJFf`(4bj7w|0*NmS!WgUUAx6z?7J;V)plEOAkgqn$7bDTsd
zpb%|L{hCHl82c|<M_0vJl7_~b+Mc34TlbpNzR9Z7E7D(I4jv7ck>0AHtoGLlFkyXL
zbjT)RjgFalc)R*L`Nef~TCFNrOY-=wn~93$Xjb)(EsP+uDHrZ8ej{7_H-h)pd^fTy
zTD$HwEjjKY@p&CHm9c^tR@pA)=E~by3le-&z}v6%rZKLWH@Jl4DS|KjMtt5j72|Qj
zNBjG~Q#fv4o{)I3S5*7g`{skw`Lb6Ri4v;<jKNQd>vWg5Ztq1gae$wK$6g8OZEUXU
z*LQD9l%u(;`unwS{|3@}FFSdDL8+GY=jgKPM{Cqve@e~D6=xpXl8?c}zg+uxw1nTO
zBmK1NFtqukOlNCArc6~iJg$}b$)g0BB`kXM(c{VT1&5}U_iF*n6$o^AVvlL&Gc#e^
zTe_&-#-|pA^c#C=N%oRB_F?nQw_KD&?8|;9v|2mCj+DPVu)?yW_f17PH`uO+N12S@
zF0lAM$az^lW3rDFv?j$_@K+o*OSLHnAa0tJc4S91^4&qJUQP7%p~x$KLghG;^g2U@
zl4$957#&zY|He0<FIG`=_Ik_DxrC~NiTY1eq5ekcWgi23)&_{MXVRgx7@%^chH*RR
zdZd18Ec3;Mov*t$DR2gU#`K8qoaI`#ZX!P2p1_%JR@pr5q&-&_Vb)(g$MTLt$C{Kw
zjE!zfU1c5TPV*MWDORcKJ|^34xE+Avsk}~NLLqi=Vct>+coT5xIo>Odv8jS{SF+#+
z68A&ezz*<3+_Xr%HfMEIz>4Y6YZkNdD4UBX+YWvvR<R-8WWh@?X}Xv@ljQ#5(_-WF
zb1$abqlM)6GeQAAn?jS36gsCeMD)7;7+$IU>g_l6gIXNH1_s|Y6D(5W&PB0_Y01LI
z5Am7jg=ELtFI`0{M5HxWac^{JZ{6%;0WFf+OtI1O7@xn%Q;X{RRvUE;$R-OIp|-B!
z`6{}jiq6N&aU+o%4Ec4-(7fGmWwhFNz9PtMUW}gj2IcFSqcSDXAk+V7>X$0mdatFu
zzcqWS0`A=(sl(tp+iq;J>1%f>+oUFjyp}rMo(^-|`A8^~|DV7K3yknUa;5b<s>5c$
zOpFQj>jgZpSvZ<(dk}Y{z2adfkr{Ptlc!8a$d;<yxb<2&(~^-BkX_jf1(?ilUG4NK
zR|Q?{9#p-;Zh2bzQi8p)pflM68_w^((dlA0%5)A>HRN~f9o|jVm~(QO)O^3N&A~^v
zRcz?<L~rMwr}NzFs?OUSKBiHM3$Duff*&I;Y*odY%a0fT^J^JsxXJRvcD(6+Y)JE{
zRJrrrA-d&7wMxOq@`meDvQ?$#y<J*Wr@9A@e}+o@%T#Q8cuVehPHa8VuxM4Ocih}m
z&bLtlbFcrTPh^^x3JG}-&}`+fdx_cV#j^^%onh?DM5+fu@3;^ATBhpiGt}-qyEHRD
z_euW6{HY~%LuD&t+Ba*)KL-W+!PoPS+B@2ti|g>=KO;>O=cKPqhlV^b4T&vN3Hn+0
z+8`)BG^F!oow}glPPG1BvueG2Xvo%lchZbi*eIg0ovTp%jzVY%*3?EPOSy5g&0NTB
zj~cDn$LzLK_Mz<sxsWT>9>UCKRS{m;<C^Y9_oyT=d-eVeV_m7n?N^<0^R`hrYAcID
zX@ft^MTcMhV-^-EmDv2fxx%vTzallPDP=!X?YGT^I3yi-IV)kNL{)j2TQaE6F-vB`
z=5^THb~^`*kSm{xg=5x^hob%W#SK=$PZ8X-;|qU{tVo$|)WmoTG))VAvsigL|CaaF
zYc;6bDFW?A3ND6H@W?Quj}68DWEMzJ0L^+XFL9pe@Jac>LhFZq<wjq9pFl6+HC7l0
zYCAn)^IJv6wDe}PnG}%;asv3|4>PImUZCpo;zl;nT`k^R2&cR5#rC|raF8gp<y*Zw
z(=&mf1yi2mCjZmE$=c4Xeu`w-V3l6gcz&T$dBvS@YJ{^*)hn17<D1M5m3cpggSRha
zq{(;;gQT_c=O=KXQ$j8q5>X$}2HzJKD@hoO`s}Hl<8E`0-MPD0`i2zj#sxjH<sEm`
zCLd!uy%Cb0`J5SzIr>s?<>&kB<WM=^%-hY|4g(AexAZ<JHf}J&7y(hmh=sRLEyph<
z&GmN;w;!Q%51W(hET^up-O0Tt!gcmQ=N+7p(DbfAFjY^VpOfThH&3Y5(5G+vInQ3;
zLqu9>_wtbH$+JtX@O{>WSvuK<ICH_1o;xcub>FXFMcB?=Dh{0i!)D0_q0{wmvM6i0
zyl-6e=qaPXay}G!U*gfXdFvvg%HO#qp`b*J2`xcqW!@<zeE+H+RT_@njcxLUMz`m4
z@fdb8OMBhvQU4<COpWen>%n04S)JH41f?_U8mFN&nzG9P4d(mMFVCF38u{w#<%m<;
zc@=?pe2uz>^juSRA64+<k5_J?6C;G}pDvVpH{^fzzk_!+Ryd)9Iv0{;@VPlj3^WDo
z1|hfo@3iMLEzmxYa%pYrSi>e-ns1SsJ2gL-L!2%bE+{;V_uO5@etFk{DrR{a07wL^
z-!<g)#7>mXH6W^9xN~&48$=26KxT@QZl*(Q7_Hb8ozPH~n??yPwcgYUd%0$uSgazC
z$LSi$1()}^106Iy;W?)VImcS=NuFp`nl4{`7M<6l^pZE#;$0HiZ#2ESl)txJSB)nd
zBcEiH{x_jqulh-1y(DMwZ4@hh_)*Qgj2DwP6D<)%Ln?Cfz7pK>38IP9uL!!4N4FmQ
zpf$rAOKrM0lC|<TV8WKn$Yl=5>3BQqd7cIAiIYW7zF<LnE5kRhuR4Jxyk3FcaqKoB
zTxf~7&S+~g;#a-;rIN}%mq@M)Paj%dR114}Na?n{=~qH6$74-T7tFotjTF$$RH>xc
zw#@jj{zW$0{#8~Sqv7o4F1lXH*-aQfVs{ba#Eluv>-GORn%SkKPtD)!_B5E^Cwz2I
zw^eeq*feF;n?HSOv2A=ay20kF2J8C*%ZDVzY8HpDpw8_ugH21Cgu}euF65B;=KVt!
zHnYTU<wI8}aQRG~N<&#@(SziYwBU?Fc9$6f<V`w6K3{hEmUH?rB!uC9bL~PHrMvg9
z3ORacFgqP)e3IH&`O&#mx<#e<X)@b^`$f>B7`pVMe0}EaKm2{iwB~c#v=O%vC9BTk
z%`Ib|U+u^~$IIRtVel=te?INwzp2T1cbfdiwyPHPc+i8Moy$d+!p0t7z!28iUQ=o=
z?$+rH#1}*k=fiHE-ny>vpxyTJYAmbY0TK3Q>I>%Q+;|WIR6i`GU`Y9te9$skGIgWQ
zf_vFL@KZ+?6Et)=dG}O6Jn%jF4@FoE@&R*bE{Y1&?R(r#!0!zzj}M)27T)opO~-kt
zD-^Z!4gE9;Uid&Y%+)CIk+;8?mQ;3Yv()Iax8z$Ct{A@iSx%1bNwHLoCx6vN`)wPS
zAIrN^k`FITs{W|pk=3E|sY@Z*N**n~$u3w~G+3y)B)mv_Ci{uBP(|LZVxiiVe%5a@
zp{2XyY>#QKulnn6A)Oy_U-_j}oNuxI8PE>D@3))tO6=vbp|{z557$DlWn#+OCv{0M
zDdI2u<gbx|!_>PoKk9E|t|JcN7ks0gsj&fGgwF<g^t<yT-WPP~wyLK5K9XfxnAIi)
zv>z&HEkqHpTLuf0T+dbRl5qE!edEm$ZT`iTa`NGe%c|7Vfa!lvH73GXJQ<UzWEUCL
z=j`v&N~E=PJIZUcM#1?WZ_03}C(Tih1&zP%NAHR)?tCchVo)j2|HT`+^^$MT098D-
zA3MJ<v^AfYt^8(;MmDWWCMPY1gujc-h3VF)5Ki1K*H$y=$E;3j+g|!Wm8Y>xdw<py
zz9;aHLC0wcySLZ*!Eeje^raD$)x$Pa^ZxzXFFyH>IdF>3zZAFm%1_?BBudq-`{CKJ
z_P52yF#^&~e?Aqb0I~{crw0<fN>Thm>9bXyE@pHtZEK?^v}e6HzO4mtfcvv?NoN_`
zQ_Gxb(>D4GxtmK7N}p8{LX!i<!zHEfF`<tA-S;Uoi{Dn21o3IT7Gt`p|FlKL=*n}m
ze<5;s1Dzo?)h8z#nA`qW=!snq558=FVn38al$E8#MON*SUCMkJU>P~v@-x;e!@^Zd
zVc}485XF6TG9{e5>b}IOz#EW~*OnGmP;mKo->>rS!G9S?#S>EV*BA(q8&dYU5#kXq
zb_^#{<_{y5uK#mx7Yy<qNm^>Ht`7dWH;9$>ebIS~fk3{bld^68{e#En=gEoU_ajr&
z+r1<^#Xhks<X7DO7rxHeg1<*x%;nBLeebqCRulGpM<1GDYo2A=dj+s~TDoax<+^FQ
zVRJH@$_2wT80NJV(WT-8OzIv}GPK%+mwc+|SeO;gJoa^9Ko>tP^`+J^CnjgtLzYZN
zz#vog_PxKG-^%Z`x*$x`*UI*MNWqNCyq=Kn-q${FUYU@TP8yM4stq?a3EYn<eQHa5
z_A3iHyXOt$uWjbFY;Yicwtm33PLOSLYznUQqXYYMT@dj4gFpU#!$h;Df#9G|xi$+P
zxc*F7p{o@cCNdjVsI^H0+>=Ggp`NwKqCRE83Ss<wrRP7yLO+D@H<$m<uTm~>AoQWH
zKUn?Xt0?AX4g@E{znv3t{$1ePTv#C?_y6^St89c)ViP5!;w8#~K*xnNf!V(WE}Q<`
zLBi2ytDDkTyQ*K${DvH=iZ%r+J}l@NIoG>R_a(Q=s{AyntB&1tZ4++W<fa_E+<p4c
z4d-E9aG4C#4UdNoWGOO4GFGE`3#wgrD>!_UGvm45Pc%&?tLqCWkz`eV!su{q=+WeH
z3r^1Ro0seTw%x9=vFX3JSFhHmQx!2VI`{|fjs_jp^*wZ}3L-3amwk;3x$SxT*ALr(
ziqZ*bc!}Q&kqo0ZA{7ijStuq3l45MsQQ?<;TL#+l9%iIIqOQs^OWbxLxiXvX_;po{
z?nhJcmt*R~D})n!%WrM6cHD!6W!LMiYq;bJsjdR)`4b8$eeH<F)w|fw-`odh#u~3x
z{-3)@F#Ketn9xhotPRJt2;cjMwb!^!{59n8fmI;#K5GbX=%0U1>ibt_TWJ<fJIUG}
z{W>zcvU@w!>-Cec%+0rjEo5~4Q<tQ_J-o&%SyB~<3J%xnEuN?Q!RkPe-cvQ}O|H6h
z$00Y;<Z?Wj_PfAC?E=gzN9OCoqVS$SFk*w34$<$#?fU0dFQtDW|5#LO$go%P!&-m+
zAdmDgd;K9x?y7-fn#xw;`Ck2>cLz%U-p20u8f&cGkYfG)wdKleECZjRjM9twuAi$J
zQ5ygMcGk>o+66xXAI1!Di@-~sGRHfAq$d_t?Ud+ztI~MzrgE_{>SJm?(?DjM<4~u4
zlbcM`na*ni1($L9!@5dg-<{Tv0zoYrHjZh2-1LGf#Laitb(|dnm^GHZ;-uWyQzygn
zL|t^IvzXK%(nR1Vg-*)Q*A>d7_gR;#L?rf(5u5X%;@0$EyUX|bs$y5po=N*HTgdF2
z+9b^^r2QlsyAaV4HpEAICn3sTd<C%Sh0-c56ijqVsk3DBJc9&=>z}{jHivBN!@p%-
z2@aI(v#~05WKL4~pN(!9F`fUT_P>oru4}0_+&S5OU?!V+mUD@Eswg9@%=_Dq2Tb%z
zFWUc^48?Fqa1{D+4ledwzv(=cu9mBUl%Ix$upgBQ)Rq`$sVdCF7QfyLNX?goC`xtt
zKQHUngB#5v%psa4))H^Ng6q;hytvajHczJ)zi|=N#^1d-9;695Ti<-7BA&oP2g!ED
zjSN`ft@!sez>ZIf8*%fMDk)9aO<B}X^I~Omith@8)67m5%(<(w{rm4u6VZa84>~Co
zoO$uov9FEay*Zi;_u-?uobsCca=$nAqb*);Ma-4&eV4O25UcaLJvyk}R%~g%Sit<o
zQMRk9cYUz7qw3{k@|;JP)hdHWSlJzks}q+8tlrYpXF_`z!nP#|aQe%K6B%NawiZ!a
z)9|sd;ob3^rpS(cJv*mxdIF*9EHH+;@qBq^K!tIt?xI%!8^3yF+!~j1vjB|Jt6H7m
z{x~c95);;?FG~4c{M<QVA?2^ps<z)FSG{yf2`e`x({8f3KkBZJd^3ZVxYx(rwUTe-
zS0GPD2)8pTIu!%3p1M8k9pfGE*X?vsw%LacI!!gP?6ayr90ml#y1m_S2mBI|e>8Gq
zH1A#jr6JY@4MUT5W*k$k_&7>r{%@6Y0Zd?l|HtW>Gct~5pDLs;>kH_w&6DV98)2$6
HonrnM+j++J

diff --git a/docs/dev/articles/examples/timeseries/timeseries_anomaly_detection/unnamed-chunk-5-1.png b/docs/dev/articles/examples/timeseries/timeseries_anomaly_detection/unnamed-chunk-5-1.png
index 9700adcd2a67b6e5906a728b383671cb635f0dbf..89fe25b04d5daa66e7dcdb099b77208f70297cd2 100644
GIT binary patch
literal 35821
zcmb@tWmJ`I+bxWRAR&l!NlJI8fPjFMlr%`U(v8AWknZm8?po5_xu_-GUDDr)_x(KY
z-tX8yKgZDJxblo6<}v4S{Zx>XM0-Z?3;_WFP3qm-4+sbk{SXiylp;L@N3ymam4YAS
z&+pXj5D-va-2Z(rI`4Unfba@I>aB>fbK*{tQ{qeUnXZDI*E^QRF^OTR-i1gn#D}qz
z<(^<-*77S4Li}A3huiqu6W^@9T^wgmdg}LU^BWQpO*o|w61mjJ)^JA{Pg&cQq!e^F
zb9Xaua}RdmaW_4*^WI5#LPdV6WGintk)iJ~7ze`SFJs61aUmSsy@`DQeo-FiBZ41>
z#}It*Q;M;MWOo|vC4AS5d)2Oc7ZuEyayv&TgDwNTWhS{ks#+J;g4%5+jKvT~dD>;W
zZ#6jS=-Mk#R*@0w9!^#a$z9zAP?P2vt_GR~JjXY|KtOmi-$tbGcu@_#8<$DB%P%?V
zVb;YL-aN#;)qKw>%h&AO8^O+{s%jm}@6;ys6C=-g!VBl-BNditrc%M}#Z3C6s%I7U
zpU*NSUkh{I_ITYXH~6uO@>O3KP2t{siWLN7cEGTMz(%rkQ~I^xHpSEsj1N`D9nqcZ
zobzR$RN~$vku?_)dVW6Jn%XXxDu7<Oar2Y4?C#&Jc4Xx&eVn2q`Lo|hEbcTnVl?Fa
zq?48%JTa7pI*8TeYzwYE)9C3Iim?Q*UAkJ!xIOFQB8(|H#Sn$e6IMGnoNOmKSYBv?
zwZKOAI$g^Vb`EwIys2UyI+a-MO~GkmUyb4x(rSiMkFIvypselgupO28+H6&tw0O_2
z5AvPPH?XVWkZK&B*+u9&FK1vs9ZVvmft7D8bLYH|yMX;LxV;m$g%4ho$R%Ho27Yg#
zU;TaPa2I#hJMP7FU3RT?_3OCLsRHhLnngP%tW%vMiVck};PSd&@+GAA6dbJB6=rUR
zHV0bWf_=^{h^6e;y6L;r{%pM5DMpUzSgd8R&3~|FQQWOl;jZ+xJBQmwV6J}OXrOz<
zXbi5U%D`v4z20Con0Se1mqbMN_RRh6I``??4!v;QS=FMTjY|oSrGnZVzt=)>G52#T
zr`z9T#9lh3dNFU1$lmf*nXlO`OU|*G$-0!<&s%Vx6W#9V|NK)eUzctA3nE`bI#=;0
z4vwpR+G_kfOs**Y;+!Ax?&3!5+m_W`<Eg)e4FTKHMb&G0oA@u=s}yj?Xrx2y#JRiC
z5%y+b+kEHcE2s0Vt8=sms|$0%J}JQ|<q$gOj&RzL+x4&PQ@M#xz$VU2%a-@mG&I_@
z#Z1}czpyG$P1?KZ&QhPxXFeCK`xG;L{O1R$?Ule}kDMchC{h_`ljS()ZkNqI+mU<>
zcg0VWpIS=ut36d?#ogqs2Rqi#D@V<P!Dyz_st&>9_5He&v$P`1Mgha_i=hF{((Rt~
zfL#a%c~aiSVUN@OJxkq<GR(Sl9bOsho4CQju~3K0eOsiwy4^5qpVwZlUF-8dSIn;0
z@2+j!p9=kP+w}eh*RH;}dd&~>@#zugww|V%o4;Nmlb<qKyIa1M(S|r*hFA(YIFya=
zs$HGQaig1Eo~_f<NzNr|He2pZ$=axobQHP+XAb=oTdmdmMGCrb#t65eSFf^_pMswb
ztj)!sZMd8y*r4w$ly<<Ibe+`3+3OM))J}WJAsCvTBF&UP(8&TC)`#v`m1ZsmmAvk*
zR$y)}UGgY6*EH9z){Z6gn{`@Et*+&-o;7e=_7SjD+Mts`LJM`oiU()Q4!d5FX<2s-
z7P*M}F+z28bVjIl!HQ_=h$K_`vz1jfuhy9`@!!A#Y31c>O;(g-1k!14?l5X^($mBU
zypAupRgPd(c52mitI@#%4S|mk5EOD)ZHVnu_iwfPxIEdnw;73xchpiSv^?(~;O<|#
zwv3|r@||r*42cVd^>y2Jy{kT0zP@nNINu!WH>;W59jsP6v1oQ`@lJx$Rj^Ym*R1zK
zFrpZPE2P5T3g6t-aB2F9z_U%R9PoJ!d`m<|ez6XP+<7e6L|8ygH$s2V-!&W^EY%eZ
zlDfz|-CzT8At)C2(<*_^WXVm2jXb|H%jM#+v#4LuW|Ugh&8%?tanK!09k#q*>UMeS
zpv&>OQ<K{*N=S9aC%T=`nA3RZZ(pexEuzKGZaE~{=E+y^`mKWEZtC_5v&EzG0Mfc{
zk>(Jyw{zzc-YHkYcLN1&a}Yn@1%|_3Xj5&|@o>xcDd_p1GPESg<`II1S}>ep|DW+K
zPPL8lEt^<ca5y&1>t+e;sjx@*_C!hdVFE=RYK;dRx1_0VN1=#51wwp`L$u?1JbWBU
zSm8wAYS8cqY%GT2Mjaxz@afxS^|s47p0hMJn=SS~H?GSmBrr->r<<wFl&KQ0Hlqle
zq=TKYWl3FjS2?rJ7V*1l=_YTO4u}2cCaM$<xYf8UV`?856i1tUvH*-@7A=9J-&9C%
z<-l4t=_EixmarlIgjA?d5`|k5sS9R1v<<7UDsGyfikUoDLz1mOF-<U8F4xqaX%G`6
zB!ev12Cl`N?0&pUJnM<-=(tT!%Xpd?li^oTBvKF^mssEu!!$5oy`wtsnxgw=sH!gB
zVRJUmtfKFh=Q3Ya4jQUte~hJ+g75Mz;WI=j_<F!3xWcGv#}W1<Xo!%-R!<3TCO>&G
zC49Y;sd+Vy@@GqXyh2PwCt<H?X59`lO>p40aMyT{w&~>9lP|jt!sl1J&(5TFwiuqy
z^<65$a(`U*x*z^EiGf%B@sG_X-!~X^cblnD>^G6*qk`1VG;HusR(0(MYx4E`R~BH-
z2sjU0!AXRVNJ#ftMvrXEkL5MnciHngV|-4bmTWxP11+V%Y*nU~7zPUt%-=Fnoblat
zGxC+8EKeVP(__iqpDQmf*R81y#*zwm7fKPzh>D8(_9pdgnSAlk`6TJ1lG5_2D;n3<
z`Bg!FJ}MU-lBkPIU9;VdC-l|E>{wJ}bo7?n!CS?XtDDO}Rb}N3N6&d)Ep6?h*$W54
z(+l^Bi3vnPG(;b4Z0yUy)2o|L(+Refulj7w%<_aXykBjcWgQ*a&JUI}Sm)*Ek0$BO
zcWj-yT9})oauFasrlY4nP1-)b;Mj`%{ufFhU>Ji*{u2Fv!;Jr5Ptw1*-P0+Vx?1!l
zZN531&CYy^hj>2*hOV>S*{ic%8R4s&TRoO{NW9JmzqOs_Pk%jn9vjaQYX1uCjvjA*
zK><BI{pPw<1hs4uUlay;Upb9`IYYwz6+6bS!7Sd;n4g|H3!wY@cp|IFJ^j16eQA?=
zGm6wu6$=I5Z(w`BVuu-B6gTYgeOZDq^m)I+h+^>_LV4tXe81Tr(fM*4?Kelm!^1!8
zBRUZNJr+DV`hj2oM|@=Y5Xrmr^J42$Quol?{C;rLi@z86QOQKVk7N6eVSBXF<L~dU
zUSaHUG9u~yiip>C<zTVxB(_MiCO%#wRO>}j6l02@vx4<l^;e2^uY&?7V;kCn!!=z8
zcx^%arC0pjS|MpuyT=G=Ct3qbfI+wxfCRHUaHv;TFmX9UilAFG#?qlP5}Eg_KR2F6
zrueM8^|IuG#$}i7#U1<aua3UHK?;-iyh+)`4fHt5GwIU#U{|S3r(LC|JLQ|=x|Nnb
zZ>O5*jr9QGD-NYpxRLsZ-la%5Wy+A`y0H(R{6j(Gx}$^VZgQ?(4lmRG;OgoemzXrW
zRHtsAiV;XRpd-#cA6(}dJ#=?DWDVo^py@w7m!Z35jl89~XyUI}wHe70@63d6sVw9h
zY=^~~f2H+Mx4c@BVjD+2Iv^kbnBWFfIo78ob!c7bJk~%9zuC|z19ND#7XCO23&Xfr
zC#S%7P1W))5ku5(BTuLnYCaCP8Peq5-Mt99oosEZn<pY`-;FM?fWS5An3+u$PBQdK
zJ!ppBG!A<loh+WZ1_^UxW@Ke$(He4K_V)JLr)UDN*>8WFeM}kN#__?tR1pfdX;f5l
zm+~EnkB>jyo}#}SeR$;B1RT`fnu{zmJ)OKWik^;cNq*C&fAbXv(nj$+ELzo+A(KYM
z`R13b0c`B|e<t88vljtFz1Fi8UR|!db4K&*Id@Cjp|mMyJrInN7(*@;R;3c1rUUxN
z4<7_CZy<6Xe$dd+DA5s2VBG6&q?Gc6!gn3FSu_+C7abdt<Z^=Tz~X?QVW0R>TidDv
zCMwMelkgDIeTC6IFu*yQ`wB@#MFmqkOZD1Qc<9)%f#h7sZtcCkzW%#Mswn9+h(0ZW
zfvCh1eSLiu<>e>5&^@B!{QUCS-OEJX8pV<+dL&Ufx!^DfpB6n?Sy@ld=3qMyp_I18
z-!4K6SVrnhjEqIP(B?BGbRwS6s3>gPkjful_>u&M=#z>X8yij4gxt<7!2^WW&z%J>
zx1xJ{rE|vehAtd%3%ZuWC~5imXWwOfd-H4ibZvEYlV@uCv|fvek*`B8HlgaXhnf~+
zjV-a@lG(6eP0(+xx*BO~>fI}<j9ry4{G12~nfOZC@69IiRs0%94yqXXTJ$ik{4<gJ
zA+Sm3U!`j{vM$!xk_6}^nkzJxUUpY;HqGTfc?O}Y3!?o!hzMV&{nFo=c;-Qs5jh`y
z`^xq(mm>Yq2fxSPVUwT!R6v`*1@`SZ_bm@y-^s@jTf45<{M05!Kp6Eu=n+v4tg)V-
zDbW>Xh>^xBZ9xLw|L~NK_Q|EJi;Ihtl+-H>77FcOrUx^Upx%16D7Y3yul7PzT1EzX
zeIyg{;z=N)Ac&l$k1?m<DMB9ci72K%91s3ERL(%kYumnRYsNeJ^#q~fKC67C3N^a$
zc`o#I;~}K{0m3i3H&GA%h9mLeW-p{k)+wv|AQ$>6)j+%E{`h~gP&C4g-kZM{Qh~=j
zyAPKiP~SPvBi^T|Xarjnn)?u4`WVCbe_aJ?#DC7rINd({Cz63w_RmZ7IO?q*-Pe;c
z2z5_h{*7S|k=Rr3Q^2nm;NK_zNB;Zx@Z{eE(Fm=mf8X|Rw)pizZ=Lz?Ti^J9OLh8R
zL;CMUsa^lv_5T<n!vhGu!aqZL)4I*|&v-*M|Cz~m8ag^Qe&k<<f1^LLX~;#GBnc}&
z*?$h?vK$N8IEsGl`8(|U{Pj1NJlJKOwd4w37WM1}u0tXIi)bhNBgm4A$IqXL{N|B+
zDLy}idcLL+W~7esIAtiOEuB*8+nYD=gc*B6Y;nbKN{{s|E&^0UaYc^lTuP~OhHWl7
z|MXW#e1lYwKkh=r%^+@C??)N&g5ba}h^b2s!P6?NGzDqgKiyjQ8KO@(CCu{6xxOhj
zZj7P^c{#&#0uIx|r4FJl+EYVSRaGOSEL+1@7=DgMTQZl4!eK<{h(3IB@7sZo|5V1T
zmw%s$NDYBZF*-KPTi*rYBvACD`%)~mB7hl<<`6f`7cVb{fgw=(wxzYD53N@KCm+Oq
zECT=8q8GZ%;}b+3E>j=mpPqXqXQaO7qT}>#c=7WFhOtJ$mQ3l?Hkua?XF9(xaFqdL
z_U4a9>Ri(_U0=ZbYmR4k9f_OY7Vr>D>^i#X>ywcI1Ezh55sr|&o}nE#nGRyfOL337
zmGdhDN?NY^fT(8`EfQF~YA-wd_^@w-cw31SMa;wt8`xG|bbMLth)gPAC-|}lb1&JX
z4J;4|Wev##2aLy(WX5yR^{FB6!XrgH%%hp9!euVyA}MLZ9&cN$Y<||;a`3<6O_CFj
zHCv62i`zVOj#TcDvui}AKA=LQOjqIvsN8tj`PWW|zp!R4Io4_Q%Ad&Oj7i;1OnrWw
z_wcU`Jns4;9TJBzTJK;%@PM37zmr7&-(3EXuEw8kePsnh6cL>$w_O<fzghmFcr0(Q
z8%W!)<&)nJ%C6b+5P>)lM99s>`|}sZKQ=htniv)q78fUlwCLgqGBWpq^KuQ`lV9%_
z!b}i=3wFjvz;lsAh<mfW_#L)R#+Bqb&Br&s<LbPEyu`+K*ce_+cH3*gvzz}ncG#<O
zY5R*^4ZvWqAOg-0=%p}AgU-Os$s$5yqkkfTo{&ySO>J##;w;yjxCG*0S4kFy*CI58
z2nhY_09yE>6O%&@1iV(G)VM6C72v9=sptszo_(xnbHr)8@~)o#HV~h^UlyC?2}c*e
zdG4(>9l*43b=}KcXdh<(WtEN(uWn8rEp(y&ou!aJi$N#n-<Wmr)~$`=o!CEzk&GTa
zxepbwA>{oZ$)EfNa9dAkgZ1Bo@~<#bNzHdz4c1Hep5Jdmzpr(qEI?p?o%*F*&e#qm
zLd4%db}_thQm9@rQ*CA8sPV7+y63WRTg{B`KoE}qxkf>MvV0ep+H1n)pSLSpR55&t
zHo~8EoctRp^{`WGzy9@n`?GyG)cNpJ&{K*??e)KR_Y`W@Si{PV9$$TVD2jAHn!-n-
zuzxR)9OiKC|M&k(TQmlNb%TFnVT2+CC(ZKB-`5eB2IFJ=jf;?KtJ!KxSovfdMsM-)
z-xt2na#@W+Bj8}+$NuI1_j0xw?q0-Agn#1CzadQz`Bnedhxc3YFKNf{t&@ZuZ0kQ0
zLgskJ4x+wG?6ZmuUXMR|hw4!GJ-CwZ^OPe$23Y`QiJySje)f3iImz{z+wD=G<{{<5
zabeDaplu5k@0C+Gw}6YT6TEjS+o~8Yx8PPk>$CinOJs@&0Riu;f@9L_2ThR$XI|%<
zmE1xLw$kJfKR4*?!dkVY_0@9I`8jXd)>HYz{QWsGQi9`k=v?gKZjJAV1xTo-dc^Ds
z5${4^Veq4+WL)b72_&-i$-On?B3mMevs@z5T5?y?Tqo!LI{91c$waub1i6*-L|ls=
z%e8HKY^XbmcWPUakvfhz>nGQ6*J4Vkdih8rm|jLY4RyHQyo3XsR#45Ugd7ib>wros
z{KX-a2!?&)yBEh)%f!<!bzeLt!_-~kkqggsme1VF8KG%vw)H8qqjek*^N>n8&PimB
z`+!fqB!lVDYg}>Is6mE>Y0Q-2xs6`5ZDM(M3UxizUTA|iUJbI%4j(32HAKJgO~Pp9
z-6<b>{bETW2^2zY#DZa584P_SA(C*9cj1{OAR~;dA8*RAoEmsY@CJ`L6<cRiaw;e(
z8DuKv7hVQZPu~r%%W$;R7a%LS|D1d&LBQD;Ln$?C#Q`$F8C4L#X{2OtUy8d5IQ)L|
z3gb{u<M?euIYXaLiCPWwW-%((O9zkFqv4#RrrrQSy0NpdO(WYO*><d*>u4K7_ilTd
z66*az6X=PHBHwv@P@!6{MrVykNFelluyq4M|8<TAEhFQ=z`(>|@E57@YftFx?xcQY
zdV1@wBY%<{@%rTWxS^5JCQn&?{fu#ik+Cs>+4%bLlZW5@vZ4h*J*cdrvbw&$%`>%s
z4iaX~_LqHjlfe}L)N-dxL&zv7#C%7Nj*bco89W9{_<cb|y5ra&v_54uzJ2=4J2WD~
zAag{I1=Oq=LouM3ZCgaj<F5Lr+LXvR`3VB>iL(gCTSfaiEzIZDv%Bi*>LbPOXFIUO
zL?VyDeNgaDFamfjm!p(&yv?K<(nTTl!O^igOakPfsqKNjzOgUGQ`-)8S{nBLutWy^
zA@93qZPbCA^W)-3WM^mMPlRpg&>y2)&CI6e3i97PMfh5e3=#m9NXLp6+6+fu878?~
zf#K~&<9bS>BvP5~=VM*T^Lh`{o}JAxsf5Lv^U<-#0De`1nx5w8GrKnWTX!r#h9#1C
zd1iOnc=px)$#!Fw+c^T@3$dM({$XF%mcU1y|Ed2!-s82ufat&Y=f6nie?VHrzt>P>
zP1({mhWr~+WWJL8gR1_+-^+YO;QU`O>%UL^4`4D}(qMYBC2$`%S@7w4(0#?0TRt+W
z{6&RdFHrGVewTHTu!GohkJJ!;85*h&0_zwY<O0y>@1g&^?0>|9f1cII@+|j0JpO0Y
z{}LRTFQ)Y}{|SNr_Z|O>6o_cyGfw>rJO6tl7}no_`@f#}FV6gzPx;!TSNBrkZ;1WB
zpZ|`kH2B|z`kxbyN`<aXI#S`H%`e5nmTTX<O5KkKuG&>8p&l)+h$<h51Bj6e_ZB^c
zuDTYz((G&r_|N)B>h(d-ykGgJ?>*k;2M|s@`ZP(#qsMp*ws+zzz&DuQOzEE<Y)2(^
zbm4xFxjscvf?)F6blh2Y)JhrSF$h8Oag(Ow`{y8f5xq~+#en!d4M^WOzsk^7k2Wfq
zDj2$$F!Mu+r}P+1d1sW{;6H2?DYX1kIdidj<ugyBT+SsS5I1%meZBCJ9`gj-@euJb
zZ5&^KYSG~jz?Ys-;ZLsS;uw&?z`#({iFcp~Q&o))2BlXu%rY)6PJ$pSK5buXUc&5q
z?!FcjK6KnTU)<Hyq+?@joE_}{@nwe|OKezJS{sd^BJiR?eW~#Oanda&)p=0arPBD<
zC&3p#9|(}{|5Sd(N#MYj1Z;;lcJen<4E<7zz)*u$v8fvR<A>HYnkmO_0wf*|0W|*q
zr(cG*G`~c8T#0Le1XvnSy^PS_mn#2nbNYX7-Uvj!c#iWfN^;+0>sSn0c<Vl|2u=I>
z5aFx+Lz|d~=Hw_9Cw&bWKcFf*yrstL>WAayP*l8A#j<2=P<|mGVBlhYSEsv$=OpvP
z6OMe-ErV^?EidQFuv6T>jBjM#k7ZMGg`{GB@XF)XZtS*wi|)BdED#Y8-U(sWSXFQm
zJ33tz*jyd&1&T#oqbyf@lpnRmsB&@hnSpEGTw=Nit7tAK2>+>`U$4Ni5?ZRN@i1+%
z^Wj^~=mH+U=Ht-sknGpWpth@7VQiCCFhCVk1!9GmMNG8AhMUV3k-^w~KF=f$`(F3y
z{OBSDHvmWp%EU9J?lPUECXC;KcfDczWlNh%eS~LsOePQIetR+1aM9!W{Q7gW7QDUG
z@b4UcMkhhp2DgV3kz5zPqJbxVt@WEWhz|BVJXz5dgF%q8X4OcDcxsZudhSKQ(FniR
z{#6GemVTX!Z5`dj<zT4O{#c8ZVfp>7CzQ&u0m6jB`g>`mrKeZAb4R@eZJGp;G~lG-
zFWWpm`CJxUExzky@BC7aGArJ7Y`jQvaPCzQGWOaImte7_7!ujH3g0u*JS0H8*IhPD
zWfPkor!-Uq$&-TXLRH@`T*zE+23Zt)%qk64l%#J`4bV9fS>Cqa5fCH-!u=f<fJ?bQ
zj~w%;?v_b(s;xUv9$j4<wQMmSOBf|#Q^@du#~|Rna3HP-TCQz8o{hGi$M4^R(VOhn
zK=I-j?gu0Iv+-E1fQ%wh?YZ09sF7M?&DLgD3RG|QO!YZ8*MIk~WQX8tKbv`*xO-6e
zF6&$xert0$PYQPsOCFKBk8GnMerc8${Zk>j4j0|f>zh@W@Fd~UpOvn=;i5Nv3rVyA
z@O_*4`YIa#Et~i6<&#z-<b^IaGD0@iH^*}J=jz5-#_ZMyi8;-c*9Vi(DmG4CuTHmf
zbSy7Vr*uJ;&*y>z1<a*(gX2H~xBSK0pTp%I_H)2z{&1)0I~vWBi7*+_+G{)>oYM8W
zA-((alxzi@OcwHR1vg)N-f7x28M%K7Zwhv~{&T2P5`%G+BYa~9Z}xzM0x~Lv&w<Vb
zrw#M5ek%l7LPL4F(kz13apLW`re*>^&d#0LqAv6LWHZN3c=|Udp{sp|uB~!~i<{dv
z-{(&sKDgZX$EJ5Z-)k}+h+k2EDSkL(QF60WF|+}-m@LE>a4vmNynpJtK9u6s$5`_|
zo`V)vEjr!kay*#Cf4&bFc01c~IbORtXnX!6&U3%%e62t3csWG4w_E_6+1lDN)ZAHU
zy4sp3n2>E+Pj<83ovE@c0#tReR$Vy82Q97Z1-E@rE$zvBU5-_njUkcoJMPpj`O{Cw
zEB*QH^Sl$B11!>N!tKqK@KHB~&(r5};o=n-;)h^?n3$R24Nm*zMt!QqTD<Om4jw&4
zcfY$i2h~kHztdjrm*v4^K?_GV{DXEZ`N)_UEYZFC?P9RTO_h}=!)ao0_v3-X<t}kW
z_sh+k7zXulfo;ud%QQxW?|@TN)AdlnfUMwCLHyElm9HS7plY(2E=@BTO75xFq*vr`
zb+?|q^>SURa)Kw=T4bOp?YV9&Ka!yC!?SKWoh)g-^v58VVq8qr0cQHQ7vKfi3Rs!!
zRmBe<%0YF|MdHTbl5H{&Uj>-s3X>t%riBJDzUo<P&DziUW@ek<g(m&6VW<IKf7(dt
z=;#1NEOfb%@dO2jh)2)#+h9DW#Z0AHI3=TQ^L5*E;lkWpg#aAQ`%6Lv7?_x{Nl9E*
z!X!yOfF{_iZ#}+zIOn_^1Z*(tZgjA$UXw!V#j(zy6Wrq*@a67jFq8Ru2Q2Zj+S=*;
zbFbTr^$f|7g({1f`cD1swSk1YbLbtWsE+4_(WNX4-t}I?{$z<RDbi!dy*ZaZ2l{Ig
zH=cL7iW%uQ8cIs+UzXYr<lL9Qk`E>F9s?hUF;aio0TvaQii(=rP_m$Aw9!m!MwMj#
zam!QEm$<lq_%hs^t4pWxPyYj+O(N_mfQQ<AGAbhz!z7!pDw`}2clN<u2;SoJRINm3
z9_r;K+MO;QfKJR8i5j5md9m6>>Lp;e){jXcK|}5NXVG`LE2OX7^0L!B>^scK{w6Um
zB(S#e#>HdrYN!Voo5$~l*L)xfJ>mglMPY6%{j%Jd;(7TKgUxYgIz0S^=sR)oCuEd)
zH&fX3su-fT7eijj;8MWEr3P86M}skVfQ6&VkiI$UWA5wi1>4ix*GDBqM@!4=dO~AI
zPj-DWCePe>&<aMJ;&G-f>Nm9R_WRpY0g|CSrJQiJBr90?DzITyr+}!g9Z-^2{xGtx
zY;1HNNf&=Dm-q_8pi#NeAIA>C2<cMN(75blh9Y@~2IhE76>9^V_XHaRk~gIkpl^Vg
zjZPM6hL8xxXZEhH#^~NI1u!F#Nuw({&6nywA~W0?&%@QW$2_KSb#ORM<h2`jS^)$}
z=jup?q^|p6M^3V<f03eScTdFYJO`<8E~}YJ*Hd%Rcc3({lPy3QqyIwPUET-;46wM$
zs;VzUDW%d!EJZL(MP5tAf^EM%-4-Ny?m>sktm8f;c&@ChJe<7?hJFVYjk;|wn3$hh
zF41|VJ8Ve!t`WEoy=oy3*s5las|~C8g=v49{-wo^{k#2-0gENYb2VPOiL}q>$1~y@
zn2J4OD9wOb7O0h|l5E;QZ=x|owJGzAlvGs2kzjzP0ooOfzjCaMwDexxI!QPLb5GFi
z3{);X<wU56Kp11_yZ7;el4fbIByO7p;0QgF<VHX<y5AXxDeW4QAreu)scr1X`XDfM
zg^<eYoPF>}=Mz!Pm&*_Cmz4?UaM)6|9W((JhI@G*UqgU@`C(G17HLEUUS-MBE56*6
z9hpS9;l@DfF4y|-q19babMum(vuZ{g;4lXV&Z0RO0oD3UFUSw`b$083K{*6*ZB$pS
zNTXviv9+jXL3+@LHs~bWZZQ-nMl*99$%cWfedX6R&>Jg27M<wctAV6skZM{fTZqg4
zLeu@8O_v!w5lIyXvvk^-{<7AOAJQd;1PVSu+Z80u3|)`Y3EjI(HPP-egD%rCu>VR=
zpOa)_Q#1|m+O3sx5-?+aj}yMzb24c<I~ZuvFq{+@R8=6KwoeVlC<i+<Iyx#IfD>l;
z>==aIsCXIeJWz@$IvFzcudc2tM|iiq@>oD33&3T*-S@haK~jf$-n9S3f@x$9Q=>Q0
ziXtTuy-hhbL^%edG<$rt1FHb@$Pn&X*Rv?g&R(mU(owoqvEBvSl8!IuR%sa&ZU>Vo
zkjn}X`yO$&rQ@=Sy+%RN8>Fy}^;*vTsONQ*?o^4ce7HjGi8r9l8eSoRwcG_+1%ixa
zDh5d$3p6~@GE;g*DFpD1&f1Un@C$m%^T}fEf|J-ydFXYkCw+j~Shj3DM_zq>JmL^C
zSvn2Kt@W9i5LKlFZX08r!j6s(xAVO$67l^oj1Q`+6412fOpez6bUFFzVl<5xAfbHt
zlu*vFKU1~6T>{Nc0U~%@V9w3ub_u-P2<4_QSwsx`JEchMCfewwM%^E<YGyUz!O^sn
z>X@jAW*Z?~G83eclvIx{O^Nv(3zc%@!YOG7h{G#MDr}igdV5a=NkS;4q*o$}6g!xb
zex<+9j-dQWkR@_OX_on^skYV`WYyl1O0TxL4UiC#$$&pBG`dvWR|t0-BN>3T%7Rie
z$G$IVG@r=FYq{XV>Q%kF8j}|u@du2_XiBWSHYK1=PmYI#)zs8bybnM$5di6RtD4%*
z?xo_(sX}!z3qA2zAobDmowk4doQLzO|E;#csC1BLw2{lNuMc`+`IwQs+5E3>ulDKT
z^do9(t%XQGULkS#%VCOt*&NNPjLEo7Mt+J;<h-8b)NNz;F)cg$M-lH6b0$1gRW&t`
zIYpA<sHmu<#oI6(+JbYF%}b_aIBfyj21YL%t}TKAaL&E8F-IwI9UuGxQ7&>48|_BC
zur+qd0%$#Hq^9(~o;c!R#k$fGRJWakHqHXc^<?!LNOfaxUMXm3NGlV>1s0}?ZysMf
z7U2PC07N=KoAQQN9$Na-_~*zarLz@6o|`Gjv!dy9MI6lOmVN5roYKm;)=*XLbR?Ms
z@o)3(i&es`9tmYKI3@BiSZgD7Re5tP@k5Z6Gv+a?wd<=arkHe_1v(#8O3NZ{5h5e{
zY-LYm%O;pAr=-<!!`&mQE#JkmT)7M7hXE@!G7UQ_@w(Xt>rJU&_LQJkr`%AA&HD^k
zfXY!tVc|}97zJhlOSL1FRQeb@GTD~M1NCU=<rX~^-LJ_FyCFf`-MuY55G?Di_8Qv0
zgY-G@ma4y}`<ce^F#033Ph<Mtna?Wt42zH|j?jt(!*PekkRTV_i9#M4ic1l%<wx_5
zQ5oeDc@FQRL8S%V!ml~z%_!iFgtmuSGBKb_VAy@S$^us`064SfMw;&@qTlCN5oHMW
zJ>gW_lSN9mK(T0yYOD5Fud`h}hkKPYm)m*}d{xZeFaJL1`w9{n6_p3g5nKbglNKMP
zQnZ{&BRA0seJF`vEv^;k#flb~Z<Fh3IO}y><|Y!(3Wx=ZMJiNmit6fdjU?>accSaf
zx!Y8yCTos0R<k%Gp)b71H&_#7ocEL@74Lc{8FTN%=^a+x)|SIY81@#LG|kvYqsUMh
zJ015xaHupL@lF%(-K~Kazdt31SkBe53gB1L|9U1B9+^kJ=n(*tH>KI5XQ#@?p(n#O
z0wnSvARiAFT*6rmIw2e=<RsCrwj2}|$2Riw^F`0H^}HSR<s1&PL(FSg#%zL#c*6zs
zRqM$?9UjlZF4+`nd%ibMCc)Zi$(By@L%s>+?uQlf$kSKx0rX?^u_zVM1H01_HkB(0
zFX~O+byEl$^+q0CeQJo6j5f;pE$Z422?ZS%SsXA`tC3<=V6X+rTU+L9G81Bf#Tr%q
z#of=p?J8lhafvLan1`1|5fTf@JB&<-wp06!BDh|x#%%Jmf^Z%yFQ$0CveFX)BD!+3
zd@>8bA@f@#e)V*z);~xpdH`*|ww~fue9H1eehq;18XGuC27R8TLWgRs+sQ^+KdS$?
z!`Am*y7&ZWHH4^$6B&PAR!bPEFEo2J9L3>HeD^nA1@%cU$HQvHP&^4SJ;DeDVRWEw
zD62}*QX14nwxsRUM$43GcM+A<q>^N&qHO@YROQGfl4l26EhduESH73yGHKxz&E&Z8
z4!gDERhTh+EPd0bk2?S9;%K$OaaW_s)d3izcZ))-xZ)hBR-&_u!H|&Z&KKP938?f}
z9RLq_Fonv3yB0ds$x&1FQJ>%DO@^W<(yT|+#45JpQG{a9C`m=<8I|3(#0$mPiQ`={
z9AXwqjTTzFn^>`&Sn=zh=vyP-SWaSP#g^OptBs_Q)GJLDi?Z>bRh)e;qiror_6wtI
zSllk3mUC5e^P}+xr9&W&A3JrVc2^+CAOW>@wru1no9q`d(X0E=I~_scxNAoXr&B|T
z>g)_{7W@X7lDPwRy`+B93uAa$QD5_)rEM#!6V?|)JGaIoG4MTFPF@;}CqaDerI1J&
zT5PnTW6Ti4k0YkKOqQ7}{6xWqcvNAOWcR++RoJYg9v#E1o19rA3DNU6#?#dYY2v*g
z75h>~7^;3W<xgwl*6UtXvG7b_^bmoch$pe6Ue{mNG;1cMi?Gpyauam+`T>ceA#8U^
zpqVgvl<(D8Q6_=oc{@a?8_8|1z~|K~*$|kUvm_2yqp|C~a%*B-9R5{z`^W;nA%7|H
z)dL0VxvyN#e>}>Hy4nQWV;|5dL14}qWKRghV&l3HL2W8GR%6WZ)$bx3XN~eScD+9?
zBMZ5&`lE#h*Y-@I<>q)xgzVJ`ef=MQAz8U=TZw6?C)`C0(P^L6@WADYo1{`BNBzoO
zrdoveQUtRP5L$+W6HG5am1bxZIlcyy&i(b3#Lih|aMH9^d@Vwa;*<M4nSXkx2Y$ri
zxB_``p!5ftgp06A%5(}^H_3=nViR@DFKOE4c@rud3kccrx8)V}81Q^u_*PsAv_bZ^
zW9<BP?I?|{S;T{oEYs7Y`MU6(AD^jA4IOHoH;q40*CzTSR{@gUF@madbO6*UO-pI~
zPi`I)FlPlW(VSb?NDg|1bYTscvfu?;$Ync*9bCAN1de*REX}-ScjTpOWk3?G>>KS=
zYUFn^-(;N{A8&8A)Y`6XB|Qb<UKcYYFsv<{^1{8jHc@xE<=dNA5@qIz@VAGqO$6|D
zyf)2T#oZ19Tf_F4ve81nMaC1Z<s+)ykdJ{mQKlp^&mKSAGWo{m#et0}KC*u<`Hw6x
zWjr>s`3zq>duO(WS)dhnS<WncX_P%^AP-OF@w^Uq@u!K{Ry>#tye7>?QePwN$981f
z;%C03Et}ZW*p6_k+4f+yv;C<%?_beHU0xBZ<*S`$mHY~9qOfPZ3Km6@TM}+2-WUG6
zcu?xRu12T*h0plt@;w0D<2SSLTwhAflYFxUIE2$;l5$oWDZBl=Qs8UPG)ng$&r<ci
zEOH&`-L_t$%@s!ZkZ&z%mj0V*ma3pH$o2fW?fI8^{*?$s!Ddg35Sx$RDofszqe-d9
zX?@o3CVE{nOYHgc&8uHuv^a|VgND}o`uoQh-dzKX%t5kg&j%V)$Zsi;L?t97-pRa1
zT)bH}<tA*)W$5EdV4V4F-!5Y&Flt;O&MwVwH{u!OS3a-e>FN3BP10qjfnmujq%ZiO
zR)lsw9a{Vh)D<8cfT!!on4j~-^kLs`Zhv?8@ZPoX^}(YgAq{%gJ>_u9hIx`h{XoaY
zP70Ugkg%pN*abLhIo}7rXa(+Ii?cH6jut<)uHj#bG8<CKQ1ornubGYVF`MdLwG|hL
zUBAG)=R)VKohnVWcrZgy<3`^J8yXtEWd}0<*k&LX7w+LD0jY{ZHRKI^A-n10rgfD@
zaG)YS;=>2PAiV!kG|W-K$}!~_Na_!)Z=YXR-Uc`viQ}%+?AzU@b?EM2e%TDCAZE6}
zbvY_?G4(Q0{-Bxd*dUmWfCo1K8m%Sa&))LrA83Shy66NPexC0>gs5LJQZ}?<qR-as
z#<M+?uHcT>pzO(wmdVSpA~dV9R<H<B|E7-}7Fd%y>3^0+KIYcnFAA&15qmucE3+LM
zQc2n)5YGHQu%N$EYoQG#8f@Ha!Q71tGOTEAca=r!`huzUp4A>DsCOz~g}%BuH&6yB
zNA$X&kc{+H*qiD8;}Y(-K}OW|-Ttj}vV1->K7uUt3=dT-qfDwZKj8#(&9C|8m-r`b
z0*WYi6Kc8xTjgcQurw}-p@~RGAA%p-v2}&2g2hdTofh>q-783BkP`S|?{-PUlJ~;`
zhnSy65SVercO*M~{j|14meeky4^8D!XzI~X2M(ZEMke#PI-|-bLlv(K7ZjO6ASVE#
z(UT`nwrnaT8yyH`xhhN38JlxT(NHmtjU^KnOxObboCv;1d<=K(2TDF`u;Ir-GQqV2
z>g=Va-T<uua#^R_az7T|^H@nxaqIm;c;wm-boebVpsR^xfac2H0Q60?Gr$W1kk=cb
z=NiHbOHzyJn*el$=glQ8ycB!Q)Twt;8tG&glZ+a4d8FRP1hWwcfB2PX>e!}9lDAM3
zoa9;M%IdBWb^b7PBS0w@%Qk0`VyQ>76Q3Dlt(!(kE?lDnf>JSHTmiELxj-ymEqOhn
ztP7v_eEd+rYcnf`jyoZf#4b>Cm^)7a1yc|*_-to9FXAZZLD~)gwdY$CGBQUeB5u57
zB(u#PZldoBREsvCGdXvr?dGkPvHo~JmSn&Zk?lGh&F(R@l`B9o-V4q|DUZv``XHi7
zgpd&1e*GX5yD;s|a@T7Vqn&;01j_ppvG5iD7bLxFJcGs+wDSXocv+BXqgex06J}1s
znprEZgb#W6aDz!UBPr4mNL+PUq|GPEuW@)_Kct8evL!Zy5&`dYBBSwz9#-HH?4%_D
zBZ=$I3ZH6aM_Dxn^%-$-m^oDTgVA)F*i*(H8KToHTU*<WtKXY$_NK4e`HGm3j!RGp
zx!&I}J4k}}eWpp6TmxLnyJW?V&>0liS*#c&Zrvq#d7RtAJ%smdY=rhtCH-u9{wR~`
zR-!{W>;Nh61==R!EDG@g+pIUo$H$-<BXSGC9Y~ziGb1J4t9QaDd-MErG81shJhs3;
zrnOqdnS*M-U6OIAl2Al(Lq6tAJkh_!)}WN8e9g=}v}q3-+o;@*)coj&Nh%(k#t#@<
z0D8yMQ|R;-G{{jRv+vR?4JahPsiw=qSjG`0C-32k4+|3s2v4-YDzyrfOn7Gfm1V44
z$@>BH?q(mXK+#;%?C$Nh?p=H%2ZpGvon3^h#|x!ziE^*4#&z6wt9s;qu+pHr&sx<4
z5%Zsg%P=Y4<p*DuCS8r@s}uw<)%(58L&IUzcuArr9~w%=rE&gY`-E8(MvWyKD;Y!}
zez)p!n-L;p4w&Wgu2@_~jZBhg1D%>qBqKncxKEerPiff=EOi8cYojRjvlm_SKY#|e
z%)v|*lO4dPMhzh{&}Iv+eS7qDZGC<0p!|VBKk)hh9J`FlY}HR|jGEPrE8#Mc4~<Ar
zm2;r730#m+KzRlf{hXE%+W-aq^~JGEGIS$7Ae?VvkR>u<U}(tepyes3(Ti_Yv{Nm`
zV8AK?BTTI?6af_Gn%SqQc%eHab~l%&*xQvfn7!3RTa!ShU9JtIkTf<lq;v@6WoNGd
z73Nk~f+JIevQLdA?5GumSs7LF&=lT$lPcmfJ;;Ny&*}>)-+a3EoTTad+?*Dl%11#P
z@p(EEQzt+^*<&&meR?@(%{_DssAG=+97fX+*jHBL;AP|f*p^}Gw;3lA)Uxs8#oGLr
zC9WUaH|?oy+TQVXFwEFLFo^!ipEL;CL+j6Rh%?_iL@rWZYUSq80v<gN^dt8|E$u$y
zyCA!8Y)zHOcS&xP6b7oQsC;i9k_^Nv13;PJA~lNTet(L(=0HUZ7B<{YC@iy#=zBDy
z<_l2*PV-FnjCNu5G&vyeXOn<^y8<x=tWfYW58s`DOY0oym;q7+WjYIc<Cq@HA3*Mp
zXm<DYWoQ=X7vO=mus{t~%S(Z?X#@M$iwjL|1yJbOj0GT6jm%@tj`5z^0TR%O&-G-3
zNxPl~?z?Z^iW6@({#^s~wlujpi6Wu-(l|!&(I>2t*<mfVPwz!?fe=45IGD{o|Nglf
zwO=_dh6mqxf-K!U;!LW^tjwZIa`66WE}NAE=x?K8W|kKu3%~|3MWf0*>ZqhVFoJ6-
z@6l`8@Eo5?thIZLd$KXY0XGYhLaJL1<~RLFAuR0G?uSW<+<F}Yy<P<r2iiq=^N$Un
zZD-us62+vXq%=M$k;bY5&D5ZKngr+<1I8IqD4P0Uay3V=#m}h3e3hxGsimc%>lZ>E
zmz9>V56Glty$fSG^1_GhSU`SAjmf%-%bkS^3~>Vq1Vnkr<|811Z2E@6y1#$_4oo7$
zu`7sBGFGAoY=HerH>T)2(6t*08+kjCZ^sAva*xkuZGc?FNuRNzWpxU~q8Fmh$@>7d
z4ISV2Ysp9H30(;qQZxQ!J4~>8LvFbI`c%s(VPFz$R?ULzHfUIM0emJN0cQY)zUOif
zceZ>=vAdX{zoXi8@O_&iU$T`j(1-vR<HSQWo;3Ze0=^HgvC*vn8V;w0gfhCAO(>tr
zY_-+Z$!6PduJx!~EMTna067F^i6t7jNk}0XoN1lLLvE)pWV>?D{7%4!_1F|I;DFG(
z0-6Z7>F_J5d3z%SO~nnsV2|A#Og}pYb+()^4-tBN%K5w#9FR#VQBe*f;<yt9Y9ZCZ
z_{F<_n<L}fz^5S<Fhn$h@oFngSA_ZK#2MKiEWVFh#xpjh4HBiwo_Q<i=y-y<HO%f2
z>03^YH+MJ7q<0&id4vEb1?b}PAUfbRC``4#9D~7QfGk9v+11ZA$N`#{SC83;1eQPZ
zY|IKpa&=*=X^B)eZHw~vR=i=a27caqx@oxpcx}+)QY|6dvVTs0r9ZNsuG8e&n{S`h
znaVi<W}^;M@hvAte;On<JVewr693JLMl*W#I|{C`k<kcK`RJV@iXC8DbX_;8sMlUl
z?n2@)z#?H1jBU)-vETfVQ0EP%XJE)=%H80guThXxq>)w;GVBgbu&)1!0hAnhTkA_9
z`bfHLqrT`Z*F?ViC}WBmoyL*^6a+wk*cX<G_#kc}`~r(82*`Hev~sa5-!Ry|EWQ09
zpTy7y7+9<6QZkR<^Va7#=kU8h{{7)cz~x-#9i~RwOSkf_Zai1DiGeMiz(DH^Xt)lu
zy0rP@?Jz%V9*Gm3-aaIB#c(sV>*%yKP7C_Z@iYFEGpw(z0cqnfD@G?Ds24z{+<a^q
zyc`)^-dwZS{<S&{nGCcL0-7{vY}3AB9@sFD$}-?%)+c}rz^_K+vRQa9Jba6VrZr|W
ze%r6W3))ROo<ZTdP*0E5s9ouY<D?)Rft+bp)cyHO26_K|ssJ5~xs8KeoAz`laj_<T
zuaLMBc7ZlQ4WU8f7muB-G-Ln#`_0U>D9x<tIB)Sdb>#v`BW<B%V#UpuTL7qx9Z-nX
zJwpyY-JMl)NwEeTE?^I4iZr?T%3E<(J^Af7KJQjd$;GjWH?^yN)5lr?=&`7{IGa&7
zrY{SJGI<?zO#$DL^DShm{r!*{r-wJ~$XTFwR=B}oTY{S;=P8H9q$SYMKw<%%A;^z(
z0J9EQJ1Pi=*=YOXXPO$&ODd&H-lG~zEA9JZftpzI?y!qA>ekowQq=qY^NQX`+VMHC
zK)a4t00y0bqUD9CC6}h+{U6LcE4TqM&8K0I7`7BEm2^ISoCcQAu-}}(WfFCs!Lk^o
zrjtg>#>NK91u(TJ0F#X|JD(Kf0NoJ~L-oipm$G5f%RbAsKsdRMGf7TkT{nDxYYy@-
zz-z`pMFUm|GS7XB=+1N*h-i@Avv{8ShTdA?n@xF0=<y(bh^LfIH8B4v01tp}5zlQS
zXTY-dDGIijoeKG2#ieO&&7^Jq;p0c=ooQC8)PYi&9QQSzVnF}L=(y$`0op=bT--j9
zUy6MYbYAki9Ib$sFFpts+e*)H?vbrl@bZHDu`HiB&12uTu!Mwy_n7iTbpF5-t;Vv^
z7#jcqH@$lXRNHFPk#s<~Po~KW-`E4|0~};uyJ<Iz2te%v#to~{n&Z#UOj10eBO=6n
zn%>)fWbCi2s{?ZRP;YO#dD{C=3NsH-eZ_^Yb}9gjc&mH~7`$oA>c5)MC{TbYc}v;c
ziwiNK;7MT1Z-Iz7v*aRH=a>P6TnE700T6MoPhYw>@9gY6C2{^LN1Pf<)nAxe14Q*N
zqTToSg{up=M|?tpCeW9TFTOn`QTSo`k<6cb=-9d@LBgrH>1+l#>wT00I0QV5L~sq<
zP=DA-2qQqlFQkNwm*_%)k_d=W;iTA6poik>IP3SvvZC&8y(;?7OHUslv_9Y9G}Pa(
z)!@i@BZS|FWu`KbBcGDQ?*xSKCb9wxrp5!&HP$zx-JtTm2d~2PI5v%kod7M2NW5WF
zNG~Mn0!4;+BXiD}3n<3{_om9pV|6MM9tk>qtnDhT=Ux9CESd}^D!SSNKKi)ne4&r-
zI@cYv3<<iPID<}QQnww}Q~F#p3!BkQsYL_)O9d1=@J)k6Fp(7Q!99L9+gMp!8~w;9
z76e3mhM%Kk*_z(+i9BBbSr2*!bZe}Mu>xL-$42FW4<2o$jE&cT!Yhx`B%dO5uT(#K
z_AIsr5r^U9L5SCNn{kE1?#w-(x9H1hHcFWl!!|~A1)QkIh7cb#!~Qu=xpO~SkuoX;
zZBX|*ZsE>g9Q%5jFUi(i-Lc^L92U}JQ?~Mo3Ngo*_-uyVfcM-C9BdCWs>{yqG_E+^
zfsI+$&@ab2Hr)Imv^Jf{kIFJe%}in`NLlFb?+4a#c6~gQBmSb<FG#;hg7nne+Z$w*
z#_OX#PEO7(U0+oE)^=$m-{+)2PfbB414txZz=e(T5{&S5TlB*LR=R%He8y@S#Hdg8
zTlobWmHN`lok2iL=p9yg3$lCDSwH*GTdn?MAde?a1gwnPP^FQjAY)O3-b!tU36<V+
zz``3Z#ASTLxNJDaHvLa8K(D{w+~+nl?zzv8$RKGf-~+bua<<F31x#&@GOs(&tQY&^
zd1LpWzsc>qoUPt&U5@SoA0HnqzZzm)+vZ%I9cb~|Tn?*8HdD{{a=kg75<?fdyFON*
z-vqrD@=1K+F#r*sh4w1Qf=s54%ue<;oHBSBNbo?g&(rZK?thzBigvL<u?tZTYFpfZ
zRphX}k5x968oRmvMB3BH`GzBqlZ{(CBTs6^l^wC|2bu-wePA^rD1SMC5MT}fp~zH=
zZ$w}_4aLhUlAhrhP*vQk#;fJM=l$Fk%Ev(3j*f|O+8F+}egR_k_Z^5z3`^g)Jm$C;
zmZ>`SEl+UUP0>DnJHcK9{_-74!R#`epAVvrRFM)nK@->?pkb=C1=6)%1LNX0AMaj1
ze9O;<&#1&m{6*C<DuBa$Ja_Vwk+m{xU2FLeIbAyzKfQg|4%CA?7LQVSqZ_Bf)Sq^6
z6L4E=S@f`_{74s%jT<_WiZXk!N#73~p}W67P;fKeMfeD`-DF^?8xR_OQe=$>I%gy1
z>gPx&jFBGDvP;#u=UA{qtgWTBW6y`?Sy$ib8WhE=1c@VX;2{DFOUUy%H7rOV$f>uq
z$*9qohL+p8G!duZaFOGav&9>3v)BF7!Gza%?Gri%v+p>jaid7sBMLeK!W!mBi7bGG
z_W|xrgythwctX6GjbsN(iw!OwiEujYg0~9wRAA9F+}!Ye0F$;I(W>+etNo=u*EH@#
z$s;z@U0Dn)G1z;{k#-Ar6f(^WNT}=S7HeIq#KsPt51~u8{TQukvh{%b$4tO}*#Cq+
zkU)m0KbmxCHZ&-;HJ$2K-fzr)8xkjlKE^*m5@s;_Y~*7}8Jo0jx{-8<`f^5Zuc<kj
z1hr^4!Iw6cFM+%&qA(0&AC{~xcaSB5B$PbDztbA@XGi9Fw;1%>yHE872`J47s^+#k
z7);iZxb|0tJGNR*DjE7~jU**b?02#v>_+9O14ghr{~U`@o-a@#7Oz(>ia-D}@)KL_
zuIExa@~2Us2)d4V$JRnULu@RyZa<eMobtYp<*eK<Ojm64YJVKWg`mW<g$<XdS^kJ6
zL}CpGWjUEW3V~93FOrcaF-3}F1qs5gPii)*B|9lDW#?I}*&A|MQJ7S{p5=Sk#M`>m
z&bALub;{k2Ic2Qk6!_s9N>4`>W9{thqRvSOeAHXr0mfSn>YP~AyUN9V`4SJmM%Zo*
zRVqP&>q**9Jw*L9{3k0ih1z#2(HZ0lhCW=a&);Tuxu_zoJ~v6|T3VOd#`4f7aMhU$
z23pO+Bc%skHUM5I`#qfV3B7N*EdD2hU|NZ)M)S#MR&3jRx*wz>NGi@?QrZ2zTt<wK
z#g^i)jZ_Px(nbJ`?75#~rfNSs=2Cpgueoosc`+P9MR*2WPr=tc{w%uoBjP{sRalJl
zW+Pn($`h{RjAE>4d6MLOEI#jARQ3}*sA_#nHlL|aGuyoy`7_k9_5Evg)0f6;>}Hz@
z7D24}p+aevN_1l<$VU7`=sp2a%@2_b`lZOx#K11gEZf!vx*znrBD{y~bhZ;a783Ou
z2OplsN5@2$cSLH-(rmu!U;9xmn@1j);*%aSZYsp`nu4M&lB9yxT3<Q@)@>xyKK?sa
zomGB$y(t8bRkW`G7k?pdXq{rlmDhMc?cL_;2Q%#6Yzd{Tu0(=7!{APJSfN68L}ptz
zeUrM9POag!07Fcl1Nb(DtxmZSukk}721ry9pv8r~QLpH+kBDgZ>lF7-nTpaN1|CYF
z{iseuK|uj%Qfa%TUo%>x^EB2#)tz2vlqphHGaC`5g4V|61Iy|FzQo?IEIoI^MgWIE
zdl2_D&5fS>`_@0kMz1hlK}^HY*6^-XWtCE!$FpQ`m+g3%nM(jpM<k^BA*pL00T$5(
za1j7Gs&4B8z}R`d{+{TuY=lC!V}NDK82}Za?5L<G(8ODy-DM?&FWB-W-8a01Ni2I?
zH5p33s{<wJCQETn0tcE=2mxpBJ3H#&ub^5CVn#Loe$QtQ=;>$UrL-B#V)hOWP96yv
z(^uYLi8QcOG>t)TPfGyg&<K9NNyKGT;k0Vr$&CbnCTL6u<j7Cpz&7^zdEU{43IjcF
z9k{w$+xq=#7-ske#yWW*{i*=%jh_g>ruWx;hXkuUgR8c2pQiLOI3lf9-@KC+OL?d?
zGaMp(>kPULQMt@(W(m10KW^)*np$Qhv2>|GOBvI^l7SC&qm3#Sin0%lVWXh<_hF1@
z6VC;71L&D!T_~2A-isPIc8~AQq~)-j{@fXeFQJjy`~~DVQ`ls(l_rbI(ln)2S1j5O
z3B_n;U28|t^wu+RdC+#V4XUaF6mt%MFJ}Q&LJ`#YWL824;KEJ{S(~fZ-VA*u(A)Y$
zP22bhM{Yd6_~I||3F;u?#(X+R)ggZq+fu^fzLo5}gv^-n$CC6EH2Ktlq8+T&X)gy8
zjHQD~3>3xLtEG`lN}qJjvavyxJZc+@?M}N3ffatg&K7Xkw5#}nZxkiEt(J|tjk}F&
zAx+5n1j~t&wWGLm0K4!v!df7!Ta%7ajPw)5m%PClV}ka-51lJlvWM2O#KZPlWRO7R
z`?BMpUE#6F!PQM~%OdjGg}eQdX^|PlI@fL1W1kk&81U(Rnb*WZY6NA^`Y~;c{}=U?
zq1RF$Ds`WTD41wizExB<(Ll{uguMuNR*&W+0GD$m$nwTseh4g+0#X!EWc90Fgqw}W
zq&NM4mAz$DR&Vqz`U*%&NC^mnlynP7x1==EDP2-ZNC`>_C>_$>(j}$RAl)V1rGSLM
zS^WO@j63d^JI>*Q3ch>u>?hZnYtFd~IB4}G(&=*A&!B81j%Rw!B&Wu+9%-eU#om5E
zY@^Z0ZtZFe*L>I8u($b}18jJ57WB7N=tEl|sfKFT2WDFy5cQ*otKj7RWJj0y5ny(K
zuM`#M?@=~!KgxQGClD`+^=@?b>f@9n5}>iOFh-~rA0=@{WjBQyYmyFI>Sqn%SxJ-a
zDS*-hbRH2LjdObk?8yFzPN(x+bcnJ0`V?84H=OHtUU_M05nn!!qHS5A<P=`f_hCc}
zqPi*PypPi&DN!7oC@3fhd8Grw;AbZ2u64T0pJ^ImL!ApqWFU8g8D3>nU?Fw!7$vzs
z4&vK(hLU@xH+3DdZ~<~0xAV&zUGoqx0eW}bfd*(;Y^+!6_1RL_W4D7dN$#M+EvO&#
z{rTFv8fp=fn%{o5InPh#-M#pdpFF~<MLr5p>8vNd{6MAG4ISHk)p<Y%58I>-NVN=F
zt~4=?m$d5Wp2Ak%{Q93KaK)eZrl<Hsvb!>4bkep5RxbNy7iv^^FB|7{Uk#)CZV5k7
zu4-uCbnQ5xQzcA!X34YuTvG2onK$FgU97g{L)oYwUkx#Re~f~-`~R)=xH9cZOywHC
zo0PRH<|vLrXH0Q7W-Pw0%Ec`(WZy*nX7^;0EnYs2sbh)Tdb>4Si}@l-XDV5wr>*9w
zM98fb(;U+z;vJEE7m1=ZcDCn<sg0V{V1*)4pLY&XusEDk$IK%d@yzOC5x2CoTONPh
z8oIZeZEv3N>Kpkz>uutK;PzyZ0*BEh=XWZvD#RIPWU2bxNY&qPRzL8Q`r>ocPB@$g
z7g5F7>rnN{_^I$Bs8AV)P{f;FkU(rR*d-vR-yUc>CVn1>rY1oX-Wj9vst^~kNxvKT
zq~wbaLvl#_3w<T5amST5D}(CZK=))S#l6wu9r=<I;=n2NVxxd$S=~9irEINBq^&y#
z0yO>{oz!zYTyEl=V{&*s&s>IUP@f5!JS)n>$=I=aT7lWq7XMA0K~VEU5`Isv_E}rG
zge?`UVnV#02=S!ZV#ZkOhBJKG2gTWhW_`MT9GytVx@sn;pVJezXbUtGky=Bm2rRZE
zb#u-QDJ|MN2O|}kHnTpBmr?g|*}P#8%z285|8N)Ncv=a=^A*ivxt_gTY{e=e_dx|H
zgJcpReJ<p7htu}=J#;jjJEk9Xbz`-}62hrUB&H5c|CQeU$V`!~OqVM1nv5OY{JzXI
z26t*hL+EkP#&BSjXNdGU!_&f`3iU7mHtW7fj4-Z>F$hv?v&Cu*uRkL-U0PvZF885P
z#{REg5~*XbO`~;>$&8;B?%S17FpuKmtu-j>YUV<y6R3e}o)@XAE_PoR3r1laLOB6+
z^GC5q1#QcqLvsD7m9aBQebX()3uLm#DzEgjm2vW_><|i{pqLS(otcTac@~Xa`k#Mz
zE=f61nT2*cK8n1S@AG;IgxXQSR|3MhzN&}En_BTy&kj3)rI;}SwUjz)>vfWD6FE&H
zr`h}7RW3my&-C`d2QGGlW;aui;V#F6M2%r4FZCirg}@kwaY#0Pu$x|Tu4(+Gn}eJz
zZ!D(wsD3IYUojD3gcL)lSzPDYcZY(i`oEp{Jb|k?YRjz8DM!>>=8)9%@F;Jw*e+l<
zDbt4dZ^O!`SKmHsVPC`bbss9^GmgW#ew?>sW-*$?L`kGPSC<tui3rnhEYlJ4G&9Zl
zNH1)mesV5*oJX%ID-^9nYR1di`cVSsj!6yqrg{K#=Y7W*3}VN>h*rG7ig1)~b}0F6
zx0r`};)@sN<<mGVYO*77w~@b*he4REyTBlq{FtQpLx*No&QS_{%PD6JX@z26!oS$z
z`yZ@2G3Sn8TDkHY^NfX)*{MGCen7sPphkXq@z>Isf`EdgH^&sB2Uu@2&@Xl0{Agq2
z6?A;ViS-71%YzL7Yg7B$Y4=q%tc(Kt7BnTdxQH{%-CMxz-rbx5XU6Tn&T37zMDeQ0
z{W5(CvHgK{Z$e7u2n(M->?@Z&mdR`nR~ufjFsa%jHMim`-98Xf9yBW*f$r98nTy$K
zCYkUQk6MaW5a%cMZQep?B#1|Sk-+{8;W_5`LP<Afa{~G}u4t|+PUn4}AEoq%`0?zC
z!)Cng)RM&ubn3`3Q3^$O)Av4!l-#Ne<_n5WN(hGI{UeD;N8d|yql>ZOasTK?eWsf=
z9wU&xm)xG|&CVt*2+M9@*A<OP*?Mw^<=x>;Kot)JwgKa;R#_6PO8x5N%XrZlcKoHO
z)BW4?{<uO1jXlcO)w#n0)#=A(scx+f`Xr*ypNu_{HKQ+a`;fG%0P$L9@s9hhZVQ)&
zF;k!O*?VP+66OL-eCnNxjN)bUT`q1x{qb_sfrp=RNYjbb-)}5=BL>Z$5f$$%bVO@4
z$I!pFdB;^<*%<T2;@FqBrg^b~01e}zgn14t--HZ$cmFCAbNhmBVQDFGyGl7QZkGAI
z2QpZC!9PPcqPU>ou5@{3ro<D~xYZl2CI~iv{W*&Xki@%jcX4l;*9`}WKEA~0#9e9C
z8xfdS)4VYU$#;f_v?O=NZDKz660fRpegGb%LKI;?*aL9-U@K>IE)SW;_!#NvFv{JP
z5qzOD?f^=%8))~}B)A2rf!gZ>^T_W)BG5jfO9ZonVj~z0>+P>&PcuSWC8coUbzJrW
zk|#TV&XD}YctEj<{uZzQOfrHt5J=igL|S*U9!6@TAD%}wWSf$d6c$FAKbbZ+*NOO)
zjrX=K{k$OlCJt#o$N`ZC$@6h5p6u*=b|9=+b#}TNTSrx$s=3RBWRjFPU;rECDGP|*
zq*Onw{j*sryd(Lb9M8%;pEF2OzK6sJQ#oK^AD5jhzBd+5=&SdM914%(uGfQ}`F_po
zUc!c6{yHM}7U`iB4(5-=JDUjr?H|yGP@w|{M-KxU-Mc!x6owN7ND*T_2t+wZTm@48
zlY|Nl1ieUn+qvw-Pwsjzb+S&MO(OVQz^5h6hcsejQzMT9Mfa(cARmfn_D#58y(i5L
z{tLIogHLSvL7o<-dTyMLt3?wEZU-Y)G3y>s>>>xgG(es_AQfgyQuQo-Wu(>^1Z;K&
z0rH4oG$^dNVM-pj3*B7(2b%dYD)iI$hw~4Bma*y~lJuu2gDz(%-c;}<6&{|ns#$r*
zYU@gEWCB?sDOvdY7!3}R6_3=t&AXiMhCuMk3P00$oXH>$BOc#XFBtBAn3O~o-b}dE
zN^5}w*bE8I&ao$WCVW6){-H<Oksl8lU6$VIZ94kcj2q&LCO}{0I8WW@@Y$$~19e!g
z99HLP_DI=1BOWkZY&N~ZH5duBdqPYbz8ZzEH?sFo1izdu-GH1``u?*U+BuGmaFmv&
z{3qEc+Fs`>{LC@$Dx?40TL#t2-jUF)>*D{{$q&(rGNQ6n6E;g2(K<Qg_o=7Ls{RxX
z<@Z~F8d3MdaD!xu#N)SAKiu8Fu@kAO9VJ^aeo!rjWkL4tNp`lCqhuPpN0TAAN<b`n
zOb4D1mk?N*(lXK>uGG@y%*0!{N_q9FWw8(Kn+ZJz!-zEG-U`LrioI<%FLX*vOaGg9
zR%kpO^_*OZyuYhlN%Cd$+3~+K*r!NcjP3umJ)$eDCZph=of#Wk3wvb*NfiWC?-^c_
z-&>xY1Rb_&o;C*e!g<`ig@$){PWzn$rMo`stKs%}=NI!nUOx>zln;6?bd5EM(L0yE
z2UVM1-c_l|9_!a7r(aIrTx(@S`Oaw}f&Q?Dc;W+Tj2TU!t^2AZfdGZD5^Akgf^|kk
z?G={_v!bvpiaNcoPD`lVrs-p+#__ti`5auv-DkPRmrN4awkB({;kxy{mwlGJc4MK3
z$La*z?D6tN>q0q>#H_s}93F$kEi%1?+C<Fc+E~Gyoz(Vo-$+;t3C2Q?nbmKaWAmdp
zj|e~pB?ubca86Oh82tU>gUdfBzi^5we(p+AEC!(*BZOg$Y65tsm-NB)kihRC?IsZQ
zJS*wJgOe(zw5=qv{dO)+>v;Td_Sa%j^q!v5<NF@gSb?N?Q5Jg671Z8akGF#=Mz8}b
zz%yk{JH4|Qd3SStcym{cBgrm{*8WX5VWah`LheK>-kL5yKzymfUie>wj|0pKu9o~`
zHgv)5=L%qQi#+O0=zpg_F39enw!*G~YF|eLh3)tySzidjBm9=&V`fQaKpn^(v9AEt
z2G^~}9-A0J>mI6vH^fjgtcwOIC#Ui@;(nX+zGS}ro4VqGz_{bv1EK4is^Sxu4%sXo
zWhekyW~)jBe_x!eu}_YR=ljwg!?|{S^|u98l>bKGjOCS32_6dos3)5yWoK}7-tKhD
zWVpImxaJ+spK~ZHC?Msw=!1eo%&(j!&jY9|^7>xASzb1S67I+;@1yrDIu*cMp}{be
z3X+lW%20~>#qZD_N(AK!xIw=~H0@@qKXS<Pg2T0!r-Iv2_Aw4}_HQBqt>BF&AIFh2
ziLlkGnb@WfaNYR!4k}#Zw*u>5Z3rsj*GuR}3Jm;7=Cx;HWqthHdY%90{@)uV3RFE5
z4%D3HSW14}$aT_(+`$v#d8)@6X+Svs_jgcdgnBm=BIze{qL)4Dg}oOfLGS@YjF@0m
z3fZU^FL6fxMv$;P&lOV;J=_T|@_|eSssZNa<``t*_s;ip39>dIBs`3GIfyG`lMVGl
zccCBI1z$$1Oa^f}q0CG0H6p_F@#!;0s3^I-eLi@vhxfO|X0OW&9o}Xl>fYbe)wbbP
zvK0JICK0vT;kf1_Cvk_nsQ^1K09p+Wax&973H)vlbL&vT{eQo$|1l@=l+NLxmA7P~
z#&N*`Kp84L*?-iBHK>5s#Mh}(xHlPow`0Vy%XVf6BHd}N>imq8Vdt5bi$V(@uv}@4
zb|&%c)9UO}srgN<Ni!!QX9x%1+xTIp%8u71=noZT#HuOndORO!ITJl+(DRrp{=`@G
zmGH^oNgNK^Jm|pQ%9P+Tn?&%%D2XO;5fFc~Q&q$;=iYg_L3OBYt3T-3|4T>WNEQOZ
z-9f&Y<m3GNo2{W7_ep<ucXibre8UP^Y`>x0pffTuat<?a(!?k|{tX2^=uWKjQ9N{)
zl&1@LDWB-Ffk_OvP3$l|e88=V13=q!8h4Ij;^bI_4y(eJ0_r_mP%mJF#hHI4q4dl6
zHVf7G?=1KdvwKEH)Ij)69;U>C)Ix_6b)d(>l!uMv+*H4tS2_U-rXh=St}b4EA%Yy`
zuNaa)_nHJqPE|FTkL%8$3=NrqA(#BiDrUAUXtaKD4<iQI7<S+j;@m!@^fpdYHLv~{
z!-%$HU63?|EKb>mawIRc<|<xM2Yv+_M<8B6-oBd2jf(G+ObDs4g=gN6Nph{gClpMQ
zkMS&&NH&Qf^*st;bd2QWAD4F=|3yj{DSSf0`A;=Z^8>)OG({wpo_!#xs+zKcX$E@p
zUDkk!i@V%`9tXeLfMqVidD7<8I9C<Q;jWH$4$&)l?x?L3X+PBLC>0N-16H?&71-(C
z|4jglsQZQQa$V>#NHVMk)BW2b&;4%Q6t(3@h-~%mfB3>hc28*F{#T(C%u2>0JAy~^
z`OY9fko(q7wvQ?WtO(mBnPY!Wi!f99kpcjnfat(IX$78SjLBHd$K!hOMn3h27yqzH
zxqkc-WMyHoTWD@-`!)-Nd078I`0(o{sgxl(Nm#a*0XvJAUSX*21V%4N*mH}Ux2LO_
zIlqRc$7XzL*99pP>^tl0>sg(4st$FF??JHjiDWwrsDOa?FMFhh=vf`L!_-}H>ZJH_
z(8KfS(fU?V8LvpILM@Wgpw(C4y0^M#f#$)vaZW~y^U0GZfC9r;$Am4ENI3X|^s21D
zy)E1**fEp?h^OBjYG;0t@4+Sp#itKETdo37T@tXH<Up^Ku7{G-1K)p#*S7mY$eDEw
zB0d&a3Od@v+IZQRNKDcycNI(}r5_BDDDZ@`c93|pmeZ5zg?8kVsPo8#m;cDmFUQ}Z
zC<x`beYapZuA!>rk58Rb#lYJy+}o96Ut5&FbGck|zyA5zrMI3teIdgRfqv)C(ed$^
z%iANUnG%tc7hiV`(R2Q;4jyhQDWkg%Ytz3I);^pUb{8E5mWyN%&RC1leNmsUSKm8;
zT_d>BSg|=BB}>c2!7<qM-Z}*5K5t)dufjm>F_0K0yg$p`o3Cq~y^3aGFc5xiSW;YU
zXJ>arSvtLf{L21#^{Ftk+J;L#A%b5}uqEyxB`Wg6^LvjuIIN;yn{VfL-I>#?y5pdd
z%#$^9qCV)TZ*1YW_c~adJ+Fj1=t_Fw^8@Z_owR!Hm{O5EX~Qh5po9!}R)U#6tLUM7
zm-k=l)uWRq%si&xb9O~n<xMWR{?s}_-P774rkCy6TskAqce_}rRO7k^ivCCJ=elGb
zcLpzOxU9;nLs;*k_*YgOKYhCSwfTvE{c`AVb!bfzv4_F;$X#FBgK!({FWM}ZzUu$N
zw>XzuXG^V%PIS?ZZ7EDo%au9`rVBei<Vjn#8HdMRA#+D|LVZfW_v&kEOWKf=8EvZB
zRqYVHOP@Uv%>tg~H2tzhSx)Dy>iakI`A^n(yL;CDoB=upvdS0wNPAyv5IB9iDG}Qf
zB*D!O>$&X{{*BQqQ3_j{e@|X%o8jO<_DeMqKVQ3JzT8^;d^-F4kXGg(@y{wHfinBy
zYMbcR+4uWB<Jb5pIORvBN_H5A&5h6FYyDrvkc0$PaQ3&FmfP6a9I+>*riy^%Q8=vm
zAq5U`n5NOqS6T1zu(#_FxeMqm*&JWKh#S?_(dpLylzC}sUOBb%@83UAZYU3A!fpaY
z!O<K%)Y`GpQ9=Y68JX))j)LCpwXQDN%dk`dHyb$4JjXBfnv^sIiBZJ+KnXm+-uVc9
z>1cBVmr`W3as#klrdau7#%ZlcPhe5Rvl&D@5bmkNvk+bbPVM&ZD(?H<q{5!A5)w#e
zs;GD<cS%X>!N$ouv_iWWs=FYN;Nj)nQi|!f;N#)hnXXnCuwWs?yd^deu1g$N4b&R;
zmVZ%B7<~<!@b9@wOWgnNqN=T(kd#!YtcppVpch0HoRC0bpnnHhZ^gMD0|Ud;2aP?f
zy=P!xAL^uR?>I;>ic3o7AShwWbQMpTfEv8Fj}IawH!rVNjy<oiFcj_|r?9%V2FoJ8
zoIx|<M6gnu<tT^Co0o0qH3khewd>{>=gYgWgl^9vaaWSmgK&`|2yNr7Rw#6kURq|p
z1;~ky8RZW<Zd+K?pnG0dB+Jqw&53CAwI-60kbI=o0nM6r%q<pT=0}e#A!dPaVmO)M
zc*yqBSFaLMRr=^nrI_KdKOb?!+C2}K-bGL{S#VfIZfV80Do#kJ)3PA)^Ye{adi;P#
zp_nS9GaziCprC*lT=zryTD<u0-y82x>e}mz{g!PH@#%%F^Z1>Y>3JFhKSyV0$z8Gf
z?r?B(BO^kjs6WXb(TC=G-4bIV=04$aO|b^b_3qA&SPKP?jb)-@2fwZV6u^eC&*AnF
zCPESk5@g%NnekNKP0p;RL<sN@IwGZqBOeMYaVaU6P`%h-Rbu4fA-wr5cSvqqPzZA@
z=V=b`erWZtcWZdPkApBRKZUq#qdX4_Sx?G9;GKli;<wwVp?zN}Dv%I~isQhisH@|*
zc*o%`Drs%ah)_)6FoDy#Pm0v9Ir{<xlTef2QX+i(-JI)2x_T5R=I77}B(~+{<;3Hy
z1Z?wwByMxwkIyFJ0wo@jc?rEa{W5QwElaE0{N@Rw*r4$QO6=!%rN2z=#Kgpa0Ry+K
z9QU)^1Kr)EoTgGxx9}lD=^|CD$xi!(!J8b;3#uqvTifHQ=l;G*#`JQ(p-#kSkbKwY
ze{%sG=Rpqh=IWn^vGK1P($mNY^u-DAG%~zRO-;eU_ok{ny#&AX)~}P@8L$d?CM(<3
z)%8)97L<W<va;GWb~?_^T-BCJO8pS3UHpY~ynh3{27dPTHo9)&&oN8XFnp4gkd*wp
zI|F*KW=fyajO=V$PEP2!cfbr}WMh-r9fxI{AsOTYA^i=}%6U~@TU%RSe-;>PUdLOv
zc~Uk((+%i40`v^&-GJ!B#soePYtV5DIm1Ba1==YNB2~!aH%R&?XJ)QJ01tkcU?b2B
z1_w~FbH2XvZg_5EZ0vW=2MJ_X(fgpF#`<~#a6144T!f06+V3gLG&$NRcs6*9Xj;<y
z_rolRq1oU9yU9eYVhr)#H5eLHz39eAoPzikxNSLVn^4{1v0#Js5`=yK6^t`H-6Hf8
zY$qTBeQ2XRh%^@gQvz0JSo>10V0LaUP=mpnpbS__pj?HxT*=$}5_$&M#0Uhiti7-P
zJuy%nB}4e7r&GZM|J8;ZBBiFGQ3cvh#3c0p4H^|eaRC94TF~f!K3Y(F$j!|)Y;d&(
zI$2+-MurNgPuD<q4a6QCOiVHk<5u8oM1rC)EzL+<yAC24)hwAN0L5T?VTvsTxec^e
z@!2yU@O%W;69qXr@Mx?;1Ox;!hvGqVGYg?E)XMlAX07b(`ntQPKuZohO}HaI35h!p
zsx*tV=IWe~AEQ8zBywB+yS({g*ROkdabn-Vwlw3|LLuzQgZM1qw)LwmU~zE~F*rF{
z2h@z<V06UC-c=YXAORKT=XZ5?i)Y=~;evC-5R~N3Pj)Zh>LE2Luo<k1xCw)RrvFHO
zlpn}@GBT)$&w{TVfQ$yziO~Z^4UN$+Cf}j@M?ggMjwO`s{rmR_6;)NI)t*==%1up9
zg2&z-aFViy>8WXGpg^Ohu8vC~aMb-MO~r*28yg#123v?3N29;gs`D9y;t&qmSXw^U
z&~Syx3ii98)Po^xXlMYYG-Is%ikQPfvoKJxVM4!?;DodlEG{UXI0x>LXc`++O+$Ch
zR+xvSq!_$>c>uwjs)51WPV`06!x?O~0uB0TPC;HFq1hilUX13e-pm|0e=rgHY_GdM
zK13MX+c&}U!t#V6L2d<lI*?Wkw#0JObc~D^oENuDKn`4rIM~cf`=;RAX;C@!9jxTG
zAbvYLKZnDQ{?Q{nZEY!OX_(1b8%s+|nT8hj_V&G*uteAYoOOw(%kKaBbpZU%^pS&;
z6XFml=q}jZ;1?l`O-!IGINzMUL`;Ild~6mh)`ksZXytHmab17GxWN1X#|&hIS!V<t
zIyD>H&z_#p?~5O48S&5u9Z<y6Zyo&l1tX1yhJx}%yBO$iHtdBDa1p;issHomPZV*8
zTVPv?qJ;xT0WVONwxYUvceH?@#{(v=A_Ugu<%h6#VLfw!juob{nmI2oZ_n!QI5%%^
zZ${O(R6y}EwzX}9V-#3^h}Pe$y|AA;IXMB(5p<^kZ{H$X2M0fPE<<1F!q29rNXf{4
zCU6EYsbr||@bFZz*x1@)u?LJpFb#7S<S{R}bin3Rk^Y9{344v`#SXVyG!_od$;rtc
zv`j7vAaz`9?BhRw)Y&V?$WGwzvPIa?z+}V)@(~7%G%9pTN=mkPTPv$mpn`s)^|<<X
zc0*<ddL#k8q6)66e2$ve5U$9nU!fj$+VaW@Eg|O6M8RDkfXRHa{?+Oa6dbUQVRQS5
z`#@;WP{TXn5%{ivU8;65aQb|~eaDX13r^dvM}dKXBqStwD4fQveoITH>gwu+XM0Oa
zNC?QpvU=>StW+2XS+_v~2u!jpe0QOF;C7c9wM3*Y92~riqJ?$bwdxFW0S%iB+5q2b
zKYu;}++lyQUu$dcarCRK?rJ9KvpGVHtQq4ujEsydDIxJTAmEW=g0Z9H{@NNGmQApW
z{6D7Q_xT*8%%PX5SYy69Lo-1sP<Zd=mKB+v$fm#(L>OCGbSxi&?Bv_GZ=XMZMt}Ru
zxgKbwaQ>Q>YjbjP!j25%>wSJC{`6g7U_xSILHyK~Js>5_bfL~e6ciLGDJiO6p3q16
z({NnF1_gM^2@u_p#Xxod7UQ`N?!W*XGvuM6LH0O>?GzaP=&>BG8-q8@gA{@4RWl=L
zZ||ma$x)>1s8%7%pX3@&z{K)fgO)sd_6%_VoUvi-zVh;NvYZTU1#!F|0K|;?LbW!X
zmjnr6(BP`duAWR1LXCo`Yi!KIW@l#)=N69u6SDv`^%+;>N>6!Su)(m`L`jOd?FTEf
z@H&D7uo-YqdiQk~)BsiY#=6X<3DD392%@(q%J>=+E?_Y{jUd0}2UKkcX4v8t_&8v_
z@UXrDt|P-Mq*j<&gXT4ZzDZA9=rz-M{Yh`Thq`T#-x7m(3oMM@THt{TK~@$`tg;lu
zHlPCdXmECMQ4~LAgMD#vfrf^rE)6u+#H6H5Ypj)NhObRcL(C6=P6qNk*a=FmjAG?;
znyj#jNMM|Kd=3u|(i@IJ_E7GLb_w=B(_y*U*(zvYR*}@;1WORr5Ad~2(ND~1ejr^)
zCpF#5r6EDFINq+Vt$kXesSyo62|6`)q~`=@fnYm*bCi?Cwo-AZ+meC@aoRv_ZEaAz
z%Z=0_?IunyP=`4M5eJ{g?-!$8T9&XAI5r)}dHR>2uEDPI*h{%!CJ6Q=nGS-xcQY25
z&;RZ_!U8q4D5|YqJwgxfjILlKARq{zk!R+JrmZ(_LqfcCbJG+(gH{&uby{b58?RNp
zY@<N8v0H?*s(+6vz8C0LL#T7#kdcsJDvsW(vR>KzM4OnJT6%>gKRxUehZSCpBxYej
zKW&v15Anz3k(dE7cfP1J87U8h8)El?$!nVwd&3fr%2uR@!pO3R73mh&-O+X=F#`jN
zElNH*At50Vkp~EXqau7Vrm=OG78i$HA6$aD#AyFqts~_$mC=G!k|*C?#NdQKSe<}V
zNbmX|N>Lw{SouG{e<wWvpv}~jdivz>FoPL(aWXi`uKId<kPtkUxr@Km)YPtkwTfu{
z1EyG_zNNiI%{OMRzSmb46I?v~xKAa^f??w<%<CBuK!Y9jlWbrd5%?si^5M*>+u$8T
z7Gp)j%#1Y_u$Nf7^;*jt<?8Iux<m%v<A7f|%5}j0gmaij-dJTuV}x>y+u6ZN;2Ka@
z1c%8tlzROKq@>-;hu~{f<a7!lzy_0AFDcKU_MBMtN8lwRhbxxH#1I~<<TY&+8M>z*
zx&z~_wDehL`^T%RD@6MI?lXD$(8$O%sNk}-+epszadO)_REepn;WO2bjj;otNP>MT
zqaB1!C5BzFWOzcQ?ceb9W%g^@s$+l`Q_gu$aPEG)bP2Zm)FVIoQX>P|`>HXsLtFq4
zDIvSOdc~V`#lyt~Qh*V|n1S_VP??jG-a<rnj&R)-rl*(5np@u3nA{pG_yd?gmbJTE
zW>!|U>hqvDP-A%~V;8|j1nPt_XSq}r1`XKcUHIL$2*-3r(^qR~&KB4r;~1a}T9#`g
zFDZ#UmLX+^*wT3Zyy($%dI15ev^4b)96Uqtk|C`XVg`2cvpsP0e?BJGE1_h4KC4Pg
zM|TNq;H?&j#Ma?tYzVXY@jJJE1C;kNavW*}RhwHLJXpDg#zpHK5IMp!Jk)$X9h1x5
zh#&G}csMF<1K7vw{mDV^h1~q(804bF#l_(y-J%<vnCM+~c1wO^VD;H%<|rXY&3Z&#
zTp7L6&;q6_KzxrQZh(v=lQOk}7<sy}VDK`{P^n6~ye<qbI;3ubc%k~EURh~ocz#Ak
zhAgexzBun~;ZPVckYr4Oh0_ftQbajCIVoFz$6o6D_wV4dGU<#-z52`0ZMv#-bY@17
zo7)A5$>0oi1zR$BnGKKfR{8f^s7qkJlPJ)rAz$_oG6Z^8zJnDOV8-EK^j)e}z@Yj<
zm-Sw?X?v)jpC81du<>(vQQnU1yw78WS3)M^FT}!f<4tvkka6^~nYSh;Az@>4b87T4
zGa;s>CHP^#0CgSkcOT!+WRQ*0XelWP1=9>J9-bqJV73UN(RW9D2W7dcOVkQ@-rFP(
zOZYy|QG2}0!GoQ-^*jj2`dC~{kf&NX6bfU<MQhIWWRmZ<#;~Tx4j}BiQ6KhVArzIA
ztV1#jJY#IKu(%RG#Ic4*WkOwIZDqyk-%r!>1XdEsS^!DlnPS{72H5Y&Po(5>#~}P<
z#2^1uRscG~%Hm?iVH`0u)M8|P(vEj4?UtW>PLzDkXj<7@-9js4<H4+_xHcKRppF>*
z<C;H<i7Ap6*&q&w;T97_=9;Q*23EgM0mp+v=qNca4^Ot|qw(?aE&r(-P{UM@==kuE
zl*ekYXx!@1T?jj9xav5oG(RW*)2AUg^Iz@H-)MXwI_Ko@l$4Z&=uKXhDYjy2C%1_c
zn@45{lI0(!P9q)q5JE$UH;8@^!>A5$prxtlTY9Aqu%rqBJfvQKzSzFV41j(Mi}!}3
zV)x!s$$TK;e6^!}Oj%u(0~0#4wQtq=lVo0_d}^CL2q$qdF(smLpu`5)qwI8lKkJhx
ziJi+$Hdcg&I0oP_po?nv;zb}APx33gYG39?fB#!_fuEy1n`J9`C~STo9?B^zJMuE|
zDagvA(}n8Bk)2Etilng<VnXnzm(qkEA|)as;^5#w#$H1%@}|x+Hj1`D3H!$P#q5rl
z0*o3Ren@@&6J$#(Q`5|4ujBCWaO)}j{3c-;3by#7{__5B%ZIx5D{U<n^@KcpfwpiM
zitOhrxqZdt<Ew>$0pSO)ZK<#$M1M?O9v!U?Y5K3#3nzph8H+SC!^YcTth*3-YhK4g
zv;fENG%-2~xWjdIS%K({fiqmb)Plq57}SA~aNOWJXuLP&@pMAQ{3AO|3;Hm!7i60j
z{SvWddzjF$WbZ&-neQ{Gw{L^1RAT{FO?5Rk7MA1;1V1jWuGunUNjUFo<>7D#Te@Oz
z<k6m<LeO?w&CJg~SRLU?hrEV?dwh0w_S-k497<#{F|k%ifehUU&5<F%GNrPyusG@G
zTTq#a*#PVTA}Q6Vmh&q(j|)kUepGWv{bMq8L&u!#pT-FKu3Dyzs;p#XmG9}08yOi1
zeGu7cyJ)_%CHK;fw{f%Br6};-J7Mp0LM0F0{tgnZ7wDQzdwY8~o_KXmG}F0EIy5cH
z3W|z~JG;AMjo64^+uLxqjae>y98p=w1F@DBA8ICy96VZ<#kRej-AAu==lZVA$Gi7b
zPtVS<!B4Nvn1Af$XJIeiW3wVwpSiQYC>Ahp@87@A<P0%9O!2A>;<IXijow`B*0ydG
zKn|z-)tyt98n#u-w^S-(!&dt)_4+&;-rx(Ur>E!Z>&rt4Gxr@n{U$ycnE<%USM5R(
z&c+73xTUG@Cilb2hrw%aUxh;8jT{krNmxh-6t28*<RP>hTnixfgcpn;x2$b!K-9gl
zx90$f2RPC}lsN-P!|8G)pcpsCdp4i&Q2bNH1O)}Feg_JC)>^wK4G7{O5(nYyu?enQ
ze&+-?QYlD`K*ytog{{LvT<K3tLqh|VX(bwfFZ)x4??15uHUhXM&doub>mf7`i4Bx3
zZ_>BLMJI6iB#ea1knc(21*W7>xP60=**Zx#vKVV@3zUi?mq+Rl<3QvMX%ve=LrhyN
zlAj8Le!1y8n8t9j!`Q$(1U7bd?2L`^@bM))Ea4R>i!VQfhY$5vS65%cxj3Zsv<RD&
z!~MWmB#n}U1Xd6i)GfCP0|Ej-ciOr5a&+bul0U*@2IBvW6jG7{kd;BkO@oIH$$g+e
z3Xo9Uzh3`a7wzcifD-}i<L~5$&t#Avr}<tb>Y}yzJrMQ;Lufx6k90y$GqYvbPvJE)
z0{r|cq{_E$-LhgO?Wsh*0S19cqSk4H8`)V|Rkbsm(+ezYmBi6+58hJ2Hvg7f<0js1
z%Bb0a+z57tPj1!IjE-^8C~zct4GAZ#qno{v!x$ls2Rats$B*Z#tO*e5G95j!%#W0k
zq$tb*^4i>d0;xdwDCw>=*j@TpOikInlKJl=7;iavc~kiuQ|s$JARyyPg=FUz0^T}w
zMTxiq_y=|X&W&V@ci-RveBf7N5wlRCBN^auJ4!}Qh=lZ$7{n=rv@iYm@uRA$3P4)u
zO+rFf$P`;!TY;Pg8gU5WhX)26Ve5q>zPH!*cx!Cbf0`wJSd2azs@*<N9X2#HynFZV
zg9i`bXqL!;tPZZZ$*@pSQ6aRMpPv_n5r=fy-@gsaTXR%{dArYpL3=bfs6t7Z2IvqV
z6*qo$bBz=I{Wpew@G6KKuy(Sgy!;bYFnqygzfk8CkZ1+i2}s8BXBPMOeZkVNstL3k
z3qBXSP!l(T6&BxX03Jp#f8cYFw}4f<7**XHaEn%zGB!4LcRvMVNst3pIWCAidXxtp
z9v2t){@uHqm<Q6UQGtAT-xR-q0N_+tps67%8miJ^la6`;i`&i3O-pMW&TO#t2~eXX
zhS|Ql`WPgw5Xi%ON%E8_ZQOo<9t}kIa#3Kj`9YSp!SlG#(;B)e^Au(*m>I$wEAruK
zA*zR818BY_fD-^<K;aKkDFxUQ!RGcxH68UrQLzuOOh_Eyy+g^w(narAo$IBjpMbOw
z3e;d{$O$_H@P$v8!XN<c=<MuW{e6CZzOu3cku42P7T`Qlw6XH$kPfqDLRpS0k%gUo
zA6nPe)&}ZrfU<zTeuL6MTIdVHsMqrzPpR-uPfwvT_X{Q-ROQLw#FEcB2TXcwpN3U6
zR$ddZ$CYk^QJp*3*uu~dC{03O3a^%^g0LP26A&P%Btmfc8k7jMazNyAsh@>!A)-oR
zdfL0HMk~ie7z$}MWL<FLdjnDj7^>Ir6%4tkp`oGPcy>{cnrCI1!Da+5X36pKo$G(#
z;)XY8zabweDwCA7s@IoKfz>)aGXoxdf84)9nwinQ2UF*^AGkrW=+)fW<%K?mgdGr#
z_K^{I!c-6ls<3B|t*UK}j*o+S`C;V(v=i*;z;RO>mNaapfDC*fd%Oko{!^F;kaq#h
zi;DZ8xcPx%7Wu9;fc|`TlYXgxCnqO?c;Ew0xBz_|93O-0IaDtIOM@E80N{THDk?7z
zm!vH%EvsTaC+2G~<*=x*5QC~15~M2tIi}OT2sq9PSR5spSA`?l8yKXV7hxFQ4ro<W
z)zqYmm2Wq70@9j9hS(z`L&FGO!>{iCskN{!b32EzpA+JTWRC@{!%BcH@bK`EVR(Zr
zDJqJZIP9@ccC)48CHxD5Bc88!%v9_3VM#_1H&JOY5qV-!xgo{0SxK?5BD+*Vk_E;y
z3bgn~mXeYI=ejSCM@CNHX&H~F!|$0&FT(y{TL-iM_OdVN5w9hxnwy*9+s%F8HJqKD
zZM^DzQIVOMIl;gb-3{++l8dq<LmK+7a5;5kXiJ$eR!~;P&d8`*YgnKGGBN3QdOfX4
z@5GI@3N#?Yr<H@3U47tsR-i#fNx86d1@RFJv26lpNo(;98%sX#xhQ26EgYhH-V7wO
zS{miB<YF>C-Nwep0hexhr78u<L?w>+{VLqPs+OOZEt2soW#AT<q0(iXEz!GDW>HXD
z%FD%N$}x(HfNQ5&6m8_~?ODUxZ62Dh`97SUoMedZ&ZlELzpKe`)0?YAR%QQPcQ>pZ
z_JKQVf>j_Ljw>u=f`w<%F;MrO4j<wT6N`Tzq*XtAjmZ~Yf6CWqD=R7*Ngx=D>gbTM
zk7`Mhh<f4S@#OI%c`VlGmZY4D3N;#WDX*>V?ZmS<2-*Raf?J)|WJcVoB(?3x<Vq}t
zUDDbi9HL~NuTI@Yz`+k_`-j!)PYUl4<4Aagg46fkkE&>18g~Fg@=h*=KQcFb#+YLl
zDzuz8|7rd7SL+A(0g0j@5}L)$4UDJn@^hHMr%^{D{x5G;mDQ?<hZ5kp`;g?#fvQ1#
zdH$7hostyB+ribqWA?%@R?gGbv<9|H<Z{+(J=G3P6oJX8h@T%1%PD&OuG?!yBbqz^
zYYQ~=_<x$$_W##9;~R(Ql7S32GSk<6XDcc>ufL?_kHrOYt@k~j$#kw@OAF>bz<tR<
zjp#Z*(hHG)>LpTAEMD>Z_q|GyCiKgIhU98bW3%ZDY|%MpSy9SQ4n!Na2Hv(a>le}O
z9upCti5hW_cqW=JvzYknZGPSxdTSD(AFiuTSG4@f_smw!XKAxi`X_QjqKIg@({|3=
zVYEk)xw^;u*5AjjoQWcR4LGhF^PaU`?$np;D#Yo})G$(W|J@7e#TGtqOu|PXUU(aN
zvIVo;vDPTZ$?Q<Zd2`}@6!!K}H)Ffdb3SI|T}`(ybOGP;OY{aTRy(CXce|1OGSBet
zV0F>^y_o%3(2Z^nYFoH$6sM2BTW1sJUE)*Nb$>lA>rWmKVpq8Oqg^O1Do~RhkpD}1
zC9cG>63dk4>dm51x?uQuj=2Jq!v>m34VHhs_-QkrS{)F<dW*+NYD3x5tNA2UY$OhW
z=n6%v<@%5!K#HGzN-x<0H`;19N^TK2Q~b9wyV&@7-kU<NI@CAaOku<C-d_hIT|qr3
z!z(Igc`qg+9nz7*e=xYeIn7$K#e+6D`Eq{{-hN+~-6^7u<X6~`{-xr)z0&gCm@|TE
zJM2&IeH+ZMLOBFtb-4)RnF*HVn>wqP$EzF8QN^+?Z)pDo1$U0|R1S0({&3eBWQY!O
zl4e4DjJ-$rTyxFR{N4o9Ex%&MzA~F)7M4=FTZk=6eV_U5ERPY<)EVw=HB(8=Z@cG<
zX?@88E}OJPr(PcNyP!BP$L*KFE{Rkwet4_v{V<YKHUA^#+G6qBh@PH*_o7gpOmt{4
z5HF;U%)1ArT`bs=w(wpbm-C2j)^~nrH`yOZ9$mGThzRYdW~wkvb;aI{rc9njN4(e#
zd;5DUYNz<CJR=4|&B@2wz`)7m9TGw;_VsqnP+fba1-nP*!CJpQ9Ri_4(XwAe>Clv(
zRS=!8uW!Qn;v7!$JmveF))G%O(e$En48Pj)d3X1bhQ%vyw?&V9c|>7XFf{CIHwmBn
zF7f$;@VwVVw9J>`wzwBWxMsnNfAW+Q1#@h*^m)TC3InW`qnTezN8cMrXx>1MjIoVe
zoemqWprfMGsg2zKx2IxSfSHs)jIF9$gF+UuktVvyfTW~{WG**+cDdL{UAZlbh$=T4
zmaY<dGUohkI8j|BVW@~-L|uGYZF3S|%*<zJb8gXVpYZC>3tkbChO^`@JAovDDSVE<
z3Q4Ykt+NT_(fem7g15A~%ctc1I)a|^7GJVa3kaC@zNgs?FWkXue6@3RY2iQt-E5$W
zvCx)^9atb~X`ZNk`+Zz#lUL)AH5$AvKqrMnZ=H{N5*L42L5_-fZ&$DOFb}J|7T4$a
zxL!YpdiNA+gUUaM`H&F#OreD(<i$Fz6;4&9j72=89KJ(%4{sHQnsN@z#>n{AGo1LV
zif#OQKy(;_kFRk%zS-!{@qAHv{A<TapC%1Wg4g{28jS|%<9lx*G;-Q8_0U9a*?FOj
zgsLU>25PF$F>&u1)Wtb2WO0WJ7WIY7)ox!q_)b`sTdN_=*V2+j7d$s!?X**D`$m+h
z#?4MrUM)MSDi_5f5Is?*-{$?%j0dMuS&54fT|FCFue;f_j_l;_sY-iT>->yOYd*8N
zTw7TB;i{|OFsY{GB<Jm*amr_8|BvpOiG0>#Su|@#`ecG9TbqNcuM3$)aTu#y|EzFM
z`TBl`K~s|1uH5%H*Sq#sR#f31bPy~eMmqW$d6zjb+=$ZpET9U;Z4~k-DtO%tcKT-U
z(PBJ8yJI}{@{&SzAd<s;Bi<FC=yus@{%grsJ2_pQ#D+gV6Tjd|hke4XIpukJB<q+n
z-Sz{DM(t%QWf!`7wwi3yo%p1|TSzWuHg=^?{Tl1b($2n5uFE5+`cW0HTfTgeru_Tn
z%@aAy?7mkAXzIwj+VD=qg3o5Hw-CQx#;7sMk<8RJ2%=Ztk3L1KpyzzHG}GA}?W^qZ
z<_#l;SdV=GnBWVzSZl9O+LFp6UQ%$b&!uXiAU<*xM18-FpyDSt&x*0Pp<B_SfBJ?X
zb%oC@U-rVJb!k${u*@*dz|)O6)JsJH=_6AuXV{wJ<;yd{S-n~;&FpW3ma)tff8cE?
zj&f4s8sBzQQ*+_2Y6Vd+<Zr*P%I2sc5So&jI<X%<PCT~qE$Jyv(3n9`RytfgQ#Kto
zM<8ZXqlpJT3+zWfT#es*9$kp~rx_BySBtk03^{6Ih59&ClldC4^ZZd8A|BZk;ZL7E
zSB-UwEdN8yeqq;N`04l&^#kiNI0+wjtqyyJ-NF(vsI)UaJj5EP+H>5nP@qM8Tz$>1
zfE2K^DOA!h74>$7EJ{=QHr+7`j@82Qpq$3g*ut9XkF$*ptf%ajd_3~iS~->qVkEjx
z8U^((g|{<JbalgqtLO$_N_ieCAOZ^vCQ558eDC7~B^nJY5hDCzv|=@7#ivxtyxCJy
zS#Q03hXe02RlI6us1R7(c>L#ZzwY<OV{g|Ay0-;JHkiVW)AZT%%C;;8*w=A%Wd(OT
z*PfB0vAASB!9w)0o*q}LMbUl18x<Bh>)n;e-1fD=RNN|}qdm}w(v2hCMcq`e!8fbe
zUiGz`!wet3SK;m?>-Z~4FX1l2F5`A>jn-`z7X@*hqs8D;d9wwl45ZR*8JyNnW3qQp
zELe2*+IPJkcqtiGPOGOk@zE(PEh!W#op>h~7iwC>J{oHp3*o5Y;6p*AzQg0PkrOd-
zFiCptwm(%RQ`Gtw38k0}14C?leDn|Nhfqp7CLyn5Z_9?f?g_yM8h5bEN98roSk_FG
z<Q8!k=2f+New#k%{p!y_kmQ0uZ_cTwu8nw+i2km)4@*>9n1<s0dpAyXqS5?TOj`Wm
zzJl{L$B8=^&xl2pqk_{!nYY*2&KJT2D#Hc5UN5`&|8jMDd>tmk{`8(6{(;i%69fGZ
zepD0`+95<th+VQ}foXh%&MVQlxeJ0_`oC8)e0obcQgr-EXP%i@jp9|uMS|?8%`<1d
z>^4jrwZ=IfifZLUAKM<NBB|T6$L6y3ldaolYDJS*UURXQFxP!bW$njuGP{zf_f5A{
z*#AlPZz24Ng^%1LRkhD?+PjmzzDFAD+bP>-m=gM*s|(SSYIw}$jf#Btm~_(IS6mGu
zyQ>GX=wf^NsLLMIbm$rsI@CyU9V;kpBK)7w&%FK_7kPQdkMOC(DSfH2-6Lr&JH_rV
zRKXD?Zi#jhD@b@2J_J_-xClR%znrH}nt4ulGf%1G{c#j2-=mesN)kIJZuOvQ{-k8a
z{qg(=X><AFF!okcQ6^=D{)Gz_nV<QHsGW%~tT%_+hLI4>OD|f{k0%KX#SPVbL$*Wf
z87UjsvN1n8Tslg132c=7gv<vj6e71eqGIL#j@ULfuE<y>)Hi)EMMl)94+e$3MIct*
z<bB&X@?_sWeL<6UhWdE@nPH5p4wv60Pv!Wl^co+fGZy-GhpqG207Z=a#@ge`ggYLT
z=HlNjKXazM!_C67N7E_iy4Bihc7%iruXHY}<|jydRQbH!GW1OcY0`6)Q4Nt^rlR69
zrmix?E>fjK=6;kl8t**r=^Otv9!{M_5xEPp@$NW97NIU=&i3!`dzOYjG~f)M4-Iw4
zv7l2i<5O0=i6gQ0O-wIhC7mL4F13tf{H#=&rWMkVxGg=Jem+0HjBO_RM(;uWyLCRJ
zne7~Vg=_4*qR95N>x8(tOe&=;ONBCwq?)6JkFm88T!#t`A@bZmuRqydZ!%P_E6cO$
zvsB0C+QLO9iz5T||2(whJiNn&YKskfTZ^+;yGL+1|J)3&&eUP`KV!$c_x(8VoqtEs
zMXcRc&iS___z5Gld4&EfQ6cQlDo!(7F8fmg{aKHYyLW398lJhgqy<CRsACkvT>aOD
zz-nPtR#xi>>o<qxEj)|0o~k?M_vZfy#fqOC=kqM5lbQ*gRo^}<E^WqP@8sxfeD!Hw
z;KWj2a{uq6(}cL3PTC){&467UZ}UYH?Rn2tZSM$0c&~A+uv!uz`PsKQx_`9Y$n0V>
z=~ZK!>#C;U_6{Vd<*FSW?d)CKUqjVfnXl*BntVsN_t84T!9uGPM|0wu-;Gsn>1;mc
z*0^rOVz9a?Z7R)$D=!tIPD_ReSH4)_xo`<@am2;8;k)(hzi6-CR(vJ<6dBaM5J@;O
zub$wscDibrIrxRj%fAcq{gILHx`qxU{Jbi?Ldq1<N_KOC@rI&qdlqC~#*_T)>LtlW
z0#i}`1<7cS7Jrm8lUmSNe0a5PyKM?vrwecXZ#8bEH^>Io#|dtgbn0QH1`n@ZNNMON
z{%IU}|7x6T%8*aBS@lu!d84n?Y~S68&%{eMndX<nH4qklj9b;vId`ixwT3Z&UdZB}
z&~=z)0%XA-cU66DUS}Yx>@glPcG2lye*U1@-uEx+Stbj2JX_C2r?P$o`<@W5U0qfU
z#;~hB1_D7C>(Sm}C{m+eI@taF=AlyQv^-h_Csjpjewb{<C;9KZ9GyE|OWv;bB-SKK
zx%{1nQlF|xlo-?^6q7MnSt#T*5pNl);poRuljEE9-ya+I%@t^;Zh7`889QOK?Cjny
zGwYG@Lg$_6>S>FcSF=iy*0E!2iOzZ)e&^Q&b?+MH4mk_<kynZR!%+K8BL?_mf$^Mv
z3jeBs@P9JV5S%ainExfLd}#llq|}usroiZg;>xTUlWgxjyXa@R<=C<^Hu!ASeF4+o
z_fY@H{Mq?8P8lJ+&NF%Xb(dv8z5cYf?D{;t^klDm`T58Sj^Vejm0q|qxpKP=^$9Kr
zLb?#k$On-hm{$Z`cl#`ndDiBhIPov3Nj&Tvsy;lAh?n-zo5(ea;P+XR*N>R(=%4yr
z^yc^a)mj*l#{3gJop1e(?koC=IM&;H3>%1j$yQ?loH|HZT{&20y#$gm8c&WZRC7L`
zaxy8OpROc*iTvA)ryCln<XiODYbG^8J2>)Z0%2;d#wtaU&{A}7tV~Djy=%vfm4GKb
z711s-r46hE351k5SXbvB@$u4kt@R{29F!lvvv_>(9x7|S25#tcVrKTHYTGtJlg5fE
z)TK&F+$2}O@mfkq-<f~Kv7wH%;Wb<cBxAn2C0smZ?R))$^sDn<O9G|)C=)({LZ0he
z1#98NeAoQ(%u02a_9wrv!_cNH(l3HGEB}%Y95~2YUN|IhJRUU}4lb9rco2Hwdr$M=
zB+0Du>2bw-o4geoBT+_mY#;M-dBm^q)1?i*j9d=ibS~pHc8=|B-eSBvYeuMmIVc7N
zad6!`K=`#UCTZ)EAtTxrEBrO$h~m5NKQL%XLPZR^bzEFoG&lJe)*TjM2;ZA{wRL^9
z#X5h&-ur8mZWn#Vu!onEo1uiRu||Uh@g=3fiv$_*Qn;y+IBn>H$c*FK;s(&Ts(z_I
zYw&?Edc640nxD||L40}K+LY7t1{M*D+GxV3Ww;m@HP>=S&Ms${CD*UJ*JsCKAE1V!
z*&@f)YE^Kt(-hsBJxlNM{A^z^_J|Fv$K!6m23P(`yG*K+H+_=@CA$;$Cu)DiM{>V?
zHJNQ}^x^yW`+dgJ`PCDnM?PGQ_ddBlUHao6)`s=p=Qkck)U;1TP`-9zYjFB^6!y<>
zXzKVqE0MwV#JSG4a}cS_{uAS-(=FStya2OQR&XXnp!R<2|6aFJKV0DW+BQ`w62W&%
zZ?vhb@uM`N4d3JB_>1vfl(veao$bvPVGC2rXKj^ghAr(2Y?LfZ!APChwFP}W1sn&d
zhpX#mDfTL=Y+1Pj<Xyebm{;7D4Jx-piRp-tMALqD^xwO0{OJ{k^+9A>W0QO72B+7b
zjzhLO|H!tlG6&3000$C|u{OfiX5n}8$fz;e1vtpHmDS)5swA$&pxG&&8G&%i1Z9AS
qS*!5>Mu^-Q{%6VlA7V`ZL|u{C>V(^S;6-!@gq-v<sZt5!xBm-0tT#UZ

literal 24266
zcmZs@c|26n`v-2xzFqsy%vd9&gzRR9M3KtAWwNgg*=1iYk}O$^h*b7{Cz+IN*>^HT
z_K<!1UG@2Vf1lUy_4~uT#&PdC&v~Br^M2mv+;c_U(Y`@L$w5g#KtOX-1AUi(fCwZY
zAS@*(hFZAwt%e8)hzai8x~~TPN=QgZL_|bPOnl+O1ribxQc_YfGBR><ataCxN=iy9
zDk^GfY8o0ET3T8<IywLV=;`Sh7#J8C8JU=vn3<VbSXfwDS=rdw*xA`RI5;>tIk~vF
zxVgD`czAeud0{XZ91cey5PW=m{QUd^0s?}9f<i(<!otEAFJ2T85fK#?y>#i4n3$Nj
zxVVIb#O2GEB_$=VT)A@f>QyNzscYA+NlQ!1$jHdb%3i;IT~1C;US3{7K|xVbQAtTj
zSy@>{MMYIr6^TTmP$)DSt)`}?uC9LL#tjV(jhi=bYHDg~X=&ZMb?f%++uGXNIyyRc
z?%dJU)xCT7?!9~W?%%(!r>Cc{uWw*rU}$J)WMpJ)Y;0m;Vrpt?W@ct?Zf;>=@!-J&
zOG`@(2J`UYLn|vQYiny88yj0&TRS^DdwY8a2M0$-M<*vIXJ=;@7Z+Do*GG>YVX;^@
zH#c{8_s5SPdw6(wdU|?!d3k$#`}p|y`uh6$`T6_%KY8*bARr(xFfb@6C^$GcBqSs>
zG&C$MEId3sA|fI(G7<#AsHmvu=;)Z3n5R#l#>U1zd-g0YF7EmB=kf9J2?+^_iHS)`
zNiSZ!NKQ^pNl8gfO-)NnOHWVF$jEs4^5v^nuQD?;v$C>YzkZ#aot=}D^XAQ)+}zx}
zyu7z>-{$A%zkB!Y{rmSHK71%BC@3r}EGjB0E-o%9DJd;2Eh{T4FE6jCsHm*0tg5Q2
zuCA`Bsj024t*fi6udi=tXlQI~Y-(z1Zf<UAY5Dl^V{2>cr%#{S+S+h9+~?1q+uPf}
zeEIVA>(`Esj&I+-b#``kb#--jclY%4^!E0C|NgzNudlzqe_&u>aBy&FXlQtNcw}T`
zbaZrVY;1gdd}3l^a&mHNYHE6VdS+(k$B!Sgv$J?Ser|4Vet!Pv&z}nm3yX`3zkdB%
zT3T9OUS3&QSzTRSTU%RSU*Fi+`2G9$=H}+s*4Fm+_Mbn0c6N4lcX#*p_V)Mp4-O6v
z4-b!yj*gFyPfkuwPfyRz&hVvXPoNy2c%)(EPC!7#aQ;KsKNWCHKma4SiB`Su{c>%J
z;-&V`5y5X;o2MSDV|xNxRu`eq`<mfjn<G}oyfnXQi>Sc`;OJ~8u~XxuumtI-@>t6*
zho{%i{Cb)eC;fV+o=KMYeA!UQio-56_D=2`O`l{EYlKI`$Tc=7?qm~1hP*=vBUSIy
zGiAd*kViic-@k=g`w~{#N&bYyuRP&{hDJ62(@+_HQg`)MgITun%fmLek?%Z5Bgwl(
z>6;j`jg={Y&Ln%dBpZotxck>n`>f3<gHL|6=yxcYW53M}A8^KMn%=V~@G@l%<>K=Q
z-2ln)yN-U>!!$3Q$%{HLqz7v8JAL5l6S#TiJ^%3HXNv*P5g&6=w#~;3hxIP#7dygN
z_G(Sl3@(Z5Vy`3RH#*H%)u^%0BMvb8Gr{R_X7;PO@ZGhi5l>?}{kJS+vsDLcz*ogw
z9o6yv>35yPIL7aXvK`L<%yKJ^>1z7%X@QH3&+@@!C;Z97MVlu2MCYXg?zDgr*t*zb
z;-+cY2Ul|u4<n>-*YfWC?7cG}Bg2fO#`BUZRzCGVRIBdnXh59}tnio|>T?$rpZ(U5
z4ej}q^$UR$Jc)MU`nDNMDp~g<U~s=%r8_Q~nxh{$3l25H_Z2>SD4zyYO}Y7T;~#w7
zjMh(i&8;T>DmadBQGtgdiu|mBsO;jlh~CO`P87QLsa)bUkvrRO5TOyr9jLV3@#4J(
zf0~qZAn_t@KBS;3E0lx5lLN=>QDs3!Tc&bc`~Bit968P`Z&8u{b)9a!C5i`A@_doL
zK3-z}9WAzm8DD(Kpfbx3*~11%ECdKRnBZzyzzvoonE<8LR-j}?@6o;2HaFDNX?(zf
zA3?hn51W~Yu=jgzvw%+~=CV3SSU7;b5f8Wf2Lwhi^Of|9$;SADfG=|^DMFI4?HmO0
z4PVm{&M{?f^$&>Jxn0y#`Sgu~38lNttvofn_{>H=mm^ud3s2sexQK)xn9$6s^prW#
zugOVt)$W9yV-G<Cd;4{$<88;ud#m(^udIwO_qI8w4oxV~$xv(>Cp3v~MHDToH!7}u
zyXI$B*d!vTEg}Xt7Fnxr8(@waS@U3BqY}B^n&~-axy#+|I|P09<4ZL21k9>O1o$UD
z+!Y@mD3+i66z!J!B+{B^?eY;VBP%Q43(vLaCh<)7H?V{|z~?&vGaAl>j=Td{w|{^|
zXTwB5j4()<15oA&FrgVJbJ!?zLWILYenAaghRkS5CNzmIHHoe&6H>J@EJOpT%7jwA
z4}{+VxRI(!>;)mh5h3sX^@9`!lkWgGC(D2x55|@)J^g&BbL1o`I5Rk~AVb03$@Fdc
zyf<)y0PDpZ*50Fj1BEnv1>H`G<%F&G!)VjOo(0uqGXJ|~v%(MX^;9^#sVaq1Wt#I;
zDV*F`{n7X9GJ(m>a~pBaL8J9<QK(S$*SyT=t4wG~7h;_8i)h}}>sF3@1k%8{(TE>_
zS6A^1nMDn@eTSPtij`MTs>fH^^Dbpo@eBB@|EW136ag6?(y)LfkDGaau>Y!gE&I{M
zI4v0V_9a_V5UrESQSbUp(89|F(swY;r`HT&;sGB7RRVu)W8$AuNCwOWUHG#b@wwsq
zhdH|=rMoFPH}F+VCTPKl4+(Q?_67`V9%Z{9*dpa3KkVDRugn$pGeagct>bA-hITIa
zjax^Kd9kew$-b|wEx2nxB0jOgP(1O$#k5jS?1KT}Yl+8M#^euJi=LhbbSc);FkHJp
zP+df9^E{sQ*l_Us{DnW?v=(i-%zY`>Ce$5r#6Ns}BuTaRX^=$j=~uc^$tI!pYxqWZ
z^A7QNjb&c7+GM9QW~kT*Q`lPOcsjR=pe_?qI5Aanj?pN%`n@<!<=BnecctJY<Ga7d
zQ#<vD5ZOn&Na`~A$Lz^M#Hd^8ik9oK^J?z;+r;+)fp!iq;HcL)2<{o$L+ozTmB<`-
z4k`|Qz#_GTmrws!3f~(SdNUM+630g>t1^iy{*;tTT*a{TY*;kC9e8!Y5+s39pXI~)
zAzYGQ#x)NthOkp2Xh?LQ5};NGyh=&%ww1=`V1xcR4Sx-xiNQWNg8V%Ai7XF5#p~cQ
zTYN-_aJzH(yJ}DhLI4w5OBm!|M}vyzKob%B2sDU7;J!cZ9KL_CNdFxBMAT&&5acxT
zNZ2q25JZwRsV?L4&oL)*`!B-r+F_$c(5QL+|LM&il((-sM=AXSQ0L-#cjuG73w8Fs
z5;F81WY=B$f9d0ez|o$Y_X09cN96p{^bk7=l6c%J@=fgD!5WGHukQ|jq4^P*_-ow%
zdx_+K?YZ&S9?1NO|Bk1W^LLmUX1u3y(Rp57V8gfl8|o1_bUs@IbUAN`-g#R*G`zEA
z?D;fvNx1OF=iO>R)08gNc2$!fd38ZjG8lc!DMSo0=d8DBux76-xmzzvtJine-r!Ac
zKVE!n1%qFO_+}eel%k`p*tk60NgKE<MVFG-kKwl2`S64XFG0y%^y4Iw;NGPWxJ&j>
z3YX1{W!jC7ZdB!fNTtDN+3wJ?u96uh9A^hV14E)SeVSE)W^`4)nRe;O=h1><xxwrA
zKbysCHn3Qi#I)FyG<Z*)=3{p&h0jyMv9fJdKKHY_R-S-d7wY8`o?;6o5#!c~_2+t5
z4Y^R0>IHAkhx09w?P?zQP$%*t`O%nVZUkq^X}TF|Egb?OSHl9C<vqG`6LE?$LZ$2>
zcK?xEWPdYQ7FHc5ruaIDv`lKg&ixtv5&<(hs*IEh7w3zV6y`Q&q5L40!w&HvVZ<m=
z^=_9b#bIR1mm5+41*Q}2gzs%bGU9T*9y6`K)Wr7@$Y}l2Wa|{J+H!jz>#t1`A08sC
z9t4*T*YWM0H2(Qgzj4!Z|I7bjxxKVvZ+l|-SIeND37P{MDma|Yf<tL#`_mUvoUcQS
z2E=!Kr9d3<s?qk+Gv7m3N(2>@6d#et^bclsaQkil)57tib0zAJ%)cQYzyoN12mV8V
zA(%zPb<XxtmC%jkKTQ6C;1X_!VCEclXyXU)VnW&6e_0h?OY}qNzgLV9JWB^wT;B*l
z4NQnqAcg4o9DNM-gpvD@h&l%q3tqI8<s6&m-G~SGp5s+etQ25vV`Twps{0fE=8GX1
z^Fy>s?LJ`LPeg-lf!e<1k}%;n=-hyz;<R%oslba#WzRGF3Y6LV=UMjm3Xve!e?5LT
zL5t!dc>DhK_e%`#bw0a97CUgJz^vjQ5BwB=C-xsoIC=ix9yEWIK)v3zCybC)`J2o{
z)SK+*6V$v1h-JePY8m0Ql}M+`@Z`TGi2%iwTm7HZrfnvMou)m-o4;(nEcrh*0tvFa
zjl6Wj+#wp@k~D5`JR_A_l#)>)(S_)gVAI)Kx()SoMixeR!Vz8_(7Y@=-W)e5utd+B
zWfk+)vHF9f<ZdODMZQDk%xiZR?=brriHOZYMPIzt+^bKCE~*Ryf1OV)9xB`Rk_J!~
zzFGAsi-PWBt+O!r2y@<@`&B}toSUZw<zr<%gn#=Xv{v3{`NCqhhiO6$NVMIRQ#r1e
zK+7Md@%)xm;7J<#{-CvqX{CW#vsFp-<NbU2Qjv1`A2i-WNha0d15z&d)XvM(X;}t&
z=&v+=5fryuXx&$CCBtupW9|Mw#eye&)~ovhKEOY@JdiQH<QaYU*E>Mu6gyzSD~ZJ$
zbi^zaF8nvYXj_P3^$;yz9exnB&+@bs>C~9^*URK2eGHU{D0zCk#@|RljHLUI!XALV
ze~Iq@s1RGpj7hv*d5%ux1tvTt+nfJJ=NXhU|I5{~+W!eJwh@Z`Ij!R_{BNAMVd9Zd
z|3>*ChL;k>z2*Hsbc#LvS0+ju7KF}`qnUT{uY?i)KTU)Z2CqD%Aq8EPqW)g`uSEP;
z1ZgmW-O~Sh%Z3>vRr#UHcCM%(B}AJI`}S1tTr3@Dh3~7AozLX|s)Pb=fz$5hd1brv
z7mWXZy6m(UX*x{*KVAv@C$%W!1p$-7{J**Rf0B&P5Z?YJ0c_@QjQPrDqSwXjH=bwb
zH)ZXomxOwS8hW7x5j%JP%Z1sTrf*mtuG2yLe;+xy=Z(=%(x|pe_p!(MPH8e`C@t;n
zAD5dFwjUX}@Z=~i%#Pny%`G2dIfr?vntu~xJzov@IEXR__g{HP_m_Pl{#80|y0V{S
z{{r_<4?@n&<-{?cVCcpZo&O03YQVye(pm$&XKe7)QEDhtjP{Od65F|!RKP_`is>RL
z{HkUDCKg1FTY%P$Dzw3RwQbDbyCKyIr^8MY|6I^B!}PNJMIGemIYDjx-CJoVO^RdE
ze{WSKFB_6GrgQ^&6HEX1)-r<<^EE!>v5L?Cq#r3*;;GKgfOtk_y*h^fTMZyhJjsRQ
z{JHi&zYiAIjSwsTZ_8xIx3jVTOzGSxs<Ke}T_@~c&x8W_(2aq+cTkhgVTvYPuS0|-
z^&uU7V747=LC1Sqoc%wIKH!PSyj`+w>VMjxw5hG6`t}^4vf$mG!$L}of<)H_-_M14
zzfv%b9{L{}zM{Bxe|LyH93TD9Z1jN7&m;FcE&qJ(O8l=akcgj$Pn#VLY3-UG5m>_o
zlRw3X^xgF`ffXBte;2*cC1L>bJ}!1Bat~&kAs-3!3?9CYn)ITUQQ9vt`gs9+@gS{y
z;I4-ja3pT{RK4-81omRlR>Y{UnAmj`7k6C!X$0*{TRk42B(Cf@j}^CJICOi`QAw(K
zi;;J%BK&y-8_oE)QL}B)iXIBg<CPuqCqwIvdEe+U$L-JV?TfI2ZCOW8oD;peAv1~y
z?hdC@^^sthKi0*{49LmjzABdG*GqSolXF)qOs_*U`;mbo?yJ*#Ek)eQ69i|KPa&$I
zSZ$h6F7kb(?7Ew2OX+}6OhD$O@3n%*Vf!bp`z_lKXg{%Iu9%N-H&a;p$ZUUai<QU-
zn}93}9X*-b?<`%FGbZ-v(i8$q+Bi2)zI~DiY-#_%1@k<p@oy;2KajHpH(I}r@D^H%
z-xdN3Sn!|k(LX@P7Zvu?O0LI$BWJ##&_W~ArzwxSy;s;+;BI6$zSeZtCXXB6y8HSS
zg`g_8?3PSHCF<7dEqJ@O;Z0F)nEI#Oh$0TvitnFFRC9Gxd_j`H!)vU^iu9gM1g>S8
zH^8_qBt>9=5ruSCuAVlaY}i!UlzhjMvGlVC<6im^v*UJk+X9OGI#{$E7(noYx02}+
zUGjC+Rf4?}=493YbK+GL8Cds>#g3_eRmxzIU^68uP(MU9jf2I^e)zfy7p3=SeFE>s
z%M<<ARgPJ=a-6dt32c?D&=_uPc!IsCgUMH#Dsz73;`aLEa8L4JxAEu;L7DVUye^S!
z6y8nkC&AxKK1M*@+OnA|)2R@DRVI1YGKjkL8|e|$qW<aG=5Nx1$3dyJ1qZkOSav!e
zWbb$zFJ8c6%CL7H?R|c}^HE)bZnKs7N+g&7n0(wGFW%s;WYbK0p$(W2a}DbEtgn6?
zj^f5|KIijsAf`syg&y3ECB{|Vp?J+2X4pKe4UiEjX*Y1begY@$qTOs4{u-N1K>h3B
z<GL>okQIETPR=Tn^hK7$Ca-)_CCZ=ZcTR0?x1*4(w&)rl@iFUHD4!!Q@)wN`xYzdF
zcf7RHl~Dufw9O+n2A>Wm+x`L@m8Lbfe&GoUDG85w%NZLyu$F!|^a65)3$HT0FF~UZ
z>F_5J<1}qCMXAt+HUdHQYNAf6541qwNb*8}GrUlSHJ=z)GMmpeMDWcO`37N_M~u66
zhv79SUHjAK6)~>)&LLCuH`?EZs>8ZD+<3k3FmWQ`%|A`@%ib!hWCCsu-<I{D4ILcy
z?iXorjWGHR1#wlBD53HI1TpvTzYY_PqDRz!>ACk^SGh!ZJ35439I$Bpa>WhUo@*^!
zC&~Eo$xh!sQnnV8O?t2LO%^MPU^F;^8(%Kej=B+)avJRQfWGLK5O^DAno&hzVW)M*
zWzt3qh}w{P>v#Io{h%(YKuyv`wpF3z*do=_$+$hio5WuYs|;=Q<y2NFX6a>4zxVTq
zeKh`Bj%uWS&L~Fkb*2t4wCAI;N~2bhME35UrD<RGfE|<oZ^jKScnk?`2p3hNCMkdK
z%n_tj(xeOF@DE6*X!Ch6N(!l)W)Du--hB~S1IUoi<fbB;iLbUY;tNPjTKNc!^xHHI
zRYV6UFnr}<?Q6us`a34ZO9W*vlV99ElA*3NOQ81zEk0!f$0xi^72>6HWb(K<r=czx
zD8-bDthx4$jy6?zcBAvSG5jhQ(#f7T-AkMOSpCuyRD2ENu`X)3!NUHD<6Eh!!Y+YI
zbCUSs{0Z7|4d5j9-(*wyiun^wh6~4j?q|Z7E2QeueH!n(lcF~crwuAS;$(hDaIXmx
zzPG;L)k-7jwB$MWD9u-f4|&cT(M;B?kDP$s&q<QW*2+N^c17Iy#nM|DGC}d-$XTwU
zO4Z3TN|S4OUB+bKxC?Q!p&G+jTaGlSRZTdr;!c2+XA#8{V$}j*^3_M*48)HM<#E%p
zi-Y$L=egfjNrN(!c3(QzDwCu;VyevQ=%vAfK(T}fU}htWjLt?w$$RnoyY3HbfTN1W
z&Y|%F+7%q%q?Q^y?G76GVGW@cSXtt!49xy50NJrx@vgG$Qq>Ml9614Bmn>iJpe&8l
zABSmETNk`7ecwxqC7Vw7KPA0sWt#v&QOJJ6;ZH*Hc5d}&_efde3IAERo&*^QSp#Y{
z%Kdctoz67e*fDg!d3UMj#v|<>dLCe#TjFH|V6l?QNoS+NV<<<-^I*sm9JiU{S<9tw
zcQ~v6t|^aJDDkKYyk|+#_4dWmg)&+r<a`XN#dl3k-1@dAN9Wy)UEXsKig(_%ee48$
zF)4Q|HI~OAWi5MF_+6O_nBhtvxa)1yB=B|Qdm;NTCENWvk!U*4?kUnT`5yazWUuua
z+b9fQ@xz%eJ~vVkws4a7+}dVoDrgXCdT1rCkZN9MIjQmPGvu?XD!p~3#=%Co()%Yb
zW1M7JKV33TQvwo2jgtqIps9^|xKQxjkRLyb7#@cYS5iCEKu+Xx?4Y!)@&v7Pj2~U>
zSoz|h2Cg{}SgA{}wLP4Q9D_Z(awXbfqI+Q-XfP6_9gUUXMEk!hk21X#EC@M>OEeX7
zkPkwL=a#{mEvN7&!46toc~taA&J@c`;7CU5ts`|;_@ce@p~|X@*{Yv?5&5RZi<BkF
z8qdoev;xim?^k<vj)u$Dp|az0ccn$-(Qa0mUg9@bS|K~U&)A-2?uE<tU+EWthBp>^
zHl<dsiW6!(tmfcQk%VTHdSpi!2x|S#Q)0xJc+T7ltKHuuS4i!-PQYP8`(O=N3SN5f
z^0ufmKs8|BNg4V;uIM%|c)c*@757jS!1|F-ty^kAFD%uGJJqLXoD`dMt%$Cu3N=W_
zrkR-KgdK+|QST$ZQCom*MGrk9M?a_U3KvzW;yCM@eeaslfpr1jT-mu41GU6(;cFd-
zz9O8%bjtk>^5}YYmj#NxEp8vCQC;^#l`aRf)z$Dz^mqk=&g`E2O5-r^7oq!tMb)ZY
zmbVLTLOI((NdGHl1*<QL#F06`#q!xgAC7Q31Cd5w<kD_eke5|lE8;F<hQIw$K{}^W
zo=k>&8M&PY$B~4${eB?l;8>LT({PG*Q@m4S3cheH@msnU{}Wh*3G0z<rGA)f<jbBg
zcKfXF4)k~<TKr1oIB6e$c5E}p`Aoh39C~hALgPB2lz11Zu<y8V+IIn+Ih~AOH?Jwc
z77XKsXy82qWCN=5N*|oBCzh*nWf)On#IL@G-O8=1A&_)w-4?x@u&5oFQSc&!4P&<I
zbWaHBA7$G0mMmz(u8Mrql@`%3<`lYrsk5(IbkztYk3OkYJsV3OiUv#)dzY~J@;JB~
zAg`&a5UxB_(u&j*)+}%aD%@k{_2J8roxiOF8`-6MbDaP!;~R%ekozmSQht){I^o1r
zChI7~WXl4H)7unx5X76LH7`dKUn6j1`yzbbQIxyrWb#cN7rPVtDjGa2`OR95$bhjt
z`iYngT-l`Dm>wTGzisOer%et?H62XfB`;!upT;U*dxjJ7QR!+IOQ=wHc7Qh*(!EWq
zzT#k9Y(5H8|Ek)Bx&2V*`!o@bmqqQiT+8_?S?5gjP0G!v>6;Sno%UCYSp0-usVlwB
z$E?^@jmj@sy4Xz_jU;2dQl1z+vpI4xcH9`dryMj?5-{?x3bT>h-LaVV6yDsnp!)J8
zU^t4ib3ygj*fv6j&Us9MT+#2;>81#kn(1P&JZYrzt3}J{du7AZH+lU#oq3*Kz2e^<
z5a-$dOPcQ3Fe|sJdiv&7qk)LNag$%K{wzP=yjLQj7{2S=U<Afbx{c#lf(Ue#8DDcA
zS2+=q11*aY!!NqSWo?^`v@M;l^&3xUDlpk_Pr5?r98KaZ5XH5K-c(hPAlpn8crV4|
z`HIZt%^YQ!dm%Gsh1q5M@++}x{)u)q6ACyts&xt|6~A()&}2i1z#gbpVn;NcV`@i?
zMJpTwKFeMFEm_3k)qN;a_A<WR#T;x{;G6tjKj1J?T&Ji=9L&IhHg=r2R<cOym7pKf
zu&`jt;U|lnyvF;ZU$D;sJJOYP(6F9<R?hz-2vIp+?by>rv#$I`fApHfdPHHvLgZ)>
z0d~a6v`U7SVt3&DdfrDKIYap3w3hL9W2f41v+Z)%z6kCknd}3ZCr?L1Ggv%X5XhI>
zO2u~mp`(_m<eSeu9*m6=r%8j?{w$&=5*)+w?22gcaxVDOoFN3xX1EYA0MEYA#J=K-
z21<fF_}Y7h)WG4kSyLXMu7IJ2y06y3)s!9+M+yRrKVj|e_9(j@F_lJH`>j>Szgr6q
z)-X32{jLgt9)EueD=?QDqDan%RIuYV6!3KHAT7I6cNkPc=E!NZOa6ieZ+soRO%r0Z
ze5EXk5gs5wwt2;HgSHNN9Lkw4k6ypLnhfXo0=C0peo`u7poP2lcV4{f{=MDfj;ob&
zX2pEaNWo3)!pN7ml$s-*M7JU<^4Za#1!o)l=#9wg;R|v}HTe~GavDppI-X?v=>X{_
zv=iMv)L3X0`C^P2pW+7awKo1{#nWzjx!8q6$%7G2@hSlFv8ag~1!~|v(_kazP`rK&
z5SR82b!o(XfCS<7kh%JtUeH@uX1D9sVLl$JVYM@Cv|}FkZD-K*2y^Y;PopAlzSi-Y
zL`j2=wNVe_=Awa}8-DP;lh{o+kMR9E$(MIr!HJ#sLMb20TTu`Q_cQp>zDlppbY?zl
zis9ZnJ>75US^pXh1okUQZQQR%Dnu&}wRnz4G*A|4vw}JqRQK++PbV?LwPEaf9omkA
zZ5xn92(yuS(t#(1X~Oh~01129TfeTWA8|Moywsn76~9enS*Pgp4$Hlx=<jod%uxU`
zdgSuGtYevAXm-l|G{axw>=ryCy^#zE8LAp;-HEF|2TJJiQC417kRKk6HtMfX())?M
zo^k#5)|t`Hqh4jOMP{B>m6u?*Ig(trB~U-s-!IegOXg2zILC@{i-g$`g|QeCH}X`y
zHEEjhtM%A1M8;SFvn2XVAohqekK+fYbe$=jqjxpgVVMR?#netl?Yy~kU)Kw4aRaFO
z9OyC8KOHkk0U`SvMThd2_bF)`_@J5**+w;nE29Y^r?=P@lfS{o84?#Wz?u^f8Ay2A
z$6H>uas`2t<S&x*d2ZG9kk7t~99tu~qV^l^+q4UL8Z2RtqsSP)aTK8ZdX3W#VoQ_9
zx(S=7t*53z?_BY(F>m1GE;1#e=()71E8UH7m*M@Pc0cL0ce9xy&QNE4{wwW_h=%9L
z%tD)zR}=v2tf-kF+C7C2IoMn;k%T~f()n2?Uyi^@RMMsVn(JnOpmk$yyMgr+zjIr>
zL=$|u#HKN%=K7nftTu6XqJW*q49A50=`Tu3Ay6Eaw#Pnyo2LktUZoeI$CFUQ{e5Tb
z=dXKsft^dn{)3DdGsXA5bC%;gZ-Yc}y?-jpyPM!HAq?Ihh6e|pajXZ9kRhHpd}Ky6
z+&6$p-ODhEe7BaS-784PJGYnwJ<6cRPv6y>)qB#|;8u@1Ry}Imx|{mp$quEFOVwK{
z^oeew_4oqN-ZTT6!I<F7binkUqS9xMy)aRnPHdiex$cbggh542kQJHMf6oH6BGCTZ
zn7hO5C$<SPUTy;poan7`N*;mf8<DlhK?~0yx!W~AJ7~hkE;f;4^E35}6%XR3@T_o`
zL0Z~PDqm>dg>;E_leG~`a=05Tid$jAqcSR<(l%Lx4+$~&-<&2vynwBaY82o#;zVIJ
z`?D*|tqz5g`ShVUH0=(T>w{22yvb#tgrmZQgg)n#z!J5^^eK$C?L6va>W1vE9~YdW
z`ZzrO1&|8)dAB>2+&^AAhUtOwxDmJkp<Cm@vt5Zn8vL1_Ala?&l_d2wFedl#S`@49
zX`u{`<_?QzO5_Qu^2Ghko}p+GG}V&2i?^gwnR6kgbY#LDDOc!-cyt2S_RFyQYKian
zy~C!+Ar|ut6g5<>oep<vORO{w^4^he<W)V}<4(fKu=GPfty<zv9nqXHN?Y54;!j4N
zB`D!=YjvIy+pvrM?EA<oE{^mmYC)3;j-hmQ@jXpTAxKB1a^i9iN{~@;9ZGgZp1Fxj
zUWdU1Wjr6}yqC=nd4>v!1*Tx;M;GcxzW1*;|DwSvPLj2F1eZiQ6x6?vNBh$hllTNq
zG2(TY@QFh1J;fLP^cXn6X$irYHm^1%2l5x)M|23>tE#LJz1g~K=h%zFaiTcV{p|81
zK<@pMoo~eVb9R;pHtlL?y*X#R<O>Z68O37H$bra}I9nmu+DASsZChw#YA*EI;rGWP
z9J1U^XD~=&;I~I!{2GdKUrF%?KDqZf5aL^AiC+&gJHrY6?^M%<HK-0gj*q$1N&lgW
z^jp1ZR|U2C=NmX?an*&cCd~%qF@aK;vIWhNqc>yFjY&A(@zGkVx6<V<W`qy*1q-KN
zsegK^Vj-%M4~dS8F{Ui7lBB*WaTE<GzdvGV%n(!dKmK?tX@nV5MEE&+*6Xg0DzlxL
zvKf3`yzhdA>-Y|P<Bs&L@<v5L75|V+>vzJ;+hm#|$|vR!xFX|WMVi+KQk5qJhe3*}
z!pK2sx7mG{c}L<8USV7=KyR1=sWl@qON{KdWgQv3f)f`#&ukD>Q#FvhmrayG#)J1*
zdwRGd4v;cSpf<REWn=#k)8p&Alp{0!n=W(o5xs6R-jywe7)(Tt;Z)PxzA>_AiE>l^
z*wim(_3&Z;gex;9NBRu~){gUIwczDpCw_ETZP97xZ&ftme&w!~FmGw69TDy*=Kj0j
z34{Vo)$5IvB`RH$^z@=BKV)@bxP3C&6%z39NZF{eXpr$T<827lYVzp9adP8$)=$z>
z`9nmw3i>ouG+xWPgQAxw%;>s(6TzmGl(?ai<QIwY-R&Qx($9Bf4hiWbC8<8Xb9#Pp
zRTfDz5~=SQ%$DE%vAp;<;=X3#mXr9OQc>LA2GYHF)#9nTQZAua<U4NDy=NU|h)B|M
z?k+3VpK9QEzw-q;mca~@QNlAeXyovHX0<>tihi60E=jxLt>ltoE7Fz63j)dxviUpl
z3M8}JzS~tEi<D(5d71m~#{#o;(>!KSYQ|Z#fLA;t^XQR}8k5m}jx{9hI`pC=$g29O
z_ALGc)v~|-`OCGUTJW~Bvy*xO3HSzf!;bwS9S2p$BrG$;I8T&cTIo!nSxug~yiVn5
zpP=)Ov~awhsPaD6P>&Vduf}suRX(xTTLWe2mgh(}_l~H3-ELKqY387Ik9_w^vMaiK
z&jmL~L~I_!Tntpv*c3Q5X)g8R){?^Z$^E>;BiX3vkk#|D{=%mnu>u*Ev?nTIwRM}5
z>wWSO%yola%pc}yC3Sw9?Oz!-@_r1+jYu^?kM;JCpL^y$YO5!>0PYW)ll!FFc)EtY
zrOVqv;OGc5qtveR!edGON0aoafJhX2^h-Nu^N8CCo)jw{mRykJ&Hcm-F<|6;ls-d(
zky09y0}i<Y!o=hR?$K21EyD3y>Kk#rWdt%(G78EZGRXlj6@Pk*=vE-^<Sy$%+Lt;6
zdhOWG(`h#LdcJbZG_~0ZWf{MM?E5{hR|S{aX7D<$^mx5@ADoTmzK5H_CpJ;0#ej^o
z$%mpyHn&%nG4i;Vd#kN`lO?RD_f6u+`Z@6nU3K{9nM;(CM9Yg@TF*49FKq^StH_&X
zu*B4#X%9z;;<BG#9Z;?uo5{Ub$gM_*mkJUYx(3?b*!ifxE~l#e18&~8FgrCbNxGUQ
zW(vdCtOua(c`3g?*-X0e-CG8H_?pO>%WviRy>)<hKzrcf$cfRfqu#zU7x|5I{fY;U
zFc#wlALGB9gn}*ZX4n!UzWN`YoSb#4{w)unRcB%F#?q>qDIxwzw%)_5vs3~nfoe<C
z^fOMQK0()bN1j!6!&JIx<C3mrh<k<aE2)@V?PkUUUzr0U^6HQ;DKD=!jx~jIq8odk
z^sY|l6j9+4jmKx&aZ?)DNLfndgU^CJ%oq$`P)e25lcCH<q2By-{AU3{9<KD*Z^F5>
zh|?YqnTDEQU)B$(05X(lr$XVQaPqyJxl3y+l$A-^z+UP}VJ)(YR3}Mh(TUVA*&t}2
zM@8bQ^F2mm1{iI`q*1S?>M_v)3KzH4C_z%@8GRjF3`EY=5F3uy>sRutOumctpw)5@
zm!)*IVC<*FBuV#YG5}ev8`IxZ3T<v`xq>J>Kg=e4R*@gpOcg7xslr9>)2wM}e^$Ei
zaTTC5gt6?uNcINf3Bs$>fNmiJZ!nNh!Om!PZmctr2nXNSvk3$p3Ei64>9D!geK+O7
zaj)V>5{#I$6uX;2qdF6kY1_Y9{Pxj7hKydBfD|xO_D5momnLMrO~mKqo0&_bo7{Gx
z{4gcrgG=_sK>iPzuP1Qxpzn?vlsO)F!CF<Wu#ASy$Xe1*%GX+MwUpf$a{>yC;0w7^
z*5m-|AM_QVYQwZ!i2?9)H}RkYuZh{uls4P;6XNr0*JLTc+1epeMtlY23Da2KnTPfn
zoC662Wy%2;bSd$13_$Ok;7J~AH!*mWhbJ+(ma@fPn7GWB1KCLM`%?|n=OYV2I&hDK
z9_M8ZvkA$&0vNc>pbfy2AL3f$ORJSuk=|WIWv{7l?)%?fQ*2tYf`JKdsgL2)p(cJ9
z`-Leuq(*W@Sq?MEaOv-Z?068HKyPA-_1G2OkK^9VFwgK^H(-Mf0d5$<l-kF8=wT6f
zPogn?W#!c53!p3Zi$O7Pq(XQT6e<LaQiK$6XTR0I5XGA%)u?jWhJQVb0AMU%Ql^gK
z%@jYy>{}XCxq@Zn>`6B-*+1^(#n`&=Ob0xLI=c2d)+Tw^*~eIu7j(3SMf@-qK8E+q
zdpeqM^g+&$2{8Q3i?=%V3{u7gZmQp)#9!z%6oRU6WETDJ$Xe?4YGl_9Eorc-5O9Ab
zMfpB-pVxlkBQNI0F?kL(ZoHKUo6nE_IOTkRS|wJrhVACcNNBw5P}lMV&oopNLPx5Y
za5t|2+is%RCsa__n;p#e)H4BI4HS4r<W#QWzn})SN`LZtZVelzfP7&BtG0D++yY#m
z+ZK-Uw_FCHDJa%8Wr^Z=_VZ*eg0^A+RhjlRZ}9BLMioV$!uLh^sVWD-Rk+d(Eh3x(
zbj|su*j1uaC=Wbd*#N!ISp!bs5wkWGW4I!#5~oqvaD`{0$+P5sGQ1oG*pbVQ2?VD$
zUuh-p>Y^Z1JlkmaN-G1WVA@1rQqaD@`o20P{yyZAI@BsZF%}-+5S@1tpOWkZtQf-=
z-o{=a2Nw5`SAaA4ZCI(b((@hS2ce^Sq!^EPyi$OB_!Ai6rd+@?JtEA#znci6E)Ryn
zQeVWXhowpw5#a^xU=-MTF7%2^w|~;@AXz$PBMIU%a}K0}nz7JUBwKXWPJFPem?3Qa
z9e3F@(dH&|B@Wf7?o0k4QNa-=MgeHKxi3(bQLWP;Nip_T!G@_jy~0Q*e&S89ptLK1
z`;Eu(!TaTxaHJ6a%+=@Mnfx1C9V)9A$u^&8jko>@PxTAEfJ6Hes}o@;T+CdZRUjYs
z)J_Q`?^>`x&d`R|YbWpk^4mo~cTp4#ep??bJcB{Y{d{X!GiX7)8!1c8@fMD2A*eQi
z(fS5MUO6%pbYXT3FbXoHky)`pT6Gh@rNSL^3cv)P7k{*ZHN%L@7@Jw4Ei0GnV7CCg
zr~Ii$xhO7RcusVC)|?P;HF=*}>nO1)!#Q+xlv0w;xd`C_obktM9igaaT@XGhlT4R@
znvsitvPtlah1*s<7=@)5k+-*E3JI@r;>NB3@;?OWtzr6=$V+ZdlhL4<HSCmh5?6=p
zy7n%E8W+DG@QfNKXivWB)&C8o#u2&!HKcU7gbaRIJCSKhup!spqmKUSp?m>CeBzz<
z1W)O!Vq-A<<ltI*<!yeFP3wzbo8MOze^O<AKH@Ta4&>m1WeN%Y!LWTA4PK=amqx=q
z3Jbv8JZh#E%cH_kW{3e<R%U;G0}l!AUP@H(BREnClhq-`&rN=QU;r}yp4K77n24g?
z1;$}LRY<F~JrJPds>)5H+oJ^h+;obMKTmhFz-T{zp%KJ2R>l($;v1(mNbwZ?46l^Y
zL9JJSjnA=*7pQQ{Cj2mS0Y#Z_Dy9&HDb|>52N`x#mI<%iG7n0%e#Vbpsj-R{#U(=b
z4(v~{vV(ECaNOMUGYKl3t!;k-4Zgq)@FJta=`0VQXb|C<$x&`VQJh$SX&jV(ac(Aw
z?z5Fvp@t|@Jn5u#wldn;vD}ap;UnJ4iy;G3LS--0z)y+x59=@BCRcvK0$gq$WlNnq
zpvv|VAm6B}rGV=Usxfv-u)ruDkOMJTofsdLLNQ{jLV%kZ5yItgG|X;Vf)%cgo>D;n
z#lgep!o*mO1uY1-J;ZdIw#VYp$~dOlMN}uQsV&{2TJ>1O8|I*V3|A*?{&k5^a$Eii
zFcB{OJk(H&60uGOxz%z-;g+>ndLAtv_1x;Fe1~xX0M|z8g9mnw%<n_Jll~BYNj`Ya
zUdfLvEH@&>41M2br^fIv*_adIEgd=7Whv+}Z*7t*x6|x8NU=f8m@+;n^-SZMj!?(f
za+wl>HOxgfBS#q$@?aQIh;_-AcGZaR@{4&ex1%~#p4~-mSVC+vMcLdHV+J_>A)|DM
zL`Ba$32603fKbvugbVlcF0fg7Q{@s61)bG6cvM1AW@M--r*a9L=-CYSCKfkYw2T$S
zISGfVHnK$rPG7>Uxd6QH8}&)?wQe(SiE<8pM3Bd4M7}Rns~A*us=v^o!XJqhL=PLQ
zEyGUlbDddr`U|hPPjez1^X~>3<^hK@L57h~Midm;vwpI>83tzLOTt;sqHb$Dz<4CS
zcDG5%fY(n|?*ht|Z%?|@Z{Os`)jhf6t1|iU_E8zXSROIj?R_tAj>~6WMBTzFUAzK%
zjShRs<<s}T7>&$@7JIPR1h<+W4`o-zb856(_w{ZVkKn7v2gq9yq7dv9EwTw*>GS-A
zrtaK0qHn9O_V<@uVNP6m^=IR9e~rM&PIzXa!L4Sme&meFfVASr`s(?H2B(V2^tW_e
zk>4v8zb@AL2=CH}Z7TW|yYzIU8b(GH9#x#7R9$!kX>GPY?yaVsgk4)ZA$8)}&s5|r
zu6ldIFnfjWBp;Wq=-1x9aMa<$W3vNR?P?!>UFq5W?SZ`_@Mk2L@zo-nqsLgtMa|Ke
z!B?oLvNPTD^@XFO!9DF|u2U2AmT5?x+Ug%UF&`KD#XorhmN6PU&o^ff2?lXS>CH1=
zW}DO=Ph3khh&!nPTbS?*tsWy?uJqj_++kvvxo%;a8qg~0(3O6tNyucO?QI`Ji&`UX
z->-!laH-xgIqY4Q&g3N=ca{A`58%rrA<#j5Wr@0Pm3>w_RO6YQ2DF|TIJPIF_X@`#
zFJV(p!dcPVX_~HJVVN!=ruyoy1$a8mO&^pM&CNUt-0T1pRBz@EmBPlEbkEcshKy4n
z%QrT;p9tSW2B|R(j#Z3(JiIbV4$S>28we`1E-a`r<Q&w85|s6^JU)UiFETa*iC1{g
z<u6a*%U-ji{D}qT{6v)4KimZIjrwi_KzAFJD|pYogE439Dl#H?^gz#;k`k;HGGWAY
z@SZnAFcv~roLe^>*w>Ac;!N0z_{x!>$q4e9!rNi&;EU7V^tTqdxcE(YLOYw`cz)Hn
zWxjn6>{CWf^Jo$Fu2a1DGQ^~U22#qx3xG)<lTQVJhwc>3K*<zzI^EV_j6acV;-)a@
zJ1K&8i}|qt*UrwPri*{clnyjp%FBb!t#Q$n4Z@8ng>je8;Ox+eMg%*UtM^<*l?z&i
z49j6{Sh+KSIHrezS+nqjpI!Y*4OtM!<>p^YX8B2pWq=a9`L`f?`=YxJXtUj!2|%$_
zfTxS4|MN3#&Z$%j1@?;ACcr2F*H;L-!iGMH)sgPV?SL`Z6UWy><GxXY4g)g6p4C5u
zE1gh^kr3l_TdB)VrTK~Dr7Tl>fNpWQ5LSdwE&m!Uq0{P;^T-oOpz=3umREKb1DkHr
z@Z{5ANOaHpI*+S%`46YtG_U`l=yNjeg>qTr8)uG>SW@J>7GhWAOSBdOC;NCC*dYny
z^M)_4@t55cd?oo@gA;t5(0>7nEis+$*{6!K^tB*cQ?R#A<<|2s!&yhBT%bWbJ8yz~
zGnx|@57{>TMME`-CJ>yTf(3$yZZ?Uxd|3tID*SXQrIl0H>>(-dz>SCX@9-mW212($
z>*oBtASAVNWIupd3jV~4@Zs*}&6y6%robY^qJX&~MqAImSqj~j!XT9DEoSsq?JJ5t
z|8FlN-&vG28G#E_4~TH}S%w2$(mYM@a+dD?3wS$La5DkeI3*>yh;xatzxWz333zZj
z2HM`X3H`%Po-;jtcy<|mvV0+9{`1s1W~P&^?EM_RC(xnmOG!XJZ_t{q|E>z$MlSB$
zs2a%!fYA2@a>>^J1binS`6!nd-CwEo9f{lfa{^D;`$A08=Rc(yWN0g|NrE2ENs~cu
zOXmsUB*lTbD+;*v7`t;T3Txh)-8KOi?ky;wPq6aR`W32&l|gpW(j0d{UkPhL+~#>k
z{cKTVr_Kq`$-2FjP+U1rfpzsT-U`+bx)&t-GSYBe{v3{_RKjmmxK=17>7gQbVn)fu
z&Wo^L6<LF6qpkoq3mk((?3%s!R0^AjA%!MS2Jtz~Ngg}SkSAx%;MQ*>zu0y<;CCq(
z^1M@XYdVX&A8c8IhlD(5Z1kNBWWLcdZ25^_2Joe?|BMoS@`75rqeS7hRwl5g_eOvh
zd~7-mbYS>qp+Ic3|4rcbaM>Pa{51%b{#8cl(@dn#{zQ(t6N>pDO~=EN72p8@643X-
zmKND_C?)wAt4Pf>+++j`$Cz9}9qk;i8;bB}yjohBCg_S1wqEEU?=RB$Mr9WyKP#t3
zE(yBAHf&JU|3Ut>ZYL)xXV}6U)|7iBk`XhAdgOc!=<9AT+s}(BIrvLRdl3^f@sIfu
zJOPx1@qkj1n%4W!QCo3M8uX(G@7IV^dn6YMn##Y-IVj&p4Vn!8j@0no--;Y{;)8OQ
zqe1g3cvz5n3ZQCt70lqnhfE|4h^DtGBH*D1`I#7^?n*{Hym>q=S)m5BxSuDW10Mb~
z;F^XQzL1<k2V7X8eIF`Zr#25ykB#+3K}u|2HP^N@6A=2U`gjUKN{@C!{-o1&1%`ML
zEozU$zsk5RQR=$IXvod{nL#YOe#{$V**xuiPK}%1hRWZSYp1|j3B&0{Nd12sJ$HP(
z9O!R*Jpq^#fY&?IlYlNJXhJBkPYLM7jRv~*TwEm3zNDR~NelKfDfqLu0z8;Xe)N_C
zqn13DRm_?VEKyX09MDpT)FMJRSLvN(3vdyk`}OOTaW%ZCvp0ayH&rQlv2VYDol~bP
z(&!VKTB%U)8F47OT&t;o5$bLLu;g;(1(2xQ9-%>7bkiFNod<>Dd%VnJycPDY#WJ{!
z6XCP?G)xG+O>)^X<||6-D0(vtg5(^;J!F%dn48zZULv(w3T(f7<$Rsm-#_o01Q=_Q
zf%C0TD?`8Tjn7lYyUzDBBkbd6c=0A{kQmsohO1|PR85*h6nh1~lhCyQ4ft!od5_B1
zVH)ei&wq#27V?SZ(P9`OVwIDn-4H|AzYh;vWPwmB_yVXCJFibrNf|Q95<@sc<cAl0
zEez4|6Ky{v)aPktTyh9&&a{MQP&F8pV1kwbp^p>2TNdEO1QT3(LwpxY{G`-Qq{g~`
z22eww)`}X8tA7Gtc8xwlK_G0s5U<dF^AlsvJd3V!PU|djpLL9|3jbP0!tbzO)oM(z
z^e3Oakx=}%0J9p92tDi+y~l#(oWG_CVUb5)5i;cS1qt+4$jEJ!{~KaxB8Zvs57X(q
z&tPG-LS`@*9Xc@ZJC;-u{e<HLzI8IXv`6>|+V@#{gX6_Ve~D+#IUq9p0Hd6WOxu_u
z2OJ+kU%DmoX>&>)PnovZ1fWv?uyatA{~lu!&|y-`zpeltb}l&;+E_>uhHLO#gGxnm
z=sw@~XJ385!zPjHu6*b@+=RiybN<Dez|7S;-1A$y^++i389M!UmtkAor_0Vi=6*+L
z)aS$FgW>(VevR;)R(4kW^nu;^q4>(G9*C5(zVPM*aIMhAL(l6uaynV!|LEdd)J#t=
zBm}=}VO*us{dK308MD)*0WATzy8}mVO}{Qvv6@=nf}UTj_`{cDRgX~oi@)uuHzz9(
z?1r$FPqM1Es!{U|bOXG`pWrI<CMF(J2ufh88nma0Us3AvC%hG8=*$bzaE%y71YYhk
z38|K>Ked1KVbk7AaR+uw7b{&k53VgxM!oAaaKgqPIjd2vefcST=-9hd`8eumfOjii
z?r;Tv!)RQ<ar}i-i{XG$Lq>8AqOP)jrO-b9QM<6&!ocCJ*@f(W?L{Bhy@x%h#gx&Y
zI-k*d$r+JpQ%1%~*TfjKO$Ms<JklaXcufYhiNq+jRo=G0j7+eQ?QNTHo>VZs&M_-X
zmo?ee-L@axv;A6!M$5z`E;(#S?^x7ve0EZNTP36i6(WP9qkWYXrw_fv*d(`vW+=l(
zRfD%mPpj)+?DEZuoK$UWS28g>-j+|MJA3-JrFl9txfpFSz)*2!fBPthNbIxeLjST$
z`P<K*NZv4cMQN=l6Q5noi2UYfzXTnQC{Zjv0pqe;loaE-iDix~Ud%06dg>V0`wykR
zg$z2X<PbCaxtP5y?l>N(qJ`Bp<w#3U^*(fbUlNP_q$cisNB|uc=V1tpyG)r<>$AUV
z4hd*F%WdG{?ScG=Jm*3ktv-X=GM2NRbWa63vCmuED)<j&U5U`%^vmn-M|Pn%%#Je@
z)cM!^Ga79A3;8JmMhcZvTcz<<w(neh56rI4JV`zMKu2pi(p6LaVK48d=5cY_A6?9P
zdV&8<E$BO!aB721MLYp?U(3s$c86V|uN&#%nTp*880fr0Q<lj<m$+P*+12ZJE_-zs
zs$I=8ySj2WcIMj3y|(1Ku~^O)$6-3AQCwP+Ek%Qk+hRuEiJH!Hmo+Q3351|TdPRCc
zCmWFHN{^^ZO^N{O0#^Ge5Gys|k?(lRNM+FZP|+}+mw_kruxWThh3b|Dl#Niz24qH~
zf;8SPyoKohLM|mHfgkt5!f+s`pAE5+2YU^4JgkRaEIO47RYo@-U4Ak&2faUFHI%5q
zj6QK}J)grbsID^hks)636C&znf6>KDC3r!CdAh<IJfyiHjkVb@$^^#Jd;t}Hc@#q?
z&@p8Oy_B=rE{5cfdyor97xOkqMSM*k-VXjcKED88Hn_Hd26N$mCTQbbo%LE$f+6#s
z8M?~g920<!U?Q!Cm+gR2TNXq^&6raI*m8HyK5E|2xnBZ)Zrc2xAc)5+#+Z0=0(?$v
z;OVZ2LzUc^YizJ(og3v*SN(b!U<uG(dHU&=98~o+)~z8X*2s(!jAsGv^^@d`mF1C3
zvgE<hTdBnq@z;tLiQ}tkzM=-Rp4|tQN)mvbFVv67frDZF9U%fN<Ut>Eg>IOjCX!!<
zi<mUNN*p=?wwf~sEn+gg;IkC*js4otF3y2i&S^3^tt&XcE)G>(8%Jiy`qK*#iI(7e
zk>};)zYge!>W(z@=6By}<yYwGI><u}WMwzmLiEy(hV=ii4HUaUUbWA&bx<YN;AyS(
z0lBp~@+yP&nf<;^k|EZj`#CY;>FszOPH?rIh)y@KGyV3<F|`|zih7b{2#e@jOjd*!
z``8so&Ow^JS$CeM?9d!4gQJ{wF_$56mR*?<?hQ+V>K_};V~b^)GD_!LAIT7YmRXSS
z{$`D&&ZdHj#+XJB9rP&N{6Rl2zCwjAK@r|O!{4%L`idZa<Id)T<k<F*tq3*-2me)K
z?5eVQRHVlEt5RgHtD85{)bORtup&Hwc!mu{4@=1B69bkU{h@-s9!60n_e8e~8cg&%
zBc7`8ApnUJXe$|EuaTB_!k9eeO?jCO*1uw@fNr9qmF`%n2*pNzWW?_gv|nX|q`{a4
z6KT5)gv9UNRdi_Iv=L6kcN=ye5ZRka2DeHy-wI~K+{%~18UJbaO;>sOIlTlAUzYcj
z#vW<)J>+!-3EJO~E)hVCk3n(~{dqmXh!-T7^G)T&G5U6JVpuMF))S$(L{ccqNWV+Z
zKY{g|TSBErK3pGcA60<6xe{d%5O5lv|6hPpf>Lvd1C<MNFQ+*%W^#845!cVERIFS{
zpfJ9=WU?SpW;+3SfNCdJK^bhfo_`%=*HtToiz9eip-hOF$<86|vpfFAiJ`q*T)VHj
z`XK27SR1dvh~a%rHWdJz4MA#%j@bWo$^}SG1SWejKgc(SUTq*p?1a#vcX2gBxPw*7
zLLnTAKz*YTa_kC~mORY^(<UlPussV>NzX*P2QCx5>AR%^jt}GPiq(U@)+ooih4y_6
z<3;-(Bt-6u(3MGizwu51E>5(6Pht~*I4X{y@T$`9kkscS)yo1<NTdhDMxTn;N(I+m
z_;m-$bUh7e><z-FC|_tsLk)i-Ws|sNeUOtB52WZ9$*_5lfvH*R4A5$mkckm<+s-jb
z0ZvPFI3LZ6o7qVa!aXE#jHUAib8iZ}^EiA|Z~8$~CgwCur38#^$131-9A~qf(pt3l
zWTye0QkoE6T$=CsE>kgF7HsE&hM{>-iQ-e4m%l#<<CtoS2I_Y6b>XSmtc~z#y2?2Y
z9gr_*I{znpS!~k~&Orug+c@GzTmK&vWh^_(G@AmxxFzCv#qK!@EXnr_cqUL4syqjq
zWW=)vE%{y)hH+T);QMm+^L-QvK6+OTFy?qVVN4g8VOuoWZ!nkDB)m~hgq5F(co0i-
zAH!sm%|f<9_Y2JNa7xAhr;h6mXR{0YM(wJ^-bAgo6xG@UF>6*^t*!Q`^{QBjQ7ej?
zZ_QF$Q7vNCstTnxwUr9BBO><vp1$w*$2b3Zt|uqYb?$SYbDw*i`}~64EK>dVQ)Dqg
zpYBUggLI)!g)vcSen7m^+V;t@jdN|BAJ9{!Sq5{OaFSa<B&=-#F&xudXYgSke$xa5
zBJ{RO;y5Y-eJij}nc3fR1;Mzf_jK)zP^r^(jz*|B()ebbT+k)S(?UQRiq6e_d9Qsj
z0cMYIPg3FSVR+BBRj|4xmA9cppn5#TlxKeq1LpldETwMH5JX6prd0ufb5ErlS$p^3
zA2GPu>C(!tDmW0G1@sz3oM*y{z4I3W7TC%jC<S^n%7nEc!LI@CndlF>!Tsn)viQmO
z6sTo+PC5-Zeul{op?PEk5Nk>1QIT;Pq{FtW9_j;|YVO_&KWAKaOh6I)h3ilKtua7`
zL$64F&f1Bd;)!SLYJ~{X7ai&cqKe|~Gub(i21!jZVTXhkGprdQf1OXgCHp+IAI`4Q
zn{nR=g%2A!3jn6j>QPo-%e|ZtIzVl?YWZR*M<4!?KBXckW>NLyS+IJGNj&d)kh)SP
zO}v(LE5zEvLFvgV{m9mhA4~$}MiP|4>KZny(m}MpA+$U<bDf+o{%n>w3cCrKlS|;$
z^&<^|JU~O@NHxA>(zg|2NU~G~sE^fWan`{3(@9qr{DU{|QX}LXvYl%nksJi{7K0n1
z6VRxd6yCea?!iI}I(7;j_?8K*Od>f@R+IxiR#VbWPXS_9o=g$UEihQC(^Lh7m-$?|
zA05DN!PWYUJlU^go1UOv-@XqxQ$H!<XVY>)Bf-c7iH9e!yL=|UA%7B_Xfe1t%{2e+
z&zSr;k9hxhKpEY3NAlM=MFJDZ<qIrDcQ5P*pHE$UU8E0qkV!z{c+DcH127M4Kqe)q
z;qQyk4Svs$5uxU9<kvUorukEzd=z7v>~HK@rw1oU8FG%13It-t$&!`1*2t5$ER*R4
z0N>^qjV1w1^7z(k{V*L~A3&oGPzS?W#OOv`7V~06_qJk(m%8hW0fCyVo?eY50TRF%
zHA}A-!$4etUozwG)=?KcXS@!^a)9mVmNtq>s(<;YcvMzDW}rktx?7ETv5SmHrH(1R
z6zt1%$tvPF@?@FHZNUEacfl;@=zT!b;VOVZ;~TnTI`io>wN$K8VQViP==iqNK=e2;
z1y#Zb2xlK)r=VCjvz6n}vCCB;+|jfUuKL?2dlX>)Ql6bZWsDx^T&X?`0T~UkK=6;6
zvLX0u4z(l23rLc26F`G77EzDliFeOe0l6~ysDPk%vQWfuc0IiJv8tG@W!mAiK)sCT
z>5*y?@)Sr7?{cQG2s8;MARs9XhDj0{Gr|oF8Ycp&3>p(w(TA+b9tJ?4PZR@DPrAef
z@*@GV=|6$NvSlLpg`+uF=~X~<9FywEdXz^(GfV{}!eOy*8>W*hs?i48<anY3e5&>K
z3JOu%PA$Hbe~Bk2T&|s7p9h&WG^>Etx%^Z>iX3defv;pYuAm%(-NMWGejlrAIF@gu
zmm~RErK+M%R1s5+q7tftvkx~HqX^G~B~|<FEX#qz!>5dwX#Ky)muTMuX1k6iKU~W~
zK?Ou3<Pf1_=qCm;qeBC9L{fq66&~34^dDlqLaq)#xa;@)Zs8wR^o&D5c#8AkKUuj%
zXS0<(#xw>DyjwGn4~0&DX&aU?9$@~1O2CzZ>sY-AuY0O$>82C}tn0mJUtyL%czi%+
z3n40?TkMIHyP&9RECcg~04LI+e5k$t1qM6iVZ~Sb;W|TSfvBKD_?y_^{zldf@?_rC
z5+I!wfMT(eS;_@75R7k}jos-@h}YX!(#WsnF6`q4hA1293D>x-D)?dQDX_dnfhdP>
zp5Z+~HGs)qodZm^^xpy)2Ovd&|2DV<2Z{>>LoRyAl5>G-1YL4pQ<Qjub`7BPz$K2E
zW8I6up!AH%FwKTLuIS8WfUQosll}0iWliMu6>^_4LRdC~#!PV*0|p<5S!cqxYU;mi
z-+OlY9DIu-vFPYI*wqNw=Bwy&MbW+RBh7l`a%j6acIlnX7%^NYyNMWnR74;{GhjHn
z_P!mwB(>A9LdqcdWj++=$sMV3o0{0v1EMHFp#ox1gPbRe;l;6jE@iURE47-H+~s*P
zr>a{MS__6psyY52LtZbF`#@g^-ul8|$F~$K$cOrUvU$qQTX0d{vfmHYhwdMS%Q5Yl
z(Nm)26Z~ZYQP8w*=!YjdYEWxpxSW6n-c2*C$DclY_|?S8z7`)UhIH@uIz54Md2A>d
z!EZ`>t|G)hc-82>ZEe<GE2@4LZe(Gs>$NgMg@ure5+W9=j;)q(&v^^xzvMV(F3_L6
z<>W(IsX_&I!jF_*2wj!}3wKb#V&^%WIEmm_0aU&{X~D2NQ;4}a-hYY;JIfNz{o+*h
zj9wKQu^*kG&KGzX{*+};`Zga***8^0MFUk#fr|T1)^r6hmWPgFpt?YQT+XRl%Z%BA
z+_7q<yHepgd3E~;u8IL;_71wheicLv3z1tS_mNDx6&C`s3F~**KUTGTOa@=_69UX!
zjRk)}LASl5(jS4M5FpFY>HY@$G0TdM-?w(ry*Lb?>I>lLqr#>zo?JIm0ZEW3C593s
zP)<sB$cKvm(D14PgtOWDO)Qsdiybg1S?UkY7w=^KKvzMVu%!lq4d6Y^FMv7eHG6ow
zsjzp#ULCmsUb0s*U;aHaeulRGMXKtnd$VDehvDqBh4>EVR25JKDI!Kf_8+&4KzEvn
zHs{i4<5P?@bqY6on!%vq(VNVsgMhRdzf$7-4*QYq*dRvKO!_#U$!vp4-PvSH2mZV2
z&A+nY@iKrg+-7u9E`7dwpB!}ZdCQ^YyjJ~;9UXhT@3!z0wgWR6iP^|^clq_+OaSv9
zMB0}wA%7T{x8%$;3>{$gKd+o^Xm#H?pK^V5XOXSPB-#Ke<6|29qQ$7X7^>3Ziy;ML
zsb+L+G>Cm3A)vQhq5G-+CT+25i8*z!$}5oL=t4HboJ>_#7Y8D_UYpzt7laW8b~WM~
z36!TuP3gpaIW4(%sjVz=-V&l-Nt$$PFQ($MmAQQ@U7J82Jkp*g_nG(%Y{wTiz*XB&
zwt%Y(9n|+A9(8aX@|~<NAiFxRpln=-JLAi(H0=g94JCu=_J>ufvDxkub8xmcZ@Z|B
zUFtoXdwg9CmWe4LXJ@0mm&3=TzV7KYYB_JIy1laER`_6i;&G!o+b;8-+DI-9T=um*
z=8VyGOTaaqEUeJZ8N>81oCCL)0;vPB>tdH4ip|R|(|x_kYjJR>a2?XmppoQ3c1&yP
zb@Tql9jY45ESP0dp~qzdq(y}}2WcmBl@$_FJ}~`Ll2;w_+y;yJI$q1$+aMk641%<=
zfXVjrHG;!}=^PT~D35Q}zfA80YWr`J^(zutlvsChR+;D}BkEoXkF1GnEc~7mTC|&b
zF_Ekqr!ZRP8*|+CIfEBtAFaG6L8hqjI&_kAT#FQ?mdM-9wnia`0TX4LkxFGSLp62e
zmpS*&l-nY>^?Rg4#@WGig(o;STB0#i;(lM10c`%-d)C~23N&uun?j_^V=_$JjGT{f
z+V^a;(<UdDyWBGGN-hRJ3*i)TwQs@sMo<o!5)P#_`J#@doFS4lQm_E$m89asc~laG
zi&j9gaG*10yvK_73nfs6>1R>5ZrpNaGZqAuCbt@*GA2<0L8?Z(XXyTNd_U9rNav`r
zJb%IR#vaKRSW~3g+lu1fQ7OgRs<57vTFmEe<5eRiy>_Cji2Ab-rmz(Wf6aOwU(`8C
zc^L0v@w;|vg2k?`^V#)f&_yQAU7IidpFec5htZNe)`d$Wy$w39J(%54eN#D;SU~cy
zCI!8caHo?EF$1>@O+1kJHeYq}X`-EOh=rOBGBE)~R8gz8S5h*x2!j}^g{pFBAH#)|
zlZqX#=+ZPnQ$y}La%>_!&<x*KTj;cY#Ro&f?}*c4O=$CEa$c8!j?JW)ZH;9=>|D=-
zKdP81GBA-^qn_D_rrK2n=Xgb+VxB0aJE8l0--*vw>-E@rfL2L0G?-(QhV5-9V|ejr
zEWPH2xj(B_EuHl(L;-}6PvK%qC;l|GC$w*=kom5@DTOF<jXkln&d~X4kv}r_+)KZ%
z`kE#RHRYzRf|%^DRm;&Y_f5@X=~n{VSrM<h<m&Q0;-Ch@X@b3x!g@D<be)j$fy^QB
zNoXYl0tMnLAa~_}l=v8dTHO$!?vKKxSU-iN95Hd(+q^4dgKQ)L<U3zI^e0|e+n33p
z?AMI9xn70JaH@uM$WF?k=zKg_cg<o<lZ3c#)G<LkSy(l1Bky+=CN)%U&M^vL8gx7k
zc0zhs-$!Kpi!7D4QX;sg`{Abi;O?>ZL?Uri-R$f^rI7gd0tveLYb5ht9W0O!QI59v
zp5ADdM0@!<O23ikTX4%063h%t$NO96w4xKMzGb^gm&=zVFn6EeDZbrpg!IF}g(?qk
zZ_oavuVWF{@Kx?FADBfxk~#Eeifn8MBd6iTm@}gBK-c7@kVmpS$};L{HTa^Z<}>3P
zg+X<?XQoS(Z+!Ys4KiZd^)~m(A3g`CKWFx&wNqD$&BzF_P$~+2D9x2RYF+Uvt{!nw
zOdy?>dsOn!k~Q>2mqHHZ1jlEu&(&5+B1PwEOxyx_a%kTOTV0egJ{VV6QT?VRe8Nb2
zWPK|i$>{zzhG!=UX=_%GvQOCDZiHjdmt~c1Nfo7;@JBk$Kd#ltU&Z<13SEGr4LJ9D
zZ2)hT`6%mdGxhMe42I9ldB1Y0q^i}!Z`}KSguR`4aBGNni+I7(-esyMZ2|aoy4P;%
zgP)H2v~5k@xU57M9l|q&``P;{gIU?+Qk6<BpFK%-=KHR*c_+kPAMlQHFg74^u=b*B
z@|>-f9;eutHeR1mgY~uFtTKaI-*Q95y19NkvbyiKz=E)v`RuMm<vSA_wMLKm%YOV?
z%U5YfMrjY%kr_fL2mKk7u`S4eR_F<R6Z%27_OrR+or))ORe>_^tKP)2Me^NUfe-W%
zW$F-?@4BRyLq~N${=M<1o*T(yaORiB5t6~V^Sqqt69ymoUO!Jhm;~=D4=8K)j-p}l
zp&U0uO9%AWa`s;+_ZYoqZ?7Mi?hB~?mGeETuHMTbyMft@>EkQcQrh1bcLl*^Ph@yX
z-fs`sDcFvf(WgCIvpjrp-Je_z<zRIfpl4HtLyrtSFMI3riz-?Ub>s42@y(%C6e)nU
zbKToMA&7*Q`G;_&jN;4T62BaAdcMvHF}fXi@Au_T!2M!{fhb?QnE-qCzrV<p_XCGB
z(6BBwrRc=370S60xz0(c?<CJ;c&VhM{t~(?k|#SRTu!P$#qu@bjkFwUh((01DoBH(
zCd22rV*~b_C|OoszkcHp^H)W66$Nb8fEUVjhV+*8iy??bkuv5^)O|OLzq5wapSeI|
z(8yjjjo6>|7xXu-AqhnX2<m#aCcNmJ-;9iNyGwTefP5!y<l{}t5<+pPyNjNi*+?^E
z*WJCTbD`)#iHnWc2zf{m*`uL&w#V0JhYRQg*+>z7l=odP>+9baj!+BH`}=}Ke&42-
zyuuQ#Z+3He3nYu~r*dJ8#7YZZ69_5f<Q+Cz@pV_kgX4D7vVE>jcfXoQce5<<U+*8Q
z^Rn93pT4D<%rLQ{UtRhWRW@s^7^kdYi`78{nN~9Yh-31qxQ<>Lne%rHPZAq)v>}SR
z%pT@MLAjSmz{-W0+dr$nd_2~3DJ`oz#ooIVZF)MWrm+@hh+p(j?P+iNbx|Tp31__u
zfh|Ei=Z=Ds-RRX%Sf|T)bP|H^XSeisTASHr$M~;QMSZ#zr~X%H?yAFsHF|L37~HoJ
z?Q@Uy95fnRuj!D0jydWaTD@*wRTW&lKb7T$dDMkJAcE?2vL-%nM8CrAuJD{XtS}w6
z1t}1(oeCB%%sf_PZSV^`j+)&+j9`Mv%+(I%*4KWJ0Uvs(xm=X@c6P=fYisSd=klh<
z(5X`td~ovzZpTsEefM1tw9=I7BXY`pK}-dg>~nZdPV+i5?LAea$Pykk3))|7wi+mW
zf_v~*9+=_k)1BSEI&J0GZJ9?Agse)*iA^ak+1t+(oCcKxlrMU!pT~?O;NOm3=$<%x
zWk*exwpX9v4w@>In>V{mZ=8Re9Wxm`H-F~eoIroA<!Hi7gqLfWMl*)3^U3!>RBvhb
zymayG$F(nO210$?mm=815#>E;5>f&6(>QJCO!J4c`r<!i9(YK!n!Tq4swqO*Il3;A
zb3W~Ky=F7VyPsYx&)mx%ZS#$aZI)~mJAtnq!Nx}}f1YbCN#~_m|4M=zWLNY&G+WTF
zr|uDN3t;x|F>#DQ?TEMaRP^eunJ}J6P@Hc`_69xmBH$900|ta}Wfv)K2qkN5@)K?u
z1;Hwtud;0>J233gyNszteoh|(8orldI~OwwWdrc7ynC%zB~6K}3whSAd#pBl>v5v%
zmG0)F6h*b4U+u4t^}qzl6%$wYX81UjiB;M3cHeL8mC6K9x=@8kB|A6)HD(0`1X~m1
z0!OJS;cbDYE^WAZdq5H7#md2gyYS-G5$jd*k?RJHlUr|u=YR%ksV>vsR|qwoTfypk
z`b^vRnu}=;vvj8*ul_vw1qjx)yoMki0JaSq89d<Wy1p|^P6L33{dYvA{~uQJe_%2J
zKsSHD&HjO>{1;aW=<GjnDqjGQn@*Jgbdc*csorO|3%L0e`nw*-{Jr++E0%kMyPT1S
zH4J86(@ECoWD_g;owT6yempz_De?k9)=GNwD{{=*@8_ao`Wndy@Z8$@-TnuY%Ra#;
z16lfb$#x1N-V3yQ-vT`@qt+Ki{V%LfjdF-#rq{W6;P#8Nms@%D(syFejclb7%tpQ3
zqXi}Xh=)q)e%v=|-X_ZzS#3!L?X@!o-sp1{PiEhY9x0@1!{xCy{5ob~gvDjbtQg>3
zugAR^{AiRl=H~sKXu}{l5UM;3KB=TiS!v|;E@H)wOGQi<U&Y-;{5g7k@jX@z>ql3L
z7UwIhYd#yBz-dcdZs$c0TR(&IjD?tq$TCpLygPSx?0&_1v6@%!%Vh5T73zU2T`O)^
z9_uM6wm;DIeM<j2@M1F@xEw}h9~lC?$bX<>d>_^5-Z{!Qyz!pNS!nBN4JiNdw~+%R
zfr|>I_vHp|QWU-Kdx<V{e^R1#T$=AYhN1YCR`1nu3(yRUIQlq$T|jiOezE9(?tSC)
z>VGkyUDKU8&CuN$=cC-F5%#&4LHw>4n`(Udc;LlwGhD+BVhAPTZn^B6qo19=E}r?Q
zXOxee1!aldCS65GFkyv88AS~A_Kcbdz}X3z_x)4&VOOuV;I4Jx`dNJE9ipUmOUPCl
z+^^^8{p9&reVY&bhttsA8kYpA(_;Bb%5<gn{dc9uRlrQP^Xg=c1@R}}M^*D!XN};x
zH#*m*UhxF`i=Q|ZA2|MWw;>yJ?PGgoU@2E@rLTq$UCcY^$R`qoqD#i{KlzgnCm387
zC}TUUkk(GIVo;RK_O6lMh3!c8ysN{=C<j5;H*76uwHcuDFO%Vh>5aU5Izkv_yujY*
z$oS@Pw1;BbkC26SngOVX4cTkgQxEOA*RFl9-~DZKQu^V~p(*?l?O&IVOK0Cq-yF8E
z(e2-8`(Q%c5uf?g3B{!^xh&rJhqtJ$^c9kkK<~DH7x3Hy`-g02(D3!YL3z;F;U`{|
zxl%snWE{G?1@aZiNE*44Ruwxe=HuDe;Oo3QBGXWM2qAk-34J4#typJSXdqYeFglbo
z>ol>SvFV>1*%z)p60MCL!7dIXoW#vP)TD+&m6PsX?%T!>dnvEMYClpbd#)>BHx|x)
zUn+^0iEM|un0nD;Z-2X`&T;cUe~-UqVd?m<0J#Bl;8UfU>4D$mEV690SaA&pA?DM(
zKB}9Zx0x{(D8p$g{C0i&dTW;!L+fdW#@%-PuJ@v!oquZuqZv%1({Il~qmYf252sUx
z<XY+qzfw~WSk-5DTtU9){*beTG#QIS6%fl<P?x*%aGTT&RSe(f+((N~4i;+=&Z$}L
zZ|0e}G5P%S>T94`$fL8HoULy6o8E}uI#4!8>}@l*xT`b;J;yu)J}(8CPeWQo<WMlR
zmZ)J!{bZ$(+l14Qw&-nEE5vGS{S1|jX&}*ftaGc=(FzpQfDAJ72F1Jt*b4;EJ3Gps
zFmm$bDE#hT&tb!BNq>8{6xb#WS!wq^@6cRV?oi|Y!}{)ds~BhmyDL=k?&1C`S+i24
zca7ivH5gYGh_B*FNn)qs=FNCxke88IcFz1m&@wRaTw@SYU9CmgXF5MtkOctZeSAgp
f|4>etOLE_W^XFtu^2GokH;JyMp+=Ru?aTiFBmh3;

diff --git a/docs/dev/articles/examples/timeseries/timeseries_classification_from_scratch.html b/docs/dev/articles/examples/timeseries/timeseries_classification_from_scratch.html
index 20be8ee04..1cb6af443 100644
--- a/docs/dev/articles/examples/timeseries/timeseries_classification_from_scratch.html
+++ b/docs/dev/articles/examples/timeseries/timeseries_classification_from_scratch.html
@@ -31,7 +31,7 @@
 
     <a class="navbar-brand me-2" href="../../../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -148,44 +148,10 @@ <h3 id="read-the-tsv-data">Read the TSV data<a class="anchor" aria-label="anchor
 <span><span class="op">}</span></span>
 <span></span>
 <span><span class="va">root_url</span> <span class="op">&lt;-</span> <span class="st">"https://raw.githubusercontent.com/hfawaz/cd-diagram/master/FordA/"</span></span>
-<span><span class="fu"><a href="https://rdrr.io/r/base/c.html" class="external-link">c</a></span><span class="op">(</span><span class="va">x_train</span>, <span class="va">y_train</span><span class="op">)</span> <span class="op"><a href="../../../reference/multi-assign.html">%&lt;-%</a></span> <span class="fu">get_data</span><span class="op">(</span><span class="fu"><a href="https://rdrr.io/r/base/paste.html" class="external-link">paste0</a></span><span class="op">(</span><span class="va">root_url</span>, <span class="st">"FordA_TRAIN.tsv"</span><span class="op">)</span><span class="op">)</span></span></code></pre></div>
-<pre><code>## Downloading data from https://raw.githubusercontent.com/hfawaz/cd-diagram/master/FordA/FordA_TRAIN.tsv
-##
-
[1m       0/20094049
[0m 
[37m━━━━━━━━━━━━━━━━━━━━
[0m 
[1m0s
[0m 0s/step
-
[1m  114688/20094049
[0m 
[37m━━━━━━━━━━━━━━━━━━━━
[0m 
[1m11s
[0m 1us/step
-
[1m  458752/20094049
[0m 
[37m━━━━━━━━━━━━━━━━━━━━
[0m 
[1m4s
[0m 0us/step 
-
[1m  925696/20094049
[0m 
[37m━━━━━━━━━━━━━━━━━━━━
[0m 
[1m3s
[0m 0us/step
-
[1m 1458176/20094049
[0m 
[32m━
[0m
[37m━━━━━━━━━━━━━━━━━━━
[0m 
[1m2s
[0m 0us/step
-
[1m 1867776/20094049
[0m 
[32m━
[0m
[37m━━━━━━━━━━━━━━━━━━━
[0m 
[1m2s
[0m 0us/step
-
[1m 2375680/20094049
[0m 
[32m━━
[0m
[37m━━━━━━━━━━━━━━━━━━
[0m 
[1m2s
[0m 0us/step
-
[1m 2834432/20094049
[0m 
[32m━━
[0m
[37m━━━━━━━━━━━━━━━━━━
[0m 
[1m2s
[0m 0us/step
-
[1m 3407872/20094049
[0m 
[32m━━━
[0m
[37m━━━━━━━━━━━━━━━━━
[0m 
[1m2s
[0m 0us/step
-
[1m 4014080/20094049
[0m 
[32m━━━
[0m
[37m━━━━━━━━━━━━━━━━━
[0m 
[1m1s
[0m 0us/step
-
[1m 4751360/20094049
[0m 
[32m━━━━
[0m
[37m━━━━━━━━━━━━━━━━
[0m 
[1m1s
[0m 0us/step
-
[1m 5939200/20094049
[0m 
[32m━━━━━
[0m
[37m━━━━━━━━━━━━━━━
[0m 
[1m1s
[0m 0us/step
-
[1m 7675904/20094049
[0m 
[32m━━━━━━━
[0m
[37m━━━━━━━━━━━━━
[0m 
[1m1s
[0m 0us/step
-
[1m 9625600/20094049
[0m 
[32m━━━━━━━━━
[0m
[37m━━━━━━━━━━━
[0m 
[1m0s
[0m 0us/step
-
[1m11698176/20094049
[0m 
[32m━━━━━━━━━━━
[0m
[37m━━━━━━━━━
[0m 
[1m0s
[0m 0us/step
-
[1m13975552/20094049
[0m 
[32m━━━━━━━━━━━━━
[0m
[37m━━━━━━━
[0m 
[1m0s
[0m 0us/step
-
[1m16367616/20094049
[0m 
[32m━━━━━━━━━━━━━━━━
[0m
[37m━━━━
[0m 
[1m0s
[0m 0us/step
-
[1m18743296/20094049
[0m 
[32m━━━━━━━━━━━━━━━━━━
[0m
[37m━━
[0m 
[1m0s
[0m 0us/step
-
[1m20094049/20094049
[0m 
[32m━━━━━━━━━━━━━━━━━━━━
[0m
[37m
[0m 
[1m1s
[0m 0us/step</code></pre>
-<div class="sourceCode" id="cb4"><pre class="downlit sourceCode r">
-<code class="sourceCode R"><span><span class="fu"><a href="https://rdrr.io/r/base/c.html" class="external-link">c</a></span><span class="op">(</span><span class="va">x_test</span>, <span class="va">y_test</span><span class="op">)</span> <span class="op"><a href="../../../reference/multi-assign.html">%&lt;-%</a></span> <span class="fu">get_data</span><span class="op">(</span><span class="fu"><a href="https://rdrr.io/r/base/paste.html" class="external-link">paste0</a></span><span class="op">(</span><span class="va">root_url</span>, <span class="st">"FordA_TEST.tsv"</span><span class="op">)</span><span class="op">)</span></span></code></pre></div>
-<pre><code>## Downloading data from https://raw.githubusercontent.com/hfawaz/cd-diagram/master/FordA/FordA_TEST.tsv
-##
-
[1m      0/7364408
[0m 
[37m━━━━━━━━━━━━━━━━━━━━
[0m 
[1m0s
[0m 0s/step
-
[1m 122880/7364408
[0m 
[37m━━━━━━━━━━━━━━━━━━━━
[0m 
[1m3s
[0m 0us/step
-
[1m 589824/7364408
[0m 
[32m━
[0m
[37m━━━━━━━━━━━━━━━━━━━
[0m 
[1m1s
[0m 0us/step
-
[1m1277952/7364408
[0m 
[32m━━━
[0m
[37m━━━━━━━━━━━━━━━━━
[0m 
[1m0s
[0m 0us/step
-
[1m1720320/7364408
[0m 
[32m━━━━
[0m
[37m━━━━━━━━━━━━━━━━
[0m 
[1m0s
[0m 0us/step
-
[1m3342336/7364408
[0m 
[32m━━━━━━━━━
[0m
[37m━━━━━━━━━━━
[0m 
[1m0s
[0m 0us/step
-
[1m4112384/7364408
[0m 
[32m━━━━━━━━━━━
[0m
[37m━━━━━━━━━
[0m 
[1m0s
[0m 0us/step
-
[1m6004736/7364408
[0m 
[32m━━━━━━━━━━━━━━━━
[0m
[37m━━━━
[0m 
[1m0s
[0m 0us/step
-
[1m6823936/7364408
[0m 
[32m━━━━━━━━━━━━━━━━━━
[0m
[37m━━
[0m 
[1m0s
[0m 0us/step
-
[1m7364408/7364408
[0m 
[32m━━━━━━━━━━━━━━━━━━━━
[0m
[37m
[0m 
[1m1s
[0m 0us/step</code></pre>
-<div class="sourceCode" id="cb6"><pre class="downlit sourceCode r">
-<code class="sourceCode R"><span><span class="fu"><a href="https://rdrr.io/r/utils/str.html" class="external-link">str</a></span><span class="op">(</span><span class="fu">keras3</span><span class="fu">:::</span><span class="fu">named_list</span><span class="op">(</span></span>
+<span><span class="fu"><a href="https://rdrr.io/r/base/c.html" class="external-link">c</a></span><span class="op">(</span><span class="va">x_train</span>, <span class="va">y_train</span><span class="op">)</span> <span class="op"><a href="../../../reference/multi-assign.html">%&lt;-%</a></span> <span class="fu">get_data</span><span class="op">(</span><span class="fu"><a href="https://rdrr.io/r/base/paste.html" class="external-link">paste0</a></span><span class="op">(</span><span class="va">root_url</span>, <span class="st">"FordA_TRAIN.tsv"</span><span class="op">)</span><span class="op">)</span></span>
+<span><span class="fu"><a href="https://rdrr.io/r/base/c.html" class="external-link">c</a></span><span class="op">(</span><span class="va">x_test</span>, <span class="va">y_test</span><span class="op">)</span> <span class="op"><a href="../../../reference/multi-assign.html">%&lt;-%</a></span> <span class="fu">get_data</span><span class="op">(</span><span class="fu"><a href="https://rdrr.io/r/base/paste.html" class="external-link">paste0</a></span><span class="op">(</span><span class="va">root_url</span>, <span class="st">"FordA_TEST.tsv"</span><span class="op">)</span><span class="op">)</span></span>
+<span></span>
+<span><span class="fu"><a href="https://rdrr.io/r/utils/str.html" class="external-link">str</a></span><span class="op">(</span><span class="fu">keras3</span><span class="fu">:::</span><span class="fu">named_list</span><span class="op">(</span></span>
 <span>  <span class="va">x_train</span>, <span class="va">y_train</span>,</span>
 <span>  <span class="va">x_test</span>, <span class="va">y_test</span></span>
 <span><span class="op">)</span><span class="op">)</span></span></code></pre></div>
@@ -201,7 +167,7 @@ <h2 id="visualize-the-data">Visualize the data<a class="anchor" aria-label="anch
 </h2>
 <p>Here we visualize one timeseries example for each class in the
 dataset.</p>
-<div class="sourceCode" id="cb8"><pre class="downlit sourceCode r">
+<div class="sourceCode" id="cb4"><pre class="downlit sourceCode r">
 <code class="sourceCode R"><span><span class="fu"><a href="https://rdrr.io/r/graphics/plot.default.html" class="external-link">plot</a></span><span class="op">(</span><span class="cn">NULL</span>, main <span class="op">=</span> <span class="st">"Timeseries Data"</span>,</span>
 <span>     xlab <span class="op">=</span> <span class="st">"Timepoints"</span>,  ylab <span class="op">=</span> <span class="st">"Values"</span>,</span>
 <span>     xlim <span class="op">=</span> <span class="fu"><a href="https://rdrr.io/r/base/c.html" class="external-link">c</a></span><span class="op">(</span><span class="fl">1</span>, <span class="fu"><a href="https://rdrr.io/r/base/nrow.html" class="external-link">ncol</a></span><span class="op">(</span><span class="va">x_test</span><span class="op">)</span><span class="op">)</span>,</span>
@@ -230,24 +196,24 @@ <h2 id="standardize-the-data">Standardize the data<a class="anchor" aria-label="
 transform the timeseries into a multivariate one with one channel using
 a simple reshaping via numpy. This will allow us to construct a model
 that is easily applicable to multivariate time series.</p>
-<div class="sourceCode" id="cb9"><pre class="downlit sourceCode r">
+<div class="sourceCode" id="cb5"><pre class="downlit sourceCode r">
 <code class="sourceCode R"><span><span class="fu"><a href="https://rdrr.io/r/base/dim.html" class="external-link">dim</a></span><span class="op">(</span><span class="va">x_train</span><span class="op">)</span> <span class="op">&lt;-</span> <span class="fu"><a href="https://rdrr.io/r/base/c.html" class="external-link">c</a></span><span class="op">(</span><span class="fu"><a href="https://rdrr.io/r/base/dim.html" class="external-link">dim</a></span><span class="op">(</span><span class="va">x_train</span><span class="op">)</span>, <span class="fl">1</span><span class="op">)</span></span>
 <span><span class="fu"><a href="https://rdrr.io/r/base/dim.html" class="external-link">dim</a></span><span class="op">(</span><span class="va">x_test</span><span class="op">)</span> <span class="op">&lt;-</span> <span class="fu"><a href="https://rdrr.io/r/base/c.html" class="external-link">c</a></span><span class="op">(</span><span class="fu"><a href="https://rdrr.io/r/base/dim.html" class="external-link">dim</a></span><span class="op">(</span><span class="va">x_test</span><span class="op">)</span>, <span class="fl">1</span><span class="op">)</span></span></code></pre></div>
 <p>Finally, in order to use
 <code>sparse_categorical_crossentropy</code>, we will have to count the
 number of classes beforehand.</p>
-<div class="sourceCode" id="cb10"><pre class="downlit sourceCode r">
+<div class="sourceCode" id="cb6"><pre class="downlit sourceCode r">
 <code class="sourceCode R"><span><span class="va">num_classes</span> <span class="op">&lt;-</span> <span class="fu"><a href="https://rdrr.io/r/base/length.html" class="external-link">length</a></span><span class="op">(</span><span class="fu"><a href="https://rdrr.io/r/base/unique.html" class="external-link">unique</a></span><span class="op">(</span><span class="va">y_train</span><span class="op">)</span><span class="op">)</span></span></code></pre></div>
 <p>Now we shuffle the training set because we will be using the
 <code>validation_split</code> option later when training.</p>
-<div class="sourceCode" id="cb11"><pre class="downlit sourceCode r">
+<div class="sourceCode" id="cb7"><pre class="downlit sourceCode r">
 <code class="sourceCode R"><span><span class="fu"><a href="https://rdrr.io/r/base/c.html" class="external-link">c</a></span><span class="op">(</span><span class="va">x_train</span>, <span class="va">y_train</span><span class="op">)</span> <span class="op"><a href="../../../reference/multi-assign.html">%&lt;-%</a></span> <span class="fu">listarrays</span><span class="fu">::</span><span class="fu"><a href="https://t-kalinowski.github.io/listarrays/reference/shuffle_rows.html" class="external-link">shuffle_rows</a></span><span class="op">(</span><span class="va">x_train</span>, <span class="va">y_train</span><span class="op">)</span></span>
 <span><span class="co"># idx &lt;- sample.int(nrow(x_train))</span></span>
 <span><span class="co"># x_train %&lt;&gt;% .[idx,, ,drop = FALSE]</span></span>
 <span><span class="co"># y_train %&lt;&gt;% .[idx,  ,drop = FALSE]</span></span></code></pre></div>
 <p>Standardize the labels to positive integers. The expected labels will
 then be 0 and 1.</p>
-<div class="sourceCode" id="cb12"><pre class="downlit sourceCode r">
+<div class="sourceCode" id="cb8"><pre class="downlit sourceCode r">
 <code class="sourceCode R"><span><span class="va">y_train</span><span class="op">[</span><span class="va">y_train</span> <span class="op">==</span> <span class="op">-</span><span class="fl">1L</span><span class="op">]</span> <span class="op">&lt;-</span> <span class="fl">0L</span></span>
 <span><span class="va">y_test</span><span class="op">[</span><span class="va">y_test</span> <span class="op">==</span> <span class="op">-</span><span class="fl">1L</span><span class="op">]</span> <span class="op">&lt;-</span> <span class="fl">0L</span></span></code></pre></div>
 </div>
@@ -259,7 +225,7 @@ <h2 id="build-a-model">Build a model<a class="anchor" aria-label="anchor" href="
 implementation is based on the TF 2 version provided <a href="https://github.com/hfawaz/dl-4-tsc/" class="external-link">here</a>. The following
 hyperparameters (kernel_size, filters, the usage of BatchNorm) were
 found via random search using <a href="https://github.com/keras-team/keras-tuner" class="external-link">KerasTuner</a>.</p>
-<div class="sourceCode" id="cb13"><pre class="downlit sourceCode r">
+<div class="sourceCode" id="cb9"><pre class="downlit sourceCode r">
 <code class="sourceCode R"><span><span class="va">make_model</span> <span class="op">&lt;-</span> <span class="kw">function</span><span class="op">(</span><span class="va">input_shape</span><span class="op">)</span> <span class="op">{</span></span>
 <span>  <span class="va">inputs</span> <span class="op">&lt;-</span> <span class="fu"><a href="../../../reference/keras_input.html">keras_input</a></span><span class="op">(</span><span class="va">input_shape</span><span class="op">)</span></span>
 <span></span>
@@ -285,7 +251,7 @@ <h2 id="build-a-model">Build a model<a class="anchor" aria-label="anchor" href="
 <span><span class="op">}</span></span>
 <span></span>
 <span><span class="va">model</span> <span class="op">&lt;-</span> <span class="fu">make_model</span><span class="op">(</span>input_shape <span class="op">=</span> <span class="fu"><a href="https://rdrr.io/r/base/dim.html" class="external-link">dim</a></span><span class="op">(</span><span class="va">x_train</span><span class="op">)</span><span class="op">[</span><span class="op">-</span><span class="fl">1</span><span class="op">]</span><span class="op">)</span></span></code></pre></div>
-<div class="sourceCode" id="cb14"><pre class="downlit sourceCode r">
+<div class="sourceCode" id="cb10"><pre class="downlit sourceCode r">
 <code class="sourceCode R"><span><span class="va">model</span></span></code></pre></div>
 <pre><code><span><span class="co">## <span style="font-weight: bold;">Model: "functional"</span></span></span>
 <span><span class="co">## ┏━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━┳━━━━━━━┓</span></span>
@@ -322,7 +288,7 @@ <h2 id="build-a-model">Build a model<a class="anchor" aria-label="anchor" href="
 <span><span class="co">## <span style="font-weight: bold;"> Total params: </span><span style="color: #00AF00;">25,858</span> (101.01 KB)</span></span>
 <span><span class="co">## <span style="font-weight: bold;"> Trainable params: </span><span style="color: #00AF00;">25,474</span> (99.51 KB)</span></span>
 <span><span class="co">## <span style="font-weight: bold;"> Non-trainable params: </span><span style="color: #00AF00;">384</span> (1.50 KB)</span></span></code></pre>
-<div class="sourceCode" id="cb16"><pre class="downlit sourceCode r">
+<div class="sourceCode" id="cb12"><pre class="downlit sourceCode r">
 <code class="sourceCode R"><span><span class="fu"><a href="https://rdrr.io/r/graphics/plot.default.html" class="external-link">plot</a></span><span class="op">(</span><span class="va">model</span>, show_shapes <span class="op">=</span> <span class="cn">TRUE</span><span class="op">)</span></span></code></pre></div>
 <div class="figure">
 <img src="timeseries_classification_from_scratch/unnamed-chunk-9-1.png" alt="plot of chunk unnamed-chunk-9" width="801"><p class="caption">
@@ -333,7 +299,7 @@ <h2 id="build-a-model">Build a model<a class="anchor" aria-label="anchor" href="
 <div class="section level2">
 <h2 id="train-the-model">Train the model<a class="anchor" aria-label="anchor" href="#train-the-model"></a>
 </h2>
-<div class="sourceCode" id="cb17"><pre class="downlit sourceCode r">
+<div class="sourceCode" id="cb13"><pre class="downlit sourceCode r">
 <code class="sourceCode R"><span><span class="va">epochs</span> <span class="op">&lt;-</span> <span class="fl">500</span></span>
 <span><span class="va">batch_size</span> <span class="op">&lt;-</span> <span class="fl">32</span></span>
 <span></span>
@@ -367,417 +333,601 @@ <h2 id="train-the-model">Train the model<a class="anchor" aria-label="anchor" hr
 <span>  validation_split <span class="op">=</span> <span class="fl">0.2</span></span>
 <span><span class="op">)</span></span></code></pre></div>
 <pre><code><span><span class="co">## Epoch 1/500</span></span>
-<span><span class="co">## 90/90 - 4s - 41ms/step - loss: 0.5310 - sparse_categorical_accuracy: 0.7198 - val_loss: 0.7948 - val_sparse_categorical_accuracy: 0.4896 - learning_rate: 1.0000e-03</span></span>
+<span><span class="co">## 90/90 - 3s - 32ms/step - loss: 0.5310 - sparse_categorical_accuracy: 0.7191 - val_loss: 0.7907 - val_sparse_categorical_accuracy: 0.4896 - learning_rate: 1.0000e-03</span></span>
 <span><span class="co">## Epoch 2/500</span></span>
-<span><span class="co">## 90/90 - 1s - 13ms/step - loss: 0.4771 - sparse_categorical_accuracy: 0.7597 - val_loss: 0.9159 - val_sparse_categorical_accuracy: 0.4896 - learning_rate: 1.0000e-03</span></span>
+<span><span class="co">## 90/90 - 1s - 11ms/step - loss: 0.4799 - sparse_categorical_accuracy: 0.7590 - val_loss: 0.9127 - val_sparse_categorical_accuracy: 0.4896 - learning_rate: 1.0000e-03</span></span>
 <span><span class="co">## Epoch 3/500</span></span>
-<span><span class="co">## 90/90 - 0s - 3ms/step - loss: 0.4717 - sparse_categorical_accuracy: 0.7573 - val_loss: 0.9025 - val_sparse_categorical_accuracy: 0.4896 - learning_rate: 1.0000e-03</span></span>
+<span><span class="co">## 90/90 - 0s - 2ms/step - loss: 0.4590 - sparse_categorical_accuracy: 0.7757 - val_loss: 0.8196 - val_sparse_categorical_accuracy: 0.4896 - learning_rate: 1.0000e-03</span></span>
 <span><span class="co">## Epoch 4/500</span></span>
-<span><span class="co">## 90/90 - 0s - 2ms/step - loss: 0.4113 - sparse_categorical_accuracy: 0.8000 - val_loss: 0.8173 - val_sparse_categorical_accuracy: 0.4910 - learning_rate: 1.0000e-03</span></span>
+<span><span class="co">## 90/90 - 0s - 2ms/step - loss: 0.4088 - sparse_categorical_accuracy: 0.7983 - val_loss: 0.6425 - val_sparse_categorical_accuracy: 0.5534 - learning_rate: 1.0000e-03</span></span>
 <span><span class="co">## Epoch 5/500</span></span>
-<span><span class="co">## 90/90 - 0s - 3ms/step - loss: 0.4164 - sparse_categorical_accuracy: 0.7920 - val_loss: 0.6884 - val_sparse_categorical_accuracy: 0.5146 - learning_rate: 1.0000e-03</span></span>
+<span><span class="co">## 90/90 - 0s - 2ms/step - loss: 0.4141 - sparse_categorical_accuracy: 0.7889 - val_loss: 0.5519 - val_sparse_categorical_accuracy: 0.6588 - learning_rate: 1.0000e-03</span></span>
 <span><span class="co">## Epoch 6/500</span></span>
-<span><span class="co">## 90/90 - 0s - 4ms/step - loss: 0.3984 - sparse_categorical_accuracy: 0.8101 - val_loss: 0.4227 - val_sparse_categorical_accuracy: 0.8294 - learning_rate: 1.0000e-03</span></span>
+<span><span class="co">## 90/90 - 0s - 2ms/step - loss: 0.3929 - sparse_categorical_accuracy: 0.8163 - val_loss: 0.4225 - val_sparse_categorical_accuracy: 0.8294 - learning_rate: 1.0000e-03</span></span>
 <span><span class="co">## Epoch 7/500</span></span>
-<span><span class="co">## 90/90 - 0s - 2ms/step - loss: 0.3878 - sparse_categorical_accuracy: 0.8132 - val_loss: 0.4749 - val_sparse_categorical_accuracy: 0.7018 - learning_rate: 1.0000e-03</span></span>
+<span><span class="co">## 90/90 - 0s - 1ms/step - loss: 0.3870 - sparse_categorical_accuracy: 0.8146 - val_loss: 0.6112 - val_sparse_categorical_accuracy: 0.6949 - learning_rate: 1.0000e-03</span></span>
 <span><span class="co">## Epoch 8/500</span></span>
-<span><span class="co">## 90/90 - 0s - 3ms/step - loss: 0.3739 - sparse_categorical_accuracy: 0.8163 - val_loss: 0.3880 - val_sparse_categorical_accuracy: 0.8252 - learning_rate: 1.0000e-03</span></span>
+<span><span class="co">## 90/90 - 0s - 1ms/step - loss: 0.3717 - sparse_categorical_accuracy: 0.8247 - val_loss: 0.4257 - val_sparse_categorical_accuracy: 0.7559 - learning_rate: 1.0000e-03</span></span>
 <span><span class="co">## Epoch 9/500</span></span>
-<span><span class="co">## 90/90 - 0s - 2ms/step - loss: 0.3684 - sparse_categorical_accuracy: 0.8292 - val_loss: 0.4363 - val_sparse_categorical_accuracy: 0.7767 - learning_rate: 1.0000e-03</span></span>
+<span><span class="co">## 90/90 - 0s - 2ms/step - loss: 0.3698 - sparse_categorical_accuracy: 0.8306 - val_loss: 0.3653 - val_sparse_categorical_accuracy: 0.8350 - learning_rate: 1.0000e-03</span></span>
 <span><span class="co">## Epoch 10/500</span></span>
-<span><span class="co">## 90/90 - 0s - 2ms/step - loss: 0.3590 - sparse_categorical_accuracy: 0.8306 - val_loss: 0.5077 - val_sparse_categorical_accuracy: 0.7282 - learning_rate: 1.0000e-03</span></span>
+<span><span class="co">## 90/90 - 0s - 1ms/step - loss: 0.3634 - sparse_categorical_accuracy: 0.8260 - val_loss: 0.4390 - val_sparse_categorical_accuracy: 0.7836 - learning_rate: 1.0000e-03</span></span>
 <span><span class="co">## Epoch 11/500</span></span>
-<span><span class="co">## 90/90 - 0s - 3ms/step - loss: 0.3638 - sparse_categorical_accuracy: 0.8247 - val_loss: 0.3574 - val_sparse_categorical_accuracy: 0.8405 - learning_rate: 1.0000e-03</span></span>
+<span><span class="co">## 90/90 - 0s - 1ms/step - loss: 0.3676 - sparse_categorical_accuracy: 0.8281 - val_loss: 0.4205 - val_sparse_categorical_accuracy: 0.7822 - learning_rate: 1.0000e-03</span></span>
 <span><span class="co">## Epoch 12/500</span></span>
-<span><span class="co">## 90/90 - 0s - 2ms/step - loss: 0.3459 - sparse_categorical_accuracy: 0.8396 - val_loss: 0.8054 - val_sparse_categorical_accuracy: 0.6352 - learning_rate: 1.0000e-03</span></span>
+<span><span class="co">## 90/90 - 0s - 1ms/step - loss: 0.3467 - sparse_categorical_accuracy: 0.8406 - val_loss: 0.3697 - val_sparse_categorical_accuracy: 0.8460 - learning_rate: 1.0000e-03</span></span>
 <span><span class="co">## Epoch 13/500</span></span>
-<span><span class="co">## 90/90 - 0s - 2ms/step - loss: 0.3350 - sparse_categorical_accuracy: 0.8552 - val_loss: 0.3838 - val_sparse_categorical_accuracy: 0.8155 - learning_rate: 1.0000e-03</span></span>
+<span><span class="co">## 90/90 - 0s - 2ms/step - loss: 0.3310 - sparse_categorical_accuracy: 0.8594 - val_loss: 0.3341 - val_sparse_categorical_accuracy: 0.8530 - learning_rate: 1.0000e-03</span></span>
 <span><span class="co">## Epoch 14/500</span></span>
-<span><span class="co">## 90/90 - 0s - 2ms/step - loss: 0.3303 - sparse_categorical_accuracy: 0.8545 - val_loss: 0.4148 - val_sparse_categorical_accuracy: 0.7920 - learning_rate: 1.0000e-03</span></span>
+<span><span class="co">## 90/90 - 0s - 1ms/step - loss: 0.3350 - sparse_categorical_accuracy: 0.8514 - val_loss: 0.5723 - val_sparse_categorical_accuracy: 0.7198 - learning_rate: 1.0000e-03</span></span>
 <span><span class="co">## Epoch 15/500</span></span>
-<span><span class="co">## 90/90 - 0s - 2ms/step - loss: 0.3228 - sparse_categorical_accuracy: 0.8507 - val_loss: 1.6007 - val_sparse_categorical_accuracy: 0.5132 - learning_rate: 1.0000e-03</span></span>
+<span><span class="co">## 90/90 - 0s - 1ms/step - loss: 0.3295 - sparse_categorical_accuracy: 0.8497 - val_loss: 1.0273 - val_sparse_categorical_accuracy: 0.5465 - learning_rate: 1.0000e-03</span></span>
 <span><span class="co">## Epoch 16/500</span></span>
-<span><span class="co">## 90/90 - 0s - 2ms/step - loss: 0.3232 - sparse_categorical_accuracy: 0.8587 - val_loss: 0.5349 - val_sparse_categorical_accuracy: 0.7184 - learning_rate: 1.0000e-03</span></span>
+<span><span class="co">## 90/90 - 0s - 2ms/step - loss: 0.3244 - sparse_categorical_accuracy: 0.8587 - val_loss: 0.3219 - val_sparse_categorical_accuracy: 0.8544 - learning_rate: 1.0000e-03</span></span>
 <span><span class="co">## Epoch 17/500</span></span>
-<span><span class="co">## 90/90 - 0s - 2ms/step - loss: 0.3119 - sparse_categorical_accuracy: 0.8656 - val_loss: 0.3336 - val_sparse_categorical_accuracy: 0.8669 - learning_rate: 1.0000e-03</span></span>
+<span><span class="co">## 90/90 - 0s - 1ms/step - loss: 0.3082 - sparse_categorical_accuracy: 0.8646 - val_loss: 0.4187 - val_sparse_categorical_accuracy: 0.7906 - learning_rate: 1.0000e-03</span></span>
 <span><span class="co">## Epoch 18/500</span></span>
-<span><span class="co">## 90/90 - 0s - 2ms/step - loss: 0.3057 - sparse_categorical_accuracy: 0.8646 - val_loss: 0.5241 - val_sparse_categorical_accuracy: 0.7351 - learning_rate: 1.0000e-03</span></span>
+<span><span class="co">## 90/90 - 0s - 1ms/step - loss: 0.3022 - sparse_categorical_accuracy: 0.8715 - val_loss: 0.6302 - val_sparse_categorical_accuracy: 0.6824 - learning_rate: 1.0000e-03</span></span>
 <span><span class="co">## Epoch 19/500</span></span>
-<span><span class="co">## 90/90 - 0s - 2ms/step - loss: 0.3042 - sparse_categorical_accuracy: 0.8691 - val_loss: 0.4023 - val_sparse_categorical_accuracy: 0.8128 - learning_rate: 1.0000e-03</span></span>
+<span><span class="co">## 90/90 - 0s - 1ms/step - loss: 0.2996 - sparse_categorical_accuracy: 0.8764 - val_loss: 0.3537 - val_sparse_categorical_accuracy: 0.8447 - learning_rate: 1.0000e-03</span></span>
 <span><span class="co">## Epoch 20/500</span></span>
-<span><span class="co">## 90/90 - 0s - 2ms/step - loss: 0.3037 - sparse_categorical_accuracy: 0.8677 - val_loss: 0.4818 - val_sparse_categorical_accuracy: 0.7670 - learning_rate: 1.0000e-03</span></span>
+<span><span class="co">## 90/90 - 0s - 1ms/step - loss: 0.3053 - sparse_categorical_accuracy: 0.8715 - val_loss: 0.4925 - val_sparse_categorical_accuracy: 0.7309 - learning_rate: 1.0000e-03</span></span>
 <span><span class="co">## Epoch 21/500</span></span>
-<span><span class="co">## 90/90 - 0s - 2ms/step - loss: 0.2834 - sparse_categorical_accuracy: 0.8795 - val_loss: 0.3017 - val_sparse_categorical_accuracy: 0.8696 - learning_rate: 1.0000e-03</span></span>
+<span><span class="co">## 90/90 - 0s - 1ms/step - loss: 0.2827 - sparse_categorical_accuracy: 0.8826 - val_loss: 0.3893 - val_sparse_categorical_accuracy: 0.7809 - learning_rate: 1.0000e-03</span></span>
 <span><span class="co">## Epoch 22/500</span></span>
-<span><span class="co">## 90/90 - 0s - 1ms/step - loss: 0.2879 - sparse_categorical_accuracy: 0.8819 - val_loss: 0.4130 - val_sparse_categorical_accuracy: 0.8003 - learning_rate: 1.0000e-03</span></span>
+<span><span class="co">## 90/90 - 0s - 2ms/step - loss: 0.2892 - sparse_categorical_accuracy: 0.8795 - val_loss: 0.3128 - val_sparse_categorical_accuracy: 0.8863 - learning_rate: 1.0000e-03</span></span>
 <span><span class="co">## Epoch 23/500</span></span>
-<span><span class="co">## 90/90 - 0s - 1ms/step - loss: 0.2778 - sparse_categorical_accuracy: 0.8844 - val_loss: 0.3149 - val_sparse_categorical_accuracy: 0.8599 - learning_rate: 1.0000e-03</span></span>
+<span><span class="co">## 90/90 - 0s - 2ms/step - loss: 0.2718 - sparse_categorical_accuracy: 0.8889 - val_loss: 0.3083 - val_sparse_categorical_accuracy: 0.8571 - learning_rate: 1.0000e-03</span></span>
 <span><span class="co">## Epoch 24/500</span></span>
-<span><span class="co">## 90/90 - 0s - 2ms/step - loss: 0.2743 - sparse_categorical_accuracy: 0.8861 - val_loss: 0.3331 - val_sparse_categorical_accuracy: 0.8460 - learning_rate: 1.0000e-03</span></span>
+<span><span class="co">## 90/90 - 0s - 1ms/step - loss: 0.2691 - sparse_categorical_accuracy: 0.8934 - val_loss: 0.5609 - val_sparse_categorical_accuracy: 0.7420 - learning_rate: 1.0000e-03</span></span>
 <span><span class="co">## Epoch 25/500</span></span>
-<span><span class="co">## 90/90 - 0s - 1ms/step - loss: 0.2654 - sparse_categorical_accuracy: 0.8892 - val_loss: 0.4245 - val_sparse_categorical_accuracy: 0.7892 - learning_rate: 1.0000e-03</span></span>
+<span><span class="co">## 90/90 - 0s - 1ms/step - loss: 0.2635 - sparse_categorical_accuracy: 0.8948 - val_loss: 0.5264 - val_sparse_categorical_accuracy: 0.7573 - learning_rate: 1.0000e-03</span></span>
 <span><span class="co">## Epoch 26/500</span></span>
-<span><span class="co">## 90/90 - 0s - 1ms/step - loss: 0.3058 - sparse_categorical_accuracy: 0.8660 - val_loss: 0.3187 - val_sparse_categorical_accuracy: 0.8669 - learning_rate: 1.0000e-03</span></span>
+<span><span class="co">## 90/90 - 0s - 1ms/step - loss: 0.3109 - sparse_categorical_accuracy: 0.8667 - val_loss: 0.3934 - val_sparse_categorical_accuracy: 0.7878 - learning_rate: 1.0000e-03</span></span>
 <span><span class="co">## Epoch 27/500</span></span>
-<span><span class="co">## 90/90 - 0s - 1ms/step - loss: 0.2619 - sparse_categorical_accuracy: 0.8965 - val_loss: 0.4151 - val_sparse_categorical_accuracy: 0.7850 - learning_rate: 1.0000e-03</span></span>
+<span><span class="co">## 90/90 - 0s - 2ms/step - loss: 0.2599 - sparse_categorical_accuracy: 0.8969 - val_loss: 0.2974 - val_sparse_categorical_accuracy: 0.8558 - learning_rate: 1.0000e-03</span></span>
 <span><span class="co">## Epoch 28/500</span></span>
-<span><span class="co">## 90/90 - 0s - 1ms/step - loss: 0.2785 - sparse_categorical_accuracy: 0.8792 - val_loss: 0.3807 - val_sparse_categorical_accuracy: 0.8308 - learning_rate: 1.0000e-03</span></span>
+<span><span class="co">## 90/90 - 0s - 1ms/step - loss: 0.2805 - sparse_categorical_accuracy: 0.8823 - val_loss: 0.4026 - val_sparse_categorical_accuracy: 0.8183 - learning_rate: 1.0000e-03</span></span>
 <span><span class="co">## Epoch 29/500</span></span>
-<span><span class="co">## 90/90 - 0s - 2ms/step - loss: 0.2594 - sparse_categorical_accuracy: 0.8878 - val_loss: 0.4138 - val_sparse_categorical_accuracy: 0.8128 - learning_rate: 1.0000e-03</span></span>
+<span><span class="co">## 90/90 - 0s - 2ms/step - loss: 0.2517 - sparse_categorical_accuracy: 0.8948 - val_loss: 0.2699 - val_sparse_categorical_accuracy: 0.8849 - learning_rate: 1.0000e-03</span></span>
 <span><span class="co">## Epoch 30/500</span></span>
-<span><span class="co">## 90/90 - 0s - 2ms/step - loss: 0.2644 - sparse_categorical_accuracy: 0.8878 - val_loss: 0.5111 - val_sparse_categorical_accuracy: 0.7323 - learning_rate: 1.0000e-03</span></span>
+<span><span class="co">## 90/90 - 0s - 1ms/step - loss: 0.2581 - sparse_categorical_accuracy: 0.8920 - val_loss: 0.2743 - val_sparse_categorical_accuracy: 0.8752 - learning_rate: 1.0000e-03</span></span>
 <span><span class="co">## Epoch 31/500</span></span>
-<span><span class="co">## 90/90 - 0s - 2ms/step - loss: 0.2653 - sparse_categorical_accuracy: 0.8924 - val_loss: 0.2788 - val_sparse_categorical_accuracy: 0.8863 - learning_rate: 1.0000e-03</span></span>
+<span><span class="co">## 90/90 - 0s - 1ms/step - loss: 0.2550 - sparse_categorical_accuracy: 0.8986 - val_loss: 0.2936 - val_sparse_categorical_accuracy: 0.8655 - learning_rate: 1.0000e-03</span></span>
 <span><span class="co">## Epoch 32/500</span></span>
-<span><span class="co">## 90/90 - 0s - 2ms/step - loss: 0.2546 - sparse_categorical_accuracy: 0.8903 - val_loss: 0.2598 - val_sparse_categorical_accuracy: 0.8821 - learning_rate: 1.0000e-03</span></span>
+<span><span class="co">## 90/90 - 0s - 2ms/step - loss: 0.2528 - sparse_categorical_accuracy: 0.8934 - val_loss: 0.2528 - val_sparse_categorical_accuracy: 0.8904 - learning_rate: 1.0000e-03</span></span>
 <span><span class="co">## Epoch 33/500</span></span>
-<span><span class="co">## 90/90 - 0s - 2ms/step - loss: 0.2408 - sparse_categorical_accuracy: 0.9062 - val_loss: 0.2632 - val_sparse_categorical_accuracy: 0.8793 - learning_rate: 1.0000e-03</span></span>
+<span><span class="co">## 90/90 - 0s - 1ms/step - loss: 0.2390 - sparse_categorical_accuracy: 0.9062 - val_loss: 0.2961 - val_sparse_categorical_accuracy: 0.8571 - learning_rate: 1.0000e-03</span></span>
 <span><span class="co">## Epoch 34/500</span></span>
-<span><span class="co">## 90/90 - 0s - 2ms/step - loss: 0.2373 - sparse_categorical_accuracy: 0.9052 - val_loss: 0.2966 - val_sparse_categorical_accuracy: 0.8710 - learning_rate: 1.0000e-03</span></span>
+<span><span class="co">## 90/90 - 0s - 1ms/step - loss: 0.2327 - sparse_categorical_accuracy: 0.9049 - val_loss: 0.3039 - val_sparse_categorical_accuracy: 0.8558 - learning_rate: 1.0000e-03</span></span>
 <span><span class="co">## Epoch 35/500</span></span>
-<span><span class="co">## 90/90 - 0s - 1ms/step - loss: 0.2560 - sparse_categorical_accuracy: 0.8903 - val_loss: 0.6435 - val_sparse_categorical_accuracy: 0.7309 - learning_rate: 1.0000e-03</span></span>
+<span><span class="co">## 90/90 - 0s - 1ms/step - loss: 0.2447 - sparse_categorical_accuracy: 0.9021 - val_loss: 0.3123 - val_sparse_categorical_accuracy: 0.8710 - learning_rate: 1.0000e-03</span></span>
 <span><span class="co">## Epoch 36/500</span></span>
-<span><span class="co">## 90/90 - 0s - 1ms/step - loss: 0.2391 - sparse_categorical_accuracy: 0.9035 - val_loss: 0.2705 - val_sparse_categorical_accuracy: 0.8807 - learning_rate: 1.0000e-03</span></span>
+<span><span class="co">## 90/90 - 0s - 1ms/step - loss: 0.2340 - sparse_categorical_accuracy: 0.9069 - val_loss: 0.2912 - val_sparse_categorical_accuracy: 0.8821 - learning_rate: 1.0000e-03</span></span>
 <span><span class="co">## Epoch 37/500</span></span>
-<span><span class="co">## 90/90 - 0s - 2ms/step - loss: 0.2384 - sparse_categorical_accuracy: 0.8993 - val_loss: 0.2720 - val_sparse_categorical_accuracy: 0.8738 - learning_rate: 1.0000e-03</span></span>
+<span><span class="co">## 90/90 - 0s - 1ms/step - loss: 0.2352 - sparse_categorical_accuracy: 0.8997 - val_loss: 0.3983 - val_sparse_categorical_accuracy: 0.8017 - learning_rate: 1.0000e-03</span></span>
 <span><span class="co">## Epoch 38/500</span></span>
-<span><span class="co">## 90/90 - 0s - 2ms/step - loss: 0.2339 - sparse_categorical_accuracy: 0.9080 - val_loss: 0.9019 - val_sparse_categorical_accuracy: 0.7309 - learning_rate: 1.0000e-03</span></span>
+<span><span class="co">## 90/90 - 0s - 1ms/step - loss: 0.2294 - sparse_categorical_accuracy: 0.9139 - val_loss: 0.3342 - val_sparse_categorical_accuracy: 0.8447 - learning_rate: 1.0000e-03</span></span>
 <span><span class="co">## Epoch 39/500</span></span>
-<span><span class="co">## 90/90 - 0s - 2ms/step - loss: 0.2404 - sparse_categorical_accuracy: 0.9038 - val_loss: 0.4007 - val_sparse_categorical_accuracy: 0.8294 - learning_rate: 1.0000e-03</span></span>
+<span><span class="co">## 90/90 - 0s - 1ms/step - loss: 0.2268 - sparse_categorical_accuracy: 0.9135 - val_loss: 0.4730 - val_sparse_categorical_accuracy: 0.7975 - learning_rate: 1.0000e-03</span></span>
 <span><span class="co">## Epoch 40/500</span></span>
-<span><span class="co">## 90/90 - 0s - 2ms/step - loss: 0.2265 - sparse_categorical_accuracy: 0.9104 - val_loss: 0.6253 - val_sparse_categorical_accuracy: 0.7559 - learning_rate: 1.0000e-03</span></span>
+<span><span class="co">## 90/90 - 0s - 1ms/step - loss: 0.2202 - sparse_categorical_accuracy: 0.9111 - val_loss: 0.3007 - val_sparse_categorical_accuracy: 0.8738 - learning_rate: 1.0000e-03</span></span>
 <span><span class="co">## Epoch 41/500</span></span>
-<span><span class="co">## 90/90 - 0s - 2ms/step - loss: 0.2159 - sparse_categorical_accuracy: 0.9174 - val_loss: 0.2307 - val_sparse_categorical_accuracy: 0.8974 - learning_rate: 1.0000e-03</span></span>
+<span><span class="co">## 90/90 - 0s - 1ms/step - loss: 0.2064 - sparse_categorical_accuracy: 0.9233 - val_loss: 0.2924 - val_sparse_categorical_accuracy: 0.8849 - learning_rate: 1.0000e-03</span></span>
 <span><span class="co">## Epoch 42/500</span></span>
-<span><span class="co">## 90/90 - 0s - 2ms/step - loss: 0.2081 - sparse_categorical_accuracy: 0.9271 - val_loss: 0.5934 - val_sparse_categorical_accuracy: 0.7434 - learning_rate: 1.0000e-03</span></span>
+<span><span class="co">## 90/90 - 0s - 1ms/step - loss: 0.1966 - sparse_categorical_accuracy: 0.9316 - val_loss: 0.8303 - val_sparse_categorical_accuracy: 0.6935 - learning_rate: 1.0000e-03</span></span>
 <span><span class="co">## Epoch 43/500</span></span>
-<span><span class="co">## 90/90 - 0s - 2ms/step - loss: 0.2135 - sparse_categorical_accuracy: 0.9153 - val_loss: 0.2809 - val_sparse_categorical_accuracy: 0.8613 - learning_rate: 1.0000e-03</span></span>
+<span><span class="co">## 90/90 - 0s - 2ms/step - loss: 0.2074 - sparse_categorical_accuracy: 0.9212 - val_loss: 0.2334 - val_sparse_categorical_accuracy: 0.9043 - learning_rate: 1.0000e-03</span></span>
 <span><span class="co">## Epoch 44/500</span></span>
-<span><span class="co">## 90/90 - 0s - 2ms/step - loss: 0.2152 - sparse_categorical_accuracy: 0.9149 - val_loss: 0.3832 - val_sparse_categorical_accuracy: 0.8211 - learning_rate: 1.0000e-03</span></span>
+<span><span class="co">## 90/90 - 0s - 1ms/step - loss: 0.1961 - sparse_categorical_accuracy: 0.9313 - val_loss: 0.2343 - val_sparse_categorical_accuracy: 0.9182 - learning_rate: 1.0000e-03</span></span>
 <span><span class="co">## Epoch 45/500</span></span>
-<span><span class="co">## 90/90 - 0s - 2ms/step - loss: 0.2060 - sparse_categorical_accuracy: 0.9233 - val_loss: 0.3182 - val_sparse_categorical_accuracy: 0.8682 - learning_rate: 1.0000e-03</span></span>
+<span><span class="co">## 90/90 - 0s - 1ms/step - loss: 0.1923 - sparse_categorical_accuracy: 0.9302 - val_loss: 1.1167 - val_sparse_categorical_accuracy: 0.6755 - learning_rate: 1.0000e-03</span></span>
 <span><span class="co">## Epoch 46/500</span></span>
-<span><span class="co">## 90/90 - 0s - 2ms/step - loss: 0.2023 - sparse_categorical_accuracy: 0.9229 - val_loss: 0.2104 - val_sparse_categorical_accuracy: 0.9140 - learning_rate: 1.0000e-03</span></span>
+<span><span class="co">## 90/90 - 0s - 1ms/step - loss: 0.1823 - sparse_categorical_accuracy: 0.9347 - val_loss: 0.8034 - val_sparse_categorical_accuracy: 0.6935 - learning_rate: 1.0000e-03</span></span>
 <span><span class="co">## Epoch 47/500</span></span>
-<span><span class="co">## 90/90 - 0s - 1ms/step - loss: 0.1855 - sparse_categorical_accuracy: 0.9309 - val_loss: 0.2882 - val_sparse_categorical_accuracy: 0.8877 - learning_rate: 1.0000e-03</span></span>
+<span><span class="co">## 90/90 - 0s - 1ms/step - loss: 0.1635 - sparse_categorical_accuracy: 0.9444 - val_loss: 0.7104 - val_sparse_categorical_accuracy: 0.7115 - learning_rate: 1.0000e-03</span></span>
 <span><span class="co">## Epoch 48/500</span></span>
-<span><span class="co">## 90/90 - 0s - 1ms/step - loss: 0.1789 - sparse_categorical_accuracy: 0.9306 - val_loss: 0.4009 - val_sparse_categorical_accuracy: 0.8197 - learning_rate: 1.0000e-03</span></span>
+<span><span class="co">## 90/90 - 0s - 2ms/step - loss: 0.1587 - sparse_categorical_accuracy: 0.9434 - val_loss: 0.2001 - val_sparse_categorical_accuracy: 0.9196 - learning_rate: 1.0000e-03</span></span>
 <span><span class="co">## Epoch 49/500</span></span>
-<span><span class="co">## 90/90 - 0s - 1ms/step - loss: 0.1643 - sparse_categorical_accuracy: 0.9417 - val_loss: 0.2160 - val_sparse_categorical_accuracy: 0.9071 - learning_rate: 1.0000e-03</span></span>
+<span><span class="co">## 90/90 - 0s - 1ms/step - loss: 0.1492 - sparse_categorical_accuracy: 0.9483 - val_loss: 0.3399 - val_sparse_categorical_accuracy: 0.8294 - learning_rate: 1.0000e-03</span></span>
 <span><span class="co">## Epoch 50/500</span></span>
-<span><span class="co">## 90/90 - 0s - 1ms/step - loss: 0.1599 - sparse_categorical_accuracy: 0.9462 - val_loss: 0.6680 - val_sparse_categorical_accuracy: 0.6768 - learning_rate: 1.0000e-03</span></span>
+<span><span class="co">## 90/90 - 0s - 1ms/step - loss: 0.1738 - sparse_categorical_accuracy: 0.9382 - val_loss: 0.2680 - val_sparse_categorical_accuracy: 0.8890 - learning_rate: 1.0000e-03</span></span>
 <span><span class="co">## Epoch 51/500</span></span>
-<span><span class="co">## 90/90 - 0s - 1ms/step - loss: 0.1623 - sparse_categorical_accuracy: 0.9420 - val_loss: 1.2733 - val_sparse_categorical_accuracy: 0.7254 - learning_rate: 1.0000e-03</span></span>
+<span><span class="co">## 90/90 - 0s - 1ms/step - loss: 0.1530 - sparse_categorical_accuracy: 0.9486 - val_loss: 0.2891 - val_sparse_categorical_accuracy: 0.8544 - learning_rate: 1.0000e-03</span></span>
 <span><span class="co">## Epoch 52/500</span></span>
-<span><span class="co">## 90/90 - 0s - 2ms/step - loss: 0.1465 - sparse_categorical_accuracy: 0.9507 - val_loss: 1.1664 - val_sparse_categorical_accuracy: 0.6976 - learning_rate: 1.0000e-03</span></span>
+<span><span class="co">## 90/90 - 0s - 1ms/step - loss: 0.1320 - sparse_categorical_accuracy: 0.9538 - val_loss: 1.7182 - val_sparse_categorical_accuracy: 0.7032 - learning_rate: 1.0000e-03</span></span>
 <span><span class="co">## Epoch 53/500</span></span>
-<span><span class="co">## 90/90 - 0s - 1ms/step - loss: 0.1453 - sparse_categorical_accuracy: 0.9465 - val_loss: 1.3677 - val_sparse_categorical_accuracy: 0.7074 - learning_rate: 1.0000e-03</span></span>
+<span><span class="co">## 90/90 - 0s - 1ms/step - loss: 0.1277 - sparse_categorical_accuracy: 0.9559 - val_loss: 1.4269 - val_sparse_categorical_accuracy: 0.7254 - learning_rate: 1.0000e-03</span></span>
 <span><span class="co">## Epoch 54/500</span></span>
-<span><span class="co">## 90/90 - 0s - 1ms/step - loss: 0.1370 - sparse_categorical_accuracy: 0.9563 - val_loss: 0.2190 - val_sparse_categorical_accuracy: 0.9098 - learning_rate: 1.0000e-03</span></span>
+<span><span class="co">## 90/90 - 0s - 1ms/step - loss: 0.1249 - sparse_categorical_accuracy: 0.9611 - val_loss: 0.7962 - val_sparse_categorical_accuracy: 0.7365 - learning_rate: 1.0000e-03</span></span>
 <span><span class="co">## Epoch 55/500</span></span>
-<span><span class="co">## 90/90 - 0s - 1ms/step - loss: 0.1319 - sparse_categorical_accuracy: 0.9563 - val_loss: 0.2376 - val_sparse_categorical_accuracy: 0.8932 - learning_rate: 1.0000e-03</span></span>
+<span><span class="co">## 90/90 - 0s - 1ms/step - loss: 0.1214 - sparse_categorical_accuracy: 0.9611 - val_loss: 0.2158 - val_sparse_categorical_accuracy: 0.9071 - learning_rate: 1.0000e-03</span></span>
 <span><span class="co">## Epoch 56/500</span></span>
-<span><span class="co">## 90/90 - 0s - 2ms/step - loss: 0.1354 - sparse_categorical_accuracy: 0.9549 - val_loss: 0.1614 - val_sparse_categorical_accuracy: 0.9334 - learning_rate: 1.0000e-03</span></span>
+<span><span class="co">## 90/90 - 0s - 1ms/step - loss: 0.1258 - sparse_categorical_accuracy: 0.9604 - val_loss: 0.6349 - val_sparse_categorical_accuracy: 0.7642 - learning_rate: 1.0000e-03</span></span>
 <span><span class="co">## Epoch 57/500</span></span>
-<span><span class="co">## 90/90 - 0s - 2ms/step - loss: 0.1298 - sparse_categorical_accuracy: 0.9590 - val_loss: 0.1651 - val_sparse_categorical_accuracy: 0.9376 - learning_rate: 1.0000e-03</span></span>
+<span><span class="co">## 90/90 - 0s - 1ms/step - loss: 0.1268 - sparse_categorical_accuracy: 0.9604 - val_loss: 0.2129 - val_sparse_categorical_accuracy: 0.9071 - learning_rate: 1.0000e-03</span></span>
 <span><span class="co">## Epoch 58/500</span></span>
-<span><span class="co">## 90/90 - 0s - 2ms/step - loss: 0.1188 - sparse_categorical_accuracy: 0.9569 - val_loss: 0.2961 - val_sparse_categorical_accuracy: 0.8766 - learning_rate: 1.0000e-03</span></span>
+<span><span class="co">## 90/90 - 0s - 1ms/step - loss: 0.1088 - sparse_categorical_accuracy: 0.9639 - val_loss: 0.7215 - val_sparse_categorical_accuracy: 0.6657 - learning_rate: 1.0000e-03</span></span>
 <span><span class="co">## Epoch 59/500</span></span>
-<span><span class="co">## 90/90 - 0s - 2ms/step - loss: 0.1295 - sparse_categorical_accuracy: 0.9545 - val_loss: 0.1905 - val_sparse_categorical_accuracy: 0.9209 - learning_rate: 1.0000e-03</span></span>
+<span><span class="co">## 90/90 - 0s - 2ms/step - loss: 0.1219 - sparse_categorical_accuracy: 0.9569 - val_loss: 0.1600 - val_sparse_categorical_accuracy: 0.9320 - learning_rate: 1.0000e-03</span></span>
 <span><span class="co">## Epoch 60/500</span></span>
-<span><span class="co">## 90/90 - 0s - 1ms/step - loss: 0.1281 - sparse_categorical_accuracy: 0.9552 - val_loss: 0.1835 - val_sparse_categorical_accuracy: 0.9140 - learning_rate: 1.0000e-03</span></span>
+<span><span class="co">## 90/90 - 0s - 1ms/step - loss: 0.1190 - sparse_categorical_accuracy: 0.9642 - val_loss: 0.3411 - val_sparse_categorical_accuracy: 0.8558 - learning_rate: 1.0000e-03</span></span>
 <span><span class="co">## Epoch 61/500</span></span>
-<span><span class="co">## 90/90 - 0s - 1ms/step - loss: 0.1176 - sparse_categorical_accuracy: 0.9646 - val_loss: 1.6040 - val_sparse_categorical_accuracy: 0.6741 - learning_rate: 1.0000e-03</span></span>
+<span><span class="co">## 90/90 - 0s - 1ms/step - loss: 0.1177 - sparse_categorical_accuracy: 0.9653 - val_loss: 0.3984 - val_sparse_categorical_accuracy: 0.8502 - learning_rate: 1.0000e-03</span></span>
 <span><span class="co">## Epoch 62/500</span></span>
-<span><span class="co">## 90/90 - 0s - 1ms/step - loss: 0.1363 - sparse_categorical_accuracy: 0.9500 - val_loss: 2.0555 - val_sparse_categorical_accuracy: 0.6963 - learning_rate: 1.0000e-03</span></span>
+<span><span class="co">## 90/90 - 0s - 1ms/step - loss: 0.1309 - sparse_categorical_accuracy: 0.9542 - val_loss: 0.3462 - val_sparse_categorical_accuracy: 0.8211 - learning_rate: 1.0000e-03</span></span>
 <span><span class="co">## Epoch 63/500</span></span>
-<span><span class="co">## 90/90 - 0s - 1ms/step - loss: 0.1358 - sparse_categorical_accuracy: 0.9483 - val_loss: 3.2565 - val_sparse_categorical_accuracy: 0.6352 - learning_rate: 1.0000e-03</span></span>
+<span><span class="co">## 90/90 - 0s - 1ms/step - loss: 0.1291 - sparse_categorical_accuracy: 0.9514 - val_loss: 0.2303 - val_sparse_categorical_accuracy: 0.8960 - learning_rate: 1.0000e-03</span></span>
 <span><span class="co">## Epoch 64/500</span></span>
-<span><span class="co">## 90/90 - 0s - 2ms/step - loss: 0.1276 - sparse_categorical_accuracy: 0.9590 - val_loss: 0.3056 - val_sparse_categorical_accuracy: 0.8669 - learning_rate: 1.0000e-03</span></span>
+<span><span class="co">## 90/90 - 0s - 1ms/step - loss: 0.1200 - sparse_categorical_accuracy: 0.9628 - val_loss: 0.1735 - val_sparse_categorical_accuracy: 0.9223 - learning_rate: 1.0000e-03</span></span>
 <span><span class="co">## Epoch 65/500</span></span>
-<span><span class="co">## 90/90 - 0s - 2ms/step - loss: 0.1189 - sparse_categorical_accuracy: 0.9590 - val_loss: 0.3571 - val_sparse_categorical_accuracy: 0.8530 - learning_rate: 1.0000e-03</span></span>
+<span><span class="co">## 90/90 - 0s - 1ms/step - loss: 0.1138 - sparse_categorical_accuracy: 0.9625 - val_loss: 0.5628 - val_sparse_categorical_accuracy: 0.7920 - learning_rate: 1.0000e-03</span></span>
 <span><span class="co">## Epoch 66/500</span></span>
-<span><span class="co">## 90/90 - 0s - 2ms/step - loss: 0.1111 - sparse_categorical_accuracy: 0.9653 - val_loss: 0.1403 - val_sparse_categorical_accuracy: 0.9473 - learning_rate: 1.0000e-03</span></span>
+<span><span class="co">## 90/90 - 0s - 2ms/step - loss: 0.1065 - sparse_categorical_accuracy: 0.9667 - val_loss: 0.1540 - val_sparse_categorical_accuracy: 0.9417 - learning_rate: 1.0000e-03</span></span>
 <span><span class="co">## Epoch 67/500</span></span>
-<span><span class="co">## 90/90 - 0s - 2ms/step - loss: 0.1120 - sparse_categorical_accuracy: 0.9635 - val_loss: 1.6044 - val_sparse_categorical_accuracy: 0.6477 - learning_rate: 1.0000e-03</span></span>
+<span><span class="co">## 90/90 - 0s - 1ms/step - loss: 0.1156 - sparse_categorical_accuracy: 0.9618 - val_loss: 0.2319 - val_sparse_categorical_accuracy: 0.9043 - learning_rate: 1.0000e-03</span></span>
 <span><span class="co">## Epoch 68/500</span></span>
-<span><span class="co">## 90/90 - 0s - 2ms/step - loss: 0.1123 - sparse_categorical_accuracy: 0.9632 - val_loss: 0.6339 - val_sparse_categorical_accuracy: 0.7989 - learning_rate: 1.0000e-03</span></span>
+<span><span class="co">## 90/90 - 0s - 1ms/step - loss: 0.1100 - sparse_categorical_accuracy: 0.9597 - val_loss: 0.8668 - val_sparse_categorical_accuracy: 0.7406 - learning_rate: 1.0000e-03</span></span>
 <span><span class="co">## Epoch 69/500</span></span>
-<span><span class="co">## 90/90 - 0s - 2ms/step - loss: 0.1150 - sparse_categorical_accuracy: 0.9611 - val_loss: 0.3409 - val_sparse_categorical_accuracy: 0.8585 - learning_rate: 1.0000e-03</span></span>
+<span><span class="co">## 90/90 - 0s - 2ms/step - loss: 0.1165 - sparse_categorical_accuracy: 0.9618 - val_loss: 0.1460 - val_sparse_categorical_accuracy: 0.9515 - learning_rate: 1.0000e-03</span></span>
 <span><span class="co">## Epoch 70/500</span></span>
-<span><span class="co">## 90/90 - 0s - 1ms/step - loss: 0.1120 - sparse_categorical_accuracy: 0.9597 - val_loss: 0.1705 - val_sparse_categorical_accuracy: 0.9209 - learning_rate: 1.0000e-03</span></span>
+<span><span class="co">## 90/90 - 0s - 1ms/step - loss: 0.1178 - sparse_categorical_accuracy: 0.9608 - val_loss: 0.8988 - val_sparse_categorical_accuracy: 0.7046 - learning_rate: 1.0000e-03</span></span>
 <span><span class="co">## Epoch 71/500</span></span>
-<span><span class="co">## 90/90 - 0s - 2ms/step - loss: 0.1070 - sparse_categorical_accuracy: 0.9646 - val_loss: 0.3976 - val_sparse_categorical_accuracy: 0.8488 - learning_rate: 1.0000e-03</span></span>
+<span><span class="co">## 90/90 - 0s - 1ms/step - loss: 0.1068 - sparse_categorical_accuracy: 0.9642 - val_loss: 0.3645 - val_sparse_categorical_accuracy: 0.8585 - learning_rate: 1.0000e-03</span></span>
 <span><span class="co">## Epoch 72/500</span></span>
-<span><span class="co">## 90/90 - 0s - 2ms/step - loss: 0.1076 - sparse_categorical_accuracy: 0.9663 - val_loss: 0.1426 - val_sparse_categorical_accuracy: 0.9473 - learning_rate: 1.0000e-03</span></span>
+<span><span class="co">## 90/90 - 0s - 1ms/step - loss: 0.1106 - sparse_categorical_accuracy: 0.9653 - val_loss: 0.2813 - val_sparse_categorical_accuracy: 0.8932 - learning_rate: 1.0000e-03</span></span>
 <span><span class="co">## Epoch 73/500</span></span>
-<span><span class="co">## 90/90 - 0s - 2ms/step - loss: 0.1025 - sparse_categorical_accuracy: 0.9677 - val_loss: 0.3274 - val_sparse_categorical_accuracy: 0.8613 - learning_rate: 1.0000e-03</span></span>
+<span><span class="co">## 90/90 - 0s - 1ms/step - loss: 0.1117 - sparse_categorical_accuracy: 0.9632 - val_loss: 1.8641 - val_sparse_categorical_accuracy: 0.6741 - learning_rate: 1.0000e-03</span></span>
 <span><span class="co">## Epoch 74/500</span></span>
-<span><span class="co">## 90/90 - 0s - 2ms/step - loss: 0.1056 - sparse_categorical_accuracy: 0.9628 - val_loss: 0.2165 - val_sparse_categorical_accuracy: 0.9182 - learning_rate: 1.0000e-03</span></span>
+<span><span class="co">## 90/90 - 0s - 1ms/step - loss: 0.1058 - sparse_categorical_accuracy: 0.9625 - val_loss: 1.4979 - val_sparse_categorical_accuracy: 0.6893 - learning_rate: 1.0000e-03</span></span>
 <span><span class="co">## Epoch 75/500</span></span>
-<span><span class="co">## 90/90 - 0s - 1ms/step - loss: 0.1303 - sparse_categorical_accuracy: 0.9524 - val_loss: 0.6197 - val_sparse_categorical_accuracy: 0.7282 - learning_rate: 1.0000e-03</span></span>
+<span><span class="co">## 90/90 - 0s - 1ms/step - loss: 0.1223 - sparse_categorical_accuracy: 0.9594 - val_loss: 0.1508 - val_sparse_categorical_accuracy: 0.9473 - learning_rate: 1.0000e-03</span></span>
 <span><span class="co">## Epoch 76/500</span></span>
-<span><span class="co">## 90/90 - 0s - 1ms/step - loss: 0.1205 - sparse_categorical_accuracy: 0.9590 - val_loss: 0.2972 - val_sparse_categorical_accuracy: 0.8793 - learning_rate: 1.0000e-03</span></span>
+<span><span class="co">## 90/90 - 0s - 2ms/step - loss: 0.1140 - sparse_categorical_accuracy: 0.9601 - val_loss: 0.1400 - val_sparse_categorical_accuracy: 0.9320 - learning_rate: 1.0000e-03</span></span>
 <span><span class="co">## Epoch 77/500</span></span>
-<span><span class="co">## 90/90 - 0s - 1ms/step - loss: 0.1061 - sparse_categorical_accuracy: 0.9632 - val_loss: 0.2158 - val_sparse_categorical_accuracy: 0.9182 - learning_rate: 1.0000e-03</span></span>
+<span><span class="co">## 90/90 - 0s - 1ms/step - loss: 0.1050 - sparse_categorical_accuracy: 0.9639 - val_loss: 0.2169 - val_sparse_categorical_accuracy: 0.9126 - learning_rate: 1.0000e-03</span></span>
 <span><span class="co">## Epoch 78/500</span></span>
-<span><span class="co">## 90/90 - 0s - 1ms/step - loss: 0.0959 - sparse_categorical_accuracy: 0.9705 - val_loss: 0.1449 - val_sparse_categorical_accuracy: 0.9362 - learning_rate: 1.0000e-03</span></span>
+<span><span class="co">## 90/90 - 0s - 1ms/step - loss: 0.0986 - sparse_categorical_accuracy: 0.9712 - val_loss: 0.3240 - val_sparse_categorical_accuracy: 0.8724 - learning_rate: 1.0000e-03</span></span>
 <span><span class="co">## Epoch 79/500</span></span>
-<span><span class="co">## 90/90 - 0s - 2ms/step - loss: 0.0995 - sparse_categorical_accuracy: 0.9674 - val_loss: 0.1231 - val_sparse_categorical_accuracy: 0.9598 - learning_rate: 1.0000e-03</span></span>
+<span><span class="co">## 90/90 - 0s - 1ms/step - loss: 0.1081 - sparse_categorical_accuracy: 0.9632 - val_loss: 0.1525 - val_sparse_categorical_accuracy: 0.9362 - learning_rate: 1.0000e-03</span></span>
 <span><span class="co">## Epoch 80/500</span></span>
-<span><span class="co">## 90/90 - 0s - 1ms/step - loss: 0.1027 - sparse_categorical_accuracy: 0.9677 - val_loss: 0.5311 - val_sparse_categorical_accuracy: 0.7712 - learning_rate: 1.0000e-03</span></span>
+<span><span class="co">## 90/90 - 0s - 1ms/step - loss: 0.0995 - sparse_categorical_accuracy: 0.9667 - val_loss: 0.3123 - val_sparse_categorical_accuracy: 0.8641 - learning_rate: 1.0000e-03</span></span>
 <span><span class="co">## Epoch 81/500</span></span>
-<span><span class="co">## 90/90 - 0s - 2ms/step - loss: 0.1012 - sparse_categorical_accuracy: 0.9677 - val_loss: 0.1622 - val_sparse_categorical_accuracy: 0.9348 - learning_rate: 1.0000e-03</span></span>
+<span><span class="co">## 90/90 - 0s - 1ms/step - loss: 0.1058 - sparse_categorical_accuracy: 0.9632 - val_loss: 0.2851 - val_sparse_categorical_accuracy: 0.8779 - learning_rate: 1.0000e-03</span></span>
 <span><span class="co">## Epoch 82/500</span></span>
-<span><span class="co">## 90/90 - 0s - 1ms/step - loss: 0.0929 - sparse_categorical_accuracy: 0.9663 - val_loss: 0.1457 - val_sparse_categorical_accuracy: 0.9417 - learning_rate: 1.0000e-03</span></span>
+<span><span class="co">## 90/90 - 0s - 1ms/step - loss: 0.0958 - sparse_categorical_accuracy: 0.9681 - val_loss: 0.1417 - val_sparse_categorical_accuracy: 0.9362 - learning_rate: 1.0000e-03</span></span>
 <span><span class="co">## Epoch 83/500</span></span>
-<span><span class="co">## 90/90 - 0s - 1ms/step - loss: 0.1008 - sparse_categorical_accuracy: 0.9663 - val_loss: 1.6042 - val_sparse_categorical_accuracy: 0.6879 - learning_rate: 1.0000e-03</span></span>
+<span><span class="co">## 90/90 - 0s - 1ms/step - loss: 0.1005 - sparse_categorical_accuracy: 0.9660 - val_loss: 0.7325 - val_sparse_categorical_accuracy: 0.7157 - learning_rate: 1.0000e-03</span></span>
 <span><span class="co">## Epoch 84/500</span></span>
-<span><span class="co">## 90/90 - 0s - 1ms/step - loss: 0.0981 - sparse_categorical_accuracy: 0.9653 - val_loss: 0.1732 - val_sparse_categorical_accuracy: 0.9307 - learning_rate: 1.0000e-03</span></span>
+<span><span class="co">## 90/90 - 0s - 2ms/step - loss: 0.0966 - sparse_categorical_accuracy: 0.9684 - val_loss: 0.1322 - val_sparse_categorical_accuracy: 0.9404 - learning_rate: 1.0000e-03</span></span>
 <span><span class="co">## Epoch 85/500</span></span>
-<span><span class="co">## 90/90 - 0s - 1ms/step - loss: 0.1006 - sparse_categorical_accuracy: 0.9674 - val_loss: 1.8495 - val_sparse_categorical_accuracy: 0.6976 - learning_rate: 1.0000e-03</span></span>
+<span><span class="co">## 90/90 - 0s - 2ms/step - loss: 0.1018 - sparse_categorical_accuracy: 0.9667 - val_loss: 0.1202 - val_sparse_categorical_accuracy: 0.9473 - learning_rate: 1.0000e-03</span></span>
 <span><span class="co">## Epoch 86/500</span></span>
-<span><span class="co">## 90/90 - 0s - 1ms/step - loss: 0.1005 - sparse_categorical_accuracy: 0.9660 - val_loss: 0.1263 - val_sparse_categorical_accuracy: 0.9459 - learning_rate: 1.0000e-03</span></span>
+<span><span class="co">## 90/90 - 0s - 1ms/step - loss: 0.0995 - sparse_categorical_accuracy: 0.9649 - val_loss: 0.2956 - val_sparse_categorical_accuracy: 0.8766 - learning_rate: 1.0000e-03</span></span>
 <span><span class="co">## Epoch 87/500</span></span>
-<span><span class="co">## 90/90 - 0s - 1ms/step - loss: 0.1054 - sparse_categorical_accuracy: 0.9632 - val_loss: 0.2135 - val_sparse_categorical_accuracy: 0.9140 - learning_rate: 1.0000e-03</span></span>
+<span><span class="co">## 90/90 - 0s - 1ms/step - loss: 0.1068 - sparse_categorical_accuracy: 0.9632 - val_loss: 0.3050 - val_sparse_categorical_accuracy: 0.8710 - learning_rate: 1.0000e-03</span></span>
 <span><span class="co">## Epoch 88/500</span></span>
-<span><span class="co">## 90/90 - 0s - 2ms/step - loss: 0.1051 - sparse_categorical_accuracy: 0.9656 - val_loss: 1.3499 - val_sparse_categorical_accuracy: 0.6630 - learning_rate: 1.0000e-03</span></span>
+<span><span class="co">## 90/90 - 0s - 1ms/step - loss: 0.1002 - sparse_categorical_accuracy: 0.9674 - val_loss: 0.1675 - val_sparse_categorical_accuracy: 0.9196 - learning_rate: 1.0000e-03</span></span>
 <span><span class="co">## Epoch 89/500</span></span>
-<span><span class="co">## 90/90 - 0s - 2ms/step - loss: 0.1037 - sparse_categorical_accuracy: 0.9615 - val_loss: 0.2271 - val_sparse_categorical_accuracy: 0.9029 - learning_rate: 1.0000e-03</span></span>
+<span><span class="co">## 90/90 - 0s - 1ms/step - loss: 0.1042 - sparse_categorical_accuracy: 0.9608 - val_loss: 0.7157 - val_sparse_categorical_accuracy: 0.7115 - learning_rate: 1.0000e-03</span></span>
 <span><span class="co">## Epoch 90/500</span></span>
-<span><span class="co">## 90/90 - 0s - 2ms/step - loss: 0.0976 - sparse_categorical_accuracy: 0.9670 - val_loss: 0.7824 - val_sparse_categorical_accuracy: 0.7129 - learning_rate: 1.0000e-03</span></span>
+<span><span class="co">## 90/90 - 0s - 1ms/step - loss: 0.0998 - sparse_categorical_accuracy: 0.9694 - val_loss: 0.1576 - val_sparse_categorical_accuracy: 0.9251 - learning_rate: 1.0000e-03</span></span>
 <span><span class="co">## Epoch 91/500</span></span>
-<span><span class="co">## 90/90 - 0s - 1ms/step - loss: 0.0995 - sparse_categorical_accuracy: 0.9677 - val_loss: 0.2369 - val_sparse_categorical_accuracy: 0.8918 - learning_rate: 1.0000e-03</span></span>
+<span><span class="co">## 90/90 - 0s - 1ms/step - loss: 0.1187 - sparse_categorical_accuracy: 0.9583 - val_loss: 0.4125 - val_sparse_categorical_accuracy: 0.8031 - learning_rate: 1.0000e-03</span></span>
 <span><span class="co">## Epoch 92/500</span></span>
-<span><span class="co">## 90/90 - 0s - 2ms/step - loss: 0.0986 - sparse_categorical_accuracy: 0.9670 - val_loss: 0.1185 - val_sparse_categorical_accuracy: 0.9584 - learning_rate: 1.0000e-03</span></span>
+<span><span class="co">## 90/90 - 0s - 1ms/step - loss: 0.0973 - sparse_categorical_accuracy: 0.9688 - val_loss: 0.3084 - val_sparse_categorical_accuracy: 0.8696 - learning_rate: 1.0000e-03</span></span>
 <span><span class="co">## Epoch 93/500</span></span>
-<span><span class="co">## 90/90 - 0s - 2ms/step - loss: 0.0957 - sparse_categorical_accuracy: 0.9670 - val_loss: 1.8619 - val_sparse_categorical_accuracy: 0.5270 - learning_rate: 1.0000e-03</span></span>
+<span><span class="co">## 90/90 - 0s - 1ms/step - loss: 0.0972 - sparse_categorical_accuracy: 0.9677 - val_loss: 0.1446 - val_sparse_categorical_accuracy: 0.9334 - learning_rate: 1.0000e-03</span></span>
 <span><span class="co">## Epoch 94/500</span></span>
-<span><span class="co">## 90/90 - 0s - 2ms/step - loss: 0.0999 - sparse_categorical_accuracy: 0.9642 - val_loss: 0.4507 - val_sparse_categorical_accuracy: 0.8350 - learning_rate: 1.0000e-03</span></span>
+<span><span class="co">## 90/90 - 0s - 1ms/step - loss: 0.0965 - sparse_categorical_accuracy: 0.9632 - val_loss: 0.1265 - val_sparse_categorical_accuracy: 0.9584 - learning_rate: 1.0000e-03</span></span>
 <span><span class="co">## Epoch 95/500</span></span>
-<span><span class="co">## 90/90 - 0s - 2ms/step - loss: 0.0939 - sparse_categorical_accuracy: 0.9653 - val_loss: 1.0529 - val_sparse_categorical_accuracy: 0.6893 - learning_rate: 1.0000e-03</span></span>
+<span><span class="co">## 90/90 - 0s - 2ms/step - loss: 0.0927 - sparse_categorical_accuracy: 0.9667 - val_loss: 0.1507 - val_sparse_categorical_accuracy: 0.9376 - learning_rate: 1.0000e-03</span></span>
 <span><span class="co">## Epoch 96/500</span></span>
-<span><span class="co">## 90/90 - 0s - 2ms/step - loss: 0.1066 - sparse_categorical_accuracy: 0.9608 - val_loss: 0.1416 - val_sparse_categorical_accuracy: 0.9417 - learning_rate: 1.0000e-03</span></span>
+<span><span class="co">## 90/90 - 0s - 1ms/step - loss: 0.1050 - sparse_categorical_accuracy: 0.9618 - val_loss: 0.5005 - val_sparse_categorical_accuracy: 0.8086 - learning_rate: 1.0000e-03</span></span>
 <span><span class="co">## Epoch 97/500</span></span>
-<span><span class="co">## 90/90 - 0s - 1ms/step - loss: 0.1068 - sparse_categorical_accuracy: 0.9618 - val_loss: 0.5324 - val_sparse_categorical_accuracy: 0.7961 - learning_rate: 1.0000e-03</span></span>
+<span><span class="co">## 90/90 - 0s - 1ms/step - loss: 0.1134 - sparse_categorical_accuracy: 0.9597 - val_loss: 0.1732 - val_sparse_categorical_accuracy: 0.9390 - learning_rate: 1.0000e-03</span></span>
 <span><span class="co">## Epoch 98/500</span></span>
-<span><span class="co">## 90/90 - 0s - 1ms/step - loss: 0.0916 - sparse_categorical_accuracy: 0.9694 - val_loss: 0.8154 - val_sparse_categorical_accuracy: 0.7656 - learning_rate: 1.0000e-03</span></span>
+<span><span class="co">## 90/90 - 0s - 1ms/step - loss: 0.0946 - sparse_categorical_accuracy: 0.9698 - val_loss: 1.8677 - val_sparse_categorical_accuracy: 0.6949 - learning_rate: 1.0000e-03</span></span>
 <span><span class="co">## Epoch 99/500</span></span>
-<span><span class="co">## 90/90 - 0s - 2ms/step - loss: 0.0942 - sparse_categorical_accuracy: 0.9639 - val_loss: 0.1314 - val_sparse_categorical_accuracy: 0.9542 - learning_rate: 1.0000e-03</span></span>
+<span><span class="co">## 90/90 - 0s - 1ms/step - loss: 0.1018 - sparse_categorical_accuracy: 0.9632 - val_loss: 1.1308 - val_sparse_categorical_accuracy: 0.6782 - learning_rate: 1.0000e-03</span></span>
 <span><span class="co">## Epoch 100/500</span></span>
-<span><span class="co">## 90/90 - 0s - 1ms/step - loss: 0.0938 - sparse_categorical_accuracy: 0.9698 - val_loss: 2.2256 - val_sparse_categorical_accuracy: 0.6366 - learning_rate: 1.0000e-03</span></span>
+<span><span class="co">## 90/90 - 0s - 1ms/step - loss: 0.0990 - sparse_categorical_accuracy: 0.9674 - val_loss: 0.3846 - val_sparse_categorical_accuracy: 0.8558 - learning_rate: 1.0000e-03</span></span>
 <span><span class="co">## Epoch 101/500</span></span>
-<span><span class="co">## 90/90 - 0s - 2ms/step - loss: 0.0853 - sparse_categorical_accuracy: 0.9750 - val_loss: 0.1660 - val_sparse_categorical_accuracy: 0.9376 - learning_rate: 1.0000e-03</span></span>
+<span><span class="co">## 90/90 - 0s - 1ms/step - loss: 0.0891 - sparse_categorical_accuracy: 0.9705 - val_loss: 0.1480 - val_sparse_categorical_accuracy: 0.9362 - learning_rate: 1.0000e-03</span></span>
 <span><span class="co">## Epoch 102/500</span></span>
-<span><span class="co">## 90/90 - 0s - 2ms/step - loss: 0.0957 - sparse_categorical_accuracy: 0.9691 - val_loss: 0.2444 - val_sparse_categorical_accuracy: 0.8932 - learning_rate: 1.0000e-03</span></span>
+<span><span class="co">## 90/90 - 0s - 1ms/step - loss: 0.1003 - sparse_categorical_accuracy: 0.9677 - val_loss: 0.1713 - val_sparse_categorical_accuracy: 0.9390 - learning_rate: 1.0000e-03</span></span>
 <span><span class="co">## Epoch 103/500</span></span>
-<span><span class="co">## 90/90 - 0s - 1ms/step - loss: 0.0949 - sparse_categorical_accuracy: 0.9660 - val_loss: 1.3661 - val_sparse_categorical_accuracy: 0.6519 - learning_rate: 1.0000e-03</span></span>
+<span><span class="co">## 90/90 - 0s - 1ms/step - loss: 0.0910 - sparse_categorical_accuracy: 0.9722 - val_loss: 0.1415 - val_sparse_categorical_accuracy: 0.9542 - learning_rate: 1.0000e-03</span></span>
 <span><span class="co">## Epoch 104/500</span></span>
-<span><span class="co">## 90/90 - 0s - 1ms/step - loss: 0.1033 - sparse_categorical_accuracy: 0.9622 - val_loss: 0.2086 - val_sparse_categorical_accuracy: 0.9126 - learning_rate: 1.0000e-03</span></span>
+<span><span class="co">## 90/90 - 0s - 1ms/step - loss: 0.0998 - sparse_categorical_accuracy: 0.9622 - val_loss: 0.5677 - val_sparse_categorical_accuracy: 0.7753 - learning_rate: 1.0000e-03</span></span>
 <span><span class="co">## Epoch 105/500</span></span>
-<span><span class="co">## 90/90 - 0s - 1ms/step - loss: 0.0981 - sparse_categorical_accuracy: 0.9663 - val_loss: 0.2216 - val_sparse_categorical_accuracy: 0.8988 - learning_rate: 1.0000e-03</span></span>
+<span><span class="co">## 90/90 - 0s - 1ms/step - loss: 0.1037 - sparse_categorical_accuracy: 0.9649 - val_loss: 0.9369 - val_sparse_categorical_accuracy: 0.6241 - learning_rate: 1.0000e-03</span></span>
 <span><span class="co">## Epoch 106/500</span></span>
-<span><span class="co">## 90/90 - 0s - 1ms/step - loss: 0.0895 - sparse_categorical_accuracy: 0.9681 - val_loss: 2.2126 - val_sparse_categorical_accuracy: 0.6269 - learning_rate: 1.0000e-03</span></span>
+<span><span class="co">## 90/90 - 1s - 10ms/step - loss: 0.0850 - sparse_categorical_accuracy: 0.9708 - val_loss: 0.1388 - val_sparse_categorical_accuracy: 0.9487 - learning_rate: 5.0000e-04</span></span>
 <span><span class="co">## Epoch 107/500</span></span>
-<span><span class="co">## 90/90 - 0s - 1ms/step - loss: 0.0939 - sparse_categorical_accuracy: 0.9670 - val_loss: 0.4274 - val_sparse_categorical_accuracy: 0.8128 - learning_rate: 1.0000e-03</span></span>
+<span><span class="co">## 90/90 - 0s - 1ms/step - loss: 0.0833 - sparse_categorical_accuracy: 0.9743 - val_loss: 0.1413 - val_sparse_categorical_accuracy: 0.9376 - learning_rate: 5.0000e-04</span></span>
 <span><span class="co">## Epoch 108/500</span></span>
-<span><span class="co">## 90/90 - 0s - 1ms/step - loss: 0.0894 - sparse_categorical_accuracy: 0.9719 - val_loss: 0.2745 - val_sparse_categorical_accuracy: 0.8946 - learning_rate: 1.0000e-03</span></span>
+<span><span class="co">## 90/90 - 0s - 1ms/step - loss: 0.0775 - sparse_categorical_accuracy: 0.9767 - val_loss: 0.1223 - val_sparse_categorical_accuracy: 0.9445 - learning_rate: 5.0000e-04</span></span>
 <span><span class="co">## Epoch 109/500</span></span>
-<span><span class="co">## 90/90 - 0s - 2ms/step - loss: 0.0937 - sparse_categorical_accuracy: 0.9722 - val_loss: 0.1248 - val_sparse_categorical_accuracy: 0.9487 - learning_rate: 1.0000e-03</span></span>
+<span><span class="co">## 90/90 - 0s - 1ms/step - loss: 0.0837 - sparse_categorical_accuracy: 0.9753 - val_loss: 0.1275 - val_sparse_categorical_accuracy: 0.9528 - learning_rate: 5.0000e-04</span></span>
 <span><span class="co">## Epoch 110/500</span></span>
-<span><span class="co">## 90/90 - 0s - 1ms/step - loss: 0.0965 - sparse_categorical_accuracy: 0.9656 - val_loss: 0.5962 - val_sparse_categorical_accuracy: 0.7989 - learning_rate: 1.0000e-03</span></span>
+<span><span class="co">## 90/90 - 0s - 1ms/step - loss: 0.0859 - sparse_categorical_accuracy: 0.9705 - val_loss: 0.1209 - val_sparse_categorical_accuracy: 0.9487 - learning_rate: 5.0000e-04</span></span>
 <span><span class="co">## Epoch 111/500</span></span>
-<span><span class="co">## 90/90 - 0s - 1ms/step - loss: 0.0825 - sparse_categorical_accuracy: 0.9726 - val_loss: 0.1671 - val_sparse_categorical_accuracy: 0.9307 - learning_rate: 1.0000e-03</span></span>
+<span><span class="co">## 90/90 - 0s - 2ms/step - loss: 0.0780 - sparse_categorical_accuracy: 0.9757 - val_loss: 0.1142 - val_sparse_categorical_accuracy: 0.9584 - learning_rate: 5.0000e-04</span></span>
 <span><span class="co">## Epoch 112/500</span></span>
-<span><span class="co">## 90/90 - 0s - 1ms/step - loss: 0.0879 - sparse_categorical_accuracy: 0.9694 - val_loss: 0.1298 - val_sparse_categorical_accuracy: 0.9390 - learning_rate: 1.0000e-03</span></span>
+<span><span class="co">## 90/90 - 0s - 1ms/step - loss: 0.0801 - sparse_categorical_accuracy: 0.9736 - val_loss: 0.1316 - val_sparse_categorical_accuracy: 0.9459 - learning_rate: 5.0000e-04</span></span>
 <span><span class="co">## Epoch 113/500</span></span>
-<span><span class="co">## 90/90 - 1s - 13ms/step - loss: 0.0785 - sparse_categorical_accuracy: 0.9757 - val_loss: 0.1539 - val_sparse_categorical_accuracy: 0.9348 - learning_rate: 5.0000e-04</span></span>
+<span><span class="co">## 90/90 - 0s - 1ms/step - loss: 0.0740 - sparse_categorical_accuracy: 0.9753 - val_loss: 0.3385 - val_sparse_categorical_accuracy: 0.8544 - learning_rate: 5.0000e-04</span></span>
 <span><span class="co">## Epoch 114/500</span></span>
-<span><span class="co">## 90/90 - 0s - 1ms/step - loss: 0.0914 - sparse_categorical_accuracy: 0.9722 - val_loss: 0.1216 - val_sparse_categorical_accuracy: 0.9542 - learning_rate: 5.0000e-04</span></span>
+<span><span class="co">## 90/90 - 0s - 1ms/step - loss: 0.0907 - sparse_categorical_accuracy: 0.9722 - val_loss: 0.1222 - val_sparse_categorical_accuracy: 0.9612 - learning_rate: 5.0000e-04</span></span>
 <span><span class="co">## Epoch 115/500</span></span>
-<span><span class="co">## 90/90 - 0s - 2ms/step - loss: 0.0754 - sparse_categorical_accuracy: 0.9740 - val_loss: 0.1167 - val_sparse_categorical_accuracy: 0.9431 - learning_rate: 5.0000e-04</span></span>
+<span><span class="co">## 90/90 - 0s - 2ms/step - loss: 0.0767 - sparse_categorical_accuracy: 0.9733 - val_loss: 0.1115 - val_sparse_categorical_accuracy: 0.9584 - learning_rate: 5.0000e-04</span></span>
 <span><span class="co">## Epoch 116/500</span></span>
-<span><span class="co">## 90/90 - 0s - 1ms/step - loss: 0.0786 - sparse_categorical_accuracy: 0.9753 - val_loss: 0.1977 - val_sparse_categorical_accuracy: 0.9279 - learning_rate: 5.0000e-04</span></span>
+<span><span class="co">## 90/90 - 0s - 1ms/step - loss: 0.0786 - sparse_categorical_accuracy: 0.9750 - val_loss: 0.1667 - val_sparse_categorical_accuracy: 0.9431 - learning_rate: 5.0000e-04</span></span>
 <span><span class="co">## Epoch 117/500</span></span>
-<span><span class="co">## 90/90 - 0s - 1ms/step - loss: 0.0758 - sparse_categorical_accuracy: 0.9736 - val_loss: 0.1724 - val_sparse_categorical_accuracy: 0.9431 - learning_rate: 5.0000e-04</span></span>
+<span><span class="co">## 90/90 - 0s - 1ms/step - loss: 0.0760 - sparse_categorical_accuracy: 0.9750 - val_loss: 0.1519 - val_sparse_categorical_accuracy: 0.9376 - learning_rate: 5.0000e-04</span></span>
 <span><span class="co">## Epoch 118/500</span></span>
-<span><span class="co">## 90/90 - 0s - 2ms/step - loss: 0.0793 - sparse_categorical_accuracy: 0.9712 - val_loss: 0.1316 - val_sparse_categorical_accuracy: 0.9501 - learning_rate: 5.0000e-04</span></span>
+<span><span class="co">## 90/90 - 0s - 1ms/step - loss: 0.0797 - sparse_categorical_accuracy: 0.9722 - val_loss: 0.1555 - val_sparse_categorical_accuracy: 0.9459 - learning_rate: 5.0000e-04</span></span>
 <span><span class="co">## Epoch 119/500</span></span>
-<span><span class="co">## 90/90 - 0s - 2ms/step - loss: 0.0781 - sparse_categorical_accuracy: 0.9750 - val_loss: 0.1054 - val_sparse_categorical_accuracy: 0.9598 - learning_rate: 5.0000e-04</span></span>
+<span><span class="co">## 90/90 - 0s - 1ms/step - loss: 0.0793 - sparse_categorical_accuracy: 0.9729 - val_loss: 0.1252 - val_sparse_categorical_accuracy: 0.9598 - learning_rate: 5.0000e-04</span></span>
 <span><span class="co">## Epoch 120/500</span></span>
-<span><span class="co">## 90/90 - 0s - 2ms/step - loss: 0.0808 - sparse_categorical_accuracy: 0.9740 - val_loss: 0.1425 - val_sparse_categorical_accuracy: 0.9515 - learning_rate: 5.0000e-04</span></span>
+<span><span class="co">## 90/90 - 0s - 1ms/step - loss: 0.0803 - sparse_categorical_accuracy: 0.9747 - val_loss: 0.2298 - val_sparse_categorical_accuracy: 0.9015 - learning_rate: 5.0000e-04</span></span>
 <span><span class="co">## Epoch 121/500</span></span>
-<span><span class="co">## 90/90 - 0s - 2ms/step - loss: 0.0761 - sparse_categorical_accuracy: 0.9753 - val_loss: 0.1359 - val_sparse_categorical_accuracy: 0.9431 - learning_rate: 5.0000e-04</span></span>
+<span><span class="co">## 90/90 - 0s - 1ms/step - loss: 0.0793 - sparse_categorical_accuracy: 0.9736 - val_loss: 0.1140 - val_sparse_categorical_accuracy: 0.9528 - learning_rate: 5.0000e-04</span></span>
 <span><span class="co">## Epoch 122/500</span></span>
-<span><span class="co">## 90/90 - 0s - 1ms/step - loss: 0.0813 - sparse_categorical_accuracy: 0.9719 - val_loss: 0.1067 - val_sparse_categorical_accuracy: 0.9556 - learning_rate: 5.0000e-04</span></span>
+<span><span class="co">## 90/90 - 0s - 1ms/step - loss: 0.0795 - sparse_categorical_accuracy: 0.9740 - val_loss: 0.1154 - val_sparse_categorical_accuracy: 0.9542 - learning_rate: 5.0000e-04</span></span>
 <span><span class="co">## Epoch 123/500</span></span>
-<span><span class="co">## 90/90 - 0s - 2ms/step - loss: 0.0778 - sparse_categorical_accuracy: 0.9767 - val_loss: 0.1328 - val_sparse_categorical_accuracy: 0.9528 - learning_rate: 5.0000e-04</span></span>
+<span><span class="co">## 90/90 - 0s - 1ms/step - loss: 0.0768 - sparse_categorical_accuracy: 0.9740 - val_loss: 0.3421 - val_sparse_categorical_accuracy: 0.8724 - learning_rate: 5.0000e-04</span></span>
 <span><span class="co">## Epoch 124/500</span></span>
-<span><span class="co">## 90/90 - 0s - 2ms/step - loss: 0.0822 - sparse_categorical_accuracy: 0.9705 - val_loss: 0.1931 - val_sparse_categorical_accuracy: 0.9265 - learning_rate: 5.0000e-04</span></span>
+<span><span class="co">## 90/90 - 0s - 1ms/step - loss: 0.0848 - sparse_categorical_accuracy: 0.9694 - val_loss: 0.1126 - val_sparse_categorical_accuracy: 0.9598 - learning_rate: 5.0000e-04</span></span>
 <span><span class="co">## Epoch 125/500</span></span>
-<span><span class="co">## 90/90 - 0s - 2ms/step - loss: 0.0797 - sparse_categorical_accuracy: 0.9719 - val_loss: 0.1279 - val_sparse_categorical_accuracy: 0.9459 - learning_rate: 5.0000e-04</span></span>
+<span><span class="co">## 90/90 - 0s - 1ms/step - loss: 0.0800 - sparse_categorical_accuracy: 0.9722 - val_loss: 0.1821 - val_sparse_categorical_accuracy: 0.9293 - learning_rate: 5.0000e-04</span></span>
 <span><span class="co">## Epoch 126/500</span></span>
-<span><span class="co">## 90/90 - 0s - 1ms/step - loss: 0.0761 - sparse_categorical_accuracy: 0.9743 - val_loss: 0.2298 - val_sparse_categorical_accuracy: 0.9154 - learning_rate: 5.0000e-04</span></span>
+<span><span class="co">## 90/90 - 0s - 2ms/step - loss: 0.0787 - sparse_categorical_accuracy: 0.9753 - val_loss: 0.1105 - val_sparse_categorical_accuracy: 0.9570 - learning_rate: 5.0000e-04</span></span>
 <span><span class="co">## Epoch 127/500</span></span>
-<span><span class="co">## 90/90 - 0s - 1ms/step - loss: 0.0740 - sparse_categorical_accuracy: 0.9802 - val_loss: 0.2257 - val_sparse_categorical_accuracy: 0.9223 - learning_rate: 5.0000e-04</span></span>
+<span><span class="co">## 90/90 - 0s - 2ms/step - loss: 0.0799 - sparse_categorical_accuracy: 0.9760 - val_loss: 0.1091 - val_sparse_categorical_accuracy: 0.9584 - learning_rate: 5.0000e-04</span></span>
 <span><span class="co">## Epoch 128/500</span></span>
-<span><span class="co">## 90/90 - 0s - 2ms/step - loss: 0.0772 - sparse_categorical_accuracy: 0.9760 - val_loss: 0.1223 - val_sparse_categorical_accuracy: 0.9515 - learning_rate: 5.0000e-04</span></span>
+<span><span class="co">## 90/90 - 0s - 1ms/step - loss: 0.0811 - sparse_categorical_accuracy: 0.9736 - val_loss: 0.1287 - val_sparse_categorical_accuracy: 0.9515 - learning_rate: 5.0000e-04</span></span>
 <span><span class="co">## Epoch 129/500</span></span>
-<span><span class="co">## 90/90 - 0s - 1ms/step - loss: 0.0760 - sparse_categorical_accuracy: 0.9764 - val_loss: 0.1443 - val_sparse_categorical_accuracy: 0.9501 - learning_rate: 5.0000e-04</span></span>
+<span><span class="co">## 90/90 - 0s - 1ms/step - loss: 0.0797 - sparse_categorical_accuracy: 0.9736 - val_loss: 0.1143 - val_sparse_categorical_accuracy: 0.9584 - learning_rate: 5.0000e-04</span></span>
 <span><span class="co">## Epoch 130/500</span></span>
-<span><span class="co">## 90/90 - 0s - 1ms/step - loss: 0.0841 - sparse_categorical_accuracy: 0.9743 - val_loss: 0.1104 - val_sparse_categorical_accuracy: 0.9612 - learning_rate: 5.0000e-04</span></span>
+<span><span class="co">## 90/90 - 0s - 1ms/step - loss: 0.0850 - sparse_categorical_accuracy: 0.9750 - val_loss: 0.1581 - val_sparse_categorical_accuracy: 0.9376 - learning_rate: 5.0000e-04</span></span>
 <span><span class="co">## Epoch 131/500</span></span>
-<span><span class="co">## 90/90 - 0s - 1ms/step - loss: 0.0697 - sparse_categorical_accuracy: 0.9785 - val_loss: 0.1062 - val_sparse_categorical_accuracy: 0.9584 - learning_rate: 5.0000e-04</span></span>
+<span><span class="co">## 90/90 - 0s - 1ms/step - loss: 0.0707 - sparse_categorical_accuracy: 0.9792 - val_loss: 0.1129 - val_sparse_categorical_accuracy: 0.9639 - learning_rate: 5.0000e-04</span></span>
 <span><span class="co">## Epoch 132/500</span></span>
-<span><span class="co">## 90/90 - 0s - 1ms/step - loss: 0.0739 - sparse_categorical_accuracy: 0.9743 - val_loss: 0.1153 - val_sparse_categorical_accuracy: 0.9653 - learning_rate: 5.0000e-04</span></span>
+<span><span class="co">## 90/90 - 0s - 1ms/step - loss: 0.0725 - sparse_categorical_accuracy: 0.9740 - val_loss: 0.1425 - val_sparse_categorical_accuracy: 0.9501 - learning_rate: 5.0000e-04</span></span>
 <span><span class="co">## Epoch 133/500</span></span>
-<span><span class="co">## 90/90 - 0s - 1ms/step - loss: 0.0704 - sparse_categorical_accuracy: 0.9781 - val_loss: 0.1186 - val_sparse_categorical_accuracy: 0.9473 - learning_rate: 5.0000e-04</span></span>
+<span><span class="co">## 90/90 - 0s - 2ms/step - loss: 0.0705 - sparse_categorical_accuracy: 0.9764 - val_loss: 0.1098 - val_sparse_categorical_accuracy: 0.9612 - learning_rate: 5.0000e-04</span></span>
 <span><span class="co">## Epoch 134/500</span></span>
-<span><span class="co">## 90/90 - 0s - 1ms/step - loss: 0.0753 - sparse_categorical_accuracy: 0.9774 - val_loss: 0.2290 - val_sparse_categorical_accuracy: 0.9196 - learning_rate: 5.0000e-04</span></span>
+<span><span class="co">## 90/90 - 0s - 1ms/step - loss: 0.0767 - sparse_categorical_accuracy: 0.9736 - val_loss: 0.1884 - val_sparse_categorical_accuracy: 0.9376 - learning_rate: 5.0000e-04</span></span>
 <span><span class="co">## Epoch 135/500</span></span>
-<span><span class="co">## 90/90 - 0s - 1ms/step - loss: 0.0734 - sparse_categorical_accuracy: 0.9743 - val_loss: 0.1498 - val_sparse_categorical_accuracy: 0.9501 - learning_rate: 5.0000e-04</span></span>
+<span><span class="co">## 90/90 - 0s - 1ms/step - loss: 0.0775 - sparse_categorical_accuracy: 0.9743 - val_loss: 0.2899 - val_sparse_categorical_accuracy: 0.8988 - learning_rate: 5.0000e-04</span></span>
 <span><span class="co">## Epoch 136/500</span></span>
-<span><span class="co">## 90/90 - 0s - 1ms/step - loss: 0.0773 - sparse_categorical_accuracy: 0.9712 - val_loss: 0.1266 - val_sparse_categorical_accuracy: 0.9570 - learning_rate: 5.0000e-04</span></span>
+<span><span class="co">## 90/90 - 0s - 1ms/step - loss: 0.0807 - sparse_categorical_accuracy: 0.9701 - val_loss: 0.1152 - val_sparse_categorical_accuracy: 0.9584 - learning_rate: 5.0000e-04</span></span>
 <span><span class="co">## Epoch 137/500</span></span>
-<span><span class="co">## 90/90 - 0s - 1ms/step - loss: 0.0807 - sparse_categorical_accuracy: 0.9740 - val_loss: 0.1143 - val_sparse_categorical_accuracy: 0.9639 - learning_rate: 5.0000e-04</span></span>
+<span><span class="co">## 90/90 - 0s - 1ms/step - loss: 0.0839 - sparse_categorical_accuracy: 0.9750 - val_loss: 0.3530 - val_sparse_categorical_accuracy: 0.8793 - learning_rate: 5.0000e-04</span></span>
 <span><span class="co">## Epoch 138/500</span></span>
-<span><span class="co">## 90/90 - 0s - 1ms/step - loss: 0.0755 - sparse_categorical_accuracy: 0.9753 - val_loss: 0.2602 - val_sparse_categorical_accuracy: 0.9015 - learning_rate: 5.0000e-04</span></span>
+<span><span class="co">## 90/90 - 0s - 1ms/step - loss: 0.0770 - sparse_categorical_accuracy: 0.9743 - val_loss: 0.1267 - val_sparse_categorical_accuracy: 0.9473 - learning_rate: 5.0000e-04</span></span>
 <span><span class="co">## Epoch 139/500</span></span>
-<span><span class="co">## 90/90 - 0s - 2ms/step - loss: 0.0807 - sparse_categorical_accuracy: 0.9750 - val_loss: 0.0994 - val_sparse_categorical_accuracy: 0.9626 - learning_rate: 5.0000e-04</span></span>
+<span><span class="co">## 90/90 - 0s - 2ms/step - loss: 0.0845 - sparse_categorical_accuracy: 0.9719 - val_loss: 0.1043 - val_sparse_categorical_accuracy: 0.9626 - learning_rate: 5.0000e-04</span></span>
 <span><span class="co">## Epoch 140/500</span></span>
-<span><span class="co">## 90/90 - 0s - 1ms/step - loss: 0.0697 - sparse_categorical_accuracy: 0.9774 - val_loss: 0.2056 - val_sparse_categorical_accuracy: 0.9182 - learning_rate: 5.0000e-04</span></span>
+<span><span class="co">## 90/90 - 0s - 1ms/step - loss: 0.0701 - sparse_categorical_accuracy: 0.9799 - val_loss: 0.1351 - val_sparse_categorical_accuracy: 0.9556 - learning_rate: 5.0000e-04</span></span>
 <span><span class="co">## Epoch 141/500</span></span>
-<span><span class="co">## 90/90 - 0s - 2ms/step - loss: 0.0701 - sparse_categorical_accuracy: 0.9753 - val_loss: 0.1829 - val_sparse_categorical_accuracy: 0.9362 - learning_rate: 5.0000e-04</span></span>
+<span><span class="co">## 90/90 - 0s - 1ms/step - loss: 0.0712 - sparse_categorical_accuracy: 0.9764 - val_loss: 0.1121 - val_sparse_categorical_accuracy: 0.9612 - learning_rate: 5.0000e-04</span></span>
 <span><span class="co">## Epoch 142/500</span></span>
-<span><span class="co">## 90/90 - 0s - 1ms/step - loss: 0.0669 - sparse_categorical_accuracy: 0.9792 - val_loss: 0.1132 - val_sparse_categorical_accuracy: 0.9598 - learning_rate: 5.0000e-04</span></span>
+<span><span class="co">## 90/90 - 0s - 1ms/step - loss: 0.0692 - sparse_categorical_accuracy: 0.9778 - val_loss: 0.1139 - val_sparse_categorical_accuracy: 0.9639 - learning_rate: 5.0000e-04</span></span>
 <span><span class="co">## Epoch 143/500</span></span>
-<span><span class="co">## 90/90 - 0s - 1ms/step - loss: 0.0751 - sparse_categorical_accuracy: 0.9767 - val_loss: 0.1953 - val_sparse_categorical_accuracy: 0.9223 - learning_rate: 5.0000e-04</span></span>
+<span><span class="co">## 90/90 - 0s - 1ms/step - loss: 0.0781 - sparse_categorical_accuracy: 0.9736 - val_loss: 0.1202 - val_sparse_categorical_accuracy: 0.9515 - learning_rate: 5.0000e-04</span></span>
 <span><span class="co">## Epoch 144/500</span></span>
-<span><span class="co">## 90/90 - 0s - 1ms/step - loss: 0.0660 - sparse_categorical_accuracy: 0.9792 - val_loss: 0.1440 - val_sparse_categorical_accuracy: 0.9404 - learning_rate: 5.0000e-04</span></span>
+<span><span class="co">## 90/90 - 0s - 1ms/step - loss: 0.0699 - sparse_categorical_accuracy: 0.9781 - val_loss: 0.1131 - val_sparse_categorical_accuracy: 0.9501 - learning_rate: 5.0000e-04</span></span>
 <span><span class="co">## Epoch 145/500</span></span>
-<span><span class="co">## 90/90 - 0s - 2ms/step - loss: 0.0673 - sparse_categorical_accuracy: 0.9764 - val_loss: 0.6545 - val_sparse_categorical_accuracy: 0.8100 - learning_rate: 5.0000e-04</span></span>
+<span><span class="co">## 90/90 - 0s - 1ms/step - loss: 0.0712 - sparse_categorical_accuracy: 0.9771 - val_loss: 0.1097 - val_sparse_categorical_accuracy: 0.9542 - learning_rate: 5.0000e-04</span></span>
 <span><span class="co">## Epoch 146/500</span></span>
-<span><span class="co">## 90/90 - 0s - 2ms/step - loss: 0.0708 - sparse_categorical_accuracy: 0.9757 - val_loss: 0.1321 - val_sparse_categorical_accuracy: 0.9542 - learning_rate: 5.0000e-04</span></span>
+<span><span class="co">## 90/90 - 0s - 1ms/step - loss: 0.0730 - sparse_categorical_accuracy: 0.9736 - val_loss: 0.1421 - val_sparse_categorical_accuracy: 0.9445 - learning_rate: 5.0000e-04</span></span>
 <span><span class="co">## Epoch 147/500</span></span>
-<span><span class="co">## 90/90 - 0s - 2ms/step - loss: 0.0733 - sparse_categorical_accuracy: 0.9764 - val_loss: 0.1479 - val_sparse_categorical_accuracy: 0.9487 - learning_rate: 5.0000e-04</span></span>
+<span><span class="co">## 90/90 - 0s - 1ms/step - loss: 0.0757 - sparse_categorical_accuracy: 0.9740 - val_loss: 0.3533 - val_sparse_categorical_accuracy: 0.8752 - learning_rate: 5.0000e-04</span></span>
 <span><span class="co">## Epoch 148/500</span></span>
-<span><span class="co">## 90/90 - 0s - 2ms/step - loss: 0.0797 - sparse_categorical_accuracy: 0.9715 - val_loss: 0.3700 - val_sparse_categorical_accuracy: 0.8752 - learning_rate: 5.0000e-04</span></span>
+<span><span class="co">## 90/90 - 0s - 1ms/step - loss: 0.0818 - sparse_categorical_accuracy: 0.9701 - val_loss: 0.7787 - val_sparse_categorical_accuracy: 0.7836 - learning_rate: 5.0000e-04</span></span>
 <span><span class="co">## Epoch 149/500</span></span>
-<span><span class="co">## 90/90 - 0s - 2ms/step - loss: 0.0709 - sparse_categorical_accuracy: 0.9760 - val_loss: 0.1170 - val_sparse_categorical_accuracy: 0.9584 - learning_rate: 5.0000e-04</span></span>
+<span><span class="co">## 90/90 - 0s - 1ms/step - loss: 0.0717 - sparse_categorical_accuracy: 0.9771 - val_loss: 0.1214 - val_sparse_categorical_accuracy: 0.9515 - learning_rate: 5.0000e-04</span></span>
 <span><span class="co">## Epoch 150/500</span></span>
-<span><span class="co">## 90/90 - 0s - 2ms/step - loss: 0.0761 - sparse_categorical_accuracy: 0.9726 - val_loss: 0.1173 - val_sparse_categorical_accuracy: 0.9653 - learning_rate: 5.0000e-04</span></span>
+<span><span class="co">## 90/90 - 0s - 1ms/step - loss: 0.0736 - sparse_categorical_accuracy: 0.9757 - val_loss: 0.1445 - val_sparse_categorical_accuracy: 0.9362 - learning_rate: 5.0000e-04</span></span>
 <span><span class="co">## Epoch 151/500</span></span>
-<span><span class="co">## 90/90 - 0s - 3ms/step - loss: 0.0728 - sparse_categorical_accuracy: 0.9719 - val_loss: 0.1596 - val_sparse_categorical_accuracy: 0.9334 - learning_rate: 5.0000e-04</span></span>
+<span><span class="co">## 90/90 - 0s - 1ms/step - loss: 0.0725 - sparse_categorical_accuracy: 0.9729 - val_loss: 0.1175 - val_sparse_categorical_accuracy: 0.9528 - learning_rate: 5.0000e-04</span></span>
 <span><span class="co">## Epoch 152/500</span></span>
-<span><span class="co">## 90/90 - 0s - 2ms/step - loss: 0.0727 - sparse_categorical_accuracy: 0.9722 - val_loss: 0.1086 - val_sparse_categorical_accuracy: 0.9528 - learning_rate: 5.0000e-04</span></span>
+<span><span class="co">## 90/90 - 0s - 1ms/step - loss: 0.0726 - sparse_categorical_accuracy: 0.9750 - val_loss: 0.1329 - val_sparse_categorical_accuracy: 0.9515 - learning_rate: 5.0000e-04</span></span>
 <span><span class="co">## Epoch 153/500</span></span>
-<span><span class="co">## 90/90 - 0s - 2ms/step - loss: 0.0704 - sparse_categorical_accuracy: 0.9771 - val_loss: 0.1305 - val_sparse_categorical_accuracy: 0.9487 - learning_rate: 5.0000e-04</span></span>
+<span><span class="co">## 90/90 - 0s - 1ms/step - loss: 0.0727 - sparse_categorical_accuracy: 0.9757 - val_loss: 0.1806 - val_sparse_categorical_accuracy: 0.9293 - learning_rate: 5.0000e-04</span></span>
 <span><span class="co">## Epoch 154/500</span></span>
-<span><span class="co">## 90/90 - 0s - 2ms/step - loss: 0.0667 - sparse_categorical_accuracy: 0.9781 - val_loss: 0.1788 - val_sparse_categorical_accuracy: 0.9279 - learning_rate: 5.0000e-04</span></span>
+<span><span class="co">## 90/90 - 0s - 1ms/step - loss: 0.0684 - sparse_categorical_accuracy: 0.9778 - val_loss: 0.3892 - val_sparse_categorical_accuracy: 0.8682 - learning_rate: 5.0000e-04</span></span>
 <span><span class="co">## Epoch 155/500</span></span>
-<span><span class="co">## 90/90 - 0s - 2ms/step - loss: 0.0702 - sparse_categorical_accuracy: 0.9764 - val_loss: 0.1109 - val_sparse_categorical_accuracy: 0.9501 - learning_rate: 5.0000e-04</span></span>
+<span><span class="co">## 90/90 - 0s - 1ms/step - loss: 0.0734 - sparse_categorical_accuracy: 0.9753 - val_loss: 0.1122 - val_sparse_categorical_accuracy: 0.9542 - learning_rate: 5.0000e-04</span></span>
 <span><span class="co">## Epoch 156/500</span></span>
-<span><span class="co">## 90/90 - 0s - 1ms/step - loss: 0.0741 - sparse_categorical_accuracy: 0.9764 - val_loss: 0.1325 - val_sparse_categorical_accuracy: 0.9431 - learning_rate: 5.0000e-04</span></span>
+<span><span class="co">## 90/90 - 0s - 1ms/step - loss: 0.0758 - sparse_categorical_accuracy: 0.9774 - val_loss: 0.1077 - val_sparse_categorical_accuracy: 0.9653 - learning_rate: 5.0000e-04</span></span>
 <span><span class="co">## Epoch 157/500</span></span>
-<span><span class="co">## 90/90 - 0s - 1ms/step - loss: 0.0750 - sparse_categorical_accuracy: 0.9722 - val_loss: 0.1820 - val_sparse_categorical_accuracy: 0.9348 - learning_rate: 5.0000e-04</span></span>
+<span><span class="co">## 90/90 - 0s - 1ms/step - loss: 0.0734 - sparse_categorical_accuracy: 0.9747 - val_loss: 0.1594 - val_sparse_categorical_accuracy: 0.9376 - learning_rate: 5.0000e-04</span></span>
 <span><span class="co">## Epoch 158/500</span></span>
-<span><span class="co">## 90/90 - 0s - 1ms/step - loss: 0.0714 - sparse_categorical_accuracy: 0.9771 - val_loss: 0.1982 - val_sparse_categorical_accuracy: 0.9182 - learning_rate: 5.0000e-04</span></span>
+<span><span class="co">## 90/90 - 0s - 1ms/step - loss: 0.0753 - sparse_categorical_accuracy: 0.9750 - val_loss: 0.1992 - val_sparse_categorical_accuracy: 0.9168 - learning_rate: 5.0000e-04</span></span>
 <span><span class="co">## Epoch 159/500</span></span>
-<span><span class="co">## 90/90 - 0s - 2ms/step - loss: 0.0732 - sparse_categorical_accuracy: 0.9750 - val_loss: 0.2841 - val_sparse_categorical_accuracy: 0.8988 - learning_rate: 5.0000e-04</span></span>
+<span><span class="co">## 90/90 - 0s - 1ms/step - loss: 0.0740 - sparse_categorical_accuracy: 0.9740 - val_loss: 0.1483 - val_sparse_categorical_accuracy: 0.9362 - learning_rate: 5.0000e-04</span></span>
 <span><span class="co">## Epoch 160/500</span></span>
-<span><span class="co">## 90/90 - 0s - 1ms/step - loss: 0.0679 - sparse_categorical_accuracy: 0.9760 - val_loss: 0.1040 - val_sparse_categorical_accuracy: 0.9639 - learning_rate: 2.5000e-04</span></span>
+<span><span class="co">## 90/90 - 0s - 2ms/step - loss: 0.0677 - sparse_categorical_accuracy: 0.9778 - val_loss: 0.1041 - val_sparse_categorical_accuracy: 0.9653 - learning_rate: 2.5000e-04</span></span>
 <span><span class="co">## Epoch 161/500</span></span>
-<span><span class="co">## 90/90 - 0s - 2ms/step - loss: 0.0649 - sparse_categorical_accuracy: 0.9785 - val_loss: 0.1281 - val_sparse_categorical_accuracy: 0.9459 - learning_rate: 2.5000e-04</span></span>
+<span><span class="co">## 90/90 - 0s - 1ms/step - loss: 0.0686 - sparse_categorical_accuracy: 0.9778 - val_loss: 0.1327 - val_sparse_categorical_accuracy: 0.9515 - learning_rate: 2.5000e-04</span></span>
 <span><span class="co">## Epoch 162/500</span></span>
-<span><span class="co">## 90/90 - 0s - 2ms/step - loss: 0.0623 - sparse_categorical_accuracy: 0.9781 - val_loss: 0.1535 - val_sparse_categorical_accuracy: 0.9431 - learning_rate: 2.5000e-04</span></span>
+<span><span class="co">## 90/90 - 0s - 1ms/step - loss: 0.0634 - sparse_categorical_accuracy: 0.9792 - val_loss: 0.1059 - val_sparse_categorical_accuracy: 0.9667 - learning_rate: 2.5000e-04</span></span>
 <span><span class="co">## Epoch 163/500</span></span>
-<span><span class="co">## 90/90 - 0s - 2ms/step - loss: 0.0588 - sparse_categorical_accuracy: 0.9813 - val_loss: 0.1049 - val_sparse_categorical_accuracy: 0.9667 - learning_rate: 2.5000e-04</span></span>
+<span><span class="co">## 90/90 - 0s - 1ms/step - loss: 0.0595 - sparse_categorical_accuracy: 0.9792 - val_loss: 0.1236 - val_sparse_categorical_accuracy: 0.9612 - learning_rate: 2.5000e-04</span></span>
 <span><span class="co">## Epoch 164/500</span></span>
-<span><span class="co">## 90/90 - 0s - 2ms/step - loss: 0.0615 - sparse_categorical_accuracy: 0.9816 - val_loss: 0.1094 - val_sparse_categorical_accuracy: 0.9515 - learning_rate: 2.5000e-04</span></span>
+<span><span class="co">## 90/90 - 0s - 1ms/step - loss: 0.0637 - sparse_categorical_accuracy: 0.9806 - val_loss: 0.1185 - val_sparse_categorical_accuracy: 0.9626 - learning_rate: 2.5000e-04</span></span>
 <span><span class="co">## Epoch 165/500</span></span>
-<span><span class="co">## 90/90 - 0s - 2ms/step - loss: 0.0594 - sparse_categorical_accuracy: 0.9809 - val_loss: 0.1116 - val_sparse_categorical_accuracy: 0.9515 - learning_rate: 2.5000e-04</span></span>
+<span><span class="co">## 90/90 - 0s - 1ms/step - loss: 0.0607 - sparse_categorical_accuracy: 0.9816 - val_loss: 0.1480 - val_sparse_categorical_accuracy: 0.9473 - learning_rate: 2.5000e-04</span></span>
 <span><span class="co">## Epoch 166/500</span></span>
-<span><span class="co">## 90/90 - 0s - 1ms/step - loss: 0.0586 - sparse_categorical_accuracy: 0.9806 - val_loss: 0.1275 - val_sparse_categorical_accuracy: 0.9501 - learning_rate: 2.5000e-04</span></span>
+<span><span class="co">## 90/90 - 0s - 1ms/step - loss: 0.0594 - sparse_categorical_accuracy: 0.9799 - val_loss: 0.1044 - val_sparse_categorical_accuracy: 0.9653 - learning_rate: 2.5000e-04</span></span>
 <span><span class="co">## Epoch 167/500</span></span>
-<span><span class="co">## 90/90 - 0s - 1ms/step - loss: 0.0605 - sparse_categorical_accuracy: 0.9795 - val_loss: 0.1516 - val_sparse_categorical_accuracy: 0.9417 - learning_rate: 2.5000e-04</span></span>
+<span><span class="co">## 90/90 - 0s - 1ms/step - loss: 0.0629 - sparse_categorical_accuracy: 0.9788 - val_loss: 0.1343 - val_sparse_categorical_accuracy: 0.9445 - learning_rate: 2.5000e-04</span></span>
 <span><span class="co">## Epoch 168/500</span></span>
-<span><span class="co">## 90/90 - 0s - 1ms/step - loss: 0.0696 - sparse_categorical_accuracy: 0.9785 - val_loss: 0.1038 - val_sparse_categorical_accuracy: 0.9612 - learning_rate: 2.5000e-04</span></span>
+<span><span class="co">## 90/90 - 0s - 1ms/step - loss: 0.0740 - sparse_categorical_accuracy: 0.9750 - val_loss: 0.1081 - val_sparse_categorical_accuracy: 0.9667 - learning_rate: 2.5000e-04</span></span>
 <span><span class="co">## Epoch 169/500</span></span>
-<span><span class="co">## 90/90 - 0s - 1ms/step - loss: 0.0612 - sparse_categorical_accuracy: 0.9799 - val_loss: 0.1354 - val_sparse_categorical_accuracy: 0.9598 - learning_rate: 2.5000e-04</span></span>
+<span><span class="co">## 90/90 - 0s - 1ms/step - loss: 0.0649 - sparse_categorical_accuracy: 0.9799 - val_loss: 0.1136 - val_sparse_categorical_accuracy: 0.9681 - learning_rate: 2.5000e-04</span></span>
 <span><span class="co">## Epoch 170/500</span></span>
-<span><span class="co">## 90/90 - 0s - 1ms/step - loss: 0.0626 - sparse_categorical_accuracy: 0.9802 - val_loss: 0.1219 - val_sparse_categorical_accuracy: 0.9598 - learning_rate: 2.5000e-04</span></span>
+<span><span class="co">## 90/90 - 0s - 2ms/step - loss: 0.0649 - sparse_categorical_accuracy: 0.9823 - val_loss: 0.1037 - val_sparse_categorical_accuracy: 0.9695 - learning_rate: 2.5000e-04</span></span>
 <span><span class="co">## Epoch 171/500</span></span>
-<span><span class="co">## 90/90 - 0s - 1ms/step - loss: 0.0582 - sparse_categorical_accuracy: 0.9816 - val_loss: 0.1627 - val_sparse_categorical_accuracy: 0.9417 - learning_rate: 2.5000e-04</span></span>
+<span><span class="co">## 90/90 - 0s - 1ms/step - loss: 0.0609 - sparse_categorical_accuracy: 0.9799 - val_loss: 0.1239 - val_sparse_categorical_accuracy: 0.9653 - learning_rate: 2.5000e-04</span></span>
 <span><span class="co">## Epoch 172/500</span></span>
-<span><span class="co">## 90/90 - 0s - 2ms/step - loss: 0.0601 - sparse_categorical_accuracy: 0.9795 - val_loss: 0.1023 - val_sparse_categorical_accuracy: 0.9723 - learning_rate: 2.5000e-04</span></span>
+<span><span class="co">## 90/90 - 0s - 1ms/step - loss: 0.0636 - sparse_categorical_accuracy: 0.9795 - val_loss: 0.1037 - val_sparse_categorical_accuracy: 0.9528 - learning_rate: 2.5000e-04</span></span>
 <span><span class="co">## Epoch 173/500</span></span>
-<span><span class="co">## 90/90 - 0s - 2ms/step - loss: 0.0587 - sparse_categorical_accuracy: 0.9813 - val_loss: 0.1001 - val_sparse_categorical_accuracy: 0.9639 - learning_rate: 2.5000e-04</span></span>
+<span><span class="co">## 90/90 - 0s - 1ms/step - loss: 0.0608 - sparse_categorical_accuracy: 0.9816 - val_loss: 0.1296 - val_sparse_categorical_accuracy: 0.9626 - learning_rate: 2.5000e-04</span></span>
 <span><span class="co">## Epoch 174/500</span></span>
-<span><span class="co">## 90/90 - 0s - 2ms/step - loss: 0.0611 - sparse_categorical_accuracy: 0.9806 - val_loss: 0.1214 - val_sparse_categorical_accuracy: 0.9487 - learning_rate: 2.5000e-04</span></span>
+<span><span class="co">## 90/90 - 0s - 2ms/step - loss: 0.0624 - sparse_categorical_accuracy: 0.9792 - val_loss: 0.0981 - val_sparse_categorical_accuracy: 0.9639 - learning_rate: 2.5000e-04</span></span>
 <span><span class="co">## Epoch 175/500</span></span>
-<span><span class="co">## 90/90 - 0s - 1ms/step - loss: 0.0630 - sparse_categorical_accuracy: 0.9788 - val_loss: 0.1159 - val_sparse_categorical_accuracy: 0.9556 - learning_rate: 2.5000e-04</span></span>
+<span><span class="co">## 90/90 - 0s - 1ms/step - loss: 0.0666 - sparse_categorical_accuracy: 0.9785 - val_loss: 0.1135 - val_sparse_categorical_accuracy: 0.9487 - learning_rate: 2.5000e-04</span></span>
 <span><span class="co">## Epoch 176/500</span></span>
-<span><span class="co">## 90/90 - 0s - 1ms/step - loss: 0.0593 - sparse_categorical_accuracy: 0.9792 - val_loss: 0.1535 - val_sparse_categorical_accuracy: 0.9473 - learning_rate: 2.5000e-04</span></span>
+<span><span class="co">## 90/90 - 0s - 1ms/step - loss: 0.0625 - sparse_categorical_accuracy: 0.9771 - val_loss: 0.1024 - val_sparse_categorical_accuracy: 0.9584 - learning_rate: 2.5000e-04</span></span>
 <span><span class="co">## Epoch 177/500</span></span>
-<span><span class="co">## 90/90 - 0s - 2ms/step - loss: 0.0579 - sparse_categorical_accuracy: 0.9844 - val_loss: 0.1511 - val_sparse_categorical_accuracy: 0.9473 - learning_rate: 2.5000e-04</span></span>
+<span><span class="co">## 90/90 - 0s - 1ms/step - loss: 0.0616 - sparse_categorical_accuracy: 0.9806 - val_loss: 0.1036 - val_sparse_categorical_accuracy: 0.9570 - learning_rate: 2.5000e-04</span></span>
 <span><span class="co">## Epoch 178/500</span></span>
-<span><span class="co">## 90/90 - 0s - 2ms/step - loss: 0.0552 - sparse_categorical_accuracy: 0.9826 - val_loss: 0.2204 - val_sparse_categorical_accuracy: 0.9140 - learning_rate: 2.5000e-04</span></span>
+<span><span class="co">## 90/90 - 0s - 1ms/step - loss: 0.0590 - sparse_categorical_accuracy: 0.9823 - val_loss: 0.2086 - val_sparse_categorical_accuracy: 0.9154 - learning_rate: 2.5000e-04</span></span>
 <span><span class="co">## Epoch 179/500</span></span>
-<span><span class="co">## 90/90 - 0s - 2ms/step - loss: 0.0572 - sparse_categorical_accuracy: 0.9799 - val_loss: 0.1048 - val_sparse_categorical_accuracy: 0.9681 - learning_rate: 2.5000e-04</span></span>
+<span><span class="co">## 90/90 - 0s - 1ms/step - loss: 0.0602 - sparse_categorical_accuracy: 0.9788 - val_loss: 0.1181 - val_sparse_categorical_accuracy: 0.9653 - learning_rate: 2.5000e-04</span></span>
 <span><span class="co">## Epoch 180/500</span></span>
-<span><span class="co">## 90/90 - 0s - 2ms/step - loss: 0.0554 - sparse_categorical_accuracy: 0.9823 - val_loss: 0.1197 - val_sparse_categorical_accuracy: 0.9695 - learning_rate: 1.2500e-04</span></span>
+<span><span class="co">## 90/90 - 0s - 1ms/step - loss: 0.0615 - sparse_categorical_accuracy: 0.9816 - val_loss: 0.1587 - val_sparse_categorical_accuracy: 0.9459 - learning_rate: 2.5000e-04</span></span>
 <span><span class="co">## Epoch 181/500</span></span>
-<span><span class="co">## 90/90 - 0s - 1ms/step - loss: 0.0575 - sparse_categorical_accuracy: 0.9837 - val_loss: 0.1206 - val_sparse_categorical_accuracy: 0.9653 - learning_rate: 1.2500e-04</span></span>
+<span><span class="co">## 90/90 - 0s - 1ms/step - loss: 0.0647 - sparse_categorical_accuracy: 0.9813 - val_loss: 0.1836 - val_sparse_categorical_accuracy: 0.9362 - learning_rate: 2.5000e-04</span></span>
 <span><span class="co">## Epoch 182/500</span></span>
-<span><span class="co">## 90/90 - 0s - 1ms/step - loss: 0.0537 - sparse_categorical_accuracy: 0.9833 - val_loss: 0.1074 - val_sparse_categorical_accuracy: 0.9667 - learning_rate: 1.2500e-04</span></span>
+<span><span class="co">## 90/90 - 0s - 1ms/step - loss: 0.0614 - sparse_categorical_accuracy: 0.9813 - val_loss: 0.2156 - val_sparse_categorical_accuracy: 0.9223 - learning_rate: 2.5000e-04</span></span>
 <span><span class="co">## Epoch 183/500</span></span>
-<span><span class="co">## 90/90 - 0s - 2ms/step - loss: 0.0571 - sparse_categorical_accuracy: 0.9809 - val_loss: 0.1052 - val_sparse_categorical_accuracy: 0.9695 - learning_rate: 1.2500e-04</span></span>
+<span><span class="co">## 90/90 - 0s - 1ms/step - loss: 0.0637 - sparse_categorical_accuracy: 0.9778 - val_loss: 0.1028 - val_sparse_categorical_accuracy: 0.9584 - learning_rate: 2.5000e-04</span></span>
 <span><span class="co">## Epoch 184/500</span></span>
-<span><span class="co">## 90/90 - 0s - 2ms/step - loss: 0.0528 - sparse_categorical_accuracy: 0.9830 - val_loss: 0.1022 - val_sparse_categorical_accuracy: 0.9709 - learning_rate: 1.2500e-04</span></span>
+<span><span class="co">## 90/90 - 0s - 1ms/step - loss: 0.0572 - sparse_categorical_accuracy: 0.9823 - val_loss: 0.1347 - val_sparse_categorical_accuracy: 0.9431 - learning_rate: 2.5000e-04</span></span>
 <span><span class="co">## Epoch 185/500</span></span>
-<span><span class="co">## 90/90 - 0s - 1ms/step - loss: 0.0550 - sparse_categorical_accuracy: 0.9830 - val_loss: 0.1042 - val_sparse_categorical_accuracy: 0.9653 - learning_rate: 1.2500e-04</span></span>
+<span><span class="co">## 90/90 - 0s - 1ms/step - loss: 0.0593 - sparse_categorical_accuracy: 0.9826 - val_loss: 0.1402 - val_sparse_categorical_accuracy: 0.9459 - learning_rate: 2.5000e-04</span></span>
 <span><span class="co">## Epoch 186/500</span></span>
-<span><span class="co">## 90/90 - 0s - 2ms/step - loss: 0.0605 - sparse_categorical_accuracy: 0.9792 - val_loss: 0.1048 - val_sparse_categorical_accuracy: 0.9695 - learning_rate: 1.2500e-04</span></span>
+<span><span class="co">## 90/90 - 0s - 1ms/step - loss: 0.0659 - sparse_categorical_accuracy: 0.9774 - val_loss: 0.1873 - val_sparse_categorical_accuracy: 0.9293 - learning_rate: 2.5000e-04</span></span>
 <span><span class="co">## Epoch 187/500</span></span>
-<span><span class="co">## 90/90 - 0s - 1ms/step - loss: 0.0539 - sparse_categorical_accuracy: 0.9847 - val_loss: 0.1090 - val_sparse_categorical_accuracy: 0.9639 - learning_rate: 1.2500e-04</span></span>
+<span><span class="co">## 90/90 - 0s - 1ms/step - loss: 0.0606 - sparse_categorical_accuracy: 0.9799 - val_loss: 0.1210 - val_sparse_categorical_accuracy: 0.9570 - learning_rate: 2.5000e-04</span></span>
 <span><span class="co">## Epoch 188/500</span></span>
-<span><span class="co">## 90/90 - 0s - 2ms/step - loss: 0.0540 - sparse_categorical_accuracy: 0.9840 - val_loss: 0.1189 - val_sparse_categorical_accuracy: 0.9653 - learning_rate: 1.2500e-04</span></span>
+<span><span class="co">## 90/90 - 0s - 1ms/step - loss: 0.0610 - sparse_categorical_accuracy: 0.9813 - val_loss: 0.1250 - val_sparse_categorical_accuracy: 0.9473 - learning_rate: 2.5000e-04</span></span>
 <span><span class="co">## Epoch 189/500</span></span>
-<span><span class="co">## 90/90 - 0s - 2ms/step - loss: 0.0514 - sparse_categorical_accuracy: 0.9854 - val_loss: 0.1599 - val_sparse_categorical_accuracy: 0.9431 - learning_rate: 1.2500e-04</span></span>
-<span><span class="co">## Epoch 189: early stopping</span></span></code></pre>
+<span><span class="co">## 90/90 - 0s - 1ms/step - loss: 0.0570 - sparse_categorical_accuracy: 0.9830 - val_loss: 0.1227 - val_sparse_categorical_accuracy: 0.9667 - learning_rate: 2.5000e-04</span></span>
+<span><span class="co">## Epoch 190/500</span></span>
+<span><span class="co">## 90/90 - 0s - 1ms/step - loss: 0.0678 - sparse_categorical_accuracy: 0.9771 - val_loss: 0.1367 - val_sparse_categorical_accuracy: 0.9584 - learning_rate: 2.5000e-04</span></span>
+<span><span class="co">## Epoch 191/500</span></span>
+<span><span class="co">## 90/90 - 0s - 1ms/step - loss: 0.0580 - sparse_categorical_accuracy: 0.9802 - val_loss: 0.1005 - val_sparse_categorical_accuracy: 0.9653 - learning_rate: 2.5000e-04</span></span>
+<span><span class="co">## Epoch 192/500</span></span>
+<span><span class="co">## 90/90 - 0s - 1ms/step - loss: 0.0651 - sparse_categorical_accuracy: 0.9778 - val_loss: 0.1170 - val_sparse_categorical_accuracy: 0.9639 - learning_rate: 2.5000e-04</span></span>
+<span><span class="co">## Epoch 193/500</span></span>
+<span><span class="co">## 90/90 - 0s - 1ms/step - loss: 0.0593 - sparse_categorical_accuracy: 0.9823 - val_loss: 0.1235 - val_sparse_categorical_accuracy: 0.9473 - learning_rate: 2.5000e-04</span></span>
+<span><span class="co">## Epoch 194/500</span></span>
+<span><span class="co">## 90/90 - 0s - 1ms/step - loss: 0.0588 - sparse_categorical_accuracy: 0.9792 - val_loss: 0.1089 - val_sparse_categorical_accuracy: 0.9570 - learning_rate: 2.5000e-04</span></span>
+<span><span class="co">## Epoch 195/500</span></span>
+<span><span class="co">## 90/90 - 0s - 2ms/step - loss: 0.0564 - sparse_categorical_accuracy: 0.9833 - val_loss: 0.0979 - val_sparse_categorical_accuracy: 0.9681 - learning_rate: 1.2500e-04</span></span>
+<span><span class="co">## Epoch 196/500</span></span>
+<span><span class="co">## 90/90 - 0s - 1ms/step - loss: 0.0584 - sparse_categorical_accuracy: 0.9788 - val_loss: 0.1169 - val_sparse_categorical_accuracy: 0.9584 - learning_rate: 1.2500e-04</span></span>
+<span><span class="co">## Epoch 197/500</span></span>
+<span><span class="co">## 90/90 - 0s - 1ms/step - loss: 0.0553 - sparse_categorical_accuracy: 0.9802 - val_loss: 0.1380 - val_sparse_categorical_accuracy: 0.9556 - learning_rate: 1.2500e-04</span></span>
+<span><span class="co">## Epoch 198/500</span></span>
+<span><span class="co">## 90/90 - 0s - 1ms/step - loss: 0.0541 - sparse_categorical_accuracy: 0.9868 - val_loss: 0.1261 - val_sparse_categorical_accuracy: 0.9556 - learning_rate: 1.2500e-04</span></span>
+<span><span class="co">## Epoch 199/500</span></span>
+<span><span class="co">## 90/90 - 0s - 1ms/step - loss: 0.0575 - sparse_categorical_accuracy: 0.9806 - val_loss: 0.1010 - val_sparse_categorical_accuracy: 0.9681 - learning_rate: 1.2500e-04</span></span>
+<span><span class="co">## Epoch 200/500</span></span>
+<span><span class="co">## 90/90 - 0s - 1ms/step - loss: 0.0568 - sparse_categorical_accuracy: 0.9816 - val_loss: 0.0996 - val_sparse_categorical_accuracy: 0.9612 - learning_rate: 1.2500e-04</span></span>
+<span><span class="co">## Epoch 201/500</span></span>
+<span><span class="co">## 90/90 - 0s - 1ms/step - loss: 0.0547 - sparse_categorical_accuracy: 0.9819 - val_loss: 0.1096 - val_sparse_categorical_accuracy: 0.9653 - learning_rate: 1.2500e-04</span></span>
+<span><span class="co">## Epoch 202/500</span></span>
+<span><span class="co">## 90/90 - 0s - 1ms/step - loss: 0.0547 - sparse_categorical_accuracy: 0.9847 - val_loss: 0.1019 - val_sparse_categorical_accuracy: 0.9556 - learning_rate: 1.2500e-04</span></span>
+<span><span class="co">## Epoch 203/500</span></span>
+<span><span class="co">## 90/90 - 0s - 1ms/step - loss: 0.0562 - sparse_categorical_accuracy: 0.9816 - val_loss: 0.0985 - val_sparse_categorical_accuracy: 0.9639 - learning_rate: 1.2500e-04</span></span>
+<span><span class="co">## Epoch 204/500</span></span>
+<span><span class="co">## 90/90 - 0s - 2ms/step - loss: 0.0519 - sparse_categorical_accuracy: 0.9830 - val_loss: 0.0978 - val_sparse_categorical_accuracy: 0.9681 - learning_rate: 1.2500e-04</span></span>
+<span><span class="co">## Epoch 205/500</span></span>
+<span><span class="co">## 90/90 - 0s - 1ms/step - loss: 0.0558 - sparse_categorical_accuracy: 0.9802 - val_loss: 0.1035 - val_sparse_categorical_accuracy: 0.9667 - learning_rate: 1.2500e-04</span></span>
+<span><span class="co">## Epoch 206/500</span></span>
+<span><span class="co">## 90/90 - 0s - 1ms/step - loss: 0.0551 - sparse_categorical_accuracy: 0.9816 - val_loss: 0.1006 - val_sparse_categorical_accuracy: 0.9709 - learning_rate: 1.2500e-04</span></span>
+<span><span class="co">## Epoch 207/500</span></span>
+<span><span class="co">## 90/90 - 0s - 1ms/step - loss: 0.0593 - sparse_categorical_accuracy: 0.9816 - val_loss: 0.1023 - val_sparse_categorical_accuracy: 0.9556 - learning_rate: 1.2500e-04</span></span>
+<span><span class="co">## Epoch 208/500</span></span>
+<span><span class="co">## 90/90 - 0s - 2ms/step - loss: 0.0556 - sparse_categorical_accuracy: 0.9816 - val_loss: 0.1314 - val_sparse_categorical_accuracy: 0.9626 - learning_rate: 1.2500e-04</span></span>
+<span><span class="co">## Epoch 209/500</span></span>
+<span><span class="co">## 90/90 - 0s - 1ms/step - loss: 0.0539 - sparse_categorical_accuracy: 0.9809 - val_loss: 0.1529 - val_sparse_categorical_accuracy: 0.9487 - learning_rate: 1.2500e-04</span></span>
+<span><span class="co">## Epoch 210/500</span></span>
+<span><span class="co">## 90/90 - 0s - 1ms/step - loss: 0.0566 - sparse_categorical_accuracy: 0.9816 - val_loss: 0.0993 - val_sparse_categorical_accuracy: 0.9653 - learning_rate: 1.2500e-04</span></span>
+<span><span class="co">## Epoch 211/500</span></span>
+<span><span class="co">## 90/90 - 0s - 1ms/step - loss: 0.0552 - sparse_categorical_accuracy: 0.9837 - val_loss: 0.0984 - val_sparse_categorical_accuracy: 0.9639 - learning_rate: 1.2500e-04</span></span>
+<span><span class="co">## Epoch 212/500</span></span>
+<span><span class="co">## 90/90 - 0s - 1ms/step - loss: 0.0548 - sparse_categorical_accuracy: 0.9826 - val_loss: 0.0999 - val_sparse_categorical_accuracy: 0.9584 - learning_rate: 1.2500e-04</span></span>
+<span><span class="co">## Epoch 213/500</span></span>
+<span><span class="co">## 90/90 - 0s - 2ms/step - loss: 0.0573 - sparse_categorical_accuracy: 0.9840 - val_loss: 0.0968 - val_sparse_categorical_accuracy: 0.9667 - learning_rate: 1.2500e-04</span></span>
+<span><span class="co">## Epoch 214/500</span></span>
+<span><span class="co">## 90/90 - 0s - 1ms/step - loss: 0.0568 - sparse_categorical_accuracy: 0.9830 - val_loss: 0.1000 - val_sparse_categorical_accuracy: 0.9639 - learning_rate: 1.2500e-04</span></span>
+<span><span class="co">## Epoch 215/500</span></span>
+<span><span class="co">## 90/90 - 0s - 1ms/step - loss: 0.0534 - sparse_categorical_accuracy: 0.9816 - val_loss: 0.1031 - val_sparse_categorical_accuracy: 0.9667 - learning_rate: 1.2500e-04</span></span>
+<span><span class="co">## Epoch 216/500</span></span>
+<span><span class="co">## 90/90 - 0s - 1ms/step - loss: 0.0531 - sparse_categorical_accuracy: 0.9847 - val_loss: 0.1042 - val_sparse_categorical_accuracy: 0.9681 - learning_rate: 1.2500e-04</span></span>
+<span><span class="co">## Epoch 217/500</span></span>
+<span><span class="co">## 90/90 - 0s - 1ms/step - loss: 0.0532 - sparse_categorical_accuracy: 0.9826 - val_loss: 0.1096 - val_sparse_categorical_accuracy: 0.9695 - learning_rate: 1.2500e-04</span></span>
+<span><span class="co">## Epoch 218/500</span></span>
+<span><span class="co">## 90/90 - 0s - 1ms/step - loss: 0.0577 - sparse_categorical_accuracy: 0.9809 - val_loss: 0.1009 - val_sparse_categorical_accuracy: 0.9612 - learning_rate: 1.2500e-04</span></span>
+<span><span class="co">## Epoch 219/500</span></span>
+<span><span class="co">## 90/90 - 0s - 1ms/step - loss: 0.0539 - sparse_categorical_accuracy: 0.9823 - val_loss: 0.1072 - val_sparse_categorical_accuracy: 0.9695 - learning_rate: 1.2500e-04</span></span>
+<span><span class="co">## Epoch 220/500</span></span>
+<span><span class="co">## 90/90 - 0s - 1ms/step - loss: 0.0529 - sparse_categorical_accuracy: 0.9851 - val_loss: 0.1005 - val_sparse_categorical_accuracy: 0.9695 - learning_rate: 1.2500e-04</span></span>
+<span><span class="co">## Epoch 221/500</span></span>
+<span><span class="co">## 90/90 - 0s - 1ms/step - loss: 0.0554 - sparse_categorical_accuracy: 0.9799 - val_loss: 0.1036 - val_sparse_categorical_accuracy: 0.9695 - learning_rate: 1.2500e-04</span></span>
+<span><span class="co">## Epoch 222/500</span></span>
+<span><span class="co">## 90/90 - 0s - 1ms/step - loss: 0.0562 - sparse_categorical_accuracy: 0.9809 - val_loss: 0.1037 - val_sparse_categorical_accuracy: 0.9736 - learning_rate: 1.2500e-04</span></span>
+<span><span class="co">## Epoch 223/500</span></span>
+<span><span class="co">## 90/90 - 0s - 1ms/step - loss: 0.0544 - sparse_categorical_accuracy: 0.9823 - val_loss: 0.1005 - val_sparse_categorical_accuracy: 0.9736 - learning_rate: 1.2500e-04</span></span>
+<span><span class="co">## Epoch 224/500</span></span>
+<span><span class="co">## 90/90 - 0s - 1ms/step - loss: 0.0509 - sparse_categorical_accuracy: 0.9847 - val_loss: 0.0969 - val_sparse_categorical_accuracy: 0.9626 - learning_rate: 1.2500e-04</span></span>
+<span><span class="co">## Epoch 225/500</span></span>
+<span><span class="co">## 90/90 - 0s - 1ms/step - loss: 0.0569 - sparse_categorical_accuracy: 0.9823 - val_loss: 0.1261 - val_sparse_categorical_accuracy: 0.9612 - learning_rate: 1.2500e-04</span></span>
+<span><span class="co">## Epoch 226/500</span></span>
+<span><span class="co">## 90/90 - 0s - 1ms/step - loss: 0.0519 - sparse_categorical_accuracy: 0.9826 - val_loss: 0.1019 - val_sparse_categorical_accuracy: 0.9570 - learning_rate: 1.2500e-04</span></span>
+<span><span class="co">## Epoch 227/500</span></span>
+<span><span class="co">## 90/90 - 0s - 1ms/step - loss: 0.0563 - sparse_categorical_accuracy: 0.9819 - val_loss: 0.1037 - val_sparse_categorical_accuracy: 0.9584 - learning_rate: 1.2500e-04</span></span>
+<span><span class="co">## Epoch 228/500</span></span>
+<span><span class="co">## 90/90 - 0s - 1ms/step - loss: 0.0518 - sparse_categorical_accuracy: 0.9844 - val_loss: 0.1127 - val_sparse_categorical_accuracy: 0.9515 - learning_rate: 1.2500e-04</span></span>
+<span><span class="co">## Epoch 229/500</span></span>
+<span><span class="co">## 90/90 - 0s - 1ms/step - loss: 0.0521 - sparse_categorical_accuracy: 0.9865 - val_loss: 0.0973 - val_sparse_categorical_accuracy: 0.9639 - learning_rate: 1.2500e-04</span></span>
+<span><span class="co">## Epoch 230/500</span></span>
+<span><span class="co">## 90/90 - 0s - 1ms/step - loss: 0.0520 - sparse_categorical_accuracy: 0.9847 - val_loss: 0.0999 - val_sparse_categorical_accuracy: 0.9626 - learning_rate: 1.2500e-04</span></span>
+<span><span class="co">## Epoch 231/500</span></span>
+<span><span class="co">## 90/90 - 0s - 2ms/step - loss: 0.0569 - sparse_categorical_accuracy: 0.9813 - val_loss: 0.0958 - val_sparse_categorical_accuracy: 0.9626 - learning_rate: 1.2500e-04</span></span>
+<span><span class="co">## Epoch 232/500</span></span>
+<span><span class="co">## 90/90 - 0s - 1ms/step - loss: 0.0530 - sparse_categorical_accuracy: 0.9851 - val_loss: 0.1024 - val_sparse_categorical_accuracy: 0.9681 - learning_rate: 1.2500e-04</span></span>
+<span><span class="co">## Epoch 233/500</span></span>
+<span><span class="co">## 90/90 - 0s - 1ms/step - loss: 0.0523 - sparse_categorical_accuracy: 0.9844 - val_loss: 0.1019 - val_sparse_categorical_accuracy: 0.9709 - learning_rate: 1.2500e-04</span></span>
+<span><span class="co">## Epoch 234/500</span></span>
+<span><span class="co">## 90/90 - 0s - 1ms/step - loss: 0.0539 - sparse_categorical_accuracy: 0.9837 - val_loss: 0.1595 - val_sparse_categorical_accuracy: 0.9459 - learning_rate: 1.2500e-04</span></span>
+<span><span class="co">## Epoch 235/500</span></span>
+<span><span class="co">## 90/90 - 0s - 1ms/step - loss: 0.0543 - sparse_categorical_accuracy: 0.9802 - val_loss: 0.1028 - val_sparse_categorical_accuracy: 0.9653 - learning_rate: 1.2500e-04</span></span>
+<span><span class="co">## Epoch 236/500</span></span>
+<span><span class="co">## 90/90 - 0s - 1ms/step - loss: 0.0540 - sparse_categorical_accuracy: 0.9826 - val_loss: 0.1068 - val_sparse_categorical_accuracy: 0.9653 - learning_rate: 1.2500e-04</span></span>
+<span><span class="co">## Epoch 237/500</span></span>
+<span><span class="co">## 90/90 - 0s - 1ms/step - loss: 0.0543 - sparse_categorical_accuracy: 0.9809 - val_loss: 0.1528 - val_sparse_categorical_accuracy: 0.9459 - learning_rate: 1.2500e-04</span></span>
+<span><span class="co">## Epoch 238/500</span></span>
+<span><span class="co">## 90/90 - 0s - 1ms/step - loss: 0.0527 - sparse_categorical_accuracy: 0.9861 - val_loss: 0.0977 - val_sparse_categorical_accuracy: 0.9695 - learning_rate: 1.2500e-04</span></span>
+<span><span class="co">## Epoch 239/500</span></span>
+<span><span class="co">## 90/90 - 0s - 1ms/step - loss: 0.0537 - sparse_categorical_accuracy: 0.9813 - val_loss: 0.0980 - val_sparse_categorical_accuracy: 0.9598 - learning_rate: 1.2500e-04</span></span>
+<span><span class="co">## Epoch 240/500</span></span>
+<span><span class="co">## 90/90 - 0s - 1ms/step - loss: 0.0512 - sparse_categorical_accuracy: 0.9854 - val_loss: 0.1054 - val_sparse_categorical_accuracy: 0.9709 - learning_rate: 1.2500e-04</span></span>
+<span><span class="co">## Epoch 241/500</span></span>
+<span><span class="co">## 90/90 - 0s - 1ms/step - loss: 0.0541 - sparse_categorical_accuracy: 0.9837 - val_loss: 0.1325 - val_sparse_categorical_accuracy: 0.9487 - learning_rate: 1.2500e-04</span></span>
+<span><span class="co">## Epoch 242/500</span></span>
+<span><span class="co">## 90/90 - 0s - 1ms/step - loss: 0.0535 - sparse_categorical_accuracy: 0.9826 - val_loss: 0.1028 - val_sparse_categorical_accuracy: 0.9681 - learning_rate: 1.2500e-04</span></span>
+<span><span class="co">## Epoch 243/500</span></span>
+<span><span class="co">## 90/90 - 0s - 1ms/step - loss: 0.0555 - sparse_categorical_accuracy: 0.9823 - val_loss: 0.0991 - val_sparse_categorical_accuracy: 0.9626 - learning_rate: 1.2500e-04</span></span>
+<span><span class="co">## Epoch 244/500</span></span>
+<span><span class="co">## 90/90 - 0s - 1ms/step - loss: 0.0572 - sparse_categorical_accuracy: 0.9795 - val_loss: 0.0986 - val_sparse_categorical_accuracy: 0.9639 - learning_rate: 1.2500e-04</span></span>
+<span><span class="co">## Epoch 245/500</span></span>
+<span><span class="co">## 90/90 - 0s - 1ms/step - loss: 0.0520 - sparse_categorical_accuracy: 0.9833 - val_loss: 0.0980 - val_sparse_categorical_accuracy: 0.9626 - learning_rate: 1.2500e-04</span></span>
+<span><span class="co">## Epoch 246/500</span></span>
+<span><span class="co">## 90/90 - 0s - 1ms/step - loss: 0.0554 - sparse_categorical_accuracy: 0.9809 - val_loss: 0.0978 - val_sparse_categorical_accuracy: 0.9639 - learning_rate: 1.2500e-04</span></span>
+<span><span class="co">## Epoch 247/500</span></span>
+<span><span class="co">## 90/90 - 0s - 1ms/step - loss: 0.0498 - sparse_categorical_accuracy: 0.9837 - val_loss: 0.1015 - val_sparse_categorical_accuracy: 0.9681 - learning_rate: 1.2500e-04</span></span>
+<span><span class="co">## Epoch 248/500</span></span>
+<span><span class="co">## 90/90 - 0s - 1ms/step - loss: 0.0493 - sparse_categorical_accuracy: 0.9840 - val_loss: 0.0994 - val_sparse_categorical_accuracy: 0.9612 - learning_rate: 1.2500e-04</span></span>
+<span><span class="co">## Epoch 249/500</span></span>
+<span><span class="co">## 90/90 - 0s - 1ms/step - loss: 0.0543 - sparse_categorical_accuracy: 0.9847 - val_loss: 0.1054 - val_sparse_categorical_accuracy: 0.9695 - learning_rate: 1.2500e-04</span></span>
+<span><span class="co">## Epoch 250/500</span></span>
+<span><span class="co">## 90/90 - 0s - 1ms/step - loss: 0.0550 - sparse_categorical_accuracy: 0.9802 - val_loss: 0.0989 - val_sparse_categorical_accuracy: 0.9626 - learning_rate: 1.2500e-04</span></span>
+<span><span class="co">## Epoch 251/500</span></span>
+<span><span class="co">## 90/90 - 0s - 1ms/step - loss: 0.0539 - sparse_categorical_accuracy: 0.9819 - val_loss: 0.1018 - val_sparse_categorical_accuracy: 0.9667 - learning_rate: 1.2500e-04</span></span>
+<span><span class="co">## Epoch 252/500</span></span>
+<span><span class="co">## 90/90 - 0s - 1ms/step - loss: 0.0542 - sparse_categorical_accuracy: 0.9830 - val_loss: 0.0975 - val_sparse_categorical_accuracy: 0.9667 - learning_rate: 1.0000e-04</span></span>
+<span><span class="co">## Epoch 253/500</span></span>
+<span><span class="co">## 90/90 - 0s - 1ms/step - loss: 0.0538 - sparse_categorical_accuracy: 0.9830 - val_loss: 0.1065 - val_sparse_categorical_accuracy: 0.9723 - learning_rate: 1.0000e-04</span></span>
+<span><span class="co">## Epoch 254/500</span></span>
+<span><span class="co">## 90/90 - 0s - 1ms/step - loss: 0.0542 - sparse_categorical_accuracy: 0.9823 - val_loss: 0.1145 - val_sparse_categorical_accuracy: 0.9528 - learning_rate: 1.0000e-04</span></span>
+<span><span class="co">## Epoch 255/500</span></span>
+<span><span class="co">## 90/90 - 0s - 1ms/step - loss: 0.0506 - sparse_categorical_accuracy: 0.9819 - val_loss: 0.0976 - val_sparse_categorical_accuracy: 0.9681 - learning_rate: 1.0000e-04</span></span>
+<span><span class="co">## Epoch 256/500</span></span>
+<span><span class="co">## 90/90 - 0s - 1ms/step - loss: 0.0521 - sparse_categorical_accuracy: 0.9847 - val_loss: 0.1117 - val_sparse_categorical_accuracy: 0.9626 - learning_rate: 1.0000e-04</span></span>
+<span><span class="co">## Epoch 257/500</span></span>
+<span><span class="co">## 90/90 - 0s - 1ms/step - loss: 0.0504 - sparse_categorical_accuracy: 0.9854 - val_loss: 0.1201 - val_sparse_categorical_accuracy: 0.9626 - learning_rate: 1.0000e-04</span></span>
+<span><span class="co">## Epoch 258/500</span></span>
+<span><span class="co">## 90/90 - 0s - 1ms/step - loss: 0.0498 - sparse_categorical_accuracy: 0.9851 - val_loss: 0.1039 - val_sparse_categorical_accuracy: 0.9584 - learning_rate: 1.0000e-04</span></span>
+<span><span class="co">## Epoch 259/500</span></span>
+<span><span class="co">## 90/90 - 0s - 1ms/step - loss: 0.0479 - sparse_categorical_accuracy: 0.9868 - val_loss: 0.1055 - val_sparse_categorical_accuracy: 0.9695 - learning_rate: 1.0000e-04</span></span>
+<span><span class="co">## Epoch 260/500</span></span>
+<span><span class="co">## 90/90 - 0s - 1ms/step - loss: 0.0499 - sparse_categorical_accuracy: 0.9840 - val_loss: 0.1046 - val_sparse_categorical_accuracy: 0.9556 - learning_rate: 1.0000e-04</span></span>
+<span><span class="co">## Epoch 261/500</span></span>
+<span><span class="co">## 90/90 - 0s - 1ms/step - loss: 0.0529 - sparse_categorical_accuracy: 0.9823 - val_loss: 0.1022 - val_sparse_categorical_accuracy: 0.9667 - learning_rate: 1.0000e-04</span></span>
+<span><span class="co">## Epoch 262/500</span></span>
+<span><span class="co">## 90/90 - 0s - 1ms/step - loss: 0.0492 - sparse_categorical_accuracy: 0.9840 - val_loss: 0.1063 - val_sparse_categorical_accuracy: 0.9681 - learning_rate: 1.0000e-04</span></span>
+<span><span class="co">## Epoch 263/500</span></span>
+<span><span class="co">## 90/90 - 0s - 1ms/step - loss: 0.0500 - sparse_categorical_accuracy: 0.9833 - val_loss: 0.1006 - val_sparse_categorical_accuracy: 0.9626 - learning_rate: 1.0000e-04</span></span>
+<span><span class="co">## Epoch 264/500</span></span>
+<span><span class="co">## 90/90 - 0s - 1ms/step - loss: 0.0533 - sparse_categorical_accuracy: 0.9823 - val_loss: 0.1088 - val_sparse_categorical_accuracy: 0.9709 - learning_rate: 1.0000e-04</span></span>
+<span><span class="co">## Epoch 265/500</span></span>
+<span><span class="co">## 90/90 - 0s - 1ms/step - loss: 0.0549 - sparse_categorical_accuracy: 0.9802 - val_loss: 0.1010 - val_sparse_categorical_accuracy: 0.9709 - learning_rate: 1.0000e-04</span></span>
+<span><span class="co">## Epoch 266/500</span></span>
+<span><span class="co">## 90/90 - 0s - 1ms/step - loss: 0.0512 - sparse_categorical_accuracy: 0.9830 - val_loss: 0.1016 - val_sparse_categorical_accuracy: 0.9570 - learning_rate: 1.0000e-04</span></span>
+<span><span class="co">## Epoch 267/500</span></span>
+<span><span class="co">## 90/90 - 0s - 1ms/step - loss: 0.0496 - sparse_categorical_accuracy: 0.9847 - val_loss: 0.1012 - val_sparse_categorical_accuracy: 0.9570 - learning_rate: 1.0000e-04</span></span>
+<span><span class="co">## Epoch 268/500</span></span>
+<span><span class="co">## 90/90 - 0s - 1ms/step - loss: 0.0477 - sparse_categorical_accuracy: 0.9872 - val_loss: 0.1016 - val_sparse_categorical_accuracy: 0.9570 - learning_rate: 1.0000e-04</span></span>
+<span><span class="co">## Epoch 269/500</span></span>
+<span><span class="co">## 90/90 - 0s - 1ms/step - loss: 0.0523 - sparse_categorical_accuracy: 0.9861 - val_loss: 0.1012 - val_sparse_categorical_accuracy: 0.9723 - learning_rate: 1.0000e-04</span></span>
+<span><span class="co">## Epoch 270/500</span></span>
+<span><span class="co">## 90/90 - 0s - 1ms/step - loss: 0.0474 - sparse_categorical_accuracy: 0.9889 - val_loss: 0.0975 - val_sparse_categorical_accuracy: 0.9653 - learning_rate: 1.0000e-04</span></span>
+<span><span class="co">## Epoch 271/500</span></span>
+<span><span class="co">## 90/90 - 0s - 1ms/step - loss: 0.0489 - sparse_categorical_accuracy: 0.9840 - val_loss: 0.1055 - val_sparse_categorical_accuracy: 0.9695 - learning_rate: 1.0000e-04</span></span>
+<span><span class="co">## Epoch 272/500</span></span>
+<span><span class="co">## 90/90 - 0s - 1ms/step - loss: 0.0503 - sparse_categorical_accuracy: 0.9826 - val_loss: 0.1050 - val_sparse_categorical_accuracy: 0.9709 - learning_rate: 1.0000e-04</span></span>
+<span><span class="co">## Epoch 273/500</span></span>
+<span><span class="co">## 90/90 - 0s - 1ms/step - loss: 0.0514 - sparse_categorical_accuracy: 0.9844 - val_loss: 0.0980 - val_sparse_categorical_accuracy: 0.9695 - learning_rate: 1.0000e-04</span></span>
+<span><span class="co">## Epoch 274/500</span></span>
+<span><span class="co">## 90/90 - 0s - 1ms/step - loss: 0.0479 - sparse_categorical_accuracy: 0.9854 - val_loss: 0.1049 - val_sparse_categorical_accuracy: 0.9639 - learning_rate: 1.0000e-04</span></span>
+<span><span class="co">## Epoch 275/500</span></span>
+<span><span class="co">## 90/90 - 0s - 1ms/step - loss: 0.0467 - sparse_categorical_accuracy: 0.9840 - val_loss: 0.1473 - val_sparse_categorical_accuracy: 0.9556 - learning_rate: 1.0000e-04</span></span>
+<span><span class="co">## Epoch 276/500</span></span>
+<span><span class="co">## 90/90 - 0s - 1ms/step - loss: 0.0504 - sparse_categorical_accuracy: 0.9819 - val_loss: 0.1065 - val_sparse_categorical_accuracy: 0.9695 - learning_rate: 1.0000e-04</span></span>
+<span><span class="co">## Epoch 277/500</span></span>
+<span><span class="co">## 90/90 - 0s - 1ms/step - loss: 0.0505 - sparse_categorical_accuracy: 0.9847 - val_loss: 0.1187 - val_sparse_categorical_accuracy: 0.9612 - learning_rate: 1.0000e-04</span></span>
+<span><span class="co">## Epoch 278/500</span></span>
+<span><span class="co">## 90/90 - 0s - 1ms/step - loss: 0.0498 - sparse_categorical_accuracy: 0.9840 - val_loss: 0.1267 - val_sparse_categorical_accuracy: 0.9598 - learning_rate: 1.0000e-04</span></span>
+<span><span class="co">## Epoch 279/500</span></span>
+<span><span class="co">## 90/90 - 0s - 1ms/step - loss: 0.0517 - sparse_categorical_accuracy: 0.9830 - val_loss: 0.1037 - val_sparse_categorical_accuracy: 0.9653 - learning_rate: 1.0000e-04</span></span>
+<span><span class="co">## Epoch 280/500</span></span>
+<span><span class="co">## 90/90 - 0s - 1ms/step - loss: 0.0501 - sparse_categorical_accuracy: 0.9830 - val_loss: 0.0985 - val_sparse_categorical_accuracy: 0.9667 - learning_rate: 1.0000e-04</span></span>
+<span><span class="co">## Epoch 281/500</span></span>
+<span><span class="co">## 90/90 - 0s - 1ms/step - loss: 0.0468 - sparse_categorical_accuracy: 0.9865 - val_loss: 0.1119 - val_sparse_categorical_accuracy: 0.9653 - learning_rate: 1.0000e-04</span></span>
+<span><span class="co">## Epoch 281: early stopping</span></span></code></pre>
 </div>
 <div class="section level2">
 <h2 id="evaluate-model-on-test-data">Evaluate model on test data<a class="anchor" aria-label="anchor" href="#evaluate-model-on-test-data"></a>
 </h2>
-<div class="sourceCode" id="cb19"><pre class="downlit sourceCode r">
+<div class="sourceCode" id="cb15"><pre class="downlit sourceCode r">
 <code class="sourceCode R"><span><span class="va">model</span> <span class="op">&lt;-</span> <span class="fu"><a href="../../../reference/load_model.html">load_model</a></span><span class="op">(</span><span class="st">"best_model.keras"</span><span class="op">)</span></span>
 <span></span>
 <span><span class="va">results</span> <span class="op">&lt;-</span> <span class="va">model</span> <span class="op">|&gt;</span> <span class="fu"><a href="https://rdrr.io/pkg/tensorflow/man/evaluate.html" class="external-link">evaluate</a></span><span class="op">(</span><span class="va">x_test</span>, <span class="va">y_test</span><span class="op">)</span></span></code></pre></div>
-<pre><code><span><span class="co">## 42/42 - 1s - 16ms/step - loss: 0.0998 - sparse_categorical_accuracy: 0.9652</span></span></code></pre>
-<div class="sourceCode" id="cb21"><pre class="downlit sourceCode r">
+<pre><code><span><span class="co">## 42/42 - 1s - 16ms/step - loss: 0.0938 - sparse_categorical_accuracy: 0.9697</span></span></code></pre>
+<div class="sourceCode" id="cb17"><pre class="downlit sourceCode r">
 <code class="sourceCode R"><span><span class="fu"><a href="https://rdrr.io/r/utils/str.html" class="external-link">str</a></span><span class="op">(</span><span class="va">results</span><span class="op">)</span></span></code></pre></div>
 <pre><code><span><span class="co">## List of 2</span></span>
-<span><span class="co">##  $ loss                       : num 0.0998</span></span>
-<span><span class="co">##  $ sparse_categorical_accuracy: num 0.965</span></span></code></pre>
-<div class="sourceCode" id="cb23"><pre class="downlit sourceCode r">
+<span><span class="co">##  $ loss                       : num 0.0938</span></span>
+<span><span class="co">##  $ sparse_categorical_accuracy: num 0.97</span></span></code></pre>
+<div class="sourceCode" id="cb19"><pre class="downlit sourceCode r">
 <code class="sourceCode R"><span><span class="fu"><a href="https://rdrr.io/r/base/cat.html" class="external-link">cat</a></span><span class="op">(</span></span>
 <span>  <span class="st">"Test accuracy: "</span>, <span class="va">results</span><span class="op">$</span><span class="va">sparse_categorical_accuracy</span>, <span class="st">"\n"</span>,</span>
 <span>  <span class="st">"Test loss: "</span>, <span class="va">results</span><span class="op">$</span><span class="va">loss</span>, <span class="st">"\n"</span>,</span>
 <span>  sep <span class="op">=</span> <span class="st">""</span></span>
 <span><span class="op">)</span></span></code></pre></div>
-<pre><code><span><span class="co">## Test accuracy: 0.9651515</span></span>
-<span><span class="co">## Test loss: 0.0998451</span></span></code></pre>
+<pre><code><span><span class="co">## Test accuracy: 0.969697</span></span>
+<span><span class="co">## Test loss: 0.09383929</span></span></code></pre>
 </div>
 <div class="section level2">
 <h2 id="plot-the-models-training-history">Plot the model’s training history<a class="anchor" aria-label="anchor" href="#plot-the-models-training-history"></a>
 </h2>
-<div class="sourceCode" id="cb25"><pre class="downlit sourceCode r">
+<div class="sourceCode" id="cb21"><pre class="downlit sourceCode r">
 <code class="sourceCode R"><span><span class="fu"><a href="https://rdrr.io/r/graphics/plot.default.html" class="external-link">plot</a></span><span class="op">(</span><span class="va">history</span><span class="op">)</span></span></code></pre></div>
 <div class="float">
 <img src="timeseries_classification_from_scratch/unnamed-chunk-12-1.png" alt="Plot of Training History Metrics"><div class="figcaption">Plot of Training History Metrics</div>
 </div>
 <p>Plot just the training and validation accuracy:</p>
-<div class="sourceCode" id="cb26"><pre class="downlit sourceCode r">
+<div class="sourceCode" id="cb22"><pre class="downlit sourceCode r">
 <code class="sourceCode R"><span><span class="fu"><a href="https://rdrr.io/r/graphics/plot.default.html" class="external-link">plot</a></span><span class="op">(</span><span class="va">history</span>, metric <span class="op">=</span> <span class="st">"sparse_categorical_accuracy"</span><span class="op">)</span> <span class="op">+</span></span>
 <span>  <span class="co"># scale x axis to actual number of epochs run before early stopping</span></span>
 <span>  <span class="fu">ggplot2</span><span class="fu">::</span><span class="fu"><a href="https://ggplot2.tidyverse.org/reference/lims.html" class="external-link">xlim</a></span><span class="op">(</span><span class="fl">0</span>, <span class="fu"><a href="https://rdrr.io/r/base/length.html" class="external-link">length</a></span><span class="op">(</span><span class="va">history</span><span class="op">$</span><span class="va">metrics</span><span class="op">$</span><span class="va">loss</span><span class="op">)</span><span class="op">)</span></span></code></pre></div>
@@ -803,7 +953,7 @@ <h2 id="plot-the-models-training-history">Plot the model’s training history<a
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer>
diff --git a/docs/dev/articles/examples/timeseries/timeseries_classification_from_scratch/unnamed-chunk-12-1.png b/docs/dev/articles/examples/timeseries/timeseries_classification_from_scratch/unnamed-chunk-12-1.png
index 3ff438a1e2dfe04320c909cdbb15fdf10b4e91b7..31fccc9f45a861eac243a107c0344926a8c4deb9 100644
GIT binary patch
literal 65538
zcmag_1z40{_%#X-A|S1_Ff>w9BArTicb9;4cS$4CDJdl&Al*5HfHX*VcXxlA-~WBD
zbFTA#-*>ziMrX!n_OtiB?|ZFvuN|f!CytInhysB?&?O~Alpv62!w?AE)r)7~l|2es
zbnt|1C!y&GfuLeNJ>W*?eeNI-GKi$eM-{iUy?HYY95Z~xB{NeO*P4+=KME8VhACX_
zy=LkQEOzShOG5UGDQ$+ZcQsSSQO%clp5h`naNB(s5BF_D1fs@vUq;ey|3>v3w4An~
z?M>IB8`sW$d_f-gGB$3i3F8C!TeLhWE;R6^(2HJs6+7@mPSvdld-_N;l;-a)1mnlJ
z1@UO421AVoLwlNNpMBc6#lS9(rh#VSC(ZD#*vXF-Z574yHuxLJdEtyiT#?2EV#)YK
z-S!-*vdJ_m6fwyIriP~2CtZkbiqjg23B4v5pLcRD=y0GO%?srF5_w9^dbS*?#G@(3
zbT}yD-l>sYr+h=F{sX4v8iCiFGSqROtF^!d_PVI!dPOK=3@Tqoh@}oBg6}0u>KnJH
zW~oHJFpBG2N9*9C2VbjV=NGK<-wmP`ERlixl8DUl4|uvS?nso#$FKkX{i9R<d3Lz?
zI8|;;x-jzv9%-i9a&x?(cQj90u!$XfFtbVCaFo`a6dn<cYd9Cd!mGx<H~W5YaCv!o
z+D0m8#7VF0#h8zu1e?Rc9wEC04wKH*7e!?9KrfGikX1pL0Sk1PNRb}x;J|!*qcaSf
zbW}@62kIX{js^FjunY;4gp2yk)3oE`LLZTp3UMshowOizK8$Q^XTn-<HOdS`bQnm)
zC<I7CFx5)+=HJmhUAn<sV!d@`33JSBAjO%M-&moVUN(j8I{Ig)(Q$haY7(RN&&gfS
zp$y!YFQa2(mJ8CroeM~+*xjkeGitls-<(HUFp`lO{EerBxMjGiq&~fE$d(K47Ob;4
z0uqP_AqVLHuRyq@D^^mZ#liR-@=3E$g+4|C3YN<rTI;+!5suHSR;*Djoy5fOHeM=b
zl`C=;>Wl&nU*#&-u&SLs<&nT62#)DBY44}T3lDM^5~X<6SK&YK$3Bfkp0<O<p<X9>
z`nhL$SeFsk!9bA74Gs!MiP~<o3l&xvn90-ujL!q*iOl)}2cMyYoR8;6Syxxz?{6=M
zc14(vK%6w?$bB(B=Cdu=C|0x7L+X7tbn4NpmpqMw62u~3AVZBECIUi<jnJAy8@%hU
zLO;Ik1covdF3Xhk9&NS23fkRk!WzPMu892?>C=f#zI>+GZRF8eJ3jGXI&adX{nG^b
zhM+`71(cRDlZe87IHM#b&1{Mk25;Q)X0}KF^@V{&R=&AhOphP!?=#-(|L#s40s`*Z
zwP5Lr%gxiXRHvS`RQV|cr$vP6!9@Rm*=ql<OYnbufV5XGXQC?EbLgLoUyKJn{Tkz^
z6P}k(>-d0RcwslSG6-!ob6a0L7WdV3+&W2ktNwe=hq_byH8vQ?nr`4^hoPH?;o)z_
zMkS)4afX}63^Z-2iYlqw-;DQxnxx(gOu7whjvM`}9>%rVS?^8zUxo#`UOdHOcHm`V
z2i<8xG1O(d?X7;rZfu62K(ob$bxv2aTGT@IohZa^t>;yORM)B3KngptgGsO<6&7;0
z{nJ{4Ev!O+-zG*Fb&H|w>!;G-G4xw%w$^ofw2K_3k$8V(#r=^A%N}Z0Ieqe9>@2&3
z<%BwvuLt&H(@$AJ_FO!{N(yasQTJO3wM-EN+pBN{XMKkFy+p+**EWzKC+%94!^EFE
zDuOdyzMd{{UiIb07HIm34X;yYN6x{aR;V&js=r7sZb~4=fQNeODA;Y!H{>gUs#b)n
zM91pu`xVEVkofuwx~Y*S$LY{?2<N&TLm$d1G4W%mP0%&T(K5a^#;dEYKNao!UlJs!
z+=<C8{hOy)1>Je*Q(1h+95Y`o>MT-SFsUgqee!!2<o!P5)6{Y&#Q4v*UN1Io?rCyy
zdq>AHi)z!#Y4Q^8M4_x<3qhv~skfwu`)2Yv1}q$Rc<;Olr1EZ#3_yVMWyZuDFUxX6
zhfeXryDX|VoTC})8XJqoEeuT^d%6;_O<2E6P$?CSyAxBPvVQM3XG5uMJ3l&7C>sBN
z#5r=!MxSguW3N6-VMhKoiES#+qek|4qR;+NyuWDEb5gAz8w07F3kAE*XG_0f@NZQL
zqW=s3yO0-^A;N2VWP&tC5RFwiTdvGj+f=>vTB|hP?hV|BPa#;3BRnHEwYG~b_aKSp
zNhdwroeouw_!cS9-drALVoyJ9B3En^7>xF<({}uItByk4$Bz(&+FQQ{cLGU>&;9vi
zOzpuBMddWE#ON}+;U6E}(?h{Nao1<#IUeEiIHPU249p~*ac#0&7SikfpCG5_5+J-j
zSr`5%3~|HT-PlO{Uz_z~v@ku@TT1ar8A1*#c_jx**V^*2{7q-*Q{3zS@L$aViTrrC
zv~|N71-wMXYc3J=v^hb2zyMEyYh3hP;OS-T>i=0l{$D<Dpgi?ih00;d6=t6{Z{w<B
ziP6dO4Zc+tX@yI!XBxtS76-}}HC9;QciA$U_)nVTW*|Rtq%xKUn2?`_P_J3Jr&FLt
zHcx3J3e3W1ios&8TTYe<f(<n%6hWj!2fRb0oU&?XAU%EzXDn4ViD0U+GSK9BuLm|n
zia{mvwNZfZq7HO)u=_xnmg}2Xj_^!O-#UmhD-PaV_HL{&k;oek;>b^$Fw;s62DKt~
zaE*^y3du6m{}tH8>2R_sE9yAO<mMMrc}is<FAKfDwINQBO%_!>z}_-rc$ywBssBII
zB5~2x)g>w>hT`((X$qIGaMTLDJkC74Jv<sRGxhb5+>^e)Hydp5AZB7>WcUd}<!jeU
zlbpy4a*?`%5Meu)y{VX1*~T(OV;<+7+_CKw9=VCpu`x5<y1??)5!<D~XD}oTXwR+)
zNOR94P%vQI%_gr+``eEck&4DOT7A6rx_eVb%mamvS4dIvU<N&7T;P*|U}X)<wMwE-
zdgcN~JCyU3)}6D9udlAW4szRUEz8XCFCG<JeeVBfAj1VqFYzK*JrZam7bjNsUKiih
zs}`xbr*D<uG3zzueQX{#e4`Wvb#ZYS+<m(G^9bq)l)<XcV>?g*C=rpi?>h6t!-G5V
zW(h%Xp=uEx*uz6+CMIkD1)%@M0z$|EEMvSjz>^c#mW0oIeT+hXFm0kdl#M3t-s~Nk
za;aW(liU*k;~MqEDAZt*eel}j&e3S}o$2h^379F^_%;=Nl)2!&yf<ArTWc#XE#0^K
z1ax}Dm+nz88h0j$^D&HP5$1V(@~on4Q+-AnWaiSv6QP3_m(MZf59htFJpAeKF>lmz
z#G`DEZ!bzox&mu%`N+x1OIM%vfzYK70@5l}NyE|Ye&fyQCE`R8tZQ%4QFHq+$?<ZQ
zeeTM7pkCeOM;ZTYy@BpGZGqDm_rAfR(>k**bGF}guB7}PSsy8#b}Ed!Bbx@E>g0<>
za3i5b5qeW+*EPxu9Y38N1gp(ke$7@LeB`cZ@A}=+vL5oz(S?6PQy&kn<_dLEhp?s)
zYvG4%{U-ilisMS%Vbm%Y!8hiItnt4l6?_p-7xj9+;5tgX`ud%@W+xfs!_=N}gagCN
zRh5SM(Bnbnq?QwQPm8bfl~n`={7vz{*wTry5T61>O6sc&_3ppIIAM%|ieeRdkij3X
z`rXrc&@xys1X|8j3+H`qWse<Di=mWw;f~9+({1+9ZhQZM|KLjzey|}(xc^~JuysLQ
z-~gDjoi7Wmqi5aWe|C|3DWak2t(!MPk;=o3YWb{Hz3Wu%)>yv$?nH^Ql9G_Hux8=~
z>wh^6=7|SQl#UZ5y|wx`5qX<0`>j}@cB})LlH=d>NJmodv=cMS0g08=SA?MCe>myd
z(X6Kh#Jn#3UR+OgBqT<H3NZj?<x@&dRIkI7)OYUDER1m<1+b4=(LlQK{oC)ib}u+m
zyNp2WKa~f*q1|(j{{u@y1TdN<=@WLGc>d?Dq`ucrH8pb8{D`<+KhH?M(RIfoTRzWW
zhtl)+B7gm++K1}cN4~X>CxKl?(;)88w#S%t8{#K5pO!kDOudk0w4I&O-yIRc(12F*
z{n&P*G<~QtB++8=f>!)fwu1NNyz4y$=>Vfv#c?q=KwST8ABor%sX!p3La)YAr67LC
z^pm+3WJrJV?9TrT>?B_FP9H?1^SWMajToP8jSOW94voNj$EwXWMI@HI!und;ve4}5
z`d@9JAbT`Q_}f+=w4>AVu(c$)Xj@atxR&DBXt$OSM<eetTZI6RxV$}<3=2T;w7l|N
zpRP2MPGlG{X9L&&t&k-YNm3Lk<f2j<A`_fUASl=h<;MdBEUNDo5Bif=dY0omRJTs+
zGqdB_@WKSg+l&5!&F{R9Vvc=b*fhgc+-`IzAyC%jZ(@`~r|!f%0U9L}5BIm3I}_N0
zgS#~H^;LEs6%-T(iETPVUx|xH`3o)l8_&C3>Ed=J1G|qj{gQ#p3CWc4rMK_4hq^DG
z-@(aLN`RW!#Oqf#lFp8SK|zL-q_k>KdR19GQqO9*C4fHjo4gbXSRicXP6u=KP)4ry
zj*d8t1J3U<hm}bHu|+?|U%utqYskg<)XS2IU&*o7^*6|3=*{fWXIv<wR@K$POm(F)
zwM5=%OdQRoa~dYQ)l}K|XZlWoplP&ng2~b#ukp9|k<#Oc;98u``Y>3X=s)`u4kL%@
z!t_25f}E(qRFAyTGo=;iU?Ng51~HF8n_rXMG9-$GUn@Q13#`NL8mndRnSN_wiPjUC
zFLitc@NtyK5IHVW?&Htu(Zb4L*W`lOE4)(nmCJJUGzX4p--uq5nHuX<sXU8?rt6TQ
zI=0k-p1ses`fvkUHigQSY9(5vB(YCmI>VQNtE$%JH~*{DdQzh;vwa~!ct|dE(F1@H
z^hbF+^+S+}tfwow`ud6hMp;-`sCStI(Q?SBikOkH+Th$;Z(3bidKglLM@Tp`YBg7G
zitE;L&~dBU9pR!P2Ns2j=%H|_Y}To82pel*N+sTz&H!tr#7|&&yhvS8{{kBkP0VZV
z9uA(vW-ghxTUJVH#?oxK>>Xqa9nihA(^JorHOct~XH*>8=m^Bu!I~{z?w5{$+bW+Q
z8^5;|0p%}nEy0(biy;3iKebAK0<HT27$3mgA1d+e?7xKfy<&O&`t{)8V7;9=4Gm51
z-D4yR@-b2l3G_uzR6F_A&1{^b8U9-}hSOIiYP_cX$>)kuTI;8&eC`!dQA4~lCYm)`
z)fU|8X-}AnG;=9g;$p7eF@G(IA9Ia4w8;0FpIpd46`FjVlmdzbsZG~Czh&X0`zwsk
zy-1_)hwF*x)w?DaZDwos*T(=xtZlJ!?@w-S7EL{DeH3u5@OH8R$nXAg@$qUk3XIb|
z9pI@R)I)F28~OR)x{br9Zv*jQGvfr@RT<%@!`ov7)R#3JAO|Zqy6mGN22@s_1|y;y
zR{O6;$8W8yoSbmS#Kf4-U_nrx=jG+CYt=+asHmuv#N-^0PfB?Cw5XTL8#iiz0D7)~
z*^Np<LJ}GU?}w3YP+qLxD!|9bm-JqODz^Vs8hZeu(V9Ykn%-p(%x1o!*-J=7<T;0}
zzg`q&2*|>Y+zaB%Km;(K$Abka)bIEZDJdzP@Vo+<+0TPZZ)Y^Guj}F7HH|0i-@o&Y
zkPYi(u?Rx`+r3I4AxuvE^15D6%4WA5d}zv5DOB?Q6X|?qKQ=a|tE)@AU}b~y;ze6w
zWq}5=+OFljFsRNaFO6$`&%$Nd2qXSOWs;JTN`>*O&uasIiBP=zE>r3~V!)!)>cdY)
z*0#9#2tOMY9)5eXTN)KcB_YStY;L$cnzwiwL;-R>)@!t2Otd9hmqzb@1&Y#4kI3Pl
zX6wq)3E9e~t=x&}QBi_~N+(g@YbWBTz6|X$id+qJEN_-#z|&^HGkoAyrBBc*QTvZ|
zOHyG48$RJgEuh+DDRAW$M$P0a*l}5KfJLd4%dgZ}(11@9A57yuy*ge6TE}xl#J?u2
z&3cbJIUoZwPh?U44W_&qX~G-OBX;i%HtYy4`Bo6|!(X9@h*`VJW}XMqdUw*7B^-i-
z(zP{`gNo0TwvPG+08t8im4QtStgudtYORuqCX84Fe0=-}G0J``TRp31u+t_9s!@yT
z|0!CjKtWQAwz09HPmua=*nqJI8#?i%gJEkXdT1YiMl0kxDy<nX#qRNJ>jwanKSKC;
zWo!EOYVSMlTQx7nsnXVmuO!Vc)d*c~mzD!~T@UWaL!>1ozm$?fxVXQ|P#2AR>`j#~
z<3Dy597d<a@_1cMlxSCaWeu2C&h9_E&T6JUwL*DM1c3n|3b}fAUue*@S}W-KC)3@3
zkewqn!jRy<qK|^mP4ju8S2_6|5eQ%Fxw>+ru1QYeC-4~x=V#dFvqhjVTVw4ym>%?V
z!2ZArM5N#DWND6=EDIVRpFB&kTt3jUNZoKi)_K%av3rp)juMVQ@oID|UTQrx%5~@8
z=fNa?#YzqmF*&QzE%>O3GQ*C0Jw3C*G`jHy(fyffxOirLlA?(p8unBIr2~Gy(4UHm
zKMKYdEnd41N$)6MCAX?sC>^f%EqVzV7%X*^Xfx^5Dg4mgr4aq<b@i&FqeEX8S*O|K
zylW_9KZAW>V8FUFjhl2q;Hi#O0A<YF!Qn(uaGSrbzWy^CAs`p5-vQr~S681*^8C7q
zT%wj5ACE&o&|{w_l=X5?hdX04s6L<OiN1;^=ad~B9Q@ywF-9U+t>{T30-$Qd9A|?M
z851+As*3IZfkfu!=Afhh#3R7A%pWmta6R;)s3M5Px<yz^Wif1T%=A6o7)VY|{+saD
z)5C*@BTOjEZC2_RyX6Fw--C^c%B0N?uD<4x=ZVx=SB-7$UhGXfW_!pw?S1<qjHlmv
zeGsKA_;6*M?vCm)=dDm|oZr*aGbdU1dF*G_r0xAXyD6{7J7Y(KTV^nqoy!GFJ%9VO
z3#dAgd>^_tGkse?QDMmBI*#%tV1-7c{8(($X|(R%9?Kv3#olQYUs|_3vO3323%>n?
z9LRXivZ);NGtIYQxgvaCp+Z6LjLzPr?oUlAV$!P*8oeB}2^c_341$pX^hskFhV{no
zm|}<X!(8Q0J(7Ay1u%j&VFlPRPo7e}&;1RM-xgZ-1?BSKI{a2&65aqT#)%sFsoBGa
z--iJO_xA(sNnf5cxX@4rf2Lg8)!5dnk~^<I@)>EoNN+$|5_paa)r^*)0)-3{8Ts|)
zC;7mn5WOhvT0JIye!#{y1^b`(ofoMX-d!Do`SiNj(*zMsg(bOx4TKmarP%y@n!b&m
z&;$*J3_#ib)$&=?&e|?C(Zxu7#bZpt7WXAem%BV2672r-hr?>>sB%c4Z$KDGP_G2&
zxhfJ<QarYXv%(^XL=P62;(BlHs=jTyL)41SPd|^nvt$xk>DpCO7jP^47=T#%TRto@
zl7z=;i!Sdy)`~n+a%LZ@Wr><4;7b>@x(2MyiV~ZH>8X|>Bm!Pn8)X&db6(?rakWLW
zQ`jvxX0iYddO<2nI3Q$UVG+YTgDE#DBhv@gIa$El1K_a@VfqA=g@u=4a6in;K`c$`
z@p@kDDc>U+cA9_7WV4(oo<omx+#K9L9taUw{PEaU{W>KClN4N-vzt$}`h}Z&j`3zk
z2<GOD{gXz91EgSt#(4oSCCd47z3V3xpZc;tqyGDaZg5%r8W|b6rKRP&yrZjYsaY*?
zgBP-Zz*J53McQ;}b&gum@<C)Or`^4Um6a8fFFu$J(Ra^3Q1lM)E>0-+ME)ALb7)VW
zg~4F;E;>(}8(|q2>T=%P+`8ZL>8Nz1MHhFmpHseZXO7wNLPQ&GqH$vKe)pQU#n=?g
z4_Ae+{QAi;p>_EFtm6G6<+JVW(*2fdlipaM5dIWsnsgbxe4>KYV?Sk6y*onMCNX7G
zbjP<788lD!YZrO=_Gdx4%R8v}6vFv{m{Xq6wHD1dv^eIv=-n;z-9e=2P>m}0)UV}#
zj}2yRZPU{1Fd9~7sP94iy+&%()k~Qa(eE5K^BVsx#yv%{u}0*o#)d7keb2;(@=uxV
zw3wdf_#}!|{Uv5YR4DlxC3rm65p8WL%j;_lMJAeWy1C20p>}gS7QgQUzXA;xtbPH0
zp5N__nSrF>y3>hH#Np-Ebh!wbGAd8wb`%4G<8Phsz4Qx6QgU*7wKl|l7yGlfH#egb
z8`Hwtsb=uj)*1X&)$Sa_6C2zg&H14d`J6`2emOhGIH-1%wDPmDpD)8{#Ky@cZsN<m
zX}~X17(EoW?aIibSN(Oz<LGfeyduu;y|jILvP<(NZfH?!xJTnp!YYoMrG<=)Mewhe
zeH_2Ex~7`m{IfXXV<U8POWAj^$7nikji9BWu_8nZE^WElQR(AY5&?K9eBr`#Lr3dr
zm(K^P*-eeV<}jl{G)U9M*>S2S=7aYPp827k7yfAl`!rwlJaw2g8lJ^^rfO<RBTwo2
z;!XCW5MY_-9^Q+h;o<X96-QnXL$i&W%s(o%&N)*58jVj`Srn-nSDR`S(-*PN8};tt
zkIwFSq#{6`ox+Dc!^{8lLueg8f<7Un^&si+0^T-kN}B3#x3pK+iWgQ`sfdo><~@~>
zZ`jnd(R9V9@M$Xm3QLNLc4t@Yp)^?Q_q`3Cmj{Kz*|h2;kYA!t7!&MVw2r#Ex;5u2
zhBAh?{qoKbjYMY$2P~~eBkC_YR<=f(YCZU~eL=@U9W37r4(q6<!Xs9>Ad!Em`xnpV
zXsX^1gm$HMJ!`Xhr`;LbPY3;`9xwtb@rT!A1D}(yQ;}}UKcw?G16T!M6ag9rH%DIu
zC12VMPne6ifX-;vd|^+4E1G0M2}i<stWEh(%@Wx^@*g&BAO@_sUT0<Id4K8fq_cbD
zO{zazb3nqV9kYAAwDX5ne&B^O>iZCUi<|g;xbWj+sFI>z(sN0wg_YKH!i>!bfSFnb
zFqaPwB9l-ZNyI(f-M<f*#z?FcC@*PH#Ppa}B3Ev_1igp-FZexO7t~=Q?G?o=%p74N
zMjWX@*DVxy;ouVg$q0RWjm|*9HhiufwLzVQ3oVuxki2`J&aXm3YIej}Y65Ayx!Q^h
zj4gTV*<n>9u|C>Q|HEM&pBOo^f&YHBJ4%}8?d;*yWTjIthp2GN^+qZJDqhomFIF4s
z(Tawhl`WniDUaU?ZrSIKP%ZWu#G7WXlm%1y3ayM(MoB#BwCPxr@;~1Pknt>=y09!+
zEv-3HC7#E;a1ixxy#1DoLzQRJq5!M-h-LpVj4OFF;LXa);NXKxk*l6B&e8Glcdf>a
zDZ_xavQ@$fW{t8R(^jUMYRSH=kQXi(AwpR!-(}kS6E)G+oxeUU{@Q=yB5p2ddRcM6
zG87_LpVa%B{s%0II>{WD@x06pVs$)qzHcad@3J*N+>NNJpH>gL8r@)M*Rzat%<AN1
z@&Dz%FIj~tB8bb#$V7>u=l?67A*+M<VenoI3OcDUCBaX%`aS}<2g=ZFwdEwBACr3K
ztlxbAT5ohcoB;m@$>Pwrvr#@{BAJ)8x#gLkr8;H1)F;|_`Th+y>4`((L8!I_1zo?i
zkXs82GjdF5z#A;&R8MRX)=v-Xf9J##s3S}<K_Nj*VN2m>`^-jUK58!aQU?atW+UOA
zV&H;#rs`0(i$(IY2Nj7&dmLFluK3`~1C#9>#egD0NNVbAPYk8>LgRFn$SVl@gf()B
zGAB9t$2=w7l}-X)5^>DcDBW9pCNxaUA+KIgmNWT1)v7I2U0pBth@Aiue=nB~*X^y!
zcD8O3?^JYN-kWy4NSX8QuSq0FYFxjj)h}|R<Y!p+iW!*8Z$#ZBU(u2NmA=wR%aKA5
zzfge1psU!KVB$ijKOl(9y)=Ry+j5sJNyy0^Oq8TjQ12Y&(7gTx@VeRFRB%n^>WS!F
zOT-^pvN+>a+Zb(*#)gK$4F21Y9`P5*urJ<8J`daLHo2u0(^kjF#~B$JhZ+M}Bj&|w
zyN&tnbU0t@zI~(3;T|>jGH2yyOQDe6Fpujs@w`8?M~l;Epr=j<K3Q2&`us{y>8iuy
zBaj7r&bAa2KLLIB%7I|_4S)xr?Pm4EpVAWmE<}1R2mj#>yR~<PDm<I10?oG<CB!&o
zl{Vf4XfNJIED32;yvS=;D8hi`Qe_HwH%P`&3qpA#NjD%L6o&sq5QU(<j!jG3TV0Lg
z)^R;tXbwQa1Qh#L>w&D3f0o5b0<d#^{)re$akbw%z4kib6O<Gb5ARY@yA$FZ*Vjgq
z^vkFA-%+3P=h&ZKqpFXMI_oVU+LGqXlUTEgr}le3KC->*z*%{kamm7}z<6f5_5034
z&jrAe1P^wcy*QD@I+hhXRxkH)H{wrx<6~kBE2kIAF+i6?BJzfIk<S5x=(ONy+L(8i
zy)iM`>*gRmo@xH*L2|zFe|rIL@%|Nr4b84zjo{0F4L|**nQlm+QBpPwV;SBBy5e0b
zhxmsNA|5;b2&maPIkWxxkg(_c9+bsw3>)jpt-jSuq2b|{t#<FIW&G!BY}@N@-$<Dl
zD~NTn1i4AJuS<K@+K9G(=}|D;`XSxuI{%*F@O8xHx1YC&tMEC!Cbf%CFMQ)wWfmZ8
zaMbx+UsV;){5CEzruU?|`JoO>X}D-v6As*$u&^-Dl<~aY0y7M>g8Xp{?OL08fPQ>a
z1t<HH-x&{fNgLFcehuU@dBqMWg>N`YWUJxmC<DVL!?NYtWmJt8r;WBR8Rfb}J~OKf
zwJT=S8tgtMo`51ry1^N-tiZGO0sBw(2N<Dzc4nwZGAp`6S0T+iR6K^^Z6^W_tRi*T
zJUm2XW%Z`DWWCY%QJ}pere1-*xF3vX0b1fF&uKIF;rvAzgU<ZY@^bb#o0FsEi%%Ol
zc#ZrsCIWcaj@tS1&d#mYYy#}sN+lD<7W?nc!xw}E%V~9}OURN`qx+=$uGDMgOO*0Z
z{(KWIRFaS^jEOneyqtUnZJgID(f9q94#7|Ib~MGKRrp~wT`_%&=kO2Y>CL(N5Xmvf
zU6qr4qva%M#K%o$fhsQug1ogn5xEuzgP+4y?X#|!k6He2pA21XwuYGGCXs`RE^}wC
zMy+Ny&D=`I`(y0KJA1|3i+w>LL~IUa(!6;SK`KbX$_;UIGchr7b8{nmgMdzGG{RS9
z6)?*gBC6th{^5;Od|R5k`4oW~(QiFE9wYOb_ep>xhWm(Uyr!ZVXM3$jM2GXP|EK1A
zd=b%(`J9Wf?tJwU*3&!2fa8<}J6YShU!0xE{l}(#B$cM}!HAF{mQN<5QYh}C^;NMt
zYPk|I)&Ypc+STm?10{?82&FbMvGMWGTfL5sAbwiTzjf+kc@esSwhBZT)hwYP5TGsB
z=2B9BfiBS0+zgU*`gOwDnUkA)v)fc3FC!x(z7Q4=s?c$vrIn|nS~;zinkCAJ=cRC#
z=WSkw3mJWW-<d`u_#2APjvF@ifOa&L#vLc<>%+`k3KHhy5fz^3S3{VNO5xXLU#G20
z>lg}AlCwkK(Jpkt{o3*<LCg6HAmxC*$Ly@<<kU!C5(5y(4&x;cleO*6cgAlo4?*3S
z?e3=OqjB@_s8Q`1blEkpoGuPR)o*Yr6beFNCRHsONBynSky{9fB%ZqLCiSfXw~->G
zCrdu$CG>0jCV$XQGrzTXQh($%8^<g{`h&k5rXRxBqgaceaJKL4jbC*R-mK{HGOese
zXCTs;rSa*x7(BY(sf>+|+AK7^hC-n)Umk7bl$VzSJQpWHrQ6_i?4b!Y`2?7*x9grz
z9z6k{)9%ujQtK5Ge0=o=ZC5tiuwBl^hvV**u5h?`TE*;(v1DT;%nuA->+GvfdJGTE
zGeNGw$Wo>Eed<4(vzb<2ud4%!S8n|0-%-Q4hg;+IYyCa!8?#`S^V3{!@yI@}`zJgB
zBjPmtV+G-IZyH^guQVN?S1TrXXcYjHr_JDFOn&%S0(SUO!y=^L;!*VSV1Ci7{VVQU
zu}uCbhjn{{#SD960F=*Wu(PwX=`VMq{;Iwr4e^1`nl$bY4ZOeAJ9wX?RETnUgMv*B
z*A^oj?!^b1TAA0;S|yPY5sWGY!+nfL#V)tRIotknzKTV>`C~M9-sCnfm<zL`)f4)D
z(XF)@z48OKeK3`C$bbbfwY`CTNj(5WZ6dw|C{`2N%-<C%hfLPcOUlb*Uf-df-JzH|
zcGe6Kaag%^f6a&_<|QFKC)NNg<8)o{;iQkbRG`7@`otdVHNDC+zhmP}+j$1O@DjBm
zz0$>U<B)L+>yhlB4;OAfRq3Zc6sc5%s1=&Pmniij|LZtrV$`lVJ388$uH3BhKqu;b
z<Bk3+^zO^@;8#1dpcaKmVq`i?OyZSNy>}@x(Gp9`npNf_1qFZy<oo$S;(7rxN=r*y
zVBqREJ-t5NWG5V0()DDqBq1=p`X^a!G4A1LxWC*H^3H{V05MOxRJ-<Iz<)JL@Zm#}
zO=zUpH420*j+~5zj;?odlJgXqiHWHh1*ue{fq*1~-Dc6e!a{XpGK!e@;<|sDR)>Jo
zw*PRxNUzxg2;$I=_}G)YTrXx<bx4%##wpLeX2eC494RPngt9SWV=4cfATTUR)u7^E
znUDW6eVNwQ&JI46191;EILbehRI7HL%X!5Tp?keANihdcD*n0M&doSj$fURG2iCIX
z45&G;SYNYgeV4hI$&t##k}L$hyvgRf-aeBG&@O1ABnPD+O$?!1Ff=sO?Pv)GpqpH#
zz`WTwebJZgnOwt3vyqQH)*)Vc#3$}ucyF&}nZtc5%Z4!a^oi9x&I`wmj_L9g-zvow
zA+7OVGCUL&juCO%+*QB;BgZ7+<I2*M=JN@RR$xKvH)ZJu@Q}x(CwkIl)qntz8rzM8
zK0&F_>wQL25`IP(c`@6>-Pc${3CZ6vF%L6KeLzxat#bicsbr>Zgqim}Kq`@i@9oxE
zGx%QP+&|a9m}7U5NuXC{CDjGe$nk1-b91wJBy?J!%@$;NRGUT7FF!&}2e6vFDI$J~
zQ;$)`eb!AuDL+QY^Lj(4FI!l^kc1$)Fq3}D0oulHRmWX|B;u_+%;&Pe<|KGDZyp%9
z@^B9qj^A=&p2}f;alYfcG$Bj%W^jKB7-cjWH1`t4K!@UTB>J!Z0#wOe*2&2!fbi94
z!G1&CISVVz$r8PJJmhkR+3?z`o_}ZSJYyR-ve>9NontiUG?5e9#WeJ)-K$&Y0Z4lF
z4xcW1#}=YU-!yn;`@jbV8=X=YjBlNyCo%|av(`DiI<^o0DdWz)2Dh}<8)yA<=mTsC
z%hg7;Kp_i+(fRJAFMDjS$<Acy;>8&npSRM;d<oDMp4?gv8J(6hwYI({dm`V6g2I%4
zOfEk(2a98t=r*bqjn`PuUO1R5f{pnuipX|;3bX|AQhmisOATs`*UGxJIX=5Td{gt4
zE8;C0Tjadk*Mn~{QpBw<2DL}EkoK=MYku-Vqbsvf8<|Q;a4M(I!RYn?Le-Pl+;j-B
zn&3|7aoL|??`s2ILf}UN%>^KJFag(^ot<4Gr*OsWzI+Csy|c5ky?s_hYlV1O^Bh}l
zEI)Xlg9kQnarFt}K_QVO{5&3KTb-REMX(>F9HVqBzD2^GH}&-GP5Ls<RjmDrn2Q4~
zU(`d;8rZ!#P(s>_)M_k#UT4FjJYT+gcXj=d0oL3|z>ilh|IP<$F(GtD=_QEE`r1@Z
z@STm!@k*CIiDLd37@uXiSV(BfRI95`H9cUq9*hZMIlw@7SSL6P8jLe?nQ)Q2+F5sj
z$k+tInqb<?D7#;tzw~bD3SLzBi3SQ6|8gU^cNuPbEdExZFJuq`{*`W~^c!lbEID0>
z2G`-q)Cd9!j@z^YVxKNsN+i{r%(3l5!f&MylBY{`FY=m#l@x`^T}M8b>bL6E*%|u!
zK3?(?l@jrLoT12*#xsUQ?5cBEjuvZH`g$P)CP6uM3HiUkI?ZtO-&P=brGp_r0RIE^
zRNCk63N(JaT=8?p@m_7|Q*ss5g)29&U3v5#|FO}B%6z-lHCb#lmkh2%{hJ{EPdG4`
zf(ZljAWPDbtMNx2sz6cr6wM2!!2aHE<G&W`>{rMADY?1bfj~9YL=3s{YV^H01EME`
zxV9dkBG&mbUB&e;!+dpGe~&)gM-Z~_04?9m?Fy(q_;U9omYg4ce<-TyOJq1YJr&k?
zoyPO!*~ic*%cT{S95HmKL%2ssflBq5^6BQ#qs6M8k)mZ@_U~l)Xq#%i;q4*I^G)ZQ
zT@T@};|;=j3t39=FA^nog7GT3>G=0A_U3m!bq&lWj}Su|yxk2tQZDJ9oJ22>z${V4
zNML|O1oXf-ZBCkPH2l>S?%?U`rK}L030kfKAJ{WJ2hI1YQtws2X34F06r0cXQr@op
z1+YSa>;(n#Z{3fpKg%>;p1K*VOQFhT2ze0VNTu;9Zrmc5l9-R?N=lMvNyr4&a%Kn|
zUtB!YDvn#Q^YM{m!G9RZjv^Htyz}UdeFK`@?mz<U_^wO&<gZKNv~4jO<g063?fV_X
z?Q*j*07Oar#Jf7f4n#AvO(wzn&ZcG^yY)Imxxa{VQV^ch{U$kIr5xVEkp&I{((f{Y
zdC5Lfl5v$*)_K`2ao}fE`Hb0Sdah^QS>J+wZr=XVC6V{ZQ~ZjRQ0cnw7Ol{Pj-1>O
z=8upq$ZXT#*;>(>8EWWjIbwI5IcfR)X46q=Lz$j6(-cvTr~c3buhsQFSjgu0Bo<Z)
zmL%vEibaulKR(=(`reILRIB|`ku06Oy}DxiJWAy#{uPuF=)=idhGf*pQy&6%;1Q^`
z?tvrY=hvl<`^5Ct`|1dEb5POI5rzwu^8?O_I)*R>nRw1S&ay?rTfBNf-ku#u50O%}
zJHFfa2oKPl_xj%iKpODqRSJO5^^o9a@0kFnt#pk>Q~`Sk`wX_0ZpwNW*)!C@zE9^S
zqkov>MAu47Io-!V$?1+F^_3{G#@`b3ys)a*d5B`pB2U<RWyQ>iUTA^|9`NXeYk~4v
zOT)Y87@x)I6H;V;T0LqV%+?-WF8a1QUy;F^`ZTAsE{5+ZHgx*IclV2i*SI*_Y;O{u
z{q>%tP#%lw<dYoTz7+I==BS<89DqEbRH)Jy1r;sf0iDN4>q`!so=m|<kR(jMTNh}d
z&(<)NydD8ajUmdWa{A_CzmaEy?(Bv;v$om&bmJErE|;Qs;T33R0DC0x^2*A|fu64S
zMvVb*iBVEgp1LjplLVknFHlkY{yayG1U}#Qz+P}_QZ(<QmouOG^M+FQd?}%M!CUr^
z+((3zhI$nRT7jjO8ch?>A!=`G&r>P{4IWdDa*ecz3SRtB5sJi5i>F3?pMY9sQH^pM
z2>OZvFtUEDCod)bI4{d=YTwmkfIyHx<&Y<r9xf44qs90ye$R{N$q_D7y6y|wOpA|O
z?+E_n=mnUb3tEy~QpuFMU*9<e9GfaR{9171ez<0cK|^D4=Yi<lI~s{;f%{>^>$7dW
zcL5+3w2Er2rdj!caNG99b8Hm#jlkvE8G}yU5ol9-dwF4y3d}7}NjhykhXmZ_*uA*H
zfcW7G?jMf6<Kw#<j|R-kmc!>+Q<FuiY+oW{-yiQ)y}uXx+v=p;JPZQJpZ>OC+_m-`
z;f3w+(h+jeS<>THU^4MLWQVK*SO>7Gi-QBWn{klr#cZchg1=AvbOw^cqbl#eju>?C
z;QwFG*yb1^0SM2Ze*<V{uFf8UhS&jkl*{1)@ty-E9Ua~BSz(^%CDhK@emQTx<13#1
zCGE!c2K_qavwJu;RCJr)6C{$7ys@!vu=%n0MAM3toAceIloYLOkuWe2A~Zw@NYit(
zWpo)`VtIL@&Rb5)_Ei@D3iN&N*%=wp7amD1>|MT`J&sq&+-%Uaga2bZBO{U5Z8NXa
z;Is{>gX3=ZDZZi*AL}7WHw5W&(rRX}X*$-tyjWVfL|XY64Ld7-Qki~-i>|V~wwqT(
zRCFMNUpr^yi7kP+d)%o8eegHDt}uM!qqXV8NMd=3+~i6~Tk>`s;62jsr2#`G2J+8G
zrNSzmdWX`3)OtW7fq~ykxyN;_{Jo8mK!?dgwubrsGq}9wW<HZq6uGb9M-e}2+v=ik
zXeF}y5C6{as>R4kh|3{EnUN^Se{h{$U0ENoq1<QL?<otGGIDVsAT58q#fa|xA!SRp
zTK_M%nP@V=c(+Rg(By||193JUi9E>CZmPqvmCZfh!v^SL2~XN6&_*Xt9PO-E3nNk&
zj#oxzHm&cS^vu$9S@0H5kf>R{$|PWjiLXx1uCZ!xxwSF#T_p9nBJU&STKEGx6MXJi
zA3$;9iX;N<s=|kW<&75K3rwVE&lbvywo@aa-2Qka3ib4U*ddtN?T`4_xG3BJeUuw_
z1KY^zpNPu2ybNIVUabR|V8b~WGjD7gIvOcMeH+aUJ)~q{D3;F@kb$)md!Aioo0Sd`
za@MCM^PNZC*J5M+rlA&;kgc5Fw^7n#T0yDfnnXK$WaYR2aCZMfJ;_WQL$}qZS)@xU
zO{BtcU+d~-SrMmQYE@C?bSfE-QLAVDgd~FSSIwghuk&bESDt>W4<r2D>>4JN&+RDH
zvuHM>GmY031EO}%TCZN(*4D<E^YiCV(38_BQQPcGI;;+PGL{4036fDmU0q;cV1ddn
zFfoQ;AF+Pl8<?Ng&lxcXI-_N|#>wW;V!4a;VUfC)dmWgpzowPvdlS|nand?U4%Z2S
z%zJYq@_%>tFQ~%bj9WBYiVGdb7AWqDna4O@9`rRe#WU>Q$zyAYQULq}P{1r&g6Zxm
zO|6fvW{Fm{B8^;@k8l}7?x>M`W)sKMnT@@@{Vag%y-Es2_G>IFD@&w)xgvWr?!XLC
z@NSjVSM6~)aiIF%>(QP=#uT%O1&Iw<84N<k5qLu=Pk-|R8z4B&Q0B-L=vDsi>1+W*
z-zXGg7#gY0ZGs_w;dQ%TONT!+iON(axKpmgz_hn<V_jdCPv~6V_M!g1^jQxI%k{JD
zl8g89M$DaWMDlZgjay`F-rwH?Hu0H-;8!G3)X3_OF{7;?=W#8i%K3w-eKLTVYL&DA
zx(X`Q-z4UtQ!jJ#Ez59$3;}PggT4&&;TK8F3375)z)}tRq^#X!aj~&qUH5~-;>fX#
zcP=@T?zn(mXg-tu!w0fNrWrx6Dc*&4{kef^t>;X-R!bYBE#9|4EBQp-<ghLkfSx+M
zFAbTh_-vV;k|&j*1#&<fn)3sP_2y!DCz0+e0Sd+BZ_G>yiP0ybd0ymU8Yh=;8u^AY
zf#IUmRJ~M3t$t;)&$s<b($3|Nf4%S7_7h{e6qMyiwFaDTcNOc17KKrZJR9+1uRVh3
zw9%E=X^kKl4au+4%=C0_cH^SutcDxv(s&OPU92#wQ#?OXbD)I*LxrP@3kipnhL~8e
z4iji(Uw5k`yJ)0u0Ihpqxd$<&BL>@CLi8nQVpf2;x))@WW+xyiVs*mnt3`@xkk0#6
znWPzc%UWF58ke1$L^EU`8qW9}XXX2Eg!{9sYfKaK(s+zktFGLU9HQFK0aB_EgFh<&
zN*0f=3Pbaxo12>YO&x*Pje>##0_Aqdy~li?rII+g`_zqh8a2Ur`-Nyy9g{FRF%b_B
z58xI$NbBPRug6(nC8}I{lYqnCdEix?Zet>?;>uHgmc;|KF$fDnuRh=8etK|*7S9x5
zslH5l1!Shc;NT1{d!eFnV1GNWZixe=D@R$qSQkB#SK+i0Uj28qpq;@;wkQNxCZdzC
zR@}pe=I7^c&v#u5G~Zs`7<C>QaqP}tq$@O_G3kFTP<pd4!2j3W!=@anq>WLie<x}H
zCzz<3n~%3t_bbRh=B4lI(no4;WM)qN$=z1H2W%b}Sm&YLCCA@*e)|;>Q&>!NGz)<D
z`^ZC$LBTTLSw-m+DBd#kF=>=7rFvdc+*|-SUu6?c(hTfmpxPzsHs9=2YSh)Q4nIi6
z(ZH2yFi?yv`m}<>J4Qy?wY8VCctqL^iBIM{dxtM0|3t|_xjSfWiPf_Q{-;uxz4l>6
zw+y=~`_;7@^sMoVlM~RszX@TxeK=Tb6*T^X764NK64evIcg2u;@rq$*`AVAP!mcUa
zohK0-gmM{|g79_MAH_D(6?(Rs*Wdh^M=Jbmb7135$bQ7}TIAPRYx7Gto0_!2VZFC6
zz0Ni7&L=Wy0$>Hp1h>Q{4l=1;?y#n32q<ps=Al~7cKh@~;Rg3SRttz3-4D%mb{Nh(
zhGRDsSoVOHQDODs<e_E=h$?)HFZGKzxnm(N(OPYH_c}+X{?--P`!f99xo5$+fd&p*
zuukL;|EV%OHKMP0!q+HMm6keV)F3?yfa5p7%_Nn`P^eUBDk>_RFMN+ktBesgRBrb$
zCQHP+xCDGZ+`D`3WqX`&!jtG#cV<gwkLE@CGvR%GpFcP1U7^nF0LR-Vb0y<CJ3HTV
z+7$En9y2hbqhTZ_&|XiOzMo!jc%xmyvX!NFyHka%MJyF$IIG6?8L<)>bu+1IwxQO>
zdt(^3&p6vY55MS(6uGNcyahmIyi}fEX;fq+C|!2IJOQYfW+FnF*-!?FE9r-vKS8Hu
zv$Hs8JiZTim8|PWETeDPKY`7F)^VnDN4fTOLO?GZj}bI}sj&FL1+I>k^AiR`ea^b)
zxSyK$l_AD}1`8jGW*eJ9k_9*0lNEv4;n8iNV0;HyE`kFCn`)N#0G*-cs?WCK8qDk(
z%QwG7fZVZITb42dSO<Enp65FW^<p>dkvGHBKwiLx4{0Aqf1AksONkB)3Y<Goqs5uF
ziS__ZCh+*;!R7D_7wY>;2zEUB)T7v+BH1d|nQSu5QzdGpB}uViS&)>AP`2FvHXpIc
zk6WSj<;!bw^5vj_jxnWXhEhzc?r<tj_L<$T4jbOaxfb`vh)~PLLgf(~euHYG4JZp5
zqF4BJIIM!;Jr*|g^3g{sngBz)3l;M4uI}#5;z04N_N_Ve5KNd=drNbH8nM1pTl#D=
zdvl=BJD^i#>~dzyY8PupY`=&5VM|JzK*0Je?oY=!t_Oc?oMJvx3L7k5a!EuYRD2(R
z|IX<#yUPy_4wLOL0RaI~Fwtto$z4Xrr>8`$COro~gt(lajhPya2I4ac)(M`$(h<=K
z_vY$PyZh7)hFAOaX!WpYqL4~-*H<?$(+o*39|%7pz%30)(<-4kyq>z8?f+*A-$%iZ
z&WHIO+AH~`Ts})pP*xpg@p0NA)I)>vqPI*yKY@<G%V}}pV3*TiO+%xEfFG*%*6W_d
z0gZyxV2;+!L-?=p4FhQ2oHXB_7Zr9%u{o##jG$*eNzz&(tnqoS+1qXApTqj!W-U8_
z)e;`=Q2!N%56!~2Wh0mcWBi?~-6sYh;-A~y^LW~zSx#3U&cWVZIu@4U@p0TXuw%4-
zYlgJI+id~7TxE^`d7r=P{!A<)D6spCj#K?^ZFl!pE0uW>-pI%Zi0w}rEFgIE&7QRe
zFn@4v$ZM!I6)}LkOc~!x5P=X~3^ODj|0@nG=yWrY=~Z~wQ+vzNVni_e|LRQs7dt92
z(9#3#LvG*|l3bS#=W@C0bsWFD$1y5zl)?L+xIq`^t;}TAUg$_&v~T0^wbyi4wZUZF
z>-rmqG@M!fGbTAbTdU5`<<l!{GnPYf2jO{`DK+1ec=#85$pq*}^|G+k?gH}!4X3ei
z7!e!M_h(jN-iL=`o}QjXD!=l5u78ZQjiYB})%g&Y%(k1Jr?WR(_8N&Z4x6)G27ha-
zAbsa4K!BJN42QPnP%!#R<)uK&R23XijlnElUBmL$D~UV>GqPiD9_KDI=XV}d88kn{
z<@e?y{b8X#&#MXFNC`0n%Y_IsQ_{ZR_b@z{Dzh^1nj0Ja^y(UNif?S7@5T8_ndW1@
z@W4!+J^Hcz)lWf-@jnAZQw_kjT<3{YGV??7Jw&6C(8i`rgFDaP-mwzrEnG9k0ZGrF
z2weue<N<}eaFt*FdU`Q*bi)>k4QLmSaN)Qf@K$P7hoJ+zuU-co@RvDbXLNnc6Edf5
z@I_{+b6U9HTrWgG!F#_2Tww|T1%VSSdf>?|?-!RnG_O56P2~-_4QM)|1o;~#ThG=!
z6_X0$Ve^y1_s^a^18O@sH6fJ;1me2)j!dy}<vxtDu?qBPwYIW0)7eUSQcu~{=gv**
zzdzvQ$nV)C3Hc%dT=#MfKn(+p#qBQ6wsyurhNVf60xm#eTwLeXKj;x+%4G&HKR-Y3
z+Y4fLZG~(qwUW+BZ$+{ft-tNU+(B!HEJk9_rBN)$eQQtYuNA#ww#WwhSMJEu<6~~u
zgSq|LT7dd&=NoVZEf`E?0(>Dp$wA8bb~jWIH+Ruaf$!ip1m)iOfCkyLFoHM-XzM?z
zl;J;6B4tb~-N9*Jz;Hc}mcCr|8rMrut!Sw0wX+u23c0Y8)Ae*|cCrefT<@=`G!lOq
zlzP`k?0J%#7KPcJ9iNo{mYH#DL#eYvHFn(0JAB}RO4zz`mU(NS+{4>{I(E2t-4U~t
zD#KU6*<w5}|5AC(Y%+`s@6>6ar1AB<C-ApHxdY>9@I2_v(wbHNboXJTdEzk}O|Ol9
ziWyoodn6r_PWUt5NUMorioNCl9CJ#zP~K-WH1VaOM<)_;?UT(cCeIm6b(;PCVc>*=
zI7XJPCqpuP2?iE&E?>(n31vvOs<zc5aokTKYT-kYtIu!#M2Jm0)VYxmqTijsY(hei
zM2wz4Z`xw7cgAjRX%Aa{f$=Aumq75<@v%lF3LRZC4?JH~FZPMxSC4k-<#mKqVh3)q
zNZ2STdD6$iz(8@0*CGqz^&sbDDUh3(wDl1|7FcUn^~ZMG!hTG`tb$=LrVOiv@&kmz
z+8}dv$jx{>#JQ|gZ@w5BRDWRbk0J#sU9VD|FD4X;hQ|D#8#ExX#BS|{3?eX^@_QPv
zQt%V=VMYqOA*Aub+cG;HMB?_n!bjrdJmTAP34J3WiwH?2Hbq5Q6ZZnAJ7D8)kB&vp
zD4#$7;Y?I~gWD2c03ZFoy#PMo7{ui75A^T^8Zvg5eo{egcM|de0~j$cps3%We3*JD
zskj+>JqqOd0_If;v}Aazhhv&AFN?<(bf0uhbwQ5*yJDNG_z^q2>-#P-pq3zVjDPE|
z5~y5UHfAI{k!js~izHNY9(weyw@3xG#yRW{@OJUdqXO3W0xeWVy6<qU>DJJx0Y<R4
zVpXFxZ^G)E*YqPboQQc~Gxtc5b}qLm^izPYE(FKty78qud<?{I=TpfAt4AgGxB<h;
z@<gG@h&XaoCh<&Og@!1KVhW5aSLw>JM6JJK!J$6%Q+5V_`JX@EEN2XkV&bBz2!%gD
z<TCskd0TGqihMpgu^EWxP(zlsgPwhys(2;(^$RvRKRPw#8cJ?HY#g`cdTGVK(bb?<
zm`ZHiVyYSY``f?qQpw-U#|g)%jr&~AZV~w~eSPn)X}r7?N(EBbgdMP`<vU^U=HrMo
zQ8|W`--rS{T@~`^g@p8~l&f#B!x6l1jE<kJ4Mn~61vT*4KA8`lzmX#M#czGldocnI
z0%bD;1z(V#!=6KdmbkPWdAd1L|JI75SUZ41RH2INU@=IJIZon?<e+Tk6~T)L@crL$
z4BAc&rd!k1ze6DlHK`KYqqfat^eWH&=*RUQJ$lRVAICORCrN*)-gZ5j9Pr&n?XX|X
z3cgQTfV^Nl(PQ@Q9m8(ZCmL@4GcKDmVsihId<|O^p}QLfoZBE35Jqe)!2*DK%F1e_
zufHK<rE^)f5+|o$hm)x=HBg<KWES&B7Jw$Dt^u%wP0I)OE~3IlDzD{P|4D^$=@s!p
zi7Pix@p2+FxgQx?LL0TY^$%%Y?+|`@(5kGDZY(MiS;>nI*O<_ycTPBgcG#ii#7o3K
z`sy?6#<}NcVW?Ile}3`7w%mMRBi%Z6cJD%bo8CIXfiQ0SdN3lG0%3e`R4C<J-z;aA
z6}PNc8d86TgXW){Moi@~=LG?U{M843+E1ZTK^oua_s4&!?8aU%`Qb{#;E*n}Kfu-T
z0JFwm&*sp(GBNS*9v2hV(}D3+3G=4m9--AmBlpsLU+Hi53Nz`??~Y&wKYVYVt7#*b
zZ!N&!CqPDuMXP_7FDFzK6osjQXgvd)CDOk{>?31GI45j3Z(A1LnnGA_kGX=rVh3ai
zZ%{Z$D%Aa(RD%6As~B3)yxN3CeaiG(g>u$JCXA@?h_Enlh{K_tZ>9s4-*Y*CJv1yh
zm6uw!>m+aKO)L_o3@$ymK`5{P53(K;JC`78cvb{H{+D`i_V#XPFO87*TqLLrGzlW2
zPnJ6U&{?^yI9Sk3D)6=2Z;{9taT!pNF`W_J=m<7MNfBwD!9Ur4$g#*BTS5mA^-yx)
z(BOF?ypZI9vSb)bnHh%#QlGzJbQ+qzV1Y_ckDm@kDlW%c0t^dw9o?ff<heP6;SSS~
z@n_AzHj*LmW#4x%-PHhRed13h-K=Zn^!%9~8Jv(L*@uEG*J$b=KgeX6vfET&zWf-E
zA(ilx_J>mZFBQ0L#^*1{Kky)LdsjBgTt8^6IUwAVA!;_?-oE`l?Y`OH&-;ET?Xqdp
z?&Vj7lCIFnqwys3MY!(Bl?%qJTgu<;N;X-lmQa4yyZzb+f1>UnRQ$mJONn(suI1>z
z!Wmm`d=6_pumB`sAt7p+<R1#OP*m>7S#Z`6SVm~So}8SZAR~XCD4qt~-_g-gGPX*C
z!I`TuTQj^biTP6D?P0un!JC!%im$)3KH{H*6MBBONl3KJ#mZA?dleW;)ofDX^2O)7
z>Nv`m4@_7dl6lUm2kWMBGGxONxqNP%F(0HRO?Qq7jg7hAcSJvjIY2w?p+njPiAY0s
zX@jQN3MxU<*_AW2{s!EU@cK*%+xrZfdKRY_35f~OWYV(K%~8zhnL`f6{~xN}GAfJj
zefz#>1f&I|MWjo*OS(lGq`O19L0Y=IOS+{&x?8%tOS+!z_xJo?+$@(1Q3qyb&z>`m
z<8y`^3_beq4Wm$Sb`N(==6pbX?nnMUb0Vd7e8M#kWo=)RJ~C}y_Zolym2i@+nXX9s
z)}(!%hzCPOmQ8vkz4mi(Y0#y~^3<2Q)`x6QUexD2e(wTiPFQ9xKR?@Brr%v8oL{S}
zvnEK%Fq_ObHsJV~5EZGv%^)H|OXjqoD89Wo5x|pwg%l_h+46AqSGZ3Ou_7#e@$+ZY
z0BHQzgHjmU2>JO`RD!AXoZhI~EB)<}Yv_34rZ{01AtFvV6@LWtp(TBAI#x;c_rVWx
zIIB=WY=4ritcm?*%@}ioC&vzp@KG`?!~Aoxu1N>e+2toU{HB|3M#{DA*MYtYh=IN{
zei^NH7r=)8?a;SWX72dsL!)rE0y*ZgiEHf-{|dASqOaFGxA<EIj<1WSQD&p0UfaGd
z=0(%;+Z1>_hV9EF2MHEna0m$r&CJXM1OyBW3>ep_=PSCpxO@-#_mL_?QK#`$Lu0Vo
zQdn3Rn77o`ljki_k&q5%N@l|eIBGTMhpKayEHQ1dLSV7Yq9vX>!>DLR$24u~%*yKB
z@asaV8vfq2swiRZ6;xK0HcOZqxcWiRw8fBJ&R8WE)FL|kD=dhk5KW;Zp9^d-VTCZi
zfA?w%p_KCp-?T)LMz<ZFV`FiO*roPcbIDkDsY%b4m^5v`ASK*Bw503v$BoJh&?})=
z3EWj9hM2}+E9Ui#LUdgtwq2p4kYIr)TuHe(1eQGzS&^m|hqlY^gBTW?+S>%OSI*HA
zB2$E8(Xe`YbBeK_ZR(^=eTB)%A+gKWdg_6jaO9omx~?iB6A(HSm<*FngyzOK_$Y*E
zDWv(r3x*~V(AtV-Rh_VB<dQVs{me^7hM9QLUdc|8f3{>*gr&ws>Si^noL7&yi3)`y
zH_|`Cuj)2!06vO*wQmd+YnwLp0v7fxiEBk~89G11fScjEhDrFc?k#IOjr3ou29&V)
zhkc%h7~3sZTDX7@8BOb><P__OiprmvU(T;yX{08RV<G)jaj$INkW)8Gl1esW!hr3p
zwrt6CHWyW96ABCqxT)o)75$9e1HKq(;=3}(MSeE?<j0#XM}r>skxq6>8+)RA=!;Da
z+Z8a^P7VY>12b|c>-<}R#vkjP1bxvESw&2!?RU|^RG|g^G&Q317?-?w8DGMB6jKT$
z{74a@Bwta-kPRs0YbK~1GGNy5?lNFj5z0q@Pb%Qb$VH4HCB;BI3r2K_E_^<f$?jyT
z`CSVX0wxQCA2`herERAniEA#zoH{5$^f&#xsf>p%LbfjrV(ecq+S|ti9M*PruJ_tO
z%k5yg7z>hML{cyfo?{N16<&?P2;uCgq3sVJJ^-InBo1z+Mj5QoV*w4Yuq`$^>;m1v
z<d5D0*_4rkiCka<ieLvFDNw+_K3(;uW0c?7r+@u==J1L;oq>*yj)sAOhUNeWEdWvh
zI$Obnd3n~{!QP%K^k6(YHcyTS9FrxP{bW8iBZHwq1K-`fKvgNkbWoguZ2?cdc<by5
zn)+6v?q|2@MU3R2ERJ&(RdL?%7n)a&&L!1q)P4#fqGs_6O!=DTM#`lX@^L{2x?)b1
zlOv`k6s%T{$`{8BoxX*qxUh}XmV^X*1^rX|FmI3q8tDo(c8ij7(O(V8?r_uGw3xSw
zP*ql!M&*HJRnGzo^-*>TtXuVuvKXX@(8QQ#`8{;mW+6VtTr0iR`YEX2VpnJPVYV*F
zaz9gJLMeob#N&j?kY-6t_Er;)!GPz8Od^Gx9C~5Y-epY!8Vc;b=e$sU6q3;BirAu<
zYeML{oZ2Fv>BcZ%p@^|1qhR%n)S)C}VD${R0u9V!$PmY2&5ZHhiQ4myUGVl@k_(}e
z{?chTkr3}jL<e_-lY3X`8ra=jyf{;X{5iDLs#<Qh(GY%?11(6!>f^d&(W?3_i577L
zZZNP}hKO*dpNJA%z>1JzbTLKF(z;?N7y$%?nwHkl>1k$ivRTbMKIqGps<VXVeB=IV
zZ(p67s_E#+COZYnSf!<<!~Olg9CjnXo`i^KI$!;vizz-nK9S+8l(e+PN{g$E%<wYk
z1uiwdM@MAEe=UUW>+1`Iw6fCDCFSLR4zJ$4d4u}(12&Q)H&GW8=)(%rj52A0KB`7p
z{k)}HYsRjtAE`?QZioo9Z-*`upm_APqXtFsjG4a*&7lhG!KA@3!k5uwmaLyRl8YvX
z@@{;LSlN?|#!MOgsEO!j`zwsFZ&9$ge^O;pVsM=)7+taWjWR!TvKScyGrj`~1=xi?
zV|OtD;7+rcpgLXxX)st#!-fTtIZK9iSg+#AK8q>Gzxkdhfk?VAbZKrUnKU^vdZ`jP
zv1rNcn-wwml%diGLzXf;aln_qL4_teJHEDD{5E#B+`T}U-PV=UJaniF>@dU;yS_E9
zoWF7`zuK}0gpwZ4B&H2D<V5^+#%Z$sjI4mYUBnEN1qw8g<@OE^7Qlp@D~*jL3bZ@m
zN_tK8=Mwbx_C6aQ`#`}Eub!U1u_=K94FdHVCvMo0Txm;o&Cj1PgwUa&ppXJKCU#4E
z^g#(nEVn8rhfF{~CtoojC@3*54sv#V?ZlZ+LqntZ<qPG|wo`L!ULHB9=QA-?78UJ1
z-d%bwEp2RUfRe$y<@MRwFV{<BA)(Id0$Gv@;iKOHVczVc6r3*pzHg)<=$Ik>hD;Yx
znDW}epj~HdV$$5)d^%rKsB$ff-fbg>kQTaHLUsN79injVM2jiyvoSjCx`f~f@9q6P
z|Kr2?Xu*W3pPwHnP7%LU860n_&w*k9Nuo~VP@T!fxgE!@(7l?Q4y7+CqRDfM(<o)?
zr|8)PtGWxbbF_|mScG@6pK*{OEBf_PhKd``xcckA1Du-E$tBBF9uVmH`GAcUfYjC{
zo!G$wpSSdyP~nB@+w*atu){<O01UN4N<jF+M}gn<K;g3bho@nc<taqaCdUL28A^g|
zFCVG=$^CDmuJ1XmOn@#GRwzlPK*KDYTryu#omTd#%#OWN=st+{a#E`Ce((nh!_kL<
zQ^xO!iSN;r%X(}fYn-E8t5CdSJcW9MeOxtGa_&M<m2`(`^<2kr+pa6gRt|p~aIN~W
zY|~gx>W@V=&tW+u9J-}jO+P(<=1g59H5@@`So-BUpcwTGLjfCYx!_zQH~nO|<W2v0
z%+6NSenn2GLs>VYaeQ%`)f~GPZ1qiU5dy9+_y%$*EQ{AUWiHI4mZ`Bi3jaHN^K^pM
zsK*cQ52NhTJs=7Nn}h8+T6cV7|Ns4A$?g^Ee_vkz-Q&>rffM`#a7mGI$A}Ww{`VP!
z7O!)fRl?xuZ%vwuP^+I1ogM><Srtq~6S+mfsagjL#B8cia>?Cc1o_W#27gY%*tcGB
zIxN7env_75kE*bw)z?h0zRs$*d@ng^cKQ1$p0jXm4T&*rhV4+{be7!4!}0fhejlu(
z5m)28T$9%*FQ>etIAQrJ*9Q_pn`x0ihoPbwq~6SFUd&l$^<qOGj~k#wQ#WrpYL+_X
z1T06wF?mx9JGFI6z+|gk;jFQkcIn4$UJTA0TeJ}tA<a&rC&t<6^nR?3Q&rcysPAqt
zOyH2GmBneArway4Zpl%M3oP;fuFWk>+E|y+z!QEiTbtWaw5XQZ$IV4CIUf{6C8dNE
zS1QCcptUltNiNtwJ3G^uXAwe2<w(I2j@hcg*}J#MGD{&v+^r4Iq4Fb@n7mWEsz@K<
zvT|4$-T9RzXOSvMYAeU+ms`WN8x_jp%nqqDN_AM-_vRbD@p@APeHiFIL_9Lot65g+
z11~rZy<lDojRfIwI_e)9`a*?WqFPZN&xZy34ks-Ay8hb|WHfLy99T-VYv0LtL@$qV
zF;WSE9d7!+?<WSX2#rJxdHS^joSc^0Xfh9hS3Hq7@vc&iH=n$7(3z;_ip!NryQ)|?
z;Uw4O6XPk-_tco_eODijJsJDDN4<0QzJrCI`8kTkfL}1xYs+ah0;qTZSoGhwdu71M
z%*+g|MmS&#Cn2er+BaXW|M~MLFna%sTfcwjZDVI)0Y5Z69Nd!~|30M?d&Vv@64Gd<
z00}WMcr#qCmX?*4F15Pf0L2Bpv!Vp_@hTn<Xt?oG+c|1$YlC_uus^+~L6IiMOewf>
zd#bU%#a$czNQfWaYZlhPbsRf_(e+ug{xkM=t#ZVRREC@_dRqnEol@iFVt7T59A+&Z
z?*$_^as2^iE>)JDZD$WpL7=v3?1ym5(C5&nrrXs(y`;xS4;>wy=)Hy~dazSc#PVsJ
zHdj|Ov#|kyz(3Fh`2Y}c#K*;ftW2y}DP?45a&q$b@83Z5JH_hhH5HisnyhW**U8HC
z4cJpb+ZmXKJv=-B$pJLzh}qRtRO&%_^ytXazf-qnp6va508en_{t8-2?m*^XIDl!V
zZ7tY03V<yc95&j@%JEY9faf&_5<q*oOzba8_GVm^vF_)VP+cG)Kq2I;1imhy89}C{
zZBo&q$tlwJ|4f9z1#||$$4Ws-3Gf1{s;ZfQh*4QN4nQEP722R0EGHoW)OEbP9zYz_
z7eSOevO^{B_@iOA_lWRM7cDREl6hS{sOx&aJpU8mG&x$Z)J@7#X-$<3?ao?TKYzOe
zO+}C61jXEBM^(U+ic3uVg1x=H4PGRB=!8tvmn-fB?0jzhU{|^AyDY6!N~)GVTz$)T
z`S5V?>$xH_mG`a=C!?-{$E9n=22H^~h$G%-gsXPPempN*Z(h8Cq|M^Wn#WKAxnlKb
zVtd`im)8bO9Nc^$ngKq5510csDj&$nwc0!yf!_&uu-Z))Uk=y%M@QSh&IH~)F)=ZI
z5_n59fPWG{v<+H8=I=1H8R!Bh`C^2V6@kB7EnJMn^ZswSdBdeEoN(x1*`IUaXe|cS
zIXWVLm^Z@fzM%(T-D2noIVugttHuha_FN>JQSe}G7w;2d-d5zc=gH$;eH~nuuh5#E
zx17@GT`4Z=pMARkdut`xM*jv;!Y>LLE;(KUO0@UhZI!;la%(#qI|S)<U_F$V^B`?S
zVqD76@YO_@{7Q5CkrWx)u=$L~<;bFb3Jo*fVbRPn^iy8#n!R3hyRfmIn*Ye353Ht7
zbzA89nk+gtAfU~yma?b{rTV#GJ|)rPJVe_Z6hJu8v`d*d0Oxo9F?@9Q4FyhOe*pA5
zNe#7fL7Sq}Y?Tm^H^vSQ0-1~3Z<KLL#K@}V-X`oWf6rOm-hL<5pqroi$NVZPxTeJM
zgzU4MXndj!O+m3j{)Ncs6JnhdO;%WVFH+6+=5-mB9~DP(A-2VAh0cpidgu0f(OdlR
zf}cMJ|7|donq}^s#7`o3ncHS?2!&((FH>l%-Y;F*sUDjvp8`?$aa5eva^IS`6Y-qO
z%^?#jI=+g)?Z<*c=wbp|22iWh08Le(bgZea22T_U{3laW00g9=qXTsgpwR@FKVUG0
z^%HMB0o?*F*NZ%<{DY~&F(4kOql}zUZnSB1EW8|7g-0cv%ag=P9Z^wKv<E_5WDC_l
zvqh@pOQuK1c71~`O!d$SHXxt^j*-~dSYXpKuTl$qO@M=o+W<%&V8fp~c4}TG#>AYg
zF_8x{V35A7G$qV+eEOot>^_*cEbKo?TO#$tI6xsv<MWmEP(kxafoclfrU5?D@UXFK
zMu9}5;2aC@mc3ARg^aN=UJR+`!@~*#Bd&r{Cigl)1tmC$d5Eoo;qnNFP2b&}7CjAE
zRxa<_D|MDDS#B<={Yp=>so9c0u=e=h#7MnnZw9d`NAj|WFfl14BPglhJ9dvfp=`8p
z=#EPUIbz1{2k20;w2vOy@NsyuQ-gmqy`S$`thO-;*}6bt=Ob;)iw1VWGtpF?{hGzH
zm+|aR56;22j&wavu2(_xtvGEOyKq5|zhK2j);P<Xn(pOS;L+gvz*%2hUT$t~zJ~Am
z_g~E8z{>U5zQFXK7`3At>vM5GmXxm8$RZ0t6@pSl3ahPU3Kr4GPspZ<mcYixAG~b2
z*1ir7yh1fzk)|%$BgFhw`lUJ1>4lZQX(A-;VfPX#<<9L~u&UFra$e<I$O|*Ik<^f}
zK!J&<Yta#*j(Gl3Y>F8I#$o=ti-@UG!{DGK`nq|(FljB5OrBW0#^?N)U`cfQce3c;
zlXA>rvLqFSzrZ5aiC9*$kcGx^+CkQ)I$*?7(@6y}VaNQ*Bx#2bC94PZ-@WpCR9**3
zF(z!^CEb{&dcvkFB*KwytcJG&9f>QWfw?re;IB$*Qh`|v>zoz2kOHg_NFwl#oZiaf
zzLk}V9yc`E67~%@lJwhh89j&L>BE3XqCX?K`mQ8tmT7wP-E~zou$xcsC8Pd7Sv7B6
zQ0~^Q>n1;Y%hga}din4HXS2uBz4Cj4$3J~fezvb7@NGfWI$*Q+mUc^2yY1gQMk?RT
z%nYP&fGvc9fuX0T2LWL{ppaznxr2bb>)iw(f7NNcCS_r90IS@%hKry78OS(AIm%ln
zv$W%a?hUPpKOy5&zadt%D_qNxh7-jwccvY>A21SY{wjm~9h=&%x5EaxtCDx@JPO$i
zx}D(Q7PrGTZD16UTFRsjB$Ma7k}+bb%cE7Kx>Yr>P${H!>;z*V*Ex@t{4z+(4UgdE
z&9SWWF7ooSRa4u{s1dqVTQCL!?w%>r?U2!|v{aq1knUX8pglCAqccx>D#><nqqdg$
z?rlQ~+luY`MnM};dSn$^n%Z~Cs4&G2_j;Nr`WQiSd1wA5t-WGkREwIf9Qs@%J&M!2
z{gt>pp1!_b!^~Lg?D!X!YCC^#qNg5liq-=HNQ_e9Qs~8W3%;_=buuz-Q#WI?$8nps
zP4)aC(3)VLKbe1WvP7;m3(Jh`**0D-`SVdfq`4Wm*BcQ+Y~vqheq;Ix=1Yb92+qBY
ztmUCVFR;VXW&W&VZAic_Yy_vSBr^4_rG4!9B2>|zy0=gIef0E=Li*r{&h_#2-rk-G
z-wP-jv1s|_!v6xRYK<B$=xTbg3K^zt<;uD=C9rI5ZOiL02aV8!rGuKV<z>jP%ak=#
zBnM~!fjR|91%cRXk`*Ym*MayYhMWs1(?FXaLPbvx@?=KQ&mbPUzPyBC^0+;-A^2WR
zC~(prf)uMRGlkMOK`x0YBBkCqJ+wGBCLfh~NE<t1`sSf$1Up=f5R~^E(gbNPRf`ST
z9cmK`ll3~fWa-?ZX&?|?11{PjdtPyfvPwz%HsK{uP7c|5WOuGnVB+BaU{9!F`jM35
zdIkzs7^2HlDwk_-BX(UYc~woB;F<<vseRk*Saa}GytuMuWDsp*6W`K*6^7x<pL{>h
zg?owI90<EiT~Vsrh(mm^oxYGhL_3$5F(W|s7&QIEBb|R$_s8|cFHP+``_?9M$vhP$
z?Q1l#u3m|zmAtiQ*>=Hh)=wMCnUfoUMgpP_?4jlHDlBaxsE982FP>q@fP4aSU?7^r
z!o&o9J8SE5_4U9T6a7=`B>F6gC~OFwHax`1{b>LT_~6B348dFFpWq)vGa#ls>}w10
zditys>@A<55|^4HVlZb*$oB(nM|VW3l8#)l{S$(|XRADJ;=aBnDI@NCvEQ?}4gw;g
zToGx4V%%+WJ`^7%Br}%F4j=OjTlkSIR3;J`n9hmbVG5OXCfTjI%B1+r;o$MY+O)nR
z(K+MQCV`8`kEPVVr~4kDw|O5D?FiIzDfHMJs{K2o>5IS}Y|=XT|1O=vo`V{PyyM=!
zA~Oj3z4Hk%+wP+$&geVN=Rzu4TEJ!jnfxnMRNz_4FZi{VxYR<So$*Dpd+RHgz}aEN
zh_OTE=k_}HzEkT$ufKn~yE()t$;r2tWZd$Fw$9xfQxg~{W{5upp?+YQ8gYc-@oaE+
zY(TXk;$dra{ibLpn#ui+rfP1m5^NYm&dzp+c%h^ay3^_BkDuZo-(QYPis~1>vD0?h
zJ4MJd(9@0+=#=X?FW043wXHl%RqH>s42|pz5B`)2D>JYKwMy_(;3d~b?mDol-|<>8
z=FzmOQ%Ryn#W<plkAl6+6(mzG-r~T_<z;1$IK;qdp=DbhtX%FAF2_k&qM@f{!mB;z
z$toRYeBb)XR?y98M*MiAx^Bi{OxvwdTSP4WPj5AdT>O)7;>gyYB)_cf1g0<YSkTH{
z6MV-KFgT$sD(xtgZv8fs|L-U$@O<6+B5c3cKjASIi4=~>x0#twov66Fp0z$UVO*)#
z%}!)j2v(fN*qS{*-euY>bl<{75^*_!U?h~P>gd_idVTV6u|1%tnqB}cBO~UA`&<h;
z8OHrJZ6+(!lnL1`vEnIC`~6CEKJxjbAo;3@8Xg4vt_Rgs6(62UJ5oRpGCPs;h9N_I
zlOjq2eF{au15KQQET7;91^HL;3)!@L(X|3R0k0O}>?DA8I_nAhtl%@ctiS;s{|3la
zK7TIghaWF`m-wC=K0e;|OO80B<x^9oXCH}l=tR~>iuN$JlJOR}1UivLr%E;5nvK7V
zNi1l)W4?VGKp3|#WtjJgN_U6xclhDjF9c&O*vH50lHTEcQ_R=BsD2`nAL}i}TJKHY
z1|88UsaEi~VcV5;GO21d9M{-i)z6CmJ(|CCU1j`NBOuP0?CLCFIOSZ^e_=xo>~XPo
zHBoKnYqnhIskm=!KV#Hf^B(AmPD<jmIU-;c!g(ueL~*Iw+KhAXI;7`=|KZFr4t|=p
zI@{&n0L)xF+ty%%`9|4RzV<f9!n*c~a3fG%OP;68-j5E)FKzZ>$uHG6KE#2Wxi4tb
zo><AqFj@M1^?zCb#bT+)=Gbuch}rp4m4u<PKgpKY(EdX2qqmd}Y3NlNecCEmm<SiS
zJ5K$laPeJC56qOZPX-lp&1~EX=s5p6)(J~|A<1mlFW5V%<Sx?0eWgnNp+_n|X6jIi
zfE|N^LgEb)1Viz&pANd9$kI>|IaRiT+SjjEKZ3h2;&1hEFD?}d8RXIl<*T_%t|V0C
zuC|;p@V>*57@Y7<TR>{&ZH@r%LL9&rn&Xq6TigpVbBmvR4We0<4O#zK%z6R)MaAh0
zhK)RwPi~OiGa<M+Kg>XXUf}5WR`J{u3=U&g>CL<7jxMamo3m$xU!KSh*SZ~~wPE6J
z_&XlsJoWS0ViS8SSdc%Z+|1U0f%pqA@GZZGi=dz&2pXnuUxEAYzz-Q)--JUC`Dn+h
z;u(Rw8=bagcOr8)j8^3WrBc?DA0{E~nMKa<RU|S4xEzlca!E<pd2({2Q(;<(&0o6^
zudN@$<YJIvP~k&CjjQ->lE(Vx2qJenxLCrU>cDm>uoFmjSetYJafq*xZVg!}s`aX$
zjPIcR+)<^5UPHDt2c%N%zFhNaMpcJz$_tb{Q`Rs3bs2&-UC-j8?UY)Jo7u-5a9jU;
ztV{cAJ$YG1MEk%Uv~aun4umv-@khial`o+LEcpvo+ROu`Q@Hoy-*@-_D&rjP&c<{t
z#e+QQy1=Lx;f@kJ0i|5EVNO>`y1NHHwplGNK?3P2Fu5!RgsF`J1x%%MsT$hq*7bPD
z*Mb51mm_gd(T-kiiS8w(guJ0YdYgT-@mNsVQ+NFL!+7dtOG=x{r*6uznjSCeNb^%8
zwGt|3D~=|gWweD7>YIRL0wmlIv+d7yI#j(2;cE4Gd~Rgx%sXC|ZSIB28}T}RpU&?I
z*qfKHPn~yA@i|YI;D$58!Vf2FYM3@*ET8a!7IODJsic;4_oO_qYFT}epwg)AemKo`
zP;KD?y2Oq?TnKrLunmcg%_%5Anb$1{2|Q*g(E!C39y&V2hcq}Q2Giqw+y4_iz3;&7
znw94|FEsQ-%LlJ}3~r(W4xgSMe_6MdKG|W!#Gn&Gljv;i=w9NUUGe12WdGU|;`ty7
z(&%R1Z=A4x+qr!$vkj*gd4`pA^;V)!@KFP#n7KuWjITTLkr9!Ebp>7fMxzEQ_i6j{
z_9l>F5ClwV+5l*x4<gv5^NAY9|GDUsi|cG2!8xy`)0~y9NY4etB?_AphYX`s1s&Ff
z7#fNS$%!X}LK8!_hIItmh4&}zPYeieMnGxa90tl3s3E@)Ww4<I!BAN34#wepK?vXy
zm0Ea#L;IJ2tPr1@Cp9-GD*e7_Q4+jCLQ8f(x!BtIxSGGjw(z!5VZ3{arL0=&15PV8
z#<P`_d7g$NC}96KG+Qfnr6y}hZcq(wUhIN*&0u6svLtk;y3{Ue){IJBEI~j;pk8L?
z%xP)ya1lxJA$R{Cu(kg;Ls4?INpFZzCv~~@jW#|6qGQ89MqrZJ`m6F&s*e3dn`C1N
z;%j-!Y1r)bxdszD8?O9@QeXiF(#S$<ip_`CEUHq?u%A*t@L#_`Ab4!hu`5EB$ffEb
zSbY->BujCoB#~E{U5mg{y8e)!bPkokAiRvI#T%KQ=}XFFu;5=5pq$xhC~QW3f890D
zf{5BVuz7af>(kjkoJ55M6iN@xr)|>m^76H6pB0xV$o$9$9BR?UV7g;4grEQ^;qBGD
z)ZWka^;eB3xqY41eGAHGXDCC1pvHd<JYIBkGBWbwnQi_3-`fRRJkYU!{3)|LAV<X1
z|Lmfxn;?j2Ljuawj{|XwC@_J(5-3(5XlUNW3hx6Z+uCth4m39ICmPYQ6-ZP%U$^<u
zeAz=oDtpH8035wfkKZe-s$GO78DU*gunhBQ8NQl&?I;yhoZjq?z9+-%5#3uL^4Es<
zvC@qC2BY{&hN4KxlrY@Eq|y>@XQVSYn$wj~faE-n3c!9)axq>LWS63l4EKx>P%(fb
zI-QEeQn(n<VPe@{QhXQgH^lqPESE0g#BLa4-P!qu;<B!V7lk;qrjRrqD3=WLz&7=T
z^nID<;0OTsasCd;AWjU*HU7m*$kqW$NyNNmYfM_&e=quHUKy!GVRz0SRih4#9TD8*
zu-J0mA|Qh?9L+B08^goHd3kx9;SViNWh-1>3%%9$*@6t{G34i7T9&_>FS=@Qb6Z;@
z47XdbF|O5rhZDATuFg99M2atSJKn=RxL2Yb99{@Yx%M1ahnCXNcU1`#tRkDe-KBaf
zh%RDj`42w<;5Bj?ysp3i7aEG>E@o~3?rieb)33SI4@UHLDl)FP5dQnK;K&&A;EzEO
z9Ea1c!0-B1AT3N(8fxt~S5Hj9ePH-+sn8mMM??qX&0KkO%+Mj+^E-T*q+q9Wreuv_
zko}@(8&GO-2JIE>Ex|ePcldPv@_HlMQ>g)g9#VyLwD$^;`t-uAgVl_nW92Uue!Lo7
zZ&g*k_`R+~XzJ{5_mzkTD@pVfM0ap8YGAw>tz1h;kH4TL?4j|O9SjyFSY*$wtM%6N
zAVv9T(EB*ohJ}P|dqPJu_%43t<{JMIU^HY?<C!g~yV5{;H$e$C!rv$G_i+gB-YX^)
z6gR46DAL^w($lt9PXw*D;6WN&LbofiIvjx`zk%<nVm`akNj`vZ!fwCSenpWabL&Nk
zba4m^ZwwEse?nO~fA;JiySdm3U>$&-l;rwI4(-2SWCHs0KrW%IqSAKtN0YMr>!$})
zg(m-9^kr-dl@9A5Z$Z`kMZjX&vgd-tkkiuAYRbD2*vNNYS<Kxl$Brum>d8>V_UHpx
z2o_R6!!$8?0?!SCgVW|p%_JoUGK9LaR1EGgwc5UOk*psm|J@JE(uLq%X%wC*l&QG$
zk)uFZiCr9FA(qejt69%Ld<z=edbwf-D;|XW!8Y~!zeQMse<!B45!`72&f&-s!T%fo
z!v&e02JavQWqeiS+oxX~v}w+CcbEF3xwvX`L@4@Pt6^1Eo2x)IY|PTdR5!VhnfLRj
zIoN(yy<eVAPKfH4njGXy!q6^M!+K!eAC|Yc9S07#f*lGJ<Ub@bMnsy`K3E6JQ0=`X
zW~Qn}^`V99smkt*8#yo?6q%%S5zasVLG-+!IaOdY>q3@8?BvnB6R-DP?1#&@LZAuW
zZ?}H~kJdAiN6mZ=D8|5P*XsS^bvZ(SlW=ZI+`zNT&*|h;kpbYvC|y6^vQThyYmlNN
zZffYwBSn!@JRy?|`$`3<29dVGX!Ppc-1H9%raJXEugXyUf@dDo&oR3PB8F+6Jq`#o
zenq>lpU1G90OI3Q7P={`=?b*FmyZH$LTB}xBC5W|n$3|w&}fdGzd>Ilq`=BC7%w1N
zNey+of)LYhWKFIAWt3<Oc}wE6X?F<^?SD22iHqMD3QxLw2RT}Ou!}r7aXf6RF{0&m
zhJ%C!Y;azi{nShzSyRJ92~Bf%BOX7*uaU9Wdv{{=Y16m1Syvqe52XZdcT|`vbW(Ly
z#VkvKw3?XRBjS(M1T*Tg+uB9Xj=T3SjrMn9;bNn$Z4X{9FHbj%qv=8G?<);>j6EO`
zJ+~hfiUk!N+;7sDjY17x;NajyMMZ&D$E%j13sBNm=WgMw3w3v`IV1FYYk%ejS+Oqy
z@-SKw{kXa2)U;?ctcDexD7hl^unufY_iy=f25q%MIU;c$a1?MXU=dWLiH}G#>s!e@
z#?#xcw#|^C;IXBA15|bfbSJyl&w70gt6mnY_(87;ocI~hU_+q%JM6xmE-kcqHs}7B
z2TDo#h-8S34W0J5v|#bf0dOMSfKpN~M}P!*4i)ey&-@D54U!-S(1_7HMQ8A_ph2Tw
z1yGZqlbhkAMn{48+md}&%7R?7sgdtWd`KX%Q07tjc}Dt4Iw382l!5qH;|Ln9MrQi-
z=5$deZa256ecr<T1~4ds@h+0)p5XwQo^013XtN~n%jRWSpmnx-e`6>_KU9i}ZlA?+
z4%l7@+4w`l!T_^k|M=L`1vFaBMVozIZi^i0+>5lDD>O4VSDts`Gcz(jw%Tp>T7PX8
zz>Dv1<lv${kCo?l{Z{yIt`(OaCND3PZZGTdxk1D>-OzJ^xjrZ%EL#I#VT2%QmYEws
z34d`u@v9&S4Xt!}t5V^!97w=%b}Uz_7UAoUK?oiD(QE%V%MADJ{X1^>Ww)#xXk`9C
z3B-7O<I5kvaKN|&0l;xNq*c)yut$Ww^<cAiAc3_dr}R|QW(L5@L>;fYxrdrg=DCZy
z-7#^{X=m{19{L0n_y;q6c>}lGu*@tUHyka2{4(0bjI`rDEiylU)X>Xb-sK|*4=$?3
zRHQxSBTwbi#Wqa|=fcXD(Y0~i_zGmxZw_c3SZY4H5kiJDXWuSKrV^1be((9wgM2<g
z39;S0_lUK=mJb4_#5@@tGs#0sPe!)U;R{392P6gn5^n?;E8Jd>zW_HK%+~-;aIgFQ
zLZhLC503j0E1pnmhm2zpL{D)8G@!!=+*uAqUlD3819EbFo1Fpd%Lv$uH>YaGn;FYL
zY&ao-x(DP`R;-Hxj}0G!`uyuXdjSx5w42ESa7)=AMu5**eHF4d`MNLC^H231cojlf
zF8w+O3mzbOYD~tiudd*9nUhfoIiq4?gkuar<IL^q;7S2yduInY?6Rt>t3hX~J-VLl
z6BTw0`F`3dWwwH8!(w{PV+DXO;-P>JaBfZOzlFI8rKwqu0sHP{GidAuUS7RLcIGD$
z+qV-{zF9H%uXB&bu3y|RHyNB0vL+S|ut1+Q`l(4p;~^HWobsU4U4r~!sz#BP1{tN7
z(aq_OZryE58pdQe$yX;*Lw2eH{vPRs{KlMce{S5;<eEUHx%n@D_IbUWXQA>vUrqW~
zUKd?M!@L=DID?>NR9^_(9nd5LV;k_V0~@YBo-qi*DKC)~6d-8wNh4petHtF@yUrap
zx<ddNm4=UxkK-ps_l6K`JK!;cU!L)fXud*aNEd#>eopC}5x<4$v;*%h&o;f3EY|m5
zK@`=swTV|!eQoOb)p8~TNnI+e2>eL@^);}uyQ?h$2-!PI;t&1(t&*85OokC2h37t*
z%fnj73QIUXK!I1O{TM_t<sEOfG5K@7Uzpi;Im+YSHXnNbn#mB)6o1)`v>wyiI*I>M
z_43u@J&}B6sal2VLg|T07&Yatm$H#_-A`_whl`GPYPYX&7@)v8ox+iLwi)zE^g4kt
z0Jjfb^$!Rwd~aQE+`k8cX%0dqK?A{4g$mkElYO}M91!%-tP(Hu;m=VONXRDzN{^PT
zS=O*e*Gwz?z*nSRW6U)S1ZH5E)LS<5xJ{jsnFUZQ1$I~td>&N0GQB#`fFB<pFFaZV
zx=E0orGjWA!qff1#oXBYv1-P=tf<JR5Ep{?7fp`S$A@&ODI`Z2F@V##V1l4Rpba^t
z01RKVsA%2U-gX2Vz?_$7ljGse?R?SaT;jU75Yf*tAi&_X%`$j5c8Nyt_xv3EKqI*V
z&#|0R9Xz`aA>@*&BRgr@u#H+Lwf`g?`Bb5#!3))8^@B?<%dQ7mpBJBE7e@zlh;eL<
zciQ=-?N3=udYY}+M&EHgfBZ-Y-p+2V{7g(|V0B=Lhg+^?&rcS*CmuIPq8zqg%H4x@
ziKXBf*ptmt;QJ=j_x-jGzsa!^j$wrA3;#~Oe=Ywe5tY*c%0CFwac$#$Fw4WhaI($-
zyyQHl&(Mh9>0kY8CtQ9xAuVTXn@7^oQHldNgHm;<XuO$)_sF>2y}kTPOdmhmfd+&?
zg6hUd$gLjbN;)cJvHsie0Czrz?mQBN)7gpG=H>a})TU(G<%&&2ME@?vCvNMJ+i{`>
zgoOvz(<GP3J}O~pgeb&*zx`BchQJiRr|z2d`7(qZ57>Mj9L6GN0ZPX|gs|{>rzdZj
z&wIjf$0mQ{;5(i=uMWizmZuI=8xLWL{Z_+)_qWk|rafK^LV|<BX44(UYvCcXP$473
zY-<CepsNubczF1Nyu5FZH||ulo$f5-<oK!;GQ$I)(7V!=0DwgRm9DL$v+BI!Do&0W
za7_Y9NA1!PH-bAc%5B0gTNKRhNEoqI{gjQju?g^EKjY%!%BNO?CR>qZy`JxOBjFi<
z<q~}KEt`S%_bp(f0V5%pHv%B>nPo6f9N?8ZVe_MR?Op$f(C;^YNf2X;H7DzU(9wFX
z63mzatFjz4rXy|kYP$}~@j%rhbiD`)e6rgKG!t9oOYaE!;6pN-h1k|I-*JM~?chG^
zls;0>FrQbeE1G!Q;LR@@YiJzL?)jSvf)hv-?d3f*_~!T4p`WJY_ldQ+Gm0P^DVj$=
z;a=HfB*DU|05oEdfpu2R)qZq-(cbqKr~J1iFH~(7sgf+V^*(<hMf5o1-|_OZX&~c<
zmo@mTHs}8E2O!&cm&gSK1a@?6vNh3^;&NS)3AuwqVj;S_PI)T<%YksxkTtgrakpb}
z@rZSQv?@D3DXHHyLtrOKO-^<N=q9kah>3{-3F*}{f<w0A`fpc4o)qtm-Jq*g5LdT9
zobW_!YP)W~2@%0$x8^nWhJ|p%R&F*Uycdd(ADQ~n+U(eD$f{iVas_#V@m;LqlS+Yx
z=8R14yoP$fH4#L2Kd0gFd$H2b+uPgGlGzO%8@&lX1Q?Q8Oo5Mwlrl+crrCLUh9LOW
zJSm8<nzjXWdjL|nVfX`3apKyN;I#_1R>pyW{i9m(Wg`awnO_6A3Fwm0J&%>?v*MH3
z9NfT)A1i8Ufxe0!U?>7>IT5dm0>wXbx%}<R$T!CYXV5&IMggi|C<Y+NbQvT?VKCa<
z`jRW>wJ=Kkrf-!bcf!T$fukAWdcVpcGgUz(S|m^G>XH`5CN54^qN(}dAUtFJU7^FJ
zfYp@7^R`Biv_Ec01<Rv|b7m_221NJ1*<dC42vN4_dYeMr{4!d{liHIG$9|!lr1I|<
zblB0%fC0p0Lh^v0QYZvqK<GJ`#%Vu)bazQ4@W;Fkyhf!pH1N5dgk?~07zJ-KsQSA`
zkI}<;@|L-*m`Gbt*~`Rr(Y^S&Z5RK5VA+P5<?Q6!(IVfUSxs_G5IBMlE_|(O!vR`>
zqt5OIjH%ki<+vd8F7rCT2E5i<ea<Snd#~Nx;4Cj5!KbVSVPw?eRAppfP*#1-!BI4$
zx~cKW1R{tf;u&3;tTd@a6M_?V5c-`}h?D`R1F(kbS7hKJVQ@X77Tk_d<Hn<eNyFpj
z*|Wa8Lg-YCU+$kMQv-nU9rQpF_<F8Dr}>lXyuY`w@J%F1GoYtQ+sJ8^DONnscIQ>F
zKRi4LJf4#vChmrV_yokIMBb5vTo<Cr|0&X3?z?rOTQ$p47MwjQtY5e0V?%^j_VQ;J
z9(a0c=e!}no~;}zq7vwc8>SX<87I{)*S6d}<Zzx}BCT3ZVe^5SR;sr|nQ0%YvhBHk
zPpw|tKlkv@s)yQi+%D8VEmEpYTeMU@(2_WI|9!$}y<h86M?|!=X4{<lpz8Uk%anG1
zGzFdn?mvm~@oX;J%76C)j<nS1LCvG$M@Rp9t42>9>R%vpI^ca;22i;6x?=(#(}axs
zk?T`lKE7Q5i&bmmz1+(NwD2IC=Adu;lSZ<xciG!v;BO1*b)IeSz^iU_CPd^i0b+Zq
z{d=aHBY}aI&w(K`=Ns0l+|!i__E_IwUYSvh3bQ(LvD!6HG2_POrp|nWP2X>M1*Kqi
zD~v_Lj)396V+0($b?zqcLrZDr#H_@gRQltK1ewpweB|8RZfT0=J;VF`V-`EdKzz$s
z%s0+J47o=Ga&M)wZiyzKtmzR5DI`>!ptRO>+*^5nqVv1#?l9IpfHPq2wi&G$%zMJH
z+Gqh-5&x=wQ)ZTHRf6d+p(2#_Cuo*9eB|T`l3yS$p~i>e^ZF?@{AdNhcbPcr?Z}zZ
zATrt-xmXWK2Hu+!2Ir&Lc(0G^y=yFH8GP4XVI?{r&7F^~JOMKXZ6YSS6?TXS@OSf$
z0rjejtLsENO%fP&a<tOQ<GE#8YmE%S>aD%J0GdvKo;{r|p6dGZpNbi*b)X;f-XY)G
z-e&XkrZDzIgfzs=gyXC7SJ#(I%n30ZlP1<hOM1&2@<wP^w1d1h=)(}PNF>A8Z`-+C
zp!W`>u|Izzb!1klSIh>zc6)2n`r5DKARZl&_%^(4vA#yS5STVz&Ud^=qNGM_H5eN#
zB=U`a1uB{Q0DliCcN3$E{vG$8H6f0sfTc#~30Sz*?7H;O*WUm)H;0$?a-k02b8i>I
zad%zr*>-uAobt`c#uDq5C=?Y;G`(XOK3_!Q<Lqs4P>}daBm#gUtsfo|U}Azm*z4Qz
zPJfW+r~nf!?LVw#<mumdRc$jMr~zjjn><BwbuEx`eO{R9rJ<%3=*zgd-nIO-4vci5
zM127mN`YebZ*<NxvTxa*M;6%$z6-OOYrP6a%qf#uP=0rKK!N5KV}I{`es*T-?3_8(
zGT%4#{90G8f*4#9V8``vA5x)O3UqM=Wzc)|+`fM`0jouJP{?DqTF4TW{yq!k^T(<n
z6pnT!9Xi_7vI{;EiTQuB(lzRmo`fU9<>#(2c8IR*c8?Kz<n^UD_s6p2FI(%kslV-(
zKt_O!#Slat|H1f6gi4?7lIUvIYdxq6oGnBcPpG`XWt9MtlPKyy!nvE`tyjs#-THPV
zDrJP;L4koAYkJ=fLC?7LNT8p<s!n6JOw+kkLx7GP2-1xr#W=YoBqfEkXkp<DR)Qqt
z#YOKP+dQ9Fb-PURWAq~TUJEE6jwQ>mM+61oLfAwdzw|fJZlz;0?`O*3ph8>#(S}kk
z4bdluDs|)<*zeTpp<&OT+W3~yi2hL^r#Rv>GM+C|#y)z&*Dg#eX9}O};0*}79RO&G
z@o1V^!=l9#UE_Zblz-uH*E-)ub8V3SWPJ-v?0rR^(db9&Ya6(+%lY|vFpe@?!Q;FK
zH=|J<pD}XDe0k`K?2nc-KDxSzs!|@Q06~YN!h_<8*Frw-F76MThI)(H6TSE9>gs^t
z1dwJLW$a+Qa`qY^<qfnHOgI$E!kt}dV!e=7Ed!{EhvkxpDJ#!dCh}_Kx*pk35SUza
zUt`nOrFp75o#eD?eYX+D$cis(mHu5ZkSmXcaC)B+e`L7gy&IIUkbFRCLn@gZO?I%x
zi3Zv-o>(ph4IAO{!#)3@u9mPcbkjF~8Pw#*pWbUr6yZCvFC0Gc?*kFw;Gk@aU;kSG
zw|wRsIq2}$gbNp4>0ERNd_jg3E&Nw!&I0*rkBeyTjqzhJ^4ywUo~xsq>Xy0>*`GA-
z9%Y|s05b;^Pp_vJsg2&a5n~bjG2lcMOci|jD4WyE>dF+kX<z3hq+a)3FVU-hu*J!$
zx2g-2#$7=Vis)@RnAbOLbB@Ux(m4bB*ZtL^vz&mp*FvpXgcZ1-d$%cLqoWyI4itTM
zz)>IZ<Huw82XUG@Tg2d^q9YO~UTbp?kH_B7P|r;ISBQv7O?scoio$tQpDI^uD>X88
znFR=CP~RZk(=6Zr8p!>y(Jt8WQ|V`${<^$4YUdEa>gmz*RZYA&z)SM~u@d~(sEta!
z+-a(drI2-#EBN0A`=byApRMqzW!HYfAHq@3eb=nq0H4WyUy-$dwpqFQe6Yz9zDjWH
zYQK)NuIX^^f(6lWvIMEC9*Yhpp-xk{hl7eu#fuxq(^o$+uyKBFxm-*$PfykZ%EWLh
zOtx>U+1id>V(jLm1yqoMd?ON8OQ$6PSPPX5_1b^KX3!cu@if@XCiA{_X?Hm<j#s?+
zUu?6i(DK|)<?=A&p=A5RZ>ATMN|xT!dG=kecGJpc@;_O;=!*2|K5*#V94`ZeDLzTu
z|CvZ`yrPV&h$jfp4*qhCybMTum8Ux>i&~U>^w=#Vk&FQ>Fnb{6cv<zFKfWFul(IQ3
zq#rs#G<}lQEda)$W@j5^Wo0m*lsLWRJ`RMm+u+TQEh?MA+X(O;V@-FTw1jfmI^M8q
zu>Aj@+=CDe57m9Ub&)z5&7KQ0ZgEXkC&Crg97gvt#sOxbr5yW9S!&g)(Zan;7=(gy
zc`i#QSWToh#HhW+;^g=W+p%7om2L=-hBg4g2oy$$SWU-4&n4A++nz<I0<n&!E=SR#
z$}2Xe$wH9iqbyYa#6kT^Q;kG<Z@&tJX>S|Mc;%(xBqmE7$}aO$LMp=n@~0%!7z7MF
zK%EAM1VDxYGgga;HFSvSG6gktF^DQ6*)IktGrwD|q*c8Hje5&jJ)TrudgRR8E17VX
zvKTH}q9t-f2fH^@gH?4H`W?*BX&uYxLLlih=i_(KSYQ7npZ4W43cw4mQXeopJb04A
zLH>K`kyP^V76U1uEAVy6eEsd|(#pT~{rbkfE4V@pDmDz}S81|PkTW-;-2zg`Z&~Uo
ziz{9cSc&))+J24;AtgB{U-tfvPg4ItdBbC4i9_1}HD46(8B0zbJMe56@;+h{<adDg
zV{Y?jWqQ#;v4&z(#q0X`!sts%->*9Jwp)8)B=u*#)C9+wpO6NGI*;poBY62tz7o4l
zFs`f1+*jhq))obf4k%orrRGx&e$;aWnb5pJUq%cylRJFVH}RFHff{3~c&T(b4&aLY
zBAb%a<fiNRVUVW7judgJG2@5V&i`ovmJKIGAfSRpLrV+r)@ndZ4K)mgw@K~@umFJ+
z>^f6js9uDHV-QV=*B=$tIg<V_p`;qk13HQ&_ZHJaIWONix3;#j)aekR^*=)U1GrdF
zP#?*WDVUC+R*Ow`zCYaz>UP^eFPhoQzWewSMldE%INNsAO#_W5wdUN^#DQHR=Gy`y
zu^%ZI_!d^bF#ud54zOM`z+~ZC_;SU2JijI7mBDdh(52b#v9Lc_8c{RUSEj>O?VERf
z{0<g6R1b3YBOL>gD6>V$<@BBQVWOwK#z8Nj5}~#E^Ee(DGU0s}*0WNMj2kdyPd%1*
zYZQ1q>!KC#XfM=!1d?p8xq|KT=6!<{RxNw3=<P;0+;Y~l=^y6PS)Ne>5acfIC4$`x
zq%PBx-Mic{TzZ|h+wXD$H4cR!(C6nZ2Wuz4prGD#f{cb{aB9i{BzQNct2Y-H;BsDG
zT-akbKArcG$k&+$5Vt~(zBcVmJwL5@zI^pj6WDnT(rlnjCk{FN(JQMe_jEHeQ$ieh
zQ1p>pcC}UwEj}T0e7_=j_^9G0{yRB$Beh#Itvl5l;)v9QUE)?nFnGJqeQKiU6%8%-
zUJ2lSlj!v2D$jsvR2@?#M;r5njeD%J4urdDDP5my{*xF(XOx>)v3(*%|A*D99Wmq4
zEK?10)Zn{Uy@>Pcax;Ez#$EJiEvZ5O)*R_1o4Rz`YL$-G_EzCAmnN?vv$Ck8TU;Ux
ziJreIBDA(;p45^}Zw2v9;@xtK&Lc@&v8P&WX}vz@vtgZD>Ige9I%Zy<l$AJDl&Azy
z#REAb@@ui0RyRSBxee(Mn-*TvYK?bWTsRV2{jRuJ2Wqk?edn^8ptkjaf&xtUvuV5Q
z%`PT!`U}o-?AJHAG7W>B(sZnqZ-#O9uq2!mZQNdxG33Pj!So3L=GF3ie+VoK($bN0
zc(p%Zd^$gYItD%_rs8JJ7U`_9fE7~)JKh+<$bvYzMwZXXLBU=@6bM**A_hq5EIeZZ
z0pfL;XFJ`dZ3SpUDLk)}kE5d#DVa+g`T}$;1!cYv&T+UbxXa&>y6^^eNfy0vd`Poh
z_}|s+Ql_tm|MjznA1#A9@Ka{`dn*opF<BU&@pgxsBCg%jnj+6mp(0I3<?^}8R+$9#
zxs7l)?8KL$1cK)Zk(Gt|^Tc2Y3bu;vTRpY`sH+w*^zTz3O5<GHdsIR>{#%37ukHy>
z<{am9jK(~vZ?So^j21IBd*LOz*BTwn$BJ>wtwFy;g{-kt*JVFH5lmZT1=9SA*Wl4l
zS1DTng}B_1k8)|8zUmG_=%&+#eFbCQi~&YqRDt`+f}>&AZc%v#{_^d_?>wXcpqA3b
z8)u!Ig4TR`Sx=VFi_sfPl`%mnV1YL%KNhH#5soR#i;+W8r)(@O-+Qa%0n~xS=_e0;
zGU?oGJAe0UYin;sku#IVhyU~aaI(n1?aDUCV#H^1Rz8I5(&ue#sVNbVwPY^qba?CX
zK5nFM3~`9IzMZvST+UaO!xWeP4T6~0nYDJC96Oc5lj?AWhZFzZ7S>lWd>lGBa8FOX
zo<zqyg8<Ge<LPoD|H=J^`4D1Pm7a_-Wsdv7I<C3-c=2?tE;@?V#^dh<Fm$&wP<0QZ
zK|R`^^?{7$=6LZ&Z-`a4%hFFNa4=LrqU;?u5)iPbq@>U!4#B{{kbD)5*}7pdK}kL*
zM!Xm?{s?yH`;mH}AO_xviHU!}r>7Abf-xQTlu`VV!q>f;#zS55<e0&jVcb-PUzg&E
zSj|&A6{c8jP`-f)b%0e0;K>91{iX}G`*GA?o;gRS|9553<TYE$R;Ira>@|jIA0}zY
zxpC~@{fUV8T+W<L%|y5G6ZU%2JouuD1QEXJV7l3@?Gg}mbGsKO@f0yKno~oaS4Twd
z8zB!+xIOkIxpX?!;?8R5FEx0j?-{?o2@|AIV2C<ns?m1dJN*C`YzDt}biA0Hocv-I
zu^aLt=gtNm&0H4acTAOEm3yX@OP)J0{YXLR{nD)r=0wvyoQBY#ae<e{p^j&G{qgzJ
z5q+*UFq~e3Q@B<ubtGU{bUw8U!%QMR?}kU(b7`d0X0V`4G^av}FsmZJdqD}BCI+<S
zbJZ&Xz5X*SaWv%T$kqo`3$jIMh}VgXo9V08VxKJ1(BM#{&`_f05)1XJ($>=Dy0F#<
zI^bdbT6y@>;xx3ihO8(!^e^hWJd>R#3{e%K_vFk6PoAg82m@QAohZp&Rwm0$kqvas
zAQuDuTPq_YHq-Ged+T2jdO*t+wbF{6=Q%znD@5Ye^0EisE1+8=&u%fjX%`pV_uzWx
z=V`nV!IhljBj!l5e2r@lS^Jz}XlQ=@wDOcC<rv${@$U(_I#B-Qb!mkq83JIMzQ1-W
ztzRdla(KU4wSP?7C@ccN^1%>quwgVda_I$^wK;<B4sc%B+uMTzg5DInZq2)*`KC_~
z&*WSUUPSO7UTG(sq0E#6y%c`%mMSbPSdo#T%LLs5FOOKn#BwoPppTduAA`4V>4}N#
zi2x0$HFxly|BaU62|57%8y}cFu*oR}z!MH+mA%0@T}&vnFrGHLO84{gL@<<TpMAOQ
zHFU?Lh(6lq_FjeGg3-EPrw5g9h_3-AiIMK;Yi<)+(aK-3tJ#^qq+ep|U>o1lKP1A4
zpE@7GJ6(@gPOR>~1<w5CospnYx+G={ittpec54?W2FqnQs|6va>ntJvZ#B7L1~v0?
za&qv)AngPp2f~8WFD8Rgi)cJnp!mUYz1ih&NTY$uM8k#)Pz?53S|1)LoY5-#^Jvo>
zi~+UW1eC<fJ|FkU$;nZcstz}%)psf6j;r*I<m5m`jas&YIuQv4Dc3GmF8PBnrrqL#
z(R2lIM<~)ftrb?THIoP04-m$J4dD4~{gZ!RB)Hf6pT^vitm6lN?spREa-c49glMtY
znP6h=@_NgH^Wkc+=1fiQ-rfo?K6LliJ@sa(&T+W#uq;!9ycio=d7qu0r#ARCIr-1w
z-8v9G@<U~afVGaLQRh4Fp!cj_q?Hc_W@ZgBE&UYUrqQvYxXb=QXKM$SrFTcOc8J6I
z9?M6J2JW#A19R199Ja;c*<oLCE1q`o75ng4$dHO0_u~yE`y2XHP)75$A=0h#>GH*A
z1GW$WyED+2f#>}-Wj?WlkzCyVbiIf$2GF*GmM9P(#aDN((7q4|5)x9%52A(!=}7<9
zPnsi1;2}1fyZwmL#pJn{XuGFUTua*~^@0B6HJ0jMPdGrL1(C<BDJvGG+^{lL0m3ci
z8|*wO<N(WS*ca{<edLs`kp4vVvtR9h%Bz~cECM$ho2M;qcS5Bh+!5C=-50NZ2?U64
z6)9~Aptmkd!16S7Wy6d-d*WeZ8|GCy(lXVtdVbd6B>xz|=8g)YM^eOb)3wegp|K3!
z3V?u)koQJ?MNJKX;J+?X2Kl#wlFOf7b^(FaN?%}e>lh(WAoV@61iZfmR%{{%hf4rd
z6cZ5vU7Ov_@~o(+C=BwBxDY0T!747jrg<h%3<+3p?C6CT(Kp4u-J5vd)%)M=_O~lh
z!UF_8*Xj~nD>%Oa#Q=kBL#q<wPjVSI)zvaoQ46VvufAhSjvmaKIcM~-m2ZAfE+)w{
z0l;9;@zQZBMS-BdKqKXqt1`BGTp`5`HXbjisjFMfRc4=t6CefToKFuJ%@K;_D}u7u
z{M;Ov1KHRO^RK9TNOIUQF;M83in7XG+{7m+nV|5lkU=}D!2*pA=XP37qb+#3j1~ey
zn0%DC_SYAp{liOqX2O8^kY&Qi#Nz+>sB$3;rnYQ^_x1F=!NF+)+B{cjeof8A*cub-
zh1!rF{cw${rS9%L%VP*c*c5Q&9G)MtN=j(EdH+AQ-a4$RFX|dSgmejr(kUR_-AYJF
zOLup7N=gbyONVqwNjFG$w{&;cUHE&y=RWVfACG@nXYaH3UVF_s#~gDEs0cg^s3;=$
z`>~>d1A{j~dPd!y$~56|>FMdVw;wd7X}FPI<TTYY(*LmEnUh;iwklFt#h{S+H8g5y
zQjolR%ld<NUjJ%{lfhc9mPp7LD_~^=Y9wE8S-ZT}N39aa&<OQ1p9jclLWB^2a=7Q;
zXORxjFPweYbt3+p0LcHZTi$MMY@iZwG5sop&&e4m*11}fhf7kp*_+%e*GDA@freCf
z2axb2Cf(Dq*id7jJXv(ChC7y51m<*R?NWfwGHBQH^3D$oNY@^+@pw*yKEq&f*J6$3
zyXFw%wl6P=v>Rb7j4C^)!7S+O?<~1Wxe9~e23QP?m!dg?zP2;<E~DGe=b~wMa{Nr|
zS|L6@zQG5-+UUd+vOW9rvzZxgklq_(C3lsu2j9)vCFRkJ9w!H*&O<Yu5_L;3!LHcC
z^Deu`ZwB*HEb&un^X*_*S^**?qq;&d!13^Ae#T+k7zQ#@Ww4yqiySYsK_etTxH(M3
zZax9n6NkS)!)Wksk03&Hy=TUqe6|&Dya15$Qn+p95DNV;BE(1g-sT#ls-JQ(Oy6-i
zAf*X5c$|J1WyngxI~ZDMv{K{VnR7W4p|}w=F=0S8qjlwcVgph&&%*sApQ~72f<>nA
zBvPDF^bvm!eUkwe!sX$f8tVln9whmmr|`}`qOn<_ef*k=b_n!+D<4hXMPU|ovA~w%
zXu$nex13!uuQ2j0>caDHUR)W|2&=!+aG+8|g#Ah#r#|C&6L%sO5$?wV#)qN(GmI=o
zOz(%&=T5FwUa673uoWY#G=IHfZmefuaOO-K`Pz<>lJeO4I~3^z7z0es^HHzHBUncL
z_>DqT9+c0J;~l{UFm?NEn3mz3`r1Sm$49bgQp6mtlRr2~?^_bHUV8wTkywoFtzBOy
zjK$8U#)GIz22%3EW82Rh_Z+2gAcdAV7iiX<f|hKxx)rYTQZ6-8sJ7@YffKvS^Bu))
zP6SA6?~SQ?_h(5?Ca~90j<T+POMM_;@OC>Nzn3RhE^5~hk0ioBI<gEuJvRbL9O@XF
zFa{q{S(O?#fH=|x2$42I@_MMc11B9#`jHgt8UYpcw<?MMeajbx7#v3JV%6q*IZ1OD
zsvf<)Y4}e+P-MQWoU{P15)7n3(7{fzKU(^Wx+hHLci*U!nhH!`D&Eq=VC>qw1x~Pw
z@EYE)z}PIe&m$e184}ea0DL~4q}PJT#)`-O<s{a({{_`nQL~dqlk0=riSyGFp68M+
zZwdy{R_jozo12@U&FuY&c9sz(Hj?<%v3l(zdigK&%kz&{{2r1X<e&@deS8eHJqv%h
zd^b7Y4~F?oqA7rgQIc9E5^3kwSrRUO%HX5<(dcWv3f20Up;v?wh^llrA7chQ-@g=r
z_VZ1uV7}BJkDz2_t+7e)eox~#H6@!mo}igjthL}I)#~@eUe8lFnEPn*dcvr)akPvz
z=M>C<x`)N9*L_}$iJO=kgNEo=+IRnGI7O2CJEhgCFT8LTEq~R~-X5wQ6C3LQ@(*C1
z3t|;Ef4)KiK-2Y%jl0)QzD$++N4G#hEEjAKryqfSFvH0NO`JKvsKC7ez1>xiKnQF{
zaQQrXmaS2(PLm3a3t>Q7Me<7ujz3Gu`;JiGIlg+0k53ydVLn~@aFQHN#wPQecb;za
zr=_K3i$%~_6$ZDhlE(eg1Ev^63h(gwauncPCA!|6YvqoWGvA!;11W=iD+e#BYxdbQ
zaXjQ$Vk2O)DGPe}9$=6>`^$oD=~Ua82<q`xllW`xmV&uolat$2Ho>w1B6atI^O;|}
z{({+T?-9J<4zszccaVh^AL4+cJfCC)L3ziNj#cR&Tos*Cg7PJGn%OwhbHZOSNDQGz
zh9&QdX4;d^;zGl>6(T;_4EbSzY#K=1KQzlqfUWus%ua)3m%~8m(?PhB{d~aSZx+vP
zny?{cDrIuvw{c`xu}AgZ@tO?8T(&C_m#N|@qb8n(6|YWV1MSyJ?haw#4sLe<3u*DU
zU3Z_6TtEEKkKG2#f=7B_V^XDFce=j{;~TQ_`NZ4XTL*gvE6`rXc!>F<Ur?SOMUm~y
zdwS8P51S(X))Jv0S<-`(uTiG4dG|>+oeCjkx#|lm8aZkKokor2PGFdvoG#A9{-;mV
zrDk6t$8=PT)k5{ho2fD;jM(%{UqeFye+(G`p}e;CDT}pmau;A*BW#2O1e$*rqJgyq
zu+ux!&nzy=?f_j$$4Pw~oW9Q{#ya0|c9-hhp&t=w?KBig)%vh8vj5uGr5~*3Q@cv!
zTE3a@j|f)kgF`R)4n?HE3@0YtuA2e$Z5X^B;Nr^)AOFriZu1hEeN_+G%-iZC78H1(
zSsnHd4rMgC0Po<*$-54<8B1pXUJ2N>0$M6VBpx!^$!ah<O^Mb#klKpFovk&phX_GR
zq-DG-aqf_>jkwss6AAK?;aa6u)7K|cayt4R-*}^3^)4Z{VCp==ZB&{e6>H(QN)>Pg
zEHj@fzIoNom6VE1#tad1DkF@HEYhZef*@xyZ02Q`(DT<F@rp9&G!wjh2{a_cBk(r%
z_wQfngWSA>>0$NcF^1iI#`bTe&lO*zapC9(6JwjEyO6v4n}Ay<$cmVKVM<DhKm^E4
z5V0s#YmY|2ivkkGIMQ(Wi@o-elnfP^^aKloe$o*}va^&!hP~ANFcW-W@3%NzBFJJo
zGetQF^auI}E^NS)RmC+;w|s-)oEd<0P_@33`{&x`rz7v-Dm9b>4Xvnl;*km>pi(xS
zSJvK8JsMT^$)=(akz*ijbC!c-@3yM)S?M-TQ@8cocpdk|LLHcy)xqwbh2Du<%FK|1
zhNR@(seKh!`${nTb|A6N?Tiw%3q`Onk~P^YB|d$st*pGua(G@~10ObemcQ0e>$ShW
z*lhh?m*sCF#%!!n{HX)6Cpe^$;Bombq*$6><|x8Vj7)#K7rE^8R0}$V&1}9xJc>v~
zMa7*1s9&I$yoUfk&dM7drtt&yOiMlYDJ>v$$%POYN}s1>&+@rQl6R@UAGziL_9-Ys
zPRL$CUKyL*?Q|VcR36aaau1ZAA95Ue>XU6xmU2(p`wLm9`nn>=p+&3yN4^^_LW2AE
zK2J?rL%S#Q=}T01L&eVe%{tO2KY8lC>~KfQR?iT{&qoEzD7?O_lg1{e-om-U!dJeL
zm+g`Z>4ZIOTa;i>teh#bS@`w$<ZhyHX}sS+cD)3rBWSWvc`r;I84Yczh6?%?)pvV)
z$(kYE2dWh&NncoOQ@H>I!UcOPYREh#X*q0+2sBd0up+_G3Z3Wu_1|`GgiJf+%>=!4
zd;U>25RSIJupz>ZDt&-QW152nuIS;<c2$U<7N?T>FKg@EME3XM6u;S_CgWyXEZ0kC
z8t2~ITedS|UBMF(wfuT&v&+&tQ~y;K{<u=0(}{BLaD0XsOh?5}FLSn<9pGm4N=C(M
zX<+4L0j6xpyuFl2C9Mn5?C_@m2*h+}t;yn2a2lH4n%PqITk2>G0n08_Oh2^+zFW`X
zG*PGAu)fA`B^spyo4lOJVwD3a4U5J43unO4^LrKxdXK~SxS|#zh6Rulj1RiHBp+i=
zBqzU|53FInLlq*MSZue9i7?8Ifs>PDdp7R#hxaiHxP6WcuCBl9>TZBtmm$ln08(3s
zJjB*;KA!aI2p$d&Z!HHbFhM4sv+KjVHoaB;-P1OklN?~31XGSb&|-%(T8C(dD5;*3
zqIK;3OB?Z%R|=x|161Xn&voLzgitIgQ9|m7s9wwLxVSWH{z=1@rP(>XkOGqG@^G}(
zyk=a@e&e$CW8BW~p$MOLO|Et);}rr+p<6n)4=EcPNJwP_oCn(5^>_R_(r4=Luz-Ld
z_>ts_R$m0$y@NH8ki6f#%gnvN1u=a|o*C@O(t=}hL>zCzOMZQ3X67aitDs&*Qcfzn
z`)go~1ar5@-eJt_fRBjMQEo$^rAoqDTLDr~5-fOl2j~o@-Qnhr$IS`!!aXx%{nK<c
z{B+mQkBKVu@seKe3ih{z$g=9G1X(^JD!+iVGP5dnQ%D-B^v=?;uFf~olas#(oIILR
z+}(57;8^wS%M6@BpNkW3`nX4nEz2DB-De4i&jm>L3&_@szo!nUM&uwPe?DLs+i2A0
zc(G%MbmLRs1OQuh;V_lfHNU3!Ie{~CbM@Iu_Lf^5Bx+WQG4~G-h+p#aX9U7X5WckC
zm9m@>qtYDe-_ZqN#2So0Feo{ayd8PSX6CL5jm&}#dN8~@MJH>|tf}?f4CHds`c%Wh
zajHT&sp@3XBc^|z+5)eh*^W#;ozCcdPL-yQou?d@o2x({g$@ZB-s;N<WJACpX%^>C
zftgpKTU$5Zji~VUSOUDxCeIt7AM+PTEWmMdiu~#9TnE_#`GY5q$8RX1r!^DmP|UEd
z+$IWOYst-B2dW1hb)^vz3TP@5K&KJTu0>B3Z20XICKMdAR}1BvrbUMiXinvNFu@(T
ztj5we79AX>BanDHKz|Zu@bBaVqyPXufaWI9?gL!O`em_zZ-IfL+@H>~g%OgHlT{qC
zb7?eumOlGp=<o3lUB{i=3@m>6RQr*cmHAA6e8VL775bBc(KgB^oJ{i86a&SI3QW(w
zQp8LWyya26jLzV-(Xm%g<GRSZE4PVF+9vo!M168&>yz}WQN@-4E92hZ*orLy{_z_h
zKO8F4cdFZL#H+rUsc5iD@sh_f$8A=p!$kTR2n{g})a-qY1cu!<Py7rhTNns6{8OFG
zmxy1pm;|em`(`!6cUW7YLAdR^(x`?n1-%|g=+}TO>cyT9TOKR~;`4AZUDf}&b&#<8
z)3wW1B>E+L3((yG5fRvy0O9hhwq+c7`!2Q^WSJBg2=#jF8VqE8GcPyS$iyVUm4TlV
zGW6%qF+K-0r8IP|z1Q%PP2{825V|lrdekf7h^XQ8!L^fa2DKjv3GfiGT|+YN@NBHK
zIb1EySXwnw5>S`s8c}GxEgZ5=hk1#rTu;~#e-!cdr8RFNSrykV6c7`pymLIWXYY;j
z`cBH?V&!?P-|>27bNggfF)zHI!`E$6A5on1tkVSd`%gDbf{nS4E|niZEkutB5yEdG
zuLqYRcr>>Wli*%%;qg`B;_R68&UzWf7qoKB*pmN>e##>hfbLg;hPb+me)#a5^yNnb
zZu-^YE2z|l_Be}J1AR)F<Qxw^sqEDBv^pxzbm$e5b1b>C&vZ@lgIF{B#ucLxc@Nq{
zE5v-R5ScHnQ)rNk_+fAeFE6kDS!2_GKiv2?$-wVpqdx)Igr)$wqR|Rn5VLTne-%c(
zh4{|&U_l@Ok?eas&=B8h6MhQ_L@MewK9{eHZ=(pu!c{``i0UaqBcAR>dqsb_WWc3b
zXVuxEL%mCVwb^JpRzULJyh(r^)v`U;58!?^tHUi$M)B9KshAF^Dj5$ZJ4u%K?Gvb0
zUS6$ib~zw}Q&p)S01}RYzCKX72h#Yq>wjJX<4I}h*73czFHDe;k@L%i`>q|e8+Cd}
z6*Fs+g~?jGo`v&4qwVS!^%84tJI$L0oJAlM{Fy7j5fb`k>l5wgQq3W_7H)tAO?q5@
zi7z2tyY<Z$|6HspKl<`HS%B!2pS=F5aGW6kfe<DT?U^>ZLb!%X#D`6_&aJ8iqza)R
zC~2c3i6&lAd1e47=@{0ovMY%xlIt|ykf4ZB<Re}G*&eFwI}%e}q|WEA1R4@Isk}p~
zjgZ3nj*CB<Ej;~`*j1*L8-G@-0|PlYIKF*HnB>nCYS*u@?q82p95cX{gxGG`<IK&o
zvAgkZ`FH780~SZ@OD`xu!kHvJ=$0m09+=2^D$h($^wLLRlj+~E_Y*p1|8(mIKsKT2
z!;#;J5X%%++Tx51{k1lb_Bo!*aOIy@I=~IbMK=D<x;Zc#G&V$CDz8s}?eJ3~E>0v9
z0TtD$UsCE%0vnIlENB?6rg>3(6wyXgDuu2ucymZ(v(#9i)o=;6G>t0L=<nZ^T*j^e
zyP)ly3E~UoKXiZ60+vk68~)oL3X8ryqKZFVTyKKlFfcn|E&4O#$<XvV1(Kg9P!1R8
zT@MmAfoYxKpbi8F=K%H|C`>9%Sf%xEh!_ve+h8Mg%QAH`q1=sbC)rY$4W!$W>&`dp
zqLt$XAEdD5t#zih^#t=UyA0~h0-LV#*ZA($2++QlcTqs3OvR<=c5>3FKRSCPgY@`?
z?W1LfH?LDqLtWhtQXOKT2tX3?HC!*NB;#lRPV*9E^nC-`-{MlTC!vL2JB)KXb6oBp
zIWyj*<Io4?()tYwxUJ0r%N4Om0yR=S$K8pY?N+}n?e6y~Q;pwh_=uvCuvh`?>CNU|
zs_@SBMt#IJq%0A{Jv$U@ErY3QPW9_U%*gh35tGMxEq-Zk%WCx!DK1dxF96^Oum;a=
z?gjRyy)=sr!3e>r+4#_=6wmy8gjO@3Y(et380ShkG)O7I5N7a|0rfIH4!+|eVx(-6
zi;apBHXjiqA0HnB!%%zsv$+G_pm@r<K3KZ1gqM|RpoA9~DnE0okBm<zOE%UUck!ox
z)LvdC-UFB)-hEFpLEvAf=FVb0-4i>%XGfh<tzAl7Fqu2`%3oRjJh}MC)7vJMQ8xHY
z=BdBTQ}LSq)^+4@k{}QE;^X~^8^voeMYl*ciF8#|J00(TmswA>k9dU5(Ev?OijXyz
zaP%GkqTK{W^M}Gub6KxMz!#eAIk<~u_mE#U{#1#ekzPZ7RlKX))@jtd@jWqd8)xCv
zRaiJ7Gm~7S`gSZ~IkvYaFGo-@=Zn?h|KS2qLuD`oH`T{v!w(;O%oXdjJi5v;O_aD4
zWcXIwt{yHnpn8Y0sQ$Lid>YLLXi&*(#?$lj;%(dKg2(vAv;E|JVrZIyG)^bgO>cME
z>vzkc@Wq2-t))9#)l+GZ5^!|;?c6!SErZuBAA0<7=M*Lmj#+EBCKf{hdMwX_0b|;<
zKeu1(B&9f<R<T)yep?*vxB`0;%NS#crrt48$oKEqrXz14trO=zKA6vR_lJ{&0KFF_
zM{Mv=-n{n`SpD<KJBdssbPQ9wg@J=WHgTRNT%oB1Iu-C`zO<zO%$*rmXuSLLF-|fL
zP$)Sp(WFxm)eka1#;#o*(b@&CSOos6=ru#6VTV+)vc9rZ_hBMI;BoeY?geEn5Qfv6
z;Q|09afKPo`95kC5n7qzE;agdvBBK4g@yJctoKOb5g`J1RjARct2~yN4_*N9zMQ(0
zE1Fi-A-xUmaJvV-zMQZQ4q!mKWoAaz$LHxU<GY;vx*fIoUk+vvTAaF>rJC5fo!0Nz
z<rPp6CU(UFCAg?cz}g9}TL2*nDA(PekAH%*A$~41QDld?36##enG<exR7VrP;s$3H
zk^nNd!*v`jAv(eM*u-`99)<`J5mEW$CjelY;$;yxe8SVY*W^Sp%8pfVBE(;$8^D>J
z4^^Xp0RPjq^bIM}9GHdQQ;lv8?*l(Krh-rh+)_J92IFHHXu5H-CJMHNB)gAiXm|n?
zm^nCtS~8_0OS0Zm`{@bt(;S-Qyu*w8b%K@Z=Gd)LRK5%|lkPF1iy^be%0d>!%jeM9
zPMzRf(rER90{D)D!ivMru#*7o{p9rO*$qwHuD=yN=-e0LW8nfON|DRB#0EAy@4JiP
zb7`d3ph0!z*@EnCt2(8xA6P)KcN{F=OfUs(s(75)25O24(6*#_fw<wv4<7=6w%KG+
zLPTj#7XkpsfW|X5wwQSfANakf{6MkVTMPtjq!+9Pv>jcSPUW+fUr4@78i~RyKy##&
zjGCU`aXrihh-6Ar8W>xKH_bS*SE9Va7ulA&5}~t;la!R~15<}Dph4e*Uy5M)NjENZ
zcOm~U%A1$3H@j1WVt0SqY`<^Fc)JoEFEovxAlb7(R5w@%qGlbB0)Z6dm>YB|;@H<M
zkQ^Ml@&RdHm8UIu3t;Sd$M>EA-Z$93>rc)ES1uB;&P`#kD}NPhD31d8qN~7Rpsa(a
zn_mz}Is#A)6%$_=qh67Mk`mZh5BK-)jnfKFnmYknQE8PW48hWQ>Ed5>LN*02bB2U4
z43-HET~+B)oc^t|SN|@&e&$U`mp*uY-6~;5_Rjq?VKUQAdigUFF?92r{rg)L#|(<V
zunT@`@L+Q_Z^Yb1x}FtC|9x@2;}jg+FfaV=b${B?Q2L?E?9p)c<3Q?IdUAN`s+ZO1
zsll)j4r<VF%E<TMm($iovmF#t34(e?3k{5R`z#)MjQyBI`&xv5(TCF;uHve7l;hQ=
zZ#pdE>=aS>b&S<H9fQ*8Hcp!I6Y6q6S>|qy(_J##CxMr&os{&@GC+VT6Q;JV;4v?|
znMX`;c6(v5J^LEcx=8NT;`h3E>9<|>_}{l9Y;P`!HJ7V_pW~tw?<uIYKL02Z5;B1J
ze*dmfYh8VNs@x2A3)L5Y33*k&+a5#m;3uZNwBR9;#0)&)zkZFPP?N51&vBX#XOE@^
zt|e<1d)sMzIzwhXoInUu)pgH5^5sX2@R5x4M*7aPw|N-+P$nK8iaTxn5K$T{<!-Av
zkh3V;PflqvWH!ag?y8;SL*R9WtvnEvF0$EFh__-;;?1sCUNgVB7N()jY_~nAs$DSA
z5vy?2zYZB`SNh#IIuw6#sly0#JPsG?03Y=kiUM2lYzw+b1DfqlJ5~H>{2<8J^>(G1
z*aLCs(HWEPz6NxRjbSRpEPKcPFvJ?`E{_J;q+WMkbv2_T7ZNfuvVtOkj0|MO$hz5J
zvas0Yilk5Eyg<eDm-AlDkYxspJLmH0binGn;q237ny{c5F7E^N`RFequcL1tY>iIf
z<Iru2k{R;b*`2JH7jAU9y$SfS&}Ng2iqsqq#R7RSAtVi){HFU;zc*v0h~C_y1#WE)
zOAJ!yP(uk1R8}6+J`R_fJTv?zO38oMrwV$w;|SAL)izb0paT+=u)i2C2TF3DFax}z
zqFhlZ&$y(ajg4T7+c<7)H%u6Njqt;#{pe=|lR38dC?l)vW$UinSsEc<;gkY!c;1;V
zOMk?Yo+04yz1?h#_PMjv=A65knn@?Z_K8;(p_JPf9-B}~N`3EF$wQ@~kX9h%KBW+C
zWSrtU{maWE49j)ugt{?Fv<>~6?Ye2TLF~Bf#~g^9c<cO%;ZfhJr|m#y4z{fm3<Loo
ze5u|H7^wo|Ob%BH&<x<ge8R*ThCfO^KHoVuQxx#J2YDM9cas=r+sgI@DNDJ{5a@7D
zD&}Fp3dguFdcI6Zed%^BMu<cQ@x_=&Uee@m+)(ZtZ1`G;f8u^r&}zn4elu5)y)|XP
zA4;~^U*RpftWFaFc7h*#!!B)%&GCJoUq98Z_&sGMFKetGfw}=u|Cig^R>5%W>WM1^
z<6T^YgshCBg(^Zs1?H!!3PaL!K)nMgqZCC9C<Fk91r>6u<P_oRO6=*mdM${G$Ht`K
z3M#(Wbm~ZSIE|jTKVq=oyt$rDSoymTR?u=R<E@eje2^%@T6NxmMW`#}!-XBZi+GP=
zvPU>U98nyDJ(Ct!V1KpTaQeqn2tp4Lau4!vj%oq#Z*RlS-taZNH>zlT7(8-|G&(XJ
zsf`c4+xBX3MUI!*eHA}kac0hQGjcvfESqp&1N7Q##uMdPVet8yPdxAa;fnz8{73YI
zxG~3zl>t2)+h13&kD+|?c6wu}#2-0`KKyl{O;9dWDgb3Ym<Fv>t@zuO6bnSV!6Y85
zLI473hUv;_>tuwo0=lZ(2%8t6<Nz^UV14$?aF>43!L*+up4f0Z`R<Z`d1-!_2;>yQ
z&=9&JD9A)UXm9iN;2lCp3Qtb_@hpDh)U4Y;+j&tjyc3Ue^*YkPwvsn#9pi_AKp><s
z7flyq>D&&+lG*OsD*BAHCpEOSb1j}{LZuL2uustvSPm{<=z#fKF(m2MK0HW<TNpe#
z(L~Yb;fV=0I=X@K6Y%YTF949V1C^O_nKmIL2%Xs5v0FM{gP3TEn9nRr@6+p-FPRS?
z$+z#Z7V49Jxt@Sk1+@Y?jzH?Rs-gneDrdJT5`ZjhKc&@vr^Zmc_Ov!zr%=R%2yfGh
zYW+-x{=iaH1XMF)Q$y8lt@NYPnnGHXfXOk6I}0hFo?D$Oeu2J@L@HeDTs6PHud8iY
zB~5`TEXK3G2@!3JlL88&Vxd&PB6ztvrBa#>@6+Q02!fE6t*t9MvU}5cm5TXVFxiKD
z=?M7gzReG<V@M)AK%fuO3ihu1Ae7WzPC~ppPvEN-#7_wFrOWJ5##JGGn_(J(HCalD
zoI1C$;`|8ylt?aDqd<e}3Bs>%%hEwZK6VG|f?L^Rp~fo;W-gCQ4dN!t6?eGZGfZV*
zDEe~=VWZGYi?u<7keq)H|CUN!?Vo%94h)fK3l*2SirmSV8w-u<e4Uo&m-<i;Ri7u)
z^T!vSd#xeMz^WSsf)q$bOWQv@eB|G>E<n%7XagK10Z}(3D(b0I4<JSuXRr_w_Q{C}
zYzUyxLA9L2wA7Cj^gmXi7$3fU-i*9x&-YjVG`YPkQ(MLv;)9UTnP2Yc+OJDw*@U$0
zZs5b%#7Vp*)*?(Jv6s`iUQVw?ax8d(Ij39#rsb;5i8$QkD=@Rq4=QK(!8{=y0XH<%
zc{!8dhWc!(#%Z}cX#0YZOwd{w1xl(0A5P;{kn6Ra#uWfuK+NyHzEI)Nx?3$v$op`2
z)Z$(vJbwO-^!;X%dU4_5qV9S=7|R@sB;d5xoc<%l<YX_=>AdI~nw$GJCl?wbhmDQB
zlFRTM1KqY8By)vHT)KX+qZ+%Eyi;kZYqs0)S8q8PItXF2VwwU30h)_T@-q%ayjwd0
zZB1UhB5S#H7+-1H$@3D}s{PTLB2*tUGL9v01FL~S2IUe3{~jPObnfIj_q|k+@=LXo
zR5ITARywGc1Ip`kjnQK6f#|Z;V!d*@7$z1LV^UZ~njl2zE)vjD=;=QKsc6vXTI<%2
za@?CbXb*WQezfSfurNp-?BhO<4%Roy9h=wZ0#_NXDz~$&jqC>geo9NPxriqZ+5Ul$
zfTX8AE0dW9!PHn8jhk{P1SyNydre%_A1+iG3PY*ma1jjZF_oY;1`^T&>B~Pq5&Qp0
zAweLD`rI;;ehT0N9@`a&Fb05>fstOd@Q?yton|vwXg_}Z0Cfs456|Y-*6{9#05x?!
z5W9tBjG7A7zG@K?lK(D8J@Io8B>TelEF)|ruqP+V_~#R6fe^rHzS^ISO;2aUKoHQb
zvIhcrf{)k!c{`wJLBVDD@MPRV8AAbd@>wlZNCU1t1y~IL!wU3r)#z}9F~TE&NWO7%
zW__RDKzKyO>y*RyQArhE4d;jHBv$u47x;8YRS1>`Vy2?bBJn*KFlFCe_a1CsXf@7#
zquD%Plcb+N3%4JN;q$t6+|KM^a>9|Ds~rECbA6|ws`}>0F*7rh%5(3_jJ<K00Dn>4
zQFR5p3wk2++3l$jc;CUlKG7w@3(KtbZ8UcTZO`V(Q7<r?e`rB8@a;&N;bNku-UfFC
zSnPlb0T?9QR_R(luF~N^zDed1ZVmYdO8Is1)OvnxX=(ZSbL;heK3H!vQR2%1?@w&a
z<_}kCpb9I0iIuPua@3dbSO1ErUhU!dHWTw_Xv!DAuQ&c{VIp>_@Ws0NFOPZ2E@y%r
zP$(XTVfJ6gLj`xpQ)7!x>St$V0ktm#Xawj&<*0~=66K<JVcGKvAU}kKEo3VNl5NMM
z-ZdsiY)ETHCh&YdJ3dB0M65Hmc{*P0;6dI5r4%*}&hD%|c=9as)#G0LfIR#3HrF?d
zmQL$ZcMr$JWb2nG=+`;e79rJJqc8IS7HAxeGL43!Sl7CF1rWVvNML2<$dt=}dU45q
za^nd~jhUG@8ual!YrTdn#@*c@uO`7D2>|uO#~4q6WfB>$@pFR2gmeuoZKuIFnWn;T
zIrA(2Pv@?`P4AEmJXJ6+ADxgG2%<up-1+9-wWhq^5{L+w%oe&kmCt$~VQ<NGx^aJh
zzp`uer`PYhVMrwKm7qQ;2fbJz+;an>{04VIQWCDAaWnl`Yz&I|tL@XYbwhl7{6*&H
z))YvE&GbA$4fNDthb$HzR71*Sd+Y}T-$zgan!9{HfMN120oWAm!FfGb^!r%nnz>}d
zLmRx@h)b88wgg&)$gGtSF^pJ|gkny$GTxbpZ<sUCh=s^wu?x-QDi_r`obC#9HTDk<
zhKEw-4Rl<$ldx(zipQ;-AN>Fw_JYsJnMVKSiGz}crf^^jG=lfw&i%g{=uK70skQmN
zYw0+`BC#~S<K+~2|Gxj^ePb{Nbi}QdEO~-Mol*Lm{#DyU>)XnwKvw$cGPSAdFn@o4
zVKRVVbqM(7?d0d@gCWq<+V3?Lirb{Q=gU4rO`gjx-x{(#(!GA6qfIV6EPr488uLV?
z#sOfY3R4{YN$*Q@T8)p9gQ;ui(?eg;)Iax2H0Y3@75W_%Hs5vE%Z9K+d^KY8y15sB
z0;7#mjoL#j_~JS*cpOG}h^;N~uFr^y`jj+02_~LEZe_-X7cc^%B2_hlHL%kLAdtY5
z<O4|nj`JQ5y(J~rXCz6<?MgQXH}xWgTqI$Ptq<{Af3?$g7TiER2fg@11>EuhgbDKt
zz?6_p4ilopH?*oV9nBhUUhk_RukqsjY+{y1xEQD2lJbxZLzSI772$o$Ep+vnOokTO
zNE1>|ve|j)@O4T+5@{dGbf<FsZi#628xL{w(VxnF9{;A_0BQI1ayzOJN1Bdr5;0Hg
zC4TVIChMznDH|!VT8?BN5a8?=8D>VxI?V`4P5o8Yq9QoY;&E35R}{%;{^f`fk}>&3
z3>ES%PVMh~AEQVNMM6Iz=uQ5v8Zat7Ea}+%ATEwiA_0NS0KBitpvQ06;(WIYz<Pb$
zP#m<h7JDfaZv>FsFSvDnE`gPZ`icr90(l1b4+FQ0PXuXAeB5SuD7a__RF;S1%k?vg
z5ju?wCmV$xMYYS*zPE={)$%d@+3<v`w=e3uf*xwfm20{Dhgur8{wON9Z#Mf!{c2kH
z<!sfQ1^tZx_vZ1BV~V?9THJ0;LE7)5KZ;I6715a~$VHPZ4NUdJ^{??}{jJDN(phb`
zsh0U1PxuE-d<gja$EB=~_-OBqdJiA#j|!%#YEpxstbX{MH{o8&{upuSqbPSVRZ(R!
z9nU)jQ{3iPRJpCk%Ydv055}>^=I0yk2Nkf{-RGTEn~^Z;?N06i%dKC{Fp>lzV>zNg
z9$Tb)9v>v0D70a%k!k(1qO>7cqs@YYceAy!)%cvm{L%x1EQ5?3<X?uGTtg3^wtV9`
zl;XTSjG+>p)LVYqk4Dq6B9ksJ^!<Q=6l(Nf&CAZu%!OGdI_C`EoTtxzSR6S?#Shiy
zz+#*>i0=Y;)L_RT7KfIX>nYCD?c3?vSn7!~hUKgINpva#;EvrJfp21DglkR&^D%}b
z!CwrOj-&A-;CC<{NQ7p^Gicqew85~lR)RKLU46ZG)@<#(>-HFQv}(-fB`1FU3RhF*
z+6aqBc$_SiUK{l&T;(CS=Yj`);l%Z+tyH0)JBrw+s1P2zZ|<EDU}@94hidhugBd`=
zg()JzoUa8B<KYPw^-q`e{WIwr!4pkM$LZM^V-Lg_!5v~5UiEn{@5GyV@gOf@W#rjN
z{CuNAi5`9oQxOEn3>vpp-OGMaJHtU;25u?Ehgd1MtEZU|2mxA8c_k&fbjq05DJ%wp
z&#&4Cw~1W^bY5HCKuyHTO<`lP)$<*X%l#Y2Zx2QFR#7uA;%GXS;dyP|O~az40-EF1
z*9hQL0VJVfdIf%%1b`ebPSHrVqArHua@EO4v%xuHc<b)&Zf~-T&UCEJd-qRDDq3K4
zT`<KIdlTWS+RuxU<r1;Y%g0D+Y!*`Isf4Rb{yZ2jOHw5^^6sxBU}JJ#@5DiT*Ot4z
zPhv_Q{XbA~N~xFDQjxXn_Xyuw35xgHp8~VVMIO|_%BG^My{Ns-Tjhj^@7;2Dwjm$L
zfSkj8FsR;@k%v;@L+5zyL~L!#h3B5gXyOW8Z(^c87Y!$~TfFxd%2&dL=<cqE8;A>d
zJ&dw>fvn^F{2Xwb=H}+`Xk;0{M4J6u%&(bvY$TLN9e}!WcJ@pxGpH3CIp~lFJ_4hD
z-+g=UY*p3OH#av~SUI}2E#a1f-&0_EjY{=7>)idW`pIX0q|^IIK3|7%Gb-LbOz{;)
zgbypK=@>}&T^FwW#2NE{Jy`E@N*&JKe7BFk*X{iSyRznhFdawGUe#i-qJ>P2L4Ry-
zOLh{h$wO|uTt}<fBLCXD`_<<0AJ$qH-|A{F7t42v6fRib`K2MkOW4rwF4+vGO25UR
z$Wk)b8{hc+NVMM<Li);|iU3sNyXCadK~c~VaNg<P9QRG#aH8=#ds^bbOy!}+J{Jlu
zoShw^>_fJR(<}cDMDf{`fNT`oc6km!1M)wjT-C+q4;D2vc|xewA$nl*nUI()Jf4cq
z?&-esX~E(G{?Lf*ji7>{vn~A|9WB2p!yWMxSg~20odRhqg*y8UyZ(4(M=p@yyt#Y_
zf=l%Oy+J$qAn?-;78bTOf4OxL#@_zRsXfRnX?U_?E`V-S)o$7I0lKMg<A2)BB;LDS
zPSyBK)YLV<y-(i1A&&?|R8$ecGmml)z)0vfs!#=}P?TnkA)o*Od%ccYYoqpf6@Yan
zX`JkZ8nSS3fLvrC%`gBXs^>o#bH<Y6Iio}+^%Bo-VOi|z9+*k=tnd@IVSqE+()f90
z6S6xPm>NyF=)t3v#N)X^9(Gyz$Voa9gZmT&T6l0~J)pw^7=3K%%T4QhSO~BKnuzTi
z_S^($tih8>e9x=nBL(KaoF}4ZtwNO!j8H$Xqn@<_B^qPnZ~7HnjEo?C1@uJLYl29|
zMT-;hv<|4&0&Ko~bcj$-ZVwM0W7;n!%gv#I78u_Gznv_7OG_R=aO{0Uq#+tCM*~c|
zjyb>gRwz3nyk**_(!ib5bKRBWxRt%n85<rte8aUGjx-TH4(%Vb9}<nv&hT6=bTpH@
zQxf#io>46zsIv1X!t8hJ-BCtcP##L_>%%{nqCt8nIVNtmUlPh*M9klWs)2o^XJv|g
z?C%nj-8F8JDiDt4ay`}s7Am6fU!EbTFmFV7%onhBR|DE!0vvC7X?I3fXK4>_*iLQ*
zraSeGYCsh~h0lTZNw#R#M+lhihz9#Ve=a2LBaE8YLbIS+<?9QpOSZ7`3o~B(;X*VV
z1SN%cEX;a7j~6phx=TrwfRPXJ6|W8n33-8xqvZJEd0i-OFG>l|!AC4&vt0^Cy&Gm!
zrvnr|Y!iYv5QrR`X|+&{oLp<bpwx5Xw%OW2me^)V%%1gh{%+;(7#FwQND%5PLbT`g
zMZ&e(ib4rSFq-Yw*4GVz932j_jygESjK@F&ZT|me1Le#QD5ba?on02~DB#vUmKpFU
zZz$V_tD1o%8~o7lAc50`+a;=;#KkZ1J2s51s8xE~q&XZLHd=4cYuWPujhSQ|Hy%tM
zL5X2Kw%2@qBzN6R^s5;PI+8RVJl`Z#7h}`E=S=V?p3*F9I=Jl3UHn8cpKzk2<r4ev
z`Hj3rDaaoq(rU!Bc{wu|F%r5zpa!9fTG5Tyvw5mgD%?1Mf}GrFCRKfdgMwD4lnJNZ
zXEM=k9dA+D#~vm;?1%9M01@y*{Q4magdsqQ0b5#-dEZK+nxgzx!@}kA%m*_0=jBDI
zW2wnd3tAaTa@57!7(f{}jP|+7aih<wWH~@Letof%VoOcLU^UOL`BunDWh%cquS2Z1
z`lm`wMd@5)lHZ!J2x);F1XA*i_C^E>8X^RZ^^?cjdrD-i(XV6Ynyzp9D@3))9dsut
z5y&_>r%ASerBiIL;Xv^-Fu!3>8DF+)%SC_|ob-Z^n;=pwThBOJ8hF(fDD@3Qk%B!K
zSkHIq;}-h<)U&J8jFapd;4vhNrWO=TPijE`jS5gciu{p_vqb!&d#Y2Z6Oo1(e7}%<
zTxd}m9<lrMdWYkBMX;_U$l7PrKBl2foD<=Hvm_F<Fr9~t%@Aj+YK_$OYyJ$0a7Isb
zcZ*#m|B5@Ie+fD#jqoFEg9EM9q5Afg;a=~dgg`n4f&(Z}V}6E5$hb5q|1a7PwkT;p
zBJO;x)H7$rWa~8b2P`!b1i}?r8X4&F>m8G4yI;VC_v!L)qh19Pf(DeL52op&(!56k
zHgZusd(*>V&~p@~y05Vew=17}u7KTt`<CEOmRgAN-E2{b<y&qJmq<ZPUMe0_E&4S}
zran7ses{6GESsVWH3d{i#^UdCXEAQ#f3LXM&n=o)|M$GMwzT{CoskxFFH;c4X?{r9
zKeXwTd;TkwJMEn&^x*xQ6I7ob(+Rr<@xZEWHt#FeK<1A2=l`(pk!8}JS;t$}bDkpU
zyt)&pXXj_ouu+!SNEPuFx%J;M3`ia@(Rc)Tx~rzo50$eR1&;}tvL1%DEPi%28vgZu
z^?YGL%-_JC!s1Xy!{6fZbgN92s=+~uB}?35^4Bz~_D#>f@5ktlR!|O;S8A5-%{6BH
zw_V`xpX#F1JNkW{Kr<&*6k{;B%@1-L;OPUp|1>-}|1N-NyG~Lx{y1jYY0|HP58|ud
z&sfcFEd6Q^48x2Qi?xMwlf<T`-g?aKC{ldAvwuEOvF7=HZblYoq?*@NuxcUzX<(f8
z0Qe2NvZDI8c?Tr?<<|2<^<F(dH=U)@9s1w*e{OUcR(}gB>T!BARvK4Z;tLBgRH#&3
zri=`Kqs*)nEe3oi`Lu0qnWWR8mzKW8O=^-l*x3nh3wjicvn`rAzoFAc(>zSSyJQn^
zc^;~ULtMcE5sTGAK86BJ`WpO$MFtRFig*g7|BE386{@hpOdK3B6iGKhWC5_yPyjmF
z@9H+L^qYO<+ad~f{r`XGt^49FiZXvp$0{{8((Bg=59C3CHtIHDc+kBoG^*$Y(CREN
zhiFAuekz-pnE@yrE`{$WFcc6@AJe;G519AQujI{Jh6LJ#{^xv&kwYMTc*eZKQ$VLc
ziKS8qFRJJHb^Zs{2$2XO;L?DvhpN2un38AGt_PZ1VvASH=4l67uVk2+bvQympa43{
z#Z@Mto2VghG||MnvW)_N1wV{>7n=DTnPxaYE;GN}DG{|Qn$-yHidUIZ(eJ;FC39Hm
zp`WV<;$a2cNiEu3UiQ03fJpV<1|P(n+FYbG^!U=ch40qvF+aLcd9{RGJSyVK#kdnT
z5=qb6TEP4MIy4gcUwFiIKV+lnj}2KUQV+4X_>J}T|8N03Xo-8i81x+0m|Tjkriwj}
zouQ0}rSV3D;&D)34J_Tx<kbvb(63GprTgSCv)|YSEB#yWcVzBpVc*gWoFB^KgnF&8
zI65Bxge$*GeT`NcRQun1BhBbOv<g{KG9Mc~n^?Vw=2mu>CZ4Gfbz)s;^LxJTL~nqs
zX1t9a$DiD3S_SL?#z#lhl)1&4ZdN;ZsON9ekti`tgx1^3epgi1R9zYN7TS0AeLTlr
z_1f}CU9-E(H%~<M=+wkKOzznf9QLw78P{Y@AsCW0&`Uu@zbbuhj?L75XbJa=-;#2>
zyv(yXI?Ji{r)ON^D$?J%=f#gt(`!K1p;E%-Dk;Lphxkf1{WjXdpT>)aWhpb;GVWmQ
zEBJRuSTVutO@X$n>b8?1K)L2|r&gq`z3_WeW-$r(GlM1x9Y2gYuPw@J^i+p6OX0I+
zN_b?0+959D`Oy+!xtYIDr`?g3nR?)}P$Kwz_x3gNOwh&k^ds}VWn}j29x5P^8mC2;
zPgaX4t)<+5rZW)X<D&B_(3ymVxVTd`4gZ)ii3;_zjO*ykH=lo1Ah+hxwN(_yfwY22
zW1tIGr)e!G_f2>x0^_eO*8rW^FeuyjD4qAWcg;ZW7rZ)ZBG9Cr8}?u^$1;LH<AO_N
zsuW5-)CW1Ebn^^`NKoj3KT?Ju9lr<H2l}5!=4R_4!1=1sR9oUeNN`ee>6D-op&>hS
z#(!+8?`dA1tX7hh-pL@bh(Z8XR!~qNl35O@3vk@Enha6l;nFfPZ$FMrPfr6iK$1Z`
zSkS`W4U~Q#LlJ9`O&yw)UI}T{uSf)6W?Al;NNTW&j|2&)PZ?MbLtUB-??MqhQcM4}
z)#JivN4SvI<g+kS@^uo9L2ITV`Q@pCp|Pw4k|{);CORo(*_xrd=Hwjb7(-~I&^g8s
z5j77%7jz%$rwxsa;+{Hpc6OHdJ)nezRB);+(BaW!PK!AJrQo#I5+l5%AioGl<Ns7a
zllnl<T2GJk-|muEgwsC|7STJY58|^DjnO}T0E#w!d=V|vqrtd?Lcr!>Bhk1G&1~S$
z*Fjw`A`jwQ^iR{+4A%Q|97_6nUGxp-&KsUiE!^ycf!Z#uhnH(xp%kt<&C%$CWrF!s
zcVcga$uLp=E}FHn-VDO__-jgOZRdk}g#xzrf`98HINO9i?tc5txb}F%=daE8tuC!0
zmx3-OshDSD)YVEe+JaAi9X-Z({bN*M{JSJl?5&2A5*QZYCS?{H?7J?h(qKF$nO43>
zUoBm6l4Wfp8#{|U$JQB6aX=(7K2*+q*>H7cAj?ymCPeu|z<W4n4rNWc^n*xLjX(cf
z^IeVPpBZ3*M1T8aoS1}{yA52^+qIjZNL?TB6k<r<y8ZCsHy$#64vytEk!cQGGf;}t
zRp}L!{}Zs1{tH+e4eHtEfGg@ke+sm)fu<|P$Elskj`8uzEdPI>J%V-k2~?y=V#T0@
zKv|Ki04Bx0O`(KUC>A-!ioM1B_)OLa`h+()JZ$ps864+~{;H^|8W<e>pJyUqrbK~>
zn^Ryj=Q1hvZ%=u{x*7oUymK&Rvr1l6mkr2y_$y!sV(4$mX6ybY9xr!HAp|eQ7hO|H
z8F_321DHH*OGxWX7%Bwei|%GnOnq2HQU#PhBL|bFrSzlGQ*v`KHvzbHAbJzXHas5}
zg_Y-r?eos^#@G=g+0TD9ux;D^-;Orwg<hltrp1NlK#qqO5tKB)R<24H(M{cD(f9Fl
zT3S*=E^{3pe?q@<fzm)Ys*a_RFJsLwJrL3KD<G|M6+nm=0|8}hg7d&bf(sid5FF?1
z#=h=H60wO$=7^kNp5Bf#o{^<x_@bE-2}V9Bxt`HyfLCC_wgF7KVLV@snvR-U*U}Pn
z7o2z_@SbCnT|1Pk#4{q*QC<e2`tNFR31zwdzgM%|kt5%oNiCx8Ywv*YQGn(<nHni+
zg1>@a72b%c5a+wN4yc$MMCfag(BYpip!+(*e_@Q0G1FtHoTK|b@8@vFIRXBloA=;e
z++LxLMa(Pv4n5vg+Kk@wb|sWDA=<3K&=@Z|P0ZK`qmElHauS{`K}t<s(HTu$gST%=
z+rkJ&x}^RYj?r^xk!%-!ZmMa|{PMqUe|r2r{ARHg8)yl7_4f2oKSs>x%)G726iW9;
z{Zyjz1bSnDKv8ew)ISv_N#}h`FJAI1em#`phU{Znfe0LNiQQ*gE*1uz-p%|O0`}l+
zGBx!NatQJiz@0uU3qvNBZEz}|%bZ=j1T#Mly6#7bJb7-6^hYwqn3?)~)nZM>Ft!Ti
zpb@roM%aK*@hsY@!n?A=uq;R@W5YDR_B|vy5An;jtC{yGbC<^Bbz+t=XzK)b0O4-Y
zVIVu<?&fC4HPDaz44Z`-ga`ObHa+}l9Q+>SO%VKgI<mVEe`3q?JY$S|N3iGGqnk*x
zW$hX!Y$TEm^04ENKf!rFNC)PL0i~skn4jVs8ip^z;%fv_2sA0Q9I8c;nCvB#?ByhV
z*N2J*k9f^TKHeiy>!q-T1=x?mlERq^MWP5X$Ve|`U_@Xc?$a!Y2$?PsFd_&TBCsSv
zx^1g~e&ibRVUkXiNIyDd=1cbI%)0UHx^W$hVXqNq?3gvhPXP(3sNq<WWHE5g00b>^
z3h49n^03su2R_2nI$t26Mpg)<>Q@jyk{v;2EZCFg7to}CwW>s|FpuzH1NH-;ZD{*(
z*>0wFd}in8@$apkU;JkIZV9gKAC_}{!Kgy2>M846eWoiJ=)`mbk4Mb9yW?J`bd@jg
zCCyhm{LNNKe1QTKXse&PHUctCg5WlwdV1ijyVuoiA9i5Nz{psbC1(jsjSm^;dDl!`
z5FQj?-)AM{U`r627>Sx}0ur%dpCPOWg$RXkIbwV7M1Twig|MrJ+Din4a{=y5vt;*q
zI#Kzmpu7p}l2LPI?A%OKdeK~L>^y92?T^wT2qzy^*rnuUEQ*pj;=xUNr8CmDSnc0X
zB$Dk0XbVjIYihD-@p-;*kPP8zmXcgEdeMS+>hI<8LKPIhCudb*=k6s^_t_u>I5Jlv
ze38KfEHuOqWbN~Fp+X_PwQdNg8S_H&QsdY%X4K$Q=8{EI3g)3>sZ!G{ej|?zq!c19
zisL}0j2V~<oVGA-wCMxL1&8&D7kn{SKJfDES6p|MI&1=?+~+*G2D1SsNyw#wnUw?7
zxe(p_84MyEc*xq|s_WC_UDO0R5#34Rm^$_93P1|Ml(wIcP>kUXIncoq_4y40`HBHi
zwdcCnx7y>LQA1wRb~X<U?d@w$PqyP^=dW8ih~KSmxd7yh<xMR0WJ*6up=m2$1la8Z
z_F#-Gr&?S32-7dd@~~kU6R>2}>ecaM7!p#W6Ae*tuW}=X2E1e9yjI?^=v8_hSl;hj
zn$N4J*m7MESb}r)O(esc;esoTLCQh+R%#TQ_wGYEg5-NFbNa5VDmr=vELi1Yqs6F?
zFfuT5Lg0aCn0<ixYPuM^ZB>g1fgpzD`Ms!$7o+<!r2rd7rYg~&XUd*CiEW{Fb!6P9
zM;j;U-^jQjjA2um7mAwA25uB9M}S;li_6=whp_)LBsWEP2mqdDIF+s%dIx5jwZIDf
z-)+haOd_`G?>?SjJNxeWJkI1K4xmqJAb(kpvV8j*`U43GhWYyXNTK9LknCkWZ@zj_
zQ*4a4!0KRiCw7whdoq$GH2u~FkE{80Xx}`jQ8t;D17{(&5jjgue-gR+DmfQUpXotJ
z{Se$=t?q36DX5I)rDH!hp*o}u{EuzXTPFy9WhFAUueQ2}Wb+7|Gb-PMU!K;F%`+xF
zYuY&GRh>RP{-KXo&aN*68WG3O=b4r-{2(JXydoG7s1NRXnbvF}bW^+v6I@@H(96Pk
zJPI=Xy{=<_I&`?;7Xq1NW;Y9m#m?8p-C$%H5XoqHLNSk@Tb;_Q)5>2H8>_h!`swND
z4FBUwYNX&l2NpNJG~wjiJY4cslKqE0?0r5WBXc+Yxpx`-!9!l|2N9wjPketWU9N|j
z(=AuW;qikiE6+ynq9Td8bmOiI1xj#Q8c2VRY(sr}A}d|!jdON9VdQ|q3LPxI&#3-1
zTpi#NA%`uB0V^smddB7y<+(r^+%bC5hAo5EglG$&!86jFVFzE5^2bv>`xXmt(&s~L
ztzVpQi(vHObylU&j?cLd#wEPYo_*xs(J{6`{ff|Cvz^FjF0T6u%eaV$!L6s!=RKJ?
zC>oA%xkSHYxw9w!y@>!9uE}CvbZpg(KQRHI{)Art@d)tAK;G)Wk>t7N_dmCV_)1VH
z{J0H%<@OR{$|0_p9Y5Fe$-=<mTX+A?xDPyu85A#1=-Cm>_jGuO@3)%JqF>hUtzLaL
zfX|~{qKuXZ_sU|?wMP$Dr5pM$uX)H;oc@3Dczy!FrLU{2`>(!i0|_}<^z%(-c8ni?
z&8*u_DkHTKV559DA@4Drmox#BH(mBtwBm2N1JTmk<;E%c@yWTlqCHLp4_9ctVtOOH
z|KM+!AGY67Wx4u5`0w_0!FyoD8gJJTS8~KK62-u~+6BKpsK9L<$~arHC8bF|I)sp8
z7U8j}cucZs&;846(u0nKmPz~i#71jk9gl4aZzD|6g8vMzrt;7%+{53Ukn67NaD&J_
z;{w`F%k*;-O1KkQwtwCjbx^V;h_7#*$P#1iFMMeBRvXQHj_`s6KWsg>Oa*j+d7tec
zn?5mNwiFp-V+6E-GxVdk_La0eY|n)f{Fg1;dkj;ePp>d?E(<y$8aur)yFz_ro4pIH
zCqAf^k*=g#G3`GcEV}4o8$Ajh51$PTOj&ZI%3&gq3c-pf=13^Q;)@Cu>WU8YcG&~X
z(IMxwjq~HkGw<LZui_-4OGA(lB(V_K{YF0epBvr{9S-m<3v!V7y`?KpN%}Utw|6qg
zx-Z~na$0^jYtrYB5z6;t)l^LNdzZ%@_0>79{T?=%RK3nrO7nFHEq|^vLjZ=zv;ZAf
zG9mZ7S^gXEqw@JA*&gNZNt-@d^NmoSC`A!SvE@~5*`xxlJ(gD7SC|XD;HX2r`-eBn
z1hV~7f=IyuUi!WeKkLdWNwbS>B0tpsBJ}B1%I4sP!&M<65^2ClH6n84=1}@%zt=Lk
zy*({iX!nuIo2g>5Upday&!@+u<HDe>I5~WWetmlmFB0(35TWD!CHON2kn?|yLx2FF
zIava}0oGfhJWkt<+`M7I^(aCWUMvxWObd7%f^AL(F_WwF;Ehh4&OK%Zjr=<&EVhU0
z2k>#Tw(4^A;+{1`DM2{@wyiTW2N_XejC*e%axd1F^kQ|9%L6#bR0$P6I~we&e$)n6
z<QiA{@U+A%1ALENXqKr%06q$wJa%<bi9G!L#)?~-Lg8Rr_gN!d|J;2bUW8cLbkU$?
zG9d_&5TQ4fxDnm9PLqyW$S!!O(@5<+sY81g<__namt;faV!Qf(&s&PA?p$bJ8KpH)
z{i2o{frpF#IlT!zuCAS?T{i8(M`UF@z8|{kx3R;XB2(Ev4$fUWU-It+dOB%D$Y`za
zX}0w}Inj<6_@j3&G<V*eB@jEjGBT$UK_Hz7&};K!kd!>yPCL!ek03>BV7^m)yq2%V
z0#C@iuzW=bRUQ*_@k$`4%KW{jz}l}D?T3W_eUjq&lYU79hhN7Z2w8>YQ7l@<ZLVK3
zY6&BdI@W<tCUm<l@tQZ5ee!{f9L$(sJL8V7&qgAU2CyL`qhaHTbhdRgl8Zs-wzp}n
z{k`ejQ~vjCC}EJ%9C7c}`1#3uSktybW${{W!@2(FoNT(_oKX`Ml_q~3nUa6^**jeR
z%qbK(_yCTY458y1F^biwwv%a27uEP$68~h$K+BcsOa@kqdhiSQdQ3^rFBX0NlKPk@
zt}AEgCpd)QV|YQ}UqlA<?LO5<<<`KY5C8TgO|Sj*?~qmCi#l#3wr!`MLL7Uc)i}Is
zMQXh&=e|BK5^S)Bc8~dzY|C0S&~>F->{7kgcpB;7fc1qeYF@3nSwcf|(N$w+w#rz4
z@>gb)j$28NoEZYi3j5SAfr)%3MS$Y_CT46r7KK8jVl?jmsq3qwqWYq?hX#QW5T#?3
zPLXZ|0YMt+2I=l@kP=iv1SFO2?k<rM=`LZY0qKVK!0-G1_||&AS+iKcy>sW>bMD#u
zInUnb*%@Pk#x$Ign||{@W?PaxbQhN0k-D{?fQeu)&dR5GemcB({V(h?H2=+$+SF^C
z_U79ZZbk|Ja*WxUVNbX>tkb6=V?LNYMxH3h#%9khgTF_!jvXFO1nCMwNWVz;uiQKN
zD9UcLNSj7rUHkWlMncoijZh>cJlM{RF3zv>W5<nf*Y<G?d4JvOgvz__{Dvp`{(`6V
z8txm{VsI+Xo3?lI_Z$73zdz6AQMcKU*6|lAa^7K|vzsJU)tXXa2DI4?`?CD1EncGw
zpl_NBm5*%j+oE7pQnTFPwY>jV^e)NAQ8LoC#+Fie3`d*e)n^|t@UOabK7%1Dcr;_K
zWG3n|#<j8R()j>F-4HL>d*b8ViG`*8pq;WfK2eqx;}w2}tqRNd8YTJ@ryD5^Es4Q4
z)vvy@zaGugMZe&XmYnXBPj8N+mvxk&{L1lz`yr$ZWHqM@=1gvb998l!zArIHHeJYj
zL$hoDt#~dN{$A;V4j;z<S*$p2cv+)aM{D3HiI!7Zncz)jd=Vud!Y#!Uvq~tb%q)h2
z3G-^!PnoU~n&OKYwE=oz$wGmQ&!?)A<oko8@d!$6LYCm~0+g_!1f$RTPsg&p!H^-Z
z$N51a2>Fz&(<tG@h%A<W5HXKHBE6>pYeCTq=qXPF(BfCBTQ}YdJyr~GMA&=3GgIGt
z>$TIL2)Ez8&TW5!{3#{H`{Ky5a$3k{s;q1WTm1DBf|I2=lqmZxUeFhJrd7V(E1OBz
zU<PcTa}HrOIii8!O-`{~Rpyi|h8dq!G)Nk}`o9Kdt=RzEgB(DOIWv&6){YLD8JKVh
z^@va1JH84VO~0D&qV+D6`IR-6LYXRTZoJYh%R^wB?S_G3sL&|L#1@G<ZBaf$W+8<S
zIseOk`RD?y+8~%RNqX|=$iHhODw?~7+QcDgJ_mhmr-emDO<UIx2p$>Pwp-)1pTDLT
zHao})b*djzzz+c?Yx@`~eJ!bGyt+Bnm(zVHvn+Fd_)%z;j#EGMkpIZ5nMS69j;JO&
zD$2QD7qoV|-HeOWGfmXL%rbby0*pqAsyZ-|n=V$IDP|Zceea%zgfNr!I(i+idsRQ{
zQam)?$G01ko{hA!Nys+MJz@2JTK%`~+W7cIV(fU$yGx{6-5Qg0j%@+qGaFOx8=hmH
zWEE1YEu{Dr3XD~67Xk<*J}BG%YCG3SD}O#tHDQ8LIZ72TPbAwl(tPJt&!O%zR%fJS
zMxVPnuq)B;y@8%InO}%Xlk4UdH<j@1$q<9Ht=|nuoIf~lWA|OXwQb$aT+VM-bYCJ6
ztYI^#TuQm_`0Xw27H%xLEwaF{$me4C3v8dg_{d)>)9!6-FVhj-tQw6P(Bo=BZ>Lvd
zgRs+>_$%W}5OU$nY*#DDa@k?Qb>Smo>j(Y4=`!=pYuPgm?x{@3e#LAStrn~EAPo9@
zi^-}m;e~+2D&^0=p8AzxZ;`3O!Glm4ERTmOn?DR!6?}~sME$d(XwV?9@7;k{^Zi31
zg^8uXBbz^Bf%8T~ekS{&fULKO@6D=#n{Q>e$7S)RD;oOp>V@zXyic#H9HkRz)qpW@
zKC*UO;c$4}ZC7isgVeDc{64_|#l*PvWlPF?E@Q4@DwXnSoBL6GANICH&d0|`UOt+q
z($cLg^!k%5sC%c`VD*R1_g(7Vep}HgJq1avj%c30)qFdSdcC(qc~~I+aFb->(%x5@
z_j3ien;6#<=~{9{t=8`M9!lk9c<uqotU2Gp_52S+!c!B-67z!V*2$WnSJ!fC>hJ9H
zlD#*!)g2{>_f##)Su9SCC;vIGDe>n0$@jczH!#nYu_IaUTIsZ}zZpvw2T112XT&lO
zpcHJ?!w@Z=l8fYtxA_zpun;8^?BXTEnn#jYwex~jM_n-W*W_I1_8(J-hD3jKcMiY)
z&Tt$<N>5)+tzYVR{xqxa@CJAZNd*fujB+=*+ETq()wr1mJIy%uh8?$*&g^dIw9jT7
zFX8px1f0nhrlHHTrgeEN;+WjbFiMpXTp!1Y43sE;qi5-q4PKfmxrsif^4-9ZzbLmT
zG#}&3h!^#G-HITpUZn<pS_t+8S1y9$YK5p)3F2YfpPy)2(1#}d>7*~;63E|Qr2GB0
zU0|T6eX~h1t^l6eJdO|#sPr1qDUK&+?d}iUBA(yl|MV6$8rZOjIg9iaY^$w$ffpYm
z78eS8wp)_>9^P{7b#g&*d2s)Nl8YdEO#W>=@RXjTrLO@uY}9T1^v(Te2pwT;iUyUQ
z<2e<oHxCUKHEToZVv-WB-#rwHM$yCSIb7g4{VH-B&Ma7L(AXom!gfE7kwtikxr)TX
z+hHR_HYe9t?DHA99TZsH*s^lpPGd%T)}lgu8~;u)dSIx8;5F3fjdv6dXyKQT=aIw;
zxEvg};6si?OL88s{nvz-b$<j^;`lb6)Vn_TeASa+@JMt47Y7znqBYqeL%ghgl8Bcz
zx$UVRbr}Npv@Nk4I27u2)QF3g|Av0+ovyBt?5A7&A%%RwCh+lA=ahLJo;VJA@clE3
z2B4P>k2qWNSq&3%u>8G0pf0j!Guq6}A6-e=UOVs*GTQ7ld>)>aH@LRRn7w?#(fOuy
z@{xE66IrjbdkhC_n&RQf`OJ@tO0Q6k6tl!!Ueaj&QjL^pF2N~C0NisBfVqPsy0|_7
z0`Ke_8SJl>!SqnO<F~$d55T-6rBr#j6{kAdC31}<*6N0<p<m!e4u*8uT@4K*S`k|V
zIyP{Qj`LY{24Cw%(N%JI){NsFBGw8&UCz@LaNcPY945D>4q25iN0d;Vj-~H?)O<14
zq_joCKXEm!tF2mm>zCj@=249T0N{gMyge>b@6`Z8Mz0O3o0GkVH<t@hsxNRS0!Q~^
z^llgyLQq0+Ii)?NVo+p5B``wtX2)ZK@x&RviGN!zE7tQk8ggHspH0OG*cEQPti_#A
zXjoR1kdow~urm8Q=GNjD?2^Ju%yY$1Ymr~2pe%?Fu+u+G$M>D_Gd<0Y<0YhM5HXgE
zn?~O+G|u}i&n7!0*<P!5`JrT2xqLDm<$cq>sxAn<|IGdm<<3u&BE+ss<wX;g6oNl~
zK$C6#Hd`MZ-ZbaH4?oL?CBwv<++$3?SQyumXp}<a7la@HTdh5CjST9-^$-w*Gp0BH
zQZOON@_+=KYJ7+?d@*Lj@qL8T>6_&;{A&N%MX_kqHAZ17fJVQ^9~@U`o~AwArJAd>
z{z!%aZOYU!Gc!wsmEIO~lfBEd4xST`BzIzlY}8y0P>E*UkB|m)7YVet*1pv|y~$(#
zhg(k@g*MiBN+{;K_+VP8235%pS9;^*KZMh1RanAT&HH)L;FLM&@Z7u_NHY!U#hDbj
zX0f8Pb&M!g!kdp3_!*Ntgx{}WM}%-%-68o<j|AD3e~kIeH)q->7l=l%VGyLQF8E@+
z(e>b=iev$EZ)7#8y-h8CyHzKG{?pSCrP0S*zI;xnYAH>o=`GdOcG&X7$3sMlF1`K6
z%I_FC92W%lc-V9p{=&=8UX|Zc|JNph#1i72T*bP9Vhqvi^zAvgPYX>`wVK4R=|Vx4
zih)hamnSz_WMo#qxY5pOq3B{bGMk$+=mb>~W8|+9oLw(#H+Zk|W-F8<q7Tj8)b0np
z-)+-<ui9Ru$#KIU<MwJw)Om<MT;Wmgjq}K8bC782^#6)+^ODqSZdQk}XWb)myJlGV
zg_Uvs!2YmsRE<w!!gdCf33{gL2iq2JCXCeC)wS$?&KDX~s&Gl4>r~-n`}cfh>5rw+
zJM%>`*f_pWy0}}l9JA>;eDL+3*Mtx^`q||9R&23~^4X=G&#|rtr(c;O+Lbr8uF%M&
zW7sV-__tu{_sI6^7eYvYny<T2*y|{_y@l(JBg<ZE+rU#7ul2!HWC)%(7E;FFf%H~i
zBvedVv5Z}7@qn>wq`=+{zTN9(i3WOFFOeipr~O}}&`EpYN2DIa(tx!_8bTzYZ?i5~
zRBfO-?}^-AOtyMZhJm%Kq;4L2^{z10)lFZ$AXAg<(lZ8Ij<q-{g9eU`x4|m<Ox5Ig
zk^wa`<!=+ci8X(olj)h_a?i6d2-a^_x*xn^wEKy|i9QqTf_Ztmcj_S|G@P588)j;N
zC(d=bAi?+;l~rNqpO7LZe~K*-#Eqac!LI!LJw7*{J|q|Xd=@hCm>7?NkQfDB&PdKo
zE;h;FvuO{F))&Hc4Z?`>HX96B8@_melK4%z?WCw(J>#^EK8M}dc8dRF(}-vB-$T$N
zu^`lK#1N#MZ<|`mSCLN{Ba5Dks)8#JKksQt;d@VTu$(prw$#XfeV)00s$*EDOu7~A
zKpF=OVXq?`$|Jwbr*KD<jX?=s8J2O#r#Tnjq1okMIfMS2(d-1L%Z6DD*g_CI{Uymn
zOv8G^@3rkYKTQ`-*w!nlAP_1;nO860{v0S;$UmA&WFKCfyPlw?=+H3mReSsPUAgAX
z^`5^0!VdFFE`%5deek9d=L+RFsW@i#7$|AV{gE6Q5{jfssHIFJ>nS<1|1HKT1TO{x
z`zyB(#`c@4Cj-vAoSzR4nIjFpTt}{GfD_|6Tk*n%Pk#-!wg|h3F-M(Gw=^%#HO-eY
zK;j)jo{3>CZ<)zsE(e0G@jXj6wG~;f9~?4PbP&^`LIircck=W3$yz?20+8F$?xw|%
z9k`$4_KBJyJNdc}8A2B)gTQRJ^ek&ns#x8!tR%4Yf$3*EorsWU?)d(%=zTNxMPo7!
zYKCptJMV~2f*%qdhl9K*i=`D62a$d!huu9r0;vkE22rQp&r=5365k*QUv3<Ob&q}R
zEXtt5Gii8<rtA>8%s%r2lQthe^bZ_4&y@nuZuWL`Sdpkh#=M$aDUPYd0-fR=%&o%P
z(&IlF=dP|r>nt{Tig|rF?j!SRxjKB3Uc~T$Bd<^!dA>rd77~J&GU>gZ@jyu&`PLM^
zR$D36+Q<rz&b`$I5-DewtU#>7uC~8o?riZQ7P!a{QqP7Dp<xmT>X*dPQh;MQZy}(f
z0`0qpw}bCq;EQ)8M6dFtDiD1sNv`_+02<5ZBfHh??^8TD>TcSz%?QEM6Ze8om=WxZ
z=*p_k#<F6}>BYJ2aGG7~KNQhPdb60bAf+m6gCd`4xM}@HTl6qDvq6|&%`!M{Hui$z
zX|Wfm!K2}ZxESPJ9GjR(6UI84t9*$<r7&zKdf8QHNH%N?!J+==<%zj;Y|>roZj+c0
z|8Iv?VnIMf>GV~=;V&auMiZ7V9DMK}V`8@kb=eAx5vBgsWkANqXaNm4mmnBSeygHD
zZCLU0bl165#qysB`GBj3{K;0Zaa0PO+(sNKeXFm(&erOqSt?QUXBxL0aPIu=jfU<*
zcMXP!W@$y@WV4`)hb6c34(}x-;e!rbq&E&Nj^mRV3@Tv?3(BinWl4(7`s^~XJm&Z-
zD=RiUDfifB8K*ZP;^W8C{QMC!o<V3$*|0-b&&mo8><M|>yH+_={it4r@+GANZ8QwX
zzg{SrItyI|OZA)61>w(n4yY8^G0fkm+ux)0CbS2wJ}_*@$AXn)tDJWWODqhGuS7kw
zs`+sg6IhRX^Os{a)+Vb@79U`0v2g4BuKDA!e{;3^!}_fz20)WaUw<L6Gk0EFsGX9V
z`L4y~am2&Uq4e@mtu*={@3|K=q7^kJ?i3AMCc(Aaf5*|TIu<KmwWbwMe(#zZ<|Tca
z$@cD%=U~$ldlUHqTl93!nVmD=A&D6vh4*((vSUL_^utZ@Ad|a2gTnYq_b7mos<WPS
z^1QyOg9(XAEf)j6!_6!+1@}bbafpB<EYat7*E-#<l#m1kVz6JLTFbsT@(gOu_lfRj
zU-8_dTrwhmnZm*W?DhJ+q=K%V!g~hOUc70yt9Cg$*Z*}IA9wSg1;l51I+w_8UT5?Y
z`Hxkv6y%%~Thg^m!iy*3uSxEZ+#OZbinY)5cs3#{WOkQ=+iL}nnFgACRogp=4X5ge
zQvX!LQ0`cg|L_-<Oc&yXnZJ|*x$3<O6##SJ81dE4pphn*uKLk@8J&U4G=e7#8{`2d
zcMd40$(;7}6^Ad<vs?rTdcD5qw;<voKeEWqR*GI?Jh|<V5)pPeF8iUte!iO$8TR%Q
z-5vaKZ1^?<z4E*MgC@SKfA9Yvoc~acb?G-~A>LtNK5Z^r8|!~D{`bNU6YY8G5T{?H
z<*`$;e_NRut)rl?FM8fxPb60>*{cuzYW24h)3)l7FTIN8v_2~9B_hXvmW-WQ*vY3)
zDyX&PDEvHUu}-1Y4%fno)dMVR=){|4O%G@N7kD=>UD;h$&+Qb4JUgyW%_|jG`+5KH
z$okes6JFw?LgvBx&OR}iZj$XKX68C|SK-^8u$W}@`B*!U{&mkIvGfVTB)Le$jg}iN
z^W3KpV)+gZt()66PrdFJM1#uV?xfM1pVz#tl!x$;AV0Rg)jQYb$lo-oE0A)NMjLH{
z$lv7lIrbBZapW|3=dsCqqTxgGUP~Ub;ls!*vD=qUKFGB$Fb$86t8pREU)P4w%DiN-
zF2m;*&Pa-bP>`z*jP{_^>+mRh-kLVUQ8OOahdh|uDQ`3{EVsn3i8{!=MZ6B&{q-tp
zlfM?{^9lnxz@KD6<r#g9x|&*OcKX+^#JQu4GTPtjIg;e;YIXiI&8Z5xgO59hb<)zl
z5GH+LgM3Vp)DbqJXO!t<GK-3iuAW8A=%(pc1`irO|0lHPA}|gm2x&J2@vfO7>eZDM
zD{Jd}yUDuKKib=scLkHiX1lg*$yr)r%tfX?x?SzHPCN1R`8Lvq<QU%=r<2U3Tpe2L
zG`S9k{I)6+V;VHW#^hR4x7AgpeC{WlK4TR<o$e(a6E|a?N2%Bi!ToLX{I>&7^sRXH
zH+cl+%aGFq`4mx;)kc^0^8IxGuL-N7{><{^m&1~Zr+-#PFhba*BBI<CQ;c+L0+<<J
z0N@Fw*xeqNMkoPz%c9$0Ong{+Y(fgd6ps!f0r|kAM}>*lVbG1q;!N)Gh0M;t_g=ny
zV3IQjmsaLY8wu@E@DEL%(n0XtY<Yc!(>;W&!oX2$@+ahZT2Eh*dsLp5lz)KN2R+JR
z7tKmovogOsN&=3aADFwjDmHyg`fgzx&3LIwso^DonHk<5gu-nNBG)CBoPYnSzaot8
zC))Jaf0*3lg$JH-0Tw<b=U$d4!||8b^Ii~>1_R!tmNZ2n?IsQgSk_x=I#8p2SZ%J2
zK8UG{aum{V=n&fT*LBRC$;1=iNDe~75O=-OY8F!6HJ>0}DNA|cSF(7(NvRK>Xb14c
znTx>C=#HAbCP{q><a>F)KJk^|@|<)7)5HmI?Rk`DTwXnL|DcVj;L<LCg!WKiqKfw&
z_ZRT=N_aPEydF8iKWJzOKK7PbMHr0}v~kMdi*GQ!Dp)dHDCmgGrBIO|BeP>#*veW(
zOQ1xDP)`#61^EczMZtQy!o-)y@iT&TZq)p&WDh$8^W0N8-LH*mNJEei+?*P|mT&a`
z2c%Dz9HmU&)-z|j_s7nUqojG;<=J2NrW4id5hzAf>o|GE!U*#51*ld%95);ozsDIJ
zKd>s~F43fM-S|u4CRq8ZkOCKU@7LFa<V9Zn4`WiyMckv#m+1YqydB<pJ@Iev`Mmmu
z)MCMPww-lkRo}}Zpnc?Eud+r1S2myPD22;c?u`&~NQW^fgIFqj@bB}MHnJEd3>A{G
zyIl!zZ-5dqYPcl5kh4i?zMjCL^uy1=^?Q3$l^aX2P$XJlkFj8=eAMyCNa9%hJ_^I+
zu;$pfvw;$z#Z{wSY<0VV3HuQPnkMU+KjtJxh%`gjt)jPUWC(-XyCiUr5yshjufx9M
zw`nMEh0Ai+G5=DqyVyw;QjR1$X!`&w_5}N<dOt>G{dCk-QBkSOawhNn`sCrXlB}o5
z(@|xOh*duEaI;E|o~?q2i<|wh-YnOKg+3W_+IYMeycmKQr8JeaKVqCp0zkQ;jd)hq
zrC@3|d7@aAENU(EhF5O)&t_b+r|07Y0+ZORpE!g$5V4HfZ5=wamW+@VQJ>e=Z^pi?
zsDfL$**3WK;z1ccd|8>VP#GMpQZU{AWXdv<uJk)}hs)N?nl3nc#cK?Lmq`Lujz^<}
zV6y+qdV^Y~TQ)Lls5++g=&0q70o=E&YU>L<ee!wGEM-#CxL^Y1<-yu{iubK$Stw*6
zP5AO}Uz~d3@u;FG?i;(9RXMsxPskH#pW_>rYu@4cCB#*SLEbRDa?3R~i^2Rx$|&Qf
z+|vrhFN`pOm{CV^MJ(DtQ7maJG;PH3l{tJNFu&N|HgRT%^m-=4Pf#(DP0*)l2JxA2
zUG<&J{o3NeWwH6h3;`NMI4<tSx#I6mOjdldT#nQ<ZXzzm?Bk61tz$3S7B5J^w2iu;
zxC4mD1roq?r>nFO+%2!Ka%8F!W+-q1_z}B#bqoaWzIR*MXON)n24gR#8hDOMl*%b`
z;;+B?wPW=HFQA4Z7D4NV?usV8bg8DU{+`y|#-?a=^XT&0Xp^;Rz^t-=O+`==4`Aui
z$ytP3buen?<TkBLtSSG`MN>D|-JxZM%n%i1E(c9iOvrPH@|Wh=E(kG7C<hv`Dlv*u
z00(-bY(E^nE;>^=yEMkbTP3#}i+5{kcheCZb{*M)X2xwXKZh=jN&Bz8#J4uLHl@;f
zaqv|5t*Yvyvl~!L!>v)+gQps>t|ZY?X>f|<CeFzJxb!_{AUw^J0;nv2uEB`FXmqFf
zVu0_!1!R*p%>yVKZJ}-77nI!+8Nm~hO21N!^x}bRdsX}IjI=d@IZ6lf9Ubk%B7!vU
zArJ<%H;bBoE|NrX#}R$`vSi=F<5V}827zEXQ2qg>=&+kIba#ICch_hj3sv*pzK^Yd
zH@IyXMtJ=&uN_IExQ+9-aox96zm51%ew$iwh8~jNn(jHN=Pr2!>^6+`8GaJVyOK`p
z3JoekojEQe@pKx#bHwGW3RU{l%^5#g{l&-p#Si@k|D;a*z*_J3*lc>*NTt{H%6KDl
zeAmKvdK`JqaewSK66j;@agrDZWkRD{%IkI9t~FE$e|f0=k`iO{Ds4Iu5s`W2w8Q(=
zilzS-5%YXo4iEWBNFEKMf{(dRgFw<m_%QudgG}wA#?x~oF}u{K&;~8L(VD<q!*)ls
zvZcBgnhv^1EPPLqJ}OYVRDV@~lX+BH_pIJfPNMWW(n{u(t|vY5Yky3XcW2?M#G`UC
zo{bTfGen(E2>S3IjoL0y|Ajk-(#wU?|BZJuZVhyr<*FWqg?#!2EIz}x_^+CKsqU8-
zxMTM7U2wRQB)z;O{n}sk8<Fd~fNUJ^f*59|Q~+=+%eh^&9=qfcmk-D6XS=8d)?^hO
znOi5id(OQU8C((Vs4o!#z{O78T~;N2%3~Puy)OXPkU9%kS4;V7BZRxnD7wG8BF{Tw
zC58J_J?>7e(5zBx@(d|mq&BjTu^Wc2IBe9|e|EDQ&2>L=-J`93)4R$7@E8Oe8bXS~
zN$Trse=z@Bk<Ulz=XrvlUI#@VFJ+%b5ywrD{!PieR{WE_sc1lC-g(04@YSm%z3U`B
zlX-<bc2P$Rr+kq9*zW~KOGO*Qzm`(v5@+e361MmTX$Ux-iP~1h*ZqSs`XUqg)bQSH
zw_5yuFZ8e%8a(oMZ>~FWTFo*w%S{Rpn)D!PKn0hR*r(a<-*H1Avmoov>>wPD|50bE
zSpvMvdtRqD%BGga#l?=tfH1GoUH5Ip=~Ttr+sfEcwo?eQfN!7@6}PPh{{1dk+tnjO
zQF`dUkym9CotyYHB$*9-MPwb=3JJC}f>*j;!_PZD_$M)T5wrI~MHO9uKAj#_TuWsP
z+^PU2AxvKV#KZTp`6f7vCe8C*MyJ@rKvuY)FHBOIAzR5`c(#qm+~l87dVD+=(9n*d
zU~hgv0r*@@(5X+Y!B*J*g-Z-bnV4^z7*Z(Gb9D19<ogW9Pgb=UnAxZ`NCPk$DSP6X
znn7STCTnes);0_)jmoB}Io!&!m~yLZ6t&itleM*!ksrVBAV%pS#&Q{8+epHDfXV;?
z-me2+-2d{S#ZoC&pX`UdS+--uWvk!D{3$vPamIw71g4%zN$9w7s)D)Z1;C|Q3|0W{
zrYgV>&q1sYb@)(d*t=OD7`yhvrYwd6_E>(uD-TeZLXlB1yv-3=8cmj8oDVoyawKs~
zB;>$NbhLLp#Lu;iUZ`FH`sAy@7f;zrf>rNon-NY%jtL9C*?Q^NhiYn61tt9Y*36E4
zS1(o>ml@VDVR&a$kY|v#0LBk}9~kBK`y&UwDyqMe+-H_kc^BF~ePP~xrVLTqJQ31B
zqzSs)5xb-17sI+kS+ZFxdvx1J(Oxd`DWu#*$7Euw&YegfKVEaH2cg}{QX7HP7`&&E
zE7??}$l~GCHqg7!SoQNx{1UV^5|g{lJcM5sI-=m`4%c6oB{2azRXP1mOT9UW5EbdQ
zUUj=F0RxbG3c(OmJ?88oM1P=46WaIT+*H!6!a&^z&|h*!HnuNE8iI^t#I!#gUwmN&
ze4rapYB=!u_#{GDn+7=mc_)5e&B8sNFSXkZ&X}tj|F_L8lp{2-h=D-$ssa#bG|YbL
zlbk40flA_#fbU&vrzvR5RbWf)D}y!rFH=^Qo{X%_AP#CNgYVb~Y2)32ZeEa;X7I`~
z!d|5SPtPP1yETsJw%|!IGnn>Bd$?+dP2T%zvI9rj)YMe-A|g84RUmbS{MxK?dd(?6
z^(p3^3)@DxHO{;Y`n`P|OOvnfuruFyIy{<#4mLO-pb+Z7+adBD>JU_@^Y6ANvJ7oW
z=U7fBpB>yx=7~b!7KtHsX{@mgbUo%hy`b6!+|K}AX~luEFke{Rgt<yEl48g>x3Xi?
z7^wK-Ww_o_{hVgU#FNT85@=*tpCy<2{&}?y&){j6t9SuzK{0KCt;<`f58TgsW?*zP
zD*ofd=nhpAC*yqtnTzrtg-?jxlibNUNTY>@1<7Ojz0_BqlSYH5^iW<{M5HArM+&&T
z%U@<Mk}zS*zj_>X|4%%!B$NaiQdI<`b(o#qKzZ?Rmy!0>+7!^REC$?ftDYxZf6x~W
zhDDn3GRBN}$KR4z7s%C+O4#DUPpivcfj6##t@FwL2NfCne<=fg-}~&J{}gEcG1@QC
z%(6ikql)@+y-so6cXzrqSF!M&?9f0Dgu{t^*ik?>8rgp7oFUZ^-$lI{&zh@?<ikFl
z#M`pRo5XFFVcT_sZWR-5O%rZKJv=SWJq8WtL-S3`{^X}qCLi_sppux}@?eoc=omy@
z1&Tk<u>!x(lfo)q$^IZ9g=4F1wBUiZ*2_XB3wiG9alG6(hR7vHY)W?&MgLq=mMU^D
zw8cq!W;II1`dQ~+%S>*34{T15Q|32M({<RR2P_^!9-eLpIL`6_^Jb+E%>Rb=t8fBa
zwXEV#6U()>6|bzUBSv46yA5qZYNNb_mK-0IhK1`tm03E(PseOuUzm@j>5F7|Ln-5S
zeyxzcxctbvsYScAwgXXo;`JtVi0LE4r2Oi0<dA2tj6Qr1A?SD{%eag=t){)e%x`&I
z{s>fVwya7G#QTGM_x)_dy_ILE;fjy|bcN^>4<SY2<FOj57J5{R`Gl$v0*2Fmq6Kc6
z4c}L6K5f&jzqC$lC%eujB<wtG*9S1K!ZBfpjvy4+TScIF?@<cW;9!t8YgZ5^gAqD{
z9t5DDVFlbFVB?g!^swO$y1R&Aww$6Ox!lpv!rHP?v*%z6xCr2)cV<(~fE7S;bGrek
zs<>36ZH0lQkkSrzwt`$QavO(rsXVJ(%&fXPu=&sW(EWinfS8y4n=TDyM>*3(SyFDS
zbCz7XbaG$+f%@7f8zX?%l<(tR^r`f_zBpa*Rs??zgyTO0LcgOezTVZrT(n^CdKi?#
z3sF5e2L01@w72NL_A3F1*IHp%%B~NkZadYdd9IUG2kWYq?ST8azk#r~9i~%dMgZtp
z<ysi4sHy_>phCCE9C(c2Zg@<Lx}fBvQBb)xFOQ*Gl}i}Nzf9~nG7SD27{HlSDc@ib
z`Vj=WCNsQKg30{n?m<wHQ|&M2C?4(s4IP)PsdemySK!}~L93zMPX$Mdc&8k8(7J^Z
z(4-aP8;9pC^qjs4XuU<4wx=-gfs`mA&lwn+{Zk0UaQ9EAC&>S-ulJPXK@SLK6|cLa
z;So5bZNDZ4z1&@sY)>ny)7Y*@S+#}6b(uAfOQOT%Y~yfIn>-DgZX8Pr&w8^l2rk_0
zVN`jOgbwWJXHe5tA?xp|OlK#B6w1DK7N$D>yM|Frz2!uxg<zf^OG?Tj*3lO08*w)b
zSu}6fFH@|U{j?g?CJPlKOZ-N|y^ARqH=CXr8wX+8%)*<hmqJ#5^A;-9l6@lCE}Hx>
zx-t$c__ud+R;*^al+UMELZdgy)_{kh^KJ<TWWcJ_?VGR8D~kMEQZ>OL;~`w4AB_6k
ze^PXD)_hwcvTF=kS~%{ESonBjcVY7Ji0{HT2Huo2qs1*Da&?$ib(vOgzh&aTVbI{1
z-gskpxH^UX%iFNlX*TYmV$0r0=AW2r=;qQ)s%@^Y<Kl-T2^&$He1XykDxvLt$^pM^
z#CR{Q;kb^U!<^9J@};?g{B+AnMyuZ=bN(U2ng<$Tg1hOUvg}d5#2*h544of_?MM*n
z#DrKLMlXooHjl2U>8I)!_+ab3stWw3Sp^OJR-iZw1UMOfKgpLli7fc#HAp%a<!TYO
zKP*2)2ocK8zOi7=7oED|I|+IL^;k*D)ZH&RW_Gqn#HqY=(RwRO1`VgpuGUpv&YSFN
zc7v|ZG%o3QTaiH_@dqmTrm}=cCGwHk4RdPhrJL=wLjtm<TAnwN{!Pb|8jE`I)*8Be
zQ%}*bfpDP~;makY08F7nuYJ|^ZoD^zA3l5;2~_U%$r`xRd54>pdgwng7~H~>un~`M
zUCQORXQ6Rl9F3l!vJoNoCW6;$sMSASca6{=8Q%7QK++Ps4^#LiK&PZmZyWbS%S=9k
zfQS(WIuTgVZh+M2H_wNt_j%_IFLTqqm$T{Z`G986svWt_X8=n<{D`7>3h^Ld#i4A&
zqDh^PhHHk2Z^sYvZ<<1&a9q2TBwAt{$87d|ge*O6XFp7Xs+l9^8}Aej=p>_>XM%^;
z#tm}<%Ol>}d-Fa6@3<Rnq39WW-?#(>8uT9xd*^q~@f)zYy2?1GF}RKqQi>qTHS_1&
zAzw##V=A<Zj+2$^X1AWdnN6cu;iH~7I`L7fz8lM>Mq~&i{qtjY3+c>Zcc<Rvt|A<t
zk%w3j+sT_VNvh0~J9D&deUTqaCqH!hC%XQ)%I^Q+I4u0TG=Xv{sNJlsoQRP5YQLXK
z!6RvJEc2;q;gg*03Kws;`2Lb6irmo(l_BeuzK^c2b=6zl(jG};7tEcgqCGagi=7}a
z=A5Que0z$v<&IOlsz<HRob9gntf$W@>7F2j-tESl^|U`>^q`7?s5iP#Yiuis&lIcn
zxKyrqo1*+hF&Hi+J!oq~MP6c_`_#<RYu4<y6W`9*aF~J#3E;2szu1j52{fVgf$y`+
zZ7&z!8y4CGJDoa1iA+%@+uvu7ymZ)3=cQ3kY}rN)+~H&RBy>*ts)(wYjsl*+*{{!S
zZOmipp{0HLzDhQZqaj$nu|xI+sj8fR4^sQTT=~pdZalZ?*hs@7ht^;MiVY+@h~tNa
zz5U<)0ASJZ8gOKnlW$x9LKd<K{>yR2)I+Z~9K4J=ymJ?NY)v~<nd{>Deb;v94L;_^
z=_BqB<+uf$D-!D)YeX9S5}%d8DOT+^4bdU>Eaq#G1`l4(27Z@1wrfWUxIH>qW3Gu$
zYQ!PqE^=O~(4$LTIOsl*SHtq{l_!o5I&~Jq3`I5v&hLrlT0yU!r^U;1JX$+X%dXTt
z1W~FW{MvD@i`ShQ{j`V3zWI9o+E2Y<qow^J8(;S!EjA^w%i@Yk4xumq@fUja$&Gm5
zy%^H{+Idi`;C8}2-9XnQnkk7FQPPU&=!1qRy>@tylDRI%FS3`K-XL<YJecaaadH_%
zub)<Xn1(}l>S=eggQ1OQ`eXA;`>hsl)4aELug}**D=LbCjkTTg-eKqu{zYMaTJLRa
zJ-)SFg<EaoN=`@yYfsC;$0R*IgRPd*TlX%E=6|NBFc-@+*+3;{YK?fcimhPLVpclH
zr{^c_lo3_&bmnd`&mdwtZp|@%?kkwA){DFn7?6zqpK{FeS}s{9E|>e8<dENAo~3z`
zL#`mY+ESz#kPsr&$Nn@Rw;>1&r4{t7hQzX=N-*7jEkQ%ybF+oTI9&cmMm&{>N<T!5
zaaeRm(Jn6S&&sPCWl>1Lp2bbyk9hUsSsFr0=6Ixlgc7gw?(A+L^N@Yk0mfgKuM;#?
zOk2Tced5&Le;C`JIQ8r0=B=!GQ3M@EMAPv+MtbFZg3r?7EuVkwJCD;ZyCo7-zHJ8^
zVy3I#)P(iV@~sVqIrrU#&iMTO)yFHQw=dbxutOhN-7;Wi=3q^#yv(atDmX3GuVGtE
z9Xs81<gFe_7;0V}>M$4Gao&L=!<@U$mQz_!#h7O6jnCObf4ePxRi3O}LTb}!GTr=D
zoa>rYD_<wfnamJsy}t9>*JaLChyiW!+Fg2UugAT8m=!bgO(&p}*CXyPyTa@5+e|-!
zeMc0qb7OL=zJ~@&gRtK;F=l2@ASXFM5O|S%R~{g#=MDk<-$?=%#o)+OC|HDo1CoGI
ezdy0=7UiYjj8+krn>x^vg~&)Myebhl4Ei6}qfuG_

literal 53506
zcmb@uby$^Ow>7-!O?Phv1p$$gl8{td8fhu%?(Pl&1p(=n7Le`^=?-b>?yh&??>Xl^
z&+~ofy1qX?F7+n%-t2p=x#k>W%rOm-lM%y2BSV8gAea*3ZxtXA_#p@c?g9!AzO(aG
z8WUVlt;N;sAP{uChre(mbDlR42qi?~t&oy)($1WNJO24S%E7sW?eKRKg>1=Ey^cqL
z;pra{I(1k`u3Nl!c{PLDtgBEzeq_m-!U&wPt`bHMK(ok`3%{P*k32wXluOe`t@(L=
z9LXn{FxL>}o=|Ul#O=GxB?x`uk8?ylG6RPr41IF4*4FCt@JZx<UNajWHDh8R`)R~d
z;XrpBc?BXfvgvx3tets9P{pIDB}Oc(+<3*orz>0$smyd(oWybjA~=WwDC8pY482Fk
zNQ7tsoQdJ7(-*I&Ak819#iIzqIxN+c;?HOZu%dsh#I6RveH#5szifK;C;NGnWcHrx
zx4jM)j5e0~e#5Gn%S%TZqA*Nk1c9<c$C8i^y@c;BiT!$|B@ElUl+QkkQ}ZN8q)t2Y
zPEq5b3gPVah{_WXVKslCf4DnW%o1x#e;ncN<^OxF=T<IKUtLI798BShgmM4bWdM&H
zx_8FMA`^jFFSQ1czQWU4ipq4H89w~Tn!t`5@^&680IUq3zXsPl&aYp;<T7a>5JvCr
z+@vI!6)jjY|Ho~uhk?ek)h~Bu&}+D@Y4b$x{0<JR|Lcbz=&gdm$1U6U+oN7)et!Np
z%0*8l3ZuX?oe|Ny5;J9zc^tOJ4js|Jd(B0>;3EO=UHEZO0K)(4RfQ9ajFFyRz1BWY
zkP?c9M#g1LDTFgG1D*p-SCQkaw1#n6FD5(9$Fmwp%FD~!+WxFseprT$@vq^Gu-)Gk
z(URF@;UIo;mFU;Qu^+!KnKq}EsI|eeilTlVYCAbH8%zoHM5dC68U&k0fhv&7zIdoZ
z1@#@APVz#Z$5Mb_kM3MIpO=w`KW{8MybRxPpeGM_lc%K45F3C_Ji^ZSMLg-1=jpi0
z^XJbS8yk6@_UwDrp}wDg>07FOrDefb-i1D+g0@ahPR=~<A@QYLv8}06d`+B+k|;=d
z;20VOcG*hInTdje;^E?n#c3pfj}~C5me96j-D}&!1ESZ;$hqd`#yv(-1m~@$rlz>K
zco-35%2koZM1~UmscS>Op<VCDtQ36=83wXd4jo(@2q@zUbt{zT;8VxC<W#RMPft&>
z@W8*Z)2{bYdXy)5U`9xik|>Ok2)*~`!)?Y$zE@IeA2hFjWR4f&PYHE0$fT3ZQ>rzP
zH&CCR8Yr_{A0U5a3GtpbUt13pdRYGk!Re{&kqWcP)!sOCMEDxbyTdLDH(soVDAAIh
zrh^Fy34v#w{@rS5O|JQlf)MzTTzugFvf2Mz5d8o3eK9v)Z*p(|s~CbPq3Lqs@?SM#
z*R%*kYPNN2iKR!WC5|}%o?};V#G$@ABw)S3Ilf53gm6V1JATIFLLmnS^AhP9v^AW9
z$x=;5?HnBI#QwF|Rcn}0)y%GA-NOsAOV>TTpb!p$$RTGU*hVhc-#F4)HsYMe;*6q3
z^<A{!teYDcL`H?w_jk^;tTDjAVUDV1gInyUgIT&y8mh&nrozFx4%tY0f4UEJ&a}e)
zeI)VMWiEarJ1*GGjoi8y=fh_A{v5Dkn8!L2ZTbCha{r$>?LTwee}3(TzqTYeAC;PV
zzZ@xG%AR03S36#yvOm1*+Mq=f?edZYyl7(!BdpsY)$8suqkDrwy=t4q>@1nr39|U7
zBkXQ>KIE!NsKM==iGkt7Q4oZNyH5Z@4IMM{ey%wDxT&BVi@nQvdk8+H#Q9L~!u9dH
zc+Xt3bE~G?i5XH{Vdx<e5!TAnYz0*A>Bahq=5X@i;bDlixDg}-5ki?Y>xY6}Plf}s
zL!{1<AXA2k>-vgB(R#lx>c`F?<fhXJ4G2rV$;S%=zS-cQ7^ym^{k`2!Aj~e{vB^Y!
zm<?xi%vd$`vwTu6aF}P+ZS|jOaEp$LQkPC4ASd6RbAEUv0y>H<tuG~y<IYs+M=>iD
z5T)(R)l#dNx}1sF6#VX+BhUVR(_hjXq(wyaBF<VdS>q7`G<5#CN{hRj%YX9GZRAgR
zu1})NN2|R>$_2vdBs}(XZ7lOL?aXdxJ3cZW(xf#Y7ba`vAsrncQT!{>DswvK!CpI;
z81^TOewRkaz_1@U_fB~`%05+yeXm<3Pp~@91>WP(Xiq0m1pWJbL_|c0_d6PVDB=LQ
zqJjbg5xR){fOZX%h&-&T#>~X>Z4|Y%z<I~k?~3h+NCAHLdwv3<PbO1>c}k6yIs_uP
zh=;!BS}X1xsJ;h|pQDD^`v=XLADe4|r&*yS9!q>wn*e!X(S771epj1yO8lp}d(}C+
zZ)$4svs&lFKgCUVJO(`krK%W^wjgX-DJexdg6)YSyWXScp?V-@WuX#L%$`jk;#GcV
z%G1XTpjK;<`%*%E7uVh5kC&4`%@mQx3Tm5qE_pW6vc~BC7$jp`SdicDdr<JWU-;Jj
zEM>UfC^%|Ti+>I?oBT1peVXcV7T?bOWflME_;~IoTDC&o<hI2?9-E8k|44YU18prY
zFxsYe9X%VzuP!f5*vUK%mv?pW!#dcdx9xdAF7#1ls8nURx!F<5QxXvox#<H3kLHUw
z)Yn>tGc-eMJ<w?W&vIp<5vWh08M(MrqRza~hg7C1&xt1H?tZg7Q$;JCfCzcSV{bfB
z7+;0|P_Q-#zlOKI)bGZz61%9Xwps2xT<&_IlzVVCV<q4szI=0eOnmDBzjGFo>N!%N
z!puzb7WzbqsK$1~T#b7Np&@@XN46?~$09*<dVT1N%kdgt?JYP`VS5Ixz8xjb({UPq
z(9qEvNbo?~R8#uhYO~V)1^f~l*dz$l0O3-L8P3Q~VXacTb(srW#IYX;x5y0K4r{}^
zYHDh7jjtgPAwg$qEvLsIfC}g>mcDUXsCPv?0x`|-MH~+gukrRw<Kg*B>+2}o+z^PN
z9r^TfnNI?{2|UVqTM=Tl`(e>e9~vROGviZGZ-3^McUcn<5R{uw+xC5X`Jv7c6BSiL
zM&`_zVTJHb5LN@He)sU4m!Q^sw;CJ8+z;gfX7~HKwf@9scoIz<W*6M@*kD2VU+mI~
z-CmtupYMBt+X=y^l}Y92V_@jZEd8u<ad~MmQ(06}GBi1Po|kv{l!~h9YU_m+(i^aD
z6ETi)J)wjz$T&>Po%XdJ>S22<qsD78beAuwsT4BpVF3O#NY(|kxx_244}!H_U+18F
zNQ&*U&ar3@oB0}R@NW_N+!a%4%{8bnU1dd)lI;r~ldkgqck9w?DFYDCk9Zy1%ajxp
z6aq0w!L}7kCvv**qC=YNU5;;f9q$g>umPaju77yw-6NhWr;U?d=BBTPs65G*8$;=L
zA8N5kSSZ=pMx%8>db=f$IZGQRe4%VUU+)4k40tDyPvo!m_6LLf)IFTXj}8tFz#6%|
z>}&BuAtokh@$C>}Sfj%4<mW08;N`s*m~m?LeFkAM9Vi4z0z__u^ON3K#+l(Pq_H1m
zyxWaBYkeGM6ZNq>;QsCy!R^0EmwPKBVli871O9}QaiC#ipBppqvyp{+;S+%TAtAh7
z8OQYD$TDSRML$;yAL5Pc^IJP}SYKaXutQQ1M-WHu84V-jZZ^7vgoKajI2=T{XVac%
zkVX_>=Y7TBtuH@zega|m5`!xGS7#)JfZbYO+`WJmUC7%JJXPldh$>M8EMKGQhtqMn
z@i6?|&7-5o&EXf7%~w}f%Eelbn6&-B>Q~1h=g3a{_;8(#ZZcQPVg4H;5>R5~^zL1s
zS#UTwSlR(}tJ&J%ELq=PIfT50hlpmDI`00ivRXK^t{mwZ9CYfn{VJKQRa`Y~&fH*R
zYP#i%gdQ3iI`45l7s-Fs&1mNJGEd2rg(T&ChRtZe98|T11+VVUPv~RNAR{9q;8*wl
z+_}G9xEB!<YqGzDby+Ml-hwJsT3UL0eNGHT%*N<^K<4U2m0!=jl1TzfXLcV_7f6rm
z>p8>5AVYTe^@sh1Q#Z5lwEy3*)bAGr5f>NE4;hpuIuF3}`?c1?tHZNFJoF&566<Qn
ze_v?7g0N}yxaKq&S@FZ>U*xh}A4mpO_RP3Y<xio|JxmA(qwV0hNB>`VLwfW9Z;(rm
z-ngeeg9=mQ9kP*woGwU7DG^mQ4b3QDSkPiye*hu;*qqkeL17xPiW3d3+THUFZjq1q
z&>)N&l^{GbX%!n1e0|}o7`i4WRf;D8s;Oe=Z=l8t88$v~48>!?pdvuKyv&Xs)ZhZ^
zuUxSD7u9?be>+;PReTO$5<pl5<2{4Oo9pW;nwouoA*JO)>(ci2cC1K@3G+SYeG3Z3
zYX>Q?XSEL7l~xOT@#4jO4gJZf3kcrDElGLKiT%I8I(HtM&qo|+0*G`>{B>};{!u#O
zo`=!$`bYO$EiEo^Hjl_`|6+a2v!mAG@6w5rpAj0qj+=g{w$^obch8oL0l~=}BiWF=
zQF?Q8lPwiD-{jQ<s>*)TedAx6@F#g>pv|Y=@>kTeUiYs-w1XF%`K>$CWqh?viT!7*
zaoSf-n_KrX*U_}HjVE`RVv$;kb$8lrqF!f{Z3rF@S%X%0iH_~h@^ZR?8N~YvQ&CZo
zul7fgMpb8kSj#LDi`DQoTa)KuvRdq%VxbTHOG*OaxF0#Psb|KVpn4eK5=F3Px}I)X
z|FMF2OA%91QGsU{7BbLLU_vNyyj5@husMlGnRh_bOK>2Iv(u*I`7Z3_52-jf(oQBV
zN0lMgpf~nIlV?N7Fdg5e)7!W0SrXB&Ydg<&XPA?C9`Qed<~`>V_SNEf^=newi_d9K
zGy0cOu~x&64|T5pybu&!jFiP_Aj#^+Pxp_~RGC|pIE!98RJOPjWE+(4(wE!CO~S&$
zR}30wBiAbVO5m_Mq{slMnY+{4DvwP4>>1chIEej@yY8J-M6Pw*?+UZHwR|g(jO0uf
zU&0Io#l@l2fJJK7%hlK#Sam&c04-8*<2pwxzitPUeGfAQ>fP@+c4QxZGK&S?Pnr^a
z5GFY?=ZtbDcWoC+mF3)#<I{&~JO1#3ZAsjAIz9GsS3>ejpjdicETR0vKDFR1m*~g=
zIJ%&NK->vvqJPcU{LSos8eBbrDSU2wYU~v1m4TT-bq?EWyMJ+2K})MV=gH1A8#VRH
zNVe2z=coMl1+i6n)-b2#PRuB38y<P^i;EG_#p0JvLtd)Jg7Uvn43Dh(gSO>|dPOKD
zlz9!jZ>7Mm*&n$;#eiW2RVm)MeW<ZrRSAIwy2GX38<9&aRNJ)8gn;)FLu!y|Z4Hb2
z`W0lxr+_-_GTdu1^ukN?-JpnkynLu~t_C0DLZ{6mA|t)F@R^=H>(DEeBOt`VS^PmL
zPtakLR2RD@N-fcNf@Yec+~j#HQ=oERVZ&E4TUYxBFze6M67mE>I9h|sPuB~6&D`>{
zOCPh5Qy^=E0MTMs$By5Gq4$B`B*~Hql-)`plS{;g@1LcJ9vmN6yjA@EE}q>aN}SsE
z&hFx1$;8A2Eui&79A+k6sDYX%{}aHFmTr9wrkA@Sqp7D0)!5ckafEOXg$|y~lqFHD
z-?$4s5B;-C^E3sc(&*5o<O}sL8yHrlN(W##?p}5(oUsN-@ID$u@4p|Gs>hB1oSZDM
zvc6tt;3JvM%Ex!x{e`Nfr3I6W-MW``PUK(c4gv`d4@ZK8i!R^L{)^Za2hx-NLu{=o
z0I^jsiUmZCZ7Ss{QJ6`>#HRfN+bUHxFIBy`oX%6oV<*G>tEWg%<M~p~#)N0kWh-nx
zWFmuV>*^1NhK85tbwc9Y)sER2O0~Lefe(1cK74&J^&X%&xW%$TezV!?QnASgdgX$<
zr|QbU#DtZ=iIR@v8B)h))B;EYUu|t`YpV>K;ojq;)VVi74ZLnLT3RV$DnSp=kyoRK
zl^OQC+Cze9L2?vtL}%1MknsI|Pfg`nzTAut7#C8-wyzR09_pA?uu(h+byvTb?A9d{
z0Fw%$<Iu&h5qevbvkZx%h0_maw;pJKO0jnb1`)buqiTie`@y1Kd=Pl!rvGql0|cUo
zE`8Q%Y=IjVR-T?F6{ib_E(w(Du}kKfnwm1B!4xi=KmCcX?k~5-Oiu@FKK^Fb81tNN
zwV(<_K}NniHfZGBo-D@Rn#$8W$EJ}^Soy0P^?3iZSKm3VUS%LGQs%Ya7(!sBsP=rs
z<K**W9|7`-BDvHq$W9=lZE0h=Z2d5W!bceHMcjwWySkyFA+N@Hg`<hmSx~`dDlHP#
ziyi~YlbV`(+zFPgCl$-TcpV;6_`%LtP+&dpg?rm^nbJSIcGeE}TB?JCJ|;gmH~6o7
zalk+*+sp{V2Jq%ku%Pl_6}oZ6lsk@X_5Ok^Ty#Bhb91w6+()j6$3IZ$IwD6V`AB_Y
z^4ZImrN=i&<U9@^@NjQ0*HZ;*8%j!WgoPj9QN4amOWVbJ`EWQNsd7CY**;y|Mu!ZQ
zbMH(c(?+ye@z(vyj%=I&D<gcxXYKh+`W>BiWBqw+BTZK%MUc}%khfqisoS><baeFO
zDlrj}iXJZrt!(r7ixMkYfYTbt08x-5^$u80-<zvzOW0q!czJPL1y^ClNPs>u{-Xw@
ze=#SZ?KNTC<I#NbvB{)W`Jvk6>B0QngRdu!8T-{T9E5?$`Jj1Hts6oYdIdxlqv)Ni
zo22&s#GBjOhWNkaz`M)(wIORfKuClH1Y;Q@VeReh)zw@S6sd3s$gh8QFNw_m9_4o0
zn|+oeQxz2~_K9Q|R0^YgZ?^DS<j5s!xW{cvWUm9;keo4sXmysH-{mQ7ow_FW$A0-D
zZ*(9>a>Q*Ms<C_J0U)ndeRu3Hef36n7d<__<<4*vOtQD~W4-sOlfeQ3e#_N6JLUoc
zO=8g%I!o4=RsskB)gYq>p_f=Y!|y3PZ}}BXn{G!8DqO|(<I}_bzCQc?BWCT!*X|bw
zfD3%j$!R(n0Dw)iP-&l(!&B1o4ipq^6OoXRkjY|gF9e=dsZIM@fLXd^67VL9y1=3~
zJVJS3$Or25&!Fh&es)d5VX@6WKgTc4e<WaKV_S&lc_bQ%<hOJdjB8_K1482J?s}g)
zVjFN|F$#~9vD?Yc1%Lij2*loF|70Wp2o6lEp`sD6nlDsB>BjXR*K;yOxtTqUUnNHF
z&S`v2na{IX-m_3gcXn~HCpModS+Y)aO0&V&nTGo)w&t+6vkOR`{~ZcZ1RlHDg!xzq
zzEo!fV(bS2_it|NicQu_@Z~1xcW{f?v^@bGU0vszCD64#l)BySDfw4c^FR!+c6PqV
zygm;N4X$%Tp;z)RGb(sraOV4L{rB#qbSz>ScZ+#L`Z$fp&2wk~#2R_=-5S2pj|mH?
z1Nj~Ti6h(dx0Jy+Xc+;%^iRLDQWZYIyBxNhF=+d_1Gv&as(>~q8esX{?2SGv7X<X^
zLLk(r!hr^AMlBCS315{<lGEl7z`?YxdcOc`@fUD<Gvl~e(l%E&0Kla8W^}qnB!_1{
z2ZBId?OvDdY7h6mfW>ci%d3+jvq_9Xd5fuX6Imwug}rJ+PVJWo#R+|reQU79TdxUf
z0{)v8V}?QKDk~{OQ%cIOnR2}!*6fkE+t}}T6*Bi#7rr)DRY~^g4_e$wyfFWJ9sKTB
z3+`-GwAi&C)tB8ZH!&|j;4U3yXaRO%DHRy$y~nq-m)ugyQDOmr-~x!-xqOx4#x#*K
zn`P08s9xkq()XF7nAHyGc$VI+#OhT@5S9%<e`kxe1=eptz68bIGfHxlJ;AwvuR=9i
zzw}uG<E#GIYt9)(bZmUI{E%-XqN)M9e_|Q<P1hmSRijZr12amOgF+}gIrh2_h8Huv
zJ=_}6UZ~_K=CBv+{^)R>F_mUfv)w(zNaSn#zK}hR0C|H`(Vo$%V1u7m?(>$~qeJ!(
zP}8>dJVBfTp<=tTocm>?JrR19AMTey96?x*(5xoEJ8Vxp{y+>igm_VnKGNg<_8f&-
zn<*l1qTyp#aGKzAi5`j2C#a+nk`hrg%|Th+_nu1vKvs8SGlG+T^}Sm(oUGw^fET=q
z$0j{HR#3q5JBAz^E(``E9W{hrn)S6k0h>QRMNzD(5PZGe&bN>+dT(wq8O)ZG&h^oE
zBJi+ZViOHjbo^0$xqMT+fM3VR6#<0pq;XS=kt|ec0rOZ?R8&w$_|t~53gtZ}3e}MP
zbl3Xg&-WK1Lqii-4MLSE>%s1GCNRhM#eJf=#Sf|&MZ|qU#1SKzAl}hgX2f}SfY4<F
zpyc*K)BW*)0x6&CN#I&hny`qt#Gc(s^G}0b$$Rx`q?GJGWb-Lp-bQ)V$b$+taNuiH
zkfg7GRLMF~Y%!Z>cOA&6?JkQXf(z*gKuVMIRz^oe6^0`qQifsxdJ?&3xX!hXQW?J*
zih=w>1nyS0iuSn)y#i%!GxQ_L_xx{pPxRpLrdfe1a&0#kDA$0T#LLF;;$zR(=aHVP
zi`9vSz1*3X<t^3|#ywR6hr``J7iY(h=C;C?JI8Y6GuvC`<>Xi_XR8H{j}}f<1`}WH
zpAKRN2DEY1-(obkE|D?7n7{=Uh>7VJ7Z;6$8F*o68Bx(cxtOwqpDELg@xYK1=dPOS
z@Yn~iQ3IGE-h*Dn+7vKbV#vF-sJH<Z*}26xU+|*7MrD1XCM0-H^&C|~94d^QhJ*GN
z@d=g0O9@u$C$!?EA~Det9y^@zb=cC~eSQ9S2yhTGcF82S#b4U;i1LaI`Ok?k6)xeM
zEy`BEwmA3bTxn#EzYcwQBEW}Mp>4SCh==-q5Qd+xkmq@GD0Yqvc)pTo18Tq<2BQ1p
z0CuZ+_v!LqYOa38F3ZjIF$x@HsS~?QA8JpW%Bm<i2=!Sv=1?Gu2Q8NlrW_`}?R$2(
zyL!ahLY7&a>2_uiUXG`^vnTO$Q@$Q^!GJ@e%%Ou!{q($Dvt7a=!h*24iB6nZ@WZzh
zbb_{ra=~e;Kwpkhk^1RS$nsD<($36iiEbP5>*$L1KlFO?8`-yfs+9||7A($C@hrh8
z6ZoZ#v6B$$;SQCTEP`|In3e}aD%qt+fvm~wh$sZWmZxE+$!|t4MN3*7C_um#oW6^j
z$RGck29E}zL!524#I;=T(SoLLl6KLGGXSGXr0*E3^(xjCA99PyfJ?h%PGmaSL9m!<
z=|IP+LglfChr<lQa=a-KV-c~0y)tYI#EkQv<MzaWl5QM5h6wKRB&1H*Z;Z)j4pY<6
zC>N^AXDbX>=)GcAMNGx=9s9b6Mv>m8>v@W~30hIBy8H$}r}52}vg%8G54a9<<HaAO
z0+z&nkLxo&<pQQDbH<Z|;72xKaT<hDp9P3oKBucSVBKe6m{(JUQR(q3UcYZh{Sx(U
z3PfZ{2}_j`6CeM+f{mrmx)CQfCTCTeNkY|hq4Em>@~pQ{mv;9a|Emk}SNH_P{sEDL
z-ERBP53)<dx-xSyK8f|M(47|+tU)8oC*$dmv5p(;<Qg%S$)Eg?T&1wPw>ObR*Y}sc
z>w2<N<Mm$MK7{{jBh&AEgpv}cF@;q7^DEP_T+CON!bQJJoMeldAwg~O2+gc!rT_p2
zF{>S!{{}k!Z}ZA;-@X}9t*)+S=j5mqslR{v<t94Fq(WVq1^`{>EyqAzns?t0Hd2{Y
zJ2ppiBHtImeON&8UPeOl!iGKIUdTo}3x4Yx?3I-}P1Sm-OeuB^QA0z&U&x;@IwS7Q
zRt}Ic$*N)36o2k|HQ=;9tpf>Rsm3`me4UvtwakTA7&R#m@PX9G!&5h~{dFhcmn`m6
zEupz^bz`4xIeJsw_q>?mwUe~G6_qO^8Y@KNt%MWYme15(GynW6vQS+fu)Ol}az7fj
zzFRW(4i?*;WMs^JcWwg1zQ8t=^n33DLpEdfp?&=BKJnT5rRgYM&|*jpW2Yo&G%9lk
zNu8nh4nPqX7Z;#7)&n5eoa!0a^M0`>+uzGtej=RwRm5yL(#jDw!dPz{vy%lp6A;2&
zZ`-^!9Aqs_%&Bo-MFlrup<UFJIGac&yr=QYrTfNy8m57|ZK|Fhmcs9k1aMIk=0N1h
zH#xc+_mP!tf{!J?NfNf6QZ}N?=Tum@hL*o;bSUz)<a9YX@VYyufMUScVB;XNTiOC(
zhnra0+{b+|rw+yNMaG1Kz{C4|r@UlBL~2_LAIj%|2pUh#xBkhQa6aFg19!!u$L@Z&
zx-WS;U1k&niY)tcm;AAu`NbzdaS2efOIx5lSPiyny!`$(vemR=Z>DO0V>ol5|J1O|
zWo3<PcjSikzq|md)rj8*rnV@`trqwvQ<A_gp8QPE`6c}Am7B3@O>S<k(SGLlyq9qK
zhA-lLZ<h%4P2T5M)+U~Lns!=-4;<F=+(kW;y5Y6onyWhpd#XK126Wg40?VZf3Ulk4
z!odE<WY%AB=BwYn$*jT`KsRb_CFaI7q7|(O?adwJ%(%YBxY6-((B{HmvpG9gOb|#H
zoRD`ag=P~PhnbKLr8^(KyT?+{ewvspmpu;uCBewPgMxrZyc^JrE)a<VUy_B-{It{E
zV1YnF`Rgv@f{aqpNN3pW++(`+O5JY`yTl5*e8)#z+5$~RzXR0`<L0hT>)spS=E)RN
zUrOa4n%ryid<owv`hC?i1_j9N_5X0?%HeWVPJ1=JwlI<<`$UV`DVLrWS)5Tdk~3RC
z!KUFV7Wp_i(UXFRiCh0AMFI=w{gBXL5jtQVn};VQTPlCfdPlvkcdNV#wK{Oq%2anT
z$s8_9jDkLWU2_fJ2tf$9GIbE!%YtE{@#c+ha;OJ~_>AVbD*doO-5Rgkf&>>KdEL;A
zn_!~H3hKdYa-sbCbto2zh=jyfBN}S}2*8PvAPzLZ!jksx=6FzdzY!%mr)8oa$EzV^
z<h;BO#dOS{Dv0^+zyk<WQWUMjnPP5_d9f%?;329F@KMoAy-NqTuZA+7=^M56lP_*x
zrwBK?o|-OYh78`v<1xz?Pj0iZvSwsv1_<l^?5M3wGcFJ<uq%NG;=V<XU-1Z#6{}HN
z@+l~|0m{5K>RY*DIH1)}OT36XK0a|u3Ti{*!VWUCevC!D_!^c1m{_S;K-=YDMn;B2
zG|dNI_YI$6x-2@6u-JYS4Cm{knkt8mnC!?%osusrA_~?^eba2`c4xd?nf=HMw@{Qb
zpJAxQ@8<Typ%UDz??1?2=Yl3Tq{L}|9<&K0vPVGY2ym|w2DhUXiJ-P&ppy6q9$fD?
zVUw{Dt@Q<U(_Of_E^Hiug*97f^Z-!5?2Q*VY}aSI=$M#~*bG0Z7a6Kmf|nC}nvu2u
zfq0WDQqmzJV<XjJDl_=?VW^g4h~s(A&2O}F1HB!zBp&JoMN08|y(Q<^6ayzBvsZ*B
zKW?1z@t#PqM+MLH=v%CeNPcqrjST_CsPfxu!oYC<`dLoiO3Q`DhIZ_SMhr0_;rl#=
zfV-};(wcv`+J93vASI;G!(;*i0ze8{sIj9p88Uihf~8=xk$$`XR!mv3KxtK2Js<#C
zl_K?(fx)zt=y3@B_#lkiac#7v!<M3CMtAH$70;Yg!}lE)(j^Wq5<X2!xbrveB473Q
zK%xb`1TJQTI#77h7w#c=sC_9Y<P$zEkeBau&r`ja)ieQJ$qBpz>fP1;95w>V8%f)B
z0k03BPJ&ij-F*e|-W);zvnnIsvkZ&8bDe?LX~wEt*e9;vxXB?pD{(C5)Nf;g`pH!$
z3MW7($LlC7MwKfxEDVSs$uTkSfy1b{@oMW=(en_Z?E$^r;*Nw*v`ietGz8_b5*oeK
zR=={cYxKfebi`$4BR_p&IApj~%#rEn=*S!2T51a_P|7zF7*i?M;=65^fTEei#2k*w
z3jEUlsp8hR=N-_7q{^k~_xJJ|4!!H+GolDFcC!8&yY;n!WV3GZ9Pyt?nz3dq^FP!x
zO*=MZP$v`p1KD#mhX>D`Et!ZY<$rL!VS81zU#+NQUp;y~Tf_D_5&ixZ6db4cQ9(-o
zkm%+mZ<jK!t3St;7BhiO=ll%n$jhQ_D<&*PG=16_5Nl?O?&)Z$SOypN6QH1h7G{<C
z^v@p#8qi1x<Qo^0)g%Qd#E-3yiR`!FLG`vgp56o!w&G-=nw?UP_J!_X7%{UYCaT{7
zG<XK%m<S%SvW78rKM7hbS~ulDBR${y03?Ry=4K<WUKR`?9D&;tMHv|xQ089%kqVM2
z8h({6grBR05b^l2)yoJ{wzW+I`1ZuUpwmu2o;ic))#1aef&Rxu1|t)!CVh4T7B%*p
z3fILnLNO%b?_SsH%Y$Q*8du-$!UY1T+ihRK-7@mg2Wn!#H10bFy@66;4M&7E`OihR
z*-unTDqXG|6dcf;1g$ba*4BYQ5N0)p*5nva?{Un~R@mU0$;FkguCH|bp@}0r)LDvN
zL8^V+KEV_+REY#3D2)g^{m4Ah+WV?l>rZvz1{w|?yrclco9C1U``)%>8jW!t-q$-_
zyIEyf@F`pn0v(!yBIry5^@lVN&Km&*>qM=b#B{<pnpTj`_W;`6Z?aB>;9WKEimqIU
z_yvj)h=r)q=zh7eu>rP>=lFbg=DcXZGb}Pv;N!~=(h|s}h@NWN6{i}{_*H}2zI^>^
ze|546UUe39f;`)3(w--b<?4pSeWPGEmeYLe0>w<_bzfOI+XcP@W(lXI(H01#f1$Qf
z|E}yF9d0q{l}A>S_Ln)ES)JEeqZAY5M>`SAuFEzJHashNYLyL=(~?XUG+#)5mKhEt
z<&ACvI_wCx6r7zbRyrmv@IHZtcpzC!)%)|nI4n%{B>%X|(8%Z>Xi&h~HCN|kloDV1
z&|N-W1bITMJ|OcgV1gj60rj-W`7ncO+w;2&@!oPKLD+=7aoN?#snZLQ)s1lJH+~4&
z3R_;-G2Io@#U$@kw#OplE;Ti-KZedX4}5@@843FZ;t(J%{6MaIG<GM5M~U~QKf!uy
zWkD)#Wo5;kOOKjC6Mj{=#gI`zKkfJgXe_jyot^OBGC9|vUy-F?0@5kcU=qep{a;Zv
zukc$IC@L*|Trtzs9k@J>PPX0yVZ;5%5+pU?E|^{r^i8EuDuzQwu-`<3u(;@Lbz%<g
zc^~XgJTe^c(Fi#hciuQhRVu(@``#KnELY6GGtDPLO~PqDmB~v>O}(?wB*4J1({MiT
zc65OnYj94%HDtAJC0EK?f6EDdydUrcZiRcAEFPxMI(4nm_hSKH#U0&|w@?e!rKX`*
z3=_un(EkOd7g=dm<v&EgCwMX}Mv;sW?rpmYSXDb?;<%~9bL~A8?8i#m*Du}2h|$94
zYur6S(a64`NDa{7(geN~P0-|cxc*HvOau|M&#0+2Ywb5DiZlQ=z%2bHtUtog@I;*E
z*9hr5M>lUJg*?-DeDPUY);iRuC%R~^u~G+$UA-%>n6&%3?XI#$1gzFHOQsi(2N!cN
z1*1MEjyj$1^Ca(KXOe<i6ho$3dAM>2@3ZJE^rTyV{G=9`z<ABNqJdCgxAv(;2UsN3
zoFlOx?@5yIK$X3)sZmKc5h5w#fsBwDWDWEse}Dge2XcTW+9v}0Fmy<a$B`jA1BDtd
z;y_ZRjbkQ2BV^QQY0YNQWsn<dOXe`i8z&ybMneP2&)#B-9~&E6TU#3t@`3;Cw@f_R
zBi^>sObhS1p9W#apWg?y9d72N?%}dD+fd@Q4BbN@?b=pkL<x^^zUkhks=SR&&X$!8
zI&lLYi6kzp=f6OBtut9uc`9V1$y)lkC~@K8<2PNdk%RgTq*S#sLq3wLXB!R&UB$E$
zMFqf_x~cuAy1JVAI4s@Z1;o1p{YQTaAs(K>)%~pmNckZ1bNq9`w~h;)0d#hpd*&5n
z%DfWPS}E!%V5I9_J({VqilzpKu2HjIRoZ%ACEvQPBv-yBP13VR)$1OoMoAP8GZdYS
zfLPF**?t!a3Glj)p*ydc%hFQV+MVGK<uyD$6u2K6g(u-gUUSD3rCQNXMnwy}!c5aM
z*qFKhF0|mdb5W0gy?*+0W*1RNj(`Jbm0am0?jDJS>PV6EHc5<S)zCd^lZxMw0(Z{Z
zTAqY{cglh|_AbOfwEU`m3&!MI??ByWDZb0XMWdpjyXJizsMKYRSB;I4%!ZU$&hfCs
z>+Fc@ylv)b)`u@8v9+sn+~v`04+ik(XimVJ{{7-?yc4}Ldt(1k%?qHW&$+>5q~lq1
zTOX7YJsSmuk;e{{Ae=iR;S%7>Nd1MUYCINcZ+SRPzL}O`f4E+bnc|_0hG!f?2@<r1
zvYR79<O(=fR=gEZKSAl05*p5~W%-};JMO6J=;(mM@KD--AttEJ=3Rl`p*Eb5WhytJ
z>A1@j$<1vfAXvbf@#)hippaLXj3PtM_7}WBak)ALtqyIksS@3~V>ES%w`_)e!66~s
zUUxSjJ*R;TMf$7P%`x8@QVrZWvo&_-X7-s0><KKo@LJ`lPyCatL0P}IMi(<FVIZfV
z@B$5+?r!nW3{Zb!Q&KF4w{3rD7trbHA>0<^-DVK*(@UVzYICV_PU_SvX4Nb&?x6;p
z=a0W@ic3jJjGOshl<6uB@ivL}DAqnbJd`gOk9!VhJ-bnrz}y%QU2gK7^J?9UP8$`{
z!SNTP^PKyEijGnPYkjJ-ZwX@HYzH|RYmE_JTf5pBj?>CNB3b8tKG*jqfy0c`c!-jn
zz5IQFvR<h(@cp98q`*hP<VxX1<jsn;`~swWmKS@|hjL$PHex79d<^mcjSEszQqb_`
z7qG-@w}6lG|1qaCt@pV%UBiimWNFcp%K;uJ)1OXsO1bJBG%iYBwOjbidF3Ml@uh}C
zuVtahLTeAGd-gmZKs8`iiSV)2{Tw&5G56!S@S4wCw84}Sy_bH|@RyhG(T*(noFR|u
zn)9NqnF+7X{9eQ@ZRl2~qN2yaO9;=dNs3`VJKaQiueZ3KTIqH53uxC~tO@CAcB++c
z-z%vej1%N4VzVbWE->uRH(W1Lcp=sykRbbGkYFhCR$2k4mU@xr7QDmVE^l*rJmtAh
zKe?B=(*2y%0;3J}A`##O06+ZeD%;$8IAul$W3~oG7o6`ufAl~nqxCv|?WCN?)L}Qx
z8yfKXEe&Rr1YO3;vB)m$T##B4MKb!EvF%itbgbf5Gs=ErYFjWxp;r99js*a^?WHBw
z&mqkdjNy*A^u$o6wH2c9R75c>AyMSFqAj^EOO}j`Cm*X4O%w%G?k?_U!bwz-bDFCb
zsaF8YK}uGD4r>Di-bvy#+A<A%uW<2W77Lo4DYQ{TI7rfK`Sq$@YKcl<#%d471xJsp
z)`M2xNT_R|<l~EX@o+C<eMJ#2U53^_ic9!(iw|uJW%@Yo3(gYZpiypeVo+cbAcd_7
zbx0nhE*&hBRoly(mwwaB_Z5UXYzd3<WM%s*O?ca>N*tLMa5lqFSq%N;2!4A|FTnxB
z(s}eb9#tK)o6vfj=G!}dr+w2^G!j;Vh(|sPj%d-A8snx@)r4z<KA|zX$IJ)It}JAW
zQ@FsO*}Zz8{33S-LdM}+c+KL4l{+Sv%}QXlidV{73%|ZyK#X?!sr$Qv-?9GOOjYM>
z1Bd6$1s=ZczeBv%!*5RLU5bUu+dVNa08Ukur~V{&G0VhYN`SY+j2m>_-2tyoLPEln
z*5;v~uTURC@cR+A`I)b6`lYb0?w3GbFD60yny=A``Dd>@r&PT53inn}yQWY@7*G}p
z$5vMD@M+^)srsmq@H@w9;-d0K`8MXTuiP;pEIa7AD%e+;-aF<~l}Le+zDRaYUc(Cu
zFIFW#Iu%hyNmS2z&Yl?VSZA<+TWw3a)*pzKyr@Nwi2P>vw_MI+j4MQB4BznWCzYIE
znX$d@kdy0#J0z_S;eoDJPNrIqLx)yQh68QBwC&SCjHZS3g8u0(il8o>`4!R@hc_d+
z?v!|S40QK*+hLOrb-Jm;!TV-EO}T3XDtLis-5!$HS!Gxyyv>FPq3}@<KR6O(E=U5h
zueVA<;<#Sms?<G<=fR%%?C$a3R^0!jch2v*9#4ZZ8i+{_d{JYD(Jhwfu@Wz?pAXIs
z4zMNY!wY>a&InbXe$Bw1mo^lH8+h4$+o2aNPLQU6#0_J;$gz4UnwY%G)gWh#Ck3NU
z?`O^`Pw@S0%V3>Q2+LI)eLEc68>cUVWc%b}`LsWQLFG&~d<$O@XU=s{;b$MJpfp=T
zE(bjLn{Wp*RVDgrH<Ulc-1+ZEllc)KauxYZe53Q}2(Z`2yEBu<>}{1?X(r*nx_>{?
zsp5qdS{|`iyCAF$u!@N!t}*)NMeDRue-_u7uqSy*hK8<lsSq3rUjxI#_g<!CMWcTo
z=nG$>zT`=BvW4(!St!TvF`rKcI14No!R={hDDcezq!YW2C92JO$kC*qL|MoLa}fL<
zhoYq`s$5u*Yt~H^D0IZUP~o?N`}Z1OV*R>vuyu(#MiQtL14-OvRtt@QB}#-}DK(V3
zSW`mU1-gs3icN_L8~0p=^JSQG^2qM;-xa}mzgVY2|0;e<{WL&M5H}0E&Ay1h<Jr2;
zTRLczBqD*3coyCl@g-}hPqfQXf@w-xrb*o#97ESj`0#!qlTZm?mLd2=AuB>wK^((?
z*1~b$Srjz%eg-qN=P2j<ep<98Gso0;Ksef)FC{b!y+JT%w_qZ+>>Z%zBul^4u%->+
zAPc2a7sSCKLOwj?e9|^iBu2E>dw0>VMWvh=+i6;nri>u*+eFfe-KP!Ck@teW9~bw#
zcUsbdSJ=`N@HhcGUo6SL$CEIy{bL1v+#Jb9_IvXBwR^KSob2tHwgOkmqpQ5UJkZ?=
z1vNN0c&g6HN>ES`2!|EwyTAy7{fI{ic*Sgz1Sv~qc5&HV02IoXe1<!s;6ph$cn@;^
z=}MG)X3dnIA-6FM&%P|pJDm<y8l6tTtYStjcDdU0E?=)K06|6+9Tonm`^sue<MR%(
z<Z-gy0&D9e?6!-L_@~j++VsQU`XP(VF~<GM+PH5Z8-<lRHA+UMNWEDmWZ~~9A-*`n
zUxUA&5RpHRM(9FgBod|1M%CGIEcGT*U`K&<`6vlS8=MhDx<_vcEANG?V*(GF!k<U~
zVcQEI)dTiWa#GT1P(Q(p2ymDL1_rwF;xYqQCUCcOUzXX)*Z?du4oo4SI8T(FPi~(A
zO#zr+2t?qTfqXic%$uW-_s|kN{3AloG2Ych%gP$1OA;eVh>xGCkO%A&)_b$uM%lf+
zy`ykR9PvFIbd#A5+h9`wX`9x}eZj<}Ok!^?Ctjdj)Y{gz908M%mS#)rClCP&6>#?{
zXuy`1-{9&va@n4C>eQ4s8*@&XE2O6l>HkPOksdXsa$iw%Md-EBgL}jfL-eh0Sh8cO
z{8KD$dc69FUBUaWdKTVF=0ADO843tqN(6Mo+_(o&zn9;1V`MH)kX&8lD(Vnb6twmi
z7mVM<O7x3C4X{o3rbI;(d^_qGTfqgB{1UvX`b7GRAXGwf0Qv|fh3oH~0iP8GYmlcQ
zD1oi_Lb<V27hw#vl&B$spg}EE0;9!+az)tPy^GHJoWTWw_eRj6QxKOoEfuH70WP>n
zb^;vX;PVHQJOFgB0MM0;p?5pqBMj>R0xP07uv|v}dOK=wZ(pu4@ieX%bQtC=y)Hj!
z^L`Q6=S<Z0Y?_pzBN$<~U?<ade+6t0NlArrbU!p}T7NPK;V9%OrON@t4xD-M?|^6A
zUpQSMuig`Yz3s^W;dJ42xjZFctpMGC&<?#Xy{l*ePwD9BczAfYxVV^^nW?CjK|dXY
zKY@rmfdLCipJ7>Qsy5KR^gezJ3k@w-XKZ5uJpka%0Uxj~z)EA7wdXCXh>3^{4GaLw
zmdzYiI{q@xIKssh0nwrB(S9oPG9iZg+oQtISulndxI^p`-#0PuK7C&1(*Dw|Q+CPH
z`ZSumPb?*GG^ob_KD1|S>+91_eW&tpj4=6Y3yJF6+q*X)+Me-00W{vVAthCt$Ax!_
zB=CvEh(4<eu!jM+>$ExWbAfp_;P716-~&!lacX#olL3{40#jVw->D!v7*N#qfNKtj
zn57@PPFx#~dG@tEPTz})25K^Cm+A#*ObE*p^!~hY7Y4nzbvIHW9AF`{oAkCDDH^Fy
z<4o-5#;+eY^*H)nNfRx>KqPQgbR7HX6UKV)6C5z?g{B(Ik=Q>FHphlxcf8hLU;i3t
zO@Pk|>_f01X(qpytQU3Mo>5ESA+`**bk5LXLW~pE*3an{506EosJWOMR<Ia)(X=<&
zt6nLAQKhFblF19{VCU_8r6Awfj3SPMymnv0+~xkPZn|W&o8{GyDQcLd%rWBO;2dnM
z8=d!ZTKqm49t0<IP$LYul^#~Yh}H^){J|>&IMBH1Wqf#*U;l2Gwaya6J8C`ZPa?3V
z)6mes3+n0pQB)xVem;Mk`<U?7VL*9KPEJ5J$R9U-tw#X9O>fuJ$wU~M5r|E3SHDoZ
zeWg0^XzQl_H&K_(4TdNVqE#6jL<r}mZw(A%B=v)d2{4BC+Dqv2;q#Q<57QSnB-%MY
zZHDL`eC<8uKTD-Rw+DDi$%wGP1?mF(&#Ua90`PN(1P7<xxy>CAYu8H`Pu3Y^3#SiR
zRvp{9!sZ7E#pMYkqHdBH!v__+STR0Us9(loq9meQm(&DS`GtQsfq<4_6!o)zT84)|
zfRoV)NPVDT%Bc2pMSjWouw!PHW3EA!fiwQyjXN)xy}POZ<8Aj{e%{bd_xteVo2Oq?
zi<JLsGUM;Sm6g%MQi8$8s_BbMUmhbehlg>D5^D%HdeFaIkl`0V!^0%!MD*SoFIZ@B
zn^fbP^*rSCrA#V{4HA!Xx97&zN;gRWb9{!GbUiM4#1GRRb|wG#pO2ANEjWDyn+LHS
zK-YJj2n)EMA8y0*F1yLI{@NX62ynuhK=Up76>vNm@qvlWNXIn5B|933t*HJQsm}(F
zhR+^r*?49M)HY)lf4%!-QA{C(1LoxZTGEn;t(Mo_UP(mN-`2uI>Tg~9*ISV9iyqy>
z1QIv4yB3;0(=K()IJ<>pzS;-RMA546M{+*ooLuaf(ex&^I`$tSiU7gw<>eJA>H@dk
zF_W*97xnGiRGE>iRfpicEifm9g@sW;{{SnZVxCeHnA$KkH3f_p_=WND@sZ$Lz~lfh
z^{R^qDj<>`DHi~$dU!koe8LT$*^Ds2atnZHLYwM7lKv)WvZj~u`SZ<Y9|SP5@u9+W
z9I!zI1cbvs7FL|eGF2|ehS9&qGR3gh%`8e)NdjHwZ^K1D0t2wMwKWiSjg5_UbQWI}
zG4b(vY>njsnJ;d-P(;u_Fc8?(vqZv(fY!tZ7<u)a?M$a6HD0U%ur7%%$gpt3miUhv
z_Zw71vd7dcc6-F$T{{@%YR+@g@93Q&1ZL37wMSSccd*~Kjgx~e=>_QI+;6gh!3-<^
z5iO~XC2JTMy8*Wh+*_{u4d++q`xIb)YtuU0zfP$>SeGO;<LKx}%bg@hG_wYHe=d(!
z5#Zs09~T}To&YVNMMsB(hK)_JV0>@7OpK1;Y;$xSjBLQRUeqWm-SlUhh5v4dc3NV9
zdz&#$&zh)8(E-*+;Qq!gtRpHZsU9@qrQ(=S%UEwZMqWp!*EU`4gIrcxVK$bF`{)rP
z0h*tmUm~YP!nbe0SoxNok3+=@N9=6{h@srvZew;<kiME5$uE)mh2Owa<Kap_7s+^?
z`CKt4Qm*^CuQ4&94Lw^yzIq3o>du*abd88J_e<%8$<3Q~J*2i^I!UXt%TSU8N`3ez
z=t-)A(TGpdO(50ZlCQ<&+5x1ObQcERcHoxWQ*e1!{qJ7M*+wame?~Ze7ksFfE*Q52
zBPcf)htv|W%-WtEp@g7gb*fpd(Av5r0tn=9mXG=B41rsvF97hcpG{IOXs})aU48}n
z7;wsY8#d#JbXPqin$&cR7SOP&6B4_At>_{;?FxAo=f~`s*t2P{S57_h8_;x})Q-1e
z{m|RLulZz4Q+c3uqMz5|x<SHN{GRM((`<Eim3r~UCqfb-J<lco&eI94I_^A)+oBMO
z)bae9SI^!HMzN-+Nd+8rvF*QtqV<iC+Y5_YR$PR<6@-X<j=B9B@9N<nSvX|=K7<5L
zgq461Z>YSkc-2`uhr(^8edztd1kkbDdH=mPKxeTu1)60*)nU?Z1VYZbR)Z_RUOT{D
z2)sw)V2a7UJq|e9C1G7^3`AeV5h0hCmp}~yg*;cY<_LI`fp=Mz!E^rJxKtGpz9rfS
z#h5(-EFiEi%uStMYx6D4*u8S-D&noU!X)9cPWz?*U~+I$q>YR21={QD%*_<#S7jp0
zFN0+LFT2%6{GSmL%uQ~SBcnVcHd7A>O6l#{#Cb&eH10=Tirg>R=UH^w!C#<AGzHSn
ziR2NVWEtlH6sn)AnPSG?gIKOfM=c>N?501Nl5D_;HPKSa6Lt!R1JbS)i#U4%R9hsa
z<(sA^jP45!At7;MzX(m40(B233F|nVdR=DV_X9`WFU)QFq!>H_w!bIK$T>_;>L2`?
z|D%}#77W~m4rrwSq0*{qCR{XAG8^6-jAtD#;m9z+7<y@FX|FFYxwVN4#$6Yi;i9M)
zJg+U*;a&lA7*H>0_Ot0p(pf48KuHAD^6aH*TuPJssZh%2Ea8=rI^^i4rlx*Fblu$=
z`0wCFb#-;Y^ML4~G*iUMmpu7jUVu&Y<YLXo8Z`<9$e8&eB3c*Ms$u~<n#zvc0E!XN
zbvz~C9{N@UV9cjE!}4iD4E(&$>m;9S(*><6evu+Wppci+fB+>yAV3)Gu?7EZ>GV6+
zjq9#MpK4f9`mpfs1Q@ctQBW_4c^VA!3rU9-b<h(*6+*)_`~!Y8rF_mG-`vGQ7NpNI
zLqYCbUo~MMY8Y2O*i{b>MvU+~{HVTS=vb`zvNX-w<If|-nEat)ft`8j$9(M*R#5H1
z0A}Pj_p&X6*s)9rN4kTchXZC48z@<T>;x7Y^Y7*0A^n3m3J1+GaLPc`frPD5?`(5_
zcgy$a$Je{nfHreZudB@*LiEQgBp43Ym=`P2vR7xj<Z1_t-3_12qq*rHKDM?(mSf&~
zZr9b9ZR~U?#!ttnSB_}CrdXmPM&A$&sm@3pF)q85dKN%q_ep(KN9~D7lpt6?-rk-7
zVK@=A*KC(QF$UZbqZ$^D5f!{suQN-dBweb+qw~fIzcGaOAl-iXssK}bpQEI<TpU0|
zC?4Q@fr2El+Dcu8_lYy;6Rx{lO>1BD`p27FCL)5#qnI25v$#qRv!(bjV&7h&y0f&=
z@q3HJ&rg{R1fb`qA*U+Xd5UHGsn>dAo?>6&npsvZ%XSh{DQn@|EwqRf^FUV}NVBr<
z-W3AJs;%u&g1Uolmm>>li|HeUrW8%4^V8GhlXKKub<BuIpi7kT!uuUi^3xRFEYp<O
z<J!Am3k&O<r%rlJ<G^4y^(Ybl){RXs56IN(2nCfD<T@Ax^pw~wXSp~y#@nVPw-?m7
z9D$bTb$<s&^Sr$wt6y$C@>SrxV-p>5mcI0JZHL#xC;AnEnbu;}ZHP<@i*OV}u@XkE
zy!12NY?Ja;4?-q;C{eS1#oQ!$Ay&|Y>9Cqb(bHDP+1j?BNx>P9Nu)U^bRD=70s>)@
z{o-_VNei{pD^0SpmcP9bR9z19gEIyVj5JJiKfU|IjBC;8qVche4TG)Ja%X|`{CQOF
z_aj~h0p9TeCe7*t&<SbMU-y>5e(23o)A;u%KYK-r-}nJt49l+#Zg<}2&2=xuRO%m!
zAu0xYP_VIrmiyw_G}im`vJ~zP9E~=HP$4>-8E_^}@7|j+sFvse;d5;D_zv$77LBmR
zjO+uU`Q%Z3<9ic%KQ;X@w}jsWpNY%ffkS}dg1uOl4m=y*-W3rIOoO}%G;e-BN0@?Q
ziwbIE<7ZFoqvI#6c(y-VEKw*ZzHtMK;Gau1!?w>)-lUwr)BE>avujcc>q0_A<nmH3
z80QKKQ6UNM1U<c!*#@bnIP931OcR?9!~;eBJ$LDEG$7T*qhWMC-(nMfoP-=4;*L=F
z3N?IA>wF?4-xw+$a5LQasmBvL$dg~hNGfaEkj?Ey2m+HFea*LIsn0lcUko=-boq~Z
zxv}?5%Hv|_J+5Us)RX_zy+8Fspd#eI!;?r5#m_uSw7Dq^HyFoz`9_{yLg`%P*5OrA
zzWR8PWWXu!eHK*Dv%rwzwp8UA5g{^SS7)cp_E;X69RZykn9CIdgDMiTpT}6qobtqv
zqqyGHoc=YeR7_HKW1KFVTV*c<Px=xy9|Iq1p@Pk!^tOcRqO`N1f;inXcS`7D@Yi{d
zbc?!2o98=<IlyH(HaJ*(e-fYS$w&UUU*!9^V?6%@z=N=W9z|AGR%@&9w<buEk<-lX
zXLsU^7tQH@!1ntA6gQ<@#14_6a_stXw8j*;)l4`DQxdQ1?iKL;BV)m!FI3>Ve3}HP
z=K^a_7dLcSNYeUe?g!^xkl)%=SX`v+)}3ID5F!?)WlVK2`Krpb1mmKUNtNirmA-g7
zWwxUi(xBJz<7Eb&FVb_OEU6%axT51*)wVA5w<?Hp)ONS(prLFxxKnc6$1N)~i%&r4
zK~|onB&~9MJx2;ghzaN;f3m><1qHPE@aRadRFzgH$=cCTtISaPO_*ZTeK$rIIQ?-9
zY8i(#)9PZ|-C>hdu!IED+<jT+?5<@c@$5!M*Z_47=kSQ-M?JbI$$oZrb+Rq;yjxlR
z<=G=!tYn)@3-w09MN@-$?n93bU{|AzIeI<EjykouY5G%ba{jxt+051NI_xJOcvwuz
zJDez}FJwjb8zCjSyMtEIE7UckMOC|J9;ACILn9M>B}SVG2W4iH{^vWDC-ZF#`~o6$
zrqe$SfTH!7-`!a4Yw8c9js1NA*Igd_ZJ#*iiJDWu{IQpE+onj?2Pml!EPnqPegm25
zyu_s7uP-b#2W^pFt66xVw4fK3nn8Ic%hj|*L_}5nj4f-6(rE6YZI#c7`b6ZQ?CcO7
zN63n;{chm(`QlQ@^?BO&z_0qb1qf3dzaEAe!P=GfE$#1neb1(1C@T4wnCPRWuAcCg
z$!cfk=6gAVhzbP{&&ED7GC}}S_b;C>xEJoKs=^E-4D>(C%NfA`S^`Q(bg#Tpd-m`N
z_s)qe+|RPKqHtgMq0QapWtqLHJ5m8n)4)tDIX0MsU&2x@!Wf3wn*&1KQjiWJA|hfX
zv-#Z4PR`COp_5IuhuSP$z^4_-fOaRRKB4q~*m~=zs@AA)bR!6;G}0*|AT15j2uOFQ
zbhmT~DgqLMw3Hwnf^>=q2uOEGcZVR&ot*c3?>Fukmp{%y&r$Z?Ypwar`KzfF(rv75
zYuj$Is%v8{yX=x~;a=q}U`r*uZe=srGCW=*D~y46qO6<KVNOqUpRMlvtMjZsm=Ude
z%^lZ<{HBIp50$DN1`}qgS$-0#oK_A1%b-cw33!)QR#srtXoRK}WJ#Cy#gt}dvs;`9
zkqD%~xl>IY5*A`E(pQI?Rp@t-0=tPkCbh|kiS5u6OIlyZr3lzfRYlNtLdpd_J^j@9
zI8f2d26b*)36x*IqR7xVX-X3=Dhc3=BYm0-LJZEBUWrH365ei}t~G_~$m?bEE<89B
zmu6Ig`umK5kX)F9uvgF!5pW?3CP$Q4I{m{X8vG)ee$t*%;?uD_%~ak3HW&woPG|SM
zkJ`~41tZ!<x}6Tchh}-6p|GZYw}=`XnE{P+h)YXVcY<0rM=>EwI^YcK+Wmv#m#6uk
zzH)44Ue*4j{n(;{#t@w4po}U_k%1BUR&?SKfn<)Pq=9Cc_A~&0fOc_be=W_R{3cy!
z{3w%%&G7Ym$~awhUz@!1wR>7iw5Ltk!_{y6eso;WPf!KCbmtmS9$R+Otan$bpno-X
z`wg!f#!e@C`{tO?^t9q9zWuAqbL_@*u`xu1JQ0s$!(K0PsJ?@-Id=}juEt3LR3`ak
zu298X*IIotvJ_A<ruP1JUv>t^zBSh*J902!W74dcuat>-+<t-36Clx))KtH#mR2P^
za}K)L39vHn^x*-Nwooa|z+e?p$s%>(ws@@26G29Tfts!QuFW&nMosl^?ms2Z(_8kF
zEx$2RMziXf&~L5wiBntcaNb=Z{MFr>nZ@UV{)$sY|Gj%J?~t(ZIQrI~2rSmTyz#o1
zHv&7Em}&S)#qX5S(9pPft8+VLe)Y6eQD~Rl?<Rc6i7e#3@K55zT;OxBi4Ee!T>9?a
zEJNBjJQQ+W|3SyS=1v(V#+u!k{_d`|Td(Z1q~g8na<4A%Np(GA?+RcRQ?#sl5NLs$
zA4@b}o|V4SB&jHfR?hrDHK|CMt{WHqba$pRh4<W7(NLp)i17isICmv3N@myO{WiOR
zifCPBKYI~fNfrji+oK6QBSm~Cmn!aWO+g$1>OU|>-P8zu{prLfraTVQ<NXUCv&;Rr
zcz-PJKco2|+vv8ky!>)^=2E=b!J(t~B^NiA$F?`!vVxiT=hUVhdCS}tRPUnZokf?j
zY-Q~iIbZe}m~x+LVP{Dpf4^$1Za)KXk~r#bOUyE2Z#B4fHHvM0qkL(mRddL~)7$$N
z4mt`{=lb;KJ$oBl2ZyGx!dI*Y4;Xg8Jy-ylbl8=gtn9cQbknu)1VWMvYG9j*<N$I1
zC}ng4Am91VcU3|Y62?YOA0R|yE`Mk;>y&3Nz791mGhPjJ%`FqWulhI!@586}cORhN
z%IcWA;D62hD(}|46)`QMdql|JQ>SX&kY3n63Bb~nre0AJR3g^a_x!syP|*4%JNpyV
zo9bErZp^!4OwtH#au?s$I^3roS+16eT(=tE+8gL&gc9dJey;ddKWti1KIv`h)Q5D-
z%P0S!yT--EEj4OZR8ms<#7R}}p7DW%oMNEzDWXU_rYRu3s>;2m)DD`$R|zHGryDv#
z{(boHA>f~GR!$Dx{ks~6Ut-%#LfaOoia!%3%G0GPmgeQ142avj>NJ!6qB_ESU070L
zXlPiHJgI(cVG2_s7!W>tuSZa$k1*$7roy#-2(?y7NC<9saF38<fp{oe%-oOQ=m`1e
zvI`h}Wm)onr#>AE`Y0*rOFEh5Iu}u2r@u4IQ=D6MdYygnhPJCXSzVj^)a$}CL5gnE
z+vB0w=PkB?&T>_ORMyx|>iZX!U$%rJxcP)@|MlZyVPQcPaC32)1SZ}=5AA~M*m!~I
zTw(L86$xT*__N$;+?b|H5ADN;rrIaE%zAf%zL&RITJ%KAe&KbP-uw5D>WRrSXoQ*G
zB&DP*$}mq>wJwK)508R~Roq{gkfp7AHSq~q2-bI3bU4CRaWIa4)K@2ppzWj~zOQ`m
zM?_IAo~RI$E^Jw|6GwLs<WPVpL5(asN6~=%GmfURG7b?DeQc+c<4o8KLnpgnBDVg|
z+}S4WL36l|mWkI7WpHFZ=b?v96MHybTbK%2cWXYUMc2EBEK#KUb~@_0Z|L_^tAT?*
zL*pMFN?%Q65fL)~SjVsN!%wZBSGdbaf7pGSR%Wzef62vbu51zA2z&0h6%7rG{G?u#
zfS|&GNo69WZh^wUyfS|)CM-EQxywdvB1PdztBTR&nwx;Rlc-m)#D0VIOv5SIouNBP
zjENbHE?^LD2_W5cFm2qN2>;{ISg)5@zMj6qdBWk9E&A6d%D;>xsny{RBuV5xtlsTS
z;dPwb|7K=7HN!P|#GOgqdB-AxoSYm{M7l*%;rZb2(Rp0Lwy^vA*7;8YFAKJ}_^p=T
zX=I-1B=NbEBFK~XJq4kd4Vdn2ENi3H+js+d6HLe=(&oWei5uPqQmTge#e4X%oSQVC
zkB*o)y`aRIE_Y#%v6@?2iWuPp9a;kC-}Cm_U5SUxx?v#!;*4)TGp4*W<*SsOR;zGF
zNsWy1di(_Y`eZSxyu7@rBMOjENAZ3^ZEe0Lg?*c$y4{tC(a})^h;K(N#c%{(wp12s
zH{qS3I_>PqFod(*)<$`uuALFP|H|`Jb+PsQ+>NfTeqnqUQKF|8S-~!Hd@N+eH9k&b
zHBfOv55S=J@aIQ}3hRHuG#IQDNXBbU?<xMYPBKCd@^s~as+zKTjhkq*E`_*E%ro;V
zH6CeZWl2Y5-~qy@g39F>k&uuaXBwdA3a_uPPf3|NT<rtj=rq*DCBXQT)6$x(20jB;
zF-s}@t~Fz7%{RGG>3-PYdwJ=dX(ruogi`GYb`f^A4d0$tYp!G($FLwUAA<<IJ%WtM
z5Wlgp(Gi6OHR5etcSh!`c7^^~%uFWc*=+FX{}>-1hho^#k*j;*3J7{cHFl7=XqM?n
zd16nB01C@R#WC(bHyP{s;e55iDx&Z<C{Xe?ySL6QzJC2)=TzY#_B(xT1PyAnjRhWv
zgmUBXW#t+rgMbZ?l9K8&<1jLsE$AXxgr4RGMcs(m7|)ms4{rRv_=;AgR+uHBT1#uQ
zpxu?$ylJsxk=_XX%^x;t3L)<sX(;_{gM>CpqP*RdOP9|HTqxE7;;$axBi0frrXcdK
zTtxz{M9BH*AaJ#hH6^*#k}FO4bX{i)0Ga#i`rTb!%q%PfBnGyRxVYR5bOnlvoF582
zIQTxheD(R0)Fxt?OP7}+cR)atKY|CPtJev5wFeI#v~+dZfQ=hk_lYJypWv=D>=?*1
zaNpJyP8V@m0;O~xyt68XiyF%d1{NN&z>l=1U6MMZJboSTBNDmgh8+=<B0=ZWz8B+0
z(uEd|@U};));2aYl_2<s``*D5Hgx*PmH=`=!6EVZ>_*R>!#RiIB(utj3WwLlGj%VS
zWMX3DSt_flq+TF(ohqwIZ)I5oqiNVSLl2T2U+jEI`qoE60=Xspi35h6y`FyK((dl=
z;^ITRP@oZVb8;+9Ov?Sv-GitBH!aX13v;fl5Kgs-eCA!3{nt%K=OXV(x;s9~wrPpr
zL*#wubD?Rpstv50q_eeAiTOU~)xR?MydJ-@zdBr4d*^6uf&mEhxi{_|h>c;z?^xg1
zfP*ICoAuB)Zl<Y9+kf391|Z#RXtel}n)>JcayL@H6_=jg6q|4OchFIz3i`yu@0)L!
zVM;cz-iEI?F`VRhp=J;V21M``R?x=xX5+UKtO?j24zB%K&8*;RcvC3Fq8GKm@kv40
zTo+&VOLy37G20&m0dL>Dkt9x`p+jA#dOy?1@4WCD{%XsH3<%AVr&D7N)4(GjSgy9k
z)~)*a=8%{Jqw(Bi#;#}zo%?0Ru}-Y6*U^TCw)Xg!pwE(*XY%R7jgvF!{%7lx4R78}
zqm>aq^!@7k^wVU9l8Cyx@%%*tQ#xIztejkmXT<l9*3>d~Q{@i7t&-$HY?e_#7^AId
zxST-sn6asi=+|>R1ZH$@qqGWciT(li@a|~w6aef@$P#ISUV{OB{%}Y)M1iU%JmD4X
z6B7o_*W!9)DWf8jUk%H+4ka=y_%;ySttz~EB#{E5Jx*LGvTYmeNxyGC)qv}!S8W#_
zojpC3BLh!V7oS*H;2tYHEQ&}~z>tYy#tlV~`|nRSm=|`nH(wltPxV-tS-QXI-`d)$
zd&A7LAwn%6P&b+exsX;1J1<tYIUyk>hBh`fc6Yi~&CqZLe5kYj$3laRwq2=H^O`FU
zqbUD=wGEr9Mo_OGJ!iW2DDEjge=5H#9h4OXTVzpS!kn*J+au9}t?~J?<aF*Nk0THv
z$aIB+;}KC&A#&2=<3HL7C2xOoIk<J3rvAeWkc)`HM}v}79QKQN?S8)+3^iFFcCL^g
zo0tY~q6+vdFU(z#e~c%$o9El#xdfGtCKhDQO^GtO{!WRIZ29Vdfk1iHA~BuSizQgY
zHr(2{B6#=aCX#H;gk$aPTPb)@Q&Ur!EJKw(@eLRkT)FBudLP+4%Y;7vX~B*<#f?NA
z)1@i+F4Q-TVcE&g&o4pgllD4WaelrRK)PkxzMz`)@_nH1^MKbweJ!j(*stKFo~9<x
z=84QRJH*%Bex&{1Ng?5~#(#*vS+uqj_M|LcHHuW(QQlbH%1pngS_3(Js3E5yFHgv<
z6O)vbpOw|RxW9+_`xC!fjg5xQ8i%XvVMd`Agz`;~J~Q6p%g*uC(^02G{rdH*s0-e6
zyiOUb&5L~0Fy)#{=PjnEA+}$9`4K#5)cZ6q%Bn^yt}ZD>{HI;_RcmVImYj>GP)zYF
zcE7q`eLB^U!UjYcqX_5joA}_6ANiQX7=m6V!|l#WD(h_(@4JPCK_osje6fMD#Lv&K
z&syyg8WIFw;v^rr`Jgty7pZzzKs<<|-Y<qIVIR(MtRaLyezbodF(td0@j>#CcUQV6
zad~&wi1QjFybUs`LXrpwdqYYnJsTcMN4_WJF0e@f4I%Mm+(=jUz<i<BR|^)z^gF3N
za8x?&d<B_Mf4?#U1J$%8=r%%DR+dijlc$^-nU2nRj=Z~vht>+Su>S=yDQR*-!s^C`
zfd7SOn-ot<qucsm7f)|*uZpXML4)T2oMOB*pIz#|d|9r0g>2E2^y;G-U*%=R$d^R{
z7W<C}NOP@d+S&+|_wqB6sgl?J&^J8$g=b}zk%92?V&{3Kz13An$^XNc-BWPp@Yf}@
zysG(b<0}ycgeayHvIq}HQqZkv)SIMBGsZ!B>rsXB>+-=be%?w+9FnJ|kAWb$T4b<w
z8rE`Q2`l&*7iS>oH_0&Hxj|`w9)#0rmK&owNlN1Ih(wzLi*(yOm-TjS5?mLUSW%Us
z(XbsY3O~Kw$UQ6}At59v_#=QaJiR?b%Q4#Q`zMl_nHh~#{@=fUdnY6qG`~bn)~)f(
zlszbHCik$>8x?Yz`(AvtV>OH9QfoB#TW}d8@`l$T7q~SVQM&WT-s$QhLu`cNIRS)~
z0P~5(U44FbRt2X-$NcR4Fkz6(-f+g39=dlAbV(PF+(wIayu7whh)EwDm(lbmP@_k?
zAczI!cN6a1-ZqGi<rJc=85N6&hyW_=UPeR1#dTlm&#@K5_mPpp_9m#MxClbDKER~E
zOpD)qBp;vY1TQXi#)e_T6AlXF-SM(TD0yV^BhC0Rw18_J?EMXWKX^dM$_nje6n{-X
znFKz!sUL#G{Ni>s)qQjn;hwD)Grxk(4<DL#RDRc8&<J9`pD6Q#IhqK~`$y))QA1S2
za`!)oE;1Lrw!hma8-$3WkCQD#EyqJCg8>r$+JtyQtf}T*NFjWX$z*IE5q6i^;I%~$
z?&0cBiV@_<uC6ZQ0b&F=R7p6J<dZI%hz{?G5NK|XT@PGN1z0*=Q1mf9BjdLr=D!<O
z*45qNqH~zczs9+SfFKDeqdG3e=Kf}l@G{PT+w5KkOPpdYR=Ytpxi@SPsBTM2c2?Cw
zk(JfdQU<De-X7tyb8US{-YRpP)e42CL-Gq%VU@d=7r(iaGJG+FmZNqwF^cqn4yLjx
zd-NVhk{!rvfLjAlEH0eSS6GCGMw}XFd~y<?iLzI39jw50FXr<%r1tH~oB3em&yHkL
zsk%d5l%o0CKXaaxH2I}#s;bi9VTw>JYqtkf1+XPgK5(H@OuHQ%`u;B%p;lK{Te}Gn
z7#O#2AqC#Sr?VpyL0VpZe!b@!+deXAftYitjYBE{9&2Z|Il$fBJw>tw$|_7lw%i`>
zJ<=-x+3@^MywjIC-i(%$(~gX~gPnLKd0yTeIU4sHBQbftm!+dhQPvQjKh)OzV)ky>
zPQpDzprW)U;Z+UU_hsf!BBkX?;6PiZVEFJHDmoz*RkM@mHAqgzmC{X&jWd><5JADg
zA@uY_L_~1N_zj5WOQj3@u@91Q3512CKCuGBi_uFycu{QuB<7fr1#9c;-1@btV7=AV
zfm_+@O-P8PnOR~|5>k!Ry!cJ*2~b9agz@q57S`4iqFG~XaBJ@G?}N=))2t2JHhLLP
z-s*0~MaSJ)wcs|ISJFBYQ)xe|9nJ6(QWEHKDER`!q@{s9Z-S^-h-;D$Ck$X16|2$z
z@O&7C>O6zy7t!>T7xPO?DBq{2FT9ps0Xj+Iu{)UicF0Ph=)W%n7Z}4;vZN}j%Brga
z>TFj*{}zcwh6N4PvgzvYleLxO)7$T$Dl%*D!PC>w&J(q)3JqnSNbhZz(_P6&Pw$1z
zhP5HAMA<MaJG})`DeRU?41`_nz*s6Ea1Y;~-|uwcN3IgE*=KFvN~El<*FmIJnh<~6
zeK$wjMg`Bn;7_)xG@C(Pw!QgifdAZBq#VQ)?;EW!eUf4lY4$xW2i^%Y3k}p-Ve7Gk
zEV|h_IgA*QpFVwpNHt`u$qH*(*{~mTd;wR!Dk`{OYacA!563}ivG_*vzK-|>w~VZ8
zGWXA+Kka8$nu^U6u71A0lU4RUFAm?n;u&SZJVLr3{;auKUiLBb%=&2VX3OH8Aq#Pd
zZ+hR~^U^D6>*?{@j$$AZx-U(SWGn1L)@)<P-d5XBX&M<_gG$nl$LCkHO4!yU3}vx&
z3I#+xaa{J36~G4c_e0_XCTa5BC)|1*#25mpSCC={ncYe9@zs7rfB*iIGVpwAb054B
z?(Q%oxL3;QGR8V~Ue*ZlmOf`h&XCGIU$ar-=BZfzYI|%Ho4{ryu0rVW>+3i-ldXnE
zengA$u)GMnu^5i#+gBa;qNZ1UH{(-MQaBbF*}1uk6Zk=l!$%%zlOqK|?di1zA~>Mu
z0@0c#nHWP#&FRyunhABa6vkP*lN72(fsEuY-pwLi!upL34U8htmz0587Cv)dVI$nI
z&Se)SAq{&wJ9xe`b8_5)1?}RUoSvQrg$!ksmeCLkirBCg?)<{S0_cwH*Eekn3x(on
zg(4{--R_g5`&o`~3D#L1BC=VL1@ErE^^nbvwzYLF$471($N3o6FIwQ)XOf_Yp>K6z
z;YMX+K8ON38U*BIhWVzxHuUpgF=*-;%9<As0pIz34u2Z-ThdtKbe!aK?j~0a$yF2)
zQV`RtpMHo-7|*aQp+to+q@8v@40H`)QdynQt5VN-Ul`q8z2d8%5|_o!O<I<y_8zl?
zjKnIShZhnW9%j=5v~Spy7r)hz!!vxM#m2-G9Lairz9~`>0@o=iDZ8Q!qZkS#E~7I=
z`U0{|JTJww2G-o_Y=noA@``xkF2xD1Qws98)5D=;-gME7j39t~POIvP&4mNHps*`&
zNt=5K@$y!~#|ple-!n6Y1_lOYWe1Yqnm8h$)MqN6Y$r*v>DM?pt(#=3%VMoPqVJr0
zij8MT>#(yqS*=G;8T%l<b4<j0aP+5D-?vpwU0sPSyZGiFQTOBA;2{yKN{5!!L-1f}
zLAHge>H^TGKF4;Ln3!))D`B+yjTw|idG=SR(xr26zO`9pd-g{ECc^7DU12@6d)&kB
zPQG+7?`mP;zM<Zu@LDRXILhmF+zW)$p#Y_@=wUnq>y^56aq|?_<_|jaZI^=o)dDCa
zKfARm*A>6wE1&QjDUp-J72_zx)|TVWE%A9RNztBkk#vYx3Ke?VSX^vTqP?GV@|Y*3
z_*c2^%ZFk(e#8}PT5JP(xV>r4cMSI=XuiV7CQ6;b5#dh|CZAP6@+7w_Nov9l4y9U~
z8!V`JhBLy2@00iX2315k+h3@wRaI4l=qMo4g!2H9Br9q$_#FVm5MUQ$lGCE4r?0jd
zVWg&p_MidU3{0-S)u3A>)6yDy#CZ#==HDOwjIqDA)gAgZx-@v(<o!NQLv(_4i)G4!
zXFmuyF(YO)2}971qTYGAbQJ9a{mFB=6g+{X%|4diwl>rFG%zD0BR0)&KrDQ0?+vgY
z#pFS)^S%Y#Y;in)&&whr6rk#5)XZ^_$}8Lhqg-X?fE7--$mKv^abJ&B=`SGz9gGHF
zB*ed6P<=7#&{7_-`iXmM);%kb&J9T-7Ot=tl$8$Gp}(xFtAnYbJG|~IRF*pPStxo#
z6+c&JI<V3cIdpIx<)LDV4fHdB<qqtX<Kx;dZ@!Lhutw<lM;8odBg4aC64C?Zam4%g
z(7wgQ#+Dk?1BzV()J;0SYdaLMAL0{5Q?vNB`c05(8SC?Je8cJ~7e~&}yPP_K5TSEy
z!GLYgsNVQwQd%)P#cti_i(%_quhE`}C;hb>OaZz=gxfZ!xEMGpPM4)$1!Bbscf*V=
zP!R_8ZPM2PFJ|)9a(Divzg}Vaa4{L6XizW(`dujPjeh4<O6kI2usQQx1U$fWgDa5>
zj3G)$%@*5g0;x>{9&X=ba1yy(dV0?6`zSFnBcRfNUDPa1$lBh~Nccs}X39Sk{?5<1
z_%B~tkgQn`Zd(2+=v~{~WHW3y>9B!olyOp;S(Fe97sSNm^N%Pg#yB3w8NtVo^Iugy
z4rjdY2wvg(8pqP4oI2K!#i40`xh+l)vd!wYNB5q3^qYO_`V!-uB)R&V87L*+4w`lD
znH&>C_9Yi){pkC(&E6DR#=sC*p4USe!P>ekc9jPBsH;=x!}E{8k2*N`A8$>pty$sY
z$A14Vx-209+B6T2ajQ}C4*#fRqLViCRq(4rS&-M$^MT97y@?G{U!IiI<YZlC<$k!-
z1#)TCr^~N&YIoc45C+4ZFCyM2mlsNJ^PrfUn*-=GFf_CY2@XAv2nb#=&S;|LE!e*d
z{y*PfzA8tRc5>Q-aCmqIw3$cYF8-973O=17!0Db5U|_+bgVFloDmp2e{4Evt$l}Pz
zH}UO7$M$O|Dl~XXwJwYD2V}M9)HQC;I(qqhNag8h@ur*a1tpF&rqJYiY5*E||5XbB
zY~;KS`aJ>N;7^Zi#ES@x-P<WXpNKvG*!#kKP8hwg6^WY3^+&Olq~W1J7CRrn6xY%X
z@`N#$`_G&oBZLI}l0sXY?2C67DRs29t#}8?8qXI@cRcA}1kvZF+W7w=+v33bktXa1
zs%~AzxU=(fA0HpyyrKXckZVJ}yUg?FUrjkZ+J(mEzn<bvG`vFAAK^zF+!ptv^yvHE
zM1`5YO81D4*XA`-LdKT}Y-@I&mdOa}&X?VU5uLw_m<?rPmVhw(R$iW?lzx>Imh<Gv
z6XH2kQcMNA&j<IIs^w_$;7m_)BGaN=jnBJfhsDA1o{_0|7@yXuD9_AU(J)h93h}U#
zJAssJgQF=u*F#<X{#)<I;{M0Xm^$j3#}P-XFqAMbY`iJ`#Xk=?qvxKn<i{|Zw}*a~
zH%U&1XPC-bj%ORU4Vas%-GsJM?vxtUKV)_??tf0r{E&C+@2Q~$ow-Tp`y(<y6W~Ky
zsQ>pf`URf{s_J6|4vL)}XCX29JwoPjiT)RVi2WC-9A+m$XP_(XI8s2cGuz4_w`BHj
z=<Fkl8$|8unWl>~f^g&NDSZl<9cFdeiAa+jHX82eRgW&-mLUj5{^ml!FG2QA8#m&8
zT|vNZh)_z8l>b1k$g|I-J&$DnD?g$za*_-CdV=2u@MBO-fv;1q-rXLVKjJF*#A}hR
zNC%CM^YoZSzZT`(B7*Vt>Np459R=~2E2))cY~nk+Wo1{3@Mz!fEQQolj&(l)#NUnJ
zwic@qQq{+4g4VTG4-f!<fLc<h6tBz%wB^?6LOxX-p_1GHo`Sja%5y{&s>7T{Ap#7q
zb}Wo4%n85m=*-Q}tBi01LkEaHBP2`y`6IMDsgbFy9d+v7wR~{o^=9huUK=gc549sV
z3z7|2|JNED>+4>leY<Dp9w#Tx*4O^zSkl}HEH~Hd$~I5`W0%OAiVCi;<H`-9dfFA<
z2{fTzPEE{X`ZTN`JN*S*_eXPP5uy@mn!2h<Hs4_K+atv)BYxC|tnfai!BZ-0pnYd5
z%kx+%P4HAad^8V7j1?{7?T?_o*M4{?PawLxK+nb1^~IGIy3)fTz&xPU0<HCP+8B`Z
z-=q`K7uy?KG`V<x4^|*!l!$D`<qwshIt(5fbfr~AP>j^f(@DV_!>~u^Bk!GW4FR!M
zadEM|y*-PD{6k1bQ&zsP0zBn;3>O!dy6@=NSg8)}mdFnP?_uBvf>Tdj-HZm$+z6Iy
z=#htQnN3>h^pYtMGDY-|47YHHW=8aoJuxGuY$N6^%NNvFQ{&v;N&=OQm9_OtxZ)dK
z>d@lfeud+A9^Co!LX?=~2=|Wk_Dpd@72MXw-I~NC4=*J;{&GRD(%m9~+kmdW^M*hU
z^uQY!h`q~GFfRbDFze4<61T7AC-3vP2x0u&jG2RDZ*cTtJ{;^I3Ll8m{zHXcDv!<)
z^Sfjh7TTJB49m^64DM1@fRxpkvL=k6s?iB&rHgKJ0}$B!k-4>~%bPx0#uTFH^AP8F
zWBx<ya;B2Vx7H6uA8BoEHz{(Jw6<S&ERH10rF1N2=AqNx88iIVXJl%ARcYM~N9p46
z4hcw458HVHY}QdwQGfJOdV3@0Ch-&4{FC`Qf3tHHrYSBKMt@j3RlQni2wEkeLD;y}
zHJ4}xzDm;-dh|B;zf<U+m}I{1ZD?2Glao1O1f%yD8!O&-m^JKCxh}QY?h+omiCS3f
z2hJ_o&7iMh)%ZF?gdZfkt2DWy2A6FnbmCj~1=iJ7=XWMbw!BnMgj>ELg0npm*~wRd
zL3l-+;V0EUr&Ah~TfdW#FneC0dK+@)jAgnac7MjqhKEHFCRoU2Bec0p-lsXdi{n>f
zDAEhLQU2V&#)bO<SUG5o2ypRD%4#6X5sYBA&tDiJr5K1%BBO{gbg5h!Aj`>hR}6CS
zKjvN7kHm7<bGgn;E$N+(?Bx=5uw^6L5Q^!%kNdOY60N`Fa%+*^&dyHXz#xu@9ViHp
z-jr$U52eN`<}$e^b0lE|j3mE`A3YY^vE-G~?_v86wHN)upP^hX7o_t{Y%(oqWlQ=h
z`}q)O6{Pg`4QlF2ZYgh@@~mn^hcWwWMv{fn6CMrDOE6?G6UuQ{KKA<N7eI`G8X={o
zV^1qnVE2lPF#HSfzNZ!vjl%jHU(y2vx=6`kMyvDr=KuyUKpLUgG;RE8%*-!kM~R_b
zb~&9ricd@=#>M@Ro1>zuD@&xA2<j%Xaji1B$)dQQe&|+QzsvwR>R4D-*6Ntce99x`
zwmG_OYyMTcD4A6FVG$!ME2|K%W0P0poiZz9oz8c-5514Kta(id?^oA<(eU?gib;@T
zrl-1t!Yg+=)^&214HJRVZy2#!n9HfHr&bmSJ}8+O^`7TA0;wfSFVC-huPn~X^2*tR
zsMR%7f~Z)s-PJXx0+uWaO9hs$-Qc1DSvEK{6dV);`H~>X2Uh^8`lyyl)CY8qq@vLo
zAh!h+QX}VZocOdW#aY&u(@uzDKk4EI)=1i_h+s!&)0DoIr$dfFAcCU1myf<g(8`$C
znXnTB;t1&%c05zuwA!o*n;RR)v2vv|wYUBJJ_)ba`HWef<x_WhzGzEd4if2pgMC}P
z#i&^sr9ZiQzhB*hmF=6^0GV|G5F}m82w-l5Ta<f@Z{AXQe$UOo^vJPlH_D_-*1S>%
zNP6EqrY2>L(c>(Q;h-OLd2IITNj|sXQiUabfe;Q6fLNiqf&&FJUv}B^{Fo(~J2*`k
z;kH{gENc3$k>$DS!noqF528Rm;q?66NV^W5LSUhD(HW`v7O>`VFUn?-cv7}^b_`#3
z0|^8KoNCMm9=CHIPD!zv=I9u+u;=7P54`D$-r?U(zb6^Pcv!N*rK|PQknLq<6X`}f
z#u;o_sr$qh_>5y>(gjQmv+KS$#{k0>0mC^p7>FPpb`Kis=|6v9`HPJv=H)+JXsAGN
zk(0BszD`0(85yIht{&L==~&q7YbqwZ+$H^iLXj&E=iauq<&i5tKwf(;J8Z42VAF;q
zL7XUcO@qyi_TMJ*0Q5f4wt<pCL_`D)Eu?R)6GzT4>{%$mLX_$qBRaN(SW-9A6se)C
zq4y*hrgW`Aom`erY~P0Mc{Tuk#vn{2YaU87P<VI0Ff1&A;IkG1&A@fZmQBC)cFiY6
z8=KVO8v7dZb|Xt7e0<lBZEaiD2=^ib&YeF`p(K_zeEy%>U;wun1gk_xM;n`)SAs4a
z&}z`ZP<#h2jGTgkOmzkuru8h3J#m!ZQVzYs|6WxI2w2v6B(Dp(C8qX-QKsGz-5OU^
zZ=F<zREU1qxq<*g{kx4p3JL*n1`KTM8+S@&JeZy)0yuKe0y`&pZQU1Ip8NS-+Sm&}
zr_BxP?roCsIZAlgwmM+muRx4}PbZP0IF>UcBPt$qtM(#T`#_PoH&!#f)p7tGji;VF
zVe!{;#iDacdE(zmMOkW`i{_^Vb#J5Z%4Pmngh}BIp$A#Pii<$V`PL8+2f5(Mtc?@&
z3>h9qR<;iXD@=6OT=U&r#={i+^o3G*P?ZM;We%E|9#_KH6uc6N_`W>DT)KP*@KD<d
z2h7cOM1Qd~R69UM01pIiU<^=iW;$C5!<O|J4icGL)CgC)Z~O`N?PhrcR-|0I`Je8`
z9th?$%hB|Fc+N?S_e6$J*8Gj?MEBi++5Ew*p%m8%EIapmthu*C>^m%;#O-D50FxbM
zKjn~aDZxystZL=xc-<!P-;Kz?c<SGD`<fYc!~}I~Wfli38uXXp2kuNX1#@X542cMv
zK9`*{>_;E(?TvqP=JfUdQJ8?Q%Tqc5kRW+yXLPq1j^}<u=ved(c+=Y%Y@oKZg0E;V
z6NJb}NWmX(ACQpf+56T+O&iVh!QwgXgsM*DGhOCbg3fG=AY(i(9;ADO-Jd$Gx-7_s
z^WB+JdW7B9l-=C8Vmr4gosgE@6mELMeGSmaZSx5Tl-SL(dd^agPazXWCsazO${^YZ
zK0jG4<w&E8-#-OLpxBNrw$tY90ZcWgc;PQ$)5s>>C`fp?r3c2wW}m*8uJQVYWb%i>
z5ssps&iusd$IC=_AoeL%Cpv~k(ck=}n=O4z9?>rU_Q|)*rlI7Wy!6b`d*c7Sh|1S*
z-Z=ApN<L!f>h6X*4#>n0s?P;j%Dedi-C*3QTeGQd3g~!bcbb_vB_HJZLCe^mjc304
zQfaIx-EBqR*H`eywhBktVx}R)-1`bQ>(ayA>J4Yf-`S|7n%a(^KZnWNT3a9P9(-x&
zCJMRwtL7>8#NxL9q>643#gs5Xi0$YF4BVLB<wwuKkO){zM2Pq4*%@H!{?*QsG%&?Q
zK|pr%)VrI9Pb3=Fr@&gLYbydHBy*?XNNr}gitJ1me1#Gzdd0~F1M5_>-449=kDqXc
zRPn62xouMt5uq7_UBTlPk@50I-j0yR*0@S>IuPOMFYUtP3UNsZadA~PwP~HoYbs2;
z>M?Ii8kg-oAsQ?Bt4(!<AN)PSYH4+T?g}!>xlZ`U@WD^oM8S^6q-Eqxb%UVna;`zh
zdXFF~rl425^YJH8^7*HW-S8`<==tgb5IfDpRER|(Kub%Dpq6oIe@_Cy0{AbluC8ch
z6q7iX;q3qjR$RL&-Car%r>FX%(+p(A@R?SB{hC)~NQZ;&52=U;2QDx!Lp*|C2v-?o
zaJzlftSDpOG&L7Pn49dM(6eE)XbY^K+__IEu7>R4P?aiz{r#xiB`+U2@a1CB>l&L&
zGfT_X+J}g*{MBzN`CRBU-mRD|3*SdaBIYD%kwQ(a9J%MlouZTuC+_yYZMj;foH3j}
zv0&7=gaqC_nK5j@E;5QS5u>9zye04kAw2W_6MrYSTkLF3D{!kc_0`;5Aj}r<!5cMr
ze)#lBUtN6|xM{$oZ!~28eMQqwKvoer8K7uF_7g0;;T9Hdnq+<Yw4TZjwQ|oq*c5g;
zR60QN^G`OT1I9dL_%Rjb<=~rx1wnnkeks!y)OqZt#l?+)M1Qo@C>;b(00($eRWl|f
zB>^P7V`bLgH>sM?-MNy0UGFS%Fl<d3<>5Hg4ubRjQ(rLW!qIdqackJSBT3C_``s+|
zW~EgX1iBFsC4gD_aUE=LoEsgnWw&6CF^nqvRxA*fNIT}oJg%?I5Fg7Nlvq^yv|Cgw
zLMpf_r#317a}O{8&CAbVI)MdJm1ahmJzK1)qm)8H#}rOyO9-W?k00;fyBGeJHGw*|
zQ+ZuQuK#&Vb{Nsk$l+QGuL*>c0P#V9J&YMh6%K#AA-rBXlHBX`z+6*PQ?l)vUiRcS
zYurLNYKLW-7tkOjT2Bzs$;p$M5@-k(-HPW53L1jUm>3v<6KScdt1ByK6npwOu!C0W
z7HVzZPJhma{IdMF-XFf27<Pr+<8>_4W=0$zAH$JVH0ck6E{kHCydR~OnN4|S5xGH~
zJwH)pS%xtXAPB%jN>O^6!7vUV_q(KKO-cf`qQtbMSnUV5`gJ0zt3IgU^A!3knStCX
z1&|!G$iachuP-EogaaSp-6)rDR%-cQ+UVi|re*JaY`1X8cbF!0QGJa{M%EzHh})C{
zVLwNYGU&h=^}_VBzw+r!qXAb2mAU=R$CCxnU|dU}uC6Wu-rGO0Lc(4-Pahu<@4tCA
zi+x+FkI^3y`9P0;Lv}deg||UcflY=@8FcDLdwaml*T7z)^ggZElEC)@1C?J`_)WVg
zIvPhZXAgd^xf$rw3rqED%MIUJuoDB92c&5`FKaz_86EmZIoi-v@2Y=AhK5I=H-pI?
ze5D$$@jl+(hG}FN2-W<KoJwG;3SQnPktvbwyW8_CCMFvz7}vpm-Zt;Lt>a;;;&nkC
z+3sw!BPlz+D-q&*)1T@KfAHUbI|vCtO_IM6#tyW@k=8e!2yH__e*Q>7Pe;e<u$73F
zZB~WZud6@YhiM~hU%q^S9Kq_UsuKC}g7PmW`&#RmBl<R%lPTNvV!&Qhd8UNT!Zyq$
z>B2mUxW<Y5V(v%UB<lNJ1%l8q33&fq14!t^*B_uxaL~%Z8PU}Rv1f<CI7L!CHLQ0p
zhLUoFv?(e=DuY+00dV$5pBwyDYHi8@sK50J5va_d4snZif*_XY6S<xS3k{U3KTUq%
z9GpQwH{ZE>sKc!3r$7aK6l@ArP*8AHzzhm}Pm3+6Bu_A)d$eg(sRO3mwQnfI#D9OE
zIKd+>`$3Qr+l%Uj@jGm51iYRhC=@o!>zqFq?(7ay?R6vl9T*<BvmSldcK>l^dXGi;
zvw)QDLOQsN{ex$JLvl1>kE)-+7eEOO)W?3#Wg;Bzz5zX`lZQuX<zZOZ-Mm61gi<lg
z0-2Vl+RU0jNQ*?=Ljdu^s7D*yVho|%Bg4m)yi*>*Z6*U&TqB=YEy}d<LZ5N7Jg19&
zj28+_P}huISnbr=pcKx}F;a3k++Z&QJUFPRs1K#yBjz?YpL!jb0oSOfrw8CB<gdX;
zqNk=d)Ys>@ztA?~&fWk1JMU+g5fi`}zyZCzy$$~)&?14&h-w#HS}ZJU3k#o5q9J90
zP2$i~F!b3%Nnt1VFq$a+yT|srr)kYGsT3K7g@snI%SvK*x~+>H!Rz=eozMA4M~8Ho
zEv$5GJ6xT`(d4_^#YRkQpgh)|cG*O)1nP%Mh>S%z9CXuLni0vElLu*An5^X%Vyc!b
zVHSG|KMr7p1L!jRUg;<P+dTE>hnBsEE5l;}SmABr*6X<<va&V$CHjWvIv7;K7Ykj>
zPTQeRt3#qFk0ksx)vIU<f7EFx*c<tL^Wz0?DWffHFD$SYD}WdJAMXP@*u-#-z7%80
zgc4)417}=d-{7(=Tgd_|%L1z9ZmaMxSJ_l*`}mx{EhOArSU3PU23GTesE3pD8ypNb
zQNv_BTQ_om2M=xt&gya0p!nq9jUM=-7adAc7^zZ)XBN~6-7UucXThCCx((x9T^pd?
zueDSJvd`2VGQ^_y=QGkWyMIWAU{M*SYpop$XFX^9w)(Y{yX^0fy|&VQfY9OIy|mZe
zwV7N3dm1@bZj(0;FE6&0O)!C68J*il^<hAknr-<(z2VcH;4nH;SM6375+dz)=cZxK
zIHCVsge@&9`W?ix_eV_rOi!nY1o*=&^p-G}_J|t<|HUjb<?->UIs?Nmeq}EDQFjud
zhwJfy_D#dS3Yx6hFgdvQMRvZBi)apoz63b@L*v=}Ul|OlJ8GQK*<z|DX?~mtiApZs
zL{e94qwT@GoSYoh7iCl;{vV&y!uNoN_gExo_Il;r#MYWy^0~cy33_+9Zwvm|#W~gL
zqXw&JY94338NQX!tuet3yNen>e}=n0r<|88s^cO0=cbB*bvFu%)c@g*P|;*!{zvfw
z%=r$?SiXLK#YIJ=+9~D3DJ_{<YH8`x)Af&bHW*_%sRTVUfYn%As{!&nBQ{qwi{&4A
z_PdGD@NgywnH<i?Pv|b5Py$E+KA}{P-7>pJ%*FS~NKndtjvL1qQta}_T~=pfMAnIs
zkiQJ%;M+ijP7bMS&I%po5Cj}zz<PanPK$#M!{q-A@2F_$eWl<$jgKE?WgxyMj-a+1
zX?+xaAB)6j3D8-;A7eDebe=Z)s!g+#pL6SUQ!BZro0nlp3T!<CEEp7yS|P2E|F1FS
ze}6LREptCUEGl6;fC6#Q;Sk^6+QK5|>(3}x4}l;R?aMfapLNb~5Eu8rS~e;Q3WG{F
z^cR+;)lN@5ZRp%4M{*RJ+9s{|z)&Lek3u`UM96q?<wg-{#edTb-M`j_cNEX8gGWXN
zx_bw6^COUGK9Y;)%4Q_^45z>61^S(R&$PghfE&q=MGEL)p7y1RFw%`W3@ZUwgV@W^
zXJ7L3ekE_w_rj>M%sp>6+wkZn;AwADp0BcH>pN_rVF;Lm!xpGN2X^B*=(fldDWOi@
zZfiz0Z#9wme=*JL$>@cEyO4$nMVyO+qd@*y-{O9n^p1$<9(CXV+OdYkpS&($q`o@c
zvEp*uXt3;~EG1x){$hT2>3~PDcbCwj4r8Qvuy@``S@o8^{nDzfjSZvM{q#PnMrTgV
zH?LX}tA#(q;mgCrgIN#@aO8rB<RAUpe~*qf4P=Bq7qvVA&`V1T7L2kHhQD0+c?TpD
zsBuRUCnx(^)#0wLevNw8BxhJX$2)%twO?}b@L*wJT>NOct3{m8H9IW-!rpDpIZ8Jv
zf9PYf1k4x3$_#d+MH3#qofO2xo)3&sF(T^jbgvfa<u0kfW>Kr@6SeJ__+l7O9|r>(
z1TW_h(D)*-$%8ippd8iw;n~^Qsi}G~^Gh7K1q6Th9<QALJ#0m5WMNE#EoGqC|Hyx{
z;&bJ*F_+`3hA59)7I&5pj*U$1QP9wKfA`*wS~WDRf=8fWeUOY4(1<9lQdkuRCN746
zkCl}#$BN$j<UdwcYdx|xXs2-Cd-FfH1Xk0<Gx$(?dV7zLj)YuSFyPkC%d3GFwDk+m
z^(!0{+=LX~FZN9L6e1|&E$T?DWl@c4>g$`QkEY^Vf-5Y0aBuYr?mYQY47BC@)b*r9
zf_I&2S-<~F7PP0|AHhM{*BC{G&5gByUK*tdpPg>B`QXma){?UPcUy>jU~85Kp}Ve{
znk#T?m~1O(oX{p1zQ|OlsI1dV)Oc<_vAj%5N-9gqG(4(PqNB+i$Fur1pqfi#a1soP
zvMM@GaXp$7q;eymt1zgtBe{|lyAi0sQ0vuX=k#}TOo`@$AaLza)n>(%De^`>e8Y0$
z`*hx!4>M01eP8#IS%0t4#>Pe$xi)p2cW%zI>=y;Tt?-^Lj3lEny1^oyZy)Sd%6?tq
zgui!uM&)}ML;dc7#9MXSb~BF6O|DHMfmFrZAq`D3*b~^3tbrm1<?eq9TX0LpfE8n5
zgko;t3Xv=hN>qoMf<2?pw<k&R!?Uf+(kj~8+M1dmEz;SXf%N61<m9^AT0dt~zf8M@
zQn1Dr8`NvQEDzU&Jm>(w4|ik8&vsf(#ojy!cCHP5b368Q+4PDihW5sb0P9U3dOoS(
z!fiThZ>WWlw_sArd}vWQ^zZO5!vEC*G;g5BofKjzhyI^$F7|#JRSJ(mZyre{(!03d
z(Aa1j!Bj-}<z|xs2>9WGAw_qZz07-{)y#0tT9TPLP-U-YX}Q{^@fKFlawQf7V7SEm
z^*-2o64LSLU2A;K%db2@ZNZu>{KRCj56vT$S^&Cm|DPMa8I_YCH$dU;-swebcLJvE
zWVSfnQ=Ahk7(pwVM|Wg+^lUWb?ry=rnv<OkcP=EXL-H2Xk|+|+x82L6BqZAqLIDE_
zRB~v6p+7Fn%gY#Ahi`5wYEQ(YM~|Qvf>(?pRs^2_wD~WeQkY|60XPSllywpNZ0(mz
zNgY@{Na+7|+m*8B*0gQE!D<8`R#K9aVWqAC)C0!2_MboZJO!ntJ7A86oqah9WM&o_
z*(ju>H6CNu?WME-1pk9|)UkR`W8+ZK(9n0L4fks?1M`OBdX~=Ksy6FlL_!za*4}Po
zYb)Y+<_1VToMm9T+y&<{APPf+gN!LUy1Ec8!q3341|VXk1LSP10FWu*xi<$blVA5b
zR0lU`3Lvo&>>2Uh%Oxcx$fy`6@haEO_X!9n_+6ec#(DerfOG6IPa4Z(AG_HtweU1Q
z#WcZn?UWYh)n1BbHOswjt_Gh&ptqr=zJN6YyhSO2ebt^$*Ah{N8Mr%7-KXE71N+qY
zj0TL>(Xp{=$f@tt8HJ#e)YY8~L_*G!xex5|T@5J<GgWJ+j|mA0=9Wn|dB~o6>Lxa(
zYDp4eC6{lGl7Tluz{)z<FG3L2A*ZMa2G=+rP&1>8Dk+r~NQ;d`K#CGPY#>7_UebI7
z*fC_o2*L6tn3~tz&fvX+Ah`joD2^m2RIjy}(;)EHfESxRiOIL4f!|9UC@e$mCqfU&
zP>@uyf<X5kQ2or~`pOT3@-8KIkDe3-zuS#yq(|??FLqjbzl@=IUP{s7t46hFpCN~4
zWHMNk8XXf;;rVkXH#gXXxl*0&mCEaAnB9bHmm<Y%=)%vpMz%|N^G?IwO3XdYJm5sK
z({B=7U*Zpy-{BF06x9Girh(ue3Tpequ&Qh5<6||jhiS5~W@^qx7LWmX2TTPsGc$0b
zO^@y+fXE0a5_~*7PeIB+_;Gyv{Fmn#So!K_8zW4hrn9MZGHUdGIaUB`aZm!Wsles4
zw6r0>BUBhfn<DeGG-##udq(baz89;#y7iUS)g7_gTxDX;WlSlOxkF-INP*PK>+O<V
zBC)1!%Uy9#$);#L*N)Z-D<a66KOO(w4V71XlP-dgPlWqKl7#%fga1Ad544A?Lsndm
z{)m825)Pm`_w9|XEevB2`a+NhWbrKQd%}#y!ON@VG7Vh}KpnoIX#ryO1?q4_Df(QI
zFOC1|->KOUXxtnf9i{zMWM4R!S|dSXeZjQ3v2oz<ze)kV&pFAr=F`*S9>hC!%N}{`
zqLcJH!p`crO=fNq*aHes5-)S&Q&DN#mccIy$z1C`bKfhGeOgv|{YlT`sYP*uySr^z
z?&Z#b=4_7y8PRy&{)Y8D=kTK}HDPeuj}R_X(?!RPu>Vfb^B)(<{U`v|z)qYl5qRwo
zOf%MOTlC9!0I<VkG7yVX;NbK_ga}9~{N*0p1IZ`gcODzV)-dzY%A^Ur>;i{UuS5YI
zc+!(I3ukieIq{tIh*U=AZqAaCLQCa2`F<-`A2*RKM>pQ)Xu7z&%LnzGt22M9r&c;+
z`<}7jqu_!&P2XJ3%xno(l#L7z!!M!DlZk=gyo>iFmM>nsI<W*~M`WkkYG6T6@lgMl
zc;BgOM$Srp?o;>fc(y8{&}Xm^dUR;W)ye53-uVWqW4OwM=n3l^iVwnMs_c>uPT6fg
z+fia608?suxk(rZ_Y9Y^W^|njn-?Dq4b3#ith<#N#`nXe(_bP~16RCydGd4*WjJ(|
z_YXtl@xHIC5er}-Vqz`mo0CNXuBvTMo`Sxk^~VqBBVp6(;@q4uR>)xl1%&aLlZ%0_
z9$4)iZrjjLTL`9buqKV#Ni5Ijk*QOHo8*7L$w>C%YkC)^6|TX9{13{sHJ2_VYJsT-
z?ozmSv_Tq^(bdEWB^kZ;d~Pp%?9kzlqp&A&v6O?0Kx&%8!q{(eyb~GeLdQY5h5)qu
z`w47yH8os$JPHauK&Nj_RVztIn8KbX$Qyw`s@cKJ7>C@Ol>D)8Yz@*c+P5N<()<@Q
z{-=#G|7f5(0gvWQwt0oMMAiTV1p*?FmXt&SA>%LCGY#Qf_`f*j?+a(xHPR*y{UY2Y
zeiCLVWnaJY5#sy6&~!ydr(y6IFop@cDqw<N)6##jB<wcBB6rUVRdIfH^974aVfuou
zWy3&80rbcBd*`UFoc{)kFv5Cjz%3-Ap5vhGjiDqm=1AuJH=%W86t#ftFD54D^K<*O
z_;^Ab9Po0EL38fS%Ln|1mAN@6w+SdI(>{F2%zL&0zgS+rZ47I&02wZw#btWJ&Q1(@
zUgp&0`mEQmW?3=#ZCIGhN=}7!voN7LuogZ&WX{$oV?WK{F~H2u)`VijE`ov(c&TVW
z^2Z5TLtULg(;oDhZtZ=$4(>cI<(5^B9m_6c4$A)(z5re*%V6q=)A$AT0l=5ed8ejF
z4}Q*m`2V1L#kyyL{QOC+^YF|KpJ!Aaz`Qyn;7xqV891v*PJa-ND6%%QYAw0`c0~iq
zZ=cIE?yC#r)H>>`BrsZ4q6WY&A9x1AnU<rNs=F!&E8{ufASk@`;N*fSvk-*8=dkz%
z(m_DrF)=aWcPInk8^CWKDx-6I$HzMZSs^rfzks~~MM)JTuYoJfo9O52A;F6#eb$7&
zRjofS0oA&Le<gCS>-8P{nI^x&;^Ld!HE)t+>URapPTI_bjL4CHre%y@XI`->kdAe0
z8+E@Y;S?^yXq_11^J=i>53rYiA0Mx|JOd9KmSH_5r{zPD>z*PVfOm@Z)(&86cFD0v
zNESXyPz2c+fXldnAt4>BQ8ehEbyx7h*u-&OS5+NHY3^71)|<K@i{U(zXjmP-Hye|l
zA@%@$!B&W{nZCg5+hc$6O}P}Cx{!bXn95+|nKKk~xI??m%0Y#OjU5H|R|J(v)68wa
z9qypqLWpO6wXvQnFX#C1T>WU)Vx&L=()6UV2;esA?guyStydWw^Mzo<vVHcGMF1JX
zyR$zKq!2jd85x=SvA}8%E3+_br0DMAu_9WT@`4y(4u05EySZg%`K0zJ$jTN-MBeo9
z@YI-`$s^|WS&0b81SjBC06GpyRtkXRz;_7v00CpMAT2-#ULUs1=D0S&?F`6fPx=Ga
zgu-O39{>il80T3iBd7G-{I5E<45g)WD=O5vuW`ZUz35U4Y6VKES*!@aPQp&omznSa
z%q=Y9KYqONN<MjlD++WCnRqlvVn7Vo*ysm53=s8<)goR8*)XMQ=)p9=C_Jp?<&AKu
zZ8#wLZpHy_G-C3@lfV+BSZen~(nHa!Jo?vQc>S`i*Ix*-w_UJUaFf6+$mepOHI=$R
zR5K?svUh9)HmGKHglT-OY9H+(euCGRrr<bRqDv3$<s}hY%a2FY%a%ka^i=%WSy>N(
z`acLUi0^}M3jH$+fS14N=Fi{6(B?xt!M~{}Le+a=;W=HC{uN=mz}BOP<>gcbuy(y|
zW7Vx#gE&(t=MT|M0A9S=5JN2nS_CMQ=-Aj}V`IFWoX|IcckW&wG~T*$S=<a0DBIAW
z1Wvs5w)YYlG<ShQ*zp$oo^Oha@9QzZ=?k`@9X>Qcdy3M#K~Z*gG59Es!0)3aP@L$x
z?_|{2?M|*MJ$9W{MozE4&!&TD>37n+FuEZw8^dF9FEIs$2Z!SEybTVugDdeCL`U_b
zeIrFI9!#V6SaluhbW?Oq^3AOi69!fBhkEWk_7qbQ!$nyUcI{pL*P+5eOYZdc5NW2*
zPjE76ah|i3{}8gfS3jNlPUbXKmP?O4wfw{9f5GP#g(f1hhYchy2&r_=SA$^#`iFhD
zg;Ln$0@enYXm`fTS{<rA*7|P(_M5O)A2>!h!88j?zQY{zYaaV;O}8-@vM3Ha+LArU
z<IVJeMg@>h=t}%NXpY|y8ET8rko@@3GVf_NuLpxcy=yw06+eZaVjB7&n~@l%Bs6+!
zSXnVBUT8e!+t*^5gLx+yKValm3Vwi!Y@B3||L!f>KKd)&G80KagaJ_vCuCu?IJlS#
zhIL4na4tH`euGeZ(9v^p9v%zXb_|KUp}>-Eqc3ZnH$LQ`+C9@!k%=F%n5!XRNy*o?
z2ipCz(?d&i`qIhix_8*4Kwk}j(L;dcu<qT13taN&(M#wOj6^QI*6{8%n~9tsY%%_`
z2H@aqHT_&^WA@Gir9R)#{m0wzQ9W?3)mMV;Kw7U_j4Orz{QMggaIrWu^@+p0%iteB
zpck5&qd-IW?qaW6w><GX`wLAc7+-EQ>X4BPu*yZF2v2u8MQ?y1)2%?QR*9+neWCHY
zRQ@Lq8m>^SP>A`gAkPqx_`(VaAaJ8BbAQCfW@Z?#4(CGzazR@bc);d=#P@WmGgSv@
z<i@-L&n`LNSHkUwuW<_U_ZD;gIoXdA_`9x4QMA1-)Z1398XX+Ze?1KleDGi%b~<6+
zxdYY?-LJ+i@G6<ZOu*IH>{x8y1R0M&dVv+^YF#PdE;{<x2|FdBBXu#c7LY$O9$)kY
zTtN|2gDs)P-viO8e4T#0zT;ls2(<@{Z#vLmz@Of7f7*AbS7lebnO>h}*!l$i3+xNB
z1tAKQpcW!>2U(+E%?lr`jk4w+<KHu)e~3WnkE}KWbNgZ%<y}1$D(305N%7GHrpkR&
zQ`5qH9*BP(c?hk_I83<K){hZ;g4gY2j_^_;J%g5wZTI*OA+t6+s8)d-X1ezc!lI6j
zh0?t@^5HOJ3jdk&<))dnzG2OfkN|bM9SEVjmKGMmd*6|vn$_D<DoOLD`RvpJhyRF<
zj?n{B`x+q1npAMt`%!V;NnVe`QcXp?g2vN0mu`F|!Lr38uRT$%@$1&~>-~REJ*;4f
zEP3BsN<=wE+oPFej5)Bj+(=L$-T*rLZHpsO^apNi{#$#Szm*_g22^QilOBk0?#>sN
zQ?pl^`%29og8{&zLBD|-2>ZiM&d;wecLR8dF(8ttB=J)0M#v`YcZLW&nBDmINzkjj
z+|t3p0rVhnh$BIP)zEiSiKr(chh>B-&}PF;_SF_b6T)d_Fi)7;n#U=^OAOHh$C7zM
zaZoB3MPX6Nz|C{hu(mz~FK}mizN~|itl4$7qx6d4_exx!LUEgx`i>i5GHHE^!7815
zN4!eA<MVN`Zy=i2CcE;us*k#OZ=(D!3~-^J$065M=?cqkq+kRjzd$zwh~G74=mSn3
zYiyyv(HyT<w<oI|o1vEj*CHkKNzmZKj9FY^**515vkexd;5f_ya0dhL4g1kjSH1eu
zd;c_lfhedg_j70*A>Pxl><J`SN?|Xm8D7N8>ePDuTIUqewfB4fR>5Ak?9k)Vpme!v
zYCDM9b5l|WLJmZ9pNX!*szX-`qtmUS*+SDjJ;uu5S=Id8@W;33s?}{?^*H8FkED(r
zfQSAjfE;iYfV2I8KR@zrEGM^yS0*kY5hp1J-74f|Fob=ETogq`00vICeE<c4xS)KX
z1^_((RODtxX6^=TNB4h1(4j>KOHMVw_t1;Mv#&+fH&oizRsJb9*0g$Re)M}U^k25l
ze7-H=w>zG$9;_@G;qo8YpYzyPY+yU8=Pgz*L#rjyf!V+NAlm$-nMA93=GbS%O|r!q
za$}&7{@wYLl1frvHkU_X0XxazU(9?qnJxFp<@eoAE|V<NPGOBH%>MTwpaDj_r>klI
zPhZyoj&;BHZzQrSQMRVtG(w3=sO;=0MaU*2L}=%3kt91=AuBR7BD+FmW!y%Roy>SY
zf1UH5^PcN{U(a<t=REG)J^thO`+h&`YeVA6Tud*!NE&xG5YlN;9x*|6L^8tM(K}HQ
z2zpIRh*}yL7Zl3N>}^%(y7Y52!u^}-0T;fu*Vm5Tx%YCZapd?Yk;vTbS(<^J9{XfN
z*xn1ct1lN_!KHpMD4v10YGm~56~msE3t_XCH3vNVL-h4d=Tq~{98viIGT-KMOTtGh
z6U@4cKOPD*po7`R7iyVe30HDCzs>I;9Xc|M)&fTAtk2aLa4D%0$-yHV!9tRCA0=$t
z!a6TN`poV*Ro^ka{^Mbus-yD@MRw<%s*fqM$@@M<L7Jkb*3-YAH2(3nd({3*YKB&+
zFXRm$Mh<$CDbc-ZmNz=B_FW#{@aC9&BcuP@fy+jLY+CHCb$W3PdZ7rMt#Ycfq<BWw
za|W|P#P_wX()PmoRaAvDJ^t+YAW&Wq@iCV#&%-I{?CcDq-$0XrWN&Jdtx@V{`&_Td
z&d7ZIxV;gO#?UD&hr3f#V=SM~<;Tq<Zp$uZv&&v~eA(om?h@{GH6eGO1zXH<u|%j}
z$ZEwK{j|Q+&&6VywcCsNhG}}PQL3%wxz|~qf4Eq*bX_DRls$|b5jfNbq=S6-t_h}?
zAufoRQaitoBOF8oAgemdDToU&jVayjSq<*nsZu(AEF*A+{COZcCnx2@z!Q3Z>iEoP
zNa)$A$M+7-TUF-YEA8^VD?Dq}){8zCWqy*IOA4`jyq)cj%Ar`7D$);q&4?4LMV#z+
zaM9K0C#}U+zcESs))W~KI_Ouj{5h9{$Er0G;Hb@w`BUoZ4NrW<@yH>=q4}Xkv#*(+
zXpe(GDf)&D#WB3?ajY{a;M3-$I#hXKlj<=tn7cYoR$NYrRO7BTQ)mrAYhXqLv^*e>
zCjwXxj^pj%2VdLvI$-?8nO#b~Z=rBXiuf)KOjKZk2>7$!Yt@t%9o>!bV!TqePsS>_
z#W)xPJM#0T3ay=-GMuNX_<f11ZEkAN=T{!qt{+=87|B?g#6?AG8yk1-*)usiK}j$A
z{dA{JMk3Ol9e;~`wH2rxr<6kNn1Z(gw!cb&Z<XQdaOQ>xyN|Kg()o99-P-3=#Pa$I
z3*8kzjlF2vR*|qzd%3ne<vcEa;6=&(kdXdiCgR-~H8hem)zHWXx5vc91dYoVsVB70
zR}LQ!eV%WU2HWVRp-)Fgj+Cs0phRl+LA^u67n_`&5*~5(rqP(bHa3x7yeGV#W_jVP
zhldQQ5bam}Se5dzbmQDKtW=lSXmC3&$cxKhr3r<?5E=h1HW=A~1oqJ;1KS_1S%cUO
zTGS^BV8q7_UXky8M3JqM%<Hw#Ja9|7Tq2Zl6%~S!rlwMAcj5CI3|N<DoGJBhJh(uF
zM%m{sV7H&b5u)k_X*$#8TLS6U<$=}<QJs<(q=<_-=gf~x3!5rMdclg@Trb6B|C(Vt
zi!IRR8aPI`L$}j6{nHT9NNd;jgj~8v+|^985Bv)2&g{yD4RnMZZ~L{(q!dB@V7tUQ
zj-!8vMkDiqc0OEGXo5ttj9OeR*oXx9q*|<?pTn?C98p$oh*Bs?OcVliY<><{27p6A
zW<~}EiS|mwb6eP5w(tNSfU#zkA6DRw4ik%*DYGkc@4C8tG0h`F_L5Coc2edburyv4
zy2hyuU0;ga>m*7loOr@RkBrM}(4Up!jt_2~*cGH)ZYgCc$BpH|O{zaS<BjPp=MP`&
zKC8%bSM8q3EfRU+6<=?eReFX*(qhN|$^2-ZflLa?ZQJ@WLhs_I=c>ppCOwajk5KSE
zxNDaq<YlIwR*{2Yr*$5*=5E6&>_yn&%<180lTHmk8IymxtUh}og_cZlmE{eOVBTmt
z9o*XRBAEVLO{Hay{)~;|K8%*W)cy7<xIxIYMAt*3EZZ`C^{#SUiw!L)yk2i&`|gcQ
ztOsY$Ap7Jn>w#d@Yg&#%%SO7oeIYjxq6Ipb`Ap8&{O3(W_E9$igvHW1xPimy=~WV6
zG(6MD7M8&_#pJh>HW}L=C1qy^*W2Q|r{ifnJl->ZxMVe|JWYM3ThNeV>szOUY)niT
z^CpeI3tbBxMP^nOm4KX_oQO!W`8joU77`N9UiStw5W)VqCm6|C<!w9iGN7zk9U!=O
zZ<Fzb-J{Jo*^HbLuqR`<2!=or)Z5U|5K1{L{~b8LSWWr8#})Yx>?2;syP`=)gylwD
zuSr08&xhgZgA5f@i+6Sw8#v`+mEe_iCHd)_dcHe$?<TGWL9;2>KqBlAEbl?Q8u}U3
zeG9<c7xr-wug0|w6(B0U>9;hjFd74SNi{9v(x*C#$L<fx^Y(nj`UJYv%=gdqSn-0y
z?%hrRe+~v+!;8mUuW0f=hfV@8I+RlTHZq}n^q%Q-_u<04Vm2k_TM4vPD2(WsHoii=
z1Pud9q(!9G{@b5Jx5$V7fnQ;+?-9grSfNsJaQ^KH!~_}~3Lr4y)#o<{B0~6ELI&~8
zn(ur365kM7n}eDQx)Nb`e8fv;%EXt_JWTmWe?x#C6EDk&6_983%Ex(>Fy)A?!JQC~
z-Q(MET+;IV`B|}3^&S@mkB})CdV#s@Y8St4qo)oZJ}@cYI3=Iex=Qhk3zde_O`+Q>
zJCpT3E;8jwedUyV_)a0gz}8Ez!bWWU1xV#7KR@eWuY}$lKL>-fFMc!nppnwjHsWR?
z{#rS>#XPLO@~2IPY(#@)Rz_G5uIrZ^aDiLu-)R}1)=LTb=XEA}J68W`@)!9ct86eJ
z2$N<eU@!(a!ND6%TZ<95xa8JIvZMS|-QF&r8KRy5t_ywo9_vqAqbmP?{oWz^DpdL~
zWbZ~TisgNztAwzf<G32Qpa#W$N|*~lQR&Dr;zZ6@HTfS-XrelV-(Y|bI|L7p9#Lv_
zZ;AtQI6P3Y<TM%BvgTfe!rqo4*W?J|5Feln35Vd6-rKl)Ge!tt>>f}$G<Z|iVimzq
z1*mv$Jh6ROn1Jj&PB^ZKsJ0k<^p+j?=-AQEBPRuB0gz8KGcyk3cXGAVG_W!aecF~2
zunV;^i{mePO$K|;z5(12+9OBsjz2cxO6{<pA?BD4ggfMSjuKi^hi@GT8O{B!vI!@`
zCPQ|U)Q&H$<CH@~q$UU3){oTOZ_YA2F!rKj%@z?H)`lF7gXP3CNiRKfzplN@(plyb
z95|-$598#a1y(?hf<8x(zU&776%Sa(u=aN)RO<$Zh0UG1v2bQqg`^UpFw-b)kCEA(
z#pSp=i@k=H1%y1RylUG<zD3->AFcf<e=p9(qRl!SS)-}!Zj|_5_V#Ie=GxG@gKX%o
z)fI7(;Su1n|F-s>+7}g6QDZX9t8S1;zkI&?E-#p!6OtirFRvc!Qpo#o`(r>RRWxo7
z$R7?eZ2h+#<6Q@GLyw~HW?*Eb@q^W9c0RCS%RPDO>T_?0{2o4>D7{)k(c86=X-(j0
zvG<y?GC>!wMg{3b9T6PS4+kc*yJ1K6!a>1ieuTcdt`6|_7>F-Y_d8BACT8Yo^XR|3
z4ie(gEDQ_|ihkJvJsbv4SOfVn^^UfI-muT^8!@T(?c29#X1={C$12n1IT(Ss9gNUM
zxn5weP27OfoD+^mqu{{R?RXKqPARw@eMgF`Px*m)02`3VZQ|ucJ1wQfjs=56!Z`$*
za8rOD2Fe-^($_aG7p$$5)DwogN=o$k+Rp5f;bD;q&HXUs%5vlI2>U5yad8Dz+UKV5
zK4;(x!Z35o(EI4&S8iM6A>xjWILWpux4nLV+8SsL$DL~{_gt#pary~8DJA+e8oJ1P
zjtI!H16neIIuo;7G1E5}%L`Oa^PBI(Tf32oByc;n)mBf6i|INLF@P}e^I^<J{Bcea
zu%%gOPB6b0UjCxV4^rWF=u-95Ob`aJAKWFnD$~nD+qxp;lh=|d!HiZ0<x}#tt>+Cb
zr)I>E-NGf6m9gl*PRXF5a)6P6R3{N$=X@c+_+sT|hl%X&C*Vi89SSpC2Gs*f9g&_(
zOaVDYd*h>3woxkSOjfIiS{X#<{mOtPp-&0+r{E0gN^xx1(78ZCX<2zUio`=ls~=7A
z8SiDvyYlCIUo{3p&U~9KI_Gxss720KXs2@mbg%Zem>*VBLOB`U`eekm#wNqoP3ke0
zDVrS2tgwGBQz><yJKuZCPJH=amv;tGd?67L%p`OskRaAnVh=%#6cY&wW$%5}M36^}
z4}11S)&rNz;_mT#ZXNrl7T~HdyR2)+7A`P0KknQhhVF1}z_tlmX>kHL6VmcQRhW8>
z<bcqa6Vk-fkev>pFQC^ph$C?aCnO{g%RasuxK~keai8&0I-iNW60-F5YvmsSxUk5X
zys_OA<~`v^EMsxKL#kD{P|?=mtL)sq-4gqm?|f4f*eh?|hQp@&<F!#`S(F_SShbJ(
zdv^EZHj@$ruT;Jd)qKM1p{t$*-E0yrxw}!?!s6l<d%3eBm^G4`$<^nPhIp3v&-&p6
zU9@~fpLB~VM^C9oPN|hrk^BV^T};eNlW%k%X$wi3u72WUY+kA*XW-aO&B2osKzmI5
zOlc2er0Kr4jd1ud^U5~&J16n3!i{8OC0}%%sDPCcKfme*#1eSFQ8G|Rq=rd+M*Xpc
z(#>3>DOsri;YYl?!-FUr#gi-yEv2|$wY8G`S{qiAc&bjFD6M0U$3Dcoan!cE$X2YR
zby)jwo)t07Na=2Q=F0AU@CGd%v730aFs_Utn&aW%!GE$%zPUaw9&-EC&#?!3DVGX5
zZ5yi4+IHCIwh1Rki)l0t*jkJ2)JSg{dZrFj@T;sAYPZ*IXJ2Pw>m-fqNs%c`s{@EF
zj!UWZY;JqeC$i9OsqmnBvkUOI8f!Ea1Si!Kdu8*$>VVQA*Gyus+<tpA(W0JnvvG<0
zXkcpEU@IkMrfFg0;$T7X<YQ!cyRw|ICDRx<3J{C=;6IPb+0v^-sStIMo-;^Fy`o2;
zKi<SpSa|1_0p|vDs_+mO*|+v|YFF7l3|kbb&`|f^*W52=!f=t{rNj$A9Np4#|HD2N
zXLt1XS6KGj>b+{c*+i9jf~Vw->{EJwQ(@tkdghXkGPQeMzvc9Iy_DO^Ws=&iqV;j~
zH*ccy_@oPQW|FVb9^;J-78UJROLKYm?)meZ1x^8?1&V6NZw-&e(~vq0w`iIkrlC%X
zQkZ7FAEa!vOEZH0@&S^cSU`D0tL^jbz!K9VX&h|fr)K*D$7|P!SjDq^Y{;|MzN2S)
z#UF3*VQgIWKbi=&5bl(YK~!@{<{I$)_9&vYDkQ}D3&We(i{2k4wOvC*B{r8x2QfA3
z1SoPaM@ws_KAx2S{>!m_4I4(wZ|3=awDnB1vciwSoeOlDaZxu^Z?@N;u4{UBK>S(F
zc8;upipHnw6>4KnK08h6jG!d09#+ReFYbS&@$^6T3e54HYzRVtQEnGH$fM|!A8tF2
zB%X>X?`b(v=F6RchnhezJ39vp&xxF@J4PQ9EV7*XHMT?ZHMeJRkk(y>@jH`yUr+7L
zp4hw5v-s2D_rV-(yqtEw?{KtR6i^<v=zVuIb&<Kl+wg<`8b!u}uUw7QXsB|;+M!F^
zDRjqkFO?5*_2u)@Yu4UQ8`@h^m~qLa@|`HRgvsj~=XdvXEAob&x@E`|ncVi%!Y61g
z?o^-W+HGEemTx~xx{fk^Oy&C>w8sEALK#gZ05=Lm2uZ9Wc6GJ21%J}2&&$Gr_O9Yi
z%h=C5{uC%^92BPZ#h-~jbX7Z=w91c~jeIDqT-AQVMtMQ`DyP6zR@cZc?vr11|1MP=
z37`p|n>&@i{Sh_$Q}Mytnklu~hAyEZO`Bp(Q%m;(>r>+E1>^;Jj(I4_D0wJ(kR1a%
z7WI|rX)k9Piqr;EzYo@CU{Bl=x93#a2HNCBm7}#CsjK(nd5g}e<GKw)lCqs2-0b-5
zy0)vNhkueV_fB~ct{ySBdFEy3$cx|__Va4OLiOKl4L#ig@fnYj>>Uh`(^omkQ$;!k
z3TX^}sk8H^Ug}7${>^+*r!mrJu0OPBsSP*3d6WhNOIGX46BLo#=I@ktgh=V-i{9?V
z5hGaV>V}$wD4^b8UVKT(G91}s2d1!*wtr+V>smC}=9to<d9cAktus@sI{TKEEjRmD
z387bZj%OGZO4NMojZM-DKV>pn^oQ2e>(_n0sFz~=nD|PRHEU^$<AeiAq*N=8^J?2f
z3M^W;)k=irZ><)uU6)!oN=2plgO#_DQGTc6v+I;u<CZlf$3xwp*#ek!U+6xxxaDvB
z`1He{FCHF$c)CVl4J~O`LC2<&KHKX3k3Qs|a(2~SA0U%d>r)=vW<S4eM&)dbww9CH
z(4jTg!Vh!piC-7LhxS&DM7)KeTXp8b>&FhOw;v*TL{iomeki%8_@Uzh!zRz4N3cZ{
zl^F0PI|p80yE6V_SNKUbgTj|jC^AYar`-e_op=(OhMLv2B$ZSS9rD9cS?PR%hU8#z
z&c$>tfAEoixz#?SR6P<37Uj5%nxIpoQ(oe_Z>GQ%gSN!448&$4_d_<EytTkjn_-Wp
znqq9`Rvqbh@x`Ezf`SoGpN#IMUAUN&yQ~}5^gzd(QWuozvdzNN!s4A&_;t!Yo`g61
z50<W_UKgLa^^M<q<R)M8Ta{`2uru@L;}BNoe%ICGfp6_-_~ZrA->j)}sbMZ#8y%c)
zAws3KTA(jnTt_;jb5F?m>TO!Lv_r*Ie+}G7+rt&p(Jd1zxQbL<;4$<eq^#nzIR8Bx
z&e$)1hL#%MY!m++u^9iswn);OTKc#Iani(3b==>U7uRomb-$859|QTBkjRgZFRib(
zZl?FwHx*-D<1bLVCww>r6(wfNN%=AlO<(?7;qg?V_(E}_n9%gbL_*5!1E}bfmKw`T
zwYC?UhNruWuI0;+o0fiO#W_u?mabi?K6uf@Am9R!nuxSbPMh4!ToFfd#~Z-!H9=_A
zy&B~aVuMgr>qc-%Z~PqNCSq<{`WU_b-tyA=Rbu+rlMnlU4sWZvlNMFz_UHV^X;Dkj
z#s|-O*Y&4cQy;QlVwE9v%UDqAKT?zWy=jiwv#7A;y*a%<-gzp@o<*E3a`AP4_?AtF
zM2xlv>Nl9^${#rzF3nxI=@wg=OO1k){MBc@+|o07Et6yy&*j;r4~!G6@*CQdd>lX2
z9`F)d81fzCKOj-9d+Yhn-{Fr#s%Xoc)cnkh(>89`wjY)E%`k3y#obfSDPdv|dXXZ`
z8T!i5QBy6f{zZ#Tt4Wz=+($MgG^w_wF|JZlS-acO#g3a_pp4~3raNse7V_Grs-~gg
zE*sI)(*t9>EVO`@mIE+~I&>LW^Y_yY)`!o0DOoy|{zZTOi->z#cZfd?kzgmzIr%<g
zi}$GYk`0wE&!))Gr#X(tj;N&?)z%wnORgpbDWB`S%k22>==_M;c&G%mUq@lQ-B+Rg
zukzR0qP_%pDf?be=O37hkx`B1YuRC2$!)h|mE(kW?$no4*<fHQ{`vf#kl^4~|7eki
zQ*GrN>FDUZ;*&1dV=d;LGA}*bUZOGEtJ9yxv+jp|yXkG)9XWS3<KO3<jwluv-axh-
zD*kgl&qO!DMgC!5etPZL$Q{~aBWFsr7*SfWptSO%y|J@<6;p{|&hXnD>p>gtZ8>8?
zbgPnik8KOg<NGd{W1rGBzsbzwXvwAZ?sE^%z1`v$N9?pVj606biU*WFU8K|hGFxeE
z*RlD=lb608HLa3-q-=#hR$s@>%(Iet8BkXGT%)C%_mLm-@VF^%y{g4(c9HS=y`}OH
z_LK<Bn%67}q@~MebRRTmuVA<NMX_h+{+!k;aavanFk#F*9V3gYySB&PTK>p~hr`*z
z>+fegE_ssRwi(;(G!qBU`z`x+y_Xd{YV+}S-m~7yOWxo1=d8!#oUDPmCo=YOkzF1C
z^T%r|M%F19H&Whgz@PDz`_r_LHe|o<Q&36We5+mDem+pNDCX}=(s1x3S|$p7RVF`D
zD!*ATVsCQnr<jB-N~fEWz%sewWFvf<Jw&+jdN*HbTAjH3s0E*7Qc_Gt)9(FM#n%rg
zeuHoJCiE;gw2;B2-@JssF?wi}?fmN`iIh(fefA8ss^&C4Th8FmkH+?2sLmTMI<I$<
zX<m8yX*}KQ&`mFoJ+pHz@zSmZdt20rmpRRR*7I_By_{JvKrZ!!NA*Xa(GnVFn}kH_
z#;J}sCf3so(ZL5N*zLNsfF$>Nnm)FW^2#$!cbDO@C0K+Ntmc_|0`NTOowur*Ptk5T
z`q4f&Yo(;5rzUyq4i~!p%_e!thQ7ARSd1e*8#v;Kn_PHN7eSX;Uk-(UI>&nK_O10~
z-fQ)nJ_lb&iVT?v0h3wK33&yo{(s>)+lkBgtW-F`z&CCvGIhu?Y9)x2y&IVV+5T0h
zz8B?L6JT{nV0Y;7@1L3k_#()GDz(t>$FJXt<)lM0OA01!%qZ(3aJ;_t*f%v?GUJTa
zU7pt4{O0fB6-SYq?j$cCczd(H^J$s!qBcX+xEZvJEPax??7V;YsYoNGJ|+{FawsT<
z&FOKygq!Sk(ypc7tU4dj-rRsuj5h^|7uj5~q_S>Y-goIG3d@`(;@Px#c5UB>T^GK1
zkUoseKb86U7vo+fCUbvg|H7{Xs;d>|pL>-uaX*kqcuPjz53=i4Q~CXVFSyipYUM4p
z`SBt<On66Ka;;f2U$#$d+x6xq4KD0_T=kP1s@@6)f5Gx#qrUjkjlf<+QWQ%wfO&)E
zg2@xmQ5UTe^K~Bf-n(PWq<M!DLASA&-|wItj|5!dG2NmMpda>!<Md$=u|mcGv{8~N
z&Gb*7-y!py@4rX&R*tHqXTJO&sIjo>D3f_Gw*Vx&oa^`T`Qh~CZw(K(iklXGHO~O(
z4p>F^O~!lyqG%AwAV7LjKQKExTgB%E<AfoU9i&}IQbD3b*VOuBm#*6EBP*6y1YNwg
z%<Ja%V{@)m{K`G_W85{cyvGrqu-omIi;AB{u=2>TZQ2BEcoBOIC{Yn<Jzr@cWRk>B
z&bcb`;p<G8xM2kCT_$Zx(EPj37ZGwITtY2lv+K?qj>zEnHUs|ohkhbT&fOkb6o9Bc
zOLK!Mt^1{=rMGWiO)@t(2Md<cVesd4|1l`VQ@;#{OJ3A7Fz|q25WXOg0L_!PHcx+D
zPp6|;YgUkaz*%KPY~{6q-T^f|03iT?7DzfERTu(fC@BZWR6r?-6f5+!et`JY`g0Fp
zH<SPKtBG(1f)R$O7!5OIMxhd?NoSyfz}G^kK7`eYosUX%5O|vVbW&D}y@fnp02(=E
z`GU3P!-d-Vd>^kxir1s=`}sBRvZ3tE*|I3{ST0;WK}Sxh1sHzY+AI1h^aT#G_+@UU
z>chSew;yt7{s$wy>|ve-h*#5D{b)zJVb;Kmn-o`&Udjd%TwU+NUwx|lF-Cm|(0zH2
z<`v<psw&Z-t=qRlN2?|*qnBnPth!FHZ~oGl*a<FncFc(Gw9nNynv#h7i208w$!Ot|
zXtBZk?Xk=*pVJjD1H5}Sv|qnxTkV_p$O=OK4twcvhw1KaLgTiwn*u3Tj!T*i2o#PL
zlI39utA2DHHar+B&#bijXLhkE6@~;2&P7KHr0+YG-dWV`F;y=OO3(}E5S*c~F@k2U
zY!yJt3b+Eqkjcpr`E94{n{9GhI%xesxK_P;C%hsZ2FL^|yRW|=FrP5Ds9G2<C!+&1
z$y(blB*g~r;bxNGZxtwWoke<a#%IY^!ngN=dxFHJM?tNW)wU%sY**{N3S*Wt9(l#X
zRQRR!t2g=5*ytC>I_R{Z2?3Db?p{j_2jY>@tiSi`N6RpfJIMM#*%7B4b;0G!m*az4
zx@i3%i~=ifW<B0>gTLYZ`zO$;Bd!Tt<@b;a<R4F(x7M{5agx5f+hO+AvI^LPKDmnW
zn&XfA5nWo)$wK<8{s4D3?Ifk)c`<N{j0~pE=zDLvAza5I|HBn#6@5Luf$D~e5h?Fi
z{-VXB_Fqb+Y`l5TpZguGrlO((^~(w%6)R-3?<uC)=+M+A=;9(&Rb8|4dsBt@vPntf
zOO;IyNdz{2PhrUep&FL>u1K+m9T$|b&HY5>bNo^a9;Pk>+2d0hD{GssJ9xy`@RWnk
z2fhloxD~`|&{+Pq7Hig2?WyOP@F5+_m3vg673;=>_39E>kHd2!M*o48Cknd!9^G+G
zdW(9((o|4jwF?98v%&^2BmmqY;rWjAVb~YYmg5!$fu1Nx@m!$9C8<|&ntmK<(3N8*
zmdI`pk%(fXW#j<WN6a_v>RRq}?*j7Q{Hj_}kX(9?%FnUyKCxgE%lqX1dkie#K&(D)
zw$QtKgYU8x^audfi|em($?8|Br_o&|`Hc?m%rhT5(Rm#UzaifS9?VO?P+mxuXPh=R
z=gF*Ft$H$ieS*%Sd_bPzqSPL2Ne&aPHZH!t%g8yI{{FGNtIMqQ;e!V~S-Oc@(ee)5
z2P<cLO5R-5ztmpxik18C3Rv80UHSmv+{1_3))Yd0w4LT?B8SzX=mXY)v#nK1znz3P
zf7)<osz4x6v%Jr`^ka$dvYa%|3kRy5dmVOh&kihDvHolt>5!1VC@8r`pQwcd$EJQ6
zi_p)J;~Ggxm$-Y~hNUmC-8j9nJQlJtvQ&`BS=)l5S8s8TGnkpyTiL@V4~ogKtxt?9
z8Tt6&&`rl5&mH$2|N93__yY#v%+$D<!=&i;X-SlLe;=rBx}$7pxsR*p6HArT<swr|
zL5-Ij!!U9JkMe+u;@id6)ht&<L*KtKDgS)(SJ+^Zy6-(y)zQJ1cZBP&RXU<|p1Ac-
z#(i|;sj`)d9BlPgTqC$T;OYuqs>t@%i(f)TQU2#Of<_y0hsQiFHWo^ENCt<)&q!(#
z{G=NyY>=q6?|PYZ0fm`oBu2!dzCZ~Q5)$H>qG-OGTVYl@<H>1WxmJFS;woGFTw`-*
zpFGgtpFU}?EKEnVnyvOhplw5!<S%p^u=ryOG)<-ds;FV%rxygBI_2%?_-ch`;_WfW
zceRFf3Cq)dGHmZZz!ZT~!iHC+RC<+boz1VT15!1wdbad5s}8*vw3v4xs?dmpFOI>F
z9}o2PA;-uO_X`+jfa=I(P}R_wv6g9;QBB><TQpDYXYan<v<{^tb3g*^F}U9#7zk-6
zRu`w<Rav2O(W~FKE(i*5eyH+rX-eZ(dCbZ!Hd6zR6E;tNXzehsm+B1u3QXY=x{?i;
zdr@nfxRxT*{z|Kob*a8n=(EqCcL!G9MCB6brNX&^a3UWRhUzO<mdutWL!rS-xzHGC
zZEby!`3BeCesh*?8GN0d_kt@oQT`5)Kbc||_dNy9j8(AKCg;jz^$rAT1A<KG(cH2l
z=eDcJUYDN@eT)(*ymY8$M?5d9Szzl!WF)Kn_@Q@?;jEh5zD}vc`PXQbS^nE7g_LFQ
zxJ#d+l9nPZ2LJ><=fF;eg`LRxp*UJ<>*msD9FYAqTUzqw6Zd?&RJ$^nMya_0$2PH2
zUR>a*w8NF2GTJC*#YDB{Guh1Cv^)v>U5vbDquZPa8tr<*7M`nDuY#T!>nw1$vP#Uf
zz?6}L$FF^GP)RkCn`ink(iy*CpV0jBT61hj+b!a6?t+aKO)s3R5lx@n$Dq{W`1@x$
zr^)BO#m)BBuNEg7o4@~UF(e&l{3!&zFbElx*I1ES`@*jwO1KLBscUQt-BnE2^z$;c
zCsOJ^e_T(Mj+yl_^ORBADzIj2+1L;tf4=L6Vxp`r`$&1!&0#&5*lbF=C_h6JSL&q5
z{DrQY;G(g8Lp>J!DYm^?Q6^C2!BZ)BdHGc&SGtlfeUaL7EV%VG=Fao6s6OP_X=<Bc
z05l#Wz>`tvbYcTU;erEC^%nYg1k;Y$vPr6^<=HRxA1|ATn@@jj5an&e!lJd(OchL3
z%kw}Wdh_mGgg2<e!_nBNujiAJy?f`*^rr>5SwLQIkgNyK=X)U_u~<k{6mrNx@=w&4
zj;mw@R?!Q-eGtCf@5e?RZM#Z|zK5P|1(Bl%xTijWC7Vrv6sw%IZEoVp_7{up-!{IP
zU`%Oej?ShDWTOr--d?g4ikDGRI#NU)3_lP&{=40Q@^cTppe`MDonV+`;E5*6xm;YI
zBsOH%gomdz(}hSO`Qj7;F|Y$Xi;U!3zkV-7ry|@)mv~Zd3A+?jCMk34)2RiI#7+G2
zYu-(!tWwmJJGud4F2U#Xw%}D+OI!t$PG3}_*CbK1lZ4&}*XX*{HjC?oid6&G`SAka
z&sLI}@YQfhI#NZhRn7V`{G(p}V0qEIH|R={-1Sw<GZTU-WIj4lyq(%r|4)E_0K3_T
zsp$d;+p44bc$=E=hn<y7oH7!B#+Qrde~4PBmdU;V;&71{qLIca_Yqs09#2!YRfdmS
zShQ*5v$WKR|9vdSP4W=s4j=mk+{;s@4!V)L+d{W3v)<I8(qKjm3`R5Dxq}dny<sO#
zNr9=Fhu{t|t%!QhcGD<cr*zXa{lag%q(Ul#!}L8e4OaW~vNRMGsWc4*c<2HL=^NLQ
zJ9zpug{H4E={`;juUQwbuIi@x@c0XR!-oz<0>S>_+P73BHc13O7O$3_T)sT>{qw$`
z=Iu8f8qE3}9&Zn%1Us?C&ql96e<9m8b!2LLzNfCDI%NCkLlgsJ3hNl8VgaRo5vj^D
zWS#kNw3lz4c(Q!cNTjS`(~OSRZoO8gv9B+W3`c4V^z8aqo$4^9#+_2a&Cj2{sQZ_s
zOST;!^Ia}*E;rVH1;-|lRZ!a>^0}}UV4xUBjt+r$4Y}r{g|Yd)35h#Z3p3U1LLn9m
z?0(dzlBv{${hXy}Nzolco7sc21m%x;lsV<9zZ8<Dr;1)hud=5;UwgJ^;%D_~S>}5-
z?%A)tSJl$0dB)sz5jrfaHl#J`T3YY*b}f0w_|b+|DaEVDS^oT5jts|279_i0w`A>#
zFlM`|cBjIx)4-7ab8`u(Z`=TD@jtaQrDp^JaatRvz-Eh{Y`KU!0||g&LP1VXzWNA1
zD<<_oVUK@<PSty(mC~Z;0_rOVkrx=*SzTLu=k{%CKRNCr)9;`&WLjN*jNd0XD#~}T
z>sRon+Zp8B?d$x8bB%+!1MjyRdyLgg51dS6MluCX;A~TAoF~Wdh`4^6;fr9l*zCo7
zg`&K!OIb@rBC@h=oEn?3WN)h|SL{S#0*&y8@{2NC{~ru2O3fXy+mPO>3nK~Lr{@~y
zdBfsAW}*jp$c~BcAFXAH+H9YjcZJ!^(U3*TFlx4Gq_!F*EWs@cDg9fmDpRH<2N}`x
z=YfBD(|d%AU<JxTRR{{#%zNS}aP!4Ke?r7TSQazsfx${tnFidZVwf~hB6_Q(*S_f+
z%HVfQ&k6Jj+#(~nt<*n`<5l-Ktwwr}imGbqnq6$+sI%-U49a)#+cfn5_&w_Ui~sL>
z8{&Dn$Q|)QPo?b8`{3vNu8l~D7;ZA4Z@Gz`!RNy+(ubwl0LPz+D;vwH0VqfIHp)+-
z=EPYMK6r2m`>(sH%GLY%01?m`^(63Lk}G}~9*Ckbg;0WgNGDW6FrLEA2oKV0_+{Y@
zGb{FPg-$1#)~if_*xl`&rgl`+^@d#meVL7=4-zkuE3(=!&IK_8=pc(vT+>W-d^R$%
z+gX!dcqgaA!p0|sOPnBpcVOr2SrMAVVvsjBaj3=ZF_%Q+oS?3pF1y{9O6}CgI}2my
zoU>NGLno@s|D%Quv^i<N&eGR)D_a?_f(TDX!b-J4isa-8ypIkC2n2#T?Vz~jYam^}
zFVZn0C%eEzBAsoNe!4ed1%7!?@u#%to7KO6|Fyt<r{H<$)3L?PK$tW7AWq-xjf<FO
z^0M<@eb<vjY1#bM(})<gb{6*(6w*-fAb$&TFSHhcqZ<T2or~+U6u((6VL!jo_H)59
zJ+{3laQ%)lJR4lD$nmiJu@lu3sMwyomM3nzrh@ON@~yMA%4eMUoNyGk;wU0w`XH>?
zsCLAme22^&_KHFrm%a^%ZIE*20q}Tda9S=mi6~mQc4}PUijchcnTA#V6>=-kHbD4=
z5`PX-uy30s?M64snv_iM?eSHpezCp5{y+=C>ji!6WF$kHuiNH4JAU?zb&i1&eO9XH
z+VIj~-8xI1ys<k@9A;mxlpM>waD&JEVViOiFLSq^JQlB@)-!J%r7F!dA7>@;#xQIs
zIZ)X0RKG20Y}C|Ie&J`<rTY|9k6{QNZlcDulINXHI?hE7GRPy5*lbmgDC!1XuOCKH
z&fGX9cA?;nh<mTVn?70h$tO#9H2Q;+_KBHv+NxbR=km@G$~^&ko+^;tY)ZnX03!!C
z{MGYKDt5nP<xJ1k>FWGbK8?;bEr4gH%reU}p0~)<cOfU^1hx;N#>dqNF&9x$=U++{
z+O+t@z&m1R-igntuMf=aliYZ!4(MK|edv=yatOhoa|K;*wPm>xP|xt`%&xB^ZY#xi
za1S#9^L@kgB?7EQOR+Y24E-jzWaR2q&KB|4>$z|C=@CVV<Yqvg%7I*eY`0VamN}iD
zxM0J$YybPZsFK$Ki*<TXRyJ}kh?<NcC^4P6#w%Bn!5LKJyzT`o<UmDIwA?EVnJc<0
z4+s$tYyn_*;R_!rx^eNLg0J4n1LMFIxrC3EoxvXosP+zWqhnjShMB?SCua1@ES!sr
zAGfSHEA9O{@=0wA!SXY2!b20W{%5!kTwT5TL8dkYOTCVX#C^^s2EZ4l>MN-Gl7uF|
zFuXe%#kT!IFABibKmwb;R=?0}znKASkF#+&*Z1lp7lFY0JMrPzRK$(t_;{%Tt<;xZ
zF8t;uYHm#$V1Y^`QW`npC(R9NzuzCOjKusQ`tg|StD!frckb-!Y)#oo`d<7;*=6zx
zabqVcmip7gnC+k6o@YKvr}l0n>S9BG^2(J>F4(DI#`wR0dU`36;^Or~uE?VzBA*Hi
zz5geux3F2tMi=@zplI7vHwS)bmtfy!erQHOXEtcK>;gYOnOxArNg@TZ^_z+QhGN4G
z7Nj2sp%!Y<IVdax-w{B*FpQLe&oL^h3knoD>6OZD25c}kHptDtL+k>FT{}zwKpKfQ
zVY9eR;L=povaLbktJwPjBzC_7LtQrB#+{dEV<gVTWE^89cP*YmaY>EJQ^xODd+uC^
zJ*%3>d9?O62bqEMD|l-=_6-a`!MFY&5MYRZ+^0(TT5Lv@j~x@GY(g*QFTl>h0mJNm
zjH!jJ<*gr8O`JE-L&Q<QwtzLs^wvpwgd1@MX^^WAe6rdXE0m0m6W8{jrzHWN8$K4b
z3$@rk)%ojC1es0X5Iq%{jrvIW+DOX2KzjL-$?p>-zec}(>T>uq9&?A4S2$yT@TWlS
zHYuU;cYLg*w|%K*p+&Q>S9yqMu2q6>Euz)nEnzIUN|CY5I>jY&?#3s=0`QjyP1(5F
zuFre2$3`lgc{L5`W~`M|IDu<+B7cqg5G-YYslZDil``W$byq@SVufxaPfO2qdwvtm
zSR=?&_5G+gF!uu|$9iSR{kEDd#u3`d6wg~r^W0BJRHQI%r6eYS>gJ!$@fN&compP%
z17=OW<oR0gO3IxMB(e1-YzU5;Y+|s_h5PXIe%G7@!zlAPT8sTCZfvh4(vY%RpZF+-
z%xn*D{joEOod?ajO%AnRz2#o6pi}lQmn2BY&-r94)fgeC&;AaYtoR3rh(#KlFl`kg
z8iIoW<(b|Rr}wY7#Kdh=>-qwKpYP1IP`(o|SVD@BXjm)(JLc-P^u=hEU+3@0yV^0I
zqdk$KAxujhOilY<{LMre*`bMU+QFP{d1JfvfV5XlghV8zc)xF{c+<LqKgtW=feW-9
za8K2{m}8Vu_d*tnq|j}g+<K^KVnI^4yjM$TG8E%jS-Y54(+E}=w|Yc=8J;$^PBxwT
zKS)%Y(JRY;X5<@7etkM#(+Kbe*45WwR3Y(z&a3$SU9J##pe>m<W%@2(Dd<eQ8z>rb
zV^Y6+#Q5L~Z3dY(TG|&o?wK6Ukx&wfr{y|{`yyd_IAOBs5SqyH9XojcKVlL^Sr@ud
z;yCUwyw1Yg2CZQ~=B%qBecsHY=b9SL$8TIc^eZB76kc0u=#Lp1hM_vwtfxN~<QO`|
z%zbdPL;`zjgOrW@bfwV6zl9pvvH9knGA+}F-uz;*dyY>|ym<I<-O52JJXnShg`XiW
zen={(J14IdlhJM7$iPDCZj9QXu<%61-d#6)Wa50@;6)#pkZ55-qGKc*z4b5W+*qW%
zPVj}w1G|xX_f|)KAv){@9!`(L^f6D%{AmR5<+ZzKv-+4AcAqSv9DlhVOPey-p7%Ow
z3;AE^4lo4LFE6O(%IOfRxNa0yP)+7M^Mb!L%edmmx6>=@<YgMHlQxyDsR>zTn@QhU
z0G)NE5cNzlKVB@clBZ~1_VNr3X{5Nb?eN0N!aq%dgsbIxmV<xRF$eHFtuY6@HHmkr
zkOgLzysgUbn8|4h6~Im%nWyo@YngKqe-U}W>6K`jC@LGv@bRg&E1*dM-sC$Uo4IeF
zmVVuzLrDi+QXCf(+X&K73*?fk1APXoXI<xh4xtu@WhIy13ps>h?T&Y*k(d&5cy0eY
zI$=~Vr!%f;%ASZD>w7(%Yd7corQy@(5zOHEv$&|s1k~aLvE)B>lHt)fIM){vwS0^u
zLuMmIvXSralo@TgQp9)bu1aR;*!=vILh;g%%$>Q?c5S*Nh4@B+*EJoHA)cOnn11c5
zs|ADC4*6Ui^duGqZSuz2*1}?y0rxJSFB6s1i8nd+5PhDmn`)-<W0QxAV_T|R<(D;6
zIB2$P*%#Oj7vAxmMY<Mf`%?942dE?A2!J&L$^)x?!P-)sU3#+I;jIQx(zX08HY5Z6
z%Z)F3>!?9@N0cW46Y?6u7E<9m$=rfRVlz)99Z51qGm19aI99^YO=rH5<_5n_3pI)7
zVTDu2*a2C$*GBg(!r0(gzlQ6#()P77R?0=W*Q3NMkGOfI2&+CU{Q8trORbfua;k8J
zcWfW|J)fQhF>kuf$N9Re`~WD}$U=VL-Pi;k<zDGLDC3b(=QBQ&(ZaLI?EDl1)`LCg
z-`PqLP0Nk^N&n1eCn2KMWMEk-Vnj!p-I-%>8A)KkGD0+#;TV7v9F171wvx)b#GLB$
zPXR&brJUNi{V(|#Xmgis9elo=oC$6)tt3m2S3v*#Z}LHt!!q%`sl!fCD(4u9XX($p
z+Eizsi=Ma0G~MmK2owuNGK<g*#R_H4J*7^Gs5j40d01fyl?<8kb@BH(q~L$FIzJ^z
zt0!lSX@ew>;XPuOzH3`{mkoqwo!SkZ2k5{K-{n6*KQ(~o=R=VU<vUo~$W0*j3dV}D
zeH-TlH1B}^Uk<LX;q;O;xBHg|U>GAph_H+Ro``yBkWpryLCiEA5nyI`d>UVYkUTo7
zy)#~oK~a0cz!)`xHUv(1e1v^k1F+PJpyx*zToRPYeL1_lXB{mY5APl>UKoYS+a|%v
zYW%MQ#G6GU379?fiTDM{Z>uVBjn@WZm}C#nfKAoePFQka(ygEJj;CC|k)ifKi<eVz
zAG;N&j;N|u!!Jauhyiv(a5Ib7?%|VeMEdZ;!U9FE`SAZ#g9qNYL2;U*<>?`j03RGe
z7!3GQ^x2bBQ!V6j3ALf2riA6WM&EoBi9AYP&)Jusu-$^<12@3W$A|bwQ+X@Cnz}k7
ztRAXx1hByG5>Qp|f;0wVC!p$z;e+27ET%SMQ1JuJ?L^)jIA;<Gh%l7f|HQQ1c-&e?
zeuMWsCAMF0h<Gb)SJz+QkLXD&wF<u4hiJtB8167Z!6w7t#P;=k(ySX0Zw*&!jU7IQ
z2)23eIcL}qG4_shFTC*#WVC1gtz00!krq3g9jQ$6WxQ3)tmi@?(A&tv^DHDJ2c08!
zV!R9k3(JdOL_@Aq!luV3t?@pSFgye8)P8e^uKqA=IZ{$?5dOo^0<>)-pEQ5zMHv$5
zzdc6iNLt7`4!W}T9vgidk%Lul2{D4pQLT!Affz&(J*xf%cfu9qiD0O41!dqC(ui5O
z_Pqy(vhoH%m1~0j)&o6GQ)`J+bMyA?nudneet<?`?8>1ubBB<*`6@+x{<YWgt9pEp
z|232l1-C5^2{v=VRrkVoywt;`Yz!LqGRh(t(R*+Z;d{HsE4La0VLjo-D5nf;kzu7H
zE;d+-)D!l~%X`C@MO<t+pG)&!N`6ewtF0U>0_r6;*Op53<RI#T=<3%a3J?9ZzW`h_
zqI*x^Qu=_kKw<uS9ASUO)rQ9bH6{eZxKun|#+WHaTy5A&aO3y6c&O3h-ckBLb;SFL
zCq~X=Fiq9nO>u4ve`PWMy=HLT!QG2!4=GvM%g|(#UYWnUjq({%@-&FG;naJXDbR&`
z7+ER{K$lAXj5yS2)T8DKTd)GGL?T?xm}Wo(7Qv~IDV8$p3gUuM-Hxmo_X^(+1R;pI
za(@8~drLRUVIaXm9*+DtQ~-$4LkYKM&mL!h>If{@i-AH&U22TvxMKk+C@j%=xad*6
z5QhTz`(R%m*t*|CO&7qTKm)705C*>r9loCpyW7~e(<Uaklnh9D9uvJNW4(TTpLoCJ
z9_Br=lOnw*t9O<xX2R%0%xl?x6bWj5FiZgbMA$%%ksXpBV7Vpp?Ei#Y4v{`!BJDfV
z%u3QmAE61^f}OTmXd}~g5SPL(fySwNYD0j{Dx9P>k{SHF-3qAu*kWw~!alahZBK2n
z2_f=UzzYd$qULyk=|X{A*N~r(W?Gb=o!xAo%eraL5yQJ8NR_~41Q%pEs{#)ZcR)ph
ziEmbT3B6%BE)gMy9~_<zdv`E8I1V9_d)F!@aySOk#3;w$*y_AEG4!pd0C7C>p3Zn2
zp<Nxwivx<o$<7X|JJJ5jMvTIwdiTy98b5`_9#(jM;)B7zWCP4#xj~6qUQDPyuf5<Y
z5cosXq>wpA*#@Z)*n%7iC$6_C0+*nU)Q0;K()S*EI;Egy+-CSVU_$;r_z$i6e))LF
kU4W`U|Mb6Jj2+8#0veg8tX@8EAmK;#sQQsKr8EBj1)y^9EC2ui

diff --git a/docs/dev/articles/examples/timeseries/timeseries_classification_from_scratch/unnamed-chunk-13-1.png b/docs/dev/articles/examples/timeseries/timeseries_classification_from_scratch/unnamed-chunk-13-1.png
index a251d01ddeb83772f99e15966ceeff664f2a171e..497a999fbcb29aa62c72df9d8e49b6348793f034 100644
GIT binary patch
literal 58558
zcmZs?Wl)u0xIX;QNQ-oLcXy|BcPOEBNjC^c2-4Eh-2zG@4bt5p-5nyxyZoJVX5J6~
zHRCt~o4udC;=Zq1k?N}Q=qN-e5CowsD#&O;5bQVv!Q3Ikf}b2wsi1@Zo;fQRxIqvq
z_P>8HlPdua5JUkf%1CN^WgcaD8c*xb^{a6u=iaUgmwb3xYu*(#*;B>fVN4tqHIIvn
zQ=zY0%dD%7gIfzleaxe3Q#$qkY?^Qz=i=?f!qQUS@<(jv4<<PmKR-Qx;voBOwl7A=
zrU2Iejt7PDz(ol5zY+4}kK83G@!vS>ri4d=`EPV#MqrV^|IZL7xe#~gq@TCH@jUt{
zmHR=4ii+%}{F4Grt^R|W(K&mVGMxoM_^Wwwhk=ggII=Dz^p54#QdIF!(6{<_ln@H!
zkcXd5xKVkUnwn2vai2t<@VdITT{|qCzz;|;<Y}<|`^BE#m~#-Ud47FL{gjyaBDz<E
z>@<gOYYjR1WiLm}x5{>u&@Ks$q<`-hm8Z>g{ieNAfd-pWAn3CuCB=W^j+o;A{p?cC
z|M#t(|KDG9*9n(Ip!$1WMPsPGj{N^jMB@48g|j(tb;9Sz2%~ZF$GE2AzeM*RUNf2K
zI&uW&r*dvbkfsIwH`dMi;?AAVn6w<9eAcI<ZC}07lxp3+-o)5VM%}H-EX=s)i92)s
zpJ#ZGc_6?<Qj{xCwaED@Z^HKq-JNQriwpUSP;(tCx-p66^PsIQOT1{=LQMw;hmOtj
z=kU4e3<;Lq@HvuH@o#w+#T|M(aByS?Ex9<w%w!X8v;Iv@q>PkC(%p_9F1%#G#S)6-
zr@+Pcfe7Mq&gwYsDj#PGTQE3BccU3CA&CJTuAC91PY6Np-@h*_V<IF>3x-1qFnjTq
zhe+dUqx=g)RGv)D*_@<C(x<XDsf&EYWnD7_Sq*r}y?6Jg{Ev*18dqTEKeWjGSkqNP
zh)<MHBNc!~U%Of+S<<|=-0b_95SNjj-oDwDl$4~;g7@z|IS9}&A};@J6IoSNRSK+!
zpR(l5!6k*0=m~~QEO>|it+Iwvm<*Lh>2%;?RA6?;wu;`k=^E4Ye!f{&`tb9)Vyl<e
zwfElscK^E_DaBQtKP6vX;|wB0AM8v`{k7j4s$9-0Mhf^P3R?eFpnWYB{Xa`X5Body
zC#(N0BQ9N{;$1)o3Mcs7o}`!jSXxXg63V6T;5ZAxM#aI!eX?Qx>-}L}@_!EnIR>Qw
z^8btgfxsmu-2XW%<f>X)Q|Fug#KgqKs#ynHTUTls(r4tznE#!ct|wo87F|inV08xL
zcE8_Av&~qvecR|T1_lOCUrpejg3tcjiVv^q>g%PYq!ehwQpik$gM**G3U**74vP5R
z*jrd^_C&rwLL(UOZhQIi<<Ri(pW9P*cJ?Rfk;9j!rl!u$9GXc^LS6n-MJl3-+Mhpv
z_Tn;Eanebafw5-XTD)z74eD)dOH4@cxjlJ|h+UnYez)_JkS{~jq7O|;X;|DF1Kuq3
z{~p6*G|O#OO57(QW3punyj2xBFE6iuf0#K{d`V-YE(@OL3Ih&`#p~BqCM|uIT;g_S
zMn=q7$n8gOeqU_gAQ%jDs@`3gt+sl@LCSQOiw$<1oSag*XXO!Mo+ry22x2qEu_K;j
zDM?9NySw~OtF1jfJ=w%#o973Ihk;Fm;jgT$ta5X6S65f<=BiA-Id9m!$l6nMY+l$0
zFMe2Ak)_h`^15SX=T{-LAy!uJeqA3X1$k*0I9OZHF0k<N&Gz@dS!s5knVCsoeO-H3
zU)#%JKbSykQ7+E*i`(<wOHW7V`oZ#90RE5efA#IG9}6XHQacJe3j?CQa?iiKq+RNL
zeSWb$%;;@!EYdzDwHAhoe^b+)nMr~jY5D3^d}=EGix>GtMSN~M%0wSiQd66HeT!=I
z{NghwSKEA5R8(M~{&NIWyl)@AUA$X-sA=2HrX3s_f`t+XjlaH)ceF6kk?(Icp|t3a
zO@3!;VDQ~xiEn@MSMA9rMqyz-vMa6gF~abwiVfrCy?%d65+WiZ1_4JW_&$S!idtIr
zQ)T`AajO{z!4TTytYOFEX^+4gT;5|s7(2nGP&(z$XO(U?Ha3h*OiT<63JMDAyOuVC
zFZtgdnt`b_ij5i?dMPg6(HPPS^~1OtTT839Me2X~@#!^c*tPv=bTmFFFxy*ORi<rG
zd3zNG^-0OeEhlV*d3CdVixf-rg}ArHJ~VoBs^45vRjDZZit>IcqoPJnPfv5d8U5@c
zMAaYjftB(4>dLsy$9=8cziBkvs(>L!(?r<3wLBuCe>_*3nTe@Fcro+cA1Fx8sXUe4
zOf^TWEdoD&-SfXL)7619nBjxzCS^fDhlSi+ZT-W#9u$w&jzA|13kyrj>F;etMLT!r
zn=E*CvlRv&9v&v18?N8q3AHse@KR7ne9DrnJ8JN~b!25_jgRzxsge~|HyL@GrX}4<
z`*HKvFA^i-6g2`k=!IGi_--mMUvhDB8nk-Ve*JpY^wYG?GSXCsP>;qZLQ}^mgCDc8
z<K}3;lUm+-Bt1Dk-eti*6GfV-w$=6eVCMPr=Z1!cw6wJ0*Syj4=m=1(`rX5CH$6Q)
z=);E(`jsrgK0ysw-NyWg>1k>HcW1dyYwbo&j-n;i?KGiA!@pWf%i6ZLQ6N+@ku+PI
zGFIkFD%xw0vZ<dfn09HblJXdWY*(aWe%~xV{N)y7X$-5&%ME!*_~yk;+v_Qh`FMX=
z{rB(R*w|PeA{7Q4a1>L1^U2!V^94NI&37ZE!FFjYX?aZa&zEL1ge<e)6S$xMxZ0n(
zyu4&LYhPYmjH39E-FOo-nAp(Jz#c{}=HNiGJj>PlWBT^hgJNoYdA8Wt+qbPr$>Lo-
z=0#txsykTvo_}8Jzu)Mu>&86!6%=Q*Cu})@1usb^>ho@Yvd}Jj(@baPeX)Oyh87GY
z?=WIvA$K&)_#kT}TUPb3o@uZ%6cvy@Wc{)a_xR^S(d?i6h?G_@pI;Y+<z#HNv{s`&
z&9j#0e=9TcW_Y)@w>djGJ2fO$=XeE(iEg%9<j18y^q7g0%KWN!q`(ez?~YS=083Dg
zNiH_Ou&}VS6nLZY_U+r*N(PcM817cRN<#!FA=O$`R1_aSb<8yNbfv}f)gN71p^vYp
zoNiB6JWp2<g8n}KDK9Il(`Nx|2<bP!W9;Bii&VYk;EiQvbYH45xJU<ea_jimzaq61
zMSAFTu--whikOH97Rnhj121P3FAnQ3HZpY5O>COdX1BFLwcRBZ4`?kcq!kfqZEW=T
z)e{MN8~C{4!9nl855Mua;n>|06J|YbGdS7g?t|fuHEj<ZCoPMW(-P9s7;EwRUAWRj
ztbY=qqknw-b35FwVruh7_;?gfkZERYwDDtqpr&rE_ov61<&Z2V?gPKw%#i`b%1Yv*
zayQ;4^E;MPd_C$0OUnRKer$+J%+kpF&;0!S5h`lAqS_iyk7=-6w8Gf0NLg2HTQm2D
zo+u`FL<^fIu|kJqwPB{|!on4mj}S$Bg;cs!a$y4Q-W462-p8)-r1(A<>>(yXoVwsv
zKkif|U3IL6CdP?<Qa_j|3d6cdR^#m03VYFvkuRSNU@4!;+R$L9N#(+-D>8&jN2pUg
ziEHp!u0G29IKBYY{EvwbHFlsP1$@665~V<MFc!(mnLk25__AY}f>Y7mnU;>Og7j+}
zskn*sY9|jo;^I%gqRe7r2w|Xh%+1tnv3^S~OcsMJ{>~eJf9bTe>ywohT2^n7RZya{
zWiq`4MqbE%da<9@p*!u+z#;UPty^1h!^I}7xYUI-W2K;~K}+pXV(8DMqea1hWw5>o
zZ3_(-*Ts)8R2TVfl_VUDk3@)&CyJ77cq|teSKib<At7Np`->vO^k<wu*XPrswow%U
zSg$(03<(V#9vZs8KBUKsuFe#6yR@FRRZ84CSo{cg9J*G3F*m1J?eOK(I4+!ha^aqG
zFvrn|5wSZiyhOaJ)5bX(?v?^_!qjv~Zb@Ys7on^bBMx>H!2XinRPoI`((hl{2Y*tf
z%V%Kf<NW><S?q<`1*1>JY5tif?8ozGg#(uD3d&L%@DLKz31&MSE^#cic<TM@7<z0C
z_4l+mF9(^`OSNnDmFW_V4i0hN@>D6%sN{aVxMM<Nq{~_B!!O|~=pjIaXy7+K$VhP(
zQFg!Ga-Jcaw2xO$mx6(whm#i$vFRT_wT*tjBEd%PI-in6j&;0yq7_I_PbU)(sHmtA
z@wwW|mbrlIZ1?}OYSO{fcOGdsJMx)x6O;6$8a4L*{{HUH3sO*2)L<S`s&<OW5aTSa
z_4faT$`twjvysDU>%#Q(6+i~Xo`@<<Tt*ISPIi+SnAab9S#;r~e75)&H4RF?e0R70
z*<FL0O5;Y(tXn}aGx1RK?W!&U6(UjgAH$DEdRtaJ$7?&lp)O&4T?awKAEK+R)fWD8
z(GJ*rg3TJjLutZK(ioswNGWoUg^TSNGhQ4EkD+oQA5I_Cc4H#!Iv3&;Q;^k&SKW+~
zS@<|2RL2D-s7if6UL7&ylSY!<{ch+9C&_MP4@1Z&_4grpRA}WV<~GBFx?j1kw=7YO
z7Av%q-V^9fFB%U0O3%nRJUGBhNvRteMzpnq?VOrKNzc_VG0FN?D611?x;oMIB5LQR
zwmII6XLMXbIy`=5p5(UHzEV06Qm8DAu5y~ic{Wo}xK+v{%h;UX*f&)d1y_2Sb?xp#
z;>kCDh=uHkk3-Bj5cu|OQla{R74Cp#ou*rUZeC__+F<V2D}!0QII)!BsnXD|yzJJy
zNElzI;*$sM-(GybWjOjFi-d$k7B@FO{-v_=@ob}elNc^y&o%;$oFx6>#>M74PuT0q
zyrb9lU#OWkiSgB_<8?7=NF7S>P1Lb94I2WV{(@!pi2XTJuE!pJ<S{veK`18BUg0`3
z_2&K&(SE*$QbgpzLF(bp@A@s9>86Hjw-eC>ex5Qm7B7CMB2VRIR(3r?CabXwp6;-Y
z7;^Eyz9w{1&I<om1vZ}I*GnoZO->Fjl*BV88*h}+SE%J<A5Pfz_xCx8XOD>magaU}
zJyP7=tN?Zb`Z_CX<=x%g^t3iHafV}nWe1ahkdV;D<>kkZA8DuX%tAK3939z0D3X(t
zi}?_n35a`Ab>1Pr$1L_{i}Lp|-x@EB!&YsM@zc@q<r3AB*KJf*)G2EI6|yux*mN4l
zHNF|$vl)Z#xPza;Fkn#1q^MO=OIz4|{VqF2uY{ApwrD40=%N&_C~`ab{=)Ux4<<I0
zB{^^C6Ktmje4&<tcG8u6|4nt!X6M$WA_J95u6F%=rEzv)lxBci`S&_0Ta~uZMr$d9
zZ7kA}q;w%)m5)79FZEtfqxDzs;3T~^m!;-+`G$75YTUlQUH?WnlLQ@SOJ6E?EkeXr
zi8_{=zEAK^pBSlCNjWQZ)A?`a7qkd({7Jc@Zsyd@e$#L|e=)HkzBDE_>S*gVFuS#N
zl;n=^wa={2D!LD=ST;l|wi-@>=>+V-4BlZuWF7|kk+8N<fleaeOab6&tJnE+3kNzg
zth*!HESYxh1$0OV{)hK1`O1|mLeQISo$3+a+vBC%&S1E{W7ciu9ewG?Ne}Bi<YG{C
zg%j3zS<NJ*K7l3NXC#b|j|Wu6n2mpYW_w<L*T9q9^!(lTY$LDYns3O`UwfqpnR?|^
zbiybbhDQ0CO09gg`qQ}Ih!HE->Wermex;3?<idHWKh!c45eI0~_hW}L(1gcEXO$IS
z%_Q!klVJ3g-2&M2wV^?LiL{#|Ffed`Ss7Z_Um_9lQm=cQueGpsO4wP_@2IR4r;>;z
z6HU*`YI`%5-B=Mrgxwnwb9IsV^;I__{BK8Dxo2IJpcjdgzF*%9&ql?l)z3^y>e;=_
z%F6O3n7Q*8^uPbQ-7>csg9dd9)3F_GSinoR<EAle00{AL81ME+AEtAte{2%#0v@s9
z%Xb8yTb)1p0^N&#psHDW7yiE79sA6ge>vCq^(zk*l@tRGD3Aa<dbCvn$``sbN?2B$
zWo&X-=jv{6&+@%MeRPzQg+&>=HPY?-;oRNfjBZ<HIluO&jHzL+`dpiWU{gbfwvXvi
zXvyj7ri^O&v7FO^S|*1jm66a_XK&~WjcmWmU4Z!AQ`je&U#Rcx$@EMNn-dl!rJmzR
zGA#;U-OuCT;4(+v;SEJ`hLI-|YVwjEy~gnLW>=zRB!XCc?ylAcUbUDN!()=iK%h;v
zvTr{!rp9-)k13ax*lwL2l<+V!m;Y<IqQ1I278;@%-BXj_SI5dChcDv-Rb>?))^yGB
zc$xr738*gNX()<Sr7J}z{|@-vj~`N?rw)eJS5-LzwkV`>twUN-xy!b<zt3v;^=*Ao
zLUPnaObccMqA)r!U&Zwzts3`c!!vf(SmHF(wQp3c7Bgi!3Fh$y1vJ#u)VR3GB_$<&
zeSM~FK9xpIu~cd7SdlwF)40j-KVfaf612VBPuN0j4S2ktDH+#R%*|rob2}alKw3KQ
z3-4dG>WPeh^{rD|<J{hQG~O4JW(OA+mzSUaA6U$(RW)<hf%ZIaE|$e+Q5}3%SXsQC
zEZf}=bs=-Ew8Or*wqF|;Ma>@9Q@fv=t<BBX_a^dbMMb>MD43a}N36@s%bk|Lf%^YA
z9Nqi6w#uwybo0DOIW4cK2oVB!6F~ReJ|Q@s85d}d*t4pz?R3XA@Zs*cBvv=uMh?()
z0C!4DO9KrzgU{hUMfYzkR){n#r9glXa@Pd9C=LlhIIDh*IaYbHB%RqCR19)4`?qiP
zb#?LZ@r$dfPIh*}U+LS}lvY)-g?8OV(={|Tb@<+%^nD~IB_$;zBhxBYO}6g{cobs2
z89g#hfjDk#c?PGION;6vhgaU+8|(GE>VXaF*B`UIvGwgx$5sdFuyY_6@M9Q>-M#&~
z!pkPQwpwisA}4#SbIObjAr}B|q0Vhr|8NX35!=raF$itS*?XeC%O5s08Bjz_rLB7N
zeQ&Mz?YJ2%+w~vvnjJqkB8JHT)+tfQ$zZu<w?Oe1g#bf`R5|h`hlcimljAG(5^d~@
z*4A`osh*c*nhvavMRL$tMQg%LF}aDHZV<(oX=|pG{4viAOeZxH9oO|SfN|YCxZMeB
zpcFREP3pgRYOS725_H?S^KFh0#C)dU#VB3q^_fU%=e18(^Sf2|pOA`swt~IihR;k$
znqJ){c~TPi(?D&*o$t0IkKITb@cW`JlVH+X43}nS>8Pm_GcsE2{w{Nd`>->Wi_SS+
zot?e4vr9}&WM*S)cixb)vpY;;)FL4vv9ttRWf15`w7;q=D~|`_sX;HQ0t<6;V$bJ$
zQbTEkBy7pY6V|FoyvaK_Hl}7~)<I9rh=rUj;vKSi_nFf|Xxr7r#bx~~qs7N7t+K<u
z-_@06WkW+lL|m31;^M{*=c)nT$jr(T_vP#F>CrY3Xj^rWfXm>fFSb0IuLY{g{!&wQ
zb+yx*CP_)jR{Y3xE-rQk&E2t_r$bM8NLxqexA(VLUY<MkHI4z#(x#?cv(w86DP)Xi
z@7M44Dd6S#uhgBFI|9WY=Qr9$Rh7v4@{y@}I5gNZ!uhLbg>C@b@l{T?wz8TpE4Sw3
z`5i`1%JWt|ARxfm+1Y3(!VqO-e%|YBy$b+0D5$ly6&(G+fq|bn;!pKQu-JL|`T4hq
zZ}|D>N7E}PN6vgNmfiF|<mYepM&kor$<@tGv|+|h-@Y+6;8Sc_TZ){p8HaH7yRmFh
zSy@^CKQ~}17F#^oOAFK;1~@!DZ-H7ltD|>{xhk4DhQ0YMtEjmliVK2oQ#P2b9*Zs`
zQscIOI$@#&MJyZ~WIFr3j3)z&Gw<DyIXDqy5A}_M^|kw74^Qo@7+0y16Fd0Bai(RO
z=wC7M&SXl#lEe+MCr!;F5Uq?)TV_TSV@c*Yt_7qf4)0!CSz7w}`Nfh5{$ORhs;_TL
z&^pUlGOH=65Tl~{(cj;XzR5gpLqEx_Ca#k7&=V0J-aM1CU}pmb8BMPYGdk@GEUu*;
z(e>0!k=Mng=#mP(8<?2j1mK33hQav@Y4~O~zlHT}WBTh?C{Uu4_*;#snvGf%Bg2n!
zanh@FY8-qq3S)IyoOT;O!m`*39fzix8XC5Dc0kQ2D=YhZ9e?x^XgQXaG$9oAZ^r(X
z`L4%JPE5onCxgc>$jzNdyb*{c2W_91H@V9kfVf<{l*%fwWnBbF?Ck8gez(xRUUwD%
z)w;R48NjQx@-cG2*=VprHj6_6ZySHnRqK6eeR7CT1E03k&(!$sUH^D6#Ib$z$SO_p
zr=+yCs?1;0uzhG%H(Tz-D}B5SyMjU;69XGM`l*2pTLo?`_>IV(cdnkn2tclvqK#Kz
zq0>s2+K7^w#HmwfaojnV3HrkD@&^a64Pxx3EUSYhVrch4jkYcorGjXL(b((Z$r3K@
zm%?hDUq10#6&S$60IfKmq3}&`;fhxG>-?a1$ncAGlO(}=4kpr~5u)9b4*`Snq=E~n
z$-EegMEsXgArY7X`kEfRO%(aLHl|MH(F@+g_e(r=g0hA_pjcXQrGa_;XHE!Okpy0B
z4GyBiG_@LAl}?Y0nF#~PB<6RQiuge$Z*p=HbfI#+YOzZKI|}yOnu4Q-cv%UFAPI_Z
zo~L9PBOY#UhY$YM6&1$Y`+N(_JoUATQx3m|l9@w8LMozM`v?gi?!~m;?2ss)*-J}z
zBdRr(XKy}-PCtLcBO4p(DGBejJza4E^ciS?wzjtKghW9H;pTQ5{~|d%J8M#TAm{w>
z78@Iz&7^sHeqJG0{hhn}T%}QQdHJ8)j0tpxD9zI8C;VIuapeC$>gMi>2jXsK=5mYY
z>9ZGf6v(gyeWj(%HLl*BoiNp9whRnq?CiyCkEF&uf#Oson~^=dBxG>1U#$uc7nWHl
z`|zSsQKDOasyO1}RJ!b`Nlpld(VprwXmWESt>+y}BeBK!r?1Svo3zrlciHC<&C2yz
zBsW?jtzIj=&Z5S#h?ttxvOA28b6-urvYxAv944bR&JmD|dBuGyM$$SCWHdcV$&LE2
zh9*GGT-exl@%Zr;Qedd|n=5}xpjB!X3<K4W)u{2qMO`u}(&<8dak<~`n46m$pc3yL
z9xA=GKo0wXiqE#SwIv`Z=<V(8&KDgS`Mu6f81kxmc6N5=b$yVWm-qVV@1rkQ_EXY2
z0~;IJw*WXVFR2GD{We4*qVSx1UM8jkC%$}Nu@@u=2nZS)8Yn0z{|-|-tsy+DB&hOM
zR)x{g!>MT|e<hIth8*B|dB&bsQ*#ESHslZ%{hD>7lZ6f#wp3;WuZPMERu{|^!u>g#
z9zHy4!lA*upB@Ia0_N)wL{O1`y<mDw<wBZR%1*yTiDNy@AH@+9m4ISolsbR&OihX0
zp~RC>IM9*&@T_TH+TH!7;Zh&bK7Eqj(A3*Uo@ugF-JdmF;`b-r81(XrbU~$_uUAxJ
z$_z*+amwm&9_36oyey=VSg-R^8Y!C>>8H5$7OpK&5h$zI5Ct&C;}aPjZBPTUd}iI1
z^)HM9{{+zBIBcKKwt6=$|G0W(vrr^?BWzam1v!estF-7O${hv>XgV{x{y>A1boYkG
zLbvHl$hc)IEXeY#F>-MF(_SU{3u7WBSyfLnw)x5YaJF(Z)k&p+%jm6zKbNT6B^J_Y
ziwj*8$_!}*fRSLQaCFU|`J2GWGbH3{GNqA%_R5-+j0JH;0rsX)TCOC8&?2d*SgNbS
zou8{=W_n^QH#$ihu+lU)Gb0-K*H4YDq_y?oo6DxqQ=vM8nTil-v|z0MJ}uG6_nGv@
z92^-L`T3KC!#pT5a?tbW1}IJ(_5wI~ou6GKih@#ZxUtg)Er$jNuWKT}kpkj{mDTn}
zcZ6p>8(en{;F>^ZoyeD0;wpU-WMXEHsJMveRr*r6-TSL|uG;J=U-awOuRZ<!TBX`-
zsdo1E%Q`;*Gr1`Xes<+4K-Bou)yxYvCT#FJ@Q9#sUc<2UoDg86KH)7cE%A?Kl8N~U
zKoyySkJ^5}a(w@ExwL-$stq>(ZRi+h%+mbbuT>M9`0<I!5K8P11{LW?%DC_aj5umg
zO)VniU_gk4n%jT?aRkZnwIuiNB8Q!ZecL@+I8}OC-7B`0g41x05cPN*o~p;%c&s9;
z$i0b8oN`C<QYmL?X|bhg`~H0@yZ-XlGBy&Ej)`GeS$=*_xS_q`!pOtX%c8$;rQ0eM
ztN+-QIe6XJ+H{Cvz8r$5s`X??g}jUn+v_WC<_`Jc=v3a#&V5hO+q1Uavi~x7dbBOc
zw%K2$-yp_gIb|M!d5wPBbJm(SBxPW;n59qbK2uL9Xt;xCr^G}|l#Y5Ts+nm0UQigO
zGZudkxF-qASpcNEIA#0Z*0I>y<lh{?K{QH-2N-x+%I#@pGPLo!qTWSoEpcICvg+#N
zv|3st1;x_u8jQRjIQ1Ok&eFeH4r2TJ`(sBY9d5*@rG0;uZvs57uZ2o71Z3Lki<6VT
zDyo@PXwAl$MFgGA)v8Rq{u@d<mseR`Z9dmbY;1BGU+pk1o{E>7-R%ZPmN>kS;o(h9
znzz&x(!llrB7~NfmQa={aQ#+S*=T8dcQ1iwvhIi$LxfswDnU;C)+sb)HoN>|U!Pn#
zA@{3R=2oQ~@7+<bqr~mTmzRx=jO?tf8(;TEEtL^{`2v|HM&&s&Ffd#dJOO67vttb`
zmC=YtKpVSOUMP=16dN<q=}7k++5eXXNGD>g9BUAUc);M}1O9@d?(92FEiDV~exxJP
z^+#c6m#_l8!yNI`pGZCpUrF%T;h@jL;(UC%1_ldbW9(MzSdg=T3l-3*FI#%HU8AF;
zX-%8RVsX37{`vb)cy)DkpuA)-F+!dC)Q>-ZD$~;!vuEi1M=+M==lyil&VE0Dqf5jy
z%~K=TxGE+mCl?tRnNn0#PZv|{vVdGq`l>PU^Ra?2@nV;Z(C0b=KE=8>2h4+J{s+b6
z0#)ko^N>Hv{6d!HHxi~}y+Oa*PPMR*FHutJPt_6Anq96B#0-<#;1I++x`D>dp>;gm
zM1IHC{t<#wu|_x~jEWQU%DSekcYm_O>NLM69JH19muMdd>EZjbMu<AC+71s9Fb^;r
zUgu$ukzCyK%9}pZnTqGEn_pO5)HKwX7`5NHNGjNK)dssgzrDTP>~2CwK5kZNP(aMD
zsI-*O@90}}G%5rfg-riDb*IlGD~~6jOFF*vl{~fhQ_w7A9p%fell*DUCf#mUy>$9P
ztDkmPw)m6q8)AG=^y`c04y<Ehj)Ss<Jz+aXvxNEK5o*0!6D#ct9yW<{_M_gVpSFvF
zm3DM=M8!CkfBx3o?GG9;84=N<;F0cs12VU__v1azvb1n>y9B5hNZfyyf<=wJOLB9c
zlaQcw=t#DV;={&{r6dYEuS=+TJWU|Ohp?|DxRBtZgsmC+Q7&xHRvOj3LjA&kM)L0W
zUamNkz-tQ&|JWSJ&23=Fc%O$xgdX44wzCuK!nLc2orm9!bg{UkWI+G48hiv+K*_Es
zAQqj|`y6kPA>(I@jph0I<+oHqLO$oeqruaeeM&8>K$&_23#l{2k&9=8-4rQvuiGp*
zXpkf9$+q4lJw8FIt3yUe$Y{V=Sy_o*>$1*+!+~a72B=U6$;@yndv!A|y3f5d^RQ}`
z5bf^^5Fmk<JXvi+t>2Y0L}9swgOer@v^~A&eNvQQ!-kDRdMAW~|HjHnu<8QoP;C7G
zL%6YMk5I@S)9}^nukMGllarc%dfUfaFmdoqO&Tu#G}j^ja~QyGMnglRvOC})K?<5q
zZ?5gA%y4pYGHb9+7GTdPek1U@=fiD0e&)LdHJnsOA>Vej9I*sGKQVAx@VYrD<C-L*
zw)kubpk@EtzPgG|6OHU;?MUO8y}x_FIq`$rY?AGTWBgDX^01xR+S#KwX?xH%fNV5J
zK|>>BZ>haF8w*DP4QXlul_J5j8WaSJUNv^b`-?iT?-nH`A$48dMh*&z<~lTP9>NZf
z<DGy<8p`|bK!X$aJBla?8ca7&Pw~@}QFD%mhritZ9Dv#llWmp>>rb8ZFpb+tS=667
zFhk}53x*=1-%q;w`u-W+KNv;2v`1=LEyI8~h8IPyj+fTg*Z21J+6J8lk?##(I-Wb3
zzC{Vc#=(J?l+0?|KRC!1az}y6%FE+Mu2W=6l==AmZ%<4BHc}H+{N{UGm6!J`(g`2+
zSjND>8#Minr6VOa<8L)LS2i`rm;ubt`s6t^s;jFQ>f}R4X=&;C?QPsHYIvfe@68b>
zA76@n2Uwqk#Nr@Tqve*lT<NFlf;YAL!zrxMRvu7>=qFV;GJJx`&exuxwQ~bp2Vxc2
z;(kHIO~J~+alISQ>0hn*e7M6EX_GJ|^Gk)57&*^RO})Q4m^sAE(gsZ+em;WZ_mq^~
zdX2d68GEb%g6}O{R8e~%s;6V%;^Hb<>UqFIuM!2`X_|;lR|tZMC$PsCYhHD`aq>l}
z`YtXa{=tChyxf`tnq`7b#dB&o7T>=Q`s;>R$blxU<8yPjlEN?MkLFiegy`jUO`_h{
zga%P4I!#egQ|nb3d#`r}e=2?!wCR|BFg}K{r`c%<3z-v`>+eojgnGoRksw<}wKUtJ
zsi0V^MfAITN`Y<j?M~W<oN2s@o%|moU(k03lI|S@D|q>gu}L{UIADc@uK)Q2)Jmr{
zF|nM}U%$BCjAn{YgvQ2>mgv|m-yp(uCUvUB0q6DpiZM%7PhI^iu~10G_0RmG*xh>9
z5CXTI{?RV34nV5eo5yV)e?>~FKSm4e(@5*(O%;@C*X`3+mqw(y%>7-EdWHVwT0{~z
zLt0CVoZH=&HKHl7_rlC16J}k)*5_~c-FYA4UtFEX=@~3|F}AFyvbr5KIWf<NL?`nf
zLpt1QWc*s1%0-Gf*@9YiE<GQf5C6Dfn%e(p>=>4=`=(hM4lb0K-_h^)4Ic=xfJ3MF
zJ)$}`RDmX;P}8@(Oyu(cCve~{CJV(mKU<cY{vKh)8+yrq5oKRIDx+})bi8WQwjUh!
zf7iQ24YEMx)0m%}{kaVgFYxI6QmSAfUfu@p%bmw(zBD94@AgJA_yISfpja=5r2^J3
z4GoP3Lr5X>hV*@LrKYBz*pnLhS9%(HdU7#8f_@i#e0*~99CqUOVPU;vX_!9uW9b=E
zQc@s|fDZYI1rF|W*F0fSDD)%`03pW2-xc`%wcYYkPQ(JPdn&Sm_6x{#Me5pHbCpPW
zZXX^V9suI=I9U}Hz=J`O3YI~HzGw!&L<q(zAfUjJc$0ZVPM!x+64WZJ&LXfL_F*A@
zZ*3NiPvr6^{+2(F3h9Z+EjXfBFNSx36BH~?De@&+ta&2}!8@+~+S#ES?5Cxqq1pO9
z#QUY->dN9-rqJ&U@0iS{K@j(`d;7N2#aLQ;bMZh2xF$e>To271H9v)nPul${6TC1;
z`h%7KCS9q#EIV_^&B!pSD-(Kuf)wQ2(uf&wzn2_>ZNCJc-!CQm`@5BraGc|tH**BH
zSP($D7L7>3k&Mf6!NhZ&zxfp@X|EZXjJe(wJT*TDvNf=^nQ>5%(22gfZV!p`WW$U8
ztF?hafH{l$Tq$Tewv=NM_Mbb5t<H(`0UhLTDmzU^nxK$SKqGtXCP2-Bzb^_K8m>bS
zP}_+*dA%<z6%ncn3JNl(=uZTNdELG}j+s0&TF)}nYTC4_Ls=IHF&<|A;7nL3QGDJ*
zdR5;K(`ocaT-;j<%>q^CBZ(?eq80P-hau)=uAb*GpNiDwvOBWZIXymvU!32bj@li7
z-6TJ}GIh9U#TDhTx5v&qp7qvRWOUwqE7|CDcckFUp0j_JH2<k$^T^<!1~^jD0&r0I
z&k^CAg@c2Gh?CV8_rsKskRSVNK&HQ#nUcXO$dLd0)n;nv6#a&SyIo74%dLCI$u!3C
z8OFci$TUVQU|&F^M-Y`~mX&vpM}uwC->b0Z{|6p6c+Rz>s&}XZCAt>{6O)~h(Z$0f
zo7YZv_a-JL=EDbMuC!$!C#>i+)X*Y-KCb<1M?~JMNK~_MlrJBLhl7I#Ht<NH!r3DN
z_KVI*lq#7YBgE<^g=rT%T;{-X0fC{FCD2wpuJ$zWqPI6Uxoh5NP(u`0$jgn6q3h~j
zCVpX_P=olVS$o0MJ}AP6XlM40j)<%d5M+hRm%+L)KRb(Pdyqr+(dQnto9_S;C%=5$
zPBr^EG6Gb`&LVLcQ~2WCbphlH*Z#E*w}SxXxoQ~~m-AYSelL+#u%d`)1X%d^O)YhZ
zjqs94+_P`kSd5<0S$_c*PWX>i^SwFpY>}Js+15YnWH>lMbzCA!@QjF%SXJ(gC;RxD
z>AAVkDl+xb8fN&L2r>PR(b3V9U%i06N3!5Ws}?UUEjf|$^X*9x=--COyCB7egGf=G
zeg!y#_=F2NJ`)oT_V!0i0^f|;6pF^Ku40G}m14;eq@}&?C=$AY%u>3;+-kgyV)aQn
ztSo{n9ZXC}uNo1cPYEy0-DpKAkn0>vwB{$e&0nL`-d0cTad@L>Yx@ZWwQuw(W`pSk
z5&+QXfXZejS2H^^0}>y`{EsHv+tKE4EzOKR($f}2_I4h8mZ1b)6=1*Hp}r9gAGB8A
zLvWD4Cv9wO$eXr@khTQ%My)jpQc{AP>Uz1DY=}5Ih!(ah0_pADJDxD9^NWjSx7|@-
z9U<k-w@cm(ekX^8gX`|@7B&Jd(%$eOqv8Zi=Sa_<wDr!L5Nax_e|i$xMa<))Jy~(<
zynz;VB*EVM{>_zk|BUVpFCU+>k`ka+d`~B!f`HAfB^&n0csgB3IFY#0OCTrt35tM8
zXxpRV3g5*28LLaSrfY%=-4wUin|^Ps$_3Qh=6%*mI<MX8E)LqnjK}PpN_pieeK(Nm
zA;?o}Gb=3(s<EA6s;aKmXNirAb801f2Bi$WB$|woFWeSus}yT&NUZdM>D-?xMy-~^
zn4LfyVx0Hjyx{)+B|m@4U3H87*TDhMw$ih++kN~yPErp*RCKw`S5HO7VJL|)r!52g
z@~@+N^78WX{Jg%2i3vM<^_bKpjlPi)fSXkj*1&E#FFS?^4vF0MVL!?xBKHHtr9sTs
zb^nn9uj>G}#)TOQ$`lzGxbPAg-1?EiYQ$!69d5;TJDP*Q%ga$8#9d6ui%_H*D*Z=N
zuj8g?KU#UGf2KS?FK@uo6)d}^W<qlE49K0eRF#3mphvWSmpHU#>2&-|QP}Uy6Cvt7
zkV(h+p+V({@4w5+=O~@2lal3qR8S1Q|5#eeP%T#b%(1ai>r8NpGM0l*Dx_VP-`srv
zmhwq)l3QIfoLH~Bpuqf*4(p7)`pUP?6Em(q^hJMe1Mbc!5>)dws?E8od~yBU3oex{
z!abN`eeKGm&f@^5Rv#^4p?1;GoU3aUhKQKr7ra2+`=6L%yvcT>*|XEr#-B5n1zZJ^
zU*Za()5R-jGr7;FsB|1QU?al=HS7BN8mJ_6babIenBmm&l19Y(M;k}6z}3u&YbN|;
z2EW++B@b4W`j_U~&rKKADpMU3ttsC=o~k@5{2)f!kBXcC<})xN!gOFJXNGUkkD94#
zXt=n#4h;<3P4frI*N&HC7PPvoG=lX4@jM;ty-yD8W^fq@`Tc$u=>q8f#=4H86R2#a
zu&K|~)j9mfJLTd(hd@vTDl)<KUw?P^hM{$nlQ~BTC*p4bxG^6kX*`&Qnrds~{g!mJ
zwcCA)nDwfZGq>Tg6JHG%k!GKuI<ED(h)_}n4_fN#>B-B<sitw!ULNg(q`>951uF{+
z^e+>>=z=`33rQ0-I)qmWn}TT8`x8aMn$tiE>)&f|@e!NF`4Qd%^Y)Fv6nUxR-L6lf
ztGg*9dH4F<ryr)-wGLYMzFaL#;>lGfAS5)<*MDnc^Do?$5qfVkpuql7=}eHo(avsZ
zVdi+LiJ3in#KBgKHvq?s#x(UTFKswq|MuqMf=I-xCNGaF{8c>&7Ai{P>|IIQoXWFa
zwSglKDSvY+m?|$b<fcV%Sa^E*V7$=v`jf(Kl+x}^?aQ)~#j~Rs`F7`uBBrP>F~J}(
zEY-|sU8}#;=uoculBG{W(rxNv<coW|5`+%bsq}4bt05f44B)?jOe`}$|D1|eWqfyS
z26uI1<Bf%JMbgJ(z!Tlyy|Z4x{E+qw1EQ@wM&cL(&hFp)u)Amwkz9DJ!?7L<3-_6e
zx$P)N>v?@1jnD2(;_2#2l{g48%%d56-4zS!hC=m?2gBRX%qIOC^mX(+=U*T~<BMF>
z2W2vGV*dB<<Xnlod|;rXHPqK=54zG;+k3=%H^)n<+1ZY-UTy3(-A~1mDqqW99^9TF
z*$~bX)R>nf1_yWHQ6f^y!9pacaYh2-Y-}g$j$&Rul7_ZHWs}YBheGHKpm{EPLMfH-
zkV1@jS<%1fz#6$cRI6XH!38s00f&H~zl-TeHeN0MKD;y2dwhB8f^Qpi6X+9AzY`Jw
zC#H^^N$%Wmt}^~Uxw=|aSlCckfxRpOK^Vso&4y<;!AblFYGgBNX&7YLXZOiYD%$|*
zK@`-~AOb9>p)ql6i1YI*d?c8|0`Ef)-yHj!xBHzX6VCKRJiL;Mgmfz5sJr%Vb)a~I
zG=X@1CS#uv2PaauAi*pho*QRtZfKgxMSM|3MM2T%4-PiEoy|_1=rDl->o@#E10hK`
zJbX-cCHXYi#8_zEGM06FNVt{1?LJ}2AWAM8uzZ<-`g0p~RdnUD<t)@0(xXtc;gjie
zad46)HH1ROu|Cnq7pTMUo~M#yL>Sb=K(4N{O}xKZO=ToM{UghT!{#@}CDm@j_s%y*
z3swF1SI6kasM)5UQV^NJwUcaGG7XK-VC_5w@ToA7AS^Nr*k`c85}FOUZEYsMD1cU%
zr6B!ZYLH6Q?*6Q6aHfI1+}2t@ZLVJMTd5#qztmW1W)<mYXCyaN{)~%DnmQlN=emfY
zj6^)Z54cu95PkT4nV@aBR5`+KwcS=$_TA|2JT<^S%4q2ncc(!n+o;ulI5s-9u%O_l
z5%HT?5)=qwWQ1u@v05G-Z&Tozq}*QMkdg6p*Jx>FVy<S@4j!6G?LD9YGYGO%l|Q9d
zTm6OHu(Om+asKqUU*5!1%LjPaf@I&n_)GIUo3AWD@w7efzO$z+0ty8Vu~PBW%JOnl
zRFvtX4-fDgGy<9DN?Q8Kdi;kDTP={Fe}Ul~G12;V$@i};ajc#p@bt|MoBBP=8LKWU
zo1B?>eEtjq0XeiOYbkPK=v&T1Gf3Bhx~HXeS^m2koE!j+$lmV5rit87mjZRq?7J(%
z4lBGCcg=~vl|`o{+Tp9<KTaEEWquuvcs#t6tSWV{5g=t1m5nV%=UaGXw2_ew?jmt9
z9|aSW729Qj4B6;c^T^)_7R+cstQ|o}=u?W2MPJO#?m0+8aeLa6gKVyx93lxy#KtHE
zmW1iaw@&oVUFkiTPw_BH<+m+Op1cx9M#N$%P!6UxQ>(Aten(kEN}@u+*AFiuFv+dC
z(x_AK?pNXvf+0FJxQuq_jDxrE`*(2Yy#d#wmW#QwH$x1ij12oZIvU=>$<c+*!P)AI
zO3W9U%zJfUVs`J+67~AxZyO&U-!Q)uVo*&@&G@f4^SAJEn-VWkxVX)0Ud#(Z5^^r_
z9l@&~m-x#c9wZHLC)c_Pqv0YjxNpr~gi#a^l^Y`b@W}lnPvvgz^`}X5l(`fZLw)V;
zw@2IBP{rOpuczSe6D0X;nFA)dEp=@wRSvXr`?rvUgf1QyME8ONElG#wg|emtFpbL+
zBnnefQZNRMSR<s1HG^b+ynRRXQSRTBq|jF(POVo_F0G{fxo!u#dQC82U?7M~Xnuu|
zRf8BZcWA>mcT%N9mmGi<J6eR}AqpB1HEAofN@jYO&VxaaSXy22cB$IT$Cnu1F{DP6
z0t?DP(k5zEsy{QDc)Y`UphU*(SL;tDiv<*i`fq>$h9!=M6$h8|^uoe-Fr#e+=BFPh
zp><c$MO_nzjKocLmuDl!Y_^UKAaU~IMdAX62S94ej>~Q0GjI^d<4S3*%6^MU$W^yq
z{;2b1LM@h@Fgu0p?@hOld`scupPwC1Tkh^E+ZSOc8l2!S6_7PoN0)oMc1AKhU4JK_
zR$0JEe0Y^b_|3@M4?AQ)qco_q$LKc1Ne<|1K;KbH_$~+_WpA2RU)04Pc?^UgBZS`b
zp>z<C!9lTFsIvkD$7IZ<F*K{-BwT`p@B2IMzTO36gNY}bgx%5g#zwxIWYkNC3K)t#
ztBY@<e`tR*(6WGhPfpRVmgalZwC{YjW!2?&nElI7byzp4es^cDuI?o`ctYaH&EAtI
zcsX`ZgS#RINm)qQra`Z2To<rpf2SURw3COOw0ZGZM`A?<%kI>hR}<J!Q1mXzrE7wq
z@!3_#b81PN_t$p9)J)B}zL@gB!-*voKAGibJ8)<Nw=G5m(}4yCS|KGe7vMF3t?=>l
z|GB>xV`mNR?w62+;jkZQ{0O$mM8h?zufg$r?IyR2$N8^bfH>hLK`PzpV-}=F7J8{r
zQ}Z5U=iBB+>R;i!o9*G${#ep(7r!6xp8?4wT)g-Ca84~-ByF1qtw<S(B*MMC+~UN8
zoe+E_!t6vOm5+uW(ZMzZ`1JXZEr&J9hu#`T_svatSTv{Jo=Pg0XoaI%H~OF1sL=TI
z+wtY;@C^qPlrO)Njd4WZhfqv1Fhx^krleS-AgE*GY{6wAva-IoUnDr(8_HuKJR_Bl
z{qgPmjD-wwURn<E+c${fZ2fQbX!nu5JrqQT7k0AX#we`Ukyl}?W!K!k3MQz%A^SJT
z(lcim=H|l{6jIU?w-$9?Gjn(|?GZ~TMN99U;BaEx--p_kn<;UcUwS}6cg6D4$sO+q
z?(Xkbvs<vFJS5dQJ`Hwm3d(1rqNBgq5J2sQf*AWaI$tCt#?A3ctObB@^LhCS-3P)+
zDH00{0sL++g1#PS47#1yUfgkWKR1vEaPI1*a#+@vvcv=jyMQLr6A1gGoATe?s7~O#
zohrNDw&J=uMfGnj-SX?FkvHQuSk-hVW6XZt)2E~(+d8IZW)>C{Go4+wePg~Jh$5O{
zz~k~Q>rvaI=7?to<aNiaw|HE?MLYZ{n3;By{S~gG6k^)S+2(BV^E0RE=mmbAYp+1;
zrhpcM$PHWQ;EOK=w>{s~vY(+u@i_}{Y1?vV2s7UQr68?b=|(J^;~n?8otc=}H!QIM
z+XDo9E_X(l-#Y5N8dnP((6vgEQ6m_^k1S^Uvwd`lZ(`R#97E{I-2p@q-!HB9X6`fn
zh1=W5KR3(u(p8+<S(wtyUk^M-G1Hs7Ps{KmCE?4cq%JQ9S=^FB!At4C*K>n8@)v=j
z(h(S4u`3kRa3doT7^Ffjzhvgr0P*8nX-0=$aTkqc2{#?egLVVlR^RDaF;xvp7M3F;
ztFM4ES<}!Mue8DlWKI|osTKK##&e$i?(Yuiz#*~ZA{5f&#KEP*Bt!%%&;Xl43=$r*
z?Php(rpE0a(1lZ2R_~nXR5Jx!e)Wt70&V<p*^a5FHzWuXqQF*ERFsvG@w1z`02z3i
z=xixz>EN-5wAC6<esvqwZ(jbXGp*R+XoOsr1D`s)sDNCR!DTfBg5T^8DDZ5zTM4}@
zCFn4*Z<3NDnmjvq93b_}otY0Xa580O&84Nuy<rlmn&`sjpN)DB>fT)@E<zCQ3|De>
zl3PASkVnlO2}X^<O>=X28Pf$rnFLi>b__HY8APG}spCtVRb*2W%d=!zjRU6_RIgq&
zNxP#_Y%qHH1h>Ng>&l<fsLC|i&vxtWZoMGZ^mJ!4e^LOv1O(aIp=oS2EpsQq348dZ
zooxZ{#k3_9hIXW(ak>6YTo3p8KJhje%*|1Fh@Z1NQjRwk5v*p+E9B8$7vu$0%b~-o
zTP**}rmkOsyxJe6azSH7MxlBR4#)0r?LyYRv(Wy!OK$5!;<{Jb*B;;dQ##Ksn8n^-
z7@U1SIEoZ-JdW-zG4Q?=Fy_nIW@M~=jAL)8K~#o!ckeVGx3ug*vx!W{)37Q2fmt5d
zq3H&X{Q-s+GW2Hr-jU1`HrcaL#!>Jy8CRZo?>hnUHv(c3|BPILVflp9-}MN$P>szL
zU5>+&t&sjDnaD|D9=4GtlPujX4==rrZVs9F{d9yqoVOtV3uzqlmZ{EPvp1r?H^AHi
zmzf&F8@+FH^vf6=2!ApBSz<?3qZv(1;m#0YEw8&@{(T7`)f+MqWi_>&r}g0W_JH}Q
z>N~!MuU~<$<9;>}ADgcGXP;Z8epN#~NlF&v_JMvSKu<4kU?9lBp{}H42a=$maGEA2
zI9glF{kpUh(nr!<>sR(DPEg3m?jX7zzHqm(0oQv69wX2Y^X-xufyptB|BS4tZ<ouh
z$Y<<W&LLCCor#k(zRMgQ5i$MqXXG#`bB-?!2vh<7=NpgUmIx3W*&~mb85tLnuCQ;&
z5rf=?o8m|W>+BjTfi)t(>Lw}AUzKOfDIg#sA|fsxc)HpK8*F~HHvw!qpuWc&$AZWR
zhgi<b!u#L9gIi&tPNy9mJw5QUmbSJwvyK3e=ma&rp72U30uzdy&KdSD{sL(;*`tPh
zsqEF@Z+8-nK6zEE58U1<>(1;I^C|obSNzKJ)4^E4`Oq)3(f#jX!(5&7BAS{!f#+AI
zQ%*!o3{ngH_VZ^m5rojX0LX1-CnQkO(>K{n6*;10w0WKX+8Rt0hy^GAUn0B3tJMR?
zq4!J%uUL`B7B8AvzvBHB<}pM=Yn9$CSFiswg(32L#|jY~k{MzUAKhnv*SEZ|@93hw
zP=BCGgIBh8FcF2S7Unr777UDZ<sxMg;6y+mZ?3j>^7X`Gq0%3rn^Uy1V;Y`qUgaxT
zNJ3*_ktmd%Y*0*0EO)g&co06v8-ldUBq;WTWYF2zmYSTz$^20-Tqz$YzU*mZPga|C
zoXD>-zomaq2BB;}WG!J#Oe8p)^Yb??UG9ihmcgS~O?ahh$zw>G8GTV#{h29dGfAVP
z$4{XJLDMr12BiDGp^}O=?=0F5hpfzQI_wwf4g<<1rAZs@;Vmq@?!S#r^9TqCP*8la
zqd5Wz%0810GCe({pB^1zZ9WMvF<{gw;&H@Oh$kGY%=&jPOSQ{rW(A%$9yOWm<{MD|
zA<wj&wAws@wZO|A-S}_l-P!?8b&>08;6CufJ3qlfY$BqUKEGM=b6q|)TEF(5Nqd$*
z*GF%1;|vqjaIWa@mkRFqNt&7loGgD|YaRr*gKX{WKmZRU1w4-zd6g7ByaO}b(gZ=S
z`XleOIFFCMtZ)CtuOZ-%K+unaF9pT%Q&S3lVhTIzSU6Z->c)mg>pX#fAH+RPx3{+?
zxr?&1IKruMz{N|?nD`fJESUe71y~);4UIBfge@ZiiD+>1&E0!ky{0a_=MG^8x3I8K
zjul<LKL!f=oEN8~N=-|4$6nt1ngEZK^bw>Hx<ZjWEBnt0L0QjoxfNqEg=HP-jw1sM
z2$1X9grx~}Bc$?&t^tCR{v+K4vsN!oNCd=fTwhNXzgN<uI{mJ;*Ie=jB(Bx?>=EE$
zNXTp5fK6<nuxv43hmM2u%8cPFUw2~`1PM1<=0ttMp;ZH%vBr%*CI>oQMHHR~neT!A
zetM5SlJN;(GyGs3p+V3<MMVXXY48j)KMmG*V9>9pmW%>DIhE*F)Jm^DWoBk>*v$oJ
zAX!@fv>tD=Uzl{$<IWiy8yT6IolSX92qIxeeI+hmD&c%RmpnKgA0In>f<fREpT4-u
zp1h=z)a~yyKy0*2f+VUkGBTzPmB)LW;TRaQ4#&cdvfapUB%YU)YfVf#cDWo4pR^+?
zQeZ{Dr@&$2<-=$?E|)4~idT$cD3hUMPE?4pOrUzELls{*u2%H{q$Gg=2y*eNYl^zM
zDQ<gW`uNBs81^grm+OC1?`=SQFCxRRYZ~YENvoJo54iuNqy!J!1)$<%zT*#l{BQnB
z-<#_M1#o`;+;CaeMQMKBivlqK7p7Jp2L+-3d@XB#^YXV4x3%QBBM>r=KJwyb4KDY*
zc2QN0&CAOJN$<G+VK51Eb9%bEh(VxQWqBh2Gkm$@mdCQGu8u3Zmy4VGk8UPb#>nW{
z*uPuQx3_^}Z^Q@)yW3j2n4+TbSPvG_x@*8aXF)+h5TJoNT|~ewVn)Wu;cZO@oM4Gg
za}JO=Zo%6+3<nH)tcBni$!PIsE4I1Kt%+*E{dTyXq$qe`V7_~YO9K3JP)M1Xl~hzf
z6!^1T{=XZrH`mc}e0nD7l5!et@%hZZa=;CP<$;a&d|yEUZOgb&#a>)o^ph7nCFj+C
z{|e?AJAIYG?b(L3tZZ+C-5k-lP13YpQiw5DG|U$rhCFJll!p|0Y~q386&41eYnt2$
z!J3XCK&!`c#MKHl_un;(qx+OH!3E7#`tF(*m$di>+#5WPG{!h;bKdwdU8?1|S5n3a
zA{mJxRkDgH>Nf=<k9rFoPfzp0hd~UDd@<tny=*M#R45;UK73pWj{3mLgg9IL+8pUM
zy+^L8hDP<r`I67bA66p6Lf-QX*Y(_-_yw}04j01H$n4(Eos5FpEa6r+X>O~x-c1fX
zmC@lJA@l9J`NaMO`x`KSE#54nf{x*xzqckpqlE?r25LwS(B>G`nVk2`zdnVv7Sa8=
zIW{6YM;ASM?lpXKTs{Yv<@}Uw+efxLDpC6W68ZmQ>Mf(HYQuKXMTdl9prrbuBHbx1
z-HoJlcej!vDIHSME!`l}-JMD!AuWAwzHjWax5J+fS<Lm!JFa>V9Pa0uw;mWL<(*?d
z8%4v(rl+QI$<%K8pNNaaKX&+!z33$^ZU5q8Dup;(5@><P8r@GLJ3q$;FCqSd1P-`5
z05DgA%U{!SuGe_cZo8#5;a>CF7|q4|*d`aK6wQog3(0h?oR#0nmJW^S*3T-d-rs+h
zsQzzwv*1f61vAPP(cR4dn2?KReNdY`*cxq%yw_zk{q<~|xyut-_vUId<S3z%#DaBt
zwdbLw<qjydA87_-VPl^ktV+TwaoB-{^!NBUIXNYWZXWTzda}Ho%zy0&+}=0vh~PJ4
zWn-f=>H}UrB_%~ws(=5t>%b&G4cT|+I5Vrl3aPIg`*wd$7#LdwQtuA3zBpp`1SAW3
zJ-6eQfti{5!NJjumo=J??yGQU<-UXKU|y;oI^=OZSoum=@-(jX2jZU2@VAHR6>SNl
z?p)3`bW5M2B%v(4f1d_xxvy_)j%?~gu@<q(=v~X1U%y>b7a9-fZesq6UXt5awB39#
zPAV%aJ3F{1!(M1kOqiua^-3!KJ8YQDgW3yjhbLpB>!RemR%5xt6tf+nxW_wN@#+VP
zzn7Ph42+HWysy6wt`pAW0I$#0%)G#IEMzvLF`z@jX}1EqsZ)uj{NAzRamGag`GP>Y
z+!zA^Ctk>74v)5ru63;LdmY(z`me~EPyZT^p?`owK*sCg0)h<?8WQq)Kl^A6Fq1G3
zQRqo?0GrD-4#G;)e7=rV;4b7U5aNO}YGiGPTP|7`CA_oVeqBaS?>9a^AjIaV3w{bg
zHGA_<SjG)pGDJcHl9GO1aU*JMc_u$-N=p7&O-Hf+$=7#ta=bI#N>vH^I#*zV!P+jJ
zz|@{2RNs9ni08EVPd(JvYW|s+s2=i)ou79{uyNObpwlDmH@)tEN;Il>L6vEj(~0r;
z$!|#xF#`4*9xKp~UH$s+O=gd<&o-d?;ObsgB;McOhn-nmQj)q~8RG-wL~uLr{&i^}
zXGfEgk`f6cIg2tw`4Aj~w$K15*NpRr6f%(zpywi~y6Fiv`}=72##m1OWj>QH_3{n)
zIYwig`mD9;<)O?W8h*L|yBRJ{GhYfv)Xn56Gnai5h1Z!1|57;eTRru(*z0r~pJAK|
z79)I-JW(uWeb!}dyD-AZ+B!O2MMW%M2AxFZz^B0>VlkQet_wb$P|@&rQX}(C?YTzf
z2W22hxTjh|zi#66uIvbwS66mc>jjc@=ch3~ovqbf4W$yUn`3#SfS`R=B}UND(6s3>
z2H*yML|SM-|L{k(*P{jGwT%JM((dPf*FGD*aF(^6uR8`4N$|TLR1em<vR%m05$wgD
zX9tF6E^T@~!wz#U<H@hc!RcZ1#J5X~0~duJibi??MjT^HON;O^`J+b@RukK+7NLx%
zWON!;g`gGc3@2YSHDo-~?&^N#OD$$8w<6eYb7`Av#4$2DIzK;8K3DgITb}@dCeGD(
zWd~Mvk5C~YAsvg^3Xo8viJ$P<Z~klT1W^QaN640i`h}^-@3v44Y+kuF-`HpGd1_iO
z|H@QG(w3ONww#sMHPdE%bxIHQtRBQfaNYrc2l%+WtgOmnnt+UK0gA-Ca-)Z<k<rcp
z%RqXD^{Rf;a&PdkwL7oSZghKC7tH@OZFEzUZW7dskn*6a>I?vK?dkN<(UEi#o3N;;
z^Ttq1e}7T#rXL8vk2IIU@RI+&@K7Awt^Fg>8i<dd!b3hityZ0VuM2(<7^~de-BIG%
zEY*u6DaFjpw!r0cpMYR{d)rR36q=Feb`rx8F#>UMGMDr#)>-5Vtls*k<ot(^TrU4k
zoIAQED=ERFij0Z^P5xm3A#5FC=e@+V>c6{}%?ni5XbH8$GPASIOifplnW40+)mzrz
z^r~)Bt9;A|i#G7I+;L#7`ZnZFByRUskl4~v64z1jWovXhSCi)~btlr$$lIlU+{CS@
zm(p$kFaR49EwKu-zCG<r7$z%Y(*x+{e@HrVe4m6dbMP?k@7#x&-tSibAsX;CNC+D(
zrkRLB*Or$L_ujUz;p%Fz<xMmwU0(c4Nn~0(@b`GQX$9#lzuT}NX8`O`mmV^sfWwD7
zR{54PI$4!m{kj606P6zFl5aJB-;6l0ucgmCt^qm(wWLOcN$KWU@7JKlPJH&^E}!5c
zfEy0;wJ=($8ONT8;_IJFjWPUO4kX)b@b0a>ii5trYc@gfC>HU(i|56X-C44stsFB6
zRc_iaUo^~WeXC5Bw11i;BuuRgTrM=EnmDx|LfLQhzVTXQUKc00In*4~D}|zd=&W<0
zu34;B{Mc!UpV7znu2FdkpAoOOAms|;L8bJI<n0GWYY&|jTobq)x1hP(FWS}4fnzx?
zAA_8noJI$W))MEjgTtQyLA1^?!`>6`OPs8qjh^)&VtpiaHaa}qZTvPCKTnCvuqXP%
zDfxFBMhhauUz^l8cWbR$yBG7L%nbk8>(y65AEHWDA~=-=Z4-D7*DmB_W%0$uIjtk3
z4(E|T2^pqr{^R|(z(#OsK^)BCPO%(iwoI~HKwLoKcE1q%`Xgni@0cn@Xi-7hc(6RJ
z(s5)eKs`__`IHqQwti);xr0aA`eMW;k42Lvl$ed6Rb;J#xv&L=KT^tvL7$OUP3q30
zUZ3{DUcH3U_-vY&f0_$E7HNsNy16Y#4L&y3=do%<N8CQzbquukK%U>bLwK%Tzm=O)
zZq`o(%3}P1foBpg-ae;{NgIxcj(30I=;*kA27Iw;gGcc!PtBG90pr+|n1oV0*GfCo
zkCvA7H*}8?H1EpZF8#87&1v`^w>7ldn$WQ4o${yl->((3HKV8_Vx4TZqM^O#^MWQ{
z`KnngYeX6aW3@s?V#HkO6LJ!>vCQR9f8)rg&^TI&TvLlT#v+2rb^rDCXRqh!f5iue
znBrB-PXas6m9hgq#?N}`sGZ~zQ099(VpZy9?)dKI1htDK*B8-!`@TMsC0XiVx~q{O
z)nUZo@_g7Q%^fGm01gR!IcHXdes6!eoqurg(0O>$F}aj~{3Mg%W}d|7cE(Jl|6wMJ
z&|tHEOh#Jt*^v4nIO6b4=5V;q5m*R{ghOP+(05%S9UYzVKd_jE?9X6vc{(ogf>780
zMCFa;Y*q5gp0~{^Q&%Rn`!w;BCqeCc&d$!w*27_LT&{vaH_SpydwY9z)_j;Ve%Yae
zV+<Y9Po^#=e;yKo1$6h<uVFZPkt8#{z2;g=J62m?^AvL)u^3ZFzfE|f`5Y@wtfcYA
z#(Iuo;}6Od-c#P!*38meMJYrDyupV3D&_Ku<x2VqOiw}x)Qo4QrXGA9q5hPZI5Ti_
z{(T^kmD*q!8DW!p2gSEZs-w%>F2((Li}m*mYjb5mL`?0tt6EdyvHe;fmU_co5!M}w
zZ>BfWQFP8n$H(^N(R?2N7$8mON8k27+XQ$E>8X-)nD!43>nF-^>rVX?I89r=BB7wK
z+`)r&q`o68VU-tuMjdP@=dKT!^dYUORP^CPWxbAY@*9($XpLtrV&eOUHaGit4_2`&
zGY>k`)fG4$HsqXsv<FEOYn>B38Le3TCLAnLb1|AD>pWo78{1uVvN1B(_18*j4lSAJ
z=8=dzWxTa^E>DBxCqKl3*;ZkdnTz9}s`y|+Y{ws$<6ZTc2QWbfX7vtblWI4=VYL@&
zA({X+%|QC8eWWxVIW`G%&%{Jkh3L`AiFWsRK+tP8mv5e;e%=cs^{TwacnBb|90PGX
zJ808JOU*m`^P=oqSpijjKtcje+8UGLZ{MW24<f&OX)$0@_QnIXiK7MGOl9kJ<IG2P
zw9v#KRh%hr>KaWm@*k^#G{rZlSgoE6^<#pS4H;hEgD1!wxW00qR)ZC-7BKPf$aX95
z_iOy#TqaI2zatR=t@o7Ai(u4<pDwWTaxad4wU_SY$2}(3nzskyG;j*|Br|m=B5{Ou
zfrODa*CHjxn@ucINo~*1>0o*tL8^|`wR&)L<N)(DDD#z<zQE!cDguZD@{jdH=tp<@
z(^M!Qc<?qyt`plmTC1qAJv@J@RmAklCnmv1l1f`!`%j2jhPWa^t%+l|LMPqml|mjk
zy8pZqs>lAhyboR$Dn%E0Nh+$`Ct-YO{(|=&J<86>Q3by+2uicEvk$gdH6!V)77F%!
zhBG7=WWl0brCrAa!8OXI6$3_<)lJuhGjGE`8wAdM%X*2@mCgKUghQ0;MBy&#)|j-E
z)a4{yVd4;c|AVh&L47$bm73qv`T*NsG>jxDIGC8>JwJ#UZvVn)>5d-o>B5T0?jrWf
z$ApA;rq*H5iZwV*nZ5>QSYGU=@&OFg-q%<7zPzpN?O0Kg{@eLnTQE3E*gz^AjoT?%
zaO{6vRF=O4mB+(>508xWc#__`_wg4Uhb;*kF6yZ9v%f}*?DW+aDSbS_K=X{14`3B<
zk3AtA6p|%x4M6!&$2B&drC<s(ZYvS%IpjMnPcvilTy=Gcv6$f?aM1g<C|<l64h`iK
z&m?~vX%n<k{+5XDixKsmFN7jEBA#po*-CC}W3O2(K6P2Q>Nj5<PwG&Jr@-hx5lm=W
zq(1IscdompRNMNVLFrri&>`(wj=xM!ffHb<x9?=?AUT_b8cYFHR=EhDa^ay7^baD^
z&vzd_k$Xn~YXl&e5PR4e@<cJ<tU(i0ci*iNiNA_IeQ6}6b@;7t9CGJs3vdx=ljW3D
zfjWi(LKFm7HS$5a<<kmzO7Es<w+G)+J_sXX&36@17j&xnW_stW*O*yjnm0u9lljQE
zf0(}C)JpY)Tf?o<C#$&*PK2D;FI)$NOsk8Zzr;p#8)cMhRg~uSw{^$!5c)XOGv|Pa
zLcyfL09?Qp78amT{*2vA|9NO=C^wgqo4a1nJEa#VxE;WR&(HIfLWtRSEuz+P6*0WY
zakR?w0dWGiDRe~YV$F6XcFYlX?Rq<imKwRf_Pn_UUGI;sE|VW0kc9n@yH^jO)db);
zBgF-4asQMhQPnC1q<+t0fyl8B>HYrC?X(|0qH(>euwwT1^=-pO$QxdA>ojX}!5WsI
zYS2<fYt>zz0VT^C+g<Q2@VM)qOMO?Hs$3pNN(GgrTjQP9vCU8CY#S}}a}QJXTnO(?
zJI%X=JVtww7ZoF(I7zF9;fL7D%^P3OWnHn^Djg@vx8Jg#nz_*o^@uZp9~;c;EgYVL
zi9Kha&r{BzSVJHfxdQV-WMm}x!w=A1@;?O!*Lz-6fgcS&?&D(|bb-4}@5Eb?(Q%0!
zdYjmfaKC%+5vO|kCz?eQ0>Si(m?c13=+blPV5}Q$eM>@YAWrui^<$jvMPl#c&tYLQ
z;!YHpJ~BD$Np@jib96g4gE{fy_|IvSqenj4xZdaLZ)eJ=7?qg{>0f!=W1raQN-d)!
zw)Xf4elt}(iVODs+!UKGl&4e&tDI_y_OZpR(Km7FwH$^BiVcT6OKS0pfu&p&7vQ--
zuMQRBrEH4kUlly%HV&N+ZpXHpKd+Ltk8l3G*2W=bvzSVW$4LsP@nCeAotRV5<&$OP
zCBgRpmOVQ3wat(I*CAiCYLU>{`T4*jgLHdU#NU|{lVQP+{O`@o@)IAtjs~RobNg?R
zP@))#L=bU<{SLO}a8hnIc5WDkgQLiK+NbtEh1+wSIHwDMbyccH&kBiDOSATY*P27q
z`X20$7K=?h7BzRc96SD~-d|{(sW4SsKrbyVh0}FA7LJe4%&61FfhAkJ&LK~wNFCy)
z2Do)1B_lP-UdA@u4_XYO*lYVunOC@^)e+r2^y_fZ-(Nr$|5WgF_S9x;G{-OQc?o;s
zRRAlR@F0|Pqz6P8$zQ(g!{*|4yp`ug0}E6*gBk;??~Q+4Al{nvG6;JuxW{pQlkYxZ
zjOjmjp8w8pK$mwJ0h(}h`}J0REW{hvhL6#!Mg+>JRKevN8z!Qno!=shG8IhK)Z$A^
zZLJ0@A&~)~iPivIdr4GCYWa4$Y5$m;3~@>S)oilPZzgMnm!z3`Eb=q9nfPirKR6Mu
zqcHPZ47}F%&24L+DCEuk%(ox~q3rZ%4s$A>5l@|{hZNG&a;v91Gm3eKdwj~EMpLpy
z2$%lpR<xMIMs#$n4<u_OzGA6#sp4~qFVD!^1WXH_mU2}&<wFN9QoVah@6c2E#`6?K
z!QvVt0a*8&dcOg!bfw=K1{bDQCJV?qKTqpI6cYH?_2H|-P;^<cIXA+_BYUH!<YOfe
z5*iBriG;mYz@a<a+j%sb)nT!d*Nsxf!opIG{xHgbkfQ>V^g(Dogb)F;0i#Ly)c20Q
zK8x-$Lxi88uGNnWu$ydEtHeCIlV(2YjoYbYKp9;k{4MTzH}#i)Sz^(Yy(Pzv8%H7*
zmFV!-tPtIzDwIevO2*#S)^)uUhlOZMUGrduX=+%W=26^=z|fB$h*Cy|0^g~4kNvw8
z;xJhC%4Q=3aQ^E8lx{`YkzTuDqtW3>;>_>)S!(LIYLnsLU6YuF^9V$udK{frBBQ>D
z)iP9?_ag=@7Vs=iX|-$4bM9<iyR$ktL_}lyBRg>Y<eS_)^PM|LeXFJAIq>zj*?M+%
zHo(UV)ny`M)s|Nmp>WtaI9&cY(Zl~S*}JzpmmX<VX*T||8Z&|d6CSE`U`*Fgl+SZQ
z37g9S5^ksMo?=o59}^~%u7XM3H|dmg)CJPbNcEACCt3|JBP3!;x>8Zq1UKZrY%dYB
zXj37wQa)w|`p(UjjWu?yat8TtUN-noc!dalu`gdcteuN|6nx`~JW;q?ZNZ2+GlhOU
zT3Oq1h4G>L>vaAq35pi|X{{X9MM47l5h-R8x*S!GYacx&M-wb<40K3$2vKU>&d28K
zoxwAA9hGtWNETANJ#vB57K%6w7+jxuz~ne+R6cpBh`G}p_3A2?Hn2*@I&22^T<qi@
z2!7;Wdfocz8wp~U3Y6zmU)GXZ8FU$wzK;`?-~LVfKAr9F?*zep2F3!4cM=kyYp}X;
z@+%fJhs6SW*2MJuOs&iSYjOSxoo`-V*16l;{^??~_ns(0JR|XsX_O01h+{FQEHfS?
zE7ZIrE&b{3LVYzkryZ)G?f|_BS8Bmo>K;K{uP_%mB3eT1#VW_n;Clv-dBPBN4t$pL
zqpBCluW~tV8SzEa$BEUb;seg9@C{6Xh*odMjUmBidwFN4=%AUMjqQYp#b)4=GwoiA
zFGh@qM5K=(+6xXZ4d~qDq*KoN|CTtbh4f1>a+#lmADM?Gu@?h$D3~rs74t~Pww}_&
zBSn2kx}&-TA%-V9Qkqowu9y5=h_L6^lhK>%Dbh;m?UB)I=OzAb=QnKTYg6raboE9;
z2a~>gdq2U`Vw<V5OtQNox`oYNG8^mn;vwo!cWU>WZSyx1VyIQul-ja~`7hn2LEu*b
z2|0f}T58UZuW1vCvQ-XV+%RPn6*c}Wo=uGDP0QQN?5L=&rf<4}ne4mmov|tQ6u!g#
z{ciI#idWrn(?0_f-i7;rdnG8Jc*$q`7#+v$4qgst;Jje6Wv|g6Mz3u~z`q(kobBeC
zf3*H`iuJzNi6LH9P-upfTV*JlnV!{rqE9uRmoPyp_7r=)xIiKc7Q8rZ*S~kI%8@>_
zz~f*cvYh;9TORB3qboO#(*2+JACnL=9xj95ZA-?oU+(TPJNDqq<|s|g&gNWhSj(GU
zx({Q*dnO)9A-?vOlKGwMWVV*(?Vo&;_y^-;@X}O4V_{(!a(@t)*01M!yp8?d<44g<
zWn6AmtbpZRY_!9D&xxoDKS3lS>4^&e8bx;H8uSkkrwEqL1D6K*43Vq9s?x$R5<?^q
z@V-WRdJ6fSqENGQ>9oLz3sY`>+Zzb;z^yc2Z(wAUmXT3uBw;oRsh)Lk*bBnnW9Pu_
zv1|qhaKyXA^bGr>xr7OF-Z$I_>?|zXyaA5(^z!LrsRqET!7u{;2Rd;Bt50$`tZaz*
z_)HHEjHM_wf<#)J{vNLgT7osKHlw?56~oju9AzZ&a?o=Aa)$gi9my2e{IOb7Qv-5~
zNirsyoTl6$(}Q}|tR1R2&|#cN&$=G3*bbS51?|_45(6wOx}~PqZCCj2V_5S3#;=;q
zQA7w3<S-8)fBezbMxi5y8`sMN*vjeN;_=ZDmHqRggC~r7ZEINfbth5o>BP;PQegg^
zo<2L7Q8t;zb`}EiMtK}1^PhAYYNxd%V968rXe#&iD{m4g*@joJQGI@&QenIGv7Z!~
zv+&|m4GyJvK487RkHH1dQ~mv%AKyiORP_sphycDvD_*ThSFKcS7fAuGl}lhQZ>0sm
zv?}Gug@FaYIXJk{evfX-46|uoSAAf)2BiAN&59txDt-Lssq@bSitf9msQE6}u|+e(
zU8^T2dZjhXBf^GB22?0ecVC3Ovp1ifyEuA1HS>F?e&ZgUG@mHN3Rc(B(5F}u8E#w|
zZbunv$FL6B5~lBJG|Ej!@T!EJ>wHk=`TB<Uy_W5-gm$5vQG8Oaqoo#<ND5#dq6<X9
ztifo6%)+W`7O)zubDrE2ia$OwLS?r4NhBl2nrN&fC@*hw(%u+Ung(pS%~YuF5SUX{
zRh7!?X|~^7>M!D<C>M=HCAsSS^7ZRYAoxq@Q&MNX<;hVU#!$kq28}PF<c;V4-RECs
z8lSGssy-Eu?2mQYpt2lBzwvyuvhTuHzOFZ{fN;5X0J!op+9hGS!S(Qw7y3*^dsc8@
zAQzC2C4E|7C|qx@JwNgf0XZ~>=$@mAfsz43jfT4Vqt!1FfbgFmnFhW6y}b>avD#%R
zWDMK1FWie&0bpa=Rv(~Sl{H;e%6kNxl_64#8&jC9Y-fY1@d9`IIk$^7k5jm2XgMst
zefs(<^=-NwM9hI9v<d)jf1|id3P=oMPHk)xBW&>CrMuJ@SUgS5RC-6{O{5fy9h3FV
zhk&G+4^jZ3vBFD5%wjSWO3V)CP_v0bPB<mxv!sy}%U2pF$r^rQsQfC;>6tnI<!OsK
z<iJ8}w-H)NoR*HXH-DWW>Ik7o&j$=wy1rD$fp74O$<lQ*-nH{^BL&!|{rpgYx2?5b
zKhHB4R3&VwZcgFJV9J%l3L~V~c1UVQG?MJ^_Sc<Q-uavZwv><%96-?s)+!IqE}_Z6
zWYP$^zMwmY4B<=Hl6-h$Y&pJqOnlQT<)EU<zcn|@o~uT>JR?}_>Z?onawhz||Ehs5
zcsjITIHq9Xosdwlj8IVT=;7HVq`bo?$)|d#LJ3>q8`lFc5P^6W;HNXGPH(7N4^Uok
zP`0GQbMd|lunYSeRm+>VqyGO#3$VFx=80{sWoEQr99`kSd(mO&X2Al@zG<q-df#zU
z8xSxEo6b`*1OKyaHu7?3cr%wNk%PnE&E+w6`p?xOlw$n+KW<KTL`e^w80Et&D@t8B
z5vXXpjK#mUw?n*9JLhWc@tytz&Adb`G4idN@h`|Y!J=`pfU>d0_1PZ-pN!_L>=QUA
zdLp=oMxDC`-T>^AJ~8=1K>2OI4i=`reG-d$8O1BQS2Q$zRtv~#8q1RZib9F8W8M2Z
zUzM)diL9A=iS+T67O<pY1<p$<whBbehD7nnW>WYp14deYIgqAOWR=$qKdEn>`Gi;q
zD@-p972^emk{lyE6MV8a@neOIfh{J67H_?er0C)Zt&jyG3*%(pIs0POhGdQ!?r@3E
zIz9J3ssMe4hSG0-(|6TY$&<dK5xu8UiA&qtG<Pw<l;IefYSpPi3_|tT$ymACk&gwu
z&XdI>Czs7U3sW@@9nO~t2W>aW`J~es*s&B}MoA}g9UItS&Nv_t{y$}w`NDq+rrYf7
zJS|+rS#tZa`r!lORSrSzJ+QvUOcobx3t27HtnKZ=j{kQgOh?$nB)3_f7Piaxf$LjA
z;^O(rPLS?isD?{QYGAlCS9=KLkN++iZA>zu!B^Qq%o}4~U+KGAwM1d+ni$f>0{5n0
zzyzY^{f903vn5?dM;NKb*pv||=jS5rr=PgP$p;{KOQFx1;W62K{X=EChTl_=l~PrI
z`YdiQg`w&FLmh1F&pm<gvb}=hGgGP`*fTsgJS=~#3(E#dIA19%lhnP>Wzzk3l!^c6
zSUq&Jcz_Vb{V!-a&7&p#VLz--4FrncU5Ul)&I#}_FaqJd(6Qg}&8~WejO2c!1A7Fu
zy$roc#aDx7B35<ma_vH$J6c+uQUZHtoK<2_Z1~>I;%Iy8X1pR|6=;jIRlF`wdHzK#
zqCPh>@132bSDsR0(y(Ga>7C~fu)h$S_O7n+a`YH5<*Hy$ZTlSFot9Wd6IgrQv{WmX
znK*86_pK`K@iQ)U&x)G%K<@y5|03_3dm2?J@y&Mc(&?69s(YbkR=dv=H}Y?)>Hd%G
zd2aV6FRg$cC&RVra^qROWu?V$36eYzmaipv-wK}1XJ2?W&#?XV@_Ik%Uy(R)vc9tZ
z%fimCD*Wp<I`dB)z3)JBI=hKdmmaD((_)jtom%lO1B&ElJ5$Otb)M5WY+eBnDx;=`
zKnVO8V}bZ4US6<oqkO13Ij(SsX39}bm%E+f;ip3b=yjWR-<|m=5^8(31r{0QabBp?
zsl@X42njLnNRH)xl%8GU-OEU|ZW1E|Bf!0YbFWw70<z@QRvalO)V~;5SXo7r#<*-R
zDRFRZ58av-`RD``9ymEUCEr|MkbT(RM#_B79;!E%D?d3=BoJriVq@Jl<$dSCGg>S>
zGb>e^(10o$ftgmx)hVuF|2-_$tNSO79bGWC{UJ04K{Sj|k)-Lm)m@Yexg0vW6ro-f
zPGTCG-@kuTDcZnPwSp7>^u1Z^*RN7oc-Wt*0zd!CJ)w;0iH?3CB-E-&eT%ZCiZ3fK
z5XUvY_1c@e)!)|cD6eiFItnfya!g*D?%M*PBh?m$663~+_HcQ58CrfJr0A`zD6M_>
z)r;shAf(e211=gKN?x9+ot<3<qy8^0w1kBDVy#=y+IG3)U->%FQG==dMt8yBO09eY
zK?Lt&qOm)hFmV$^dq_!0n!K(+)D|a5T`T(bZAH>v7*^C@TDy&$TTuk<Q~*Wy(2yK&
z>LQF^-$q{F@u%p9Wd<fc_|`5BY<%S?AyZ;(Xy|)+c`tF&88lJ|vIp^sx`u|N*a#ob
zG9}+t0)sJjTYUH9q5>X4)!D*m|JpN;rsKc&;yIouh!fr%!c`GeBDY!u3TTKR0)#5>
zjKTG>AG8dEUQG|>Ps%B8n_f*wN9OCUfi0)wveZ-^VbSjjj3n*+?KV*Fs3=?H8-qeZ
zG{3+9D9lAK5y5EQIRJ=yPEHQH`Na0-reY2{r&+1ms1<7X;+d+pYKV`I@7bEadKazW
z%m1E8LAtnUZ&hviZ_x3Ec80X)4vR_tx!Q7d6*gUI2asejpEo4b_TXPI3a^oo5f}y`
zvjU8-@Jv*$;udP{U$L-M0F^8hfb)8XJ`jSiXq8*=($?#gZ-2Gy(%9Gu_@{K*4Qd4{
zv=GGp`L0?Ff0?G&hfY2Sq3+{1<!3MM;NaQOPSg~6SFR81<L>%;r(P+{=@M%5+_bcy
zz}MF+ooa7?VJ%<L^NfpY2GmVLLbpdI#-=0(Rg_WA2j0y8c?`Dp_8KK6y7^fNR1Ye-
z2yF^<1|3S<j?XQ5*to?*^}!3Uvcdu~<c9i{ha`RsNuJBt|2-%9P!=1MyZ1C}z?)16
zBu~aU#@{5_5mR-W^Bu?jGC0u*5O!U6|D|rb8B9@qijh2q5}fX0^zX`6_(va^%%Og{
z_n-CRl(I#xDdV^vH93?iu#{O?Xxb4^gzI%Lj8z=^NWB(EW_UlO<>1({VK(`?5!XcE
zz%UQq2pBuyjgTm(rxsP3tb(4i%6cg9|F*O7B(ORB$1H?!BdIFlnpKybY}xzm%h*~=
zvlf;^kDHu~fk&b?)^ucj?<;X!P@&Y2$;iac3)1@ZAN{I2I^^W7_6wQW(Pn7r6yK*U
zFh}h3jbF48tHm`T&xM0L>@LkayR|m?oS90MLJig3E7iMDg?F3v9!g2qXXfmHuoaM=
z_G<k8o)ekXKm)@9vZq9)<Jp^k@FphohP@{KX2^lr8}cp%D;@5VJ9p@(Cu_A&aMS9*
z7C&LUX&0XW+sJ&az47W`lbqaTf}Br=8#7!Pz+PfW7rK>tbIB^MaVoC)jZP!9R}XpP
zy}}=nw{O$k@7ZxuAF=%zXO75JkkptP6qm36>s|~R+g=D;SEUtoik&oxv#XZWV|aX#
zBb8G29<RbNVCxTF;fI;$@B+enKCQTWrNU6}wDa>^`hEGHh{z>??oce~D-t5#@*Ay%
z8ED6=A%8$@*KAp#yUo{R<uz%zL7gS!azG_w|0q@E+?u%cg2QQd4r>rsPPLhkxE$V2
zJ7?0Yv^2^G5;jWK=y)V1dU^*?t_Ad8(ta-kcyYA#s4rqe&75bxIi=Z&^pCm?;q%u5
zSAF8@>Qsn#qid17cogMbuWf;s2RUa#4A3YYG=`bFSYVeWf><&&n(M|t`Z13pDbnpQ
zhDQgxx+tgk1tJMi6SQByiAhEeei~q6F2g)z{-ZdgCJJbu+~=I}eRp-I6+H@2{RNR?
z?iZwJr%ysc!MEpWdK_NtpyKfK@PJf65KLmnvg<dww%|3P)rPFtwBcQ?9$VxaL_F+)
z5FQBCEP!$W>*7dp(lr+&3DiHW{WVhc>Xw+^>)pdc_4Zz|0}HnRRqVYaCG-zM|8;C`
zNwE(VbQeyo2R1O0Ky`LdE5-IjU)~|cXO8MQJwAT|0AJ0&ZsrQ}aC6ncJ{#(|%_md=
z8IN-x<52|&0R)oIP7svom+-((Dc_0u!x<yaho5ayW7e2?dGA^>3c&c$%3VkAH!DT)
z>bS;jdaxvu_X(93X$#U4sA)hvzA;x@4n~Kq`%fT92aQ{CR4<DOp-E)v)3nRtIsa{2
zpf}D#AOo~1Fd{>ok``dRei?D}urmRpqy@#v!lFmiUePDxF=dMY7T&|?rQBjL1pv5N
z_Fpr0W@HuyKH*;eTUi+c9bVDpoR3et*htH;gYjvJA%g(QpmZY3@KEkl+D8IASlQYg
z9;sKOeZWmgb>5?ixoV|<!Ex~XRkyuQAFX&^e%-4Xhq{8M0RtYbd5JGXR1#_pJ8^<9
zqZ*<A{P>Y(HDCA8?dks=?4es0ySTg4kF7gdOo%@H5(Y{Qpu!+F0DkD}e-}NN?yYzY
zB3*_QqFQs#3<dJ{qqE<5GXA@`s3+q=PYO^gJ?L&~d_)yEM3n`b2PTflpQ?!7h{#Lg
z2l$GSe+Rp!zV-!2h#_qZZI0qNXdOGSZpK(5YwUFO^<6IBTobBK5_EU(h7!?0x=+67
zTY-I1n6S5Q-S3r6bO=k<M|(R-hC=9$Y(+by_bOM0enH*Jo4L!BFcaF_C%cD4Db+80
zl`Mv9#p(7E>_K|RTtg-sbE~ku0xv+>+|>K2vGHpAm8_FnU!a1kQ}=J3d_gzUXCDiE
z`ubltgzI4;=t9|qb(|AF7CSle=Jl)q6qK%2-&&0EfmV{oD5$+w`5cncsL0_xR7J-p
z+io(OK8me!{??phJ0^s|#bjFX3%786Z*gyD2lqp4N=K7RKuV$%@}G<*K95IiGP4ZT
z5gN4-8a}8v0fOn2ohg#R&CNFeLSp$mh6tImk;K%SNIxS(_m^2skAjeLM@{Y@I~r~L
z*;M|G+++PgFrDAZUyw^HPf2CdhLZBy0Xnc&X?gjwT_atfAItYYT6<%!5vI@gd)&jx
zGY{Z?lI@wIJKRgi&mMKZD{M*4K+fs+Rd`b2An#J|w7QYS2yrb-j^;y|uXh7vrdE};
z=VN1m$eh&gEbY0^|7m>-5Vr{s3FM<69ySmJO}@wqFfv4u`|p#j8<#!KG4AU;c{iv=
zH9uTA2=0IR%98PJ)Xw=!IuxqgL)}pU#|vmR#ABR}+{ywyW+t7IzZ6@u6sdgOYo4#f
z6m->=1H$Cf{IGlw$#~{}0N(<zC~guV67qN$elid-Rf)Wn*h1exK&?B<kn`PgxdXJA
z<4QIjF=Ff#S=!xQ6i7rvvXb^$rMjkfimtE?jEukm3%W)CLvCS3WmX(g(&Tu_Yaj#j
z0XBonzzWPBKX~u}n)W_7WKuzZ^6Z!+cD2_PighZ^gc6E)j|Gvq8-DKK@~<DyD$H=d
zVCw3wdYyu7&`P-T<|ZNlmsqb|!eJAg3pbpME$Np^<??G3ET*%y6{Y4Q98Pfpem=|o
zc2=Ci@qC?7f`^LPQdrv)_i>Zl`jc-w_EIE<yeY&%U&_bHSqr7kwUR!tbrXa-!sN+N
zJ1L+`aXq_>3AdDq8b<!p0fs`$oLCya(dfT>-5-tB&qFWpk>r3UNyGgE!?ZPAp*3!-
z!%FL|{ssmk{F|Ab<>d@efm$xABp42Pr{MV8Ih)ORX(x(_BX4A;H}!R;Ww|}`{=Fp#
zqJgK$L3pG{KTO3S!O9QoKGk^|SUdv7kvl3G`p)KqmRB|OBH%M=rlE=+f1nV@$H#AJ
zb=MjBr$@>IH;qB=F>r+|Y-ybyhh`?iai5r04$4_h<R(pPlM>l1w^o%6ksiGWq!9NP
zY<VBO%KZ+)lt2{}eGx(R;I;U5a%|Io{y9|77yXK==DLPt4&%NvpFM#z{V!ipvj|v-
zp8qFZw_06K>kPK#UnTd{%E=H84p8d>lvw%9ONoXiY?=<AYYDKpO5SGj`LgMR@L$|+
zPSW>-v_!(WE-tgEWc|)g9$uWV!fj!#L=`RneXVrNcWcP%n4EMz`J(@x(YVUA5fIRP
zGMRQkWK|}n$(Px0#SIZ=%n64$$C?h6n{WM<oQe$MrPhY0%R6Y}QDSFSjHI`BcBs!k
z;h>{@^OnkDV&-&P)f|og{~y!fdXkxGlav&h++d{0-RocX?p;C_Yfgz;(;N+x=8i5U
z!c|Ms_9oqcRg*;=3`;xCM6=GwQ=T^7C7t23mFBABzWJ2928Jc#Qm(IJ5kvYz>qJ|!
zL!-J^j}-ok1FSsrXj3SXLjLu7`+Ej9JV*5kz1L=A-v`CW-8ST2kVL}=skohVh5$~&
zqK-SRV&J>lx6f80N}$6`<+{GZ$74Ad<~E~=C^MpM<GENdFjHc8c|#^8Wxp`SC@UtG
zJMv5E!CqS=YtU=wvN9walePz*jS1*(>dhs(Cabx)5AVN{A$jA%5Z+l{QIV-Uc!P}n
zc@cmiC4Bhm@KMn(E-ps-Jk=i%*fqIh5DAP;j-nxmr03=ZCK?=-!%Mg0lagk$UIxB#
z!<3fRY@V1teXAUv`!y)0d-7G5rr4kyvV+g#^o&*c)*m%K!S&as5X`S!;-ZWFnGZML
zWwt~H6j{(wA!^Ppm5^Smj)K$OPOrQJQT9M$rpxv-SC*I*xS|Lj$xE(+(i&JiNVNv~
z-vz|~QZN6E<=rn=@OTOVNDxgfMnS}85f2b(xO48~o<<jqM9tFw{7;?L>!K-;I85x~
zo;;cQUa(k3m~XsaCE-i8pn{{c&zA3!gbnb$Ajr!=)#(lYuVfq@EFehVop#g(x-32M
zE}{hbAtDlRTEtPV*!lE*GZ4;w;j{tG@yjeO33(rVupw&vm+@W&<Cy>+YpB%(6I#;z
z)nOo@VFh6qujImfgmRU3*RyV7dV1w3kt!6{3kyxBin-S775-;pkjIH$N<$6AHd8ny
z)C2IVNUtfZSi^WBtOzdc1FBy9BYW71Gh*rLKw3^NA|vBJ=F1HAAHv9TV})6a4$)jW
z9&?EgvD+uyPfH<_=KC~zvscU1KEh`(#q00V$K+(KT9Q?K0g=eg(UIz`%on|@2%2%V
z-VmG%Gh6S18aWU+e^w9_R=pB@J733GXX{HUtHV~L%MkW*mcf-^J=M?)_oX6LOquo@
zU6@`dSFr6RhyRwY@R|KbMP&v6Xu#DuMA3X4T#x%T#6pByJ3a{{R~BbUR;0oc{jc24
z14HT6>H4W=zb@r>BYv!g^vURadqWTz1>yc&p0!_Ev?LF`qG!_5(*7-{8*z9&QwvS%
zfJ>*=_OAj@Fyx{N7`GP69r07Q%BKmlR?vzQ)&EK(4n+Fj;$21Lh@3%xv6gl6)eWWI
zYn1}Xs@zE06kqyoTfj4!;K@S2cgX`^bl{B1xLW!q&LB#^gX#6fw%_2oEVmPhMkD{-
zKTbOvC+C$#MV6+b3Z7U4J=$^6=PDHFY0^1^3`5D^N;=c!C4e&aLC$4p5x{QZb@?l=
zS??HU#A+XRvzkY~%NG`rRJA)-XjF-lSYQGx4y*ArQ=Z^6Wy%5XYq#gy`qYlcPhWWE
z_x_W_X1rKcM2tMLq?jlDyXYsBE{EAlgpeELIuYTY4g-5yCe#CpmoM3@=HlD+`Y#WJ
zsbWC<@O`B_6y`*Rk|bufKj(1yq4%dH7n%n}l2WcT50Kv9Z=>*>e~0c?_ab-i97MJ7
zLPLa1ZC#y$)g1K2X7-2AIihXjoKm`~_DTpV^ixfy`<xO{`Hh-6zSB-QsL8G+Yw`Ym
z!91g2--pJ)$_i;~CUSCe-r;DdHlF|9Bmechsa>ufw6JBv6r9U&PTwr)Ph_q4+{NK?
z@VjY;P8a8j04f;_hOt|Y5O*t4Ump{r99J_{Z>{5-5jI;r6V9!&_C<+MJE)B~QbDlH
z4-t>bC@5+!fa|ONqSrYzG<2ud;fc|pFaKW%AcTAWhBlRO-QZL%#(w_vOYY?x*P?lM
zv5QzdA$1C#D@qguB`@FAg25;K;I&4Pie)~$sOz+FKCD{JYbX(OcXw|g6;hJ?PNRYc
z@A#+>XMS7c3D)knMD;uG63+P6*IyK$t3(MRhac_m09@Kb<l+S#0e%m{Hj#dh)2;oK
zW*l7H0yB!3AEKoO9#RcLZym&9BhYb1hu!2#c2B^2m&=BD<MHn<UKm3$S`xZ5ZYYd1
zz2+8RFSt#y(_q1eaBpEW`FqSH21Ma3CLO3I=K-2XI$}P;2&p<vN(JP*hId>VqzKGV
zZ6NCrkSs^XV;IU}moYJYAS1t7tiyW0t)l~&B(?#GoA|`FU-sXjEFmC)tllswV~-cF
z*g7vO`(-Tk6wdiKJj-mIMiOoZlZ8af+M2q$*uT)tGJY36uGQRy*MmCuP~rn<!1D4Q
zlE1p3)2Lyj8|DiwLB)KZmNI}HrH(6^x;Il}tG~T~Qco7qdyf?E%gid)#^V$P2#T!f
zf}W+)f4%ONd*kiRA%~v%a-Npa5{9mh|F~P;Gr;Gyt5xQS(7EUH)PzNSdW%WnQbtJ$
zr?a`l&Po<1DP@QL++yCkMdCUledEh0D6T$%7jN1^*y{6J&U7Uz?@I1vaCV+<PYc~`
zgT|V~(#<$H1*0OKWK;mUv_I2X*Ul^enYrh4?uRe;ujkbXzC_FLonDV_h_E3Fj~#&e
zhNu;9qsMSz<6Oy!HA+;s5><?_Rw}#TaQ44U>PrW3uxV?UyY3{?vWy``|B9?K8#gPm
z99<<lOjk-dR?E%KR-yBbYreTArk?a^;XfsVO@9XbLW8GyK2JaC{9W84{3w!f4^`mG
z_#w8yUGIPGc~VI?IticelK7?YhIU4^Mv5(CIriK8u$ol5)2c17DvHHDwrGBzmUV6e
z^0Iu-p2>f~t-k=O*Xc>8q@}ey%l(CMencTI1)_F(1~pMp42bq;U|B+5Vt8U#t4TqC
z^ppv8%Q&q3#q-G4dne>>N7t7`i{Ca;chujG{p|=<w(P+XxXU+<@&Up>go7WUWoC1Z
z=hp&Ex!5rp0J-A^Wj=Vqd+m`s-3`=F5~%Wk7AnmR@S=9a`;b~250|y{C(mA-@qPZ>
zQ0?e1D6c*5!*oscXk5sHQ+sC-tLGnt;>Gn|8an1YN5$x`u?+6hN3v!)CihEl*zSCO
zmb%+%!rsr@c&0r1E7_o>#LUtDQn5@4#`FgLPQ3H#60V||lYEVwM{#%3a%!hRVTDGY
zLPfg;IUwer2N!ImiP44IPaMe5KD-oL|H><JKQoF*^UI-0oj@o^9>5<RNl|Nkj1Fdk
zeBM`xzeh(nkGYT?`{+F`)2dw$VO-C5d;73G!(w>gfwr6q8v@y8Xl0-zeer8+@s{mF
zP<;08uu=~da+2}HHyGM0;|D_ZejbSZP9c4KmroyL73A%t^VgFhC`^*^Uy52R`3$L+
z2YUDB7Rp~uDd~H!M@D}A+5iu8W9a?OGn6q6R`tgoH}WDKN3s24OL{JT8G@xQQ~c>t
zye5Oi{l)`}aC1oTAr%Kld~D<!Maj^YeJW<A!2#jQTN}tjqoYZnR#8qJ@)nbp7ACbv
z<XYcf_Ad5wp9kg?Y7WqkWKiPiQ@*5A*fKMl!*I`>j)_3xFf*mgO<!CABH-5O%ucKB
z7Z}K%yeaaKIRiSSV{Y1GJi*y~Me+jgg~wGJ=9q|VOOk>@D%Gsz-Q6aB)%n0)Wn(79
zQ7H?bU@or9*_WrE>XPf7o@OGnU%Yr_e8+meE+#T^gguEbvo=)eSvuCLgqCOep^1WI
zeZK5Bu|#%juZt03ht1D_CY3%G^R2*LIEq&2dJ*?ic!PlJ0Tty#p02WI|A{vrg)}RS
z@&>;^JVs>{)3A?C_d=;OQ<QsY?ZymMu$><HA)BUrL@J-d;}P$}jP>x}1br1Or_FBM
z>pvFZN+#61j~-RI=Tqy@aNcb8<_UOe^VC-FjbFqJpsYx|2$l(dBzYtn`fA)eC@bn)
zJiB#DOR0Xzz`rW%MK6d9dNI8U6)v6$g#=Hk|K~(e{t{BGzm)%hjmJpwzqUPv*9g2&
zo8fU=<{{>+tu94qYnLoz$raz^)YL`qFyWnSa&8?s`sdYlPLNVnBBjBXuFr0%1zNbg
zFiVT0W1@{eq+S=#h?ho8Qp<&B?~higWX@2=)v%%aGV1>HQ`pfnG72csfo<C*&*G%y
zA<Dg1QKU;*<m)*FRP|d-8}}U?HE|t|y>}eRZ>KCXROOVW%b>6@ZP37~OeLB4=)zSL
za2FoXG=x&YTrPj#?*ux3C04s*cwkXG7twlMn7%rc8Co>B0JEfO+Yj({K#jaobh&Gk
zs=YD{{Q74O33*_g!F7Y0`Pa>69IojwHZ^^hAPVH`^L=fyo5-h%B9Vx`F*Pez91#;U
z6Qy`%@nioUh1cl$0Q1Pbqf&BpdpL$8dR7iCJlTj<ikKVC<LYFtZXXgjur1j@hg4gl
z?Z1G1Qt?NxOE<8<o^)pB1wqB<-l-Fa`?H?I;n{u`hp)pypJj&@2pMp&zfm84YEPrE
z<)$_#5BH{*%stQK4BRnFWADj~E&E=f29E;!bE+Q!$!ZW}Fr1utJtFWZc07DiWWk*<
zad2^7ncD$(@$SuxZ+Ufo+AgGF%Te8kdi(6}-|MMOtA{KNKx(T^U?9*@NF1T(W4PS}
zMXzi+w0M!-6uia$fPOUcyKeJO<>V6L{aw`6I$3bB$=|!Qg*m7PLK+)55^G!>_xt}e
zxC>Bp?JmE3IKu?D4elG>VEKi}QAMAlKW5zB;$MKKl!QlaW@M!Dv-Q^A3B);&GUT8o
z7oY37^_{irGm3-@k5zwhl)7Q5{6Y~&{(C?C0L|164Z$XWGdZ4;Ut+H8IU)<-5?p+u
zgd^(8WVg~aE2)Zv5T798@o4}2@RsM8qJg`84Ab4^I#pO79_$h14a^B=isVT$prWCL
zayUE|3ivhNJ8iKBJ`ahqNgy^+P*6fcLI8w~mohTNSC>9}8oRx^PmjrzDM7%Gi?08z
zX~2btv=Nyx9ps$~dC}?$uOIv`Rtm(n9tS|C54|Utt^Dmk?E+|thf8=s=-)L*aC!n6
z3m3aMmz*Al6`4~R=&;uCaCq(p{U0sBLW5f-1#9pw8z<*G$_H>k<@&|k;v%^fb6Q65
zIU6@&ZJd}D0v9Zd(`9gH2F>O>Wn-}=8KaX0mh0p`i<m;IE|nyswDruQ?r##r@FvPB
z(ZMAt<Q|@$%d(_=lUe;0#|31G<D{0Rze<gyTa1A3p2!T+K9&l7SjJK1wiv4wb!q)*
zxwIK>9E2Nle0oX6BH)YYt;bY=p}*~|kSO81ca0Un@-*Jh6<)YL`)`!n|5W}t8b5-k
zh6u39dOL4Z=li#3Z}G8&O*DQrq+w5_vj(<cIz8dqn2RBeqklNF?e#vbI<%HIFzR*8
zvFq8J^PZgydm}=_gXGxQs<TaI(DyBkyld7K5P-z0S`BdXUYoA<5gW@or{5!^R#j;e
zR~ALZZvRqogwk+CQA>gVoPE<odVI3-Y?ZeAHJk-6Yk>H8)zsA91O?~i<Se`Gutmv$
zH)swHr`cHUmulQ+QI5gEW9v2uld|~!<>pjA`-QHq++1FN!Dlr$Qd%aa>lzrt!^50j
zmo-XMhNZdjq6o3*?$g76qGy{4*e?(~=$;M518+T(iM-V;EVOyOuc4cXl&V=68?(bn
z&L_izx=h}m>P}CmV>Fnuv%7o!uWdT%KXi@Ftp?duOR%6q{!}e4X6;w|>j&)Iin&T;
z?)OjV5^^SsL>a*$3lejFF5Arz@`_5JgTw8O?^K=bjCb6sf3zFTtgLt;B2v2lr)pS}
zM5cFEcJkG$mlb%yo7pAY3aWuZ7o#YwvA`8M4<K)BESyXmKY1_)v|c{hJ@{CucR<hh
z8AvkFV~oLd;oz-=@0gg?EIOtb?*sa|TetPk8_OKI?^lCE`Wa&AQ6yaXK^bX0apt<Z
z`>od?CEFu;0I_=Ge}1~sRgWdmj3-A#M3fp2=>DYexH>Rw@wvJ<U7x3l>vf)Qa11KA
z5?|!ULYVZaB$?*2^GikvwEH1H|FS_R5LgWO)YcIg<|S-y&+0QctK8k)-J>M#xOqBz
zulcU*0_lM6*c(c3?N~}0z%7&w@GSeo$B{!SURH9(+~kRtOh>j|$Wk<Vl$uxzo?b{1
zU}t8g?pcTI)>FvGQW#iMy)7$`w{Wz0Dy-ap{XFxeNHA|bm!eE`S=Sa!{`zMkQNW%(
z&$5%2>GA~Mul1Kh&i$4}efiPH?8ViN$!huYe4gjg@$pk2zti(cyggkKc*+dlwl!;P
zbOmN(6%*N~=Nj$XDyiYU1}|$zSLtjF&EZY0%-48(Z`C_Z-IqsXEQvL#QFE6%y7!2Q
z&4019oh|q}QLA43`xkw(0b8fSruFV;fckhKhYx`);Otl>Vfp6K=Kx{Pz57q1-mARI
znO`6Dv|7v@6A^&1ea-H_AGQ}SQ<1I!{Fcin<+h$-+8$`+LcxH{MWs1R_>9*iU;`=x
zgG!s7|5l|`dgibR$T8v2fi|SqGUMbQ?;qHb3Aiy59DN(=Q{T7}MCGIl2q7fr*JIvf
z(^FITj*lO+DdN8F7B!OEgF1cob+Dup1TB0tI$l5Yn|zB0MwAP7jqg;yJR~!?gEWL#
zO;R;6yW_*ZqKY!AuFNx~bMOBB`{2MD*@Z&I7BJV+QG|G}eF}-*z;)B#v_M+-r7nPO
zArTa&=H~T1i#Lu62dYg!dP+<9pF(>7;B3JDHSHfSOoVA?@yMg1!(;xtsL|NUI{dFM
zwZT{5G&?{>$|aab()`}4$C~0++RJ#FNslNQrh`8YT%Nb?-;e8p!;zU|7mIpGLs)v>
ztJXAyuEOu7z64lOYhEBxGiRDP-)(!9UqxjC%(V1vuTCb0)$5%s1*)rigw6&D74iye
z%zR+M_kJ)IDe*2sA`77K`dYmp!`P*aFWv0LRTBlqyH^)*7M@k({`dT={W6o7!bD+K
zrqHBMatJB+BMeWQEsBYv1ui|8rXsi>685_pc#Kz<mj{-R=gYu?iXg%@Gn2?F`zl-V
z`BP9(g8M2Jn|0^fjy)X)tf$ETC}TrHgX}{>YEx+q?zUlS?O>>^xl4x>jjI7*7J^@6
zXE<PyrEu1KbF~KtUfC0!uNMwHy~DN-Zj$w-^+Dp#kYy?iuH>dNS!)Z;z^Kk&&UP@>
zDYFvXR%tA=fD9TiLZ3up<xD#wzjxnQ@#-YAG#^)rG`PV{fveCFCi{!issloK(Hi4~
zrw%j~_x1Ic<9p*kmu9K=8xZkC>w@}v_uN^U$B!2lDqGvyGGY0tO#+^$!E^Tw&Xa&o
zB~eLibmG%hW&%=DIE8<Nj^9{n-MC5>5+*xF(#u6qAWfOu#SuMWxh(M}6py@8Eo|$l
z9Fl9Uw4K+w6O4{~T@Fcm9ifmT0IE@-B+e&gpM`z?ELhTnkjv<ZSve@AQ&KnTiyyAu
ztA=ImQGx2^`D7@7y6S(EE>)D7z%~^Xqj(k~bC*c!C4xv`2(AOCWqo(`^R?hh#rVWv
z3><-BpAW6SDnJC7Hq$jQfS1K6*NVwP<TeGKuQz)taAPsy#i=Dlbklr2ICf(XYn4d>
zpJ^TK{1n1}#6HHBP@LgSL#rha>6A$HG1gILB5(ft2MrTVO%fWK^demJ`^ohnUE$iB
z`0(Ka<2oJ5*6^%vTvHiWd?{QewLGQ!SHYI{y<5Q}W$t*OeyYHJ9cZ=;<0zQxTp=)E
zHRLn=Hs^v^VP!**xs{O+>08n|!RKK#QP=OYRR8KH?*MkgsP6X7r#_rx`zAlJ%v@ax
zqHU=W`!yisO}nYZ%}h)jlzYBJ;$#s#6S%b*PyPTu1_T6vKcji{ye>O?E#vBf3;Um0
zN9Ax{dB|H3&U$N&GCN8o+D`fJD!@DOR^+J=qL$=z7YO^cdAGwgX$S;@SilpWf4iPp
zU)y9sZ0L~|I3F*$&c0y^FmpwI;GsT5=++miZPF=oZ7<lv&->rTNRJDiCJ0Mq1$Otz
zg9io$MLZdbIbf@S08>05jow|Ie^m?&>ch#Z3Ny*QlNWo)Tnmy-cW-hzV+IFq$9aHK
z1Y#oKQkua@^V@(rcI$zzt_^UE-HO&W+w!B+3b$984kQm%5HxCqF|4@fCRb^cq!8V|
zP~`~CP9e`>WkddKLJ`vO_s{OnRAKP-%gg2<2WorW21LNcEXq8eBki`vUc1nsC54`l
z78mOPX1DRb4YULrs6?nHo%i1*q-2@B(VTE9ub)Fhw>Xm=)%3dTuwtY*o+!(h|5(?(
zyKocUU^r<L3y2H{F=WZ!xJ0l&{g0J(yuY8;FU@zbxzSMZ*20qt%H1{QiObNS2r*TP
z?75yk<TQbd+9w+WjAQh=XlE~_6XEj0njvZUbuS<Z^DLOq&CHBOqp|Lds1y<c@OpkO
za@NpieGmYv%tHe2whj?YV0Z~qY+jfD=pt!q6<jwWTGE9A?ey8E1=t_U-%A@6MhT!g
zzU~iSJv@7v$oV@Xtm+X263zaCWD>!zqAl4-W#y3daFEk4X-plH6sd@RFk(m{{tA;&
zYX`_Wu;wFf9iNX$^h%0u-2wp2#-!T=&IZ?g?ojVBDz3l>ebfB<#(1x(HIb_x@gdc{
z{3-tj24ugir*v(!&oOr!9|{+JF5eX%=CajK&b6E;J}qo7a@;Gzr0iKpLSQAFUZ9v?
zJ<xCSqp<NBJmm?eaQ>R<=YFtonBx3~WAa0r-xF$~N>V&Qu^io>`<?)g%t=uEYcJqL
z^flX_!mtFhG6FGiH5;-X{_*MCWcMCYbB{X7P$E{nz9Kx~Mn?V86H$>p%)_grX~<Ao
zIz&Q*gsj@Rt~L^o+SKM6IJp^>y~Rc>h-@2HP3+#p`fJzP85*jiUsj!zi8;6q6c$Pn
z7ARCo!~jbgdSq`rIy&n5hD^pt0P}UAaK>VwOpfHW^qn?4&SV)3gvkHL+FM3d74>_g
zTa=WN?k?%>RvPK<?vj*7DFNv&X^`&jmhLW*MmnYM#OIuI$9v!N@j8ZsPi*$uYwbDz
z^OwVas$leuGU}>si82m>wGc$yz2<Wcvz)bB$}8U9-p&wQYWMY>uRtc#?>tH$N%dVN
zC@C6@8R(-bK$4P`N|1DSzD0&@p5h=_7D^5{19Owb^eafmI4V6oG}Px=BQfcTJ!e>0
z*kMjo(X{|_g0%t=I>%>2V5Z;95Ko8jrwN=~6BA;3c-`=5q!%+QbMxdIJ#lex5CQGG
zT(DSg@7p<i@SH5pR|{f*s;P|U;4wNTKmtJWf;%hEi(mF;P1Aj$G6mjs$&ryL^x7{N
z8KV+ri7<c>N-G45G$5)872}fdLoEl$2~4+*XwWGWCIQMnA;+kIfB=x$K)d?>)azcY
zM5EKFk|wmt?fAO}{kCHxJytPLJ(!3P@p)mBq7A0$zJGwFvuw=b19|QWG$os{z*_>e
z=kZ#a!5Hz&bZ=hx;fFOQ*iXgS{;kyo7JZ+Cc@|I>ClyPjsjaOpmqvgrK^7gCkf1iY
zckvoPl{6Sae=CcM=508oVv6Lw%he=fOjwg=Zj8lpCIs=9%DywN|1Ik1=s=2?y`t))
zUTX=vvfu=X62PT6JtE0z#>xa`432Iv2+*ESu2mJncD?9G$hjI%a|zOniBlFCz8}@a
z0^L*;x2Opg78ZmeiG7Qexhy82aAc`OWb3Pv%z9Nn5=onO1-yfo<&Nw4_k($zCI@!y
zXRR2~Lz9hbNcl_9T*WCUxXLT&+E%}Fu^sLib+*K1hx1)f6bnnCk1G4oUuc9A?74@>
z$M#e99VhGEuMrSbgp8e?ofjgWr@Emb<tLfRm<s=3JLs_&_QVFvq@?Rj#ZDEqMb&z5
zMB!ACk{rA+%WKyP>yOXh=UEdA+Rl{`f~M&PoK_7+`?LgM8L4awXT#&qaqv4+gdjJj
zd@K!mNd5cq<U^#;BQSJ=I|yX_;2-ELRXM_aZc*}HWMX#T4=gzVH?S1gh~2QvmcWn+
zlp|p4x&cJV@#5*XOp6SAZRW@Nz`2Aq>q_UE4hMkXgVhuc4(`*{L2{k32wYYXOqh7P
z6@a{eVnE>L$~JgH2*f^NLqe_RtgHS&Ui37$`p7Nd3ZwomYsGNAA9HQdAn~ih=xG#R
zOiwjM0akaEr79#Wh8KRtq_4^FIf0QP<M0zB7xD10feX!WO6<>SP0#xuQ_<uA{#3h@
z;RCD$u-NG6>Qc+4S(unyj2u2^vC_ZA#9|Ema{n7ds}S<#4{ADYo#{`qk|Tuqbq0Jx
zl>F)sd%fd!aMHbQ-2<ph^OaT<C#)|On+GKexY(oWfBO<dv*y8}7obM_bLCIxV?xy$
z8B8zyQAKA^BY^$_kd2^a92EM1@!^FqJr=PiHg@eJ%H!aSaoP=p#Bz(S^v~l~#FSQ@
zpK&nhk$}x(4lwqbqJY?8(e5B&b4dCNB>djKKA%S8D0w<=FVszhUkeb(r&{1(jbGHC
z15&Gk<yK$Ei?QMcH6VafHBAJYqJJVb#4q{ArBReD+RG_1OsrS1^7Kuh82gtbE7Q{w
zk@QlXDH-&BkO^=IyD=99eDWtY@ULEo_Zb|6Jyv!79S|1s{T*qJ&FN=?)>75dMIjCF
zc|}E_Ukj7~(QZc!^|i+hUcEq&118+p0D@*=TOqzqfr&CPqPjX`aO(%Wx}BYZ?`^Sp
zfe+E$$9ym_^6`5B^hF^GC6Z_sv~dm*nvbN^)=Isrzt{Qdfy9dYKFxBb==qYsB5v40
zVu1g-V6OA}wCLdX))urmxiD>D!*x0$QXH2q>f0ZhmdV{{%1$PZ=2RQ9NWswjGM+>L
z(e(kyui8|Yf)L32YVd-WmLX-RbunxU5C0H|0;T|492|%ZXU(6At%cVdFd?X-x3BFB
zy;SF0VJjRtAb#O1w;3<pkBD@V4>n^{^H~SjqTjF9mfolj5=q5?R?-f06kto18A=4f
zJjo<q;j@N@N;RmYs=(g`z7k+i(Ctc&1)BQfiJagjeJo&vfc-A;`HGN~uDY30Jzf{o
zP}l8jrs=c~_VrPb&`46C;o>C-rC;X#aQ2e+DA9m|2qmX7R@ZV)1{Zcg6o75b-&mY*
z<(4nFJGsf{A(EZ5RJ9n3-xGm>pOmF@Z9*?xge)NHvCFbjp93hA=-ABtvq)P!cs?E{
z#C2usd$nk3Xm3j8zPV!bmUvb4c79O2U<#0~uB{PAhNhJk6Ck9N8q*C7L=QJR8TLuq
z!qS6TnYe_!_xc<#&=mq)z8}Tw0<vW-YgAolA7pv-`Qk>EhH5mj6u3d&*URJ8>&jY9
zFfAnZ@_HZ?^wNHwGK%l8tmI~>Lj+ICB}gYKbK3dLQ!mrSmnG}b;Ko&!^;v^4^77B-
z=2mH$0-b)~okKuSP??i!APf;nY@Kf9=I%reaR%7`@^ek@m35#XZG8u8H+6N0tpxnt
zf&|hAU0L8E1V28*OW^7@(*9so7j=0hf`f<m_$y6GNJwZseTmI%*Z_N{QCrgi2?_G+
zgWcmu{V|p6Zf?G0tzV~s%7KrFrRCcsrx%djZ`Ps{{ZA&oJJ{bMWrh<6G$|VASM5Se
zYad<bGbb!$sg!fV=Pt0r#dBz+-SVyeVDP5Pp$2$7MX2Jw5*fE(zt<qumEs1aGt-~g
zvBD_PLTK+*ft3Zcb^ihma}aU^$VKlrAM;J8gZmJJb5=iH6J)Mh&}jxraD$WT_0F>U
zcr#Or2S;<bzLZo*Qnn~-mD~3AOJTy7vsj)>w=ncsg6%wj`<LKugOT8lqrqPv|I?0`
z;!jS%j(2G7Vm}jF1`5?ofp)C;J=*S!{ZnIgXxEp@Dzq;Y83@u?`b~?D8kyzCzOl+c
zpbl^nd!XS0(ix@hndqCF8(_==0TX~(o|Lu@2K0~M?Z!2js+(T@_4M@Y52f6QnIe<v
zg)ol~JBJgFMNO||AqAj{rUY^jY~k>KI*E4=gLDW}d0)(RYT3LDYz#apn27t_S&&Kz
z9Gk>kL@<Z}HPfUwJU~zQ_w;9bKIjy1@e{h(gB3OGH;)9SJG!*BAEpfP<TruUlhpwT
zf$-rv*5*@l)0<0oyDv-D6`;T+ci<sYIt0A5<{sSbkL#80+j1o3froG|x2uKO3bQ0<
z3qEF;;zLm1ym`_6GAo`75$6*#j~GC?8FK-E@h({I*Bn-pYr9P!2&QP=9?jE<)?p!~
z^meogQpmFC;%wOdtz%mC@2dRkRR&o=Wa|2IBZub28%E@R+YV;J7JVcLJBHx_D=r2!
zB_$%vJ10_*O6J3j4cXjGN=dOg(vfaKjgOF%lZlg*@+(wZUYJdA*6~LIqY(nw@g*!R
z3+a#&=4xR5_#^WX>zlUhfS<_MeYwF&p6H$3(RQUVyH@x5;i16&bfNbUmPp|zA8jG)
zKaOZ5AC}@%+eyf^MHTsZALKpp^^NI&O2Cok(w3~#^ryjZaLraw!e>j8!4G1V>DJ2)
z^ns`x0_6T*Vfc(GQHhED2F$XuKD$`r&{o``n4y`|!?A%-5DU|u)op{J9Q6?uOzZ!A
zz7;|5;7is!@FG}M&#O+!;TpYCkP<22AO=&onhOhq_2t8zFpTPXa+~{?YV^42X;{j%
zRFHfP&EeX*%ynA~@Jy`1qeqNYmZg>Vaz_2!0y&!uV|p$q1UZLaYxX343>`f3B}G13
z){G$69iDI^h{9zdbOyYH2-C**_4H_HYA)=UvgipyfBCezd+X+<^ye235FE_8Ixmvu
z|M)fi?Hx3D+XI}-+Xe*%D}!LH*@J2c5Ec>X1mVGU+`L>r2ilf8Fd-BpS{^lfyIF@F
zH8fJRgvoJNN-A2e_@)uH(4*oaaHJRH%<F!jd2e}ld8I`2oZO9xC7{=l&-3FS*F|LM
z)H19@SxKp}#DQ=r$Jka|l!(eYa#8SUUo)>mLf(rerPzdfDJW~CQAV>iUzgT}Au2*r
ztDF+ew`m|&l}VliH+J;-b^hQtR4@$+6nEF!pUU0Ju0|!}6{q11#)pgcHuK0DjZBml
zCx;23g@u$yF|nzihjzF^85e?e05NiV{jE@;CFFJ>0hLk>Zm)F+Ui@;@>3SJ-D!+&<
z{|+d`!N6r|XGesC^LMLy!f4>6C+<~9oCs=|ZArgYlB$Ov9BDvNl@(VPX|$wxBvGrM
z2_p-O8!=}KH^a#KYIs!{zOJqnTn4`vNO%Bs0zl4@dnh^c7U(N)#x~SX4!(p1=u;8A
z`SQE$7#*ELi74HtDCngbK6usXvPDV)Adulow-Tnax0p>knoIb23x`kh8SGfBfn;q8
z6{cqaP4Y+IzjJv4q52}Snr)e0E5ZZp7YTBUrN@%A+xa<F5fKZs(I2_fXo1;0OD9Fe
zvD8Zn%~@mts6{bFD-4))baav*81AmSV)$k9S;TV6fBuy9d1R;MVhSTGbFs7^HgPW7
zTR#8Y>v$}<OHLFC>qZRu^ioX36)U*=g`ZKNk)w2<rpOz}-_(z@r?!X=M|9w3jQntg
z4+fs3vn$&j=&@as0i-L3mwxB$3DxSwDXW54cW&md#gZvSjq?kI=_hvg%i~MH2t{6|
zx58)7`BmCYK7H7Pb%SzpU8;~J&xluKxUvq+Pg#dJ&(pQ(jPog{n8*uiFIZAEJ$czI
zzd@Qej9zcNTP0u%MOhxZX0*1=DsP~j>W{6K8!GW~CV9b=R4@gUIpkGU{gegRgIiea
zi*gct_T+VS@1`&tKBc1lzyI+WSJm<tXB;398jQXjd614znyyG6U+3b8qf=g&uHzxV
z(z_5>U1}Zdwyg$xEKp*!00Uu8mW`qC9s7wFlcYtH*o<Gn*vxHM_Ze062bxN=HZ7#1
z$vZtE=KaKOq}6mt>m4K$2I7!3tUY^g-2aq6=k#<~zg9H)5Eq$?acS)_+tvaF*|oK*
z3bD7ARLiy4018du+;nhr13Y(alkt1-v~v_DI4JCSiX%6gaB(~IbX?VrTr?q33p^5@
zuM<L0-4143+WOb=xnMH{)oBtg;1K{EQc;&PmFbo2<t)8eYGk-@Msr@mKo@7z0XEIx
z;}w(@&yFnep`4ulaQuQJ$>&d6q-fURD_h2QJ^~`2U&D0bVK2z1Gd`)vA;BtAFfTU8
zqlV<V>?d9w=wD?@QUJ$^v&R<F6(|A$2i=N=csC|;Ju<?SBs97NW_Tl~4+VNbB?!b;
zAZgZ|N*4O%OUQ18*91UGrm^#{43UD-G`h21%{tJ0<{;4MRurGXL&hqXTC@Bk$N#Od
zx{uDsyWQg+Z}r{Iv3nB3oQ3m#o&t_|8pV6Ez`(o)nes>!d^|igtvV6KX}~F4M`rA@
zfBKo>`>4r7-KwZFx#w+EqS2Kz^;EWwV{rPe{Tg5ZQStB)1(hBXqhlwqn9?d<iWbv^
zHa}cf0{SN1Tg2U2zrxp`AZ5Xq0gqSG;dF31<yO&su*EI*`9|)2VQiIuxJ*=`JzS_R
zsQj4680Z0z)0Z$d8VxuQQLNF|aBx6NgeS1^z2ytfo+fLkBbPxS-|3BzPAV%XA^KaW
zY>$U0qPERsJ@<oB61)M`??oI0&HcB;I~Ta@TfeF}?VeU;3fpN^F$%^*G;7!1%Saim
zv?D>t+L7Ox!yuVJ1&b5=9B6=YcGJ~Gu#iFM{-6kHXd%3U8leQBRDy}wQ*<mWWQCh5
z_0V=c7f<P(V1_Ti1p)ZNm?%@U@o(LFIm3t(aQG7`i-DD~Eruuct=-7wDSkbd^sC_n
zWncW?MX;2)?+bsPEL$qjnAo}KReei|!}w;Qz@$!LL_zuO?3@RP6!XM>@9e;0k9L;g
zJ3QK$8&<}~#s<l$NGlr0C=XSI4<szFAiG)(a0*t{)b!ZA1Y7jJN{iW7d-~T+e+W6t
z1<-Lyu?gYxZSPI>Lq)5z3L`rM+e5mb1_Q|Vr~SLvO)y^PIG*X5ZKdH)QyUp%y#6x>
z0*}<?D8N0mk08*q{dqsNU<fj$0?f2Rc7`?|PX~Nohg!a>oSfzD=_@iqR|3mDsf!}R
zdS>WvwE5Czj_t`8kfi&o^grf3RC^MgG#d-#9yW-m0Y9{fnN9e(xlfv$Gn`MY*tCft
zs1Tyrgtu?s#$%x*4j~~TUR+<#DfF6tdB4aYxYwr8Lp9(X;hN)H@*aaU09ENVrhQ?O
z=UpXopiKi>dO~XK*}cin#eO93KYJom%&gR5E)T2_Nv)ef!t@~`5Hxiaq--wN>W-b9
ze#b3v?kcSX#=;uhyjws?HMPilQ74@2Ua)Zx@FuHRNzsFgxZt3rLaO0qi?GQG4nVfX
zk9~KB=bG+JQ&NyzhBAfjm@LvG1<W%`e4*Np`trkg99AXVjy5~1t+G^3-O1JTYQ1+#
zyAhyRl`1bc>}?TanEUv^>UT}()I6{sOe$k)w|uzTcF-^h!+o$WDm79;4{zVykz>>2
z&qG9%35v{<9hv^*MOY(HZ{XvR50x2SIUxtjhkRIUtZu`pO@pCkX4A9{LPbHsR0{<S
z^4V1ixWs~Z_W8du@HY2(uQ(r-pzwVSH_BOy<k)H2h!<<?N}&PQX=yr8Yw^a54#n9c
zz@G+fJcs;6W{uNWQM#=HM25|2JKvlZ0+W(V<oug%fD;u`b4U4iauXM|$;#I7<NNmk
zb=?5guoCH|i#?e<UazwzGv;))Yo}#BoEh~?DCH+~zAX|@b_gKv3t4iGYn+*IrN^Tv
zfPbLQ`)@74kK{sJ;9`4WP!hO6)rRzrjZMD@G2}gzxVV84h`g+^l$HJVJeEBF5lz}I
zsL$>U>TGu8Zl}c-zpl@AbotLj=OolH&2*%km~17oL_`CPG?c79`H;S-h%&G3N(i3P
zZx0zTab+=jH*HPxHfX8<$OLK$pO@IzMWafmRv-8<s{X{gkpJQ1>$cZ#Jkv2V`i9MA
zJD4~COF^N}NNX6=g9W?vRxQOgy3B3U(H%&DeS97h<pTo)KZJTL_+DSV?Dhqjhb8y(
zV9UW%YP3*^aduf_Fn)sE;X6^4c-qYxuiugozrM^Ph_BWv(aWLPH+DC!QrXUky4Ry;
z$%^(QIPeM&R}qqPp}`v&8zTlHMI>u&$qzT)k{zHsU1j-An0S0gwjyPPg$g`9UT^jU
zD=ig4S3t00t^}LJ%b5si%-%G<6E~B$Yg`>`L)+a4vhG(K8^L9LQl%-^NEZ6B4^IOK
z;0y^`?`&5Gl!c?hJe*V6Ehn`rdxS`NKG0U@PoeNFaWVgG<7u5CU7rV@K{R7!iz9dp
z=yut$wof9xge>52f{skd{O^Y>_uvnWzlC3yd2PGNReg1H75kAXIPtkk%R6sMvDt(u
z*eS2C(ox-#IjsXWHUxGq3K*JZfwO$1*#~_#pmX0PjEMjC&emqEA(y>yIAnlZd+OA)
zI^0T>0_}4pJTRg|{0kL+9pj6_Stb{Sw_CYgFEZ!sG}xstyMyFg)IKB+l=H2`)bgLP
z`c#XaC|n$wwN|u3Cvubi{ZN@8^ibBx32P+Dn)buni7vcqvgjY+AlBEHe<uGx4!nn*
z12^3$4E)p~ret57jaedj_~uzIG8c-~5P@I*vLPeO+U={SP>!mMjyW=EK#g24T|!C4
z{4l`Y{FsF#;2^@1p?UgCNwjD*+*Ge1oh%%*v^3d%Vi3dal@*WK<zIXyP+A4OcEr3v
z&+#y%q@B~#&HxG&dNdLwlZxqH_wuOU1>~sb_SkwV&JOo1zr!!T=m|Fr0gJ@)awdAX
z&$mYgW-zOzVy`UqmH2tyR;!U;hnDt<#+c0PmJ|gvHo7k^d<9?g>D|SFp+9_Pw9yn&
z1ye{!NF3OykF<28MGD0oMUf<MUr2BeZH~1gs@Br>?<>1RW?74{!UuaqD-U8B;A0F8
ze$qdA=I~q1O;QWK4r+}4GDOGOx!Fkg0Y0jq5P`>x<1+C|oW)a3$NA1Bs2=_WXicWZ
z3bSMf_*Jczo;)scu(Fa8S)#9oUPSf}pKyUuq=R>8Jt17jlU|GAxNPF#1yeOZeX)>O
zmR2|i{GumEgurL7WtCAsfjKk@JnBW~BPv@Ud@fh1@IBdtD@oJWP6H6X7)eZb@fJvv
zzL2~pi>`Y)A!h3C<?^Tz?+rWaim_O*2nQSQ{wePGOEVe-(#T9ru5Gl}d~i(H`C(`!
zeRF|jN;xnNpHHB4I$v|)LQ~8xV!3nbi}<)zQ_7ZkA|weS3^2)|D@FSW8VHCgjhwM%
zO&<Q*NdgZAIrtW?x3lv{eSJwPlzU@U;e~<_)3u8pW6eoGP`I3scNH&n|L{R`>JOup
z3^WX}^}BNp3aQ_z=Nt%uUnr!!oCd{WOgMa<EZA!qUS`j!$Z*50%EPYSUmVNFY_?Nf
z!{^|8+bYp_UhZ2o`3!a1<hWTfh|gxdReD@F|0YzC?%A`-5R8vQ$)8i7(N{c7MdZwz
zzO&?d`9ST$(AX~I-e$E}BF%pjxtNsR%+3Z2fg0vM{Jr{3Mg|&^N%M1TDhL1F^J%Ct
zlwp;sk+*VTT`#iU?Zyh}g=$J_8tE4AJkFFKnpw+`X)~|GrWmeNG8&sFY+2{dO;CP8
zuM&m}XgUJaBsg%L-gt3*m&s8x-zpm%eBoD#Wj18%j?m}@ti%$m)xsq6{I0`(FJ96*
z4DT~D)O9&%m7sAmsKKXU6~B@T>$R*jRgvX>Z_k+Ckny;_TH125^%<!gMyHv%Q_;5f
zX+C&EhK|i?)@%o=!$ijJYxqomzl^jrFtw%Yb-tywl79{^_q{u;2YKfp^A?B;)WRjo
zypq6d3LlL2?*}XufvHN#denArdxY2ZItr4=2I733am{TN(gPZ(_V&EC1ENaDO-aBJ
zeMtKPr4jYLmV=8JlCPyn|4Y;7wxn!DDj95!TOBM0*4JZYgs$deuE(poOK-p!G08(a
z>)_DhgItO?lxb>ahJ}qiSE3pFW4x#cXi&eSQ+4GQk|f&W32>fXfZ~eMRk*i)jeDAW
zj^{6ased<(iVyUNEZ0K_{O$M=e*W|YWU<!w_JW0q0Nk3k+<lrrvU4R#ri{!mw5ZjT
zKO0<j_Hlx8PYADHYyV!eU+c*L&pjQd#2=vI{B`(}LpL)4pWX88_R<>D?OQ~I&+Ww?
z7l=!HYo#T98?o#KWsqE28)1$Q5sopTMmLs}dJ+1CE)f{)iVnYmv1^YE6CL?9+*HJ^
zwR^kP94hBBo4m6oy!~LR{%W(gx0flBXmPIkya~B&#UvYalc@h9iHL|ep@M<<Ws`8u
zuCD6Ew?2!y@arm_{&qf#F{PYMFgRE!Hxxemz0$DlzgrL$I=U&<s-^vPuy-|tU3%eg
zJX@OjJQ9%RhJjl|6Dng+Pc)Y#0}C-YTXHLFKXHvAIJhkQ_KqwXhfV0>1v`fUrpgbo
zNFcP{k0AWwn~V%;k3<A^v4?F&kRjDP$!0i90BQL2dmiB9AiC;e%ZxQNB`L<IsN$<Q
zs-bHO#Aym4pa9cL&=(NKixE&@zLit14C-E9ik#sxc~`vkBD@r1fD1GdEHgKJ|IV=4
zx#d_06YksmG~f5t`<Kgb1P4ByYZ{=%Iy!`B>eXp5PAdPrPS>-`5$6&6AT>fW#L0Q)
z{&M7kYDpN}LY1uj-?lDh1Ph`l@PxCRRKc4Z2fTX<8uHbH2P=6*KzsMBj{`W(st_Lz
zKECfQhL|D{bhNl0rc_ruDJY;O4yn#bfrt;7R2;&^O3Zoj!_{`%>!ELBoKecto73Mm
zprM92qvSy9)dNr?i;*b`3kw6X{~!0;u3B;m+Im`1`T_qzE#Lvw4YmIFO?`19Ks6WY
z?e#gzSa1KH8fPi&Zf#vIoXN#!*cxmwQC6VzOMn$>E|2b>7e#i&Ky^1f3K~$uU4jJq
z{@wr&6C{`q&XK)L$;<8mtWl;m5Zl04RjrQdazE~cfzRD{?hQa<A_#cUK&kXmHCR8-
zFE1Zhmy#~*>ncO+{f+S@-Es>G{Qb!_&i#77ToDpK!z|zlye-8WQBoZ3?tXc+9JIu>
zz7z6m6VX3d(C1#`as|{#EP3tM3?%c|s+Yo}ps1TBR@rRyv@Zm;<Cal=b%q0wQ0{vg
zMwmdTqZP~X%+Jq5;2zfuT11s5?TTP}Ud)?Rl{~e=XU)4T9ydkLJ1+u=u-uiK%LSl#
ziHDO>4ta%7%7N2<ZCOuFaYK0$9EXGsS(b!(0bi}J*IIO!@Zm6lX>q>r9Y`^*0TZEQ
zyK0Um?O-&;bAT5RaDzhK2hprdfhC94P|Tv^QM2l-A8@kVByW7H&)$iEZGCv^nO&N<
zv0nlZZoiR)BsrM+Ye06Rt;A>4BC?`6`gr@Qii&wacSF0TfqKEmjQ)2FTwuDeeB`Ip
zR)SPdvPEKHph2t;J|)?b09Xi_mQzOTe~tl@ZLOtN>|8QiAV3)4>KxKixpUWGI*VOQ
z6uglni|*;|<>BR}<$;D~rWUX^w-bC(F&KgUlVeJOS^5_}CS_Y%9!M(Uv|qjd-TNA-
z)Bt@yVPSQbKF8PaylBga!KdYYsa6^M!7G#Z*T-URu?|b8EHE#gaxPo!q?CU=@=XMY
zV<fUw*)e(go|7~UR1oZLWK^-51#*O2@FSDM2URxY)3aK13oy5OAkbjQ@9H|&(n*=E
z&TggX&;eCcQok_C0gmZ`$gY`+)y#P3s+ep_?-0>qVQ@-B=r)Dz_gZpQ66BYiX>v?V
zcCTcDxtaao=u5vF{-FtvDmjGEkPyHmeKv(Wz$~|VsxPgR1VptdeM|_g5H`>on)$Z$
zQ}>=0?{Dk9J9cEtcx1HdohxCzP^4FSEZ<Y#3PAfT=)S3H@)$0-LzpAYIaGJ%{KSX>
zWZZe@27$<DG&D9^@)3YFU%iNuAfQp*dNC_Xx4Z;jb8P|F=WZ~5hZWuUp&_rzWB^U<
zTF~<wJxp+!)l`1Cn$ceYYI_$oM#w8|ZP|Jy?I~v!cq~SpUEfps;pEGyFlrzaUqpr_
z4Y-IO6w8UpExxnEy@Q=!OJXxzajGVIJqLTEAsH41Z*1IGfBi^RTF*I`*6I%3X_<`C
zp+Xl1h&|h2f3ciF18b-Mn)y|s!o({rYKBot^T-6b?B9A$RB>AU<Sw%6dP7P>6q`m>
z@Iw)6yL~~+nvVbi==cANg(nWo@Yw&7`ANqyB`y1Yggc9qH~RUB6V(FyYW^}UR)myX
ztda_L#Hn~2QXw%W6Ikt{!~ys8Mm``LK!HY~XB=k74i!1*w{dpw#l%*?{UD3B;6nv3
zLE_u$zL{OTRWdFIZ&bG8Q7$A?Z`_sT0|47!xzk30*|<v%8RQvx>I;3q{PDnPv+sS?
zHJ<ej?E5u9)B>mK7TR@<LS8+n$*3Cbd@x73<8)!RbwyhXNoD++D0$aUy*0FX1XI#2
z=apm6u$iy?3aWGa2aZM!2Bb{C3Fpll7K?EbTiC#7`dRg_jlIQua=k)9z1xDD5bg2A
z(Tw(J>)<QCvGr<mx1l9wC;4IY#R}-!F-B{=%RS{*og}S@Hk~y&ZnCFENAj((=}>07
zanGcB?Ypas{pOmUL{?MQr+`hH<mLEY{_O*UqPahIsKUE5^N8S`<Er$1R(#;(d%WsL
z>;DlhBrofH3Ox`W1=KIINAtg^xX~}Bm?XUAMs=`1c?I1(<`Lp@<$GOC%j*!{pFbQ1
zyh}<<jy~;u&pkOUB|4$AR68T7;vq$%{)!@(LOQn&@mA3sRz^U!mz!X;=ho$Exq*nj
zbg`$;xj%0JtXHpp9DMbcUt~fco*za{DMJ(lflZK{t*YikI9eeVPDR;)bEmM!KbDH4
z%D2f5H8{AFzt8(k?2&`T)LMKcC<jJI^*}t+$ql-l%rPjQ1Emp`M>=^J9PBKVRVp^p
zIfRZ?ZU(!5T8qH?1N&ZLXJ?1YzAdU#K|-Dj-1fC#+P7#aTG}iS2%Iq=pE6qJW^aPw
zhCGY*UykA-7S6}^=1b}Cla#9Nwe#Qw#?~w3E{}QMYgO5;@Q$QQ$|<5NO&-4%3!?Zq
z9-_bIcX|3YQ1saj1QZ85TU(Zlkjo+<lK@p;%+Gr-1L+}NW0;*@cs!IJ_wDS14YVZD
z(TcQO2zTve{f)7<)=$BQ!Sc<w>Dj-rPkq(H7}EHq(6Cl-X5;pbU3_cjmzQ<i3Ys9l
zc7Cp5+|(`GVj%PutCs^Fn2?Z=&J_#6z)Z{Rz2O^w5?8)6?Fu9zZ&xsF1#_Dj7{7iA
z+lsJeZ)~hwS^d053u^du%IG$^T(_;oWTY}~@ZvzaTa8oy-Ux+zcQHvayftg}uH5L0
ziY9Qq_zcY_kmz$dEibdEwY!|@X9eHiKV!6Y^gDfmxhPSnd?*@QglNS1woe4OwNWHm
z4^@k|XU&<k-43p{U_eMCGxG_;7}nXD+4RR$J~un)(JC79JA-F%V>DuJ_3zWsB1nBC
zkRJ2<7z=?Q0&Ol?-#1+4ti&$>ir^(phdK0KT-TLF;GSN8ZH%TNKC8w@T=(_cD5qng
z`cwn+t$RYF<;ZOR)2Rn6qgVY{2J1wl%}bRJ+;C3ivo#{5%}tMbby_7Shfc}Kqvcp&
zl350-%2#clCW>UnSU_ccV<*r&l(AbOtaw6o+y<Ugy?#Z2ywfaK$-;L%mc|d!FzkQL
z%=K`f3FH5w_o=RmhK6~2;{*nvX_%GtI`~=;hy(sCGz3EK<8S5#fm9SNfo$)NuF}9>
z_3x7cw8)C%`*KU|K5Gmt#Y5RQTvWp-Y=5fB8tTf%x(ry8fdE&v>az+h^lbYj7X9pX
znY0T<yMv|~GUSz}=C}2fc>E6gM^Frr@iHMnmom?r+Wx6*gZvrp?L#TWDi+t$WKbhP
zZ9v3<EA=Mja(k5yEZ^b#1r4h7h?3ILG(R$iKg`h)HUWH>hMpcMkfx*Su%0*0|MNe6
z8J4#AOqaaac3za8pa?c4Nh`-(zy0zm&4cCIqbdP1_08~}KSOE;A?t2)RXBF`q@h$H
z8uZLWk3}t^{7oN-F9Z5x-kATa##u&Pq-~Y-yJc?kEmyJzHG0kzNp+`*-fumo+I8VU
zu@9iJ>MTyhilfVAVx)s9@)Kw<!jX_J><l@<g-Oz>q~$4LBzD*dl(8ftLpnC=k9B<4
z4=H%&i*3#5)62Hunb=2aFvsEtpwN`7O{!Apx1&?!2n1|S#A0$6*A8j@N#MME+k(V%
z)`eH*q9~+LUp;Qeg;7Z1g{!5n5h#^o%}f%gniqb$tGo@ev8d(Gtc)kS+3u^2#D|$v
z-m#DLIQOp_3c^Ft5YJUW43Upuz|@}Hqqd%@G2Gic<6cv`TAy>p;4M1ff?dmSjMr7J
zB@B${S|gx587b_5EiiHk2Z_a)^Tm_TBgyDDa_wExtnS6W&8U_jU1~yvKbHBULIOsW
z8sQRbtgJE8dHbbyvC7jwHZOt!p%ugbyJzQR-f9vZQn>ik)K9t>VZAW}CVwB}!m?@T
z@gDH-!o?MDZ@EOK2HPH0ceo08skJk1UAx{m`RI#=AIY6!uO)xFl|P~PRJh%|_x|%)
zuWrU7iH4iuJT0>%{YQFm*5`j`n}Cafdi02{=?3<)d46&h()Cz-&fK@<D%9IkY7s~a
zjd$m-;r)=jQaStZbY1}n39x-_=?<SFG&kY9US;3Sca9JgZ(-j7C|J#88`44WJ-$7;
zs8kXY0Tfoh2hs{qp$&?$*>Oq0_ayg^N(*)NKcp>Zf4*+3O@aL_rS1#no@M7CO!Z4b
za5&(*9XLltp7S&t2MZcIb`~-yu>e9sZ;;p!Q4{|TAg3WpUYGXF@egw?Ye$l*pA>9W
z81PD0R;;_fecxzuaonl2gb3qs?FtMh>i1f6qftzD^r*0i-QZNJFZtwj&Y8Q4i+2vr
zKp+I8fDHptx0%tXE|IcS_(0vp^PP|eqZMeu>TL1l;K3gwPE1H}z1UUcOeI7`%^V!m
zj~P<F&yW|uJy6pdp*7rQbN#uKPB`cd-UNTy{})%|h<~{k%HQavASzv^qu1|O+dmN@
z<+D>lQ25G(HFJuTe5I>Pl#OLSn_B*R+QvfHi7@1^_0o1>bTo*kSwuzv6Bo*cE&Dy6
z90=<3^4g1ev}4S3yjlGf4D1+Be{~~_DosWi^^bfUpTe;#s~DW|KNotNeLKClIjVm*
zm*!5#Zp~w8sepfWbp@*8HlT}VTk_2Xe8_%XBWbeCXQ#miPk+9qPGsOAu%DTZAKk0B
z`62<$hKA|JX36R0qincX+|Ts;%#!nnA*DG<^-xnV593nV$SUo<QFkV><7TSQQ*J(a
zwK-A>2Rc1{^Y}nVsW!)(Tstx;w9O~iEL9qZPk^DJ?fW%+8VMi#qJ#e!<wIpEM7U=l
zGW9KesVVu86!$ICIp2ksG;t2D?PKBj*b!u^_?&oF-`SvXAjFUyg|dz)3J>=sg5_wd
z{}|LUK@tH+&QgZk-3aY*C$3^A?BN_iQhaN}Va3Ij4Bfq>%7`+BVk)e&KHl;E%D6Nh
zF}vETuSCyaM1pD!JV^L~f4dhMKnA{9n<n;SCMv%`A4BcX$vGV(BPYj$*Mf&Gpqr|_
zZk{WD!1)67ZQUD$0?dADRlEMN*84RL#>=<vz@jpPNezZ2I>+ya{Q)_^yg(xmZf}GL
z`FIoPr9KxqqK#=LtsazP5zg!8VgT6^$Xyvd6hWU{ZKJUuf7Dp+v{@{zMhe$e>eji{
zFOHKZ#t`x(v8-NBHvK)Gliqzw{E6oo4-YHm8pjs$U`dmQ@;f=@Z{KS+S{k9f?u&ZV
zPD>_K)Ob(*H9f)=%#rrTCz$MbAFivBAW0x2V(+N6vvzhSC%6$78B~NENJNOoY=Y1|
z>h@0}ITJ<lTR^^1nC>v26!}C2;iT8S-Tteu)R?^#XGt#pE$W9C0Zi-*lGb|X$-*GM
z$E4p5zPEk_8G_n+b2sutxyeaioz5|Sk&Fys-v4|L!JJciW443W?0F*Sb#6Lm`&-SC
zN#cbn%ldqV{$F!BB1lP1t)04nmR*|M*sD3PKOXW4^0-{Uq9a2{;IhH_&FU*6X~1|p
z7QF_^!=?JnT@O0F_-tvcTPjTlB1E!`fp2enApO+xnut>7sEs&t-!i#WqIQYhd@yGF
zVu#71mX`JR)8``V!WujCDx=xSz9C;L8`6j;^M|v7?^#GbziOKH?LtoGE@UA|76XKx
z98<2`@ww&cBZb$;9le;3jmw~}1RPGlc;SWE=ceP1D;S21)0&uiVPRkg5uQ!{cyC~$
zQdhW?JCMx8I&vthc0Xv8ES1QRi3^Fk|4e6Ar8!HaFDQxK*A$Mlx7TAjC!M0(QT|u>
zWtLUJEBD#f!y$Fzw)3_qOXc#BpYG>S`-SIAzP(sQH+Z1N0~M~#Z6a8sGFMwi>9fvi
zXAjMCwEW%oE!WVnFyLgpFtViMOl8185c1wMMv?&Osvu8KD}9KchaloT*;|Iv2nO>u
zzy<x*qSfy<tCotPXJnjx#+@moz?n8(NR1tDWLd{Dxfk2n>XcLSKEypt&OG*2@oDZ_
z{wGITk1S=5b%mc4z=foZ_9}cdxA?Ls_cDN7o(fbl;4nou;a8AZ@!}PFO)${X;tBcO
zQzk>5oy#%k{My)1&WfTNON58Lh5H{81#l_j*8%TzRhmnb7O4B0mvC_LQZ(jIhQkpN
zd9e6XQy3%;zhPK%w+)giD&)jW{iX!^6r!UZb%7;_AISRG+aJfpqW=<a4FHhLMiM}1
zkv`-t#>f9^nsc$^-q6X;<ZjR8-szq7<3x{K(-=a@sOHi<JzR0Yhh`>cIF}0<_anLD
zIlEbN3PyxfRabW=pg=lI^srUeE)$<$j8cVQ{b##dPUc|+GBONAIE4deaq92I-e%v5
z0Odz1#}Swx?$*_g0KbTVXcO9`R<wJ%HvtEF*Y?`_)@yDr2d|Bwl&`4In5r9XghNvO
zg(0DC%jxf-fSS1ZI|-~3&;4~v2lr!!*jvUFt0q03QMgbw(=AAcD1NaWK`a767I18G
zi6%1r9GG9Jx&he_GlZFapiUF;WAvn?B1@m2hw(ub-Ih|l@eo2j(K*?@C#pn%$Z?)e
z|BhIIMZ+#3Hp+oLXKmXCVpB(HUTEg&W#^bJw~|+P$v{aA$Wr<6&#EcL{ShDd-SM<M
zIxmv@&1KKe&!=30%p;(PD!!lEV)lb)B#!lw@%Ge0U|i}F&csUJL-xQ=b_N+L{O@s8
zN($d?x>a4*>d|cCKLR67QZ+;pt)=DK_JI%`8BAn&*;ZYOw>GNkPY&>@7FA_!{e9oH
zbve`Z=X#~>CAfMpIKYFcO6y`Ue$9QBd^;}m4jp(HzxEnfz;&U$^gk}s2ViP|O78D_
z`H8LY^)h^W8~g*=oW^egkJlaMS)QwV){QjK%mN%6l?o=ZF|cWb^n0omBF&+o1qt!o
zH_$fKu(t{X!dN{69iplUq17ty8*hTHt`1N(AhJFme572<%H+R3Xk{cLEx7l;enrt?
z=DE~FpMT~L#3T%QEuZfHo&pdJ@2lbvzkmR`-1i2a^_{r(ry||vx_{Hqg(`+v>!(vP
zP$|+M8CO|*eS-e`ccQO;#GU;CiEsJSIC(~WV4Q%~)1OY0&L?|K7TU5rquIJ<)5s{K
z1b&h4m)&2@Q?Vp&AJ(&)tqZ~C@C}b6G&U{Dx?3N(Pyo6tDCi=mfRU1n1UYwC)8+14
z<5vMT+V=%^<Fz{>Bfe?B8~~%sfm6?kgNr*;s;$1H2p$Y2aoYF@T(<4BLLRt4;}Uw&
zckkL8hr(A1+`Dh5vxZjhYlT|He3y3mN5!zTYw?(WE+GTak+cq($qNNBMROEAxjHKv
zH9rQ3>iu?CG>W#z+p+*+o(HP>fq4&lHwY@Vj*!>3yX8oPfAieV*Y6okWXxlp_YFOH
ze=77Y+ZWDsdbMeDr;=XYdF-r8MYS#9R#CT^YP|o!q+P39Yd8=?`Nw*mXn)Gq*Iz5+
zQQcjc_Ip}N76iw_jJJ1xX-WwL%z}1A>?e@AHy4M#z=iffJRCs!55d?sefw)q3#}o1
z*!<XiAN=-2_<~@xyw6nrxWR87F}k-Hj-lg95+5#Jh&p%=CQX&(3o=fG<sW@u(*6-Z
z>*c)b{Egyce?_G)%YNhL!P>`nP$b*}d+iKtrKxw*E;h4ya%a&~ER!v4u|dZ)g@>Ju
zyr#Rv$0l6`&D4Q9oR?O#i>K#<{cCy`!o6F2&?r~Bthamh{+~^S=+H)w+92f3+`GH9
z`k<wKw9E_g3%61%8|84$Y3r<0s!+^P%&D%^qfL5BNixR9x}!-=11>**^m_oue`^6w
zHb2_B5x|dAg}jgX{@u>nddS3QLmUSzMj3i8>KT#rlw;zYL>0~o8&tK&tDcYaJVO1s
zC$XV$v`-o5D!&q<2}?0Y$&nde{knKOkDBUd*dt&gf2S;?Oa+lVlAv%)Z?1I`NKx<G
z%Ti~BW%+5=ZGLsi8fiVGn<E<ETaS*sP)#VZWGBeywstaFn?gKDmv=gs!M-9SYK|TO
zIwps8?v}p7xIB5_&*huzL!y+XcXMk@LgK~@4qR%~y|o6x8kIbhs83IyvY5R9C^B|j
z4kaoO4ehpw_(SpK?~U?OZIG!__Kb~7RueVN-2b%?dCv-(FYn6i6vz&!uoyvl^M>-b
za-9i8?Q7=+jE|yNGZr294dIW<=(lpj(~X|ho*_^5m?Cix#BfxxB7Xd48&TypH^;M(
zS5is{(9o|^Z6&?DF0{dUBWf0Dv991f3Odgr?<a4XC6e98_8QBgW&!bFMBJ`*hsQRj
z0;#z+;4yyXcZ0j&CLKWkW&Y5@{8C^_X@9m9pNOceQZl*-7^Prei*}^F4h#f!x<E^v
z_)jD{;%MZfL^oiPcE52AbjLsd=HcOiR;07B&?aEJBil50*F&cD2>)62rJK*5f5lyh
z8E)?r1G;=t?V&XCYaRvb%-{2$C2yEQ?`xjcVjv=e&9V{@0}t0q-J2!GF_!HeS2*5B
zOPST%w!|FKuFmTNwoLZ|*O!uIH|^ng&k&YwNcl{A{*fFFhz_v@kzot%C_C6kgCnqh
zGhV5ELG!kvOrQakndJq9nKMC>;vXDKi|^lQu~c_>ePB54dRES+B!EAOyV)Yd+Qqaw
ze4<xjKAP-SIBe#BSB0A&;eY3lB-eUn7sCB%uXg5Vxb_scI0K(=lPSWC`$0fry^<B;
zzH+dkqUVra^*djIK}$l{iUNFH6}0h;fmb9Adj6goK77O^G@-ZK;t&Ndp=sZqzTgET
z63YNlG7hWi?Y_cFju}@OO5h^X4TggJ8*fZlaUuL@p|?HMGBj49m~NzF(&6S4-XH@V
zXhML;3}|`pxmANl%w7j($a_tCgN2#A8%4DF%3))%?!^P}Bi?Osf<On59%@>>0Hkgp
z|FyV7llyu8xmgU@w~81rp~VQTip0D{OdSC(j)n$}rO&%$F%-$NW`JY|{&^>-^W()l
zqq+CMYld*3YN^t~K-O~M*>IThGpTDJmcnp7-Tp)J<EpayYi_~#x3S^Ua<K|3_{KTz
zv%C-bqKBaQY>H*WU=ky9s$M?%cKg>0#`@oXdM$K;rMYbG$0?|argHLOsP`i$C+~LK
zW@5I5feQp}n}VQIay9{2JB}18PZR%Ffd7(-*C-V4dt3@>{EP{!8~6NVk>=AG=x?-`
z#?jH~9KiMR0v!>6{Pa-)XAWSpeXMny06#?UkPVI@cD43A+@QR<Ev6d~<+%YxLsOo$
z`Gg1ymxE2wm+9&2_2oM5jBG5kmw((GtooD<3t6O&Tx!_z_Usx%g1EW4JDTe=6(C|F
z&T-Y%)pg{PC4lS^p0)<98c?h4^|iG#|E7k=oc*&4mx4q6_Tz-48Pkb-eDv+$<V(#r
zI@$$H>pZLr0$E>Q&Hpyg?)S?(bg1#`->}8yHMu_^foPyJ{{;QSOK&WnLo5^AIWvg)
z(ft+$EVDu6!@n=x1+i+FIL8Ncv7Cjx-`13rz@K8p?}(XMAeIyV^8e5Xl?qQ-#OGSA
z7yRFRvt-W4LVuq0poO4WLeL@d_hcR9&`Qb4&AQTqNux%*H{hL@vF5n$afSO~a)(9h
zt|YVV2GgCAvv;jV?(%cK5WR-O{ojv)qRLNi0gK1Cz8rWi(oS@=z(qqN`Of^A{Jr~c
z7;2^Te_$vfa?U*o^&&AX3~*Nev_6SdK8yanb-BXOKWOyqWc}UlJgx<M#fE;)*O(YI
zr1WS0Awv11yN8p#klI=lQfUG#dz})++H%9y$L?Pq`fH%J<}ot;8XBX!a4h%DhY@&4
zRJ%Z%O~1bpf*<WDn)~Pq&W|Or0@W2zrC_jalR4EnvA2FEhQ})de1z+Nr5x-bH#a_U
z%fJWr4xCO&b2n7d^OVf{q8Tot#tm(a9VPX;%VLoOFWJe3m2_lr@Yw``=*P$T_velI
zz=vVt5`CW|@0_eMc1X&SUz94y66}1iZGPR{WLTG_3#u|uiD7G(N$93>OH3PJYHXKW
zO?8%M#wmk>f~wU^#>X|E8G6BZY=fMI+Yf8oTn|q~==$)w19~+jBj&ilWCC)(rl%vs
zqg?j!R;Azg)7@=2P1c*+C10iu+>LjW)6vP$#>Y@t8f0!$%1OrLs<yA<7Xi(Wc+SXo
z0Y31}cXEd&{O}E{Oe?SP8c4ALof$qg-Z>LN0(U=hgBL^b0<xCo_~w=3)xab;0>KHX
znPw2?hG6O!{x)j-ObS5_3Zggs2oHULas#dEup=rrway{<xH^)~$HmAv3VKU>LeQ<g
z`H{elcw3i40X5I?@Guz}Sq<UWQPStKPp~0a@uBAW!hPwmyOmbiiHz5Ypp1m1NO8lF
zlWSSNg(&2aRc5`pGnP=Iy&Kgnx^X^#(uzBpugp0hPSI{b@`kwsl8^t;)6xdS$biO%
zy#j91kO2S9<|L~20i|h<@Nm`Xg9n!4JVzn{Q}*uzYIXKo1H;20Pmvi|;Njpp%Hn3!
zfL0IaoHreBfRYYWbv!faaH^Uq4VCY4IEa5Gx186j{wdSEI6ROXTq8(&=zj<Q)hL^S
zv5c3cNKHdySJ37!E87q>`z&y20k=~fc~JuxQNMou3Y?<y#JM#40Dl8jOc)`~?e9`&
zK`<YSabEy4Qldk{r7xbEipoDC{Kw#A4;(CHD(gJ%CSB-J!*kcZ_60->mTBhaUfUYr
zJ37l5*ll4TT*l97*1$dV<1mx<H}Fe6{^1g~)}ODoAN%<Rc<h)wpG7#qNkIm7s~&%5
zf&J1*Or|kAJKK)8S$Ied0+9-Nj`d!uFMqaWWEJi4iE~rnjxiAkd=x6>=F6!#U}Rh=
z_4YJxrRU#1{S*~<u^t?U@jxi^96U#_@k+O;^F9h_01^a}($l?ZxQL-7M2_3)y%u#^
zJ}EYf?9)FR79Q%?Jse|zpkM7-`dK79^Cn`_ytwH$SZ4lxtjnbU=hgCl)!uy*qgvVF
zq7^%D`02?BT+w=it1BuE1|7hH*7An;oO!_!#{i<V*RN>~@{)UgqiDy4YLx6ZXq3_Z
z^?E#WefBOuVi2#+OZD5^{^>tyWqN0f#C#(!_&0owZ^Jqc86tz-+x2Jm%#H4ZI%}sn
zTe9LcX-ye~FFaiTf;gQe*s(7#Nb4>pJ=uZ4mu4CXDpkwELSL|*GeS24D&yX{?mOeK
z(lZHGWg-O{PC#uND7EW0@LJ_BV;$IX1aXE~FGrOWq09?uK3A*6b!gIZ%jA8tA}5HW
z;aVGC{r;f^0)bGqQ?+jzmO)J?I#>j{mOEL2Jp4Dlzdza7loS{LDG0;^<K4|$9!(O|
z>0))D%A=GR4~2e>mA1QxuA{@G&u!l5eFwfvwK#|Xv;LEd4j{}gq1J_9PO@7dH}R*G
zk1~H?{=Jaj8|~x!Y4>5K$%+}YZY?*QCPo6aw2BWiEpK!T3}YDtQtGXLu$cc$yTe(g
zg^K0;Z^w`Fh>Iy<4~wrdNS_`NX%I%myjq^UU_TImXy)BU#gs+fL?9mWXW<HvoevU_
z>OsI|R{(rd^*H1|?f_|ne`W#T6EIad>L&HVue^2{sT9DGFqd76*YTV#!hB~{?B2?u
zV(Y-2Mk$kE!G{1|K3<~NKA}oN26nWW+AK~-qmpLU+~cXMwi?-Vw`croU5S4CYLPZd
zaoGy+T<0EYfGZ%?Ezs%`!4wxyM*O~WOyK@txe<--iMT&U!$4bh8&Lwna}9c%xixCB
z)XH$1?7UrwXrIsQDp}yS9IP6#^m#sAj8!-x6M9>QV3QHypW7B6%Xa<!`w8gGA%5a7
zB?>@Mg{=g5b8vg?KsnCbvSV`)?S7VU$wU|AXm5)^M^H&NuBndjK$}rt&!xlty727@
z?<5nGL?U%2nJNZ0HYj@lXvmaD>o;QX8myS+g>dLZ;?YDS1RQEqMQ+t9dTm~~L_~I0
zR=hx`|DY8rhHX~3i;<=9cS>Gp#cBnn=XaT6IncOK#7ox_#xZtj(*6@|I3ys9D!tA5
zO^+$IIcqrwRbFNrj;<^SYEd+5t>Wu)Xfi~zP{Vq`KC+1XUT7IOZ`E@0$1B+&Y4RUH
zxbMf0A17QDfJ<fON0P(%|1O<XUu&l2{R8^a$h$cG;0rP+-GF*3qb&Q&psA4&6!Z<T
z&(E>Dns@x*Hn=FdXxSK6Cd<nc3w}t-P^K!!kw|wz5mN-e+L6Mn(Q;r?$_ujO=3L)r
z0CRpS_s4A2$nRjH0P#%YuG|pFNGcbxoC0bSA{JB9QcYz}PR@V6&62%ukYmldcew73
zB^FY<4WmP?R)XJgN#!Xqa|TuQrF-3#46E2mTr`4=Ho`yS(92fDK7^Jvj$4EAbU?n1
z`l=FSt7;}EBf@kqEa3hT2Xr{PN-MRJmr#wht@dlixvt5<80!hLziLFg%Xqn82TYKw
zssq2Wot*I;EKRj~x;?)(fUO-s3AsYK4``97>%KvW?B8%~{HG%;q6g}^^}Mxi+qw6B
zHV4Z-swRuY=`R?x-FXdO1T$&@qcSl^`nI|G@{_kQKq+bnM|beJ>=56W3sZcVqSmgl
z$oltHR7IPyD^&O|-Fn%wZVMZp_~qG-3(pzZ&z$5znaa$;E(2yzl<?z_rp*elw`u8o
z8Fo+A?&sTJ+Mjmi22V9@h!yTj!=srLP8t*SxwZcYRJ8wIH4YEs*cWCV_ZcwvMG}`E
z(txK~l@rrG>>q8~hORxlUuDsDxOZb{>8<vyRM%TO^_l+MZT{gUAF#K(RD5(3>$^JW
zM8s{cQjY8bMqI@iz(F0XwP%Ru09Sb#5HSBcI7JaDe+=+PTTYoP8i*QkQ+@?~%tXIG
z`QxeV+HpiYsD|$CTg%Pw-cQ4tpLllFpE#@mh}PbU0-m1>Oo08tjoR{SnNK}<do#d6
z;Bf1U4u7>W1v6QG6U$Lm17S(*cAyb;d{1HQ0Ys_+ec9cW9k^-*cNP5g%=dx4Fm-iQ
z#7EiVB?q5i*wZkz^V~mu>S%3Npzv$q|8&y=B&&-N&*LI^lL-iO<%v^8Ebqe2ku9w#
zL+;G~^yQkpuCmi<mDZZRv@;1Vc2N-lAr$7=qkyuaOP9-NF{=gb+yCb4+<ujVZpD~1
z^VRIn^tWdd77&OYTjIfavnzxd@0;Z}cnQ^+Y_Zn|BS2~*$)IxBJ@Ik%F#-=c4^5}Z
z*>bn(QaFlD%M&Cl{F5pHIZEJQAWm2Pn^bwRf!s=1YPaKY&+SgGyc9$kbPBYAagIN~
z>T4@JR&LW;K;WG<A25Q5{%BHgalifF*UAN}%ErdVj-8j`Fc5_TJ0qj)e{VaW=rve<
zF#gxIQZ9deetLLzw$4`S4ZkOpREWbi(}4+i|8S~!Gib<Pl!t2n4tD1F2Bkag?#jrt
zpq|br*%$kkj?~*Mu+ovc+_!4asg*f6I!^w35*1vsjCL4<uG7K=w(VO?ABIA+mbG7Z
zAxM<KInTC>6X-B<adT_@L*PVbkVO=E6+@$AV{`g)J~rxn7AA?^TlkPM*lAWBD4JE5
zGp_hw87BL>QqHD;&_@V_5>neTs$QUw0=ipkK2+WJF%+=_C8NZ2c;RMtc4db&HS$m5
zZ=d&vTsP+RXGvyM-`?Q9HCPR;`0+$g>9N8(4St*^slxboDjYZK#skYbjK2tKz_XB?
z9WOka=3XNnWD=vl`kYbnZ~T1(8m7|1(eel7rWl7G*H_<xj!lC;VQ9;W-z;7v#>Ay9
zxro5O6NI7x+WGmc3~bt8zz}WXjU*L=AqHsVVDTIfLk+wq-ryB1vRzdr_6=Q-#1#LI
z(gYvNH~#Ro<U?Hvw3ll<0gQ9|mAoMmEO`F`?8yHBVmM2Jf9hb=Np4y0e;4*K$>CX;
zCAsn67iic8%#3iP;oSj%9R!eaF#P{dFA!072tl2Ei55b^&(9B<-T)E~_&q@RfREP~
zc+j)#wHqAP50<|W`Vhh#zHSG_?*Aa)2Y>Jm*6P?;5t>|KJw0)Fwt=qeKY}+Da53K6
z)XM0n7*LQF?~TpP+<!vACj)TR|4(&S9?jO;?l+3lo>Xxt)#!kl=W5MkHPn<+>L{g#
zQc8$6YSziDbRbBns-kI|Ac&e<p;|+o8k$ofD%2D$5yb3!_51Ex_gi<}weCN6t$Y8-
zTFFjU_WSPrJiq6Ce*1ZT@5o8eZtidS;Y8t!X2b2tCFZZy`1D~g@ij7#1OpX&jytw<
z{}9Dh5nQ7Ga!sb<ehf!lcl_S?o;6a>d6MrFDlIlHVYUmVZMsY03r}Y1(-$IXnblSM
z-#NUp7?-yiIM+t(`>w0Pn>rZz?oONgb9a;e+dpnerI>QD5VcA&e4$5o%lskx(QVl%
zt~GMM);r?&20x^4Q$h3JE}jz~Uvq}pUF$#duUL$NmENE|2xm9LL;z9zR5G3ig8nCm
z>-Hql9GqjTGzE3puRiw;SY-6?x@e2s13`>OSYzbe$=PtmWZd#zOXCY)ZqPTW%k1Q&
zH@EgZxwstp#)I=m6f&dPZ}e}s;xRlIWCHKC*BM&tZQdnIuLyToYjQHCR|Q8ldt#HK
z7z8RZ&ErkD*?DG#;rDbUGce*KE>rSwJ6osc?UxT&rO|nFmrisOeH77ljwkqLw%7gU
z_{?&%>5F#?gGTB=xB)%{iiucb?d|Hf96HEBoxP;kzOOFWEF5kIlcIB39Bhxy);pa?
zQtm{<OPuPJI~}YvDhG=*trk3sL=Kmq)*^&dkr6T2B>7Ii>L)1Ck0NLap>*h0>d#u)
zW=6(7BM&`aR*h!X2e1vBqLLC&kphqL;WR<)<yuv-)GCUWE&ZInm$(~Z_$Wo)rtdnu
zoq2@yK^;~sa{SdrD+S;E=E%{rE?*JeZ$;6MCGH%kJz!;Se%Xhks5ldl^0jEb;e869
zGx&R8jI=FeP*4!;kMW5KfCQE@3j-s#=#AJob@4?ogN}}lJ@IBLDk|>$$rj`w{d7uI
zK=#O-HmnGb|M<^y(#Ve@F65Sgx>N>c&)oEteBXEFE#CUed_@EL#&!o!>YO&)J0M4C
z<ijXM-3UGs8Nf-OX=<Cazrh7r9=bs1#N6?VtKmc;%fj*Kj)pU}ijmPYcAU}pLDulW
z$V{WnRHvwv;HKCYgs}*4W1}6=p@pZqY8)jF9&}j`5_46Ye2eYSEiVog2*l^{xGI+T
zPmGU;=eI=71_Dd`ArFTj6a78>5cF4i3l{`c6@ixN|I>#n8wBRJ#{v0=G+G@8HU=uJ
zb<qs-vM~s{@AmL={MaY`IXCySRlZ-ADhphyf%^NuzIqk{)Jl!>12R()mMCIo*C6R~
zDitURTdJP1*NHYNsM6}3mKA!CVbt*L_U{dLyC9R$Gbt@cr5?vFdt{gWDVr*$Qr@3-
z;CMqikY5P3Co&`RP0gRz*6Jo?U~xE!J@N4N<WX-~SV51NKw_hvP-!479Z~f8O8<@>
zXxUORx#xe7l&uxox&pEtI<&eiWf&Miwc_I92+4O*tkCQZP!i+4as{-hc6WDcz(;`E
ztQ$EHq<iw<kb*+;041_zj>CV)wCr4~LZ8;D!n4-asZ>+OgCJV^n9M2-VX+b0>?vMh
zm~jk7Q4gfqP558N$;O{zl!mk(r()FVV{UD%3s3ickHe~|C@aH(el4{Sf~IU9ewRou
zD<Tr@>Nxis3^Z(W)zP5K%w3Kfditm^B)Ynx!2tg<b1gC%@x5CGR4UBt2|-Xk%vI6D
zR-yVwg$3BSS^~!B*Oi0*cq}_S|2OEST|x^Bsi3MFJCW<xU<b_7uMG=;>i>(0{bwJr
z?xED}O&p2v^RO<ne!q7P700T^Qe54KT|`ASNQpMJp?{|Lviku>{=DKc8j?=%VAYfI
zo{V;97_6e+2CaKlg;HG${9y8`s!dPjjZe4taYHwCioN40_~<#xtkTcM>(7VT^W>%V
z7H<SiiqSu?%I7v%b3{p6O)e`Rc<009^p?>t)Y?@_j1<D1nwId$)H(1N%G-9Jkf25+
zV5JK#XAf~U?Om+Qfl{2ki%ag4Cr#ljhPdomtJTd|D#m*yXE%kU+^4l;lggc%Fv=u5
zLQ>|?frO({>0{yGU2X|Jv!knPQP0u<05Rdn8gX=`IW7hr>G0KC%t%wC%=%?NYUSe?
zZ3bc2=`3|BQ9uO{v}exjPNZl=N7BgB4}SA|AEr?3wH_cR+^1FaS-uPjKs^+m=s%#k
zgSpaH3Z0~ft8MmH#m_^I-NLpMhO*P@>XH`EEn(i}H39T()y2&Id99#9^_eEA^7+Z$
zWddlW2w1oHg}p1ATW94(lFhw#W}=>k#qQs60UDVyst!&RNDKRQB8qyBi=+mOf*JOS
zMcee*Adwusc+PbZfueS8RRIIenniBJ-0XDB*8y@==r(y_Hdd4wS6w|CH5OBUlGh@3
z+u)}?7jz`zN^NA{6sG$}Wu2rxymppcD^{9pe*0dj^l*tvU62p6lwX3s$Mrf!jVQqj
z9Z7znQ%lk$yDEb9G@!CxZ<0Iz!%6X=Y4NOlX4lo`Ul_cy8B0+I8>lr12xq`3<w5$n
zrCI(s@PO03bpDXi-W?pay@3z!!VCkUHBh`ruBuvzZUXEF7W*%Cy>Q9B&`qDroOo}L
zyVjs+sw&;RgnGEaks{Rxg<?kR$CFPo5FirHV)TIB*m<GwV51(>3A(bVsODAjE^f^w
ztLXt(6s`ue7L^roLHE;yD?@-<?}yUFV}=)BDxOyG4-^#A4S5Q>T$vzcktu3zux>-=
z0cnZm3tJIXZ&A5A&eR2`2*$=%<=P^kB_e>HEv(?hafWKvwQv9S68-7m|0stVGCny8
z1h{d>r6C9{rBUMurs-DrzZhe7!UJ!g764MfB_IJp{s53%EK+7;F~el4bTXFwvpvOy
znRwjJt#G4O2ho0o=`<PRA64Cw@>ERYC&4H{Vni)|wpi!P&0z$uYu*@%;Q-#d1X=KN
z=OZX2e~HJ}a_CWw0^F@?0b&5zRNcvrUd$9F;)jlo1wCH90>Ivi?B5@m&z+A1L<-P#
zX|&^Nt!^E106`XgOsK~3**hGGguB$wgii&2lF)_8T$dMqp||k0B4Iu|u{|T!eS|@d
z`_06>HF~NrvEh^8In>g`WovCm?dWpUIrjbF<xWR;elxe<<CcB>y-hvE#%3QpvZt7s
ztq;9jM;gwLTNF+hGTrt1$4H|QqGbjff9sdpH8h6R$FQcj>u1v_S_uMw$$mbslea(!
zF79^Ixj}F4wv1tSYZ~P}JB9i-?;7XE>YSvPp~nv+>~e11O~1PwXs9>yg;zUvF1?nA
z(;V{@Gg5LfURyQi_;z~miH+ha)zFEfui$S-j$If{I`539^D0P%mC=2J2xFK-tr{NL
zMwDg^`LZB^yc3}<V7{%|8zpUhzr)ZxP6LQA-*S>((bF0Q#mcA7i>=9T4<y1Qzgh2g
z*16!?=9iqb;DWOL{jkmFqgl8S&h4gEe2@~|?_QoD&u&SL4bXm5x^HXx!S+!9wdE5r
z(gclC;v@v^f}b%rb@c2co0}^u0U2H}Jf(jQubf26dylYa-Mwq*_3PFKrkDg6lfwba
zDX?x}rX+d<;=+6GXuIWj`uLXeALJ$`{c%N;L?V6C8D|{+H7Gt<r(t=efrwrm`x4cY
zC%oXT@$~Blk2*}uEO0jnU;CFPqI^9vgOih<!w&hzy{VLqTnpA$N=*=5=`paOdra=l
zQ(?vKON^Q+!t?|iDW1;uFdl0OG;;suHiXkI82%FTshX!dP8^s;RS$u_z--r+46fIq
z7>9mpq%OaTnu`~s<LwSa59~e1zv5Iik#;I$j2kV@+F8gY=_iwmxj|YW`0uGj-aCc3
zF=w^8d*PBiCbxkP>YS>Yz1Qx(aWJ08B#^bvd$V;kZ>W-$)xG|>E`4jct0JZsN8hH@
zef!Hjal20czP{ImkI%QJrGyEw+XG9%+4z<UAmuyjn_~GUv|Wj5{i-j~=7~|nC7R@0
z)N|5eO<CZp&co*x^3l8s^esn7m|F>May#UzM?I#u`>JB}0YdRUo5p}uIf8Bdc{Rn{
z&3S&ifDq<pwny#aqjJ}!mOK{w3%usF_tlB?$84`GX7pD5L0Of*LwduE-1TB+kjB3y
zC*7@4+x(m=S9v_0(8QW4_~~20V*)F0;gaE-tQ;^N(Ol$Lk`t&JZ=rSUH+U0Cea(fs
zzQ`CU%c0roQdi{e#YM9-n>vkg6$_h*2%6?LqeLh*wzB4h|MX|(cKK%0(URkpE7@w=
zT6I!h*XzSRs{IuAN?42D=<Q213)PU@19^x|3LVifM><mLme$FEFV9LnjQi1$S|{Oz
z;F$7U%SsYAOb|GE@{q2NV}Y^nO%*Axed3Oe)l9Z@sbWtvu|Pozk%7N4tSaRxRC&eZ
zfr3=W!m7c>14u@g?0B>FlBj?@DV4bP`5|Dy4b=T|th%QIlFnV5|4YHyJP)_YPZ^Hl
zC{2a=Y>$>FSx(Jw6-|mR&JO4|Z}cs^NP6Q`W*ELbxW~f0RbA%<l%#TA=Vt5E$gU>l
zLObd)W~4c`p83SDfQZ{hU<XT|(Xv0e4uxPijXj3+MM`!lY)%NSH;S#ztx3a>jaY-H
zIpFzn$ttnIpN*<PNt~}iwRtqxPP8Y$S!)(t_mfs!6-^i+Gsa)bQS7a65#GMo4eQ|g
z8N9!9)<|RMVrg*JDxYm5$_iZ~oa{JvVY`F&-9abp{i6M_ve0g)OkxZss@QNUWzJ1k
z-dgv2ut7y-kIJz-D(H@rTakVRhuyXBB4wc_)jj<8Uktr0ExG8#nvdGnxZMQ3y_T1B
z^KRB)+(v&MocJCR-1jHQS)oDJDDZ~=F16+~SL5vV`b@4UgP%fYPTQFi%`V;k2Qu!4
A`~Uy|

literal 53428
zcmZs@1yEIC*EYNf>5`HTLApCc8tIVkQo0+FR6s$xyFt37yQRCkBt*LNzkQzf{r>r8
zzCELdIm3y)_qx}*>gJoW;#+hSA`}RM&}F10-a!!T7zDvwBf)}a4qhstg9l^>X>Df+
zLdE{~4`yQ7=N^J6AQ=f!HILMTG*=Pay}R~|`FSJvdHqJ+MqI3(_FfV8ii_^{ZoBi5
zH%+KYqWqPc=OY4=xG~BSFnD|MBl<kFKE5SR_Y%+XLXSQ@1|)f|dafp=CGB`)Q_5h|
z{m;=#^YcguLix`T5|8=HLHWPOCs#WE&jJ5AHj2fONpSyr7?R#zk+l<BSzD)6@*;fp
zH`YQu3zk&D?K?RV3i%B$sRCJe##kA1ieY`}F#5H9dL$J28(tI(UKA<ddIio>ao0sa
z6pPvB9rLDt_PB)&&z)Dq#@$&C5()OVTtY9?>Uon&g02d)gR`ctZKsqg25)yh5{v+f
zf3G;=p<j)go}OOyq8LUF$wosEXNdDT^5>%c^oG2PElK{e1#u$v=XXRDNfp^OXdqWi
z^gqW5dL}%^|D0br#{c{L&hXNtI5{~L6cpm3kRZR!>yb1;Yinx-d3hSe)a#28x&L{q
zoD+2}5^9<J{CvB`T7@j-DckSgzOi$1>eg7PYin!!_&m<?u|E6H(G$_HIqo2xH<6p0
z>mE=rIXSslXO9VmhllU&?MX>VL0C9ARM_7V<?};B<v1}!lK3+f8zwkJRAPHfypL44
zXz4;;B2ApqOm~{)%4?5z(n#iv>#U@Ki-v}<=@pTkWJy?uEE;(v^o2a2{G~(1{6Us@
z@Efnol1Hg1>F#xhSTD(!QBIqG7wgW>&eYY_1zmQ>noW%=wU}8?G*R!SCoH(O`wqM?
zbNMYj>ty&z@$llny)u)ccmDM6(+5@!LQ|)%$a)ecI^RfT4A;rs_iQ`$n6N|A)YvQu
z#Y^>{Zk)EsKYaKw?7@PLjvggFb&BoFZU1KGD4YX)K#JO6lL8l!g1&y5?ES@N*Us4o
z4(&m?`)OD*I$X&l`QPBJuXYwC3$?VhU(wRSi8wh`#l=nY3jXiqbKlWLA}K5^yu18+
zc6|J6czC(dwKzULe&YXBr7!#zVkI+olT2QIeh^3D{hBgOTn22OOkQ*I{lD|dl4lS`
z1$8E-qNcC^@UPCqUp;1L{^z*fSB6FWpBtqajLrAIi$Nf^{lB+b^kh2g|1X;naQc$}
zttPma_@Qy%Dqz6dP|(r+sI9$BF!Zrpto?gMs)hRbf8NOlmV<+%)^3>r7scD#I~arX
zcEvRJou!>!Sz)0i4r4`bem(*M!hSq{+#VqY0b3?`m(%S*(Qu9Z*E37noHObC_EuI4
zB{A0-LI39+@mUS-3v%+x8dgd&g-j~vW@c(_7i%vVOQ~sS%2K#mYr@^+MJcP}@Qz&5
zwaJg-7&Wu$k`oeM;Nu5n|7dEuTW)f9b#dv9ARpa4;jvwC>73UIUNqZ?bLfc+?bTEC
zdbeJZ^9(X2{hOVWL+_tMnDyzuSLIDoRTW171vWOeV#b@aq@<g|VVUUf-@m{1y0np$
zlw9uJ-?wVedtBLMWMGhIkVW?_;)3rlDPrMw-|sjozJGW?4v3V^<1`y0SS<VdLnTKM
zW52G!%bYC6bE^NeQ<?nkd~35WLQ_+dm4L&+;>NllHn+kMj&`tU@pS|y5)uRzsugu@
zM?!J$=I`CgT6y9>m=vbA!>sujz^Di2cxG|^YfVpo*7(wslZ&#lvWkoETKZ0Ll(n?@
zz&G4Hjec7mDnLl+k(9!;TuaEyuBoE3H&uue+7m6C*A<NUG}cp6R5UO;D(tlNRnjdv
zFxhC}H@X+AAy}~~!oIDqA1jqv>MjU})*Xh1h7iz+)MjP%^;<-!84B@skOTjspKEx1
zgxVVsE0xk;e~b8b6Bv6%`o`gRnDm>_XVBmC3p6!#nD#E|C#mj><2l4XO>Rz;lj|G=
zLPPQ{d^U*83X4TVab2Q)%OqAG-r<uzcqL*&KgRandVhrTTiqru^(8#|{reKVZ089>
z&FU;8C_ZYV?#;iS0`M~y^r*2FlDYa7^kP)?C(V|=6si{Z+?`NUQ^&@~yB*A_>^0KG
z^pm~zpuj?U({Fi_lE-6MYctQ)JR_rP-9GQj5<&0<Iu~+g?_QKk7<5E?Q=9kJTphuA
zyM?aFpoN>8n^q~kq@aKn7X=Ouu6WagvBqV3FeL3~%>+Zy2-*pPUz;lD6Sl0&v-?`7
zembu|-v^2q_q5+=;7evNIHB7f83-X4{%ma4d#7F)kS%%RiAJJXK-s8=0exlvF8+iv
zXa--H#*u>cgIm*A&C2;zZ2QUN*lFu~@TfYr@hn{U>3;IK!`6^k4kB>4%rG6VJV#iq
zh>eZaZ*r5#n?S~)arwJ|_iuK+eEAX$4ejH_zWQp|8mCsZ89MWC|3befB^{~b;YBnA
z)6Or*aJ%mdwe(Cj>%pp_xty$T_cxwvdwSGWSD$Y4^+iVyQz9}}Q&W?cMvCbt5&G!j
z=5`r~`hulV5Ls)eaMteda5>FC+nT()<l-RGkSc_piNwrQojJyE@YjK<aS+aXdCxq7
zh_`*XZbf?jD8(Xl99xCa5TDFM+d(7u7YZ46QP`MoRa)EG3q;@G<78~PreM}YeMBpz
zODOtvSTM`!isSp)0~{u&W$2i#xBpJo`TD*K42#NDX27%f1BLJ$wbS{FBfoOE=la>q
zN(dfte?tB_+1W3JgqC}Ir6{oI>ZV@sVUWF^6IicIXh4S0+B_~qwp!d_Nw%DB5;6XY
zBYK=94)tqN`3O0NQ1|Ci1&uga$Edw?E6bKLG&I!J^_nhH*U`}-C;cw*QoG5`cI9tM
zGFh5NsSarT1@33Bj|3s8EZ=;Y(+)hrj6)XTSn<pcS+jFiuX#%1W~~J_SS*Xr`M12C
z-62@e!PR>UB$xiz-CdgHmd7_pL6jWyMg+gMxf;GJ(?zGhF7(hhGJ8Ka&<!8e52^Zz
zQI;3>%g9!Yk11n)QKo#srcH+~nqsKoXO;9la6<MKHcU)je&<IWJ*z^NHTI;02@dZP
zCM>^qmDB;bn1tLR65~?`se<F<hj@~t=;)BqL7s2JAt^~(YHv8{;CqhMa~UkErR8vp
z-O>(?B)5l|*cQap)YRCfcozM}ui@H1kfr?|9a%?G`6rtvJrQNsl*MlTA)$K_iMMYP
z5)$0^CIUxy+Nz6vs=o9_$>yOUzyYY^aeci0c;gJZ%fT;3t^2nQ;>GuC9iQPaNIjUb
z^E5T4iBg!gDnyi&SYN#gj8E@$vZ{9dGZP(+{X*Y1Psy*m^n;$B9smtRBV?p#7dUbQ
zKGhr-HI?#o^!hCyFIv6tRs_kC<U#v>=U8RhhjBDua%2p*i4r<nrr%^XO!VEOdVdk;
zYKWfpNZU26GLIkn<@h>3(;?lRyg{4M*Vh*Sft@pPF)?^gZ6dqBGmn?Vq0~;C9A$>B
zB_R_Z-Tn@6@`y>k%$Eo-LXlvBYk=**yBW>E?AvB=3Hliph93+Efp+WgdaGsA1{0za
zF($+PJ#G6QnzOcqg=EFmcMsebDUhVhBVB&JqbrJHL?TkqQ`KjVE9M}uyA`-PL$TqS
zcI8XG^L!@OkR4jtvHs-+N`TAM#I+!^>C`En3(HkR4{GP9+X<}qZXUVc>LQ|QTqPvn
z8HKhDR{kt$KSEgdOoeTE{2>%QH5*#^zJ8@}ke1dwgZxmfB8mO)9$)4X=2^~L)4p)`
zAgP>jo>YNqhdzufHsAB?b^%K`NEtV7odaq|-ayoLePR`{<ntlYS(!+h0BeVin*<pd
zIfd80?8lGY<)*9alg+k?23R<_8bg|7<ZD{PRsj#HY(--V>~Cg6i5>m2tO?RD4whFd
zP5T;Y04f1k?!`9=Fh0m%d$5IXp8Xz<dF`o3l@Rgh+7n4B-Wx{j-Z}Dp&`3=>Z?ZZR
zll%?Z1s92@rzaZ1=V7zeCikP8vz_xFUe3#lXiW|m(!Q+D+@da`#so3EKKIw$n|*1-
z6B)yqN)7nz7Td|Wxzu;LwBT0cJhoo$_`P9p!G%7J=<9gj=QcF(fA07}OePaOn83=~
z=qwjt{3)UKWz_EWV1lJ-&=?kV<<EPgPXT2Nb&&E4o|rxpr`>Z(o_GS$R2jvZ$*CVv
zgP;>#sPo6YOCq7Ymd_8MzL7`Z;tkB!Aw>;6FR>wJA%-NJ4rJ}fLRskwe-OOSH4HVb
zt3JC=`&({ic9vLTM-xQdem%&0xmu4buOa>iF31>OlmuKAhJ(@Ug$3hU4S&+5GpjFf
zN4)|90ss>=iU(sPB_#o1Iy*ay=y$d=nz7mp1AQD#<hGok{y=L|dA-pSO4#>w?p$^E
zuuc^~t$*S2()rZQp!vOt$vgM5#KgoHS|uwBi=jmJ%@`7&>gwtj7^F*GQ+#C=bM7}N
z<m4Hw%|T(}mVB|wuzgD6gLly?;L-ox+U{h*s3NBjDtAY@d(Og${v5awX2MYtFOw#V
z@Af)YAO~(C#!`KAU*D&-wKd&3yX~zl8++5$zD~98#_7C=uKQE{!^5_}A}JxZ=F&eI
za38H5=N?Ayna4Q+#KCc$hSSrHfmX~P-w>WzfRSTPs#~e7Mwg%@&x(YO{tU`%HNHLH
z8`>G&;>((;=}YFe2E|o5D<ZBR$bEhP;lq&F+}T!}>ICx?yb5(ue!f(6t`?J9UoJM9
z5>;k!@_zB@0Od?6f{wP{;Lu1(Nv+z8Sh-y9AaqSr_R!N)XO0PN>MW(VVb`xK#YCgq
zZ8aQW#8n#928`LDwVEICwk8Z?leAbwG6gXV3?g9F-ge{|tA@F%E4$2Y7WCAZWXZ2|
z{4y%}Jlsk<np#uSlt=*k4}zqW8Dt3iJ}Qc%!>LcN6tEs%>`UcP2B*{5{#{3gLhY7A
zV@*DO{0QJeIESOZ@>Y|}eDFK_>0s)906F`*yMGiFu^6^4b#`vV(0G8AVm_MA&&T)E
zc9Dmcwr_kqYS=E*<VLUC<RtV3D}Y!(Yid%8M=?c_No46_P@X;OdE4%oniOZZXMzgl
z{Iykl<8xmK8!Yisg$3&D>H^Hh>ta7Jn=X8_qoaeJo!xffCo&R}h>F^&Ta@yvn7rIv
z8$i`?9UrVlS)3VVbqZ0THR;Y!N~@))FIxwdRKB<`Fi8cb^uo%^2ZoEETUDrsyW6@S
zTfSrD=7vNdmxhDp=4Qvu-Z1@^?1;em>GP63K{v`oSX`7qt^0%i7{zVLM{h6K_VL`U
zWL;hwo#eQ9yu#v&csB>R@)C}XLIF$bFy{(MEQu23u|y_>fK2$u7353%509(D<>_#K
z?0%KEz8aJmo1e206r(*+mgYKN(!YwM$1~+i7dJiMpJi>T*wwzic!QtsMJB8}Hxk!p
z^7wch6`CGiTy_~gcYS&G0$j}!V($Cv_Ew1PaKg1_h9xK{2nGhmhDVnI3(!wM5Kg^3
z&{wXT7(R=-+qAG-m;%sKo+I^he0pNj=}K6J4eywvs3`R3MVUd1fXDG#e_wc|b9Fsh
zV_1HpmVXN!bHXMcj^s9crS)tX;0}<8v2m6Ng+;b4s%WI=*>7&2@A5QFPTL`3S;r?Q
z>rM?9?x5G9lfBlW@A$L3%eDPoA-HS3zNv}VJ_GmP!u-D)af{d0%Id(FM7yCk>w-_O
zjlKO*{#6C|q=Et^3Z$2@gMESv337Ruo<x+ClwMbd>Jg2C<MZoBQ{QQS&oFA1p`oCB
zbiDmS89^K$)if|o-<au;SgP(Tca|8?NgA>7qSigd=RxB>w-O0b>>Tr|^loyOLvE|=
zt@c)|sgPjWvozCt6}LPzaKXic#el2bsbk8L5<cc~;X+8=FM*{-8T{L8a~+K)tb>9O
zEjlkcKJLTd2RJ|6u1>ju`F*1A6B~9kQeq7o5!c;5r9;`MD0^HB66SZ!MC+$Z8*r4A
z#I?3N!r}XNR?V!3^Iu(w8(sfMVM)l#qlIsh3c9HYWG;bcZZG~Y;WJ*`a$~}^>;{Pa
zX<cz@SnBQlz1|h<)Szc*DExG2o26$wu=DrN<mLzMo-tp0PvYI)7kGk}*66jBuQO5w
zc4y<jm6Tdtcmuenn7BA0T2M;L;=|n)9*SgtmzR&?<(1P}Sa5jiYtO2Tj8zs+A!qK6
z+~Rc^Byml(Wow=Gt^SylKaJCfT@{8tH{0>Lf<|nwk0jIi95+Zwr8wUxk7fwJRhb$Y
zDbxOVM@@$wj4iu%ZE4|YSoK8ePsPGw2bj9FR8bejDwsXK#<<(F2wfGoj#>Qt{5Ll@
zK<OFF5(hm3(0H$y!T<`a1~dnf4mJv82|^Oe1h{Z2YX9cp7uxD-E<#&l@aA{suku7h
zh=pt+q|pJ#KS$0xBcz6;=-V~*1{(7@rf=az0iRZ|{C4l^-;gX4U<?M9>YEITZ$xN2
z6<YGq7G>8+*uPM|#eU0vhsH|PF|gvVrUHWiw~H3t1B(!oJINZn`elgATm5kRXG#Ab
zAI;fuZLInd@)!-~m_W`3KqFo2PIt$1M!UiRab#)meyUUaI|P3}%<kxW{<lAC@_{9|
z?u69J$VgdL6^EF(I6of{D-Kgr(`f2)gO=pz=!q6DcTLUN-nY8Cx+zIXmd3^%+fFjk
zBr`MH-R(X^HgOHSN7r_9rw8+#1qGD`zi9xqaN*^?<l-T5d7Ln8h1=yyM(caqGK7MP
zdZ#>!iqE7~D$Bc72<zv*6v}BnvbDW!yHsBZsO#jUbMJS;VW%2liFfbbMRfUgZWTh1
zZ<^p@p6b=%@*ncC^Q)EMPa%#Z34`0<I=1{Wf(ef>$@IEBP+6(ZZ=&?2jBU^Q!cnNK
z{K=I^A>}mT2vXN<VrH+@EH@x!Gitvg8*ifELoh+6#3q}oOp8kEkgyc;MyTQ77?6=E
z^E_GdAH^Bi_WfDAo=wM68f|aTmBoEgU5?1sF;509?B1KxAKBWWu;K*o-@o`fuPI0R
zE(XRArX3FAhJh;fTu{M7rk_<d9fXXo7>e`$m0x44XesO<6I~?sm%y~N<T8WnG{H0#
zWmRLC_9IMk=D8;O`J!oSCNye-R-xbhwdwj;B4y^kUr%u#PUOiQt+c#4;`BUsJi-E+
zR%!z8MV+gx+*ABo<ngfsI$m7lQ-+V}{q@N=78V&989P>08UaB_BuR^dfv00)g5!A1
zOUx94L?dC~5V%hI+P9tl1TYN@4i3YAvc({V5OnK4-kuk>w0L*-ID@+uSYv|oM#*#U
zi&J1DO0g6TP!}w7%`t*iSqj;=3I~pPXI`3`ZUV6=|NH*n`q_>03C~7j^Y^|@Zz)=_
zEH(;?wMH$J&Sb_vGqv`st#-3XoaSbv7|%#P^RTcvazLpV3xZVq+J=U{zlV}|w+W`y
z3|H&@2k35X|B0F*qq$X8rvSThB-w2Jit6p{B^Qu?8D&wau9k!b9dq%#CdB(-Wi_zv
zG@2&Zw{!OQ?_csapDavFIGnbmdl853Sq+*Y2oW)u2BQSvV9)@#h-?q$Di;<O{5HSJ
zhkWe;|4Da$)sip9duhV{68mM%%iLQ0LcEtKfmMIsk;3fO!$Qp?{<HiVkcJe3{Vg6c
zj2}XKCN~xj)|dC72-%VdL}S%8Lbg!w@bG{VW>g&$9gU_;q!@i7`3m8+`m=D(@88o>
zQ*luwhc@f?@0(4~GK{Q&SdbJqZ3!Dz*5dW#d=tX=H;$ysvb4k}X-jFsqTL8eQ{+hW
z%Z88Pcbv#&qqWBPOqP^#RT1|9+eVKhQe4*+1@oigM)mh7vV*Hn<dx1s2M#aKp!bc~
z@Q7I0Gu8oy&gkeG{(~GDR!rN~;=;oDUp*z{&*2~v4D;i)jw7Lo`G3dFMz}43D4jYg
z4vjp4pqmZOk<R4!)YMSvb9^L80lkr=gu=FK#QMg@H{`p5*XvyjJof9QUYyH102Ags
zHT*6qjnmSpy`1Zp`iP2zl*nxzKX|s4@BweKj*Gir%h}k<N?>*7<4!W3i1~F2k*^j1
z^=Cv(APk$Co14pLAlokb%Qt$T-Bz2AGF-{#O=$ZDb%ZZRUZY{-to$@)V?#qlL22YS
zV&khrsPtH@WiC)mvs&>~X?`(~dpaqSq~=3HChVi9tc*oO)HtM^()Yf^2g{l5KAuBx
z(#w95we7x2O|>SjQKt@Pr$sl3-r;3Sm6gJah&#`6@-`NonabtKjc8x(W}k=K8oOm3
zCVcbv?>#&`cK7z8sQImDm|BL^hyr30YZuP%1n}UqObH4jsZgFNuu!o-%WlU?g2&Y|
zkou@IC9Y1R8W_G{@q-fST~0ffmb_bxv+y8{G>Lf@9VGDrxAA<>>q29T!j@YgOygAy
zX>t5H8Q<=Rf6Js^D$LqDD>H@BH*g5(QmaJEXQ!td8yi5Fuj|)Z6qR%S`+06s@&gZ&
z@mOg=PEL+k76He5X;;^pqN07l)oaJU%W*1{fH!`Kiv}<hPDDlJxhNJ8U5~<n!(w7;
zT8AJ%`@UG3spRr#^=T!}(aGsFP0+ohv=p>OWN5NLDMO@7&!xt<>BaD$g|4Zosl~;n
zc`97HrTXX1ua)%!0|jd4tId`4v5S7cqM|ZgthEKhjY{hK4@SOh`TMxYNGw7^%MTy6
z_x1{zrXI)HjR@k?(rk^4#w6}wMX|tu!6bWKQCFA2Z9S9lHe5|Z0~`Ltpuaro%0NwR
zD$|tK=r_q+C7x8oGjBM^ue~QvlSN%q6ZE#b>0(V@LMAN|BtDGl{Lki^?Q9uU8<CNb
z&rtDG2{>S5&C&GqK5kFEP|R|qiwXJ9kj{h;FNYDu0_~icj_&yM)M~nDx{nwt(X9(7
zI}=d|1Tt3X(E>6A*2P{|AD#Hks#gjF0|N#LUu1H!l}>D1wdU;RXv{oXuQKuAY7SXg
zWnHE6CM|w~5Cel<d3lPOjvd1q@tNl~_P6jdvKR_ATqm+(&o?^oqQ3|6f;2SvG`Wj&
zf7aK|OpR;zEv2e$KW89WtB9GLn2jK>t8%{kIZzZzC&<Ug$H8&xu(p8_bnEVPHv1)w
zjgvEFF7Irf@<Dy~hwiS7oGUkdkf%b%-q8^`DXF*7W9oXXhI!0n?1Xs6<O^1ei*>eL
zuZ{G=h|A(~0wiSQ2JgGt+FCawqs-RU$NlNzVZ&Fh>^w_1d_KnmH9{ZHEjGTt=|^D2
zMR~g3j5s(rczn3KyS|?KK=eQz5)v|!E;MLX{WhFro{c*RJDM687@I-g&1e}J|E&5x
zsjI4<U0ggY6kwpE>qo~UcS*J9XgwXAE2Qvvw>Wd3t_Xg*cA&@{YlWkJ)!D$w=EU_{
zX+I<3@7fv|OBY7F(iPRgGCpjUC>9nXa~U0+<`&l004wXYerf;}EaHTMMvknsxl>r>
z`*5L&*2SV%9|_=GxnWzrHBJiBD+jtP`0xw_*)<D+%F4<nx4*D{=|WyShx%<Ud+$<8
z)XYgs(<0IzZ_N{1S{{PKnOIm1YOLa}=W(EVzNrGemr0Hfm*Bd>7+G$8_W8rZFx-xA
zkMlBWtcWoxd4J;k=hB~90b=5*Ut4tO>ppynRgT?>0mGWbSl@P08vJ?~;-o{gNUAmB
zm5x^SOiZ*G;+IDHi7m~&xrHVp3l^j_&F9mCQJ3^<8f*fng~HP>2@U`Tn#YL&jAP~E
zdT%0f?3R!qCMM?TW`|r~UmrmJt*!TRa&jyQx_Wvx@89PZ6`??Y%>hiw$jp3UGBM+_
z2SHS<28*9=_JLJ$MQjO3mYH7LUu)C(4r$+k>68KlFE|l-QRDi;i~F0ixPpHMmDRDo
zdO8{XHr}9t=9j?0bv~c<&Oor1$$-AQJGOqtGDxc;T;n-eSW@znj31@<)?aOcd7~G|
z=tLU&g`cVWfBc|p&|7J6-oYdsOb%^semzUC+F=hH8}HdCEC33q+C?t<y^W2Ht!-&R
z0XY3{ZEa6^e*BqS+1wc)MHu4VcpUqU?$iM?V`oZr)Uvg;#looB(V3^Lvr+3?myn<d
z)IY~0*v6|bIaw8!jF5<;V!>MG;d0a6?JW^8vH6D&HK>Wmn=~ShQ9BQUX#%cQm6eVz
zF7k?stuA{Lo>Zz8KLcX=fW+$J=x7-)Lk~QwSFxCgl5*l%$~MS}ZjL`NgFCbw4yUNh
zcW#f$ZzHL-J+9owl*u8FcWPAw_DrN%doBkhdg%!8@LIaMv!u;HL|>g!vkAUE7e3>K
zEUM<EJ{h!e`f^%urNF=esRi65K)9cE5^?b&meLi(C5Ty8W>b{g!PvnXnb<(k{YN6y
zXQE%XPQ{|;WX0Pvvt~J}?sO6vkT#}5HE~i&k+6SDE;&w5eq&@Jdp|9!ePr66B=`Dl
z0g&}IsXjp>DZ}XmPp4B8c*LUy5|&?E8QjIEa!h6V#TxVy2InWbRO33d$R8{mz3N`o
zk)zeskjYEqo9B;`|AvDWNEGK*eP>_rfu#%%gVs;{#lEXB@%Xf7D7UQIJeCu=jXDfu
z|2Z_v^rA9^%9c#ZdRbXn%{h`@Ht6XG*3`-cvuN6D&q|-|mZWGm@FeepelnbTZdf6%
zr?;f$Wewl1p$d2Snpy^1avYE|-*xXN1PMt=(ejnG8GPvQwQqWS{ACt_Tr_ooLW-M;
z0^HmIneY8`eEeRMN~8b}Fijg9Zw2RRT2q1e*m)eawuAtc355U|@_Opj=I8@ZpbL^1
z+Be;W94MHWN)(gcn=^btRQlg6fVunLiRtbEGXH$ncl5S<2?z+3z|!*_oqtwT)C*K~
zJvOfJw6x_LZO#btM{!kE=gajjpauYqG_W1dk3u9zR0er>SEN%FNdCgY!XIJ)C??>>
zPt}F{*-DP%?Xo`w7<L1z7W`i!iYM-&R78l)a~}}86wS3vWib!~U=9QNwKckH-eYD#
zk_?!d!SEP>j1iPkAkhac4>H&Dm^ftpaQhwh%jn3!B+z6bG0ONVXa8U_`<k!G_zkpX
z`horS$U2_Liq&(U()>~EeU-i(e#2DksW#KOHX3@~wB`xpi&_qna&_Y~g&`z$14Y*w
zqz_uPu80^fRJNlS7?8p>X=(4|31&M)o~)!aj;pC@ZA<GrH%>+{xZp{+xF-%4e_qur
zpWNx-hdjo_uN(iU>+I+h7NmO3SWwrn*G4!`Lqh||>?)wk<2i4Oi;J%^_;<FpzEDVm
zdmMkZQzPK&%EPHkOngUTLHl|M5((c#2{0P7;GgCu%AXksb-O&!#t-E-=|=Jw!NSIN
z`eDe5^Ts}?_$+g>;19Oq>^M+~3yU=fPCU4sZl)PjCI;f@LKMS4&W|e#m;La#yVOrl
zPX}lL$Qds}Luu_6_YMy9W@mbOw##!U+uUptRJ0Xz<{#HYJ1@q^$Nm5CuHK(DW<Gz4
z0Iq8&66rk+Z2RA~5h0J`_T^>o)2;rfPuz5$yge4C{NWIJ&e}K9#qKafqLmFEJuAtM
zj)Al=$^BC6H7ayi^kb>jyNM;!Y^Fr}6{BX5LDTD@F{6w9X+s|t2niFjZFU^z_8iU+
zhuHV2=I2lPLvG--spB81cW$0!{={{0G3utv33NUUlQ*kzEj_9$&dGseRc$Y2#)4PV
zaN<h^vTWx7w@88b7PB&)Ux(j)v(iAc;<yV)l4R!@8yQ6o%u2JAk8L_d2rC&hJ7`lO
zD$Un>imYM33AHTEk`j;X3+(w^s%z;yM_rMJveag6Uhz}A*`qQvG}KY4JhYI|n_1o(
z4<%wf2Z3*{tbOY|v><XKx)dY+;*MI~^GiN{R-rc^_4~$RPDq;llZMkwho?W3sre)g
zK_V(s<Krj0Ey09@gsQ{CKob!a^*1)z@Jn0dnz!lf@I1<vbP-sdp1$Y|L>)a1d2*>T
znFNw>i(W+PPs=0!O+G<Nj@SHqZw3S$*8I{}e9ax)-OC5__&!}9?@bn%YtAMjU&EE^
zqVwHFP%2&2)TVpO^angcbvlw_xpaszsecM1mW}YvtGp1W(FA7W!)#fGpgRjAqxDy_
zej?s~a*<lYlJ)igy_wknCRxU)VTF-sS1q6f7#|-#wm;k@Cnpp0+TG7R+^2W<^w_R6
z--N5ZS?RT!%vU%(JghF$U#$CtOoy8igl?P5!}PfWx>+nLb>dG`6m~n%dVl1zHvSx`
z(|7!S!VHhExuvB~N+R{zmQcDoIA6&c9~C1_F}SW@Emrl%VqSK3)rY}&W`$k&u8KUl
zgrK}!*PS0s+Euwx6u1bVPgeJEQO>JTx)Bi&;+b_^USzq#Io%0AO_euhzh7(jud^Q&
zY%w963K8EK5>ttl%7F=oNyC5(Zl4Iu95_?ijCj&e>3FhMenTRR9*g<b8vw0(y~AJ)
zC!iRDX<uAYa(;I9XQ{#YL7y)igclelsS{#j>4Zl?@&}-D!=hp!(LALkJLa~{&Tmc4
zW-mPZt5^KKO3~O{r=iZhw#vn*nyZL$nR}Z1>(kX_G#wkyXADR;d$SiY*oljq9Sy`3
zbZV@oj|4thqhFkrPZ;pm0pS17vb;K4jWA9G6I>8Eoakl3V2?><a<a}t7=VE5Z&wm`
z4<a?GQe#v}@bP&DmM>0^MY-TT&l`8MYsUd15uP$OFjxWl!}xcusS#tKbAvUX!eg8E
z?J#u0={rS!BOgqA@*R7N&uX_3n?^YCWLw-#*ZnLUoNpSNacA(rr2SI8qsNUk@CG3Q
zw{_X$haXZ)M6C$+9GRgKT|wy9vaXn~6wcSj4i_<JY~gu`%M4aN<d*9(4sJUUg<XzY
z0w?oa+VfS#{F|CiU|*C1|DvuI(VF-ledtMe%pbQHpaLPMI-B`-`uai^W0@*BQrwOx
z=*ApL)G`b!QhdJr#Tun31H*`9<@O(Lz;qL)+V~CBg)`KU%r;D<KsMKRN{C({3P92a
zR^(!hHLv^Ol0b(TOM+4&+j@RDnVRNkC7W)gNubl6dGi%Dw#=qKCr8rv4kNnqCs4^j
zHgoWbBgagSGc(K@8XCh|WqB9_!Q#ky>zUX6%5=D(lqZM!>G0$l<&##F7cn6qXn-kc
zSKm@!&qqPguAz1F^T(IL_j49M+SJqm<4Oq(?CJQ77(ph)fc>o`ZSnT?r^K{0(uEE@
z=x_(`%gwp#^dt-f&!)Ts;|~L6<%JQa7$Zm}rayi+)I6SymUIf-^Ala5p#d^TB8Ym3
z+q!C7-jK*yyRE==9k$F2C90^ZZnik#l^7tp^oF?>XXt2y_QB1eN)t1GjCZuNQ^zCd
z^~*4U5M$)e3@|5gv*@cG9Fx$J<TakwuCK4#9&hD<ufEa!4c5PXZti-k|LGkEz<_*T
zix!A~=1*Ff8|2gRnrWz8TUrtl5Rf!_3bM104^C7?;Zi|shlf5^cfV%@bO__7oluMI
znQSAe=RGe;u5G!I;GZwnC8nhKa(ytLq(p*ElvnogK|@nY6NpMq764HjV&as1uGMDG
zYG11&Jr;Kd2NA69*;)6y1AaCazTCVK2U}HITc|zymE7S-Dr&OMb1%2##Kh*9n;gf}
z)6>WMGbMOYfGdvH){aLo*HE>q&B6?&f~2XPoSce;*YW{!hYPobpvP7yhGzA2A1?HT
z-_!ewnzoBk$(1N7)qHssvi|oR5#++@iHUSrNJGQJ#DC6BKdG630Gg{WC>#EZ?#lCX
zm&Mh1?oj%W?B`%aY|w5F#4&om&+Z3)ib%$Tm*?yBurLMh7yeR6C@2f>^N)&VKkf!H
z2H$xVR%$hHcgBHbyWFHHF21Qi@Jw{*f>aOtK^gyr9PG8$qvmRr8k36kEDR@ipTe+A
znbq<o93ln~O)KYZ^1{7$M$&>0n3<;Fr#Wt}V^b><`v$twg?%v;M>1`8$0_OQ3myG8
zTgLXzYpiEG7iwH}1;R7<?vBnt+$X%vqvmldsqrt+dH1W$Naa&tdL$#@U=>sOP82nP
z$61IMM<80iw>Q+}*>_bw=#;v<FD#jKD+W?St*-at=YWV-R5%=8`~fXpVX}&sHZLzP
z(;kvV?6(uz*xXD}S<4)^XubXTBAd+b8{vJcfZKGV78$F>%6&IgaGdM(DymF$o&72c
zJ^dblFWv>FfH)Qei`%WX+8!*|wK~aBV=qe*W`cSg%h>@VIhn-)DNs!FiJr|C0Xn$M
zc;|K|RbB94VN!A(48(wi8dF;NbOaSlqg8SUqC=^@grJt@^e+gr$Rv<R&Uoy_dMo?>
z{=OPsojn|gPu^$7CM4L;Rb*9EjP>^BS5$zNivb+8+1a}M{NC6>7{8soZ7z6~hL0yx
zS)IYHRLSqovw)WwF$x3ybC+o#i^}=zdk!d5T{CN*@cclYixxFXW*A09l@zjKE0JYl
zUq3U5)3n1HxVrLc+iPhh5~0T4+im_PRcJqJJza!H{mQ={Us<4AN-+H4)3&F#*+^>A
z?p{({9Ny1TQ*YrWP!{cD#zrP6H2LeiWguCG3LkKbK5uiGdy#_hKqM_5xe`Nz2oFzz
zbV~U9>7Tbp%v(licDe^BxcIvDj}oXAI+SW0N>cYNS7T66ROI2|vFo1Cb_9{Cb%*QY
z!M4$d>1tk}yg|E@tPBiA&leSpjMkpRKf_Ta*tw#nu5TswkBU;!ua_V#Dk{=<>{+gQ
zH#YX=*J2sYicscrY#f}d7R0$ZSm@`~*NBL#<Kyet8*kTz8f$mQnIYh4z2DsfoRj$D
zM;QOEVz8_}>pmw}pku-Q1}KfCg@xADWE@uRcj|o%<-GA-;fP80XaBh`OuDr;$QjeU
zJu~adTHOSOw$3fz1nVTFq)t7W<&@HmsR_@ijeRo?XNdMqNo?-EkW|&wY@c}CT^-Rz
zXQW)57&JJ25O1pnMKzkt>Fh*I`%H1D`8j`g{`A?z>F~<RQl$6Kfsn>7X#qH1j*ej_
z2UX`l5$~Pbi=-Xa!n!{dX2Z$(-#-CYURCqqGBjto=6i|HCe4iP3nX~W*^&hinUIOb
zF(0*BQah2Mi?ITq)ZGoAZPH%DY!rg2UI=m+?o@%lfz0B)ea4H>=NL^*TsKaE<m5Yg
z2r^sWdQ1!rS8vHH{;_bG$FbgL?^H4Gl<u2<V6h|!nvalVDPPa-vWs8=jx)eGIYK^!
zo2^V2-Ygq<kpe+eQ!EC}e4z0}q<;Xj8C)rl;y5`u0a3uoU<_*yL4d6(g#Hc-TNZ+J
zuR_WUZyvBp05FIj+y=kw*}CWMA1e1Zv^)Bm&s0!Qs?Ensj?oG!o2akYQA{2iJD}th
z(lZ&2^?EHD);MV>Ph8{8SZJ~nOWwUG+*%eFv<AY?L}B7NCzeQvlJiKNZ&k?9Rm4@e
zaq-41tj>9TEyjfV-o|aHW~EB`5<SkQO2zG!4R%L=a3@3Xy22jV?qW62|1%FGO2Kxz
z?w3>zVkZ@Ps?!QeJ7yY+%AZV>IAey#BTQ+(b118$`&{a+n2U5xQ$x8Rytf=?Q{FpU
zShS`YpXJ)-3*S-W>UgG+sLj|pmUz!Z>*?W@5*-r*6x6h|G~pT^2~M%-EzpObzJ?-F
zj56y_034BDSNh{eqL7zcy~IB97N}F8H{DVejF)!<#Z1$%0fCFU;_Vek%lSxQ{(}%s
zwuq%nFnvwshWeU@;{63QkBu&S(NB*Qnd9+Smqmj{Gg#5ouD6$;ZJQv`HM=buydZ$O
z%+Vf6!SKOn(Ifb_zPz^^&#VjL0Q+taH#-@jN@)mwwx&{}=|Z|RG2G?EeFZOR`^BDV
zd^NstlFWkjDpkfc@gNL}+w`Ch(rhcmyvpei!9vHK&dq6bQ43X^l^EAvAZi@miIm_E
zWB8XD6{ce<m>HEDB>l<lZ2jd`%|^mDUdyn;BH=EKUqoDBO_;*d<AXWo@uATV^5pB6
zp8FoaW{8Tj&IH0#%h2~aTH2?!LkE)K!7TH<x)EvTLZ;vlv2C;H;*b<=x*?r)GkoeD
zxk+}AZGI~uvAe&2dwc6c_iBn4PBhY0wr)ZRq#%UWKL-W0ba`hvfQnRb{B0rS<m{~Z
zR<dh`6soY?;Q^Z1Xnmqn14-<ypN`7=chR+@nT{HzGiCb4LkYKqXe4|U^!_DC0S75u
zjx$xSGRu$D8A(X7PYHLde*GR!p(+@Hyg&bnjt1=wWI!wIX1{>Qq<rF7g$%I&L+}_C
z&37J6EGymqE|jlhX!~y_ad_71MKNo>XT(k7&<-bIwCv<iij$fYQE?R*cxwrZyFF<E
z8{1ouRYVu|nSz8UOh()T$#$o<bR5pxzXulW5B6zaIH@4yry!nqVMAvUpf;t;8Hxl;
zCG(wH5*`K;3&6#x|3Xphs3#wZ9gX{30xw7tM*=TgbKQemQEi<MqLZ#7=n7#!*j_Bl
zw@^71J^Q!sKp2Z4Xu5EW*JiE)m=V#}@Bh4cFx$q{)X*S`Z&UgFDmE5)rZcm%!i1zL
z$;m`@J}pjy2{LUdDMw}HFK{z<`Musi6p66I^5(gDnMsAx%TD;B4nlBCoha26-L4nJ
z8n2&TgVEpm;k5e|nXO^>lHv7nRu$~LlB|pj@;N&Uw02Uaso}d>{o=)o&*QqYh^8^X
zhsZ}F0jQDBWfv=e0^n94jDeUvY7#H`$B!L*F%@=ja|`655vt2oA8dtu7u_jj{@HYZ
zVANGsf_Tbx^YO1Js<$7#yfoF-|3OMk#p<FWhOucdvuE&80u?N)z`0JQFB>O!otP`}
zr|^{e-hKbX(?bydGA?fbuROO3w~$GarM!EkybOd?c3r#?^fN?xb0;qdT}TD=LBVI|
zEc`2cbMC)YnL=J@`qi+no5WnEVhdvwl8#PT-dSVFNk0op*!18JPcZhV6jKS`+~&(V
zv)wFxe4AND--GE7Vo^X*etfBttIUKUFjuB8NJ^UK-Z~HH(B6xmn9%J|V&N16XOjHe
zw|#zoe(a(6=WLv1vA+V?U3{fFGEW}i6wnBLGbF1gqUz*WqVochw2kmJ>I1hE7&(cY
zwg=+y(iWQb_gvVfo~}CaNcx0XJ;x*3Getj-{9QyB59*BR3vF_F^!RMn1Kc?0&EC7p
zalZ-5egC_StM{0y8q0%2LkIKK!22-;G4wjgP-91m$gOWTs1u{9wLqtQ<M^6~=Q1<O
z_$LQbXAkqk?X=X07l>ay8;v%Sg}VEMPOj`hkP=j^aBO}pC9mC*w4&l6AfEq-g=Wqe
zklKmwuEdVsV0jh;=S0CWIwJ#flN5q<>iQcmpRCeASU6SC9Yb>bIhE9>ACDwb(Uwk<
z(Gt<rQLkh^uzskz3V9*lpvMvelK?kc7EnxVVbQ)GS=hJ`)F=Sx++C+^vwowCr<HM3
zbhHV#lU5h5!SGiE(&@|`4Z|aNTzeqphL8u0jg7H$^-!$Bm^|ZgS*&$uI+UeRNIqXF
zAqE;!t!5eMFXR=K<E0D*ZPK~s<{}0~PUR$hs9&;k^3<KL5XRf`QFDO>BDH|3DMb2A
zbjruBsGI!D8}C~^71bu^opA92i~)xF;Q&3Bh>(zt6OXW9+6t`_U?%~tjRViVzxC}(
zgABlb1uTrUwd=b9b<KDdkbfN;fpc+zffj1L9`8MY_4xDW&)ch`r}MqZ-LDS2Go+6#
zGxg^VM=L@E!h~}+9{LWday$9y&ks-qIGXF}HI2Z9|8s`Y-Xe4`=lbcs9(<oqKV>Ef
z1>H*g{X25rMl2y<a&dBoe*J0zc2)=yuCH{AkF&qLM>yYw;Vnx^NdY|-#EDShVWXtq
zb6RnICwNBS<2{iJ&@Ka&rDL>wT>s{)36F1FDlZk{2i~JE-JKJ-5aSO9_53tuw^h^9
zV9FoPqy&Py5ka>^m!H4i=JDx`XrD^^MhOWg&)=o_Uc?*9qT*t5VILb46BAooTT|1?
z&QG(C2{$UKQj#xKI4^s*onB!hi(-)rpYQnhzaRMX?x_|jQJyA%LYBM^=qCzs8rs%B
z;M&*D5=LK9dyBnI42Vw;YL?E-YYC{c#Fyc|&{G$+=^lPwt3fStg|h?CgfE4K`WXWx
z>1OPi<79zM5!+9j{8i#*VNuaXTEfq23~4;JL^rRq#28ZQyzFr_vxVOv1voS;vHFVj
zn4EyHioNc9K!q7>Oso6R3hQ3jZ1Gi!Q)XbG>d{gvuYLN%IiyvCFO5A@B8`k()|3S@
z1HeuM6zy=Xa&st&Gmb})C@d&@S5St(4#2ITUO$#)qeq*$2F-a7qp`yvdihvfeEg!q
zLV9LaP`^#jf%h+>C&<Lb$HjFnXLdyyPQuS|HNair?&sycff=l&{m}NIAtVGP$XM3g
zGfDnmzTs^I86}bv9&N`0!r`)v0DgQ_6g%Hb>>vz8h|g>PDvFB1<zmyS=cLAD{K?mt
z!}6<8%HJUk%VA?d=Q%eUJW)8ph1Jz3n<r5K?n8ff6xApjVDLT$TEak(CToU^?E~S2
zm7-$!ChKOV9Iv$8GB6K;2Db$wR4lB_%-n2jSfM>NH8lhB9m3BCf_MMB_soJXnVpms
zilqK#Oz@cVQNowZMWDsAe8Rn;zCMv!25bt0jxR3_gnUOPFl<5kT17Az$f6DP^`E%l
zu$hGu(g`B+si~@X9|@@=g69P`E)~DD8h2EipY(-CRMa%a#<RCin5cO1(g$ItzR7!-
z>=xL(1-o<ZFAtuP>azwiQB&vACm)0G<L#3p5?Dvk9@5g%`p62nk|4Sri~Snxe*0}U
zZt>GhCei!vLd{~?;n*r7^tklX$I|C9A@<jP$A{H;8UH-ka#SvE?u+yDaQS1wRv+#F
z3ZIw<aJGLC<a5UNO;rETWDlj>jO6AkWu!NK^4x8kMUxFN)YSYe0!aoE5fP#9u;7P+
z9Uq@aMYBqn-Y?(+i-0H$L5%X)Ls>a9FYhAm@NsFA*7%G^1k1q@9RtIskMnbn$vU3_
zZwJ5gZ^%#Ts8#k#*TF>qIG>9h;+cwn*i6ptsYO)2oeCoa@KtKtktkZ%hT!P-Xax5B
zII<DN3PF&o2Y$9`+k@M>YgB}PtIN?y%6444?;U&#%h>Mp1814Ig$12FDx@n2Y1Mjk
zG<%+by&gd&ccf@&wiEB(O>v1R*hZq9fz<tQ`u4$}6HiYaT{7X2_VI&G;CM63uTjWU
zy&AUY3BlvOI%2cnYOeVMAj;D6vSrg%TB~DDs*#mYnDwY>TAKsUu>AzcMNCvMzbyOk
zTFHo7C`Y|HEZ@u-cg77x@a@vCtuP$Aa6x|lG&jeqrcYE2B__H`N|16^&PP2vS6A0G
zqxCv#UYliPu7%;<AE0rAERT=N*5h4((V}AJc-d;$Yd|1D$mY+{!^iPKplq$UJ;AJ9
z7xdQqP|7|p^7Qd?r9|U<4D=q!5PrG@fUayj*k2I?rPFb#udl#PXxU>FyoRLO8*yxJ
z(ChLC`KD~aWE^)tfbYO0p1qUTaECRl@N(~32?vd<#hVs1VkDE(>>Hqb=rwzsw4uK6
z50Y|+&8YT$n*FFIgDs7X1<!^i8tFa)F#p|(mwn3a{<D<C!<Q<%d+K$gUpg!Xjhmd*
z^yV2xZ185D64>?yP%tnIeAexu)8p@Z7o5p}F_V{&bHX|7j%67pFu~I}2UX_)pW0jx
z5ndGPTam|?j;6jkn1>^-gGHCou9Ya9x!oDaYXxRKG0%ajV8M0kJsm7!&l4^{K^{Y&
zsUBJ&qwSVmP=+8NJTv_705LbmojaF;NsF{FbOimT%O`k@fK2l|&gk@U&dFFqCJ_}$
zM(d$MbTccMqq8KfHXm;Q2&}C3AD=l7!{2`|+9y&C6|nqFEn~zc+HgHKQKa}-Pm0*Z
zW>nm4o2{=Jd6qYuZ7CroWyi=8sIUmBq8V0IR5<V0>(y)x?kzXf&GyIeG7e-lehgAC
z9vyis?C$Ey+3Izz)VeQBwn~!cF<W$-YbM=2NB2EA!orzk6D$A@9AHBdFUNyTEUc}p
zL{<sF>VV@R?qZ{*Etoha4&$=K51>&%(rSUVv7tO1-=Yd8=I4+r^H64NWi0~1zL#nt
z&p-JvLm?DyOlE?38iKX5<}l^-g)~3`Xt+1#vRQ_KUw_gylHD008qhmEGTjGT+GrK2
zkv6JTdqLdN_KX<QcA1jOv@dZS!lI9dVvuWIuC*z|6_@I(t6z9NsD66uQ{|2?ISBS1
zYKTip-qa!J=C#WD^=;yXScTm*b5!Pz=)hpJbF!!393a0tqmhUin^EDbRpI+dBaGkl
zBP(lteck_XrS9h5)0Q)(<pLKQDL~$O^V9L13h<;+cY+QVZpNb~_s+lZ%w!cP`V|u1
zySg3w+pF~oFfyVZ035Z8HpWZG%KC)u)$I-@D4C!e6D6gnx_XUIR8+5wqT;{JtjWm>
z05?xhO-V=J?_~G_;Jvg2umRb<fh85tzslPlCnhF{gnd4V6KeywPLSevDsFug33e%F
zxvYhK#rTYXzQ9HW1%>-99em@_aJ^Y;EBXlofsGqS0O=_zD*lOU<N3Z&s%*r|P-Lf}
zzFLa6Gq}3Iv5SV=9@6e*Y2~tUw%|DT={)SCFtF=(JD-8Hg0OI#lvH|L+%q>Im2nXv
zP5Ivat(z&;jUeVdZ0kFFIwPDv`IA##p5cD;PE0HS<TECo(^h!ZDFkDP{NagAgkFsJ
zhm+mA*80wkv+~PP0U!xBoT<^_h7)lY?Fp-e<R9nog}qCX5|$0^=K_#*GCDeX44C%t
zrrEnur$lexzIAoIpjXWYs}CdunDy$%a<jQK#k~L#aoe~0eyw#oNd3WuvERhF7Q^Z7
zi_Z26Z3OY?Ex_af!E9-6PENE=rS3mIJ_g&n0rVmuY8CuG+N$thMFjkokNwEE^Ye2I
z)E6Bkp2|8C4@YN5VH-Cfup9C9>yOgXi__EF$S0IADOF9)r`{x~IG}0&L{CIkH5ZZe
zjQFEDn#zwuLehM%PIP`6u=w1Xd58#qOh!rRh|L~GJ`Lklk2mlQz(yBAL@aoAlkTrg
zU2i`(3LMN|P+P3-1nT|X*$EB~CMF~VmJqjQYuG46)lZ9UnOkwGSbCi#5z^+JF$s~V
zX{D+2r|a1vvgp?MV|D<H{yG7nbIVd$Yxm*!a6E_n-X|eGe&H@Gc>aun8Siij;i)h(
zL2$P}CsJxM`I|&GaK2d*7QySTq->(7Jwa3!)Y88YO*V(h8)bl@6y)Uo0I@?>R#sa(
zsjluy^OX1%k_iY}x_?4|i@*%VpeDh%p>{vT6#YP?4!m(;bY}52t(K464=S-qe!i5i
z{wAzyfDihb*|O;6Mqh|mKENvEewcc?tgoR{q^aHe-S_+N4~)dvM9EGs+=nXBZ%n;H
zyjN^dFfqx!*a$!<1n~+ZjO5MxUy(t5uRTw<&i6!Q{ImZz3or?0wrYW*`o{$yegJ$Q
z``QWMVe5UEAMP96i<;<f!eQTPY0Zg!BN6a~g=|)%>({R&L-7L<%OH4|=<|j8OE8L#
zxR60ZS2Xwc@`C)(|Gc|8vc1k`rKbm*qn>}0$dL-OGD%yvN0W)Br>B2SR-PqMoQvB<
zB~im`{l(q7hx^*MHI+tQidWsvUES_2M$^eH1yB<Whi9P|nvkDsaWw`aZ}yf4aF~Ip
z2%?`ReG2f1Zy~?9K~%+z+sMUyWjf&Yd5w+ZNMyq_Xai<xNf*PcT}6$n%*N)&n1J0G
z?z7wU7`EyKQslP0HUaZz5cJai@8mcMztdI%$npPYlSogUI8Ds<QoY-XI|L=7TqlW9
z26rW`+1O{%R4C)jHOte}(Xma;no|q&hIxw=(;-44SVJXoouxu%X5)JFqP`-EX#&p%
zHKEXF>G~t~+D``l;(MDD-)X%cZZCuhV5p>EAc~(~zkZbp$@GES-$_z8v9Up$Efc2&
z8x%fph2-bcY;6a8qsPT<*4zi_`vAo8wG}cpv;o=GlsVbAZ_NSx$WW8p)&-{G!^1;4
z>Qq*hiA>52z<e5Zw9+0lQ7gFLUVf=T#<pCRjgrf8-Ee9wcKUSgqF^D7V_gW`oUR!2
zwGEU+SO&&7@5xn6tgV}XYz?BJz4>|mpQ(OrQ($vrQ#*kYMkEywZ+nCsKFnT%PU(&^
z*#Ew2>>N$22?i2Tc*yX5w3MycvG<FQ$KTp+p`S`fH=CH*_KNr+3HICIu7V^7+s3fh
zyP!=pgf6gQ=l7p%QH@eu_$XC<(|56>4Q?=nWw20mhZ<jvMtv>SREjzG6IGW<Wo9kD
z7E`2j>lpls%bX3*)OLSa;cut5lm-EIcE{ml?kk6?HX>i3c)7k417Fe{3|0GnvG&Re
zPiOF_0DybSh|k=p!hArr#=KaSXRPUccbRs1yx#E&6{b&-RsWPT->?Y?BCcMv$As{Z
zuEyiOraGyTxIL02@0aYVk9%)Qwd7gTtytCVyf9K^#38>?4a<+)wIISRA0CdO-H7FI
ze^Aw)-)#$wiu3V_ytfG~@FY)O(*siBTMm53Gs`GnI)f(A(i9QDZ4C=^PwUnEX1zG|
z1mvn~vf!3A8I!ItQ*H-?^o<r_;s$3X_`?F|%w4%vASzQ5x&tsqu>1%J0o;6g1nt7`
z(X}igDa!D!mcfZ1ij>dq?9u<d0yB4H=t>HT7Q`_?&iL{E2K3p-TFz^v4EGd+mXF)~
z=44trqEb57`C%+9s}}k-Kz>FaugokG_=@0rf70hgWf)5nH#awjPR#SQ#{?5I;@`%b
zBQ_BIcXl2TQ&tZwP{POG0tr_h`zQ39DXYrWXiy*{OiZ3Q&B^gaCbm@GckyPF(zGmn
zp?wz{g*+w9ojYf3UNu^;_`*TQK4VRS?cV6IW08wzvh)}N3-*QnJ)+D1N7P$JRrP&g
zqX%iEL_kvb(cPs;hk$f<h;(<Cl#<d7f`EV^(jkp>iiC7XrwEE5aVP)xzV|Zx;21g%
z`|Q2fT5~=%-CLyj!RC0ax{;ZByz~wn{!Af1-sf99@kYD>{hGS^+WtNpJ_bI8&Xbm*
zn5K9+I)u1u<FEPyB8P<myuL}9r%&;)MX+kNBplewb+!F}T;Qc>ZnZX)0*nv&35A8g
zOE$obJaS-48v+;&GB-4rdXN0c2L}fsWn}rQ_smmf=ToprxNbffpA7pXhUaz2^$D{`
zBP;GtaEQ}HgtlhP`QE8uiQgIRZH&k|^a5~_0hs_=_eNLqI764VLicYqC|vxNHUHUu
z{krsIs_n^?W0|4AL1;t|)zcN##H=h(dXkrxs6&W{o*v{hEWMg>81g<+yAw$OK_bT$
z+&}>d2%sfO6XZ?Mbn9uQAjE(D{pCqXG|EADMT-{xfYUhm=~HohRV~zj(CX@HDzUNS
z>_={0=(r({t;$EsZc^0F9v&6*DVdqRd(-UT69LKN5kcsCXsf;^lL*B<9bA8Vnz8*I
z_St+N)VPeBzNG+kK9_nVM9?9JpZIrjBvfhq1%qrYyL%05^^Oh>bsP%@dtcOaa;BQq
z)D{dVN_1r?_L-|Od4;{suI>$4QNDWfCiv)xQ;EY1@CTn?OaDYwY8d8~bHZd}SHIu;
zcHsyh{w_$EQ6Jz;8AcPcKVC7!lfCr*wY`dst20&o`+O}?#}yS3oOBoQJ!!X1J9U{H
zGt2c+o(ga6N>B6yI3GRS+`c)ATyt{VfKpS!q<fh#*_ql`ypXQy$gt0?8WFtdnldt9
zl37>yn)8x>g70d5;J#=c-qvVnJl^`oxq?F3_k?P){uk9B6o|Y9P=HJm1IZiP>Z_=z
zT^=VSY~0|dD!q&C?Y_Tl1JEvfP$W*d2#l2#gCBdWNLpv*Bh4+|fLKRFB`?m8^gLej
z4|*25{q@VLA9;t1MxWi{kiS-O*-{`n8eg+IQ@H*@Ads!Qe*7W)brloNk)r=bOH=dq
zt+KM(a+R%E@`7zCFsojC`PzABci(DkaI=Kw)pULsa;dhoHrl^GzrCW3^w^rUgyowa
zs1B+U@<aLJM#GD+Rx`A2vv~LV^?;LvgfCl&s)QlOgQzHR7Nmz%G<nUfGY2AcbSMnU
zXjTub7u;1=*pQV35eQQNQ0mDAUZ7Z=9xncK(+OvZrG5qKEyy={98ZcKm|1oI0!*+_
z(N7+RU`rP(P^xP_znmZTI$?34c(>yz8K{!~_I<*5s@{40#+NT&dd+(!=l^_in<+L9
zyu}vRq+W)GuFfoXOcIju8<Pv&rKfmx_%;mrD|?-OPD+Y}jm?1ehwo;B>704UO2}~^
z$OL*n|LF*mb?@OA7Y(|;%vMDai$e|W`{!x$P(b>#owUd9;L(SfN|T_9Pu>wupP@=h
zU39dy#nRJT*}r2;RD#nyn^jb_0ghI8UkuzxWSQ_IvY;0h8B=c4JF8`RVFjZ^X}N4{
z%_b~Gh>i)YcQ%h7m3nN>i#^3|*bK<Gbd&)k7rj^++3LYtbj2)@cQlxh85yK6EMAr}
z_rEX-FL}pYAjPyQ?#{%_Yy&$UPEijB&DZ$Y)^wm127MjN;+EAIEH^l>ZV(Y>h#O_+
z;?AGw0!uT6@=5W}G?6;u#jZ(2&AO%x1gSaCzAW6Zjeyz>4-6DqEIG6JSFlpcSSM_?
zbahEcNW57@-lY|t14~~~Uhb((B4#t~&PDuNko!<XYRJy)lu4aIOjFgp`BP&#YRiL2
z5%Cw3GvE8`%>vjTVsvwjmm`<_v<=P+yx7@TA81Io?;kmL*uY{(3#FzK|Db6blPP52
z2hCc0)^Yi=owp0`Zzbu_UJy(mAs)s@n(3mdL=}|CCo$xJwf2)};u;`Lq?2pOJSpB8
zkaGbf<x(Yygj%xJ*~6dzd`)Z2ZkEhNK0`Wz%zT1&@*zv?D8LF0D))cdW0W$7a)w_Y
z_fa9<s+Sh3?=afs+N%97S7&A@UFj;2LY3j%IMR;$&O;h1BYPDayYMJw-$5IfmiAqG
zI!T_|yL=CWSAnf{LCe9xQKR+uqI+l(`Vq?-{Ksxb=GG}uXk_p-P+8eY8GX|n`X_BU
zLXi-$<!pc2{q<hAt_rn*;&~t;0Wl7NFPZ#ZYHjgiucM-)JG%(@Gz?k-AWY^o-FaJC
zU!M#m27xH~p_w|)l`*VhNe*2lV8d>%u2>itOz(S(LcIUe*&n?hNto8pM1~+Cnbzv3
zGJC!t)4}L|{(f(NKimCc5-j973^8>L4M8^_3knLrUk+{~95TLsQ1gYNr+ynjk*Mui
z`K9;eC&U#grE{5Se%J&D-Q}s1bf%?9?8?#n(&#G{<Q5Q`3%mT$-X%{gK9g><hi+;k
zF4|1@fBFDN!>sC96@u5&Moj!I{z#7=Isb{NtgLLKoxlW`Ovsl6>+Xu7+&St3C{niC
zeTJqP*ezddjbsT{PCSiy^$G(QcXVK&^z`B2`{JD%8TvqZo1FXVam2(!OmhhX*;W?E
z!y_AS-Wdtzi@%?Zh)`v+pZU0mK8uD>gv({tX)F!#$}$NIIN6N-*e(7hKK~gDf5a*_
zyK*L<O~R`^M@v`H|IXEn;_(JC2p-Muv*yE|g?O`>>f)HIWBy}tVd2)cLPd2OUjTn>
zc_J#C3)MBO_bedyX1U4D%F)qLT)boJ;=K;uqLF$&(=MRN5L#eAS@IPgps3!KrukBf
z&u3vzRHf*Z4!x$WF|9Rcc*I6&iD-_8&c|Kva^}hTtl#!LV>zeuku$6{F?^G+j*P$^
z*J2{EV%eo$*{cof{Wr*bnUavO${Y0Ga5@gcS&&VxpHkG{n?GEOKjNfGhiE%73az>D
zr9b~bP0#}R>78H>WnA=vcSDYSlPsI>(D$F71=HD}NzE2iRBH+Qy(IZoFM_DGn>noU
zlNNn$X(<qN?H}+o-#{o1k6fUfhLssa;MB~R_6OoD2g8-?ws9U_U8W;w2n6F6Pz>Nn
zZ*l#1E?#a}McYo@_(@md8^Tn2MM|dEd_*x@lttv@)1%Y^Yu`kA`tQH@^N(J=ak{4<
z^rq5sxoIz>@yp?;NB#Bv*?b+bk=U3*@6q4SGTZE46EZ#i<DHW8gM@tktpsu_TJ$6u
zNUsB~hd8Y`B0OKzY!2rb479#k|Ln9dF)`+G_?HEbZo<I_bw)vf|J@fy`)MKGBncYY
zwpFFp7d5kfj4EWpZe*XnYh(&c`_dDklw!Ec7xi*w=gTx1wgE^Fywv%{1;7SaX+z&<
zIJwP@k;L;2^Qi~=GG-0(p6neRUa)X37rejO@(s9sXI>t1l=`!07JuLg*aBGLKHep&
zN47uj1RQ>L4h}}1@)qznz!(8BciS9x@$%vf5~EWxVPd|b)wxvsbGS;0a+x=*B#h;)
z)#B|6`UtU`%cSBR5E6PEEnV4B87COCX!+xQFgfs+g5O8Rv|7TAeG=Oj6EM=|c`r#G
zsrxx<fzCaZi>b=bYu7(MD&tn{+ao$|)|=X?Wc+{eDjQ%%>Q{hiVg*XqOhG46`|n&8
zUUXtEYhR$AgEx!|5LC{O@m#}cr7YNtt0m1{U0uV&ih_cI#L-&yY=40E4poWl_`<BJ
zyYkuQ3XB}0XU?cfRLo3ssf@~6Wr*Mqv)0&fv5YK%$zC#@D&2iiS3));OcdN9VzI@R
z9Xr;~<oYK96(y8TOv%pHBn#tI>Kdl%m|OeuqfMrY3`H?A6B6CLSq)8T*Te`6Y>hW`
zh;f8%)WmPjGFa?m@?%c~yuZhMS%-KFJFhKhXwR81kE=Sni3c=6Q~T|3G{KgE;PsOy
zPekS9A-(6VgI@<17dK8fY2KwTYf*{VuP;jGNtRYG-m_@kp~!n-NwB)x@@*+1E_xA^
zXmAfckQh9FPUFoF`cD!?>$xWk48>8x()d!L1h|Df=~b2I|LS>v6d-kLiYgU~DbrBq
zfCSa{aOU&OK#icrx8kQg9~*DHzcdYryuU8nGtGVCbLR>zxE}=#8y636=5W+$sRdIi
zGPED1*<+cA^!I>+Tq5jV-shzkGyu@T%K^01grnvw;yU|Tf{(^%b!1D$UR~CrH)0#D
zJ+`h&%En2Jg++Qzv$l1QLerfr?(G~_FZ532_JEPlH;w26@u%~vxQDk}4r)9Tar6;H
zuXOGjdU<(yKg8^jp=J(`TwPjP(iiT=8X6r19xGRu?c=o+?eZNy8^Uv}G!CPv<{+n+
zDX-rdVj+?<n#%n+2M=$M+W`gvZww29DP^#vq(s8g6N>fbIOrg-5(zqsIKK1*8V0_N
zf+HQBv*4}Fv;N_I2Z!@xG=(D9)9tS_`;_^$?u79*uu`i(gX>H9uu^Je3O}`?@w0`-
z)x{j!^H}R=An^3Eu#B)bGza^vTX#hDe9hFZgU?#%8V#llQZQxGc~aAk^L_I}zdb{h
zY#}6GD}abbGMDw5UQ86qn><+nLiO{Q34&d}tNHiS_t%Eg_m=wmx94`Xpu0l{Vbs=G
zU3K-upVxt%px3w~;1trUWV%r{#F{$u)>uYPDP+@mlEk+?m6U>F7+ZvYvPLw{j4-&_
zLfFD!ne$IFXJq}RGsqiJ(a>^7Mh~p)AVA3WU|KO*p$Jy;qtxtOD>HDYSI2XHFaep@
z-*=4TP6$ND+>}dJ=pe^5<VOA7yrd8Q@h0QZFW2pev$kQ|H*<WE?-b9AMtC)ME1J$D
zv7-1`jrOYa+I7=^3lda|vXpXLjb;ZN2!|x`$|?9=R_%bp-{i^Zm)G}I;wkR-_P$m3
zFlfK{x!SEIMj6X8x=JEo|9vXMEo2~Ut)LD4&n4)3=CI0HPOqEZ6ItU%-j)<BC`m&`
zd15wO8~Nu)G3o0THN^Dv^cBXf`ZNi(Hs2yfELjPopitJn;pBC5I}ri1=GT!?pC?NI
zzcvI0B4-xo+mq781w9V!>k#MhTMxKhfxpL>bHZMN-s0l?9I1mm#rK)xb;3&;dL`sw
z#>lIzqKrJ%zfEgKpw_kPnk+T*|L=vxb^Yd96h(5e&G`PP>4WIW5(_ptoR<W+>qD=S
z=?+uIJ!OwFKU=0KEUO|AAt51tyTW?E4q60IEA|$0$6j1dx*g?`Hvs+*r!7P=P*PF?
zL@apsZuIVJ4K`vBPyR<wzwz&=y|P|13SQh@*Od+&dFJHQI4&Xx^GLp;2FGw?lRdY$
zUw5nD4?%RBy;tbC6pfw-nbFZhPVH9_fdX3k{{EMhTbx~Uu?<YS(*-^{Sq(yOi<OCq
zKUk9?$Xp(^>!ttZB1pbjgdoe?iZeT#f|s`jiz7Z05*;#7qu+Yi{oTs@DyH(ax74~1
z4*W_EJq)ERN$<a58_#}ID*MpT(gG2T1`1y#$+SRQ{M(~vd}y3TwoLEr=+i?nL}sz~
zC4bcb9+e24ks%nWY@D3Ln#}**4+w(HgrI$o{U}@qgz55$&Sn0)bj3~8k;mQZg}MKA
z-jY>}ULSuyrhw8)c>W6A*)1~Q^O@()`q5Ekg-Du8g|?XQsT?&vNT~gP^0fM#F#Pvl
zJ{&T0gV>NkC~~}nOu~=$-1zLXj|0>o`m+v?+PvlISy*<VXM&Y46@krtS#L{*A_2^i
z`AX>172r#$xkE|};!&U}eRAkTY-tpJ>FT9Q4^)Yoni@F5rl+U%7aXS2dMqt1d%t|~
z_Vc?9ZQ%Y?_V=$_<=eL%4%^P@?HJD3+HBE}PSaMU|2?`bR^W{ljqV@W49o30IYEdo
zb{H%b^^H-CjvYBrySOVeG4XA<CSomDbiLwa+IO_@SnWdH-a*GkYiu@RV=81S^PMx!
z8yQ53hnghtdR4Ge+n}q{lxwcu;orb#3JQw0`b^1l2)XGTqr$lJggjT&U3F`xHw~z~
zf}t)ME?+h}MzLH`qm$1yvet98%=?RY%O2<)4|b<%k@SxndD%uDW4Wi>TBk!RWm%yD
znsD3n4I>%w6phz1zmi5E*bRUvBi&4T?T^hKEh8FUtfxZT82M-I+^1th#M6_??RYK<
zhg?*bvjM<2BKop{*G|U~v4KMOCeAdOX3LDALIyTrr858#ypa9T-?*p-%wdV})S}`w
ze{*q13@ajn(a*>Q>~)($$r^9O8c3ajxMqT$9Y0Is=zR5cWfyI+IZtk^gt<GaRAIxl
z^kq6PUZWx6=YK!j_r8ZW3u#N$hiHEN9B$4+RHB7{ej1%(cCR}`R_G0$84u~!gr4kk
z3dEZ@Wjy=rKb)EE`5K=avjt(B4Z68_1py7S{t9*aN@~fL(ybsp0AM9Pm6HG~{6za>
zxgiO>uB2@~tU`|(#(Y1+#ve5-V^n8LWHnn2-#|8&3>5`M^RN-k@yn}wJ0En~b*lh*
z^InZ2y1+#TyC*e2>aJg~0xFN*BT5q-=>o|nZXN%!lJr6wgmw4UNHw2b-`0*WM_~aQ
zJA1tHmCw;!3ZK%AkWE2FMMWE}Bn#FE6h>uMxabL~U7|rzZ{(PQ_b+fz#ZHYI%tC`K
z6?n;65hU)LU2*eXbsjlwmyalb$ZF<y;^rd3Ke<TTPJ>{=$w3K^3eWO2Ka+DeAx)Mk
z*dCP&t%4POZ9DxzC~J#bxh5*dKX(_EcUWm|x<C^-ML<v&w%LgM8w!b90N%o=F&-ZA
z=6xdU!D=50^nSpkkk7E3oP61Ae_p&<0X$dt!(}JPI4n_3X429c769@Md}Fj}L!UkN
z1b~fjt#6@E3EAALPN>pksTh>Fw+K!W5KqmQYm@)_-iux;Yaq%wK3DMW^|qUb5I)LR
z)b6Vo@mf2u;OYrhmNJ(~w1>b34w{94AWsr)^<f+lnU(-##F(3F02}*m#w%HG{zk7O
zW6)vhwE6D@@t_zkKG+vl?cv%S%_09Ygw4g>)Ew|3GV<x8npR?uqk8*22J3B{2cvjK
z2KiX%ui8lL>l5CpqFo72r90S6M~r>*jTA$XXzLdjW57S~{j*QHmG`vo;*?E|rM1DS
zyUUT`0g)dL=|@s?YIOo=aZGubM@)(9_G-7c^2Sb&!p(JGcq;3lqa$=bnIf_2s{ZCu
zc=oKDgl)Xs?Bw7tl+&&RaZqs92mYdi?dmF5>|lSt3<LmK1n*THhez<3S<b1a#=Q(N
zejP|+_6gqV=R4&v5Pimd4%=(<W33Mnr*rLA3*W2C`|yNT%}B&WSB8Vt!eS3PeD3;j
zm`IwYPxfB_p3KuE+&CMmh2u<H1Up20JVAYVbP(n4vqsON4NkGN_JA7DVe{K*|Nb%U
z3pJaKjg6?Ns3C{Wa@Dp;tzi|n;EkagCMp7P@V7;eX-XU)IXJZ^@V<Vn^}T|^u`l!y
z?IY{F8JXnl-cKa@T_n6oaK(`gy^me@222sRNqGZ|W0nUCi<0bOg5x(6E23t?4R&_g
zM*wT1=RBGXSS+Iy^S^1DBkI?A<a=5&R{7002gl>|C?_}f<B{{}PMsk)!+}FzSn`_!
z0<5>8!Bo3z-o3LDN3#)diKUtsu}Bi9%8axSvW$j`Do;lM+Cr+JOEouh8=D>Y9YyM;
zA3p2>CG6p1xbF7X{~83yppw}Yw5c;2`#7xrI$N{yrj3=AKckTOPB(XOe(rB{Y3XDU
zI#$-9?tP;SbU|J<5C}Cj`ApHmvr@(>!_lI>bp456zKB7BHlj}<<ZK!uz!5kl^c8Dk
zZhzr!Pf@0fR}M;xUMlg34~0l1s`$54V2;mvdqWJXJja8p$D~ls(RN;h{Q$od@X0{w
zj1CR4{u>}7Bbx&qIaKXYJ!_Mbnnn~4m?NvIj-j7sP|0n%(bnxgYF4qI>nH59Hvf8b
zp7?5*4bhmMuHN9d7#4<-NQd+mvLc_F^ew53HD)LON@sTlxb=*bIxl4PsM&#=By;94
zi=aX^PTp^xTlC!Dg&AjDtG|7J&AV%$6Tcb<(FmZrabm>A2eZZA+5fh%?%w@PdGF<R
ztg@^Sz%Tk>6+jpp8w1hz=g%L+8^~TRkhzP1-w_=jU#amvjsrI~8IN{K`^;(9q8$lQ
z7yY5dD({}_IXK*OQ1UdUT~=j{gnn&b-wm;bwg5EqSz|e3ftNy&-Lt&zC7m2n;(<rY
zg3iCON{*1s8lE0Z8$5G~D)2m73b?rr;N+}?yRY4CpQzjP{F$-DELJk^oomBqHN{W5
zGkX%#($MQ~lzr;!IoQ|;ZERi1KV4p4LWT|)hJbqqR%ct7qVUhS{Awliz(6hzfwl9`
z!94hAGTX)=S_9$*udlD6Vz*mv3&;}<K$t>gABg?I_Y3vHH(!!|^z8zgseF;sEX@wC
z0;vbG@y17I`G@=K-e#&!6x|=`ihAp<1?{HG<IkMIln73`^S^&j59S|8iIn5W!T|uH
zKxnK^|60`Bd@J6;92kGk85Vhym{#X=V!24U7b|%$J=vE{ScjMW)K4R+M;VtrHmq6r
zdC|t^roQJ9l2oLWy!;qrJP3W9{yJsDOPqH&hrBx-nlsN4NA>UOY<W+(fGnO)zLtR8
zov*8N>x4fg^92uS-pxlr382xy=;Z+sMzA120yz>3RWKyFcYkvARnva?`n9}^iwirs
zabWw}67-~?z6t&D7nTepfU71%{@AfusA1%mXC8UZ?E6RUvrmU#tQu3)^ZO`qW(d>4
z4d@{PE~B!=f&x#Elb~=KHNlE`mHx=hl4$JJ(8ijO@!dg;HWkkBd;insKos=k;XgU#
zK9>Z!Qwj2%c)%-hipTDt-JkpCh?!j{hn|24&Mt1)Tz486N>k%QxZlIFw3Fr69ISlo
z^^!VW(UPC=y36HqBR>>cID)7!WUqsZwQssalKHBon%Y$pOiJJ0^HwkMOPn6L9k0oZ
z^w+jcW-7dk5UbY0aZhxH_#N4&PXU+~d;HjU{H;VL`H!_Vu(zV)kimIaT#bojvGdCF
zS(Y7%+?1z7zKy<{ED$_FBUx?>?dVF~pJsWVNOJ1y>L%alZl1i)CQl2dW4^DUmnV_j
zM1OfU(8)F2s&f=>cJ^ds<m_py{(~ncKdd+b&<Cg*>MXw}dQH--Do(TVQ)G3&V&2|4
zVgOz)I-W`*%kxb-{)*vN=)rCJI*XQ8bJaJ5RkT8?<yJAP7GnHOx3P`+ROVQPoPFN+
z`F0jcl^(b*r3{{;{l>>4soKhAck}#c#qfrQSHDWBP;p*ei_F9cYO)KVR*cJceJ#`a
zAa?yU;$-pvY61S5L#`TlDq$_d!v|Q15(2lOCjwsY2Y9!(w@adb1kyt6jqu|+H&V!k
z^7s@T8_UkP+98)1@cYThh8se4#v|!wJkm1AmuC?~lTdcv`UlM&#u>e=Ei6!-COSqa
zqoQi3lu=9hVPn(&<bB*W?iNaM@PO0k;@SvoOACa*33SU>H;cijT>gbW{DEsvOU!<x
zIw>NZ=M^cjrMZ$!Sf@!=NxpWuZl_VxFMr3S;XenU8?HQZelv&6@AQFtFoM|4&F%fa
zfS~@{=kv-ft--j>*7oAHzy{QCBaqjXM9|kShmNgL>sV?34XNxWX-;0A7zA_LxrTG1
zv1*rts-@&|0MU__*^tuY@<X<x`FHSru{!H|@8jpl?I^~=vcbz|5-0Tl04siPy6fy8
zc;*Oo?7y|!ulw!AwPd=;187<fPeRdcEt7RrG36_U2j=rk#F0KZ+qaYsK<^+ct(7*T
zHBX<i8<yEX$~M`egv>5d;={(q7PxfT7ccDfoJ7E3J`T`qNPhQqJ3V^jkjeesKB`mo
za$M&?d+M?Sl<T1X#|nQrVyWYzL-aK9MaPkysS!RG$viHVz1x9490Fwr+x(LNBk^na
zgW`L#Sc#ndu&JcPx@+m`69<N~|HkwPWv+-Fv^eK@%<Sx_KY}WaqZL7{!V=iXN?%mN
zTg|I!i`Zr$WNWB$mXNZ;P&dyHdnXIoKU(0?8Jwj)`La5dGj}IAk#4K!{VdkuIeC`8
zo?dIa5(E%IGy-TOKSP$xv6PM#u#nIwnA-gQ=un;|T(<H{26kBFP=U^ZBJr<jscf%A
z+<zu9+ibkBvI<SQzaoSP{_F?(y9f7i1qB-%*A_nHTZN4L3AnC3C|_oAefxyJNr?4K
z+Uy=emxrW%UjH`_f#xHBp!YCS1S*k|6*iVy{Qz0GwS!05kNID_%LBvFUvi>D&P-=K
z_E33eRs=rRJ_~$)_E|ht9;YWI?9gO<e;fU&u4{7gR5&w?@!p$MtoJG@-}Z67zIy#S
zq<AD`1c~DT8UO3p*r~NqBSXW-FD%3--Q*O)>~vUo{3Y>ktxRXW)60jZexPNHn%ol7
zF`N((J%`Ig<C1&fUzvM<PyoM>Se%xEzXJkh=C@jY*VGu-xZjgu-&trn>6<SsDq3t0
zTCQ_6icj|5o0RT24ytx_ENbOm{S@ivgAlKh^nrdiI2bYV2ck-|+V`wGe_TAC`~AoT
z+j4kZ?I!>oz@x&UUvYZ04-fumx!ok2MA)>a8ZudI1Y73%y1S8I3P`Aky_I|AzbpD<
z=o=eLT&Jl$9CGl-waNc`>Ig81L_UPkJ-^P*D*ID+6sAsN_UDTT5G6yFsI&TZ)A`gG
z>UZCUy+iBA`Yd8E&WX^S{0JyhnwYW>^!qc)B*OV>Qs3T7+C96A5W1h0K0P<nqV)dX
zUf_&<lD_6z`OaND1Oj?wW|=19CyuYx<)=NS)YicYv;D-4n9ueO=ew6yV`ZiTZ}478
z7l0<hPNaTraBlA6v^*oo(K!Cc-(;D)b>1Q}-4M0M#ijcB-=vVXDxVnZ`}D$el0QjJ
z1B%bw>djQ}W%1WPybbGie%-Uy(IGy2VwIrYq{|_=;hbx^CFCo3<2?jTo5SAp?AO0H
zDa5F3o7v`l64sM%825c>zJ!fz3OMl}wl<LgK7EqRtYM!gTa=cJb~sn@PUCyM138=F
z9h&>9OxFvI#jcp>zv{r72zSsg;OQ~ZugUe8nEmJ;dc?|rnNv-}@w(=o=6@&>Wk?D_
z*lLh>0!J!1^{c-0*6suwcXA((yyWCt=Zej&|1RXX&^P}X!+RQ=0HKh=LT3G6^c3q3
zd-*JaZ8{qLjChb9Xa@lv9^ae)Ha9mJm9s&OPLqJq=fR&6c`HsuLSS@>6PwaTZ2G=_
zF*jdsUq^;cV#lEW++jcxnH)o%JU^^^1;S=3NChoj?;nY%&}7C?At8`=KR<j^HkK`X
zHt>5}T7QraDO#1e$Mw8&#&Ju*0*8G#q!V@Ij-+2?OXQhozrdh%K+6>}+&vt<DwH{|
z;J)~XYaxaAeUi4>1c~~IIew~!7>GE<O{zZv*1c#iC!d%6OhbYV24I*?kPD>gVeDmP
zIC`HZhnROi*`tb&8*-X;?0ktz#Yi#yeP;jee<Y<0>}=+lb-Zq#mRuyzf84qShrSao
zEUf;_FC76(Ljk}EHgkWpYrx;v-`@{I75o>d8efLD+osN`z26R?BD|hR{YS7f=2SD0
z!h&rYlQHm|^)82T05W;WPaP#S;U!hLC_rDh2*bm}%|E+SOmwu;Jr=Fe`gUPiEXIFg
z<FT@>@$7T);#DB&qhCKFKVAELY4aem&U7zhJ)v6p>lBU+*_;r)R{PkEfu8;xSILo|
z8$H4LJiUq&kMi;@VWh0Bt*yJeJ1F4FCU)1{pyY#}+BSp2O?CCtxrIvrL<xi0ld5J%
zo8JXj9=(P3_EEC)HkHyh7z5D+-%ppSvh$&1z``#7tk>Uv&C1N-J<&Jcm!Ka|5sh&B
z!E>UAf+%Up>Tln}yvd1_!2c4AgbdQL>6W+<Wg`S4DhkRxVds?&Jc0w@H*Xq`5PVKZ
z1~)?O)~h-xIkKWpE3`rTsi&`BP|Qmur^>`wWF?w5fYdRj+t}GVm>e?aN!u@oL{H%>
z`|TEy-Loos|LU&cF6_;ik&6qxwhM8KaB7~P&JfAFazt<ea};6prp%MWuspZj`f`_l
zIpAy&ojgjKrhg#tEHCNl*8JR9yJvttZ5~9Ir#ZN-_utd4B_zS#{_b<MB&OD-<^c!`
zfG_P<mPk$$9F_XBwKf_7PhT*_Js=zz7*0@4M)alW8^rL*fOTY@9I#XoKS)FRHb2=z
zInN?j-4pW8W$h(J>7-kSsm_ebOmD{|i*mobtm~bXzxVAT(|N5V(f?>`OyY&q2zv!e
z;@_NC_5V1o$E6bR=&gLq(D|NQF&!I`HGL(@M&C0qu(OEEjm?CG5U;pcS(V#BK}B86
zdO`$Lpn5aZg+LMS2-92*F%H-`)o?sWyCcr~&?~Yv_I`Bf?V~K!W?CKR#r6|)N>joQ
zeX?L58T)Xi{$Pti#MJfC*{Ryv0Xa3sIbJ|_*P)+5CNcuK?;z|YFwgNiIVdP7;LH38
zmh$e0Pkg<SO7)-~n?-|kk@jZmhu-{QRw_YrpNt=DmC*7jIf1zejvat3!2rrDC^+LK
zeFT#`KuL%|#GCxE<k;(~AQ+fz(3`d8NA51xOCSBjW&Sq25)cFl%*b3HZ|}=hO{dNr
zbVZ6F8^<e+<+eOZOD6W!hyMVSEY48$$&;{s-<(6bPw#hKOZ;<*Fl}>w=6r!n=Os1l
zct@Y@bkgNeShUiHPhklrFU|v6t?P6I-Dmu{Te##Z<C$qXm?i4@c+odGL&E;TP7HX>
zws3)01k^Q-BP`q3B4Nun*Z+=wQ^tuh+xYruAAY>NL)MWnu&`q~pWXM>OSMkI%xo0^
zdx-b$2tkHZy5z_izff|#S~nB|66F(L0J{8X?DAK4G}d|NQ}OJlZjBRxNXa>&n}48+
zMMXykvsXODqv}g{37UkkhzOnxMQ8!XJp6I)-W>w|<a^aSj?@qS<xxY=t%lQG#w+o3
zb?+chMh8WLl(RauQZ0#0U>SRRH#RhQ0Wfk(sivjnt7o4r`DjO}_ioc2#AR$4`9XOM
z&Ss{Ok_oXeXl?fv{a$7~x@<aGEsllL7`2vdIym2=ip)B!o0`vdvLxH@^#aA=YbolK
zaY@>pj@9JZn@UP=@NN0}YOmw_i?)}$%`<|C^@YJ}l<d8({;zmm_TT)&NUNcYqiFHC
zkA^Dln%JHa-qw8%B9z-7AgxvU?nYT!SPn7i%dObim!n0A8u{XuqEzYn=a@>?5F-!)
z|CXCQhJG$rVBBF^<?;uDU+^y%LsCdnc;PbsobKKG|56k-(DzK3qUQdrGheph0=dH?
zB;*Z&Vp?KvlyOLT=)8tfBUy<dk9f?WReyPv`e?m$z+}nNm^A?*{_KD6KXN8jnHFS_
zzDdHd<p-?sfyVo1i~&!>sEJ^`hH9res9J7v`O?H{)$P32ugC@McvY<+brR);gD-@z
z@HPAp!eyaXC0r3AB*H{_$<p`HO(;$!t9WE-Pj1SHgCi$5A>};-DT%<)#J-N3#+?zo
zDFL)l7yHru;Es&oRj~%!;~jMojOX5}UHRzu?%jiy9UA;8&3|pMp@Ogi$rRw&z`#I@
ztb5SKYCPpCO02C-#UbmwkC8t(!vjE2-0b)6FL@yavcuV_ixmcXTlGfw85Oey$@eK9
z6RpQ`rDLYcDXA%_-Q{(==3u&mctC9Txz$7Cp>0e7E4VQS0xHjWUru5GZ|w*p9>zrd
z=4QS+UT?;Br+vi7K#>w~3b$ZI>UQKX`8C<s*V&1LcvVYyxMQBKAg@0^h=%>HY3_V~
zba3!a-(v(q2)DP_wJE#rht1*cOy#oIazI*M-qH59?#^`;d^vQyhlCg?Al&wJcb`-k
zezodHlX8`un^LHGr-M4-Sl{4T+o(Q1Erz;Oewfr201h8b+7DM&#y{wAQuI+>)*bC5
zf)L)RIc|Ydu4r*mb|$7gNC+~)*U`~#&d#QEJ#Lc9&kHkay^?>EOWu4gkuKHDee6*N
zRL1ghyKBz(X0M|*_c;hlT6OSsGJjWXFD1o9sBK|k4<lg4Hdsx2DVUh!as&?G?7)ub
zyb2KkSs3IYK-Ls=n*SCusSDR8`C7RJgUmky?Cpkj?}goBcL|zdycB-nMZU>@&A)++
z`NM0)$zMnE;9QayM`J@%M<P)nF=lx5>VwW%tdX$zI&c42WS1mH`_aL{NKEu*M0ST9
z|Bj}PfHTR}62UYeYB#?G?5a8%kJGCF%1n9H2a2AhUYr^Yg6S+0cr}#lwlHC<q!NqZ
zSnL?a&dxs4pD!*fwD!8ug6ODekMqHe&FOBZ1^v&S6>Z(6$HqfMiv$`o8|z7YSD^_%
zb3gDxO^<R;5&92Fe`tTc0k2mjtoe7x=dU46HQ__4{IJ@WOOrk_(BO6=5at4Qq1;E~
zQ9Yxhqe-cpHAGvkqj_ji>d-fzJruux6BP36<iv?vZ~Pr2BG|Ah=;r#&pg=PlMsJnt
z)zrEN-mGqJqCMb{*y4DrEu~%naSI%A>O>GU8&UcVF1{sT8hw=KZ@5szGnsE$u02`5
zex(qy)BQ{l(3TqS(m|0J3@5-{1jspbSh2(aZ<0loMfTaRv0i$9Jd$asUtw!z!}eRD
z;SnXzb3Ppeai*K){+`!tgh4=tNzh-X6uJyWwCCL8e5NEXEZ2zwJ^o`bIssqf-n=6~
z;YKmxNq4rK>-dhOqam&5KUU=7VUW$RNV9)sWmaIwyAar`Gekmw{)5_pwe}~G#2Ze@
z6j={eCZ_G)kbw9nE#hpV8Z(D}4;T?Y{m24Nw_!TY<zE*N7C{e;atDc)Bh&Ai?@i%c
zQEtna>BYOu7(<Q*8_LM<k2$FMT>b7ftyYFs=+~89^G+17Xp>dI&Ufp0$Pz;Sb;41h
zUmDHxw(cF*+P|erzOsTHN}o0UrXw_gR9(FoE~2R~uetbJC$<^$72cuU)gqgM9<7-l
zBu{<uRCq&*?K!E3<0Ybq!nEX6ttRWQ1>u`iCc^><MBgzbqe_KIyD?3|sa|70eswIE
zUZ5xho9p$@ZoG{&Zhm|KHsMn^IXJ+Tua(Dr0a|rPc!iN<ppW+2)%+LDDI;g`jknr-
z0}HXuvhMV$uHK>m59S@RV8uFXMuIj4+UwzrZpA1ap(&W^<Duu}jrc#S)0D2%^CRPt
za5}>)*b5AsT*2q=1HGpOy#x-|HyC}^$nE>&`&<`QmI^Ko9CSZ)L|QJM3f0OnBt80b
zcsDu?DH4Y@O$2kUq_pgLFsBC5s*o}kO1n%SB{Ne2LK>xNVUDKbi|(?E7ADzh3A1e<
zIb4$UTPmjJ#N6D66clVQ*6Yq4zJpCrdMT<+qSJGb$YWByAO~gdL4yN2c&fivzPp1A
z#`u!=ynpW$sryLfqm1k>I4;uC(6kAxaybKg3jg}4t`2C}#lABJM#j)cbP{r8MD0mz
zOw1%5MIE+Gx5+u`hk_qHEc}SSq?myXb=5pAQ==F!NknVev>72!s`&TeM%Lfv(e6*&
z=Do={v;pZ3bYEPh;>_tvKkhN?{X0|`gcQ1!zADu|OyAQdglTUhw^<{4o_jp!p%z59
zChIy(UJm-$3*Cv*Rq%gets}dBn+xC*6<vA@q7Zy?WOe3}sh?a6Lqi`fLMQ8(GA<+6
zYcrf`zOdBsM*GypT?IZBPIiE{_|bNa6eH+~djEQRI@h8Hf7<f`Ly+k8`7=63*HKht
z`7$HJjaKy33V9c2=io0WpK8!c3wf7Uh<#g*&X$8rT7B!JX>RjmV5fh7<Xp=PIH7E8
zh-@s9>DgrmJyqT3A1CJz(m3av`Q<9K@hBg)q}J^XGY+KaHAUWyVnT4ljkvj!bZ>o9
z&EbQ76G2a*C8;(EI63V7;43)0u9bN9Y{+U&a->}&Dhh1LV&2pVxr0-N@-AhpN4(ot
zdIjdaF|RWP*#EADD?d?QS!kMaCQjL0TPxVPqw(*j_Y)>&G{jpQ)qXKO9ssI0;jJq_
z>~}9uFZ5iG_Sz&$RwqQ9@)3NjuKQJAJz1^}V>?UvlASdNKse)&D@MAp-)Oy*ZDh-T
zpOK>3zu|dL6oBp~U`SjpRn5rUKbarK+YP_}NN<6+kGn3aw(f>{9)S?ZYK2~ta>|dU
zA03Q){vP|^v&LxuV`a7e`$<7CtyJO#VJl<%O1yuSuF{=g)~vdP-s<S!8}@ABm()|Y
z_kh5SWBm-N74cTH6pM5`ddT$hc<clM3ZNu_4MU;SP!b>Ic-jAD|IljAU-EU=!D(dy
zXQ2^kqNG#tBJ~5HtU#>}|C3B$|NeQHIJFSO_C9}J02EL~p7F<mop-1*d#)!nfR%kR
z=LxFeqO^TJEIy+O4c$Zd?wH2y61kcTrd5hU_O}|>5r=Y)n|Z1A&=1#$;0Ldo7_ZSX
zl50P~A*&Eni=-|B;VhqD4}_S#Q_jXqIhf@&>k5l+=yX|_nyR}n|Da4y=TZL2+bUN`
zspt;k8Jc->(X3no_uZ$ZXfLVtBha3GcmX;~T4Q=#VXqyts&xf*TUWEM)qZ$DVmt;B
z99WR>1bU;Qy%}$M>57tHXV|YeJ9uP2Wm$!Y*70w6F){y{TI=ejw{?FFeNki^Xc?zs
zxYjbf&zGVgjeoa%n06t%FC$8|K7{%e$J={xETHfMZ5>Q)C3yV;!T_w6+URY3-`C*8
z50LyjNv`{Ei;qYMfna6cr2Kc6m8DRcVWN}PM%1>fR0SZQH<@=lvR5pUVSXtpmCWs^
zqI#e-ZR{V*Acv(WTn}@$7^^EFA%aDs&)>a1X3|JFe!n|UgYnWjI-ib0tCH79<liOT
zOKQ+GgCiXbF(>3Ml@u{yi1GQ64FDP-xf_AtVdnhHWvnSjPCA4Zk3#!yP)6IqO4ET0
z``eH_yK#%IsMq0a`M!h)h1J!ypUiXbKl`&@HCw?c9_Se_07yGBo>k!<PGi<l0`d7{
zL)F4kC0s<s-xf?u29a2ljj^fm>^QSr%_6g#2`_4WZ37jB<^MFbZc%ow<gn20RcjX~
zk_bqQiz=vX9E*a)Pm`n!TAd`tJMB4{qBDIk)W(#B-er?yedA8NN-RmA0B_~OC=6cl
z%%g8L=OQ7A*x4wgprnKi_EHNE8shtGCaC_MImzqnkSR4+yOWefIyY2Uu)`YO6SNgw
zN%%kQ&UYcv4o$w<iI#hLIte1yZQa<rg=M(iTra6@XXR~n>t?Et1f=|=w3ONX)L2GT
z32-wMi+fsKnhe7w)vj7KzuW(5<lnnEnnOh(2VISfjw);G4~n>9Pb`nKrP_C>CiX>Q
z<&~<JE&xFS$U<vtD<l<L+S-;aTFD>rEx+L3C?v!}wdM%Z)87Cth{K+u4`)5bcz}oz
zD;gX|8)fn%?7tuF=hOTMp`mp{Lm~?NwkkBb;fk|fSX-q1M;$}}3Pshazty?31vRzs
zQ96HUF|}7NV$-jd@PpLp^ajh6Z={K&OHLW(<(qRnx(fR?Tb*0@wx6hDK}N@ESrepO
zx8k+PXiZ%x9zHrPV^!#PpO1Im@7)p39*fdol8TCwic(l3QW#<HMfp0WOG3@#*!AJK
zo$ww?X%qAgQu|pt>D-lwzKy^uxvff$fUC8hyXg%ih_~ijI;?5dB2U&oD8k&#lfy-5
zGVAoX#^&b@mRJ3mWJ<A^SsrO<US2FSlc7YMV|j<h#s+c`yN<u&cu9Q`L4a#yw%9u0
zdqKjbc)f_nA{ei*pAYV>nOte{n$4g4rC?y1rj>5f2Iw04hsq<S>v*(Ri36NOX~xXV
z0q0J+A;`7sZvRLyuVR;tU=lJ4Hwr)vn42{if4cn;Z+{+?|8g{J%7HkJr%nFnODjLj
zpO6n5eG>nGM8lt+S$vZ)Fh)#d!j{^iRXXVCwe{&L76pl+H?G8={Uvn-Iz5GnM3MRk
zj776n=PA+b+Zq<&=l5`R1=%{4eQ%6UPm>ENB6-As)u6WGbQRe1qxPF?zZ6MX+jzR7
zU)rzNwM$QPrCH9sP>JvM&HeHbLAgAXy(64#MUC;iL$&Le2r>>`COa`1E>RRVKnH%;
z`^&}fzKh+o7Yq%tC#PW#58y_zhS%oZ_PAD1CrzEiJ^7_1Ij0l_rxdw~CGuhXd(t}p
zwMVLs>vcRJC^3dtnjfJc)F$cc`*`vL9-`wpQX2X9xu;5Bk#8<**j77ETpO)ap8uaZ
zY3c5`qR9FlQPQMT>1ms^rwL5}v(6j7iY7|*pR2BqG5_X-U0NFA?v_VT%Z<Dwh^rh2
zwDtG1)UNd7H`$YdoY5hIy<C~sXa8V7Tg`bKHDFPg65DbHw#e25toupp=CAkjHE%!l
z*f<dhA)qMBh}51Gd^EI`Q6T>o<K?6-)9PO=kW%OnbHFe+tq)U}3LE)pUuTlOp<P~(
zBTTU#olTmRH@@N_9=O2E&nUrRQ-4TCEs-DdX7Hgxp~{P2)tRJZ+Snk9sbtTaj~M&V
zH@&&64h6EmzdzVa;XJQvG(fqnt1<g#AgUZpAWDdpgF<-~7`walRs!0t5V|HS6@sQm
zd}*xYT-f}UQlh2eN}@?aHwlAzgD+GpT#wsP#rgNshqe73?H)E%)F;L80Fy*k`yU3G
zV8{k52fGO<czBH{P$lr!lT^FZ<URlQ11|QFCg%t_rD9TSz4rf-OEcP}KI9}58cAxX
zClU}eBKqvUV55=Bv?sIDSp2f$nH5v_N&F?&O@fVW*J0MM;;R4Chuz1}pm=mS22=;(
zMXbRi5xHXfAd~<UvsskpCAAzW9HvD7rnM~h?K)hVZV(gG#>=ad-UA9#ZW&M18RAn}
zztHJPs`b$6DXK@E74yRw^geGAjxXE~CDK<Kk)v&fUg+~;jELE1VDkp>8R$&7yO;7A
z8sal2fpzX!A?2do%z&3vsd0h9&{N}K19w_?<@IaPOxou{SIZqxkGtd0QxGL4B)I%q
zmwy=l03V;45NpWMN+j0w>I4I^Zsb?C*Q^y}n#O4&3c8i1rgPXHL4sE%7F?#Ge(La=
zPLgHDU;aVjdKwk-+d^ylH{o=_3OES7BC2{Xf|W+z=$+x`k81Qr*@`!prxhbS(Pdu{
zUa51$Cl0(@k7CMrqMfgKbrx+Xx3WZ`Y|inXIetEDs5HVX#x>@l$i-k{_;@Yp6Rf;O
z?}Uf6@Nme(2yhX>;H8cX4+og<@4tUPj;^)~Z7v}0wyCKph|`4Zd$V4tI{%-!I(W?q
zzy=lbMd(ijE8?yFl=8;6!i(;>m>3F|pHAuAcU|2y-mBo@;!Yo4-?ALgmf=S7!u*5y
zy(}%G7{zNH<LiWl%a$`s-W9~}@8z<Fj#3=+QJwf}jrT(Cx){FtA&b)4U8Nqy@*gv)
z(hyVG8#)U@c7_4>l-&x2oKkJ8lF1!~l&}H?=G}0^W6A5h(F`_hH&bp(J;kRC-$DjT
zx}VQ-lNO_xN#YafArgz!8CkUk6fCOMQwmjknNyCj=+Bq2rNr@b?X_t+-rr06(Y-Ng
zQRN)k^!FQ}<xDDLK=1!k%&>;(*?=9QoY#Yq4a^%))_$sO3nFr*Z>8l>HCcGw(&vGj
zS+#44X1TwGr(8gM*l@%Mah#t&LsgnR4ncM-XJ}ELIil@;l)k6i_uTi8>~X&f11|xc
z`^a#uK&UhuvEYRZcFaR$p4kzXAAo-Lnp$sF79|&q5~ombFh%|K(v_6P&p~-Il2$92
zM$2?28TyJ^+N9Y~T`sXyK?}ZAxh6d`A+v&3qI$|ZZC0DdYWa$nM~V12ce)iZ43qqw
zohcB7gr%2)X{r<MW$N@-4Q*Th!cY!-%>FqGEg8lMMjn{k(xfS{{^Z+^jD??(84q?@
zs7A4pojYI<h3ViJHQybbhPNGh7CW-~`kC^Q!!I4FxLe2nuNEMuYuVCRPHuW3T~qai
zqhnl3s`D5{pQggZoc7n6hBhpJA79_D)pzj1C8tU%*LyyA91$5Qm3Gwe`5t27weLeJ
zs-so43}W}AA+02G*OPc6LK_x0a|C5fIFe*YUe%k5sV!A$d>W3`?kg2RPiC1S7A+f=
zDe{!ZTKWF()ZeqD=-GN8$tUnBO}OV~4r>IhzKlZ+bvCQY)>C{SFdyD?y7tNFd-<6Z
zqivVHHf(sikp#G(s;WXJHGkS-^vH-P#=yGdB5~!%L{uEur!EA1XB5(uPh7sajBZMJ
z<M-m`=9wec>2yZ{XiYwtKmW*K@@2!-s@x>f=J$uuxj8ZpLm(^!oq&!ns}^LIaTsI$
zORG{Y&u^a*gGe|gA!mKS@suV(Q)Dh`WwjAK@}Sc!$Y19_`6ziF6D8s|g}O>CHu^OI
zvm5wm!N}1Y!_QO2E?!b&y!6Mlq)wWGDMYo6f*Km%*4Cca=^G`)!Wbr)mX|g}pNNa?
zKU-}9zBBLB?S$pTqTtov&!Xl&Vcy3T?p?5iiojk#n4bAHmG5AR=7fE{dw+a#coA_o
z2UXr!W&M$5EYR|(i$MswCbR7~x9UDeMmjLAuo)-iq*Z7m2ct)d%uKXoy-_hGiSO&?
zD$i6=m?^ulnBb@A1J5+Pz`rS93EV^Je8IEf+SG>pm4TL)HXLL$Ol>VTBe#?piUdqf
z-4@2WlEUm0RwzpB@zrq{C@{d&ZAJS|(dZrc@cz+U-3OQ$A7KVO^sI#4<?bU5fcV<o
z*D^iyW6q;bfAwm7m=cQnk#eOM^j@=?iS_Ij6Pi}+WxHEr#?Wi=fD+CWT40xp#W5p0
zB{3tx&Gk!JuU;aN&iPTtCIZpeJYa4NrJkyrj`22WyYDC>*!+9<Ct>p`TZfT%e7Y&3
zoz_jy1GPsN)tM2XaVQJldBLmkah2`6IBel%I_x)X<tR{y&Fki2ea3<A<7fWi$sPU%
z-o*QUrnYa95R-&jW$IHG_V3m}MgXe?><@W)R6Sdj1Ea#Ks)fZi3=y$hL?hK)yibk#
z;3L}{wpmZ^LktQT)sw#rmmhETGcre+8-?QCt_fwo;YHJbK3p1OONlY6(qA?s!%C!s
z1haq&HH{VL22_VpD#AA^>WFdU4g(wWjbkT+H()qRP9}tbBBFVfGv4=5B=AMVG5Z)H
zN8t5+#?-3^Ndt}V<IP+j+1MQLAM|agGv|#x?IyPXJ>JKRbRhN-uaFUm)#f}~`k6R^
zP6K1`4SElS5f2}Ja{k&)IS(DphX&`{j}0i<e$lb8V3^Zu)+m?g$3eo0|II(7{KzzO
zfKNa`cDDDQ=AHj)&jQY<>U_i0ufc?RWe;{dm_z(br{hxsSHLDjL6k-^-F{YUnRVA!
zfGDA%hP+$lBJ~|UYUE=OjDbW3k_JXcSf%xOl(-55(7wv{ngp&lgcYm?H-2)eB5Y|1
zDPEl#3b9glwBEnkVKL4mCsnL|zNij&_RU;jLHi_RegJHJ4e*ExWLQEa1BB|Y<KwS3
zFMMmC_h8Xneth3F8-71X#bynegyHCG^o1s)ZmDK71n6yVTZ4=lFd^%P)6@_@dY$3H
z{^?{!FH7zR<ahOFJ^9ENZj#8)4(87`%Phvn(YqYuG3Rt!6QIMSzBu#n5CXp`Fh@L)
z3{24nf)8-NOw>CVRD8)ldrV=$6h$`GY|$TVj1d3E7hx7PeUOs3lga~<KV(Oow|Z_u
zKK?ldivaTJ(C_~u9rbl5m8l#XS<YqD{>F#iU96w3T*GA^z&s0jo2dABuub{sVR~vt
zzLORA5VFP6jO;eee@sC})LQ5Yc`?9~c>DN(geE*90>W*MAhmL0iV-Ktki!JHaj;$x
z(UlH#abZW(>kigaik-ay>-u;z4p+u_!wo2LSiIjhHa3E?)YeO15rzx^o&f;@b&X>x
z&Bin~M0vIJf+kAea!@Zne;lo0A8Ai_Km<cm4zkR;`JbrV&QOEFa8?Byo6<I934G}r
zmP}^=$lRc2fBpJ3bY4GCPrtadK%mG6otd(teDJB+rT*OYMz(rNYbGntjG$An=u-PJ
zv@<e2Hj1XMfj~SI1E93LU+s3<yywm@HML5z7!q(yvsOqyq@+CHw*OmF0z#+T$mcxm
zKj&3<_(3mI{$J@1DN7xHQq|1YKaEMP_n$dcyue3p@IU|M`1uU?hH;|FsU1uz0g<89
zuX45cR<{94gTz{x4e04v4=B1x&=o2Ibnp`v_EQR_nr%%k6%BSq#1i}4Uwf39RC)D?
zHLS3P9RA);M!L?oGCka{a#Al=%ur82HfnT6y0aCUCpMh!w$g!AffDp{MIv4F{^h0F
zpM^)A@9V0^@C;cK+~pJ>8LNgNVH1!>PI>$iB-mXOmpcIabiMSG5E%T_j)zU|vOCFq
z%ZR}xx=(CFfIfKWXn-5K(+Lx}K`1I>JIRzItL*@8?PBGeL4~Lwh6I_7bsvMwA0%%O
zgV#vuHa_6%g<(s#2~=A%4-*SuEKjK^PaW@*ju$)<ZmX%q!kFIrvsM|BiH5m3Ubk`#
zTt(S8IUeXpWeef{NO3>jUVA-Cl(Y-lT_!A2e*~bP`9NuocQ#Cb1K-M?Fbzs+ViW@f
zpOS%o+g^;VqoeV(Pa*Ml7&B|=X2zOFwT6WyAltM5iLVb)NZI@{G>HRO5K@@3qBleQ
z0=)mYujMZ4&u+m6;1xr9H#SS$e*Y8qySi$XSTHWT@l>D46QGlrTUK`rX2)OD&=SFD
z0oS1RaMsk|oPfy5b&1EHCv>7Ig`8K??E=#Oo>WYz+iwg40sjA87>t5Agl7wa2VJ)5
zfh0y~87jVgD$~%`PR)wNL^Nik+uP4mAglUah(`-New>|`XX5@r8!|m`QaLtVoAhVj
zhlX}%jk`ere;Fv1p&r)$qbI2zlJN*23PeYE&z!x}V$}N_>jbAS&_;oJDkUpxZ)fMV
ztPsn8{QN}z!pH~=6hV5P+gD9X=619R&AbPA=;;VVK*0>t{5pb>K*})VID!dC|No1v
zNT_Mb2!NnP7?;wMn6wI`>s3#I{eeXQS_nr_---7V1B3a>D-6pj-3?76z<vO}dXnq;
z@%@v;LA1ga%(OwS;MbN6|MtG`Ja=VE8u-Fho|=-#dSg-cnhX1fYZH&2McYq+!e9U)
zX4i`T8DRPN(i_^NI^;l*K3^6s(ZS1o@(^qN7HGcyV=s_JH=wVrTj|-7?@Tc>JvKyE
zQ9w&UC<CRPt-dK!iUMq%Fx-5X=hN;t7({EH|0dw57*qDEf5>DIwS#wF{FJQHH*&WV
z!ObZ7%z9INgMIsEtgkURC|^S%A}70W2jrYW!ou-}HL^ey%W1)KOc!!_4tY`lYZ%m8
zA46V)zIMpFQ1z}ZCg1{F(I>i`ypCnc)cGHYfWZYFA!fdcM$W1+sr^vm^!+UYEbKYy
zMTk*AN!04>3}_z!FHo;rfFceo|3L0N#`*9V{L57|T}a}`Mk<4&63eUGBg?2V2lo?c
z3tf@ooT?gTk4&Shap1_3Heh+&)oEFS)bZ^bFAdFe&_ct{hP`6q3su!O(ljt44I*=>
z_yu8bCRB;wBOr$iB^%+FTN|G^Bku&4sqt!c_g>&W6)Pm<XZ^8wIS#HcPd{?a(Yt`j
zeEk>|=M2FdpW(cseC+-m$OQPh4)*qgvoPoL|5y1xQ&E{&@IE^<f>1wLdxQ@jq~%p^
z>$|ud=SxLG1mj00j?LV=y_yPx)5c$$K0lP206WVPq;A0om7&~PTm1K`o+=8$YPw-a
zm$qh}qq9gB<ZyVsn$~g;GY3s7IMcvU05`lHvVh<Q4$caJu>_7r#LC77Oqeuva>7Fp
z6BEOS!d%Dxv9SlQrM2){UFir0Ze0eMSjo0Ln?BOQ_h~cx(k}4E|4D!Sf69B$u&BCi
zOSnv^2r8%~K@?P!ETEDULn$Z-3P{dKR>?W3NRsp_f<!Tq1SIF6h=5Q8$w`tZNRXUk
zkMi_Aef#$9pI`s^>N(FrU#qCS_gZU)F~=MeT}&kh&z2^O{1d~vF`5b<VhjI{{TeFb
zg6!&-Ze874($U&kzt|Q1QU3VZvkSOXXcb|o&u4RoOF^=b@<H!tc|$`wzL6r&3n+qS
z0^l~-NAm3B3ebw@$S)rEmBvQ-+GQC`^J~qmYH6`W3lG!+4ANB$Sc=Y=_e6z<R}IZT
z0bO$`cyWi9uK1ye-~bC~+J6i3*poc*eu*zYSLM@=hbhr?L0TeDpFaII;JlaV;kOet
z?eFsPRDu&ibom+b=IKT`Pnc?{E$7rQ6O7sVSS_tB^)j=wXDHcrY%b&Hd8ycP0Hb;s
zM5(-}wtKFQEKcDY4fr`Do|_pj7zk`-@X}?7EB`(UmSTzZX5FtA4ol0hMzG;6tuMJ`
zXS|h3?>&sy=S>xEUrh)wPcT>YJTC*eDwr>DmLblr`1H?`OG(@ZFPY;5Grc_T#i0rF
zSMS2i4*ix`H~T-G2M8D+4;-QWZjPO@TBZtjw%S^iC^JU?zWupT8(4lrov}mY(Wc}X
zJ8Nr?=)%`G#Qq^2Sqn%^9Nu2PGjO2ODYWtVdtKj?PiapQWO4+wb&W`$=Wgz~JafbF
zrr#J`$d>5BOMamQ3-;rpm_(sl=B1szou@>mg<C=*rc?*gVeS@UUb>wro)**Yf4UuY
zsO&ekPu7o(-Onhz_A@Z*($-^vY|j$dFY;CHv+ey-e>=9W(bSjgi6V<_v!<FY(d7M=
zTW_-V^T<bdf1$C8p%t0$>1FMTbNr!}9^4UKr0p(VZ)hhlbEv7x&2G^7OL;@sRP*5n
z5xMQFrO5|OZEJIX3j~(w#u|-W^Jn){S?pu0CpqyYR0CB{oHDne``T=74$k_$&MfHA
zeOP?6kDh+vT?ozx0!L_DJbV6J@l<J~Y`mZnfx_eUk7vJ;p-f#t-^$$XzI3P*sHM1#
zdA8wzvh%1*C?@{7_wL%4<;V64oFRtaZL^D4r%BSCCE`Z&{>r1mMTHinZ?`O_ahruE
z3|%L^9giIvwV6*g&dxRB+b&kDR!)+5D0Nhiq2ki|S^$>p1pC?qtAHSeCdZt8nJ=qA
z9f{iTyz|1flyF8`ar9hNoBI+mHt>Fhhn1aOoIjQ6-V18fCe54!LKjZ^8L0Xa^#Y7W
zPWsy(;oErmJ-u<)d&*D)t`g@q!y5e<G4T;}jNJ~k8UMa?jce4V08q#K7E`njKuw%n
zzHCp|N34TMRR_xWDLwU|0(pCUB|Nv!!p@GDm)GrXY3Xm9jMUsrRYo;Qj$b~O6r+iv
zm}-VLsmwO#_tD$-oc~&VYm_?R?mO+)nfsqcyOr}b>iBE?Sf4xcpn>fV>x*k|HJLli
zU^?qpMh!)7oE`&Vquc&Z+^GRypU{fcDNFy<iH~b#SJ<8g2P?R_mE_wgYiXTF6>38I
zdSb$f`*quk`~{DEeN<Lj3e<ULa$j1-u8~{ZBvBEOspv=UaXtMB#!qx{&+&-Crt}f{
zGo3ai5)sF`eX!6-6R~_;Tg@*O`@X~@C85*7Qq7}(AZqn={t~gdTsFO`Eps>1=s8&%
zo7wHAr_1;^JAc~PUtV6Wc<NL}l9{FDsrbU+syjGid|X`k16|{}Lj=<YiQ6i{rwic2
zxC0AJT}n$!J9N&;BJ>H@k5S)eUnG2V70s`XL~9}-c7|0$Kokhc&TV3KAl*R)o1Ke`
zOF$q4_5DH0yS$!i-C<e&G1ApO#>svIZwkh_*S|@Ge~Rcnr=X&uQfWI{*KBz-`@OoI
zA%I#_o%q3sh3$l7&%_`_PEu_x^}nXiGFmr(Vn1-}V#zA0)?Y1U_ihExk4})jujeQ(
zPL{5G{%CuWA9;+C%hPj#zyWVcIZsc;@*I@oQDfM@e^yU&lA)y}+d(?^kR9K-E3u&h
zHI#fo7iDj&Z<{U`U`;GK%vHQ)6%F>_GhT~rL?Pm~Nw?v=BxfigNHN(nMC<15zNb$b
zW^9)Ptt>6m3qH*FUuE~}txRZf;$yvC$xD3x&p(s0k)h}3?eM?IA3>PxqwIV}BILU5
z?~A938=v|ZfBS~WEwZ=|-sPk1XlQEc#Eje0c};mHTZw_!vf9dQJ^iEIYWK~0bPiMD
zAfxrVUu+v;M|k2k$?_e)h&DjBm9-o_BgHk+&#il$;A=Y^95Od{nOmB;Y15{my5m_3
zQ=Bcv)1(kS0sFLo0O}C&quCMf<gb?uuhbuA9i(Mtm4Li1rEz<IUCP{On}lmvTlJ4a
z`{ioC%8yd>k?z_rJGO-=#+Pg*5F9My+&1GqsXyj$c(66<Zyb`9`k+CpEqB!G*8G@$
z!i>Fo0G6J&z$JfS{j%u>MW#gW?4v?qVe@fyYeO2+)zfC3;N#E-u1Zw6l=z0aUSezL
z*f@T#yI}5Z(0PGhxo3XS*VWcE?7Py=m9ym3p*zV;AoK;{Ucr6?)S$?DJ5jOXVsY}e
z$jE!vBMdZ&pG8g9ws3|DQVE*(h)0~Q=XNHGB#bw(8+J77`!EQE?h*(;DHm-^QRc0h
zuRgQO*EsEMS$vOexiK3hflwqcr#^v#GU#L42#%pDoM$<JK+1xNq`?r|INzsrHK}E3
zDXC0!nk7DhZX=#PMMYNP(CZd1DE7$`G2R|gYz$IpC`!CfPCmm>5pTeh6ky-2(YomH
zrjtM*^n`axALHleKXRnBt1Hq6%5@+>U<iVK1#{G_e1q|w-f~~_^|n`OTr47WuhIku
zcum`3sT!M_;3oeJ4k{rcH2!)s;=CV~)3VSRgNc_<`V&}x+mtCZ)xQYIq{{r><oa09
zJosn$=3>R<%}$9byJqFnRes0n-5Fs&;8BVpXZKzHjEtBb*mA(<(Ef~t@n=$shdO%S
zCsDH@gQDEI^OTkYveahY=l(Y&yK;~&N2F7mhhFXCJN?H?K6bBf&a$Izc9~QGMAc5p
z@CZ}vW@ezFX+T3*q9PL}b}%wlnQ&)*6BBt6z`$|+WBkm#>wIVH+lw9)S#~VLEBk2H
zY>P|9_W7K+kuq$(u+EZQ;c2k2xRUkk^$Hh(pt4tCS0GR`2xv1)xM?J{z+T)(n^{za
z0KF@Vq$`V`Tk=inl9G~`JxqB<4dPz9Uj7V&!^4i+Z(T>n*A%lWZq&Fb^iCW*$d$K&
zEt*|z?JuO3{{qkiDbu~*vAO-{z1God^#TftBTzdrGS2Fut$Mh8ezmT0UgMir8Rsi%
z@9YZ)cl&vsVZOR~B9_hJ?DO|M_TzYC3GGzwoGUCRf$%`ByjhYHwOq`8eg;|@_6gPq
z3wbq=xynR9`sU4>kh87>ndQi3ujW+km<i$A$3nSQdT(%j8)oX1`Z%>P_xS>JWpxr|
zipz8ZiWRl~NzcCSpV{@kqPFvfgV@=#FD8f^0}C0`x#K;2--fj^vX?%Tu)e1z5Yp~k
z8jqcRgyCbZon^slxyRU?)@rrgmSPbix_B>l#awUs)I6&fUE*1D=EKXN%=YTRMhrCX
zyxNCvd0bd1*8NeEV#_J2*mn;0S<i>{@1MB6Hu0UWaX6x@lX-F>QU9Lc0_F_x+%YYE
z-8KpjOt*LM1)^zo0h)KwBh)Xzuz>=58}-5EI{(C@bRX$^P6`W$O%RDh`{oF@M3W^^
z&C3UewcNE8590m~V25C=-NsD4M0|Wa#xjqC&xN7weL|t04E#Ds`?f5~PNmue#7#sy
zt1{hlP5k-lW9q?|?^IGiuVY4l+S)qbZQpyZDUQ0=5)@A()6yw)YxaJ0cL&iH)-OFf
z8z35Qe0txle5QLFg(eH9)y#c=MlQSAHe7EuB{RUL``^y?yU2&KuJ?J&PV8y%QfepM
zB;g>IS5$cUpmp1B>PI=a!bdqctivfW6=t|csvtaBzR<U0ZT);n!kOd_Sh<rz{zSHC
zTnTO-{{B=Y`!*3Q>Y5+1!+qhw;|kyl=78ym2MBtX`k@c50$4+{>0FaCSL^UfcYciG
z2OHfEkM$DjQ&#yQX1JYb(QBN?ehq$$?m~^zg<~_JFDg;{xpInH2tebBiH|@BVN4)h
z^3dmlMt<Y_+g!W@J3GkTz~5z#@TCT(r^Di=hUt}1W|#6$Yc0oI4`C;br%rUWx5jx{
zKk5J<&!PUYrFM$uvEZt>RBaAA_13lbpdg*tPw7kiY`>SV_xVtm^sieJ#S2<@?<T8S
zHRt3jDsqO@KX|x5)8FTJ&9F+iO>@Gnpbv(JBP=9lF5F7zDO;VC*V#p*P$nYcnXkwl
z08TPyX}%UXcJ9RY>G`b#7Gh!2<!U;-2mJ8Q{@kNUx9`n;$uPgDH2(079!&#<b}GNK
z*d`4JohpbM?B16BuC3>Av0F`z37#IUuZ-&mC1KKL(d_dcXlD$+K9V)Pud}dlV2a(m
zc&_#BsLQQM9ZjuGG(6Kb6T>>n1B|a*AvAnAsDMahqv_}{5n>{*3z_PH9_1kN1PaX}
z2g}-_#54Pui<bI4_S6-f9q1eFMpN`3(1L$;v$BR3#e`ob26snQ=J#naH$R5AYOMQX
zGtB1dtn*K9e7ZRkLFIU?e7;#5Z;67qx7sa>-*ZkR7lsDl97x>7myOnf1b(%zFp7G-
zwr5!h`rhqOMUj0IUU;KD!!iB6jrBKabpQPj+kN3rY>O$!jR<J2u+q{#3k#DXQh4Yh
zZ^7PLn4M)ldq1jF2RZ)_0FXV7L(!7GKg8Atc7CLmymaZakzicw?GuNVx4Wz_4T0l4
zL9DlsOksQ$q4eQwmh#uOB?ro+=o2cUKq00EDrcfMzseyCL%=HC?<rD+&p?2Su~Wy;
zkj+7U9i;L?dp8=Z02ZHzWH4&ozz0&Ml>iScoqts6U{~7k(S4(mH^|S=8Di_T(Uq%b
zYNFO%H>6II++wVZP+gfQn%Uj!`!eoA{HXnqTZ!z#vy@s{3BHpaR$B<KT1C+8;zPs!
zPCn7wyV5MR_$m?WqJiRgo3=Yy<r&aRwQKUr>Ch{VcBQAl+Tti~fr4A6SO4SZ2C)cN
zs<!Qsz1L-Re`x;dxplHPFR?!7RiVRR%eVIO>cB~wi9~M`olOMOlTW1-8zgIluQy@*
zLyz0aALz7GF$w*KQfY{?TeUjfua0}d1-N-U4Cn%~va(`8Azs_Jd!w<&I;K|KrNBnr
zdbnE7ej-|?*xu%w`qbLGzWs2KfvoMyMZ>!MMk{r%rO$0$q-PX-y$A^YIdT~e-hO5N
zUP&f_utOW=;uqR(i4DeR`5zb<@I3+001z86n2@SF&^<Y|`&LTR@XOiIZM5rm(^2Xn
zxB6fdA4OnpKt0E!@0v|bO^$(cJ~x&;#V8N)6O&v=azeYNS%$PF))xmP-Pc~650iAv
z-_zsn4y#W#6?^lbx&`?2tV!b+w1o?sEFmW4z@#y>ntuB93783xlyitNN=uJ$a&n4_
z7F4T;F2$-XW|lu8zCpLA<khRYnH~NtMmzHgblp=<buF+s*_*#EdFh&ojS??St&RP#
zLgm8=;f(=M5u4KmPf_LP$21zZn4XS1vyIUAT!mW;Qy)S+5r?H@{De`Gb0TeS3GL8h
z^aAUu$@4hRyofVzPjNko<Ofey5Gx;49FusO!AZEi-p2ZunCxv_n$4<Y<LF7Al*kmQ
zLO_&~LSe#!!N?Ez(LE##7<rmvnWF+d3Nn#@V*l-P&BKyC^$xF^wsmwSFC>~hjMf1!
zw)Kv|TAFwTn7pCkadlw7byhT2bx~1V28k2Ex;xGb%qR$9-+Oy|{Y6Z>e%`w(>Y>|S
zJiBj?boZCBbc#8bcXE+23Jk#vx4o*mk~beW&!h`0s@2n-KEsYH0`(1|M#WmJ%*@}x
z<v@rdtKUM~pqkHdDD*m`p?6GuQxdM8kYs3DnhwwfyAWq==-038dtG+WA!nBi6w=@A
zc}ce-<rEcraMi3+F=MWY-OBW?GlTTB4V}ww{rBs6`;OB3I&OqGaM1ZNdeO9B+&LkH
zcNaB%;in>HB&T^5=&H6>R9xlGd39)xPu786@t)xbdkEGV%|qtq=4@o?wOzZ&1E!Dx
z<w$Z?an70L%1h5KZ&+KZjSSzfPTeuks%PNvPVc~^RJj_ml=9E)2iO(NbjdN{`ohqg
zq$I8a&LSV$Vz=hvS=!=~0F9h9-a$je4jZ?M`T9S6plhp77eIK1=Ax@a;_PUf9A`*H
zc{#>iS}{p~q9RVj8T2S7YL<_k(Kd_biG|#>pkPy5n=;$QrhP~j&ivmZJKD5*tC>9q
zUikR5Iy<X1FaC-{e)ip+j{E4wGK+PFQi|HIp`l89HpD(Q^@kSC7AO1B4ULHs5)P9(
z*RJ`>CyKi)rX8TZcmGyhXsBycupE)qDer2#x|wuB{rMGtI=<Lz3U&UGd4{!2B?E}w
z4mix89RqL!<u8Ax(@$3TH@|q>BAZ9vpFI|7k#{fU?)Nj`XZVWXAVHR8Q`XRC!ULws
zu}1@QZQ5<dLO9Xkr}VE<Z>8l7DCJ+)x=Q(m2tBF8gloi$iaAPM7dSPC{+NVNpIx6?
z9!-r=S=yPsE0$e?D)Myj1jwC}4BNJC!+uOwPVLd}?MBN#I;BuLipl?`A!B;^@KY%_
z$aR&tu0+_|G^iQqDvg^P{IYub+s~<IcXFaayGk5{xu6C(jaeL&YqP=VZzkVWGbJ!x
zSmj{SFTfa977p`5G2qsaYcs%7eWl`?ZZkR<q68JRUae0b;g=Xd2cV@36iH_0yVyY*
zwDk0qs+_My=F`3iSF~|x9pdZQ@mg|<gb6*<BtEzl!iBF#&j$aFNzXtHZk|uAJ@?iF
z`S`A>9#2rFpAg;M?N7J28?Uj>bfeAgUPZs?u3fvb)&9UZNcuAz39f<$;~C&i_q#<|
zSz`dwEiO95%2#FFg{&G>Bw0086cvSZJ`tj2EBR|s=FxKe6*9sSh+GfjI`EpCzeO@I
z%?fVHpdxy(usBJjWT+!$YZmXCm{;D<oI_+SX{=i;I8Ec~{z~t3aZ=%#kyG2qGHZOA
z1wViPUWBX-<oM9>ZQ1bV3Py{d`&{nEjT9AV-8sd$&UZkQ7*d2Q?VjX`Obgxiu;ATT
z1$q%?1&Iw&AO=t_#9@OWK%dlLq<P6#Eweqv4|}fFbAKCNCBB*6W7wkbUTioU@;R2k
zUKRtfDd=tt^{d{p0NioDre-oHm;Ftb=anAngGyBTsrD`v{BeN053@NtuN;#O{rYPf
z4q54tuNXJCzmJB7h7eP&K79|24<rYu$}pa1{>tvk=12PI{q1#`sZ}>GF3%Xl;*#eg
zq997&VaY=i8EGH)uEtCEQAvlh!*8kSskQtxo#Dn<tjK^E6B4YfjDCGiEkR-uJx@Wp
zgx1?8`xtQpluS+c<>*8ee-Rd5L_mC8HmV(qn%m=)G%Xqwt(Y+V8PX?Uo_EyOPauNl
z6QFzn=m%ae9OI^6kbI2?Yr?Tn1SkWIM3vP_1acqv&^Z^xT)!5lRqX2%6-6SEL^|K9
zrSzofmSGeZeDPVkF_nim2DTDzM%G4)DbK90K=1gA<91gU7yPa!bVl~|>9CD7#n)#T
zkp3lu1)bEf(NVJ>ZfHCtM<)oJx<Ef6Js3G<JDuZ2eGs-j2IJ3ts!1gEY*SmrGNu!A
zw?0P}@o2~UN##c-Hq6_kb&U_X0*jjRc6)W!`CE9?soF?n>*J1RjjNAdg>z22Gr#}P
zMM<~Co4R6%Vd9frkR8B<z}LQmyOW8fX_dzQAL~H*tQBc5q;-aBO+;mS3=i--MrMEd
zd5<oMDL7ZPS%VW=Kj_$~m`F!U>H$~DmoF061ih&bwqjdhy&)rBG+oUFbo)cky6Mg>
z4b8L3G6AjXY)j2ayfkyeO+#I$-Wv557RN8RzrS@a3{b4uZ0G3FkVbbW?L3SAlV<C)
zO>*7J$r^)gheuvmL*p$)<wM&jE^_UmTJG0|*k1`J9=g^*b#lfOfrFjhJZ6w$+N;VR
zQx6&UcLHcI1Fau<9Zds}ND-z3U!MK?`BUmB?Y^YuImd01e$rp4GCSHx*H#ybQvoW)
zb{6}}K(6M>#*H_}`5Icd<XDdkbeVsIdHve(G~&$4?ApTlMWZFCrpk#F%Ew!<y;)tw
zlUo>W@@}fR^rmgJsj(4V-C93CuoE6!!Bj)&t-DP92~)FJT^tBvdzt|ES;1N#Bf-kR
z@cV~e`G9q#Y-$@>d&j<Xi#%uVo?VWEK1s@?5NWGvALGD6v$XFyR|VGpl<kpFl$4z4
z$agjNSk9#5ajK3$+uU=v7-p8lte@GU8%oJ&A$uOaw6Moy;aq2Szfjsg&!0c{XA~Ud
z@H}T=aQ~24965bPWR-R}E~g(xgDl$&Q4k(lUtZ0EX5?}xz~ldSGYWj+7U9mPu7sOk
zW)_uUze+>H^V0p^94C-ujued#XKjjY&#J}-rYVn2&CrCKQ(4cPCvl%R5hi5PXqg@Q
z#82)EuNi1d<?8TqQB!mGGtwD*#f`DY20I7dZf*4%K&~9a_3Y-#?ssiXH75C$fz#1n
z#eq$z0h=h#m_Nfjs0X`<?&qIB?+kn!M41+9<{&RyL9MmswuMRtm{L}D>nqzihJZ?)
zkL?SKiH=N`7t0vC&bLQU>7e1z*w`vu*^3u14j$aBd<K<hl+Q+NMD^u3KfoT5<)!`w
zfb$j4%2SU$Wd{u0eE1(9siSNMBPF-fQV<eqk=Fye0+ZH16(<&1#C<a}b@as-jbbiV
ziq>TTk&uv*Tk@;Vb7<lZh2GyV+F@<(`;J0Tsp)Nt7Y!-n^o%Y{hp^LvR>a(q5b@Rd
z?vtvLK`Im8Ht4+2>kTKmE;pn+pW=yBpLj?hSd_x-0R}?lvZCP<X12r8t#la0Kz!9I
zyjH%wtE;z@bphMp>#y_OMzTrYJ+d-3N8h({3Vmq&&)ds)qDOKEQf|9#tl5w3m(@EZ
zB6{oG2D133LNE0Q*bS|YoH^Lhaq|)gQ`Xr7V<hRB9^B0)3eF;UL^h(Dc(kZ$Lr&83
z)vLqS*%i4z&oCcf{^-oudq83~Sc;ZuxyzcLHJSgVfzt68q<#DLmA#?=<~*TrZ6{CE
z3&+Ln7}_S|5Wn-DaPqcJ3l;9HetxGwiSgm#VQ^HC9>Srry+Mj%8CtzO#>u_>Vez%k
zZ^4)X4aI|(T44Wb*H*^W4<UFoOja}b(*nmL?wjb|3%PT9ToUTe0^sN_&!SreWl@WW
zv2C$&adly*$_B@IG|MU=(&KPlP)=}6I%VmEptnz4o=C2;4#8`vsnFhh_V6~sUU7ww
zeiMKbP&Nd%W|hDCFFbchsMQa(=LW9}y;~sGG4`hbWyf*5cW1x-oUK8K=viYI*Yz<s
z6IXs3SPpg(=K8)YEZzeA2nE9rTE!7jQL8^9JCQV<?RCjS3lzdOW$y1dTu4Hal2uVW
zj)+h*s2w>qx7+)A?EA1hrC9ln&eq=!N-Z}vG{)%rt`Gw>qpUV$`tM67O;6W{gmwwk
zP_W7kOR(tTF~=VgyU5COID5H0s>8(l2TI0yB*fyM;~I>~b6WBO_l1N6SXwW^ylA^b
zA|rDh6#ZXOk+zF{933;N2~LQ(z$=7I(o{LUw#-ZMrSgT^MCA8hm?gf)Tb4z;A1^-d
zAS~&eAE)U03;||?Z8uZLYS*<z1YvLb?2ghtP2t(L6@oP9TwejEdcZ0Eq5GTi`ckn-
z#YXR?tY4iUXX4MKXNxp_+i1|nOs_=*5K+<5Sh+*{_6a)A-$j|Tzdzli>G%L8NM|!s
zQzi{>$~V?}HxQ(vNC3DSQ{2qZjeWrEepe0;rZb^-A}(Hp!98#wN}M5}K`#FA<HOKT
z@JUk@IaNagf`Z%!r5NMR%OhN^Kp5st#rUSdIQdOqn7E@=@h$zcUCk!v96RBX^V&Vt
z%)_3CCZ1yW{Ad5h(6j{4KOP<)!#{U&o_t+HRbZQhAV~vwD#76y7LHZK>Rm#DkB>^I
zW|r|S7y(^fUC4L*YC+BgEyTG$xxGvrh9v0k7Y)Le;;?q5>0V}21IPnN08!RxU*9B9
zynVOuUr=YCM}Qdvyf54laOhN>*hI2=?gzND3RcMgra7#Lc%@~Vu$L|bbMVUVDv!@!
zY760XR?2E5y{-PnjLOFLxEtLiCeh%j6vm`V4(&={gt`yN2%ifpHa+m=6*9d!rIw<E
zta#0(?5d(#tLE(-i~i3~a*xU}u)N2AhExID8<euTy1MvOGl;w|$XZ<0VKe|N1Z25Q
zS9{}AML;wfh=Aw@qJo{fcCoX8e(>Zfz1-B;p3h*2fy+6HXM^Zbh__+b25)j4<fH65
z9cy}n3|YM&9cOo~dG!X>1O(yw0Yi^E&BzT?mF|-ByIseA;?I}Ecj^2YxxH@j^!n~(
zb<{gm!)54bX{X;EIl(nEH&<I<zcJrso{KXG&kehTqIZ?>10w?rC?Z!!5t2H>jX>W8
zWm84fOueEH2-?UWA-@4q{lIA5+}5qB6IaJ1IYD<2!@fhJUB5m6VpMj~5cYrr2ake2
zzfFz;mF4?C+4L%pFdu$lkDIh~Xw!C9mdSZBC&j}Gm76YX^Ci-vrJjk&^m?}L>_}27
z6O|Xa*J!xOXleQv;2oqPWOS(rAgwh*j`mn-`T@-+oMy|5i)2){{j?#FFO)Q>_w3<A
zOLI$7+sxou+jqB$wPK>8?lOnfJLsc}SFpY?Gd}IZ)2lD;41K8KNi(`4v3*fuex8E<
z@u?qL2Ad;!T76QRa6T3qn}=uDy@^C4+0?8Tkx9d}SDCWu7)+Y=bl=@#S1>OCJ7gX6
z3}xlyFoU`q&3nj6N=>+>=i}wG&!qn2LnqO%OL?Kz+SYb0DBx$|qQ}oM$e(;%pW`J%
zimQ7RoD*F())r6~iUC9yALHhPq77V#tiCFDcF$zp8kK(i^HPVf&xj@O!<SSFKRO?F
z9}%efBdsSP*PhwAqj*yR*YaRP*R)iZIrlwgJY$H_-Ntvms#irbucqfaUgON(!lm|0
zGuwzVtQ<^8F&80ws~!*8U33+dl`|}r5kBQVu)0Y9R2T2vCL~15xW1!hWbA+y0k(yR
zR8T~uqp2ylNd-Do`O_2LCtn`wp%s_u{{m`_NuBsJsRaFaSCYQ)pLQw&;dBLjDzwQ^
z>>+D6wxvqIgTNUI2N*mQq9`jg30L!Y9a*fG_I69J<yL9V5P6fGaGZ`yf|V&-2z!rS
z?1>l3?$cq21J><Sd|p^M4|jx0c~nKl1+y8-fISN5=iZa->5f3Ru`^d8UP0$@>`^(@
zoaf{uSmzMJEEKksu~kUb)ft(Ycx*-bmAhs2r!J)^rDA8GoUH9ScRQr#bFSjZ0TD-{
zP+M88!lS9@!p%UKvMx^<^+wxBtakw)m}1#ZAk>sK*Vh5j(}t=9qP@Dham33oOu%1U
z8mmR4@1qI~!AhuEfaX)-iB^vd*eM_A^1cJ&bb!QDiKMetM7UwH!E$PQb5|~GZfBl`
z`K!cXRZfp1D!d2#e||Op+WD!Xf_%#^@vw&00&rWUI_Oh8TViKx>$+SvhpN!Z$_ge%
z0s+F;G1AYEw(&u5=FFKKe9F9Ns>1K_;8*OD-`Qus_PnH^7I9s+DW7KX+*9`=M4Hhs
z_J#KjO;A1wFJltz|7&$^VYXxo3=pzc;{N^npFewsP)K2_6$)9Hn9U`czlwwa5e2>s
z6^ReyN5m=0_~jF49fzJ-SX#bzPW-7H^g-($wp>aI4~>C|(5Vhb;%td>cW9u??xG`f
zegRLnUgo9~q*7RVg~63jQ9suDeT$D@lcK!59zTQsK9N>b+}=kW-?Z14Sbuu^<J<U2
z>-%N)Vrt4G{`K*lWA79r6IrW}MWKEt?6#Vxl*EFDb&Qzz`CT2t735m2B`=See7<O@
z4*}K0jqCh*x0<k12PD>Erl4Gf?x@t_d9}GcKP-V9`@YRlRzbV7-SVt?W2595lhuam
zkc!umhTlt+-!*NkGRPcV5;wV85>c9weCnAFMt7sXXFKLh3ArpzzNPT8vMN#EdoO(9
zlZ@_s(|#6~wW;vTFZ_a+J0TYH#b5o*(<$@VPwQRmY)_H&ggg1Qai(#T`29aop446K
zcJA53{@N+fnSf&f#**BxW2l8X-7N*dA7GnLLQ#>%>zj()?u(PJ5x9uITZN_b=c_sY
z;`+5~RWo(5xJ|McTD`9iuouk7aaCTOEPQl{reV6^zCn4pp^OUk=P-p=8P(Q_$!ggn
z#xIml3O~+@*eKf2x~~$ZYDdNr6Se=ay}2<^ab8rCjlRY?0)vMoO6KW#mcx9`t7c}j
z<y(V0ak{PI#+@kXSU|#F3#5}Zv3fj@r|g3aRBfg?nz}5+UqH`oAg>w4X$a>ZAi2Yg
zh4QFdrAzYeADv}29NgO=Odqe4?>`~DSy@FzYWxt9sJAoRL~9NKq1@U!KhMVV&=iK;
z8PeF;-ORf=$Im%TK3F#qyE1+VT3|&Ln+OT#WhnzOei?`fDryA<1u*htzB3?qBH*WE
zTqD1Q&vJ1{znu%Y-m|c`kH8%o2l=-wT^`MX3)+9e+T*fSb1)86)3no{!HdXdIxUax
zSk*J?!4S>u5czdHTKwWp4zXjJ<g!_>FM;shH~7dgmC$GbgX`r%WRgpB0u4=Vlt{2&
zw`KoblaBGvY;0=%x0R11Z!Zo_#0<_M<y4G6b-2*xbHBSF+#nL&>2KQQPUNweWY1Pk
z)Si#f53OxI3K~2jM1)gtW<A<2M;^37q|;27Dyj13L8?rF3~J&U3k%!WPnXI>p3Q$>
zHF*~?&MY%Y2{BoKK}ULwdBXRouTM_we*$=M3XoafmDha^{`J$8&u*J?oGMKFu#(wE
zv8xsFFj$4;HxORNQ&&Pa6nsr}H8m$Zt$}B`y>5|_iY^67t;y9@9B=Xu-T$C>(UyM;
ze}P5+qpnf&ytyZ63w@)OGAOnixF14Ts9E{>`H5K%QkC&zs|E!I9)AOcDU0v`R%?OG
z5Q@uY`G0O|c`rTcJc%e0t&S+1LNCczHh#dE@7$Xqg&F}6<~4#1k(2&3<|7b7!qqGw
z&+>mJ1a#UZJWdX)x`|=BN)9dMCkAiMoL`+RD>k1OJ<s73d+|&EeqX(vj9oDlLdjt+
zh=dhzd7L5F-!>=~(cLjy@)lhyN@iw@==-t%j@g`k*Td7xykaj#miE!uj=9W@+zy-<
zc=u3Z&EfHQ&+9+x&a=Nu8?*=nPQmOxasoY?`q2$Ta1L-Op_@*3`e#*C#gFo5W8eS6
zCgH>5pY7=qVkw$uVn^Mdm0s}P1Ny&J)(dbjf7RI@->~eT99zJgHj&O<gbN1#!}>)r
z;*J6sfz12D9^n_{Z3b&~=U<Z_WsI5eOwzCFHuK&lMk=wjqF$RcL45{mnu@(2LvA1H
z)Zo{e6av?=0=^(F&Tjm>BVxKx{-Iz?<C)b}mlBV#eO_mB={_$OR%YD&C34P`N1?DB
zy!B~B!zfCO0$+f$A(J=(c*+L194&rmXf*hn%Ikw2B`NvLt56Fos~-klkWBdyzASYN
z^&6Zm34pyCydaT&L_^OflVqlxl*!TwgI@wPncurSCb;1<Hlu3)6#X}c4!HrX{Q6Z2
z;1pb`B~#LI1G8*aqaoS`ZhK`3-^XMh{P*27%*^8Gm_W^utP~BJF!H~siuc!cWmx&y
zE+5Z1GoU%?kMhQYeON~{YwV<^K4<mQ$Iow989(BVK_R7p{F_Q88y8Q}xQ-3;raz!?
z%Jx=DI)@+^Rj5;q#|Rhh>Bzsm6>v2ZJDyR{kilI8YVI$dng))?T<h$4uity|;k8cF
zK-u_%=bUC~HVCwSg~{3*zg}oxkCiXEyoo^YmMJJZ@-I{z-)O}g7k7{AM8OBeDq@ZJ
zp0g#H&Ac#el2IDQpAL^SP2AB5?sZ!Um>>f7B0u%cuC839Rl>sUJv}{nM|pVzGZjxC
zkxNn!4VENM<mWF-H7I3Wy%sCa$+fcLJuPtXpkUDGGN1&1wKc}{v7c%GAZrb`2f}`n
z8jk?6171;A6LoiGT0OBTppYMSP|iA#VgfVt#Oi(KT9e8%DGHw}gDhk6DM~?VGmNeo
z&xmj8OWIqfniElf6ui$00|q?fEo@CyRaM19j1HZHhk<f%nF0TOO<v`3*PAj#>7wtk
z#$46j2vy}sHO<2|U<nL%Te-QNcXSMvA`<sA_*7L92;2{W!=iJ5T;~Erg-l$t#KjSL
zB2LEho(mI3hNHetqV2ZX3X?4H^O*70r)O*}Cs)sFaEyab^Du|elZ`VjZ(mU)ZXu+}
z$JzO<L0m2sHDRz0<*?z97@(|x-_>0AdAs}bllK@6UvpF(I1q;b#%=Midt->SM5@n$
z#;{WaLMy3RjJd+l6=L{>4pTpn<e-F?FH`>dQp)(im^fR+;P?-&Q*)E01;V1DP>Kne
z*pF%mpMr!Xk~-|~Vq84R?<k%*J2_b~`H7tpaoi@V2}%fyvJtt`N@%9T3XtIv8JXFI
z1#M6y;FwcvrgO}$w<vpFA?_a-*7e8%uZD2Cy!p}Pq?UCQU_gItHO&Tc1QIK5<g0hM
zh^R;ZZRhf&d1EdV5N=un%L4nu1||Sj``XcGWHtwCkT9xI#{~-~p0KW~sU6o1^R(?=
zs@tqX0ePa0fglMC4MBZfnS5d59#{kf><aSoM`?G`P%{gsAxPJV6Oafg<(B~BQjsTo
zCaW}yMQ0BA`T5Zwtj)1pe<ae_&UL?ce!m4L#57kWT}9*OO&u3aa2<2+Yu2$@>fO^S
zbq#3sG%(7`pN!u2i3(x>7tKFjzM=IdmtzfxhD+rpf&;l|iqZgC79EijQyD)z$>_J&
zWmq|1<Lc)pKYaM0aa_nlCcy5-{Y^BrQx+*@8TcUomhIcOS63YmDtI0ZF2q-IW@5yj
z-S-<f0l+T*RAAp<R-4K%bLTvI%}r-86goUtd-tA~XiLA?t_O}Tlcn321dU!cj<9_h
zTj5fgL<HW)D8oa)dCL~W&A^E=Km?6C=WGXaw}pbLPbfTJgOjUw<5ifCo_M#ajpw%4
zlQT2$FfrF-1y(c)2?P!PY-&kJ2LgR`UmoVj*H{NrK$RrSOz|ct!-|?%%si%b>+4eB
z79GQcGlWlSK#hO?om<X&&Phl>ppitv0;+>3_M+JH%EuQWUHmmJa`+w)1_Mw;vapzh
z0yKVk7i=BU1oiUADzQ$kGO~wgZrIpP2TLsmZoy`7hT{W`=eUs2EZ&ZEhN^6=r#P}n
z#n4hZ*3f6<`WJ<72k0JbG0j$sZz4PJ<SpnuS**K2L>6jPks=}{h6=G+cBeq`dZ=2{
zccmTm`F~l$CwouUT;qKqNX4G!LEB>jg$T47pgM3n6uCm`#+txz7HRuz(32-Wj;Uv-
zb8|hFdVc*_@5P#`zhvJ;MNT-CV1C{cp$5@WQH!#r5?FVYllf9d{Y|%gdb5h~d6ad3
zQ=`fgO_9;U%Lm?W-VBWJ)2`iwzDu&QyO&m*{vnaNitJBpy6_LTK|61_j3K!mjkY#+
z8l}s((pB#ExF75BE!XqX<^K5!qG!OvHRa{upB~VMQ>i+{`|5^C$|(R2`JI{lbNr%#
zK{w?z+Bj8`n95%0GK3m1>QdK~Wh5(d9`M~dcUqo-{`seM>7+B#CS_B<_x+&n2YX*b
zgF9{Tu2p_fa<c!#{=uWcRYbO@$HZTqW}rgQb$;r)QATdOrr1W|%Bbm{cD|w})WpFx
zV|6uthJiZt^c=xe5Hz}LWK=1gW4iCo!QIZz&e>`_X=j)}=e<%1EnnSoN5oz@FTt7n
zU8n&2#Y0bzF)Kn9d*_NM3OYnZ6%;$ZLATA+yp}fbtG~bZ%<qdWDeCqJBSPBJ*}U8*
zS-#XfcDm^z>2Pba@Yc+|&9!GAjG(jS-B|KRcPTmOyA7O%NB}{#nq@cPP)nkM(Usow
zf`+ok)H}jnFhP>zDcS2>@_4l6fsZjvFoG-qL1EFDfwb#{Fk$^TonN`01`UA_1%@Gv
zR39_tzkfJTWUJNMZzZHT;hlZH&{AuNmt*bvkO~k|h)f3*Cg_ykgut2nUmT*_X&!8S
zTEnMll;|fUGwHBfQ#A`K?CH!h5OX&v7&ZO~3YudrLenS8!w(KzF1^J9|J^at`%kD_
z#zM>4q57D5du9hO)PHJMMk^8>{dUWR?LXy8(QBEEn)*>m4mP%rP8}T`WHZC%_2{iA
ztSKMm&SPwp#(4|l=Sl`X32MLuPs|uhT&Sx<jlhK?g9)i=mbF;y(WcH8w2}P#qu@sV
zE;PTE8n0a+XwrjJW0DGEBWz~*6Q4hKI(sP<3W|MVPg}wS465m)#oV4djYi&LcUl;)
zg<VS13l&gLc2vKF{;0oaOHJ^8ra-hT+<F{{kH-KXfgu<|`o;VwK>$39o%@m^o@($K
zF~k|}9sw@~uEo~gJ_cHhxw&WU#r}Qx2BYi4?H@()wEl}ue>X=(SL5ct+4LublfC$P
zi>ggC`Vvlk7Q2DJZ=(b+^ikQ4AVn<R1M4u6&4il7ygUa~b3j-?u`gW6gmLJ(8iBC)
z5Sa@=OUfXZl>8ouF%xuTK?v<`PA;DkY4Gzq1o~z)c}Y*+v>-eXadjQtoo#%DID0kd
zZ|w5#f1pknlCC=I>*FSC-17ba2#}xTUr}!-gf+dn;&*fx{ot2B_aSPOcQTl4yh1l`
z!#@RQX3l^Gv?#lYK)3yWUMaQ|`GWty|Dnfw=6%z!(gzL5B##UY(H}Tqpka*Gy!+h|
z!NHioEn(7l)Jr$79bzK{mBX^*78rZc)7y(#5~Aqg=HW@vd;Rw901|1G-oUD@Eq6ow
z_esFg%9|Z<?$M)<ePD9r<>di6`~3d6v6fm~yoPdh#F=~efaX9102k5dKwSpj`wlzC
zj#El*TwkT_xwv!MN9vVw@Tc8Rwl2rS5CWwL-a2$q`<d}8L8bG#BcKWviKQ;!2bg4N
z@dGCu9~^AWFpy?r5;T;tvdRb!ZUU{YHnJ3>yowHHZAGu4izL59Kg;vUKp%3celbwT
z4bufCb{HNHxsEYJi?X_goAX}Th_dw3aBt?==6&h)2`(=7H-J;0NUQO_fEnuP?k>jJ
zBZq*G^x>B(>>+hf!|<u2BOfhSNi7kUkwpq8_T!dZ;`!kY>i|sbzU~l@-a4TFpv26u
ze2G#C4UswuX@QBugFpNF&^QbG3#W07fqp+J19#&R-^a>vPZ(Yq{XISy=6sT$KTS{9
z%1VF+<c7i@PZ0nSnkn!wa2#<Fm{DPCYdbeJm5-<w_}S&lm(lWszVs+Oc)<A9>Du@}
znv8#`PM_EonnQ_U7Z`HsxXRcMBt4MiY~38B$W3=BkwLc)IpqIl-~NC1x2UiQg1f$I
zYtg~|zC^>`;~yEkXnJV+g;dW!l_D~0yk12F@MzUs?w{?)9<U2&5p_;(^Cq&L8feA)
z{K_|rRe}ytv~j&iUo}*h;UPAf>^>-xiQnKRQI5UxkEz`(b^_*#s%(;_luhM&>1##K
z+Pu^G8XuIBvei-;bF0Kc4Qf|Xzlc4R`to<<{1cDpV;4TkT7!?0^)!wD^Ijou-u<il
zn$Ka~u)hBfOhpO8ts%0K$K%J3!D9m_B7>U*bDt1k;l1TTALe<-_l%xRHeH_<+qj{H
zoDt^lRp_PCO@}QYQufAjU*bi=+;DDx5Es1612xKR+aG*#baZU!79$dm9ZMB$C!@lq
zC$sv&mIR`*l3Eq)E}Zr;^YPkY(FoYyo6@iQ2z!OWz2PfaycFN`I{yS=l5E>ew}HC;
ziEls3z=TgC|J{##HBAd{?UL^)MyaTIqBIurH~YS%<Xl*O=)8+C3JRfaR>@gE+nqD5
zgIG?J3~~ApoW{#1f;dVK!d2)!KtC#lozC6^;661MP}4m99LUTJ;tI-?7@8d&C)?F!
z-<Nw!zmQKO>gm&|A6=-v?S0UxZUN-u;g3CyW!|%Gb8UBz3WSf9=Y~f`$y}_hs*PqI
zx#=B<0;k&)`3rVj_~ZHx8Vp_M^si3z=C_oU={JtF7<L$<?DqTPwRcMUNR7XGBWImS
z+>wZr)9kWy`wqWB{|hp)7XO0?#{Hem2Zu=dKE`0opvq?W9iHUD-20Yo+vzc{2CpUf
z8G??gsrlcCCgIe7yK$m~m|BKy(BX{fd~-81eovkV@Yaw>k*D;%(2zg_46_zIQ9n)W
zhzHw#yz3KoUoZOj(H+4B2Izn@LN@m&Fo6V~9E~AA^@6Z6I#t4s?=U)$cK?1N@#4jc
zFl`l{h*prU0yw=PCDo@#M(B0Yh4ag9{b$gbfxi0JG#!sx*4+*((8Lkq13u??140hU
zxA+C+U%nulE#n6iP2o}M-}RG)l%bdYs^SKXFOjNj$JndK_7Q(OS&_oJY(FLVRO*ht
zewjU`HLYJcH>tC;a~gK}vE{{`V`T}Q!Xi&8J>MYSC3BJvP<mSG;@SH171#`04OCF;
zmo^~V!ad+P;`}mvf_5O(K<1nT?+4oh#ODloLXML|uKx!ml#ygVqxHIIpD&S{WD2%B
zQUaOdO{UrZVGN`InJ7#K1wUd(i_yuGiI6k{HiG<%pZ7a*RX;i&%y@no5y8a3V64G_
z>Yd9|a@Ini^t|-8!8jX%Rj|EYx(b0_%K&yJwQuj=M1{iRqR#g6jAW4P+b?#2<0r%f
zgY-8SSqjvl5jH79A<zkOmhJ5-AXbCjM=t$o7Up0fz<fgB!^6o52GTkD73?ws>5IzO
z%wsb>{r~E$q-$pZJNA~ES~Z^GGt#gBc*e*L{<cA$CnrG;-j{XiebFO|{shAKcE#iV
zyPg=Rg2;~B-Lu6Vdyp_>s$UANh(zS@D1ks*-1qOH%zo5jf7Xe%SnDI2k&~2D5sQdQ
zNJyMMoelg2<-Gcun$_uf&4`0HnEotS5B2oqw)OP9oJhlO+|1Gbw?gJ;HJv~*6I$59
zbmp0;ZMT3eZEdvN_0yis#}2Az9HBIQ85Q-tz}Bp!jC|yxbXXG+)u6C*OpecL=G7$c
zQ96bAy5<oB)82=?Q5+#pu1kZ-g+kkOH1LlLi(8$&%c`n;1sbbJky&b-T*r(!t{2KV
zopWqHmM-itz5>CzV7fYca)68}1rH{7=JbV~HP@P|376c(YLgbi6w1ANC290T=DC?3
zAzI%aza24X+UebdXJSfIO2-V?iyS5Q;=_aoQlIXTW5G{Wuln$VPzMwE|K^A9((cK#
za7lUDD4HVq?7645`5{XK=pfV-kbI=ZBFpH+SLyKo>kv-i3}L&W4OL75Niy~%ch2*V
zBZF)YkQDY8o};1pn|15BforCnX?UZ|i+XAIX1PqH1RzE&^f(7i9G(<{Pz92ZkL$}^
zy%+l!F+&SBI2X9YB+J&mshf6xcg<tFncCk0XL~Hj4|tn+nKU2zq_pbnDIF*vFoA+T
zn6EE+xVdZU>w^GT$sMhc<isP>xiu+8-ps-rS(2RprMT!)`S3jgJ;R!?{_$~HA}Z}O
zlars8GA<zM`=Fe9h?H@g_k@NvUF;8`{*28@hqJ_(k&G=YEC9F1EJJB&=>f6bijdGC
zLlpq8ajXy>NAXGHp3%;+RK*d4Bk1(=(dVU%y|-oe@p7e6Ug8o)T#O*i=LBBIj(JW9
z-@FYIcJ*on!nl33v^$1K)0ozcl9WPx6Qg^w)gy}%ZemE&V-jWZqi>XfutND7vP6OK
zWDF|BheI@Y%|Ssy>5lA)@v(hFq{gpzGT^6~%<G!i>G#*Y@Jz6Zuuy$?d#~C-@)1^2
zx=MtCDLItDm;Yx&7sa<r`PbHd7uDk$N;(G~+j;-XKKc5<fFm{qsB;^|qS1PZjiYZc
zx=~I}?!SCO$pWJ&dJ64PNo{RuX{oIACVvHH3u5{lhLfrzYKPF?|NbHVNqQ`hg{-mt
z-z~rS|F+j3xUYeK0y^*zxsi77|H{ex7g0_<C}jK%y*>~}Vk6U-FjE5>-e`(AtNVHD
z08x7v9k+3Nny!vcANoxKKF&2rvNCR;!y~7sL7hxzvab~04caWPv{HwrEC?5*9-qk)
zxsU>ONFoR?t3X?6ecu7e1WJGG4d{@_sJqL{e_=zmXBhZ>{I~)ALGYY>PY4MK9X<Lk
zD@*9ek$%uRu(bjDt$}$#MpC1q#wI2<s<~35#2qv6t6W$a2|G67P96%|jwoKdSc*P7
zpWmRjpm?%t_inej7G;2C(nK7;jec)d^0Wjr)D6^Vk^8L3Z0;vd@Esf>*CAF55EnFZ
zW;!}0bbPRxVmAYW{<wV`;nBlKTP_^_X|;rj1?YN}B_eL2@aRuUEur;Yce~5^R4Pl2
zhXxpv#^;P=t<*NtZ283IyrX=4a<TG^yI}ifE+J@m_`5Zw3z-Fi0w{<bG%wPPDj5ZK
zj2l!-vBUKOxy1|RdS%WKd;|Uf{uKG;=LgS+A%1ucr|SeL2|9LpWt$G~#b3d(xeLPc
zZO#^`Ly(6%WAeBC?{7yi?gy+X$d`P6yc6z~o>#*djBpxi7SY+-v%NwLROFf3t;vwn
zUH%y?M^5X`?<VX$F4IGCSOxq98EI*-;b?j<0=E8p`18Dx**}`=E8UW!G&(X;jh;Hf
mzv9Jn;__tc^55InHHl9u*8_6e@*dzVglku1i61WMJo+CJNe$-!

diff --git a/docs/dev/articles/examples/timeseries/timeseries_classification_from_scratch/unnamed-chunk-3-1.png b/docs/dev/articles/examples/timeseries/timeseries_classification_from_scratch/unnamed-chunk-3-1.png
index a2016e7967bd9864adfa6c3b86abc013556d48cc..bc06ec3e6a13a02e586f2540803795e6b667bfaf 100644
GIT binary patch
literal 54970
zcmd43by!v1_6E99kQ78x*>raaZ0SZpdIQqZ(i`bk!cBK0Al=<9AR!>#-AK3e<}Uos
z`JL~?bN{-3TppKf=CjrsV~#oEeaD;-sH7nC7~?qx2n2fk`jw;#2=rhG1VXw%djO2=
zQprCCzM;cjY1)H8m{0F7q>*{A8xV*B^jcC3>hfhb#rYk{`t98!z3YLHkw9Q2Dtd&U
z{97bh3WM0Hn3^h&J_c43&vZQ5zVLZ2?XR~Lvgx}*y%^L*exh`V1BU_U5yq~s;v8B<
z6jr<Jb4OjQrmvb1fyJ+KPhLBpdkG&5K9N*L7JG94Pm#r_`Qe|>Z#-VX{tUH|J_Y|7
zfx(ec|J*f4+4k4n&i}u^B}B1l)>iim9Y<N*-7cxd!D_A<T3kDM^ZxUlTjcL9d;LA#
z8e`eWh&OVwZZ?{?2Aeu}{WVCd@viW_W})U9F-5Bbg(s3$Ss`cGPmpm5#HBAoP%wqa
zL>*puHkM8%>j`_Rn-IQZEc#LXR#%mXMJGWoJ|1L=(WZqzn1ZVkoI8;J!@tq_Zr^i9
zuVEE6ZT~~5_mK>b8_D^8!*v!)>pbM_YOrRZ#;?Q$nbUM=OXkTcUSCX_@Y%Q`(zn|U
zMW#WQInGP^JH+-xXPfF(OEq(s&1TG6X~5<HTIZ&4-F~`>`N!J5y+Q?{%}+l|7hN`!
zPWH8vaBFKR(0m@Em)%gJQ7!c7VjYCd#$}=BkFwV-$IHazWE;l~Xu6)j8-<%!ck8pK
zUuJVFHf(2mdQ}=M&E}Q}TYH%(nA&s}t$Mo+(x(W6K9=5Q3V6Bn?{53_hum!Yw%B(X
zk&tE}PCS3&fYaFC((C^2_qKg$`{NW-|3f}Y?5jm7a?AC6zT5o}@!;CaP>sp(cUecF
zBm<f{8Zn7+ajLv0b-`qBXX)K)7u_y=7E8!KnQiXDH@y#}h!PZ6k13xZr%TeH8G5X`
zt+rDkF;;pt+t3_Flx~)<Z#}{Ax|k{ZHSM*t0tJb?b>3tWDh`x`#I>~C7GgSO23X|@
z-Z2VpZn35Pvi96)k`0b{xqd;e>o$D+@{*9!e)*sUjNkn8&RMI4B5M-txZp1$=qXmN
zRtCMj7}rbNKS|yfVS5}mH`8E<=~tx2j@9Q(H{z^(lfEDBrD^G(y%OcD`w8Ck*qJo)
zQC8aVV9lO0X(BCJA`d*}XudK$Io0x$<1qm_&UEeV`Tl`?b-Pym%7&S+?Qb`yvQ%Ed
zAaWux8fN@uihwN9o|o&0GZGQq>veraSt3c>pR}bmn&;D#?0UkT+dGR0>*&eGYkhCe
z`gyK{B%%e^m9{%~&oiOibu8P4yAloS)Z(<#9irmV!JD6Z=OTz18TJLo%T6<e4`(62
zg!J767C*Wp4n!==&jnpg8?4x=Jc;zDL<+y=DS;%-yfu<@y*kg?>(4`ZN`=l>m}^<s
z%Uot_Y-`sCgtuaRgIJy%o3aaR2E2|F+Zbi&5_@uUAM~$3Wg9NsaAErStYqL*m`t>x
znjkSsbzTyRo?dbs?{!&|4OgyjNQ@$jl|gUTLcNs1bM>Ag*!)LzT!qd%x4DX}`3cSg
z9rBLd`lp4JxtFJ@?9<%hZH)0cPA-l%(gkWQM3F7Qv6r<yPpq4Hai4tn&==nLyQDB~
z(z@}a$OtEuYyHN|-Sea37+Ro?y&^Rl)xOdQgBT5}4+WZAE7B7dLZnx^r@`Jj?hdAh
zZF8DD`D&PVm5vAj_NK#L%hJ}fVVS|X=Bvb3kq<ww^}*RE9<Qm<5;Zk@`2x0eTk5$_
zUl3*(bzG&8y|pwOtZMP@H0UE2RC4YbpS(eAOXw)3U1!ShYqGl>1W|X0gi-`tNZ?}o
zP=K=fH96?jo}gBmmFiSmj294a&o6r2p5-XmN{=QxFStxrUYqJ6!ueXw_opt0dJWsE
z&rcjvDyIg`euVs5%pM<7J#zFq%+=Q0In$(P6)|e7;ai*&iK;kUX+$_|y_5@{adu+T
zXL1wL!#%X(@_H`zWbP{(kc&Y@qh5!7+=#2pc!b+;ba}IWvWw$tKE89s#@$$bvyJ`P
ziyl%%XQ9*HS>~`L8eXqs<+KFe>pCO4!8}7@k%!XOb%0l>+dnSf%+?2AI+D$KhBp2_
z!@BmXAn<P5uiIU=El~B(ajZ>gTq*pao7Q}Nh8D5nR4<A4vi{`*4|&_U-;EMBh?9)h
z{EEKmDpY7xrVQ;sX#l+~81z2fgR{S+GdpKi6l}F|89=FRJwi0+5^4qwAZC+sZdca$
zyEg}gu6io<PZc%2=0{8C${_Namj*Voku8`;cq^|aOG+n8?@TR6->L6WGM9QD`{uG0
zRlha6xj8u3E6eL`yi2x4#kupoYw$9_ke(1ZnPV%>zS?DGe_6z2y(n;ZRi-)Q`8?iN
z_K75#Se64L;3Ga1kUuL1e5jM~@x32F+mBfK^W8^)26F$i&k{B|u&DpPz2&|ZNlH6A
zpTo($AQSF?Dly)Z;LZxe&Y-a|C>aU4nUB%g_IPQ_)%HlXT(_GHFqMEnQ6nQG35ky0
zSY}}4sSl6k1dC2xSxZZ5cUO1f5(qY?ETijj)T!J$w|Vjy3u|p-qsuNnfI>KJaLZw1
zeLeIn?9CfV<=umW+>SnUWqeX>880uy+k+mzPMcc#rDfelr@cP*Ru9DGX82hc4f4;Q
zKW}(b9bSU<N}XXaSkbv*J5>-Bsh(Fr284ryW5S&?(vCmHs2LOQ8GqAjg%R|yv9Yn^
z<Kyy$VD#W|i{iXI;+^J%vbQX8y}iAE1}V_S$H$GDiT?QjcG}?N;-U!fCWE6(4{W&6
z9~qrD-EQo3EEx<ld%uCl0<lOJXy~_)Q8k%#DCPW_jq?vb8zOYlJCfIEQkE$GjKKU^
z2udQd+e!XX1iJ!SCmx5P(WA1Ap~=+rU$-tO^`|jAxI}-+!UG%U|MOvj3aE<&obShL
z{x-4yd-z@?pM#5|U8}p@p84G*BiG$%?Op1ssqXFctD;$SikXGh)i4~ZF2xnG5%j+a
ze%yIeFAnEJI-GK9`q1H6+p@?Zcy*BC;N$z3s(`e69c_OdCViW{`h5^LSj@u0z?5d!
z&Qtor{T9c$xW!<4EW_tGkzV(#U5^O#ZT{k_R54j!RR`L9xT`I^d+NVf=Y0KeQ3%l_
z($wILHTEb<2?iM9A8)>&{bSOFLzk9@^CUB7o9Q8wow?)qt4Z6NZc`kP3yHpm>)O@e
ztUyb>GeRaFv72`7acI@N)6~bu>TW0}GaJvRakJ^8FT6-FK~p|}=<Flm_4Yh>=DNJ2
z#p+r-tRaqMA$Y#fh5kb&earAm1PcLw(+!M21s$Sm)y|=SbCASyF`8yLBQF1#%yVo`
zRh96@?W`f>n5s|7ISn(;t5HG)7ns}SEFSX+@|R<{WmxL(U&*L!npY#i&1UnSmy{@s
z43NF6G<lKT;_7`h`tGW<rvF}>)zd>Vd#nX_+p2{(4hODVGlk?AdE;7~nR>)s=_4I+
z?Mbg+J^-QBN9W?tx?KtG6<hC>>U#XzhTbt3#$V0N?78yJyUrAN@19`1WG2^fpA0n5
zJT)u37qq%nFaEE~H9BE+3lHO9pjFo^cyA@VRu`JKY}s`FJ^J^n@arorY={&-`ZrN~
zRsF5qikOU*sC*I6o1J8w*GL==?@pY~W59cl02>@M3Lu=JcaJ=y*x<eefDfoK>7+9e
zEGRug(fD=vS<2yY<46!w@aXc^#TxmbH5h#KDz10P{e0o*b*)*}t?%7={2;c_bxg&W
ztC>gl$=J_uD<_v`AxnhE?=5F9>l_=WH~(!U2bYd0uDcbH;<>ZFdQ~&_%jFXukqf=H
zlX$(mbdpzz^)~^x8mugZkhLJ|T;ZdsyD`lOuZvLjIi0gzXEQY@+f8T9dia99QzXYk
zBl#mSe3{lunNz{@Is5tfd6B!z!OVpR{}-P3@w7mblU?DoHxsY#by+iOL$X8q8qKEl
z_P70@Q0<?D70uTEuOMqmPY?knPxtaQJ>;*L{l7@ZwvCDgV&=KGo&UeCqtC!F_amQ<
z$}I%>;sxSur~i*Xq5daY@lQ4}fRq9ECdbylrVkgslv2#=Yji#+wDr=m*O-``cBI$8
zYW98R-y!%{KKW+?0J2Ae%f#I}Dto`%#l$*Fh%8?Dt$$(3&P|$(w=nxIskXPk{`^7l
zZ?Na~Zn?PyTboEb<=c68z4{%TTA^F14#e&C`N87i)Pe}t)O?k<w~<V%MysUhWQp#e
z<=bYrYKEW;1~)-6cn5ALj#K0OGEDOJ!xiQ3a`o(<Q40y)87hBM6He(*Q<9VUgl{4=
z8_V(mdFiFjn?%X8#3|xPqF|fpx{@}VoD8U~Kx&Sem*?<&n%|<dg@r|3uH+yCTG&Xm
z&Fx|Gq?hJ+72oEX<=xeiFOE)ql|!4qx9efswYQ^Hi`R5n-}?6G5b!lVP8&o8DxD<h
z#G$69R_PaMh#$r25^)hE<8RXOj$jM>ODV<;k&Q-9W$#!09Xlt9#FhL(b-$Y(yVncj
z(i(2&g6~d`OVYn}q{h|Sw!qf67Ms^L;+;{+91gs9!t}qieptWMSU6WMlreK(m=8z5
zhGSlexYnz#wVoI}k-=2#P7)XtUONmX(>y;mIb*&XCmNS7X*%vzD^%}v9lS4l#_K>)
zYC>G?bi2^&9NZLb2;A)+2BxMslWBVOpAC8*uhX+5_Un2x4`%joo3HnU>FAxeoe_t-
zs0Sjq$*as3-dECTil@~=*SCz7Ezakb<jr>ncS1SaU-i9(8D_(<21WH1&p&k!GQIt+
zl&olr`e4p!*81_|$6pY=!^zo(tO8dvMCxf5gZj6-6+{Ar<eU}<g+BjqMyaeWeL>v(
zBO<e>l}8#bytZAVirrQID~30v^`Gd(2b3oTJvYqh-}HT?cb?uGd_i_-z(f&*db{#T
zB$aR6XEmp)RP%JGur{sXw7%8*!&vcbi6Au&q@Zf~mOJ%kC5Jl9)M_M2O(`ZeF<JPx
z=PMa!m*fcV!_IJTw~d4IdfD#nsIBayQ$bvFK_NxS=Zb@gcKjlqO&oeS2Wvkr_MMGp
z#MR@b&9s<I$C{iE2o`l49i3G*%#s0;+(CPSlVpqplKV-CN`z=E|JiKj?t4<Htxo36
z4szj7YRIX@?yicydFE?g{>)_5G#t_R=qnRJiBc?vMxLxcGc>zB`N+-X<QVoVJ@Z7p
zWulv1=ki<D?7sWb9Vx%bMWm|26s-^&Hsq>#BwWwz?%nS(9XU5?_1nvHVfS&Y!qd8l
zxaeP)<`y?n-|;MNIte15dml*x_WY9GK|-j1G9V<Cdvkh|nH!&TUZg$yqz9ScdeM7(
zotJz`=&;@BAP(%ln{|5G&9dcjzSmW?va&)oh}x?2VF7Q?D2>3PpI8#oMffiu@Y5yq
zG8_-cu(T$K>BT;w>zimN_Wu1U0u7B>{l@j>v?ee9B7$fwgvaTM%(mdHGlAY!nkBZv
zr?TS>_XVMye&2;ef%(S9^urM|wxUYW+%jIVSN{tsyw>&DJE`3yz<4FJ`&qF@r_6Ej
z-ByTZbOq{fuS&TuNm`mti-nsCX^9hc)tX;kxSLAFp@52_2VLLaG8PLs*J#uIOw~+f
z?5WM9+nXX;NYGOftO?8nGUGptPEY`(I<@>m(<b%&kO$(+AsraEBDsX}LUpgMIK8(s
zIzbWMLk~a+)p>8u_$?L2zep5L%__Hu{n(|JTWK7OjJ`*7N!jgqk*4{Fl3AKeU(yZ8
z|KH)%UlL*df0e8MFYwfVKrfP@OMCx6sowKH3ILtKeL|cC()Zf~De-TPum49ipyMR*
z9-o*9%5b^fs|BFz_n_H51v$B3J8BX>o13F>Z*ctWa1>BQOaZL-*LC>$vnPy}6UWxX
zWW57NL~z0#7gbA3i%vcXe^nZ>>5ZUg6X>AX`tTpxe8y(iFw`LV14I6*+y4v)03+5%
zUBLMOoT>&$`KP70FiM7&*JW2qv!fOM3!CWykTvRmjM5;M<^S$k|0s_iXpbgG{zrvm
z2aw+D|Dov9yF<Fa1LS|8UO-LDt^Wah;{jw%^Ivc{o;lFl<==F-W!M-5pq|d_n`(!D
zL%e@M?LGj0KK&aG{>NzScaf%l)1B{6nbs?zL1^*+9o+vx4}i^p=g|KJ`Tsd8yF$M3
zZ@T~GIqVuwy;=Vkw*KJH66^P0h#RmZj_tqcE`Q4Do&318nQ+oG@gsntW54Um(f{3e
zCEtzni?rd_2jM&WZ$$qBDSy|GmQK!Ome*?f>EwLkM;fnwZ=LAByy5>~(IMSc!d;Cu
z1CzL!nU1qrhFDzCm%38Y|5E&EbasV1-vB?Rc5RnqINfvOeB@`u-;`xKU6cQPfnpO_
z2-0xiV4t7Ks|_-Eq2I$V!=KwW`%g?jY`Onyf-#(T6Y5ZPLxhLyD#vG*SPJ%LzZ;JL
zL&4g_|MZ90U#zivvz+d{+dJjXwoT3!;()Gxv!N5}{VBm0Y~=spZo}G4dwY2WKI6|f
z33mLbkBktR61MD2%#q0?n@z+2weHGYH>B8W>z(fUbQH$<uduKv1V}C~14_kT{ij38
zp2o>4!QQeQUamB1;GtJQX&#~0qXEA4cmMk9CVc#Frojh#@-$CHHbnSgL9qXdp}=t!
zMmA|hMGt3lDquRunoK|b23cvb<7Aa_-P~9IDd<r|tW=!rQ_=q;f?!;!@>UUbIe#kf
z=lk__ppc{0{1^KS6TE9Wyk3-fc5-k)-)8U#P}awXx}^^eI8rV^_d|I9MKmM5??{xp
zZ-mI+7K4<aemOaH`xAaXEChjVV)#VS*-^;<B5~~e{Ge~k0Xz)yxY!bZ=JpCVtVyr}
zO2_}N_XXE3wD-ma2yH0(bUXnBLuf<k{v)amGqO<}9U0h426_U_1)$l6Xqrq<+t6e@
zR%8F&se%U-Vqy(?@y+m8^<<<YlD6;a>k^RRNP*-h<fK^&nd;0<3obUi|LI;*X7s&%
zoikeJCAzq+6U(ArEq9w$FY0uT=Bon%Q$JPlqu&3|NI%Rd#O&;BAIq0gMvi#@A=$h~
zC31-dNMwIcW<+}IDk|T9M9&L2OW$b0AJ*4{f15vuy5PJUreuIfa;7HB-*hDZomF3`
zXz$FnimYhlQ-NP_sFMQ*nX~bpQ8cCd3}3C8{?EW>a}j(c_mE<3!_-9(w{^}&11)S$
zXUJL&wE_zLt^s7{zh`J7egS1I3{hl3Wi)Kp5r!vsP6~Jd_i=~qvvfyEO`YmP{hyZc
zf#P9;4yEZKYq()`&Ke1#5!o43LMbe<=#T#|HNes{C7yzPD6(>MWo|yiE(Kh_DnSR#
z#<W7R47eF^-M2mNb0C|8<7*!Yd?`izlmsYGm~sBEXuuT8pD7^=uJqf*wEs~#bkHOr
zAeWJmCtE}=@6i$y)AfI-48vIndSszr3BbvGtEZ#5Qs*y0fDZt%_)i|>FNzJPKzn9l
zg4<318BX9sy-!`=6~<$^(j-;^ee%naEu9eTybw5R+bx2?osawRXTdnz_riviY5vl*
z%~ZmG5~Pej|Lhs?4!ZY#M(;pGdc>R7bU@_bM`*(^+J+aye>>?o5(wXcsi~E%UjEZ=
zA$50NbDqvDB>$Lnn~jXvr&A*(SkomP;s^)paiYuR(33T8Q~IE#sR|TpYr6$HO(wMd
zoOmU)Fx^HDTQX#%lMRE#7KWNHKw^t2U-&fqnT3FQ!2|E*%W|uae|29_lI1P3DAB|f
z)M9Cf^xV+dh&+UK+|29Z6^GEebL3%-tAY%`Qvim1{+z>;A2c?ZnU2J5D%oIC0ex`s
zne?`oGAK12(H2?98^u>k{e?Al)+U&L)vRdim#7|@o15X*f#P4Oqy4Z9KjVp%9fU_M
z3X_%fVmQ2?^$7l1oLT)1xUJldnvh*uM&>>V1IzdhrUWp`z>q7q-D$keIJ~>VFrOvZ
zf!9Vw2h(wIklapK-Jz|%KvrYo^{xJZ){(b$W(lNDi|O{Z?o8d|nl<;Q{YLd9>}o54
z2UEGf6yqP+>!*@d5)F|K+t@@><2*Kw#6q)C22JSxT%b{SXeH1=L!tD<d`}G?%|ZHB
zW&ZwsfMH*tZ9a?$Y`E!N((Ja1KDbK`+jZObK?n@i?{@wk4!87Mvd9;_Vu)jtQpE#=
znx2@lX!=lxf8R`#X?_e~?x%^_TwFejcd48EZ(bmaF%%U>Bf}}aHi7r{==A7<%6izc
z@Tl6C&&2VSbxF<c+ijqDW*W%I=rYD<L~@dn>O)C#!0S;`g|tZsZ`OKOnMb*qup4gz
z=?AAzlSvlJ6C=RvEf*_XkJ+0~M-t6ax6a()a5zA|$>#xPEfg0|a&y>4vZAI@Uu$Le
z-Lo(A(06KLgNenexP-73!N0w?J6k(wT1nfU#;xql;q78QsSEEWceY*-+_?U-B!bnb
z(`H*bYH@aYYBe<keam2Kj8$8U1NXixJANs=SZ1W7hqNZLyLiCFK+M2;R$UeH-LB8m
z(-Zv|HEbD2LPA*VkX)^Be12wT^N#8<w%e!W<>gx;p^Oerb@=&|QTRFon&f~B3F5=)
z7<tRsa;(?cv~Zo_z}SJvsNltKDK&b;?ML>CVqMqu*h2N<AQSI78)tWScNT4qcUvmu
zg>r(MV1?w(e!K2Kz0#?9)>KFJvDrfxL2Doy<`~7pCU<~OJZI2kd7TER$NtNs0fCnX
zW03AWLPh@9SHqhpjx9%@UFIE0cmrCmVi|%~%Da=M+DEdN$#JEW@y`#48XGg`2ZsSb
zb2K8?i6eq<TSu9iY;A5HxVh0qe|H&vvE++L6S+~kStSP^=xh<T>kJx-4inKkb)lO7
zU~g}~+vq$jMa1*s#rGCcx;biaT)xQF{PAOv6z8FezO;oM{|N7P);omBqGNEp&K)A5
z&b6r!Dl0oQ+F^Slcy#%${A(UsbF;GTJ#>?Nmv6NYk{RBWpW8LU_>!?~*0)j*MU0`i
z6h6N&w;OJqOa-~(28RoMu3ZT7!0CRF3q(2tJG~q&qE09MOPUh4jKst39SOyPd@+0b
zENul3Ie%}>BBc-8p85wDTifBkW7jl66xU14`irm;?=<&y`^BbRSZSDmaGLGAcNBbl
z@r&BumDVCbXyPGiObWT&>|y>6ept%**Votbqzm%NW$%za*c((|9grd8B`!UY8fHYR
z$CgpWv#>-0o=!zwNEuf9=UIT#d~|h758lpXgcqAnhfVD!YM9f=$_Gl9yVp1yu&V0x
z0;bcf2L2bkhR16GKnjC9JM<ck_7bHwY!^w<eu@>99c~(BzrAG@5usb$ouROCsaMMn
z3AqqhGc!%vA&OD@o<FDe1*XPCO?|OS-t9*JAYh{q%4_qz-5;c+FJd#AAcK<3I(3FA
z+Dh@YwKW@ML~Gb;ArGUFpokL2$~x65GII;oee@_I+P7W&^sLjq5}N(s+q$?`(%t6J
z&x~q<<0TW#XZ!-x^IU2GU$|k3UC_fPKoYgygEY0*)>6Os|AWPBxbX&6>TjI{&!9JO
zSxFh>I9U#H0dp1n@@X1z^L(g$TG*#gU$dXCKNVreDU!QeTB^Hb$f{GYT2^>uoG&`4
z5~=?QDJCwXka>iaz9svfz3H!(9<f6B`Jap#5D+M{?tDa_yy&y(c0X7S<%2DH8Z|S6
z`uPyYZU*HSmOPnUzR_r!R*LvijU7m7>@K{1L_nQ+HB3CpkGHL=vgFil5HEw2izjHc
z`TgJx4@Mb^j0{{HCP}dT>MuNzmi-hT98H&=?cXhMm{rV*KNKtmd&41TvZ(;%x<u|}
zhps4D625*drnWeLiZ<LZQGI*#+);Efw3*Ch6nH^tJu-9q@FW~4_La2DXr?7u*=?(V
ztEN9xIEj`Sn1iW+{-<nv+n~zU))8jdilKw^i3@6K9<of8YBF6yc7<iUcU_yZjKO?E
z`2xvFTyLsSdWiE=aM#q{NZwnJ4@I6KLKy>qD%J<=jJVYW7{$fqdxv=xw4o1rdd0yS
zPXHs99%&pRD8=G>zkf{2rYe31!Y88tZUNz1i<32>y~ls%vT*V+_EzvtGpB9-3%=_@
z77$cp^&3)&J5NjbtnbDrh9IX#-PD>VExn#D38c$I9&c2rw%2~FQro^DUhWx?24_Cq
zPH64(-<w^nIkWQ-G=~n)=;-N_j#tFq^RBe)k@?USLl6TS8N3*>h9m?OPvqi2TDdZ>
z+qVayu5z?*C(zYSjKzIBC`(m<&7=(C2J*EvzsC6j(^0uH7J@DA{6Z2wABiY~qfl5L
zYpmlAY*@-}_6j0Kwmc-?7uxGi4Zx!%v-~M5#E_L$ab;J)z(5^Mr~-#vcJ;VE3A+hW
zm@Ay1GEP<%Z+m+Mu}_9G0HNO3Cbcwv369F<qRNV03x5^9w!zgWZ=6sA)zVVZ4NXcq
zid<TExRS}+T%!nEeV`WCd)P&~5UM>iMO@2AjMGC6_A<^^DHjFSj}t?U$&Ou*{PEge
zC=*kL%{g})3xP70|4Xy}Z~OB`4SXA{u{FF>5466}9LBK%J>Xq6jW5c)h>F=gHaSz*
z)YeD^mo|eLmB@;$u0G@1r85d<=2RaEf|CQi%G=?PwBSKuqCYaDrEl}Nq-j-DsD4h;
z0&^zzXNo<F3MdI~*9Zv-G|qQ$Os{%zy3^!(@q4y>mISoHN{}z-83AD}oGkID2nk1V
zLuMd~SI8YT`N1-H<_+(-R(8Y14poTMGJjnxhnMy>J1T`#u`Rs<`h8D;>{V+Q;-=Bg
z-S_CDW*crjpW>gXhPFl}0-}24gh2gC{-gq3HnyOIDySftx4c<IMMa%M9^zQ;i|fV|
z+;t^nmJ|yn7uOpRp%!J`pjM`b4W|p^<oLL21*|V-BF!V4+B><Jl%M6YoksS0ZW5Z4
zpX^4y&?xlSLvA=wWY%KK?|I1$3&Cd_QkeuYQImd20FV=XO5`rL-<;Ze42q5mieaIk
zS1|8)|Ggxzm<aQc_-vey6_nVi8HpBF#B%0agx0-8!No0aoK8m!W{v&ysU`A@THy-g
z`!19aFt#g(Zp6sQD|hcm0O222Lf4E<!wM@I5-1738wL&#CcTQ21vpKwc&qGaOc$YQ
zTu_swR7M_yQyXO{I8oS>3KJUySaa(qOK-@__H<lu)K4BgfK>y6NCkdp^|LVf%<_}v
zDoPFZ1W3b9PcP)a>yisrG529{?&20`HxZOs%06}7qgb^HrHjhW=v{v0HwFgeIMuib
zwuG^=<`v5a^lN$ndbIKta?Dq~CnGb(#7?O;b#oiLR5p_EK!MMx;FA2VGL?Tgl%^(J
z$6g{~epY{NBX~MAmDXdC;N@{a$6;|#r!F*^KCF~yfu*YS&oeT9F<IXW41!T?E9)I~
z#AJV1`e-=7vtpEg`a7cz5?Akq9>~yJX-O;BxjfLLh{;+vI>Nz~(Tb9j9Tm%C5Vw2u
zb2*~>#@*Q|8yffEFfPW~Ce+2|l|$1p#zawDZd%wtV<MFh=|Y%IOibP#_3;Opd&aI1
z7j)dtgP;Zv$Ng5Ik;^WWV}*1Hh63WLY@ajZhF46z+gyj3T+=*y&ON;4Ff*s=non8(
zgzjgwW;PWX&;qDS&I}RUgS5N^e~{~}sjYoyfeVt8(p(H1#yge)<95D%5oT;Gy1Q>6
z38895186i%(k$OE)JOK2fmoNOpD!aMqd`iZLr$uCo7le1pjTxWBkV#7^5lT{ns{`P
z983`WG277IWaF)P#>+PoDBN&r-je!9tvBm$2V!IS)|bul>jbVUEQ3JC3T)2%q>s@l
zs>cH@Sf0ypaV?RI5SB<%|DmoKWl)<sq-c7hKElk+S3I`&vNLNnSRom@c#1S~QtIqm
zX8o#k@|sMfq}x4j1Z|!JG+gYHo>x2M`*MV$K|D(=lD=s{w}YSW^o5*n0cQ{=D=YD{
zwv)kRPp`RUfJbtAr<d{-XP}ANcqTTwzMulDSeUMLl6(B<8A69gW^*A0w5wcR7~QN8
z86<h)U^5|iLxN$*jTNZq@f|<}66uLb_+A|Wf&Bj2<2Jj4=0}-sfHpRK;a9uw{HR5|
zA7tB|p7)`kf^tV+S}BL%ppROO>G&DJIAXSNYd#q(;wV52Sm+f{Wjd)A{w#*YfqW^;
z10_hIjurXiHcB3M`R{0`s<5?98K%8^a+pyGZ6{=SjK!eR4H$#^7#x<H*0oc*%Gata
zB5c4yun`q=-QD0RuwVCZP%!iS&eSX)CQ7!PpyVTAlwVgA6f?+?!hK2w?j_4UMteV{
z4is_*W|F=mv}2%wVGmi03v&DxsT+y;)ym6N_Tc?@Fijh-ykEy=$0DMOk)2+^`n%<8
zlN)&F(=CB0z%3Y9V%Pi1-{UU0b}wCy)SF4e@>pVz@Rx@$!eDW|Ir4QI&%3Cs+TSwG
zNKoNkN>MzuX~Jc14dGiDy1qM%3!20%j6HIx3RUw#rUv85!hkXrFEva^<#F#~F+Zi(
zCsDbJAze;nqrcuV>9HSUB7@Awvt3Ju175x=zR9SEw4RfZc@b;qb{dspcna-*xj4<E
z12Q_$ktuBndTD>SU3hW5_&Ak3r2QFY*pYR<B-!#*3u&bp_lAo+v7D541{J6I)1>73
zW=k2=I2JJj%A}-(JMsk;WgZ-KmAAy1Z+`8ZDB3A9$gO`uHFR(0YDfB7Y5y%LI==$i
zE(uA93sQmgl;RjPCSyLpz~};gkn>bkeO%~6EoqbQBKXNDEz0n$^FZh{l5J!7#hB|Z
zXVV3h(_WBJ`_@VRP7_9xPzhLmvNZ3$F#+i8ycoKOARHg>dYGSuv>vI&ba}aIM3+Tw
zN{)T2Du=&(M#U4$xh`*+kA=1L?UT#??ysTgA1!wFZ&qQD8=@tXzB1FtCydLiv0>O<
zN{_y$b4_fXz!jd-n>!4W-RwJ;qkM_M#TI)~QCX;RG9DK+Y@%bE@rjGWa6zSfHZs0@
z1a8-~2ltg&4^4~6rV7%teQ%*>vHiYi2=D<;PtRXlZKqoh#Kn|=%X{vkW5MPIA8PS>
z8K9gk*Fv3kMXigyHND$VTz+ddZ`Rf%1p%%=AVy5C+QeturKVA?Zhlq#6lr(bH(wR$
zYmIvI+@u!ML2J0SP2eI~I3Z9#!+eZ375Wp9Kxb!Dwc(dBhjFs*0!9oESIm!T!?KUp
z;}2a7uOU8EI3S83LwkE~H!+&Hpwy+Iw@hs)fp$GvjU=?RMSQyD56C@@4|Bic+Q;$@
zKMA@#KR>7DbNTs3jtKGy`yp~JwG{7#XU8M3BR&Z1&LVtC)8#qjrGG5#_PN3VS;mi^
z1ZbAS7gbylt{&%&2spL~Sl_$&{HBH%s>W$X&7h>ol-Y*3<s3-a-Rq={f48SKai_`h
zD9C)2j^!-~#KW_FM5QH{T|d3_i;m50;(N%msp{ZC?fn`5L~f*az4#B?aJy87d6y#+
z0(^}2v~HGA%RL)kr~w(Q<C#B(9POL1+!_}xaS?b2#Fz>l`z&ZZR%2*BG+c!DN_z#h
z6-}l`WoxfG7a#Q}ilvRZ<wIv?-YEe{Mb2`RH`daSgi&jpIf_h4qZx;)4D6To?3*^p
zzZPi09fv+`*Zchljqlf1{AEzPBnuQ5`w;uXoD21U01l1_xR195@f;{Qr>6nQGB)!;
zmnTSHOB7P9zE(q%l81*ll8%-xzQ(`ccjv9K-8_jCaK1v3u`UL{2&0@oBg7CW`)VnH
z)DU3b)}N@$2~$ZLf-Lq*j`=4;ZxBU0B~SE7Mv?go+kA)0B%Rd*?{4POrC|2cbL!N?
zV3;4kd=?O65uui>y$4oHHsYw<1=P&Y>GFq_rbvg4J>~Ob)jrhS<J=LriA%7?TTUW^
zdr0FYw;rdBPwp=Ld12RgLjA1BmQJfJU1#|_(91t)hT{l-1G>X~rdgEwQc}crGC3iC
z8c^RFxGh*?yQIP1f;HB`CW_sljTf{aP0hIt5GpHy=PGYgb_PAfkl8(A5rz5y`<s6{
zWmeA0kVJ>W8m2E-J7tie<<2_=b5mrKayLn4$W%H{Jm&Z~tWid$zVM2!J5s5?wRSS{
zDSz9D+=<({-JLp0tbw^XO)<o2?>H?1M9B6%N4|X5R~=@Ia}F7$#se1@Uu-jXrQKs-
zAd<p_wT^zx)nEM9-Z=3kI{X68Jm$E3s^nv2vO7}^+2vT<pyK9!Bf=#?!^%_a_Wsq-
zo3sZJaXFK^m*0?D(MB<_i?c_j0}nh<cP3?;lZ|}m$$H<bD(m#;GDZt#|Fke!&}!Fa
znxD5~W~iz26I_U<zzu75=k#>z@6aH{9`Bx;-y2|3Srdl`0(;T;;17M{F1|uSUe-jt
zJ=hDGE4U+kzH-y=wWn2R=zQXWrpD*f2pjrL)z+qyq}hK8SF{}*_innLlsQ_m*cKCj
z6D5M)g+K6YQ&#$jBLWps0H)Yxb5s59Q>x9GOGMNIzw0myYA_h3+p=^ygd1^6Ki7C3
z(uKMh<i^d)n(zXBRCe!HoNTrAOrt`tw3Sg^KD41l#xnoH<8716O}5=HO{Rbj9vM_u
z<Cc;zO;3^qm0_p+_<-UX$!&DjbXNK2cs`oVrZZZ<s1EYcfddd|D3>Tf!<Nc6BBpY*
zG`r79lr>O6!Gz1nwp5F;0m|#J@8twCLn?O8Gjr4F>=sTY3rYfu<Mlecl1elsFXlPC
zy{ssCJ}>nU*PLl%P7O}rf>~MLR%4KReD(Trk4A;$r1W9P=*$oxw8DxZEf<EK6Xhg)
zD{G7F(<dsYM?$Y*Z`i8NFW*|aro<LCF)CRU@S|k_UG;i(4L$^uku|UPEIO+WGJLxs
zCz|g$P=~DPUs=s_W331oNe#ILSXCEL)un<!0U1-%TPnBNiHXv5Nj5U5p-TaIvM9&y
zKNp%21!X^?$6wEQDBaJd(UoK4z-H%&z%TABuKlJTw*?pb(i!p~nI+*HC8s5!({6wX
zUzHEFXU|5NBpn1B{Eab}!{_)uxOO#}Mso}f^!B2Cn^B*r`T5F`>;jRHJ@2H^I}aR&
zs9=1ycb&Nq^&+W}Zg7F8{i<OHElkK=Nx!sFJ~o?b{6g|s%tYHxXt`MqqqMX*UV~bH
za_(kpr)Zg>qG2!o2GPhE3xO%Nk(G2nj>yCm9(h}OdTw`gS}Ejn*Yx6St<fOagg1kP
zY@~YrM@Go@wzLc^j|yzJYeQBAotwu_S8|&E(bsqM67u^{t--7dQRgG%&S9<fOT#mB
znntV5;4Bo&-LZt{%ExGgrXKoR;@<UYpbq$@g4z8Lx?<Mgb(4Inxd4Am0e&S7_*m3U
z5w81KHOGilHP`5bFmYRw%Yh#CJ&8*KdZ>7VJ}HHzU4ir~zcP;+7q_x6KfvWoiezD9
zW0L0izi_n+aw|mC#lK0wfCIBf#+#e)!q?HuAGW?r$L59u+?-g;2rcaRgF{GH1h<+W
zby(ntm|W+mks@Q1uG8@eGuIv`O_smRb{-wBn4iS<4TH?)+>1bUV88B)k&8dKm<Cc+
z&vU;hxS_eVvl4cWtB!#aD$BHxmaBqlH#-h<eroDo9z_mZA;l`^1_u^`31^|HGdXcJ
zWkCNdabigr65q-v)QF^@aN%|${dWt3MOp-KYiThHCSVIn5vq)io$}&}BQB`M`a?vI
z7HPh6b>S<&7h&(1ryv;E%J_|qGE`jQI&4-8{Z}~ay1*|BaJG1(_E{15NQ_)$jw>pW
zEC)1ZQ_Ya(cXG8w#%rYPyz35H@za4Bn<W#1XZEvBHoFb;rRg9JZ0#x%xYlD)K(t6W
z3kvmVR!{fOc+KD{G<|?fl<IR={Yo&pY1NbscJ?CsIs5AQvyV$<0;(CcQ0Snc7L#4u
zW{0Zc@`UzntJmE%@$(h6^j>0uos(rhyu<y*5nI(<F4AyfCEZB}>G$w*=-PvqC3##D
zPt%oXv2n2{z6vzXn=J@Fe<r4tlusk>>POAo#{eOMhETTQg2ynw>%<vhSo=}y4S!dP
z4VsVCD+qwanDIY<zH-9$CZm6ONE(98|4POvNt_iTsa%F84ZJduPWNloGR{3In?E|`
z;hC|KD>%q&%hsDkz%{VDTRd&8(}3PrK6g%)SJ%F*c}NT=2FU4JR2KPlW@u=F;LFp(
zUSp^x)A2*%)tTMOLD*U(1}X4g6F(&<7yCgY&e@sE3i?*9KW4d|k<OwD!Sc}7C_$G>
zBvJHt0>F`Qwd|1ZCVrv7H;f38DS3EwpMVO--=Gy1Pe&&v8|FxBks<L+V6~0XIGFw@
zWtlO509KNMgp8U}Q;l6%zN09Fnu>3($7apcV_>Uc!!_H|t*!9%+m>Mg^{x#*zOh_S
zn<n3v4&8fRLmWn5V(`fChu`wGeMbZ5bX*Cg5I#R{wSUW$XUD~fN^m!NLWI(WPsM;Z
z<qykC&8~m~yS6?otfZ-)gqtb3F2-9bdL6XHv`hd2fRl-9C$AOS5rO$f*E-%Z38KJn
zCcuj<OTRcYOKtpe(BTY0WItSbp{IzEXmaaoUB>Ax)0LCs*^t*p-@PN$3=))okf=Q-
z4mR?mCO$cLlQidTPHSp<(3d9->DLR3aeM<Rd?rehhXVik*~cef=8J>#_27r%eV~!O
zJX8&JK!yv#sj1r9IyV=p%w0AfTsCvqw#D@SeoqBX|NPmhv6>-`lC|H_hNA}RL+=`5
zX{h|2Z}Huar=7ks;7@@5UdfO0l3}lCnbnyG1o3Q}kW@{^a5J8Q`TYQ|D|cp9uIWd9
z@j5e*0^%b%!J`u9A%y#lc}sA8x-kCx(D0h<&(~=J8o7}+va(7%(hxz>-#G<cOqptV
zRP<rEVPCUX&X`6cPENRWoY<a~`}&T^Fy2*FpWQ@8#t}L)S1lW4Fa6rV3&NnDd3&gm
zuKV<!i4kiuu>e#JKa8A&_r)5S$FKBsRFxcgHqt5b&Nv?~&8J{2o^KU6*1V3|`DJlI
z4pH3<ke16waBxP8g_)_LG0nn{WKfllQ8)HoOf;=r)_|PdDZZ9gnGI>78Y}ZpZh%&O
zPEprn;zSQ~!J3VoER@mD>wyjUg@ibWTD$)m_jh_xKk{Lux6e_W<ui4BqKja<p_O+d
z>$p{{GA_6P6r${xS%^ubAJTf$xfbUOaKFNv7C`WO2E?;+0Vlw=f)^CTM=W^nO|LHl
z4Kkt%f|;#&<&?!F^_P|=)!oJabRmrM6BM7Dtc0iVzWC1S>Co|+{WBq(7L9vWL&F$>
zya)S~*fS@03v;B^_MSh9Tif!K1t`jPHd9uS3#AJ#5BuV|$u=D>)wG$C`L4THd)>S`
z<vzBYJ{@=ockDBkufn?T+&o7QBUTYovs?d#(%pj=CcyUlSHwbAa?g+?<m$RJ1Pcdb
zY}^xH=9W_0pr!Nda;sqxc{+Y!tr%(%n&~n+6o^4F;nMQDs#lM%{b;{2uN1(#33DHM
zRF2S@K0Ou-u$3bKU7VyHO}5JB>kYps9gWL-#q#!v<8%0o%GOS09I-UyMQ5oPfutJK
zDo9t(WQ1lfk%g%9kwURYGi~lyPz^Rw%bn*JCoPX-lJi|x+r=0DOFe^wN75&>n5<uR
ze^*5bcBKS3lEz+H^e9U~pmDN?(ga$&(^imGEFGAb<m)VQPruc0GeI)@MG@ZZ`%PNN
zFA7n+H!I_ztA(Yo?&5{!&?NRJeN1S!`G>e*rVJu5d(+1&x28+e7~7kz+>zF{q_^ny
zjH%CGSn4G|*3`85Xpqkl?xvV5P&gE`#9^MF7o<sy8`djZRg*NMe#5~OJYR+0O=`9v
z@>`qf;lqwNmNO5C@)urKwj=gu^74>N<NQIp7r0?dMKA(PqR)Grt)2n`hdI^Ib~Nmt
z<|O!FXf_u%5Vvn@zUbhSyx<!8+(FyhyX3ZO4)CYCO)|l2t5X5iWhZLd?Bb4$!z+|F
z3*5|$B>(+<VSTT3|MU>edTW3g@RL?mLTSV|yUr$8yGZP&3|uY@q8l4*$%BGp?Rr3&
zG>{sb&YMd_nw)Hw@zQ;J)C{~}dFSPE;5VSIn0O`f*2k!bg@9lfGpw7eqxCLyyfD*s
zZ=(2Z%^TTS<xps71+uPDJ_IaoVnX<}273BAa7sWo5hVv1jv;hQNd!hTsl_fSeb$5E
ze#)*e!QA-SQKU(y_^35|1qNYto%vKH7r&C?Of|3R>62?`&6<?-_L;!)5c3Wnr}Da@
zx1r@z>yf<bsUekNk6<h+;CLAgSeXyO^-UCJEPRg4)_?{l{p`3yH7!p#TPsjtX4$Ho
zpFd3V^M|&4LR#hU95^xiDUHK~)qGI2LOX%RQg;05g2|bMj-3<(?D(+bVtR0DFl16!
z-ydk9*IHxfSgssTczQB0d>ooLQDj#o!ibHtSncBkcF0`)B=9-8d}g$ajU}m{c%6_4
zo?z`aU_MWy>{V)_>#?Z}eD;h@)Kh;_dqrD_jgtDgT&IOe?PfeCb}))?nlvl~1NSY{
z2w5^2`&#&=0Qgos^(x7{{EJJIqCf<T8^A(UT4r%&0D1^@c54n7{iNqd&H7k4&5pjg
zOPV9tX6z-{(^EWhAm$Z%+|p3rpJN?>b9~)OGN1W9>U0~G@MU7VpFzxqjfJp35pSob
zlvTUtVk6Zd>dL^<O;%CnRGx+sZG%400tH?U*hGXJ8(#F<UIDvEeWUTpqi&zS|4U%9
z?W2#&1|Bz@ajZrMgH`6cmHg89xY*<V77I;ZmjdQhG+W7ni5||+=f~1oTt4;BHvBU4
zWbcO8nxu*WMdt9>^I{oSQD|?^pW@xRY)Hu0YBd5MCf3jbWL>=Fvf0Q~VXQn>f@U<C
zf<gaNhj|05iiai5{@g~79zHx?t_HpJkvQm-1>krqg5bAi&QK@}Ioa@MiskXjMlBSK
z4UZ^oG2K|pYWPxV);+Q&{zF6G;K&fvZ#SO9WF#QqXxVK#IFT=ZJUaHBtRk}pij^ZN
z`(8oY7Y#TChz>8we$Yj$H(U1Qb(wr!Q2ieHZpw#@yZC6uFs>-D@y>+v?1m)@y!Pu<
zIeW#KVV-;<Lp#N3VR>ZLYv?0j{|5G|?;53hdmOTid?!Jl;&;9{*<vQ~5a6}z@kTr9
zb50}*0*@zkbGNRaQEVv7THzCYQVM<n`ZiauVw_pPgiQ_Ov-(B}2kGk8tICdX&M>eL
z)RLH+H{6lz`MK@sa>%yf#v}ts&TRR4xd|pA8}i4_;Bv>pZ6;lnr)nlF0K~4YzWm6q
zs>p8sLf(OrmDS2mBDpw=h<j+@V@zx`N?rk1bl1_<4se3CY|!xQ?b9u32(QhlH$%?m
zvny7D#`#ZzYH-`N^|s}@wWl{Xn-_?4YV|CviBfpXfx4>iLsV$eAk^7D){*FI=0noz
z;Bv3#g`GK@wY6W;5a>(rcDo2cQlWjBv~HopeMdHgCH8%EmLoOTU3f}`@Cn#{<2$hr
zQLWYO$jj0ag%cOMAEHLaJB>qUZtp<T(+xfnC**piR|}=R>2caj#}^Uaav$34w(~Cz
zysHoD4IW|(6N3Q2bh5FW4u?%>N_)#07~ITSF;i(5^nUra<g#@|U__~}ikI^SsiCoN
zBPK6CfFC!EV|D9sz=X=Mf=#d*>`f}Td}*p_I;~dVzE=~j?FaOJv$7vQ$cd6}*30U+
zcQ{75Po+7(2}_8$BQ)|ah5^*{dLC#$RvdjT_{o43CGEV|zB?_7tY7>!o9o?ARr-px
zjrJ{_CgL=&AmE@<yx#K@E#SzRi3vza`O|Nw@qCMd-lD?bEi<7bWIZb#ztWZwcU^n#
zFo=(?LKbWAW4z^|B0L|U`KYuhdtb-_w9<fb=Jg^jI4zyhiGrrpXwu2{UB6cUv#E#0
zzqf*kAB8Ka^N)e2vuPc0E2ou|zPi^dUD3LALO!g;VMxGU@KNU_IG{dlH2vpU0Q{-X
zj}dCBVn_-iq-~szZf}xvxXA?{eNmWBYAMJ45KG^9wT&eh_;JJZO%k$$vm$?84DK@8
z9A~Q6%w1;w(W#qc@{0m%>4<ou<R!F#8Z<OQjB;cdpkFHgJbgZ_s)`aQaqml?-QM=}
z!Q^~=J$EqrZN~<7fY#QsLqWKgBafDLsQYeM$k#&VUXmJ%2n;Ni*d;6JgyIf#EvAR+
z@(*%%Ap>6m378Kjv^UPRe0A2dvjV8e)LNBPeI&*W_`WqB7)G0wh%LNV#iLU+)lkT1
zCFrdF?B9~zKBEut3oKxFR>x1`K%h4awmZkzpY&`-FC;}~?{`g`hbendRl!q5_bQ*v
zQXGH(24|{uN#I1-85D{2gPasC<dJ)<2A9~ohT*nUBD$kN<;vM|C)W50tC<DrPq9+F
zqwTlwyh3JY*}yAmw4QG@ygE+LCW?>@pDE&a_wWZznqM0w_E8Nz!3Pr%;F8zf)9mL<
zinGeBn3-0*y#*?X{ME!tsQU@HVF<6a?8XpenWa+$4D4nwcKy=3#qWg$u%&MWYSJ&m
z^}PbA!-j74jjC|5=a<>Nnz6p^)-u1%=1MT6fq-rf;pOXJcNappMtxgdo3{_j^1;~Z
zAbZ#oCjqTl)&wpo+hS@~m4rz(M?Y#7FFt<1Pr!jIpmA=L-&5kNSK1KRvJmT^(s6{i
zHeY<Hj5{8ZtuZ!HyaV6lbXPW>hzZce#l&u9x2jD4DtfM^{FZ?TTwwf&j!BNzC2Dgc
zqs$1^^rwnoGS4H%q7@BJM<%V7YNK^optLCn5jZcBwgvwj8bgi^S5xz^;q?m-NX)r!
zI1>gRUlH~w)^E}2xEG6fSOTXApSg>@&|r(^V22I6yiP~&#0}Gow<(TFjE{R%D<B*f
zWU5s(k6l<z&Jp`!8%9At4uqq;Sq`WI4L*^@K@H*J)+}7x-d=7lPDznI2|C~JEYeg~
zsGO9BIMh-r+HvHz+;Vz1&YO=TTgFa6bx9i<jeb!#c{VQYBY{GUH)3;`TW*x~Lqrh7
zW?zut4<TpJW+uJYu(3&Q<Un=L&tiDqar%^Mr0?%HGecMiO#CFYYe_ZzR()}b=+p9g
z>O;~sVpa=NQaURWM<8Y&0<-BW&QgFKK5I@DC`~PN4GH@47_%fb+ps#L=1#jmrO;))
z4Ylw?2XsBmG=ZN$F17x7>*eG97Lqg%pHPy-e9Ga`v8)MABpFdgC2G3U>a8uY*MSRM
zp0m6<p4E4dXPQg~as{U|8vYXWCH*<MG9sQzsYv+4XL`aVPefn8E@;-aF?{5f$xwG?
zF*e3=66IJv9TCL^N~i$p;9AbJ1)&H0v=(x-?DpT8v+~lQYCLt<qNX{@k$$JT0SX2{
zcd?<z%vb~ZwOMGF*(XkR3Jvw+>*#(f71Zn{Vzz<;Lj1&<<<^h<<pF0q4~H{N`c9GI
z_yj3}&#uo4@8krq*szKu8@?(G{8aM*j;+&xfx`gSG}EH6JbOr!o`YjPGr!^c`Lcr)
zT~!5xTyEpZ*sODznikCY$x!>O*f?1^Ol???D%u$f0Vo=^rSr?nAPuq)XuMVtB3~zE
zr#>>8qmH<1+4VAl-jy%4T*F4?cXdeYUq#P|=SiDU(LfBVpyIgHwIuoZj*jmwR77Mj
zQeX24D!-_LwwtC&)h?OPhHft(y&Gq`IY}$JQ{^c%&<g=}PX-S<U-hTz3go>%@A{Hm
zZAZP|#9a|d<RoA^Py_;w*-SICWmbFz0bQ~u%MtK%S=B4)4?K<oCMKy919%E(?2rum
zBp^n43r@bz{diIZ4Y7yN7(Ul2lfYFZ?I|rdWJH#^U?NBf83s-`$9Gux*VbVv(5sHV
zwyZofF)$s28*IliB?kdLo{k<B<y3B5U>}3=>T+|$N>ys!6wAlxT}Mf#cP}AW-28Q_
zKu(tO{3atSLA?>P3>)OM07`+5t>!0zcN{1V%Mx-qsEpP4>@_&p{mg8sdXj232&<Da
zvUJ(!O(K<l<tiNn1gfr1>MNw!ouibEm(O1fh|GTavj|O<*rkgp)#K^O=FXT^f$&Vg
zvGn;t_u7wieR@SUXY;JYN8zaB<B^`CzP^VpyC0{~WSoNXFY5Reha9kF%EvbbM!uvH
z_8opbI2{c|hC{vablHl5o66APe)VW@{1c)*Yz;S;Hf7b0h$U0yRNdQpwe<xwYA`=f
z`A7$%b1R6z<<@b}J}X4c)TUbe1a&|&bF2FE4^F1a@PF7vPopYw4ebMj%oAS(!3hO^
zA=vZ#k4w<U-%e}pZb3OaFCtl3Sq}mJB`eoS(T*#~;ueST5%{AE4Xm-94+_=DLId8y
zT$<ljni@_22nql;T~>~|si%V;blPyU^K;43#(~2kq-fC&&-K-XsrhORBwF!_g)iP&
z%^1JW@1cL(MbM?BM!_12D>c(d#H_JBw+1D*mgK`zOIK1;3zX9(Dv0tQ;hXzK*R!oj
z74#{gHPE~qfd(&MDto(hk_a^9Ov<Ii7h!bNpqCs`PB+8UkGk23mjkG(I{`fA_!7ud
z*&RE3_5t-dRA6Nc>{22hiaUj=%EBPsBV4eGOI^iNYwK_In=AfAr@UW}1&EQ^S{uwo
z(MF%D(f--|2XYg`X)bS+6>qyiWqP@C!AfCX#4#okXncJ5EXZwx(e`yBf<87fawE71
zK^ybpc^P!786OBav)tAqU4Ox8Ih%Zj8E~DB4W{LM7Wi~-p=5SC{AV}0(bh=P|3lMR
zheg>%YoBh0?g2?@X&8`B>Fx$WB!?1ELh0_1?(Pz#86*UeF3F+0JHO}s&N=^c@yxYj
z?X}naTlY4Sa<#1Zug${K_)Q3!^|c^K<uIBD1}6i5GocGzV_~sddY|N2_RpxP2}o5`
zja4q;al^X6p=Qe-3n+p52aNaS;{S$+F&=$vf~{LyjYw$1Y^ht_nQi~_6N5$W|Gc`G
zQ;D*99b;?o_$l*PHNOtu&yN5=M|w3`=0hhbaz8fKes)ue`hhDGZ@@v)NC47RN(0uF
zhqvEfDTkVpgs<D@!w?3Qh7WP{d+580wQOK5>{xCJ)#YWCUf`Z9AV3SLsDwjz1^hqc
zFksn$4VH>ZGpjB2nxAz2`2I^vVSls|{HP1Y;H@`ikDJVsagkHw7SEaDZ^*>`D(W{i
zRQtHL%oLKF@G}UATH70pg+JayxeqQcIY{~uml{hsD|1}^Oj_}&gwO(_Ah5ICm$fge
z*8DZVP5XkOQ5PTpzm4$>U|v1vxzH~Cf>vQ;8MQ1FVT4T|DUJd3*;$&zeCSl49OTl-
zmJ+5yY2$AheOYi`=;Wab`SfE?RN#-pLqi3-pS?hN10G|q<#b+^I*|)93LSp3o)Mc0
z`Y*U-Nk12p$BCJo-0o3~$VYzjCfpJp6a@!+)6ZY5Kb9ZMQtru<W*5Vvq1;CJ<>1+O
z6*|6VJPg|ItPzE4@$$pJca^G#-Q{!odI`aV`Q@F1GzYbYbNF2nW|CKwa*(`M_VUPE
zodDGTE<_2%Nd5N9fb16z^}QnLy2m0jp?xZueM-Qa^xq_S+&wj&U69AE1AsAyJ>p>%
zj3v{BfDFflGFCbqI*D3<nJBQC_?u0A$48DXF5g@4_m=cc6~M<fo}X=v6y|)Z!3Tfd
zpl84$S@Iw{{;AN8ay?jtjLm8T!9>t7*xlPtdh>Z^#?EP`Sx2u{-Y3-0Po@<^!OV=f
zv-^(%q{tD^W?LPhnB3>^DDv-U2?Kc8#xzbtYggA8mF+>25G-%E{nuNx&R{+j>tgEo
zL?INP5RBH@-LwVzEDWy5^vf9&jIVfB=`{1OnaQu8$itA!V8b88KxbeqjD&u7U}vT_
z%Z=OmXD8yRAhBW~zCwX8&XoEdmvuUQ#TUHbK(1KW%l;vcKZmD)?iQ>q47TJX*>cdq
z)CCSh5lfDg#Kg6V-i~6zVfgJKpNS9C^W9k9YhCQvcy7V|HCiA5ua>XcEV)i4TGe04
z>(<kR@72HXFNDlWVhwQ-1KBCvfY;Ifn~dJpA+{Zr{hxx4l$1p);yAsUCOEZ1ZE7p_
zy}gf($Tuygu)c%-w<i|MB0p;qPhY)ya?R`N2=twJVSb1GPE2_#+Y362d5zbNNd^zE
zxFKs)>bJNeZ6h7gr1^c*_2=p`y&Yno$m#frPC(9YTvuo=Z*BbPD}SJj8$x<27Eu`T
z-l=~1?kAx+K=w3b+rM69@Oxt22gi1a<D^a(l(SZSj<yB?V`g2#8TVho#=6-ymdAqS
z!-1Y|aAClB20aULH*0Tx1Qdk}<gbj{lgMPyuJ@d^cAu2qmybc1`5XRKI=3RV_kTxi
zeSJ;1!gPKQwdg5z_=nT&sLSd7!8vy2zlvCHCGVT%)<fQ+V||a&srBE(?msd*6IizM
zQD$Q;)jxH<25Xo(ZKXz9T`t#{`W06JLBj^m#140z1qVRJG!^KQwevSOlL%(29Gd^z
z9x54(AG1zn7y<3ug+@(GR~G>+{Ucrrw|ZB`(y3;3GO+>Wk|QOCil=?@M|&mhpI=91
zB7EGHI)Exj9%_L{mLvxO60ObP&U|zr1M$1KzmHF>m?fxqEX%VbwCgI)Z_(Gqe^V{^
z-!OKsag)LSEB_7A)C+U$V}vb(^yR22sc=!kJoNc~==k4jf5K{C05aVqUV`f+Q0?4E
zB(x4>m8|`Rf5j}I`yNJU2CZ|01jdEPLyAiZ@+4~x1F@@5v6NE#EImYqq{Y~-)OFb+
z)Btve0zUr9W0nY<PC@^si(Fm2Tm+Luu`pZ(>U+l|eun~V;e&&!AfXnEx|u`!92>JR
z3871%RmY{AMM>%A<P}RMO#D=nxiH|GB32l!r$~{b*L5@Dc0G#ahdmq2Qrty{63Okf
zm#vqet!jiGT;Zbd0F_d$S+}+^V$04ht3W!uer!XMx06!SNyXAeWz%!X6oeN<wPl7C
zIq(2E-y{(*8})<I#2n&r5p^ckg$1ukcMpO}HI7JCGKdMmZs_ttX_`dv_(zYDVpvi4
zszeSATB&+40f@xVITP2u*RrwKST|^UyKa91!sWIo_{ox!50?M7TQwF__`pMihksLX
z^gRYhPblTo=bDofz-~g!fA9D0dhntseD4mQDP$qP((ql6BK<vf3@r8%LS@WSI;1&D
zTZVM}24nyyg)jiycD&+WNcM`M5PxtMt^@QIzE3l6WSukN_^tPp`?!;A&LpTbC$Jxu
zwO1thhl|`<h#m5s&UcyJcFjdF{J))oC7N@2NPJD<KQ1-jgb9C9!hP$dq()?nVc!;+
z{_9Qc^>|g#6AC>TUN}4HNN&RedhgV=l^a%}*R)`0kWQfVZ*)W^K%!6U?8dBR;oRXO
z=1Ig{ZDn7-fLgGUkr80$0Z5sJh1`GL@Z%gTR{doi8|;oFde^T?<&j9UWoNrE?|7P8
z$y}nRA0}HWI$z5U;cqiGa>{7Y$x<tdbmk)fUHCzC-s3FZ%gP5cRB}_lNz3`nzcAu<
zV)kj(+BR6mFES>YdqswnX11`cjq>TK=2CRKNtx5~+^og9|L2a?hJ2Q%9!vSvl)CFJ
zyn^}qrZ2G?vdhW}Cp^+cr9*fRf7hRFCS1^K0SPktoDmHj68~roG~p=02mlx%4!T>d
zjE0_Ywyu5Ki`Vp09Lqa9^KlZFxZ{e{|Is=Nip0xT|5-pusVSSwkeIaX?LADCS#lSR
z*|;r~TC#ZV3j01)DZERZn0WV1A}L3_+T0MYd~yG`-ss^=k6T~OCa~_DGVdxqK3rq=
ztS)}LqO_=IC#lv;T1f5*W3a0$FflnkwyFa54H-wdzZNz7?|K_}<E>v5!%6WVb%MSU
z4~Cr2svp_v)0^khu&lIaCPh{GoJ8JtlM~;Dl?*D`Dd#A+V(wqYw{26_c8@~1%Z;vR
zW*X2bDWwsmmhqEc!z^1ExYRGviv>_^%FFk$_%v~rC8O1H$vIM@^SjJ6tG%8A=j}QF
zP*|c7|MB>mWWu?9Yu||xwmvziQj!`Hc2&1{1ciiZyQ?<eW3ca@v|;0kha0=}|7PCE
z`SOU@bpcWnH@uQ9C7PWqsB0%nwFvIBO0anu%XKk>a(|f0Bm@6rO9AN1)^;za`6au!
zK~y$PK(pB0f$lEza?-Ta`B8zyA&-k;jXKZQLMC50pC8Okd|lV*n_XU&sPL>*6qof(
zQA-8)tp>|{vf*Wb&^PNyJFdnqG{&F+stCp<j^Id3WLNQ_<B;SbBR$|T)Hqru3nb%a
zP!zm|hUyAwC`#4(Kcf)v@{F#**YQENW$otFNQvK4#FZ7O*1cy=%KnCJe#;_#)Pf2u
z!w>tF2K7j&sGSP}jltO#Kjt-FBi~%*=OGwsvx^8)V8%9%4eZZEyO^otX)#yI9Gxrw
zFo6Pmft`iEW+(}miD~y(P`G1sM?36EyW`D6U0eP47i<Af`@1#q?WcYf^w!QV^^O#j
zl;kfqY_N$*BvxWZ1tIg-!Bcy|eLTq-AGDz4ascBY`Jwcs;>&`K_l4@q@5WEnf-nGO
zO+L5MyGjO$h#|}uHw?G=BKPLUdj77^iN1#paggYedm|aJb(N5DNw{r~nYsITi9!EW
z$&I9O<#ey-XA{Z01H$KfO{r><vX(dHR7=j>TBe#yHs4VMAQ)-b7dz}Be~-B-3K6uF
zu{*>PjM*9-QGp<mH&9>7Osmb${h^%2d3N!K<zKMrs>pcVly^-OW$~H?aRmb85vq&y
zlR|n9`{_QxBJneMFWn1hG@Zf4@f9Js;US1E)5c3*DwVzIs)DVl851Dyiu+$7VI0`_
zPQ|x8k}JqI9$l_GJi9c{MJ=}%vl_OK#8#X<VOhk)YrH;NaAe_-J-&?0_iGEZevd*;
zeLG@KUf;Z8_FlExLKk+MfE{bq*^a~hv8hg@478$<au;_<^b0VJu*?-NF5carl!x&0
zlJ5jw2U^#2zk&Go^0?c6Jo#G&)PirU{(`S7{^<k<85UScLv{4^m6H}#jM(H;I%_)u
zFUD?O9Gll;-$F+P=G)0^Y>e6!h|@iH?%%Cy;vU*s1R_)lkRxiVx~G)7a&x3~h{`7N
zi$Kco!4>@3u6E_oRbSTd%)!q+s?5e+{+A~Y-8V{o&&_i<FGmSi6L@_}9Gj`bWBw)O
zy8O{t`xNb0$eFq}t8RaUdY(^ZxY5Q>R-CeUuxP70H0tD>f#TWEHvK;hynE3?21g_A
z5CH2cpfHkz?uMo|H1crc7p_@s#?W}gcWlox=6CDV%A)3DD+2T26_K35;?bW<ZKxZ$
z^p!~;8oF7YQ%uOQfx!;B{<OpSef{=QyKtw>LJ%%_O4(|OOv;%^b0>JTHhhh%F^ZR{
zfdyjyP21iDn=EJlPl_(h3M`<<xx}f|TM`o+s{Vkpk8AYBtvA2&R9D|Qr5LF>Wk6gC
zNeXkhfV7gttoA=TCi*kzC>W(OCE;q1tc>pismCr>?3%)WF9QNF0sA_ihV*Kc<m#Cx
zut4q0vCq$lm1@BZOmZo<AK#bzo+7G=@?0JyGO@71P|4O>x@{!7SqPQoKGf$g;;4{z
zCu4(V=H#glunb~=nh#KrxcV6_quf6?CBW~p2dH4}D-S(B>yZqNRVmsIS=q~)o<8p+
zfpO0e0B&4c0bvhXykje@e>t@n++Tk){)F%(C3eykrL8_AO1Ud&DQRo-%|eI@ju8`?
zlC2{jSlH<P{+As~iD<uwT{+8oB)~siJ2I?(J~nY&rc7U(ua^<`7qM_<Nf*y{($18v
zJxB|tl_c)N0Nust?WG0R%r{VFcI0lHa!>LRQ%Ks_?Cx3^0C@@vI|&`eKwx1p>d&Ne
zZ8Yiva;Yia&V@_=-m|?@t$h5iqw9|y1DuD|bxRFPdn6QsroHvHxUu4t%-O%<vHSYo
z)!G!<qG0Q8PCza^Q?x9AW-`re+KO%30o^8}Uc}iH*|#45>}~L3M#j)>{|pJ&CHVv-
zC_pZCo>w{n-Mi^y^hFQ_T!XA5*e%BfXbM8(5D^AbZYa%ZS;lTWFRq^Ml|n8xuD>`0
znxbJ^d=8Ykp%id2tJ_xa-_!PcYTD}~+-3krnjvS5*?^7M%F(HXAEv(e-^@|f__ba%
zxyN1}-kluX_-P1ug+?^(xJr#{EwtbPKwl5yR)}&Ym2Ds^^1e8qo04)00qBX9@U2(H
zU^&va{%|@&UDVm<Ax@HX6CqMCI*5GPXA6Zh<=G#=OGOg0orvMz#i$lr6kbD;4|*%u
z^n!!xXJdnB^+GkNi;7Uh#4PWjy9$uU10B`hAGt@oZ=7c!7<-@p*6GL+z}65N8e<$@
zT_4MB{Slo%A{foWCi?Gk1omsXeAmv&Fe>zCmY16PU#N5H@HtYNv*KB4E`XKs)V9?w
zj@+x&!8Ww=BNwY71wh&CA~mwYpywU5sPqe($!ptTdwMW3a_W*JKn3hiUm}0$1Ip}Z
zD_U0b$*j@Qyg)kBNevvY>1-rnh34b+I{ayL%NT?AS2&iI&GrXM(so^Z^0`JVQ>xz{
z8%i`VudXIhKC%D}b_->KnaqzlJuG1|#X1A|#a88aza?H=L%(QGH95RI?buK}tUlii
z2n$JgFW_7peJ9M!D?VTN0NG~&k|U4heB$>$F=*bEwI}z*TZgrU5%v>V8L&$!pjWJH
zBKc1|KOvZ!dK(oWD>1V@zfm}!5rZ?L6YV``q!Eb7q=X|=D!&yYz3h;GE7lr-B92cL
zDw*Qw={hb~=;bCZmp14;3X{H;AM)~`?S731Qjqmi2Wqt4(K4hk*Ppa1>3?eOk9RTT
zA<~4ERaH<~{o8p*7S`TBRoi|vdNmCN?|R~h;kVmVa@T^!&^jFHOwQ=y0u!YQ)Uac$
z<_^*2Sxca4fdMR-sbP?3`g}`-tF2vK5auIYj)LAu5|+%Bm^&6f;w)rMQWFdH@ZhdO
zO-haIvn2E6AW`*lw{5^i@V(84sUwQK^$e50JunuSaM;GTkhs4y?mY=v&k*)x(}tF1
z#IRYJHgH)^2O=w3ih?*vXp8TQDo+B|e?QbhN_5_(-$Vs;`<{fL?=ddN$^ZeF5n?-P
zMOKIhYREWgz;FKoE*IHLMriU;x6J+|{O^w^DJv7Ip>rca?0EB}M{u8SFI%71!VlGR
z@#qZd(f|6%h^bqm0epHAe+Hoq-SaP47a7Nz_`#y8v7*v-P4YWx(~?%gGJ{MD%BZK)
zwWH$#5{kRdd(WB4Q5l=^+d^(cQw!+U{;uR%kqQp396v?Q!kkIL=Q61rQ$~y)ScsNk
z{JrxVB5*!<;qp>vV$?B)E%G_VUd`M+Di0rwCbvFS3iRx}_)rH^SJ`gqR&b_32|zzV
zd~KG=X&*$$17)bj^c5f;f5XNRIoO8X`F=I|DmA*Ea9#8%O%s4@mC~F8oiR$MBA)%*
z?F3=@QRmkFhdhAPu(&f8kTiRkBh@%jyK;O@y3ZKR=k@7)(;`38P+9J^5fBfbz#x*1
zaV<pdriYYF|0c(?!}QYScF!rI?#`{5(4o*=W`3#v>}CFBhNbgu6(b`&t}+m@%S$k1
z{t(#OyJrO`zlv$GD8~6!M%%me)T}HOmrZMo%S`()z-`7n_e{%N?SGc>S}8y@{r|F_
zEQCLzsZGdVb&%6~+HVvW1gL=^FvirVVevH5GPQKd-;Amo8NB{EbL5_uRjU=z0Xa=z
zl)GsWdo8&unOXe*2;^QAlED$6;Z6dvVuH7oETM_E({~~b4F-3mSsC=aCmM82OG`^>
z+6;5$`SyA-U&Yoje4mBI@88iu=Iic%{#<^io@U54OdGkcS2k%m5XexOO9z1gX>b-E
z-EYrnO>|*$%${?Hv@@0Ia1@1Ihcclv76B3l2EgJb&3tzGKl2NDh(Tab!bkW;FCYn^
zz49ZzSt>L}I_W0;$U!33OBZ>2KAJOV7AFfG>O!>TBry!@JtY8TvyuEa+TS*oE7jZ2
zFaEt+xqV_41DW^#ImeyjdYK70Drx*f2=+;d?1!~Ul93g*8=r_&K3&HhuB_@?Of$=5
z^H&f(eNe_G!URt`V{%_%vbv$YCjjY%|LeB7^|fRG_j)q+X&JVFlv8OPy*4i|=`xj1
zR!RnB>8o>v-RE($JT*)<ah-o~Pk9<LK#;6Qi34RQ;22n-?ue?7+yGkrm}_a#Z)#Y+
zXfH-?qv83$J+MX&{&ypu%9-SUt&RplNu6=$waI^fN1&$I=;6uT^$8t>@;79A`wxY{
zIr;k`l_>)qQtVW&!N1wB+kSE{J|GvcH0mVR6U=144Ft!YpF4gCn`N8%8zTsO*7dL+
zeL;LeT$DSnEOC5AI|z<x%XJ=4Xz$S1VAJ~i*}F6H=#V#q$>-&%^+<4en)SEV-o-9s
zOhf1z4}VgFwCed+khsB$ukyE-zR6AFJ<M2SbnfnkhYc5Hox6&PP4i|EEDmNsz-jn=
z%N;0lt*gTnQ$Ze<8u<{lvzp^RH5KqY(I6~GYM=yqnPGUz?K}a+S8~h+*xG)uT}(b|
zZ*2dadKt`^j3RyVPCJq|@V*ftl57hLgN;n{=YA)UeqjyCl}}a{5Ed~fe-DnA9=3J(
zF}JX=kiN{ib2BLHK78X>sxtD|cI;v<neW5jEa9DpoD;v58=nBTTPiMJ@3Q3n^Tf#A
z=aE(K$QgMb;n1*|0YedKDi^0$7lL-a4C!{&gz$JYLtz|3P+xri4@@E#&Rs$fPft(0
z&cA=Rxp^Bkt<Zd~R&|CgWyDmWh>l+}umU+nk_tw;19CdB>^vKttsOpSrA*1_5UATO
z|BZ=ZRV%0q`W0!c-qgeXe=Gn)sjM1gattoXp+~TXnKs7?Dy&by<KTOEe;!X1`@Pho
zDurM@e#lE-e_{U~xE8c9!g|l~pN93JGrDO0DeHn7l8HLWNnbxrmgya&?Az<t#l=F$
z7cf0QVSk&we-T8z{xP1A7=gK0U@Z^%^+lT<Mi8?eW&TS&0IQS;Poio*P6Un-rfCq)
z<xdds%=e|%H)YsPMJz0c9*n~t>$RGqP=Cv2S6oYTR?PComZM_&UpXLqQ$CEtJhy%g
zNd;%UgC(@qMk}#3yv#|~N+SL{;{1?bWaSER+OS64W7~c`@aS4dBy}8=rn4y}<+I2v
z`OZ(P60PWiYE8A6#%$hM%=_}-$Q!!G@jnoFoe{2S1J-`tdWizPL_paWr8Ma24GnAg
zkkuRjjxZ^2s~+nG7-LpYF;4$=NEE9rmXd5?kPC$FCO7>~V9`my80<mF_SM%54R{2q
zMhLWI`CD70?WSjaOfdK-nItHN4>m|4-Qiy=Vj>%SzM;5r{hezYA&BAZ75jkU7I`@>
zcuSY4I=VwM!uRHK8oamIZT?w#xWR@1l#93eRP)We-}a`?;C*5wF9Clx801x!Tf<hW
zodMv3iG<Zu{9Ii2To~7hf7JfErAK+6N!SpA;8f<`hyZxh>~Pk30`Gc=JkpND2vhPs
z>w)&_OS}@fF`4ne>Pj1i{u<?&)nEX>|93MG;|8n$B9zKOkb%zK*JK~`(<pc&NnOMK
z;vuz_=wlud2HHFMI6}DC-Q8TGjuzl{w7D_Hk*sq#Xy&848G9iSu=r~Am~!!Eo2dK#
zUQS}~WhpIIx5bkC>-T_FD)q<b0RP)gs;-M{s_vH!bMf2rZL=V2kCI+xwsRk^33?k!
z1w)F^(CNc(dxlg0mI32Oo+n%B;`3B7wAV0X`A-M0Ch=shE*Yn4gB9fDq%6LQNix!c
zh}{(Xw~z##U>O3Bf+E3yZgrQW$Y?ySF-lc@Z)GQ$0h%G}MIw|dt8f?XB+ti0XaCe>
ziwksq&Y59<jV4J>E9^Aw%?(Z$=PB7YuKpr=`VjK~EhCI2Gn-%Z6;y=KzC(?SkfG>L
zFoqaau*pz6!sw#2ChSaQayr(0-pjJXdQl3EzFXbExz8_H73IWWQaeyPo$_RP@<Bx|
zQ*yxikc;3gCcEkQr@367r^~2cIi5@;Je4}^u(P!w>AJEyd@{4|FZ;@qmP7PlD#5iO
zh7ep~(8J*HUlF%=4A(n14+Igit)kbw*%exdQ;{+YlHZxZ8ak2j-D&CZrSgzZC-Rnb
zl$jjARy^pz5w&TnH^K6<N^I%0;B#{xY$g(j{_11gRm&^k^S|Q8J`dRHCsohe*at2W
z1K%&iu$Sf+B}6~|KARIiRa*=$9zSv$2=#it3g9yBJdX@Bwv{Y}Gm+e3`J7qVaMW~w
z`}*uLqc5v|tE#?@^xP$<_+xzr{R9fL48&)>RPi`4cS=l4NyveUR#fm0llg?-kg%3_
zyxAhLZ`q9HF!)C<2ppeqTs!>OyNihxXS=a;sU9e6-9A`-XO^jfxhRi43cBRWx8f&M
zAF3k{QBwYS?A-t7HRCg7ybg+YCpi`0d^IsWe7=?7Lr%m97a2}}pJm6H{LMLE|C#Er
zL`}7vSBsTpX_51}@2IG`%_f31FX@obV6PRO%0V*|$!qcv+sNf88}<Pv60dI_=;bAn
zh-GD`vgFN-q4fVv(>3$3!w;~l&9lr6!78WFq=gb&0ZIA3O`R8)$(y}5?T5UgcSW*V
zOb3!n0TMJmC<z$PPcDx0Ujy43L|^JcpJFm<GZ6HG3!8L&6h}xH5CW@d^PD+25^T2;
z%2WD}m!krGV@tTum`H>%V5P6i?G)Wa=L3227{QGiT8I{`W}kaiy(I2#gjRhXn#rC|
zn9W5hIB=zyx?WzKeLn9Yzx~mDo%Y~=b+H)mn3Vgx-hwB7bIg3Q%>CS$+g-u4>*?jy
zEuY`m*!bf|>P9PIs*8iB?aeJFW@>1-Fx8ims&-bZqaXHfk<Lk5i*WCNiUQtY7W>?~
zOp~6K+({UrGRSdvCt|+;X;5CiNWXP`{r9uKRp#Hn-^UdE!C*4<=*OCO`3?nNRBjkO
zyxeDIOt{H6Lql5|c<+o+Q8Cacax*xt^`OZAzI+lX9US}#d><56Sy8cmc02U0sEj+d
zWfer@59wBgfTQK}f8-`0nd>$!4sqe0%zs10-faOag|Dwa4wRB|tX&@+9Ov-6B}B*c
z9>t;uN!7GxUfeYa$@}|@l5?;{M#nb%*xQRpPGL1GuZ(JLmMa9)ni!E&V2Wh_$|)@g
zBvoD~NsFSQB3eE7ra$v8_uY7YHZY_ZynCW+&)$H<F^hBK=i_;LBwu*9T5x033W@V-
zX%ijo^LmWnfZRfsXq`KFo_7oLfmWt|GWVDJxy#ysQBya?`o+oBSKC6zd+Bf3H`dR*
zi)vyQySrPzQ-O((YoQ;Q^)=~^p&rR2V|ZsS3%*Py5*w5I$J?Sgv_jlHDQU(Wy^@k`
zj=+e)VEBWsXOWTbyjl`orndrDYach-+BXCQF7MK_vM%p@Cv&N=A{Gs*lX~+!$Tv2!
zf~WwylI3Zzbt?v1P}Yq7mer-Pk57Vy#@X=vyob~`$7FJ9T;c4H5Db(er~7+pnm7;g
zYu>rF^NpW7xoIcNlre}VqBMB1F`V<a<AGi@j9tqoJLear{V`iEA$_iyoT27VZ2m{h
zlcmEWJR(NKRCs7XOS>%kI>`{sv)S4C$>OC^p~3E_qj{$m%if7%__1^B27<FypsS#-
z_$!2U<zn=Zkc0$+^=cT5^=^$~wN0SG_or{>f6jiW!m*h`aSk6XqY?ZicH9)nx(~U&
zXMDE+%kzV`98=pwY8tiAa$}=0gf!PT;F6NpXZ!nhAD9PS<74La9PF%+8m2}NVVrzo
zkfEQ;kH%`uWLV`U5|68ZS6)m^|Hhd?w5X@r2c1Wf=H_O3XZs^`qt1azG&!b%x4DnM
zs~P>zH~;FixAcdh%RA^-GYUspufW4daPUH=W!g<mO*2_>aBz~-(%ivhG4e9`8d!L*
znVFHlNqTv86fyqk*3u-{-;Mtfk&^uD{y{+j?#3g243!O%S{N`jMby`|q-*MKDq@^m
z<?4j#>Z>=m5G_VW|LQdBXNM;c)a3Q7okWuUl=kyqj`{LM(~tr;M0%-5vZG4nSx=Kd
zNO0hGI==b6zf|{iPWInDNrcBe>Q}Gg<O}@+s;Zo{eXsV%FPmC<@;(>H%O`wj3BIK=
zdq_K885Izka?#b3fBUxZvb4@Z%J*a?NzHA=4V#`I`+|=b_H(&jgC!Ul-D+k=<U?1R
z$n1il4!@Yl@U7|Dw{PKdmA5*Yn!fMf-#9zNrPW#q*UvN{Rc9oC{hYP+I(vUV92p&U
z?ldsCAtwjkxe22N#cnJe&yQ<Fuc({9(l`7aB2D-AC^jxuPfzU?B+q@~9B*I{et+Te
z)1fBfb7wCJ`#M+WFR@Dv0CRoV*Eg-X9oGn6ZLQ@p(i@N~oG1rSksqX_1M>30E%fN1
zyb7fy_z;G^2JPdfXzrp2zhuBi$e;cVx|&8lGh<~FK^YknP0jBctp<jM9W8T1G3F+J
zfJrBIQB{*a`gF%`ZJjkjk_&8^2O^<_J_2mCekaEt8nm<$oLN(ZYhch`$#&i+axi{I
z7URl#+M9Fz*RQSa?i$o%qE$qI-C{yZOa0i>lf203q-{XH`e{Ijv-{D;!%b`HHBB6_
zB5=`;^i4EHSG1SUogWhvHdn<(C^s;}nq1ebC25UFPPUDV3KU*d-;$7-dBb}6@-;9H
z<_6WpQUVT&)7EqEb$}nfy}irXqVjkXFj$$IB76TJ9LvrQ?Df*!M)=Lc26)ibbvq)G
zgmhqFAiM<&3wHhx^>EYOH<HS##x)qy2g4Yj>2JLq9Msey#txCr8YH5kAzd9F$;3kY
z-EQHQSlNUtFGiCY6D3dbvNunD+e>$br>3ehByOfjaB{pIr*PDO8M6KrYGl>dmo?be
z-A@FnmZ}8;7Ilnl!DYESgr10Ka!&waZYO$5shX*ds$-PJ#MNEaYLG`(|NEP|S-^hr
z%7*`q4f5}hLoNf)Gt@q^fMbbuA^+#1K1w#A@9OZxY2kc7=;h-5hS%zO#mV(|>saB?
zecq+HxfZ8I!OEF^C}enK1d;Kf=in+fIvNM3HI3uUpc)DRE2pJi-zYB{R8}7KuFV7f
zQ<G`|9ZkgWI>(LUV~rVD0e#eU+U%UQH~B~!c5Plw4I^>jx&G$XHwl`vlm<zJm>9Hx
zw9YDie(sufSMyifqm5rgw{CB(b##BPy8XnDfY>|yR!3dN-alGwx-rq-`2KxuOnq-{
zEpKomLSFNFjJZnR#P)Rc<Hv8ERhARKP}ta;g+2k65lvIOE!$%)u`vq~?gnONv2sz;
zFH1nP;g9Q2hYK49KeZ0HAN_T7=0`?GCT3H;mRv>9&5a#ifvYNwMFa=t2Sh|NgaX9R
zxcZ!5b{P&d74kbevTj-?@37(qNpT`;`33)(e{q7rS>peOBI~i^<D~w{#j9*?n54ww
zQG?Z(mX%BS_y{X0oxgrf`iA51SkPm{fP>}piJ0%p!#cb7t)=*>xlyF|Qh&hQW>H}S
zAt<b`-i^uZOZ9svvw#4emsD#jB;~ZoS2qt6(>kzj-GI$N(+w_OwQQ2Dtu5~xQ+|FR
z+>wzn5E88HpA0tGFzK;ZiN8WZBqzsOat`hr{<0vwQfJ&!&_oRUy_lGoS#3*42N)Ig
z+S_EQq^3s2e?wg*ud8Bw-D6li`JZL<;>ZXm>AJ?qF!-EW1PqFmajn0^Li*NFQc^%a
za@pi@s#n>!k?_N{oxp3XFRg2{btmUM(}gq?>1~xB#JQWA%#;|i?GO+VcKrF4-^4Yf
zZ);tG1joq1AtXRy9#Y@uy>!+byT}>P8u3VCehV{fG1D=7)&*<;e`GqZul{Qs{Jfp+
z^EfyrHiw8nol;ZJV^Dpu!$8haT-%ou3#y$;7aDC-M|AjDE=EYFrXv%yvFv423RJe{
zs#_u#Pchjajd5{j69@)!kOA|z?K84&(@#P((~#>!I-%o3*+Pe3hQ`>H&2{+w{afpx
z!GXH>iaD*WPSw^M@(Y8VX0nBKMqB!@4xi$rq)EX93^w)<A9QVP7S65Jb%-o|#avt;
zTvC?=>6^hH_*wjvtdFoX9XtgY9fYCpa?V$HYk{|?lA@?V09srhTY|bHN!*@sz4E!J
zUOZU!x2bG1%?6#%J@1|b+*$Q5dYQUEw{4B$#zg7mp5L8FOt!6ezntxo9f17Lu+595
zA(CuvyTik`01lonp3*n5BEJ7|+?}SBu$pLe)N_30%~;H2X!9>33cYgKWU0xo<y~R+
zSK#GT5%a8L?Sijmosqh1J`Bir?LUTM@|$=>A|pOqzcyt{d0WBU{L{x*c?6Rg;EHBo
z!UTymZ2{f>Z8c+IIW+P@pJ;f^1u*;j-%5PPZ<EwOJHlZZv0`WA(AJK86oOBxZbCJF
zP$LzIV#T3bTYGv|&GK=y(lZQ=eVBb7(FQRgH<r9^`oZP~N%j-kG{pFkp)^LEPQzGC
z>*0Ck1l`KME4m~zl!0dFN;0C=nFQZ4y2_<s>H0CgVnRYBUM$_jUE&{BM9aEry1>lt
zlX@L9U{$ve4hGSO>Axdr-Dm5#uUKF4IpAz+hC7(Ev%&0I0bkf~eRG@RU`SMIn%SR5
zSIC9RxsZ3K-yN>^mG0p}5=)-dGCii(#peMoGz4=^g11z;=<H!ZD#}{h6C+7WI{Pp0
zb=j2ILAHj6Yqa~=!TcZ6{a7S*%egPC!&O|Zx-=@aCeHAJp^Ln#kj;&2jw-Eui%PR`
z9fYC@Wv1HuTL=HEKX2c@4MWO@LLjs4E`S%*ZEq<VWg}>e8%Qf@@#7z{5b;A8K<Fi<
zhI+vWVk%$=w~;}mLtxE;^PAQuB0C2@HHc$z&TgImWEpHmA)WX)?=s?7<TCcmX`V>J
z#*vK(wiNY~=a8!}LKJN6s0h(9bW3*8(6VX%3frEaCP~Od%vly?URBBR&c71+c4XA!
zDt1sBQ?5FEf->LYy~X)zR!d5wgr=!CaCNQp(|U5^wJ+Ut0~wR_+7;|X=pCe_)K$fz
zbN5(IGG~-|YwFIoW24<;S_ziFtG97QsxSoBk0@or2>W+vgTKi!(B|(=`Kz2NfZ+Rg
zNlTi_Jq&CWn3R6`_|cvelOSQR%=ZgN!ACe=>$y@Fq{yuTd-X;9xE;?*9u9D?Df-{v
zYBDl)-`k(PJZ;|*omf3bl4ZDri{0Wow<-FDOsEP^)XE-lVS^zEaSEO~7Bev{80Z04
zL*iT#Xw7c?<x0gx!L(qH$=x+t`$HtSouiKT`95O|5)>l+{P8Z=uAb#s0d>yi?w}HO
z5VAVvQJA_@yzQTY#Z)fqC4Se){3+a;QfTI-myb(u<<z`9t$dRdfVyMde{+5AiUr=l
z*(2%tNr1}>o0#HD45-nMgC?iR95a2*pzNrw!+3B0<>;tGp_Bl^F1F;KP0C44{auLb
ztf*4ik>jrnCv1Xh(q&p&r@VFJjA`SQ(cku|wca3et2ZWr)LeTm!?qucFi@&1QwO7>
zs@26m9wPtIBjEQH?|7(rASZ!IEu<o^?e8nE?0JZ!A)-A_OA-R+2{q=nqpT^|b`OY%
zGBOcL^0jFIUXzh}mMh}Co(nm6z1{AXX0|6$Z}hvtAVv~&z^Gh$)PooZjTijrJAruQ
znjb6$>rF#dRZ;Q2HP%dm{@SRMzv|8ywD4CV03l2lkz=sDNnc<8WnO1T;Q#W)#7(hd
z-tQm>f2AYMTobdAU{#aFRd*z_`qZ*w(t@n(f_ihOW$6wn$L7X^yv9uY)0p*nKq7o2
zX;+`ug+|cH>X31I(fIJXY^95%tEv^5FeY7=obHIQ_x$tL1fgxp@ER>;HOy1qn}44u
zHX~y%b_E8z)A$z1LtaZ1OboXm1f6jsefv5HiAcvBA21he=QEf7=GBYZ?S1vzffA2o
zhs#VWy|#lf`U_X$L1Io$ZwDm<0Vt$MS`IR=+xlA;l11XY+=cP-EEFrkcE(umOPoI&
z``+%|OBXFu3OdcL@o37WtsvF+(KZAWq&oh}>4}m2g^qAx1r>GcqP*FEi_3YF!^z*C
z{cX26F>~?6#O&G7_+a%Cye^&}*R_;$pv1%m!fHmkl`Tz-bgK6>c#r-;s{Em$+!7t=
zQDl6pB(+wFh@>Pxouq<LKkH>ORjE*?`Rka!szr_n4w@$aEhy(0FMd#k<lzsRP1Sj)
zqDnyr28Z=Neiq4-Mc*VQi2Wrjz*v3?*sD+p-M{j^Rw(TVNnyX6W0aU;i0zrJevGm*
z<Uoi+hdX_N?iB-g)#+}bRAwYPOEK+`VHAp(V)5i&G;x_fDiIdp`PraBv(ob$S3DjU
zc~m$zKAdGESUG-IT8aaEA-J$pVPlSB_pusj6l0JhQKt+e%;~qB&?+yHf?Isg+>7XA
zU0{FG6CXBKRxf@}D-;VmMP~5yw654B4Vp}}B9I({kO{3z9kORBv%Qkju)46#^82)o
zDL|dL;fB=LH^lB{&&9MgU{3Vy>%+na^IG47hf66^TH-{Lh)oO>jYEO8aV8&Io}Q>y
zl+44!U++DMjEtSGuIxR+!ieX%6MK3C%mcdF;Cb(bWBX2CL{7p68K*CEXuz()fAv|I
z!Lb`w^{p&WUQG1dSb3C(V$SkdLN^tap1tJT;sBTKKI3B2Ljt@Qe(*=f%jh5INF#`T
zeX|;xce`Er87Y&+EQ)LqiY@twcw#6BM@PdFY0TiyK^(t}ybgQrc8C_k6NhhaTaAK)
z$o>^4<sWW1w=}vW?0t+TrApKxE-{NxY^c+4^(3It^PrTXHDk<8D%$)CU`*__r*l+v
z<vLF%SKg;m;%>Azr?ip=N_8#?zkZfbE@ON@Jlet%`}Zc$X#NLo?n}pGNDhr$t+@DW
zDT5isxjX>HkHJFbs^?~eh2yjn<Z^w=)3+8GKMpQHjYIdBXmroag5EwogX5qq1A$@A
z_kR^s+zB4`xn2%msweqB=Wwnuk)%!fIIoC0F3^85)7z$@LL*0}U@eE5Kqsw&p)B;^
z<oWNi*mxwwccMQqv*iwEtAQm%sLnFevt`9>3c{sx$-!G*rsV7FNK<-0^&0Fd`<^){
z`_%Hi01WQN)L_bQ2Y4>Mm5{^-mXDYvlJ}(I<JZDZ6~$;2Xw+L5=lC%dgdnARpEX3C
zYKL>s4~#JiN(gp?2XE}gR|5{)4Rmo^Beq&?qUKA$&Yx#{(k*y34j4Aeu(N#$1cTK@
z?zT&VWn{)6o0yot$FskGZ^2xqqt}C#?UBBG*$TF{!?@UHbrYkcMA`Wf3$h`?vtBWO
zpE#Pz3CK!TRIZ&h@jA(%)5lJ(nA_p0!OWP+U~79UDSZJyhl#|<^;i`SsmMs;h<GC)
z@KHMKYOD2{dcnJ&v4_`w?IQ<JR~zbOjI+Wle<b7Be|C0~bLab_TsFvCiiW1KrpgA(
z|JvY0Rdx^+r^up9Mv)vGlzt@u318=4g(M{FTt4|Y*so_c(}UZd2z4wuU_2v?DUo-$
zj(=)dl$tpN=V*VFrUhXjsI?FKf<lF9!z}|hEb8mrRup$d=6LDBip9i;a^i6eMRE{`
z+P2T1M7$ZslnBiEOmZH}?GH|Z#tRQ8>bd?PcY?x+Zq@IHWL%-2G*e%bSddmUHPZ#8
z(*h4UmH|c@PuRaN|7gE;K@oK9xft8HHy?J)H8vK2fQj@fXLj7fH$9>Fy1JX(1N>Ae
zZi$3p$~X9r*DbC<&+uaL^?JxFC!AO1GY8opI3rA=P%<)#HvMe|ShD?zSZgs`SfBam
z$Qd5;`jIkm6EK*{b#y5bbcP1rXlq{<4Oim)ON<=i2uZ0)G%s7;n3}5F;VUuEHVefx
zb7c9bgZ8ubCsIhr(AW-Z^3sXhXxNRT*w$xP_81KGGgGvnuP91Yvvt~huwJE&cq>dD
zD^AO|Rg)Bah)i&c*-#u1eSma0bSS9nv!NDj&iH-^1AXz@Yz01w0ZL&bL0uNvk*KkD
z;uG-gG5;=~qNGM=DP9&1{gYd(aU&zQCBDObu@Q+eIF3I&9E}e`yVc}&CC7m!Y~p6?
zZ?i_*5ybA~^qmbd=K$|CHN95rtU#Oa7NUt8oU6SFIQwji48rwthrb6bo0r>d(3SE#
zshM-Du1iE?G-Zvc^lCrC8l4a*shW9Rivuhd1${TDDh6r87M%dI1gq)VnEvqQKb<%0
z$L*Fa%<|#uRx>ns`)iPz|B6f}Yr^hcz%do+XlkT|RK4ODp_$6OKo5_QpO_7uQj>A#
zbN-1+W`=>VGRyhx<Vk14+dFh_rZ`yun3pU&CU~9;X-%7Ln~W4BxK33eOjM`YgZ0yi
zts>i7FMBWzLUgpw)74#MtZt7cLY)IyDraE}FMavx9j1Ulgt(NFqFf9VqbZn6BE^*O
zvNI0@@){$|8H6qcVcBq@0Nh<(9!sSY=exx8v{pVONTW~^fao(OS7P1AY@7!9&{8_a
zp#cG{A432}=tXfD9<(#AYw~J&CEP${W;QaNje{jgKK=B;a)<m%I`q8Q-oDCD40Two
zO;nTNAl=}bUbT*=H9_oBnvOwQ|NDbQ^{2h(2g2LY7_uzrlmp+i1Lp=1%GTw0NtO)F
zY^)U^#K2ikbL-1G>~<CT9We;XxcI#*uZKe4GZGKYbtn6F<%V#qz3u%K8<GdmtOb*G
z_J(XjZy7?+z8pnr;Dd1E-D8o&UopTYR$~7Nz+ehr6tiS^1n=~D-?qLpRYiSAx2s!O
z=S0PmZ!aZ?;R@clQj6dX?4e?X<>wU9Sx(URr?1o@4$@#FXxTe)Lr6y0FFxw}pLgdZ
z13p<r_twB<%&q=t%kUOrhvkb(RO(qyy~e?1{Xox<I4sna@QSh%dVqI!*obi2iw?QB
z`^=tz!}7Vj4g>v{ELI-@s5EI}JZu7ZjryNn1QZ|JS_J!>c;DofJ3Peu{$u|aujQF9
zd!BBczrBNHnnf82<SYj83K$rYKE;v3Ev{UM0zbX=vg*gWQb&^X6!3Y=Mo)x_+NL@g
zZcRGlk^)lOoLdY<58wN0`lJASO1l{m7ol|-5G^`ovMe=%gQ&dMN0|Nhynq|(`p#<i
z&WW*L_tm_qx?$Jn(TI<pQKNL+Oe9GDA8b-W#JE`L!JVC_cR2Epp@(<rrNe9OzX?HK
zkuzyd{)O=U?Qv@Pjdt)?tw0u)Em(#;9G6ONLHv2%XUX*t=hG-g82-gviq7V!JS#T?
z@Nde>N!FYG@g>J^l~=0~0gC0YO0v2*!E)l)KD@zH&f!El)uEEtl{0M?-;^IQ1HV?2
z!SmeTA#xapnzk2}0z9evUjiEJr=-<@fG~&b>x$Soi=@Oz(aPd#RW+x*s}^hj7$-?7
z2ud6{5(?NeByib%g?3y5RDzSx>TMs%7<?>X0N9%90Wd}8399qPh95=r%wT7)L;|2j
zgu$cUo|Z2fCZon0K95LU^C8Pp!7B6qvx*v4H)78><lUpKSWzYMHxy%V4F?xiBUkRh
zq;&(1@1?BecHRL;^HPk^aUyO%ITja?A3hJ-gvuWv`(kTQQ^RMDU#LqH_Y-Mo9}P7)
z#4ppz$542GagR6mG&T|^;qejx*84K+bR+g$+ikk@xSNy*_n#F8t=vZWo94_3J63W;
zQP6_L#7^%@VisQFL)eCuOg;to-8PjnN6H%h9}5tX`-F3|8mI23bWNyBgq?yYZBQm8
z09euc#-TbtTx+;Db{wmLsFx|<yHCIGjfP@;Q%b`oB0Ai>tLO0#CqA?hawaByK<iFq
zrDSK1HVdFG&B^eFk+C?h;PgUu@ERmSHy~*aT924W)H<={#l?(p+8FLEwctKkj=iUw
z;#Yp>zX8sA?Wjcpnp5T!GQ6{Ub(7@eIx()nHv>XgY#Cw3H2V3*WX&Cil{~YK@#rv_
zFNK7C&0~GPGG(>EnDk+{5CBKose28?b3S9We}dn1f7l*9Juq_G;7WjE@Q~X~mKpV_
zTlyL_g+sHaev%Q+277$5-%g0i+hIi2yj|L9?G~leNPT<U<JBdzboEb{?O(MSDq|Jo
z`<6yRetw=Q-^b)F2^KjBKcBpmF^3jCF(UmYj~K3g4h$sM(Vd=h-ZW#tAzvfWKNK&`
zlb`xsUSygNGwe}%*l&Z&NUPp};|yYv38<4dgKtYM&6}Km=LiXnNiyG6ePu0|OvS{r
z7}wg%w#9oN*<MqME}rYZ&x`Ocl3}7e3mu4F*0A0_+pS&m?VG~Vut?eCRdu{}`{2S3
zlbGVhI%|cxdd;KPbA~f#<8@WDG@=Fs+N{sh)4@*<6p2dwkz@X(WF}urb4i~=VyNn!
zvsW}<uU+J~q)oj2Y$2%zS@F~VQRD0^uYe9HnqjY;ciE>g!w{I@?{IDjLqn5+uK=p=
zxxv#AhA0hdFHjymjSHad0vhk{4w{no{ltCGF=3WU4d#v8b*Ck#b{K-0Cf23M&&LD2
zgAnB|9k+FuH1H<^WcFeCrKJ;d%tu9e2;EPd@O4R<L>=H$r>m4nI{yWT`w$ekMXIy?
zSMk2cFE@$G>v$4I`#4r3q(JHB4`0OHvPE!^<sB2fYsU9r2A~vQS3(J&Z&8PSQ2rV#
zGCsre&l*j$m!Cg)fL<=4#jby1@Ext?_rk%AO-$laoRE4dl5zoQNbn{d)8^#Nf6#)%
zn8e46n>V);2Sj1H^OteC&vQX>%bsHr*)bQOlp;+5MY$ZIf7F&u-fQ_~oq|?N$zPp=
zgOYWKb?-tgnin4#-t#^ZsjoGRl&RhT4_q^MBp@JutVyJAbUYQXO-)Tb_dJ+(c6-7=
zFT*htj@}Vt9A8Mkm38O4KDqHOlR8c7{slB6o4>$Jy;QH!hp}Yl+jRBWQh}*y&$P5G
zxGdQ{NMhPE;<%mbKE3ItQq@kw?bWOF?2mNIfxdHKL<VlLyour45!$d&Tr9qze?|r+
zuugO<=G;OC(%HF<ezHA!ZHv*2tzNuRLFd=+8AR61^(<rL8UMafMAM03UG!7y!@oXi
zN#DoB)pX-!DaUxJ=24WJx9>%wmuE3_L(FrMC(>lK?Z(3$|6m|x7#W3Z-g%7v$~?5P
z2Da=Kl(8y64?h1A)={weF6(v4&#Wn{^k@+eH^}#3>)=A7`j{dk#TXrL76;3SjO$*h
zKU}n!=-u4!mDoHUSslH8R!2(4zq=(~@ShHM35TLUoP6w371>-~oZL@z_C7qH4&-(>
zEM@Co-K{S9xxHM(xm2|8#l8GIkLQMeZ7ZVB-M+@<H4^2H$AQx_fSa4oH988)@xI#e
zW6mw2(4!tl&}hX00Yx!W|8Uhntd+lkUx@u^L9x{Y<8~u_a4@L1i5^VJsgC2($bi(X
zC9L+}tS}QTKtO<lmy0hl2)(BLc)B~F<?imS>sAUH)*7J(9~Cuud|XKMqx(jdJOzAc
z?9~ot*yU6i#$a8;uSfzoIi)o0l(OgT3BKS>tMll<pwXNk99tx8;6q)OC+QW?vZDO>
z3VPC^{3~+zs%0=}ua)>aX`sr0iYTHd6LJrGZP~PDJU;ulqN3c-4`}c_Ux*!sL`a5u
z__;dojc9^LRvI(v7|wiaBfw)&^LX6bxYRtW107Ohzas8T8DN@*teojhIQ=8T7$WQn
z{DhhZlys6IP}#R_AH#%RW77WR7baz`9QX*+bkN6--5LpzW+^F>CW@8x4|!W<P!-9L
zxL&I$1-w|7LxXEi6l?gpyyJ0sipqgB`~C|3I+Zg+-=cm@;zR+j`>~}oz;BBnN2Wit
zFTQAU#qzforG`>CDKz6<krq==IBX<u_p=qTYuezdq8W7g($BsFMMhj-lgAS24Oz*j
z7T1`Vq@*N!8hU#IQ2)CMDF}`%BzZ;xI1*Zb#tX1dktElfP7g0Q!LeWF*~@e8wLS}A
zpOa6A29MtoU^I49msW(VpV6xSo5lz`!x%i@0`A5L4BxJ9_vSjuvNMWGm}R7Te}v?E
zK0OCu!lY`uIY}>tnHso?6v)*rtV)hf=NtMhjA}#&7s88pA>qTrjFmch6Q=PmW?OPF
zd;QVPGnB9rR%*Io6_q@PiI3~@!KNkh!G#6Vi44uDMiIp%KvB$!CwrVT3YwfsUu!Lq
zUkVH~zE9!%BVjqn=Bgkh+`cjp8eT}ipcx?tdHk!SZ;&8IPbiKGe<%2nSX!`E5dPgJ
zahSUOUS#!tbZaz-Yt>E9eRPtq>+>S=%aX8f*?X5$Zu)W*q=D2{G-F>xjUThb7R?o(
z=Z=;un{@b5qUQJ}s%gOq80_)+R&qVT{QAHVKN06;s8>~1Q=<m3z~6bu-sq@m2uMQ0
zkyeh6e`!_KCwLWV1xx1mY7u5}q<&DU9l+*d+(G@mt2;moju`nPX(CUgq*z>1GGH&q
zSfB*i^-<-7xgP%nGD;7Ryw>j~B8%el-dDM}AO1F+0&l6UT#t!c#oM~o9{WaMV`+z<
z9hh@|^l~Ho)ZW{z^~ZHq;CY=aUQv-vD<WRwkwHq_JN`rH7GHlyb@&%01ugg&jIt^7
z7~5CpR0>g0#+o;wzL=0~H6U1~2%2dH2GWFb5xDQk;)C4I3K_w_T5v%Hq9qojf>aRD
z8KixNa{1YbzqWwoojgQVk*OL(!7yEn8z$;Wzn5i`h_zN_;&YwSvi#B#*VZny{L;Mj
z?W!f<v+I-P2lI>5H>0nK8!_w?BL@o^QQAlr!RS#T8_|MqY;Glnk8u)P0096AUYC*>
zY`rQz>nwgJJNl4TqCUU??(bi_ZgEpp-PqXpy$xltRZ%)h&HD)Jl@S0q#C(#CgbqQO
zVqcwCq*06g2|-1Us7XL2^{sZb!ME(Ux|ry=&A`v3*gtT++iDHj&EXdZg)uA;(<8~r
zVaWe({C(Mb=*1a+OaNrPm8<P(W`7%`F6RjIasF*?m4Ctwj)vOlmxp_W;V%?1l2jLI
z4i#Kr4y953@s&jSvlywNVmBbk=<6`Jh(nq{Ne8e@DSbQwB+U!_fh-7%@+D2g$GuMm
zbzi-+Dd1%_dibD3rvpVcCV~P!{ESRpbVd;v50EHk4aF8|>1}7w*w#ShCWS{BNhlnT
zOFTbhd;MMbkloghcQxAWH}6-54-!yBjrf1Gy=7R`-PS)21}%y-j5LZP-HkK@NDM<s
zNh{?L(jgtv(kUe=4N{5@E#07mBaL)7zm4ZU&$-We&hzSj{ki7F?cCq_?!DJud#$xs
zeM(?xMYXaz;@0s&^|-j@DN*hLUpuasamM#*UCN%mQMn}u71cNo^N5G)W51IXTyCr2
z)n;b6VTG)J_WSulN(xA{c)O2QBXFczaFJrw!*>dh{x;?$$SKA~TC}C|0Os6a`L4Qp
zh+p|Gq3+(JV|Pa8IMBtEpj6i%z+b(xM*Z;bdBZ4~(DqI73qVnQL#D2!^Lu=*dU#0C
z<Z<4iz%aIBQK?Y10mYz)yP8H$dz`M?uB6+hSba*4fj-pehP48e!2|>gyOf~aiJ+fn
zjp+c6Xvic@&t>u=@eR{=%<*e?BU{7!q7gyCjCLR8QbNw=9P0F04tW?L$}6=+ii2K3
zPB^1X5NT<a3?|%;BF&@Y{>g<Wd$0J+KCmWn(iN%&m#TssHZk?u%bFw7DrGn(Mr*s?
zjOV=Ntmx%d!^x}5?dQdtVW$Fz)fXG#LeI|HQdMw`Jzk;cr@ma>cDOdI`|6@nQje1`
zP0>R`Qp0Wr^V%Ny*rWuBw!I%cK2oJ-wImOnyw+`Hkk+N#5kAl(XT3W9W5NEAMG7wp
zOp28ZnWu3Ni~3Vi28I}P?VAp$R3r>sf={-Np*dny8}#ezUA4;Uez6Epgk4-_Vj{(d
zk|e<{f<ZANh4Fk&i1#A^WUH!=a0DY3VpS-ai10Vcf(u2zSH-j84)WKUF2y5-<UHHW
zBp778OQI1vVfk$!=8I5-T1^%M2Y?WX=wqRh5O`z^PH<^%_7lPCs;)MJ7t9e)Kb*C7
z&;iQb+}}STA`AmHYAoseCQUchV(}&&^f-cZI2B5*v_7CDE_^f30=5l!(G)U)rag}X
z{2hD*NYEkW9%La#S1z-C7w_MGIjhCmUfUTTZsuQZm7TgDG#%vA`mBdDGY2b%;*Btq
zVm#7%))9FF9i6Z6ke>D7UH+V+bo{tyB^6{GWVqN}!wc*21UcpH#MJb-*p%p>qYCjN
z87dQS)}oMSd3Q-kAJg68fKt)YYEzGlkfqFrPcrR2)%c<lR+{i&HHGi!=XG%F*Y7Ak
z^cnE)*XGY-f_&>{05vDu+uNWcpSKu$JyQkd=-WVw>aOBA8^aPA)}XO5l4~|MlJ?Vj
zaO7bcFdGFyedGJ4c#FyfFetFKU$i_6kftvz|G961<dmd~!%OoH@J~@(s8tWslcS8i
zJKUV~$t45_+lz8IJfKb>g2Zg+4!&k*5+&xFn?02iSy@Q?3h;}S9#!U^Lh}Kl9Fj?O
zU7lc5vl|B#LI*TnykljSACumf6?>NsiN6m)mOBcvfV^>_jtAS_udJ|_NrsNob3Z&p
zW_(T(#|>N&Po7fQMM_D-iEulb+@2Z;+z_4|4=ySq5&5{S)0A}R&NPQ9H?L2o|H>?7
zY~T*FWI?dNyB|jg$51VGvdl0YkOG#LhM>kLfI*}C!mrud%3Zs;L;qU2No0}2bIbP*
zI;3sj$D@x4ij^W|1T{6x^vr0r6l8e#^T@u2cTA+kq?zZe;-{0XSR*&36J&@6N~7Hm
z4cu(#%mKZ4iT95=Nr_On=jR2Lw_enZzkSw_>hg-{6YeYAVAty^oewObD0QtMF^{{%
z6MNNZq$^9v*Kzm@-PUc%>JHX}#fIfG+J1HvN-y8tLvv1SP*iFrsWdc=hPkFaEDXTM
ze-$g29FrXr^EzGMrogS3I2Ml*F&u{|5U)WI6^@qk;K(<j_)C%mWmMkYM~5^hed0PW
zU^CPL$G!gi_|)$BS6A;wwdOh`C@2@T^b0pMG$r<1tISX=-gVqmZ*I7isHsY*A61zb
zJPPf7L_b!d?xA71tXr$>j~qdY`8#L!S`=$Vc~ZqA?pDy=jD5l0(od2r^s~}@qDHCU
z`;T(pf{78`>KD&0JP&p9{g5n5%LML&Mc_zdh;F^M>B>tx{=!ZD<XV)d)IKZhW;=6y
z3f10H{}R4tw)(uSI|pvNw4TzniR#d0oD$p2wgki%g=}_?j>X>4%LGus``(kK7S3ND
z!%;?>t3G+G8jfT+-NhMr1<A+7u1JH0Aqfa8#RNz}A=xsE{}|zI4{}z{jKllJmlv4Y
zJoZ9`nuo-Z^7GUAFu`X{co-PJ8&WExTP5e;+(kc&r(3xjiEF9*aKfcVLD(`mmz(?@
zePLXJ$*j2N9+6gwnyMV}=GHtec8*u})n+h3R$=MFTdrZ*;KzKAvG{GdRe)2;Z}~7!
zM)qAz%NR*iN5|+e<wzQi0!otf&NbK_!U~QI22+yPqhTmim><A8{<)AC%6n6KJwN&R
za~3KaZG8>JagLU}VQ4g4fE!Hqka&l{d~Wpy#tnZ^ZY&RpK^foWWCt(`6C1*0)ciR8
z@y8iKkEgUripL_6Apz3b@cLEg*{U`-Dh3gc_gb@fYrEr@-p-^-3~y0*Xedtt6?))6
zeEb8PXw%(5GbRzSXPp4kTeuMv<OdRMWATV>ZRZxA$XEqG72vZ-e2Tj&1gW5Cve&u;
zyVsCzN7sEEkArRfd<<DWM}&-vZ+*o80bG@gKM;+$>H_Gx!c0OEt@6%6K?nl7OW3Xm
zfVQ2SqQ9Pf(2H=Mr27ft@OgCi8>!yLjAgk8n2iNknA@n&UT7)tbl7`e=BChQe}AWZ
zH?mzWJ5eC|2KtWtel34dm~(H+H!td-Ac}k6;SV6MUiE#w4&=i7jSPsH62cHZcpfli
zgQ$O+u>Q85-TOc{8S$1ZDHd<|(eM%giwTaP4s^`x3Z;CvEBDckS@5mhP=O>Yub$H9
z%oK$sJ_#jdqPqEj?g7M%P@Wq=EX2X6aaK-{A(R*mgs+-%O+t{IZv%pY{cqj`S|NqI
zx+bug8VB{Y`eL5`en&K6A04CA4&87k4iz4Dcq)Sw0O(Vru|;K3qqjN?WbF^&ZncU~
zM^Fc{B5_#X%9oX)^Wdz}-@l+lISk3UQ<lj3HABoS!1UA^;+!;?{tdld_shGY$zKwr
zGZ<IB$ju@S@K5u<M|f*}c4%ho?W|eV<gab38Up8<Hf)&Symuc1!4KUM{f89Ky#a%k
zX?=FtA9Zs%JDcwdX7(Sv0m$X%<`xngz-S+l_gRmqatKEqJ{TklpgcU(^}6|&Jr^lG
zHyjITiwN2qBNvuTjb`9$-$#GEX;@`ph?#d%WS}97(ujcVF+vFDILb{mGEYcU#3?ay
z6lJ@=r*#bF?aVWM0J%K@{>R#&7%(|pQs4z5m(_h?)Gs$%Rj^Z>5SQS#HoL($I?Nl-
z;|o5CJsp2s5J2@!<w*qc70-tF0tM5!7Z&#5&oxbsX5-gh2I?!7o}&gS0gQLl?$ujb
z$A8m*9h95>+8v@_=bNKGzrZXjNTCJN^h}miVv);cnuRSZtkq9KU4%{jEE2mE=~##u
zKlpZb#^!wZz|l-k6SX->jIdZ+@mD`rB^Gb_UPghQsEsJfC@bp-B+ua5qS$D9dxQ|~
z`Lf#iCdmAr-EGL7VRfFzf@aT@pd&R<(6D^zeNSq~a`e)nsTH)4Gy^MkRhTT8l&1en
z#-J(VOJ?#JjQH&qI8zOf>*T)`p?GTREBlUuJ3C+OO@>0X2(gP5nW@ZgaL!jz56iZ=
z4i(t(Myl5F)1c{yLDqHqJ;DzvRiexBy7Jgm6hwqIN}5iWN~;^x5SdR;a%K`So#(@s
z%t!EnbRnHux$AW6(z?REGY;WE2m)2tWIXb2guUS+x>H$AnPIT{Ohyn`NU8h}apWy6
z@eB|$6YJ5$gmATqCSARZ2PfWKcG{m~gF`p$eC*dpNXBPqU<sByS{B9Hb2(AD`CTl;
zCg^J-GM>|~W(|&2g)1}3cnXS((_$whx}Kt=`7u^}=?j%rhV~GKCbB8WEj9i}WLrDk
zf|qB?_lj!6B<Un*V^q_P_I^(1+ty#7J<)QQm5~t?KYGp0J!~wY*rD1>pez*k@HQxF
zzyP&8{*WRe$hW;4?NYK>^;r+zA_u17A-`n-SOHOFgYeNg$eH|fvUdYbCr$eHt*?%<
zL|$m=%j7gCZSo-^VCl;`T{2pm@No%-thx>b_}7)Pl9{K9V1(k`07ozNj2h=lf~Bzv
zEc@c6B2)<57AA`sFBSjG1bN*rrDze|8$dx750FEyW0QzO9zH-~yu$%D)<hGfJWb)Y
zy~%S^*2rT;n)|=Dhwy$Z7qC(6-KR!B<Md~Yj4qy@s9+2|D0*y6@4OlG-MIWaU#>g=
z$1ZAFAJRzPmbC!q1Qy8G)kz#Z5&P}zm;kjs<yg0cQod&eCyE4BPm9^brg{?2_r{`m
z(T3FugX>WM)Pd>HHvut1u&l#Lf`0uAl|W1=3(eye^EE1Hom;zE+GHL|l&RK`J?a|3
z$Z@-0BC3OS{20$ZR8=(_<d>hw>a~6g2Z7ju;+hc<f0%mJXrOSkydn3hMC2Zde&W44
zjfxyja1tyh=rINtxh?ta&dMZZ81vFI1l-qZ=D9tW%_r<5ni5#D0O!d^%z~16EG*`+
zov+wfdivbPAC(dtP-bsCHl}d}Y(^PJLXicNJi#qU;YZMcS=yJj>3zWdrGwD=q#+Eu
z$D1SZIS}S{h)!Y!2QHcHtSqSVLlO*bS;4y%FMz3$Vm2R$E5{tKhLe+11YvU<AR@fA
z8?@WkVsvRyU$Mw_csU)vCqOiu5|(SiG$o}71(gcC5W-QMU_T_pi+9E=(T4?uDwJCk
zn^nEP_N!d*(G@OB0Swghx>2il7D9ul#0;5RZD+U+09L&SD%_J}Y)vmGII%6ea8X<w
z7!Xy;9v3tmhQt|^v+_hs)1Z+IV|jPc0~v|eFk3&__}`0}lDKR_zxzdC+(#72#1#Mi
z>H8Pv>O^VR4Vn}(L1Itx!8Pl#BSKFWqDYq+`l1K1e#7f)6CyWX&lT+TDBK`><9w-%
zT8JWw(EeRAHt2i2drbYIiYH^Ox@|c{E{^%-Y-IeSh|MiOd{C_jy1O1U_N3L0L^NaY
z@{TCi;6KpURf;ByB8s7?r;H7oVf~tbcrHD*|C`5jF9X9iIxA)i)QB0n-Bi*R(daY0
zYjGNiOF}l(mj0cH(0p8Lk|3D?t@is|nQRf_OaXAdXt^0}d&D~C*6xzB)c6=Y3BZjZ
zV<IgZmCE*dZE`n4=GjMTcu9zF<%e6hZmoUQxM>pL?{)a_P*eLs$)m;wH4=;V>zb0?
znt_2fu)I83XFJ__^^bbb>E4-^^Y8|oZualvij7Bc%qd90)uWIk1o;%`XZmdNw9~vt
zzMOY9XQSM9V=JD}p-XlSHV*t><bTQ>w2Q%rsX>Lb@7fBi#MFGa6%j#7N<mYLb?ul+
z7vBU!gLY?Ihl5{w?&P<ow5%0BW+Al&tKk&yekf-i?#$6OZep|otJEz;XmoG94J^>v
z+<jZtbCKr-Z2?(QsgY=jR+=jVz_lGUDAS>DZ5b=;c$h!1X?=&#jga7B;j%j;1qSRr
z8yTz47wgBlze#S#bAjk_XEFOoXZ}H8IwGLujt##zM7=d5C|F%f%i4@hY93krF()gl
zM7cs?pq7A0z|PjZCQ*y9mEn*E63|j=n^kn3J})X30fm-(e`57D$W!vKogl&a0@o}Y
z(?_keCa?+abjBT>NVcXiLO`BDm^=?TTJT_kvE7iICSPose9TquNBfsJy4XWv>Sttm
zJiE7k{BT1J&i2bPs7F*h?zM^Y&51-|(_&zBRNR8VZ}OmG5@?@4c_Le_T~|^mBiP>Y
zH4cGD$TJCe{_HWcwPqi?Dc9OeIwC<CZ=lxsV!zD(>&?QAlk7KJJv<m&KoTu8oOs<r
z(M$NOC_t^~-pXO7*<*q3=~U=ti+rL2a69IuxbEY|qsjJBFKJc-Kl(X{Jq87-8Hx+9
zYCo!pehg$CHEy!h5AvkT2n`LW7jZm3Ay;&7+pr1sorip&dJrAV9Vn1zf$fW7^NP>P
zIZm9`&;#&S(C~W;9LGo`6MbR99_AycpND>dhcf?*I7z{s-9u8^$itbI)*Bh#RF;{x
zPA;rEN}|Ua-khIAV;?}2jTA}@*(UTmK|(=$JF&_a4OQPcom;qf)vWVTS6xUK5;33c
zF6Qo$7z_YgHK}$iH|n|fw#@8}$M=78bH9K;HX*XohZ9X!7Nu&SbZIeW+^{=nGY`~P
zUwa8L|8ywSI$xQaFx{T<NJ^2C$F5>~Er<2U9zUP8u)tR`zF(BMtmzuxoBj+M;kWJU
z?bx(xUNJ0yQe67|^PVP}+E~h~GFMfA8uLgemz-MX=PU&rDMZ&7K6?Sz7S3lYv+y0z
zZC~`FPEj#-=VFODUg3s0M8j*u7ak?8PQ>W<R5$cFC@X{u5Bg%P?RwdqLf+Mq{T|Qx
zJW~Si;$nIL0_^hkTvJ~S#Us*Qtmk8uC9+9hd^oW_+K$?eFZ=mEb?DQaSO(tdMC*+=
zND2++z2V;>ews9FY|pKVN?H#;TfVRx!;@tqLhDpvrSfBb8CLCnnqz@cotvEyV+rx+
zdt|Cti)j`hJ^Shl&RuI`U~Hk>+)!Dgww@QBU48qLmiF6iFKF`!d70v!jB8BRpQbtB
z8EbP$%@rkNN(AM`@}r+QEfo~`wfNTZ$9O*KBY4qy8$;q!XycPMABz4xh(no~WMIGb
z3fp;lpAjNX+t5fV1TiouC_8lbIuXAn<MwVR=vS6Hara!B<5mtWfP>Q6e8`w~)=-&`
ztt8fE9=@YChRCT)vY)KeFDQbaFg=i936IKoR~Ym}P|(iUezWx6HnD0?c&N$&=Ms+m
z-Po`uk7@#b>Joi!8Ai<a>JK`Mo2~gun`yKSlYO&g%0AXl%?|k)e&^~>Y6(tC$W8gx
z5sqZK_IW9ty!q7bgN&dsDogP@D_P0Yz$d9&eIigs`7q#j+Y5|By9MjjdfmWibfJF9
zZAk(GjOB_qaZz6t7KkdGKMZ_kq5)S_*biGu46XtzCFsg8^5!qI7eUl7%|0h*=Vp&Z
zvw7*iBBgP~2whP$MtzM%fX)OOs_N{{xFJK~PXk)k$)LK`RlMB58~Q<3HA)J;vI=te
z&(AS|16oq>E?WM0#)%PPJ<dP^$Nlk;9CG&Ym9vR45o2u}2`|e&vFfs8xXda(wW%qR
zO>1RqKXo>3cY>5-!AMxlmda;=A~qWF&g7CTqMa30$#vdV{5TBowq<(!#%GMU1pd09
zS+`)tgQa3E0$wQ+;vzyEw+fQwqu0vOKYt8cuVorqTfIUaEedS6(EbAM%#d21$IP$H
zUn7-Sb=J@38N(w{R-`^rDK1h&3M6XWOr#(k^Rw4#M6jwj<;%~w7-h|o#44RKkRTcF
z&wlz>)Ip&K21+T?m54iQb5ZTOmH@rJb)oqe*)!ni12w^XGaUEXUfa(6GZ#R&Xd8F?
zy_WS#Uz8)I5>;TdbS@nPG;=JYrAirzLUwY<PR?pD>dhP;Nd5RVYt?rBc0t_GBXUTS
z&hTiUoL>I7==__<dCUfA1TpqkLfWTYCRVp1T|pkPYm%<Sm#Ti6y2CPAJ3Spc=IFOc
zjsc9gu$zw^d7kQZk^lH{n&!0X7}lxtkSc0t=ldC2WV5y3r_|FRUOf6vublb86iJ8$
zX637DXAk#smNYW7PW@e<OA))~>CPBx>}lDT&mJ4mjmwq1FjoITD0;%uC#LPm^#QIY
z#LGk)FIcA_h%hKnQoKW8;!bs2EP%sDh*%!x$JyDLog0e<k3%rTR9AYJXn_uR-p5K8
z$-l_dU{s^8T}6txu-pX5-aB>XgKqIiAo|#~<LRv}8Vv+@+1N~M(?wSbs@`gPXSe1{
ze5gcaCn%HQ0}!PeB8#v>!g=y~r$B4Ob`idIgGCPLA>+ralyFZ|$FXByQ&lwPZ0R<1
z5rPJq9dlr+r#$`ni<O7^ccLCFS@-!nD``)JzmKY{49sg0aaIe;2JO1GO&CQ6qL3IM
zQ8%WYzMF8+PW3h^DyG7iU9<eReXMV{>G{tdJ^Q(c9bv|ro8~(n<0Ww8^v~ZEon9IL
z*6M{Xzxut^IGfcdGHGMtPv@Cg`KT~AjHDQ$aD8yz9yGg~yEQ{(#D#t97PH47=s7Xy
zp!Px3?>3FHYJRFPn{OXFDI=(bi<g9#*CI@X>hvH_Sr8$)4{MyiTcRI|Rmye~MQ`*;
z0d(g~w0c}~pDZOnnmafD+&04vE752J`R;fp=RswaLnqzUl2za@?AkBR=RwKa=_4|N
z+xeB=c_%m~HmofHE!^Ba6ix(5$SttzjSVBX9fR_7pAJEPr-Kk1Ic+$M9c5o0?H$mv
zfQocm`pMxF^Wd8q=ql+TNT5}y!wSkifEHMs^=>1jpzRPlH2Zajk`8NNO%YniAPWmb
zYhpJ}*AWXptNAtFe6exyGT?sw#zw&7Xt=VnlYr$t5;{HEf+x=6pgC3Va<lY2l4;dG
zI9NIeCs@a0c8DR&kB{#zx^u1Et(Wg;?Blb0BT59ul6^&aIP!`C(k;{aRPUC>UVdjv
zX^t$!%5%JVrHdaFaz^m!J$?T>6&qWF60$56A1*uXc4vKDrn>qaK*1#hxBd>AHxbl%
zS4$)!WQ@g=R3M>r#Kw2neTi{?zH*n)_b$s5I^gpP*Dl1U`-P|PoNRH&t_|;OSzZ_K
z<!cA)zH=aw+l>f3BIY&32_>37tuA+dx*Q*ke^=-}9htOMDPp5=hncjdtMok<NQN3M
zetT}wTLz(uwHT0zMHm~`({tz!x?jFkZEXu*mdwE9<VoOdIvk~M?i`Ln7Tla=Mycnp
zB0^>EgoZryu;7SKlXi$zG-dq$y^~z;Y2D(@>(>jBH=ll|ruH9Col>cOoc~~JqIs4M
zeTNYO@}e!7O%61a-smL8q~tnm;bO-Kq<9BN&yY*Yv|s4!4zakow;0ba@kp3`oI?lm
zY_AuZj4dE@9~S0Do1V0PPhF3?c;Pe~UCujQ{V-5V(Fx>YuoOO@N0u06UK<^hF*LOI
zM$}<6+{l-ahUG-_k;m;fkuk>;%zqkOl`V485&uO6fR+<~f}l_AqF&D4#rN7LhD~O*
z`Tas*YrweQR@ytn_U*(`F>9|>c^^rFaGw>YeozH6Yv8wDzMS0(9OSTI7~>|Dn`W}b
z;u&KVd|u_mvoH#`nMX#)DiSj0Gm+M$NVCCYb66Ad40F|FMP6n(p0-H{o1fnf!0Bkd
zOqcTF<|uqVCGP-}WeYYnVHgg`sh^ywHyTJ17I8yy@$;t#<i;M6+>Sv~JlYz+v3{Pf
zDOq~6o{+%+t*QAmKss+qAr*=9h#|Xc#ADNVT6jBoGDCbRIxY9ww~I$){=MyGQRf%A
z<7K*Fx4t3^qB$ydPxlvt<|11}HZi_<Y<4yFn%5~ZThlQ>zJH($e+^O?C_s{rECGx;
zSW;flA{tpB)Mu^<AG|fhp%?TKVmTiyH|=p5@$3#Z;WEY3iZ4n4eYt43TyU*`==Z7>
z0+AcdRTv7;w5<W;^pXmz9@dwYJ*p>Ewl6z=yqs$#ZW|8^3(yN+xRJoL{N1}2zd?Ln
zI3x=v*xj$;sVkL3FKGB-EZZq9be0mU*hen}yUE?GMB{OoU7u`3YbgWo0_f%{rem+V
z)UzvyqC~STUwO^AxuS1k0KfpZ#*#^f!si4yh4hU?kutgx((o?h`MBKd#dGH7vslgx
z?)!zywHQx7LqH~2#-b5gZx$s>j!w|4_PI$M5%K)RAav>uA32|ZMB_c53rD=!5?n4E
zyl1->5c?aiV`9qb0a{J=XBvs~!Zak^S0V1lg3mXbYY~cwXsP%x_s>aS6>06h;Ri{7
z&79<_CB^TtP@>4RZuB$nV05@TjrI61?PicE_`(ih80mKnh~GImp|-yAZS66-+l4Am
z3e6>%FadU3JC6Ld_qnm$BRJSZQ;PYDfir2O*y`^ph(K?{$`Trm&mkZm1pq_V)ZiT{
z+ef@es?0L^ndS&TvUt*&-yT3*F-yx8$BTCbi!tNosV~#|z2<1E1w>kI&_^?8Gv7Rx
zWna(;EkyE;?6wgsAUg%DY?&2e9~RC-ViORNg4Y~QYa^)br|gp>kY-89I}vFyGVLr;
zHP&35&mJp&{)Lz3LX*H6^CRW`{X&Q*E#WbqOiiZM_fA1H9W$xs<#8H@`XV`A@BCM*
zy3{wFb-CHU$Bb+I=0Fo|HQi(tOaE+!weRr<4H8uaW%AnO7zDRM1(JFfx!rFGLh9xS
zrDC3bBm@}Lx2T8gY(!sXAZ-}TYR+N4AN&V`lo&?z+N=`W6iwrnSx@^!WN!TN#=(@J
zZEZ<uaVbRrBwhB+5wonG-q0s+0INfw1YKBQst3->62l1VQm%np7sHwDw3iBXf;|i{
zSzvk=3Nrf+4iZ$s;P5V2>FsbooeJ~!Q3gevQi8x_;|6%&<u=I;d1c0|Mv={z2EoYO
z%z`)wD_Nbs=5mxb-SR_>GjCH^4iu#1PI+*QmB2h&pFfWlCf`wrZ?rDI!9ViydPL=f
znu2pZn^9!C2<9Cjc24Ga$W^rGHBW!*8;)yVu9C{1Rp@-cxiK&rKO#y9t{jUzE6*Ph
zCX1C5Z7Jb<O|0uQCb+T%LgOj7L5nfOa9QYrBzEcW$DFvIP9|1iaR}~7yGOeUgGE9j
zVBU|Fpuj)hA_Qq5{}M(@8KPXG<J$)y5QqJ)?d?J0&YXnAFOO9tL}}dC*(%0dJVjFo
z(JXATFCLbD;$OdBTt@!Q&Mnt?we;KFv_3PNp8*D3n~4UEadA#<cxzLk&q*4ciPG_}
zTv*Z7xz~Oc6>oJ&T>D}gh1SUYWrQCHf3f_*xFVBlARy;Kd~J83q&7T@Jc8#gnuR6$
zB+MMR-9^sL3E0W6VFHRD<H&CiyjBKXprAcXFc|YXj=W%rs$gj&VWNI^RRlYs2*LjT
zQ_S!ocy)EI-f227d(&1)DK<#62A+m(c-A23S!Xkq6l&2Uv$TQOCwU&1Fz|vQS}%_0
zot&Wesh;QQkIE(BVg)I6pp;?CG4xsJUu=$n(99|&{5TV4nekOqd%NtO8xwo8=*mkM
z!OBr@__Zg3geI(-+HlNKc6+Q3y~m6(ZuogO)t~l)is<ktkRwcq<|Md}rqik=m@C_J
z5DcGZTv~d=&R=Lc$?dkve_a&#f&^08NXS0B&ENVU;jUwovUVsx)HByLp+jG`csl<!
z?%WyJ9jv|yA{B22^r!Az@9gH7i_<{1Q!a0kf9l2r5$bmhR01<0Wj^T>{#Ubx#~?63
zlNl=N?bO*2{On*F#gAETwQ8Dy7v2CGi2LNuyGMlsYStJrwjqSVBF?$#{#dW6IN;oh
zTJj(Y$P_;ni8K?Kf6J5<;D1oUYjtL>4Cewy3l1LsD}8e%rS18**kUNmHQ~kL9Ienr
zW#v{!=I-Rp{2u8sZ*JX?3wE=u>{>1KK$|p=Q8*{>6?o)632^i5_G_fhxIOSb6cHX5
zJ7kVm=EF|oQ;-GSYifQK8})W#ZYfGgkP;D1_dH(nXl+6Ms11Lt4ez?gY8Vbcuk})W
z<)ETCs2I%a7D7{Y;7a?0h|sObn<>`!m8UQZt66|v2#&miCUoR!DkVCJi2KQ0iZv32
z>p;HdA0R_chsmGHv&(2`^>bJ5_~d)Ki2eC~;ksPU6oajWr+DZ<$bhv{uOngVt$X)a
zpzyMcsvq_j$9vsCx;la>cCpM4!?m(+2P`Sa-iN$Eh2+LE{=}8XvgQlDkxnq)3ep~g
zAn0hui0Y=EFA4#S^-$7;T&b<M;W|xbH%3PV&-ZNNeXL(2e@28Yc=Hfes4>%wms1N6
z4q3>tp|a}hh+0acBqBMyxmHZUp-+<zi&b2t`<V7kk&%=kGSlpdO^SjfJozE03lZ}u
zx7KcN$h%7vxp!K{^3%Cx^8s_be{o`-e@qO4UMv$nqY#9SiS&E5b^sTQ*@XeZbCUr@
zq#=@tP0N7pjLuB~2^AC78`CZeSbcIar?KCF&+8#*`Q$H(5o2%4g6Z7IyR$y?Bu>Sg
zRE~+X{-LE{NlZ<;0qfHM(E8d-giwT)M`YgPd1xnM`wd1Jxv6PxE_X3b$Jag`gp^?-
zlV1q1X~M7&MTih7g?+8hjyD4qe7wE;w#q=Feb6n8t;JNq<9VlpR*vn?d6(>NaF^<*
z{j8rjcLyi8l8g^5z9k&A^um#^W%#wf4{xbII2dW5dymBkvh5Q&10x%grPPV!<bk`M
zk`W^F=-Es)v+17@^25ULh+gE><J%^6Y;!cvyRr*dj-jM#%$f`c9Bj72BoH#ZNgX;L
zo(qRx%O@5SbN$VhhLjTasC-KFMm-i!BWl>uJX3=9-aQb$#V3F7z;w0sN%k<=4g2H|
zH&>h7mX|SJ>kw%D{<`PF%TU;f+tI^5C~@iLx$vmkLpO|$DPHC@gy8G%wh0PgQ8e98
zgv~GE+XoH~QnJKEOFw}R7QubpQ_Wk<EAFE^)nPZkw2{vRQFDr=zkfnxbRj1AoG!aC
zziNl_a<jRt#Gj8q1KuUnQ<uwd=--vVm{@RwX6gnZ#al!OP6t6>((Hw^@cAPJGf*8F
z_sq#@3<i2l$=fZ0dMk050#rzgEVT->;d6xq2G%<Xi|<k@tIkz)3{Bm7*)}+RYv;T1
zH`kP(@*B6Pk$Z!MkUx^nLLUyg3O6)x56JmqR&YC?9`xuaQ^-{UE2TdPM)y0w)FvFL
zkCpTc(vSdCEg8X~L<F4kLb#g#0c38jvp@eSw4biY5P;X5q_8&!t!=#5ijtXON!I%I
zSe;1xf+9;Ttflr!IQ?UwN)<up4bjd3MqQXW2*xA_c3eL`&oD-zw>g<er{B-<+ydC)
z*tpp1rY3+l1FuRESy+3RQ8Dzq{Mj|6Vq&1}cR~32@_@?EpWELaxz5Lse8_rorE2hF
z!zRQ&m#G%d!dIU^|59r3cu|@6R?E9w2Ey~nqTyuR`31=h?r1b`7wGw&lbaj6%kqf$
zmm4UG=hn|N(Saw=0ZtLxK2_Iv%P5|cEOT{B(ZLTtai>p=4CX?`^u4wPv8rIoM8ukN
zp)<SQ6P2Bv@~b8qx2|6otlQPVi~UAVjD8ohZDRoRmK#DFU;GrjoRzHW01*Ic!>6w$
z1slW6Bjp9>DQM-ry=(Pgy%-?LYOLMLyxf(`c{3ry*~#qvi?_m8nv-)bHVkJ5#jjVl
z2lqtd1J6#?!A%>A7{rv_!*PI}8qExuEVu<%fTpBBa0HS@3cFaJ(>VVU^mo$zrU3N&
zI+<=Po&nB_`w`#Jhb?Heim=RHQ0~-z-sgcIw!GHEkG>K(S7#2RsAyn3`15BQIa(EA
z+y|NAZ#eqClb0LIz#noXexIaK6oN$O*~>8KF2|bIJjH~;lW*aEWI+F{^B=3wA$2P7
z!{GuBB|aC&;TODaN9Rcbn|B(15QFyb+T_KF4+gWuPl*WsV(M!EPR(;KrW9M*y?k=3
zm3WdM>(4lk-r{s*#kF82&wU?IBRkQmn@*eM0N5E)g7Y8VID(L!><6D8d%V-(eLd)B
zQi!}W?}>FGB7;6befILDlP1xwd(Dj^Kdybe;xh+Hxb7z^0K<G8yP#xo@-}0vqFa3&
zzlcXe6s45yhsT0g*t(B0c-7&kh;W72Z#Hz+o}4eqeB%($@6!8)W$i}3bpSG>6B>Ut
zNY^pEhm4o65E&YTRb7MmWsi4fr}ExH>UK>vJ8(yI-h=^|Uz;BmoR)40UjeJ#jih|Z
zkfJ*>iW_)HSsJ>4-hbNBK?=@Nx4Nfor;YCN1K?oaklY?&f<$Y!M@HoozL29{H3N|r
zXt6xkOQ#?FCHl$hEq`<Yy8^VP=DBqXuuJOZEDj!XgQbKubx|_bc<|?xs)Qi-_Fo9R
zP{d_sp6cwZn;N5AVkV{GG0a9WN*K{c4Mbe0jK$kdW-_Mok(3hzi?JbgPwiI(I6{5G
zZ}!`V^8BQ~0-WWX$ZrS6f9@pYOr#hzw=Z@j98F02Si=>dTGkuXT6s7z$h9Adfi#~C
zIojJvt_TL<r^do|lsW?~qK&LM8D->{N#h?SxlWeAF@HMaCyxYkqMtukTN!Z$^*QT&
z7!%bGO`m`s6LZ!t40?o6G5El+&CftFC-&?~nqdN616=>~$wDDQkVY5N?O)29pK@Ed
z`R~J^RZJ*#b>DHL4!@f`UJw%WyGOYbv3LNf<!&OQVWeD@S_$tDHQk3JR!8SLO@h>N
z!9oNdsoUC?m%n6)VP}1a|BD+i9gI9xiNH8a`bk(=1bf<Pnk=j&Ypfm~RngHK`wlrd
zcrbH*eYw%mx>GxS=SgMSU`*84QaJ2>u>Y4VCi3^;DO6v7dYD|R&?)UW<B95KZ&}Ho
z%YH+M8U?-nFgPoIa|rkxv@(vTZRc%Xz8DU<;}~vpw4FTb>b~_Zc+61-E`$Jw$#)=>
zwL9=D54JcL^KOgMi-$qM%rNFh@tn!>ulN$!L^^{$+H4t!$#Qmw0Xw$~LMR1(m)DYe
zRd%E83L&6Li47y`lb{8Fc6N1WI!{gZlDwK2FBI4IYQ)ma-r&y5i+<Izr-RNN^-Gs$
zIh+_s-_mq8Y3ybIKCA!@h7u@3n@Br+uS45@S5eB!f~rdqvE!m*w(Hkpb7SLK6c;O;
z*-9Jj-rv54@zX()#w$mu5=1#uh70iG>Cy&zaj-EmRjrIIsx^DF0{pJ&dmn#2hHAhi
zoeons`KFMwwu$NXi+KWKC*zj~7wGSe5}~U;?0TI<um!b1NmyW6c&}E~6lL3zc<}Gi
zZmDb7H-4zB><xXM5(*twX`!?Pynw+O&HJqsZhV)J6e`KWX%TS!@skJ=bmw`}Wr^HI
z=FFPB;96wUmt~60;J4#@HM3{$mNQ<S?c@w7Plp1Xtw|AwIKt_mhY7MbL;U?G=^?VR
zNl9we`2sNnPL6_KKtmN~(tUk3DPJx|h(elkhDS{PxD2p_tXE2iBbycy4b_i=-}GH9
z`bg|u&|&}ZE8e_J-t>Lbci!LOY_((}b~xo|sph&-zr<em(@%$$j4=4ej~^PY1Jlz6
zhK3nC9lhA!Z#OhFK%r140d}^C!vi_zO~MmT9-p3MuREaun{_!$^KYwmNYTi;%1Uks
z;m41)4qDccXY-Rg@9G3WZ4#-`${NJ6ac;*}SXj8IxOkzA0PEVeTF4Z7-fYUw(p%PB
zeCbfN>#_`Tyv}p_daf=jt9(=#mS(ssX+b+OGLqYzNvtZ$^(NqTxA?EQ%%;8Hsb$9(
zZ$GGC1S%ZrH8-Oj^Y6Kg5}(_Ts>(9N5aEsMJN71i-edOmf6kkN@?Tk50qLi45_aMP
z-k<^G`r+v-0q+&&jYyF?V%6}DIk1qbWy!PLw`YvN?<m{YSg*v9zCD+0aNE!$;<QUO
zR~uXpKij67q$)6R^Yh>LVLn)^nlf;S5rNPwvNwbrlNw!ckB{B|<S+!={<DOiBO3K%
zS{pmvbFC|Uv}OmAaw?mA<NCuxLnw1>>>P!uUr!w_UR^qhUox$3-m6BWyY{9(Fw8kG
zY-(z{`;D!0#R+r=+lsOc?K6XkVQ>lwtvJ-fPwS2!pZbnkrKhDiHBNu0<}e8J*bHnl
zL7VmoIc(KmVvv$%m9o3QLT!<EMn`$PMXP@M{)l+|*q!-K`4%i^$M*jH`=X#3q8K^Z
z4vaGFo{DM<aL)kROVO{8DoKGXl3DSp+eO6{rr3&#tba|M?vBFco{Fj}qqX3KIAr^}
z&k0JK_IGBzKD0`axwxFN7_M860qW$SQ?M4ehDO^P^rug{TPhN*F=5+@Z{u6CW*G9`
zJ^r3Hs^%P-{T|C@CbZwvLPe_(B9o*E5@fDk!l{#Y?dO=D{qGNx;%U;i^fgD#tRG?Q
z?SC!LDaNI_fqvI*BIK_LT@CP$LWbxigEFckPYRm$mHd7`?l9emDTz<msYo3>swiC%
zAFP)xNxFHlPmt|J2n@k=!Tk{%+CKW=)zy_g6^!Lg0bKb?I#0o0_95_i9O-r=4GpAV
zk`evvlua(fudNm^g>s!9X#5((;rI~tkL6kTesbo;5uH!|l~d~K94hBxDLsieoQuMI
zS9aSnLufddr&~>*!0ic_ZLD#1BHpxRLQOp-`VK&6MO_EB0rXG4iy~Av&>q$%gZHzj
z=zU`Y#=DQ>!2g0@OuAjS7pFpRq5!>lGggj)frML1um=Y_Vk+vm-)jD1CEYQgB@R)E
z2TZ0=k}snptsq1{9vl{yES@%r%~h1Oh^g3qhL88ysJ}f{4hXVaS0g(3u!$|<1V1b^
zBfXlmj@Re2EczSQkixy~ODmvF9vZ(t8Zl2GCBu>{$2^Zow67dhkN<B|z4|Gn;|Ygh
zxW~rV?T-GviXJ;oz4iATg`LIETa0;7k_?BY7gbdCKJH79D{HF+bv4V#Ilm6tD5En}
z$rkJu*_-HgfDrUeu^r2-XVy+WVD-fZ_Jlh`ggn9ipvKdTw7xkZ<FDmi4PK&U-PK0!
zKUeO&u1woq5};EsG~3_lYpeiOE>MD6#E0<%N)}(455OVtkCh=I0(LdOK73TZ-8}--
zvlftz;W*gRaAPS0)|h<H0(=ff@_O3c%+b#CVUfR41PLK0z^ZUqP^Rs)34G3Pg@9ug
zzmu(E)rv#-Ny-_nLjKV1f6wp9lU`-oHdeV;d?%-v@0v3Q-?SIxX49afMY2?tfoCD$
zuZcPSjUqk6sf1S6_CGN&2naZFu=U_sw@|#e!2pNu@kYm)C7;9PUY`&s7zGS+^=rRe
z69&zZQCNWgi!=+C#WFiv_I@-Dmb!l0FMFDztM^`UCJ3PbyGHA;U*(222s^o4f6RnJ
zE|@4m0~m!t=VJ$yv78{zUrz~P|FykW^y3A2Nl9aiAdO{X)}rSHu-AAUBo-x#V@~s3
z`u^CnS3LXYNp-m2pt`fEJ56|!ge|{SiCzgz@cGp{uT+bWyx?EYUF|-%AKExBp3Bfu
z391at7<9Q#?nLNcPlI{>+YI(%5<vpo#C?bjjF}QeAFNBJ#Q5L%Ndpd}1z8-fm~Y6d
zX3bmA7V=6(-{<q4d<i2W9DWfO^gmp5uN2&qu4a5SlOcJ7do=DiA}-uB+|WXb>m6LI
zqKp5W>c5x=EaOC{fkfg6Q*TWz9uWd1eUii^ef_`j%hjVtD01ZfXwfDWNCO`mp>lUD
zy*ij-Q$|Ytf7`yR^JWH|_1)6E-!|DC<(&(@TpbKReWAa$;vd^ASB&5-4CuJ-O>0r$
zFZbUV0ldjJm!t41zeAcgtVt*_rtJ3BCjQ6a{_|D@d@%g#)@;+`bZ+uTUi_~Ci#{Rd
z0~qi6PM1~?M`7pWSQP8<{~x<<BW(m9tfJ&&gi!r=elsI2iVd3zmX1Ph2|@0opZz)e
z|7*mcBE%k~`;(z0oD27V3$&}H0PgwG8JBC~Y-f7)ivLDSiXu1{;8<XcUeRqH@;8R7
z#LI~dg9*ZA|1auM#3Fbd1r-&0AcUG%WV+(n%h{RqAbDM*TNDy+lcR6OmNOa0$fEAv
zVNsZ9DaF*4kyU$Sr5kr`<JUFP!LYo_qoNv8H41q#Squy^CYX#QB1~sr^YIrgU-E~~
z9!thlvfLSCQSyz(WlCm^Hd}O)xbg5i>{?e^@&mVx*?eR6&NIiJK+6mdi}ERyBwStg
z&qvb)Uhf$!?qet=rFhJMM#m<0pwRbVvZSy!Ck1HHKR!(lNE;2NHZIR9OI!*`epJan
z{dll~F1oR6v|BdPBVpLh24N!A{PVFjH`?X+Et}IucrSLb*UK@tC?Gfc+gX1e+ec(H
z0ul4)<E9F2{^<adky&ZZD_5i9-I^>c0{JS?*szuVe8$XdrJ0#Q%6)(5>Y4vGAhY3a
z^PcLE(O1&Q?=`xDO^k&F+?S3o<lo)?`SPE$(qz6LzdFjyvH9zVO6fB+s68&!)j)g3
z)BeXo6_uexL=Gz)5s!2PC-_0mzV$_X%5jHveV~7CcG#a!?<wD9H+kgR8aFv%+<#eH
zd=jE`2M=2jBm>IAx$ga06<Ao&BSolZ(DR!~mnlAZ78l!3?a*|E1q@f8=Z{1RLqd8N
zKkg#9z=Ril*lqtd-fCDZ9`yCq&?dtFvmEU>1iKd<c4@|O*|E0OB5m0M8R0)i;@=zx
zc4IvXP5i%KR<xG;%lS=<18u^x^q-dvgIn^SNfbau3hVUYIYF7Kn(_a2l5+-7?JbUv
zStS4UwWxe=P2fHFH?}EdDJ(}KUjH9+Y-!0m%1`l8$OGm^QO~9RcL#jzs~22-G(%x+
z7L7;S)cWWGivM5?Kwx%b@b&+}3rl#x9Ds?EFYBxSMwFtvZ^J@HN**rtRUi>}{$rks
zN>F?I*c7N`pMq1h&uz`T|GNI|6rXKW#mmfUtA$x&&4>Sm3e~yrCgl7AY=8Aa3=#jI
z48R@x?c=PXa=4EwV)XwSVEBEdU3&G4x!E}c-+vw8?p=nMn4#JaS1<W<RFs4N$(Tjq
zw=b;5!UJ>sdVG4D;XIeFe>-y3@y@-M3yFEJtR*t?@W77$?<11OOiHve)*EG6nDXb;
z8@fi8dO1!WlAm3EX^T>U56b*QU>r2(`Gpxb6@hbq{s<zsvAubi(Idb<t-e<l$dgzd
z`ZA52a<p2yc;eYU@tLwcJoM*o=qr*3@tvKWk7;wQpeoRCZB-Q&6$AqDPF3BU%TSX&
zv3q`QzXcUjYw;}_X5HVxZ%tuNP88Erz2~TRKf>_tciWQ&+2+WaLRz%LAGW=cHRepD
z2#PsPSy1*CMt#|J(5$**c=`Q@McE9%bKaT$)kwYXbDSC%H@$h}@#7AF=J=lj(q8=S
zY;M2XTCj+ZawnLtI5Xnp_xkyRGh5YK^@GeiM2X$;Cq|W(l~yAK^bA49BqStpbM42w
zOS?Ne-_ASg)lVYu$=Nv#N4FOj7qQMLG>_y|R3a%O)iqy&FR9rRi+>sAA6=F!o*iVX
zdrXV>m*`)*pT06Zk+^VMBwd5()!HQQ8^Pgl_rr~K=SF-yyfhU{Av-e*3psiD#M3Cp
zoIUELzVtVtp}0p}MXM22mi?du_0>T8a^UZi`~EfWaHyyMcF|0MQL}ejvx=u@(+tfs
zp72MFKY#MnpKANWv5IvNM3y=4`?#(Q<*DR54&}lYyj`O@-lnE@zmbb~70pOXgBXq$
zv~m(8cDK34Qv}7Nr`zr%PHS4}>gtATN2#`ZT%I4QrVH2TR#}J(4RIWu+E&P&BxbIC
zI*629nB@0F^<w+H;6D1v@dS(ZEH7O3J-p3TGx>5M)Z0kVL1WBTFllp1{u99_T)nLk
zf!Jmz)ofCdrJmCS|56TiDJ=M904*Yz$;<BgX#o93Q3l<O8lT0NSMzUNYUbi@-`ouP
z1f=UfWNw7@MbXt05{U<=&3SOPT3JUQiN%Hg(N7A{@8Ey`U3<hJssUsDhhQu{E;a#t
z`|qL#NE<Mzzl&<{BK|)tQu8l8IoQtUqZzIpf1IhL_b<tPCA>|WfRcPK{ZD-=3m<F;
zWbkQ9&=dUU{Jz&MmW6E>T^+f<X-dGuw&B|UCNTf84nT;%i9+h}GtH}|{JT~Fbcxt5
zVkk@^{HMtL$2!daU9-%kG()eJ^6%>RAH-oI{g?3n$2$H)lk4wl|Jl)h6Qutj&c7+6
zKkJD9cP;s^9jpBp5rIOyB2En0|9|f6Ki0wh@3P9i*NAjZ)BIv-=x;;>RQ~^ElV9C9
z!O(sHg<384BtOzFWj)jec)o<erl6UkxF{(*?(N&E3JMcVZjSG6!}$UO0-S!2u?YzY
z4PY0h`1$*L9BzE_K0mn<Y?YHX*Rcq68l7#yljgVX3GdB!y(K+k=!l}djgP<E<#wUR
ztS5*?2oXg7{1WCFC>>1Lo6Pfok+DTv9<`v7Cdl<_713VYy5a=>J7yc&)!X}Nw@-w!
z>@GOAadEA#`+L8~U(U5h_=?NEU`<3|V7$JpKw@BEykdo5U_8EukBxz$)`7yrz!=fB
z1L$GkuX4&|T>$VHICQQxLXh}1ubB8ZV54{l7zZr31QNSJESP9n0|5^FpJ)Cs8f_UV
zIUxqdPfLA?8yFa?I)Y#zVKVT|yU4)n7#IbnXfWpAzwWz0;d9(LA-FI#TfbbYjMB1I
z#eXqTV)!cI10Gb+W2Zz0ar<Ir+_0m&r#tKCNG<86Pn}h_Z%N4eUK43s29&GUgUeu!
z1g|>J>4UipcQz_{gEORD`)|{Ysqmicf%0eG?p{!z@ykBLauZXBtG^xXd}|k5y>WU5
z*_B3Z4&8&}!(^f-zf9OKBGBi3uPX<9Oc^~|h{aBx95)lc__0Xvo8RL2h4^-fRJaCh
zIL*;eEh^J{#kQ%Z2_jnP^Xb#@*7)bHPKg@3r<ajA65<QSO`kQkrb}`SZ~J(%JXrcN
zE9XMF_f#>9({Q$XmaEF=jYnb{j|@fG?RpIfkNpo35A=oFcJOZzZKqwhDko-DeHMRZ
zGF>IGm)1nz*eY_puN<$xnVg`+ajt33u4OAtkE=*BKk<O<=81CrXy7NM3~C>ZZXEf!
z;AR$<&)vUWlih?!Hd1CjZ%=4W1{qUiRJRF?t9UMiq;8)meqHQ%>hxK#V2bZhz)HYY
zo?hb9=fx8%tHc!JEPpZ~F)SaUqoHF>)6)8dPd%B_+SP}7D*GExoySZmHi+@D<ZkP1
zy^J~pa}w4T#B;Wto@OZYIj1Ai>-~IPnC5!}+c3!^BT%_ISTOgxf;gDB!l>cTgxqcz
zc+4>#I+f8gn7k>`<~WnOaNB%q*0k^}ihBP$^<ho<n-KNr%$G8w;RL_($MeHWZVhKl
z2%T-`M!#|4G|b#igNadTJ#v}*QTAEbXC+~dTt+6m<Ylw?i2dW{6#kfZCjv}0Q=iP`
zeaA(vd$|K(B*sH!cf<4suC`g^e$oWe@@de6FTq76kcfs^_tULh{u*{o$+EduJi89a
zYHA)AhvujJME#R&RTt$&==G1)1SU<N?~c41{#hDfBXTJUEp$A^_HljIcT`6Q*r^j)
z>b;sgV(*hoayk5_k)mv~F6#XSa&I`0mexM~$N5-6-?@e7+XorpJblqg6IA>}G4GZq
zmd|)rgzR3rV0mX%^`L-hpnp+B*}{3WeZX8cXTG=)=kaWE7&acIlgjzB>#glQz}ek2
z0!I~=`4T*WK4`~H8Yld@ebTwTosN7<(X%Cc`-9BsURSRcA_=?ggnOwoK5i#(cC|Qs
z=niY@Gc7`mL)I(RsZ3l&XEq50yICQ%s}&V!$`x{#Wl>9a@AQf1-JXZ5O_Ae$J>6|y
zHf;ek0%wClCS9dZ$_TC3o;7_oHKUzKON)J|Zl8Win7VgT6)KWgDrCoLNc-^YB;Ha$
zrj^FcvVGH%U9}#0wCQMQSTWn;>eWs*{N5$M-I##^{^7}z(am&GbX=1A;8J>@sI*M$
zoU#Np6*0B(p|j=B*?g1qOVnto>Kf8@?!A+=jJ}cG#ALSo_*36ZpLq1EtMe7Eu_?AU
z;j3QUtj}Aax7PO2hTAZ+>)52dvoO<w<WkJ@z$rz9yQtw1#ywMP*}VAv!Ts4O(Mq`p
zDn6+-liME*#ah3rmi5=&Zsw|M@h`5DmC0QmaCtcTw6@4HYHccKq3>3zqfD#9Z1wiD
z;0)YRo0$!Q<w;qYRtpN>P2me;vzKXUf+t2^4{qH$H9@boJC64^*RxCfTv(bpg?}`*
zT-><S$b1%ax4pZkEG1@pqj&sbMr^T<aP-Y)xd7y7IrsA*d;4$V_GH!e1J~tenf4#6
zS*L2|KgC^a+NBLQzuy}EZSAHc$Ls3S8KY^sI&swf)l#%WqON#6ZdgE!vsztl=d{?5
z<{WH~*4Orxg=Uy%yO4=gW}52k%#8JR1Em<Ft>GrY-Z5J>G9+Ya?enWlTS2vny8d$d
zb<;KTOU$?)h1sU9mMP5J(<Afw-DX!}v@8;R=72uie=wc6Q7%wtj5-&~u?-*Bp6Z*<
z{BX8(v9|i8fuNyA`|ZFl56ZHar0WI;-!{mirlzxwP9y*ay0mBPTMMkiZrcvJO#D0;
zB+u7Ms&(uw>h18;n-*j()QtNOCvyGU(<_yNgY&y72*~{#1b<U7|A*fB|M>M1&K@Sl
z4a33G@y(1gSi{oJsh<_$@$^Kvz#-x3bH&4Fz1aPRJg<5-?$Wk(PMlm=W~6Q?g;%_w
zx%QeLC@7Pc?z_W0!UmkJHH#}mzumo$zLkFmc)8Qo>w1kOV`HBQR|`B05)+uk!yeeS
z7|pO)`KX$N8yrJ05MbhW^bq>_H1-S*8%W#JkG+jd_s42N6*iXMe7t9~L1gc<{VVB%
zy;qdi)>ldSsw|BQ<nPJj{^ot#K0a7fziYL2_!B0V*`fL}pR=;+fiY=XcX$0nVg5tk
zKw`3+w?mIt&*!V7Ed)e47ruVpUt{=XZ|CK4;Mae9zJn>;b+0StLyIiIlCM$}x3hc)
zk>#wLsgk$Tsa|H<#p9(d4#Ttc>^T37ZBweIMr3E9Th;7$BnwB;I$M+qWxCoMC7Q?P
zC~HBCxLnTlQ?^qcQHG4YO$j%9Ifr0N!{o80y%I@V`4Sr^(5Q_Waxs@X?m`xNy48gC
zw@6o}IGoQlOnx~dd^Q@#z@=Yk^oeVwkga#q<JZ@Vz|Co)`Mi5K8J0diQKkBPBsGg0
zHA~$5^7}0pcHzb2$(*(0y~8Nh@>vnbKGS;#Vwb79Ki){)rOoux-mMc^Y&!Y<?&D_T
zOY@_x2l-D{<AhD;cT=y4K`^v~A3-om19Cb{t2z#CpEX{TmhyxJo!Q%LHVmDVM*1AD
z#^o!J^`F`wrCw6Z^2p@AF?C7pqFwiz7b4P|h!`dp<zsEXFJ_(Z47nJ4|InxM({6^x
zR9JluvEM@b0eP*#;rB{fIn~nLOl&UE@fm^N^?E)F(Y2<;OQoWV34KKB=&zC41u68F
zkq~@qzWSFYhs&+U)(!7R38q*u;;`|tUq2#36gxjj4H5LaH@e!Y-$cpE07ts|6gl?u
z2&@;~XNdJVTkHCqFM5~lpnd%Zsb$npeq7t##e0371Q&y(vt_T!Lqgap*UP9f1s$pt
z*!2Wk2U*4Y=-Z4=m&T3!8<o>CTsuD5M^HKcmUuq@^8Za<a`As%<`1gCG%T9T&&m7R
z;Ng1jcYA->3C-4B;%55O&t_fDbVVg4x4LbsHB@v=mPE~%`1q8m${n?&Cp$hl9dfGw
z=+v`d!6d8vg6?Xuw=$O5_pg0jEq;FA+h<$8yuB>4Uz1_qF<|-#@hNax-|~00UHP>Y
z!OQ(Zu1yM*RbO|?(B(++qs#nez$T5J%%9uU@!MZrHPu?=y}s<W{ht>e3p*z6{d%}Q
zrR3esXI%FRvnAvXpNK5XpM3E1`71luy^rnpmz4beUjE8{eFnRuqChVQU-@AE|MR0`
zw|{SszO^p!+dtutm;cp8=gwZzwb${l^5v>;Glb{<{5!GDUw6JlKL4iP`nNlKm4C<V
zf3&M(@$Yl$e>0zM(3p58X#Xx<Bdb@(clz7i{q?l^y@l<n=;KeXA5VK@c`4x(mwjzH
zLk+N$cUy3t;ebxprEfLGYP-C5R~=MRdS_g{eeUY}y8E_Se9yHMzWjIB426$N-v7QO
zd3$f^7GI#+yz^`Bo9^4@>%z6U@11>#-k!|YX{AmsOEP}H7n19X+4OL`Vtb$1mEQb7
z^Eb;+uRPYe+5XM{k2NLT)2_Q%cvXoQ&e4p%oV;s$-?mP5?b74>-c-G{Hd!;RROjh~
zmy4o&qs{v{c`vjuOc4f_nXKaCefrC{Jo>zgp#yj}$O?|Gw^pCF1=!8zKH$Ft+*}9t
kjd8cGaWuM#5dJ5>qyF2ZcJ(ctzy>>mr>mdKI;Vst08b5`&j0`b

literal 53086
zcmdS>by!qu`vwe;2r_^oAw6^t9m3Eer64^tf;31C-JpathzNob(jg%&-3`($F?4t5
zyKwL4xA(L6^Bv#6?>mPh9>Zd-`@Z75&g;A{3|3N*d4NfZ2?Bv0Jbxys0s^7*gFtsq
z!6?8dn>6wdfG-$`XIc&*5EkC;`_ACB_az8K1$r(i4tGu1Omy=kJBq&%`@l9EyCYP-
z09&NOphH#JM$kXOkf|z7fv^P($`c(|eE%L$-p@Au65P)w^HBC8wO>V38`Ga<>3)De
zgzkNG3x>k{X&qbVKDEx^<u$PdZ@Y_h+Ja^G=)pB(2y0{;ZEw%Yg=bPT4HyONO9lLK
zHB8(jzx@Z;@La;z@Atp9-_QO2>MKMC`+e0MHU{?ZtAG5z{u0C7V&hIctWX`BN)vPQ
z4#Og2Q&WXwQIsbTm(0}Er{_El4i1U}OCN8pcW#c|YuoCV5U}&}^QO8mDsVx%!Z9`u
zDlWal%1;7HUVi?Qlao)evFhsTDA@%C<=Xk7L>vloa$knY@8g^#<`fq8_4aP<?97%~
zjupJD#)iOlo3J>DYpyR3`etiAPJSi|ePVw}TIfXurcq(_rFwBJ3fvrj_UxxilGpKs
z$9CmPBRXZmrWaC9N=nMiY~^C7-Uk^#gyOoA=)zuqX<s;2?QyaltswY(ry`bB>nY$K
z2J&J!*pQUQg)0IgU|6KNzP@f#e|3C$c^S`RCc4*xuTx?3vdYoI#%3FhNEdN=YPVPM
z`Sa(3s`cCtA3hWp7n9l646HeZ^w~}no1E=0Iak+~mL7Jo6kg7Ua@^OO9T_nx(v+2z
zEw`C=U(d;^yWH=F%Fz$HaLobVuiNipK@)e~YbCl<b28&$mSFd8+J*xP)h;q@-JY(-
zg_D6=5(hj;#BqM%f_>xWf4MG%gMeL9tdUeqJ)okcmB8_DYm?qH?(6E({{HIiDU6NH
zdiz%;@B-ECh1-IC=j<!Hy1T9Urq)Nk^@(0sfj;TE=k%>?tgRJIZK;2AI$9rDN%j$a
z)78}#6@^<n=Hl#(($U$OQ&4bkx1#<^zK$oVE_9uPje`Sy_xfT-u1|QsgBCQJr!_P@
z{4qY>qGW=-YBhC8QH=b_%a@5%VB7jDXKQP=H#cX=Hx?Eazt>(v-6)QuO2OmXY*+*l
zafjoe?(z9t9gSi6?#_7Z+==72O`-zVpSE&<E;s70zx@c{=@Z6p0PMOK>2rDaiI>+^
zkx@IU_-WzIc_9!PeEj?ywR(>r^Vg1!j$x6J-7XA+F3MLK+KqdMLyC7{D+vyK){~CC
zkH>_M$Blfc4o^;|toT58@80#LV%4i5ZUDjvO`Oveu&OS#TmaPvU%=;8Re~TM(;g6&
zG;Rx(zh+&L%Z?tX{&I<<mGnKEPC4il@K!9Fj=GA9{bEN{RaF(fqHs}R;Z8yIc5Xqz
zs7LLo^V-@P;6qzCH@DZhG&~}ME*KXs5ANRwW#{DyUu>4_{8)%EYL7?+qO`=cmz0tc
zeeU24UdVsO?MpR}t(Jo$sq(4}s!qhr%$!RU)HOG%V=Dq|C7p5QXX2Om_=*1h-9b5y
z+UxTTGJYG+&7$@s&;8CsA(!=9J<s=FzI;hc%%2W_Hfw@x3TX2giy#*Q4Udd??KYyb
zXcXk-=AJh`^xi#Ph%D4BN-Zo@yl{`^pas)xEBONm1;QB;DlRD*%hS4DS=Lac!-_&^
z>;CbStr+0#va&MU+9P#a+no>s7Ga9Mc1oYTlk?NM(7jPUn_s8v-wL%^V=W6Rm&5sI
zb}z2@XIvMoE2bI1zUtCJZxg-)kMSjn-dY1PmBrK^X8JRM`Fi!UWx<`@&5}M$-~Rr7
z(5vIk312E$;?EBs@V<?>Pdkmhv}+{*t#52h+nl_6Kn7Y|T+Gv}J>Ss3sfx3!w4D{L
z9SaqJw6?bPBnBV8g!|;gHFKH$Oi)!6aRNefUTX8p>}4huI8yYS2~<7py_WIG)Wjrs
zFe^V_9><@hum-O|qyBsZKoruNTa^042;Zl+k^z&p66x0%G{4)Ls_@y+nLP(gR#H;J
zQF}PJlH_4CQ{x6$|MFt?rp|mI^=M;^L-&W$&e5J1(Xuo)rDzR+a|Hzj(TbvHK(gEn
zC10by+L|mI)APuu0?W(G*I#Yfww6QYBe#KI$dLQweYw|eZEbz;)<!JDr=fooi{Hxu
zs>A<ht42<A?cJz>si_NOAf>{w9uBHFP780#>oPsSt9)O=+uO3@=Ei^6Em#)TmV2J<
zch2D6R!czuT=&}A+5rF97rW3=kC$2u&9{bbY;3q$d4^zMgUk}0Mo;QaYO{T&cTOHY
zc(Am*+-CO)j8Q+k<6OPvd||xPaVUx7-`FTFSd)N*+PGxDw6r9D9PvUYLBR}#Vg_(&
z{rUPgz#Ch}JI|i>4-PijE<FF#j667d7Z}Ls!I8=4OVy8D3Zw!{NJvE25)A<;D>6~d
znV8c2Wsxp%?)!w5iHXVWH)212{J6gltUS~rbmVn@c6QumZCtq~Yiw?AF7V;neexMP
z#x>Ny*&Rv<B;odZr&r#ex*tz?a9QITUEkh<P#h;gVa0f-)$tv+?T2VbUq4oYg7+Cs
z1jTSy2`-<4ex$tp2L;3*pZs_li>c9}f_(7DebKd`x6s96oP!8pC7AiwrM67VeFTCv
z&lLtSu#4gU`C&{gXw9EHq58iczI`N_V2%6cM2T7F2YPoguj{iF+aJhx;}x_(^;0aZ
zsF3;YV@I=w`qx8u-x*%(rZfw!xz^4+6g{157rWj|XRV3;_iFWcgn^qlIXO93Ujg7O
z=D7O)G<|BT4sb(gNXSpvLKsKg=~K`Bv`A6y8dp0hsn*YrEzd75daW)qzt~h=9Z$)~
z$hd9QQa@*e0ow<5@aMjF0i$U~P>Ok<OuN+YwULR4h{(mVx*ZP4w62U5>icY!jX66z
z>$$IGSdQl5qNV^)o%^lee5BBOWO7nlRn=i3od5I0b@KILme<8G0TGeICD~!Td-If`
zWGUR-m5fnRE+8}1?rAA}b$7pc%b9V`vB;wNkx0!KKBSpgs0G%@$Z0d>i`wGU)cMoB
zg^Y}htE(%%FZo$nGM?T56VShy90Ogh%80oGNDVNH{k?X|Wami>vt%#6H&eFtwV9a=
zdt-zD+Lmbu#f-zxMvoo&eh;nu(LhS?(?XvsOQ5JIC`33`<mTrSP^M>Qjt&nyIyse<
zmTpd#Sq`L1%^xr(xzg}REw5!I>s8p0zd@&Pode+V-2W1{3W(gVdxyFyf4ys-_v{^F
zv;yVi5J=JO$+B<R*{hW+NmJhq{7yX`97;|bRt^B%?~UV1%geI@NJ9O^^bQDcx1!ip
zX=-X}b+t}e;}_eyv*j<gNgZO1Z_>MwU!epAN->xk*|GuraOp-M{tVVOHoE)zDyFs+
zvJPB;Ta1s#on9VqP0hM4JgA?&*=>F}2E0Z`^_(Y63Tuxi%dMvX0NAd-x%Swpb+JGB
zb^l)my7jxw59GaXBib)t*1jIe-SN4(&?&caI~virqYK<e?B=XbPEKytZ6%RN&-Km|
zC#NGI@VtiP`D0>Ytf$Ji;?0ucH@zM{)PhP1|1F8$XQZ}wGA7qpj<JIPbMJKGNf*=}
zYwtAz)hO#KU+tTEzIItiCApEHprGO8wE6QY@V_$RH%j~~w|=ML>&sSxzrBH!d;W6~
z9Feb6q0n{V%S!O`AF0nEgUHjZv{#uS0+RgQ-=HB-BN8KyP)G~kaOs8r+z%i{cw&A`
z+w8q<v?UkyJf(jC6sSi4;-VIgZP-_)y3RQYzGT-Ju2*>2M4^{8B(PS^^^y`mIWJdo
z?Dq7XHeT=dae*$AG7oB_UIdj*SX~|))1bEBeJH$}Z|h~bHkgs@eQu%p`=zcPd)hEp
z0dRmf?*Km~B~ggufKEN@m?!`m7Rq(A>YYeIEVk=j>*!i3io+2RT}L)_^p^M_`EWX}
z;33|@?)jBB9iCk>p-?5hApsb_^|k;YkWKuvo~k(h2?+_SeD)Vx6`_9nw@ng_GY}5W
zKz=K3i*C-;dhE@&`cm;eer#fFyg5~23y?pcwBB6y-LQh2l5Z}}Kpgd##p7_tMqf1V
z`ud6relanhp-g$b!V`d`RRO$vX=*@4Pj9BbAIDj7G+$?Ie4KbpjJ>7h@paXF2n)dO
zjuUr0R{$0h_nm*_X?XpQ+c%BU0!I=UsLed8WqV^YwXVp3xN;@9|IO(`9*%U2BD=Jj
z8qel;59-f)ZEu@`U{8GkQBuV|{I_jG2a-VECB&y53}s=e9_{{22qs?1Gc}3#H29U(
z%J2w(dh(ZH(J%Ln{E8X(4b=KVR5bw4a{;_ogX*S031(i5G+R~}tqlgdefyypD$kU?
zM}(^g{H1{kL=yg6QXjk9Y%tf}ENzkruXp`xslk*VlLq^d54sPN2rY3@4T?{wQEn**
z@FLRpv@3S;2be}2w$*Ii)rf}v%L<$6+k&{;FY)lD*U`w)(UCA;k|XZ_ztS<I)RWBO
z=o+1njt_wYg)0{T=s$MgDiO%v`CPD3{Ew>jCv5ZodF>xn>v!P^Idcg3XZ5=+SGKFW
z3<Ccv_y2zsN(T4Xu5^k2ej0|yZG;b`$pDod52ZlvGq6nFq$HNVz4?O$lIk)urE&ZN
zDcsEg(xDE7_2A&3;ja#${kmmq?Iu2@|BatD@wtFM!NG2>EKq0?(9;3+9@alODQMId
zM$ErIz2nW#F4Ro7X8*5(#i3!SK~{5XmOuRae`pD$%ZJlIexStu6#z9U+?W6dSdyJD
z%=ll06)@k2e~|;d0=jFY{NL?flM*MUvASzd|G$dOJ8TYsR6PG53&p$HWk{U+3(5X;
zHLy&}HNcF2dGI$|3H$^w5xW2JX|Rv$f35vbUZVT_vdr;+P?K)pZg~HSy6%`3v6}yi
z1^!0|E12ZBo7EQlJ1_Z@3@riihVd6R|L0TLdCHmpT07bA=;P49^I=Q;JN^1I0PO=<
zWBmUjW&in<e(#~}f302qkjp3VZecyem{<He0L5ck^yL`;ZoGmP;~dX(X797^AKO29
z{eNTmfBDX^7>*5TsPnR$PPBM7-+r0J`!B<#MxIf^-ZVXRbZK{T$s{w*=?K1<r2dyv
zxwN4A-oYc^ihO)@Pt2>tV?$c{aefN2{G6hbqeuSb&)X%PUcr_gZBCPoiF~%9^X~C!
z8vlN0Ys+rN^q+S6!en&4wUwRSkSg`D&EUQk6tA$xw?-KJRJbDkpYag0$`<2?&gCPz
zzF~K0UgPpw8(2ffIPv6E(W^W2)t*%Uy$ng)97pHWVmxMMJG~+WAw02VF|Natv{aPJ
zO1qzA@|VqD{_Yr8e3PEpCEsa^o!^$G;pX+<v|3!EuEYGdB?fe$ALhPmBTw303N!JI
zbAs+abif5d^zUH`ESBIoVsWJz8MLD(&-nZ4h!DiTawvF2g&F!(U*FTkLJ-&vSPR;T
z@_$SeL603PYe0Q|u~4lsgi#EaeE05sj8v>F-jn}>BM1P&2u^jm02;`<TT2@PR?zy(
z84)#yAC_i3(!$OjVt<(9ysHJJ`7jr>vwyp(+g1wy!Cwzs?DLTx=JF{7WCD<qX3+BT
zkL}S?fIdVa4IU9hWqd&Wt6$)#ssc4Eg*;}G$HkWTy|*A-?CM0;m^JMGyWg&1^F3WR
zdc(-G=s8XxGI152rv=x?D*P9Om|II(vxS7@a&sF_0zw7x)7X%pN5~-Cnh*b#Ag%+`
z;e#r}Bb8~yCxW8bU)XkkWCx)J-U+5WNG5JkNMmN+J)5tb_-8T`O;0ni;J<#|hD{l9
z>hh9(aS4qgE2D8Cc&9*~u|owHmlk5j_|HHm5)om77WaHsBbEXMGgQO_1PNCG=%D_d
zu&&4h>c?tv0uGz~DM3DCOHXas1HC@~;ehw3fmAIHCI3t_{nxKA<>hg2Vxdjqca4sy
z<)n<QSl=H<#|IA#{IKTx=PI!DI0gqnn)VJnH8Vjz(`#^W#4p``Yc=#@IJ`vzK<B?_
zIZ;)$vI@!*P>nJe+ij2u=f+uKD5xd!Oz}RU+?=?yv@n{?KNAnFh=C9^7pj2h0Bo>R
zz!^>j=$h_=TF@cVzq-1o8lRY8tb}sz?(Qx11$YoU%Ff5XFZ87SN4N(hQ79t-NiaMW
zKRUN9ij~0YHYyr9QQL)fBx}q5=@_g=FgnOqo0R4XB{GbPhv!hg7=B1ChgOV7V2;*4
zQ>$5qsqC11_ercQ%AeioZJh@%BY-wIY;Z7W0Y5Dbvk~_sN;kSaQ{HZdN(%};eoUl{
z6(K{>DDe>KQXgv5{YOF)bMX<Tr9Jy2VSJn4`l5h*slrT*B_YXinYa-t)W2Mf5&*E_
zy1{s2)5wh4K1U@eWY<nUUW~S|pll`ZA4m>6UE<hVO)Ii`^9Nx@%fdf9o0oWws5v6^
zJ`r-*?&<(9@|b3Y&W%?*Bqz6<VF1?jGm}Ue9MK$DgrH_6WYrBaYEK=v2db_rqX#+z
zVCXQ{4YtL_mf|~q*qjK0T@>}M)vz-s@6l2dOJT#p+rGdZAF*GBibLS`BoSh=(1WkU
z#qd`+NOP+e3`WosFA-8QE;EI69R@ji|7N6;2oLif<xFf>m6NjB<qN)-FTYO>j1kHC
z`26_*X_;1>04i`A-xnq#ZJoV*+mw=_mcqy|fAzdLQD37In*?Xls;c?Q-rMK_H8-E*
zVk;XDk1cdWF_kBr2D$+Bz0T``kX2JjQ&Ur4|7Oy%Fj&!DIO*~gu>%eAb<Tb4`9^d=
zb57vviyzZ(&sB@#f2iD3i20`%d!K24W+D=%7#@zRo{orkI#X*@MHfyX=<m=MUIK6e
zG-@v_b6QA=+1-ZC33ID3!|q_)9H14^tajcq`Sg;HlXDkn^$TBC9|JmA{2Trezi%1)
zfnu-D?0NYeCWXz8w_A%%U<WU(Ch(5<fTmv7IJWB)wq5&aW%~s`5Xj=z`L_Z6F(y6~
z0Y=Fy##$>4LY#Wypu*K!wSc5TX;{F(h&Xg~6>bEM6w<0lgG8GEq6ylZt-oOuJ9`dL
ze4zI`8CF%Cw(n+JU{FEmd!H>4km|WD#e6HM>RLGnkBH!V{8;LwVa>5^f8*Z68BbM!
z92a?A=-SQhqYC2)3JRm4ZK<g>nI%irCIX_Z03<^4)FE@@?m^(#o*(7hgqeuPiNd)k
z<Efu=q|Etom8wdOmshp(W1>f35|XsU#5AOj7yOL0-e>qReIaNcU0y!v@-pQEo{!~*
zE2Ki0lk+{5&m5g7G0C0|>)W)?AGk+nL~%}O9z48wGcYiaStr`K^f|Bj_CUiPeiy|@
zL2xbIW!6jMd<T$6Y-*22fs2@CfY#4ezr<9YRfva=Z**dUq?_=*W}WWp8t(O4GR!)O
zXC;yG)bpur;dghZ;VGakBL_nM7xUAIz7<PbTmB=JKndXT(+&Lq846^Ls%k<J7cw=p
zB@pPIE_(R`C@)`)u_I@%<#AAvzZ~SP49@m<`)(UEdHJWdwxt&WMGzQO-b)tG)0Exl
zWB^U`^Yiz9{Tg1pxmI-D?`&9jDE7$vTovd<4PHI@F(1+&`cq7Q*PV9S9%#=ut6dIy
z`T6<y+Um#|@NtovKo0^OQ2=Vt{Pndr=$3lBY>wNFYMIG5=vX3OW~k_f)roq4SqYwf
z_>CYSV09<b9biESdK5X69lZoEKK4qwScT5#Z-<zB-ODst>5q;Of;!qLMb@g%(!`qg
zdU(daQ}2r;-2Y{>xcD_Mj5X$TVb9;*`>6mVJUqMbVQqF5DH~gqjldpTOfXX@Aot3V
zk&^P+&ah%jclGtn*0>$qHmiWXi{a`*;9c)yZqCHd|7ljIMQJe#q{+&qo&z-D9uTZG
zFo`eKeQfN{f;Aq;bBobYUqU~AY?$_YMS~CVrTW1ZtNZC&CCTH*p!Z+M`uxUzg?;>Z
zY8Aro=ckJrIJDM-MX{CMy71QV=J_Xkn4qAN7KyvEHCqH_drv&BHGp$Fs4=ZS<G2<8
z0lgDjopQ>m{99q2`*f8P{=v4r=dWCNTl-z|eI6bz?xqEDSG+8H4}(dyoUq7+EV0DR
zL@-TUp~^%pStt>3ta9~H-Y5wNg5A-zbdQbjK6ZKeQ(czNy<}=pA5cbdevGQ~K9;IZ
z#_#UOvA0)Jdh_<Jt4(Y!9ONqeOb$yJdzy7}iu$2mY;0<^&>)*?jBHb5pUT>fi~%)h
zc=$yli>DbQ9SfZp<>yRhhIsGTE&%Y)Ru;tIG=enW;pyP5FcRe!(`ax;hKgGg1~Dl>
zC73>)Ityujx<(?I50r!K633eo?{}!YP0FCDN4d+yEUu`SuA|^7M-MRm`Zj%49fYHk
z_@Kekk}})I8;h~i?zI+=9b_**pf^U>vVeN{ulLVhmEC*&IE~vc8mwidh=ZEdjaIKO
zNY;cEv1gb5I?N&w<0vNw!0TpqMUV}@>-f*$+4$EeudpHt*aPN*piNPKG-YCPQg{#^
z)30SQ?LxchX|#{}L&Xs<c;uG)sabX2RXQM@+ds&X@%Ly$Rl1+Pp99N$FRToj+IrVY
zGm(Zw3(PI71hX`8aRTgJq=X*%o?ByZaPFwhuEaPeE)?-GRvH%|=Jd!pvz)^tG*lU7
zTz3z*n!Ac{U)sqW`Jo;a0Z$m`uerRE-L<UfTK;9W7+RU9Bs+AlFgw9k<suL(VJLK&
zGGnfzcjWjQ4_}#uWnhM43R@9uILP!SkJSNxvj;%OLsFmTCl7Ip;hE@{tORywcfV#c
zQL8scQCGmBP=IAcWilsv3zl1d*{tQV%_%KK&1D6_)l%}qH}vGK6BIgy0X==qv1^F1
zNux3{qL>!)Ir8}n$wUE%*Nfq@xDlIHEnLu)h%*mim@B7^PBKkLv@}6}@6yqTGMi2(
z&|;oyL2GN7peS!rQ<=v@Bq;i6<>-Tq<pKM3GR><SjjBX-38uVGdSQ;JTvUVFROsM5
zJ@HFOKvd|ASh?fIX3%&ul$EJUBZ9bD>u|BlGWOT9ew;1;u=Wa}ibSktE=e!d0_+#D
zTLOnfJTd1hyZ!<D&n!Uss8+P3i;M0D>8$vS!5NC|aH0|rV{LPYmRv?9Nm-elk4quZ
zCw+>Od<DK@2=;&qNkuxr3Ysa$;xc!)ym4GPTw6z3Ens~;Cu(lV8C^PiWsxdk0Yxpg
zYjykH?6w!oTiC{yli<A!45dRh!|*LYLXA*z;q!VmqPQJXWHFy)rnk*LYe9W+Fdcrz
z`*GAK4By(=(UA$Sf1d8`$yRe)cywXad|2_`r=hh&O`h(}x6*<@;8@5>^?djYjh;^5
z5LZ_hMhk>4&sML^QF32Pop$Rl9N$yD<2GmxQfte8_WE`DS=B9<!dJ@RR>K=`Qa%ld
z;sUsS62-@l{@2wA>bg_I{o{LY+~lM-RXA|7OpcDXzH4T-p@^V-43z^E+9K1~rF(VF
z$SbB=aG-QnP=0C4Xhv|PagMQByDB1^l>qswxo~jDvF_03*UT&ED|LkAc5gIb@)u7Q
z3XM$xQo6755OlDdl45ZCoab__8xMvZWk1HYWk?*ATmM+k?R<F95?{o{+`y&{rH##-
zw;Ia#J+g=K+J&W18=KKGGU|5FaS@YYc?K7k2&m_{l5dxZF>hIXq3L}-?=Kuk3sOZ$
z>XFFB{ze*u)lp{uwL9>(7mRXJPPlPGM1k1yGczDgi=3cN6#My)ZJvu@_FBhD3~Z9r
z3~p9qLK7((SaI>NVlfS*xw)f<H!9%sD?};$-Mf2j4x@R&SzqJvIPGDJ%#L{U^ijvx
zG=HRIM`urhidoc0Y6gQcX)`Q&M!Ae?vW$$Bn~<T83`DB2AvKax>EAh>8>s=2AL+lw
z3IhRkpn{`>lV=&QBO-`z3|<7MgK0w_JdjaYAtW?pHGwK5ha~OUh0m>SV10yj1=mb4
z-3k^vdxtctc?9hZZ-y_Jo|G=YGov$7yO7WPhatdooXg#X>mu^=vvZyY^=Nqq#ddw=
z_(nBdYED6Jf`y$}X6o)fe5M>N;fWUbO9huSvec~i9%q~ypC@a9gvF8Dq4MQ%Mex|G
zq_y@yTa>hSx1CaQ1`Q*HTo;Dqg%byBSi#_}6LZ1nsT4~DVlT9y*47~cV?mJ_gR_yT
zshZ2@^mM*deFX$;1Y?@gdfMiViLvkY?5@~2*<$kYEJ}<cZLtG|muG#l8Ou^p-xU7H
z^YCeLkVrMUa<`iF&(~pGo_-OPgsrc^5fVBg7F6>q>UIgh4r>tdS74dajM|}2m9;gX
zJ9w_Rz-f}x@ue-KC5=^my#|tHmE7{gh@ooQrd>03n(zab!Ju%w8k7+0@*1|ij7qC0
z6ahT@mx>)^t1np%+@P%|*w1PA^g4_ul{mC5D(pVE)+-Dwvq)qB$V(BctTEq}mAGhM
z7=}o};(Tdyb%uFM;P7t|&xs(JtOWUULHho8_acfEx&8bGvo%rK2m%>-+DY+x4T{<0
z1RzCrD~YFLK8Zcw>uZ=FipiPeSyEM3;XGOH;XWA%eh)I#Nz86bwWFiGdyQ+2l;!>1
zl+t3#Ewzia^~8@?^sPQN?ETX6C=_vc@(oR<)N+0e5Odb)@4j&FN@teK_$EglAKWf4
z5d!EE8%vBY#C6%v9ytS`v~M<9lSvEdY11c#hbxC!Sht?S^xQi)Hx{tP8mS}CbncxG
zbz$9E9Q_ng_CnHM;ze&*&8Z&;35g(*6M+3Y%9_wG{0rq%O}rNr6zbpzW<uoqs$x0+
zZX_{rgB%AJsR0n&Qr(i^h;~YT;t|U(6%w)q<zzM~*#3T|GHLtVS6u`MQ9Yfe-g^3;
zvsb!T3id8gBA&v4?Ah@V(ie-tVZM|XO1SqE`CbRg{PX%S<S7X4pv(VfFb>49WK1#)
z!HFiLr<XpM1UQ~nP>$Z}?O^|V8Ye5j4K#ho)2Z3r<u9jeiX(lBWO4zObW7Jo6v9D)
zlx?9L>sP-vSqZR#E_g^5Va`gR7(%ne^J3*bfK<gYIl?HU<pk}gFj+Z92Vq~~Cfr#F
zc4fu~?!_+lJ~XAOY}PU#8bj>Vxgdrn<jKNV{qo08(TRN9C6mNr-?eGIWNtObfDHRi
zde{-X<{+YRqt&4=Y)(lFs><AyLI6Rs#)x<zRuaRc#!HLfrGb)k`=L+hE?0c!M>O`9
zzqpG%w9Ps7eD>Txn9|ya0E4-SD<2iTTV+p^z?!;A28b|;2`jj_7w1rjk^Ksnd|1Z(
z0X4@g3RQk?%XWNFoZomC$LG)kfV$*lA==4>k8>Ml`D5{uDMoJLoLX+FZ0x%^uAT>Q
zz!2WJk-?NuG#57+E4zpr@mppyiWs|H6+M&&ASf9DQ`OE+1{2(m9(}EhKReis4as#2
zk`xyY<2HCj#n~#7!WyH01udV|kcNRAT^?pE8t1acaN=TTW~4rX>~BbnZ!Z&K?H@3z
zB0A)}tofRA=fF`qP(9n-@vO|u(}fd?^B|%FbJBI;gZp~kB!LgxDUgR3BTjl}b)t$9
znHw7LTgk(pHAaTAHPYKZ7rO8kKL2L_8Cy(EO%4}6H>l)c;B^i?c=zm830G!~1zSuC
zb!l0vgBjzyZJXopFGi$+-2^{Cg_>{Wzr=B105#Hy>m4^0c1)4C50&Ew*z+`*2;i_e
zf^9B6>{eyi+?3TxnpEMc3fqw9Rk|nLqXSi7+;ol=4O_4lgFv9RD1f9TzI*mm^}wn;
zSTiNqm)7v!z5cO4e1(Ry6{CY?;uG;N1|}!R8<-K`B?<t9Wg1j0%%8~1ztRE*lz<|8
zs5d)f18qtvhO@@zaJ72u=>4)ow~Wc0h2l0u(Z)T-0=<4c-0hX0v44HxI`{T;YfmzT
z2sy_D!zOP^5CAFZM;f*Nu@h%*U?)io`tzF5sq)5+@U4aSk`$mPIR3G*jfa=dSP4WL
zQ;Oi}Xu)<J=@mq+t$F;l^O%<ll=pmOAzT`6_zzOzZsz8w>@zG)wW|V*<~W0R$(f5A
zcdDI{c_pB^Uk-bF?OfY_#}|f_J7k@MxDebVtJ?^=Be#_CY4@lQJxTD&(epTdW<v<*
zDmoLwzdkX^d4Vb~&-F8LbEjSGhNs-t?5gMWW2G0r`<6RjtF%S*BrO3hNDP^7k?Q*#
z$AcdD@W8FRIgrbR9T+Ym0Y`AghLmFjb~rwQOjRsYZ5UoSTjav4s-$@^AH~YXK@_d`
zN`t(y1MNB_A3@d$y3dbRw_0T=Nv_!T{Pg{4jdKo9gj7T69H{B}ztu)n6R;d%N90-j
zSf#wHS;c#78Ly65j5vB=whqSjvi!mvcY9QU65HjmW3bfKpvq%PjC$1yM#o_=ObJ1}
zae2Q`cd;98JC~DXIMHU7Y%l^%&yx8iY=7g;s*MUND_b|do1ZZ_=-bv79ERSGUYkB}
zc3`(lm0Th0$V3Hu{rb=z1{A#)jksSjXbRU~KpKl&IBaUT%5~mwHyIRZ4i3t4W0uea
zA%c3-OPMT5dv&0-c?@Hxy|&-Iul$`Bmx+*oE_`+6SN=Zyi%Q(sbjP%REO#y~3*5rI
zp~ek_!ftQH@035+v3K~uM3Mp4)`r4GMDVAF-l9LE^pJixv*~y2e3WO)pxv0Fi)<mN
zr=;|J^H8*5%C^0j#I8fbfV!=nMpZgI9EL=F<776h4Q-EHxID0%d&24=h{2h~e5a#p
zo)29U(CNabfaFh_3(DUV_a+Lf5dslOOTQ{HSR3RdKWq9L2FQT@GolHqA4tTt@+bx{
zv7OC(7sS;F+S(YTE5DsHv_~oR)Ys3(kP0>q%57e?|M2-nXQHWs_nZX~%I-2UylD+B
zGE-!|_(}koNb}l0>_C<A5OUspb&2ukjy6jO`q(ua%ni|=IN#~(<bpno=C~sTQ*X1G
zjVSa)tE}wkxjz)ygQG0hTR=)uWS*&b=b7BlM!~O9LL38N`(^N|YP+?J@k*Zc=>$l^
zPbKLuA*nQ?^6`tz*38U3I&!XZ85N(b(OL;Vy>IuxqieD|WKw`pFC)LPe_A)%-XiaY
z!WW+g1|wEe9zpE8M_m29yzW>Nb#=sUEAN5VzBHy;mF8P9<KchEpgBCm&J5Tt)>?9w
zIcUk4kl{OhsdrX`pg!J-T*@!=rFD?R-BkX{0oy;=Lk6y|+_p30;q?llT8b6!wa?_y
zn!IGnaIqtN7mmj4&X!JQLfa#By*J0mQfFRRvY7TXY~rnaZT)##bS*o&g2mgKTcAF6
zV$KG5H!G3vXb^VDO3;G*oaQIYt*_IaSNFg+$Gmztv_)Nq=GdQbAeKQMQ$@WnzMQbC
z!{apyVJM(H_H}E7>4n-~aOps)4O}i^uPyJ)4Xf^*twim}wq)DG=7A2Y7~WPzZZwg?
z@!C1SD?NH=U6-AVh6zvOvyEmcQ~7&R350Q&;62s-iNaA_NPfO%$vsNXuim${*-u1H
zN+0pep8x`o!FF|sPQ`YQ;7P+BUC-HKqm8u4onT)h)3*){ecIg{dB#ifZ>{*8a3oD~
zg6#9lo!8ac1R-zB@wI<|u}7Tb8EtuMGEyfaB(0`-?$3)v2N;V;mTRj~=QDx;h1b!=
z1MXvCni&<GIW-_o3!}+Zd3&knaA^J&YPI~r=>s1nTmWLXX~W)n*)Di~d6r^QTSmp<
zkYMXCaZMi4zz~fGlU(ivX0=$WfkM4Ti~Z>_cjLh5yHE~O)TQ5UDuJ8!wGE_A63Nw~
z<EZ)j$&qbF;ihNgU%0KOMH(K5RE7P_%&wq|SS~0R@@Ui2cOh@ITLiCfI922n!&^v`
znfwL?Pd&Mgdw5$<UCp0Xgalw6T~-jkhaO9=ugB>>Q!RnV34|4qP=CJGj8^nS4SZK8
zYO9wtPfTj&Ln$0YE2Al<^zI`~zRDd{#PFyBKr5sG?@a3g-f@2!i!x!-Zqv6{4Ljad
zt1@f5{^3h2!|CC?T7NEG^1RfSwpC1YeW-GD6)S>-XHi7}blW9NsSGIB?0@85h$}5;
zjs28S?7>gSvAzD(n$4=!O>5z2wxvN!=xeFWGzXaP<Gq12X<rGGoLiYxFbPcvn9=^R
z`xH>iyf2MvdG)suIbmuT5$o$rYy={Pa!2svc3H417(29e{>XG?Dd5}lXyo2%i*!s@
zB+xfv=5wLSXpvMg7<R-%n{eADEjm^9js;Y)Og<&n)a<`7yN_Xk#1+a7mc&8g2#IwK
z2=L3N5<t|yd547kv|z!+wzdHdvM^*M7uFJgD(OzUzDzBHyJ7&7s*~RoqCaHvaHr}X
zJPTfRZk*T}D1vLwI^}G93d&7=3WQbwXV4sqO-rFS+Ls46r|9K2txW-^ZT$1!dzrpZ
z%pJ6Pq=v{MR?`Op0WM@6O;2O#lrB9|K!ZC7OIQDH_Xtql4v!;bj49lN$E!ab-CcUA
zT}Q=L=ObrQj@d~4`Z9Q(F*_X5GOdD}z^)VZ0x`a&R>|(1$E47KCgWUg(U`KOP-U*n
z8vEMVS7L>RmG9AjG}~=_Ikpd#B&Y>8=WS(?BU@MkjaGij(`F?(D9rhcen^$jNp&Mr
zroH?9N5b|naMU>REiP78clToN^1yYsvB9VQOg=1PKUc{lEad3It`W`k<mH}Ei76L(
zB(c)VF-K{0U^j~>0tMOJ5_l;}^w_xAqhE!xhHPhqNhxKtKj+Xrb@QiX>1Kj4GPk{-
zdw=f;h3nH&+*2}SkOnUoJY(K5;}a7N#3a6;t?jf1E1JmxR*D-DtuQzcm#Kt>E&g4T
zO&YjmJcHiXQsl#17CgOoI5IM6BbU3A-I$>_bseaI-CkrwhCTY+x0x@i;0oVaL)NGz
zOc{g_3{;_gMk?YB;W2x6{f(PMEjHv5)b@k#Vi7;VE%PaQf*V7j!n3fRZpI|w_~ZIi
zX3r;-$mnGIM_mTMriQXu!ZK($7=l@H!7?<Z^Y5=mSJ48Ojer44KAoyol=SpC3<IZ9
zR~`q|-r;B6v|rmkiwnv>HH*FMN=;;FxTEIZ4728=0mgIQy(^YAW+<K3vvRNp@Ct%m
z;wv#V&QRyuyo6`JRpsj!!%Z`7S*l@vyS<lk0ZTnn78d|m<~MFh+Y|_g((Ek8`(1oa
z9apim{GJj&B=9+@-TeG3EeuXB120t30?v?wsO01V>o29YQdsqWTK+&ylUxFb|I}=1
zn#AGA51{q7*f(G$Nb;5Nwv4bPL^#IfY4ATC@|OSxu*`)`pzc0voskg?im#ok;!kTX
z`HTjMF(Afj#HLU8rPc1X%@ZTv8(xbRQvJ+)_^M(>hmp6*;MhJr&j>}c7eph@27_QP
zuE-28r&sVJVlAD-fCLIG$VjDyEeb$zF|gT1Pm)B^928?`v9=h;Oq`|hk>JO?j)JO~
z2gHL-x-0jL<fH^Ujdby`P(*Ja-LcF`!VCAUrU<Y29w_l6apZt`cEpI9>w$b6G%W%r
zX0D$aa7<*aWpJ?x!`J7e@~Dl}xLRY+1on?4agwtSKAIKk;Ynzew8uu*y?U!aJ3S14
z{#Y6&{Ny+@w+)J~mQBMLfr|~m!r4cQ4mURv3dT=i<4A+Ow2wZMlkV3Y&*Va@2qssv
zRS@^@Yhq&2OvX9p^s{5!rcxLyK`hXsVYWxUtR{8%#QKudoyO_7=R-;+Q#oCyu`+^l
z7}PE52phJ}XwrtxKc`5WJH2#Bun#%UO^grT+jC{KMd`<kc=xX1W3)+k%TGDG!$-j3
z*;-bfdRZQCUmR-;v9q>GP757up36!@uaLqkg{E1c#Mn6f?&~RuLiW;c6Vb8J>gt*J
zhFp-!o|cUV1V-qS);Xr%UIfi4p}h*Dfsm8`SR%aKx%vF77?_^$?Aq}#WssJhPsW|<
ze`C_UyBPhdrR2re_$?x;cfqo`>;(u|Ucecy*V~N3{9}LM><<~#p~98S+eL>XWNjX#
zbG};eQb&d`**$kNJ~tOnk#|od$B#QfLFc*XLQ_XE{PP!ORii^FP>t<uV34Sp!WmEe
zPcPeClZKWwtaE_D4dO_ajWYs{JLg?2r&&IR)7)D4Y|WYg2bK1veFs-m(-m-U<PwcI
z_!%l9n9je(z|D&r(I`$$+l&PwA=#(T(_8tLq^>BfSGCPAew@MY#zLU8z%;|_n2Hkg
zILQ_-zj!PhVf8C$qNK>Xwv#Q8m7rdV{<}d@OGO4I=yJ>lNmP%cTAOO2m&K-xD{!`$
zJp>z$?P_=;TV&ubp?V3;p~_W>M_JW|@|c?U8CNp#Azm+SunC-<AspQh%@0o3HTF6g
z2X;zZqG)%>zvSaU+}s4}%<GI(0-}fF)ve5a1#6Be7ZtvPsQCjW>=}P<zhNJ;Vd>Mt
z`39{l%~iV?;`%IX#c2DYU78#@UV_e;h{_c4yr#--bNp=}<srF9!5K<Xp6_lfCS_v^
zjeFx>MSV&gg>5at5*cu;bImf+5)vo3H~dxWKoHW=MS7Pw7V{wc#3{^#APx`0z(%0P
zg+<$>to-)aS#>IeQre^l=!<F&dM?`U$N|3yHT*dv=^-Py#H^kEnm(<PgheC7D<Ulz
z@vNjtja3g)-hSEWx7qMzR0V?;#+#K-2g{Z<!Tf5O^W91SFon7C*9U>2A7h&ida!<M
z3kn|->$x+ZClZp7NxXh7^GOYPq{GUrP90LHwVJekvO=Iq(iImol?g!HkhqYX6rH2Z
zFE9S<1F^lBLZIP!vAteo!~|7oqxO68er~*;GL#UFf`}-zv%z6yzCQ(4qoImlQ2Dib
zS||7VjvV4eY)0y>J$iUf50Z=QjCkv-eZ<uBaE6@Wv}$Rr;$xan_D{eiSV}rj_h!k2
z>7*V{RbPpKr0e%s5l?1vWf4pg90@@UDiakCv4{+3grgZH8dU5mf2mhHQ*G@p0l#ha
z%tf7Yjkh32+<}M1h!Ns$a>c-qd%0<5C7&u!D$mSa3g?}r@w*+M#gG{eufNZU1Bv&!
zqP3|b(Lci!z2f~aZEUyu_1o*ZV=|Kxsc2qC+s6jpKc(g5jAKJ2SMw(7DT!YC-<;g+
zcConc0KK<nHQytc+dA_vd#sC?y{e6Gz|!~2<r^*Kb!XgPfzHmpA0ESo<VFIX($i0#
z+P_ePMpGbhP{ShAGVhU)41oP9uW>l)f|73_q{-A>N;M1}+X&z!m*R<+{+P~uma9l?
zm^o|A$4AZ{-j%<niY!sv{c+xA&}<j2?LB=PSKs~Eyy#nDy!uAK%0=1T&eWBXKC0#F
zM7w+CHfqyrN1`Zb?zj=9yr)-fiH8xC=q0PtiOpP`ub7A9@tB+NAnpRcXEmx~F?nC|
z$WM7s7U<LP3{6gxw086cjD+AThB+;c0f+jRSnO5Lt{i3EU+Zfe%F`JX#U<0_5<uDj
ztLRbavTH>zCqUBEIcG<9+C@%Y4bWVk_}pz+WfDCHv#5VRF*~+8)*`130SB%>EkmTx
z!3-<l&f6^(S$Aq{%R>yyJcFB5xD>2|oV~x4KhXZ@D^cZ{?Csv7s(5i>P&C)Y0)WY}
zvl=)zoh}#8QQp6KP160G>1#RskOsn6kG{56)jV#DuIqti)^`8hyTxY9JKs?QXE<0a
zESr<BA04Y@_J<>$$rll_a=O`J-Nsn=hpal&<;8K~?Xv?T3;Pt~c5Gm@0kr(%wQ5Y!
z^mJ93(ZR~bFaw&ojp@S)WgX}}RT?QEm&AdoycmCpF_k+Dc!k!QKP6w|7Z{4Kpv!Eq
z-@i0l>SChp#Jf1DpDKFB0`okRDlm%5Cz7!4LcMb@ML#?|T|SljJ6u{uLR*vy#%I%l
zMueJFwMtWSQ+F&vMpop?Wm8LVabh$r=y~+*$AyjjY#$X@q>5qM53s*>`W+A<D-JDL
zOpG1z+iFW91p&0xwa$DN^>*{L@9?&^_pzB5L*|UbmOH-6nAp#4etvFE2(WCa^U|y#
z1}O+VxH$P%Fxc9vtk%Yj8kmvE@KDKYeT_am6N4f>BOn}MUJ`R?cyE&AeX)HpRZNWO
zAZ*6<uFs9fxCU3hcd#QZq}l~t^wH44{kyRtLu<Y-N04OBLb?SyHjVX{SaoN;K?KZx
zi9*!N)oL1P`SKZOFP!)_v|Plis`}L&NS(KbFKUap78h5gVL)58-Xuow`CAD*tw8Pf
z!I5;VZ0YmCD%=QK%1xFBm=Qq8aiMZ-PSy9;elJ%)@KoOG-|HB4N!d8+$0;sx0_YmL
zX!7|?VOL5lkZPw&ujK3-?K<gQFRZsbwkHGV<v??-M8HUm_W_x%Q_R4uiW1HZ-&9CP
zY%Vt=45nA(-KyCEoPJjceyylDY%x5-Q)X<!H5IPCGoTHf?6fYzjELu<C4_jK+}n3v
z>1%KFvkL{J-<Z-;g9i9uc3+^Q>-(evArt1)k>Q!9Bg1?9&=}ApI8Y$xV!K+Ij{;ss
z5+FXfifoP(qYRq+My~~xV0Lz*k+Wg{wGHSGfD)c1);d|nRY>pduB43#T3TA&geT@^
z=JvR|$9a!Z-UNM$Ob#IzI`l5gKr>5dWM>?VZG9JvFdxNG@ozR%UXi^$`y$qYR-ql;
z;f<Ah6NT$cMOV17E3S{|(e5D`m)QD=uV6wf96R_SDz<~2EhRLr`R%3Wba%iY#{f%P
z6m12Z*f%W$?StW<@~diRG<A72G+?6YU^&38eWJu^*{}W06r0Wvr5JW)SM*CSr+Z0~
zPvo$G-%xb-Fk@vlkc9cVgc0e1g?S$*=|!4HOPBS6H;>0rWt2*#-oTk!@F2?4FkTCP
z1;)WaY0$fj)LPT~*iA`X44Bx*`7E${WuZbAUvS`5zUaA{nz=3ic9D>1L{buy?Rc&@
zloL3ip9tgB38cL>+h<gT3A+R#A`RWQ56F6U%I;Nm$;lJ6)#iGqWTFP<Hru>+nQf)m
zLGh~@ST^lR=qyN3(Se3oKEU#l3n;4x^dMp&wtj#EXIws#BVmpm+_pC69$d~8PJ%-3
z4&rK=2BD#~i36AsZ|^*Mv^o>~5nn`5XhapQw1bTxW{!AUL$8ySFq4e!sD|_TtBB9d
zZ?~o@OHGH4DKH9H4U-8eV3Hytl}@eekyK1~g#>r6c8rD?sABgH!Txz^ry|xqtPWbJ
zq;!qSdK7zah3rv*v8A`A<RitLU8ZSy&={?f5RlVc1PPKD;x967L<jTDdZ>z6T|#uA
zp<giQX$6;;Kei0Br>L-QCNW#R9wxT-mvA|id%|&-^J42mF8rNpI4b%ZGMgYa^rdbN
zyC#gm8&A<AmvJ~YI?EdY;rm(mTZ!t!BWw9iEmD{VRk;xefBHG;K%m~;3S~g!FyYj{
zqRKxoR$ctv0XE|!wl#=FpxmqJF10mV@8C#XLL_DCRMd}OGzUIy!baaea4IA~L$hy8
zku`mlS`1H*7Um~Hd--7970vR#ie_t};V5gVH?l=d1KCP+n~sMAVDE|hy}e3jnlUk}
zt4~gY;w;=fly$AjpoO-Lmme>7+F!b$HC0FhYGDy$N~L|gzuVi;HaSdrrU*n&m8&H|
zZ%rw0Sa@r@%D{_IC&8!>+~7D_Qm4xIMh8?PFNflpgIJqV9y9lVBYIwpLMwA$ADVl(
zR1DP2r^#^H>U|vA1f;R)tqxU4PtMMhp4m0WEOXy7S<BpV!rEG{LtIf;&Q~WOP#ez{
z{aOfz&oI$Kq-fL9r{*OT_DadZ5RFHWqoe!cscOjv{VmB$8qd_wfr&gkY+5-f5@i3V
zwG1Khbk6T%3=!0x-Ds6yY0@sc+FVqXbft&2{8*~IP8CSd4?4mI|Ct3~lg7kaiVYEd
z1C{Y^94UdTs3`HAq7kwxT`qs1&;7-zR+&a(+S^o2bVDw7NloGPh|(u)x;j3GdSK{R
z-l&$i%oAoMT86{2wG9coq8aY1hC7!R_2?1NQ5ksU>9U1hYB0z7F01qN!y)t~k|($?
zca|0~?>r)!N~pER=NTuHYP7B}XbbCb4-<eqlP5<31ZQ?4`=R?g6yk|^Wf_KjkP|JH
z5sutE$aOIr&rtTWr<d%KS)70gUoP?#cXtm=kR0;^liXJRcN@Y;O!;#BR1jf=>grmL
z;>(fsZv=~i5{3K*(g;{~x2t>f3DQz|%u;inU(s%H{^0zMw?>zk@a5ZS&~3W~9F<Hy
zD}Fntv;vIQt*$9+IJWdxA;*ORH`tS!FO{9G?ap5rZ&detO~YigEpkfBi_rpw1C;ty
z?r=J+VkNky&hsk@pjA6}Cih#?TRnNHt>dmQ#9l`X%2OBKL*1@VPP8<_wo}mokF37%
zN0kYSm1Q#I<-7MKznrkQcYQ3Vk!SB6Blumc?D%HMDq&4ty0S=(UY?fU>`l@Uk6H=o
za;qkAbb_K6)^U0ExH-hxwns}F9eZv?j34u^xt$=Fi;cjmQI$@GBjOk|G(f-h0|yG%
zXnYXlO!}7Nz|MA5?xK$ncWuN?97+#Id{<6nTvoW(slxRWi4$yr5ci`=68xy4wXFe;
z*j0lHlBC_vAWO(%dP-#=s^dQ^=za;{z~V_0y{}pbpMc+HDv%%h3Bk`0*@^>fI=eB2
zPt+<%{N{)!olZBujQcjLs8~^kGF!k;oMh#sw#U1ou&`^P1VO$M1jv4W&r}?Ug`(*7
zr;sU$o4)R5poKXaqn!%5-%1@-h@@LzZ@JL?g^CXPnkntzR;G#l7(8#Hx(PEh&XMF{
zX?k=osVHP6sX0|j@bQ=2NWRpXflP*;1d{Weo_G_AaNeQ0qtskz@k4W0V!c9qPBLVN
z!eatTjTMZ~Kaotbc3~zK3X0#H9LMWl$3)HHY}lrDXFMiDzW<B}0;Z3GxO^Jd_R|B2
zAR{C9F>DBzCLYJ>Xf20wHl?WIvxt&wCg^z!%(zNika8Kt#OvhuUp6|7n!D#4*oSp(
z0wX-PN6R0_2(0M@d?68Hmv?5*3E_M`US2JoXP`Gf6&BqtFctePXtY#5>@3m3KE{TK
z`%4J2y2oO^(cO7d#TGLX^6`-yFO6X%^1xsyb2@{C05_b>wxy&`x|jehWZG2E(KuF?
zRRBUA+GQyI^`>~Apt3T&`d(|p?g#qulJ<U;C`HmRmavrEC)%X6A&!Sk1vM%w3|Z&q
zv{muhm;q3hP?7a2r5M&n^+;)carHhtad6wE{EN^BlCUtez!A*)SwNnKhO(934>RRP
zc`vbGv63uegSbow%RI_qGo^*(cE1%kVx!QTeIwWLm;f210;J1#&Ba*^GiQ@1Mp0jh
z`wunXU(@dXIDdV)W$R9+bVh#P_mloKFO|PVa4z0&l$Vahd~~i#_MDFliGU$4e`vO9
z=&>&op@4F8Q#S~X<`3%TRjw4Fw6wjHT32#R7QE&;-#wV2Y%$y0PqeBr-HJc`Qc*4y
z*mZ5o%X5mh?{_fEvN03mqxAA$>i%ZQ8iJ*2La(BdYSW$3|Eg!EtuZ)}OB#maOB=`1
zf&)1_Yd?WWlI3RJ%g+bu@PR14*J+x)TAZ)nL`YiNT{ShqVDH?>@>CiWA3dN7vgN7-
zw5MpH1LMAx4iC@*BSf)UE*}pW#y?LZ8GSRI>7mNR)X?xmGB!jP@kkRwyB5}0+N%GU
zr%Y|?oj47lT&tD8M9iQGdlNN<UcCp5USsIl69+ew1Y8CId&#K7yN4MXgkjleGI3#<
zGkTeG#AJTGgpVL<YKY21(t4gg+yHKDcALSW-cinPlh?U*@xQC2X--ES_P`24_!?Gs
zK)#nU&&wNklNs6W*V{+Qp_glMn(=ER=mX(X9z$Oj4QCccc}i^4#H6H*RM;Dt2uI@)
zo`T3*_0eX{2Ui9L{5K_t-p65WG5Qyu&!Q%vz%=083h{#f`ICNrVjDx)=-Js8I>gXl
zV)HoN3>fM>oxOmhts!Y3?rb~{3>d<>>p7*}mCpt!PCiy+JFV96_UvFAtEruiyjC%x
z4DxNx7N2;-Fmq3nwef4|po)PY6+01k{vGBNj}s`Xq>s<=^E5YKBUm^!Y|^@HEwNor
z){${Ah`)H1mB7^BZ@~B2(syZ<jQJA`lK9VZ&mm}mFck&q$o#?W2;o3KdYI%>RgXRq
zex;HEYEPe<uj`LP5KrthR0#HRt}*{8jA$VvRe)NLiySYyE95+PzR`M<<K(mGSDNfi
zIxBdE=<;+%2?PeljM);O#$z;Avs@@fvynZf_orT5iMb6lRzb&X7F+CvhIZf>o1$gD
zb!5ccLRr*j!bG|6i3S+E)-DP+0CT&Zvi>wc9q*puU}L~X?gUg26VvYT{UD;+Txe;<
zhBlOpKb5t)vV=?5zu6`-_2knTI52K2rRNeO^$~5~Wo?Z|;Tky5UTe&ZnSD1WA0@_%
z3KN@_*TJktPcz?2F5t}KAvR4j9v<XgZGGtyHqapFeyN?_Aui4lQ%!y-qE!a-{`D0s
zl>o(Q2cy%{rG+5kELj+Vpi5qwG#7M|5u4Ezogo4B(>^i_XBv*5@q|q5aDI@^U@`m7
zo5YnW?jD{4iTDnnx(ggb4Tv&91J;;%LyM7DGb$KM2+w^Cn|kv4o7(i}BfhkBVe(wU
zVE_ha90hW<KSOA$=0{v-Lr49%M2?dgs{2Ww-8l<*YLmm%a*AxO4R9bP(ppM*E;9B`
z-Y9K+Dv%l$5;AunHP)%a28;7?-n}N&b3OHRT_NB6l`~~0=>y50O-g1&y+*aWlVg@s
zbAYX`KX<u%fWVE&cK6}>2EVUvn_^i6S&H(s)hTdk_|>W$ZG}W=Tji|#6;xHq`bwPK
z9Ry<!+%eXMzR+b)r&Y>GotoBVY)Uc9$)2)}e0V=I*8z6#9;dFZ-}3VJbuM6>J?wp0
ze7P_wN^D5-$Y+7JHz^5M7ixeyi2Mi-Z93$phw<WKug&9phl79#&8?}W!l=>Y9tgJ7
z`|%gp=#S3fW%mI+a{$9A=Mlm4m278&ETZ?=noyuX2mU`aopn@|&)3BPX_Rg*Ez(_=
zmhQfQG}3~=r5mIhq(!B>1wp#IK|rLtr9o17=lfgh{ntf#=Xqw%oU`|5OY~AW4RTi4
zl?!(>t>-6f+^<GdDo*jM#|VxbI|};JYmQ1|R8jV1{qa$S=Cm#hvlI30we&=wDZRnN
zJ{zWLoh0woBn67G@8eQT#Zp7VsgZWS>KC-$%0aD1arxD5Xz}CX-ZbK-8|ASeVAC2Z
zl!og*&&hcuX!b5j9|^cXkC#80uHV!$N{sd%l?d~4mBZ?Sr7@+yMZ^rzK|rzP(!DYn
z!Y5eX?)~hULVc-wD2s#=k4o@Z10T?1VRrY;zM9s0i;oN0q5H1~NHo8RovuT<4i{c$
zuI@YAL=Bd1EhjT%5nS$7dn-t>KIx<Nsykp~{puF{`M)G%UF;J27bzex(>L8%*t*ye
zmqxas$z$OUeoeK>#0Kj`DQpovdu|`Yd|(IJWib3qAfVvs&OO?zV1)AUBKw{AdL0YR
zDwZga%UT9oEDHENa94Gh{F>88Gj)7~x6IFqj<jM2I_{;Gyk6DPOi2;of06pD2FGt^
zXfN9@*mWF{%K}vy4u&s39OKN(i|8Ob7jk{HG|Fiu$ntx&95L{ccFN5B+u5e*gw%JX
zZl3wI_;BG1y9=^h6d_F@ejqA&({?l;^2O=cfef#@?|m1ovik2jjNL4(F7qp2)zIyD
zYQ33Wq-W1giA$D)$#|uLrH?!Jc6jXQAbHv2b+Av*u~GY{d)6+aKmkE#Kf3<G01kzm
z>geVdGF15RjmL<>KN!CmiG}ZtJ0)94jRQ$vY^;|<#`CVOgCczKiS~LT&ej}-d8@0p
zZY^ZN^rNP|2bwQs#R#)+N{fdl92McV+EIO>)3{ysFTN+Lpw9;}C|gD2fmhZ}=crrl
zq=JG>a=c0v`p?ZB=Y1vjgzb+`GdIEniFlEd-iwJIJF9UwEt;+E1dq=|kWt+BV<O%g
zK?<4wCh#aGPCgzi_s=EvU;dd7ISnifmyVDrj$%jVH||r}NYV4P{ndpSj2C~ZY40o$
zTT6;oE3kS#A+5u@u)$F>h|X}da6}5Re8Up7a)>Q&qMB7>x;s(hWJG=RO`m~w9~c$C
zI7$UP?omI{9UJ^U+c#|3`0Bbg#%i}xn0(13&pXn;{-Ih$jJv*!<)!d;SAcO>?$q*z
z-E?c=hDXn<*o-Lnr<0ysI*S=@8ZJ7BKFa>7*X2z9kV%?+LhU^jFFoWy`@P@SGpgFt
zZ&W%uI;4=xg2jw!j{49c-{lT!J<~Y<dtd@#lDE6C{u<DH5Ec^h&6*Y4$Ih-%##cuY
z9YG(e<y@>6Z_7nGo|BgKiXKv{zVPOM2DZ=!*d;y0=P+sgt!3+jXjO~aJ-vHM%@5sV
z#~ibNK7RL$jcZ7j*5;dXge3MNrnO8=q=)3_7oTZzDZaknC-*=|HJRhw140E<)Q{QK
zEegW7mS=8pY!)yh2<d2vih9H<wd{zdd)vzk_Fz10D@q(J=PDmNmWO~_A4WIrnFKh5
zVvtkq#s1*1Qe8baTj~d5x0V<?*VvQGN{)NhiDtU$6m#6@^Y(j=_OlmDo~Glhf%t{5
z=+za;UhW@A$)aeL4JSAW4pY2%$Yp0|>6%)_8Hr_?^@R;(*}i{{g6uA>@fkg08l%M+
z9l4Mn#Y*5ybGHhvlpt>YG$X1qgN0>|sm9^*2O2d+=0P$WRKiiW4+O^SJ2!Sfd${p)
zK&2=xECM)9GT}{f4oyV{{-H#d5>G-|e)u@zk*Ta_zb%3oSG@s(smx^0gTAnU>(6nR
zY&85$W$m9Q*_cr*?49>M7s1kog*=pOsXY-m&qygKd;zZL@n%56XVWE0>3O#y%bgI!
zr$MFMbJXP&A&m{==`WBZu0BzDE{DCKIBh#Y;nW(4&#fBwD*x5(=j4Qp7yp_7dp>n2
zb}e#73j;`;b|wVzA-;cUe)IljGiAC<tNl(JQj;7B<VvFg`E*_A@?KrEl$KV8I$>+*
z27Ubem_j~{#Wd8p2|BICcI>IIlR{`<y85vw10<E+h5O^`68uYVT|@I)tJ}g^t6m2L
z1bSSFj?i)$zGq4fLIbOuP@%tV>>G<bFxMU)M?tpQ+AjXvx=i}$T=|C)T;^S=t(l%?
zI1#`%?H%z8wT_+WMf}tDr`j!Pu6eVD#c!`Wf6Be{uTKT7F<-vM9EF^MOoE4K-PR$D
z6f-IR3nxMMkw?!qdCj~kCU!o$ydgz%%QL;@lR#=P*@1fhaGH~Y0tc>vZf{xEPsw65
zgck6(r#sO=XKFa~Q=>Jhu`>$uFrTTTDQ~ymAB!B-O(QG;yBNS-bXEskl?UE81b#WC
z+>|9%mL(>dF$e-r{=#|)D*Y@`pK-XNMo-VF*Eie5+Mwz5w>)x)zm?AY{eeWX_C3ln
ztM#}3nrcoaZ9XoWQYOb=;4}p28~;PI9<|LT4!e}#5X};#waaN8j4f+)W6sM%`GS{M
zT){xfB77#}ZE+7q_{-F=c{bC(TQ=TOhSe1^@)5E?fA<4b3gK;A>;%P8dgf-d?T<_o
zKlangzqhB`=^H5(oXQhrC##6x<P?~aZy?x&gf!Qo=ZO~0!4`fHW8<6EC#&9HclU|>
z<zIC$2wrnQqc4wnDhj58k(5*2sC9NEYSnSo0kOsZE+r+WSRULS2hF}>EnvmZ4mYkL
zBM&S!G5m|ji-$wRz!1W+x=a;lrWsgXA<OINq(Ym<3JT3(^DO~lRc!*NPzDqniCD1H
zPUSPP&n}1scUFkZ$v?!~nF&7o{^ny6oopv;+GzV7<hgR6qx?(7{GJ|6(8rq2NJ<+9
zPQfOd2`a#Z0jR$Y09!k0y%=A;Y-vyv+DaQ{Zy5A_jet%-G{ft081c7kxPAMR(}#cS
z5oXVGp3apwfg<6Z6$-KZR$BC*&*i>-Z0tl6ULmvXq7gs_ML0GY94swZmy#Kmn@fs<
zM(=c4ki5m;@CxH!p$`KKy0JnxI(CzC!Hi5U>yJN!Wm;R|6-~sR3RwsGELvppZ@Qf8
z<^o(Y?JunJ=Ss{3)<3wtSgOY=&SDY2kX=ZM9yit)!@&pLe(oWfmiDCu&|C$Rb`sOr
zZmH*2ab>zo{{DlNHqQ@IE5w@CCUtJgSR015f)<o4vT3?<LLD`gLDxaN63mpyt{@8-
z7|hJE6%!v`XNm}RDC@x$sL3wk5Wr`z|NS?uA?PUdWy!EYzp62!qD5%A(LqW}%lMS(
zmA6B1;Ok+zPaKBf-BLz@VgZ-*e-E+fYhZ4~6wPCumS@UMYa0FzGc{E?v$iW*nzB$>
z_6V#Vc7Aao=(J*_p~2;OyfihXnJ|y}gZaaf)_fgCf<)DWb0a6i#h*&o69S41UW}F6
zkUg9O?`vZtm#Ce2vj&W3Jgn#JYTxnVCrjSUD_XsTHn$86#D$v=YgaIU`-h&W7w*L_
z#qBRFVzZ}5AP|%e@q2Ka+|Y38#(4i_;iZpphYTArx)f;_ZO0VLzobbbLl2AOYgY)R
ztWP-IvywVOW8>%lb6YSVu@PHzO)W$(k$IQ)a7aexyvta5fA?YScdI@dA<{R!H<l9Z
z32j7ahH1W1-r>@E@c-(^iGEoU?yc@Url0ntrDR&U(W2xw?NwUOpYH*e1A)veq_;6z
zmlfLCoMaoJR$TBcqJMiE(RF95j?#R^bPEq#EH&_w>EMvtS77tOOb$AO!kPQkhOCG^
zm9|_BJEFhwXFcrGW+PxTy<dqfRGvSn#PA?{`C%Ued)a=bJ4|aV;o)cT-QBzC2-#72
zC}A5aU-n2!9jpqVudp~8*A`bF|Ncir8Xq734|E8g_WyRBFw0ViNVvATtl^jQ?!vJb
zypqcwn<rwwCgEF!N(V;w=1Lzpm%WB_n=9yA1q_>k=Ut_W0xSN9K)y&dWJrq)l7fNV
zKrufrhZp~!xYZ2Em71t3DlBE8r1nP&woW#rv-n`!J;a)hke2AidsA$uq*lnb$=Lym
z3&E}4eXcBvYeq&T<?+M$sAnp_<dc2GG}+W>6Bk~*x_H;TP0WOwFwf72{P*nBvxN?O
z)|E^n*L!J%PAjqC)9x+{GNC>c>B3Bi31hLah^8k`T;p;oEr~{nMV9wz_8hb@FRFOe
z&CO>yHukF?W<93lBtotW-{U_eCg?dhASJUJMUU_i@xjC9)nVb7%aH{gXQOa1^b+yp
z36)if{!#TZbVtPEvNVimnQ)NP9y3+LlvC|vxE8$RbDZ6G@9q<fSQjWv$x6w9rQ`9c
zI`A@|TI^z!_j<{7;^yXNIF<G3VOK)^A<J2+7AAg+Y?B!_Tb_T2T}U%x!`Er(??1=M
zZWBnU3X9o+krd#3sfS6$6<LNCB(Cip@Nhyw*VXM8icMpmHr$p+scc3EOH#!Md7DYM
z3N;NADcb@$O^LY&<+W)xi@Ro7+3yb1N^#j1b*oZj7dDmHk0%FGCiRpYdNTg%m&o!t
zvC_odXI^HdUxupgjt`RPAhc0B1_oSJtbSkzC$VtU9pUC=)zjB}iTgSh?)TV4=lg{n
z`UW+(2d`gClY?P4SlTnpKsW7pkB1&QgNGf_qxG@c_z@ZZg~JcRt`X5(*niy}=~0rQ
z5sSg?8t#4;Bg*x?1M@dn;7XG|!o2r}<@52*s3i%)&8c))Hj<m?A$A=4OoLtC+$e1)
z;^jp{!AaQ0XCZrH;Qaa=$6{_=%lURD_?gmMQ3m$Zl$4bJJg0yRm6@42J3H&S?5-DY
z^7DvL2EZ@6xSgn4Nd3J&&HAN>O)l=%?iXkyP40+=xi4(n^3LfO5bxT=WyHZlt5SiP
zWd}=_ENwXGS>E;AnLoBGqI`Tz!j5FqlL*d{cDTzSJ{wVGsHFW&f4kn;=m0Y!`zVru
zW9i^=Vy<EEPP0Iu_}`;v1rQ>A<9e?P;MWLZQqdqGHEL>Z*5IQjpr^-E4YvWnfl~5!
zDK;efY@Hi@=K3G+961Oexsdrww=Z1%@8ej|gHr9rUn`f((n6m{_wJ5%wfhHM7z9&R
zSa0i;v8TfEu$!=Xo8P9SS~ZgfOH)kf$wS@7BPJ2q*y<9vd+Yp^zB`|AUi2t|Ww2IC
zcMNt!Z&-}$k}Nh8=+2Efw|MNK23%Ih?2b-VgYUkmV{)DkZF71#$z0=0^BKH_Qbgau
zJ6soz2LVMMOHlzZTdfB65$SF1?p6c(*GL*WV9s_q6SS2TVI~#2tRWC{^e#1>|3pLx
zI!TNoviNW&hW7cWIRFzV`;^=#=<xmuM(GC>F)%VtliZRb{VQ6a#l!+SgVvdJFw4vR
z&O!RX7Y!#QG&7?C)4x}Vy4U^H`;Aty&DBj^UDah$hK8RU$5{%A{~Y6%x&5+i<1~(*
z5P$Z8v%RxBEH)-@SPR2XGB~)$Og8~=O!%|0a!A$@;L#TJ<6?Su)ij|=)-$)lwY6_<
zi{C2~eex{%i0}B0Y_K%#a7Kf_T&^<<I85JnNj&D_Lf*gmw*=<pSsdSCDUyV<l^TUT
zu_*GtXr$qx6YG;g)KW=@72nldxbZ)1&v9&Kl;B~+PtIvLZ6uC`pPhlS_n^Fa5s#f{
z868a-&c(z25YT9JKDov_u~swSro8rSjxMe+j}eNDxVUoQyA!8?9Px(&%nK6TNuN`%
zPnK$Fa)SpMuw(%WwxO^PrL)#*WL)%O^gqUNH<+uQvk;Z1x(9Kv0Q3BcyYBrF^;7LT
zn$;5ZaN&0_#=D6$;#c_4-sRd(7*!6P&1>Ryv#^q9hT*H!0^PKg^x%;$&NiRt2|bGc
znkBw*y0^SI^GTuL1$-mw59+uu6URBKpy8+e<_bJuGOz|UP%K;_nBI)eo;<dMEw$5z
z)W}&6w)evzJ@0v$BXvJ1B&6`!iA5jgnX?%Woa<XfRx&dd!$<<$?sI){%vrs?JysSL
zJ9~y}**5p!$Bd$yk_ufT4>D{@o~g00`4wthWKi+O#+4i|;cdaebKKHWr(5dbST}?&
zAL%+FHgs0=cR!a_HZ}Bb7E2i8r?x?9rIP$-Bqot9C*c#tzsKL^>nC+QF#M{VR?Wm`
zPerU&`^jXFP9@FD%sL~3V>^U@<kgE1grl77RLBTjSol_wcb(4}t#eyewq;r`jLE?W
zjM23|H=xq}s{><*EBgfMJ;0D{bzE*99#%pqEiDC%w^gtbjD7=Nw7^7tZF<(LT_F%1
zGc{N2nK>(G`%nH0<d*lvie{#5B@j3oTW)_MgcK$A{m@ZujYDUJ79c$9(Gv1dc_8R^
z(-U0}9nR%^q18_K%_n(KSyXB?gFrJ^3He7nSywKmA<EeEXI-=cWti+9slh$g1Q6^#
z$>;BB8aD(6nG><%f4f50E>dF~5{M@>{>T|;lD@GXA0h<f98;6rO(yzYJp5I5cG;M6
z3UVn4Obl#YoT(E(v?5~p=;_yzgIby^$g#<CadOou>ATqOcJz>iKgbT)@}EvNn2Sf9
z1FPav8kci9DS5kTGrs`sZ8eGD(g!qZ86V-tOU6ErOO$~<rrKJdjr01e|5vw<B^5Cn
zxM$?xTn5hXSJ)EFqjV;R+kMIeW!11PGz$gQ3Ud*Oeyk9TGwB)&!Dt!l*U+={tl5d?
zckRt}%@8)v(zJhSF|)ERvbgq$-~Zv=k@>?MZCg4R5`12x{+i}Wv6@`NcaAWg{{`_F
zpO?IQX{g$Xv~_NVch}qc&v(Nt1tDsJV*F%eZ4O!Af~7^TXVRkCJSf2t?qo6s4|z;T
zwJnmAI%$-HUI@G4p^GHaF+sai>V`0g%fQ6Xvq9jtss!+$<Ge2rN=S11`iQ@%o<&3;
zM=WDYci{~hDk!rKQ}V7NDC&U|&f683KRt+k$F12^G>)@WCt`MK8tsS=4^q5{I{EuI
zx-eBfJ~V)ExYi0Iepu;YAn?;o`R@ZN;+Rn_%~!$wNj-0=_kZf9+G~F%#fuL{$|vCC
zpS2%>9B8nLG!0vEY!ptUFhh|}8Uw`AI0tI28h5EguVC?G`H}FwUJ}E59;dsVL96ur
zBJIfUu)F@FQS%qTj-__LQY*^=B`4={GBNW);00H3?ztE^L^QIS&}LiIm+J-J2PMR^
zeEF`^tcM#4eHKj&Z1(fR2~g%y1wNbsVx+O(SsyqgSMOJ@+yB;Ai(O7ipZRnEKy8;F
z0Rna;5KXs>*u?Y1<(HthwfP(c)qASi8;;mUsZBe*5>9RSV{gVKYG(a?eC5<tdSQ`1
zLyt~ND@o}0So>?_L=Hk1Wd=bK(B&`AeAWIccI}fyhX*v+5{tdrJO6VNDQc)=#9ho9
zLKhuu*|*h$8%XgheowtCUQd*VFFGL1)C6ys^YJrZYOkqLw-$vv)n}!n6;D4fm#{*A
zRZEE8m)ms(ebC8ibY{wjo)gEX)&5M}eBXuF=;G7HYh5&qm34m~M+Z?Mo$V%fmk^GQ
zeGYb9u($5ONnxdg@w|VvTio7m0TXMvIb;4XhwY15)_xNo)%tDFb5GH+BU}ggRHUxF
z3bqQ9%(LUe+~tq5PyvVTB!(aoOr&^dG>6H8J?_1$a3(1=HHp^zh?{CE=BG+%%Ck_}
zE=1YF$h#A(S6llS@!_bJR=JL8vi0?#N?ltVnea`OFZco5poMM)_JNfc7ddGKP(ObA
zUrQc~4r%qb_5ZI0DE{g)49o`r>q#`?8yg!7IsnfA-(zfW5Rjk&yWDkW^mcwCK!8{|
zA=DO}HMf}F$3|9cRgI+n{yrG73oS)!WXdB`C($UcDMl^!;U3r#?}E6pQg<D(o*b~Q
zYI%4G)Cwv_bNn=AgsXuvwm%NgR{Y}56pqj4FAn9#!0zGo3^h3a2eGPZ(GRu`^iLcp
zzuE@uT@Qoj3378-mEl0+Dex5dcSS@ksd@Y<(EVCP{rGsu+p~MeIL+#<r=*JkA)yg2
zhR;{Dyfs=<sK#-`x}RUF$m>>VWI`b#Ne0ZXOl)-TJcCfutXw~TbyZiNxO6L9rPeWC
z*UrYIv@5dSxp*iqZYgdlD(XJG2x#kBjN&ig>DJWhcK2jGelN9AjiD^tx!d#m)6ly@
zT$F*oN55Oi^i0|L2ywzr69V0;g-u)A6>kUX>-Uaz+5V*bRL!c+(9y=*&s&N`IU93^
zLnDi>?$R)chHLc;0riLIA}yLxT{whKI|m^p-IqSa0ACvqJ39gWo?x4wy6Va0*opd8
zyWxsLh+S!9r2;=rNKb?&>}95vo=KF%KSlgft;5%%=jc>{kVN#AK<dRGwAEF7BN7$U
zk8<VJpAp!>U!EsgupN(wkXuQUv;5`_{@5`4G)s(_AE|fKSA+IXb+9emVD%BLsqb4}
zd*R$Ugb>%CT8J9Dx*mWR9Po6%+OZx^aZqpVk+Ea?bY%)|#kYI;34j0om7_}x3JL;u
zwubxblc$Bb_M!{J<>lq-vc0{%lFG_$C<j#nT1Xcm72rEXM@MH-RTWr%4lF9OE316*
z(>FSX-rt{|{_Dg*X`q;h2#*l6?<;(a0Z3C5s_DU5ubf<gO=EZW&||=c7$s%RXY%v8
zE@$V?(N&@$&d+;Pyh|9%%SJjxB+pTsa%rpTV2@fRRKb$NPO8Lr!%c|))W?S~5)xd`
zyhI5xb#!(6H_s9i6(j>7PVHny4e^2MOoXv=xmQB;-u$Ukj=X{IY)?$s^z#Brn?Hlf
zK|8>H2>*C6@%UJg7j5V_sc4oL7wJ}WOVelae}eRaf>h6m31#Hu-<|C5N2R2(WR;f3
z)c=snhtNO_NXamEws5mW#VFmN_!ERg&K4FM$|g=PsW=lmnmQzO4ThZqo?spV1I{K+
zer7azD>GA?<A>Y9T{M2}H-if9?sea%XP9}RgWIkljaR>J07L5v$eC+L8atlE9=>Ng
zr<<DWIbU8!+z4J@hAJyJd5Kb$mT$&Wva)TVqJ>?s&P;WsrpHi2g@?Cy)Xq4Y?4p|j
zhI40^$GqDl9nC(lvH}ZK&<D5Q&tEv--fBYz`bI|RmzS*{i)Cd`?^B-|kB`0F>wTV7
z)YJe@S>b7^l=(~y5V^US9QtpdyF2}B9?FdT@=Zp_A(G?!ZZ6zc%KG|=W*TRszu!PP
z)jd*4DKm$Mh9=}`XxfUaorVun1?XGqY~fe$z>RZAf{{E9(MXgECq9mS{&BI;NnFaW
zWp!eEC%jK^#1u1ZgG2AR#J_g{b6v2?!rHTEEBRiO^tnrZzLBJ)Ib*M(fjO(5OT#Kv
z$L-GV>y8A5FG^C&%a8%YrCww!r_<7)-gG6L^9y=<`d;PR+v)Qd8^5`#!GqcJYd5!?
zI=(^Yr99cCsr(ey533VIM2ct~e@wZ=xFzRfZJCCfAnxu)@(~+>!aQ_eWui=RP>^a(
ztIMK2+)mGY=il=3b0f|lZzHUyVBlc8P~kjjsbLH97dk(jJ{1KcIlgD3N*FNTsu~%J
ziVl=@aHz~kHh2nJsi>*TMUS6;S8zf3AH)g7BRGS&3t|!y8!Ib-*UoksW=@cJyaC)@
z_ki;6-!GmKfj|a8Q-=)5=1+kdkdO+j11eGwSQKaH$iSjjzp^%CAN7hx=&x09yyVdh
zSGO!rnQ8pp86SVGX%!%T*-_UB{kh61)}y1N+E9;oLPi#K?-y+&MNe;Jk~P%9YM{Y4
zyDRAD>I4Ec@&Lbhsq=N~e=m<DqM(4>Z6&SE*+y%?|K1)nq&3fzn;S1a@~eA`!b*|N
z!(XalJsV#T$0+%EqAT7H5C(Q1J{6+jxB%@^I{0anl$uvgZ-<h=xEH{rFHWA~6?)B{
zVq@V{>hJCBFpZ6C%2-0gOm&D3_n4T_6XLZVt8O}9_wH&x=t@fZS#GUKI__(!LGd@v
zN;MRAS=Y{b2l{($FtMbKjcIRv`v*2rP$0Y45lfEoo1$62CpA?53HkdQ$)X7I^Y?j+
zo=r_fN5l2fyCg>>NU7zgoE{@u&iYFxNQ}O}dhEtpata_g+*_+^`%%Iq&PTYux->_^
z&c2;Dwddx*ZeVbJVL?ydt(;iJ0n?49hVpc8Z({_y=*A5Q@#Q@J)QF53C~l08_i|u&
z{_n*>L7|P`G%qYXTQT8f40uZ`C_t>;$|poNHl8NIE6`6w)W^uk`0ue-am!;MSn^Da
zV-7|`v&wE1Pv6<J{8MpE(w-M;X6KUR6=m8yn4R${a(ibuz~0i5hmlbt8w<H0j+*q-
zU$27et=$37>?i(#wre>?#=h&`P3QQ{ye_l9B<&i|6>abLz7I76TdU&2<eMlFmpi@6
zL9gq0R>mmB$d~IB&-C8UyWDk7gEju$%pWi=^q|byiu<(1C=D6l7nFDE@ZX4#h}ElN
ziN6HxMLmOIcTe^5>+sc+1Xr*rl!~yVWm8j7(3=d8U=2-h9{;<RF4txD;2a2(>L$wh
zp6qg(kdU841Bh{b-fcC&%~V)Xv2XTK6zdgZ6Xu-R%%OWzqb=-Af|BA{ptHgQS~24*
z%MKG0s*~fL2KmEp8K0w{bvs7Az;QL<e8T61UUK+^g)QybzQor#5pEFS;if$@NxsI)
zL=*r1?xOUyVoVDNnz%uO02YMT>6ARRasG0{;&Z@c26*dp8;N`U+ho99x6b&*q=wK5
zfaQL*cEfN}(^UX0&QvR2JlY8qn+MWtCre99v(E44VRQSTi$^9+wFMofJ_+B;SVTx6
z6s|CsWqDv*0dnrgKV!?w+NsJ|n2<YD1Ji`{w-qMrq9Fh;Sjx=w6kWk~x=0W7l;}A%
zQ(LR4#s)n6T#h$8?c!_(DMwdn>zfIh`pki)CFqHfEkDv8DsyZU{r@`tQ?IJ>P**v<
zFSdlkCl9aW@|pZ^6UI4>-ko>2UNY4xQ>3Y?<m5>`jU5Lee)X-oE9{HOw6K`(6=!2*
z?WwiW=5!J{lhHGzz@~JmHZE@<3y>Hd`93l{Jh|}M{h$<;!qo7K81#)^gUkEabE@4v
z+g}@fy@6!&^VI%ZqWdpXQd7_VRX%U;K6FP3+2D<fo3CycUia8l^zC_3On{CGLWjbq
z^)=VOe$M+7`tM%TIIgyqhmgcZRfWy?^Xjt0L5ed}<C_Iq`@QiQ8L;M1-*7TZikkEW
z&TSR6<RLjWEVg;DoA}pzS?C4^(mo~mnj-=CPsj0ferf5az|zNFhWml+;y=QSAXMk{
z?;;?@v9hwV$-Ja%omoxLDUZ%ngWB8SqIdQdc5X^C6^N!&Zeaj3fszs^M1D<AZ(iJj
zZ~T~^PMHvMQd-WMntek+P*OeWxwSRunexjz)+Rkw2+~`Wme#yYz30&)gdG<4lNTLX
z>C0zKOvw7Tx-;+?9Hy>pp3xlZhuT*;OPs!*KWe~A4}K5Lgg@=cH(vhi9Gd$_EAGB}
zGTukhzQ)0_5y(pF_Ujks6Xp6?6i)2G@?8DX;MtnPmg!j^rwM$w(8_9V!zs{p*@NQg
zXRh@4*q84&rX5@Of(ULpgmH0Q`$ny#U1m{t&_C~-uD7==!`|&Oudy)?#mhue&Q7ic
z+&j!W&bPJfjf?bMpJK2+D8G0yo`^Thh=#<(JhXxH_<nvKW_%>>7U}0Ik47FN=VE8S
z?zXd=|LvQ~LuU<l#hXNi_NE`d+WcGo%=wGU<imxx<-7A_8IqHK`qJUSVCdJBD(d+K
zp@C7GxKXxL5{?bPqk)cW&fPGlyz**zn2Vo5bb5A{0n#?KO4f%KF*pv(2*qFjodO7b
z>W@Fu+y6^!5`Dbc-1i#@!4dZYY{LHlmf#;-LEug`;JOp2LDvCv?-nFTq>c|1p_&#-
zxqX0bO{S)u3qk6n{B^M94Hruqx=Jn@S*C=!mzF;K5b$<P5g#{qvV4<|uNl6DSiVDL
zH|y#qa$vV{AjE(vbD~qBHAGG}xcrpc%AF#fpZ}>uaIQc(-$mZFSU@l#(ChMY+Ig7y
zbC@e#Qnk;YPtMJbsxN$en5KVwTl&ww#KgL86QNZJ(%`o(&S8sT=k!!jXFXpHl@?6w
z$+cUgrP<c>Y9pm6FVBz5%GD$!lXT>$|7)7%!o<&IZzrW3Ll+?}<SCsMwpLbks+PPj
zsf8(J!zh8OC!V3M6qxt|S@4D+vKPUC<gK`(k!E3DnRt>Z0}>*kwfYEJ<9<r=^R1b3
zk?ayz=wQLro97%Pi9M2Ar<lx0bEXM;CYoPGoLYl^=^A+&OCriidjn8wb*t1rTysug
z6EE?RzPA_F=T4`$cP<pb!3A|=(tIr|B8$H~cE}vajW9z<IQjj|+ZK4D43c;fOtQ9$
zG@lsDwKRWh<fQdsW9NyhE4ZfB<DkHFSY{qb>9oH#n%V*`@g+_ud_n6Ae6V`awN@$2
z4D3A?Kd1>ZRlf+-|M&rtJ|x7zco8Is`HTzgE`fvz5WRCaSFOHlp-m6UcO=A8Q(A;P
zc<+%MRK;Evy@Zp%*f{^TK3sM3Mavhg<wpzy!!ayuy(mH9_h0$S_73t@q6_$!5kGLu
zq^gngMdN=T*^FXIjlp?LzYuB2U}-@?h>fG2u~mcH&hXh;cM=h)Vygi4Md^Hv7I{-f
zJcOTrZ-45KkR+v5o9)yZ8sz{bTBofpR)UsfN}5Yso7VBCy85VgT!Z<TB%0IPwd*g0
zkaaFK?Ds@1zIcP*ZwvgIKa;m_w$#lkb>)82-Mk=@8-(a77Be76{8}D)xn&&d@o>r$
zIqmo$%f%h^z_u`nM-J=Ct<us=D2`yMNyLS8i%o2Ihv00CnC`?#vlNM(EuU<xN6=am
znz-Sa_(a{&Mi*@9nJ_<G1xk4K`_FW2gcUjU23#?<|8qDa8|K^YZN;w))Xyo$GO6Ct
z3DOk=I_HNe`g{Ld0o&AkMIDyDzR)!d#yGHI!iKnjvl*S$imI58uyo8E_~udF4NAD1
zFLn6y?$CbYqU)sCBR_4oaG^sf%)E4vS777p21LL);wxQh4Q#t)daiiC3s8_-D~Ca=
z6apqM+>K&LlqXJm;$}YL0pCjEQjDKmaDK(}a}$s(*Z(D&IOx)%x-#x$>RtF(Ois^H
zGZ|%H5n^_whsk*iE4z5z(0tEfFAXfs!i}f_z1WW3n@h1*f>aOquKSpO{JU-{-Q0`h
zY4$_|+*`j*-WE$_+@l<np!4(d1Ao)R5P7il><<Wno}TH0uDsefOQ(zk|3~oKV-M5o
zcT$E?F7FW$rKO{k<gybI;G_cVn}LTjQR&WI4l{b&#y*8lQPYZPeFw-cwj%lq4#viz
zYcz`mXR>GqR3u6&nAq)Xw&v#DBiYf-!SqxSymk+>q9IMpvQUH)$LOFB8y}h3R35O>
z*n2nwf+hmRO*GtgbD3MuxH|IikL4a@a(gw|seg1wlqv;h-s}h_rh8syNo<ncsYG-g
z)LY>E(|pV>xI9xFW!0}k7xB~*w=8RDnB1Hd^?&E1?~M8ql{e}W-ceHL`R&#ezFq3@
z8Vz}7x3k=(EgvzEh_C-$d3flXe4R@=I)@<@4c9=4Xe0}I-Dz*qJBmu!<veHXITU#5
zB|3n?T6>G&RLZ7Hh3_XhZ9cXsL&vE;pU_ck1J`$ti$ag+{%;Dsr_(qj19ce|JEaIk
z0~dP6%#<728b87*a(Ny;OFxs3X~SaI4~2Il_Bu|kO?)S@K?o4mFQ-VA@h0)dr*w~0
zl58S9Z)3M+g?(En$PdQGaraM(byt#RW`LAoq%TJf8ZS$nsqwiaQ`rQ*wJjv1c6U)H
zZ;+%Q=wD3|&@n_yE-n2AA`E2IrPa~$+@%Oun4(o?G-j53NC;m!<ZYuz#*!V4ep1iQ
z4cx!Jx(XI&6q5^3Nz^2r-_-0Ud55z1AogZei#8)mfF26P4r#jDyZh)R&Y1wIEQ^G{
z{9C%9mG~q`sFOZGo#3$eVBAH$q;;yY5zKzYYurKA;_>#M&)~2*qE{RGk$C9S13r}(
z!r>g6=Ol@Nf{eCI7p)#UK2U7Ss;NQM^T$e1G4Vi%5cJXOw!Ez^Dop?ZzZeTI5$JLC
zx65=xH(xrUTg4&R+1EIHxx}wwD*#jBtX5IiV0vah@wz%&3~WuJF-7HV7{i641I^pg
zhhxuZ#-ZB`;JJ48`GUmL%`GbWhl#fiHXOOaUq+1N)+K33XP>Ai`z0o4D+Gte66b@9
z$je>Rm!wp3x{~qMR$>TwXX&d6Hd)@48%o1+b`62($QpaRQ@P3q8L~lk8y;?EwPhIW
z?5#^vlF{~`6xY@|syj@<Jn_9yd=KZN(^S|1LshT!=L#k#D5%0M1E1-@Pxoz+e~Rc3
zFl?!)FqQrL7ja+gpK8!W<vcwYP9GPyG9|G7hR5!=u#Sd+kIYmNnB(FL8c!)X*=tu}
z4mL%)DKNBoa@Ht?3n_>3Eu^|GegD2+tN@Ljk^=n1yi&c5SEWB-TVfS6?~sGMf*!J7
zI$RWsGc>b=7HZvQuXg7roEylj^n`q{WtNnX#lJV;-;*<|Z4gX4nLVK^88}`B?<pN*
zrU~yScggU24korKtnm~dOWM3f=AEFHHo@V+Y!l_4y3|z9Vc4pA=F+P<vA+#4!r)D0
z2W$I#o_J`M3HE$#VciS?8R`oyINS*S^Q{~l5IR`Cb4TG(krU2%855w&FIE|Ud`xDU
zd(GILSRX0=B>S%g+>wDz_wEW0&FZXp`d(Q<D(AVHM&9CK=E8k>rK%lLiEl&di$Sh=
zO>tJNsgfc9uaig+4IbA1>?bWBmd#8BeKBtwzX^Sv1evLW6{@q-$U^a3Ui&v)-q(2Z
zZv8EF9r!Von$5NJn&-{0-=C(J>At>_)BLL8?7sGmZcgvxJWq8&uH@cQ(I2jNKdwnK
zEHBn0G^`S$4&6(6Q_IK*ypLIzA(hto)6?k0I;9sGv%A3?u_|nmDhzSMlSa}Y^2#L8
z_yO0su)XT=Nv>Qk3X_fSXAFF>qr4DhZ0ypL4sQg;E5?O8e({QET_Xpwh&p>#$W6u@
zbpB9W6^M$aNxFf?m~`IPDB<Mp-dZldTh045JtxZXKj@?4{<IBtpXn1tWunLhy0d88
z*E=6kPJ3Jk2TP?h4r0n_$_7st=tSBIoxBXsv}eYPFVKqq5KLa}tE3Kxf34W%9usF{
z$s=)AQK9km7i(^IlfrUQF1*b2?A|s0<Qsb~ka_FSJr;250ue4CnVU*QnvqHcY~%_%
zsS#ZS<@Alyw`0Ae`yG=%k*3>h>O#r8emDRTujrE>`?#u6kAZ=k2QyLew7&Kx6*<0=
z8u-r`tD5xRycF3WO?OGc|1x5#vb`F3F?KVRaJu9QzVrP&62dLD1GUICPv+6OjxyaG
zouqbh%2M?xsZQ~#%y5}>W!s0Xju$V$_R+^F41qV{M&(qh>Pu+qXSjvZ6Zc%!-4zYD
zJO_Q(YBnrPd{*Pa!tU(N1L6f>)^RmP|JYy}q8}c`5LsHFNp{}8gL!@d7}be*DJg^Z
z#*SVmL7wgBB_$A`F6#B96p>rHQ`P+A_Sbq}4(f#_>z$&|BwW^OkazU0-*OaLsY|l*
z1sO!fQ6myQ7}74Muhl&#@>nZkI?tS}muKly<JbL#Y{x2f_>8<xLu6(`b)yzQPGd+|
zYZHqxhwWTcWPnyr!@y1suA#9~#1!0P36qf2Eftgiv)e3x;^NG-o1iqoV$V&!GCN#~
z6VTe+`&IdPWuuMtC6*q-s=JnRjNb4=AoEL|Xq;w2JyH8rH~Y65&FYb+OVi(Ni+^Tk
zHl0xhm}8pW#J(93y~G8kh57kf*3e(EXb#YLRu1f#a-a3p5{R*Z(et+X_fh3+>$URv
zwhnzwyBog@L7ZrJa(fz%F~z0QkXP8;eCcTZN5L>%Ozf+nZXKP7MIt&)^u@Y{k)z~P
zR-m&S)8dYW13ysvje@tY`O4Mfzc=+t|2TNll;|XAxfvCJ@{yedSwD|uLnj+6`h$fO
zoKy~q-j!?jz;1D2-TF|wks;=5-e={H%`B5cuuhZ5{F(pW$6*EU-~?wIxVCMZ4yAIH
z=!)gBYutF^;iPVl5vvJ*&<jllVeK&Y`tJ#`aE)@CQf`Mf^ynt^0myfl7dEBvCAhI%
zl_94adt?OQoHwD&YSC(Uxel`!r<|eD{9+Ij14P-WwDRio)o$is`uVQQ1#d=#V96y{
zoc4>%-igf2S|1|BXI*crxDqL-;+vKq721dNTj&(gv5&S^K9ehj=2yO`rw9#8cuayu
z4LO`w7nrOUlolarp@Zzq8ec(Y=~-zH`VHv#?iT?<i13fZ!CI?mu{H+^@iPnrTU&+5
z6EnDQnvOxf-tExS6;MFTyqu+GMTzkFcT`p$@0pwyZ2*S^ih>RB^cxTLWNe|!W?WNl
z4SP#_giRLW2jD#}Qa$jDlBDZnBYM$Ai!N&rDh={4V9)%ST#L7_DdN!4YPfW5XiKiY
zx!t^<4=&EqYV}z4Z_DnGJr>MJfZL`;yvzduUy^R1D`lt$pWJD{2)0-iC#}z^D5F!C
zm$jthoPaBk*No-eS8{hnJ^0L>A~=iW=slGp%f|<N{yM-b!A2Cl*0}T8-iSj_Utd2k
z@F|SM1j@w)KWV!je@N3u2A}!qE(|8Tcg-Lhn%_Xe&krQlTiyku+J$?|mMCbQL0L-;
zffC~q)F4>&kNNTx;%p?OYz>#Aw}?WvN&FGIJEOf%_NN&Sn|5cykVSYfBCZTnH^9hi
z%v#ZWd7>eeR@^S4n!^B15o$uij$BUq01%fE)GW&3Zi&mELC$0ngX~5&4e4(%bykug
z0Y>xtYh{V{OMH%24KKD)^?=>XHre96>Vq3}fp?{Y-wg~5w}0xkJ$i6fdAwjkj+msE
zXMnap`5P&|q@%!$Coy^LP<i;*u%lr||6xbaYmXLSaiXV@w@*E-{FTAXR5a@)J-`j&
zg%>`uNqb{Q7-``;h6p={)<*Qb+_>{mQB$htHa6Zn+Ex=)3^3-&?#bm}(R$w}*_K7v
zZ!QUbMDaDqT9aAQ^na#HjW?X!LrN!YFe)v`0L8@onoQ7(U4Z>Z=S2M3;_|i>n>7=3
z=Tl+#_rRw|VE#Wrlem%OB7G<hyekG_XJ`K<Lf^7~M=vWxvNGBIR-VUaEd{^4%qe>H
zzUm*2en}o!1bOJ_>R-3pvxKt|(~acp6tTjxsIYz%sW&Bp{Ofy~=ltYLSz_8_W9Q8x
zLa6QhgX<+jQgLwfCNEjpPoHuI*3eLIp4ui&JnlW-);OKBmKN32cJI~C!N-N7sBe&D
zWrsd;)CN8u0Qr{5U=;=>xB2S}LJ1+3^s!$dp;9tz`<b2v5_bUces|ajQWd|rh6k|1
z#DkYt)#LkQrKPXjC=mY4(|0JAWF9^+@E_BLnG>WHP*S=S)>^-CaRGKzrjEXEO54{R
z0}bt|nQ>-pB<kW(z(~y*AX&e4<6fz#EF0Oy1Rw#Z0hn_URwI#BuBf|GdHlIPqWtb7
zvNP8&!!#N<FpqHp315yZdv1PNeL3#)kA>);W;IfjE@TGzXmC=4YthE66cm#Av6_=f
z<Fh)oE|GwF9-GG|D?)hoP^3FtZWz~RzICS^`<yu`SVvifm*AijV6kXG-B}1SvG=C(
zK};qH>g?(1K|w*0e@?+SPFa$MX_;|~fJ#Z!&h<~K2zm%`4B8f7o1ed4j;7}>{Te|B
zO?ubewdvw=&@p-<ub#ky@BQb~U2Qo7ba=Q9CLGu18xj|1z}VDP7B63HRsYx*Ld>1I
zp{}Y5^@ueumB|OjOz{P_2a+F71I)#w4pzaR$uIwES<B-*A4C~gbQ<|JJ2qx(-(!)`
zWqFqVV+~gWGrkwa)ZA8q8T<k+QMQEHgxJo*R2Q{zF_OT0ij74NzZB2t04;yj7ub;3
zOa%wMKi>8@=Vs<kF*h2yBp6ekUh}0j54`(Zs70{5yV;32+omD$oDw^~TK(>0t0*{x
zG-0K3pr1ku4SlK<Qtq=14XH3yK*zS*FTy?$tAKiPc8ET<tfp@v1KK1wigw0&|1E<r
z`gyMWvgS)kFg<93!;8q*p)N9e<7uxc-8J+E_Dp#rF^h6(`G}dKTl)%d4UjMjg?vQS
zj$fap<=Xx!ik|t52a!7eh~MPS&&ZzY<bP<$nCbSjv$K;(&#1=xVn;pTJo%!OC<1`?
zfzCW$Lr(ORDH>yp&r!*Ybe)EsNLHtv`RhE#6M9BFa3+)?O(DcnYDcJ-#32{>64keb
zJa6wG1$kfd%*D2uTm5+0B|Ki+pC)`2Xs%rC9s6r6OsWs)yiVFc8z}He<haGDRzl;m
zL|B42PhB*Fo!#WJ>btjje^jUD7?ppzLr)j%K`6;sQRxUrms+vw*k>LSxPtvl0zB+6
z-lWK#sD@C7MfgGK&%^q9xJraii@8m}a3|Z=NK5;**en)>Km~TYVB5cE-2z%@56UB%
zQ;`Mp3?r{+bz@fi)rx8Vd~woRCg9K8Hcb9o{{LD4%b!|!j}1G3a&N4hhRZEg%0PnP
z{$)xA7A<@tjdn$)tB}^+-C5tTbas~7_~^3~C;_|n%tViPSfWi}5{g=uao|YbgsaFc
zqY6P_UEJNJ1T|v?28tmtuax?4GOG7$G{n1!!6BxkTI@vb56mgXTh1&<l6h;|miMJq
zH$}+uAEp~3z$SJc=i~?SW+*{&>l_XQNRe)jn-8lDo!Z=i6i_|p-JvTryRvM+7ZDx>
zva{E;WL=VS&>P1C3_@vt1ettAw>DMWp$nDEj{&`h)ygZa8y91<A<BAbJJwdd9!Yt0
z$PdoFo10x`26)(+iHxs`=JA3|OCiNgWCQVuQSTuU5>I4;O9%}V4NETLrcaf(O}_AD
z_g9!}__q8Ml%yX`v6w4=j@gr2h0+UpekFnY%r1(C<Fd1XqB@uMCArBnkd#!5d6V{H
zk=jXUJI0jnH=lTj?E0^>3d4r(Cywi1i!v1oZI6Ee3Q0NW9qpiHRTN{+ktKrHK7)~4
zV#Wk6>m=BzZifg#wSg0{sM?f`+Xt~c_NLp1vMOU{W!P@2HNduPdMCa0+ME8qYCS`(
z?v1){Q47t_lQQ7h=O^T3XV|LvYWp&epI_>=G+bt}WtQEll<vdt>8*MNAuKg^sNHv>
zV2dleKZ|I#Z&>Vr4vGCblrTo;EV|%kuE4C6kT_gE<Sy7EUY~`4fB+mH&eGC7APe=I
z@AYv-MFoftIsm!Ewcfse3%D#w>*_^ZTveb9z35P5{8AVOlcpD0bSa~C6p8mg$}PmX
zN^$kEbhE(|ow83{lA3r$I6qwC;*yh-4a8*X=pfOorSWTAc<UOvPalIr{H_54yq5sQ
z`|Ngtdernp=p6C!Hv6hWO43^>Je(<5I&JRb`ysI?EqL#yq)FX74D1WMco6jGD~O<k
zbog$4&M!SHxuBSlp7Uh0FrDb%h6F}eu1TN>WN##~xBVm4Tk)W%&p&#=&F$5+%eK(C
zHCt+zBn<XazYu9zyyYMyRG{i}*(Sz|$~3o7W?t?fkck}WM0eM*F9RJOVGz-BDouT3
zpot>SM$C)8{1aJT5u<h?0WScP4;d{{Q0Z4zIda6qBDmdb!_cwM>|_*Sc^KG%YO<ON
ziL!K<;$dqF0`CZ=1RC4dK3Sv<(<g}joSVCDS^dX>ALV!Knr&}qcX#aB(PBSWHNOC5
z1)nW4(oS0AD|tUejrd=LaAAJdY8+~{(FG->hzXG?0E<gRXaATx75yRFt(u1xnn!)&
zS)s2DOew&B?mS@B4C9rY1^0DKua*`9<(huEcA^M6<nv+)#qi{wGq)kV(_IP1IV>*j
z@>n`3endu8V=RY~a$$4x!oJMQJZ9+3HGd`gfGi1svQMtW=h3<B|MBs>*9))^c1`OS
z`QkYL&xP7KvB<9a*hmozvjO*FphNfP@z%@Q@2ZcieGr?Te$xje9%f}(7Cijuz?WMj
zI4%>DcPg6dtQ%heed>k`Y2jSY=6A$H?-T(fDnG{D6iG-(2yC1>3trCL=RQpSQ6<WZ
zQt*#4AlZec-2skk#T`})Z|Zc89N5qQHM6-Y_k9>E*fLbQ4)!w|gveNfT#^<w!c5mu
zEyOjKrvJpO+5i6793360i(RFRY&UNuDW@?6UVLXaNQ<5;Cqxb>T7;lQglvvtbLdd*
zA`sf3mKtJHDoMf{l@$JC;2v3ELo_Cma-8IN@Z!6r-~$GYMMY<4XO^x|LQj>{MfTjq
z&=7gv?nG!V(Hy<EL%7^+2{(2`aU?xwQZ11RDhk?w^XBHJ{V%gsan-I|Io57XgXlXS
zBY*zmM(JCu%}sbqvMMbd6Vq!hrDIh)>e0tMhAHXbHC*$n2%Pft?{0O%V$2}|`uaW=
zZZchOp~_+Mk?+gch}sG?2lFm4wnkpqiPIonf%k}@-kb)%{c6+URgA(II(D{tY7e&<
zrF)ZGFcU&soQfibXvn}fON=v7{%%d3LR?6}f<P@Sn!eYPhELGm<l_up{98TQ$58$>
zrFZW%y*@Q}6Upv^yVDj3acTz@CBRyfl96SoE9M)dVdCp2cPHi;J2QT&&d+D!5P*7H
zmONWFUtubA|GnO0KrieLPxmdpC)zD5gHwLUTGQZ+FZ9)+lgg3R#4|c#H`1Qo+On+m
zhZ3d>WuA-Brd37=chg$gH!q_{04;d?v|>{7F>^`^tT9}%i#8^J!PKchh8Y^%wHE&3
zcN0O22g75Wr3nN>pO}>uv19)gn_-b%SiL*zxR_uZJ1U=bp!U6)+8n?75nk0^x2Q<7
zR#EOOn$-Yy)0ccxBNkKS!#6B+HKbq1w*YygyvDQom%tt}RquP8gD3q;M)W>_a;qSM
zM2QIZX^~OIdx;9aqdPZE*M^1iNHBCu);qGY5)nX#uycXA%oaT?xnK=+BOkUHFVA0H
zIrDE-41VEyTUsKp{05$q;+22-IJcoU%B_sjIXE~7GRzOhz6gS_Z;AUq-{*{6Re>0d
z*Dtx<J4kJlvwc4W_s1MDq#I*tG3LPT!hei;SJRc3nm32dD4t4Z#9P{>+;3enth=yb
zN5o4*d90)-<LBLLdvfdGpIVmn)UF)*TK~QJ)Y#`d@)<qoouij8Yh^R??kY+#<P(%C
zp(g%ByO}mOLH#o`Fd1LB=7>EH3&Egat@&?=Aj&i<D%7x`SsAx42(0jjWVP<6$La$L
zQ!hjk)35!+&&aR7&Y(xQzWYwPj=M9TapF^K^`A)}@V_6T$^v9${42h}Y~zc2Xh$w#
z1}p~i<-&N_LYR~}trd~F@fqk5xrKRP>i>PHFX5*$xx4UfNCF+`dYm5|#?eDqy4K-2
zFW(oP5$Dvx1``+Mgdn$)E;_CcXaD74P(MlDa@4Pp$~5_of<zB~&VE=Zy;Qb7O+X)b
zKsj1ZyAKWy<f;8Hlax93_x8MQx6_|)v!CYkTf9EI+Um0){Ho|$X~XEERZrBSV`Ij5
zkmF}0ValzjN#i6zK`Wi!-b__VHa&`q1LsdH{M50apR8K#w-#Gjq*-CoxeE<$Bncr%
z_O>^ksU!ZpMX4<wE8S}pU1g#^0~GkjEn1<{P4RjYM7+wXs{3_Xa3-9XQ8{S;z(6|Z
zz^}^2hx1+v7_WY0jDUAaQcp~3uet3zMcPie*Oj{ka?t26(H7F(JD8q4WpCAFN5BJ@
zwkn_^2PI6ye36Y45`ux4>$V1(JY|NZtqv(0DVpI0y_cK^e<<m?bH!t4qeG#BV}zFH
zkzp*Bqp_Xq2=#0pwoI@$wvhE4j~zXR(RWspczVn(d9{w%1B2+b7F+e6)34qyD`oFg
zidWq=MgQTJ4Gqca$?c9^n~a7>I#ld_Ds<V&u%Qz7^IH7j@N`%GWbl@-B`Y78W%AbE
z)G0U}gwrtI-)ZP)Y-|Z+ZI9fVBv^`e#%3z#m&B_~lap)d<1VbfAjKA+HO@;-?=PlT
zi{E_3$XH*60#aoZeB~tV%QIdFJ>b87wUX(_v*NE83GX%EJ4$UxQ*sfmj$K&TU@lzK
z5LhQAn`#rL0Bo<XYT#tNS=z;S(xBwBR3r=v2%A~7WVg2iD{IGKpdGRwiL17EB422)
zQ0yy)QQzWi!^r(*QUe~zxEA`bQ}N%i0O+5+FW4Yy<#d}6?2jI~e%_V8d6gX4|CW&$
zoUrCjplY{EN_4+<S>m=V=L&K2E+#mG&H@0&F2&0@EqYmc`k?nNHMz5fncrZ8IiITv
zr80%nlk@d^)bJo@E_2^D`f2Sx^P$MkXm7+!P1v1%`wR~EqlVRX=up3sD>kKa*hC)z
zK7l*P)Qc)$C&G*p_-9!#<JbQ+snyQ(Z$V<W|B2Z9-#J%#U4=HYER<cEV^`0u=NgTw
zSdIpZ^_3a$?lnJ-!pR=hBn8ln{_9y_>G&@Gz-!P6k^O=lL*m;nOc*S!FJ)6pQu&Gn
z>@@TVgK0*?BGk5Ha>>96Vq3Y9)ZbdDWgPdo)aRG0#1~zWDYE4liFoiU2>~>e`r~^0
z_t3GM3z3xY-Sx@DZ)nasbaYBJw8aTew6MP=a#j=NX$sw{$HJL*FVIOnAbMEX=0eFS
zYow5Z>m`_mQS6jvbOj~VUv_<S@e|dJ-!(yb)*JAjWyg%dREaj(Rcjw^Sr8EFHtiN9
zG0`+Rm4ZWbgDsh$uUwRrUrUtBP6-XLrG|ZoFVxZk<ad1kSM^C#y}Y8{U6N$kdv0`8
zj3^NvTE&PaPLeQuNFCx}=XmB03YxbG?nQe_3Yx)!vFe$ZpOHwWWcmM9_SFGVE?@r`
zNGK(s0xO+Mhq5BwolBR}B_JV)gwiaHGzbzaERBFjOQRs&D6n*jgmnC7z46}f``+&#
zKmWM*y7zfz<~eiDnKLuzd_JU?*kEw%K=GUYQt9D&U3ai&-RQ-dgq_Tsi9NSv=vA7B
z4{ZSh6?VWAgt_9xuWq;1%4_d~KzZJs^I&1AiTa0^u$pNJX*#?bV)vK<Ioa^=BeemR
zvL|UjEBI!H$l9JGHAjvC&8YJo11PxQ&pd}S!{Q1wFF6B8b!fiLbo5;{R-hmvEX;$0
zvpH+#<Xf4djPN>vvCOn4Z;3wh6qMx-CH~S9r5+TJsdaXzh7J=J+u9AYRo$F?04cxY
z{nUV1fvQzOFlO--2<e&a=<1_RQs<;wodom{5Yp)50xQG<+q?Romv?r$RNuPfLZTS4
zaV(htUgzZ0=SSxy##VU_#E0HN8f>8or5(As?lz8eDfhArDGd8PirNQyo~F-;p+<-_
z*0YuIqdH{G5(Q%dXNgPKii-=NlLS=IC<iLaY-JlIW9Q+K#T(CTWnlS@FI>73^Xrjj
zhg3UOIbTNJZ}6km=d3eg_3znARKfEb`0VY<2=V~y$z#)1J2U*X`T^rQ+^-hN0(OI%
zwvHVjQ^e*<rb7b$%>aK=fG3#J8mk;A<=YJxB<9;_rL)4i^DKX!lU&O`AeB)-7}C?t
zALX3SBxP@(U+{_;GCZ0oZ02%fqM5=%TAtd1l0ri2{N>TR(Eghc$~ZYw%-QkjcD6sv
z2PLe_3AE_k^O(Iv9zr@9=HBAx;Y<v&KJf?@&wnAvVQ!%j&H&Y7crB)3I{Tw3$JSe1
zaD2>_q)I772kUxKHdQ=-_HfeRg-aSZz!Ww?CN!5+9nxZl)Mr+Nu0FeEKZZ{rAa;zq
zKFVf#@<sRkkMqy^cbi%jSV+c}hOy;zbZ%wm&dNi~QGP|CMFNtQ@V!_$*orx~o&73-
z5I`N$HHrl{5<Ft=9D=vy-s_lMoKKq!Ys-*?F)nM}+%EGpx|iGrYmu-$#rASo15Bd&
z)|cMGQ7cvVm-TBjGtzUDS|!4r)h`luz9|=&@hDT0GlgqBuX=>KbDhv7O3ELNMkKJa
z(Z3YmyjfKx1HdPe)H^3oCAXp+zf9!SU;FHn*(@QjuxFPvqyCmgx4=!mag9r*nFQCy
zq-Gi1a7&7JYHGZ_i*s9b)`m=9A0ssOs08+MCmJ@`jrRqWpFe;0PzkpdTD+#h{W^h`
zk%X!ERaaNs(Bx#&6mfJX(OrBf|9(Uk{Gt#$({@W2NP4=sAfvRjGm&$dNI0r)i`kd`
zuH~$owT#qzylP`(Ga<WcS3_>pBC~~b78P>#zQxOgvy$$d`YNljRJ9Dzp-|>eVd!6k
z5Xv*@1p@lBmOJ$GAxrXW*9o^d?pMOe@;ip8mYppuU(L7Wk~TmH4H6Uu6!E|=({Y%K
za_O}!LC#a}mZ~FMoRmlW0s=ptt*fLiYLxa&P7R0y_|m%tW(Q5Vy^YNh!*y<JK5YFA
z*7aOkB6N6!2F-7>#K%_sICY44mD-;H$PWC_ZACm#K1f3RzLHI(t};s+rYeN@&;xo^
zPEHPA3oZ?SByK};vwk+?9PbPB9)!Ep?j#5U;+C_Zs-YnxLqkUWH!Aum#fBl#qH(^;
z9Vx`uZ?DkJ89E+SS8*3IoY=B~JVS)7ejV=wE95<axR*Ld!e@ml;e;f;SQVIOd@BOm
z#G54ZNcr7;?U{DkuCFo+TDNZtdtKDfOh=|UD*i|d#qF%yDxgfE$fvRFRQrBI`N5ZX
z98gk##jX=7LHHFyXz6v?GK*eny8ks4Haw&mIxj;82_}iE6^e)ixr>o&L!|xzH?);i
zubJ`A$1}V8bT8!QK~3+X`vLpM4?hc)BCdddd;|+8EOh=0T0N@fN=&|Z|7}R{=cZR!
zM*4(noC<pvDM+)2dz|FhltIv-pxJ9z1+@!91j%^?pc=X%C0sHzQSqc8cjGD!1%Rr&
z0+17XOG%UNf0;`zUC6o_ofKauD`(vNmU{vhM~0%vl-r}O+{5YSyf(m!g$Uyoe|;Ni
zFR$h-AOH`V%%!o6<mRJ7ztw|QrbP%}3Ufqpm?)D#`b5Jy4f(223ZR0<v29(cG<x&+
z<I2b2T<imI)ne->(EOOWhR9Ayrts2T0w5h$LI;2ckgEwDB0UUJ6jn8z{UVo`yh>V7
zVsnoM-Swp<^CE&JgXa+cX;^7v*y!je<?SaV?EFyytyw0;NNTkDQi-+obv&By-}4df
zc?__?54Hu=s#aG+k`;1fXkY5Z)*1`bcAJ?cA~t#KQR4hgm1V>w(_FsS>@~kqvF=V`
z!QswV0^;ldmcVNUE<iA;6jGc0nYphyK(AcV?>cb#2Ae5TbPEgdNjXho(}OpA#RCI~
zAvAvJaFcG`CIqLUKtdXH0gtif4!x~S{tXlaA1mc1By3!uT{G*u#X=CC6=wp9Eh5U!
zu5v*G)a#}SdLIuRHZ{Q<i+#WCuAju6GtYf^_ZXFoR~-YG6+K+<IaXT*O3`5okfxk=
zi54F~5L0FEXIPqM^H?rY%iuSFL%RV8-UT!&((j-P8^cl&N2GJvO*9D!Ydq1ahSzyp
zB%-;YVoe2#@`o!M(h97_xFN((;KrthF><W&q3*T-WN1t271o@nSHNA3Z(;}Ot8eEx
z%4e_2-An}GSe31%#3h$&LeF1|vNcQCPhNKzS0DP7!3dl73eV4wM?ln6PtdklkUYh{
z2<chpmA*d4+Eb0}ZG+b%lT%;`^via4YkS@BMC08%63fpo_8)B;?9v=g^vxZwKka?6
zr;B><1SNQWxxZgs<U8|*vn8j5fsicY)5UjsP+1Q=;EEi(YTfOvC>vPR1-tqcJSWR8
zaW@1eth;FTn%w-cPfFU=r+eFDO%4{XWoYTqWdSnGHmu+CEPtFYvnINTp|s8MbrDwO
z)_65|BDg1=ANrma<ylM}wtWeaX2a;e;h+^o{xr+siHy{hvpLkOJLtEWK41C@3Q0#3
z-n?PO)tG;kn!j0a>RpR;*~@pB8EnfzSN4m4dfVLFPwac7R5#PV+2}*T0*a9yw!{ji
zn*|2AR60x!3mld>(}_O25q^tl=>99lyuq;-{X}*TH_hXphF_H)#b=`Ii~bHb)i*R8
z7}qn2LO-^qjVXLxMSG1WGS)OY(v07(@fYV;^;n&>m%~(69H!F(KH?vdcn#awe57c3
z4oMXQXv}x^)kH)s`_OLRe4NtYw7N*Xag%4dYWmY_y<%b`(V{<KbkrZf23*wetTt*k
zX899PsK-uMNn^|=(Vya&>T#r>y_p;M_zv8gJRb>4JRBA)dkvuKS{CO@84%+`KY0|D
z?7tStghBv_gMa|2u(^eI%qi6-SY6rQp~TMJmc36634CbS6jX3YQZe(3CL3mJ_5SL!
zNH&6FD=k_qIm@6?=dg5$_aU9CFfUVsXaEPSIhI5T=`7$1P#e6>&c^Lq<}106k4B~k
zMtRjaLl;iM2-L#|QAfc~p488-Y%SBr=J!^K7f!GSfvs=d)@7{D0V-H7{=Jr19kKqk
zwH=#MXFYvJu&)C#k3b^QEYRkWID@@*370u9L^v6qsEof-{p@_A@8?9{yW;`F53{Rz
zmjFXLVDDNRbH5tmeStJb*+?))UWQ$GJAB9IY#vZiV*pW%iyPS$KTj=V&S<%hO?kIy
z2PEh)*#J8Ch1SNPRWa&vFiI-CG|47zo(n-$HTxbFI2Wu<{Qhk<r6=O?!<D$5FSb5I
zrsIR}UcC%BOzf+2icrmkrP*1{Zj-}*niir1Qs#fzvm>as)8e%(xZV~#kVoTzl1Wlh
z&owq9&#Z~)95l!NlIn@X=3;F9?A!gIwpw?|wCfLwi)|c)&)kHM7-3QOV`yKE!pQLf
z<-l#ztALR0Em4RnxjUg8$Gfe26r^7{(^t~4x-5(p0aX{{t*j+Zat%)VXReBFqc|_C
zr(10EkPQjf`ib(c%E+W@`*nEFk%AEn4cR#5MPc`Ue#E4Tj^`xjWYs+Kzy5Pjso!pU
zVkoCKciR(eu?uSCp_Nxd?v&)&iIx|zUr||;_7;tZWOD&vD;kCm1;_;Me+|997|Wpd
zJ`lHY%_3qa-B&2XQ<8y6;raP)q9UWeX4IM}<=ee^b#?Kuw>ofI$^^Wq_-AgkFE`rn
z?d?Ahf5Q2}QN~B4P_vxkvp!UySwjz3tt?56A@9TfP|I@8RE!_0mJnbepD&t%ve9eB
z=Qp;!x+^-FnQi=A)7c-Vo5+zB?Er+DckdAR-pdc8maH1orEii%5m1%6vS5KK+jo<u
zJNcaC#SwrsQl~W1=OYy@1Ej+krzb}5x)!*y|Jn_(a&6~9I_IWQ=JRVw#)<nnlyA*Z
zx#2@$1XRxZ6Nh_wsM2-F$j+HuIehRIBz_e_PVQVcFbPFCwP(;#?Oge!=Ib9qJs9B}
zy2@uD$j93kP#`Mvg7Y;=R4S#L2Rk5Af2d))wL5a>fQ0tTl>`>DXjK)CzM$K4IAWhx
z^M*{cWMN0{mTE*gUa!0FENYCxbIM{|#>vRD-fK><JU9w(IfU(`7OeU+P#WuuSNJ)d
zM)30uNp}j}N}rW)0T8s&iD;MR&rA#gH`S@!T_929#jRfj=O^v?)#F!o?<E~YQ_Gar
znrGAy(wG^<Ss`eEY2hCF9uIAwY>g=eYRSO_WYSt+%}00kz5zEVk?eAae9nZu$Y`_-
z3+$QSRV*B$R2n|(#K2h|eWZWj?AzR(P={}<5!_*&;CB17JE<VfQ;m^o_#lR3*7V$e
z$}y)gIUAXMFNV{nq=_12g0*cb4g>~Rp^8i*Su%%j?1qg-=mfUTlR)+NDg^i_(D}tO
zYSAs4s~c_EGG}L(=5cZ8U^mfnM|l*on2=&PUUpoy24wIR=Uh>`&RUMl?T;PyJIyKY
zkk4)G2b?7XE|GYq2QwF_&Za}jAIq(eM-RKL7$2j=bIuF}C|z;8JN9mu`m;9vl=NQH
zTJ^p4GA=GS8(lrE$#k--&CYy{4uPj_ENN1N@9r&bCI54N)m_xIVRSfc7SzHYP9He`
zsIirvV#|-K<)ei$Cl+--_(>uwhjv5wMeA$@<|>e>YjOZ6h|XAs$$*ePq3fkhgjfKa
zY)Gw78WPMXs7uH7i!;HEjkSNfg!Q!&sooP*f6eS=K(0BrWBL@opMxAL^ElqB-c<Sh
z@F3}&Lh-%9H%+~k9N@x&FjmqT6x22P+Sm-jTIj^cEH5(bCUzRDZy$4>zfl$536*7Q
zaA|H%yZ7dejhnlT(av5Bjlm#E3FRDZWowAZ`{dD=3>lsfLhgI(IG+0xB_ZdfpXTvD
z8{&<}VY9&E0Kg{TW<v&fzb`K4mV0xYPEWxuGtb6e`1Nk%GBr3`{jz*-@k-dZ1AH{Y
z5Sy1n{6o|zjT%vE;1DTfeH2~l2Y6O3th&GEWMm8SJaULxC$%?7ov9@P6_kJ|#G9{J
z)9`^%B0?5oL<UJU&JPbKxj_%<aPkZ1_F*G6RFD+tD(lx&2q5E3kY_9|qRAr_{2usX
zYtc_L{W7IOj!<Y`PNG?int7H%h&SSC2X=^1NKStKAl!aw!{KosjR<sf^z-W|At9SZ
z7DcW`J@X)>0p9}%xFJ@j%6gYfNh{WcWH8D@=uUm)bGWa#+)7k#)?V;>r*P|BXOZS2
z`E;v~^}4DvP!ud1$9I))U|y9hehR`FMy9UR?$vHGX#;y~oS^2Y-lMc23vfB;HGb>C
zxg!U|g80?bM@l8$%@)g`ru1hpPpW}ujm<Mk#f}|KFI^B-PeV)j2MyGjFIH={p_Sit
zHD2|qZcHK#@qNCq9Uk=!e1GF`zg!`-QEk)!ii$H!z@HEocwSO}6Hv>|keEItqNG&Z
zf6Ge-G3%Bn-oX|1Bqq#D6XMM3MdOB`pAivYs~GdQt5V8E<#NCTG`=d-ow^hmhqrH!
zl7Z#!7rhKjpcZ~5^<H^DIYd*+%iS+6FSBDrx9+=5qt{U2J@$|r??K?_lBnp)A9a@@
zQ}|hrX`_WFEa6w7T@-Cc2OkFv!q6HY9u>w65jLwB=v95vTy9JHY$L{3*i9bJ4?xph
zNTO2c0c*Af$v#&tvEN~Uum-a^xFT$Q{9dsg{pr`(d>D({#ir@u<&Ms$7iTHs()exI
z!v;!B+M1Pc)`qp1H%}T#g=XF6Ks{NtxuCL&%lGei0kp{E<niz6A2eP%kY~A96r!5Z
ztrTZl#19~0Uf<AYHoyT?)F9>s7r-toWNMC`=w6BxoLe^yvF~E#4GcEYRw9XtzZ#<I
z;ktSmdA{)GVf#|4d8&kMUD@xi#jh$~8puH(FEuyAxd_)J6Cz~3KM&our<aS)u9;ZC
z%qz$UbC}^e^IVC!g9cZC#QxH;yn<R?X?|B$w$yZ;apS}%uF}u@w|XbwzvTA5q-%nr
z3N);^CU$s9r?j_?S(CQ*-jrBNtn@s$;)}=yEm2iiR<VoLLYAq=GiHl|fl3Cb=y4zc
z-hk#n?S!^2xF#)gG=4u*bdn$?GdxB4S1y3m;9Y@O#)M2$tAeDWaj%O{m!9%6?9=vP
z4M#A&nLe59BH96nm<G)r3eIr$>)P}r()4CusMEkMFs=elNi{I3DEfSlb*$dv5iNP~
z+YkF49kaVk2XtQfL!_Qjclp_1ZMY#Bv$NV^$!BhR`FcsQsYt$E|9~zk6|ok&%y7E6
za=#J|+xTw{S;Q``s3w402d+s;XELC!Tp<i9N=!kJAwP$Dk&}aqhh7%)dZ|CYrL?dz
zxqpinuY|j~z&<#tNUW{IC1@v8krTdD&C<#SiJo1-*S}bqgeNLq+nFBzAyeR@=i3h8
z%=@{T#{fHhU$w!+iT#1&^X;2dYEGEXaB*c@beo@__9CJ|%kO_=2~MD$Cn4liPI}B?
z4DC6-pC4;ja}$RUi#YLR*s~4Sc1^m}OLnC~J(6$2AIlKp*rLYo80td}?uB?IB2@J`
zzBb5bA)XBkri&z<&2AKX;F3fg%&29Frc}e-dKqev+2G2KH`ca!;9L6lr^-K*g@int
zdGFs+ZP8RXy%o(0>mff}&;D_irz=qs@(m|8>gN|lfX|`fn9B_dBq!muumi+FE{H?Y
zKS%%8?t(nLd2w#6)7NisUAG&|Rx%;^4%0K=rhx`@<<QSjL##6M5bsH&;6gY@0004r
zw&J-0R%PDi*pFly9OP6vfczqRwto#~jY68;$u+IQcly}Z_lEKp#FwU-wzl>LrNO`~
zXTtuouTy2769YF#wY6Ef)#%JMle7!V{F;7x&WIKnVAHa6THm*=EGBSj4CC)5&SYmL
zR^>HdDPz<a+aIcXOT};?p)3QdBpn9lJ~1yEyA$!)C8-w|lk}k<7#=8LzxPcN4k5~^
zNin6dZqxki9Vqpt!(p6BPQjdk`bAIv$S-g|-RF1h<(SFq=<ZbZL_{w(-MyknXT~qb
zY>9}-@&sh|$cd8gA=AKNcRJDxcG(6bKFmbpr1~LB{q822Ap9;o&YJuc%^9MoN7%m+
zi`*`D^qrUhl*XEeR>p0#tQhX$<6l^^vXUS}MU`-Eef#1()5w`N>j7AYhZ`;owAeHx
z@U@|nM`i$P*o6XzZ?ctxd@42_c{yQryzy>cI=w8+0UY0*5e`x02vZ5p0n(im>$OHy
z(T(TNS2o)AH(kQspgT+x>4;Du#J7}Syi`>M@3<;=*$R~#kvL21Z@3Ca<<F>w2e8^W
zv}K4Sa};-itynNsr9s7dt2JW%Td?+%eAZmSwy@yS(c~5BjRQH5b$;3N@G;o(dmBkx
z%*)mvm;`f>OA>e89Eg+jjZr#+n?chlH@$ha9zb>kt15}<!!QE3Rk8&27*o|xW}>vz
zzB&XlwE73!S%_fdlYI_CpieLH*pIH8e##SzBxhfEq5_Ctts03!ys@(%sNH+Q#|4vD
zlBu|;=cN5|`tF^TtGRn^TDZkGP2P=Bw447tU~pI0i0!oq@V~q<leu3jRxv!5Q$dOq
z9Nb3Qvk2vTapU86z%|tpE3`-k*z2B52^fT?S`W}ao|3kxI#ZB1kQgRNlCG3%UrTCN
z9Aho7*f4Y=A)&QGjYs8jCbUR`d`c;8Kso;2@-!~H>fWlp*f6eKFlWMhK20MQHrQx!
zYouUm@-@Hh_QR1<0CTdlycB~3a}(#59{0#ipw%v}63!>8E}TQm{-LL*T#20Dp8-#a
zNeY~#73$bl@7?>lHUd(z2(HQ<hbewWJAax-N&D>?H>+I&RcT6;BaeodL1No|2l<Ea
z(SBMLt_20{y7m04LX&x~RP`NL86SSrDeY+V!3u@k=22=7W_*Ie0!Si#VnkR{QmmXo
zt!OuTVsUG0&4Z^enPGgH3VDw0boN9J*L-asD?ht`-;psxDNE5(=h(SRs$g}jZ72WA
z*lKl7eAnmZJpo&kh08acC53YBihIc<5J3C}vYUn;K!R~TrQV|mRR*LYQG^ZNL-oRY
zdp+{Lwdw4orwd)AvAQ$F_CB4RIjgu)aW~`RXWyj_PL~~fWo;y~M85lx3Z-LVNxARb
zBB#-pbas+(7a!cEHu{9hKApwlt$F!}!eC=oTHU70FFR>&%N41pAo=<Y^V)qo)UQWn
zz&p6|g2C0Vi9oMnAOp7=uupKb^JXuJ7l9zg1*Zg?43PaAEKO`VfdZQ)aMWN6P$I8u
z5jfUhCz(3_q?U?$NciiQ*QJQKn-Gyy$4!qxP~<ML%rM?SwVavh37v5u47oN{ZEbdk
zpSR|rwivb$l4(r<;3~yyfxY#`uGq&TtBF1kBX0S1y7r-4u7^FstAIPm4T3A~FN(t>
z=-!7Qzx{A0>1bJ1m_*m~&0clgcb&-0crbX12qdKeR6SBDy?u1)G_lL;Ix@(n*ucLt
zVEZYcGyvjN$Hpmtj}$mp<PK?77Ikqf8E<EH*C~$6bJUu}4jHZ2FYM~`J8N^QKArWL
z>$hkUi?gCu>nI`Lv1`17qe;!eXaKdqtE?xj3;i({{PCsmg>bJZ_^#+5j^!Q?upWJ-
zglnTg0{%$p&CE|2kAgs6o=s|!HfK0LX==wFaPmw0A?{k3UnK3<v^Rg_-Z3#xC4AV1
z;AI(Qc|h-UgI)#m6&iUXZcV%RA{9wzRu1wPa<C$TBSt)J=lmBssZUNVdo_7U(OtMl
zN7sxkl@x#GiGB49yfW^$Q<bBtu%xWqwi&nB)2Fo*8=0wmtGwwtMSXs;w%f|wz}MRh
zl$64Uv;g86fOxmx)^=(Df%@t4nvuteAvA@O5N0jk4svkZ=D_~%{iXV$z8OSlsG}9}
zGWB*c^;9Cs4NmftGVce3cO{=Zxpocplh~i(3PJKBKyOiMpC%!EUf|0Z_sZQ{Xkl5a
zg;P3^p$1AmVvt6OPA2CqUh1HFdc3;BHO?<>h9C2?JK^;ayNTV;=Q-X!;DqdOv|97z
z4zFvt1HYCQ10E$wPWf>Cl5-8nVDi7bxmL5`;Jw7ZmWMg7)+~qfAx=M7DD1sjdGOkT
zG)jQi95sH`#Mz=u=e2zQXJ?fnaMCl68<H+CxWgBaf-C2u4HZxd>Sv*_-Hsyx*)7x0
zbrSs6#ocBj5sSVRQF{3|qt`2`1c@fBK&nqx-A9tv5;fshAO}0Q`W41nWR#d(y9`zJ
z`Q(E_=UMbg)0o-xpcU<y+bP_WgS?IDIAW5hz2yqlpPsV^uM@xY99ypP@Wr7M89$pb
zhg=B+XOxrdX1$u_YJuh-Kphru@!5mjP=5xgL_1k}$XtJI^-Q9m^ZoEM&IGWr0%0y+
zh%i>{c8B`qiHX64tyzg|w=bGKdeEJB=~*$VqaUgOu|aDHAhDlR3B37}zfzh{`;>lm
z_pw7nC44D<HhDEdhH0mkk%YcQ!T>5{14j`D#Z2aS-^Q2jzW8P|wNm<Pz5EFkR{gn9
zuMqniwYelui=BtOmXV$L%S5In-A{3^YU+pcfs2%$TEM95tVd2#TAFmNx2g(xADj0c
zW*|l_WXwOT5dN1VqUI*7`I%v%8lSv}5+gJ)*UE#M6jW)8A77D}edL@M>5oIcr9Fa@
z=PMRS3tSBWbO<DbN0+F=#ybGjZp`a|3~G=IsVXM&M$JWtBpk&-UO;0#oZz!<R#`p&
z6NGH3UlM`e{XLYer?w-P-3RC^tFMT2T+4H1K6#i5Xr}$mpCzs68VC6uC6O(?8lvM!
zXgdtO0a$q)Ko+&d^y_ZMni@%?$8K)76d4yl0)m$|w4b`dp_U^Nk1)*uip5oTW9IyR
z3tC``Y<qDgAb^=ZRaTTebNizA;TIvjp}%{_G1r#k8o*1RxJsB@0SEUt9Mil2sL1ex
zg>tk$I}8!dD<rhcL0(QCBH{`Qg=R8xkf$ZIPu{AT4egB0bjte<NX>9Vwh!lReQ(Sl
za_XoS%B8Bd-O#1qrak}$A^3+tpls%&oaD~~b1L|+n3J)FCII4Ai+AmTwsBQ(2ft3`
z%iY+W<9EQa@!q4BDb(LwRT?%}4cD|ez1IW)0JonmEEUK?`{+e!GM+R0s3n<7J@Y8k
ziI$#bP0>-zdW!`Z!9IT`<DTl#EWcX;&`22|iPxiiPOS8t8SPO$j)4pno((aXTNJk-
zCfx0sybvMLDew#^eOVJ_?$P|r;nIvgU=W*81*$7>KO2cHB2s{hqd_q>U6T2!d3R9h
z@N7%Vf#56fdXI6GWYVc;#_|N6*xV_p%KgM*YM4{s@!Fz7lY9%;+^hnMIJ)X?Sin^H
zK-nKF2)Z){HR&=0*Zmn<wV?p17@Ha$vqjoOGct02l|NPC_uj<t(eT*y8aLskt8zlO
z9WW7uel`4k1#<F1VQ~6KBfYv?rHzPYiJpx!Hf$x!X=&K7g|J~#CEP(xJE7m&ZxlPE
zm99#JoEK9@j<v-51jX?acnOTu3`7*|ZrRvL+Ot22Z{Asid8!|mH;C{{rByx#t6j3Y
zl1!k}V|((<3*p8?nQxvsmZsvJ5y-i1`5MCc`CT1nJq@bdGPoOy+GJn>^s#@!-PE$e
z3x?*m@8dO|pll)ybM>KXlJbcH(9SKuTQ~sq<!UvA#_zt_;$$5hcw9C6=#O5o#VwhR
zkNd~0E%(qp{7*;qKW^}KH@=LbP(l_dQ(8aFbb-+-&4)^JB*)FW30?ZEnQ|SkI(v5Z
zBSo5CQJ4^1BJd7~o^?E2M{Nb3pZanb2t@bpUp+cn*~tC6!^k{rNQ3^Xr%&6R^EuHs
zUVO?8=?)S*ANWFye-~7H*`gR>OIn0D7I(*jgE^87(c8|xO_eW}LuX^6xjsA~Q<Aaw
zHG~yW%XYfJ@<U7FK@pdOp>z_zss_o*!Eq4>>Y=Yk^~g0ruv~`#S^)QT00XnKrfGSv
z&j<JoOH0kkjc+NUR)|Rh3FjMxgF3>tnnpd55%P&&jEL>Md7KD=$Fd`#%SL4F-ep-V
z!^N>kQ1lghU-5cfiC5RgZm|C*0fEgE?3aMPQeQ{z$jXQFNtZt<<XIUU1d2btGrJ}(
z-vL0u8qPh(!~;E=x-4YV)*qP@JC|LfqS=97PHA{Q=`zw^VR?+Q@%5jRV}{WdM%<;5
z(;6xw94?ROkc}M4NfxYqAvju6<;9;2mYa=7sNvD<D)K&&kRjEy?*yNPtvJY!V)zy9
z4vgNFsD?W4{aPimRScUmbF{E9njZbMpNQXAl$8H-VF|iH@@rNcfs{N|jVtRgjlm;q
zqTw0~9GF^WG(G3z`uK9cr9fLU#Q$KXQ-MXlfdXTe;7hPP8xA`JH)u%IDER?ds_jLv
zHXA*D6oVW`^1wRqYRZpMJ=o1?m&eZiw?e!$NEYk~>;k)rkl9bKlB!ie$+4)TV-!T9
zu>op%M{WmyUt3}6WsrU|{&8UKYx1wEN;+_EHQuaw39x>HM7N1)-0R@LB2;fA*kYg9
z6+@e}vzT(eH)qy4o@{~YiN%$B{=nn~XZvbWN)Rz*4WF$R2?~g8+7`hHo5poS!V)Dp
zxnyzhO!Q)Rv5lLJxiB@^t?Y)oA%tH)>9Jd(<Rd;WD`O=i52NpLkbgQ~dVxqThsUh6
z2=F{IYn)4`7c;f7o|v+ikgV<NjEsVEB-A+H(bh(vpFC|RuV}c-d=;lbktqd2=znB!
z`<=3TRsSr&tSL|aZhoG!{DuAL^m70SLjj{qMqb8xvA?0;;N`BMpy2AdyW1z$C-y)@
zWHT-DSBeTp!V!W(3A=@nFZ_CdHIXPU({55y+|3(AN{I-j1fG<<0{lXqug<N8LZ@&w
z5<7k1s2EdPCG?=XBSOjuOx+EBbU^cwEox;ur4PdtRNW+U#ndvSs#27nWUCH9?{U?x
zWN%JYZ%tOM7ZjF2+1Vfd8rHI&eSsQp(@ZG^@O_6mfYgApecd$$?4%~lzbZEtKs!IZ
zv_R?Il=%!cn7!g3(%BFQ+UQ1<(tC^<=V=hIH6NrCS22e>W96e_Ws_6xYV&7yb+?Y_
z_Vwt3^i-_<AKy<JwtV{5e`UCj>Jy2b{`@s&N)y&I*(06?#!mh^5#N-;ex&T1{vv|i
zxtsGAqqDfp6x*hWyQ!n0%a0n^#VMG7Eao`bRc$B$*q~#=#R&#*4RBp062g<t(`AAE
znFfA3fFnxzY-7Ax44|WhkChmoe0B15z1En7UjgT6&qcP#!IqG-pjzn$p#cM^a^vC6
zx}(-7f2|cls9p#M(5uh5Ska$=G2g6vGiI{z!_tE6q!ei_>HN)o%-HY)xY75&E-NdG
zB@8e|AA0-{aMRB4*`L_e76$pYm`rf<ay?cuJ9~R2w1kI;2dGl~($q`6BDS)z!NJW9
zP|_3>>Z=7^em*LDXSHAbpn`}#w*R}wLArY)lPL)x(t|=Jv(ycht}nYyE~sU&6X9Q<
z&&ko)PKk=*jb`EUQ+x5^1<1fXO=%XV6}>0pJG)>Z{4nhpu=RnoE6xkQi3O+3?*P+^
zfFLNnSHbMyJx+>`y0P(_^X2rX9Tc|;E;-7Ak({>Dob&MVHecK!49WCbLrpm~887S7
z$99DxFd+E=eg{2694V^>&SOJ5J3F<`-S@>WBw&#Jle1taUo`fQA9XeSy`i1+6-{4V
zS0qk*_|F&^K=FKCT}_{#?`XBPqlo7(?H^+2QeI2%(&p+s*r!RuqyakULSI|P^WeHI
z(VZ(HAvH?eNZ<CP#<r>tihb$Nt-dk)ootB0mLVm`pt|EVSow>IfF9AVt{)BC9aPcO
zB46}dZzQyjj*J*TUj$^7*jO20=o=dweLg>20XBM4Z>a@<z$@(f{5eI{Gq<G8;(abP
z)~xwC>P+ax+lx=DZmAOug*s@Oi<hnFS@Yie6P_nx;?IxZ)+|_Kv$O>TRz5G5Mz2KU
zyn4B(OglUKT6S>@ZGJwuou1jpV#zAX0|}l|IYWFiQdUvugWb&dQgpFdaq_kR|HVOe
zZkxo0$v$lbk*5;V*FL<E*0Kowl{&Z1yB<BKNr*Bixzf{=LQJX}9t%&%LtB9Gkx)-D
z4IIhV-g8xGWRnZP;F)C5+}X*k9=i60OTD1*X`jf)!#QkyBwqk{eX6+tD6a)Ljs>(I
zZ#)bd9_$j70P<=iHrDa+quLzyfDqvA-s|;RBp^P_&N4ve>c(lwGg2vU5Y5-`E{RjL
zr+1v!dEC0%=>dw3Q!X@P?|_TQo4rP$#dJuzYt%>@|5knhzK0u@xStrvJ(t`=hj!8#
zXw$!=^ixr5-F)X8bh_yfgXqFzneBaub_QdzFTXJGx<z?r3S`fJfusiu0vRx2hI`=a
z8l%Ooy=)Y9m=>N&<L_kU2&{jn_nonIgM<86y8h&pexTICMnw<~<`obb5n7@25YLr@
zVlNw6T7Cq_HC`|D0~z$wRomtVqu0pj5y-eCmgkK0T3S|DaR9egxnGZ;s`<tDy4kBZ
z7^AL7w#Z*9sTJzI9@QuspVYU%)FaZCAFM;9&bO4Ag!60Lfy*{t1~v?WW6h1GL=$KK
z0DWj;3kR5$(cuL`0y<)GRYEH748qs1;~Rg1u%U!j`rW(A+(2R4>^H0B&LFHpVghT-
z`(Tx2gLDRg%<C&evz0?uFASIN4{Ow|0&Z^Tq<W|fOn!rAH*;v*?|=)|+zh(FyzC6Q
zExajkI_6i7pE>Ld3lsnx%y78+w0Os9MMYMI0P0C1F$rN}s|47CD`CXijE9lQZ;!R=
zcFXh`1&k2*-2D$d3T=`J=e~-BH$VnX_7fDC1|Q6bFpKdY!qmeD<shmo)1Cpp%9IJB
z<!st{Ve(<l>Kx?INvkm7MX4!I{g2~fOJaHD7e9>uM-pfl*PR<?o0AjMF*S&Bhv6dR
zWdn(DNBtrj)YX9a`eAHBjHJ))SHY(5&mPGD+JdpM$dn;(2&zERPYi4CK+9?%xe8v%
z0NGJpAthCIj`>q$5C3{4YJU8OO30*voBO3tpRi<{`$3sr0FQBVGd)D;Uq(g3K`vV`
zJfh$qFbVq#?=h{bsu2!hxFh}eF=Fb0DMqy8ew+<|s6o*0IZY5i6-s#sVoB4-ml(Mz
z1+9c)C~J+JfkVFpz5qP*;jhVmoT!)@N<wI7?+OxgGG?qwhFD1<cK<ScJW5O|m<^L6
zjdGt<*im6xWNO;2?yDh`7`;6k8w4TBU#bB%9=NTotei+;wQ6|2zZ63Y+F8Dwl=%<k
z15)@~65!xfP3Kl^?TQL)B~g*jUy6`DT-*F2`Qo>w{*MBH5wB`#1%Qe{cAb7xWZUT?
zvP2kN)(g?&ZD|&YEcjap%triw>SvFpLT0Vd>3;Qib@Wc0zEV2`B4>*(|Mb8081oze
zROr0xRv|T{Nml@OdrU8d^VdrHA77=A$?Z#GdW2Xid{+R^zgP9n#ApTc&Hrgmm<}w!
z3HldFfiHdrOCKB>F$xWS-2{<5#)N*Gk5wVu4H-Cg{hVJfUn=s<{M?oDWNZ2#dIiL)
za-4|h3n}N7&gn96ra0>O_r8Fx0p61K_B&MO#!bJ+^PdvdLx4(Cu%nfZc~iVgXTmKj
zH<7O=v>$!cXS>n#;&wV_EA!W|?sX(0G}1I~q5HuxN}}0C)gB1d|FS8dIWj|<u?J|L
zn-Fh898w7UPlf*5>M+&;2een6E}3d&@Z?$bf1gU7*KA5bDvmiA=kuS2j!|E;0!8pI
z+kX`rCpj}rv;Ca+;z8t@$JT$0t*9d&nVr_EVbtRS6Zl&ne)~pDtC+q903tJ`5SgTE
zRBy<_vbjqf9-+`byf4PnU>=pMh8uCUnGHtEYD0xz9GESBiE#W@|LMQ<3R`lD7x?13
zfSbtixrg~H3q~gUrb^uZ<M=Sswvvc1LHjbB{;XP>J|FdYT<4Dm|LL+Yy*r|auU-ah
zdo!BwK#)*?dF)?JvE5F<{CrN{i<AT$Co{wT(!$=<O@d_GKW7CjDtd@_NKN&ZpB1t;
z+5%?}+qq|V^~-!*Y5s7l4w!Z0AF=_Ftcid!F5$cTm1Y0AxcvE0VOHnqBB1*ErtZvB
ziP4(>?RS2QI4<JP{|mnR@6`e$X01g0b-L7R|5fwwu|mu{K3Hd97L0#Ui!YjwbN{dR
zLLiet_G~aDV&Q)+i2oR~2=y&+_=*_<2x9+J;n_L><8yEk_813mz@Mi159g$ZC`aDJ
zeOu-nTx}=%_+Q<Uj=4J0%*=&K3g_SaU^$t&Qjq(ZBGJD{oYxY?(;WNX#Icv`=PP=c
zlS-hefO+}<c-V3>6{U7G@uX83uAJAujr<*wz6DSr{mXg+jAO*6;}k>}q?i}~ulKS6
zM{FSRD~hi<791k|9i9ATAKBI@fqPj<>?>DT+0!N>Q+5A)AJ3b?Ig(CM@&sq9?!Wi7
zq7SW~O-`4VXzp<wuJ`}LQm6ik88GkM;7EA$P{TyD!cbu5zv$&a#F5~Avvr*!6OoEE
z{+a)xn&6_@Wp?|~*VL5n*+<v@(EWd`g9zjufpa@YD}M%+|6+4{X6JwgF@fFM*>H*O
zPj#xWdU3@Hvj+Y{I#{ry1&VKpN8kzsV_$AyBgV@<a~(W%<4=n-69$1EwZnk{BFYx$
zBNi56mx8ne9JsU%+$d1F*!SfCNa7Z_-o&gMBW1}=$g=PD7`M^~%(z3aF*eV_%*@}9
z=gN%gpMLVU2fqO#g~{JfR>Q66L*_fh&bNn$hZ}uPsJ(wSBM|slDk>_RoSghWN9^Xd
z0ronF8EJA69d)#tva+Xe1_*nO@H(Hq5`VdA5<YvEy>#s16cwM-_Q@F-X$Tk{Ku`iK
zZZy7!=qbm#&9%Z(-;1*Y05|(a72s|oeRJuEaVztGo}2{*1!2LWu7|X;3F8Jyf*N$T
z3(W50k(XEMriqe=6VILWUtg0uiI|z0H_9k>e_fhtNEF4rdU@54&@ZGaG^HC2!v64T
z$M46xnHQ(QRE|8m=U}IPk%zc~Lx|bKmFsap2#E*?kUt+eIDk@ljzv>zqCJ|3sHnuG
zq`qUTo-rH%R!&Y%UeO4eZ~pnySzCLg$f-lv_tUddBvC<O;m}3;Vt0}u%M}QuQ~zMq
z4i<IN&h5Jm2poj>$IaPJj{!;G0|9}(s!|`$MS{PUNEnzyyiN}C@ZZknb?o<(MQ*<%
zP+9Ki>gp<cKkU)FLO6s_PDVz?&JKX3>m?u(oQo9O6=6C^Gv1<z7R~V=Rb)CyNp#f1
zAIV4D<d3z%$^>>bLE1opz-EN<130!h*M?p%sd`|4YROi#orH7-{{XFQC$;~c*_nn4
zT0l(xW(%{a`qS_+p)cjF-D<MmuiLNc6GH$v;`bgu6z)G<{BKnPY6Kqs=ZXz9RHqe4
z<v*i!9;L+#r>%4If`SWJR3^iL9rKMEmo1sE-Qp+ix6^{kR?G}YZRu?WgM01pzfKfb
zx>jwBEgc(=zOT`%8r9(mVY?QI<mb<0GMI8Vh6bW}d^dDv$+?XgOVYP!tW7%*PQTiB
zmomDai=NkJtW5X$L{^Q7aLe0%+r2GQpWSnzKj{(_pddr`8o|m;uAvMa4Alkyy~=|M
zHPa+%&!tBw$WU%0`26Y<*&)i%>}0q!FG_wsqltBn-159{@E^Z|7h+}W!4t|M|Jh#r
z(R`H=`qAIb9h}OLl(O#nC4TtFub@+bIr8>gGoinF`Ln$qn$ksloWF<nahV(8DU)(h
zQTdNwL8rN(NoMk)$A9(mM|+m%o}9X{--E_9*CFDk(Z>GeG!yxcUol@9=_$jDN+$fP
zmp|Lv?X<6HY5pw(Omjbu!s$&kmM%D+{Nq>9scjoOLZRL5uU`IaFGb9xo1E~sOfk(p
z?K33<a=fs%{l~8uSu&UaslRLetCv68`*ON%P~!aeBt*VA&kog%KN3?i{Kv0<OpTc)
z@7=$8`Ln&H=&^rH>i_>#f6$|uy?C9Tembbcy*bn5_pL7`CgxGaglanM@hpmG4)`NL
za47n0L$8R4?*YK!-5fP^67kxz!Tfb%a~~Y=oc%<%mWy^R<QA;$EOrx73*T7}P&Q!K
zE!p_-qp^hrlr<-d`qa`b9~HvbA;61)Ke{_Nii?X&o6fzQoJuw1T5JHt4k+1WFrfqB
zLPVG<PW+>@gBgGq9;({~9F2s;rR^^@^HC3@AP;V?Yn|ThRsmI+kLEOBSXeJ%_&8Wt
z#ByjXEUY<J2o{!j`XYX3Je$@#<??TJ89ZRdJb@Alq;dB8a`QDNi<xK-RqACv&IU;f
zxw*M|nP!7vs{ee773e{G%Oxx<<~s*Om;R+~NYE85tk4{1(AiTmLXdFQYY;R)jMAEz
zm{85D$j&x<^oVw}GwcNr%QFv;gD#rRh@_rQG&qJe{pkGI#l^+f*H=sHwQsCvq^7=p
zN=8OTT%4wn(P=k#)5`6zF}+xM3xE^D;Kyi^!=e9P`mA(r&g{(WTU(*NtN`U3K_?S{
zc+7ZHzshbToV)SPI%#`(R#q0EYoO=lEw8E3Z{oA6w-6<~d-pECRS(Sb=OY-z_jJ3n
z!K75SzQfke&ZX`<pl~o}B{Qh7Qjw7flC8gd|H-7=h_>!(bl8|UrmWx4(9jR_=i<|+
zm$!7c4f35KLT&*6T*=S-^l*E<q>k&ua+p1a@JmJTyG6x>EPiM5cH9wcW4htR(az%W
z{Po*h`302tL7`);?(Xh})s7=HTN4%5fNhX+ra`u_uyFUtn;|H`Gm?*oCnh=hFfaTm
zS!ob+0$}cW-e7W8KDI96@x#cWk917*cqK=iHYPPy$i=e9^RqVu5gzWbHC-$QjteO$
zDEO^=ftUc9;l_A50~x;0uMypZgaiOUqJ8+|9-QZn_wi~TSV-kch)Xplr>36VpURdA
z*VNR+$H(u@5GicauD=C^G&D3Y_?4?@Y7WlM0%=8U#xlR=ooCv8n#^k++=^BwLpC=0
zey?>0i2d&*>t%OO&Z0&3$5Z&N!CaEf-HUo>OC{*&y_leD#5L9>Gk5Pxd-!dqy3G%%
zsi~%ZXCv#s*4!UN&D}ehv9f;)hI)SPd-fCt`2e5_!@|l%1}%P1n5d7BPnw{UsL$~p
zkfM)hCg`bdxA)<hub+y_*U2h-LjwaKPC%)v>gDC7p)t^<HV^eapkxsgtj@}k@T+%U
zAKlyA+g<MK)TqBz+Zvma@_ea>;_NAA1(gjhDEPZ4LMbUJV1R@n^Cw$PV^hJB^&QXW
zBL$~lY>AvbH8Gh66O}W-x=o2JFE6jH6=Gv!<LBQUL@I&X1QykFT9X`s`}e8Ee2x4Z
zs;5YA-h8K$c1vx$<2Xc~SrzCfIJQnQ^`(fE1(y54X@(l;XG3LW<v<g1XACocqDKp_
z!MUnZj!m0=y3fh}^M+}0+J1l|)5Q?YVN$QhU~;MXk@KDtinx?+e`{vTrN$S`VRsU@
z*mfH!<|OtHg|tcC2kQ6Td-p4KbY6YEx3e=3#^!RahwO{ZG3{QAfjHwRPYwGvmr;M{
z7_1a9ww@mC-X-irkh23t_!Zt}Uo;RT#b7eO@vX+S*&mxLsj#wA;LeS;+w)8q_mKQe
zANYXL$61*euTn1gKfdMv*nbqpTeSVVBl`b+_~m<t3BJ1TfSy2GyK2JWcz$wnGCdu#
znk4#sXJMvEQ9~nkj{-{>>$d~D^im%XM8LJRkGw@?G`LLa$G?3`PfKGzMqC_1pVj#Q
zV8*>0Sj<=+Fsz8;s;XaEA=H{=iy<_=9~S}5E~nvr`%#X}K980Dtor);8rS8<)1B^P
zU1||8H*`x-g>@enxV!N?ug_)$1fduPR$x1EuV3HhHu*L=S>U$_^c3U2!AYHg!Q{};
zZl3PvzCP0?--fBp-tO*)E(@XAtV~Qy12zKe?DmfyJ=$IRs+u>n>Lv=>;>ElADdDX%
z)}gg;(*-(&=Eugyus|^wMphR3)2l1o^*<#UZ?0}`8aH_T0ubUt4<Cw9_XQJB?fF^G
zH2Q#rVtsX$jg?ie+VM>`D{zQu_n#CL6aXX*MHLmFy#cw4Jt(j}cSlD@o;yb6W<k@?
zwe|HU_YG-@ZES2{?rU%*rS9q(-#XxN5)%`{2(UgSyl+1S%ZE7lmICcUxVCYvn=Ru_
zV5o#4=A%XWO=tU4=++Qd7Z(t!lgmuJw+!!`2Z!;9?1iNz&;%5*Gpwc7@N3i%!1ZJT
zolU)Gzn3}gd49AkLaheE<Q3f%A$L3Aa0?0mf*!b`qd1G!)Y6j5QU{Ac-}3tUdfG=1
z!cVkdNx5)X>PaamfFj~zVrJcheud-V;UTi!+}sY&&OuOqmOZw<xwWNB25>)(o6IUE
zgsIhZb#-~}0MPDZnWS!HK|y0v1ORN2S5}5RUm8f{(BJuVg~(yNjE)TdJAljwXgJA2
z?gjm6*tD+{us@MMtpm2?<3URd_6lG+z+c{oy!epKDt2*NagLqDZTbVqRSENFePyMe
z8&GKx5)tL)=0^BJ8-^lK6%}CB%Pub71gM7(K$rlogvO~Z)B3kunnjS2lLO_scmMwR
zb|=kLwUZw}=h$0kQOhVQEv2ER_63w(;9Fm(r04^$Y-sr6cyHB4;CQ2=uf(Lm>r1@f
zugtg2&CS4*0$<fsS2wl!Y|7z-JR|*42G*TuOb0z_f?I&>ft4E+-F}aZj0C%=EZ>H*
zvY#M$@Tf7$^x4-}QHd-l7&CeC1TYop>rbRVa0OUAK&j5Z?qc{^f$;<Xoen%NaDtw#
z1cD0-3q__)COaaRc&IOE2(dZ}OG-qYXFmbHOUQZv7SHt`zt{fLM0~9I#pUI`Ofj+1
zJgi6b${-*ShrGuB&rsz5*8}`#Z1IP0Im5Aa;I?mBiB<u=2undmRk}p-k^lb#%@!G>

diff --git a/docs/dev/articles/examples/timeseries/timeseries_classification_from_scratch/unnamed-chunk-9-1.png b/docs/dev/articles/examples/timeseries/timeseries_classification_from_scratch/unnamed-chunk-9-1.png
index 45d0436e03763e43c399a94f7c74d867ce8874e1..801cd19b824f12dcc04fd2ac85a6dcb16de37026 100644
GIT binary patch
literal 321147
zcmeFabySvXyDg5(T4G>ff>@-20Z2%wfG9|ZG$J4k(j5k(7^rkhcZh_vqO^#_OSg1O
zNpt3dYoD{v7<>HAH^x4HoUwoV{nq#v@$$USbH{bfIj?!$>zSmu@J{k0<YZ)IJ8#|)
zlqMtF;z~wFA^rDX_{oS9w><u}<*ul(AlVx6zjygDo@8Xl$!-c>m9+^PZMRnO99UiZ
zl`3~-(|+0e%EI1E<5Jt1bVsfvSXR3$jU?ul>Z>N@>YE-*2rsVr87sdjsfW*<Q_}R^
z3vRMKf>*Cxk#}<LwVo>D_#OOuQ{JORyH(0$mEYbHTheXqckatpVcCe!$Lshu8O?)#
zz9e(FQugzoFNa>2Zus+M?Fo;6-Wn-zZ0nz&-EiaC_2;YCyKJfce07|9iu2D`SGapk
z|9s_Q@xLzUcQlub%;lJTidw$itiI-LTwdg$vT|!#*$v{GAC$OIo@i}d-M+nhU#M;-
zBV$lCG%|Xr`WM+1ip1Oao_PJ|ZsGs>GyNC8V1Ee<#yroS+hXv}y?b<qh2{g|A|mcH
zvks|kZEaN4)UIQ{e*L;|;X-XAu|!nYha;c*vembi7Nw@uK40}?d=3hWb{(odK0aeD
z8T&H%9T(Y{nBvPSeOXjn^6jlPmQo5=7u&5ny{|ap3O3u<*UO8N&!4-$`Lh<yIj=2s
z{w#UvS#S)$z2UO*uP)DNEXpg4cTG$thKH4sRk8=FLraT``}+F4ySn-+eN&V&^j*m*
z-@bht!socyUlq8QYMJ-m-Mi9Z*IeBx&$Qt76&3vh19k4*i;jsYbXd5^&Q59~>HVDK
zcXf4TWMnKaFIQAlq+MmOva-rBXk@vatYT(nW@u=rp`r1Ne-oahnive1tYq<EZc@D5
zzDze)S8W}gO0A2xZ#Q1hEZK4RVkCZ5Vb99SDiOky?XWPFntGo3(PeJ#ob2pvtu#C~
zul@WOJj;WEf*c$j-F6(Vjg>mpm(Rt;Rr-p0X)LSr74_w4?MdA2CF3@QreoC9!LMH*
zx4ZfL05fi*{)oAm8H;*B;_`dPrMUwK4qWvA_WAQwi`VgT33hgN<gKARwo}#oYr2dd
z1?!WPvouRz&925+jkRE+!g)ja9PLILl04}p#yj#ab8#s)h3S*xXNO7qGKB>N^BfkW
z?GH06XQpRn${RSWEY6Y=ZfB@y8yH-+82o{UvK;*Jxw%<!mL*%o372{O`n8LT%hRWS
zk9nx|4pbBs6*V?mk2bv@A0Mx<Z%kBVHELlSJB0;Q)=TDMUAnw5{c@#q=M_57@{6pj
zLj2+R_P-B|73!0iYE~^r8rUyf@aMMFmzVd6_HfkD=&tZSld-=pLYSgvyRgSGX3z2`
zI}X3~@rmwh9Pax3i1LhlJ)6x$N4`)>0H<kqw5V}M9^>fRV9hBKAEr@Br1oxens)q8
zDqcI;^wyRPgN(epLQ1K39u&A@u^6?amp3%X8^tcpjvT}wE(kuRqM|Z0H+Ma_I@f9v
z!1Dh6d&Nvc|G6yXu+-H1YpW|*3k}jPbJcP$kFY1J7qaAK`K9pOIgc=Rl7E5BrS=iA
zf?aC-{Qbw;vduA~+Lb<}`qHmowMt$buJk>}ac7y!ynnPMqw@QAaZ${}Vi2FBlIZEJ
zs(yii9!EJ?PM*AGbh7u+RtoVD9>!6Av)&uUJFbc?&J3Mu`P|v5o?aLj*ovFDR~?eH
zGA4ii!i9z5gp_2p{NcVnIW3(N2M_vA4^(p)G~BY#+PDSll;#K#wePSU<uInYWa!7j
ztMkpTCdbkse+vzt!#q|<qNoH12eCf<14rI_xT&x-CiqbZRt2zo9OI48EUT}smT0PL
zZf-6u6|R(AS^9nO(sdyr)f|hPO&xi*(?5Rvz}St*B|Uod<=4+r`?=A$M!9?U?rq&g
z6|0t?qef(b*pozpxn371Lr-lV5D@S^gjYG`GuF~O`+)<@N)Br)zjJ=C{h0a~85wE+
zdmI~3HP7b4$&)9UnZ;zPCHNM!mgj%PGk^a48Ih<c1h-0JyHV^;dD&`|!SG-()xhxZ
zkwb?V+VsRv(9+VLJ$sgihlieC*e1o_U-8Nnm%&}u0k{0lhlvGV3K-XRLg4tZnaFA9
z3yO-a3AIfRlg@}a+RY4x4b40{DI_fH_T<TRVPS(IR_k#UdU|^5t;@ToJ`N5J$|cBK
z4%e60)Vw$AtHAW@*F{&wggUiE-rQ$zzf|(_s7Xhj;*6yqpJUFjmO!>jfyS3D8XSmF
z0qh3ud={Mrj;?i-+qZAu_}5>LgBz1om7_!`HI^{bdHMOlK|w9)dUY63>~|KGkCHa?
zalygu4N1yEmAj+t$^;O0WMyS_bq6DaJ*K9oEom^d4qcq;lw@RwnhPA4qQwHEIg`_G
z`Cl|2_`cHZM)SBG1A!=Lxjb*0zlGmneiGM5w2qyBf3GUQpfP!jl;E2h<Lymqdavoc
zI7F}dgqeZiN5T&FgXh#_4bAWA={08>DQmcW{rWXFmO&$TlUqWCR+@5_@hmnMC#Qnw
zsLk*3c9X)DxrOP0>HaF2+{)(WY{d`v5GtiI=a-hwspTn*H7V}f=UEdWTpfJ*Hs`#{
zhK-M3(u)V}4$&JKAMc2fpwSp8D=R}-iu?Z49)afJ=IuZa8Nd%2->JDRCTC`N?B^2i
zQ;8w?`mm_x=+#9d(w|XEOiaY6SvKbbrMOJXZRk=hT8L+$6&ZGr!@%oQAM4&*DB!m1
znbiT+wD=*aTiGhB1EJ0V0r4qY&a<=gn)N=_UE4}=EY`BYq9DHKwU|^+P)fXPwwZR#
zFAkHAL_G&NlOYGW_Uw-zpB^~xxM-0Pa`cknZY%r6nYjDMf;>HCoSX_$KMcMybLX<V
zcI?=(@Xdh%0VQ9*{=I~_E}0bLQ!OVh{%X)n>C;nUjyjmv0nT;LkPtq-+SA#0_BLS7
z1eZn%LS*SY`vwLi0@!YT<a?i*Dw(cI=ZSEeA{`}?vxCKM!TuX20GaKl2fuN1TAxm+
zWB&sG&LeEQ$JDcoRWp2MM;f>1%Ed`XAvo&4RCk>9_V@Q!($v=<R(D>Zr3w@EWhp9&
zjg7^!cy>TuLE#V;)mg)druPx&b4a~dzTD>hFJ1*>mp)cV0hE2w|LMU-%-rrwgO)+D
zlwud1>JUm66ImN<!Q$r!g(w_!k?kZUCEq;S&ieJYhpVd)GCg)r4IU`Ra>R#OS^l@d
z*w`3J>Ex+X0@tqP<>l44l)gA@K0m2JI#5#_j~(tf-TzWa6T#7d6i+*fD{D$fNQh-1
z;XXUi7J1VXu}U{bgz@ra`}Q15Au9xr9}fvUdgx^|buJrqr(*Q2!|huO?PlYB>{C>8
z=>kgSl9c3pKX)$i7*BS6=3iU>5IudsT{i66W{rr@(9rW9OUu*M7}V9?@_oe~a`CbY
zGj$TKjqV|d!#zDG7Y94mUyE{QJxPg*GQK?Hqs<?b#HleVKQ@oZA<fr3J*Ifvv-~k@
z*pA83W>)6kbhNb3GPl!PIl43?Dt@s4ZT#p9r~fUwHcD|i&&``RZ>x^LI-8xHm0N8}
zQRnZ@skw}3l4~W#c?W@OXY!A(8YC*khZGoczFP#E{-n!R#mdw_6h)q$ovk!Z+P^Tw
zxmm(yi|gB{s3?|GUP{>btK+#79VKN5ki=@t^f<yUlC2Rm7`79Ua5uGN%EDWr+6(St
z;^Hiyw*pVo&^080i$s)HS06Mw_oi-dh3`3&Px*0iaXCATH8jG8ShjB6s(PIv7?A?=
z7CqGQ@Zm!@3K}|k`hmVaMx|S~Zt*RRvbL8ZbS;GoJx@$Zk~^xZs#-YT^Bi}RlarHa
z&?r9kc6oVOHP`CdtGTH@(RP}g8{8oSuMv9x-NSXQ$QC*|ZXB1Z{K+=94*`O!W)6Xl
z`|{-x`8#^`+2%UJ!onK%YHA%tuG=|`S|SfTP*hSH&iKg5%*=fJ_?@<orut)u`@D8D
zx`j?FkM~hLdi02$nMM5O%{@lWSU|WPixiQo*>2Pr$=2%xjp>k3Q&Xd(qYHl!fSUoP
z5=zOj|D8PSfn^rn&gN<imQwrSTWyr>bbt8R_5)`X(zKqXwwIUR>~vGeG7diP0caEW
z{Q2|2^7wFiE(Sqi;eq~s@`HbeUbZ&=`gG6Yfh{0chv$ggHW4#SN_gIX`J%t!k9SwG
zzV>9|Ho1|@Glc4PSp$C^xCFq7q$-+nlHJHmM^iJRk&{+P=KCQeu9glAfFn&$N4|ZG
zM5w$08!M~w$NL%y@=5hAGSbqR6t<n--rlpFYpb`LR*Mg+4`w_MefI3x9H|PRnf5uc
zHS)565}U$>wr7+erOq!bJe#A~woU4?$hEpOVF`)qWYygGOfN<my5q;+A;Q|p`Mr5_
zqNM}ij5>?;jADRzZLCs8d}>Wgi{1Ait}Q(7gCi3oBWF|+a{zw()GWhaQE{C#Oq7kQ
zEnJ;Fr^<v(Jk(jZHrASzYcUiq>WnbF_p_ZG8QEk<T2>at&YkT5DbM!NOG`^L$tMnu
zjqw^boi{iCwYItCqiKYaDB`(;Lta8cf>Q3YXE*QM=_*`XnE)O=uz&x@?Ch7vcyoX=
z7v=qYeSLj=<gt-6=AP_4I*yM7=(~CIX2pxMmoHyFd-j&E23905FRx`{cTe&2if?C5
zoVc1^{o*ideUZrJ%eer@LxKEEo3?BL+DVm9QWBc0Pg9WV&gyhpT3wnrb2~Y)fzNr>
z0a?RhNLtKP4J^iO7d5xT{QZp`?0BlX>)>UA)!cmg%_(7wLRwi!Bet~dRAeeIC#MZ?
zZjsgY?a%S7;z6A6{V0KoWY{i`)L|<59A%a7YHDH~B~<$1PB2<2uQtoZ%igp-|L{I-
z+(TdAMy!oY!{)t5j~d*&*8uPi>ToF4Hs~6VFCYcBN?g9Ojm^iFDUC2BJ)~L?rZss3
zjtdtQKi=;~?xr}#V}s1u+pCqWGTznIH9Fdch;+%Q#bKsK_z5Ki8JVO55n#GjPUqAW
z`h1VL{&FwxvPPv3bGU5~9WcxLw6xm%wQ^5-%h1ZXOjkEIb!1?`)}mZsoSBi7!g%)K
z*^>z^nMSSHo$S;y38|W;yHuaFq-l#H?+bOF0_en_AC&}$`=ngFy8dD5x;5d5CBwBc
zEWSi`tCLL}X-qL`%@ix$8jKtZBx%Vk1rDHv;8j*$UNiDGOMj}jJV4xE@|16-D}B^3
z13SAVaFP;tZ-4)I^9SM5ni3Bhn%%qaD9~cv*Pi;b^CBZpQ&4yyyy{NKOG=)1TAl}U
z(H&w1R@;7%MVE&D+&Pol$ea8tGoA&Oz}ptX^>K@8vvVp14z^9H8W@0@4raUu6Ip!o
zGZ0ryUk4^S&*s;)(lByz@+TBDd_du11Dd+No}Qio0s5a@@$paARg;ONj~_qcZ&FmU
z!DVF9AKf+T2iwf>@O<;820K9AX+cOv#Bz0c0TYs_Uf7Y!jc4b%^U3AoN6s%S=S;CE
z3_1%|Kv~%>Kid4di|@0H+m(WZvyrg<S#{Xj6Mo*6lf&aBs^u?&GFt{6To|m00O49)
zS-PF5@T)#fW|0qY*5z{(KLWwv7$ym`N11uXW3%MGqetz>+qg9Bt*kP2*V@`tYF6L8
zd9&!nh;c_8`~WD(@9b0^!V6AuJ40U%I|d-Usj2CU8%1Z1rD&;X+J_Hd3<TsIZ_mv#
zYHa`&C@Sh1s*915k^;(q`tqgRT%OgKl9EzDOCPRO$G3`fvFjL56CMF;rXfYW^ZR#D
zbTb6wc*n<Xp&bQ|ttBPbOJm+W8{8I&&4g89glLyzF+_0Sty|qjM@NA>fW>K`OoTeG
zN?CYT_JvP)y*f7vjKNLdeu-PRc$}6$@=5fpiHeFUXMn%TC17gn5a|Ghxw*MpT3cJ2
znzVFuK=QH0c`jYLggfBjS;oC9J+b@M<2ooKdg~V3yYrY%IlAzr0~^UiVm2casC@j<
zBu5q{3I_Do#*MCHnk6q7`!})V8G+bfH6+R>1t={ce}RM<dA)V9_DV`Jva`#1pMc?a
zpN+@m&hOf{Z=ZpIfky6z=fDqS9VgZmv-7V&xZ1$elP<8a)w{Ts<E<0B*JZUs4>B@l
z@!PY8ov9j+W{UBf8_x)pgY3Ag-Rv*{aOga}v9vY5Ji-0w{do64DU@C!WVExe^AHIi
z@sDID?T3}d(diJ8UroWk{5ml3j2JI2Qc^?&XEHDHt5>h)T2EZE0&yTSp<dVaUQ8g;
zV8VSX`$|~AE4q?2^O>M<WaZoZ>H&)_n)~`yaOnAi4cHCKD=P?WAt52gqe{3!<dyYQ
zYZMk1hM$!r|0VR?-Cde4#w?o$;hBqT0pt+j)ZN2_!`KM#mtK#iF8lwzNO;F#l;%Gn
zJN_4!=>8+6_5X!OHYY@7h6@D+1<A?Dn0TOb7G7RQKpf(mNvk)d8$K<4X>3Hu$o%Km
zEG(}8RXXhdTkOS7_-DWF?}jEtw??odP2cIyT-M`3<!c20)jYU4&*b_ms<=hvdqzgo
zG&Clo;RqJY>*H>drLCp)`0?XSTeo)R*|I`2<l<_st*tFDj}!4`0&yM$mIUmm`u6RG
zD&1r<69a?aMQwL|(Y^|A5O!rv(j)7l;LFy4GiJ!S^YinF`VSxerK2;zBo}XEXSX;%
zpRP};z`jq}02l*#<<S29`;Q(C^z!n0^F}I8nig**fad(Xjjb&*Nq*G8Nq->eG3bf>
z{OZ3n<>d)tG&kC;*@l2Pai$+?=?*V%?+he4-~}XN!}gq6JUqlvh^;pzBs`uy%f8=T
z^p;k%CCdas@C$fcT~qO26HLy|YtS9yGH-?Q<r_Aqg$EZd4xia~0V|6%wVTj{L<24v
zDZQ20xYh{KRIrnAY#2GYw+z4HwV$&vH(!Dv0@dj#v<Rd>K<UZuFCfOuifNHhg+iSc
z^#HM6ML`M%Xv@F;@(3j9?K^hlSx<C4KX?v$AW9ZRU%q4+HrIk_c<?RiLJBOlzqtcf
zGX?m<r+f?LckcT10GcQAOi^Q1Wu>TZ4e}AI&Ey@D4<ybfuCDpPEgQ&axDoGNPCz_V
zL#2eqdHDi(vVAoKCZtU$A?&=ossSxpOqH)xR8?v1f9wUfKGnxj4xo4>m+Fw1vGF*7
zuADMNU`UD)6P&*1)aeea?WW>$_-OjG1Tp}rW2`nx6sZEVN%PG{h*=!QZ8o6W6^feS
zJv}{(zbE)1<YK;8m{>x<{4#ZG?v}ki0cCly+IiI+pbkm-pw6c_nV9&DZ-E>soE!fl
zd$$eA;))4~)b!-!jT<*4mvxPe`E-B0+qiWXB#J!n9WY+3yTD6Ei=!W^y?lcp$U=mv
z5G5PDiwuv6mNGibWfmD69K1oG%AZxZCvnqW$bga<Of2IF(E&)u+xF1`C#i{s4i4%9
z*8wI`Z*Riz0ro>Xvm8|?XE$o8|MEp3qp!Ez6QsfUr5puW7cas2en<!k3Szl>neSH0
zG(3Fl7?hw8jA{cO1o(rJNmjQhwHu@K9cv$xg}uUfZsymo@~SF!wXsjRk<edkj31?>
zq-aF&V%~7m#Vq7eve&K^mAyGNJvB9xqmL=YXztEWc=zrO#tbyfZs$jn&VrpsFAZAW
z)H}6I2`u6*#k7Geual^C*p*dO9O1lw2XatPDTqZ7mw8E5m6*&L?kj}L{QBW?Aa1}W
zW%p|jtkv(_xpV8*3w_bLXtC!Xk_f#0AuY{a^`J-u#MSM;H^oXvRba{d!U~v~nfY0Y
zwA!YSB;e$<3QQ<OFTS|2pqg(dmsC?!ba$vW%GA^paIL$$n~>PmICZ{Lwa3(ZUjr4%
z$jFd0+O}iIGGwtA!~i-95d-+;T;cNG!-wNOssd1b`g8>lX3UpOzn)m{6(5!xl2z>q
zgsMzCskYmgdY*J!ol@r$HVr^mCg-WHF~vNa_jwHZ3yHPuxz-+3T;UB0U^jrWI}e@z
z1Od090g^#7Bs}O_?n+sclaqLc1or-cffR-J*XFORuC8`EujMBs82De*ZfkE}gKl%K
zVE$T?1PV1!a*kE!9fFF1)iQnc(H3Tf)H@hDf{vN@-?~tG7IX>dSyq{R`}UQUmFHY%
z$;kAHNyw=vFPCaJ)Wufi9@p&a?!KU3&(P-)<vdmmPDU>lP@k?R0|~;wV1(G*VD~^)
zY*wnp-hqKj7@=J>d~H=#Rd@ibP=Hx*Q;gy4^tAoza<*Co?dj8J6;j^;b&HFOYgc&1
zfoGy5@q4s6;^(;3(9qC^4I4ly4BI}whi(Bq2Rs5YK2X{|(0|{`S*GhOzM-6^cR#so
zz&c7t)gdcut=CiH-uEEe-{Y$JTev`LAdF!|5@h4<YHQ2AzbgdQ9*e^5@#8@`iC?_V
zt15nPpNb3DRsxu+<=IGmyDCepVI*Y7#Tf!F;zRM#!NICYN*Ped`@i`ks)bJ|#}RtZ
z?@s66b92d35l^5reE2X8WiUsp{B3$V8?~^_{A4#^H^Ro+uTmP|I%Op#`B=&B2oJvD
zj{GIa+e8&GXi8*TjdvL#Qw>IV$SRMHjS1;6U}vZ%(Q<OmH-FI4e`#!NEbwH9lo0~6
zVva@J{ZB3~WMc#TUS8ngggRzgF#PW7BV6#Frl!oYM&a_5Xoe3~(!puZ9b_&!C)ZBw
zB9pX3s8z2Fok0N4Ckc1H*L0hD5BVjgF#}XsQq4d^<4sV|+0lqwzoT#YhdZVtqA=Kj
z55#9~zkOS8@4<ugdA&cv1Z-_>5#{SFygfW5q@~|R$0J`wjQT?(j&dUP8bEoBx3-+<
zQ1xAObac!p6ufqjMMZMSEV%?*3yKqpOGO4GHChM<cZ<nEL268)wnz2Jy{+tnF21xl
zl9H5|n7+RbvU?zhiT)7{<lgvmvmbP-`Bp|mxh4M13os*7WBRSC3Xv)}G*rEJ#JvC8
zVQXt^W?EYt8)!}=lOn!5x_dd1Gf$=rLqPC}d3WA&ST-XeJR-uiJ%CLweC54y+sB#@
zIud>Lx5bXI=}s<96dD{BX8X~cYn`IPLhKFfW<j?MEGqme<2E7OrxYw!eEW7w)V9CB
zA4>5o^or*WMxs9V3%psTf3n3v4VsJ#Mu2LP8ME2F843s9O)`G8W#?Y2Kz`?f!3d&+
z(X{aO*}g6O!9BZIQmzbq51u`8nwC~*{&N8mnR2}1jTs}Rw@FD!T9kj0xqB=?7*Gx7
zG`(i4Y*`S<f;npp6WCfj28kFE!dpw5RtGx|gVkU$0j-7GayW8=C#|*%Q4soU`1|M5
z5C}n4<Ya`zKwQ}k8qSIdn<KCnBiUA-@BrD4Dl9CVlQ|Aj-;7JEzGA;6Awh`U(IE>9
z3n;(QpaeD-Aq0uBx34csPt)ENpA4Qp4j>Z#8#%e7*EwA<CkL8@pVIS|_Z~T{>?{0h
zg;~g~r5-d#6&ltOm9vsjrp)=KmFK*+3Lvp{$D!A6-h?#@jJD@CwzewQiEBY(1ti$9
zb0^7S0jiI3x~^&0rwwB!jwl-x7CIxro<r&Owk_>A!Zu_|78VwSjrW<ER+g4Vq<9jx
zItX;^iq{hWEQHs1_5eRWHWc-mva!ezWb|KR`6ARw8)cjIxlgtLiakAWyd`<5@RirK
zSFc{7JSA<ge^ul!GD#|eL8z*Jynod;S0+Y$QEhx^C@wEa^!rJokPo3-n9kD@8o~MV
z=Yb-zyP25up;6Dx&AIad|4;^VNe56KIIy_5$RyD{SQk@`sR4+9fQMJl&!3MTKD_tM
zuX}Vfu~K(!q1t(WU7dr2!>ePw0jO*vd_r>P=jVsgEEL+BWrF8cS>nX_Ap;xVzkgpv
zr46#shYugt76w8gYJj{laBFF4?Ks9Wk1|M0sXvdcnU+>$T-*?}E&%}nh*RDGT|hH8
zJ?R+fua6--l)t47xIVw~yAxBCa>EF6_VpKsOtDFUYT;u5W<{1cbNd-o00Pos>k1X#
zDz=&%Rm|`KP&|669@8l2?0|Xk;9u2kk|aMPpXkS9o2I5iJAo~CJ4G!gbk!C`*zPZG
z2I|u&)m}ydshO>T4nH|NTlM{W!>{hW`}SSNlI4Bv*orKH6frw9R0p+3OIsTWLM7kM
z66ps88aXW8mE~nNp-=*4UE=2A@;;}omR>l>qe!|65HIL)E*5J{V`IVd$A!k2{YxUW
zWG=sb-qR0mAe*`KbvB;RKewi(r{jiS7MzMT=IBUPy{zHZlX6?`$c4L)!v)@mE<g6<
z?EfBoX7m6Jje3kYwRR{H-S#D~GCR4=B;F*ar7?~%iLe{-SBD|&vsqSFRkZ_gV^2M{
zj>{K;&Vfu^WDFcGA2K0@HI|phQ?nH~rt7_CDVw*Hf&wRGtUk?WEN^5~Q{v~ZY(dQ<
zHyEW);PVQ;-xHk&=WHjtuHrBEJI2e#nL)v*iO)1{k5?(Y1T{^=4NHJ=WT8LMgpKzy
zb#X~a#U}AkKGj~=@DVxWNCbi8mb?1;NA~Oq%M(eNjmONMGkY0N=Se5FcNasBnw)h}
zbuEOPj1(zSBmg_sRsF!ri$NO(HbO<BMdg^YK25vwad5_2j`6w}kJPpKUq8L6*~BR8
zXLQ77W}aZ8KhCGT-5{csAFsl4FvLN+z+nONmL>IJH%dNT0-u%v)5YI<dU|K6s;IDy
zQAZ0Qae8Hmj=hzjp$SDXY<nwyvAQ^N&fMI**hAk|lJOvJ_ajQRg#SaVeg9Ia<L`EB
zp<Mu8LD)6qttf2qAa)mzA=>LHiGIXlgyf@A*9Bn;=o=Byx2F`ehSK)RSKNL5cLCW!
zM6W4POJn2Y!J2T8k?$21Z_BmdRR9<XkndesToezyBxZ`LT05*dEuu_x8O4vtQ5EDc
zWP=8kG31VN5rL^T2w)d~_q!m#;uou}t<8=Yi3%uy6dm>hssa$1iYfso4^dJ^Eqy`q
z3F5WC1oH|ltu_{Hxc#p9;4@6Vl~|7~DiQi7CMdrRJFhO#b8!5IDxGUJW@c$AH!_W3
z#=ad<3;7B;ye7*;y_-s)CW4tsHujw!RZXt!)*1`>C(#=n+^Hs2>r*uZGJIxcENb)~
zK|go`b>N*R=nRv|h)rVr5z>Sqt$mT^w%Wbv?E3YCP@?7AUcGpcXEF3vR*3R85Md8U
zmFt|$ehUo~lN8ztDb;VvK_q9Wn$m4gJv=;Kyr{4JfZN2PskN9usP(?k02PyTDG9pY
zGAgS3vp{EBf&G{%*Z_`;6Hsoj1{J3cdEEWj%_$%-5V)kdYXjt0z?r0$y;M|&pDw)u
z$I3S!c$Ml1RE%khznv1&6Vw(a;HCyp15%;i4yDYZIt_zzWbfXd?rseuqp`ukdDx&<
z0BDPh^+!%(k-O6H0C2*tArZ`-0mz6+g`I`!qE;E?VAsoWNLh@|>LOu?+S_!#M8(8x
zx7S|~JD4wOsqt~p1xB8LexU<gUAbk&#X?Q<PC+Hx$rxw}xXUfo{=B`Uq{L;9Y|7r1
z0v!wZT;gz(KD*+V?g#Tbua4c8z38QL?w+<brL(9R4Gm4-UlL(kb{=8yeR^OZ*8s`^
z^VzfevocHs*#RN@@`O&v$kgHan2dF_v|?%5Pn<YWL0y4@OmzNfME1T8t9VftN;>hN
zSQW>2eK&SfBjCkm5QUC0g)NTEi22MC{=D{?1&=l%P8=l`E8t4T^@9fw*2G0~JN*T4
z4RRem_k9-?rz8*u!z18G@};v%>9NSXMdYq^LUna@w0uU<_?CZl{wI*{&!0cxfx39{
zqH$+IJ}{U@KwK(JL?mbGH|1I>>V*fa>ZU})maC+r0<QU#xe!139^v5jw4~F-{_a__
zu1D7h#`8=+fNs$!=|+6#Z#E+~V-<wS#*Q#}Jt|B>CVmV(K%SEK0g&zjNXp=pxy_a|
zD2vi=lEVA&9xKyh#Xv>2++T6G=`&}iR6%Znr!re|qhFAK>Dt@dpC^iB_jA+HMVEaq
zKEdNIE4n5A6wqA;&zZAlN5N_BcA5e2Y5c^OLYEOX)jxVDU%U2@B_#w)@qTf_pondT
zhrg~XQpY~W{iNzAb9+Qoz>N^{DEI6k$_K974+3nFii?XYDk@4!hyh~N`JRVTr>-ri
z$AbqC;+exkcQOm>JU#X`Z1G@$;1>A0F4=^BuDvoN#872UzkzImiqIs)si>*Fvv#%~
zx-rY7T`wK48%jm*S$>w8*`Pi)>|?N>-~8Mero*UiA}6QySV1MS?sOdJ?d=^KTbZur
z2Z6cPneFb)Bv;#Ub%9rYX~>5tEFi)Vn4V3qr?*!&(@@cKP1N@B(>axtGHWr*R`v|*
zG0dg=Q7m=<WY#sB4{H}zXoPM8nn=z#w{A}+HIOQ_E^f3k+`l*!6Ix%g04*myR3rYn
zg_9(9y5Qy3mKOJrytvBK5vVq}qbNTR?|K!?w<rM{)A(e6odiFnmCrqo)CU{48n<R*
zsfo#Kg<}WK8+Y;RyHF)vk+Rshp<DRIl4|^;r@LY~LyA*3bP9z(zi-L^{veslNobj5
z7U2O!r%#`LD*p6`xTTNG8~;~QgHxNOv$w0kYN5;$a~d*=VTBiC#Sjw;_}Cx-mZ&8r
zJbEO6?L^e^ex5j*GjSzmsMY@7y?$sHgdYJmzg2LmJ$v?4hw@jW3dZ*8ewttW$xe3C
z%k88Nc9KHU6`}4m2H$t_FZG)COSNxP*V^dx^Ra$o@=6s)XKW_VADMz%EhIRYIY3x<
z!BSq_M>eXGUd*QH;lnwVi(qS}+`^_nLi^1hzm-NphD5Wyl|*;Q{aDPqJ_@fEv)h|V
zou|ZfuIe87P)0nr^Bb?{zc!FD@U1V3px|Kd&euqB-ojSH;+Apm=GZ+-!Yha4tu2I4
zqbdsYjFP80R4RSXecQ>&QR%s8(2xk6iV;aH`*Q2~5E25a0~m(7v}LiEb8p=p8q@Q%
za~j_*`O#kEJQE+e_Y<VU=9ZTDOw?W^OLvG?P!vYbW*2QvjTec$Hnx!R^>J%Q`KAZa
zgP(JW5))grS$|wS0+ud~WTl_bu6dTdD^ceSCuGUU1lK2v1oYoG22L|#zX0axoyEU2
z`k_Zw8pf5NO7<FU>SVLA$o%~L&lEF0!{=)0Lmj4CO$tGxm%%Rp>JDAhdOCbS6)|JA
zv2c|gwZtr?FVHi+23R?{xT=D<reSM!fAIpYPHmLdhP~MOF6?iiTu6>%v6i!M^h!v@
z?6NeLwuW4z<!QY@OB*E^UW?2SM|~Hr&yOgVsU|buP7WGeKF&PdE~w5>bJLtQoS-tD
zx+${S<2n1`1vA(9iR8e(qAw^Uq(>8Z&6ONu-SF$0|E;;M&l{-|?A)F{@d$T=+<xZ}
zF;&U1M0saH_?XSP#LcagZC!xu=;r2@>S0E!^yY()psz4xXtYV72o`7sb?M`niMZs*
zy4C$1(ktBXzamKtA$wbowansQ!0JTV*48E*tklM4UfFICFqkpxj=8mUA&3uD2Ecr<
z%{wM0>hR-`%8j{`|7Mv|mD#zyzJtAQj}PETvHF0*L$9sE46ryE`2ULZg?|eaBO!CN
zEnBA4K*;w^e6Ui2m%smEYU-q)$7zK)ZQ$0Vv=1E`L>1K}g~zcy+g!LQc5V&<S#0LK
ze9fb0nwpxf&d$L2(0t=tYGGxh-s)AZ1va|x60|1i^y%Dd7QO_O6;b<}tSf(A*CN5y
z<b71A9e0k2t6gdFK69Ii-fqVcwi^~-82iE!gp7SmKIto}sz5ohcf;f{J-XDLB}Nj^
z%;0y;k}cFY!eKl*Fz_27gh?4P-i*(&5s0i6jz4_PDu!#n(tvFDN;&4>LbB_VCrl;@
z`34n?p%{MXNcDZ)k-y66L;pTH(OIZ|@)3%X&{QQdgns*5V_OuJfGR$<T7~j@Du|r$
zBf;)>6Kn@A2`sHwG(fo$kkZn64fZxcyMz+s&U5qBg$q(M)rt3c^?*TK+XX4ADUC*J
zBHS`EEZ=Ng1(zx+fu{0z^^f~zW_t_vOdRM=34Hld4Bp)r(mOCJf<*3<!(bDXY1y>s
z@iy*bRkqzlu8^6ToZ(5$(dFoqK0anjf^_J<gU(BaG^2Fb4$Hk}h)1MwZ8di)^S*oN
zur%G6#MK#tBqmavRqa`0YS0fe9Tt)&hltPcjue1w1<`84gTdL4u#=G|XqN4OnR0K<
z%*@oMe}Kd#$92ktkmdNoH$k*Nv*V0VjL6}8D8>_3;I1OTwe3eKsJZtaIuzS5KQkng
zQ7_T`{9M8LE>p`!n1VHa!qg(zD9cpVH$~hP>IBHx4QyP-<89eRKNTscIPWOD0XqnS
z9~#>{4GwoPddtY@Xz+3ll<aMMC%M3F$#WN$mJ)bJz!kC7w%Naew`zX3{)b7ei_I8;
z`e<pOGZf^7IlI7<u=?tcz})sYr5pN(tFP}Cf+-p3H6;4SF{%f{l=EaQmW*uXJj5K}
zPex<<)2C;E?#twyP%*&c%80)TL_ub-jmE#iNdP|s6Ckm5!CRAZazu7qArvxHW3n>D
zq6UlGpuc9`#$TT7M3FwG;d6Vt>fv#eebwtYkyiEuhlh(;mad%C+3xc<X~Ms_^MU9-
zQhTY_8E4lJ9@~=vN1ob&2?MLKwZjc#rEPavt890MxhTAMvoP>88ACyg?N7Q+AM^~5
zC@=j;SwUeL_P8x7P55}=jBTw;vm+b`jedT9f7mqKUKvKMm#_qr0{#7);Y!#Ur=SBO
z4x`00CC$5c?;<gy2yX@71Kgy&0|OunOx)ayklkdWZ^bERjYCv*!X#%Y`AKSN^+H|8
zmg9rSiDrQjkZkv94G5^YDC#2~Cy0;9f?QDE3xIxD^qPg9zO=4x8f)u@m>4QtumRYh
z#{4*N1tCn9^Btnw;AM7b63}CTLxE2oySe=yCXK);q3;M28}_71Z{4{*!am<Hm0Uf)
zk<5$RZcJ?l+4*Ep8G{8+Cs4?3`q*qw^kZR`9wFQ%!?4h5kZvA8g$LBBre1N@UKQ-Y
zl^10RQhx&%H@Ew)s3Sk{Sq8S;@i7HVkc6)^F~g2z`KBF<O-NWK*_jP~ZAfsg)bQOW
zWEPTeZ9z9_mZ8~SziOllS#Z<809T?SBXx(S!B3b(Tk>o_wB*O9)B=5{^9Pw-XorWk
zeq8#cBJSzzq6T5V2;?3Z8A)i+Ap}Nhn1o)@@Uu`2L)ob?tf{R9K$dPAz&?V)BsI5(
z&sq*)DMmbaxFur*6I?&8ge{X6Vp6bhz4ACWOg+zA=<7(NjIxBqxv?|*IH80Mn%6je
zGX;whr0IbQlM^I(R=?-#v!V^Kelm?_TTiIf2p=?#HPP0NY8b#Mw|HofuGM{51yXw)
z3{aagR`tUWHH#<Sz>_Emp5pF@)oTamO5L|_u)-UJZ$tqgp^BbVgKvbk?pNK*r%|-&
zi<-dZ8j8GEFt^9+HNoH1?Z(5#md+V9+y}SO7M@ee%F5+To5=RlOoQ_*>LGmr;dxyU
zR#5QE&0TtZLI80((y#LuHr~?j36tiuC|C%wnbG1RJ%g$sDtxl4&*Gx-)0am%<lFeL
zTj8%gKt&Y<a^>rLkcOrZL><g*WvcQVIN3o}wipr1sTMFvppd|j5SDYpk=H<<{BLF6
z`TP+6WE5m$A|nY>O-I-7tL|iKZZ5;3JqBA!qq;LQ_d_%Jrz$KH5I$ja4i67UWf8Q7
zFb~1f0S)hKX(=>p1$)|fi4*6~-$w-p78Ir_zwKfWmmsey1HPfuf`8@+zZ|9yS3q5A
z@7}%GHz^^!_AE;2H)^tB9^SZR$NA9*?lVA_xw*M#j6=CC<Kg!~B?8q2Ea01>qT6@w
zgur_b^NQ8cSm0=1!#wPYqz*V&hcp0d_ay|nQ>RX`vQERtft~wnYD)ThXAq=2=)i)a
zu!IHnSfL<^of5!dQh@A^Rhfsq4#f|}z|W7if}kwHwW{Bc*bI}SP_5G56t_!oUP>QJ
z3)_v6JA7DC`e&e@-&ys7Hka+F_?p*jO1goxh<I?R>zX6sxJL)V!>w;Spr&ozdE~Sm
zd`cwPP)K~NLI9cG(T^k%1gs`Hm{l_3Rbf>{IVZA_HcIlwji)UWHh44MrJZPpA~gNa
zk_TjD`Cd;JOv$0raF@`M6O9a_qIKE|%nyQZ(S1lw{hORT;Jn7e9$~TRpS@<;g9|Pq
z;1ce}`mBZSw<_R>^9l+|>|>l6Z&zOW8v^PrR6<u{^_+s{NajV1no8*V5wyQ4g4n}r
z+VlAS<F^4c^uCo4F1@E~*oeXqY>#GiUULhknA*KNTV&zoW<bSUv<i%eA&p566-E71
zUh~VGXyN-7><e_UvbTMplo7I)P@Q=_o6N1x7?7ou_yDJsMaGM!VS7VqV!9?KQYIYU
z8@J~O2?{=ab#G?D6uLgp<2kcl@SNfK0>6L&6bC7}x?sF>P06N1)wbTuYwR)`KU8lg
z{a_@|rJnk@bEwmQa&sZ1dI2~9m>>p!ez;l4>E{ZHIw0qhC=|6eHxFGjl$3l84P!9>
z`}gmVqEIfWiz*^?CS;798QBcbA0njG<-@bf>A09a*qtf@-cRY|D|3X~LKlC{qV9+J
zo}MUND=0IjjRgZhg=KPX&MCp87RdwEM|59^QNs-2jBwn3iVWs7FxzyKPPG{|EPJfx
z&6_sqs(<3V`}HZbeS>W=`)`SfL3FW0a6;SAh=Q8wi(|Y>NyS|oH?b)eIAm))m%!?<
zSeWX=ia%&;i%d*3Dw%ZhkyQ84o;T|)gF?51nmfHro#V`z1cVc{kM~b7FqDBa{&qWW
zBR7iZhQH{BgkHHohE0z3L$r2bJG+mzK^p^XKDZ#923G+VSk5i21TY9B#u%a!)gPRY
zFaj6^^&??o6YnGSFIfW+>h#yBpmZ|X-G!gHIIPq8;s!FwwD<4*q0^vt1Ql%$6_o?X
z6DS-Cr>MO_F(wTED6M1E4O%mk)H*++h61N4A>W~aiIURW$EP238x;aP1Hz!}?``KG
znG*ou%9Shk?@t0G!J3<vmewo8%*w{WK~z*9Z`%g|ZHA-`QzOKC80t{1H6HvCCe+V<
z(LWK*CD_iFQBFktobrqSwjJJrn(*afr5hk6i?ar*Aw@vNsM}TI10c@vI?SIxb?Qqh
zZ35V_Kc{IA0xgPFrU0!#T9EppqodIg<O6^P!JJtseIAh(Zr+_82XCE2VFVsS6yd=t
zb8zQy7{X`=pC7Kp1v~M^iddEGDd-=B@fND-)2B~iN5RSfY_J@v)vk|~3Ll{On_aKA
zmsn)B)A!&r1{)+w4FoLd?e)Uv;>}`EClEM*IHGR$>_-Gdksr><1jB7`AUu8g!{Vg-
z2Cxlom>q!C58Y2fdO=SMT0QQA_QGW<c>Q_-ymH{V*!PzA@1ym}ju<M-VU(F@s|G}3
zu`>~RVRGb#`mL+03r2<ASyx*Nr4X_UuzDK=UkUz|bRgJ()ty&v+O$OX?Em`<`G?HR
zV;8W(_neZJks&<ANTf(-s8$ngi>$15FkfH~*3|s+l;G2A&&lY~e7Dqv$;PO;KYhxH
z&%+B3h55dE6;)svj36%$4}OA{qll%zyE-~q0OBy49^*1Q&B@7$WsOP@E9-~5#qPLr
zRUpT2aJjiT7PBXDlkDl~=~S1k62~RPj==sYdAT7I3;Id0FbcE>=!BeDH;{Q9Bi!^K
z?}tU(pHt1LM;8&}h%E)`yGa-w_f}vvCbJAikw*a9j1<n;vPqI*4^2mG?O|b$RS8Zy
zZ#&fswT+}O>)?l8Cu(YHBuQ^(+!nI&3alkB6iAn2@TrJ9@yrllk$Xbk57&JC^5rL*
z3*(uQDLv=&Up;@GfEKiZ+k~*-B(i}lii+svAQOqh+}#GB%qZ^c3{7d<n~M@!iviH5
zO%Z24CM?cX1^d_A=4Js4nl$3%V*mvv*vYz#juVb!hA1dvuwV$2d|teGLAY`fLMUXA
zY=%9cQp4`x|K8nJ7%VWJ<u7(@tOCbe0Iofvdue(8YI6C)@o(*CSB7S18`U}VY6Ua2
zF(dG4l*A8gjPxm8LFeIONE>-Z<9%Qyz|9h->z=UWB5=fKg0aTp=1StXl1)0TJD?fz
zzkT~wf9VAz?Y4A1&-cR%Kj6h#oa(Erph4xjbMElh30e}*IN+pw;H;v^S+zV)<eLRe
zU<vs|1&yynX5{!zG+;B3;0xe%sZ_xJj^Sx+DlP^=aJpd3I}r}I2|<dD(dGoXwH;-u
zkF?KPPz-|MifHEU61WORNk;Y*f$FG+?ggTx<*x@<;O6xn6vFVf-e>OcLn&SN#9AoA
z;L%*FL82?N!a51kMhOe?=RYXc9jW;0|I}^%|AGe+qRaa0KkkSB7w<6r-8hE+&$$5q
zdb9n%c=3OK!G9c^h^I@SATNmy{PQY=ZCMoUBbKNVPp&or&=KF_wg17okr`k8`yT)Q
z!^f_7;?+W>!s<JkOHNke<>@*7g@Qje6X+!h;yBjr6}>wsKaK8%HR}8uIF6ICCCPF`
zBqXvve2_6BvL2G@3t3p%$!1E{P0RS|KQ+Steh~hzOxb^55L5DBdQz78=f+uC+!Gw<
z=g$fW>iy-u`Tbp!rrhhF)lvoDH+~h2U<}YreBYsOEXr1+@U%ujljFeG1KSl&SzQ$p
zlocFec$z*?yc(L%!TO={uS&A1B`1gXF#Gk7rrrOpVve3aGP1tiL^2^>UIIh|7x9wZ
z_3Pgv3fEU~+V$6e_{LXEpk*l|c!SH@VK&_#kTWf)#Uv!0r+S}Ri9Fj&E8@iv7V`<v
zAKu4O1rk2$$}aytOYC{F$#cM2XwHb~ULd5JyLT<9k$|tESn*Qo9DaXjE4DEi*_1D!
zBe2<#g!B^vPhw@6G`#^(U4l*#asG87WN_jb0EH;<mofz*{J|%Nr>K*V^<FL^4+j$R
z@Ft#M>SzzB1lfW(@*t6^r_F#=%^=L8)%3Xy+8eGEXc`!_B8^fV2$zoga|MC*ho6cF
zl29~)L>VRG{X+UMac6so@o`y>MFAE9A{XO>V4^`{ivG`MNY#W30Xu$h%t6B+5avq-
z8sd+(um6z)6aB_EbdC%{NcdWTpMO&L^I<Yv&*Cm9q8;LR=gIX)lRyKl<MOZ4X3h(4
z=Bjw2*9m{#sos3F-)B1iIR5yx{Xaj#P{zu}Ny}H?O+(AX6{qXOTpnMOj*|uM{#MHf
zGceS$s;EW$(eZ&le`KIYJ&LXveo@g%mG~X_h1sh=BSCaU!1Z(}B(xl!X|e^XKacx;
za<?+@KHuVCgi6x-`=74A--!cCp0|kh4e@j5&H=8qC8ODk%<SQxU*5)17ZA~pzV-Zb
zF`D&@{q^bf6=j@^0eL+7iZ<T<df)nBlS5Y80**LbtvR1goLtj;Vtr6Iy8ti(vCRx<
ztUus`E5rl7yYqv_7<fPpdh-4R;+E20u7B$1AM4L0sw#Y#3d=-Na($IE9D-Rn$7+np
z{zjBEj7y9v#KdS({JA56#^WD}w&6qtspcH~{CME{Fzr7C=9rq1;T_FN%+5Qq3#=lH
zU<_!MjFEB&vT!*D!3+JFb>grSW<xH9#1ed44!>bIE#k|<3!}xU{azY7P|i7jjc9(s
z%Vjwp&avyyA}<h~^Tg|4-*f-+=KSwB{(r|VMYY=)_~XK|Omfhk2(y2zR0R4CI8m#_
zOF<#eap@8tA0Jw^2$9~n4Hm1N<m4426_{HxV!2*<AP1%jW9N$_?D3$bwGr57c!8~e
z%SpxjQx!!S^4W_QvYF8|f5X23VFIoeyolDs8*fh5@?d8|c)gJN3C&Ls;8o`&ade8v
z`U~DRtXXtGL!E|Jd5qtQYWME@5Dw6FF<Q6}D=E)=eQA}A{6>f%-vfgrTE|Y)(cO#a
zeCy=}8sXhI@$NW$l*7;zGjVyy=ePJfAQB!@n}Wps;n%)zI2uVzj2CnV>`qJbA)YQf
z66>>hB>(C9#@S35Hz5@VdL~ywWuIOY+kZ&t`GKlhorGIF;y+OtLRFwLKMVW*U6sNw
z&N*{jV)IR?@Aw#5u+6;Wzo%RQ#vp$7L43y$*{TwUK?&?3vd`x02>Kq>m*v9+pAql$
zg@CxYwB*aId@1D&#`?}~A_ly>lbV*+0xN0A9Tn8N!41&Y34%L|ETf>H;Ds_T_y(9j
zJ4%)o7U^#@T#Hbj9vc}k(9!X2q{Q%?cagf6pGF+LQD%h3)4IB&zh-AGEiFmrc8J-+
zFzbOW?`ZI-Bp;jo32p(&cQJHrIR+Pn6YurP)kE4VACcqYSuu!S3{v4}M}(@PjxOmt
zciydv6T+|AfF>Sn_onqRa1}v}>nL<iNjQ^c`tv$$Oac`b*x6Tlo}WVxsmYCdIIsmb
zodo>=R(kMAgb>)EP>6y43b|?qJQ5rk9uD)l@2jP_rOHYJq!hR*pq)PR@SxbW3vG-`
zh?=)<O&N+ma&;|+2gGS*wh?Wa1{C6`BfcN4F8cV3bC0(OvSA9eq2VFlG+P#o0q^=6
zJ`t{SAKi5WAiyu}LB=XDkY};%>(qXr;#UZ79BxfVsUL7UGKSxN?v|qk`c+{S$VuM2
zd$(l7^__bD_)MRwPV{cBc$->TiHnMQhRnl?addu9-%TEI7}mS-2|0WTuV3$ua3+Fo
zJrP@7ii*B|?S+FaE%-HL7?g64{5C_C?&ZsuRnqE%cml|-HR_4UHbL-RLi%OR|B6S5
zJduq%!ZAxXSqWRWvv#z2Fq6GE%J7Hg8T&MN#CLCkt)i>zj%p~Wl*ZZ62F>_nJ9H9l
zsFt{h&<l6@brd1q@TOxopa^s5mlEr}Z%tTPkac{%$4f<&V2<H&|6s@EEs_C25NE%f
zt_z|S^=U&HM7OsQ>KfFy6IG)_LwZI=NihU6xv>vf(+*bdTLy1wV3;PV){Qmj4q#LS
zW^hSdPwG>-P}c{!(=A8jPpRqRj0h+cWgXd&`OysN6D$`m44k1YpqU1lrv)iYDtwe}
zGx9U0c$g^$D+TGrJNjVaVG6W>*HH3WSX)<rq5O^KyC@_9P}OUVS@IF9E-Rb-{hOmf
z4(cOe84mKg191h?6|!_9(@nIsD|r#8GHu$l>4y2PqnB>oJ^>Zan~u<?AX;!Gh7_UI
z;B26=k&)_$dt$C<uu1Q55}WXNp}DE)8!=SjY_Y}OCtrlA8>aJS0VH8qmP~*Tpw+`=
zH`%NcioVTjmtq0~14AC@*`e^6hJA!{tg!pBvTNA4Wg+m@25V`5c}n9rL%2I5haj$`
z)uH>W%^#;Z_1O)?p|&v0EsG;+XblhvUbA0UbWOffWsE1@{2V50LyBI2i;|qjm^w4F
zv{AXa@%A%2A9yp$^g~@qTY$k;>)eUlP#AyS$-lo_3?FRSJs)v)^qpX#C9w^l0sl<W
z6O<+l-$T!+s-^eITH=Z&#Yzb@3UfmuJeyCf9)Z+Q3;_ck?W|VaOU{rH^c*lrM?Pia
z<>t17atot%_{`+webqgoh@0Aam!IPHiG~rcdH9Z1JujfwNQe;vJ~7SxUcqThmlbDW
z`9<OMLTt{$O<7s5WC3i1+i7(8G{tKKxaqY<o!(b0D=qC}11-^O99V$y^Yc5@EP%2Z
z^t!jvP(h%KexDRwgIo#O&@&qFR{2t)VXA~x+BK-M@HD3chZ}^t{l(pQ;<oQF9^#)}
z?*z{qeW0<4iAka;VUWADf*or2of~Hm_=tE~&J0R5Tc|!N)Dm`rBMLE2B7o({&PJ40
z(TXD#f{k<7I#8(?D%~LW%-oprpV^^Z4=7Ftj~rnN93LH32^ZWJp;A;+BXvo&v4DKh
zW0*;diI%nx^_Vm_oQbAYXStKyvz?@YjX+1YHZoBzF5iyB+K}=3EvuX@jAl;!{V;ha
zq$;whjel03C-nmD-w#Ei^if5VVhJ-S=2-cXXhF^HumY|4Biclo1rR^C90<?Ut~@aG
zwF6?NesmAK5h%ESuidwOP)qLW?wN1fckS9G|G$kjgxGq^8&;ogUm6V9o8YWWrYdad
zP3Mgw%LuY@&I8w2At9kLRJ#L`7oLJUL>?`M*LWzvGO#*MHXaG9;uyM%DNxG$$cOuV
zvQAH?a-lWk;~29!qPL7Nb5DyxqpgS@Ij7Pnr5e?DsrijD)*&-~^X`<D;{C0q5KI#w
zL8rOl+%12SE|!DK+x1wvOtiH^=1Oye&UVS9!%$y8;EO0=Kv^_;Tji@2i9^Js^|HrM
zpQ^PDke8RoCNqVPAD+v>A5dxuW25ZY3mhC}3c#B43dRJPa)N<qQ4#=Hk_=ucCU$nY
zIY~$}I1fVS&W-Q&^%k&Xz_orb3WYn7OKY&up($RA!V-%fj)K}5$gtR=N2Cu&R3hli
z&G|&bB_V5dzYVESt<WiWY4+kpVc}A+kijEChp;d%n;`KYUpr249C8<g-(^@)b6&kV
z%-WB7CbU?R%tK@`padLH<+Qq>V{dO?Q-gxb4Gw|bn?84RbfC8(XH*G74sMd(J`5$7
zqXkYuHF(gQM?PW~9%8-sg3bIvV50uryFEigM0*YDT|*Vb`EROW${10Hb#B6N2#xbA
zlidQU!cfR9Fhhu76o)T<C6P!C4Gr)ZppO-TIa;QO4n_rqCbYas;k2XXx}QBggmt}H
z0F#IIt7&+LEi6#Ju(q-~2B3%_472&kKxY88W*h60DL@}w<>_8AF|kVTGxQ=mwr#6n
zXF9(o87GT?2&*DICq$Rs#*G_Mr#|O+?g%{5S_IQ!AiMY?>=UmFg8ODEfE`p<%sVXo
zxy+-}Cy}dZQQuAN#WqinWooWaNAFU=8cGW}9f|O5j?$LAqE1+oL3dQ{2|2<Q|A5B+
z&Ye52_`F`cfU+%@*$c!ukg8>bfLp_!eTX2>D3R)UIc@*>lYqCTkWn!L@WuCE?<TB)
zZ1qv<hK7V=l=b4p7jjXfP~>a&^~)d!ETQp%GVawi;KWv<(Q%Oo(uWSNudUZOhYRcP
zBAPom$O$QW^aqy!T6740n=SkPEK&Pp|1JL${SP?wNiv#`oD=<+fN1=Y);I}(&u~4k
zY%pJ2bws-v5;1(GEJtGs@&7mitg!~gm$`{f{)D3!VeCai&M171&9M^3Sdiip5{;hS
zj}#4i|G?A{1T)ke)EY6OpCs3>`*|8=I24Vub8=V|(|S>1-x7k;#Uf~&=06t~Yh%;d
zrJ=tg3;Pm#*k-z4ELxk-evYtNqQM0vq09#)$xi@F(77$C9p^^<VfubMdm#sE^*^R5
zRCiENsTvt3_S?&Um`5PO2J(D5ZVV^oB_bQ5>khjx9(^c;f$F3X%I&ZhfBEud0?`?t
ztFGRO401$m3O#$oxp1fp;R0;vL`Cm%M+Z^-Att1z230?FP9T?GMuG;DNG_-vZHCPq
zP6;gvd?X_}6S^0*<gQ=8{=XBAZ&*t>vk{eifIAQk?JZvc)c*R$kbBMHybN(VCh<}n
zB@XE#Ui*27AzXj`hYHbulhFTP4$nPul?d2K0rkjL{sz!z<l|d~rzy=1;U(_AHe$;v
znpE~sQ!|U*_D}dG!mx({*}&oi?L}JO3i+N!LqtwkM<lzhVTp;OzheN-No@)Oh;$=o
zxhzk>0jZX47R8o-!~bL-osUcb-xWTuBY}$+-E}HVK@0JJ)})Ksx`;!3Q*mOwGK8-f
z+z63~c)g=qSC{bBf7v%)s2m~hdW4LWOk^c>oVM9iQGw1DnI?Gm*3e&DD-KJ7EeQN{
zJl`ibdUo~<=e1Rsa{~4w)S?ljSHqi(w&uqG^AWpupTY(v!&;Vq-80{3KIV0B+qr0}
z;J3GD%R|)I?QKcZdufvkJ}&muJ#Rjf8ZSVy9iJ9_7V}WTR(<-r+5?(nJ*H*M2u!G+
z$`y*kGuK+N{}OB_5MS&NGwGthZ{V!6RiN#4Mr#?)v6`HoevWGT)2FGasSK()>Cc~|
z*;X~p`jKx)2)uDoIO_`MGM-&OqO%Ig8WIdaLePhioo(AM)N6tUGF*{=dGZlmAB<?o
zpAauSrjn%>ZoqtivLQN#=YhD;<PTg6p9fGe1qB5M2Zy(acO#{sIG+2&5j!n`Fr&dv
zAX8u_l-Y4G?G(5)?h)!QdM1$sh|5CkOHvRU<Z_4&;O1byt0Z{2&vfQzUi__0?+nJ)
zL<d%>p1Hu*5327h_Za9s*!;`0oR{VXk1@_U3pmydrwODdudLY(Qw=zJB7JG_;qh;1
z$8T&m0k8+&h#UfrIm)K{I$0E_W%PpB?x#i0TZ0L;%+68cHQ2+_{Jhp|B668C&h1^C
z2GYc}&5=cMr?nB_xc5y>iM9nC=b?9J2hJ!Z3cjK~XW>7|m>EO^9)LGw2{h_{1HnEj
z{mwKhD#{hddu-l}&XeN|3=F{g=&SX@NlX_A#V0)$?Yl&mW=r8!biMY$RdMSU`GnbS
zG<?7up_rzHBh}s+K0$8v#_5cJlAR9IVxVFBlaEbmybD`9Ui0IqDrN$61BWbe5OLt=
zuU{jP32`}X5bCtWcV(reI`A5#xuL|1(?A*RV_*NKTCH2)ryONEl#ssjHb;=HLy1Dv
zhKi*tJ*HqLq_01ri3DbOm!D}7D?&2{Sg>!_>{T_0a5!x%+X@DFIUXF<*7qCh%iaBL
zG>#=r#UJL3`ykb_7g`)4iU}ChSeSo?bVA~YBb}=6!+?VBIh?LJSUwGq4Ss2YV;lg5
zfJf@py8xx^XNRNC9Ed=#V<9pz`hzqvcU1w;LIe9kaIgjlAG(y_{Z?}pNAryUd3r_$
z(fbOL!p4YY3rTqZmot~G`3ty986%vFlPp&a3EMD04NQqZyoB3WdzkKT;?Oog-_2KX
zc!A#q%_nyo*$*7Ci>>+L9IHF{l7X4o2z{c2YQ<+(=Y#EnK6`-Z7x+9}_lfTsRz05P
zx}e}-PEPW^?Y7ln#zQ&mq77{FdRr%`+2p0JgfQ&an7`HoIHa@2->jp~a2iLS;t=bj
zV_xWa)&A+<iatR3WEH(UyA%A7Pt}2-XMfkQv#8}Y!NsSlDv|N--8)F9BNGt@JkG0a
z7~Iaz&UZOSzJQK^+=l9%XAz|n3((Tnj}N{GIKxZ*i=qkGkuZc%zkt-!frYU)|I-5u
zWCA`>m<>%(1Yo0S#3y40(Jy|xSl<S=EXy;tz=p*v`hsjTk`ZBU-Q>ywDL!6a^=O@l
z67?OXDA8k`JXH%VKYf)#gu-sux&lSxg>yY>s*E81K}!tDuf+p0EeTT7Jb|l&SOb&E
z@rz#2q0nwtc1%n}WVj}x6d%3g-1!qH!Vt%>LN6|REteD*M}c3qFhaF!Ds*;MBhjY-
z@uG!+osUm#r4wTf!whl$H{OL~adbi8x9m7X7);PqhjTh12*-ZFW^+WMgEQ${uKUK#
zJ9hH!xXpQV5$+DiKgW3do}Y@r9`gl}&|0$ipHF~2M|b+PZVXP#X=!fOPDuw3hBkXR
zLaodihuj0ao}p&!9~|UF@WK)$Oh<&)6Ji6CJzQUvJ_{&}^VrSY@Y20^Zw{&rkgkc%
z;Zq=K>;H{_-(X@m2E=px(;;f=6&Nw@{dVl#PMqEswLkL&JrYGrmIDwauIp@(B(4gn
z1}GJpxUMv&)?%WAmzIfqzXRLMc}WCOXu5Q{JJhkCG2jKD7BE^@VuPdQ8Noae%A9;9
zHi%R7_Bn0$kS|;BeI4KP$a-hPDo3BfhE^3y+n$Yz{JUIQ2*Wa?np%%)!iQdXLl&Ta
zpb6GY`fsyxVzc6;Lw*Ugr)kxJPC5Syx0yD>ac<Zy5L^^jI^o$S#;C<~Uj{Y>JofQ7
z-Z&ebLMPh^rRp0-FxA53B43EZ{ss_d+i*oh2fYDUM+eYq7Sr_I1%5H4Mx4N=L=bkl
z1Q?exaFio6{pL9k2W<Y91Yh_Z(Kdpq=H^N^{7g(X#MgNB##pJJc#l6z7t+x%k{i5S
ze?No($_958)_lL6(1V8_PYNW%!m8s@VJv5%z@gf+mFEhZT3Ze1<G1<%_w->aq&(;&
zh6>^dm->ZwKjx?k_*We-m@lxOQNC@+ThMGHbyweRc(R?lbnS_t+o2C;T=)B$z4Ma)
zB9AOO<{>IB-jQVzT3EOSiUMO3GHW|xB{WR<l<sB3!s&xp(UxP04#972ZA8w4mu_xy
za(9N-<{FF-PK!sTr9OPv-_K7M=VaiN3nVCe8yhXl3qwK9UGSN~Cs+vi6b}W{7Yu<a
zDk?+}gHunn_0Z5SwD;g6u=W^uc-n9dc|MgHut1vZ4LGMgJgTuei7xxsuW_0r3d5U0
zc5s7Cc{NMOb-*#P!P1Ee?=if6k>6mVgAdcFBacf*Zpjx`G|ctV!UB>Pq*&ZT2?#y-
z5e6J3USsIj-}$rf#MGSkkE5H8g$2H~nyRWKOtU|WO(G8NaKOC68Haccha1`^nwp!v
z&nniUYwydK2pH0!k-o<Xp&7>Q-1Y-@_V)jSy|)aja_ypp*|sQ(3I>R%2uLG>7=#K4
zlG1{NBHbxSnkW`1B}hqkHv*!lgdp7rg47}f79HPs^qlkk{eFEv&N*wp*S_|PxYo0t
z=f3AX=NMy-DYI|rdr;I!RMJaEQFp9n_0*Ll-#v#8-NmVa0Aj`p3D-HGD=Ifp{tCF+
z=b#e;dO^iPqwu`Ed0??2&5o~C0ZS38OqkNl=^B1^`Bg{*Vi(tmaMUc1XOWa@(JI1=
z5*aa3v?)!My&1PW<{?wG3VSmwb8Hdvk;+`6hBEsO#Dh372Q9VjJUxF>^UoGM^VE6h
zDWd#1>-{=97^B5DfbjSO3C`&G7#z^Zv5m~llaCER+qa5$4PJM4a_TB}5tP~I)J$pd
z<JVkuXW=gYi~ujrIHPFYT}2_qX6@J8X5;Il3=@K5QlyQx9ctCpS|?{ZRcfBzm$k)!
zu8C8%vAL|d<ZW@zV+T-LpOzmjIR4okb!1D)u9cxh(S(t{G&gRWT!#tmEXSC={)eAm
z_{#cC)a&l_e1B<LX})Defto}zoxhn%u|BzqnZ0v=4wp`}ldb7=r(d3;m*V}jjg#=~
zSkKv`R+}<$z4!*%a$vkFFr9Oc|3TWrv|}?LSx<beP^>>5D?##4Z7nbGdLH^*Td}B5
z;@LOWZ<4H6dR!bPFAqO9viZTs6WM!G{7*jaTZ(p(+UgRJ2ta7yY~iEsjLw%kxd^o9
z&z{AK)5G%FvfseO<Q#0~5e1$e@rNm9A;e=?+K2;5sgZ<X&aY3KFHt9|r6hX%7Yc}s
zXxdqIMkFN~QxBu-!3=IM&OgrP5Zu9XGP1Ke&i308a2&w0_ygW2Q&LV=_B?d#ko04*
z&?#x=idtWyhTRG_5CK<!;TWmMR)D`@&D&G`NCfM0R2LzPPIH%vUyFw_gXlH{Lglr;
z;(E5RA`G$4n73U+)RFC}$6k(A2wwwNeRVhCKZdIZ7vU4S%|`~Yj9HoxqMn)15r<d8
zd5|Ttj6diIvrTL(`19=N5>G+CZ32BWPG%m($VaI{J#O=%@%<;aeZpadCUxSR0rNmK
z<aRv4wmkzX2l@6II(pFF<7|5H$dNU)m|!=`E952PBC%)88e?CE*-jT=goCcGu41H@
zS0(Vq=pYQGIKG+<=$?}lYj&J3(k{k}IKu8U5~;rJfHWir7XIMK%v$(6Qgw*4frX?B
z#AN2-q-s)%PO;``<s9rF*e38zm7^!Y4X0>uEFtM=aOwheMWAHv#W%|~tj@>aV58i=
zeb1)%SP3|+o1@J%E#$=R;+rCoc`lCwqzm6hJZ>BREb))VC9!Y?np#irWjb_@9NLAm
zP&DufxB6V|LEvvVatvu-K0;vf*LeX-yB}rO&v<-j@A|D<w)(Gl(z7%bDIov8W%>D8
zFm3Z;TL)`ouis_ApH~|h)~L=>eXRE7`7nJwW9Y%+^`C3uJZp!;4~M@DHjbN3Y_CoV
zb2>XidY;51QapV9yN&#}bJ~dxrxaJsgfq6+utHO3QkB8>Yx}Rv)|`y=5$z<$JW@BE
zzlFZ39{W4VanvR$g1MGdZNUMrLqo;b{o}LK8$THid~sn`FgZP7TvxvshlVM$M*_!X
zXn6!O>nkpiw_?W+SJ|B#o2D$KRS$PN>dQ{4D8vo_<E)2HAZ5`6mpvcUvY(b#0BuT8
zibF}v-x!JLe+6CzuqSU1fkh{XiDfL0nEsE=D5#>i*&B`ckf*h1skoSD9o>dg<3Nm5
zr^+R2{yhFI=kHGIBh8w9@eN3x=$)}Pv}uFYfz~#)G&{t#7l9ikSmgZ4y?geYf~mcz
zXeknJvAOr7C<O%=9bn>uWE$=;`p*e+0|yKaE^=uC*dFX?`5$`=a79G*J3mht4jb=d
z&VfS8*gY0|cB1b6!?5Z@eK@WoX-24w&P2nR@7)<L*6NnTRdlq3d#n|#g-0FT1}Fo`
zHKfJk2)XtSkFB~3V-|<`!TgpYChTFla)OmrMgb@%4j<Gz>A(layv3*=;g$gSX5r?R
z_n<=K!z7lL`%H}>B7U=~A0IAs9qlxrlG@s?17QWLiP9wpvH=K~myCA6=HY7L%RsnS
zM@2>P+d<b9?G;(v0Eif62S-f{jFC<BaI;|mC7x#rYMD4&;J*-NI%I@6wJ+uoPKHts
z(ZEk0i_hkQ+tGb|HK3LO@?oHjNC`GyI*XeRWmbmXdhpk2Wd1cgMJKD69N?ltp1(FL
z*J7E|jD&$1?W(x2%(%L`#=6m}i()*N#jNbHIc)pNx0@U#nBW6BobCG<7=-o8w~a59
z<mKh1R(|q4`CWMHZ4**g_S2D4&v?;S?po#96<<r}Z>Um>GBmjd*qMH3GrD0U#(hio
zySr@vDPgm@!yeYIAD`djY!>&6OP13x5i(n@8Bcwc`fl0kPNwY}u7U!YneK;<^PMd`
zeJ`CvdlGl4?%s7Fp24muD${MCoxyNEwcTZGEu-G6-ZLb!K3<DyvgBvW{+3M1BGt*3
zwt_!C+G5dG=VS_1Ml<tg^t9FMjnzup<<mlcwtff<(SB3hrdOi$w9lpk?FLB`d}oa7
zZo<G1ckCqe#3*DT@OMYn?+}K<I1zMRN3op-mSa?*w+>_ww?2EK;;t?)KY^|)tV)D5
zl_9hv5VK*HXGeVBO;%Rm=^@@ljzkp|6p#^d!6$+uzyk#ml8;dXunB{1ACy#>WQ6It
zm5)787ff_&Xli0>WFXRjLE%!26t}gVhF%0miOuhdJ>>B&92x9r^d6}yM-iYtUk^_R
zv{S017Do`(DY|2U_zQ!+2Shpqk?a6!75EhTaNUCGN=Kd!VkcMx7a$#I>;v8c7t109
zb#fde`<ieMK?y9iqX1?Pv1vgw7IBxw#3bylH_=w>n4lp-7;d72rATFw=_Y6Q+{_0O
zh#b3fY6X<SI*=3S%lkp4_p+fVSc9;Bc5*)ygq5f~t5ZJ}!ODkCIWv8V1QrF&u3&hR
z%%VLv4E_k`d1rb1HcrmB=n+#>7gXd=zL<iPJTu6k5jz4fc4Sb!9Z2&L9dq16)PVtE
zNlFQ(u#s5oZq0XJbuN4P4J|BRDl6HJ_vvagqj1?>qK=gI5ZZj*Ei5g=n(x}-D&V{n
z<~*SwEnS8G<1oJA^%LV^li|&D&Kv9MhPKcxasHW^@nxAKHOo*x!VX_tSb$If1W5yo
zd@!d4TLeWP?*|b$e2(C&3#>B;T80$<Hs{C|e$tiu@%DZGgRFYG-*r5;yjJ=mV=<PM
z@grA6;dc07gehNDyWhL!TLHIE1zjw-a%hzA`FSq7EUOnMyBoS1KHdzuvdTvmoNebd
z9*USZ=|kUzN_rA+&+U>mUX6slm|Uu^mfLpT@#W^0ZNj;K+9o?aten<Pz5M2v$0fSf
zyRh?;`|05akv?Dh0S{nUM<FOBqI46?z_}id{xj!i(n?AV08qg1r<9%~-yB1E&W(=J
z(-cK`pn!8~Ad_y|r-1fO5p+=@ckpVV76`!7M>&fdVZ8JJcf_B2;Z2O_1h9@e`@^7G
z3HKpIPuLS7Iv*Bwcg3d%1q96O{s7Dv&y%n{c=^(4iiECauz-Z)ABJ;WqDFas9uY*?
z_Zmnh7w{oaJ7ImHP*Cvwi4z1#5f!7z3M$qIM=`9#ZMu@R;r-f!qgu#$2-}`g$dP8D
z@e`kU=!tG5P&gP#0K#L(gDK?G#3sYLEr~=O_I>3KYOX!ZB@?9_R+Ne32Z&6Z0XdHO
z=OQiOB-b}XjDt5J4GpQ6y1R)@E4<~s<iV&tPBcWS>$)-1-LOGrW^V3Jz^OP<&}+|=
z#5$M@CpF}ZSknf`1+gLg4D$i<@GIuZaBaJcx;RFmp!u<J;Yj9Y(p_Ut7q0fpls+h1
z2*0`u1JeWMsZL7wv=I1bo<yR2mUwQMqI+`NWnhB>W`BS=3|$ejL`1^TwtrI@{OH_<
z`X?s1wWn^hMvKxp<DVtj`Z|smo;m@>vf)R!p^gP}gm2v}z)_0OLFiHJLlO|6=einD
z*;ZRxyjFdjHRO||HKxVhsa`9j+tRMl;nq3W>E$8EZ<(Q+x75+aKSSp4RJYX4(wB=$
z{;xGc_ThgHTg{6znm?}iYRdDSt`ss=Npwm89?Ww1O*ltHxQpt%$#|RrIaxk6w8WI$
zjBKJkUvEuOs^13mJql}#_TUj8VfQx$RSj=vQjQD}-p~lR`lB5EM_77Xp(HCkJwBx1
zP_H?Zo<4rO4?sYWfLVZI&$o5O-DE<8%o2mD&@;`LL#8o|z&bl5)s+)e&@4BC2D(_6
zodZ)F9j&clEcrfcKoNv5T9!dtucCeguZ?9j8qCME?SV6M2Z%Z2`TAxI3M8yR0nQ~c
z4&#yZAy01zq9MlJILBOvFEQYpN)Z)wk**+3>F(+R!x!-Sbq4%|yC!0{Zh9(tjqt6!
zVZ_&`egG&l4^J8@U}Gh;fWr|^XXHsh0>W{XrEQu{IWdC@^&mnd;Vl_287bn3`2h`w
zI5L|={UFbMK`&U_+)PFRK+F^{oPG8cXFJTxfH@e+5xW-Ub&P@=CYPYj!&n^f=*!r8
z=8#M|cfx8n60O{7NoqAdbnKk$>{V!G2gTA5=B(oP{WAyZUTq2Vl1mqjbUfCNy4X%R
zTBUom308}<3;_&bQ}^qSa7LK@e(a8NK7gy{Da=rH!6G1K`*j&G`e&oXdgKDg1|-G8
z)s2TGve%_dvoTPhkm|7~j68r!8Px=nfjkXY%^<tNX(>I^b6nyXuCDu}-OU|FKaQ!J
z9G5(G^_0uRSiv55CN}2EFSRk3g0?t>l)JPVZaXCS+3F{j&BA%!LGADFehV7L8qNjG
zH8UhRru@wLU~yM*jk~$fFNbaUxJY%U<M8`kU%#CAV!pl2c6q>J$LZ%HIZOHK4uOMX
zicwEeow}B@=L0`+cYEc}oF&<~+PW%9zj++=*w<gVn2%hjI?BS(aVGZ9!O#6RB=1>`
zt}F||yv@bC_yh&4v{L{0Y!QoI{(UE*g}6MyPgWJQQOb&0kuzOG37c%b^=gxf+#Kh%
zDT`<VY1FF{RnmfsJ5*={Tfzm`S3VYRywY>+@>P}~qebfG1ZBzGo}=3?gZL#Jx7`_F
zfu?gvb2mLby|Bv+7c6NS=`7mPF9|nCUaZRw47xLq3UQX_=CR)k8?l&v*r)UXsxob<
zX-D3<_8qlVOHN`oy1wS3mLVMT;=?=^lZjiz%(kJmM#yqzlG|Ea1(Cgbq}bvqNJi3-
zSB-LEXL~W*q!+VGX6=pbv4LLNzj+lD)J`~0;cXykLoH{5u10K3!f7!&4x?x$055OR
zkr7rQTrq}GdqGQ!XnW#<^+o>jUX+>smM<`u&=If-mJB+0>pp+Jjc!8tegujfQ$S{n
zVR-kOm35HWK)Nv7Dt%|<urSj*{noE7+~vcY_lr05<U=h&6Ct|X9_40dP@&hR)M1n`
zuUTV79L5gXk3Dr>dW!6>uQT)IZp3L4Frt1Qk{MJO9Yqu-GIDYZI92YXH%xI^@~NJ6
z*+rQ@Jti$V`Re%(?zy|Yc4vN%G5+TvD{_9Zf1J^b@5Bk#<Z0bzL+y9&ol7;s!j7%u
z+9kW)R>POVcbs$V{du2xD8OlD@Yl>oNtI-;s_4hhzw13eM?**7F+OjZw4xpgPGac(
zkNl1OcYb!T)u@%e_<pu8JwJWqTb$rxOl7=qR6wY1E^0dEvbYtz>0MXDp_I@@Zw1Hs
zJG);}(S{D_4e9kx-!)wqT8|r@)Na+)@W3TjO0pgZ*B)DZVkE2df`5ot(2ZWOl1|Gb
z-G2PT!f&S^DPdXi-_P%L`VCkEtL|mjX{GZtwJHYZ-%H%N@_T^sG5--k3@;Ln`kCFz
z-^}?usu}=ivXR|M<ZzH>CFx>(WOu36qhqmEbzMi%*DU2o^@E+sz1NxO`(8bM;)M9R
zTsQjcXAc8xD=_5RK61MO)q8?24V#iax=@#r0J)3q^Ic!&@_pq?ER3Lao*^?KQ8#v-
z+6Wwb=h)JbWJK<U7kou$=Y^z!r@60`-XRA@ld!lKfi!EkYRm+Us%Pp~qK{kK#a$Zk
zJdljjCi3-zZkg5DR$cjuE^V(m;Ddya9o^iR)!}FfZ7`0jR*+aoK?X)gi*V{f2#ubU
zi)ZZg^djwdDZ<Z;KW;dcvr+8NpY=|os{gU5fDoHmJzc!iP_mejlT&kLdxz1N@`~B<
z>BCgT$BdEXEqO$}|FNBJ$y`cS)~|c`^F@MBoO|}HW2EC(H_`}w6WYtGl!#b)#qP__
zU28TrljUW)<5Wr)tVk=q>k5TuxM(*WW9BlbF$l7_JL-RCZNpRLSMqx-<O@f_<r>Vn
z?{G)@Nh*4G6?Z#!|0E9<*9zz`?PJR8Zex)XTzMHC8|Q8}`>xzie7wAOxc7^Z!Q2K*
z-GXBK=y&ejiO+9PUZMu@yZTw`k8^^k?#klNB$vqY+SE|0Ao+Jzv<*%nYdKchT5I##
zKkNnh9UX(}qdTA`jK8aWPi<43t2hYR;LVP<I>E`DZoOLqqDqUb78J&D_K})R`T3RE
z2`qhRmt4)ypE_Btp7xb5fL!IRucO|JDUFxC%6QJuW!LkscoyQ1zlUDHJRn7N_jb*z
z>NB+VJ#3Y+NWvR@%w8Tmb(wCL&z&4tCaFQ8(tshnhY#Cx@o*IvIW5mw<R0j%+zWv+
z>R(X&gry^RFg*QL2%FIh3aST`;~r}2bmZ*q7QM&?jxE)^8D}27dymsas)#!3z;$Z4
z|6kp=V=1Ck*2Mc>#m|>Ma~hfbY<P-p5Hw1NDTt+shW-r^th3$^;j0WA;=Hu9NkT^s
zd}(>5cX{$lbtSaims~P7;H-oU`k;-6_osHd3JX+!Ibv3sd*7_{lxSs$e_3-_0!CsW
zF;6=x)M~NlR6TdUvw3~6pqlfriIvH3RXwZH4{obp92EQdhrcz1ip4a_`zpOrYE<5Q
ziB~Zw-`|G5<L2;pZ^KklhN)OM)s3y@%^&upIqNix(yd+L7AlzHoCxAyP5W3Xe(C%r
z>ak}Xsx}q+qOul%LrMO7B)8o5)l(3n;jxUkGnnDT>u_pD&(h$_ZbeF;4Tq(_Zy(uP
z(UHB*XIS|Ddg%E+uVH7t3T{&$QRFb@XsRduTm&EuC#MFa@+b>&&vQ{^YIHAkUuOi1
zR0v7{`93DjL`6nI29(V>VwPqZ6~}C6ZC#fWlHxfWr&ch%9<aIjh1)j5T6kW7<Bc(o
zT%>!-$dQ4$H-{R2+MU_2c7P31D#);M4m54sy;~QQ1vpbG&R@uj8e#%5na?5Rb2oud
zL=Lm%ac5S4td@^J#uK_m7aJ2}(sR=cO7qxDh48l=+vw2`AN`=Um_E_P-@fRrAGzI$
z{0zC43D5Qu+T7X;>!Z^ZZ(e2@_s|?F{`f%Z4%;Dez-va{@k#wT%VXJ-$H<%#eyG`~
zE{XBf)tP}<IYPG-k`DX9Q>R{sj-9ja{Ya!?ea4U}nnN#X|Ew7H$Y3E1cbUK$81+O>
zsSxUf*7ZE@q`y0TQN*u3MzZ{Sd9p|7+U<8QQdA<90x0viOn*;K#qb7Q_i(<l)H!cu
z!MR43F}lS*^SvQ8zMEA}<j{4io7<~p!;W|zT)Pte>vGQTrr*nH6N2omQhpP9L(?av
zKNLRTMns5ZxAFK4b-ed~9u<Q;x4E*O$O?QjRS(w|n0mG-ckJc25#22k%kL#)_br_6
zzACpoaonS}{Ivnpfgl`^btWFz1V(ff4IY>j#JTM<3Vqze5m?UO{76v~E3-7PvQa@^
zUI&dGpr3cT-b;&nXIU}mSm<sFX0~<K6Yw$oKn<h}_)^{|8sAxMtoXW@R|%K`z=D*y
zZzUL0wmNe!TLJ>IePnwB{oQa2D*yV`u#n$>Y_C-2XN&B~Sme@NHPo0X*4xownrYQI
zymOHx50e9|!UmpBH1y{bjbt+DwI`f(-eqz?Z+63u<L&LTkyL9ULLDAlpFca)F4u`I
zC$Ajxj(RT`%}?jNs>(-6R5ABGb)e8Nu(04-)d%$fuLp^T9q|3k2O2z5=mMYHLl{G`
zM9lRN6I;)lY5_x@nbGOx{<oCbtGqY->$AAI=WQAtDf3!9R3{F78dhcsFBgKcJ)C9N
z9ht3$E$ph*Md6MKeeH)_Ef(m#oT9Q6_seCeRx4E~-FX}K^08djbj?nl6G<5rt}#Bf
z>H?9n-?Yt)Dk>wlRGC52hcJe0g|qOIGzLV0`a`_|%X~iE{Pwt~UukJ+o3ZsKOO|V=
z+WDQG<2`r!x000`8^<eDEO~0#iw{)UUl0<zj!BfcJ6=40jyZhq@5h;mMzqw;F!2fs
z28M3FS>f_JG}m{8dUrdWq3T)zkpWsYNyWMxE2Mv2T-K7Ka}`DvV-enzwCwt67qy>c
z@8Eo@Um6pX`arkN<Ohv!XjTB*c=2!jq0oVI#68`J%n5MD5jtmK{jFb#HaM5WqF_j^
z;U&tSHFu^b(yA6IWk>aEss)^}s=k+g=mq}0pL%cf_%{8KuR6|Al=fYRms{|I8;|H=
zJ}zycMD_CA`0?1acbNxXRKM}#4^1+Xe(3q&m5poR^|yy^z26(PTT?~p7z3k``1bAF
zGnen08S`ib7Z(pTMZ08ke1w?<QBLDfhtUoYBNx=iz(!5NF2IEr)jBY<MjR53v5j|a
z^^kl7)7}LQfKeWGa4f<1_Ji&lg-)<Y4^S5f2ZBvZJneHFR+~`Asj~Yo8P($p0Xu4$
za#Vy?<@0HyhrYgzz>Ra1UywQH!B1#IVm6m+33cYFsp4vpk>=A6jsmTnOR}tpKh(bF
z7Hj5u&_GDOq+`3xor#C>pF!?2jXdiv9Bha4f^<&=F&^<#L+1PzqC7%$8?$F`H>=#Q
zbgyErA~dA@!WFlhgm)pDNY;u7uLNMd(*BveWG3h{U9{s4%WtiY@qFVM=K1p?|18}<
z5)XxAUPQ!E#mLkcdV@x?Y<#J+z>a-=7}pzdH^oTJ>etx0--*AQWW%22B(%lKNd(n?
zeO$MLl`(YZ^BpqMi>10TZ@$Q^HTARhtN-cYdwGO3ebrat@N=&VtZXK1Pd4t>-C49i
zs^O4JPu7mP(;p;Aj~@AdT7Y1>^S?Eu3oX0XchSB+=Qq;l)}b$GQp3Ja^}i$fSA@6;
z+>hMjVr=EfJWg#KHgfeqE_LPS7~$tMY?1CCn6_O833&11M<kjMNymrj(|{+y!@~3h
zCT`0_K+<M-NJ^!a*gtsK(Kj9@Ap<?AbQ3NmP>hg8g|J7^Lj`%PvM3Ygi>i6BEIyib
zF;3n?1)CvY7gN(qJxUQI4GGT>j?S2E+qYkpmWDyO-;qK`g%2jLj&1`CSdd4NZcYWB
zZ4Iqm-=;2ygl451I9dA2mr$Q@^vY!D173yYWp;+L$!E0&-9VmmGo8pSZ(uVzXPB6n
zKuBSN5gNl?7-x%F>Lx*(g^$FoUeiX3Br=qgGGqr(1?x%#j%7=wgx?G#2EXDU&N803
z0wj$!^hK=jG;<oAG=avrKJXNAF&-8WS@0|YUxs{{hM7c{%lR&KedKvPZEcv?Pm)r~
zbAFqco~q^p>d*I@`6a)XdYO53=?dSY7kgZ;XzkiJ$>IN5P^tNn;d=_<Cy{%z()~>J
zBfl%h+?%?`_+!=B=@`#G%Bg_z&1e1ID(El8i@U77vft=X!`LM|zuw|*9P|YJg>aPO
zM@t#jRgG8{ZjPzKS-w<AP2zRqRT6PH4wIsGJUH6HC>MxME&Ms@N7P~hrc@A3EX&ux
zsx<Kc#0W_nWF|(}`XO)7ORnD!2`9Bf?#`}^66A{w-@I}60Nwc4tP{NBKFlMybzLAU
z3iUuk*b<t=!JV;O;J6fxPG+>gWpZ>JYy}fd74+2sN@fV3r5hzqi{na9XqT}bJNAnJ
zU+kJkuX@4LP-eLly+eq)frgDMYlfOPfoI#f7+<w?S-AFr#|8NHbTkX)GKQJ(r(yBE
zjoEH5iut^QWGEonYa-lkM6Um+#rLw!7mMfSpWQ&^wsui0To{{<(N(M`=DBSA0MSs?
z*eP%!&)B!tyhinL?vT$DeL?Fs8@ng`r(|T_vV9jf=+k3eOBH$L&Pnxk`-8d<t}`K8
zqfu3i37dlAZ2ZSLWJZ0$Ub_ok9^CE6eJkTugO3(yZ%faf4~OrZYxDr64G80X#{102
z2ey^>+nR!}SvtJ}s*H31YYL1NEkiTE3<6<eX;3W^DYCAV3-$dC4@sD9fdu*eVw(An
zk0<%PTSG!abMi!e-;7gNG^{IQR!&MRF<bsQ%VzYQ#_Pgl2_S&*79GX^8lL!qL=~87
zVO9p(h6x7k7Zhi_Oo&NU(W3=I=pg~B_S$d)=hqo-i=+P1%v_@Ad4jlLNnz)%U1&&k
z1-P0WcfuGoTx9l*@bN!9Sj14qbU;H$_L~Bc098-r5mHw{)|>|#AOxyu&6fX2{`nkV
zS9h(3@stG^*jj!G>PG}bF)<8u591_nN!@>N)CsZ*(a)SIA5%`$3-Mi=$eVuHgbt*d
zP!hPgY(di|LK+F?mDGxAAAoh>VEvKAqoI$Z$B}mirCOw!oXl#HZm3Iwit`-Z(%Ad=
zt#h{oXMsOhw#Xh916fi7vRd2=1RyeoJ<M<!sG9io&IM|RW4kMBOE_;c{mf~stQwgy
z4Chda8arvIBU{L20$g2IuH1`DThlqeYQ^hJNLI>B#L~m)5n!R3!3P)4Pt~}%6~#Z-
z-*#vy{3+w!Hrkwb9cO$?e;};`D8#^UH#@_wY|a1Jkw9g@>dlS!F%V(})lP~#(fu@%
z$%pDoy;Mp;VV>MvAUHz^4ZHH+m$E>R5ZroB<7Jz*K~=?=pt*hU)>2~d(~%8ZH*dD+
zE^PkDhfxk`L)(AjBoWAyZE!G!a3FRh|3|O@$L_cTUi_HSROtR>&y&-Ly2o6)2QV#G
zdLNi_`0NRemmLK>h>pTQcLxh|bNvf`SNOrFn)D&|Au;D-IDk1lnSj^<>*PZ-2${r>
zP1%^0k+nCI-sI3pD7+yiK?h72@2yX8M$D14fW7IF^Y(;wdgcz2pm|$eGic2Ph)<+T
zdC|Z~sI~-p)*%u^v^UMAYZ``lBKsnwkt1y{A!`J0Vq&O*@teERqIlji24SINCYt~9
z<;&6G@5l5w?s9|E<mJ8Q_x5vh9VI2k42*X!fT3Z8s;4oF7}C2qv_3jQ3?{e};N$PF
z3h6~tnn0Fb41{r1+nJ5&syK7r;81<}_jql;B9`{2OmKD|i>wa|kl;1255emqPE!$(
z#59DffY8rw6(hvAhc;H6(Lo?<*00O4u)H;K?)UrQ6G7twgGE1DkC2d#4y0ZxlHUKL
zbmq*@4<1b2hGiX}oE76Fg6~BId9`D$w2(DeFBEdg4BCgB&Mv!F6Oepn|Ct>ZxW?3q
z-|-JkTnG~S<k#Wqmc?)5^!eB5QgV`dVc5yy4;G6kJgrEjBeQLdVstxSZ(-hXxW)VQ
z0|+Z1HoN*n5H8Co?38A7+iRuK$_#k0NudCII-y-2nLdP4;P7cf+1Tc~BYy4(E{D16
z(xEPa0%chDl%&gjPqe8F{I+#Ao+yVaB#5?7h#5`7uq;p)(*6AJ+4JZ080rVzXmt1-
z#BgY4jdr&H1qV__aSJPJ99kZUd0!8XqAv|W*$g6*r9evfss6zeE?J(qptuA@GXHK0
zXeYbjjlFS{r@Emb697_h8B1{9X$+(dp-Hkp7=rq{g4FaDg(-QWdl_gz<SZC=tUH)f
zbfcdb6k^APDl@{pHr4$_23EEcrdBmP`HTz#m*|7}QiLr7EysQa1_tQ*ivA0F14Bu)
zq~nljA^OFUkzqk$|3I|qb0<~+yby~V<UM`BGP1Xp#^cB^1snx#7YM~T><VK<EkVMM
zpYx={NrPJ27;=?iKZGJRG<$(0LJER=>Kpof0f{49uBJ4ilZ0byrdbyP8E_D#3tt)<
zloW(2dU|wcSu4=S*WE2_<j=(mgVPIWHAVc}{=g?V_&n|c{(FS39PN*&LqSOJJ8Gdy
z_jjii0CWRuJ<e^PkYNxOrk#|onC$^od{uTBzAzye_94I5Ax<!1s$xt2*{>2};rk$$
zS%t?tzxUN9JXnwiDwr~YAn~od93!cvdvb6a26-N8LdtF6=uexG?$dshFc0idxTa4#
z9VVtz@Tc#}C}oO-mxD~J@VP;KV_D;|qBh$z10l(j9|P!sdh^>@9+w`|jq{i%b&K7`
zky(;Ny4IX1#iDz8V^GlB8i();1vS6F)q6jGyWhsaT0l@B(D3=yXYG!XLi+`j^K^fT
z?6r?}ark5JG|xUUq1x5&+y7bb^BW~!pdaV<+hu2cUrW_kL_3siA|!t}z07`oSVR!w
z_rQsv3=j|2fp#?~Ho$|VT{I023=r_BmhRfN4Z!x3xrgOgk$CXPaaIr^0EIN^`a&pV
zdOj4ArohzQ!OT#iLP=qUTM7TSlcAKbI~20)IxD9<P9nkn_SIm_Xnm48PtAq~rYkRq
zA?<U%%^>p#R2;lMI`TW-T^zyuA#*b`7m(NJ%R=+LqHmawPdZ>vT&ZcSIw$N9avu^i
zYicgDBBx1$1u@#*#XAe3FotO>I2wqCbFlI-9`jvx2y!0NuuYIHR{2zo65Y3oRgeVO
z=_Fr)@u_~eg&_(3%f2E8sxaA`o}X`RmYi&esYX?lW%DIjv(&vfHDx1=%0Z2~L+*<x
zK5&;&`v$ZDICD`_5ywiLw+P_;x$`Yd>o7>w%-cUNwdoR^`k*x+L7t*-RS8$bRI|$(
z&@V*iMLjr*p#Bqg4%(O3<55uU^SbmTje!8)C;TKZbexQSXT>w^0*?IArDmw4@-i|`
z88*-40Z{^Kr?`3^T?eStut(1#u+BZodGTT=Bs-vj33_Nokl3IEO&sRM#&>7zdJW;i
zjJsmi^AXMJvt#XcAWmV#^M*U%L}bub8k|UmHFO)}TpQ|k5PGa2U^`U&v+WT|3B*%e
zc464Re{gb=hyDT0rd!zmTh<<*+V8yUu3k^KTfzum(v7!?vK*i;ujjsJIq0|?<8;QR
zBy)6*yWL*Ku9}Y7jQ?cg#eFYMx6o%~q?=rey_BMqqyNwbtr_P8SY>=w1q>Hzov0Tt
zwn~^?XtvZ!wNZVN$f(0BfAmP89U=q}rd_*t&reU=+S{9hFab-8c8hjM<!f5B3FgeL
zDmvG7(Y9ty8S&2A!GTaO;65X}>hJBvG5QGdF|78=GwvO5`$M@K<?e>81*L5hm>z7<
z;GBcm-%RL_b8(G>?jr;|IOAZtgf`V~q~;KqXcce%iPW#T9lh)5S<Tbxu#=;Hgb-jH
zrp<~;*hBo7>6`=V7uR#g+Ij)>Ce!ib>}ts;i;jEw!CLkF5S|^dBMk_Ak<7J%WyC~2
z$bIYW9k=X0!daA`e{>r(eW=h{ljjdZ*N2G^5Q4=*Qiv;5RZkZ8+Kcx`5s?h>N~{a%
zP&Hs(@s0oCgyzv^JHxu4TkGvhhB~MQ-CSJ>xFIHs&^?;9atk|-T+s(L(cjOcy}+i7
zya;FeO-q%^lPn5h7>%x{EeZEeXs!Zcg7(0?a)l6&=wHX5wCYxKvc4xiXlMAnfbgqE
z#_&1dR4{#`{%k)NTJ}$kCwL&~`!hSs3Hc)SJ=$cNJr1JwTg3zhlZnT$qaFvhkMAwa
zOt>Rvk*<MLh3z!X6DTPO_JPl!ir>6#3Q|#oe)Js$*3#0#1<NNk2iCopVTC}5)J&KO
zc-dt?!m%)wTdeW$3?)^u3ZBy1+Ok!<n}fq5d=|+&8*7<o4^N95n-Bk3_b$HR?LDHi
zA}ZoFrvLiUSj&~oY|KxKX}(CXks|Y@6Le(iHE+C*eV~+L-RY<GGDh`Tq*6(MEk_2k
z-!>(e6R-6%tJ$l+PkOt0*bQB<GvqrR6e8;+%(%uIZdW_xYB(rc8#M1)J7jmMe^K+7
zd+jPM>lcv|v$hK@zh=CDs26Ff=rv842`o{B-L;NY(pfz>(POjzO1jD6g*xr#ACzku
z>x`|9i9t`Vc8`DWl-#%3&^tl0qTe(TCloQ%x7nMQm|lzUfps%IJx#-bBL-h~Qq$57
zM`jHg!~MSPEkQ8C8IF=ugM48HzB9z2x#}-fSgN!`VHhEU`w8VWhx^-<%5poHH31(&
zwuk8K3|>F-187Nt4O|(hd5Vd+V7T8qT~|}P1IQCl4@Sl(na)li6Ij9Lk{Vr1_)How
zRW&^^kY8AM6>uR+z-vdM;dXrp#Y)q3_vf0Lafsd^?a0YqnVj5+E~r;8U#6J8ou;xu
zbAtw|_*#WlYaA=2(h$OJ;YybTRt!jNgO8rrKZQ6>k=bl=AtzzdfhL52$jQoAUe{Gu
zbHIxSC-eD<S&S*|#x6tan3;px*~jQWDE{-6esHjGt=@ekCXld?z4dv^coI^MVUM*T
z4+LJF@nwvb7$r)iaM98ZZA9!-&=tXZf7uk`0K}LD=n%O$NvjpxafY)uW}iEI7R+sI
zYKO8qLIyEKgUCiKMg62sK#_w4rO5i%gDOVMJPHKsPQqCMtw?I_An1;hKSt;A|DpUR
zIwi>Jwg?Ak_@F7C$wnuh@|qdSN;fpye9@S0V7{+Wc_L)^T)kanD;4V!U7xs>I697z
zDypof$D<*B1%ESfLgBzZ^<56eGmmR}wAbXh>z!`9yE?C2aOtV!5%4b)+g$lOv;B!-
zl7eT%*u_uxAD419P0Kpi47a@Bz01w##+xLQ$I==GZHwu3S?`0@>Ljk*`kF7?e@3@0
z`S$Lj8x1#}yWilq9jXuwU(xVgZ8<l6mT4Fe0OsW45v#gC8Nnk(!3%gJC(s)?kerm%
zTlkMxSuLZkZO7q?OqTA9dAy7aI>O;UEdppKc=t!dFbvathEsjJngLwlaD2i;dN8eY
zZSSo%MT8^ZBJhjH;cdwt$c7JG$NV_VBFy%_7e&l_Ku-+9qvEWFkJJeSMvNLg2Lu>A
zOX8}IK89%T-%qSl$kA~Blh~=uFcdXByhjvVIzf3sXH5Fs*tD0N6wh#B)7Oe*FNkQH
z()V11@dA7wR*=u2Ap-40@wx{(Iy+0Tgb<I2;R@&*G)?32WWi^`N(e$4fKPvMaEf9S
z$b{+^T03|eVQV%!{=6$$?U1#UqK`g}Lh;=Xpk*T71LzK+VT95I%YM&?LNb{giZ}<b
z0c9f7*<YdOVOfL~jLd!+3rlsUP}<Ggw`cN7SGl0T!TC^i|LGktGsW0SKrD1TRviY?
zGWU9wcXbJ&`=6hmFl~lXwPEd&@{8RmL~DnOV9g=fRb`mM*fXqKih7_0(E}Mb%oc1Q
zyC&`N4-M@`j74ne#2MqU@axI>JD)+-3d1KGH;Wi{xDr^_-hLMj;fNB2j3GFph}k1}
zlO$Vx({ElSp38~*S)IFWBGb0Z&+^oIm0<6oP*{TM3;9NjNHF|}8!=GhI1>{ZOU^<e
z7(6Mjt97m^g=cL+?~}ql_tN#{tY=151Fa3IDQ7eE#=LLGNL)F1D<YRRfVRY?x7*u&
z`J{-j4$Jx^^<mn~XUg;Pa-E_U7(MeqaELV8w5q(yv}9F~KA<xvJTToi-52F0+0#WH
zjGcLBS`sL&G0=KScYfz&0ckd|XL3C?j6$t~JvC0dXMd4JdF>viJFm{#kNQk!rN(`q
zWynyyCN5s?GFcpFGD2uF&(<XqZ3FC{k1;+pskaBT+wa8x-ZSsLiUK5O<(e|$;@Zmk
zoi%e5EK6?uM9@J0NYYyPyLU<XxdGJ!5IkbFVA%jn>3Fh6tD>Q?h>|NgnG;=8c%~fQ
z8-s&_5)OK3HrlDM_)8<<M&D$j`}F{+6YlQr&g3pNG#q05WG%#GKYk$R13+qnr<_>)
zOrUQ^>KKw%f`(#ew{>tJ!@S>}0j>iy#q>?#V|cRQ8BgrJ3QdszP&x+>j{-m<1qJA`
zj4=u~=XRQ@5|}tx@nf=7U^Tkq#`^oCx@a%^Oxtv#lc?Kcu9X`gKuq8_NTHgIJk+0h
zc_}g;D*~*g$$gQ)EHX2js*8(@b)83;89Fu8)X)wY20K9&mEXO+b#V4T$1j7GCfu}s
zR4RrtB%n4W^^E&_Q*Op^d37Dd&@{<w@JYb9QyrZ!#86ZTsI$36mt{`8!6}yTBKo$8
z$=!h}S`QBo?K7?LiIIS98}7@kTkqjB<Vy@$zEN|54rsT5zCNr6(P|6`AG@Pv%4i(+
zJ_#zp7-p}b(T)j#U~vW}zji=58LyRqkD+g|5xX>XuT=sG|0ZQG9E`;C@ZiWwPtS+k
zmiysFdT6_r`MVVj{Pq>!`fYm`*NL`g%=yMwhW8S#cA{8x6SOy&2;mnH;Ec1>*_pj{
zKRx{o0h60RZWD{GQa^lH0V<C_!uM0<c?~?VPl1`$S$>vuwqE9JJ;yKcw40`^&X-^h
zt&{+xlRN}R$nM$Wyo2*`p4j$uvJ;+8w(;*O{2z)R*8r>lh&M&Yvip07nWcbO<_4ap
z&ptf+mc*DuL-fVrfFAT0WxX3x!LJ?r+0Ud<)?*EJ-sWxT>!;UeF?Tl1`&c3AH9#P^
znQ#OW)-)$i#^X?9gyRX`fGuzqz0WwT`sU;6ED=qSS}Su0p;t#-{82>>Z;epuTA#Qs
zBX)e!+=sVk>sz(l&ehY0dTE`Ls_3U+Xfi$WZ8h=Zj3L-k_Kl%qn6B#CICV*vJu0zu
zx{9OkJL9~N;$4C<Ke%Vltp}~cgv4)J3}*ZY#@VTK8+#=_di|uaj!Y|9vW*lxpL&==
z<(T)zFfsOr%NW^#6Qd-RZe{~UWFpB_6v|FdUqB{8rc0*W>qT&ZRl6~Y0TbF))nC^C
zMmyrQuE**Is87I&;({rhF5#VTgbcro)WN@Aacrfa{B85mE*?u04@`-B5v7{nP_<YJ
z1iR9f6CO{Fu3|?cmC<@Kp5V>S!LdFQzJBru6v<|20e5ISBBx?0hZ!CHtd<nVd&PD0
z{^t#0P~SyCQy%6@cke9D79q?F9oT!D>c(?Zw)LN*Lqp$;`L^VvKyhDNmp^TF`!>3X
z?(7q~$HBpYh6=wPUC3vHy|bw(3h|y#i1#c$TTf5Crys_j{fGm2Upm)e=Ft7@KogEF
zTBc>9$om%55s^$-u-~{(Gz4PN=Z)`ainz2<#DIcC$+~pD+^b}e^u_DvaK0@?#0z*~
zy+}rGiT|aHsx?``i5xG7ZWxek!d*HiKTC1(h|yD=P+K=`Qdunp+U(GIaM&0?4{XI=
zKg!!lag9NKmb`}*4&&FneZC#~^o@AMrQY~&6^hwb2iPc*H?l4*J>4fvK@l$*wyp^G
z_`k2le{SOV@25`)PcGs`{r`G{ANgN4&O*d=W8!ts3Cru>>(BpksqVwl`1iANvHxj8
z@xS{UW>DW^#G!%;l;)NsoW0+o35cJoj|MKSO#{DwZwsid9xym$a9nT;ubjUEhgrs*
zf>Y*r#QWb7ro4DH31E6l)CrTW|7kVz|NV|}zEF|@bcY1b(vL6#iI9NRx0Y6jtrw()
zd3lc7IWV8<{ZbG4NmSEqmcZ(lN_+5Y;!_yxg`#8N4IpV$tj&sqk7b151Yt@~y#CL5
zdjAJ5*#9lR@_vVIV&SeZqC}8}$yW9(&Qo?~P#Nr8T+K?qkVUp)cuPM&5!?g}u@y!O
zsqmi*#LPS=w8ec*Bxr46;cg^%*A?2;TK}gOfEIR)5uW*kfAPm`*|)r9Pjial75a#X
zgDj7UVEHlI9>oyCHr^wDgs_e>ef9U}zYlVV+c`Ke04h;T#M|7<_!}wThQL+=rq|6M
z!hmXaf)^Mh<1c-D`ZuD=%A;Yb2&$tQ=X}D5`Qs*H!Y0Mr$Y4nmq^;<r;FrFQU%x5-
zHx$Pj!)*i>4)Is1U@X8dIHU)2+K<1jy|&<2o<ULw(EWxl-M_cu|Myn$!JABRx#O65
zxQCty*th?#30_5Xv*bW5X5NVlJd}3|f5N-}Z|L9G3@NPOPp8l677#L9b$unS_xiu<
z4ZM|?jFyt_*4EbHm$=e<ivPZ4DvHiCrcF{kDQ_$B<B!BqiSNOUu|jb5QcjbDgax_q
z4ViKL_+rh!KjnA3P(%jCU0t0NK!;1a7yNhV2CoTN^E)uGb#t>+Ly+RN`TONrYCts1
zFDsz&4}F)~Nq^so>fbw^?=Jd)yykm*dx0A9CGXK4Ck`iG2jp)iko0n!*=Dih$0Pqf
zzyY}=a;u~}0+2)ki+x^4JjkX^#C(j4!Vxmx@^I0OXx@KBeD?^Jzu%p7A4OiCS*!f=
zdE9)OIO2}QZzt~9y`@_yGBpDV3k!)6%h`W#*v-EeP5z9u52Oe`mm6S8v@LuwCw#<r
z{yj0Ti6V0-g!U&mfRDJmi*A2U=4)2Ql-;P^;UC0idmF#~cLFKJJ?Gp1p8ho+OaqQA
zIM?ccKO;YU9AJPH7abP+Q`m@0Qp0^lZ)u`Z8&YT(`!rp3!H*;URWS<vdIg5~g?$;J
z_#s{&XKtdSN0U!*!tCD<{_lD4?>_whlo9=lW!Khs9udkXYP11Jm(QI$*O#}I+>P!7
zSnjJ*!e*rqM$7(}k*R~{NovJ#HxE3u4%6_2KHqIvvcw!#r~9(N+6aED6{9~;@?dH;
zsy4W|0w0r+mQIY1r;d9}{(>6#04Po!fR&a}(G)<%;KsUF%N_&<jAOVuJKEX^MyFT^
zUP{N0M-EzuS-iz0JAjlbl&Fz6&*Vc-L#A^FWO3lMcT%&$aV$D!toC~cDnKr;&*hbl
z6scalS_z2|csYU~QBhH$r)vBHEKy5aTa_{^Gm{u5%Q~K*icg2%(EHv<v=f61sTi=8
zBQ9Xupa4Gaxc9^DbRw=!7cbs(UIgh+xnsu{D{LR?hgw;PUd;}e(ZCxE7}50HA5|H4
z|Jt{wqEhL?-@c9i8I5kI^CsW%mp17}2I@Fho+KaGzu)vO6L6R`gB$4PDNBK+kFj6w
z`!i^&ODHu!T1PS?LfF(Q1Pgf!HD{i|`?k#J_r!$Gcy7+>RaiD|p%-#W57?nOh!pCg
zI^59Pk-)<uMyHkPZar){u3xYFMi@bpp#;a&L=EUu(d$ntq~+m97bft6RtFAZb#469
z$%hdwFf#Z7M^ug06sPxa%eDn~hI{VNP5l%xjuRe`#X{)7U-{ZpfPk~@)v0g#IDp=s
zB|^>#+@AC9EBSrz!3l$2^4y7r?%|X91ObjU17h~L+Q#?jf3idUg)=5;)bpY_@<s44
ziKA+7pWzMBMH`7~0h3^^VD;Sl4P^_dX9^l-gc8WSu9L}NA)sDUw>UC1IM{*%6=l4-
zx;jBznVF$U5q+TWe!(a-bk}y3dg!wF9vam+7y*_SM40uRJ79k^Fcw!w=O<(=F#Cc&
z;iZxXE*70?edFU6Ffb1aA}$V#fPR#X2hwVYN2{x=PfJ^n;M%oe!v;h_z_TDHFp2>4
z=%bZP@N*3At#@n%RRkj|^qRpU8;$6Um=@T>EsydYbLTLA)gLbv_Lt9}1GNs0^+(Oy
zhVKR0PkU`GP9-Ah6LY3aOcHcl_*a=e1i%GsVQC5WAfV$h1Wx0|4QS*1{rqTXXfR`S
z9pj|HJ|QG)XbUQ{{$pYRCMINM*}+`_leBPoC<f5zY!<a{f(VOtevFJZ;M2fBa-gC}
z8g9=PBbF}_hY7XXwQEzTozXp9h}I*L47%zz<<XtB?uEt^#Ut{3`mI~Hju)L|XFm@J
z5k~nC%;45Et2^Nw!K_+!vIm|8nHC#r^D-r5%HrbU#vPNOh6r^*hgljiuhsf7xLm@a
zpfi^X<9L^q>wc%?{R|fMa7ReNspMCUt%zwEM2~5394#<&9W<yA_QKb_Dbe@`!3zwB
zh}lALnjvvu3@```#`E@sXEYO2R0kytUAwZ36v)#A4N;x~Mi!#d5pdJf4Y*$Xrqmy6
z)as<IgpnxdiIcCT(Dne)3@=D%e-mz4fuhH54K_e04-^&@z%**LpK%?7DO9<6*w_|v
zhftCO>dF>k=s)sn3;JO+;dSPG*VUCkcKl_)-Zdnuay)dTIdo`oxSr|DX9ylruV+M(
zyv&vsI1qQRxd{#g)cS+hHh6-F2x{bYh%?KiG4q3|%taB3xtV7G{?^r9>`fYQ@6p$a
z$%ci{Z?H2quw&B@7T0&iF_FB9IX)C7wQ+HA((n7Ikv)YG?qi2eV^(KUW-W+eh`iq0
z`fZ??T505)CxI;?72=RSNZ6`vhGQUD^;o%RrG#D<$Z@E(JkY(VS(_4!$)v#OHHveg
z$DqIVPKz5xYmql+oaZOp(C5v-!N$(M)cbJvrcs!kKI~=TDV}`;WEU$qr}Z#qdka^w
zVPtZ~<0D6ouy<nVZcj}f5l{r1mkkGB?K{91q3|GX{@eDE1<k7`-k?>tLZ6RJ*n%jf
zo<MVpTPW7KmpC5;RNueLa*d<4IOn+aA+VCb6$AVM_7To?TngXvCG#(%<cwvFj|3v|
zpr{RzvMLyfH(x{oWm@rEu&o90@BjSu4GdOSCq2|X_M=JoRQ@e^W<GPyX4HOwPJ6<y
zMM){Le**2*_CX>N)fb3N1-yevj|@Q3o18e&Rt21<ZgLHY45CWC$Yk{QRVIPpcDnBs
zIDVJ~5td|3EeW5#e>cZ%#L?+C;ID4qg>ilI>7y_ABRCU|>ll8Vlhy+098YsZHB0<B
z8#i}qa<YoaFNjvjDUv=ZDdR+12fF}^R@Tj+LC{}?vjdWfuJllx5Py9hUL3Rk&Vq(*
z4+zZ4N_;61uy(_$T<Qc^kVQyPfcRu56HG5SZqo}TBQlSey7&w7^I`O61oeqSx6gVr
z<cXNC0EnLLiE{@wC;#!|dI<RJ4D1d~IKtp?1Y}@5sRkZ45F2f2+WkL{^?T`<t>Fg$
zj0b{AhWh%1aPn(+XiUriC`SaCmp4*AfnLNm*%bb24`JvFN~kb-@hiQRG3j>vRYV`C
zG*l=NsUdJMD<&x#85$DZ-q_pp=@vLfa00ahFLo0gL#r8dL_6D}9z;d;!+A;1z(938
z7>po2<Y9P3px%<S+dDhq2-qxO@;0JF<dd9+lvD-K?$JWpy#SUGFH8%C)r#=ulWiGd
zJ>2H6UcJJcNx-*-$%koZngC90?AU@#w|oILl@$7M{3UNe`aCl?cLbe`u|CZF{3@s1
zefMMA;qetf(+PF+rAwD2Bpw$R6EXp~NpHSzO90X++$Lgi&E7TUN5eGk?4e>!2UJak
zX+mINQ9BK*dR)6!hsm3MR(o=?vk6ab^2@oVB-)UCF-i1!f_MS*9-+nE>XwA+;q5!K
z6$6#P-Y6{X^ApuoRU~||!d%(oLD);s6+w=Yoo0AmGQZ-t?IIo-<iKGNKwZ0bO;ghp
zD`#Y+9f2Od8=nS`&i!<BbZB*L-J_?4Y6BlNPbLfGdqOsdx!JDpq${+FTt3GBS_(t6
z0jHYG?m6}!1+N@YHAof0vJ2uiV2(iXckF1)yo_{ex9AGt)`&p5<JChLfQsOJC3-E;
z@@h~I&=@XsY8Ja9VfJ}od^H`7{4iBgR#HNN44?kh)m5U*98Wjyz!(i(tOo9rCx<aQ
z8TrYkhjAZ2egp!~%FH~XGmgWWV5RVJ*9xpr*WiF|<O__y^YiCVO-nztoMNVmKDsa-
zJ`_Wl0nNpU{0<yDSfw$oZ)cyqljUG%$47&sfU@o#<Vje6>N<BN(q2iybsqvlx0MAO
zWC5V5mB&+_f>#6u&KCrf{1&Ny6!Zf^^sB??rk^mnQ+IVWy_(WgQc;OkIv!X}IDldw
zBdIkBzAbq4=uI56&rGh<ARr*Ll0nclOW$zdMef2izXM<>P_Ln(A&f!TO~)^#Ecg}h
zEGGD>T*%2jU!Zk_fx!**Mv&~8n;$9U!lO#jlkqo+{aXbRGxOGwo`HtXNw!;<`UNK0
zIQ+d5i!ZiVYgLsgv2_V+QRmrVg3wER0$xr=N-BP79a(OZ<XTm2EqwZ)FLC>A{iIuH
zpWw`?#0>jNWMeop{VMgV24Fh^Ma3r=q~{3jkkjXD$1#P$B=i!M^j04TgOjpOb{sp2
z>PJ{#OF<yG3V57-x3yaHqSUDyF(rh6iOm&R;{f)o-|}T}x$l4t?jIxXJAByu<Kfqc
z%E36w&@1J?10LS0KlgBfZWl$PY=Vj=*&1638p_tD8F?08xo{C-KE6!SFOaQB*&%>Q
zE-{25*XEhCICgW0FAGPaj@Ct-Lg;jg)G2{j2KGGz1XYw{)w2f1vV#LHJ@jy%1$Ep!
zNyW%}^WJgqBEam{*4EAIXxRE0BXfY9Sh6xLQ?{t;FcOj^Z++a*OC0-Ws`kPp6sY2g
zdbe(6;BQFG__d=50nv*H;5>f*^l3YIv*IF@EoGbEda}cnrat=WzK3xRTPPw#>iJqX
z&LnN12s<4-sNP(U(1S_V`pI97gSV(Owi=clIs*-DQnw=fID20m5q6FiIhVkPx$<&i
zawfseirj%XOs>1Ff(X}4W92woO2XDo3`o@~P;w*u!jb|=0;P#SaAGU3LWs=MmW%LU
z;yghrj?6Aq!%c25xG3o5>8Q&ch%`Cz+f{7l`s{DWf)kRvAXbvIXpFg~IqdML_3KJv
zGcf}Grm8^r^<HGwq-F?6)vO1II>Ru(s7SqtiD5{cIL9`r`&JJGZ^><(4XUZApdS0c
zwy>*eU>ZajzQFLp-pc0KoXr#wYYif$?5oHp>zZM6fQhmBXb|>T>1QO(XKuGe297)u
z+ZXvW@?V$%Ck^LECv5cvUR<xq`PzQ0t-i5Q*}5LOCWMP@^Banadk!3kYn{Lt3kc2P
zb=+aS2R%ztAU!YD1M%ZHCi)3NA0!|9fmDsGSO~qB)X1cyG)j4SO)aftlaOSbKH49g
z2m!NOJ+6uHjOtHC;);lak;ADjX3dYEJV|n70mt*~X1t=BS}weH{69gS0d$9aL(9R{
zjI^>%5L(tYd=xTEjHwY{RaHU;Ff{u2g+c<PHi87Z<I{a1fd#w57w|!wtAF6+#DtGB
z3Dh{)5&=v1>%qc!c&wwhlEHD6=zBsNZ1D<0B~)7~Jc46Zfa%D){+4QopyPV}yc)G}
zPM1Ins=9yVu9mem_xPsLN_tt|GdS+?eOJM*zsQeb$HJ#|)j+2JXMRa6y-5xXB?&_o
z+#0_fIE`>%N)2>Q*rOLqu#7M*GCkB!3>bL8lN{TzX%z-oOFxsa0|K_lof?Jc7&MkD
z<$&!XeiJ&$7^T$JlvANhUAzQC*DDtY6<Cu<UNX)}WSg*2Xop`eSEG$_hgzu`op&tL
z4Bk5UOl5U7F|S~=)yBz|ly?(%Lt@>Ze_eftTD6$Ko9_LFfCPx1x)3qc!QOrhEudK7
zT%4O~{gHX<esHoeHjZC94lM*wgkp@A$5t4Tx~rcZtUMwfYd!e=5YvRjrKzxMUgD8#
zoqJvHI_!a{1X^oIrcWM#218TXh*6HeSOoi|HPb(q_p<c~J~>l&v3R7kHvOK?cir7f
zdY53z7<aRvH)Ln<PdeobpQxq?QH|_&CBnG**s)`pXF^b*z)!5~z9NH&%|mayG9mf2
zey!}%?ZD`GLZ&b=hY+D!%O>G-SKz((Ksk*VaR4(Mlz(d6*;!eTPx$rTyjs|3rdMNX
zvNRYH9Bf)%eA1XSrK+Z8*u{Y25VZy{QL%z_tXOuYMs=AfLsc(vB-T*|9CBHQj1%R2
z^kR2C6n0xunJO`nOUv!GHs9;EmwKPjCvpxtw#&Svq1oSL#Cxp`H3PCc)6CKHu6+&;
zAqQw^;vFlK^=AKF@RCE;R*K%o)8GmLl2Xds4EOd%{Io;S9POzn;TD(h<M~9^W0ED3
zThePaeD7ZI)TKAo1DXMJ%-8Gvl!xUTk9r2{SjbJjnh@?LE$=BoqoTLSl2kEa=Si%L
z%wLe2nI8Na6x$(Eo6|L`P45BrFJ5w(5fA;0Qo<~jJhVQsON6-&G4AD0U1jA}Yw<*O
z!h;@yHk2fKEAANXO^9-asnp(!w0Y3V`3g1p#HP*X{OP{I;+t|k<FWxo4(|8|c)As;
zM<#6bK~Wx%!cVm+HsQ>r2aaAO375B#sh*P$Opg32+(AX94LPy%BH|nxoYBN!diOX4
zXiy_*NlQPQreDHwwzRNtuuCN>DXCwp`j|9<VU-bh?zNZD&z9NxOPL`jfXL^;J8$^O
zv9Mf;>1t_Fdz5{d-Jdc5;^@*`N3I8uEn&Tspa7_POh|Bwx5%Y=_kQ<o=5F?L7h#m0
z+19Fv92G@vQ+4S=-;Od{0|X*MN6nL#oE#$TatGwPGl?IgIee@7vkwnxdcJWz<qd?V
zylfRbn~IpSI-kE{o(?LYvSy+=;E1Z!i8n}>Mw@rKT@X{P=YQ*rG#a7u_Ql7shR90U
znVnw-V@!%c*))oKB+O<VB-GsvNx4H}7L31gSOM*%Z|wpY!Pa(LUIHw}rs`7MRlxMy
z-FcBR#k$PwTs9t+XndV8A_X}L(q}ZC)g_$)j(dTCf3j^H9b23HWn+3mVo%iy)Z9Fa
zW4%~&daV64`IG5ZHK^;Ik)riRBGSRo)ubJRUp^q|FU(7q4*{bND&n-$3@YE#FzzWP
zY-$_t>7g8?+_{rROJ#;lSx$qv&=V(5N=>D#Ai!_>_C`5wa^ECMd*o9STr?}jX6<SJ
z@iPPgVPEX*S!2mJ#`Z;-M;gF<PHQz?T<QVhQu#5}XyBhD?PUHeiq@*#0#vCz5V?>)
zK^V-1!#C|KTC!A%>ws4(pNS@1t0)67SY;X|8qVS~g6hRbQ28${_pBeme5bSfViGif
z0+b2AKpu{=WYYn+*?LLHguSGBQ13}XFeGEW3QY~XWvkN-n|9>5J!<tN&Jk<UZ$!2c
zck@wlIn*G679G)c8=3b~F<KB2jqUVBSJw}*4+*dUYWQp184x%iydX{D$zQ_<i4WIp
zrTA#){Rbzm5A>hx9wvaq5J_j}=O-by2uNUs>c@`}_Jg|CIS@taK(h-9|JfUy+C-qj
zb95F^D%fH`pl`?=@$GP=(>%tts<>8S(9JlgD^{li7>eEse1mAO5IDMuW(<H}gcUuu
zuWF1B`S{I3_thoj4rmTHZW~0W7wUjisB)Gf(P|43X{gX$K1i3!%HE@pz!=d66@_cp
zd~Q+2^O!Dg6D$OV{N&N2_&}MMh7TWv%0lEdfG^=%P6UT9AXGs9ZVlQ*!V4b0sEs#o
zeb;!oQSt&me>U(PXwiTxKl=oV=Igg_qpBYM;FPyt`7t*)hvugOtUC-1LYRN|t`ldH
z0F}GCre-0bW}gFgsX~xNM+towf*rZz#QF0}kZx8n6Mi|>11OW&%>WEf<7cQDvB9nC
zA}$fA8<}Kqk%=fa(7Hvo1!n_0w|E-Zyq}!T2cUrOgBgqx$~BNcp?VRFxEkA!<DT%N
z>owolEPeH=2HN{$`D;+R0$giWn?_F;z^RIR#jQ+O9Q5ipZ}y@gZ1VvEGz*-9B0o#@
zZWmEg;5%TECM095HW~^N)>no^2_%J`9g6O5{Nd(@+T%F**sl`>8Q>GxPq82PS>*U5
zU5F6y5V@fJJTc)a=f^E21E@C?!h}xUoWzm5naDewlj7qSkW{Ns`o4LCB%1Z0IUGL{
zykm%t1G&=@+s?yr6sb;ABenL*sP6hSncQr@vBM6ZycOaRWe;V0n!|^?@TurEIG9V!
zH2e0Fe!i=xT)k5=-v*l7il;e%<<=norqv>R<FRnq0)dPD9vnOa0r<D7H0>f-%ByrX
z>H^*aVudwZ4|Xq|&K=&#Wow?Gh$vnV2Z=it9J0c*WqWGCANU0aC(1CRpj&!%2TA8@
z@a#uTh^BBC4fuk}1m5OjDxzG!tEvolAilCj<45Zbw4EE^fr8!e94~ZFXlCm2Lpe5l
zR^y2lcR0#*<Zt$AS}eX6UHSEOu_MP=S&i`P#5B2^gn<f-8^{H_L;s_oruuv9V5T>Y
zu|Ba{oB{&mmx4d*_&fYQ;qh<Xc`#yxMSu%9s6VKHNi|i}R;=N!uqn6)MIk94gCi<2
zTWM!!=mH>VGJ~Y927W?_`%xA9OFMM^25f`GOBE{)nefh^R;x<5o0T9?Foo+o+6%MO
z)RdI?upnUS>o$rWof(*qP+5c?!LI`%h7+$YjxBr=&=>4zK?r5>Cz=-Y6cju(L{Ll&
z^!3g9XQB!-c|yCb*+TTM;v7M_{4P8^t%RSA%>?H=<j0;8v?xDtVi6NIcTT({qs2Q8
zSqsE-9RR@$4P#@Q39KFJULqmFGT}3qbcke0ll2EA39X7NB!{T6t7dMWL|Fl>YZnz2
z#shpo>52QU<}2KCA;<aAyG-QhP?yglg#^2Ff2m<*Wd$~L5CDF5)<E@vLj##@JA0H;
zizhHyq#*I=HbusUk7#NdNP*lA@5tynbqUq|PN{<!v2=%6IZpRipyL9FT9^!-q8b(-
zP`)>BY*B#%cI=*i6CPfQW+LPvgjpd9e4y*#K`?Fu(Mz>NS(e3j^VY3NR@@5~y6@A|
zt<gwF1bYlh#0%IFB!LA5q}}GvDi3c$uuCKjUI5#IZ*ms4bmz{Yl*GVTydmNi(E{)B
z3egt`whB)x%@f-F`zb{<<d_<tczSATXaqRYVwm(lUw{Of!GQr2D^+**HLMN(8q0+K
zch?cjP}!kP4I~x#Dt@yTaL(<p!vuB4$|}k3{}pZ_1_mQas<dvp9G6hSVp^RlThlBz
z@Q58H9Vn{zX^FyozP()|=2~@<HofEm-mZD;yM3+?fO!=jmO<wpwU(Tts_(gzzLolL
zpeA}=fV*Q7X)A^gV3%Te0tRjSQ#b;1AkF8Qmfq0f4uq@{ujQqsI}=^)aCC>Yk=wJq
z{&6N#_00a#*hN4Jq)|ixjRDer?!pD2@KdOAupOaq1q$Qr@U(H2A^DAu&(Vb~!X8(-
zDJVYu4n_izoO%PwBI4C_8ezZXO@U1**lOZ+^#6**h}-ag^>M^qprGi>;B~5oU>T&*
zm#1=|7IQ3s+UUCedF)((gaE7UFMrEx16hKQP2VHu0t@;Zfv>s2cJp2`P^gL)l_qzT
zPYKmxY60TF7Axco<v>+fu3O%^Rc69~q6U6V)SQBOr}AvG_7<GFXY$6sYPS2)vMN4|
zc>SL!PrTmbu<_ataG*VCt0k=?xynuNrQYi=%YsZlM2lP957-jMY3A;Fx9MULs4j2~
z1AC!z1ytRn>O%1c)^|*(1CGDuhvoCpqY<g^W<FKc)D*j~7TR7#ngLto(L`a2#4leq
zP8~lgd*2;R)XV75QC}7Pf7pA=uq@ZEYt;3qfC_>lf*@d$qS6wAh(&joK}btThhl+(
zh%^YI(y4TcNVg)5gp|@HAic-sdfxZ{{<pv5+xL$Zz+KmMp7We@j4{U)6zm}8N%kzf
z?LvZ#vo1l`cHq-L^XF7kii>qQg27m*s;eX5-~mcCQuR^JC6++ww*SpNk;)~<KmASu
zHW^J5NOU=*;eH3y51oQXY-MD_Wm$m3zm$$cwT{{z$@|FqyeJx%v@l%u(y_754?~EG
z<Ap+JUI9eeoGqt(`h=jG`qls8f*2nJ6CMEA@eLx;_YXO4VDkHrpwLi|AkuLyFJJBl
zMvQT>a|&I$qN1Vti%xg9VmCM*xu(yo3=gy3ZDD#)(!a6czvh;zd~WO4IvO10Vn45b
zj$Oz{DuIMmeS`2q2YHa%v$X9lr?|K$_@26xIELskm1m%?j+vP#H$gT;I4CJA+oI7Z
z_+lY%Yce<{uygKt^Ru%~$QM%d(qdu=JR3>05G;J1oeSRLN}#*s&TvftU_96_B~7pK
zyFAJ{!O3Zka}s|`j6t-q`7nPBLqyOzQw-D|Xs_TANzk(ZW2tpp6BL-``Z>KGJb*2B
zf>T&Pz&>;iAV+s!-dwM3YTATr0q|b<Uh9C6E3Rp5w5)k?6txc>zgdu}$BpgdkI;BV
z%qpro38oX19}nmXp$=cMF1m=b2bc7%DJ|?zkSK>rzd^YmbEYKL)AdcDc8QY{usD<z
zsK++f7ZgTpfhqx_#8{f~D964JnVBu_rB^cEhhV5Xe&3951AgMAEQ?A|WO(?>>MDd(
z1o?m{z#lV;{vA{dNufPGJs`&}e^Wp6r&7}K_4Jk0DVOfWmZ!?c)ye1{N&U0|Q3Mne
zAUvpSG|VgJO(jtyk3bWd+7_-^5e6#v=91pM4XHk=)B-~k>$r(toSdF+L4?6|fn9Y~
z)s;Ibt*x!d=u!@}M7>c@L*|8#w`#?YQWpu|)WjqT7??KUi;q7;ZV7}Gezv8^H?f9+
zxc_?TNLTW?8yx36n&XfwOe64lamL2QAss^lhdYQ`bj*eWwa{OLi-pWcWZFf1a@5~D
zXcq|<IYS@m+S8XH)$wwmB6TxL#1%l~pkU3$+pMp<mC;~fLt2A`{W;(uyonV^ikrDp
zP;sbjd16JK7ZttKlk?%u+_pzhJ!(3K0rK8c>9=>h5BAo{BG1$?HERjlmG*2k>?P@+
z?%;pn*g?SHTT2&NbL{6pYk|WeO3fDK`@?TzV~5fGK(I$hVi*n{6fmrLVYYnx_il43
z3`sz(4s*5OTjzEb=z1XNBlO^X1hEm&b4pZ7cQc9;6Mx~^rEtGneG9*h+p7Tz^Pdew
z*U#zyv8_M5{TRuKk&}l$T>Qiq)?C)gDDF|bwG|o>q219a5mnpJK#VE3CHh%ld-gkx
z6%vg&CehG?@ZAO<RW>QBg+vXU5#0TU>*}NfI8!N5c<dr4*ZpwwQT$nXn8|BjyXLLG
zIMbDNRfSF?4MZ=S_$f}#H7q}hlyRXK0Rdk@pn$^Tcw&P#MD)6x2dV;3Mca-_TwmBd
z-UNN)We{j42}8lb2A#X1MkVEy+mq4bF$vqAx|7fpg2-&oiw2yEg+!mBcNS#TW#sA<
z7{Afb6hv5|0Mpu9?I^vJ{yqd_eAd?1uq5<Cvl<Y{Fqb4*V*{2?ROCbU-_=)s?D3e}
zChA!;bZ_+sG!od>s;9%xyXM|lFh!>>%b7FH*QUJKwebb@^LPoe871mXbxj5o&>aY-
z5PxVVi=&P})=pwAhA$I5^9deAQc`CI6>^Yg0+Oso+MUdVMkvHqLMVYajr)b666-ie
zfJy@lh;h^T0=Ns4pr(Xto$s%50=|kfM38>0mIEG$UXV0C<?UNyfK1)HU{lu6dJGN9
zf{*@Jj#R>bU;-2)UdTLY6W5__Sjq-psUKg-B_=WP+DKkL^g9@FSoyvif7Fqxb<5YG
z>;NbP8(0R6kDz@E6%ba_!D729++B!xUZ10Be+;<Kk5NSZ8%;8(YR-(}7JOZx_VfZ5
z{bX4pF_k@Dg!AqVE35SR@U+B4r+-9JLV|-46s7?{sZnBYU;soehSR})z;0nnKd`gA
z`{QD}4qg^P)S6~ljA+RQpf#9xC@}%FSq2Xabo`dCzkl7(v_w@e1RpH!?wu;XQR;2t
zC>$RF0D`7Nd}x(T>a<xn4RLHncQ`!C`GH^owN-1YdozYCLF@@r1(1oYd}6*1L<GsR
zop|)@$xm4ct+Kdb8ICQc(Wj0L<*i}AL69d;Vx0BJlOpW9UT`P_yHRNo+;&9kz3y*G
zDy%rXrY3IF`)i#J#jcRtc>D}e3p3?X#u=L%8>7TMHwo^)JseL1WukiJ0x!-FBIp62
z*EoZjPDH^5FZu%=8_)|P=3xD|^BjCx_~pwNpoMMk1Hh?LP*A9yURee|iaBa1cG}*L
zx)85{D3!KainI<ozijf;Rv5Gfy^hGabHxiyB5K>O!z+$Z!#cZ}Z;OEcq*PqRfWdq+
z!Rq0RdAyT4(UQNw>6)~(@{BZlz_o*?G^zIQSLToFBQOkz^mvX++StrtMfx+TMQ!`~
z3pX993ySaL@9xI}u7P?LTc0h*dlOPZrLe$2JNQ$eT7mPvZuKgt6f`73-DK9;M7uQj
zLpgK&fZgu2SLhB)6fT$bF!i`;{5D=r){y6kYu7T%bd+XC+Pz)Q(F^%#NX6UN8}c|h
zE=LUSwJksv!@|m{qpgjBSg-}4`w(y!0x}Vd6pe}bgi;lQwb6IOqSiLM698X`blk|u
z2yzu*QT*Xb9hx)RIRF24+7Z2LPo7-WD_KKFY>9NOj&+Gz;BF{0y}Z1Vqo_V@!xM2>
zp&*01HFOhrZS(|riQlJ>M|BB(qPTc}TU!oT?8kESFs(xAYiDOSH8}}Nc5r&yN;yjq
z^SBC7G(Wa<MI=Cml6(N2ag3ObiZNm1HxQl>qrrx%k!-K4sp-Vs2Ha<!^SOH#lL!|3
zpQ3EUAD{(2RT(oV7Qj*w`<KYypXa@R4<?6aMxr!bTgE_28;CC8Rh*^K&>o}dG~(Rc
z<pLgj7iEtGOwzCqAF~E4junMr3VxT41=Jc6V-;VMbniIdPfLk{6EGL*afDeMhUw|)
zkK2w1)&h-z)X37(5{?P<f>x>46tI*;i)BMoQ!eWMzg?CvzT->fR?CTlcUW8;t`&}L
zIyc6b0RcyYrJ>M(6bFOsD^|2FUbeFOyW4l0JI&pJC=i3Me<1Zpk~-;yg#;-f&{%v5
zTogS$JqoeJ9c}A_Z;%a*&&T7&BCCZiKK^WYerANOs0DymqsQhrBGI0UFv7vKy1I%>
zlYv75scBsbqC5dn8J*U9r=iI}O!E-qxQVVd;Jjb4QG;z!c>@Us0kVluN_$%OOWY=L
zapCFp^a(CO+uqNdzaEu1eR-*q=6tKMXVW_6fWAVhnB^AV14k1UZo3b5HhGSrGQ*U1
zOb0`0&9%Q(HdS^tb@kQhvxK5scQka=<e(9KvB85b?QRz#r!JkYpFzX1!wIT&iws}N
zWp9pJ%}n1O>im-Y10BDYjV^apb}7786W`37PNmm$?sPrZBWyxr5=R>!s<B(P&+-7Q
zo6;%t7%dN{(QA$b4;#^#(1>y-dhmWzoLR_6FD?Wv8XKipYlzUek(jaqq#|nmQ#gDB
zTwvaVub61`lzTB5H?{opXT;j*fPgoeTt|)!lq~l}Vzy_9H=v_8<JB*q9DyU^ZBqit
z8f->7trwiJjtoKrNN5<d+_3;mhG55p);JO<qyV^c6;atE!$1%{3@IY=fP(Kpc5sR!
zSYfT_Kd0p+DIHLQT9^`O7UuLIO#U1mW~;-XSt=^3vv6(#!e#KvQLYPgD^#qw>Co~7
z)kCt`=Q5E-D2ZmH%C19yq*SU)CuD0#kpf0aH~V$~T>dF0lCao500qL^>3ww5it8zb
z6b78Z6O;kabxrk44Gs+j``_mPrkPq;Hj7ghK~D(v?#*a`L<vZVNp|jpM{RQVL9|sO
zANdKP5d;<--&!$@tr*RP$m*gX&z0I~M*?kJk;@Mp?lfImG@$5!TyBalP4k|9nqoOH
z`z7W2<;(kAC2t)#b)>WC!#zD}{@g^Lh>F{MRZK-JOBIP<J6&5fZ)r^Kq@Z`*!}jz)
z|9NZQN4ZwK(7?;9bFX*joae^+kt|*l%A)$|l4YTvVn}&jL=jMt)@s8qiipp!DGf5J
zWRVSvk*$)VVOJ(3y5D-7Kz#xog-KE6acpcRK`Y&Z^Kg)Xq(CX8Ul;SBqA8OuBQK_9
zu7ygsJWn<X05!Cke9Wm>8OYIMv$J1aadP^DEQ4>W!|m{Gb1vNC@9F6<SZVnu76%ZF
zC~+M7T(#=|n3n3Pij06e@Q?mZN+^hkiJidrPH<ivhv*7QVC1M6G*JC19(NNb$$OVY
zGaEq#B)rzvgdGBKMZK93q=Q`mS&0^WaC2iF9#DZp9sr7NCke@TSXG?!+nu$N?f_j0
z4-XIu7L8R^@qjo9ESP{a*W-wh7VwM*3MRe$1)LP{_zk!(LVy>VlboTIg;fz7dtxjA
z#*l|z{?lQJ3I^3fWA|szID8}ME+M);28Elzz9RU?AYVXcvzMH_m9aRS&&m7wbA32i
z0nEqZ4+RW2aRcuvw(<L$qkEN=i_0n@^r^<z=MqtVd&viI{ArzBS%-DtvVrmgJT4Dd
z-oCoPLG~G!IjHs=J3OQCDDm3@o{RNF<7t2AjwUMSkax?=qE2U@3ZI;)5j(J7g*xN1
zF7=K#iFRN6_+3fW_So93D!klKYIqmV87Zxht{ZFnF4J$^wLL>QR{9^(m5vLIt;Led
zd#4HuI9L7Fr>@m?wRfc?t}jxr2rK6r=GX?*r7z8orE5g%e~rzK2r{3SYE<}-G^r-p
z){5WY%faX0MKT8(yxw=7l@Q^SN$EamF39mMQNC7~X1B(j>;Q!izTW=%o9nC5I<9+r
zMS_ANV!OSMj=Qu@siY`m**2!0+FGa_m#d3BzU+TWq;zuhS)`Vgdb~whX0O;U^*qDg
z<jX1ATI<b%`9)`DqFvm~ENi+w4K~bEi+F>{r|fBc!UV0OpPTz#EOa!4+F@L%W&aS4
zDb!`@if_Df#R!Q966x+~`F11_3HH9mfK*{+sH>v_Rr{A$>hG>l^p?J}FAJJ<<h!Ga
z+AF7U=AtPfRN9^d_5&&krI~j@QKX{CyHTHEDRuM)?msD=xO6E;yvw<r4*?OXp=1$1
z&6o@Kk+9r0|AF;Wz4^MYh_?c1y&+?wc206%3VD}H4l?G<6pRH*5tUo6*UZYiP<ZIA
z0ABJnzqvYUklc7!7$p^R;KW6vKy?OC6O4S;@Aa?3CjsWMxbz;|3CDBGS|@#|ZT|y=
ztjg-@e&~2`@B<TeMv;z*jYs8>{v{cUX6P2Y6<vW5MzVoA5&^l67<L$AaV!iWBtm=X
zr_?<I#wB|MStWp-MEQT7`}=qFbYq7ZYGA-Bz3rQQ==E0U!pIcZHQi`ggXuwVdYmvS
zkNDHX)VE4=Vw$w*5BQVrSAeC#*RQS5Xtx-v^$}Zt$0LDH7CXm-;)(W*vx3)?ii%5;
zAJ^{w`XaGNdw^N9F$20)c`bPZVt_2tp>CJ`2M;bG`JRxoSXo<}a9RjhG=U?SF#Tb^
z+lr8TO;@uD2;>4eD5SN;Naq!tyxrPoU1q;P2&+X3Z;<3=4oMbUVtiv}it0~#e;E2#
zGoJTY*X)v1gS$;nmWluKc-aG@Cr+O9V>;jc;^ckEik-oY^i{X#`_hT(LQu3KQ&3o#
zOkz<d6$*1tXXo&22^A_%)@zW(^|ASty07Iz(9;qEW+$2c7J4TbTCQ7>F~JHdb+Ysb
znLP8k;3$>$hAn}u60NRsYD$G)`6t-bHh<m`ANS}IGFr2mVhE*UEgG|FxHb2A*AtfS
zE;jX_uCjBVv}yOzq8U~H((WfeA<Ul9edq3j$!t8IO(CTy+bU<JZK=O6W2&z|zIZCH
z<n{Ol>;Z3$6~qN(Q25opt`uj;zZLVm+HR?~t1jj7mQeBM;@uM$+fB^{__^+Gj;(~7
zDJ~AX$V@O6+nqfxIT0#7H!YysJHX#xZ;>Ipd`9)7HHPh6yvn-J>aiCcOVBd56nBA#
z&A?%fAjqXWTxwnqHvp)0DZ1)b!C(No`05)~n4jO8s`ag>M<+sv7*z#iE4bXIpda!K
zLcj=GqtPE65iUM7+k~kLX7HSm-@BWFf|xoy9iO>!52v||;3|UIm!L!GYAGRnhu!9f
zr2w8(ME$X^*~|)g`LYFMHnOc}<wNLOP)k;YksBcj%ijSJNjwUhI{sLC)MOUxTMz^$
z=k>U-?@=>!KG6;V!s-wfH|o2<N~>L7N61hlC#obg;S@A{Fq=$#AjrK@rVF1wO>OrE
zbZRet1<1x+<Olu<w4=~j#r^$Bq@{rhVq-mdGmz^p_H`gPgfBvzu)di*{?qYqCYy|B
zBUzFTJv}uzs0?Bj;ik@bsTL*ARUy;*?r!b7M>^mws!ezKgrA~cUkv*?B6*$VS3+`Y
z+=iqI7>fPo`U+AphWU%Z-5983|2PIDkS(3K3dSrNJsX3Z$#4Ejz#4j!ycuo=aj3!h
z`AT@9?&`<_TD7iBhV2{qG4KYd6ZC;p!~yZ8_^LAuYL_?BViMYS{yDyXCq^$d?#{vY
zY{w@A8{Wb<NO&M@ZMs^9l9G~UVA|=c+k`m=GMcswgRKXYVlSS@fwBX(!7y)=1+gBg
zZgcumuwN%UGSpI77!6l}1Yu`J&243+1VxKF56;Y>3x#FwFb?HLCLSyRY6Zgg5gtN_
zM`{e*R<kd_=C1Q{tIwXrT9S2t&WS<C1DLkNcgN8i*MMlv6mF}b>LO=C769%NZonI(
ze(n6`Z?POetBO?jhh`oYO|0q!FRs{l24I7V^DLHT3uzrv9k21Kj6v<m?CfmRML1lS
z0BJ>-iH4iGp);<9MgWn%zG-vv^BSXQn5v$CVJXRGM6<k@*a`PEkDl2c=|>b%S{5dj
zf`w(jw9<y{*nK~Lt+J5|<xk_tZE^4_$=PUGlBuUv)|${OyfAk3?i03W)21q|1-~-x
zmpUZ%om4rWO?7HCC%j{#T)$iXeP%ZO55d8rE32UizZ5&(O-GBlYOCHpDo@#J<9Oce
zcW28jnGbH2nRfmx!3py=Kqh^>z1c9K3#c&CS;7wqM-p&Jpq@IX=I#Ii!nPc<G{YD*
zj4A<0-2+4bsT;uyqL_!pl?7lZUlB+I2$!MA>5g4wWVg|pXKy4ypxGRTW+k<#C@ZGM
zAty%;_Xmo2;s)YEAvr}M;OFCmgHz%9^_Zh6OTP2>LQiSg;s8Y8gATk1X%uR8<WK8(
z9{MLpNl46^q^@0qwN+l2WGl;Bb2sv202Qe#o&>n??+kK6pN5kP3KyiuNGCgFX(7yJ
zMyG(;hSQKVwCG?sCS%}!(5}4(zbeD46&;cX0juSiwMpm3YGe?d#7O=M3Kb*4PmHI1
z3kVA%?7x!x`0fl`klY-;3=D>7Zizq5U5LWB=2$5PyaMY+^Y<G#siHazOpgju_Pa-1
zD@!p87~t6p-uz7>C&8ca9hVS@E~<>y-rf@YH6Xlpav?O3p^xx3mM}Z}s<|e#P4>_+
z4EjS7fa0c2Bu+yR+h*k0+~mxGIr$fx5nu<xxpo<(Y4h7$QA19Q<?7ohsDwlm+fozB
z^xQSvu;ACl-3K>D+6I20FBF=wmi1EHI|66LAHqZsu7+py3J;(j+qZ8HdJidPK}+cH
zF{pD0a5i9L$X|cp4iTzvw`r(lHGI$GL}X3>k2)kgX{alJR0(o$aPaZ*p%)IiJ&?k<
z0VW9Ud_Re(B%G!Uthf~-mz@%`OG_x;pL=_sgP;$yI|21$8C34*hg!w3=+!gy!U5K}
zC@IM=PJ-gDeE}0Trf@4E{6Shy{dx|~Fo?8u$wuLyac%^Uk>A=$3HMq;kay6Pg$yf<
zZ1-;E=kI<2m*qyEhb-6i2KiYkTG|dk53=q~sJL;t<#imIMSanh>Ee<ct^4-Ut`WzF
zKnP6Vb`d)|ci;yW6`=K<yALFyE9~`a%q7oDNFcHVbj-w74%j1vd;(@($qZNBTzIxy
zEG@b3Ws<G#YGT-5A^bmD0Ozu+dn{5j^=(xnFJyd*=Jzd)aFU-8;_p?P?@)GVcjy;v
z7mf+m`WB&(qM85p$4}4k4b!5!so&A<{Ry5u(j;706_a&i=jT6-JUYy)6quK+(-thj
z$!PKWz}Gr<XAWlC{rk}o>5h+uYRQJY3^gh022d>2{S!HtRh5C)!9ZFZ2PQfOE3Bh&
ztl?V`!Zf|Y?9m!Rb-eJz&j>mkKn%>h#=cI5Ks>5x;MHj56bVjgUji#qOJiWPnKL0U
z$u)-~073=II0%{2igGN(2GD1N9aai44{j5nDkGW82<SN|9iX<l?XeE700~-rk$9I|
z`(9=-QO`}c#x3-f;F#`)`RAB-3FOHTqQcrUZbd#UD2T8I9Gg~N0=+1yMU@8?mzH@N
z66VaVQ8>9F$Tn%4BiGZ2ay>(K5euAXQpmz6ib?R8XbvIid!{)G8Zr&Bth-wj;wo%r
zLsD=Xv~9atypG^Bt7~p<#yH3vc->>9_9B&?KQ=GO7DyWubDX-$pK$cV(Q?M$y>xAn
z#WbRvSP~ZI0)C)Eq3zkg7Yb>UmzS0nMh6EvSYQvD*kCUMJ1bm(tEm<z$v&gBG7mZ?
z0Zq&ON$<=HhS4V)dnZmmkFSGRUD)qf*71KxB2d^O*Qop+c!Uf|h_c5#I4<eM`C@ue
z??7Ml=K=HSdm~AzrTh_7)`?-MU@D=}ND}#$&_#9={?0^0$m2u6j|nNOh-3O_b!)4t
zW#AX|@e*?xiDolcg{#l?R#lC@^A`g)U;L*BS4CT$Rf7el56E%RRRr-4%>F()0ShsL
zD=R`X4Qm~;R_(hthoE1nXmbXycg|zo2{mQxQ8|cc5c5xY94&?6mc3<;4>5kzp#sR<
z3tQ7<Y+{z4nF*xnFGIBWz9a4vom%((j*x*}jR-yn?nJd@2`V~4(SHJ>25um}CC{8i
zfCHPb1lQcHkT_r<*ka^|Cb`9rlThMUEpcbxo+71u%y)Zt-9TA;`x1~DYYfFeJ$_Us
z0;bLDgy|!CJ6d9Wrq3TDbZA&sHp!`v9#x{mSM@AwaIt<s`pw49mip(jxcG!VXXl^V
z6**yfE{}IdmwI?iRHn%&oYwQpU!6(u%$a(7HoH&poylsk%({u}$#FT=YnN)iom-3(
z-;CQhue&(;B`hT1;6Vn5p%p(o?UC3X_s!LUaZ~jWb%WNm&$CSJY@4s=-v5dHW4^Cr
z(`TKPU(cQHx@}gid~SFjcWqO*l~%P;GA&CS%Les|0$w{MFC!>V{j{?4#lT(RB_azo
zDl6yyow${Fu$!1K|BVX70e9@)Wt@O&;#R27km7P}A}W7QT@<+L{jL$9FABm|8)HQD
z%8LTsh713kQH=S7O=Yi<Nw6Ap87b|-p|wp}m^`oA=((HrG+`)f=~_h?2;N+b!)8qy
z{g3Z2_!uiOH8e%5ApH!D6Dx&DdAUi74?$!UzSYWlxB~1JAI2H@1h$JBCwWsiZ}lvV
z4k82c0!cXW;h+FNPz>7vqc^W1IKnGS`2#(<Qpg&gh%Jud@$2s?6t2K$E!w2T=GKt^
z4o)1U6$U9yKzJ#vs<c#NotT}%!fj;z6bGitmcB8xQ-!4m05T;4c0mL7lU!<?L5_cB
zZ_Hg~W@c^+RXNDE<slcOU1u3vD<?TF#6fdLUpFU5_oOu`C7e%Qq|2*~V}N<zzIO;t
zTw0lz?w&G4R2wuO0-UN`^1Y)&V_E(PVi|GNk|&_9-2f|5xQ71oO!$ztaU=r1Js2ug
zRawa%`P9#k=5-ySCs6`+J23r1b*p*L8r-f+!n#Qyj}se!J7d<kr|9i8vc3oAC7jPx
z`CtSVX@J>^H;mzr@$e{TwGjVwBZ&hgOWyVHYEaGq(OrtCD&@9gTY|>L28sqlk2Dxp
zJ>c|}c49nkuF|LP_XZ@S4~as%F~PVdj>a_<1QjI@g3xPvrnlAvly2P-Uh_UkX<J)b
zzIl*uel_t?m}pCT&`!E?YR)wCZEck~L)^~_)=F!N-qo=s1#_<Sk>zIwJ^einJ)LRa
zR^|OP_|gj!M;%E^Ri;O^L=9}X-Hk8ps@-KUy^`72yeOP8b9mPLllS@KEiEGkgQs@v
z=~w=k94hYPI$ytHfZwi{c~R1HS+{yvu30e@-PP^*Izua7YOwAhv?cTSNJhh#;)`}7
z<{N6CFDH#91wNOZ*BYeG7EA1naJU=N-nBn@{<-$2Hy0AjR?Va8-PV6ppVS+XxTfvW
z-9XJ-66%zv^DTX0wJ9gGohipN$kz3{e4B7EdyKLMYxPKyn=U}ryW|@_BHfs%{EN#9
zAqDj!gJqssyUXoOfY16}r*9R6|2gyu>dFnxFW!{-dv9_tf(jg$XE6GyRfMpJb};@e
z0(!{tYc%^k)uYjHak?i@!65}rQ29r$>E=Zm)%}n_KGxg&Ee&7^boMWp3ec?5m%%BG
zegb7CV(JHIwYYr`P>bLW2H!2hVj}m~f)Kb9eK>P6@<?jb>hIKq2w1@RSIETxNj18$
zu9<Y<jKcY<S67k^<!o4v<yclN+BGrv8)dTa5=|FQTjW_rq-qZ|7}>4cB4hpNN+eWj
zTHmrJLGyrvFDbqi0$oCOJ9ib(F%ae&={&=5+cVyYiELfD?#?gy9>iOIk`x?ME6CcR
z6HUqFJAaW8(v#f0eK54MNq7X@@sux(=>aZn)rKw<WfNk~h}OdK8y=wNZ{ED7Rd*dl
z+0F4AV2lc>^#=#La4xX1=7tG7DfT|9x`WapKy5Ff-95<iQl?|h2WzwR_Sa)Jm5+mi
z+lWD%+UJg)JJ&gD<g2xZ3E*o?DorOSkFF|3<h8(_l&6<sEXxn89CxY-f`a5!#|1aI
zC?J7V)rca_H>8*ee{Lb-1-oV%b-6N#hIMPWmTKEt`S872`qMUKM7ybT>d#L*alt=1
zWnm9)$%}~|ie&TDVmZNTvA0)rZuXK<eU8j8EwhlF$CrHnEVREKJG8Z;*k@<nyqYo*
zoS<eCm|^u?#+SkMj>XW3-_KVO)cg0n42xhB`LQzb(@wroX|%~A&BFXZ$_*z<R_5?x
z0mHO&T&D%N9VW{@kj^lvMt8+&k3Ri|(KiVpZp;i4R37J+%IBZe_s6fBbl=5Hyypxe
zRfp-0i!|1K9#@%B;C1ykeu1K_uv#YW#>2jhE~Ik&vqW0jruWd6)rRd1#HefaCw>3i
zn^44H`$kX53hXd2lF`KYr&eIdrM4p8BtJhtWUHfFgI;7`K@wH&lmPVfwKqt=&iCa}
z*VkhLHiMVC-V;?kHTB3$y)br2;=!+Be5Nmvx=mh3nZnnG&aFoG7pCp0;kMP1ZKUb~
ztx4ZWC8-{wjOQ@1@9ysA;H-fvCEQ9P{3SXL@HPMd_oD-lUCRmgBR@YM@HCPl{rnrS
z&dr=5PnL494CLP-p~ywe-aGe+$fu@X6kfRhAN+L+`T@=Iw{Wmvk8o0=C{$T3f-c6O
z`btZ<TK=6IBV~uBHY37+2Vbvw-XPtQN87e&C3L!(loDJ{y>aC+lm)$!olnaoLCrfT
zhuoZvag;gK+Z+K(Fzi+dTs%_QBI!gSWm{WKZoc}$LF#Z6kLPU6YSXQA@HfzU<Abbe
zEdS9_NTJb|c3R}NrR90hF(U*iteST1_V|EA*GDj`<5v`n>VlL$I)N-rT4$7dGH}$K
zPb6^sTv>T}d2@3HihyLPC8VbuGHgW4JoHzC{>aX{S^A6{iRIa9;JDT8>Y6ca7ZPni
zANspr@mo`Oy8ZnYwcccx(Ao3}C5;7z#f#1B<NVZFlfQ?)RWlt9WE5S<kUm-AwRb*=
z!QbF~-~PhOC=*}neYD+Kug0mJ=1|)`cxHcS;(pc8x{Qx~2JE5&x&k?hxBA$F??0`(
z%puZa7A26P`cZLZ|LOVE-{oOX%rB}MJ58392#Tt2svlMq{xn;}%W%G^vQ)?ZOu!3|
zT^%AP-sagy{C<d@5a`0`zLdS{K&Kie+ndGH5aJ(5R$jX=X@RA*KSSSEh3Hv@bO9YK
zh_i9b!!7Sekp!SLY;$mSi01$BGAJi!5fwtEUt)!Vd6#DevmE1HcG8db<gWB6u5O}=
zj^kUWz*tP2j=4{-Nuz;7W4aBZ<~Yln*HFs@AAjhwhs!+CLq(bKXJ|=-D(52HDw3uX
z<AgssjkmJfByfE*MQnmM3Bmpr*LHUo0sx0F#S$r@El(xtVw8+5QC$hIp0M5C^^9Nj
zf!>&y%N6S{2ZGqIQ^?5F17$#nl%@WKbdns!^M=a`+s!+7W|4AZg|adMVziExDM4jw
zs3M&uwW#pNoo+BBgtM``V#k=h&!jB1Kok;pVl1$V0sqO9aR9y{vmXa&*DgSM2TfNS
z-)p@{OC%Km8=zy$+QMQIBQ)bi#w#h45I3=#Onc&_RkCE4^)Of573ogI8o<k77N3fN
zG8mk{@=owQT(R}DK2(7>ZB&n(3E}@IMma=T=c6hi+`0%F>!7Om@<eOWyR^~kxNg`)
z*4K_pBf4U?K!&VQ2)oKC6h6s2?bLqCy@uc&aQI_;9uYF8Oi7(#jM{#4@exi8w)Pmb
ze91%26O_e6drA|$r_<aXRk--#SdH6$8OLNgK+%0`>wP4YEO8N{%9$TBP1WA01=`BF
zN&h+E$e!+~qd=jtmU?REmmB?-E6JxMs8nT2P3!Gw?y!AHmoBO@V$iTlQUCJw%Q=qH
zk~ccxjIMWN3kKDw9<F^|-f~~eIF(^ydSgtjb;2!dT~0qzdpakGbf&oY9TzkGhJbc%
zevay<vX#75z^=ljNP$w(V1E@ZLzb8EuYT_>xZSh;+KIG1k>)9#Y!IuT(R@DuMD^=>
zKHytKB`h3>1PN;aP=7>^#F^tfb=x<QkHT;OsM8Q?dY95SW6?#?Q6_SS4FDBc!jXL1
zAy&4;SXyhljMm19J+P50vFPF%3!6n(Ey>=)w+BXATIk-w8T|I1WwoqCRJPLc^6im(
zLejZt*z4=aypHH}#Wv9G_KL~lcpGHq$MB9?!_a(aa<bU0jiW7b8gjgr7ut*d8r24`
z$78RV9w(M&cFq=rGg1lX53(s<N1z1jU6)A=GNN9vLFtFOE6GX4Xa7kw^jm%3o3gp<
zXVd<Hs*{bM4K0H|*ypNFH&g@kDMMsJX&%A23!XUuM^H6fgkv#`Y?zr}V_p^35yQFn
z2_m*s<d9S#b4R(2RST*|(pnLg5D8NrcP;3+T9#5#-+I4(3|MYASxv>KH4%Ez251lq
zCmtOkgTyN>bu(gba@m9DHtKDWstihX+Osj;e^7a=e$(%!5*^+gD1$Bq4TCu{&@u&E
zIjW%Vgg7P=6(yDgX^_KXkWW+oD;z}G>zA?q!HOGy7c0#xdoUJj5^Lactbg@Nl2idp
zV1&GKQ$q{qo;8|Z4#(#o&&OmcM#LBij}H_+5YJdHc<1L9T%!Lx_p!e18EJFVN>`=j
zpoK}E967TaSA*H#a&cx`XuXO1;g>HI=+Ar7^1*AZ7~%KFPcmjZKau`K<~56A63rio
z29rnuBaWkQ4ITvTs;3Db@4jL{oSy0L{_~t|jq(p_E^t9+*KeOB^RkHYmUW2EId!!v
zTzfQpXI;yQD+O0ERREc!)bbNRoZO-(6ISnxdB1}364$9yK{DF`x5n=5R!g#yj>~%)
zY7t`8o@jO8#c`-mP_@RN`p&sYW1c8~!g+4+QLeF?^-F=j=wYr@cA^5av)?WlhY<Qq
zt-ZRdpX!YDe@}EM<gCY;QF$UFa-YPsj)W<hkK-3*Mhsl~j&qgRqig0=_L0UoW6P^#
zD)r@M<*0kWf<q`OJNMum505QYRz#WxG%YB<(JIc)$sHfGQFHY8@mml`0omkp6Y4=d
zl07o?Tg=F7$%G3feLc-|-`Bru3;`EF-F|<_sN4W2k?2Qv4M;3b^?thK(C$V0&aigD
z4QK8iV^*gxJu&RV&94nsc2;zC387Qe;mbFS=ozZ2d(;5Digv-Kh5ry<8wsRRdew{b
zlk@<d2g6b03L8g$B-X>F5+7wbW2w7|EjO$p3qlSSysgss%@CZYXvq*(xBg}Q2@`+j
zIqQy3EBCyPur25OnUSmI5G?34x=C;0K`NOrHWs5>8#y$uD%Rxv+aoYE`g%pnw$Xib
zbe-tYK{aX(j0^cui}P-fg}5sCi2;C4_+44I>>|S@nC{9-u#aTL9YE_8v<(&LnOWz+
zU+R3Qc|NU83AVPvnhYP<h4!#ixaYLL+0L$0bkG&!42?bc8oLnhWG<Pm7WFe@Xih;&
zO38hiTubcT>T;Z8;5kOledM7jyH$fv>uDcSuD__+c5+zQ>=~QJ2v%nqDRWr5gi4}n
z(#!EV6pNgc>Pg4W+~NGS{_-qJ`*O$b;Jvp9ORp5Fj8#^PamSb6KQ%Wir0OK=UX0VQ
zut?31i-MTB{h&;QHq%}vIP(DbBjOC91V%Z6AQ>264mUCY={X1mpnV!&3GvpiInfhB
zO~{V}m~8IcVFSv70RzyGRj{Afw{PE|)plVoQpqb;+2_q$zP~Lha*a=x9hcKJ4jl7(
za){R$5bxCpZ;V@O!2YqS$qA=mQdLPOIbD3x>C4%(PA^6^`Lo^It6`}3{CT|>>5#zf
z-53g^JURp#FGH9j4Z4Urw^z9XM{%?Qp(1e@LNFhktjEt$gvj)S&sff07v5FtqP}*}
zhJz<=<v^REZbMHP)B-k?|A?3uQ9i5Bd(2xn%opq^)7$@?pFYGik?YX%iD{|fiEAH5
zj>%uHJJ?#N^G;ttUAtP;B2D#5QeX>@5(|$Ct9f8j$<4+GZm$Z_|ImADrsb&ENKDF)
zfW#!h;*Kf^<o-)I*3{(xOI85oxQku}8=rm%*MWZl$g_HzC@di0fIiO%!&mTN0_=#s
zZ<Q=#Iiq!Z(4K7>U!;jub1bQQ{MpC>xc77gpiq+xaURFa4Y(4h+RSAhA%h+rIz|i1
z*jQ~9^}F8{s^zdxx{U>lUL);?x4p&vEVL*0x%4R4rH$CVF!Q16%1<<Vd<?QeQPE;U
zsmkdAQ`6)duQhk{9zYK6@9#fu({|?8V`e_a5oHzz28I&q%M07a%s%3<EOA@mQQ0|)
zWb6z{6x@;OW`+xX1PUVWM$78IdTvj`{%ik!TYMElqITXzb>Xfs;stVe04bRI!L3A!
zi*b@=2<kc@v($8S$oDRfbiYE*clj=_+#yNhYIE0i;99Jwom0Rdg>cPVj@iBd7nwRA
z@!_D7I_eRyj1iUf^qyNAd6|3(9BtyxjEpKJQg;U<KO_oFkB%B&Kaw0E==Fp_P*uHZ
z*OtgTT~!Dqa2OLCWp-{3<3?g9V{)DD4IEY2@5tU3?;gBVchO?`-HgA`+0nXhTV3pl
z$S0~v=n8H7*InE@Wt?YcH_F8iZyY@(AnWqpd@-F@yGb$HBIa$=c;$ti5_jyBy99L$
z4AS-{f7{MnCF<)I<+GT;ZJnijrZTpsIti&o;WYg#kWYhKIP18I)DG)^D-yntQxPnr
zrKzckodpaEfWb8x8G<g&zxNv@6vD^15=!k5(1kMOcaRgN9>3krefKhpNOQ!WA+Si%
z;-2Yg+3#0;in@J(7=fz{s?LhR;zMK)JeZ{Rm?gO`gv;E>=%h{pHB5?rHZits6szW`
zpRLP?k$kk923`ZNlc|t^qDB#TfAJ0s*^&g0QSCS?Cf4_&O&Jg_JoVI>LQ52pt<Q{%
z#F0BA<Lilg=B6+Qy9l=GS$?j+!AB=AecbF9T0BkLSm$CgcM1vyQ<d4{0~ub0OW$1!
z(NGh>2HCjv5i5j!EFiVu)8QR5+BJloS{-#PwT`{#pNQ$e<&9lT(9I&nCiW2IAZQIg
z1N;lFJrR_TP(=_82`mu?rAU(+Jt5tN=UCD$PLk~pp8TuwGL39Zelw(n2|KL~p97-<
z8Q8>cK(}OYi(d?KejMW0`e!M~nN$;~qD34}A3rY1e<+P!G>orDEnfDJk&{ZYDtq}=
zvwnD+qIDVAeBpJBUPW?s>B<!japqv1?R&fie*QFR0SbBdo_dcCXIc4L((@;E^FNf<
zuIfF@XA#k?Yx}}@&F-OX^@Gle&L5t86{Lfk&n?uPc}@0`Jz3{8nb&m_ON~l7Bh{gP
zjqpF$T@EZ<!}=;I=T&In8@An3_PMM|VrI9|!jQFQRD4*(UV?Pl%J}|V@0Dn3p^J>P
zu1}Jm6s>y=#-BqmDn%-^D7IJ2VWiIU3Jvy6LXmN}tvQ!v=0d8^&LcY{I$^ALfB7~^
zvfZ<wonu*5pFiKM@=f)<4{kA8UV0S_rfM_cNf}HEyNHlSMFOv~4`*C~<wLe=sQ_{v
z_zS|1A$U9u*=99OrmjLB!Yw94eJ%36jewtA6vg%=CXSkk$k2FauY*-G<mC75i-F>S
zSxi;z@ne%dBlI7s8pxnoR*^|{L*8iokQ+~3!8YUR3wu`yDbH3jff@hD4AlpL=Z<b6
zD?~jRK7V*DlaVncd8dlU{Kwa2l903!(hhX0n0d&8Y=Je4&$J9V6Vz<+t_G?qJ18$d
zJ;1KB<au!3yR01Nm+EqV1AARBIlQ;Bax*?CBR76b$sGM?4+^@(^-E4jV}Nfbi3(w4
zFEsgeLo5_b$i{+q7DEtm!YVeU9NGkEcbXyKWw(aVS<z$M=tkGan*1`!uWv60yPn5f
zLbinS$xH`!5C#b>mWt*06<#X%*?C02*altvLE@DIJ1Cu_$Ai*M<~W$YKwW0MI&B#6
z!0&-=h(SXSoe@tHs{UKS&$|lFY_)7=a>?<FC9enAGW%fttSK#1Sg*Tvt_1O|(58bj
zbaPlGWi>14=<YLjb#GpEY3383w;3n+v?15es2~Nc)|;nmAOx<Hn!K5hF($6SgDj`O
zWKYO#Tq9j?%o2-(j^bvC&NshL&jN>l{HNY(&H$c<oW1drK-NMkKl+4Tm54*N+N%0M
z`c}k!%t^6*?!T}42c###NqXg%aGohM!8+$dX8HoyK-H^F!8CS$GC5D6h0%pWLaq$%
zBp<qp5Oigugw=XRkHTlkJT=DkrvW1%R&dB_P}_s;GOs^^p`sVp-*S%p`UU?Rm98?7
z$DCU+L<je+>b+wD2t5S2S0?~vhV?kfP6<s6ouB7crEs8#bvYxxKKo#j;s^TPaZeSa
zSH50l_SpiLX)n1w@ig5shCaG`J)WP@xVkYaN(goav!l>?HAa#z46jTX>N~_b8&`HT
zeS68V&<r(I;S6{E>*5)^*&~eq;g+0@alO)mHQ)-n#_c_=B~Ru-Zdb_Z1+kNiMfYkk
z^z-a9%GLOidG@S1PP4#A>@UE3sE9rGI^qVsL5o=~=0V2Bx~+~S)Ro4&t;~nJoXx<8
zsC$49s1$WO-UiYNp!uq)n#guBwfqJ>f^nTf3EXq|>ZABvkRWqonX7&<OQ-_KlA#ya
zOwOU2_`-I<<((iKorn`G(SCiE)K@~w)GdfhG7l{SX?FQD`)0BMnW)sa2x!&c^wmf#
zX7!;@e0kT~Wg({;cGE)~B1_q6Ck%uI@}~y`f`3~-41;l(n#Z2z|3tSv-E6k^a5Z1s
z$;@@fv9{e<t39M$Ja<JY#D!BOrXi2wqgu2Z08MfU&pDB=kwJp2%(bOT2OUlorw;M7
zrD^PxfaJtX>&-~SY4hBr@7nM3)47WeSGcI%g$e+`i+WV+ak9@C$5{C_4CxR@!fu&<
zX2OsUta4H=FYK<`GWROCSYs^hK4FJPf|q1veRH|0F2TrS(D1b^&2sI54-8O>C4cMS
zhtGo59|1QcO1lr9!o<u0P=HEQ5C2LT+B66gA6I|Gu-x3-%&9`ac}M${(2A2K9$A9K
ztmPt?S4hEzmH?b)k>}GZQ>dCDAA>?5fI=o7w+&t}(4-LEI>14oEXpJ}tev_NRPYnd
z3?b!v1qpx!!#N2mh@%h`31hIka6%)@b=&w6E_rz72|cDiGbIH@<K)ow15hO>Q4DbX
zj}~AZy*<VaVM;jZKwaUe?Q3kDg6xr?19B~O&|@~PMzl<1K2!&hgdu+1XxizT15~k>
z?raRYH<0Ur|3)UD-ZVDR68FsZbJSM`zDGD_-NE9u1b28Hf!ZPgyA8W<VSXMM1f&-F
zzzKM`zv3OnnkaGnqMXU-b3LV%8Jo#52!0|eYUKC}9)vTsc#a@Dp#{v#NdXpuS&oXl
zqjR8YiwX)F(db!y5)vO6$>jW@Ne9%~Z5G&;s3ZyH5po%Da!F152oEu%qF)?`#M<OX
zA%=*h43c_4+W3}>u{HqeDBOMe=tpR}8wN`Zw~M4=b|E2Wf`jv~^C9H_^JsyyCn!wB
za`p5P4P|#xKL*ht1ZV7}%<t7;9Q|_0&!WW)q#`4~nX2&_$f5_K&oLwg1Z&=!$PM#X
zX#K&SqMCuuUzUG;4dV&r^gel@cL-ZCc;g3h%MJk9z$bC-&uFB-diU-YoOlp-;OFNA
zi^=2ZJDU!GdWKNq=J|h=FfR}4Jsc#{LqmZWUkhlgjXMpdA8edShd{az)*b{GW1WD?
z&;w@-$nWtVKlV{mXTd<ZO&)!@%B%Z$p?WBB#8o~&@^T7Q%b`sA)$^20wsoZ$n`v6Q
zVjWJh@3_^zC%<xX8f@#)IH@P&klOvhp;mTs$8C>V)7oYO@lZ8~EUHuO4thn?<Aq;L
z-uNW>s6AC3Q+oI0iFxk96f>vL*}e|PesCAorNx%!rV}bH2^~*Lf&zzjNN6&=z5Do~
zz=A@vwQ{<TMQKMN(_McGQHO!1`}GQ^uU1@*jZ2`hKc#uffO}?tUye+xLr(LQYxE0$
z6%$KKR08wIE<Z)H-Q_}0$cI;<BdT2f4q;!i6*~<T)xh+0dQy=P?hD}Jnk#P+@ri>Q
zNnKB8=P&&+7DjQR!`$`~rwT6PCTc;PNIB@fT6}l<=+PTcG7S31*N4GH8}O7O)kCO`
z0M5bkvI}lwE*CMNmrzv8MaC~5-EXv@1%J62;sD&0j2=!x6$Vkk?LFSG_1eN}CEUQF
zGR#z3MX7`q0KEmGr<eX?<x2rVzA-uk8MU7p?_20efIbm8eSA8gO?c3}vAxZG{=7Tt
z$z&PY%c$NXO!`}~Q)v$$K1@S%QiN&${<mE6jS`62AZG>6c4=eMHO2$#FUDgv5F!MK
z9$)Z(ge3_!tEoA$P`jfKyc4`_57`Z@QFU;R8uIEOQ{Y3`W26-p7AENPijp<vx`1^^
zy~A8+3v?@+o2@~q4W{}5n!(_>TL=cpaYg=&B50hYvWX9S@Sq%@$|$(zGKwrhTi)7=
zhP`$8t`bOXmk@I4O=9w^M`~wJPbj;}72czRFghg4u{$d!`V>qsN5`DUbGN@7kt?_|
zvI~jZIxZ>SR$gxI((-b1Q}5Wo#6(JX+7h~52=g3V)zHPC5L+R7Y@oG^oSX@*@v!vA
z5j}%r{KL>z>NlTf&n6IA!W1#5F5F&W|5tPfGVq(lc3mk(HwG3yqmXT^nRjqNfDP12
zxIj1)rC!;w@<3J%Zni6O7I0hE3yE)|&(3_*#NEh@54M6+&LKGGXBJ*YtuozOe1r@-
zkxgLnJRS87aCE@nX(?#p4TtQ||Ax?}!a{BLO38~9h(3BIoLpV&H_ZUE+0XnQ0UoBk
zy_;a35)ys_whelg0pS}ScXg3r7@5ZRk&`jmR(xH%Dxw;o(3YL~e)l}aNUy$D69Ylr
z9#SBfVf%YiX=F3`w8Lr1iM_#$e066Ae~fd-|Dfyrc{ty0z^p>8_a_@f9nCWu9y2;A
z8I$?%{NBw=Icp5}b;vcX;$s^_APiUwyq^1fEh|D$4hGlZs^UDVV5fru59Oj-RJyg1
z5tOEEa&hBmm_k>^62xdwVItl2@$(~gBhDGfc(E=C^ITVkOyxLfHQ=Z!L57g^5@8y&
z8xFCG3#AA(kFCWZg|mSs08RdAM8tIXPBb3kWw3EDi_P2Dm*`Ewrm1W!#o@@VS7oG&
zb;YKYX$V*QBV;)2xbzBtL+J!;5>`R0sMl}`S)n(Rde{$Y4~!RioE_e1lBW3PB5F%4
zBF~N0oFil)V7eg!#*IO*C<I5ZUcCZEga!dn(++Tj9GP_aW{ML9et_BkL1!j08Bwj2
z&ae@56V!+P4e2dYeI;vvqbb}RE<xUngm-IJGSGVWX|jvxHA2$Z3kKBknZn(gMx7cQ
zR796~x0#TLurPtqA&+;FQS^Bg6@~SeX1Nhc9G55`K(l8;M4c)8T^Lmr0*#PL(Wueu
z`64tR737V2``q5W^{ju_GFX3r)VMHEh1TJOX=_`ANeNzy(3Jm<%Ei|uNa*PdW(c^t
z(Qh@%PmHOC9E)%Q!!-q70~)0p+GS(^P{>%HA9}~CY=I48d}s*$;)G#8&N}RT@2oet
zqWVe7Cr8wr`HAk<zZij9iZmu<qw50yXBwS^zz9tUdNC}~R`~tFO`K`*J^$F@U(kAW
zoZ9sY?~YH4>x~+^3pe;Y;sT%r+z07v*DzUuN!UKIZ~gfZGPIVWMHKs&lAPQEwGCX}
zuq|+a5sgYj6O^PRrUY2lYooO-1RV%T4<jQYgppEV$-ux4+ULC6U*94jK{B(OlCl&D
z0m3cceZTQ%)xWTsS?a67o}Pk5d|`-?2vjhT$bQGdY|zig^H5+XHbr&3xfXB?B4F-v
zbtvYEvB<dP;oWzx{GKME#NmOt33_;Wa(@zl55?#Uw=Ab2c#)5P2A^Rem`ApzCIe2w
zx8+?JP2NE%baQKJ<8{Yc@y#2-Y2${YuU_%y2Gq^Z>JEnPDc#|`vrB8j`D@k5%Cv&a
z(LMcVGTp99B}uu=py020ShMR1@A8Hc-&AMz;L6jy)Gs_)w8mtIG6D@ursU<G0{ep#
ze?OYOksTsm`*JYn434U+|D<A1q2qf1J_X4A<Koy&$kES&ol*N@@c^>l_6p}@YGu$(
zgzX8#yo?lD9tnUaH%K8Fq7fj=9ZFqXk=V%|fGu!Y&6(~aGi-X5lqpyT19Eu%-MC)G
zy4I`>F)biZS~|MKmA!#sVfjY&2OBN+sG)WiPQG;L89dIx%owT+Lm_Bd;ie0?hXj6R
zmVzXVm|QXgoFA?hCC?K0zW4SL2`2!^hGp(&P`kscz7z86a+O4gUFqQ3<a|Xe>-{ly
z*kmIW%ZwQ0K_i3^P8M8feuk|vb$AovBpuVQFoc?!sDp7-LNxkKXHc8!A<B?6t*ojB
z^YeWl-BZ?gCu`>jH^Zh-7l$Hvl+F1O83`)1|3Uj4`)(i<t_+A}YI)WnQ37yd;Qr<~
zSg~v5l^3*;7-?x>EoN2Wy$><zYpTb22yT%=#^x=}^#3(WNtt(>rGk$ve1I9%&WU3v
zgJkA)E%aIW;W+kIm_vG3x}mNg42aePUi5V|BgB~Us{VdLYd|=Un7xGX-3^;SWBLgv
z{n_i#BkaT=tdv8|&oHv#ynE-R$mY5~D=Rp2rZ+dX)a&K1T|0-%2%f|DE-m8t3T`T|
zPDoRCw+nzW!sh^$oZ-Mp=vT0nmS)d>*5=sm4hRdnfkW`M9z9=21z0;mz%s&YO8XtJ
z((D%u<iNm?wmNrd1bEED4xP^ah$tltV;h{2Bh9ZpQ8OmzY#BlhRDwQ-wq3pPlCv*V
zT6kI>Pts@RRJVIQ*Y3~eedv3r+T?Jt>si5{L!WPzm-a0$es<WCb}BVHD)7at*)HbS
zMs}WUrqu~K87+MY59{JR-(uD4N4j1v3CPgrQG^<Wwy2a&*|CPKbf0>NwkKowKy_J`
zLes~&^(*but+AcW5>XITBD^5?PDy<ax)|6jVdzsyTllgXI!#;_dK}jFl}{5<65y}Q
z`#+W(jlviIgjrU;FX;)CF1O|C@ULHg5T?=a_O+E+0O}ON6CoEjVR{;%5}{W7L1Db5
zMZKe}zJ3A}2ebtp=Zg*Eq=P{uJbd&hoHdyV?=xA0?Ly(hKrToqT3+7RzKG9)kB399
zLr<t4mNa!i?BJ6GO?c!9%^z47M2kn$jemAX02RhvF#z2uigVVImn>*cjD@9TmZ)qY
zQd1y6dk^tksun!2gv=kC*A-Pk?_Bm3d;|#7G3Gq(-3t}KQ{aIKlvUE!<cfvj-F@Lo
zqX#hLZ3X+3*bj5QV7WX|rW45Z%AO2j0z+m)PDVyXU#O9O8=?c2o+nzK3;Qifii<Jb
z0)oz+CAH`LfDRA}#l;a-u(e2J=rrvy>&zUB7~7g>*iZ>c93eW+ivnAXgx3(PO;v<P
zTq8~j!rr|w;mHxQzP>)BMrJr(e^4Am+XZrHR5Zz1w!|0=-Fza&D+)mEs8{OYj+s!x
zeQqFw(CPdh`>E~xb6JT)qHtfR^`3%*fjoTP_t3(+jxgjyI!A%lQ$4qLIF-PQ2?Jjj
z3bdg|%J2v54y@*4UDNY|gVnGgkY(Q;s-m#0J@?LYJp8m1kt`zZ24|i)MA=ndehUB!
zA#Oe4Xk0-eM@ulu6ZlTZ*xR}JAu6&(Q7`Pinthb=oy|Vc;h3x3{=%sfcUC{Fy3Bh2
z@jfc@B(*qOd9hu0<Wto4*A^?ob1Buyd5<qWwD0?}axp(Tt!MG6obQHVJ*zm$)*qJ3
z?nW(}B5NY4W3xwA?~kXENH<?Iu(gtk?C-Hybyc}(_Vzii0c&aB@8$UC6T(Yf-K>r*
zLl1Su#G9ZNKOEc=!j2idM03mNR^fxo*tm1Dj*c5TPc(IO6reuAy#_4x2l+CR1rRak
zMMVkAdCabmG``%`)wKa&3(3KE^dBImpl5dsbJEd>3?V75+U+KG<yh}dPR6bmI7_ht
ziQ5WX+#OwN7cY99TNY6nx&|F%Bp@Ob@sbxWq8eXB{qyVBudpyWXqmt_;~<8yK{kr3
zs|vnr_tsWckRcH+--OAk=avVA`$)fKg4p13{tab>`A0ttF!G-IwPBRCmAcDoU~*Cv
zKpvoY&<&e_ZBWyAKYc1KD{BuUMn}gCrFW0vp7^|yrKQ7iNzm>Kw$j}4a_GS<vORl5
z-2S{D7c;7}^0qgmL|;U_R!(wp67rD3qMu`91u*=^4u(?w^tp3?aGs(y2WSnrDWb~)
zT`(BeML;Y=d*Ud?{#?+VJE82}9-FayY@}N|Hzx=F)6g+&04F3GhlrHGX>Jf{0?IM$
z3Ty{W9IVJQ(Lnm9=m|(8wWh4dvp2kQ+n66G{!7vAkjTezg%l3I-9S1F^LPY^o~|y$
zS%a)6p&6LJp)w5k2z!$WPV68cP#hg*VnSBE02v-(+UvLs^n%b;s!)siwZ&vccz}NK
z3ufe0Pwhmof?j48t}B@KLx`Y&0A=Uqp4NPS9mgM99*|A_-YfHKWMYD|r{NOSj<y|7
zTS4|4Ud_y9;%L5OWAoLGgXp$_8*#*xyMC#MUZUbrs5S7yL>~FVI+6!GJCPFu6@oh#
zTsUDGIRX=fpFca0Y0IU<Y~zaC_OEk)N75r%1fm6(>rJF71c{!?$H@GTGmjJD`V6%k
z+HZh5jq-A%5)}|=2bGB<@ZYvabBJD^S4blR;J2&(?GXbrXrI!ET=BMlwv%9)695c~
zV89(xuuDUdjBdo<54Wy<pu322wbJ4D-tYtL%7#c8h(<t!CFJn<u<^T|3Oh_D7Ahri
zz(x*Wyz^*7IEVr=xF*3l7!IE>Ji#f9SJOkg0j~78Q>WZMQF54i0a=N-st=FnO&GYg
z1$<6K(}HdF9mZx0#2GZZt7zR<q9|#;i;#^4j8E}jXE?b1!ZQuHETVBKA}lH^GBUEW
zv9b3+O%t#NFeH@O5!xq#J1YPQLj4S;2|&3n*dYT00{}`T>_<luq6rOPAyqlT7@Klz
z4dfH(sBoLR#mB?b($tiWycX6i0G)m>Ej6{iRKCT3yJY|_VA@zqe%{{5v6}lmw-_;9
zo|<}Scvv-2@eOXh*<h!|K`V9ll|z{>Q3DcV62BiT-1uokYF+xJ@JzuGGTP5)IyE<&
z1p^Z16etQZ5^NM)t*7+TZs)L!M5vz9<%STjGw(yQzW5%G!-v<|&YeH5NA)na^@Ne9
z)knpyAQmwpedp0~x)qX-c?REevtD}I#>FPZ#!{<LQ@F7;J&Cf;&D46pLxDE^<vWY-
z^Kv~`Mn?iWSs`CtXPswm6gsnCpG{|v{~nFpCBEsT`G!P6C7@9<O<Y>knJh8(7p7k3
zJQJq9n{DIk!QehKQnKN=#ObM0LSZJdrQcd&<LMXaV4>|d$PWsGV@$iHt*@=CeFa_k
zzw*9PzGK$+uoe34oH+g<Bt`RAr|c{eJYf>LpN)~rNDw09HH2;<`?(q3gYaf4nNg$G
z_u9k~0U*=X15gTvXX(f<{@nPD{gz{~A`X3Gd#W>z&~IszO|Hb|d4FAEj4cn{Md)%v
zQ;gkv4PzMA3?W8J)%T!a{(%7KlI3(|8LKMU%1#MwU~+N_hIMNT+(DrP9*oMDX2%;8
z*_mnK3t@y70^M&ehA5a^I(C1`|A_X8u7>ZB^WVgegxX#^?kjuXbRxm)2{6k<kC{8G
zpOSvD8tWO64{Q7QQ@=hP^3z!-`}Uh8$+u4zlRp)`K^`4(U-2sv27p3AlD*Cfwc|SQ
zon42Q)QKUU5vWiyPqJbs)Gf?qA0dOxFKCLN8)nuE)!(-DasHKEMN~jl;;S?BAD3FM
zCm=>l>}<cgdJjlkn%y>ha(L*TV-+HOtT;?>8efZxCkQWiz#IDg?`ovjVJ7c7i!UwF
z5vd1?NTIevSpXP<Eo+sriJ%nxsp`3{m=!zWlunZ^t~)j{e$*!k-vhL91TbMv;pKLY
zrH{>IOnhWa_e3AN9iDAM8;%ruIe!z*`S2I8O%|Uaz;5Yh)o}draa}LF)AFLbtn3Dq
zGvOSP3(qlC?qli+LSCAwg?Y5(b7+(fIi;oSVu&3tJl^&xAHCKbQ7?#wI~;za*FF1`
zc2LsqpuE6T&civN$Aj()tPfMV5*!b}4M(nr_AwJADzh_?Y+x)#X1qI~oIwaj)%Z-D
zFne^_<MHz!92__ng4Jy>k%XA)YWH4dTOTuu8C+)F_t)?LzJBf6HNeug$ucuErpJ<C
zS%!soc1PI*lCteDmS6a9?>mXa9@(p={PN&=nnQ=!uUtl9jL_`RwDeOCUNeky+yL5#
zVb~v@wZTmZVSQWFvwuE`LKI@OFmXcw-1s1AKh%GEy1KTx;xos*NVhrDteo3+wgxSt
z2!A2pN}TdwasySBXo4$@k;7}oJ&*1oW7;;XHaxUhUtTVG*!4zvh2zd<`fW!@=eU2L
zMzi@UzUWQ~Cua2$Jmb61r@2y;<V4N>YI-Nn6BCMUFCP4%bD?bv<365H1oP8A$H``l
zG$9^glz8{Qk56tj#3)|kF)AqrLsjB&mU$j?rijPA6<mOS|FhwV8!X9*$91t`Sls`8
zEF6V{^WVq+y)Ws1HzWgnsY_33IDI=YrTMxUQl8}M*W#Y~Ic93SjwIW*4SV45POk3n
z^I#BQZD@%4$2P$?$hSfN<)YhnP!jml=#lz=wq2MGh=_1Orhz-SZJU!xENfiC)%c(G
zXYM)DO#F|B?f;8Q{>+y=%{7ySh2^8u+~-$VCfn{=0iH>RfynO~4fxie1#<+uAHx`i
z2L=?O!ne%Rp2%-OQGwKUhaOhT1;W0m_Bq7<R9Sd_v83kzjHds;_lE!fo9F-Ns4?+s
znuu7N)yNkxS0TGM0%l4g`}gf5e4iL8LO3-}15m9RBcfeq<1ql)32XSzdt2kk$dOC<
zwM9VN2Pqt6lBTI7*cC72QZa~fTa6wj^I282qx6=l!jFeJ{>Dk$Py}UKHpNexX^X$U
zmTMBu*Pqz_hTY-QAa_u&UoowNmlSA$E2g{gR#g{?NPMJ&yFE4+#K1ug)f%?H&(i$&
zS;Bw`Z5<t(Jw&RwE)((h-4P-TznB2deRryxKhp1<urOqZWo#RGuY3Cfh!|eumsx5^
zSOx)1yfQ|PXZd;)5&eRdDi8$JR)z^X9f*PIHBtY*i*i@eraClsM0dysUi|%8n0Vn|
z1*(9-vfPN_kWh8`b{@jd==Kl?%01m=WL;1v+-OfX`ap~<Prv;4n=IF9Yd7}xo<DNr
z(=Uu@Am+mV{YWJQ@@X1Ab}tP%7?A(?U&Q$HO#i>1YZ?`g5{yq8^eb<w;W^)Z|Nb~V
zq-$6kfC5|qWm}2&He~+yz0+|0fK@+!-Jhay<{O5;ZQIs$_3ztja6%A|uu36L*fGaX
z5MTbkZ(?Qg=VcoZ4z)Dxwpik$T@d*9YE9wvj(YWqTqrP#SO~Td<ltWgQ5daymDxRZ
znGfs}lBdWw;`njR-oL+{*(S;274U)Tf3yG};X4GMix%>~FYNI5J3EmQQ{7P{7dDT9
zyl@uAt+_Y(_x{c9#b{dsbg}WmU}V4ojqSv^Xv0H)zn|lS>60laJWfF>()IUQr;ZbM
zeb(wW3LB`PvU}^AFp+56%<g|b3<*9Vk)iehii1s@EAKL~RJ8t{@&i$CEcff8{SK^2
z<M+Q;k>cNXxq_S<YLpn!roR{6(fcytPPzQg>J%znAV@;Jd$7_L0BI0?Kv#h#0Rk|-
zaU(#5I4%k<A(aAQ6ZNg+o=3Y-?>}_I0BB*YRy(Fo+(NeV=QlwA{P#sbdxU?}yC?B~
z;+gV?aM?TZG-Zn6PfL;XRVs$dr!<dwJ@r18xbG;3(uYVo*Hb<pDiSqs7EFCueth2H
zz2K?DBOEO|BrZ$D9k(27Si8~h8eL#w{3I<aW~)D8WHQ{m`Jh9Sz8w(&h==F@W?tL2
z4N(5QrT_k7UE+W8G+a^f|4rYp5=e<#_}|Co-zD|$|9oQp--K=3wm%PuZS>#A|Cin*
zBH4=gRQJYFJl{n{*lq^z)QfN4!`&3cIVb0{Cy%8*17B-x4bnml9{uv=F-tz6b8uTV
zb_Q&vs(PUMK5U$UFsC}g|AU5>R;Fh^@roG-|5+@9xIlG73OoEy1ZMAEb9DeqQ4fbl
zOh5Sse4d}5pGH88k1r_a5m5b*quo#)mC<tv3rDP7;N|7D$xs6rV&r0s*ZY6id(*g{
z+wT3>5R##&jG-thbA$#&g(50NQs&Y?RFWt%hG<ezNJVo(gHkd_ktP*Vh9oIM2?_1@
zC-?n)KiIGTYw!L0d2(HstG>hMJkNEmbsWb!mh3PfBHOH~;r&l_Btx>^<nWl-Q+c(c
zenaAfQZL@2pYUx;#s4i(ZUWH>fz!?3P2TZCJJRr$PX88FEIcXc2Hnco@tIB+5q_y&
z{oi^?Sp%)aTQ2NK%j;)-?0C0Z3W&Vp<Mm}FV&mcl4;canm`u%DtxP`)(?DDa&Tp^m
z6<)@$@ZTP|Wd)}K2O`&QvUDl%ydWPsaYEAGgLWL-W=CKlT+Yc7Iq3Wp#_z&_g}6BX
zgsP5LS^h7_m+iJs7V$59li;*;@#5peHS5=pS{o}Um$10Oy%@?eG_?K6*hhth*hVx0
zep1unZ=2+~U8o_n&hBmLs2eW8CuD(_TzP3_XJ_8<29w9#L>6vTF!UdUtfBzNE_-i1
zd$xmel_#)^&<(dBRkTj{Jh4;$9Spx5m?pDF2j|K?Fv5sj_*L*yESTl;&I9@3!rlfc
ziV<sas;~(<)ep}XF6mBw2lwvnY~KoleQ#J~bo2lj8O?R~L~TddL!EPdz?m}suVgq;
zu2UzMKKDDSqXh<*mQ`hCK`lv;nuj&)xPGbw9N97;n}WWz{qIcsgZz`F&^;6_tge>j
zzhmZEO3(8ed;3S@mgwvEKk4hdefwX;6{p^b%weSM?UIsY&;fQVQPS1f8H#*;`56@}
zeai56o79Al5Gzy5^wLh9wlX(;K6d_EvW82SPMI|6a(cR8bmZmLBW3g)2^x3gMfa%|
zfIVG|j=n#RzQ1}Na-0Hw-jpWpe}AKmdu$s-(@*2e?6EH-IR~9Q)w}!)80d540!&7U
zt^;y~He8dP4mUyDTt0)_Cz-1(mgoF24;U;p)oIC+C79JR-BmBNW3ReM{`-8nV^vkl
zo<2Rf-Xk5UR&uc5qh2Y8IzI6o<wVIe;bQMpFLkV=ob<DtH}5`J0RFcq0B;*T#@aW{
z!oc7nrFyjwlaz|e%cG^g*O%FYgqNgCb^NZ_zeBNMqm@+}2p^;-N6OJ8<;s<{<1t4#
zmFk;oVe;>&4E)bhdt?8NxU;tKT%pkeW|td@$J{bZw6n7_d$eNudB4kZU0l*EHVcpU
zXXcuJVeur`e|)_1`0?WjE%ZvbxCB26;WwZs^)Qrjsh6^spnFO;?E%lgD8fuyO6qbj
z!<`*R*sNJ2yLl0><+(-j-?r(Pkq)^H8!Lz;B6WL}e=^jCEwN%=$<F1gR)s}Y+zpj5
z3CFU(+=*80#;$7p4_2=eK9KwANgZKc%Xq{r_fu>M6j@Z5r+)m$JPJ|nhL_A<zdPg$
zSb53Sy}RCuwjQyMkBf_A98|`xZe8{YXB;%f&B(}q-4NmK|6;23?={r{)&VfBt5jOh
zu?ZIZU<tPG2%@yJuDjUYSfvq_YP^g(mOY+${tK-iqa5ZYY02X2WIcY*_m3fW{_}#j
z+)2$kb^0J7GTFX-T)Brh_D~46^2agJv%7sPx!|05FNe}tf0Gs~Lo>5Ml9G3KK1G;w
zZ<v*>t-w~7XAPg_9(yWS*pTky|8uk}P&yQzQR7E3H-R>{IY@U6amjkwDj*w7=;eKx
zSDOVs4Y$7Vg~?zzq=RWZW<l4LAK^@X{0mtVw!CJ(nLAtJ_q`ClzG30o8ziNjJjX$w
zF^<iUFK2$u=5Qs@Ox!_JQ&ZOX`Q^1;FD6GtSutXE&6+ivHjQgI&s&KPnzV*?1#BBi
z%geE~JbC>fl;(%Bgp&t?D3yVHbkmukGHqi4mOjI*)3DWAkKzrPsIRy8SS6*&s;Vx|
z&O_nd<KmR;pV8L9$;x{nB%)soWW#5@HcJp<Q_cJ8>XD7(@+a$l<0M<nybvFMR842`
z^5sHktFC7R;ASpHq2>5VOd;ex#=DmxJ(0#rNlP2q<61KLbiW}(l$T7`zL%eWYzVdi
zpPC-Xi{5$lYPH-J^3qNV?mT{MAh&U=;k25!^*uUHRG|J<_cJLe&ISg38jOgk*5z9`
zbqwj#wCwV-$#Lh-od6Ukr_@sVyxU{XYiK#;J}4V#Hd{DrTJsP9gJ5jTMwn?tAAIVa
z0|!Ki7v%?Q>O+|ujA#@;gFKlxe4(>wW4fFc!p4`f|Lnx=tn?VseJhe8BCcQ?VciO`
z%bKDra(O<y5mGpLnBsou85q|Pm+%qkOcsM(qIo+pGt+!<AK_k!{rsO@Q^jkc?#$i!
z6eU%Nc8y4vIb3P5NicLsva}r{ZbhV~enSmQ<>t73J5C>$u8$9%$9OmJVvTKem$$r{
z6=7&*|A*bi$k+=YhA2VaM!4}Vom!;6cibNmJZd0P-<Bl8MB{qqs#|Z`l)rO1?~`K{
z&1U8v(Z;N5`SPbE`e<wuxt$|t@DN}YILiwdFt=!t``7Ajt`*cP<wZq)S^jC&I}T%c
z1|zum^t6i<X`qWP+qO-eHcdD?Nl6YyM#aFZ%s}9V9^9{=|Cux6HVQRaryJeu5~qv{
zPZAVo1ZdN$|K^XeS#)S1aE`2BruAvkYJjPKfByfJGXDQ3`Td{#P0q*FclW%G>w-y@
z5F7jo(06zrJ$mY$pkpt;IJEr}mxE_!A}0R?rg(>Liv61$TVBF~nEvM%$T-ZJ9g$-Q
z;`G)nLatHD{bHN@z{Bc)xBkC#|NlVU=eaYlsK^l4X6zD372_QqrRO4<IDUtTI*L;9
znMR#%5$1FkF7*C&PJ?v7bh?=5Y!bX0uXc38@eT#k$1s$Ji|Lni&F9aT>gRRV8c*$T
z;rmU&daQol?WfvH%lZ(}5sRof|7f(d%Y)BpX=yPn8y?$+PbiAg0eJCCN}eFcVYUcn
zWC$Yggyq74!$quI*4-okXoAK7OV}+$0L*aUPe(cNjJ^ucY+L~$R4mWNEZon1Mc13!
zjxVw+k_8L16Qk2MzX(3pz0`{Ok*tU@^!nSjrPiG-DKBTn1VGCb8Vk_&&^&eM(4qL#
zJq>-)@W@Y}e$qU7+XrLmv~KlYXhbm7wLm#UQqSk)rD|(ykxrq3AgLxBrs9h46DyCF
zB$H$iod~@q3?o6H&UD;G1_rc?GS+T@goI#4{c;Gc2d7}T6vm7>p+F-`>)*D|E-o%^
z_Fdf{i^hs0?uokN>i64nqn#b{)xGk)d-vAa@=)mDstVq+katI26WhL@8gFidUz(H9
z9o}9E{q3arFxKe&DKC@*tJDg6Vqbx1cJy5}>^)CuY{G<?$}S^iJrxwSmFtQXA@ZM0
zc4#RmDJzSKiH*D&ch9!uLLOWl(trzMVW&^quU~(6$W~TPrUw&Io|Nptz47GdBM%o>
zE~WjW>$%5@N=hr3%*>dv9~$pozs7ZT#jK&-#eUEt-P9+i=fRDn%8^#7+?_e5NcP7z
z+>G$|H(IuAz=QLbcPG@me?Rt3&(3os?oqmB+*HfaTe$G8kA<P(7nC^mc20fHUP()1
z9((Em15-XG$+}>46&gC^N3Xi-YA5z9{~ouG_4CfUOh5A-$JYoAZCdd89ra2wsF=(r
zWA)$wR#x9!wB!zX?U7?6X0Mv6bB%2g@6ao@gv7zN>Jm5th6;K^H_|96GDk7V(r{o<
zU?9Vk=$tq>ZS=l*ic%BIQ**Iqp6Y(A&{H2^`3=VFl!nZcYYaugPLsgUl$b}@*61Ot
z4vbjKHe^yrX#n8lK`*aI67>yV7}Y(0{sV=D!DJhbPZ!bSBPxpj-0HP!<z{b|beG=v
zl61^F><*cZY1IJ3D0IflmM>qr5P_+y?S6hqn8@yUe4o0qn%Xsz3B;L+@$uEMY(se9
z%hZxjW|?b>m;_KArrMR&^EzqY{c!i84GLgSL41{&nPT(Op~(58q7>3LkXNULP46tf
zmnzBg0yeNXrt6I_E-4XYj=dvLIk|m&Ho03`ulkV@hChkIuW*FD^9v4{>YmKKc6@li
zfkTIwH-o{5qS9ZUB|yd4<~RAO(<w@8?2SBJTIx-yE>%(}Vbs>$J4{WBo&%#ZR(_1U
zJj0F<SA^BbQK3?!`Hdee_T}+WU*Fjtzow?)G4Y&8bjXmsvAga2#eEgsT#THYk}-3o
z>-9j3&{x517IwhPF3Ng_hDAUFv~&eY?V+?$@!eQjTpXfeC6>lX{()e{NGI#{N`de6
zyg)B<NLbk7@n2h7(vp({##bOY8LO-;RXib<b7lVR_3f^4eQ_!dcljAO#BeZ{x~P)^
z)3Z^`96fXT*|++#M+F5w=7R4A#saMsrtRoddJk4PAam;(i$?0q9?%3B$24X2>YL<>
zFt}l~SPIYWjM{n!hsDF|K7aQ8JfTzGk<k%L7n1YKu8d_y4d+@Z(#UY2>GI{i%*}&G
zG#=){*P&EsmNe8u9&^RL)#+W%Z2Bu$?{`PO(92eAUF`7Us)5^c?VsBPbVxRiAoWBa
z_3%XJRFvoFxpT{8W0}|$39xi$93PHG5_enM4`k(nbf#8J31oVG)eplrorC5mO8E?$
zye}lqX!-IH>?@3vNckCB&U6*ZS*C+cqF0MrI`8drQ&Xz4_>2s*g6%#-r~52xx@h@s
zfEZxRUOb>8XOp9YOf-D_I4)_omewsY?7TpM8|48N&e!f-f0IexAVf<S(qm?}{76s{
zjU<Rm<y^DECsZU`^>R5ozu)-r7m<qL>KGIp%uoh*+@kWkLi6XUcUjD*1?6~yuxL&2
z(7}U+IbQU_<~@7}+Q4K(vgZ$^9f*vSO@C8w<nGy=O#06vL<euMwkD4@F1Xz3Sz*NP
zTf6%z&;Pp{*ultUrD&Hnwb;2JkDqQ9lF0AMbNBfpqm!SXPl!C#W8Lf?e~Z66<FV&A
z##~FQQ}eLj=uvITfE@28)>c>F(blqJ=EmUWVjdBtfnuW{zsf^I`Dz&O6ZP^yRh5Qm
zTPxet4x`sUx;-!~Y}MMe5jBM-awZw)y1ES7hv^n|fxMbBGaW;};6P)vU8A8QWO?_R
zfscspCRGDIz7HQlbszJ`m+=D5BSt#R?h)YYE8ej8S>fJ|`K`WubAL5+`~(^Tdlm1r
zyAT>0O1o3oicdT$!|hrM5$~?)P)X2Gl{2jW(4mw`MFczZ{d5lO+PTwsRLb_@89$sQ
zo?mkw^C=b#Y{3tccZdGS+?X|JM30APz$J$aQKzK|Uw*_n+}!x-poE#Os%lzcI5*-8
zcG1{Q;3H5}R>lE?)}kX|SbM`rrAh&Z9lx^n)2FHndXkI^VDuA|2Def{sOnh0ub*{x
zJ9j-1>>F5@?UQKkRY)YDVPp^|wvN*?H5Ht%NhY2H@M0jx3#?zizDEx)gmRxL{!&v-
zDq?~e*MkKLFKfPXWkY>Eoum0rpZ=hhGJfpzi^6br`|l}hZgtYo$YK^<T|>ikN;>*V
zKy)7F<@sOLa}w*@H#$fVvKJhG<Lu;wlz8kC5$juk1-~&4<v0$^ZtC8>JBmXx07}}~
zNNImd_LS`3pH3lRFe>md2xi6e=O|(uk*)V1zRJ{8n&*ajHc~_){dAimE$)MgieC)a
zX+t(eH}r*g(*x3qBtP=<2bRS~M9f*~ZYV23*PfS*vkvhg(#lDCxZz-8^{KZr)6&wI
zw}7c^&M?5IE2w|<?uSsT`h{-ZR2o~1dUS(*JYITWMrjP+5isIp-2_*TGEI2%{$2y6
z<WuT9h0{P%IJ<q(-@JU4QdKdvK4k_-8){n3Szp2zS0$?nqe)MU@j>C~k)<=ba~-=j
zA;BJ#6Pu!NfjZ6DX#yIK_ZmM!y0dZJx=0Lb)~<bnGL;sA;?FB>&zaLJ1XP5U&;|f*
z%^K!sPNF%5ZJ{qM&sr-g#)2Ya09W?pi9K>%x+8#W=$^(?F4Z7FGNt*Jx{RJOuJM7t
zQ^y|RGT~?Y3|7I$Sg*7vx8PbLI93)!h#k1I41TGvyib;-Pa2P)c+;`uA?w$_WjWf3
z704!nCozWeHkw&fyvVG_01omaUAyjQeRl2Gfdoi?<jA>UwPpQ~O|6}fUH|dpqO5rd
z(a~3FwW)nW@|pT+h!iU#T$Uua<`HMlRuQjqEYtdnijW6tU<J>vvO6ve3k5*;-;<e{
z`S7>+Po@|YS<4+DHTZIO!0SD+(}s!mk8gYXmMIRpE*K7JN!;V12s*X$?(E>kiZt)H
zZ)_qEzHb6~9`f+sy&51cnTXt1gVOqrJUz=py}@~d&({Gl<!2w+K790OWBw^Qho_tS
zVIfL>Uw+s2tuW?d-isez8?3Ba82;uHykX-;B<KRDbY6ioMx#%S=v@0kpvRoeYH0Lc
z0BM=8KF4uH6I)kbf7IR>m}<u$-MKlxW$4oKYLn0lkR_Ky>k>Uo@$SzIA{|GzEb~c1
zT%4_;A(nYc^A)WFu8`@9%n1w*7O-K98c)wojhEqpgsX?*@7hY6EnD``J<>RE*<h&|
zvjj7v{P)bo%QFuNeJwklcAdD+L3>@Sg6YFjmftaFPUla3kgnJ8K|_2M^M20QCrdUx
zM?+{c5wnf2F;Sh>NC>J~k6nbm%O+#v{%(oE!F<C3DL(N9*N>|AG1}OrT>Wq&2Ou((
zkxZpch^+KVcEdU~{#-<j$$qa~%K+U6>YEOz&2nHJvCd4KgXILYk(xOJ(}jE2adlul
zYd|rPA)5M~p~p`7O3gohym#=;<zj_Z$S=M?u)_x5VBd1{$@t@++Pve>FV!DLX+FM^
zq9VxYlj7ncI-VRYWb~^{eySes=;WmCR3ke73_#-)-%w3Al{pf0$PRQhowszUGhOkO
zm*%Zn^>WM2KMdr#S9OP{Fn``WCokg}IX=EaC$AG}cEfGt$})%N$qFe=L?0oA*^PyI
z?OU8w7R%kHMA5e#aJQ&v67W@;@A!@fc&e8M0dSR^L6F4#2Ob{a)KZ^`a$MQyCe#|T
zxzA}sjTj##S3tYl_3Z4Ni>v)ko}6x18v84w%fLAjJVpw%cf-X3Pn{A3f997Agpr;s
z3FT7a4?ZhayeEH+I%<vJ5>W5`a0jCFWN4b&dfP-Y`6)DmIBM6EsM1oAacRTuzrUY9
zdE$(guBK_;B5*d!q>xE#yU5jsUgRY!_T=3xTW?_@)9m~VZNT8UaS82(11}jW3hsJ|
zpMJH&YY}BKC;`pqt5B>lIV&HWw8^_<B=r-NP~u<-f1zOn_sGg@wlK(KAr+S0Cq1GH
zJ^brXG5+>l7Y_VB7?9WIo)?LBNVJFS!2fCiLPF^11+wLKIDO_!zQ?8w8{|`pa&iu)
zso1NH9})4z^C4phU92M_x1Tr#dqdYq-s}e~iTls3Ze2>OFNuom6D$x$^_qff-~Q{5
zL$lw#rYGRd8@Wu&6)W_HZ8MC?-7_TqbaF}x()kq)Yp!tXFA1F*S8fCKLwpF!DBcGh
zMExn#E`?>^EXv*=dkol;US1D?UF|m?(~=!TcFozpSG9{-)G(5bNBTi<Ik~vdWT-j+
zEQ7E(!x&ATYk3JToBUpDsP5R)6q6<TGz$-tk=Y=s+n}kr>-({{cG(A*gj3IW_Pho6
z$$4>jVJ-kueiy`h9rpG8%*C){$Bc$-aCfAb49nP`ard02TS-R~__VTLf{6c9@%bL}
zCmtMuf|3u+E{);~*j&{oO}o!XXv%@>*EJj07rRL?GCE*S?>H@4ip;^Pc2a3muDLWU
zt2nf{WZm}Rr`q3kGKMvmN@Nf%oFZF-MW)&^Sg=4sjO}s||EnL(&7%?ovE=O8=QK>9
zO}I<+^o}t07v%a6&9G#)w6^kFeFjbPO#BXuSN{B*NVN=&-pp*<CfG^yKRv!y>vI3`
zt=Kk;7UlCWIRW<B7D^JL82L^?@e&^&&#gkf_mij2R8MLbaum5sF9u3W8=O@S((pNQ
zBo9-5D*K=_XQ*%ujf}oB){B!sW@=H|d!H;#l;Y(}HB60-OStgida|?sP|fm^hRc?H
zhlHj9#qaz>avquyK#f<75<NH<D@dN$k-vZpl>iH*^<XnZuyyOIY0t~M?a6TLpBOGL
zZAg%xU2C?p*nQRDxiIk|?)Eyg<4xSW{a8Jj6ENEeyXHjHxKwL-%j+*DB~hHN$^K;^
zjY}A@+1ib2mRg=uCBNkjvvROe8Rh12aFi%c&@+gavuED~JBLJ9P<Tf<MV7w>H=>(2
zXE((5=-2dQq1Bcxh2|kGOLe@%ITp8mU&9^k1>#nA1c_oSS7W4_Arn3s^fQ<#wNwHM
zi>`6oo!QwlhJLz$-oxLpzVpWR6Fb;4dCsV=^%JsLjFRDtv-`@e4Li88>D;0EM0#BP
z-a2w=W&yV|%-&p4im(q#!b99jM?`9~_Y8zl{rglCr!o7G9)D73ut0$WP=C0d7L4+p
zhtks1@v|+4>SlKUgy`&gB+{eDiC#%8<#8Fc<nVA$jRHpY>9p69XTug#aKJ5k@C?Cy
zuyfOOA&9Mb&~{i}US0@#*E8zgy>nqWDzz9~2cYQT!-rAcc+YF5&O=BOa&E~)5}V&m
z{zasyw&(i|9B(dPcrXDJiE?pi{D&8GZ2ZN9M1D^Zyv>h>rxTG*WL4nhmLp>`T0#`3
z`3eTV*XIm!V16q0zSX()O|RLBLK;L`0traW<2kTD$j>O0Nt@Z9N)xX9#2tVLpz_FS
z*3+??;VM7_liMO$P|`D^ga8FFbrwR1R2LH6+xrm>(h8p!F+u)xMV8Q(ta<wKkxA`)
zguO4FxVKaMg_2@7*|?okSDRjd%w=9{zFn#1C(?=0dWwvM7v3zvXt_FM_odj_HRb+$
z%6q506I)@L3CI>{gW&>Nk?MGb(W8k<%S=rte;zre_AtF3{wpr>=+zS$#O2dsL~nB@
zSaMU7ljCF2K7ACjJgcK~i!0vJr7k6A>q3%?WG_#h@Nd;eb?qII<N`)~tQZ;`5<*q8
z(P)@~!DB{`Sr2P+9xEYwN^)*gPJW*jzFPTtv-b%IKe)X7m(epf+gLo%{YE{Paa^H^
zGNtr}Kil$B6o)r_*)e!`v-hUT45RAw1*=j&_@${#ovJBt$%^G}<<+Ba<lMV=`F>1+
z2msr`^}|w2zyL^&itleTRNY+hjU!&^54ahZaN5?+E<p0jyO2xz5+bdm&fRNHlnlyW
z9_4d^XHomJcHn!$>V^Z9LsQ1a*5@b0KJlAdn-CzCaZa23s#S(Og}JFpNww5Py3HcC
zHrkd1FSB%GIsz0%<bvZ?c4gcp*7BwnLs?a*+(oknG)Sq{7M%ufn4Az56SLGSl|~)U
z6*1SZpILO3A7Wt-r<oF&YX(07>u~FSAuLU|mul{jYb7}W3JdLK9&HJ@3y_N+&~$UT
zixbDUx|b0t^_+@NmXY6|>TI9A?!jiduTnjQ`+rxzQij<#aH>nLFHBx<Fv^KdwZ5wN
zd;TV|CC6a8b(K>#nf$|EvAr=_p4dER%%W4bs<wROL|q&zR(wBi8od(}Cob7`QnT6g
z{Q2`C5A>7!`bMs~QUA0@YsRmL-S6I#eBOI^s;Rk|@zuIz7#`U5iAqXhw9QgSasBK?
z4>L1YoPGSjzgJSq_L&NE{=9kf;lm0`%g5%>5i2*OaSR?+zUCwNJAy@DdqGh-$|2?s
z&T>v3*g5y9y{|{TFnUimlUa_P*8}onHibml8I*M?`_?r(k}SpiX)ri5<4XDYd+*+g
z?gM@6aZz*R$bFiIP3~th{Cye5!nx-zQyMpp+H7?58zBN09#fhyBIeSi_H*sO=TCiB
zgOaX#T}klIe$!oYDOtHx$mYntXh_?TZSy>QPS@o23!6%6-->kE8y!8`dDuwDlyp#*
zN53+vGndtL(H0Yr?cMZopkP7dxE5#0yyyNS=|3IyIcUm=rx#aFued^wT=0*Beuwm%
z97|o}*#b`UD6rNni$(D1wpiO`+K7Q8Ml6}ODD}pTRGZdI0Pfxi*C29}D?op(9GdYf
zkre~zl(e3IuzdKQJ$oD-sS&2Cs_K>@J9wNZP_8Nkd~&r=ovNW;ZU(wnT4$8E9;xg#
z!u#-}2k&ZXing!5a{02Nvhqmrz)2%;uB~ELo3+8PVP|=>F$xMnCr@s)w(j1eN6LiO
z_=pJqz(5`EJbzz-%ZE<o?qlq<(EToahGDfY)CHM>R~S_axheY$$<3Ck%(CoVuGXqA
zw`T{+jTj>#N-i_o{ijwhK}=izFyFL`F8)4uhOBS{@~SOvgc6r|+Kk7S&L3&STZ=!b
z<el(cZ@FOOs~a9oH>LW|oz>W|zcoI52We=(^mHwn)pLiX#aXU$L{yqH-dR>c^l$UI
zcE4*w9T)B`HK+-T*|6PsIG?!R^e#D%9(lV@ly|$G6T)_RZer)-Yl?3}h81H0U6v=B
zk?mPpdN2{i{2)o4_xEj2N=hO}cMhSMZS<0`ahZPrZ~8ZV`fWP4_Pbu<F#WvAV>%;D
zTU{YbbWJ?>TS#W8xf!ZW$b=pz9Rvsaims!4LzR(t(4j;3+1@l~ZM3&fKt2O>&SeNR
zHpRZgditavouoxYXrt-Xr%y{$Q~ytwFvMshH~R}Fk!nXK)l|61B#hg=R&$)=ljM6=
zYB@nC9EGaa&T^0EU5;qZCl$dDyS_%BnN_u3I@b9kl9K-<$FGj<*`;T?X_RegY(KZG
zf`T7`@28q<G;&>d@@6wSjO%M|cX9k-e986U6F1rEa~3aIvZtk9+SSU(U=rnL8zt5e
zhnwIyAf2gVpxshcIAPST3SUoEfeB5&Dwb=i>xr6szk)_Cd4C%b;nhpq1a6_)z8YX!
zkWM^8LiUO;kFUtl?|HBX-`M@9#{8<Ux9jOTuAis~k`HF$IMVk7yg~c_&71A0i%ykc
zg%o@43#n+iZ45YX8~xFg6N(kHZg5QjbL8i^bK~u3|Lf6nW$sP>Zt|r$*ubdR+u7TD
z551qeGbtvlPQYq5G-^`$G9l_{g|=)zk;ycm)_=f{+AaCw&-@GL&)==={CQ=are_n-
z?NMAJavdjL`)o(Hb@5dbx7WLjr|?u{Ns%uAWh!@#DplfIh#bnz?&m!AE2x5hLh2;o
zbsWu%wyg|HVUB#?=X>6*S~z#*%G0~fw#aE8@|!JxDqEq+ZWX0p?s7dDr@^<$dYQFa
zkzG3YNwR^PELp?kELErbDC{)Uwra_b8@H%JcP;%{c^0WKf#^|8Q|AK3QnIROtR;lW
zG=v)&oSL+^{Pg6hQ!hN8FtGL82T3=LDN|0g?wve+`o>QiQEi-3JLsn#ytfCm8o9_T
z%$Tv{B5NZTVV5$CH=|EmQ&W@8Y>u<OnSH>14VzevJx2aF7YUmSvTbDH8v!tNK6xxR
zH<v8Xf6p|VqSr*$%3a0_54KJ`6Bd>f``TrI%ZHKAozzRuo9+1=cjin?gu3_6#hT8w
z$IcX!1i&@sR5jGp=rk@K%Px8R`1tR6)s33lK2?+z7yF;@lBT_27Y?E&kdQj*^N$ZP
zZJ4?!nQUzGD}qw;`&%EEFK7VX%Ac#IAx^y*-Qoi(ZrIRe^Hfk;>j8a-Og)@gltOE3
zpOBl;0oSvV0piT(3HE{c3i({6u_+&f2PreO@%{S;i`J8zUs+ZB=2~+R&9y{()<hj|
zWU!XR#E;b5a8{%Dv=jl>d2M$)-C{AZQD>rrbJ9d5CAmvWgbXeBiJRiYiS2k2gd_~c
zu;fVQ2SHfQhndiPev-%Ovpo!l*Syub4e3aQSYX6xxqa6kpY<i#SYI$EQqC)`*TK2T
z)0h2d0p8s9mKzvUGjL-+N1w0zN7nS3@LqL-N?+#-gMw00rl=hc3___w&a9Je>uf57
z#|blmqs;A3&>uhNW$>gp`Jv@Gu4=pU7T5?0cFi3mfI4@(D%`VBG?$eSNRgJ2G1=KO
zLrkIqwpv+~dkl)(()q|QJxK$JaMLe8er&XqFql8z%VL1vx8JSKR;$CLZJ#Dp0ct>=
z!qcdcx)t}og9Mo3o8!qB)!6=Vki80i1nWy@=XgxIH{Nr+!{9Y4t2^on|It&XO?$~s
zBiUt}>FYn_^PIXFoDd&x^Hx#!MWoTh<&Y@*+iFIRukDpk8fZ0a&YwYt7q&Q^zoB3G
z>U;diLfhC0_9Yj3s?FR^Mo$N4^{ipeZyPmZ6B16W*Q#mlj=D$I)$6sx#jGq1S4ALm
zmDO6ip2pg$#;J=#OP4ebP?omZHvC&l3nZA0@10PUx3?<X2B|s8+rL|T_t4~<Z>Vn+
z^=>90qXIR_chXZAPaMD2LkiD+>j_6x_ub(%w|si>gZh*-1GUmnvTR0hBE0nRSxn)N
zY8kqtm1;G4_*lL7d)GJgE_cpG0_%1Nn$+rbo#qS;jn%ks(ne16yR!1p%x`0qKzqKc
zP_>V<$ng99<kqd%Tp86#OCQ$0`1mjSh|NnnHo9-i%EZ*v!1K}AKwR|rLl?AvdQ`<~
z<23Mw`957UcmLVRcq4jYX=c5^sRpTD;G&@4H1JGY8PpF&h2YOY)67c@(1548;Z(((
z`e`>Nsr_i#+^@+}^7!nLkK$I^{2?%hwnu$kUEi)?*T3b#HRBkAD-&KvUA#Dc<1<pS
zCrRJ<utn+vD$OqC9?2RswsC^qga%uC`;_62)JGlr6_)JqCKF!-)ybz)BJAz#<X^2c
z7|tU?jonL4{(vn*T=LUWyLFjkXIIZT$eo?@xU*Dx?=u-rDn56*DtDd#h3bS5#_^a7
z1yNAoAne}}Cq_akl};`Yn99G;;7a8i_DyVB6Mga8P5GOBOBGx;3_Yba^XQ_qQl+Ld
zhw3u`dr#avp1$dM31$)4IH1X8Kx(jy;(#8yF2iwqMh&0}^NA(``C;9M55+gi^5!gY
z+CDrrEiEtSr_G<4^=fx;^(rsu+LFI=goZdqzTw|@ck6PSjDLk|iPZe~g8lI84F#tI
zwh2?D^MAQpHE5D#A4+>qrc$tIieAgUT#zekxMYyRn;9<aDF6hjFLZRY%MLRJyi}Q(
zjY`>_89PS^toL%#FJH#=p$*Xia>{KkE&>Qe`*pd&@YpBk9zA@>5CJ`pk>J(=HOFLw
zwxl$7dWiH{@sx?i6sIv^@@hQ{2PXS(>^v$=sAR|^O1ByFlsUANNv~+`=6Qmglw~eD
zeYj@gj0XwXAg!nhcm~^$)Cs!J+}ziL!V5emuLyHJUNIgRH16-E(DsoV{U@z|IIT!~
zZsNq`xx4RBm>&!7WR8DrhP&8O2=f%bx621OKkl_x>sIEJ(Z`{&8_IIb_bgw~|N64T
zB})~e7lqt0U;J3+LtWivpTCkny5(fPPWNoObOFpkc0xu%!dT~5zLefmV)AR=y)#;U
zpU~%*j-m2ZkgRTL-`XcFx2ny(zgso-mAJs;zu9$6{b5}B86SQ3oU=W^IKUPBHI_?=
ze5F(bH2y<{NCttg{wcuR1J2!QK5KH0(YdBj>mi<>3Gu<Dw|)DxhIQ{piha~9dZ}Xf
zJ|eW*{*xvq4n@nN2>XHVn;%nXlz-c+qAp%N;L+QN<}c(d>d8LGLqjW&$)5rZ|4{mO
zc+0{g`ti;)TiO>0M6JbdK%pl|Kj;~F3`4Zyg!%(T>W$3L!fedZi_gS{2md|y-^DAn
z4Cw8*?7Q;chopZ2*ihky30hMz<#9p5ET<aKyCsNY$a(^+0?4roPEFQsT6nKgQu&?R
z(;4%MAIU70GaK_u>4M~z_3NiFdBQFaG&9J5r{9T;R#h>JeDD3n?<Mtes1;RJ=e51l
z8hTAOz~BG&kQ15f`sO`*c1k^M^_$MpNj#Or_O?XlYJ3cH)hqU;z3Z~~HDv`n@;aX*
zYd&q<^6c{7MA->@lt)s=E9jpsSvlX#+&ti>9|S$d7Z;9eomDn#*w<6}tBdP!07IdB
zj}pc|>s|l=JCh!`wFTHtX;<hTFgqYC5KHCOBSX*j=y!12@I$=^&t0h@Sn3D@*c!80
zvwtgp)RYi0a<%T4+-N=a=c|IQ4M$eVzPt7kMoFj*D$Bnoju}{f=i|psO<4!OKfkf{
zO_v5w{qhh+a<mNx={9jnZT)WubvY@&)HZhJ;#&PL0Za2#r&7^iX$AYmyubIrVwlbW
z*UgN3+8bE38u!i8pkhdoD^yI$GfBZUu=fDUn>X)}p<-fE(&$Ud-Yd-x%;S)#``6>K
zVEjpXyn|ccgD`HLogEHlRPIfs7eTA74r4syJEkR3*&80K7adbQ9I3eB=c4Skclj3G
zmdHw6-~A^POuFFgfH}e<aPj1~T<cc-bm02&Cy_AQPTq53lB>lZWw#wW{L(*HUK{o4
zO{w%4AZMi|3j?>+{;Agx=laDcWe2wpattab6@WTQyc4bC_Q~>}C0x0AUgnS1kl5Dz
z?M)_3{*#T*3pg?<9z|K4c8Oh8sGd%|>+$cwMjGNcxdg>G%T6$glt^5ybhKWxXuB<{
z;q`Ydl%e3t-)PopD*DQg7fOdIc@zBJT3iy&aARa&`!Rdd45xNqcKdPlx5wLuYw7v<
z?$5+m`FwezNK4J_%Fl=LH<#`FWK!9C;%;EQpKWcI%+rM6gvOkkbJWvK^ukyT3%jVu
zwJ@LxQ=qjNjR5H37Eu<VjRoaGSnYbJLH!Hr`niV2jC`dE7-09;KeOvV@vdE??pQ2X
zWt63a8%Z6k>4o5$+S=qskNfx3C2Oc&tLsx%X%3uzR!7ckKQ>6l@vjy_ni;R^6~)=Q
zqdxReNlC!3nA4|+rubxJcUy7tM9in_0#R`5mqhiAyDJ_AsqgbW(mFBm*`Ja1<7nn-
z`hI(Vk0|wNYHA@C8mkGcv{kH_|8C<dD)oKsXZ4mp&=99+WCCzL%Yl-(m7pNV&qCs`
z`G40ZZGxgA)@Oco+XdiLQj*KN;q&LuU|)#F5)u{;`xqTtpslOh?b>3Qi#L~V)DTb1
z8lV{8T!gcn(|I$0o{@e=K(fhMuj=Y`wSS*h?^6`BN2{}yY^QqkrlyIYacO6^m!GJ&
z7$Gg4GcGWuSyoT$pnYY=Wc!C<X(w)(<5zR%if!T@(8I|ar94KFpZ2eL8{Rllw3?4v
z>5?l()yi&)s9LycL7a<9bZ_VFf%l4vwsM0DJoK6q082`?5rz?lgK%odoQZV<H_28L
z9~>{_5o(Wf9gvooc<f_y@#1D|lQ)cM1Nsiol-a_N;w57dW}BFNL16x@qQZEg{Ppsu
z&_&j7P-K{S>B96}wzf0dFMKCXvVN0DFEKG9iYu3P&U9}^Ff9p~jvd6=#CNGH<fw_Q
zC#3(VD;`xlF$(0dys`njHVA%=uySYIhEJ~sN=Z%HQ$>=}TzTH=ufzMleSh2Gj~K9i
z36GwNNU`+_I)+g_cET_1bLs8<u<$xrgh;!F$o@7ZG%0uj<y9=|#?$*s0RV`hTPw-j
z?wAV?;28Q{6Rp5_W=8ixm|ib<wzsFa`1_<U;9Dv)^+`arXEm6-zkToC0ky63_MY`Z
zrCxNkpI=P#R?6G*hphfOpRQ>%)>v-WGTQm0@$tC8t#=cUVo@M|xjbbTX8=sn{_zY+
zkt^6ATf0G|+YIb#Hr>za&wQU_$2K^)QTEQ+RVAbl@}D&?4>)={I6*$K{pPC$?>h&I
zw9bfqg0YcWqrs9TW74Mp1Z-S2TrO))<=MXn4jeGgnqQ-I;#{5*7EFTSu5LQ<oO0sA
zvH1~;L)TbOi9=>0|3<nBkpprHXOkIlQ`aJUIB(LWU07#28Z7%yXPb|-o;Ii~TRa#x
z8a{8)p{|tXj}|?l7b@ghk~4-%1on_Aqp__Fs)#~YFxfd-8^r<7QZ0H50}~T1u9e_m
zg7Pi#Z85Td9197ZFttatC~d}A3VvqB4Kbe)vlsh}iE`S$o#IL<?OtW2J9tB?9i1hR
zS{I}-9Xa$oh$uPb;+B{%o+H5&*4#Lz@;U6y=g);QuLqEs(=A{wSk#0pT}lwmdoMPQ
zsxQCQQS;cy#9lb~lMBuJi@JYX&oQ~PcYN`*c%>yPad>Ff#il8j&cw($J3BY+3TjhY
z5b&j%IuUm!r<0573x*!mJ)4wM<@S2_s93~Hchp6_x7FnKYnt}S5{b&llvd;jn)2)9
zV@eD5W1T~pY&D~Z_N^5=z+wITB+TDHBM6N=iNDFcAAQLPX#V<jZDZ?Y$QSq;XBlav
zKnTl#>ah?DOY;@FH)>aWW?LkyLR>hpbva-38A-X;b327iUf<n*<PkS>d^BY@XJl*1
zV!JZK{pX~#11%pXN{B)cX*WnJYldsd#>U65yK$%~zwIx@!Ks7PxY5UFjTT)8yPF@H
zY83hKzRcP1sHlVPGP%0(jyJ=V*{;TKG!`u0Gf7oddbyWonrmIrUX2Pelid42LPtjh
zYRRI(;+tK|Ff3h_OBEX;jg8~<<hd9QXUss9yFt~l@GBid15JyHizAZaqcuy_j!|t#
zRxNb6gd%8M;|z}cfHK5q7pioJS)`z>bL@w-23Bfbc`EpHx!j9U5~A1(0w}J~CVu(9
z{Yz$zY|2&|!G~Ufvn}YLkx{=NDQ_Rc<b||*fDzZuI_hE=Aw+m<il2+wgN@n)=2aeT
zfxcs^vXt0C>QYjzb&ZZn$=v(ai)w?n(qX;MV;}2ab>AX;r)pXOIqHfL6=h{D7{KwG
zR8~MVWTZgIf+-iNK-{0%<r_VkYYnV#fg(8H943>S7FAYS3h)g+f!H*KPJU<*U~xWL
z%Bq{vIGOdDvPMToWb?Cqxs=avCFJ_JeQJOemn-Cxm^{+f!iv7CDum)c#cKCpR4;kG
z3E+It8@v#Q6XlIyhnRSL!=JwI@BjLGn*Ja82P#iQ(6!8sF#I)UyoTf00)#koo{Qs+
z^RcdO!K`&;VLH3)oY@}Xf0N$MX!%EDNX(I-F}=U(<*QedV1kaHICpM|!Mz?LBA1tq
zN?9*xFg`7pYH~+oxgg71T_?09A?T=#iwaa;g=MKvC0S#RE&$w2m%2&}S7%hPR^7#i
zkDFjJcWxKARW>y0q})M&kvn7dpKp>smcD(0>Px$=2WHtxEwQjMVuOSx!o*f|tLEts
z9~TzJ3^TQ~G)VHE8+HDC3Z_ALIFq~Z7+=3$Q&Bd$U!Ok5FI<o}UsOJLJUhqQLnCfY
z?0>ZY`v8R3E!+M3=j5~kfdFxV>|@5<PN?JJ+6zeNDRH2_RwZxfxp{xD+LY0v`&<TA
z0NkdRyB?5Mb6YJdL2=jzM{i-pS7j?JDuVJXp*@l0Jfr1lX;~T4$lJ%RQGi=p{YMoU
z6LS((B+PtBE_xF5Xq12N<zySi{pAcl{66HU;c_EcW+|@D-icrJ<;28s%F6X#BX+MT
zrwtG_b<p=WX0vw-ouV=evqKedG*oJl45q3daq9TSoL|me>t$&v^V%{^0MJpA-~VNr
z<xoXkf-K9;O)aGW)+0|HL8Wy$UoXP)&A=Zre96V-Zx-~^uO=;FR1$?oUzgkKL%)67
z<}_0ZxtO)V!J4<=V!2<(LCz-5UPu!F`r6ouMi*OW?A0o|xu>0C$owpip48`r3rYT?
ziq}XD0$kXcK;tvQEEg{>pm9=&W!JqkZxOuCsqI^Exb<y_YKX3a+bkd1!xZ<E$1c6{
z*B!yysFo0YVNg0JIRDhq#}E36$Q8Qx2xDBG-IM4(4fYEpL^i*-=n0{-z+x1b?VHw&
zqesPR^Bwds&vSGiadgy@Sq_LvQ@g^ojrlVr*judGQDhD_hWmb<V~*8*e~m1v0@cBc
zCm2)ioOBqKk(g#5qaGrNm;6?|;ryjuQgv+2Aju3UG_tT;GM7XYzf`An>qgEWR{U&j
z@hBFrq|wx5)`VHJ!j5qRjwlzuOYUW({od~JgPtP$&=sp_d;I?CukLf&B`YR5S;g)w
zL2t?{k9dR_#6^|WBK0d)RvuZRoY+tub>YI+5yFEAzi=^U?c+r>PM8MF+`QrsV|?;G
z#R?x-SzA9$iB&9YoncG;dPLP4abn25lMlJHT6@MGbFYM_wO#$Xe<&K0-r{4H%<>7z
zwcfrnNUSyE62=8Bchf_Bxpewl-lAZ8Lsmk5L*mTMwWTs+PIka~y08Zfxc|$HTeEcX
zd<Kt!PZluI60OL>g+O?xY*wB_d+8^x+^A-J^4cGK>|wE18XC&p6<@GVHh1t~2xmYT
z7*!J#5&U+1LIT<cggQ3_sv%M{(fid{`4RJff7a^m?A(mRj6M*8Bo+Y=9z5Vv7;UpH
z#Xb;dKGLqTHG@;`9X$vW6+wu*t#dQa7^PuU*j-dXd=s;w7Q78tZF2P9y|+mtNZy+W
ztfYB}-}x8HU0~d6s6-=s)g}&`b@2V$Qzz}1b$Q?8ut_T#v{>)gla92tf^c}$ruvuz
z1w!fURSmz>>`3LYc=6(rtA1}+JoR@!6BH!<U{D_UboqgSULb79=f4bK*YP6E?~5{!
zVEV%0HbtisV@Av$Hf(e<MgoH?WW~#Bp%q1Y_AGw-bj*n53m4jxAd{LhH|V0FqA*T@
zS__{JgOzp<pFa<2sqUj^x^Y#vutN~@;bi*ds9Lt^=;$DTF;sN_M*!)5w&WG^!~-(i
zD@jUnX9$;+k56rZq@(QYFGOea%xxb~LcF}WXEl@HOik%dmK-pE36I;HBtifuNQrIw
zNd@_iQq&d&YXhwGl~Eg6(Kvec>~?6IJ-@$t+5EZ&QtIaDuYtMS$!}l3UgM(0BZmjX
zfCy4`gGGyS?Vts!hy1QD8(Uwohn9U9%U<H*g?V{#1TEMI0|UW5A0j;Oo3Pn!=17J4
zjuPGH7PO9&ROz)K$XD<Ues)<dW!4eZDY_E(h*mRv`p1yjC5BkBR-*fG%MToHFe>&%
z>kOv!`N-uFW!EiMc}Gvqt;ze$u5=E9qZO?tZAv`RX@)R@V7tF|t-Hm}o$6D(EgLKb
z=X7v@?iPQ(Y_5HaL{qq9%?s}d?W8+07*^D1axgf!9N;GJN`IL`f0vUZ`~R3ULAg=d
z*oFz}h3dcN8MS6qR$7$*xfu)i$8ORK%g$(a1Y=%jc{P}^RfnH9L~>zosJgmDW_Ocg
z{@7M{5}Exvq92+a5o1*E2$h+Aj$^X0YE9BIIxzC{M$316x1YaTQ*dCd880O&LKOj5
zw!!0I!`xW~%;}Vq+pDpOe({<YR;hGCjdG1X=J_y}AlQ$SQu)dJ3}9<Yn5ev*^DH0r
z%=tQWCKIrdRkf{^3x%KR*H5l&`Fs)`1-A$jr&`mGwufD=EKp{KCgZnNZjYHT;$%=z
z&Xq|c8(K|d%EhnG04G>8&v4DGM#{U=K!vf0J_!Cz&IM<EHBb{`k$SBw_tP{qEax9{
zJfk-I+&g~GUB+dW*8<+=nz+C6i1`xtgi<FgI>TLNjs&Lo{`EUwBI+aptk6~(#s6m~
z>#$1{wnS*j+A#N1=nET>KB49<IRN^qTf@}wu&?a7PifL5!^#;RRw5?G!o5H@?xd3m
zk`L&fuAF~mnacaOvD#O4>Df|VIdsDz_sXX|+SIhpey?(Uh5hc*%HaC@_wN%YtQKoT
zqL>p#E4EZC2VoY)sgSo2`^L9<9IATaLiAi}5`i)0Hr21X{{nq3^>}q~x`e0*kqil*
zpYjP6Fmt5sha2LMg49A~q*~3gVS7`FFtXDQacSd&<kVCoj33~1$0OjTqhRi^3~pai
zFR82`B$ib~P5HZf>4ZD#<3m%)(Xg=Lt_Dk&KB2LQM>6R_4<+uoGE|1dg7JBa&B!eU
zHvldZAwSOBqnxYx-b7XcK_BxBR`g$`bu~HJ96v}rTuJb+*&|yQhO4lxt8eVnp;{q_
z8XzrgYTs3AxSw7{uhK=JxyQY|PeWY&(7=lJ=YtV@baZv|G`AaD=a*&SK=g3tts47x
z_4U6v-CSJdnIVTzN+RD)T77duUau(1oZm4KAl5j}&#P>Gnv;V8ibcR7`u|3#IUP{%
zYxJCX;O@VFeia&GT8vnpA#`_5K3h~(^@p$$5YtJd%Waffd1DVR0>WkLb>D{I8}i`q
zz-t{M3`s9aAY#7SfbOQ@IA#_lIaM)Xue{1DN`-Gu9x}M^8UCI`4#~Cs*({;72Hic2
zlnUS^dfG$Uy;RlJe}VBNy40l6LV--4ECdiT_R-RE{n`R;gvxX;PNO^AJZsfR&Lf7n
zGh7>GTx7@Ku5F(4;6Y)is(0HMEfiV&A4eo$=6jSBHD!-$4x40*77?gV*zgI!=-YTP
zM8Gwz=M~p9XbN(t_tS1YoqyzJb7!eYmYoiBY9e*q{Anp9DgY{xn6ekXH5bly{*kYP
zBJtIAw-nMG!C@IYO%5?vFnwsW;|P)u_xZDFH^4>(%{@PIFDC~j#^C|&!QTfzfAJzF
zI{F519jA!Y5VTn^nQ(I>ZzVC|WspEOppV7dvS!5$Z6_pq!cab{6voVO=LVfUi@S_1
z^XqDAqy`NVCe-w78*IX1`_|kn2(#G*+#%=Cb^PlGpg=%%uOG0sxU>|@&fC<O;2f@8
zMlij6G2AndsL@)&a}*qKI1bsg`OQ`D+&`2P1a%dsfWn$rU;0`*aaFduyu8k#vp+dY
zj8VA8ECZAvh+$vz3DHQjwzk&S*N+}Onr}iXLN^KZq5EHz3-q#_o@!gdmBLgO0&5#D
z40ZMH{riCf25=BULqZUdXEGJn&CRiW@qo*pwm?wLM2i&iqMU4w6n6_C0o#Pbj6{cN
zhd`Y`8wdhbCrz@!KY{4ZaV~*5#(;Uj?w-rFwbzqNvQ8XfJ4zzx0YWzSAH^^|Ll~4{
zvR+4jT&v6y-ENpm2%TusV5|WePsYe`&^dmBEfH(Q?Ph?}0s2EFC0n{Bu3dBE34&gl
z26f?fyqb1KOZL>vUDMjf$BpXUrw@GOZuAh`4tQ(4<Dx>Eg-{i3JI{rH@%_seJTbHg
z78iaco3@55|NEo}jh9#N!9pi38KKaI6%>TcZL0m#Midd>TUvsCmkrX~v32g;J7`s5
ztj3s!?Q?liXa0va8G00@U51m==(hi4Gohbl;}Q&^fp{NJ+{k-x+&q|@7KjJGoj)Xb
zV0wR=-gzTDOG|UXjSdXq33#3D$uUn#O9lINRD3@^K6fU`Dz<mnyqPoogMje#*H}vd
zsG)=f^ye95?fQ{aH)WLiqf@6(rz4}J-Qd-$b1u`nDqKlTl|gpS;^hWV9`nWWuF5Np
zojEg`J^FBa|BvK;Ff)(|Ha2ynH|XYqPM;<*rA>yGmJ&uE7Q1m^R@S!E)&MEdX&KL{
z5s+5`5+59h;bCFnK(T||97t9vkv{VY`FhM;&d{E$A<mH$+GsDXWluseBk5kv)7i0O
zhnblfF7LcEbrCXNR9~F(W5<tUKT^e)b9aYUo#6I`rX2@hcJ?_Bh<4_qbyLkPEH<%;
zY;6T`BZ@3!!RB6Z%9^qb`QmQVGGeH69`i<2kd7ZpqJ)*?%UD?<Kc(XZQykZZ^2hW}
z0(ZuZlD<b}*r!*o^T9m3r|R(o!?!CBKi4C`=##r?nBJ@w>H5BT?+eTAQ8}k3{DGNI
zE0`>DaJ^6Bn`g3v%usfetP-wi$49mv27E^CUDYyUjg?F-PhI+@+B$Es_G-zvT7Df#
zY3hb$;CH4~v0Ho;=gjjMNXg6&imBgTHU-Ex>TC+<<b3_SuWR!H6kH<SA8CzWJwk8+
z@(a9rb=u~y{wyCcBR+oSt7!g}1byj3LGq8P$TTe`Y+cgrbdR>}AwTLb&S%Ku2=hEr
zWYd!Voqo<^`}$PcE|(uY`p~rU?t$Y~RFW@W_A?ha4W!MyV2jknX5zDKSxImUx^cD5
zD~yRnTv&1OGry<`W^Gzx0#EzN^9nrL^U)*rZbGX}hP>!6g|Hhlr1Qr3?aKvZh~3j&
zL_`?A5O3Fe3roua$0`WRzx}Csr%tv4!#VAN8`Mv{U4HtJ?d2%OLN;mS?|wuN4+5hF
z<9Snqi*0v2y3u5ZdR1?H$n-6%E;e-PvY^B1NeL&LC*Sn}a?oL&jIU_!6X_C1J^YGN
z)KoP(j6f9SP-_ncU#a>rAr&u28fKfo6@>keaCv5@FUPk!IfbH2OOu~E^&2}m-mY-^
z;c~ZUS+19Adac;EfQ>2WOk=#fdi+S*x!Q|e3+cUdAp(%Sw;jVfTrU@lk(CW6aFzxD
zJ0nXsO?$sm`;#$Q&LKeD(s~Mcwo45f$7bK&8du6eoUW_w{HB<;fq`l66B_z<dX-E(
zTKZH(R&*qT<qv*pjRu;AI6Ls$nITnxLCTw!%+&ql$<@R>fz_s2?q8Hpo+D~7PbPZg
zxt4T#edrH{x)<(S^dJ7y$jofPgw819cC!|G-}OX>*qy9+`qUK%eC18Ow{XKSEm1I5
z`8jX!(4p^ZYGN~`UkNI7|5xplAHs~%&AaNA?B1`%7Yxu-_$3-@QEu!oP49BUT=4eL
z$}UEZdBLP69mP-W09!m6R!0%!B}5|XmzD0yKl{A1*5ZG5avgRJw?-WE-LF1|#)2|~
zPH{#*vT(+DJ$SKpi;uT=uS4THce<6{`)!At8bAJb#I7I1o|d-!+@m?JpH)xAr0N0b
z#w#!Gl3kWF=d7ag(W~aJD_8a?P_K`TyjY(1;GpN#0(0F`^Bz7&b<fODRGtucTxG+b
zzqy}JFPU|1x}0B4mtQ+){q9<rt@cBHZ%xhFDV?;`YyN8py)E~@@A|>-zs}JmaviQ-
ze3A>=!v&`${w(m)?C|2k&!ex~e+&m@hc*4jpzwd>3r6Qub*i(&HXui^Eis(I>5qJe
z3SzmdD9+01b;Gak%6+S!jihjZt0aymyU```XMpQip642~f1YOi>;L_S{PB{xI_(MT
zkmThE#z2O@Ri;dVF%4SC?eO<Q%((%}cj4ibXT;2#YKJbV>jKZ6TV1)C_wc+*y0dxH
zroe^Zo;R^8U1ujOr`Yk<c>cj3&;Q*B9nax^_kPD?`2WChf1rpaC*y^%ciV>m$xtR-
zY_y?05)3<NA)dDk`s>z5T4+<Djro_|aDH;z2J=b`X*%6%x1sxV>(<NP-y=EaPn?E1
zAU<-YhVz#E|03}Y7sSAE+J}1Mc~QK4GdVg5?5}v?w=+ix@lkOF-f(oT#b2rb?2o@B
z9Hj7p9f4KB=f>Pr<hfUz1Vd)Mjv#7SO-uPrY(p}izWwkaZkK}aHWP&a>e>GRZqelP
z>dV$`*q|dsT*EH@LNE^0$y~C6lz%gI{I`-p{6<|rA^2+Pl`_?i^v%+4%@}^%GvJ62
zox|moS6)r@@|y3z#lg-_ed}wjnOBAI{cCQ=YtH?ka_ce4aB;Db?{UF`=|)Gy4>Zi&
zV7EwcidMYRaT(|<QfQzPk$!VXy}*UP_nh<Z(^%*Zt&MJ5CHT%wlIi&50SR<XFiTkZ
z<Hb{nLO=@m&rkBKcyvmo21xIrL2bjo^GjN;|DKNd+8$Xn-QA+0OsnIAH2!;E;gPA@
zmguxKhjhHyoqsR3tz!Nr2C&sLJbBJg!69zgfByd8npB)2FQAXcuCC)f{<8v>sETH_
z)<|wH*pRF&#Kid>&p^yZGL;F)!eGw+Qw2-uOxcdL>;`hQiKg{dnNKlk_%((P`u84P
zm@GjlQ|8|HgkadO->u_om&|dl1V2fkb9_O$uq;x8JC@}GQ`0(;|GLZ{EF-L<`M(F;
z7u7a}tJF+Nh8lPlM}l-G!>Jt)BkEA*oUJS|Xh?MGX5K=p-@kVrLUnWX%9Vb>Q&!;h
z7A`fpY%2^7$-F^;s^BfyEuQiiQxc~mSFU^|pS$_1KB3x9utqhRYE!fZ4{1B`_X)1J
zxb%F{lfUK9^<G>PNpjBXJ8TpAi~r#l`2XP~X!7_*VaL_o%gwE$xs3YA84$C9nlpQ+
z3zm~G88ECDqNDLT`AsN7u0vY}+rYtb`T$|Igq<3-jLToZnh=F3o5qX!;&Q<sqQ@Uz
zGx%gBfdM!PCG0|X_v@+SJ?5i>VzOAtW8u$5jw=$Heqm3*B#;G+AFBz)n3a`P7Xf!n
zX3QUqSs*`A$6yNr8;DT!CO9zl`C`zGUo+XFimO`d>4r<pSji@ry+e5lc*G$bvQwp=
zExu#t&H+DWDjL+6U(|pUrBjapEB*5e197$6Lg5IE4{V%-!xyqRx)Etui42it-(P8X
znVog0pto?66s_u?UfKgB8ToF86J0k-jT%OFffP)Jz89yuow<W1pHb^;GqSLTGn^XQ
zJmb4F4=75H-@r)&<mG&|gQ)%i@mW9D00{6cKprMEUgc+OyoE>^I0VRJ7BgTNLUuGu
zPH_d*rZXefzySxiYV%{M9U)U+m`Tj%w-tR0@`l2N2{j)+ezf3XVX^q(X<J^pBPoZ|
z0?G+YPV+OPnnU>ve&wAd$Lp9EO*$DQ*d%M^T_H^N`4fzCfwmUg3^yqG?CfmFD*wV?
zjt3*I@3wGo5KNYL|9sN}WjPP1c+A|14G#pIboah1R~-4Q%#-}o*mwheIyP3Z&&{rY
zTOeUm^Y8LjhBH3C=O6mnLsadqd$cXHjIrWz&{kVfx_;d{0e<c&IoifK`S|h23<V~9
zIkReXoU_ubI*UyO^apw$?GIg6;L%S$h5b263lP{R_zn(2OlLYv1qmbp%!d)s`nT&&
zKUw48@E6*e=h;_ghN0WmY9<Kjoa^<c)=?SB1Cu$y)z<f!X*R8psHM=kV|81m_V&@(
z^*xcco81$abGgCbMMX#3)hXr8WXB+GXQHk9UYA2&UcI*bt;v&YE$%mJ9;jtpmOeN)
z$jAYgUq62y4toU%`S^GoIE{lt(w9Sg-zQCS%E6k83tqkyMwZtPPoS?kyL}c9MruIE
zdshLfFw|^xg(c3|=y{@dPkajPj3};;hkOKTXfQz?=d}8!8C3mXd;@nJVegBLx(0#}
zbK!#MlD;_lSDr#tc6e135ViR%{k(vj$^X>?<eYktw?PGS8Qh^<e9KVdrf%Wbddk@p
zynR;4;R)QvHcm8-I$9UO&YDSwx9`)YmX@mYE$5gL5aA&F%FXt_S>V10v(M13PVGO7
z+PO<%qS=*d{=;P6?rd5Jsg!pl1B(L69jf5)9yo|#;~LfI-tkI3VPs<#g+w;iL7PNv
z!j@KRt#GS&1|$&x>iaUAsP5}rFvpdELd6dDYuA3|nIA1sE(-_IiYz`A%iY2atrfME
z2o-t!1IqHuo$$dgLBQ5tvoTdD$4)(B!px8GN0XB>gu705C)aHF1>|d2Y-kw&OWE_y
zC5V?3#vY#>f|3TyA0+4B%H-wHHU<U2STf~Jv=Bt@K&C!3<|mAE$1#DV%;)h5VV0rJ
zql<}&XoVFW*Ry=79T&L1ItDq#O^BI5bn6?Gi4AlUO3d8$?qLv!|MGZ~yvoY!s5yxc
zg0FF12p&B{x<}hGiB6AGMW}?lDxB4}s7k&POueUTfb{WxINBIrG*?}F4PdHnTLB<B
z!l1e7S!PTGoL3creZzw1+^n8=iZ5T4X~W0cGvGB2h>;#a&;=frckbUGJ$CHEZ^|HA
zlJnLui_h)L>%DcYXyzhMN>mljSv)AVEOCV}&@6IHub5Cqa?%7kZ%1wVtKDNa%>#4Q
zN!UzSj~ITXnx7Uly7j|L5Hd^GH}cKv26nt<$z%A#R14p?*T9sGqZbWc>qe(PS{@i|
zlZSBQNfW*@m)j-c>i)w=j)Y9L4bTUa6fZNl4@RD;kAc}krr#-RF5UJpzHUPq(4){F
zXRewq56Eoj8Vkp2e3e2r+)&8IwDjmdU0Lh|vCkL8{Vr7#Rf>|5-nSwg9e~Tnj+;Jt
z@`uv66W>n~7d))3qc01WR^)r{LmrVUCr<@WqaWg-Lj89)MN17e#X;NRgj0C816xqA
zBy+L3Mq|s32}TLa#uqnnAK7J;q0AIBJsk^WaWpt#=x+t}CXAszLZ;nu3M$4=qlL~O
zW3@3;rYybrqEw$5!K<nw*p>fQXYRk3uHv_K1U`HfVLIF5&h6WCQZ1@(h=uybCm1Md
zXtW^*j`X5CL+4RhA2G?QvbCsy@856#^!do*Wx^<OOp|&!*yo*hI4>TuDB9sJDZgic
zXUDa%EppV&4Jskn7?PB?IQ4WHd&73PbyQ>Q_l5`PE|3Es9saPOpgg(%HPA`1G2b~y
zrjw=9Atlc3{uTG-P}H{M_(h|iOqz%o!Te#c5SGL`!qNwNT4Nji>O3!vG@P!mtodrE
zdLQBQvahPRn>ELe^<Rj>ZtH8$%NAcH+;;91n|_qkASrW2x&l$c2;^BuoESP!S68o6
zD0M#lJMa@UHGc!_MS4Ljm;onNP3$*f#6<I?Yfdin%gU^6B1ct0P_It1PzwqDnEb7;
zPXh=soRyCFS!ViPwjIrv5B#?wPM$o8##Mf2ZfWT@n0F4K`l^wmN6$;W4v;>?0sXYs
zTwW2Kd(74_CfB{&c+O)PL_5zQPG%=(1D+(x?(8hJ88gV8=T#LWtx^&adP6fI6APF6
z34*~rt0PW7qa%r~`8+4y1S07S6G8P`*>^gHBPiz>(po3SVcxiL<61W^O~}Pm+hYQA
z)7IA3>k-%e1oat>|Fl#p?pfdRZ{L2yf7N#%Gi&F!1>0Uu5U%4nLOr0rmmf}AnZCHH
z7(Nh@gfPryh~1>=(}f04WU=T0xg1S=a9J?V7`$)MXtZ(oM+}^hwG0}CV0vU#9T!8z
zr4cc%YzfKkC*hdzAng}sQ9n9i$k3rUNm*oVzeo*HUl5|^9GzK5Di?m>UoLl-Fo5~W
zrA8Z4HdLjOg9c$2VR~uROlmuBU^qLJ8T60ZakCMds5b2?Qse3#FEut6I*tzyuV=oh
znb|4sf&KgUhlH#^+JVW;z~RGJ#Vj6H+{_`cJg_T9u+bL$3{1q(TadOy8)Mvl-oa!|
z-dgfCd;678mfWF&^(i7T1crCoBDdBR$&F@Ifv2(T>TDKR+O>gQ^)~@b8vB+ZJ13fl
zBLBPGM1;pgpuwyeh245ga4LQx5P#E(wS|%EM(8NriId5*9{UE6Wll4oKk90*_%0KZ
zz{TSFyZ;g}bnS%}6d%>GS#HhInRz+9N&k*aaCa{O5Y>tke1Q3I*rg7Q6axQ%K`yqz
zXU=FsBf=DVsK(iVt53b>?u62Yy?gS4T7SVSh9}zRTNW2?`09?`HTBxoH*;AcJTZG4
zVFnXkSERO?Z{K^!)LgATV*?4UF3wv7uB{0}BHM~ldkLO+z<;;QoDWu8HLqj8gpeRG
zJZ;?S8=6z5q=-%z+;4q+mbq?&rhtqd_wi6~pm46d76e+}MZ>jeb1Z!H=*iZ(aWC@b
zoCO{99W*%-ArW9<PL+_=Z|*p!CB!tr<cRfK37Z6inR>kp28s@k*U+j`QWxH&k}gmk
z-zzNq%tg|zyp`k`Z4&KgO*}^PY1&=DsQdQmW9s^i_eP3fe0Bb3gmizIUyKc98U6nN
z^%<vy12;p!E;-pFcf#btzrvhzb2M-vdi;FjMRCk_$fOGp6HOGabdPeq2qSYLXvtqm
zO>9>;Gr%*ON6MgE6!|b=4<V1qsJKS>Ze(`sR~nUPQ2p~MIW~QEPWQGZoj@}!F%vb)
zRi5~f7fa2^VId;<gD8ld1=&r^*hxn`<R~W)YGAO0a8xJv*){X>O_B`ZkD=MW=_&kc
zpw$22I_`hbabX4iUyBf~H?LWVn;`a!AFe0?^%ao})Ax4Z9i<S*>%@(jLIB=HK8Tg<
z<;rK3&+gsV`Srb%g5mD%yB8ZTj&Lykrimo+MjL`jZ+8ky*=$I!J<X4&-*F-$6Jr*>
z;)XT6gJ`sl6izaD<&2dxe6|MOn5oq>uvM}7dFIB0nLV2Yz)1M>Kbu$)P>=9Hv>1hO
z*EO@z_}pVk&W;z&H;568x6UgpTpoCgE;9OH2b~ttN;z}p4CN-4(_{4^RT&F&^opM6
zJ)l7s^wc4XmXR*QtqyQ?eyM@Ml^K_-Qy#43?B8GRR@;etb}4q-F{Ui-QkzHmXH&5y
zy62k2e@qPiF16y-A&;4--QuPGx_^D|?K?5pPi$eYrWsO;=@B6S*35<ol-jcXrS>cL
z=Et}zMg81Dg-*@leMwER8*voAeH$tstP%zJOl_^;P8w$%MXwvF-n1!G#(o$3zu0@r
zurAl_TiCT-CLjigfTE}%9V$qpC<ZOv2-1jjmjNiXKvF;i6qH7hPDK$x0qIf@q`R9R
zXFRU6&vibY_uKp7&$ah;?T!56d7k^8_nc#nF(w5j>ulk>;BQc6dSLz_Jd07F!L%kI
zKpGhkJ)ik$4QVhML61jhPhf#>i415!C_qkX&`aguYC+F|v5XW+COJ77R0E=->Fg{*
zfJS$fV``@aC8RBk(;_Th`QC&Y0c9RS_#w#~h|YLZ+$)w<<hB5a4uIGJSvd`j$=$oU
zMYcw8gM`k8inFHG&1Yw5bhHbM_rRX60P}~4v9z@4eS)cq>U08JOVm?L7JK3zIfbgw
zrneo>Rb}0{FxzWj7RZ&RsYc!N<o%ZyUL1I*WJ$mJ-{>i2KlTK#PZfzoza^zYsT}fA
zbYBNUL7{>@NZN}t>|3iW7C0A?Ew$?O<46E{%jQFF!jl4X2N<kVEvaH|5-}Wk$PS)&
zUTdST@wK4$ka7lvCK(Van(j}}naH6jn^%<5!+t&loB_O+s4sKlYym)!`UAL8D%+8f
z1xTnc96XqFr}J#%Wu&S~YTX#8jGK!lJQWk9xn+)%F>hzx%FM55YIcE54jcz0T4{%2
zf)IpOF-+xvU@PWRIGf<8*A@&6t1-k4R$AWGnyn{3sK(}l>Wt!;kn&F^ShocOwMnA_
z{9z4e93Lz0T!X^;%kJ(QFueyO3Dp9l$k_NeY$$Tkc*9X_wwjZ%jKv}(<O(ha02~>4
zc^c6V9UX=EYWNRmBUbvcX;TDM78Q3`u~yk|+@tr5Yu^mY*M6rLw%fR7?EfBcFs>Sh
zw)6oOWRb6&>6Ns@#n&f+p<(=CRW}3GOYAxJt@e(NmGta#MW>v(_~0DuLYUp8ts97(
zI(6BZ-S^xPF#p0w&LwN$o%rId!Bm0&3u*{6fospTK>V`6;>^@a_hd2}*-K%FoC;}q
zKyYxodl&ZLE4jEeFwl^K>aEuTun~kKWTC(;2|!j>F+Hl>K%j)EYyCq(JA#iI?2}~H
z(<^RcUYIbn(G)kH{Q$rM&R{WC2<$IFhS7`3hjO&Cd45YiW(7X2aJWq_%3^4Axu=aA
zimeqP4c5T#zP`k%4X7G0n?`>Xj0(aB3nfgW7Hkc*SFX0bkPN6Spy@iRFx&@~9y&YV
zW;I#Y0~ElRNvPv(0gT|WOKB%EqGp4iS$9JNx3I9*M>DW-kYhrfwt&hSxz!7gP5gh5
zoxqVf#NfKQmgh@fJ{-5J&y-asoR2fTx@((<3wPzBB0)kmAq9`LOF7G<kXXPTs%1jz
z%3hh!h?NJtXe3*Q#SfKpH+VXAozazGp@Z7PrCoT!$Y&?U4K$%#r#bSO)s*B7>BaAt
zT!HN0AxlFV6fS+3OFM<J(yahC@Rdl&4txdy8>$;(@gczy1ZW27GYI_c0BEXKA)Q2i
zrQ$+y{u!oZRS8dHpC=ckc;Ytxz-23o&B#ZeP2T;`0*se@<2t&{JTj(6z$b{ODTBhl
zaV1D|Ggd`!PE*rKV6+$TLe&iQM8QBMRR~7Gl?U@x@}q#|cb9EFmIF2yM^8LExm-n~
zN0pgIRBH%d!-o^SL24fPS+a~Zms^3eRugx%B}lmm(Ps`0xGQUGbXg>1vRi~Cu)GL^
zI5E2_O&cT2MI&fOrv0#~2qpSkZduTuaoccW*I1RW8e_B^)(Yyg6+kf1M|tA+KvZkw
zQ6G<hs))A?$nLwy8WtLwWApp!U?>1dWc4|wdANBJ_9I7oD4<gz+JZ~b;Tz1z+Le0@
zza`*GA?n5kpgbmISSTGxe?iWu(mC<ZB{I39V^Rc2fpAeRO2P(1l_!~H4Tc)w4J`V~
zsR>pi5EmpU!rB--Bz4H^Q6E4vn1SNb=+hM-HNg2n!aJSveLuiW5y-9xxM>#GOl&2*
zrG};^s$I};$H8+0bmTKO3)B7A@A;heo(I?uRD&&waC(3sQAky*Si9U^Ly&)SEInJ9
zxtY}t{A5rHu|Wq<{YLzU4Kn6fwaUh9Pq>3@9umY_CQLB|3@(|@$;4!Yh+3;OJX4Q(
zADCh_fq<h@3%gHPPW`~`0`V8c57byG&#a*z#3l`<zfk|3n(hmV$170|?0V%JSQ7{j
zYQ#wFbMH_RK^Kf?Zz_26qCTpu3esBoKS4?JGh9XS)?OV9t)rj0^`zH!@Jmt62_#S{
zta@rOhY_M1qZQXgZS+4dArB41#!MZa$3RMMxV(Cvg7(Ja$m*&pzgP4av7iN_;DGq@
zXtN1y;1QX=e*2~@c1A#81(}+XhL!@cRDlIxz?iqN7%(9baKG)@TT$QN1-=7vU{}i)
zfF%8ozN2EJ*#yz^Ym}PqK0e?}*f$WuG`<~bk|4>j`bv|-mGknEGa$*q0do%Ci$o9P
zZ1e;UdTb<`tfQ0{uNTUd8t8v|p@DCO2Mgk+ed^ykUq}!4p?U_FSD7`ppr8U`2Yl1P
ztz+j+72v2r_%hgb;Z(R9%%B4R@1d&cZb}A`{l`?NM@I>97fgr;C05@z0smbGvAQLP
z1biMQ0wX>3%lY#szGYICudD>r=Lx4%>3Z3CbkfcqD=<8(v8VG}-(ojr^>NcUu`4_%
zjdjshRc*t5=GHDG47q_0*G90T6$r0^a-8x$nV2FWhzWooAfieGdGNUV%m=JQ!rk>l
ztal=zx4;%h+Xtag7HuEoBo7~6MrQy}XbTphgi7-Ib_R@CMiUD)Apmk?6O-=lZp^I2
z9{{M4^ZB`{@anY**4iNI)Va-i%ntyZo2H!F1e^#|KbqB&!51~(zEOD=ywc}2=3HG{
zgC$-s;#IXP6%rdTYL6m=0SFHr|JtGj5E0xS%qqk~^g@wZtT&Dji^DU}&kr>tY&fiK
zY({Z}+z&X!ju8_eM9-tchLs=#qOFt^8+K}yKQ&;_ULt+u)C>>#>mk5l9JbtDmbf{L
zfyqF^vcJ?1h#{3k2~BpH!pmz5#V~EI#AWn560Eg+(eZsI>{Jh9#RX<0WY~nOxE;Bv
zz?wVJuT$D+ud5?W1$|=Pv#a5tLB@s6C11pI;>5+DzM0d{aagfGXGIx7tnyeb@M-UK
z<=)^u>5Rq^JdO@l?4xND{^3#x_98f_NNV6G+T7d>$SYkrb*}N+aiAR3j_r@3t40*W
zMZ)u_4UKeE&S>vJt)S`I9Ud3wf$swu+)oHR85!Y@)&hMn(n$OY7td8WzKSJAPep}s
z7Y{)g?C$EST^3!zVM3;I1I}~TuYUoWGmlga7D8@r>j0pjh9g|Yf~^X>3Y}+P)RNr>
z{zbHtrQ=YyBMJWb_V%N@vg^l@(&HT~aGYVU;GrXd&;k(a5}XT$J$UlPdJ@(3m$5#H
z_#P)r1DgzVC?g|{0YgEFfP58*0u(%BJrZT#ljna;Pjho%q)qur;jma9{Yo@INThYE
z!Adl&5Tnw`+i{@yzCqukZ?~A06=7VRUUCCzD3Xlyx~l{O9Dk<vT!_cKg@uC8Vy4HD
z;rxIqL(mnmT}&p-G=gI}`~_1#6P!LuQLt;N)<+F)%}>M!MXBt%{FE*>Kc(q~ezNc4
z#CRuEv1bcF%#?PEjf{zQ74P$GNl!_a&mI~`U5cjol<Xg%C0a1X-<tU?GxTie-RhrW
zv%fyQ9Jto?qb(=U=&Lzr9J^3^mO}Td4B6GQgYRr}dJHRWT$g|BO6@RTRVSQ1HPRzt
zvTInkzD(@RX2JtWAMXY45$}-@H%8tYyv6Mq-@1Mj8|6hFS6zRTV`LPY!*3*0d)kD#
z!gYQu{`D5JBVNIGR1-JU$L0&cH6TQ|q<J(^FwhxU3?X}nSS&^9WR80@JbVG(+nnhj
zzCMM<FFroL_UU7cNhw0#1nkEV%&{&`0~lq8F{A&d;4(MeLYah2pL@#%0lxO@Jh<5B
zPMspM1^<Jvmev;-93R)}a0vi>vxfQ^lV`k-zZ(Qqlou%qdKzX*T2GE!N6SRQfd*k1
z8H(h!Yh=NMIX9pqWZlS~$AsEk-P@=D0Og^{iRBMhgw&(Zu*3mRzR>GVfMhvFOkJ<w
zJ$1$9wOzD~v8AQX09TZRz5N2>xR0LnVPf(ufX_*cq|%RK!B)dwmVe5VDB-hrtbP^W
zOnD6${v90kP@>@vk3pb=0zZNh$ur??46qY$Z6-FF?_qeCZwzA$@cjItP1&$?tpcKJ
zUshUbPwT8c0edGVjs@5z9)=UP_K|ovt)ij$xn-fkio!Mf`+j(Xrz?We1LV%pLI2qU
z{^)vy#_(!uAm9kbFW($*q(<hF+>{KKspoJ>%ih}H0RRf=eXSS&jH|LAwOv~<i<VhJ
zpvEEW1u+EM@CMRi{~5KO;|gk8=%Sodc#fb9sQs0|c@T(!9I^#3fa9NReKi7*s{v_l
zz1ao&w`YL!Yy0=?oh$0Y-;icLQj%q`0dzzom|(oR#N(PR5rmK%kq~Cg3W>ZLZqo=P
zh`=na5~zRRSDDD@7AoGDv*nfo2|3_Qv`#%E%q9?6Cch_sgj^{1h7TC(f_UV;URyfO
zrYP57m!b~AEHZJjjB9${zrB!MYI{!$sQmr+-!k!HW$$g&y9#mSAT`~`Ad;e=x(?Kv
zs3380)Ad{<Z+9VF43KSHu<S}=^cww9G!JcmZ{p=$6pP`A`ZlosVk0Q`h%nU8Ck6+Z
z+;b7o6OYD^mPuT`E-Op@>NcVc;XkmqG?PE>5j-_@T_|4wxBM2vmu~9vY@%#hWTfh!
zOTym*w}z#kn3y%ZIv#h|*kDO^-Rf##phC~RyyqUGk_;y1cg8fUTVFo8Rzf#f8^xO4
zchGOtZ`a-9>P|K~=BjDMc4J-bo8{fhY(B4F$ge)ylJ^#U*Zj>mLZ!M|M&LZ7;JG5d
z+hI&Uqt4rG#5X?Kq+43gYIbtVp57!iO!p3@z4~U`Hq^FBw>UodBi*PqlKFFt*u+Nc
ziq-qPd-opS`?2w2;;3S}L?+&0cR_VrkN#R?$u*kIqLD#*(Qgfw%BhXR$GxVeCb(Sn
z+%?w*hK3nR9cG3D#H(L@8uJwV$hzpF4eJ+!f@y#(D8bXUI$(LG4=pLm|1s4MYFTW;
z!rj%?*M53$;V1!GhKk)`EGH0Bm)dl;ne{Mi7gnw`k$oD>VL(0PlCP{dGKJMZrk%ce
zkPT`NJXwzTm6!EUYk*NziUg={7Xx*%$S;6B02kPnZ%S%w|3FoPBa7HRYgIW6@q9%{
z9#L|!c!l1a#ZpJ>7(7{w-o%8rOU{*$p-{`B`K*frmQP!bp3UOw7sDK+0pm>P5M7=?
z{OBv6+0}v1jMDQ=U3CesF~p)kq`>so%P?ClzDpDZ)Rb|R$a4|3eoszv23|8dtATtG
zD-Hf4mn9@xM46E*92Ie%i4)%8-a7y5!*+y96tFO6Y)7MUzGf#zxU3-)tp4-~mQGsg
z>K&b(sBQ@g9AX)#p01Mk3BjB^tJDAG0tl`m6R4a6F0z*le`9WDmU=LenTZLfvKFs|
zDv_~*qr-Dz<X<L`L$(O0hESKRo+^eurg&AXYXtpOsT!pI2s_OXV4ZB&(8Ntuob5{@
z#*^tNdm@g^iO4aBL05BqWoZFW&>-oaZyFF3YLhhg0oY;r)~0_)r;t!rC_bF~60@8%
zk7jrc>U{I3d2vxuL|%xU{K&`0W3J{5U7D1<JVYyxinXWl?GY42F@vpGybo+4_`1}^
zQUe$Pljk-T2e5!cqL*f`pd}C4H4!-QAumi&Bu$~&gjPhW`)fk%3ENW&>2`7WTYY+b
z06l>fuuLR-Gf$6*U<|BAwGf^JXq#w5ICTEF${c27Vgl`)bZqhnpsv2j$uby+p(PJ8
z&&7qK+Z`D2<S!V1I}*Wt#fJol78!vj+O|-hvh+;P`y+R<9}Tsxs7GY=Te1c4bBaT`
z^QI2&b|uuq-n}Qi;8N(S5x;=)5n)SZ1178dNY4Zg%ezh64PVmJRIyU|JmBS<vW1OD
z8+KM}-O5t!Wn67d?e<G@L4qnjUlsF5y<s=GF!YkwgKjC>+En7xyWNd;8MW$r9;SI6
zQW1YDtY&}sxv{aqP|y3#>gVEVKMn{UIu({!_eiKyD2Qa4%X2N=LT~x|<IA<tFL&lm
znm3FKW$S(PyiFM>D?_u@BOMpns=2njTyU0V>uPC~U3PGq*Faviv~yIRiGktS<jL?o
z@iIL8O}Q6R{hrj^W8VIQzOX?^Y<wo}qh$HB#E3N!xlRIYwVH%j@5ouh>tM{(Msm_0
zJ647!68dwif_QeQ88ek$|1*?^*>9}V@|fpAJRk5T&+_yC0~3VZy9wEJ{5gzU!769-
z%GPCryT#Q5pTX9#1VV=7nVMY{n++)+v`>IZBE_{+Q_Wv`2p&5=gXK^weh#vx!s232
zEG(j_1V%0rKjdTb&je8Qdwmi9fpa2NrzC%lK9=QYd>-NedQj*me%;c7<Pk|jQDGsN
zyo|6C5D{?))H-*=9Oc!*%nZR2^49A>y)50x-SU+b3gAJ3)vR4^(eko1wk<TS8T}QA
z1IPsx$#Ym9fUD!j^&-0-unXYUxo)cpGx6RflsQ=Z?WOP5ktHCyaTfm|VKxJd&ayUZ
z^$5>4d~a~dPIp;^e+H}sQc7lKbs+8hWI=D**N23-sE=U>>S`P?ho55#NDFi!Rq<)I
zO#Uk}SaT0dv&&F*p?XBTg(SWjWhY_6Y-fjZ8w|N_WCU2y7J`P;0o>)zYYVCU6*z^%
z+O-W)N!i%j2L`OiOUg)GzTEqWE6<kH1bOc2`aess!vL9PpdtiV52Spd6pC#<C0#Dp
z<>cUtuUHyVTv+J7)M*0CCCJU|(4`CrD8ckgWcG+AK(o)2heyH!=*iYujI>KfIgD!N
zTTKlJL|q7p173Fl8GxFH67WUuNAzC5uI~f~7hl@~ArFjSR0;6Sy^HLA=GUdwbOkVe
z)bu=B5pW|i_xQj>l9lo|pbY+7@9!a*F9%@>&ks;6VZB%I8ixy34Dy2gnSF@aH|y1j
z8JSa4J-B(G+yPF2{R0e%wMZin_J4p2G17=isfBJA=Y^3c)`u&~g(&eJqVW0UIq_W)
zCkyZcpWm;~7h;n9wc{T4Xca_s{zEs?;i8T&3qU2$st?x*A{=hES|FJoRndIHSx*Hy
z`%x8#K^#t)#Gx$n6M(kw{=<ivIl?QdfGmk(iUYtDF|i&@O#ASGP%R@Mw&t7*i-{Q<
z98BIQDbWA86XCC~O})!4UYK%cUWesdo<JpiwR*YvJ#1_d*}04FGUF#n6E<_VkHs**
zu$asWDCR!P+Z4WDklt87d@7z(Ek7k(>#2q*?Pu2W*-h?WO|*P>QmS0|C70rB86<f(
zpvA|=HlkLoOVrS5`dHbarhpv%g11j**F496j+qO;ugEad5O)%PWV>&xXf6Jvw%mWF
zl^HHzS{^EAJX5HNZ>~;SLq`R4CS3kZ&6h8rfz%Po<|4M$jCA0V9J;0Ivu`u82GI6c
zgtT;URT7UA&!rPWvWP{o#X(@9NQco=_V?7fc=4XKO{7HG{%ewwtd++oDT5OdPDC`E
zn=>;d09-K2?(6LRd3*+#E^2yAh}6*1N<SI{#}jmOfsPE0_94-OWSLL?NaL)Aw45A(
zvo|7v;5%rcrUY6L6topvLX2n!I*!drpkOl0pjE>PpcClHd2uNr1ou|N5tY3G&+f{^
z2`xfoRy(c>CgIpu?urcMRK)}4Wc&7UDIYl#6&ouIl{Ua)lm`}eB<OwlmR}EgVk_cY
zNp*J)KpSdihGpXH2C5LW+{G}A=tgWIMBih>!v&~pakwvzHeYkDL=xfSL$}k;@3VPr
z#0XN${dK>g0RW2=Z6e9)8`v4hz%@`mv~I8sqx=vN#URU~_jS8+8#6S{MTBh54K5oP
z=?&(U1@QE>fBp;rU1+Z19O453-aYxag97BwM+g*X%1H2Ln}d~#o(S9P)OB`eHDOA~
zTm8LF(UpSXpMs!12H2x_jvSnOMsya-8aoQ|Qb-}EQCqwy1#|-(24m{Of`>{FeSt3T
zB0q9Mvj+R)JT!QO+VSu4c#oy%>})Hr?t+7ZHzK^i8^>*+)BOemI%-;4v%7c8K|mh-
z3O3*^l!WjI5piF6Bf~6x<Hjn81(K(d01mk?wyMp=<9|SmFBjqW9#1A7g(c<?0Y*3#
zR)Yfu7n0@POL0knrRmF;lrLd27_^T1NO&;Y1OvPhMKHbz;G=d5kK7a@B=Qv?E4KH^
z4tyDHa)F1T!U6skF=)w=3nF)SVAg!gt~q>ZfkKBlEP`84SKzR&uHQ+qW^CUR6C%J9
zFzAecD}IjrL_p)zEiHU+^9mb4d8!!v3yEqE<$<2=`iFf2okFeo&HVk`R9a-+R>JPW
ztIIWOnU4AeJ~1B!iW<a8?%l1Ht{OF2k`ad=#_ro!_^R0ak&sfC94{Z2fL@w%dDHUG
z@_6S$mH_6uF%ju?9yxvE;ppZ5={PU@^4)6ZbDh|NVK2NiKM$#Ss1#=9HX<4#m=8){
z!)ymaC><m23oSm-jNPvXfpIm&Uf>oJ(`Dd+qfI745E4Jvvg2Fv?sD;SWUe@i^7K4Z
z!~084V3r-q<1~8h?*O21zd*P$%aX0$^#jaTK!0?RXL0tSF+kXM0ziU057OHV7xRbz
zG!Pvx0|U|)T^K=G(qa5sS!sagB<glR*H2}b=jZ3wfh(VH_|Ny7SG>QNnmF|G9!^0`
z2}DP&M^cQF(-MQcb}+X{U>QXVIi$VOE*B$M<d|B7(-kPU_wRM=-E&5@RDBZ%Z_f2?
z0$-(t^c>5aXgL6_sk(m12e2W|&v>@qC(4n?dH?-nS{Hob@xH#B_&Z2IvG=hxbeQr4
z(T>3a$FciTmW(MfB;;3LpAF6({cO#Jdh@4l>_R&giVl{kneMK9ugM&?y0VgTXDN+?
znRy6js(|glb;PXVU`JrB3eQJXVHEuS{lzlav4g8xd)#M4;ek`|-1Uzy${zFG{h`iV
zezAyOG<4VD+6rahRb35pZrFSNlv&7(4*J><1`KHLuvm+rJC_d;J$xa&qcbqL6CD@;
zWv7F0Vk(5>ynvI_GVB#iN2);98yXy}>HTyZCOiuxHyg<%<VgJ$Mqdy2<>43wP}N=T
z;kJ6Ogy@~1s4DEKM7MHaS_wT*BtZu^f0BMZq=Yr$RDVAb_b_ym;P-JvDEN>gLz-H4
zmQg6v0ETdkj?#iA!~%z%Ipp>&J6_S@iAfe3m4dXElS7BWuXx(v-_TQ2=ef)q_pu$q
zpomqZVaUi+>U@wm1Ic{~rk(YjLa@FI3J6CA=+^PT4j(zfB0~+uxey{L3N(vm3BS>T
z`4fQ+_7fteS>N|x5vBz2*8b+lC6Lk5)5lTJ(4f6N^$ZlhL8fa^bYmiuYaHF0*Pm_2
z68bVJFsb)0$#HTuFK~kVjMnSJDw-)7^~_;4b4IIA8ct=4S1>HHv9X>Tb!f9tI;12_
zCJb=wdA+Fko11a6y)Qk3)7n(xlMlO2b5nei`DS>wvuwrf(vwS+VGOD{8RydmQ+%ma
zDs<j)oAkD&PIf*rsTvC@tMhPUU23W;jNM}1h>eMec@n38*L=!!v?D|6o$jLBo=5pX
ztEPcVF3&a?JwoRUod%{#vZrFApL!@yp2|L~JNQj`y+L$r+4N>|Ro-y>r?i{tH<Pcv
z)cWUeWQF^@?et>V1>GOcvFrJfI-K}(hY6C2?Q`Vk{Zj!tn>u#${y8U~Uowla(;6cv
z=I2j@)EGjGf$3)98^dU1N>04^2V*^be`Z8+XuMeID8d4%3wRpN;=jxf7A^*I)}5j0
zuxg(D3#}z~Yn?u%S8Q#Cf}n&Bg*mvifrXo?9)iTR=Y>A#ahBm;2O&z}c%ve8aNpev
zXV0eT>ZGfXUkO4AZ}hYeN#6PFGc02`<qycE@bd8yp6sw<gOsfc(FCpqk9>Uzl&CTm
zY<bnURLb`UKJEqly#SKtKMbO40N-Ee&83s?!eAQ$9G}U(ckdnmm>>qxF4!@lONHYa
z6S5B6buBckIanvHK>ZRtR?Hadi;oyNdx36f!7uA}rnVnC`QEVB(@*r}k-H?ec6C9c
zzNTgtNi4}ZVUTUHYd_ch1wcoN3u20fwa2sk^M%vqGFvzY^h$eH>o4CgdJbvC{ob1%
zvsE->JFmk9TJMP`_zsc_N7>oa|5H6pbe(xaO5YcZB7~%OyVZhswz{5%sy+1?a=ll(
z*)bjEHKVDcwY5SZ3nqgF1@TSHHC5I>EUs$L&2^qNh`1o7r`LP3RN8+hEr7z{z(DP{
z20qZ)Ap^yI#0kUKNYx4hh4tm|7Y+=lcUgXjb+@%Gq3Z%{4F%ZaSXfaMEW;GZ=W-vC
zXp}G+M)OJ6hA%~1N7o~{{Et`i{M;PT<>jJ6Gk2B%GFGyYm-=RJ?j0|i6gkq>D^GG<
zbkcD&*7#Bua*o%;QF3B!Xyn4f|7?|{>uaiZ$<ug_y-EFe@8|Zzq$2BvAkoC-%iJR6
z*8D>gAG&u%#p_Hh9WMZ!%+^XrPg`JfP=o&FwKELQM&7+^sDB)1a{T$`<dD^NN6JEa
z?wl2~cWft>QVo<0OpeEI&aMSt+v0a^Em8r}B|z;lR{5Lk^_J){cTwl1+0LIEC3~Dq
z9daC+Y*$5!fQT>s=&_scjcxe!OmOwrK+d_ZB91PnjUF$wUMn>1TD;lP;;0m4n0D+Q
zJ@?(h)sj`8)wn=0m!-bWV_Or%m(klICEIrG;y=D%^`qjATKIe6h-yT1(~%df+%$C5
z)X7C@1en~hT!*p%2;p2jESkx*&y)`+xIcw`MQd57tNVLCczHeAcm(xF-SPMKE^UAl
zGo5e4sG)R-85|Bh-gZS7N$Mg%P0vZ<#o5&xD{se|wj+P4oVo6XJ}eviEO3bRwY9_f
z3=x*k2Z+un%t#_C0fPJA{q+X0rzMy2TZ1aZJ@y7BR%kj#?2>Yv(ndnZK%YM7i!pjx
zV5}&xqCa&&=kK4^K;m)8qV#M6J`X84gMjrdC~#1r?-vym71a^YP`8&!_w7C0c)6DB
z+StG8WiBX3dm3T>_+UqEgD0T~%L@rHMggwE>H_aKUs1yz+S=ouI7a*x9?oUjyYx^D
z$4!esBr}q8y6D+ZVzqxgTyQfC1lvz4*fZ=sD)B<ZSbHQ)UhL)-x5`-s{|h;oO^)ot
zwER0dx>NzTcAxn~TReZleULRAh-<ot0s_k7^0Jy!c?z0AeIe)w<Rg|3iar2r1AZs~
z9%lDM!q^XEkonu2tcLZGr64s?WyN8Z1^ZN~KI9{OcgFR1+cWr|z?Is+q~zpOe{->8
z3)EPQE%lxg{)rPq(kY!K27S(33fVoSbWXRlrCyn@q?cYAd)+Oj<EB43pF2FpR#RU`
z8qE2UUs_3_=n($&`O4&5tB5l%j0X*O=#o)A*Z8RFkk`Cx4`uPHQ@7pihVcUPcIP1Z
z;9d{Qf;Zah!l7YN^NWqy&F{8`%Y=n^T{~%?l-(0<pW+~IKhI*Knx!P$>L7j2?rep~
zLUL_fZEejO*MM+ej#C-^LL138)aCV@K%I1rfS@3&Czthb4%p&mVjrN6n6+b#Da+p2
zPjOUWrYo;9=s>GNtU@A#HRpKav+>cn4AzFY7URQy{leCLbhNlRQyX6X`CPGQ``RBC
zdo{EHMQ_X8GWm+LcImNPVC#3bzMRdoi}B`HzUECRV$e64+d$>rw=y&8er?%Pb_v!V
zEYUKKV=b(fnklDXe@=A80EMAvDuQ|5NC*MGPQZW$;K2Ae{ec5-ao{&r1!Mf6GAZYZ
zTP|Js1hr_=pV<31#bm0+yCEPvpivJ5{N;C-sT4p#M8Xzw?b>+?WB?#BrB*W)OUCMt
z+~3A@e{T2V7asc+%Oggd{SB7uod}m7!gr{Rikk`tRb9ZNv9fuETT`i<@0MWd7tuF>
zH3+<k@5)=k`_G>#SL*h#J`!8D2+SjSqYyocJ>{*Tg7V32`S5C>p_P?5Or-RSE(ueD
zN?e)U4^QV-O9`fpeRJr7Na|{2D~yPYNX!cqBBF!xkCO`X!eG<n=eP|#!DPi8bi4*b
zvQW6K>D*o;qGz~QVTJ7iGaKjl*6>$8_Lid^hQAi}HE;gd<afE#C1`5$E9By8#dL`u
z$5Atr#9<YK<)2-zO#>@8|8C86L$}Df%j~M7ugI5k{AV3g`_GnE%?RoYZM3Yv=8U)8
zbL413;i>0!SK5SHNdG+ZawgrVJ*2-neX}Cn+x~3yxBLvtW9$D}>QHVQ+j{FXJj?rY
z{de^lH*b#6>t77yGZOP3cjy?r)|9r?$td7-bu&A_q|BMux}?IU(o)9Rsz9y$pzWa6
zZnKG&!#N@QHLm2ZZt^ZWO%>VI%NU17hk3Wp_XW@54vK{do*O<lJ-=GoTYf)=x<s>q
zg`Q3J>y7nutMZwj+j=sMq?;o%qu9Kjo!5E$bVVzpqhC)p)|{KJ{oVaL|9LQ%3Y9!g
z{<Z<J5|m!(^%G;Du`t1Vgnl%qqxGex%;GL?OqB8S18wmePC^*6!hoVGs<$fIbRphB
zYCdbvZ$imx(<#h&|8ZLon(Y=Oi>ZafO;56JNbTj)$ho3TE1Ri`()8!dMlH~WsPnQp
z7*v<_fF2F%rmLCkO~n|_q|(Jw(jjZ%_hN>@&+ps=zZ!-Y$RvCh-h@YT>t%^&5>{Qv
zN|@u~Rn*m;AylwgV=&X9YTar`JSEM~e3JCn?YUzy53@G0IM7DY6j0tteuT&HO9(nX
zri<c08qS_oRL>-a5Th$R+qYSFp?72}>s;iWgF>=zN26j_1h_lpch6c~7fpR!`~9ns
z^AV){*bjXAg#duybL8voO$@0=flc70X$!KzBdUBtlgN*YKs<XCb_Bpt&{srCjVX)=
zw?4OpUsXv}M!gOItevmY#?G!e!dN$9G0RWdeec6ZNZa<>h>m`_RLFVsBel(p0-y{a
zgZ?zn&@af;F2a1tmiMC9rOVTl(~xhB3Y~@;6%3y|ItmU&jhqh;CuUx~d82Uj292~#
z`08MENXUs2j`z7oM8bVRJ_VI@!k_^?*G(jJ666FS2F7qr8|XLiVpJ|oTL_wG{^E4N
z(NH6tv($K_<~Vapb(a;@R(Ew*ldE;}hu>X+=}LKWZ|MUByJJ_5h&kvnw{3B#F+Vpo
zVX~`hs_HRvkDdJ(Ly@kT`ZWETCb<{m@bTwSmR>h!`lZGqRC_XWyDW-ZN~4WUf9iKu
zczrad*b<d7_DzsVZjxert(ZEIIT_@7)u>c?R$!A=`V-9sf1NvecUsTQa^IgTGQS!H
z+QJe*nCoAOj%Xc|D{?z<IPAyUEWrocsanpyd7RKW5wfpx_oDkW-HTh^>~$r~G9iuR
zO5~$~j5=(hnf*3t{aK^5DR^wCC_5p!{C*G)cE^rc-eOG$kRnaL_=&@C8Wvt&_>{M1
zNruhTHF7}*w&uW6j`FY|=Qx?jxPDf)K7=j+BeA1FOIV=Bw9uO(Ns-4&$5_9ebc=F6
z!`z!FIMTI6v}7IZ^URu!fvTf{*49OEY9vIa09*r8sVF*!wwcxg@4>YiE~BQjsb$53
zUfapu9(?)uxQf_s8}9L6A8KmMKwLr}wmpdg{KIY{xi525VT%A62YeT#pGeouztARi
zOxXI8Gl1O+V@Y5=s@p+cGBh0Vucn=&$Qajwz1JZjf(fLiKf?o989G2s5o4_!LqF1v
z3WFRI6;U!XqiK-Lo-j735R0iEyQcP6Vn+#E+b;}Bs)S|4S`>tq>Uc?;cOjs1V!lO;
z(=@IK)5!COmT<MqV&IqDit1q(cIJtHw(-nOJ}+v?!X|ZKu5P^d`t}>XJV{L{H>boJ
z<)&UKrY5DI#f8|@n%)ln+*`hyd2KLMrksDnC&_ShLoz-lGtN=l$nZ;3>!l&-hU(7h
zlm*AnpHEWHC|3QJN$C)5>ZgzjyLZtA@ED$3FIr>3RX)+M0VPB>h#%BY2ag;nG^pCU
z`SqqWH%zkuC0ym~XcUj(s4G#O2*088E$+I?TC=3rsL)3sL@R(M`$-L89BPZoQg#}y
zx{^rXN`ZpZWNw5_%gxVlrsUJB4(We<ljgZqS6B1y+`iq6V-sMPMV`f`ZOA5X#+ic0
z6FPb3n->PXYn)8?6%XY<Ho3PKdd-tT)qd;_=EnrlE6K^@GxB=3IQ3IY;OwcBC+#=Y
zLpOKF=E>LzGU~E2m;BNZ&(q|re|w^2D&t$(r-zT;zkiQD6}}<?<@6WnfYJhp4nM%i
z!BxM2uKdjy|I5CQ0|Lf>mK5ByFkU2QAlEqbdR&J+pS+BL+I~g>hCqa2W6b#*rUU}~
z=GOiD)L}u#M85bokh?RIGbj(yiclRsTw6cIG5hO7$g!Bq2x!T(g-8JdqP|(6jYG8R
z=z^R1oZq~Ci|G?la+Ber_Srfm%;e|H+cF5nmCi9wqg|9k^8>4g@}$-a{30Vf)FL^K
znC$CxIor%g)>=hF?vU?9x;x`EwQtIJq-E<1cT2RW68~Km;b3{1m;7y2o%N;f{+7Du
z{Hzo-u*BUl95$hip{;(+Ay-+bZdq6$!vy?s1`zWs)*WIOe)_{|arR*}7U*6>;8sUx
z(-ZQ$2R3+i=v|YMlV>1XMh^fvgwJdz@LlqWt7biXeHGxJZ>>cE$<+Dgp)aMmvt7Ae
z2L&)VFic{z@xo4bk*)iRsPcbL7&iAp#GgpK`-PmQGxt~(^Smvltmox%8uiTDs^m#I
z#yx2Jm7&5Oec&?Wiy%KwNHKK|+V;mib$>K;LiTkgW9+>f`8VYSbk%&WX<)t5+#MHO
zGQ90lb-2`X&tn?vJxr_=Q3%=YkC5s`8IRpxtuN(X2G8}U2hd)yNKr$qqNR$VcrHl#
z`eG|mQbxenBj6Z<%P>*=^B@{&oyg$5Z#{$508vBFZIaK*$_g?`(B|}Y)eQ}ko@*BT
z=dyu$aStDqlYNRx*yMyo%16EHxIg5*g@1!J3fh;K;_y@R8E~z<(bMv$34%Vd{(bx1
ziC>ra?<d^1yDiP{%v1BJ+h4#rBy#9`UINkeZK5%SQC!vueVn@lPcTcTnM)>>-QnfU
zhmWAr`*=nfM4mnL<F#uqW^4B`d>fX$#H&^%)Hc=eHve1K{BNh=YrTQx@itp7Brh?k
zRyH-QR;!57l9dX>mydh{cYQ9P<)-*NNHa8DR=hm08J4P%D)A=6F0TI2c8a(;4yHP-
z<=&DzMpJD3F7X}B!%Xt5>^%Bnrb>(qXQuJ()~eU~F(E0ez#2fLw}f=(tG+wHC$XLQ
zX!Fp+ItKa!1A`~gG?GF_A%IH<m>0uu6MBN5-w`)6rW@xa+S=ifvu!-3^%t>#^bJv_
z|3=~+!~grT*>$~}paerubi+3gC@CT7STR4!G+FpxF2EM>5(<}uiBW=tT5P*O(098=
z2p@36G7xh${d7QgJ+1pfi=(rp$Vk#vUzKT9vK?Lk__Zm;e#Esy2M>Y=mATjfZ8|`H
z>1Eb^@zKNroAmn(Mo-eE%bm10*0!@SpDABS?8BVKHRbU^Jpg|2YvAh(CfpqW_T|_e
zaX6qS65^4t;-<A#)9!B4gF^wED?L>(K`|9p!+;V!(IH2dxX!#EpM^~YEIQik%Q~{8
z-Fg$)UK*fBW%;D++{b{yMrbB3#0T1!Spcp?OE{)R0cz8bIB6}_r0T>wuihO*z*vPc
zqsMi;La>6b=o-082G)|Jcfv5&yPYQtofsf0*7DP$v6DZ*=g;_XS7)<5{4U{IeLgRK
z;!HZymVQ4?;!uS+cMV<R7V{Q9%jgdu(^I3oxXpgo9%65Qrv0+D<BjHJNmpfwCYl$j
zslA=6^`q-uUt6SBCB{l>PVfD0r>6POfmG$=!(X&j)|RWhu#(zm;<UXu!gNdTlqoN&
z$l_Qz(4WWO)Ja<zb^C2zF~v8j^n2h|$;6yz<>CSt!6qgW)JFP3*SCM4Javli0vu-v
zvksjP0ufTlDqm%Sn@TMtYQeS@G1_~Z7A7QDo33tWXQWC;QF7dKC9$yT{n#AhYM3q?
zUfM!Exdyd)au9RIU)<7~S0oN`O&9<}+twKpRs(d2K4*-Lk+Bc!rpYn0^(cmL<@gx*
z1jjyQUJi>y%WanD=Ya6}N9byO<<f}Ytme=%yICwym61eSN1;8Y(ulTO#d42ZrB{1K
zBSZbzljvvyQS7V8Sxx&SF1_Ft5k38FJ&l3^xP(@o`67ULy%I;+r{8P=N}%f%Jxc4t
z9S@8V7k*Rz7~A$;l*nFVzJ|E&o{oVu6p;D~m;Q`3dQ`S~+cbonL0QdDeWijFTX}rL
z^w+OTOQl{z-)MD)FAeUZME@zmcK{a9Js1&q8PzER1JtUKvN|T3ua1d8Ju2n8D0=SP
zJQ|Hn)wR#Z(W$!ChES5Ndp0qWU>|W=nas?XhjN~n`6wB!S8eIh9aAC9WN4<Z(JZ;H
zzRAAHZN?*)thi%!XT(rZaoeOhW1Y@x#a4$*_Q}gniboXvztMzq<y4(?f95XND{lFv
zLW)1acFyj(2siiY-+zDlJwWtg)-K@l9}W59@*ogyE32Gtz@*$wpFi%|oARg2e`UY_
z+ZlUtmhWQcyTTj0=S{IvKPA^D8G5aLWveS`dwGqrFaLa*GA-B>A@#G3n%;_ZXt%Om
z`m9a7OizXf)Hk!K2x2$(Zf(Zf{|2w-c6>Vv3jc6^&xOWoYx7>>5XV4IjCNxbGwJ(N
z5clkN>-HD1iIY}dVn#74i^T~>7y}%w+8<W_8N&7J7rNi*KWqz5MuDGQz9v<iywHaY
zBbB@cu3)m)<(D}|0mO2cuNQ7|v{|`2GJw)9NvoFO+t8`a4YvKOoM*&HE4TZSmLTV`
z@b{oU38PB&G}P>|=uqA$IB?A~IX&bP6ix|Sqf+ie9ysS=jD-t3NJ@YIi!d^A9m|DL
zIIXAss8>74M?N;!5nc0CxAychfFF#|d47hL`oyLxOjr0f{`1-sd6k~$s?i<>Vvx7l
z(^m-z7u{!e6Sn1U9xWLUcj0c0lF+vTt0ZgdK8UMp=z$=|W0uoC!NS22{(R9su9HH)
z>%&9n>lC6c$`l*-Bhpt!J$>3%)H5L5!ZZ0R5f!avw|TV!1{j?@c`_-EFgsFxi5J|V
z(3^LO*hOh1>9}Y_@tRJP^SO#huhn|_Nl~^vyXbeZlTG^QF+_-FaOBZHwAi^k!`>Rw
zQQc|cJm1%}6x{8_VOhF1Vy^o9z@8sguYS6(OeCt_ek~Md_?p*0Z|-iM(U(Z8-j#gU
z2!q<xs6G8Qb5Izio4j!Y(~2NP+La&|ZKC1<n9=?NlhB2n6-<r|;wJjNQ~G)^`zgEZ
z(5;W?w?{irC;D1H;XiZcTEmWRG*N0uo{=gsPMO$El@n`tInq+SObr)A-lo0>dtNhu
zYjaq*Hk=O&$m}-o9Gvi<4qZUn&{^i@G@1wr+G%0cbBPG4H?nQ<k3uXgra0a-N6SFw
zKwY`m<2LQ$=7#!Qhu<XVDyVwL59UZ_lP|za2Wi&Y%8EN0)JQ%1X%C}+iJSwDBYAh`
z)<OJSTU|Bd!voY2;yejU>i75lsx(N3k_4>;>DS)Q&d!gcPDGnMVv88V3D{JTc%NzB
zFzC3rZcL30oM^z(IoP%duGs<%zaz3dZJmf(r(~wrsn=uq@n42&sRWO{<Dwexqi4^K
zj(!rgG02M|6l&rtzqa>PpywW;I}_D~W*kD+`2v_}JM8$6_qlA;mxlsw$jOnw`&|&f
z?blG6J@+f|D#c_x%g_n(BMa5+php|5T$zKk=xT%VOE8*tH!#&16Vn(OB`)9!b7Un>
zwt;msS|xI59|PW9HK*|D#IHUOh~i+uZ#~4x!SN&CGWO2YFT}p5!kHbQHzoyv2|z1g
zonZf;2_VE3>r*A*G$Vj|fT~uk8k62?KC01>vXP{S@F(+>=aTFDD_5@)`@eG53s(ii
zRGQlytwF8($I`y>IDgkX^<wtJuH*39-?N2Y+8Q#0q>rTt9<zR)&}vrwbJWnQaRv4l
z^dqc8GyaqewNh*I_GZ5~rsk`Z${twkv^rj1rfD>vwSzS%qm;ueddD^4Y9B^2pL;L!
zk{4%&K3*X2Iw4$A$D;2=M!(p_l6X2jEUx)M$PN#>g(r7cl_Yqhc#`u!*SKF*UDgwQ
zIZvL+mL7RHGDwsz*h2JfZ^^nNd2eH)>aN@I(!2w~&1kLKpqm*YCn+Z-<qF=i&sOa|
z4irKLAVC(LgiLP(O-iX#+Q3MT2s_?+`*XkE^-}oSu%6?Eb4)4_{x!1Kgg;Q?UM~y<
z5{V<&sLelsGq2;Mg6xCi=MWeTsAP+8KKTa+FS4%b=GVbe$(Wur(p#~GxBpZzl?3_@
z(Od@ePc?H*+$RiP@7@6pfUic|GGqHTqgvsYmYME_Ru1Q&SQTkD*egS)WU24+B-y^T
zJj-rJ^s(Tj7NEKgE<GBrbmvWf4Gk?pHc{ea#AcVSoCXrf=w9`wQg{e>?@!_+KEiO}
z{`Os6-QD2B^kNDJ5^(U>SXonL=3bH^7J*lZfF^`r>*?u%KLau@kh1aUx&i=iW<YBO
zVdwzjY09E4A@D;Aa{lg*4mgTKkddQO3_len*((M_rj0&1$OPzLow*Bhz_n^|5-(cv
zm<%ug67=cQU)t26nEl7OQW{Iji0q@;lgzbtVvd2`KFn7(^>~d#Az=FhZo>W?N)_cp
zWwKIIW6;on9>h^LcA^90#&#WdKk;oQ2G%86tb`XciA2I>sii7Isym5njKhN+?PVl)
znFj4`1Y0jlTVwuZV@Ikk_u#KS^xuyN+9r`&ocm#Qe@w)b*<}-F683gP=6h#MC$#zN
zy8)nzKsr7)U~+5iaA+9;F_W(VlBy&n8{o=kWMgvCK#Y)bKtTqJWumtsKeGn=iAfN_
zo8`q@%`Z&;x)k>8k_8)@KG&cp)r|S+?|~c^M(UrR3qg%t3VO|G$Zgn5q9139wm+B(
zK;aF^?rN&5bEv;b7!c)`LE8eaWsA$rgW4MBF<59VnOf~1J}}~2FfmEy$4i}U{gsF?
z2}xt5oHD^?Ky+#c_yCU>uwSnCPDw?u!Rtl!{;`;k$p82)yH0B}H)7F}?b$PnVf2fO
z9w=dfoF>O1v!*JiJx{*p(AObaYkT_%gNtJ~c2XMkSAHaG5|F#NcyHuwqf`73kC^qA
z?wam*Qae~uZ5T%-=;Qe{gTO_Z>Q0#c`P#{j%yzt_HQT@<CZn@0vTeS&A-2nAOmEby
zQ?qleXth(lhDzLhZsW$42TR&529`nHLd8|4R<F0BO)QP{`O527W-ULzHcW<+geHwl
zhrP&EdDi;PlT>4+eZ1mwx&*`b&>;{y@~wKg+&8JIs3yAd*jZSvKb_fX)3U1Zq{fi|
zAXLjvn2QV9oo9dcFU*R=`GnH}tU<(9MK5_m3G!DzNGC}}B0;t*Gku7YfRt@PA_4c{
zyVR4A=g$eckA7QdP*BjskHcZah=7oZ)i|e~)w!?SRWvg|d(m}T!!d&}+B!p)nC#w;
zL6)d!wxVU$(H=%ZgH%N&hUX+7p99Fw1S*xCP0TM$UNr!OeOfzDT2_|t&Acl)Z0gbn
zv6XxWD0m5A1o)QJ%zN~MYjoc$rz)e#21-~Xb>KLj6Q%*|dETgwBD*|G!x?8XaA?Fe
zf@TQZ=PUn_GfPOvkf35eX%|{wYH)T%LR`Gl8bLpb)TETEj2l5nqUz(qg5to&lxZLS
zleT~VuE6*ttOLpvd~F8^Z8~phgMicHrPs)`6Qad7b4WA;U_}+`!J>t(2{I_EN)<3z
zcJJHA049x$OyJ6GnDmz*lf>l_M~Hw*+3HZN=1nzD5NFzde5tP|b@!5(Nf9av3U9FE
z5)u-K46@@~Hs~F1>g2^Y?O}qkUGFFGZS@MzGI&Mf8&rP$m~6oS_7@*`aD8WT@mmFj
z)PvtgX3=1>xNB%);xyVch>cIxa9v79<{9s8!oUo;2l-JE^brMwg~3z+r~Kx|RTE=l
zNL@?u(GmxJL0S2tdvy)5(I?Ot;g(|e?iG-XLc_wUx>H)m5D`7rCkxPkSXy50h3MU0
zd@ngeQm=UOY?noL#4qU{%<jaa$?@Al!3oi{HGmMEOIhIRe$7(3P(&QxYC*q{E@FNq
z36(j>MO^c@o~e)SqJ-4W?9slnO4re>Aa*u<89JXdVLjR>$MLT2ldZ3;Jc$`JOLiX6
z1fgTx`ItzZ@lESQ<(0@DK015AhV@0lb>{el2p<)l_F~R5&AlHA#lvJHrM?`^-zw6{
zFAJ4T_HCsrSN+Y$c*G=)rr&UQK{KeG&h308#phJzv^SUauX3ukI^Ucz=+E-&beTIe
zAe!#XpWRrRtoF4_#X>Fo)-!+SXI#+9V=fFPi6ER7fZ~h7tQ8ptsFY>1pXnu{GeN)@
zDZ2L#u%>?Q0cvVkUV&!_w1H5}6c-f{?maN#1W_}m!nO*~0jyw8)J*~&XaiCS&jJvm
zspgm;4=dBNpjJPBJ{PQLWb#;;<oosk2_Y5%J`?Z8{H%%~25lKQo59OXxv>6OSlGGS
zpUFXM0s{!_?s8<;z&x-uJfW^94j8528n(@ne4OHp?F1wN&3I#>&27Abfa)Q+2jT+G
z@65u&QG72*c(T9kJrN;AkcKnK518j^rpQ?wL?&n1@qNTq2>o;D(M_9@{B8!1b-p8X
zpVaBM=ouKeVzp(&acl;Qt$jJV$?Jzc3Cz>f8+PI&4?p<odj!&Mk`GY;)nL8}%nWEF
zhD(q=V+-ZxtJ*(bL7kAN2@P(h;jY?;xR`hd)>n9yl$94i>PDo)khwGv*7oe%2jX5j
zIHgFoJ}EsU4kC<--_$#XU3)lEZ41>qx<QL`b9-^`I&+fK(=D)oKo*C3UunWsB|?f1
zI=i|cII2Rw04NoAO;=!kj7@U$ASJ_IGBPi$cyK~6l{gz8`PW~6LB+9q@7}qkCB#w^
z<ZdrJztz{@MHDt>%SuyBMZ=&Q4C%T_Tna?WKdP#dk$&LTLztF@Xb09vbbNd-EGo8N
zp2V|>(f^FB9i1WzOwy*Jf^zp<=7B~?q~WA-qGT6bD)IgIv~LHBLt1U{YQvF*<N{Dv
z%uQKzW091Iqxy$Jo=QukT^YTGG``={Ahq~r?7#2>ig&4yzSBHBJSR^|SXkaCN;(iz
zLuAjH5IRphyAWe%4po$a7?fxH4_CA|>BB?N<dE71Y5oUH-`a|;&1wJ1??*w*UWTrQ
zUL|*#ASO&GQ1Ljk9CC{9(Q(UJ>5M68va0h}Tnpi;<&N_5E22@FFn*$Tsz8UOb!794
z-nG50Rq?;egEsl+(h~Rd=V`mx8)<(kyw-V@TRFI#`dIu?y__?A>?p!uC0U4}t93iV
zaK6VON`Q0FAFc>iEfH_uo&M0<yM@6vhq)gX!#6-i<{Pr2^X=b&wH-fjK>xb30U~mx
z%ZoQ}W^fA#6qoBor15iQh=`y(LXAax=#U*|vg%(_KrXm7y+yL>$~9x^dB;dhrmyhw
z>Z^dc1Y#xnry%L&81rAjg9Y!LBlCBv^k^?1T3R@SK>E}zm4z({s40Yt3*7q3`zOc7
zFp<c7N-~PBDX<haaA*5SP*Y3g?@R&(!KD5fVfQ#OC(4OUja`6oOBJCs3t)v1#wGxc
zYq=~!v8ssBh}=^u%(T-QJ+Mg#J)dU`XcAt#=%T?(g}P)AVG%Fl&~jFW9Ka0l^A>5j
zB7i~~P$Tp3#Kci*YD}!kHSjtvBp~1p`X-7&z{IMrtmx2sYmKo*jkz!sYsJ?2$o2)W
zbU^fEhdK6p(ZgzuP)h^A2M?7>%@-BIERnqf8#54R#R&)cvV_>*5mFK1;W(FS3JV45
zsHtRMF>rHd8@LHTLo_S%3=J?Gw7X4obrw(RKKbww@BrU@0LoQPU1x&U4p-bxTHaR%
zCME?S=*+ryVC$iGf)kb~I%6V169+l^<IHp`()ldiGR^rjvT|~eo-M?CkyxK^h+(<o
z2951G3!4-M)Ft>L{Z}H9PY`ys#>%Ja*%5EWVcN5OU6l?Is)^MW1soLb*YZ1T-h!2m
z5tL`NekrEmu@D{zg<Vho1FAdpF1#l7N4G<ctC_n3I{`G5tfsnV>5S<_Kvo}w=o41Y
zoQ?{(hh;be54#OE5#S!+Ks=`#;v9<TR^YKWWyq@?#DSDMG!R2!!k?ld6xI0YOm%RV
z7~83W($ka?DoI?fUd!v%Q(veA4sSf0nHOK)RBB-7`!w~nM7(43{&TulK9WA^w^g|}
zJQA}g$ZeQsW#TzKJj>S78Q)e&IP^WK_=kLTImEAo`ITYx9hc}m2ZEbZLF+@_gU$WD
z*tiM)5J0Bl1#Dh)2;4-C4qUp&W%yX;vpO6uKX4S`geBi#9T^{Y1Coe-O_Z3%9P))y
z)5aH&{sShsaV-*4ju7rph801ec78k(9!H&6j>8B8l<?Q3Q-ndBBl<V6C-^M)oqe+e
z9YfeBI3te&z?Ft<p~)~8)d-M-ZhRzKq|%N?g?}eqYs4dl^M}j|l~w8N)f&|!f`8m5
ziGh2FMB`w_+kY7x8R<X@kG>Ej^j65%hGH)yRwIqaL!&-$0288icRnOYW;l!R7t{E!
z`1-^-QH}RWU(12PTediOzC|OJn0?L5I|~B}dek_O?A<x?74CdEL<p)nZW0Ce&b4~s
zX>ItUBAj}yclYMj9#hxU+__^1C#)H8^x*08Sr@-Zb!4G>uphLdXv6zjh#sKYx<c^+
zuRY+wnQpL?hH{Y@+ktGa4sJCwdS_pq<Xk13Mtp8R7`%oo7IokjGBo6@P&g{B8G%MU
zi%M$?uqN_P7}1=wE<o$o55xPAORMg@fawX?@qqS0hwBa5RM*mipaT|=Nk|i}dRUZ&
zKYWBsfD$hq`zxme&JBvV`|X)pPT)K4m<zDz>E|wAk+8TK3rA@*Pw}xtHCDs*sO2`c
z4Pq>j&{a|9=I2l0TPZ8IpsEen48?S9EmTg}W+48f1R%@Ira|aKtRS)%gyM40kM<Z_
zMg^WPX52i?mW;+_6o<Si7!fdt$QHa33lwlu0x~CJ>I0F*;^MFdVqU!a{5C)TIfSBc
zTfu2d4D99njb1QpH3bl}fDjNihHwWjmSb;v_dLzs$Dr<la|$deaF)S5nNZ1_(HX(;
z0B;BHX}x0mn)er~yoJyA{we#g6o~Q^3X)`|dobI=%6W&B29JhJdMM)4uf#i?g2jd_
z!5=Amg*ION@(QKWxn9gO{ww3d7Q?d`%LnA!haw}Q_=bjgo0nt*<p1GT+$tDzy>#=^
zLfy!+WBKyveR(6AVLJ9TRr~gtEDQOpvgjr8J%{Y(?3Q$Y3eT+HPd8AFAZaw&HEq3J
zKCZ*>s9|SY*ZJvP(T*Zbdc$F(;}vf1qi>UsU9dk&!#lO{aZWMqX03#LTM|z2-7Ydo
zAkUB*yhzO=WoyBRihCsOa6p|}FC!+ZC3>r(_+9w+1&%NQeya^vWWBt+Mh9oHL<m`E
zOg9WY4}AV$o+kt|cgR#RIIg{2%_~M%1H}Z0Jng7CP_iS9VIrk&jWnKORLjk3XNio7
zSp!%yF>!}6UJgD?fiXs$@DaWa8nC>)*W7>~s3~D#N(j(N6_0bD&QNlpjrk6D6dg(k
z3eXM>*xHB3CJqtmQ2s(y%(TuST}cof;3tGeE9#@DD7CkT`ae|@+I5Qf-q*d*DCVZ^
zB{D*JGi93#f1~JcZr%cp->Ll$*37)6Nv;T{C|D5U!QjWI6S|$26#ZL;XAn_DbxJ~S
zLJo6&gjI~~Z-jWf{@j2c3}DxQE}*_ZnuAOmCapi;#~;F3)*=;ca{IR3)DJ<kk=LL0
z?%fMV*LWOe5fLDe4*wzMYpl`w;OqyPjg^%Z_Y2(*%pv?t(&g*hhw(jmS6h*4ad)03
z+XOm<iU_l&;Fpa<e{6En^2-&u%MkA?#t0LW_+0lbqF#kvkl*hjz`lUbc#_<`&0uY0
z0OOS=N3?u9j|bHPB6o+>08=qkRU<(@23gtf<Btc(WuR8m($H94TT@M6>g_TCB#C<V
z`wwKgVWhPt!WRy^9PM{TGJQ~+yj8-oyM8gk&IpJqV31R%KKAqw`7u%+t|e&_fTEVu
z-E2$Vp)oOREtlYpP9;wrE+r@D33&$mlmObnE#cFd0D|F7cN<4ouGl*EYd~NiN^Kz6
zG&Jd$zK-<c60Gv@EO562GQYnF82WmKlM_@ixWl(%p>2`fjJBNF4}pV2F~&lNKi|tn
zz>95-xahFwWJhp35&fk6^_7()M@Xuv%4j;^USrXMx#I*SjFS_Mm+_*~otY3MyO;sF
zFscN#4N4P8HjX&Ufzd#dqh;O#im#YhdB@92Yn;reb>aAqfei4Hgowk{**V;H&dtm*
zrQ;VPZU-M9Ve*RP#}UH-=H|}U>=5bv*Z``GS6STLIcLzmzu?qKo{Wf9SvsoompR?G
zy<f^=*R3-hKfDMD2?2g4+kUy$K}1f!*tdnHK{w8667Usb^a1}*02c9#aR5rcy>dq(
zRT=A}z4pd=C^kY>J~|r_GrhmeBsbzI;B1bl`E-K`V3-|5x&4vA_Yk|-d&A~vgVPIj
z<C?#Jy`m;EHTVu0k?f@N2T&)d>!<w+dR#5mm$rtK3f{aCfTW92g^%&p%n6qC58G!2
z64aR<SHIPzPC0_I5fdY*Y`)?UJasaViz=w;d+og*R1!vetIB0^NxrxN`zR^BMys0v
z?ZayB;1&ecj@ZUVt+d;+ALE>~Fgo9j>DsL0la(x}^3nx{dbdbU4{Sag%p`pH^_O$a
z{^0FwfsnmfaNg2@p#M2*7EdPT4M1$Rdn+$LpTO>On^F?hiN?bgYJUaxJuT9s*p8KJ
z9zpO&I--92?`=|-*#o>hJ<GHH^MU6PI1`WOAEDq;V&B|N{^8f9haa}jgfcShQa*y(
zL{mip8eADIb_!Q;l&5OKwhoi84TQ;S=HJoq-AsLq(rjzxH)UT1P@MSlW{R{+*LXDp
zVzB;;$|)|=f~ttGMpZzVtY{yhF;1qw-i121hV52q?GX`E9_MYkV#!i9doJpuS58=m
zH29_Okb1ZJUoHS3EZU{vh?5M8&|E*xwPCV`vMw-iE8H7a1_bsoCn*=9rmlVjN>s61
z=qF3X8`EziC&uI8Z1&UA^_lLml@M`uo-SLKQc_)vsQUN%=7{S%jYcu78MwxEOJ5M<
zf(`pp;air<=1{JPy>cpJ6^g+5<z?<}63fThT7TtB+iDdM09OyQu_Hl>1<pH^Pm+rH
z?(<>$y(sEgfo(k3kz@fjRc|Xx>44s-K2UDk`}9*REGZc!3VpoJ+h%f3GA!%ik6;#&
zH{=MJaBcoMp8eBw+Y-oa=+Yje_y>Y;p4y5QgFS>l{U&{FCw>Yt-2VDKll9lO+WoH+
z{aN5moKXU|R9y#dFD{us#;My9Hck+(%ft`MBj^7;buQs(|9^+q**^a7$hulw`v3eP
zaR>hI+mj)^@dY|Mob}OfFm&$y4;-9^-DG?B3S`0sIu$8eO+L-w;lbnq)*rM9?rhh&
zdDeK%e<IDY@|D9BB;tqr9hyv{ycgQw1^(yf|NU0KZL^C983~k9>TwK*+U5^#%@KNf
z_PevtS<KF|rjzi?i)r9VxK8CMq__GL!=|<+quF>4$G?p<95xWtzNQG{w>)&?|Mk-y
zM7Y3j@DqOdze5QX_F<*|`MD_g3AW*Xe*WhV|L6AnyAu9?vMAJ}9Ed12BMy`eFh|mO
z4}@T*c~`Do8y*@O>+Az;ji!T1Og&*+-@?&i?x3+um`HCe)|`pwfKQgPn;qa`IXP(6
zcv|=2pA@7rFcfziiyr2u+^sBw{3PTSF(9r;f_U@^?gW5?$Sj+8q50}E-bumr&+e!_
zrnI9G__Eu=h<>OJ-sipYpHLbujDs!r%^Q|YPl8!|`?k{BCSLX4`+xr{(mBHaVRZSU
zfeP&7Z66as`J#*RnnKLKBJvwT#MV4?l8D9%3>3g9YirZayB&eZCz<IXoVV}UK4PT%
z^BPnGXnjN&Q*!*7Js1D)M<^xlN#Vo6in@w<0)uO|$@u*F3_(5&90Fh&D(VQkqF0mu
z_dgMS6aXC}C*I;x&smz_Rkzswz0T}VPGW#wCOpn+go&>f_UA-V@PwW{0nNJ4V*XON
zQQHP-iTH7E?kl<sX!%|hg-Nw7FfrpXOqkjx75wKr6d6w#7JwM&>gE=kdJey55&d%%
z4QgJ$PRO78y5Ff2v)Z1W`g8eNF!wuBPHI6m1j|a}$v^Mp_wOAuSqU#dMFGA>Ox~Y&
zV*2-;bYPKzI}8HJzjtNp-@D?3qJx;H=SYl%Kp%|t`*D1H_|89Hdhh!XKP!}Y_7E^8
z{CP8qKW|2Tkx=_yxSQb>i|I)$%{YGl`x1X*=85Z+!@nd8gE*{!i+%@)u$poN-W_K8
z`ibK`0Bhjk%hLJhqsae!lz$>}7Zl=64%`t$IA+`X?<e_YCB*)_04o2y7m?B7y?BUK
zX=KD?dHm1w|1E@D?EgWt!2^gw8HQII5KHgkTX;Kpq63!snQ&JG+ev1Pcv7r1G*JKr
zMoYCayMY-~Pn^Lo@iTLF_ntkChwN_a5PyZA|NF%IbGwLN{=c76|C2fVe^0E&bBMu2
zDbAo6p}UYgi3Wz8yu4buc*+P`3%_8E(z=2=9%LF*J|G2f^O9QLkEu_<(X&VqYVHUB
z+Ln?)JS$zsQ&$5qBcr_{NHen@6YAh&SznLh@dIi@fQ!WJbs0)_f7DB8^GJ1LAct`%
z%SBUDVhGHH4&X3!j9cf`f!V;x<a#%{)lkJq_g;AV5soJ)q3}iE<kjh{4Pm-H<<?qz
zg>IemGYxS+L^YZJ^9@G23$ot^yTWA-xG=b#=v#-z#aaII53K=A2_Z7j3_^Jc1lBTo
zp*jOl8@l}*ZffZE0Ex_0O?MD&>*>+u*R_Leq~C_+5s<xhP=?Ib0hXeq51%IH)IHEF
z08Sa@FpX>`86fYN;4C#ey49e{7`3kA7D|=aBfy>=g~Q=j!>ucLj|(*<kSi8S4<N8L
z&#S{6v-29TlG@WA4+TGc{w&p-16*S?60pV-4=>!CSMCM>i@o;_$GZRjhr3GAKugP>
zEtDCBs3b|q-X&zDVQ*K9gskipviII2Bb)4qY|1Djd)<%s_5J?-xc|M6<NojG_#DTl
zx{C9BpRe(JKGsuO2kmucT0F4mYLTwKh#UCHIv@mX_hffqmO(jD?0<)@<G8lHTo<iP
zXH&%0f!<(2q7;jdCmzWCP{iCUqPa@8o4buD2q;k=t1c78-x<bKj$J$MXQSyCI(IIS
z`0pX`AVAInTurNs_C*`0S@aO6O^_v`T*{+I%;3os96vLTT`nj=+(@11@+fHp7&aiQ
zh(tdFAP}BF|Nc@O(m-X16rRLaUlEtU2W$L=vW?8-D?4K_KD;UH3i<^(EgIyr;hXC9
z@5dR>OB_D&xI|bt-~9RcGap<S&|j%MbhOZTJkOXDkdbM@b3pH2SX6Z7znKtFydc*X
z>JD~vbCB2K#fyuJ$J$b{$^v>PSkyv7VAF{d4ZKtU|8AJdcK0rS-4+C0J&B6vv5HAa
z3G25e*bK-EVr}cdLJ$rV{GW8@rL$Oy_@;0U<2ym)gRugi;^N|BV{t?Ab8s|`$|!>a
zf_@iB4PZ|qu4SNX;M>`|;Oo%0L2RFzl41hj26G8O0hya1#bFHZk<Z|3Iix5Y95Y}t
z@Z+&p_$E&_Y#(~U2HO^(7BH*NkkI!0c?O29<m5QuL`AYM(z^+$27%4cDP;bR71Pkv
zL?#*%)Cq|=aAOSVgGjHcstSmB@WU-bG=Z2>Kroo#L$M0*uAuAaA8pyWLv;1O&(H>W
zQ%>9K1El1q5VOKs(v3j!B{|>&&{QvPY!Y8U0{>nWx3pZs9RfCqkO061@M{mi3KOVr
z#&dC7n(xA^?J2khlRI8aVR#<45)v>le83IkjHdShwFVA2fTfDWs=s9<nm?>ZJiz(J
z0|mo%IJ^W34qU4_)^m)fl0u4_1@KrbNA@wt>~G({!yL5+Bp&3c2gZz<@5)hg0I-f%
ziH{i=X3RHQc=$+2Nu|Dnb4hW{xA@&>12h_B>XCsf246qoZ|>g?1FiOVhc&<q&|(Vb
z`~m{Hu{WI$oQzT&9|moS;7LGQhc5vfa(?rs1S&stzlZZcvu%JCfQHo$zVh&ZhrP_j
zBtY~q88T)D?+d9~!~WoWia<E9I!58BL&lNB%2RA@?1>ngp+b=@NO@m=`}PemIg!GU
z$v6TV5Gn?PCs;jdmZw1-`f9qZ5uxixj~YBVZfE{$cxZ?VQy#%v5JPmQt|xk9U>7y@
zI)WYW5%hS8&frJ{nvi|!&Gin@$~f78nHqmfkL=<{IC2o5v45A`z8&QaGBPur|F4s(
zK71Y#J_%EdKrk~YPm0**PcK0t-f8xvlNE3oFygin;7?g3cJfJ`h%E7$(E^wZ2-aZm
zF2uluM6+iT_3ND_MjUEbqR?h(|NXOydUx%*AfUQEOX;}k)jTbpKlkoe=^nT-U9$WG
z$x!^G37tk3U|8|^bst*p*-BA3he>)_uqk|6E`_ddI1DN*#X@O}?!6r|*<Z9wtd+lJ
zu*+X7kHPJLQ@bl`4sV8<*WlyCSl|-bah#izwlCDcXE3cFMQ_^P%YFHBA&DjxocZF#
zc@WclxPKb|&JhxfPie&Y&-yzM8Bgi-n-T%+>gwO|cA;PFE004464xjDci~p)kU~Ij
z0z)KDaewmtJ9o%F$4HBxg3EH+C<w!_=kTTXrx1f0_C^5e7zASo>YZ;$BnQ^2#;_~U
zyf{7WN)CEKi6CpQFxzMe)U+^{AU|!#W=PJ{0Y?;uH)m91?jbA)MlGYoXU(C4Uw%s{
zVGHMC8R6{Hi4)liAsUq2BiKhBK<;#9FhwK65y=>t56E^sJc1wDUt*Ux-6tsqVH0jd
zJhRTEZm=DQp!oZEdSv7XEv@Es)uo4LpS=!{!HGmnRl)PFFJ#`9gG3kK7in9HG#%|N
zm6gm$bc1G2?Z`Ru4phY;003v*+SZ3UA{?rqy-SokM0bl5r@jp?C{<+mPV006jsUaU
zE)oXP2=@qn<ip<neg2%9rvP@Xo8B>i$PX+ypDy8rpA3E&-+)iFbbHG#<qLq#t|aA_
z_|03YDk_hGjpBG>wSJ731*{O$kd3By5iA*K#=}C<<$!D<7eX*!^%KIa4B;Tu*OkcJ
z`sRTd5E#B98~nbGd-;%7IBvi)9C5f){Gi~#JdD(_j5WlV*rI&?a`%CmrR95fck}1Z
zxpYU6A_gk?!^fwXvV3C{HLfTmaV*Uh<Ir~?ryJ<8rkFoY?R~2ue?g=ap;U*60NQFO
zj{v(MEDXgx@lUSoO#+AxMl}%dqNKLVUC3y*uqdQU)z3l{T6rlPu?&<~H|}?AK~+i&
z$JAm%JGOxReuBE!_qS5+GnW_8qr#d`NV{{?Oje>Hj?*_xSg4O@eBlY{&xsWcyn8~t
z8YwKF0AY6t=tKbfn<0J?%o-1ja@K%O12@N=3&~z1(!&)Je88sQdEyX{`Hp~7*z+%P
zbGOAaS7pPnf}izkVnTUX_`peS7F?b13S8P?bf%9w=ZBexPY^N3)<U)yx5N$NGgrjc
z!vBm1zW}?|MxJvxIUzqtbJ{e<t%~Epr|A_!1_j~D2D%8@m~lU!<@7h`RgO`Ua=im*
zv1OxzV0WDIGSFf)HCb3#Kr?25<37W%^)HUyUR*ACt}wdg;KQoj6OI!T4;AV^%}Ogf
z6pxS>tkG|;xa2=aw*$99Ot@Oq8lpsgs-A(l?p1tzJXi^g#leMIw$y8!83Ntn3`9Mb
z&Ydnvug(7WaS#U-o)8R|NYZ<Zk^&L~_u#i5yuCAUqv3cVrh38^#cp|an6OtOYY#MM
zoD998qXV2|xVW+GHm0<<eH^T9u2eL!V8lq0c#6^=tW)e~kVJ(>)7ZVlO1pOQn3$+&
z`M}mv9!AGz{q5WVuVPo0mH;HTco}f0fkY*OSyK(&amyW`q`Z)uo&7;AMFe><n45_s
z5lXSxDYL&2W)iJjG1Gt5Ts-{f*k~B7x@enAoV!FSb=fsNIk{H6Z!pWaPZBKxN=h8?
zNQwC6)jv3Q4j(?un+L%YOVfs*zdsS!^-KQkks-n!d+Jy#;`D>L^@vftaB1b2ppS+a
zavM3>2@{Gx?<pwo!l;R7xt9ZVR$ta5&*{v1NMTP<c|s2fHOL7beRqM+LHhIFvMFsU
zxI+tHdSZ(b(7!e3D5zvqvTRqT9>Jo=8h_{{Ev+yNd3a=Q)X1aG{&7fHM@I+H_J`o$
z#MM9Wgy6U4O~vQWojV25H0n>j?JyAsTL@0jmYX-=C)a)%xnuV_q!V=+4)#>15K57i
zk7y3sU@4D-v{b3j6lD*(Pa-jEn*x+ExMM3!`1#d4i?em_91%y^7;;I*tOy4M`d^>l
zL?C76q9S=v=y_K4w>X#e^!15BjnEffbIZxGK!@^Gm<zQa7Bu0%h>vr`aTW5k)CLpT
zpFiRiB+nrZO}q$J&Zz@CM8%`xx^^w1r@nAf6W{hI6;&s^MDi9ll~2lQUJcPZ{}`Qj
z&bQOHm@GSeq#xyiTBt%}Z?Bqi;#o~a+D@ZI9Q5b<2`|9S5|y5{L{LSoeODkl`Bj3l
zqOH$Z!9fuPiDtDV#0BRN)}`&F*=-!q{H|>~@7??N{49|nw`+>hl#(KZY?#PT(+Zm2
z=;k4efNIpFd-hCDhA`5$SH?5<(JU&u<zPMZX!@0=m*cGaR76Ba>0LtpRLSE@APBUk
zkDOG|&OsNE8G`Fu({;w&K_3$<bx%L7FMVZ{Q;1U(H_@Ig*AAad%;Z+kSer`XzKS#B
zIxz+W+X|eFJv;+3#yz?Pg@tSD>rlwG^z;y#F`~X=<NOu;c@)%WCS4N2Aui9)`;<(f
z*n`)gZQ<L0g_BBj_;2X^kP7UTi<zN6qdi4@asxgdczHWU4XHgY^o~D=h8Uk{WE;rg
z#_{M^glSdR&3Als^k=<q4qgM48IbNu{+awyvW`STa{Eafx444FCnl(Ws3)L~kve^Z
z$iU{0N1c_2n5dL5D4`L$ayuaZ)JoTf)k9i3C<ln@EoO#@S^S&EGKPBsl1+S#j}rq#
zAtZA<T~*hm$60wC@7$_q3B~P9f06Q1Dz-G_Uvjqh;haHEdR0vg^-2fyyCdKnV`gxE
z8vQ6P$S%<x*o!A9K%iY0n{lXNq7a($zsaXi*A2nVMu;FnLZz<uXwS2<QS;wRP#{zg
zwtwUIM%;Rccb@6wk2hQQB+ZyH#SxVlwkHx({qng41gv4<8P>CMCOdH8gNf(6cWu5S
z?2!5`Q6?Eb%|zs~^26@Q+jUBYU#awkadP9@&pw59h@${iK~iVxcFNhS64rG@IAo?9
z=R{*rbXjgP!r@TSBaY)M!XJgl1k8oI^0k8Br!w>4TX$av%}$Jd?9m&93emY{#JowJ
zcg70MC)IxdzXJGoXlwqR)ke?^*md-4NWrP=RU6mnL7PK}K`_O=tq9j2JcWsbH7-ly
z<gL(?DSxxsy!T=ZkWR&;L$#@Uk;PG{qn_!cbv+qMODN>97jyHUqi!r@TmKm~^#{Yz
z#f3fFsNB7pZp2FcI{rGba}r^e^1b00=T+pOp%MDH=&|%3&8N~h*ro2{G*Dv|<0WK9
zPFwcElp9!ZIEj>p|CKvgI(wLYB6Ch4`n_3o=@E3SDgs0Q?D$9&5|8gDL8t<*(K9nf
z8Sf3!mwMQno<}3#0T%PenJkqj6;-lS3LOQnTF;ihZejY$H+25Og_dlY4=C=B9m|EK
z^Z7$G13BF7$E7SY5^u*Kzq@<kjArKn^BHM>*@WmS6jE$OPt>}#)%j^d>I2Dc2?|ws
z<Upm{bBz_6n1Gb~4EV@9;aa-A(4n_d(J(21v8EsH)^%K5{qNDcJ#U}>XNPqR$e2f1
z6Pc_OggJvp0>dD*>t)5o<BXf#hB~j2<Jjgmyk(4IJM)c(M~qU^VxO^3`MXqsH$&#E
zTpf$uU5md@h_t6f(GNkS7|w4Job7-oj-$VIp&Srj0QU)Hk)U;f?yR5RY5rT+42?vM
z6@mFB=o|rV%i1d6zdwkF{AL>g$F=&O+7Nnx(?Y|@DAB8I0GKUWHMT$L#|$%_$Th|3
zZcBEz*qLoCah^Med%|C7XAbeWH~%<YvOis@zXN*yUp607Z`?0jVw1A;i<kRLD*x&B
zD(8n5=Dug3kV8cr3bP%cVk5TiR?VahBcTL@=nCz|3OOyVF@T<~MMkE27NJvuH(Uuo
z>p8I%&)mdhhQ~!pw+QVNk_a*>iSd!B4<b9Y67I3?j-nzusTraLSe8G3AD0GnlAx7`
zzX0VrRZ6PqMBh{IS^XpF{MxGNnFxve7kJi$?9g!ChgC}Vdo?tLM6$Wb1?W65#v7Pd
zj9eT}kborX(C`rrnR9i@3j(A&L{Ad*z{eAXu|9@+{YGCuzL?g>!RF_DPT`y)?lJdC
z=M3LSUUOFYr>Lt?l1Qaj&g)|ah8o9By(auX_s4c<q*1q82v2%k`f<_d=><|V`k`?i
z3|Q(p`K%$}LO>W-T=l?!cJ3xRUI0tc)1;`9f}O3>bm>i^e_rOcpshoJ2&ppq`<pWu
zrhVc)fq}0nU^a2L0(XjyS)pm@RN{RBNY$fl=ZCBJ8$Qf%?SzRu@WTfdyKa=V$R$`5
z>cZGav|i53>33)VNkR0^>S&D#kucq7_hV3i^0f@yX~GZ}Ev2>v+~H_xba)*ii6&=S
zeYI0ryKdaamwV|=9)#e4gq)XUm!eZ9Y@|fHy@>=*v_SPg_^^<mI5M=uYI>}w4TR|^
zfiV44Kue2ChHy45{UGX)?WFSsj_93ZJW%Bj#$k`iMx6bhJkT)}>Rf+Gpd*<xP;v<3
z7KV77og=4a7mD$xC5uW3WBX;eH{$<j0Uoo~&$B^?U&Z6zVYG<KOZQJODi$I<hHt+q
zb{h<%CDThzP}<@alIrr<_qk6ZUy#W5&W!PeI!13L!=z(>9z=CGwm;J<B$(T0BUmN%
zmg-*I8B>=9h+#?CTyE;gDJfulH!<9!?g<Q&6ua5}{<CAYjoAsGMj~M-_q6<<Qx&RH
z08YQo3Zv*eyY&YUgzhDEgl=GM;z~W8oo(Fx8-h-vB}()i7Jf@BWc>^mH{NR<f@V%3
zkA!ifmPqM9-KwYj9CV4qjT=JtE3&>`Tgd2vrV5k4aI19OU_5{0-(LWxh&B_l^xI3b
zC@9{a<oVLveAl)cflCnioS>otSB5IK1NDBlmYsCBBjMQTdhxCTPonOPvvYIARqftz
zD9s@1=@AI~hJzQww4>WLE3l&1(WamPxMcv`rKiv$>2ao@u(?Avyba%Q#bS&S&f7>X
z*n^Ss9lvReUR;^~?=;XvJS&`1NDRcR#LD(|9t@mOo-ab949kSyV9!50{yp{|{z$}R
z!PFj2XNo`w&@FG(37@o55^SN_xykZz%`ln)Qih{O4+F(cocM;dcG!F#Zw_4Bhg`st
z#PpaXE;vx$)(AUp2=MYE0xPF*34=@sChV|z!bTvLpBLkG-6Zm(=ef9^<3uWqGK5$S
zfSXpv6+y&+;DeExou+zQ;Q6c_8Z<Q|yi99CXXZ>I<Fvhr+_8w~(z+(}|GNATlPi2n
zzWZ~|j(b&F&rR;vyevtHHx|?Au0PkMn31e`boc(k`<1yb?~zo$OD5wd`LwLH;8?rR
z%+u2N>7dWQD`k%M48B3WjC<>Zmjx$89`xDw0mFno4<8R{F9q5Vn3RyxE{Cm&40-h6
zkpF)P{x<dpAARz7JOf-F_==~#YQMsuCc+5=0Wa@|?&)>Bg$3=(*A3sluVa%j34txz
z@s)v|U=p9gKPlzU9%%yZ6R<0{v5piv4^=5*Sbee5yZcl$G{uiv=pd;-0RbsLAKw_?
zKPe|ykoykfXW)f*UPumy^OT4~Zl#bx_ag9ICo7`r^e?M?>x85vw~QRXeT1JQ*=iAR
ztP^^1$h-I;j>DAKYy4Y4Nk&I?B`-LA9(o3X;~)~FjEv5y?FVf3fc7aFUMQ33CP5&s
zB_t<@3Or{r7GHq87~;mg?YA(?&mpPb2}mF#b2@Oe0<0H6!`|u!!$9M#J68|CtsFv_
z_bcTG0$b>jMPQXdhD0mf+B#FKm9@MjTw09Brv{%1U_IWwbLVL39H3>Q80nBta>2L2
zz7TQR(s=z4cf_CZ@vdp#X1t<%uj~nOR+t!sfHzb;j|X6g;%q|E<lSGCufPfCf<Jw!
zf!h)i?n%3&%9@(}fFVQxey*%oGg6+taDh;08c?2gz(KH>5-p>sr6q#f8vBMwbFsOl
zr#FVhjBV!|8p=LD;;7BCN%1rj*!34rdMC_r+$sWMT2xRF22!tdhF-`TpS5g9R+0-I
zCO~wQ`br84WHz?|6k|I)Y7CUq(&B721FyFd`QHc6prWOP`BVeIq0y-#y?<r;Bt898
zJ-r;%$rTmX#!J8#VfUfn3^7P}2bn1~bqH=%SD{l=(9%H)Mbr%k4!}h-o}&&sJBS%f
z8n8l%*-95L$2%G{bQ#K0SJyf`1q3N3Og)r+E)a_5-UMDKd?WBYun;`1Wp=b;S)+Jb
zkL8WE3q=!_*6HyQXgM*o4N%k(lMk#em2hFS;WRT;Jj)uHm}mqt2T7u!HE{G4I~OY|
zEB|k@0~P=kV<oX{HX<GIA#I2PaY5JlrY9>gF%iyC;1XDV*x-D8iYpRl+1WMFr9(cA
zhd>-FIXOu5LWUirjPX3xkcnb>R|thJ8bv7FyO#-Mgt#<t-QtC{LnQk3tCwGlRl2vE
zU^?2zf2`)iP<>!+=H^<W>-pg42WSeaQ@j9t&y{Tt6glV5o_+9S!jFRZ|ILRsim|wE
zi=v{KOdg7f{j;5g{1OE5v!!)%ys<xo#v9?(Kp~!Ea~m5^;XE`lHa0diM9>1L8~h<b
zw;;qIFgAa3T_XeCm@GapLu#9ndH~Z89C85<RTq|lG|J88H(`!=q<;9%EZ{^Qx2BA(
z7mA9C6~zI7zN-pDX7k+yWji}N;WU`W<pLT2#TB+F4aI4M`mwUU@h?Y&0I!+j`R$%z
z8j1<58&D;9lrx=40g&7xDALUJaBsxf`69#=8siGbfenH_HzT7JVG~_lO7ik}JH(hb
zNl6SyRxR06wRd};DaNMwFYY#S4AQ`7W=Zn{E!L^^Y5e+szDjw(&yDz+RoX)p@xNcc
z#6-~1{?C`Ftrx+E|Ie2f-6$%=|NZ(zy86GmCH%qv{U`tTiT%&I`2TUicpqQA1@k0e
zsdsY<_-UeTXs1vWbB<dI3zwkIlvtY})~a+LzzJAiYWh;!RAV{=mD;Y-|H~FXsjlen
zq*Ow2NSGe1=A#D>80UxZw;3EjfhB*(cAv{3KEyl0!NIuf{q(37kH*+@?~s<)6xVd1
z{QQ462p-OVU;ke=yZ6)SBme$^J-8mIctIIT$NGj54i-f~<DgJ!C|vrY(2(lucfE}x
z*|E{t(E)^3Ht)o#Q-JOZGMY(6*I-9L{P^CNvdYSNxw%iK+gFD$D-}a*gA1#{4QOkZ
z;jir=xi-6FmRXd0_v8fPr$BrUZ4tL{l(^d$;RyOBfTjSEWcp1PClyCWL_gvJ!X*g~
z0dOQ(b$o$eW$XO-x}&pmY<wKG0w@m<PS0c%>n9j4Tp)Z$>FF0u-NM6R)rpFYjs4K_
z=`3Pk8*rP~*Plb!R%fDLYAR=`cCarxEL>x53~x~$Vj6s6hRlS^$nW2dLlwK{V>2_k
zd-C>`OTb(Ku3KL}7B?a;WH_2bLp6MjWUnPAC4mixh5|==PIflociBvPOhVaoPv%wl
zi}sy{)Vs7uHypQ@Zwk_Tb7X2}$No9h`(^A>z`&jSo*Sn&w0M8nW45Nfr6vEe6$}G>
zd^ul)>Hqx#C2PLprtPFZm<MndS-<aWE=MA3gd3&j;z4W!^z+#sCz=`Lz#kCRlW=``
z`9}zGcUqn2<P<@y3@R`%J9vF{Gf;@Y{|K#k5S&@)j%Gd2FYRFkGeJCkNH>6x;}Y#C
zI&k0s=3`s|#|z1<tJih*vu9mBJoHMEU|B9sJ&z)w4K~wx;yHk*#`Bt5|Jei<?>8!M
zVw}sdV@g_DcT1#XWS%0Dl(-oYB+y=bf7c2pa%EQ6*X8q&g;9=j!D6De_SWR1D|M)`
zBqUtAuS3FssyIA11Kb?o<izdOm$EFUgfMHOy1Sb&gd=+T(-UPmxsz}`0N{&7Y!l3w
z8zb*Z#P&+wx|I)47sl{nmbs=TZGk3K2^}5guqyyG{N0@vLB;`rgyYKWivrE-*9lpA
zw;b@U`^JyYf9cYSD@%<}IQa0@4xQFd3(qk~uh4e!-10JLM2w2Ug1iYH#oNoPe6t_-
zx43xB^A^%mtvJ;1smBx00pdketYWv;WlvI5pC*#s&u7tr)Zl+MDB64Tz$34CR1=|i
z@}&|`nJg0nHkJjQwu)hyv9-OtkcwIxOcCA;#sPAu=dx=SDj?VK8}*0RP`e<si6FkA
zoUw3$i~@LwcBy?&dHH>GaQID{@;YFoPntbG@7&x_x2XPs*H#<BOK1q|I8fhVMeuQP
zC3Ht$I*lM&c&2BUmN?njb^93+h6V){noxYFDZCpA<We@7xxIfszV#zU>NqA2B?xEd
zGg$9v1F{vb$BD$3I#-H2sEfy%*<*tc-w0!dH7&LeNAIA+O9q&~O3?<aMO+{gzxMgM
zxOg?_9~8Qpa-V$)&R|o*T7YT@K<?1U2oEffFFel>s}RFp@E@?LI(TVZD~DK6(Xz48
z!S9IW1LTc{|8d}~M}WUSs3QR`E^nIvKR>FpyCTCblyn=Xvs~1pPMVv_nu=wfpe8xE
zKY8M72e^wzgWEtFjHlaQ-1Nn)Vwis>k71q-VTI2N`|#nY+Lf>DHrCcePsC?+lh|uI
z)$gG1LZx=QKumIFX(l8rtRJh+hIJ$!v!Y=!n(TBG72T1h8!8IY@BE^^qGBh#(*~oO
ze3K})12T}~;!g9Oj7vzsM58B9pANu}GWoR{t+5qo8PIPJOYZo*c6UE89};k-VZP|q
z;6M2&@<1%dtHXEt4M6i+E_{Sgweie_9p@EC`3%_&YWw5@o<x|0pc})$N|k!@+qZD&
zqV&eV0glGJMlb*<j<wR3;E_}A(hc>hx%YGniZsy$v}fhZ4}hD%P}jkM*6x^zFge!P
z%a=R3h@Sx3G_Bj*83s_euMGLm?4=j6N9Zp?U_D;<%HU@gTW8d|AVjD=^#x31zP?UX
zsl9+3d%k}M&l+kcfAgk3&`?|qXk#%@{3)nOG{^G`3sAt^4m)NJAr*8Be7D1o53c6`
zNA5iaK~8J18dPqtWJrTkafRbu-AnSDs~LI*hTayV2om_tFcOhXm(2}ewRzo3xo@9Q
zE&ZFpe?Hn=7oixuBCDo$dy0rHA&evjDafY)pyoDG{P8R}rXLv(NzICii|z4v2zlCN
zvl+Bq*fRciU-`(YstP`+AkjLM=<nYJAb%92WAyde)ZPSR1LCuv4OTsS00c$v_G4-3
zCQ1R79CTR2Rj!bpjsM`icoDM-h0x03lTMy?dPu}NLq7_p3CqDPblo1k7?>id?a+Tp
z<q->%9h{D15;laWoMUj-A3n_Ve&J=T0X_{7g!)gqzkW@JZ53HJi_F2<NZXiS&}bvS
zU4iEZEr$*H0MEsX={UBtvWy2xh11q9ZeV61Pn>D@{+BNLvH6JtY^=^p3xe9Ve{2kl
zg3SR93?ibICS~La9y$gHF<l~LY`6ybKJE4j#T|n5)VTo{Wnp~^7Xn%xf9-|@DOjbS
zYtMmp0`j-My4t#=iX$HOEkmlWRX;wywzjn`#l>AfL<*r#h0m64WD5-{BF6#iFC@IF
z!r);{dz{@9{wXBnJGOg*I&i%lB$#1$?8)7ro%{ROer|a8vK1;n1k#99SL3WibNeYF
zLC|5%{K-GbFhEhB<bJI-aB7D&Ce7?HajO-!(Lnb`jBPM*-RqU@OhSoUyD{<uiP#4G
z$Jh~ltc$pYF)-tfx>qFXg8u%^ap|?4Z*`u^9tJjm2WXfjL-tZuNl6P#9IUs!c~51Y
zG(s+c^9e^DusdembsCDArX~)s^Dib=EB8=xo&C!?ev3By%Fp;#<-oJaXYSPA9kjT0
zM743~`ft*1sZ<Y#V_#qMPS|VUzQY+iH97frago}`3a|-Q4ZMIr#b5$9L}hV8CbJcF
z2_|M?jU*)|f`Nz7IiFFQZ+5SYq$Em^#MD%x1c{7%a9yc`syj&uQ)Hof3kZnAAKkzI
zRV8f;1+)PW1&vc}_ib!eaXyf6k}<qQDFi_sFc}}8!>|aVU&q$$;*@PgmyJ9mg0VqY
zj*lgZ<CGWrW5^sLBGS<GW8J2u4SZ613W4KC{KlY0#Cn!NL?k@7wYK(*fPl*N>#yvx
zuwrl$qol~z5hY#^r)DrDgEw#&rD>OOufIc?fS(5-MeI9pi~rF|lK$z<k*Yn*0+JzG
zOZgd(dn1|FM_ee-cgRMxfGh>{tK3b`$_>3$Z0S9ScaG}ETvb-?#ySE*d5}a-S$Se;
z=(EP2!o5a6K3_-J3r;>&)gaVuFJ6dYEr=V(p;!Zaf{-lSD|nQ6or@-sv9aS=h#*;T
zQ36EfN5Ge??7<n`{1lv?*f4nptrg$Csf-G+pm~QD5`kL)&9FV%e))fdZ<CUOBG%;b
z{aL~&k86||70u|8ehwna`g+uJ_~eXWZy;WPqkM*u^{lMCd?oo)+_dW129eRx?MIIt
zi8H!2`wYlL@4jvlF*t+AA%$Rhij(a}{`1PdzT*7+z{2l%Z6fw7qUuu0)7aw>>;x(v
zvM@JC0o4r-zSj~~T#&Btk_tkLfnQZ*69^$tPv1+xmPn2b=`CV~+RrQfs4O{c5pv`E
z%H@&8>GB??tDy%N+i&Oil{$;K4B9uZYba@cF>X;xOvzTv;XBGVDPX~>$ohurO-~H|
zZfdQ<Q3ZEi58k7E^Ysh&_RAX!E}s9%c)j+g@lWHk#tCN=6}uGtmHfv2$BrrPOD$s;
z5*HNTMZU{~Y9=%@)ZNE@Qq8}*sQQ|!RzZ%MF^3+9xM20m`Iq=_!}U6!UHHfc97H%-
zS;qjZ^bH^r7)5GBeLY4w*+P+EG}MStF34FQfs?lPKk<;?&)XY?00hmy&=;6q0<MFo
zB}}LDlzBIVQ=zZ<Ai55~JR%bjVPVpa#n?fz%w;%j&;*z|@V;b#D&wBKydeids-}MZ
znzJEOg3=SbHTL0cNy*431mlFp1AscDf2_LcO!G&Y74O|UeDtU}UIV5|0RuHH0@{Ym
z7!+m4In-FflszVa_@ZtiZjkjQ4hw*b1#WH)(7N5o*FUJLqr=3b0eQ)tT_MO?LlV1R
zL4nP#??d&CjY0sZpr`@Ce4BbHEg|98;NV5m(OR503uSC@n~*OGmxG=!54%sqtOGU#
zBJ8PgIu&R?D)XLfLwUS*lJEz3*P^1LxQK7uATbG2!(9Z858}5l1tv$5b^HbbaSg}6
zpyKu|;NetN?LifWL)R~N0_=&wFP@A`XZ9k!P}iRdTp2UZNr{-~=mkVc?tNUGoo$Zu
z31HPIfm#wdFt6=t#15kRiXOWJXm?=xx=|aLoVrJCJw2)`tnv^gqNr=!)Iom9FMO;C
zF=P37<iOR-H^bH~%9C;?Cyh)CF3Zcy8<F0_S%uQ)=FL3?W6(o3u1W0KvqzV0&Sf7R
z3KszOtys|Q@4SKo1KnZQfw~az?6J%Q2=?GDSy@~2NwTR&biAr6ctfN54`+8$%1B8O
z<BR|{Zf>laP3o0lUqwfc;6x7%3tK%fGF<wVWGsR3fa6u;mEOrk%<Dy<+uBZ>7AXF&
zZOP$nUB#z>XPsbdP+M1r;sqL`hxqAHiVjpwcz7t+6V!<W{X2U?%ZLCqWY}lBZr_2Y
z8(Wevxoc?9m@wped&>c%JPr53^mOA@YhQy|FU+<^fq=J#L^4#UZOzZ{Ls3$1{g9aM
z>g<Gg<9$JYW#y~fY8-W)or{o5WY2(wkEIwqZ1?JL1&!$KgH^bDgX#o`!2<Yt;ZcT3
zI_O77rNK&})^?>S*@o&u<?oaBkfj(|Ir+KEn|xviZ6|Gc?3&Fp%v%-LWqdZJwgNuK
z?UA?QZ|9=^`aqQ4S7iM1IEUiTfERs{Hp^H0C>@W7D&O1|h`Oh&koPsZ%qjDBrh&Xt
zlX+wIhhLSN+V*yzr47j^1q^PsoOJlU=*PIVaGXl<*yh4e#4b6~d;8?cevImlJ?{N>
zz~JgftB@bDDbk8mKd-a5yR{GKmzD%QTi9}0_v7<R_8_b5`qq^a@ZIbE^-<kzvj>Vw
znoBJ;PVcK}>;g9Lj_Mm;Hn0lZXxIJDpR2i8SYNl%LZxx<?mcVM`~j)5wB6d(i_d&2
zhCH0S-(q#l3o5b48%untuFm*<or4Py!S>ds9afvxL?6ynRMT3N62|9*ghFMW=0aeG
zaxN_v8CDJbDbSp3Y~P%GeYSVtjBZ+TL<Ie|%YEc3dX|e4;V!tVj;yfrjqXiLP6qU?
z37C2%!qwaR&{R-*xXg7eK6GW5FI>2=WmrW-0v1_%DDOuTOup%~lm6}rv?ZUyM8Q1m
zrekevOTxz#LkkMAs?THld*GF<e+y4AT|f>G_`gT!M$=*;uWS}NhZ+dCK}$q*RFo;2
zrmYs|w6rv|-sL>Je0(KN+ok0bO2vUwbgx}p;KRVcbBUWbOReXI(dOYIui#ZwS8oWO
zBkC!$3DodiZk@@=>~(x+_F_jjZsY^&z@4em8-aMNoG}#M&jo0@7|W5rvQvE>u~P;T
zXV0D6XZG;~<Uav6^CV&?PM(zf``p;bDBdiAXf4tz!8S?KU4fN)oC0kpvsG<N%S5|^
z<&x8S1~|7{B~$n|`VOWBP@zcC&A%d{L=jw}vF<^8#eb2KB|_zQ#Id7CZ#4Xpmqry>
zSQy+LgvJH8s^)#2hSpZ^78lAHKpplgy!l9)2@XEB)gtrY!4PyID>9?JJUsZ!Xw!~W
zq0hqrio5rbP`%Uhus*{CC?ErgjQ4yPydoGuisC>&oG}SZHgq7qGV+J_(P0Y<3kkgr
zIt+^cBRZ75_D`y%-2Y2&3Ud13?|;m={0oh_<iXdVQ&7v0WSZ{m#34}CW^H3r5EH|i
zGln`O{mVCm237EWJNnXP?-90(EfEZ4IT0O~r++hyMMyp6<Z791osh7nNJx&{zGC0F
z`R##~i>k6dmEzIX-?y7(*eLs~ogQAZeCgG4Pu!xqS>lTHa^U=BkGY8&9?yBFF`4f(
z_~(uNk3GMs{mwgJLtin~+K|*Rp%GTOmf!DCNbf4*qsa95PPkks*LJVHNXtZOO;Ml(
zM^Q$jdD^elfC(Eub$RuytENsRr*!*!e!uPSuPU;j2|w;PTDe*_pZD!4V~$nq&^`aD
zrU#YIbAomHIpHX!>g&$zX8k`}0E}<Qr?*5qg5xkRC#SP@^-teCU=Em}ANXh9y>$yw
zsw{bb6y)UzTmVm_qp=lr3DVRXr{5c|;gRQ}tx8Nh_gh*}umHbBZUw%G)E^p}qgb>@
zk5-i>>=79PT-NbML{UK@U(eZYaZ;R|6qB5w$r4Sis*$MF#<ekt+X`cXUQE*k1O%Y}
zv_c|Eb~Z~~InY9xz>6;gT|O8?b+l`H0F2`9@LbWxYgA0jRVeT8J;f|8Bni1%tC5Hu
ztio}MoKB*}S350~glZEqt;>RfkT*VLH%F}v-wblN>Wy7cBclH=v#a^hb5Bu`4|gsE
zgea7y2OpwA_Bg@yGN()-7(;iWVq)|jJZQx?CKQoKUc=+SqC(@5?d`ye8lzE69p5aO
zwg4G19TjI-j9#AukTHbzuTifR%%qHdC)9`-S_VE5;@LBe)PBgkHSRd{5>t>GpKRD>
z<mJUl$D+JE+(0VczC1rBWh&+lR*M3jQ7sqw3nKKwwk%(c9EsMa5o!$PYZw@aC{Boq
zs@QZ#OBj{9d7OrZ{h7xI;)3qtQC+z%EhCc#qy=85sMtdoG2#f+LCXc2h|sG};WP#b
z@cS4V_TUX6sM57J;Mk!<LqC6>Y3%ku*gGUI`Q(#RQ{Pcmsj1z&5(G2>9PFLuZO>Gl
z+fq{S14C%RRN#sIYDFG|58bB-$-e>u0?535mvd-t4{-`KIvshwn-WB6QunezJ1#!N
z7%`4c9t3SiBW%lS5$}ER)CnP+2KaYKcuh)np9tk^!$)iK`0=StGs)4x7kpoS|0ua)
z>{x)M*XV&3q~r8$l*>D(D$+(ZreX5Kzs7}(7xPlxBTmc1&Kjxa-`weh3mK=fj$E~z
zWsuALl3NqzX?&)2-6j4{XU=A~r6I?3tq!XVtEVH~6^XAk_ujrqd^uU*I&ssS?z1te
zp3NBZ!}sP!)!f}hnhln2CdYZ)e$Pe4rCRl^u2elaGG*b+{krjow_V&~;1^?3LDhmW
ziUucY25Z}W^!Qnq`kzqn(B7n=99?)OUnEl^w3u1y?0B6{E-F}21YSer*_1#Gg?pl*
zsmTFYHljE6^~V9s)*YrKs(pZf<*|gwO+zCY>56~;pwit0zYALqd=@}<e4bCB9Ccb9
zJ6^XbNI-!DWmhfq#qd$D^K=-ON~)@!EYK(ce&Oi}z@{1KLAcO=Gi>mz(AFb176!5X
zJjOEYGS<K~Ay0;1M;*Q^v>rel0kXqACgil$*V7YJBx7j!3k7-|4bug3r%oXuAvjg4
zHVKj%`G056oLPZ|1U(pnO|beA;m|_MsQ{E8!ql#{xnV_ld3ZadzZ<}@rDlxR9&5rl
zK*<D?0y=pTliO*cSfR)>-)CmYcmV>g!zyvO2m!lMC0oHDp&jUCa}RwGd^nR9Z@-~^
zYh@}!7lu~}&=7mW68iFv0$+vXL+LMG{F6494vu?l;4NTMlCDrw)CEvglp)3$@Gkx~
znuTXYG$6Wh=pe{7VbbU`A3N@m-20z*&c}jGfM2_uv1B75KK|lq*Gu*>Z~yFEY{|~c
zOM5D54d~GHhN6+tB#@&j>nB74B19VlU0wctZMf3MFI`BJbZX+WehWoBw#M=@HU$#b
zz(Eu{tn)+Z1I`#nim>n|Nc@foj{vmmpftKWI$&%V#jk+Z1HSM3vq`9KGHM{Jt5Ypc
z#?d-9GUAJRI3$D`hv}h1bLdMUMIu65@$8CZ(0y2_uqL6SLD=m<k_X^D4ju9i52r)7
z1u)9bk8t8~e@23!jQUupf~Mxo@bF=j5+#^Krv2Og2Ve`-6oA1Q!DQQ4!+Qw@{*5Gm
z*1af`AI0)iQ4z!Cus@}$c$nYfynqJs;=0@lAPyoeyqn|;z!-GkxGj^r1C#-Tfd979
zTmE#t`131Q*S%BC=&c)#U*jk-ic%#uZyE+^SGsU=ast?%29Eve6){f1)m0KrH3H7!
zs|Q!|Vf2E)q&8%2Z4Ej&qrsl7+Rv{><IgA1pRf=WyZhHCx>!J6pvdz0_T%CHyG3@V
zGDp)AK8*S~(${8PPb$qN@9!$Gcr>&eVZvvZ-DB{gs3|9W-RyW|Ec3;nUx8uIL|Sj>
z(DB(6O<D?%ncfx`6Frpjr7d~>IU0uS%uK{7CuC*u^tdoEGxOlo!hZ9OSS)5bmk4hl
zu1XP%I=})N^^3!&h_Deb4Dd1JtHXR&OLLW-6Z8r6H=zDU{Q)7o<~o4!BrE*es}uC5
z))f%4AOyj%+aKQudYT1%J}6=gA3m%s%fU4Xqyx>vHiQPi&Jee0ATN$ZU;#iLYu@xJ
zVjRn$Qr>8Zdy&?Q){@wBoBge=b1-UP2SqJ?1Eqy`fvp;9CymBw6_elQuf{mLd3ruN
zRy9LtR#K1=@bzX0?gE4aYG!w#bZfZ06wdG#0L}OB-?Q`|k-dc79f2vw=H+U+WpI{&
zDk|<R1^|HtU_cFZX1&_y*9>p5tB}YEH(9z;jB5J*e{$$_+T~PL3veu<`^q|YVSOV=
zufW{F0S1L+E4g+W0**6~%6h(g&mK$oUSO7q>bwio2rX4BW+MTwY-}`_N0o}}KQog`
z>>!krR8rm_iAyT?jVeQ?YSj4e)#DNmSuu))_NLrG_zp(Qj&OPN#=GG3ONOpw)yj==
z%`AQLsUd7e2GWxWiHR^t64j{x2ju_2cVUSW7l_dI9eOLqx9GuPP764tf9Cx8G_;uL
z0s2BvI~VVX3<(T``DO$F1gzNE@BRQnKzamnR2jD~!!4y2rV$K@Ft#E<Q!H_~I$Ziv
z6A}ubLq#*;y=x2~feI-<4hp}}Fof*S0~f|@cG>HM)DpuQ-xf1K6c*K-;_n4Ks`nUv
z>Vx{249uX2OIe2bPQ)lkd<Yu@N&Y;$s9!R`a+%M=0U3^}C*gf~BO}VoVvV6{|4f~B
z$KeHJ?$xVZt7Uf-H8iG&hnem?^8D+{WevP9=I$FtR~jL*_Kz$VUy%YUBW9og2}TEf
ziNla+pm2y|VD)1JMr?E=Kdd>N&-j?20zffDMDZj$-FMhb$L{-?_{(Rc+Gx_~u-Rel
zxU%L8$G<9isrm$Yg&hnzwtjB2v-@RQv67sMPTq~_Ad&OY+kX>A{l`l0IIX6<b(r*g
zctzqN$3N$JyG)Jb_sI|Qj67`D8Q~u0mf}{m^7ZvoH_B63;hBCi8?lu>TtD3XD{Z@J
zEid+c3HAL3FAk%3i&|Wgl3KsM<v&a&Rq9gM3R(Qq@Hb%h*zQI<w5<FTy($APzVG$@
zJmr;?{!{BkIzVm1O^(27);qMnVIoX-?8lbFiQ;n!X1p6$zTXsE;zz;tp8C@JmT>gf
z%=>$0+saQD14u-FHKASi1RRZSakH9ocr0UDFtqQ>0k|$nNl6#fcuGM<=`O|q!S2%d
z_rxZ)WuphVs+kck<m`?KB*C+781UV@bLYU6yZ-N^<Cu7f`V?p^4%<wn6Mb+*&hq5)
z^72B21rZSZFFw$O9z0mi{<m`I8O;6(>cmZFtm8u@&O@L4y8wEiC7@x<H@}iPqCocj
z<HCkHE=KT*HufAzKJs=YIr6jZ@u19r8?SuIQ=DYX#_FlIj*EjE{MDT{0KlZlm{7j~
z;{CVS(8Q!^&GPU+3){-q5@KSVgZqxghBN!(7Bx1OzP&&}hg(J>EUcavy7L#pH=$`m
z9icz7ACY#a2dJUddMMoY<45@RZ3diI@PYUJ&PFwc4q^T(y|gqD<JfgB9uN*ooFpdn
zaD_m-#iCsgqdZ(_p567kAn0S<Ka1q)TZ3NPdvx|t;+OG%J}UTtT%XyM4)q)GFeY9L
zl<-~ihd@{YeI0Jj##mD(ZLxC{F3gQgLM*K>T@Yo1fZw2=;E)h?G&ZPyL2`nLLSnFZ
zn0uoi7}})c>1Rb}MXL|c0r4hT+Qi^qObB+<xboHz0)-9$j~wb>4seSv08aH$QdeGo
zR29+Xk`iVgroox7LZ3?qDzSWO$D~NLU$ydZYGdi;qV7m73JF?Ak@bbqf9pmrxYJvk
z-fwv&>{PT)ZN?`9c$C|L?PiQ(%=ss&>FW(oD&FjM@Lbk(G_x_G-DJA!?3MA<xW(35
zO|G{!F!~7Rot0nZGP~{`I5Bcb-?V)6`9zH%jm0y8$!EOdXIwri>z^>XtRSiN%<tJf
zY0ai_Cw_JMT+4z%?zO(3KO}6m-#=WvVUo~hS{yT|Ho@C&cjsxeIOR2?DI+VP-Uk-n
zjQ$q1;zy(l>SVF$lAGD|&H0af&(lpvYX3<Tq@A69GTrnAQRFU_hgABidV6gVbW{9$
zta)*8%aLGl&pnhrxk?NAq<>aL)?G$`UYBbT75A(RHQ%Tr?i46Yxt5m!naxjzjDZwB
z7W9TFcxJQ!T$x@wDEL%axAC_ax`C~@ng{@(IOOBihet+sYo%U<gxjm26+TrKme;4?
z{Sg3n@ULnfs?b|2fqV=M_lY}Jyct-4O2f0SvHTi^=B_<<U{PxTw#Z8KAg-5RKp?Qd
zUkP);3>w-ld7%x>(|v&Dxte?)Zo=V-35oBur;6t#ZL)rdo7_P4ajk?+SPr5{nGk*t
zzL+g6bClk`mmpigMl_63zk!)qOqVVN<UZEUy3zD;ZgO_e(^jFK{d3p!E7FW&IbaIb
z4(<RAz0Y%0Rad)yRjvZ5ATd6T5^v>CF-l31N=Mkyq}iMUV-7$hE)z?KB4Oopacx}|
z`r^hH96l6uD$0?E!cozsnEpi1d*j5dyJ0mFU{!u&ht!><MouUa5FT;Rz_ZV|!Fv-6
z9ee4b@ow~C;kg$-V2}vV)Sx8cS+$E2Z3l%qq%<|dYWWUANr@nxy311=M94K9Pc~V&
ziW>QPZY&Xtdub{~n^J;4maW5QzmKQz=F^P9vZ(DztKx_W&6b)7`r}Gp2mG;Y1BN+G
zY|~wPGCjQiP0(jjaWq@?cuj2)x@~_J3Kg57&fVir#tZ-UG>KD*PyJ4%w$>T^J==ew
zU0bO`;j3jWdtD!GX-Lo&`495bsRNoTJbm*m$I1d6N7M5w-c?vwb@xjbHuRZA>21Di
zKSVOr<yQKhT<oCJ>#gU*J+AYU+XsqLM>;g!%EoPKd~-uW-TNLV^(;EvIBTG^DzLQ~
z({b_Ipx)9aN#ECfk-5447*`eRI$fUx@@&O?FPF%W{a@_vbt$0WsH0Y=d%ToA)D9~S
z)_=kUPC-G@H&D0t&?l6hUPoTS&skM=_C?@H)(2f8BTs=XWIvgJ*hEp`>(zi&RJh6F
zyhxJ3?Rw|?H^}XuSkRr)dPr}8EQSh=Ij=?67z+#T?#~Ql709m@i~`5&ttu9qo2z7c
zq&^VlWiY3(mbgK}`lPM`ivrsK_CAZ>javB+aK2?5@9AQ>lJ@x&iCDV7>!(y}20#Wc
zoNo1J>PaDW3KptN=XHWHfl}a!xrIf{)(qa#M3$GiL(Y@_;DQV}pR#*59Zr*FG4UK(
z9C>3=svd2;^cf~D^IhkUVdUzpc9s7Dx^wWT`ed@aB7GrMKkI~*-cd_41)|?jyr*{X
zGyK^<!X1QmE`F%}aTaqck|TsFd{Hz|1;-~OL@ue5G5@Cs_RED#+7ZosV_1y`b&Z<0
zY}@BuS%gNr>rFVg)E4`NV!P${tCwz5yd-UXrpz{$^$I!KsO^J-P8$0NLp7v&jSF=#
z2$)f6P=;E&p8j{_9=5pf>qp9R-pU^{mhoG1V@}tlS#4r0IlfzT&KT+G>KQ+7*nA-T
z_0w*<rW@YkY!c3n`(jMo*Xr->l~aEsV$NK^uxi?;H#x0F&g${LmzHCZ-K6)Mm(wk|
z-BPmsxx3fxonkvG6nej2OD!2<r=`-rtUviq$fb$f!l<+EmT$P<R_OL>pOpV0y1X9)
zw>dsJ9bGo*R$g)OJ@rpgwUhRWjQ2;a*ZZdrA5yvOQCNwx0LB8gs=fTB*r_0*%v`9F
z)$=&n*z{ome6@#s_^RQD<5+|mpasz2BN{)}(ObSU82l{Ee)E@$8PiUwO5TQ~orlL)
z<L$eR+}O=)KR%duR8OxAx6hnqad6l~fRSwF06r#^_w$KwtNzZ7NQ|p#;!r5w;|w~{
zx##+s^sn5<r~D2ZU%fhG@<lB{Sz6k|LY$+E;tM8dn#OR{;l_^u{SI4mkQMwfo=_X1
z;=>Enp@?0<7#e*13C!z(BM0}=J>WWSoGzs)Aw%?LJi8lFS>qbjf-K|7M|8x02-(}t
zlH-J11z|SL&F}j%G1zEQ;R4#xUaz)2I-AG04i)AnjNKNe4iu&3Z1s9j;nS}1$^|Y6
z)2SpSWo4qagyjx)=A$OhcOZ{sD~*=J`rx<SK(ApH?M4j^=Gv&hc3$cn);%_L_c_JB
zV7(3c3e2B^BDAjt{Ax6t>wnADi{@>OJ5B8XA1`lyrBP|k-~;py5Z`soIioQGC<iP%
z&x<47@el5S5ap)NFk~WWhOC{;*Why{G>kzBjcBb-v$AF!Yt*B?L<=g)Vp8G3-Au-V
zLcAntwJ#XnB8+D@35gfq8ALnbwriyEvq3-`h#;@yvXF*DF4yrs3LuE8n%cX$P^WK$
zVo@r-5_k5Q?qj|qabQxQwzsxdv|V&CGh9QhKI+cy!I34qwPf~}3@TD(bM<C93WkEc
z_J7MMc5zs6<S4%Jh*mqZzvzSN&p6p}cjcE1BI6DlYO{X+-M)U_IbXBhN=7=>F0?hI
zA2)lbbC)U2-ayNS&cxLC`!N%XPV=nWt0|W_9GcG2*QnLH-;fsEJ8e3Yv*)0z+VR{v
zrHlvn-~d1}JQhy=B@9)<7nT+V69Z)<7(9S}8@x;zLPJ0&5W-AE9H5LN(Tbd-FZzMA
zb6PtQ*LHUr{kCO1i}0g&s=M4&H*b)@<pZBn^wM5PZ2%N#*H$LPaNPn&J3;k3OJ5G+
zexlkMf0<w=CE1wCO(OPHpP!4znM8~%`Q5E*wUL|CwI_Ru<CBsi1kIfJ5;2G%#<?ST
zyV!nfU4>Nq9`zwR6tPL0aAw_e5hg^2Q2c=vYs|cj7MyM2Ie;T!(@`>;0B`SBEVL@S
zisbDI=`50bYq_k8QBB;j@BNK}SM$eD38Ailzbc@JNlaY)4dnj#!d-#=i7+n|>&6cs
zGV9jRZP&gyc-i{a-MdSRix_ohYj0m%{t;@zx7{Q2suzQ+rW8k>35@1Fz<7&+(#4<F
z+jn2N0MfjtT*5%6gXTPY&DcI2eEgcK>K`4G*OJ+TFLLIv7&Br^ti@)({(VJ?=Zmi{
z`OA3O76l>bwMlpH4hjx#+ls^L#!T+oLjeHiV*SgWVXGrL`j6l?(ROt_pVVppDK&7B
z;Nlhrn;N}Uce<u){o^#JI@jpYc+HT+b12^J@$pTaWw*qg4*n&giSH`t<{1MX8CFvk
z&<Bo8Pv+b)GFJ3wjH;A&p}a)PY0q3ZSz9A{IIUUcP`c%`tlSlo(&)n%NyG?R2HbLg
zzc2@fM^Q|6_FtTGef49wIYAC@BFAAxp`4L#V{aK&2|S^%52TR2{Md9P(y%s(%~V$-
z%`lwlzCYWc24-OLm?jfmDDoJJt=Y4JU2$$Qqn|*lz}6Oe$}1#9isNhZI}5pUFZ_xP
zqS?hO3uQ&4(I5lw*FE-|b=$2#h13U8*b7Gfi=0Nx|G@@sH+=j0e)6$FbO;opnVs~m
zgN`Bd1@x$uYFj>hZz(A$Ps_LAiZAAzS3Rx9z1<!{Z}ck$E5-HePQHM1z3&eCbh*EB
z0cnli9>FZHHkg|0>j`0})h|T|X8@N$3etMpor_P%6t}-R)*VK(&bIfU6|){2At}`*
z`J<eYo-RAO&uYMM7QXt6kS;<qW{TKY6$C$cxpZH$zQ1XGoZFj2ZI;2G-6o5YPG!3s
z5ZO-XWuS7pctI2bWx!kp)Y5nE?4h8k4Xxc+no;Mbmk9HXMEb}0_?J)*F}vOZEQ&f|
z95$=&{F@I@7S?ez=t=9#dGOH}JsPD`4PuiUIiKb}#{8jsXDqMNfz5|TR)=b4_6gE?
zF_4WH9@{QSdN;-?9b<dol}RMi^hIHjKpq=v&8+$wSMT!kHj@-O?Booz$fbs^2Bir!
zCBR?~S%hA*uvo-42C*ruQ}q_`)h+WKgggZ;Kth8M?vg2d;m|mL-Zvl<#fh#r?N$!m
zsrf{2Wio8}ou_H!B%P)yjd!aaVpq*1f;BpPT2MmZ$5hZXTyBnNS<u$hY+CEx%j4j@
zxGgov^xR&bi7ZW$D&D9&m6qpI?3?st`P$WvV-K3^8Oer`FyW)|%I7QOr(H9U&oz3C
zkN@47M>60>0xot)xi`l0A%wyUi3eNOF0lWS@rgxZ*^6ZL`p|2vvTzC%d<ui0(!iIM
zHRU8JCAbnIRT`tzfa^fM1e8=-+65xza6z*#-Q8A5f(NR3(1up#L4HvY#E*f6gu1A_
zTvT0MT~gAO<{EO`@22vXpSC_2JK@}&bhC*Aph8zt{ipIOy<L)=*3i<H+AOf~9l}Vg
z)rJdYWM+>YUXgZvNb)-J^pMcAy(fia&5|{WOQ+o4y;HHPG5q|>8Y=^^`o)Fk5IO76
zI_p1q(%e`xw~r1w?YKKZOrwU-zxqU9fb78s{j6~dJ-@YAXxq7evK%Zr%yTLF>x11J
zF0{y;y~EgjP}WCz@`0%Ksk_sOMyetltJ<_?i%$n4l({)m&Tu5LxII>tkdAxB-_m(%
z_dTzXo{7jaHEmLws_LHH-iy6K0k$br=V@zeGGO%@8CbqMGyKr9a38(vZQIXwT2Pb4
zp*RK=OyRQdKV$)vH)vYC<Tubsk6=y@;*=ue5w-CQJZ+zDXmav2bQyrAZI`C&+#}GD
zsa%K*BG!udJ)6$s#_Zs}vN@opskpOn^cKQlP)3Iq-#G|jcy+8ad&4fSjIIxQGs`1p
z0U)>2yrs$GFi$QfMm%IN8QSRdAN*8ihO0NL)R<m(U8oO^1RQc(TH2)DIpeam35W4F
ziDO~}7;+zAM`75a&YcwxpUTksPtcw{`)I%;8rF22mNpFe|7klH0NLt(B~X!Kwx|0B
zikn2NzhayIPg(|WB!DO=bx$5UwloXr-x#k;C88e!e0<t~2eF#(gx*z`8=N4JoOVQp
z0KIN35ClM^r+LTsW?<`Fu@)aWIk{=QI#k%4+OZd;y5G8cco_P1SMj;F_4U~Uw}g#Z
zgD|G~kLy;iU1*u8Qu(7pyTZnq>;YBbzLRfF`8)?1MZ)#;Y39n9>m|{^;36s+J@tVE
zDk)So?fmoaNKwE@jh>4scITq6WUpG{`V&1t+j4wh>ru?dT#jDn<|CHlrGKTTxI7Bx
zKG%>s32iXPTzpgb{9c|URAfO{_Djlym7F*ucCgTheae}qJ8A5dh4tXXMvE%zVxUv7
zHs|e<qpsIP-uL5tSnVEk$F=436q|NNE7MMTaC%4AVdXX!X6AN4+NgG$LoTJkTZaxk
zRCEn^HvZ9qHmS5OA_sK;wfp&%QACU<BJw;hxlLzn2wGzxn8mmEOVJl3(n9kMw(4S@
z+kJ?WAOVOumZ0a%mJAag0}!;3IA7pjfq}J!;m4=i_UWu1S-Hr^=i#v#k$5mCCkKqi
z@yu(_(2RXL&cr|#m7lK?NYJ~A1flUHu0#$>J;3vy57+>8<9^QMAUy{d>Eilq0mrRW
zG0F6t9KA5|Fh4)>l(idAbaOu@H5_KTg_M_lCK%DUH{#7iFgZv(5Hw~rSCnx+udd3)
zHSWK`F?vbA@x(Yq#|w^)GJDqLru7jLN}ywa&hdqSD_BfKET;pZ315~XrV0`<F<)2K
zP!>OPl9QFS29gpHbK>;@I+$h9$y<Bqj}0#*1_cH_lNgd)Hry9Dkd7cAD&-M;^mEJ0
zagKxgPYR_|FuFsYk5;wLpbKSaPbz9#p2`p)^ZN$m*M;{`LT`*yuVe3d5;6O=zh}>%
zcU#@8)i@Q%Kn5Vwm4C)FDE0qn0nY4Arp+!EQ9C5G5ok-Db%$Q-Nanc3nsM)!T(yc+
z(cUwG=jo0r9px;_NX~Q?8E(wq?Z9xWB%QNB^x$isVxIy|skATMFKz9)9O||!`c?)j
zU(tG6l9}ZnV2qCX$n?xP^Y4&TAZ=dYDSGf8l&e?F1T9TxFHTK!oHJ!(yKp${uNX7@
zC&R<nW=4;$2)&)IcH91Y4j!J_G}SK}YHk6}#YVdSkiBF;X~8+leEXlNZ%q1Pw$Q59
zTHCp`k%FYJzI=ul(U80jk~y^CIiy!iLfNzNW7Y+5hbF|V*_`Fz=m)9~-@nQq2h3zV
z@v3&w9H;E?uU}`MooWS^ucCbAfu)!{sgMuQn+u~vI7XYk(c2O+)a6cFjrJ68T8i6$
zPEKZ)660&D`f6>ZS$`d}=WWc`aZlvM&1{%X=9eL+*sm=*1D*>>XQNwa7)qZFT^kqa
zFuY=~uXw~~V|<hA6)8En2WXo#^}P3X@QrZ8jDO8jd=Di;mxszY6Iou8;&`H<-Fn1C
zc~Xdlg(W?;tVbm<>Ex%_*mt?9T)rrM!PMUnNCz*|`1CcHO|XI&t(qGe_8&N)`y{|%
z7H4HO`ydrHwG`*4h8k=PBm!c3Ok=TI<<P~l#T$XSGKZheevFH&(U=OtAGru;K4Br#
zK}mJk#7nC9b0w6kjiJ-4V-3gk$Ac?}(r+FmqYD=9V$?iY&942`*D08Nwf&IyyO;aS
zK4?4Ii13_@V@cy}w%52$*N_p$Y$7S2DyJVEICTrF#<6kpjQt4LS+35Uzxz2K?SDkh
z`Z?@K$pJgfFv}cIN%|Ldx9R7fPw(8dYgOYk2u~7{jYk%`5KN3SSaNHeCo)I2gwrg+
zGn2$R#?x-K1w|0fqs$*Eqqc)}pPxhgHJ&w2_{qVhzImL*nAZC>@_JqCI($zF35ODS
zV~^rf<E~kGn?dggE7GDGpF1zS`Qu+u+j31^h63C-XChRUOBrJwp8*R3tX)ijjGYEV
zoB`qZGd5OjJa-V&TV5R4P~IAg&dfAL!vx?OL2tX#4pcy-Ten~H9AxgOn%nsyC-hyV
zv;ako(@-uCE$9Zk{*C<jH|Kxt+51w!V)E#Mpf0)Po2s=lP8)NiGrJi1`MaR&D44)Z
z#M%LP#=Eu2l8<}=le_w-yn17MOunoj2riqQf{#xN3x^f%;G~C?tAnHujt14ZLir=C
z8|{kFYH!fXjXwQBFKGSI^2IGi>%ps0GidlhK^)JN-ZDOe+=LGr9dCkkKBo3?sBQbd
zzL-@);_3P#3ex<e56C{ju)HA1LN<)aj*&`PtYlGzg}A_b8n@YSiAh?GOG98%@Ej_~
z<XuhqQPI(ee<F+q2M>x;AN@io(l)w6U~Q^BeRgT^#iDTsZuR8ex*V+1?zCUbuRya^
zYZPlulOG<Ypi@v#h`k%5+%f!204-(qp=Ib)H#SyzOk|)bH&V;OCAyQDwCj&8B(Pgo
zyc|adBQrWoRg|Vz3-&uqeq;JB&Rj&MMV67#(Uq}5#l|7czPR;Bpe(a^{)HUr0Y2(%
z760PojR@xLPcc7w^B$Mw4ycVcM@C!|d_6d9^(k{Qvqt-;@`n^fb$V(mcZ*+&FBu?e
zbPl1odTFkYs{4FD`O4xDA5){U9!8@eyzBGjco&AZ1Q1Zd1uo_h1y!u02;{xC_*9c;
zlGSqF<s|Q-T1;fd$J3HHTS2Jha;i?81)@wdEuS_lvUn{VF+KaHW-U@tW$!OLpbcRm
zG!9VwZeqqbi5O}<M2x24F&tfa3&{h#71?8jK+Nk%3BgdML>9P`4{Gsm2Dx&zby0nR
z5u}Peb*2;jnB|`z(o_D)si|~)MjsG*h27f+l`;+n7<%+~YrJHDmG8V7ovbx<T-kbq
zWG~|r6ES6b94ZL-uwbl>jEPZ1wvC_PBCZj%V^GJhiBRfz`}!^*e8t+%?t4!UIR+_1
z!U9$pFdk9{fVzNi))gj(2R9y8gTfau*+7xi1(S84w8%xdWWVxk_wF}PR}(v}=C5?9
zfVbe_UEoa>C<`zSKrGBg`rr*{w?V)5_xB@h0G$NzCPZgd+_|%Xp|~A>^x#zyh{*oJ
z6TT>FeZ?wF2(hrRrDTzYe49t{GfhK}20S6?$y2e=`b0c|Vn9*`JB?clqz$a-KSo(Y
z0Y;|v>f&@h?Pe1uXr>QXD*0hmbGWdSx_E%(N~9(4Rnu8fhbY(#pPe?DlY9^N7p?*A
z74}iIjR+aW0UBh5TB4w+$oEpvE*=T@6Odd~3!elfj44;+8jfsT6Y|BMuLDRy*w&2(
zhZ{F<zB4{`A8Jz6ig|b!Yji|tG^(L?ez+O%nK>W=7;P(>Oemaysyrm$#uQt|Uptpz
z(E5?}h;oMhDm<WQWRY7=xS;TN@DhmePdq$ew2?@__NAbFk?G6e=j!S=$)`o@TUv;4
zp4`VlNCiX!2l#H|h%Ws6AioZ{cv#@#fzPK^<Sw#|p@rnf3_4ihpbo+@krh6QH50GR
z1+4k_$}{vp_wV-nMnLk#tVoQ^ip;R*Sm|y5SETEk*}riGsjJZ~p8M+e_ZsA!?6wG%
z4zW!pIpd(1qZq^*lwJRX`?dQ3OW8tNbB$h=VuwQDRG`d^NT*7Yn6F|%vU~EAmO_W2
z&Ns_~f2j1RQogImpfnW}c+Wsq79b>XYc}($>Hovtn}+qgzVE-*gq1l-M46@0P%@NO
zh71|fEDa(wP&6P^)}qLip`xgi<~bFLYMDZ5p2)OFvy=uB>i>FM-{1e(`^o><``PdK
z?cI~Lj>@O^aNqZJUFUV4=hb)R^kM7ftm?7Py$XkAo!?jQTHVq4wzE3ZWT(MSw;^tF
zljS5QODfbxX4m~r^Z*+W46m+UO+|3j)U$z8gL-a)hErE(r||_}i2Sf?C*Jr_2R#8K
zSF71O7P2ptK4`a;7<~cu<-HxOtZhDqS$GG_16RC!>BjW%cncRcBcT0chgFJOHY1s<
zsaZ07DQF3v6jX1-WUZI@Zy*t|110}CWRmVL7U_)JQkXgOjHEtMLWb0aYWVigZ&7=J
z?NdqIWA~()_fS+ZTg&h30oqt?+VUUUaSxi)v72ju8#~4lgAAM~(OQ)^%fJ!93Vd7D
zW}+Mg?1m`CbNZ~sr0j4&PgQn|(7E~!PImikxt9LMjgojNSzSwq7M0W2yRg}eZjjKa
zCIBq8`v&uzg;Y}^EX{pcGGHt}9$&~JwueLYqe?claG(|a$z}u*_HTRkeb}~j+2~%w
zjn3NXj>G=u_U%XOCHdKDA0G|phT(RNA802CGK9$wIM{hCOXHTVW&RHynesOx-LX_f
zADLo#uYdI-!W7_wSa*6@d^`g@4}3UujKd6vU4uuBA#gVYDctLvKI+GtwPOPo9>gXe
z^Xi&jMI|L{*Pos_TEkGd;DshFoG;KYh5Ktc2yn&*!_ZNE;^6@(q|gRXjB)*^B|mF)
zICGZr<7LytXy7b1JAVYj@i}-Mf0+H21P$9)lJw0(Qs#{-r27VBE>7NbI%Bf8`aXdJ
z+p>hi>ft1-;(Z%3onw4_hNVW1@}Qq?o@PBFHFMee^&|Sq)ij3#QIOIlzTQw7a{PE6
z383I5+Z+Nb=N5g@IRQ@_?u)9?$A*+wistx}#raE@@;xGyNX9G277IWZY$BZ20?sFa
z3G&rF&U)K6t%7pxpZ+?k^|7tyUP@hOXrSY4M@4Z(Me)8l`B9OHTISaq>kOpa4?Z%r
zFwk4+<lQ1#t(Ed^Vy;PU@OSkutxnTF7RGfNefw<LquTsBRjIJBN#(?KGXqV(FJT9j
z0zZ6>+?@cimF2*(gaS&iQl>&dn8#FKe8pWI2YQ`6`2=OA@zG`k^9dVLy*9>VqJ?5>
z5DQ!!9ZO3}u#6F2lYrUkV#um5uyH)bT%#1>Jvg2?_x^QGPI$N#G!xVulYVh?a&mez
zS{k5DFmJ+Q6zLWl1Xy;VOH+74g2Lp<G@?jeCCMxldtc&p+oY`ck+rk~6avl3?|<E7
zmvhh(V}#fh6=kHO(}sYVW7F~QZK5>xx$7^^mF$kiYOsl3FKq%s!LOR8xA)DGk+DYW
z3#US5?p(ovMNttjaR2un7kzy(C+Dnp5YBC27Rw#N`5J!DHFJq$+t}HHde?1x<Jy4P
zQ)kRr`N-%flKH(ff27o>WXsIc98Ra@HV(|Bo~~;a)n1*UG%+bI4r5?-^YqIc+LZY7
z@tpf}YV-=C6B&8EN}smSmnQVB!*Cz(RFm=ci=f=Gd-RhGI>w%MRZw+xEK8U_Y2u5!
zeeYBRArw1B4NRX2D~z_n^oF62^fTx!uGg@^lFbDd+1?A%cqn8Ds&(~bWY(B$JVY%-
z(}FG}aRy%ho5YVycO1HI-A`xVV)S*hOoD8CIYAeC8>CK0<us&r$o(!ZSE>3Z^xCy{
zL8?`FncyXiUWLQ`z%4=`U+e1Ly>~CkG-Gk1W(v&D#|eirkSM^o*%kFGvyErb$!95W
z1ade$dh+BTmSvm~Fe(wJUy$E0X=cunA2fVqj$o*%sU7gM5e~5r2#I)8eQ0jhR<w?$
zma#L5JAC*a%kZY(E<_5<xc2zx@1LNmgU;aqL6Y%rdxLsjZGY!6$*0NLuJ`Scs-_(@
zrH)*->^B!hrzK#Q4rusD(7V)yt{9-q=J5-Qj-EbU$YBzZtpBs-aDY*s*!C>(%M;yl
z?kqw2MH-+1_Vwnf=A1dH0D|T(M+v(XjWS8qM)Sjg8ycINX)A2ZI}tZ&pS9S;)+O#K
zUn89C{3DvI_h`JzvhSDQ**5KML)le!?yB~mU!^VcTUKv=d82KP{<{IO@iAiEGQVnC
z`EA>IP$#E(dFl1yo&Cr52eYsG{KwEcGh$>%v}jHh`7&=0i84`n;NZb0eSGRir89oA
zZN{PW9nY?@EFczLm~afTW!<rfplIv_kvSo}p2L#$DMME<7w{7Hk~2NT4A3Jp&uOQu
zv>)vqs?aa`d+a8lFCuM%D!)7A&&!uDk6&n9sr8I<8Iy3&C*8{544Q#G@Cbz*`FU()
zSlC;tqZcLO@j%*N^u0w#<IP1grqt^ZOVN9l^mbd@1qh4H<7Tx9hyKWs=i#UhysY+M
ztF^Vn@ZrS(ustDHj}b|ThjmR&vIj-bg!{t-J-@anYT%r6JXz3Y?YJ3hx7gcDVY~#w
z^5L$>w=ohDzxMCn9&&#Q9dk>{RgQQlMI(&g22L1A3l#uvrE4M%9rHUxec2DGH`tPB
z4TX?CeL5LD2NLN`$Bgc+h-Gp0sI0wYb8F|bYo1R!)mb?6BmXWdW4da_&mFr-W;Lj8
z_I_$6x8z$(1kb??x`I|x74HupEic#C(9l?9Fi-GGSS`L<NG@Mzy2yPI%{lPdzQts}
z!C0OOA2MgVZ-?WD7xNjWL?uRk?4z<WJT!EKsOVkOVHcD+7$71QGKutg2&~-)9^J`D
z#6u!*=5KhKLhjj@;BRJPvL8}2_EIOBCxWW>k{I{+M*xN)5UZm&_)J^#-9!K%Htp=W
z;`dBtCo-coOyG^x71X$E{eBfD6=&F%1T(okXk;*-4v&t0WRl_YxUy0J>O4#=-VZq*
z#8FRfV3KB?g1=p5wTCj~Www;n_+o|m1Et1lD@5(t-S$>aFUdNt{abO}@Num(TW_u0
zd&zN;V){&%+7`b>SKZ!Oz1!EKXJU}q*6KOL)WT)s)9bV9w#1(@7XLnZtKYUYx0Drb
zoLqEZ<qDU}nGsbN6tA1zeqot$S-wc)=(I`mH&@k|eGK>*rXIc2QhmT4JWiP0&AQlP
zXbhr9KEsAI$@H~oH#Rin!uOFumyeMP2X$oZ4_=O17taR(bPyOGoEtl5)~q)wWk<tf
zV>7Y3MM|);Cx07J2%XjK#H7JPh5!s=4<r6Yaz}TZmT(ja`{ugC>%yD^uX)bTpD#pY
zykLD7UNqJl@zR^L`o)CXWNjspg9q8L`bgv8Xw+ED*ETZR%T>}m@!K~0bR300EkJeG
z_9=Z2x(EJ6O2a5;Ye&a+Dw%z!CTv=Jh%hOZajE?E>psu}{%D<3#Bv1wu(07z)kv>%
za%n7`iZu=8<w248HrCb{GgBdooL@L{%i;L*kM7To+=K_dx3@RT|4qm7G}O)a?$vON
zl`7qKBn+6;RWl^$pXb*y0x3txMLfQ?sP;-<r)KEw?d9r5T|8N1f+UtqWy9;WR49;}
z!_O@HOcBNa7s^LD%RSSC^PfHS^Yd9VXEL{S7s~XTUKSrQWbP4&m5Gg-!;kp+8SLA4
z6BmZIA;`@asHwf67ID!(jX0BZs->mni|TCmZ{2j36iw2O6&>|~<j3Eo=rFFDckb71
zZM$$ko^@Rb`hnZmJtKzd>*#co2!qkSMO>NF?G+JGOZ>qbQ@+4zTou_X9<;By(+=ro
zB>SrHv}wp;B53}2UW!LEtr!WMW|+#2%U_>Q*4M$u-kN+{l<oZM(h*C=hL2LwdT@O6
zWSQ#{qc-ZluN%8jbmQuBJN<I8Q@O@lM6YUX*}P+xxTv=Ll+g!<?0F!%?6>e0%YXgk
zno{(#?^Vz<OUcv|?ixv(K0SGMPjTjU52xezrLQ5J+56gUFT!uPYjfrp8t)G}N&)_j
zph@UEBM~@SLIOvcJd*Clj@f~ug9Z%Byx^Wx<&$;#`{YH^-*#8le7u#LwD8v2&h>HE
z`s#WupL>1p=#(y5uB4XTR#&IGZn|%>%K`7%?Q^_#N`Bh<{z694gYDB=Ze=7@4a@ov
z{&Iwy{<;TwZ<E%#cje})UcbNdvC@+ZG6801vZmT=w@M%IdAjKCf_w8k{d|1vFy-w@
z^i2%^`}J!LdHK~{>CW4kbGzQ`8aDGchN4C4d!Jf_zrb}47BB#Y*3jqo@aQ8c!6(!p
zWc_;TX%w|xL<Z_oH_|>VRKUitaaDb_k0~9d29exsli|wniG&&su-Cv@>?@(CcK-ZT
zsjoYW-NLcl=)?y*oBV{VQRo<B;sDt+IM|-9#)l755)u;Q#=S_lGxGMo`1v;&S_OOx
zR;RnjJ4sbhsvy^6t~71f6c%Xgy2;46^{VvV4zugI6{#a5oRT;%lUWH@iok-idnH91
z^O^VVO@iiu&>5U<`mAS5J3d!-7L3zpv>4xr%DDId`VAptO6IzN<T19QjiC?+Y%62i
zLOE!)Rh39;zy@(HpGUUD*d!_eMs84mK<<YmXzp6;N4moSWViF+M8>~Q?m9VLrDX4#
zjvI?RoXu3Q(Gr~d`2i={0>CuNIhJ}m>JaE+%qf?a4xwFq#LEl9@~Rs%m6X=aE3_wN
z$=v4AaFsSSDYUMP7$*-ISq9IKutU>;vjeA7S;P3~>U(k6@OWvxBj*Loyu|J&G;>3~
zj5YK1B|XQ?-i%KH<>9*$%72%cjuJ5{0|rgaD4oTh7hHTSvcHia2_+2K*vgeGj3oUS
zUH%mhp^8rxb<p;W<`+#a1;~hrl=MTn_Sz%K<^IhtI!c3i#1#gO+z}wS;m%fEcUowQ
z@>lVKffnD`VWXG7u=V!HF=I05@gQ(wuGqqb3t46~8E=27yA&ILACKkl!<N2HS~O|W
zTVPcdDSD7hA}b@t_v;7VpZbsId$n<tXBrxP7+2x2t7A42VIdI%J-F~9+`1}+k7L$`
z=$%y@T(<9xy+?~N2&>B{vUy!s%Z=Ak%GSluF{t>2nj8TX85J+)Dhs>m!ElDW#3n}F
z-#87DYcCWCf`13+k`Uu(&01eEqY1iEgH}Y?&p-bx3<qKzW1AcT_YIW<-AHVa$=7((
z*0i!PhQ6O~J~rqmF&tZibzOOJf`18}MbQyj1ZK~LUfQ=)!3GX82PjQk!eGY7EIdIf
zxH=YxsHEfpP*>gtE_fOU#ZXF7`l1UNI%G)2_LYB@Hml(mJzrL~iOB=H2?ltjRRI3M
zmO!P!Mqz$@7B34&k{Po^Ma{b)vn;LmY)I-{FpQfsM?XS8q5MJ#PN{gH;AhVa0KpcR
zL!kef87fAe5!J5NOe7-#=NQJITTmF0Ck6+9pqOcF+>qn$W}s1d-}BNkkMErUS!?Sx
z8`f3_=?z<JGye9+rS;kGL_FU*-?bU=yzk4DcA4h0i;B8rI!fAF?nxTjNJvSTSH;=8
zMzxP|9pnAXJEiY_?yD0DhT?K!$G3YH9cjt<wA4y3F#a!j%g47Gt0lW%?Q^{Mc<tA`
zuG=f7oGVm%Qq>VLuj?QED3>7fxGyHp4_)sa^)09+NY7DEYMYwcuI9fh-o?evUgsD*
z&`$f&>fx22WIE0!TiM#%YOcRyG(D&1&D~+4DvLLEr$@gveDL7TR0So4MN{2h7r5OG
zNxLuN;^(vORA0kC%R{8Tr|quDEr?a=i2KLrPR3uCu5SF$@gnVZ+AW*CMe1sMXEdH`
zII+!O$CSAvU!Xs`KO~o|_cq=g5oNOZ-EU7QmFK^RjEZW`>=KD<Z@+Dq^ZnDMN=HIz
z7X!<WMpSn=F2IJYDzPthlO{Wt7;UzH&MgZyTBHo3D7~HEQdhT!vj<Ff(DhaQX{D{(
zwCPG+`yMp684Stro)P@ajTstY&rNYqh^_2AqG=wSUwlF^&l3LD;!p<FrBqfvQM$;m
zpSBZ9*~MztIIBhLMgG3Mt?+V;%Iwcf!rGYO7!gseCtl*JBb}mO9}21(7P8P(GQ)7p
zv}XA&f7{mfo$s^u+f(?QZC9~iO4!9D5#J#kLq;zL8VJ1$PDJ_g%NKkrdkjH0K^CW<
zobxEeqc=vS_8l9@<IUGH5l=zvXL4YT;DFuRt9<A}9>+zx;d$figBh+>xBl=>TQH2h
zGjsQ@7uv@P504ER@XxZvQUiR)AP9=7y8$<Y#4A;Qi}_lNGD^Sg!9~7(hZq*d`-fn3
zhwj!PCil^_$&;7ZY5PMRz+g^0eZb$^@O8EA7HA(l_>ICtz9RU7jsS>u)T{W!o879s
zIqHDLHfJgQ2DN@l67BP@JQ+A_SZi67Pu*E$9iIS!{B4W!r;!pmeY^>UIQnz<;a=ri
z<wTy1xAr7x<`?^}=c>v~&e`ymrX?=@rC-lQsxHWsb!XX1E8~rR>3{`xtnSc#&Vjw}
zpO*a9GUIB4T6D2%Kl7QZQ-96_%lC)WVz=iiDg9^07v=x8X20t=1o_|dK;YfTeERS7
z>gT!mtgo*#UCgc<AqU%6dJl5X=?BJz3@tk2xD#t<p>xBgJJWvJ=SmEZH%{5i^#l%3
zGQC+GzhREHN@GOWm?^ZcZ4~MUxD~tVgm(11z3)-Yp=;rwWGt3+XE}3SF!d&LAxKdU
zSU&sYyNtA2KgWya7%K4VI}W_F1bfJIn$_>oRaUA=P4mP{0I6M`t*`rfZ!mFh>vLfV
z^=k{8gd1%Zo+W^pzx&%SdnQ<5jU2h9(U9)*$b_z(O}{P}*3Vr3-R{1JIQzR%HH^#`
zd8N|t&~LlP$ey6CC$^0ed5}M|89RUD=?hNxZHJQyFFws#Zj;3S{zXQKY-*f_EB1x)
zUsL~Md)YqmznpO2E%~pJ?T?NM1EYkmGs9f5Ll(X!NVosT8267w@BVAv`@u*5<(&KF
z%zsUO{~!8+$>!oo|HTFPKi+<dYR?loRtapDlWQ2K8v%Q6P3Jz4AWpHZ_RL96b_Zh$
zd)Q(BPTlz1A9cC?Dp!6nwP~1<TfZBil`pLj9)J9DydF3z<X<wXU1yt^hMeMB#p}*;
zhprv}68rBvIrRVd`9IW(DbJX0=kDIi%6zwjRnjjUU<{1SqT>-F5@h_FV?OcSKcX4A
zMBvPnO204<zRt7PuXDr#f8{5W8K`7C9+6X2QGYJTYo-O0^n|bfe~tWj4FBKU7XQbs
z@Zalbd9UzNcv0Vc1k~!lr`1(gnImT+WNQb3p1ehH@Pu|`9h}Ri{o88ps)*9ko&5ga
zecmTezF<&AhAmTEzhmh_S^mUeB0pMv|A&2Yv*?&HHo4WL{mUg7j<Ct#$Eg^vO<+Lh
z`o!Ql6GxsH2)pIUU&6uqcP98bMA9mGgZcC3bpm+FGiR*d-=_cmyVDu{)6q0nz?b{K
zeSa;C>^OtCZTr>h%#I7^^EOT4z~1D7_W`o~O`8Ej6!El}t$lzWZqoX3N>@JWoDs)2
z1>ThQmjBE5qX831ie^Rh-y@8BPhR`u*A4T0D_(Qa_&aUR&E?ns>!Qjjbxtk1L83!(
zyKRRs?fr@8kDFN~&~&iR$qmZ465Q6~mi@S@tdl2@v&0d}Y(M?@Ao+jqu8N1TW9N_$
z$;-5~utV|F`f;%fegv21T4wK}C0bIn`I;<YN@cPS&ezGdUS4)WX*+bNV#JS+a{l*P
z6Sx|_*>gd9w$+M86;7pok^lJ@oM0f5l{w$Aa9Nc!i=WIKFGR9K>17^ywj}dHdm;7|
zPyLYhzt?33(hZNdYsTZX!n3XP`EhgAj63@TY3lxFrp3_Ro1MlJ?-w~ixGtGrI>&S~
zyl){G=a{j4qo3Z7An5xOdf7N;H#!GUmF1U7@RN%_Ui_88yPTchlPxX%iD?U}uMg^7
z7tY+jt$*Cl*2A5?6`*t&cjr|K|48DLa1B>>G!%y(rU#q2{MiN;(Utgr|GFgeYr$!i
zdd{G-U%K?jlNRA_cK^7WyuoIN{}KEV>guLB{dnu+ga>?R_SJvi`Ut7*!UO(e%;u~5
zvI;Nz`^qm^updRYzrC;&ZsQz{RZ_uZ?iqG60H?pLDBl-_W$+wJsECb!o_n@2^QU5M
z91UP`DF_j2(<w;V((3!IgruckJ}6zP^T;cs>*AOdCVc%5O$+{;o_`0lRe9!rx97n&
zK~^eyg2tBM{JniU<uP20{NkBzzy9_cgSbU}rcaq746`%2l#}B@vDCJY=^wc8O<n7}
zYnS!&>j@1@fQ1VR3+JX`jMPs_ern!dK%pi*s=xI+G<pL*M4?9Yl+iS89qILfe_W%$
z-&BKV=CX}9cRZUt+4k!{hlee-yS49QSf{s;TbRdzu1j>CXTjcM;EzOg#GlK4q!WKj
zjFQ#IULk<#5+E}8^o`>DXcXrcuk84->}Lvd0b#?ua{szhTUE$Qw)Xq6vBx~Eu6Ace
zA#`JZdts7C{J$IinK16o-%R`8P5oUsXu}y-6^~LOR0mw;W=4i^PEl)RiZInM36moL
z<?|_j@)0sye(a`jL{v2p=b9YQ1tG{1pbR#bx4(b~g}Pn_*JVsAbANf-ebi|5KcGB_
zj~;!g+@4P{miuwI{UYO+ck$vy5Wt#$yRCPBrq93MQ26Tt<HE}eD5m^XugCtaAVHKE
zf|CK<w}(9afKVZHC6A34hVN`jhE+h+GVPqmICar=t;M1DJbF7qcwRy!jgeZhNR60?
zCw|$q+-OxFAaEe`x)}Q4$A|nlY;td6pw{s5#S2sKRN}{?FYnH?lNmd(He~z9uCDK&
zo;%#w@eFtR&8y{jWSiWmVENBrBO{}d8o-|lm*Wf!2IZp2vubc+6kKstm8zN=fO5T8
zFvH;c+1=|F3Wmeas;kk8=yrzoeXG~VEDY4}$WEM@c=KjAGkkz+`Ut2?2Qk6Qo@!T1
zS%{LAUO&#>O-$o$`}}#d6i(P)iE4^6GQ!YEQ7I!?KSmG5=titUv&XQfjaW~B+Y!|V
zY2#Or2$5>qb)jkzfA}Lj&j6*#FDx{a5#uC-%s?pl06xk<(ByF47={Cd>Bfj9)Sq9s
zh;=(c3U=%_*c5TD?|79-S)Z!8BfIkGP=k43M}&gh?-Tg!WYHgwQTNk_mvQT6Md;EO
zeg`Oed=U)3ietwHzWI3ZqC~>XwYb`-V?+RyYVwe(n^H|6#H>5dM{ElE_nF25@IlKk
z+w69OpB<C3NZocSQ4j9mGM~N7QY^0H`Trd|Q<z@z_ddEAJeF=yG-PB`j$_c~Ep;O$
z#Rk56Vk%lBfT#4Kv}fEQoO#_4<Ihz7n{({rB&S2YdfmF$9N|l6!O1GA0b)KMS|nqt
zFYJ!3gF1xWv09hBg;w|)u$8crkjXQG3r`w&!_XF2SIiva8T=b%^lDq?CjNcG^P_P7
z>Eh>O2nCQ91D;IH&31Njjtk3ja|PWeI>YBgL+hzDjq7ZE7$lz&D~grr#QnD^G)kw#
zlpAX@ozKem{`a$JRg?(ELpz@vg2+*9hdmFCJ;<=WSFc`$4TEQdv^(t2D`O&}qE1Kw
zxSs3u7S#+*T)>r^uRSTm<>D#?;0SRqoH^6>@uBn}Y5IJERAvGFr8i439u6BNppahI
zHTTyHENirl<w)c<6PJ$?Sw^O`)Q99yUQRAw?GzujZ{m-bE>TyuPNStf=6r*H3`658
zQ7+SzjB)u!lNT&UT)~4I)wPL80bejSHI-i=l0+{fbJ{yuvE!U#-cqzlFFza0?rmGV
zDO<>GEPYqiLIGR~10gk4poF${lPxTZV^>_zTxXWDdD-DSNlfSGC@QY)48>rtVL=>o
zeYv&KZZdFT)f9xQTmSFXy=P1mX5L2|marva^`cH!-r+O=GGA)E57-9)v$B2+3_Zk5
z5VkCrT(Uia=Y?74K5$i<m-6El7~%X3c>)I~^Uiphq5&Ch!jc&GV@aHx1<!>u>iN^B
z*5i$aoRwU3@Mhf|PC@`(#SPb%#_a-p<50$50@B=+9(izehua@>;VSSt@lM-^-NyHg
z4fXYmHb0;~NE+T5yDBz`ad|4}O9@DpDU1OAhlz$y30Jg)og5vR6tHCRV!R+fz&|%>
zAtd1s*Xb~a#iFh0eeI9U3F2&`OeAJzMwZ?)!r0TJ)gw69=PfOK_UsF|aAqE}!}{(G
z(|8Z`27!l-rcPU-XwldAxOQ*Yupw|NL@xOqTD0wFyc|3V(Nw{LoSr7RvSxi*KMV%8
zxgO?KEy8&IiQ{j}Z6^E3NeU4`#vr$Lx2C2h1mm#nJK@n7<NMvisU7M3yH$P-s@%Qs
zf<`u*LNaW0px7>!b~W8_r$^W<TBO#cSP;ne6vHb))|KGW#ybbs7BgE>SPe<9>%&Wd
zajI37%zJ=gADFU|f2TC~M--_MyuV`fMSOs|NXNAIsm+~xo#_)yDwz495z@WY3vVsX
zc6Ykk%LeqU%r7bm0ahXnX4kH|YAL+be*@#kUi^}HE+-Z#Gymz$@cAHS7cX`Lxf6;$
zJWW}W<@whUSZw&U+tFL$cIj`;N`K&m!NZvOj#eCL#1{OrH*Wqy6QeL-&UE@dM6=n-
z!wA<WH+u-xz%oO_N^s2?J?PCsJ2SYdj<&YpZVk}AtNd0!K;)F0i)&UI3=PnC8rdkk
z^(%;|=K}&<Y3O%%cY}^v+}x$P;JP64;_o?6)}HcF47*tdX1Lb}?H!B%Juws}`u=_U
z^y7_7o;{l{C)Z5CW~vsHq|ZD<S}I#NvvT^G_n#X!M}E^z&KF_cg!YYp(^KOlA_&lk
zN#-b-F+bKkzuWJBJgD{GG<*HA(s};>%^($@tO&_|df@6CGMVxG_e{YnidI|OJ}4z7
zH|!NWC}j_+<Ue_W8B0vW#-E%f`N(x(AZ90EuKn%-pBpOsG&4N}nyJ#izkGo4Z>}!}
zdHugf^!}$pv|EKOeoFW%z+_zErT>NXP$y3Hr|C+vs;w$=$QK?1jL?Fd7oZLsGS?qE
z2uD4xx0WXQg@{WXr7N|yp?{Zt8!a|sgk_`FPcnDBhDh!}Uy;*m+o<*Cfu%!Ze@fp}
z0yHOxtrW#Z=7|$0BC3CBz7Q0Iv2ZUDJ@>%_*pLyCk%rMap~PLfh0R1DqC5$m!bZy(
zj|&SI;hasXqG)>a42{0am*aJ0fKD*x8A)Frt{DDT8#e~n4*?@MfDVR;o7MA3$a+tX
zDZ8Tii|g1;1*p$xFn6*b7|*)CYwdjM@B;^gu*e8n!WGg>1akiN({#KFIAEDR*zt&|
z&kE(@zC-AtR#OSkciBUNk0uZ=7DA5^^Ni8;Pnr}4`IImJgk)a`f7TqPzy&$5LTSV^
z4VfdBd11`SBgj&ad~S+M%B#GH7p@zlY)Juir2`hY*z!O0inrSRVs5p{_cu7w>HjQ@
z?lr9^B;PC3lqQ{yRo#lsmcM;|egs`461G5(!xVWaSHnEX2Ev<__ELt37BPot_?sC*
zG->>{T89b5^ynrhhC)|8mv}uoy5W*T_Sv-$;Wi4j$hU6?-KuRqiD<7xjiIma+oR#9
zNLxiPyL$Y1R&+beG9iH#q2VpN`z_^Y+S?q44Zlp2F5j2weuNg$qh}|1C1e>wO4AcW
zQdFpv^KH`hdz%D%&h<H=raD$A^;Hs<-r_lDions-?JP>hi;k|2Xwjq|3sUziG4}~F
zeyeH;lM#JRP3_EyX$*u)@XDjl?ZAPU&^hpgNTJG7W40wF`@Vo14pu{zQL40Q*RD4H
z9wzTQbZSXG%3QYCU_&Z89U_N$+&40(HD5JPIrsItH5zFtP=QZUI1)-d(6AMgso05m
z+^oO-AbA_0B$Jm(^wloN85I`q-}dBUsJ?7Lm7?)g?oMdTORBE5TYbbIG6$6_L`Pkf
zgR2Wgv>y}nSc!AUQt?h}JV(a?kblG{=ZI4bU4YP}r5w-glXjG=U1Q0!8OE%yIPr$X
zMH~*AvwbZmoAYB{I3^@XkhKB=0@~Bod|YhmDilLKBfJ=R-+2A+bDVD&q=vh`qgqu>
zJI5OI(dd18@YG*^1Y|>&SI)Qge5O`;Z-&V%F>}-3uz%Ct)2*VDed9Nz*7+Hke<4ZN
zem@bh#&tBDaBoX;+8J?h)gz6BnAN<j;}Pk$_HFy0djtr;4d2j%J&jR;%i%`8a;+af
zIkt!W<I&SNeNOtIN*Ptv8(cM1-ICOUtR{M&AQO`|dPYoPE+zYO)~poYCJWXzuV-^-
z<6RY1d{tN5t>1vJU^Y|GT&~$TIA~yu4O<yp1R4$9Mh0SbVHWA$l3rCFHIBKd%F5+8
zXNv8!K?c2*6%e?b$_{qfspuwV2V^<bpAM-0&CVIV8x#`sN(~!&sX)dumwET@lpUk|
z&*9SwjgYxx@xDzSz!s;ZiW=U%Q?`FGuV6B13&pJJ@>^}!1US<T?&Z{p6E!<laQfiR
zcQS4#iZfWGb=ITfOL&OX{(Tb8dEl_=3l2t@SlZl#X$*aAmBWGaoRuS0h4Ro%=!2i7
z#^<0ZN!#P6P69=KX)tWlj*_qk>=SyIFCd-5-i5e;t!#ErQye*-Fs_+4?0W9A%bS$=
zTkFrAQGrJiU9F&j<Qf?0=*&cU<fJT8$t_tUDn-FDw;>)HBIFXIc75Yj)N$(652bM)
zJPwWC?yECYZg*8~G)cAQoHN+pIMG~+TKY;7(4@_HGRyE1Vfe2DC+;+rft5{g)`8`R
zOG(dne8=gA#!+CIw0iVyytz1!9kA@Py7B?fh>3V&!&|%?o?BVDhcVsoQ4ivf$3FVk
zLYO4-b<{d(0>}4ehP00ZGTyW2x_8Ss=H^tMz0jh#mthQ6R#td&G(Cr%L}<XHwo&#q
z&xHSvot%NDP2d}x_WF6N1w2LQODaAbw8s)f<D!t9o8H`Y=iQz=W2rV5I|VHt1;YE=
zAaTh14A^T077@zC;CVPjmyFoOX`zg=rt9#qMZ={iLB7#~gIkWxp(!k!z`*?wCt04?
z;4<~EV0XHewXW$8-=v|KRiLX|3LAEHy!a-Kqu2iYK2Cn^_b&zbcb4v4FZ-<Mj;Ig~
zyQ=DeezUBy@4F9`<uryxb9dykB29Obuk)WGC)eG``Z&`14~sRyU<Y;j-OWhPXX_h}
z<+i3D`~<({er|5l0(kt5)KA|i6}n>jfL}`EH#keqYy3V=AEY|c#Sj+&gjQXBD+})a
z^>x1W&?V7^3Ds~KyXMPHo2E6V?bL}AaxHsIp8Hh|Z1Rvty_@2^@oCYW+1cH12PNcu
zq&8j|5@x3{Ae}dPBKVQ}(AgskJs_-IElFq5m^$Zq+N!IhgF}(Q|9--)?aqJ*Gs_tz
zc^Uc?Wtb*72RVXt5ZWxDWZcigID)cN*-pGgZAEMSm3GPuyC6XIh4<X`&!>dkJ8|^m
zZW#sYHpo(}egcO*cfj42Q;Q3X4>;UjhD-svMIhvos^T{YP0)uAqgc>zQ2jm3rli8b
z{6v)sYYg-Cde}V;4aT5<(9YHbD$n4yZ659K<6ogMXymkGA;UYv$Db<`S2B;UH$KPz
z(VSm>d0q-{7+`~X86p(zm@)Hq{X-v#{AKX>648+(N6Q-*Q`^{9#Li#1Z~!X+J|Ua3
ze}UtESbl&HulLoQV#`7&W7k43<gEZADNmjHpm<t<H8cm4<>n0c7^&*TAf*jy)@Ns^
z43xv#_WMSMtO}TS6>_cIKx?b?xN+Y;+-kLdcI`o@_YrV{ASQAHtjX#}W4j_QAu$if
z6Hbq2*ZuedI6HS9m3GJW?KwO>%s>{-pD)nvh(HWLe<kon(vK&Pi3ks;@HT3`5)>3v
z`Svh`79vv9tmU>KniZ=Y3(PHTWMpK9pZ<&~8$x}1u;#>6<ULH52``#ghVyGd8nDE1
z?qW`24380bHn{FAH7UKFu|^vM&WD8L*jC7P-QaJ>>V*EaXgYqj-<vJ^bUox&kBL9E
zL2*fCBV^d7=5(?ha=efb3&35@d!n$tjsg1f%M16~gDCS?xkwNaa)B%+il`g;Frwx~
znoSu%f#8ZlQpL~`q;2FlDbY(}0q`dDTe(zIf6K$zMX$&9T%|a+fxW>BKq-tNT>=RQ
zO8VaJnqh}!(R^%%RfrrXmPZRqT4a=LIjeH^Y)5VminVIm%4wo|q?R6?Jmwwov$*eC
z9Fg6g*@kIEC2n2M<!3qlXdOCN1QdblyTJL;u^YPQbz{`1-mR?O$tB%pm%M%RrUZ5#
z4gD9lUG<8_LiKa+esdJ`R~c}bLxQD`(uDy(5WJy9&v$9^ol%e~n$y>NYgc0h?jH#U
z(nw(7ZVnI5wkeyI&@%Q3@2Q&zOTHQF!ocud1e44UCNsp2lUEBb4eVoTWzMcQcQqJ+
z{(UF>o>RTQya&gsls_Wm6(8@2cT4-9`XzLFv$QnSr@aON;2Uy&c_22bc9pBs?f?$H
zKG<>UewmGd3q$R&NG*wFQ+5Knt<;a`J-Pnvglq7JjXJlBo&H!EaX%<Oohj-&c;1<N
zUK+;mOhIrO8ck|y@O(~6Z}kcE@hPH#FvSFY%Xvh+5d2oA5zV)AaiV_K>Gii-v&+(J
zt%1upE;qhyhJdLK5`|pGYz+oe7c9lRr@d)?0~tMhUCH!c36`$Ri?8(7I_aamYSn&V
zhgB~cY0=KXK@k3Fn5^Za?mW39Fn*RN%%$eo43e4B3LIV`ir%JOP^v%y_zBjt{P(;5
z{pXza`3lDFseyL-qP1ngp(Q&x^M&9&+$>-#OvY_7YEZM9hg3s5<kqJj)N;#xIKwXw
zB+xqODPyowl?^=n$4cKoJksS<k+8Aqij%{Oo1u%U?%bI;fBpvJTwNn`Co?bQ?LYg^
zI~K2bxPR<92q<N0j7Zt+RxK?dA!C+P3}lTwUEXXbP|mWWrv^TJNc-RsZ0EFd#29#7
zE7RVw?|?l0JJ-a}wLu-W-e`zPi)-(<jQf|bfpj;1iY95K0klYT4M$ZaXK~lQhytt;
z&7XQ<+O;ag;BEIj&&7g=HSL6ik*6f21yLBMgh<tbVqH*PY4>i;uwW3&fki0T;&o%b
z(G+lg>GvP5b-2YTn3}9<UsCx@PZ;C#7%b>)ux<ORkGpw;qP?cjsKY$%Dq^R~6ZG%s
zCL}B0riF30p4paB8)P+{>Q2^OlOVvu2@Z4H@#9IZdUJQE#lerB8v<je=RVBZCm}tc
z<CatXS_%^6OiDe>Q|v|FljdJqTuZu+|4?bVyQYarHLzXYTM^S$5!ZObXit5mM4WlD
zU~j(r9BhROaewEERjWkZlu0$ISOTTP(wItN2M@gTUIB;tfFWb2n5wTS=X+W{QVDh|
zUXs_4JVvxyG5%tny!;L<f|_bi9U-z(;4H8$mx(xi;5uP^Zr|1(Lmw&{@m?uF{#VCh
zckJw|mv*2J%R4x=v#M|O$dP4)GLFg%^UlpuQE_4D3q@qlYhOdg=!py)HhRq;`tC3w
zfBL0=!+z`&9}2@(m2Gsr)xE*mSSI*^8}_ar97sDCSTAwx$2Z{mwY3IZ3BggxB&`9h
zbZI@pev^GOF*r|4ZrGimcK7`#e=lO%z+r)Qs@{BTA9-Vw0}}9Az-)T5<$+mYIx^%q
z?dRm52#Hx&Raj%f5o!cPDY4n^ET_6Fzr=0s+4^V!qL6p#RpxTLlb^z21HabWypv;+
z7ZaFR@)s-Z13>(J>1A~6v-^4vH0A^nnfVx?ME|D}o{MJ@svNSvXjNUs-s1OR!!86%
z>_%6|f)?BZM7Q#+%E-p3bG#Srhau=vq-iIXup3o>luJ1M8-W%PCYw-r4}I*E50%X}
zuos1Ac+yHxFG7O`Y`y5`x3yT07DfbIFY3FIZfRIMv4v?(mALevSmAO7s8=-PGfg8}
zQ0E&W9@X&Zm<<*~u!L@CkZLe??hZ5~ZHV_VfLj@&U9b6oK}Mak!bLnMO`pD<bUHBT
zF{{INXM>fDWxZy%>QIRzYD2d%|F9Uf2t_{?zwl1jq*UA#X~I}j^pX`1rmensCOBBp
ztO6tgU>-nSXiq+#9%6cHj{->d)Tvv61c2$iGM^r3&Dx$W{kyYJiqW?<upl#cK{o_t
z90M5{R1n(QgwL;ammJ7?^xw9u@?5+d(1AGwh-KzN;3g?Sx#dA&0*Z>_AT-+{#LrI!
z50S?Goa7WXH%Yc$K@6Wf`6^a!tPuF3f(fTV0bqdi!L~meBFxI*Yct(7T1N(89IL?|
z5DQW)2B@`>Z|L~Q7Cm`FJ%DyD)_Mr{OCGR@#!3?{re>mQI}Za?pJ_T|`+WkN`0!7{
z|IzKa&@CIVcAmevN=VhQ*g#c>M0Y!VUdDFo1(7pA&?#{H?+j-8Bm*HM`&MNE^0B2y
zW_?W%s&?X9Y8$=)Ot@?L(Ss1;0>LlR!!=b;)=zlTINR`ZQX@p4P-n3KW1aRA5g0yu
z?DT$}^?jdgW6f?|x3(;KJt^TwOdan2qW$w{52_^o#6prq`oGb2f6$W`w41;HcodY(
zaeR5(HMLIRGV8K&&c3B4s6Z>vQte-pD3;K4y6_59(<SSwZVwCHUN~t^nbHzNJ2p(y
z$9V!*gZ>K|Af&|1WDQ1xA}M(b6Xt~G6N==$21DMSz;@#FX+wnz?yLj>c#qXzX;6n$
zV5+$#Peh$-0<`*084gn_Hft*v7;e5xT~5x9Lza@Fv}?!HWX-Paw4NTap{AkYudU^J
zome{Lv$3-&=0}Wgy()0+$x5Rg6upifg_B*xAxUGR=+PsLUdoBk23Nn(o?sH0T#-Xk
za>s;k^;e_V`nIREF00u2&rnT7?xDQ<?b5@M4h|$7#QPo%VZa^<Rp8~z@@DUa#_&*O
zJsB~u9WuVFbp;YORLBf1I=jMU>n?7OjwaI#R2MG1MUq2R!&2eB3E>89(tYodF29{1
zgb9^<+s%@y{i#OAV^LmDcz=RX?a#1UMd03gyJIco89hw)2oq$0c)0ei*41l+)i$K=
zeTf`gZ?2`@h~Dp<MA`l9Cj5O4Ud^x`*(DtY22h00o;x>5J~AWjC1J8NCJ$j76~^7o
zfpX)?1-Yz0jB`6T%(?4-X{$jaE!9|CLCL~A_QcIBV%Eq3S}1K3yY7$kBf%JS0tLGD
z&^Y}3uQjB?5&s;Ql#LZlK;pKg%a_O9I+Pjt$R<4s;Sc-7WGhpKk<6EGrJbf|LgIh5
zeCsQ>&t@fFpqwfWuS^YcU%l#OL)@L+^m?P`;)$1s+satZ?X0#-^4sU%aIk{*2t~Qw
zc%c)kvaZ}CyNO~z#Y6Y7vQ6pI&Kvo@%RVkW`cU#?fJH6Fe&t^kX8Tn-KN96{@5Hts
zO}50fGILrs*{649()^Ld??(j&HC*xI59V|fE;QM)KpFjo8|P@L3hGa!eRI%Mb^Ur!
zuP#p3Z&laPgDL^rgN&`RW9wGfl81Zueaqw=SZ=n2cX;GCd6!z1s;!UYI}9qG+zeS_
zZXVpTovAF%RO~K(h`|rp244|R2vr114ciz4I=}M!Qrh8}!7?P^LN;5iV`?fSlwD5V
zYtN1xniasROEzA-_^-M%{@`DO^7-NF@z<-LKX+ikrx~!f4J~Ft81yvyqag0OZD{|I
zzZ+#<1U2*C?)}OUMtctxecDjlu86LXq)ggzZuULL17}+Pm3&vU_36q>K+cAqFnz-g
zjPQMQ=gu3}k=JeQ+>xh7L@@s2xV@KC`@5AT?U|91C+suKDc4Hll9e&FGwTkgC_DE-
z^*SZt{UM%dxsNd`p(`FEHG6Az6B;tl`li&OJ{dDDx*jZ%GMej~poZs_wo$GxLX^6?
zNLg_yDf7(+FuT|(C0*fxH&!k)QcbN^TbuPyrgfE`bfB~v)hWAi=?N086=pj~TeMX7
zQDism7rez@N&&o@EgV%>TN_gqMell~{blElx;0%<pGhfJD9@NU@y^SaW(JN#+`B&`
zt))T??}`#6h@wTMenJWh?r1Y!gUX|4>GzI#W|O5!h@2@6K{R2Wfe<a}c9sPz8d(`5
z=i{_{_ZLnKgTrfRLSBtqn=sYgdCmAHCOQFP0>!ehA-y=1T=HiWNmH+$%^YPP&H;qJ
zcU_J-7rdhDAKlQ=)MOGfT4FXh(H-CKlgK3^<Eqh;b-iEVhw30kT4k1x-OJ4M$_&UO
zDqWs4k1T7SSFkr0G<#a5PSo1*mOt4&s#%>ec*9Pqz4WqO?&&u-ORkdZ3k0-GJV#Zg
zJNhg5;7NNY(pi1K?k`_{UnpzR;(hE`u|W~-Y$Tor50G0KSpEt9B)0hrX#14F@C?<{
zZNbM~JLZ*q@h%-C9oLXo15Z3{j|KB(a9q$0iQX<{#F+48%O#d8ksT5GUXzpZAJz^_
zU(~F<qOkA-)s%_Ls&~6-pAxE^2tf?DA_O5IJ+k$PL$km$2w_^dw-C><ul&)v&=I!I
z1FPp<rN`9PhU)0Y8MctRn;0-N%OWj#m){?LLRy|yC#*aP_3A8g9nsGnWVzvAX8c`Q
zX_%UCwMuGw=9xn9iZvcHI+sd)eruU6`^14anfXaP8gcdh<vR-$lN8}iX8O<9krAU2
zoU!0mNxIlE7u{CZ*tOHwq~QE!Hd%+{x1!Z^k<v{?4s|0{gE=)LA*@B%Ju}<By`Fm5
zxXT3xOl~t>1eyBk^r%air*9i5^RBrWDuHoD6_un>;Pnn8?<3Erd+(9yntu6LWd>$1
zk>fdy_+B7%fdzaM^LStqU=xTnja2p|^`ib_FV}1PGQOTh3IR@;;*%#$&MTm!0F=hM
zc6$!*8m9AaUfIylK*X)+jXgD1r3X|@qo-kV!e;WW$fDKq>d<lSx*BIjZf#sO-(|c}
z)fPG!OT3$htuW^FPc-w=`1*cM!DBzLq*1-+B-V7_skph$aO;U}e_W}*9%~{<6ML&|
z>kkAvj=5$Mcq`)4D_2}ZnY`$gxLl=mZ(ymbj>BJ<&z#vs-=guZ=%r1F^ZrS=@k-at
zjRccNXE&<5JzvSd0&3q%rDCFl4}mi}`_=WGmz>Y9p3<ZAz_;))beGEPQlZOnBB@by
z{CL6Y@6AWZAA-~ym@U@L5zu@;THQCE+02_L8m|qTEF|PL1vQDJoI&h6#$=5ck;3pS
z%M?%w>t3&@hzRes3yHBUJlsplHes09Q2HY(z5C!G&KZ3@%}VMUWpbdk{obmH4~&R5
zXnfPyT$%j|Rkw>TW!){^<Xk^dNy$M-dDVYafl;;j{=kC#G9Z|f8$WJ<bg=T~FR${8
zUH@pD=yUiB?BP{vv9I>3(4@KF*G@3Hun(zt4*NGY3AN(1EY-gZVhj@kn@(_50XO~j
zVIZ{MPkf@J<R!}3=<K<*wr_y;Kjf@Rnkr29%PQPYB&d*{awKbN*UQ^N{I!^BR-QjF
zGCW)u1=YO1U`$!=8dn`=G>h)BTOFTDLDKZ+FS6XDm!im5EWieiPH6l1@eAtUw4sO!
z4}R~ey1UjGx9s~1tOJ2uA}TKJc5*5UG)ru6D4E<<#oD}Uagjvi#3yp}V1>(zqv!i;
z_NPpxkRnUpN&A8sgD@}P)&OuM4*80VHf@nV4k*X`^@n{Z4$J7cFs*C)NFiIHAhXSD
zly>VpyX(S95LiJ&`90y*t*`7KnXKvQnnyhD(gmuEePL^(Xl7*^dDYdwp<eyOoR71u
z(v_3E*+?@}u2ayH1e^I(#u=^4T9_~+=qQd)T^>^!3zv|MG*u?Zt^Xr+Npx4DI<7Pn
z9kEGA2anNxtGwEU;>0o=)U+x1aI%GEG_|A1u)wFT7u8&^8Maze(N0w?j;w`QH&O~z
z@J8L@B3by4P2Wt;_rCO_h_acwpJa8kqNO3z{pMfl<+6vL9@6?OHEL$A<E~x9v#w5d
zst+aXwPt_)0P=|K8^(|v|71qDcGqZiMM)_LU7yS>1=f^Hs8PesN}4y(t?PcQu2uDK
zI5zuws&bCRjUgG;cPhIC*3FuK-LQSc{`m`s$*8~EJ#!9}N+Q+@F}#WmJrsf-G7oHE
z{Yabj!}N*@3JgBy#=qvAPwaV1#HP-lxpl;D?~jLNn?}F5f55b9|B=ZLl*v0|qRt+7
z+_T5NIxVKjE+In}=uB5h{W(yM=q3HsZEL!!_Is_$9btCR=<WHF!N{_%j8epKXNiiM
z&bqsvVGz++i_07w=^SqyA8T9cwAHl4O+w}M&ngA4l`DSzDdA^2F=2zkNXug`{HDLq
zqxYcA>tF+$r^#=?ijbWObDCfU#rO~3vujs0R7o;!<lSEY)V#ek1=?TTjZF7h&I*ST
z?8a<zG2g^Nb$IFOx;Jl9h|ZLiw6lN%#?09^P(y*5$oAA#{hcQwU2knBl`kDVddcq=
zPjyepCmg9u`ZF=wD6lSW%J<i=bcF~x2ReLL=ToKZ?#=XtOD}A(DGeHc2Thm6aJ}AN
zpe@DhJ$lBTBXDO#9OL367AO_(uTtKka(KZo&Lpo7;<jf>8`Kcx;dE@He!|*nnr2$Z
zOjMsGcYo<WczEwokM6e6Mc)+>)Keh-?$cc6I((;hL+Y;s1`JA_QZn{QVQy|DunV=C
z_KFq8>9z)B|D<X-F_K_F|C&WL$^OXs{OUkheN&n1nWqbubeaHz$7bIR<RF?*;yO@7
zq=$Yc>^qQnp`Vh`8D+vouFRJhJzA9>PS*`O-+)_{E+M7!l(yIZ6agP&Tcr7zHKPmq
zHdBC#PNh?~>V0!gd!AfV(vWJ4(rZH2AvM4mVIFCHMf?UQ-Z-Zx*kG*Y-mp%OI}sKw
z6n2715|uGR<QS9mSjZ5VKAq~{HOZ^g%&GC~bJ?c7YgBqyxS2W)N?5n(%l+P*_r3X_
za`e}fe%oapTNb%=bep5IvvNwck%gN<^I8W})60er7Y8<spKqqR#YH{5Xs0>}bcxfl
z?MG~~qrx`cOx$>*P;0=O9SwkU-#;79Sq8W-w1uQZCn@vP-<VV*Os4n%ZZPzSvm1IP
zaqB<M|A}pfm(~KYjKNXy#*G_?4=>rY=^sqV(iW|b6@Y)IlN}URDjM&HNU0e+-x;kp
zaEo1|$2q0`1yfy4N?yF;>c7A?MK;pZEB*!lsyyVBQM~bbP8nh^K1$z-WMNBErd*8c
zI0G-OK=t+DvC7e?`RSJc$dvUYB}#Q#ZLYrqTA3vgV(781V{8eCjp)K*fkv;cJ$04f
zvbMDvh3Xvq+Hxaop-t0fo%&Gv`(X<mc<a`$Zf7(U*_qZ`Ld^(M7gng{6RYAuoAHA@
z#QA$0tSGbcR0>vsDVsg5SI1Yp8Ig4|c8}hOgbkfHv{OZeGuOGIwWVb|kQ>+rfe(Q`
zS;j%I)UZox)@wZqi;ASx_9fR1yXM@e>y|jMzw7@#W)AP!<~uaFGo!z?qScnXpS5Pk
znP>G-XsgtO(}HlLBh1i9Vd@;rN{kKHzn#5PwkPKw(WFllk2DeR9_tet_49U5LTSJG
zP-bU=mEJ3r0(+3^-6bBQ`Y!Aj_fB40(|>=%%FbnT^OxTw&$apO4+2f4U(`TaGXsaM
zI+!P#=o+dcvvhB6IcLPEl3ORHx+fM#wTpW9UKF22GrFWE@G%CL=NlR{%=)e@9sSMF
z`rC{EHp!K^cJMTj&)<dfeOKOUj4|u-Z!#RBtBNd?M$KG3?LMDf2HU)qKY50J=@Z6U
zYJHO~8&|8pIdaPgrCs%*ggiPcI<2NK8j6AzG8pwOl9;lhqNKG!eGgv+4ajv?@9hiO
zF=fT1iNAZNSDu@YrdzI<^3XiRu%gtkr1q4!LqtM^ZJe%weoI<<(o~tiGsmXg7oGBJ
z%9Lx<udS;4`tX~3M{v=Lh35kvKM7i?Chf8NL)P}fCsQ)En>+byL8C6-o%l!N!eO-P
zfEVLv*Z?x!<nasGhtPaXar$lY9;xzIvnvXhoIT!N?V6mDvVNmq7#!2j=_=qV%W8hn
zbCyz8@Sm<(KT%!%<>Jbelxx=lm2J)?tI@Gf&xz_;zS7V9^Ztt0ldl?Civ%`FtJ+_#
z{;UqGMM#F*R#(i|0yK*K`f4!79c1WQwvBza1A_)!9O^nw{(|B!;cK#v=QP5ll_bxg
zCPG)EcOX)TTCl1SY>;s=uVZJEKh4D5LgwwUEZ%MLR-0EU3RfG3B$ml6o!&DhVk8E=
zD&)2m=ink-oU~tN>Y5I_lQ%D35T6#bF8l%<{$pBOY(vJCYbnYIrEc2VicwC!q#ycG
zpCq#uU**}HZfMPJxGapmE-4WYQ@HTh3O+|PC)BRhN31mt0B7cnRIZH@Atc#OofcGT
zHxaIdZ^#*F>O!a30b25J+NLw_$mi{?x1_<#G6kD3_#hf5z*5LW=&cdEs>;j0QPqK*
zwqt(M<X#imab=Tl;XUMM6_rCV{4AnX6_=l`+uL?2rDMW{e1+}ep$+!euk}559iCN^
z-+8-Al}?xQsKCD>6IzxzD293Ru#9?luK)UM5lBG8p;aXTGBBZxf|GH|P}u3NvT)eU
z@$K`B70n_`qd(2_l)KYhDE~UtAl~>R?fIMRT$=stOo7J&7$4)lT_}cOZC!NUnnrFZ
z@PD5LbLo4DIP@U^UvU9scD$~8T)d6t2$cxeD*fq*f18u`w~R`#aUFV<Ats`Q3B|tS
z8=7$#(D_<oc5_$g>A3LbuYTzbf6QUZ#sisq+w)#b-~L8+-|T?*z(irwPi*^Ot0KY6
z0Lc|iEK(@Doqf^Tb%yGVk_&BfymvjUaEu}2&hubsjMnRX1Eb6DadojV+01d5(W-_w
zk^tTC)CGer;L_#NZslRi25k+VdLbV&%-G?>uYA?9`{l{y`Fa@|he;b5iZc`hj5Wth
z9H@5{6K$s27ldD_x)u4!KW2}V0C5`CYwGE277O#xJz3$Z^E|>#@q*p6nMMY5!<QcI
zD0ZDW-poXo2yFRCQSL$3s|Mk%D7h?M_<NY=@ma}Q_Wp)RJL-R$hHF877kG}S-tQHc
zj_%5l+<M*M?cS)sZTnW<tEf<+<p`ii<5qj+N@3`TV?iBYX33jpx!36aKG3)v6oj<G
zol~Ba2MSc_y?giYMsb-xc>MNG>&${U-E2Fgk+EN{HVriu9Xz=F#IA|1TUsLZk!lDW
zt@J7Q6*#xfnOG7qfm$`NaP;?f(=un#tHTox(x}?+W<GY_t}Bfu(HCe*mpn5Tpv}In
zj*gMRS8W_LOKRU$?0GeFzh~rxp5cMLUMKu7IsY1Rrp@T-&9BRh<tZm?bBGG@Rf0*a
zL1K^tvTLRyznQ-+r^CN)?v$@Na~kDucjy2d(54l7grwDwXv`sz&JLhsqxw>(nHgF5
zrLOttU}5nZ5fL*-uo4#ZLLh2*Sh*x$N?9<B#52$3ZF$cDarVN?^uLXRj<R59s2(=}
zX7xQSH>0`2l?kTeA3PFbM7NAcz&CwqQsLpq-!GNka{r!bol$MJF6n?9TEgUH<)8s(
zvdWp@+{d*(g$4auvKZx$!lm4?W}s`kv`XYkT@QJmd3eY1acw)@7h8?xu8Vo={C?yf
z{V&T_d7qy(<YMZUNd~d{0$Ci?+A;Yt1?1{99cxI5P&(hf2K3jYADey|(Mp1L5R{B+
zz4F_))wYJLYnl5Ra>CZ>Um<Ol*1nli#!^i`hookU!6G1IAcr{rhHf(pn72nf*>@wy
zi;s|$lnX3BX1G~C%#V^_<Av!L&z_w{kDI=tM9zkv+4EJkL(Fd22XUkEXXteOe&IQ@
zl}5SiVhKFDz?_Q3zEot5CD``e#U;INT3TLhWoDitqo#Z)5z)9|*C%OgPhmadVK9E>
zaKLmv+i@0&56b=DTz~_$fB4`#dF-_1tgk^gly6LY);IpI(C;&FC8L!(rh4_{I4Eyy
zER2-=T;IYSUEOI+BR+3gF?D+Q)<cX?ia$O6A|+SLbY^#M7~>P0Wf!|-kwR?8KUe^c
zn};9sJq$3)uaA&fQ*X(Z2%AoYw0z!BjKsB_pZ&C@kzw)4IuC~=ZQinlvtIsA;d|l>
z;Z1JVZvXsRxO}7iL6a?7^81wxv0CPFmc8P+(Hss(T$a67Qn8qGtJ*Pf+dzN2NntrT
zNh|){ohJmkyu$L4qN2Xf1NOo~>h3yrO)@;I%mYHd=&j{>8zYs^>#+|34r06aYHHr0
z3$*&9Ozin-M{FWvA;ON&DN$e~I!Sf<Wjy)aYmZCaO&JCKQ2*OxS=re8uQGg(ggjGr
zdv4S3z=5w2rzCb+QnAmBbfsOk%d8$u5_*o#ye(mrnA2KxA?E1Kdo~feEbR`HNCyGE
zWv~7Rm09-i>Ti?-qko-IaJd#-<c4+0TMqN$_NYxfv%HVtvCwgdmRC*Q*45E%(MT_&
z-AO`{V0e?6=$1y=6!*4n3$GeQx6!2^WTAC<ZxN=FVncTg8a7Ojtd)z<BAa?SusTL>
z`V6B@uAL2Mce!8w_3A){KXKvYa1H;~3IyzYd4dNPinTrcX9uh<d%83AwX3$PBw}p6
zUbBi`G3VcJPx+vhW?Qi;Lp!pmP%<}SftmJ!dd;trp-r@5!VV|<;ujbD?MC&j$}}6j
zt${hbhu{B_co~q^@ec;Q(0YGSe2;N+ve4pS0UOF4gz@0f_toOfV>&W(|Jm)rbaSN_
z9hS7SXr@&;8vKX5LpPosapKb~2%K`an+1b)FO_S_$!!1`JIbR=XnCu=TmYXc5{m8E
zDZe{3^9g#A_Dbg;+mkwD30NUoaw})`G|4&JPMU-iH;l@&x<+SO#NuA*ldr)7)V+(%
zZF|+H){*2O-*bEHDNz-bHx0j%CAZJ>J#Jp(ZmCH1Fso)^^yOY1D9>x^WgJ>y(v_P(
zCuQ(G1y|YMKh|n?09<e(>c_}dm_H)ycup7ipLrvJqZ4C-VL#Q{1P0e^4{?i}7Ybb{
zHZC^cbkn$8WSG(84eXQV0tB{nt)@q_JCfc_^~8!hD%i87M~+1=$4L^Umdasp*U7i#
z>|-m?S;#2j2?D8O{)eL6+-c--Ek{UbW$QC=5S+wp?XRilpU`0U%yFZnE^+s3xU!g<
z_e}sz!>OOPiG9~r2{ep{u4qNj<*XyuLnOP)N86L`rbx4x{qI%bD@T@0zX@~0GKj;|
zP<=ZGR@ktY>@pyb*BMR-qL}6*G)dXKupb*3S!mn0u5&}k0+&_BZ-m}L&Uh1<D*x-(
zu7Sxs=lQ~4Zii*aKH+o%_&e_I5tqy`LQ7AfDS1y^jPBY2y0p+!$Z&Dy)^q+{;3Tfn
z2{lzjLVCFE9qwL|-)2K+*D(&oAs@Jv5r5=w(2BCZwQHAX9q%1dwYKwNE3N&iS`*9F
zn;aZM@T==PKlgIvsO|nr7WPd?d3yAEQ`$$M)(Ly6xA6p3a;$q|nPc;;s-P<48xu23
z@vM!Tcc-i@m6lgpa9Wyj#RPX?zp$toWwQvEEiJ0zwci=hnrnMh|LBSny*36B<6LHA
zt(d+fcDz8tw2OU1LKeHEUf^LX2(JwmNU44W56|zjCp>VoeRm0>eqFbz@vB1uec?b2
zNVnJ}$Yqmy=y7lT8OUBV_zPL3=#9KJ+S+=zVnI4H6pzQkd}TOKE6WV`A0f{MJFUL4
zMg&rnPyxs3Icw#E{VWUlhbG#Q=|5=ummyn=6p{JU&VS04jk<Immd&I^i+Yieazg4o
z|7US`e63?fgJK$r-sRq{$7E{Xxdp7QV@eumgz(aTbjByMjutIqxNt&XrA6xX>!(ki
z6syc2Iezn)-W5w_?TQtT3JS6uYmd)-bpEjw%m8=Hp?@89yznhd?+nYN(7&2wMOM+=
zLRw3+{0Z|Vjva$`eW>j+cb5C`Ve{8rVH)ln6_vFn6%MIl(0}4Tk5u#&@H$DQRh>2c
z<%hePK5NUX+dVrV9ymD6!<Nl=pmsi<(N74J!A?D$HnsD~d`AWA0o8pPk`q~f6K}2B
z`6iPVn=$yetFf#eXUEt<QgJI+tk6k3Dh!ZOE0b|Gh5YQOrp+O}RKuSMcEiTZUfMNP
zc9svf0vk3DfNkM)4Gj;<-aIFv8(jSlon_e~Sd{lz(rIe9Kt5r1)aCSjPii~U^s_+7
zt<d#8eMCBV**$rK%=+Dq8Q-8>-aT*N?S2plJEdt(g*K3a$e=-<E@M@F&ga)g1hh`r
zEftkSZ^t{^%Q3@rXT9lo^vgPWlTVq4mUP_2qUp^Odoy}CHeU`zue52u3H2!2>?t1G
z-$QoCT;y`|YsnL1a~fe6n_No;Er|Bo4VW4B+4MK3m#(MwN$QRtt}H~P-w_4cdQ>Xc
z?uH4kw`lpAlHk*@BWA&HA(>IAm&qL}CxzYiik9e<_M4zVGR=*^nGM61fZJ3TXNY#p
z3s^lDc+b*+l+bYeLxs$jdSqZt^9iux*<WKph9OMI92}wFS{-PAaQ_Ii(MouBfBo?+
z{GOVOEU{{$b)EU%My`p1O5D)8Q%w<GlJbe;t1drmv;(3GbD@W<JLYZkR>XUy`qszQ
zJ^=UFdi?F84AS+BRHCx$<(ZGf_v~MDTrKh2L^P1*1!{>PG{=?NAWg#76vMq3pz7Q?
zN4l034vvTb5EnX>JVZ2XINk}C>|Rv8aJWDf=>W7KE=)=A0aj?&R8e1EvDF?rf*aJ4
zlP3rNvgnVzTF2UxCpnmzL%ow~ify-c-8v;%Ss0n$ki-l0&`cShhPR)!9@o6<+dJVf
z(_FEiB^(6%`S}F}2@F}-3^Lhc857XL<9w^!;%Z8MEv7dZu^^Fe+k?>;*9EIR%XYif
zr2d3iv=HXf?kJhxwyucvYkvYH?*VI)cJu*mh?h(|jpq$547zD7#5LLMnWJ0U#;E&@
zbT{}~xUkhaXK9k!TEz=-ihoTGZ>%YUPVZT^V3j_-TK3iP9}gWh5znmWx=w!T^)qUT
z^aD(CS769+qQAKbP4su5)8`C$JA4cPU9;ry1Fo(%J?qPb$KTX2)06EdQS4PZ(?M$b
zg?yN+rchFAYp_r`g~cOqvB0UINwRJoVY3oJPlbPA`^S$XzPD>VHr;DJ`IVQ{qJ!2B
z4oPZGybEFicS}<xgJ0%g5NkE#Vf$*V2yDN}p(5KXB<<*KK3!fhsCL-xq$GjCE|@@a
zVqzCSoQLc|Q|&dPlJ!F;0mNtSdh?gbklehQcd|<cZk@e>V~?^j_UVUd_+Ke2cWU%~
z@!8ztN`mNXp=euF<6AQox-GzX51&j>Ldm4K@%odB3IkWPbT5@vx*-R|+Vz2@Di|kE
zV8qk>`5DMFz+^dBp;QqEjB-bUf9S^DZ*aYAsyNzR`y=?ajqolMAwWNSwqalPIqTUJ
zxHzeY)yl2`RhSegFR3nUV0<FH9w^VwFRWRx=Y90n741NHN6xQKmz-^N%zwoF+=OB=
z;X`?_w;HDOQ=olq`SilL{`I@*q1+X_bw6Wv?lLK(-ZKyVh9qs1ssh><sDS&X;}m(y
z|0oW{Wl{z(?Z>O{*Pm~=INwR^WyFe=UuP;BR2!{WUlwPiR8peK)1uwG^mCuCLFi&n
z<XSUY&8s_ZqFjZc6PE|D`LtQr@8U(7%$*_AO5g3i>(Gv`Wcpyyv&|t7J-xhaT?0)P
zEecXUVCZr<L-b2`_rU3wWLTBl9tUGf$<?DD6nr0Bk*ipMVK&I^_}%p~#8|>jy3r-(
zSbyC!hX1cR7Zeb-Snr*1cCv2nzD!hcrFFH;;n&QW^L+Hizq~5Md2DBm=M3KVNm!pp
zJv|${q9#w6@W+MGdGdSBYhHOR9eqN-wy|-Hv~<$oZjL*8TPEUKVp$=!lX`whdJM2P
zIsVQu*B3Ap3?^%@C(R<1>GWQ<D-tNX8jkN)w2We$xEgZ<v@fBEh`&Dn+#YT@wqw2+
z?xATGVMfBb7xsRwjV-Ip@e2{Yt7j2}3`^RtVFm${g!^V;<Jl1S<G#5!jeTub8fr`y
zOQ|C0;dp6Ld3gW6b*AP(YGs?TQ4YsN*8DH_-a4wvcIzJ2$3hfAz#uJ3N<u_XKtx38
zkS<9H5$O`K6$Aw77L-m22?@oZr8f!)N{57`^l#oi?|Z&~zdye5jqw}ryT=*hoFlOJ
zz3=<FuC>;jbIlc8g%TFgY-d<{D+?c=a_ci=@7-VeuROlaMYO3rDtLc&z;2LoMdV!q
zCW|ivREd0Z#l_B!I6Q842Bslt11WxGzD`{ozyQvSR=eX(SWH;Hnqh)|l|4PTp<KW*
zaBB2a4_E_eM@adl59Qg?QR+h{I4`RTLUInO8H=%0$f18BbXwKjN`ccFrcD@>y4rzq
zDUb+#{J07`1)@3K?@`xsU?U>s^dnZD4RoZyq{a>KzYG8=HXm??DK?BMQ0?5jd9&%Z
zIWP{;T%d5+7!R4M!Y>k@yaRa5PTzK>qkj_mE>AtZZZs#Qt;lS<KOBuL9t)Y6c3X8K
z0IgoX0lY)QM!?~njTu0Rp?X4<e{yZ7Dn39f=c*)r>P-Rs<NWe%NuD;Gt|&5<A_SyY
z&fvF02-gpgRokD%oD9D?KZ)wi^3YXP1I%4n0IY>ge8Nu94~HR9fIzz##(N_8+f->{
z1qJ*kJi!`9PF6spco6^sF9i@+!c1c9^aVAdkjCGmnzL?^af2i5o<rgpW}-$oU(QQO
zIX&BCD)0>eb#U)tatL^)a8DC-mpRLTqe7Dwf!+ph;XP<6MVV)*eX#8bipZA>JRLZL
z<+?q!PQWqJ8#MXm1Hez_QHVhbz+*RQ6~v}MpiH=D-vL&<I7OcV8Visg(Fp8)EITtJ
z1KCt5P&fd{sk}<sNAI7Z1g|au3uCB8<$k#z2d4&Mk~L11v@IgiLOAx3WrK|BSl{~s
ztPoB!{8ZCBj@H*8KC;E}T`(dhMzj(;AWj1cTznTW3d((lq$9M3jer+Z7j7PWwo_tw
z%f)8YOprkU4!DN90;~xUH39=Xks4!^?rA77C&UCXio+cl+8UJgSgC~E+1<V5?OP9I
zCzyXtsF7FvZ=*~>LkdI#@La7TqAFTjwN`a|uR(|M0p9?UP`uFY1@9hUgLuJ10_M8k
zFrTWci)g1Llya8)zclHiW<(EYKCT9(J{HPZuk~&ciG<uAu@J;IQm3}paq(98Ca6X?
z(5{0CUULXCXz)Z&3!acmkJ~k9Axl`ep;>}n`x68KQ0lrsY71=yN)|{D)Kiq-qpJbS
z8T$^IJK^=<*Rr#znLX~709cIn=Qs)o-17+~hH&&jEWnUM69<eFI4H8oD5&5KF!mN<
z9}FZ$Lg5OrQK2V7=!e8~&+gs9GePm3TwKJC!Gx~gQxzL%-NOvV12C9~gDH1Jgf`YX
zP9%5{6hVW4w$0Xc5pA|N73u(xu)j@jcImviUX1hs;q{`lGzM%9;n$Jd+@2dzL5T>~
zAFLLk_(Sx=ahZp30I6{E8kZg<3&=oy5q*)#;UFnTvW>hCBG|mVJh=P--C4oHsPOXo
zKI}@LUWs~*D7yARZLR9jlKc`MO0?!+&Ik~pvsei5HbxT}$_XGgphaQ6+zLd}LHa6@
z4O@#u$S*)9e5sdumkPdAHW~r*8pY4x*dUgJL0Kf&IGjXiVF79c9END5#+6!jWW^YG
zU2nKO%(G?#CuD8h)%DTjIK?>eHH2&6dJr80<VR5p<5tZ@O%tT~07i+)f1rPowPMdZ
zc8yS@f5?MqH%OSJ0xJ0Dlahi01S*LyUz%WRKoR>5=O;@3Et8$w@vc>PJb*zRk<vh!
zgT@=APN?#5{|I&!Jvq%TZUVZxT_9WBYJ-mx7(Q~dec8omeFkiZWK_Q|*>M|sMzMzn
zp`gSt!(Qy5eG-B`Z+(%|78bgIsRn&4?IOsxmy(je_f`Q3)e=2iplT1kVM*aYs~_H#
z!I@@meqPLepc;Gr8Im&KQAlh-wV|yLje<obC8Sg}89T#C1m^eS4I$5A|Bt+NeWiB;
z+j0)q2(h{sGERs-6czm;xQ4j~nv>vlHGv64LKaQuJk%?{NMW;Lz>JyJhm6cjtU_FF
zBP<D_&_tcKg2SFjm}O-#ABK~g+Y!y)c6OGi$YHSyz6f91DZBgIH#g+^V8BkJr4Ub*
zwy+sl9oD0Vt1A`*8y{a2v@{=$(k^BzI}XWMo+{D=VV#HOZ7wM8P+;L9Abz9E0qrk%
zf3!SZ2kVKIySzAM+Y(3~iC!BbJxpW8x@Pqs+a_y1^xLh;&uuteR(>%xJBP8PF^;Dd
z>W%jD35;XovhzRUt9Uy1V%_%lRQs5jvli%aELrO)rk}+|@$IDP$UPczS-bL9`^><;
zVu}!U_XhkqJd)ZgGXR~2zg5X*ls=cV%&VoXjV1A=dxFyh(`VA08=(|@Tg<@v5cH?^
z`Np87v_VSvdz{mR(U`LCtyd83W3LWbHKwNm`@*tLV}*m+GPL9M?Xd!~McaFEAXD_D
z7w%sA`S?(qEWE$0^YXX8<ErfH(k*VyszfH!_PO(28zgK<P1WVZmq%VAYQUw)ELdka
z9^TRJJT}@F2Zx8#dF0S)Ckiymp<u`0PPmqh{KPHN?AmOCOL(Lpe$NQ*LE$kmnXB2E
znG-<1(^j^sazNLRzU~5igpej`ny}C?FnGYjr~NsW#OPaVo=qdxnAB(G5DoU+DC=`+
zWZR@Wl6s-*kH2YZqlX2Gz?V9?rZvy07awx%K=BEemlPcV0OQmho3}a=9+K|xIJnno
zV2Pv+znxdyIZLiOuG$zWRELf+J-_2G0a&UF&!U>zH%YI-%cEC>;fGpFhS*rpnMAx6
z+hnj8k#ueP1EsngP94GzGi#yrUYjT-7~Uq&BXS%F6$3*<oEr!!G81pdLMQ>25zrS;
zs$qrr{r;=mL<?}-5vfPS2!#S%Eh~sb;;~#Tkoyit=AS;y9L=yB!iovmSi_^({nfjm
z-uvm(E~H^rH6U&|26HDj8JH2s-dN-&<g(b!0wx{?$5H5j#d|(efm6G@qN0MXOYLH+
z^E-5p<K!^S(ieY~k1!5ur~Ns;&1iYh_7_Ym>Hpm6c_3FQMUULFmh~Mp7(`=*{=(%4
z^zFP(h~GFmY(9U!lmzGys=uKd+wN!v!-&z|KDSu+w-54w=$WaCu&@q5a}8ZsNd>dh
zIOv3}+Nyp10_!2-MBu+Rsqvzl@O}=a15S8>GqDhC<Inw;n>Gc@;Y1Cr8yvbJ9l_om
z!F298|9|Xu_w@H)wZmODo@6^`o3^6=&!)rtH>AwIFxZy94`AK&F5UO9lNKyT?l=-|
zMSowT|M6#T`Hw5>tN$22U;2OCM8G@$`xl2nivIh*&WZl@I>Z0@ANhvY8@_J}CgEsk
z_zu6p7jf_=Xr94QDA#%fc3!m!9?C=&=!TX_J{KC?BfDrAZvi62n!qiSfBGM9m;d)q
zV@<v85i$XE4t95E;!TF&PbQ4<XYJ1+PKg!sh)frQ0gZhdx|6#V?kHvq*cTUf5f4ir
zCe5N}&ZLJDu7^mjh1DGJuP=oEW5fBsPoKC=oBnsf|Gf?WH|%k)V!a!ar*OGHydZ?y
zi`-z;0x|}u@vLzC6R%Kmq_rax6E+!yRpN_aV#DIKdI%X~7ldn*!>AGPvJ@q9Ya~bv
zFZJ<rzCC|;g;pq}NpQA1m`?bpp?!=lo%pn1=D)V0X1}MW(eN1c+KK^25wFu=7k1}g
z(>*OllZO`Qmc2Si9FuZ?&(f=Q+bNl^+u9l~yQ6TEHH1ynCf}!j-7Nb~pE<J#WG6TT
zDM}z=maldA?*S`!W%+nIGDi4Jl&k!;5k5wQgiSdYP&_#}JM$_~{2IXfT#fj9m~#h~
zm93#|j-nielr!vRE?$S-+9uzB&6v-JnBUO_5{Th0zF!aG-w*Ek`!ybNDkOdWYtQ`j
zUwh_wxCdj%tNFW!iP7V5_S_Wo?}zF=qHy3^GoZr15ULpbR~_L=48UOAV_y*A56MmU
z_Z<geT|^HbK!Y&C%o;9_7YP61bEE%$wUeGZ-cryBU_{OP3xvz^fB$<vk~LQkkMY~$
z|GpRF-}g#v6*G2lahU>leVGAY{f_Ft7vl*06d%(vSdX492=vqn3HRm1e{HWvL42){
zbZtlN+5Ua6L&STXqkAFOq)~W7A56+l;%?q)`7g6(!YT&yGAD*d#4`z(QS%FgPqp%P
zkaX~AQCJa?Cw#CuX#U>i+U*l|MCuJ+09I?l?0WabzrTHI_unri*{ZUgc!a_G{zgW#
z`>3-);klcMlm6QmK$;Ru2`JO$Zf5-X!M~QKFY234l4Su*0HPE0h6n%7NA>rjBbvcM
zLt14d^7pIMQt9$kp~3?*eL`CezovZj#EERcSMAgo{RD8q6a428;mN$I4SoHNh&J?;
zbwX!uuNFO+K#U?@wfFDb*~AmUFiSi$;`P57z~9^O|4-86pIOX+k5lGEWe>v296A-E
z%7*tj`Z!_og&EewoKGZ|0BZe!w)psnB5A-e(C}Rge|U#DGlI2aV`BlZyTMe$ij;Bu
z>=@{;fdO>tWTL=TQ?UomgA#NVU=DDLmN)s)&?f>{@k5gPuQU+u9f!{*1ASpvjBa{e
z2}$W3O7A&koSy(n@dChswpG-dC==1P3O*Pe9uQMu*X{NqT<eBdJAMIm7`nlPi+Td4
zfB?Xdn(Md=L?WSwMNY&DT`()iiB*_9!wokoG*qLnt_uM7cWZGBa@jofLXX!AEhfMX
zVY<K-xf4fdxIVNUp<i|ri=V>;oXjnZ<NILG_z?Qt$d${%bZFmlFM9+0R~AeMBNPUs
zvEsM@qx@mJU&FYpT51W23W&VXI9=aOaaTG8SSkQlfJeANrju{*D<eLC?;(M3!_l`0
zaiW;vpSl+4W+6wx5HSbBJ^CCwv4LM)heW;>q*uWMJ7CY46S@f35VWxXNl}FpD-Z=>
ziHexd0!0JdnN;8Gr?&1XT>^TGXJIw@i1@g3b$S{k;@}Q7f!;x9nFsm<2q^+X>(Knw
z3q|k9h=s^_)u@!yS`i>UI6M)B3BWSIt+bxUd~<ITE|#?HTDrP#i;9SuTsSBo3^bCL
z_YIBPT&#B+<6K)yiNk2wNP;-JhT!^82o7B>*$mYL;IWc6wA##q-}ahomMMuwNq^R~
zIsnb)`k3hp1so3#?5OxeL<nzGwEqG5vRz$UoA~uhcCmiF3ceAT`@6Be2Bj$^u~?f}
zCWMU=%q}uA8%`w?;tE(KqePqq1fafyo$!90E91MtoctMJ10Nqi&c{0LIoApQXYS2^
zAMR}aZvQhF#|1w>yF=(iw8p)`JL|a1lH<v)qF06!fUI#0AmYk8=!<A&;EorB3UyQk
zD-oO&XNChh2@{49)O0;e0A;CPxx!L<6Yb5S<rmCr`e|{~oYz;~6tdfaAo{J(;{Loi
zK}0U)H`=<o67ayyf2*pcr3DwNe(`QVi<qlWFc<g$)b`-SL}r<ef?G5MCWH{MPYO!F
zoyUp+1YR>WB~$8wd&I(LcsIz+sbJqG_NSYEKC8$UgrNZN2efq0BznL-h>b2U>A_re
zn{a4uDZmaSY>OXbGcX)FWc#ry628k=T;NLQOVaRDP)~q+nihY8Ck3D-0C<?X(Gl2o
zuG!Ph>TH0iz=-8sXAtEHx~C2O^&Cb&(M1U38rkdKtd|rNV&H|4^b#ZQ19JI^1@Sjt
zh9XcqNA9udE&u+6@+5F5bhMFjteF@Y2^X4X?L-|{cXywQCJ8XjJ_9$tf=X=LC#TFI
zc#9mfL@NfS_~q#3b3SzfB&jfJ1ahvvs1M=mocBm7K1HsBXzd0@$P82V44L0EGo3|t
zCYDxE0gqut0si)FXNd>CmmY#<K~@0)Ro5XjubosEfKJKC2)6RTX63$}^=1;WzmVVE
z-259Aaaa-Y1xo(TZ3xW*a*{9zrdv@#L5TDcUCnSak(_J7cL#L&YjV;6en0?iZ^H))
z{Ow+`){YL%x_2jeAgBX@XaK1xlrt#FEh8Uv5h6>)+ys7OAX8<fr4I%H>n5~{;53W@
z_vP}oVGN!bk$VVOtu_?bFd|deY5(_h;g*CsxP{>$0I|>RWN1Nz9bvP~x3@0YeE)!W
zFo#NSWC@NC36w`l;_lz$Q-(aq0B&L<HI=DJ5Wrw#Q^ksyeBNyYh1SRs9irJ~-u+Yk
z+qnDZ0oDqYy#kv|8Xv`n9z3ujf`ctL6MHLR@FHVlI{+m>t7LU*YHIBA;nOpNas@Mh
z+pJ`70py$O+|!3HO=42KjuBGeq?&LLJgQtZ)zyDM??1Taqo}<vCc_<>FWMAiBCRij
zU_8l`rdOOB%!5Rr>2Eri<A}s0VG!C(ev%X{rS$WOo&^Bk36Fu+;@Gh<D)#eJKzMQT
z8o~9T*LP<f;pg9YfMVrBp)tDLh~;ACo#e?}OH$k;%ch@(B=p_i0{2d0U3E1TkBgI&
z5J{>HQp5aJ)Xq9t1VL<2;0?Al+@d(?y<x(!3RrkSOLm-FlYv6`>bp(AZVMdX@8<hf
zA=4DwI%%{C5Yema$fEBU5NX++MEIrNwQvJM2W`6V3>BEo13Vr1+4mM+Mq1GT?kq+B
z{zyF3%FXr)_bV^BI30waT*3UtvPVLY`UY;%+}zyrE-SZ_HgiXUYEIe|;htRG`R{wv
zFfFkL)nHEo1GWu*+?3xJwP2(X=|5I}TMt9dRa62;;$dXS#HI2TB-BmgY%oOjoLRtV
zAyRmFAhfwA4I;m~(jP@dr;qC*%Dp&1979(}f$F7<28`S}vBe^iI^9!ulY)#;J6mvE
zT=pm;@oWDO4m4b00KB1A)-7+;4G$PT^<Y+ZYliNwYi4Hc!1bG@aAx9IrX~{Lb1p>8
zde<TiIt5z&xh5RO<G2&ck1r`I?&s++C{DuN5EN<-ya5P43l{0}QqJ1WJF3uG4s+3`
z%;bCakc1l${Q{_oGdjl+B4RL4)MBRA)^8!~1K*%3{t>6CbR98melQKqkYnmmso*Rw
zo>t0(8!9*@Vmv37b^9LJF<wWDe;SES4QVY@k%^-t*yoAFjF>y+#H~F?2O-)RGy^`&
z&|J0><fHEZI0y`#nrW6flp2CHr!%(-Wc>>IZReD6)FbvD<%Z;gz|na9_(hBL2s0|9
zB|!etJF1xAk_Ny%$u=wywgE&n6uF&U+;#e#V-GJrbntJ+Z*~p|4Z&105hLNCT_B;0
znnwr*(RN@9px=SJg`Ja=`WcueX#X@P6NrHLzzN&4(yRkt?lpQ@ns-1q)ato*`eheP
zIQWX@VAp~7ddMDDT$6^=1niZskIxvyb4d<_p`AiAa{4<D|NZz6U_v3JNguU1ckUem
zHGCvQ@V=lrVdaK4I5$^UaB05K{2+MasKucig#&nG5XKrnm(oY#p(K0!-~k##h}#GK
z7j}9ihHQXF!~MU|NO1a_8^34)4|2wVU2q=N4Jej^O6WYT>GxTPU@2ojrw-f$;77o%
zSH#E0W{&V)Ih^1K10FU^X1OD9F6$c(?1t~6xbO{2$n3!fhW0%5Csd&s-F0<bI8TFx
zHemmSOg6#q*^?(!v2g(bFGg#S`gx0ge6W)yzd{yAMtfo^(OdM}?3uUs@%E=j!w8>`
z1R=;EUj+-Lmq1^hR71?4px#AB*3@Eabc;y+1QD@FINU(os9jk{o=N?ZccP`W6@8q?
z!|MFnSw5hv<-A;dvhtNHL{r-37DjK3se$s~-319E?m~ls9Z~xnhK(Qv!R?19%mDxQ
z*%Q+$4O1lWmGW(TWwiVDSw?!KJAy|<Qg*48c_*W&V+t;MaFlNIraLo#=j;UY100ya
z=nL53;SkzgMDSHG-)|*gXS91<>7}4UtIfpl@Bv1~>*qY=2~{<Aj*cRS!x<5a86-c@
z@9#dycUVxcwcCa`Z1??5fjsbTSRGIC<cYp692Tv=(=d4iE4U+7?HW#A9mOuhmcqhC
zBp#sj<T3=09Qg^s3?YC5$KiDS(N>6eT)_+}5E^X7=+j5cJbZ~68sN3DdaTj+iLc-A
z9rtw8+SkapM9H^Z_R=L{qNaIcEUp<X2g68^g<;r&C4qti{0HPeHTCtkmgh#n9zq`F
zjye*{al~L2TG}^oit+N=z}5#3OoB=r1S`l4oI^-}v0TY1C<s?ikWbLMC@i0($OE;H
z_wZ)!M3#WV^5aBvm>w^zHt~XP5E-|UoSX+THEc_qEyw=W83(C|$H&LxA(m5tsR+dD
z#fb$?0yxEsdmC{BiP!%PK-FKg1lRgM94;qL{J*dN?*WIe^tTeB;7(6Z-y4PaD$gS%
zB!uoqEeyo&Lc7Rl7pgCAJ!BQ|F3#yLNht8k^P8L4%pFjwye=@2Xb^TVwhGGUS<u;n
zf9$cU${-WGg<&9-_F#6wnalPf+r9b|Y5fA!DCBiak*8gKe+QI{-Pu;_5wK|5KvTvG
zkBfi5-~aG6`DW<GrrL1GIx=(%6@O0wTfp?|Po)!X?U?NAg_;e8g0=woM#PsWjIhWj
zdIXWGp!^D9g;*A4?@5!OP2R*fUX(o0si7Ld;r_BXYPR+p%5?O5ywJ%-UW{`q3WSI3
zh2eNTE3M3*9b)AgTxXeF+|@p^<>gmhy_>ybdjXj$*Xr-SX9o@$<Oh(~k<XkS8HWD_
zQdqUs6_gJ{HuYR1(5(}ZEM%jgun;Qy<fX{_0xxiiLg52#4bIC;^77oAoHNtYDai7H
z*+4L83|MY~T;T$57#wC{to{e@3Z4|Y-hWM7@l6N^cIe4jti3@K{wtLXVZ|r!+Pxcv
zT)L@Ejw$F!5P^tv8l`H$1yRKU@3e*Z7N{X|w`-HzE6H%u?~B>d-WVEi5qeXgMX<LG
zW<@^)R0TpNL06Z+!eUtaK5A+-Lg9s(wVz;B06|(Z%8y5^#c1ETfLH^W>IS|jazkzm
zyAc-d`1Y+Bbrl-Q(HvwA@imL}&X~vcHLvfzG_;W@sPsr(tZxZ#nRc9^6%UNjq}b51
zn!_D@`lTPr4=%I0JW%Jz*`f-??zll~=!_I1jTKE*NTq4TwC6iyLf{R8{>S9eBbZ8c
z7R748`A34`ZRk@%q>7(XN1%;?hslBivMF5Y+IN>;qP^DD5$OhaPgqtoqY}e<$f&zd
zUv{1bK}ipUH3(AauGA)^QU9uLno$a7*DAp@7cDx?QV_UzT<L<shnUX{Y}Bj<VtZ7`
z8JUKo>c;?Jf&X8IstzT+rD#=Gmj_gMQ2rF==04B0IamA}SCRS%8u#mPNQ}H4j|qw|
zmcwK%5d%?HH_cIj;9Y~kOFKCvWkk=wc(1|8N5`1Q0GJR<fgT763v0SM;$6{@79rt9
zm_<6S=3Sp7S771gMP1_tn?Quvy<+b>J4?{(3Yixrnhy(8BYU!;2+DvFe5-o4E84hj
znwaxrboNR9GtE!-{d3W%*{vHItl+zqD1Yj^4mzl;{VFu6+Jk&#_RS7+LRj9up`)t{
z%k$(&MUAz26(^W90U}p0^o8v+dA9NC`<Fj`{P?6?tyAdwSLKzRsCg-N@AmcfZnpP7
zT8K=u3wXrZYFou~0jGy-xe)NcN&o^|?G;N1m<anSxNo>FPU?T>Qj(PgKBs4#;Odoy
z7Kzq}sBdZ5<h3w_Iv0-W?nu_L_1o5(>ms0w^F)HL^ZfKf<P*;h@)I+aVJl(*^~p&a
zC{?tTn>K_kjUl$fk_Z{H^Lz4T3+?luASf5&by@(95}t3O643ghAyex%-Wrxtgel!z
z8nzC(ttkq>mWr~b5N}aRJQd=)(03l{faD1Jwm&#O(3aQK(4f#qt`HA5a%f|)s*r!Q
z+3P^0LI`DI#m->JGJbQ!U<3ajeG2Gd;64lgpUlBZ&C=P!j6CY}3vU)KW>D~jY97@)
z(MqJ(lN;>I1=I$63C$d&1XeYer?dOuNzSmkYYmZRrOxlEze9!qJhC%SFcAxLWrdjM
zCaWFq%?;l-lpCy%>|{G7j*$7GuT$-yW&%(k7>lHYa%<(BlY5|9&I8Y8bOF$P#<VnN
zNcX<wn&Eye!D<B0iqOR#I&^4o2_P8|V$Br>3sK0TYtLZtwo;;-a(}|?riW+=P4oN_
z2V7|Led|~><)byWzhGCW1+)rKj<o;&0_EE$9BCF6?AT;#Xd_|5NCq5z1&{60=Ely#
zfReEOJn7rIEoR4yTL5HgxBsLq#@?gTXvY}?WI%sx|NHY8Be(!XC9uyD!{;F(M0u)k
zEChN*Z*GWHu^p1qy#PO8`w0D+fm`I<uWxDv1Hgx@1e*-^VvVdpNff-&GKYy8NLs-V
zLHM+EW~!8iii!_#33~qshf3(BK*#yC0>1#@16`!htPmrI31MiwzI3gf8eM`AF41dr
z;TuRl{t|zy2*(r*mmM2l(T43ty6zOjAEaSHZeL6|!U8=y`CW?W0kf$rGQjZi!6@lQ
z&22AkLUV^1Rw<z{2wqxQVbJ@Di{mvcuXmRpYyoTmJhh$N#>NIn3}&05b@u>0eTjJk
zeZgKDq`|oT#L$tg`y$V4X??{(^%rA8Nc@E$0rrj<0tz#CWawCVj~+Z2Y43+PU&Gsa
zGI#_!H*Ae?s~P|un5CdoYgHEtaJKy`VS20{C7(=;g#$FYlZ;H+*SD^rfr6ae2X<m7
zE%)3>!|7mdW|kUC)P{gWAGL=D2EImfT6?MkhcG%Bak<LM$}cpKkI&=Ly+Q2^Z$<=P
z2t+l{{3LGpV3;8r0O~Pv1yk(Ogt`*-EIl`<j9~gUJvWz+Dg?OmlP7;WhiSG2V-2nX
zR}Rh%bY!j#dVAbOW>O{YMszUG=z*^}90PXX-4@o~pd0Yi_3+MVC&Ddk5WO0QP_~SA
zZVS>Xb1xW^7gGUVPs^huFHfr=ypH-1+nZw@zKI*C9h^p*;QypF9}WFxBihq!Y$OTG
z1n7j|;<$J656rvjh|=yxBK@I5gijUj^N6Y!dQD;EB_}Zo{`5Aql^XhcgjiSzg>z3&
zPhv)KhH|M|?N4g750R)pA3rTB8jFE{&`05LLDvzXSc4VY1FotqRMc5C*?I??`M|Hm
zh<m!j8%$Phr~IpvF^^HM5Pe>mIRJq{nh*l>z6z-D@SES!i`S+Y1(+Njs+#fVw-IX&
zAbG-vF$ikm-!l$h8J(1aaFb*%2iD6p{tU9dQ#LxIT}!pj%nyn~T3!jWJGo^()%jU)
zts^v=rq@o6El9iCe9yOS-;ADB)_*703Ydwgf`DwQtMCLmDFCvg>_p%nLFmU+2ULvZ
z>KKOy`V_;3tWld`8a%}1&N%GQV!>K;Ia~*F&%k3cK-4kaFghee9n%vD^(od40uxB=
zru4LyiXQ+);1E}Yu0iZNd{F=n1Ym=3hwVH5b|t?4(v%LMFtjroSIk19D<e|{goS8m
zLPh>#DwHB9ER22ihC-cjy(oct18pbR1?^|5i1WbU(Y=Oy31k!33^-h0Umrlv3^8jG
z&vWjX6O0vrsUUxpfJ-bu4<uLc{Nb>(GiC+<fgOdt3NtP-m{x+zfV>>)LhJ<yDrG>#
zaGSn+f%FSFfz4M!D!TWxDgZDI^K0aQUt&d_()ilGBCWy|%j?7+hu_AzBA)xGx>4UV
zGBHV?fRA_&@|Z<zjWN{?=qiy-1aHhQE)oMM4QO2V{hFA7@DhFe2JMwNh^;$w3Rb@r
z+Q!YqAY~+$IIkwK9V|oxk)wZ5K*S1O^Z)4ecwXKA3?-DVkOi5->K;@Z)SY0A!C}C@
zpTf(_%hSd0(qC_bDnj5b<~}gJ@%3$EQOfIV${%!T986pHQH2~$Na5g=dwsyiNnKe|
z`%~`zHhm|X^em?d&&(HE8lQTkCG{Dc_ioF0@<et@y7S$v@w(EfC{qC=`jm>Pssko$
z4(x(-h_xP#?JZVFvr+HEpu@Gt2A!@II6A?3{p+X&Ik&H$ABv$K+%@Elm+j7aVY6V>
zA(EiSD<M1kMC9TNBBod&We~y4Ps9S`!@!iWHXw$^p#x|=k5Q@U7O1o^=c}j?ZXnS$
zv&_rScZ66GBGe$8*7aksS3(59o5R=v^OShGxCHt5sH9&(c@5SsY>mfsH?ypq9C3)C
zB@z-sFk0x&^Yp|@7(XK_N;oATKIPiR>s3>q#n1T;@ap70sgLoNhX)HT7rKxayy$`?
za^cLkq1_Bu9zdta?=ih?>(&R49{G4&MnuPkslAqv=i@_;Co$G^H}3Yb_e--qy*AVD
z0t@La8ts{ab)R!NrO9ck6=v)A?q}nnw-8P_nihNOc5c6Tnc#y+ADym8HoD>YIsCc&
z8P+K;Q<3A3s>II!w-!KKStG#k`3Bo~L!EEfGYPS*lvh>t<b7Y8UWX~1a1e7~QizE@
zA@bYu_g>p$&+UW$lxpZoy)DyT8Jdq07PLsxl*x|0Sk~gxk}#8y(U%cMcfv%?#L%=f
zz;jm=gYcr&q*=VJ(F?Nx_2`g*5UO?2v;6ETu;_3Ua_WjIu)Rjgy6e29Ow^q|jlP(*
zm`<8bNnEUW<>!qXHJW_vKQ-f?0l*<tlq0`^NEH+iHWgx6tU6*-7_EqXh-AqZvq`S0
zfhNZO>&`LSlBx6&geW%;52VJ#bjGYyBCg|EZj}Umxnsu;hwB^Imk@ifxR+wPSh?K>
zhYD)-RZ`M(Iq(?aHL3V8O_y%tix)o=^%DTeA+tA^fG7{+(PN~6c;@A`hcsIQV`9pX
zEMj%Cv9R#PVcbq)YU(H=bFzHIQQn-;euHE1!R0nAzlA3WjZc0u(J-HxDTbf{<uF?O
zUs<|Co0}>aL$eiK&0J2^(kS%47FZvcee;S|{MMfwoGUO~@Yha|S5Uy2O^EfcUOjvx
z28J7c`-ApwwFB-ZCG`Pmol-d8PgJvQ$>?q*`YZq{I|BrajErQ|{evPC5efL(-T5#!
zV;h??0B8`QXp1tFig2j$9e`0Q|AFO|6~WePNM1pmy^QRvNZwv7oI*z6d^~%}w!#ad
z@3|d|xJV2bCDoOibPNnQ^s`ICMp%&i>s;~#f@v%MicO`W1_i1eBZURP>u=bn$hT>>
zs%U8yAtVvk{pJg<vt}ThgPE#B$H0I&LOB*<S7g}L$v}3IS?F6SH6j@E$y{;s@CfP;
ztBCt?gpG}jgX4+D*}E+C42V^^NV+axPlJ3NhUrKgl4GtB);b8oAkfjF?jMG8!yI7{
zkpg^i1{cm&b5H<`#q7=gZ0j58U(m0Kt{xC|ml-~n8-F5)WgCL~Yz;Z)J`zpVsQT)Y
z0Y{oyS`tmp*nZv<)w}XGY=Ma;7FAt#K<0*77qY(|f2u;<K<!BM8ThL%V(L4h3r6>B
zt#Z3y_g)HWcetwoVXq1FLXkU;(zE;fcahH2K^P{&82r3$N<sqBK|Vwb)~dJlz}e6?
z7eSl{p!Uq2iSY<ndO?45#NQMC8YvD`#c$uDD?zj57US>lJ1Fg6+m47kg=}=wF6m9X
z4Sk9jOB)mHHurU8yc~0<{d4uB`l{5_l=+-7{-)b6(i5%4{M;FB1NEalIoI1rZ*^zo
zW<O<`*qD~R8m8{7Oy8*~#k|J;P%ODFnnW?OX2KQ|fg`4?HotwQr%pFwJ=ta8&i*a8
z$7i~lugQez>8i55x)vdg{k-EV<+Rmgd%22-`o65EF)rVx_K1tQ@-4r}k&TRyiljfO
z-{T?T;NBfEDR`p1q-d_QW$vhmZ)1V*>q8wz(Jl_Qq(*_HwSz_>(~jMvJT8SS!>o1d
z&OWp4!~F;iHQ&m)I62L~L?6$}A?9|X!5y8!N3+Pq3sQkA0NsAVFdY{hJTN#|fHc;w
z*cYN{6*cOqI5ao}*vd{kmZ$oN#b_ZHp3HUcFdDe)*7lzS$V~K#Q>^YChWrhk>4?<E
z2GdQ|GY<i+5as3qoh2Sql46WI@(gU1oE|OgYZ`cJsDQgj(G)ldg*8gyauoX5b&wK7
zaG|bx()|jN8`V-w|HPjc?Y@V}DpC6Jio7J;akg%~saG&#2eUnx&e%`M+&heqxDE0K
zaU1bOOYiIQuWy1g>BrD9jCV&o;A?%kRg&ysC2~w0H4h#?e!8Lv6BXF_8I%UxOdrM{
z!r2U=>kJ%)0|WOi){;A+dO&W(%EmS{G-PFINk>nQT}F(?!I-a9&1YPDP-S5Pk_j*R
z1&w$yPrZpmN=j6QgA-zH4Rfx&t@2(Bd5G9~4@3wCIv6Z`MrO3-pMO%qM!+xkeO{Sv
zPLw;)`5XixFf#a0oF6+?u^+*9;FI^GM;D!**5VA(2!3{qtPfN`eQPUDsnp+_WYh)+
z2L~ea<;u|@)#`tq$c@*Qg)+BUHAYUcxIXdq1>Go7l`?@Q5mc@Ec+AM~P}wsozylvm
zAW!PXkz-;iH4$n1)*k@>FDol6X6FKi07ax0-VSJsw&r0MN2Flp=1FM5#*wAfQHq6u
z(+zvBaxETwKdSAm{w?!B%zzSrEW<dHUE|+mcfOax9yhxlbodD_0S9%pKrw|R6|Nfh
zb5PA^643#78)@}fXo&E9#zo1=nE@y5J$Ufa+atAe0Jt3CXGwTHl({=K^fU&Ex)xz9
zx<k1lXfP7TA|Aq3-d7DNyo=lS9XWdRp4$O8G@03_D$(!TSAer<reb+{8D*jA!l(Wz
zk3Gx-i7(w`DB3q1RlLI16s6RU{i(XDDW}<4+va~*b4zyQc*LyRt`8-T?ke70)0vBM
zG*_J)wjen<o^%ZNPILJ-w`&&BT2NC>Q+@HxtigD3=z`>hE8EgObnUjy=yjd@-CTb*
zvLJ9iBJf&@h|Hdb-BCQh%5R=~y4gg@uYuRRVv1GKBX8k|#G?_5+}^#2>4T<|0|$nx
zR;yQ2$*&}r$j1zvy)6AsNH{g^oYT$p_dU)%4ksGLR{Ap9Y?CbvJ;nX4urw3YJaCM8
z!VVbSJQ$LJxJFOE471s0QPiqa=sOAt+2^7L=5z0-R%op7O!5m0HC0sHPy>Q)#BhTP
z<M22_aq-06a4nq*cblyr3Dhl5Pe@S^Wt6Bcz`a*qz8;CHw4vth@~>^>PkaqswY4Kg
zA|>VI(oy$g9f2H$K^5XL`s2VEqPgxK$B#X`cHxuh=0ARxT_{@sl?yb2_P~Ls>(*vw
zLJ0Zb|F>+pkL&>NLAX$#yfgtK83zI}4kldqdVgPEb&)dftlY&57n(wgN?Mc{&N^Wc
zK+!wk)!NbmoLI&=H=q)TfcDdor?WV2P*yOZ;@NQRf*77)B)bLQ%3QoiG&L4<uVYIi
z6gE`MbU@?)r~*<5F`RlLKn`n5^fWbv;XH;$w!Rrj#4w{TDyS}cF}QQ6_Mf}&zI^dt
zRZBc|KB;k_=f%;SixUV?>|}ko%|r<Wha(_U${Z&#TjI@|$(BU4axpaWdccclV{VjY
zsTX5>p6G-|;`B*PwbbeK@(Cv@W)HH|)TJ;a-QHi$tRgV)+WJp3<qWDD6g;xyuj+dX
z>CZxShfE@HhboFMgslRc35JFhuU@h6@Jt{{hqwrv4e!gylqe0P?btFuu#I_qRb<LA
z)rQ`X7>YB}m<hlTFHzCRd2ziPH?lZR@$vJILPvl^s-eCf=q~JWa7^J3ctpw7)zt|@
zPLzr-6cnmY0_0T@$O4>(0_z4~Y`Aly!bSoZh8YK)%wSVsZy_fqN7@hk>SDudjCqIj
z0IavAwKc@p-+`5)8)v_#2l9%J=4R~zt9^NUm$p;l((6@LXULhMnQ(<B(EQ$=WO&iT
zDNA5Q$?l1d4~nM}k40Xa{9`-l?mtEy4?ZVoV?Q3KI6wby!fsUS&-t+*cS&B#4+`ee
ziUyIHU=6ZAo~$jClrC}@-s2EILQ52PgtK;dIGQ;W6ckKMh{-I3GUyr~hKgb4h|Jmq
zbW~+%)2UB={bvQ=^PvvOnof{T#zd<lz+)SIkUQJ^`}cONiH%52?w~v6C%tI?J#4`s
zjq^wC%3GZhe>?g-<k+3|JWf2ZHK$dh9@#ikc8aZ5-?U?z4CkKb30u(N<fUUbR&5-x
z?X6gu9FJ1xZg}vW^PiB4?W^r|k+wTFM6H}ox!Uq&S(gs~W<K7dxU)R<RyP_3uc)gd
zY@(0bbZqn{Za!fJLcxw5TnEU!@cOsP@4*&r?f8meJh~SR-r(6U66Yj_k#l&dsH+P>
zZ-$-MH0D`bhfw+8ypgNxX}UtDv#tT!(TO@c`k>T!UK0Z>7PxA-v)D2EM?{k$dZRAy
zK5n{EiN!J0NZh03^mI87PK<B^3{3d<{Q8M{48f~thi`RJ2iOY4&I>iCot);8Y2s8#
zO<9d{LHqJFyrqCZ4P(d9Y_&to2W&%R)ZM+$eSz=6+EhuG!^znMidhUlOC*&MvVA;S
z1$p^b785{w+GRg`lRZ4|<+XtTfvodU<1+5Skbkl?0GBgpu<QJ!Yi!&HT}Zm_BMA&7
zsH$SeJlr=FdJT<@y$}UJLWQcc3!>{Ga-bw8yaOZkpCUD0o;r$XaaZUN2F$<^@g@){
z(rr~#jK~Qv5#6NtnLVO%+s`9Jb$WVf(;*}~Nvv338@!w=s@5KI{692N=hW89I*O@j
zXrNu!BsFLkBY|`wjZH%B)1QX)YII^kL06&M@I7`jBW`g$RVFq+?s5LUf3|E%PfG&;
z$^AR}ATVDV8WU4fLf?(>98`xcUu59<1k3I|i}*~iF-Va!{2;+cQ(jYRYfz}xA(+CW
zK0*VkfO-AF=0Gn@NyCeA$1FURu`3l62s5pyWoK+%02E)4<-y@tj_QJeK@Ya2mf9sk
zm<bHz!ZU`G7KB$@=eeu6)r1Oy@$@um{jHJ!Nk4aWMJ<2ZclX<I)x%Vm#Ht;U3!!Pc
ztxFJCXOjP?^iWL7vtsSP067ODlLv-adOsw)V7ikT5b?234?4lP0UK|A&N>%pA*5LH
z<L>+eMXC$H4#GWuO8W(j3KE4;{SQfX^#VA@AT8)~pTaQ$lMs1hiB?>On$#jNt=iMi
ze5zd-&T=T4tbK0Sy_EDPUD{C2D|98}g@fdbjk8_jr|z25fp$!vCp=3vc|(46p1MBX
z|HEPWc$wAX(`x!jX{ndKoQP->)NXO88OUp)Rfz0z;@vQ1v{dX6$rd#1RV?B8bcVB2
zIOG}Ei4(-)0pe#-x9y+z*laM}I?Bo9gi(xG5e|lrpiPCl-y;!jLuqL<wEsdak+2nY
zb0^W2h<dR2w*SUnztNY>O%PUntXSsenlT9Q@u`K%;-eF<#rKkV9HVgDD$))5y-Qd?
zz<qXzLgqLPPI}0N3qcy7?UI9bva{f+hsAwg4;xfl-rgJO{KcrTN#xon&{07`KZKO-
zD48)zGc-K!TIAS7CoPjV=e*zLnMwk{bOIw0eR>PXDIz7Sc}fIMSd>qn^?1`L`_M*X
zYKq)08C;p*$|53$LX2pLNXdhc6b~<??ied|6(VEgXV@2kVhWd#j3SkLIe-5T)SP*_
zT9ajS7zBmH8#ysp3LGYTC@2!mY;Z(FH%fo-VD+_KA>5cDgkJ%xV}7wc<4~x{FCObd
z=d6wq332#NYZMQ@bW%~jLg21XjP61fOE`HH$-M>Opej>hX<-2c&n4auT4<`jyz=tu
zgzWT0{KCoL$-^Yfc*ynITp%Pu&B~o3k%r`8_R!#B=61?~hv{WyVuEz+N2$mFeX%Ua
z@SZpUSIY{=QEEYUhzdL)u*TRK6?%fE&R9Bai$)$pLw}ANuoyu=O$c%Yj0)Ahwgc7C
zg!>!FO(Ld%yTaQ@MOau^<?&mLQv%Y(Kh8tgXE?^Tw6|k9bB^0~RaW-lbOvDz_%$Of
zO$X#z!mM5WkN5B-zsB~orn%V;pMlnXwlIL7#5kfp>Lf-SztzzKv4_c+m|_6gfBW<)
z5vc(oPzQRESNjnq5zf1}nCB;M%i=UW4Fm!)o`2<$wWTEr8g$aOH#JQo@j*gvY`zaH
z+fQ&9*4A}+0P^MUw+MQ&nM}X?j_hSSC9-(N!9B!eG{kXGFsooZ`$!nMOI9#o0dgDm
zA15MC&24xVzQ5lW5b+XtRh7vz2?e+1#8o-$JR62LiR#ymktO(~_+0jO3!V98_PH+l
zd30sVc_DQnDV`5QpBu!YJak5yWm-MHx}R*ZnG>n#bnvoM)>KE@<TNz?f&X#uaH(+f
zsio94C7!X?sVus7A$!YT_OojWZ%#cCF#1+D;PKk>^;B-pVAzn?@r{JkB+E3L&JrW*
z6S^nNJKmg2KDWMfx><9Xf2zNi->ah5H8Cq~(sXE`gOr^mDt3DzXWpVKXQDMY`Fg_>
zGh1Vck~NE=nYo0ymyQ17N}dL6+}2iK+p;s`LVkK}A4b+4Hg7;W_UdttYNU@V`6r>F
zh4j`mbuQA@&63Ef=?N445jzcy{SnT4Wi&OzOiqpC0R)A`K!`lVYpj|cFnkKk5y0-r
z5Z2E&0=Ivn`<r^49DPhdLO1XrKb7Tjdq9vMz+L^l(AJP-8vTT$i=%`Z7OkO0>H@~j
zHx7_(9U3Z4JD5Yhd-p3RLBJ*KE@F}zNbybfq^PK=Au4qfB0<VSJ@`-@9grvtqB1fq
zU&XRFJ<j2M;X650mMz$4_$ekUJ)38f$UJin9+24K5$TBe)2JKq$S2#=JJ4c|YMO}|
zC-mG+q!Zn&DOSp6W`j8M!o4&y%x;|iJolRPI_YneJi^ZYxH#Y#oecGSxzYDJoD)D+
z+UmB%s*f#Q4L#=$m4K!rb0GbEpKQrDmDFr%&FTts&)N6)KbFSgy!q&H1-bBn4#^f0
zDYX7JaJT>`R2WR&Ym1l}DCx`+<{Ihj;6Nk17@U;EHh;Sv0nm~Qci^$S2J-8GW6cnd
z$}ezKVDOrKu5q25;QbwTpE_J|9?(r(5P@Ub@h{Iaymvb|%p%J}R!hlvy6M9Q6NufA
z(3&+QU{l}S5RoqVVT63Qel%}>ejd0$k>`ptEOrrLv4P;g8u4wPA_pyE4v6*Rhwkm*
zG>^cn9FE`Y`r6vWP*=OlKM^hAy&brkotDN~Wm7S>4Arf)$MkQ>X93Jywi&y<m~J2Y
z{rT?57D3)G2i-?{wJCPV?m8u7<J>zc{%7j_(`=qS9?Z^z4IwF%D#A3jvHlWeV?${M
z9_|k-_9P3XA5~t8=a;j&Kqp=5H&R&9yFRie+}nGW+2hu9?|nnFTv`d6?*V2U>h=aR
zrFFBStL;XpHOFscp4sO$W>wjaH|^-vmzeogji0(dnlt^<)Tvd$-#FuT6hQRC?)mQd
zs~vtH83JvOY4&J|Ouu^<eS-V-@+Fxp?L*oM4l!(pM>fBkZ5u8-?`~6+cSPXG%+N~I
zO6;$3T#CfhdVHpN{I%aEb%)Yd{AH&V2K#Ytf7qL#Quh#$!t=MO7!YV*zmoLwr68uc
zH20%7$J?pTnsUaNO+vzOp&_*Yz?b(QK1j_S=I0Mw`A0sq%z16uY33smCJ6`I=b4#2
zBBbX!D$=H)t~)m3e$9tV+<Fnl%qJjV%=YY$n!0+TdLN=P#@pR?>WWvGWuc~}1v(|I
zdm88<X(uBFv(*gyjdf%nqNkT-l2gSbCz9uds52o2uZRI%;WZ=}Z-6WCeuLU4uz;+j
z&%)Z8m}B_KXCF{a(65BJ1BfRi6Llppy5M)n;j@m@@HgBwuGUr=NQ5h`S-Z24V&@D!
zg}$6}tOrDgM*;axJ9sDUwYiLT(ke!Q-rZxLo}rnGmPgr;weGfNw5_-+ChvtNvJ{iK
zbeIUZLb=UWS1fyXH#3X#k@y9=YnUFwrsh$<t~}r^FC}FHGm~yxVjM#**>;C>(Fyl=
z3<W0ZfXTYJl7PhuP0-OBo*H0a>Un3kN)nbTZ6fz}-h(Xm>syx-v*%Qlm8t0HYKF@O
z7|y~_0!6~}H#fn=ASC5KgeG<%oFuiE>g=U>q=-%ll(9s^i216awl*TbrLl+`_|=R&
z^-IHVDVS+FUvijKrtWlOof|mebgxa^W97O}XRvC((X2pwqop@1+^sVL?3Y4Rob47e
z<nmIssK{~2Qi_Rv)N56ZGnkmOA$7$b5&IDPcweY@=I==D=%vNhiio-;uK=G~AHx&t
zI(qAI4ld`ND{C3J9X7kzkU}E#`js`yZiKPTl_aw6W2I@7f9071I;6MK%Otc6{g7cB
zUB?@cp!M^zwxT0t=Jm&jIsoG}W#_@|R96``k9dx0o<u8DvQm17Oz=ygBrDIKXS;rW
z8L!FhNHJkw@;!QF>aL)*X+BWz@r|G?UWR?M3om{xe!Go0s2H;@Xr`p4b#VNhmVc&k
z=mh&m!}c8BVLDX|<mNkOZ>}ZJSI$LR_@uu>U|b4(!TgxIxGu-ByXTs~oI+cJ908{6
z<_NApKamm8L%xm-9Ss#O6P=%QR!1Nm&adDmJqhhMKY?|1kw6q~URQ$G5USDx%*?!;
z+dqni5Vl(u-s`BC)v6c}fF=Qv__LO13rjUKZMNECMM*Zw`yeITctS~~zkKNjh5)-E
zsmPD)Yfs3`Y|uY@Ws*32**D4{7LXiQRNM}Gu|15bT6?9Wq;emSIu4$N8s!eDy6@A|
z^bkRmzS{*giAjOwp4wS~Fd&CYp;j-e>BkWXi8wc|FV)qTx>>H&djx=?SJfn?@83$9
z`(ttk1W5a-se#mn09=BH=Iia7kh>@HXhS_Gba}Y1?`SW5E4Z~6%154^Pe~Ql*p|3a
zcS4xoLf}<<0B~zzSKF^5H;32bS16a5sn>K{f)a1OiVuxW{}}t^jO2~oqYC>y$C&RO
z4*#H75$v^>b>_-ePJarfwXWrjo{7Z<PvI%%{FN7tk~8;rTQ{hV2W;#pURz(NDt%Tz
zIBXF6l692qUQ_l<TQe3Te}y#v{Y=uwoGquSBRLy5zqrRPN55&RsrPoW>WrLMXn$rB
z%9WJB{E0<zy*+K8XVd8_@{Da{R>`2g&@|~|6^B`^U?H7>v;kiQ$c-t=F&~>>o}5{(
z<l^E2d(4dZio_2>e9)F-6E0S)^pko1+#Y_u9RO5PTYCbxU9h0Ud_SJ$?1Va9zGM2u
z^2Z7OeXQOBzko_=$k1>JoJqYv4F=ryv2_DjHsXEtJ#c?AGR^vj8TY?iNYU1KUXact
zyZC9D2H5yu-a`Qt%uFuYi$E_R8xUN%w)0^nK`7NN@(oKLS+97shl><VHviAXNea$;
zMGMGM0FfxIUx9g#1aLScvz-eEXleC|Zr)|g3=Io|g|yO|8}tQFSF|+sGo(i@1SxoR
zrBpbK@*YX)2w+TnV%3CN7A&DX82p5j-O!vMNH@K8u|+wI$PNGP1Rb?rRHk=CRj6q;
ze&C$>1F!%sEwHVvL7M?L_bUUFr%4xC$r4ISS5fWs9c6?>+|F$1a(>yxn7*Lh%+H@c
zuP$X4v_E_(sN!cfziWzN#X733lr4&Cs;Z{|h=8<=<>5=d#Y)x(^p|Qe2sdlzo;^uC
z10t6b9p7)Uc0EQ*o*Lnp_Uw>$)>%k<v>$J)K5%b`HmsxIW|p&Of$(p-^@lIw{_Olb
zTuP7Si7R1jyS=@D$fG$VP;C+v`ug8dH`lu8<V>A%RZiiuWiebyNll@!6f(X3__YE>
zq^w?qY_@;2*N4UD+V*QptlUc%6x2gKLYJN9s+<lEcXf80ldJrGjDG#lK+Q0FMjZ}E
z!&em+wxrlMrLS-I7i!-cW-?z}pBU?6nlT$aa*1kd<^x7aDyO!wE8|Qd{ZX}5c46G{
zJ>3DPzZ&#qTx;XH+bLpoF&cmd{6YiGv~en&D3L}oGBz^u6|^ZVb*7`26c<jEf_sN4
zCy4ZZx4f7YYQ@m9i_QUiA0EGT>QN(GjsGb(#v`M#%u2vI7G`Ej;9p$Ka`fm?0fFbk
zS2R$_)<0wv*+ojOt*e6#A8Tl!F^}fQPN7HcgYcvMQey3@c=Sl+GXJ1|BjZzu#UPzM
zK1%};qOz0~=|_l(@qSWrieEJ_II50yK!1uWP7wSUeSESP6>Gi_2nxlaelQ@h!E?3~
zzA*SYP9ExD!c==9>ibmG1+jmEil1A>m!~-o^_RE*kkr&%1R@1)f+mE5hl>jhrZ`WB
zciv0UEzE)m53~*NcmESD$)XX~PeUl%)=<u%EoS(e2+{>5J#qiWy;|J{*rdmV(XKJ;
zLB?!eAO0H2pH=-EY`@9P6<0E@sr%o}&rkxrJneR3^xh8AXMFL!XD{6v8z*Pp`p;c|
z|2?x&AZ@X;lANBPY6qozEN`+0$heFjJ*)dFq{eDWll4cwat?PI|F;$(cLn+3r{laQ
z)&t}Ybt=8gPOPy0!E1ihn^IiRPeV?1PgxOzBvrji#8yeZ$9FrUzCPZ)hd-5FVc2^8
z5KE@=uYI2Gd&_>ER%r-MG|~^fA6p#jR#4leE_@($Yws0tMSqrsz@)6ha-07d6g+hb
zxGCYXh;V<tOyxK&Iiw{JBfdmB?DFfY9qwQE2NkohkPtM&>St~~)7|`v+e5)P?6MuL
zJ_cisR*E?>c|k$Bo02m5#zwS=LrPw<-Bq5@#@}lwmjO->T{i>s9vJ@2gvt2W@6cm$
z_UqK`vFT=j{D1`;zIFC{6YV_s&YizrMfbt{ak)A#ImLh1K;gsAEx~Q@Iy4ve7gb}Q
z2%7~RM=BQ5&i-~*OsvLm2msF~eeT!mJGmSqhji|Vs`@&;J5z`p+sbbn$?lraB*PwN
z^b%qoS0RE87-_5CEVOI?wI-}a+$Cn0;0xx~pZ+L=*?I7LEBIV*zm@XV=a1f9*3`-I
z!_y(#DLXqm;oB1~U`AdHhTRrL&)IA8SvRKNMT++?Nm8Aq^4ZD!;QSHmDaF+LJH#nC
z&!h7ye5nW;0p-W_Psz641CYu$Yu9&g2fi}o66PI>q_|0}h-$u<IDN6?4ivs8qTOi`
zL;Ezc<$|3J7pMGMSo>+?Q+yN8yvmWCc@XdJG`Fzlo3&nt)@FwJLxbmEw+m{|KFV18
z^D&iF+c#p#`6lcRZNM`VO=abCiuXSI=$Kw9aqGJfPDjp)1P`;y@bdBYE=I>1POB?d
zXy;$eKzW5rgT@7d1Fjo_1f0Uw-`3uaIuX;FW#r`QK<uIY=&^4j)V?1b1WmtL4|p*M
zS<Nl>a))q#QknBRK6J8Zretai<G67Ci~U9yJ!aGUHd<@bP%yfzU^1GP*VrxhqMXo)
z0IhmCus28~K9}?tSdlnMcI5d#GCVy(pBxbqDtPtkfzvDM@tY$51@H{!OKa`od(As=
zsp@Ie&%kz&6E#(g6=Ed4*pEB2D>k=n)*A>8UWH2FdDTt0s%e)nb4=E+*zBOhU}Y#A
z^wE=#si=hV3&}P}o0)2)s&1GngdeV<cA&1BkmZ3xfY-*lsemHr5!8HW!WF`MojeJF
z6U)j&bk^TB$<n2ci;gTROG-|b>9GRGjTCGD2Bm7nZO?Tsx%X{tb4#-u!CwM3r}>~S
z#^<T-@Vn9V#&7sqHr9yFxqxfES~wYs9ZT2fk&?GD_HA!tXJzG=cmXw$)ApSNgozY8
z+n|{}iB_}Zwa$AfH#d=uJp|=EP}`SQ{^S;crvz(UD&iXI`S`tTXWiG?uFHkTdM~mK
zvjXGjjZ<B7pBJA0&a$DfdV=)pNSy0cy~lR-rh3IWtE+XhE<saC2a<wna(NZr%8yhR
z)(O(&-S}RiWSMi8^|k9E^~nRHa<$I~v?z59uz<pXgFiBjtFdPp$ClQt*w!DVJt?I7
zh!|uERC%v@o2<NiDF(8_*HvDAw<Oj1_?y7q;7u1~`;(;wAOUb<F86N8Zc(}mCjZ|5
zQ=;(>k<9qwzQX%+Q1`dw{(7Y=s;bZhvxWuOAsIjw)%JD`wBOI!^Q2ZUOtUVc7yJGD
zcl|@qujD)=Yh&Z{T)z?$6Q!9Fxb|9)c1Kp0^2_mMjSw^1Gcy%-Q9V3QOhpL_qM0Z_
z=LFHRb}IG1z+s-f$9Z_n*eJ9;SnSx92mIY<r9R>$SWq1=z}Yhat=)yuXQ<C1ce`Ie
z`SNsI)Q*f3ub2yqmCAk2%-*Jae;-Me#h1<Z4#V3M={RJkG>nYg^a_wj4o*!Sf&t}=
z02i#cfxDilz>sn66Q&BeHXZQ%xvGZ7XHNm@!Vat^WDU|W@erzMue6Y8ho%AwS|Vt%
z*Fe8u9zcaqg>-?;pVCkv*Q<V1r(;zu{<&9R2k8STqUe)~<AG+YpEPH4Vw2)v<Bade
zJyEqBW_InG%tQ(+$cXnfHN>FW{D*xdJjY!Bh`W7t=R0ttG|#T?J5T;Y=lO}?TxnI~
z+X0*joIhOOt(h+K3DHSBDcCC)zsWy2*C$;?RwyrgFsmg#Tx!@bcDT2CTt0NBPiQ<U
z(oh|-!uy4PfziU3!<7Ei!(&6*AsRa)PuYl?)dnLa!gKZs#k{<S!z0qikqb)BojXhg
zzCue+!2d8=r=B?@&J5F<(%f8uL3!U#YOu=*-{tn@!C@5$o>5cE>sLBByCKF7Wu6CS
z!rKL|R%GbB$d}EHJ|f-^IP|}UKq<C^6dcrzVJ)+CPW!;zA2?8@r^wI{pO8=~s=~cA
ze33P6?Lz;rU-A^KgQ$uKvd5{KbKJ2$Ly$*cM=}R(UV=ztyn*2v95cw!f;PVAK4ku4
z4^rxDq3xk1Lz26CTOIYJR7BsG(q7%^7YPY12#Tz7p%}a9^Q)ekhGrHl&9<^*Zqt^e
zjBmhnMTX-0=*|X*g+09?Wq)`vGA%9b{{1cU#|2yoRyULWPW^6X{Mlf8W=mDoW9Qog
zba;A;ptTN9o=sW(5f5QjeQ-CZasuLKcLlJ3eil<6eRjhc3TCvHH<qp#C@bF_C-Dqf
zYd6j#anrQ8@AKp5`{!0}uo)hocHT<a%TC(4bx&1}9+OM%$g4jMfzN%;CB7D4D_rJX
zVFCw9^&szI9`{sxrnd0rJZC1!lY-YjcR#94t1UVix+VB@Mv!#)cx10qdE;<gXo-fM
zo86bNPF1Bl4{kBOMLlYleG`y%^A>gWb`TFOv8>q=INPhmem4NXEbe{c{=sl3(~oWc
zY@TO6Jg%+BTYdW^u>7{v+HkGxRc(~a(059Un}>#_Nuq+R7Z=Z$<j4(|@G7U>=Ro`w
z!!aV=s_*0ip!NqX4B=cL+uoK;>GD9nWyBkO%7)%$Z7MoC8hHrNP{CbZPrwfHOWjOe
zNka+o5x!O^9~pUQ_wQHDJ9wqO@id66AuFJruk^V=u@D-&Vz${qd)t7+Pats^a?#4P
z`gR3luOS(!A8ebWC=Mi)9a%NYCK`-Gz1h?P6$Kw*@Nqu2jp&_?{1x>Kiq}9XCsEj+
z!X{!!w5|Sx11+^Wh|tBZ%br7P$k710Bc}}7P(MG7*-N$jN>C}VZWta|G=NCzXnz3C
z%%+PQfillv1dH=<_LdCmRc3WSn}2$L46sp=LD$CmX%}S3HS(<p%F146UqKv@*K<hS
zV)`%@S@;Bw0C0~{D-XfWq13lRM^VAe&6hc71sdJ^zWlmXE$phSuFg8l`oh>h0#>U>
zDUGk>JCt=Z&<5I0HvyDVj2RIL#;1e(6cC<4JkNiK#hT&phx5xK<JmS^QgaxE6GKaA
zxK)(muahdrby@sM#+=l*&D}L*o0q?$KC(-n{_R|D;Bl5y9fF5?%uhQkoye--J5XNl
zs;#Br_s^tl9=yvp<aNImTV5mG@Vzq8&^g6Wa$=w=V0$NZl3MoO>Z;f9B0VCL<AJgQ
zl&;L&SI~mm|A0+8m9PVxiDuT57?~v?3T<O;87aqxOj8-_;~+CXFSFMtK-dzdJR&!~
zmSZc<FGqxMdo<M4c>mG|ShqZ|bLDzNZE{l5=iLVezRthHv2Gk(1G4!^G2f|EB#brb
zwCl&Jg5=V4-4V{D+9pCG=-xK#87%kOD9_8Ag;LCNunc`*fZBy%Kq4rZhkwe0^u^oX
zzx@3>(b0WM81Vvd0*zd8=&Lt1Whr<JFyj+`Ehaiz>uGNxh5>#4{OC;%9#s8}z38UH
zZe`!Wh2t6Y>ho``7`5|2bn9hF3hJ$JRjpp#<E9u<Wrc84FY8Z$^ZAxEB9kYCg%{zk
zZtQ<s;MA$fZ+YxXy^t7>ic)60yaJiSrL4m@1Ri7fs(F=|l~IAMF4h9_%M0sJp9Hvl
zh#N)14#b~EOhZLwZy4F(#?q~lHwTPW)zq-7>el>K<J>BSf5aO}EwlO@4cNTJrJR@s
zv3lk?QLn#DZ^pW)Sv1zjgXs!OM7%Bx+Z*;DiHDBDxTikIWV)%$!)0{FIODw(#QA6_
zs1eLo<^XJQ*m~kBfB@;ReNm!LoDM$$dZ0z_-LFki;o%Tk8BkJE!Y2Zm+fxQh19kOJ
zXUEmk#g7Nwy8jr;Q$a5%vGu~S(U(@d0Ea;?7A}lY-C`k4hvV=3xO?j21ycwY(9bmB
z1?WV~qUo@Ccn9EuG4I<59@NKnkhth^G}v$RuVE~o{SlvJ>Q!UTtJmppydXVH&Lk}7
zr_ucQacAx)+~`z{XyqQU7|R$byC;7-4Yjzf;H}{oF&SlDX=W<Vq}OfMYL06=2|0x=
z8ceo#KP?`5t+lg|FJeI^Z`JSmWTBg4m`J|1hWb{{1-1o~n9l~QqkYXkWxjKNPg^mN
zy}xW{jJ7DEimK15#tH!KLxZsLE&-_~%#Rb!v_L>Nzp!9#UWArEnu7;9>;zMUsi9Re
z&G?v)g##wWK`9!&^_K<IjjXM?>bmT5a&xK0s%vXSp~L{`5<Y$lZSkN(+iM-Gb333u
z8Uuz+h55(QxN*c1DJwBS-3^iU3j7v!?_L+Qn}H6bI?4%p>|Z@)XEr9TVfe!3w<|8Q
zLke_VuOgAt13HxOO6tA&1E$l-XtY83_UH{U-~p42z*d7@1{b2HN?ljqy;YJJ`>0oW
z-T;^ku$rS|pDL=s8}8VBP#KW9wY9aFSQ$L4pa}{N4yI3irU<K*ZYX>3pOZpDJlx#!
z^73dq!lP_xj<Q9@SpN|B2DOH>i%YU+G)x6xQ`gqhW1z1uCN3W96q<ztf`3;aQ48W`
zAlK``8IigR$pWY<%uedVT|-@p3Ly8(>m1~yR*%E5DRLBd%X|<WhjshgkCFG`yq|9Z
z@io2Bkv|Q(ASv;+q`!6H5+VD-WZZjh9e48~@B%3sC{{{1T4c%HD=sdExiZ4@&U<7#
zcM@$-;cqea>?WX6>MGLc_;?2<pF1{dea|xJHJ#3ZEEyv1$re{6$K({u7h*~w&WJ`H
zExM&2i$9g-jx16I0AyYm^EOoI34{eb<xiN=_9O7N4;QBxAz%iYeyb!48yhBFY`_?{
zdz(Qob{%r-zfx_z^7r>sPH)NN8o~8LGGl5PO4wXsbF_;&!DYjyr!<1j9>}yAq%sig
zfssJ;q8NV=_0Qd_s90NU`6bX^18lioQI`lQXvr=wzm59=hr}qf*@T9EE{UmS$^H?K
zJArcds`Hg@G^?R>f=mzr3sS51-QOV|K!U4&9=XWWRK>MM=9yF8mtgt<J_8RJFB};L
z<+p0OFM;cusXkoEu28RY?VOBEGS{<~ATnmFkFAYL5pIl+x9<tO8+cjdGMWv{OU#ox
z4c`aTE2+gg^G%)8J^iYkYHFzGa`9yV`J{`aPWqYD$jqdKH$}sJbd!QTW)-c<?P#M=
z(KL%#<r|j!QlsT=c*T1`X6n%B=@$7#iM0&%ZE5|Yz6>2kdOAirR+4EsvKA?376~@0
znlb-}z4wmix_{rs)zzeg%t(=~WJJgeB_m{Imc27lA)_U;WD{ALS;^iN*(HU{WQCN>
zLgGGNb$uS+-|xTsaX;=q?ms^MxULHC_xtsFJ)h6>JdfizkJCaVPJhSm(H^322_YFe
zM3f(6!}%#GBRJXcS93q2)2x7*^!4l4IE@rSkn}^GqAa-uaHoBL$?M);F(i_Jat91X
zK7Eqe=8uJ9)jgjRh24&o1DW6q_*GbDaOcXOgrODG6ivp2t#^)>!B`4WZWc_yq8FBH
z8jhbP(6ZH2@EETK&VJ12kXA$W=c%cc)}tHKD!-6hATYg-5xA|#&nhWN$Y}|;%coCc
zPF&+FG*&S*WODz&Cn@=8qTw>;cab=p@${5PtYDFVO%qgjih*<pC4NiWIAkdRci|~g
zMF%fH(stZ2f%^M-BlKree;~_;EKo#*IJ{5*|6W_U{F-Al6NL%_1S1X2688UnH==`J
zWV0xQ5T7HNWhwA*i6ykQEMPa&26wH2p-Ms~PvlF!@tIfy2)aMVn)iaP#zd0(0Hyt=
zPH4{T3xg5?_krg!qis}r`e&m3z_kmT8AeSIg5%v3ehZ-=T@30h6jVerMZy!9rK+@6
zyn_eM*~KIvu+V*WX6N8qaQ0wI;82;f+(@kML5c^{Oz4%w8%szJ0%t~@Uk4zduI`Yi
zXpGor0Tvdt^zIyEF9P~qSg5kqS`>Ye>VIAlj}2Zj$v+fVVJ?FcumX|G(6CL&jURpS
z@81Un2fN~gnwf;_610Tw9v)!|^g8**uRR!TU_oxKP>4P#aZyoZoWL<6E6xIxhrI-b
zk-+;2#|Umz%r`C;w~Nd^<{c#t2HfQHcp1qnU#RVdYms{3`oYYr8HU-R6wWd08oz{V
zANjoF&Fj}*5Ov`I866$9lVCW_etmWpeQV|A1+FaITwHE2c*o<mODzJX<7{CCLTF&;
zmRkUrx@)J=fS#Gj)g8<tOQ9TfSi@CXb>Egf!XxGb&4Ef$9=D!s?y|i5x=+~lG{<N{
zMoVtQ1J*l|d(4dT((){Hhw>ciN&tx%ACN~k6aI{tm(whI>maR>kyh@DC#-2SX?xGA
za`BjbKj;W?MY>7c)q8!y0@m}zY@<8IM(cawr+fcCF<t<w0Q8B0jrtBtO2jF6l7+@w
zW<w^(bleHfrD~@bafBh*KmZG27+SZX?IZk);ADg|MK9ka3%V|ho@Vo4g3Jxe8p9!G
zX3lo2CjH3z%neN;;wiuraWE`!!R6z_i{RCOT7gXCaHoa+(bk`@o%Aq0Gq9>1%S-~C
zFF<`nD_z|WNLThTFkn`P02&2x?B%32o^P)C05z{%cOTc0BSy8}5VTjLQxiu3EW7}L
zV0fRkt*tl)1wnGmWwnJumI7~FL`sQ=KK8Z91=4eCE8TA^uDf>ch8`Z7y<V;^X5Dcj
z{MiWgY}>xQkd|)L($rKCleQJ?Lks~Km=T2xZ;oMkm<r#*@aSlrB){Rz?v_W;?!(5Z
z_>1SY>F#e>nzpt|-#ER|+|m~YO&ibUe{BDO(<@?ED4h_}Xs8Zx^-xs94Z;<cVFd*O
z{QUICtSO4^X2E0MM+ckNB!ECxKi%5vKl5;o!{z~WH8GPTIoSkH2hb?yoUz~6W#3^6
z&^={45GdSJ)rZI6fyLflHIkmPe)snFWOX-06d;#?S{Uf*MbU*_UXGr|xY$^jDI%j(
zE_#UFjdf3btei)d;^xhp2wHTK-jI8M#e|5$pxAb)Zjbaa7<0aSX^J=oO+QfR@mO1A
z98|sxXvqUDW7G3Axv=muss<a%4s>ldU|M0{Exa;vm8M^tkY~XH%MJoI3=hNB=L)pl
zwQCo~T|leB2tL$#-q0N&Qu5OB^|_I)iIgL2uu3-!B;`ay7P>6i9PBq8zbV5{hyVmH
z2PS1?fJA+hGs~rc#|$YuO~DzL7yBhLv$8-_Wa-Ntgq(I6upw{mpLApua`Pv`Hs6Up
z^&gsSEN+>U2v`Q+m1c5dA3WlhKG`cBU@NxQhvSP}M12=uRh83hmP*I3rE4F2wZqz-
z!E5qWjTszF_^i+nRKVG2Nv(Z7$;15QSYfBtyXvfyEC(79ie`}Tq%d-z0)e*N9mj2P
zY6PklL>SS#Y(0H_Xj8r=zUvF_F47zrai(1V3^*H8SlAMC>H$+WE!<J(@`Nq|he&f=
z=Cj3c9&T=TWEx-#!92o1PMLhi6LhpfdMJ!Jepv>eQTp%4YaH4zfIklEW*qeg4jlcm
zy8|ugI4A`s=BQ8qs7`<Kqykw4?xgyM=9U%@WQm}J@WfA5k2+Xe6D*s;o(gIcoiUtq
z5lsEM+S-wqb}1s=0OHPX(L&uleEk@}J@5uq@JbMLoXpD3h8qI2n9H3j=l6!k#`Z#X
z2pcPN?xMCK5O{clI*rKqhQ7X4_+R0JM3xbLBLI{2IGNKRqC)`v1|bPsdKz&KLT?DU
z1(C3hESVtH#B4Xro6v*3=eT`b6PQCAqf4j<0RZH*Ib^b9SJR{8!*j*uOUFSp(P4NT
z0tuljezVKW(a{I2JT{f$nsP^HC&16mdV6jj^hXMb<n`3)(?K&y{ya;NyDN^+0xOB?
z+slcy$-@&Uu@D${bmYWNX2_fseg_f!>3+8!=x6=?cXTrg%%@~!xnmO77q=c1=n&pw
zE2B*)Fbe+BV<_m>Lb4&3LU5t=LO){`AvsxJ6DwZ$1(h-W(jID#yt*2_G|+7nKNyF?
z3t~X*g3K%9Fz!Yk*m}9&+{D_t=_@Pe(5GZ5LvzCSD~}BiQ|{mYTKomOCLRZ({^>^!
zDm&4>fr~6e5}iO9gpo54c4Nh*OqP!-P*Y1l@B#|Q9cL}oZE;LELg-bly39>Lk1t*f
zBS&nLrcH-+8PN`ZOv)&hLZUMXonv1;pB4sEoCyA`_eE%ff1@g_@0>)0wuSb1{rkO+
z)mE{eVq95&)aTRrj4!8p#^0BgoLXeNFQ(!f`SP_!obFoT$pKPNiRI<zc8MR#qbHWf
z*6(P)oc^A-NAj2}v?VsaE{w(;r<)oB;5Gq2Lb%m3^+|@KArY+XC@zd@K^}4%oP$7o
ziXOg3!!R;EtlDvRZa9rnVs0%pHQV)qF=DC@61K&;IsB~2sVLwhfUHn8c_M=@{1P=_
zV{410{e!StJ#jVFml~G|txrF|?kD}SYCoz^Q4kh)kfhflam2zA3hH<m7x!y*^{%fk
zJ13{|HLLTaZ5KMAU`NaR*fOp3`toDfL&3<r;iMRaM+q;`p25LK3@%fcPc-%33jLyh
z#o(J%GP0`bG`qB<s;EeeWWynN-bV}D9Ud{5lI01hDQMC+8e4<KzGALpmj2I`EHGmT
zJcOGu(ioH_sF1slMZvrlDh>3w66KeV;9KOMnDFh`SCoZ(s%VNF;8XmgM^JyxVOKE-
zS<ZgV<I3OG9B@z(a%vZbPsn$wA<)Ofp7yxo?Jr*3{E-7#90P!%T#!b8EOg83x`OjB
z<L8Aj9<N#KMW39qGVG7t5yjz)cp2`EG20C=0u+87xQQTUt%w7^*B+OO;7WJ+yiE2d
zOEa@VJif_EtT#KUZ#a6fiF1sr`#yfG(LbH*DCmZjPYk%LuBtj8hh9C*zi!2YfVe5`
z&gw{tQ57n&z9bRQY(y=KM~0uzsbdiXatO-a32u~~bWgD9+}sdYQ4<@C$O0m{jCu_4
z8iw6CySk?CH%9A(L6J2veii3TwcFguK;!;*CYHxzPO~sz%%n_(crqH;A)luDEsrSr
zTH+uvv8Azb0`CTn8r4@+V}nk}!-su1zKI8cbPrXzV8)z=t}c@OQK%__B_X%li*8ou
z44Th+ArBl<&cE8a2lW*WCm)>Hx3-eSSVT;BW{rOP26J8}VBVKAnhIp`Lx{)7kQcEG
zVBzsRe)D;V5uuJacI+7RX-tCoDyXx%zr<zx9boW2GOTa&$|sOoQq^62qh*c7?`|qx
zbMng@DKt?u_r(M((WTmcE>T<O_lao_1q+3-{-%z`(;KwwjMJY6z67fHf2AF8m>$e^
zEL_**FDR>82A%W|`K})7;xj+4HS@hm@b>ojFlbadFgQOuS{D{b)klg~DGk4rUtZkc
zmGP@5+$_;E?}(<b;7-ve9KL>j;a5~*06P2x;>%;1z0S$UcR&7E27Cf?8XrOU=R$e}
zc+!1l1`7)d4ULWIOtb>ER!L%*Ssx(t0OpCYu?Nmh$h9DsM|sfI*@>F7tab4j6{#;)
zT5DSyC>68<q#A#LIrD?!lviG-o&b?RWrIK#y5x#(0*zeb?o`wd@Cm?Id^{7r1H}g5
zNWpx6>#=2DP3U1bIXGx(XwX`Xq(-CGkI+-s`dnCySuvq0{m`Tq3Y`fYGR~b#$<8KR
z<pTDPWY!v7zkVG}BW15+h5SUaKOs(urJqW>v@HwIpQ(RiZuE4wej&)@wG6GJm=KO0
z3)mU(@E~|$&sd{+Xm@Rgn&(;J)SHpJa9%w~az<WOONgPotc*WAq4UDl=C^O3;<Y?K
za^WUH_yTV!E^1`3a3{VoF<y@}+lGAsHvj_iHkIFsxhW~E@Ls@FfTvH@Au+}>fgQ&W
zW?$S~1V|*)CzIa-o$Tx^s`!xYV*mE#%UiVCSfFwhXK4rE!-d}gX@{GPWEKrQ%dTT%
z1f1H|7mBl#N%~5%y0wkXnb#B8Qq}GYciR^!=Zn!oisOSn8uzWh_!y6$I#(M+U>MDN
z`0z30_E@3fuT_-*MX0LcF&004nt5SksJtD>6*|2SdDxqo)m5#g5akV;L1bk&>G`6D
zxI32F{QZa<7dJPOjTR(Et*zOb8TW7Alm!mT#ykdXCm&A<``n$dgoLwiZi9V1ZL2Gr
z4%!V5J`!Vbkgwz5=?qs;QX=iWHwWcU((i9*^Zkh~qb;%oVI;2S0CykMVmJ_x832ts
z?0lEjS$=yW;gI@-2h-hPfiSb@B-dd#r>j@fAMt%3tZWpx-GaO)X2BY?RqW`tQ_ve*
zT5?<l3Lw_LcwCkO2{djq9s&|vq&p}`J8ShWDUmVb@PYvgv?5(yU8qRmh|{zXPsPNf
zqEkwZ8R4k6W9u($^+7)?Rw{lN4KnaCL>C(Ro6Nti@hRWNi=cSKzoStYh#ZzXj<-YQ
z1JBKVuON&#wU=sMzvYLN2!3nv<yFSh9_fc_Z;9<DiAZA08e7m=y`uo-PmSlwv$E*U
zFP`yX0c2UpBEP(U5aum7m7C9u?7a1~s7M7PSoK*%4(>2TW?1x49<#~)TnkM+OM&y8
z(OIUN!f1JLG<K<@se)*u^u)B0PIjdH$P8MB!un!EetCFFXCylY!04lN0$E*3`6w|i
zSt;-PPUP-nWWP!iDr62~IQYIqJ>i>oKHb#PeDTk`GYQq1(2zPe?Hl=W$VW3^y=N2_
z5#c&-Ud$P|P46vF68gRL#rD4PR<<58srFccyLau1+rxwYu%b}k$WBf!xcK7M!#;YF
z<+E@!f5-EXn6lfjay3~W@!dMyjA*!0Y|w_Zi*m~zk|FxBQIeW{*na7|FRmcX7po|{
z<*BWs1Kr)DwS1iA$Z~8fI%QI@n!orUP|EUdUQ(@aU*Hh)zF-b@pTq?HCXxqRa|3<c
z-OI~%-jbzEkn&Q>J@5`<Q~ednPyf8Qz2nyQ8;-khX%}__Sn4~Ri2<SWk9c>BV&0-C
z=fu#?+wg=ZCaLTjNXQRW=c<Cda^(UFRZMp5{Pcrsw}X0NRu=b+<3BfCkS;<vsS*8k
zpgde$fRBs%NGM>;OH-E5mlIgrC}d|zX1|4Iv2l`c^`gSWy3F(%*>}uZ{dWI?QS#kR
zphGgfqTB?AfAqVL=1an47LZO$;+WRTos+`=jx(iWkkK_8TQZW|$Uyds(UMUWW7kE0
zbW?SHGJ!(qnM(HR;K|v8w?1w9)PLKP>G=nM(dZTA6RxHi%0ky^sJ0vS&X};~#5H27
z^;paWoLLwL_i&JT9Hf;()<Kd?S=6hdq+Jo{m?4}L7Gth+s8Vcy@ehe-_JfdrNIt#7
zW`FZ0FTHeRx*J`!i<4VOZhZH{&QW!Isk`UjpZgV^pP!FO;gba=C0Acx+BJKpdK(Fe
z@Q#e+{0{WQpmSVhb0sko!Ipg9c?0knzA~qBZ+v-S{13^EeRtl!{m}H{MRNLa!V<UU
z+K~4VmA~$t@H_7NX9fP;|9*|jO2NMW@0W}_@d5w+QfJ2U`OjCx_W%CMf9}Qq-8&|n
zbs;t?ik6O!vr(0JRto4IHZwBT(75>?-qa-g2;7By`G>O(EAwj3H;{CmrRk;V@>Y#>
zL$Y@6`6<}J{{8Z7g(czZ(91Yp{_{^H|0kbRW7Lfh4nfMO)K5*k(gQ^<YAKIJ|B8AP
zR09K5=_Rl(#wZh{9WUl`lAb(KvLQ@qNk{}++S`FGKIYRydm~UW4Y{~Ki>d4G+!Nya
zf4*9<tN!!n8}kPb{CO?nACe@Rh{-~Kzqt0`METFxe}Cn_ckJ(4`0s-G-`X#ab>0&m
z@gyW0#8jk+2x{kc6nF2U<l#~l5#e-96q<1`jw|+m)>V@V0+;st{=dHggsmH9dgEOW
zfp$Y#0-$NU>mc#HrX8byMrzqc!)>9ppcv)oK?Y*`s%QV*#_E+t&<(tKbJjSw`FQ1@
zjr~mSPjE@<4%MZ4-P2PBCnFadQsULdw*CnyH}o4}k$#Y(>EnIO06<;-^&9cs8%!*J
zu2g**L*t4Gf0(QM=ZWEENwg>bzM@E&WTd5u$uRmVqKQFkBqU=0`3^Ci4f)XFBf<0n
zCl$87fVDM=Ozv%p-Nf&U*n;o`j$~o8>Gt2t;rtusY`OKQY=L`25g0n!K)m3Ms(-Ib
z857;)Q8-k5fB{AGH8bLyGSC$gr(0$k09qh<hX(CnKQdlP40Rznc=YdC)@l51x867S
zksuu&IF9&@>)!nLy8E(0nxSy&@}jYXBR5{?^}jDz4GiVP#pgF{tC4$Hz+Zs0kGKMo
z2mAk?iN*R?PE?{~LiM5|OH7R;A^Fd3)|noxECoisu_W;KZbkij|8+EVbtk0Pwa`(8
z;nSppe_xRD-xu7!!RKgfYz**&bddhfTm9!dk3cx5rLiXp|NDAB{(ZfYy*sv`z>O>;
zRtZS8$fuagiR&i&doOD?dy3cWrsq!rAFb5$_a(RfeX|QUQTO)t_g~H@I^)KPVHsFj
z|K61R+xhuN8w*iV0O!!^`}-6+|K68napZ<Ow`_K9UtM0#grV8x>%_0hi2uFSp8JuP
zL1lJyMz1Z0SYflpGJjUer)(C!slO{FW$f=tnfm)&XS?XveZCG_8EQ^!NU!soA7v1H
zrEPdazt^8E^ocfgY4^Q8+d+M;BG$B9*OQ9#^y_6`X=r8Y6!i&id37_1wTyyd$Mx@^
zHxq|e5;hoEcAk;(k~}b=n0NWnw4s<7o9@SneL|L~?>gRN=2TXu|5c0<BYEB^tJhS5
zDhIYkO}RY8CA>{Rg)NN;8)og`;DwT3B#W4ANb(j1Eoc_MD?*8bV1LHXs}cW7eBDsH
z2_Ei23}b`v-+#b=oA!+SPsp6V?e;%0W9=X@E8_3hL23X0iL(FwmH$WYScr>(F7dn|
z8A(V00k*h^A?IOXR4X60OI=0xyNHm`V>TTvt*?WX(P40S8-Z`j(P4Bp!*n1@`*$Dj
z8YeO`&r}K=Q8(I~fX%_0ms?^_VtYRN7l6Cx<q=FQ1WM33XljOg+AbHs{FbmEmi&#M
zJ{PbVn=LCdlm8M*bs!tG&xmpn;{hh8Q21l0p0(GY3@z<ZV|xg8&wyaF2yTRp@Ox!Z
zywL<hNcfZwD?WeI4q{EtGQ^@>ffLJ-8Dqj`ab^HQG^s0Jw60t!KFq@r2)q~uV&|xW
z0s}E_q?2PQlp7c?V1Wiz!a=ySbYfw}{ucllfd)HZR1_h%uUdl&2PV0=2+{{UOmnRi
z-tFKX&#MFgJl%(Qps1;(74hI`4Xtze0319@w8!DUKc%gtrsf#l0+=*zj@oPP72P@R
zS|qSdi~^u8S$=()ni80ajEUijthjlYjg#~J&@*6Z9%n(uHT?XE8#A9@q6MYqojID=
zn3$rn{G5%wcrO`KJf@TTiq5waf6)Tg032@v4uJg?tcjKA>9Md_oFlg2jS!kYn{=cX
z@C?+={S~x{!idBN4kVW3{h9r7<BpRf175@)aw|=Uiwr)D6qeuyp}Iq`D9PK=WB`6h
zMutT4Hy-zVzDisr%JosE%jeth-bs(-h}nNP!T~D4L}(-nM|xV?I8bsZ1sN7jh>69<
z#(FK!${N#b;ejp?iXGg@$GN<l=ez&na3y&P*PVeKgP|$8LJaDZvHx>mI6?V4RXdBv
z2i8su4m=*-TetcG0WP5rLB9=*X#GGmprr|<H)8LzXD-MopyPtC5nxS-N!%q8ii~}?
zpbf{w%<M7zq=08Zae{*oTmeiKp&3z5^F-|f{R2R$-MjCBKZ5<m5FkqZqwau90P3UV
zDO}0M6=?|`K0vB&jA`%c$_0MF%{}Z=4vLkFk1uL<7;1-^uV3w9?v$DehjTCVRy$XU
zg4cyVHX)3HM$3@zIOYMO*~UH=Ae7%OhIg3Uh<62F46BWGSU3YQGu>bC`@*mw1{9b9
zU&PyDJQ@yOTwqGdIgk(_i;a!fK^NeI;-i#a5mtapEbK$yW@&OF9RLN7-U_T)Xnw)V
zHKlCWfG_^FIAvyH@?vxjaAbH`7!br%Z&X@0e(aee%5qEa-Z0~RR9ae!E$@|$cs^>(
z%*+geitye9qH)F5_0EBmii%bA8sV0MCq;g-xC<p4IMMu1cnmC(xA7md3(0t6C}N)O
z|1dcC8#;^qBKGG2JJAbC0Bi^8fOQHGlX{)12d(_h;Ox=k2kPx$1|1Czyv(OT59NmF
zxc=>#0)A+3Ax4ecKzs^V{NBEO8z`dWKZ)%?_Z8rUh(Qvvva&&2(Itvh?%jL2lP9-O
zmIAMXMgX%NAC$ri3oc|-YJ@>kEeRe2obp<Solsap8X*Y|wMCE)Q4dfWM75k^9heq0
zG~@%y1qz$7MeKkfpN+wxp;{R7H?@Y&(C}>n(2c{wV-*r(0NN^N;Wu;^`cQT$b!+QM
zaGCxY$YA@6Y{EbJ7XJRV7s50cK4Q=(Xd?!>2FNT3lWmET(o(C#=J?5AGY!I_3DyfQ
zMrrf;L_mwdSs@nBqPSRiJW(qNw-xIKgcfdm$`)HhaiF88&am;pV5$on6?Q~V(P|VS
zLGWLY!H)A_UoS82)Vsa}7Rtfs1fH-B4pq~&mVVeh_wyO@MLS0m@bc#&OjmpGT}K4?
zE(U!atS-VyucW>&u^qnQ<~a!vYbYrN`ul3>>K-+!h$tZZ!i8q$U(sEdoSK3^tb$uJ
z=<=6xV%T}C-ETWPBcA0aBruy<mRvq50+tnqpBm(^Cl;X4BknqV;odnzXVG{F66f3`
zR=Z)$_JHh+np(yLT1|$XM8O2|VQHLehuV>H6CIq*0KK!s!ln^c(ClO5ia$v?aYPnY
z%a9VsqZ}c#3|jyXLZubSZ{Nf3Vgn>JJm~@ClQ^l~Uw8%KHd|f{RErw-XS0Js;r;{M
z9i*6ogrThLO|I6A9q!vPM})w=*SH}32sV)=PRwW`)A=rylaHkCxe|P7E?h`-eveMG
zR5y%9?g!S7azwd6g%sxzY^eZgmVystok>Yfe$lgZY6pyb;noQPe7F~y8)`~Q7M*4$
zatN%`%gbZW)d2<cHy}R*dJRV`1mz_?O98+%Dj@Y=US38knc`J-kPLv_Vg+`wnA)4d
z^9VZbU@y*{Sf%Uh>rCWU0I7X`FGT|&e$YlsOA8R!wqP`_Y|P2GR1H`loarFHu$>NE
zQv+rLXE|u0l*zyB1U{L8&3VVyS3_JYu<6w+c4&R@@Lv`})ZvM{M-_oQa5nVhpEDgG
zQ8%uXVS6?$*lX{#^A&#6Z<Ff)PJ%8<H8^a7SH;H(K6*M3gHRmJ%nm8N$*)-r%nf0Q
z!*a*|(4>OMl9<b=Bj$?EIS|L#$59tAoH$Vls}QjKft90h$_41mF3L)@nu0*q($KIG
zn1JIG=t(ri5`6=vmSki>SQN$SrorZ3zNJ>UZYV4qruCiB5CkX1p1X~t$B_r%JPPRd
zRfV0PxVS#1U`4dy^I?yLCtboOIB=+qDDj><!|<W~WjfS4CME)~*DE=w4DMgRC!AvU
z?%l9-!g<Gw`;wDmZ)*$B6v9%luu#a1GY21-FerjNG^~bP2Z}Jg%B#7ZL21+A{3m09
zW@BVJpUw*2Q^>OI;2z3vxLdfQ!<Vc>j1P~#+|P>GgKpmxKfmpX?Q~xG*iO7W&ptBH
zu+Ju=<{2pfe+)ZMR@>L2@enQkT3JE1MB3i%o#!Mry0C&g9ntF7?64d{gOSlA9C!$>
z@OH-so?5|hKuGHlI9^qpOjVYZ9R_WRT#G6iLXil_zaWG^vuGq=D=)@$A1%xHZkp5~
zHiX+>cm8DVjo5_E&1dS{pyC6MOBsg13cX?>mL9^BW&Cy{h+|qJxDgb=wxrq5&&~A$
zln-*ID4?gO-|^iK3=th2T{PU*b#*iH^WlDEXJz$aq{#YZHCmexI1puEF;5S1yxFH_
zOv(Ysjc}Gi_X`0E(wQqzFpbp%K7jB*u1P=TrQZ4TI~cg4Vdn_OJ32Z8)+S(}00Qt^
zVx&(Zzsarhm*4HZtrGX}Az~=l9mEM3WkY33*v%D)v*Ht-R#2$++4%VGoeM<pN5~a0
zp+XjNzmP^oA<JD}-d^N>k705>K7IuwQyK>Q`)T+N;p!r&r4l!gK7u%xKYh9|x4;ha
zl`oy#rVO;Vphbj$0A(h7ctzkv(9b6u$fUhik731BRSneFo5MKTevlqICpWmD(nd$6
z<I&-mM04EOoEs=V`;F7s8mEaZCbjoZs^iZiExm!h1#uX#K9C@+P+TV;%+m0<OMJvK
zKpPHr6IUE~I7sh^DV}coC@DjuqBz_Lrhc{U0{rRVJC$<amIUQGkujT4-mI#sLaIgB
z>NJjqz6X~}G+kqw221{}-64$zAgr*n10t|pzkv1%uHykaULV+q(;52=^!yEp7qLkT
zlm+@}WjMxrZiPUDhX#L+pT11s;t}%bum9e%WsBm|Q)4&+k(RIh_-5de0?*U&{gkZm
z0Yr03>o{owAyi^vxtTk$KcsPCZf@u9-82#q1bY4H6Xul6K);un#MuhnG8K?~xQ?Ce
zg|1WQz;Jl1B6y+IOXe(;R*}WUO9)MO=wjmIB`^a4IK*m?)3s~YSedUOy}5CN<i#j7
z4Sl`69YKACR(nQQpQO0HNi#F_kF6Wl($RVIP<R@bDIbUpn3#A#LsAflAbLz8FgMJR
zqobug`|udIh)9$n!;AR4)DS@lqZi*z&x(i!0S}6qpAv(}JWYGLb!jwU$cj8VO`D;_
zLZ;VSL@4iRY(m1vk(*E_G3;4{4e#nmsyX;odrj4X+%HWN3&_f*RtR}YcaVkQg(YnI
zTe-NK$&MHb<9{*x{B$;YZJr)ETVeNsuc1OUhX3Zb;`i4Tc1r8wSf{c$d7QkpvVCF!
zTYqXe6|F8$_dA0_jA%n9VSi<Bn(0aZE55C!UR|`9C*og_*?sizEHn`w1;`03+C0aa
zC>v~X*cQ1Rr2L@!4gNf4B2tO%A_e>VH|)U-n@y-TTzk+y0WKW*lD}Kva*+iwQBlhH
z<p*7b4I(|%fYJkBhE;CP$cq*W?1ORY;<W93e0mS&N}y~);+ir_Ww3_>RBTpPOx@H_
zZEb8U!)Q9HFqj1G>@7vYjG0F{bvU`cp@dBE#x8}D=iPhu^V@FL4Wq35i5~v%omqRu
z1VlyM02z(YE}>&UkngO%re>XpnhugZw6x+N_6mm@8mXy*ge&s;kPtQecd|I`1Ps)w
zg*xD+$8dWe4s!7L8ElwTnv|D!U#gx>rsmgwv;f-mrRlHpodSIC-0_4>^&+7GdEZV2
zrRZH8QTRf5L9j+fRxZahTG8gP{8Cavg8n@euaNuEA9sLxATryu3-rj<o)s4(vCZ=L
zTgOmfq!*{vYT<2wML9JU=EZ@#=jlq+#3U5$Xh*<gC<u#ugZB}rSx87bq%I7c6IA!-
zqO0oZ>rOF|6NA+7krJu*usNYorQ!Yj`tR!vYxe)x_`<A`kV`>Q))asO9q8GZ$+Z+G
zS66ct%mJkoZ0to#A%ogs1DjxQpByvijl>(Y_w~HYL6rn|;0SpfY1s4lh??(wYlXi}
zkVdzmnxfxsbgB25!gAj;FzZljEzZwF%}V)<X?uUoNY^=!p1d|_X$rrg3-<?N89(_y
z*IZqxC7Bz$9W^QvF)x_p!2bGC{7)2x7Y@CH3COXLJNNbUK!7K&7du?(f&>QRMUba2
zs^UAC)ph>q#Z6v6xF|!dPT5H+aS9Sn;HhrZn@3~R$^GhR`%NXPQ$#UU8+#UlM5VK6
zfMGPSIv`>{ftGKDCS&^-I~rn}dWyVdxR9%=s-|T`M@I{N#wa`iGx5=fyX+lRuUEcg
zM>837OW489kQd?3Q$HJpm;|P&Y~=LC!(`FR;#R?p0aqTieiOoa|1^5*s01$Ktf?vM
zXuJ;2fZuqa^wGvRt$rgw+1c7Q1?=0m4|~!y{_q)B(<3>(tR)xTK%{U^g7ju?ovxhR
zZPz`WZ{MP)gNPL_DfjN}wy<2K6(H9ccM^rI@W~&h2Cr}A9y?X9w4W+FGxH+T@z+;N
z7&6R8{r7IN@zX5>d*An}O#KXIB8-x5y8~@>&iR5AVZTdtbRTs-xT7huS~>kjr}SQY
zPAR^TD}~5rmUF7E|Adkw$T>g`rfweXq0gT`*SX6hb=BU>SpLEhEmuH+$$mwfH*cn=
zbD^O_;_*zI=O#^wsPAMChk*Hse*5qDCnX1^e(-u?PeLaAIKvU$MyM-JUS-1AFW%eE
z-@<e9FSFdPv6zd+JI_+e{y7#*$_!^v{h^>!8ycIOL?@BLNEXun(O|93m@Dp$ZpG+U
z_RrsSwX_I}!!%<P(pB?~&GtbOMO6Lm*Y(zZe4cYdnwp}L)D11}_nKd47f<$hkf3PZ
z(HrEz%3t!1Uf@0zP0;K`QX}3W-yiP2L3~_X)+o#u7xBihv(a0wIceMk>j!L22soM?
zCn5O4AmhW%+G%;Yd6!wDx&FXGv422{B$#$>4W5d4&<!phdFAu6UvTtAcFaA6W$EIt
z<5X_gu!ud<y870|3HshUcebEc!({~Qgu>CA-`4o^)_O*CN+u^VlD&+FSz#53{=-9(
zUI6!lr_9l~3&G>!BL0u2XB4-$n6K=U^l+^VY}{G?##e=X#oc%U_iz5@A9C`@#c8Ee
zu~$%57zg*lTvB<Dwv%W)(tm&L-OeZEc0N^2zXa2M7?JHHwRc2!yUZC%yY4kytWY{y
zA^`@(Ipo7}KTx|bL$+JD>?^`C|BoB}2!CcFVu6ra3Z`$s>KD()YON6oAa%7z*l)Wa
z-)}AEm0?H%fgk~EKc9xhe`E6R8P$L$%UoWDv>~Q`rU`pqdKAw=+Tp9K9v-Gad$TiA
zQ*oFJT!MXf1P+x4UZ{kznYlc6_siSce?<z%J~x9W53zfChMl$bQ&lbzc+#W!wm~)b
zf?%;^_!819qGE6yCr+M;u*Z9U`XcHeJoy{THTXENG)9v2M~b=4e#)<1f9U)kwH<U*
zlT$>+s(k_Xj)LqDnw5iQpsT<V3B)cJ0%LY*i#Wqt{9&;k=NlxWf!c6dCG|+Td2^G2
zy}uK>;Tsl_vO&K1w!hy%BeaiMT|*-lqo-gvG^!Nw?OtMHV!dSWvZuGa{Fc#la2-S|
z*gLX!t*z6jyn|Q3#Smqpp0E_@vznI{CIGC0l{pV!u+gHOK@MHslP4aqG!5l#8h*NF
zcUhFFL@{;?lq)S)o$vbjg)kn!AD@zwvxXA|`F)sxkG%XHCh<kwdCz%%9|AXT?Cp(R
zdyEtt;>Hi?$_5E@GAbeX{(TN<U`fJeC0OuK)1l!uK@rL13SJ3<y!t0><y)QsIXi_k
z7u^WS$u|`T;6gWd$On5FE&--YuWT+|@<yWp($5_nsZe+THTvF_9SynYLDChCEin+&
z0-3-e!QDH9BN|O>Am`E?vAEYTEegdI@Fdtf3=9kqO{qHfYin!GX_n77t&&EI!}kMN
zqlb8OD{wNi;P;`vckk9*|9QV|{Ur)NtYudWxMzL!@+B+WOq`#IvGlALW9|#=%^pTY
z0S1<1gZya)#ky(aQE2jX8r~M7=ztCiHd9FHA!h^31d&fKM=jxkLrxy`X%aU}=H~ym
zV#zVWk7Be#GdIG|r-c<1ni26R`XFP({ezOO7;qTG*eEpFGRFWi)U14F^r*laK@rat
z0BHH(h%<mg6xW1%D{pUGn?3xnskE|wFXP<8^qZGueKJAl$AW3n8(P&uG>%F8cnA3s
z)`_T&sF;2qVx8tStcZW~sMMjqp$tX#*_176DA#!X8;RNn`317R40EixaxAS{ctSsX
zs1LOyyhu@{HayRBf_t8+3QkI<zqUcm0Z<Bu9cI1;4J(oI-K&A@y>?buWeaL$D1{Hc
z=peT3=0EUV0pLVTHH-+0LHbQe`GUUZoQ6hnd)m@)-S(15R(c>c767Ogt^K*6(WV3X
zTe8VI@di3Uk;})gr~7kZq*z0PB`Tz_ZwaxnVkj8ZyeSfM@|~toRfU<_APBF*=#b&~
zZseGdrj5ZVH2vK>B2T9z^FhBYx*T$%YjN{|XPi}0DZ=E>oIA)S_<tOS1e0Xv#dRW)
zoa_AXK`mC$*ZCCDe1UT^@SGwc^C}!fjI9use2Si-LJVPK4&ij(`OOevuwUZBk&%TJ
zpazz4^k>T;&cDwj_AE73{+XPQp}7a306abl=K^x6#9H&_Fw7?HAe)$T1K=82WNQj0
zi~+8SND0|{2rL~9=)cEqzriPQ`NLM-peM=`*szt9oyQ1s-3wl?MEa?R3CD7jX}3G?
zEMo@%#D6uYe4}v`CqgeUO#q^-GdO(-Gn{5}P9L<DjJ$cQrek10NatZB&KJ!)!iV-8
z>_E7BtK-(xH~uceesL@X7Kqkbc!Q}9_uAUoA#PkBso&|`3diM7&~CBkKB7m3VrXOp
z&n;BZoFjawim;6z!hPJqAy;J@gzl`KEKxcpSs<f;(31K1rFfLwFoJmy4;%ns8G4uD
zPbTJNhq*FH7}3PQPNofm2Q=3~=B7FF@jSvpKBVj@V*nXqR$NmQ#yekuT9L?lkaV-1
zP>-{I9ZLzZygBfrn%_VAmzUk<SLw|?r8e+6aZKPRjS4MQQQO$vU+xH<qWPFWIC3*f
zu2z1pxemp@mmM6b0bT=NGp|0p;<=8tQ1p+SjEy`;i!=%+`v(9?$W;LB7F=<+v$x0e
zs*eyh0@`Eu=LeZ|!zo6ZsF4rhHYtM;Vz&Yhfl32lBZl^3Nnl6HODjQd`qztp1j=XJ
zr2PB~2#q+;V`I-Sb2p66A@iB66k6}&gd#l`Kpl>5PzLxy!rRx^7Z}*dXkTwi9!!+<
z^qy9zP>|>2pacq-o(=>h3KaoB?Zm{RYr?313ZFa)On5^}j(ptIv%p^H{dU4`KU1Bm
zwWa0GhbL84o^V~a&o&5L=IHL~B2@aw^8fM&kVN1;0o!21;^`Uk{e-13PM#EM_JcF!
zrr235xI`t^{>nA<i#8PgMjh9%24E7u6(P@JWk9RDj52pJ{Ix*M^SgJqY2>A06Z2c{
zeRNS^yC8E2?6Et5i=ln&gs3Pz1H;_P8XzIi5=tcKBMo1i%YAt|Xkldyc9DenM@x$>
zI^jWm<1Q?+D{&pw9LWx99Y;Appvrhpq%-&~{(yChSdWj~+uIAugCbile)4;gjK2Q9
zQO3n71s9w}F>}4hl&f>?kdyxXnhl-9*kY>xpFC{k&)WU_eHrgRNyp#6JUH;Di1_=L
zvF(4(wZDJ)@2_AJ{Qdgx9s3`x=>OdXqY>q%2Fx9Bg<=7rkT_@uY?yGt$$(K3)FuY&
zOGLcAdI^IhvE@=TvVUqFpA~%I<3-Y!e0R_8fy;&kS3@P3tsoJ0a5yqIL%fC$Hc#*r
zprB=aVpE8tJZo@O5LTcuqY=~HmgF_kHCI2l^{>R~&jbGNx+k)PzhD2G+x~QUE-%S}
zShS#khbLpMEHCHg9i<B6We%ZZWNe*S2%rXbOfWPikt8x@$Xd$EL|p>WVvAVW|3r#p
zJ{p-A+SG@gP$6OHXn%!x;v*zIkg&NZA%jDsKp%eWE$cC65~=Mw(r>sCGvCp5yMsA8
zDhlj<v4fY?aaeE}7Fkn@jU#^fH_8U>V`MBhyo4}$`EnsDC44=3Vq3|he#%Y47|zyz
z(&hKv-LzWTZ>}IgAm6bAU5k97ruO#np`po-A0L`{SY#!^oS&KL-#U&Egi#KzUpjXV
z9Zd`aYKdma%cAkXB&6NvQ**Nydi9V*BP|KYK$xko)+S*=ZUK`_d?DoF3g9S=Hp@Qk
zn}Y8^Od;Nny8xU=D23a^*hb{NhC0qE<@ZOAdS4C=qfel!5wpA;*Lm7>NX+YJ-KPUw
z0lbfOPNkhL4&N+I`%Rpg@B@AM^5hG%?c2Ab`=%s!k4!%%n+_2UJQmMyZ2I`|V{mNq
zh7?f{<FniQAqP-ZQc7V<IdNMB2F>8zgDRsjiwP;NejAL~+@bqN8B9E-f`U**WWlcR
zdw@eAvTRVt^3r*oWD%iY^#J-Ie*Z>OA_C}`DxM}WSOw)MA}qT<iw-QDI5~wb?L$)6
z|2+`RdEkMm)Nm&->jnPL>@ywGlE9eJdjX$HLdb!;&>V->dYFg`Um7R_Az9RZz@Fp)
zkik1h7DF`&y@{n|LyQ=7F>AeNpO%z}z!3ceF~9|=7<T};w+o(gzROEnLz{}`W>I{?
zCKA^o<-HaY50q|Zcv!&X<sEcgAw}@_Hq4^qIs~?vm|F$Afph1mLK+Xf19Jl<8RldV
zUQXss6g!pPG&ckFrjaxP-H7Q|4*Bg>bDuhU&jhBWjiq%8cbzGaA<44-FemEcvjG-m
zCo?h?0B%JNUVZW?W59)fDu9w7lc4&$xA(mTZ&hZ<CuRYH4{PG@!vW$3UuiTnnGNtB
zKYk4mU4*==)q^b5nThB>9jJz_E8*ZJLCn)ZBR4!KC*!bu;mLq6_!i2@JV%}&A8<{v
zT1F3#7{$yG-U#R)y%1px59&l-9^@;Ak$8jVH;vr*RyuUfyO#<EfZ!wOk3-^U4xAUw
z<?<L4+JqOYaGhf9?vQ53$^@>kgPa^wL#)~`UfT&P1Vt6bD_sNhjT9%+Y7p{}tsx+W
zA(G%Jl~X3$w?}4q*zZ8odqXZ>-o5|qpG2ekQF=WUq5~0?$kT5$`}eC(RkaU0>7WMp
z+3=p5CSwiN7){R0BTyx9nOCmt3JpRognmV&w+_hPk$S-?#CFdA+)?#Y0QlkVN(Fm+
z@;XW&1rXg}qAK7bc=tRtYa~Y{*Flz_m&Yt%3fb#6ukc5Y_V=qwy|Ef{6Fq*fyqulk
z=iV7T3%zrPDP(w_CdggC2CdQXh8pI980c>AQ-QP;>0mQAo=n0qo3OHRb1N&&mYZr}
z(05?VF_>AIqyGtWVTft~2TMFzgfFUk(Gw@ofEsXw{xs~P>0pYavjY0p{zVyWu5tCl
zwW8$YVKmuC<z%YKfb_=jKrlp2FGS(F2InI5zaVG4J#d8>NQ;kS0H=r1b~e4H_IA!1
z_pOf)ON(1e+;|QxAiBt#&_JV7@boD~X*|dnK<hH-7y6NjRfg99W*%H1^bEQR46j_n
z{4y6l^;0O!5e}i>;ch8yLurWh!J)H){b9MQA)xH&%9PbdH%Jwr7MykNDYR14)}|4&
zfE!3yZU!Y2lcA0dN1BK1btsoO3hoMFP9)LXlCZUfaM(tXgl<@#$Mc2Ho`HyEotweA
z6EXB0A_+ac#L?V7glSxzgy*vQtd^{-32cjT3<C(qd>b`IMK_>FHa1fzL7M7r!N3;7
z*E>Z57B+iUp<LkN#^#RB$;4%sR*1j##%Xg8vxX!LVx2ste2^XkQ;RmZDd+;h3Opo`
z*0*G?!49{(ii-#n4j}bBw>-(t=KeW?bN6#w>>)iHyfb>a-euAioel^HAjpXf6yp7T
zedJ7{51hyPaDE?RWgXK@I!qMF2@<inzIX+K$OIsSqfuEz4#ZuY<aLq$CM69n9-edS
zRzMM?>fSs7*?|Y+{_98XCkX9qJUxji(Srj6)HECUswsJSQt-ZlE(LRC^6Psha9)7G
z&|bf?ZCR0>jZOXY9S}05u!l2G!?d~yux@`oB>0^bbVW~2VM9Rrgf?S#ANjp-+EegM
zXswXZ!wu|0N!?rVJr+7WnE9MI0fUWu4&L;`0E{N(%3s|=SV@1DEXaK#VE$)`miGsV
zqT}B@00dC%AR~J`P{t@=dI#uv{Z<s=PM89b%e~3m3h@K=(aVLWuN&JY7I01~9&i%y
zT%HAEO7?LMff=L`aJ7eFR`?AFX_DJ+a4c=MFvrmbMb_dX=;K^W%=-8-!TKkJTCk0W
zg~_QVLq%6aV=RE&oH0x`0S48?U5m0-P5O6ga&kvYm^3Sz2i+DYLqFk%=CB3tsPN?_
zCCv_A8$ugk+e2Yx{!Nf6Jpgs@?rleM@iNdprNH`Y^IHfzbchI=H}5@Ymun+oZeh_;
zdCo<Ya9rHJZ5wG+CvYgjhtWdzF3KoyG^Z6b=wu>1Za#xx<`|l)S|;~`g-%Z|ffW^s
zYvL{{!oCNqgqVndcrSlAv3&wdzZAYbkxOg#jC;=XcCvroF}PI8Xmd_ChC82E?Wl@X
zf7+wNpJ$&eceO1DLFV`NzGur^8FnD3#|1YDq(R;F`<FeXqFd20JM%XAHlPJ;T58_j
z&+roiafjt7eC!ySR0%u1hbj8*F99YXjX^bxpF_%Ir_&kPU5Sv68=;P--P~LUTiZUJ
z?|)3j&%}y4o7$1WI0RGN9i3%Kc48po=ZOgz;UWnk0QTu=_}TG)+#E@a0mO>{`GEuj
zd<vVvu^X-9=tn|E1-%;zS3>HT-NlGfbMo_Nj@}yl5PCTo*gFOW*1+LlN#4ppWsT3O
z3i*)AHZ6n*&}9H$VrjX^0F`A7@j|SM?_+M82s<cB2RN}DIu+c5d5r#Yidb<n>qad8
z{O=5Ak9_GZ%z)Ht#S?|T!l_el;#bOx+5p2eeeVRkF0VJ`h#ib51@P&!lc)r9`-czF
zHM?9-EZQ7rM)xT+q)5Afkm9ni6WMwF8b{GV1=JgmCGKQ7!qrF=p7kPh3Gf~SzJMYh
z!*ZZJRn%dSwZR?$R007TR$Wxt{&VW;&&OTxFhE&B<~n}a;8sde(IPH8qFzUY8E`GU
zflt8Zx10JU&fbTu<LIzJ*E%F~9k6a32@2GhX552YeCxBSg+&@r{?Jg3>*rPCbA<2t
z;j?Jyl;JW!9|9y}Pte!XL#)ds8hj%o$L!HciL5LoCyf`IY8Z08MLkNG4nuSxcJVoh
zQbkS8I!<HcDTFcvFA#8U9eF4~Vt+#3bmZ*I*2zK#i5&q3TYYFGx~}rD3jms5f3-q&
z;Ox!E7n4jbrZ~Kum>r6$%C*TcX|+i*Ny<;o&rNwfy6|M*la}cgvTQPTsl58m6Pa<-
zF?QM3(VbCa5fjH2Ud6qY!waN1q-3XOXQk*3X<ZmpH!{>T)EJsyYfaJAI4960(xv<1
z=o9Xi{1!)h%kxzRJe?6;yM%XN>2lSM&}y-54YUp9ndUY4nCrDXIqD1~11<x!5{RUj
zADEYw1)#7A>}&I0D+mL<FolH}-TJJazB&^E2#BE(z9565=tYr0hrUmEe6X<GEh;+p
z;xz>zD>SXMEJ4`FCS?pIMbnX5=m<Z5wlwuRM(zh~Y(#jt9CtGAGD^ae<S)1m9U8^p
zGl0kV7-Oy&#fsShZFv5`idaQ-*rbLx1LeZEaRB-U1@Urov%HIph`?KC;MU_zYMwAU
zL<oT(f5gfI4kK2L7l*V9g6iBYwG3>;;1NeYH+T2qXFW(s0k{4D%pN(ugPNL}aQGtJ
z1HKPsxc<UOP}mafhhPa$+joP*9foTw`zV~QT%p;;k14_g-V6L1vxLAmKvlB#`!{0A
zN7LOXjNy=xZ&*%6M|WB;UFY;^D`=x|ml805rINJm%qIeJss|NDQ6RJlUY}_5flbrx
zu!Yhg@5guAQ;m}jGhq;>RBjI(q$Xd83Ia_Ap+HOIQ0t%L9C2FuN1){c?J;L=xiR_G
z#qyhMX0LI_Zd9u|quK*4qcaDLr4%j*3d22#mG$K8D`FNF(9IBu;&>T=gvb-NglxUA
z-9AlNM1-of3v#1S$Aem|G<W}_1xQOz$H9Y=WG^?woT=?T=J1D-r;^TYsFU7Uy9_N4
z<~X*nrGm;}z^Go(i%k3XD;(xIapD)0sz^5KZ4;1Fql*VwBQh$Fph}EoNs#tg2U(7|
z#j;{mi=BbA3xX?rSYOvss;UmAcUj(04q@vc=KH_HJ?6*iL&PBFK`*DkfCjc?!vs%I
zb@wjVi=d!Z>^InZ9PG`>4=cyo^YW4uO9OS7BLCd?7r|1TeZ<rI+@}H}A*=#o9s*K<
zi2GQBpNSet!u=TprLzrjDvWyo;~{^6X5fs9N={y0dM7cg3+d5;>mf!cn`voJsr_79
zN=uE2i-Sw|0xWbQ7V%iaPB@*ZmSIS~b`Wef?8X)^GQ~oiU0i%6$QSx2sK4*kqvwAK
z1dGyu$3`3Do^!+2!-^~CGVPuyH!DNm*`VKWxh7ty!_t)V^~ToTp~xD>x&9q4pFf_e
z_PcUGB(pV1>XeuG6HT9Oq!piERRxyMk0_nly)^2tqO8!`7vu5UDZ4C!CA<2b$Mh48
z%1Wu#Sw)I>F5icxezo@At|Zs_wDz4wyHj_-`^qnwmp8x8+moMQS76^x=4Uc^S4>}2
zpF!WoHR)<WTXvf=uVRo=NT557=8#tVd4H~W@e{B2i+*_Q=NC4z;eEdASoeUp*DvdL
z9dy1&+4!PTD!qT&{7x6s6I~LX$xfZ!__Z}!FM=VC=KAW~)K81=X;(z`h70#fov8=w
zdgm)I%~<gv5QYp34>IVF9y#K*GM@%!#^4dAzbq^)^pc<E<`EImX&SUg$iuXnk`$s;
zY3i=^@fpq!mADu<zfYlx_c|=fhW(QIVP6@vn;V57<lIGWQTmr(PESvR>4Gervx)=H
zH(1ytN6qe|@FqMwr2i7ZsD_ClmW7cIqN1)&Y}m{>rQn~or)@?O=iq!)Bg%Ncw6}ys
zK2%g5BE!)I7tJ@h3&@Kg0^wlzH0Nfo;I-7HyI74p4z-$Z6%(E+++>EW4#CmCtu5!(
zK12=78!2+TlonJoBMZz6qw4_3gGxUD{t8tvC}WL4TZ}6Yb|V^j6UuXPg6A4}qFW=4
z{^ULKAt5+F;OMTHwuTXi50p{G^CY%gecf!Rr`Kz0gIY5EYZ~OgE%(pC*_GL29E3Fj
zrAkPDXQw=~@)CBJfAKH04*Gl55_H~OTpJ|={1$zwwa+j6{47B3p!jhSq?J$t6Dbk{
z)9CipIdyO<EsnK>ZElUsNIj{sKAecz)73<fNj5Z`DtU;KtNo#H_N(VnadB}E_8d_1
zLa}>?x(wn5$bp6@G63(R_3q_%7YLevAzQO&SFE1nhv%`<R+!25jToo6jt#&L-Qs9N
zO#^(ns9lnBat?d!W>~^#a2_5dMGZk)31s*LM=i+q7JL#8*CPoAQ(>l~!#EEX>C&>Z
zTYD}(zE&~jB#PPIf=AS8{bTnw;pk00=LUWQs*-Yo(h@2ufTyTD_fyGZ%p9<Q&zJVA
z>-d_{>!o{5Y~zuSM}l#k&(J!rdGspBm)g4Nkq?8{t|VSFF;+0zYN4rPLt8K!6?0_D
zcZxM3t)ih~|HjPIK8II2_s&>xbo><mImX1^U+cJ$W_4Not9jwj=hfC}`_uUjIYv%4
zn?{{~`7KJvyTur=+r~-M3i@=6E1Y~Y@n+-dyPx{b4!&K?*+L=hPigu)2a2S<YtL^L
zw7e?yb?sWikc8B9h|-ADL_&PwN5><jyzlK9>b@g|Q)7u-OYOdlag59&_A#I8&Kek0
z`fRMj-{jP(jpHQ}!oppk?IY51K*lv<(019iCyyU5fT=>^fdUNVCW?tIzponpA}A10
ztVm>cB*3K&cfAfFwzqdB=xEJB@$ES7QM8jDeG<u8g;5z*Z!=JBTzohTRaIErZ7eX<
z5r=d}_NFX-*T<x45X2Zuc6#JT8QM!wZldUPu($VT;YT%xBQZUMg=Y5W&jahkxO3?v
zI+Y5G0X%Fqd$S)umh<)i=m7!A&g@q>`r=WIii(RP?|k60H$eky9Ca3yv_={%r!GvQ
zr2)g6P?!~107!Xs%;sGXUE6U3&}5c?DWP*gnP23VgW4MQD*_*nMrI(ELsYMCV6X>n
z6d-Ys!*z9ZFo13ZX_{tr0D0`T>i1F4o(YcL!-qDSujkt0LZ<x%wKFQIsn!&@9|ib7
zL6V9_7-*`hy?^URkVSq)W62ovEhHl8Lk)DO3CS$@ji77NlH%&CMbOwwW_n&<pPW%d
zC1<c%S_(?F3*0&?fl|W_b#S>QT+k%6b#>`HPvMS-eps)>#K<5!Cj8hCkv#>gqi^S`
zt1A;)4a9=!nVD<c7c5H#;ol6#OwP3k+X4U}aS;Idntr}R?E=+-{7Quw(&#`M-Yr`n
z#>8O$a&J%114lvR|MvL3wT%yzfqbwSef|1di1hlwc^FHQc0ZWvc)UIIh@|9TSJx7%
z;b33IQeyX?-Z&nrGim~Tlc`t8d(1{RL&3?KM^5;4!HO}oa$i77<UK!7W~jPx8@fB6
z-q$-sJY8^M4(sJ?&Z9RA?hWFkZdjg=zTkaEO+`(OGm#6p9$7G1EtOdT&JDQzz}4m_
zHv9{F2~(Px>1kzV6CvP>-8t=VR;n2H2#$C5PR0BbaM1qpn`OSZ!h3Sn_-@LN!#dhJ
zOZ!JE&8Uly`f6FbzTFW=CGuo~t7MmLi>+TZxsS<n{_aY<#ysWtsJEPhW@b4So^8K+
zsQbB1yVP$!erna0;JD4RE=u-{hE3YEQmx66J!+IMvkEmk2e{OD)l#k}J*ay<DSgQF
z)SZBmGj{j6G!wNXE}=p}`2iI36>3vFa(HTm#l)OPMFoh&G|&%O{v)x=>90~~kLYXZ
z>Dic?y5Lu!NXGI7+=_g`xWYBS`dWHgnirn*v+H6=IGDv<x!lcfLUp0Oub_dDE>pC2
zVOB1HRcaa<_{@9(q(PXfx3xjBeJ?O@{>Kkmi_QeCvt^J0gU<ze3GaW{C5JwEFblg&
zpuecUphqH`M>h&ENk4y<XEwhpKTl6zN9Ffo6p-CI)J~|b0Pvu(8>+w~+}yozlS`_k
zfa3X?`OeX`xp$BgMa0A?o+&CU49_XTNuhDhe_D1kav}hIDZ8c|I?(SUiF9Ul@}vlu
zVms%A`#2)tGAwWj_7kC88}oM%bMg0p;-F4g=AF~`2G<Y4JS4pZMMZ=Uz57=-1~alA
zyi=}p4o-^MlSHn|-w*+L$trdjK&OW&r}g#0$nUr$Hu>cXcGOp7#UR1m$%C-j4;xp-
zxIM|tbVID6tz0C{$jdu>fbHXIB&!tZSo9P+&}SlM;m|4{@&wE@((y9|m$o290NyPw
zCPvAPUglJvg=f20e`zu9!Dl1fhCVq%W5Rmq5NK+^DA-%*N<x7TTo<D|&qz7iD=8@v
z#yOpx@HTWt9*wdGp2jbJ5+;rTC5=dJ{575jZzE@-r=kkO=QlKrM{5mI7>sB!HueS<
zhWJG`51sCVN^L<NK0d4qZoj$$l*q_mTU(XUT&Su#?otkXfo-!VSpCO2E@!R5c>p=E
zwSDqw3X_z!j=B@tpV83^)gF<L9;sH1;syfxO>2u{C&Qgd^%F|%as&L;hf4mA@Te1p
zut__Q$gg@)vn+eAqr`r-?|IL|200WQQ2Eub&f|kXBf(_gswS1%?>6}^jm*pvXveA}
zo};Bpk{-^)%?6{xM7s?>2jFEOUw|uY(YuhQOeZH-2M{2P^Y-2mtV%pU)&v|vgfc2O
z=i9`9+k^c_v~&M4+vBPp=Gmel^0E7)&@W}5V{>^Ol{LrA{4a!O@nm0Oe&ZphY}kM5
z(zH#F=HlAIx6dKdl6kM}rG7|V_(^3Wy4lF_@yEvM5Jqc%0ZzxEvtB-N%W)olv>Y9_
zyx5{M$X8HcpkDj=;|IqK3ns3>$9V=4IaaCZUVSZdm}6O5T4FoJ37Bo0Yt;m$g4P4*
zaS;}1QVH?p;(FaX7YHX&x)IZO6~%CoMh_7>1C2ie@ei2U3BD6KW5!xZXLmP&6rvh}
zYdXR445t;(-+|fzcX|VeC~!1<n22On$WyBzZbI?4MGe|GZ|FaVmu$u>uR=o#3C$le
zP7(?dXeP!^{D3DP(jjmyG;16=&}dY96wmS$e?-<expxGBj*`wp^a7-nx*Hk}pb%o`
zecjmTSMXiZuWkzX3?h-4>)pCxLrgFb6+K`7nBu(zz%g`@#OAg^K8WzDP6buuOVQgq
zvE(p+51qaxC6@7#mQYZQtx)ubK*o@il!P4uOXeRxj?}qKad6~ZyF-6Gw2lezJ}sAW
z;Z8{0JG#4npz|ZG?UCVy3mZT{LDdjLDHauvGbFYr!ZBQ+kjAf04_XH;j-AB3TXtV5
z*)h%lrb;>~v3-Xi=Wsbr>im6+KbNTJGPaKV*MPd=wb{Boi;f=;crV{{{;5_xC*vyk
z?(JK`suYgI+_LUK=K#P%;74R9A&Wa{Z=El58pctiddU(7j=6Y{c)YhV0PEO4dRyr_
zML^Fu5YfO8FF1M~>k`i2aS;*nRTaR1S|rM;#s0XZAy33vcmVdrp?S!dz+0l9W$*rR
z5IPJeMAjGItJP0d0aC?j{GqPy_CPA&eNop*in-eoLPFsZCHkjuN^o<d6i-o!Zm@=q
zFk-0&o9uLgrG_hDabWI?cX#6%0Z+V`>RVGIHCjVhtxA$~JZz{#$;%*Zf&(Ad`R&UW
zrM08sXlMefq!AG(InVj3;WjdScdlKsBMrB?unnu{Ff?c$*+K}tzz8z@)Ge>k^?>9G
z8KN4xJ%;TBp<{XzcRXa0CmY}_s$Asp&R!+=w)ak@_6AHI?XI_L;c8KAR_4xi7}k-M
z)t2=z@o>2`A6~`zz4c>ZZpzb6_pHuCUROVFRA!tr)Law&KIcdJIa&IW(V+Nr$<vc&
zLbYO*62J6b6pjsyb*^>7<GaCLDNw03K03g-ZasaY@_>%sh1js@`S@Q|0gaxF2cI93
z(Ny?K6Rs7h9dcH*rOeDr^Y`VMg+}*~R@>G|gX@e-A4h5w2I`s{IcY*r6$bhHUyQdN
z0<E`0j*X4YUvSS)>~DtJOK{gXDJK_@0n#a~t3;LJS{$-i4waaLZ16unA`rSePA6d&
z3<}t*bIU4MVaF<e{gJb=dti5!9RS)}+tI&(-XRkMA#Wr#^<R~dHJJo&RqOz5Iu($s
zcbORJLbbZMu&}qfbRP>4W@q`ri#ryu?LqDU@WiRhU`3CK_tBGwN{Wh$0XaifkLXej
zdUAHwX;oet`Y&X8V-Yy`=O4ywIeoejS-{ROU_gFgRcJiU++Pb$HDEe<KPbHz)O22p
zTSiG(S?sL&FR`Bd@r+BIcFV$$CNbh_IV1%y4H9q+Pujed*kjP?Q>uOY_Ko2v=gzkq
zJ{HvawvJ6NR5d<dP8AXswm6(?v{J;lNAB(}mXNb)jBv>dx)zM$lMppFTxY>`aOT~^
z>JyxqL4@>Vr$B!!!5du4AjZys)O17Jkt;rbqbS)I<P-rWVNf+Kra)#IQ0fAJ7>u8p
zfjbQEw$^X)#GbM(X!YUkVB7<yn66ZlM^;t!-u$NXnwo`>s_qOFhwIzuA|<3!qb}vQ
zLTMpVcH7>(>{-JVG)uu|Cn4dn$i>Qf%eW7O0|M=ZDmTgl2R3k<P!-HyG{3pH=vq>F
z<J+cvd(0Ob013+l{alB8qma`l8PjhE#Zj})D@pO{<Rl#zWMAkNVK&c={ZMuMQvUCw
zKOW1SR+Bi<u}*g#l;SY=@R&s8QY78Eqiwavd_E=L%sXvZE48##G;0(7=IlOi)8**x
zwTr*{LuwvoWR-AJN`<F<x%)+1BRk&bN`mLr`A=rIbx-Bmf7bA+=XRntJZwaNtv$C%
zGR)O{zPI@|`6A<0&zKK3`8LDvIpSkKdDM3;`A>(Y9`Yo=VtF@I=a6=!K>~aH+j_e`
zkL%g^aC_U&GSP{=nR{v`&r`=~dwS`9xV?VDbi8Tz4L#b5&Qpp)`R>1J`@eY?Fw%)!
z!dsC0?e{KQ|HsRH;EYIk#TD}y-Hg{>=jO^lJd>6$ozAhYn^-W2AE^KM(Hj7wy}?#9
z$t}b<3GVFRVypUVP7wc3_uo1vF?^CDJZlI;0Ni>qg}<W@tQ`rV^ovmdk532VXb>cr
zDSg-AnaX7VRbs#<+4?f0fDXq!;MvIX6P!RvfG1|lkibxnSBbo?$YN7D!Kx$XJQ^?s
zAPSRcTCa~Y5SDlJ=YV4+nmoP7R1&2gIqxHHd3zW8?EUwT$om0>F^(CFU{_#2Cg-Qo
zpe?lb*<voJCYG3IPf^*P?54pC$yJnb+yTM&bXiy&8&t>sq>28*lL2VinTvC*G1-*`
z$vw2l0d9odE=24tokyr1=Wo?8G*r<m`5rERKdE4Fmg0Vu?yvY;-@+Xw6vs~@8MXLD
zuW{}itB$^+s%qpX<Io%E=cgo+u)PWHh=wOCe#5r@{*GN88G@phKkuUS*n#F4=2_Nl
z?T|CPJ5kb9G55TT^xh=KxG2YpEg~@uPdX&#&UD}f+<B3s)%)*eXHSEsLEtPCC0yiI
z`A;?P+XKcvshvjD<{ewhOLepWVBqj#7jqA7Td43c=K7bTb8HF!2uMr&;1LEteuM7i
z4?s*Q)qsgj**!b+6|_N9-<Qd*TDJ9*<e3(c7W=C=e8RVGJY1fBU?wNBRy|x>x)J$7
z<7g|^?n2hYZ^xDQ%2)XvO)_J*=$y)}it@hKxG@y%qUF)q%V{;$yXu;9|59FNcF*$9
zhT%rP?}39$qTij3H21U1$!fpI>oHpFi46=>@(Ev0`N03_>hr0_jDSnd<5$KRH$E<3
z`#iK5uW4T&7k|HQxNX0pQ6qAXHP4BO(LMJ+h^*)L3x~8<u6>=q%gD&p8$liMx&JlY
zPk)!$Gp|PP3>{@Zen@{U7MAT=ltRaj9~a7F?UbS{orfsMK-%zRAg7@};097=6h+9N
z&MCgZ`8~$1VPF_r1hgBXkz0-RH+Gza<A%V&rollYH;p4Hrz7O`eyKq2;nU(^1<^-^
z_?L-i6rNIv&Q$l%e~IOxK?N3KYvF?W(<QH@+1c5oe77byQ|?mnsp*Tr9|)O#$XOhB
ztGfg-k>weq`Y9_1hy4^3v%q(;qF+S#f^;TCD#5{Dd&AvF?&4+7c?<Dx>+Kb{a}s50
zuwiZHWhS&96dzM(VBvw%0R?b#&p>`|E~bQTm2?I$KlT9~3f?SaXM$SazI~9;!|~#R
z*r5wMQ*xSAPOrO}Fn#a-EVa@gXKHH7Yjqe*7e$T#kq$R_E7_?wg;`4!m8AyHUQ>lX
zjDJCTYN|Si#i4zBklOO`wf&}S(q^4G+gv;Me7X6MIMS>Z`MpSo#(vSJxS@IX>{)Vl
z;41DY?HAkja;RL44^{8Jx7~wi2M8vUM5TAnd#=iD?$-J~i|zNeH}BBlpb1Nc$OktZ
zrKshuXI7S$|BJo%j_104-^aV}=B=cJLZM+qC?jQ$R49ATgff!sog&FDB`cxGCPK+7
zdu5Lh%HFcF`5jO9`~7(Qe*gXc`+gsf&!<1$qj-($d0ppqp2vBd$I%ook9r=4mh3ai
z2==M{sfNN`<sR3%YxAksxilSKmq>ZBjcukUCMKnHW*f8@My_;*bzQWzBVRi=^<6U1
zvF?(0+3m7kM{BR3(BH51nJ&57nzs9=WTrf6*)uz$X|6I!w<2`M&;00_qfZViDpWKi
z<g^&jP0NKSy_Oi&WO>j0^4Dca6a$*9Dc!e53(qbUYkScI-`83>KXFciEN~>G@#0`h
zVBc4r-M;rP<eKPbF9rJkK2APB|BJnc+{AK>$LcX)8fRFh%YV(`gWMGe+F~Ew(L`oq
zWOQAf{0L%u@oxXjXMO)1Mu$NHqa>IrTlRhRBc-aoe!A-;XV+A!WBZe^*cZXUAWl8;
zOn72pR=vG1F-1#fo>ou%uCq@iW}Jyp?I8<vSio56xOM4`GiFp=Q#nvZ=L8%_Z*gAW
z5xZ>GnJ-H+-ZR1?gIor$64mfjEhX8@Y5mY)FO}BH<Tk8)O1KBp3%@Fq!jnU}fetOE
zdgJZiIV?`;Kx8=gHhls839mSfGhe=cXH0*~jIQKEy1%L9m6d-WyrY)hNxF_|7RW`E
zkW7IuEQ&)yLr36Y8JmI-C)i?!I_L#QTx#hO=i0>|eN1v{tQr+Ntn9tpY!1DQL?TYO
zD12Q@!k7sfrC_xXA|r;r+e5>`{&AaT@%N&#5JU1%1fM$quwZ@*>Ils+s;H}f|M8<X
zM>_aM9oH`}kfD&QyZ=;2%VmJ!V57rXZtj!(eT`cCOAQd-#6E@xX3U%gO#m~tR&FJD
zm!2ZG-o{3SQt~53i(rLz3?0STCF*bTxC8tMrD#mP^WuWy+RL!8stfuY21TQ2Cql8~
ziDFfoXnzdvcg_7r_g8%R!XPO4D%1mwGXo5)%<G0B1`{JI-t_Y5M04M+JJriV)XDgE
z6ys||3gL6k;ZnO;VhKyK!j#aD$#0X_ZLeu*w<(FLT@Q-cCcd^1#Lwzm^6RYZdO6kb
zSVE7bc9dmF9@)oH!?)?jO$DUouD8V1PUYuH(9cx=x~ChGe_!v<Yw~s_W%;c0IkMI7
z?pJ=GHa`FKb*RFNuT*60iR>}^?q;`qm_Tx_#eS#2WO+5V&(iNu*ul!<@l+PM?^FM}
z#8AJt>+bE0nC7M?{(C>apwG2a4xK>!L4BbEZ>FP1p`tMW9>B`k6N#^WLkb7F8A$dp
zG!b2<XI#9R_Hjt(<P;}<oKVekxM-M3$1NGP-B)<kdj|>~z#h_t%y^l6*sUG3XO{C9
z8)-|W)v}H{=c9=k3A9IVo(UDqc_A95iZCyK`F!Ka$I$C2(>PdlWcbrF{+pN1)240)
z7OV!CjXAA}-0zT(Kd^)Hz+n#4RG&9{BvXApEv97$sG4V|s+MUkrnP>CPEYn!>u1)M
z!0nXdZEXs>@7{}faP8p3wT;T@XBOekhs3Ym-KJMxG(XVuTtfU4^R)Bh=gsJ4oN~4W
zGhbm^QRcpbt;@=(-I7UaH|I&Ni)j&-KN&0UuK<-qD;KbsLVijU;1(6f9{?_LfwQ(s
z;MhUJ>A5BCndO5V;W7_sL!3wqzMnBzXjGKs4rEJ=^g{iMJ~NB`DH!5uf@}x2Dh8X;
z;Q;3e|4GAZ8awPS)Gf4IY`U%lB|O1{Lt2h5>QGlAl-67S<ny2RQ*Hc{27~j~*562M
zk00OpX-4;&02z`WwcZ8s1F&#lGEOOa=V3T3R^h5z<#mefAa}|fgT^x8)KdnW<|Bd>
z>vbORO2Z$+=iuTc7It3;{}QvWpiRWjgrSM!>+P-J{|T&_@Rixa$h3gXaovsR{a)uW
zI&gZnPN?O_%H<TJ?3c|w5lnPp!JOOwur*AY7OPmsIK=RX)!b|O^{_<5WY+tw|InFL
zmy4Za&ef@xQf1hBr1Kf+ZmUI`D|R1aeEj6`vZv>==ct~y*CEGOr%m#fO}16h3XHW_
zbW*%!ys<+JW?-U;ig6HdULZs_q(JvWybERv=>pgCZuiP_IpM~PzXPOy>HUVioH?4s
zPUyaiHH0$1i8(tR{#B^|@9T3s4;)O$-wGOFwVAJn#}<m*v9ikCc=l?bChqO9^1Ex)
z(4LxUc~Dw<X<=t4qtspL;Av@Hnzv_2Pv$B-&{$esEm~;P>hp9~iMq*H_3#uL0?n$a
z0}^AwM+(H18or;>9P}XLX%7W8mG10~2M!nSEgV)g7$`%Y#u!rcVxkn2xZ0Fb7*a7=
z4dT4msJJG@>DUsY``K<X{4d^w+lh}S_vo84ewzD@kwRc3sIKK~kIP=YN>F92##_0J
zs%<b031R|`jGs4U;c&D(%R6ul;7zk0|H0_iEz_F*S=X8!lhc~e3i`_SgR;3!0{kGr
zdnoTxSs}a@;OEyeH}7socWN_#oZ67K&GD?Xv~HgOS&+yLeSWfKRbvw5i+!97==Wf5
zubBn%0_KfxJRMHV0h&;De6{jwA1nmOgqK7GX(jrN)UZR8j)9hyV30LR+vRvRvb6S6
z)ma&x;m*h+^TpIS$!F^QY^Rm_<Pvg7ukqHJ3ME?zZLgRdyGVzFJ5xxtxhS7gDiJqH
za(18C)khQm<pQ*oiS_C`@jtme8?NJB-S5ZJYcV^v`-r8<Hb)!yt_cbfm?c`I7o6r3
zRxWsga1N#slKFK=W}ubuR%$oHo$k-s;Jc1>31SEb(s+g1xmIA$C8cNLe=D<RzI9IZ
zD$X}cvLL+5mt`*~GH-sTuc9<0CHnuMN+fVD!^x)!#kOl*@X_J){slM8FAg}z*UwW|
z)vFaol#YgaM;D#Q7F9uesZ7VzpCe^}!s?ir?~~hud)$8mmyAdgS4z<M@XpW?K47Ju
z*=A-O_t-|(g3f|!wr=04Qsh4aT+f`XRObGp>+2!0EHJZDOx0q2EgqZ>)_dzqyPTYy
z$Jbx-WRw4JgmV2Wcx+VAyXC__0I>2al+i-_Wu&B#{(r+<bOlliYz2?SnU)HOzNx0D
z$LPY&Pd-N66NFy0=Mo=3C8UG0yB-A8c?jFz%_=Y(>$nN>-rlTg!@{pd$HUU*&H|}G
z|2A>=Gw=$cTxPw6hpDpW7^ri{F3!BH^9Xt!xqHD+MXD=Xn3q>&cIwmR;jf=4!^JU+
zQRb)(9wMVx4D{Z$UR9Wk3+7{z5Zlw~d_N@Rk{Ux${1JbZb8;(s7#a?9cYV}OwBpvR
zf4Z+fNRO!^t~Bw_f%hMBoa5bx+U{KriDa8;S6uvP5ZAyO(^>q<E0OF_(;;Qv5H^)G
zQ;pO--l3Xu>2V=4EjI7GmUOGV@2?p?+~4+2)VZwMLr_)|b9htE(_<f(O8ImaroF?R
zN}SC>+riW}wu+mlxFv8q_-%~?_QR}i(&_v7Pk?*=0RiU~()k$~jSxht&(<|H&Ats3
zUis-k285AwN(7@-I<st9>iRvrttS#Y+YWOd20q{)gs9>A&T*&)6T`^9ld5Xm89+e1
z9?6;^^fYuJJTqu<-A2$53_f#1rC;cq!NKgZf%^wA_X@|&lMh+2TU-v3d1U<KaPT!2
zST?{S4}6+DkKrL{yH#c1=!iGsN_+O~d3q*fjq9CT%;=MNF1*NqfPB^t`KWL9U}Zyw
z?y&L0Wytxf#B(z+h^|HEvuHcYxD0;<sf2K4EB6DFy%n8uxmOq9m;@<6{lRt<Pw|;=
ze7)a0I^fD!a*6Z;>|I#>*g&!YXB@YVSxdt4*AWC<dG+qQM6syGna&<wG%H(Qd_;@F
zyz5;qdnEPp(vopX{;d?L7q}AXc#NM~&1T|83n`VcmJV9095)UfkO<n|>nE9MMMg#@
zB=AO>@$KBbg{0Ej3%`m^R;!>dnyj37^E1iy;|vTx!H}Tw>|EJyLnnQzl_RdxCaGI0
z|3KGdY}Y}BgNNzXZ^uVVa|8$MYqcLLTw5t>*3)+Q{pYKO*X=TFpu>NXy3Fo=*Pi+6
zG;0&2yHEMRjzC57)yKYz$*p2m4nMmG^8Oj!8R{0k*7S5i@tTjIY$qa{P5M{I<yoWF
zA^_tlGYMmVGz&j$KNa#R=((q-B&0>Kyxd@4p$7^8Dw*ji9Xg(fHlG*8xW5BfRyp<q
zdGGiL&-jtgTldXa5iSSNMiO)4JO|@DTV*9<<V(xSylY4s5T+nWnZ&8cHWhv~Ni#bM
zY`=2B$k^Ce;APTN!;HDv?h5yJ97E5iE^@U#=>S^cXc078d;WykiIT-7hCSk%n&HA9
z;=p)BgN>7;eM!>R(sE{T@vyxkDo0@SL$CWlS3A&KqSkhUSX4L|S)I2L7k6Jtl1mb|
zB&DMx*V)nR?d#jjGPAJInlYp8{cCiT(2yE*7{&k(dJD;NVf7&^;`j08$?2no-%Bo*
zwcDum@Ws#FKsyFv4Va#VgTi@scJqI2TdRGG`FdUEAbc5N?6Qs&>c!^!jEswR6QKdA
zN9HF4;RGw}ID4tfSe5bEbU21OtteivoRfUGM6&wx>vjVbP~hh=bdk9Ewf(~!!^KCx
zF=s&HDbD5hazZeaXEI4K&gYY}QlcW``g99bMV$nwk0`TSnN#+--@|19ByVOxA)og>
z!ydFnm{J=kPFXaVt~N^#|N7SQ{mRXi-=8WiHw8DYynU}o?P|}!yLTpF_i2sOZx;jZ
z2fw}8S6d-z&2aVM)w{V2d)y<lb<D}U^JvI)DSj5)a4x_8^!sP3$6P>N^RUg{uf<=>
zkM~kn&-Qblxt<#uz8~<U;<{G;u{_h??*}H>ZIi-6J40=nymqm%C1a8`v3H<$(e~_j
z&%K+QY3_g@r+8~hwaiSELhY|-Z+pbBhj_Yk)J6;!bFSty6zD_IZK#fSfbJ?hbWiPO
zTt~kmsA6-eKd`SDoNwyiEh|cT0cxk=mE)Rt-y8yx@2-V4!G^Qy3Py;Ju;2Xc^chR(
z{p3ldJxyZ}qs%FLfrGv7s`S4TGWAP#3GD^6ntV>FWQ3B${T^z&8M6ep*)+rmiCKHX
zb-PczYFhan-Pd?AK`POqU7?ML9Lk%z?{7TZ*qd++-u8fNtKFa`>WXewpFvCf<VdIr
zd+2QS>Mc9ESCNrLm^eV6j&*Ytk}>e3jTp&Qj`YGOAMH8ZN)hR0HCU~Zx)S9rm{eDS
zea5)h%X4_<?k=!;A5WT{85$fUY|FRHf-p>}cM$YD!u4zBD>`Vn?hk}SWX;JVQJeW&
z7tdVDJT|!mEp9Ld_#ojrBZ2+m(#n0Kv(U#<fDBb+U%_ii5>JQkZw+G@_Ml0|W8x`Y
zJ}H<5O>w~v!V0)=Kr=@E>K~R7!50oe5}#kkN@+(s7j<{de>?Zo!{5W<h2Ct9q2-a=
zx6Rw*>_UnK)$R@6@Mpgs%-=G`&~xPkBio=|pZPkQ$T9mB+q9v>U`y&mPR<rxY>d9?
z5xH)Ee$M>m^t0KMfzNX=m2#~25ZrC}EH*atPOdI3J=vX+{Migtc`(QtROP@XE^G{3
zA$4;cihuO@aZ+0zZ}%~W)}I&9Z8_$AAM+;b80Bzk(+IUs1%#5+xJo3tvH`j>2Gw8i
z^aRT*o-F<X=Asp?Q=$wJR1OYkj^#S&45OVT({u4~dZ?Lp5IeGlF|1d1q!pS8cuuN`
ziblAvI?t1VRH?l8V7I{L*l#>J{E}L#rY)$Eo#c<u1xzj+NwT)FahWby<RL+E%d#Ol
z+3{f*h@=wdMeH|drgElPLo4H5n*4!BsDw1uXr%4z7EUdnL3+}(<N@W6T%2$3zzH;{
z#(EKS_qTcsJeq;MYxiyzZ;>MYHVY1Y%&)}I<jf&0S|Gh=dllIw*-zn(C4;KOsWCC{
zsZXZ3v_D5R{$!hu_~9sDwDW|-u-M|0900^?({7v1A7s0oS%@)%>_p!SH|2XjbiH5I
zVmcw~yp#RVcG5!!59)t8a22=Lj2X+GV&*89H-+J;40}-IZ{U=bMKR%@lzMB&A!xhR
zX$;!_?FIab7rySMhs>h^NC-3@#V;^@RHjxyCbHJ>^D<4r9D^e@!OAdgm5syQx;Ym(
zRLKB%VP<IIa=QFkyJ5_l#8~Kkqs*otDMPNod-(^aRu19oOLOHVr=G4*o6B&d{p{q$
ztGeTnuU^^EN6#|Z)73Jn9nUb3TMZ9as(8G6zl9q2rvHI&jC^u^@xN^sTx;e;N|q;6
zCqr!16*FvN&evq-Wa+lteRg?XwJf{(kv;PhGGVLXC*qZEGXlZ2?O_#DCimak*xTr}
z8U+uuCA%}IYiOe0TJM}|pIpyrPmUg^drf^#PX&cTH9jNb?u9+>XVEZj?G8n<>Fq_5
zyy?0IlGjju6sHEEmBkohD<8j++K+x2zny3M6%<lA?B&Xs1XJ;FCCU0E_b?%x0HiOv
z=$L>Sv|>@BhJ6b=Z?r-6K;B^3gIsm3&zBQD>zC8qw+s!xcXpcIX}1N_rS(uNM$CS{
z?LZSMwNTzS+UQQvSG+U*ks}H9rqN@U1CHjxp3lE42oDb96B6JkLHvDsXy^}O6G(i%
zYF{XK96xoc^!-tJH@GLYeBUK|;1TXMfW7w@yX}$O;I`z%qz-z#;^M{Fix6uhN`!{?
zuPk5;3xjYDJnXzlQ!I`ecK_n<-<7qs`5?=n>j!Uo7RNS-1hVRiCyRTMM7jRyfbAA<
zHAdigktMu;e-<?oTH(kEd@JS`ARr<5vu*@wG8)Idj>E?m!rI*#en7tfeFxOjvx|%M
z=s|jWYvfwRT?jSqEi726Bl7^%0{&HD{mA4byDsH9F0Q63sk!~Jk1;oDseGKSMne%C
z>UHn~Kz*)urbDYAPS{IeYumhnYZx?x(^g2-l2zHw&7R{M$YqDN<5Cu0897;X?k0Y?
z`Madrwapkinx26=kCnYfO?4htO)mrPox%tr*yAhlPeH(dcCHdux0KB2_U%9#Ug-tB
znJ=|avVaDXd`Bpfy~kc&zQeP=QiZP(R|&;TED=pd1p<p7FKI(rnGsMw^kCRVyulm+
zS*z$WES_D!I(iIqPYMb{h&h4_0S=Fp-KH>D-b$P|L}1lxQM9hw?}5C!7v`dPzZ~47
zn%SnP)Nm1?ra<{~2fBp^hyOnO<e&f<N>!y)=4dD_|3GGf(ShANc7Xgc0q5S?*KME<
zma-*Mqeg%7<cT`i0_9Q7fYgbr3~4P~GNoi?!F~udJc)0deBWCG>UxE882(|Oc`=DK
z8{?*3A)OWst?@WaHm25a)a+=Z;G3K)lW*+<3muk=^Ze{X_;qL;58q%GT{arNC34*@
zDI+t&^F~HtJHoler?mkqRaw<ZtL)Xb@o}Yh^3PR*X`Fa}h!}5{7xDEt+%IWsI{!r=
zdQ)J%%%gvYn7W(7_?~~vsiM?>WNOOzuj_7i{X1}<0J$yq*jmY4U^CtD(BMnkGg>{l
zE)waF{qt#pGdRbbUq7vof@06cXnYjrQ}_(Z_bse{0BsnhSNuxE_~_{A+&E>-D4^U6
z9>0q%{0YIB#{LbgCJAvrz)NkQ-7a3^TUZ^&v_qeO=1wjIW6Tiv(eSdu?0Hr(<k_=x
zP-tm&yB-qCW8O)CR_+Y&<=zJ_B{o!?bg-fbgMffDt`i`_b&HIOrw8)>w5sV5Dxw{U
znuRX7RO+ET@+MV+5IDopN>$J3!(~HbKML0TXX|0v!6L+kluas`PU4b&Re@`d+U(mC
z0{4NS27YNLKPm-95Hu-!HEU~Xy58%1475m6Ub%8*FwepiT|>VuHhl|A5!79C?k7=$
z7zS5i@%s1KigD5q#>yU$K(}xS)CP2T-zqG_1{ldnT&g$}Q_U@W8rgXGVL}ZH)i|&a
zX-;Z^Kgee{p?-`RmfRfb@4u9$u9D3}x<FG{Y5Ig(ESxyM!IhWS)BwF|OabWN&K+p#
z26#w~p*qvpP!Eq1OnC6Bo=^k*ijb$Zy1tH9IJnJbxLj!}QgrhE0Pm?=%)Dt~p%*7j
zIr!iSrpo{^k6c{;5Nk(AAc7JSA~?8%R{nk4r+Nxjv8JatfoEgBMLRUUnQ+3bZZw5H
zO!&WA`>BnuQ4NBSI)cdvCp$Wff<cq*@jmLlk&*9lHEu(rvG3mPr{TFu$pRyBX;(S&
zU{q`8Eqb(b{T*=LZyI1a=@?r_nIa1X<ft^VO<NHTl?|qp{dycRpBq_D{|i4s{WRKS
z4jT^YQp)-7g6{%=AQCnQTVGTv-HrJSd+>`Z$XfQcRl8f?yQiLUQxYMNcd8zbb+Wfe
z?<iAvqyj$~y0ZW_!G?+{k7qh}PGM124>^30nJ6vJEX-m7SEf14+Jdx(0s4SLX01ta
zuUl#!)OlzeEp+{4ZT>ALT$M_-do9m)ZnOSO)T{iXW;JgjlJSq~`9-ECc}*v`UooYi
z8QeW{y8qBYnwZtlga`#UkqxcIb2m4c*7Bx`KZ}v>GI=4;dBLamMc<o|jh@hUbkr__
z?~4V6VZbW3OF9f1!HC3{MD&ZMaR*H+188s6uNAbb5#yB7l!2ojqoQg9WgcxmB)W02
zv9CbK!at(B1KKM7M{@zG22Kj9yBEGeAQxT`5-Nbk3nq_~0tn*XLa&Qxj69}segihd
zL%G}dMp9H%6oot34UobB36sdAaLXm1-^4(7A>?lR$FQwFezbRS!IFbucm;qIijuME
z>FLqYWl(=`90_H~j;aVCNB|z@6F)qihNEv6id3jrfy%v8>^ndrL;<9S6y)UKpC~0M
z6IS&Yzb7=&!EhrFOyi5h_d;QxoW2dCFjf#us^Z-BI1JGk>@3{|jjh7N-@kRPTLGSr
zorv86)1fX(_rwn1l;r@w#z`=Xa7mI-=n?NJDe1@V>}<0F4;egtJ$%189e3%nswxu{
z&z(bVM8r-&Y6emYmg9!M?I+v!c-7CgW)JPYu?f!D=k^r%PKBRjdZWFhw-9*`1v%z|
zjIm(If+2bah2%N(Fwn07tdqDIXKQWUm1mzml}?Z6&;?9&*8T>5hFB>8h*&esFB2;Y
z(a=FUP9%H2Pb%Z%zbC1%4f2$uJ&Efs4L~Zu{iMbQN>*h8+FMEq>U1OqaB-X=nFVj#
zZdeWC>SpOp1RoS*&nSVQhNqLgy9{HG*FJ6^!hhXg8D$=P@dZQNcR<O*<UL}Lt)1O{
za7-~Ul5?jc=-IPh0|PujB3<T=i5<YC7uYA5#_YgnIFHZ^1EBVHK5p(gOp4)CshAt9
zqbf3Q*2Fk1f#Ylc!^g~g8!BQ@#vTPWc;g1_;L&{ghK?7WEhkt%^CJpI6_;m2Dn?)(
zm-?*Y7{s7q;_Av_t^d8N3wNw5b|6~Gn6U;X`0S`ax&)dV;LI?WsQ%;502tO5E4{Px
z-sQL@J37Mp8JQybAmhEXDJscr$dj_V)6P11zWa>`p$mojuIO7*$cO;G71c&0H#wg5
zt)zVx#j_myIImRImWJx6*4d2Q()r&f`n<SBZ04qK-Hxp1G#^WD(c@fOF=>_~z5hn-
zV*0ErmDmAtF7kIr&wO9J65Mg+mfF3v14D;Z8xB#rpAZu0e$UK%diH_|mc0lU3_;S;
zEl5gjFNBuyvAs9^>4f+pii#~ovEw*OXu+G60PF~jrP}x-BKPw%qnIx;7$GHrJa)sd
zmm)MtKEs95Khb3jN7ioYN1}nzM{z5}vX6Il92`E)QaWH>3pV1;Z?A=M#o<XiF$sIR
zwc}DFFgYA3)}j*NfJkAkU+feIh3|QQ${W8j7(nAV&`1oR*J4yESXW<f0Wr5uJA(#&
z+zOCrg)mgD{p<@x+}A{=Je4<1mG8d`9dv%AJ_ynCf$|t8faDrWeWcv>Tc<$h#!aGX
zaNUv=djyHn6r885`bE02PYFqc?&qO?x-qBs9lg+iF+)7sWe!XBJ>sAdg8!iwHta+Y
zAl4|Aty;&(EwFL0BjGPPj|+RJSY=h!a_{CwyusIWD0v_NO5g2GvHT7-Mir@EWLsqh
zB#?*4^%7)3x7cwzibzVUJJNxt;?G#%eyB^}g!Fxob{Re{ER3XgfnW6(Q3eX14*iOi
zBC?fMEf++;e-81e#btc_*qdV>;cz+kJn&hwd-s0hEis$VKuwO(c34RGES9KH^V@){
zFb*Zb7mIV`<m5JF^yeFgKF~mA3$ey-_0MMN%gATXW{Nh~({WXmJNck#af8^gIxp)w
zJDs2;1Zu(()^%QxF!MyMu`qM4SesSI72!@LNtt+qS&J9Krfn&p=-#QyHuJB$6zPS0
zsRI<*v>(h=M~;9sdbc~HkjNPzi;pFPv#|?W`$1-#CdMrZjVNwT?Y4>7AGzDQN@XD6
zyU<PN-btfl*&B_jdwg;mtvCJFKd=m2&YYex_!FqnX_*-88JuIlmB}r<vEDB}mCi5V
zWEbBV`FH-Q^Xp1dl~+fKvfXaJ8=7Dg8azEK+&wq?<jcXRovycx5*I2O4_J!~MHuvI
zyE?TJXW)<ZQy5=mIL-|r8TV0BKd>VP2@r0FiukB8l0aZ*h-FaWythEj#tMKFL$_Cv
zgw31OqN2iV>#Y~Ywe9U{UDxQ5TvY%)hW&V3%i#!oC%TJPm%0{{F$Y3u@FQrgsHrd6
zUq7$|bQok{&?309IJu6n1KJ`V2Z!uyT{P`Vc%>oQm?l>w=7v(BHIyqQ9_1L8(-w0=
zpVjJSfBz;hYSdx_`}$wt?7+S}428>hkDuIuN7#2@>=I*H#2tFxT~zR|YK&z4?23hK
zEkATc5LInlnw3zoRRP`rCCzp*J{Zpqu<jtz&T;ihFq8>m15Q&5vjF!^&!Fd(eq8UO
z!J5Qh;lh0%U=N7a2tR^ED#-0uFt{kDr`nMwe_L$Nj2-LxI1PSw(aIQGx7(fd^>si_
zHu#7xB^Tm=h9Z*{pKY(|?uM~as}sj(+`UDplK=#`?pHmHjxSoe16p#PB$Q~(WI$is
zfV8NI(0L>p#RISD+!=W8eE*Hw8$H$)wUMc*`}iQJdZYrXA@(keju8M)TeT*9C}7n4
z=!LZ_=;>u-YV3Pm?Qx)joR5aNf0}!5R6}*L%>oY15J226o0XA4*l}X&2C|V!&NKcW
zdz{SBt3h)2oZC+cX~SI6=7s>Uvu=Ze)RYu{y9t5_jbzMpZ@=IDR&Jf)7^K?7`d}C=
zfGiV2+0bd22P=E@c_ej)c(W}X3UfD%pkpLu<VZLmqikd|kx00sLK;7f2!=Km3^+&f
z*$7+eISDWIXH}fE^Iuy2Q6CgYby|V%ZBOr?jrGHejPwOdOE&|r3)Phlf3S5fRLoQ)
z{^d=mWLSFKZ6?Vczlk-oVF4-oQ}nG$uTNN%B~nhdWp`Pb$f(M`3^>{Rq4}eBXOodc
z2w9GsJWt7mm?wdObBl{=-N8*SDVCO&fTyu;rsjPxHZx1+&%9Z-_HjR)bwMzx!-5mf
z5v2@d6~UK<858bTxN?kgv%;5q3cO}|0FOXxaA*F;ZHw#&1A{rZbn!i_4Nz;vn+6B)
z5QfQwr~zVcuiNH2wmSqcTfa~L-F+n-=@E{E1q6yg`GXshh}(uBU_VSx!803g4g7pB
zT0499HaC*@7yd-*i@wV$2A#+t@rjZLuLKMbuIGzA&gr<*sx3|;mrQb+UDw~cnf8zS
zPh<+v6+z|zISy1QD`<^EvI6J;^2anJ-6-*p<=I|4udrFbzIJQBGziN;rN(?Uu;Gsh
zI>=i^aanRJ+r)a|epai~q!dxcvF!dfx3Hj-X;=krfm|d%<)K5?fDF;6fQA)TA@B2N
zBs|=wPQ~Khr=it*QUDQ;Bx*U<py}`x3(u9&PSc$1YzXTE6O=$ffzE{EuVPQ!dIb(M
zC$^@<QZcgtx&g^Q|IEzJHgy-HOY2k7aXpMnvTK6O6|;9ln*m4(B!-d40VXBlEx0)l
zRYIl>7Nr$gIs_1?cuv5e=NQZo089~YgWC-c9g3V74{~yFP)<=l$HS8)=}Wbf=BG&t
zNm=qJR4h8xN7aej578doUJSw!LdQ<Hd~xN0lZuo;3)RcdpFcs>aIugXNX8{W6A`w@
zoq;+!LF>aU|I6O`b1~AyQm~~<qe>#2BCMRFz$M0^hhq-#3QsSu@i}v)63BbP_lZzx
z04ju%8orijnW1z<Q7aJ^j0`nl<H5*M3sQ#D92|FXH{ruWz4tB}J*LXx0WzZA6M{#D
zU0~cf4yF}+hm>^>J;$(*q9Qt)$OEt)k5N#}<2D=Hql#QC`KZ9l>78YOg%{qMtR_I;
z0_AieA+v(Z05#eYVkBM&oLIv8BdP2FvV`a)=3Xr(q`^pyhf#8&96U-(y9(h;qxb~;
zR1j^j^z6dII+(n}-y(?VPJ9_+@{Eazm4XCx7@P&<(IPxN$`v(ted<c!a}Rn#n#*AD
zeN3aFp?&P&WnjUj0*}WXZsSl98d0Wa&L+K>cJ9WL2AmB<`8LR^%yNhL&)@qm7vQza
z+LEP})htd9I7mObyT@CT50jI>sSo=5-MHWjHnqDrduT@~rm4)L3<0|i8>|@W4&B`a
zxbYCbA+=S_M^3J{fY!>XVb76zo2@p)I(jO|hgEehgLDsc5PksQ=)oR}21Z%Yr{l_#
z8e${8kjrD`5hp5e6fy3h2zn0WPC-UNHxTdqcA|pH>esC*JrUPi>CKJ@@}Aoa+x#%`
zFQ%=}nv&tzQyjL$eg5U<GSe)}<TK&=1(t=zFO%DQXI+xPn&JIncA@KHeEv1Zpp8>Q
z_pj9!gbGT&1|n^x>)mBN8CLp^$E3rg@8mR-+1;Pc{T-wNY~R@0$F;q$^0P6^)Q9xh
z{1{ag`RMeqI`5lH#PW-ojE^~Pub0>Kt4l--HrM!m=?$K%=X&jJ<}2raxcv{y!n38y
z(%RCA9}|@Es%uU&?N6(Z`{&a-@;mZ(USKM?_IY7Vb>Q388M$x^&(46NU<b3Nhf*H`
zYtz(XbUl1X^J^SJGC2DU$6qY-bB`J|b;srn)4wiSo6j8ZRgG02`F`yz2CFMb%hBNs
zpyn>;N#zP8puFr_a{?gNN@r2*d1X%whc>9!rQ<kY7o1gd?zx7B!k)dW_3k^-5fZT;
z@*BNn7a|CCHpEz!qMFHp0h_^UN|--!=Sc5D|2`ecTPw46&{spGtl=x;{*1#db=UgW
z;9#6~^P!!yL)wp}gRtv~fh%{07QmkdW<%t+5Q4GGQk|+$>%ymgTmQ;fayrlDA=G2>
zIPlzR&ORsk$&$#eZRsF#b9%YpZA(|2U~h=+%X9LEM&0#JS=!vg=FRW?t&WjP)vE3E
z@j^YsX7MNqNaHx#Nw-YvXr&EY*#Q3nK=M4BCK^B|lyq*;5H$j0p$Pk)VnhqxKjERE
zQ?0F=Y_oW=MbTo)Fsvf)PqTmMVq!zbibGlX)zX52pN!Rp?m^Ha!Tl`fF?^i5(Tfd-
z4nsupISWE<FKLqxLL0CBZ_1GSx1WXT48|sgO#SX4DDdPMCL*Cl4OiPrx6NV?t`Gmw
zGepn>%$eIcw+pV0h2Tj#RU!K()yi1BetB5TyKcLD@zaXG$}VFmh<rge6EaBPutp>c
zS%d>mO6H^+o}Nzc>?d1MnpT9aZHjKeXVIDl3MHJPHuWGA$c1@od+*vd523vs78Vu@
zLBw#hdit(%TW}2){_@61w=)_@wclUg^r`b8kK7&oNI#rX{?!@nPiNstQb2o*ni`Vx
zH@MG_ezSxGZ!4RHkx@2y27jAB-X?Yc*BeH;DO=rce+v(ucJE~rg;X@I+tflIYve#$
zS;qS1On*68W4xOXJjcfw^CUdl8loTOhR<JCAPdSbDByiL=OH{I$kOPzO`2}@Li7ck
z<Z6uH)pq;vDWgT6q((tU*OQ=daxN$S!e__1cVoQsfBtM>@c3I7og*T*)RgwVZL=et
z*~#TgPL)go0$R4eA~3Ir9O7Ho>%J#-+aCn^Ko@$R%M?wcx0l26D9g0A)M2u6wf@@n
z;2KZcru$zbg4<XUEBiMv6BanuvzL~M>Ttf<wr$*fT*^mVQ?>Yngmex(;?r-V-L{P-
zeB{DL6t)+zv#uYhqF;Zf{k5(3NsYWEkNuRkH6P?q^GCl&ZCOadDVWUhKkly#^t1mF
z@e~yv_>bA?1D8s|A%ys|jA#At|C&AiUoqDKiT}FAR(cWA*2JIxi7yzK{q^?!xo>>P
zgwNkpg4}E(_|#G_A+C~zejpd&<_rhXnVFzLPK$g$tpm)~1`pmy&1fn#DDA87$38_-
zq~sw7ty#${pk^a0tnG${o&24Fg6bNXN^X0$2Yo)vN&hbJ|Nm!FAs*B4nVC=%2ckD{
zhw!wvk&%-lkV&Vi!^GhDVZG2~!j1U=;8yUr+y5kF2XJGK3vk%AhEw>5BP<|py^TfM
zIp2Z*>vM1@83NzLpa0zjTg&i2d*lD#A3OrHa}g7_J=pvNfF0*EODG&dy$Z9)(2%*0
zC%i}A4GwF$2@wTJzKO@khvpFnWeJCOz5_0s8Kxhyvhp!T02!tXR(K+a9zUlCd--T%
z+*W}H&ERdp<OiE57R9Yl>GKj4P<YQ6v;x=HGf*MI>{|vRV6tc;X+&*u>qQbqACWMa
zydX~WjWb)pRx%)CS2l$GK;NQ{W5wSe*-Hdw(Pz!j=Kyp<RY=#*AxWrF|L4`LFxrXk
zsY<0ZbgeL$f9J&?{NtA=|9*-$#c~>xfcXI+D$wA~x1Bt(^%;C<`0h>&537aJeAn3e
z694}78<Gzz5t!uxI^wtG!Tz6jKiG;4-J>3fc05X`gIf;r23z5u<qdB8g*t)hBX%6m
zRd6%-oFT4{2XAWLtEYk`idopp7KCBDpZk{S!3Y46^f;r30(Y>kK>N8Qo$-(I4*!1n
z5Xe3t*C%nJ`uDZCwqEN*YV<En;3IH;y4G?8KTbjS@0uW<$w7r^J&^F$7$v^ngW{80
z->(S?KfwdooIw@KrDO2_dXLBrB~`-KAzt>_^Ye(v+dMB3x0ig2Es#Vs30IC*$r5*%
zB>mQ#D5#;Mft+x_IW#WL2)U7=G2ygM;<5F0jx^l<sdew(du)n*gi*Yx{?_$Ku^o5}
zvd~u7YXDsbOo#>e?{B=%&c+)h2W%&vKAbhRhy}QDcI%t=i7mcp1rY$HV8_<u@_0tv
zwiIO0tNV>&_YaEz5qBvLZ!K4u2R;#+9*p!>HZ~KeW@;z#!w>ZTee|T}9zWrvG%+x6
z%9{B8H;!z*#m%>9P$C8HvtrtMsgqPBp2a`RMG55vKHz%RAMpmEdQe0}73t)ScR;xd
znd#bA+ynBLc#Iov;4O#l?wq}CKDA4f69$%0b>OjQ!<Sch5fVbEru_RFQCO7Rh;0RD
ziDN?i`Jb2){GW@A>JfR^4cmsm%>03x+J&qJV}no(!z4av`d6<$jnQ98(E6ALy_-*h
z!onM9vh7TTW<xw8Vc#sY0@1q~z0g=*{tis>ATpHp9B4X5v*SfsGALN~gIr-c)e949
z=}f?^?z)kfkG4c*3+R~Fq^T78GJgKOA1$H`_Y%pps5#LaH#7&z^>OM@cXxMof$DVr
zT!5Getjk$KVI$&hXL5rI)zP+XXD@-n2*q#aYoYgmtf1}f_2^M(o+Q2v3*FY{CP_wy
z45>?uTk_gV;P2)4ZJoO#598tn0X4Q{8p&918P`Z2BTn&yMxdvl?N{N$M~@h|so+<1
zAY<_C`SX)|+!1bJzydrCKta|qZ8Y{#y`uxGTfPftAv$y`&^5&2%|MluUvGyS5)aQj
zFurt^<d*|8Acb%k(}4pvJT;}8b4LUJ1K$`!$Y%gb0DqI8M3fO#klczg+ekc5@biC0
zYZ0Ia@Fz<8pQt`>t3b{PG8wm#@S`M2P2Fr}gJX&?I(fjZ`2g{bwX&VeB5p9rUE^CJ
z;X5eBn=9j-RbHs?W`zn+%A)9kp!@*vaAFYXU^y^8*!bb&0S^Z_v<gBxm$)yMK;H1*
z)#}4AOwb@sFn=A?1BhNE)HQ(lEjUu)bE7d~-jP{`Wqztktiy;tVWRoqxO6ZpqsD!3
zK+WSJMW2G!CqSZ|u4yPWU|W-Yr(+i7_)ek2#D^vA@5g_U)B>_Z*a;#7+@{^70VYKQ
zRXM>;#3B-d#28g#B!gTy1ZNWex{fU%s<2g3$lt&`fv~V}-z<#8jcfXWQ3+r^*)mr=
z2J#v)XigRtx#7Y<^?UTl5eHOxGtWR3vp~b<xE2w_KNxI%W)9$%zy&~HQ&)e%s*;2Y
z6Xc)0Fv+D)lxMjdNS7+la%wz`*X$+s!Y3D$#MlIioao}$5^gvEAc0bu9VHY?&|y3B
z#{(}N{bcJpZ!iM9xpMTKJPU|Ob6|^J71qS;1oD0F8uSx*i!a21>;h&QMz&?&3Ib6C
zJP*jSEnW&Q!V(@zm{iNFqewFYgDYU68-su+H3%<CQvoA9wV*+O<E+MW_+k0jv%>gN
zU+w{|1X1^UzN59BT~0s=cpi*32+`qu7HN|ykiNqBntw4UJZ@mE?X)<JU-cm&g8#OX
zU{1Y_<n%%m@&t7$pae6Kr)mI^syE!G;AF)$_@K{otHAJ71jP)y6mF(s73VN?TJFSu
zQkC&EC#PJdwiP#t5LxB7Z|uqucu{Gpa$-l0pt0}$>J`90S}?Fcbdizq?h3;|I$Gj{
z;3(n?BO97@e5%eM?)$3*#{%1{vVuNMH}gw8lW1FPviz_kgu`>R4BG4XIz6zH!gf<h
zRFt$5P>w?{XNz$h16d=3>?>Rcc;3J=sA?9t%$Tsk75RizS$qS&LtC3}egRa7UnO1@
z6%|myR9Z{;SA)}wl$PoK2zom~lypK)S<&VV1(uh7nV2OZ4#G_@=gLm81RQHaLgXBS
zsea8wHbee!>$blEK&2av{VH}P5eULS_YibErMHBlF@S?-f`rf#lSy0*p<|VwFIc7E
z0NCPn&Of>C3<O~&a}Aw<%VtD6=l#V6G3bH20t*FrfMElLyTJw_B-TO*N4m@;VDX1X
z5)qF`Z1VrYTnF6SPc1h97s!en1mzxiPn8#Du>G33@G3$7rdD5oNk>6c?W%AKQcgJB
z`xoOCg`NLgbGe7947B0r9WSF(fnFKZy%M6Y^L{~aVXTSZHOm1pa;D)ThHufU5)crO
zc}+=8Z8Go&kCnpu-<VtqJ{*y(>g&fSq#ZrM2a$_~VsI4@?%`iA-+u!Fcq1btsA0h$
z)|l0RY<yR?*&<fLex%_mG<kp-t$c)eUmkwLiOvO^1rd;KldxAq4+4Wm-ejvF{ot2A
z0y9Uq3RJvLxw$)cL9psV!?$nOGdpo3NnW`!-g}F9%I<%cJq*a!#A5#1xnqZM9?5v<
ziO(QmR4D;xEtMDm#7$7OCYL~qLhP{}Z5k|C>=|#0t3hW09XM-H(zCL77EVF|_r%?D
z#7C0mZ}L|5_G;<YXrFtgE5*pev!jZ|;=s0T4OhU0P)PfEmUs43-_~Y$&UGd*87Zc&
z?&|>!P|^<3>!A;dqh4(Dk6}$erd6RkBwxvot3Z3{BW{3^K*6+6H`a+`xE*w_Wl*0(
zToN#ShUnbRvWEW)Sd%69skXu#Fm@nOfE`PI(xCe(3?kXd^Kfd7F5L<_XMf;XP>^_j
z`cYjy1~m+-kc!gMe4zB`O2q5S5gAyCEYBB!eSoA{E;h)0M2&z86Dc&DBHAfPw<GXO
zZC#}M2*Y+Tm9%DKV9Wi)tMVUrcn?1Zck=P0M;IBD$o6#K4~i?{*$f5{;%^sPC&f9R
z7%)MATT&Tf)A;<DydQF2L*#|mUf=k5p3AZobvTh!U&|;W?hHc56wHw-FFYEXC;_Om
zp){)Uf*o!$dzcTpPTt;@Ha2(gjnS?InvRYCVgb(|d^5Nd{6y%$ULmM?Q{ShsILKc<
z(}r;#ld}YRhCY%%EalJ=#vPid@b)oOyV2w+C~(!(>;%LJa`7e(2C|5X@^Y$7q#N5T
z{@z<fO~qo0ui2JoZ?ToG;;$)mpu&TP$Ysy=ROEjF4y?$w|3&tQRrvo@E@swz+dO%p
z4YCEHQq+RDg?plO5GvOuomq)%L=7gXVQM;t9{cSn)i0ry{XPYVXkmH83BIA4uFD=)
zfiwe#6qh$pt3P;ihJzzOFzwM+dPTh3|9AveGy`$e5KKvZ{TX61ERWsd4H~PO38(c6
zU2ffamrJ`-Gz761ql^h|RwAnt+}3;~PwNS?Y0(;b5M1`wJ%pPz(v~qdcW@I8r!N0h
zSpjMZ`^Go|EuZ~cd8@7ZNWK<9t{;uoF-Guu;Kj|O<}5!CkE1J+l5o{nMB@dVHaIYy
zQF`YWAAcLU7}yTqVbF*YeNNT*1hPs@k)XE?Ws;ld@u3yKum_4HsIbv4$8EgtpMNs4
zUW~TgK=dVg{7AABq(Qb@ZoERjyt-P`jOWkJK8aifBc_|khl8hIppoq5bp@mkyc|Xp
zFu`C9S<I<hf0q$K7y+$S22uFdyPRkTI9@a&@g@enU`Yb7iie#&6*J2iW5WAm3=i-O
zWEdkI*?69z>k8Es{3$ChXF7FC&~4*KU0vO`Z<DDXd*Bv}Yrq-EgppY-K7|Nc(-;|0
zsX}ucD)GRAjm)-7bM5r;Ks^yYb4=Ma9U6?!e~m^5qFZ{p_eH(%YV8km#}5!-RzteB
z&J8jOtVmBg{sucB+tmtk5VnFcAX}EjuOTUTw;YB!bP7<`c4<o>xToiJqK-j=CuYw-
zG%|7lv`EM<yLOsDJ>e9l1yNWg6)AT*I6D(bEj2au&LqKe=NchOnOThVVeG;=2y&m?
zFawA4*M0dClhuP%9NH|mxp7V1stx@$F_E$wIXgE;@HbzPdNjgSLUrM>mDn?a8p%UG
z3Gl;AzyYMiAbJVZZx{<aR)1nyjS*~%9U=#C*Y+W``@Csodg$=s_`D(b0f2O*t~E}>
zkBM8t@l+_n6rXeeO9%=e<SiQ3|K7D@2kbhzQAvpgM}gYX<sXM;%-Dq)aD*E#+y{lz
zb5)FV9%GE%>3-74FElitxZf;fjSdm>*mc^A1&##R`Df#asi?QBywHN#z1z#Z4dX~L
z@+^Ms92AmR20RUdJY62ThZJ=ji0ZarIt7<h45Tz%l8q9c93JKa&;~9@_m%RB3O7Lg
zeXrB*(!0(#;<hF=R>!VH+L543PDyDCV(#`NR%-a8=^^<C%``SHZd91-{P`U8kZ=~E
zjxw8do#mif4-#HBw@1hrLWvKE;0C&UQV6n=chSN4B~aU&9sI=)Qi2H3rey)}iiMf^
zN`*e?%z)!)pmI>2{&F8>cqLl8>L}kZY%|a$A&j;Z${=JENp6F9uU(sjQ~2EcyfUZ8
z`x|0~?bk)=fqO-#13yIuoE!oIgx}GNO%IZY#1~{gnnWEIANUAkzmn1S-YPorUoL=f
z$}qvsGHrZQh7O2fa>FG&?{-uTAXS7WY+!*MF&DTiibL1h8XL{iOu-WQFEtHKlJt3;
zMWTZbG9Utu5!A++nTb<M9A%V-$^3(0BLn1o>>2Zgb{sh_HlPwG(MCl(lr#aSPX#>K
zbBcNL6gYJd_$x0f8+BZRzS!!*IK=swwLF=)P*!3kmXJs>C2c~J;+%uv+sMdn(1BWI
zaN05ZaMRE3*Jg%&)I$Yj^O6d_*r&UWKcx6O=Bn4D6IWT94v9+dIB|gAhv!*es{78^
zga>=vpCrupTzb*tdzG}{(4MRRFdiURyvo?ISW>tsXgyskn^U)IX;awzR=u`!j$wU=
znXF^$GXSRsu-;(;_0A<0)BK1ZLz1YwwNYg9j(NHpBUGZq4U~!}GW_(C+szMh?t#OF
z(eW;G3vyaovtLZ-!7xvkk8xkB#7O{7Xr@u1D~KlWbW@z~*#|TJd_5Pi9YM8(jxC#V
z!cQP3KxVK?gp44;2FIKYhJ5ecgo}5t&Vw17=M807U|W4dHvvx_2VL{6BFhtu(qU){
z_1*U8b(-i;BZp0q6{QUW{0&CPox^SjV_-hNKUGXTB0gU`&}?#Q==xg4P(S-20-q^3
z3CMv@fB;7zl4hphDK8%>=xBv+)6|cDDFR^wS|Ru{BA6~3B{>amp{L!84P$?q9|s7A
zK-zTeI;woY_&LZwQQku13VK)CAV2|o09FQt)$_s|Xf5D_2ETlHOJ4}jJKlakC9gUX
zV^)-u!C_&Es~^BrLfD7G?>S{*E-ovCM{(&XFk~@xukiNG7vE{{WC2Bx(_7lvz0_fB
zyNb97RKl-pL0pUn{L&fzKz_t<%%tUl_k)j&W`uz4*eRhI$fZ9$_SfwNJR<<Oa>C1(
zobs<eJ3*X$lsYJVUqnX2p7uFmHi3aAOpTpAy@3yYrs%u|sN^8#JqawWs7QwwQ}O$4
zs~7W+e6RmH^{CeZV+fh{I$4^3tlxW~z8qU$g!K(8XIT3CC=Xe0az>Dq<Crs}JXYra
zSmQ#?It7XT*!g2FAoDEvT#w{e(F6g&iFHSoCrCxo^xeI^o1l(9mh}z{+=QGd0Ty0c
z<1AWJJgQ+wdtq?>s14{wZ|tGxuLXUQnfeyiiBC};4oAdq-1v#4-v8n_PGH}oX_WNl
zI`!^tv*BV(H%DQRUNByKfB&CzC^iZ<X8@`XD(#ybX4peU79#X){irfi<R(GV|L}Hj
zV4zYqw;?s2sj{ZVc|7%FY?<)#z#*;OSqBRQLu8GOjUkjAql9vfaA5(ZX^>WskMAQ;
z(Z0&e$lb&zBm^uAKJ>WEU2lG!RQA38*I>!gh40e3x;;o+pwpuW18*b@D87+{xPdl;
zP0Pir?Buk9No$z5LkwwpY>ei>fd&je5XWHM_F2jlf&`w<zP)>aE@XLOWX{(Ya2F>0
zFv{o%`>NE_<LmK=wm5P?|KRz38U%ezeITM1Zf-cAtO&vtPe`7CVrUFD11y#C5}5eF
zBgCX{A#^fvF(^fK{#&>;ffWsvIZ(0?8BuoLM{&jU%?dx+gXr`fq;wKWRtk2_7BHkR
zl8Jmb3`Q|XcoJ1I8&H)k1Z1I}D)#>cUKEj@NbNBBg3nd3I+ceb*HSQx;X<$!apT}L
zeFxfjChDpK_9kqdT2eHcAu!$C`~icJ2>Qgp#j|)Y9$gScqwcSm;8LprZUI&*Mke9P
zRR{$Rg4v7rnaK8ePYEA1%tUBRM+0)}>GPoR9|C|zRetmzNMk34RAoDfYQdY}Wx%41
zB{7;HgV!UCgjx|X0B%qo20Y~IEAum|Dc>>5{w<k^+BkKr!WgK?!1Vjvn}U0|SBBvO
zw?<H80d5({-f2te04;@i?esS;3LaIH$0YBtj(aC<fffSa#y&YlM+Y0pB>+j`6ogzl
zGE9ele?CkE006_t1rk%hLaZm%0;M6xQt0A{<L}|&fu_5*wl;nW+21ZQx|CH{voWQ-
zd%uctAAqr^1DG8=LBL{wSuJp3GVKJ@d;*P3=qP-H+7Cbz=)tRizR1dIiuHw%@H^Ei
zaZGAX+&lFgSzg!Dt@b0pra@$#K!S@eYa1zwGzHI)&#Y|%kpotKfz%dpp{mK<fSQz*
zm4U&bqgGH*fcQgXu0RIARqm;(s>1UjnrTQh@qq@BaR2(Hfb$BO2Nzh70FnL}>$I~J
zg?6k=Iv|A+owv64?uBw0`OMEFFB~^-zd=MOj3y!Kjvq^rk1590OsbRJWXuE17Pv|>
zf<@sBs`3>i?t9(Z()D^L+`eFdT+4#081UY=CH=i?WUKGC>&Lo_2&duR`pg9eW|L_)
zU`+>L%SLPyV`CUCC}I_SrJDeAn_WA1Rz04bnIYz8#aVvwT`u&fgry?W>d`oZe9|iW
zVJjh!VDT*Zvc_{{u{tn~JmLJgXXz8t`H{)hlVIHv3PR%7=|CREP+c&Kf_MxQu1hSA
z!k#y3Q3yq}omN>GKSY92ySN}FBeRMG9gIs-l|)P$vHG1j`27BU@FC+}e9lQ7u(w}C
z5it^JT)y-P6jY2HF!V<10=po#500p(eA6idJc2AubA)(=J5otVGAZu|WVwoc{t}un
zxsNH?RB`=e&+awbt9q1vvCE=JDoPEZz;Lhcp`-4cg-gZn{xTP{GkS^hrq1X9W?c_$
z7D!U~url<Gp);16sdj+GoS5x^*RiVbzAh-QVAsgJ<dA}FFT&<7hUo64K{FC77;7}q
zI;)vXO<upF*8YG7$hbC)$nVQ4Knr&^7}JXp-#NIXh`2#YMy9qOh!Jgv{>i2p0^G5-
zw0uUYh^8or)1Ki-B$i=@rBcYEg0c;{8I4c0S`<KJ&x=Mt{;L-nbKx+72u7JS-)WYP
ziO`ZKYMuPx1;^Qbg8tR4#&SSlvoR0o&EF6{1RDKjEK#_OLfz6Pb-NG_u};pNF!5J}
zrBOXxW#6kN7r3tG#3XY8Lc-;xdY>2h6^10=o{;omX@8G!g?2}EOiJD7{(ho4>#B>1
zpb+*oyt&lg>CWsLtl=+A^@RveVaueDua$Zq>N7eA4k$EijaI*x`vo?eW;b~r1BLZ&
za1xF8j3p<1dBxWi7_fB!zQu)A(-RREb`dVI1#vw4aF<_GR}YO@*o)6W<TNMJzQnbf
zVli<`Fb~FvVm5RV*JEM(JJ=L{M|%eA*5OPq?D8E0Z8L!^LP$*EHxe{^?*r7t|9}|$
z*(p2)5GyG_(}t&?5r)u_QCyZLT%-_lVU)8^-|=K3Di{QH2k1;|Zg0OM^Md8qZ@rfm
z3$V$N936g--@o;URe8j%&;e6^=)>K(Td+}t=0;1Jjx!GHa)6-=I&=8)P(Ai9Vo7}-
z9^Q$t+Oh9R9fa;0!9<f-1acRRCqRrOePljJOWTE~fQD1Dh>zGAL>n}XJ|Cn<^1>+n
zA_0AO9C1W*yY&^@L2E?uxZNP^GHa>-`J^jcjJ!)9yP(OVt@HX#Ogv#!l4^r6j7Iol
z|0jXZS@Lb;1STygH9{Q`A&bpBCQ{w8ZTU=^F4+s9_7EJ27R|_~W(~!Wg~LO$%WhxL
zBCIh(w|tov14{_|r2bTX0Rj8>?p0cfAHj$N1|Wks&0y8ng`5(-zd<c(|I5vQAyH(t
zHWYv&VcL-yH(UAyU$7-zZxczVGK)2u9SC4x{mcL9!2u=_+FhWKHPlh)n%DCTO-=;K
zu!GX+bMV7G?U}rA8L2=N+Bg2xRm3zPF++&y8kpt8dVK=diq<0E5){Y|XQ=aUfTRUE
zrjmwHV%%IQz7E3ybFM##WOaEcf~0`L7Ba^LK*-Yz{%AfsA!+kL1)V&UE&>=wh04^R
z^{C*;jvYJHg&!-wZ6qX;v9<c@2GA{y{TTB+F*b7d=;nnR<g=FVGm&Zoe@6C)nFZO1
z^OLw3fOBB@08t7>z+3q{2o%6qs$<YLgs}(aK~-a~KC>uX9>Suda)M(8s!C$ySAGJ+
z@@L+Q{IgFZ<7OGMA)AjFF1(23fl#+(X2uso9Y2G{2r;yU_FQ2I8!yc_RCAawT*qHy
z&*HPpqalKzj0PuBA>8YRj+ia>LA9leiwh>mGBFPaqBopy(1wMq<PBUL=#RFmNF=x3
zpRHofyq>3i?OH9SeRycD1KfxIpAGgC{CH5=vTNmBJEHPBD5web1p;>E<*RaX5S$XR
z9eXiVk1s@|r6?bz;gwxk`3b#4B<P-BANW$SuYfu~>l%E9`It<4M7`PN1QY}$7XEce
zQm%gs@WvoU0*K^H=C5r*<wdNeE>g6>%q|k*iB^jYB5{L*wiCuWh^{%<%x4Q9V*`zY
zP&6}`Dx2I2cYrnA&{%3nHo46XNZq+JfvgY?N=nG#o;=ypp+Lr89IuuFrcI`&k%*+z
z!4$d%s#uYa{|>Yf1GboD%mMiS`Jy~nz<58xVXr+!!@#iG7I#~p%mr+i@N*tlUU)u!
zJa$Y<PR{M%A)Z?YrklEOpq1nf>v^5!s(-^(6A|<G_=kGoL*@PXNWSn8bL}QIVL66K
z3d34NCI||>X)ip2XQA26PzcckEbPcjnIh~-)a%G<Y>bdYKegBkkK!^c_1&(VNuVmI
z5}PO4KcHox|8)V9%GgXy%*<>R8kvUPQ@#0!AW*ovkM04kKIGqs5kr%cschDxO^h8g
z!u)8ZqM@37&oTpi8>Cu=kakBg2IfDw_gDOB;%L8xL_`Q+PuL0<AP1VSsklmP4O;&T
zXsgI@XyXA9-l9y`dv%eO<JvS|^B_VO{vN`kiX4BSb0EDt+=M3_$vDx*b2)ce1NP>~
zmMJnnvwp(YLqu4>O2a@0f*0sF>V38`LE?poyPt%F1_S*urG1F}IEbf(?%HWt5GfD(
z=O%hQf`UK3e*Fdfk>2IE*ww2~K$-FNl|c><oCN8Ek;`o>uz~w}Y!TJiJ04xXva;!n
zeIIgzbjSDSI#-6js&h8Ds6;{!;^gt;Vw%_Tle=GCwD$W4j^s9TL)T5t%}DP?Vo;0R
z_hp7ln_t$Ds)h<-hV$7Ny$*u7qdWo70$o41jkVl)4*i<}b;vK`%236Xl_Y`hz_Q9h
zOgsW%!JP9x6$2!E;)fd=C@vPPz|R>S`0?PX+FHLSPYzViPj;!DZ+Y<7_D5WxY(Gy`
z$Kgbq-PQ<gLQF_=u%C3f%S+>LBc6~Ln;SzsNavRmF0GJn#c>}yMhM!JGSi6YEg)C!
zyKNeTTB+v#=QA1<a&`{+)J4m-NceZZ?L8qgdflrwBtCu^KVT$LRrldM&&(u3MdnNy
z_g|@W4P%D8Q;D%S4>I8f+UDb`*qBI_*{)z~ymYYTMwHP1yO42CC0jG~_es6z*_1<h
zwQ7`sv38c03QI$LhDcMyA7S;S_I^(u%vR+#3|>z9n{8_N>9c3Y1uHP<1$6odE6`@y
zZb302CkY)e?~Fbw!qw6Jw<yxU$8@l_F-7QoJoYps1Zn7nUm1q+l~P_kkUmq@1l%tx
z7;6Napot4HO+2#|v+H1;XJ-2Z1W;+#qRP6eC2E<Rs+I6nyeXKY5aR|e^+5##XMtow
zQ}F6^w60#~(sB|p1~K#|Jq1~?%=RJYGT9DLKj!}i&Q~)f`IHklCX$1&S&lHd^Zp4b
zsH`$0e!)8QT26k<XnCeq8ZLltJLEG&CsKF>&+WykKX8Cbdx_%(mYJ9Wlf8DWrQ4{=
zr4^Z(bUp8_V{~Tvg<~~dE<b#Kl~q?D?=6-Oz_n1-+^nRQg7dfh7!yGasO$F$B`k3z
zPK@2!d7hs6%qNbn1)j%CURRKg0=qHt1(EOy3v_Z$eP9zr26Vg*&{yy{v;$5{2v9=R
zXV172*!a?XvV_R#@9_e_X(TEzIcC$$N=O?(J`}ezB_c2xlk`1BJ=m{vLN)Tjh?ls4
zBAf^s496>D7vZ{w$_fA-N0FriFU02rPV-Xu6%n1|&XErGqoOHVof7E2L4Y7fd=fvF
z2F4^x;4Hv`zyh60G&S*%4LLOl<E#;pj_}0g3;vh;L8#O5s%)4bJaTG;(_+$s-+y`9
zy1mn#aaGcfPN-8J($d=ccCmFxk35~(Yb$smdo%q5n_6o{JhYf&`NE8V3aY)mCzX?w
zr8J`%2KvvSdb}e%g-|bEW&@jLY0gbvY#RwiLQjJN!7IAM7IRJ?ZN;QRIM^o>Mp~Iq
zoG6wjT-;$7HR^){eO;5Vy%2vD1++Luy05@8wp|%(^gJ>uA^&$qK#YqvD)pWzHZ&%H
ztrmM<9_q)VK)D&mh~Zm7Ovzy`3z$3V4-{rK9Lr-PBS#J$dV5&uIK3JGAv__1T!e)L
zBzOjZM3&#YIi~+(CnN~77i1&2_u*rNQq-g+;aOIwX`XKQGa@-hpaBHhxHlW~<Q>IL
zDu3G|4Pc_}z}EXZ(Kv<skfePM-b9=ml=2Dk$1Ovuf?Qrri1Y8II`e)l?kxH8rx3C^
z6@6DiK{=*iCrGeN6AA27voyf0jf}&#`(K_z4e?Jdt%S@&ifhHM<HhZo$T2L;%ky^}
z_GGcMe(m17mDn_%Nj}ytnXu}#Fz(O+%z?>0O?v?d7Xj#S+KKBquB-W5M~B9%9Rx?H
z1uYrdkAQjYQ;h`N24z0G2Ho*$a%hqg1vqz(_bQfif_YP9bGaU10i+<Nfn_2I0``v@
z6#YFZnd+$o#u|*xBN;*Ji>rVY_zVI&uoc}*Ln#xff)3IqAyHnM>!^pY#7($xyEl>m
zmcYh{=n^EVond4wn+@~Mfno#lw=)vWnfr_`k{&zUr!ExDK9INZz#ZT_b*GU7vhlvE
zV8qdv_pjD6t^YFZ;Y>d4dqM2RKu(eqv1C6u-p>iKW1pRwxjXvOo(?lDMB_|i(I?$r
zv&r$md9eUrU-<p`Sm}FSKZ_ZZ8DcE(t{!M3xSpJL60x`2|LM~MgFE?Qcl*1X)?(yt
zIyg<a(o{j~(iyRH@&G0ZT;pIM8vA}J{5b5?a?!D$U&75OjG;^<QsaNBh^89GW@1TQ
z@7_&&Ch*djCmg#Tzr(}IT8*o-yIbhQi9S4;AS!4`A9O1|bO?ilk&w{&2@P9l73ADq
zU4rt?G0<c_hfX8y*iR>DY)IN_zq#Q$%3BFek%<&Ec_d1X7<D;WN)UJL4ZGpvCr`d5
z@?AQ2(d5=Y<Lz=k1fl8BnNJxNM{=0sO!r^YDkFkl>?%>IA)lp~qWq(C9inE7S_HR^
zWD$@-wsE7lJsg`$s+~_#uwYMrrp2uv;;H2w!mg#N9w%81!8W9!pSxaWQDKmcY9+ne
zObSXKEa@M5;6wJy>rvLnQ`+AkLlgX)mHNb)lcCUL0dc#)xs3bWw}@E=!tBG&?2I4k
z#FP3zAJk#a{hAI1VH?#X$p$$ec!Z|dH^2!KJL8#MF^XWyHnBsr=%(melp_LL*yYwC
zBzbmW!la4J4>0;25ob^;nod9J%h<>Idw!mfq=5l48az=r?7)BVb&#YT>3q)#{haCB
zRYZ40hQvc){)znt`Jww^#4ZzbU1<qwhCZXrC<isFW5h~|jWe2ABiL0h7GuwQ&+A>_
z;HcC5y7$aXljP8#P3i5bjh+tzdyz28{9wYx-&1UiVjkGg<0<q{limxYGDdpA<?0u$
zRPG!~Fb`}EHi`rGdlq`K5u$a`U9iu?(^{4-q2MQLqMrj2BwjDpnSQnv4Yb(Z>=S}e
zmy%Am2XyD>a@e?;u=5%HN>J_&JGGrwNqX_Lh>E)~j|<>lfuAl?kT{u8unoO$H-oFE
z#eY5@_hJ*zONeStP#y~wAR{Ju0ZAfx$D+dF3I2~jt9x++k`y`3qp*~lp3S+VWswlm
ze#Ce@3JI!qR1P4;_px(NgM$}(oTsJKxRH0gb7+phi^zuaR83rY=EN&N_Wl@APob0y
zVN;Whc>o0Vp4LVB;h#SlBIRPQ9C?Q}EysL0Ng)8TvAejNQ?Q?3o;i_r_P+cThp+7U
z<&34z0w@j<OvQK4s;z8oQ!%tNmD8ggjJCmEnR>usBE76&_33^xH&m<=%=DdRT?;zb
zgf{Vf&|ZRCcGn^LS?pd(GoEmiQI)y_Xt%t1s!YjkWu)EVdZf(aq3D`a^YsG_YzGb=
zJf6p(3G~v_=&zlaB+gkF)QH3+GXNqHh-*RsfKWGKKaT2f;{Rdq&BJ<bySDFeI+2i6
zD)W>kLx>_{Bnr_e6(Uil5J_YvMT#b*NHixzN+K0PAx&tOG-!?@dOu&T>)oDhd;fo*
z?RmDlKd$>c&%5FGJ-)}Wj<xoEU;C=B7X+p7c9G1onkst-N2A8T<FjOrOFx==N9^X=
z2{&0DeVucA_wActdUH4uPE=)6r+%Y=IQQG^0J&z7`r^X78_Qlukv|)4HP;)nf9;kF
zw^W9w4(p~BW@%k!vZ(z-k)?1k3ZB5dv)A#7(8KCfd^YvpQF#S_>J%Aq-G_&mzHqE?
z#Cf|9f(@-cdEUUL?oRa=)T1`6s4Rs&xn3u2E3s<VP%e8CgP89Mw-hILzZGzTG9yE}
zE;s68%b?#}&FL+Pr=-Wp$;y(U>s8LZD}T#lMU3UlP2(3$J)EPVu8wNciHr^g6HRCp
zi6`tls1H}CIdCUyp}laS3$4kwD&?~G>gP|_x*LW2pSYQ6M2|b&k7Y9F{Tc6=CQAru
zFlj~09|q!s21R_IBup9yaTWjF($HWUXX|~X=KSZq|J^y0)>==|N-ehHY84$r`$N;$
z8i8Do-}3gi*EhlObnW2RRK@}V3z7zHJdRSCYF*yKb<*$dCI{j`Z$L>Hs9ryfkiPyo
z7Jrxd<U7RsORE~zT~J<KoZG)ht*~uWX}n|aC<^%Ws^r0)5wBgU!bktM-<rM2;8t%(
z>`6>z?jj3!GI62~jmeAEm*2M_^L3i3)*jiNE3?wl7JM@=kX>8$-EepBJYwH;nc)F_
z0(^aKI*M$41cxXH(+Og|Jgey+*)?v<Y3-d{iKG!dUF0rw+qh9yUF+#Fb>n^;dy766
z0QohKL{Kl1zLE)JkF%iu>9tJMtZC~mEp(R|@jEsTOuC@^YLyX7Rw=tIdN?>$x7u&o
z#JSEsXKfYrg&h+Qcil7jG^PQ+0Pe_KwgK?qVRE8YU!{9)H9qid4VY1UWqS~iK-!TW
zv^>l?Hg}oy?xA4sN7LF`e(WooI#N?pvpDlee8qm;0@?iH-w%LI(3&itc<%ZLd558=
zBo#*{HeXi_IKi>cn6<ijdOzq*&#xMiN`^WYHcYDOuDa^46H);Vt6DYbWVIkltEMiA
z%q#WWqV13v_DtL7*Dlv1LZp>BRbN@Thhf0@ZA8G;AtM3U`~`Hqe%fB~WiFX7Us6cl
zXmfoDzwRgfcy!E$$}V@cX7x4R@HW#u!*%Ss_acq+ruwL?xpoU4fbL@5OH27+xzt94
z3rPJwE)4hi^`jzD@8X%p3GpcrGeJv*0Rh-Daq!_avG=}`BH~ct_T1-zUD;N9Y^&;A
zDt`G+bek`i{ov40ZLj^FQB4E3BHwiC6@;!*Hsgj4?2Ar6rH%}*=)G6oY+S!5{atx|
z{e*oEs>Q`p>HlQ;)@I*vV{XN>p#yEs2-oe5A6c&lp07PzEcYhaOURtBlyGo;>wc=*
zzEh|=?+;wn_#&QRAEmolJ%Ad!kPE+uRfWG4Ey(vh-uAPU1G~Z5!*!uX#l`GcH{bZ?
z*Wd|6@t!kGHpaat_Y(%$(q6EX%ud-T{F9GN^nbMgEg)`k8YYjLGPTeLf(SvI2`1fS
zT}b|a0~*|N3#c)TZ4UbG<2s|?{gD&wDjzfIWt-6{aOv{A+?UUuon104|7!D^jj@_i
z)B6#Y#tpB@S65dT!y3;?lj`nc@c<V^SO0gpZk-+i9wU|}G$<9d4n@EL>S}7XLi6bt
z2%EC~+u(qvAf;8#Y3Y1$A)rTo-*$CL^K`MZcUzTy-!%ZP0n+%5EEcy;r>qt}9rbtb
ze*L!LpaUQuWjASi$s_8&K`ySQd*_9WslSk-;B<SP*|FOBvm<tu>3S;(x{*dtv_T(6
zwyqhW)od3T@8{d~V&>RMiwFek4c0j&4{p&O|AjN5KlQp}+TcTA5RSIVIo-@GXv0|F
z`Q5rQ6+Y?tnY(AB?rR6%2crbPxwdZP`kw^|`Nc(LG6JiwfddSQtvX-SHKn+Cb=$E^
z0Rj7&>T59S<PFi!iE9p9Jop%Jf+$%;lm<ae86SHX^#){hj~+d~*W_j8W_Qq{2GH|6
zd)aK+;_cacU&k)~<09i^*CPh6Lm;>GOA((>6rTy`5oNq;`-c7*bsHkuIh~ueVpL<J
zVn{5$c4a3;eYGAmu;2(ap!CC4EwerdNm}i}#RdKfus4#T2+FmuWd8TyboW&?*pxbR
zY1QI}^xH4g?ob)_Lx_Op2?Ub>)j|I@@Q~8O4UT+jkhc%JtnUy{xYOo@E+cElmB${@
z0MPq&@QBZ!MNPc>j8nP#)N9WYO;7YD3*^5_ZP~cDZ-fGs>#nWGu6#TcHjA`7$)ipf
z7yL=7>bm7EpCKFN2c?=tPICTy>}Q3FMfYXFZ_JV2-@Umx(_+G5;Z5+HLTYx}iN#Ei
z-}f=))fT-mzm|+Fv~LZ0cWu#t(r>J+-b{-JWdd|DF3k0`S&{F<=92pQ&epuuf)^@k
z6^M|69>bjVZ*N#KA`ps!tTh)3q5Oo|#kEAol;2`hhET0lJSB=~CP@1McMSFNsn6A$
zE}vxQw9P-hc{`ZWx#|6mON=od@zdKXD)UIu_^)PB#%rILtKt?=P^!6W=;o(>7bKO9
z^)9T>Gku-AN{DL-p`$)5Iy*CTP&c~@TSX^p@}mAt_6qSRm;AOF&->`8^F6Wuj#cfJ
zGt{O$(H%mzK)wh)!?l&nH0xVr7i(XA7NFU_S$BOP-mA7c?%hi-i8HVX1r-IDPjYfH
zbeN-t*eIJ88QafkNB6v9;`}8S$@)hjDUary(;4V>FL&!=z~7&|)U<^dzKl%#?ti{k
z?<`2PjM9XCjUQj{l889l=l9op#TQ!k(Xz4hZp4*Eq=eQs>yYZXb>c4W?W20<UQq3h
zCZ`$OgqZzo*3m2V+Y8F44tYFw%&&oW>kjnqZj8g*_I<ea$Or6$&RF}^nY)bga?y+Y
zq>8v}vTr-;^%A-k?Nz|Dfri2Z$d^ch0428}A)U8DY5Mf{yS1V?4}$4wq6NeVK>!?n
z+{lqaw`EaL5jR8G<E`9esO@^!q*k?l7u{SrcXwV_JBOhc*sq%<Be0FxI?80l)Uam6
z*Ybs(@Abv)Jo_AgP7q$8{O0g_veH{mTg_&kd;f#AUfho=aH4!rj<?Fn6;rJz2rukQ
zd3o^7O_4baiCJtTnKY94)KiF0W>FAL!5P*^yOPv%9P$!P;?H{3uh*OWxi<eKGD$F9
zVF<R^)PmS`Cr(R?E?KtB<II`d!l9Ga>u}>rimr~|{@Blt<VipCDrl|f{#i*I9wH>1
zH+CQ4K3(&YEhb!Hdp?0p=8EE@(hw9hys`hNm{v91MxC|;FW!1GQg}0y)o#s=Trjl_
zn!tVUVF+m|?IQkYYs8p0vFjGDD|$`e*@AEJMUiJolyrSR%-HeiWQ*?#V`HDqb#b8o
zQ~PY}Ei`WoJ^!^BWKh?okaQ>XW3cz{ttP+wI&Yf4U_q~S%WwlsurJpx&k}rWjvWG3
zYB@XPiPd+XJg*Hp&B<C5&j-!}1?P{yjJod45;hUX;LbhHE~EMG$myXq6wxUoHwv7i
z%o-WAQtNws7K+L4bfJy7wy?EGt?_JD*POi#jg6jtFU49_u0@FE^r5He+r2_oxVaSQ
zd-2mlG|DXAUC_mD6Vr8Cf;<OVA}ZTm^>)FrfC3`kEIg2%Cuj0{gprR!(d|DdN)svQ
zv);PXQ=rsVS+=#eS0!x~QmynoL)Z6{yWnFt>Q-lRj80{jsq<#_?f#zRQL64VlP2#9
z)>E-9O($RXSy9)5?pC9?&hz#9@W~o!FJ9CH>?wLv2zSS@z=z|8F3{XG@4_odVQD>0
zc2QR4MXkD_Ujid0Y20%5Ow>laPwLWvj+(y)L@fGo@zv{`rIX_m5_tAE?gIe##x2Rl
zt8-35^iFIK1bA<<ypnO`#_aJ99Pp5HL*NYrKS(9LVn>)Kf&9mQL(s|W)_hMQf35dn
z(@8da$<MjP2TxjmGR5REG2v+AH7lff2xIyWV_*e5>JiIWI-cOd=hc@mC?dq7o%E=>
z^=sr%{Yu7}oPX%Pq5d+lXT*>aX-1vkC6aUm%+hP=5(ulX3(7Y?ChaqsK6~l0<wv_8
z!U&asI=n+;bN<9pW1n?my_NFgU%Auiuk*Jg-g_WDRSR`%%J=XlgNgE`A=E6O11aPE
z%L1lF>S|i(-fs@-&_2nfq_<n#XAOBy%9f-(QJn92Q(Yhrltx>*+c1fVjzEgmSL(5#
zN^g(Q;ZaE{+}f7`bS^?|pRz5f)koID?g8!QRhRn-UQHl?VcT;_!;kkJVup|t37f%P
z4^`NXO%eAF{Bmkm@;`U3nf-}fVHJFb+PhsiUNObbe5&W@Ex9vxeF>WO(UL0&jFSAR
zY}Bo&WbI(?Y7fgm6mVp~$l5|;u8e7Z$i1?^>M?b4_32r;wsE8bXMNrc8e?ns_DwH#
z$EWf&{m-TUO0mpkBFxi0`hLF!j7ZD+==CHZoQyYu7T)kT)xYSKDRQTSgUk6NeU3?d
zpNB=*sBsPGOnP<4b9$q~()sg`>c4!wvm?vpcd)X4{Pwkt;kI{2U4G&S>XWqZ*w-8+
zQDK8^u<%Oo`u+R2(;n9zd3u!L`htHuOugtB)P}>yje55*_zebHAt52pk7vAmDVhFD
zWBPPKk}0LWr|#0ktPsL9@MuQXQZ?exx!>P~7Pxkmg1N^?z}|Y%mEZnt!>(VtGmjfb
z*0Z4*M+^mq*k9O%`^>9VNDSl*62_RL)>t1{LgF2wN>9wl@y(!D&t~mH+l0VP@T{c4
zmXAgN;u`BXG3)@W&XM@eu|Fe`O^oZK&Yr5cr*-J&-8)B)Xp!^Tyz~>AYbjwXKYfCk
z-By@j^zBEKz^sLI;pj(C+dTIMSQR`a&xKdOXCkz<@5D~@A%;Ceb8H3Fb{yt_>u#@3
z7tEF9UO#<m17ZwVR8UX=Nzul&XsxXHb4aw_xAN~eAPw3XJ!J_Uq!%t<-umO*)W9oJ
zHV_bNYfdcH)r~N}f2-jn1ZLjXu&5G(Mp?0g=n{`4GfO))EwiY*U@-|@0m=B)x9l<f
zO?7QUCR$)MZ?}r3ruqHfxg+o7q36m?)YY7C_ASClPsRa{!aYfesYC7yeM_gd|NOS4
z@%4Qq#tOna+g10mKtB|jT<@>*Q1@KxPAA^D)V)Z-y#p^?IDmO^l1l^px^tGms85jq
z81nRaK&tw1RRz{wRRP`?FDgg6jWlM?V+@1}MrQw^lQG^d%8?&}S*4$!Op@QN$!2J-
zRDS-07-P&hXJ|)ze}P|lL4Ge(NeIUhUM&Iw+D!=y8g&4blhYt`oxnKb2=BsU-{4ZR
zz7O#nqgRx-aSG5jw>o{%jXW!i8Vs7iRc<brq*W7x74R2O^S6F5gTB(}MK4dXUQF=`
z%}MHU%~cbsp|sSVb&%?sky#c)XVc!8D`y;~x<eE`Lu6n8m@s)Vy`b$34by!gIo`aD
zMyczwF4aEW28*12N%`T0kZjEM&IR7=j%Z_4)B}}*z(EQJpFK^Iu`F=Dq#dld>TYDM
z$yYw~g!5ro{&6ebc}a0~Y!aI7vp&z!)Fh^C`=TNE7|oMCCV52_p5P5iK!TC$K+W6%
z%xZ`{At589Fm98#vpKzR6EBZph}1WmMP_D#?z3%7#<XEK-fX1P>mq&|Ch`ooA=jDb
zPsW;I9M6WG=TE{}Ojs=~pp0}zu*DLeD;i^2(vDa}bguj8UyUjkUrJ1#HS9)?E%3nT
zaVqjZo2&Sg|KSKZ5IXD3_%k`S<NWJ`cFf8&Bq#CR-v@rLDD%EH$L8FX#bz^ua+>Yg
zk8k}8(|h*mW7IZlN~)K$&>G*ebTb!JRaXAU-7lYRihEB!c(OBFOG`^?hmBA^sLPsQ
z7Ev(O<f-Mj7^#aR8V?2uSBO~&;mT>X-SH-SqJ-4t;4AMgd(zo~6;%ldo6`9=r-$Uu
zv@$e2UzC)a@suQE(aCHSrv?3<fYY}e&#bs8&?B&`XOAPYyzDsmUhVE>sQAk!aRhuG
z{<H!E1?LWGSs;Gz{c*Wj=blBo4FoiD_GxTak5bsO<>clK0?n9SyGgkQnZ@~k?>NmG
zKWkQz*MbE`nNR&~!B%gnA{-b$x^c-QRL>4w-;W%&ha5g>!uC-D{_TF^_UK__m#o*F
zku-I|thx%VS#@a*wx7$xjibU757skP2g-+Mfs(@ld_5#|nX(2hl1LTK?vH%%K(o;=
z82T0!HHQrDoA<Z}E5L*;`vKFGG()dcDp-F0CKMrDS18V8XAkG4OdK^zL181DJ8H($
zuC9B3v~1#}0JpQAB>5_q=NIm7+!yhw{7&bLvyb!g_CRhBHT4)4tgDNc-dl#K1SL+X
zE-x3B(?Yev*&NlqgMk=r;6fWtVfUB79Whv|emvg(;|HxS&y3p$)}&C<i*a@oLzlr1
z;0TwMk+FE$vSgRe-|RK)!E%tQ6?WfQ&^M61@>Vxam=G~5?V9r?;|Qq<heug|sm$gZ
zL$aqgNr{UyKWa$9LF25aGxs)K<KDiWt6#cu;8|z$IdcxC$ULXFwBV@kVVESLw2l+L
zk-Y_A=yzfGcrfZK9#<#nHrbcGSv#HkbXss<pPSnR4pUHy2}7gJDH^y{@X9Ua#z%hx
z(2#_A4Z-{Ry>|>{I9Z!E=nxTCd8!utn!SUA)Z%Y-g+IX+NWbxSO6Hq|m13G*Zj`d~
zS+R=p4Vjk|Q)wTrWHr2-=ZhZg=ZtbwUeE6)PSaonsEKDH$Qd!B6jGHWd%(Y*V4ocR
z*$7L#J+e<77q^{q0e1lvG&A=Km#kRv7BWXrOBc2@bB=ot9O&I_pW^r&!*@0X@<nyr
z#tF8>xbKaP{ZA45-s&F{I&Lk0#vuBFOkj1kqjW_^!Vchv@RgX35!{HGu}-bFNyGmv
zn^~37DX3VbK&utA>hPrnm3rIMwczD>;|M-~c*@G-Q>Y83-rn3QyeeT9--IQ>iuvvF
z=uj2J7CnU4+T|5?i!10@|J2+*8{LF&W9Id-+R-B2>1_FI@oD0WVbDfnUs+YoFiB4!
zL^}Toldqi3DOkA9ytnCUBK<vfV#a}v_#NRX*aC9{q?q<PR;iGv>p5qyjIUplsG<{*
zPI<<UbhGw&p<b=)9P_!X>=w1ymDUxRPia(LvAe)57PX<yJ_{@E-di8O$zEMLYpK(i
zWDq~mBlGWEuPi?^SmSEL6sZHVglPv49-BX8eXgBb^3UHbga`|#S=rTEr?m~P&UY1h
zroKsy@7W%I_Uu`1NaYip6SV`0P$aerpE5W1e=SY4Fs2?|P_;GhxFJ;W-4PRy7eGUQ
z89D?S6kz>=NuS#VGld9BDzWvg$wD6O^xiAAZ=1~S$$D9+tWcRav4P8F-r+3$S#`js
zqP<6D+`V!9*@2Qi?PzO5`bYYC%s>;XZmp=Spr8N-)^<s`vRt?~*uzRaB9{lJQyfWF
z)A!vg>6PdCuyY;*vMOJ1;yP?nl&>hiWA>G11gXo{uL!S(_N=>$@|p$q-^**{pp3sy
zK=zVnx$jk*+}}O8`D6U@Z}z?C&5Gtmq4LY{1TM1;J#_ZesZ*1tEG?GqvgN#>#w33E
zj5o2-A&0!;VfFB3=2O=bn`<z#dGuhq-)ZNMFHZa#tRIwws0s3E>Yb2;d-uM>ZC%Sg
zlM1bi9lUX4nvp8VDQ86XzT_A1><oq~=q-xa?%*Iy*TSb@bM}@8D4aNdb{qHhKCL}K
zQt}RN&F3a5BG<>r;PREr=*DusfBkUd-pdn1XU|T?-dC`GpYJM>;ZadvKYMM;Q2~B$
ztM-pVqg8fjpWJDAgn$XF?js+#2^O*8Qa))N3?0VkskFo2Uk}27ue2~NP4{V!0>e3T
z9&s}x6PUJccM5~=GPu@VxJ;3ZJZlRS-H@e`vvCSwf*hp1=fJbj34)4Vs7Too_d)Q4
za6@DKx-ZH)PhY*-0>sM}etEB}aX$s?>h88-Y$%9nd!WaRdaC?NCSmPBlXm;*(hExx
zoc3@#c=zyo5Xqtm4LEpIeoFM$MyCczg$QivI!gu0BA;nm2qu{)7JlX1J)RWkZ0dC@
zDkYNK0Ula+;^>HohHC(Co4WrzE#`u*f9>?O!h;OfuyPii6Z{7Gfnoe%g&f#l5|QUy
ztCT}cVv5L`unO?o_N2dr#I5b6^Nt;x8x=~>bTYaBt-d(0zV3;<FszJ5JK3>RJA&;m
zT92$fb+#nyJ}N7P+>C?iXS8rKc2}Kb)>Kk^AC8J)UJCHSUBW4W+RgXIOW%Tk6X=7G
zBL|bfx9?73Na8$-nIw7z8s{usNnv%LYl@QJ+okLI@fO0XXBTFE_XW(Wj=k1R{^Y4s
zf+Ii2%-hXd(cm*BEd*9@#;^sa%`4YJ0Vw5lnJ{e7@5~;NS7|UptS>14$$Y`d5&xRf
z2mv{QJwX+o1_pKhRYZhBuvOALZZ@>k?}@m-Iy$JLWu&BXfR@Vlw2-tnQ5G^j>>*nj
z?FcvH7wOKat+F%=G;+Kl0|d(xE{F26G6czdk?DTM%c`<+;NZa@si|3`mAAGZZ_MzW
z;KD1SYi%9xntg)ZZY(Z3Yu2nmRTzBrs?c@2ezw+To`o9MoDYoP3y<Oe#=G;|BqZJv
z>)CK}=hkA=`=j=DmbSYL%@)UbqnJEU6saRDN=r+_kqFf+ihv)hwk72NcSh{$>OKWF
z3-by$4`!jn!{!=0pZq%(qrg3&4MK;89ACkhnO#H6S{7Ravw!Lk7VQ%qn|uH{x%Xhf
zNGR1@WKVI)0p4%qvy$25lMfK$1O^z}Y6Jy1TS$yqER3mw?|?5i{2dVK`Jp%?EzOz?
zYeQZX#Vja+Fhz%@<BW41-v>V>oLs+to!rUM(Gf&6lBSt$4;r<>Xfe64=@?^9(ZK_^
z@&jaXfZwxdc;U;+>MEUqwED1;RO4_P3(G4V&6L+zS{9+rY3&Q*Y@m^Ca}5M9%YQPZ
zp(@qgN1Bq6hZ1FlV+`G0Qv<S6$mc3h?tVSTw)L(L*(P|vie&ljM>)PE3Z2b?ozO%=
z*DU3(n}>%BZ;>9Mz`)f2FBkFBCZbS2Uudzm1M6zqu#vw-BI{459SJzV1X}i$7^FcH
z3pV4tyYyr^y2xN7n`n&32xVJgljN-r+LbkEbhglh6x~cGOZURr2du2hgySTsC?P4y
zr4+@UT`+&X-y7SDBUDs`4@IgZczCm8amF-5T0?z@&@q@JP2oAsVeh1|V{Zr@uLLCO
z9YG~kr`AtQY%_{UK3VfG@96TH8bSUHZ6XjR2wxMj+@rn0g8KCBi!k8;S3>R9f2Wo~
zXR~esfy79<9pvzvioAku!3vTZt+5<sO4Q`Nr>3Uz3M+^Th+;AboRoBhA3yTO;E;mt
zeDQ#OX}K+P-hxw-{-n#`?&LB~6!!R{kdTfYGls@;x_VJn190FITD+NshlM$>wP;2^
zWSabI)Lto&3Vlciu;6pv-dNz4qPqu+(U6z_%(eubprMCGh1PH1*pWMEPh)%WTHX8A
zk+0K&f}(quu*dFh$yN(Sk%hj#;rn-_I$Lpp;d>uGF^n3jzx(rvF@rw38O2!c=hbua
z3y{V0PC5BPbKu2`HDA9Dk&$89o*>tv=l#v~QBq>}VH44%q%L&czrT4%PVY}yndQxE
z2LH@7G4gLTSiO9^%xsr$$9%lKL-xcK7Zhj|YaG8hz3Q3Vu|9`KC{LemR<G95+&uZt
zrQ8?m5fC^nGhe$v>7r84W~0^eF*=i5CaG@vUEBLBG>p^pNMi2$#1%1vEe`8-jHmqW
z;6Q$h&g;E_m2uj_r4?0cp{ORb8Fjx>iAYgUiFJ8oii@L@n+Fn2CW=n@7M>NMICg9$
zTCTED<o~~kGqpy61JCd3^N{ygP@CkmdN@HWxX6=M%z_Yk5zGFlO;VO7PN!=%Z;)NL
zRc>d_EM2sy|FCvuIfJd+#{75?)H1yv&1~}HwzMD4$jErR%I6(m3WJ}&h4-_R|5PsW
z=pR8U@u5pOQXSQ8P5oc)gkqr}C|$;u<=x57J|s!}ukGInv)_(;Vt&M=aM#nP3z_cM
zro)?CQWz&KqNiqGHa|%t6R)8oNlMMOZ|z!z#G|P1?U-9A^^l6VdgujAS1aCMYS|UT
z9Cr_ofx~2HI_-K`Bhoc}I7UkH+wJY&d{7pMZsKP|QgyTF+nN;~^l6jPehYuuTWQoN
ztq1-wxC!sdyXpIuYv>|9$sJY92B2?pMsANs$^zxUpOP8|MEQbf!}6J&wX9r=qCI`W
z`d$GvO4*>;bV1HJf3HcBuNLmg$WKxiaz$j7sGIg~R~MPM1Tz=(@~I2?!vdbfBHE@}
zRC6urnzYP$C^T@>?q8ZEG>^Ckn66eQo`FXkYZG*PIp(=T9z|@%*S}1y3yC6u>#J#?
zNd~zP;&%BCjWv%B0)NiPt?WLg?jRT)hD5I|6P-f^5q4~BY}qI-!)XV`_L(F5GotIk
z=ht*ow1yxDO_3<!@R>1JQuXKPc^>h)g#tsFwC_Xgr0_5VCQFx|NwVuwB31GAYrxsF
zYwFD&TonRZ>Ca>;7qHHo+xA=Pmt{VoKZGi8{nL!l>r|3Ru<q>{*YI{E>+iUhe#!Ia
zPgQXD$rq07BXZDp1()+3v#2Dy#*}k)FZ({4_s3pIu>E*P&#0t2vboqjQfnqP9`F?r
z33cD_<rpmR%;PFuL@ce{Y0>`k^$ybi??37HV|o94&6&x@`}@^;_J6+d?@IiCm*?e=
z<Jjm87g;oat|8y^zbsYx`g+kvD}ZBe?!mC~;3mX6>pifL%6@GjB2wbL;CqRHj~tuE
zq1<PJ$&2YzE=O)H>v~M6L>}G!<K!m%+0YZtf{QF)K|`DV>!9uQOfX3kzN$wE7H7g&
z=!suOtHlUQmsQstXRC>b)J~vYela0gau}771Ghk}(J*Y3wx3s7`L@XD;PS!3)~OMG
zb{$jVAVFO{+=tCRkaD!w?r>p8=%ow3V#3$|Y={3j1pkM|h`we~xkV1_;YXcS++27F
zT-n#$)I{kqW>y(aG_?PSZzr?oqRtH;%J->!;^^{!OI&FC*p5&!-FF!SUluQ(v}+r;
zmfgG1wWpmpU5}>!bz8i-tG=iZDXZBVDzgekNk?tur*94xLg~T7X~o6K=*DwT<pwY8
zXLk%&!h=SC_L9Q(NbmVH9mp&Hap0WtC*F!)LRSfwX*P5$3g)Ka!Y>;C=hSyU<Lhfe
zfKU-DD=C>H1neLY;ZhLM7hgHcI*y(!lKFd!t@(*H3;zDDs~8P)s8>oH3J@ZQsQlkw
z<cTFSDd@{th0mW?LA}OW?H7J%+}|Hk(ET@_$}pq2xcHPePkAuz@23q?c^;<v$KCnS
z-#<Ht#1WI@9QOg^F4YL@b@A_i-|j!{<*TNo;ll5n@%MK|lk`(37QMUe7b*Bto)z3d
zL=@t&wF9c&`c{!7W<OR5n+*w1BBp1AfH&vRhUzu<z}Rp)7-WRworuq<KaU%f0|}w7
zp5NA_XN<R}Ldl;{D7h>BMn0t)_wU;548Hr(yuW`t0aSp3OvYdm3kt52+bTTNnE(6`
zZ$>C1RLA)L`CtuyKA2+{nR6SJxjHv)3>E#e27kZ(fd~Zzc0!@%>1~^$_!~Dz{aMio
zBEUJMwMxEgZz=JYdmj=W;E~&0jY4B8Id(*jl5o}D{`=ci=WA;xV91@|`&bC)6GZ?1
zlH*i34_k9r-P*&)MJEtuE-Z%MpXmRn_mkJd;{gd^ewG#{@ZTTl{dqh~8PI3EoV|wH
z{Qc9Z&x@oU>6Sq<74QEZAlw3vUI@@APhK{P{A4<vSqq2lUmv5?MWo>rJ`QtHgnzZ@
z{Xp#RWpL0}AYk*e1-EjM1_@zO2G&O#>&;56|Cme)|M`C|ga0j;L2hYXT@gtFr3K|&
z6>Ko@>QdkY;JYPDn3Oq}rc{_bm<KfmUV*R4$lfN`Hq$`c@Nh|>+dE`$)^;lxfa`5%
zrHONM!E|Vilp>>UL0N4RWPitX5pnFLUXa*T<lsO{HZF|+56D1<Y%Q*Dg<)=K5$oAA
zlq!*u!l4Jikuax?-BO37Jt8N3)oQ=tbgF?)9j~KxcqwJAV~@4hC{$>FUAGwPP!&Q3
z?YRIAKs+BwxIy7JTg;T)j#vy_T(oy@#EAVwMZKie4|F(!b-KF?IGE5+l`yR`8Dl&^
z7rtokaiK;B6alPBOHK7_=e)H|5YAC^_Z~d}i4%z^*DuRpfg%XK_y6cX`$d@#aCl~1
zc#jw0Udu|0kg5VCSy4KM=k$od%7rfvU;*HV&`<?*#bB0<K&DM3x5(&7I0gKLDVroH
z)ug4V7snwsbQqck2S}4o@EZaxz~AEfA^>%MVmF3d2(p&nm_?Q@l#(JFeqh?b3^Qx(
zNlPOEG|;8xhD$_7AfpBn%dkE_VE4he2aV~2D7vYh&(U#_RMWLpNXRIu1D%@|0gV~y
zrdJ3*L4}?`cRdwPw{G3OS_~oRxDs?8t%ptn2jP(u0ABd{LDhwc_fYHX@F$e=${nh^
z?W)bU=g=ek`WD?YND+L%(C{6v5qtN|^~4rSO(C?%Uz}`n=En@BIs1WOUfpxYLJkI&
zaJY*KE-lhXa<CwNLk~byvaUbHk4=muK2Lllgp9{MhABwuEL|$}z*68-9=3wVQ$>S_
z(h`PdLKp!sWy%y4l}Lz+a3h(|g2@%u_o`!XHoAw$&8oZ};<SVc5)|O5J9`$?j>3zs
z<mJbs+QIgiegKcLff)R^?aybH5vm!#AeM%W%SBDuC<uGNG|;iLx0K}N<x%gT$pZbH
zI(0Xd6#>!$fdiKY%m71nHhim4dFH-aCK|Z|_d?tz>xLaQOglJol4s{kVO-EqTQFJc
zMn~oB&gHj{EA#{f1m~Pa6lF~cS`N_#<uGgkZa!=x6yYd-t^l;Y%FHxYuc-`l5$5_X
zQT|uN(O+<0UUL{sTSo3ZgQveyv)5woRy-WOmd<mqr2Vk+Kt43pMOiF4%_HyFfrf1*
zBgXIrdWC-Nk1KKvXA&Gqu;N>|P>>9P0xVj*_*TXfWM=|pproXvt{zYP2gl%@C$eX_
z!vN}I@NTcjhGbB315$pUmzPJoR8CGDa2lfgW%P(IS&{@iBlM+SQWqw8a*iY3nbYR7
zlSbxiPr<|nNlISZ^O{B_ju>t+r@vOOli&n#6H>ij1B;}4m0b>i1u0tvvA}~pM35pN
zxXA|!n|kc9w-?-pspjF`z<`(!jtWoEmSm2+pP0z3ghh@vIQ3KlK3<*Q3r#4b8medd
zx^bKU>lfM`P(4uD2$c7t9OiU?rR3(k0DFmxi=Vx8$(c6|sDPrk8U)>EHj8P1$de{a
z#>d}*8sHQ1g(ex1E)*I=)A>kQ{FdflX9rD~IrA6z5z;~q?`&akLq)}yaoKop&?7%{
z{|`eAh7_1!G{B3&c>#<Z{Ekiz<d(qQpkGh}&gh5$-Prwp%a?g4uit!wJduweggd(U
zEFU;Rd`7M6^ho;Y?pF5W=f4l8Jf630<f-nxwN-jrJsF=r>fhM43%beKU3Hb)r{bC}
zeOgXRw>D`&e{sY43&d4U4!O3dKB=zRDsbPd=0J_fGgg`EC+9^}ur4+>K8nd26~W2L
zoe-;KWo1Wwv!g9$@Q&a*rrPDceEoXj%$YMm#<Q~OQLv#a_l?;KO3sTzZo~ibX1yN>
zrtO|FQ&!?UhASCgWGBDaY`Af=y<3}_MsCP``*u3aE5}JmS=q>d=|61`T)UhffjRwM
zmlq=X<<RZAix<adb|YdN{g7sr7k!lO{Ng}|F!aN6riQ=UX|i^0DY8mIA{{4BI2hye
z8cngM&J=`*cqm3kq1>EXtrD^#A|tw;CwTel)!q#7+5Yy0ToF3I9wKB-bx2A|TAk#S
zwzN`oTUbQI)3mhM<`eW!Mm>RRFHkl%G~{h<kFv70O`vxHnCN$siG{^}`W1M)4JN}!
zjy#;Sb;Sy=^_#eP*o_6UVq#)nj$!X{B6*USk&zK0t(u7&?|zMvaAhMR9ilhosgW7b
zDfi-nk%+6^PRgPnd!Fpdhb(%yf0b<<a}sfeQ+7D)o02lGteOvM9-*@O;U&9ZVP_Qh
zgoISk(1WF9i194=@*9B%Z1Zo9aH*&=bn&`!u4?tgPL=nB1N>L)jUL{)vx`f1-fkUk
z0pTczhbviW<mBYAsGbjAaP5k!dgx`RfD?=N-rMrUvtb^IcKk#2@?^iYTxs6k-sUxM
zGX{=BL`rsks;YYK@UO_(US;+{gHJx~;@k()?umnha04&dy!?#grKqTJs%uh5;niO4
z{*R_d2Gm+aboA%q`3?@P&{}&le8W=CV}8NI?(h!N(G)zDYo##YI(_;ynv*3ecRBN^
zSD`f)yeww>T^3~oDl(g!c?r(#lk$`ZO@f>Lp3CtM9ys&4G#yG@p9;-jckY}wRasZP
zY-PDU&AC#r(Umjijr2@N_(6IbBJJ(#n};>IMwva9LDTKaj_T#$w_^A9TCZgR6(^r!
zgXn6vGgbbQgb?WI57qN_!9DL$mjOmDUTZ%Y8f)UAzZFD#g>o{H^7n;7va$=bwHJ>Y
zD=&YY9BcI)&84Fy0RPU~6=Dk^8q-b0`HZ+nH|x2Yeb0AmR6iJVC?;k?qZlKw$2Dje
z8{bMVN9%>;N~L&jEz!RstXE!e6ZzUdKC3M(UQM<Lb9Rt+OT#er+qZA2lPb<Ul^rw)
zX6ClZ1N-vhI-_l~ulI?~XFG`NG8<_0-7zbZ56oOPgskJ_Dp|&Es&<m`-GBeNfn7Gp
zWJr9O{nV*byLazy-d$<T(6Nt)JwY56^JoXtxgq@LnWsv5OjcFJ3fnU{lXRfW^-e~A
zkJ~qImb{nFOPvINbT;_2RV*E)=!+g}{xmlnG}d44)KfUjWjf_^U8yJSjaDsOScyKO
z>k9@LTkGqK+rO!l@)+PcpS<(4SrROAN<<ftz#*ySb5^o26>~KEme@VY=`(TCq+7@D
zDi~-A{#m!~o?PoUacx(VCGkVOt-S&Rv()MjjZCA#ZNS)7=VuI^21`|TiloX*EOn02
z)~)07f|h#SpCQB1+xz+C=&7%Du?=|rdW{V{!X4+w>b;^H#y8ptQ|c+^HA6tW-}GKS
zdZ>x9ae7XcJ!H*}fw~LbpKh|VJ8}GYhQ$UwJ+Ul>bE!s-W%e*hpWlne?*9nH`YHOp
zsR}$lfNp8HUeu-awhBYXo<p=3ZM%&z>5WZkfonC-L>MHSs?c5S+F&d4$fif5y1w*F
zVZy>;V#vg<ea@5rhQ&|X%y=8pO!+fE$A4sJ?oF3BxD*{-2bq+1Zp4fkKk3{T+Wb(&
zkejr$wyLSA(PQ_TIKg)dy*!bT?n%bXC~!=X-7!2f(#+i4nO5HCm8@{=<S#WfJB`Yr
z9BNa!S$8<tt4*0Q`tA{^Jn?<s`c0mEo7fi`F<|Q^W~8uLIEZA0g6SS9owTIn&D+0V
zI=Jda<Q{#Ml5&-g+;y@6JcGb%f|PSj3*|oUyt=x&3m4vilterOm5k=cCI&($31-7T
zAUZ1Q7YgiWW<5#;>i~-C*lYs2U?T+`iYBL~s)|^ypswy05XEcnNZ#q+&=QKoYTY$c
zGqWFUZ3?$Ex^+9kBlRDZ7MQl^`f*_zESXEK;Knn-?2b-rYglhknKZS+@WD5Fa)Zd_
z?4Be3P<-RC9t~+VMM*RAWP?5*kC85#wH57#c@85@n4Y|pLI7_k8!M|xbTD2%pbbwd
zpL;ZU|3}UP;~6$k#79LH5?!gCceL6>hKJv}dv}BHFgdyD{D4*Y1OXu+CdRsPpHG=_
z+xcrb9qDt-Yq&rShq5SE`|iM|O1xkflG1Ldr1+Svn#|}4`#e;iv6JibV~?1YB15?O
z-3D6S?bNA!e^R83@&5U<&UpFy-h(vd5o@ZetE;LUke+5|3;0ETzMmx-$wE@t32|RN
z{QZT#pZry1UTbYGPu(JC62+tjdLIfQmPr`Yw!DA$;lqqEV~$m>x*2T2-JAN`4|bw<
zrTn{`9M^R(jR--j99-9L8)jo?_abL#)YYqp3r}vMx-l;nTD|`G45&q!#9ZcK`Sgv`
z6z2}EL0mgY?ftCrvuAgbp>D;4>h9g$+`y2+PrQZ61+^*nf6R4DE2Cl)>?qhLY!Y5D
zS=uML9m~sZ+YK8lfwfNZGV`R*@7=r2u{ik6K1k){(?dIk59P&S(N;RW8)m-8?V<?k
zJvC0~1<neHPiSy7H#L<}t>jb`oIA@265fJeGBLl+^w|IpbW!Sw_C80BbTj#~7XER?
z_o*HfHn(3~lmB(?+=QX`7&2aF2F-fiX4LK{@K>K~f8M62V)t(*mTcLwW%q_MgF2Va
zUkf|Vq-waJBz&rtH`ac8?5R_tyi)C@mmNQ{QBFRdyWnNO4Bz;1yVZ|JOx}d>f+H&9
z&tQ5a1J7C)<Tt8EcNS0tibq66L|hCAc$Jwfj8NEcs4<>R|Ka`n=#{sR+gy4Gg*{P}
zEPAYVM4DA2-bfsB*>=XLAgy2alFgP8?Yt#mz3up+zUR(m!@+AGh(0S!QkB(kaB#T4
z;p0Px8tI?Iwqv1|dPs51tf!72$=^;!40!TIq#=#SdcONFipV+QRv+gd&hMe`jY&AS
z`bjr7S6baW<CQD9L;XoaGGo~7=p#!n0u6LB-8eiVLbi%7gk*-S+e`Up*bo7R%y>gB
z9WgCdj~Xm=ZrK;=4B0KkCoPN#E;s||sT4eYx==-WvP7RgeQ1+?th4#e)$XykN5^Ih
z{-~};a!Io<Mkp6>I~2HHxNre^erm^NWCFYqra_7OtsY+HDdjrfJ+_okSMueH&Pxq+
z476fuTkY5}Depm#`Kgyv2;Sy96$VO5(w}xZU=}L%(m~c@15NR(1S5dSXiD7(SNvE<
zcH$Zl)t)T>Op0SvmfB93x3Y(08>nd}{|Lz~TXg+1I-V95)O~*R=Y6$T#Dr3)qh47_
z=>h}Ee-4|duI@-(&}4jfJi%||b;cm+r>QXtHLHIaxWVwACcCsoOFVxUeMMeOX5vA0
z=X~;w`{{RIWMsgts@T8bxsPr8K`b{f)uH&>eTnP<{L_1E{3aWldld9g_{UG2fS1Pt
zcB0vNA-#*gx}waUwPBXT+8ZdhJsZ0r8K1XcLH8MB?Q1nWbtDH4M6I!9_YXKu8AX1>
z29F}rY{9qgTQva4$gSfiPCV;3UcwdiqfUqFPN}rrh@>u;e!b5W*jN6OsoE;3RVZ92
z9o_FPAUhoy{Y?E9e9yXRtrD*DscO7>Z7)CvDvKcTnin&vkzB03QA|QRB<m`wcAo&G
zU^}hyHiz^S6T7B8`P%j&kD)lp(8LD{945F)?XEt3PLnRbF|chMZJ>4Q_HEQtDN6xi
z`ZNwZI&-@vkc;aWjH+BC4&@^lSs%F}Abs*EK*WQF<dh-@ho>GNDzVu0AH}Ow)~UOG
zyUNG=?C{p!MZ-n(0&*~_q%}9}?CfsOaIoDd$U;Vu++bnM$)~c~0Wd)MIdS6aw{Im@
z2rhx8IiFLcmaOww<{J?3wZ1+pv*3zWkLV5aQ~%v6FvuMTPWEXL(jfO9_6wY`fOwug
z<6!=&aN<_#)%=set&LxZ;xxn1WyR_VdW9RyhAAs6KaV$?XKXXLfMm7*?PEue-dr_&
z@yWC@j7mm)z@Tr8YAPnvnYU*{DC?*!X&djoXxFY5803Pgrhp#t-^7CNm}x$LcNXJY
zD;*0J=>boCYp1R@pS|Gneu-hjVp@)e#GO<%9UT%Hs`J0@JwAqJo`BYe-jd%8P?6d~
zhSe0A7I!{k%E-%wDYtuR|AVqPiz<ySuTz>K&3En(0L}4G)wwj^)>g@0Zi$tzQU5*q
z^>n`o`yiDI16BB$AK9=P7=?RsmWNH>b<*^@UJtjxhO3wacmHf?U?AW(`1%E0y7Vb(
zVDqyb$C5Fg+$&nTdhznI--b?`BoHGG#ts0_?uJ*26`J~PB#9}Lo3<@l)N6+4kkF_%
zfn_#_?8Ki_Cx*q#gBWYf6J{Iz1A0(V-5jD5Iae`%NZ9DML5kbQRVXbvvxE&lAe)F+
z<wEU#fYdW(nwb_RIs)Xe`v;Y`>yT~a9E&>+lh%x$I!AfJgxR}mWo85vS$nH@m~Y?S
zfcC}J!KZ3XW4Ud0iavIi^kUq6$+p6+!0^!G{@We;`PS!U2BrQ|lz5=q9r<QcLxapL
zXGbTe>8e_%FZcMfg-@z3>1CCuaaZn_V(xzKr91cTjU7Mot~*4-8lo+a(0}JL3#>IB
zUGAc9g$mtviut-g#gh>e#&rh0H`m?k^f@EJ6+xkRc9m8pYn_R2L7TK)Uz5bU-~|%<
z6uF@#l7nJnRrmS)+sn!~hx={O^oN@^Zq#E4R*9L!q#-H?y8i6mJ22@QJ*a}&YmoLW
zqk}TKAw37Lv2^`uFxt^qLW%(a*RBEaT;hU3{t<iob`5UiCg=2A<!1JKN7y@l-rL)L
z+{pCe0^61wrr&uK_nLvf>CBNb;aO%a{B!*D&4bezWP<^lX`1(yb>9!4$2YcnmYKQk
z{l%kfiyqt`tXnnC05!$pm**~COvFYQgAI@YUX;@H=W52jha;n+xHqCA-^S0|2aF?s
zCVuQD>B_P)tF2ozP5B58R0z?1p6|MOe%_kUd7~vx51V~C-xVh72kB!~UQ8gB$o1>;
z4(q!wZOF*c^Ny&wbo5d5>+s@0p}S`J&((ZzL6boi%x(~J5l~x0+YxF`LqpTf=>-@C
z&rVD5sBD$U`=|m1vFq<+fpk{du3+N&7-y%&A>|il+#;vcw6h;4r9L3!pJl5p@LsXH
z{g}a^KRz9C-!wQ(liZd+<Vcgi+__!$U-~UqU6n31_gog|3J~4Z?d0jg(T(a>HaXxF
zjZ;3APB%q#XFEwkJ@N4N)KSV3u5|nf<DP)ONIIyzl`Kj;E2>5FXyXp6+4f&8z=X?j
zR4AjaHE@t6#vT(KkRyK@ELkFqofp5g+}g8V^HH3QOh^uIFw?92h7f^&?~w^~Lp4TY
zkaxTS`YbLprD(>3{wX^}M6O`=aR$WS{2Bv6cz|eEH9NEI+b726tYFpRFQ1^^b~TuC
zYW`K=!P1N08X8LE=5GQttgb#<vzb`~QftsT_P1{h3JRj2m2h*|wM)`+<{?0gFGu1P
z>BF)E9Y|fkqS|;Q=OQuzI_+<BjwaI50{t~>hN~t?2S1VgrjjtyGxX-o23o>WJGy#o
zc63y?cZ@;0U>~=AFokNKp`RE*h$~^;jl5&O6mzI4U8^|G1n}YG5k~4mW9C-@rGN{B
z+z>W#rc+zP2vvNpEeLAy8xcIJtsEtV>}}Gr$i29hg-0^;)cGMrrih5FVB2)_=GpjX
z@7}d={Ug^NbEaZ-FqhJ(kt0vK0l?*33@82SH)Q(!&1VcL3?2Jc+HBnl^k}J;`|=3Q
zQOK8+?Y%it`hoY~uis>AE0N9awiwVD&n4HmC-K}9s1nyI=53SquD`+U<TvxCAqbC9
zO&a5dnepO9K^16B#rqRt<gU=iGw{(r`*PBVq2{DaRvQ>Q1=>C$&FB1i#I*efqP;K1
zLDzHRrcL+x93#f)etfX^T)@V4>&icWmcMzFG!8SXJ=Cu$)50Z*+Nr~iczV78Q^`+0
zK`#VlVxTF|mCU-6_k=5EB^}YDGgEEXub;1}+*fAe8ek`-Fta7GckZ+_HfoSP=gn>s
zGMKJ)92c=a$3+Z<Ax+>Zad&vJ<i_6#r3Coeva+{m!NKvlh`}W8+2sj_9BR|-AHycO
zR#7|BxAPpRCn_>>w3;9Z+qh97R?$ai^mJ5`4PAbY!4pwp@L<hdU-T*QxO?I@Wq$?&
zp`&fQeSQ;#4>Z(rmkkVap-8z6YILqJA_AK^s%D?Ctz8xG-3~t*5MaTZ<^)<8yRKW>
z`0~^6<@);gQFJNG^%ycEeRlMNurPu2g3MrRkt@7N9-+Z_kk|nr=2^eLs%?B-0!qjA
z*;QBmfVHR2I~EV$P|8y?Q&ZM`rJ0%2-o3|8oOtn0Ydhzcw$0iWK3FSqZkq*aYH{Q`
zcnGv1xuRSjaVEedxT&cL09Vhv1-U7PrU?lN7ySJARsSi*aLs%F{(@hi<E$`ZR3WeD
znS(aBO;3C}=^F+@y?giW>C>;EzsbMCo=Faio40CJA|*C^5Rh4zQDScXkwWwSBojkJ
zZYYUmRlKf$Qf0G6)=Lc;V(}r_T0~zE#VLOuoRZ^Xiz+a+gA&E+6Sw!iA1z^rb7+r{
z|E4PGIv+I4=;n4;q<a$=_wCbX%x7m1|D(MIFc3&6khg64g%pSu|K{^Ahdz%6;7PqS
zcIl-c8#@T_^(UuKWz^z`<P&rfh?^8AI4I{SK}$XOxkyFYjoCc+<wHXzPn$MDMP;;)
z_^@HgNLQ)VWtk<kaiffkj46*5AI~U;9}ps7b%smJj8uS9Ej9q9p4~rE!WE=IC(`J^
zf!_>cu-dYPK`2{rL@X_BWQ^S9UpFyj(0=~@{d>ya6?=ljsixQGU8>2m1ezrp2;*_6
zaK3y2Ib3CGiW)vGJG+^T+$*+Hfk<ub`_kbWHu&A^j1Iu3fMAQ(orCb&{bBs_uL0}x
zpXiHr<oSpnkPdlp;MawKz_u3x+-+p!{c+!sGZE#TtdiseV7T;JFXZ`>6Z|t@yl}Jg
z?fS?!6)Uv+HvQUcdsSy39GLSRFs0_pm+gJTjg?ca8Yu?|4x!I>^`On~TGpMQU3m3D
z7pxzBdAh?X^$ww{$1YTD@R5rb2Q@Ad+Qzs%rB|fz5BNLv*qw0V#8V8Zstly{ah_9f
zxbv7AKziUn-9jh)WYEw896TKwMQ2&H;Hzmiua5?nKme=-o<l4w3Q5foB(JRVqDldX
z-cQ+(ru1wRN^kBt!imkbh_Eo26u$Ca@0RktbV-xOJ*yKN4^i-qU2<Wx{1V&7m6z5V
z*0wiSVgMRi*dABYHP%$<dy76E!B(9AMZmizMTCd5uT4QD?GNtM5Dz?HEi@rulOyqT
zvIJC!l$u<pgj*W1__T+I&9!0l-}BiRjQ9Mx2IkN^l+$xo8nqn<Cg`ln9=lsm(w`8M
zoo8M_oCOi>NNdXaG-nUYFp|yXipe9#2<D};yug1y%e`>_LI6b5KETi4MxxhM2w*lZ
z1j^zzC5NXR@BDG1-IWv-?f2}d?4_00cup_MlnEKIm5*N1oFIlc`^}F~OAugQ!gWD%
zv2uhkJ-12REsZ{kI%*xdDo0k(I`;a>6Rnk5p&Z;ZXU=q8<**z?o?A)i6MZV}LSJxt
zeEoc{H6MBKd^hd@9D%*@HHm{~x<s_Wzrga0(+=+30{l-;x*dI5PW6p__Z;BZT3VS(
z0_TkwIdb0hKw8)nIr`ORFM?s%cD$Oq2oeJOq`}eqpIJj4&%3(0y}g}D8;-v_H1e*y
zR#7VRkpiopwE(;b)^2BA^b&aaXpaKy(k)L_FPr3GzG+h!mijzT$Y*zEAcDNe%~-Z_
z+aAL(@-S9f80mmY$S!d_nR+`oHY54ILr$S0|3g~NDP3k@6{`-3Pu)d`VvT9MkhC|R
z-#&lg!Ve!l2;S8clq)6=l#+stevzANoU7lng4aX6$*nH8>@v6(=1dKDHsH7&z}Q3`
zou11^aM3fU!I5?V=hjcq?vlHwEu;7a^&K=w`N81ml0uS+K4M}g-P$T`&rUzrZ54er
zQBhw7+l7S{EHv{iz{zHF!K|D%cK)J8&+++VkmsZ3;{F2%PQ(7*U+(E5LlcvIPENm>
zKf~IY?}d-!Fq0u*E4=XK*>mTf#BGIP0*WGS|5Q_>{NU@_dGk`J|Dfz&4pmz2Ew}^N
z*)g4|Z*+T575j`n9;<EJI?K{}k2n_5Hrn7jU8&!zvTX>g=E*x_O`VCy;L^@NT3p)<
zu%RFLRwJoSXVIc|T($raCqjcJe3N6x%^D~(o7I*aDecDaRpQqe1A}bV2?sAh3=bml
z?%f@74-k2hH~h*_nR9;Gz5ojqH1OKCbh!>z&kYG|CFmc(^jYC8*4DyCkBM2$R2c@d
zUA|1}Ury`Z!Gj`~9goi8n!}KO7rY$Nv4@xppmT3ZJT5ybwj|hKE>vU+p6?%zQ&o*6
zc@`SX)a|YKIr=rO1_z(>^z`)d+Q9>ZzQLm@nl8I4^uJ_Vf0YvaD=AG#T~y!3#l;Ey
zI1uqolU}`h(;wK~zHjc;t0UVl=yC504HE#hT&?x!5!PQYL$*K}zbKYB#fBogaly=h
zdyZ*at~*_omM>c5O5>cMW`Pkxkqz`+T~h-laqaqb?F9=0X1f$&4OQoO$5f>V|9)aF
z%@u;BG4~H$Bo&pFE<ain8LUm5$85pt5~dWX64$1_6oViU6%~eev4u!ga&XgRMYOb@
z2=XfWbEKu|v4LqU<-_vu{BZ5ig19k?idX?0U_Mj;3^<EFc*qbYpr3YkXDX^6PK+$R
zpI;j=J&7b56t44UhViF3m8;ov*kpdIE37FkRb>YQGms_yW$wANnH1$^o3GoUD7tBQ
zq!rjKU)-SFBdBH2mJmWY$%J{Ryf?)AYO@}D0mo}YOU=42ZfvU1UA5}N`)HJKBI23{
zI^y|_?y1xzB%op4+Pf&|LK2e3<ZszQLkXh^mMa@|6*=oo#|{C@Hc`4^%$z?1%nwfR
z=N%>`&7AQIu()UXAteDo$e%k)K|!b0o2xI>LP#P>!=E3(#nwvv6Du7aBjzl4Phea%
z*ZgQ8Bhs9Y{YlbA+ByUg3}K$@Saa_bHMKpp`RJE2Ee%(!C@U{FSQmFB<{58|TdVNq
zTCK~P<ZWz3t}dy4Z%B>WY;N<%+D5LOC8(PZW_h@}J}0}wXX+aPYxU~lXV0ENE&{ck
z_wiw8Fr->={vt+^k;VTYyprr1ncK9DWUM1IY_=3X{KW_UOKDRm@$RJD(wLw|hNg8~
zjA7l<d4?~9uEAL|ry;K!2*da14||jJ!RGU&XEzg0lx?bJIHqkn0+wPp*PI-q>v4eP
zG=Z6Dv9cneFJ7-k@GVlJ+{6vZbd)vnJwhGW#lqCNEqQ!L&~AWnzM~;deS68l;NW#w
z7Hu*4=b<fW3ogZeXpEe~yNWzA=`A3>A|emBgR8PO1*FsLNuqay(TmJX=J@6l+}OK+
z;<-p%um61FEU;5*pLbkR2yYl{@EENJ#mMSEUn<g-{5qz@L*vu1bm9aA!3_Lt??#BN
zzD_LEJ#Yej#vO2rDO1jO^aTNc$we<q!hoCs;D~JmWCVzAmzcq)i|hBFKf4SvgxNlP
z2@kUyK?N=G3$u0k9{)8R|7k1WUvmE3JNyG*a~%HemOuZZ50UuiAB6ugglzkN|K)#v
zguCPK*Z;*U_Ue7Kh_e*D2D%QrL}{LP_z_4-X1A7I<syiPXF9ovnouPZA>%B1x0!K{
z!Y9J>vti%u9>@Q8&c8tXgJvJlv|o;&;VEU{0{v{-tTI|hiq5I}docg3<Nw9W`2Wu#
z5?+>wb1b(~`PZ-G+8@xXOf%a1sH{9DK4wxhsYn|rLiVB?jZ9POa?p^n^npv~tX3P0
zAjss3xqZzdB@QaloWGk8<dp@az7hE2Yi_08PI4X~8n?^00PG3nxyJiT+`dDWLy5gL
z;F4DblXxHA@1T;9$g70}1(gs9y|O@^;#T>`ye0tBX`}TkEI3%0!1+F+uk*ho;`o9~
zq6tGf=Hp{*WMqv^$lA4U1iy}7KkYVcdcOQK*DEr#+0Jda+!^cZuL0UYNLW%b5o(Hz
zCB&tBJ8S@$&_DnF%Z2sVr4e}J0`6V(Cu(Tyrw=nc(w!`voVZGn--iS;K&eLG7b{f3
zd4t3XVm*rq8iN%{1Q;*_(n>n^K6<!<lB%j_!roAD#lj5dEet3asA9^eogX6j6=h~d
zOGo^pHd<NvMRD<LbbY|HwA-mj50aC+yD2;U@nfOUJKVX~dWvguq-1L>Gut7Ux$KOL
zdmMa!*m1D)h%eikk#noFgzuyJ&G{<Yw=W6iq?r}SW49rkQ56Q{gq|3dz;cpt_}dL2
z`yfkqPv!3X+1B>Va<zCWECeX#S%-1Dx-Y0!n<~RXL-Qf)csd5;=Ic@|(f<D(3mNR3
zZ>@cDyz4!T3tK_xk?M;3+Jb;3BnUlvREPsj*DQIKl_mH}f!m#2ZGII%20xBstiXhy
zQPR}V*h?LpdN2hMa}id(WoG>Mn&RZyzn*_0+q|S^Cur=~zyC4tIz<j$_^dgJw3_M@
zBoK1bwrWB<Vno^NRj)3eZX@z^nh0-)DZ&NimorI+aGlvLSTs<`-=A7qQ}Yw#d6EQa
zHl4nj1vR~!go!M)4{Obv#}0CpCEH_K+ZSNkXL6^a!ovwJa^+)3s~+fRLy@{YiaBEM
z*-i#7pFe$qxc|(j_SsY;5;#Wn;3R6O=8l)b_=jw%{G@%;;_eg6@>)JuUuR1m=&WSR
z#90in2Tyg-XI78-hn)5Va=s(KXNN{dQz^ahE+FG<3yipaU0PQ5c^uQ`Y4v1t+~4p}
z{3Nt9YJ&BN>2C8!dOjlyO3o)ShOqCcnDO>)6Ka_iHC)gw_)&@~-eierPS<~1=UKpj
z4Lv=*f9leC$1e;WQFQ^V++Un{;Mer51jMK|p9IHesh`n11R^xgMBiEsiZ3_JVC<oV
z;RvNV5>2YW>?hoB!d+&r_Xi^%UcriK{eW_K9Mzeo<W2@W%PDl86P<dHshJoV*;iEB
zKE!VH93?_@9N5P*OKUWEYRv&9E_|Qh$X;^aaI{1KqmHjPF=6P5(L0WHdmeRUP-WiR
zw;Ydioe@fp-o9N?ec4m2XSX2rmz5$76$<0VCH-`v+s-y|y1t#@`H9Blje)ku&_V_x
ztbEbmEe+h($TBHtyc6C<oc;Un2dgqWigU=Ur-Ngck+0LHOxTgX_Q=aig7J($dP{i=
z4b5nnkL?M`dj8xqtOLIRVf3Hn1W%pO4CR@RDyvEn1s;93-Z6Kw3IV-;wp0m|7L-DW
zg8&V}_&t6K$L7Spea|qRIzAxz&vSBGVM&&+>_dbaeZT1;(1!IOqpR=U8CHM3g5L6O
zM7OvN<fhaFE$6%UmJW@0(Q&5@e<Ku4N->r}gN0GZ`@U~Tt}h*teN3403deCa$+EMU
zgR0Yc`|4FI4$ThVe;e)#+*XP1le)KGAWc>;Cm&{$7{_btFQoQi{&5rNL5hi@f`Sa@
z7BFTe$APMtV6yDGEpBwfCr;GAZY0^ls?s@d^>YSARu!CMg{8%kC0n2xTFz6-^61Lu
zF<DuZtwQ5ZVBm!I+Y|##>jb!^J}G1!EzHf??)Rub5r8D`ZB9o-@5OntrQG{6qu+4c
zzlM%F6%`yj*`eC7eX%e-3Olh$)!j2ZQObZD^RaXF+tI%fo}dzT_?CSBOyM?vnSgy?
z82D(Qke42uAT-va>iD8a9rA8n_+TFts@Ipi6<zH`!~(Cm>={Z6JX!$^EG!duZ=u1f
z_IV#nCz1T>sJza~@=!M}q-z)+=KuvPk9Ke4o5KN*KYRf1G`(CuJdqI{*m2y>rW8PK
zgwE(L`wN5DK*pK=1>$Zk?XQo!8ejlv^a3Rf@Q)?0I%PkB=yg(u7+rgrwUldy5QJJ`
zw9B`uY(aHL&-d^RR43oQe-E6oXT0~VHIKVSm-g;*@G%}1K|u#EmsM9QeUF?$qK#f>
zg?7>$jlaU<f+~hhS=?Onm^uzYG8_3Dfs)ql9O{7FCSgwGno(QP$fVyqLZQjCFTbXD
zFri<w1~>Zh{G>aW+lO5?-yuu^0i5D!$CxGsX@gEL^`CTXcE3E8e;^7{<xN;38AN%X
znvp~EGXjHRX0|vB@Z2Ap)MGQ>HiPx1jBmiK@Y>WVSWhrrq_wW_;u|f<eL;cmUZ<pX
zg?EnOL9K=~{RhqrvMVzTXRSXns6I=_R*ctT{Giy1LXy6Kh1&vv^7&kgoE}uU37ZF5
z@N=6%gDw)GmR0W>NkD|gV!G*JozWnI{sia!qYXq#ZedieogtvSird+P0JL{*-CA)?
zT>liR^xKJHoWw{NXk04XpJyi}C2=$3SK7!n@2%2F*~xui@Zd7dyleg`;hrK;-gve|
zWzLWjj)M<=0iwll1_cIK24o@XPC<t(J-CBQn#GsKD#XtZ4cvR89VGB+*~!{<_uL#7
zGOw2sn*kwIstCa$#@1L_0S5aLn|lpgckyaT-({d@qN1lxo@{HWGFZFz3cw`eSmDP}
znxZ;Xva_Zv1IhUXjDxm@1_{6zhF_-^6x`=erS!&gueX>OiPr&A8CZNi*|B2_fG+?a
z4V_!4;>g9abapv77~RarGr}1UME4m}7Z@5D3D#YsM~^1WAu<9F@~O17wT+F9i8OT>
z$MdTWUu*+yVy}|GUA%C?cFmgCu!iJm->Y+UOR9kqr%#{G@8Ew~QSw80Csv3fGDOo+
z5R0$hGTBJW?C$18fnIv@yu%4IR9c~0mOoZ^W#xp)A~S~+fm6ur45apM6VyGM;nm&u
zIZ?6lYfj6k9<~S)EwiUQUb}38m7U!n35k@}2s_{%(h*qB@z**4Oc}c2@gp1cBaxW+
z;}smvr*|(Z!vM3>smz}Bm=1LE={8Cx508F{Ve&kUU<C`)2d%hZ<nsCRQ#txQ`p6mH
zb}({ZF<#5TfC2+uJ^K^eM0#yVbX8PTkYrs72sq~Mw-lxeE8P&FMHT>LlJr(EC@enk
zGm72vNP_oHT;gx|T)u3%eEHhVn>SlqkMU2Nmm7ADKZy4V@m;|Ayu7T|tXZ>V%h&2^
z28>Vh7hR;MhwX)%yE{ew1O<hOQ>Q{M-3OK8EqFUVmJ*u(+y|@!$>SwUD&CHa+)Nx|
zbz`iIehvd<z@iLPi#r2!i7VyD58=fE=hA*5DBH_^3JwpfsH>YBwLdCiR0jfz2$#`4
z+S1rqnvg3Y(Zrv@bM)}xu4KlHIboNO&1wuCfBx|P?gL%t^W>LQ*Zr%S%o|605D}ql
z|J(dHw;30TfeXphK1WA=n{ECr)j2O1`8I2~HYSg_P7!P?6#A4j>%VAT^vJjDw%_Vj
z+O^(^3Ff<2v27$*sk!2OD>DPbko~<MoEQrCf(qoe8y1C+3BZo0=yKFFGKAsJ+CwoJ
zvcnQ((!-6!b|H5}Ng5~ULBFlzKlJY0x}Jnw*a5*GB!zw;j`Kou-<iqbA7_}WNF&hh
zBn-fceL8gna2POu;<RbL5l8u(ulG1uTUTHVwD>`9vgLM{&R=7<cd9V0bJg0N3G4r>
z1z2^_Gz4rhiuqz-E;`E>+oP4x9s+FEwkOhW-vOfpq<I%(eVm&k(CZ4WE$BfsKfHbO
zrm?AsF(KoYX$UNIXA0YAoU|KD4JF~KrQc6x;_wk84kaJwmGY<W#l@*D>%;XruGfaf
zfg!tU+l@%xChXb79a0;!8^gBUnaD|Fcgo4h(FO7PHtDxVz!r*x+OC{V*8vx`=g)uk
z%i*8u>>K^o1rska-ftzw{E9lrVIXl~d_{h-;j@2&1hH03Xj%JbTl)h!r1i@`2qs{9
zHR$QjnC$`StrerpS!bTQ7pbR(*G@pzmR;2s=2>(_oSa`7Le3a#vb<x`9ji3aMH(#R
zr<hJMnb6ad<li$pdm~QxWZVn4<k5EWORD47ul8w=vFtba)+x8(?X(W!O{zmdbHq^f
z?FSvx+5RA5GA>IP9?I06M>`kOFkBpGXX4<XX41=Rm5Qwi4Mre<{kJV3S;0QgZp6<@
zB8g_U`P6MiD2dDGWh|HeCjJlTvu}cQUPpLKH{$8ebhD;cF>+9y2=;ylJ;1A%WQJ(4
zH=~`}r69S1-jP_-!TtNM-L@@jyfFYWIRWz3<l}56vxHrWovZTMtU~Toea2{fnZhHG
z*cFRQNmkr_4<o=IKGZ6#Ib!k6iyg}@PF<pCB-OV$;;>NBW;Cu$Ete_1!Zp=^S8}md
z^FWf7@xO0>6RMP<g+Qrut<Qbjv*Y=*XC*(*)cVNJUs3;FmTolKnZ{iE#EaD})IC+Z
zJ=T;a%Vi_fm|(na#R`eoe}0TRz$K8FIKyRmZBNOcy&+{yZM~ho)%tsS&Q&Sp@_F~Q
zgrm_mZEPypNmhQs*;Te98r30ywI7(CaWuz^{?Y35I-}p_=2DFHGg+RxRgz>3YU*>c
zCk<+{c@yTVNGCI?i^{#L9RTMel8&%M6W9F4c>3LhTNafMh$}4(@sOYY4}0$!$aVj|
z59?}|qJb!@(lA3Pvs5y(M`T1OlE}!eR8&f2@9aH7HYr(=>^&lTZ<+V;cKyD;`+q<D
z-w*Bw|KIJwHHwe-c)ecdc^>C+949gpd*uB{b=SjTmgHLV0ahwwRY;~vV#2UoYtt=<
z>s~yTbEN=<>w>^6NcBvMKMy=ZD;G4)OE>$;KzP!cL7sbt@(nZ<ufF#p1aYZ=-J9~s
zh~+9<Y2VPsD<H+*_d1a*Fdt8eic(fmDk&$YptwfCZlhTefr`^J2w1?Jh3<z#ydW0z
zGVjrP8EciO(EaS!Q5oFKvqEvB4ei}G^{yaa7F7>|u86=03iwaZAqml4H-{r!jq(R9
zh^O6VSW$piOGINe5^<DZ?fv@ZdmkWj3CA^S_~kQJXKxB@Mlt%;sPxU5?chqhOtWM(
zeHc~RcQrDQ9C{X{8;2eryLQ|oDvC+A%DDVJ<Crok8l=YG<{v2Q5`qnm-*7Q!$d4N!
zChk@)+LWm3qsgW$`P(g!6GmX5IbZ8F8?$*#Xr#XO^K5~1foV?lMM?wm`weK>K#PN$
zC1Gj}ao3kG;LFS-kH_)n`R5`KHQ)x+dT%}n!A70yyeEhkXrmyxN(Kko&+jqla74o8
zIf$Mj!0XV?u}Px+y~8jwBLnjh^U({5oWG9tK7f2+#h~R}J$``aFD=z~>+Hx`kAMJm
zoDZ->ZUhe+!>Lhmq5y{TD?a2f1z5x3gg>j_x!*t`N(SV;2+Q#q*BLlm!Q-$$(6mrq
zRh4KLp>U0jzJ$IMpo<C&t8`@oqCGd->`wEv3seG}lAzu~;)Zq_rodqjc-~{d*)=pW
zVhgiXEADTQ<l&|yh1*+NrY5g2Vk+8^v&e3IQer^`sWYsQmXbnQ*A7hNPck}^J(%zE
zoXlKXMTHlj0veq-6!+X>5*Bs@`35PmpbcE^3Ls)~aA<vl9^GVbFW%bmhmTPCpwf4m
z8#<8o`=t?F@m~5RfBeV^6b9W`YtX&?7`eC<ssf_Fefjd_h^~sFVtQJd$7(wFo{zSP
zzTVyi_J2JK*wZ8uPX_Z}soQ(fK*a~qHT4k0u^JxYUO_=)qoYSDDN*7OuojX?Xgraw
z0gswKPafTlTf&hLCn8`ua?*Z^mL2~OrYz3Y!7Bd<ois~r1A|jMU&HX@iE9J6S(NrM
z*wi4BVU?t&o=d*TVC2#3`!V+{A7AsgZ_fN4ZLO_HW23RhckI{!9u00DWD%&ikp*Mx
zBB@x7LcRngjS0A7sE3edXQicKsr^OI5ZOf>cOB#}=)+#UzeQC`i#>-gR=a4O=*R)^
zfanjWDb8~gT|2gITUuULQB{3&H5{wyGbZMa9NvaB&S7@oEh=wJVGj({fEg4n_4x7Q
z>uZbJ6<6P*EX1yMSQxvBn(Y<SSZAJZ*JwUA2s-MiX=xZKDUW;{??e{Evs&%IEkOAO
zyDaX!_sAK6T?nRY?mbLTU*lLDofznX#XgC(h$Z(S&z6`B{qxH=CIC|)>!SuD7gj4n
zB9Bg1wFuu`EE9m$aMm|m2uFsxIrDxWid8x=y%B>ON$n3n3?Po~iiNp3($M{z{YZ>a
zCPo<I3z804*Q1Zvn)1x`_=yw4F415+AOSf1`1_x)>%Z}|k(uWMm3n=6ublP25}8oV
zl?xSt704K=C@Hx<q3iQH_Y$^%B%7r9ZzOsi=dELA=jREli<3xgY0?zk6xzy&4FAFf
zlos_$F9B$6OW!pn#(0_Dy`!<wJE6WGj~i7T^1uqM$P!w1$~WU<V<?{y5@m0}b?b(c
z&g}H`?ZJ;l=Dx==QMPdLcObUa3h?_Q7b8p9lvmvYXL`QTYNF$&vhw$G#ap)oQO?Jo
z(Jc~Po@{Dw4;;<gdo}+oRUN5Ii|_>S9T4R|#$ba+M@7jSS2PGuAoVjRi?W8ffz{Iz
zahIvzKCthrB$8g0@&w1vaCDCyyT+I2a7+aAw80@I*JiD+5%j_<cw}OW?)=!Q=YsaZ
zxgFOn=DJB~Ijf8e4P%zHdefjVUl4;Dp9j~IeBBl8Wuz0AMt=MS#|T5<jM!*tgYlmg
z$FH@OnRH}(q=jQqhudAI%xn~XCp`BULZALKw}E&l?FQDO!>TN~l1&t&ncg%>!~s3V
zG6^X}tg_c*Y6bk&zfIx#4GxH@i`Vxkw71uxk>Sfs@Ie^{;*ygKZDymvmbi3@#O|yE
z2lf+)!Wgg^I4z>4sVPGyU4vvT`y<-Z09~<v;mCM^8$KcmNuQzEaLtlX!UABdZ)z%h
z`1_v|vRFzdKXLpX8<Xznz}AZ!RYC#%iuUQDg9l?5&mgrcv67xk;8i_(hY;MBEluLc
z`B78jn|pjZ{4e^eVt#SB;<?N_M)cQ<(!C{s*|$>Ja$aA#@RW}XC;ph27uJObM<Vtk
zbFquKBOcCJ1vywS#K$#iU%qlh4-`Tiwmlp<v9YeTDsxEdJ+Hmaa8F1~ypxsOe{ZYC
zp<lxLmC&@5dso9w2MrMz3OSMyWLjV7{yx+6g6cF+z^w#{THJkzBJUW=NebRL&d&a>
z@)GIo^5;gJ*emSwT0xJyqc<04&V%T);A<)=()P|CjW9h>6J>$aR_QYeuD8KYg+Hwz
zT4Yz^Nk=cIfJ~F7=TZ5#>_C~DukCfF0oO8Y{trU!G;99XqaPnjdlW8ya-=<L68RfO
zBUqxpi_XJAz2hjA#(wKOR;TmFbX(vmfUxFlM5(cQO3%C>y|xU?0XxKaC#MZWrJ4Yq
zH)>aOSF=u))3|?7`*de_UMw_dpB`bzu9=A~lF_kM9Ar^kXeJ!Y`bm)}XghaCARW!E
zUb|O2%>s06kG=hYPy8`!fpVwPuk~eDwVcP}4tz$-C}2a0-7BJgZLiB}2VaF?tkE&1
zUnxnzUT$M83ceG~d=A1^4~OI{U&g4CzLP_zeQ^9u`<`k9_{MaN=u@m-1gopX_Zp`$
z2fSyUvu$tQq;5R*6S}k`=K*?hu(J=Olp;RHs>X<{%^echswYmKd|*^kN(!`^*!v~y
z9~y#0F|Pt7l|S(cVq!ZQFTr8;MeyaE=?e>Yc;J|D=+kjdI#mUQuM<O;+;#T2mr+ns
z2Kf7@aZ4GHUO<+C_z8L?ipZ*;5K}KNjQ?Q&aK!yXYG!6<fn8Q;Xg5mC9Cbb5Dz#E9
z?Cb{lIbKAV028$)=pu%~A)?Cb?$w4i1x%>&$BB2ug2pt;mC@-AbqXWFE!eK339j(w
z@T;;{5_;OO=s24}b?5oxW3dOdo!Fmn`1$xy>WP`~FY~-Spme+~=h10P?y@NYfM{Na
zYme@(hzEc(J?+t#I$##Rhd(a;Hgn-ORQgMNA0r~tF`B+&5Ih9z`;aO;2*CmdLKiZB
z3GBD9NbbE+U?~TPm=T;!7n9v+@}S6()CvL}CjGRATr%3@=`0MX^;>E|0Hx*-!T-xz
z-lk+16b$O*mPeC9Ev*+(2i*`Tq_KZ-+V0-9ONXonn!?lhDb^^*;l^JG#0M+`6gZkW
z<}ZYzL&Q(ot}bz|)xg3;x2F#k1CrlhIKrVPg%pU4P7sE<WsV+&5X&uGjj}8QnAKWM
z@A@Hnx+CJy&Zdu4qO5X?iXYYtv~-^Gz%aul7}yoF<#$N^5u40t`;oJVIIRaQKZ56e
zWx#gt3$we;LR>gVEw?~!G{ch=mBK_l5tc0fhfB#<zU;<`mlrx@3Ol!_XJ+cl@-WE5
zfN(_UGGK%??~~VnK#ho7VIf4TD@qnf3tYUIRu<^{tEnj(Qk0@!d-m;%%B=$66_CJD
zBz}^Mlk;QlmgSWdXtZaVB`=Q!?p(j<8`+k6duyL)4$8@k$(}g4^!Zks8yZ0MT*jt9
zbt7n~x0g$=l+?+|sh5RcVpjHPJ7%APx)X$D^Pcev@~~;}&(QPvl+O8p4y`Hrit>{d
zNNteRAU8OQlon494FFJH=P~1k^AiG{37Q=`^M6h)Dr2FU!oB#GX&XQ*D=?NvYEXmJ
zZ77ki3@-BAn(s+?f!<pb%11H#2&P(qIPnyuJkR%085@rR5u3g<wXoo04>lrZ&2Tkk
zG|7>1W675^8*|C`?jQW!+bg>7=;EJ>Lm@#En0ov2<t-_&G6?qO-ylD~@}*_4#qnl5
zYv#^&y@Fx9RDph!d+c$fmq+5S{6c+DqV1(LMji>N^Avl~eU$%h#TLLOLE2$7zBs26
zei$aQh_l=W!OoIB7W9JVj$Awdg=P>vLOx+`q7`px0d2wBE`pdirZb8xjTql7rhDtw
zu|B$jXhf<xm<|I>0iAeib~X>s9qSfq)}M{W25991WS;W^PGrR!qP^GbfJ2YCKE)Y>
zw?;8nOHLnpimcn!-5!P2+~YJr``GgRU3<1pXH;x_`u*clupY{wDfznLA>*^=9memD
zBMo)SwSnXHL1CRURG_tl{CqO^D&%mE%@KZ!*gscrQBHF{xe+HuY9n?wD)~4RUDqrJ
z6K4=U2ii=)21XoVRQ){70_S=|^*6M-{=V^i?I8I(=k}p+P}1FK=Ybp|d#q63KRrPB
zl8#X>qW5y*=OD|}KygC@64Zz}30}<Vapij~`hX62aI9FS7k(6QT3SL`_IZ}z<Az1Z
zM}fcB*OVTF$1YJFtpfxkq)auf(kf%2-OUdhol_`#&8ED2>0?whGy;tnYYO*slF|wK
zi4a#6msf$*yVx>Oe@^o}rY%)`X^}kpRwkPEX9QZR0PH+#R%JO9+-ZbLD=!!|FjOMZ
z20)}>e63(>Ojei&1ao5>PQ?6x%rlN-pRXyjvq=)zswD`cA?C=8lRds>lXs_p%(Yk&
z!`fSa19xO(6yLB9_G5AdPtO3OA3LS)=5Q^Man0((y9>e7$mpX~WY2;FSAfF+m>Ev#
zMZg3YO?!mvPC05u^cLXNf{wIg(0fK+-X^q9LPSYjrZcn|*u$y96aO)`us04_CR~?L
z8lYal*|Y{U<<)Ej(GYesoJaE;S!}(YE4p?}R;iw3X*4{YuT$B>_(EA(SUyB-TUKAK
zKh;Ee4G=7~Wsa=0bP&u1Hns_Tl{$fnAzYZneD)HDi%9b6)mShi&jox?<Rb$=#ET{$
zwgllf2F3y*mf@G<BYPeC(5QFdHlR^AQJjI-Wu7vns*FiIzuLhlugkAG%l;DWinPA<
zF~5Kzr$sQ=g;$h=K#OXgq8dA=D=&{K#{GGN+yg;XAsp=hr9#mZg==0}nafg<50t_~
zBFLFhM}_$NOGLpC{%X#lfm9BA?QhCth3bCV6om>3E<3{5cPO0R_>`I1UlW#+(w%Wn
zQNr38b<dt8U6avYAHSg!d9r?cX-P>-Yb#!z11cASL`fKs>}MdA?D&nd5#Bdi=-|Sk
z5QXNwduP{f`x&KvbDH+CZ5X0Ha{cY=If>}h$htVAbHb2&s?3}&*r_6;(>=V0`uNp&
zI8pyH6g#xp74?pbI-M5C+pMSp{x8HHEt8_aIWD=KuRSoi3w0fi9cp(Yx1Tf(oZ%;U
z6AziKUj1A4ONkO=A5b#RRM8TiUg&fpU;5CQSKZz1gv)IRE!p2A`1YnN=Yi#|q-mEz
z_%;;wChoq~pJmd{%gX9z)I5upbS^+Cn1fk~U%pJ(Y7}G<e(FQP-eCR8?xT)7T;f4~
za`u{IYb+njahuSRp^cdtnMQ^B!#Sjl`ijxg@X9DSNooyX6(lz*Dqn077@=pOK)prV
zk~7sU_$##iuPGft0zq&F=126k_4O$ZAO5wGH(YNtyV~XU_HC5bOFS;&JHnQ;&^X9`
zpKr0=-(Wa`OsPxIFCYN)U_Y8`S5F=UcMf(<vrxEDsEOXc&uaHo5imySIhw-@?H4?H
z9N(1$?$njFS+3ri_7aieR<HrM#4WE-{+9SSRDOM=yaq^lczliUT7S@S!78u>2gEN!
zN_e>x4Mk9Z#K=<6%(n9Ne+eFu`)`V7*1qSzyLT<^=^Gs7u)N^2tA2P9EI^dhaa%fJ
z|8O#82-N`cDNHW7fWJVv+I?2?tiUROJ?vPA>E0Lb*=9t(N(dRv4f2)BN6?g`(NodD
z93S01nHEDMR~cr2M@9>z&4DEQ$&)AWu>xIT*UbWW33MfRe66P?oKeOeA9D~s#D41=
zMVdz0uTXo$5?7a(UE~VP%Xz&4C+;w?0f7R!S^W601-fk!8xr}~PcK~Tzj>A8xmuVS
zD?3CayfdEaIx;=XPQa4f9NLip5UB9!;9dLoNB5e+J|Ub;paq>DEYCCPxyQE1*v7<j
zP^^&#VS>l?-E)~E27)&{e#mKpf&&1Bu)mZLTW99rD1wLs#rW0Z^5}nn-rGMmCWuG;
zRF!-^aHs7f_2fVo91{rBsLf?`q=Z6PGQX?{jc~8UN*MUOx5rTARhkEMlu$}S=D+-)
z8yx~C7j6DtW$)9%Qn!@k<vk!bov)L~w_*645Y;V6ZYqTq+-O7f)@J1zkS;ivpb{cP
zFs;+sK!}hU#_t;j6D9T<+%Ya-v`-y}#Ms{dI{lXu{y8IAzI028?(cGE1LdJZAF#EF
zy^RL$++2KoJmgDw00c>VvP<av`HW~|1;_+2>H%@=Y4HV*j68|B>u*Dznl(gz?&(vg
z-)f#7sxdUVN`ZlEPUqV0NDQ2&_{83vS}f@D<nbZaKY0%=(|((JFy7J8iyLXE=W>u{
z+S6C@bm?4G{aL?k*8mmIgi9e@QIu+jvfDoY(s-h<yf|5NL;A*zXkl*0i-0u|7sqQ3
zl9PX7H$-4pdI_?6@}5IZ3kuq%_FIO8zPma!MTV0xN{(5pj1qg(49GC*gIoEJIZqsf
zugS3^N3v0RBk{p|IuN+`EZSr$!22;_f$Ddsfz1QDJGtG5)+XiqY}y$c4Q*$Cp|_0C
zTB`^S^AKCo(gBWYL6=KXQg1GFJni5kQ)oALq~s&Jr7G0>kOfo57_=&fsf3qPa&Z83
zb#;v;&_%pSC_%?5juddzU>x_#dxw3hLQ6&;9gaN-f8L=dKkob|@bryvF_(*j7|~t;
z9Sl068~6!iRY;cMj<aR$E72uDRzr&>7@Fcds&A3W*Tpa+0}mv)EuT%$#s@A0>jjz4
z=g`~iwP{t8le;9NdHt$s-FJlNPyibj@%E=o<?AOGTk3n5M+=Mr^8(kgA8X-c?nnFp
z+D-9^fPheyte6;r0Fl<}<*Ijtfe|J|FX{T|J*8Du3FnnP?*z1B3`9;^ncwN?X=`s!
zp9Lu%=-oj(<^y>@YWLp5bV+}IYCHCPLVX%YC(f37D%GXlhhhJhtI~gVmlfG9-dG}S
zo7?|$ELe>3bJ2VL-}6*qK<R)|d;Ty6Jb=a_K{rX!QdYi3-z-r)p~GV_t^9}U<OlT}
zn{%9;tpFI2;^Of<gfak`p0MqltI?n`)LKZGpf82AQhYosVHff<qah*mrJruyi7qf2
zE**gSuU~iBGXcuY-W(T?VMp1yqU$m^wxPc@XMIUVe<LWz^$n7ysVZ}{+6?o@l$xrY
z;q?S)VR$wRnWbuv_36)If9^5+`TAa8jr-v+%V2p<G3tDN(r};)c0UwisT$Oe_6giI
zFeokLf26qt<?M659f*+OqEpCg1TVHy6{h@-->J<8w&Y<70DYmCL0D8}Soj{ezW!Ed
z|8~vscLpd3K_?ZE5^Dh22=LiFrm4<7A*4_<Gc@OPo-|zU{i7%^e|g`H+q$}#;Xo0x
zC1kspGU)ApLnnsTn5eJ6&E+P>tYg?^V#M!;V+(xDY;4t)l?s(?eP~4DY?N~1khI^B
zxMN@-=Kfm}VhVDnbBGuvwH-!t56~@ht5t}=!kc5yRWSQbX?zVkNWf$ID$u7qv=liX
ze8Sa0rGbL{$f$w7zJzTD_qnfzpdgGnenNuA^zO-xAd1N7KC`iuQ}R&4z7svDadLX6
zoY%WG_Y=JjUN14#{`&RM+&tXz23Dg+y20D2MsP-APxGTNv%0Fb<gKctN2QHy>v)@L
zY|`>xqI#2fbi_kkNmG-Sg2EC$MKHsVuwY<eF~-)AmzU=`|3(V!a&)Xth5|`L!}J#n
zdv$L-{csp@6LxvCS)P6TMrgI8!164WgMy`GAS?65nt+l@{WUBVpm+fPpGtaI&{KY5
zpAN?)Eno<u$G%=Tb?Rj>0M@W_*4b58o-ri&W~Uyo<FHiLdRsm=KF&m2I-DKQ(Nz7v
z9<Z-zwBG)}=8SXEvKl3|Jz>*(PUG<M-qN;49Lw-Eg!~Lpg_I|tdUIehG;JN=EyXUZ
z0>V8ldybu(`-2m`LL?oJE!kN_%$HSgBuE*4m3F5N6H=_gzZAPgA)o|+p(AvyCz;aB
zA~cfj!x8kvTQw=r;x)=BqqaMCw}MtAk7|bop7ASA)+nqc3a)Cid-sOH9BLhQOAbSY
zT=Gg6nvK?1V;VNluEH!@%xx02CWTrLSPHG&I<RxQARv&=%_5IQ`6}4jm7DqC;N9q&
zZ?XT3?Xbq<blbA6PzjVGp?Si*C3?RFYB;|6Rwx;RFE4p#rKZXiQ;X)DF{eIQTllO&
z(Tq07Mh*u`R@MgcFO+umP{K&`B7++edgJ|u|1XG9!NLv!EJuLN9Xa&T95hcI(vk`>
zS{?u{D3<G}<oU?JC9sU&%C$1Uh$&q+tC@j>mt_m@$+t;`$x_yD#IRmlWx~$H>iG1+
zm78c#e&1hF;IQ%;Wjcx{EOF2$a!L!7!x8L19eSl&0DG6-;b((RF{h9$6Sk7fEF@u)
z(4L%Oh82p_oC`Lkd}yonj=7y}sNbi&AiXHlKD4~oiDE7U&GCq$t<X&A?=x}jI_Gks
zXZy9s&e1s3XAQ&N1)zA2h&bVJK<LgFx`_fx%dr-Z&eH2ku67ClTSN<gFF=&|^#*jz
z<SRZ2$ay$B+)AM2YuS2(<v`j?Lf2Wte;o!s>F9bR#)A<{&V3BN8V-Be>&0s<u==s}
z0N9ian0i6U=+Tn#bd7e~a2D)o|D6#4(r#-j(Jg$e7N3|19pO%a({@B7>r)^bT9)oz
zkK}X_RmT1)TA2{AipfWONL~C*IEiU#?gGM1lqB`2*tEqYK|R3FwOlmj*De3Dz5(%$
za%FynWjD&xxYLc1ktdh0zFkGiDz3>+2&9js-+I>7;k;3K6!SGL@7s;^vebL4di;3r
zYBdXzaf!%Z1_!RI$0@o60lz1!r9E3#+|Eaay%kmSs8Y<{<Wh-K18UmXaL}}>j_^O>
zw}VnjEBoQ)9!ljWYr<F@;yP{$%qWG>Yx<FOeA*s*K=p+$CXjc5Vs_im3T+6z`hRe+
zr~Lr`8I+8(8?S*7FTtY-Dnr5wM)xJcZ#sv)IiJP+)Rc}k5BN7N)m=5fl!g6K0L6WB
zxMgL<>(FGrn<9T^TPz!~D&q6=gA8vJ1O*)%xmFMtcXq{tpzJmnTB&kKVw5g8u}j9H
zlbkW-<;7p2Zg0MIon+4%6I+eqb?iR`v28xobaX#wUC}-_M_aIB5NO&t-EA<^RZ7W4
zjfYZuPsd%E`i2JJ@d790U&)X`qY&Ape{fA?VM0JSo_Ts@Nt#AN{4}HH^Y=Oj)nh;I
zt-Gq`x<{P+um#)R<gia+$xq%K)}$P!-R5zMO#f`M!gkT!Qi`@Ij0IE*gThW#>jslQ
zgLC|)PgyGnPf|IZ91O9W@RukL{`>Q6O2I<F-1iPNZ9yh|^r!rQW(8;pX#U-QSXGXB
z(K@A;GJG)J9FFNg>kS14@Y<%pZO{Q0;vqmb4Q%;Fh+u1vk+nJ^wE*+&O0l%kyXOOt
zQ$fUj9xfBe6dO$D69|wS-4wVM-Lee_W7lGeXwpV~JC2^_#*~Z5(7^)SHj!t$Jclp7
zoZxneADx{JKtt+;i)JcWVr~T2Icg{<5N%pwqFPFTdJ^xFBEp}SuuC*Y{i*pfk#+f8
z&pr&pqhwVJ3tU8FbABMuRMJqy#?J1|U>C+4ExYa}z0n-ZW9*YGie$`fUS)YB%TwxO
zBLkv#Z`L%rd50zqjH?h_y1)?c%tGnMxpbrM`*%!xy}FWF1<ih~-$ACK2Za>5vnm6;
zkp~MN|GIk~X*FNdPe5Tx6WQwh+JZ|fD=ZHwI^VbemGZCE0&08iNa4{8h|v9~J@HHG
z3nV?*#rcdH&S;7*ey7q_vp#Sz4a{zdS(?3tkB1r)d~I_4iUYp_3WcbWfvyx_b{rv|
za<aq-r^u|6KbA2Ii<J$7$~=mxxbfsyCh{hcOv4B)7Xy3!{RhxEeBZT1`XI|>5BykR
zt5)z82CXn4oUYFXUC&uIyQ`qfE3XS<@s67qf>e{chu&e%Bdtr)X&PLy$PY)Jv|S-#
zz05x=ejfunlslio@_LvZ?9}7VmL*xG$WSZ!S`c|nssX9};EPg-Lxd#ylJ_#fYM!<u
z2C8@TWTI@$pumv5l;9I#r}_9)O;<#m-Q^={hpZTAxDDJ^x3V^)C5%h^GwLeRL%f<U
zc`2day=xjc>J6vn@bfsiUx$Wgb=gX#yz8<x1PFn|`8_WzpASv%944l_HXR@<*}H33
zHh3Yc5zwj!K=dId_EPIF!ieo^sMBg6{mffGT@V~>H&)Q}{2?^zt*Rw2u{x@~;n~D`
z=FF~LyWXobK^^E_EOq3VP80=ZNda<z#9vQp7l;H>ZDGapH<4c6pBkxqK}-UQQN12O
zK8(@{f0&QlmgqI$Ll){=A}HOr)V5-Ta0l79Y*@Pu@#yR6%~o;OfT8*k<?jr=Hi2#F
z1fp@eBh*SjAP@Z=&;XixxP+MdWG!*wY2ND~{<nd?pqD^hh<Cf===to6acM<cPjP)d
z=pPe9jPeF6*Uh0S74#u)c0Thz;`bDB4xU{sfk*H?3vdC2xtDc#{Y^ByhQH)Z7AE9{
z7Zeb@Cw=BnaNY&!a}XiX>@8(}GnhuRqWu>o-5dHDmXv(m)V4Pqg|SJ$Frn|Si(Ya2
zFPu^q?ONBbT?5i8F$!Z5Mbl>s=*UJkS_sg^D#Zib{UP$)yxNbm59#xaB|+)U>qW^i
z|L%2Y6bL&0RwW4y9GJp78T*(;R!@ZB)fPnc!y_$B^}4B%6HbDC!tH?M8l%*~l!4V;
z9lz7xAKE2Fq3Bd{x!)jvj*Ooew_G=a*D|6dh^Q!DctBA+=6t}6VN*}{EPyacW1l?9
z9!w{{J9Q8_7O#~glp6>qtm!O^-Rlc{_nddQd;_vCX-u=)69cA@Z6h#qgT$@i8$$3i
zjE4ioV)0;QO9XHBIVsdB7e&bWkhhT*ipwQ50<6cP=l$%ai4>9dXYzZTaT0-q>$=QE
z)#a~oc(oKxCEmJ~<}HK?aW?YV@;)7E%Zu*y5B3p0^XT!t*1iBr=4XH<xkf12;ET-n
zm~C3B23(nA7tOC)$3{lpLE7Htoeoio;)GtqzF-W+@+d4#A|JOakByB5bx$+<AxQgI
z?avDd5#ANKvbLBp5|vWKHvin$$LDpo{OYv36DYZb6L-zbrg3Uy8jo(<xpS2Fk7rF^
zh6oS3hgfeg$}(mFEe4H+EBV3=#wi5`iQizN)1VB1tqUU@-~p^$zz=;6p8PQ$BQL9u
z^uv3Frzexr1j9umS+wFBN00>kO_#3e+jg1U)PhBy)xYlUdCV)1uX<IAUwG91<NayD
z8|2&4MZ~jI=nTW5WItl>9I0RGciq{Ef(QtTwH4oy9An3FwWurqF+;AL`yxoX8UBI=
zBF?9{1l&LD4?!=5r!PrfUropzjwHi-Ir13_AK)PYmtk7>Zn}FO`v?!5qUbmL^}%R<
zfA6N?+qQ~|d)T+Rc2600q2i{;U~0)IcpiV8AahL<ydn65AY-F6C}lTlZGV5^G)Rdx
z*uoYrAc;YhA6~%tT3hf)RUg<B7StA)-bN?l$U2+v%>t9Srl!wWJovL**!_3aqVRw7
z0qi0<*r=<PDcE#iXIK;+1UCrXQary}S~#?8eZKF%FuS$W){41RkR6)myt}|UXiGTK
z@C!>)T$Pv`;kvQUDjR8Z{)1Ow{^r}v&P|xnyd5C4iLjA?k1IS)u%aW!T_@X0AJ|iT
z!b#t_f#aQHk+pvI!ew&Iq-0D=)MNFBnH!(!_bRlPkGl1-g_Pk&{R^(A>e>V`?84<X
zGcBEh=iLSSrBv$Xcxk%Rnk|C~DTQ1SWZswQb~1UC6?2W4(`6wUBxLaqG$B(*=(lyM
zMDW2GsyC*Ne!OjZDJVYR{D4vwrIC6O_avxa)uPv6cdZei3JTj%;j3#1z<=m7{krVL
zf!U#zNCYxu4#@i&^WrwvCY^kpe3<_}A7!gI(9^>l+QYQ8ofwnddG#oK*zs0}Lu~#E
zPJkVjJJ?iRorIqQu+|tzewdp2>};zgQY;X&lr|F3=fdlo01ydaA?imd%loq3&loiI
z^??7)#3-F{XJKRWZjJ3y<6_z+diioOtZMF$U3CUU8cL}9shRQReJH~B!Rsak&`xRT
zn39WNG7yvAtgs!gzV(}<XiNl`5l~FsQ_WKFz>e5+L@HAd4-~p&3iZJ}d^4mH1DZd^
zu)HP}5@xoBvQ&-`-vIt>U7Qn;DqQ_IZUu*oh2{^G*pg)@q=u884h;-6fe?na!O*%b
zz*#`Za#{XvX@V1oGreb0ac_hIwbKS#Y)c+CyQ8@<bne>Kgf)k!d|gazWPK|Zi8by#
zch3%DApqq_PS68eosO;xRMpNam=fxgBOCp$t6Kw`WPGU?fpKJCN;0-Ox=2xMYRvA3
zw*5wGby&N$xR~+e$w<5;n0n^A?+bBQxH)~!-d`{o=UuOV61pn58u)cygMCMdDxpVu
z>BbK+R$a~35SsyiT#2zj^x{gQG@Wlq2s<s6J!DXS)0w@>(9TgccyUOt!i!-S2ySsh
zFK>YI1gSZ1OTc&HvWY#W`8$o6Va>HrY=Z;g83X1f*UriQ0QIBFF-O;m&ezW`c5z)a
zN#%`*v+53;V_sM3cK(R{Vbg&}5h`eNTx-V-S@79&t~6riXM)BLF=ou?L@m{;ALAxe
zip<2HmMrpsH>vX^V-ztk`A)?_XEwE|-BjCdm%34Ng^t<4;sdi7pADU24lUC+Fu+CK
z;`6(~40!3{#hea9VoYD@K=&)WY=?zb!#ZkqWsfom-}8>s->Ld#<K*a0@A`^o@K#8{
zU+{)8EN5&3&1qnZw1}-hV<*I>BtsXI33pllzZ|;ITVOI@d37~Ulb8N%HlbntI}n(A
zakwAkiY3fO(eI*sjdlIAr!)dicn|PWy!}be5Eh-&vB#W|r**-5+&MkhBK{0sAR`CI
z<Of;dv)0?CF^;`xZQPXWL`bE+7U5ofD)^`)kihi3+aV+G%kDvlbZvk=t{%-+jA(D_
zu@H?zuIijeYK&Gds6(pc&&`QfBD^xjtvCMs`SYRpo2$IAXAv1Tp<&>~ND3$$rzy9U
z&SV3gqY1p1%yn44;#_}HqGXiqfWM*Jny|(>cNQoX&fk46&4;1Vj*#z|%^VNKz9e#%
z>HSA+yI}Qfz9A09h2vE5&S?9)2Pi3#rSsq0OGyNkUm9~)q7<lum|0Xl5^=;oN=JVN
z*96lXKo%*!b~JzBcmY!z|D)QQYX#wqkRjom=%9;NibuG_wk_{a*?c&^QMl%EZsS6f
zw+5D$o`S`6nl|R%!|$H)V*fey9^>taLZFA44D&A&MUb^Y+ZiVv&EWIl!z>bYSHn2u
zJJ@)D?u<_726*CbFy3SiI|pZixbH};9Gx`+8}MBahgr{`S84;($OQaBm>(`9Vbx<(
z<0ZBlcoS$K)M;2|qb~uLPNPi4c_HF2kTs#`sI@y(huRVOf8hE8)G>=dYQ|9(SX}%A
z^b-j;XhV%HO8q#t02a=PskOAVRj&Fye?5pT10AT8th_&Q0?NtBrNTZ5=qUIjBf$?$
zP3dWA*IErjsc$JtB#sbI6MP#Gz7HHdijmHG!^%j;7n(Fou>`U@_%!Qkt+4FSF7XtR
z0`5SJ2SRd3PeGwS<&X^m4gLC*3kaP`7P@*jZv2D|iPq!a+Y78;p`1d*K(J!A?oAH?
zfJLOoz}=Q9nc$fcfg#_(gepx+dW3mV#>dOa-oZfNmGl~6C~IDHF+t`66B{nXPjjlr
z%)T}@K^9lW@j4|JN=O?|<1u}NR}FfI=u_1l{(|B5kS62w<K~@i^}{bi`G#erZXXC=
z3AoMlqNL5Vl&-*%YWZxUfods=@eQJ$S}(5!{s&WfuT0iBLbrn-iXZh$?ocCE0fuG}
z1@Oa|7km26kXir5WM&j&I;oh@8iQ1ZC_+zVrs-feBVrJTw?kwz=s<}6SkQYm0`t8B
z@dh3OutqEQ&Cb&xwZaOnyGPQg6jo3VAxZ<DV@q5APWBiIM7-0tYx|Y1(Vgxb<%WmT
zy4ULgGjP>&IX~L|nZ1w%>=tUbPtae~>HNeX6=WbV^EhKI_?V*KOdV8QD1|^V46{h%
z1>FLdV?R@I8krBrq6hYHx@`<>_kNa;eXCSfS4aFPRn@Y&wwz8l$z3%6gV7Z%HoMbj
zFAN+1a9W=Y^&WGhF7FFFnW<x6#Q4R6;NsmBx{WfBD<MbYs;KC*iBf7B8lx$Wdtd@S
zXbu6R;RlD4BZ%iXM7$*%(N)H*>VgQw#6}su7Hk&KU@Tk!#nB7d5-x+7odTGs$T8_M
z))pGeXi#IwF{L}9s;iZWJGXG;JKA1^DPdq>LOX38ym+YWl*LgD04~9ofk8BM=V0cd
z8y7YGA!DnL?bq)J;9w{OdwchfjL?vihb@{<A=)CRWWGWMif9g;e(M~A{r$l`JibfR
z+<XSqMQ~2RAeW(@G)UBgg(xZyq&8Uc;EQl=rOzlS%EEu;1`9FY!~3$e`{$=W_jLpR
zRGie}d}}bNi;~`envk{2M+%cbzR#HqZ@(;#``MJ>Tj5vWYj&1paZz#<FBlBDMiw_$
z*Sp9=Gd@jZiI$a?iW!`YQ$v#q)dv2FTmr`eTq)cNg5N_u-(;c`pK1Tse0q7#`{b^#
z5SyaafX{3&%FN5N!?%(OZ>3OxHxnAbK|!}vRGMIRgI<ST9>1S>L@oL@&;((o3=j~^
zaSpiPdMQew05d(HSv%~TztCn3DRr{66DcgP;I@yW0LoNsnmweX=k-}%p2c=Q)%iXR
zmH&*B5s_Vn7{n^XK?cJR{P||1BMi%?rNuZ|3ZER_#)u9zOfgTK_=ZM4jLIQa370xC
zEFmk4Dcxp(aS@a+U3!w8>9n!Z-6q40(Lj~Kd;s1F?<Xi5OnyZu#39OC2xzNzn!u!B
z45cLAH{J!tZyYrB2Lf%2b9=lQeRmFPHxpbgvNvQSa3%YEQc&=g!Ca6V&;o$s0FvO+
ze*OA&yavNi2Ni>d{+N#FjwKMWmW$epi+c=f1>EeoHGWoBX~(>e_{AuP;pzmdvYk72
zybg6iJvqXg`ChmMXD2a$EJ`60-0=LHCdhQ~dto~WPX_K9!4r-@S+1^>T^CSj+4S94
zGpLI2SxPn;D>Zxxg75SI9kcfcDA08d?Uo?tGIq|T<2P|_P@`=b<l}RS4fDv1&`^7|
z#tMX6QnC_+znU8BM~`N4Ba4e1#L@u=rnx=fR+IwV{crNCsIO{#?CbB>uUz8tmh{G@
zBR}`{);BesL?JjiY1%8)j|9Y4YM`~f9n#(#%Q^B)V11%Hg0DY=gL!#*K$g`;sLo<o
z8v0qbFVKse(F=lgU~-nN?A^OVDCk-KW_6MRY5}~8f*!VD2GunB%g*XQjU*V1s1n@W
z71dsdIerW{w{~JNfSrly+9a(i1W2s&70l8nJ2i-{f9faQi+5eD>S*A+scMZI00KJq
z@+S7z&Hkc?nKbmOdmDn=<xGy#(yMB`T71_`sJt=-ln2EJ@6)t17Wh!}JqYUBn)0js
z)?QuvV_uOKU8E$j#5^V9i2BY15D0`rm&|2FzF{_dqRWeJM<A!pXZAd_fboK|?;!h{
zIbmY#0`~dPTYi5`EqfIqLh?nBqi`v_ON<YrwDoefzR<a8%RvW~=-G5Z{Ukies4Pnj
zOdh{~KhZmP>lqg)+KoOUERR+T_@Q?H^3V1?gfHn*J8E&Bdvzu1!EqnwLcaO=`5_M6
zOyGW9v57v%Wk&)LAOP0P<*i{8egf9XOYH<Q7Is642gBb0wsvO01P{|=f`?bUn{VMb
z|HZyHY<96|L-VhH{F9j_ww+JZ%K?JoT<!HlcL1c2diAw9V~|GQ(Fw(`5K3@D`Jn4$
zbjFMh47503g+>*E8jLaD4hbp=00GO(B}w)Sb>yUXQJ#k=4C3Zjcw)`)dq(R?ad8}?
z;mXdqe9H2lKjD~Nm>mhb60Sh)t(mWCDHOJl@U-C>U%7JSyI${{Pp@brOemxw@merO
zIfpPi!)baA!hRSAN-vH>Bs^NqF(t)OK(v=SvrqwJdSt4u6G_8aWI(o<>+zkbqv-I7
z^~219g7!c-RL0zNv5)73&-rw>?2DBt?}K~Qr?j*KMF}pRFxN;7G_MQ0yj5yENFxR&
z(BrM?*H<!;Pa#$(w$p$2E4;!<q80?NF-(gx9oK4*$MCgWrd8}u*34>ap26|)>Xc(6
ztTV8BVy&vFNqI*)5YLZ7EdSC@k_Orup~a3zHA05tS|oD&ZI0=@Lw!2Dqv$-E$hpB!
zUlB#r(Z@h0bvu7z4l>PyD?GKJ28;atbBIEOi{wd^N7<HvhnAN5Jz{SUA!9jq>^?q;
zWaJFKQ6Bf3*twnL(s;>C(5530$rE$j-M?S`@BcGhxss8v945ZJu56li;lDOl+~%JV
zSGD;cBw@#hdmz3pUHn5>OycW*&<9#IO^}Q{$FYvMtCmLvkF$SJmfasC&$7fLR#aCv
z8fOpN&*<)au#nKwz{#bp08hxL7Pi;M)-X8>+o4y;y!q7rHy-6}CuKFznhAA&Qs-V0
z5~CvImgrx;mO}d#j**Ftn4miH4(9#%9huM7$+sCV5{7Egm;(#5P%2+(pcO!@R7MDU
zEoQgB@&8`I|DYH5KX9%7pLok<Kig=rcE4<>!DLWZH#{u@S+y+_`h@T#@VYnU?zE~W
zF7B$&58uHjl8Squ>5L2G`+JIjYoCV+wN&N+Osu@T{va)j+krhICsCyeOUYO*B$c;|
zn_xE~eFQ%%m$Z2@Dsd2=;1fB6tc?Ns^%(X?qC`iWgqJe);>fF!S)zYJd2bK?+((o+
zHRZ<PETfR5qFNY?5R9xxwUD@6HS&Ld2&$wy^j9jgzTLsN`6ouMgfVRxf4sdH<*{RM
z-!Ob*LCk-q1d@c$qJ`<hS3ta6Y$c_oZ7ox}aL5m9`^^NK)k~ItCMPlLXR$Kl%jQRo
zyxRPzfOzS!1%wi0h?UQRaeGgq{=E-Hpy5V<$*Kgbq0;afhs!((>x+MXW@c_~408)?
z1|d=D`1HU2l=gtkO(<scFj?xm01+PEFc489wh+a(NOpF%(j2RmIsQ2I=VnBS=K$D)
z3Ql$C@KIXcGco_+0vrug)pp2yy^J4yQh8$Y6PlR$Tau8O0e4bvpvFi1=eagv(4DEN
zY5geo<_PY9|N3@x^khPvMtXY44;@mBTEm@P0y>O;DOpF8*&GE&3DYUPt%#@R{qNK3
zg!NDoy13B|cZh+}m*%f;KGaTh3(&^%(WnN?7sdLXIXr3-0&T_<-|ELTWdcnb@`XK{
zZ{k0H4n@Q%l){BKkcx2;Pw0~V=5uWzCA9PxN5+9tSet(5C!SII=5;lY!r5CXp|SKG
z(1xn6F0G1n{M{G9%}vR31ru3-;J&5RHzXE9z_ZOaQF@Lt7mlhxByavkp#SghE`=Ny
zqn@Jrfim#&<|h-cfO;&kbIa=Wn>Pu;Rw5A;<T~-zN09p1zCap|F7+o#CtRq^0ptUR
z58oNrLd6x=@Qgt%k4!F3DIOaD0IcyMd^+*fa04JlAkxxIyDi?Y|2sZ-ZoXII8~*=H
zfp`l4A0<ccf-?p1%j7aK8xSs}q;!}y$^r-mUzqKVFLW~D{)6thx%sEB!Q1efJ2zLb
zQ4XlMzFM|%qA@71Fo$UxA~-N5gfTRdVV80P3|~naE^nTAzf@7DuxsW*89xltCl;Lz
z<{4wHBzB%9zU$!TN;OCBn-6pWm=0mz+~z>6eUnR@yQOeyY6?kz5k|Q;O<;^2iIL!D
zZup_4y&X+PTc9-m_2;|`kMP%)N*F?fUVNfL9PnZ&Zov_P%PN2*6T{Ps|J(yJ3(7D!
zTKFn-61QZsnTU?Wf)j#cRx{-n;59e|fNo@lbcQq}N+kj0p*)OP1|+pd6LK<qb->64
z5CRYfSPGmzP1v1qbJypuAN_;g1`sifyoMMXG(?Om+rZHV0X2mYT!jAS42EDd4F-$(
z!s7;+J<5I5L~xh4vRc4h60MM1x9*EALcEEIggt|Ugu5(a3F=wk0XS<2{iV!(OfLKP
zRZMB)qj8;WXrH09G58sD?_pvPj37X!hacZvMH(~%+VSLJ56};1Nul`KyYK-5<*7Rb
zZh~MLMF+!i3w()te*r*%tDB+_2QMArwYhrqVJ`CXzh(h32~Kps(Im&bKA*|>x43}&
zW9WIo=;EHKDTksWCQFmiYhhBV_QrBEWXJu1%5h5QGUHWuL8z*cX{2v#%#s8b$OTj`
zP7q7N8G$*A2Zs`k?4|klsOHDJ5g(0W5Nv4j|KbX@*GoVgI+ACG9xq|5u|5ape2}n1
zTZ!sFz|dY}5a&G;oUA)!?NMhn8qH&x&rQJo^^Ur_y0Gw;UO_0%0YTzbLF3=~X6^pa
zZ7_C5i?4rRAfPTlEP|K`?&9KNA4r^lIF=HI3h0!duMq_ZX*-@K^z@{3!0EiZjlpY6
z104u$p!3t}zzEP%+(65u(a7=zKL%qG11Qj?Z)+(pNS>XY6+p2Ji}d0ML>=(W2(zFU
z?3(N%B1Pzc%76(0;LK~;DAN{tKYS|s!Rhy9@mu35HrI7T;qnNJ-kTJ{cB<v3DJGZ$
zq6X-}Y{8bPE`)v}QOt`PIzZq9_5Hc1Xu#62Kp>@4Pt}3c!}LY620H{KP_-<g0}y`t
z<%<^(AhclE)h8mL#R{tt&goxb(LBHn<pUh6yMoRa&~C_hZTsP-elyd^ySNFc8;EsI
zqfKnU(uj83Xp@Y(A<}!q(=~W^OOM^g7_nD8%FK=UJ-E@tEXkK(KNjET1bz?k5=8oF
z{656Oo|?Lg;ukiA0Pv)9nt+~aWZb<y^$UdyWc=VQfqrs9=L-pdlY%|KA2%j9){0;p
ziOIYT8P^eQ<s#JVzOt@6&-dX5jvJ)V$KW(DrOtCWdTbSa^_$M#U@j)|X=dQsfT7g|
z;GFT8IC1;_+J#_@;{m*b{<$5nQ#7$k;|nkp-j=Osb>>H%5w80@Zw^!*bJ!6`b*#<I
z($ePph|N1Sf{%h!`fxK`R@T8U-lswb2G5u*E=Y-Vv`m@)k6Gh(BIQz!^t8E!nFnd@
zi`W%&%w%|VI@Qn>9JzHQ86fD6(IT|DY<c=;v(Y>N>vWVE2smNAw8L03>k2LOkQo8R
zj8APb3<;`T<Ei(>NsIs|F`6aH+vNUzG%rsjIbp7yb6^?*>Fj&_Zeln*3Bze2q0Sb@
z;TvH{1H4oYZ6h%2XAI(G@B<^T;5zf+GrZJL8`q<s!lpHd7NH$n*UqWsfT(NBw1e*%
zC^}binybsp(J?K`5ai;j>uDkfS}N$UoRM)WqSk)CQMGOc2KazRecFiubTkA&L~;p7
z*1NPTz!rK<afMKZFeOpa(>sCjr}y?4KyP*MiB8vNIs08O3qL<5fYy6)W!=qK#FjcE
zdiyD26i)Z!zPC`-s;a6UvJ2P0|EmMJfm^USx^~X}9QjkCFWB+><W8i95*ghwOe@!O
z!Hn<k-B{6hj}i%hzf7?nOKI-G<EdDYXmmkXV4i0U{0KO&<PR8dp@NM(u@p&G)<YuX
zEjf=}OT72C6tqswp>qK^3XaPJ7BEoR55o#5efRmm_&fzIs?whT=hL#Xnpwu`qTW=!
zz!o(+v$;htLJCp3f0%6pLfI?MjW{mx#cZnyL9Ajtb<mc0Z-`o|`yvId5qbw4|47?>
zb0jYjUr!!nw1Lq{FwUW`b>7=UV@C%`?U_frou%y$62F6TAOSDAob!UC%qY^_UQ>Nf
z0dBvR04@3wptv&mO<?@MA>UM2-&+y_a9((!<ARR{1sD|%7JrgOSiQuv8KxNqp#kJm
zW8d`^5h;Em_f&G5Phg<&*n6=E&S`ac!i<4qA)R6PJ|$%mms53Q*)aDtyC#vMp4iMg
zj{qR1I&?^H>K7h0Mt^;0!j7QK&j1rKDxlazBNJR0C1z!)8}8is6S_VZgMMvAMqC-V
zC<OKF1*JXRf9|P?$j8qC_`PQv7&`)BPcfyWM=T0hF*<{&s}hHzV293ZE6If43FZ`%
z0ev8!5S5}sb`9txPG~I4`wv6+Bw-Q5>&%tPlY2q1D(E7nJ7<V$#@}@y7Uxr9aehff
zMvl;9w}T9EGf6W~C1naNao8D$goN1kgEGstlAv-0|Hh0KydIb25_muYrbZ3`c3X~?
zJ}eON0~-*K<$6L(MoKSyt8EN56)~$fei+8tmt?-$Bc~>Dz&Vai2?1)R?Li!db}&Ti
zHu7b(Kcc(|`B7h=-g2X}vojoTrBhO#;5Y)H53jK=J@N&bR5)TyQ`@lqWQxMolYF{o
zC{TS394DUE5>e+Mj1y*rKw&<^k4)|Kzp{WGWJEt(9TfI#EF!(6S3+oc0UQMp+{*ib
zXIL2;W=LjudHai4k-{OZ``{yj`ERS^=&%BaXHDXT{U3hZJVn>g5VW*Z{4nPIWpq^H
zo_G$-5~ERRF)%49SS(&80nR`FCQ%XL;bUWev>0|jJBJ4cheYS=IOQFGf8BmB0CKBY
zBDjNpOxQz(P>_@V0>X|rA+)hNTREwNNsllHT?gw8l$LQgvS70jCK)Q}Sy_PXlZ%z6
z@z89UwC$%~SU3oQ4}-&38deP0rvTCG2!ldIvL%!$a1BOZCizy5?s^3FN^B`nKwP4Q
z{}Db@KxHlw4OekwoV%W#!-<HnD{-X&t413c3VI77sL(O*lwlWQ9=rold+)|fn}?HZ
zY;bQOdbL9l?#O<LsUYBMfD3>z(CG640ft6|xS2LT=^9jIQ4znvVMT*y9lhKUZy1<D
zm3jvrCeUqXTaK!rSD27sjF<;}mY}yGTaCAUj0fDfn}`}uBmf0t(}=)5S-lYC$1;!`
zIuRg_2)02=0V<Np<w@4krY7E#CrjqGj}^k5=rVa^KW^vkn>R$`c<wz~8);EP_{foe
zHh0*WZtPUp;i^115|fFGgWW|Q8tCzM`j&{{8eq~DUqSB_w<TaZ7yDI{czElZ)z=?l
zelKDm;bi6R&bO9`3&}8Uo<QnIQ4k~+0rSE<*gap9RZ;^937v0Pj)m9Wb!&uu;ur>G
zW0*$(Mvm1kC@LsGU8k8NjL%k5{I4PalgH4TKsyf+NC1ZBJ@!@R5DgMD5b*?8>O<EX
z(13#a&A$>@H-^ubfTpuJ-i`@N_aBs1dq5%p5dz`^6bCzY?)1&89l@E3&{}}mFG%gZ
z+q1<YP@?`s@=1gh)F%W2!g|m*G!$6{BrmYMDF)rI{3MPE5}tsz{AS<O?EkwueWFeX
zv{_lJdKlk|;TDK5M06Y<I5m)Y!(BCAyafDtob+%O?Lp(Tw|548E`)4sX~Yl!D+J$n
zeeVC6PMc17;a1RQ8>N7SbyXl=O^Y&UPR&Txal5MEAq1lRiTn;{g?64*bxYT$=vZtZ
zoFz0oMj(9rMw>$+E`trq;VA@7NYG9>ZzTyE0oW>&mm`??PB1xl6EDGiDTBndth4q%
z89M(z5suu<xbS*6+cw0%5hMOzG)VLe$M=6Zi6&hzpJ@mX7?Rj4g~&^hg=?vI;lu%5
z?`$_9ehZLVUSLeby_~XHI`;8fUXx^U`}%{ce=PI-Xe)Rk|Hq66^Vc*`Wc~Xg*aD<J
zP@9`c{X?RL0m8qnKYdRAyuWb&9_t;kA@?;+k~+7S-~Qk4NBkeuD-tlaCV3(csu{Fq
z^^Pz{dIa=wbRlMcE~sJXFHUG6x}bf|vamB#14$Mb4%sIe7#MIMsjqGEvWD~R+?<8C
zWItw0CTrr;Vt!)mWfxfAB}^b5Aq{UDA&U=0ngA!O;X+F*tBGG9oheqflJL;j9FueC
z%r+xB)NLs^r!Hc=7k+hej^mmwD7mOJESZpzZ?U;$0#`&&@D4mZak42|TQA~-f-?HW
ziVo;}{l40$6;pB#VptHeYv%3`Bhs7r2m0(lJqU^%z-ScxO%sP%2SA+b%rXrH2e~o6
z5vEf~ox!;n(;SPe1eZ|jopHa_GgOrA#4f}}Q8-{BRw742I4deD0y6R{W)ej<lKWg!
zl=f&w+uT!%C}^H8kmx)$OyQWl_^qIMQ9Vk;U}d6lGFPNsGv%D^O$#GOBc6NY?~l8G
zbAm~5N3NyRiY4YLg1&CKvQ>?k-B@xv2ySo~4o=#1cw_NE0$4?ei~8(lpy|Rnzu+tK
z;I`K3vYU)BH0VYMDO#UTSz3AwsxymfLL+Q`0Z?G5<>SSP<&;us4_{xf-)=g)K)-7M
zr^~5dKp|Q;o53JLgX*}n(!KFOY^RYw$PFN&!#8;n868Fr#w{W1Hbqo{2l9Em5ItGe
zC+4UNWRH~sHHBqILZi{WYD5NtKC<H|-v66|f`T+DOm=6xc=2Y+dukG8Ha50#coh&P
z`R?uz>WjhaVJTKUY5(56qqPxbAPQH?G!bLGoYq&tI2q4+#z3gU5NLUq>=afkNGI;b
zMZf;>nnN&0C$5R4K`y`ln-gKg;7lg`%dr3>jS;ytvP4wmVG@o3x#P{iDrOfKvmQ9<
zTtm3pz>+9oGA;aTfU>5zvUfvR7cmrd1<0~5(h^8%`Oh;lUQRg&$>+529$?Xa(52x?
zm2d{VW7vg?9Vp8H7|e*@P&#NM+-Hf76IdH!AJ^kEFh^|Bc@9tttG^#rLary~TVY^t
zAyP_;Tw4<bHMMM=;%A^9zB%pw1}Ge|O=x@P0cId+CWbC}k7>vh8Q;Q`#j|H6q7;xC
z5p`YD#Qwxn&=P>Q*y}kp^-yo?EQYAX94$&QM2-TGfZNI&sRaC`<u@O*As}>Mb{8#%
z-Y?Bw6MDwbk-ce(fZi<H<Aa!~hb8Q$Y-MdtkE$NtQ<hA`0Ce0aAu{-#tSS}w@@QdF
zAmc@joY|S5K1Ta4?fdSy|BDms)#g`4+WvCIrh_fXysna@LS)YrCRZbBj~rFlIf`<R
zaNb6ly6>f#EwqQ(P(|;_guTCk-tAjCMA^9hIXioa;IgO0sWX7Di-Hz($^$ND02T=W
z2fyW>-MiydQ$mtgk#!&!C4^M>E8QE98x?Enp-O98CuUo<IXvZY04#?Ss@D^~vVNRH
z_dpNso&T<*iG2J!s^;T|N{2z669)!KxI8Kv;xRzf*XxPJ3JAx)TZ$w$0jcDrZh)`H
z=;=Aw@fS4&-#|=}Kp;R0WNYA1D+LgPef`1T|K_!E42bAK&dIkj2Q~=~8?y>LwK)~F
zloQAk+W{bSe&$7i1zQh5qcv4k`M?=)7!a&CfBzg5JJUhEAZvm3>c;2tWVo)r9;=U)
z9nENZC^)EPuqZbA=r>a3igdOEG9Zw%Dw(qG?hUA+4*22*^l&%I-8>|*CYx7IXb>bp
z{xP;PMy{HC%UVYQ_07AtZ`oK`&p&SH?k++Ng7X%R7`&!D$KHx@L(-<l%_4JKPmddD
ztC<9HV*rO$4^m$;v$Ml|c?jo4MmHR`AVShbs*9vIGwttW0u@M=I4l0*G|+@p;=yz_
zR{tB-V>=m4G5yol_%+d$!x@s_v<`JkWb~&l8y<q@ci=!Rphm+Gn^HAQnLk$G(PoSH
z6rnl;eyMu2lU&D#pwrXtz4Pl)RFN8a+LwaAESr%`i6jFkhmn@qE5vPmpU*fix>0Wc
z47oL&{P7b)0I2p+kK!<dkC;06Q(*EuAXimC4-Ydytbor@^wdOa>!P;7i%7X6Lgd(>
z1HF>MpUh?ztyRMy-;=GG0c-Cf+%y-A5hMUfoi}c(Qq#{=U+c1)#L^;|c+VI>D8`Rs
zaDhht#T_Kh_~jU)9IhI>7<cIf0?ALDTi_#mk5aM1ZGSW=s|&M##Vsr{CQiIXl`0c-
zVH%iK-SWojpa>Fml!A9tGpV#5Gh&uTJ_(%k4Y531a~`p2le4YM=cipsTGMDwhkkA8
z%t|eu&W=$-!RpjIHO#jcGduHA#y?weJmG&WxZekd`67-}JXL!7P|(EQ?Br0wv}=M%
z2iR%rKmQoD@}>Z7NB01FTa@GTuoDL(zPd{+eMa^%3{bIVfSRH`0LR*+7(bvTo54Cm
zj1|5IOd4KLh;Z)iVC#n|e;y?!A8J)V3|(TZ<o_E?;G?i!{ci0DW+4hRkP9)G2=5<V
z2Jm=1H7U@Tu>r9V0oujoI*4agd!2!8DJUuemuWQ0ba!{h&_n_-v{jQQ3ZjYKLw$Ye
zs7--mjC$MI*z6;;!BfA%ZG|DR*V%Ezl?{UUZ`p0cXL3&OEZp@{!NtR=HYABpM)62-
z^r%K{v{5YJ8CREWSVHQj-0Go!5JGu|V_TWoTPy;iwe@hm>3mx_G05PEhnGI)HKUx?
zE-K8lyTb{^9r3)Xs#wpSot~S+@L1l@A0%?l)5w5tj3<l22YghZ3)Mqa*eck;V5sI7
z6u<*t|MM0$RY+IR7Qq}OJW^0ls*K|^GL-C^p>BbD1ZQ5bWE&bRA5L!^qTaS`n^#^<
zbUvmoAmn`n&D?u5U<-`j067BL2Ivm(r>m>Wr!x?&%|0M&Pzj;@2N%#FV+#4scoM8N
zW6|`$AHo=pa_siCze`IX*c0%7UEpI-oM5Zs3sEm{Gwn&BXcu9}4WN79=X==x2;v1t
ztN~bM3cJ7QG@4_#|1*A~=J62O#g3U3pZ@6|)rCPxZ<A8qn0{N{H!C$WY0$w|e?-gW
za+1PI&T@`GZnJf)wvxJ+c2KC+`w&Ghy#>qaf!{+1dMI^_^#XlDef$E|YU{^M#@960
zG)HszTQ5?09sWbvA3YF5=|dCpONWkXr!Pqs$#sD;uT`%<9d|A>sim56X4yW_`l&VX
zZsHlk9l>hxYFRdBw)k-4k76g>uWZM58)=%ZHMv2YYak7}s+MB`+xB~_M$3jixX&%=
zECUH7<S9ltlmHwxDH}J$*5Z}{n~z~WOky041B^+~Sg{14sedpTe80}*7f6$OTZXVi
zP$jns4P+pcPQ!|6B(n!>%}P5_A<%v?Zh#1pFyAama6|Es2>~ajn4l-ZMJDHNBb)7m
z=Zo-w<%<9nnZ+fprq+U(PSDZ(*S4@NDIOx?2u!xY<|ooO&j5ap$p3HwJP&H-pf7^$
zwLarp<To7@C=*dD?)w}4wTDcB1&H5hs&v^521|a9P(~hs%oVGz^!s=HJF7K=35aWW
zAP^ul3ryyp;HJ11r67dOgIvV+1=5Q)vTyK$ScJG9vIJ#5=ZfOukASs8oz`<+ZG8uR
z!zzY#m3;q>#UOR-zcdJSpa}wztSq{Q!Ie)lh#@XsFU@q(L&sM?Z%AmU2!iYMr&jbp
z*KyibYqZ&;4cRMrBT8WnBX})Oy9a`=IN^^;#GDR48gR}LzT>DBfT1Gk0PuB)gDpls
zDevXN;r}iayd^D>)8Imh??vw2*aSd(5^)*%6tEYz*>gZ$Giqd^n*hM7Z`6e=7I^u+
ztE&)4ZLWD|_oL1$XPBAc`3o{Gkt3gV0=5YiX;X?eW)<C=?s$l<8jcObZ>Zz_3iYS3
zsi8+t1&TAC9s??XktOmcf_bX=*+N=eyag{E(j|Bh#Mp%>Gb<xugCb%XAqVZ%a{Glx
zION9FpXbIbhvJ0C8X`M(A&mAFAP}^2^>lTGH&!404Smr9D1=Cs0EWHFl7nUx)*sEl
zsvzsv(-okc#tRtoCZ#za>lx}oK}EGOpW?)0r*Sk6w*|Ou8GtJ~kag{;Oy|y};<zMu
zD;PksTJ{0Yv0k7V;S8VLwx4J)L;pm4FRcldKH58=_}r-+^#>T8sGd234+81*mqrnI
z>YtF8AbtGMnw9a|4Sx?UKbJMebG767Ux!+TtRh8&xw;Qk#b>C>_d1?Y%gecteRX-H
zC_}tLD#zCbq-&l<6JdU%xhDo&gT^+vP3(V6Y6y34<RnKYy){`0Td3i4EW9G;A~z{#
zXt$Qi!^t;oKGTm<h%VdVlJM?55BKh}YIYAXjp=R*{M5!6>)aLk;%nX4?t@N>aY_rH
z7rtqbx|`6pSFIGKv>Y-&Xr<YiIa5`cubZQKnSDRIX6SfZgqi!9jo4FKGWSi^2O~zg
zK8`sKtj+EowASjh8K|XB@8dq2S^eIXF)c|iVl~3zMTJep%7ZZt$|BZ?FS>dZ0z&oo
z#7o{=Jsl63w^zBP6@Rv1>d#k?yz9JcpU-BUX)bS#{jg1m4Evu;B$}Y>#0xOcy-NL8
zWZx$V)dI|}SFl&8N7KRDMy~f(^F~S>j+^p)`u(J&u)Fq3i!NbNL2JEIP;A=hl&!n`
zz*E-(^aJ~aJeD!|?C$-HqId4@HDn^IauexF{-<w|KL<(2dqt%feF;=BrRu3YvNDD^
zux9#(VeYoR++1XIp2j&w1ekdFMh*(89A6gS;@b0wzcYsZpFXRiZ{z4#*!5}$eco!y
zv?r4kOe!l2f`WrP(LI2K;B6*1OvHw7qAsS~wyCST@|Q^IH}0cj%?~m~%P>DGV=$F{
zO)EcHe4zin?qs(}P4&HXZJ+&6!5JCyhP&IY9gdQef3_=)rG)N2qfg6QCM&%+rCQ->
z)=Kx)$kU#@4tpZ5%*lMbE>fJzt5=xgRL$yOh<~t2>X9`EkwP`hig5|<EU-c|C<rl(
zr(uEm+75K(*~!gaF$U}EgSSFY#bJdMw39asXU8KtOqUrHbLh&=<zH^qgGBQx*#?TR
zxUzMe6k`+tMh#(fnE%5<z6(fy@u59I=kLKQEjX1nLMs&M+jJ2P4vZ7c=g*xB<1jq^
zdd~-M@9!IjNm11OCVL3MQhYo=CjR~8uqsWkfKNM|i*C15V+I<gDW(clR94O`E>iis
zAkJ#Z$uZnNp_JPTgF=Wxo=5FHunhqu4JM2pDK+TGRNQ5ChbY0v*LQEbsFKoQo+*db
zCHX(xGYLsa`a1(r!+hH~WBW_;$oB0optCRCdQtle+=fF2!Cyhaol+iiVpllwZp^z8
z4Agab4NJ^z{}kh0pl|uygxHPfi2P~2|KI_;lag;_VxpkU%$)^}r!`v!<#?Rd?L%B(
z>c^RzcyJrS#=e~K^h@Or&}lvQD{Q><v-!I3P0xz1Z~HusK2O^C_KLn`G*p_u)k?2j
z#v^b<q>+1KkL2#-^mp^VSy@+2Jd^y`txjsn(sd*MoQGiz-_-{?B1Qw={%(!JO`Jt0
zb=^@}oh_f6pK5I!-m|uSJ$kO%dZnn(zN>4eU8t0hs<I-}0KKPQ;LxLTmqD_eKr`R9
zGlQK!-{l?aOYeJO@;Ng8PTdQMgC_4E=@%Vxl${B6U1g~h_Db5L%~WF@U3}@?SvgSR
z$!GOlN?$teXNZEMJ(h}Qfvq>gE+j($Fwl!#E_sgh4~2B?qn*#;?+r1CDTbX~CWkFy
zQ&SU2Yja8|NZSb5aP@;r&2^}L5N9DAk7WXogT63IPcRE2PwfYB1Oo)RR|7rQwrgiS
zp!Gf&9vRsH{0dSPtXEq~WjJ7BL{W%PxZ>#1Ke2cf)v!3#a*6~LiMoHXDe0)CEn%tz
zruvg7BtO|8v$tD!8aqY$f3^4J?^wQH*9{`1R3t-@%tVEdDVb8{AyEiPWJsb&6eWs;
zB=eX#W0ZuHsf-yyLWaz9=Hd3PtMB)DkM}QletPfY^TX#5?%R2t=RWsdd+oKfFY8l{
zwdW*M`|xpBGU6>K@gNg>ggsL-Gf7auL92jEJ!%{ZY>4AvND|$zc*`){Q2;vcxawpg
zb|cyUJlO@&3Ghjv%AzQvh3X9@{rdVkMB6xi%BMOEk|1poECxetg5`vs6}mwHE{@*h
zQ2=sBD0QT4z(Ng%2>RYpoM?pZ7jDBQhI7J|jzJGaGc)#9T0(D+=npFF0zSavw3;_z
z=LTsnKny_=cF@=ZL!uxjCys&|SDF}<X&=PqBsGZy!E+pHZ#S>8jf{<fdr{ZbRgB9^
z*YqI(IxLUx_=CH+JMjsE7_V?~as8tZInOPB1SJn{IjPx?p<e;FN6)iwZP3#E&-57X
zvEe8xv@md&*3>9AOh9$R@bYCgmm(}Z791CtH2otybadU0Z&1jB4$#^vDkBqZXa+77
z25!d3f5GwIx)*61oFxv~gc3|v8aF>WE8bu5)>>B>&(L-_%AUOC!KnpjB}_r#+!Z@d
zgp2co5rdoj2Aa8hq!kSf!=X<qT^G{}Oa@F9h9BF8kzC=MIpv?MX+YWc9gaUt($Jvs
zUC`Vqp8WN1165mNBbyHONV$~ur+$VZlfi{o=_#M2RP@=<o3U!ns1vF$-EZYvB69Bd
z_Y)mMMfMklOlb;3H+tTY_e3)~RW>zhw^QwN-AA#$GnapCH(76M{TM|AYYN$^zz^Zh
z#nb(do+%yR{3$Q(R2+O*{pY*43<aJg-lWO0qEh>fVuc8c{gz^a*L0ufA5)Iz^~wyN
z%r&klnCpp;Zy{@?{4ER_^fOk-_y8FEfcXUs7NKd2iV$*g&G7Hl)<$vt)d|%u0a4u4
z@xU+{=m}|QqX3XFGY3Kf_8^3U_A)Rix5n`Z2?4^OOIX(5I*MTy5S8-sg!9Xrnwq>6
zn|y;2y^`HO5dtJA&QULdvZ<w|g#-_n;5~G@z$7gXqxjAy^d5$X(CHz(#&D{U!Xw9p
zZ7i-&$lD#ckbVCw`x3%FgwN|sWpi2cvBm1bgdh94UbD8w8G!<*(BZ>gJA6JOTSRY7
z_cWm)Z{cU><7?#G(``IVTW=3^2K#T06TS_xgBMlKve6>>nZ!i#9_dlEn+cf^ZUJ>d
zixuU+P+)pplw|1NY+xhT03OV#RIzU(C)Zw<^q&A9X8aMZNo3d<b0c_BZ{Na6FaO;;
zq}lnn6VS^`$a{PgNW$FLYV;{ld|>h})WRP8os-U2NW(9Rb;>=qtT)_(hRCh%NPmAC
zdIk*oAecFAon%&gsnw574hWEQ`{(X%83Z)^dHC!Q*@}(53{wW(lwmL#fe2ssi^52(
zvhw3!@}t+EL<ZM$eIL)m8F`(Y9Q+PIEV5>Rb0At{*P=%W^3NOs2U?~+!Up;8-?sxP
zg772|dT4dvP~r4J*$GeRJK|CbOE-XCorY@6P1zycgf=SRWFv$Pc)dd%0|_>cmyNCM
zS0_v;scWPU&>cL8&bhlM8(de{0pkHoMolgSSQk%!?pHfRced9YAi@F0PE(d?Q88;k
z+zg=<FJE7P2u}k77H4PKr_|J;UgeTot*_!dZch&zs><J-Tu=`s!hI-&ZvLX&(w@4A
z_CrjJiHeHv(yrax$-pkLocq!%S1%Q7BKBtv@ZODuU1};Sj2OSIe%wsJn3!Chac?{u
zmy`r@m)JPLBJ1^tjfy&8{#F0@8PVv>t41D9LnOf<2dZVmseCCd;gM^p)(55~FURu6
zZ*l4zwDri{@bx$_6XOswC8NiA)r>2E=H3an>uhGO&x7KdlH4QOgPkixGX7@&DSpY%
ze^!&i=0;T+|J^LQ;)8D6;Cj!?$;EZz`VWEQ$8RDdnDbQxucyRmY+uMth!S(mhX2K)
zpqmQ_CiKWfrP6&v!^3w#RvLY0<m!5IXh#uw4%B_o#zn#wZg?C^D=C0O#^4bF1Vi_D
z4Js_unPuWz4^dkYU^P+(7U8QqHhx_3C_{FNkwgc0s$e@+KGMW2qbvY3Jg6%&>M&!%
zGhbR%u1tVSoQV6{o!J6kUlnjvaS;u7YzEN)U7Tl6pC;6Yf?Xo!Fw1a7fxbEcHpB)W
z)W9wT8OE;!S&0qM!zNo>Y#20h{(OYTUu{LjYY-kHOIv}Y%fG(ws}KTb;wC7a(4O!}
zMpp(O+V2rqE<ojaFO?!|!{v?AHZqC@dd8nJja&A&8{CDYATl`&!Lko|HBW)^1Tw_J
z%nTu=XnfE_&Fr}D*?_ksXfx0>{RqA%IA@}|_vIF8EyaOanG=|bSn;!pe#>A`ecT$F
z55R#TC!wUQ+<g7R$ml3RG{q|DZnr7(+?LCI$^jvWeK8whYG4l#t_4z0p`&sn<?IWr
zbg&pwadck>hli_|jPZ?zxmUiT76$jT+GG?Amh_aAVS*C~8)c^sU$*U7Sj<^#dqFre
zD%-CM8<(JW0A)to<?@T`np#?nM=zg(nzUZ(1P}|998#<DC4rD4hG^~?`|@9iyI6Vc
z&9Og!8Xj-|b^`kT$KPUN6rwD*U^-HHI|BkzKzdWkX3*zHl`d7(!Z!TXl_OOdO#l?Y
z4dNLXQhXs;mzS3I4RHvzX}DUEG>_q$M2oM!3dCpK+E1EDh+ya85>e|ySD?0_SO$Gp
zLYMxS^YfP0R>Hg(yu<pd1^}wSV4$;H95^g4{wiucIXxXpI6@?fT#8h%H3L|+-Wt0e
z@IR>?<`5zpg7hwtK1r{2R?&WJ+q{8=0lrW^RGwU-R+m*fbV*G`%gS|egr?{n*H13p
z9sKD6{KtfH=C<Bmf2~!JpkOy4-}Trp$nV8F-WiD*4eIEZ_jvj*)v31Gq||B7+-~|Y
z*YnqB?b}-UFNItw)`+AhgVs~6efBTSEUx8O-lZ8bvS9rdG#jMdlXUaP6!UkNa@w&&
z+ERzzOmP%Puv4Et32DR*&<@^){e@Lgup^v^{jR5R(9?(DDC(46U6}0Rf0t^Ti?}5!
z%J{gHst9W=9$IGMRt6)jFFjthRaKU_A_gW9u_+_M<$7XUlx@*K)qQqKQ<GM5mrn~y
z1{4UvF0{yA5fKM^I2l2g&dK`n)FdXt7RGpfk3kWMC_QyN88Vplr;H(D^!dvd@B%^Z
z*}zjpm(Ouu2j#9^mJEmcb6|iAh6jo!jFID7xcS4<)cyu^lg4Y`gM54m>>1D%A@PSR
z5QL=2USC{q&7Eu_dvYsfVc1-U{kX&mqKz%QJFPRtIuQ#;DA{o0=JuuMUM^i^;(a?;
z7)K$dSdiMQFm>**leTpN5-e$#Sq{rlsGM=nhk%mgj+<h6GF&x9Hy>~G_eC63vcu4^
zude@%_8P#FzDu^i{9EpNbuk4E^!0u2p;S;)n*jJRuq|z6#WheUA4BcoqQ8cW09j(}
zhQl7IU{1mYGLs~wA$(@+zeq<(T%eG7i0iwDs&%l3vm@s^Q}yvtj9iB~Yy+&J45G8P
zjWyQWFEc9<y&ML_QBF=I4>%y2%For=Xnllbz09nvpaOYrPzsG_B6BEKhx^&#y^i2h
zj?Z1hb2-i+bLdbn;8sBAXjPg{^+G~YL0#cFI96xt8B6d0lQqT$%iYt#qd6Kb{($aD
z+LNn~YYR4N+#6@)fTs=T7S~Mcd@;{88y*{YF*NSmB`k8|TIhsPN#{maY`>RMB-62h
za@scO%*?x&7Rx71-^ondP1Q-*>nbsFGP*x-@0r-JyY{<r(&M)5+%WCksFxp(bktd@
z@Y?JA=r&$m@0}{LnQB_SY~QoD<gmK`{yR3YT_(<W<M$!;ftMw-`%n2U94$TW@vFlD
zZ}2#p<tM&(O1|5-ki|mJ_3en<l=7jZLg~R=;eM(f2GY$DhQ~%bWywfK@uBd4tE`ij
zvd#saHcA^*OH_vrQfRLrn!2onqx0+YK`ra>7Zp7{C$v01Pug~WT%H@*FmB)#5cuf6
z{^8->pI3(GlQBdf)c=ENdfSj(h!qT<26~<eD66s2YI2SOh*%Q#MW5+Z#jDfu@{EgM
zP6mbRT3GR3KPWTw@dU|v?;6sI{oMADZMecw=i856KOXJ?*Aft|;!R5I(4PlARQ#`R
z`G~k2bZffNZcOd-MlO!hh~lnjWhX*)KX6t*CqygUd#d(#lHa^3HvQsFxKig_g??Y6
zjsxfXNHcfpUw)v$omnQb^&CC}0|T7rNLf8$FJJxuki+QIlUV6K1(;QRK`6L^Roauw
zXs_`~0uyhksuH9s;w{O9L5~VJ`5;td;hKCLc?r2of*>+fqodJ?$_h^IXVlc_4W6j)
z-@A7&-&O9}*<PZQOY%DcVSx{IGS;XUx$K8WMu@C-$N-xu#b6D8&J~lP8;*_%<T9pN
z0k6omY+l_HD3EYajuy?}t7PG*i1@S)Y~5oy^AP&{$QiU-HzC$S=}SWPFcLO15=&EC
z8%CIt&&H*urNJ3E%rTA2+$++nA2$y4Ml}56aF77|CFnD7LI|Sff_8x%77BiT-rr~w
z;q}w>_=Kr=51;m>TVx-e1kfMW@c%eFezq&wBTI*Bds#xHa+!wb*NvARJSMm91rE74
z_%EK|8sv0OF*lZJ$g1tRW8WX9J19l2$Enx3kELziLW`91Yr{{r!y<&u=jX~#S-Xvy
zjA_QZYR<O#Ly8R^QhyVsD0}1sb{4xW<{E!mW76PPjc|1sP4Cn9`%`elrb4LX{9@{4
z@?Tj+3sO(pD!$%d?_a~)V45Xh6(UV7O`YXz?DvRuOh%n5`c{91zlNs=MS0rKwE5d@
z6kOYKS-X?(vHC{uE|K!7dhi0Y_^Q9@El}28$uE~}eahwoo+Wj}fm87&O77FYA8-_P
zya36FnQuo9G@w*!z6gu_rmiPELR<c5)?a-=XYlMNu7gxFe%ZBYuj=CXEzZbnFMsqy
z6Q(7RFR;}^p)Vt^GDerD3FSH?{V+z6D@Ih6s~Ei@KY0!HsyWvG+h{Ax@&e-9KJLwG
z8X8j|d0Gk#MCN?17ByciW7}?q4$hpxRl0hrp3lcGF;R7dy|COyO-+qC2`I<haBoBk
zo@4rjyz(=IqY?U=n`PmqQ7OZt-S0N_b$0VcR;<SDpp-j|k7R_y9ZN0-?|pet5vf*u
zZU8Da*>r84HJlpdZgG$*6%_N^5h0f%+MP#Rd_RKIN<re9UrS-Su$B-FIAyWtFa25)
zL%+G7>y`)iN4cAc(-OAorxqv%JB^N1XKcxMav9Qt3yDol+6&{S8I_CAzcD<P<_T|x
zbB2^NI7{P66v!JRQ~O~28mE6V(MKYKU>5%R6<MMH6392_#DraLXbp^i3dVer)3G2s
zbpF6|4YXTi(f|Snxjk@l7;9j?h3;D_7&BSU&#iSYCcLAH&{6p&P$w1a>V2@dNvqka
z?PHX`uYW>eLf#|c<*c=>RibB1Rjjoor>8%4KG|5F5nv<p)0NQ{rn!3ec@g7n^X|2V
za3w>f37XE%@SoZqkGu-yjko<h@H#2u$LWj1+m3kiTj{sYlr4%*EtcOD$#N7KT5A24
zFeJwE#@;iAJo%%VsZbAdmDcvoV*etAhV3Q?DMxp1Y4jBkYJ&Fa>|#!JBwf9*^9Rs;
ztpQh^KfmV^eXPtyf|!B~cC5^#*tNB+j4%>JZuc_t+uf}tzY`%A1CAPwDK5UoemVds
z10(z%dp8+8GgPhJ@GpTBz3$m9`OJRIEi6rFS7PMo+ntp%FX1@(0XG_VKwE^)D>8{x
zPQr`${f$E%eEO=9rq^7xIifJ~0#52-axG(H5(Ka6m+Nl<(XiA=|Bl+UrR9nO*?xQB
z`ty{7gM*Pr)F4p>bMnM<BF?+7Men`5b*mSW8!d}ORC+RI<=dUjDu4nJ6-t8=?)-!I
z(m-G(2o=BXuhuSiuQJW{51RxvS%_*+7*Kj%`)I!-HLx{nsm<ZmEsk?aKib-olag+F
zGt0=Mm^~~j2{aZepJPaql%BEhsxOkCK)0A=t}J9^Xlz_>kn%LMq=bZm4c9bkqq(uY
zy1&o`&Mql2ZWrl9J)~t6AkNFfgGsw!mtTHbQ^*~3d6d!87)Vjp@If_k^>&U8kN>gJ
z-DE`age7m1^+t|`alS~{WWT!jn`ItQ;X~2z{UX;iG1V&0Y}&u$nd4RW@1@Vgm8F@P
znN1U-0|R#-rjSJsSV$;9rqDPVV~7%W_gPt6L(}@}Ps3XgXZrXLCNgQL-K$%!8mchJ
z3scH_aDc+KVE*l~0;^2k$)m|xf;!2(1?d|%*1|0mhR^&=`^CE9T-Ql6d^C1QO!^4T
z5m7w_<p(ORqpk)PZuGLey(7Q2U1T>_%;0*(kS)pNq^qr)w!vZdIX{Xq?=__ytqZ+N
z{k~uF&RxkTD+kW29KJvOM^0*IVtEf)6x;UE481F+`hle_s4U?J44)z;x6F|H521m3
z5oym?D?2DFA*ey@2maQMnD2uw>GU+M$N4i}AAe&9eZX1){sYH_KX-_0Jyaw@K-u4Y
zLia+~byL6!%v0mxj;CohC|`+V+oBmhQtApx66`;I_pcuHB703`Lm@&UaO4Pr4`a>W
zzlG=4n!cz1ZpMi}*5}~YVH(Mt+|a$pH&0!2D;NTJ^{?F`+%Ksrcaxb~S-GGmh;1g^
zy$ju4TynYvzX2qqQeGpIBY1h^#qIaSd+uthuC6r1c~o**zc~|UdG4O~j5vN%VDpYt
zLFsZ2>)#Qg*UzJGc=jw6*2~2Or4TY?Q-CK>%>>pZEhUxl=1s-;9v)JP^7KwO@Tigg
z6@tbMP)vcu#w#Smu<)@@fj%{10eLRqPQEWbGNTjdt;bJ2QHr|JbtVCW+o&9=hxZ<&
zrlKN*hcWGN?A<4cq-io=@u@wQEpX@@%m}S25m8*5`czdogAZ{=_wFGehbnyzNH_tB
zn2TjM!Mx6GId@4`z2v7{+Kd#x+j#>W-kbTn^*TvTF<kC*PvTSgX5{baww||mc|%nr
zGxN`xbCwKeo3&dFNAhoG&tFb^^l0g=u7pfnr18go!D~weeYSeG^!=jPH%&Ti{F!?e
zFScnuJ*;@^8ME@{v!<$VUt3+JG1*hzf9kB1Q#j{qsN~-~U}R)8>n-4@Jpf}HL>CH7
zV#R)7$Dt*t-B^MvrgRv3KpFx9mQQ3b=PES`$r`Ssye}f3ryHIJ{Ym+H)A>+t%|iik
zOG}YzVZ!5!$l`_$5%u|;l-xl)&&X0cmr-kn!OKq^&kAJ(##eh@%dMiunL>Mur)xmr
z81Ur0hS<0H`7ox~PmSj11gzYmRp*vw%=r#hRexN<a9{O)vaLhQ8%Wkt-flrtz%-7r
z-uGo+?V**dgvA?Y&YcT3qn`VyICAo|2%m3>tnWt4r{veKolq=b!ZHdF%1Y&3Y%<jE
z&sZc*(<%Ud1gdUX7tWb$n?9ZpeTwY@ighFp74}OP^z=9d4sus$CCI!n=}i$sQG(?2
zyU<&kpKpDQq;&V!>t`HOlEw+B-;~3mB<-1A^PGiXf<gGiQjC{K&nPM5ZoBOpNk5Rt
z)DY1m#@;qCD&+k?TmXaqVx>su*O7GFa|9pzq`psk)g|i5wmoh3z%D0?`R2nKc~8HI
zzJC(HQsI$W)P77&T9mTSp{_;olGq*<9Yq~ImHZ@<Y;z_p6HA{%<6OB_*x5JDDam90
zafsZ2_rS6P0JH-oq%JdfRd~VVs0J-IMRk*)3LqPpo*4<_!X;8!S&6C=nKz)mmZ|S3
zSf6K|@W?r0@Qi6)pr*A|FTQUvGBkskYRLXxJs&A(6qCC^F5s%Gc!Z$cco=8!Qhw$#
zoMqWkAFSb!#74R5=38sI+hC^(3q;8LS2n-P%{8m)Ck938`Pa~cxl+~NZY0WsRCh4r
zSCwz~D{&zqjSEH|^^4KB)@eu=F=dSXbGAE4@TyN%Muy+XR<yb-=x(gBoS9o!S!*LF
z+%TgsDhK{4fG1>3=TubU0=RINK{P~tPzB6Z*>C^%l)Es5^dlZyC<wRk_QGMkq^9P-
z@*DHZT2X>7<TCgx@3rI=6zpSsni!l@v_;l+Q(zNYLNp*lNZ1b-_XFdp*a#Y6(J86c
z(pZ2Q^|50)z)r{G8etNM^eTu%yKYU9T6y@mM14Q=F4MC2k+YGW-?s4OnkQPc6u5Mf
zcb9H(3Op(IAeRfDI1wSdUw<Q2CV?YiT%M7><+^2Ky{67e{L8enr}Mr`Os!lHnw)1=
z(ch<JWN={SeHP=thg)c;`sPc?jU$7#45vS`Hv5<VoAX9ii>4tc4&Vm?4Bfz((?Bs`
zTlh?>tElu^z2i;*DNnmh932uo_#)VPlCD~V1S_s1bw$E@Zp$aIWf}5ReOWHRN{0cN
zfVSQjiyW*iwnuZ%LvUesCs%nQL4gNv(Gy~DDBJf=q@yh7$*nKwhAP4I<4BX|Smoe#
z2bchq#P88(^-A&38VP~VGS2{GtLfRCiAF<4uIrhvFNXJnQa-I=wlO7mGX(|3LPYjg
zWtxAdbnug<t!*(uk{xS9l7n7=qOx)ie7aAZcsc$R-aE(eOMqH5CqEI(0OGGzCT?WZ
z1==l)dJPQ?weP*;!Rg%1ZGWyQL8+lVzVtyd`fOHl-jxY)ai-U<RaZAzr;K8PcQYpV
z;@n4R$%r;?{m$Puk*qG!2Ub;;q~U|(TY_oNs6(mX-a@+1nHCN{5Uus=ayZRV9~vdo
zc94Vfc+k9g8*2x({p%jXy*b`m8n$<addEsk+MY|c-FQJ4A}$#=`J^L%qt{(YsdLah
z;_-!Dfh&c6`Ti38b*a4RBWaG8p7XqU$#y78I9O0ZD6DJl<zw-Yhtu2SrFMpQ<8EHo
zM~9{1+sb0Zi*?5b?!G}DL67{3yO|yW*m?A*0)md)J?|hYBLD;I-|opHAV9$5panAb
zt00N`@@3=A)su+ZV6B>7C)|T;dH3hCh^~0xj+BkL!_u)Q^!w-eI%afH%fF*HiL8AJ
zl8lC@WzY4tmB8vIDioWdv%NiAzy9mj-d~n!Q>Ny>ZTLPw{gAZa^8)uF+7jfE+ZqTH
z;9|}8wk<<9&CUDCySa^9xP4$rf8AN6w+4)uzJUP+cd;KWEiAI`ZuKmbX*rEg9UTg!
zt!DqVJ8*>llx~ur`dlkK?Hg&>KCm?gwbj2s)qtu#FE<zZ-#5X!@nn<1&@WPx%BNMB
zvwQV1E<Sz_BjeZLr7yvYXSr{-9KNPp2M<%yqssmB<t`@+t=cjV96s2lnGhWm)C9c*
zxv6;Lix>H?25hT81~guYjrIn4f<zJjjeXwNc_QzMUu$GL=21;a=7|*hY0vM|&On%e
zUt3TNqjW#GXAgVs0V?hVHbq*6&I134;Q0`9F)E%_%@cpW)do1#DjcD{_w0IY3?;uj
zzq6$;YK?aawy0UnH7-aB7b#2B#x#=2-8wOnk@#t?V>xrwNbe09O`U{f@1$Gy--Yzf
zKB-IciU~=X-ral*C8teD4X#UTT5;^r^eJPC;|Cq*ID|-CV*?!rcFhRyWO|s8067|-
z{0)Rh>qPE&G1mwP$TfsR9kB0;&F<U-qUg<zE+8MSa5=e*0EHRUD*}lO@j>4a{L52M
zl`Gjdob7mL&pPDPQhZlpB^6Iv#CU;Ijy96J^-7?E`<4g9U@%<9BAboD9h%Jcaha|C
z0oiJ2^<l)A4(=Fy2f9n+N7_3aM4GIn#{YUqOr6@hLXb2%nW1M{dfyNlyb20rE=b5D
zWQ?B8O87CcmrkcnT_C{q<s<Xl(^IlDVKi>gXGM=2YI5AmJ<zeJ<o6GK{`^$;a95Go
zhJci_iMl#;ABOy7G{JLCy})eVA}lI8(HOVYMH0@g4z0;nIh{))zf&Ma(FUZN_uz1E
zZ(02-Y+KynRCh6x(ucv{oKVM|Q8X1_zIYj}`~?`fR-AnX4;opk<WT7SQ*4(5T(wm@
zE!#0G2^Lj**?SNvA3lD3j=}HndvadvY^5mE^^j)G@0V}$9!V86rRZ;Y(dvEcg-mir
z4&!ylOI@|;zxUS*&>SfWcMkbe=>2nt#k(Ms_6iKoJSiy)Dfie)ofHl)zG3q&hn+R$
z6wi|LEX}U%45i1mxuwzucqJ`DZ@%;0vf$yy-~&L-_AfSR=q_nURhu<Q4Xf;(4eK|D
zU}w~2G2?w(?~&7ZXezBG+2^w41NhY(;zP>?xB|mpl|P@pe)Y{s-0a8*bv8sw?uxA=
z%iCUHc`@GMmop7&TO`9!#}R6#^N<|q2)+%4&Z`a4M>2w72i}3XgjU=X;F0_p7`Tz?
zh?bGYzWOs^;vym<O8#c&jnHj_EaGK_6Bv>S3oQFM*nxJtD`zeQKJNHD$kWG!jN}Ft
z$HLu!b<&zP&4H6Iz75yb)I>n1Q9J+w(6JQ1xiq3{4gRb<?iXDuwL$DkSzsI23W!~&
zgw}-A-v|(M1&?1=UZ<R8oCi!%Jxaro`elq=Yb)zU%Hm)hv+LjM2eI?Ow3Om!K@|89
zI}Zwn@FWxkj=CI{lw1XKuum$%ExS|?0LcCWZ!Kofl{q`Z1BOriZ4w&K#-;9RCfXM-
zN}7HqbH1#vtemXX?*OXFcmkS=VEkV;Rg?wHf@<M6^g9O!xyB=sOG=`pN7Y^(m3wJq
z1UFmsQaeoOaE+i>{^IremZGQM+l0@E`mSiOu-5*lGwgh|)?h=4>}_IeLBxo=bFm*!
zWE+QT!ANTQg6!Ti7e@<zN#A)#*QGyFBD&-pS9QAb^Hirtg9OD1cIngc{#6!LImsCd
zYe_?B;XiDB)2!9Z*lBCi{xDnZo>*KVAML9wsu*9~<j||Vxvz?tF~x0K-3etYGYwkU
zsq5+J9Bx<!njmG>#8IEVj>k_gaqje5BBjHNmoG<0MtoJJW1~}MzWFp>-&IjnMa@6q
zfWio<mv5;75HB#us$H(YVhUtQoQHyHlZPk^(=sxy(8b_bp+4(T8bphJeEft8q-<Fw
zo&8EUPMjz};Eqeu#vZ$kI<rrAIKlDOEl`|<=7Y3<WE$PYSp&i4Ek>>&Rt_0>p!nlE
zlu=k8`@*QLqpuGR%dxxKy(s-~-eqmtk76wK@ZkiIWd5Tgi!h8F^-0fMcAUVa_k^u0
z`nt(NJE*9j@E3N})br6JMHQ8=h<c#de|5*GkmCsJ5Qc;>lxjR693Byom*5Xx-ChZ`
z5#;{k`|tIuO|Jgp;pIgxT8qfwCd(@A+S1;B2k?Z{u2hKL;={Lo1&L)RP=mfpf3a&}
zSA*X!foX~bBM9uGn-@!}3F{joZn;AsEJOVPR<xnT>Uw(g(si2q`K}m%ccY{8WiQ*l
z5quOZ<Q@ajAkt~k(ND@rc{$vxI-u`hB%8;b?It>%$WwbGw}pj-U_RP#wg4HI<eV%9
zcCe^H{x#@4H4`cz<QbqcJI{Z@OkUc^2@S8S$8t*bfX|`jkJ{FvL*O++0&t#E*ZH)>
zU6*pq6meacn}W*o=FJ4=%aB<kB?mN6KqNqdwV2Ga8?W2~cKr@YO4IT})c%kX=pljG
zi|!yYLg0BwJUTi$FkXU(oA9nRC;)`wWpIwh|6t>RFIS3NvcVi+Rgv?bb82Q9dr7Y3
zsa#q*Cc!3Q4cq&8XsN0mv&shEVxPAief(h;I)%@24_L%4E~}sC+|HgS5?tJ<+BV+z
z>M8dgjgNdTuQO7{uT6^Q)|_#ciKTPT-AFz6RFL7INi}c7_vXgG)<dEx{cO(Z9ZkDl
zhc<IW<p0f??4NA!Iwu`^{AIv9=APtxF^xlU0X;H;nS$rFb*(eMf6RBk&P|=1nPlW*
zsQO8N`;7QCSNvV*Yl6=KH5|X5C{TjH1-9?i23fDiV<T$)2C^K~!Z8w#-fTV?+Xjas
zJJ=gl<8iR?P|l8W3vi>=izzej+NYG%)X)+N5!ma3Gw=>G^5|(3!#{bGC8#JNzm8?D
z-e|IPIFW&?lb?s@xhx|*wA-M@PW*zLdYowtav=I(vb!QIm#<%oS^UOt(AO}+wcXR_
z=~Ktsw+S9LFj(=Jbc|LZ5}`s~M+gLXp;F3^#_UN4R7A5FV^WxuC`HP2^lWuVXec))
zr>O2zCHO58#{;w3O%wpo<_S)V?@^@e;|I6xz^#2H=DX;7TmjK1ou1sQla-W=7!Bzd
zz%WQLz8jd(@t5e;;WZ<bjYBa<>&ueeya~xP@3`=+BAj7B)|<zGQS}DFsYgzUK{tlg
zF%&5O)g;nYQ-Yfyk(6qYxL8H2vpT?=Lu>MoJZWgpJr|c{9O~2tEkufut|L4U5JDZ!
z>2ep;ZlbJk7D2X1Xe^*9r%0m2spYV5X>04jpwy@Po7oplUSW<9iaq!P2%UusW#8~8
zX@+!xtin8%P9&<T-O!6Rz=s5LoAGf{O3F1zvS2KZc6m=A)*SYlxO=hK`z<s~O57wM
z`ja+B=;*h<`J|?*dLF|mU|f2#0R=<ZjB_wHHdeEI3B4<{!&dN~KOwyYbpJp1k={s1
zvRvLPU~o`C{(R=v6F)zjlshbPv`C;z!8%LKyM@vzNj2>A!5=8(mgWB<LmY0D`(Sg+
z7mO%d=)_>zECo>dN}=!p(!zX~qTB&=)%rhksq3F1JAx3W++K{X*A)}nzK<t1(eSWK
zdv=-V>8ssN6V5xQ3V_g}qG3)|Hc#ZQ-VDF<YTJ*MH=%a~O$BV8%a+Hz_L4uhyY5PX
zvVCaEuB{^qO}-BfkpB!np3eTZ(P}1mT_ebjrsQ<YNHSdsAJqRLdQyX%2JO6~Fj?m2
z=00#B3Y8~FS%EW5O@55R;u`0+d%luIT=+Zmb%Y*b9%xM=rV{2@#CQ+3D~>>ARn=ay
z94y!)>M+Kd9YArol6T`8S{!S$$AASg6||HOO(DAG=jQI&&_GQVaopk?Y8@;(Rt#gB
zS1_fNnQ3fjxPlFX>`v>o7%aG$guK3@vDVhM2znQp92A|foMkX+N38qx>m#_t7)ixH
z4V)9PYDxs`Kouc%hI>_uddChBx$eyl09pk$3TPO!IQUovxw)A5Nk<G86m$nC5~3QA
z7%9)N=9UE9L$GP)Uc-FC`$^SqvSXM!!a3G9kTL^11U6TfFg?o|rLzEH8HC2Q;vb(;
z*3%P(Rz_W&|F;WtF)v@9Hy6`?tr{b<jw=e+@7|3=pj&RIp$Tzx)N?kg_SwynuU=vU
z%rGM>%jCmc4}SwlW=P5L;V`0`lbzkz+zd+(Hy~NK&H%QiDl_iif2r{1BU$H;w+Tw1
zT7rc`kXEow`}_NmA7i=!ga_mU2PAGjv^GdVQ{Tpra3%)Y<2KU7>B)xls0%~^%qJCo
z_+V{i1-+Z#ueWj00=&5iI4Q<yY-B{m=;3~x8a#2ZVzBTj!y%Z8Ua}{FN&z}4CUVlS
zDR6TDauGdt40nbLRs-7sI(~=@a&KyPdYy<xf7SwD9_WFj{w&Zce+<up#0F=VfpfPo
zy^$MBA`Am6Z*TWpf*qRuulEQ7rVxZd1jKwgQUfs96j5Q2lb_eu4^9V<b^}0T<i%HF
z_Qt?{JL==HMo<>}`O_8gf%q&(kEXzG9wbCqhT@pxXBoS$w>%?zLAXHnIbeO)TCZ|K
zaIk-Bs-aq%p$dT<JsU-j2A?n0)jX&-0@ZrX5KZim5Jgf+kPiqMZVTBn*%Z7WRn(tg
zXuQ8rJ~1;kkenpYCt0{=zMS*WyxAzJ#xg50dxBJOV5_swr6afi$C}!<O73{E;w$sx
z9^LZI-DJC?81G5Vyr=8TdXu#-u{4yIP+ws6L#e6ieCv7Sd~63P)EcmjKYunuSO9eV
zc|{vDm}WKmo<&P?0G9<PPkYD)3u2M{64}vFHnvj(VXbJbASx-<o1=8XE1`!OX5J<;
z-a4@JH3-yir$XY3<RUP>`Qs9%8<r7dpsMJ}1_%(H$MWcKh<X+~4yhmG4ykr;!6CsY
z^3c?j#(HAWz10&MYA10Q)N5<f*r9{zcjp(jA@lS{Ek#8ObiDi&WF6*ULV5C}E0Buu
zrmK-Sn*iPf6a}MOP@bSJ!?4ON45%MtjnBwnp{1P$WLop4F4EKUGupJsB*biga;TTU
z#I+v;4++OyCJmU6t<v2Zzwn*jvs4Pe6tuM_z$5#*?>V^IXr~YGF8zT%D%>n?a$OD%
zV8YhMow<P;0Q;_t(XR3P_d@3hriS;^2-7g13+g(96rjE53vwLDn$U+uS(}<3vu;WG
z^y$>UIG_4f^6dHZFAEDLHk8_-y3(HidKm6;bq0)iad3G73H8aza-x@cGz?DgxFX97
z$rNcHqH%>72v>zWxiRQ7AY8a)3c^#a91jpI+$)tUxWkqsxNmT^iHUo7Y=Gm?G-L!T
zjhpHrAwP*(Xira$H(Ie+mpyxCA&=%8b7i8hZx0<^`56XVDPVC#g4Z7)i#v&z^hh4i
z9ze5=w$;Dc!c!&na%byvYjQnk)}5+89T@M9eSwB1H-!c%ay1=u0`nbuz$9jz`(1vx
z=&f0sO_y(zW~iQ32o_TBZoKW4vrn0V^4pj3$2x-QyJ=59&6(xT<Xe#LE&5?;E1T|a
zCX-0gSr2Ow?wrE45ykHo*x0aHh8mf$IL2RmDV6ok;YJ4s9K_jKhG~K*hZrP&{n36V
zCRn0e#-HCeu7cnM#?@h`bl#H~?hOoafNvS>c!QolZ%o!;?T86{j1lZxw?v8-j;iSB
zID(i>$Tp?y@E=!xgR}~BqLe#!yv@n!HK+q$7=$BG8a2#I&n6;UKfd!4Oi}10Or#(h
zKlz)D-8MQH3Rjo}06?~^5rA`qg!@%c5PQRA78Mhd4eTILt)S^3v{6oUvtbX+juCv6
z$e#x<I;gW>e;#xSm)a060F)~(&bf#sAY=MHdv^8qI~yB(7*o1JM?}Y|-OJmUF-GF*
zx`OsXhqU}K;dP6v<NbT_V`FDdOr!vYyff7;Eh{Tqe>)bYA>gt;eM;T1PgP9~<FSyz
ztZ*T87XX|+RAaJO!iiu`{PPsLLgEZB5xt$^-o>Z=;|@TvTySnp8E!%p7>APk``k%n
zK5)pae_{wQAH^x6Ze^vL<vr{bpyB?{o(*~!<H2Bkx)I<5WJ$Chi-|u#Gjv}k4T->t
z{#l-Rs({9!t~-hG@wn+<;!apvS>fZGjVeIqOH|v^w~KOs--0(NA|c_@SMIx3u@~o#
zTP=JJ9U>%K+o>x>yHnb-F{Zgc|G0>#=<a~vfb1bt^LoEtfifjp1vBHm#$fsWT*X^I
zNGuu8E)@6ucrqLCXS`^1+k9w~AdR%d+v(yAX^&`n8RlCyjcBM!e6h`+&o44__Lme>
zuY7-DxoVbC|Eiy?7|E{Ei1gmr&Smo1s$_}ebXQZvIg5?4-_-Zl#Fz9TP=yv>N@wI4
zbi_2reEjf<B?5=q{f_g0_G6Klg-YFfV427fFhdPa9d1e#MX=<q_;?yFdM#jq5uU|-
ze506rhY}V*Dp018^l}dU1kUcW#!Tr+bdBK>gSldiAK>gNDsB^%*Np#~1|b7L#<gIO
zsjA|^g6b@U6hcrCf?LhszvJdckv;*Nc#vPSkBE#@6Y@6rBBnQxlz`obOMY~C_*GaK
z1Ws3$m$f=n@WWTWNIH-%bOeOQ$M+G*pg+?*k{@J<=$yUBNBa2j(tm~>Q6ytDtfRxA
zJ?qBG>iLDMJ2_CI@aiPtHt>dCZSNPUA;r0_*l98cKrFy+-lA_=e|Oo^4sQyghCL1@
z(qyE!$h-iv8yOgcd3!6tC>PZY^hqHeo|cgCC2zMU8!>v@ZE0#5)KcX6ne_vPp7BCU
zp769q*^3MSzX)SpsC^LeVItKxIcaBWdnC7MUr~6;yLY$cL!K7o<vsQBsjjJ^rliDo
z<EK%Pe{X6!j@ie8g30d*kBf_?H%4N_+l3Cu(Xw}XX>Kdv88>D$p$GU+NF*8;pTzQD
zt{p5~sEHwI`}FBkd;5DlEU^1S*el*!DbpZZKsN*RFlq@Mg44db@Fp|U4tzx@g`IYa
z9kjHw1G(QFXAdbIm{Vv#FflWGpbUmmz&PN>LUkl*Uu7Q*y)af*9)ju}h{7~9G@PXm
zQ8uKEYt3+{BQZFRH-Zr-ib}jtKe5b%9+&m{wFR~a(uzNSDm*qP5yo*LArJTKh=O{b
zns7MdJYml(g=+`d9{j?9CbU5F91<r8qXdcOvo&@_i?aeLhZx#bRfV4@356Vxe&k*F
z9SBBJ08D1`t{Gjr1fm^={tmOV<A%k(4<{kC&G1<5z@xxa9njI{ewf_(r*ZO`3sZc9
zI_v(9861JZRHxY2uPYA9EqP2GG05eRkO&j@RUHtDM6qfDc4>C@Q5*+U!)T|2RS$d!
zeHTO~<jc_VjpaKuI}8jrVbCjkr~cgSN7W&)mMbXeCMHhaTttw^bn_{9+G<u|{0w$D
zHV4{8>;VjSTXZxvFF*!mBjHYC&yk=$tHB#k*;Lv~m?750X-^CZ@Cxx}d3LmNzmw8$
zU!fPJ&oRjHd&J0Trgq<_?#aZx?As0Yt{?K9^BVIs_0J2X@qOd1DJ1o+G<$mJ*irW|
z<i3kBimvN6v~v->TX$P;YX7;u+<sj;J<`wv`d=PKWvg2j$R$jk>{a0g$!GQLqQkY3
z%kj4Hr+5ZR8$9MGTSF}MXZbg5@4YmRtnW6gsG?f<LE>FgR`XpxV0_W!;!4~txpvPt
z-!Dpcy8e8deXK=9_q?v4o5`yep|{p6oSeOM1DWXxNEn2jEH1H(wYqG1)$+Rab-Y7*
zW_nzP)<Rsy`wvS7y?INKOQDKFuhq-Fgo3sdt%{2H=H?!cF1&-COz%Nz^EmD1CZ)$G
zjw)ZPkx53wq(KWH!nMwVBhqpxeCCh=&^Qfl?SowthN>i_4NhIWxQaDIU1ImSILHez
z4!Q~7HJnhUqW<`yspeiGhV&0fcTOns*D8!SyI7WN+Dg8IP5vMrD~_XDu_W6*TU*<k
z0<T1`?;EP>ecC%PfM_qj5(FY5(qpk!l3NsL>$TOj=&Y~cD-qTn;*?}t&@f&3xEa3n
z$W(m8o_F3CmwfeaF!5B@kY%Xu_C}6x>^0p+rb_zy85hbrZtNP^MF=~-^7Hpk*4T()
ze~)d37K~cHtVuGUzvEXY*FY#RaXb2q%?G9b(b=8u8|$~1Df&14PS3S&<*N)5k|Zs;
zn9)_)G$;8Be0A`L%F;uva&w5z&hDz*yvc1d*x#FeXJ0PcqV$bV!=CEe4uYPr(e~v_
zKg3R~(xL(*P`eQ{p-^JJ;mIV_#`Yld)IcIILb!|&>`~p-zw{VGP&soZRtO3U_ff~`
zOavUo`AXW*mzPh_y34tBlUv!pw`v5<UMZ#w{7@)vC;g1cqLn_F92W=GlFfnhgC(0V
zdT#E=Wn{2v6WJvd1v%Z8m5!2oMdKFZn>H!@EU;x>?;HeVXyz>XfE-gta#JhNwonif
z%HsOPsy=Slf`At4xQk}feIfD{iA|esM4WW8#)YP(wL5<RC@SDswfE;t0wB`_2vldo
zDDSF&b>F1PeH>eaDA(OH0Cg}BaiUjn^QX21Y!Vc0uaz%?>}9N{SJ8ZF3lr@+-;2V+
zWsHb~->`fLSzfMt&w;BGc7fqHAgoVn{`Yq#Wn{QPVKAxtX%_LJQcc!yOAwgi&-K7v
z2o(dDx18E*#^LQH)!NdcxwQun5o(DDXAkjX9~!o9x-UTdmirCrm~_OiX>Nh|8^nYC
z`r<Zinyw&vqTbJ*?R26+V9vI>O<w*mfqWBNK6V;Gup^rhRV^Wz5%|c558<D*2sK`B
z?>j6aG}wR~rE5dKQ#NlpL-TrO<0;b6A1y_Yr`u%0_0|U8_@CEv)^g}`{`WONG4o%2
zLVRHvtg8R{2ZSzs$p60n|Bw8ytk|EV?#;9{2y1WNyoqoXf`tixP)tlip%f?QHhHKK
ziySfa(jAFvnr*)+F}35<(1^}_a#HuU-(GiqdB~$Q1t~yh3c~H1w2X}W4^eK~B%?7;
zKac;^is4!m|9?gQuw~XzVgJw5Ws~*!A4e!C9q;P<ufr4mdDiO=p}_v{b$^X#JDF!v
zHwD{vPulHdz{dam)_?ytZ2{n@xwX}Hw0M)}ssCpaMISYRgzkS{|64nX#G2*D7C)gB
RQAND3ilVwg=IP6x{|5sL%)I~r

literal 225954
zcmeFaXH-?`+9kTxQcDmqf`WpIh=70yNRVJDK?OmwWCcm0AUTIpF%U!~2@(V)C&?KU
zC1*i`2uRL3=bp<tXY{z^j@$R^zJI>2Z|_s1v<Q2z^}bJ-b3XH3=fzEF2?{bAG7^bI
zaqX(89Er5Ul|-VH`*SP4GU_Ctgg<s@OG=26Hi`eeD~R?Xk&ctDiT<Tv6Fk;oAM9_w
zAu<&hRC{C4$L&<~%RT%G_ht9J@H|5n{`lj*JT=)ptY&9Y(k0(E=`oWfo^80}YPOJg
z@W-B1F^L;NFB`dMS9R#RR;`@}3kO5(rCX}|me)on%pH^J=0jIQU)av`pVHIQJB^DV
zk^Vj-|8qO>mru#h1@{ntdBotg4S%@&D(@x!^ElrOAMxj(FaKlT_EEt5z1n%;wA7n(
zcP^VMZz1`ft*EHza$0?W|G1RjXG)5!nCQrJUYjoKN!m&}qw?Cv$35>4(#$72#s8TP
z`u}#B|FwU6=6+qMP+-H{SX)}Q<f`TLUnG*RIQGd!d-kV!oft(E4Y$F~pOITgHx;nn
zc<nuEdzAR|4+*-R#GjwN%eLT;OAr6`?cfhR?AYrQCr*Tmxl^sL%#Z%a7;TWC*T}OP
z_x1IqpuLd4*l8~xExo?6I!VrFv$`<8IzQSHe&yLy3fkrvS)I}5MEvuXOTU&iIX~5M
zwE1%E=2f|PrNjeYxQ%~i8+YVp-1{DPD@EIt?9jKyM(fGWkf5O4J71pmR|Pn4EDxFw
z)^?=n%a|&!Z}J$_zcXq{Dq5~F`H}fREm}H6HRGOQ%*{JxFORinnP_F+Z+o<ze4yrS
zmQLBrM<Ow};B7ni2u|kD2Jsjv<(Tz9XwMW{8G2tQI4kv8Vh8Ev!$6WR-C~!sv+1Vw
zw*7c5jI|w@S4-Eg3zc169BN>=xB2(Bou^No;x?%NHQ%CDH{8DQ>F;fe9oAYN+S-&(
zGZjo6DygTUKjW%}EQadO>zb)^PBqEPh!vzL#w*FLw#3NdJq#P8Tvyg6tc$8*L)8@%
zZlxTj7Y&|K{T?c0A0-tiSLC!F^WJjeC;x~>R+p3g*!ju;jw0M=IqhI=$lFFXHnq&)
z3lCKB8hf-j(Qy&0CSW=8)y8gYGM?-Zi{ScP<1<S7TC4fk=uo#_il@Z^!NEt%C|OC2
zuJduEOQr&|)m#A_T29OT0fSB9mtD!6R~o}YomPe&c6UZf2Gogc=3jZXhu`w5v^2gL
z9W4_^$MBj#Susu_XE^GuQlhG5gBZ1#rbfC!L!}=pL;Fq&nuGSO?(DfScJm|N<Jf!8
z`%+R;YhUO$KHhIj-B@1@U{z*cDEDJk-XOk)?KQ9EXR@k`VNr;g8hC3We#7j*fdjLg
z$4OI0z0N{gT(lcAAGEuYQ#Z#ebF_rwVhZNINzQUFwdscmSx@|oQ;Z+@k)~DT<ijLW
z+xROmFwk~n)95B5utn285u55Aa9TR}a9b;W@UX1m<;#~RI|~?St;brYo7GHojn2wP
zReK!aD)pod@b!IGb8F;#e3n5&1mf}hhFa=cKZkS9R5AIi9X06^QR>cIxbk%OHEHRW
zX|p<BLjKX8ckVfQD^@O&?_!rl;|<P%A+gtGVu}c-qRsV5Z}YF34$~z_h4Kx~sB-M?
zbZQm5`+aAucH_5isXQ-fd6Q(qgl&^~!yT4qoHtjy+S3gWY_fIaS<|1peMwrm2357^
z1pamX84<eJ_@Pqgg*N@P@mi`c$tS$pbIb=tHm1GQjB_Fp*An!O7V{(DeO|w2QBHCg
zODVIO>~#G0;o440I*<H#-70_Yyq59ytVvgD&F5o3^Q<RhSmh4v-_KsS)Dxrfpild|
zJ4fMCQr?7{KsnY#z-sJPf!+LdKb8=?k=Qz+MfJX&%D6R^G`$j$6n2>$Hqs>p)Q@*>
zBBP@A&<n3+_r8q3mD2NqDRil&XoDv@R3q1-V)Pg7h5OmlrL?p>_Y@Fe;+`}eHoczs
z;`1MAPU|Zz>gG3szv$h)+n4cL?(XonNFn=$1m)yT%NC7AlqJWd?q_c=Jor&yzu5id
z+2Ul^<`+tl{Mp)z*oWbtIp%!|8BdNj2X0`aUN|%ow>_TGBA$3+#32j!i}W~~&&bI+
z9sZ2YO`!PQUyoN%gSwy5PjnWnB3rAFP5u5@y?+1lP^M5DZE4-FQ<K<;O4j82tsn28
z?z|LQXl2*P6`cNh;G*?J^(Y^jp6oOo)0Ru2d|H0{BR<wD*Hl+aUAt!3k!zW`;k;a>
z6=CIsi<seFYSD7mKsCF+j*TlosK=}auxkV*noAGH)1j5g7Jaj+Ri+5w(P5Q)yn`ZQ
zp)Trqt+HoVf=b%SutHRZn2^ooI%jdKTK{C-SoIvUCJwfltIbIodE$;cH#QcF1{G%4
zmS-7Ooma=R+Hy;l3M;Ft^H(SGW|x%yYA`EgxiqZpbd+>6Slr|AG)k1dk;6cseuNc@
zglXswBEtF{BEy_FY^Q#GmR$|A9SB5PzSQqyuy;~lYfex((AF9w`@Z}zH!A$diJa(Q
zvq}~H5WmFf*OKgr`RkdbL^`g`RJG9{mgq7*Vg*k|2Wi<2N0yiR$OtdD>Uh-*`<L&d
zv+eWdySR-c`Wg+SL=UCQyZvv*Q>PL%3j{_kmj^G<xR1mq7p{z@)kSEkrrxbUvYu>5
z7}pBUR@>~TTRSvHAMonc!yksx9gd3~94G;UVa~(~A-B@g)2Ez&)EEZcY7#W=E_Us5
zn7wh={;nX6WyvT41exxg$T_lq0D+%;E9K<9JG*x6B9=8Gvn+>3fQ!p~Zm3~;?hjJ=
zi{JT8PIZ>gV(8{wd!U`BUOt9Fsgf!5w32L^9Bv-T#;6;;N7l+~dn^6f?0Y?VleCI9
zaOE@?t?qXfI@a;^dLX{PW7V)*(;hd|^lD<p4=_o4+|OSbiT~nd93vN5$rzx%zm4sl
zMLYJnyfoKxwDcAo?)i0029HUn;63{b^v-Lh(rSCRlD=L;<30EmyE0&5xVf?J?fK%x
zi)qRYq`=qgAo&mbEwx*cucJqzI4l;Eb9_B=LtGI#7Bk)$Ez^}@q?}pGq8K+f-p)H$
zCdpBdjy-EzK^fNV%(wNHsgM_?RSOZY>=|BI9t@*yKCs0_iwzzB=6&s>;zuky)kDWR
z@_0+XJFYH>J)jJW@uMU%s(hUBGcG*WYFt(H@g#m=gga}BF<w?iX}wHj!#gxIG%*Eh
z;Cf4K{6|LbaHeT*>1BsAq^Z$UC}VClpC9iu&sCV(c0ju>xN`XC)25~-n;kOOue<%w
zOjOOVG3H3ITbyw0cBQ5oW*`dpayQY9P{(gvy^1DOI27T2a?;K2(t|#QyhqQ?nT><H
zPLN2PfSoQ3@;;|;%$V|&`W8I&CTB?tw)l2!`lqJ0hV%NoY`gXN56j7d#ZapshTr29
zS>>aodN%D)TxM6E1Ctc3FSL7)BMGlxy}C4R;;fdUU2;D_cy%l-X3*rQpv}33v*i)L
z3^YjZegXV^@*YNM@-)pr{L8i%ZEmbii41T*=7@eU&?ERh?dwQWto+?x+$xIZ&u4UY
z23Ydg*w}7;mA>Jx?XRN4oCH!+Ou7ouMqk*a-fOBbRY`T!R6#rSdt)s-*uRBza62~Q
z;avgrZy=KDm;eDl2xLY1!ra{4{k-|jnJNxh2?LazyE>jUJS&TnVbLck>4lGG6rhvX
z(nQ}>z;dGGc#peNvAGEdv7I>4osd>rS@w9KE^Ko=t4q&FN?Mxk{N1Zr#vQ2{^n4F*
zmwFuGr^vYd`B8Z(;)(Zu>p6PYEu=rLAO?@75MTt+eBYDDs5zM1P_Ai%C>)vQbO2WS
zGX_yvQ;(ep^7ZA`t@K4m55yLuj*C7J*;wkG<(3YAu6`znF}h}^mVYpKt=|+W_&X~c
zumKE@%lXX*zP2(=gjjdJxKA|X4(qOC`ixkeBf<_#NO)>}XOhb!9~2LXgyuvwqJOTh
zE{crAya(v8ZjzNCwphzyS_~_Rw<e%K^UQZ;EwS;os2gvsyBxC+!-rUv5Vv|>bV9<w
zd8|p<CzB5iq-%7bk6SgX==WO;iSBF5G1u=q5{36bweb3BJc*X(&iQ~%8TsTDZ2yL>
zr+WXqVDqcx*+HU~q^@VS8>5rm8l#aG82<2#-f3xdvd9^Y7ce3~t%5{yJ-t$7wUu=A
zNYI4`G&+kZ57n}aJ$JuRN>a~3zXntjU6^c&ZTuD~Nq$RHdh`-eL1aYMxz&1!{^#Yq
z#t?d^3+-}={Le_ABS(&eIWDO5)!G3-YJ9x?xnQxAXlBM;h3iA_y(YfLx0HJ`FbvX_
zngHpeROR))yr^vrOd;QNDE0)f(R*>vpo`YVnG0XPeq{=CNCo%+#9PD1^BVuO`Ss~<
zfZ^s?`Di3MviBm;ajLY%cG91rD2)Wp?*?WPdid*7W+}pOxG}m5S1C4Ek5Ye1NJuDp
zd-!L2?9Ka73sVp^>bn=z6|l|u$Sq(xow-0^N4uoKz{2@vHQeJIf{@NOmLLm1GFWf=
zkw{_$nfp~vgiyHy8xu$VRPktX5#9W^rRiRIIk}#rgVSa7UTOON$1Yms<4tUHBd^{F
z6|@E9F}0GucFnfmPkClNA;YLeyknhBJ=mNUH&Cmr<;c^8cmG3*2;@NF<vyP(_Pn#v
z+YS^HQ1bCk%G%)(1iVS^a8wL!ylYi?`C)_|gO}jz76E`~IvMZJS0yF$@UvMaU7<#G
z=s5tm2Ds|&wm%zXM2_{20C=VA)dbm;R+N`#qtZ|2jwYcmJ!;xU+Pj;`<gY;@&Kqc+
z^MI}WO(%GG=3ic%$lh2PC303L7zv<m)uxL@k(WM(25*l72yn<wFP5Z|Hh{{$22k?W
z%ig^h8w7qs?0V@?n$2_%S1lg{uvIjO7DuT?J|9z=Zbfyq1K0<)oy+jshx^a^N<RXB
zU>W9vjFMh#3MwI&Sef5qNe;T26s>1s2f+9zcv{pO|Dk3|$(oM;otjk#a-l@f)bSCE
zxr-^dzQ}n~l0E0cjkgypN1Cq5$aJL}Hm2%U=~chsH0dPhLVsSl7ZU%;0qN|KxcFS_
zNrF1GBx{NA^XDAq^=r>IMeHBTvluq3;b{p5{s&V8^XXYYqryAju8^l<p*-Fq2w4OW
zfm%*-AR{^GC2uMxYYN>hf0Z{`m?s&)eoVlEZ!{r|V6Lc6sCW2+0T4bjGjo^o<~k}-
zAr_fY(mz2x=X}fhFo{H_p^eUN+CsXNPiOP%Z;|z3nG%idzGsE(=AH{`a-BGFr~chv
zND7Ck`BzNPAOJq&P?$3p5bl7VMYpa$`gVq&-xh2Ozek!*)jR|S?E_wln<`%rWK-Hs
zI-@}_<b!X)T<ENSnzqppEdpG!h_7-B>CZ<<CKqikySbtAQY?IJC=2P+J|Z|T)uUTm
zV6Bjr1>l)&?BgAIEv3%rH#N&~h~U1Q!O*<rfwww~M?{>~YC|qQj@?Omck2y3e)n1m
z7*`p+bAEO_iF8Jl&|j`UL8x{X*iQoaWRq|C*H_)Tc)J|TJ8cIOZ>2>1{<7Hu!7#G2
zI_+3FIa9LZkBcneM$vOQiq_0Pd(-smgaL_6mH!~^{ezGMwj4Q+EjV$-EsqR$qF2B|
zs005`D#8DYi12^D{eRU~Uc2QXNNZfSY@^{dR{sfgYdOn)ae}DU*x7dm6}_aFN4M?R
zZ2-JU2xWa;;I93qBniP$MM9nVpOxPJlWJP!B`Ep_Hd{b;-1368LQ34JnKrZ_5T2Qc
z_;uC%3kC0!O<%=WIbF255{-i-iL(gI?Ww2$L=K?KvZ<!WKz;y&qq%V3W+XNm39SDk
z-H`a!#MW?&iw;Xu$E<(u>-+}nLh$G8k#L1pggWAVb#c-eT!RM`5RwSmi&01JyrQxO
zkK@W5U{6l#&a;0{p!t6I@S&c55mgKCrJb<Tg+WbjLW6WX2>q}Ry}K7}!8g%ymOxV`
znuQL_U<#_hOVEvkr#@|?py58_e=<D%EnvV*f7KT?CkOz_aSF`$245Ut+H*{RJ5>mU
z1{e>@Sj{lACoS&=8ekev)Q1mE5RWDqeewWa763cSM>W9Qdoo%yY4c>n#Y_7td?Y++
z{MZ=U3xRyb+m(~l`Ry0RfR?(<>x7(_zR*K%c{O^4-0{<8Uw{8hwBXIPL6P_`qE8>A
z_ur6_sk0;++WdIC^~&6Exc?C@y`QFK$MnM-txmL1kOs=OyrM*JEF8~l|B+=vf234w
z0jTgBE>8rXbNgP^`3FDJ>x15-7w=dex-9!+H2UL*$Vks4T&MFRu}MJMm7{#X#BHVB
z&(k51k+WytMoCFwN<f`A0VUH9v1~^D+)DTHVTaS|IGB7sn5M+bqx|SPkfx{}Ms)%(
z1}64*+E22TyO3TTKm^#P0ajO`ZY4Xf8^!d0Js@+GORt(*%OM48h*nQf--ZY=G^s>j
zD5YnR6%ZGLb<P{+oPN6_zxy)FZ>)B0KF#|1@C(KFc;&jXn;;dcMNW<<zKdN;$D2bU
zvB5i-E@*)wfHxIkfj{jyyDTj<-y|I(;F?t(CL+?}Y`fLP;T2Yi7k5fPuI5n$iv*qR
z-Tq3y_Yz*l?U@gttw7D|$~NttMZX!24%?W>n@lQP8A%8Q();q1g51XcsP8@e-FxuD
z?!)JfJHL;_64D9feiD;4!=EeQfz^#{AiOmw##?DWDm-7?M=RpwkfN~lllKj}otNmO
zLImdf1GH-GLIBOp2dcdpB>W-3-5W$_tbN1T3p_`Wu^WL%uXg%#awGV1TB4HhJY-7T
zjsqnher0nxY?Jw+hOa09b_oG@+MJMm31Amsq@yj#MGzbS@6s%7*5_LWs{;>wa)}g6
z)+~TR|3z(}HpIx7ndB;si^-Rhm;bWZd<0w>-!>~S>iPO2<f3)0T>;=p*XG8em?HgA
zl~4$~_t}&+ZFT?JQC(eqlv7*uFY$NJec9A5aP8sTV{S9oAOX-6oM;NTiUfav%_wm@
zgf<Pmsue4IYNitz!RNBY9l3ibh}UGJy=%j2X{sBPZ+YqqW!8gsrO4}kk33YMWnrf9
zEbH#wyI7};+kRUwS)p(C@Yv4uK@w#goi~gQv1-w<E-_cu*6#Um&FAv7i2_Jj07tBG
z`^m59=?&OV7PO#WBgK_LxR5f4`{iEx&1?jD4}0P}IXStFm6jr+ivrXjqBxL?fR1Vz
z_bjoD)xrF;(4lN;#!(w`(b`d-G!PhleSHU{`R;sj*)naCo(>L;`ak2wnq&Z3#>~tN
zHw-zg3CWEOb(;UKY`gxAtY5yWx;hvLsy_UR%*FBa_b&zFumngKNK^BPS)IAycKQAa
zmylp#C`g&55SghSwt~!XsAqq4dHiG*%QFeAf}(rDs5t?5gJ_MHkFJHJXEPibAiU73
z0~U4W*Al|l9vjgPjR6w1+acBq#y^+7Npe6=xF=YqtG7JYC_`^I6b=dh7?dtBQ5LO2
z2b3~3D8<QIMTG3VA<g<7Tek0q`0)=ACA@Q53FWO^$Es+l$33}dBlqggvnQQ5X2XiG
zE<nhF_;lV5i$;JT+TOk0OBztE+f+57HX%+PdF>``Y6)&q?Yy+hbaQ>NY>CwoVxri}
z97^Pcxr78Q=^ksq%_kGbz}K3Di`K_Y(gdJrF{enp#RR}09iJ$$CKt@bKxnoJ8@;8T
z<V2;#mGu*3-%e6a?%VTy446Zh$Hm<m@#^Z@SjcJ}BsIk;j#e;*Nl8k&=kvDb5ycGq
zGQeGn7OtHy2?#q`xN3|>#t@L>n%CpT>V(Nfu=ts;GHiWB)FZ~I-hV*pNK{EX9h#MU
zQ&jq;5$b!AwGhO4&cY{Wfy8Rc%ge2FK;KT<<C?zOG#e!sP}TLcVW-$JWhmvt(z@px
zq1C8pNfRYzPdgwv_ELBuZyL356w=yD4NXV|bPiJxCMXA1I;oOQ-w1jVuy1fJJ6I}+
zn|VfZ6uL!IE!J56?x6idPW08CyXE@2B5WP)p`z_urfxzZl@;ACQEBBS8I1hi7Qff!
z5tqmwmu!MrqhhHf<H}7-4k`ByFQCgk)vOB>DcVcUDhhp2cyn#mJgD0*G;{;HpoVpa
zrMFS`^}^}iGD<o@z6FJMs0Bc!A@0;1FNK<LjU2f{5%m`yc|&?It76OQv#^BxM1;`C
z+kfd*SDeO%tw2slLuGF*Id$+`^a5@Yos4Ip^B&%0qCWf!u6N)3$k!JK4s0rikLhpL
zUYxvZRADuyvsUT5EV5a&h}u{<Q^~q~mKE&w#2r*LOohfkq_KKt<M~LKm&Y#Rp9>32
zbcRDntY>>VFimQThpao&n$nB)kI=jUb_kBkk<<5@;O-0ZxCqNG$3?U!TeZyl74o*O
ztPBjjE2GIKke*N`QU?x)J}lk^g~S0oxg|lxXLOuk_IG=a=m+o1FmGBQ1}_MIje9Lg
z9k`VHp57v_NjLWr)oH&|rRK9e$5@mS-bA<8VFBt>D2WyTIzXP5ec23JFrf#7L?02f
znF1PGLqGfw6-9&{E+38bzET+axyP}KnKq-<@mM~D1<R-<swL)f3!it=L(5#py`nv%
zGKX;LltMh#g5ql<4zZ&Z)%6y@Ogp$^ua^jTcN-WXpansC>+`r#JSl~jjxX?`k<){w
z5ZoA;0x<{C-5!-o23k62*-@YG-5_zaU~WnC2iN4^3OlZ#Mg<*eh1{9syt#o`ZqKze
zQdRwd^+$cBR%OgZuSLz8XiZ^gi6fg>U+k)Dem$y?91JajDgCx!)lkEG%3}h)MGfcA
zpNEbpghf~~;RiZ{B1-_aB@^4J&pVl*E9TV!HorH17Jz|Khh<9%m*TlT+|2|sfHO%0
ziav?}HA=eQ>({RlvFOmr8hHz0KuM?pgxEu92<;VgZ4MATygjkZ4o$U8!4EGdF&tcR
ziInd1YL5_ecZ=LldU>nlq7bEvAam(_TsK$f1nR4U$3w8ZXW}*xBu`#ESZP;xw%gvM
zOjGA@;S49HsgUqC(KJrx0mPm;sO-m~wSlE#JuzRV1Cp(B#gAEj0HwoPy2A>i!TYVg
z!Xmu?_5yqaQkKxHTwwp6sKz_;EpaL}Ie(@yMg|h3Bvc{dmgJ8h9RU~Juk?ztW$zWq
zKwEmIBuZ!5Buj{TO>smIu3yQZHO@Yfdc|hNe71&naPE5v^9Y?g)Wpj9(wGxXScYO5
z?TYiJJzuA2d$p47F!_{N@MuJ%W7z4DtH~n`!3}sqKXVH`O3s=m^o@Z2etDKx_eT}Y
z710sU;~AqHA~v>F(3<2U>ZRYVpwcA{?bt2%xg{PXoxmY&J7U4U`z}wO)GlE8l>GOy
zDe0>&x=(mZ*)eN%VQ~XM`k{G1T=MX!lT%Q^^e(D!Q~{UKugsriV`H09&Uk%xuH6h7
zgXGT-&i$UCGK+FO)|#R-*0BG;0b=$EoGGrE+x?XG@@jTX$VJ|f-3b_G7opQ*1Ii88
z(2Y^wTzkTau8}f&?M!R*1;`x<hrCg0?U8Kes@}Y}Ke^OzIZezapgEc)#vS+P(4uRz
z`F4$|-~N;AdvV=%A;cd2PYkRz%i4hYp6ohsvK{e%4jB^$iR9Pk$ADALkVM~4tX@FS
zF4f(E=zSPjj6sqk;0}apBGbK(?K#T#>cRRdvf_Q{QX2v-kW1gVEtQdt?E$JVga}WM
z;7<HxvCvq=-to4yXbeb+F4PwZRVwCL3PB%6Zt`b5quQiGw8|pa(lW-7o*3en$n7Rc
zoC20M-KBMhX7!d!@RQBj1tb9V<r}W!Z**QBfy5%MjKSw&GEt)YBC~fWD*xrG9RPuv
z$|0ScBcBkQ_%$ew{WQgB33cLvvllV>Ao6=&prvC6`AqBs%peA;167hSoIJW}@d*8f
z$>{qaDlES0(@rgW90HDGraxfw4hot+`%gbIjDY(D^NHO@2@%ySvxUmVP&A9rDzF&-
zhJHgtFsuP4KhlRGyw75stk<5!j=GN9nl{nk#hzXBIZepoz(vrn94B%HA;auzV{6YX
z-*w>hSXYsVS{SB=gkS}(7mH;BNDSgOG(&U~FvQz_K_Zl1)$BubG{O7KWFJUZ-A`ka
ztIZIC%1wETMl6JrX8G|h((-AqZO7SBLBm^?-7SkJF~GDw(WKCd`QuxtGL7HQ>sG4U
zk7uYP%gf8_Os;E1a6@n(Eg|YP>JZh#bO1tu((hA>72B6Pc)06~A(^EF7$54BC;Fd3
zZZo7;>bwtt;ow;8qn1lnZYoZTd0nnb^nw_iraY@u>?>1%a#wD{DucRCwIW?U13vVm
z&fc@EXk$LPycF1#*I~iPALC6?p(nhxH>SZ;>2l6p7+LH@BRjZJMo2=bb*Qcko6Whu
z9+4L-i|a4{dQAC9${V8C4**}x;ha#@F&YH`A~1)VNr5FwqQrO++!^=XGcaFXnCPHI
zdW%qpCkm$R+P>&2eB1CK&j$O*ixkpY97Z?KC6`Rn_}4S3)Rrdoi0RCeAhLFJwgY)+
zSjR3WOEwJuO|4-e+w6{Pjl}z!<WGx-Ht*K+X>_YS3w};sz<xDppRn}M?XDQInOIlt
z41`Ui^hLIoxXZEfV>R>S?1CG$`z@FCLlr=U+jsa9J5#W|cN&CdeNbfc#eJ`!y#6LS
zTWK)`j20-5mF?RfxWc&BWs-)5>44pD6W}&#J}Y>75A{ZYcmGuP7w<bpvIgJdzJqIn
zQd|I2+Olm&Isos#U^6CznDot{>lZ;f_MW_S`EoIaaK4F*(UYFRFUO!MyZ@kDku0p=
zdDb(ca95<joj=}vsF?2Z*>ap<2v?G4vsiB~X%N_%SPx+E!+BShc4%Ug*n7;#YJgpw
zFt~7p9{>XjVx*ar-Qxy<81N$lH9$YuR2;ct9@M#C5wJ-|onn*mf?3<`Y?H#}0d(|D
zz@<WvH`IK79v+_Pt_7VzyY0(OPN+Z$)WgHW-G2IDICCc@WjV56svl5A(|<&H5p@o!
zLX2*k&O={DQRS*VcH+JCo_pZd(p~#y<v&92pci)FyZ7y25<0fGCuCG_&+a?*Xt@j*
z6V$TU%{1=q{yesi%92hi8(?{(ZgZ^;dQV!035M>t)|wz5@{GRpzb{D}jhNq~^pqdk
z<}WTw-CeqIk%Bbyb&Y^Q8?4qF%_pALnYo1w9}G};w%^^gm4lHn@_7|AH}HnzXkwOt
z^~BM8*REX)#8d-#hVTIaLWA=`U=@nK#@^PKuB`Ck{t=~e&|!AKH>5t)gaSmE2IOQQ
zDs!^l{Sh~)$dzBoun8(%JbJp<+13A@9xa!%yl5}=^69eW;jc=NbW(K6o@O0IVMJv$
z?JOju;L^h)Rp4GP?pK$W|Jmhno(cNR%lp+lzGr<OYezT*mS{uokZn)9QI~Kt{Bl$8
zP4ktDEYXW1bMF;?->ppgpCK)~N!yu-Lef2GIG_`&2QY+I5Hc^o$;mk~99+WQa*dmR
zyA-1q){L56oezc>gc!G4Ga-WjeMSREVJ+QrFW&0L_Ss_A(g`tT$pmwX=cXCQ!&E)?
zp6k+~aw~Ura`hu=0=kE5Or-)auwq=G7`|Kcc|;ObwWX3q`16p>9kJ+wvgRq}t(+rm
zdBBk`1jaCqW1t0BJW_ZyAAxL{`ck7dN+J2v_IDrP7TX&D?DkZXLnD_vIoh3ID%_u=
zVZG{(*P87{4sFj7V#-PSLZQWncdl^|iFrO!r-%t1(Yr8~D1dD4B{*{xUnw4Z31&SE
zis=Y%g<Fyf22gO1xDL=~+;5xlmY`F~e%Or`LCglxke&{_7wd>V8~4cJUijn}otONl
zxj5YH$Q1k9US6~sdmbWLI+;>NN4}|p0LRR%H-{63k;C%eA_^rjVi(I&FSvQS)4#M;
zXr$<2-`tghg}3>m_mMe%l4H8Jba=jAkyK7kbVz=<LcI6~-XyeO;PAUB8Hb(My*@O*
zomzcwhKvEHYEqhikIJM!WscsZ?jkGk&_d<0`LbB;5;rhq8h*3Be;f)a54(#oQ#my%
zhwFKe)j?$$t`z<TBO`RwR87jm=WZ|0^fO4j93uo8y&9s=7SeET^nXISz7!aVNuY-Q
z;U;h_DqECJ{S{{Ljbk73!B2EBrGN-i7e-&_^jx;(mGO^s=IE>JBk6gBI4~5IBbGSz
zh@8D#zI6qX&mmb`+YA(s7!eO3jk;ztR2B)VRg9Y7@1eDPG#jWNs`6o*zksZ@B*DeE
zas1)J;4iqx#fAb}<(%Ex()6^<455`7G%E(4*EP-?&y4A~{48l8S7iedL63t)=vjn-
z)iX?o_~ry=vTC%GMd;zAAZ$!%HSwU%Dr`_dYUUlwv+CP|2%8`5MB07lr;Egh$~617
z>uv$4FqXzCSZ4Pn-1SLX2AMrISp>_F3f!C(`FmQb(5UreHzrkk+k|yWJz~}gh*w?;
z&k#q_{@MPR_b=Vl4RSDLLaXI$u?MJF%L1@N3^50N+wdY1C{$(^gF4S2bjv)N_+{_t
z)UycT^Lj!yQ@5LRhXD<kzip8WDW{A9mgp`sAYIx5UH*9sAlw!2M%3kH^c4PD+Go3u
z><{X1fU1n1E_^SmUhZ8Y?yo(QExLgac#*HIGP}bJ4>=q2;ZhHWaB*>MPJ21uEs%!C
zOz>iA>nO+bx5l~5a{_C9p2dfPx(RPdt9?NchK!~!RbI&uZx=p3NLVEX4{Un~Iw`^a
zTr<Z^_l{>5iW+2s3-r8CqvW321AH<R=~J<*ALj9@y1<iMKMxb5nk1})P;j)z$g)r?
zF!~LJ*NYesH6_+V>NvoVj`q_fR*2O@{$cKO)k9$do#&;1d@`70cx>eff<UenbDcjA
zG!=T^(@S{8ps$Z`?S(>J5fiw5d0)P#7)yG$VL3F;P_~jCE{~?@50QEZCJgrfY=h6N
zkFWsr*|iXYe2}dI1hC&$qR{SD3Y0I_Xk)^bnUtmh&EZP2c}ODF45UxOoL4bC0=T(%
zr9THy&70{qmHqnmeNHTwn-b^9#fB`3K`2UY9a@FVSt6pAh<b-{0o37{gckUG-e4Xb
zOi6L6PcRiIAboEcJ(r%;nneH8pTJh$*6VAtbr>hU0a8O>z;>m>c?yBCk1+8Cn`5w7
z=BlSn<xO<JnZ4EEU2qO+qX|N$ki2n22nhaP(mUD!oR4r+OhvslX8D@UMF&^|R4E5G
zm1{A4{!Z}|xS#F;;i2mi(i$j*Jzx!lvKSvK!c%Ga2|}NElo&35w5w>7*dQQl=sQkd
zpYJ0)n<pndrP|!IKYi<m@v5TP3}l~Lk`cy@ne^}-q>@oMaJ{CLS1{?{7IVVkkQ`tw
zVL=1F5*yD!lO}4Z@)BHFrH6^pCcGCFBPs|CG%fK3^iQRwpD9nCJRwA~R5b&iRQw$%
z;H%9Lh)l6W{LS#<eB0TRc|^qy_l*OACA{0aVnyiSWyDxx(+Tt?SO88Mf*OP~IsHad
zPl_Z_H#tKO6$XzZ-)gLQ1o>!%>@1r05nZ^RyZS7q8rmLqWJ|G#6&8ypx+sHTU<!w|
zG%)2Tb|ufV9L+?tzaJ0^_u5S%VqHA`>X}16vvqG0fF#M<HDKR)?XS;cLI{1LytX^S
zso9inV=s7M+R2?n|GHZTydtrno~C!>ZXNKO?&!j&ZIpd^#hB1D%;d;gKOrnrN%RuZ
zGX;+|v1Mi}s~5PqW*6E`>dM|g#M!O)>9R8|m-*6DuSP6~8I2Y8(0eB@;a1<y%p*fg
z6b5=(uS-ksNw`cRs|9?S(W~@jhB{^1Jc=!#pcl@^I0C3)9XpT-mw~_k0xCu!Rug>g
z;_wF_f@md#LeH?pC@U#<B<?T+6BAPxrW7j40Qo;4L@~=ptwDB0X#D^jfnd-Mnh#DF
z9T|x^4zHCrQ;3D!&`>qLSqsMjt}%ywEq^SvY7d=YD+VEi4H9tp#*G^=Ro9`2*@9wQ
zKu=5W+Jvw|47K*(a+U_I?S$ZiX|o3S3t`d#>L<h{bYzHUSk?jfOAXpH$6>fm70`P4
zdAoO>{2_nBadYsM&|Li$-1jstFApLxZvFwpCRjl*_lPGHKg3I!h<lt%m?~EK@-t6N
zjVO+s|A0iUgB=Sp6Vc)EQ6N!}Er>cutPt7}`~))?py*Mp?+Ye)I_!hRYH*GcJ`<`Q
z26)mA*?)hifV%Op?|La8y*1n-1HGmBxgf^ZR|j-uVIjYbg=s5fE%rQiaRJ7JipDos
zvu_xgQMbB8mJs)|?m_Q3M+~QwOuujY<Q<#6?Jv^IyHnL((B1{%(Lx94-88x3$D*iR
z=Rw{EeH`;6ZLbUWnqsEDzmpvd(ACp>d6@HL>G5!8<!-J-=xKX-Qg;#3G&p5I@JjE#
zAFb@lAgAk4!)f_1->D(qnlN;jv-QFWvYRqgyUvSsDcV0Uke?W`oz1Zkt?gu)TQ-H%
z?EN+a*M4Dv%{GW0(iT7T^}||v(qZ#J*~a<PrAu8&*ezjf0>(<R60HCH7AQDjsbz@f
zE4c}0QrG%~HKN1)Xd>>57=0qGd`8PbnWn?uG~oBk14G4lZ!-tcQ!mR%0G6D+8e*&K
zbo3v2rcVVzIc6pPXH~v|&9@4n9A31U`URR=8hV6U4#gd%axndiHIli^i+;2I3S|jb
z6U4kOzyKS&ry}dWwXNB|AiX`Mq~GXq=O|s$$h8^eco&bzCpsw%T<&ws*mzn@yZRbD
zPayCB8F_W2wpK>{3R8v68&a+atUl-t=N8hCY>nuYFv%qU^>~M=zw%!ELUr?b%qoi=
zzka;)1vBC|EqR1l5=&eiDs<7TZsRHh=?#!wM2=syMS=^4bt0(L+sUcE`Y?Wvl^=wl
zq>j7csuPq)3i8o=Px3^AiK)!;LdT((8H8>D8}$Gth6WHjl0u*v6C)T5cbma+#k1h7
zhWo^M1vhAx7+$}#I_kKI&b~HeEn3QfPB{qQpp*xdg-AzZd)C7=NaaMnNcx{$2C3Yf
zPj+Tg&whwG1@@{PsuEb===XSQM2eP0gBbFtZZui{Mi93O_z7X`{$Z4?J8nwYETQ{b
zz!{DXLU0oEU|dK|n20liABnGt9|u6oM(t1!aJYce4d}HFP%3dbsQ|7BO-4q>W5@z(
zO`Gl;yd6!5i2a(*3NTY|hHYk`lMpH*Oup!|lF$ag4%*Nh&_D^9HneMjyo4C^K$Uht
zoREXJVD(U^FnwqQx`7%oH`ygZuM1IF^6FK-eqZ@ccmPr5A<Zokny=k^=wCN+AJ~5a
zV+;nHOGDaxe61&ya8Us<5L2|6eq!97qAdoHNH)3;8EN<H@0~N;9Xuc4A;P|`U~j3;
z-Xh$+4?sI{!SGomLT$tw$Ux!kS{vZSDsp_iHyZ~=h-lCR7ss7IFj=YL?IOT9#ybBR
zkwMXJM3RB;Xrkh^Sy@(}o{6bJn<TPnnSIPxUYaP~p~z(13UoySgnvAY?r=lhyEg)S
z5U&s`2YQOl2}U!6$Bhu=Jk|u{(0xpRr6DA25EE}`EkM(m=v25W7!Bd9+U)a@nP_-l
z78#&Pqq>j2)x~w=0+lz6JCki`O!@urYgY{4dPpkI`dS^)McVFCJcAA=Dpbv01wa5=
zOn+BM!Ub52z@F81A+KK(7t)jX%x29_ID-(2vO28PY^qR6HvnvNvPYIJ;ZFGp2KPk`
z0A~LWGjBdW|Ev_uV<GnbWmDBt;*RS-e3ycj*I01ijn0gnvYFWon74Zb9m(wx&YqAM
zn=trHOFSMbU>T>Cv+aWO#@dvK5*!^feHA5&Qp1WYouybQUGiRdjcFbp-uWr{37k{!
zo?w!=Mwmi7Uk3%P0HaYW3l&{P30s5uTdYQhz=Qxxr7RSwU}#IrB=zQ1qApbP61iRQ
z$bht6s<4D{Tkqn9J0L7@dT=D!IXytJ+P`7EIlx;~k@@J2jjvmlDJV6p1Am(23<Md0
zSSqW}!;njOP<_>jSj@ch6Y?vjHf=p|gDs~W9;YFTZUmlZEPt%+bvUx;0gMMyc6N3{
zUYKV%0SUOL(2-u=uT_RUtO1UXK2lQ8e!5daOTQiyV*^eqc{@LRD&%ZnJk|*o(ozQe
z1Efn9#9+uY;NhvM3tKN;XpSdN1O7gNwHCMDP3c^|riDT9nF+J!?lKf47dcG*@dN+2
zqgwyJaamnsnZWTtZp`{^0WsZSP&>YfQ!pAo8A&($;L4MD37B-ydKcEHulPTBc5qg4
z%XJ^58z5RDJXmgxkeUfN(P;*orWMQp*&|!7Kf-Y<;(h!6`j@{D21DY5QD~<8;R@dT
zdnetQC^HhNoa=upqZStO(SK48*?+N1djb4bqJd+yn_ckY5&}fO1rGy*4q#3<8x%}p
z#_q9&(0MbVI()3$MtbM++bu}i@B-1S3+CtA*Z`lxi(lgbS}WQN`FotuBzyL|-b#iE
zI}%Lx)_bTs9B_Jm6vtZ-=NU+hl9-U7uc+U8uVG%p<H~ao*7eIMB{zExDm`zGx<Px9
zD16+QKfCN7gbyVKhcZ8x<p6z?3;&6qmHhp)3h1loq1%t1;^dSizOogn2<dvnBMy~u
z=&vYj<dKJQJ>EneKQ+?uUV_l>&p+~kr}X-1qBdVsfU624-htHxDEP#w(7mHuNbf#k
zNI<$C^GNF9C>j>7l{|%bANPOy!PO|e6c*B@7mTp6o(TtPB>0iT&s(1|lHkH+{$2C^
ziVTgni;!<6xkM1rdL{?9HNv=@->FOd?lp{S@I}l6SLcR(qSc_VSz!Q7Eh~g~2qr#8
zv}*Kj7GMLIaqM$9T+=@brwVj|8O$MPXLr~yuk-!B32NuB@K6_EW>$379Y5;wdyO?f
zr||*3!?rN&FflS7vLxR8cTf9ZgzbVj+cRuPTulWuU(%WFJ-dDoqbI0vJ8$kLk+u^D
z0ncp5Tw4PWt<|wdtC@^+iIX@}<+~MIIte^7CQOA6;Y&tr)VC8>Ke2VhDb4Ll_<dp|
zMLKf~#0|d5$l#=)qu3$Ua$;qlA)quc@s{OaJbhZzasK<3mTja<8^6Ej8GQP)<xqV$
zOdAxLR3y?2aVY8LJy`VK<E^~Me}CY~Yp&-jiE}L&gIqoWSMXsPo_lkEcu%rmsnY`r
zgc_}H>h}j}inRfT%Y6*u2a7KSy52#vcfCoRc_;ZQfQEDUo~ro|R%L9;f35^yavtl@
zS3lU&`Dr>b{?C(JPqJM97!dQ<V@B<<4x`BzrBsq0Y_dAn8VtCbPl-PkqA5w*r_=KF
zDMj9`&u((tBh$#oX0<2NtkbyKZwGE)b`&YRRTs){u1CeEziCeP;mSYvOCr76GyVH)
zJz#*qS8U$z*YIBK?eAAbA)~>+y#W8=U;1D9!Iz1VtoX!NOiD^h#1SNT$JW=@92^~4
z6=Hs2V*cybtDqp92s(J^5EQ~hf)c@F51i|tbQZsNiMXGGs+yXbU@)t|ub}o!4lQlO
zD+l9Qyro<uM&Gwl#b5~W@bU`Cj566pOG_&vQUt<7{DWcc!ootaAh5Ea!_r6Z?eIqN
z2?)3q`uX`q-aK=VOiWx{T>JM+N~#*R%i=FzIN(EDU$Jz(?7hkgzx2nmk^CsErQcfc
zA8vbh?tDg=reGYPUiXK}fXh`|AtUnvGxVSnLh5y3Jg~#W1R{$E2S0cab=W((Se*VO
z>=h^lGm~+6pC|X469V%(-Kly|Ti?AR91OQ^-O|iuhjyPjCw7nq`M@>YdHsNweBQ(E
zM@2=>8ydoRdctY*lHS%jyurcK8yg!SMlp)<{*kPYwtnvF+C+9Exl&S7Q!O;GejZ}K
z7mx<DnlR_X>&B+B@8IzZ7cPLq!GV^S$i~XBjpS1Nw<b9b1RTBYe>TT{(OBqnc0xh|
zW-G_Awm|7ADJh1AhR~Hvt*uQkyn+(|b!cQ{ghki&L?k>p<^418VyWex#Ak}1t4dIA
zCL(S5LB(rEJ39smFFNqBxX>@|umIvr0Eic}ib(CjpwTj2eSIwTGTcHqrZx|bC@S)2
zS^|9w2H}c3)YR0htrt-v;A{Pbk(Q^IS4$Uj2%o9;#?Kx2q(F~#^n^>^a<OuZIqFIA
ztf@F}$S}zAqqWrm^$<!K&*`+RtVK|FC@2p>eQ{nQ^1}yMIYq_9TwPrW7cp4BlarHe
zUV?M1Vtg}*8@p!y5}eks3l0qpfq3F*OBr4s`ka1`A78PtS%OueInUap=JN~4!+Y1y
zR?~Vl%0-C2K%hlNM~|*tyDB4-H8u_mZo>O}Z~tuI;+p`i#j+0wX`oXR$4&kGbU#wV
zkw7fp;h$3Pa84!Snb%>?%cp&8?d&Gn(j;1#VNk-!6r%=hM@L6<GqWep&Z=kE|Kbt0
znYv{23m5O1FGlsEES5OUi->Kq%)4OHdEbF}=!Vqg-y46Zxw$#Fhl9*D!mPhiPcXpK
zvj(DGYFgTj5%Te_BCSpK19m;8$Rf8FFF0-7K*?3o`*4`%Pc8{@aaFa^_ROlTF0Hw<
zu+*QM5mz}!7aiv;F8*{Vh0L|Ms;X*903UU@PJNnh0gQ#TJtQnl3!1TL7mN&xIAB+8
zc~VWS9iws(Z#Cz0k};5VL95xc3iD(2=VoUwn)W<-_Uzf`;NCK?$sZZ<=QQS02*!)4
zVQe&agCpg6E-vz_U#T$aZD7C@Vbb>F2Ydo*JuCR!H&1M^`)7vG{USyaag^wb2h$2B
z))y^DeDicr4e~Lo*w1+VjYSi>P)KmFNA+0&a!ZJv=(_u(M%xUVnwv8*5h$JmfoD2(
z>afZYceQ+*6m@oL$9M$)eq{vNr&C1vy!4lHp4Ec}i%7X6`}b=nrQz`y=EwtE=5F}K
zFlrCY8DUUB-ePub#0XOoMrhaG<lt5zVo|e2Lpq<Bm?)kr&e(blH3r6X(XN@r#enSd
zxR28(&?-q}0wyLV4Rh2Sns1B*12Qt^u}>BSv<D9!+`IQzPmdf9Jb@2Zqxmr~T;U$X
z9hq|)@xb|yW2889B@ILbIFf1Be_ewg`Nbxso@bSKuA!phS_{klh%19AYBC`Lx)y%q
ze`p-CGB-E3vhx4^@udvEKRy@7eqThrH8s=4Q2<NN;<tjfGoaqUi%9mj9v&VaKYj$X
zBy1#HT=H*n0PTqU8g9b)!5G>N>_v1UPQihJ%i#O4A`g>QU}@}&zXC8me|g}bXC6O4
zKVhQS7nQ7%CLXI0q+ssu?%~li@1L}NC#o6Q*5A4E&ivuSy*Ff}r7yHF1ATD>ZQw{0
zhgzo88z)$PFcl6|@MQ-^gr@JW6p>UzhBFGsgV+VWr(?&8Lr#hgrm!4zgDa>bW`ycg
z1o>brYW;PU16XLE-|cUENRsX<<T1k}`RBFU{3`C+MZHW?&A4uHNrgDs8^Hdc&lYwR
z6pZ6iP_PS!6h~=TvT?8QOdBx%`0-=U;d7T*Wi9G|qU`HvYkS0*{-v#}t2<1bZ|d7Y
zwB4KayBKb;Bqb#c2)KHAc^RzWRG8(l85EW0&;PuWn~{-`w0-i)OL~!{n`fR3UynQw
zAp9oDs8t!0dyE=fhgBPRIXIGm!pYJOC50SqLXCPqn%LB&z-)x7a(n+jD`gMwk%1iy
znI7~nqMki|{MaCv6$dOYTu`d}TP9R+-wkVXbFP^tK-lJaB|E#^bFy{M&o#VJMFD~V
zH+*9f`<s}Uk>^Yp9jwOQbmvkD+VzB@lK7?)kMQ|dVCE7#7^SA7!t>w<_hf%wcJ}iq
z+M_y=tdEmaA}`xze%O1AH@Xagk)&C`BDD#9r72mADLx4f4ko6u59WvVLdV*B!^_>B
ztM|tvysw_b@7iQBSQpkZ?{DBzp^!W9h34mhOUgSmwv)Xu{ucMz{czyW-W;f~vEtt$
zqDBK!t4`|a>$|hxR-Ur8wZ);En{$aMnb*{h{P&&JFfcH%ut*<LL-m0|ZAMpBS6;sN
z*sxN*jcNDBhaig(0byZbFU86b$h>469}x0DlL71Y?}C3ogg(i@5E&Dk<A~9Lm6cV5
z$uWL2(Vj}Y{l6;>JoS$rJreCo&B$1X#Q5O;{TUlKh|}&-JOB7&NZ>0fVoQtin@}I0
zSr{Azo0+UZGa#;?_`NFVs88$@E6K*C_dp#58c7(!qN+g1_0L9M|MznLjUvp?$|^aR
z=;Gq?>C+{KD`w~$fIkPGnn?%qod|M<PL=CM18?ro1R12o106JqG!fGo&;ha4g%Dp=
zhh=4Dp>_Fw&>Uz2j>hz<wY}ZFFwb$->Vi2wYWvaOVfrU>{4P*NZs#$aWkmyrxEK-r
z2jC6<u7!e!?;(m+Cr5(1iASTL!R=}kffj|~sNCG%DrtW~=M9Hjx85j4xVuZ<zWwgp
zEF6)RV*KQ%B_zs-vqpIE1Z1#7HLtwAIqetjzc50VxXq#_sf>&GbVkd0-4c>11QV7O
zV*m<RfAgSL!m&*dxPL$gKA(<`4jr113?Tvk?C7vO<kVO0jd}#Zjl4@W*TN7H4a=%!
zZz)BN`o!oc;XZ<Z!^g+>Cl|~~2!-=4Ee2GGP;jC^rHHJoEG0L&SSZRN3VxN0fEm=(
z)V#b|km9*<w3KKGx8kDUejW<410}lYa;7HyRgxf5k1PP^#1}!MY$tQ8gLouw*`Tw{
zYNnps8-**TZdhXw_o%l_0RlWQz^ga8U0J!s!(;bjMPS+(?<|e<)2EjqkBUmd4_%>C
z5rlK^lnlwB#>cGz1`Yk4$e|*UE`^(-=ECb=;Iu9vLe>S<o{pXqDPV~TI1ne3gok1v
zzRGTAW{ug7Kyn7oyf>_hB|Yr=yCq6tvJrIQfgM`i+8GC!Sn&gu9Vw^@!zeGRszJ8#
zz881sJTNsig@-qq+pod|cp57|4wblDm7EgTQulY;J7JDn!Lk5%oMT{Mpnj$g=p-uo
zoWRcCk@5r#CX)O-<x#$0gM*8}it0^FBG6k!y1TpW&aYvjY?h;*y3!X0B*l$68}w-<
z7C-+az)hEw=U}BS#Y~VdaDI$9uAcqy4ieZ~!u8oR!k*sQ+4<LBf91~H)dz4BYVYcD
z#uk=}_Fukog*dQ=EPnFj4;`bhP2cPN2!{E)^_#|@1A>BR4jya*r-ewaj)%*9`;PW<
zj$##@*QJ={{CP8c7)0Bt#YK~bh!V_@m2uJs!W<tze_MMy2BARgVq#()5bDv-0bVJ{
z$+=Sd>g($P)okqTVdTrSogIKGfi`MxX^DD%1CKAddesK!hv2crlrs!J>`+juj#JFN
zfaK$2V^2&%;9fw>_@-QO3ayaL^>X-44D*;y6EATI37FvdPM+*RlK^Z`Jaf=Li2=C7
zI8ayldDeZQ+t^*aE|C*=@%!aU`9EZM|0`tef6N*G&pZAXSn!vLe+-@hyiQ9?`-5pK
zP6A)Qe%)@m$8|^z+Mq81_}=|;fAQkl)vJ4>)Y??q{QXbfBpuw!e_kyAWLTwtPs@EU
z62MizU!o`iH!z+VYDo?oQiBJ{`=8%_ixZp(QrF>y{;?`A4h~00m#t?;Et}J9>t<pf
z4*WYjzx_tFM5JDQMieFDb^I6sT!@#81(Eu{U;i8b;0!V0O5Eszs==laa&i#IMwB))
zEr#VJmC0<{cOMf_n9ZtgXy815{@XXdt*0(uF;U^3yL3mit~)V3eDUdXj+5ydITv%)
zay6qkR~7S*Wie{)^V-&V8O0ietX^t*ZVvM!kI0e7kKk$5MQz0Mt_<tn{T&bx066&U
znFPjDkSfzKEkaRI$^Z1o5N8ZCGc!jvqaq`3ga~|pG|$XeTwF|yo={L*;uNcDYHB`z
z{#;cRj}a|2RN^@dQc@m~tbZOf=AJvJC~G_f!k&X~qC8^m`wlLiJ$v@l*46@A0fG}x
z@k8HX)zuy1C*4fr364CM4cLJTQA~zgiQo@a%{xCP;Asvp-dy2=sK88+q?&vH7YrB4
z8^P30o;>-p!&3Nj3>-t?e(4ZI4``^XOC_J*9j~4<gXgT=5W0sR2@hwp&j}_beGFsq
ze88p=fwB_fh0<-0)nL5t`v&Kd;3Vn(_3QSOhaN|LexI2wC_H{~!f3D4gO4dG6_u6e
zZv`bM6KI(C!H*BpyU56b&fS6DQWQ9<VT0j$Hxx=iJa0_NPJwteVH^%ncSlEu3yvW|
z3}|d@Jn8ob*+rU_to;-PZ9<MK7SQ2AVr5zUf`-j<1o`=wD}0!67y?EhD}WWkkOHCT
zmWs*|EgPKPxOC}~l~v|c;;&!7FxY-hr}))=&(5buV}>;w)gx>8&g-lujt{U<yrS#v
z?WDu7RKd$(4<%)mh2?EdvmA$spE*IyczVTAezOm;{im3ijQk#F$=thl5Aqiwy=?MZ
zfmlGl>cqn?T3c_uAm`2_j*h{^30wky+X_bHfDAKByPcsht1?3ys+IiPRoTjF0sRa7
z`sD7?UFJJ>?7-kvta4D_<S*7`D)p_^kDk(%vOc0H_<{K{hSVE4mSC{G_X)Z)cKQ$%
zm7YU4eAQTMDvHu0qIFwb$fo&zJxIfl4DH-73JH*oX?D&TmUHK(AoJ_RQIiGZ=n1Cd
zp@O#G%GbcS2ve21dhp%tRAfM1=+~}#m9Qr5-tnPR7?KpO|5lD!q(CIpb>J5Y3JN^@
zig>ii#u~BWz~+-9BXO0`7=M7L;`~g-WyKw2tTc9qW6$u;5yxa8M+ae~BGd0h%R7Pi
z%em+ux{Y+G$MlmLBNLNEcJM6tthsWKI(r!WZK15LTY=P6^lO7ThhKphLGM#d-AYd0
zo!p9CXOlZX4!`c*hYvrFWF#hz6E-3M74YWe<z>VjP`EN;8ju*z)0qWJ!6TWHD+7Ie
zxJ-LQl>$i~5?3gD54AmYf?WVIUPaLi5IwQ1eUX3I?IGR$+Ut@fg*R_L6$PM8JS?SU
zJA9%Lwv23(uy{ZngoN`3T3<czF{DZCKZJVaVpj#?aIs03@Gi2mqEO`f`}#fv?mv1|
z{ln8>;u&_(UgKI&7!WczW0xBe5@MhS?d{#WcUbK>dGVJ=1Kk;YTq0gPr4wgKrC5j0
z0q~yuPOgtnBWyg<p@K(fX;}lRw183};6UNPS>M(VGhssZ{)>My2tCWHX37(u))<SQ
z=U~rxD6{47KRRDSy^Dd$sKgox8;LFR%~%ELw-?4CZP+sZvHZE5VyA`5$J>8DZ1BM#
z0O2b>|D&u-A3F=U7KDpbB2XCg9HAoeY9`bmV)_LU6}Sw+m}J-8+lvRT1dX1T^gYu6
z4msYwnHBL+TDlg|?p9tbh4ThMlHE9nUxeDY!wV*m66}bAPpD>*6V3!46tCD)`YxP9
zz+lLHFs)8lQbwlb`rP7T7o<tt;df>*6TDEzh-Vr=jKisRod*wWV&LGbB@XDU%o|y{
zK`~_4E1_x&y%Nw>l9ltS;%<kLrPepY7fKO=YMSPOZ41sMvY-e$uROc?d$(`jzIwGB
z*$1L2C?vFP>sI19L^vn;>4O;%E_MR&-5PZd%(=a-jd&0!S~c80i733nikdW#oq?CJ
z=LMFdN)8SVfcj}^=1}C|Txtcwg|&j3RjGRE+l<W@Sa}M;h)_0BhZ!Imgx|yvrnJ31
z4<ZZR))8hNsa)<fhkMr}XW)Lv!?Rl86?y;uJ=Brl;9w~6El^7_hRZZ;Ed9XJFRQNp
z6Hi~v*gxC3N?6L^U{VFcDl=;$jD)aJAVGxya&d$f2QZNMSUv32CJZ(>l)nU5-azD9
zYilbpq{8DTk&EDBrsn2G#G`gXg+x|{7-}d6=Ck3+{>PknVBfy*r0~2vd+0WJ+)+EE
z`=4x$@(;h3mBmL>#kIAzLN1eGHARPk-qWDE%*f3>hZP*rNSuBDY0XC_-Vg5PD{0DX
zJ{6dio!75D>SJ|}^#k%7YfSrn6vRg=RIu&Mdr<eS+qS`a^nQgV4Y3LtM@y^AA;F-t
zzMl5QQ(>)!`gr_J;V$h8s`DDO0o50Ug@vG9EE>7#N$;W2rRkLI(W%tJ^SBU-O`DE5
zeeUSskS;(HXQ>q0Y1s5IxgpEAYmv23E9q{|{sqR+0;gtYwau?{qBVj^jOyE1b~bS+
z7v<VpPReyk?~yJUbBnBKl<Q+|sA#b6ozQeNoiG2wVCa^<mh>QSzwS!Ycw^@EcDG0t
z0e(KlD5j?+PmbBz+?@<5Xyf0JN#Q-f!r~v-y4kT&!^+Dssowdy`2;)9g!r82Hfze@
z=X~M5M_SL9oSLutZlV;c@}A6-B}aipF~RqJfsUEpOnkbswf1DNqus*Lm-iKc6?%dp
zuFN%mtt*H&0YlW7Q3;{KM+$2|W)+=1tD1f_gb626cz7(~?#&&+3}z}J;jBh<BMXaw
zfT}V_V{@|=3@vtUD|o=h`$`o&x@o;7+k8;kG6j79n?qGYLqq-GS)Yo#5L%m^$-)sd
zqmwVQb931t3$h2$=1bb<bpI2iusSw$3D1wU&Eki0ub=CUj#_|z$ZJr4NyZOus_Cij
zD>}v*Du0Fr;(@lEzhvP)a}SlgIXydDRZ-C(aC7#^`+<V*=1EY5X}O}oAum9nXq1;m
z6?c^YDU4;07Jaclq6ClMT1AAeOqN2#(0y2BOHCaBes@<-k0w8B0HagJntO=lzT6l{
zINg>R1`U_FXEBF(5~yGO#yffYb*UxgrlxjUdU}qvZHH)Rf@BU+eVf%iUL>OdTmF0w
z;!*os3ZCF|lO7_u=$##s8LzqL?s3m?&#4*=%~Dc0c6wq~U^4=3=S}1kTy;k5B{%Va
z$f>C*$Y#6h4j(;w|0B2i;!zY19Bp^?3#hoer*qHbSv!dyXP?cWvk7<fl%hhfNDB!w
z3VwBtbgt3mvF`lA><~X}azk-vv!>>%`(rKjm-)f(n>O;8d|6~P91SfKbu-Ll4=~I+
z*sry7+XfA_(SH^C^VYHBc5)R-5tThGtQ-l|mHSM%IRx|SwJqJ%yQpfMzgSV`EzAUM
zklXr9>kYR>eh*Lz7-MR*kII^|{(xg}J0^2D-G){OV=9dFH#w{ri1Ts4ZHMI*@L&u~
z_ADPf_wn<~!0{Ix8#rVA_KsRJnk}AP6G=CJoCPO0fm(vRUWoQjKb4S>AX+AliCr(q
z^MkQ18qF8cx5CQ6)fi1!B4G9%(lYA@7RTlSFdTBUhRmh3S$CGTgHFKWDyx{$?CD-Z
z&(<Pn=@WG8v6$&#EE9nyfriWbW*P6jWy_Y?!Mc{s7v!GA;q|3yL&F|QXW;0qTeosY
zA2xVh)sH?*xEnb%^J6y80((HyN)}n0`Z6?R;1j2>8F%CW{Qjm84^W`Z&EtVzptKNP
za6B+)L}USKQZc5!v~GG-H;NpV%?hb*4WT@X^ZgWGVqIRah0=ieib66_vE9O$;kWJ}
ztzh6Ly;8TpeH20z1)+k!qeV~RN*FmFhtqH!T!Fhvp$$*cM0hYCB(to9i1#|+PKQ|L
z;H$3ti4T+)Z*}CoEEjcRZFAoC?(rz4W|OJxml=xh%yzyXSvgZ1${ims<{UKVG}q8Q
zA1JkOW7MS3aQ%j9UQfiNl4X2u;^qbYo%-Es-_%_*{N`uatIC$VQc_a))%?<Ga4!3F
zY%;r0M=nKm?EYAx;e1VQ=;%lQ*OX{k+h<$3v}=E@Kl*vBW7e#6YsB-?UYt|08K^#D
zZ@&tdx&{pqeGO+Gu(7~}sh%~?&b6q~<|}iCkHs9JwxA+sWM<OA;)bV`bmiOn1P6na
z^dmpx-N}IFaSljWQ%3mt^XG5)&7(##9zXtqlz{4ejG8+7<{9D{G+(|ve)LEfdR$LQ
z$>u`?&K3%jZyzcUWr_#HF@%Cm+_*D8=WfM*lgZ5;iNq64&HAoBp08MqM)!*%Fw)92
zL}TzTJV@Xn2$MQHlz$P-n{&INe2+q$b#Tc4tlHm!J9_V@J+<p2JJ=P59raQ;IN`5H
z{gqIK(04V^GpC;siWj;jKwBw28dETi$q&^L&+)~l?P~P9-jAb9cdP&0nt?dPfs)Ic
zi>(fk3M@d(N2#d%-@JL_=SScmR0DjTIm#>Q;Uu1--u$R52zuyAsRX%5H-iC)x-<cL
z&oSZp=l>(%?B}uVRN}>2yx)5aOn7TN7a$|v`}S2lPznpqHGk_dAn&PDr{1KQ>4B|E
z84~q~dk#@%5(i8o;=G~GngVqU!+vym%uJxr8ADJQA0NkH0*{5@LoMr`^kpZW*EjX+
z@0B7Pk|Iuu)3K^^+wWO^pz3Z&uW<gsIJ1s|W&Da!mR61C*wX1<5#y<p4Z;54${R)}
zX7t&p%}fvZh`jlk^WWHe>##1@uv^f*w~dMg24#WLNC=340Vp6P-6E(oN=SpPv~+`j
zbSd2k3J6G-fPm7Cbi=HdeZDj2I%ocwYv!A4=9-Uxwp;MW8_#n;_r315)@}1}-TDe8
zoLf4iOC|04)2=D7C(?%c*{y{=TYVq@O!3A0jcYh8U#F8B3Ug8u0$k*@@7}#bArOxq
z6$=Xswmw>Y3tqJ1oag^O-t=G-*d5pcsDPh6dqxy`Po6-KsDqjfNa+ynDBkuHnoH~G
zY0i$AB0M3yVw*P8N6MbxNlI;K7l;H5h$o%5vZ^ZNtyZXC&|^R+COtEgSvl=PYN`oH
z=lFz0wEAv`@HBLHYdGpz7yQOE!=Xb;LPD}<4|H*2=q7@PWnf}5)!$Djv{BCVpka?o
zNcfJ++o{@m#X%pYDAI!ctZ)lL;s~%3{p_jVzqKAdR0IzjhcOl^ENwFrWQGxlme4|-
zSEQdp<vBOn!kGM3Ew$<Q#6)uLYQ$|im$z@><83i12A`W&P!~-f|3Y>LZlaOSAduM;
z-bQG;VIQhE;r0tX$6o8&fEivv6I^e43*Ee(yLJ(WY#YHSmkJmj8KDo#MeSDl6##8b
zwAOcxyISYgEb>hUoa1i@?xm{Hx`w7R`~0I_yLQdn%QE;b&lxd$B%ed*Lh!oAId?cJ
z;u!7a89bdL@InZ(Hdxh2vVqMfOJ4w41B-YDIuz(S)6kX(GN5rnw=jG#JPknZTghKs
z@(ts>{n;s8d80Or9tN{q3&uZ7&->r-vN$29sZ*3L?V-XiM#fPW=@+B%YWUWTh`rGa
zxAzAG<ukY$u-D%GCiN{gqpnxd#mut&c@#ntGC**tM~8;a+Zn*&Wwb@)`HL5B){;4(
z>I13Y(}Ze27je4907L`_^$R}>tYzmTv035v8j)g%dI*ep<*QepZV<UQ&NM=GgP=dc
z&{J8dhl65NUea|$VsJn<n9njrKk{X0Xcw9=7H%zAL^(F<vk;gIiHMv`lGh+$lK{eV
zUF{g0w{9m3zaqR4yu9w|^I{8X=9$N{b-{iJ&kUVBuo&z>!osT*@_!Ks618Zs3!eNa
z(qL#iAemi(80ggr5DhY7%8=izgP>Cyf@e+4M*tMB#^MO_72?h-ZAbV$I`)3u)0cQY
z8oC-H%Q$Zg;s+WWS`NpKP5&OtYiv349IyJce0t2nh9+u%{DQPw_s<=npa7gy%a<FN
z@T;#c%0d0Eb{Y{lB4gi2C`LqNA3bTq969UKmaH1<Sl<pGf4ThB(W--FWYM{K&(RA<
z2YP<|fL1|dyV<+)rZ%bKD|7eLKR@8|Z&`a(^CoK1O@R8b8l%S*Rfj9ROWYO3BZZlk
z-trQoYjUw2Ui$Nn@jU~JEM`)*_MUUsmBv5YT=o~OaWUe~w)pu&M!w}e#kJCX2eJ<2
z?OYdqM5fHE%*dKzSeRnrcAqOKNuPao`1?yv&prG98QuS_jPefYO^=HkXR4K77Vq6$
z`#iM4xSq(GO#Sil-RZ!xK;;gK6WP;7P0ji7@ei4#oprB3a`nV!;maLX)=-$pfR(o{
zY;9}nK_jTqmXBtU^rAOXJujD<kIced6pRQp-Qpa5``gMI8f4Z3+J2o8vR-bEjEqEi
z6Jx^<4)?kXd0oK88(b%ZBBaMVv+L@G`zyho$h>1{WMsjg%_)M|5hJ<YChLy$lU6(<
z`!qLUEL|GD28AMo2ChhD&>BUI173$N$d6*j;zzN|s2NWbN7kTMA<uW|G7rzv($X5F
zYXGCh6H>T#K12}539Mobt0DX8-ZH>{zRaqvwY4eep+W@Sg2u)x($<%JE@YJGX+k&C
z9Apq>FzVL&mG;HRs41~1X!HK${y8?w9L4*E)3WoBzJt*b_jUT+y2EbQZb;o=WeRz%
z@i*sO(96CvZa!gQ0=o65YRHZ^Bfi46jx#tGWz(qjYb`Hg4O|^|sb?v+Y(J89{xYvh
z7_;NX;PR(G+ToMybN3s?<zM%I?<X5l<kV<cYAOuhc)admz?xbXQy#<EnC{z-iUk#u
zaY5dvRhq8MOWL3S>}NclsT3^g7p8t0Z+q7?shoVMg<)&uZqMDOp=lNwch2QI)9;F9
z;!W=t?L8u#b><S+lg=8Wf3ipqPPXbB7<FbFYxcRwpJ;ugbWwF;*J;u|nuVpP3eCk0
zXDOGus!AKrxqB^T@=*)!hmHF$r%dz6)+xO|+F+%IqhP*Z6*|UvO^US_JTOh>r$|IU
zGV#0zaJ(_i!#P){XJ+mc@<8|Ca_o$jg?GU8>#+!jHEJ4~&GB{)e%{%ccq%~$we_Qm
zQ62(~nzZOR;v7iSAn9C!6z(>E<uFEqG->)=EA^=C>QetP##M>jr$ulN15p!G9?*70
zYAPPUgYIn2wTlcmu*tiE>%K)B0ZdtjJGLIJ^Ts&r!tcx$N6H!^c|N!1PQ-@!m-Lyy
zVZ~3}V!)yWX_YV}M%j#bpQ}bqMMXH#;GlUsD0S;r{ZBJ#I&aK>a2{-)7mvvG4WOI%
zw@yE%bj!(sv36>jo!mcUt(bYp>{x5V6$X0x%6%5>+$U|c>+7p^hd13`3$y`)VP@@{
zLMHZRKLs7-=X{jwHM_Q<Elj{<^U4`VYr<|gh|7ME;9<LVoVhV$VleCekY8bCeSWBn
z%Jpwe2a|>T@3Gtu6^hXwQdvs;)5D^lePGw5(^}=sg^>u~Gd1!ruaF#SJ-;Tu^qP_4
zWuVzbm1JKNMy+0-8?rCMf3?Rxl;UsfR!@*A6M1DrmV4%;M-$i7)v_;rj)G%5ReMmz
zjMUX<jBGkwAZgh7oTy_8t!4)$(AoxH$)Hj0Z{`ODzeW6_x~HV%Gz7RTOgh0mLI~Yy
z5IyNSs%+NA#92(1__lWMMNy~2w6NZwNKXGA%Gb7Z7?z;%xeLibx|V8+d7k1E2TrgH
z6^dmC{i%LB-Xr(`#c?UE3<#B6#r*n>YwgmNwYB8-@AaRcyg=zAVQ4sx*a3xHii-CS
z>{pzZ!0x&MGJ#(5(()>uw^4hV@N99O1Dm--onkT@?T3w%+6#Y^&^N3~|9IiLQ_n#^
zuju1QthGx>sBji(I?w<3VveB}y)`!&*PN`n2vH*<KB7cKL<HdeBZGTLN!#+gk&=d7
z)PB8E?C$Q4Y_NsQ==&?|;cq66U8{_j+v7q#jgpMk7+0kB2js<61ZjV9f8|cHKlnKL
zwD8Ln&7qTLT!I2E{ag68Y}0%@`0NE8qY9awZqqf9SyW9W?k=KW3M$&1bkOR}AL!=N
z>krM3%1=#Ppew7>5F~NgbEL*`en39?C<SA%2-$6o;+$Xg%l&^Q1}9TY?xhcEiv0fd
zgc(*cP2tzf)IH+jE}Yd@JcZ-34+xH;k<kWj3HTOaeS=yNMIXREu+})@>Dk-%6V5gG
zCcX!iTp_AMz{O*uyN0JGhX!N`YR;&zf`Z?Jnihl9y_#|Ti&vO_%N=@AZ9<`q+6~N=
z7PZOCY;2EEx{9zIrCoWO3OZ|^!#N?o3XnI@;jvj!Vt_915Zz^7Cr<UCE}`E7YV3hb
zXU>#<*r6G%Jo4q&op!`o0>$qw>tGBPu%0=GZ?-KKCx@npzRDYH9}!dZ|MyLf@b0AW
zEKRu`$UfSZOA+wi+9xqF@x-VOIy!(j{Tu2H-bd6?om{!nC|~sEO$AY+7~fl4TO(RR
zp!NUGGX?%3I~5cJ8V6!jjRDkxfhEPvbo4>|Z`Pg+I>|K5c3ap;>0U0WZm)jS>c}yr
zn`FQ${JJH!N|^dlNBdjOs>5o+N=a6_<1(dZt+Rf14{Fv(@5w)0TNW};+TJ8r8y~{f
z&{@XU>$3Z-I8ahDad}Pwvum<K6%`{Y%~!wu+R44MldQ(~v1wu4vpGXULx}t86IwQ3
zd!T!c1_*&K;%R}=Fam)fa2s>}jYh9U3hG0L8Xy=2Kn@6z59b;XN~kYwo^&1*xY6q4
zFC~$l9h!W4Xfys{jIM8BzUSHUBi%Y@1*z8YJ79^C8W<^k;X!SPUjOgu>FH7-e$CFG
zKYzY-Lai2LAUAR+;`$D)$dW0B2g-sPbL`Cbuy`Uzf}$=_Qc}Jgldm@4eAZ|&+fsb<
z0qe(iht<ba3f=lT#b(=hEoTpM-Q*qd7*X20zwlyo0}<)C>g$JSPTKIB`wD`qn>@si
zk^mB=LxasfBj!<xYH4k{2oaJ@uwNb|j5v}JG0%2sG@|6nY)6gi4=pX0QYH4l?;TYk
zr+wmdJS(i6DZi+SRei|Y`+LTWfwzjXAtzXL^>dyd+m{zgZM{3D^-?^MqSU(X6HsL=
zP*up!&kyH$e1?Z72f(n3`e*cF04_sV{S~w(^M-@7d5`-<z(hjTp}Te(kHDt8raJb)
zcfb4?KRxnN`fY<N_v*n`neXLcKqXRaBX67WvCPSRnS1GD7t@?0vyWG5YO1b>@>u8F
zVF$a&e6LEKw19x)TKAA%aGiG#Ec{%0E0ZG1uT)@9+o_s$PP2>hvMm0vs1Tp;sHze(
zx0G~=iinNiKD^)FA&OYHV+7LHVBWB2KGF&AezcfbrMLlWHT`fDSDAb?Zew)c=(^TR
z=LdDCi)kO8Tr!$a%+}nroKLo9h<0W23^07+zOYXr;%^JGpbt|=c{}*T>_j-uQ2Q(H
zj2+Ugjr~?#{-eG9wIlQF4GiT$0=$X73sDJzusqxt?J{_F07fIo@kZNEU6}&w8j$e^
z2PS2ZnUMJZ#z_kr3}9@uq~ru+$kmI4i0D7x&jhk(H2utKWx}I3F-uEU>zazL*^ho-
zui0eqtpZz+!tDL9_9K3&1mP(!Utjr+ntL48Rw36Hjnp_tBgVLInH}~{IlNU=tz-(B
zk7Z<wqO@4CxOMAoe7;K<pQR-K=UB8tg3>i5d>#LztfJ<e+acRvIBdbJL!ZYJuN$eq
zFDWoFT6L>ZN}~M7Ai`bmib#Of^m5rDTakj|nRR{f#ATKR`SnhUEN;h9-JM?Q!Xv%C
zR1^k`1`QF9L<8x@Ykc25u+qu6M#n55U4OJ;zVG`m%9fS3L-mI5UfK@FRs5i^J8tlu
zZDUOq-3oF-(b@opGicx8FoDHuPuMjm7Hx`}2B#9ay^_Br{4HcX$cWMzgtC-f5#@D<
ztEyuaQhqu<%mpbKed_^x`uk>wQXYo*`jSymC1;wbOm5%4-5g^E%=$|K#kjUNNto-4
zg5{0Qq3)*AoSq$8ATv`V*}D7D-?GIXwY#icw3n3hSm>&B)_CzH8IW;NSr($dr<n5A
zkN)y-yiAm1a)>_8!Ltp*KA0HAWG#FSUL;(2aV?VmD1;$lYALRZ@k4nUuSwVi2Sl98
z07lC>Clbb448-~X8l$Q`_5O^Cwbj+)2A1l9?lt+sN7U8T=M8HNp9rT9-BkuCc2>=J
z-1ujg#}m7AKNTZfwI^wg3^LNEMrPeKYUm>Cer$i!V45fJwqMLG@hY~{)aUbN%aM(|
z8K=B=hn9r3ykejQ@onxQGQ@(F%#T!e1OqST?EC}FK_XSX2>opgvS~p*4@<d3F<kaV
zD;iUY!U4a+!gYWibLMTX#lk2t;DDTB<LT`kmAjr>RtP?MsnGA&_j0rDmi(RUkbAL?
zw6LH6W^N3(N%!tOTvLqpl2$K4+Lt3Ui$*z7N`D-~^a!F!i8iPwI1SgK2k)0^*&y~{
zxm^$vp~DhwdF4n6cPr|X?%8vQML0Y#kp8lKG9LW03IPv1h58^~Lw%L`lEa$S{1{jP
z6oTKz&=?+-Q3Q{jpg<1N@5X8_bbVcPck9Hm$&9S8UmS0j4LE-Cq`pq%aK_IySsmL0
z<oyHZX14WJxiFTpDU9?+Q~0vJmvV_Na8BXMcaCj4C|Z2NLYAIGi}SIe{&>{(i7wMM
z#wCAdWd+V+I^+r)QwubGtn88>51vLS0HR}4WP8E43LNtr^Btg6=S2J1Tr~yf%g`_>
z#S{vXM<16Yzvu^5w~KKM-A6lhv@M9lG}zzf6wxZ2PyHLP&4IJk->YL=!AzJB_yIWY
zxV6Y0G{J;<q(*Vf4l)!2^hT?^w(MCbpUe0sesrUL*L_D5O(=f^Hc(Y%)^2T}7$zUh
zRU>L6ewQ+^$KZq(fm)2fvxd*mqw*>x6G5}ddrS;a_HY*_<nY^Mgw?K~+(*BI<#wuS
zb_&>!51bL||0F<C0JFmgUw{~Y|9*NDk{O~iZ1612F6Yg%MB#y;4v^y>ejoNKH3r#z
z^j#AnQeGK3f0mRIA8X@t6Oc71qi4R$9#j!MX+w6>2I0Am>!gh@-U=aQ2p}fi80^0a
z#viJULxbO^XoQ&oIN!Y%neWp&n-rT_{46<h;`sp*=T5TvLaq-5&N1B*`OWnkU5*v=
zp<4Uj?BRHGh2&!9nzu(rHg8t?C)HN#z%cpW&h6x5>amN@PVXAb3?3T1koMsE=~5G0
z4Nx!UAQWktpG2H)F&YK=^(BS2Udo4$=qITUz#E=*Crz)O@uKmc(*?@9k}b8x?UbYZ
zukUGV3s(#!Qt~|dy!7S=_5+%y@O+H0u_yA1;7cd0@9f!4!)h>j*w+$yAD6=}{!9RJ
zBp!HR@<=y;TpU9vlD+_dns6O5A8VCZx<Cl=Wk~D4ork-}VdMMA$m0R|AO;TO>d<~h
zmWM+Nuy!{(L#T2K3JQQFJ}5W2Rt{Yf07&34`jEw-8Gv5eNgA4PL_7)#TL5ohgMzGj
z%be-O@RPf!!;U>!Lc^U0;u;{d=t2o#3jy^d{DuE0X$|oS?qxD(QOkMn+XxAgOS8~D
z5O%<<QP?D$`*C1$ST)0{r`a(>+?RoNrSO&l+rMvg<@wXs0+zbNE!``xi||WJTSx)E
z380bj<qavkWk9dh009H=!FehDn|~qg;k?mEU|ttVZXUn|D!GAqW3iLJn&)VH*oz%y
zAKt!f8?$Lu@4j!%?4eRN;mpU&lR0CQY4)r3QjyVz)G*I;Mb3b5Z*SZ+rS%ufYRxjH
z{khUIk)6Dmb@)j~;q#=BB&uvzd&8J40rSE~vLu_%8#e`cFQYx7LY+xDL1mO=s3xAH
zV3ydLFF934bu665%t7nTtCpX0$zn1}2R%4m07#$)vpc{6uLJ!BwfT<jZimvBdrGeW
zV>35jGZ_ZiEjQ9-_c*LF;10pAw7=$4jrCLrCTv>cUE3%O`c)6WcjnHfBF~x6YR%~!
z@T?!I2|gb*1T(crqX=p&Ab8p*YTS5;R1NRZM)HFk1L(Ky5jS@I{?dZoA4juuw4<>5
z9BAOpV!R*>b-T#VE-a9TO2=pdlRhsIbndo?v-3$9+vDsSpO`?i7<=_}US%^3jb6>3
z{ho*x2d?Hw3=TygkeMhZ@lHYE09_hq5*0fBNQ3vK91U}n6O=(7i5{n!rDb+{IwP=3
zAdU!|XM>87C1{tw-Pi8`L|SlloI`5`{|YFNI*x0oF^9lpIGS>|3#5mJ%2N+m*}`=<
zy5sjOUigkO^FfpAWj!YsmweFe74D(i{oV1e9gW}$v>fq7Py}3kj9JCUm$vA=sh!;#
ziYa7H4=t)-K55SEFME|(VJG$%q6p!f8x+0~eiKh>v^Y;6U^iUpqR<!vB?HWbCD@1V
zB@F>5D}_pB_Qs(pUZtwS99yT6Gvc04Qo4DU`hImDG1jDUw3qPUnzC_DB9(r0%Z?0d
zr|+Z(A8&8}dgUUkdm&Hp`Wic}Q0s@zCuXW@3AHtL=1!tcdkW4qezUo3EopL*A0HGP
z-ebo$P%`kzpL9Qe#>X$E)S0$Cd~XfQ@7y}%QJshw4Qrk!Fv|xCp8)XTz$Pk+H&lh)
zgZ&mY`2@7^MCkrYzktW~`T$je4b_n&gy9q32?3)-NQ0h01Ev<J`UKCidVm1SG`$H5
zmF4zev`|p4BOICWhYtYAK|^W!&!6Lv;UQ!YZAAPWuxRLU&?m%P!S1@cDKsP#zN70>
zQ`|z1W?;Mn9o}URG-lIy6`r1+dtaSlVVT6^N7G^o3hO&}yrxMNXP4qp`HC0r%vEE7
z<}xEZya(MNWB4oqZN;y^@Gt!M6jn4QFd*>b*5_d_cJ4>unBeBJN9)N9d~~Uw04u7a
zct>{6IFNDn4DlSd&O@(jsQ0R@JqRNvB`f!@!)Xu^rYVzs8aOiA3lHu*BX}Pm!kZTw
zAj9$merd4ou87H}xbAE@`wv(Fmn}YAjgTfLE?{w^$%UgyAL$(MXuyvsP{Ek`if3U0
z@WQvLqr(R0EMk9yyx}!nTD{@>#F~0tU$2Vl>Yay&6jI6fbO8*zn#FGzlJym8F#GtH
z@_sF_9@omicYcpnoOo%h0%ZOwaQ?>aNg~C}86x$a^>p!?g_KT1W#rR!`Posz#1@ek
z93hxva;v)$(|6|fHHHD;Kz<b_V9(%PsZn5cI&ZWDl0mSuSo3;c2SITM6KuK87DxzT
z(Rh{d3m&h8I{7(zdOfrOMp-2zan^8hf8*f8JQX~{`0#vcs4>7gD>X#x2C3MTtC}+1
zRCosnie!6V*-S@)-h^InYk=$!BPl7N>NiM^$j5rZ#tIrc3R(@~0*6K8Ga%>qIjo{=
ztrs%PZ*J42fBLkH&pQ{~b@%@LOuZ&LsElBrNU`~<s<w6#?G$j-9~8ARx1m2kh(#Li
zXUS7iQNjH1C%jGO?~SgsfcJo^Mujg2dBUr2-Gv>tm82&*<8Cqk)i*qx1H~QsYRMXf
zD!!(`HbI6TUEcB7-2CT9truZ;L!8T0vFEpT{b|lQ@@qFzt8qf3(njZ1`vrw+BF@bT
z*b04{dK{!OjBOrNZ4jhr+#UD4Nc}(~*+}^0s`*v^i6>dV-i<SmhbnXCw(C$B*sI>H
zh%VpX$<HKln*HyN<6?oPRT1<T<1dCU+25m0XSaKLo6fs3%``2m`^U=Ij!*;HqkGp%
zhX{T9#>T3Nx|B1pbEL0%*0T@zgI~b|2q%MVUMovxcD5$;fSw)(0OSaT{0Z+SHJq1b
z;Gj8^WoD~hN=1GKy{M8D1uwXu*k2UI{MOxqT!FU<ohu{{83t{Vo4HBJ$*6}N=Z0mZ
zq@>O!`X4JzwML1^K1GjK_wWF1TC9fy(p_Z2>!>APCNA&A$v+CkC_!pf%eznvT>|V;
z=djbu$7coy2QR@>7rT=f@Cad{NZ9pFDIbRfEh>=(T{Kz|+$oR_|DjNU{uZqUXoB`C
zDk}1~J3~i_r0~`2*C&&nwv#k_`P&9K;_!kU{)fPOMn-TWd?emdLG%*-`RAWH;E$LJ
zxXufq6X9e+O>N9}{5<2ft$MO>!)%N61l+<*{yn)F*BtI1*PmC?qlup7=g*(9NoCER
zN$#mP`*_Ma%kFE~p<Pon&Az|e#~uPW&smvC@X2KN)lvKeab!gPYXM2SB*rfT->U6j
ztcaWw>K|ZoHY_VB9<tBqi*l*k&syOb5SFu1wQ^~!!_Kld|7FJZ7t18<<rRU?Vq=G}
z(A+Wd1~Cd8A;`HQ(^J3!hTaQ0w`lNYl^iCfrlI4G!U%AkYNmN2&KD2~A*$qr?Ham<
z$CbSSve01g4AEzOGLeVG8YvPUj~TZ&5sTY4dO$Hl66E{t-5=~?5SD!b{{j0&h6w3Q
z_=)g}QEm{LQRHmjxibt<3ck3wT54*NaaaRdCnjmKBzP`LNzi!{4SB?8a7suI9N0i`
zk>h4Kd-e-*LU6C~OVGa?>n_aaq%!x#Z7r{?hzPAsk@n)B0Z)k62t7YVyiEjAcg$rt
zR<VF!h_Z{A3I{?8h<f2yti90c4hk9_ABPHN1m8!bDuDUYgQR344@LA(1W6Qm+?$s#
zfuNno<VN;)&V(9io2bw=Iq>k%{Wdl>h6)GJ+MeCJVb_;DDDvid>Hc&b_Q2r;Mv(-?
z)NY6Y{Fzms;Msza0;8%v^n%ccgoIO4O3KpQ-2Cz5coaJDiUK<Zk6UMrLxX}M7;(a$
zpt~HP+I!$Y5w;%6DUgt{DGB=#{4{~APISogi%X{cUE@GHTZgTIhC8P$44awqRCQ(Q
z;mm<#99T19dUY8G0?r9-ZEZ|}1m&;q_iy;=yTMQl$Ux->6J2r>91OHqZ<w3|&>0D0
z=YD&r(lIK^6k!&Nmw>+5+3#>w6hoD*p`n3$CWgQw6>=k;3qHiZt6I!6)|HfcbG4;s
z75uu7b~ik{sKGe+(fs4B;vvWG^;z)YW(;$q$8-a96c5cXefaYFKxm?+@kD2y`$MYi
z!^W3dFFlTb%pGaGy@y2Lki&I)jylu*L*b!<o8xx!*$K=H*;c7*w(?bSkLd?{$A2Y#
zlQ|c2zUa1%^~6A?xTk-u0Dt1QOZvkGB@gfWMYEve!kCxrQH29w*0$}O6Ej-;psLFi
zO5VPWhyKaX{;q4}Y<#j>?j3`Rg9McZWal(Fd*>K1JVE}K=ZHp9GU^*}*Y&zNDQ50O
zWPVsdc8u5bA?XWF4s&L=b0t9vCgG7yIt3EWwXq^MG!Gt)6iN8ZP__8+)EuP_dpqIb
zM^u)Uf1X?rEGqcw``6x>q&DTcLH;H&aCQd)>&$5I%*sE@CeFbOI&!$Eokol2&OCh(
zenrf(CZvL(94IBQzmn6+eS`hbHhL#puSsNuh3jeLi`PY`E-<;tT~pARI?cap8v{Xs
zb^~~UDla+ED<lN?w~Ba6>~0LTCOG;Js^Gtf1C>cWyfp4$*B3P2^`nVK${DV|q=k?W
zsGZjTus12vj&ua_1ZY$WUV&qKB)P>0*<46im~s5dQ7**Krh-mHinbu7?JxFL^SgrZ
zxp!~l?Gu=W4H$x1o@Sfy_WgzT$^1kTPBEq}z5`{Pi=d2uD=!x@pbCtOgC~Q6a)N-c
z+hF5)BC><j<Fi^(NQDmz!Jsd623Cq8Pgue%G-jp$qfdjNzFQKd;@v}*;tk{XY6Cf9
zv17@TZl9J{n*5Du1ZOo}i2g1(>kc!=edN5eL)o+N%^Vojkce#KfX}5@bfWa|W75C2
z?I3rP>&73R65}C>FG77YK*%B@)*Qis@k=@2S{l1^8{1(C5v{NI!?wML{{JP^`S-E>
z=gN}|X?bn}l7w3JUtyLrp?xQpVB5CsAjU}_lu%^wnK_=jwSUGxjN37(>m5J~wz7iZ
zzv=z_6wu8}7itVh-7>=YKclOvXl07Fg6d^;H8}bL?seERG?eP($%Mc=u-?GQr<6eG
zkA>fWv8G)0=?;VM1A3eS+Ya9n6T_+L^PmdWp0hY<RN@G~2euQ0NOkM$e>ar*FLB`i
z2gHH@@6mT|R&?8!3qQ^zrJ>1dJSzPc3Hdm!J{$S?4iW{<dnH09`7%euBb{jJo*ZFv
z5_usa!MX7KP-NVXBh7EiyFV#jl|D4`r{KqC`eHX<dUdN-L(gI#$rW4Bpt{<-3{nHl
zwoIR{3iN~5xb?OCj<BO4zWQ$nY<&}!M5N3A;gIy7@Pk=9x`;ip=0UW9fD{H%V>p{M
zqm#QklA^Z8UjY2jvzKV%#Sa~?VL%=BeieUwx=HxFLO><a66E4i4PcLoib|s86u#y9
z)e2)?w^cVn^>ZYv6@ZKvA*LjMWe0WT)BDz_`H=t`C$!!U?!wJJo&H8BER`r_X&c0R
zE*v^|@UB-3e*P}Q)>C1UM6U=U?&%50{h=i}(L+3z+i$l%tZaIIeja`9BC4H)oKxx4
z4Dp-%zAeeA0P0v<UzcbkTo(ckZaqP9Sbzw^aY<WcxTm-{ldFUpzrF0V5H8}Q5OPNx
zVr(%x{{1AntxqDa3V2D#bRar1D6PdOeU|(8V&fo~st398qtT7)*Kc=j{gS;~zl3C9
zaIh8V*_W(|rdSU>KH_!Cp2kSyZFh6`QKWUi5*DE<pnwB^&$j2^>rDjR0Fc;<ahVu#
zbAGN{H#cBvZVu1Xe2riwRaIh+(t}ax-66dq7RKku5!-ZCTvxRZy>`sjrCI&^yi@>T
zKu8o}Yya<^@cerxHMO;pi_@{7Q=v=nU}Ec&gtk7Zam1lo6}SVCcIB<hJ?lv<!jI<2
zD?rREVBtEEgh+mO`QKZ=%*dEX04J(7Wo1ZOSlH(Xi_Y>>#4Sqi9yaY$1=x<oaDNz%
z1lwG%5x?X8K`Xp<oD%VclBl)^M*fFyJsQSy^1DfBpHkHM+Ak9i?8Vj<T%bgn4S<J}
z|0Fa54<DX?=!oy#J-&5AuU?=-8y+4Wb5mKF48P&yzZamZE0zZgs1{~GgIiApbI|aQ
z_PU`)g3jh`HHuLaBMCL)VtIb(TaP2>K{eu=OqhRtQ$&i6_Isj&652qZ{vN$W{KiwT
z<(P)Kldc_)R60<VsvlDLTYUX8sz;ZEnU}kq>)*fFHna7niU0ZbhIom@*D0E<3&x+G
z+HE~T;xGT}2mkMUJaesBiC5^s0bm|ZiI8qwckTTtTb2%oiqpbae}4{GudihGn<&cX
z=H?Q<-_z4*XP}@Xq6AcYFs&lVo0=kLPHlz;C)K_a<+Zr6)l9#r^<`qC=n`A*>895G
z`vb$nO3<#r8#mLq%Le^3f*_091ayArn3I!|_9kefd`Z^{szh=8K2Ppcx$6m85GS8c
z?zqcdkong(t6jv}zIq;_wQiJ;a67_YA<1K%$~YtJ46HWly)6-skGm7k$7Z;@kg#?j
z$vZZcM~pN^na4P+K`fx1XRzU(I$#`qH){RolVyB0<>e1R!{5;b$BJ#Aul^g@_l5&*
zN7<P8-ga6Ne%gwVpaHZc$vgW_6-nM-l6|^*R+r#T^&v+P8};;s3*W?EL5~eI0^1@v
z-U03UkggYC=@VAbd-IOrv!9;c`dJH*sPKhC9tHFUoHWYF0KvgIZDD3Wyt1aNt1B_#
z0uX#aJNUE9QTkqZ=!J%^qrL|#7qsFqV90=u4jnoKy!g`W#dx&&&_&9GfgEZVgqDfP
zN#K+agv^1n4}fbLfF42Du%3;%m>&ny7!=9CKOpcM`u*DzS_(uS)W^hVLzs<}(_F_l
zK><V?>RMm$oxZ-lCRFCYL)sD*>ma1SRkgNW0y~7BeguCD4Xz+YcL1Teol^mXk{FSJ
z;t29M5W=uzP_q-9<w<6UV;5mL3XTpiT2`4zA&8cNnSgXbLrzZTkd>UAOpvB=Q)rUe
z*xJI11W-C)4)!Zo>`=177D(lZ+yzM!4(n?7!mG7r=?-u)k`B1VlA7`n|M2-F)wXRh
zbC6|Lre2YlErBFgZ^dtJdb%Ow7;{gu7i6CxRqF$xdn}7FD-GFP(O5gTo}_{@hd!D*
zu!kaMpyEVt&y-UQe2{b8&@RNC+6}P4Im7(}9w=~#j*p6BHCU`?yR)EF^zud9C*D5x
z`6*5e4#2?zd?#<5SYCPi$eK6QczJl%mij#KwDdh7G!V-4L{~sx?*jlJB7P>lgFU}i
z>GqbGQ4IaV7)OZ;3&>tHH8q*vcu#ReiByjyucqrL)qrU}(y1zsJl$>Cy-Uu-s4ING
z{7m>zxg+rmhB1>`quQUT4~aXQ5b*!_AV^G4qN@Px1|AQ(375D~+kQrfCse(#!m(rI
z)@$qs6k8t_dT13A&DwB5nxjWQ#Dn(!1&q>F80?R>B*oa{7_VM>vi%yE27wU30X~5L
zp^>(BRCNka{1vp1hG%Z;V~u_UVFKg~71hKyI(bri(UHyuUIYhJ^%)sHzI>3r8B8+A
z&~k&m02&x|iDZmg_yxoh7uEWBYT0XcE{l`gYJ7)jSoLxstWdPJ&OkR@5N$Quqxv&i
zx@djjZvG&3F-lJ$XnPLpHn8@YohU_{1Pu`2eO1|mXlY;tCn)w2(5U@tmdbBPqb)=k
zbsriSpp`WV{*BA*y$*|s4hRX;f#+Q`EEZg!lSKxwzf<A&3sQj}8G7CbObhy0CnM09
z6&G)!YvWALoKTnszd^9gB=W_c?u*SXOAFCveK-3vDI8wXM>U8;Q+f#a?a<IMa`J_;
zW16=Vl;G3GyD&C9yo6b)fqI!=v4(6Au?BUOTYwK*)GabAeY~1lo}QNW^+g8&H=_Gr
z*l}nTb8;X`1)~MTL2QgC=H|_oUx2%wz5#89Tt(Ae^bjsj+JIvNOthBUEVdqa3>3g)
zcVsOe@E0<NyrQ&y7Bci;{Wi||KU;C{AT@Q$<zO_3?tnIm4g{^EAW*UE{!R<f7Z4pn
zAZm~@5(IKIDWPbwFgGXWlHI*~7uJ<92DBoCY<fZ8YmcLsN)X5;x|2KsuQtF%qVN(i
z&k`ZMWiK?45qfD?)Cn6sT;)y^mffXb3J8MpV4vN{ZdXSMA2xN&;OiPQs3hA<{=}v_
z=^eH0wa_7(i@!Js&mghtpQcGe+qPL%59f$IC5Oa~9;0p-3JL8kBq(*32cr@1h$|w7
zUfpTE*z#fOyi{1T985cznV6DrrlBnbe3K9W$wZZ-`wp#3a=@gzDZ@nxvuwr!boMb1
z0T#xQ&~T$CKoB1!xX>m@SY@;TkN?_1JpD4#coyaNZ8itMPr(&8HWvbzfeB*)?D2%1
zj}nfOSVB#WjpqzNT+%yP6T6;URb9;oejk=>L17`3QP6Jopyyg&FP&MooilmpJ#>z*
zpc(~Y|M1}uEFt>)b(np@sv)8^#G#PI0SOWo61t^;Nt~JeaUw^;y7R7x@5l_hp|76+
z--=~-^}%_{y=zyvH;9LI{pPC80b=P*9n_wAj8H>IGnET1)U-PqO&64*cyja`4}fRL
zLLufji8Wn=2eEq7<!#ST5h?)qdEo933w{{LwnWeX6{UI9U}d=wjFw7ugUX6M&mWEd
zxS06{9vz@Ik$H^zdO#@c9rMo#L5G4I2T;Ock3l<TkY{nrHmn!@bT5a6O@BYTpV;vs
zp`r4Oj+jNEvcLlq*On@%_(3jwm!upevR{2A#z`YKn|J2%vZ;lI?``?mm>6hqI^Jn-
zZ1!6lkX#m|*kKzM+`n~;Z{~^6_LqUd7z!?8k|*A73}F!x+*=CaLr#8vu>P^<r_3jR
zDyOX>IYr+c$YD^-{wYlJy!(0E^x?f75@A=Ck)r6o3KCgD4xx}*PF95B6q$$7WCMEo
z^Jxq?HK&vvuwwBEF4YJh8(3~*Ek5LswgtcivPd=l1bp2(M$dr58^o{)H%esZGRzpH
zx0{sIKgsY%V$`|15hFIJg@F;GsS@uCzbRxo6WDF#oW^<;8Xt#p_iaUs%l|Fa*pHhc
zhDL(&e(KaOw1JmfaRxeoCkJX320zM!k_ILQz7R(3xXmpsrNqS8h5dJ6h5r5bxoQBL
zI3B5FCxn5!ikg0ML2DAUjgHRFMBgF4CB-`c-i3r-cHD7AA6648B?uKYBr?HhOCwa2
zL6?-#hDSrJ%!~xTxTmIunU`0&LVajp0BsM6imITmWo4il$tC|mc7WC8^@RaP7c)x%
z#9lgm9hy`myoLMw#>WeA=)sXI&NntT))IG&c-+CwYF?r7bSU@AzjsV5(t0UmkFerH
zRs|<72IZ3S@<ALn=&#yfukP=B6&xIVh)V7}?=QK7FVXc5e8jpJBzSP2dQHm$xTcd|
zA#Zv;p^Fv(uxEwPU*Fr?t<jn%I8quDx;AbE>a6FDv;<x;PO&ZNSy>Cn0r;iRKiD&{
zTkVDD0tYMW6yg+4i^GEFB)j)KhX*c6WjU=Tl{jDFe(XnLd^@&m!q%<8fC4)9rq{0)
ze~|pS;y<w!pI%J^Yu$~SP=R55d>qlN3+W_+j%aHtglhoJ&(UB}yE!|nq(5*gDgLqJ
z+M`F0*sN3^%L0(?ok4kqVm?Tpm576lbJ8$!$dNYPp-N?$q4i~FW+@CY<l>~SEsB6}
zbIx%MYe=uSPEwXS8RGQ3sdc1(d5Wz<rfMHf?e-G4d!VCpYvBqjt5n*d=|pY4*zCVT
zLPC(!HbK*w3H_EUY(9|S8FYO7Lv33UHk4*hBn>$PXmV8@CNMasp^Wvw5l5()y(~mr
zkDunHegrNhQeX~~%{dtuELP6otp{pbs+9MNcAwRDC0a6Y3eEfqb`6l3yzKWQl?~=Y
z+C^k@C-or~_@|4}0G62Im8O{wMTiaM70<1zs=n*CQpZR#E1Ciz)`Bk@M~oN6waQi(
z*k0EXSa{A~H?T&WSO<xk@aZK`tEy^h$oy?>Y>MDFs{ab`0d#`?(VBUUoJVDHPsZzs
zcKb!2TIM|2hr>@xOH2P1w5jkRYW=iBqB6<u=6Ah>_wm<6VTeyWGxn(%QFvwQ<hom)
zQ~ouQ><?p85g{QWFeYE&;2@}tm<QC5W)$(iUJHojNR_?q;?NS*r-cQ`C!HFCBoZzF
zC%+Z+<#9?>J^KX7%EiZ)!qD8En-#zjgz|xVvAf0ycd0M#qN1vb#;an*R@j`{#-NfB
z*WYg-79>&zru%4)LBv72vrK>HT0_sAAh`LN#99Gx02YvbPk}TI@&1q;9VSqdnTR~y
zrwr5YlM&6F6YuNjU^uNIehZRiaZynjDz*!CV$VT8^JALo%Fo_s%bkdTkKRx`8ICZu
zMRLw$Hb&~kI5{{hBIA5)ZEQS=Wn$4|5uqIzo6X{Uls@vfLyAHeSZS1$1<0K2R4)`%
zq<pqUSr`59>p>WXmyNgGt%)Lzhg)BdGqDZSi|mxsULlVk?S@#JveMMn)^>t|n&;8g
zG&?Zvkd?L-{^Li_$=Mu`Z-L2U7nB8{eUk`wJ;^0z`gEvxb0cvz()*Qru9V5_<42Ev
zfSrJfO2rya6U|#lUVTkS$X8oisfGZs`MoBJzLSU$XngDE2@(lj!Y1>b4{vq2B)kV;
z`XrZ1Gok+vhc9@(N%mimyvwHjM4V$_NWe)Obcsnmp_#C}1lPQakG>dg<z2htoUzKJ
z(h3j*5jKzca~rgdBVVXbjL&|7tKz^_A>@N@^J43&&>6=R_IUBL|I<~CqSanlSO`w?
zV;n08^S4?d;Bj|zlV=tc69X%Qyv?xjgVWNb3>3$3w-8O$&{K(dRPuZWv?%DUHfQ_>
zDF=jfn#woG4IRK-!^49esSkrZ{&j6jOD6sy-dYb3NBjq~CL`&o^mx&;GybzJ4-6!|
zyfVFpNMW2{y6_`+adAPqgWfYNwK5$y9iY#BJb(y+52a{*bG_sz4rCN+pan&tldM00
zc#qwL>apz;m5aT>nENaqwYc`kAr!ytcmQ>>FuI;c7XRnZqe{`Jqv+BnT+(Ia=2m2%
z7#JT%W!r@Xh!Ywgy1KF=8N^8sE1D)YB?->;5aasygj)(a@baYBnBA5zQBq$&3;8Gd
zx*f%in-DQU5e7RlYZSJGHfO3wb7oiWj;e3pmN3?PU*{!m?nU&5Tjz8Ej}gZ;ieyx2
z$hf3}FD2uHytoL~5k_2oS&_aFusV0?4c;o~f68jc+S=*xX)0K|AjKf=U@-P}>&CXw
z<HCV~0nv?|uAIkFI_s#CiAmy+l|WC6ibn;GML3~~`1q}7KJ-gm&f@`>2Yevp>C#I0
zKB}30olmd>0fn(Yn1T@&GOC1&%t!YkCzq&&Dh#K3jDF3%o{qe}E4zS5ZiQhJl*t}W
zPKW$cRWe7B16;mD^gi!blXqDLK|yiiM7*s%7cq$lR+Uy}B>u3Jf?MwkZer#Wp+NhQ
zg(GXCHMtYDz`kx<Ev?q3&Qj$W7_5gNTqqqNu(V;qNsRzjVs3ldaUyneRs07RfFm7O
zh48kVh3)ziWgMaF&s>D^S}$lt{TQaZ5kng<GApIQaR3YBM3xb@BGMD${(^%O+g^h#
z2DRSQx7`9};oW}UOM5yl_Y${fx^;W!u{lx0z#Kem;aNC%3WPbh!&;8tWiU&P;Aj`a
zMV`S8!+P5nZ7c`3YvJ4hfq@Vc;9brp+kvnH(ghNc69s&MdVzGTnpl`X&tX>IfcCgW
zrNY<pa%j|%cStnpteMcJhjo+J{`ZTxev7r~z_!nPEN9R55x!a)g~ZOls{*ET_RN`P
z@<;XkkTnqM1sr<V%dka&@9YTh0k|Q6^4kx%3*X^by2NtrAY6K7kUJnK@7lSOn9hhG
zh*Lc)%f>FEr`Qo1YA3?^7C*nfzD~iVFBfzB1WInCp2&>g2da^0PESYo00|y?ylgq)
zJ9<$EqWBy`Fezcijo^iLR#conb?QC%Vn`mDWg=e!W=>8vLV}LDdO(=5*kO`;D6_?)
zhw9&AuQ=190)hEzNmE*aPstG+p<7>5Nd$kaGWVhe;9lbEKesPJWO!R&|IfE%{_9u&
z^Y5eozt{uI+G)ZI`WdjHA1~Dbst%$kBgSJ&s!fiJSOXMXUd1nB`UyrXusc>3R+Y@S
zo7|-MTc!8kD*~q}`sO>n{;>;*jeQ9I&Ax6N7~krensDT{w)&-7Amj{Uu!OR)`qz%`
zCJnS3{OGy)6?-I(|3~Bdzj^>OkL#s$b>mt#3B%>*leK)rJOG7F1DxYKy6CqpALrD4
zH`>`+SxKBr_?xQ&+evr0{QB#H>u&09F#|z0ibd;?j$$uKT9Q1?m5a+V1yoWle;JbG
z{j<&77w+YlDi!Z%Wo-?oh=lhm|5PvH#6i6Z107HrL9-c0b%bJUc5W^SKqbOFL<i&W
z9ob_P^aa^a6buU@BVm|x0+NM*fMH-5AY%b@L%~b9(%?U777swAMAGrCwpL+n5fKZC
zd<RZcL#mHDkg9CJK^jQF5{g9Rq{z5XgTrb7z;QB;7aW;0b8}3?0)M|cUh>b}{TWu9
zGrPVVyt2)N-=($!ys>!vh;uRFH;Op^e1cT^BOy=UShH^N0%H?pg<RSUhK-cFQ!%nw
zK7I@iJEQWQzLwYcI{}-yTghy^yu|G970VOLiSQsrcWSOnAE*Khx$5tn=H|9U#{k!1
zzp<99Z;f`^_3Nc*m0%|C<nS;L5JLhkC@Hz-9~u-TVMYwe1XT@${CEETu_|FOy7KZe
zMWuiW2dOB0Q+@#!fl(3kgN=pLFz>+yVm8iO8t1of36)_#ftDkABUnUfIw1EF)BcbH
znfJ_h4)s@hV^VGzPLybc4g3#Kt`I>-2SClRjYq1zj!X!`lD!|iuWH5G{<+#O{r0He
zuV>$`J^NL%&GVnL20on1f16ho(W|l;Uz9g^8qcQD(zuQ~uPWci1mo=LyRJQhd@%$j
zMfy(Z>4b^doDl|72`5t(*dsNAOCj7L<H<J?a2VC#KEc>OPD7)43Tpw$QKEd}0-(Lb
zSIzfe=njBA&u%$w(v6^Ve}4Lx*R*+2xd%<E0y`uS|8!XM;kIF^nTN!pS0^Or0Z>Jh
z4+8~si{y5*MLXVX^JWvpE&Oq3TRz6``Qq=Af3``W`S<DzRuzz(R26nO(t?+HqFzo(
z=?8K-iDp)%5abi!0FHqAez~<6DoZ>Yco^y!H&H@TaS_B_SeS5R3PSLMb$mR_2v@0s
z2_!0}@G!7TmobUP2W;Q517l0nl<y$#K#QcTyj;L`feHBn@&dHHP6GD95gd6PwyT(+
zhNl0Uzem}nl>!;8PMKPb1+KY-eY@uL>#wJD{e4fbpF;k7Wqt;r<30ms4i3VfNyKt5
zxjz02T%yguo0w%p$$Mw;l7ulJ7_fjjHCNH|Sq9wXrgv>E48CPtF(uda(JDiq$GO<Z
z$Or-yOJKcd1fa12#I6d~Ip|yfX5RqG49%O?Y!gl48mO1Syu4_(<2~PzouonsO!&PW
z8;<Z^z~-rcZbvD-8}{yT`*~1HU}w+oGk%1ZiWDZbVf}fTlbhQ*KC5g1-7dUFQ#5Gs
zh+$^SI9+faj}9bpe?z07pGcep;5zj<7Ze7h{$Lj$MIgXv5b*zzl(C|u#WmQVzSywu
zb36w#qUy3T;Y`Tad)4y?bdcOt<;LNb@YvUo^LMnjiz=Ww(htamAnV<)#2VZez`n<d
zJMa?mcwPQ|yw869b#JqMgQiDXGt+TTylKygy(C%<W8nD_l|!=_R7*FsPkyhiCT5##
ztgpK1Z6CagPKDa7>TO7`k4Z_h!#5QnhJGVO=~FK#u|(Gm|1?29nB$=&ylndQn);C-
zd;9stDEt8B04O41%BI{4687X|H@NKZ978ut4}eE&>$0Xlpow)LCup)oFrgx3yKCqY
z6{UUm?k_*)%}_}m1QOirRPEQ>Kpzrhe}kx)jR(JMOxPMB<EG!qUuZRb4+-N>+Fiq?
z872RCzMxIdAa?I5D>fDXo)7-H2e(Jutf40P=U4*SgF&5EC8n{zbDmCj*V_FA3HVEC
zDdbv{J;mX{!D9mhEdVo!ekba?ro>Q01!zwY!B-$DLa_Y^X%L*Gu$rM_mP=G%=CIJ9
zhD;3rI#F!_qJa%F?A7`(_-SII3wH}BZ2>3GEzdJ~*$?`U5Pk_*5QHTj-ZXl9$m-FO
zfzb@Q0s9UdAUX*1mpQ!)^wFzBLc59}zllhLUMC$bZEjAEm8~t_c>#(BJeH<JaS<@%
z-a0$qVdNCJ!~5vhXY4g(%?3JrV%R1g?(V>*i!sp7+s6lOy)o?j-28k+91+-{I3jG&
z40hEM6cBL0{o?c?CgGx703|{ODLlLz9OfEsW@TNr_rVX3LR~?8sEdp2!bhD-FO&hK
zyLW^2$!;hhQ0*BI5J$N)P6wXP=uu>r610D{9BCszKe+6GB+j6ic}X*v&Jw}^@Tn8Y
zR#sL3Hnr8%NHi$`YU<9jNMxA+DVi9OUbTsn;3F<J!SmB3=7}_4f;!?0O2m2f_~ds4
zF&h_2<_IFJBUEX5hL|k_FJ%Q{97kOM%B!3l9-uKCPnvnQlU_b!#l~p1os`ala{KdL
zBzLzC!X3}wT>oHGIa*@zoWjpwcw&4)cgkwBY`rLWl`fW+kuiraFFUS2GR(42pEhy%
z=<MUikLd3y_Dc)~w*{t#$6P54dNe;CmA+~`Y3n^N^19oG$7U|2iSNDX)!cgnYLseT
zO<t^m`2|yhMSmE7^X4uyQJkrJ=EJCvGG%KYtf;F|p6U7``Nnnq(&gBpv#BD~u6zOK
z`_uKEHAb!53{;9ncHI}epjlw|=}cM3c<3GWJ7LAv4zwFpaLeg%G;<Uaao%J3klZ?G
zqAOpq`ZW@c=Cou{wE`=d+<b&2(bi?6BVhs4K3Q2=&`wh0>mZT>XwsU&ml*??ID!65
zMrLLxoC5t~&}dDGQ${y&q9O+!kNfF><K*nWRaVj$D@vh`#F5@%d77A-G!gM4KYtbJ
z6bA>#ti7MBb?NNP3`!#+xtZvph0P{BrE_D*Z!aa#V_HfM06mpFVyHC=>d_gw;Fn7)
zPhh<?(?y8dRSsM$sjFi}lsCH3BK|y~F4@n1;>V{O{?n{j*sEBznHf!BGfgX?J1j%#
z4pUv7bi8=1Q1^w0eb^P>Esd6NF~qn|%-!f2YR$l4@zjp};90#0-b+Gqghmk@30g*M
zTO~DXVBPE)dZsw-$)+rUEzs$$0Rm2lq7MxX#Tia^`0$x-Z3OnBWK3>!7Mw;|E72%V
z#jF<RAgX-LpOpNpJkaFad|)WUnM8f-)9FsZEvXjqzQ}qHwa;$qy|#Z&J?wvX^UqyM
z@dMOvDQCLNe_d<p%9Nbqm0v0H`o2lG#(VAO7uWH9%)yj}W^x6hQ3G!JOevoxdM{1i
z*brP*Tq(MfOq(gtGrK(UCf01WS#a^|9S&K=jr6p^-k=V?)cEt4TbJopg1G}t&hb4u
z`&-1tm-(Ef{>(oO&0ne_eEq}K!a^C>qY=0({TNkGu4Tk+P!wqYDfZZGaX7`=q{*^k
zAd@NFqQl1_?m^G&;6Ze}#n1O7EcXFtg}-oINPvhTl24>fjnVN4vS6C1H1%?!ACD<+
zOPE@;Pc<j+E)c>4Gy-bYZ=Pea=m)|<QlrsR2cZqm6U+snn>UB_)YS;O0f0aIh4H|J
z+TC<EEfAX!G^+`Y4^T!o_w)z?#F}XUex$taNEhg<5OP9MRaewfpZiBRuUYxWYI8A|
zm*aI!U@$l2`!mS|5QgB4KkGY|l_4)hANvQl$p_YEHMs2qlTG;YmmH*r=t9P#ts#IJ
zXA6a(ZuA1)GKRt*$LjoOm>#_dcXNt*Cny-O`av%Mb4W)p@f$;HU>hwl(MhD$y`>26
zw-o%C1npNc_Tjij54>3nhrw<34q7fQE+85hKct+7IR)C2-)H{(IoM+mwLaRW+J|l$
zQd#*mb^Kz4&(X}>QIl94Apsw?>4g4V{qsIr^7MY{+tj?fx061&D^v1?Z|je*-azAB
z_1>4J)$bOU${vO1r}o8h|5}yo;l9vxby{<Ra)R=WJ)2?MQ}r^JqulnIXk5KKVf663
zE?-ofx=iKGM4z~{+^GkOyA-Ey%uv3CyQmk7xCqzD``Y)<Do4oVim0xdB=>8Graox-
z(f*@GmqXj4u4TSpHaD_w^w;I0ktGTHOTOj|df%yQMw6>5(4n0~RszZ%T0$d;<On6`
z(1I<%38n=_@Fw4RNWCmDHpecQaK|IK1zXanKcKD^*e$cu(hguXz=%=)5;Sx%cxOS=
zR^Lytuowe|axSK#qy)DJ^2<-KTLU-`Ru0fIV7DV^W3i~ntw^Ml%Sl<G-T}{(>;)gz
zC*qm~n!f0Wt}HEGV8G}$V%GffLdBAz5x8^Vb~8?A2D8%BDR>^;#Q=hTaZen^D71Wf
znt;o}(eV}r%VjH;^O^R)RS_*xG~hyM;Q^9@ekF2pxi32k`v(VsfZ=Xb7A1fa#h^q&
z(%r>P&^8dqGTKnPm!N<s8^GyIV?XD~a%{~PSXKiR;}gZa1WDBd@e;)u8f<`;9dOtY
z*(LHDR2{^zi`&%qHcP~KSJ#}6(8PNA5(A~gNd|SBGN?f|LDMI*^y$-WpgN++krG(`
zfNOxj2(JPZ8tB=Na~p>&@kdvJCt&<DCt}S6QmtM4PwMErDVx$P`>t!k-|}h!{L}f9
zKQ8RPjzY+H@g%qiw9u3IgmN5uzDr%@tkngzk{5bsccHww!o*-@RYCkQ;!nU&Q`7XB
zm70gXsSLCUYt0kqo3i5BvMPUgj5#J1x@WsA_WK>CG1$rW>na^*aDA@I!wCy9USAbx
zE2Bc^LaUe44kwZCWLaBUw+q`PKQG?a(`@PO@<Mz=o}^UO*Q2w*+6;^=%!4CRf4qMj
zGzG|u2$VACQN*akYMxy&k%;47k+Gu<MVO0PSXe+|1(X|Liq)^rf8GDei{cspWbnu%
z(2+v@{j(u&x$o4^_6Dti;qF0Xz(BlrT>F@n<&NGV+HOeCk<e(NM~pt>*cg!b9<XyJ
z;F<wU9K$c$x`RBWa^N9@^p!=ki1Nr0LOF}i!PI%QfHBPtt=)%6G|*_o^7W21CC&-W
zmm)>_Kp&+w-YqUHxH~y56N44OT1b$i=i}o;F^*t}(T#)v(HWF5tUr)hbf>(g2?6oi
zvRUT!7emc|rl)1p<E0P2%*n|?O>nXm^-U96yZe7Uf+aj6T>YmRXEH|fcmhEJ%OSHA
zDeIU(&6`F%9Dxlb4=L;R{(%8d9Y8VG1TP`~iG<qlzyPgsJMbd403V+Zm)u-<J2;su
zgO5QYkK7Ov-zLcG3BeokRpg(9?GqYq1hE)8F{1x0>2yjDr+Na-eZ*&9$`?_r)2oX}
zzgD^zWsb*plh2jxq8ZN5#nzdL$3!=drc|!8#W)55ofQ2l$?$w}_0K83<ri-(c})-E
zWb{pG$=A2}7LdDqa+T9qx#4rkZIzeJ-p8ky$mKZVlnyoB){mchcB@<Of<l&3yC^f=
ziPq=_36Y`TnInnA3Ku!qnm_Ufbz%vD;6nf)Am1)UfL%!caHrGTxAg;Q-N>b(#w;7a
z{?bU(aq6A={{8kq&nqPD$BsHe)RB$eq%0Q@WZW^~1#D{|BJdlSGpvUotZHs9LK%i9
z<Kyi$e0c4|Yyz@j@r!XKwYA20I6;$%H0(NWO%KfA@V{cW#MYy6q7T2Zw-l|f6F+}p
zb+piJ3qp**vYa;>LJAJ+gTwx|*49ApMsY$CeGfP@LxI`f%#5hkXmI)?U?<c(7y0?u
zkwU@d@}MGSqAF69E+D=DCU62PLfH&U7iz_t@85~mA*!vi#Ssv*S5ZXdSxQNQ<8S~r
zF&^pPaJy1c$J$sm)K(PCBnA>*SblGasT4hMNTtvwb78C;h=*KKL`-}sun)Oiw^USB
z@ozP%DeLkQWp3C{r{b{3K$uQ&I@CaIn3)pu-FG5E8U&qzo(Jpk5B5FZ^?X;Q`q9x}
zC33{em42i{Lft(;lvo!+CJS>G%eq;EUUMaX$`MJFPZpUxd4lR#vdU?W^Bk91Y((Z<
zNx2<<=m>jv2sWqWebg_Ep3hrrz26$4?4iMbIJ><#rJVPUTgGzfpk(orJLVQm%M!hN
zVKy3MyyE@)YMNQPf4EBb6N}DKmqc#m2A$E3cu|vcJ2~ePJL{fa{16=?=B9t{VrW_N
zP_}^ET1LZ+NDH~V!KXCg;puu;!}(XPoy%FXi_>kE9_&_$GmlYywWU4rHDwq^&khU%
zeUmDT%)n6;L=7Ee9D7VVgxG(PC>}d{|4;w;xS^c^on0x{V{{EPd}}OAPibpp;%ORo
z7bs6Jf{~Ib#dZagFv@<o;T*&68?Zbmg5%Cz#R+807z5Wllp_J0BaSJzz{E5N>QQ&W
z5_GAvLwrDh$G{2DG&?c#BuR-F!hi-3IqN>h)di!j^Y01M>w6Z$m^Fw9K+s&(^KK<Q
zO?guM9cj?l>K_<I1R9YnH2ol|W}q_FELv04w9m9%G@ce_n`XG{cNTOPAPD+>_c-@^
z%TPS@G0D~E!%@@!HT%-=1<<_CofEP+7=1@WqZ_rI{1i(dr=FJ|=Sz)q=LL=xH}wc4
zoV~qY)m?i2VenEnjX2U55)@9yt^_b_?Q5Sa$X_UkROXfJc+?M;VvBZb>!Bu}*g)Pu
zQVju~M`NGg{yF5K$xpuhH0{~@?N0aI6gjC63G^<S6>qLC^(-|nWh7;U6omM=`XugA
zbmU((@aVCPOZl2m|2|2QBXh){LDMB>GTi3u^y|pPrUSS6j-K}vN|{hrO%}@fCotz>
zZ)qRN6kShERi(3(TaTTVSuyka*Tl>YmCl`Qds+*XUzP0NdF+#X@j+9L=?C>w*Bmug
zIi@A<JuzO0QuI&=QMq`<F30M0TZ7xpvHY7S3gnk_2x_UdoW;PZonQemdFK8ha{~i|
zfvX<CS0|blReTzwH%%%%kmg*mIp_Izo6DZ?N6bF{Rme05a4kCKVyX%}foGh@;6DDL
zP#l;C0=2>#^M)8T=1`eltw-gOR2XmPq#5HyEAD(&uFw@_gf-m-%ny{FZzWS-1KF&%
z6q!k4_vgrZ*oanI%$bM0&H^F~7g=AGi2mHgrD^S!O|b)SP_~6GE$X1JZ~Dl)5k1HL
zklp1N`;C}^sSVYLeweO_(-3gmNyMO|N6#q>g5YFffA!iy;F7ck`Zlt0EUVFC^M1R`
zP#u-4UB@_3SZRD*q7K12P270BI6E5*NB9z3nX}8@KYnsoD?Rt|*<|50QBt{SLCC2C
z-%G`a5DQ~h#KE1UbWTK-Et~%N*Y8j_{PE58VTsn@x-)`D9=iBk()>}dG*!mhl$mmW
zj*fm%!&OCV`b-(W`nmFZ6n(9;s)FHnHlMsLpquuTugdjNak{eWu$iF9&v4Z`lVtvM
z1-164`7IqwS%;2E{^XNc+q+>gl^T`ZqUgW!{^u@o9+#guBAPfwu13hkZ-)6VJIDf+
zk+a?5xVag2l|AV%9R`#Pa%<C2W8)h*n#U|Ui%%XA_<cDmN-bE*>tbpUJ9}1UCSkW#
zRCs5nyUJh%*Pa&5k9)n&^33%Dw$?7+msv;*wOk~}B;TaVQpsLA!%6XK6@ia2xhgJ1
zMpcwDhh)<>2qsH%SL~f26d7gZmrodaeO*%0J!ULCMjjvw;Ipy{VseZ1w|A$V>0e^>
zAx7B%94r9(f+OZ__+Qr!0swixqA79yb<FQ&=l1ts<s3}|q9nY%?M#m%2SIuphG!og
z9SxNA2|BiFLtqQ**Y)F-!`3`<EfonnQY^0ZyKGMZEVR;4lV6|bte<-!!PoMzvPJ&h
z^2fh#{`a?5e)GWJuqIVD{PV|H1b@8odu}LGFg&Kem_5xcQzFXLjV?1FgF(0DS;?+@
zH!C!#EqNzrc?ODWTWB9}ax60>^07{RuawmsSuGKt_Z8iKt=`9iees!)n%g_39(#-I
zoIm_ii*tFiDIa7-H7)0bou`U(9GBbgSFSvy372QY420$PN!_!@f`iM)p1qQ(^y=7o
zI?=(|$wWv|7VNzVGP$8&5Jfm0C`=~^V|Ye#Og9{my`jX)%gqIRl!E#X*)e#{O=AXM
z<fN96Oo2_ZjNOJ(VFJw<z=uIa>6vc^%L|A(60&PhcqVt44P1%u6+HT#|AE+*q_aKo
zX%-jZN<k2OBqgZG$qAb^s`D0J1(b?l{Yk{W)H$_^^BH^`=F^_nagu+1#lmt}lOy4D
zsHeDF`hRc%tlF8b9qjGxoeK()4^!ry32n_o7zRdcq8qou^Z^y3n~Tfr{Ccqoodjgj
zZca^dDIM;}(s(7@{Nb&EW;Ozc^XAI(GQf~T94o!;T2K^3t1U6{^B3Z?5_YAbvb7Ub
zojb#O>3ok#boSFXZ{Ez9e3_Jx?h;$BQaOrD|KOF~uQ;sO*ok0$yXqeDF}AI^o!n<R
zTNmc2wECrPO^|Uy<#JDu{%r@w)?yD@<3y<%pPbKn56@0VGktlc=I$@))+L!28tAQO
zGApQg!~KbSim+N{Pg-nMXPK(nu<XzCSP4|BfLPT&ndT|_Rj67xhQ#dI{dGp&m9A%M
zTZ^*czOWZp9nfh3zrm<uZBd~P9SpD%-KhAo(8NHPA>Fr+7&IFrF0b$ESlZXy3;O~!
z6iWz#H-&`Y6VHL5Rjp6aJ64$BYTY7twvUo5cQf4dsFO<Y>#l(azG?5IzQm|d7|9?<
zk{3sl2&G#>-yqEVdapAJeuBF1G|Tsz8u?HuF>e%d&Lp$z>RMj8-L$0XSBWk{n%evU
z5)gT&e>OfJB)uRZRW!Q#jnXJ6B&Hn2)d#<{PibjEJdY!DVn#0;5I!h&%f%%^s22z`
zvdFLvno6e>Ujv?^GC9FB`zYST+4y@egh*aHZB-x5g~0e|C#lH)#@>4XRkd|n!gviR
z22{XAQY1&k00@YPN)8ewtAGI|O9oMti-3Y;QL-RO2_`@!C?HV~m7FAJP>`JdvGDu)
zz1RJ^-rv>TRbBm>x>a%yaL(RquQkJ%V~$DA)tzA~RtE>lm2~UNyV-?jQ&cuh)!JXI
zZm$x-j(Pq<A^fUi0E^Xb0Un`c6YjOyTTY2BhMzPyNHuO#OtU@T%BX0$uP!&SSpTyt
zLK*g;G28TyIqz>Etpdum3qr17dA+e5L9Gz+BBQWnG%Qf}8s=xVNJL={FziJdnXQe@
z8!^+$lxGzIIdgBXZtrZKRHQ2+5`iNA4?cnjR~4y||0=UUwk;}Z%)=`1%My)}63S}<
zH7}Y3D9heVu<qPOv!(-yNasl*A)#7HgG;902>2$T%`d{P^zy*IeUh_{u~WWnD0T$X
z{yIS?x%C~o37$uXcj_x-&_3PsYD;Ct*5sk9dk&}*ED<Q4cRpH}Shf0iE!R*3v*nKa
zAa!BIl}K)-;&*VUXCWEt7rC-8Q|2RT3Ciu$uN57*U;V!B6@FbQb#A@pL3z=wY_oS*
zIGj1crXLB6TYVj8lzVz?>cgYc&uy&7-8P01byD!x7x7`>;$ZAm8v1;aL){>wNw{$%
zLyq($Z$m`nxFG}R210L9<l`ix$a4wv_m6oW$175MkJZTFNY4pw6l=s>AI)<oIvj~F
zDmpbH!FMAQ5`neLy*)i-I`+HkoQ`EW6&faGmoYuD++}~2g7ug}%zO`TSj?B-TT~fs
z;^NoB<?1}{jSUDlRCuXkMepO46Cc)XvAxXpTbm$=G%8Eb#d<n67>iLDnd9rC&>58M
zd$TrlmLQHuwdQ_&_(4MObanRSSPtt0p7uu_4x|k0nc2knnu^KkIShVCiV^iZs@Il2
z7@M3IljrT$5nA}UPMYT7$cYwjIm)ATnsou0A9KsU7@4KU3C*_eju6OYXMR|sd&Dh#
z1ISxa^N-wW1*m^H36;-xOypW7SMW{us>R9pW0Rn@Om*vBuw=p7<;MgoD=Sf;9}o@&
z_7?I!{SqmhaE=WTJr^|eKl2HB=huI*7Yc4m`P&8{m*g_h`{36#_D9|hqaDc0QQ%1C
zv$;g-x)6p*ZZknGW0iiLVIaf0afI2>OfNg9Io_EKX~x-9b@{0Sd-t9PW9?GYQ*FRM
z^48(*34P1F4iI5{ii|F?7TMT+<0k<^oUT7_&+#S@d<DG(c;qkMo`{UTu&7ve(%#;l
zFNNoEc=skbI&SrcFIKOOM?!~E3^Z#3R|ej^Nqd!a35aNUK1#Snbi5r;LVw);Az2df
z_{pq0<sEX4CGFPIiER~mgg{t~Hb`zXxX9*#zGBu~O`Ttn54VHx?wb$3v~-I-)QLBh
zO-)rP%MEDiwsPP5zRYQYTi?$myt|y<t#$s|kGHp-Y`=cBbPIo-zr3XD?12cAb9sl`
zWln*(Zc)QA`1Rjo5!xuFAbV#A9wsb&?(8@kCtRW^?{I~dpr^(VAWFy=x#@(`M_MnC
zL3&HzplgJ33s|Da6Pcy*ekSw-k!#*@X2?1_4FWDeKDUK}nJ4>l3AUlw1Ig;wB%gJ^
zps`0Cih@Qzp6qz*Q83<6V8(Gcq+Guw4Oqa|kx&BnX{XxVhcoWft_7r1^z9Rx&1AOl
z3;|Np@q#PMhHv)X2(1}g-XI2TqvM@gEsc%ZB5$Ue&5&%B=SxXR+2)rGA^9!ONrn0o
zSZsW$MjP7Suc3yaG&ykX=lOHpd$)=`^7WuZ5w`ayKPC0r*w;QqwLXXZs_xmGGe6yW
zoYTlE#g4noaj*Wh8^Qc({AX`=lzE6r`sH|asdZU!B&CIXOT4~c`Lc4^T=Mnj*XQ!;
z?bY{t&9B+WFe`i&MZ2%#qWi#PNAIQ9!boQ)i@}nt<HS!ySDjs5N8Vtf7Tvaex!mai
zgWq3`-&fv^z<AuEw(OOnyyG`plrmmdHvlowWNbd%t3jw|<4d&wOa`)~cdMuBX^TBW
zokRYaX+9u4up$neFRY4St1FB|I6V_C;_x%tScY-et}6vMZHa2>n*a<VdH8TEA5BhN
zd_2Y#z6In7T;O5&7gkYGT`_tYgaVsKyTO@jhvq0Kt)?5E1AofG#3d_7l;<GnqAm}8
zSgzxPs9<Bbdm<e=(hOZD?nOlz;*8W%eFi2F^$9tr_eP8k7kcA-Z+v`w2frR?>eW%T
zvk$DdACAXL7?$M7Q{@1j;89b;Sf$wxPv#pfmQZOd&k@6>IE(%oD@OU5UN#36r2m$e
zHZyJVC6T`UeiWKvghU)Tw+6@u)lDX+yv-uor3=0jK26%AANN=@-OKlL3mqIBTf{+G
zC$x(<l>za6`xJGnrr|(C)hBbSzG9tA9y}Lyy9=9`&uqCL{_UpH1`aO~8Y+gd3s(`g
zk(RE*;t-VD(=%;|%BRCNo?2~cKsV2lB0w+cb^f4edv(f1;L(%2gXyLd58kV!Qlhdh
z;3$xL(%3j|bZ7Vj$69X9=ey9THzRuNm=hM}*r@utPWca}KF4-n1?ciF%`Sn!+`lwY
zjhu$+F^c@C9RPb8ZOau!*bXGlD=l0rSzXrgkVn{k%qe+4{2gc_5F5fih!VBu$9Qx!
zn?NlJ8vwy6Jn?z#1xidm`gkLOAMptWp<zgBBQ6}VW`wx&Q4C;%_dy{RfD=BDYQqNq
z_md~A2T}>EnjxI9O<+*bFd}#6%qN&wcIC(g=#6-1(ovT8Ui<bMa&$aQ1v9V+@LiZv
z6CW36Y;27ASnQmf{a^M?3Rok&sbvl!humZ|0UOOq2+~_Ybg%Chq1GR#BMZvMr09>6
z6}7cT4j)#ZZN~>&<FwD3PlD^jmA@V!1pDkane_V>2JVz=-6`eI*ZugAzEWmqL~Xn+
ziG1roFYe6eubbATaalH$?TCCWuc_Ae;t%l;ES{JxX1<v)<8Yh4!7*{0caxrQOyk}0
z;)-US<`7HXngpe&&#pqEO-a&g8?zgA1qPiO6|O1{@A(zG<3ah6rsb%`!sEL%U#Tm?
z!_)K7+e=R;vnAg|2>V4ThT}}cxi40dwcDF6#dDskWKO6-_XwoBh#JD1aab`@67tSR
zBN>tJ{Ph+--41vTuL(TGa}=R{IN}<R_yao$wh&SulOm-f^oSWq$2>zu27HCo37&Cs
zBUC|MQfo=eWwK2@Xg?QHXiPLcbghopDgqBj=BCBf3zaEH@usMSa(2#NDEu248W}?P
z%@-#PRkH<1%Ppek&-9xmFD5gJuwN^>w+&?aR|cMY>%imw`Exl;Zvx$yXWk|eSPJeG
zP~z8XDJT_?IeG}=OnXoe^?%=;cnJ*m&WugZG?4J2y5KhPS;({pjx`(@j<&toEtIQJ
zKJzj<u85r`Vj5JWSdYXB)37T6r6+3UK|pduf)=#;eIj;+3TBV>z$Ahh`MzT-YF8H#
z<i56;!IQtLbNMsF_U+qK4RVt;ab?nt$5ehG?6pI!0#U$v_lGOJ4Pr1AKsx1>iUJ=!
z5{Axh?b-qC#Q3Z47&gwrRv-mwL%re8emUZ^85L3mpC6Jp>$j(}C9wJI<m}gbfy&<Y
z=~*4#FzTYZCc7J5HBaf~D|uo~Cp<oHNZPY=qNW?dco&xMy7s0m=@0jGPMXN;YwdV>
zxP_U66jp$6Kpqv<OHdWm(HQ_V7R6!0)I)-(l>Os3Tqj67{B148#U_}T>YX1Y?p%jL
z3c5?c#}MY>;W>EF3e6GdkHHr>Xdsl;w*?|7;L;j<j07FkJ@j&KW^%U1gMgxC2)72I
zXyJj2fZ79s3|bJ;$_E>f4GRUaXpAyE-Gg$@grYA|{qLEaK-6pxZB%>;9#th2`(<Pr
z@Duz$!2TJ`9BUQ0yJt4J_PMs2h@r!!OTEYi7Z(;#J+*=bL1pO@<R~~e08zN&z#;_(
z`R*i8{ppcrjzVr7G<Isu$IoGg=da9{2T2y>`ok{{!O*>u{ry=v1mnN88PX;7h7||a
zdM#j&khJnDO9wDZf|jJM0~_txVrCRP_zRr)zdbw#L@l2p1BFvOsus$6T3R~A^-2&}
zFxcJJQ!xcZ$#V7@2M^CR%;~^FLt1|nHwpF_fG1QvZoqSZ$^_6uG3-7%$$-zxqGomE
z$hCj?`a)3}v0Xa|invy~7rdPhch@s@hdv3-nX1&$H)YS2J!V%QlqvTpWGz!^VJP)Z
z2K|R_;#~4ZAxn;m7r&N$^qKF9BMc(q3vn|9UmNg=!2;=+CjxR;GB)l8IdP4o!|#6o
z?$h}=Wpyc9%GP$9*dVb7XUMG?s;r<e3~@po4^Uc$M14wtYZ3g#spR%Ka5yz~zB(i#
zARtgIzRi323~cM^(_SfWFhUdE$hB5Gc6AxP0afa<i3t}6_c0WwIHFu)D^fovL(z{+
zxp^{1SpJy2c%YWQrL`5{@$)eX(uRpWi(U_>5uJ$RJ*ZbfcodeQXSdhh_3jWoKwyE+
zwp@7kR=0(Q#zvTti$H!ZjXe2403h@=lyPOyA^rxBlw?=oGkQ5-#xROGsEjCW-i-hZ
zHXe6~!!F$ZhKmmGM_7hW<~27SG0jX_@4j{W_VLuu%bn=eL$V0}g|<7@v(~6zVu?&0
zzo9(<QHZ9zpi)f1pW@D4dv=EdyekmGy#+$F+OSQG_MS$9){(R?3wK^Ng$T=3$kAt9
zJC*-#(D3Ds({ESLF3n8SJXT82sC&D{4s@3vjCb0Na}QqM>3FiOD3>NG{`kCKyW8_M
zDGNed*GO&!)-M#<KJ!iXfsGpN-nLI)u><X9eqK=2vPBaR6rW$CTb*yCGs;)Ppb0(h
z-3|Rw>dl*Le+`!3-+c(Rv#0WkY6Gn~)2LaO2df$GX?#rzjS%wc;W5KdI^gSdu1Bo8
z439`|UAKh+m{J+x_P2+kPaC8zkg>;ldg6_y@9$fZeG7HhZ=)A6cOhP0{(1gWeNcz%
z(I1YNe^usOM75fXa0URck+54(2rb9(<TxEeFIjXkp9qvqH!LM$k`P{P)e`+H5>H>N
zwYbL^{T}G)A$|v;yo;Nnh^Q#$u?pzEt;FOh`8NReQJKsH4fNfO0tT~W!QQLtgP8P;
z(>d<CjLP?24-djGOt!gRwin}qagdVP&@u;lkN+#`9Uq}8!OJ^5mcLwO&kzqd1ep`a
zEx{Jh>AywXwmap>yi<icWg*F@!6En)vu$wqbPos%6uHs&MZA9cO0p%GP&akwB3xZG
zs|<FZc)PmM>S{^!$emHv_8Rq$--X{kJaO*|=)B)3uc=fQUTAZ?X|5)Ta#q><ad^o4
z-pa>zDb23VKbzx(WM}I_nWsyxqC#nlo;8?L)Yez1bj-&=K;i?H4jeaxMFj0ZXB-L`
zxKq-*i`MDl!zt(oL{}ByAqLQi98`BHLO;L>j8m8+GZaw)_5gyBQ&c>1{d*Z8zU9T`
zKeg2f5K~|(cW|i|L0A;X`O?x7-1xzc2U<;v;lc+wIXSURsG>oP1*$Ai8RSGq53DFq
z+$$%9?z`2lkxbOh97RxohTWapwppQ3q?jf+GIDfqkaV@+Fu_{BM$R(6<{K4a$cf3x
z2R`#-o@D;F07T@&WlJ2e9|*G;gQd(1-Xf+00k<9)eeBbxgwcns2u2?!e!at2b~fxI
zN!sAR*oR-_kt#3>&|tY?qH=Lj(Vq%Fsn?=s2$zepG5QN+5gVX?S$!uPEscOV+xm}>
z4jOG5J%`H)<<%zmAfgvopnlXU2~`e#2k?r^Mq^d0pC(t8SW#~fEzR4&`Q{!~md1;X
z+?>&z?(Q;YRnC5h*P(l^G1TO3YhpE@h{09fy20M+v?^z}?NncLHu=Xos=Kx6>rVDg
zOW%wbJA5ciA)qWmc3jL&u`1jdVwSR_qGP#zEwFA!!88WKhfA0lv2bgK6~=<HyU86D
z6`c=|=NFkDer4!9@&@LH|6KZc@WtWLuHn4lpJ{k)@W&>{+9d>^@|7M`%cFm4JMgs!
zj%#`NRkeX*PcRsapc}upd~ATF!n1QO+9nOPgjYaE%iOe-V<6sWg`d>$aYWouZPE46
zq`|^~rCNomZ~~}FID<&mdb+!>H@!~6(jzi%x*X-IB;0Wqml-lhuhZoBKERVghABY3
zrcBCF(DPw;j1w!hpyzm?SX|Askth5SP6ISi#}7K>fAD-Q8mCdHdF12~>!W?V`Z>Q1
z4Ht(VShO|d**L#5Qjj}u(Y?srF42)eqtrAXXTaz@(tN3X$E2?DWs|hF_j40;QOf#a
zwsis89V(_F*B;GxwPZ_FjeL()^D~?jPHdHQKHC0Dww3+C;NhmT=|+onlM;PP3ysxL
zuCEuKPTzE}D5;7|y{Q<wX9IsBX#;nk<OPYxzL-u+T|s@|AncUH;()S;AGSd|V&mVR
zqy&?z6Q&AQQ|nomE%QEbY%~dNwW$^#@J?`gH<X&=&D+-kCs>*+%bL_QN=A~<YAb`w
z)rk0lTKGN!z2#qiZuiYB>Qh@$PFcO>8CBwO&zR)6ITzCaPX!e-KPN8ri0KP|vr=PV
zhp+Q(vpb9QuaEP5(f;%><LzSlNZ-60|5pcy+vcNRy4dVu9=G}yz;e8T8~QU!m@2>u
z)e@`y2Adx1b{*u8{eEHrspq-V2D9mj!~<&-qIUbAvxik)$S?^?PKpi2Yrbjq44_PG
z`rZ|7od_A82SRPTT3;gYKppO|{x5YYFm`~07~H5p)|6Mm<Q&HnDVL0U!@cahy8F;2
z1s#@E$<$?1O&ZNl4X8BwTsa+YB`b<=;eLR1$t*1|0h<XM=-Ru^By(_=X>v0$X}g~#
z@RrP($rhp+1UjYljf&lu!>vGXN#J3Mo_i=3r}eL{Z5-G370n#{FMBKC*qB^?ez;Mr
z)_O`nO%-Ml2W72eS|;~Xr_oQC?A_hhCw=od)a)@JBkYsItwE{rhX%D1iJ>~RcFb~2
z(d9O78i@+X$NhgW@v0~G-}nhrGYW6C=T6f5<UVs3tS#EBf8CH?rLGp>c{n*zI`@0>
z+j*673XMo7h7+o|_@_5KRp=TXP^dHeI`v#;{!1hHkNoLYhRLkuzH*+lP0Lz@t-!cN
z_E*xXhiA4(1^X*+rjQYxT#*G_@9TudBnpZ((&7+sqM&ep%0x;2{cv_gO7PZ?L?W+8
zal-QZN=MsX;X$I2M}c?$zu?c6J_}kn_C?r}W#rnlmLl;Atd>u69+Cyn<km23qIfuq
z4q~QE*CDsJ{5jXO3qG8<L^ZZP)GqL4eHSt#(4LY7Ew)e?`(lb*THYLY>sbnl;4_j2
zl9Ec2@1EWQXskh&>pxuO@9RB%ayf<oB|s^N(fBY01uGf2jjK47iAE7@k(vQ-=m4C9
z10wBDR@HS1=_c69_RVIQRO>fT+!6!_pYQ9}#IyTF=qXOH;V_a9{!jT>|1)pt|LESo
zE7)UOEcglJ&9Fd?GwU)M5%l6rkRgG6DeJz9@NwZ>zOSLAIFVhu-m`GP7t%V*eISUs
zjb`_$9?{X%tRopH+^hQL^%N&W!qBhyP+JMgQMA5x-_J^OeOg6fObhdHtJ1pDivuCn
zVU4E|R-bk!?V+N$6{G^YQp$@NK_Zji)zk#lafmQbJUk!<O22gEukHsX`!o@_28wY(
zIFu)ni3u_eH}VyEhKgekGsheGnpG4Jw-ZaW8)7i?_WASrB9uw6vs?+`0Rawpr+ws|
z6x5fvQWC<g<se6=wjH98TyZdN552aC|Na6vH~8`zd(E{Jr3x!=lq{B#f5fcmH<D20
z3s5VL^M+wO@%J7~Z4b*tLS_kAMRgb6Bq^4>Nz4?2o7L1}>DQYdL*MdY2C<KVT8-tV
zxJGf<6E!CCfidKFv#W%=?mvI7%5fdS<W30wP8UIOWHyQ1R%)wV^GNA_Ki&~dc8HJ~
z3$84IW>aTVr1Pj_={`?&6<kbE6uFo)^tcSHSHPGAIxD<4uA{g$u11#g3JOiaqgh=6
zW6L@1i>aONt0<xuRxU_?TuqbuSbsDOiTV}tMuFswo)LmAnwtU!29#j+!<1qB_kM#8
zSu`?kHoJ=Q@?(p?zpcIpgELN^SK^w0?4KivPkl`RaZr4KXbKZqv)}?Rj>raNQ$?<r
z0JBGdyuT-Te>N4A>jAB>_p2Zx{-z|#myU<Rd5GLw`MzI4wZ~oO4qLQ|rJw&VuXZ#T
z`I#ZbwyJ*jiksqu30XK55n~gR%}ip#qiV?MC1$$=SNr0UUJ}=oc-T8<{b-Py(p4~s
z3&<rMUAd%BF`i6K2u2(10-=<|?Z2nPQn>Y?%1hvald(Sha`v<oC*~ls3Y_8KJ2Z);
zAc#a-FQX-chb`P6eTPJ)H>4%YOw%bCMgPeKcqU2_fRb|vnABXv#D^tDNB@AHl+w!E
z0Y*69y6++FAHC@RyIkadWKaGl!h`i5Cu6Ys9)amtSzsVYv-5s^qKCT8SlH#rMr2Qw
z_g#R@t)eFKuO+RP;MSS=tf7dOOAZ@_5#tVANyoI(@aOqUlWBgUqgg7PK~Yi5X>N-{
zP1n*0-u{gdYKXzNhHyH|<DplBT~Fi`vil;RjGm*w&L+o1H5xxbkOZYb%Xh0+;Vh7|
z8Hb$n$S8z1@o&87RvhogWOJ9}!g{}}DJ$aHQy3p6d-@b$h;v=}&@ODp*4xOR^#d~l
zIa8NqIU;C{vGdiBjS?gI9O)GL`_Og8hNjv@C?I1TAyab#f>K%EI;4sy3xSYcL&@x7
ziL4N%Voe~p1ZO}M31B7Co^YHTyskQjV=E&;K3i2c+q#EFS<DJ~LecSB?9%N*V~`h&
zTKb&;V5|-hdZ`+068;-mMagggG95sC#eNp7wr{cWOvVjqkXYe_w##I#mQ^z#(6I9<
zkgL_hIRF?+22@@@n0gO=XNGtMmvwBsbbkdC^woEtf!T=Wh3wt7OjkINVl&L}JOYu3
zoY8kFmp-djp1p#OjVBg)j!jn+f{&fT0tS!;)W*1rBj)p1qH`_m_N}({=npObsRP*Y
zCj7+|bP)xwZ-kK|$DH$xzJ2TD^H){g^WP6ui_v5*7SSl*&uP=+q<xG~9mOOJ2+%|1
z4O9FRN1w{G@bny5)XN?_4LEB(&}DDTPFBgd_Jyu}wfOY=pC30$%>1~6=Ur^~2zo5Y
zENGhzfH&)!XS0ny!kJz||15McTTy?*nz=%ACVzQx%x!tTy=1E8T3T+E*hHmu-dYNZ
zuT@-{c=iT2XM0$kF&)Q&fTU3bAdsYFq-|b;Bnb&DGe!u}aOCwj2uAA1Ee@un-9%U)
zcpsLmvU?8KM*9dp@~kxD8vvhK)qt{u%;y1Ud(;0!oy^y)1^MjF$rQt2t`k^653+)d
zbhNZkTOw-#$kWiO82kP^QoV*xK*0shrAKEe<{>5P_WX+?diy~`I*RDwj~$jbZ!Tev
zZMPpn0pPk3+NzAknpI+fjfC3K;GMl!&Imu)rf>ol`HOZ|(@L8?PQp`P1<syqkU<Dl
zdy<7s61T4t+e>jG{uwdR252k-fGxVbFwu3q8Ic2Ro+ir7YQj2t=os^DHoy)pq6b(j
zVD9WyPyH8#IuEhaUYi(%({ZlATt`Jz6wMXL<~8-Iokz1Tr{zw0n?c7%imD63f|^v)
zKGzWFS~WvLB*>w#^BAoVK7Eq7a+Q~s!l2YjR73a{7hSJsqu6pB5poV}h*2{d+q`o+
zc@KdKc-(v*?lkPb`s-sj0VNPK+BKboW5wV+oRKj?6i}~s0m3s4YTLjO{;+eJrVGPH
z?l-sL@UGE?t4KteXcYw`a+d1jUlfVl#DK_+ph2J=4J0iI=9r<>@Ch=opN@jN2K~Xe
zT)w<?ZP-v*Be)Gh_O;SQ4d*$*?DuS?h;~w4+2`1Jvw8EQKJy5n81Wd>SRP;@bgox}
zk}R=2TP>S+5w<J!-C#1k#rq6(tB}t?t{JC&6S0J3bD~hfJpUcGlQDDaZYqx*=*vL8
ziPm#Hg-RH)W<kYx>^{15qtBk0bAJ_tqUYL;JBp^>cL&JhIRwvmOyO_y7e_3}-e<IL
zse$2md0Lkn!$7-u(%x^MPI~HlY!I`v#HU+qNQ(p>EE=qaA(Dz2C5y0GU$AOfPiVEW
z)=(toeN<nE<EHn{O^qVj8{G<OEl+=9keGLKca`WE?`T6SN)ap(^OHkbdVPdyz3X7?
zAL1m!iA&g<C_uQtY=B>82aaWpIN;Ew?)}Pb0EEsOS7P+!`*5Qj5MU-bP{r~91x1<g
zaC)Jt^_qm(q!NuS*kdX+<tl#HV1Y=%$NUH|{e~r6y0VCxd%>VpjLH=-7<w3vqQC>D
zZO^O<OzBy_6y8d$wG^CB;f`=e2S(KZQTXN73<9Vo56Ih(=SXVa+#pi3;LCThEaSkk
zlr_XCC^Y_FIhe3!`lw<(m{%X>wqCn-EdstdIxW2n$TN6Tzj%)w;aJX<07&E+wCG+p
zrvjfu43mxNXFMzi5F9BHlRAvda`KB5b+KzQz{wH@8wLMQmcn8eKHf!Tn#k&aw<9~Q
zf$GLj?EWZ()@AK}N7%@lcRdhda_Wz%7MoBAI8p@H(PGzZ2Bej5)E5ry`}&fcF!8Wn
zu#?dYA@lAj#jWJ)fY(5}yl2KNZ21FS-rHB^DMIIje^h_trRqpV-2k5}fHGlYv3u?M
zV?0S>ivQpon}CF3bzy>*Q4{5ZAO*U+(1kJ!*S1|RPE)VD4C8icu&tkDDI*fZ)ILYx
zD$D>)0K5|L6Q*NgVnPVvEGkiCh5(-I0%UmGfS}{(1as`7h_2isI-E{@d}eR!Srx&_
zta8Z}q(VkPJoV)P*fxCqrvfehH%4R-=l)-f_Z<1gNYP{aO;IRb6p`BvR8;bT=QT{B
zM{w!he&6-SnJD%KA|6+IbYNfW*@tq2R>5tCOXHz8JdsRPJ@q@jCzGijyfU8DES4Du
z<aD5{VH0qF`u_>6{7=NE{~K0=xW0cMJmZeXS*n{!gJ*~n4IeZCOnYRh9{ro&7-?19
zZK^~#=BnSKT?*9%WqXya3}I4EvP{u}0b(jKrpQiil^dg1H`MT^;|+OGqHe+=0KMe)
zKzyx;#sL%?(fxZ4a1}J0$w(y}W!I_31i<VBSck*@CVGDDEbIc78ouziMcynsIAds`
zBL2VaJ#kIW4HrCNDH36)Z?Cl}+$<!9BoQJU9unq_FME<xQk{CkF8{T0Zy~Z#VliG_
z`V4syZ8TA$e1~A54ITy2^gN;#v0p_rFq09ZyJBAA7S;t&maWz>d{q7eak+dwM;4@k
zORtg7iZ%3QB{tQ++60A!fh`Gp9J~Fcy%#pO;l*OxJlLturN+^h75u1lZZL?4smsuv
zjpk8DAyCeKW)o73FKWu-!43_e+Z_M>Maqk&1)*%abp;<nVtMweAnJ3HM7s>Z<tz9>
zFZ&<+JQ>ET-IiX#K&#)W4m$*dL3H#M0(Qmp4&?m$am(-wDtSUWc2>)kQZZi;s{Q3_
z3|BABXsh0-%nd-|<3<5J^j#HX_@5N?JRBgVt`Xq*JwS|=?0WXLL)$zV?at}2#J#T6
z@`w|IfA!xOjABo~sHKb_I-C4<A7XzO4||AU^ybwPs7V|7TO|fStI`GWivEq7(_kYh
zeldfWB%T*h@8iHgnxD7mx?x)3#B&L(f48usNE})zP~C|};KnB5098w*Rq(Y{f@_Wf
zP-Ni~YO$2s{V7S1IF2&5PzI}_Z3JES66a6~&+3bGhDN&2hoT+H3({@nOcF^fj?wC@
z)B%*q)RV%uj4?JYsnjb8J2L?X+GLk0{1Q4c&AdBVXw0x5@RVJ^W(ge5KmWnq$WPo3
zaoO$7=OoSF4LsW@@csb6C-+bJh(55ha$cG_|F}&Fjg9^9OdveM780ABj4TB~;?LWZ
zO!r<7zo|Ds#1i&r6M=>U?&=h<as#J<c<fQ0;dIn~APC+9nExu(Zf>@+)n$QV>R)C8
zC<TFdoKS`8(8W*#CX5+wJkt#53U1qQG`dueEMU%9o^G-tB#<k7_}_7+uj2WLWWo~H
zuGKPw7^X#H>6hR5{jKvt1RERQbj~pXkpT7spNk&0c(ZxV2x7=A#M|&uKj6ACI}}4j
zr%;|4#&*v^2@g?IZu(_O$qyj5y1$p(!hNJ{bv%n>;(A<O2)R6PJamjd*5F<punr-P
z1&H~lUR59GvDY(>#jUmCsb-Ayh1*mi#zGhA9O@rSv(;|C`ulp~Lk%7an#fVC*1QY0
z*EkUW`)2JXq55b1m`LTa7qu~<Hi)uGC4tD)NG$wReXxwJ7_v<N5$Yy(s2lZ%AaT_!
z98$zZXOvohN95oFgext(_c1{3;*l2QmEb241t$LxB4>i^Z9R)hVmB)_tMd3(#qkqw
z9*(XZmhqk!iEG8v-qA=6jef;<g&Ev7dwwG}Ej4WC%WMg72<!|o_9a+Vr}-7x-XC<@
z0jQqmh&{z&=bWO!kV0dBbA*)6*E6NIh_K$m!$2vCetr-~rU6ANpJro(7$`>>ApZ8c
z15uP$rPuwz6)MWp>tqYtxOul=T&X6Ws_Q_TDR<njn_DomqsFFZ^!-GY@mp@t=uq1=
zPN7hBR}dW!(R@}cU?a1!OsbaMkyD~fYP_X7X3)x&fv%rCF{Fyqh_MOf9`*(BADL3U
zo4whyutTWL!3cdnuqoz<To6GaE?Xwmxp<x8LnPr?!lT@UDS8k$Bnc{FpN~%*qi|TN
zkIxSy7neJtBVs>;MP*lbWI33WfSqL(s&AJBqX_8CSGe=uwbblhTm|2Vu6$X-jxC-r
z%r^((LzjAm@*83MkVkyOUH}hm48a!YsF$uvar+@}fc?&u=|ZSoE;5Wb6N#=ym87@B
zbgR7uP96{j(y@>54Zgs%CbrnXd88JiH-JuzKt=0ekl^6x1A_&7q!8DF0R^o{NT<iz
zm+=^US~ldJB0(CS<X2n|&c-W5>qy+;08Bv52cv8X_$ld@?zw+p1UyWduywo8qqCTq
zzxYC?i+Opb1SfaLRY+AC{Pue2y(^u?1qrrA?xdza5rfE^YPwcT7YU;ct1yq?mPkDn
z{X#Z6h36)HQw$coVu2EoY5@zMvc%$zthm~zDFM<<j;!OJQ%~9#dcAp4WiAo?LID(k
zam8N98bN3B>S>R2<{H8_>*EhQJ`_<;EDsPfzS+AqK<t#im<&_mv9DTGakPTVb6G4L
z`mix#&`0XLNUR04JOiXKS#liUr?;pQYz?YBgKMNBtjoDP#HodU+$yq|0M>ACQx1Gq
zf5ILp(iIu1mf7?BP5vyT<}mo2i4~$@R{%HB*~>XeOjL|v3n`bPj;`cl^9#NS^B3L}
zKX~|2nXmIjX4~9ni<^Oa<52QhV^{nE@sCb-Z8KUMIyKoMJGXe8gzvE&UTmtc^PZ1d
zUbrl%b{?>TIDnxwU4YCv>{39hq0M^=A6LRiBvwg2+qoXewH>@S+m<v`D23WxY@kEn
zl_T^<)xp~tb~|G%1&M~0XcQ))XG9?FY@%XJXF*Sm;FPT)Ol;65($X)j9FI7?109ce
z&O7G4%fdSiFspAkKL-{np7nco{I9tz!^J@@|4^QVWB%_CRSEBZIRcY+Q8ccnRM%C|
z(^4dWGE9oIp>m(z?f}oqYKD&io(`<kWr<E@cZRx>nd(dYolxTC2)8_GeBtpqAaZj+
zUZN^haEM_}uUqw9MYb@O)#dxD>R8J4=~SgOL`?>kAJ%$R_aZ+xN13eVvTY}K-xe}K
z0FexacJEel_`45AlHgWaFLZ*1AE|gBe}i=-!&1DoWWvOq?Xb;}$LyT{js(lFD_`{`
z>j;n@&A9l|+Oe{V^0_Ft_bcTQcjo|D-(#5bf>7*pltg|}6+k)C&5cL*c!tK$7Dtpg
zC1hw`^TRlxQ0B>lE6=JXq!ySF(Sf)?2y_(|rM<+<NacS$GV;~N#U_1}e*u*w7=33v
zGc%l>j^MwD(ZReDl#gdW3$?ShO<_ZLn~8Nll<KteZ6*&ztN#{V2fUfU*RKF~#tqkP
zr}99gAciM})cKSS#$e~+V8Nx}28jxnjL~qrgT^G@&IPEdhu@RKSFK@yLQE(6$CSV;
zmAP;}N!8<!r^+CCRxOuPG{zs!hDphc?{(xvlossn#E>Ke6a-o;5N&9J(P{*R2Pt4a
z6k1VL>cj~o_3_~eR5f6AMcKC4!oj!#T?A6GQ_u1Pe%%e-HE5|NKp2+ms&OKGOS{w1
z3%Z$`+y$UprvXCzSc1@VHvkvWvG=cl^x}4|gI5CZ6#eh8qOXW@e&v;<rVeRDE&w+O
zqaGcDXx>3SLC7fhZQQ-{1y4|JNcPV5No`4aq>Ll#Jp!AJ-$votHUr;8tqJzI2~h!%
z)o034Od?81cAfrkLb(nmTRVZx<?}koA^EK^I9gmUZxmy}nIrgvG~v?6ZcgXmP_gBg
z0ZAj*O>dsHZ&s1OKw#=gSVaMy*2mpvjC5Ge0LYZ<z3WzlB>-pF|I+{mUQ9}eB#<<Y
zEH*bvAUObt0bm=Fk_Ey!SImvrsU0QS1WMxeW)hcu8fk&I-YzXOdy6X*3TpTSx~pRF
zsdy~Epc+NIG%nc?ehC(!75f6d)EHeg8E$IDWJTd08E37){Y6BfB(mCfao!c-PKC%N
zlWnd~b3u&<TAHc;gc-56f5m1J8R6N+3udV&Cgb?Vm))t{cdK$=(1$Xq-pgC|l}T~Z
z1`snk{?+~}Q@l!cZ_3{%V_@e$<+=ztnRSu+Pgsm;eX2N;{swcT7MEDN&}Kh{E-s__
zIO96m>)kz30|y{d_%Q~w`ZjEY&UfTE<B|ar`MPp@h`OMC1^pK?>8z#*WKnn8wVT7d
zfNSCV(nQqqsoo>QV7wr}v1&<V0)k$8x{KVJ|K+EzG7)U1XJv-OWnE*<#dvcz^Ibw-
z@8<@I#nX=?%jMUWuC9g=RDBX9c1m9f9w>UgMVmu58Q=`^B=1e<dYv0^{L5tB<a#uW
z%VoVF{WybAvzU?F{B{>SjBj&xmaQYiX1L7@f_#>CB2Xy0*Wa^j#0Y1J#7jmD9x<OG
zR~AIM{rTrjBCWS?@kHt}%0hn>dhd4QS{`!sF17tWkt8}6yl?qMlJfK!2Y@-=VoyT$
zTiBK|eC$zTGT!2K$kWnRj!96$X1zSq$(V=T`yvO!#U7>UOlBST{Z0nFAdj*6`GK<J
zxNVh4->DsxVaIGu1rsLC;X$#%KfPr}eCNOH1^pDi#pOq3EG489g3+e(V6p8rNcTX@
z78Wbg;kEVXPc$Qw2J{Ufr&<wC8aQ-#$OKFir{>F{o~LXENlA~1Js?0#0vylD8XA<o
zziXTJodtl_Xts0oIvd=G1cs?O3t()hV*v<BsilJwFgp|2kmU!6rWnuw=4LWs0uH<z
zzg$&OF^}M5c`7r%SU;Dfl-#wZ<-}%y?v-iQJcY2ATt@~YAxaoiHG_eJ@cV5;B%mip
z#H-*UW+U__U}2rE#REkKQnLtZA7jxVWoX#M2FD%b)^i>?!MSw$s_8i!FDaE%K3{Lu
zM~RGv>z88}iQ>}>x6JoRENkm|!kwO~_65h=2}Y90VMnTLv_-NfBadx}Kvj3i{}2%h
z>i$KFPJ;K-;ipb9!G0WDdoW%nxRT){mm(e<U4+P8L1D+zh01-bjxz#fQoKlby<kf1
zVFFRb$jQ3AUtSWkMu?sC@@xVIE0Fv9YiA@QB4XATj_7>=oGr(E-RV?70^Dzi+iTj<
zpPd-C^#kD_zl&a{yJXMocQDL<fmyfXc-{~9joZ@>x{SQ3-%-^H<5&V{PtJ+P1aV+H
z-PI%nC4dRa#m*!3PXnghPYQ6z<AN_FZlGZ=LX-(jN;pGi5$o>MoKYPFrQ%Z!T_~!#
z_5^ErKR-@G<&lTrvwLSWV$4cRRwHkWID8m!BDMH$6^N@{%%KZwJ%B0gILbQ)z9CFR
z>S2kn8ULROU>_+uIdu6*Su!^Ed}>?EYXia*R5V|gzJJvDtc*77vie15<8Y+t+dwBZ
ze7hNL`WPN!|L&k|9hMFlBmndF;I*D@t;Pn^t?}EZxdR&Ve{*v={e88w`7c_Ixof(=
zDTDSUoAz#%yE?wtCb9ot8LdJ$S*A{44<?BSI!ImK{qobk?4X(DYoHh-{ox&L1oBQ`
zfuFyS0??%Wcybog;B;rs<LQ$zSt8Q%9b1(5TEBHcROlrT!e5<m6xe5}Y(0V#9xEBV
zDhfxxk$I;G>v9LR+daK?3jlT5m=aE9M4ct&mqXuGVU5+h>!S&X5Syke`xUjti}V&*
z9e?+4JV%9$nP>FH?*Kiox5(NvzUg)@jV>Z6u8}|tnv&rf37ga~7Ffb)ulIZ5*~zi@
z6G{&^e!w;-i%f`Zm=TUcoQV?wzqs-@LfHbZao}0hQV}FDWL~y%qph)D^4>Aq88N<!
z=cHp*6!sv(4abc79}0lv=0q1D6QtLhEuYbOlP$I1$3(CQtFhm&LbJ}RjT@cA*yidQ
zpZ~GVV;)uRQ#V@)?M?=GE(eIwN{^qYjb@S_q!Ut%CS(8>KY1}#oOXXNQ4E;l8yCLi
zOc7s>N{h<O55QNLGS&xEsh7;N>?4EawGJ=Mo$chVPS_#qi#q5mLi}e%i7eSh?vqv(
z0&_#xVzLei%f*9bD%5&I&0ovw`R1TF_<9kiq^5gWjW$*zvZ?n)QS&T~ZQ7QR?cDF?
z_E{4~0!cXwBjoaSEt&}E{2n_KgH-O5_;Q#vz=$zIRD?P-4iYI#SZ?Uhd3G(J>7QJH
zH>Qx(n~MunLYNmCiLRsu@vT%I<ePqGpZQi!bEhSwOkHoK^Q_}s3cR=4SJO5jfmuQ%
zi;-wpd|MPW;LEMAN3qF=&^A7RgUCYOvtI&K+aG{vxJH{#?k;MDy5@`hHm8=^X{SB-
zGWr>75lPb)1*vKr`9~Uamm3L<iRnMWv#qBQp+%eWOexR?X=8WMh|w?l@lKum&4Tw%
zD`Jcd?I6Rtw%3^41v7&b2|P0Q5e~Dm_~dt(V2JILp>UoI)}`c$C-0X|wwu8dnf)p+
z0FNpS^-lv>0X%#k(MudvRTUFSi*N=W)Z-HqPxn)M78Z+hs>j$oyO>}LX}+9=Lag0I
z@WoNOa%15`2^5NfH^zC66^rk$hbO{r>STY(qxrlc+vc)+P~w+u(hS?0vnu#=ijJO1
zxRH_hnablNDs2xUbOKHs*l1HEPRC5h|G-5WNg?Ys`l+Y!bI45^b@&x4Qt+DV=vj^U
zTd5vDf11<$Cvcct*e3!GE&M`3ArcS?i1tRIf1~A7uKUwvz<>1M_WY&~J8<Eo*YY_z
zw$Y;S?bNawQZR|w-*#pve{1?v3)!gG!fJf&mh9b{JkwYg7WOZh<6Ty2^{c6V`zO71
zZ|&9eFXGhnD!J9KalQQgI?10eU%f*7R@?uClE48-5wm`OmG?tp4{fgBNU0Hq@{X!&
zZ0l+7z-@>lEkFZ^6Hrx!h+~r|niDlo_Z+jL$ngi7k;;6>MA~-8|Hx;dk@dc>{We7`
zfHuc7aN)}KimIx-$vWMqx&%Q0QpzUr5mD?#dK#Asr1z(<8=<@Rx07lyX;!*!<YmaG
z`m)ZsB{OWtPW>BaTc-Mbc!3SmO#5CcUcAF?-Y31}`;P@Az@AL*Ghm`_3th}c&WL0@
zYy5Wd>CnzvAv7k~q&Lgra<P8TF>C6QWWG7WQ;pA9&KULwWjznv@!#L1Dl+3IG~$`0
z&+z;bJLg50d!JAi;Eog{juD~c(}HgS{JqDXg}^MvuNbu)Ij%Yc#EBphqz=!KFN<4p
zzl{CIW>eAn)mQswM*Uy{x0eL?WcxQZ(mH{rb{xob(56ToD}fq;Uo8%DtV=s~#vu**
z^x53U#pa&#{QE4SZ5usK!XSB@fjWNmw^&*b^&@Mr-0w9r*(>R_kV?ArQU>j;4cR#l
z9K0)4xxND^k?}0uK9)gcQGD){M^$+lLrvy)-oopj0o$TuApNHms$!jO(T8%AEqk|?
zKf<ngxT#EP&mqvxCCT{|RPo#YG*~#8-DytBanvZ6ps^OZOu)9C;a5pT5AjbsYIb<V
z=lE*KFcJl4j;tvA*&g^t`lGzoR2lGknV^lKQi`?|5?K=QxGL@veKXe7aTim6FLjdk
zWF6-I&3k3IH3+F>;7rQtaU=-7{mg|OKabSw1$Mu+s@CBUPGPVYAcm#$Qh0oKQ<&Uz
zkBGTPY*v1BzlbzIiSboxsabClJNfx{WM=#n459VZc%Nk`)7d8UVsHRhQui<9IEGF1
z_p5-DTR%_+1{8ZaKSXmaqhpE-yskaGu8is>cIvPw5uBE@tkhogY}D#@gI30`w;#&x
zzau@&XTcI$6zVW~I4ot9M+w%E%n15?q#nn>7N-{o>oJ@Sswgl!LiMa=TT-5?T?9nz
zFQ0>W^A{=>hBlv%pj8v7mJw(iv3<0x5=jYSQB5RlnK=(xajR;{_|wBpg(y`NN1W$<
z0YisIJE0Y%S(g7Qtv!1wK^EnocSVY|16t!5HMVV8PuN7Nt|sqDX_oW-R~>Y`wMcw|
zJC$B`0F+WZ1v;8?r|<6RzvA1MrZVY{oi86j4szOjR*Q4=vryMj^d|6rXr)u*AElDZ
zpQ}wnW#j=d#fXz~fBqtT8W2ZR&5olCTH4j+E6V>?3cU)otg!9^0I68H4h0g31b&n=
zz_aK<03|ml+k#E#2HtM}2*6{v)$qan)-3TwY|cA*&#BDybDsjW5S_=w?6fbA+I#R}
zdvjY!1Zj6~(MI&e1R&CVn0r=VJyLlDLa_WUDk>^lT{FjjH^G=1(Ho5|Xj~%Lt7hHu
zf{@a<o*`_csO=za^Kp=RvKFHpd&3$&friN*R?qzeyqqI;?}|vT27=S=nXRZSpm^&7
zKC*_wmF1^6%VjRn2&T`sOtFkvq2VVZB^hz~d`5c^0Q^oR9(t6ZJ+_?NrUVDn@Cg_a
zTjMT!9mQg=#Via=*rlzNQ=S8xx!(^U>XQXq15#b+PEJ8oCeNSE`c|2&)&ZOKYoCaf
zG6V#L{WBS*EtBdz1E3<piD9=Hdc!v4aiADNjF^X?J&=e7(G#arnGGOGgs|PJi|iYi
zK(GB?$J^;EX@h}S2#gCm{Tlx69iBiKiJf%11Uz6Z;jK=n1L|4IA_%nCEg<b&%@mTr
z1`$Q%+ozL1wZq91ZqQ3J`s_U{O#oOJ=0(w0{5(tB)}LmWi$EkSe8@hSk6u6L?x9a$
ze2^X)f%xw<aJbO2RuP`mXy=+nAS)FN-N2vTVzmrW?%A+?=yIeZ`)Jt|Jeq*@o3ks)
zb>0x*BJx`@T}X?VoEM4Tb}3jzsosAphN{66of!AcT)?RE){~;$;)zz0oz1Ue-Kc%;
zJ-@ZX3K=Hd>?K_feSl~F$=^skf?5Y_cE{<R`n1{QMlkV&i*r5dHM3)iP48AJ)|EYJ
z^v^v>Vmn078nYw&o{~6VdD?ARs|9^K^X>af@&RoXx0P@^)!xCa#bj9#6l%M2*6H?$
z+#;J^u5Gz*ty<Pvsj*ngM<ef8L!)beJ=<L^656NMjOqcQ)6F=bK?B;U#sLC4H5ZlF
zO1IY}Ef7)qh1StneX!j!=n-bT1xkQS4n&Y10v}_+NIeE8(^uSPX~=N-nDxQaYj-2^
z7>-B)nV}UNLqIgwK&B&AP;6Pyweo=m5p)sx1SkjM2%}YJ;@Nhi+Ve=*>LN@Uz^5Zf
zV+h_1iv%-*TI2H4bUsvtNKTm*%2br=To0>C^CU_f<ve&U0)Zo8BegBeyqH67cy0J?
zhL`o}O^>W887xl>4;Lc?B6&Ad3EENNGcahwGY|x@jSTsfOeRvPQ*J3@y^o9n!hBXT
z*8?aL?HRDdVcWTLgp?rTda!-?TIo|O0e2sU|1b3e1L`wL^jZe-`F;%R_XwFYOsAaX
zT<!ZFVk+y`29To?`mc`4o$lMoytA@1aW$x%g{CSWm=h~NNyvN1u3q2%^gBYG`f#hJ
zUWRbyygU(-0h0?uZ_@Ql7z)vVD;PI2j~nzJ34HRT8GrckIPwhxRDG{UktbgoQNNR4
zZ&x$l+)^V3rYQeh7gz%kO^!&1i?A)lwxtynLtxl+(;>Lji<3&WbGR|*yf|53!q|b{
zAI_I|2TAdEZ>052Q!o#`TRf5hIg<3!-Cj#Y<p2&Biq+5cvIXBQVyS)B*mQX8WZ$(<
zmL!_}Zs6tS6}^ZpyqAS^smm@8(Bwr?X`azLN=r-B83agZST+B9fTr07?8OZ6e4s-O
zktlw1eH{^So0SjS{~qv%k)hD&J0(J)>6qX&2&?~qibMtWTBk4Rw+g;c0L5C`oyMk=
z3bV&Ic^0l1DMRxll9ZY_WN%j+L~m|)QCp=I(ngbeF=F1jtDFE#eps_rd|MKtRODKh
zt@0~D_~L(=NeoTr--;Ffna7N`bXaqsPwix}$(D}6pp~dIL@em<?tPBI#~fz>{RXy(
zkV5jq+h;F<H6iupMM(q_!Wb1!9-bjko#znsLOcfBTo}FV2&{RXepMl`%+3n+F=%~;
zJ0d9}HuY=dhkpu3u7CtRmd_zE-!rg7D5a{p+OMS`1?6KPuEb!X8uBqh>-opD3kY)n
zf{{)Eg2W<SCT0}PGbpD{;>B2j1>c;TX2>9b(8N|pMOl8_fQ^o+tBuHwFsK~~$5Cjs
zeWbZo(|3~Qa&6D@ckdRjh3Gq5oKOIToJkf;2oT@rIN!fIi%A(4I2%*;_k5YE(D5KA
z^ciLh6~lpsdS!BJ$mL>N6w$bO&qss5x?7Za-)?MQ{gzc~ks3Q`X@eX$1SlvafrT~!
z;C81Fkr(zA8!uJeOt7hvn-b5XSMjjKZ}BWdeM&-3)aofJ12EJAh1Y^b;(a4-z9B{h
zrN&JMgQS~=tbm*x1A)P3;o|~YBIjXk0R54oH0JpLcGAaX)d;BqGa|tQjUU6pzj~ZS
zVNzMg*jC~pHfq^Gu^_p4C8qDFeSp{$(^S_*Oc+uej$iptNc}|8E(Z$%5n@)R4n!9|
zk%8NVQgETbhHc#?FVcKo6Oey!3PsyrE7St{o|X<2n>Y{NC;S>5Ou3(z)aH{s3Omq8
zCT5)~x|>elZ=+{n#?XrvpsZd(TpDtWTRc&Jt6MX6@B?Kopkx>Vq1o2phzQpTzD!zL
z+9JjmEH4KkGJx*#l1MFuXlz=ynL=jV^F`^3yFpHX7I}V&)*>?uH2nPz!Rdza0Ya<b
zvK3~C1W9-Judl@-h?Z<@Y$~QsnkXfaUSZ=D6gZz<|FfU~bKv8F^GOCUYjV<>f^X{1
z)5YR0F|O%}TG1%NXq8hGlUbJQIR3eE55<Bs|FLpICH^%qbPHH|I(y9oR6MoxZEK0)
z_+i-qU<H-KotafvQ2W4#IOC>RX(|<kT8X^4vD341bP$6&lY>XB(Vv3Bo2J;x8@Akr
z`*cA_9+3aUVOHXe3g=;f2m~4={UHlFC}qG<5V{?bhd^TolN~voWG?2TrqEr0bXw-~
zLK9I#AnuT$@eRTDMi{a_0c0h6^xcqL@Q)BRay=Md2gt}+@FNtn(B?HKfovJjUp-PH
zR>RoJZ)Z>mC9WC!oM2yJRY1@Q-iFo~$m@k?T}EdMYJZN%g{2J*Xj8~XqH+M<ht@?3
zF6Q)m<HLpbP({W7r&+e8U!yRR$ix@w^OxvyKWS0`16n{tdO8z*7LX=jDcvO~4G#zU
zhcMn1PzBp)81})3wL8&@%sRDE_W5rn#}vnZMgs!>j0Pynq&5N1RCwtrD-#N&^*#1I
zDqEjm3BWo^M-qdabo6brIq7CZ!(Bb1(e2X6<_po)90g?64Jf`x#dCn+iI6*iJ<0`s
z*_3G!Wy}f?><L;a;nfVVkfcEztw>-K@i#Mm!}Ohr>ZKtR>0!-AfYlR&fPZiauaQ|H
znIJKy7v-?}ENaiBY79w%DqIU}INrh>2{QhUxOY5U!MkUdlrf)@(XwC_hWgIo(m-*Y
zLq4{1G!cOTq5k3@c<%y#X(o~L1-H(R52$e7KzRlwbt3HHA3{O|l+lqCxg{IKlBY3S
zVJK&tE%zGrLPQJwq!AQC9;j4J16@snbpzS}V~UYT1EkgipoQcn9;-q%5y>N~$;R|f
zDA$m{ndA%P=B_soAPqZa7>^URuT#K}Nw&gr?GW8IXSWm;&flIA3o(_q$(<mWoCc^5
z&9G&VL2jbhP5>DcXWjzsg|Ghaw0Bbp{g1VGC-J4PQp=<M_uIS0z$E)KNdoIP0-&b`
zM~|en%lhIT<f+0LoX!lt@m1Dpj7G;5xS$zy(=ibx91oIYxqP}Ifq}0;T0jjGD&VF7
zfkZm)(#%bb=q6iS84w@WECWV!fz+jBcSrsMppjt$Jab_n?}T4moK>Fb&wwap$=*cv
zD4pTfcK)2<mp3Ri7+bL!PanUB2Nf@xDEF*sv=GVRPRA}>pOQRl*a5>g<%#Si(cI5g
zTB0|dHKTUnQa4*^iSA9&6z@&>i}e~<|EDie03A0%l^VU*O^**6o%!vDr|kf$+E+RS
zHVyIV4(z6Od{}=7vgQ)@MhY%g$I`8}69s4mwm~a!BI&CwOa6Cyo|XTdp6BvJT<L$^
z^UR+3mNJiar91dUF?uIBJ^q`X=f78ao(qZJCn?l}U+QP@>F5o+fjp^(4_(inKp~4v
z{nM#z_pdHb1(z595VioG_*CX|jEzFFp@8=8FVIff?}t*X5c1et7Soz*)vH~FQOZL@
zY^fjGw*2%dDWh3rn7*buG0mu^IE_k`t)?qq)L12#fC?Px&N{1ZSTpbg`4XvW9$u#=
zt5NMb12apQ=L5cfB|c-N9?%e{c1glI#&qANif5ucVKQDL%^WX!SFZZ_nrI!XA3PHC
zTUD$6g%i5_zwLbH_mw1_&$mno$NE1zX;|%0>narkVx7J+{zaj33|rT#nFMd0k$+Jn
z^9@MWK4sDxSe#6AQ?y5urzwm%;VPT~Wt<}VB5C==*GiB0t!2O5dEaY;gUrHAdvAH)
zi(SwEy>o|FP3*Zhqa$4gSKf@C<uOgEtBJ7{w|!k_YE|c@qg5=fBVFu4FC=_c+ugg8
zVs_ELFO)-hRI&8^n#K2qOWTyI5bv!i6>4{02NLd$QnW9ue0?RuwpYGnJjtwU^0k+7
zO6xN-l+$uxDc_HxFQ?RbU;rjlxt<pmLlJq#M6v`0MS&egMRTX+P~2(Y3NC*?x*C7G
zX|ne-J!%-A0|>(iIcmjI(pXk-g){-Ie8sIyLMGon+(Jf;ldluU{>zBffrsebSo!*Y
zzxe-_=g&@{_&;M)7$vB4PrReXHO-`<Jf@#zDOvvtzy-v1zVdRRP@HkG`1j!%;P@Eg
zrjzW9Upx>fmWGyBw0qY85=L;a#{f$6_(Dk&5<Y3#ub#2Zk%Lijo~R4lGQ%k&qj%zU
zHg4L4*#)oEuaS3QUfBuFodW}nv8Ol6zd`L*LjDN(JD<Je{GPYr4L$`u?4vy7TsQ0?
zGr1{_CZPDVmWuX3egq-nw6Di&yguZR-}-E2V?%^F!yZ6E%by&$bjg(yc-?<iO%tN7
zkb=D8Uxhy?a9=SM4YBC%hD8he_Lry#UT$Dt`G~iPCvN|RK=H_|Wwu@$^x8H~5o;21
z@;LlK!I`i_1uu~PV4%p3VYF^j@OrF#`U63Zgippq#o@{6KfhSkGmd~E%rZL6hRg*%
zukq(%La!FKRJL#Y3%^;nVl@ivNXUo*c%PVxSKdJP%*q?sp=EAz;9E(els7HcSMUOP
zCdqa5?q6A)Z}%C8>mYZC+O1B~A-w*N^?X7m<6$6P;m{#%G`$pj`m++Ye<w6RFSHSJ
zc~b0PEh#Rsth~+8U#nL`lYUcDBgSE7>zR)%;b#R5e=b!Ys5l_jD>90t=bWP9_{#ds
z-rDE&7D)K~{Ji{^-T3uKcgPByIQ}Ka-U7mti?*Ah>R|Y9iJc<%v6uPh@88#JVm+3Q
zDWHv>qIBoV;&)TeeKSLRKQ%cS{Cws5ca!~aYas^i9xZ3zCwcF5;myZ4{`pWTf0omT
z^abXEA|murq+%N-dsbm(1^2dpIkKJ~^S@7eP<@>uJBWSd?jBlT&S@F+WzJQRg}vo+
zDzz^g9b;=KDdp=(M^JaNLYN@%);ivd%uG-b<`4ppC(Gj}UsTaR58c&47oDVL40&N2
z=D@4?`pJI<Jx!6}^nbdz@yh=h=lDO911g=sWM*vPj90I&K`;oJZ)aa$UssnJ07We=
zt;?7DFk}qO6fzJFa07-Xh=Ym=0CD2oEAkA<Mtf)x48#^WJQ*IGq#A#>8VX`iyJo@J
zHv1BNJDG^@BKad6rIRP!UyHj+3kwU+%+8{!kL;=r$rzB0?R6-aE~0;W)AgUI?P08a
ztTF~va?w(Gq!(?M`zIIRq(96Bt_g!h=;_HsAB;LCh4Ne{jTn_jH`#Daj`Az}Y~mf!
z-Czlih;Tz8_4{{qasT`L7><Wr_wx`>sWP`l@?lsD&=4Rf)-yK_?rzo35S)i=A4L<@
zx$zBQ*#dge9!rMN!DJvsdhrjNR_?eOB85Fjjh^*bsU!0Cuixo!LCdL9kXWzz*d@FP
zC^u$_y|ACf1w5;X@-vcvxp^3ciAqqpO+|s%v2|O+8aKCPP+2h-0~KZdw0W3?^PJ>_
z?rl%he$`9HA<^-K)HB}RjQOsQkivq2)orbL1}0~A_1TR${>)T6`7!)LNxUozB^qFN
z3`n?<_WY4&0|j&X=T9`FNCK`0b}>9U%5UPiFCI&UaerHO9oU;DN<Nc9+(uhl0_LT`
z7r5hqm##0A{Xvfv_uR>;avpyUJX7W6&G3VV>-|XW8#@7D2>Oh&A+lH_^pRz#1cYi0
z4N3k`X9ovNcLjuf6WRPVJnP7)Rq?7Apkf>miX|)%9`3`6jOW7SUA}Z_kV*&AJ7z!h
z$gCbTIJkh{_P095m_hu!6dMDi5hO{0K(C)AB;?~`vEazh$*>)avT_Wal)&RV2-pNo
z^Ac`LVi;%1&fbd>11gxw=Q3i~R$t`B2F3;fMEnAyKOWzqfee<woWZ%H%n9)j00nR-
zBQO7LGVozYNJyr@P8c?-mFGJ)u$uvTqg=4T6_Kg1X22$^%gevga3$$MFC51p+noJN
zk{U-%q333L_Q0}#D1r{rA`k}O?!lE;5JZRX|0o4G=Hbwhwj6t9hx4aS-MuYvHnO`*
z9o$SZ%uHXP9;a&a7*^$0@JeJm`y@9vH+8}1WK5z;jQiT^o*zF5=#Gg>;s*o1;2=$o
z3C3_R*8};76r!GT28JotgaFmze~w9yW)^7{&HQd|ZZ)qUu-sO#lbd?NDtRd(A;E_U
zC<T-;%^aBVVb3Mqj{71bq`3703|JnG<e+{v(*FyFY`hX{Xa(>M%r;Y4EaQy<$JP*o
zB4OP>Mb=bTkD;FhB$~=My}<dft9W@0-<Srcr{m{neX9TVxId2)in6mNmW@D;z&hSp
zlhAGL2xw)+&D=V>;<9w>p_-(x4A7{|onsCVG1{BYH)TLiz5$*u=A_5{V|cb_A!N8t
zb4P^7J47lFCZwvE96rn)T7e3}e%^MlST0_Sy&X|z1Ka{4;IyvEKsC18<9rlw60dvf
zMGE4lE7__^n>KEI^<5h@%i|Y|*4yaJpn3)cVxy#`^%~#}0`ArEz%|1^Z`CA<p1_k%
zH(zaj`vS8;()Dg6Mtt^u@F3T+!@XY(iirQ_6DFYg&gX_r8UM+qvAMY#LvHF{?c28x
zS3-L7@J_C-uK29L$Vh%28{86mrH-PK5<uE<Qc#LMaT{ic7vsORJ_B>1+K8RS$9u!p
zw)PFxlm0K+*t;kM>}fH15}i^=^cGd)?iEGgC>?N*eAClVjqe)V)6);soa;fGVaT!5
zpu5_hK81a~wa(&eoLB1Nnp2*o15MSEN$Rk2!0_c$mH@SJ>o@`)id_U2dOw};9XyWc
z>+56etw%7K()FDuBy(OQ5zwq$eF`|+8c+FQzzSZjd|<Z)Ml{CuV0<$^54qmy3Ls@;
z*c0n3&<V2$6k$(o5I=+=v#q-R){II0OjRDj$D_l4toAil<zocPslE(uT5kXq^xyS$
zS>9Pjdi(hJ1c{ToS~-2^hi>pSLze%7e8kBcOo<BY-wxT{K)pyVUiF*-o)_Y~-41O<
zMMdx(8!A9C;&H0QtjE|9bX_gtZFM}Xv){u8>3GBjYG5a0vh6h$KQk{xI|lZ|4AK7C
z9W<97t3z3q&(<3y6Yr3yW43bj`ygKK=k2{c1jEIKbbc!JqsW39+__YB?!bV6yFKsz
zc7^sNnr}{@OAU7A0geMS(W7>}UTKRd3~v1-6qLRH&Anja5w8xZ&lO6?4q#*h@8do>
zd3oINH=7>_I!{%QA&Wnf_Or0u5DyLsVdwpRh(+OWf+k`pFnyDjoAce=oSjF0jsbl_
zJoh;_?X9$O+XN1m>eQNl+3GpALm^kPaf_I0H9G>*kos$}`h%M9xceV#aK~MCJ6B_6
z;{-j-9@yTG<KUwuGy5oM>VpM@5ld(LUTy;lO~dEs=a>1T$sHDAhct;e8{Pqz8+4L8
z{K1%4f8v5oE6oY5wfdW>OpJ~DoyzYrg!OODE+{xrn|RF7kZ@72h(C#l;Cpu~VuUOW
z7&WJ(-r?N**|Pt5xE~7(G5O$B)F7$^9#nLd*CoWo^>4)K;;eYg2bzAh9mLd%LpjEe
zftaorhy#=jny~-k*a?w0=t4UqEiGRoB0mpN57@iYQO8_n(*ModBa}FK^5m(gfQlY8
z!P#Rp`UEXBqiOjtvLPu|3MO!*t7|`Rby*oG7G7bC2Vu4>-deQt^P(L;8e@$u(=@d6
zTFdrfQU)B-N)Uodlv*bV=Xb4SD5aDk{Q8fBgW2ZP0e2uB+pJI~A}oBLQB?C$xya3h
ztSgJKB15L8XpV`^;lZk(zHv?c>b>52wBr-;ddNgv4JIB9<KqH;R|KFWo_VdF9|F}i
zmqEJzbcv)#f6|TF0&EhPhG#>Kh>e(ViK_8(PCAx$c1&nf0m&3U>m40UoqV3U+ZQJ1
z(SZx@=?etP!wQ|EE^96efal9~oOq^r0pL3>)*~1cixi%Ut55Xw^hl=x9x+CCdTU~k
zB4!9bE6PPwaNx0OCRV%Gmbj*+!rIzNxE`mIJ3UTjAgU`UC@2r+$;r+Z!6*PHCn(-r
zLmd+ZRbLD_uMa<H^tUFNf{Ssa?{q-DoInv`7^B^XhxNc0!edW(_RPBuqZb9#p50HM
z0PSe|3+QVd(64uf^6CXwsXC{PR8(IlxRW3=s*VB34fc{|sHen!UWw7pwB@<-G|AvD
zH0a=3pc+g>>ydjPW6pP(Rb4*q#jcEV9#>ByHpz{DVpLEh=TUQ=J0$q`*4{2-L2X!u
z&lZkoZiy8S6=d!k9YtUnc}>fYLFD3(3jQXVrwb@V#PSW5mxNuyV_s*l4KC%)Ky?X`
zccJK}2fJhp^%UIo^O_kDBPv^3e&BEc2Fp};QwVwVC<}!-y_BAB5ezqG@j!^?T752-
z^T7I*D~f_B@!B<O5U-T%d4LY&Pu;Vt02K_P8OPK#*#rkb&&CX+?opslI2@7j7Smp^
zbkk@#b9><dbKG{Bn6VSbZD8|JP~kh5zE@nl^==`Q)G@vHR21gt47OxXpfLY1FwjW-
z34RT}o&-=E7Bl`vQR>v0GlNO#2yQS2vuw^t93Elg`t>s?S_H|N{D17dbx@FP*Da2Z
zEh;7|2v{JJ3IbB1D4mLQD@cQMgNX?$DIg`C(kUPqlyo;JC5?2Sb>sVf=X^6~=FIP(
z-^`iwGT*!(4<OueUHjU5?X}j{m`CyVS$5-3tl{*!3>pI7Ijn9@(9wPW9FAeDh0Ei-
zW@Z!c%T~7r_Y#TN?}-V*g7{jh1tx?-Z#Ry6!5JNvpLIF%W{$(GA^+hS0i)%cH*aE#
z;#?tka5%jCtcgth)#>UHtb3Ao^jK8j2jQ7d&(E7-+ymiTgt`JPI(Yu_aC0L97a%o9
zyu}K`GQecwQtaYA#@u{-pKyY9brIGvrZBs}(!!zHi_F)r0`dF6#<)RO$d7RQm}I9q
zOCdFf))BIk_S)JT3k`r*i1WnR?-l6C;j;9Hb_uDdsNj0>fhowKnc&e9#zTvLbe{;m
zZFWWkKJOEpPu=p@76bo9zDyMO#7E{sqL?Q>PEc)B-uUeQ2(A9-2eEUBkN-XL{D1%B
z5i$Kkl&~07M?MN@Cpi;|OiPQBc8(egoa9qyR&oEFEId3|=gLZ^6UGxg11FqaNhl+3
z(WiN)zT){h<J5tw2pwbnKfT!h4yb;EQ4%G0)YnggELk^?*wB)VDi1{MYf2i)hF7Z}
z1mmm8SAV~*4`FpQZ8-TKu$>4r_#j#gNA;j@w^^US;b7kxG+%{hJGRygrQCF`9ErsR
zyGgA|P1Peu{Dy{xK8+tbLG|$IC!tA_K9MbC6A{+@%|gz{B4?|=d^`PdS@-x$&2M>A
zsqd}?H&c;l#UNrMX*2S*v$uDr=I8miw7GpAWE}L{pSZbU3O7F|=N!%~<gQo{$jwKH
zv%rH2FVvLOMCA+0T1aysj1F6#pD6kEjrPo$OT4_mIy#!0o4dOUkqzRS{Sw%LB4FB|
zrGkR*a3OrzQ89F->b}&WO4bE((*X)Kt^&$j3f#o|_iBK0ASec|kpUzGNXz`p%#V@F
z|12IY+Wbm)ip=5E_WFH%BolmrRVBoHY?KVM+4xf+=asgvk=PT?V2HW(OsK<e5|t7o
zZ1@S8Xt9N&VslIkJsDY$EhSR^<<VC5#9~;=Lz7F<t=pFbND*}^=UfbQsDK}?$UFfP
z5u7~OrL5vH1oY#u!v|tI{uceMm>NiPke^8C>FHgQHgxAq?8SY6iqZ1PKJ8{~15-85
zEoo^NBw4_(Av;z^%8c437oi`tTsJOia`Gnp3Y5;ha$#X%M4G{m=`e%^3ram%zsHXs
z6Gd`&clV@Rp(E<t`9?*}RS@4*`eEgT?CcvrRb+<aD8WHGgnAHaIYt?toqe`$#B8la
zQJ0;V$8rl(B9qN(3Wll4CGJb}@|m=|oA?^;Nj@|X-<I3$ijKQDHztpL%zMP=<?qk!
zIA>hx32_~A(vB7z95WehMR{joX^(0K*-_mt*Nlj$sH%4teuYDMi~-wToQc|yV&Xw{
zg7~5!%6y2Lx(i7m(pHMD1^n-}-A7MQ$P5}{)Vc~S*22ZGAlqv;HghOlVBmH}^7~Gs
zZQHjKgO|!3Ehavno#4H<GkM3$Ex3h=oiAN$4syrfh1VdpeItP0{{S{X04<U~8%s;}
zu-KlS9<Ya$Vl<Pvv9_>Y)<-@T?oJ@4Gysn&DJhZYh~jX9_!ab>AEl+qH<w|q=)5o)
zW>7*Xp`bGepe5=S4$4_+%p4eNN^D1#)JTglYJ?8l&d$!**cii^kT((97;q$V^77&t
zux_BPA0#A-;VhdMeIg%jKlRi$_?y7c^WzOWs7Z#<Lnz&jWdE@QD|SgkdK2cR=>2%7
zZ4<2(4oJ>?X~rOYZw?+Ra*Q*k^*=!sy9|E#NJb$j5hz6c>mN?q+F<JsTzn@3Q8FK<
zH_a-s_*DmTv7S0b5Z=MPLse@9jVTs5$~=OXiW~`x4f?jezAam~rdVv%x<tB-c#wSm
z^Fac`e?#`z&CO8i5(KRT#wlfb4@^BsB`@URF(NmD{^YPPW=}5Q5dxAVKEu5~83`Um
z8INqlGsD0f|8fbO(kJ~uK~T^y;IU&iD((_+j6D`7(jT~Z)x^w&@Y&y9^Lq>fgY4W~
zLtp&m0<&w<c=*v8FI8ClFv$YBTdaqN#|IyNK0ah#xVLbYW0(!GQ-W^O#Mz3C8#w>(
z8@Mu*MisKiG*i2DB;?d=scg%Rz^>{|_sdbFWQe1HF_3;F9%%t*P!W&g7oztop`>`^
zaGLqx4@^zVzETETF~pWg2iWm^0}hHA4bYj$#}h1i$ZZ&;0?FqgA&8EQG&sDpyiC}M
z06H2zfT@DEw)5cV%RmE%=MM?R++0q#UwK*CBIrk$CKkzJZezo#Z7KnH2vITFq}|o;
zU{zTeWILV9Y3JoEEK&z<UIGbe48>QH2&%_X?CCf$89)XB7Yz=C_mNyv;>57EWZg`1
z%pNS=LH(o2Ab)}L2%CR`k$cih9;P~mA3v%R3a;1fxwmOLKX5S{-cNu4pFVlQZ~y1a
z*|Q}T74hhB6N)I5G{(1h#UuZ~rw;gB4$^3>U@9uA3LnO1lz4z|G4-TgJjhQNxud8k
z<`Id&0r%~jH*W$0RDh17jB7;Ag6<gpk6AYKes7}x;zI%FB`X{OAiWpZjH#%py%!9l
z*Z7p3m1PT16TTCxV2e;#8<5*KdyDt?etAmxgaA544snK#4%snAE5$(@4}jjRG0x|P
zI-8u6(+Ud5^j(#&ZxLg2pbtaE7|`m`Mzj)v8H1qml#GUsjt=X(sj{-NvojyK10n3j
z>3|BxM0y!`GCm>v1Cam&SW*9Q{W&TQj8Soh$iO1!s%tq$=D8ve;}Kx;My%sdZ|@B}
zcC10$k*3P@*#%<}pur9lp#hH`eZfjd-$hKN(XTy~D6axMz^o%XtgErj!9=A4A$JV}
ze-i~UDikq@oKZot`39z_0)XX;SOw~V)qZoe8-yC7&lm7Er58wXlywg_61=l?N2N^`
zab(5gbWL<1TuMPdqp%<vUH9%K6LY1Fet>U;Bpze`5?XFd@8diAtfwe0VLzV|pT(Q~
zd@UsZ-UfZA8~Ng9lNy)9Eogig)Z*UAg`NEvJk#06b&gp~y(s^UWBo3v5#Kn~#_+hT
zz^wQ*qfI6M#QMjYv3GwzQ_|s4=)Bn4|7D&mn&Bqxkja$bH^;`|#@w4D7N-4bLF!(0
zo~GZs_?*@!$5r#vEG=3&uLqEx8&pU#8!M1vxOh%3;#FD)_hs5Ol^+?-S@Tq{d6K;Q
zQ;lBS81{{8&vyE?O_$P}Ri!KOe&9<l2@Byr6%Bq_KTTJs7eY?YzdoaPMs?CkA*^8x
z{Sk+~YU%oKgYPgC{hQ(X*`eBN=~aLSVv@YjMDg^j1HPp0ObuZk@`t2)1CGO(p2~R2
zyP_WuUTu9t;o2;#d?m1*p~COr2VLtN#qoVH#KPRXY^>9(OeYopIV|v`2F2m*)4fP*
zC48C60LRYCvkh)lWbnb_GR0C{tM<f$C(3h0QgWx}2E616tfAk(Z(?DOcNdzKN|@{~
zUT)Vd_b!`_0r5_#nl=VKA|L?zR5A7vveOxjmPu&VP_V5>;w(h9Sz9j+`0<HIzTK3d
zH8-H7116cEqT*m}ST`K)E);0a?*)gQ2ix=2t5<58suc_vd~%oT`0?YCrVN+_YNt+s
zUWt4<rTet|v9HByqQ1cmPSUOmTT(e<3W=CuRTL8tf6N5b(h8k?g)|~3)A!Ar6Du!|
zP*a<}O`^aW9#bu6a9h#7Acsq6h;B!Aa-(E!q*=z<SXTB;vY>|Zp96d``}z9jd#;w3
zUKN|vqE_9Dyuo|o_GGA$RicE6ADPR_m7lH$=VcDfoj=WZt@`9yn&3m!KMOkwmpq4E
zI~J}j-1=;`iDqTtYGsI_i+xYU%i1HB-fgb)k8iTcDQGAu-x1rh8r0l)aPx?h?Py-Z
zk6X2SJ6;wxTj|Qxda0}kTFoZc6^mC?Ggp`IU$5bt?^8?jsg!h4zPP@pmAkKensJG>
z&l&?o)|rxWv*UCG{38WenOCUrNZf;^3*<&hfcK?i7+n5&_Zo0H;d5moDu<>ZZ0d57
zDvjVEfiB1Sdp%d`Ig_eXQPCg2aQm^t<A*VbOAxo~ch&9J9U!GXc(5Lk5$VCXSyPq*
zgcPjElJDP#Ctm#!oM;0G3rHZ+^@%qU*lk<3pgw<jIS<~0dFy&_<qe+rn<u*a?vCd8
zyarSusxwMXLP7$0oDwh-4A~1X>c=mJoxBxT|4ZHGVMrvfGFdp}SA;784uC-($4zxb
z4x`~<Xj%(-KGN$EfeOGk@C^xs>;(bzd(619uc6H&1d6=8yzuZlz?K*klZc>1@QM*^
zn_WjSQ}!`2@$5(<lZwmKPobS{>Dh>jMm|XMm31;D#Kpz2Rs_Rj{eZ2p4!ruofB=3k
z!F?@HbK(^;c>#wR(b;TJrx2vMhVNBQNFzPU&?;eiMhwZ7338(NS1eL`Ahnfj!oKgs
zX}eN^zPUd>oeFa^hs8MB%0_kATJPs>b}?SRv&uFrzux%q&7H8`J#G`nlB4w>eDIx!
zmqbshz+`bOi@A__NJ)04wVam0Y%DM1ndSaoxjMCH`9Z5PK`s<y0&PWASJNN5a<7lA
zX1VsPJ-AV6Xl`|HCG%UBQSQfBRR%v^9-;RyCH47EM#wyS`tm8aUV5A@0}VL-R>-i<
zo{c69XaG!!iHRYRLGBJ@#p>Qdb@f+7VZ!p?NRFUmBX7hUN$3a7fq()zBWCe^vNK27
zFdT)9WS}a@=-Wuj!<d*D982-kFaX@Xbt^J<hRc^V!envuiU<q4J$~#2EzVZb7weuz
zDGj7u$n5Yk;^=I#J%KU~AbY@m|1&qIAfJsq08_jy$6hZ^w86cBNO@l5;CP0Dh^VO0
zJ41{@1%P5KRv;SV2|-H^z~OB4@rUituR9Yx)h3#TVYx+v>Z8&V3~+rnQepHwh$(5<
z0=oq;UDgpm5|MAOyg`<Wd_&ST2nV;cKO33@w-~l#C|2nn*9;t+5GG2l+e*jZWblCq
z3YU5pIRU0RAiY4-lAu&$?kq8v0ePXyOG_h+%JNK3MuzYK0B@49C~1>^eJTbN;qS>M
z7UNWHZ*6_@<cT{>ZINrCH3EGH)DUqvn-Ov*ZN>dyr3m;9-2<#|q#}W>+PxDW4RzCP
zg9n5iH*MM^EM@Fnrh^Tl@R}nVvv`5h(emn@+(PD-E<E|B@Zt6f;k|4Rw~r*p_Vn_<
zKCZKcY_Y`quH0o7vknJ^oH(@)e?F?aWEclB{GxDQPVb6d8u7MZ=7{t!4dFQW?%v6R
z0>LdQh9axh;<?^x&3DvKicGz&$Vv5TWgbo*swcCwr@Gk4sLmQ;_6+Cdxcqc6ug~lD
zT<iF!PeqiJT9D9mAdN)bNKP+H7{%kQ)*VkbD;*pOvHZ=k{BJIR>)P+WzC#og6vvMf
zKYvQ-#UaR8kckoUZikr|o}Vj7moR?TD6+zSc`os0tQs6JbFW(ke1TirZZq~XXU?#*
z%h7qEb4P;%5LhtbNOI%DV>K2+Tm1SpF<=s_bsY;lH1q@(?JAL^;pwB}O_V>{+S+KY
z8+K%;pv8*e*YWX2c#Cw>?}?%E;2FvH*VNRg3GAqJYR2goU2swp;|-IOlTjAtqYNR+
z3o9!o6q`86pn$=&GsDJbQvmD<Q<9}TodT3zP$b`suj%99z#wi8TL!$g8#iuXSzJua
zz_J`|&#HdB86Czn5CJwbepE^+eq)HRC@L;q$3WcvZq0#fV*u|%TnxWz+LzJ$l;e_!
zlQ%a89p4?K`JX;rXLNLM*hPBq;K2-E^AzDgUs@i!qO!7BApKY&8t^c<e*G&J63|d%
z<539u2vR%l9;Q^HmSS6O08_4AkyAjAtCsuH_XBHd@P8|7F{fF4JCGa-68=%gY*5vf
zBIOu-6qYjY4xD8z+FbM{?+Bj~Cy{fBN-y>FI(uv^GVwvfAZvczsy3%R`IMaT`agB(
z`2QJs@xzvUSRp)6byEeCp~2{rXC4<1F^nkKSk#wASsNIKr{qMfJZ2z|&rbd#SYM@Y
zGI8rNdX%Q{WFYiY(Lml4G>)CQc(LJ4y92^LVEaeOM8JTlh}W?B-Mdjs<^}gwm#iBf
z-M@byN5OT)3?!G#%Bc}J0+&?Tr?&zm>41?Rs47+&6T`zxNdIGw&|WyRz&@xR$nThG
zc=J(s_%F28tlZ<VlwutA{K50VjKaY-MoYCk8!07N!ODA4Vwym1(?3^;WF8ALWsDE`
zK5DsQ3z%~vtKr|EG5o@gg-B(;daf=A5W*%~IALytz!4E0EdV$Q^!MUj;2EgMwP593
zfc2PmCG-r&ow*!H<1uX)^$)_|VUPM&Ov?jivS%XL8$A=sqetoA>%GN*Ry3m%G}B`i
zWG&VE1H5b#YNNhAm)@_EcGrDv`8!Hnxv{XQ_+q3#7te~ewYMWZzfnz5?G~@0?l}Ds
zm=d4e<mVM%6ZDVXFHLQydB-LnhUYb0qOclwcP}w6u~x2cq^IcS+efc@PELrD^L?t`
z_|(;5Wq3+)d!ga+j%;K$Aqh)54Z$}`K6n@?l>VsVBex!HztcU}QI}TxLbyO+SwZ<$
z<T$G}Z{ozo<&Uk^U3>>KTMIp^IBue%SZbOp%u=-G=u>%?{4lLUxj9r|C(9Yevqnf6
znVZuSG9y&N54%Uv_y@mZU<(V=4056SagzD>O^fuxQVa68tfvcG=AW!vtjE2K=TvXC
z>izxXm~tECv8+D^hkK~BMvu`>O85VQ(kNVSf?@V%4d?;jp)_@SXJyTU_+Z7)i++{0
zWEw*GLF7D)!#%NZBOJ)OTXaP22tH=jTh)&2W*by@#M?!{mVb@2FU<8pFBVknfGPc<
z7m%9t+3de-52Qrlxf)Be_3PSwDCtA^7G1|hzkd7nTcun5HPKaf{eu?)EDI_W;z$6g
zL>^OZacKzfhfQ~{_Y`gZawHU3N~WQ$5cn?0qt?*53gUS_dOwLn<1VqhOh%Ce+oOes
zGsMop0q8QJoCBf&UdqQwx_9@EoNs@N>8Yp%674VKMwRJTvX^J5rBNBB8<-b|?b(Kw
zH5jK2@2|&3;bhxG5y73SgSF{Uw@-M#`hwa*;M#cYdu_uuo5ul~?T_!c8gD9kbaDsv
zAN5Hs47kRT5*@LVx~qNTQ-LP}q!Z}{^+|zww>|H$rLl(;xHw$=GgR7nI;7B^#jDJd
z-P12fz3HCNm$!K$rn$Mlk)fKOGdFmVZmesAVrOijy-;&4e<Ug?fHx(sfz_x++~m8;
zO7+j*B2nEE+0nzpufo~aUK|iRwd}BD=p$-#(rd4zij_lcc!bCcV<9#1j;0aSWs{M!
z9f|U$tP|?-S-q8UyJH0+?yD^em##iu?&P#?chz$hG!(?7&NQxSh4-2lZqM)N6;8TC
z>A^Ag>VRGJA9J9mjnfGORr5J(Q^Vbbg~3uaNO2ku@&nVGW$iYS6Zc!I_N5_$`JCjL
zt)_=a3wj)4%zVzCIRo~`br~tkt_-BCu@l*Qk6)5?^#c_FaXZF=X+?Am8A0R0hj3))
zT9xV-U*NR2v)gg{RyZK1Sl0u$$eC3lxUIhJLo#`U9J(DjOOlu32zEmAwe0tfwE?Js
ztN=8K4+4&$Gc>P!9I|?cx@JeA3>5Gx{5fJ{7>RI`Ws4kKp9wWDhKnL~dj9-5LIVb>
zqwWV<6{x@Z_3QP@)lIr=%HTBUOQ{zg3TSsRs5|~g%bk<C>b>d^HU-@I_Ingz!pidj
zTRRy0BN6g<B4_@f`8<(d={xzEv+K4M_Vsl<0U~HYfB(lh_2R?r*Y@$1{?VS)dP}SZ
zEd$P@e4ny5!UK)Ce1u0gm&QBho9zzHBcALM8qp)274#ezM$O-$ef^IRzq-)C?ZDH2
zbo<N(GX$vSI`=Pn$*0~EOFUq}a%NQQU9tdncC)(5*#zTa{S5NU%#WXV?Ug=I@MxWH
zZLn|iSZ5@Kw~^MP_|km2#Y^gzTW*Q_u1r+xCWUz8pli}Rd7`IK;L3r~O*-eG6_n$*
z!<m5G!)Bt(q#@arncS&pkFiKTR|vgj7l*YDlV1g4h-mb&;X!!ttU0sbG-bkzq1}7K
z4>8;tf<ms5rviDy9<yGc76Bs-anIeVa?AIq9kT_eZuoFU<KqW;qpq$$Zr{HBxWaE?
zVL{)ezCL0xEzC10C~;&hT6b}2DM=-RfrZ5suLF6iJoCv>G$BCpM)iUU)uii_^1`lC
zom8AOMr(@AWNt~nF#<dE@x?+5rEsa22c6At0m`ukMuV6g5gGa9=~KA73MrjsW4k%x
z^9hL(;PHlaYjblr8oo?o-A%X8Sw3;;+ZSQ|wWC8z#Ze&x@wXzLmCk9~d-N3T+d`J=
zbd>KM4S}37mEV2zbe^3!zWZdBCq`e3RXOYFY<JwgZn5SSD6P?G(pY;^<V%TL_wL+d
zv%20(nk9y(7z31@{2JA{Y;_qkY?oY?R2=mV_)A7Hj_8b$?#!oRP$`(7)6mu)yO{xo
z<Y4d5-d{=iVpCp<GJr}BCAuyy>Lwo8&iizmuDeEIOPf^}*^k$Jd^1vlO)HFa$Dbud
ziu*7uG3qB3A>xs4cF;QrWM~eerP$1qg|-tK==XzANlfQ~7#Vbd?&|95nAdwG@_PUv
z0(NaiErHr`8t8OipN@J+>(Vt#Y*8pi<Ef$6G?-J%myVl!@FHVSa%Z%4U+%5UlaLAL
zV(5EiWGpbyvy$($dOk>h$ohrmNBoG0;Keu2C=YFHGK1fUM&1Y)E*4Z&eEa%X<s0d2
z6JJTXmoHwh8Ja5?DD;<FE02JKMHr~|3XaK(V*K}^!-xBSO^3+PsUIY<u6$lEpVaEX
z(UyKm-5s{&e_;Lwpe1g7X>oC^H7y#6Lvf=v3ajw01txy}d^9)$ju1nS)YdbP3R#&-
zCxwVl+($<%o+gl!^XfGHEU%H=4(i=kCr?y_6g~1MHeb;PV@^9eveVm-6<sT^Ub->f
zIvrfVU$Xrbt9pc05=VxT6Md7<;HfXea%@7T_jq`oInhhpJIVFytK1#312j5!^?0~f
zs!L=~mIw;%I#es5BO`aN`DU@;#D}gMn!?9g9!k<N4k}blme<)ue=qs1a{mf_bEUq;
z>h2n~Z{Pbkc0>&J^d!SPfSh@aw}JiKx!*vjj>0_!9T5yKZUcge05+5>ANg(N2p<DO
z98v2VH33PytFJ$V0m;Z+XGb$ne(H`ddv7bn&Q{g*W_rCLO*FmjqKWs@-p8z|4<<ZZ
z_U_$_#Hf+!2K^x*m0XjPU>a$M1%FfmwQe6dFn#@`vyl?m3{RfqxO_}5RhjtU=YJM_
zU0RhS?&9LsR^{x>bjf)ddUe)<fr+<&UkJc^B}UyldV0{~oza%6bZ#QR(CGo!N6*0u
z*^4%hApa^^?USTN$4doYF0NmH$4Pd>(%Wn8y|l(jyYmqo54S6??WX?mib#MxE4)x}
zZG5`ewCXoMtN5<|Zq-8@Qpr;7G95NYUQ#_3jWaJhzRUIX6-Y%&#*yx45s3B-|FOR^
z?^p&|g5ycjj%U`a`j)nq-LX#ZxAr^GFs+-ij`jAM1NlYf4@lhTi4i(Es4R}dIvI?w
zp?_$62}@*@$>0}#p1g*x$_$@TRo>lF-@Nac2K$S9mwb2qL@t@6Q7Wv9QpJze==Er&
z<%eI>miZS<_b(L|i<Om>h+NEVTpBo$Ya2X7cbm<hiIY=7Nhv|;LZz67f?<7XxgRSl
zSb%4x)6b)*U(FI4;c`H$!{+wmgYux`Anj~C+#vCZG&b{g!|lfB+?;}z0#XN~4Da5(
z+m&s8ZMLPVO4`b{ice(n%A{5(P?A1B;s80}RrZI)&aV5^#>(}1A2yMdA(MF7*=R&{
z^49I&%BQ$nM~{ykOAySoIcZq8U5G`R<v_><#lBx(`+0nXEyo{m2Pz?JT4U7N1H7RP
zcZo^|a%|+)pMYN&bgj92cpxsMt*w5l^!D*NeDr8c!4oB9@F*at1YG#o*!}<w!dW=1
zTK+-8{TY%zIhhCog`oJ-W_$HX-Z?aGht>_Xv!TV%UsH^|g<sqFX0wo)@<>iR(-K1~
z)|J_Jf1&SHeR*b8ACB!(FT5IH=`oW|oU;<mDz`q2ynId`=a!<Bt(yDX<?Q}Wt}&{5
z4EUM-1?uM=+avs7aXQ?WL2+_XuaKg`A>`8*GIXb%CbcXG@F-cHd_DB6MY7vGr(JSf
z!f|J#Fb~s9DoS>1zZbRh+=yAo3@_ssm}YLQwmVy9H7@CI@>Ca1dHC~zdwx*<kZd;B
zG2zilpLYSvr9*1RG8XP%?24WJ#WaMXa%phI$i;5QeBO=MPpfCY$DX`4o)`_y4{%i^
zoc@w@AdC}MMu<J~Tv<Ni{pd*&&0U8(l}<pt<X?{vz%!uvq(EP{>Y)pR>WuQk0_=6+
z;o%^oR(7)&-F$3Brl+fmVcz;CmUjaXx^0GPu3CB*p~%_O>MCVLA&L^($?20m6QjfQ
zC-Cx{^5mG?z98p@ljc`(z*VKCqP6zBf`ix7dfyfnt|QcE46Ef@iKR=`ZW7-^q5shW
z&1L3z{8N;+O>HM+qVB9q)e52?#h=Hh@_uNSIC#l#5;oft2MoAHaX~EL5+Nx_bPbVW
z#(mxpK$9;c`606tkY02Y>~0#RY5eQqzfsfJivD2JWY;z_`~J=IWRH#9J3R{d?=4^H
zjoS51U9wO_-{hu-Xr(8Ikgejv=he-L-w1YE+^Jcq-2ID}YJ<4e>{qHo1-2b)6jS}O
zaLJrNzBPN1SqzsHSMQ#62f#9t@3NXZm7lf?y(%P!zyv-5O^EX(7~-1Sy<)RUu%$sq
zej8j<KIvEb-Zl|EqPXoSB?44^26OO4NJ*ary80WVLQv68Kbf@*nxs08BW-(8(P#Rr
z1QU}K(;yE%%GLKo4YD&@;yrR$(9v?JLHNuEpAnluIofDF9P!odQy^Rh_c1r;e#dj!
zsVGlIN_vomGUV;slRSwD;0UYG0?+yhIyMW_ZZa~iWLc}b?z20(r_dFT_iJ*=LM|q~
zYKPXW8`&R+$({!e4L<q#t5`Zutc`oU{CbGAkurgT?;ZI@DpK<O(e_i@jqE}mJ>|R6
zCf_&yEmF_#F?Yi4tuHy~?^x#9#EIFT42riUXZ=Wb(`EIWb>xp{wO>(CkR8adt(amR
znNV(#OD~{9YEh7}>@j!b*x7f3r6tcvckZm6<_B}yv@Kn1wxzLAK6};6Ve%gO)hK+E
zYygMo$1Eg$HVlxc7H1dGg$3wvNx(}zRbPi1Yj_^cd~C^eUbflXesNIU2yAi`zo4+~
z<_q6~Cf%9$^@sNF*N&aTyetl_&m;%;j<(#;5g0`Rz0>us2l)IHH*JS%iXKl>Bw_M}
z3N3PTOu2U;w-J+&Kv!8EU0pO_&*g@Sp>K1KV;_WSUcC08Vd8DR260<Pi?9~KFf1`8
zjd%rSGc&Ww17u`mz4C*7eUg|_6MqjcySV5y5E8V9YZ=mo1Ib;F5o!h|wL+KtrlY4o
z*+X|jTgqsi-tY$N!>Bo;_joDyOd%!EeV%?R_%Facr}gt@5lm~O?yp4(g$=TEq@@@D
z&!b&M*Midw|FsM31kk?!*DE@iD`Y_Rkw+zvvUIF}y1kNhNI#F#NYwu1rmjsvaR%$-
zfT4@tEM0r-^CkH4vGz4hvUFP>8>@S2b>`25<$HIPlFrPG<$!R(Z8G3}Z@)78$?7Ry
zrg7jL5S_#)=@IPKwG~vWhfHrr<_s61l^GTvgpLwzGo@*OYTBT~myS~xqvci22A((w
zgl1=7s!f?;K+(g#;+DoeLBOdXQ%5;CLD7Vy1G{<W4zsOBCc`>WSn!ku?LlqGztDDa
zT4V!97@-a|)2=5!5s*L<7^#bVIa##`a(^hbq&6+V#5d)&MyU*hW(5*^&=bM920G<|
zllR#(CFGuva$vjhqYFhE#k&V?LJuE3GHQsa#bN;<fNnoG1O<$Yj4UkU;0puT+<|+e
zayS7V4(hlkMXfMjz6_lYRt5Srd`=6**iHgroK(^dHO@z9!+}%0^HoYp3gM}ZTnZH`
zI~yBZ2pYi9C|s*sJ7H^M)3kDX5f+(8Q|IK+zeuJ;uWNXh!`<7#yg;rau)n{n-z$(u
z=bFIW*WpsvKtLD{4G@Gh(rh6J_?qrInNwvFlc9o>TBQGNe?FaOAWDT#FK>6h=xZ{v
zZ(NG4m*Eg^q90PKX>84SBJ)yUX}z{Q+GnPd{r%S?sWD$ZJg1Ri(oFx<p`r0X_W9+E
zC1Z3i-h|uSdgJ4jvFa9<RoT$w_VKW(VXtP@da0g>8`n{daGgCz&K}v$yM118TweUv
z@89i~+RIP$L_*YSOt1PJSW<{ARZXCYt_OV%DkZ{NIPnj9O)gX4j>MI~nht(hN2t-i
zvjl+CsGtDgeht1Yv}$i|qa_cizI9S(&GDyB*f;m@lY}P;O5n@Tq9B!|QNM8uoh72J
zy7qB$JU=o<?<N>5aB9#)T3*b9v+nR}N8OkAwkG&*bOvO<&7)@|gANwAMMTm;=;I5a
zqUeUl87lD%-(ICmV}A$fCa?xv0Rn`xT#W(TkY~GI{WAa-DQsy{QaFf^1gn3tOSvI}
zE`cf7KY%uy8C<7+J{e*bVTQ-;zAD}11Weon!J;2h`+A1g9v)Dhq{_>f{1bU|5uWrr
z?_5?VL{tvxrJ*B)Nd?$Xnz>f73X4bvXIyW9juJoAv4;_Kl%k>{5O(X@u1un@hIUS~
zYY*PM^;p{7Js@on;248(b5f8FRjDQJOaSg}==JHa`3r^(p=E+h9dlPs<nEI}w+B6n
z0XHkWg1QVf4|n3wL`B*4sHZ6XUx)xD=ow1nMUQ62zjlj>OB^tjtw}K~@sJn#zSzd}
zob|dXD^2^o(H6H;BQNyZ9Gs1uUQ|^&Sv_~Aq;xuI=X8HgFU2pux#6zcIFpXiQJS+!
z=m#J_1DEb7b`Pp?AoT$D9TC6~HDGTdp5Hrz$t@@i>G(Pt8?6yV<4Ukyu>FaFLwK%-
zcL@m!7K3mB&<%aN2{eyTv5icKfwtF}XIqKJG>UW(CFrX8b6|}{>}g*n!m<WY5nRk!
z5ZvHk1k@HyM9n;##E_6~RPncbuxU;5zwO6im{j1p20O=ev+gx?u4-#*cRe|E`SKdJ
zJYGF)5j#Qk+FiKca3D`1TzJo{LoZGQlt-{IRQ-$4L++54n(FE4iH-*`4hcP=U56Rc
zja&cV_Z=Ah0Bea=iIG4<E35QLxz^)Yw#@=lA`N&okEi$V=2J&9E_JC!iIA^tk$59o
zT~R?$k#floAOe`8o9b?u3^-w_a9KHAssaML+sNBe4sxMsJB#a9ik0{PleVZ?k#Gne
zFjV)gw{xi_ksZ0Vqu#<kgFYFLX&g@i?KXCH_7Kp&FbV~ii3cC&We&vv;qndAh82eq
zDUrmoJ$yP(XLp`?17@IlP8{>3r|jNnc(CbpMq0ZyNJ>ia&6avzqh+et#(HMSdgh@Q
zsl~^$1AfVxbKU!ut=QIYJH5|FRtK1K0+kOeCjyhm+yH$1J|;_yaKTgJcUe)Q^Fk<-
zoQdkjX2xq|%;j+-h$`qQ-O8%~UWai?GV#S&_wMT7ei6-NF0-cN<|l69xmQ+>GJzH-
z!9gcD+YOKgi%e805&SYD06&QT{BAvI7H>REHs%OJF0?G~F5KD8ogwc7dnYu2R7N_-
zQNW`e<ID3TFK-#1ORs`XXqlN6;>xf-l2r9-H0}K*2YPxY;a()ce()-~RjsWiEy-ax
ze~JD@ucixxFh~Ws(MpJ)tJJ|RMQ95g@@xTg#DXh=3Lwp)*K*Ym5b(-^dE#4&13-0=
zOrRV_eqfN~C>MucrVGM-6SZNOkWuyvNjjXLPTfteSkXii0l630Yq1PDSpIlMvnzEu
z+1YT4la^tVrmYsR-$}iLT4zFPQcGo{TYb8Y;75{QpL}yxL0zpeLeOiQ?yQOVn<!Dr
z96g;vo;I3PPN7uN#i^Os2PtK&lAE)(+Bk`?ya~E?^Kfibkl;BvX`V~xIBDnD3#&(U
zY;tH&<msd)yiZ_m&RlVj9&9^g|4j3aInEr|sAeVQ@mS<I;fO_xq#-TE559<?Fj^<F
z5BK$@-|H3JJds536CsW{9}yKfGu$|W<QWPaGk0TU7#oAG&@la$K-ZyIpeP8;UYrP!
zq@#q<Kj>>j2Vh+dD-$mZ%I9Z(etr)+*`H1_wqmU_IljQ+t&0?z>#qHXnQH_|D&@}8
zd;RKR$im<ASR0~Y8&I)ZInc|i9wiXyR(DH1i~%-4rBjSBLdYYLlWmp>TR$r<(712+
z=5QG_f{<oFc{sRa5{I4w&KS@S{lP~FHbeor0^ZD*FJJcjzR%Itpi@;5(}d8AigjRV
zd1P#i>iF>$P$mtNQfME5-2(!Zp>IojI|!1u$}hL<$Wfy{e%uL+FSWrNJP<6wfDFvk
zW}Wvxwq4$Mhch7~RHW3BF9|4u&3|-t2Y4z>4|`621i-Vmon!QK=jL1nKKt2-?o1cI
z5KZ4yMY=bmf;PgV$3NYf{1WjrmqG1SK?`q$?XBD1+<G^37zbAP^?K$W42BF8&>P-V
zscVxkQcX0NU7eaZD0k2@k(D1{5WJMF@u0vTCD^Ap{}B72;6oH4yw_kSh7`-vBn+De
z$x|n~FWAZQ6$ymM9A6`%I`YwbH;$_BB4@G#2@QK0soO8W;0Rw5VqzCen#vG`lU54<
zc2n6dT$n<|4(khgm~z1F42Bl?`*(^SC@v8*_7IYB*a7#05Kvr94E^Bi$mSt<=GLz{
zaWGeVEf3@hFe)u!Uy3L1=H?cB$y{(&Ffuy&2(AfR3AvF0s13+OBX>s0qnrQp*RO+z
zXS)S7pkggRt>!>WHGP;#X%eS0kxcI_nOhK;8|N*EjdSg>%axcX5%(9on)ymW_<;ha
z_ny@KZzJjHugUsfX1Z}~?-s^Cq4W~MZy)|Rwd2s67i_!mzibi)VUIF@FRi}qMxeFu
zx*L3EY;g7GYYPilb9Z%|!ThOeZVYoddcKEn5d{MR2Oo%`5FGjj1wjz^YhpqZsVGh^
zAt53Elq|Pl%;ZMwfMN#jXh&#hTya>Yr>8?5nw_2Pjk`e1Lh|VG?AZ{)%kYQC0T+*C
zd6wyb{&6w)4j$to%>0Sr7{uh8%T5cRwqvi_q7DU~1IZ!!<sl)OfGh|%vS-hpfjGYE
z>S=9Z0h7cmVUOcEC}2^(;^-A{S~$<lT-n^rMNdyS{1LAgu)D?eKz0GkJ^$tclyX+O
zP{?!W*J!P;twJ(22<#1XSpYZTN23ODcaS~;6F^BtsJE9~6}iH(rC~4->x(6G%_6pj
z?Gzv13Nm}xr?$YV1kFucEXX_x3JO|UTBr&D_BZ2}aixz6!jxBuLX_m;c7A0ZyYsZf
z3FYnH^PsfIY73cyH*IymA>QB1Y29STeCHFym4rUN*vt0Rd(=CkL`*<)ZSix)=0Act
z8Ot$8w6`B{<t=E>HRT_k8#-E-awAf>pTC@!n}@&L#cF&lr-d=ay!qD)-2wlk-o7!f
zvt}U)as|?TmTxQLFMgy_YOn6BE~z|QNH4&Dc2(nhzQDxl;&M>;>vc=MLfMabnu*`8
zmYB4lxT(Jo(vN}yj<$yW8p<W0P=TtFw7?6=7;y01)d!c$tI{%YzGEO-NbChD=En#c
zhRW)>Kv=uWW&C~*+$QEd$4&=^wCD?HAeda91qogF`t=>--vOCp-dU8g1`ra0R4^`i
zPS)nKo48l%G?<kw580Mu^@Md7kGF|V5-hPYN@IS_$=wb@L-cn(-|b)Xay0Ab&nQMc
z0@j{pJKZbAt)pLX<TP&Egy+q2`t%p#GWJsQ#Vl#c$W#MbQi^Avi#HDbj*3r$Ie5Gw
z)@#9z;)*{}ca#rA2iyC!T8d7Rc$-4Lw%jv344t}KqbG7Um`*j}cRd1G1|@b9EXK<`
z0adBQ3)sZ{2>y=LB;E)1iJuwxH_6)U<5N=iUu|FyPj%Y6llp$0yVG$kE==#^$nbv+
zcmyLr?<B5y6W)!o{Ov@)h;qvoXYb1Tci!IUtI#!iVh7=gmaDA({bh=+D-G-)*&K;q
zcJnpy%b<T~7;wsQU0*}6N^A^C6}ArW1elVjo~O{KViqaf%16wqpr@yoU2FdG<rl`=
zr`&cW@uu2B)+Em<F#99^NAR_W+d~`fILm(yCiHwtkG7IWQtSl+U5KJjVf581Ltyk2
z!VLT&ClWeINpuetJv!`DDZ@X_kT&?HXA}J_1o`(hT(CEdY{M@0T>OWmeDCF1$rA1@
zvMLue2x}L7?0f3e;hK-%@ol|#a8Pe-{0b`{oQD{H-YKP?N`PW1e#1(K_|a$oWuJn7
z^l$y!|8(*?{$C2NON6#?<HLNz6M*>i=YP3-ef=-vvHur;;V<JEU+`EZB_%K3QG{wR
zlHX}THuY0z=>6B$u!B2y`t-HbPxM)$YqBia?DtI=L>bT91>PRRL=mG0(*KelMk-^G
zTJ?1t{c9cf|K-h8@lZo5YSfw<nVf{T5_%nEvz<G4#u1{8Se1;pjHS)^vJ?){=Bb!#
z=H@9IKlb@LiY!pI<7k0{EW=|Vv4ILdcJ$R$9R43Z?*HeNZ#;+ptn!V=@PEO^{$)Hy
zC<S|5iJ7Sc{dZ1ZLnBy)z|dgyZGG_)BpuK{Pg@Z`MQ2IR!jfr2_^V86qPix4lpGZ}
zgt!31Mna!M0k!b`=NtP`OBKCi!|zSVJFl_fAAgS8h<wS+s4hP$ehki0VK=iOY`LQT
zdX!oI@VU+iC=gw;%DO}N`j5X}X_bVG2hHc-W4kFXmVW#8ZN>)Q)Dqr^t;W)#qN4ix
zF~7&oD<P17-uBnPOeiY44iz^!2<*O9`03BDHsWu~2iz4@r~Y-tD147o?VQ8+jQ5_T
zz(q6f+PfF2-ucL_+cxaW{@S~7mU}-|0k#2k&F$|Sze(-yZ;HVbqgH51vP}r~f7k!|
zEqYhib);q#S%fyBvVFt!Y(t?>(yrU_rV-kdrdYX>QNrWozVhE^qK>+3?#~}KD@Fwi
zTvD>`-<$ITSrn?q-ECiMYo}mUo+;}}+&a(30mE5Y6W9rpGtEfx!=%=POWVI5WIeL+
zn-EZ~j<9b$lYg1I8K6{NdM#jWZ9UffF+^yYh;Ik~<@dJK{fGSU;lum(Y1MAr$@RZ~
z9}J8!U}s@6Qh%SxUrV>ls-hxoFfM2n@ZyQu{{Dsy+=k@{$>{ydTm^&?-Ip)duILe0
zyKCcWZ9)OB!oIdfT@S;JZ}Sj;UGDGAlc9!dn$m}T91zztf9>5U2>$?*<?}##w{PEW
z@_O&as~6JQxc|Qg?A%FkTcK!c4^K#d%2<o_?~B=EA_llC4nz>ral&!$)nD5@)7?<e
zm6eI5IpNQHcznefLo!#STyjSym@mg^G4)@k0J3^+$GHR*80PH3cMN=pTf5?gJRV}e
zjLT{XooESeojZuH;)8w<9`%RQgWt1>$B&OI|2g|Np2Po->CWyZfY`Us0VvL}Y1Z<a
zDg2N<cgvg;n@EmW#JZyB9rf8N%A@E5szmrsiiu&%uE7I>m<fu|W;CB~TstM{4B~xm
z*6j4OSH~op4!3!T*fd1AE8bu4g+vs3FUax$1!4k(3rb(`lL>wi5ESr#`+9n`5<Q&3
zN%J&_LW@W_UP9r5`A<S2pptjwe(Pl2Hj>cYgwNo84Pr(F+Kmv7KpCo($Eq6uSPWcU
zB6mfovLi10{1xnbPSXzkF%55P%uZH8FD&uz_eD3pFNyENs@e$OHvp~=geTpGm3a22
zxmEM9xmrR%$API!eOo?#J*wh;<m3?bXpMPXNe3#UYEB`d`lS_rP1@WQ8Eh(UggXXJ
z7g#TTBf7_p8$-~?uX+I^6rpj}8Pgzs_xGMPY|DnojgKJU-TeHkm^{>&fQQpNGJ-~d
zTv;WchG4}TfarkCZrQRW8?a7J&S4m&;R8GeFqTj#1DJg7?p|45o;rCUb{dwC0ITsd
z0GLCJa(v&up20zYT+4vpac5}6T7Z%Ptbtwk2x7?5%T9J^5fB0X=+Sq<!7czaH*eny
zux=cb?I;>Se3KZ%cjSnkgPdWU;>Rv{K%?%5bmrpu^XE^T_}tpc&Ck!z%#4!82<Q{x
zu7Mt!_IG|bmumr8!I(@48lNHjii?*+N=|6BS5@y>TN7e+0N(){R)32qZh%O`RObW_
zkIJ+$FAvY$?|MDd-vHuv?A=R%reb2;*2A}8tbxA*sk8NHDGQ4k={emJQ=gHJ4--4_
zrVf_%Hy$SP+Oe)pBwEtI&^>AF+Xfzy?%C5gcTPFq!P>;JcB+-(93!I<E*N-TfRxnx
zOSLsk-XK)c<|*P?gVBA4nK|(*Ibaw7n9ST1NnmAq1`>unbY945#oji(^$rX~>8kcC
z_#6jXF?j=-XpDOL$OQ33I!1%c!o)O+k(C%{U!&KI$JwA46FP2*smF8Qj%XetCy(P_
zYSjyg<%DzHJ}UQ%CT%mQi9KOYvxway9|-8R6%C}wv24&a4<0!1;nzKJK6XPZH5A*l
zzYMG3KT^{bw}hOT*>-fm!C=Id^VajiDk)bPtio|1N<t-1nSQ~q?YyvxN<dD)pP3nA
zxF4v40BwFWMsc1NewB?M12((5x+;;-frW(Ltg@|*a6M)u*Z~H?A~uNc!37gjQ}MO)
zG0G;(7<GOB_gJY88-zTBwi0t@aCTwKER8PVc9)>ko#pmw%g)Im76S}bJUoU^`mcFC
zoE`66#+?B?^KxqV_DvU@sd81qhQ=C*KIZM|8!_XFp@s+W)?_(Bd%)hlor03jCSGmu
zR5uQY<RipJts&-4c=iJn<6J(U4p$-cm7cn}1!O796W05%uIU*YGs#hhr1|@|Z=gM=
zd<}r!0>!f5RrhY-(tzX*(T-$kIGY$6n3gt;%mV0Uxc`y*M4&u)$%!{3Uzjhi!V3w9
z(d~}RU4&c-B$AGf4lBh#3>yMk1UMB>Q6))@yVD1$6fhpWF8P19sC98x*m!g@H0KCy
zT(Kr`#j=2sXrMTuqT&bKUbYLHEA6WbWYw4=;z*k4dkND}AURd}peqmaZnw3?0`vr1
zLcgP&;@F-6y9P6Oz-`M2|4oeW0Rhmo9n+;vv3aSfsohME9zM)*_s3C_*TeS5U`?P=
ztxk;XMz?|Gf@O3K`e_JjLvnJi<oTaW-n$>qlTWgIS#TD|dm(YWpLqHF`Hi5;lg!M_
zCr@gF?&aAXf@3<i`uZ}YSpx8OhbQZ)Q>PFb21Z6k;7-xoJB4{#2>wD4D)jdom^LPB
zUo-J)^#?5cM{|A2uQib42F`ZX3`5--I>76E7=fYEi*TGS<s1DP9WWR=bPZL4B8-v`
zreiuo3OsHT%7XAKhzHNo)tM?s(BECX>tXBf6x9(Pu4nr<HZ>!aWkpKxW|C0vT1y`i
zD)cUXj*P@RvI0sOLgM`V5k1}shB;ldO#zUiOJymw<N`NFlSUwNAHu_tcu^cW)CYKP
zWJDFF4;WT~My14s&b+=`x7!9x6=9qA4oexhac4&dklJY&I~c@sL;k-SLcqIdCIH2&
ze0s}|>AXJwb`g;hLh3ovh2_+qVKj(`Gd|vhBp6Ezq(IT8A7y2=oB=ZPIq0YW%bbLI
z2t9f@<`HdrWX<R>#WB-`n(%Lb0>Baq=*5Kvr~wg;>EUfA(vPuWVDiI(s=cw1Fg-<U
z(X$b%$ulAx0aJgP2APzN_@Cl^Oitdbpvvo+J)Yna6u1_Q6u7fCKX^0Jy(`CVi(gJk
zPOcgK#gD}rmlKX)C91Fn*8o~xh1i?4osRYxj+#!@fWA|r2^ERIqyjQ?#O)L)!~)2E
zFN_BT2ZKj<Cu6Ctzy(Z8b&F=C$^iKR2+KLrb8s|*!8a9ibuk+aP9xqoP^9TMnPFv`
znkMNun`G_%T_xv)IoEZ1>#n~`OM(PkyAZ)M{$RdQ?48$u0u>TR2L>?Im`OPev1VyT
zit-VQnMB_&@jm#i`Nt80obH1B3neQww>8?>@~hpfKU$`JB37@h65*c2e$UEk1;ls4
ztrcP`%M5!QN|nXn%HW`UXGEy<(MI>|$CCaH6AARZ(@pGvZ0<iTtj`h_+Kveyw!Ot$
zGXzaeJK?A_l>M~}#E1eFF)ghg0ANPf<&d%aFs;LFQZYd(OxXqD3DZLwLZhsZ^==&G
zLkfn7T@jhNr4YlX%U?zlyKsHgtkDzpp6}irEi3=d&nteL#Vp+3(QyJKD-di9`@y0@
zfwG2QD`n8p<k6YioO6F3mENdFS_Ztm?|0(9pohJV6CTwYy8IIdM!3)y#X1E1?SR4(
zK?eseKGfBqO~4U+0(kdvbzdw~v^pewA?hN$PY2pw3L^UV-9AEez0_E2t*zx5S<NRu
zMu0pH3!#k2qtTN7slX3mj{7KJxFscAXAzq{-6R!(8N?!kD(031Wp-3xGc#8Fw)ks&
za3kSoG=dV`YG&w0MlnJSIXSub-gros67Tk6U7w+BUd5cVev{*9OyNEW1wkSo7grYW
zc{B%exw7)^Wc{>e86f-{^ES$fFYtghL7N;+(8{VR--MpNzKh5q;9%PTM}%y3*x&}%
zT($PxEUnAd3QWE-`^XdY8Zd3s$d~bZf<c!GIPL}=9<X$24HqG6+*yUkk5O_M#b-v@
z;E_2vge6;_S%#}Yv-@InqotJ<Gz__G_Ak)<U=Zn{kV*|fDU8^Z1RL0aYXN1v{*rTj
zWe3Z?(}qx~=6!PS`Td({Fksk8j1NWxBQY67zoU&ay?f(kMv0qQF1C{=A<;qjvy|=q
z`4g77{a6*$mz|VSU2q!0Yvkl2=(E-MA~9*0a@Ad(&I=ys$MO^J%7L8`(2u!w#L?EQ
z;SI~Mn6a?;?~hjsmK`>p?WhI4qG1;@f?0%r>2+ZoM`sFGz|0ydsFqfxPDEd8X<`-h
zktM|@VGFeg$a&R%va_<vqz>Z7Jl*v#^Jd!EQx?JU38l6H$Ym<_8X7P?ScR4()a2SS
zzs-ooU~w05(z0%xTNvJU4`&^cf*e!>7*P|X6NJ``x&>bkUIA)Loz}Hu#nsg^sqoSL
zav-n%0=di|>SJ<0f_N2K{NjCKEdo$kbDcar=(vgPWnZPD&@db5VR3;2V8D3x8|n%7
zqV4J#%cZC1yj`-Tt=|MzK(CQci(J1R<5SU3*o=<6vz579U1HifqSogKoYLko0TJiI
zXE8h`Ml_9&o_+{(du18HOqK<D`FKV`v(4=kgW_op*0T20LuX{d&mnvsrlLwzxP&+i
zX7=yahnYGG&rc7e%hxo8M!*timC@+5zimAo7nfqHS~ahy@#G&)QM|m{ckXC9{HD#Y
z$WgltUIt+?P?BlmJld3BCA4vfiV%lrXazzU_9RW&Q;LiFp$cpkeWGOnh&iD0NP`Ok
z-w3qZeKy}APP>6|L(J3j`mv0jSVO#Gb)92YI>C-&W*#0I%7!w4pR2mE(&y9W7&J3I
zy*{2ZpgbHH5D+(*q-*o;RdP{<s_P3KHMH3p%8?*|7L@HcZ`v8yKAWOC&jTATPvVMX
z|7jH3e9K^O+lyDP8WMc{{0Msy{1qbNfC3V?dE2*yGj_ke?<pf}J$JR_2}MRk5SFru
z$q0B$S)nFRlq*lViz8s}1*4IPm4N}Y#LTt4Ox{d;`;jl5o$41>)Drmw1YD7g;cR6{
z*^2eEHOUh2&mID5u-dt6SDZp(K;JdP`5+9(!H0nYFVF|5hIL4h&es&41w!meIMw8G
zK^z#Wvj`>w0gkwCH?BvoNh(M6EDMWtsuNgBZJe0@{IHJ%?WGmDb)mCk9+9kO6__~C
zj>?B%ftSPl(QWs!R8UC|s3y1iWv*pcZ6ZnL*vJ^m0k5DH**5;#5vba*YbTQYc!NNu
zLPVQw&IXE<?_zI38q-M$>_sHCqr2M`#S{M2r_4Pdr7mD@ClnwO>(Vd${QYCsN#vy<
zqVi)Rj2v;r7b{Rn@glK%6u>~m<ZQ3{EHF?NAUsaNzgGpl2mB44DN_FXRml#Hj>1)<
z1Fd+1CieB&eoyONyLKHoaNw3iU0oeMy=X)P&S}ncl*okZaj1rYYSbiLAu$mPXw1{h
zJ@B3U11my_%eO#)L0=2JP<ha0hzJI=SV)n<PDj+p1o#3k6!=97tMb&@B&cMR4{fr-
zO*GrQtVhj3SOS6A2D&F=G-_{jvyjnm22~=!2N@NT0bMnk(T8~C5~$cZ!Gc1h#?6tE
zqV6v+Y3Jl`|6gt{lUjQ-$%!j?1Su~tUI;%6BDSbBMdUr=xVWl{%+b?;6(<AY_nu>H
z5h(7l20&MYO&74o8qVDclqQ<dUy6&RLoUNxHPK;*p}|)SbJbJ@1H&f}GzhF&#8i5}
z3ZnQxJps<uZg_C~A<fxSUQt1`5!~A|V`B7C!n9FSfu;dk3`h%rU!nOxhOdsiy(|EI
z_yWiIBy;CvQs074%-VR5pnBCXq2b(?sw-hG=6%G6xiEe9t2GddM54EJzX%m4ZD9ux
z+#4GKoeSO(SbjKbP`0|ROqIav*=>9VRgNdu@1b7^mdJSu%-h=t^9NwNW!WZ}LY8v3
zl0t6I1*|?o`d0T)J$b0##0JR@vM)A8BhKZzCYok$So&??wcAuN)EH#-yNl>L5<WNl
zlwvY6-B4ex@k&Wab>=%}t<5(6fCUu~PuozXiJG=H!Sr_TS|NrtL91!r4vZikM;E9K
z><$xi#UK2#+|&C1y+^2sJ#zjE{uN;|Q1$-YVvHIKob8$s)WnU?^7H2->wuu0OMe1y
zsD^Vk?{b&jFLzK^1M40#qkaI1DBZlB{LmrLWQ@ST#tYtNf~YhO=?EL!!|FP}EatQ+
zzs%r`ExGaMY9w4-)=+o_z+uC&XAlLS;I(U@x<rLn;rHUzrzfU_=1rlydo<tXfg~NS
zKpg-gPR2Eqs|OhI-gA+U0(laikUmdU-{%vPoO};;GGL$%5ScEwzJO1|)koV<bE>xM
ztVDVWv6@lp{bDuYHxJrOqIEjL&tN-=)NT5^|K<XOzN)|n{7*|u!4~<BECxFf;lGjm
zUJV`6wNs});0(l)H38)b%MdXHk}E45k9bTtUmAznv*ECT(`=+6mK`(fU`v7g3yu`X
z`*?XNCm*;V?}U;Uw7)F;Cg3}`)&!KAg@uJcL$TQm4Tr(q1+M}j!DV(>K785WFgRMR
zg|4m$Chn%#A_Pi|f)H2|DC%Rt&;d&m#8QY+W@cu<wE$%W{!F9bfEvb+_)`5X`mlye
zh&4R>TS>lrXeS+kJR!+siwDP%$HEk?^S@vDf4hJBpB^-^3QqiQ+wN7uX54qazk(lG
zKnf$^r8;OK*FpT&_r(H7cKLh)*lw~pL<ju$dhXw~&l#-PI=A;wH_hj5WfNs`oKZ*9
zpHD4jgZYNzh$0`cZ67N=y=ZDjYpef8cbbrbrFPVI)^e1}-_s*$zvZo}GenR&{<AVS
z*b#2ZXbgUFg@ER3Tjx1TxUvmT7DA72Iwrk|z6ho^w=^?h2nr-A#U`4>d<S*H%brci
z@EeV+%k^uf%@2&2`}k0AtI>SL`MUxy>daMM^$=~!8bySC-ic4)DmS5Ig6)!0yg;%u
z08_-Om+tPcmRJEXn?}H8bYz6`&><qtAc)CBwWrUYuOV+se~c&({0*Vl86gVTteBLP
zwWTG9pvJ(gATQs(`>1T8i?enRQsFARd!#jo)}3+kfjSK|o<3o7Cewur<~W&97@{;l
zI|2DAc&3Muc_HfmWlmEA>>j)PkIk!|O>S(<+xg!<!X4hfSsgWP4r@~wiKhIVz<_7t
z_RPxM)lCV6-6&3FIc9%O6JqB08AiC|S^$X*=tqr(i#*JrznT$-rWOuGc}AFu&reUY
zf2yEw#Vp2oBz{M88-XL9$KVI5!$92NE69%Gn^5)o!ifp)eIkUTD$bdwPfJj6p)yj(
zd^%?v!{?mLy(pJptT2b0MY14Y(Qjc7mOnU25PuQ$VK}*9Ky4^&cR;0G;-aRZq1nIx
zhLTduqAWy0co4K%yLRq8hNK0oMju*uP&o_+*@8(JyPhNwf*nR!p901=>_>08&~<$c
z{Sud@-|(4<Mwa6OAjF43K2`keCdX*k$t!(YBBQkMFO)rDcfNRB(YE=cR_2zWAFJ$H
zYAv^unn*0QKgA^yJRfh{bGW3=w$2BQrTu?E8jaboprz!hs<DB8WMkXa-(EpMo%my1
zU3S-)SLCOWV}Nv;1-TC`?KcJ!m12xUmjliO{OAUAM3eWMH!z#O4?loa%IeZmf^VtA
zr~UyDQH&3sS;;dS@u$Vb#aUer@i5#BD7#b{01hWI0rJH_wQ~!7%v+z5q@9<*EyxPL
zTYVH<di6oiru&>oR+f{|dcdNK_r<V@a`@Ulivqw0!x8N$m!bi{6A4(AShY=QYrv=R
zgX>tYMYurZHyS9PKYijO%~;1A;?7U@_1PwP!W74j#j03<!G9p_Mr0k@UwHKBc{1ph
zLEBIWmM)SB{KA3$e%iBVY0xaeT7n_ci(P1dq1Q@SFC;eTe%A13#G|I4{jqjcrZ4pR
zB*_)ZKj!zT=ouL5-aFW%VEvv!=U0LJ)5L->9EHilXv>91oH+pDF}Cq!wp1EG(?$&A
zVQyYSQ`==_N<>bj#;KAJ-FR;sYd~UxjO4H@oUU9kf);D=J?QZI`uZRm0$3#UDF7A-
zv>L{Ga9Oy~z{J21z-Q0Ku1K=PeusFJ!~XyEDEHz~^6sOgjOXl$L0*Ddh$FWMRDyK?
z_;mB6oM%s->;>ju!$gBo$><tDfdr8qW-D*5w3opg`xKZ5X`2Le5>$C?fJzJay5sF0
zzwC60mR5+xkC1cVMZ>R(u$8<zh8WR4!@3iU(n;uCnbECcHp+TW5W=~SWB`Xpwgeo~
z89JJ>a$=|p_;3VH$aIX$^!<K0x=p%X_8{{=z4>F5X6E|dU0cX%X4*2F1%2IORvo>W
zm#PTUQ4N@m76a*tSNH=TwZJ+pbMq+z<LmjnNsi(H?ImvRb4TjfiqU|&c$WSc5+$V6
zhg@O0X$k#5Xh<92Uj{s+X2;qUB?Iqrq2mYyCZiaygFqrttCV}f?*|{)$GBTKBb!6p
z<KaQO&LV&77J*rT?n*BKaOTy~_Mg4I`wt(^(5-NdKFGqrpn@#6)&Jc&_(xeYyhpA_
zcJzGc*F=RRHT(YZ!{0)q@lPSMbI<D|ael}3-&{u}o2DIq>#{)P*}cu#Pl#EL(fdgG
z$j5!TBrICFb^fn;0m^M?T2~N`;SvA@6jH${jE&APX>SU(K`4UeJR2JaU^8MxUUoJR
zw?w3ikb@XP8SsRMRoKB{3CjRz5GdqN8X6j~QUOY1TpP?xu}2dT=3ybQly&##P8n(G
zE<g&1I6xcy0I7w9&>TIggtGxSI`VD~m`I0SW3Nb4P*eMaAccT<918*_MQAG^2!kvQ
zJZuC346Vi8!=BG!GYS>U<qX04(>~ZzXx&(kw$y&L@kW~u>fJDKgyC1sjy?q=V+DA<
z7&nKk8+2UwR4tRTe$o<O#@PWaR3XklFe55KRmJ&OfC>vze^<;1e;+jKz=%S91{Z-+
z!`It8VpJRUi-hc$h@q=X$GB)h#zdv0<o9mA^)K<ON-vfnUVap(gapJqfKa^SRaBFt
z(g3A*z&JWi;pl+_sc?#Lb#*mS#U#NYw7}}3R{`B0=XbcZ*bP!Y5KK7^wkCwv)do(&
zbPPLuk9usSbRwwhz`<S=267m2qN(YeX>sq~GY=2bhM2cMjZi>KZrH#++sbQ50n#x1
zM!jhtZch^-T-GSH<_O#Etq-^J-ux(-f7N|^MoiPG7RHt@jlbr++OKhviB*d!D(slt
z@yw3pgTcM}+{42P^NHhQ17qVq*Cf}NB!Xv8Oc(Ww_*HlD=dHJ`=`B`onc=KCeNXGx
zgCHZ@qyg2+_DWv<v;0%qKi7<w@34G3MDCMp{OsN4*a20OQqpT}{LM)(WDaxbWr?)D
zuHwIxwftv1J0knG^3M3EqY?OvoE&YRlxDwCxNtaiI`gaZvrn_p2sk>O>#v(~^`ObN
zIl<0a|0*x-*#-wO{>~R5<G)*)m|({i+R=%2$(a$ty04rE0w}~;lz;V2gJ|1e*u8#x
zkcu4Ea}D`^9$wzU`7TGEfMHnEez=QHpB2q7pp!n88knUydlKAV&tQJ2%9o);wVDvD
z$0-%gYqev?4v3MSz_V?N$={+FXbDc!ZVcz@451~~%gRDZCi#zrgmbItmY^q8UmJ;q
zjqUQ=GOki^dVUNW%saagKx*u-eTPuIV$U^j-dYuto$VNIX@*Do?6LJabjh)9hxDWI
z4+B&?3tctMFQIAYS&7qq67ep#`g<_HqqrJnS9P@+!U#Yz+F;e`xEtYah6K^~l_>+!
z2hYaf3|dBbR;r}^K}6k}ZeW87vaw?Vd*esl0#d76B61G)_Ug#Apf3n;e1+JWl7aDc
zmA0Q?SCM8VFH_+gyB?7cT21-#d9dR+cmFspL8~Sc1+Ba(ul*mL%3V!MeYAG6r|gp9
zIjYkqD><n}vEe!TLgLP3=-9YQ^$|WJXBCOb!dS;B!@2VxPso(eiI`D0c0}~Y`OYdf
zbUSFr4_xcjeA>K}uJEOrqF}iqPjCGFIR*vS?s{+L>a~G+Gl~TL(7fq3hYxx|WMs|@
zl53Kwn-yu)-52NtEu!xV>@(p{$WNb(lZ)T=B0}JzOy}jpi&7D_(L%G`Lu29VpXQ^-
zH`@m+TP;rYTl+O1w$W7xR#Kkk09bg0Du_vbx}+bnkdY=S?hEcw$RwoSoxd6q1>gsy
zu8$^D0cfeAv%27;sByux*KkAy?Oh~NKDP9tu&N4V(|s{4!{aH**Nc@kgy4iG-3x0L
zmI6F$fd0_P%G&Db+;wQe0J=_MrLad}i(lm9<2!ls>YrLbr93N7mG4G2{&*G7oE{;N
zK{>j<@Qdu1OOzZ8sfoh|Yno`j9bkmp8$JB0apvTPm`h>JqLN42Q76y<Ls8J<WA<*B
zrS+>I2=$1I@PAX8*9gWrj+o=8;qV&)Y@2oUr$gbC5C|Uvw*#F9-~jO3of&JN2jLrP
z!hmrTLOW4e47-Q2a?gHbp8IQ}TE{?^PmF*Z|L(Cdcm{zYJh`&sq+3RG1j>Vyk$4Bp
zw+*e$#y1dW1+vyduFH#yP@{E+R>~LhQt{c<yrN=1P58A@S&v?{JMXyR7C{oYi@Mxy
z<5QX{kE-sO|H0l{M`hWr-=a1uDi#<B7=V;XNeF^s&|MNL2vQ>5pdcopNC-$M(xuV@
zA_ho<Gzcgq-66f_jo(_oz0cWW?=#LB=Z`baVti})eSq)#yw7vr*L_{{nsZ)r-CgIm
z$DXGMbb~Ca%JNHei}2?7?NNV{Z_*#7@6g(pVev&;*hYAKS_!%s-Esc%;CeGHz5sI{
z?>bg({qZMdLN|qis2=idy+Bb|VeicIvTMy@io-=4kHn|V8#KtItuowGldyN!;Q%|j
z?!>{obc0(P@_PPc&PW<?Ua$TsMJ>qw{N07eY)UhkKx7yMtV=PC2skc$C3G=-2&oKC
zUz|LBofAv-UvL88PG@Wnw#WbD%z$IkDI_mw454?<1#>)XIy@wKo;2i1_=}fEY&rI`
zvKpW+c$9~3F|$nQJiMC#9<m4v3*+~ZK9a1G&s-N}$1!v6ZvJ1mm2fzWy>{yYdM>c8
zM`jvu<TCoLR$}fTK6%jcCVg3azL<bPh>O7O6u<lcyD<5~Y~SDh5wK2}TR>J5J^upJ
z3*?YY;lCvY{Cd#{f2Azno}AXSiC(Y)(^k`TQC`LUauq@40x&lfp5P0-tQ;IK{+i!R
zwCN(!;4N!35A*TO<W9F%a~wN*)EOt_#7w?+FNDrdpFVw`XQ8f$<OyZaljqMt3DT04
z9l>S+%t@?AhjXX3P-G#`PUPXsh{L8jn?ka3A3va95i*qiH>z_W7{^!bMx2xbKYwmz
zU|0e}3^iXvTU#4i+E|0c3G=W_ouiZiC;UDD$FeO6067ECDzM>dS$*;~y;OM_tCzWG
zal5VKDb{_5ZOAgr4bqA{dYXI!`pj#nX5l@0MLXZkj<^uHU2jkvku6yM*9F^SW4Y<v
zn}RFi&RG@L&%{lubekw#mG8fFDA*I9y+Pnl?!|{0+=q6X?N?})9+b<tKg3Ucnmvj3
zP5EcB+48RKeBbu@iyfhD%(*)r9;eo&7ktln%zFH2Rvfo)oZ}CKZ+68P^0AOFqgW`0
zBnCd{fZ9>^D&!hyhgH-QtApBFZXO;Ue*X7xT*7=uJbip@VsDFG?6>VVQVO(pO)2T5
zp&7yr2MNcQ$>;OF^Thfu7^?^~$xa5AR#lndk)tNQfjS$+PT=CXIXTcpqv<gQm1i4_
zCZT2|&6LZWqjv>PokA$!g~@=BkTdj%Xdj>-V{T<NIW-0FdK3UP##J=nFNt+QF|}QI
zy_91e9FgPZMT9VC6%t}}zxzSoA(e%O2Bs7f{=|+Q<Jd!NL%Z|v;DN(h*%Vn~(5Ed9
zOZAsk7KVTaG_4{<10|7LO<)o4u5Ut-h#K4E%RN}4p*YD7^#jKyQ6>#=IMiJ6)Hv*A
zmkfk_@!tIAuU8I_ecHKe7yO-x`)#ub>_o&;h=HMxpPQ>I7iIPL?=?|oZ_fpHGue&1
z!Z;OFv*uKT6!ZW<#=z!g6c%3k_4z(Q;vs2-f8`uffq3bM`wf5x#F<{lGw|Ezro&QJ
zKTPOpJx&Z$7bnxhiMw-U8FsR7bW&z!gJ@fw<U9_;{kAMl|5{nM+THHI+ef@Z2E>!?
zb5l?27XNTejLUeGV9<GmneHC1-|pdrSZz1w4@2F7HQy^&Kh3F@RZwb+<2<OhyFM~Q
zJx}Rcv5bP$QjkQ1|K5xX66{Jdr;u5rq=FdGPzYmXS-ML1gXJ+hpO68G5h0t5A&+pG
zYSe#}*JYkHW||HA4ZE)lfQ~_@^GP>EvjHmtLb%G7pofRi2xVbrjzaR?)HFRcMKFJJ
z`Rz`w!U~~jsOMSMaiil%?4_HA;u$#gJj}O@p>F~tIr@uMU_1T{B1t}KK%~sEQY*NC
z{UjuCoNc}&EsZI>S$f3E1`*E!mb2)0fY`&Ko_f)|&mV@eyHmF|AUMGP_DOKC17zQ%
zOPnGK7D&_9bo7}>wHm|?dF&fx<`7f@VF0#sL@P&iTqpz$M<9s3ZKt1t-ExnLA&IYC
z^k@J&?1pvI<Ig_SYFy6h`$;E(q{rv%-8vX`%-M7R24Ss}tycbp{&3uCMtC%X0YX7%
zA=sbn>@@i`4`)7>`BnQ8o};KmXKWJH3pOdmV6x|!>0@Ul&b9E8b?ZDj02Q9f@}Wh4
zHUF&IvYYHwZ`k1_$u@ypDa+tl_P)H-?nqwi4`wXAPYl=Ey_}c*{?hJ#cW%yTS6%#N
z-r1b0@sd!}+3VGzZPOaJZ1Y<@vV%;v7`?LcQrux8l$SMU(?YG;VtMnK-dbLLUK8Gh
z4X2*CzH(KHR<Y7(e%t!UK2aqj)ar_I6X(N~o6iPs;=Z=YK&;}?4s~%!eY1tp$qbvk
z7IUSN_a*)39e2E)X!xG9YSP_9z}<3VGX2dS`n4=L;D8SFsAh}vQCn(hwL*<%#jAue
z3}!tB4wk6Wi=XQubw1CbVREGSXAycF6LqDwpVCGrCKMOhF1Ozmi4rs-xi2i}+lldO
zf9Mrk@Zlc4r3DoMv74iJzohwiw^Z78xLTX!o#+GfcdNi-X+{jVk682k5I-n(qT|I>
zn>KlPd*i4zhQuFR6G=02jx3BQEOY-whBYj$;eswrHX*5}yIUDuaWZyCU*;ioW$jRa
z8Wvdh$tkQynA=W(<&W(SF(taeqxlpJgs3%s;dq(SN8<}3BKfGGY6`yIpk<EWDUbPL
zUTe)!71e0a+7up*lo`7}e#<LGkgl9zhpaSwG!T;oy}0g*Fdi1FcZ>`WiaR9fdD%;L
z(kQq)S2+FlPZ5SSRMxu$pR*G=rmElbpD#<7ZLH^fZ}GnTJ$>4Same<`%o!~f8RsC!
zk=guu9j&|oGa2S**SoJL31+>k?Nd%J%vtQv7GAU)u7F(g#r%s;?Vn0>N=v+ty4Xx*
ze-NHh(tN0NAR$hRGi{F{a@UkCrSF66m8xnRDfAhneJ&p4-Dvn)`QX5rsD+azjc>1K
zd^hdd)g7%HZ7h^$*W>>(C_ar}fS<}x@Zypl_4)gr);7vZ2RhaJ(wv2-nk}x&^rh5b
zQDyJ3Q3=yJEu@#zEt+JSW?xcydZTRIm>Gbes?nIf;%TN2cDTWkg$H{u*QNLlJ%HBt
z(Zgx1e(Xz2J`BJfavW3+uD!XHGxQdnudp8mE+T+lz95D8fii*B#mosgf~5^BN^iu`
zM|s@BJ|N%4?Atm4Rzd&GxN|`=WSjxlAw@UIP(d|2CKe9(n@`qJJK+?#kV4vXalCI>
zuqfrs=}w=Heo1m2VtS*@VmAoRhs<Q2&*!V?EF{l3kw$)TzLYp|U-<U$3IYxcndfo(
zgbKR|9XTQecnPk?K!-uKd6p37?;rJED16yE%s(M^Q0(RVfdIDJ%k*Sr1di>n$Ibw1
z%RQX0hN7>}<_|iPNWf@wcX9h@KRn`;;&J3#>HM(IctNC<Vg_1($Iph9dM+=GlLD~W
zLn?s5FJ<f8?w_PxJ_0IJ=JkgMoG)-*ew{iq-%Wj3S+J}&%fBim^^Z$??Z$;43$8g#
z1=oVxXFK|}!<3gS8=S=2kHo)H5lh>Cf~a@3QiSUnw6&)T`OAvW|46Jm@Y+r9S88rG
zS6F_>(mrAPvrbmAyW(?ixTSpai*tODeqz-|SoBW6=NNtV4oKa5_Ph6&rdMr$v+hZj
zB?_L=LQo(f(cZdahfbx*k|B2Zr16C;o-js*Q10uxpb7qfPJX*X<feUArQSv~acx(q
z886)0{RMMXF>KS&LG8_zJBvrqEtatgRc9)yRJ)uN%GrVRGro9FnjhC{o5m$AJyLPM
z_yCc}xyj&AwpydJFW<b8Z5|HmCNmk4UBbP_7_SDP^AYP^qz$OIfR+Oz^1`~i5xO!^
z5=?}3wM&NC$5S^~sFbdYyfn#vpQ(vUpy*jdCD2V`xCC>1;8O<<yhPp*WR;|ymt@Th
zM_NoBmh5r?-wg>z%pj%PHpYrLu{_&`^|d8uns@QEC|3A*dw&_bs(fK2oXV9VkZv7S
z)G|<#@~-jcOXI&=I`wyPC`BMsvd2JYPt7~^l0Dlv>n1K++?vS0o3VZS;ls@6lg!sg
zgnseR)6>3gb)nts)3r&z56{Blk|hs5D@wf{!r`?4ZGa9_iH&A!q}m_R5z2Kfb}IQP
zq8D$^9{idS_Cwn7Ig8rTKyHd5lk^tpW6IpNi`OKoXC1;%$H|J+@7JxhIZ;MiRV#=j
z`aOG0dA`}wqSI;Ej-q0ye<eow^<*dVBNS=)BFGq+x4_5&FKnJ|I9|UY4@fa;eh6=;
zk-!T&&z}O6*sPyERQbbKe-PQG+`OEuY<QDJmtbs=3yZ3Ksc-iiukF#rb%pQ{0UVaZ
z&uwLCIRV<r#EY6sW4|*&^)t+7WL({dQ+gK-P2c*5JAqOrs`U&W89foP_q)N-U?!L=
z`}k}MdZARBu#ZXF=rmj}K*O=<m7i3*Nc;f69Na#(6eiiUZiON3j_5><wB^&fX)oa*
zqGh;h)vBFKn>TMp^FSY;iA(=}Ba#mDUnu}dkuJy5e}nzBxTEQb2b~QswhQ}--@$m`
zuC{`f;PCf;V}WU`$ny8wK55h+xpIx1*2md{H`5niSw6TggZBU8v`bc1$?)Agg~n*}
zhupQ#7nBo(hcEq54m%R?=T7z4OYLXXKHg!Bx)@z>^L5LUUB#Lqi@$Zgeg68nu;s$^
z5r;U}TsQ8b$5~_l;sQ{{&nd~+^Ioc;0p7~pW;m-t6QS6`5N)z*l@!y_^9t?yxBEDF
z&R2-+4~0nwsILN+e})r7fm0!?2|-gHSn0z2{Hs^5NH80WX5)?u8^Pv4Ezf1#K;j_M
zJkl@29Qn6R_qvC3<0S3s6c2TXQj0CU+$Z*`F3QZtu198^v43Wu2W;Y0(86kcK&C-z
z)|+LAQLlP0y^>XN*w=Fa;+GGR5f3I$lE(i}?k5h5TnYLprJT_MTA!{s8&weTLE=18
zy|^xNurvOH*Yas;<cR*9AuBC1P}H;rb&8t$gUIL~K3v+&3>A#P%^#;YCnuIJNn0jr
ze15xA!RhhwoBLk%X-dDh;I9)M370W~*~i=_ddX$m(NAhjwk&STmasQ99UP@>+{H9$
zoPG@I95PoW#Y=u^p#NOpcHxe$U4A)|k&-ocZb%D9%PMLx2HJ4mplet`#W}x~R<^v`
zx$FqiVotVEo8xnuTJJIa4{br*EF?7Yl=Z;M(h(I@w4JySgUns2#)(KDFgAGqyHMkz
z&oUarS`j29F|@ZSO6^8#l4;WZ4?v%b9~$pWH*PeaN|+7FRX%Lz;`~t9BiMZ?`PM=A
zyLT_1&oFG;$5>zfOye9#lSDS1QrXs~ZS^F;y@@Ls7|TU}39pCGILN5`v0Sl1>Rhxw
zC|HA56Ax0fM184!6>^N^IljgeyjpExe7Da&==`QyOhbY#gzWs_&!79V_@isL!RE22
zNZO6U;o)Dalu~;noy)c<*6f@z_3>7>I`YqIRdWA|y^^(dJ743cgOl45f{X{x_<o@f
z3rz}jomM?KX7*itpLhiSRKBlR0wtw{^9D6h^bBn<S^|YT>`)w$Y6EOnT$q^#GLQTA
zJvTJtA=<7av!37Z78?#eOL7_-KM-+1qX`NKh@QWsW*)cg5)#4UJBpx}9OJng>9V#~
zLyDD~TWMN_a={7pRNT@L<Wz_}lZ~=thmGj+*d#)kC@3gOJWp|!<Z@b1m#LjR`*M)_
zHGlWT3(EjNYMB@lU*2=wm+d?>oHh~yMZ=6f{ZW1=G4Q204>Xs}nwk^v&M2mJB&<pP
z*;-luxxZD{WWHO~@@i6EZdc`RFE$>zBDYrTeYa>xrJ<|duinmT9UBjmw{c&`OTwts
z_P+YF>)CFdw?aX8-mD%yc9^c=6#psl){OyA^*=Ug7*vRV;Cz2>os`i2^RYp~<sPKa
zbtuj2yMMpip0{}Jm&E)p#jwjmM+#e7PNQ;wQ3dlHtW^Z}$Hf5%zP`(Bwa7LpP7O2o
z^zNFb=2he<%XQ>PrRPrLC^}+x$w^#3mBU72$1cU=u|LG6%r2F1%Qdc9y;AM>bm4r^
zFxPbX@zNhzoFC^V>u+`c;M5;lG0hp~Kyv_45i_$M7<+xe&1`<e_6NXoK+ape5Ac*T
zDzK7)pDYK5WSf*khyKL#@mQt!sU`i_nCW?4PcJA>-&Ny_5fua&UkeNSjp1^Nr#%hJ
zDF@$tB=MebIW^yV5h=91ypJviyWMyWNQC}yIc@xmU4+4$rJ~6<<j}44c@Zs})|ba4
zXb5T&Fb%6djII)SCsnjkJUQo8c5y6IlP}Rjfx~La^KHk(Kpv|aCbeC?Vk6dgDJguS
z{qZrjx|K$;IJS`S*9vGxMMVu^zQ~ABFh8cWaeKpgg*E94rHU583=e77eCI&2y^nAG
z8mdPr%A`TPArcm|{sDqo+mpGc9KH`H4>>F4#EM?$tsdT{9^_;CTF&phPd0l_=crG4
zyP4~L-EO_8)5k=Gq|JYOh7B(X-mJ(xX!X7BxR7)c@b*nAr{HYc{q|4L%dDsJJ@&bX
z9%92GVn8)4hg%m=T40kX1|W&qX-_X1)kt5`*of|%-#StD4w<KXVd^-cYN)VZQZv{x
z-lHA%i1I6OIfYn>|7z-qj5*whKG=O7g%#U<-38KZqXaf$sJKf?K2e=})bb(VGW3dR
z?2SQj1F-NGXjO^jp!Pb9_9=~?0azsA-=9+m($;8PADUD&$7sVsLXPkyavVP{8~tUw
zN%2jzyV^gYPe-f9!k1ED{w2&7BnB=CqX1(<z!TiEc{8{3lX?(6xgwGV1_$vXN0W4x
z@9>?U`8=vXWsjI|T~FVs(~-aWsT-4T_s@Jb$6|>extSv)uBYyI4fQRGTLm|#DV#0h
zJ^AL>nclH=+q>R+>nMqCJHO+w9%JjJs+J?sQl30YB}#q?$z}QFR~4(}OjDokK77Nf
zv*R^}1b_ZGy~a;`5+F|u16X-lISbG9Ms5L*0=@TxHHemn4!*WoTUV)NuN*kT3P;OI
z!8$F$%UVGq^$0M$IoJAeY<e13zmQ2JdERI|fJNRhYHHbNX}ZxDQNRL=Yn_3)7I@Ja
zeY&$SCoyn87fgfEVM#dA>0WF}))wBORd0)CawRKhBLKHfzIh{o@kr>^K}#rvZd%b=
zDOk3AiNC`I%&}9N9;S&n>d@%s+^R|lD_E}WV}RaCdoKG6bSd*mAQ58=L@Q)sg4pQK
zr+6)bLq%8Op*Ziw^g+g>^H7CZ7Ho}Gj04{P$Vh!wMee@dHZD)J`F8WYf1OFX2+UoU
z=${gq$hXzR;$~0R3Egi`fX+XF&8Z(8fINTZ_i@0Ac6k>>9wLq2yWjT-s=9a5M3&_Q
z_Zr{)d1E*;q2i><;pP@qJf!a*^yXG>wp^Oh9<wzNjo;8-9KFP^Su-UA##C=ZQNDrI
zQtdirclHidJ=>076oqseKaAc+2Y(6Lpj_DYCKb8e(y+@?_4EKf9d~^mlrp2t%1!a~
zq%PAIK(@j@aW8nPZos|uc!2v<s$of4PUQfF@AKy@eEwQQbYIYRi|4mOJfNC0XKv4$
zDJs?gSMzj~Tl>ItZND*GPNDkku*t4|&TZuXCkCY?>E)s`0>G1(cNz+BFb(P89R8<#
zM&elR>phZJ4}cG}(^LS113P~I&cGxX{07Xhe*gZBxwe{s#et#%NE#m<CCqP+UTK}q
z?(Tc;?f@k~Khi<i1pNV{e2FAfQWDLX5x@c{I2iBKU%c7Ejl96@))W-}UbMVJNUqU9
z#5xPXn?iicD5SiP1x;K#)GU&McTS+#K!Xmb_uoz<PHnUYVb29i(DwFM3t!*irv(WM
zXgo3wp*wRo%Joj*b2P_mSs^OAT-AE(trCRHm6ft(-Y8qq=5u*@As5P%AW9M7r%ZW!
zH>8_&p|=KG?X@(UiB;MfSBjYi+Y#BggA{juzGM`)Wqx|uw7o^Q>^8Tv(!559MoG5s
z@KCc;#q-mP9~O&kDr`6AJ@s2qF7G=PuN<lFHRTn0{q*YKHADN-QWBosRD7cHBxCN;
z+l(?Up0GBK)|yvklqO!e&c<Wx&lckTWbtM5HSpQ`PR9IJGsOaV>2$Ut$4t*ItZPzm
zbE5cFO&FZkV=Z70LLg}xGKLTaUXH{<Lt`Vx!L`vaG$&pL!HIWLSs5f$wC8D#U3`oC
ziN<Sk<;@%3XN7RT31w{JsO7|Pn=mxN2xi(l{hmEbx%L%fMBB5c&8tdZqlKyp!UTlC
zndLa8n36BA!@-6W45SfNx;7{#_#nxJ_-<C~Vx;^6c-+K%7WrOe%BDg*MQ)3CFr#{p
z9Ohu|+P6S=hnlC8zj6cay};RALqH|8)01H~bMZp%$R|oXZ4KDqNzdRwWisXKQw7Ro
zd(<rl0%jp#Q|N8TMhjw82Kps5P6j(-6ywmXeV`-cgiISGr)!mj^sqD@wv+&GrD#X|
z$Gtq3EqhL$GatI=-sG#Ps;Wx*%{*Ik%)7)9Q8qLDf!wISS-)&MDkfr&6!HljB$m3B
z@xf4BqwoUdd5y@ovNAe8bK-A~i(t*#Dg&+oFte$9mc)!O1NNpGFxFe+dS8C4AndAf
zQkthU!X`{aYYkO#_EXD4-0Vs@A|gw(3v=}WyMG&{>~uba6;a(R9oDLRO>>)xfQ}9G
z^`4f}3t@NlH(brTo7<u_tv;aH>Tq+Hjt$K_ZV&HgkY$Gh6#(fW7j*)_aS0h27u*LB
z99o^{Ou%L+yH^dk4!?Hi&j>UE5JqV&oWf24<pKaSn$gARHAzbgp~5F~YQ6Ra!XOwz
z)X%45wevpcO(%f@F@U`1#EGiTtiw1X=rNiB)HFo&Y@oo{?qojhv17vkVu<$|LK8d-
z_fS(uCjEewTFDWw<CZ<LWeekI*=8bML&xWd$<}yNadCcZcR&tEnV$Lk>&<)0atd=J
zohq~c1*SG4JY0ynfd7L;U7NSGC;+hmH#Zr*2K6sUcf`BbVWP(jZM3!4ReJ?(-<0##
z!qRzEsc|1pU0%V_3*Yf@zK(D6aLfXp7<}uBoD)=x8#zQ?34-K=G_C*gMO#dMgLs`q
z-VKQ~?|e%I4vBcQz|h?h3(5R7JiIho$Z$DE!Utub<)T3j)+2g=%I~O@F$FYfCjA+-
z<H(I5A3dh_jw}Z75^?o#a01Cw_tq`U7MtWQI_2OnhbRTtMN#dP^jnA`IM~Jinr6al
z%=;5L;XT&V`>KSqHJD$0e1coyq=IwK;oq_w>>7X4KAaP{X;(V-ssECijeW~Y+U<{}
zX|(b9({Z6h+>3X|{Gdvkt1!7uy|c2GbWv<ae&Q3Z6|u#XsZOapJRVGf(J;5#v>C-B
zVC7hw0I+`uUq{-VV7M{ywwM!xWg+G8<&G*?J!!lt&q6XXGVCSq&K6?J!%N>R4YW}w
zlqTnsR_QE8^I+Kaj~^OkqGAZ*j^ky#o^}S0<Iv|pZ^5cAn?z7x)?M}+AO?d3aHo~)
zZJ4<uV?e6&;re5_rLMIixI=$>BD{SAAMRi83T7fAv6G`C8WcM70Sq8RA{T`CtnHN(
zry4dXJ}*-iV|oy@Il8)?5VvqshSAnx4fXWg2zk19V=8MlPDOmcdM$^<%3Ri3bWTYN
zOVvKceM`tMHrM@_qvV%9sSV5m1gYadsll@o)dVtLMv%lm5f%_5L6dPu_eD@`^71eR
z(bx_D#Cmo)Asq5gw&lAi8P8kkMDG&FmH%8xL!p+XrZ6o-*CIK6&`4pY!y$HO^%L=*
zIVc>?UcPepI-}GdYf-(MQ3l6eZdV-MXLtC(p<k2!$;}(A-5Kzi_jgivn&jmyxadzN
zNnI72-y;zZyahomd^pJqgXIv=i6m$>tph~xC%`Lc<Bx#G1Qy-rTQ$`(Z(Rj?>KV%#
z04*%+e7oIuS!SH~SHW$C`_CPTrq`FA;e%e8zuDgHV(7Xb4eBAhA_~fr^Xl+6g^*In
zdf+)8^HX@6{$X##A>ZZn5Uw2z3=bD<Hd#O@1imG{`;~q85yn0*VEo++h+(Eqp7rk@
z7{~$p7A-CvYh9RCkoO1k5Gr~UG}k*Nb}xFKlsvuW^S7Sr=4Js##!fJ%+?|8rWTAo7
z0|x=b^P3d$77JJbc*>@d6hJk2ETkg}@JP`RQS50mMSce%3DMXOa;NWn^7N@$=eKQx
z+p(b_wr6Xkp#9a;vzvy-8pn<3vb8C(yv8bWXYN=-oB4MtgLC3`ntx)j>waG^5`<~s
zqv!K1jI%fU(tT6<I+b6^(Jr2?+_UWwkHx2o*F4I7YMSgU;5_^%JJx1W=@0i%{ZoPS
zt<qU0<?)@}zmG0n9gWY{K0O^iA||JhBG0ScFu*XhBsXz^HIUW4PKCasV*+D97JH+^
z0QA9biJq30cGs@oJw1OuQ63`#8>rzR0cJ}j#wnU$uwWn?8QoS}lAs(T`x06$#0`>#
z+RE*A($LgI+$IPJ9vA2by@+P;TZ<~=heOYp=@OZEiVi$jW@NN7T)*Kh0x{Sd7zu+3
za&k+~$$=1?)(yypC@CL%dEvmd#z@uZXnh1$M9>pJiHNspH1m2<oV{dJIAnI{N8&Z(
zWNHvg(INlY(NPWT879Y@aB>D-Q8$5N5L8Rha@&NH5pW6$3JjIMp9qVO@4wIBeCm4f
z&K+JvvxR&P4A?^GXaH=^U?G9uhl>X_^p=ecrl32cQo~X>tYi8(4*h5}5b{8E!0skF
zfOcM}kIxwFQ-CFd0tSXhI+8eiTwo(Lt-i97Iaz-D*DuXfxJV9>xh#kqq>GGGDddz{
zJJH0MwmOdd6aiTTb~lI`XntbFP8PE#lVQPl_kuPbt6R5V88B;<oNJBwiul}UrGqI9
z6I}qBcoJ#o#S#P6Q@6tP&IkLDA|eg>D;<a;#8DNlmY{56EQEbfC_G&t*Fzx*Q*R;s
zpd&B7<OZT4u$}6oy}*Mzm0w1lLyAQteJEGakb(aKdLa8@ObZ$|s*xnbPq%fOaVS>f
z;~#W7n=(s^jg@Pt&cCx9xgy=Uf2-i~63%yo-RK0z??#UYd$cv@4bMH>hDLUreA1(k
zD*?}++RhaXqFPvN2rD!nvq)&U5WzBaUV6#ueTBc!LjUyeujYuiHaZpWLbpt9xwPj}
z0E4zy=C*B4?CMP4QZzmX=9Ryg>6g^3c(187@cotjV}8GWMK*<4R|)qYu0LuLYAzUf
zJ3CA?A2Z9FcU=1Y!!hD~p8DgEJ69W!f3K@c0a3pT&w`_xy?_1Y&Cp0M#XG+U3Q~t%
zLH`s*0n6#ICjBrc9;ewn{{lzYz7nHRqseGXbu@O$&C7{^FCl2nA-|^T2C%|HCTn!A
zT?0y(aEliYWU^df_yHN;OuuVa^euhy+utqruw4?bhz5s*QhZQJ=6vMLvf(J0|3xd)
zZ%VbqeTeJf(D|r$I&W#dYW4!ZI|3j-#>9R&`#gE$PAZ%%-vigAg1MNQ%0(`N-+4ZB
z${S)Xq;jds`j3*o+wDE3`Tq3|nYp<+b_SW%EK2L?jYWEM`ikOfmVG#NVb->j3W1ry
zapaEmrN129gxWB976E}dBr$bI)i5L#$11!&uf^4b7NGam>x*Tk8hLnyI9eU?>}e{e
z_xN=$yO$bwcHF+!fqq=Q+RP~~E-sK|;azG1`<_UrbSSo8b8c8)z!DNY3T6QqX_ZI!
zK*R>&V`44Cib#rlH!HvL_|P2jK|V&?72<+pW@p~QsVO|Ay96mT0sZkAQ>(P1zJS#t
zD5z1C+bZsXPk2^T6s0zoFbd|WfBB8l6TeaZWcx9;v}MoG(FSd57HPzwYgedfRDG>a
zk#W{wyQshA{HoF`=t<052n!9yNGy@iuf$ktO?E07RvgG*E!$39)glOYp>9a@(no_e
zD}3oSCQFImsQd}y13|uIx6EWNjYG;1w<w3zVq`;c{+Gh)6{_0;Uba7wf$6U*-LeN7
ze)S9=(L64Hb^Y5t)c4;4FuW2mOT*WHe(`^Ed%Tz#$N&&yOuH#a!VQyv2Iwh=%*Bf%
z<KqgMoREr5f!uM)U%#c){!ROxPa!LfpRuVZ)tr^AD8z^>d_8x>y?#Rn>MK(9>@)|s
zC_M6X9dLF1vdU<B6-yHK+47@U!SZiw*zo@{g)=8=_vNqRf1AIu6iP>a&pzCb^9M7R
zd<XF8S8PJ69T*lCW?cj>AIdI+#AP2k&;o0#s}<T(yw-VfC`6SKBKe)8c*T$Bz*FaE
z!eO&tZ#8`B&i~)IjQ`8m`=4Jl`}+)U|4$$1e{RqJliOpKY7JM26(h$8LSJ59H*Jg*
zPPz#39CWQK;Ue1FZBVroRYQ}x;(3{Lq`X=-@#*^`3}FZefB_nbm~;-EMjnha%(>U&
zN1V$c_hUNLSK<YA@P)hhKe&-1d9<>z35ZXFOs=Gs_>Ss|!1Hg#$8_+|U>d3DANAtJ
z3-B8cNGRZkPyRZ21|O;qPe+o66kp9h3g6$q>#r#%;m!j_?9MneEN@AEf9-N`4(|B*
z^EiAqIB8igrw;(4yxvPLTld`ZThRmxUndh;kSRTjL=;-_1Y^<g$K@KFIQS>1dX>Qc
zU`qO;w(Unti^F6ela4-q?_Y0j2rdINg8<S2$ddL5AQG&|`s=YN2cxDNtvOi+g(uZ;
z0DaW^Yy7Dh6-inp0DBo7mS2c+`BS7DK{YrvCAOsjcd-ja(nAt{J~}QRL&o(8JH$Hw
z75!IT*vG86Ogs`{?@G;`rHmvOiHlkc`4Gn${(f|f_Fj*?wxb{7TeYPB{wLj#N&r0R
z=!Y}Jck^XK)S2Z7*}&fsosfQ`3T`bGua>_b`1kwo(TD)Dr(^i;89v?Z!N1nA{U~#h
zxXRw(M!5t&E^8%msD186oLA1AxKjEt1Jw!2=d*W~-A(r`e~w3YP$1ab+U`ywIMDTF
zM^cOByT!8ZMMwy23^ys>GB@7~_t|R;h_SbemM>u^)xGEhxb<bjgy_+-?JD!~MaT+*
z*^-E)d^YsDK5onHcGu;%AH}E>VUnOlV`i!V#w)llJFiy%eJ7rY&91b9wzrT*?p(Gj
zm0Uj7!y9gZuZ(}bc1bEjRLpXIK#|QaZSM9bP}*;SAzL?u@4oo}3H`*xK<$*pzqkOF
zq+)b{H3md?MpndK@By%XNTfbC<(~PqFR^9iZYD6kQCR(b1+lej*8&`SN4z6%Pl6cp
zxk>Z3+<#7we_Y-FcSraC`?212|9TWv3VRM<nh<Iy<iTWG3Mwy{guv9nB;_|zPG;r<
z)svvlbXjt)wukAE>_I+t;sUzj>%;ivFIeiJlTm}sCbAW|FJOi_pqdN7{OX2?gZZPC
z1tc?h{uRt^l(JxZIKay^v1J#2wsZCJe$o920D_)g2y5BM2IgNY(!Iz6zby!07ChI>
zFGbh>{XP$-2;lq1Ik~y;{V`m&4*eLl9B_^^!hCax0R;%a#Ub!|hT8IylbvBp;Wt1}
zE`x50Xy}iFTn=ah{oeP~+qRjZe1}5xasQv$*}aWHz;OPujUX~c2c#G9ZhN~AVl*7S
zP|u?P2ZV?5#%gUO)k?q=0tA7l=sUZBVTsy#yD(%oN<F^-OU$BP6Ts$2bP9<^9K;t&
zZ;<{oF-W08XwDeab=@w!ium?!5`Uk!!`CH%13`)RmM=g&u+GVZEf{b-d}hXv{Y|9F
zTJRM^y&Gji&Jvz`%lCp6_lU6s!t{jUtq4)j2!e(Bsp{=|i=D8WH;17lcI2+C$M_p>
zE-l|}$*LH+Lq%hN0D1*<(Ut+g!JOj{-6`4&wHS=L!wdK6ug%K}SzYv(9PsIw!U$<y
z*AC!5Q+@D`HCM4#{xR4A#TD2e(2{WZSj#@&BbTkQti<-JlEb$7?<u5uz1Ww)kYL;6
zXZ^o^9Y}hE!kV~9Y24J;pT=U5Naus}0jjq7iE0_7oM5`;0h)mv7_hwQ1EV^)KN7>3
z54boKFd3?N1e+Q&Ov9dXgHw9~(}RnmfxuURJ7zY6Ck#Ux#-o^>NsAy04w>`}pf%#P
zdT2RDHh=j;PQg9<_$_0=%0s{9sbe~@44*E*2Q~FSq3HbLj)Tn>2)R_>e*CKO-_y1l
zcC#22fz4v`@%GADAGG1g1QZR7t=I_#@$eIX$)Ep_V_VkaI4Q_JTC(}E0o=!FbT1x5
z$sbXIH2!6dXJ;ImH)5yBYUgnuC3#Jh#jRVrp%jNsXL2xvZM`eBc)`KJ8p$wpV^vX@
zeg1?=F;fJO=Gxy~kOk8uC}wfnbc^)aF%+}INC`PZK^-Fo8D8{<uywakCFX9v%c&f7
z!|O+GTs}0<f`x`PgLXN}|5p4DE2=ks1~?mTv5dwdprSo<9i5$wIz~Rg!x&2dU0wNe
z#)*1@<UF$c%e#%`g?J>n+vr`v*x)#tUPb(nTp&ZC4gfp4pVmw-SYAPasRSw*o&VEc
z4gz0mxFuU<EEahVM!!jEI8P60n*D-OZQxN7W}KePV|a=loP<9N6<R;aq4+=*^~|vk
zY1m0iSQ&@E-VYNZtWJz1vDVXz>QsQLV*9T1Vqy;|0ps_;Gtr`=bEH~E_!4=E57_Oq
zUZ66VTU#HMU%nSR$i29I5ZGN6vLMDV`2FeNw6ffYeBoe{fvM?ItHbb~nMVLH$Rxv`
zZd8wIWU65oEN^~65&H}vEm1AaD*|L>+VRwNKWv*1LDbTH6q|Xc{@;zZZSUSVQ>9F_
zlKOlh`W3`XXAKL$QqcU~(~d#Q37{X0mxh3{VSw&6b<@Y4z%cn?9MY+PjhtgU+73{Y
z3?N0)1xG6M=!qj0&}TSY_}uT2(7>DdYO(F4K-7LWWs3Tc2`!@v#^g_g>umGwaG2w(
z;U*X(*@P4mhJeZpiaxaV2&Ra<U>_YF7UNzXV`2vfCNSzmqk1yGbo4ZDE7%EEO)!#3
z3^a*Dhznp<D43VQP2U{>+U+}y{Hq3?9giqTQl}cUVI~tYP<S+6q(>k7P|^}sQ_!C6
zEdJg*)*RsCswvfjdzSt5kV*M{T=Ap3jv8k9jwHd`!Nkn$jGmW|jMdN&wVbLxTp!x0
zCNs`OI$#GP$e)aMw}Gt$NXzDvc=zxx%TNBwd*F(?*RNj^V$u>z3J<>m5+4FE<b`h?
z>r4H@!T<>VeDdT8`rxnVgMEBX;y=+nQhlX^H~sn-{-{RTjT<2{77Lu*a<EA$S~gNX
z5mtyGN1%5FQuz4~#b6WTdH|-~=$O{=69bb4Jnc#>8Sq_b1J9!G4m8;S6~e^edU7ZT
z_|@o5z2PSPwN0CreY07Lz%%OZ?q)tpwSIjA%7;6vb8~ZnXtHM+iAYL%oMMXp1;bRB
zCn2cdTxx1=Ziajcj6Y(B2tS6{SX@CaiU~lQfUm4zh_u?RW@7S(EzVLSSbqHChYNEr
zwi4hgbPjt52Q}eg0^}7o4KUgT3>pnxgD^CcBqa_NoSC1oupmjPWF2TdSTFDqXW2KC
z*?xzdDEK<=@$UzO|N0N~^S^<9mNz@OZ~wo7(7ft~R+Oe;LZFci7XHpM&~Cq1SAXpS
zE^LTVrembAN#{6n!~{7ODuT{)LP-tFb8T91g3EFdL0@ShT*(~JaVVa~)^2zQLDUm(
z_N)J?5&mw&6{JQu^LFRny@u1?X!xpU3q*qA<I$3bj!JP;$ST<fIDd_n8sBGS0b5kd
zR=92cU?1^2V)0r}e?^Wwp%pt1qtSw6GTr)NM#ePKsjG>L;i|S%18NA@=gysj)eh_&
z(5GU<ygASt(eB-}?-a?-5N^_My3vdw^pe)~w{L*GQ7D|?<7@o-H4-%{6ni(pl|*|A
zM@WpSek*j==zb#yj&(tn3rm(Mgf!~@+nzX{nn+n65y2p6`;Uwk=MT<|8gys~2W@T&
zpD8YFZv9S`HB`q}LYB5#T!{LS8;DkF{X0M_dDNW-BO$%s0Jo>O(3dkBNG}pwLgaPP
zy61OsgZ!^j#SUeyk4rkT3mMFQ8ezi`wu|8_g$WrMz}D2Iuyu{dAIQ!(^9F~7<w1b)
zdF4&88A<#9cSRU&wM-*z>JuEto~Y>2h8WpkTpL%5{?uhzS=rYG=r;kc*xz^syr(;>
zwb4rrnFaNUd)PBbdaOcY>s{P0wja-csY2INHF6lgd>o?;;g!VxgZJBmZ~2eFS739c
zu-hiMdkV6^Jwlv&p+<cxP9>*Us<PH&44}W^I<~g=RweKrvg+*YL~G+lfG?&KL(eK^
z0{=nHXgc4EI1+GUUqAtjhqg(vq_kAy%@q*<1aNeuzF8>>)0@ZW2D*Suz~)zci_G7p
zqUC{d%&34j)Zd>6hBXkn@8IWwjwHkfRY%)hYjF<{eQ-=<7}XD<({eOb4~!lhH2@>x
zCEM<{yL25HTDY&Vo?e!54b>Vdmeas1ZodVq)#|3CC;F;jtub>^qlNstoGx&7?PcG<
zK&0kv$TW7V3G?uzquv6k;}Q3b17zj`MtfRYr@J?SbUlk^r4*c0$GdR^R~_YZp0|A&
z8{@-<l63$6pD7hW)xNNXJ@&WpjUifKuR(i*#^e{+W`R~$Lk*HLM7d=2ExXko7JsaJ
zL01aOG?|gVwTFkitn7X*dpp!YQEC2QVO_~zl_IdVISpP$u;`Dycpv$W6JQTq+=eix
zsm51gk2&gzBjbmeZdwcVtxIK4r3O`B8RbTK^5DVxNgWuT!|4Y!4y%0EwMYJ17pA0G
zpy)$dK=>rs76cL0OaYPN)G5^>IrB4b&yU<bc)<1z=xz-h$eFJA;-x}1TtFIW3hb<C
zGg_(RSZbWv+Hb_L7|D?1Oh>rT<A{eB<9ifqaPK*+b%2bbZblNZYP0$$6z0Tm%ghe7
zB#?aa88P_-<tk{0^H}H50~jJDc-pjtEV6Xz-f?6*NHvWDu<d<6jWQ!wsN%RV2(w+h
zJd9s3zUY!NW(74ts6-5hSOY=80RzUbheft@o8U*=HB?Pl0&J=dD+PADFYHUW32I`{
z_y|@ELQ_Jq@mi8&&h*b=Qu}6xSEiTrJdyf{(|_|REv{DcX++xlq7Ivwd2tOWq^}iB
zu=&~zvG3)>MI?Ia5`+KLLc_cR9m(M*D8oSYX~(K{=Lro9OXj@%<?C0mdGmV#y2>zB
ztKl=m%@=csH-FRayanYVls-Z;ojY{AFl3?!6P&wQ){3sWoIOV4mpB>rEuRg0kH?i_
z9DC1uVV^jCL1DftY^6_OgCeO`M9GX=6C^kEykBj2HmJl_i+g}|g<?V#C6<lNG%_Vv
z>MR}k&K8;uHNYg+ChbYx2g%8%D4|0ziZIv?3<GBn$WXZ+lQ+@}9l>!1XKj0sn9=1&
zP&%d%C5gijCi%R`#gO~cOePkM#ezbM;0p0@4*Gg1w=E!k$JmnZuZ!>yKm${%IU0Bf
zEq}1_qi=nCazKyA0Sg0k9ZuS+QIt4GG%_L5XoLM1P#aKytE*G+t{AhS@xjr~*f=3@
z@Gvv;QLIzMX0%L?;3}|I!NShGUP4nQJ>rh8@DFfOpm`exT#H#LrvwBX?CfxE0K#)Z
zGm<!_2cAxS59p(e>kMH(!TX_q=ZEof0jT)--LYWnKWw}9aNUt#SYN{PBW}_!laq%*
zhMQmtT_bV>I(wj+5CUy9altdk84|>RkYF;#a0M({=^|9yitdT@nm#bFlN-N3h4v<h
z3MJ1YP^Lol#nq-u>TS&NmawWaW;$iQRd+KPts9rQhAN$b&;r%hV}!D-x#}_M2M6CQ
zjGFQDYI2@DS>UlJBDn%VsI;Vn^Ia@_-htJMr+2{ALce_9YN^fim{FMmCDFaMZY)%;
zGA*d==U4#(A6oR;*p{U+Kqh_U1SQ|UUx=w07Ut#~PfK*axaB8q1;!khfuDmPV>s{x
z3ies`C!-40u$*Jia-O-qTC?wa<l}3HwjfBq9>-r*IW`d^YHCL5S_wPdQRl&^mUco_
z@eEvr(XlQV&airL0{*OyHE2fE4#$9F4`&=$DB-T%bFjq%!VB#lA_GDJxdI;2?CJFj
z@W{vnt>?=yF)#J;JB$JGNrI-Wf{%G_Y(O`Dd)^T0ro$2)pgzS<DHh|r#_7lougUpG
z1wYy`rG#cFTfk*8heMM)l)A7m7sp0WfPzcN@nwN%;W`S6$Ly+I#h2l&<hO^A>%xsY
z!`ZgvyA590f@2I?v;r2p{M8*DWH*U8BQm4ZsN#@31g_2^Ae-IQit6f+(6vmoM3<G6
zT!t?T#&#Gswbs^A@7>D}cFSj*aR3Z3+KqKmx!@Cc7$oTrD0SrI<k;M7Y{}3~aAnXK
z-96;7zdCvff&owmXtfTYW{g6@fK|6kIL&#%KHKHnJH|n*HcN5K^Lextt&7@Awxp-C
z|JTje2?+_H>Opon0COC7Wv@q%FwuJ)zGs_U+OGkghaB)cK@d_9X)!)%90ea!SXoh#
z9VivE5bO{A;QO2m{!BITDg?eUIyx%W7n%bw0mDqwKpS+hwT-#3u?Ev|u{Q}z7bG2n
zKMl*+RzsW;JMo+$&p?be^)8CMK3`i0hhX7N=eFMU@bG{SDCmDuML0Peot&g@z9DJV
zUheO#JC{5I6bTy?e5D$1pgUX`JA9BXX`zKU^_;YpJWBRKCnfd)W7rX;<)^ZWi23(m
zBXB(NcALkH4Hvi}!Vnh>=6=-GC?w%$iRM=bxapXh@V*>ZPYfbNgE6ftgRpb~Y{1Hm
zkkH`&CM_dF#y!BQ|D}N;;64d|j<-MO-rk0fAVQq>Lc`auOs7u$LFACcC?n8OWn}UY
zLb1olm{%1Q6&>|w{dbX49OdGgz!~8MyZ`cXX$cAVX~NIqI#J0`0}wrdgH?16WXN><
zZDO5BXDz_Thvf1o&UUb@0il8pis)w%A}uby7bC4f6MmU{vKsr&N_=CkQoPKSE8*qe
zsln2%+_LLY9N+-3kO1BmFu;vX81@Y$%lUH_dnXmk<+ByNpzU!)0JAKBSim+tvwkI;
zLFo?|K-Fyap?Ksh4$oxGZ2zDjh{C2JCquAug5b-dyA+h(MNDb3tovPOjd>YnSCJ(#
zvXjZil{C7hvew8>05#l^z6zXka9{w&-3!F*#o+rF1ZV7#=H_g~V4Q^Qd8|%&M8p`W
z%o_uUb_+Y_1k%wgEnpn>r&6BBl!KmAWCUN7WZ*%%)|Y1~%5J#<5{rKQ`dcF(pKr}K
zktf`sQp)Z(-)ely_RYdKk^Qz?*WCGe=jTeM(8+t<FV{b+cT^JD3c)zz{b9po6DR|2
z=>!1*6SC}ilGz06AJ?{MpuV`#U#xTktqu-l&2Z8|xKbm9t6T!*d^QeV_%OXI{PN{N
zn9u0l!ootTJtx0>^ohZNiw#(tdZ_X09^o|PFoxaSaM2q?TI)AZi=&dutpJH{@_2&v
z#Im28Y1u^|BSvz%R<34pBhxHOU+^hJ6KZc01&u**5y@_nj9+85Kce-0VRh{_ICv0>
z_Hf-8qn`qq>*TD$$$)a-XQIfm$;2NYsY}eKa4H;DQ)_!K`0(LFe<NAzrX^8sW(e_f
z=krvN_~_{9Y|hAm{$lVfanqk!%V}4Pr>mZ#=UILSnBrEM#dj=5Wh%Aw6J*l;4t4Nw
z+@suq2lomZ`LUZCJlLnGM?rON%e_MsN(LM0jfKr+w+XIyJu|XDl8b_>RON`>ZU%}O
zJXI4oBBJxq9$kPBt+tO5V&&(}_Q;O##KJOjuzZ4+9%2!<FL|d+T9g@~;K65@RI;$K
z(MBGLq+xT0UiStN_KC~UCXR>dh?dCmZ@A`wT>I&k%bIbaG$>0ZbJl^-3nSWoS`}fl
z_H$)>JMlSz6SJqPFlS@dVCEa3ip;#HToxTeYF)74Ck7fhcLdp^1C09o3vqo#b_)5X
zoVE27@OzqLS{zySB|sD(rv{CoYm%gqH3p6;>@Flj_?0w%fzr#b&_IvH3H6qll(29f
z{#HR@A*O90UoZvryZz%RSj6v!Z4LyW974wM@V6R@my1WEv_L*8vYzlW>K^{q8qRdF
zoN6N#6|ZC4AG{qZpBbEhlS;%hXKLUgA`hU?CTx-yNpal94YJV~IyML?Uw4M_9OkHV
zdz0}KJJe$|v(vlHpyk`-<a7C<$G$acu)?H}xUj`_GsQp2z!EmXA=@*TM1{7DVd{7Q
z(Ektm+!7+drOEu5d%<yV8$|gqaasUf11Z%M<V8p{UfI;~bLA_bEG;#aD%g?a9UR<D
zvS4S89#@fWM?++Htx#*vj^ZwDN<Kl3+Q}cDXb&Sp-ZjN6X#)s5Wc!5+7tkVoMw@{9
zh>w{jfon%6-#}CWNIT6^H~xJFhY1d7G7|!21Ndye%Yun7XGpF78WE)8UZa$iDE;ob
zyK_UYg;JZ86g<Xg4cyKQDj&Fq_<_mmy)_0HpM<O(St=6!zRpf!V+kF~FrG%cyZP7b
z+l^k>{OG9kls>NFl{&8<D)+K0$XeI5ayx2gDS_kc>?9p<SQmLuJaRAG1W-?MF#!(b
zdE&;hy-1F+J#S+BC5L;lzf4w6R4q0|Ev#&O;Io%+{kwJ86aHTN*3fPtdjdTCMUj#H
z4-t6Q&7AtSJfkbLHuSfA&r&kp{S-fcpF>@1E14M7SBS3KJ(ZC(RuCEbC`i!)Zhc^4
zk?p6=l|PPvP0u5!+Vn!^2pgNYMKwaRHun!EP~VXt#xeA3_z(%oje9Fi8{XvK5^&${
zavtet`}(wt2&CM3>kLWnhCbukYjiXoQW6W*!Id0_z;2)X5U=>=cLO((CA9V6Og7Nu
zLxT(m)htOVR{hGO(*&Y5tT2_ILnF#hV|^63m*|QCqI%@%DRt=*lIn-v-hh<DZJ#-$
zq0<w>s)91yhkJ5C?#A$Zq}2G=%Tw!)<iD8vjbH$c-&XH@4A)DWviU<)c!D;UB*)m_
zMpBeCJizRrGM(!0{7n}kBO{9{53Bk8E0S+L_tZ_+(3PUu8=|QDE|5r{9l7lTNVYTD
zH0ZWE1S4GDl)3V-0L{PUj>u<3&;Ewui3Z@^2<H(TaLAmT5Ld+Z&Lp2L*l|-=ASic+
zF>pNAz?D?2AQWa*Y8hO+u)4n^NM^1-KB#k(*FnM<f$x@;mRJhUlc!)Uzs^-myr(;n
zko@DDpIM<NZM2%=<+J;;;mpO|mh1iIUtEC4-rjj;9rvWyYHMo`9&>0)b>XcU8y{yH
zb}3Fm%Mu|5mYpzC`VCcct;Bj9=SnE^fGMF~Z6Uoi{WDwdV%r%k#SGh=ZJU^fwQP}j
znaUO*Fl&l9BkI7pN0f_>jKoS05q5C~Q@WiWx#V{~Psmu&Zmgb%c}_ZL*iLOUlKbtH
zkIW{1w!X2%+%xw{@A=fjIN6u3h!9GbJpYLh`JbM;U5WoEO>(0l5?o(n`{8pdnbDaz
zIg&6Xd~)=RB^}B$P7M)<JzhAyXsZN*w{G1Eu-|Bsr;{W~pr#-|VMZXr;}67BJY1UI
z;sdawK-48&ijre_=mT{iF~K^YO=`RcF@#VWREO~lotO|etr02T>kzJ2#i!72KnIAa
zn^yo(1~xPl0tyP-X=n&2)1hMNlXFNV;{0Z(yzM@B22c}VzRl*l=L``mcI<+Ag_cGP
z-S>wXW!u;M!a`_|Wm?CQa{PX0qOA$6>4{DaJ{eZU=&+HR36E&fdzN`cUocH{vPQMn
zz-D#V4+ZEqnU8f|qN1Xj^ah6U1CRo~Yz>|xni=s3IlvhYnQ>~{H&ic_3x)EGK07Qn
z<}&(i!ueY#$YWycIwcoGRd4<8L}V+6aB~xn+u!9&abyh@vQcp^QW?zB?B80w!Q;!u
z+;umSuj!2KT~mxs=PB%_scsP@rG$`(BpevwKX@RLslIHb$j|dT$fO7uTk)b4g0-9X
zg9m?LL#-e$ziA--z=y8*vKKN^0;kIB<wvN>&TS9jk)rLo$~d}*XDs4GDJI?EaDvt)
zD+>r1?+14juVydrn5v<5*8a)Wb=BbH>xo$Si)Oy5Os094FnM%_C-EA9AfK)#HxB5<
z8Ev?#r>Vr-it6X-ig0}O7!f>%e`c?MK(<Y*-35(tEbu08<cU!;W=Q=JJ^f>&m%=*P
z>Vr=2li(S2*%l=wRzu&Tdn!?fnu+_Ls&x$|&Z@+9|5dowAYqq(>g^`QJ|U#~a?j~t
z;HQ|;M7M3*3DDm|9vlIw2D&mL0iB%m)95&>1qs<r*!`lf?<xE`JpALw6dX?l##xtZ
zc*+jb#~)MKDog}pqCAGUr|)e?UvYSrNSHrxemPxm{pz2~FMLpHFhyJrG2P<R{QJ_)
zgAgr<in?Z2VwRCZXu!bc;v^8Qe7bxr+8HUq8|7M3wOo9G4wC6%F@66>O*mh;sktO_
z9>-=}tGQaG#+QE9K|*SWo)P>x5IDsa8j$6iyGi%6<zvrrlG5=6Xw17wp}veBZuwT2
zA4SdYciv&vjc-mBO4ne=1U|5=V0m~4ICm%{g5W+7FzLhUedv7RKCGUz3x;})%u2AJ
zd<!BLie66$F;Frf(39afS@9fMO%E6?gt*Jj#er)P_vq6r8<DTstZ*u)T{OI!0|drs
zDMmiFyFA3G^R16f;^rSp{(=(L*4F;PX3QlGf!C(Hh;zg6v5OAnPd_ZD^RsoYd^9v9
z-^rf#5bM6#dw9nEoOxTSPc#6tph6-y?^_oEV%m3)(b4ho>d*ysMVj|x{(ruSys;*V
ze#Yia|D2GiTLpI?WwbqGc0p(hPh#ao>ACg;tz+jo89!7yL>G9x&v=1^7bKJG00LnP
zffENF@Ystg!jbZ^QFtRhesvlIrTR^SC=lB5S&%d~qp=ItBBtTqs9-(rEbk4~KA`Rz
zDhzjI92QY?w;lWik9Q~D-H+pN#YncR3B;QDRATR?@-+`1Kd#=TD$y#Kj7+5Krn1D;
z8zgXIv&E^H%0p?D9D}Yo(@Y_n0!o7n0Ok%QS)k%6;$fm-$sH?Zr!kSW23l3!I3^=y
zJ}y-xideEbo4r`td~GDu{&H`(sLGLEP+dYS(JhcEH3#V+<4Lk3sz$T5@!ED0mpb>W
z9zzv#+{fg<+Yy^~Z5IsB1obT#k|fs8)Mx}p*l~Q&yK(dN&iA^e=vY)a=x>955dn_$
zd9gccnE(#Ehf+K<u>OV;2S85bou1-93cBUdb#SPibR!$Sr8qGjz1?=v(Q<79eg@_)
z<So+D+9g*M7DnU4wvVCdez6XRXm;isl==lZ{sxHh%%MlO!SHe9*Jn}hvWki&7(MtO
z)iexDS}4I*@SHh=zKBJUdC2tcty}kNDJ>wqYc~}~BH&Y9jO6S{Q#cbLdurIiv+xil
zQ*^DW*Po7~DoHQ&GzYhv0;CEmgcPm7dmJdQmtKt@1b%pluT}}pT}Rvr?DD@@NhYYv
z*L^q)!LI3tnez*oKr!3t!nr!`V@xhG&&f21D`~fFyPA2pY1+Xr@YTtypVpM=41P`N
zip-L^jMC#Xd-_moZu2*uvY8q*=o8h>>0z0i|L`iZHdf(+V016yzj<04&SL<gLDmU+
zTzAvq3jfeWAI%)52*VKCfp@gXOr>FtC@m}d<c_;0%nM;L^BtNuP=3UZ-tY+sIIOkx
zHSz*xn=w2q&<8?zOoHxwv`2LJ;cX7)GI+Je<TaTo;9QUTft>EFM6PT*e-R}VEW!1v
z)~%&{@fkSIL(06_KK76k=in0pN^5(N|KjPImuf4eB5+asZN?IBfM48rCk^HLAMl7g
z1qq`uNRA$K2waR;rWX{6k$mP`cWwZ1;-Q=JPeSlncs>1c93E|69c&HG*b?+Mv!OVZ
zQ@Uq&LG9jJDDF_}hKj7mNFOkod+V+0){xvvhwgWkTFt8FA{uHp($|Qi>2&;U*Ro51
z#sAbLKw?MR#=6$3B1ga)1;f_SCg^>(uMQKm`;M}2WMl*oB?=0(ppKv&LUJB~-^Acj
z@3^}=sh4xcG49zTEUXQ^eSj6dILg4}pIB<h9{=+^Y4J|SGhLqt7eyba9()<wl=13m
zOpFt&^ZK;3$QZ*~<s*iyRFQ)UuOENkNtI|gpvdty&A&;9QsmNruFSU&6~o?kp?2MY
z4*DlA50x^8O_wlqlxAVR;^Cr&^66F!Y4b6(8!-HejQs#gjOovH9@^>?nAZX6U>&^S
zv4IjHy9CMxtQwQA(C|UxLxl+*+=jkA5frcDZOTGVU)~*N0`duHHz+v1RaE35ZH3Bd
zan##oeFpWq;_Fr6jR;}lr|f|ekEYdYJ@)WGFR>KA2{sIn7c}drJPnaM^|i5)CY)2j
zK>%jr4W4MID7oWIH0mD&&4F8-2y7ZwHlGA`kJ3DLk!Xw&FaqA!f`z8EYZby(Rr2|h
z?3rWAo!>;5FJYI33-@oLPY@>lhcO3WdV0n#YH7ctY9;bKj>!#iZ4~P<0#JDn-4tad
z$i(Zb!e=?ZqpquK@s(t@u(tN{@YtygG0}6jN=tL|_jcn^Z=}I72Tk&w7n-j?|L$5L
z`;B@#HVdlL-P)S^3N5I$yjIs-K+qv*AELDhH8N>H@kvb`NQAi9p7-C4zn_OqvBTsS
z641R0F~H2^F1Sd}*I*WfV!SMHg#~n6_`Q-|zwSh{7C~41;zikGq}Wvov4cagWokH?
zM}2834>Uya7V%tKHo{g+oUgjTuz@|kdP<}}Z_$m6NWQyys;^c;l@m-R)p1ncpMvme
z-(%7Nk3=}2TcY^OS?sqJV!9e@*nR8s6>imd^6+J=;=%CLYv<g^aPPWH<2*N}EMjC?
z5Lx19pBX)Y&}8r1Z27Z>CU66|`<O43&!XUg&~wq|O_k`11VpXwTm2MVBqBmIoV$uZ
zQg^Bum`QD2f>=@x^#OC)=lcY=^062^-nWyQ)%MNeHxZGME9+e;78)(FX?~qOEhCLS
zf-jw0)wjaNdmDqjNko!lPRYm!!&!;dfMx(jw?w>p1vKrIklk-I`BDd^oved_yQ0-M
zg@he}2Zt`CcPJi}7O%pW;B{Qk4Mj~w#n<W|<3Uyjo(Px;3!7giRJl0$Dh5D+_Co+}
z9LC5=H&RkU`HWL8_ht+KBt@Rs?_}=!4c^|~DtRugc%JC-K)Db4AR`kKh(4%cuNKo9
z^-YUfU8s$6G2F}kG!OkB)bQ_7@1qLG>%rw0y9!e?{1Lq3=qKwR%#_m0qnoqw_v2jD
zUqgwTHpasJ+KOtvYDxqgG0MV2JvzuBUTqFK9e0$K^@RP#?M7Q-F%8iQm4Jzf-u3Hb
zm_+zGT5dzed1ux0oBhQuCd#rnFHEYOrML-Spj_mW51jVU$y46fM&BYCCct=7pW0_K
z0i>f}dt?41$G=tj=nw%FSwv2b94PL?lUF#p$W(#s>%GbU_1t>A{=3H(&g)9y)?hhu
zMks0i0E0d)o3YRkGJJrLbql}sF#D1JRWq^>ymIf&tN0{+I`Zct4BbUfVEi<l=la!B
zbQ9XUvqjw2!5*N-$sZo|I6EV)Bk*_7Nv$z^)4fo$bK_TZ85O7BPs-KJoN^aE@WS*u
z^2ztSCs<x9v+P9XSZn6NFVyqm#S4<iFIA&eKZ5Q;fUoa^bY&VjE3mw90=rLBTal^C
zKthoP-%V6Id$9!zM?OyhzEgd|y;j&mmS3v%Oy78j;tIFX(;t0tMm-qgm%yypopC7Q
zY#%rehQ{2ONYXKTW#i-D9XmIQVayZ;iBHVG(LEW6z~FVbL-BRqnDX6kB3I2x2oiC1
zeDUd#8wG?5JJ(RzV_K1~6a`ghM=?XUx0}*9&IsJ@w3gbZC0(EzpxTV^DtU<p6__At
zWt@P4Qqeua<R*=<a&QGn$6~gID*zXQO&5i};Pv+b$bnGuNft8QLP3M5SmnZPRxQjO
zD9MA^j+F=Z3x(+qAc#T*Vuj`ta?0Mu(-9l=Utr>8ww=uhlP0bzt2T+Q_V%%4mjxpn
z<DUX;+_7%$(Sk<Rzjx=)pl?M*2I!2!4x{Su!34R80^Cs&k<eHmqpvG8cb_&s$NWqa
zE-28Q1DBjtOj3%kgZ%@xDW(9t%GLwq0Z(^~WEP(+e0&!r8961+iHZ3&txVyK$@jI0
zISJk(Kp)@Pl75WkqM@0>4AA(Bw@i0{=OGlu1T-`^d(-kt7YnIUWogj=)wrW8-@;9|
zZy&Sda}51qpBFu-+sJOZ)_b3>&l&(*vnfqK^Jj^3I=U$<@>dt{aXR+wc4SSQQa?HY
zU?gJ8Yr??<WjQ8VAw1eo4e)_fp;s2<*W`-B85J_&d1KUR7gs!@gtN0V!1557_aGQ7
zkN4ukWuvtRhJ-Vc7<B7(_4R2pt~HTH;Zh+EDeP(NeK@=VLXCvlnj1hR3(F0Jj7Y6F
zW!I520uo1n5CAU`_XQ;k&On%P!JQ@pVk6qB&CAad!&>9vQG;xl>EOY3FkFjw_yq*~
z9vag6E2cp#Fhj}p{Y&-9Kx5FEVWSO~!8jl65NN71ki=<Z-!e0UMZ!xF@Aw!><k%ZF
z@By(vuZ_%a!1+!f8hw3;wp9?hP?n(`(udjJdjyPQ5Lgk(xS5%mPo5+VG#vOBB_v3-
z9}%I0>qQm1di82FqCmoZd4l)Zx?i)(n5h1lN|@`t4&b6$G)@3yU}D)AdQ2L)(^xyi
z@?TF+4~E*Houm9fQyE<nhUM|G+6$(?_)ndZ4C3kq213N+AUZ%;j$=|?JqUARGcs&Y
zM$aLUfv-kwI8V`Dt~ox%dK_WoE}>hFrVV1W)BGPG%<{Mu$V9L>ckJHXfb$w*s((qd
z#UtIcE!591Qfmx%;BnFpbu$=ZkdF<n0(Jr(Ne>Qs9CXNo;jg2GD1km3k>%rTSO+os
zfM13NTv*vj?`}f3E8lH$*`9S7lG5>aK0{^y_NT2<Z?ACZs=5eQN5fCGZ8)~B^-3$9
zTblnZ%q{c)hl1LB%o6+*`WXr;&()YGlqqI^ZF7fGiRVW1&20K?<wVHM#5Ukw;lVb*
zm+JHL&S+d0Rfp2bGul|U^}AQ15jN}Lu}NYw`V%~4-psmSY!;?J>BN>d#5}UwLdL|k
zf7vkUgUhCLGhT5VgD%+RP8@UWRyt7uA#yPh23H2a^b;ENC@r7BKup?D$E#g~`#+(Z
z)JTAAUjj0jYMgL-3@#LW-5b6fu-ZR(dZnxx99!HrI{e~{*yylj1HG&@;S;TFwfjxy
z8R!jOHGBq)Viy(9V42Db|H92hK=4^Rr}LPto6qC%V@fJ2m1BD_6&|~h1A1+0=(WEp
zl~RV;DV0L}efy*sWMi9tBC2B@#07jQEQB{@ll(&LZ89Umez3j$9nR>70S@dGctJ9`
zN>1)+Umi9tO}X3Z71!_l%x`DwcCX06Pf@c6UUO1faa<AwH}yvMNC!Ey*!ynao0DNt
zYXx*7@JnQt5sn;=7RnXWJE6e)KVHif3F6Jsd$d8Y{J@HE(O;kR`O6PnvHtf|$JdZ%
zDgVYmW99h^Sky;a=z82C)V{*1;by!oZ10M2*JU9X;h0|hw=ifw-Lj$%-@9$OtREvk
z{{C<A?|**re|>u{-0a=*Q4xH!R7*&aZeK-;c&-FJftbM~;HO^QM(-=!*<su&yf{iF
zK^U&j3vq~okf7VEh5H<Cd_oKA4)XQnzeMu<|E2=>{+lyQp6$PWu|K})9<(W#tn%Y~
zF20F*<N@FUX-r}=lLIpE5_!U!t&2e1^dDL^aV=Vwg`)-3*cCnOm@EJJ^VwWo(4NRO
z%i7?(SB@7D?GX9;&vXBWyT0N-+v5LQ!Qia|%TE6xs1a5y8X^BDT(T*!MT-D_)=D6O
z0T#_ox0%@TfGo+YGL2unaxg1W0Ds<I+O~UlCT4Xn^kaq!zKk5|D=Lz~=<4nv^1@QL
z(K!53cO8in;KFW06Uao6EOi?aCxocwn0iGMVnQS;hDmiGVO&ij(yc2r%lkONE6e8w
zGG*Ovq7^e*-p3#NFkC1T1~X9^oFG(G(_+8gw)-o2<_EfoOpp8RHb`WbgOBoqWsM&9
zNvCgy|Bb!(0IG7^x<v7)peSO(te8-;iHv|4P)U+=Fp`4;o1Bh_h!PY5iIOA-0m(rG
zOhm~D2&hPoA{l`m3-A5!-(6kR{a(H5dR_nBbpz+H*<pR_n`_Q7#~2eNVkB*qbvyA|
z0e?i;hG0L70EduQyUIlF^1(l9xm}QR=p!!^Gv$0%;kJI~_4msfX^rLUs=^|_?>5-J
z!4NC{(P5h}hy1J2l4bEHR1lNi82tG-a%Q>q$lk?_^uDVo<JEmxmOoQdMb3vWlG^J?
z;JCGXm^e?;VoWXk?IP6wyg<;O7tq2|;aEGRw*?pqYCm6DC;ZJ-fj?iZs>ZDyeFLPQ
z<$pBUe^fPzP3gBP;9&)~|6;QEN`IuBne`&?h%F(3t@L%t%fP`I|JQ^Of8>(PHNg2W
z2UF+<m%r=QKNlbr^S}{vC*sSlD&on50MGILsQaGkp(D0#&6>yI%bM>$(!AYcV4d;t
z@%5!dEu|{q=_(tS1#i2*4iC3y8Y;I_b|gU|b#=?~I)0{Z@6Q6T!vjmV^kwa~-tv1I
zor5VO7!_GoHhkpGl$HZ|DjgDe@Y^;yR8&`+V6@z3qFA7M|7Yn>WeBYTN~i!PvHm<+
z*F%36a3`QtRBRjlPc2`U`yVL<P`f3QL1cM&nzLMEvidX_76VN15d+>2?pe0NGyHvc
z1kh3dH_w~08Cv)9-9A&|_>VZ>Es7kK^5e^nf0}{(`<P$eM_9D~Q&|v69PoAlZ>y_Q
zaJ#=OoKgb`2VG&f?g70G2@QSq>J<>D?_FJJ4qE>4s9He((tm^<e|LuL;m=7Z7)YT4
zSFw0dTE;Z3mLH6aVd&_xNstk%DAN)BK*74yR&Q2PTuclnL1qBLcM*IyOcYB6yUYO-
z5hsm6<*0ztb0QfQiy+M60p`cV@&k-;fKRX%!xYGG&^o)iy1r+Gdf)jLO9qU$oTKfq
zk@uIt-xRpQC$yb5mFW>jMg%tr#=kjiAb`be6I8QacsaQ%rhtehO*h?rj`U?e)#dY|
z3NR|IcyL7~LI!d?uzKQwUf;56z!#0_Yk3vxb#+;$oO%K{wj99gJ2rcw!S+|3+ixTW
zI_r9Y?7AOa3f0erc5mD$W9bIximQ9tXZf07J3y5MNjDKYqG6C}aq%hxgsafMWy1X!
z@l|3PGNLou<eqh|(rZd;`FkY|P79rWuQv}71S-M%L-W9l|I{HIEp2BSaQ_&c1EPRR
zJ>t+#Wd(%{TsxdGVS-iw%2+_Fg24jlB!K2d0T81HfDs$fL&skb;w5B+z_se@_9EKX
zxOk-{RTBAW(g(1eDQ`Itp-3&@RYho;xIn;MSOP?<HxB&zz_5M0Hr5?HV({D5;V0Nf
zbIpIY3*_s#r2Wq-i29P(<n;Bo4;US14q%GVhfmIDBLXn2yCsAgD^QyH9G^bb1`q=9
z$=196C_@MMm1>KNzoAV24OQwyJO+{ja{i2>RvV--0N@~b!b1o=RUqiN^v}^-$NrQO
zOj@c%r3*9C5YQJO_5j9)avs2$W`>?DIht~|pW_!4#LO{Z?A}EC@5J{LjE1cX?3KaZ
z^)g^a;jhRzs`<-sZp|F($C}C<Br9R`+{>*?I)o1vSywv{Ynx3F<_J`JkSQA7caWqr
z8s4oU9(8%Mps(IKsfzmEa^y?A@pgPv$Swl)UIBbyVqz+hBBNI1yont_2_3{iDZMP|
zmxQI`0fIsp+-R8xY$eW#1mE&CB)ng315lOOtqn8;WINOM4*Yz4n8y{DlcRFc<8X%V
zr+tvH^nLu02_E2y6KM^a_S5ZFksOSeNvkpo&}?cd&wrS9&6<tv6}aq^r)9TQeqfmo
zL(eMD7%EWIo)e?E24_@jLcMRG4y3v|+zOdxzNK+(HkO@?EioUy#QBT}yg{1Cd#ZXC
zy9)D3<j^OgYrlb|hprf$!zwvL>F1|&L4qk6hZJ?qIxJBenb(=@o?`bnEX<UR{u#J+
zBY?2BJ4kRo1J&~UcXAl+xJQzRR7)_DT07>z$gRqW!gh%OobcXQ9n=$)4hkt6E#b-F
zK7th!qkF4XHV!F`xRexw|DcM#0$vQ`j#S9}PPnrP3EsGaTLav$jSh*p{F|+?)!yu1
zzLPRB-91n*_iO~8!(2x06p_uY-Et(Lt_1s_;AYuo^T~^GXv4wowT2lznI&kJ?)ky=
z3|YYWwtkT9@JQ8|kQGey3kV<$&)RI<wTFrSh_mSOxog-vILM~9_@iu#xzMLurlv)>
zosCEDySdTO&?qOXHG=O(OmY3$F`;WqGQg(uhs2VB_@D{=%XPa}oI-pZ>?SfG=rZQc
zrwkrp*ML+ewFdO#Hg=;(@9KB&--E5FG0LgF+IVjY_#MEe-iG47J?`w4+pY2mKdHFj
zpnde772phaGbbTXaU2X1cF0R5Mv589NB7S?2+5k<1F#mRhbZ*91_&EA($8w>^MTeS
zOk7*bHrkebMWw!~isR_fiz+tY4pWDos7UDz=fdYT*3ASEM!gqd-q|K?`35#dzbdjD
zisGKu3xFcdSCG+>u*tqQi_Qb_CUps_(?L@Oq|kcihAfr8ffVZH_Q}+vETmO_xN+|X
zH5C;C{0=s18Xgq5^V|o}v9A#MR)21S;S8xZ?PzfZ^Wn1=d*URI)Ss}h0GEl@^{h$<
z=(vCo`*S;C>4GS+t^zQuX>HUL5))%xY(kK^o0&1A10>`c=B;~^@~|f`tiI6JND442
zJCSQr-B`|onxq5t<;Eu_xSBcP`Lkcm5fG7aFb0)zadJwln3!U=Y>8k5EReTu+eTZ#
zK-MoWRio{OHd<6<epRZ*p?TutgCuZGVoQ_w238=R%~MKHgZ+131{~`BY9n1;QA<zt
z1RzFv)RPM%eSq*5M%K9BYR55&q%U_6_q(AT8)^cTxGwkOwIOQQ^9ZL&bED`o4-a#f
z7D0hS(#na~YK(hrK7p?x-m(SX#^ukq838UA*^T5=R8%x3TN^xw+HtwCv^pixwm7*1
zdk0G&3iHO665)<!-eGnpaO4$UUZ6=#QO`!F?QOV&syjMlkNm+;558j}ir0O$Tu~&o
zqu-TL$P{$-q;J;G)aaD8;mLQis9n+4%PhQg-b->~Q!X|LOIU&-W<;UB;p)IZk!_)G
zs^`#ik44H5X$sDp*7xfFbn`<LfnXhJO1Jam3qtYDA`xotMKChe+YWM+R?AQypX=A3
zdHllo)wT3*Q3YY2rY||jNOyyzfG&-Hq<PO24C+9~eE1rUP@?X{#Odlrl0Cg?#pt4Y
zZf=iWW7aO(kh-`Rw^-eQ?!9&#5#2Z#k?{lOz^G`+$&I7mwq^cC2*e8jn`tXLZ3j7U
z9Dh?wJQ=K6b^z)t7kXg?mbf@Qa1)l(<bT8#x*1rQU=-{5PFfUW=x&{2>V-xU(DpUf
zUSu_qlZ3KqELDj88JWli^^&TEzJmUPo@R{pByCn4(E7e0p^sv8ua9G(JouS7L;7J=
zF*kxtUZs-U+x*L4h#G`8CuCl&9eV*g1G&|<|9N3A2li6YY)>$m(*TAD3dZ;OtzvvY
zTPTPECF3wanS^WOruMh#E=-?7_@Stbk|m;8WbD_vofZbQThl|D$}m9}P@OQQYoG?x
zb499!9A4rJ8`)X^DXAJWg?R)p{-#ZNxF3k#=N}X1fka0|MFEUPYyp)23h)2o#bL!N
z#65j3+(k|ca};9^)7FfB5MU{2Dq32a;2+Q$fduKiA&3&F8<aoKVjs9SL337+)m2Ct
z^iO>)Y+i_5N$P{D95NbxQqYCnEeG?HrPXLTZ9B}yCXw+3i4rRt22bK9!axXX)ZEyE
zI4c?*WFpJt@g@2v(Ez$w1O7A;RS&GJu-%lTm6#Y~iG2u$7{eyKpTq6`{fAQ;<@O=)
zz(v*+uPFpo!Sh(njHA@EAkkh>Z}Rs(RSimaLhmMA7GG{8b<1E?@TR?c+p#JLGJY4j
zaNcp<x#NO{Lsh6?v&O;+NQPQ*S4kx6M6vPf5o$Il#435Dnzvmpwv5V<gn-Ti3|w)W
zrxV=B#k}d^PPFK}VkLATm2FI2zCnf(($b5_xTE_eRoOTp^t1j6gx1KA^E5J+C(Fed
zPY;Sz!a~O!3)mKUd_E0!8N%3d4vh^%y79<Is55iHP(WuKGgaR{So9_8030y`abt+a
zsQE56g{7OLVFn=&Tzw=XWK{CtcSSY#CoQ_7rPYB17R2KjTuS&&^yMlV84=2@gM$O`
zyv#-xPENFW9M7D2gM??YtQ<)b@u3Iz4ZV{3?CD!qB4mi^{>9bpqiRKn$7py1lc%Ak
z#-DZ6wUZ;j<4YYeS~^|_XdlZe1X{JTvoltaG?bwqh{S0EBbld9<<VLf6BDzy|BZ71
z=Ma%51_bO7Qim#P7o2b)W<;)n$^^}>L~IXhTifz(#QO=H9Jo>;kb|O$giv(L)<dif
zD~c@%rlPB9D1L+HjQ?v1YV16WSjMArSrK#vk4u+wtVXpkXbDCr2>l8wDti6G$Q^e`
zd{CoL`ez}#<Kd(tz72U8F5BdhpZ*_UtIM@M{^ZR63WWXl6UoZr@ju6?|9`tYhj?uz
zpsm1$I%SER#5WVW5_!tCTn$c;DIz8ouzQq7kkTX1DlxOuwWFb@>7QSza^ONlBy-`%
za4C;ZAF7Z=MtKXLv_*ZnzYsks=>N{Cj0_JubiJp;lTp-t0m%mOiId$*|Ht<Q|7^T&
zl1zt7K`dL?u8g&4q5WUEc>k#h@&DZgz7Ip@z;>t8c2kvg?eVw07)4B`i{Q?1tzUWt
z^E>dz8=Q>MSHuxzgnv^9{X6?znm5}vo%WR&bW8G{=acRVdzSbtdj8gqSBw{C`8dQL
ztLLqxno!ixI1R4Rqz{+}BpdJ_pik0<aoZoqia#A)jQ;yqM;Bv#(2PYs-@dDreI4pV
zPoyUv9tD_Aheocr_(KS7kPbtrwhxT{&7139T;5V9a0<#moR<s?3@l5Cj;A0>>vWu#
zO5dRJ3Uglo4-kKHpe<msVM`G<43K4<;meLXR~WGuC8jr}UV~qSZ_BZX^BdkDu%;>(
zO052%2@56g5zCuF^8?c-+sKKkLJfr)UZPO~l2f?K9aJN;w!j(R=AXap-eF^-U)TnV
z_+O}uJWCLBNeJ$^<(q<29$l&97Kz}<z>^?YfEBjtpv>uj9Rv3KKsRVhcz~J*4w)TZ
zw6`+w7mi>gWrMHk)Ae4lLy>qDnI_UDIp(bc-9QrXb;zV+paal)L61HSzorelxITio
zz4RW-2X+K1I(B=ug9jtwiWaFpJ3kL&ta0QLvVKSu;f2AyfB%^!&1WJwa29?%y#hMG
zG0$=E7xBPC;dE{U10Vbng#08ZB`yf8{VcFRv|78bvk^-8=3<vkQ(hfyZ!3(}*6`Ex
z8s%oa(@0*?Q(~=Xm`#SG%9lgw;Xks0JqCKtpWojNFaZ{~LGey!d=$ezqcE8l99vwW
z0H_MZW)h(j)y21bSuidRdDsQ)17Zq+%SFgdo{!a-Sy@>jCdT-OrT0MRI_!T*)mrB%
zpcHIo4#YTyO~bzQa8ZFW3uyS(BVG5%&;%`{jVz;OelBdL{CukdTvhMV9CV{C8QKn2
zG}B!gs#a7IMx|hp?Gv_-(>e*mKfw8if&fP3+fApy3L_CX9o`}gCL`xQ%>m;Du@h8T
z|D3AyR9H|zuXPNXI*-;cF8w-)A3fTL2o^Aa(lL-kZpXL8nE{&81r-!Inw($3M4beD
z18`h}Yl`~Y31}e3w?jdD6Jha|EAj(0q8;Xkv$qYLyqZ|a0NR|dola5B7XG-6r>Say
z#a&$~E-rpg{upR6)j3U&F(VDqHH*C(jsLfTBE*F!7uYj&J4`0^L+HH#Zi`GeBv={|
zF|%;2anePDUPs)8XQ17x6Rb;_*ZvjVNUz@^r+|{J8MzYJ_wbUKotXilofuZ2PHzHz
z+K*d8TpZQt`$`7q#mXK3`fSXb@Y$ery1Ei$Ub(3EWMcc^PsDi9+v!3175wFxN+{eH
zM_<oNs|e!;!UISvwy)z{e{6}(6_f$FcGD(htz7v;-Uz8qi5dLTkSUVcGq``T=W(xq
z6|9nW{Q~g1^#)7dO5I|dE`!l|$J|{0nEQhVaeCe%N!mqP+}krXI57qM7cw_wHy7C8
zh2ZMuMpDfLAE<nqAFw=ieppt3U3`vTXKI}g*~}~Nb+jycEuD1obSi6+6WK}2cCB%c
zpGpQbX|h<^;{)d;^#PGq25}GKc!}Y<n+|qDxZ0(PHPDGi-@faEV++wmY&H+nHe|z-
zV`BlB@hO%GxDIwaGC@35BS8F+Q#CTLhmBa;jf+?!MJ6zS@qjKU82uSsfbAS=H%+zY
z9OA)0ft|=aK&b>WuKdvZEI>rLfQ;0$ep?BA!>{pd<q-Ek@FE*yOH^TEcInckc=cqk
zZGcpki(Ww@8GPaf^_8*}bH_~y1-&YXV|jMV>42opPv0vvKHg0KTZ)ErG+!esH9RN$
za@>n`U;m0>3!BZs7XZv0ijRavMiPw*3MHRoZ>W`)!=$21t1R<Y3<lKt!=nOO2HZ~w
zHJcGPknD2-g7OeJ=7|u9To)~O6vP~Ok)ua3m9h}E629$Vd;T=aTHviOoD7dV9|LIz
z65Ivwvv3ArED|d_JDfN2kk5j&(P`b~iNMtoe(qF7Y>W)^1~l-n5ZKOah|e4xRaI5V
zcv&EQn1FN}L>;;TA9tUpPceU)n^wwOVFRNu1hci+lYyVT$^qCQ5}JV)1fUxlJ_yuo
zV3}cSOZ*y;<PYO+$IT+8FT%poue~(ajBFiBtZ4*VWYhp)b)dyHEu8e%NNGmt0K*(%
z`x$(2LcH2%Yh#lM*dJeqNS;lEg$JMb_(blaj{E-gYbxM3&l1>#uxr28xZ&@Q`vE|6
zyiz7-J~jtv`lwi9#UhG|mi{2&F~*6wr7@ORF>Hzy0w@j=7w0jygG;aGjL^xZPLQd&
zc)s8hfz-HOGrmeJ8M;_hvgOUBEW$jQiI(!}YECRA?!xkyCxEUzeDvto^z^Y(ub}&1
zMX#))zm<Rk{navkiS6n??gSS){h`Jg&uyj!a-)8&iU!J-3BSi$O1}lW#rDP|vRW%m
ztxsY6l4m(6rsZibWZxdr&dRS;&hoL0^`rPl$(%s@Uv*C<Xh!!`vQ$0^&u(N>W^BE#
z=e^yo;a6O5&yfG~;C)Ybn2%K#Yw}t$Jo8OAdcZ{^_bIAl%)eByaa5tXFy+-3etyAO
z(54D`r}udZCOq~_H+U$j)@;&V`|^fe*p#E<!f95U9D!V&9^IIp7af=Kem1)lZD{BB
z4GhzN-|n~8)skkzo|-h`WkNAB+y+oiewY(;b~CD>Z%<BmiaRbeR0x*a#t!>1$_w0~
ztag1azsHZmISv9x1Jqkr+px6Ixx`|a1uFaW38=NC8ixRjH*}g&C>+D}Ps?tm<2WDa
zaIL{~#211VV-pkW>F79^nVBcq1I<bpXx|mP&>+60%&QR$wxL16m?Pk)ygX^rK#CI;
zC4n>IHS{9Fv90X)rn3R#sShXtKPmJ&CqqnsE?;(ke}5o*RpX~*F9$gKo(ubVjL2mu
z1HN9C08*H~u!7@l!$&wz##zWFu<z>~cu<5Y544V<??jpcXgkFBHWk1dsv)01`xf#*
z#7nwhpz%tt8+s60woielF5cy-GQmM#vVb65&+dtSho}SBEwoy$-?(wna~IQ>8=&FE
zC!>+IX6ZZX+ra2VTea63zx>mEgq>6*N!PS-1O_KJYq9`A$`}iRUV3e{geFG6DPTm1
zkoDt}W-=H6_dbZE4;O6K@5>O<9Bi$dbbYD)c&78>yzu6!<3-HcAC7z&-Mignb1G~2
zWxc1y%~@+ZY3EM#;aPf9Qm?$Ho$SB;blvRQ(<u}&bayDMdZ~dcBYvdXaM1lZW!pa0
z0Qtv<TK{e<%XV?1-PRu8LVsSJURz`8d%~;4hvILK1?e(AAO0<`8+GN;Tv3|fdo4#>
z)#<-af3kc1{Uw8^5|_`=(O1=OlzT;o0LM4fmYH!oX<+?;yC`m{UjwLY1b9SqY&_CZ
z)WCY`1u=1PeK15J)*Bw`Ij!lW7Y%hz;}+V;y1F_@zET_|F8aD3J&jRwh-x6x%Jyk!
z*dw4pr386xL;KFaK;cZ}lmI9+r`NMU9fonO0N6G=SRoCxT)O73s_z<j$J5n+BKD{A
z0z7*>^QB^a(A;zUJp*%0!kkr`1y%|>3WV&&1^0DQ(B8pq3{C~S2B1wsSKa8gG7hbR
z0Laro>vM!1O1qs601xo~rdkbO0@K*hAoNSl%Y=Bq15x|K?IfO}Q1dx0{tAOswzhB>
zamCe@=vR~3)f^!sTZ^L_q)P)|Kn)e_QJU$mUSY2&s;I=V&VuS8WpgqSyR<~t*rIcR
zl@~k99Kdh)Ad+H)@6X{?jYWb8oS!}Ro&q!RF3nzw5r)0mkxA&xE0q{dD~x><1p`v>
z$JBTj^#a$6+)D|y(mF?<mUS2{^fWuXHc`$F9WTmLv`qSKGH+#7l$&o5ozOFyh^g0Q
z+bq;m!`uGaesO!P)$S&}9Ge*(D_yH^>UDb_6oxMFF1)r8`RYI;NK;F@V)CynuP<|V
zd)HBOCl=Qh-w|zIJxKl1P&40GP++8bq+U=O^KVDG2X2+`t5C=-|A1;^7_lBO3{03{
zf+2|pgNexqj$)CyzC$!_PD_%=?0b-XMe}gOGZ_U7>_D#s9RCLY7U1-sDk}?ae=EUD
zC`utE0WO`f-aB?Ztg!%}DbKcIf-m$==+|D7mPX))ZpOpYla8L=2^pcL##^T{Zm3Cs
z9l;Gsl0WAVCI#Y<6A-Y4ffy{R%4s2HutkO~G3?zR)_M%5MllT|^9>W|$594d6cYoN
zciYyj@a2F7azaWy$_}usB`;rQ$yAOZx)-s_@m7YJJ9&%(&;qwDLk`tYy4nTx3Ndj;
zjH-8f3TYx3;-EunFG&$Y7O=b9DNvr{*k`|n#*;PfhY1+6n4zfx2y1Y$S~UiQen9n@
zUj&26M-Lyuy#Ur8Fr}JC+~PK9&;xaO*HiPvd`MkW$KGw%-kJ<T<~av7=0`kLgZShC
zwd$CMYNOm>tc{#dWag_2-Wq-h@cD^Hj!W_<DALBP#Dj7Ag5<6~*$J|r*UnS#-)>mv
z<m7iM%Fk|K-7vDqom{E1N1lQ>EjyW1j1Z1nwX%i()B)?7SZ-+dY4=s`=Kr{6U-^B-
z)$|OfiIV8EmujeP;ehUEd!=Xtm8fs@D;<xU28%ykd9<j#$GDwhkl*;cQT4~CpiLfp
zkG-`P`X<F*4yMT0XZFlJp7h)ME=+^dFSOFcrnB3$NTb|Fra4ABc<kFZlWLP_<;;6#
ze%@_RslkLDUkp8#HZT+wkhO!M1b-cvF6GXho1fFr#%fui$Bofksi~=WP7WB`;Fcq|
zB9eThX^NYA;nA}QBY|hN-_Xiu+4>OMGEw}z8-D?+&bjQJ`}S$UOYdnJ`mNYC=gytO
z6eWcyF@Yo{!+&xC=yEk+QpMopM6zqr<AU34AXXMd&`1-s8tAoP!+5y6^MgwR3l5CT
z=o=m$hFwTcPZQD*Z{fK<q_Oy=tdiG*2h5Xa^}feiLb}pOHw$_-=nl3ZB!LMI3y&B?
z8r1aADBxgC!GBr9b~qt+5zpg%i5S#hCcp^9B5KVI;O?IAQi1tt88#m(hNE3-#C8vc
zxt^bILnMx!Tk~oh8PsI_6#ASmz$4r02>5sI%C87BP`a}_dA`0ZxQTus4l-<%jJ<Gs
z%D(4{w03hcR)d3G5?F1(`)Kx7u4A=s+o3BV<qHl-1ZR}b3&_He4?{hKA&<Y&2gtYR
z-@Ic-oG{bfIGo9eODq`fyJ@e>4Q33>M3km;7%ANrFS4xx_53+rD@#kIPyyGovSOW0
zk_F4t_<W1L78j7Q@S<jQH1rJo%iGL%(R1F=uEIBWR$#`m@5`Hy#gYOZ1{^0lRP|))
zMfdqto{Eo-pgB1ABI2>P+o#H}f$u5$a$jG_$vd8<OW;;%vI@Lo$8Olo%7}7d8eODJ
z<2I$4CV)5!r(E#8`^?oxu{c8@_oo)%wCJ?|<vC6f75#f4Q|f`2m(uGsw6wOvnavr?
zj6FR)Jq&VPzUGj2AagVTGbC%O{mU1ke+%+GANKW=?-gT|_2r^;US3{M5;RX~EWkTK
z!Td4le4rkRo+{spgO6bwGV70h)HwJOPpqiHE&;KAy>Tz@YnXm+^lwg3G{ja1Y5{R>
z5t4Fvd!3Ip35485blu-lh@IFuU>V?)bO5F59s?QBbw3rB)=uEIlmsCrL+5D@Dx0Wi
zF<SS}I5k=Ar3sFXxY1Z-Cs$rxj?CTncE|<v3AMDIr#dz!eD@AAL}pqbzs)!(v~Obs
zE_eTHQ+v3ocJDH1Wd-9rk^XT;c6KMx=kPzDY<}V1^IaRBA@~&Bg}2xsHF{yYP_pj*
zdPOo=c7PJ-sCa5MQJnN~8VOh*S0*P}>YlQ@ZV&g6cw$uVqDwE#KX&;;s+ZK+;DW54
zO!xew``iU1wIafEzFzwBWc9^u>t~HGeI0ny)5+tcKP!=3WZFL3>-cC%<@ez(_tdH<
z$r;s83?i#)gXfL<qJ|7+8gE})t(*HQ%Wd&6-{VN1NX1?yih6VM9lfqMjencdS4mZD
z+F5fvvT@C!_m7(_l4l*AKU!={i%*lEl3uIAP@{Jt>TB5DXQ3MY+Ane+-?3U65q&Ls
ze?;~#Y@uc-y!RR^$H&~Q#x;QsawG<74+ve7hTCyv6R(q1-<ftB%GA5DaD52q2-<li
zhX)Sw3Hj;K(mQUJ$8KX}#2@N(4}yQt_x!3g`v2u=yKx&KE9M;DFKLGEr03>bjg;Yi
zQNm=7v3bwZmnWCR#p`&A@;5*fRuc9bw=@YENKO$REX~XeYQl?QZiejTkj9Qh6lQZ*
z&b+ezdA)n!AsY0s`=0??Fh99Pb%kb^O(4+WHP6yaK`p|81$KWAB2`~sUvDo#%SJFT
z+<J>-4sF0BfTdkf3%<|UVO)BX>-|nYwUiEN)eL>hl8d`1e35M#G%!z!`m)lxUPl=@
zaOD}1@?P9kvOjW5*gB{B3dYGzG4DAp#w}(_%?x>S@3*djoS|wpeb2)^H&@YbW0D$^
zxkltjF2)aSH#g#v3f{0ha`VX6t=muAF819X`ZjG@IL*}+FL|V*@{@**&Mmjby&Y|n
z9WRv~3x7|T$yv&oTr)|F<U1wuOJ}@_?itVe-<0q<@g(tK`-!0VDjBv#wG-1dU)or*
z{WL?ZtUDO*9K8Iar=yBn{IKBTo%6cvBDudE=N3x7m%JEyf$ekAXy*ZqWJQx>PEqf6
z&+#U5Wwn16H>=uxz)Egp{T!v{wfelAncPG1htF@GR&_9$yr;S4-nv}rf#YR0LgL{X
z-Cg(T8B@A6C;dhx7P`1mM(uqR4$t#MYTKL8)jfaRoX`;eYm`ZZFg|{`U!wi8U+45U
zfjARCc%Q+<Oy;7Lyl^H+25}=fuP2+-^YVXS)=$Janr_2yOtSz<6zJB>;W|=y6<NQ1
zfb9yChYa0G)D(?Z?$)8CYh=`{{Q$Z3Jy$T3l%1R}#6xNtIoQ|FPiGA9kBm|DJX;%9
z2+%HJN!fz05_!<U-@+|X7wH?n9`S>n<R$)1k|=MmC5YQAwgegxV2Hu82?!ow)JqwT
z&>PLoIiCz<7Zg;_h{@8XoxD}l5HB(=L*fNXb`9vO=#I15Qy=BxQdkhQN5w{<n+Ywh
z2aZgfgKoEyM0I+co`vvoCP%$igob$M+)a~J^mF^IAGcf`yXwE3-I^%vIIZw``9~Ks
zTJxBZP}a(A@kU-9du&-JY@SzVdYIKoi|{3*g)OIO%KIekbIb!N&xM~MJ&o5!Ii859
zawj%4FhA)zBQ0{8W9W*J!O+~JL03OdYqjJ=C*^D3JXW5%7xq<~eo??t*_;wTHQB2E
zjUBGdjIqWZ&T>^jOAEQ|lj1D|i(<HYcX58@;o@Cqdg@>c*8K=HZY#9aQ9Dx$O}#Ak
zII8i<Qo-<-`^j#3`3vvOatjauqNl_e_yz(Po7o$Rido&gz8=4Xx@9}NN_g5cVldiS
z0xYGmH(qfQU1+A?ApfZmumluO86>V*Y1Oi{7=YSiP3*Cs)6>dJF|pDMy656S`f6La
z^2?y3U=|o8KrXO*Ph&P~_tp{|GB}CQke)+UNrr=*UJ20~Mr02xHBPxi7+b$K<6%B9
z`Et*#+L#vi=X;`&Ar5sGkpvlrpscJ0NHR+V07>uCJ&BANn=@f4FE5X@^;=&b-yu4Y
z<ln!Yo-JbWM)>f$V3PsevUbgzv&SDDTE3ktsa<I5mv3TKRpT+46MydKRx3m8vpW53
zhXg+hu4-5pu=BKeRXDfXhkaf_TNzGq92x22Yb<Ks{isq;XG&4Y>irPAl3vQhLSJrA
z`Qej+lIl(IN%V)m>)g|ueI9Ig&y~Y4o2#!(@R;&3rz=$#9x@8s?RdK5klIpRy~Hl2
z4VD{*9o9cD%1P2MWYl|9P+&oKR%wati2X**QV^<M<lmj!vj?O>+lgw{^}D$G{FJq1
zInNOT9v~4!+=r7Dbs|z_L(GH_eMV;<ItHv2gUx>d+Ct#~20>M2WrJl^)83%q@pvKY
zu`{!gadDRldlPKrT#xyd+gAGCR=F52e{1jtWUDy4Wx|34*x8NHPO+69IBY9e{o%vK
zHt&0S`*$-kYHMgD9Xc#Wp!jWhx$YV|4d(>DrK{qO;^eFjFH?;U^owfHPexTt+>dWn
zaOoZ>#E2rykS$s&dMVjTO=YlqOKh@6VQ9Ku{ZIxsk>R0rZ)0oAx`dmp4S5$SoB0GQ
zaSSM~3xny7=N^8!g)>kbN-VTp-&mW*AFbtA{x{B$M6KP^MtA3w60S5Qz?RG9tEf(J
zv5PxNz>1TSc&y``mw(g=Ezq`|vcL4}fV@jg)18;{22yl7PFn}Z<)3^Qc(v}l$!!6F
z2O7Rie2)VT|7`f&o{_)$-D>F#OXfQhhE!S#eey>xPBk=ICUUS(BoFMKz7z7C0}!k2
zzR8J~?4D+m_90no*W44=^oVINbe?195&I;%UkFybh&h(SbDbB3>g(&XjoWxozrT6&
z28A-PoN3^mwU&wEF8xPOb8?P=J4lW*T(3h152B{V%=nlanRMS4v?ww51_Mv-qy_mb
z$#@>D)vM2mJw0*sbeXNAKpr7fqHACHA~p3k{>at@1BDN=H~eHEs`!!;AD_*WvoA8R
zbvAADpDQe3H!pm&FXn<|2%l_*#=_clNGHF2sXICB6rK@kKiL`e=XwCe-{I-&iGmDe
zz&-B9W>pXjG$bpXT-*U)67N@?f&L<QX^UjF(zHGDOIxme;X6MW8)DIG!S4E1G^}aO
zzMEM)PY3KE+0kL$CtoA}f~N8=O)%x?q}K?oge~W6!nCs34<fTJ3=KTR@iXvX_D@;~
zaqSqnBaMHri4BuymGynl)4pDfrs-AsPLsE`c`EJgUr<k-ijo_3X&D)>1Fr=sAX*%K
zokc}rIdBg*hbjYkCv0dRNsnFALUuY_6HyAb$Ydueia+FI8Ryt@4fTR`4%lmSYKZ=Z
z8HnPR7G_voch@mh9kt_-o|yeDXX+8`I`Z<i-1zkLc25)T*ZdDsLXYM>2#`DifB&u9
zloFMs93Ct0%+%KcgcR8wdE@)}b*P~foz%G;qp)v{kBafhTsZP{a<o%hQIz{#{$n#1
zF}hm<$3KY2y^Pg=D^rE@h|rL=+ge)OQ>yjv5aT7=^xCh*N(V_Q0doY6F`3%spR39%
zn<v`@lHR3$<Df`cv!)dsd=g@G$GvBb(g8!kjST%R2bGU1yHBgIj^Ft%zE?a-c&ga<
zLeis05=#uM>mYGxgJvB6f${4UitmZR<|_+Rqkx-{)W5V7GTnj(E3xRHtV=_^PC7+c
zLKN^|XcK4eyl8fT#s|#u!rYB#SYcM?dZ2ucGwwEbG<^2jwK*oH0jkF@DWiD;_mD8U
z3#O`rIf?AH{)=ISg5KW!bBrRwElf`?+`IRHqe06set*{6&G`!KvJEL{TqQdsE5-R6
zlpf94nwhA_wC1{Ugr%b4u<GH%2kpk^4;6i~?>%L%d-7Bu|EVX<zrK0%nUCc5cI1{h
ze6kzi{{R}j^`YwZ<3o~Z^FP=FBV?v`zu@b+bZ*^yz53LKstZi~MFBUj72jZ|9`U#J
z3A=o_S<0?-W_GsBNf<C|O5!7cX>yCsSJ0_pxL7n$Zf&Intjdj7*t$lLlxzWPE}ebu
zdpsiEQqzn1J&o1pY`8P(CC_nser)KY=Aq!1CW}qu+4g50XcvcHXSL}$a8X^Ok$bNt
zkeZf}>TAE1lS$q!GuKCJ`{GTD2#X;5GakGKi+n5RP{(t5&ng$##HCmra@zO0=9qM*
z&5sUiOt)-A=)-@?Z>FPvhAtY&joa;WA+v#mcm^Gg%Z_PE7GO>KiZly>l@8t;MO)24
zBr=wHRLzxk$?5zY&=jpAN6iWDE$c5(M!uoeMXAKXVIISXKIWS@XWqrG2SOKDHD~Bk
zdIq?T@8ie&cUFiy+Gt^lTbF}8M@uQlNk9R1=zEJ1SBkPbwEOlgz4|H(HXKmwScP81
zCm&;R;qZxS!SYujyP~8@Vdr-iW&(cOK0b|Cb&^jfbl=i}htx;;UF4vbb)~#ffX8t%
z<PF=CECNG93Q+u44H~horKf+kv}fbSaJAw+BL(ChP#bCZ?6LiH-(C_o?wwL3jji=e
zJl4nk222_idw1nOx!tv3JetUvAk!qfG`hx9>c`&=Y}!}MD6V*d-w|&>_b*>*$KLFF
z?EO^3rt59d(fcNh3Gzp)=stb8ENC<SqK6ilqyJKp?Wf01PvrXS3&A#QDzp;2cI#Fn
zKqb%$R@2aI_XTFhV7qVPi1xmUrY1|cqVGD(Hhe`^wj&cz_}1T#pFNAsQOGST1f-MT
zn()Y}-xx@R__Yik?P*2W)x&Kdp&K4nPm)NL)L4Q<RKM{;p+cODC?)4Sn!~vN*0uBS
z@>-!bCOyVGc~iu?C!#8!<j9x&+4kN*ZKR;9NLm5|89zY@@F^{a<Mm60>mEFGDd><T
zxFX$A+a~UnbYfql6%1hO#^W<PjgSHRF73f+tA%q~h(&%Plw11xWO75JK+o-7CTp^@
zX`-m<=Pr<dTzcZy*S|j&?+`Z|ySkYn^ziFKzNMK0gAIuu4uMt$l_rhS{7-1=qNM5j
z$6GgdKMT{Ue%3$6b;n-qcl+V)Ml&@9gKx5}=h9yvWLuZG!jyhT^V#qGqE8Po9yLan
zZI|%TAHBUxV7aXpfyrb%=h9&(Yd%g+PV_c31wjwkw5cxfuL>4vAvPWRa<rhkMcAaD
z8>@`Fr9gVY_33jS4or?V_VWCuTyx#oUx~<b>dd_Vf+p=8MY1074PB!nT5;5H^8auj
zw?~qj5Uh0@4UY&2sGWR)5l;;_idqLTZ$R!c_$g41-FMdnNJ+>Tak5e9W?qukLn|>#
zVD}{i%E^%*I4jVW9R@aJuqKJjez^XQws9Ofd9-I$(_TP!pKvi0rqc+d%nY2W>E|cE
zJweNy2adS$H9L%V{pNy%T`rT%Tu`#T&nzNB4D}PZJsntBpIo_nq#48(jxbJ^5i#$5
z99Q^3BVGiQ(`Z(TH!Keu??c*yP^>T*P#XH|YXnuXOW98De?%i+kAGb!9|>Qc38T1-
zoWC=pCw(|2hKwt%uF*ZA`L5oQZD-uwxqy4Yjdyp7+H8Z1d|sB*Ov<?U2OcG!PK$Z5
z(xltbNfL$1ON9>Sra1;OKDNtVuU_@E%SxrS2wWa?K@#oA{L%4wZLHV8L8XVU_8E1)
z*3=Y4KZ1Q?9q+s7mURmyU-oZxQ`~L(!_{6Wd}hYFOt*g>u%|2Y$3h;Brx1(@fX`|X
zMga>wAcATm3RI0R@A_?{utu(4w@t68HS`N?dKI#57k}rX3v6vYg98dZ22h8IW&sF*
z{A4h3=C5FkIqrj=ZX%~3CYQ-0k0rA`3an3Xb8{o_KXBk1o&>Tac_Tz1w5?6vSj6(+
zrN9m$jvH851W`1-4nS@LD!8!lqPNH*z%bgiYoFLpoH6T9A;eH_zImVq_c!dd#XbNo
zk8D~B_a6u=32h0~<?jvaQ^^n=0p}B(oWwW-MnoVi;HE?&HCiA0>04<zuNf#k@96!J
zg4gSe#KI1`(FDNUM>f43?vNntjP^msH)#3h9cDS2Lu&N9h|Dt~3WD|l@Rj{L*MQS=
zSY%gOWmmPtn2ht+;ypWX@sE+CXopuI4LuVp`LV$LyVm6&O}njmRB9bAw2xDUjfX;q
zLIh?kJ`3yR7kL-WjObq(`<s2822b>)FKdco5=V0T-Y@gKeM^JR+Ix+Bqwa<jgcvJl
zEVN3RXbuT9Uf@u(&gL5{{#$m`BSHOjN>p-`pQ~T6OPl(LDcL49DG{+BLKhP%ludID
zgZxS|0z{;ffx{!PC!$jz-i?K2beNDW{*;xK1-47tKo&$=Ru&fQD&=Gf2;gA3(8t$N
zPXoWUiD;Lx-!Kd61{xXydtH5Ss3A)#a#~s(h>U}W5iW1}QsUfjVZ_IaaxbEz(Sp#!
zizHBRYBAe%)`c^VwnAc0Bv5x}91#fM<O)|++3YB^L4i8hV|b~^((hMwTzjwyeZ0$d
zVTb^(cwGH~I9PxTj<heG%I(3fc=$Oc5(g5f_l~Gg1dO|fhlP-vgV4e3C*m-tFwS<7
z34C+E0Tt>}K#~xk%%RKWR}N(o7to5R5jA=!w6NinJZBq#QH<o=*-WpAkYQhPZ!JLk
zP`Ozk5L`|&a5c;R@tN*Wrv7cTXMtkLsGS+ZJ7X6$7JP$sN|H$^_YJT+fUe$`y9Rvo
z9_7ss<nC}81#es_P|`m%6pXI|BuuyyB&6Ry4G7=^E&tu>eoS9YG|Ag@D)-ms(>O52
z&y0m@6KL`{*DErDWBsGAq6I};7Nw=HC~1a=t0$YZ8(co!YnZi)@c|;ojMWEkC*|u}
zn=TY-Epi`_8on+(@V4;by7^N>H?+I$(qf<X>KbQis($=3W*?G)oeG@~H8r)@Zr%}G
zK<s~Tlm$T2!*>`O9X+Of+ZIhBKzT~-U*IkTQTJ+*5AWU;qF8`A7g`Nv;IBbr`0ybS
z7;M;34d9kb?{Uv_^cW>x+ctx(?1Za_Xj1s$xds?7pg-f@#Yn0#;PqgM&mi+aLW349
zAR$2fDcFS|3$ex_g}7Q!wrOr^m~Cm%{#%JF*4FdkL!75XyNvl+goJhVjWasCr~~wt
zF6eFQ%*1P!_pbYd4<TqtfCeXw2LZ(eBE6G{6W~XIo`GZx_J<^gK@(yJ0|V1l223Ed
zWpR*mnCN>lpqa}Iwo`=7SML|#6UKX8tTD49vfzEY2MiFy{)*YZ=~I2ZIk4)1BbyPm
z?qEw=7^I@56(lmhlO#^~QAH&nWhL+Lg$2-`^j>gY764Pfj*-sPbQB-p(U40SI*An`
z!|))w63mkfx*?fL3Bxx$Fe?{%1mT41*S!E46Eqcne}9zjeeku<)2aG~KtfusCyHGh
zmup~)Rm3l!`ux~_EiK26xnF~s@qrTp{huAq3A0{P(Xrm|ttC9i+LKr^cv;oPpnKFw
zLQ+80{wJ<JxLqT-e`g_ogVpjy3h#Z?fd<y$wwUM&5MmyUsaq#o@M)NeCcsM)1F`S0
z?uW;(($fLH-Ygy%M|%ZXr9wjLoh~s=-Jh2(UvA0JfBJn3O#<RbOin;(F^d0fr?<$C
zb;646<Ht*l2B7=|@tQvOfAwbTg63{6JsAj*>?0#VS+vKsY8`5gD}Q#Z6PZEutjRQv
zj9BTt=Y~tFs*C|y<ZalPnvp@;84xr!zib|OS%9J{J)OK@z>_s611n2PuH$F|imU;T
zzbQsX_?ANLaq7nV^-)t&rnpZ(=TQ=W_<mju`V&wlxF8(}o6&fMwl~a|*o5K_giE==
zaw!ep5Hl|#4CS{$P}>VM^7_z44qs-Wl!H&paE-)Ue;)(5g#%?Tl6b&gwBjZm1^jp$
zx*`kw0*6=gV-L2g<ATRXe<A5DV8rFjY|AD_Q1EtaEOxn^B$N$?3eTuQUB+8cTg$ls
zp&QgZ3a`t87_QNzPt4S@B*&<z8mRpxc!1;DO!YS;z=PS3>{S<!&WpH9Jem*hr`dM$
zQ3>zkvYRJWrM0SiB!Y`XqAX2lw|s9_33~LX0)9~q4v9C6jfFrhLf6zl<#Pq5hk%RO
zemV+K9dNyxWBy497tos2ldqRS6VPgILf_Q1JNgNoA?Uhz#x4_m#9y%fUQQ*tZ(q<v
zCJuI(5W47)lyQw#)YTC>6{tHP(=?>pU5J=_VuVBCe)~JPm$EAPsyJOdg)swY)et@m
zrAJc{GDbdVIB`+7#_l}HqpsdY!hwLrrC^4ZFYWE<SJ=aM5fmAJOx~=})O#Jt#b&+g
zTty$C7IPdLNRM3|gIG25CXm?UPV0vFG9m~jcxZfs{F2%4e4<rB!MC277qNlNdzzb9
zhl&;*Si&>D<HfN8?c=G@&X+3JkwIhn^%IBVTbi1h5XSM|>=t{X{rnn%c_J6r)H^Ay
zL%m!;(p1gcmcsn-AB8;I@^Bte5M;!aOqx(PF6GR+<<Zh>o!*f{`RaOf`}rwx`jQPd
zdmBSbE!XEiv%jM0YVKXV+H~#frxvlyc4gI7S{65J=i53&=2OQm%Ra4t+8nCg%Qzx(
z%vP&z@Wt7|M<0IOF~H3OUFC1E?htEH(!hx8NoHmyR7_ZXAo^N})o0b?<f6*~-9mFk
zW{Ta_SAsJ$tj5vp<bzdN<K=4&Y1!GtvlqdhsAtgt1Z@Q-V*R`WKVTi|w9t$cl^_GY
zz7qiJ(Z=?Jqbp?JIL^_W<NfNQ9ON)wb`&k*wd>YBf&(---ay--@_hD88J8RJI^iPk
z!^IR(tZ+VJBrq)IPQd;Q>$v(TD{ORHT5>V34ikYe-cJ$yoOi40SHr`tzpD!$d1>Ki
z9vBwb2i>Row}#?c;!p+(l>8wJ-9J3=7J^WOV|)q*jE8~R;P2lvG^7ECF2qNW_l_Jt
zUgKQYsEE&jKzkd>8e$D12RPwEU<6`uczAeVCV^dr;e*jIxPW!#_sDw*3fr@N%5GwK
zKu!J)d^iM+chH=hV};;^uMIa-L~EEHN=iv#&<?U63koF}t1UQM(AC?!9qr@=<#f~_
zh>(mYZ+=9b0gR7R`!)Qt(QN#I&iA)(vS`Et?!@?mbHb5QH-o7~JRc&5-Oa!NWYB52
zJ)c)VU=HBS>e7jFq|f*SASGeoKm-<`)k%proD3y~yJ*@AoR2;duKpWwRj^Yrp0OPA
zpM99tL|6AuF2LRe)LUSK10?f<i3*&FadMB~8pFImI;=+05vAfr=dYp??>FpwxS5pW
z;$o=AWh_?FADLhsY<&B`T||2(SKGz0svyYkm}bXWolRleoO)+NJ57FDer@_Y=?lC3
zv@Bc6h94+GG@GpYl@ir2hIvg24dx5xsH>_my<pO?wPpD}`k=3^cJ^ecWb^I0V$5-p
zZnk*Yl4a0#L`%xRH~E`dd1Fg8zu+-JexcdR<xh*A|DNoP61no*?x3J@Vyw2*yR)S?
zJ|i99)$<Cb#4T7Ck9xNHHFR`1Jq(KBm{5kcWy~;rhOAi!VPknSE{VwBaGa>Uh?_h)
z9&>A8=BuDD(2I>zt}AudUwlZPj8(3)A3$OYeD7{P89sMKc=mkemx>C)SGg<ri<+4~
zJA(IenN9PWeU+l_g@Zd~E(^QqY(7okwec0|AF}rfb!-<X-*H@h=aSB8)yebmsn})E
zW)1LMD`1s5?SV2;e`5^-DZEJ~8^e710YyGsqF7iE!F0TUyuDv#JuG!EL|eB)+G&59
zF8urf>*F*`zfyEnn5On@K22rHgYlz3uNoS{I@jo(PeVfk@t=zsxUv@W9WiZi?K`yE
z{RxdmVSyzEUkj4WT)8r2&ysapsF={jEc5%E|9pkt516u16|_MA{q8-+Cx>(Wx&wj$
z#KBmdI;hIzei~)#*Nf?I+?XMI@=6z|jacYNm6KqQ5x%v;!U`EP#5bj~g2A0i?m}*>
z=-=%2w%;RTq@t?IQZ-A=1tEwp#;4HV8q0zR0}}rB+@`kOQ8%BzM?DCB$;YKZ*ui0&
z2d9cqiEp?|Q_mV`B~;3q*8BMD3hS=4*Q_!j8DKn=?qWaA`tJ5^Nckbs=enl*@THH<
zE)3El2D)Vpg;X+U0dL6s51)ieuN#(|T$Jq{t3t%6&Xg5vzU<?3qv}xj4~Mt62N<WF
z&|9qky#xv<H^nNPp@W;sdF`FutX|g5P>Po=s}hN}2u~&U26#O957o}bf4RQ>%WVwP
zVgEeOnQz&;hWx{wWsyDk>HD(Cnf&x$@deYG>?(3{a)5#FOocG_&1VF_!xhla<|ISp
zr3jItPq)oZ8`V3icLi4l#D=`s6PjJ7Yzo^CvA)A((Wt1#j$nBG)B7R2|3mZjzrDC0
z)JhEz6;NAlY$7gkTUmi6VETL7+lq=rVjeXBqZ29xVzFMDxD91q$+92NLGjC%!2_3N
zhk(2q=Z;hZ(RKEDEW23Yr+;5(ECL??wYT>FP2u5&@c=~RR9~I!!9WFKr#NkZVT*EQ
z$d}=_$i>A6ZVV$MW2pss)&-p>ut6CRCV#Mm@nhUvTn#eG@QwhZ8uMI=W#}Lp4u7z0
zeRWj}9y)q&nn<;D5`X+AX*t{-KPDB#MGTc9)J_>$lAG+pAJ4F%9jxv>BO{RiU46W4
z@y5IyGApHlLAttLsBK%jVMA708jCIQl{>zFIkM`&YakX7zy(vcWh*nCKW=XPthTIG
zAgo~E>V9G@;=E!xbRWNn@-rhnomG4t5aaZW!=Djojtkz?!wdGropq(fQGY@I@8wtO
zabNa@3uF~POQa`A_qeX=;(1s9@u?ax4iA0P*v>2{RPGPMJ%wt$>p%Vk%wbUBCfl+A
zY5jM<nyoYaXn-aP+}ZWVs_`dp_Aeg>0WXnkV`7C`bFxYqKJS~5Kc80{5DBoeI{hsY
z5)}}leY!*xv}bDn{MIl}c&G^kM5y8qQHU4OKNd0x>hLm5ikY8VuLKqDaq~ax806{c
zIn{Lq9^>;9{WltjZ%z1vKW{HM%xHs=y%y%?rwquKlsW(BPc%WLOi+lAl#-$*E|viz
z%ZuJqT~b0chptKp)&oG+otIyWdHJ=ru53L$(*&1f8XB4vWbMz$E?;(LY8b@IK~Gdg
z0hE3?<`mwBnRfa0-kn2C=PnpP$_)r9iTtK!H<7?Pe?M^e@LPD-w<lt6At^sd<iAu4
ze_n7OobrS4`9>6#$k$}%{PX_12YgIGBSG4WW8+`X<{_K$)~#qYVZECD5)$79$jcYw
z5ICk)Kv1{T6g)Rv2G<$%?YnZ<u<Y>eGWH=~?TjCsjFC$=ybgKUfwX<~)E4|(Jgm1^
z3^ZImSM7O+xE12@@9_U;9sUo7|Gp&PDzkE8UVHQhQX*9979W0Zm4DlUiy1&19+*V|
z)-lbtD9VB>0P6}kD<FA=V5|WBIS?J^XJ!KOMwuSnw=(Fjiw?@BP$+(1#=zyjE<j?F
zo)|1PoR?7ULtP^SeI9xZke(CN89-hhT7^KXp~2u_XP0bD1I8$OvI=B3{RGayw=j_h
zaq{MX78P~rOCBmJre(9kjB)Uhp!uRe&V$=KW<<hW3HKPJ1_8?+8qSMkABVf(V;grA
zv@{el_X~p!kJ2^p&+nfv|2`PMpB!m#!|!W>nQsGdh&RH+Hl2rp<doTwkl)~~B|jZ`
zejP0>83=(%)A53T)uIQXTcW?GicWVWmGkw7y<ISInC!3ra8?wJDk8wf1fUf7T{wRn
z5%qI*eFuh${-4*L`$}$0Ur7D%z%p_IBE@)X#bih&zrt#Myj!BI3Jl^7IT<k5QF3Cm
z7if4yqXs+`e}k#qxPO3eW1zLh`SWi9wtCo0mIGdc2zzvJ5IFHD)akVl{53XmA3D?p
zQ3~cU5-vJwGc*bmeKo=C=0#Y*Jngo&d^qW&yn|9&1eFLXDd_l=C(eycPme-?i3+?n
zSWXUuY(aNN&z1`_M)3#md;x-kKZEKN-6R3@)ltzwyJ7l92<>evOr`}U0K)g3+qWNh
zduylPECW`bX<CdfFxVvsML5CkSXn{OTZ9b&5EL{7g-=;f8UkTKocRoJ2@{{dm3;r<
zLm0m~QEs8GhwpZCYSi^RHoMDqTMiw`t@fPbU4bZc-ppq5FEBB9DWiziFHntN?ig>V
zqJitVJjIH<cVgS4`*w-sHvodkG-Doa=#VtkpF&>gdvhAQYyx!?7MP*61pY*TN?7=J
zrQp8s2_{A6k*5S#eL*>V>*h^D(I~PKCnJ#K^OE6Fb(0J!(CYv<6{^k_fTgf#$!MW~
z6Q8k@HEvb|En3JAtZ*g8U#6&}olSd!@d<i*FW`>^4C!&wV?RIWV79)Cfv=N2idD)7
z3KU4l80g&O`vPSnzl}F7viB=PxgVG}rBsWY+0$z1{#;|zq7n0Er_k*SrZ`#NNfUtZ
z%gvpz8gUi|D{y_IZ&Xwl;0gFMHo=1JTxNVVs$;m#fw<8CrPEfmT=>lq%c`Csj|2>f
z=ml;;R0+z(u>VBUj`Uxos}kCj;p$r3_N444kQdNNMNlZ?yWy6}Xqm*fMW>h7xcY?v
zngm#{Pzcqdqs=bGEWEgil;$*$OKb%u?aZ_^Vx;g}HXnQml*8%Ja1W@F1w#li#N^uU
zj&)4q<2<}N*5txoSd>Ei$=Q$i<E)1L;?s?#IzDqe1IDZ1XT3tka^F~~7FQPkz7J#X
z3N%SWNkAYEFoSDT2;ZISK=dsPexuog{=Fir2K@bb^jB9UB`0Gq2||d|SP-BCWj&k=
z@50c{aee`kET}zRbY`u_T7&}5`P|Z?p8~zwJuqqK=H}p6+1}BiW!{aQ2@Q2eYT~Lj
zIR#bzvwQL{r<g}Af8rSuF`d(~yyWA6_y-NG4@1*SZBvg0*!KU1TtrNb;|}^70Ktxn
zRXlv}wKvh>^(aJgb@4xkhKw3TjR7I;gStq1IBpIf3F8Bh2f^G7i%YqEyM|H<=G&%2
zVsUng6%|)OwpJUl1)~(4Ps517{J)3@4;FPi+?l!!6>XJ)Wof}b1b7EJS5Z+tJVug*
zkCvky$$Y`*;28mz9sq_+A|hzgss+1G!F<edw)+{R@u=2p?CkPEHC0y^<l}?bhWq&O
z<1oB|DFKCu*5ZRJQ<&iqC<yMU3$OZlY$;+}Kxw@RFd5NFjy2?zu2){!ct$OPJ%Ynv
zVJ_~Ex)){8HdwTamyQYDh*L?6gt^qzB=-;6TRN6D>y#wD+ZCE~r6*>BkuuzgzGlKt
zJTeQeWoR&+aRDhOFDFN@qXL%{f*8fV{o<GlS6M09sELqtLF<Fe)7J<UVw#yyRRewo
z1r|5@=g*%}BE#ED7amYpLXLnDO*7UL6!Yxu=$-@u-h#pfmI8?Gv!rdGNSX@6o$-h!
z6*tqt^=W*z^%;AOz7P!(NCex2!4A^W%Fyrn{#_2I4%UK0Bj>n~kY>6Lz@EHX`=C#X
zn5_bB5Dc7H72*IbsRQ5zQhM+HeO8Vm;U+?Ee*s$tufoF863sM=+qZ9<nWf(^M-c!^
zKr*f|IoTNUCiny#PQ)1{Zl(~m;+Uy%wNHGLIUJKPB9XfzZJpMq;vda6wM$2kZ4T*a
zeS)_DRro3vabZqg-sZ1eLRj~dJfaJ`th5NKIFn#ce}BBqG2D&vRx)U@Dr9fK{Gdc=
z;n8l9!<e+jXI%DEknNd3tRd$>VJw2GktNrVZk8`l5;*pW=x3O=2CRFXRwD~zGt7o5
zG$&!o&^U*FN7=`ZkCR@G)E=m^c}NDc%ndCaqnphklaG*;r;WQYI>ue|2FR<NWCZCA
z{ME{WuxDyxOb7ZbY<3@FRAxpkP^LgZf`LxhB!AD+=3j{%^>ao6Qv$pnm$&cUCgs&s
zZ>ZQgIEV%e9Tr<_>m3PIfNA5mK~#?Ty3=6$$-pFn+jb)<?W>Ef=yY02HYo+!3y#(*
zl(nm?pgF}n5{BrrltcE2{V(LZRtwJ$or?BBU?D~ewp@+qg_VtyF8Hr1-h^&I%>bt3
zrKR*!5p>8>H7ENL;-^042^wq^zbhepX*dJl`GD~NTwg+|X|(;x7IT|H2=7=<o=itm
zQ@k4ub`lLC1d!dun|*_Kn8NvWIc$4n#-kcLlXeZw5wr{tPpMa}`U16E%^bcVfs(<c
zfd5=Ff(~r$@M!(~d23!oaIiWYKn~=wqeHY}#R{KpXqBo!lU<+r>@S?kK94l<BBx+m
z+tj2KUr<wA<b(bK!=_CU8C$!Eg+Q_z10&;0d3m7SzU-LKotY=)h<+fX_i@!6pOMca
zgdIM5)G&7tlmI|A2CAw^S4QBp%glgDPsiX$Xr^+@+B(PJ8zgj0yPJQH1eA3rhv5T*
ziBN=-MZ6nQAbe>LFR!Kon>0boEn2<T7ngS{DL>BeqQ=0bM(meGZ{y}_HnzkA=kCKj
zA`zL|+;F61;K>Vp;U~*h#Dc?)CsKcr_#3@2Okzp&-|nF*?~qYbTPs_%*8vJAB!luH
zUz5Pz!cqJivVIGLi|uw6*tC>jBKAcrNT?rBRZ~*}18M&BJLm|z58a+T(((!44(t<R
zmN&`N_#;&Eh~3Zct~a}UM!j-1J8Ur;iJ0smQS~i{p9*ev%nCX{*}bjv2Jbft=g{&c
zTt~38Kl$I^7^Msb4dRT$%-FTEIS4mlbQ|<T0@546>qH$oza<|%)K{<S?bI7%(1=&4
zz(1IN$+;F+_))XaR<T_IdnLiuhwZaLZWKsSEDVrn#vuN;@?O`~Jv#eM1hM9qx#hsP
z9?HGeQPR6<RaUxartt#8UXAak5Tet25x{*s<6aB)SlSF0*Vg_7c|#9Vw&ZNUK0vhJ
zu!k?)r1ZPpP)uR9>jF7ytqJ1doQ@9!W*;H^V`KaVAbfZbokKCR(_fME!v5K_O++&%
zFHhVbI+YzataiQsep&dC{8kW;KtzzkxV=o!v0hJc#0l~bJhx}hxGQ~bnApZR+@fIi
z42x=tg$^x@>du~=)G<yTc3x|qUqHcRs#F-)LZ}-(j;sQQtwDg*6{QgG_KfD-k45lt
z-g<~s(=|7t`-R9Yp&R+NY0*NK+%<$wq&@d0B-97)A1qfs#LZszA)WTp^Y@W(%X0)N
zK<*kiA-IDwjYKt1Ce~~+A7G6$l#zpIa>}pqNo9Wg(<e_3sHSvpxvXPtGx)ZnLl|yA
zAU=q9!(kYmFtq92?i)XvObeb8*gi{=&FjlfwLx&%`HeQd&oN?C7=IJ9aecV8(S$=r
z0~s$!OzLWC<uva()iOT5X;*v_7|0ShjaDjwZPKPqjL?2pP*ZEDt20CIOU>jtKE>>+
z$AyK3KYkdR4T!yi2-zUlQ1yhQSiraL<ifi*HPCDGEyq^DWFJI!;i};6gCvXq7CKD8
zjPQ*QA5k;Q?acuDRlU&wq$*?r#10$HI80h%pjrP=^)hkQ^XbNNY)Q3>dWgiF^fImy
zwWq0QJMF84bGy1(!FzxysoHb!^S&0Z9JbON3h3=l_5`5IuwF!&^;2~<YA!?o(+G~O
zoi~oX!6$IeUEVxLa9MCq`y68tvQk1X51}AY2_D;HNN5hL6Y2bvP+3)#AA5FYCP+Jw
zTrgFjv8I7n9u~a%9HN9r;~CNx78y5Z5q}ztPeRmEw&FSHEYGGMNpBN;|GL2Q15#5k
zK@jo8yK!E^Lhv(Wco7@{!!NU@a;jHQ9p@!mMW&*p<RDZ>?Z58e+48?lQA{09j1Oin
z0;^8IK~H&C0E;q)Pgm_GGZ%ciAwzff@UVw0yTMom<Q38bLYu4k_0#XsF);=rXA~9D
z<GyKwi*N)2KeZ0<UQi|sG&BSke}0zz1exM6V4P(p7x`E;8@^;3KI8HHA3v6kv9V^v
zSy!lX8A0fPNs12z#RA&s3=&A$@uPwvx<Yc3Co)han<A@;)rwb-8AEmmxlb;}NafKX
zttI4V@z)9{h|q6O0exuy{&1wdfOimP(9zig_Y?3b_no#rj($6(|I)8C%&0A-sev&?
zyxm)UP_=N)4G<5&w+{`~MD&Cp3FY4k?Ayk2>>VUJO)u%+-Ay}YdLuP6vmHkwR8D^7
z@acpRGLnIaTY@_CiOE7K$j>T0b1#0TqFR%RuxGy-)ecxnEroX3KVHbLhQER@wK0Sz
zmtzXWpK&^%&x+#5+Qz1(wRLW0MlzU(11*F0<0_N}hY%A~1CSA8D%e(fQ5Me`_%Fm8
zHg$A3LP`V5111(!l$R47KjEg~FouoLWyk%z=kS5C@{Oi#a4E&#gL(-sDV*C^Fk$24
z#~gr<KqDY*WHMZY$1%ui|MA_3#}Eowg^J{)q%4#6AnhHq7j9J-wW@s=prLy6W%-=a
zC9#yz1Q`>%Y7rP^ASBx?59sOvLl2W$6LC`Z<<;Zhxg`7|$`gqLeA%^&j32p@4mW7z
zMaEExW#Usr4lTeH6I428@pbR+${sUyd6ktl0?9sr+%|3F;X9!#sjez5gA=cag}DGl
zJkuV3inne^e}V=dZfS#}dYr*1n)S7{%b@c_;XKjtr4p<RsI}xrt3-Z}m4ZVsm_3aD
z#T>HPxk<yqX)TQ%ljE4X{`;SojE2Av7LOs^1Rn+VnP<lv5qeRYA+Bw0mBvm)-iCrF
zxxOP-Dmcbuz{(iYP%SYO;(~s02=@9s$ZK=PaFnY@iV4q47`FCjd_(L8ps03MY;5QJ
zIFGK*Vq!AW@)lWszYL$y&_YyWSVaAOGd4I4$u5PF^|?^c^noVDtKSec?*dL(XE*#;
zpkx+!S|q>Le*GI<1y?0lv=Wck(J#4C(<7_uk4tGqz%p64DyR`f1wNhL)8q%Rg4nPD
z!qwQkvYF0HVub}&A%RN<hVFx~7e95NP!h=&lZK8?4|X^zAPn!cLpVf{fXS-StM@e6
z%#1~Vt^fiAHr!{BsxXDtpH-0+99iTH=w$<CX`Vm9iHsQhsGxU|51IbS1>iQE8f-px
z{CFz7!x0I|gwG~{FExaD2faQ}gM4VS0IH&4Ha0%aDj(wkOcYL~*ROv^sDayDCN?q1
zVHZaogWjTI!iYd%JV{pTaBSc+B2hOxbLI?gTySXnaP;@|Aew~Y!1ndEz)})U1pMCh
zDeXR`anG^?PyPDmj<VgQ<X2~R7w0H>gug{e`v0!u_wW8JS->;@Wm?i+nH}oV*l<X0
zsE06B=6Dt&8sP>nXJGL8`%bzEm4t;(5=u3bC4YyO=eOQX(;PpidL{ik?}xz*_r+O0
zdU|i%#{Gr-R8$*3VZc}Vm)6#xHuMi<y5YU9YNA?R*Ho%>TbJsW@3SYdgGc@|q2RwZ
zOCWD89bM+kOIT_I<Ownn-tW65H6(5Ef~&(ff_N$aI}>CVIJZ5)zj<(O?5f%-`ZIUV
zkBy0`Ep^NZhIPo!DOab3#Lq|Xc42%Md`E5KDrr{MD8Z12kfv3>_xJN7&61y_=@~^7
zLXEO%(<W3yUXLExK~@0gbvL(*ID$ajti|brB01<>@cS`TkjwnikEpd_GJ$Emu;n2P
zC`4?;Hz)%VFo`_18r7U3Siwk%Rxddqh{1>rYGwAMl|4hy*cL4<+F{5jP=T(lE+nZC
zyTHt47UdrWZwXoh!f4FMAsLTNr_U%;7u9az?{gzWmOzRy@qz=plzE09BSR6b!A-iV
z1{Nr0Qutga_9)j)O^>PCfcyp}j{=7kCRKw&muqHFWK5<jAXxg*BKd)FD`oT~tuinm
zn8zqLCu8siADl$#h=B%p;euerqda8uKZ<kl@L|m7q~Ohv0275D*fIv_Usg$b*2bJ5
z#Q-5R0C9G7e0|*pSgC=*0M6j);Wl0r1vuRtV7Kz$*n87>EZcT%yzXXcCd$-AWGqR9
zQb|ae=P@KhWD1$ONh%~`hD-^WWynlQ#zdKiC{t0H=kdQC?&n$WTK^C0>-&CqJip(%
zTZZep&htEu{n+<y-?nWBzWO>ny}o!1JSOB7IJOgusdnx9T;yO|m4bF0I(u0#2DPsu
zj^yCyL!f=L<QG3I$P7Uc#OT>AvEwMvz@wi;dJSgvKiuFx6oJCR!qQ(PqH-HV;9$sl
z>9G_cnt^+^W8uMYlKdx~E%n?lid*j*Bt@&OqjVoj92mw0_??Y)gg#B)b`y(#U|_aR
z>0clG@WMWjGPc}-2T@>qY1mOtPR;`dbbhzKmi(UP2iXr?<1=BG2Q@5<Q3Ea7BJwDZ
zwVmh0*j_48)6vmUQCVXq>K{7fU;ALBt}mqd3YZ;xAXbBv&z&wxAT@XZNN;0}Ph8Eu
zp`xM!z0O7Fqca;j?Yr+?xF<?`gj~WwO6QoPmaOMS*PO!?$a7;m85x&Rl6>$3bexx)
z`|~qQwp)R?Y-{8W4#&qrmNf;uTdldoagJEp?Ag70eU1?&aC4v<SJ0C8ZHG=_5g^HA
z#~Xm}Pzfq>0CdrbZhOW51&aq=T`-f4?W@$yi*PGk|G^Fc>p-A@BGwLrf6@8aKn+4G
zG-NCA%n5x}bI!ux6+}hfy$wg{;F3vSx^9gPQn(N2;G#0rSVIO*q-7(w!V}z^C(k2~
zMIwYzrdPv|8am;gx^|q*#m+<Xq}%Qfx>{3c7O_cgO)SX-9azjLJs7?$3Xs5*)}iHs
zo7|6IM$QJ_cLTVi9qsL;_l0~AX#H29UBY1Dx`v`khRwtE0g7`D8!4e-=(1bgK2!(n
z)<mtHI~_#}k0hD>w*kF?V!nUy>o?*V0D2^Ko;J@J<QoL~H=cv}Ib=FP5fRRS%n7Pz
z)8&ua7^COKRl+vfVn4&!Pa&J=6yQmOoAkuT2T}m&k>=ng62z^&jAqnO<1#$Z9xorn
z*aXZ=mUw*x74}PE%M9nGyK@?ixG22!FF38;(9=7tnujb03HBWF^nPZ1wl8|sN22sk
zm|`uDZseWyj;Ms7Yn_5qdhfbO8Kzj5Ek-XQDcwVOeofJ@p_;98S&mT{?{XNSjx7bS
zgN%*o^zA5C@3*vg0DW7H26dq?OxzQImX0MN%!aej{1bm3i0R0D9=U8_b!Z8yUkTgD
z57>Wl(mM7Xu}FaDAI3Rr^$d_NtoKYTXJBSFxX_>=cNcC99OefEfG}Mt1uS`zwgX<4
zJ2AG;=GTh!(dW;e<sdHl@gA24?$D!f8dckN3W{UaNhn6i%&ZM=5zsY^A#9BX&_`{1
zzY+3+fsdWA;f0L(g0Ibr>fMm7Ay=9U%tj3Ut_h&Xw9FiGJZ!yIBkE_@M@Ql~k17E0
z-W2*fg@r=sszIi@Lu57zG;qOe|Hw!eRGRP^1sZ$Sakd_Fve5TKvWWu(X>f$EAonFp
z%W*^;oZ`39OF)5v?SxuT`ISsG<jlaR2eHpkLuZ+%JUqGxY#U0B)9Cu4b&CmLKqW}%
z2J4%^mzy^68-2MLe}T8MASg1D3A|f`#(&MI2QrDgnEIH)O`Hm_at5MgGbFbQCKoVU
zJ4U@%SdQ#RbR7|8JzJl2qBn%OO4e9is4da`1R7oq|6-V#f{BjvXc%H7X#X?=rg4I_
z;2Y5C;x~&N=f1#46^#sR3@3P~rKDuf%ilrQ1xt}IynzRfGlKA*ieCYifN1*JTtsBT
zH#nG;X(OwlfdLu775ff+Ref}HB{Q?iFk39oQmi;Jd#wE=@qT3xu)LU1Ur}KQ<^|9L
z*&Ib|2UrutNiP7zhfc$HB?mDQ2avp%S9VMMHKy2YaO5}%Wu>OFF?$>imj&!!*GTwF
z2%#Bw>rF;_`sS@$lRyh8tq8{Lh9(2toQj7GocmojaYDWpcpPKpSA9r%bx5qau-4V%
z`pY|c?WNmX!yNe2Wb$3zQr^4K?AW*{ctAi{RjahNv^FhKJBGsBI50`EEjGhHT%IjV
zE-h25r26g@$5fm5eL5OO?T7i)Q-TZG_qu{?SUb{0OYZGo+r`y&_0H9*!K#*?RI#s*
z+nt<j&9n~IJa{Ou)0}1m64jgX+1}4GpM_^@o>zMpYxr#XYb%Y@Q9*5`%0xCnBZX+$
zq+=;v36oc;TaIzXYS7<ekmr?;EOB-gUY>vCa*aOM<^UgW)4c-T$Bt7(8=(6hfU!|)
z^8xS!4s~nF$UM~#CnC0AfS^H~$4P%#Gg91())z8UGh0b%z{*!NFW!XK!fD=mn=~xa
zlA>;4IrA~~0;lY+CC_S%8J2$Ux`acnuynwoYC!W922^<u)R)%Q{s}WLuBxs+HpxNV
zfB(miAL3$SF#k3wel`g^2{g=y)OfxXu#h%3usoBWayVxZnPY1cPHk=)$$_et;;S+C
z$^Pxp=dg|^#>WFrDJ)9LjCC5fwqR*x?_22^8G%om#I`@y!=0)&NjS3i(mw%yqo0Y-
z3U{Z7fy^;%3!3~>g-J=jAjtx3yHz;nfg-@L4`!hJ4t~Yy7CdA@9Lh2<kLJB1JF)Wm
z3#$)8lf+?L;yPorh;#pO3>(%X7WyBemK45p@gh~b+V$%%hO`S?TUFew6{Tu|*j!>Y
z&V`-j5L3VapgCeh`m*5-P^tj)Gtj^_QN-mF?b|LPs+n)1`h8EYjmn%Hpoeg#ujEmw
zTYkI=L-(EJ;;j5{F7nQiFWTRKe12tHSXrm1?AMnchW6SNCuA;F-)Va0=rWm7oOQQ|
zdP-!ew*mT#{(knAukwLGfo}%t_-slNlJX{o?!GVx&+`~4{&bhK?QfUe%VX-jNAu*}
zm>%XcHg8UOJ^yj&xzJ$PK*%P;_Ee*!86(+6e#al9V64Z^ozuEesNsCwKBYhVzGevZ
zQ3Zu1hCx!yUM-fLs4%#3<Idn0F_GimiHLRB*hUQsumQ(?Y2wsEt~UXX1!A;6&#7@s
z4QDHGi=ea72gno|#Fxj~gO`u>Y;l3R8JK4XIss`G7GpT=kQyGt$P_C>{f&29&?4F7
z;^I=v`w}q{j(Z5DlEwM>4bPPmmC0F&Ar`UjL0R#W_m;4VrWQi4atRANVHp8Q#^Jz`
zH4aW74yL`_TFG+^%6ynI4~<2!Gv|_6E%IIbgvw(bE4<j|(ZT@4h=`?H>tK%)3TO+l
zu>al08=h~dvLS|Us3mO#2!NhmFDBi?_XR)((ft9;`cn9ZP!rTzlT~@Es{Q(LIeY5k
zN`d{c1;WC4zkZe!nrwi@Ap|Sb$1)a!4=FmRiFNHam#2SuTL4Tz_mgJv_s|f=&TEIC
zQeOORjbTdb?tHwrnQ=4Y$ws!gQn%7od!JpwilZ{`$UiTP>E1FbimMBq8IpRrpHqJ?
zx;D3CN~H?8ZVfage=7+UXg%$9*RV07)ImckZ+~$b%zaqG8Z16Ne`59f)^BF*ltHbD
zmbE2<9Rf+kM+VMuaB?VHYadVjaGHH9dldy|p}HwuR!Y9QO?W6wcMW*s-#;HM{_cH@
zYTWxtX81&(l7G%*x7?@K8nV+j70eYx%pj-&Bm%5?2{#i=Gy;_%mBIdl*hWQlX~pC^
zasV7O>{$uKlM!1Tx<9N6l3e4cL_ty)euf{z&g|uplBgEiz24<hu^H?G$lJVJBmeyV
z%_1F&9@A@>V&T+>bO7KGvRVVw%hS_eO8O)zZQkK8_wC=W#*=#(JDi!hA2A<iB>{6p
z-AVm}6wG}({^v+L$VMW-?p%XA%FAy6CPgU*nl!o)<Sg;^9bd494i4xS<~42oqxczP
z8x#^=)|;C&4WiwKuI%+HfSlCtH-y6t4CV_S9v&to5-9V~zCrB~yTk(O2F8B{4hTYE
z^Bj}ljP22;Mhr@;rCGKlO&^!vUu`w0TH4=^MSHfg9=h&X`k3!?tHb-$R_d5?v%*V_
z7yWPcc&i+4wu>p-zsJOD_v@$^pJ~-!T#{`>D_>^)hP4R`%cXFoxO3xKO6JzZVx|?9
zx(ES>J9~o-BL!lSZnu|Mg$ACFMxT)hLp*_aLt>7mi3QS2WVQy#tbkq<sc(BvXChkT
zN4S!eQbP6n5#E1&aTRDknSv*Z<69*q9q9}3ZXEYZNEKnM{$prp2rgw98-?Bh+V*23
z0W26Mh*pWxvMo4A$R$r?yzlAmE)4ukA>8CW4M}u#OW7sSooa7P6X;ED_m+EO(2D$>
zPZu{cS{r$Pi`Ln&BhH(Cz3Zll^u%w`55>3tReGWHc+#I_CB;egx?|c+!ig*KGcH#;
zZbY!v#(UWr-k%oimkNq(QtGJ-T}x?qxItZdL^ji7S}=kuB5c8&Yxsb$rGWM+*V>HX
znoEUYb@nn-PiD@XdGMQcP~pT%Vvx+MaSOCqIn>`^w;Pf<fyguoQ3qO7WOrCL@(fvN
zHB+QQu&}nK+`1LYNh91eBxaHUte<h=s12B4oV?Eh-m@!EE1hcSuwR<bJ|7K2Rc6LT
z%h2~YNo>rmu<8MtyB=t5Z^u-^G!QY6Dx%uPh+byNfYGtBeUMi{jk<zDA2(B6T>MYU
zKKpmKOidvVhLp+%sT9^#7QzNrG6a}_tCN+}x*$>k!Ke#~SYb67e{))!y%!q^j9hq-
z4DLjm5PGl=b#)N*Hh|Uh_3PKVItc%qpi`@)JC1>gm`pZz(czgFP~sv)%vaR=3*#;!
zMk(4JqgsCu$}LP<;7!g3*AJ=c+k5NKw;)5ReyP}FFXt<C<lC5_Ew1(*C#sG1z=c6%
z1BWmD98=ahMd5ykgX1?w(SDwc0}YOuYjS+tFIfZBI162zmhC6F25vTF9ofW<E;||=
zuaS@dfha=aj#z*y8%-KwMT|x)UK5DChK<B}qS#DysYcI?RbZ32HdYr-;=wb+F$d%r
z_D-S>8Fa|S$ur)_Mn%Evq|Q5DjiP*1W){kvOIS{Bzx#GDKDO*Tuv3VNwa9qt%VfZD
zlL@}0&g5H)>yGjo?W&i#UYr?RlIAjaJx;YzuABdmz!#b1>w}MZJ3gLFmhum~_KCm$
zs8Un5@$DI%_gbGjDHKIMA$;J>Lq)y&EZWbk{nZEXB=S3()&JROANtX@&<3Ngs~6mv
z{@%Y1f}}no3Qjd3m-vKGq$9w*(P@n-;yZlJ4oN+(7xOHW5|*HZTApmURxKuBy=B>J
zt`YuA<1Bm_L`x4jX+!QjtUy>>4s@dd3G43Sq7&s<jvw5yrBCLTlD-V*nv*aCL22=;
z;}ZY?z)W0RGvniGV3WZF5B+ttCvyNnRK==e6lOxpB^~@$2^uG$RN_Y{$<|xA<JP~D
z)<tKd4WNG+A7+e!rZR`;jXE94J%>iFjL=n*l_ONN6ggPl#h^JbG3_G6fcoQw%BE@p
zd>M^X6EtkbpCF_NRI!~$4&Lb8O_cHW4zG+7M{&$HZSjL<7u^)MAifA2<$^c!1KNZq
zIsBEk*npjeh9=$-F1HV^D85q5P8@%&T9UZW?c3kqM8}yatVeYcD@6~$+O*Bp;|0CI
z=Jh)^G`xSd<E+4<M9tHnY~@C?oZ_!#^^bOa-jzsw!s^7~7l#KfePl>?IyaGB`ztk3
zT~F2fLHEhub53<pdql7KMLz7HZ$8(%KjBjQzP!wiATxW-uNF2Qxvu_2{&Au%CklOk
zcm6i#cI>dwt59P%=F{$Qoc`SQ$Y}ST{SQvj8o#BNyGBzdA<Z0=nY++4EWSy^c>Y#H
zXpPC6xm4-V!EausQ{G-@=B$ln^W2NmSl!MC`uzG5>lsv(6d)yPsILzqiM)e6w44zq
z3?9cBB6cKs8KKg<@wU7v=mlj2Vvnw)35(G8P%y1nN2&)3be<lMY{zED$dZs5J6Myp
z)6g!1`Zh6O2G?cx2M-2jSiCG@cAmZ4!NEa#hSui$@G#6x&kkRPWss3tx27i~<quuP
zNq~qDMzPt7bMuU#AgaD8Y<E~)l-TLbqS6~IULy^9V%P9v)RicwHNlU^wAV9HN#4&?
z0hHR&Enw6e$bY&hfe}ZnZr>*IUBC^H7C4Vd^NexnG#@^5#sNkP#uL|1O8d)(83vf=
z8J|Hk=qtNGXB`j`0YOES;3GMSXc@u<wXr5AZT8E9lZWy~^Je*rMrFfwja%>BPt@)x
z({o8)Tk6KOnUO-{->ZPt-qqs}-9M|~!%-Rs=WZ+W#;zZ1b-Y$5m`d(C{vLk4Wr0OV
z+WFY#^OBdUL#<3gQ*NEo?Y_^Hwm93;P|_pU`z7;>6_o(nJ~sYdbF;j{*byHmMceDO
z4Y@C_vS$yP%#G9-GKD*x`F=-B`3HTN)giS*k2G>LG_!ZNbya#@;0+4bnm#Zns&rLJ
z!?^QPW%vc#ZPX*p#^xqxUv(Y$8dl4jKlbzb^br5~;Uk@6E1y@srMPV|+R~QShNYKW
zlzwNic&x`;<0yYv@(6o%lnl&GDfhkTS8WO(nSi;eX)6zBSQLP$V@K~2V+d`So7y?7
ztOPM3^K=-$cxXc^&7O$coB=u2K;+P3d>I>eJW}oA#rxw=ev$m6ND8PL9Gq$#IEs&~
z{L)^LySiB|$1Hq0Zk%~JxEfP@pU6xRFX=^z%1qO?6HU?V`2cC7az`<8T881RUpyLX
zehWVbb184(6gUeySN8d;f6Psnl`RE7f|C59`zVgq<ma6)CxI7K<?X!kXg4z)j|zb$
z0?9@;kMIGc7kJTIC`B`zNddKd&-K8|xH!q?#!0E@{6Ei*H67eRmW5ZII|v)cA@&}l
zeU63t%Gtxb@zrN=mMNX7`TVpW)e5VW?|=Dn(Y(g~{TIPRam;lIi@FwfX*6o9>!#>6
z8+WigB8i23=DTmvmE<J2ur?ql@YK{>1&cSAjSTxfQ5z~1-i@*0j>#&^+)<ZE9mJ*-
zry=nuueF@);QXya${y=-jx6ij2aPJQ`~L8L?qsdh8Ck$RY*w75s?m1uvoV9C;KQwk
z^brS_*SomPE+jY%6~M1(P?%jqEEP}iWbLlX-od$&$QsAA0O0nGTG3*jD@$vt(l<R#
zAn5H#DWD;Uy9qO==)qFb@$Xw&UNiJ}Ir<^+2l&2*XdJ3+fAh3}zhi=?ETqp*(U~$<
z?W4cT28$6?rA=DQtgM%3_C|;}CFWmE@zejAoS;z=XaO_J1Pxhr3HscvD~rb+&||Qh
z+YPRBllk351(#Ff=yKo8UolccJ_0ql69y`xpsw)R=bv*w-OcR+jw2Y7yheI}bn;Q%
zy#t$3o69ey#W7JzH%v?@x$Tx;s@8pSO8P2ZDFGP?3NCn4A>9BWuobKoi*QumiQ(V{
zR1!pUP8-s39Ec|K?{wBwWg0hgGb){*syNp`bLx*mom_(4?hX%C`qdqM|1Vl0S|+lO
zSq!pEFMRgGBlK_`#Ad0JAMf7R_Nh75C^il!v&Klta_dh$nxooY6kd;)Ppi$PiE}=p
zuDw!RGt~8}NPuE0yqKv(t~=!7&h^r&N3BYH)5CQA`WMWHX=s|cwYAfHZ4ET5=W`Rx
z<raVT$GmD~t(u2tm=gQz%-0TSFS>rwgoTwa{JlZ8Eu)>~VA3-6o*fsJF$8SeA8Lh`
zZ+6qaL$()XQef0LNcL474;FEL1Dmq4vH1fUEa@MEGW|;Sa1j?>oZxr-8@FWyx(ibG
zA>|@lIRA#pj#@bK0Q3%?IRGwG_K#{Cxeq(WgB?Bv*r|yZOBwvSj;VskjFdG)1&<sd
z%q&5!p(FNZQeis~>U`wLx;EtI{dkyL9*yJF*U(VOcv)s}iSOke^DySk=fU#P^31Hx
zJSJXTqlUrVzX3o~Y**5Dp9C!kokbCm6&HgqC)dg}JU=bnpZF&@M%urjF*iQ`2cTEz
zBgU6IfL+(u*CR20r0jt<nWCBFG?=;A&k4o8*pz(>Ibz&?R#xLh$Eg28yS%#Mfw#1b
zu;g5o8nx>t&7DG<H}Kl=oUq)vnej@sv75&s+SL%<Jh*Ajux?W$15fq`g%P=~xn>2u
z@n{tn;XqoY$WS&5w}#TIJuf)k(>;3Zf7Sb|RN=D_|G}FxJ~c99IoA{Z%>}UV=!tzC
z_(eqjn&O;a{|EQL8iglV2o-zRi@s9G_ZtrEmwi=UVWf8N97lKU&GDj*uTrn;jqYDR
z4k-1^?QiEiwy)3)g}@#H+7RiG6J%OAK*mAPE)dK3z9!1)EpG97@&w!waj+$dXL~$B
zct(K3<TvBCTNfNllmm}Ph8pWXk2y3`5|DYhvHhfJfX8szQPqlT;sACFfEgzS;ViC&
zn4#i5;&Pe08oE|?b`Q6L<D=N(|K%kfWY0UiQm}rCZB>rZE@Fwtq0FHB9kYnF)Ya-2
zH%T^BDK8115D#OCPZ&cdAt4s_FQzc_IhdM|U)PTJc=2MH<_bVHTQGQ7h6Y#sHGZn)
zdFUOCn?L$4-Tq->%=4^&yiT9N>-4_sJ|WtD<SAriULKWMe|Xd1-=g2<PxL&Qrr4U#
zq4LG%(OLlx(VeEZt~@ug4?1K)$IZ^aa_iWJysn3Kb9q8}k2R_a!?YIO=2P>RdaLjo
zCKZIKqo!gB7;#E>Ehw@^X^!%daqnKDuFJlSw$>k%Lm=!-j*a<he0!e-E>$~fF0f1o
zoiCf2r3M*d6Ii!dDtDifJ};!vtsv@%Ts2^8TVEUCn*N%h!|*z_baXD}FPA>QpnO_j
zZg%$ar5xjK`Xe)+{5vmR*?AujoJBN5B11>Gd9#hwFy7CX<Jt2)M^$M-zg_HjF-^y?
zf4}izF(Fy8o%e|=5<BZNJDQq~$Fkn4M+a24G%cOn7MzG;jTAVA6{var3ar%7Wh!-h
zLUD6cM&=OilIYI$H5+z_agxPMksLsIt0cwk*EU8u?GD%Y6Au2IzW%jUijk$qZrn1z
zD6t$>`QfwJfbvJ0xzk2-S7T+v7j!<dG((8ALJ@KS=@zs=D@eC6w-M~cSJ=jhIWCw>
zf);+P(^>PwAmDCVzn<0nU^7VdFfKx#O(qiHxIiHhkQ2cCY`?Gm$N0GCmlaWExQl#r
zqERRelRJOjbvkWtzIulKqahc}enRsxvF2~ToaF3I#ULRmT-ZRE#SOpLyc>qYGW8#v
zZ!mv4^!zX<Cw6|rm#n>kee>7--0k11T<PSjJyPO-c{og|5<Rd>bF+krG|UCttbP4!
zm|(8j)Rs1XqC4;z((&l}O*=M{%H1vd#GHFXWhwn*(@`Czw!HS{^WlxAez~LXdbjG$
zJh?S~%YT23$?W=tzNHSD>^f7YlWW_$9wud4>*$&2u?tXf30&|g%G=#A&a^L_yJ#=d
zQ}JlEv!Cvh($?uy$<T)jh27qkLXO4!S8euLnu~w~3<qgqL%Kl}O)mjZ-El`J%*@!S
z9wUGebCU7$&@m{o;^<q1v10(;O7nBtM1I!$7_U^kxKel7=g}hxDJfp(y7ltA(uALf
zhc7M8S!Lh)L<z(>4AoH5lN0du3FcZzleoGL7||mipJ!eM%;pg&tGXO*6PX=a(S!q>
z6*@>=JA^;*o1m%=T&Q>;4pQ+~xJIdkqga)A!zOTUuxO@HS%6PGEKaf5(H<i|v$u_}
z-?m^b30LYoqc+XYeMgjK$P6d_U-x23&k(va`!kv>%*;4BuPum#8KZ6f$`Pl&p~VK1
z+|}_NKWASI+$jE5`i)dVPs3wLGs}uKqtq?7@Xz!cyOh^qWBlFSOT~8b8*FLbkL(H!
zdv5;ni6KMPSL&`i4(hkYPK)mRlJUi_`=_suf2V!#j7N#jqR~sO@T^l_KK^&`u>Vja
zGl+NOD89jJ{<5z?%Uoao<#G}4Rzx?5EdUSbfCXsJjBe&!#3nbkoC-_Wf<S^4R4+Lr
zj|JKcfVFTFdeMpMf$ADKAwRhH)Rs`b!K{RA9<^s3PF-DHC}3%xK?rvtV;QID2Qu8L
zA;EJan|5)bU49uDpb<j$ej~5JZy@bg6r>*;vR!^cf5R0yfO49-rl~8Yq@=`7aSGMY
z<7eJ=kV!mOI)xrr_HL^zBTKyN9vbxiWGl{Bh42j=UM`agV2##LE4UqoT5BBgx%6ST
zLOi^9dK&|Hlt6A_!LWLV5ha^TlS|!necjHD>If<^0<90eAiJHB<6Xhavag;}*SH>8
zsib~N9Z^ir*E|#0=Xkitc*c(_jaim)jzjL?&*Pz|x%h^)Ud)>bNVA{fRDSVX-%0V#
zsXG^jc7!x4Tz?+*0XP1QUd49K?md`I1Oj>ROs;;FU(cV&05=a<8$$qdG8?1;>b)#Q
z?CE|(Uoyu(UfU;pH*5U$P)}Q2EWI_Tt{eWazX8ckz2X!iDe&!pUgef#)tFw?5ukbW
z2uK4B;8$D#C`%{en7ek<^v{E@nWm9Rm!0KBV0lKFfaJvN?XRI!#35s5#6h&{)b`$y
zxOS}@15MEKIeI1dKwgvtz-&9dr+B3MJF7q_!OOY;Tvz++7nW$6MxLAo@opkQz#yc%
zN{gvD<-j7f%6<oawC5%tB5VYm)YzTm_wg3Me|p{lKhwHJ>3cp}1*;}EF3*LkiIqlF
zo(zw&q1TkF_hve)6=F$<&fp1J3*E;1Rm(D>5rgJrXzmj}4Gec1-zlW>T-9xp+v;I<
zb3N(o$^^J=@_gF)BI|GEb=kGse&6~-1D5q~ogzAu%DGEw3TB@(gXTw;JIu^|%!Jf5
z=6=Uy&cAW#iI~4y0fT$?yUFz#$25*FTwgb|{WQ<`TUV~FLK;V!ucV)L_J|b*-TLgx
zoKKt~nvE|Dmp80m|B(N+oaWRw#M>{8pWwFSKeCO#DWz?qNY`w4Vnbpr|HSz`CK;H@
z_o>Vak)eP~4>D1op>v7u>;$9u#fzp;jaJYE{ynRd-X3@L5^O)eIOUv)MSI=T=Pm?`
zz>W|A2z-JzJHcRZY#>ALIrD?C^euVZL%{QuE+pw^JVvB|SxOXZu$2NL5A!52F8~YK
zqfLSxhk$iTH(#Cxz&C6GK$irs6P7!3Ab=uE0bYz{hwa^1;<9q@-kPfDO@8XOPSEZ2
zd@}_=inF{B)_9jLm4kT$V(1ilK978;H&qXqG-Sai@j%2|S~|K8uo}=J!Uup0B7z`#
zf03u7oezQ~B)4?E4YVK4W@n1cn|InChA3H8wC2i3faghpS?1kZCHxvX5ab>^eOhI%
z8}x^h`09=UHz%=hX8@#5<4L|c49Z%r9zxilZ~w;zrAi>Nkk#wDwp3s9FZ(9-HmlgQ
z__RznD~WhN)N6|;f5%@H3GW+;7}j?_T$9fo)@6J*@Pr~3i11?v%2a;4A5X(SFf_)v
z0`B?RVqJ7AE^DxdWh`#5G&`U0eCcP45$}nqNgE4m6be@C!%FXC%J1c`b1m2q8D?!J
z(Zb9pl$9c>tD4Ld{}~7dNJ)b~e#9#|gAa=S2kXdd=tc2}K2*Fh@9u(gojH3_R8$nY
zArEO@f@NO^&?Ly5U!me4j9^hM?!8;Il@9_+S`b`8Jm3?P#qPx2R9seeoas^s;AI1U
z2>9$yDyk!?R*>RU{R-M!?r1D3Dhe8XwdY8MH{G*{2oa!q(%dg)STVvF{H^7AKmy9@
z5A*;CR12s=C8+`^K+9Z|$hsD$m-H-ZVL4S>c}h_+RRhYND%2n?zh1%i8^;>qav1j`
zh4C-!e{};~kDffrI7JZdL1KU>@rn{VPE|5ANmaXj4TU@DzV!rO+XES)g#VF^n&2(;
zk!hxAH#AkFla{#Tc^%+1^AEG&1ZjVO?I<WC+BdqQ^@>_4Y<eBECy7Rpd{)pnlL0_#
zy7-B~uX`Veatn@$pvvrCe~_Ks1P9U9y&S7bpz}gowroLTB9ZP(igt-sgQb9yWj{>9
zHZwj(99JP&aByf7VSvPmjUFgbAItIk;$KEbHTep|S)?CX6e~#$@LvaXlh@W)RePkF
zS~i>4($F!nyw>sLBPRyBaq-H=y+bY6)BLm|-!$169yEDBd)f$`r%#oYKT$uxCXh=b
z_o<)XIJDk)TGHShgNs(ZEgmDa!Q<i_`wlFh9-O1DUhtpLI=a=FJBF=}4T(>V@_irF
z1sJP}nYtd^=;Dfw0^d~_?#hOt90X$JA3p38n(4zGBx&{JBs~VdB40sUe;m%F=#l}x
zC3ymp7$DS-A3m&1BLD9M=ac$0?X&$$6X=`T!n_az8$q_ldxB(!)k4f6ac_a8+m!4w
z?u#D)9<9MG#LH`iXNVWV1mZuCp@EYVh*lWtCiE`QK{&(1Gn$gqHPrgL7fo~hR<R^d
z=Zw1*kj<7U8|V=w_d*$y1fI>#I`R6k{*9EhI&c4;pNH4Bfr{%zXBZv;;Sb%r8$udg
zgqJDfbli~kl7$c?rHMGeV60NShaU(d;dSuZNXP-zYG?AsYUBe4=97p@C%TG(-FGm}
z3y8KV))a~&2%!W-L_~ObCy)w407FShiJ~L!#6Ck4lLDkx#1quGd4DRr+<^@FZegVK
zBxq5MuK-waJO^Dj&GjF%<kxObmZ~kwD(hwpwtRKpNcB72BYW!1zDB>rUlMZzzxg+r
zOK=FP-aamJwH;4`G+D4_fvOng^H5@ZUd_EBn1&%s`UW<S>j8-5E{?}_Bb7!ghBZ&{
z0jT<4p^lii^X%H?%d~>luhiUOKPK9(C<Hi#@LU9Y&LyK6k{H$&0)mrrSMbBa74(wu
z<(XuIk=R7hY(Y)ido~*-IR*%`(HtY%Y$61$kqQgM6wS!U$o9KAxWTBUPBls8=H+3}
zOLcGg3&1;0eWX_71GR!nfi&wONFT=Q`5Cgf9@vaZ0ij@OdKx5;*kNsFdU|@6E5SU_
z24qgx>L8Oqqx@N)U#mTwpx?cJ52<jh#;isG9uhsI2De^5pu82TfsOz!@Zqk*<6py0
zT*Fc!eymnIvSngUX`m2;d;g0!%V59Oz-$4JXc1$;`vVuj)j}L>9+&#j(~|*9dd{j`
zp}i%m`R-;h7CRmeQh;$>TqY*t_wldmOOe0SOPc139=E?aeYs9Da;Ko~&2>|r$%?*z
zyg7u6tnbzg-g(ph=IxVf-b_mqKeLigzSlZ`ko5+A#ICUY3!*VQOxK4k?%26rbZ3TD
zW{&pP<;aQsEph1woV{ZMxikr5ugO}u0ty*;1FZHcyjN8=Lyc_Gzr3Io1sUs?ovi_S
z)?iW9$HpaK&h{uaT;Q<;b8k~M#36Ly)VPcQpne6WK3wpKZY5!hE01MgB~CnB&?y#7
zh*D>64*ljvaNsJ2_gz>FG)0Jwb9#i%pt<>@CEF~kJ9oNJi@}1YjD6%WB$D4Tmpcrz
zy@yU*7#R5ZGZzmY0t?MM`V*M>4fY8-i%Yfq-6R#Ocvio>Cswm-CnZD7HE!-S968{}
z4J_PBy!dE0$jCURFuc4$uC}dh4r?GE^Dns=c*MgJy#%}^FFc?GA}Cp0QiArx_AOhW
zUe{J=Y4sBY_ZI6~QOfkzEr5rI4VjYf7Zo{OzFY|!=6JZck)a_!=xbPI17Ti!lE1C?
z<)W6dDOCRzgEza%oFpZSKiW6}AD@4G?NGruzwk7U0^Yv+^G$yC$3N@;9k^EC;V@==
z+BZhX&iD_TB#s(>jC8w0`S_~WsR6|d%Z$B?`QMhlj%eFGUwQFLVo)NC>()@#>jveD
z63Hu4A9IQqX_?}E{~2ouSyw(dD}W_~gK6R5iV%WbWB}SVYR9S^cQ{m_)7p>iGduSq
zaw15T6Ho#Yk$l?`5sW1md%+WW`pg+Le>qN`w8rCs=Lb$l2@no-Gp_>Im~U|br;JQq
zM5%~W7(L5i=uc44sAt_gj)G**_QEe_@a_d}h{OxIH}Iva($W~x4ieybG<^{Ijn46V
ztu(F|#Q{)?fQqQTd+**mID`xLOaXgC8yc<w{P-!Fr^tf?j|P8g0OrcX6#4TfF&zaj
z7e$&SGefVynKLUmkio?(Tw4JK7}B|AsJhkRXo9pGG+#GB@?ccpCG+z03FQ)@3m3qz
zYd%FLOdtikC?*Ckk4R9#5Myys+}g#E(j%{Xg&PGI0%-IbaI`c@Z~0IbroP2>G|gqH
z=Gzm#oAo-#KG9RpMXUw=hTI$+7oZiK2E$NT`0IxcvCtml_S(aS8o*`)$k|`btH5w4
zlY?O+0H*79__88-gxvSa6$}ot#-l+nY(jJ(7>;39Ra;!uwHS<g>Bc#@2x}2AYVZzr
zXpDB96joMN{>Pcx7DNwtlH|K6^GD-C3Z0kKvR^{7@2Bp*cNd8Cr$JsubcAg?>7wDG
z5G!!!rFQhi8X7Xu4I^}12oJTo>p7f}Mm$CH>B&b1A=*6zMiV<s4h+Upw+%&~9EF)p
zt7o2-FsYd9^ip^WOWCz=K4p<$@tOODOWs_~=QHAHM{#}?53iiQ^hrdp!^WzGk<Y<;
za!{ouDOKEJ#~Jm4*~ZzVy3_|>Db6_0<eulLwcPfDG5pNTPYoNNp<CU?3uf-NNqNfb
z^2t}&;^kEHRI%W%`A05ewA0z^%d<=0M^ps_y`8hE=s4e}zkmNeJJ~P~`~jpEVe!gA
zyXff`L4fTOzKY5A&_)?7tka|lzxD^+nsA{xCaG1kQP9jY6kHz1ff|^|`J)r!z5Yar
zdRbp#Z^GqPzqW+*6=np!d34AMBaiDn$Ss5?84`&TwsAXI1)oK~d->DR7>RjQrPkDy
zSFDj2*tamkZiZFhZF6j~RNjn1xB;a~3jdC7A<YSelyu}rSoNQ+H%}=3?W&8S(AfVT
z^b06kOcR!=q_HW`dcYxpREo1H#5L(=L^(L6uRb&Yl=V9su~FtF%0IUk_D#Ggha9$V
zVZ8Uw8;+hlU>FdejYMW}&v4wP<C3DVm=F{-v-Od}qK}!a;6m~0ki&j}IjKs!si^dQ
zBi67A;z&vT_a}GV6m|PB(<UlfCdbjg+p8-O$RQjL8oDM+2=7gq5OtblOPs40`z+)Q
zKpj0)8P0te@BpqDM-$T;F*uhRP}e3!7*V$o=TfS>eZBYodJl8?TeZ(nYp1|9MsA7C
zSm7C)1)@6$l+PWCsP~IaoZPr!10X>574QkhkgvD9!nxEM*QxZma41B{mpGT6i>_Es
zJSX}EOaM-Y^tZ_h?@NJTE2e3Ebq#0BeLP^)2_Xp>!yI3?t`Vcc`uzWWkTvIy65(}h
zfeh{K{cG`Am4$mPwy&j9`UnSz)vvw_BEv2aNa%fVfp>EGm#=?coo!sRW^b9ns{=jq
zP$lhMwZtR8`*a7<I+Cw*B19X!`Zc%efJXj_SAvF`eEsJa|M%PDQ&cKWiZm=)9$jTv
zwdKLwTr#Mlyd35@)t&ob?Y9KI?~$5@m8&j=ldnI9uTAzpBo#e!@WGE>&}Z<qO$0Jb
z6(7_AU}C5dH?F9+m9>?%-Zk0Q-1l4`&n5ozDDF8D73}KQ_<vbJR{kITA=9eu9e#e#
z(Ugk9+aaUaeu70rMC(%*zIUzw=no8g5`zFcARHJ4tvZpG0YeoQ5;`etlGa~|<^}?M
z6{RMulkWYEl(tC{C$88>m;lP$<IyI5ruh2Lium7G#D8|y|6B1y&wP-?vTsbO+9hBp
zL=Uz#HY#NMA)F&ck&tpj)rmIZap@oUC7qLW?&oy}Mab{FfY6a)(!wR5eZ$0L0X;D=
zz(Ny6@gv$tt6Q>}a9UYRQ*bgJA(xy?u^Ll<eSwo4k{S8zQ`SFa<oQT&_9vcQYjiYm
zNa9#D^ov7%Mw}!4;_ySAW2*})`ANdYY{=j83JStrvf=x||GZvO1F$#rA9_2E|B<uC
z_j$IgZgeGL`vB14(kVF{IzXg?U!EmDJ-GVQ=E_;NBye`wWMABu5Jc=>vxm5Ft<mu&
zhMAD*=&9)#D&pI}E|O6HR&rnFd>C?H;E-2Z5w-sP?(Brh;^xBu$q;~sEvw#_@~hED
zCzd34>ADx*KQYKShKSMh&l?m`N+luVMOok8x9XBAv3jheYau~}$Hf^Px)b^j0Z0hU
zlTX#VYC4*(g{<8e+P7ly<En@=Z$tn2w(&d*rhF$g=UV9Tu6mJ{<8;SGG~wp`{Pr!%
zVF`HW#Q*q-kdP2jh&wC+xeiGquAe-NtIx$Gej%WVxQgQ|dG&+&uEtZ9Ga#>nyQXQJ
z;g3&rZ|a}7_=8Frj&~vEe3%Xa51z_;CcGA{8!=-&w>GNgZ4NR`3}lmi@Co04OS}3E
zDei$qiSCI}8ZT~}{px+!TfKbNHE<&YcATvMdlp(AoqMLlbM;>7>PMiWY>!VyJ&8wO
z(@btYt=H<?l-9#L03Js-?*@g#PG=3pKhCBb+)`5hU>*Z=<T12KP^TLV6H`gb)k8b#
zZT-p3sep9AS$&$fdi|8-`tSV}0}cLa?}nG$;;1#RWm&0$hS!Y~CNNU|)syG#IKn=B
z;}uAe2u%lC*Bnhy^Xo8op_c>vSUsPnGO{T4%$jhFP{LTFvM)t=>t<Gd{uv~#LBm+>
z>4n%SYWZ91w@i=^O}<t$i&D+hh<|edR=@k7EaU%55_WFqdSXvS!GI?S-56%hz$3vL
z7!g`Twl+4|pCTa7RUX)gp44G`9n2Hc6~<#^4fxIZl`L7;vNrNX$`o`m?tssRcJ8N7
z%Rm`KRu(5f_gBxi#q(udL%X_HsZh8BFoK-OX|!}x(C}9LjAG4dkYdp>H!nef4{%Q{
z|1iFverI(XU3m|_E<O+};xeqo@qL}9)qN7|1yBuS1Ck;mW#U_Xk@nezg^opKlnCgy
zqwJqWC2E}5jj7!N1&9^*{_<EN%g{OyItMUSQSU!+cL$9M<)Wqrv}EEH&9F!T64r=D
z8HAtUdjNGq_YwlUvWg0@f|+4Dhp=++VChR(X|7olw3<erGlx7C(wPy+1h||NaXaQ1
zwE%t-eH@6B&_f$vLmb2xihmt6a*T|O$OJ$ThAXS_fv8wk6L+4n+Aw}*Vo;w&Z=k0)
z0arRgM8@y;=syo7ldBfTDlpvve6*8=F?jd%6RY=#QkO_Wm%+-F_5fpj7`P<Ng{G(P
zew4)+93iM^?yX%nIeZ9zk-U3#on1g7D4wwhx0TTCs_4)GPuOP12|XCVg<?>;H`0^4
zYO)$d&Y7dP1UzbnIkLUjsX*Tz;rvKT*q-&esPB+IvVrxDVRL9TLSPs#!`e7R#jUkS
z+G!>_#JWCt&+1!hf^(D%NEjW9(=N+}VPRpp=WpKp31ZbjuAJAeLtu-g9|n;DvgcNW
z51cThfr%%NPz9ovghnPF2$m)PI4558ZBOn83Pi42L`CalVwHkvYmy>O_`itj${NLh
zLtw`6um*NKK&|T&<d0^LqsxIAykZs0<OKweuO4>mfgKPZ6EM3ji*yi~`8bi<8Q%ss
zq30bwZxP1JZ?xHoKWTPdb>IFDPZgBs1$YNoD@j+O5f4H<E0>p#kB^sEqje=Zsy62M
zaBDW8F2>WP#7!YWjuPjTqEP%&ZaNl7j}zUvV!@qTPCLj(Qq=wDrIGmKpdrANv_m|F
z>o>6ddU~!WDo>7%@+iuO(mrJl$P*<C8Fe!L^BTX8GLeY#s|;%b*!14DSlR*uGVw!e
ziMb~H`SUEMpTt5UVq}*5*`a^@!%qRj0!j=R8A{WiZ~+9TjIT)p(l@jU%K8Bmoz!(f
zT5_c|6gsYpRpUUJjLyo)Z;&SD>r~RKi*6iO1@huuRQ<p;i8Y{~4JrtrHJn^%<JX_;
z`6vjyNFoRtV(EpyLyz4w#!Za6xtbY7MWTrUFkO6{4N_cC2Rv(ladC5DY+J@>ZFDlE
z7o9I+e$YYHt`}lkImL;^Z%d2vc*3h!tb44nHlxy4)2OH%v`ph-V<klRV6<tWi}pfp
zItcM-l(GzmutCP7mQjdS9L~BHYXyMXF)=Y(|Lr0sr=iF&FP^(Jo`56V=69=X$|%yi
zo>1HyqIiCi_Oqww1ONl#YnbM|e*g*$zcVk|+uP0Bvl1{lr6(T7m~0^rsn&l7^9)gd
z-+~4Yl;Xr{JbJKs`}XFnW1T>hn<PPI*}HEaTdDvaFOcCs+zNmi&|@G5dT7N(L_|E8
zQsM^A*$$++PaX!Bq#uA$DKgMJ!rT}{Oi+w!LwLz-Txe1JJZ?&AkmsRELA4$*SH|Qu
zEIRj2w2PV3-T_{0vaSRGo-)|}HTuG)>}ud=jS6%4C{Sm?X6+6LzQ6zcAut2}7^YL}
zH*JE}NH%)_o)p$2dBez&Q=4z#GkTfw)&cDX<^0BCJklcT&oFWD*C^FB>+7Z1NG>!u
z45ImK*Xoc>SH*yF$edaN&EMe*f3=NUxE{a@)3up`fPjX^aVC7m`mJ-huCA+z^Tx-X
zo_PP1K7vc_GS3pke-M|$O0?sS1KEZgd4Iz@IB_`kRqlcb2OfO=KlyVmFs``xcqtKP
zEio8++9LmkMxyTH$CEBCUV&h=A`}BE@9flY@&N|<8{;zpYSw-}j&0!j<e%M!nj0bO
z=$rzfB0=1t^h$;m-q5l*R-Qh6nw52uLAxKZ561)}8a%@buR||hz4{62qBe91vCDr2
zU$}DlvP#xXS>RbG|7^yCUqk2Kcw;*K6!{xiJH$GyU2{%%8Xi-~DlPI#N?dS;2CYLj
zg~3tWc_yV;zGZ9z%*x6wSRQXMd=b$OT{X-N!GQ;%qa-FhNlNB}iw`3ZAe}I=vVvfP
z)2;F3l!h(N0z{!EtqOphNU6~tf?Fg2bLEf|25=KYX1xSj6)-Y%8AlQ9LEt`%)dy_B
zYY_7UcvR44#J9``wGO!a`&r@|&Yc;z*s$S$W^DNOf3Tqc;2o=>pDcyBR`vXKJk1Cv
zoL`aXM@3?t!SnKAO#%2mIJoDC^-;1ad=pRsAV~|}J(v3zh?C1Xu?6_^$xOJoR5?D6
zjO<1P?jJ$Xl70i^Qx+!W%KyMi{|%45_ZQpg)2#y~Xo?#R4pVU=tIU{-$IVvHmtQk}
z90Il`ZFL+C=!!|%Qas(^>t?V^Xw8OQrRzRccYR?%u!BW_At)?~%!n&UrO+UdIe<or
zAktVER)dpz>eMvaUWlh$Xo}(3!70;(l-btSw&v7L7L<?MFiLC3juIFIp%4a76}eeK
zfdc?^0w1C?1twoja|m4Wt|O459<X1LmXZpC1`!84=pKQdj~@$y0|T=Ph*QZdIzPW2
zFwO@Dk&xA*N=6JzK~E|ORz@<rgx;<pGul}fsaQATBsWO7Hj3A;ZW)4%H&1)Jpc>4E
z$E_qJB{QsTA3sI$lC-oPQdkA~{?&85<qbG^5C*Gh#iF&*6EhoCn+W7`UQ=X)=+673
zMemS_gdXqM!1xV1b4PfTfL--DHj>4>{tB^#bez2nkDfnAF+;TcEGTf`Ics^tIJ+F!
ze@1C~IuGZjC$z`Ur!0et;q4Ifz<-gCX1y-B(<f$#%@A(jmt_gu1?ga(?@;xfnAyg1
zu}?>I{SJ5PQa7VrLN~!~J^aV;QUV+r_YNMM{NSi~?kHr?vfCAk9cEHv3)a<Ce*8#D
zK_SNp#HI|Y)yZ2h-WN+@ppwKmT#!g2PFwV#2b`rKh1<+)cok0-ldbg4>aSpaY5@)b
zGPsz+{k%ghehbaRv&W7}&XnV+p&W|v_t!gD0m@07(m3Xe)7D>&hG_&ZP6ae5NeMI8
zObH^7Qc63ry1v0MnbEYu4Mi(F&;<N+2V;1x7_lm>Z$hHpWOfDlVj}kY#*@^$cUz8h
z3GbQO+RVzy34N>C(W}qvUpwpT>qBe_^AQ7+`YQt<q{24F%ZrjXo18=F`O)1mMd5&D
zlcZ*eOwv(a)?E>9$0a2z=#VE7v+qkXGWA%V0LLMKL_gFKj1~ygQJ3{AUY3?-0J{q5
zJ4d5I)VV`;_l7YkuTi9Q-D=TM4}Ev4-C#d-m$c1lkT>G+^ctp0Zoz*ccu#3L2z5s=
z4EasBg}%r85wrG*>=MXYQI&iS-dk3ZiLO!kNtN3mEOBA8j1XB5Sgc5mMBp(*D#W{Y
zK_fuY@EQphicV5@JUw_}+!#hUm||QlAi15jeluefjq4_>19t!5VuBc2DIa#b5%xh4
zweRozhTF0-|BE|jVS|Y^YF8wyo;Bj$`>Cm^Nudf*hYSxdJYNkDg9{ff7~HI<M`}EV
z>d><WUMd1enV>y@Fc1z8#mGsCZU;PN?5#=T6(}IeT`v4Y0~tA{qSU*0?;d}%^GBV@
z{w(HAP7X{|exOtv`u6RBDk<a92AhRoJdhccm;<~X=@~XqoW`Sn@=cVwblv0iQrtS5
z!qX$&K45Z)sy!o!6MBQg>BllROky<dpB!`SruH45vyNoS0SsW?p)YY`E`hTOmga_6
zOaK|txVVB!p_oKS#1`3DLpAkL0)#fWI0Sl{_oGLjtt*k`0Mf%8<UsRY>{R1SF+ind
z-0@Bl1476LmChI=37Riblu!TWyd9F$M63Q{nEPu!sfKY$eYD#+>%Q}UnG}o<)wIuF
z>(kkVjYn7NHU`yth4XfISgGk;LPiw9qkEos#{=ZV^bIh8eD?8{->aGi>kXd&cpox=
z9VxPgrX~{V6|`P|W<QR%fnmlx%%8EQf5DOs<D=muP*_`tr$yW;LEie=6zY3)5uxs1
z1Co7wuHnWLxh80K*kWXgAl|-tCMBd#UY?Z-gH$oh3G)<eF=luK^&Td4f_|{$>>|&h
zL+87>_8kT&EhMCo0-;&)rBe>rBoO`5wLJ_a>2CO$9FWuS8%P*|GN3Aeq#69`79b0U
z4j=Y5J<G*qh9!l}5RF$dP87U`R;*L90Bl|BT0plNz;O8=KOm<kv+e-_p!JA;6<pS_
z9Eh3LrRn;f-Me=OG^`0kxw)4^B}(ZuN(lT~A~m6APhEN_u>T~PY7L*NG-G#s8pN6A
z0K47mqY>PZ2~ST=)#fZ=6Gy_ab*>xbJniDvpCcn_N*T90PKh}uiPNy)CMz$$D4O_m
z2G`Zn&q_xZ2F`y!e+c#!V5zVq&r9rOzHT~S2zMx%LUL4R$rsG-C|44b+&%i`{;qeM
z^6`-`7z_;!6$Iic?ZgOIGgKW<q=kVu0Gapu)7sHd2MU#xJkao}q_!y0b$`q!u>9pt
zIY83a6p!Q0dRDsH<%i{KjH14}U|bC4B_1T=@y8jCeOB+F%K)zf0MSb)t1;jtMX>6Y
z-^#Ms7(g0-m=Pi_^$b%V*gGD&0eyH?bl07NdR+Kew|b}R0Z^zf7it&i=H$Y})KD|-
zEhak+f@iAMJ~lD2hiMAsaG|XNMCesFFmuof>fb-Y2yQya(X(ySz_<WsqXzt!naR+g
zE#?*;_wPUKRz~DxW=G8k8Z>mj{3dP#Ja>Mpj9~(OLCX>E6;NRk<=McQCl(UG$H&-*
zEIrR0JSe*OV;Fc6%V6R#$7zvaZ~*0sHM&~1eBB1>o4qNfv*x;rr51`EwUnXV1?K$b
z@k`quyP}Eh(ia7}!uioC&v!IJJR+&104}a@A3hv|pjqv1R7Ri(IOb@EN@7kzcX3hN
zLKcofL@9d!NZUFJ+A}c92t19nB{MA;yMjETFfw%c#Q6QQF3X*Db=+9<(No*&5b209
z2pwHg5wGhbk!-|-JD7vDK*5Im$N2M$4_HheuBE7EmJgI%$vE-{T&5<;#!Dr?J~dA>
zx!A#o;3f0-)EE#f0SMs)TtV$Or}X6EL-=adq7ce<>A8+=N6I9eHhq~Wq!2~z=f-3E
zGDAW_n4BhMaZ^DAjapJf=0#{zZ!owPC;nZVx-0-d@LH0El$xJEr_lE=Y?6_VoDme;
zPCq^|{HY-0BzoQ<x=ASoajf9E)V4<nz*6mE3PwNH%<3LkmCa&@L14xvRXjI92Z5&>
zowcoP8eAdOqGxRj(_zQ>2~lXA@fi*~Ku32OQs=)80@-8v9h^YbQz^=$3D36!bEdu&
zjLBySuWmD=*SXPqe__nU6}aq<BCUq+1g_*WLOiTQzBDwD**5Tlf&ipYkOVat^o`#d
zit!Q22RRpg1W<AjQDzVfLH#DF4+a2N#}Ts!OoHbvX5xML<O#BZJ6(Ekox+qAK$x7^
z?d<IAFmAz$!{I@D{Q7-p^te(&hVCO83Y|VZ3+|B1@}fOR=WtEd*Y88%IJUy%V9Q9`
z&;-iQ4e>+wTXzR0CNl40K6z1gm%@HVhn>{fwqlVUa*C3kB5$S#U$h%1WhbertFtOf
zZOI-vn8KWK`SQi-k0U;F!V}?!!+vM{L(cg3EzO%)4s`Mw4j1K3gw%CfU3J?yPl5Qp
zA93K(65EpO0@538`+jPt);|LSd+F$i#~UIpWEWXkS;(vQi^Rj!1}=yGK|xB_(<F^n
zmX{#nSyo^_im?*<Lyxh+D^6?hD8u;(3||=dr8@h=s0DCkUGVV>ODv!_5d(<8U$6w2
zz^2;$?3Z;ujkcp~?T+P*V9E?)G7p1C<WK8-B%;KF0}~+J-Q5WkjZ_y%Oczu{2#=E(
zyQS}JriUDACymHlpTA!qEEqJdZ&fm{A0nfNCI+Pa-yyi-d}KgvAE>Nl^!bH5u*HA!
za!|W?;mb3&67U?|9PmItjQ&8Ah%DOb?Vc+x1k|rSRYTG1K8=Kcj1X%c00EMx(+n(S
zci8%w-1)S>+NAG*3Lh9!^#?3>(x0}wy;uo){?7G(lZe|Nv1C^3wy(O`PEm9xnt#mp
z3#ntB<?x%AX$UANSU?WKkqVx?$c3`O;bG^&FU%HjMG)LIQ#g5;0c6Y=^J?%Fv10<0
z;GZZ0h#GEN4<j>rwnHmt2LDc~xQn$PKO!c^M>AAyS08(&S?t)xlNt^B7}34_Zo=Q*
z$~AHX-HfGfzqmQ_6@LelV-lZykxmrv<`1AeM5clOP6?qR;AA;nC|@6u;r3J?TS9u=
zvyoXm99E}tvAA6SjhtHViy_HK^pP_Pp%h>vq(hV?N5cHB+%|2`lE}#DEEED8>5WPK
zkQaCKM&gzev8<hp907TtteHK6ba_}~ns)bXS5=|Di}4NEcj7E`cO4u&H@UP06qP6K
z@18t~VNYAt$ef92cLH(=nLNAJ^?0YiPe}&ESXS}v`H!iDPCNHS9bUc7qLO0U8EMnh
zmT$YzCu7F65$ZBf4}0P-dC}?y)9WMl8#_0LopnK${W7-@_1goJ>o8ZHo&Q-$8EFq!
zlRDv<=Nf#V7Xoc1a|H9U<%UObdXY=gm25=bY}T4yy0gy-0j_#!m!xeYc{EHd-v$Pz
zYq(1)zbc<af%%@k55zLu4y8v$83ddE;0{3kbuQ*;$A^7f=7rY?4t%xxMkE!)#fvKq
z#)Hb*PBusZ(N006aY47oFEa=6_Cs=80NMt`o}{yrK$Sjp{8u#&Vj_?JW(;fraxhWG
z2Yjj+`vIB5z>gpB&P9H#On(~7y=mx|rb-Q(0ED34%P~GN!F*I6*;2m#8O#0$V`s2E
z_8zI&1VeO<Ion=HWga|ufK*vF^*l^(#(rKlH2i^$SBi<ACpDV@FNR1+pejHODb=0P
zD<vUuT*$fS(NT_@(a6jnRG`>W`stq?SW_vQZj4~7^Q%(y_P1}PV??X;y<{_E*=9zW
zh^IboQ#L4laB>Z+zE`M+wJ2_af>|P(&Y*wgcV-gF&V4@I!s?nD-jd8|-OUlWEt1`#
z)V8_wKuwjk?8R1c%eJ^lS~Vh2!}&1DMF^*OzrX`T*jr1j-Y}BEy?r*x!eQH6zRiC(
z;6gkOwN4%6b@5k1u4@phC^#~14ON4DboOJKYUTirL=Yu4)3k(bZD&wt6k{e3$}+wX
zr%OBU<HjdN-UH=c0QfIv_l=K^-fT_(j!?CWnwmNF#)*7gWQ~X}EH;0wTepJ!0`Dj+
z98_IUOcB};3h16kmLHPt2|Pgv0pm2(%hQ*RMeoF?S7nwLR-Xi*1&$6LAlED`<#y|1
z|GpaU>Fo`F`-o_L94r(9n=>zWi5*(L#{JAoJop&*)&mu@CShZ1EqC_wS?)~Uk1f$E
z_}?PWEVt_>2>=89l@+D)u%<{JO&&Y_Y<HSAkWVsG(V-d01G;VV6_7<#^UexVVPeXX
z8~CN2Xj=uOs9Jci^DL@G1O>%8W?CCPQG=xGyLbx;&v{Zvn{T$HhA+Ra5+657E4q=p
zZI+m4n8jc0(Zg_aPK*1-K?tS$udXXXJ7HgU&WkT3#Bn_biE?Zk$6W-1Kn=-j{a3Q^
zNcQY7_Z6BUwb9}aB8#*shLPFWOchq}J|FWhn1b7#^?Y-AO$|E?n|`I0z}e->-Me>T
zH-@MHM=ma{J}qt#ZS^Y^k-zFo=ebby29|j3E4EWF5>;od9r=MphU=p0cFWYS+TA-c
zim<TPM{~FTqNM4sFa!FFN=iZBUmF-P^rk?Dl2U&(=~x<a=O*ONqC3U5pLgHFNWYph
zEPH+v-Cnw`degW?x=5bxUw4scof6_2^UuIdt)|A4rS#!bSy!nR;j(PMgnWeYzK4f7
z##tqwS&>$;@JKSPNIw<%@(VrRh1<jq4-sXzgTX@v>VXPQ;oh{gyMBjHe?<2v5VZ2p
z!pFL12Lj1$5mHR~TYr*6Io0w;XKOOI@BM#s0a{hwzii*w)2cb!-S8WyB2MoU3H2~z
z=;d}O_o~}g^y0;#s_wHXT&%1=L3YQeY*n0`P~1++Z!XBo!z}F|?lfAnn9GB94s+K!
z*8^xdR?l9`v$|krhuOQu`gz~kJ(}pUHdG16w4joQtr|0o=degcM92ZqBZ%e`Z<{xw
zx;>uLHWxe>5TMnDJ6ty$nMsr`u{piHmsfOnYuJfihW<Z)Zn=x7D%!&aEs02yT)hu(
zDwTEHRN+GeDTh~atZsdmM}KiBVGX$(dUNad(0!H`nrf7<ruK^eepymd?(-RsdOUN(
z2>VMVlyjizgBA3nw|5UruhAR;zy|lc;HfIz6b5IU%@{yWl0e)S5Zkrzm^%&MSXyg^
z#!TZp9Au)}yLbPLMtnn+aDC=rFxy!Ac5MDH{AZxvM8*t@u{N0ulyx{26CyuLjML=d
z&1%AT`~7W-VGpvhzUME$Y{9%=3tgxiKe(4447Y28p#ZACvI|mFH+heH^j#Y(f3Tf_
zOjtp7T>nxC<l%z{qXE;*pY7kcrh1#FKRbVx-qglhqN!uR?Z~Vvz$Ei$(T4>yn10u;
zgAx00pptpuUWYUqJayy6OF<l}ay{-JIQ-UGNiJSK#)Oudk)3H<(Ux-cecMxeS^sP8
zrZ3$zsi6GV+Vz4(1;Rl~?%D0lNBP(O9W0@9M5)K{-WbQJy>+5d!oV5PqoLzw4Dj`R
zjew3DR7ARBNy=!`E+Q4|cTPMZJI@gD^y%okt=#K3Zu}H{Tn~pTTA0$;t|_bBy0GY<
z_8M!u_gik|<uR9i8~uW_AopfG_Gl)(`a;Y{ATC(V<;)v4Y(UYf+<hLkM6S$t^`Nx0
zwA}RE+*UggHUXS2G&Di8P4A9b6-gCK7tn&zub($2y_dA`*h8x{qx$X?e_yE_+Scx>
z(F;3lo~oA90s>1Dkt<X!i@iQ)y_h1;x^#^iA}NCWKrc38*~KqYgRc2DN4FREuApsV
zCyPHz2&CxI{8!lAeCi4Pdhu=l=RW=1B@|VN#;b=?&xj85jY)d!j_hgdFVMn~P=#DH
zC2dF0#li7GYIEerY<eoLC<i4Ex?Qp*7)by6U1qF+v2jo*Urz0pJ>5y>DF=#CmOOZ5
zI!Pv)qGO$O(oEAY=jXH8j$B1n7taPc@=7u4{^r~LYJNy?9b(yh9%KA+;R#$}_xu_5
zU_J`cD$A<l7JlwHY9DO%tSZ662n&B1>?QXZmW9uvU3DMZ9$%94VlfpwR$)QQonTaY
zvli((^6Vj|GXeq<T=hoU{iktAupm#Bar;*Kgn$C%fA=!(InvXsJ-T4z<#9IP2E#8z
z4$mA;?^my00iEnZ&lSl%-VWIGiRBHmxCWp>aw>HNlN4LGBE6tIdl2~DC!`|4-w+cV
zVDJgY^6$z9Rc)7!1P;8~w^0sKiF1bA!<H$IX-Ttja!Su7x!*;85j_?u7;Xm0{a8-m
zVv`#A6O66u(=XU%sFY_L_tD8d=%Y;K(x!br0)8f?L)+BCJ2e#i{5>W14w>t{w#@01
za6ag(!Jo27yj#DI;UKTo)S0Z0zVN`oF_hojOlZDI<J5c~=L|yn2np%T|9wxe((D$B
zN(OPwL+Sbv!^Q6hwZ_YDOF2F&2~W)X)pZBSej^G(gch@oH*om4h0EhE9Ccl03%@!u
zl$0jf#c+!n>Dlqlc40&ftO874;S&-n#1=<ay<Cr?amNr_tt{Z4h~-k;$|Ejr8C%CX
z-=mJPJj)ZB1Uw3MoaM}T4OxL$-@eOFk#$sUXa?WE1vc(?&MHYwr{h~UoD7JIt7)UK
z6%J47@yOR`9GANwV|)Ay<4MnFNLSGX0824x*qDL{e3wOPc~F9@m`zm$PCo8U)G1dj
zH^!nftdY6B^|Smn;U=ItOTCZ@07wkU#t@{MZ9d>=u%znDT<fCPEwFidD~#S1EI<A$
zfh*ly>SlQmXP+Q!1a2Y_$)(;#lli^gnBjx(9s#@Ke+yd;c+C}MZb2Y{Y6h`n2W*zL
z{hmM9=y6UD)^+DVY(+A_c2RlARu=gL^4h5#=x2K?v*)CJ`=(<NZP(B3SqtO(wmFy7
z@?$Q)CMG6KI&)v=IwDE%n_8txI1cCiXe<meK+%1ak;?-7JFupn<=&E&{fw<AfD*dH
z`j5-xP?J>cT98x|^z*75enYZvr!<8;*t3L<%#Pa5?)l`&8zZ+jL?mjV>*!cmeZESU
z@6BP6PBeWU*k{)iA^LqCwZjQAG%}Izyp;bvl^xkIvUEj_is8sYY97m-S=;`X8-#3G
zb~YlaVGt7n6I=*aWQ;=Q!GZ_63g8I#wX?CZw!XO)W_iuaWu)oG!)gdrxA}M<t$V}u
zek2QglWN{KK+Uxb!VpCuisrd;Xac8nqNA9)!+6!v4|<_U@(oj_iP#F@UCkT5&TYD;
zs1katZeWG_$!1sW_^m`kO>H3O2h;RDdt!KAOQ9JL)P4aKH>n~l&--JL``G;h5K!;6
zwZzi^cn_>pc{7|hwK(VW9H7Iew51!w*(SA^3h^SzuKnKu)bSi)f_B#kq*G}+D%12G
zNSryjfR?&f317rkhKUOpAtF1ZX>Xk4CFNTce}|<hV37yO4M+-!%p#EP3DeoG+t0)i
zr6U~2B`g>UT*j69Dn;wbs0NuT-G3zo9xI3J>JQtUv;WYx&V!uUPSb6MmUYe;rTL()
z1Wmj#L8bCcM$^`G(Nsbods)WR>7{JQJ;ra9#1qk>VtHO8iHBoLfS=zBo^%7i&_ga(
zT(z<$$a|q+o$`VDpCp@s+{Pw7k*?m#*2jv_hkK>{=Tj>@z1g}(fvll01Sf#h*|<{x
z;ZSqk#o*WF{>4W}BX~Nm=pIHxX?L2C4HhHfF}F(kH_*k9RFm;&2??m`765^xA+xl&
zNQR<+;N~=@qN0K?YchZdY-UZ6kRjd0oHE2xtPUdP0qw7%tn6~v-2-`dTZoE&tPMmh
z8ag@$tVxtZz&8rPw1T26!cN^H%G49P9z?93&z^wzQO6RSGq{Xm?8w#Ur^(D*9OWSj
zkW2C`E6sIU)g|onq}sFR@NhY(9F){34>(#nP<Vv~pgr}WkY6JR{5x)BETcd0=0K)2
zdX&h7VaJ`w?xkT$@NLci$KHEDRk>~3nz*+S0|qb>Oh^)p2#BJX5ZMYSh(s|b6hRRb
z6vT}fL4sly1A?Fg6#)S;pd{1A04gdHOdudil2qTsb5Fa~+I#h0Rco(RdsUZn?%n&~
za;^3M|D0ov(R&|#unBaWdyenICV0N-KShZ@i4o&!c2zAgpys)j*K2%>TquW!^9!JG
z`-mgR$e)IOpPWniZN6bn&-M)S0ZQ5z_3eGy`<|<BCMOF~jg8W{ad@mPSPXnQkx!S-
zdlkE6S5|fz(sLP$aduOfs(D}1(c~cbuJGpi_%Yi&Gbd;No;{;eieP5{QeCvpcaU#d
zWy%k~<y|R^CmfqrL$9OBXJZxr*R0^~Q6?ycg#)@|WprYqPy&nD&G3m%4Px5M=ZEl!
z(pDy4DC?zLEbvt^D#%V8)X_1u=E(9=TrU>bygFL2b75M`u6&Bc^zOA-@Ysw~Q8Bdn
zL1-#EF81uD35<o_@2-s_!S}K`kISU{_g{4P`=M3eCuvijj#5@W<nZI5*1OoyJHU*Y
zF1a_mTWwePy7?y6<{*uvm7uNrI*hVdI$K@PnJGs_&-uvD)7NA^l1Qa^T|d!&Qcva|
zeiI^rf~=FrkFUL&K&Q&6yvw$FTh*qOgsQnvg@-Ej)A_7AzpAi4V3%Lv9ohC$Uk`sN
z{D5$AVbUgMB*vHbA2~}e+94vgA(6%2)w{5~oPi?sECyzKu3WK#c|2*~9}IdB<ocNY
z_KMO2s}jD~)j`0;034k{qjdbZ0ACTJejRn0GDy=b=s+Z1!ri;NHT?P&ZFy~kj!sf+
ztcla7tX$9TKWc0J_U^R=BR=EVatz!MZ~ma5O>|Jg(GPlGudd^HC4P&Wz#l4^4c^+s
zI)-mP#^>NM05*(waGB?#aV_d*I*}U9rsFL4robWcW(|<Xq|b|aLs&oro%|Y3`Oe0u
zG{c9hY7R}*(XpJgxEuy2IAN^X1*(_R)wdOr{<JkS8!qo3`Tzh}anSW`X0|pq4&SAR
z`^hqx*SZcM+<85aVJug+QZxU8VW01H(+6p;-hap{{245KZo$q$bQRNl&^P7C{UPyg
z^)HtBPhMzdwp?Ff&b^9;M4%qR^0Sh^8wz(xu@?it9zsI!B)GEJv)SeD-(4PbRhg)(
z>k0mqt-h``%evRlPPUhJ(JNe<sc#YBEsCSf?WH1nO*$I(S0|~G;%u6gR!sMKltH%?
z<&If{^SI@iJ$*jc5;8Ee$yWdCjNV(fzxFLRyYlP{0I9mqtn{51|ETG@Ov@p^N<8;>
zMqvdX;22cm22nt#(d>KoM0|zmvoD*)oR$xO9uBAc`Cmm13!$K+QubO7iJupKLf19=
zuDhA`Q*z&JU&2D3S88z5gI1?8*88fa^^6z4r^NJbjgGj@vW(BO10w2`@Uim{7fzBn
zP4f;PJ{-0qw+u0#kK1eVcjXH&zij59>}5PyJ)B|^CgDixq5DH>z+UIxI6l!Oa>aJ#
z?#Z8uH&3+H*S|v@XD&~9HEHr>nTiABp0)%f44u*A=KM}nk|V}vW_<Y<=xR_Eia}RW
z_U5XodeKcwTC}x=%W^Xd);aAf!)E6l8!P&rP3pCA1AooBE%`XW#nlxTk`KZ@^x`;1
z>U5KmvRLrU3S!j_IdO)(`@@SdV?kY6J*pn<{kQ`>z^@j4Mm3xEX+H!-2qY-?(qe}?
zSpt7rP7^Kgg=+GuG2ahdo6=H$-evJpnE1f5TOkqj35)5<SRg5%-fA&&(xfKQ-j0R3
zWxEY3ge9ycwHXmGcgE+kGUh2<;Pmn2&}VYM{D$6rr3UqC{{g2F!f^!j$MI~|`6*Nl
zOnZXD4qnITYNYAH!{$LkzrDDcG^EzO+wbo`D(+A4z8Pz3&)z}4OqOb1G{tdO$L_%B
zH&*2IE!Md){t)vNacC;;%|(#Eu6fld<HOglH>rSG{Uod^d7GV_20&W?d!#R(onGFU
zWsPsYevrtjC;R?K1=U#|&vTQ<5zR;2mxpl|fI$jy3?ap2($l80m+Ihg+|tXqyn)JK
zS1~fq2ZsbQVPU>5%#)3;ef{=Lu4m7l0|w~8r%-iAl^YTx7+4@CMIkr%&auR)#(>fr
z4)?E<cDFVxCitaM<!{^e3*?2wy`;GK8J*r(8G^?3o)UYd^h|5hD$e~7E}8Y#=CkjB
z!yCm7^fkj5tezgsRnTnna8H7v;~P=&t*7$&C3pD5i85w>e)^OLAw#=3;lkjAn^n)h
z37T4<zDV3noM~NW@gU%|E>l#sh@M3JzqY|*%u7T+?9Nuw0Hii)BpmcLf|P((F|Z!W
zn*;()bpHVZ*4*!LLJg%6YXZ6h2gXS#5m|JwQN%VTG~iV71}+7d%$16Xx?senamRYb
zI4X9&D{wfc4D9C{{xMKcx4c@yEh6L5g}f!UiXh4c3<@enM?~$$4d>{GC1j@f{AS}s
zW){WV`t=nxuPd1vm(7KUX({312$HXG3=6@*zI_~SAxlh7tQ1Gpz;*9Z8n|)7ikZrW
z`nZm|5ZIr@2c8T3NE885QcQvi`BnLK;ZLe_Lg~}Q5<sUzHcHov7*M0T`SVjEH}U%u
zYXq)FOf{f}!v;ps1rP-ywlf+1+O4y%>Vgt%X7%vi)qyRqD(0JWsQs&cb$<9x>$AR$
zuAjw=)4#q;Z<LiOiP&r5>0W$LtNiVyeV@fCnYHut>-eZtUGoQK`P=i+xCd$NU`lQB
z&-7joqK*ubDH#`W@{+c}!NNQGX8BTjYLT&xPxZY!@n=&G&?y{0%U^!|*grngLk`z+
z-NIA#G9uyrt@KGZU<~ytYTOF5`vtz_EWdy(4;az=kgh*LVo|R<yLLq!Rq%LeKBkTe
z%0cv<MwC`mh;dc}Lzb^tu^lqJ_N(#zV1^_Uf(C^!@ih$yhTmD5Evfqe^*nbDRKv26
z(-INdo#kcqeVPleos&$K>LPEii^lMFw^6A-1G64xFbDc!R#qHeEtD-gpt1^?EKfCR
zOF8c}rf$i!X?M{(9WbInFPGjp_nJ;q@9S6DyUP67_2j1NN4?>0UUNnN*f^}WJ?Nfz
zpvZ_NQnh&Tj){rd_lh|X5xm<Jm%hv1xi3jdxcn?yw7UKpqOqc_8zpY{?JQp&y>v&F
zr3JU5y;Q8qOYYncF}5cB*{kG!G9gad+SIzRy>H@v-!EZxM{c{S(oP~Gl)oMIpZ_*;
zX|Jm#g#Ox!X~#13+Dn)zdsH2kZ|oU6@j(=7Zt?S)raIpr*fAK@Ee5ViaCHrpH0Um2
zKgy%y;hD0U5>q?;yLtO*kAL?KiTn2Z=i9%Vx3B-ZdHcU!_<wzS{MbfbxvZrlsY0Ln
zFo=odmbd8nObSO3?K9KOq~(ym+7D(HOm=rg=)EX|@IcI6&r_zG<VOD((ec60>VJ1(
zhccD#_wT&6wDl~B{~@03?_aL=tufEpjrwF8^<`~aw_F;W7^K>ukj>=p6rr`EF?*vd
zyx%p~exVrZhF^0Pt%sot$z(Y`4;TpwhoG$MBHnc6vJ}<d@BKggXd!?XKmYZ@|K^kV
zuiNu`dHi=4?*Gc$qjYNCs>-&m>(*`q#)?Sk{G@NcevpU*Loh?+M|pfA$Utc5;B_vd
z*EOtUZDm_mw<pa14h(y~mNArppUlaF$hi9VUwv=~8}|0Zwk=w?=-m19lth<bSTQXY
z^eWy8q59M>ZB51-N7DM&MM^4R;X0g?;ubF8?PGs8+V;q>GQ19`x6+T7ZyetiOl0TL
zDMDUm_G=4(gUzp5yxiXQUw_2*M1g1YYEMwm?*_O|zq_k@9o~_-mXV%lKd<D92H!=!
z+CsR2w2X|o5D#C!Jx5%Q-(GQ-y@oNpaz3Ea$ae)VUbx)$YPYalT=eDNom{(AA9deA
zR&j;IS*EQadtF<6DGZ$xv|w2*(ZK6oTf<dhC_+ZeD_7sy9VS~QLe+`izfT;Q{OHQh
zb^83~_U|aA9ZfPuY4x<=O8?8evM{r4i)d?*JbnJW`r3r=wY6e^_(9RM?UmcsB3=0O
zVd|?ZB!%!-H%wRYW+wQh@DEuldYq|XfBovVkT!vh8-M@y)Gujko)o<TOP`6J>CoWc
zpK%ie3@%PnlN1|1ac!As)eEH-;sI*I7kD8HxO6GY-ke`&v>k$>-Asa3ckSG{bNB8q
zLPd+Sa`(2VyW%!kC3?Q#vIvEj;(B%4uGgd^Nu&~b%A_8<%z4x7_P<}f%lW1{gIciU
zzt+~ny|nFnX747;MK@_#VYxv(cV)G<ZxY=7-Me>WO2Ho4s4_5j99Sp9?~%4!Z?ZD;
zZTo(G`_@95uX}@ohs2^i;z>?D2DQ!u)^yVO{mDHo++8I~(>X<OCY$bF_lJa6^^^DS
zTWQ97phz761$k`zn!wbX?K$xOAuZ+R^QzLFzmEYE`IA#TPiwyA10*C$^TAt@m2RlC
z3!P;DU#CFZO_BJ2A$p0`ROa^eQn($Ur>H*!o`B;j=QZVnogKwhUGF}9{Au>0YGy<f
zE?pp#Q4ZtK=qm|3vkBWr`4#l3$zFP9r=(Hyl##8kB_x9XjIOO<BRE9i#RW+Tqxu5m
zCQB080_68B=H64*)hJS@AdY$u)W^qyFP&SmYRLo|EMc)G>gKd~<?$z!foonV2HP^_
z40)v4f^oXKjVw8Get=!dYXANFsj#kJN$zCuu}BDib18)=dgC!Y6>ZYtK$g^TX74ZF
z?P5tBW4a9cn|5AUAAU}r^Z;DsL*&r*vDe^6``+p!A<<r6^zFsssa6IMEi~&?UMJ$u
z?%!`Dv~Th2w8WUAHQO9E{)ZOe$EP&9876!kGN+P81+&=?@ZF;~!*B5A$3Ut4AQxr0
zHQUd7FwTV@WKja3b<GRC1mZcCDY|@b<kcT{ybRur!#s`_RNSelseHNh)$}9Qf)of<
z3}wJ0HktA&aX`eTQ`GI?IiSFf9l!Vx85xSzEBR`R-cm{$U8W`X!rWCR0nD8gUz?A>
z^y5fs;~xjjBF(;VO%xa3@?y>$$x7@^a{z_4aGN8S>&~QdSDN8hhz@M8R}63w(~Z$M
zySui(@q*G7s9i0uOXtp?5eZIcZbuhR{HiDNC_{Mx4u=G|$&{7=1`U>fNZ4zLmG)!&
z`q!IL4;Gx8--`*XR&fLEWq<;fKUbCY#deF)cGRz*zGlyuDj_khDz2Tx=v_>Rp!oxa
z{gTNiVg_W>l#PYA-+Q)hC*`DzgbWxrxpC+=kF7cl0V#%#YnF{3`LUuR`b<L93>1ED
zKw<PVuYapJ6{XjP*7}SoHymCqt@O*!;N-wt>>pedM^BmJ2`N!q(ODw&VPOlGJb>Ie
zmCvRTmC_FZ!{0tWkC2-}Y=Ve0qurEgd=%JXdNzAQ>z!LM3=w^{U$|zV*()oaEQzN7
zw=_L!=kMNshqnsB%isZA#xkaTfUK;5Cwzuivc4{@kXqsjYJ@u~ovZ$P_S6fi@t4)^
zKDo@s<-tL~4Ptr=G6wo9Ci`3gzZfy^<mzXB?Ig1MR8TNRz$fnFJ4~;51-8uuj=vtw
zoiF+Zpi}7b;eozo2S(L?Z^iC)FpG`82JDVE^TzaGf-)UvGl1KYy93_P?b!;%wb0`}
zcL*MYO9x@a@M|&}t06;{Kt!{4A+#30yTRr*9&ib6on%04?@RL);S`YKAVSAK8s9(A
zAsVSX*z#=R>q|e~A8v3~Z`)60B_)HCLSxm;{6L4-3N>F}UWuCW6KxPS(L@SJ%-M7d
z&7C{h<*V86!7)%Yb6Tfyf!IF$-1KB}9?^t^h8ZVR?5zx&8xwv0PQbT<q#rc3aK8#G
zD=R;I2nY_|%qFKP2KnHS$&*umIxcv+G2|=n@tJ2ufOt!w0hzTE$K5ILD)t2(D2tO+
zPS|z2o|c(En+t|y2T+iAFOTDvtz5pG*&H~bHv&d5n0WWEEthww!X1EVQM&U(;N(T0
zbKrDW>aXjojPo?Tv2*99Lh$egBaBZ^Lh1*#&Ae~Jt1XaPVo|=F_osxJ&BWUS1-)3g
zX@J)wTmxrU4BgmN?*x()mA2H(EVv}+APV{=QJ~M47pL9g`=+nULpTmv8nLn@#M$`x
zg!S7#KOK@}k@fJQ2iV4`Q->=IN5w^rU;s=aP1-2q7HzK-#G=}iDT`CS@qL&%($#4z
zx^fW7;U~m%L_^^gB%(_C2AbVwrkuU2G|%3?5Troq)uv^r<gwYUs;P;0tHgGq>C==}
z6;;)-TNiq~@pCp#e=cje3tpz;b-dBpOK@vt#+7tW?N`H<BBZ|e!h$d?YCY;roPP-B
zW&KzNDJS)#ZO_6@-P)#;!|TLfo{O%v#2;1@6aXb>V3`M-jah+FD}7gQR2_(j^j1mN
z$#{0Gj9+%Ai%REdI7LEZmAt!L+%&e`?OE`ADy|(Q9!}Hi@ls!}hjQJ?Q>RLA?IdTN
zjF{<$o9U);b3oj9_I+0w2CwJ1NG}~Ivn>~3l{MRY@d`q(q>9e9i4KR*-&HkV8{d%F
zrO$}j8{Qx!0F0x7A7y&dOGGJZ(Vjr6rLrq%l`iQD<DqZ!Dl6;D;?`dl`uaO>P7Vjj
zhZ7{Q#@dTpl^G^47}%gK&+9UdRtLOTM?)c4<ledK(Y8K%^vHYoo)@l{>`i|S!;`$C
z6r^Zq(%x^;e@<9$)97KtCZY&1DsS>^`JqsH`pqvh>1&R9^LqZx>>QfMv(}@IL~-))
zYe%oE<lL8i8c1s+S(VSDER3r!u?NH~<>Hn-*l==6x**ZXwv>-UuyAV9*4+2+-+T8X
z^1p=|CuZ%{WYLopsCBiQrcb1x{j7iQ-b>%E>~QQ(BwBDzLeVJ6ft&uMhk~9>v}Tr8
zMP=ptFr$7xdb1xNI2{tAbgLQDm96|`mdogM+wTdH6;YhZ;H%a@jpf&#APn13v2dlz
z_e|FFc!9gOtmaQa=np2zJ#`CFrEBi6`<&`;Z+=d+57~d<pn#1TEJ##Td?{8pj=+j_
zm#-Ml%=n4E;^=|Dv^HV|Q;|jjB1A+ICSk-}XJg2s`8lUE8~8{H9rZW0bUnPsR1vR+
zhi^A>f>H}Cv=p}Wyj-(RG1&al{1puTcxme}Mp-$6hMb~mZr|4o^=Rvac}?YLI22t2
zg*4e&ezQ+}XIooa+!_=c2cO=v_KtUlz{5r*6&1HIrlSh0Il(-1zk!-_DlFCHJNpbP
z!*6)$eM!v$JyU@h4d>mtI5o#M@~<IFmLHw_@aWu_Ker@h=4aC10Q4J)nESJH_jxmY
zg(r7YlSfv#!>O1~62Wj$JX&;MbB-X}=SZ|Ie!K_ymLunM*qXD@62eL{{@1Woka5m^
zc6RlKn7hSMg-|02Xx)Eu?INl^3au!!OVy<R2D4^qSl<{qYLsz`xxKyp-t$op$1Y9p
zoK+2|cVg1$AzLj@4ekud#7CtjrcgaOzj_GQf-c+0M&s5CG19@=Xg!SMMc$p@!c^R9
z1|D5G)A0DhJZ>{iDWNSIqjSbazMOnDCZ#%i+>=Sm^tH@#dC2IR0^R4N{I!pZUr{=A
zSyyo|p>Rzsco}|f7$=P$!%<2W;4VGTwWEZ|h*z$OBeT1I^#3zEm5BngTN-g+dA4O#
zr1SCNkSFk=2-Oy1P)S&Xe4^%6p~nQjXLRWSmyZPn4buw!D_UurSkbhAz-}fdC)3=G
zgEUcjh($wKh+XdlPwZShn~D{P6xq7NX{hAZ6bQeGk?-6`P^>{Erdz*duDjcTcR>92
zU62r$Ts<hX&*cXSr0oT_W^rp<O=MQ{Qsx=b9%0ft0h@d!OUskqW_m?vHDE*NGmJr<
zL@NY&v71{x{XUF3Sdr`Z!vFHMYd;7{CEMcJtF-ct5<fBgOV-=fgLT$~eA{x#zGXAy
zoPOZ5gI_k^M0hRI)basuY&(^^1;qW+<g^V|kI_Gzd}n=q4QM+M^YoIA7j+rBe!Hq_
zY$LA_COQSW)wQ(~qLd`Edw%ra6^}~X1ynXW7154SOCtrn)rLMe!)f+Agu_ZwU$Y(G
zI61>@iiPw->kU%8q)UmV6?Ufe^MqD@pyt}6&xNWRs$a28LC<5+k|`mpfe?x&C@O8f
zHd3J<&LT-K==wVl#e+7w`kL&e#OjERcP%MS-#k>XUm?KL4-2*U{Bda2uMVQG?NHeR
z`gzi8Am}2TE5E>q%w3{jHLEn>8}|J+F@TFI8)_K|8xS?>So^rQ8Zu_#Q=oxlrKLqH
zgU{jlv{pIhCW`%)l9{ie5E2kT7Tz`Jm$OO|Tl@I)>ou1mZtGV&A;SrAH=t`^Oxqnn
zZQNI&-Dcmbst%2OlAWEKsI-*q;n2udtbpfUoxP3v-zL_6r{!Fg{+erp-DX{bTg&Rc
zUJMOd^YPG4mO$du_wN^<>~sA<!O;alV<JEHu#{Q8&;<+sWH;~8Sx%QJJc@!36)d-d
zs|_rB!0t)+;R3fq0+deZl`tA*)#1Z07HnN!+V>9g3C@gL<(JsBOG@w1AakWGjpWUu
zqvOLDZW<~9U6EKGo}b?OdO~WC{JNpbhM}8yNv1Pbvh;{b1UZA@UJ);qbkr}NhG9B_
zOIGmAws4X;*XGK~HKx{O5H-tb$Me)k$jS{fayhp(2A}8`k%xaJWp&Wk-dcU3K<gnC
z$Ehivb~Y7AsquzQ>S!ImR+CbRSSiXWl6ek^Uy9P*N-7>+oud~`jX+w|H$-Xpa4MjJ
z#1hC1+}Ne_%Ktg<TQKo=>qX&CLor~>Ei<xYq7?bM%#(g2wtg1VyHDy2d(~<C6UCg;
z$3y!}JXOJyHeK-S?|8q<1?L7($Ea&)94U{zJH`8GZSwO!VEROtdV}-Aq#pU-xd)Vt
zD5$O^HbI02TDX6o*zJa%jNBQQe!1-=%-RsL$UWm!w!OTm9yNX-uKAWNVzlIXXX?zq
z|8DP;glQDX(et$H)mD9-r!TlKx5de6D?~gxu^kYPq;gbG*Lh~h$B3A&n>H<Y)+fs2
z9>&{8PD;#k-Z3$28jL-VQPu<U*WrXy!`eqro)}l(8kF^>j6K7AY>KnHPrbOYhFjP_
zD+5bM_{r*=8-3U6;a7uN^iGE^U)*>@ZIa@}W)00eK5}^8j`%MwxBdDJ7_@i)el=y~
zp7d}{%<@@xU(4gwFSE@a)JMVYs?7Y$ISqU?RFq-nG`^0?a`yD}{Jy=96}=RPWCsM3
zRZxJ(p0!zcALpXxsGM(oOG8E*E~O=?l6;yXC7|(f%<clOrPm^4FGq6xT4lZQd?<JI
zH8m3d9oWLu;E3;cdu<bQ>)ED%NtX6VU128g8r0bS@T;QC>=m~i39Wh47{yuAJSQ+F
zj7!LYAG^8bo2vYf4bRHFwvQS+;oRsSj`QaEC7O7QIy%?*{#gAXG27@;RCoVFLV3S{
zv&@n<e(+u{p2ng@M<1LTv-0YLbNYJ|lrqgntEEOXsUnXE92vtD-|uJe=GwDok4fvf
z3AxxXX0=<5$<C3JCt6vUY*^`#P8~WBSgjaK+VsOX%QPBGh;tv}O6*?{%=f1~$dTKY
zqg?QH>BGw%eAYWHSL`z#e?%|~=M$f~#ohKFwcT=Q4&>-6T)=5*S}ytUVqUvOyX6iu
zPCgrdD8+H!!;9fdk3Ge-JgXY3A%mls7Zw(_U4xEu$)17*MGa{}zI&?00;%o`-4R-c
zZlu5QC{fMSTzw}h^U53hCiC^!WC~$J>%JSqTo+qY-IjJ#g*DE)e6js8>k{g~pRVuQ
za~%$^a!X!`c7wgq!mFoFwXccJHnZkDhe2|$(7^2LI(gEhbS;^IXtnp|WgJ=_VypUv
zW_#$xei}B50EUBA7Er*wAR|gLpV5BM;!fr60dc*2qT;fx^BuOSPSgXKAw!$3Xa4xe
z6$y#_&H@)urDBCuI_$9)T!Vz-=!51lnU5YFPU#XxqEY0ddoWAA<tQn1%Zoq%_+zBo
z%lgN`kjq9dlSm(^_xVJ+626s}J7lieLD$vS<Zv7xg1*mP8#W#b(j=RQoM-nl)CLa@
zrC`0=`c8ROEdN4S*qigCy88N=3x>_BbqC2=W+kPbr0qUu@xb;Bv%4_+z^fG@lHPPG
zEDkD2CB&UMvzU|yQzGvUUCH0W3cNW5YEv_{N1iyn&(D$l>uiw(GQ9@h<EfTr^YnVP
zlen?a#%83BXWHP-zuT%~ixWmBec4x1@ve9KJM3HMcfUX&(pp=rn~gYCuSq`1K3Jkz
z5nHk+9EBj`b!tgCrck`n_warE`UwNSyd*PLdpxGHtZWbmV1pUmTxcuqmx_hQWtXHL
z2~2qY#ZdkiD1h#<O5(3YbBkHrKLLkrhE6>+<yHIpp6$mNybJ65(<U~zZP)v*C|mGP
z0NRk$>TI<@W0}5vMFUweIP~wm9^WE_`F;E1HHR*b&$;{Jg?X{Ymv3*AEv`1vCw$~h
z3IBgf@43bro$o^Rt89Jddy);pOpt}um#8fvE}Ur2JG*4AL|ghX%IOpdvWWXvUx_Y|
z?RC!=^k1^JrlaDIb)YQ>aeV7%<xrB-t1>hC5@?mv5`-}nC1!QcM>x1HFP-J*Pbz%1
zAS=sI!)b8V64y0*9k$8E7SDXJM1-tG012pP&J?_Tt9XV1&~p2ABqW{-IL<_E%$V<F
zl!x87Fs+p8W9_??q$9SuL4#za<>mEg<aTPf_C!yNRl9b4ek|h&fTMz@(NG8hWB8II
zDRIduUzl>kmjO8rTRuO%)3?nB4Ak5@;SpF((#+xhvN}T_8v=FMI{R-&=IU%|uiq05
z%n*&uTeh5@u>N@V4yrYwVS+R_w_L3FM(!2&!|(`$wOG)zY+6v6`0m~vxyJIze_UDR
z%m-t{l}*Jojd?_Ag!K90^W?H5i6`S$K75*J0m(jR!|s*0M{?L)t4r!OwJT?6s&nbS
zyDvf`B385Q$=ueFtZ;8g*fHd4YHmgd#6eslRKz0~nBY7Ly|H40DmlP}=#S;)9u);T
zNgDixudh~};UAIYf8ZD>27ch^l69O9=cHO$Goz9<$XGp~AEu~L-@XGmWH?ueYTD0S
zqX&~gEqmr=WoP$B)|%eOSWHcEezecvab$tJ_h4W=)fC<RsT?WJ23t^At+i6d!+~tm
zW|;S2g3T#XN9TLUj#V9$J5M8B)T=Mt6&x&migb8<)O+_pRs_a>Naj@&l5aIt=MZDu
zxWn7F(T*$ND9L0koWOHzRXg7$8C7)81x?a)H$V0}i`!$Yb8X(du8&JN|M!;d|Flu0
zaQ4`G1b#cJJL6`0pWGKOUgYH^U|s`7)__u`lIeR@8k;AcTJGF;s?Ip)lG&w3?Lt90
z7vU5WT7(7OEzj0lGNT5EJ;~_ejY2&ww+@dU#gDb=HiSCOq)lIe00TAY<D~et=M+U(
z4+zFwSEfzao-NHSDF!uDvfRXw)V1^WB5<3tA-YN~nb&a3_a)B(5!h1s&bBB@P4+kf
ztOJ<RaMati*{BhL0Xtt|fZ12WhMA&{HqX8`Xx}+vP>aGv-6`+V5LYjU>qZ?0eJXB?
z(Jlc_`5q_j8yA>Hk6YPLw0~5|z!2Rb>L8MHlKTD8*6#Tq$-HVXyBRZ^b+%J42m1GE
z{{e07$hlf*B297Z+vL798i$Bu*B!2my59FwG1Ju;A~!M`f#_hZ!sj!n-%4)1q{k$C
zeV9kufp=&GUu-S}e4emw@!nTM6;>sSs<xU6JowfMpW%DdDNP#rX2oS*wi;voPWnG9
zG#jbbkVt|0W?e^fiGdJt>ktQmEdw3MGIxZg>!|A&$5}1CI^K@1_pD1X&u8O=N2$NW
zN^4NV#}ke;lPLw=6*L7YoT@eAV=+w3#a$DeJ<IYFZQ>yuH5YwAJJ-LzWm~PX$--sy
z5%jU>>onz#!cLVOzf!v*d6>1CnKWh%JLY;(^y|4*O<l2Kaz;Ve+Wemr3+ww0IJjn`
z>cm@g$^)LxFfdSixVejjSATKdDQYlwn%rFYozMhaG)o@K3xEVXlJ2S6s*W64j+qG(
zqQ!_<U+xCykm}MUX?x?14x?_`-|@!YDDqMg^(a;8z8tzMfG!RnKW??OD|xaI+fwU2
ztOz?NPRJuMw#f^(4|7hs078(Im^fQ69#_56I&g!1fB-B%8yh~mP-vl7Y+s-MGt=(o
zW)&)>x$?^?v{+Zce<i6lg=>G*FuzA$l5DeF_9RJvCs&{j76Le4p4F6*?Icuth@7x%
z>Bo<pi}9F9(Gv85l;ee?8=^TLmlz1p*l2uG`#b6S%dhm^{X?Kbx{M0YKyz_bQE_pn
z?gO8rE}=C&$4<P8G98Uz+%&tfOiM#!%f=n0sv3BCBl{VvLLm=KdCJD1zg<j{%lW0E
zG;-u+KD(%zbAlB5lH)2GA}lQ}i>fIagp$~~4-G$5TWZ%YHC;}Or$Zc>-ze)Fe>JPm
z{<XjfKgF;8q%`->^)Cl*#83cePd-T!n3Cf6<@+1OrKO6|p(m$1J+zJ7P3U;Nc)9Eu
zaXW(1TrvcVJdsUn-F&s1M8x2uA0dQQVUf_lex(SI8#wSaxi+Wb6wD2{B_h3~UZ<vq
zDB}=^20*MY%m_d}z$5<B7`H-()CiLR2NF2m33*}u)fGkzkRi`U1PeBNAv~M`XElgH
zfkw4lO{W-gy*MLLNZw(0K{og9-yaF8HN$A=#{%{OSYTC5KuDbw07pg|cnOSzxb_7f
zAu=2qS20p0kMAL?E!>)f?)9gwOwb_Xo7AIxN|d7#-jkdrtdKraG?%9D0W_Vnp@yRj
zqt1*+kLET#hg^FhRXaWroEQKqeC~;ttV$LPXZ=M`2!w_2-o2w*<<_!l0Z?Yoo-Gt*
z^i&Jbx@^!Uhi%J=nu0p~blQF1`^2=gG>#rRHc`L9Ib7$P@pZa*o8TwAYSrE|_Ad&O
zKY1maDT20-zm=)22iSE`1_w&IG^C|PD_YwR8deL>NSK5sTuu1`wH3o8LJwRkD|b<z
zvgirqc!sn{s$OBA<KgbwU8ZaJP%me%8d-aFAs&=aej>UoEN@}*aM0ab(r-a835h?&
z(w8Wu+w}@Ei~FUp8%!-aQQNHVbN^Cj39m72f4<@`FbaZ;I9_!Ih3NBYdPB=ADl~R{
zr5#%1QBkIsd^mQ+N1Srww6!h^rwkyOHjR{wH_eqhB%5$3f$0)$oX8V}ULVwljGXbA
zbejHRm=LUtg~i_c$&8j`g_L*&eLny2gIRkM=F;mp(;mjVW&bI(d=ko0YxAPTkYRkn
zBHmx1dx9Okd@2_G`>;0k`5wt7$@t6*7t+g@F^e55{X)+bo1xs>SRk}qCPTytj$X0%
zI?KBi+A{4V+8jzv9+Ulk<49npuI0*=7gHO(B@c>wH2Gimpv~sY6L|yxh1K=d-$@0`
zI0MPxSq~&OtOqIiNyOwtJiQ2TBstEIWJjvJ)i<ib01o*5J~+0(oF-kvoe!c=Mo!$f
zzV?%aAc3E$|D_6q%xnAk-$DynjFu^Mn&9-Ylr0zD=Jvbc{W$oH^j^Tr9z^8hb;UMy
zTYPNn(|4&;XF|qU^+$vZ^!Klq6He(VG17DSaJR$nm%oTvytu+w`S#cUCnClFm0GX1
z70ko=ZwfVDoI4!+A6kI_8=}g#@7to_UMHtuept(I%Fdn_>V}4Dc%QS^|0k?`n5y?>
zW!0hQ3ZpOYiCq}GbG_3Yjk{Lz7G`~4cFecftS+6LxbNV~^D4@x22bv?WQS)-<3q<S
zE5{xi9(1DTyUKo^e?30w{&8I0h?AS9)ScecPNH+0m9T_F`+q5Nz1!qa5)yaX^kfoV
z0-Xq3LE`$QbrrS33c1I6pJ-n&YIOt>9aR?UHLcsg@ygXt1nWx}R4oe~tld#+MNZ*P
zuP-(EAz2yi%%;K+2sf4GhCq_75Q8D+lgecf4Aj!KZnOEHnHjR1{-Pj^r1=Bdb(sI8
zs{m;94weXiy_7rinbw&G=-#R=y2o@r9z|$?+gm(*4-u4funuu|3)2SM4|s@Y5xjM;
z7AKMgt%8ly548(kMR>Yd_s(6w(0nK)q-Y)|zvhfa!ruzafn$mF%XW9)Jt|zHa5isO
z_0aa0VmZ6pr>zuu$~Orb<$AUx;?#bmHY4BwqK;7>CWPhei6^(`knqaata3v$Eo0Vj
z)>nRCV~903f(%8mKA3Nsv7vNP$mv}@CPs9YYiiw!vPN>f&Ztq9v;+t!$9qepMGQ{*
z36-zo)sC4gcxSI-iOU^&{G)_WHwAgu0w;>}BJi_)&+82owwrPICw8aPf7qSff{Hl*
zRkWfgHlDjbp1uO_TEp7`4-qAJwtVvJt#FalOA4ZqEUR6{thECs1rOQfD`^~K5s{d7
z##Dp%y(;N1`Jma!#0hU3MUQKBY{Tq{ve|#pnu>C9m=U#y&hpGd=WdAkd=XnB4vE>p
z6qFp<ino@m=Qbx4IOjNeHooD{I~3TfY%Pz=AdCA!Q>*W^v+0_j9-YWbaC}dnvEuuB
z1S=XTRd)eHmbRZjvJnwz?U)+h|C?Di)(9-rR;pdZWcQ}6qX>!P9qH_zHaafMe9!5G
zKRi0jzgu|r?z1xett?WLfBrf?9#CM%zL5;9H6hyU`{y|l1zJgtoJ&;pC%tldy(XM2
zMk!06Cn8a!A4Or&uCV1Cj{vHV9zWIyt7U0XJB0Led_%f-#;s^?fFg|+BgYRiAM!0f
z?mX_X{*$U_Czd>&E=3+!4lU1WIYEvHoJBQYv$gav0qaRjEHyu{!IY}of3A=oTYjaR
zyy}Eh`QYh<2$^i2-qH{|N`sBxt#Y!ml1h(Q&ykfb%~MY(sJ_!lL3&LP?iOe?M#gqD
zy2l0>%VD0o&hyvLFG7-GXtk&LgdF}^h64+Xx(^t;9FDdj1~<`)n`P1TqP%m?QRgAA
zW|xwCvfSsfcko@WB2yKV_>XnS9aq&?eb9fW*vVaYmoJ=Jr_vw~*{MEbF+z;!V!&2F
zpt3z~&`FAIo`Yo(caX(fM~mEpBeacU>^+lXnpEHw4U+wnpFT7qVaQgCY5&xI%_7=m
zUGi-ES=+Jjdkvd~C*UfWWO6427EREr572bcus<}uRbVxfQuFyV(uA2uM9Q)_`TcUx
zHY`FNTFPjDr!W_;tbj63YV`4}CZ`Rt-C~)@DjY)ox~%8UyYHS9p8bP$kfg{uI6H~V
z!&H&Xa)aT(Za&i}bVeF>4Gt7k!nOJ3?4(5|T`6u{%jb^@v2o5KSA>K~2Kfgft$wUq
znNk*=yfh)T*iZM5)v@DkobJnTr3X|YO^d0e#?g9-jwFc_4bl)<3$aB`i?vJsudygL
zlu&h8RYq^^Yw8OHq(r3TzD9&Hs{?}ta>D;b9D&lXIdHLdJ}Biu8`IBxCLQpOXRRl*
z)VWbPo~iUu(Jgf2deN?Z_y2uu8Mi9%@L@LZ=KuqFOVUK{nVfUx5UM4O2i_gdntUek
z31tp)89M?fg^`@gb5qfj9k6hk`WZr${MM2H0I+1hBb33Ih0TGn@Y62ZgSit`)q0E}
zX0E;Rx2^~B1~J&;)5%SI2(P|^t}KRcMRF#O^hTHrJ&!pOc)qWs&pTS>)L?6mF7tcE
zSPfqLn%aU@TYWW0<AmnAg{IS0&$1`dsL)M)Ks~o~IdO`u0zYV!Kl_4|UcEgVkb=??
z9+rBCyPC@zOqZH`9XO?qUTD%~qx=>eqAKtMrU+%$@^aE}0e2zt;%=r;E)2J>f8yD)
ziO?yU(1;ac8$NvKfKr9$i&R*fQCPHe#{b>xNGE{*&QIERgMfB+uq%JPE@nM1GyA}9
z9;)DAJFxqa-B?}yf<M};>cAwZL5?W_g#l{~WS!E5sEA`@+8ZI}dyfLF%0I37o8$iG
zPWg}r5rLu#dO9;v?EpZ>q$b}j565KKY<i@b<$(3Au!XC7Rmky`g2*RVa*Rwj`M~8>
zCXwCbtIad-?6Q-}j?Z;lt=1UIE*?>@IY`gLZB3`1Tl<QLndl<)EgMfEKVf@H^&cG?
zdK8jET)MZOM&u=UoTewOO~wUVAeoW391Fd4?eRplw2z!Dr|NaZ{F)Cd#d!9Hh3qHE
zMr&iVQ|_EUy*+1KrrU4i_xZQDWoPh_rEERxerdSwBxy-8Erds7+~ife>9Ni#K*kR7
zhwh&5Tg#%A-SluwmuVB(Zl3iYe3x%=nUik!Uby+Ol7Bv(-sRS`TM%C9dbN5Lvo$bP
zyt68`zG#2eL(ebBqBy#OLr)P1A8Zf?-AyqIc=U;naHZTpF|)Vs0fo~$s)Y0pFR)`F
zFi<}DRJ?gU@pI^|q4#E2w<LQ1by9mwZPY1Zw3FImkMH;64m;lba&i88Hm!4(V%(X=
zT%%E87rqloMmkSeL5UVJ(sC@~z<lGANKVgQK~FQJ){|%cF>Ltr;d9UEO`cHi5!U=q
zTEip9yJrr2&N-6ORU>h5%KCy{vn^{<I!Z0+DAiHQw43~#y>>AXy)UTT6DOwE`E%z2
z{fBAR-t;Jd`Jbo#;qUa;7d-}yy%6fZ-{EW9GN`)ON6Y0-hknxI_0*nB|C$+dET5p*
z$I<St9PIzs;lp&wWA7nTT<|$h<%4NXUFI-<GluOxJw&DV6L9wa`XSCEMI+wd2E$b1
zek3sg#^m!tC*S9BLDCoJof;UwBw+V*P9x>q?57QN=J)-If5?o#aI0vaT5m1)+wD3@
zOD{nlimT!=kH6W$j}_88jZjqxIeogtcS_S)>N*kvtXRdIcVv4{CYD`x$HoJgYSi`k
zMBife$3bhI>UJSZ;Ix%ra3;0uB-^v*ED=NfbKd3I?1wF%&-%`~!36IT6J@jVaX}`^
zo6~z4%vi6@xf|$vJmqS3#&#T2KS)K=hX@R?|N7PvbsYq!`>&YDS37l-TGLUg%YaEG
z<J^>oq}rKgW%L^U!~B&&cLhDcVdGbu@-;GwX6F-!4+~>r_J<w2>VtkR`Bn=>bHCTm
z%`I$e$ohnAl?N}XNMTn;AYcKMb2N3+(j?{)->*rh8LKQ}N)WF;*?JSWjkHb-1R@_e
zbK(SoyqBOyf1R3G<L}wD48S#Tjk>vR3LnezZ=?g3=}|$P_;3cu7ZAHNcanFr-{B)h
zMsX<lJXlbDzP2RD*lx?K6%9r!kO`HZ+pn<=u7L1*A}EMa0lURX;*s0}ur9tG-n!BQ
z&9hPoDW-}(c30GAgRGrjC?OI9T&4Esy=J3`qY0xc#WX`?*L-!4vMUKA(y}-*<pJc<
zT`)2k;5ze!iSkb#>X2hm%}e;kli8g)8!Of-OK)SVQ<w*H0|^<Z88qlXMTC|oRI|Wr
zACIk?rT$p1looz}bq9{2f^1(|=h<h~FQnWF77!B$=y^%^pPx~I3Pae%T%CdZ*5zzl
zJlQxqwiP&K@Vq9~Iz|rWhWbAoN;q+3Nhx;IuB~*Xu}jfmQ%ui>J@3%UmeWx^^L3MJ
zTBB|Kk%MGAn6RlFz7w|R&h)WY22ASDe}exe9wv`{+-bgPxl3I}SXdYbk8*|j(*Psi
z)2z>|@rREcYoKd9W1%pOcS}de(!dG$;H6=fVnu$Yk=`RQ!na{A=kymsxk`j!qIXLv
zv>@EQf9o)N4_uGa(4BLngHIff^VvLI^@4c?UJKlw3+MT&3%fx=(n8hpjVFDdQU9Qk
zQ}X`8+20f^5_5)=Nrpe|UhGGHAv+-H=(bz5tZO^pD;eeV-mUmxoMzBZ%9->D{p)(W
z2EzEHu~}wnYLmzn#c$zW_bpXEJo|aGlMm4TM|zdJu2tXQR}9nUXPp%4vK1)NVq5o%
zHYKg+mT<oJn9^==G$$YCJ5!uQ`QuzN0+6M1tKZ?0c{{}cWIBjER9CT1bNHpp75<Ho
zda^B{@VEWW``~vlq2R87DQh<TpX1mryO{n33I>M?6f*QwCr@rB)ZjQiz1*6tKr*JA
z$Oe~3e|by)r>TGYY*z{k-^RnxsQG#qCv`oe&b;{23qkW#Bdk~iBmO%26d*AD%HfT^
zT5ECzX*e(TvIDghUV9B2rGDiv^H=m!y0YA0Jx_&Tg?2Y;epq%|YjY*JmPUm^H~C50
zDIFUSeO+TuPCKu<mkm7eMl5&k{)Q^xeLpLS)tm$z=W1(p@&866;^82yN=3i9aK6*k
zb!Tn?C6YlwnDvIl54%f3=Fc-SN`|K5<ckFh<soAUJ^sdB`cIGp4Y0gQ{7p1&a(Hv&
z8|KZ#U})BBK<I3?aT7{^;vjW#OX^o?<STE$T!jT|egeL~MbY7Qptgx~Wz6JvT(HU>
zG_pHluD$9fnLt^TQ+Wo<9BLtPDtW73h6r-rd72DXXiEvTYK9RFOIJN$cGn92tLT_?
zl1$LHzJ}D_=0IN>#0TtP==U$_bTi3r^4;W%cKm&3y526|KvG;CB_7q+=L?LP0`9I~
zxjV$0&YBkPO>Zri!kwK)ESgXhXr$4_f6eKwACF;gey)@CguBDSIhc;5#7kFHd6VZ9
zlb7Xq%4yo~GY>p}gX<^7E={||P3kjhw8yi#TBXJXaP-?t4_in!_Ex+K&>Im*CeH5Z
zn|!BZ_kjf)?{yDQ<fKSuz{395kN@t{XQ}6pLE~2oreiT@FDHj5r3W~&IFTv#HMe{@
z#;sROG*rr_jzs9@#_?dr;^4RbdcQsoSaq_Owfb7S5+G&ZzX9gqoJy??PdtG>-!q*E
z5x4B-r}{!fk5A&j<ou20Kc?sSlbVnT4c=n+m;SgF=ROXQv6A1QR7(0rmQ=JpkpKvZ
zFl*kqZe4w*^|Rgl4+z3NMF9kOkDn2co;pXNF~sisjK`zrHZAh<o-x8>nE$+nB~^#I
z?dr74H~&F*`M~hJUC#6L{I)-bYClu>Hgl_nPc5^hx&dP<e}2m&wRQX8(-6Z__{D<n
ztN8Hha~utt`tixl{1u~oD@j+kKRy_hl5q+f$KKCJW|@klxPD)mLQ9oz`VhdZG)RGl
zm}S127B)^Z>$jAW8;y5b;InVCm`AZs?rh&I2-kaQT^cIVT3a%$EdsxbsMlARD+g%B
zmgtrSFDTM8sH75`{c%?h3g-MIbtlg{VN;Z;u-N@ueEvk>ljY?N%hA3JRnjVT=djyl
zGc%g_2h8k=M=YlKX)FWwHhE-9?dx@b2YanO*;{crQPhHyH<Ws}Xbe>d?W^KZajLqb
z(R9NNw{jjoCIQ;Ntn$t=vc?h}4_wFgP2R4u5amklEz66O@)3*FyL```7+}p}SL`q*
z>^Skgn|y9J`IZxRrU!hk2|XF)zf8gWsq9V<j-(!G8rQ>{-d64iNt}iy<lPvdnizgg
zt>2E;#`mAn>Y;ZyoofMr`hB_4s;pA*!SIIItnD?8oS)wvCfw4N`RQDI8gci}jP(6w
zAvPrg9Bw6!)82Q1RSrtw_ba%ndTLfZk7{U0WB2{Kn>8XPWxenjb@TJNeBYP%dlWot
zY0PsVRTNfQW<Hc{Gf#h+xv%MMW=Kd#zq*0W*>X<ly+}FgNi5h86Cl`{Rvbkio691l
znr%iGx{dH&q<5o|hLgRjDxY1?hYt8f@#~axZ<U;iW9<0hZWljL!8KHRwz{)LNK{VJ
zFv6<`Sg(RdHg$J~p4NK1c|m;!`QIdNL5Hw``-fE6slNzbq0_5jory`LR$OMP8F;*O
zdhptgQdy(jW{-2LckC!N^BfB{`nK96lQ?l&-boHPrC)yVv*|hc8aXxR?tbXblzQ1F
zJ*8H8-%^`^{pK1gtJic_c|O+pgakZ>+<=`oxgxQcI@9FBJ1uiv54@bH?sEZGgl_`+
z5hwP3hg08_<d27xL<slkhEWjKas#$1ORIjKc`NtyI2qk5!^VxfR`Q5(gl<js-@(Id
z6PH8(V+c9h*)%-SLKAB=)2A^r3U-zw;X$j@yPMh%7bFQcV!7>cJ^@*w`3C-+`s%hf
z$8T9{_{W`*8+zBx_qf3J69)F=V*6DWgN>m+L{=!Hp|ZM%z5cTzNmZ$YAhAI855hgw
z!(3gVCU5Zv5pg4;p(dA4zWbPhwfBx41)Kup(mfkya(bD3ZdISz+jXJi+0^+RNLzMQ
zl@+~x-z&+yKJ$p$f866e_Ejck+a@c25=UlaXxRRxGY{@tDzmM&f6~6Wad}Ri?Z}hF
zmdorkdIh90RTp;1E5Nb1WuiBLR<5g|Cv46Id-&q&{}WiuF)#WpIORkRTbR`)Y`HF{
zZ}f6c)tYbUXH}A^J?if1);_0a_I|PQdqwqZyYhMSsy`23NJ1u#sH?-7PD}Wo=lp{t
zGQIBz?fYwKQ*!t~!Z<W9NOd;w&(f}4C&$drvZ{1STY0Si1cyZh)e*WQ!6XLOIYo46
zX7{y7wW@@r+{raDDdyzgH|Le09=V7(%~leL8Mo|sRP`AmRU*Gt+4r+`Z{wZ=RpVsU
zVl!KZ?$eSQQ>_u6u5;?@(yBv$t!vBWdv%lFHwRSdq=#&<+1b21l8-(gwkDb#r_B^{
z7=Jc7O&esv8I;ye|IF-1maSUxhr*>Awm8NqU*5O2xK^K_kXO2stf;d#Ym#$9#QKXf
z%4ejRbuW6N^XX*Ov<VFtjnA~uN_XMHcj<k)kIp{rTC5UKFRd|~V`bloe)}hH+S6E{
zlK5mPg)TGs);laAO&R)WD_0t<OBq>oB;!<{ehC|C{rS*peXivkUiycBx-C`e8Xgu_
zQ}I^UW=Kl9`{gEsF|$-&sHB$8DETpd{}^-3byuO^Z-}8~uUPNYwTB{I9iCNq_PRM#
zEjg=BgSPUeY2mpu9&cC9x^|nTy)SjW;Yp_l1B5aG7)=;zBm>qR825NjN>eh)4u`pt
z+lXyh;#9|_?t{@D(M=2xwerd7RKz9V7Mi-cx*mo)74wHV9WIIRQJOF3R=D%*IpuFr
zp>a}SKrYjZ7VIy13qBDrW%o4UV5@j-NrBPJF43wPZ;jBlhG`X#RIGbkju)^~kaUOU
zUNax-s7Ck%YaMm&aAl^9>-M-rpN53E3)+{x?>}1kV0)8Ihi4>oRb_;`#0cYl&9t1|
z)vCYFh!f}9b*6{yo~)~~=Rb$M?o6NaJCg4TRNdUa-`3Xy8uLS|nyY;5wC8?2(wKXV
zXe!Ed3{;Q9NdmcZS;7G6Q<{vVj|PZqqU+J{bV}lT2^nyN_0;~LOlF*_0(~OPWp?S1
z_2jckmgyU+tsje#f2&rwq+IQtTw8^0Mc13Umm*M!zGi=A27)$@xOTy`V9j8lDAm<w
za&_4aVo<_1i^$?x`J4hnPmLmB_EZb2+P9W?>(tu+#P~J6&2w3e`D;<QZ7o<?{O<f_
zgw-Kycc~<h;}}e?9<{jOlj8$*`l)t!hpSz*DUb=?@#Dn%_I;;X>Hu~~O{%)xlm|aA
za`S)*XGjC<T{vx*ss;O=&MUlAc=qIIH|5i*ORGZLGKwB^#GHybHcOy7Kjg)&I`Lt@
zsZVsCE|VqB8Ca+yN-scSmuKhxE#=4KoKco6-?NQp^Y8j+WK#?CmHKqM68L^5TSMvO
z#%V44*;X_3zNbDEfM1+XoK&mo4bY<G`C+YJH3O5Dza}Y}-a}{j4G&O1r!sM@#`!3n
z+B|>$kiI$G90lEfs-bZjNm;tSod6vXmmx2f-yEgut+@y{miv8VN7f~l#=VaN0*~t&
zMaG>Gw`vZkjMJ(kH4rCeq`<(1^Be?<7C7>Nu2RC!bouXOp0o5q&i|2&{P123y$vBv
zbL}|wy1sXzFb<=ZnRH{`*3VnJi=-f%w5#Zqbw?>8?30q>&<lVY#|tM7k(Qj?{PS5u
z%(@qmGL7}&Rg?2c<x&3G>VLypP>>%^!vp%c?>ctI8OK21#ReLuPg#A~e;x4EI3>86
zJrT0}1ow)&)GcQ44@!8YhP~}@zH})+c@Zq(Vq~+xD=FAUKG^*sdC1U%1ICT0bzKQX
z1ju4)k+%Rpv&VZ>Y10~@z0uy=$94+`6<`M(0es5%iutR#Nq)9wrF(|?DQ`LCnDYIB
zsev5x<zw528wx19VE;E!EZ1*19}6>ff5&&eO2Ou5b4`8Ppy=jd98cQ##~Q`oxjsMt
zAi%Vt!-pv(z!C0yrk#CC8LD7??YgJT`M>+=ch#-UUv~bngIXk@`}mX2zYwh{5DbQE
z$r9Y_k3<QC7Z4-PC3`x1^bqJcF-rQ4-59s0kR=~O2F($$RMY#`=J($3nlq-f^wQfZ
z&vlhgzdSV0Xd6yXUDtksyHfq>ZC!HyRo7Je)4c<_4$83YuBCX_-<(!oTkpN<F{!my
z)%{MAc~P;vBB5x?6QDCiVo=jliF7+jKabH1vkuDXPG|^!>thRC6KB01w^)RDsQj$1
zt(;5C9>Wv`YeCI7uJ&pJ1(58mW9Nbo^dB>{_KPj%nRXO=m8@hU=UPrsXd%Qk!U+!8
zUSw@#u(lLgDwCEO&f@JLeE19Q*(%`&6sEL*mU9-~BSMHaLDH>cGFc=y2i)K5>l;rU
zM2IOL?TUB5C=Kv%V$cg6v3kxWnQ1LGF>!Amtb8@+JZA;ZWtq}wa)nx|U4i2en2s}x
zl-<H)czg|XOkarObnC*cxrR!L<K)&5!{e4vst&XHz{44({F?lGcw`BnkjCqN@7BG*
zf0-=mg93B5YJjGsZ@xgiCs)y#-Pxz^=h!PAKY<$?*bsk;4MFz7hR6#%^S(!y;T0A@
z7<RAQ&9IyI1=gToBIZns6Xz{lxX}CM&{Y=A&6S?U1!h)OR-oy^#V&s>qn|dog$i^8
z)F}NWC8rT5adp{vV05yTI2Tc%M`T1}j5_#+Rmr$adhX8@b{%mZn_0onx_*0YZsJ2R
z$&K`#@*N=IVC11lIeDGhRPd>y8&$v?2I|ursY(_1g#T@@9z5lP!6FJk3cK@NZoMTJ
zA>hA-5uv!4ky=Bz^eG+yL*+dy@F%c9SaSl%l=MyW<^7icryQ_uIKD1=?-f~<inw{s
z%^wrR_sj1HVLQND4A73#6nJxfTPhK`<<0s88Qr@Sg%fKx6>EHb&SRP;WuU(=wevid
z0)o}9YilE;qE%;|5Y)ry?iW3@I36iGeOmL>0|#wLJlkKv;`1>>eIS>9jiWQ*arzG1
zy%f8@ZTjxRj!Fy-A8i4G2WF)HIb<ujf@$)vq)(^SB3@nnO4OPrFyRiP-7>@T_P-uo
zXg5k)Qt?Oq9+sYVA>Q02#j_7?*HLh#Bw(yFNYolu<QWg!)7^0>87yHqL*J}A3mWtW
z3NRetqM~oAI=6eyAu0XSDIn_2c*l3b#I~N5#uRnLI<4e-wPksu(l!koZ9g^9s$d9-
zO7cD9O0vEWbhi^0lS$$4Zn5)dNHryi5X1)-h2mI3(}lj2azm($l;$3n-ulC%pl02I
z_<KBPxvkAjbzm{{aR{uBLXZd|RJ#<2Ptp{Xs;8|k8sd`WZ}t3H0XqJC5zWYgP?<e`
z^yn?niI_J?(I_6oB;(OrZiiG4siBv1S;$-LbDfH=>>GdK9ZeQRd#5=0yP#z)zwWoZ
zfc5EgZ+63>ZFmaYw@bM^=~FSi$-fA9N~^c<_|h8RR_=Sg5_ECt6DgxhcXGAIb@7=d
zNKn&HX_!h)#>1um8FK*V{4W-rWH(ObjmIoizA^Sm1`c`Ou0NZ%|JV8?;3YwSa6v0p
ztwNgj6*Mpj)3sq|&Mo2$67bnxYmK0`bDEQSFkA2EwgB(SnrYQe?Tz&Op67@BaH73T
z8<f6xYs_|#{Nepy!*n!+?&>k8;4a%rR%eJ}#i5DPk_TmV#_A7?JO8kKpl6~R2b~y@
zm+pQEP7R%-&u%ARSmBP`<(9)*e=B9_<BaU8pXNIDkN)%F$Mb6zR&d~_pO9WNTTaqP
z{CoSw*Fu-!`sg>M`HOK@Wi&7>>C`uN!`?Sm*}l)unkWN%F*YbJZ)Kkd05us4n>ciM
z`7g7l>s7!cjQshxs<b36%$f}|V;Bn^GI&4;_0gAy!=)wXLZ;Pz>C-{y$HQ?64zK2?
zNU7EyGo;z$+zm=kv+uuerO&jp*hYxMCTrMXEMdvx+PgmRIc&)3(le`Yh@sd6+foM$
z)V?f3-%yOL5P<9F9fzpM*<v3qg`b|YQNtF_66HlqXuK6OD#!&~Ve6FoJftOn^XkZ7
zkIcU0#k|LjKuWvNYkqSuGFObSHSUZSnAX;s7ZyRDL)}?|RID@m+5TjP;Ym>Ky;Fb@
z=Ww2S@NTS><ICL?dnIqI>^dst&})CRL~a~Q`)-<3Vzhks_e5{dHEvyP24(_p1Wv`X
z=(onpdK&Q&3CkIvnXyS1NKu)ITExNQ4za<l2re;@c7H5tWnr<vR{`Q2ek>aV;9n{L
z0lQ37gBjQahKTxiV~O0XIE28pd1);!aAH0W4pnzNAG};^_bH;;;_Wr&XFb#44GKB3
zT5DoFXGX`#ctPw5E8hS`#MA>(2OJs`+y@OMn$ISZZy1Zf#Fa#BQH;Y-q@R9%U-d`Z
zsvYuuRU3IS8;J1!VTZauhd*UYLai23u%9N-L!qLPIUEq+1RW&q8Vs%ziM8bV53FJK
zLd@ykzZ~Ovu$FDB0N6&gbQ4gJIz3?da84630ap;cq)8s!0p!R($ba_NC0-X~X-?}k
zd`JOtgu{h!XNL}<uy-X?gA<2&wzg!Q{1%>$>+BnH1@!*(A-eMioMDc{M8rhKI;Fw;
zHEu@qUlQCDLE$gQ%qh{ADDXpyV-h7TPXj_)6q!B{jX0J)>yqD}!gBmhw~Iecfp}8+
zF;90*VhQp0=oHTulCZ;({t-~p>l5BCyr8@A?k~PFY^p#!cu+#hObUX^;W~N^t{Y4L
z7#^xfs?jda@4sFQBqtN!KpibkevqHy-)zMizVYvP{$Xb)`FxHIks#1Gv3AKt<+vxE
zDLRuO;P%PoL%=pCq&2T+|MT2mV6l3OxGGL%o+WQhgYYU2WEQ0RQ~geIWY$7y2~txX
zatA{TY5{?Fw6b<RP;3faCshJpl^cBtPJk{{K_5QWE<Ql@=&Kap;iY}H>#_?V*0*49
z!Np5OP$pmq9^X1agAQTZ`MHB-Y&AZfV*FH7(l*Idb_Owb|41E@?P5LE%?kg}5(GNC
znm{ieG37YsQAq?Bf-A!!LJ?&9%5BLfvuflWWmFQZKTeLhGhO00tv#)DOI1%pK!ACc
z??Vkbn11|W{^yaV<&xR{+doCz*K7VJj^(n@S{-77<8xDOV9e}^5e$b-9>nzzthq1f
z^}>6~Qa{Yq0=I_`ikw}0>io(o6Zy|I**EgK9*db1`_Q68#A=PClE4VVspz=hk=j_^
zzvg&cXVc0Oe*iEys+1hnqp5kfk_M%*%6aTlu3q4PjQgZ2p#t^JdsQP$GeFj6Hl7gR
z*^sk%s1c{}FXXGi=93yj$$NyD|7r>J`v!6+-aq8^D%WCw>hp6R_{mw12#{#(vCDUC
zcwX0MWpVy=MMrqG{0yHm^NNRX@PLRs!uuTOJ}Nsqp4oV}ix4&qc)+`8xfTOn5flen
zI&-El(d^RYqIR`%2<D7pLX2ZFZee#B`bQ*f@_$NA>e<Fs2)bm2%#&LyS-l_f#IB^&
zJoIE+)acqrZw1>5B<|}8<*=>Xd74BR$C<;In%tk;%9l-AEEBp^u=Uf=nphgd`R|jR
z&y3iQho<*Xxo2N%*Z6V6$+1WzRCoxI_6jSiOm_d=78`@^&r4d2B*>Zg_rBX3+S|WJ
z5x4zdgAfr*^ppka@S{%D`02^=wSM{GkhDQ}udx>fy&j~qRw7tf1M%}eBP!%&Q-7PO
zNetHht)mq0{=ffUpEt|0(``Kg>k&d!%akR&hT^JVuJ8EeWxEJB(dYAWNUT(V|FF(N
z;)t}!@};SS#3f*6gJ74=5)$rXeh6c?|M6WZZiu`vi}V<3Z4qxV^*)d_xN=HnAdpi6
ziY?Zb*LDM9K=#D;s(XGy)V~}fA#vz3*{3j(yDD^W32(Pp5AgX@Vrsy#-v*%nS@`+?
z*=O`WA$t9<YsXCFKdr)(2d?7k=+#<%`%fvN&CO6(s^{}_sK+^(qT6>D0vOJlOI><M
zc=a}t@|m_m4&8cd8qnSjO5aOsuYmGHnO-7NMGkB)5&Ac=Tta!VMq;72^o=bqEms*n
zz;r-Ogo4r>;3)U}CB9*i@y}P*(cu8ZD&gg<WNvt!OidiRigVjZ98zt2@nd*epjo4L
zkC{SkLCZ^E0_#~4>9ItxE(wC@&dYi;hF<GX*+9hf%Kmds&(SUP`cvdrf%Xy-vxkUx
zum0@aRU-m{n4wjFO0BRt9o_jRenSRIOx=1N0FJ6Sez^Kk93@Tk+TMGCqG1Gw;S$)p
z&62z5nlKlJ3E7uGpAvw0@4wjsDsZWLM+q-8A*NOiCp*fcCN5h4Ce5?;c}Kovr0D$)
zUVQ(=Rfbn(2*9rN4m-4W+jZr^5Ft~3bZ!F!a`-C8P=-%!5Q;f;UQx482;mE2IrT8N
zf8U@h5n9QoJgT$6!A6B~W&gTw->B3PKP{=QnTv^~@~5`%84txed`jzhSQ`+&uz|R8
zDh?MtweJ)?bF9f2z5={N@^wt8ium}qT}YD1R}>dKTW-$nGS=&XpN^5SaV4)i`+_TW
zc`#sA`<~)`gpp)yzu6xjB#5sH7uT*mS8FJ)w!DKM)g-yfi0E<lx4rDW+ew5D7hkBh
z2bHIB)ds3our>Kbqa`GyPjJb%-!0)jf@gO~b$_?@6Ck2urnq8x@z7GI6W#=I^)&f3
zPbM}&I7jENlMdeeuLs<Fqj(>G@jiCN|9Zd<4Rlkgnvad<rMk<5T_l>OfHMhd%+i3z
z;mQO3DJDsp=ab=)=N&Pe#J4E&`m=|S!CD&VQ;=9~gZe7#FZ{?Fp)f?wD8e|6HqWST
z9FxD@xa#M9ef+;d`D}3qe;=RXpZ~u;AOCx?H%qyUTQf(9EpZ#`Rd<EN<`e$<A3tIm
zhMJI{b!<T@Y-<MIi)MNtG@2AU%BpkMPxFLhE3={LGc!s5t#~DsiEK>;>Axtq=ANFk
zl?-Y{Z+TuED6*OFFCt8r;VXsOyNP3%e!N95Kd{>M#VVB+c6eg;G-@lX{&ObwWr1!J
zPtix17zz3g?BhVUobCLLHV=&~w#)n9$rLZ$z)NrIMY6G2xcrF1p&;=)LR~!;y}(kS
zGrQavsJ*ces{hZfy(PT%W&GV$!hUfA$tpT3L;}KX1X=r6me&-dfUtj?5W()ZZjDZ^
zpl7@#9-YSUH|VI=;Gk%X3=#3=co1lmBn0BO0HLVOL=sF8Nkxl`B0;&K-x^>N(M;%C
zxr543c>i%<akZj)s|8~XI<SR*DMLsEnIS|l@ZO?KqIGA~C`WPg8E3#ZVw91XN{n*e
z6Fibk=NH$8bS-F&PV6a6hx>Il)pMX3|FUO4iO{9Ik#)i2I;7z9VqZ|B)j$JCKX;mf
zJSNLyGs!=r%mWePxb+t@dX!i(Nang6pICfgT+s&iEX9xuh{4vs{t4Nx8qrL()Lc5s
z(@=ITotCKC_%6MD2bSJ=v!ScA647T2Hhx20eg?}oBv#}1CNanGd$rS=HEXEc3*y&V
z6bO(INyZWTHyDVDWL8T{tKrUVE>o13GSBV3Jqi%Eq)~QGdm~2wc3qpolG^m~p`qUM
z*z|_P;7ZOff>yRv8EYlI?hL4DTkFTdS+gF16<R;tO+hbuMIrtyF8Ibx*ihgGYTV`c
z!6+IJI0FQyepXXhUQxnjY%c={pF}da|7Vhqq_~FiHWy~e@w^UXF46H6#d?SJE)a==
zdg-qa0=SF%`A5Wgzd1XCxLFhiYFXKcr;QyxEb*0_K(SUbWsISIMSFteesOZ_L@(Be
zBIUvMDAu-9j<N2y>+7PEMpRg%Mq#_GW37Q-yrj)V=uxC4XTIR9{)Ien@dfwnz_z-c
z;S`ahD^lJH?J=LP2N6d7d^#Y&6bOmq>jDC9A=!cCjpIS4E0~>j{X=5vE48vN*P5oh
zq_mhnzr3cn`YnJ~2bcG(E5FE-$P0lxGp?L$E%Z_5_8>j2-5u}!(my8w)Rz3=7XM&H
zPcAt$y`|2xH6ipkV#rNqISVfSwfBl>Ms44CL5QtDgjkX>Fs=EnMvh`7ENM=xo~{?2
z!l}R~-}NY9Q%HWjh5WIPc@i7VX`OJx7#YzFhxp*7z*^5mDrROt<jWzPH?8S9Cre0}
zO*KUwpTCx#ij8A+m7;UsbZI-y4l8`b1ZwzI%3|+#kub1khMb#4F9l!Mb!~~zgQ!oU
zR2RhIaGAwcuUV`@k!`-{X^=84lfBHZf;Qx@1&4A(PZHQ}hDuP|<`MbA?BSCqPj1K#
zKj}Qj;R8GQ(~~ocpN>F%y1jbteK!3wcI&bnp-N9QKF8phd1i2%MoXhkxxBQ5Va)-O
znGDQ9UI8Lp1?ql<%xmnB4r!R#)@m%w>L4-u0{X@dsWOd=cCozgFZ@5+`|@xs`}a+W
ziZ&s#lPz0{EQJutzKgOIN<Fq@-?C&&3sICpkv&^kB3n@*YnG5Dp(M(_M9g{je!s`e
z{O0)0AM?k|F>{PR-h*DA=YH<{^SM6fbzbLnUhw_Y+xPW0WylfqCBzWJvYvtl?4Hrx
z2NCIRpGeaQIX|M(Ww|yJ*UbR<;M$>gLgnZ;tQF`K*HKUOf#6~QYd)yba7tza*_mHo
zh<<+M9VnIt4G@B<z5XcWlI)AKyJ4twPqPTMIEJ|{=!31dT01SSAP%E&kkDwGnx0NX
zUt@DOf<6Ih69!+%+OB_YkH_EJGlGQuhG`|DUB=yyNJbyLZqd4AY4HSX&hELBeJfwT
zl!4zMDhE75-%v>G1fNOoV{Lwa;=gVQ6f51KVtwQH15%!wf@+iPA0X|C)+_=E5Hx_6
z4xQ+vB0yY3R(ynw1>%?JR(@En_oA7lE?TUI3fUSX0Yk{m;c-Nn8cZm(GNK6?-q+_1
zk>Fn7WKfRq>;JnwDG{_4Mx*hR_I&~;#<DzYQGq8BG~@+rH|+*xB~*IXSGvIN6Dq=P
zqA7_3Nugc{f@_3`a1v^6|D@|_2!^mDslW#%ntTvCe33gMm$Qb-E3*Fy7$!YlpA~Ij
zBnRTeawc^9pb+&@py~4f7soz-AdDDI6w!2q4YwA=&~?))SwfrvrUJ~9<=6Flw{K4a
zG!oRMUMB$rK>^qKd=dOWyy|vh)vv~ZB{lQtcZ2fH0gJbf8ezN){L^sd{96{MS_asd
ze#bYOYwb=DW_Ew$xl@W0T^^-O=fNy2#gWL}p4g*Lf{Bwmk5HqdqJhxm4rWAlrjA=4
zq!bZ8LE{FE52AF6sY{V7T@VdmCvcb&7Yg@SS?4`i<OJg%D{6TKh9pGDRJDA}gB`O7
z;*wBdQ1)-z<Bqnpw+!%S^})e=o7ZO;A7l%G1FDaX;}h?Hd$L7#wh_WcYHule>VDKI
z;KLFwsH3ApG)O?5O|*tV7~u&D4<8gwdlmqyVAnrlEyQRa$wc6K6P`XoG%^0|rCRxk
z{vk-&VQ#Xs5~nXg|Dl+Khl9SeT!b?)*SJOx5UcMTE8zu^vAOi;O>7#@zhJSRsLH}c
z;m(U@GzKl;MJ(JpjS8HIE*|XlRN7O5M6)#<77_M@KM+^1X+JdZ5=)BwOr;>1!Vo46
zf~_J_4cvCd(_W|;;P#l2>D?{Gt~-h??mXD*;g0YO^*s>J7f=zPXWtXM8#`X}4_^2o
zp8rQP7cDf1Tzq#j6h%8%L|9u}+jo8*9K`Z}L$r$13<6|<)#^oqKejSa08#0cM6u!C
zfgv1F@zTy;;OZ6KM*@S2OX#CyG>ckIQGH(vG~o!O3{DfzT8ApjXadns*B<8t*@8Lv
zOAOP)7eSYz+IjM17q%;&awU3c2oXIfU#M0^)K&<24FbC(LX&~^VNjU+i3$(y?`hm2
z9Be?5uvL@;fLtF&E9Vq3jmhlAf$^R#8$2<+l4v*teCI%C6(N8Jlm~J&`>_sTCxLk|
zu?a=5HNYDNQ43jI6?*B2X3hGA#Ryz3&67EvbaUO_D=;|&R*H3_2Z~jAa|jsZljF;r
zrzxBwO}>QPKb9L-I==pAt+x2z|8KHl{~78hLOE!Kzb0t^=Rdnx!x6IR%c$647;5Fy
z6tXiMm!U}@Xu5*Y6Oh_KXu;wuOGe`k!2yP0Nfwv?RCBXqHx^&=e8k!->GbKVhg6&!
zJ-zBt$6qkBm;{2niv=;Egb>P6QjVjqjEAfNC~;N1r=dvDR<eGZ^qzf5f*-TYy11gk
z|0~H#tdsxiv&1`=h_xE)^4LE6e_UEPOvpQa`?EmAe`>P|r4%4)on<Lb5kP{hr&ohM
zk1aLC-ioYs)!*GR;=(Ul;drs5>bL~+T{kn8*cnm%vkTh@4n%<!$@+inC%=<OoLofz
zB{(1t^x&b8UJgnE@tNT8(KSfqH%^tye066WQDQ=waw-zUDNZJqply$2QS)j8e)$Aw
zj+_U+wE6gB9E^78ibe2;>*zB)%V#@i795XMQriPA!^6DlYrLH{P&9%EJ`5*_6hH+~
zFGS5K4C+k8YNRT=dv9q|bB2gf=e@2x3qzO_j>?6W|JDK^5nTHL#9axfa*k!^og8@8
z5O#^Tc_9UJz|r^RBmsViM4@{I%Lm%nLcJ97yeCB$+P<JNc}uXdp#d#cj%`Y5byraQ
zFm)^)1@1)A{5D};g8yR<$t;c#cXSUC&E$aR)^Kk{QXAv6ie@S#*LFm8-W2B@DoV(=
zkV<4jb4%nSL(w>FEP#EHI%%$@2qhAfHMHpKP*kNf;0y#AHBMBDkn|9FF)wsX&}R^q
zpPE6z73MQ+Nfu5{(|8-icsGDny%Um_knP~S@W=U$PCPC2d857rl`wo#VSjgCar@8V
zgbq8#3U|+x8iUSWEvNoNey`>Xi#AQ%gk8ycm4HE7@ULpiXEti5!j$e43N0CTdmA-8
ztgvlN`W^15<uG^d7=4E>m@X)`A!0rv$i!%ek}NV3a|l~!ZaVcm8wEDF@!*C~2Mm6b
z)rloxRcLT<WOURV>j)*BTUu}5bYYCQ6^?_e<%xzHyl+xZ?N+^Z$JN)nVkQmeE^#CK
zY|NhGfg@G^4N<g?@5=%_YJvt#Mv?A6ZBAFckMx?5k;%xsywgiyb*PXm${kRrB&7Mr
zV;eC+pw-?mq!5W3a4&&|hN96Q>xxDioS}ra743Zx2Ee+_1&|v$5&5duxT$dTSqDFR
zN3SlwMBV6J`o~Y7fXu6~3Q*@0$_V<`)vdh;YOspS`;QIjnRAd2K&_iL*NY|c8<^NU
zn!(C{5M30;(VG_aVJvhXos@suw2@`mVKt-sUjtnL%u<dI@89o9R=ExoLCCsRp%Ees
ztXAW3Vlo}|ePP7N=_N3+H0Qm$8p<oW!EYHc<f>H5(gEv#d(hQP|CVY0`yO;vqEXBe
z7_(p$TY>qYRV|VM?6x?x>>%eNbas%MV`T~IuMlt%AV7VO4Yn+y8$3n4uHR&A47L=Z
zfBf4qPe}N|E+KcR&FO<n39-D<*EST*r4+h~N2o|a?Nim^PCVo9__2ymlHi5^QIf!1
zY62fk_`<8EFm~??==Y(NNR7!-U7LHIbo!><?s=y_3-jmL7$@jtTQ+>I`XBq!SJ!)A
z&c5KwE!z2+&RuaqMM33uv7l;aM6#~pE=uRpFS!{eD)x4I9T62Dx=vDe8Z+b!2zuW>
zX;<K=`-*adkRYcZgWeD7{Q(m%x9}(o$Y1_-NJ_jXhN|w@%5>x<X>k`mIxjKXX0K-4
zwb34`eGMSVx;}i^t2+koKj22hnP|+YCB6XYz%f-JT}Qa<=%|%{H9j#x82$M9$(>5h
z$q}h}Y6Aaq7$rh`BGcpo$cJ&JEozM&_~oSJeS7y>fzPli=>?BOcx0HDDbpzX`TfC4
zIjFuvfi_~hZ@9O46CM{z?L7G-K{+`Ny1K1UiW0yR28DrI;84-HjOw=H%Nll;Cu!@N
zRSa~BG|wk&RhE^h=%2}SD_PSQuWckN_iN?YJX7Yk;%;wGOg+JFfizk~M5Mp3j}V-p
zTC1D&=Td23VPPTWEWq;m%NHvxttPlJ_;`7VSXTDEdZnm#)Al$%5Dceqb8sA(nw$Um
z+QhPP<3_+jOl)#;`h`60``DOjrPSfUDcE?Tyi9~Q09s7xWZm%&pBw{8`}_UStU>sw
z8>Xj4WxpN0Oc)VkJK;zG;_mG|?dRtwB$Pcp{0{&62_v(!`06vWvc_TJO&)>+BHE<{
zR1`!W&_qVN$Ugypl9rxsi`D_3g{Q^E3*i1jLqqY1&;_`LFbiD$8}hC9&(|9o8jxi0
z^70bD`S|fG^t!{F9w!!j-ZDc&l>@oKV~L1NqRasMPn)fP&S+JHYSji4ReNTp>%SA=
ze04fZ;c-GjKRQ`&-n<#&KxkIMKWIE##|E@Le?D6N7L)kfuCCbiTPl|^cY~Y-j_$4q
zANo1dy?$)6p81a-2QGT!oxvwY*$6RQ-_61zgOT?Z92a0d3t!NLgald^K@N(eR9sU6
zcg4t)UZ=1JR`bUo=y4ozt$XuE#e`Br<2|A=91=r*XM<^^`Vq<P?HgrPO*6|i%l0ku
zyZqa%lDoOFKi~h3+ZrkI)Gm`-=-ycTi|wu~XC<uim(Q>$vG_Uov1uhpO-SYS<h_5;
zz+ou&!rWVCUTvJ^R;%+ZYMU&(BQ4GYBOZK<QWJT3IhG&wJ*vK38IS3xt)i7V^5&_-
zo5pK%QWI5Ee?o>g-}+_OYHX_XGOKi4dpY6GW&x%UCbMIWWfuBOXC+>{1^gJhlCpWv
z_5&RUJK8_kk?gX(3}%f-Ek9bAy<8fxGBAxH-?_P;ow01w+BCf(GGZh5jmR}M*d4J;
z3f&+cR_n^d{2oU~i*nC`e91hs(q|PF=G{XS?nV|C`zK6`LPOu8YO69a4h*w7?<+`Q
zm84;aX6pKpvc|0&!~XKiS{}yez(h-Vjw=J44(IP5ao4V?X=-kJ(8RE^zkW8<rL+&W
z!pK?M24ldZm>U>y+wY_<Cp`fovBneI8;=PDYa{h^Y6GN@N4Iv3eJaW8&Ot`Q{e|8W
zGUd*m9)WzW^#I|u#gnxKII?-`S@-Jbv^;$(ylZ88W+vIu=;X<_GHlpLz5Pd|rI8J3
z&b@SAm^c*}Hf4DH_;D-~J3BkNl>xz%3JPRJ2SGNRtZN1otCxT948aC``26f_3j0Ag
z)=dm#WoKuun>zKErADQx?~|6kw=i>9Mn-6*6_Q<^ov<a;%Phu$y{pvW(xn4ML0QS0
z$T_G2^ImQ~%9p~<7^qW!L`LRfl4?kBaCmo?5QRW*B;rS!xaQ_b*4nA6?Y2_7rrliw
zd{!}yw)Z=<9{RhoGmFpF_S1D(mQHvu6}m*#><?>c@w@VT<UGIe)H<V)_{-($n;RDn
zE{GjcnjsB7@t+O-6I#3>GMoJ++ocq_{nq?5+9NXZ7V8BSF>8^t^Kl>hcfNQ(+Z?ri
z^5N;}Z`txHNvcYAN^?bHuiCCMt=PmjA4px}=qYo{?p|cQ7r5%onNnOV>Ez@jR=U#&
z18mS6w6{K&nS}-Yk^Ah=rY$Ti0B8vV<E1Dyw%^^q4*~MPfhg8d#q8o@cd!Sm)}#yd
z%b_UrBeH=bb16+`-vC3^UU>M<9a)s6E+rvdot>MbPdLXXF5dp;%_xq`!YVKTqCcue
zbb3{2YO(Pt)hQTu#XPK@ds|*E+w)fKHexY3IXR{p#Uv$(7GL)EE_?96R9QKAOh|k8
zU(Y=rVY^0@=MZs&<>b#i??K5o0r|C~WA0c=$H$KZWQBdmK2g!QEU(MzT3WmJ6%-cU
z@n7Xuq(K<M<S^D9VZun9BW1atW9&_)wL?ZyOUVbs*hG2%W*-L|n?pVe_L-93ez|FB
zXc(NSq2Cg?Gf+o4!I621o_&4`#nI#WCc?eKml2U^3ygGj=&PqbPIInatnNE|i0<tv
zitC)=yxg%956jv!^XPt^XmgctGrnnYw~(h^^_}Xp1Mij~&SLG}GPe(17V&dQu^vqW
za>Y~k0*g1y8#g`-XV!4|IA@+D?VsGNpPuu%Z*Y0(%kWjrO=?A^8{)^~P34N;Hr--5
z%loxHFtGTQ_s4mvt+b8vjiZP6mP_Dp5~JR+gP7_N8`}jm4dL$i_&9DUsJxQ1HJjY#
z*H<fK&^QQd(b}IR9297I(KRsGx@{Xq17Eyw;R1<N<~4g>SC^(V8>Wuo;ahkUBWdB$
zjDujt0vl@c=FOI+ms7K{B5H;XNl3`~{XT@5OUP@ItA0k&&xaDCNinf#DaEv$oE#YU
z!l7AEu;K07P~#y;Dp4V&{lvbU*Ng9gk_-m?uBWA?RaLEn1`-q$?6!AzbDPKabr#WC
zS%mSdk+Jc0hr6r$Eh~j}_byy3Ie<wmtzBJTMVHZtxx8Zc<U>!-VMx+mypXP!!jXx|
zUVk7c<QcTV{s*hR+dLf}gaE1?m-olPwNhVglU5XZsNvtNGFT#{3nPBa5>g-}{ilRa
z;zS2VAIES&d;6!Uspl?U{5~>5&&sO#M|KMxD=Pq=^~^hD3OLBknl~B9`;<3caH_zu
z8J&kQm`~!nRcV5=vxN@pP9qV?%E>XZ8K$QQ)QPU(nsUplPVde*!8vs><LT2Ir={V}
z+h#i7&6#TWpkK;gTA}Qryj6wW!q2MnYZ?VBBUH?ch4v$2ZEjEJawK22uP)vW+{XWG
zsAhZnCLKEScgI@xUePYk<@>YcRDlVFPy;`yXS3D7<}2JeC-{UlTR(O&TJ9cjAB)_`
zgiYMo*Cz{^E-G`t@DgHTAKKcSFI@_ci1?DU#98lg2Ry#HhK3MQG)YN_xC-E-3--+Y
z@@1lq_Tj^aI;Tm3)<PNXj_VN=1HWD_g?*?E)uiEVC_z;#TQGq@@_`mjaD03(o*hnJ
zm@GM+KcDvaaT}a<kXhhgi8<iVRla)l3ac@Bk?$KzvsD3dR5zrwAgQ6<EGsKpz}$3H
zv0+4tO9ZEhOMU=EoreeZtlM$R;k;hJgL=j7AmGsag9Q|W@B@H>HI$i|iJ?GNR<Jz(
z<1}C&j_U@#zUCOc8W9meB9SPl+1S|5eR^>7=Js_`B!wcDG}=xSiYxq({9!D)nEf7h
z_A4kVQ3+uLkWMrSzVx}-*>w9>p}phLeD?MiU=)nHr?($xFTBbKe8;C1KQcaEhPDpe
zTG%+dNI%3|!A}ek4u-VOD66fuFJA;mnqHTh0B?;owYoj+$^x+26Sl=Rw&|T)ndQ6-
z3m;%x#W=CWy>EK|-oWJjQXP{7%@WPhK;^$z40v9ljy8X0W@f9|JqyH|5f9zc)Pun(
zl?9U4{SN0IPH_bdZi#=wYFN2Y8G0l_EpgxEwbomQo{Zaei`I%L7xU#Xl^nN^<G7ns
z;P6D}iT#P-5~|<B3de;RexNqPCPAW~HwKpFf1Zu8?iw=}ToJlqrrI%sV;@=h^~Lz6
z^D%iZLxO_H2ex8P1jbEm$v$sy&q_x({PioJx6|dyArzqj018V=7>oS@zabosx-ZSM
zOfJ%5v~}y&VXbX#9UZdhH^vOtNiG<MsC{*o_ny0;rgp9SWc@JkV8^k7mM|6U;N7ap
z==3YbEz@|J1o(^nNGeTD9fkWVOldp2x=z9(nVntx!5SYA&l7-`nHtJXn*dHqii(Qh
zNDd)d2Q|VEc{|zrhK9=ZH6N0ekIxEL|J}H_=RFf3?O}+n`|b#=p!m4Cm1e%y-*AtB
zqx8eEZ_9YsZ#k8MPmi5C#k5YLCLvLTwvccWwquu-!l?H*bai)|7#i}dC%$~q(4amu
z|48wfCRagG5wE4Ok&(b7E%kuvu<8>&;f`!e^t*|}A`RI9$ez&CM=>SE#pU4acW-;3
z%j>@xA2d5%zUjusEbjXH6%_}+-$RRg&6a`vue3!q2jeT<e&)Hnn`Du)U(8=j9$+c3
zt({9s4>PA0au81a=`wxkajEpE>(kHO9Z6|3@$r}V15YcODw%dalHa8Lq~iIXcYPfE
zZ?u?c#cld3ogXtQ*$uXCduVyc_x3gyRgpR2qt0n+52gyge($|g)GU8w=PmXL_iB+D
zl8vd)0|ho6868HJ$96diI|J9GquL-PP?nYAZoi28D+fdBxm;vF$ZdlENOE!ptXjb1
zRc|26tDk(ClCaI1`K1-;^x^<u;z*1`fm+;1DG60fq^QFcJGO13Pk7>m+I{)5DC-V>
z0Rgo>wYpK9#^k(M-i2yLJT@c}GwqDL&L=@3A=n!|>DK|OoR4uLk<j+UMniMv+&RRN
zb9QzwQHq6NVvEfb*Ougr4AmK)qi4^aE#EDOMd0Az0PhBQj6oww1Rgxm*4Bm`jogiu
zl@+=Gls(ad&VjTH)Y2-7uCgv)x6N;G^7MQW+!V9f?ZVYy#3*<@FY)U|zc9<Ks*+zH
zi^g=xUHvrrkDCm&{@khbTe7{^P|W)HcH7|cxeROnGIENeYMfJUFRRlt{zmyx==;<6
z%Q&7nqwuM7-=lY%1kSxQ^0xck(ylWoZ$4ny{PPT>6n(m!mTAI^-$uhiL7efu@e5xk
z*j_qKjMsKnjjYcA#0fnRGL;r#>${a!oVM9$=asTQ?N2SwPv_M(40q@xYdU<0sAORC
zJ~-z*Z*5^vrMWVr&(&nzeAX|N>dpc8y(}#bH_Djj%YPnDziQB9=7rUgJjkGDY}cvV
ze(#sG<xKy~(@Uoxr*L%-tv&gD?U!h!Ek|O`6@&hP9*bU+GEoo9Ggh2#TpuN@KdSmD
zd%k}pzr&{R($KM8Rew+fs{3fQaTffEiZD-JTH0~#*o!0*Z320zMnQ4$*NVFOdI6DC
zuX66)l$Aq!qiVJTVOv)E3M8k=8jHxuCHRssY*4>VOYGo5kY>WaB#M+qKxuRH8%+*Y
zS<dCwCR^}98!=wun+Y%Nm?Vlsj6hf$l8=U+vvAQAROMedC~C#g3DH}$g;%pc7P1tG
z5R%A43X>|p22QW5%A}4Vcj@q%0qYV&SE~yB_*u^Tx5myrRYAlxEwG|=+op}hwJx^`
z`UT_z->e=*>j15gPrjycYe$E2Wo(XVn1A-$x5p_9)!Jy5cmk_WxYkssNBw=mR!FLC
zUZ0@7{w5+alH<UcfbJ$A`zZY^sfSWO^t<$YuB*PIzV*iD&7r~wGx2X)4hz@q1CMdD
zr5ye3N8jGC?tOaJ)j#*M<>NbHht?&8y-l~=IpDIlNA!8-q}S=$Lq|muPoD~HBeA=)
zcf7v&O-euSa}nc5vFjAZ2buf*4?U}!zeF7GR1a)lS)ZL=95gE!l*I4_72`|J%PWz?
zU;=qBn8$sdoNV%GYi<3EL6i6H?d$BI+Cc4F!`rvBw3182E0@{J18(RCF=n73XK$4+
z_KTtYIj^Zj@~PJ5pxY6-xjzBfDR0p-Fz_w2gMoNcrAu~gVL{2HH24SY(#lHAsd}}J
z2OMAg<<F0fj-mugpK%5SIscNz$o1#1+S}R`WqxZGWTYRz?COfpZ)R<6jcgmWXxQJV
z<;=o*2%usn4zM2W8zHNDgn-Y|pVyc+ievJfPVs%q{;li%_AV~ZxA}bbkK4BGLadv?
z;lCSgSO0;jdXB$foM^p44v-p)5pn7-vfLuK>#Msh4E4<LO$$wBWoODPdhBrCdU(v|
zfP+eJjzr1N&4Z$=<x#6r4qLR2i-+Vr^Zvc|eMj!+h!1bJ2ie)~qTqV^QLTqRdX@8j
z@oM8$@s8JvIRkN|k%ZF3(u8XX3PT_Q7KMJcccI6mSmD|c4L?`;%-NCGpOnW8j*_)0
z^WXg(+1l}Hl^kt*a&5bpX79D;l66Wf|Nad&<11Jx3Psr~tDZ=8!QsGI74us_fuX1!
zJzi<9=d^@=dpEdkqVR_)l=*+5jhac2d|Qdsj9C$*sKt~8&e++Khm9LF{a&R^c-eMj
z9<*w|sbK|}fIzqUy#luxr<mwjE8*S~O-R+~7H&VPVA>}n^kZy{rj6^3xMQC*>2c-J
zCgXh;4espY$*F(wkosNEdx1B^U!OmE+Qh`H(zge*RTqb`H84s}R!%O{YZTwq85x+G
zn&L7V*Y9Jl$F7z2V}j&DyJy1xT>s5*Kth7o^1Wo%ZGWMe26urizR?Pwxyx)n$B2*b
z+TyF;)z${c$K2}{hGC_3b0JoL@vm{x^<Sw8GWI`@#JP2C>awz<)?LzlDsS1gZP4+2
zyR2%vVfylLOZBzByeGeyg|bAfJo1Y=*1xqR<fi2Y-U=M!@>vB)&Ipk_E*X-<sQzMe
z5a(}}g}8FDa=!ydci1SiT)lYB{DPXjH|LtBf|zZ9RbpA0x4L>n*t@}+dq2j<Veh&N
zL)dUnxVgE(GJ->#^{=2)T$3u#&%cC1!Hyj}h@Pp-muHY8lI-kMo)#<mop07j%zo?d
z(e^5RwyN&FH2KOMhbK2iGMt_T8=p9YEC#958tslSr39|Mnr?0qMY^>PS@-WV73DvF
z9wW~mcL2{Ng(0Fs%0RW!x3}F;_$)^+WENXpd-6@50uAfi&E&PlKX~A?zBXt%1E37n
zS&*`5PX4s1X=ca>^DU=?Ymp`Pvoy;m4qww!%caH5p4c83IhGix(@exC0Rua`A0Rn1
z5}bt@gpM461k89YHzP!jOI4sVCZ0c8C|HMTaHv8l^r&zUXNL)O<)WxvvX@!yclxS3
z9IraA>9)5_yIz*LxYGaR^yV&Q;|)7kHND;5q=#1TnR^#Fe$6ko+b>QVJYP#)ovOBW
z;z8@sYU2T@{xM5^0K+<)n{P>~MOYCfb0igTA&K#gX>QiuRdx4UmfEwC+`<*ZEu>~f
zmHsVFiE<A#H`UHUWY}LRB26+Mz3?qDoXzYJO|sL~-N#HMS$hvV7ls%a8I7!O4b-vL
z*QdRHtD4?85i`G|cGuDWGAvKsL%EMf>k8w$s>HNy1`<8?3qavsd8dRKCdVHNh5?iD
zKSusurqRJYuu#qasIu}3+D)*4$wz@ZP&GJRb)~}4=SP8}dS-mDvY#M>^nR12%8c6h
zZFUan-(x1n=pVjP@9fkLuJxVjZ2w@r6yBN&aYKd$SZe<T2EnyuZfs{Uj*Ju~Ur*i8
z@2JTo9jz!TlE4iDr;ELZbbQ1$3tI?RO;mG0$ElMPe0(t4kz;vJ*Q-}&A(}#+fC)zl
zdS!3j+Of>KgrkdA`$f#RtlKed09|zgvgRAyQuC42<5nJybN$>?U>)}IrOYyS-1|S6
zu)@<5wNOwM&Ock&^68Xt?9BN|0h`rb8!ZEZN0XDAM30-+vAj)=YPrhO@cyD$U2CrG
zVyv@^3%4amb$~myquLRi;h@+q2Sip>$gNKpUlb4}aZVw9-PN$wSL}owd7#c=bq$S*
ziVBRX;nzim<nyJL3X_a3dURV~y`x{`5z$b(T;Hve5mNf5bh8t?@IKKDryTy3*L%~C
z|30!Ln0hUuy8GImMX|3r8bS&;Q;*Bg=N4#1NIW79hI0f8#Tv%+4NYEMlh&<3C;0U>
z!N#PZ;Ide{<-4aRYHZcgMD}4~-7S3JUoS7hnH@zsSJy*M0c3IS%|U2GiRS7`!Ps~*
zys+@lWvhPVO^Agda!Yr8Pu`I=ZfY2p?_p3^S4RPkdp({i;#kvQ-@pJ$j{EZsAKF7;
zhQh+q2@*t2YL{!2WbUS6QWThug)0H>WI>?bXA&3Mi!Z?;81c{W!>)7Iz_ktzBKZpJ
zTsD8SdXLf3(e26>5@dDQlnU=Ju`V(rFBF>b+&qQMyE;2j@QARmHBc}!V_o*ft8`GC
zbN#)$YM#313@kp!9R3?7r>2IKFwJV5`ZYctQd79BiA_8Qf;3{X5etvEn^5e$n))tI
zpV8<*XUv~dcb*;e+Pi~cM`5<mHZdliDA`+_;lLm_bM)ulwdIyOPri{@r(4sPp!F6*
zn?oluB3g4Ue*Q3D<es0I4~-mPC^#I6$-GaT67)0hJUu=C8CjiVjA<%d3w}{drL~xn
zhMen8cwTyRl?sj_<ifKSm^p>May3uoO@zp><JdSkN$gXl-^=Qp@Uc=o*{-fG2#xj^
z$#QW87O#O2Q5Si5;B1hx)}3pR@cA>iB*KjU!w3C>PwuX+tWX~VF&k^QQ?%tWd0=QU
z)z-}POv8>A3sguZCn$_(yoE4V<8~f`G%v@9k)a`4)&vIZke@*Ts{6w=Bm?LS2z<Nx
zcHd_z#6oQYZ2(A=nc!SpTnu}ZXKrtI$;qf@1{WRG67{mU$X>Mb{foHiu}X7~PHjFN
zo42h@bE8TZ8n!l2zX+yJISKH6okVGT&iUuJzH|?cf=^k`I~Xggrs!<ror79afQAi4
zxwsT-D;Ti%Al&lIo1dT0)VOea<FF9OEFqzg$*`Is9{%<`PpvOHgYO$zlOVc_%3G$3
z5@I;~_z{?c<aRyG3^9A{*Wl6#=Af>i2VBD4sqk%{kPc+bzRnPrvH3bS_QIx~mPzuV
z<jwOri%5+1<i%06DJ<a?YgD@302D>3-fdruIkDj*(#Uf_&Y*^EQsM2UqC)AAu(04x
zI$nIXCbYh<IUs<HG=y5H@PW|q_oK&ERP^iwR~+aS_!PPQ-0Sx4AtT#GNB9Cj$#<hE
z)+ndCTH)Dy0Vt2a#Lzk}L5UnA@D>AL=V1u?uY3OPYIXW$+;aqn<n&EPLr%yAablbb
z0T`6dow%{*c0z(k%d^jNhgCBvCTUY^M8dWTJ3e}xR?zR`BifYvGJjJH{}-!aF~9K7
z{+)_^S5HdkJUZEY<-W+|n&@tG+I?+?A8*`e;NC0r@K$JZ-&q%fRfE^*!4JN-HKOuK
z>3>jh9(9cAJpVr$-A*WED<Ad8?4p$OyS}t^`XMx(T_LF6?y%U^_n)q_xjv#<71(%3
z$|uC^2WA&XyB`!z#JsP1j^6h!o%Rk6etzpQRD0n@iNt}H{mzDtci&evNDZPGke;P!
zB8xK9BS#|xgW)e<Xc-xeEi9UcE`c;&gJ%E+fW{p8eFolLBLVchr>d^Vxv3R3Hfn-*
zKnefrw{KDs5^k=phDJsN0ggE}I3as`y`grep7?pE7Vnoyw+G7xW?;mq9>6{<T2SKP
z!ZhM2s?J=rvzr|>(Y|*eP-ug^ygW%98ufbWz2~hgs;-znBHYbBBwQ*vcjgQY6O-E7
z60E6sczLh&li)Km!f{YZNvT)cDA`VVJH=6UippGt%s`#uzbHY~*S9<`Pt<LQawlw4
z8#TwxJTvD;I8$kN=Ux1LiA{vfLu>Gy)7e3d=W&(Am0=0*uD#)xos_-rlsK~I=<R(;
z8QVD=_nbMTprlo!`O!XbILLZy<aZMF@uY?3g<%g~KUs{r|3TxXxAogvGoyUj3(}mR
z&sHBf5jnTmvTM2ZKIeZmu}Ml@CfPLeAC@WiD7WSFF_tdZ0+Vfi48wa!B*7wCad8(g
z)#V?Nl|urN(sf_jIXQ8_pItFHB&3b+xikj8@(8_{jVy>;e?NvZ07}=st}ec(<PVw#
zN?rJux!ov&lcS#N0Z!G_)Li_5_}=Eu8qWEbnBsQXgvKmENy(G7fiKK{Lb6w0QIWYb
z1}k3Ur~>3Ok*Os492DoN?+~Nr*ZQLo&oA(|UhAsz_wO~rz}*JaE8zJEN)RguP6o$b
zcK1MJcu3+RyNaMLBmlrfnj)eC0#sB~OouN&2@DJbn-1DCC1n^5;J8}N^X|4L_V&Wd
zj8|RpzIZccOjjBk!-%>+D7bcPhldQmeaOjk8ChB0Q=bDB9+#9zS#Q@Zg?3ZKBtKv=
zzGkQ~N!}kIYjSe#8cyYG_)X8$?39n<C5DN{zRkE!JbDDh7bSdC^$bMfMA-xB`D9WP
zG>S~j%pdk|)uxpjVpNRq+L_JXRn$82@}Z>X>6vFEK0khrZXV^BlUu6fZQ_~RA2LMO
zrrT~J?%tFdpb;r4+wkX*)N7&FhH-rS{RKhRMoq^UZ|+L_;-)gOJzoB6?C)617yh@)
zg-g#kyc)h4>2$?=c6DYJk9z6I67z-jr+YTZF@TDkM@rZOVt;se7TyYQR+5yEfD!2t
z#2zeRq)8(mpAtqQFoEl6LY#$6rP;?^PwyZYYgF2i@jFXD1S~>n995;3B>AMK^P(2z
zCaxNotfZM6z_4qVGhiq<yZhgMq4)e9_TTH*R)Aape+?}yd6}6PQ4K=~z`=}hQ7A#e
zHN80-w$e7EscG?YK9Yww4Z)NFgf!@6lA0P-SPhElM?8M^p`H#v#KOYD>wi!9o6+-<
z5`LFl@Qj2;7dm!usgNdEU=RBE_?Vj$L^?Em?5wP^Fa(FIa%x(dSUzo!{Mz)+z~DdN
zi#wbi&JZ-HoL9t2Pfuw#(v9G-2QnVH+gQ|ppa+4^N2g58)btaW$cy|Pm`8X8Z7=F1
z+8m5H3*6}oLjBzhrWzX-LyQK$H6if%k~9|=7ufXc;o&c!Tre`){h(>Th)4xSP+q!(
z(WqCU8EVMRM8rWFjp_!~CkNwj<}JPy<d%_C|D2_Hc1lrJ7W^gRM@227<BOa~d9ubM
z!!~cmgZTA7=sx3i3A6w7vwpy3Vd&JG;?3SWp5{?!En@j0(Op8R2U?)PDt~#01dQH*
z8K+7&`|Fw-1ATo&Zbcg9ix&kPA~ITATQRSp_52<#RHrvnP~c%*<qXFY#otgFN@FH%
z;eB1yCvmar1N%NaD=Hebq|qG`77@XouK(7ENwKhIA}{I%TUq&M!IVn;d!2tLbh3>@
zm+EYEAcS6nu$T96q5dfo6HhO%>DYB{-0SUZM-27C4jeq#Q||RV+#?hd`FeqF&}2l%
zA=)Oj2dPnKC-i!z7kBcAGhBCDnmUWT3R*g~D;rUBD23>(wN*E@SStYMZ&}&zXXoD6
z=4#c8qeBNWK$RXP5egnYKI%J5%gghH<+30};n{@3RucfCCoet>H6HYIU|W0p^~gxZ
zHl}weGXOH6kmv4I^P6^X(=3%qQV86Y_MH8^6gYfhk*=s#aclZHw-4hgHI(c3JOpke
z)HP?zNWE~>HKr@-xNPLo8DStaaOuMFUiBc2OHzr0CT(WZZ5fd!6dJ3WC~&)dY*+F;
zuKfAv_V_T(qp)fYEnPV!r5cpwYdSmcE|ef0!g7IHV^@K6{Nu-GQOC*mEkWqLefu`;
zEeE2492OSFBeDA&P&8^re^$2cU&%2+VF$mk_Xs_LAu=28sbDjT<veCZ8jSG-8IdG*
zE@I3YHw=PFc=G5qEA!JP4|JguPQM@wV=aq|(@vSH7lLDg8QB=Bamd0MO?zr;hJoS1
zQLSdi-$6ANdF&C)v5YpF1a>4WiO}E(3Cg_(4>YdOw{Zwk5n^h2dCLjgmbY*DBqbMM
zy6~-~;7tHp_Yf6k<1#WcbIeMk>m^-x;e(<+30#P6l57Wha`UE5P?{Oq+p@M%67_xZ
zq%rG?bN@OhcTqfkl#y<-ML+BH&oi4J?vP#_dzeox`onVnZ-=%?<C6RKcem}4xoxnh
zEh1mKda7Ui>~6={zK2N%s`86<6H_i7ksk677L}_!RAS=W8&(!ursNeSJ1kq(v{bxs
z^YbyG*7h^LtgckCv9Snc6`s>pW@cuX(}wCuXy|clX+rjC^@bYd*8(4+C-};hwcFe$
z2q`1N%lJ5H<HoCCvzL~ZC^l_^Nt4d&*g2Rz0}ToYe8=rvUS0-kj{5hrCIr84%}MJ(
z(1y?pA&#`RK1xZkwzmg`s*BnV<yMA$mp<RU00L@mejdV{Cio2t3)?$7N{WmBz@Q?y
z&0@goxeKAI#y2^`vAxsN(g1Lf<$~b7zj(Z;xERNmiK!_F!M@H;H}pC-_C-xWN!X5d
zgZ!yq)xNl9l#;|)Zm(kh5YG9y>W2JDXtYrHcL9-wZa4%ZOv@_=3@R<{sjjZ>?nd4E
z6$*fh)6;rq>{3UEh%aZ{o2`udZES6cardD9aW=%q$0Ku+l$3<#m*}sP^%7fueXB6T
zu4EU*G<?Pdo&XdtU@@Yl;2YYlAZ<WW=-EDZ<K|72jbIM+ZD<HSSljA<b?zn>BT@7e
zy)Ws&JgNU!F8KO&4Kx4%#(@BSfPOPPsZ*zxf5XiRh4ii0A>Z=^>RVa@Fh4QlmnDl7
z?;u&A&as9azB9^l$^4W@Q;*A09*fd{{Bh!=V!fjL;F?$BuxxszHbaGo2}Sm^(r5W;
zQ;VCcEPXHE?tgXCjfMAh_cfcU#)VspiISeZJTKL1gv0KCkI&4`G?p;#7kC-7df@Qz
z#Z7U#nx_LE#vMK^>M|hV^U?3o%4R=#&v$cd+-%$%9X4w7r{|{rG#bv9xBTOkGV&ID
zvNRiQQ%gW%z#E0r)KSVxN>{Om^9NCjGp;2k_3PQ#j6kL6nFp3k`Js#}o7}f7!f}^;
zIsG56FGr)I^mmv(X{)Nb#lIfwqHe^%>7a*)N1*RG3{AjnZj-VG(FOT=ax8>$263Gm
z3%r|?T1(VgN=aLM%@sYad7Ja7LTDE0PIa-JDJ|6T*%M(3&y_=F7c<lt%w+ul_~0aW
zQ!svG&;iTcWD2?p3IWy1Sy<?x<%zck8!^02FvmE^vPj#kXJ%(9f4U(3xL?(>A%LAS
zFi&WVL2subHJVF+D$&!Tp+TIYUqw-`_zl~!oUDa8O%#mqk%mQkkiTLHWcAkEL)9@b
zIzRuw-9id7?;R9J@BY~T-@J^%Fg!GA>)6?tn0h{ZNXBa@Ae5{+Ey0_G@~eKuUCd6$
zoZ=-7G^!jzKd({I=D)AfIsUJU2_!B?K?(XLPABbCr|`xrhMkG*;qunn2mq*h-ZK-^
zF?#*_b?>K7^&D<)ZtK6+*K|TAHv}YZ4AdzqQ|VBOA}7=3#8FG8(9++}CNz*ivz}0o
zItJR_%y&zM9wc<A@GE;Vev(XKXGnGWvHrc|T&f$$_R;)Cllvo~GUwsYtOt<ySJNoV
zkp(1f`lm&l5d4$fBRH6wNEIKhSVB#e<+wwY!gxEG!uch-UbdkuxPk;kBLn{XyZBGQ
zVN@gj{ll3Z#Owd-i+G`be^2)H|NqIq@5TSFJI0*li2h9piB1q*ojG_?>`#)DgNn-^
zJtE^m*OZ}#NmN#q+2FlTb(gHk%F>!+wT+}UC@cD#EusWA_xaUEqY`rb)rE@}6;;W}
z0&+7`GE-7Am57(3fjbfw!araCE6=&+%jME6$BmE@qDUScP5bw){=VFRHc})AQL-E<
zX(;~N)fD_sW6IxOYE<<9+r{+v2Nz;L{p+imbq;N0|N1(x`Oja+gTFuiul&if|N3>y
W(9&Er*f)@osi|ly=N>x~^j`oicdRu4

diff --git a/docs/dev/articles/examples/vision/autoencoder.html b/docs/dev/articles/examples/vision/autoencoder.html
index b23592401..cc0a7b4d1 100644
--- a/docs/dev/articles/examples/vision/autoencoder.html
+++ b/docs/dev/articles/examples/vision/autoencoder.html
@@ -31,7 +31,7 @@
 
     <a class="navbar-brand me-2" href="../../../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -225,108 +225,108 @@ <h2 id="build-the-autoencoder">Build the autoencoder<a class="anchor" aria-label
 <span><span class="co">## Epoch 2/50</span></span>
 <span><span class="co">## 469/469 - 2s - 3ms/step - loss: 0.0712 - val_loss: 0.0692</span></span>
 <span><span class="co">## Epoch 3/50</span></span>
-<span><span class="co">## 469/469 - 2s - 3ms/step - loss: 0.0691 - val_loss: 0.0679</span></span>
+<span><span class="co">## 469/469 - 2s - 3ms/step - loss: 0.0690 - val_loss: 0.0678</span></span>
 <span><span class="co">## Epoch 4/50</span></span>
-<span><span class="co">## 469/469 - 2s - 4ms/step - loss: 0.0680 - val_loss: 0.0671</span></span>
+<span><span class="co">## 469/469 - 2s - 3ms/step - loss: 0.0679 - val_loss: 0.0671</span></span>
 <span><span class="co">## Epoch 5/50</span></span>
-<span><span class="co">## 469/469 - 2s - 4ms/step - loss: 0.0672 - val_loss: 0.0665</span></span>
+<span><span class="co">## 469/469 - 2s - 3ms/step - loss: 0.0672 - val_loss: 0.0665</span></span>
 <span><span class="co">## Epoch 6/50</span></span>
-<span><span class="co">## 469/469 - 2s - 4ms/step - loss: 0.0667 - val_loss: 0.0660</span></span>
+<span><span class="co">## 469/469 - 2s - 3ms/step - loss: 0.0666 - val_loss: 0.0660</span></span>
 <span><span class="co">## Epoch 7/50</span></span>
 <span><span class="co">## 469/469 - 2s - 3ms/step - loss: 0.0662 - val_loss: 0.0656</span></span>
 <span><span class="co">## Epoch 8/50</span></span>
-<span><span class="co">## 469/469 - 2s - 3ms/step - loss: 0.0659 - val_loss: 0.0653</span></span>
+<span><span class="co">## 469/469 - 2s - 3ms/step - loss: 0.0659 - val_loss: 0.0652</span></span>
 <span><span class="co">## Epoch 9/50</span></span>
-<span><span class="co">## 469/469 - 2s - 4ms/step - loss: 0.0656 - val_loss: 0.0650</span></span>
+<span><span class="co">## 469/469 - 2s - 3ms/step - loss: 0.0656 - val_loss: 0.0650</span></span>
 <span><span class="co">## Epoch 10/50</span></span>
-<span><span class="co">## 469/469 - 2s - 4ms/step - loss: 0.0654 - val_loss: 0.0648</span></span>
+<span><span class="co">## 469/469 - 1s - 3ms/step - loss: 0.0653 - val_loss: 0.0648</span></span>
 <span><span class="co">## Epoch 11/50</span></span>
-<span><span class="co">## 469/469 - 2s - 4ms/step - loss: 0.0652 - val_loss: 0.0646</span></span>
+<span><span class="co">## 469/469 - 2s - 3ms/step - loss: 0.0651 - val_loss: 0.0646</span></span>
 <span><span class="co">## Epoch 12/50</span></span>
-<span><span class="co">## 469/469 - 2s - 4ms/step - loss: 0.0650 - val_loss: 0.0644</span></span>
+<span><span class="co">## 469/469 - 1s - 3ms/step - loss: 0.0649 - val_loss: 0.0644</span></span>
 <span><span class="co">## Epoch 13/50</span></span>
-<span><span class="co">## 469/469 - 2s - 4ms/step - loss: 0.0648 - val_loss: 0.0643</span></span>
+<span><span class="co">## 469/469 - 2s - 3ms/step - loss: 0.0648 - val_loss: 0.0643</span></span>
 <span><span class="co">## Epoch 14/50</span></span>
-<span><span class="co">## 469/469 - 2s - 4ms/step - loss: 0.0646 - val_loss: 0.0641</span></span>
+<span><span class="co">## 469/469 - 1s - 3ms/step - loss: 0.0646 - val_loss: 0.0641</span></span>
 <span><span class="co">## Epoch 15/50</span></span>
-<span><span class="co">## 469/469 - 2s - 4ms/step - loss: 0.0645 - val_loss: 0.0640</span></span>
+<span><span class="co">## 469/469 - 1s - 3ms/step - loss: 0.0645 - val_loss: 0.0640</span></span>
 <span><span class="co">## Epoch 16/50</span></span>
-<span><span class="co">## 469/469 - 2s - 4ms/step - loss: 0.0644 - val_loss: 0.0639</span></span>
+<span><span class="co">## 469/469 - 1s - 3ms/step - loss: 0.0644 - val_loss: 0.0639</span></span>
 <span><span class="co">## Epoch 17/50</span></span>
-<span><span class="co">## 469/469 - 2s - 4ms/step - loss: 0.0643 - val_loss: 0.0638</span></span>
+<span><span class="co">## 469/469 - 2s - 3ms/step - loss: 0.0642 - val_loss: 0.0638</span></span>
 <span><span class="co">## Epoch 18/50</span></span>
-<span><span class="co">## 469/469 - 2s - 4ms/step - loss: 0.0642 - val_loss: 0.0637</span></span>
+<span><span class="co">## 469/469 - 2s - 3ms/step - loss: 0.0641 - val_loss: 0.0637</span></span>
 <span><span class="co">## Epoch 19/50</span></span>
-<span><span class="co">## 469/469 - 2s - 4ms/step - loss: 0.0641 - val_loss: 0.0636</span></span>
+<span><span class="co">## 469/469 - 1s - 3ms/step - loss: 0.0640 - val_loss: 0.0636</span></span>
 <span><span class="co">## Epoch 20/50</span></span>
-<span><span class="co">## 469/469 - 2s - 4ms/step - loss: 0.0640 - val_loss: 0.0635</span></span>
+<span><span class="co">## 469/469 - 2s - 3ms/step - loss: 0.0639 - val_loss: 0.0635</span></span>
 <span><span class="co">## Epoch 21/50</span></span>
-<span><span class="co">## 469/469 - 2s - 4ms/step - loss: 0.0639 - val_loss: 0.0635</span></span>
+<span><span class="co">## 469/469 - 2s - 3ms/step - loss: 0.0639 - val_loss: 0.0635</span></span>
 <span><span class="co">## Epoch 22/50</span></span>
-<span><span class="co">## 469/469 - 2s - 4ms/step - loss: 0.0638 - val_loss: 0.0634</span></span>
+<span><span class="co">## 469/469 - 1s - 3ms/step - loss: 0.0638 - val_loss: 0.0634</span></span>
 <span><span class="co">## Epoch 23/50</span></span>
-<span><span class="co">## 469/469 - 2s - 4ms/step - loss: 0.0638 - val_loss: 0.0634</span></span>
+<span><span class="co">## 469/469 - 2s - 3ms/step - loss: 0.0637 - val_loss: 0.0634</span></span>
 <span><span class="co">## Epoch 24/50</span></span>
-<span><span class="co">## 469/469 - 2s - 4ms/step - loss: 0.0637 - val_loss: 0.0633</span></span>
+<span><span class="co">## 469/469 - 2s - 3ms/step - loss: 0.0637 - val_loss: 0.0633</span></span>
 <span><span class="co">## Epoch 25/50</span></span>
-<span><span class="co">## 469/469 - 2s - 4ms/step - loss: 0.0637 - val_loss: 0.0633</span></span>
+<span><span class="co">## 469/469 - 2s - 3ms/step - loss: 0.0636 - val_loss: 0.0632</span></span>
 <span><span class="co">## Epoch 26/50</span></span>
-<span><span class="co">## 469/469 - 2s - 4ms/step - loss: 0.0636 - val_loss: 0.0632</span></span>
+<span><span class="co">## 469/469 - 2s - 3ms/step - loss: 0.0636 - val_loss: 0.0632</span></span>
 <span><span class="co">## Epoch 27/50</span></span>
-<span><span class="co">## 469/469 - 2s - 4ms/step - loss: 0.0636 - val_loss: 0.0632</span></span>
+<span><span class="co">## 469/469 - 2s - 3ms/step - loss: 0.0635 - val_loss: 0.0631</span></span>
 <span><span class="co">## Epoch 28/50</span></span>
-<span><span class="co">## 469/469 - 2s - 4ms/step - loss: 0.0635 - val_loss: 0.0631</span></span>
+<span><span class="co">## 469/469 - 2s - 3ms/step - loss: 0.0635 - val_loss: 0.0631</span></span>
 <span><span class="co">## Epoch 29/50</span></span>
-<span><span class="co">## 469/469 - 2s - 4ms/step - loss: 0.0635 - val_loss: 0.0631</span></span>
+<span><span class="co">## 469/469 - 2s - 3ms/step - loss: 0.0634 - val_loss: 0.0631</span></span>
 <span><span class="co">## Epoch 30/50</span></span>
-<span><span class="co">## 469/469 - 2s - 4ms/step - loss: 0.0634 - val_loss: 0.0630</span></span>
+<span><span class="co">## 469/469 - 1s - 3ms/step - loss: 0.0634 - val_loss: 0.0630</span></span>
 <span><span class="co">## Epoch 31/50</span></span>
-<span><span class="co">## 469/469 - 2s - 4ms/step - loss: 0.0634 - val_loss: 0.0630</span></span>
+<span><span class="co">## 469/469 - 2s - 3ms/step - loss: 0.0633 - val_loss: 0.0630</span></span>
 <span><span class="co">## Epoch 32/50</span></span>
-<span><span class="co">## 469/469 - 2s - 4ms/step - loss: 0.0633 - val_loss: 0.0629</span></span>
+<span><span class="co">## 469/469 - 2s - 3ms/step - loss: 0.0633 - val_loss: 0.0630</span></span>
 <span><span class="co">## Epoch 33/50</span></span>
-<span><span class="co">## 469/469 - 2s - 4ms/step - loss: 0.0633 - val_loss: 0.0629</span></span>
+<span><span class="co">## 469/469 - 2s - 3ms/step - loss: 0.0633 - val_loss: 0.0629</span></span>
 <span><span class="co">## Epoch 34/50</span></span>
-<span><span class="co">## 469/469 - 2s - 4ms/step - loss: 0.0633 - val_loss: 0.0629</span></span>
+<span><span class="co">## 469/469 - 2s - 3ms/step - loss: 0.0632 - val_loss: 0.0629</span></span>
 <span><span class="co">## Epoch 35/50</span></span>
-<span><span class="co">## 469/469 - 2s - 4ms/step - loss: 0.0632 - val_loss: 0.0628</span></span>
+<span><span class="co">## 469/469 - 1s - 3ms/step - loss: 0.0632 - val_loss: 0.0628</span></span>
 <span><span class="co">## Epoch 36/50</span></span>
-<span><span class="co">## 469/469 - 2s - 4ms/step - loss: 0.0632 - val_loss: 0.0628</span></span>
+<span><span class="co">## 469/469 - 2s - 3ms/step - loss: 0.0632 - val_loss: 0.0628</span></span>
 <span><span class="co">## Epoch 37/50</span></span>
-<span><span class="co">## 469/469 - 2s - 4ms/step - loss: 0.0632 - val_loss: 0.0628</span></span>
+<span><span class="co">## 469/469 - 1s - 3ms/step - loss: 0.0631 - val_loss: 0.0628</span></span>
 <span><span class="co">## Epoch 38/50</span></span>
-<span><span class="co">## 469/469 - 2s - 4ms/step - loss: 0.0631 - val_loss: 0.0628</span></span>
+<span><span class="co">## 469/469 - 2s - 3ms/step - loss: 0.0631 - val_loss: 0.0628</span></span>
 <span><span class="co">## Epoch 39/50</span></span>
-<span><span class="co">## 469/469 - 2s - 4ms/step - loss: 0.0631 - val_loss: 0.0627</span></span>
+<span><span class="co">## 469/469 - 1s - 3ms/step - loss: 0.0631 - val_loss: 0.0627</span></span>
 <span><span class="co">## Epoch 40/50</span></span>
-<span><span class="co">## 469/469 - 2s - 4ms/step - loss: 0.0631 - val_loss: 0.0627</span></span>
+<span><span class="co">## 469/469 - 2s - 3ms/step - loss: 0.0630 - val_loss: 0.0627</span></span>
 <span><span class="co">## Epoch 41/50</span></span>
-<span><span class="co">## 469/469 - 2s - 4ms/step - loss: 0.0630 - val_loss: 0.0627</span></span>
+<span><span class="co">## 469/469 - 2s - 3ms/step - loss: 0.0630 - val_loss: 0.0627</span></span>
 <span><span class="co">## Epoch 42/50</span></span>
-<span><span class="co">## 469/469 - 2s - 4ms/step - loss: 0.0630 - val_loss: 0.0627</span></span>
+<span><span class="co">## 469/469 - 2s - 3ms/step - loss: 0.0630 - val_loss: 0.0626</span></span>
 <span><span class="co">## Epoch 43/50</span></span>
-<span><span class="co">## 469/469 - 2s - 4ms/step - loss: 0.0630 - val_loss: 0.0626</span></span>
+<span><span class="co">## 469/469 - 2s - 3ms/step - loss: 0.0630 - val_loss: 0.0626</span></span>
 <span><span class="co">## Epoch 44/50</span></span>
-<span><span class="co">## 469/469 - 2s - 4ms/step - loss: 0.0630 - val_loss: 0.0626</span></span>
+<span><span class="co">## 469/469 - 2s - 3ms/step - loss: 0.0629 - val_loss: 0.0626</span></span>
 <span><span class="co">## Epoch 45/50</span></span>
-<span><span class="co">## 469/469 - 2s - 4ms/step - loss: 0.0629 - val_loss: 0.0626</span></span>
+<span><span class="co">## 469/469 - 2s - 3ms/step - loss: 0.0629 - val_loss: 0.0626</span></span>
 <span><span class="co">## Epoch 46/50</span></span>
-<span><span class="co">## 469/469 - 2s - 4ms/step - loss: 0.0629 - val_loss: 0.0626</span></span>
+<span><span class="co">## 469/469 - 2s - 3ms/step - loss: 0.0629 - val_loss: 0.0625</span></span>
 <span><span class="co">## Epoch 47/50</span></span>
-<span><span class="co">## 469/469 - 2s - 4ms/step - loss: 0.0629 - val_loss: 0.0625</span></span>
+<span><span class="co">## 469/469 - 2s - 3ms/step - loss: 0.0629 - val_loss: 0.0625</span></span>
 <span><span class="co">## Epoch 48/50</span></span>
-<span><span class="co">## 469/469 - 2s - 4ms/step - loss: 0.0629 - val_loss: 0.0625</span></span>
+<span><span class="co">## 469/469 - 1s - 3ms/step - loss: 0.0628 - val_loss: 0.0625</span></span>
 <span><span class="co">## Epoch 49/50</span></span>
-<span><span class="co">## 469/469 - 2s - 4ms/step - loss: 0.0628 - val_loss: 0.0625</span></span>
+<span><span class="co">## 469/469 - 2s - 3ms/step - loss: 0.0628 - val_loss: 0.0625</span></span>
 <span><span class="co">## Epoch 50/50</span></span>
-<span><span class="co">## 469/469 - 2s - 4ms/step - loss: 0.0628 - val_loss: 0.0625</span></span></code></pre>
+<span><span class="co">## 469/469 - 2s - 3ms/step - loss: 0.0628 - val_loss: 0.0625</span></span></code></pre>
 <p>Let’s predict on our test dataset and display the original image
 together with the prediction from our autoencoder.</p>
 <p>Notice how the predictions are pretty close to the original images,
 although not quite the same.</p>
 <div class="sourceCode" id="cb7"><pre class="downlit sourceCode r">
 <code class="sourceCode R"><span><span class="va">predictions</span> <span class="op">&lt;-</span> <span class="va">autoencoder</span> <span class="op">|&gt;</span> <span class="fu"><a href="https://rdrr.io/r/stats/predict.html" class="external-link">predict</a></span><span class="op">(</span><span class="va">test_data</span><span class="op">)</span></span></code></pre></div>
-<pre><code><span><span class="co">## 313/313 - 1s - 3ms/step</span></span></code></pre>
+<pre><code><span><span class="co">## 313/313 - 1s - 2ms/step</span></span></code></pre>
 <div class="sourceCode" id="cb9"><pre class="downlit sourceCode r">
 <code class="sourceCode R"><span><span class="fu">display</span><span class="op">(</span><span class="va">test_data</span>, <span class="va">predictions</span><span class="op">)</span></span></code></pre></div>
 <div class="float">
@@ -345,212 +345,212 @@ <h2 id="build-the-autoencoder">Build the autoencoder<a class="anchor" aria-label
 <span>  validation_data <span class="op">=</span> <span class="fu"><a href="https://rdrr.io/r/base/list.html" class="external-link">list</a></span><span class="op">(</span><span class="va">noisy_test_data</span>, <span class="va">test_data</span><span class="op">)</span>,</span>
 <span><span class="op">)</span></span></code></pre></div>
 <pre><code><span><span class="co">## Epoch 1/100</span></span>
-<span><span class="co">## 469/469 - 2s - 5ms/step - loss: 0.1009 - val_loss: 0.0939</span></span>
+<span><span class="co">## 469/469 - 2s - 3ms/step - loss: 0.1006 - val_loss: 0.0939</span></span>
 <span><span class="co">## Epoch 2/100</span></span>
-<span><span class="co">## 469/469 - 2s - 4ms/step - loss: 0.0936 - val_loss: 0.0919</span></span>
+<span><span class="co">## 469/469 - 2s - 3ms/step - loss: 0.0936 - val_loss: 0.0919</span></span>
 <span><span class="co">## Epoch 3/100</span></span>
-<span><span class="co">## 469/469 - 2s - 4ms/step - loss: 0.0921 - val_loss: 0.0908</span></span>
+<span><span class="co">## 469/469 - 2s - 3ms/step - loss: 0.0921 - val_loss: 0.0908</span></span>
 <span><span class="co">## Epoch 4/100</span></span>
-<span><span class="co">## 469/469 - 2s - 4ms/step - loss: 0.0911 - val_loss: 0.0900</span></span>
+<span><span class="co">## 469/469 - 2s - 3ms/step - loss: 0.0912 - val_loss: 0.0902</span></span>
 <span><span class="co">## Epoch 5/100</span></span>
-<span><span class="co">## 469/469 - 2s - 4ms/step - loss: 0.0903 - val_loss: 0.0895</span></span>
+<span><span class="co">## 469/469 - 2s - 3ms/step - loss: 0.0905 - val_loss: 0.0896</span></span>
 <span><span class="co">## Epoch 6/100</span></span>
-<span><span class="co">## 469/469 - 2s - 4ms/step - loss: 0.0897 - val_loss: 0.0889</span></span>
+<span><span class="co">## 469/469 - 2s - 3ms/step - loss: 0.0900 - val_loss: 0.0891</span></span>
 <span><span class="co">## Epoch 7/100</span></span>
-<span><span class="co">## 469/469 - 2s - 4ms/step - loss: 0.0893 - val_loss: 0.0885</span></span>
+<span><span class="co">## 469/469 - 2s - 3ms/step - loss: 0.0895 - val_loss: 0.0887</span></span>
 <span><span class="co">## Epoch 8/100</span></span>
-<span><span class="co">## 469/469 - 2s - 4ms/step - loss: 0.0889 - val_loss: 0.0881</span></span>
+<span><span class="co">## 469/469 - 2s - 3ms/step - loss: 0.0891 - val_loss: 0.0883</span></span>
 <span><span class="co">## Epoch 9/100</span></span>
-<span><span class="co">## 469/469 - 2s - 4ms/step - loss: 0.0886 - val_loss: 0.0879</span></span>
+<span><span class="co">## 469/469 - 2s - 3ms/step - loss: 0.0887 - val_loss: 0.0880</span></span>
 <span><span class="co">## Epoch 10/100</span></span>
-<span><span class="co">## 469/469 - 2s - 4ms/step - loss: 0.0883 - val_loss: 0.0876</span></span>
+<span><span class="co">## 469/469 - 2s - 3ms/step - loss: 0.0884 - val_loss: 0.0878</span></span>
 <span><span class="co">## Epoch 11/100</span></span>
-<span><span class="co">## 469/469 - 2s - 4ms/step - loss: 0.0880 - val_loss: 0.0874</span></span>
+<span><span class="co">## 469/469 - 2s - 3ms/step - loss: 0.0882 - val_loss: 0.0875</span></span>
 <span><span class="co">## Epoch 12/100</span></span>
-<span><span class="co">## 469/469 - 2s - 4ms/step - loss: 0.0878 - val_loss: 0.0872</span></span>
+<span><span class="co">## 469/469 - 2s - 3ms/step - loss: 0.0879 - val_loss: 0.0873</span></span>
 <span><span class="co">## Epoch 13/100</span></span>
-<span><span class="co">## 469/469 - 2s - 4ms/step - loss: 0.0876 - val_loss: 0.0870</span></span>
+<span><span class="co">## 469/469 - 2s - 3ms/step - loss: 0.0877 - val_loss: 0.0871</span></span>
 <span><span class="co">## Epoch 14/100</span></span>
-<span><span class="co">## 469/469 - 2s - 4ms/step - loss: 0.0875 - val_loss: 0.0868</span></span>
+<span><span class="co">## 469/469 - 2s - 3ms/step - loss: 0.0875 - val_loss: 0.0869</span></span>
 <span><span class="co">## Epoch 15/100</span></span>
-<span><span class="co">## 469/469 - 2s - 4ms/step - loss: 0.0873 - val_loss: 0.0867</span></span>
+<span><span class="co">## 469/469 - 2s - 3ms/step - loss: 0.0873 - val_loss: 0.0867</span></span>
 <span><span class="co">## Epoch 16/100</span></span>
-<span><span class="co">## 469/469 - 2s - 4ms/step - loss: 0.0872 - val_loss: 0.0866</span></span>
+<span><span class="co">## 469/469 - 2s - 3ms/step - loss: 0.0872 - val_loss: 0.0866</span></span>
 <span><span class="co">## Epoch 17/100</span></span>
-<span><span class="co">## 469/469 - 2s - 4ms/step - loss: 0.0870 - val_loss: 0.0865</span></span>
+<span><span class="co">## 469/469 - 2s - 3ms/step - loss: 0.0870 - val_loss: 0.0864</span></span>
 <span><span class="co">## Epoch 18/100</span></span>
-<span><span class="co">## 469/469 - 2s - 4ms/step - loss: 0.0869 - val_loss: 0.0864</span></span>
+<span><span class="co">## 469/469 - 2s - 3ms/step - loss: 0.0869 - val_loss: 0.0863</span></span>
 <span><span class="co">## Epoch 19/100</span></span>
-<span><span class="co">## 469/469 - 2s - 4ms/step - loss: 0.0868 - val_loss: 0.0863</span></span>
+<span><span class="co">## 469/469 - 2s - 3ms/step - loss: 0.0868 - val_loss: 0.0862</span></span>
 <span><span class="co">## Epoch 20/100</span></span>
-<span><span class="co">## 469/469 - 2s - 4ms/step - loss: 0.0867 - val_loss: 0.0862</span></span>
+<span><span class="co">## 469/469 - 2s - 3ms/step - loss: 0.0866 - val_loss: 0.0861</span></span>
 <span><span class="co">## Epoch 21/100</span></span>
-<span><span class="co">## 469/469 - 2s - 4ms/step - loss: 0.0866 - val_loss: 0.0861</span></span>
+<span><span class="co">## 469/469 - 2s - 3ms/step - loss: 0.0865 - val_loss: 0.0860</span></span>
 <span><span class="co">## Epoch 22/100</span></span>
-<span><span class="co">## 469/469 - 2s - 4ms/step - loss: 0.0865 - val_loss: 0.0860</span></span>
+<span><span class="co">## 469/469 - 2s - 3ms/step - loss: 0.0864 - val_loss: 0.0859</span></span>
 <span><span class="co">## Epoch 23/100</span></span>
-<span><span class="co">## 469/469 - 2s - 4ms/step - loss: 0.0865 - val_loss: 0.0859</span></span>
+<span><span class="co">## 469/469 - 2s - 3ms/step - loss: 0.0863 - val_loss: 0.0858</span></span>
 <span><span class="co">## Epoch 24/100</span></span>
-<span><span class="co">## 469/469 - 2s - 4ms/step - loss: 0.0864 - val_loss: 0.0859</span></span>
+<span><span class="co">## 469/469 - 1s - 3ms/step - loss: 0.0862 - val_loss: 0.0857</span></span>
 <span><span class="co">## Epoch 25/100</span></span>
-<span><span class="co">## 469/469 - 2s - 4ms/step - loss: 0.0863 - val_loss: 0.0858</span></span>
+<span><span class="co">## 469/469 - 2s - 3ms/step - loss: 0.0862 - val_loss: 0.0856</span></span>
 <span><span class="co">## Epoch 26/100</span></span>
-<span><span class="co">## 469/469 - 2s - 4ms/step - loss: 0.0862 - val_loss: 0.0857</span></span>
+<span><span class="co">## 469/469 - 2s - 3ms/step - loss: 0.0861 - val_loss: 0.0856</span></span>
 <span><span class="co">## Epoch 27/100</span></span>
-<span><span class="co">## 469/469 - 2s - 4ms/step - loss: 0.0862 - val_loss: 0.0857</span></span>
+<span><span class="co">## 469/469 - 2s - 3ms/step - loss: 0.0860 - val_loss: 0.0855</span></span>
 <span><span class="co">## Epoch 28/100</span></span>
-<span><span class="co">## 469/469 - 2s - 4ms/step - loss: 0.0861 - val_loss: 0.0856</span></span>
+<span><span class="co">## 469/469 - 2s - 3ms/step - loss: 0.0859 - val_loss: 0.0855</span></span>
 <span><span class="co">## Epoch 29/100</span></span>
-<span><span class="co">## 469/469 - 2s - 4ms/step - loss: 0.0860 - val_loss: 0.0856</span></span>
+<span><span class="co">## 469/469 - 2s - 3ms/step - loss: 0.0859 - val_loss: 0.0854</span></span>
 <span><span class="co">## Epoch 30/100</span></span>
-<span><span class="co">## 469/469 - 2s - 4ms/step - loss: 0.0860 - val_loss: 0.0855</span></span>
+<span><span class="co">## 469/469 - 2s - 3ms/step - loss: 0.0858 - val_loss: 0.0853</span></span>
 <span><span class="co">## Epoch 31/100</span></span>
-<span><span class="co">## 469/469 - 2s - 4ms/step - loss: 0.0859 - val_loss: 0.0854</span></span>
+<span><span class="co">## 469/469 - 2s - 3ms/step - loss: 0.0858 - val_loss: 0.0853</span></span>
 <span><span class="co">## Epoch 32/100</span></span>
-<span><span class="co">## 469/469 - 2s - 4ms/step - loss: 0.0859 - val_loss: 0.0854</span></span>
+<span><span class="co">## 469/469 - 2s - 3ms/step - loss: 0.0857 - val_loss: 0.0853</span></span>
 <span><span class="co">## Epoch 33/100</span></span>
-<span><span class="co">## 469/469 - 2s - 4ms/step - loss: 0.0858 - val_loss: 0.0854</span></span>
+<span><span class="co">## 469/469 - 2s - 3ms/step - loss: 0.0857 - val_loss: 0.0852</span></span>
 <span><span class="co">## Epoch 34/100</span></span>
-<span><span class="co">## 469/469 - 2s - 4ms/step - loss: 0.0858 - val_loss: 0.0853</span></span>
+<span><span class="co">## 469/469 - 2s - 3ms/step - loss: 0.0856 - val_loss: 0.0852</span></span>
 <span><span class="co">## Epoch 35/100</span></span>
-<span><span class="co">## 469/469 - 2s - 4ms/step - loss: 0.0857 - val_loss: 0.0853</span></span>
+<span><span class="co">## 469/469 - 2s - 3ms/step - loss: 0.0856 - val_loss: 0.0851</span></span>
 <span><span class="co">## Epoch 36/100</span></span>
-<span><span class="co">## 469/469 - 2s - 4ms/step - loss: 0.0857 - val_loss: 0.0852</span></span>
+<span><span class="co">## 469/469 - 2s - 3ms/step - loss: 0.0855 - val_loss: 0.0851</span></span>
 <span><span class="co">## Epoch 37/100</span></span>
-<span><span class="co">## 469/469 - 2s - 4ms/step - loss: 0.0856 - val_loss: 0.0852</span></span>
+<span><span class="co">## 469/469 - 2s - 3ms/step - loss: 0.0855 - val_loss: 0.0851</span></span>
 <span><span class="co">## Epoch 38/100</span></span>
-<span><span class="co">## 469/469 - 2s - 4ms/step - loss: 0.0856 - val_loss: 0.0851</span></span>
+<span><span class="co">## 469/469 - 2s - 3ms/step - loss: 0.0855 - val_loss: 0.0850</span></span>
 <span><span class="co">## Epoch 39/100</span></span>
-<span><span class="co">## 469/469 - 2s - 4ms/step - loss: 0.0856 - val_loss: 0.0851</span></span>
+<span><span class="co">## 469/469 - 2s - 3ms/step - loss: 0.0854 - val_loss: 0.0850</span></span>
 <span><span class="co">## Epoch 40/100</span></span>
-<span><span class="co">## 469/469 - 2s - 4ms/step - loss: 0.0855 - val_loss: 0.0851</span></span>
+<span><span class="co">## 469/469 - 2s - 3ms/step - loss: 0.0854 - val_loss: 0.0850</span></span>
 <span><span class="co">## Epoch 41/100</span></span>
-<span><span class="co">## 469/469 - 2s - 4ms/step - loss: 0.0855 - val_loss: 0.0850</span></span>
+<span><span class="co">## 469/469 - 2s - 3ms/step - loss: 0.0854 - val_loss: 0.0850</span></span>
 <span><span class="co">## Epoch 42/100</span></span>
-<span><span class="co">## 469/469 - 2s - 4ms/step - loss: 0.0854 - val_loss: 0.0850</span></span>
+<span><span class="co">## 469/469 - 2s - 3ms/step - loss: 0.0853 - val_loss: 0.0849</span></span>
 <span><span class="co">## Epoch 43/100</span></span>
-<span><span class="co">## 469/469 - 2s - 4ms/step - loss: 0.0854 - val_loss: 0.0850</span></span>
+<span><span class="co">## 469/469 - 2s - 3ms/step - loss: 0.0853 - val_loss: 0.0849</span></span>
 <span><span class="co">## Epoch 44/100</span></span>
-<span><span class="co">## 469/469 - 2s - 4ms/step - loss: 0.0854 - val_loss: 0.0850</span></span>
+<span><span class="co">## 469/469 - 2s - 3ms/step - loss: 0.0853 - val_loss: 0.0849</span></span>
 <span><span class="co">## Epoch 45/100</span></span>
-<span><span class="co">## 469/469 - 2s - 4ms/step - loss: 0.0853 - val_loss: 0.0849</span></span>
+<span><span class="co">## 469/469 - 2s - 3ms/step - loss: 0.0853 - val_loss: 0.0849</span></span>
 <span><span class="co">## Epoch 46/100</span></span>
-<span><span class="co">## 469/469 - 2s - 4ms/step - loss: 0.0853 - val_loss: 0.0849</span></span>
+<span><span class="co">## 469/469 - 2s - 3ms/step - loss: 0.0852 - val_loss: 0.0848</span></span>
 <span><span class="co">## Epoch 47/100</span></span>
-<span><span class="co">## 469/469 - 2s - 4ms/step - loss: 0.0853 - val_loss: 0.0849</span></span>
+<span><span class="co">## 469/469 - 2s - 3ms/step - loss: 0.0852 - val_loss: 0.0848</span></span>
 <span><span class="co">## Epoch 48/100</span></span>
-<span><span class="co">## 469/469 - 2s - 4ms/step - loss: 0.0853 - val_loss: 0.0849</span></span>
+<span><span class="co">## 469/469 - 2s - 3ms/step - loss: 0.0852 - val_loss: 0.0848</span></span>
 <span><span class="co">## Epoch 49/100</span></span>
-<span><span class="co">## 469/469 - 2s - 4ms/step - loss: 0.0852 - val_loss: 0.0849</span></span>
+<span><span class="co">## 469/469 - 2s - 3ms/step - loss: 0.0852 - val_loss: 0.0848</span></span>
 <span><span class="co">## Epoch 50/100</span></span>
-<span><span class="co">## 469/469 - 2s - 4ms/step - loss: 0.0852 - val_loss: 0.0848</span></span>
+<span><span class="co">## 469/469 - 2s - 3ms/step - loss: 0.0851 - val_loss: 0.0848</span></span>
 <span><span class="co">## Epoch 51/100</span></span>
-<span><span class="co">## 469/469 - 2s - 4ms/step - loss: 0.0852 - val_loss: 0.0848</span></span>
+<span><span class="co">## 469/469 - 2s - 3ms/step - loss: 0.0851 - val_loss: 0.0847</span></span>
 <span><span class="co">## Epoch 52/100</span></span>
-<span><span class="co">## 469/469 - 2s - 4ms/step - loss: 0.0852 - val_loss: 0.0848</span></span>
+<span><span class="co">## 469/469 - 2s - 3ms/step - loss: 0.0851 - val_loss: 0.0847</span></span>
 <span><span class="co">## Epoch 53/100</span></span>
-<span><span class="co">## 469/469 - 2s - 4ms/step - loss: 0.0851 - val_loss: 0.0848</span></span>
+<span><span class="co">## 469/469 - 2s - 3ms/step - loss: 0.0851 - val_loss: 0.0847</span></span>
 <span><span class="co">## Epoch 54/100</span></span>
-<span><span class="co">## 469/469 - 2s - 4ms/step - loss: 0.0851 - val_loss: 0.0848</span></span>
+<span><span class="co">## 469/469 - 2s - 3ms/step - loss: 0.0850 - val_loss: 0.0847</span></span>
 <span><span class="co">## Epoch 55/100</span></span>
-<span><span class="co">## 469/469 - 2s - 4ms/step - loss: 0.0851 - val_loss: 0.0847</span></span>
+<span><span class="co">## 469/469 - 2s - 3ms/step - loss: 0.0850 - val_loss: 0.0847</span></span>
 <span><span class="co">## Epoch 56/100</span></span>
-<span><span class="co">## 469/469 - 2s - 4ms/step - loss: 0.0851 - val_loss: 0.0847</span></span>
+<span><span class="co">## 469/469 - 2s - 3ms/step - loss: 0.0850 - val_loss: 0.0846</span></span>
 <span><span class="co">## Epoch 57/100</span></span>
-<span><span class="co">## 469/469 - 2s - 4ms/step - loss: 0.0850 - val_loss: 0.0847</span></span>
+<span><span class="co">## 469/469 - 2s - 3ms/step - loss: 0.0850 - val_loss: 0.0846</span></span>
 <span><span class="co">## Epoch 58/100</span></span>
-<span><span class="co">## 469/469 - 2s - 4ms/step - loss: 0.0850 - val_loss: 0.0847</span></span>
+<span><span class="co">## 469/469 - 2s - 3ms/step - loss: 0.0850 - val_loss: 0.0846</span></span>
 <span><span class="co">## Epoch 59/100</span></span>
-<span><span class="co">## 469/469 - 2s - 4ms/step - loss: 0.0850 - val_loss: 0.0847</span></span>
+<span><span class="co">## 469/469 - 2s - 3ms/step - loss: 0.0849 - val_loss: 0.0846</span></span>
 <span><span class="co">## Epoch 60/100</span></span>
-<span><span class="co">## 469/469 - 2s - 4ms/step - loss: 0.0850 - val_loss: 0.0847</span></span>
+<span><span class="co">## 469/469 - 2s - 3ms/step - loss: 0.0849 - val_loss: 0.0846</span></span>
 <span><span class="co">## Epoch 61/100</span></span>
-<span><span class="co">## 469/469 - 2s - 4ms/step - loss: 0.0850 - val_loss: 0.0847</span></span>
+<span><span class="co">## 469/469 - 2s - 3ms/step - loss: 0.0849 - val_loss: 0.0846</span></span>
 <span><span class="co">## Epoch 62/100</span></span>
-<span><span class="co">## 469/469 - 2s - 4ms/step - loss: 0.0850 - val_loss: 0.0846</span></span>
+<span><span class="co">## 469/469 - 2s - 3ms/step - loss: 0.0849 - val_loss: 0.0846</span></span>
 <span><span class="co">## Epoch 63/100</span></span>
-<span><span class="co">## 469/469 - 2s - 4ms/step - loss: 0.0849 - val_loss: 0.0846</span></span>
+<span><span class="co">## 469/469 - 1s - 3ms/step - loss: 0.0849 - val_loss: 0.0845</span></span>
 <span><span class="co">## Epoch 64/100</span></span>
-<span><span class="co">## 469/469 - 2s - 4ms/step - loss: 0.0849 - val_loss: 0.0846</span></span>
+<span><span class="co">## 469/469 - 2s - 3ms/step - loss: 0.0849 - val_loss: 0.0845</span></span>
 <span><span class="co">## Epoch 65/100</span></span>
-<span><span class="co">## 469/469 - 2s - 4ms/step - loss: 0.0849 - val_loss: 0.0846</span></span>
+<span><span class="co">## 469/469 - 2s - 3ms/step - loss: 0.0848 - val_loss: 0.0845</span></span>
 <span><span class="co">## Epoch 66/100</span></span>
-<span><span class="co">## 469/469 - 2s - 4ms/step - loss: 0.0849 - val_loss: 0.0846</span></span>
+<span><span class="co">## 469/469 - 2s - 3ms/step - loss: 0.0848 - val_loss: 0.0845</span></span>
 <span><span class="co">## Epoch 67/100</span></span>
-<span><span class="co">## 469/469 - 2s - 4ms/step - loss: 0.0849 - val_loss: 0.0846</span></span>
+<span><span class="co">## 469/469 - 2s - 3ms/step - loss: 0.0848 - val_loss: 0.0845</span></span>
 <span><span class="co">## Epoch 68/100</span></span>
-<span><span class="co">## 469/469 - 2s - 4ms/step - loss: 0.0849 - val_loss: 0.0845</span></span>
+<span><span class="co">## 469/469 - 2s - 3ms/step - loss: 0.0848 - val_loss: 0.0845</span></span>
 <span><span class="co">## Epoch 69/100</span></span>
-<span><span class="co">## 469/469 - 2s - 4ms/step - loss: 0.0848 - val_loss: 0.0845</span></span>
+<span><span class="co">## 469/469 - 2s - 3ms/step - loss: 0.0848 - val_loss: 0.0845</span></span>
 <span><span class="co">## Epoch 70/100</span></span>
-<span><span class="co">## 469/469 - 2s - 4ms/step - loss: 0.0848 - val_loss: 0.0845</span></span>
+<span><span class="co">## 469/469 - 2s - 3ms/step - loss: 0.0848 - val_loss: 0.0845</span></span>
 <span><span class="co">## Epoch 71/100</span></span>
-<span><span class="co">## 469/469 - 2s - 4ms/step - loss: 0.0848 - val_loss: 0.0845</span></span>
+<span><span class="co">## 469/469 - 2s - 3ms/step - loss: 0.0848 - val_loss: 0.0844</span></span>
 <span><span class="co">## Epoch 72/100</span></span>
-<span><span class="co">## 469/469 - 2s - 4ms/step - loss: 0.0848 - val_loss: 0.0845</span></span>
+<span><span class="co">## 469/469 - 2s - 3ms/step - loss: 0.0847 - val_loss: 0.0844</span></span>
 <span><span class="co">## Epoch 73/100</span></span>
-<span><span class="co">## 469/469 - 2s - 4ms/step - loss: 0.0848 - val_loss: 0.0845</span></span>
+<span><span class="co">## 469/469 - 2s - 3ms/step - loss: 0.0847 - val_loss: 0.0844</span></span>
 <span><span class="co">## Epoch 74/100</span></span>
-<span><span class="co">## 469/469 - 2s - 4ms/step - loss: 0.0848 - val_loss: 0.0845</span></span>
+<span><span class="co">## 469/469 - 2s - 3ms/step - loss: 0.0847 - val_loss: 0.0844</span></span>
 <span><span class="co">## Epoch 75/100</span></span>
-<span><span class="co">## 469/469 - 2s - 4ms/step - loss: 0.0848 - val_loss: 0.0845</span></span>
+<span><span class="co">## 469/469 - 1s - 3ms/step - loss: 0.0847 - val_loss: 0.0844</span></span>
 <span><span class="co">## Epoch 76/100</span></span>
-<span><span class="co">## 469/469 - 2s - 4ms/step - loss: 0.0848 - val_loss: 0.0845</span></span>
+<span><span class="co">## 469/469 - 2s - 3ms/step - loss: 0.0847 - val_loss: 0.0844</span></span>
 <span><span class="co">## Epoch 77/100</span></span>
-<span><span class="co">## 469/469 - 2s - 4ms/step - loss: 0.0847 - val_loss: 0.0845</span></span>
+<span><span class="co">## 469/469 - 2s - 3ms/step - loss: 0.0847 - val_loss: 0.0844</span></span>
 <span><span class="co">## Epoch 78/100</span></span>
-<span><span class="co">## 469/469 - 2s - 4ms/step - loss: 0.0847 - val_loss: 0.0844</span></span>
+<span><span class="co">## 469/469 - 2s - 3ms/step - loss: 0.0847 - val_loss: 0.0844</span></span>
 <span><span class="co">## Epoch 79/100</span></span>
-<span><span class="co">## 469/469 - 2s - 4ms/step - loss: 0.0847 - val_loss: 0.0844</span></span>
+<span><span class="co">## 469/469 - 2s - 3ms/step - loss: 0.0847 - val_loss: 0.0844</span></span>
 <span><span class="co">## Epoch 80/100</span></span>
-<span><span class="co">## 469/469 - 2s - 4ms/step - loss: 0.0847 - val_loss: 0.0844</span></span>
+<span><span class="co">## 469/469 - 2s - 3ms/step - loss: 0.0847 - val_loss: 0.0844</span></span>
 <span><span class="co">## Epoch 81/100</span></span>
-<span><span class="co">## 469/469 - 2s - 4ms/step - loss: 0.0847 - val_loss: 0.0844</span></span>
+<span><span class="co">## 469/469 - 2s - 3ms/step - loss: 0.0846 - val_loss: 0.0844</span></span>
 <span><span class="co">## Epoch 82/100</span></span>
-<span><span class="co">## 469/469 - 2s - 4ms/step - loss: 0.0847 - val_loss: 0.0844</span></span>
+<span><span class="co">## 469/469 - 2s - 3ms/step - loss: 0.0846 - val_loss: 0.0844</span></span>
 <span><span class="co">## Epoch 83/100</span></span>
-<span><span class="co">## 469/469 - 2s - 4ms/step - loss: 0.0847 - val_loss: 0.0844</span></span>
+<span><span class="co">## 469/469 - 2s - 3ms/step - loss: 0.0846 - val_loss: 0.0844</span></span>
 <span><span class="co">## Epoch 84/100</span></span>
-<span><span class="co">## 469/469 - 2s - 4ms/step - loss: 0.0847 - val_loss: 0.0844</span></span>
+<span><span class="co">## 469/469 - 2s - 3ms/step - loss: 0.0846 - val_loss: 0.0843</span></span>
 <span><span class="co">## Epoch 85/100</span></span>
-<span><span class="co">## 469/469 - 2s - 4ms/step - loss: 0.0847 - val_loss: 0.0844</span></span>
+<span><span class="co">## 469/469 - 2s - 3ms/step - loss: 0.0846 - val_loss: 0.0843</span></span>
 <span><span class="co">## Epoch 86/100</span></span>
-<span><span class="co">## 469/469 - 2s - 4ms/step - loss: 0.0846 - val_loss: 0.0844</span></span>
+<span><span class="co">## 469/469 - 2s - 3ms/step - loss: 0.0846 - val_loss: 0.0843</span></span>
 <span><span class="co">## Epoch 87/100</span></span>
-<span><span class="co">## 469/469 - 2s - 4ms/step - loss: 0.0846 - val_loss: 0.0844</span></span>
+<span><span class="co">## 469/469 - 1s - 3ms/step - loss: 0.0846 - val_loss: 0.0843</span></span>
 <span><span class="co">## Epoch 88/100</span></span>
-<span><span class="co">## 469/469 - 2s - 4ms/step - loss: 0.0846 - val_loss: 0.0844</span></span>
+<span><span class="co">## 469/469 - 2s - 3ms/step - loss: 0.0846 - val_loss: 0.0843</span></span>
 <span><span class="co">## Epoch 89/100</span></span>
-<span><span class="co">## 469/469 - 2s - 4ms/step - loss: 0.0846 - val_loss: 0.0843</span></span>
+<span><span class="co">## 469/469 - 2s - 3ms/step - loss: 0.0846 - val_loss: 0.0843</span></span>
 <span><span class="co">## Epoch 90/100</span></span>
-<span><span class="co">## 469/469 - 2s - 4ms/step - loss: 0.0846 - val_loss: 0.0843</span></span>
+<span><span class="co">## 469/469 - 2s - 3ms/step - loss: 0.0846 - val_loss: 0.0843</span></span>
 <span><span class="co">## Epoch 91/100</span></span>
-<span><span class="co">## 469/469 - 2s - 4ms/step - loss: 0.0846 - val_loss: 0.0843</span></span>
+<span><span class="co">## 469/469 - 2s - 3ms/step - loss: 0.0845 - val_loss: 0.0843</span></span>
 <span><span class="co">## Epoch 92/100</span></span>
-<span><span class="co">## 469/469 - 2s - 4ms/step - loss: 0.0846 - val_loss: 0.0843</span></span>
+<span><span class="co">## 469/469 - 2s - 3ms/step - loss: 0.0845 - val_loss: 0.0843</span></span>
 <span><span class="co">## Epoch 93/100</span></span>
-<span><span class="co">## 469/469 - 2s - 4ms/step - loss: 0.0846 - val_loss: 0.0843</span></span>
+<span><span class="co">## 469/469 - 2s - 3ms/step - loss: 0.0845 - val_loss: 0.0843</span></span>
 <span><span class="co">## Epoch 94/100</span></span>
-<span><span class="co">## 469/469 - 2s - 4ms/step - loss: 0.0846 - val_loss: 0.0843</span></span>
+<span><span class="co">## 469/469 - 2s - 3ms/step - loss: 0.0845 - val_loss: 0.0843</span></span>
 <span><span class="co">## Epoch 95/100</span></span>
-<span><span class="co">## 469/469 - 2s - 4ms/step - loss: 0.0846 - val_loss: 0.0843</span></span>
+<span><span class="co">## 469/469 - 2s - 3ms/step - loss: 0.0845 - val_loss: 0.0843</span></span>
 <span><span class="co">## Epoch 96/100</span></span>
-<span><span class="co">## 469/469 - 2s - 4ms/step - loss: 0.0846 - val_loss: 0.0843</span></span>
+<span><span class="co">## 469/469 - 1s - 3ms/step - loss: 0.0845 - val_loss: 0.0842</span></span>
 <span><span class="co">## Epoch 97/100</span></span>
-<span><span class="co">## 469/469 - 2s - 4ms/step - loss: 0.0845 - val_loss: 0.0843</span></span>
+<span><span class="co">## 469/469 - 2s - 3ms/step - loss: 0.0845 - val_loss: 0.0842</span></span>
 <span><span class="co">## Epoch 98/100</span></span>
-<span><span class="co">## 469/469 - 2s - 4ms/step - loss: 0.0845 - val_loss: 0.0843</span></span>
+<span><span class="co">## 469/469 - 2s - 3ms/step - loss: 0.0845 - val_loss: 0.0842</span></span>
 <span><span class="co">## Epoch 99/100</span></span>
-<span><span class="co">## 469/469 - 2s - 4ms/step - loss: 0.0845 - val_loss: 0.0843</span></span>
+<span><span class="co">## 469/469 - 2s - 3ms/step - loss: 0.0845 - val_loss: 0.0842</span></span>
 <span><span class="co">## Epoch 100/100</span></span>
-<span><span class="co">## 469/469 - 2s - 4ms/step - loss: 0.0845 - val_loss: 0.0843</span></span></code></pre>
+<span><span class="co">## 469/469 - 1s - 3ms/step - loss: 0.0845 - val_loss: 0.0842</span></span></code></pre>
 <p>Let’s now predict on the noisy data and display the results of our
 autoencoder.</p>
 <p>Notice how the autoencoder does an amazing job at removing the noise
 from the input images.</p>
 <div class="sourceCode" id="cb12"><pre class="downlit sourceCode r">
 <code class="sourceCode R"><span><span class="va">predictions</span> <span class="op">&lt;-</span> <span class="va">autoencoder</span> <span class="op">|&gt;</span> <span class="fu"><a href="https://rdrr.io/r/stats/predict.html" class="external-link">predict</a></span><span class="op">(</span><span class="va">noisy_test_data</span><span class="op">)</span></span></code></pre></div>
-<pre><code><span><span class="co">## 313/313 - 0s - 1ms/step</span></span></code></pre>
+<pre><code><span><span class="co">## 313/313 - 0s - 775us/step</span></span></code></pre>
 <div class="sourceCode" id="cb14"><pre class="downlit sourceCode r">
 <code class="sourceCode R"><span><span class="fu">display</span><span class="op">(</span><span class="va">noisy_test_data</span>, <span class="va">predictions</span><span class="op">)</span></span></code></pre></div>
 <div class="float">
@@ -568,7 +568,7 @@ <h2 id="build-the-autoencoder">Build the autoencoder<a class="anchor" aria-label
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer>
diff --git a/docs/dev/articles/examples/vision/autoencoder/unnamed-chunk-5-1.png b/docs/dev/articles/examples/vision/autoencoder/unnamed-chunk-5-1.png
index 1aa75cd9346d10667b7685ff13d5757b23f96959..095cb80e36063dd419dca38332e7c74ee8cf1a93 100644
GIT binary patch
delta 53605
zcma%?Wl$Ymu&xu_H9&9(?(QxL65O5O?!je(yA#|=aCdityX(fCy>XYrcdO3td#dK=
zRIOEOW_7>a{k+@7(8Gn$)w0lluczR`D(>AGXDsDaT%2e2D28$&!dJS#Nl`KpY9W%e
zU+8ht7$nT|$I=cw#{?%`{v37YxI<pNv!(8`!JZJl;tAi1x5*>zh1au=5Xht7+eNII
zZsY&?0sil={{P>61MCJOKz_O{&hw=WCp`AlE3MI)2?-Gq5$WmaAt4a~nMpw&qlKs^
zixv7E)_V>Nez$<4{?ERaPJ#mgd{`<f=bVUitX743PYOrh@wZY4gJy+6o!j~&6XovQ
zwOqNL&odNMfOv23m;8WsJzD|yDV2EtLkGPsw|NJVrzYmcihEQSU+4*U1p{GWzr)U4
z3{oBejy^Z*OS|*UD&4EDgC~*q-Fcn0q7w2rkQW&J*a0wcQ<RkibIYBT)a75QBif^a
z5J+h_gqWz77nleaDxa6TTx-sn)NQ((I<1h|_?RdNW8+fm+XDjfu^bKea`R8K^`X|v
zqw3YjZZj6D9!Rfz2O@A7F7AlV6zsQzE=zS<Ev_f<KNDo68nE*I{aaY=-t+<f3QY$o
zf$2o^<rK(iD-Q87Y%-d~bsU&wg9h!PkxcVdc<+~Q@cBYhtnY-P56L39)Ui$}$74F^
z+=(1KMX1{>IIKbCxKU@X*Om_6jFsD+of6ygf!sPy9LMF}`pIGGj11hU`6DVa7M@Q*
zZK>}Iv>9=*A>vb^a#Sq|&zo4tR~ghb)^r#_fXzcw-{HeKOfzjZVaH>up)j8E*X1sE
zHq^&X;~RbqcNHXSeza-d<083AexY-}(-S<5=hLsvIVrO@$QCgqaVrCpJ(AxW&CVuY
zdm=4)K|Gjt6h(z5n;RX5`h)io1x1U{5Gh5UoJV-!b`$quQC+(?dolZDVzpK7Hx^<q
z;PBC*tHPiwJT)o6T!5mhzH|b1wM~)2WE(3{b%q_ZPc0>QQ{CmG@LlFxR8E$Eg2(#g
z>p-MMD#S)fqbV!?tMs>*IatC)8Y9g3>6(jhnJU$MD2RY|+F|i%1hNyyY&`Z$Eb$|V
z^?=S&RJQH=#l2^nx7bAs?10(t22_X+AW&MnBLNoBIkX~*TOJsdYqNX#-vxNE5>@Oh
zO!pF$gq=1PD9ET81gdh2|FqY(a*Obt4xgP+(<I%=AULzD*u9Wm5<{M;joFXSa-$OY
z%Yku<dvzllDfhRjrQNZ$l)S>;-2-%@F-0p{8AO+<M7#I@%Hsm%$`@ES$3oJ9(_r?~
zmJd7u`(Ly~Lv&W#ydL4D{;iJ8Q22uNxL_d9_QJR;c7%SnEC$6DWY>(M4@eQbU)Ya{
z8TFv@6KV+_HJWn1nW_D}K|_g^?|eHyI<d6Te*A8@coIT`J{*3<s%-nXEaI0JqJLiO
zAvKW*1!{qH|2|C2Rt>bCU9mURHx!f=IXHU!G%hWproq&6PnonX847b|i+fu~7s}HQ
zJW*4gX0N>p2QOFo+&53Fr%2@uKe%>4OLSkPwI~`6t}uZ9!Flqf<_HurGIfhLJXpLv
z!mjw_pe2DRyB~;c+rK2FLMAp?uv|(7+Pt5{)&^;cya|z1zW~{-t7%z6`SqbW4-m+c
z>{6ZR00SQmlMn?5A&A)umaAem1kX;NR77!z-kawGF(4vv_i*B2ArCH_qN!YzS0r6w
z=~bxPdtWbj8R@f|$FUK<giKx5$4`)iC$Wo42XmJ}0ZQhAKbS$jmZrAG&R*u5`q~=L
zS5offZ#+JsKnML&9a+dQBQ%EqPuDq#XM>4_9HXs1_$ide9Cy(*D#{;P<Z5f@pbC54
zHJG)-wVrcI1#2gt;&Hjl&!`c~ec6s|C8SoHxpSQzp7SRPvw{2dxhF}6hSEAeuOK_W
zFdLT;hrpah)*ChbVcw`Zg`F89_zZ9Xy<MyxHxP0I*#0&#$^L19nwCv}qAr#76B%wR
z(V_cxRGno{be{gv6m3lUq&{6goR~VOTA7O%SGBDNtCp&?3xol;Rfci?=zcc@7`O5n
zGPFmc)GTE5<VBUf;O!e-5@GUqnY7^faUrtfMAg};m+8uo+9ny6K$4AGUT>|C9HW!T
z*9T;v{`Y${`SRsslOP2{Q+K6MZA9tdWZLSQAI3d9pHNeCX-$Ru=OEwPzO_L`!%l@8
z9?kvEa5T?dQ%?nnZ@(@er-E`P-FEQKgykHQ42+$cTh%e~!5ynRajAc<+h<nxE6G3I
zETeb7gm)&V_l{}Ci{07E$BZIFC?A&57^gIV%3vX6dVV+!67rf3BzUd{R3`h@rw}Uh
z-<pd&O?N0!3KRL&z~l4Y$@P2aP~I;~@dtBL5AMD$OjO;w(Tk0LNNn|mN6TTaY2Di+
zjZaclGgzJ5;NTR7&dPbZznDEJ$(tw~2^jlBONxiAZ^ouElXKPsWjD{S)EL;z;erf=
zkJ1zKyRDo@6j+F=SsVE(iqM;E0QU$5uMbyv825P-|1R6>gB%^CT!{bbGp_pGR=u5s
zK%vhN2JtS%kPfHnx%goaC2Gtqt!=FHaB_1&>kaaWaI7qDC~05cj=Lxl_DOgz4f{31
zBiE~+7)=_(bA<j%J@xOSxt`GF-#`Z<#_1i1U!U>$o}+zzMP_DIR1*>jAP{TU+g#pN
z(A`|seTvg3=+NB$*FY&nLPJ--&scAb`8br9st^k!;qTSxiG3Y*(er;FYC4;kiGFY8
z-!C&iYkoVNDL3daA5U+vF<S4my2AakT|ri2vbBx(5t4DNO#FhJ%-z%u^#ufcnsbYp
zd_plsoA|){m<UabdL9|;@pJTf&cACMFSfyo_09k4g!==?|D@Q@%-HBa9QKq=zu^0#
zJ_+2{BAf+RqX$Mg8w79{>5*Mo6|JOUdk<(YlIEusx9wUpB8(5RxSI0p3<qn?6;im&
zD6x`=m&#=S^D$~s8#<H$wOI<lbEDvc?SXL<OM0*`9CgE~iD%yKg@~1THMsZnmBc|%
z_%QV?!?AOAn&Z@TXBX!aPD;XWO(AC$NPBQYV41tp?n!2Zdj~7Hjy4C@2wsZBLu4AX
zp_HBTN`D!CwKqG`ru~ZBM0`4v`;3$6l3@48bl33`;dvkP*N2Qif<Z-qFx8;b+<kQZ
zD&A09O+~{=f3<tA$^Y%z3EX!&_LLX%ICHu<J(=CCb!0Pw^vYEGh;uks^{-WJlY<7j
z6S6csGTg_~?j0NTo)z{lR1YQ$TiyB|z|bIz=_*k2Tm$-#V3aY7>AcX1*q>bLyMG5l
z;ny0t;li0_LEBRW%&35ziS#fXZAr1!(I!d=I-j4C#8XuKG7jP3D)^4{2{!cGzampS
zO&<5y_8fG<pTt4FUZq`sTgjNf1GC<@c#35*S1(M*9=&$(y)im7tN7Y{p22Q1;NJqd
zZj!p1pxtepj0|QldNz!pcUoM!s><w|P!z)t<xP=14iTX#9aSKegmnMFbyiRF#bNtD
zb%eUDfa1VQ!2d`bB{#P8>6^0p&OLrXq=uV<JE<_L4=yhL-Ld@RYVT$1er3<avI7+Q
zj<66H+s3-q?X9|Gx32|5!47=?8_7DuQnmygDyHOuEvN@Zg+Lh*-}ASG4!_qwyb4=l
z$eQ=z4k;Ec4v>?SuCJxM2$l8rYa0y;9qlD_`EE85TPXUkcDRuugiq^`RQ=P=v-eLX
z%}Q<bZH14d+|=e8+-S@sMu_-dZwmMix<~g9Z@YOYzX%`6>zFI}DQlapx7}v<)Rdru
z;9mPs9DQy76g09jQSZi=tq*VzIJfayzM;KX3%Y%BEdktEYn;y!dgL^&U<5B4o7TWj
zFd-dLGSUj;V|27VnWR~dtfAIgN<-ZeZ+_O!KJ^9EZ{yzqMXB!+cIx=#8pmA&aprp~
z(UN4~Rebl_S{LW5OXP^tc-3hbnP(gUi&&|TMm1G+5waunWhVq14fXCfXfaWWlYW`#
z&oqZm{lL=s4;wQD=Vnkx*YB$(R;YHDV%?sJar8k40-A|*y;iW};FXM$maVh7n}Lyx
zr>nk+i>?Kvd+2_SaiR4-0P4*M?p~OEP9+<M^tpR@<UFV~a|vWd*<roDU$Kj!_p#Fz
z<3@iw!MSva?B6WSy_vX<fwr%=lqNcN1VK)wAArujG6{n7Gbi}hQB}gpA>9Ae*iceg
zm`@2(80iYUYQ7~xq@yB+M)q2JSEfQ{rhe-^XFanA%2hL0qbC{Z<ZJm3KR(Yx97%#5
zjJhR6g!tY!1JX*u+S>72m(<laP>Hh>I08Y1)mXnI+gYOZFpyEwQ`1q^8Q!FB1lz;_
z;8=+-hGw7CxLi>uH?tcYR=BSG-_HH4eN3yniaJ@TW*{5mqh5n(BjCr3UzlXZ1U`^I
zyPRZTpI&2?C-sJjRIQC>Ul8Q$*h$j>7n9ItdEWz}*Zf<B6|7HA8UZQsF{EN#9^s;%
zN8oR*p?Fwg{NrMXL4}ro0~5FX-`rJTxQFHo9TP)QalYIAP)*$d)0w@u|0NY2MOS-X
z2SC(e&6@FWun1ZnOe?9`dIh67t_}nXHoFd-rylz`ebi_*e3&bsNBeUHb5BVmBzp$^
zz}S4uyvpk-h=daBebrc8TH=BcOdOr(=P-3ge&}s={5@w?15WBz2I;EWwulUP{Ow;j
z<qlIxp10BiXT(K+OrWz;G;v6np|ir@B7%ev?U|KksDYetAKt>J$9<L6J=sA@3WiwQ
zl#VK*-L;tbzV7zCBy>HN4C;ZZov;yMi75!1x|wCPVdI8!1Whchtz1UWwq`9@uvIs3
zHRhUbQ53ODn?S3fM;CdW6AM7$N@mcHDnfI9cq&Pc<<C~qJZ~eky0*_S$C~tTXdj)l
z&7u=q52ED8nW~mI+I=DMZ&yP>Nlmb3T=K-e6sj)GKli=;-&nc$r{-6*RI@R$53L;h
z_Rp)(*s^olF#|a!k*q&!Q>oSE@N^mb&JOfFSxHOUI)AfwGV}eu*u4Zifx28^GO~z{
zAhsn{0i@5w=RKo{7rAB$(OvA7uTAP0a*eKpd_rxxLG3FSb$HDBHJwf;xB4uQqMkfu
zL<jaOrH``$oUbILcMv#|48^29nowc48ulrbC93XWImNG+W%e`ACm%wxNMmy9Z=%K-
zD?!bBUX!alhhI&7oV@@AWo=i9@0y@<7;dkZ>jw}tI_Y!TMGrXnam4t(2gE7h>WGfV
zKtVj;$kcUBjS5#wR`ACRCoV4Zb0A?8IQMleGNq<!!*usklEwm8gWD}ywEXxsO%(|F
z`Sf1&pe=poIqL8Oy1;BAeL`2`l_hnZ*ZQ|ze%`AD3{4X^R9*mp<T-3!rLcup&DLE-
zGFTJpuCzUc4i?pfKXt*AP<iEmOlFsC<7PmerGRM{>nxg%?2Mt0J6aNiwN4CKEc9no
z1yPR`x(^BF6Fn6NO;>YyX=6=G)~~F8Gs}?odtF~u(b-PM!8f-$e0gf5Ee6ML?VP|w
z2#nil_v#>M04-NCL!|s>ww_num%&%^t?6(F_Bx}C(#DUU&V*hh3b1F3t>3O89F~mW
zoTAXSjIzi&)tx}0GnQkbq}{?iN(%DFfhIG}iq-nxD5mx0;@;|tpLN$;(ZS8lyVs|^
zwdr2XbW`DEJG_*87J7(xS>0WvA1ClSe6a~P?Y&={fo*$k=a~x04DNFM7Y1;ciIZWt
zoAKdiZC>{?%|$+7{K4-B?PW(UijC@)PtGr)NRXz}eKYXL$rqL(DbnIQ|H0>cY_$=~
zZsR@@GPpD9ZtwJ!$#+!uE^udaR6}`jZCQks^IeCBd_=Uk@I;Ncbcc%l=%BLGN5ZF_
zsh+YAtam!~JvP{Ayrq|8G%pLRh5KhA`;`RJL{+W(T??{NF`!erj1&7pdmNsUQeHK`
zYf;W=kd7)$o?CSYJgMM%)v3USh65#aD6$w7`*ACtgBR>T2NPqJ9T>*y^Z?Rr<9M6Z
zt%zDbk845ob$jr<i}MvK@@eEo(vcw~(NGO|Eq=vG8O$4o-Q$wC6AIjcTxg#z0>-z|
z1#bi%pYIfRgdp_JpngGbDC_%z%|II67A!ql2EUSOYz+95>Wa_THOOJIBeX=*ZKlAa
z2>ajScHG_bzC@7Rh9g-dI;r{AsgswrwKI}0(z&iEQ`RJg-l|Gg4(?s=9a{6*Eh51F
z8c<*v06`FvkyU8P3Mbw+-%DzUF3%(pj+~^gva>!R`TU*&fr%~*4C*P*!<TFCH73#O
z@Xr0x=lXjkk*1SxI+tJG<@ITo#@a=og`QzHHrkpQjyf8#kd^7t<`&*7eH#?y*=N(K
z?zUt&yk>eU&;PYETNt5MQpo0<2l!R<#=pF%7k|Znj2Dsw><Ek<y45-%p03m>%E+tg
z-u1es!1qj-z&6NN<o+bg$<BBybvjS5+I=26pCVgH4zO%%;rn??zJk(rYzXJMl)b2d
zVMhK03mvN)vLYvoS~Iiljf2s;Tb_4BMYyIR_tMtLtkl(>^F)1axyvui4+t)SR@bK&
zE-OERLvL@<qjTYI;dv|4PVMLe>nvjl&5P#$`0=4Yi4PSlpA<6vQM(r4yAK-pdvSHk
zzylFR3-?RZ{ip{aB+ANs2x}V_Q#ugt#ugFgh7#Xcot>Xvh=?FcJ)GSB-ahPq+nR8?
z$TkUC&B#()8ITRxe*F%c4OF9i?dzw4ky0wUL7B|ZHK=oOya>u5*Zw+RZT~fQ>!G_~
zc=hmH)X-h(q%CBn)R{NO*81oJEc5fJ*#6zoiKM%Plh<jTlzU}Hrc(TF-dh9In*)NU
zTc^H?Gki}G+O#Q8<{HUk=)m9o<azg>cSa3_Zhd_WB&ZouA#jhPQ-E$p27z_i3FyDj
z6(hH~!H;oI7dM}>SNJbf#F<m%Z#yB(t7hcWGy5lsJLa^&!3l$Ad+*t!n6{Lr*tf`H
zO{<o_T;uD_`7asWKqjmcDU6YWRe1$H!-#j}iCw+hZBLyD*VWeS;zea^FHihOgBaXL
zortcL5hCkFQ3c^Jpj=$kicg(MAKL|{;=+%B2=k7*thzQ2Rn5T!d)Dy)FHkX{eqSiF
z4fR)KWkX9%Wov%@X9?r(K}Q1g)(jGGI;cPi(}F_xViY#J8fAd_OPLo|^4W&4pV(yI
z7ZW9i#?rh7LSdtG*r#jxO$%zVwYHUwdG+78`=)=|Bjjbrff1I9QU~&Sp3MYX1C>;W
zzHZ&SLd@^NJ%?jhMQj(cg-SposFphNxqD6gb-<$hQP<$EvH(4Bqz_dT%Hz{Sx)YDS
znYNRkf`$$^k$(Lhiu{i!XHpTs^SzYQQ_al6I)GUJEu!yHcpV+cszzxDR6O34Q6=O7
zrA#}850~Hr4p&Iu%Z}4meppLr0hFIJ+izSOxaT!K9hi5@L?HsWm>Egg(DWM%G1;$1
z*UV9_YI~G|ckc;KxP+Qv?3Td}VP}O4LkG^&`?E*?9+#5v0pTYCSlPShnGv0(-*8n=
z*WRauuLp}P&E;qYuKH(8(XLDCchq!XS91fa@wN%Tul{rN6fH+VM}@1I=TTL_deu?c
zcjve_XQv#)5G-|GbK4yfjEqD>Efs^{A)C%&ZQ(EHv^|@<6>X^f9kc>Wbllr%BAF{x
zh30sgT#Y}kQGOlRgHJ4VzvAvnG+&=wkZ-PV+3;Y+i(2AGR+m50ap>~ooG<;Uw?!a4
zJM9qxw3}bLs%Xzy!Io1S3u{WM%9%M?uheYZ?40~`6jkMw4HkZh=5FP0tK)Q^?{m(s
zx(JQ%WmWVikL`bm`*9%MV>GXK*-gHDZ-P_U`1Z)ZG4J18ME$J`!DHKX5IMg_#YJ{X
zKm_KPbm&(C=AQfF?XwjsY`7t^Q?SiqE(f|3FuM(3FK-Ke{QCC*SF^XtE|b;csnFyK
zKV>o&<+`3{Q8NW^&pdzzycd~UG+B_qj?9q3j~H$H{EMDN=(z{g1l39*B8S*ra0qG}
zAzY?0K3+8+*1bnR!5jTVoByd@jse~XLP{i7@&FGdOEhYsj0k_FeO|Me6xj6QD!v;D
zh~~h%XPs3EY}}wBM1CKpE3>yeI}ZK&W@+R3Evc*R%#&yiYyn)n$%4dX6=b4UvE!1I
z1z<_|a?W?#XG%kondAOqZq_!3pq>M6UOuL|O?7AbLe1MQbu$Ur#$OCzo`gIdCvlQL
zSb5?^ue0Jv){AULLM)$CSYBxa-UzdSW>^)EQ`5c8{lGtP&hTnn^(YmY9fs{A6|gC5
z7~qNC9Iv$#ZXqu2rPXC9mX`MNxj)TAQ$M<-49JcbG?~-2mexBVq_{cBhC1s@3j&;M
z$d!+mQI^2F;^F9mdl_}O-=`efC;O33xQfb`TbSnxo<Zfw!9&#kJCDjl?27?dh<+fz
za5JV3@mdA>H!;z0aPg5746Ln=zQiO3uHWe1`D70k?yGY*xf_kvo<}=KIQU?A?JNV2
z>Rm`NdRy(DG=w}|Va;>GusE*cCj1sCa$~k&K`-1K?Uf)%_qZQC)yCp947e(4(J9`D
z{8bno(!vL-@9=sExZqpIegyWpnbDv7u<I~rZGI5J3Ulyx@B9lkq&#OJzn~r4!Y8-I
z&2s08Hldk3%t+ODzw{wUj#Ji}T?$}t#qaILA@=gvr-5(Vm`~dM(-DCL;ND+f1bW*u
z$J{n9R_S$GTH9IycWk?&VsCt2`{tVgtRu$|J-#w}|B;rQ13pU78c=OU*t-58Nm+fv
z6IR;lygt#hJj%jIURT`c`x*ZC>p7|{QNJ%1Zq?b1;~FmJ^<ifyL)Zh?XygIvjY7`U
zjGE^3Oq-?4labO~MpG~j>Y1+z$9p@Lp4|irvYe46<RvQmE7)sPal%$}r)?%K%~nM3
zK8UTU+1h*dVFO#z7*M%EtAzjc6L$-xdYa-`xi?1MUDeD_N8E$#aV!tAk2gVxe|VJ5
z?<Dt@khPk4aDh)%oq0x@_I!Tx;<zZFH7NDSYq+3}`PAYc`geC_oHA+Ez4kq%fhA5p
zrc{9(e>%0@ZC=cFpEoC+&a5nSIw>Xo&?f$9c5&4AG+eXg4|G<B%rwbhe!SB!uhPkl
z%MmQ}lvdYPevhXr7em^T9ymD(?_G--j?32*ErFq9XdwrKxmlHDT+^@qZ)t;m<Nqp_
z=A3x+z`}nWtUcE!ruQmSnjAzeugiWa#dGb9rnWrbQ=Td-#6+XqG^y=(gd<&cLQB|N
z{{%-(I8{&(0HJee^bB;CP>Ybqpw1Dx(sBp?lC6}Zyor#@wYhs%hrvjGc(=NQS4ogE
z#F7}3Z(Tvbe_bUawEl)!YZ=t?c5fvrO{)j@#c}cZq#K{hQp}hG1uJ7=BvaE!*xn6q
zGIM5OMoCi{3KK+8qpPO%D^^PHqIhF3gTycXy@X}}(py7}HW}(1uifa$&N(GT>bzmm
zUvO0jBz5Wx_lD>snhCa!&alrJ8R?nn&g?$Q$+xnx$1b-xV9b&_g_5wjKX~(N9YY_)
zl4z)FEYY-S>d2~y+ukwrMRV-mlDx(jXWBssvmV&FmJBp6Z(XlBstEBEXd`N$`@2nS
zUJ0dvVthWBe$ymQ#lV!|ae+fK+b?TxBP`5S_Ld>YKN3M2?4a8tQJqDtYwERwTdsng
zLqZVKm!ZM&;hujZWBtv6*}CFKsqw$B`VQEt>b{={z!hOu8R?z-4t{?eL<i#-7mz=O
z2O{00SJ4#yvDm{|t}{I2ZN$Ywxf4=lcR~6NFxWVqh0<OBGpeB1+q5X$%!5EdG;GN-
zdHmxbAmmKE&8@Bd?cL2E7A83HvvYOrH)ND-sBLG3U2^Z5`jZ*uH<;$tRH9Im<s5>0
zkl&LX1sySRzS%5S$No=|&ApSg-kFH>M|Zm{EmZhYB`Evb9GIaVsimS5Tf1^k!W{_U
zrfw*a9GemNJ1sstCOsfCD(WZUu!ga**M<<hEAvt-d{7@HBJ;=&G_WlClpDoL$TY1;
zB;VWh1cyZ+Cn2`FnUGcH?)V8RMq|Fi<^=pT7$u$FuZ<K6S_EFeb+gunUe4>a`?Yr%
zx-A&igI7DvZAwL1Y!6=+2?<F<*^>z<uyOa9vRho@<GeAe?fQ$<_eGCGu+`AxpfSgN
z)yN+t*V9e+#aHwyspRHNiUn&UIWRD$fFK(()ouAknwd8=DIbfPKO?jkoKle&!imR6
zhX_ZmD^^{B0>w_?m8@WyuD!g*bK^=hts&BEX5=BRu92FRe18B-^3oe^y1ER&Q0}V5
z;_EfK0}*}rKW}2?#-73DJ7nDFikC(N+(oPD#~a_C_%lvu!`WwMl@jdM+l`p~Db1Ov
z+Lrti&2J4R^AkbhZehbLi5(dJ*u<AZ%a@t^!eWJ_q#masgV6>}@nTEWy?NMh`)-x7
z;>NQP$NY$DeL*cdQbegA(0PH>%+JSS&qCUCM(D)M*-9$|;+Ni5#9zD&=o)76&N`m^
z3_F-?j<cTmr`0pw__s@*P+dAGB8=CrDH+IL58+$iq9N5jMJ9XLtL{oCQN!wk-ko`f
z7#6?!WB}@16DPo*Z_FjE;WPNmva*bEVl=|f>)^PhkKJc8onpP!YzJtaFJ4Cv=(6VG
zi!APS!X~`y4(pHVJx=ZnpTzbZ|J_2zHc8T^bws*MQ}42h{@&S`%6H8Olc6Jjq3*i(
zQ$O$3cXb4?u(A?#FDgStx#WH&i)Q$L6V<5~eaxQgFUH$wV!TEhMqvHb{NJ4VX2d~|
zd}V^E#JRVhV2;ZPAD}`N!ZxJwCXk)k*%W+zmQknk`!=Iq=4ZNl;49U(fil+KW||b6
z$=zz}{nTk_>7ygO;DDdyBNF<NQK?;u$UpqU2eJ<`l45Gpq#xjXd_F-jJ$A=e(GQP%
z5W^oeMaMFPJ1HrPR-ZpMlvRtp_Z){(zwXwfRIPhHeqrXSDg|2bU_0qqxNc!h8ACN+
zVnYnJG(If$s<Tpalk}|D(o4C@Gdiq!Zjur_Q5V%$`es|-Bgga``S|!i)lKKUW>4HE
z7d~DFlrPq7^#~ipBtPmP8E5{a1>0!&65ccY%rutOfz!+c!+-lL3RhgEisAK}2p@sj
z+i@MSF0RS|l>&GRT%(CwvR#RA8jh4^u*&4*Y;wEwz#@u|%hueOnqKGOg`ycUWc{#p
ziqZ=DN1lt6eK&EtuwPiZojDgYbl|tayeOwYl7#z%I#k-qalJxMEKKl7A<sAcpprMx
z0_>{rRdtdVLC|YQBB#}WXC|SK)m?=3G6>8$WTEXLdIq48xEFkHoXGB+#Qc}v_}4r2
z{Tqs^VwQKcb^NVqtSv_f5@I~B39ZX|;A7eof*Ml;eMEN1+gqZl@?0Pi%)uOf;L0)y
zo2q4{U|R#N@3%M3*t?Bz*;Eg`{Q%K5a=0KyoWZiJ@o~hBi69}x`etb>8Gs_9^WR5I
z(m%@6-hipL6<rY5oVsT6-e0P3A@?5ZjqoKA>E*OELf|yr4y-{Pbn&ieWzqff<?m>5
zk0*B8CU;9T8285k*FlbNZ43Y2=Oo%g8n2t(RlWCa!;OJ=@*3KzgNF&xiA9rjv}?zP
zcbm`US-%R8PX@H#yVkDwnPTrl2w~?gGv8rpDL_J5HY_XqTA(#NqhvYAY~EF_Qi9b>
zLH%iI@ve_od~{GDWwKb-S-I*Z-Z&Bli?H^LvAyk<o@7SY!fk#FOmU9Z=P6LU;vI5M
z8_AO66vc(19PE3Hp{Svu7x%=?_n1~yRCi`ppqdVU8{y2MY`5I*1g$pqjQSnl=uqbK
zJ@98<A!EJyj+2wT;7_N`HFkbA?4BMOj4pj|_nVRjS4@@dEN*F5yYbU}{mpoVgQp-`
zjL&anEe(Cp+D9211hu|up-^V`AH+lNW!-zx)Gia|wZC&@>w?%^*{-OEF@jj}^3v26
z9`RLFXRo<4c;{tvsXU@l8RlCt)42S11i-n8doFJ4uRc7*TY<1AR5&UQ_9U}5P6fw7
z#i>3iOwd5lK91&G_c9wCtABGi%V(QWUcp;lGJW~4Rz~(3j~SXQ&>i{?TAy=fls0I{
z`-dg#P^)l}I=Ye4L5<)K4X49b3-+-&c|rJaB>A?9U*0=sa$?+LvacrCy1ff5#6W0R
z`HIXA!)87-4N2pI&J;9{FqcFw!7Hx-m+%^Bb{eW2f2D2ZGN6erTsg1X^@)MHRVwN3
z9c`$aQH{5KR!DCQcc#htlV1pi6>xtmj;dXVe@Oco)YKPj;taJ46nD~-XB3&>ZFrwI
zzf!fW-sN<L$q&=nc)%}$elSTm0@+h3XRt=VqL!wrwoYndTKXqa;w{KtlvNS=M>gvj
zrf&E{m<{C5p4(?@SR+d5J4Q-svSRXsLsA&`E8C2Yx*p3J{xdnz&!PJbya!@beceq`
z4?!2|c#v*K$L&!<eoU+JPexeKWlOOf*c8xY;Am7-)SK?(;}K+bWd(N-Fz`?Bt@*ff
zUGP-k$ZDw3Ugs%m+mwQjqG`$R2D>^QGhSy#Flyn?YKHYFGWIh>>yQ=^@+-gB#!qb%
zk&+vwJwbL5g^@_q82g2v@Dc-y*9L)uH>v^Fz}*f*R9ib_JgcBr{8`9BJ1tWUL^IOU
z8)o%&yK~(kiwr8fVq4i4h}6w~Wl6j$T@4jcxNe#k_qd6K=V3XfjjU0AUW$NQ{RzJ(
zP{%qq?Bt2U+`3j964O%xVsZr@_<4kcVODFoxaQ~Q)z#I1{|?q3X3Kls05?^p9g`@p
zQFI5C{pf1DZ1J~+FDj0@oUX%dYL@_g>L3iU%OO{S;ZZ9J_A(FwydCkdr4hwc>W`&h
zNu-SXzO*Y!K-fKgy<;5_*1AtL!drMg2RwNgmY@G+{mLEl$ig7T-M+hjtzh`JGNSS<
zFr3SH{s?jKnV@OxG^%o@*A4x#c_QngvMt_0Jg!(4R`-_M-G_vm?BsZmLQv~^mMr27
zj+pJUK*V51m9lsPTH%6D)atv+=vc-SzJHgOmzR<8yrLHVvU?QR)?~{huRzL2#?(9f
zfj;f+@EMNX3(d%iYJ7Gd4LzF%j~jIynw_H$vYM&7J`Q8u{TE&E()Y$X0EMI|wHLEv
z6fOIs@^BjLZ}}_<)gXg$i*kpLKBV}E-z@3F(&QCo6rv*0008&b7Aj+nsh_}_+xjo(
zR?w<BftxZPs{T8VkuxR9cn*DbV@B2y{vo^qEQW|SGGFE=N*DQkpQjfImrh!mgxCDX
z(IL*~tYlr4^|@&g?yKN#9$DHSI$K{vPG<yx*XQNBxVX5@B6J~pP`ka6wIh}cZ0*z*
z{St(~%m1eU&^ARf(8q6ndzV9H<e&5TC={3~4}))fv^Eyt*_zq<d!0>2RQo#sZ8cD!
z*v@jxagqh9zoWgf!QGS~lp^ElS3y2BB5?@icOVM`8z1wLaR`L>k%{QEQaAl1M=1sS
zz@(6TC5mCY^C~u=JMvV-=P{t-BoQxs`FBMhDGfY7K)`UxVes{#Su0!p_wFAZXo|O0
zEJL@Tz0ZR2$i=ngWL5Vkg5^<ZsW0N)c%m@GjD`d~<1r&Q-XtVousfu+rG;<Vu!)mn
zW>!pR%EuN(tv@&Ep(%v8M|5);hH7tKF?Hh{#Z!HXH^|(J)@mr`y3;6^c_OMTu=6R0
zB^k33Fv+D(Elzswp|Fsv;>#0eH#)#`rQIqCT@-*C(!@FVLo*Nr|0I0d2{{6LtqyO6
zH2fYIxcYW5VOcUFgHR30T|)83d)pdxC^W)ec0Wui8P6J=#I9xdu?a7bu}RAL4-zxa
z=Ss(X42&bB3?UHJqntt+$F(CQMuy8xq=8<5{v~*FWa8vWwQfARvA4X8zWFi@4b4g|
z9UZOIw8N9^IsWoeN``RWjG;!`m2(KHKtKiPTPQ*ApqwoPmTI2|ZGuIA_nS~Hk01=x
z8~oeD_CqJ`pE}Avp9GeiM%gq^^*aLJIRhhIa33~Dv2Gxke<?qJ_Xuk#6M8Z|qo4!G
z8s}n&{EjCdje7mlwZ4H9oo+Y`(Yc%6zOVJIc@xe1^V6K`L>~#<C&tgOyj*iPm94x2
zRKne7>96Eg_UR2vG+p)nhArlO=EY%fH0+n53WtX6;GdHR+G81O&QOOhqdxqzFrdbO
z%rQ+&t!1~z>k?Idej3GW-j6l75zg#@cYn|WZ#=^7KN0ECMTY{jE$l>fK5Nr6lGM0G
z*8AVxH_3QcyM@z6h{+I6uaxe)n~d(XqI?tVwSr>LmE95frqI_sC#BEG%%LdJ^1h<$
z1Rl$%cdm_`x$?RFDLsl{zEz#n&&Mb4C47nRY4JaDbsQ>KzU2;MaY0)YPE$#M^|b7*
zkEbmyUHx;?DD|RVdrn?AkWO%HUrG0#A5ku=x#yNfp`qTO>9si%{ROa|v5*LgCUyC|
z1QGqx!oCf3yw8Xds|BL-XBO8tW<K<0r~JO<C%-N^s3^n0EGsU~C1YWsX)^q#qavp!
z{VOHXjwYt{0R8y<hLZDL+TH>rQ=HBFxEMLe&?1p!C|P!}PQqya2of02^R$}g{0FZf
zm$E;8R7?3Av2A)*Sz7jac$@DF13^xqJ4F-wtr7W|*SpPAMFz~=DJZUJWo4yAN=YD=
zHchYCpe1HZBQZZ_?!tE@Z}ncvoeQBoF6tPOf>s#|{zh9^9`&w%?l-V!K`nAs&$S>q
zUPg-P5j7eV?9xOu#9Qv*9V%%Yb*wK@kn0snyu#8vcB1dAZuPp#iSYZ@si{d+3tvp^
zqo1F!umOp^f}4gpHJleefvnP)?M5>=wvh3Jd`wJI=VFSFL9?rQ?NLmPB9Eyq-vgve
zP2jC%WbmAGS@89W_yT~;aMbwQo<z002)z~ezYkX9RD6F`gMXdK-?=r-{l+ux2#@u7
z?SI+FG0EyfB3sHQTk;Xq%OpB2XB-p2$8Vff>LqYHBb_(NDAs~8!txwA5_7cG@Ekoa
zvTFY_9Qni}7c+7>y#>`(A0Mhm&8fukO-KEpJ-SR$%b9xUI33`B{N_ZInHv|p6hZR&
zZo7Sq=R!G8tkeB$Y|?rp(rJ2ndS*sRRaIF<CC=oK;ERV;fzLBK9O_A<3Wt+e$8cS8
zQ_jErIrL5P^bUQha6g3uuDj|NW#e|sP6x6te^(~@|Dj%p-e!wAEeD&;M_UdA{-Vmv
z=<(C^miQ{B!7c|_>c6i<EL1J^f#{0(`e*%|+NWIv9hi<m>Z!6<%iI{KD(AHibnIsN
z-Lba~JON2;Os9DrB@2G9q_Yirb>Tq?*=dPDN<i|Kx+W;4gi0M1;}+Su)gc+gdOVP@
z7PY%{j-k}D-dyahr4bek)$y&Y8vJSEH3Rs$u0j7zp%@tXW3#nQn5fV5i-rw+!tA>T
zvGtOSh(?HMc$HAD)(S!J5k+NaCZ!}K5hGhB`@;oOKL8(8cx)Y*saIuUa4+<;j%#*)
z2_4Ui$0*B#SGi{yxPtHA|1@~bMARN&th5P=y)|J#U?Js8B~HQ4HO%iese@|&Ii<lF
z1Z%eDBmpp3@G=BGks}<w!4_q*a<R>!p13LX_kR<-^4>l28U=4JQy>|y2{gV8Xh1U_
zyjSaUF*!{gc+d1*m}NU}e={Y3g}<IV7k%!65n4Ln;;s4BPL^3&UH+Q0bv_h_4<(Fn
z(`z;Dv@9>`b1tc)x`w<ilF9)NG_Zaju4rV28-Z^A*C{Kli4N(p5?jItl*&g97!wb$
znT0cq-pzL`4<fQ3T-2Eo;<9at-=(3p_oJ)$*U@8S7^pKJClMG?BAao2XkJTr!XBe-
za2+wonTEc#-AwY}P1mf-4fjLS{CX!_6?;{cs{E|$yO5uQzb6JlbdKmF*|$lfsCaK2
zQGk|@zaepvYTu49#HQPzo2o?aJkbn?I2SdOcg+eZ{{_JbU$pN0@O@|p$3xHb5?%ix
zeEsR6@k4@u*@;RT72ccbfs42xZWH6Ti&J%`0OPnsourdgbC)Zj@`XZO7S5KX;&Fy{
zn;=4zSC%d!ooonMEtn$+?TgRXuVUuToXx-<xRrOwpu^O<*j-jaMkgsU-g(q8T!7a@
zlG9e)5ecqJz?M-!KzM0h4>`tnH&GX_VGs5w#Z)>FdZFjomw^rzXN=<$n$<~KZzeZ$
zL35tPhs#URsA+hm!E3ih?uBF=)4!CGy~Ku1x^5c#|Ga3pl5XF>sf<J|Sm0{^tru{|
zf)l1wCsxg9Q_LRxi?`?=Ek^$-?5lK?$EmHq5oU63%-~MhrUucreSp%;j#r_^`_vIB
z#wSaE<yUF5(~u&2V0YlA)ubOb;qhJfP2oIM{X^m3Jo+xFpwP7BpR<rq#WxlZ=Bp)x
zP)h0-O6jX^3dbqIPDU0Xwxzxmgd~9TyYu=B;lM|VltY8Pg4TnowV%KxBXvV=D^AMT
z<b-cJKIy5IO*x$ekA7TiN?G7Ij!U9TQH%{y(kewS^78J*)`J5;K5=I!7f|bnam~ug
z&eujHUPTSRavTAomI{5Z4~}TxnE__8gP8NizLI`4b2JjMB{rVg=B-DJGk_^&q2UIb
zR=P+^v{73>0ltTPDT&x#j94upG5(z`Z%K^F^mxhXN7!PE{v!wFJ+GLt>?!)e#>!+8
z#<WoqM$~AdiTQ{mY4eXxf~rW@5t_CZ-&Zqq5~=Ek+_3)ZfD!lg0!u2CTAf^Xyd*6;
zC!>$CYYxkL+<fkj{nHZkNI<POS=FtZY=V5*+nPY=T~^drwb#^|h73T<$Lpi`vcj!c
zOxOKXk^N$HEL~Ww3S&=PNG`(%8(vxq=8|Hk$LD$7yKM2EOrJ0&LVUtOH_3f!IDw4P
zh>gPjuJewApJ%<=OiSsB2}?|j-J?D2fgwZ_D{n+-pS>+Ls;Dh6<AFg6`KvOWN185$
zn<foB(F1+9gW;53&$S#sAxu+7lY@^{(T)C8nZsn{?e5Z2P*RMsNeWw=l)J-1G&U|0
z#PVEb!M-zCZ9yJ<TAwZHq`9jMZw@L2+9988jp2VXZ}^;lxkV+oSM<w1^woSHva=uN
zVfW7?b0jkLXyHySGy<N4H)y!_@9^y<L6Lk_iy%To?==h=4b`?m5IFXxaqd$e!Tzwm
zH?QMb6LJty#lUIq-!Bx)wNC4$B8Z6nK8AH62@-mH^({LR{a6bhG>ciY+Z?;z$wXK5
zGn^k^-n|jaU1i16-F&+hPhxCwtNA3(x?^J#z15TBzC;ka5a7d5Sa+zqy%yPh?maWP
z9|_t*R-A@6;YHz%jo*{YxqCZAEi<9ocZ)DFDvT|pP&n3|C)u1j=+SE_Elb0Y9&{m|
zyZduYPkftQ_PA)m(EY*>H$^+*&srV{!G^h1E5T^oN)e&0Hm%I%nPfDdvkI2&<?X@I
zZzq;<9c%leJ^*fUJ#cfy%8p?zIrygqo$~vHIvH*-slVl@iMfXxk;Jn)Eh)sGc%=AO
z(ZZIP^z$lxGt)=<G<;hz7#KhM`m#LE6ZPgd--8oJ7hRpOtl)4X0VW*Sqo(fVu5AI-
zNdr^~o7m0WC?CmSju-rUgCqj%=c&EYn#Kr92wDQw2Vi*{E?!9nMuQz5%|;?>{wnBH
zR(d3CcZyX5o;Q-nx{6LN+V^(1@*P}Ee+L&2!?;--rDGJ?4)5<sIb3{*(&XbZpE1id
z?;Y%sAKY+BXd{bE-1<0p$&1|ll!M6Z1=kF+-uT%!J7?7sRYPwf+xehxazPm+&jZz)
zzD0YhK#`f(tiTtJuKL1~I^gu>pk^Zdc@fi(<Tm9xhCaf{j>aJ1D*<ODMZupui$L|Y
zE$joiig)&Run=cnf|p7s_A$U!5&o(50vXJ0ZV~%_%p9urI%HY5NIp+L4w(05HT9se
z_VHt<XDHXQIHrV~H63Ms!qt$UYN%YEmAB^uGf37tMJy{VPhbDy@StwuWwRmjF12}R
zW@o_+2v_~rR*BoR5L-gB)w^HJXNR;M`m5qLy=-58$7I^PA3K=7EzRu5i*+y*K%YpZ
zg+Eg%55sPq&7X-l&i&Ql+;^kwIG3>&VGhMW(aB+`nU;ce^&H;Tr>?qQFybA|=N<>-
zHM5Y-|HCv%P;rzc=B~c+if5nnc!oZl*pv=IK|($clcg&r6)^Exe#g57`I<^mn`i@~
z3qq!OOL-|s3b}FL4HgWt$S&+;wy+_vPc~xYaRJV<#4}BucnNws!K4qd_2QtdLRKR0
zni)n|x_$y_k+FXaQ9Wpkr+R|EHN?QptypjG2T6V&^^{$6%Ag03drgOFai9+F7$@a@
z^V_^(@c643EEAdgGgR|Y-TO&OIjPa&k>iZaIBs^A^t(xmv`ZPFTK`z#^?V4XG5^i|
zGn(80y_t@kLwIg>MM*zB2aEXBcW~i(&4~+^PMO4<e@32AiN<Nq(ASr|1P$mFLKq_E
za6ONamihz@U65Fp6QV0SOYyRHCg;k9`8KIC)m={UO`0$bI-S3HoOOstQ)hKXLaFol
z(yzI!<co;Vg;{RX>Nvx}q@@uP^5aADfZ@yH#d8>=F@0<DW+6+eg<!}}i?AE(@T9Ng
zJ@9O>LJYBCyMB)Q(Rn1BCelEgp5E+=NEIcU(AunuZh8jV3_r_(O*ID-1Nm5=xMGVd
zbe{6v;@TiM<1DKCZWh=5+1P~;Gm>MdR2a2`K>;zemXc=fF7kp^+1>yum8xGb?`Vjx
z8Rm)iaIdWXG#{y%<SXyBA@$Y*_@sDK%AL3}{m80};S_UkPzM+J@);n#LVv%d^6#ZQ
zi^|Bs`|;|I&QRoRG_dixYKaB7PeaExqmfMprZVaUbeIOxMqy3Mk`=;*IeE)nH!a|r
zQGE>-bgpaSoZ(|+A@3dRY0t7=-ETnF+Yjti?F081$V=EvS`1UAe#6%5BeG?b_IvHV
zm^q?=zZLhmREZO?2mbun#Y|KcvErQMgttk1r<Oh036=CewTGje|7>@g{=ije)V9C+
zeyXpF%CM+g<q<T@{KDZQrCAn&Bvy3tfujT8{<HY7w8pWIBy%gXUz~0fc|EUvm9)FO
zREPVc|IJZFtK-YoFE*_Ks$Tku$8Ekf9fWh=)nwG5iz~xt0I@kVL{su2D6MZN1z$a`
zigcrvR2XgUfC_HCjzg$2q&{dkSzG~c%WqDBKv%2=bhyZ877Zg0LHB@FdUmm~&n+ln
zWcWAP7?>R4LzvK~^TW5r{EC~gPf5ULk04d++};lv8%7wXl=aA_><lzM-rVdzXwLC2
zu6o%^X;7GeGkwy>P?|q@zbHTp>muwIew{v~_GKgTNWihvRmAWSh3vcP8*<+sCtP?P
z7G8&-)X>#jMM#%sbvO@-FgJZf<jO;{0`wI!yH>fNqoLW`JkzY2n1VCGxctl)(1Y(Z
z1LP@~%JkEGBWg_4_Cf&ygLoeVDKTfg%c`&-j}c&C?)Hb4t?E-}M}Amb?Cr#CitFzz
z_~&MjN4J-C6*WjPPDN)Fn^naa9>&R?xhy-$^wFnl8wPtNsXc;P4>q!anf3a4vn`=s
z8i*yto3tu>w_DXs*?*PJZ79=x)g+4)sdkn8w7PZs@LKS%F?-T5rJh`$WR4VE;YoP5
zV+=UKjT||%F-2!Gvq5)_83`fNI1oLI?ehFo_aH1Ejq%$y^;5Be*!$fQC-p}bS)r(Y
zM?=^ab{!t3d9+*aeiq@M(R-$zdrBU_SK^Cs25Uuh_tm8C^L?j*Y!>gH)`nDK5>=yA
zN^Q{!3t!LgQNek?Qvv+tw6MbNm-3PG-vR9!691U~yOIweQYmwdyIevzZAJBe-yPuV
zu#slk?w4?H&|_ZBb_MC`!UAsp^Uks6RM-%vK8jEEi3_e`h;FDPecCp%E2gW{{pCW#
zw459~`nauQZWHbg5l_fX?d>@VeGghs$IjTBe)nwiFIZ6Jt9YtHnXLH6>wk-Vfp=rw
zxtZ}dF?E;8C3qs6`9;9*Aq>|n1U>`FFjOd@4$JZ6vSupYUe9q;bMu=-2Hj`g&Fi-y
zy89EF-Gw_7J^H#j4vwrnx;%r%e^RFXFfFM^LwH;0uV*5$HzjnulxW#Cju-x!nZF^e
zAw^F9@F~wkj>JX^^fO;b1kz7vfk;Hhn$z5sMn`SK_MmB8Bsc{lx`93#)N*?fM5*t!
zF#mQYBBF+9n3FB$asEg)yU)ZiZo7s;e!qi7rP(c~*Svz{OiksT9Y8>?{macPqxek|
zPhjM0$%!lW88o}`W_$tla`jXppTB8vP>j5itjyK5r;R2kY;L<yOweium{ZerQaN8Q
zJD8&i@?8F0jHf9?<l={#iD5~w>?Q1-6)bHkWudf*=|6iqN86vmwa^|NoEB18{(%Nr
zCrYcj9O58d8l4&xl`SDHE3=ctrt-%J*`Mcq+xz*u2L?QH4R~*@6mt1T_okdE73+am
zGpyJ5yMpK^L`f>$l4H;SSd6&5<Gitef<~ulDl%L2MWLskA+OkTKY+`~1xer{5qoF+
z;o`N4NrCvo?;Qm-E|!FYx9&mW=c`rN-G=d=r*u`l`KgH8?Q@h9Qc&zi5Min8CALKS
zDNCjr?!i(^ARLYp^2=hEa%OT6GK0oD#>N8rf1V=_lEs65SfJTpUWBp35`w&ekYI%(
zemOW!)ZRCHC0k0Ka(GPx^>luE{{ik<oG8*O-8nzktEdQO{z;tF18Nchwe7I8uNBXB
zKb$o#;iH~}-B9?g6Q=-ksIHif+(*0K&)Ldqdt+R&77p~s)!V-I*){gg#ch8Z8AT`_
zf}0Wtw1-qfLI5x!ZskD~Q~A^s$3?^_X-G9X%UhVH!zg(NvHwLgJ8?S;o+8PwjxUd2
zg>HiUDgD@k5dG1jCw;yY=XMTWXR2;XknrlWy?X)=LcZ+(ktEb$;jDQ=ML!LKq2({+
zwpWzcA=CDS@ct`i=jIpG%D}&vnvhl{ut~AizE9zHU;sAxSv^d{wJ6c>S41L4JQp{j
zO@;^eG+kVFEQLh4grA4cY`*CrwJn!yej5vMMP78FMo13>s|r{-&kr-OZmUz%Qo$F#
z->}HU9M6@{#PPv`1as#souBO~Bml1=?IG%tN{1bsbd7L|>j&1ZCt<EB4Na($ND}_O
zwB}DDvw)j*gq<WIUgaj*013A^C&!-*)^@K!8g@3JcLRip<0bZM?Ma!O@0O>&qPp$D
z_UJ!sx4h-QvFR-^lqq-l+C--1PZ2p?v2Hx-yM_Io9>?Y;r3JO2?t=>;Z0uJq-46fM
zD7aFHEFCFN^Y$SQMi$D^Mq4Ool7r-6;Gs?D7$Es%VYH_3wb7L=YcN%~(VpU9&3m_f
zCoxs<bZRF|yQbS@v*<l>GcQh@6b~cRsBA5C<NB?V3vzfIEDeZ}yN?Pma<vuL<`LU(
zP{8o$P0)K2`z&78C!D)Uf2sBCGamU@aE%lZ<p>e6@?>EwbAqb!459neBWcpy@$Lx%
z11+&_S!rN$ZF1a^;E8R7Z(0lSvw}8$SJ{9mfixR|z~i9ByDH$IAQdp+Wu`&MqA^fA
zK=j^he`;W9ful;<`hoYEjUjn6%5?#!@Eg20sMFr-cuH8=P3KEgoXJ)wNe)PKE!uy)
zBkoVa2>%+V8ybtOX67OAlz;1IFfz~z7#~ElqU>w(+g$NhHb0FDzA(pR{dCK1dA93~
zq>1}sAYs2~Ec;rqF8*z_zo+XvLC<hbrySkKCZ@HS^e38s9w;+bwEYEuB|1Dl;svhb
z<g=C1BEa37W1u8+SE}iu(f66U<l+d~;Phd(D86yzH0U7K-Y`@tOh0qAi5YhVa2hCW
zAZ*OFD$gz5XX^i7n4lNYC8fy8lj*$Ef3DbYQ-g#yM?mMw#Ft2(^Y2oZ#ABNh1+64f
z+2QI#f6C7t0yk+jBixZ6Xy}b<rZm0H87v!;M2l@p2zy#$YdA64b&ft5SSjk^f0I>7
zJ12T?${$CzGqt)_lZ(`Wel0wZr?_>Ct~_W#fW6Ji2dOepJlnJPQXtHGJ95yyTwc*e
ze1n=GUmZDAz!*B9A)_%t;(v&HtF|~AE?66v;1Dc$&_J-@!6A6?;10pvoyOhW-6goY
zySux?;BMbM?|1Ek{R{RS%;{KnPp?%~cM)<|j$x1!$cgR-+9|TwXBktA!L?))>WtjD
zFSwcQeD4dE(wAR^Mf?=gIv-*6OqF==(QVQ&t*}Kga3vq}Mtd8QcHUnrmSFe`;KI$*
zh`dNgR|M(qL)!+mX#MNhuDl_8hZ?a29<ROX(5w)Q+g-1=4qz=|V=^}Nc(k)Ce7Rxj
zz@p=f*u>^-$gp`Psra6!Kdqh|Vht_)Bgo8to0TGePd|UH%|=0ft#l=@BAP-;xAB&Y
zR|<D?C|pw9IjWuYV(^(;5Je4mWM3J4nvtNyD5O-yrKlniH%oWEbyr%sNs;TA{2hKP
z9O(F49HvhT%MO=$)=I<XB3z*Cm*n6W{a4D$Qdo%uUwsxyGA!2Et0x{dm}8wEad~f=
z@{=ZmJ6L+z393!Pq(Bci1CdjOJvsS=50XH+7l^E1qd00!0qOO(M>hf?4zogs(f-QE
zM&hxwv=!O^%t1CE%VljXe6J_GqW<!+af#@AdIlbUvmo9u(LFIfmA0!JO+ZQ*@@92z
zm2oyft|K3d=j2xs1-G`v(=gKU_V>6Ot3slfA#!GTr|z~GHT;w9iM1=^b|=ln{mM;t
z2^sTqqtnz&-4P9K-5C#v!Zk5xx6E7V?xAmb*}|u5+{)j%KgNHRvoZ2Ozbn6v?ajF$
zZCoZF8Ex>hI}L<o8Q2Y(3odcK7Wi_>a=^j!nl~%8ff5lOX`o6iZVlf!tIgxkmS5Hq
zl$HnqQfJ*f+$MRE)oj6fZE1U8K^1+4l%-4+y!3p~ua`r}P!|V6+hq5LObzIYUIeR;
zDI=QbD;{b|23yib>^FY#yL-HpPBRCrpUs7J{(x5;&4`Xs$Kt60PYORb^D+>JX1Ksp
zi#oi!GR)?*i?73IOTeU&%_QMB-l8KuHcRL*u*(ml@}`t~V1Sycn%lU%wVLO}@y`%B
zE}by_la|GwQyicyJhgFN?sr&BgofNze8Flh2R73so(nj*Q&fP1$wlTE%<|>Bl`91w
zrwF(uu%b)~jX|2}JjK&)*#;r;J(jXJ6$A_?MmWr27Yw#{nTg|*8C~<Y`aFNIC+k;T
zU2Gu5Fm7NRC#<t@){PHF#RaROKZ1}v;GNg47;r>?N&*VGDCRhMB2iACzd3ypCz@k#
zf%3^n#v7btWe>_Cu&VqfF37g$N~C1!lkr!oicX?9Uui=7&Lu{D`5}dr*@tx!tTexo
zWs*EjNQebkBG9)v`y9sW?<YkoE`l8*>-%~(zujzwCqFw7yYkPeC!`T>A0F%K*_?U9
zF!na|>i`6Pb#!ZAgPr>OIewP4JV1Z(?d^(s=qMYLia@%vhwl(Bgm>%HtJOO%4S0X?
zA4X?;VQ?)Tx<rUAPMC8<9v|!CrEe->TY@?{Jj_mP%0}woU~r42R81kzE}g%t;O(6$
zeYW{hd{MMrqfe%(d}~+5>HJ49JS@_r`lwL3M+>k(cZOq#X&(`rOmDPnJH&<LdtAuj
zs>^}31akt6EQW*+%cy>yy!-;MUT}3H4d2**;&@$o!q<matV8eEpFSTTh?h;v1ZSb$
zwHpJ+Yw{wtAR-F9n~9z<2_b!{LGl8pgPPr5xu&m(*k3T-p_(^0RnwoZkF_)u*U2S|
z0|8Aay-&9jw!Y4w;HwCD4E*`GEdE^^TpP73OZ}x;_y5Hum_T9oXjolyub9{1yw~@$
zi?4>`1DS|IizbDYIhWG@m1*M&Ii(*yhvU57F7{cJ*sQ2tul}>>sRTl~!5ttIXT$-|
zX+AC}7`>qE^o1vahpeKu{_$?ajG1L%qeV0Lz8OJ{3Vp>GQB{Yu9WRO|AGbAhI?VI^
z6+>*|)5QvSpNE<l*4*l!ihJ7NZ4PHv>sxvLJ(krCCoWcHeun=}c8D0BZQvy>R59;$
z&02F#WN3#-3X*9ZfqMHoNH#^a^{L}Gohfk-(nn;J<J?|*qzibqt9iGja1sOBwmML(
zpn>_8h+oClHHxZW3B!@2^QaMazy9JU{D2YZ68wxm>K(^tiXdLQj@-(U@Ep%XD1(q`
z|1y(|dYc}T9)2wG#nUi_JSZN$HKhkZm(#DW$F?Y~()o4V%>k4TWrVzz9?OYUwX@8e
zZ?gU_J}`v^$IL#N9E_)?I}kvjls8Ks)+cRSFFdFe3Nb#<zWd#0w-Y4A9Q0yX|NGD@
zQ2t8IR0j(2MUB-cJ8f_Q4vxxCBZs^DxT?LLA8X(7W=tZG1!AXiKL=BDaM=ogZBKqN
zW)e(^E>>(bu$b^%#(Y@G6m%Ox8J%=+gQy^^L*oJs7z;)(0{p}(xjsN@@v-KP{y{9-
zmb?}3I(xDU=S@Jx+(24Q)Z)^!Sh=9-s7uDzGG~9S!sd3(5MK5p32zx)1`U&t`*Cu!
zhLy9BmYQrxMz#a|HG%>+n^NKv5!=dtgo1f3WX6_4$|Cf8lYVXiES!!lv2`>+69#|)
zUr(+MwMka0{`(kD9AF@WhKhfA7Rtt@>-*NlKwOw8*8|#Gsh2(X&nc*gdlCM?m9PA9
zEot86$ieqAiH5~ln#UQvq=rwdlK6Ss&^TXBvk8A9_{jCXScq3qyj5Xliue7gckW1+
zW9<{HM1AAUf6R-^aoZo$(chb_g`+ur7*98#v;9R<q&3|zfHQ@M9=*%K<CqPnA@2+v
z6b5y<-$NHYRS{C#FZ{td`G;Imhl7_1BE@EPm`EuR;Ip{tt`E{y&`SAnl95n7B*<01
z24UD6%liD}j)IBxSxJ53RnE$&uCbtyf*N*!=yh8uKM}&wpJS`Fys%^5G7Gml+Nr?$
z`sRCKpjvDa@cDjkf-Me<Rw37e@ORx-nudvl;8LH_9++SAMuE_J+6(rBj^(B{Ojs=y
zWTrlGGEYs*#bm|!6RUdCQ-<^pW{wdc))vQ3h=^TW`q#Bx4@~47GX|Gq+BLN!x;r2_
zBCFva(==MDb#4Kfa>G-05;R$?^^RyR#P}ayXp?h*yD1EtqmL)A;=B&l7soLc>m>6-
z*-{X;Ce$V3hPYEGO?s0q@WJ(28U}x0E-9t@9rYL4CzU^p=?bfKUBf*BQXP)7zAx<l
z7))uw<<F*yr>jw5dWb@{nFHfOg$F-3$HJ^;LRKa^@z$|R|7BBXT^Hu_JDLK!gd?>1
z<l#49Swtz~T+RV1(`~LW63wBJeJXxGNR@nNJZ!3wtP9B*;f&kYWOx%Ca7|ofZP?B)
zKj3?_2O|h*(MFDh_WV-AnNBfv(Jvb!M1Jpw?-TJy>&4Hp3KxLw)WkNMe~bFy-MKXR
z?7`wI{zH)tQxB^2<}i&Zb3GW|^nqtHjSVQV)!jvg3kgzlq7fgL(@lsY1mq)$QtL7c
zN=t|e)!`A<Uf^*hjWOHfe)PF!s(VE7-R$pb98iB2Ne;#hqfcc@^Vt#|<$k|{El2=o
z1htYK2!sS)<FYBtE#01{_%PU0DR-*<MMoH1ho<hmXcW|a@COzjgfk)xn@eo<q5#cu
z<jA408U>;DQ<N}>x*i8tQ$cL%N<)iNjL%srn)0jux)rnH{?1NLIP7S#Jy{#QFc!Z!
zqrJzUKS*_J<?{9vr|P}XQ&v~zxxF#g$Fc=o1Q$-xqfVbsV$c4D-BVK+%h#SyGda0n
z#x1V+VDx!WoJqaex7AXi$#m(hJpf@<jEK7^8$+BA#BBC%UvyL<nPs=8gt=x@du|_@
z_&Ln6EGd~$PxxhJo=Gg&cF^)ShZ%ATjU73Tl*m3;^qL4={phQGlP1kVYjVDceZTS}
z1>LUV06fgr@gAWq3ut-Qs|X`a7eTEXO5K7@QMCd8YVhceeQo=i4p;-TivZ)lM$i*F
zcL)*;$6W?!q|t;RlZc20m3^v!=cZ*anJ1K!G|~IRS9eddW^l|Jo+6=V&<Nt|NYOKC
zue3jZ)6pI4Ls*ppL-e+x7lrUI$#%bhFr7aNN)Iy?gBYg!qy@q-+$=Y_d=dRjS@3M-
z4K0Kei_0zXGs5w&p~e<%c>vOiEk!#?=WnzIVFBAY7Flk^V2|a7Y$xn0C3B|lbd-OX
zb}`T~R`Z+&dX}kw<ki4au-jmJU@o#BTlmpb%K2Z<D5$-zF99reAkP)Bko8RokM7~6
z8I0?>6EVeWUkdS4c>TS1iZgIqz4PEJ({jMkE`(`k=JpIHr@(RN1=u0a{wQUXj6l8h
zb<JpbCDxM>mCP;pOpv#y^3&jyNO8%>IQ!9erYt_$SXT4j5f}pU!%9n_^;1sxg6<Xr
zoDr(F9&?GOWa$?-oJQgf@sY|45<J2$Ujmm=1ShG<-=DBlTBQ!_a4O7ivm$)OKzrCx
zxcH9aoD8j%k3j<mzQ8{?3LR>#R&Q%mqUU;=l%i}Sp_i&u>N|Qu9_ZZJPy(Nep_`E5
z7$2)(d)i?D3m;+!_4+5*5d3ek2G?)<u!38YD}15!Ctqz+rEj;EKZCCe2|Me8uZtI4
zgo0;A#Wp=7{hj8Me3W2PhTDLg10}JUHS%@^t_0el;a4Lv0^r7eFdM`~O+LR+V4}Vq
z;j8jK$5wgSD7}O_d^T8VaE`48{;#}0a8svs4_H1HUwFN<MwO;i7FPay*}*K>oB&>x
zftg2M=y*?5m8F(0E-run{&jG0u*fbeAU0k3_2~0+;wb+M#iHK2ph%7FH%foKt#i`y
z7Qh51G#l{s0H_0w&eZ;Y#>c5J7mwiCKt=p&3Ky)C#Bhm8<`he_$9i|~p@;|E-aauA
zKJgdAql;5Z;rW}g<f@`fnMLVILM822>E_MV#wV1%C(q}fD+;k+-k0`Ans-a9L}aDw
zo*4oACCf%9^1mWR`un~&=)AS3GMOEW`WSaTOLzdn^U%=crJOU%SEW#|ZTrQI9svy{
zW&f-={1enl9m#qZS|4aPj?BKK+pQlFvO1)3e|K)c+lb!|bGUsJA)9=g$dEEZJ2mjE
zGDj)e)%HGnLK15I^_Fig0RvX)<=Q4)<Q&DCg(Vd9eZ3RE#Ht;4zk-6^Cv1W`ijjz(
zQ<{NE9kC5;ssNXj4E}#q)ZbwCAwRC=PzU``jep$QP}QpJH@)u@SjlStqBg|&!|sGE
z)AI)>{l(|Kp`}JsJmB1O5Ox11`I70|^uR?317m7V*9%4by%0NhpFZb`awg(&|JVef
zkL#TdiT|#vPzVh^r$5C#iT|x@Lu0!~)*uC7yqC1tco(^I4+0y}%|dHpzhV&Gio9V3
z?NP!m<C>E_poL2cJf*i=>#OmL*4<v+Ean8RadvX6Oq33vV;pQqv|8@!WR7n!&-Lv^
zW2^nShKNtSx{dM=pQ<0vcr##`joZ!Zy<8G5mpwz$v6)XfHvM*iy`5i_^Xu1CSrP+q
z#(C(8Ir^NPzr^~NxzBb&HB*D;5IHj~$v-gw%}nehMXPmx8(P9hlJ-=i<yowVzb7Jw
zFL@!s5Ipj`wr*50E%3V8WZ{XQl49KI*CBm$6~ibY6TZck1rFN6MC!nwZiw==n7Uc;
zi|UEBy)y!Kj}aJ{W__XGi%OiI3RM6(3nLL}11V#Y&-XsH4|=L2bAtbVlc2#%xy(x-
zR?k}ZvRNh0IDdXj>!=)oN4QPsm4>bKy;#nKpG<FpIE9jfXzHq@E;#%g!|U?8M!Cw3
z=9jy%rwsN(*re{FuthUo`-Igmrx=)J$NTW`|N2k^oT|gs#pj2<(bQXcY1shTKY4FA
z9C?-esRIqF0}eRShL4h6w+_0F!|pSyLjC?2?FK8GS{a#HxIn~LRcRfbS^-alF1meB
zswNVNHP*sRHYayPud%d212#Y0bOOKJV39?We4f$w_ztvW3;i53+06`2_K#1G4Gj#?
zQP{rxW-yv&wZW;WhfAZ6kw5?tH)q8_)U;?UGNkjChC*L_ASelbe(VxK=7CGmsfA<M
z#`OCd85$U(2x0o=Si5ceofXU&fK;-r;`$&abmfGmrx%Yvblv-u<;?b{LlNTxeb~fd
z{i*XGv7pI4*Y`_AALmMf_x^VL*Vh=|P$%wQ85}1WFt-rCn+<qNERqGp_vo5n*Jo#E
zS63GoH>2P|g8}e74)u+!)4Zi^Kokv0YUc;Uu!+akAnv!z;J;8mEV?4v&VPR=<bi5^
zpe-hz=Q1|4on7ZRQ*siy*&?;j>QX=rY27QN9^?z6B{6av>`uYm)?1e8*bfTsQ$!+S
zZNpAh;0{#Og!Hi<3VQ&YTT1~|dG+|i$v6~@KWb`A&5tl5{S*(xQ#_UyuDy3;+8Omk
z@z%WSS{X!^CdS)f8&zkJ73w0xN!A=^m5OF^5_G{0kloUz3g<Bh#UD^HMo)988iiTW
z7N<a3(WB-57kfT3bB?9i;XJT5Nk>)4NZyj0Mn~}N(>qi<eLYaX_3a8nP?jg%f`K3d
zt_0=t>V{yu<*dTJB?6pmyS<#*8%*en*k_mIW%c9k4l)j`yDStQp9GwHBg1{PVzVQ>
zWDRcjXPz*hv!(>_ylb=HnFo%^6A!2H0B#=>D0D1-ibtuqTSf*#vmYec1y4FK@15^}
z-g&B{Tu)f65x|YaL8Fr}S1)s%)#)+5@dWa8SvC|H37Xy(j2?1!tZQ8OOtx>?n6@|;
z;Sg0q2U@2e8XtjRrj}Z6B*Uw(O0AX>cO%zkxF1dGeokL1YNVem1|^~I{C%CZ2sI%)
z|EPH&5>PT)UYMP={DmH!?iXu6K}#dyE6EB0Mf+7;4uByi4fe(S4D%;C+uPflG&Y8i
z@-W?A6kTNX>GWgb3Dmw+aXx$8Wq9(hRK;4mJ*8x$+<vn0fzQ1%Al*47!+oyK0S!Mb
ztESUq{lSaEZWLDq0{c0~mPvYfeM3`6>HS2+Le)E|)~{e$o|hhbs&4zQaUr|{Il<I^
zOedRLJn&dP8B79}4z608gkGJUF_cqhb<e*XE;)BUJ=1q~MGqArGTMjI60E$4cB6vi
zbYzVQ*06jN`HaM(mz+m~uk0<23-i9N)__TPt6Sda*85KsmAmvU--9u;ZZ8L#4N{BE
zj-%M6IV=i{c3|#_zrZ7hkI`FB`q_?LF2$ei0I6_IFucq+MrJTSt#J>16JQ^Z77E=~
z%-lcc|9a$9!#ghiHZ0z8&iSUI7vOTH{(2Go0iqYQXoai2W*}`BLsG8v@<Hp(+sW!2
z8)5XBH6)Ld+k{PjK6#C&;k2|95EhXqUURU})qsz$0*p7|;N3!?7qsPJwJvOCtOOcw
z0iU4uD}c)&1VnlMM+DT2b(wjC@(I4fu`;dYrR#wI2(1UQD75Dcx69#mZMZkJ^)IDJ
zmp48y_QLdGRukww+J!#;@j?f!zs`JX_Y_r|43W~Z3i<ek^jYnC^!r|6ndpRvGk>y6
zmzzf$gG89mOZLeR%&)#jB0*+e<sbVbKsFINa5;9K+>>eeX8{FMCmP*j9aK!Kpkn*>
zsK9Cii=!mRYvQo>yk^*_;d~(maly{H0%LjFmE~du-jJ1o&Di%j9I6t361I0&CPu`i
zL^={oSPpjIIg%?qAd#=60hBB8<qrI&q^nO-v?hdn`$v##qo2TNO!j)mJCrCO_wE#G
zt<$PRn0CoOS|dpYpC(KgiH`PDO@)T8&`Ym!-v{LprK3fgLD-8jMc82Kbrx*%GNE!J
zE-r*u3comy^27V>vXoULlu$0_jGa_zT5C`dj*3IbX1F6q;$cTHtp1dMj4Av)`;S<9
zprm8`VT)yl8GGD1_~Sjm#-bH4LXvaBh+Qcu2d>-1<Q&vZS@1H5voe9OI7%<HfkA?@
zQ2SZp7cJh3PO;>WBa$os52z!htj%Wsw?4pk^D%AnD&hmNtmkF4#qpHg0)~aXk3!>^
z)89wD;0|^%%}Y>kI)u^3SjT*FYLBNS6p@%3Zq+=LB`%Za(vjD>M-Ct=@9`B~*P#R5
z5IwfVdxnwmJsf9@QlkxpAllT-yy>X%vU}0QH6?@-o=FBn40Z_1ZIOR_a1?VUF^c@m
zVC1vm0dP-Ad0>1zg4npRq+R)*=eLOzC-zDA2~!t7o=tNrC}3NZf!fHyd=|Vi*RNkO
z6_^Z!2fO;{JEVFqn}GHGv<@G#sNL9Eor>Ssz8Ou`G2QPQg(DvMvCYn-`|14{mi-AO
zJ-|_-MhFoZ-*+ssca%aumW4;!g^$t#kc<SGmDBUbt~_2@Tg04=KznEiwt1@+YW|nr
zOvCuOhg&rL!lImGH^3U##%TO^WJJ@2^LcqQh!{~OHKMqF7~q$FwI-q6pa))Am$egd
z5HqPdDX}8oD)Q8;G$g>IDi}r7Xn%bvbuNgRX3x;*D?~C#)jq~RQ})l??9_ijD4S({
zeR>6feX`qp&XDvOZL?kCy%hG01S;@xz?5@695V|SnCu@BkcU&EeW$wGTJlpw;{cX9
zT?6{|?#)gB{ERAFhx-;>;w)vk`Jew|6`=-(Z}I6$2}>K23AwvOUqqBS9Xy-WKv~2S
zyu;1=T|j`;aE5ueP6=7zym#o_UYsN@U%OLjm4{%7X!5dijdb-6_pmVIpjL~Dh`NhK
zJnYn+_^$1(tt>Cjj)i6TTq41K&Wi8!;s6f51Mrd>S6`BYW7pJ_T=EWQn{@#Z`N&0h
zCGLe4+%a(laCbs-5Odz^s;9?hVzZxYnIu%il^tdT$zskT(Gs)@YPeY-*O><C(dwv2
zXJF!DoSoYCZ`XzVOZ7#dhW;u4)aY!|*<~F_-+eS+bo{S0y$N0$SB6N$D*_7Gg7nKE
z21K-=OpkpObAsdBazg`in$HW;-t}xvmt944tvV`Niy9j%0bLuDrpES?!Zs!jUM{LV
z(D>**ju=|;c_PNPQ31J5#swiMafjlVUI5CVfXFT??U(Rh&m?cqKCf^Vh+PJ<lLO@6
z>lH<$CYO3ZJ@p!pU~ZVWk>dOu7bmk1a6A=x+raTBVb3tXPi*7iuE}HJI=SP3SPjDp
z$1E1e7kr_dYOZbS0b62F#xi^a!<ih=h2PQOm~9k$kJQnxU7>NtB+XXb#J*4bbcYq8
zGD`gOXP&LW>Gl>44Fv@i75Vnz&Sy!dfIlh%`QE8j*V3rC1OfK(_?!4Us?_5LK#;{N
zr+3dakcJdT?2^tVxQ<{l%>LHRkKyEoCt=wjKLf#{&7|YdGcY_)N6PvWQ40y0A%{gV
zTZEg3b7~r4&p(qFvHA4agBX=)NJ<y}^gTp$9oIXqkZ-zK+S^sP73Q}G(ze%a33=DT
zb(tGNQ=<x|ouv9+;vs~XwwlxhQ2b;%XwLVL7iT}3rq12s1|RJXgXrmiH+<A0h#hzH
z^<5&kv*?CP%7BOsR(`4u^2=74{Rtq<bBh5B=9F*{{T6IZMQc;^c*{blFlLf$szW9e
zngNG?3zZMGwmjrT1u0cMK_n?%d$r=Xa_<D;Dr90C2ah>#BX&I==@wl;v*j-TOwkMG
zpI3L(tXAxp>aU7uqT#cdLTq9g?Xhb*O8@#x*5bPBNg4QPWL<~mX5MoLGp33d!KPW}
z<1@ma9DH*@e;wH<xfrAS1_p`vgg`MEP+A0%KVrvERrbdWg2k{_VUe$npyvg#=im%)
zNHC*bu`0pD=j^S%9=d9P;HhU>^ZK&&gOiMi^G~h_;q^XD>Weu0=B%1ZY8)ok$pxZ^
zn&ThrzNVBTQ(f!lq<P&p9UJF~DUj$<*r&7H>!RlFgX`}OBll<p);2~rtKOVn!9bEe
zhI&7-EhSvMm}q&0MYfh;rI6QKFFm+1!aDYNpG*{wIOb`k&2NDKMI_M7aD@hmegQ}O
zrE13KFBr2K+4!}f{6PeXSGN$sW1~al)vXFnB5dwQ&-Kak;Q2o{SdqTYu`-i0;*3F8
z)((xXa^$uN*$CKLDqUYyjAYTCj&c4E*Ir0;>lgg13qHu{JP#%nw3ZUNayO(EaV@lT
zd-hx<FmIiWXekhYN*@LNQ4lO$++7ejgkbqQ%k2Y#gvuIt3dsQ_u3Yz=#jEa>+IRl7
z)+-M~b$vz29pi^`c%~s@@q`(i*lOFB9*h%1Ev5;+u9CyBUKWC>a)s@h@%12}#7<5;
z1oK)G;+6PJ97Z6Mtp|$^66YGgV6$AyA)PJy7Hb_X<!As7317@y5n6h$boXz*YC#7I
zYP1WQfCo#U%Ni9*)7f?pEeiHNyGbRyKGkaAXY1RgWM`24{8FX!5cpV_7=8+zY)8KY
zX)13ra|Zl+!$I`2)zj0@ZvS_tq}~%3?Y#<0novt2;Vb?}Lq3M``su?=Ylj4if{&uY
z%HxaF>u<oCD0Xx0#s;capq27z*|fN%qy$|~NDd;KIx{XvM^#hZtFfkKp*xTIoI&eB
z<ezKus56R8haeL_KL-al_rjvKj(qanjv9qRuT}%UVQx{i80yK&D~f&sKTP~1x;T%K
zfkRs*Y^p8HM%!P@HXVs{b8Z3_Xz<DOq`Z`F@E0&LJj6i{s9%MA>m4wnmR-%XZ#e-u
z!rxo`*oqz>8pC9rf<jrzxli2*?;GsiAfAx9-`dwwbS;WUK;wqq*Qu0x70$+<=X9l~
zsBr41CTitAbnSoN!=OOISeuv_7#J8C@%6^{4GBRB|06L(LqmHpyTHujcJ}z~MGDji
zO#<|j<|HYz)jdS~&a?G1jOkDw8*6*G8$FasBjPKDbFkF4Hc|SNf~^}Y`)v@Tlap;p
zUMDjv6AyG@yL~HfDPpS!wroGD{+1pL{(HIR;frY(S}J~Qq-B(I^MAYqW_p%~Hrt`6
zj<VJtTT|fs>2JqeA~r)t`oydUhk#Gs#4l%}=;B*@erxyVrs{1$X>2AP*aga=54ZpQ
zRPWoP>SO*_*J5jy4_z-SPqSTps&WF}Sy@q0NlEbD1Q6-2luF}%aAHvzRdrmaF3bCc
z)>C;${(mTLztnNH2U~@b_WveuvRWISm^rJ89MFE6d~H<8(P{ME#Nht`K(nhU(;Evd
z*HBfWj_X%-(_WOYrnY2!GGKsy6)1GgM#0Yg{+N)xwyt%{R}LJx^fKv-Tk_vaN@<3r
zTp~+@J<4ZAk!((^32%aLWY8eMKQSw#yGTsJ#NO*J-uVY|MkS(`t-#Od!|dcl<d<U1
z3V7OVGV%Paq9h|Ks;&-5DkumE2@(2$)h8|5)RrJ=dW#7zE!(@MeF7W)@3eei26r(O
zUl*#+-6%(GvPjIzSCd~J@A1bNyq0W~u83sXs5IYeC9%&w+4L>y2Y=iV#S4{`{Yv=T
zjhXelk?}*{muEF)x%ttb@^AFTvyZS*(WF*02<wsQ%+|pJ3}*n8t^Nb%I%<&{4H><k
z|M+g`T`hS|m(2x?$f?;DmB%MnL7>5Nvp|2}eUE{><{wB(V--NjCmQRZ;)P>iEo`A3
zaep|ISNOBm?!Nd>iyMQ6rVB;DJZKb@$@eZuhB195QIb;mY1I4J4eg#D6=8r3+rnK+
zjVhX|o{hISCl|;kg?+}c0eSnI^iBdB$2+pyEIZtCJ2{PCnf$)F;1ir6B`<DPcEC61
zx-{4$g-Z1D8t*FJC+W?HzK+>nhmDq<d`ShPQd``V^HKTQpi^zhq1!X|O{3lzvSw5;
z_YSvrbd6!3p!V78A~n_(W$9`z{7?OGk8m(Zz4{TT`2kAWrJL&%gP-vC7mEusb8}0L
z)>$nrE#Sh66QAA#<--i?$93Qq&ujIQBwg8QH}@RG0U0Wnu(eUPkFp{dZ<~BE&mx+7
zt)MUDp(!wiz~WYAmGI2Y=|KhWi_RZ{j^@$1Wm_>K>-mhX_g=S*0S?h1f<MM6NO^vS
z*JEr1xLP^i8U&M!4J*yNDYQN=hn6^7j`@O!jJp%60uZ}F3xSeQtl&`G>d>&bL~x^S
z*wQ@8do4UI@|B#Xorep=&uIwwue*6-QG|SdO-!;YD~pQC%F6QcpzU1HI^kt86`5zG
z*W2HYe1~{sTnL|I8#v~b3!x1X@?P?}meufp<D7U5hDtT*g^xS79eIh_tC*aI($^&2
z?tS0&5`#dSURcSWc&X_;_S3tdyvN(z6`64P)oT@=L0iwXGwO}Qh<I32Wwwb*$l7nn
zxT4$2vjCsxuXEl*_%Bj3xqiHi>S~!aoUS`f%_%7*$JGW@RN!k;A}YgBj_lZXA16Ja
zH_qO=TKS=KXYF9~cf_qD>%4MT49>1+ERf+nrvoR@Hhvd0Y)DE*l$oycd<X56QCZgU
zor8<&79^d7B#s^Udh!7BtATTpQQ>_f=`Vzm63;GTLW22I#b?>o@@X9!Hvd0_TYG5x
zp`E%jHQKyc&@(Jkk1XZ-6uj}(`XKP((W737*4RD)8%vYKz{$mx>CEX2I#Y9ZcjxBj
z)>sV#@eLU!X?tI8OtWyZvvvC4)C*Jy7Eukzm-wZ#>h^d@%1R^$?2)J<KQoi|ovt?V
zyJ&yPcZk_-^>2yv@o1~8FT~^f{^-`C#b@n4a#nmSVeVc;XT{O(*R!j=mk2C0poQzg
z{$UEFy4a-8b~Y3<hbHAc-M@L%y<DRluePigeuLWb85!m{z-GcBO3MUZAcG6@s89?9
zEE&nZPL19iL=X@!R<^d1HNkD2?2}xBo%Ee;#N6LeSK=<R8~ZhuQZ7yd&1{c{ZKH&(
zb*lMVWFw>WMo)15=|ECmFaU2Xpoh57vM(!c*&(da^zhh^b8;$*O13g5wbW4{_(zu(
zt#<FZ@_&o%9?-tqek`iCBtBmG0y2<?=FHs8b73Cqw4pZ;$I}=&f*ccBQ%ab6<QnI4
zoe<caGuyS=7{_c`Pc#NS$!FFT3}fORcKn>2G&GkR?cUxH05b8f0f5RFCmxz@!bdOr
zb6~x)s(NbH7a|xZ+sZIIZ7R_q*v+e!zu+~7e7IkrNdiHgj&H1SQf9q|jsCQGx%;Eo
zpE^&}5{sW{rBXL6Iurw<0sP1<r@Q-}@N;MK)rchmTAj`2HU)=w7-sE^3~ecRXJ-Gt
zjgE?Y;b7r^IlI_JqyUQNyA3A)J1DBMnrBiABej=$yOxeS+&Ng*)?q$78e|Rh{Sp%s
z5#jA7LoHbji|JHGJ4wDX!nj6#o<*81*3u+<toPrzxo>jYm^tdw{^q4T!cUiWzvvfh
zT$FiNpY7i>(t_hJF2EBgwD*YWK#uFAwq161i#wUsFDk+l0))u^mXV(9`4kf=z3XIQ
zJ7@^=6o&nmLPU6GJ)QCAQufG~I>4{LFAD~?(Raf1(0fUxU|u={dbd^z&p>`Jjx8#4
z^K~pdJpA_dmX&Zu@a_J~7kz!g+6`X<D8eU>n_V!`KOtb5afgKewiDKQ2bnb?AGTpO
z&Qdj@xy2fi4A5R2VPopZhI;okPEuyD40m%?)Ve#pj?H~tI}0yQZ}Z)m30<7O8$Vxv
z)72$39uIyIxNZD{-4>44dFZ&lKjXSOOC3qXmHm}*<PKgqt?bh$hUPbFhG!m`Nyah}
zJ`EOYl{Wuw+_Iwf>cZT%HstKOy3g+y*VlGTsxU!f(*WB;<HZ&UQp2HG%hneY2b%2l
ziJo6N=9O)n{OAlX7b5ZR3=liAt!rVUbd{!oMa1q0n>k8E$%r73g|(%XS#eoOnYD$L
z>FJUGs9cpSG^V=hu<nnQVcR+uU~qkUW2NV;xLZ(qUMu;jejfD>$znHpfywa_Sz_dS
zvB?o(I-uUSRQ45slADb3HQ0Q{UjaW>!Tl@SMErg@=h^Y@K^u-NiqHZxlw?1a2H#@e
zz8NqnQB!sG4+u0n&Q1a=+1HhDZ6M95Q&d&hwzqO7=%0B*Tmz=DC}>gP`Y)`-8`sxA
zG@JcU`%Eyvf=o)><<R6^dGi&S7JWT0x8Xa0_*VLXy0=O>A8xS@>dyzW{u`l&oiwqa
zxCR9*?e!gCb=94qj+lF>=65hthoE<9xA?>VR4HQdq7-!5t&KjdnLdIT#`-6~sB2Ww
z@XN3}Ag52`=L`S12=7;;4X*<18{V#Z&o5!xJRQ4c&c6v_8HQ^80(=rNLp*yoY4?E}
z9iMQm-}vfLKHVvWnbjOjPU%5ES+)Bh?R%;}fJ^Ic^$3Ge=dI+#Yl^9xYy2DG755yb
z!xNcQ52QWr{JDtpO8EZvAqKbgIX2e%V2$_>xbaT?8E@Efbpn_0V~~kuY2~mx@$rQd
zsEN-d5X*1U21t<J)`_eb^W}aZsRbObkV0f8zL_5p3*(ycdA#0|uyzKGF&cQUQw{73
zY`KYf9qbAm<iJ0DVQ2H6;olmdBw!J?G=yc|07QF^J7K<+Rpo-NZn`d**c!|Kc|s5g
zK+}rrY!e{aRQuhpaiTjP9iNz(o0gZA($J5M4hI8!hX*#-(SdDsO-%d`sDQGGEF#Xv
z)x}&K)1`RHQA(EM9jaGsd;}9(nIifNb$S7Nq@#Bf#3ekcA3R#6+*01ZmYQ-HL_YlD
zL@u?o6LXMKv^POfSq<5(7OCD|f;<b4>&E&+cU*q?PU?O4v~769F$p@FuUJ7#AiNv*
zYAS&l#z@$;{SR#_>OK)+NT3KSh}Ag^3F*?N#?PI2&AE(2{t3k*XPK<cJ_|BA4mtc+
zvh|rry>wC{^YL3y<u-4;AxyV(J=@}98w1_M<ham@7R6(9D@XU|ihnI%<?>)Pl7Oty
zo-9}dZAiEaS4JMPyvMW)2D}x14}ndh*44Dr(vdKX3B@da_h*W)d%$};_r_v+c{gbZ
z1qlZ{_gtgue#nS5gbPUGY+;I{zWL1c;_4<n6i2R5dV;=VFynF(LQTx;wJQVV)-Weg
zN?zMZT~c3jY_}?U%bT0y-%nM+0zOhux1_8{%Sv9|Cd3?bv<jEY+|2ZwjH-vIgs{2F
z2a}xkEA3VCi?yv2A;4}#R~@k&O$uyk5HJ4I=CQU!-CuZGQcK2Ts@pwv48po#s_t-M
zxwO5e3+Eqwof)iTK|2@21v`GDSvLQ`_cbWS?PU7%BD9_E2EYK_p!nmQw=S9}G>OCG
z;-TQHfAXzR=?M+pA@+-)K<a}(3-!99wjDA&9?l*tuu0tl06zyA)y_7rj;!;cROG#}
z3=s$OQf26`iC$4LVd>@w<kyv}`#EdUoWK6>S1%D2jIOX^MPQCTyE7_j0oYK<>4W~A
zDW(&eN()-6;!er%NsSRRx0mG6ee;(uzqPTYUL-AN$o07{crq*`rQB#+@pIhl`uX*T
z8CBI&gUZDt!1drg2o90g)0-e&(DfPo(AxU61ZhZwY0z`A*}m_!-&Qd%F`F0oQ~gBJ
zMmZ=XFx-I-ky&{ehS60_pSnO_cYhB{Fqz-}?!LDv{PlVz5wu+bYXSA)YUgKaPCu?!
z|9d7@zsK*wMN3#5zMH_CS%C|YER$%gZtnA5exJYu0Jr(<NGEWIaGbI%@^F-C(}qu&
zrlu+nKOXVn?b7u!GbS!8<tQO@!ZH=_)ar20x{{};5G_i}i8@9~*>d$UYKE5d#%xm^
zMgu1C9G>7N`ZMV)Zt(L(9ZPoS65(=tw5#=~&9-~o@8xtZ)`#5}jfHoeF-8Bw+}eN?
zX|yE=M9a9f@94J@b5wubA7!-c5#95YRbj4nzQ($QPu|VxifNfznQ8uto^-6D%0}Bs
z#DrY%chRT^5YzRL(}H&d?e{v|p>T-PTEq?#fz9%#Nx){G+vOizTICf+cgLcjIZouH
z<fj|k><Zc{)W($<Pk5WtsPNa~mtAsF4Miv*u{CH4?)(<nR}IF#iW?SJ2OQq6QPd~*
zeY4`^UP)(~nO$pDREElD1uOoa<qECn(W9QXFyf+r+>Dk&u73ft=i^nhH~GA4KM67O
zM|}nrhf6uzOPgxD&CFiDkUp|dN1m9`TqMT%k5+C?VOeifi3Y<XO{TG9P1}Xi5%B;<
z_WOw+&~q2{HKh$5)ddxKt>BGkWhbDkYlDuA%rV!&z&IQLRmVrHGP$Zs$n7AYB_SjE
z8Qa{iMJtVC3!Dv<+Opzyld0BH52BxFc3byE$47HZX1kNFkZVTJ#mnw2s;&Qzn$5~c
z&c$%8=iNA{B&HE&5C}hqy<gD31vK%{CudWxG*MZ#VuwJ}vP0ry6JgO&AD%XX;P~Az
zlhRXTGE`L6<+W5cr>2(X*4AcL<WS(uXJWHNct2g=v^gE8-yy~c|NJu2`nS*ju_;C~
zG4jjz<!$5ncgm?%E39c8Gj=ZtN_t^?0x@H>tgfkI|36IxP=66I>Y(e`=7A0du%PG?
zbIW?Y&6`E=4_EIJo-N*aM?F4Sc|Nu2TsfKAc<Xbb;s)^OpOCS;Z_d$n;cMvAl#h$w
z3>zTQPg!8Nk1bA>7ITeUq6^h#-K71~`CzZQw`JwT-FY||>B{b0*C0@ZsCe_TmirCX
zt7in)(qM^<v|dB&kVY=m31~kiz<Yi)?FmC>vQ|)3jL0r6Ev+McF;qB<6YF-ic5l+;
zG>?F(Y$QR+j7oz>%!G^5P||@`9tieNO~D!>Ae`d2uv?os_>Nb4i|B$*OqBKd_0CXT
z)JWM0G}@QUv{w|Fe$P@g79sGKTZa`K<2(OT)a{GSxPimLRkhm-Ca@&jnGk9VpOsvm
zyW-_ZZqFp*@pixdqBo7mSaXeZ4S9Xt1^a=gdC7n~xlHQX)-8ylMB##n&cls_#oIyy
z$(H`EH{I}oLr0}I_nX@#py2c8g!y_0WK@E*jE11JNi_1TUxR%EykM1<82Vv#Qk|GP
zH|n#syAu??@^AA*IG}i{hXe7|QrQ#I<p|>>WN6AwTY-4wF<CvG3%F}}XrC%JE@4zu
zRMa6A4+-<IS)9G|C8F7I*vj_tr+KUZXYaDRckRng%DRqNtRU<xe_McW8Ncq#*l{Ps
z&!ZBv3p+|c`+@k8JwFpaJwgspe&RZ{JCZ>sK*EBix0tr11O~o`Y^~ZXH+pm35ioCj
z5&%@48m~Lvo~d*nmbsbEz6l`XH6f#1+BxzVcs)Wa`TA+(bMql({zaOWEHjIM`Cu$>
z?V<X51R*uq*E7-G53ZL1C)MHKAr`G&Hm>j6cLBzuhA_W^X-Gr0=;3bJpIm2?IidK`
zH#(Qr14pZm03vpI@6laOWb)1r?s<oB8E~%j#tOkV2#ti#<+AJHS!8uLsP18?Jl`(a
zkxBFNakccB%0@pSnIDQ0De6(ikrzt}z9=lIvZaOWRi~P5RwV)ZuZH1yYA4z$J4PIs
z7mtnM%C>H>AY;{m0H$WPwYO^#$>5)Fl8;Y*3ZHHO{>3niKS1{Ki2jSQAOQD##qaR(
z8idU*us0YJO>k?~F>}l<QBL&w4`PV`s7D2|lYIl!)$UpE)w8G*3=<O*Dj1qv=6Vit
zBHI2)b;S$6lMY_IuOPJHQ@?14H^fduZS!qq82<GbJ9|V#l424prhpyjQ&DL>xaegh
zV8?<6w5X}ln}aJVP<&a8CkXlN1RSYt`ur+}>!;)1Wwu~Sa(*x}BT_<hM+M(9U{}1s
zJ2>2-pc3)r=KZ6g@9Y2FKC;3Fkaz`#p!V?C9r)N~`5<~Td-JhA9Ow#-urH;)l6-w<
z`DT}G+7OT)7?S~C-bZ)7TvlFM*nyVXV_Q7~SoU^ecXc9s1YXzuU8TWaYJ$xXJ$-*$
zflya3Z8hwF<We>Hpysz{>hBM(vR;FkLsE0TL<kL)5|@-59U2<xrl$EhQej@I`hoFE
zVEz+=C%`2j`j$9>UyziZwrhOimy}O^*K5{SV)LI@NVmiOJrM$t>0x1MK}ktbB=OA1
z0PE)|0t>c!^LM`2Tb6^0*v|=O``CSu#^C$53uhrsl=8=jS16I@MtOzV>DgHprXN3;
znRhpcu3T#NWFgm_<vOu)ya>P$S8^mI<sH<C2;WNX0rU%i(dADZfydc)8PPP3zCI>q
zYz2b2L{)|hPv-Y0`>y7*>=<jxen=X3phxEQ<0ysN8v+V}^#8#V@PEDLyRr9O{>xL6
zV^@UTZsB-ZNCSGD1tsOCsDy|#(ikFu_EX)AcEob;Z@xyO6(t!N85I@P3NmlVVxbiE
zLa047>lWpPq@-v^1XM^yU}7XH79TV+fp;QYNCp%Nafk;J2D0b*a_!FR;{r6fj6Zm_
z`U#LUQ+E?JS~ySV@$mhsAFCtu`&V>UkUcyB5(^d>7zurin^Q=-*<v^wsI!o6Nyc-J
z$03ByjrkJ`Y1Ii_0>%ePhQVwLgNZ;9^C9?Nker+x91TH&hCArQ4n}f`K4J+7QB$qd
zy(iW0F9%l>YNMTj=c^gh=&oh1CRf;B6u@66+LE)`g(u9Fmp}1NBa$JAh@UQRuC8tD
z1(THRz#`HI3O$_)cku<B+v=O|FEpYeQ@rG5&Zeh;UzCP>&fmV3nVOhiY<7NtW90B?
z#GF+mOLFF>=ZfoPO^Pz=Lcdj`vy#&6;koIsc)|RItAjiK%7QK6nd5{N1tsZQF<^3p
zmvhB{+3Xveng}9`+f8ghXV2=rZW`9pvb<|=bs8BRmXwkR4MViEHZEKjtwD4=)g<pW
z&{|luq!2PG$~Dwo?xrWD{nlt+nR)=(E-YF|{arSzp{Ay$fzN(mWo6`~u3*1JaWIOA
zZEn4?Byft6k^Ecany2P!%5lk12*6SC&3z*)d-MCXDY>9YkqGLmlNBU}MDx1)eEn@y
zYF0z%mR;}a;Naj(<E2fr%fg9&^){p>J&Ge+Lc9BC!+sb}HIv~GFEz!nyv^n;TZYv&
z-<=<=@ocDH@KV%NSGl#F?wV|G3X6)87ArKgwKF-a*Djm9{#zVym&Fx|0bpL&d_scL
zumpH0s%Dmp!Km<?7&E81&+>YQz3d1`4);yX46%gA%?i$M%PmcuATvPHpvIl#ejwN1
z-O&f*Q(v6hG#OVkQd9AWxc+7r*$ZeG7Uh$|V__pAc#RXfTX@yRB!|aFhD7*@shSnl
z6oN2Vq*iZ|DqWv9-0?5~{HMpOxZ>iYqvB%fD01?i-X744`=(9%p00Vw`z_m_hao8h
z+uCVG^t{43bBe*2<3mSpXH11pU)+qGWcFf{mJ4!A+?VEh4tnZWAA0n`wVjeuq@gNc
zF`R&S&;G5Z{XAv1NG3Hcm2h5um9l$KMyiFl_tk9hPV4cQnza&GZ%gn^49N)7Q5BVc
z+tRBAH+5;<v|th9WJITC#J!v-X={543VLmnt)-`*m5;?gQb$2~@>Y`Zdzd4!<O#P7
z<s@P5{4FF~uku2nd`RvKj86~Ebmivgtnv@fU4iU1z-fq7X@~0BnYq?!ycNFnj7(gF
zJ)ZpP*H+oA*?a`-=~|mxn&y@jl8*F?qavl%WAZeFlYh<A*WKO0O;A`XGXJaRJbiu-
zz7g=RQz>AhuduW{CnCzw0{h|4Ei1qf-H6Jt8pYi#e;Crvm(fTkW$-$PFS|uC7U1T6
z60IxMAm}FVBX=c=9rq&#_mfgar9Y7$ETF3Jcn}Y?4FP3l1gS4Dm#5ZD9BbN&(KxtH
ziw_NpC>)DrN$D~KB!3bQj0={k5+&6wM1<QS5(E5n{<tNz{0k}Yt1bG-OFm2qP#WjW
zv%j&%!NJ{RXm{_tb+X+aUppJSl`}Rl6ckk$%gg3&{Ku`sY3_M_B#7-Q&p;I_O|U3*
znS8N2Ko8`{*M$!qO3VFL7nD%C$?musW<gc?w|NiXWuz-EEU|PvexSd?Z(*S#?ea=v
zyMi+VWN5xnq%TPpW4H5r0S$`azD`UmOU;KHv%v=xUH)?dc1uB8d^2@8c8ZcX5ljLt
zgS|r#2<tO(TC9@l@)8$kvv+i4)w|bI@AZqb6o7DdV{h<weuoX4GHfeLy#DaeeC*Ue
zy)q^zB_XN%A4>=yaeCnHk&ydeIC)OC!Tx9TxaE2kEsepcq1*g~H&|md>ks&C=Hcaj
zaN;n|`r`)nA`6Eae#KnMUU4^5ls`mZ39aMMb^hHyXMlm(La-ARU~$)kl|t^c9xBoT
zB=w@6z25SSiq06dZ?1ty&UX<fJrOO9iy_uXXK|j_d0AA)#)U*&qavz)5%7AoZP#;m
zWLZ%ZUI{FTH`aPDaa3rWBXmNl_FMj{`S~gE{B`Hki2AcwBoV*kLHRBvZj^?zuMl|_
zRol?Fe`Y7dnQG5yBQJ-2of-JYrX}A1&xFL(Yo06oZ2rfFGQa+qH+Kq=M^4rMP*(5$
zjh5dv$%F+5p(b9fu)!2rD(dpHj4qJ}(jWJ(&t*F!&kLaLE495?_%`xB=z~qj1LFa>
z7WEU_Fdf4JO6op7L^Q-}r&b%d_anWtRdBPllm8}!_<^2+SfiGJq{Qgd*a{dBOKIFl
z(D()g;tvgXO-xTrP*Go><6`1_0P*Wwz8|lzJ?g61Te=Fqm!+SLIrx7h2YG>Hl&M;1
zTl4*!)l7#ucCB%dk|t-WuR=%8r>EF{@Eh*h4A1Z0uWyOT?qnA2|1UV>h!!q0o>xcS
z2EdH<lDw_$4gD|WB?3r5_PmAVKPJn4q^ZP^v`5l@;a;ZVK_P^Ik^R+{knS6x<8~p4
zMOSpH&F2hF!V?-EDfQbmB}GF{Q#&C2$*1}_;6dm5{rIitc1c%O!$5lebmY2bW+%6&
z?(xz)P324m&x~#~#I9BjW7qoQ`p}Vp!*8lY&9&e8d{%gd>lR2w@`2~$?##!`X)m)j
zu`sj35Pk`YB7wZ|es)Hxjmz`%)UuSmW2+pM`gCLW12b#w!LifC{uLdw6#<up5@SQ#
z3E_&=WyUg|m2gKAK5|*R{SB-w{`Qy&t1DOeuVs6_9%Y>L{_6Otk)&hqZT5Qg>G*2<
z-RgJ1telQYERgWXiO>v+smusXpF(Y^;NP_bzsoyu1u+>3;mL0eQoB=#FJ;>Q^EmEe
z%-n>IXfLwYIe^_Y5i-9g>AwpuK%Pgt`akbKNFmwJV&w_H@oJq{jH5WtSB~lw3<Scu
z{Onyagl(LS-MTPFMIm-5dfH=)u+Vz#iHpf{^|A%HFr5^oG{jV7Hz6>e=g!g^FZjKN
zUy9H!Hab!=GVut&i336bv)7G{!t1s8qad$EM%6UJKZ!erUq4{2xfeSvNWYT76RAUT
zISPl}fSfUAvN~Zp9c<RPTVCLT2B`{-4mQ3sv=7Ec##gCcZ;}tct)1QOJsw?d3FyeG
zh-m=Qwnmtb(>hf@_^QNwgJNzsA75nPS7w=9V%p&nRo=cTc1u`WU7QdQ5)ou2r>AD7
zWTb&z32u$ASBs1$AGgQ5L58YpW8>rF6BA=&zh%+tl(9({-*_jb7DQLL1D3yCHaa6o
zs($(cmw2x4?huvrfDrDRzsTl%iwe$6lg9&-zB@#7pivUi{my=|<gLztg7vZ6xxGGB
zSp$)cuq83bOx%h{Gg^e2=-qP_Q0-O$tMue1sx1A55JTf|%2WRXiLDqRMAa+hMOMe7
zfk!&YaXI6#8?oc&P0u%BZ1Cd9@$~YpUw@qn9J--!uu_vY;C|>z9$UPlM>yVPK?0aO
z;j014vB}&)!`99o#*dej?)%i-u|Dc<j*RcO522PtT}{{|I{(K^AhyHY%bwPtFjH%H
z9VUJ*tdABtT5?W84BghLp&f$mT!*{yN_+|L^GdAz{vB!8lU_+AMMD1JA2;EmJf<Nz
zbT1;o!ke!c3s>^MxRC;_DBwg)hR7wI$1>rCF${}Rks(eCKZC#a@JilbC6<>Z=KTvL
z)g?J?OXbqtTkO3z=xI9hU<8U{5XpwyytG}>+Z1WKFyb_3>MGcMsZF)ft5=zp3f!9c
z)c!@3oxv-HF*Fr@qv#M^ocRE@@9vp5FrjtqA&3HDSHt+1*4o-+fS0XVRc%fU>5Xes
zR#-)KqAvuvYm!}IN7`j>K!|`t{A_M4-8Yt7sV-ttvzZ7pOKsy*hfQ29k`xIO#J4q#
zsccw|Cr->aaE4_F;eebIiSK=g8oeF8i_DdUdFnr}M$GNdVdy*=+c%jJ%fA%5gopqt
z@AGqqEei%LR#6pjMl9`T;55OPgUFR|j&M=Sl%_`J<$64Rwznh)a0o0~PjPepV2<#@
zW%}=ueQsV>`F$)Ba5;_LCu75}ix6o$PMVLzN9y8E=giUF!Ax8)votYszPo=*p6gN1
z_rTA9SR_hq>J$rYXhgbUQ53%p&Hv}gyS^1rm{$vS_W^(|`a#>JS>N@$t*u*rWBtEe
zCOY=V0atq>5j`WfFQ0yZ$xoL@Q+{4V$<?{5WEB4&P2a#CSNFZ$IBmnGNt(vC8#`%|
zHnwfsI<c+Bwr$(CZB6X_=lQ+Y`vK-Ud-lxPd#!b=Fh}3wc<TW3=RaAEO(?IPuhs%|
z531F?T#Rdr^D9fAWIwQ?t#=6Q*yy%ve5A$g(S&#`5fw>Ir+x5{$ZzjMKikzZ%f+FV
zYo8}oe?sH}*i}SZh5jg^hrGSAWWr5TR#K_MZaFqq-QulXnfP6dpUpiE>3~_^?&jk1
z`ktDSj+UB2^7iKLaCTf>LLXM7Sdi7ujSHQ0e?2`G@5ZJ*nVMqKz#9TrXIh_Nf&H~J
zJTxQud|q7ub#f-mY$YgK!U=TV`hd?%{J#FWXJZYx>pQ9*W4a(>r!a@%SB()D8W?#L
zRo3Qr5RhJ9u(;3qwHp3*ZDf{<-IDzH_KRA#H!@jTO?7eXK6J6Ys%fq*binEn8@AAk
zR6jHQC-%vym32jBc|{2gJuTfBJq;Z#psA_88GLL_;|ZQ4A@yynPRsrK_rvW4GIlfM
z%034m<|xHwM*4uQ_QRw^V$An5y07rkt)>K56Xl+}))=;`Uop1st*nb^Muvxn-*LZC
zo2k167H(t>qY!vKou?*;Om}Zol&ty7Tv8i{|9Th-xvrUxwNwM|?&AY^!hjkRZI!!5
zSK`t&hw;?cPXbQ*WpzT51e0K675!bhxB$5#!nOyehF=QGpC;^UnP=r^neDaEzZ35H
zwOG68IiXq}dE&Vn8yZPurM@xJa@a!9DRF+`He2~{Ul$?j;=-kmompww2=R!~R8ier
zVB@^G$IW`{9h^K#?5DiC)ksQ+O-p-yZ5$e+uQb2J&(t1ZiZbxE#yl`08x&O=WdiiT
z90ir1W|s-*;7GuwwiQNwA5$f{X`9g^d=C40wlh}|EmcYThbfYaDFCvsw+!iN!132&
zI~}b{LVC>oj;r45>|@{F+-=#~YfdN_KkkSrX3M%467P)4N`=$2rW5=8&|zWPq^u?A
z9T(zuIspD}1Z6H-iv*qeC)`SF0-y*9+EObEsDJm6YTaJ1Z91AsN9N^pBW`$Th$wd)
z|Iuv0vuz{W56Zgy8z<2CE5DHJ$?m7T0<OG#lKw$4;ky5pd^|V91b;TeVGFv?-OCwS
zIhgSB7rY(nrHFR$LN&O_eM_gUNAIs>(kdX~asP>qh4Jm3os)H&qX+O5aFw)Q-0H)*
z2)f31k@a43H@DmI<(^13qZ|#psh-`Qw0x+{WJ9rB7m*z7Bdea=OIys1pOMOo7WlT(
zc-c8q?83aVE|VCCgLaM5?}rZViZK=&kfrE%Iq{l49vz$|!5o>)()8x;j`SWblMrPt
zBbu#jxVW1fDp3m_Jb(nF!QtLFD5-HY(~;z#052aG569fBnzm|u>OD>sxNwu{O83^e
zN<;mZ6sssB8?+Wv+vJ8fJ0qf(X+kX-an)*OSQKE~xZc^tO$nY{a}}<=?_zoo%~H9;
ziVjRizK#SBkKOL|EgfCSfT0Abky%j)a2MJs+LgOB*EVth@aDn#t4;?cbgoNSYj)rK
z0Um0}kxehY?Pv93?*Pk90pYQUArV3%qqnc+Rq->C8$L0%GV$-vx>#CQ%tyQ8#_IO^
zil$ylW;?S((O_MkS0fM4n#P)yi|giRc(+FiPE@4bI|6_`4_ZU*z3DohaalG7Zyy+6
z_yZSi$RH2wpn_8WG)yzp*j}H{@HjqMDHu88KY#a&&%Y`n&Zm~^b8xQnWgF}nJ7mm;
zXspoW-?~+Adg2Zi`wugPNSHcH&EMbrmI-<BEH7H;6lS}F3~Ewc(QLu^=Ye0S*maHj
z-3Wh!c0!VPUY*{>yu5;f(j(!HiIR@$Ou)OcFbHTpofpTP$gJmbfA29oG+(sdxJGup
zJ~N~f^dy&?ks7D<!g%>Eaaobc%rdn^+db;bm&rlmA}8jigv7hu9=JgFkZ~}F6zM&I
zXlJuYoJ@JNa7%!$>L_gcA26t~_r{e`fq5pS%AfX%U|i`2BLXH!&-tBx8%Ejil=AT0
z2ExdF*l=j0%&2$JLB9>=!Ywo)^{9f%G$C`COSg*CA*?II{xB_mjce{vPA;C@)h812
zgYsyvXw|UeHn!HcG_~zHOIi+*FPvP;aI~?zYkNV-epfVg)OVk}+FP43?sHZ&99(~J
z;B-}SkNU-j&GL@Y>GhQZarVD^-;tOLc-S&_Mcjh;)?N`4YS4twab1rt+<ZI?lGS~<
zbZge+HTLXT;{W)Xw5)M^g&ZbLUr{FX<vC>1FY~w*tQ!7u4Q(ASwxe-N0zO4CPD2{I
zL4vB1akYznyLRR;K{}^cA)a1(RoG)_hMgWrUPlie?c5rb8a7q+H4KC0;YUXR77nML
z19&=%C5p-K2@YqcltV+1rme^mnO&c6o)omw=Cv%#C88dZsx_?FAAo?JBqg9Lv^=n3
zCMm8mb<neOGeY`heGo0QxuGC08=wA6Lv@=G6;vDdcDySB3wM9MegkulU(XFr^81#0
zUy}FpvzHsleg`{Ldtj~8@pufV_H!p~0tG!HlEjB|J^nUCca-LI4ItZRLx|MNTn8sU
zr)Xgskq(LkLO;8<XmVJaJDxNZPZPfRt)#rnDJDEnop02un^uOFv;-#J6EdIP&}F<6
z{_$0RY-x73z*}wmt#?@r6H*>l-rUgw$xGTm;1YP=Ml{pvD6Q3mn;{3xa7K2tb*ieQ
z!LFy2#I}X^6`=t1%H*kNWI+z~^Y;{g!a}@!G=Bz3acbPS!msenmwTdprS6-&XG_fE
z@w9mfEiaC&Az_2sK<SK$4-61jacf-BJuU4(rR*Ok-aRVTR!KWDFQ)qAj)8S{=z9LG
zZO;K&r$KKzK}Hsp5LFgnwEatorOL7Y7Bps=R<Vv0?vkhl@0NjYYsfHU9UIek|3B3e
zh_d=^F|d42A`SA}myy82%4Z%g)dfZtX67&yzK?_PJssoOQ*a^`bQQs_iD5})c1{ri
zEnVN>@Wed~^!_}eRGK5`GnnT<SAO|3{R0t{e_Tz-I{}te15|R{F2A|4aJe#@d`G-y
zYpo!rO${ye<Cf(4R#esF@$e6T-sS0oyy}jI5h)HVN?Tb%2wp2XsYu+_#@gEIn(A8X
z&F=2uep;$ZyRD<Yt)qDF+mlS<*e?Z2KM86*5x1C~XOF?G*0$#{ANk1<TB6*h$)Woc
zKtjY2^Ba2?@cR0yVr?#II}z3t?4F^lgGfg;G}@dV5R({-1F>w~Jw~5f0<3NHKd`N`
zQh=Z&4Y5GkUKPesr=?;yU?n{Go?ohq@DchW@>hPdXZXfTL&=bQ3zz3!h!%bMv##?x
zZuxi=j>0M}DRjd|a`KNO??)I?n10qhWPYg*OKcbbhLjZjE6y~x)AOt2M=*F!p@}xS
zv~WIRX{=vTi<xQFnjmlOG(k*3SJ;9n!;9REe{4npB>cq0bGgZ>XJoGgqp}9-jTUOi
z!gO9i<>BB=)aQfyiVi~C<NSy>=1jCrjf^Lnvx?N-zfde#+(cf?FwBifzA2G0<BFN0
ze=7kqx9Ci+&}QB|6xXxM3!9xQh`djShPlgPZ2>0C8o|%|#_TMiFxNozRMpp^vuR$B
z9*~HPgNW<t(a6%o8o#|N884oj%{;Tc4<U4~JSI23w?6U|I5lBIOF1$1Cs+_;zdk(j
zTU1rVJHib67+(qvW!5&7(hR@e;kxO8Js-%bFn$)_S;v+7Su%XJEq#?bbnFZ-cCNI@
z(&!9wy`?G5sYeVON$b1fvO1(iBK+3|{ec7<N}F2kZgcb<Jt?;b@0nJChI>b6bYkM}
zV)q{@E%8wo28kESfY;#oHnbP68%b$lmhmBKtilwxnd9{7`|PhKNKFk51qJ`Tx@|z9
z15X);Cc<&93yEOM^Jd^qo7C&;xD%l(3y=KXAX-Ya`Z`NJY%ts~E?ewcF*Z)4^Su{J
zoSHWgA0HpfL3Y$BM_8}g$d_0_4Vt>W9~;1vYk)J(&F^>#4I+a6+Lj%(47s^quXwX3
zaD9#O#-#q6j(ot33~r`gkR*c4U>HDs&eqW4va4k<bCZ|X*VpIoZ!9f%;o7RrYwU=l
z;k=_WkGt(T8wmcCrA>HnXo9rZ-_xm(8#dnzFvKcR0kI<>+>4Zl{0m{sl5X<~_>PGm
zB*x2)*zUetL+hd#8;7!Iq@T6c@|or>hk7z|a;O(*#~Mb5&PRVB445vT0`hzZwjYw5
zIsbSFJL&m<IEsb|oz!?pxS(-aEcZ0z-)=yOZo`lGy*&8&`T0UZd0IDVRMr0CDg(!p
zONAqc%UB4Yk9CN=8hs(7OIlxPT5}C28A3Dnb{WgY{9R$n69<OrG9Vxz-rt=sH=6Aa
z2NLWP$=rU1$`QS4ScK2*0evz%TVFRIyq-Tl8F+L5Rb3t19(eO_dnR-!{<d~U0g8FB
zVo&F6lTO5s3Wlq+|2}s+b_OXYmJ(=WoL14SY47N0Yun*AKRq$?T9naHl2gWRWb#GB
z8K^psfbP`fxBHhe@p*5K9J4XJ5@IJtNLs45Yowc+tWkM%D`|gu6!=H4*#e(IsDoJ8
ztgMAI@<RYr;ke`(O+T2&7CTepD;<~;16D$6Sgmgz_BARUknkY2xprR%2jcLeBEvBb
zQJNlgg_iwCm0nm-DwI~w;54+Ixed-}NQrIBgAdM#47C_-Y<RWZ6eoT;?fwiSsSn<p
zCLuxL(V?DzJ@yd*jQ$KprKYAPr{FO&vT|p2h~lkXOX!Hjf(haDCR88u+m8|I#oPYZ
zctX^eABD>-P&)oF?Wc!<YSi|be4l+&(d{6|@;*G`flq5W6tvX;qUWU0Ej>MXG&OMv
zb{eEz8_WC|@X4vaag#1xek2B>RX`OvuW%6DEA}4!B)|i>oJ9JEBBuqWz|`-;ih`1o
z8g(f-|ERbKA8;6GJhTOrPFzdn+ryR}3~tz{mu&L<+YuyB;NxOtY_6}hW&>{tWt4Q^
z?k|WgUKq?Np{06x0Rb&3B@q#gq^L05*SSVL-iu8&IEW~ovH#HdU*q(TB8H0*We_{I
z&j|)H7l0_gf}9-WFa3Z3XxIM;*(MZ!s{3D*D_v-^Y`&0p7KJx7$35)cmUNJw&HWVg
zSUpvyoPMy@=Frc}_Xa<>%8J<~NtD-xdrhN?dUyhXc(duERSD=?WyNKVr99_xTc9S&
zHF7~j0R!b3+s~5fe;Q?H8Ea}7m>HR-W@kPDPZ-~bEJ{y4p6PuiTZi%q3#@jw(BErn
zFinhT==FmuN8C|wx?V~msvCR>c)vw0r|S#<Fg6x?ha8j0l7%zi=se|nf1=+oj(FW6
z{8Za%;Vty94&hcsKyR#NQTb~U{U<j<i#5U6YdC4GYt7~pKAnnZjkdNE+b+w|9Hzco
z;MV)$67xbeY|{n)&+D_KOm{piv5+K^%SW!#{J*7O&2CPQgO&EBw3W%Y)@}@v(kO5b
zh6k6%{wbQE5C(vpO4@AGU3Z3!GP(;r8Mw!LKwA`1ih?A>t5vB+?C6M&iLoN}X{&sZ
z8o!g=b2tO17g4>8Pi}fdCv7s^BDER-U-c)&;0JqQk&p=b1MCRq6pmidRBm40<)aor
zt@GWr@44~yvV-~{w1zqXcF{jUu{Wu#Rw@s7|E*4<yD!9a`8lxgJ4;j<jWRevcg#5I
zqA8tAH!Dy%>KFUg3Fp<8*q2vWhQjw&ze5apnR&(r$MzSJ@-k0ahYK5im`)(L;nRrE
zc-H|w8d+amDdIS81)NCXzxqg8eP=V?@q4y}g28~trma<)o-a1B+I<9WKlp;*;d~GO
zZPb1Q2h5JYRI65UEeclR!Htbv6^z60gO|4!d_D>k#zWluoQx@6Mn9%4;_gzP{(gMH
zi-M}i3d-)?0)9uJeb!PDSO>b+Ci>}VK#GzNYA)%$5PvX*M3G30{I4C0oCcl4oOLqu
zaCso{$PX7Cq<0#~yW4bS%^a!@h0Wuj+)1uXA&p^FRHbRWB=SO&BE>b1y{qR*C$ow$
zh40$Uw|Tr^RFE?=l>=$K_?!LxeMH5CC&85j8wqQBUbeHFp)!uOC@}iL6wPoYR{ArI
z{m|=B*2743c9G2HD(R6APR<Czbv#$;Uz;x;-(;}~R#Nxp9-ScCN-h<gg%#RozTezR
z3VO`6g?(C@_U?;Zgwpai+he#^&PW;0Mf}`2?63A+o;aq>B#L0KP98{NG^S89uE0Zv
zmLKqBttECD0c#z4hCog!6?qeG)WOLINi>lSTA6oa*x3d!c2MGMA|WQF3YApTkIFZ%
znVyPrVn6?!geOz`p7|M18Ls-vC9jSdaBkICTlYFrkTLzF%?Rm&pE$TW%Ir-L+?-vI
zL&9tM<wJEV4+3VNC6eIZy62e2+hY$H!dM;*iTpz1;s8i|11j$a-*KiAF~MrQsId;>
z1?4s&w3+LX(-k-}44G->taN6u!8cLQ<kQ$TSTo7rg$E>(eIP`?!ttInRM}k-;6KJm
zu|ei)K?a;wWEb~Kvr}e)7Gg2CdH>Pgu_PGyv;Sy(vxbgxUH`;PlLH<Ab8WzTrO3)?
zt*$ULwLek=Qk=#`gnY-@|7Jt{%vwqQ<A(1}dc9_TI<m7S7%1By#DBEEc#jzl*k>94
zAMKp|xDq*65+>w<XCkGDdL;O&Y9@9kimETONBXFS7Mnb?EwTEX<zw)_EGA1HAvAmO
z&YU*-c1@h3@sO|Jv?RGzTH3L3!5cnQE|2OppeZ2~wPXB@t=jP(QG6`mWJ7<}dLvs6
z*b--x%fvptv?y+r7fT9ChluKndPnHWhI=W{@fTX&n(JT94smrnD!58@rjGsFu81Y;
zrA%|@^f&Cv>SHq)*U?}Nfpo(#;U1U6bNEzOLrGZF8O%Iw^7{BZ`f@E}d4OguI_v2I
zkUQU#e?sQ@(FRdjy)t;V4wVV6EAAmsm)NYcl`yE1=QKe~uZg+IiSh88NJeVcv>#B_
zxje{4VH9AV5m>W7*nU(Odw^PoN0#`wzwcA>PZzHtFm#Pe<%hR4+`CEp+ry_mYodIJ
zRtj+6koeEDYhwQhdWjw`Hd^I#+Hxc?vlo5pBpM&!;#4&yA8R#$-(36glLFK?SLPAh
z`f!}O(!3G!hqCB0D5pV&{zTcrYE?VNW3ykc^E+C}MX+dW!3WEftsA={ZDpCE+XOK?
zZ)3wHB2~iO_HLyNlc}zkQ$aZc;ZVPvvQJ!kNMvZZ%kB#2ZIe{L5z(Av+ILOh2lR_C
zx6RYpoGO^$_gEIXuq~X%i5D%GoSOU-zkJed2ke8Y2wkzU-AX!pS^bTRVF9+=Zz(&M
z{VVvUwW@~-l9+5EOxCoWo!a?!Tk6``)>g;B#=^@zwW@kgbe7o5h)Bt5<aT1I9>Q8i
zyJMUF<qH*<@c~PQp~Tum5WWnc@Y?m<=d@~@oQ>wJOjI{cmBEc=(zJtH1#s@X38f3h
zo{?eJ5VA*qNT+)ZlyucME>admTpnVyuh=h0U;>DiSCx82v9UwXO81+Ui{8cQ{dA+a
z%9wNnk`okd%q_Wdx-|9~I>h~q`DKF;7rH_nf<)OFjz-(7AW_cFw^%@KhN<S7A8ji<
z?zG%}+m`-Di|s+uTPUjK7p>lSaxA_pM~dmFovgP{!DYr-7n}!g<X~S<hhJQ(H2)t3
z4LLO_-G4J$i0i~gQ~DeP+=R7t*~UA#(JVk~Q&STI%gpM=Rm+~D+bz#S?O@&hlb1y#
zqP91Pz+k<5oScG+kQRu7E-%X|0Kd8cQ;{X!V3(&o<zG&S(w%;7elJ#D9$C5BH4HDp
z{+Ei>Lg=<mGx%6KeQ@5$Pc4#t1`JMgS`fO7Iz*kK!kYS>ksaG$Y6BO!an$NNdjwxN
zHdYnEnRz)a7-9W^ezff(C#~umXNIl9psnAC`y;&|Pk*_a7@!~vV3cEHcX%uUG0NYd
zu}W2Fy%WlYU`?f1*yz3UATbN2-|0;M9GXuxp@e7f3CXz3G*<KM@^ysz(E?W;CN@Gv
zL0M{!JI7~X;I<QApWaKln9BQrC&Dc$F-l!(d5WE?J&*?uh4DMDCsm!y6}oK6Y-G=)
zBkzKAMq@3@2Ud}E-Z1RcFpzj<=WOS;ijIfn3(GCfCOFx(nqW<(CrgV=EiSQq1&d_X
zak~w)<%O;6ZlUloxtP~^zNOWrF8WVXA(<>ghO@;i3FqyhCp?@l@)%ec=^7r|fKKYM
zy8=MA7Biz|uQewHk?+&T{m3flW1x({nzOvY?nccCAcMQCSv)>2;}P63KZYTaD1fIR
zePGWXu&oR}m#N{U!*OVBY3iV4<?4fAF52wP=qVW43mPcM%7;4XZ>amVpelbwMD@6U
zuFlM(s0yJJdS^bY^ph*KIypWi#1B726Nckj#Hes0t=(q+yx?IyEBZe>rrWigYcm3M
zUT%Q#=3sqr-#N_s0A)jTrSv)$AI?pSS3mFzwp<0G^Y$i`h2M4X2$ks(4yTdsuFX@*
zp=Bo5yH7W=P;Vx$Xvg^-E|iLWh2haU<Vu1yc%sdmvGLI2u<~13Q7Z}Rabi20coKhy
z+<_>>1e=G;LK5%SFJ7;e2~bJvKYF`P;ykeb7e*r3gVEu>+@;CGrI8)^S1DYo*i0s@
zyH*`1m<?FSjw1K8utNB*hsM#1Bw#9xDAStOdggU{;Wu?PR8$E1JgIGf*&c{7N?o9V
z&!_V|=Z6S0iSMdi&6(G(Eah)M-u@@2?V~LqHbnkMFdhy|V;$7N2*g(DI}^5Bq!aL8
z;yhg$d?r8XTu4lJ$4XnC=*GWtq~1m>?#3T)U-*x7U}M1H1}akfTe4qMZ<JA1{Cacr
zA&%*1YwbcAR_>{zsxCK)g1jtTvM@jMt}QMP{057Me&A#CQaS-}Q_ZWAsUZ}}u6^GX
zkbkN!bs5lY3H%<bjSpAuPpBWq<pH3GhY&xw^g{)*{8K)-cOpyItD&we17p5K8Ju*&
zH5tmKoD15Ov<#NfH|c0N#i7oC;fa0B9E?%^uJ4P<+G)7>CzjSF<zrGVdxv^adydyT
zkT@>+(;v9rp4dC(2CGRfea>LUSbFo8-CLrz4ty35Phr;gCb?S#ged88TA;1>InML-
z(aET+Gq<>fhL3}3i*xER7NZ|vJEr$BSJMSc*2>r`4i5g4(kCb@%{MyTMM?RHE=B!t
z-fsrIM+PqTbI|MKxwI}NpOy}X{^#0aNvtK0EZ=FL%5&;*Dz)WTMP403&t&&SjLPh%
z8kf86^Tm=tDY7*g?8+$sj-SmCg<WgXaaAz?2M=zPs&Z82tA(l&zD?EdeBZR6np;?Y
z(ftiR>dZu}1WaA~PfJo(o&yUStZtbfK6ANO`Oj53=5x^R^gvl@HLyT^sqnxlyH&_@
z+3UT*<$uFIN?s25evhTbu5t4frecMDCVSXMQb^q0nQy?+!W0X5d2MUs=2}~pfz5+>
z<DpZ*#$TfD@9iGpuZ=K>;2H6GQ#IKlXp$dY8dyG*6%<ojDD9aUNeLV%I7n=`kDQ=0
zF$Nu+Uz4w$b*m_Qf*nKqcqU5^z5ms;!IQ84^hgpkd{yOeF;oCg;G7?#TryH(-b$+S
za!M+yo6FJM#PtEq!0O59%S|xFK^c<VP@8RVmJBWdEHAQVHm7#7C`^nHlIx$kX8OTM
zNd@E6Sm*5caC?7;jDVykiO8FEk(;IcxsQK4aqCzgHbso7CoSAz&c%q~30Fn*w@822
zKtFwn!_-+>DqtX{`}gV8jO4yy;sWe<qcDrFKMBicUj+o>UY5{G2=m%bT#b!>wx5}m
ziG`w_VnKfibq(n8Vh&10&S>u$;Nj+EW##2%<ltZlU?KIllAMxksjMKtN(ns?g335Y
zSY23K9qZ(ct65~=-1Q(Q$}g&}Z?A1_1W-b7OKl?TjeDQ|M%5`A^ar?T*K;5hm<cx}
zZMDu7umBibb|-&h<R^#ea}W@4GK>~)q2w)5SQt>XJ_WW=cLRM%l@kt%M!#QLS8x>N
z<I#5ybqWYq5zZYJRc61emMP5CHrbp`-}DS@SA#+*DfT+OMIvsBpKmjzb*>;A=)l6W
z-)jO$?HRCki?)pN;Pm=QGz$K(jATFQ-)ggCAV8^aG&F-vW$;K^dnpjAkxXZng?U};
zm3wbMRo{^jf<d=Ysc@_+X9L{aT?kBS%_Fz>x{bUYif8o^Z#KWLfY~`FtOW#wRemwC
z29~U6HZ`n23l33tsNFX_ba|Q~5kR%xk3oy)kws<026%8sYKVV%{y&#Mmw==lP(LrC
zRbV^T?mNNC?*0gT)aK*2Zb>el2n|2-Hg_8M>Au(Z2!>z3>Gv}zbEV-!O;i|dGQw3}
z&4pSejlB|r2p(M#Kn?#DaEsS2gbS)T9{U#}#vK>Q>Rq0TQ(xbiFZk$$w1*k1SkaCT
zE%fW8|0*F2TgmJ!ZqR!P_m2XPYvWaz`<geIn*UASbiJ$!+S!k@TgSKCeb>MgO^SvJ
z)t%xb_Q(GHhaW0hAkWvDHhnXU=UbX~br5PlY7{ed>9{y_C7S3Qx}mV_&SJUj&UCEE
z{Lxv&RD2EH-W>hgg4T0mUuQqcN7wqn6rLBI6WL0=IQNM>Np6v`EdX+I5wjg#+M7d=
z>!Q`UZb2W^<bRz>{g0z_aUVMNpI;9xaw+9V@O4e4=1z5@1C{6*w_vO|7wS&8<_F?V
zOq2(lVB0E|5Gk(|>p=ZtT@PfToZD4Xt}(n!%}?Q)=j7$5N*4>jm(I~sv%JfTY^ry<
zN5-)+9G(QP)TJF**?#@e)rX$M`EA90>HfL*kQBFetLp<5)PxSn-&WskrB$ciY-;%}
z$_(R+*pPW@#!VA`ruuA<!qskEl{HAX;O2e&_6SU-@>H~vd(_Ob&WF_6U-$*$&yQEv
znIy!1fcx8j&V2K=sQ4Ld{jzok4|=FnJ-t+0b!X0WMO^`!VFnC3E_~lOrwY;sH{97<
z_nmLz{^JZ(if1h*rZ;CWIZ2|NT2}qHP8&SfP}l{l2edGh9RaP6!9M!!;KrHN;s2>G
zMcR}?2Jac>qB3kjK&;Zj!aS5(;%?kkOVA+xA$jvJu7%BM70r}?uOvgKOjC`JeYMlX
zeU(75DKiZNtvb@i`m<6+_h(_t*($!5cjREvMd(9LvDk~&j0MF{*39wd=a#T=8kD79
z6uuJ6g{G`>wz|W4!%!|oL0L0X3$j@oi@2$qjb1bF^j^4WfZ0S%$^vr1?ucl}8V$$Z
zqYZ(>%c;oUuCdiuR=8SY;`rd{c$)BosjUE;f?b{jdSWZCWYh9-ENB`Xz0n#!Ws`pm
zArWDhxqTW8-ZH9+@|#Ov$gK+wrxOUT%0&T?V!kWA9^P{^#3Wr_G@RTg0~to^aYp3L
zyke9C;rRcWOoYhsG1mlT83heBE!o+nm3cmH##-;vvu&>aZx-qYeK0Mn*u>Rg*Fep5
zjP*O-wsjkn4Q;lO&U!a-ewn_R``yCZ(UNh<q`cfLQ_YQdY1h*9kH?i>4a}E_N~&=;
zyt}T!w;n8cq4M^c9u0~3)K1+4-_A#@ktX?R#_N%ylS~_%Ry2$!5-Kv*3spTiH7n5R
zz<tNG>e&7V>J47h+P#qEkx)QPB(Nf7s|5yc@n^$((R6;Vsdh~zPcTxGJ8!l^fQu|a
zya3SwQezB-x%xX_IM=4{h&9jEK44RZlb-NC@`pEWy4Cbg72A+!zmh~#SVUes7cHUP
z_2X^!q$Y32gricm8rwYIE@0Qbx~HTmf`jhhf>5_=X)Yw)%G#L@Rg0p+fbwP^SQSzN
za2~+?kJmOYW=60t504pz^J5dD25Hkl<m=~Gq1;Sk2%p02=@3%}yNBbqowL0^fV294
z>{h+G-qwr>n<&BXunhksli2~5Mo}MZc7r{Xc%_*sw(2@geSPQqHP^zgi<kZb9W+H&
z_;{(Gi1#n~bQ-reJ2TG!Q;NS)fCC&2@_8JZ&iehRngJfq?OyyZaNb>Pt3n;3gv5AM
zCb#3tN%(m__IuRM?3dD2cbW63ed$r_Z9N-nV<AJ`gPX$b?v4s$=^sTS*UXj8jsoxm
z=54lUs9@ztbaTvdd5I^#sPg936e{s9^S^dqeJzanhhBMPG8+b93{c6yvH-_i0vM=A
zw(oiGS|D5n^Xq^wY$l-x_h2`flb2uh5BPJ_P!q!&GA(t8c^tK2mwego-}QT;Y({>u
z5K_$zF5fS`uM?f^ZC)MpX>a|4m=eIzBDDQ$ip3F--kJc`7234f^0AhF8(Wi{Xj4^t
zzx_<?bPF0+u@(m$ZJkDDyH1UxOh-Ud6~h<kKWKx*4vI9tvUM^H&}eB1%)*<j@Qk{y
zPpz;paK0V96-JGUhglmHk`Kairka7juQs>4%V|!&Pbf6pK(AQ@YE%NtK#Ifk3^P-0
zmE~sl@Hlbj<{D@r_fWW8(^E)aeyhdrlOipJnffW!ms&vYHNE^c0s5G&Y*-DWRl?Wn
ztSE~Nk{QK+D_ViaV*d>DDd$IUDx&m6=gTeiQX=&EiS4iE$w|n<jo<Q#FM}GqSExnR
zWK%br5@5P$ypYpJ2(`1m-BbyZhBI}<Da^0_a<QnUbCVphfjV8<imw%)s4Wn!KM@4=
zP!-tqE${#o7yqH<gI!B!GvsR)Pp@-FpJI3C<2%D|+dbrrG5HX|WHh~;2=gSD??+aM
zf;^>7DWK-m?BV<-`Hv@g^w=GzdX@v;cK4^IhFhl{7S)+oSQ~k~(tuuqCiTIco%5=_
zvYf>Tncf>TrC~c=*p3;N1%k!&QBx?=?0OEv7l70T+fbs%9idF~HMjOlOen_1L^;Zb
zlb4%CQRUYo9!;+D(yn$^w6x7nLfj+VxRq%IA?@n?b2npPF+c|>K%4LF<GzAB?bOnJ
z(_Y#S_Ag;AJQh<FDkYInY<xW7-#mXA1tL;;F|62FiZ7(`$tcN8h8c&+66gufJe}=)
zXu#pzaoWYr+N!eWCNMdy@A_bTS%0r}>IoWnX3G`wOfQe9uVl)FnpD3~#PGwg8a+(A
z9sfs=$P?E!fD0$>0-ak6O82ui3jVmV(le;4EN$oDdp{7S<{|CoD0aOi&tjgvIq2O|
zE>DnK*1y<7b)aWusmv)hM!>}kKp$7v04^!XSs!yEd`kMvj9^#LM=|KrPW#Ztn6HM5
zuyO1=zQG2xl&ZfB8ly%nvXX8^H!>@r>l=N7TTk1(@anB?hmpzl44P&nzr;obCHL2-
z=lYvm2K`moFAI?2P-iAroalh`Sf921>m56B;ro%Ely_)|$fIAIaa^#`j)jf&6F72D
zz&k#5k+J7V%ku4>wA}f8nD}H9sDQ3;W9+HJqP-mWB%4LNvr8;w^Sh0ma%`~Mpp^!b
z>nqgxb-w*=h#kh&k#fz~QS%dg+go_}C`tOVW1M7R?injCP9}|1G&Gm{E2$?A4z35%
z;4?1{=BP)q@%Es|R;~6M_R>1nV8A6a9ro^Z=`{O2W1^-mxU>-2bae8(z<;?5S}?Kp
zN}T%i2R_#>`uC?_4~nZaa4Dv*o(+_I7Y70Q_U%tr7+$>!8pn(9BNo#u&PRDKcW}kR
zh~Ta}G8|d{k}P_?f{A(LqpY6}rh{W0xF_#=q{M0#W-B>OQEAm^emh#m2gU}fW{k{g
z-VAqerPX`O?cHm3qP(l=vkmuA{pi_5C$&DiU*|M@KyCQ3FssU@-cTH+?c?<~ubaK1
z^IHp33QM;8-COy@Q0mN265)wj3#;`R!o^v}g3ay9ZtWctPsC+pm~F1YMlGj*HP|CZ
zju&6_2M5(E{5jn2do<Y%1DaB;7GC1PBYq|3iRI$hN}Xx3#QGfU28+$Grs4yohp{BH
zZ%5^SqMFW3awzZOP@|BSi_W;5=m@B4bIiZoA-_Wh9qB-=q=r-%ygsU9S{^*JPb>2a
zj5isj`?rNIK(vMlFedMV?b`nE@qv<bTYBhjm^)C5^UtjdtA<c_0A1`E7kdH0Az^G!
zba3_g=T+KvjIc|oy)ATOx>lJuHY|1U>hDTs<|B-_GE<orH341i`d*&1ik|E#G}RQJ
z<d}G8vI(Wfw;iunauP2HQbc07sA!UFF=sB`O0OU0obNEfq9R{7T%h|E5hGiJ2;vNv
zQiA#be%O8Wu#7b)fX}K*(I20rfe#H7Yy|nD2EmxnKB!SSXrS0#q5m7@0FwZZG1GS!
z)}2Pv4-eIz48M>MO^n}CSXzk30@14^6t=rfIG$RGvHV~!VUgG2+3dF()>-#Yfie>>
zw;fQ$O}*X9fz<o7DvRHc726#IF&@*VQW|`4P(DpO*Got+AnJyiO_1mhRyi-G_vzKa
zneVK>WM;Zhz&7|s`WV*Fv7(8BGrLB-*!3W|r1;Qi3x{w6)3ZyCCc&5ep@6vRNrJ;E
z>#>>^e5CvI%*H$$JJb6;c+D{%xdp2I6ghf2&XnBsZN{HC4s^Wy{ys@L#*+lNE@Ksk
zekt4ch71Du5JCPc$0>KK$mw@c6Q(BJ=LcS1JBPGcDuRFOeFK3TJnE>azF4{<Egq27
z$u4=$VpA;%$Jpq`pL24OcW3{n{xi80xXDLh!$r2n{~T>Qo?wDM&YhS1n=Ck3CI{9N
ztr|1F*Mp5v+eL1$HBQ0cHOJbjS7%BR0rKGF4p1HJeKe#KH^;!?c`l%YJLC8?iqZ+s
zG)Y^iZ~HEyAs}vyyrCa?es|d_Cf{S6Z2tbMdXiz1D%)#oZvJI5tHAoQQt-76=4&@)
zW4OH1>e{}Wg%(=urbJtB@JDcKNd|WoAt<h&cQ&S*sK5w8ZgngrFmYm|Tieg2Z69yg
z8i3TZJ7HXZey9zRZa>_$*(!9;N;&?mhoQKBX~RL3`%{FE{n{KQYM(7DgUbaL$DZXX
z30oxSYt;o;ybDSzJRjz&^!Jy}U@aVxnDRh!&x?HJdA3HV9da*-9b<g^fk<V3LSzh{
zwR4*th@16&Q-XWH)qG*p81GE-dtkf#4}j1!0!vP}E?N2;ZrBTT2lb$vvCzz6=FrMa
znARa%(3z`j8{-?WA78xS#t88ncc(A}7iKF%%OtF$-Pz@Yn&fe@6BeYhOBfd%>eKG5
zVrQx-L|wN=9Q^T^-AcrBJ#s9>WpI$gHOL2<?mfe+URm|sq!-37^A#p|(3#6k0LX72
z<gwzT^-^42n}#bS|9s&73Di+IUdQ$a{<?~JCOQpo%(`>?@5(F8NAlUp8X4Y)lfd34
z%Wwo*ry;(W2d)73d;dQ_`uBLs#~qkImtlj?0U+Hj(vKL8(KND*ra;J;T)*@WnMz|;
zP<9ACHR=xeyTd^59+JE_QSnj;7GM(EFHTvU*!!7Hp>4ahkNSWP6wO^m02d>4xcJnW
zYsUlNzk?gx3<^*>G)P4$)9sRp;xO-TWtyM7_th7p?HtnhQRS(r>Dlr;t%mOtR{-XX
zeWHuQ3ARY5#nTrvDlzuO&jG{lF=k)ge*5-e5OzS%*{6F}8ly?~V6szO0$(xcd{f{`
ze<Z1bLN+eZic?Dai08)Dt4uy;r8}Y1-j4gAwg126JI9KZd;-0J)0&hU27Z1MY%^E+
zds>0Vx0y)e*WgxmmUc!H?l7ct=*-9`8Y22|^7LJtqF8x`k%2h}t^NCr?ZL-;=5C++
z`d_NIFQ9zN?G5t#(l8)y?Gz_wQASRPqmjk@sTuY2O#b%|8V4NvI2W2qVTV^G#+oX$
zRooK#kB856J=`K5cBXe{oM03#;1Wc#6zn60BSRDD6*^YR;kpo_Lu?h+A)Ri4g!b`r
z`L)|gbZ}^7qQ86Kx3myqLn~0iDv4tKAi5Ls@EDuNAk8I*gAeoyOD4+AG71>l$x<%i
z!Fmn$-*leFWrBABW4kGvXj9E0(vz}ZP$B0O|NL&KJRfYvV7IIs(z$DDo<%e!Y8LnF
z27_oMBD_jcMM2G~c7zojX0O5LbA}Ue854vFGp1Maf?M(1fFNUK=oE(#ZAvpSl!RS{
z8S*&IzEgiE0MIGj-9h^vw^8+1N=+jPKhnOp*6UbLX2qj4Z;<p`^b);jp7H{l-%^a|
z&z!>xL%pKb@ebNNmS3NR)T*FmjStXbqh)0lma<C$olQE)govq&w20)4z!Phur;s>b
zO|{v%VS3Kij^_4OYZnAUJT(9A=ZQb3X+nBI5g?gJO#p0&pBsC^YPp`CQX+<JNYZWF
z2rKCj-Eu*X)W*j%Dl+bsD9l4%NVXc5>EEP!OzAro5}+`UH{EA6eXTfUVchOJ(va>Z
zQXpy}rH0SCQHppT2kpZCzVP&u+NZc^apb&?9rJtY;@SdC0+Ot=uEJac?tLOY8$muJ
z5secSCx9X4g=)&*QV^Mu9n!q73ziG;cjUCi!}F0-bS938(%W{r75W5sbU$Z`I|Vh1
zZ}qMPE`ZOk7!cT}Rli%h!<<^I$#FaBf=p{-UZPYf+MfY3agiopcN-&CNoE~Fh0pBt
z$ZmRYeE9vS+mj1*^cNilXNh$O^(!_Sj|OQLP~?2JQ|>Q#L=|yAkJd0#9w049r*^qY
z^oNPpz~!ut;a<{g#T`Ks3BT5!+kLw%w>%%50cBNQj-CJCb|iM^;!WCpjB#f>jkMY>
zPJivTeWFe0dHXZDK1L#GF5pphI<1NQ_YWd;YjM8b*7tbXl?vypqH{{jC~6<V2P-Y$
zhTHiX?EEZMFDd7%AwiAdd0MCwdVoNqr;d2u)*CDKZa~eG8=M_Pp~n#IPXvp{Mf~py
z7Tprcp1-9wwb6R3oL_tL##l;4J?Y4DlZbk3Tq0s^{R8IaPbhyqB#m2s`vP8LX2;pl
zF<f<L`ZK$bhq%{Tggpk=KD)A?eH;n!%jRm4-AsuWY!UL-doN}6@6~!eLh4UMEB?;m
zvE5El$VJ1r-0PWxc*mVS3bLP~goSw?izJRG@eK7AG&<D=uf9}V(26X`wmL458Wr#5
zOM~O7qC<F8I#_pnhVMast=rQZyeiP4<^9qN%hmAbYt<~BS`$j;aVPunc1|JS+o+YJ
zGp+qan({Qz3fCq}&&Z_aN!TK~_6bQ9Nl8w&y@y%uCsL7?!L|)(<ZkLEDz5g+h<4tf
z@J8Fog~OVHw2(l2s?W7LpAjAmsT|z((B&__Q~Y#AMA~HrKHD|Bnv192Q}Az|w=3p9
zC^bsi;CcvwjafCh9uX=&;-UrI;)qEFL2H3*xXp^{|Awb}F~am~Gw~rOe$*k1S;YQ?
z!wE*E@^C_0zeu^pgUMi-Rg>oAbbdtV$)3T)NQh2+sVQj&$4b)Q$+|>Gm6ji0l%GDS
z7Ii3&5~1Z*y(_QjAAd<7M){EJE7{q7`P*k}S476e*)^=8jzrRcFq8mneb5Ybe!}D8
zsSFRyP8h9kWdu=9W)WDbdkX7HiFqifR4r15d73f3N{D07rs%#;_U#^^GpPklJc=4G
zB1I7$Xy6-vCjNE6i@>g5LJ7m`^3yC<rlGIG43p_<)H28nV_KT|KhO-8tYEMOTg?7j
zuC=y3j~Bmt96Z+@TL>U(VeAT>omW11P7X8fzk5Nh@nM`rb<<rx8&qfT_aLoT-}u{F
z+vv{Od~p1kZ%xWUg-><ybO$*8=5?%3l#K@$>3of5OspB&Hty+U{msKf+3(DW6N4@*
z%hW%yRemZyB3cm3+68addE~cHTCzf&MGPF}Bf=<ce#_!@nKiH!7vA>msu{UG8Sm1(
zf@Vm<#b$MIc!pmHqrxpTfcgA$XSQ2KQ9SLw7%XdrW6sRA`aKSZI}1~NUWAQWM&QJV
z%t?7Y1t_xd&CBtx-15cMbRoLN%t>GarZqP>k)hno<P<B#g^zV=`l^J2oR)CzFudQ2
zl+mdqzcxv<m>f_`NUX>*-rfVBqw8K%+j1&I2NjVNdJ>(P{yle4p6I5|o@WS(zq5)C
zndD?D2Sh!Vh0z@)q}^Z#4ffj0%B(Nl-<;g^z=p@a=y?dKhI+d-lvQI>k$}=yHi)Z#
zP){`HRCE5eHw)*j;9za50V73+9j#XU!fXL?Yf*6_N8rAf;E0MA(v>_wF^bI40bxuq
z)p1NpyW@tWv#zEhzx3lS^<mdfb9|nZK7(|#^}wQ8Sy&J8j!CRf6#nA=$KA(3U+KRJ
zgS{>En2n;G=FNq3HwO{bLF_e)`<uaGwxR-4f!Fe|tSr>76dW&ZkbjEbs}r%ga%Zb_
zH>&dD48X(iZ1sF@!v@bwPYlY~bX|qwVht%1`E5dpN$($%+hzY$@B0n=`@GW`Bd0?T
zEdkN7uq5&b*RSnV9PIy)w_*yZ7We(|<VfF7OXrEb<dgWgVDP{!Bf-zTy~XG5sSEBn
zU-51nw_PaLnok1Zp>9FZ8|PAB%M}}n#k2w)7%hfBU_3b-7d<E_D#BU>cWTtuoB9^J
zZnUp|tsEcJEh=AbpL$oP39F7yOEZL@xKc0U4@o=H@o~85)`-b+8!m^A#R;RBVjA~4
zwn4+k&ctuH5%c*^%TYq~ne}sUsXWgz;>{+!KK%-HU~z+o;EuPGW&7KK68ZXT51uj5
zVR4j7QU+I@DIfhtv9#{hSwJ(%*ZTT#>34;qux5DRa_qEbzvl@t!RzUA(^A=9{pBtF
z{N6fFRU9fwJFtOpXHNeRdYsD1>od%qrbv*bW@RAn$fxQ~p?|N1(wT*1f!jIs^kN9D
zdpbFky^^=JZ(B>K^;CGe#Sv8QV^S>u+`2SfQSaqaAvx+zU5hg5F=MxuPOWmx_rMVz
zhdFV9s~;4mKbBqnyA~@d+7}<IB67omfTi5I*O7&8O2MNG#2WjyA&EBk`DHm|i&UA4
zjPJgK9=@bT9=_S(gQwMv^7xyPNLd9$pBvk8CF;P+S0$oN?#Fh_ow;5uQhqLg_-kd8
zZ{UW&DKj2~$(hiS5?pX5&iSQ3zO<6W6b!bf;%B5i(P-bdr5tn#4Z)ezIY*=wUr#E+
z!W`VtMQ^Ixw_jEaGTPDKu|Bc7pG%rpGmJBZ2C!!hC0Et;D3$ZPjF&^4cd%Uy6znC=
zJGYCN2K7lJIL!GWJeiImM#U8XE2;jf0%za9{^`nwe7!F-hauf>JW9#DMhY4CnqOXx
z+GvyL#Kt}`wJxI!8k_}J^K2Elg#nw~w+mVPMR05KG(L^2V-`3)c4jRF4RpNSF}9|z
z`(QrEkrZ4A@X%72U+|sg`HEVK;O}=`Qmn$^{<_j967)Ie%SMq`Ic-+pR%np_ck>~X
z#GjyN_f{(VolAn$_8r*I0#>Zg`(g0jhQ#pyBD%Bn$<ui}{4Rcz<{c)?HQYrImJ#<w
zJ=80p8<GzAKXKwS&qeR+=MkVv{}ha$etdNO7|Qc~hEQmQ?c?PDwRG0jXSZ;_b!67|
zUyS>PmzpT&cn*e95IlGT&%*I@bqH0)!e(KS!yx;P?5M4J%p(6mK99aW3%*X=6AKDa
zjP};{ki=I`vRKfrs@NnYO<~LE(slSl1l;h`yymV8E^EZkuBfPpx_G0$wJQgP)ppCU
zw8s_}3F?0|0|KK1aFdO17*57(blS+CD4B`!D*v(1USYR!x>W#aeOIfI+^eoIa+FPu
zmCMup=SzqLUB!i!uQjRa)vEiJ!D3i)-qP(@xMvW5RUMVa9F@g+NOg^1>mbn1lz!YX
zROTCNje~ukDotI+mLH@voOD$k<wZY+)CC9cZc=?3Qg+Vrf^4R-i^>mG?>seWKZIqN
zK!2!Au~jn$_6GoCl0Tn8+ueaZd+<Qj@+r|olSQ5QXZ~8*CJP+R_BWx6!F*nNfwslK
z0dnK_r$5G)wniR(5#q))h%Ea7ONk#xNw)mkhNi;f?FHm?up+)^1=gW5n{QW0H{XZd
zJ-`};j?f>}Xan3Dw-gUm%AZZwmU7%5<B`8#Sz4T*L9qbc?iAYLdTouxMD#4Iube8k
zt?_+}Wv?AtcOOKFOg3SbD-DZd7q2Y84S=iOvBFUOrhJ-HP<m-aLr}zBox&CuDTJ=-
zKJK&8;AE4Z<~zjgLi%x253UF~T*`<YIBBPCLIvycKfSuveXjn14~~86FDJkBjY};(
zuX6-YCV(m(({XX1xC5?uWrnR2zN4dl+d9-7t%vo7@=j03wxO#R&<!iu&=sBBDw@j)
zRLUdkdZtJ5Y0YjJd7TS#ex>8Y*(j_Qh>m#9j68T<6tK9H`e!9lh{<*<o&9&&V&^|j
zi@ctdHV?2eF*9fLa9^JK-KOYbj?duS<z0>+7-~;~Hb^D^E_0>)2#yVTjfd;rFs6xc
zE*=4c;q-6QjJqlly#v&Al`XSfBl~7cacwCZSUY@Dy1bm|%eDla97au&l8T05J4P^1
zh$uBE*yxILjhA!l#{Mg%2l0(vhfB#A&_CRzHDw2KxxgW3>}8et6y3cN!c=X3=aE8y
zPW!i=icbx#uPYnQe{k>N4!~8)h+V9n;G5^#WO_~QFXoL#%t45+tFwpnPn#{6xc}l~
z)D^U#Xho(Kk@5Gl<X%n}YK*HJxACnDPeg3BSXZohFW>0}(m<l=bU|c$;0a#5F8($T
zvko50zk7zg5f1fPr~wh!XZMGr3!*9jJYDJ=+ewOe<H*ED{X7|J&gCARqjTN+>*vdl
z;C60u!nI4^w`@-}bpz-{FhTIk<%i%bbG_z^<%Ex7ruzIFhAj36L+T^wy&>fv2W2&C
zb~l7_+gh6RG8+5&$uYP%=<D^5A3KB@dXj=y`)19Yg%f^dU%y;gj36&0a$*gzI=s?X
zn~*X$GC1@H&TFjQlw|z%b&;8AOZV>d7%j!rDMh++;M{k)_6Lu`=NTbc?~w{8Gwg4!
zkl-WSuFI-wuo^PFOFm)zmM?0w06cX^{7x^(k<Vvj;!zKbvQ-7C+K5ZFemjlc^I#Ox
z2XIf693^$dJ!7@u!uFe}dMN>Rf))-2Dnk+8rO=Spfx^5Je<oZqv^A7MEN-fB{JQ~2
zpanFoTcxRFG`MlgZfg_N7SC|`lPtxMg!v~@sH~I#C5)owY+MGLel2w?cuL+z>gTvR
zuOb@_A~})D>fBV^9Od26%Aja#YxAhk_}YYrV@J--NqGJ=T2er#0uSWQ6I=+)-BEF|
z_Ki+^G=8J;S5>Ahe4J)P^X~rIwS~OW3fFHC-w@IM<oh&S-JP)zSTfVomS%iRDZM*)
zuRiqY6LQJl-vSEy`taOdv1u9L{#;I0adm19_JQ-);{!fl_qd%)R%0pdig`x;l;U@y
zWF;<ytjI$Za=OAYQwsnt^FYf{GUwk|=GCDB|C^4n{QPL!Sl?0*qb%};=vpf_1_mCU
z7v`4Mp5KvZ)NUqOKDUsEVk_+6K(!r;Ev_Gy;OS`mpCs&(;IG%wNo~0Oc5Y6*vRJ~7
zLZ$Xz5Axa#{2M6McUU`x8<B_)&W&trb4toe@!<EojX>tH1_9V_%PlRm{-lfR85Y>*
zx8>ftlJfPNmHW!ilD}1qv$Jw>UFI|pKe#H3#*7?=Qos64C!<4yrsSZ_6NRj=sI)qU
zM#qOI+!B*B5)(5rGGbzqlamwTUUIUEn)1rNg>nc!DfPuey~96}!I}9UJhwffNY5EZ
z|5ihRr^(CF-^T!=qvdDn+3okO^mKR4>+BESXm&zoTRJNn3tLNTb0hdgjoAo1Y##51
zj(+rT;62~s(->+ZUB;`6GhL#w8@UfZiW<f{cl6@Vf4Wim8`xyvYgwA#TwI-9(NNG*
zQqs~=P|(oOP|;CcoL}FbTZ)}_Ub!8y2?u@EVrTby_h}D!Lyoc&|FbO{*##-~lMEbS
z_O}^~>$nAwwH4v_q>~yY55I~g2}<OFeRF3npVAm)FmgVov%P!*MvqjQp!?V;PVhL~
z&%M-^X?8x42g9$}%O(Z}MMOnJL<a{32Fb~SnNgc_g?z?)_3yV84Tp6%oSsK|5J>E;
zg+4VzRoMa8pT$c~)YHGZL@TXqtg}7-bS-+6%3-X=EK6{+Fvw~xuWo5$r>ExZ+8?5W
z@oK@xJ;5Z{z-Yv;dVmiZEywM0JTR`<;N2CHiDSR{acaFWae&)D2qrl$ZZ69`l#uX|
zzwr)PIedRqju~r5gxC}IWWmh5GD;UhI!bk*rV<98)BARv6^m;1texR3+K0v^{suSO
zPCIoDxLKwA_b~n!Xx9LehKqWKdp{&G+;tg!eD&Mfj&0;eHsm5vm7u!L)V|F1YGM#^
z;-r)FFQtFz2YKYYkzd+u*8b?fsj0>8dJf3_R_dJ^ndc*Fu4*f&E;lx@xONX2Wc{5q
z;R}RUbhMNaaGekL22T+@panTmhW=jwNDR00aZvy0P`?1|K4OHO!3v<J@eF?W420ea
zvb6Dc9|^}meD_8SMiA;D>45cF*$lqfAs77cXop=VsR=RnLatx5HwCe;SwV8VV}lTk
ztQ=xWyUZNDZa&O}?ZxeVBX7Ttcw$ZpZ1L@lwXaGFf3j2Jqr-3dUAl0>OixXA9V^w3
z4)qHlvjB{&Yu2t;+jsP=Z&*TZaYb!YXW#2L_|5I-S!o_d{X2wnz&hGl2LC_eKf|}&
zmio%l!t7KKdoYOI9>y-fx}4;ApP2$#w_%%sm9y`i__XK6l}%kkZ@!y&_=r0(HrU--
zUs;l$f0Y^^6?Vhd!^OeML|0v2gzMY$7(WC^qbYn+7v8kvfZc_Fh@@vPN~)SV2gYWs
zB2ngNxSfKGL(<>lH`||PvVhTn&gL4p$tM8XuX>(yus*1(zEPB$k@RApX|LuKSJE-H
zcMFJkocjD_O<Vu?cM^~EkAs}9uf6_N(ew1De-G~lU-NW^vv=Ama$>x!G>d<9s9yl_
zl7Rz%yz5o>SU7s#h<cKqTT;`~GlCTK)6+A#I-yth0c9L7+>j$-8p~<lqU|=%l42un
z`FR7{&EWHd1Rn`KPs{)n*<ZYrl4acnb^Rl!y>3LrrM;+V>K*%D;*mM2k>T#9s^Xl~
zf4B#s0he4ItpM#?6{LlER@0NvZ6yw3%9UM2VY`vlS)W_?<J0oW>N|(tVIHBMXJz-F
z2!)4_L~{|^aX}n&gB|rqyG?k|Ww+Bd=7zh}6=g*E*cnLYd14lRL&wJ!(s7Ax);6|w
zzI^lklV>mMx<<c`cyvx`WO%T>t~59Ge{tlkYwk|AhYfYqHj=Ix2O;Lb$Rnk^>!9s9
z-&^++G7GC(2i{_r)}Lk-^w^h=ef>D;81Ylc{Vwb=d-gQ?PN4U>6Gso|Zc|*3ij<R<
zfa#yGjG1r!*4?Hc`jF`4{HpfX-%b1*_$JfeQdy7@7ZH5rg5xohJsO)N`AKDNf97kz
zq;FQN7Tln&XJ&ui_tt~N=Vgt(GZPX2_XR$G0qhla9L}6QPl^up_i#FHroTg3R+x)a
zhQ+=Cq7~;H_~sLH${PCyhK5JboBJu^p%y#l@OvzHk@b3bcxZ62ue-AiUgp0l$b1rY
z+uzH@{>Xtnno2Ul+{hFL3CzuWe+`&0!g6L_3B~RD=Ju{WHzMLQi>lkZ;p6i7+bO&i
zYgTqY6a6R_{Xkz&XM1aNBiQRwV($iBzJMNa^5XpLEDUrjmMtO4MUMmT$S|{Y_P&1a
zae6^jU1JMs7dbY8&b{J#E@*!RjJLb9qph{Msj<GU`c-LhL0(Q~TKt31f51!UPl3c{
z_ckS2F+L8GnHKvBFnPOW44gtp51FmAS3uaq<lM3<XtU7&di)(xtj}<coM>q9?QQT5
z9|T|avhc<8tR#@p`d)OhvoO}v+PYCvkc)+c7d;N79YuD_c0E(;)9!w^qn>8HC@Mv4
zCT1G0@rEL_>$<(AvA(A2e^q(u%fkG;?99}pgxHw-cW?T6yPmQ+Y`kZO>SkH7b(}1;
zfOkX^_eV$gHK5EdrDNk2lT+5(YifPU?dr|Q`1I$6Wz~)Cy+cIL@kb)|(cyu<?hdfM
z)zwtN$I9YDxZaWALjx}Z+K(RGr=zh&UQ%f7YGmY$L>D~{iscOKe|(~{iW<7cM^Ct3
zx*qm0A>~<KaYb!2^2$zh6d@7^1HGNCjdfM!B}MtU*_mm{Pvaj&MIbF2K5l1ETAA(F
z(^gZGLxh9qVK+KSOpARL$jI;F6>$7*gA$1Sh`ozfKxkBaN_IhcU2E@aWV?fpH|Ur<
zh#h%quCIX~d46urf3u9#<b>F$h&#7}{H}PoI3Bk+uuo^3GNu#<hXhGsSnOLspXr1p
z<y3a;GqrL&?-Ljr8Jn2-qNKVB*RPs+pY3UHs;ew5%*{$qej4}i0p#4@*ZZQYlcU`+
zGvj?ankt)QByed=IB`Y_LqlH&WS)YGmU#`Iu$27P9s3Sie>u2%2fzt|ywbY%0VwQ&
zyjG&O6vHrf`0Q5vA_raY(f99#2K!&}a&tLlZ)JW!f7f<ZML8)^MBs^*<X)n`0^+r1
z{bu!DhGy2M&R+_+4eHw5k{Y~m77&j#qV@N}>7t^%XDLr&qa(uLTK9HyIdjtHsJV&3
zo}JsaZr&gzf4Yu`6HRSMW@zXyfNK2FF|Fq27nV`f0I}PgI)C}bz1XxD<&9W(31A*u
z@_k5OM)8a6v_!DWMTXxF4)nX^e(tor)nO33Hi%tLT1=3Sn*%CiE=Sfp2@H#U3)sXn
zD+j;W2BmGg4G&qJx^U%2#N)KQa!jCeavEt%fEou)fAC^5?dhZYVYjYdyX5Zb<Z#^5
z?0~+mwuZ`P`SlV|6q1Dr7re(rL4S0B-@rVIrRcOb*zdP!>Kd6_J9!3%Kh7wu?&$9A
zA4Hf(i`Aa4j;5OO!si)Dad3OT?swVC4R#(MGSbu5*rKpOT3m>K?HV>_21NaO38@T=
zeGAYBe;i(5(z0-_gU8*@edcy9mxAuaWfYcIWAAmiwnbZAd0`HGhqxJV#q+$=Nt+|4
zMtVCnRW@&s6kf;6#m>q|w-OE&Bje<lT+ok>?_2oyeCUA=W9Jf(kXP2+ZFuyQM_|O0
zoN`dt^gwIqakOz1lx;1wAa?ldcst1NvgZZoe^d6>=EfkUQQROdz{|Y`)eu-li4BI5
z$h_1yfhhqkSO(8K5$TP}ItQ$ryl+H4eGc2@JNrl9O}~FTj^uSsmGD;fe(3ef?k<kD
zM@<d)>`+sZmqv0q5)nQ4KFq`pm%Xf_+FlEEIHtb2trO`?nLy_&TAQIl3i1GT)!WVa
zf0W%ZaJuPeC~XkqU&FMLq|?~H1g`zHbOk*sXQQrbVs*yD|IWiFDbI3?z-JHYx7w--
zn8~L<eHalEaQULM!||hs_V3lv1V<al;h|C~DHm^l2_p8DjH}_m{AMli&3j%4-+W#{
z5$q?e0flQt*-H>Q^0h-Dv~w`2F*^W~e;RcqISIkFZ1k|Nj?`1RuK<;wo`=Xpuyd~y
zk=hKZ8_ScRvI_~1jEPT7%gTG6m64kCBrX=V7(nSqkkK48+N-l&O=+Wy1XN96UPao)
z+g|~XeF-%kGY5~LxQy~nqazL%u7*TCjEzf3OioSDOixXIngBwNx(j0Wx^Nare`SsK
z?%n}&SqUK?4klX2I|<h>e+Arf4Qh8WvvctwatQ`z){d_3-dFsBZr!;T6?rc_G$c4E
zz|Y6s^~?!taIzzTD{YjO1oaCSJG_`C5w(l&!%XZe=@H%b4Jzm?=vDu~>o;zOgofP>
z3k|t><9dMKRglA<IRU<PV*|Zif7+UA${?E+gMK6QD<JO+Nj!)D8n71&LeEG@3g$Tm
zrj|B#4yT=6FM9fTd%C-wKj-Xp`UG6=hI%?$8e5eVHmsKr5#ZxqgP1yyjMT^XVaBL{
zhJl5hi&s<@R5nKDHjZbUoLyYcySaOS$X#8|I-NdckALc#>MBZ`<YlBle<BaH$Z4ql
zGo!vAj~QPUL<5YCi(f=aUP(=Bx1PR{@j){SOY38nM-H1GI%r~Suy;3D>?(>IH>{Tw
z6BgiG%Za$SKuN5*89Wg?CF)7TBP1!Oq^_;E|A5IsQ#13!7Dp`(o12+}(DxhYf!G1(
z3i7hjl42r)pp1cr2DqKYf1C^bU$~VLTGq|BYf#*xqN<^(rLD75d&hRoZ5nE-%8Hvn
z*b<^b0{lGOoE&T{Ot5f8j`uklJP|w8UZ;ZsaDrm%<rTN8sjF*j)7-ALLkmW(uBN(G
zSrMgNSa2O5oR?=us&;r|fnHe5z1aVSneM@s9H`$0ZMfJtxOn&le}sfZ#Ka|~q@|@K
zCB((WFpU7f_-a-bCPsQXC=Y{*f&GVuw4?wmc8bMID5;jy&@-`daPtX>h>D7diAzXG
zg2*Mr#X#sVc0Rb;SF^D)Gchu(qC<r;khE?NRPBEayd?+mTv$%Ml9p~2Jp%(H6Eial
z3o|nl6C)!d13d_rJZ9wz>g7~aQ1KC}eNy0tP2b~ANNPaEZDCgq?8sV4y9xx($jHP5
zf(MZ!p%bx#;Fp2mq15SYsn;JJ0+Q58lQ7f;7cBlCIQCY27#+W?00000NkvXXu0mjf
DZDRMx

delta 53889
zcma%?Ra6~a)2$)6ySuwP8+~!N;GW>_4vkB2cPBs~xckNl8r+@W?#|)6IQM^#zU<Xw
zb=R^v>#0AbP&?&NwenCvm%Bmg>H@6KeKQWjQ9{BwthG3L0y+%jpco;tDmpYI4!RVK
zxSHyJdeDhU2}$Rlmnz5P+q%G}XI0n>agXZ*2K%J>>s_ws{V<`(O|I(zSKd`T@O!nr
z>i={D|HtS4uaCWbEWSVPPKol_Ewwt2HIsl4B0Un-R8`g0)z#G0G&GWu6Vb7DcI`Kt
z-N#B}iQa+NUCWO-O}>GLSRy`CKf0d<Vx4g38VQN#l#~OTZQoxOShD%vRGv|VHXFUi
z7VIAf%)Yw$>a;yR!azfZg@#E<!g#kWWgJ@5%RxSan(db#52oI$D(xdLz;|b0Pn#_e
z9T}PEt&ys1WPWytJDM;H4gmrN<i<Tq_WOa~$1spj6nL`gc{)J%)52sIBP~HgU8KCC
zwAxwrCUON*Kn_Dh*+P&IQJwPjOrQ7lbhCXW{5h^AIWQ$WF;-jWTYge~{iyh1cl&h!
z0fMQBvR3%|SF;XykJp-~_hB-?4!q&Yt+>`V1IO)#GKo)+ou#Xj2Xge0e@K|=?)pMJ
zehYB#4Lq*$zMtlItlZQ2ZC)Zo$ysN%Nd;aAmVLG$B*@Ol%FGw2t_m>8Z$;-@&09~Q
zhF;}+`|L>ttkcK$U5hP{-^#<~Y3d@sg<J3RpQja+mh2Xxi;Y&7UdfNzq!@O#7g9YJ
z%FYSVmxt<Rdx&l<^uz=Wj+-AB5#1nuz{uUB&EqY+@<lVdA$%CMa^wtW>sw*j+Fj2-
z(7;=~l0M`)*i#R6`Yo%B`D~r+Q8)k-pO}#JN$BtJ`t~b9ZV2bMS7Ki)*{hb<Q?kyL
zX6G>vKT35(Lc!`T#0Zt#CeH`|!-o;kEo&3i$j#gn56h`T>(g}XeT!u!>DoI>;A26P
z8~!C9a99bKk^N7(D?8CxYj@APtV8AdE>@BT7pLn6Zpo0)UAxC09c?M)oc!FN;7j0-
z_W-X0Ot;40zg$S8)_Ls^egFpE2P2074G#;I$mul6fkNps;g6X<1K;w<mjC^&AK_c?
z2NvivN&N|0icd?rmKT;$FoWhmDi6wpeXekmca&*&HLtWD+`j>@OlyxfHi7u1Qv1h_
z5y*35Ot`zLHnB~lXoWGRO4}2{A6z@Z7ac~tq8^8VDdejTNoS%jE!9PZV8>o6>Ypnp
zGyIuUwA!8jX~+xuK1U3fTHuNLk=nAz#d<mQKc}Sqh+&5#If~1Fi`HHRcrPHjXE|Ii
zN3F93#Wq3hKF_e>(nGYRy}witeYZT8_gaVEyVhd%db%!_Tu_p_gQ?>JQ-0)yDTsZQ
zHyRr_Uzlw@J?`#Jy!y2q_CVI7=%HoN<+XEU;;~a(1U*IS(wYCZt8H_*3!CJr(*j|=
z#wsspp$y4!^tOV(>H7i)L_nTuaJfP$(GakG>8jQxI;j{%tW`aKUGDJ@3ujs^NM&@9
zCzSROQL~1bry3qV6DF1ViOYi;7o#*e&JytWXkO8=e^zNJ!uLD&?j0}3SmeeERYqBE
zYLtqZp{D4UrG@7#+p)W8Kl$UzFdgLijYL&yTNFmTaE%c=e-ZsCpd9ki0K6fdp?R(J
znnoZH)1Z)0aS`Qb{GOp$pS93Z&Woo?a$#DP9-=F=#M|(Hyftrc=_aQc-JIQCJL*TI
zJe})lxEUEnRom<&2*0Mro^y*~Uh7jwv{|8xnG)EzMp7Av?2eCBQB<0moM4OdSimhE
z6!twdbqp}jO!DOedZ#>0!tM`-CiEdsLq8Xn4uXv3wx_V4Xv8z1@7B9}+eEiQ0%gDJ
z;Cw`S9s`bM(=*?Pl44rXjkWbP3~`)US2h>CsT~ImJ!H8RrUoWi$SG<n{AcDy--o&a
zBSU}Bgy%*5`DK0C#>U)DbYaanB^N-t=exU;+5hrL5E;-6pfNRy^4TwUCqo#xawC(M
zVyrOHlvR8;@qWrW=ubF^B6z}y&6U^!f42X)yqZm4?p^qiQ&LX%kTaD(ld%1h?lV&F
zAe1nYUJ_VduYtWr-zD$R`%!wbX}&rar;Y<M!t5jsy-f~oE>Hyc{k?xqOy6G3V&<lx
zBM!IyM<3uD2{dCdHKJs<Y-6B@=;tM-5=ru0j}i{YRMl+wpkGt+R2KYh{%PyvhwAKi
zw0pqj+JD9abHvt9+O|q)`v-yjy|j!p<Ru^e9F#EnN{YX9sx6E-`Fht8|KF~9Jt8Ba
zAnXw2A^7QmOoVwN#*I%Cw^PpNePsT5Pn;voP!w2=##sqxsl}3#8v~D0QFp7z8NRe;
zPTcvb6eYV>9%rp|z)90uZc@|&ueVQQud><_*2<&N3BKAaci3T>0aF_BK_pTCAkQB|
zB~y?yMuskf`;|7gG8VmQH1Dd`Lv|bB_Z9xWJYBuVJ^u(E-%IGYA0kY$D-m6d#UcEy
znFxGY%WMmy-qS=q$>r@h^%3PiSCX3rUmo&D?&EDx1pv>t1Q4PNtLmDcA1LZ{tQoMV
z5kE-cQJioY9-5XZ^HuBV(wWM(Y&E{m*gTcD$^8x`R%N2=C3MNsBzl&r%l<A^N`M2+
zkh#f>|4%?O5mRIAss|=j6C<SaQ;5uMEHJs!{w(5~sy{Cf=SpB{j7tkEcszEqNe%10
zLQtVloA`Qa{9zdWXPwUu9NI)}c1!%<jC{nM((l}hN12~{^KN}Y7Vqm3@%bN7b!^5W
z<4VX6&3p?fG;rQJCx2X%KkWG4)3!Mml*oOTJI%*%_w4)W$f>`GlOW;sL=K41pa8ka
zv<LPz?N89drFeX|%0Mpxm^ddWo=kKy?v}fsO=LeY`|j7~es2F3a?U0)iHwo^f0W-a
zKj)X;voNz=&d5Qvqu<8)is46O=xKeaNJ1+cSZJRsSYYR)Z_Y`@)?Z4E_i=|MAe}8h
zz<Ie23JoGIp1q>n>Jb0F_69L(3}7Ukn-efKv$nt333w2{j9&ECtP0^WhD1{})_cC(
zew!2XHI(xhjkerSwu~o1e`4A2;<gRha2zIAmRFYR%FauSN)5?2mR0;uPg(qop8v*!
zm2^Xe(#RZpz7Z`L+2Lv(4qhmfB<JZTT0?P{y}P-$BWB3O5`pzAZ41fo834RO{5lZj
znV682s5Z^X9#qT<fxvsLH>{kU*Wos}JaO}046bcmY%`BFq;fTeLL&w~9Gp#mm;It-
zJ2P+XiO7v=?%Z<3n1@<Zn}a;u(kd$`3D;y)Hi9(ekXwF}w~1(U>&@)(`uAM*5+72G
zwTb<e*F9crZIjJE6kI&n0|0yoMs4*i)-{x520c#y>83ojw4Nm}qA>qSN=VdDRiFOb
zM~a+(X=L37_1UYb4C7Y6`MU+-&EcRzwYn$*Z*%vvH!P$vu*4$d>5XeYiThwCLZV7n
zjc%?8_1J^{-W9qhv<GFNAP)-(6&4X+&!y2X{~wHmGH!P$ip*?F0KdUQ>%--rhQsul
z1|%jT5F9N#0y6~ZAPK+d57Otzz}5RwnFl1!chr`3(f2iIC%VuU&}B}ZGaA=QnA*Z$
zL64!e(aGhv^Zqi5?dduMh9g}^g$lye7wNK1e;A<w{{RPz_MV^D_VS9g7fPU;?{G8$
z@%1!9xAgk>RkE(v0Q?WRKMcP}76hD!Zr7xI%ZRo{Uf{xnfy|sdLmi{lwPm!N2wh@D
zlU}%Z=YZO-Qg&gHwU6rAr)0vJPPd-^!U#>mK#%|qi`QJH?lOO?S18TD6dreF!u%#@
zx2~HlIUU8FliTr}*P@xxtfA)fpy^+255`a*0nPy)7B)p?Kp{1k$Z29(!#n|t6i3`<
zky;%jsn3^hm@XyM6N7Y-IgiE9mgAGF{01G~i;JTCocreLHWm)z*5-z;zEfLEbFXkE
zPLM$#w-`GjAmRDNuW7mX1Wp$#QpLscz5+nJ%Uw7N2p;0RLtLzX^3TK=8$D%Jc`3@5
ztg%VHE>aj7phtd;g{9{AFIzi@lf!c&;@pfpprsL6=+fWR!NSI`v4K|f&by$cvTE5G
zMSw!9VzUqwkwz=`he9JKSzGn1`h8$~b9dq3Ak*U38|kyqyStWyxU;vW;dfgZm5;xN
zd*iPi@k^sH`+rqb;ymYZO9r8RHfuRR)!i3-S8rrQK+8EbWyl`lWfj!iAaw~g&tK{B
zx4^%6a+zV|1xH6j_{5|n5`vsa>_u;ckAvQkFRHdu%HJ|JS%zI<S4P$S6tR0GL<6Ww
zADVs9h!pkB@0o0m2{QDJ+X7bkwshV6Y&_J=C0uP~v@gJa=@mB*a82Ihsto*|pizYf
z{7b4h0XmA?ppM*Sgo(|F=vrr@cUD9_X)RVGND4J)1_A*X6yf3<ZG#J8)M%wYtkw2s
zL^6ysm%93_i_D)fG<ER#=7#Tr@eQi0#+Y+PBOPz|8~F=~4n0xJPS9$c58tU&r2HQe
zRbjui!<?!ymVCPHx<8q<l$)YKJ1G5LC1!X3fUzE8pTHMYS2-WWz_MN4w|@DV|4!XX
zdbsZp_iJnI8bTs-(xO6CB;~b?)O64OCKmgITiWLqUXaJJRM?!8=E0V;QPix6_Pl1Z
zuuLk&qCGsgef>u(cAuD;7akl6!E{zL4wA@rBv0(8qAV_rh?Q~9d}AbB8CE99Z*X^k
z1JDexTX&I0PgZ?o5&}L?+{Vg6uVmndqpQTAjme4}?WuF)9xE>$U9FUesONNA{r8}4
z>@)Jx63dWALLMcS_1&_`3>~q}3-Qfa7qAjs@FM|t&eN;XmMF7gJlcbM6EE&h5wzAb
zulvyU;1qN^DxB$>W1$&a8NF&Rkf+)-(BOphV~nxF=J+t5L)aICmIEy>{Bs~21d@0E
zv^7EA>JGdo>wzVGBfB^#8;*ExUkSXJekZ-(A5M8V`)679(L^wGsESM~LO}xsw})?)
z%g)X#YH9B~c=XJ2AtR1aCr98I&2*yg8&12=ZhFcW1c>qp%q%Xe8>f<xxqM^+cbQbz
znE$PfH=05B(AJP1weRc)l@ka+4?Hl~9_Mj$i#oav!c?gbcKYOIYOCum&MfnB3dGB;
ziU!<`6n-q0sa#mlg%@6=E?@O-u;MA(j-KOenLQ6$B5sMH_x&pRj+Y|jbeOUI6rE0D
zxaP2acjZe+h?5Q-FEqXZ`?=)-wXFcw<7eaQwD!@Wx$<FBkJ5JeZ(nl8-BJd6_PG&O
zLY5oNL8OX1^aKK7C0oQ9VX0NSP1&A)q!gMJbK@;iI6SdchW`D#T($q+x7i>k@sLmR
zMOc*M?>%`J)9||E{M>Hpf5j!yi(cmL;N6jY&rl2x4%ZaB1}7%(^RfaCEs4(td(qkq
zN6SxxfA1!Z5?Uv_`v0+QLcEjAB>tu!4InuE=XV~Og^kBV#qy8!)z~EfdDa)I8ta_0
zxa8*ndYamTpmo5XKX=WG{d5!<Yzd5-YPy<+8v2$OV0cBtO)8AZ;huZS#Phx3us?<8
zX)E=?<E>y{<pbq%g%4Dq!Ij)UqIj|qdCeCGHNUae_3j=jZ@5$b&&VG(#>(IJR@N6o
z&SiCwXMBOiImE{{6^xT|vDdCGJVKOgBVr0t(lH;l$}vT;vW+f~-@I`l8O-fNT<;w=
z;`2nJe<&!9{a5kzPN^za_zmot(8LxXX7cg4NPA1yNFl(tv1w!ss3oKlU;Uj}kW~=R
zi2C1@L3cc<5n@N=B%oAhOc;SboBi8T>w4plOOTzA1+&roT$<KA>0o#KgPlRt)TqS~
zqKAEY(@6W7_f7LsxUAuGSk}MwaqTR@-Ztk-I`!qo+ScqO&K{5mlZN+Y*(saqDt%WV
z8d-UPix@`VG+hDW7Hr?Y2*E!5T%OKLJ1#m+Sps)xPmyY;?iH1i)6jxfo1e6WZgu)g
zFV-+e+i9B}x(gxhUT2Pn?l(+jiSlx?GNm#?mkv<GpNPQ41KE60)A4Ke$UNrWo)=5q
zXWE)!V0k?|I`YD`u83HDll&nBh2hnBs^rZ*+|)&0&s!iiCQkhP^mI+A8t3LBe9+(8
zqP(H-)c$zv67ExuZq=`b&<ESCzbZukXSX6QE8Pzsr6Dh~HgR7+(Cu4IstCW|Nc<J1
zvRSZba!LfAign{j`PUsc*t`w=i=)<SUuv|5!yrJRB5K((qA}C}#Sg3fk|GbiI3~+A
zdVGxKL;+Z1J`la2AAbG{YR>B8Lr_z>;MXJ7#yBWgflXBnU?IRiWr;YFPNXCQ{nlb2
z;c54p0MYXnH&zs)sKfUhcbdxLUTpS+#4bLy2_Wa=2Oud}F`OXT4da^AN{T$WgOV~X
z{J2&UD2hwMJ1Nx|ylLG1kzTR`wh;tU?_`g8AAoaPH_;+!tizWtJ0RAEm&l!NZs8>(
zjZhdPZ#K|Bwn$K#MaZu)*S_$MPR>)&tksQ6M8CWw?e)Ja^JqCA?Kk1~D+d;qrL@Dt
z(%})X;pBfXw@G_#7BEX@(VDam`WZ0!i2A8qdUlw5$?$&&3GR(US6(N*pqBS7oWKn0
zqX81e-3+j;52&$9Q#~KYlrl>+wVOgb0vsZ|f}Al(-t6~RR1n?G^jsW*m#nalEZKm-
z7DvA-NUoe*Tkq5UsVKd#(+IaXU^N<492C1gynMXvi>L`4_Q;oxfP3+67hK5_pId~Z
z%tt6mcp~+%wSd!Sa!oGI_h~cW_m|z*2KJ1%_JioH9zIXQ3Shx?HGinth6NA4b7559
zOe59#m3a`qAcAFpbN;ih)0Sl^9Ua)4jQH^6?EDx^D?lb<qBBos87_GIqhX79P!rk<
zaWCZKXW+1aN>_w}th}1xL1W^wuI(z*B%jbZT<Bo;BumxFVK~&3Ycycp_ALt?coWXG
zpFnBSTzcPQzg`X*x!C<wxcl|LwYTFPKsv{ja#ydx#Gxpc-c-;vL?le{k&VGwxQNg5
zCwv!=Ffw%HeiiS9(@_LV4ntiIwTsBAWXBh0=XQ?v4mRZyR`a#LgRYOn`Pq^3+LBI7
z4tgKNRM9XA5d%SwTL8mWUB4<2v-Zv9*eCpeB&WJ#v-@Bd4;`BzJxNzpZF3`DZl&i-
zz!3@99~-<eJ)^3xtM|yuR8#)D9daX@{ZFpG-^bOMoAGx!AG_N4%Gjvt>l%K}2RP}z
zP#->`PqFT_H^H_HG95AFKsainnV&HH>z>g74AdRs9+c7zhr>O$_woRXI>s8BMtWb3
zW}!3jXJuocpaLNLEdAv?Toh~5vAzgB(cXr9gar<xNYq7-*yaRUcNmLpHAEgRd6n=+
zJN*D-h2Lg>_R<-53$+8i35VusnSi0jD5SHjsQFj1r-_#x=96Tp?+KEzjA+}5+jdM^
z(IJ~mXHdH^c`e1ylAnO(PO|FU9ADR=+t^7EpsM3$K&faVF1yObjcS>Tk=JX<9N}c~
zCFk)xZlUx!YqWkRxHD&5i295m&j4)h<GQp>0qw;BAvEx|_=S`US!#-|IACXO9;9N_
zsVJ-LpZwNbTT+Cp^)q59eVEJ{7UU@5pIe18Zw@1H^yg}-|2tsRH95*wX?=W3fKMv3
z5%w!@<+TJ}6b%%O6|on#98h;;wUo5nnQMXv!vcjgXDeuGeX);4{QhdLuBT<>ETtr&
z0C(L^MZ%8^U>2%OD#rz3ZEhGC8>;DLq~Fl8&`_|HwN}&>1H`M<5<xN=x(BY$qu+j;
z^$?$r^{W=i`vADw7cf1eY>W&w6;cDX&wqEQvq~ta|NGrLsSn_RySUkI?kY1b$(NcG
z0+O}0R%fQU*f}|%D0}}@9~+cMI9;XBNr(jwLs^5Mkm!1lWjBc(gyR$^KQ}LNdbc-x
zL%7q|&2)frPc87}_R$H@*2j7Vn|CVvJy9}>@RYTd0zWOG1IRBD1g4*+&!yr0=yAOw
zU)W%UJn-e^XGhK*WYyI`Umo#W*Lzr}-jF)>-%bZqRrIOAZ%Wr=BPKxpsz^fY9c@o1
zWb%h!AR_&;bC755WZr^$((63Fwk(zh663c%HXpI!xpNOOMQL3n30t}u@0a7>--iop
z_ok<VfOC(_KAZz{%TDD;U(%~n>w&Sx7ITzb39gYhh@glD=zoqHyZ>Eukb{$pT0exg
zK6G#$D9EWdq~xW8kXw;{$en?{nfY>UNm<|o`g^&3L73oZtY909Lgv4e6<eLxGlto?
zZ!B3FNeG1VbTo2x>Oz9rj`6$f3Iqf*ZdE+A0P|s(byvaY!j-pEwgBC37&I~xT(RU<
z{~r~@ET^AlAML^XH_k<G;s$tvDDl20slB)h2gmA4MwSu(8$k>cQbO#@k;SoDHfma$
zmfv+m6&nS5=p%UX{=2D?x&eZh<l&pJu6+2ToM3Bm5qg^fFI5QzyyF(hQfzK=8>L!I
zAS6~5^3rPhTY!U&TWt$er-OMI_$G9z{4ZEW1j2oIOe>9dvc)f?2#YJ4@mQ}NcGzMg
zsIr^!#XKSXrj1KRv*algc-D;1Mx!Vm;Ad;q_URd=NNCIV?EHs!|9O6C>mu4SPbQ}_
z?&CyBC?i>@ppAGD%(7+r;leeWwDX-3$cGULINberwC%CJyj`mD_h!+NNgp<1TM9uV
z@F}s{Cu34v7`h9JQk3j5w)P()5B|0ZJas(dFKe5VT%qqa-rL7#GTTuCI8{rCYaoLb
z0{``Wi!!>1EeI1+5552F+;mjI5*Hx88ql(_prff>f#Wy-`Cak9H+u%5DLkC907U)`
z=hE1}SGF%gvKw1o@H8lM4jAYd>51Cm-%W<n7SbKpZqR_c(sEo1c@@Q}F8#bKSUjOV
z%zXS)n+n!RC6&;O7tzb^UFY4b<dCj2sY+LJoOUwoY%Lu)Wt5~_;FXha2B9pXD9_7?
zV^I^^^t}z-4{ZT@XWjESz)F+^hBwJ8kLyEjVzWhprsV{FqO9-YSKRuZ7JkIgAjQYJ
z@%4J>k8^qTK#tce>E-fp30BoqR{biqHFLjP=(lt}auk1F73OaG^noh{=|l!zR$wGP
za{^zOzdnlI#r9J>ej2K#0;K#k9+DR+iGwh*wQVPO*Tid@^DO>{d>xlS0K#H&Z{h%m
zX!K@ghYG6jgfEAL)mX^V+QG)o%1QnsW7(W<)ACC9pM3`BCg!Qe9B#Ka&NeVM`Bg?d
zBNH0s7>N6@;+IgY%SPz5Q-9HHx<z4O6KHMqS-T3je6$|9361GQL@!kAM~8_zc6uw1
zzreF78<Gis&SKi=jza+M@0=Q2#K}bq%lvgDG<DUCwB(=4DJZ>CGZKxBHaXbqF3l^5
zB=0C5YV-+?KE3gJO7I8sULt+~oY`c(6J@=f)BV(O{q9}1r#~57SeNW51cIC`+I`|n
z?V44@Pey8INBS|jIMmy=Bd4bs7zK@fjbZmV8<K@NRc+z&-?o6y^7?1!li6#fY*t+*
zBA*q?^E;~Q?gJ=F^AUtFhaG6hi_Leimzbc!h$mMKJI-Dnt(|n{q904l&$4=Wa}vLt
zbv%8F#omp{tk;yoK+vR?r83Fl{LEJk$Vdf+a~vGpt=&Y`1&(&NbtZORn#%I_CP(5g
z#>vr%EgDc@4FK!)YBoW}e>ma{{<3zAMCgm$wKCtuvk)qBbew|x7%Tnf-T=YDP;C}u
zE81UxD#@GsY2eExzI*0W^`exORsE_iqHj^rlGc+~|Ie?cJ}<04x6D~r^NuCMVc}F&
z-#x&_ZUkykL0a!RnEq0fE?0uapczX;{JA{m$GtVuDq!pRn%Cu(n)3{+UQr=-zak95
zTJbOAyvAJ+eW59(R1~e%eef0kI`#qwV{dnV&*Z!LM!&4eFw3tZO9#&zQ%OHPBPork
zL!lzu4D{m@pXt4fXAufsn=gCgA3~cg7)i5rPDl!)l}GqGOTjv~48Ny;n)i;ANZ!1$
zXMK1k4FWGuM;wQOzyEvQbg31BjPHmk$zT^n<Kx4lql1G3b+jHw-7NZ38KrAc4R-57
zyqf6av=WDWYI3jr$egXbd?caAS&*!i?jtyHk(1wZaGmc_N>5x)%H1E!FDHQ*m70ow
z<Fr0+81wh+ok#OrPb^ydcgG@{so{y<(Q#QB1wcw!T3)_BIwLY6Ax>NG?KAV`+dGw9
zuW-CwAcXs<P&z#SFLc=oK0wbe<hk1)?^HtWTzi&B2pD{OpDiC~JblHyO!erqnSd@D
zilmul;bPvHGL4+kty8%OlvR?$CiX;46Tan6;onUWF|+anv6E`Cf}Ul>r5n~V8#({3
z0dx>SAG)caxu@Jc<Q&R+xF(fc*Ume7&LewGnJEZq8?Uzipr+pe_HgnK(*PT6xpgi|
z|C~>?-Zml|iQsX*)b71Y#!vMPJz<u!-d(0Wljlsr)wmR6`54ZD*W#J*y6==bRFuF=
zXlDKQ>(0x(7`w~eLUHZmIOgemjKWF}Ab6_dG0~A(C%)xl?kvCLPlseEMd9PLCKu`9
zuo*n6@37qNPK2@<u<p7@Dl*0TyWIAS1UBewTucEbC#mbi?R}0yH(<z6l-TPwt-sO8
zGjXO4NuGgZfv?Hq-Vghpv7xaJkB;XRX4Ogh7xY+6$Z}>~P<5JMLa&9SXs78<poVhj
z8U>aBlB8z5a8_$kOg{U%x!&FBeoWC=LQQXc)|YibS|)X{?I;pxyaBFq@xhaDrlRH+
zyDkS9LV5G>;sM_*Fl4eN-yzx8vYwvWmNSoz5;yL{M|aj@N>uZ0wdm%5w*F!m@xTby
z=s}=Dg9exFB?;KM{8MKCZLr`8+{RA<;o#z|I6P%4<5#rm`R~@6kqIK$$p`7Ge?L>Y
z8>H0Lrfgh@6|1mL>izuxFs-8hIw?lSl^!RxIEFrA+dx2&L&!;re@)t@d|R`NK^;S2
zhgkEpTdlY5*N#I!-A_}WZE@>_3JV;~hO_7%N99hpqyY!KE@KgLzQ#}kL_qn&ShvIw
z_a%vtbHz&8_-~T6uD}~>$1$QhdpBm&<9_Y6tgE`|?ZTk<Zr*bB^lMJWrI!_{ekw#t
zI|xroKcQAdq$dEau*@V~734oXe<6?}X?;$R`@)S92&e5|@W5jm(tjEIF%2$Gsb`6N
zrX`BsGQwZdS!rJjK*t8|7=?R^>U?!=9&coHDj-Z$RHh1GS0}q_KXf#6vhR<*tpejK
ztD>rt_~}@JU-n~-q2g7>7^=VA5N;;4#v~1OTW8D^Bvxgk$XbW$js)1L5dKtg*ocvl
zR33x(hfAD5(5Pgp_IxY2aU44~$#igYvDkjQ_CytzX*CPM&yoVDu<2q@fSr>gGxPFd
z=09Vy?MS}s6pDWE!UWOfGaL%T9pv4A7^_e4vx2}9I0@wWM0nehkh5s8q}sW(B?aHN
zA=^M+YZGjO>zyCeU$%9)18ouLcw)Y5Y}>W=bQJ$AEUm`|lagF7os@EYHB84fDSd@i
zBg_w%I1OwrylVn{Rlci7AN!V}i;pFp?j49Hgca4N=YG41FCk-jQSQ!$srUi4q{M2_
z=6paQ&&~i>!y!+i^LlBWDZ|IZ5gv%Di|an0YEC0iv)h;dxaQJk=dGtNrI!$!2H#6!
z<O!G*(k)Zb*O<HgxicL>(rG)(e~yIsc$R`U)3yx~%5wvD(UkT+9<d~7B{%r!>+Axq
zDqh|KFI&HIyPlC5o;U3-eU%8oKRL3hu4HBz`6KH`dVQFey(@K-tT)Q}X_QiLk|SM|
ze8Wmx6tX^YI#H8!39x^dOAm97DfsvOLBW!{q_w>HRkoCYQNPI^H1ZuxgXRyHF0>-|
zrPt3Ig%}F}^J><nuOo9#qOrAE$kx-3PECw@;Se19jx7((h-n*&t0#m6eOdEBe~1Nk
zN?7^OXK&;~EYC;nd;X?7$`ks~rj<VHS6-`sPy8~e24iY5L-sme=wt5nn4;LZgBTH)
ztpjTJ7XFm?f5Ly+`l4Z1^~$02UlY5uL7j;IuBrptf0KiR8CZq3-X|*YCRD5|&(qOE
zPeQxJNd}6|FcLID7T<P{iJlo-+HJ4LcJ@pxp;{S|^VVItXOnh+E+Zy>sO1aANg_TJ
z27!l`5W(;0*5qkl7kHb*yVkA0CS2|w#FMeNtdE$Gwf+OQ7#SHPWM`)3pyNG1o}+X$
zSh@l8*d<a=OZerHo?FVoTJw)8|LIM3Z8bNzv=O1FCK=BPghS-#-*X}>M<2f;Z%zL)
z{!b4T<4J;!nYClX7aKDfS;2P6V}MfG#_MQ(?e_6fOa-pi!^$2-bI_41X}9RVSD_li
zW^Vq9c9s7kvs2@eyoUMKz{MszAvYEWxfd|Npt-h(vPlFjQp2}bhnz(X?4>t}2*h`=
z&E!mN^a#leUrQHc_dPv?uapFmbRC@Up(urmuq@ZNf|iCV1h7$3ehBw!zXa);OS_fJ
zYeJuqIO)A`BlZBO#n1hi;}^3CPvu{cyYHpV)glyE%Q*JMWDuA0Fh^GykS(VZZa7ka
z&PV|Br4^%uSYvdMik{|X-v$&f)dsc}&jV!<k87(kGP5o-%cNoq#F~Cqc6N3k53d%^
z0Xau*&9J6z@^`|QhwECdX|-g`ZTlx_6bCp0s+QK^((b3!RY(rU(beHZj-&I1@QS38
zf>mBR&CsK5_3a5!L@xtEnJFgsOa9*g1FTHHS977Ol^d}};_#hOFBc!&H(i}n2m9Z@
z=o!EsxnwUJZv23>zOH;AK?Akh2_M=UsZkVEgo+;j+C7kGG{gBCfcXV2FRCmLb>9b0
z5)itSeelO)XsF#$<Hy8~9pihqzdjj*v$L=7uCA_bZf-6urRL)iVH~;0TTKOAVP#T(
z!swX%`vMFvI4Zx})BQZaPSu#_Xmq}bmS?jkiZIWw>&?60ZyzszK@C6t<|`)K>dxgy
z48&Yy_nOPdD)v#A6xh2w?NePg9i;ZdAo45K|8qepMdFF9g&&LFu=q(cC>17h*Ku*=
z;OKpP8Ne~QE(r}v)CuV41Fr(g^1~F&l|}w5&+RHYF5<i7(A}_cD1AXczRM@LK1c)(
zE5nLkv*WNpalLFn$;7gCh$5>5yIC&@zQkZ<4bC%y50<rzHIvdapY|uRV`5@%ZwE4J
z%h1l-1(x;lM0`}WLX#nrdjnk8+?St-&?wav^oG+AG6kmmYY1&m*P#Fxa1$IU(^kk?
zugj^XppD(hJvs99hTREKwqf6bZ7p&1QyatHh{X6!J(oJ>PEmgd6sgaqN6)_3^3EIv
zjhu)Dqu=eK<74y1E%E2+C7w{n_yI+MAry?(IFD;dGaTwAkUOjWq;`8OTC0E^IWmoc
zsSbCS@NhtrG<b`w!UF8H_8lt1|JHQlT@-^gG%%Btl=Sk_(a|}ajWBAd1})`q2?zZ~
zG%H+(iF^Ctxo*of<LgLhyiQEUt9}aakq*kwgdi~W)o>VxqDz+8QEYYcQt}^$e?;#)
zl1b<qD<|P<aKAx_@=ey(n_c7K=V-ab03C;B!ycbmRj{y){{dtJaEM4quWzrQK2@S4
zzj0lipsmjiysQ~N!3f2<zDekHtY|ybng`{}SNBN+7k~6T^*Qd_0Grw_REuC!kww-V
zA^P_ab=|d_@YQJdnHkjMA7wcI8O?#0#yh5D7vAS(W%ZF-mlNIASI15^x;i{B*4lOh
zCVKCK+ot8r@Br>eD+Yus&v)BiR7(giUh!X0S<eRs=Mg`#TVg1Lg{iGZqVHTe=2J{I
zqP~1geE2mX(`|)h$BD11)mg**tGyBAYzj;{Oz@t1A{e)VhPQN?#WD_sdwzNH^4@TB
zZ~H+-Oe{!F)i<HoU<)G53czY7F+(;ea<OTb=5;`$G6E#F%0B04qgSitRbv)49nPcU
zqDb|i^<U0fq)3S?N;gAAJDUI;kJl?Md1mpMg=&wCp09hB)RJ(*ov!^G8<XeP=KA_X
z*+^-$uyk(mR^=lKQJ3``>dBvs;m2Yy$z=0ckrZN|Pnz<^R8q8_bb16U;tr`H7YI4<
z1MTCf0(gqUx-`W~P*q(N@0cod{^gx|H$l=RvC9=r#_@AG%$qyf+PXZ<QiSm_adCo(
zHSWF{Z%RLlfG9_(y7=d>YD%;g$k{}!_p%h|FKo0m)s}WA=XjZVYeKK4KIN4dPk8=&
z9)H9nendY<P;(RUI&MbUNq&dR-!g3_|FaGv7Xn1O3J_&{WFl!~7dDnxzG`b~<>V)0
zhLqZuspJc~U++a>JwGSd7XGHEV__X1{-da(B;UF#8v1Y5W4iV-YOL}o1GIr8$7n|?
z<#s+)FllJo`Qd&;mBhSBe_I`N!4FBsKvSiGejg#uURq`2aB&_{OnSlIKEwW7+-E}b
zXBglJ&IGo+C(d)ieqo3iTpT7!*ubS?@0tIk3H27p@aSho%?)_jXX*;cN%Xx4h}|vq
zAP-k$f&@<?dw#M!7PLK=54AjVDchw_6u+2fl7qM~Uz6nZ%BEJ@e4Zk@1DAaj*!EH;
zCVTXsV+y{|1!X~=h9w9P-raU*3!4ovG675sB>V;C<>YisG^~u&;IYAQO{XU`(EuV2
z9<G|D?vRyI?Q>{4T(K=vQwTFz!$AZ{KIX;}oh^wv_o|>_#K%indh7fk<QeqxHhuNk
zU<L!`&f&D9QrhiRBMsP%`}FPi2ZyCx9)2H54MWW*aDV^-LsVB$U+`<^+9w|B0(^Tt
z%Vu|eYI&z86`(>>G@m#X-+`WKOGM+M5@wknT&~&8+i2zzVIEV6rscLp5vnln>x)8c
zTf0VzmLd#4^Z%Zp56#q=&yo1agV{?>uPSCV!mf2#ggBW)!@xW~Jw1APw+bxlW<Ie0
zomSUKD9RoAK+pQJmDws~mWoES3V6fPVspm^^w!V&$YbTV>e|RPV1i`=ZdxQlHE&3m
zwy5eqB7>TuwN30?hE|h~hvMF+5!@o*pqdr=I2JaSzZxbz(Ec+o-?XO-C8VdJprxb+
z4?{&sF)RAwAA0T%*uF!B!N&Wc#fR=}W8rF>kQ%QD)^meaR`sCnfZ%fjfMY&scl9lw
zbxf11+&tI{G-8t@<U7wrF31DN3GEY3d%*FtLM5M~<r~BfDypB)H-otbI?qoFxd|=O
zb$Vvzud0Tcn!0L2a^WfA7e7VWm;4ND{iyPLr(XZ$@aC4h?omfVUFEp3WT^Ei8-<3#
z%u7P3`1#FLb&&JMAklFvASV1D{*s9_VLG~|J$1%~L(mp=H@xrRw$pvdegK9CJTL#@
zSNzSOo5!WuXELkrwSiEn+sq_F+x*HWhcl--#`=c3??gJ)yrPU#DiIG<Hu%|BkqldA
z2^mS%nOz6Tc#Z`=Di@c~kw@Kx0oh{78NsJ>+Ee{v`}^O_og|GSfiU7}ho#8-B=&pH
z_d_j8d1+;FIT&{+qIkM1?^r4wjNQWbA1MXd2XUjd7$&QFwuc~Pu;qsu7hH`c`4Tvc
z2Op(fC10d7J3lP+DCqwAy7M;lfFj<}ux?H}b>Ey^+Ib^*?PP@Sy~QfH)*!aC!fF3j
zNq6;dX3Yzbk&rS4a7)f~j_Kk5)~GSQEasL;%3|S`I1i7DHDV;@7{hgvh2Ge)`$wRy
zts%GWi*4lP!_bjJk*z<txi)hORZveV^&t6=zn`c>DIP#m<Dij%PWDK$)aQctuK>A&
zzl0-MomUzAFKH}~CEULJkAo#Raak+th+bEvUncGQkJixuD<lU855g$>m2CsSbm1gD
zQ9Qti?~FyoW;D8X7JaJ_<sepP<f8{*Y}R5c3s@Upn0Q#O%&Ca0i+8=OfNT5DkNm#E
z7-^7A@jHGCe0(^93*x5*$8x&v|G9*YIhB!`U2Hvv$r->kW1^iTIk0NmP3BJv&@+ul
zOWbba;Uxk9#S%;iiI@fm?c;zzwu&@Yu%c2HrOVc`fN?a^pIr;j>uU&Z!+T<2PFyP9
z^1GGC@*Bk|+Jni#o#-}ZX57Ay+bfI6vu@x8318V5E?qov!j^f_QX3Re=$q2o^gHoq
z0%8fWiVH920(@>@zc?_woOo_w)^PaUlc-;UQwU&hU9Df~C25w88u3X`Q@yZ|@@k}0
zWaak}f{#*QDi6Y>`JSNrfKb6m_4?y@Cuc4_F$nF#XU(a;$7^!@@Dt|!O^-(f^Zu}f
z`ko74zHAaPI<Ox;#ns62;^5O*(^Xi(&$qBSar_7=-p*q^W5Q#_(Xn<U08NwVj;QS*
z0uO}EIzi!oIh!eaSQ@p{lrvg53B|*pc<=UHH>@@gvi&l@2eW)6TgRepH+6UMc7-B&
zE2LiW$>9o-3Pq36I~-=d@`KivRnxK&nLM@1=BV4s7b$c5uPe8{64R`q!8#bkewRKt
z-O5{CYV7>6Z#2%m`4S(^wHqRQwq-q~001~RkKmj3bt$uJC2fV>z(#m9E);|fKmn&W
zG6-|pIf_Xlh?|WW3hrn)<b+#bZfx2>KQ7%vDM#i*aC{@NOPu&W&QZOxar+#Ke9q$X
zW>;vszBc~Y1ZgG?sdkx-76<-i$T^faGEI$y(BR{NL$<X?Pv_z5XAh3+i&-K2e}Kz}
zwVS%Kx9cW0w&;JFHR!n=O^KlXuCozAR|`~9aUBl67UwHTYdD-(!!aL&8!U-+7WaQ)
z%YQOlVj**agTGUJ^5;2<jb@>%?+TdQX|uBRG7$Vg@A8S<S9N8nF);Gi=FmmMZ;kn@
zQc&)4%JQe8$2Ni=JlQGX0ce#4<A6Acgx7p;g_4?*t*DuP!3*?09V{rZZ8-E`>a|O_
zdRbpL1OM=ZSWP>K7|BB?Qxd6?=>2>T?!G3O6N%|CkdKvINTj%+)ZXcNcHD$0#Ruum
zwa2cgn7&WG8@B1WW(9MOJqIUM9ro&HSN6l~q4qBUDNUu^5UfK&--Pk;K%k(w`PN?g
z+a8wV1JePDyQ#IZv`bAb1RETeFb~0L(}(9StB#T^CvGBCyM?xcXslx3_p<&5<NfgM
z0jqCw8Y*x|!Zd1H^%uxQACtzzHd&YQefn^mrY7^H%}t(ey8h_LV_G=oKL^5L*B+ee
z>x7Eon354+O70OQ4t~iZI0KqlJx@e+@yrC389&|yEFC4Oe;2xX?a#Ua3R220w;GP+
zvut!DlI^hT0!z^nT9lLuIH3a%B!30{&k?WJ$X}ofXla~C*R4c&BAG>E;%|F=Cl(Yx
z{swSB2yo&`s7T1k$Mk$Z8Tv@Vh@@#ycakm7ZT9ewEDz=B$6$2J+5@~l4I-wH{zL9R
z74aA>8rF%_amHW8Ph#eIYQRg#k`YagW{8YxT{5=^r5r1+y=V>{VC9C2oJMG;{D|aO
zg0jl4ZAV(CrJlI%dTipPW7zaHvi@r;O{D*7&TSbd{BZ8!Ck=j$T~;A95x2mn#H3{^
zd{!Aq-OTC2q0tZP1_8fa$Z~Uh<C{FO{8>^*nx{aM3qyj8pJ1D;X<SSw{#obS17k-<
zx&;zW%PEFNgzwZR)XYS<wIC&V&P>I+G(5lXRYfW|JI)FvoVE@i$%oGIVGZiyB&3@H
zdBb##_cfZ8;-xw9&Mz83qJB$=@92l2GUO3ZW97x&5<$Uv0yw`v_q;?cQ70i1l(!Nk
z<>&^Ye)Y>bwx?H)M`;K8ha|P7k|OfFQ1*8@1_WE>LUWCYm4p9_)09*<Qqm@~7$q|`
zBrl1sJAY2%tLb~Y5Y*3|*23N`pjm3`8~wsNQ1DOtZ^za%KQCH~%e73ok?Dt^1-y<H
zMkjAx5L-iKBEWoJgY_TQy1SG$Td@DyDYFWR8uoz&InB^tn_8I4i`NtVa_@ifdG6lm
zxtgj*t212OZI{*+W!2;~j9ZiQNU(Z8zU4HA+Oo6#xZROuC_h><!S$5fd?ZfM1yXfa
zpgS-}yyh$$eFO#P;z^u(e0nd-e_)Gp!E7KRA@M6yB?B49M`L-ta$`S+NB&ah<Z=mD
zcdIxld7Rlp2r!@as?RiB|M+|@sepy^zlGhyxuPM+oXH)T5>-A9!8NVabj-K8lklo1
z1{QM$NG(EmkbLD+Hy`{nY7`o@ucW53(X|(5JgUA#=0#Ms&0z<~OPMDjXqu<>%!%-%
zsD3nS7>J!T^tkfFCQLU{T^Hg0p=LgywNGS=$hAZDda+(BYFSX+u9A_Qht2ZUU-S~+
zR1wofIg_eXStOC*v&*7l=XO|%=2nN9pYWC$8d&Mx371=lqov*+f8u+B2X;O3XReH!
ziDKu9BwZolD-~1Upe%x@r=sD%j`MBNX`Yf2U@vDTB9`PltTaWlxY7q#G|uzue{W>z
zHS=Ox9VTrAqi79Rh#`B1i{P1|<=uBIH|Z{4Y^IA#0?u6hlToDl<f0Hi;T+bi+&T4S
zx}knWrI^7tUjS$2g|2LarNHY{sP)J-9VecSL!Dmh?-S2ousF1cY88|-`Z9SmdCmGA
z5bu?$tG*e#{@xmT3KP7(7~HsIiRNh7#JK<}p{5$`|4-q^+)+Z(%tKcXk*d^JLP2q<
zu7*W2Y0HJJQug7M(TxG<weD}9)h3kcm<AyfSyiF%v(o3}IXKR45nT@@d`~ziBvA0@
zwQg75rkyoS?et7fQvWLiN}tM`*3?b`aJS71spxd99b)HY%rLaMYGL^Nuj+R=Qa}=4
zYZfdT=wHK|F?w;VA?~S$KQnv1Qp3GdpSfV$71%98l&^oy13G)Us+lj8)aNuD`JCoS
zgz4EB#{1<}qT-|rU{)@k$@#wh+!&lkv2zfPo$8Mym||s(ZB8la#hB$1WRElh$cOcz
zvS{{%DAqB)^FzNpE=Q+~ch1<DiIM$SCilr{P)(v?VZEJRC%VUriL+&t>vz}b*cSYa
zv!A{c5pJ2^{G_=#5>sJ5xcf($p4RE1f@bg6bv5s$Y$|blOq`*t-Vr;zon5YQCz+^k
zUc!6xZOsv=h#VD33(Y~mm~(y+5F%$7AMRC@kqhp@S&e+1g!7t3{ZCzob2nMkX8(}n
zzP%9(^s}CO#vp^(XY=2y^8}Q-Q^TR2o#r~^5!sI{y7BbX{RD)<>0joiXbHMOr<9M~
zCpb*ASeI!LI7oypJGrY`Sw@t;vW3b8_1&qJfPq7Q{mzx7jMRQK-KSv%T(;|<`L|o8
z66Ysn8q4dx-0H%E>JRZ*r^I(Er4+yPa4)ZZO@ra{4jCCzl~FQ^iO5BG?2a}8G0&Qw
zTyNeG@$O<!^<{j6Q_*P?`SGXXoa{eBsGAEBWS)IG=yhlIS*_I7vElzZY+0Hs&P%@G
zB!S%e`cPS4{4>*xmxEW2fH%?(NbHw2z3jX>tyH`N%i^*^>eVoG1+5GthmAI*Sawu`
z74L9rycQM}(=S__1CN{aGO!7)1ll-Wik{H+(F6vqKC`Q$+_?2)s#5d6yi2S?mb@(8
zjnwzPPgO3ES7;>R?i7S{WBn%=jl#sl&`ZiLOF3vpaY9c|(PNFw2tZR3RluHXb|YWb
z3KtjB(~LJo*eGd@?##N#5zlcobbKPvxhaM#e|g$F_wg%rcsI0?2+w`r`s7$Ny%8W$
zk<|!uk#v_&8ekS;Tf?XyPIUNRB+b9xu`-dWG^SfhigtT@v!uy>1f~6nB6u`M%N4vp
zHfqh-JKf1z5Q#B(2jIRQL(G)R=*e^C!L51J^80_)B!z9Xy6!fg4ewcX`RGBG)V{Uz
z`&KVMJlXyzS(@rTK2FJo+-eDD(_k#*(j-_&%{0`a#bk~^@c1tdf4TL=bimT$!(9!t
zc6z8Y{r!A_?yHu66ALnn0D08QoYX^Lx+9TH+Q*@!udWX^Ca_h0j`gp^*nBIol$fDN
z*6Gg3Nzp~i*?twgw@BC+xz_vg5Q5^3Anr*r;XZ$_aM^H9yI!Z)lp?B{Rd@@jB%vQ4
z*qXBvU=Vie58P#28CxQe2sn$gTIkhl8Zks~cv6p3$`4No`ie&T=jtyk@o+v6lr2FE
z&_#|&$Q+%p18g!}_HCN__R&>Zm__<JXr@QE-#=GYiC7pp)d6i4332s6_XisXv|)!A
zX+mf`!?z-=L{{pY=>GCrN<p*{u&lz>iR|WyRv9;TNoec6>cvX;w=Iq`2PzR5s`SH8
z|DddDYsA2G7oB*<tEIKIwf3(@BJTM=6Bix%(Dd9yAj|B)=1cr~qRb@4uYjQ;D)M`G
zWjQ7N&Bxbk2YokKH?M*gL9YF}$~!qtXGm6DpHV|5WBce_kzqQ^E$lQk5tqKX^Ry3F
zGg;g+uQmN#lIXvqsb-R`+V1Xx)mTasRK80C8(fDtylEeS$)<zK{+?Q$-WXe^mM&k#
z5viTs0PN6JEj_N`#5Iqzee{j83R|z<CnJ;7?D~boe{utpGbxsN=~Ly=L4Z^H{3G1Q
z^92<UaD!`;At$horZOfaX-0>Kv4sIs=J`n#h6t0e+&N-o)iifP&>?Z+*X_(VFX?_&
zAMsNS3PEzB`lm13*~>uMl3ptP*qQ@E-4tjBH2sa8<jLG#U)(1JUnIpgvjyRX4x9vk
z=W}4%qmWmi+_(E^8RJl2*lq3hZ??}wlg_U!wfn=tVD;RJe-;_!kbbd%cJ;7mucGI;
z+#gIH4c>B`Ny0kQk=jyQ3wr2)K_za1y|4|?wRffC#Cw2gxOTEOwSm^PFiaudcRhOr
z5Ht}`%>*H+otfTFjM07vx{c0E-Ua-PzTFAmIQm>xOJ74xRD6@EDtiP;78A^%bxWE+
zBb3uW*)@>H2pcaFf(_lRjlN}E#70}9jv=|!an2-QZQC0iL7_QE*gJh8MCFjOm$1dF
zp-YTfHdpFOc0$FO9?!Cx1PHc#j8uYv^pL2yA8wlIvb!snF-10()m~v`{em2d8=}k)
z#TLWQAVS9IdJU?#PpG$0nPA3kv-{eAGNWnCO%;6$zQq3#da$IkD)`!ie{eU8`LbIl
zm%sJ7a<)#MOp<x$*z)v3L?I}NSNKy5U!?IBuqq<EiBY8O!H2@QiXs0HOuzzm9akG?
zz^Hu5qwwlKW<M(*C^#aQ=C^KT+&us6Jz7beBZ*of+3PkM1ovFC*)P9~O_Gp`Kt}JG
zHrU4RT(7pjM6j=Ji+m4#E{)he+{hj{cyB4dKQQu;=f`O9KN)=x*u~9?JQsEAo7q^W
z`-HUMN|LBtKPQ;^ukDALEHWT9LXhxlWxF7WzZHC>+|p09Y{4gK%X!MzD+V(%>Y;sU
z-z>%(F3ciw>-W|QY&KyY7zPafTc(~fUi;;^ar5Am{}%hM98COd9OO_U+4s|Acx<?e
z{fL#gX)+0{8054)efiJV>c)-@F5#7{2qk@-RqM`?nm^*O_F7%<*5m*lYa^c`0FU-n
zWVMk*i)ovXZ2o&`x}krtap``_T=QK#8#5Cr9xb%@3SqfIE&F*XHI1IqfvEyIS~GOp
zp4((Q>;Rsh`(a(;*{Q0pz3EV8B*awZfeJ|ymKkkC1!`s80C6?L039Vo@m|Uddu`j%
z$Cq3U!aRe=##_{WqZ&XN?V%o2%qXwYA+LQ}Pf<q7`J)X-dC5oSpn04;?1?6$KS0`(
z^0Ri+f&PzTBz)_EU+-3JCpyxgystBVKgtPM=Dgig)s7aPL(eC?Y?FnF^k6{5Ll{t0
zbK=7hKxnnpa&~{t|0KmjbCOilK6cI$7IElz&93_CVxU=CuYj_b?6C$cDvZ&nv|xJB
zU4_ztBiOFJ;9v4$ozqpUVuav`e?L~kMVe&-lac66r<=N5g2mn9&@Qj!btS%$e~#+H
zhqvwwOtJE_LrK>$3uQg3pEkDsL@KZpg`*prd{DjN|4a$RsvoMMNA?~SKFe~ad4%y9
zdE&JFPrQ8q*o34J(>|-||BG2d0*3qlFo7;7Wx31BB#5!Ylz0`8zKy`xqU$(8+-jGw
zWRkG|*=5iXr4{{pJ3Lec+6tG}gEdXnNOypvO9#)_!_Oj2Mk%Yrpx(Qu)52_Y7PRC?
z<b+0=XjW&M_uEHeYjHWF$}(RJW1pI=!@dA>ZeI^*sCJ$!7cqv}e|Vl2eG5GOf5e?-
zR~&8Br6IUBuEB!@3GVI$CqS^^7Tg`Wa0u@1?(VKZf;++8U4u?P@2r_M|6usQSE{-f
zb)VX2?`tDalNb{(A{ZjIa{m0u`|$RMpC@ZxO*QGA{xa#zg41Eaj*hf?%072b56yT8
zfwffuoBi-3;}4Wj+t<u9J4QtS^U&oiuQuzi5XE;(|3n5kBO+Lx%grJb8dp+Yo11c>
z*rP<BCBw|g3kqNhZi2}~@|iE4BzB9Of@ZbVc*XLmtAoO^y6E2>?FcRmL8OcQl1gFU
zTh_#JM*`hDzjC0ux?+M8XW8D;UpNXlL57NNWPC2?We4#G7LqayMCWHu9W1<a^AKq^
zo%G-4)4v^AO8E_C0lnNUF*z-paI5H@Yn>wdObTa8M<c+xrBUF-l;smYC7+EKohul?
zM?c{YmPdPzmA5aA4(_FX@LkH#cK&t$0g7}#CUczA6t4P%-3n33FF<bJt1wtqIA(_L
zgec`3>m^ch7xRQX>VKi1yl@Gq^S{dNp{ZtzVOHimTGUwDX118AFhp(2YDdfEr^pfH
zmoLvsPJpq!9t-WAX%6i-_B?$b3RLT#BgO77{wuSAc}*53W*^u<NqD(L{UUZ6F1ALG
zFb{vsWO`E~qj2mv?L$nBmXeIb4##SgNz>G(c>Gtmn^{<wmNnHAZunM*_A=I?w*;+N
z$Rqggo}HOjQNg;rmq+Ye;Q`oFXUb7H&%a4xFo8YOLMnZaGgMk8Sn0g235Ng_#jLmK
zYLsJ71Osol)J*onrKLai8z&3JoCO{~H2`HCV(qZ9tfo$qxF0jD;rXN;A={qh4c2tt
zBzvIZY{-a$?0B9_x!$!MPO3voBNYgR!|6PJYWHu*B%v|)<S+Rpd5CqD#IBd!Um}wA
zAwaqr-cu>&NlX$!j|dK9DY!@-4czyIXGr#!S+Nk{KhO~7rBNQ}Up{9f@vfw^)cOnw
z>#DE&r_OwMAfenzdt-aKV)AklEbKg0D;H*K_Mk=LL}ZdIj85T!93)9@`P&5;kAtbT
zz2{BA$8{0-N*Dg0zKeg4$PV}w9AU7R2x!X};zUPe!5F@#=mj5p*jcsZ$Q%lY-J}s(
zhv*DZX1nugl4`_m7U%`=s$9;AP;=Xg?x{Ap_05)3ToMQtkN8CeU%zmF9d-?6*G9N^
zWMi*Td_7bellekTl^j|Tw(zw+{L4d@FQ}^tMbRSH>JN=4kz73%<@CpIilM2e#6Z{(
zLjx16a@Y?ES<#D%aJ#wmvUNDr6|<wAm)-yJyQV7V<>nCcWB;cl;?krY8$E|!+kTn~
zN{;LSzkb>=8@p9d*f^cVseX}%!X0NK1@Tvwf9(H=#$>nIEI7rzs)xBb#k1`hXnuR9
zmm{-b<@t-qTNR+$7Y8N8&ksWs6Av6FhVFvAeTN+sbW)J`&KAq{>Z*X5yj4Yg6FDah
z#(;W}Cjzwxp=;&a?bLTMS$K3&B#(L31>0ynlf_%+uAzvpOyiLh-*A?zsa|dMxOmIx
zMn_xH!y<kjG^roT8D{ZnN6^=mKg=F{_E10TryECk`-hYwE`+eLB56#53w$Mok5fp|
zik!bBGUoZ-SoCY>^>FNOt`glj%8k8dEi5|2(`pUtf(q7w{|;L@m874gtY1tV3Z|og
z)xj|d0gRCQQi^A^bN)sru|@fq2kt$FO6n4GpQv2Rlg<!1jQ@WL@zC0}v*Ihe69fl=
za>!n3ZQP!X?fi&D`j0UJ&}$QZcv01ttN&ecKUQrahz}~}N^N0@Ju273&zO5bCH23U
z+;7QvWAYG_e2O0d4x$N{)O1}JB|(1zQcl~pl(JbmP<SdH@f0oF{-zw#i?3>&fp&)H
zUg8kYvpJgmIPM2duao=;f6V`MC~5s9xIdCvZZ%LA9886m?Q-o2B&khtyg;wf!JtlK
zjVt~_>jcJ+8`?M~ZC})}vPHMGO-eE2b29b`PpsBs&eZ)GDSOWww1`AC=g*rES5RkA
z7e(&iypAE8to&=EZ2Jss_eCMlZI@wtTD=@%CSh}hLw8#hMcoV?L?x|hGIl@DmdW%a
zqM0(a+x~FTs!0fZ6l_rnj0ZEN3@TIlRQFN%s6OjN=g-RNNGB)71Wnen&%)SnN|LvE
zdcHqV=jVqRoz5|2V)lO?w-%MUgs>VUaia({c{hBp;uEx#m*)i8o-scY_L_2??WNW#
zZK^#|NXCet>tTQAq{pr*vii9bH5v$rsTchYMo5$07=QydLRzd&Lv9z;A8%6pNl1~<
zu$yZV{L>~Z!sQfOQZuur{{dic1YGRrw!idsJ*Eh6%Z;GXGu_1P-fm)2>hMjWyPYB~
zg4*0GYRx9UkG{G0$Nsl1Q_^l#x`8K7kEz3M74}l7(l>oYF*67*krle<kJ2>VBh1pf
zHNC<D+PT0*+2}9Om1C>DHbi`9c_6jiZZuG*nkXk>4=NYI%Gn>I&4@F((`h-ls4LCl
zE5NUUFD2$jRFoMUprR@Eoa@<+%U|h~R(7DOMm^49?!$UTtczuRRxe>*{a)B{JfN7d
zHEVyWKEv7t+jop_A$cblkBDG^idAWeGP@&2>JISevCOHj(DsQ+*Ufa{eDR^plYmcB
zi^^!I=>#*AQ$l~)k=&~>9IWUrudIcC>u>){V8I~s>#{<2YnM1!=?`^Pp1J2T2|am_
zi+;AlwGnj+T21*->7b{Q*d%q+drVhqW&E?CNxG`5p`?$_#qe!A(3RaQLqhG_Gcj?R
zatvr+qOt}vatpZh4nA8rJ1R;#z0XOjEtS+a=JHT|8CWENP12@+mS{~a3H{DT6|7?k
zJL@+5g`?g5u1Lv1P0Y+eQrF3RV|e!kYnTFZnH<8r*Rdg)sq3Oh6r^16Wt@<ZbD(`;
zaqZdo&}E_|BrR1#;=ig9vswlyi^ZfT0$GlW6N#z}wH(aIcoe+jss|qheO|=#^omUO
znYs7yXc71!I0km|b4cQK-h`!ZsC<O4j|ekA#=okye~tV!t2z3@Y$(L{gmX{pYSLSm
zIlzgQ`SZZIgj}0bA-fSz8a{-d5m8oE+gLyJ>RYpva0>aClE9h>nTWb;TujCXObR80
zy3PzrtE29GI7?l5sLb%xc`f#i9|Rx{qOw9aYf>#>1tn7{8ocW|Eh3(lo*0{Cg#R|G
zs`&#Fmz4W<H^2*V3}e>*8a}sa@$)n>a&<V$Mllvt!1b`W+5(emOk_<h(T3s*=}BHA
zErg~HqGz=ozCB9Ha;ce-rUw92wLY8a3?p*L2><xiHla2?QlSxU-6!H<+H)|Q=r@&C
zo&?Yu+S4~yR8^u(D`x0EvIx#CYi1?KV(fvm$kML+7$X`I3MBX!`yPOpny(gVeCyG}
z!wi)+N5OKm>*En<B9^9xM#F;Asx}^PYihggh;CJ4$l*H1Qp`2)bd-QxR@AxSc^@hg
zge5cCx`lq()fHIBe4q#!;LQp(nbH=!iIIpa^xQ*=JDEx?&i#>bP~yec#^<T+X1B&7
zTC>7<XxU+yc-;7#DjJAUX5B*hk>aPoOf4%ke58tn3mh)!1~hwkDn;0<yl|-?Y4RIC
z$N7KwJXAYzv-S6gvVh(~>EObllb(QU{elMGvUrQO<#fwBYC>mCCrN_l50%-ezevwD
zDO#12?*le*E&^^(mtah;<HqS)iHHQJjp>^z=;|=~6P<Q=iduK*)F5zr7RL6TH@bl1
zM>Tr@{WP$;!98%MBUEO1nR;%M2|GS9WSBBwh-4m1$NK@?0TkzQ+^7e>>zZ3Zhyhru
zw-T`B>yd+maM^KwI{{1nij1#2H<UAo%o7SIH;ydR2}ZQKjez4SmnD}4Q#kS4RZym2
ziO4CIt=@6=Z0B9j=gq2s7KYH8>R;Pv3&S2bia)yWW9NNw4Ic%B7iJ@r8OYiw%SGs#
ztG)+7iQiD>1KY9umJVlz@~$v#R|cMMFt!HBh8+@cJ&K1arbv*pao$dekr*z!&Tpr`
zg>tt^@{kN=$H%6z29wl&_P2<7Mxx1{4U(D&$aHF!m973ODO=qGvK-UJ5!rxsE;xFv
zYgJe0k`H;+1s~N#tzc2)ecOuaKwZG%fFyHvb(x<9U_(kQ?G;zmNJz!_H+K`(niWUD
z@Gl7$Z^S|CW7}9*V!VbT3bGD0UPFUhZu=tsRvk>sutiNHCGF>J)|P|FySd#!URP{8
zy;m&33A)F`$sc!KTC0&euVc}!+)3REktp34-so)lu9ce?l*R?b-1#TEber((q|lfS
zN8;wzfGzxh*&-?`dbHSmFC!%b2glS(P=-g~USWCDaeeK(<!%RpqVCnN5v%?(J8vn@
zyS!}AcFo?#HTePt<b4%&lLofviY+{>WECDuqXSHgc$M|(wH1)LV==3G{QZ7#+}I;a
z_ln{|uqV;;phs#c;-bhXra8_Fy?MsVwwz}TfQ*wJdr40vw7RiV2SJIDQ%%d3CDh2b
z5>mUdvN&BLhyOX%n4a>tN}|-q=R|{O%8`*uz8dJVl+E0W8p-CgQjcDI`Wy?ojb+pC
zHp4a#<W)7;`7kxGiDa@f#wrS(0pB`M+qnRMBbwARVN_Ohe0YKuIa!N>(lK`AUt=fW
z10#O9qy$+u-8*y4t0(OicOio9XSe*fR2pBgKAhg}K3cWvVL;g(-X5x3kSWeuo8`d$
z;8juT_`m$2H!pM&gtazo-TNM9nofO5)q{fApx5yb!(XlHBs<2M#1h{@{f%F8!Ohj$
z#3#L-{TUGn5>_e=eSf>jo`BMG?N9Ob%{-;8`U%F|(?yA!=pez4ozIY+N=slNWOi3f
z*W&jQijo=oY)1$Ik3k=Ry}gY}+8faaW-K72lMdZ2DPmJz2j~gGx%Nx0iY@K0h#l+&
z>f%LYKWgi}k8R|g69##^yFZ7_@2`nU)TXDZLV|cI5QT5}I<6lLsD?6;t+2CnH1f7~
zHgW%1Q(oA_D9lbR^mOFo7UE>?tfm2cJKu=>wwV(f4jAz&j-9iYH9HMWUneBR;RvH}
z_Wr&&3jr$>78)BJNbTn^S!dV@B7(CrR2QZqNgq`Vl<^48k5Q!NTogYFkyDKQ=?V#s
zuwNlM)Z9MZ$PAGM;zlr|1*&>nP#9F=2IyffJ_Rv=KJ9`mkBBqdPfxebK<b`<h}wc#
z(~%LLSiK1H{K?k*(*HE5n*?pn?9FF_Ub*velsUWZKeZxb){8&h?${Js3_A)l;kKY8
zdLOvG&l=md+rWNO`;UP_>1gRG^*xHTt9Kgd$YT#JGhh~IQQgqePm8~%yuIoqzCS?g
z#ipGI!1mXJp^&Sf=t20TjvA$CpoU7@Ugnt7u@uBEZTRE$HPXI<jRhJGCM2f7pD$t3
zscrA(6QvMmdr4WTl|2@_+s5LWihkNNAJ^E%%G&g(#QP=mXVLq4qTyGN)aMd2Pb!v1
zGiyuu#JqPvVBuQ|4r;Tnh#@VWb{xsZv(kI=3VPU1*>UQRHrI=<&a!4;Ed{y7P1r9<
z7j~s%UL@ErJ^j;vPYdwo<V{3m&D?dkD@jgMRLNp~@WN_@jrXfYPiNF}>Xq)Je$Pux
z2@V3CM{%^-4~{dCQwrr{EYI)?@OC%wgxk7WRE^02kv{teM@GhmIWQ9fh$kh(--~}d
zl10eMtc@}8bb_(sKE~G8PXBIZCTrI^Vi@Vs@|MlJ_%7Tts<0WWe6fYI)DGf;uHFE7
zq-sTx5Q=_xj_Mz*sG-6+-LVDfYQ$`sub>s?Y|ZR7-jYDSJ;f)KVP<&mxm!8NmNd^U
z8xtV|SRX1HHkbwdp~1vRB*mc}PpmuPip^v~;d0C+T2(6su2I~J_<kkmuA-BNBz)v(
za9N><Zx^ri<)&(4MEl&?rd;FOx2wS?b$J*4S85cb-{I-8UYHm{?PAQ)vm>Gs?Jk7&
z`pv)SJs`=AmFp~A=-2`Wd1rRIXp)~PFs878D$_N1^UXH)R?J7DA8xZPCslO+QWw=x
zVr0UBkR~F<Lql3<3RWaSIBHIm?u40GRlKIfi$_7PE57EA+G0?tQb2sL>A5+rU&+MT
zsE0TS8mh;;v^DMv&85mZ`tOhdxK_=>QrEK=X1^i&r<_v$&%D%+t8mntLbNSb)GS;;
zet`uBMNGe)wd;rFTD4!!zH497PsCG|Nt4!09<8*RdkMV{egw=M(_nBtBj7w1#I&QE
zZSZmjFvzOToSGRJ86NB|GgPBjc_45Xpdl6KAi%@JBZLG8htv+d1|IZ+1gn>|i4IGQ
z7QA9!LKCOe`iStL9`~z#tH&d5+l)$pD0vL!H@BXC9@6@OFnfI7RbB`uE+mvPn@tcP
zbhOYvP;bm9{8%w5zAkEtIlN8z;as<wRXDemtZhQ15yg(_go{Ufy)`BxxKzvJ0MQ;?
z(aOSl7}Ys~X{x1?^{!`)m?<IgvcHXugPSaNO`wB68A36ARmwX!KfXrpLwpEaoSYpU
zTURiSD*g&b^mae^jmr>^t&Q-|d)6+))9rI|%&hs0FYXI^Ww^AXyjadw!Ijj2+IjWa
ztmPaffpYoQU`;{a3ZkhAPQtJLF=XNpxwf<Pckrj3oC2%8(94Jq#xxO`a6zbv!_-}4
z#mXjX=lMFvIC%>*MGyWb*!3oWeMSON|GhDdQZW&4!E6uoac+vw!t%`h1}wr;1wztm
zg6LlF$(_QWA;+_~p(}IYb`{es)2sun*v~ieVsxJ{V{_wVV(Aj#kw!FeC<0A}Q#`T`
zJhE`}w*HZ9oODz@cP+R7z4#lya^Tke<`uQG40&|&U7Y-|%Czd#(>((!YdXK4=8Fh}
z@IzG1F3t5qlLH9BukZNQ@!d4w1|iz}r0~`@I<nj99P+oTJER0J!qEZbD_5228DUGq
zokPdLgN4@`+OMe)OrCCk2@yKqKaxi3W#%k_jvb$)igAaj3hZ4`G@+MxA7a!JNf*Tp
z_l(7;rtsGnkGw{ny;z|72FYeXF8q-8*G5x%U>LNF?;pAn%Ok>6ZB2gN>n;)^7@s{k
zdS7J0gc-#s^n2$94z=WNze?SzjAmX3d3`xT4x6RaBMUkiL{|za#rIRe5RB}_ds~ah
z$5UGF%yBitAgXJ!jdamwtDI<(pR*2#{@xRh14s2$S&I}4`Pl%q`F^4rrwh7#PiA;*
zu3s0#$Wb9dm+IEYZ;=Pmy-+p_K!43XU8z9P8(Mhy3O>Nn9>c_Vr#yXX-AN0zJeuTE
z=+(pG&rWHG`<+WggU5wFVUWQ_PpVnXop;-^p%C3yZ%XZ9q11o=qy49ORdr<<d8dA8
zNTy#>ntuf1do(~&{Qji^vpa&O>-^66{icZCUt}^e!+tRuJK9naTXl-9&LJSEhf3sj
z<G~ffKv1wfr=biwwIV&+4|(Y=D{A5TbRcvk>wpmrf;}p{M-p>GvCn|degE&{{yBgS
zu|?py0zESA&}%axLW)Iz<8Q1b(VMW5l8<a0^$q+8K<490uS!pIRat8>U=Mo)g1i?s
zbE`p}J%m7K9aKbcnu1QTKWs~J)MR232t9HOx%oa4vwECe_}1ddzsRhe5Og~YCBXVl
zVJ(BOqiq*`8?9=XYxG4|mS#p?r3?n*0}~=L`syslQbRal{FF=uHUHe$3sg}#iO9z!
zXQoC0$hc1n-HfO&;yIYJf=)LUDi`!pn^t*+EJzkcr#lCa|CaSOkr^2)^Q;_a&R40K
z-@S6Dkt~)WCNACnP#Y4Kvvjf(o60XEU&BG|ds`;C)s(&cy?rte^OU@AilFt{b%^T<
zfZrwQ{k`F0IX`R)7lbM5`@lW`Qwd`v>&XQiCdJQl!27g?D9zT#K3Ccis5#_>>cHh!
zh_KXBo>)wAZW51<{Fd*=f2-Wdy-ovotYJBx(N{LYn2+@18A<KF$e`ZWaet4)3M!LD
z_Rs!ATImHmRir1sze?ABO@-)*+MZHH_>0cJHX<K55)V2fqM?d*+O8=p40<dcHp&3I
z8>&YIFihw>$G2ztr{Mfr0T3}&f*I!KYfcKuzPi6Nvs#PXFa^JZ%1CP>fkiT|1Sv#I
zqfRg~x(b&T*_QNHINQ#CPpn3`E@eK{meRk{-{gP$4O+eGe(`@Buvu$!c}8b3u1CoB
zO+%Z0=|1$>QM5Pt;V4C#l6OXur45LDXT{$?=sETpv7*a|)5ypr(hP!&Yy*C_oIXGL
zI6hp>B03H+pg894r@hBS=r6o9``HtG<|ScR0?kbEo9&#!Er^7l=<s>MOVOC&DPt8v
z6eB91Gqgl%k5pz#+qPu_I%M|TPkL{AORQ_<9%|a}|NfXfuQUv&SG`M<L<7h~rkvd<
z$REd+lrIW*a1!TPU_mO)WCETUF;B<iE3%I|+^o-}rvrZb&iYQKqEcdNG2KJ0s?v>n
z11*PBXn!NgN!2HMH=bgjPM4tYozPgp6YuR&(#(>%oq;LKB8ugv29~B5`!K{X8coj5
z4-`a}H$6=@?lslLjd)8lG=Koud3J3=LOd)Qoor9{o*5hDcHXDYDc@@|04mS!l6Ofk
z{k_`x!Sik-8A<wm!SL)z5H?VGcdj9V@$^TBvXRs6953j7^K$5lS;R<Hbz^lM4#W+u
zlA1NqRP@x$Tk3Gw6_f`H8C2gk8hu|?H5dCu=AH7<TJ8Vb`yyVv18e2dWY4;E8(b7Q
z`%b$pjrk|@=^)D9I4N=3gem_MA6KyF2kXNI4k*(ld?hOJ^HuwLnEacMa|9Q40Fy02
ziuv-$k=CONIv0E>N}KBf1pV<n;)LI->~J>9xe$W(^MzxN_8kU)-Rn0iX7*JHKPXM*
zg`1k}EUe5gZw7t>CaA8uc4d%op?4wyw)ah0cxG!Z9^+?;sE<#;a~jORv&3FQ&+Itj
zRiww2pN91<=!25A=Hsr{JJV{8q?+%QDE%a4xmR9c?7j*wYh$|(idL8iOiu_9r^?^K
zR-LVZb%ogSuhPeMTS9Ym*3S<rqe(t|XV`VOKYYJUAJYIH$7`rh3wdXCGt)1pF7obJ
z4;P0YNI(R)w{+WBGLkUakFVbxM@8J$&ZG=X#LN`ItSRYs7e{9&CuIL|kdID&RxvS;
zdL=&HPInR~XDE{J7kDa*g~nsrw7l6NC^6;1VP>xhgiMnYT8#p4oAff)Dl1_d&j@}m
z@Mj~k^1%225+4^|umRE>Efl@JzR5nW24}-VlrO?Bq^T^jBb)LN<fVv$CFzIP>NY;@
zbm#yIgXIw9*A1FCxA>wz<@#p5<Wis5LMiStMN!A*SOnM<#D@@3nnubkPsjva?!)x$
z<*!egap&(8Y%)7xsQUZG#RCFhUSACwPRGj=9)Xk?9X<RrMFPvwuVL3tEHg3N?=VT1
z!ce<{q#ddq;+z#jX(;hxJ{j`!d!IH3To2sdBugd)cpfAQc5grWi}a0*Ju$XvscRxP
zK|f2=6i>l?hdC2zKO3AyeAL=bDD7(VFF<5wU?P0WZYp@S%Gp$9YsJ7wUPMjF&eHz@
zM+0<8Uc%JgsZ<6~ZY+rRbaZZG$E%3ydvJZ}ePxr?ei~v_)NdieAWBZtSYKU0qUM@%
zGx0=@NJYK9bLLb}-tF>!+f~WANOa;V3W3B~L}%gW{=ARY4`}qD8j*X-k&$ofgo(8O
z6_ZA4OnEy4FuzS&i+&|nt-lOs_(`1E;0AQMqO`5dI?7rSiE;@G2nj|L)*M@uF%HW|
zAfa8t2XJlfsX~s+*x(4Li{gs?ccGXYdV=hm=n3ukQiOKK)XARp^{f7{XP)lTAK|SV
zp)y<R93Vy|q;{vLRWu+;MR|=Y(8EKjGQ+KD{=-pI$)>MW6*_HXBkm_ykwy5QzJT2`
zciOBz3<gN}>m+kd0@{#{+>8M>dd_bsGt4eI0rxo3PcfwyypkYGA{}1c`*yM`4r=11
z^?(R?Qqur>LX(P`{3afhF#=s^cBv`pEV5rdlM^<;bpP?Be}?qL)RpJsALQG->*RmA
zN!QjUqxbDI+$n|nSWDsfvOi7LU4UrqzwPc%f}x6|o;n+XO`#JB-4#%^J#SS_zuBhh
z{WeU-Zs$xSat5yK7OCGpSBz8UjByy64;}JroJJkfeMN>ri_>R$RCqj`wh#+-<C6k^
zZKO&%5P%^aEbRDFWkH<sLt;e3=Owi1hS2)+qz@+cuGa{uwOr^;{fuwZxG*4SbSn)t
zz+XZ2){pO=4m&m<U`x;V9kNU*u!Z3~K`*dGpe3;uhn@gIM8yCKZ2lZ4U!uCB^uWRP
zBtP&{-+wzQT@MvSZSUOxoyFA*#nfX8ang5RiIYhv)ltfO@Dww+uAH70dKfuPlsb}*
zn@Y|0^lwl|NHaE!K+JD!04r7`2o+5bn;5@3=MX$7PO~7gM)`%hye6;K>L*bcNl?(y
z6-5rRj&2Mo8F+DMe*Gh=CXy5DH_E-|Zr$E0Xqt5dj17NE()zD3Lid+_^whkBNY>TO
zU(C#0ZNXbU2qrjc9L~3q@?*X2H+Q}{(!*82!PG6*X?A1F<J;>lU?A<YoIX}S8-2df
zvF|#u(5I=PpePoD{fPUv);1s<)c9HFw&`4+OhEhmWZ!;9Mh+cOF*ZInHZBGO8ygc7
z8#_KGCdPl7(vt?y^-GI;V$k*?SCca=NqWoNVo*)&EMw@$6d9%-hysTz*-F?GvfRH6
zc}T0O$l`o*dEdkXE<w-~Z_{!EQE(XV;Jog}vP<>rcm(d%Q+bzNNnXvA!v{!e2@AVt
zoEI`km?;Vwwz%?}7+8Du9XIszCjVJ{IqwrpXKVqJWa>)se;yZm(E0lPdIu}UHQ@hv
z31W1FG=60C{An+W8v|2(+3z#~y1!g0L1%G$LrlWLeyF{b72q^t>&h`X53FsFcYxst
zExe~E-``G-IkmYt4i2w}nVA_;VJ9I9R)^96SxfCU`<JLHy=D<M&_@eDl$(!;Boab`
zoDf-B3o(gWgzdt><+O%gZRE0cS{3shhQ3f34(G~Zk6;_WIE0UQ6h)631~X-Ls9_J%
z9RU3A@&Hjv7}LuPffu7VM|$NWGw|rd)r+K?lHG=mAH^2yWZaCNXm0>T`Uu=qihfFV
zW@_4LDFRlaTNN*VhLDA2NqMO^2B(R{%fkuQI}XI#UQs19T9V!)Prw}E$_<rTU*lUo
z!9|nP+c!QqGVyz~Z){*-tZ#Jm_rwIy%g~^NC3GV_&RB7lk)d1tqx+S7`?DrqiIfC9
zUXtiHZkV{X^WeIU2J<byE`#60b0TtIku`@|mE*Cj<xh;zx~afh{mF&PwOvh5*J3qN
zdp=tYzoI%CT1ucm{Oly0nr?8(X$AgIhCA3?QTT%7qZhD2!P3INlw`-<Gc_IoJVBEi
z(w2{7Fyfi93w0HNA#cg2@~C$W^9ks(XOPKEtaRJf`cgkSsXLqKw1mMc9$L!Sx1eYn
z%&KBY0Sy)+wI$Wl69PkvQW!dDQAhUw9;%$k=(Hc{UtuO^;n^O!nX7ndR=>>W*YrRT
z^J5<#6Bs@seQf2|&gLmAhV}uDHibX2;QdWm7$jmDSj-K5e}PUCE3KADFkyz&I+T1K
zl?jD9VgD}*!isdLCv4)peaq{1;fr7x8@-c7TWo4}fQyr?zA(pKnY!M8Z-|{Q{_0}q
zyWi268<#QW_e$yS{bLiF=*!MqBJ}lz@hdW}kye)gB$5N&^ca_ap$~NbL?L+!VEYz_
z<qO7CsrA{$Lip5svZ8{rQg5qax7zkh&PdNi!AnBdHpkIq=)IBq1MF()ZBkR+P9uzK
z`-^~V;ahAx4)&&*-=Ybk>&1(Pl*K#LN$7<{sJ;g`J9>IAj@dxS2K*Q?<X(FEMIj>G
z_GMOmY2z$)Hh*q(FEA<4oKw)z)>Yfu*jeB1rf>|EbdpX$a(Y5P>Fs?G*XSo8%#4pf
z%uY3oKP(f&XDj~1MWBpST5W^L^rc!XkrwCKz)eT|E<GWIoD4tRSo^qaG|cxE=GzP#
z!rm@!VuDMKIIzBU;A-zaHY@Qcn3q?S+uGR<5EQm{HFSL72Y#a6yEY}6{T93w0=_9h
zdvmt-vQYDWPN40lk^<TV&XoKjo;PDTw~NWQ{Vo<E2sI@HHGct=fD~agpF$BK8VG64
zb2BF_DeKB%1x3RySU_=<{!4io3|R~GC61qUTYj*$?}D)9NhjUs-L%YH0WxYh-p@#^
zkcbm0PySjHu(t@9LPmWkkXjZ<|9f;Eqo~j;h^sx|ZF)1A2>bBNdvtj7lEOxT6PFrJ
z(iE4|MS+Qgyy*;CDY~e0lz~0~e7NS*qX&(;)S)jSc$rs#tV4Vq-{+mlgg$cZP{%es
ztsH~ZS9sOU2{~sT)7&LyoQ<4YI@`!$7+ZcgHwXc?h;}WZvZLiBkoQr(ont5F?SJLl
zZC{p3(4MM*ey5MK%_AO;f0iYFWjn&;zi=t1_g(hUox<G+F9{HRGV64KE1iDmn9xc~
zCuQf$D=tS<sVXl91`ZEdPn<MfNF`{^a5d&2oX<%4iF|2iOc1TX#gzkQ35jzfmzud?
zs}q2v`7_qey8M_JYGB1Hq%f&hMV^Tfcnht#QRpGGIG(-q;uiQKrcQ8MuAvrUiuO_W
z%QL*xKrxJ*jQc3m&#&H6yvqistcEkhkig!?0D?!xLi`rx#-Vce2jv0#kUL;Ac_dK@
z@k;p{AOF;krzc(vdsjyuV=Eh%nbonEfv-ThDp!;1J$~+ArY8?6uq4S}+H;th!skFC
z&iEa1k%Yzyx(9XZoZ?rNd7p|;#4BY_@L-02oSfYeEm3~n9o>~ppJ?y2b^}(r^pZ13
zDwIC5H`h5iIy|&Nt5+g1-^iZu!AnnTW6_DnCxR||$Do}%8=)Pdxgn6mNpbIqv{E3v
zKGxc~L?60CUax~#e=3T|_Z1QI9h3Ez<h%s>^MtM)^CZK$Ek1LGu-C-DKFLM7a}5FS
z3yQD=m+><}7^=<g{v~3@-Vb9_XA8=AO`9&B*Yzo08ku7*=*VnuHtQJx@o<YN+QbhO
zAC+%wOY11<GE@hyV7#-U^v6$>w_w1xdT}pY`E~o~%<UO0<8J*;id55qEa+9nFXO}S
zMI&3s77%J|(k{%*4um~W&98Gjg2IYv1G!>9f2FIGn@U>VNQ{7d)tm70cbKZX8qIfi
zJJrGT+rW~Xo_4nnwom``9=gE&i|=MVXZ#Oh`Pqdgd&%=nu3{b0N|W6~!$4whTFzP}
zntI{CYJHkPNkLRtJt6y*bOU1@^$jS(<ch9L4H-uV+`ob^*nbB-y`ik{a67=MAIXSu
z@w}OPE*5l^>xX{R7MkOWeHDF}(msK?M1MG@32D=@{BvFZ4LkdHU<fSy;|C;5&da6x
z>0B_JBdG^fF>x&6RO$sGQ8wVy{b2<X8FsTojP(xp`G`$Ictyv^@#%Ie|1xy=FlFfK
zbE-@@`W|X)DBHZI_J`*W_N*zfFC0&Xn+n+sUqn>*poCHO^po4)+eLp<=i9k)@|2d;
zkvbDYU(q4=64`+?cbGimZ;nl6Aq$q<_~OZU)meSh{2-Yja~A%llpTQW@9Yotlsigv
znzfU#6Av#5t$<^)nv}<IkuFz|J`}rt%!c<P*>6Ss60bgED#9b?*>8HEXsKzayJfyA
zc%}M{#dEsd`uDjI;l)F7Rpju$+dAfSR&g-DddM%m>Ifw>^Ju1%O{<Bxj&rvtJ#dn5
z-&qTi2{2dq0=-UutjGYH1Wykq9|iS)EB}!-D^6Em`r>B7%F^I(msqR`;J#^(!Ld$h
z>!w@@&<{_p^p7uZHSAsGha?mR@^-N|wC0lUB?x){rJ~zHOIUVAQyc>xKM;TJ&+=k7
zwindDZz_QBZ)(Wj6~Ll~Dh?wn-_)S_dD#4}eg(X1Z$eUu<v=b6E$8{P?nZf)xhQT$
zE|i|h;V=?ci~1cJ)VAp5zr98#<x7^U;Zo_cwpv3L{@Lk;nN>*Yd1GpB=I{D49J6?@
z578|$GCeg!yDVS_9{$47wyV-ckxb94#lqE5zK72C=)-;vDTqoW#$c@~vHK6KAvJG|
z@$fS{VqqZ(*xxTI`u#ijcTv%P(n8by{TH=lOLSxWU~Phw0Nws?*@3IKExvkhxVwhN
zfs%{dO^^#CFMb2Rq|c<88Q|75RIxn!3a}Y+j7a%%5tk_ij8#<REr<&lhxOFnkFFMk
z@@JVd$w%HcsNtG?dP93H6Gp&tMPdp^dBe2eg)mtGK`M5+pZt51*&4B~&jW{ah+YIx
zGQ2&b&cpNszw>Qzui~Jq**fnA2^}cj&hL{T_O)w~5|Hg*U)J<U8+9KTU-xtdzsYNi
z=+Naty9OgLY)X@o>eZj2p0=;VG@04qiG~*&ww~|YhM9UkT0TRYbKA)-c*iAnZ$wk(
zAF<hhV_N^u8^tvI3{F`g|Mcqwi^`G9&0^1fuY}fKk$u4FNGpwZv+w(pv0b)*zGV?1
z(?gc&J6Xc9`TD^#1Eb>2wTY8^ZohZ{))V@qDB#<(<feH`_n)FmI|R)58vowXaIm2R
z^JZ-6XZJZg+rC^>SB=FRbW9Lln96627pm+aK#YuxnDq2NX0DZm=QI5qg+SMb-@jW^
zN`AT7T;@NM`9ea;iWdd-O7^!mw*rr=#u6J{z1&2b{pb@sOUNHYo13}jH8c|N_72Ph
zqg!@d)<<Q;Wq+u}B*!OWHWddpcX9L0Rf~uWF_d~hvfH4saNO_BX!4V}KYXSa*N*Lg
zD@@m5aZj(ydWO6c6@49b742^c*~!<BejC(YPCmF=4iuz#e8_!~VCQ$3%~?me5+ZG0
z{r1<(k?_;FzfrlhNC{0xZlMSvq&MeF^XoI32rB<EjuK<j5-=Wk2xKsbEl6>2cXy1;
z;Ws?6`p}jFUolT}_c0QMO)L*QU-bY9Z!<%MXO`{FC=SXk0fYd~dzEwR%N+<zL0!uM
ze14*FR66qO%E7R?TDL*4s+zu2DJDHKD-DOt%j130R`$E7#T^wb3G*|&B(-Vc%U1gz
z^II}lS<hJAny2Gytk9d=gaLD<JhCA8wN97$!Sex3v{oO14C}F)u0ePuwKo8Iig=9+
zzvEm$c9^m3<T%7Ar2XFNk>r7aK@O!})mUUUkMO(hgp9`<BNxrJ%3(kczmS2TjE1Ck
zAPDpOQlyXiNBEj3zl6GT)^wD`b19)o&y}SGK^8XN2k>6{K)#S~xD&R$aTC^BiVx05
z_mz!<V^a-tev}>Y8Yv2xH9Y|5jF?~X$>;Rb@*nu}B;V!3OIW)5`}4QhU#V_?$b@}@
zqK6aTp?Us9e=z^|6t>+lhJO`JCfMTkTvYuq&tUuI{yWWoH!60Ch7i2Z?fG{bR1**r
z2fc6%ot8fQ1z@nmIanF!zv>omXc86XKC*DuoU+-l(pde%lUXZd;R;Ad@8z_RCoC&}
zi;2bPo!wbV+q9<5$uU3Jpv||ozpgk~8~&7I40WHipA^G*K6}9FEf)6zB1^(Al1A8E
zR|J0_tuL8s2Wv35>+&2JSozGX%4q09XB7rw{}ps!sjc$V?_bcBz;QOXQBZl`ZMEv@
zB^CVq#I!xVHK~N^$^}3)98((MNaW4~V0;ZyZZ!ImvSrA`nGge_#-xCLBW<Wd`hJbB
zQ_^3{aFi`@fi|e8rh*VWX1<;e=Le4e+q3+_rrdC7QCZFC5g+^b>9nePD2SaaX}?Gc
zzD?T_d1kw>!+;f;>b8c0SWFmJB045GOG{I9^13fF=5bLn2Z-e4yQ%ai{Y8olxHdF5
z?%UXUZtv8H0L;zx#j&KB9SuBcy|uaUpSMU*P1=z8l>|rU(tat(Bb6O{i|_RQ5SZHX
z3i-U88%B!FUly22cqTq@mo1GTPB@+#|NAL~Lmk(II`=q{m}f8+>s+k5!<gaZU8@;&
zkh~zr?#z{e4WMo>!CHs^M{_ld{rmb>+1v+`hY)jS>45!dmQ%_|+o+GIiEU&v|Hk=-
zl0f%>ffZkOrpj990t%J)(AAutVr3BmO{$9wyxXB=V|v{UKB$W8DyR=m<whf4x1XB0
zKq#sEz9|@Lt!xHDmzsa>`3x4wEIcqDh7iehZkpO!0jr+`7E=>rVkrSE(U=D(7FAy+
zzJ^Dmw>rI!!g<MI`zT0~{~hdy1QZoSk^juuU~OX>B1nm9*|3K&KyNvGJ!&0Pk0gN+
znS^>AWHewwNw$CO)E(!v+c4>J$0PAvSs9<k5)eJZ#}pZHg^q6{S*c-(ekCHP*P(Y+
z@mSlr2hQL8K$zW&qsMK*MGN;6SzTRkZ+?Egy%d_s)wNY+FE*r%BE*|c%gb+b4*BAi
z5O#aeKSOm{7Kt`uJAS+-t1fs(8>p1xz4hu6^YhK~ko&gEic&jkXm~^M{+~Uy;->sl
zOkoHYRf}trE0pvnhf#19Ye5g7ld*T*9+{6ZfH-I6wIRDsGQ;}YgBWGjr-CB$y<Pmo
z?+_+82NxG-b3;QD=Ynr<+SuH{Qs^dsm&V$}vQ`=)$%1lRqVI%&TEB0sgNZu<O8I>?
zbQP~=e@K|R$T!K!vHK69f*fpLSkKfa(jnb5e}8{Rv|%B#QmO5a=Uf#0F}a*2P*j{t
z2T&tVPxK+0e6F>tIgj==%W`LK&nbwDsx7SC36j&O-e|*rq2v3qHTb7OV2<C}11@32
ztkAC~cjx@b^*mzU!B`ODj@3d^fwo2ePE!!Nym_{f4ZUwAEj392Yt+nh=H9!8vOp0e
zHaR@O{H5Ag8A&<r{mGQF%(?S)+r-?|7Vy?o)xwkASxzB5y8q_4CidxGL0o{+I1n`7
zAhNz0<5#=<mH4^dpoDGnt-qw@pp<!*ovmKa{9xDglW=4p_&BTTU_w(TMZ}|{t^*Z=
z3a;Fxfn&G|Dvv!na(+dl$NMalj*tiT{tuni2{Bo7or_p38H~hFC-Pk}s9NcuRp9QX
zZTA$BUIRlQV^YjxT+dQgA`Dl{cMT^5+-ZXjN5as&35HAxoR$635eiBfuKwKF{YF^>
zf{{ux@}9BRIO}%8R!%P^&}to@hap7Lt9<Qx`$LgR^zfBgeQ2BHz^Vgk5WS!2@J^D5
z>(lSd&CsQRf`Z-MU3_Md)81wuV0CguD@ANC9U`%C;0W+sIU6t|26hochWMyBx5vB1
zvz(%!HKiblL2A?8M2t0?9JwcQ9u7tUO8WgP@dUYqu#SjNaieUkzSSRpPK(A;{i!CV
z*t(_l6S`FvRFVxM$T^P}C@CrOd0hJ1ev|C;!JjHWFE2me*oaiTR^Lq+Kxru3to|rt
z%DS@i=%qY^nf}?8fWQr|21-eMsAueV?>H@_Q7+fc>}+@6XKnF);jG<Hw3l1ZHz;@p
zI%cF-gO7DJTf8rLXPulnTNo`N0<+}64~)x6g5iy^%v@Sv^V9akd++y}m!IFm1p=I)
zyZ!9!GTwK4%+0+(+osM0-~>2Gnc`MG{b?{juaXz^ADoNL^2v<rK=%i;?)|O*qnjI#
zVUDA3wnadr)f;9{yS0vA1=^k-yjj=lby_f<`Hzk&53lFPAKl=c;P#!b;x2@SuH^Je
zHN-rlcDLP+b}(Tc3EAkFBYIqKua3u0Ma7CM%lsLF7pL2nc7`PY1;eOhbL(*u8La$E
zlrF4DPh^(rA(Xw0fY18!zdktp<Z%4lB|i`u8T$bLR@V4T;Xl?KYrH0ZymMRc50vx}
zTn`d2r>-&f><+D>c2*s)4Ojv@EHeoH8PER>K4&}|>FaRmPfd4;Pq>_S3XGMCnRO5X
zUoEfC3*u$-ehw}0%)JWru4@2W>PLEaF8FBTmD(COV4B$Ij83w^&SPZaeOr^3(E^1a
zI0$dG_#zHPU%28*&h_oqx`Ps7SIs4gfQX2YkdRoswj74gjmGjcli}ZuW;`WAnwfH>
z_@6Pj@7sI0IEiX%^K*u(Jm{Md!7eDG07&_|0d}BZ?p7JFXjhWeQ&(3|P|(%Y6&F`A
z>n6u%XW}hWy|N2ec(3<$`jMcu9hE84855fruCJ+Py&>d)^b01=K-kwW&ALCx>y6bH
zpJA$#*0+ow5^sW5L87xpjEf_8b^YCmBhfzH2T%NJ_p0N^mbvtyji}~L2&2sOU~_YK
z8~#{0qap(E`ux?<j>VdU`q}Cc&Sx4kCI;r!2i=wwN)Pfg#xwt(9|9h?Gx>9$*~1tg
z;hz>943F3`U8#U$cqKEf-^cuO7Hx?-F(WP*ZvYJ@j6{G3B7sHdbHhW{zip?O??G!`
z_nSLA*dpWu(r(+ewiJgKAD)(|qPg3*YB0%b`vHi0Tq_p<gU++*`~Tw?@c;hdmv_)q
zmUFl5GV&@Nq!PA&rO^u{^gr$=UN~G9+84o+QaT6&HS`yvpBJ&vTb7`g>sD{pK>Bo+
z%Xl>APMd<f7;=!}SIJ0p@P!@*CoDa^ES>W=41NlF&zH4OmyPOfYx)x*=eZg}D)Nk&
zl!OTXFd#T7D>f)IJVal{WMdHnbNgV=$lTN@r#L@9=#6e}4yRxMw%yHkqps9QpWPrR
zlE95Jc*pA~RdboQ)%_7RP(otRK!hUv*RQa!P(&0Iq{}Nur@DGJb~cu&+1Y8at$Z{F
zKr#b`o3DrrcGS_Gc;Jcjg9sm_WD|y+n3g*`21tSM(d*jox2kh1s~DM`47x+1gvppe
z5dJj+Kf53r9IMcJ)41v1(5JaxnuN`<_;1q?N5~b*MSM(LqK2Bv;tVfr4$9UP1&NIf
zlL8u=I<<$^XQabZl)h>4t$NmJ6`1h9l{HEyG{4^PZyiE@A}o~8HJ~znTqSq5@K9$G
z1gPrcxzJfo7yI`MQfsrC(()=I5_X}lUY{i-A9=M7OJy$IDA60!sa(ev(fRJ$uX@L&
z`iCbN7{65J6cv`+nK-@_G**%Xva_68E3If|*VS=wvvN(%e^ZU)r-KwGd5DO4ce6lf
zI!tg^S!bbN&-M}DlHVy0eDH-N_#a1W0sCvDv;TQQo7#_UH_S>#pEQ5?@WJ*KPl&Y2
z{J&cQSy)+=l+rWM50hYQ8Jp)&t~Z*pvqm>9ulpgO?vIrIOlgH3G*xdvlV^e`_+MH)
z+@GIBFGK8SCm(*hkX=}52r}#Pc>Vs{Hz%X?5EGSD#D~*W(ZT{^Ac5N|as&$B_7<6j
z-J(Jh{DY$)Rff_(vlBAH77VFoRnfnYcDac|*4S9vK*%9v$j+)#<k~l*k&cWyBoivA
zWUP<Qj*R+9W;)jK*(gNVZPP2HIo+4wheJe$7y=Fl8U*F6yV_|By9$c*#v&#rM%X%<
zZJ~|M&!_As3xZp_zjN%~1b$jsK#Jx(#|V)BTM8wFm@<F1R_Sn=#puSpMAx-T;J6(-
zos!lQ`>v%rx%86oI#|vfK)EyPxRKNqq*520(ePNZ33VZ-h;FW^qazCV8HcEw4!Wn5
zB625Eb3bLRz+IvHoTV(oqSILtXFJ<>?gPx9D7jeAT)?8DqK1Zmp^A!%Az07${zN17
z=&OTD?!!xUt3D5cq@9)1=uDrcw3cj2xZekxvChx_@61pBRK>aGTf4Y8@iD4`Chji@
z{Xu`g4^-G}C}r9R2DjDHR_a^j11Xz28iK33x)c1oOUo%orLrhaQ&|0E?c<Jz)Bnav
zDcMFl$(xsHt8Ki2<5k@E&Eb}Uhz>}X&jbX}dVYSMm6dfqalf!%6x^0J!<{<ffZ5wz
zt>1p^uza>1bfZ`tMe>c0yhm06DkTJbolfZ<>P}Z3Q0GC<XNE4c1+fcQwn;OcPP9M6
z1`7J5>aPh2yu)qY*s%!DsOY4mVt8EKD|DaeWhJluP)rAo>3`j@#o2DiFRlW;g3ZRX
zq|iukXaoG!ISp;G^Z&}5$p(w}a5OmGK$DD2%)sNNp{c3qUH{B;P-|-GuiEt_qRYW>
zbfh-7gQRyfGUjNf;wZ~MzZ|Bq#KtaGqs|8M+5VEWBaM{&OwA#ZlKAl-Eq}3R?syE+
zPnGUx=m3<qLQ?VA;qStT?=X-4z0bea_rqRp1BazS0T=zJ?aSFL%86J8x#UzdtMyhx
z3;NuAUBXa&sT)yMD_T76u#(gTdkLWQWxW>HgX#L))$Ow16MnxuLa*lMqcMa;$C$g)
zA?q?CNN?DGT@~I~B!r}R*PCx^D|(QH7$5R8&~o&XPUhGBrp?g)5^FZE_cu#O*)S%O
zmL%s{KcqA6dy^u5_2VkCH^ChTXK_K3kG<s)pnvsd>R?%u50Uu<x3oUuK+Y%%(XQl7
zMl?<n$tk$k@2BIZi(QukI9`OyExA<8)Wro=mFDiJb6O9?ep{0>*Ru4)e?6s()+@<@
z21u{01*8@9=(YV{YwC1Y)bFx5$V>Lu$(?K=$Wztb`5^6&MZ9>}w!o=;gW4xpcO7oN
zK4GXIgQ8@uYXx3f?vjOx4X9CK>L;o7%1Apq05^}ozShj2)wtvRtk~tfcZ!Z`TCl}`
z@0NINXEU?k;;^A#DYykEv-yxSy(j=kNQwU23(3+-_W%ytRO-Pdeqw^^{EXm-AUmi1
zQ|b%fn8U!^4oKfvQ&#TP^25)+!ah$ePtYvgEZn@SE9;uN+A*1lsaI^cdBPMgRS%Ln
z?1yF8`=-V@G|Tlq`BX13z9yuK$0b^i_kXMivud0iUE8-^UiUMfb1_+!uvG!}`Wrc`
zw03yfec_i2hV^dmlx%`y+foK*M)&x1m~U$h#fDR@&78BUNo&lIvrjT2q?1WNUPx|y
zjES+d<d@o+^5dfID$*~{q_{-=?~AL`>^xk8-GN3VG+;O*@YQR3>Q(M~)PCD&Gqog~
zOwqf*b=IV0`;MpbSB9Qmk*`3D)BjxLrkh_C987gfUR7#kc(Z;_CjyAdXd5eTmQC*f
zHz`4byWtKsYO+KBU)lpR!-eSMnL^{ddLkq|41pV`PUfF)*t<1w%zcl;**78IxM5ge
zd*7(E-%^zA9cH2_uY!EX9zFjmy+D1GJ6QT%UwpymB!aK~JF`x}g>a7M`|UsWF@YQ~
z*we@aa`G1Db~P2X)#Y`Zo}CmN<`5^JDjix4hojU?$DQ~6E>U^BSRf&<r^q+KP84=t
z-MSvH86Hx~?siX2eJ!dfsHiKdsx5@9*xZfG43BmeI{+<R?G(R$J3A|z|4L0tfb`54
z9)hP;o%BO`IlzBre^I}-G4|%9DO09DSXbhE|M1LF9oaE^a$Ibt{x_ACxpj6PHX&a2
zcd*AEc2Nv;_rH1ELyhhh7Jg14wxz|j`CrK09Q?dA!;5Rveafm}VG#D9?Ww>e%jTx2
z)E}xsi|tJZT6)~4RQ!Yljd@P4<|a4ShkFouw1VPfC!nMK@0RmIUkI$}$<58qE5s(i
z!GSpc=;JfK=V9Ss=+?pfS2TGIY|FRJq*IglPrrF*D<BvRCc7XlciyAAJ+}60gt^`k
zMXubESf?l2#~M<Oy0x}_Jv!t<&G@(~zG{5jc2%oF_tM+z_qKn%nlRK0P9{+jJE~+M
z<`h~e0=Q57@xKWeIsH?V_av0?kdt<MpIUkx*qwu%I}k1Yd2kQJ-0iLXf6N5s*+PKi
zb277i8O)m7&CQLIQ+<6sJ9~i5e9+W^(D35;goKV%Oe&6-S94AvH5?i-CY)gS&AO=4
zO#KE<@*+o*%N-~(LP4aUTjz)NTTLDCbTxax+-+l2(SDkHh>w?ZnY)RgO@I9@=k`7<
zKd7sYhns`~4?7wO++EaqGhT9geGYw?G5LkD%HRtp0r_dQ$Ga{+!iy-LR6EhSS@A&d
zlU@Ut+WL@UtqVg_mx?X4%SSXAlmloqRfOtihidxcv=z}W%{4F3Ma1I|c9d_x_`BE!
zcvve1;(zdtCnEOz^BPgw6ahuW%JrS21phD-_7b~b3hbgfWYvq*>H<Tx<NE+Ax56~S
zT7I|yM?Y-RLW41XdV2^@Z)oejJV{!|Jp79f*$YYyjp16rV@w+n9a5C(O8zhO@@;l^
zY;Ps@Gu$0hW@mu0s#AE0EgCQ&@K_dr;^Tqz(v&OQ^Wo$D^F#Q?BLO4CxX>~F^_S1#
zf#+XM1^;w-P{*vH#yccvX1&&w388UDDyhk;^bf4>D31NvHVb`7{pTh}CRBy_?qEu|
zY8JnP@f#fet5oX}(9@*?J3wH_wRM*+vU4OL1XU&rA#F=XtUf{hX24L1808gDT10em
zqFYHyYvp!~`n~3{g01Iy_*24A6{T;B{y*OaD@es(F%4~xj}RZ1kPs^?FOvSd9NGV(
z=`8%Bc-*cn-QBUIq@;p$3kcFkm$Y<A$4EB_NOyO4cXvp4cXz(?{XL)eFWA}LiF@vI
zopWaO-U`IhvF|$i7nI~h{QO~49>E3E{seXT8Q;0ne@!4{3KRVH%NNqZE=tzM{oEe4
z-oH$k9re7pzyNms37Cl&faDr2Toj<c*A>PBucFztRRxW|SNtPuqkUghd}92)1AJV@
z$CybVpNRLz_;L9H*P+C%x^W`nU&J<b@n@ZG&?4ky#H9GUo{;^FC(=^WNXa4I(Q~h>
zX{X>Qk+!<p_6oHx{p{J*&MmOK@?9k*72k7i8jJV3#l0{?tepfXMi<zh^CtRWt_lS*
zl6jx^T2KAxjAA*iebUdLX#NbBewf{(m3(YoN+Q<hC=s(TU7NTc>H0`Za$m(C`KxBt
zM0D@9CATdLQHTzlaVN{H{yWqReFXUMbCF}lKv{V{Jv0}><lDR~xJ`R#TsCq+Dm;?!
zW$k(iscD1|VqhO^68x>Vb9h8ToSdYkxwM$6jG=>a7B7@#@OEkJYVDCxUHEvCoFmD#
zA{@{?1cJ>H4ekiDq;B7*L%@Q2Z#7YkH@|M<xApxVwcc7tcZLTPd4*#Spq@YsvXMm$
zA6}4E(5wW+aH>PeCr8BcjhqY`+KX#5*_d}%Rp=am6<_SGWu_XVd-dt|9**PvsQHx_
zAF3^|mC@!n;AddO%jPAcc9CzWr=7luy=4JdsZrYqvMxng$5w+&Q&pECLq$o8dH*YR
ze&|)<CVJOX-jYm2vTt2lJ`{n`=rtyKM?p@9BEZS;ne!wn4-bn0E;*>>6bZ~{jJh!6
z0cdKiUU1sWf3$gcx35CN!KRnSMxJ&tMR6Sq(5=julZ}+6r{l(j@>2TPRD2cPm3ck;
zMOj2CAvrh%(c&@wZKcX{&)S0u+g=r`QEYcT#Ql+~IHT%6cI;tMB%f?LJ~fpTa(rTG
z1)CF(yR)}r)cSul(2OjFB{m1d0AaFgy^jHzwW;j?>Uvvc0gt<o6~$UlpXCxzyJ)W5
zwLl<tYJPV%v#cO}wC>8aYb*y=D!aeUA7g6|dVmkMaF_i<aJ~W^B;UX-!UJ;^^3V|<
zJT1N|_;V+&Pdz;>6b4V$Rh8Uhu7!RMSPVgcT`-fn_{?j<A$)ufP&k#r9QKcCdrv)-
zJWnmgon|s9<WK~@;tU)L0|suEm>k>1(xE7;+eXGYx$nVAs%>Pz@1B*h+~R;JDW(6p
z_4Uro_+W*!fr$7rhWtUdf}iOFOV4N(L1>%P+k+zs*+*?UdYd07UVOBzC3%GqkD9ge
zG6S!cpDpRfHpL9Un4Bwez}Tk29QA{WzOlpjKIv||s3z~X&@+5^lF(`LwLo)TYUYO*
z{;=)l|Bl&frOp9^5($p8^mj^H?9u4J%;pg9emx)W%PEYDhq!p<CDhn1UBLfJ-lTub
z#-s~{hP#qJaR^dkWxYA-yL!5>%`+XsAYr{d;QTZd0p3*P{~hIfr``5{#|S~O|EPd{
zI=`R5e((6i0!{D4OUv$9?cjxFmaFq3W*AeA^6i5~FH*q0v!Ot^abux8xj&^#L%=*i
z`L{#f025WU$u+(aKyX?)`Hku_B0swq%MtX16$$)zO)4D!_f)G1%POLcm?Nhps0HsG
zk@CPeYiXIqw1?l-1JC~0sd*s?6OV%Jgcl1j`{#k3lo;gVUwZk2-Qymc3aXG$_dL`!
zG~h<b8scH<?fbO8DN#SOY915~oo*RsQuZZToC;|<9aB=}=NX#B&+xdiQ_xmdIsg0x
zeK62HN3t*G!w@rgQ5m%{_Bgd%V~Tb6@(g$Ykxy6`2M@8#FBt`JwH^Jn?JJggp8C=<
z>$TOY5=j|$;ha6$0pWxiG#sO;FVxf0d?Jd-<d5<S#}lJ%Y#6z}x?x&%PM95IY5pS^
zghGSCFwI}!D~?|JQQIpQ%;|D7Iqn6=j-#TA!}vhY%rW$LXs<4g3qkq1_&muqz)8px
zyq-kC!WvkMvc@;dg<yG6so|b6>S9yR1Ks1JB(H4-lM9&3+jop}nV=V`ZZ}}DC~J;E
zzWeM+uNAR#r~cdrIdf*Ys&BP3>Y}jGdAPWpMi9;8s5H_Xf~((8wYd>Y0<X$)AQrbi
zp>~amO1&b@Ed(Vm_f=3$L-QJdrxJLz`?BvT+!)_)%oTq{psl^KT<)UX>U1Qs{fVBs
z%v+F9Y-<`EaNfFW&T09EK9a9pRS3C#1>Z}3(cLeT-z<Bn-)HL^ZDD=qsJTLPi__xb
zK<4E_2_07O`iRoR*{UF}s;n5D9+Mb{iQ_gmca@cW!$7EdWP4k+=9>U)&)JZ4>|qaK
z3YO5dW$y39Ak4AOlTOEV({9#pV*OE?thT)zJXmTF3v6ow2p<<kdRPs~5<6YQMdVD(
zH>VTM|8%_5vjO)tdk$bX!Dm%EW_gYDy4q@ITl;2wOw_zz8P^5bHb&+fFadYt%Bi5B
zD>65hbWl7PV$c3ZUlt6LOSftcW;jr9fbE%)AuacRzsnU7CND>wZc*SvM6~~3*(3cJ
z$hY%<{&=01K9}p@6qhH7+SY~Su9n_0uveb3n1^Q6FbpDZHveVT-nRa<aQmwJ*B2hh
zHk2?tViCP;UH23n92omVFIaDF`^y760#el7kI=kCTs({nR841&yMvk}kX{_mn&w<l
z%j>$?aKLUT1|#6<OD)kS8JU=%k?wfc_;fBcRs3AFIJ-KTn9G_VSzK5)bZpu=ms-hE
z`7&N7@O~zY)FUf;k#20w<vOdt)zaSF#EFc<W|vNK{YxD;odEl1e)^8RkkILa7+Oj6
z!ViodaH)pC=P_MpvR>0RSFPzOP}yimyYGt*%g=nZHjOx946!r7xERRX!(T>#xSc8N
zPejVe$c>MA_wO|Zcb4rYO+A<Vy-%p<NL~eR^JDYsNy&r<<Wpc(Em-5A^e0|T;k%Nm
zqKfhyS2d4wFBjXhnYojlqs3cKb8U-DQ#0?~gBT#BW)UMMjqNdl+swslf?Br~b7Bno
z$%P6`9WPA2FD}&~ZSO#ej7){=^B%3dRPF!s%E=SBe*^0YtW@C4?65Y<DT<+X$6>mJ
zpYrocfy)=-3kMikn@{NiEeEGRywuIe_jHdb$w=Lt1fyUyclT`k@^Nd!<VDE&?kC1k
z0cS2XyuB7bRxS(z_Yca3zTJz&j3Edcr30P@SyZB3R9RUDPq(4p=zkKWaXU;Nsh)2c
z71SJ%Ccd?lWfi){;F8%coHhs&a_09?Z6eX{&p-6!#G=aUwOw&%mpHgNIiVwaZLBmG
zW|r+1MK52!ub^<~oLv?c(~#5D{{A~K6-YezaQ>|jQptaWbxzO1nO~FXZn?qL6W<mB
z;M(>Mit4!v5!_;8pbLD8U-u%?`L;jN&SaGiZ3=#&9Q_Mk8W#SExh<bm|EP$M^M3d^
z`0H>a>&Kv<aztiigo&xd((;&8R0%BZ{(;U%Ps@|vV*JDDq}C*_zHD85W~EpTpaXXX
ze)ARie_zJYMZxrcN+cdfoWNpl`D(^yk$p^jeEhzrMe}d=dbD4<rwtX*SVqQ2$GZM2
zK1cQ5ug}<vRbb~n*tUD(TKbH5;X7)qG2VUVk|to5K{Y1(tzwcryAqO~5*MrXMd>vt
z{3(?#5l^T3>-t0{`H?j}+}Qz8U*5PVc315KD}+JEheG|LOYtlMHUf%5zaV}6`c+D5
zaA1I_p%>kpl*g`b$892}rmJc7ptcdOAdA6=k(k`HAgi;e=T?X>s36hxOKz;AT*?)a
z+(7ApFLhue`H3|X5R-;^vmQ{KT8REPP7y@8v5r?8(eNAZA&5Ug?G`{~O5wK=l6Fy9
z<X~$~yN|n^3yuPTMlFb9YzOUs|0oYA-W8DwY!TgGJ36xRYe8-55n7kCd<}IcTz6ds
z^QonL(i5Y@{15@I)!WkZ0ea$zw|mE7{B2|9xt_-?o1Y#|#(Ud~J=lNOaP#~9@0%7A
zAxfCtSv~kjWT=z?q3GLjAkspt&b!pqR0!k-+{7SFQbMh=)R_ck&iDnV_+gR<C3{#T
z{lMUC4obOYCm~4B=r{AfY5k;!1Dv$PW){I>;A2G&*%)D@a_>Q&!7?M92vX$M?7V#y
zgBj1YVbjuxgQ5;RZY(!6bhs}Za5Cr5&KPji<88jg@VkI%>uJA9Tx}%qrCqUXhwFAX
z5U-M<-5Z-!iP(h2+jA<Z-%{sAG~KxC!vqaYZ*T8H;e{fo8-lW70+}jYC(P<a+b4;t
z<Y;kWX_d9*S9|bHLs$0hnwElW=&*2q!Nu>kn6%RV7AZuQ6<*&LZof-Sik-5)g=I0F
zLokPm8W`%9LAz)p(eQYI`y_OS67slWbe+<rG!z(iQnnoI^-$+%c@m60XwwB<#+^3N
zUGxlJWE@N{3;`!FL-Qjy0s39>p-_Us!9gjhV0kjQl_sZy6lO$SP27?uu@=DQ&Vhi3
zdfdyEnflw_QRlcWSoUQFy}7ZA2SZZ-yCz@~U@$Fj#tOzv6O#Y>^|$lON=nHnDo1Qi
z@BKW;VzCvU9`(^Gj+{~D8slKbE6?%m9Ae-hjAmQN|8;;F%~ela>j2jR7^5G`To4(1
z%i;MwKs^i_oAi0T^W=z(0AeLaNUj<mkGlayjoXX`S8an8DJk6#CHKPa9@!k3IlcmH
zgTK@#?GKE_`aci$Pu+#~UtgX;t@3o`s`>EH$a7f%<Wu|*SxI2<22XYN?FMhAs}QeT
zdN>r)-sd8!_=A%a9!Eq50*8?A{BXwnnSqWMO#4`r(!|84tCRRSzaho=VqG->0RjFp
z?KY3j$-d2mwz`mID8-qH8S=1@Q{a7`{>BnrvD^G*7T{tob<_hMPp!1@a0d-lsr~%+
z-|5_M*4EZ#Wdy8Ezb9^Le#fDKZxawa2B=l%8OwJUd`JvL!qEjDGp~ofek}nTcq2G8
zK>~C<k8R`r$`a-XF8dX<PDYr%31QL1B?CpIAL}Zrs|7q4`xYPn#y%)|0#+)ng`T1<
zq)gW{+h3aBz78TVq7ZSz^<oqya=Qk;$NL;<1HCD7{DYi~@h4oS&+WNlXbe6+-N<Mp
zIoGGzX%IeNrW-(Gb&bJ1_7PosgowxMIbjMLQ$|+wh<GLs-=*!&pp|AMzI_y$(TsYc
zqTTNkCox2m(zT=_6L6O31rQ;cM8;MyFrEUf-NHSxItS}e>>^}N61>FHAE};o_+G!(
zqD@212zl;+%wJ!ntT2DcEzFmd3?_uf{9$^tpFazViwld2z}}^ALCCh%pTLrvXe^xL
zPMD9eYmpG$@?9LB42LPs^}(YgZ;YcLo8tdm9>;qo?lQ_Dk!k0^TEs<j$4}_rp2+AK
zffS%(!%PEeHlWKg(3j|W21z0QiQKB$r;SVCKM+7RG_>B(;ZTwb=3;;TLW=(GW2FO1
zhjgr{sE-dce@cnWuNmBo2q=PO)$hk^=w3_KRLGj3Wlxc_`s#HuQRRhPRRM=zZ#QtM
zcr~ZjfAyD;A&^BAF56>Rfrr6A@~H|+MH;<<PO3dV`mgO>M5$2C6HFR+;ekfOJziw$
zP3Tw=WC6CP``s_=1$2`$db0`hlJ9@FGjw1y@r9}${X$Y>^d{#o+`hOOlTQqI_n*DH
z`1in?E_##FnplS_1cQ$qxtMC`M`&td0Gm}ZQ{gp?aSFUEdFw^6jMn3aZ?|3w9^g?@
zR010h)0o&PUn7h0t(Ig(dpr85-<KJ0*dN)c-gG?bd1&FOXI-@$nP&N`q`RTAEmZ}f
zvZJA`Lz?D+Klh_9{f6fs%==PErvu-7RsYTXo*Q3JYcG|gNH7+maZcaZF8g@byM>@&
z?~L0V%?F^eZN3xsAr*YcV8<}#6vF3cERI+H<amiByGp7d08h_}NhQpKg@s9uVAVvv
zdbTX`WweLdXXYYP0n%QU;v2*|E*FA)?G!M?gJj9L$M!B6PZ-JH2io&A*lTx&+VNS0
z^Sp#Ss*dN-1@*NVF6U0c$`^kr@MafKw<M4J8wcp@|IVt5Ds>nBt|>LV&uEelZ@tbR
z31-8l(8M{OvUSWju6N?8F26CTNboE1%1ywtJD#`Py{|4Y{PgRT?H+FERNu-AvDxXQ
ztGnCx(*ya6xkP=jJns{K&@iJ7dH*qf2lJBJg?>E?U)qFnSCOs&JmDZK;sE)2KCDaT
zx+~C&_bVdJz7dN!UFTmdk=qU7k3(KWJ_^gVHmjw(Mb5PjYetYU<b**%h+M*wa{Z?h
zW06b`4p={qN5)UWh@8SxOZ9cm`DPK`e2br(88kB1Bh|UxmV;%yl$WZBboXY@Q5FN>
zLWuh0iBQ+4ZG#v{KUO<{TO&`6?`wRub$~a=+?x9H$J<V{_-R^Rb<Uqn=!E3>xL$HO
z<`|rT1<#bqT3nbOc4BKLE|Gc&e^Jm-jrum}a=sz{bbUXAD$nDNAVJf{VsV<TvBaNz
zVngraj4VfF0m1FdLrQ5!xpxPj@e%52%}!jKJh;#Gw6+zA_wBHv@LoizTj*rMm%!)t
zCeNRzI3^dCb>(doG{hvPr>|6O_T%4=4MN=T<<gup*HP9J`9?Y@TNJ!^(wl`~Ms~jT
zGL)HnERG5to0(giiQVl(OTQv@eOFVFgI0}DdlMHHUzaPurJyv#;w^Ake!VO`&o^r{
z4*xOaWL4LaPm94P8TN+l)j1;~(*;}xZiIPRm664DJT)}&uIk1;Y`4YZ+*(aNi~nZf
zT0Yx~pW-$@+9e1{Do%Haq#}Dnk@4ZoLRj4!iTp{r+&}tTeCa(v+)e@U`s<BU|GtC}
zFEvS7``-*FN9#S&>;N<|WA<ooVN^Yf{>Id{9uzMBb4_)8R5(i0dZ~7qP!KTs4tZ{+
zdpcB)ZLK-DuAVg;nq`6T=g>FA=XOkf^G)-kF@cJY6xBgl|6_TB8*Pe-`F`iXO7A(j
zx|$Vi$Qtb5T+#iPfYz~1c3UF#2KRfv{!FBFRfOQV(ZZX(P#h$<b$oFghV>>|08$GU
z8T1<-i#%F7%AVe@{^Fa!I{<xuWAWraiS8hg$0fHml1~RE#9c*qZfZdYIgWM`nept=
zqADHUZJvlhI`pD)S{c*wb{=Zo5F<c-y`BgHf_U}RF;_o2*DeyB$rXqrCuF%6{}7P7
z^h)Uc4<=C^^nT{acu{j+k5QPcwnqvA#V>8-AZcdv!+kJ+5w0l60=irPFOj`oo7pk9
z=bk6==hsTjhF1jd6nB0Db6V>Sn9D;+`Q`_I8Y8t|x2R$7{EuQ;6HrYXPH$qg2uA)@
z%AFpMLybO_yZGVO{Mb>_?0-&i23uCpN*LJRydJ1A6*P3++S!DbG?IwDUP=m+5ZB&t
zDq$!tx|@Ok>`P+!A8Z0FS60LHniy9o5-YA1ywp?uoeB{#vHGkae{hRP`U)O@T>UXs
zrpY$$GHbS_5lWA*^b(zdIYoNgyCrOO%*s;GUR1P##j~y`D&-R?Zf^<@BVG((2RC^)
zX78=+*K`Fysn6j5eAn62U9<{=SC~nb|Lb;oQUQ4FF+!Myp59m%lNcUC@KcaaX(vm%
z&-1dqfYBsFC&cR%O&GP=gkjY<{Q5geR>?<4{k`oyb;mEpb0<pcO}{VaeH8Nq+t^X6
z0+t4zvsX^{SH~fBQJhkG6uHp7g6y+@Ufy>s$cQ8AwmxkPqBC`}(oZXH-=I4wNj?Nm
zvw{8}kWV@wPUm(%d5sC1nw{LiyLLGVymL?%Mz3tW7oT^`H)N!`9?0ys<|``yeenHH
zw6enFajxMHBT!y4HZ<M)%@r7-$=jgYj6;^Ls6EHoRs;LNQ*4Aan-dBJ$G6ooa^M15
z??OO`f598_dz{~6+Z&S#ui~GH@z(nG?goH^xy)KzV*JCv%z5lc_xok?;TY1oMBusr
zQ^BNy4)VWBYgIRd2eXa}7FDx_SHwRXr1nLvzrsc=NtauoMZ1EiI-nS6Ak?9sSkhia
z6p@G?RG434==lVL9(pzC?`I!}qGu4>QRr$gZ2uUf`dqz1{b|o)P73<r$|R>N66FbO
zBIFpMgGMfkgKn~6HII%Z88gE*+!3CMR-H7GqqCm{Ok+}j$N2O{LKEzs3IFwxQ75<S
zy?<|ycT`h!RZ&tv9{!i+)apQ+_3Q=$4LCg673xpqm?RQGkN&}(pw}`hC4RD=bqP#y
z|7e50)XNDbZB?wmGMgc1pwiOQx&jJPDmvfxD!exor(^O{waZ+d)=*qiQs8<3Q>?(G
z++2{T^@ZHzx*m)hp1P*iPQ()Qi7s4{cdGJ3e>4a8N?uVf(N&;dB=9<tkcj+~fVH*0
z+1cg+!9D$F9+Wy-k|b@AfmAgKY$oo=;`~$0H66X1EmjJ##@~!O#w$nzy4t1`HtO^w
z6m{m#6BdxZk?;jpkHwd*A~w9SeY|ybTU}bLd%b`QJ7s3uqaTHDLjIwlL<9!ZI$Lq^
z7NtY0)0zC(!jd4BjQr1eC1_=O=5qV&AF8ImsJ8<e&m6&!bJ2q0fxFZ0?|_52uK9BJ
zJB)rDH)Mn3?`v8@-Eu(2(2bSL{rZRTjxLxcyufhzlsad7r&f=DOLwyZ^Zp>esw<@e
zuABEqsF(oc?^vb?T8x*v{9fyOr$6kz$V&f-wI`VWy1?YMN!Srh$^Rb|S9={6skOk{
z*wCnVaedji;rxl7pXEVC#8r%L0zL&k>4e_<ZfEDBY_Ti85&&Ht=ycaA=;!Rw)=OXN
zrV{KSmL$UUq$KebIa#w&Tc1r?U7^Rw`NKR=qtX(U&+Dq{8nc}dIX(z=H`Sv!R8O}$
z`m?6<H*B~q%E+t0)1z3@n(lBOQ{)(I`=Gb<>1S`{ACK!b9lC#7w_f-CQ(?Q_rLHhJ
z!NVQA-weYr2SgSmTi2BGA88b%li;?5{G}i{PwiIi?zDkYDLd<tJT`%Qk6RNix2SjB
zghvRXxTmhtk4IeMxhjsb9bp)D1?n%)NM=hivcJBY(D_mMtaMmL8DPxvmcLLsVV@cA
z5X4VUAk+@$TD;zo1xWVx2)d{Jy~K76N^y&1FBFw=2Rc-^X4Dg~?S@xo5s`I+AjvcQ
z0POx}wCGh^lpx3q2mgV~fy9q5mR+G?mivLaYs<6$5Ma~SS`kW7CK?`Ud<ZqdYX9bl
zBWUwl<-BL^8<&%>Tv*LjXbrXc!eQwXJ=D}Mx4_pdRulN8uNP+Oldka|es;~t9GSja
zoI+cuBk;hQWvy4qHLdg)dqaQGh&5V)7lSFz`2_wl^4UNM*}TIAmqn}7rnR9ZFS8bh
z5KC^1g@XCRM75^I7lzC@Qvpxglaj`WYe6VE6XGe^TNXGU1XTgyRS6~WvxXIMjP&IH
zqpC%7+Q&SV6<}Lxv7nS?OhK!yQ+e@-R0z<pwE?^1sFYO^H;~7BldzK1%EaHH`4mL-
ztJ!@GJ=`_zmAAa0rV0!qEhey|bVH~wS&v(pdY;*Tryx2i>#==iZkQ9|*@r6xEz=zV
zWwl_xC!+p+`D$h)rRBM6v)*_o_8*$n3XI|OI@T@DP=b}0K}L0D(R;3~gKjOd2z&s(
z2;s~9JW-+4t!2?VJBo41^V1FoYQ$2C!(AI5E|hWALoY~!2EpwimUuKbgq`Y?v?ar<
zUjEB{U{G`xfJBn>(szvZ35!WISU9wfc5pI*3naYeg8$-SW3}`rCtl*xV%5|&brn@~
zwB{z}msdZr@&4xx88)r48VC7LNl&xWVZVYWE*72?HiZ~`&}WrHxn|T!{ffqME#t9?
zyeL_kvg~;VE7?N}Hbbj`Tzak3uRAoV)$Da1bl=8{*NA|SEK3SAm(IhW5Lki_TeKd3
zV4teL82xoWcrdP!IVp$zBE0kM7Cf(gt?C!~c$=H_`9ht)73&FR2g(6`%_OTPG~x&k
zx`GHEIk|rGiTeeE!MB!Wfee2}yQz(zaM2Uzn_SjHgFMq789_j!uCX+`+4%?9mp#iD
z5Z<KVxwu2~qf$LBJrS0XXUR~rO7tfrRu+%1NqM-Z4kIjK8C4!c&r1se#@#X47D-5n
zMU#m=z668|&zU#WQUP@An`Lv2)BTbE&0JN}(E3Ov9}}LTGH^#-{fufOQUl_@{7>Wq
zBYvLX0|F%V&F)-#)EtfRj0Kt+xF_e7RQ@Cq;vBmVu4=2<fvds?)K^5dUA!sY|MtKM
z_wbu7t;PA|j{}{nNy^I8)0s8~G5?<>_^I5*3J})ky*_q*pa1@!vHm~)?JUnynOYN}
zrE9ZDLb%{sLsi33N123;EHwf4autQ`U^wU!E>DGTRz(lm^JG0D)w}4u&?h2Ib;Inf
z2ZoOPG^ELkyM{NWe5W5SxXTup3l+(yPrZ&_O+R*F8#qUu)J%Fi+rHT&d~sRWX9zwA
zoKELGbUyqY^X(DQ3B3}e{DD$qgV`WI#v<&Ot(yk5Lq7R=^Y`7uM%@rs2iGd(zHb<@
zdir^}kG<|)={-Z4>!uIzOG5mNug35?zTC3AJ>$8~#FSTBiL6vM(W*`wJ{&@Sz#E%=
zE4<cFWJ=aHswU+5y5`nlp_D!uNQO8Ew7K1$?;5+Kh{^J>v$##k-tc-uNz6)gFB4I<
zRo$(q&Bj;=laxYjcDFzx8EuA)$Y8=cYbsOwTe*dw-ElAU+Du-j8H>c%lo7P!J+$Q#
z_ltxzn3>t7AvQSJ7~><w<U(fzvAUC2=vM9bvYR65ZcySa-HJ3&)~X6!)>9XNXjn1t
zxsA(ZQ56YcT#56dPX80Ze3G@biFsBvBP-8lU49JDSBnP?v!Ks&&A7zEYlH*!!3BwF
z%~%grE$;d&`LD@6z6662(FFQ)$KiMnm17cjroUbPa8$JYda{#nB^WRuh>GK{LTc`N
z;{EjRat;aOW!W<FuxKC%>ffFf@MH2-@&>O09`i!MzQ6NBiGlsCn~SE#9Y62&8D_g5
z)$0bQJ2jjxx`{fBQ#E#Ebk$t7eJft7?ko>l5!d!|cHA0!3-EBSk99$4!!j6zcIi|%
zHK)K$(K9G4QsH<sj>z-_g)Q#akR-&DEH}=nbj4wfnyhJ?k8-nhjBv;XIF+)e*c$I)
zsE~i0kRo^+sHs0uwnain=vS98tY|`AM(PiEJVH^*-f*Q>KT3(S#5hD~KM10u$kzX{
zgRQ{fbdLVbb%_Uo>W%E;{2J-kV0X_FglAy1uYrKdYtMv+Y-?!um?TPb>NAh;^+BAP
zg2_fZ<?9M*g^Ax>=gW#5KvGPKOR#B@N4e5!tlZX}9n1TG8+KeS;GsAL6{3D#`6PHT
zQBOuCcwLn0eZ0GxEW%~`o+C0H#>s;j=O>ro%=!Ie`t#uwG{4lU$^)<9cGH!VKqE|+
z$MN;|p61UqXBK4e_L9dfqA{VAn<DFLbY4VaEv0R!V7(<8&t#e`@Qq(-=v0|Y;OL62
zvBDLV<9N4oP)}l^r@R+rZ-fNET-+7iI3oAAqU^-DoP>9DSGxA1{OSs(%GmOtOxWi+
z(J}k!ePM$q{ONd^N1)3(K8DU^s*RqXTiR?4yv--0<P$4+nE6ULM5SUfhZg)K^$A!R
z`nlRY;3`H>BPAK|@j8bv_nJ*QFI$DH^kitprG?-Ed*wp;S$;J0Rfb$xM4S+uWWRug
z1rm%H<Rr}{J}`>@31Oq2p`c1fBJ{@V)rTE&Ryq>Yc3?01O*;WW-OX~U_lHXss7{Bz
zWt90}g-`4?a^9fHj+R<MQrAWBD^vP{rVN>ahO5)yo&avpQT|Od-7P@&sW5OYXOME&
zfBQyGw*d?LjrUF#bdO4zxr$pUYu$AWY#+PJ7-I+x`^1NMH!b@i%>5V2gn{VrNfBr3
zCFq=M7V3m;ELipND-au-8grMIQ2{=*3tJIn{$+VRrAKXBTQ;=k?~aBRERjX*@-kL|
zWq13)zy%=3ilOx`PXUo~*Q}A|J1Wz-`$~A2EB1eyIs1dS2{`{LOTNvM9&WuM-H>5;
zo&+X`>Pf8K*s}4IFj5bVeyucB_B#1dOh(~AMPF+D({bz*1ryarV!t;g${R-!S1|T{
zAN5!e6^|z)0%2$QT-Z=*w;Hqko2ksr3_5VSWt>O*+?I~?d8=K^dHsa)J*BVY>GkM%
z!{hS#RlDQUSnHEW)VNahH}j>(jY9K#joVLdWo|9mAGz51JGwhJg91v}_|P_XlZqsu
zc(2_NrN9zOZgHa-)R7bImKIOz`^Pop1_!QVDSJ?g`?hvJGeVj`{|3wJd<LS_VE@N&
zG^vr7qq4gGn}0@MxDUR1$w6?n-i<GNdy8p)_Of$D14;R%Pqz`K<2=pU*KH)=TbBAX
zRNffHu9{$0{f3%qo*T(pn;V(kUD6c;wWE}vL^N}9l<VI(E+ROX{0S&HS<mVJWh|h}
znU3?h!|{i?C+hu!-)2=1qc1K1>>kqs>dSGcVal7+*yogaI{Lq~IVcN`LuqV5_?%BV
z>9YOP^HgduKsO2-2d}w-+itYpz;xIqywMmgY^1^K*5R>&q}60`&pv|Sm@0foR>^j9
zmHQ3KvCeP#uV19b;h%H=*8cj1aYRB$OwL*9*m^;6YpgFoIHZmv=uKV#z%Lb{e8Q%J
zv1wRLIwZx_&tCppjo5y1nCai@b{qWdtMoAGU$u$)7F)>uo=;+P;Q}bI>$PR{L<qK$
zoLQ7I3`)7*c03*1$qv4ix3Wg=x@o<&@1;2;Az&vO$GOrf?~XVSVIvjh;NnGP`61$s
zhI|QrkCrE6%qW()=Q_j>aPl11d%3sM@j5p5>m0dzaHr1!rm3+z7mH`Dch%{8uB|Qy
zyT+N&N5y*Z$>RwNjSu2~=6ARl7>bkTA-Ao+xN<AXh*X{)!f>~v=8DtuLa2UrlkU3r
zbjlb~vvwcjc1Xd`b0D2AiS&2$RHcRc#<;_jw|JjKRbyf3fr&1F(=B?(I#x&5zo*R8
zxFjy|MzczS7Sj46M0TdM_*^NE1X$i_Z%8`XkVr-nA~HN527<QyNMV76H5hT8c>fF+
zU+XS3q4-uuaV6z3<7%@9c|`UZY4HEYkKzTdAlt6jNp7%k8C%TIV1#xoFy3#Hp#j0Y
zPx2c5ZH_6+xzh`bQ^7>4TZDWXA9odjZYblV`t#13-=bj$^;kgkEGJg=Uwy~Jy1d@G
z!|Bz2b6pdc{{jOBHFV_3Y8GRY*N(i5#vabZ0<o|_v0q+n=vsKr^XBE^E(`qu*NZJ$
zdgUUq#fP*tfiwO5_9$Q5t<<A6uyL?zpPfALpCxuu9D+Qc*OsoLpcyR9<JGwr)c3@f
zTqKQBrDVY3MqV)!y_=NUigEy*^4Q*e9{(w{AnBCPj*^zSejkmOkE_~%J&0d>k5>C~
z)gr4x-zh@wJ<@j)7#?sAF~U;)Y-fuR=8}-~R79#+Msv!A<10|=h}Zj|aiR8tel`=o
zmz}FCR%s6`M7!qwz7z;LBJ=H_6>Fw`ots#=k9Z9>+oTf;%jO*Nk(?VRWmCgTNxx)W
zgT{XQ(8mwco)Yoz6^F8ZGu?zh<?~&HESB5oSs>WqWIh~v6+`VS;AEqSrJt3b2z-JQ
zf*#|Cl!txBwTyu7Y%el*<@)UFG9LlmZ~U!;^2{8_W%w)<i2RDWT@}dWrfDrz%Ul=O
zB1XOWCqFETA#3M;YhPI2-B8w5O4Bmo?$wGo+H&N|E_xS1xcki$3oRe^>w&g>pm?8<
zyQ1ov0B^Pog8kwonAEg-88)xr%-0%j>w^q0)E?Y#X@SqI*%BxFRc>iyb;syDyutJV
z&q@sN<LH>lw^PrvC<5DKY**I)Nyww$?53929uRfxKo_X{<6SdBR$jVo=uB``oKeTY
zxXCM0vXa5@_U+n!b$#8jF;5Ipu%Ok<*Dt!P@kQipLMvD!R~BYk;(hv>m{og3TtF9{
zVrM_8%->z-tqqwUgJ^I#=`~Hkqh90HY6d`h3zJRihVUo<Ia3mp=(6-xp|#|*Jnucn
z;-3q!IkAqiR@!oby*ar5ZZ`VFK1DQEg}Lj~r83QkKQw9(?VZ?mroSt1_;Aqph082?
zkvAQ+@qyxqq?$cP6K0sxs#l&GM!SS#_s*zSCL@w)xPXwyKL?L9{CH(-^{boW3STgQ
z^>%ZF%c#HfR>sS9@t7z!%X_6vxIy=OHAJbPelvG|M(yU)F<dZA`JR=VOhs~OSq|vt
zxq?O}GiwaQ?=pFk3+OeDbC$;rzfI77nEz@x8B;*P<u1k}MVSUir5tI%fQM6GkdoBq
zpKZ#vk@uKiwv|>-#17UmxhbK9xs(7`no6q!oz(9^I_(MvpPRx9nUkK-%IgM~MprZ_
zFxCqStKBa;EmcoczZ$A+J{<jz*R)T=kKn-SoRsIRsV2TmEf1+D$&w>{8nQadaL#r&
zC%%EvF$GmoOiHiJo``W=M$6A@dyay%JC4)zf>O=r!Ts=Iu)Q3m+RruuuK$1#3|!I8
zN|UX)Im8S2E2XzR(|1b{0g}H^zA8=0k(Y<?8#SE>^C8ve!E}9SOz8KkF5l-lBE~TR
z8eR;oP{mE-9@qx`{sYlqh#E2$=8|RQ+Md0K(+*z=hMD~aPUN-z$CAC}3bWAcyMH|`
z0g48{lN6xMKgdD#EFIc_u~mSkLXTmmgQWu7;*K5JvMuHgYfp&D1`fuKl^M~T!vHa_
zi$7gU0STEHLP2FK*~c|iA1tuyIe|4{9{l67j6YzBFd;Ep%vDFp>(y_#m7%1&h@#3U
z&E3xXHyjgl|M2^@wAlkZ&g({FqXo^yfH4(DnOfYHZ%1qGBHuKm83AH18s+@yx#-K}
zC0_TZ!2Z=Yn4rK*Am}!;U^jxAvNBNxCgE-dSCqfYL(<Ir*MuVGlc;+K<&4|{*WuYE
za&x-YTG-Mkv&ZIzIBSOa0XA-`?qndcv%KhLJBVP=U+_6XU7#@DnnjYSG6?XEhuZNt
z%9eusGd0c|;$~$T;Gk>lVEQ)jx60b3HSzwQ=wWk`;7CsRnTJbj1@6Aru-6P<3ZfN>
za#v*$RW=O3UQ}GUG2n9VF4X)RyzASVI}xbpe0|9mGQre~gAdko(Guw7ulBngGu%5f
zp`1E-Bix*fQ)c1WY86xi$49;nE+O8F3$4vFrth1R5va5PJ2T1M#QezN<1YHV#j`vG
z*%`NSJi9Ho9m9sP04dAwGA8be7}$%cEDw|;RrwkkHlN~BE!&Z+q9&7(uMgzH{FV*k
zr(8`=Y3m8a0sW*YCHIfUv=^PHPLC9+TLv0mUHK<N#aOfS9j}*D!-ksjJ_|qJ3;n)7
zFqG3?>SKNn&}NgG*LM$wkmIyxklDWaFF8yi7260wJ0xiuRc(@s^1F?&hnTBF(8Hu3
zI*s5c1mUqG3&ogH<`v(an)%N#VIZu^eJD4x^cO{XgWU{Xk^5@3MERB+sn=R+W_d<L
znk@C4{0pod^>Htz6Q<5+8r`w{Ztm(R_CfoG{0*rGFqBn4g8e(1fv)W(C9Z?=S<Y`1
z)fsA2EMlSikW>o$ZJ71T*?!iWY^|X#PNa-fN8Eu1f{<*>*-ehm)X?Ls_0JKp5hbji
zP!!v8x>`5>-OS+8KnjP{P!oY=?(|?u_^x&O>!E}7x1!%L(k&s4SBWHp{x(I#8PqsU
z|Kvw0fK^OvdGm0>9#oExu+mU;%L?Ut8o_EKTAS>l=eZNc7R(=45q#f<-2}KV5MHP{
zi1b$G15z*a5H}6Q_??KNe;|yp?i(RqJ|c?dXXbY5aG3Nyd(e<Ue6?Fp(c7n|a5~?N
z9Y2P-pa?(5%<XW>p0xUIrMG#d(h5?4#>I}j0hXf_s%buB!NxtC57i#+6N0Uog>xmk
zmA@vBzZC5<T<o(4B9pSSRadf6b+RV*Dn??Kn?qUO$cKI~Pv&zA%nN=qtU6+TzuFlD
z+iZEEhByLAjrENUPBR%TG7dGR>*E#*?l0p~=*Os=I=40G92ZSR#b!k`z-YKB$-#C2
z_fNs;UBP#-Ju%vP0*bm?Qo#d_c2{<k*f}5KM&!cm+OQ)v{lu`uRONIj1+71u@S2C<
zeKX0F3%~+CV|#WSkfaF*N!~r4))P68yYmtbe?zvS;Mt@+8_r0l(!P48nh#o<S}2Se
z)!WPzzdx)kQ0BnmFIv1LNYtY%s2j!W0bb#IR{gQxVkusED7?5RbQ89NB)`dgpf@A@
zQCv}gn{sBDk*R+BR3iA`D#b5&g;J`<?INDDog1}Y$PhOw9fytyQq(SdXpcuucabgW
zw67R`_gl@f8Oa<ZQjJpGD9kk9t&PB9me@G8)skinEGb^146okH+&9@qSb1bZpqInB
z!M}&Uyr{p}p&|9&2!ukxn+yrWj&fg2qA}%r(A$oh_E1QoulyHH5!odLNeC?KKzG1T
z?h@^~`}*{<r^L$U?!4xJICS;Y<P_Ipx?X;=JgU$*Wr|ZAQS6T)AQzk$$5(GPxzu8o
zO;3){uvrGjwVWPNjY9`lT$2X8T!h{^940L`37;cg6r|qgM@0WEs17XU6o+eAAT*<w
zGSI{y+{712)T9N=VF#Kbq8e|?Ao#RR$+7S5CWMhvZbGBFPldK?{pgV``CuydW_>u<
zxHxX1Xvi#jzS1<JlU_?S|22T%(6!G?^o}<5?atmtAVl(>RaUZ1Da8SJIO`%Vj|XSc
zHRN`~Bxg5m(Ft=ii%7BPiq87qXX;LFrnO!J>uhKtAFPf)w*U2kU)lIq6F%a6Y;P~H
z+qBl%2L_hvU@>{f?K&oOcC-T{0;}HDA<APvBky*4o+T=F5hVI9&`CWe5y)0E(dw|2
z7EUTIUhymv8`l5SQ~m%v(?g#IKka!$lCPT9kE&~ltEQyK1e@H*N9F>CS%K#l=5OF7
zD?Xt@d{RtjnX45-CYfTlnn(NfqoC3JLYoh94@+)={ttKqXMx=la_sf3jl-JZ#pSQc
zikd;8iEcC_dq}5m*UCStckFW(lqP2e?|6eChjUv3&I47FHBi8_PMyV{v8;YyJ=uOr
z`s^Yb>-zDXd8r<fmq-71XW36D%;DgE9hQe^mg!|dz7u!!Afbn&L>neME%td()lCzt
zn^dtUgBng8_1g0F#2ML5j#2}p1>EW44OgKu<8AtuTS$r5Jx7ewTm8M0O4#UNHVN)X
zc0XL2?gB5KgTSQ<g^Q+otN;m;0YXbRrx2%r)bt}ViyI2|<&W-SJFtvCx(Ghb9i99R
z4Y?W1Q;3K8z`CdRcZ}riLee4Qw~NXtrniMikyXK2_KJ#cM+;4XeMT+*INv8K4eSGW
zWXa4Wdxr<5iOasYz`)_5w1cM7+UzB%_Vx21zKcP#LEysK+Ao!F%kft^nAybf=Em4V
zjQe!>Ho~7mm6f3>ISza3B<XXO(++1!I%A!ll9ZNe!64mc=Oz@E`kge<Hh8wyQla%{
z`BLJa;rdfSnL!qhnDi)I_m`Hn3$fc>-jFQqUmt*5Locl=>_He;+%RW%|CRAFV#2OI
zsXS&n3qbF^wqWC&-!JW8FA-NgnQqEFqo1v7pUyXHm+T^ys!@iJtCXWpC3JbnZ7tJd
zy`u>cMzQHR-kz={g|7Oa>tI>j+)QlCYx$fubX8S^h#yfm>5{-X_JP}4ah;0DknB0w
z3sOI0@@YSe3@<+AwW^`2uy{jOchPrAZS0|89<cd0KhR(EXi9`eikIj6&>oT)gYk=2
zEVXfco8qv7?<0pos)Nw;y*0HfOT{#`ybqxD=MBp5>l!=}^RQ?C_)3_pM^d7i(!%_t
zy2kUz7dww@$DhRcjdn0AQ#7FjI3sVt_iOP!;uxiF+hI=bI*%qlv6CYKq1C_XE;~JB
z)PP3CZsyp^a(@IudqRY~E_sp0G7znCo>$gt<6&M?+{STt@RCg6fqufFgsq?Z%y1EO
z!W2pIx8|n(p`jw!Hd+$m1a;;an|R0Ka{TenjylGrSV$vXnoF(;uF}T^&tq*P`@Z2h
zJhT(f?1X#U+3Fq=(-kM84@m6HdQGqo`v71>h=?y>GG=`E>rhjvv;OL$;J2&cs_(ac
z=YwzMra9RX$P^BE9O$;x;=>i%N=f)~x|xcR>Ao~;Gn{PfoU=>c6%!IY&UPEO<dz(b
z9th)bZjoVa8J1$d7Z3*dBqe6gJ<QictESuu<z=kVUTavn-f&=>a75=|^V^Z^MFX>o
zryh*Y){miTHs+uWwU4g<F)H%)A!doPA2K@Aqpnu2?JYfOh#-(brpEVVMFi(@h7%Xf
zwd}~m&70fcxVc3ac3&t};Kn(>rj>?;<MAsnj(f1ctqW_7bN%BOW08RYz_G6ReATiU
z(p0yad8hneds(h4(V(5$^NLe4Q1s1yQbDmyH@E-;r{jfpU!xHY6nr?Z>v<beThkI=
z>R>J-urXxk-4>INitVw>gw@Qj2A<5Z5y|Qk^|4hyFTjZhIfM=s5rgTSw6t{ais3KU
z?i=|j^%hbNN{S`A{^p}N<N{7SnljwO)O;qQ_(71Zm*F`fLAl@JYyVaO?Z2cO2p=`{
z2cC*{-cuNarHq~@s*0>FF6<dx$2>=eEbf-IMPVqPAv7lyNnfT3wK`n<9hMWLiMr{5
z*)8WjZwB*Y{Zc~WG81pgsN2A-;>+S@#nnuB%oztOixRJ=M~z=W`M$wf?+WR;sR@+a
z&WCMgR=qm*3}l9ixm3`A((|a7$UCvt6EK3(zcG03CGvPCeoLp`2~GWGJv-g5zq{{C
ze7W&1ew50;PqjST4p7ff#m$wa=$JweUPkYj9-D?r8qQ<f>*+5QVOFC~<9H$<wK6;S
zm@$)!yZczU7ZlYsRb@nO1)eF7Y|rTMdVj9YRV(6SXy%W>M6W4e4QZXlWAnEY^%bNU
zgY>4A@P@6rq3w=ZUm)lur7Hzt?iu}TB4+2oBJ%C+j$))zrS90C&wWeec8NHrg5&lg
z5`+3hhuIM?S0Q+&yYq|GAMZH$m)d-ryL*QEkO;A_5-Sr^8-L{YOA!)#(ar=t`)<Yr
z!!d4aIKQ`pXW}7{(p=ic#5MS}uG=$2lJHpB1YPo_Y+06}q<3vTUob+%G<fkcI_%!U
zw9}+rLn#46T1lzwU8kXXBdF_1B6^%GO5lHXM{i2M#t3!$qpC1D^%jiTlq~EN@jQ}0
zxhKz^XcPCN|7l@&vEcXp$xTUWD{CMBhZ0{WQQJK7;8qyGqagRdxx3{qu{acbAXAzj
z4R4%~nT>aZC(_TdT}c%aX`DF-Ik11EEX}BL#M}c<ok;2*lWK9>roDQCo)aK#iN?eF
zwKLP;fEZqm(u~FmhmRl~r9Q@Ctxn`3N%8*Jj|6oFPcRe>A-OdQTT@fB85uSbip-{K
z>b%mS-x@F#cHjTE-!l_$?3gM>e{zn&wmPS)1|P0b_DEjKrG#yw>#fN?+9(F+#{G@4
zAo<h@f!t#@S#^Z*7j_eL_HG3aMp+uqPtb-v*7c>`nwh2{iK$cBt%O@de3YfAz@~Pb
zdi`VAi>&9|z&3e9%}!wEsfmlWU@-E<a<huKpDVENHq-PB{5GCBa;3qmtv|ID#3H>|
zceQ(IhSDC>zufAXKksNx8w**Mj2bY(fG9Pp|73{6em5+E$&!51d`m8{LGDKLk%fo4
zJSVT#8CM@d=~YtSc&TzbeP1wv^HloF$n`*~*9|S?BeJlfge|vTn5(7&jGks@2Gmy(
zAz|R(!EIcC&$gEL`k(k6eZ5FGR3_gMY6bzBK-KQD?__3Z)dHtbYFlJN<E+J2Fo|q=
zhxa!}Pu{mpTyW#$Ywh-Zj`?l;LIOzyBc8`^nj*uLOy!NeeS@FQ4_p8G6pgWWkrHwK
zCMSVdvv`sLoE@Y3#brq?+(LXX|1OQW3p4;18r<6Y<N2e*0}_vi*;dJeMsXW3;T(Yu
za}4W053XzC<N9sxCrQjT9&;u3S#e|ae`rh{C-5HU#?0M@TW6U+CuhLBiO9bL&>Iy!
z!<R{h-j{sexG>4|>!B)l-VnV+y42gM5IPyvnM}}uhGlE#$o>x|Wf}+KiC%j|J77@=
z9s2rOy1w;PXD011{U<E^UK*ZEN`=~j4ny18L(}IkCi!VOt8_3DOmb^sT2cuj2FeXG
z^n@iY3yE`jM@3Mxlo_k^(8fF=DHY}GgRSJ~(TF6wnlFE9$u+K@I{hxM38{jMhNk#<
z*DKq|7i4oc!7MYxB3=RS1V2@y6`=i>eR^UM8#&kIz68|NgYmvoyU_#Vd}G%DRi5?f
zHpB_Uy(Iea0xHQW_LZ@-y5yS|i9Uo@DUCUfz^YEzfzBmO?uN7`%m=pSnnJ+-QMs3W
zUOVoE?Xo3un2f~1cQiT2uK^Kpm}*W#UZ`St?VI!{L(*wO7eqqnmNRY#pzYV@Idbjj
zQ`a$v*(z#`AVRPoVKB4t=kvUtem+byLt88ylt1gwES0c&-75GoE|6y~efcpuOVop(
zl=S%6(z2xFXs+7M{t6i}So*7(CbwI_7eO~gq*@C2i9QmU<D&Llej%FIUH@rG(ve!D
zheWk;@W_KoUtK0{o`}kUqcu-aR^v~><krPZU&Gs7U;hRc$F%-a^+cl)Cr#>$Yff2_
zhoj_VHcF<$RXAsSb}@AnIc&A?PLunr#j)eh^HXiaafCar)OeMmEJMHLr6YKvpr5t4
zN$AG|*I$yr^yh&IVZ)0nzaww5h$WeDv|qI|D!4m4+n}G**x1Mj*v9?)eTK8<*G=gi
zIYtv!`fE|d&lc}a>e&6U!>nt+<jj0+XPaqd?7h4<v&<u#UVGxt-r+^?OHPoRX53`x
zx@L^8WVdKn**Z7IjpLg({;5}2g1d*?KdMKe`^Q>Y$nt%<td|?pua6&w)oh9~dw-Ho
zxxj?lD-?Pb<T8D@08A?8ygnF?%5IXV7XX^8FMQsq7mvFUX?xZ_7<ET5kK2COkxkuo
zD%%3h&ddYD3ufiMsp0=VSC?wNlcA_f(ool2nqTB=a6?!N`ZoB%#(RouLXE(h@TX^1
zA;$-X%V!2PJ_i9jv)4DJ#aAtF-SL{bWA1%VvXaN4vb&(gdth5;vwyW#m0&NAwr{x!
zpUi&tuBWmamfzz~o4?=o%m{)1z=Is3CUz~+AM`%$Yn}!%d5A9B0Ryl&ae*Iv4+axp
zc1wK``R)wffF<Pg_d{Bj1kU@qQ$oftK;0&vJ=HZw0gu3LW27)fm&{sH8~v1S-D6_s
zT+umli*76cknGEEp)VB^4Gia+0L(m$3fk+NdEd-r63+D9ajcBZDCHy=qk|zDH&j(%
zEStLkST$kNZbjbDdNb`JG{2&a_HAY`_|7n!sw^O#&i26VyQQf8&=mn(`7q-jE3!P_
ztqBQIR{38l+a1pGd}8w&BXZKIy%a<{n17aYpwO;>GQr<A$7=)OKmdY@o|f#va81s9
zkcAqHe<}tK!TuTf?@i0~#y=A%`Wx}(t8J0Hc3)}3YA9s9NXd!(92x##=v)akS%ijw
z5ovT+fSKNtmNzVGZh@JR^Zc3g^A!j*2A?IoaWAf|DCucv{*4EdVIQ{@4fwqSBb7he
zUxonEtEymlf2}BtBPQP>02ONQ>q$Y)=x^QSF011AHSZsKH?}nB`M2ZYY=yA2#bY_Z
zEo<sG8Jjoc*PfmfFJJdd1tVS9IU8nyv3k~PZ4d~=laBUQSKC8$vs<!I<AEj?jUuvJ
z6EhFfVbod2O4boh-a63jlvDZb8(J~&0f<wW_>V6s_^AasZ*$=)Z6LY~AVihHc|$_U
zZp$m$w9*(X?(%>Z?=`01J4sz>^+Y_vIg8@Mrtq>PH0mmq^&jK{SKKx4*D0uIkiHut
z{hpt*!CWXPsa;v1<7de(vD)muri-i<kHly_vIMP^%!~s-VscDuVicG%6m<bSyPBA&
zI?0_5w><k~spTr1A&g<Jr;o(TH>6rSb>Zn<Z$!>`z!Q|`>F;9y`J=|UazupvQd6Lh
zs8g42iYcEmd*<d2S_%0UwUBC@e?fIUf#rvi5?Es!Xi%-h<28Q;8vE7Lk`wCFj}q)K
z<?Ha}=vbCzSyRDX%JuER3n5@YK07rtub`%-tfQf*p`oIqqpYl@wl*<8vpls3U&eU`
zA0SC5xZaB-%&)myKra=Jeo?{ocP}$bTb8(%2)ZCD;!jRX=E`Jsee~54ysc?4k;%t5
zoC%e+?^=m=d&^NLhoUtOkV9wnmJs3y>;Xob6$gF}+S&$35BLkhb7p{iH`g$m8f+!9
z4eu5gWM8rS?JnmjPB0SBx^C;&+ZyeYbJRAr%ji>6JxzGeSMjr#SknYuS6@nY7-n_j
zuA@p41&>^CJkI5p0<ujm8RYcH+^WFsikJ1t>t3p#ImCnGe;-~G9RiMP3ZimregDoL
zeN#eC+rYGJ!g^)kLAs^K<NpQn3=Q){UL+9e5!%7#He+|2!ki_ENA7p+jn!pExfxF)
z&$;h1l;@@SPeV9!2c)9Do|6|gp|}OzOnOmO--1e!&XRacvlZkq@WHF6s~xspyv$5Z
zj=y@wWrxm2uC;R=>UV&Bf?&}LNoZI&9uK|!qyQ_+g4hYiKx7gI9e)JEs2@YV_V*t?
zpdANzF$w~^&vv1e*px^2qtBn%Z>cH8wU*>~pSBc{)v-F@106EqjBoeA=*QU-k2*2H
z8yz^=o9e2{3UksQC&b*k;&%k3Gh%F{^GuBcH4)<E6;d*?bv<)AF1@%M8*`oBzrSG2
z<J3=Jti#X!8{WbB4u4B}$Xy=Lp7|{CZsgSq-UnfyD%%>8;~g7>pl4<iQ{Cd=e)f7o
zW`0RUJxER7&yILvP6}-CZ4EW$C51VukM7^Teldvjo~dyVi(W>{!qM|w#Dgp-PzP51
z_-EWJkq|aOsfXO-tOMHdo&(4__^(gnL%khMAohZ+XAke*x_=zvfAWwW)|pOfyd$Ge
zh^+>{oGyH`j(n8<;&nx1=fK3QiAUOq@mo!H+M~FrtD*iU-458AYAc9v{(K%|PCzVr
zQDsA$gWl&OlhSgE-qyADj(q%#H_f5Uqgr(+_c-a0^-26|NB<YrZj+q~X%9Z*br>m6
zkYw!B?bRIOs(&U(eez-2i{k3mcVn|89(m#f_nO|ey7E^qGM*$vU%L=+0?ytUZI&0?
zz`S<ZT!;D{&_z!}$H*t6Y3gvqH!S+Wv+SbshR#9!3F`}fPsh23A2v$)%)=e>4zGW(
z$E^5e#^d{7$@_U7ak8^Cgt3!Qd14yBK*en+>Bqz`r+;nf=yCRHOj3IO+s2;J*%FV;
zNevHnHCDdPO-s6WD?G@@{Q#g{Ls43oXFV;&lDQ7`JK$b2)EuIUdfOe{{VzquKg|KF
zzHj1BdEFO`{HZ^F2=tVj@xFWUthd{NU6#gL$}*xGSm{WoJTU{9i05J&PBCR88|M?j
z*Y7^gDu1r+9GM;Q=$zE>V1HXJi2Y&gjSGIqUG`X+Xl=suFGxZP;9H<`?JHKUX5y7n
zGv2Y^BOp8~AvN!9V=q#$_%@*@_@C}DAs)D}`tw*%Q+Yu~Qp~kbU(Z8(theYWV=i+f
zqJQyYtYYNdsIk@F#XB_Oeo8@Q+t93upMW))cYn<lh0j6kA*bE<+nVVpN%E1(+{_Pv
z&rngb3Ce10vDxq8AAb8`=Ih$7G4$OT5dZ(a!#B%-ARc+&>8vfzPL916?0wY9)?80j
zMwpA03DM3XotN70LA2s{gKz$6ZdpTb-@xE7`f~q?DAZ!-9JazoM@NQ-hXw}+`ulsk
zI)7T>V}5yI=Hs}VVP{VscHC)(#7;6pL*E0R8>C|1AfcpVYP;XVFD&vwdO-zJZbGK}
z@jmHk;W8rlkr6cZzTTd$j<%MjhPu)hDe;kE0bVY9ZOru5<i+{enCWTY6FmvWP5|%7
zaBw*hc=^u5jKa#=hGx_*a&#OW{+z;db$_J4VxX_LyQ`zU6@A67DldIqn3tQG{^;JV
zOF^fOIXhTw)m4=h<7Fe6p`q`AO$slfq~{RasG@6X<K*spJ|gaEZfONFS}{0=y~+ae
zF-a2`JUr3f^>lT%x3xCI;A<-1mK7Hjyv%u)aQn)+Q{ZN^eT%M|yo3N3GyOU$ihq?P
zx#$TX?I^Np`sVh|9)1^Z#y!c-N6SuK(+$^{Gx9jvb=}t7P*+`9{-zWJotKlD_AEIu
z{%-X3i-A7J4(_orGt^X;731e%BAKC~?}4e5|GsM7dL9uO6>XCpjz@fguYzTtTZk!-
z<Hbh&odPU*q}JThSclXqON(C>f`8!CpFK`Wh>N@&?CWu8&kj?45WBb__j-CLiA$P`
zp1=|cI#ym$Srr{qn|()of-XlTB&TNOy?$HMgjCt#n86ewanRS((b7;``KII*Sm&AP
zDNi0fh>Jm5G|u^Y9X;T%!`w(qRUUIyB$c6|?}6)eub^UJ;}zMcxY>C7ZhsfA0I=*I
zrDhi*ZQ{eDSj+Wi^cj9+sK2M9wGoNEFz-cnW=85WqCX|T+Z`-<bFk#)rQo%WfrJ-5
z0kljfEGfU)aGU)B_fzLC-GVX!`6X42xPH}ieHN)TmlnOu&PaKZlyDDn9u^vS#^<;@
zl#AMCqPJN|MgmudB$c6|AAbNFbY8lent_c+P+Ue;&upjDQSZQTI3bW%TGI;6(2?!#
zGnTCm^mVm0)|4Z$r#^f9Fya2)JGZWdBf+~J+-qyOMQ^j>MhRgMJDf8_rtL_m68!@Z
z)tZgU+FPs~oR6If3cq>xQTof0YP@k4={W|xzeDE$^RiMOC*F^Vx_@!yVo0FxNzbDP
z_wBabZf2~jp)3a$JP!xB(n29%QW+Zh5fHhve=lA^Nlnkf&LbeDsAaqz#t!!T#I(G!
z`mUaL{pdl5E&1MVSaW>!62$&6{@$I)h^v<_ob&VXJaW(xisR{Ps=$qo8!AA<;~g2S
zp8H|3p8=bAW@h6PlYdjy-D<V#z_HUIS7RQg=e@x^l|I6us?i~+aex#Z)1N%J8+GII
zg&;q$*bg~5?6fpB(%G!60M<Jn_|_S4&39568u}p+b%)=VtXK^dyVh?Il~vW-YQ6iQ
z*V#+A;RUy~v%9x{5YAP-2fTwTU1xg(T7G&Ed;8|KOXtr8_<wqP9Nxcs=e8~SnyQ;*
z(Aeo|Y1UFA6PB2e#9W8@6ToLIDB4BM#KA8vr>bMTZMW;mpeuKho);lg<FH<bYg@F|
zzA4I0gEg4&kbqOij~;Ziw>CG{+pH`nDa^ltlNCu~R$&qnBr`PhA3=ySQZaJyODL%8
zn_4@01YC+q&VMOwY;Es?%#VK<9|gR3ceFKEmlbBGCdJ(b`#V_jUhapT_w3weqN~11
zT0~$27n;C9`3z(*bXNM$a|ArMSw+pr&MP9lN!?`Uf#ZQ!W1qZ$?eZP(Mm|h^em{oX
z>l!QIt1K);_?>ij*=KKMW~imAAdTGRNJR8tcEH7GSASA7aSBL)WpC+tEZ|yPYGF-d
zOFP<`0yCLz_*U_@C?_Q`_Gb7wbfepCYqeDmju&ySpP#W4x@Rla(4ulS+NQP#JpIEW
z6CS5#y@X;BAb$fXE4;~PJV}U&2oFAU!tKyLc(mzhD8r*|p4?-<2d-<f0=DD`Ny@`%
z!earK?teVWeu?*!)>M_36~8LT%S}s;kG>uj<m=_;dcbj)&2}>r9r(5LvNO)l*a?|O
zTqZ(F8Dw@&M?q#6aXU8t(bM#7ki9%l0|{$l{GEu)p#i5n4<FcLXAN#NTIk`%1$7Ha
zH+lOBaM|)@6f}(N8-yfeHH@tt-A;$zxEufAVSn<Il(gqr&(l+$CMP8(#6{n@6yom#
z&h@>!c3PQlG1NuxbDS)6RP!=+BKW1NFqZ~o?%l!J-N!F5<l>bZk$2)^@7%r>aqUX@
zMdTjq3{G~7t%kZ9s+(jbMFqjx4pq!ZMD1dh!1KCQRJ6?OJc45K8b+252R+V&UJSo{
z^?!QAt*GdzTM;*|UJk!_KIja%%^lclXT8nT#6VY5U3rs?xDYQV6uF+CvE#uLdPYK0
z3hH`Wx7#}%banGM;d3V7Tws8o?<wyS$32d~%+1Q&#6U+wRZ(uEgopqy7ds0B?OMu}
zFm@8n+J6GvlV%0QS~{p$hh#R^4$f{Ko_}5^eNOrMfyht$oCKjCJM6sAVJFz&^>sAi
z45f?|OxoF5=xNr>)7bIg@$Li`PCgMS1ywDBt>%{2Hg*nsot*bOIquoL%g)wjyTw)`
zJ*~|uo8&f1iU|wwg1i-)R-u9sq?@(>1b=d&f(_@Y+D2y9HaqR?9d_?=+~>G=w||4Z
zo$byYmT-I50w1~@;9N{tkRPsjbXH`Z#*X=mqG{b4>UE6kIe7)eq~(-U)HZAC=ouK9
z7#kVt>*?xffu*jbAS*2)DkQ+i!^OeI!o&cjfsiAPL}X584Wt43H{5)UnPUS8U0y{)
zQ%hS%S5M!-&_G{LPe)rzQv>_VVSntrAb56GsE$Lu1}Xr~*Vr+)Mc9(Fe8npGq`}C-
z#>vAcAS5gzCN3!@EiEM}AucW^DguJuzzrB*&&<R?PfJ5hwFcAw`yUNyNdjo>NM{D+
zT3QAcHm(f<!Xlz#V&W1Kk|1&kaWN1&8apRDD~O#D1iubMPl=uG^EY+~yjCR#)$*dG
zqNZ6#OGih~z{tqN#K_3NKu=Fk2g6-UwT6;n_3BkCS1ezSii4mYZ6yC-@e|;k8GwDT
z=}}Ry1A)`iGcbVQLF7p2-^31rN2ABgs^)X-lkd_)lG6$q@qdl~4{ok<Ao=>@_y7O^
M07*qoM6N<$f|@hZbN~PV

diff --git a/docs/dev/articles/examples/vision/autoencoder/unnamed-chunk-7-1.png b/docs/dev/articles/examples/vision/autoencoder/unnamed-chunk-7-1.png
index ba436d5fe1b46f6d47be16a09e8bf10cc0e65c1a..e86c38b7a166d775ba4176bc7676b7f3199bc144 100644
GIT binary patch
delta 102577
zcmV*pKt{j2r3TfZ2Cz>9e|S_`wsm76?(V51RjGI?S`w?cySuwP5kiO(NU#7QK=1$o
zg1fuByITVdP5aDy^Phe0ty`h#nMXH0-Tc$Dzwf=FlT@X0_c>eFUi*tc5Fd9Jt{uzF
z$Us+9RZ&(_bo3XBOsCN7`svGeu9(q~8}4kZJmL$1jFFI4*3@H|f3eIMG)-kWi7{XN
zDiM`aqnTLq+ygv#HYPN6IWgh4Mu~_^$tkL6=xC@a%1Mcj9rb^{kBy^8ke|1^t0UWv
zWx+JiQCE}^ANF}fWmR>VR%};aH=eD9k)FDul!UnG*fFC=j`-*A`&W_wD*i7j|5f~7
zRQ{{@zo`6I@qbYXe+KdMbm7`sni%S8swv4!!xLAavjb8p+WXed>!?f%<T4b6k7~4-
zw1S$pzLA-czP1{E&?82U8Z9zbR7_k<bganeQRuP4SH|G4pf8OT88c?|s8Q(aBhb5y
zl~LBFo7+0NJKC8uv{ht8g`Z`Vh?t~|yppPhs*=2nq?pKXf5qc>M85>R8U79EeN?m=
z=5|h=Zmv#Tww<+=InzK_Ls?F8?8uSCZz2?LWa;4I!{^zWGj!GDC6I`Nh=Kf3&-k;H
zFpvN*SFVk@5e@WCJaIXFTd&yshN+9EjW3Avwlh>1rg%{)Ic0TieTKe{y0V;<*yt}t
zjvgx}DJ`p@f21HMEh#p3%t-td329k*MFn{|8EJIS#6bO^`!7VK6x8*MENppp7DhA;
zMX9mE&oXk%SaB&?q<JNISt)T*^gfh~7&QjHBl;!Mk`m}QjDq)3)Swwz*g5hz4tBQI
zRu<+=bZ}J^q{Z<u8%Z286&*tHTpKgGj;frvgrt<Te~hf0yaLD{_?SOGi3ADqcI8^b
zlcAottge-7SbBMLPisX+xEqTmCsh2Xu@ce>Nb`Ey>M9B{5@ScBuZT;_E2(MfYN;v9
zOT$-2qp!%JtE;1>p{9b)gACR^QpgD8Ddd#3Xh!DN=0^HjDsmDc!p|~tw1}7l(!3)2
z6cS=0f1`&fe&iTYbS4y(Rh1QFr6u4KqW4kJ(r236+E}x!EG;d}&6tLIT53wjQw_HY
zSn=Mj9BWfUZ53Hj2`L#lMP)S&Es#I(F@Js%4HDwR=deulH5KJ$$R{qNZN~GD%dKf2
zmlq%4Xs#nWOz~il6_nL9)sPafc|gWrUR6_%e{M{tX{jn8<3C~y`ihFCt|60QK-1Aw
zS3z$qEh#=0-WfcFs+JzzgaJP+EjG+06W1E420jJ4*25H!&z7>9mL7SwMv?DBW0(?R
zW`-0`(^gkTXHgtf51Pk{x3G6cif5S^XemqM^N3EU0R!X@e9WJp#DRqQ@i|sZJym2#
zf5izyDy3m$=N_6=P+OQB=3&p&loq}M8+N!iigGeYndtI~NXn~f8yH*KS(zB>s3}N|
z9f7{0qOH%gu(2{XVHoIYsSAG?84MW(6?JWzwgzc1C>a5vf|$6Z6nqL1Fb|ZB7L$@!
z(a<$C26HYiEjF5b9}OLS14BB)h-qwMf6AoOP*|uW2i`ywzZ?|LK#KQrVY7_&kzW}j
zE~AL!6BfuH_mn?FNd$@TcjH(xw3W~U7Zt%lucWGhrK3+|YGGQWualLbnw0Prf_o`T
z-OI7kN}2{HEDm8%m1RUnh@!9P8(TRzbM37yOi(l+jMtFSmXJoVyPArk92o6ke-RuJ
zz~LZKV5|iKsf30e!<1#p(9=|wMX7|~ebhCzb!i54BV(j^eRRao+X_Dzp?D`R7Y8e2
zn!1AIXmOO0paW^k1Nq~g@@FV1AW;GC91FUpywE#}E9sipx%x+@MEUdWP4tu{gs(vA
zf$&-!yf$G`MP=3Wj9FYaKX;xrf6CD0#Yc+Csp&x;<>TSPvu8oPtqE5V@`az-^Wlyh
z-Wtd5B1CvL>^mZo3TnCxGi#2G8ADH9K}tmUK1laky83h@6DHk2Pa8QPta(vzD}>@v
z<nQcY$<$LrA52_MO&1w|Cr^++>KT8Qk`59R<jyfSP?Lejjl<zD#N;&%e=QvNei8nB
zwk2ItL44RtMiBZz!SiVNarDCqnsiH!TR>!>JJ$*s4vCTC=qnZuu72S`zV6O!8w;ke
zLLfLeV!~7Q-dm3x37H@HF(?@;t*m8eYRz>+;klufvdma2zY~*`RaDi~(Px0-X}Tzy
z6uK2E<#V#PFw#|(6B{8ef3IO+X6xjIv)Di4`F^&N2@)6V!8W6*ND7}d6%V_hyP~Tq
zEBY_q0?dQ4ovR-*4*`4!Q<{p@NJ(W~V_Rq6@WdFDD!OoOp}3-<EDyQgZxdXKVR86=
zZXA}eo|@dS&oc(1JQP}*Fd2sWdOC#W2`56-4D1{|o$M?aI#4qZe^bz6u(%#Uu^_+k
zv;1cW4(>v{9IcJC6^C8xu>vQ}M&YoglJu}_o4B!J;u6T#i^8=XBcrBoX7B13kre6A
zcQDgOmP1NOn_<Oq^9v6N0Bi4H1Jxl7CE4L+wGYcRV7*1fQL-*8D<>x>FE6hE(I#QE
z;l`WUaNT&eW^`@1fAN$+eyB+^GPAZcGiDH89@0VdvJsTW;d5=x>AJ|)D`*&)+B$g#
z;bi9D|8zfN!8Xs!$<9n)Lw;B(3j%c=LlY}|mMNVG)TwXbG7Hj}iZV*jWuZEwplN81
z+(LLlxSy-NnZCO0SSb|kA;Zk~_4ah<yEyUK_7L!^%1es8fB!4A;a5=K8bxIPn!37r
zG@8Bv3fIZRd^Az^VVK(>1>(}{FvU|9AqN`^6L<yEJ3cRCMoB1W=$qQ`+@U~a3k53j
zG<ya_fcycE_wyAG5Jx+6`uo*Wl=mp9qwEPiN?i>UMT@*wz>$H1I+#^;Rb>S^>=oF2
z->|q)9~XNwe*<)aq!p2x%&gfiNSr~OP*|=1zU2@@?$+Gi#V5?y)xpA06CFFjFSWE{
zS=-p!**mb&vC~ykKxYCs2;f>H_5G~uBe3TlE*u-8D1=WVRF5T<wCIq=hxmEA@=&r#
z*QJ|S+VNaH{X;?ifXDm!3LC`Q44$vlu%}7XWp(t8e~@oRd9(PK5mc~+)uV&Xgqbm!
zKWXS2BlqhY5)<O>Y;R_$sfbPnuB}+w+OxTi&Mtg+p_iouXN<m)sYPi~WX34)FD|Zp
zcMnf*A76hrt~KG&M++P|{e!ZPpj#6N**Y%MNJ)}jNE#hU>~Vtqy!cLRTT2rQYkRJ<
zo0ne@f5;#5fIn9;1EJb2lnAV@4oZ*YWl*Fvg3`Q*lp<kT*|t_hnH1$&He9|>aAXkj
z8Rm2?B`B_<9Ls=VY|gUdIPpDw13Vmm-iwpbHg^n6D5#rKUzil+WT7jIRH&|RX6Nh`
z5*8jA9TS(3m>l8fVrQ(YBq1o*Xw!5)DEpvze~oU%aSx0R_HwpEA-^K@VZ_8h^RoCz
zAdemq65#8}cjlsV=;Xrp@bd8o`D32&XDT$14~uAmeu;__)W*fejvh&Ao(lNgow;^a
zrs&W>(T>OW4h#?Q0L7z=Ehn$2tO}(SI`ZL;ZazW3kQT}t+WRIHH}$NX)m)O~=Rj8=
zf3gD?lpUmHWaZ=)6qS@^#s;`p)77Pf<r=8|O1#&&6Iaq9UKJJS&Vk<{3WD2jkh>vg
z5*i;985R=g=k39Fb3?Yo)63fj<PUnppQosRK#d+JzLend2n5)KyBve_dqVT*aT*TG
z(ao9q$n`on^1TB>{cx{UR}DgP1qDS#e<dXd#U0)K!{bA}ootvoO2ZVdWa1K8&@^Z5
z;i0+Bg;9J{6=eJkt(^U$(o4!JtH;%kpU~V|l@sg7HPe=X{tb@z<*7KI5(yQWIoC5Z
zAvY=9$BD(zlp7<0(tD(NadAa0s4w_OCMU+nMn{B&1O@o}`TF{JdwKr1fFlV)e^qLQ
zLiNANZo06^4L8iz#V;Z`HayVV6>@Y_D_f2ea-6|FDBHqOx4fLJj5PRKMQtM%&oekC
zEhf;-!SuZfpsJNmYF*FTBlnLD&2C8dvsV8?TGQCUD>A(lX}n`{*R+|ldfThBBHgWN
z^1~YZf|kGVFHkdNx%kIqlx4*RfAVdOG-b)cj<|%Rk`5F9_Vm=0<iv#d*chyRe_t;T
zH;_N*5&wTGaM4F%Ar&v7qHn?T4o}X{OoFx^Qni(h1B#fCBljk=CS_>MBNH+TU1EI;
zu2)!MZc>;x&%!`eQuxLi_F;vSR~)?iKlcx>>?#gtYkeW3W8oZ}T--Qqf9CAD^A|2z
zzH)VMYgvjvo1r-TgN%5;rbqUZk_+kzQ^GuPKZ$DgsOXv4I{SnrWMyS$W@MzNr==yu
zMTG%o2ht6fKkgy^{~;rW3t2s<pf4l7kWi+Xu|0z0vr=Ni{5_oPEt#g4C{}jjd-%9H
z;-nm9(gOH}ye8d>=NS^0e*w1M8rgaY>dLCI{4yGQH=KHUdQ)FhW`K<b^nx7&;`3`e
zr*u!9KC@?b@7zf>*-;)g`to3b;LeZ1<_Fh_DI1_-VCCc&ol)9YnjY!RHqnwn-x>{$
zN<vCSmuc+?_9i1UJ0~~4pr|-MJ3To*Iz0Hl!QIaVU#n3<A=+gUe^SC;C)vM=N{x#M
z^um=rh6xJZ*^aK*<)aI)A}1${MMg$d&)Ayh5g3&m9gI>vbGkMFDI_6S{6g8xJ+`!C
z;nuU;7Il`#dzh;tQ*XldiOMW*Zf<F5X>Dz5YpcnP^XGw&7ofX@=LVxTkkePvL&q>Q
zF}HGJMP{@w$5fzre^K<Il~Fdwbq|P4O3Td2FD|RBsjI0dDagr4Nr(aYv!C_nBv|dE
zp#w8!G!~?GXu--0F&g5?Z(BQhhsFbNhXcSJ8l7ooW$VBL;GQu?JYsZ{qLQ+T4&BVw
z(cLc$#RFch&{`A1`V@_L!5KAE7HwV9U7Ho+$W%f`n`!3}f0kTWUspS>rn;)KqC7h;
z(A9>4qIwi;LUbnucn(pF7{JI(NI2mXMyD5zYpq6+AbCo#EkHhA8(Kv^p>fF>xkVM@
z#y5AgOsGe?FU-X)ia+~de@;T0cA-ElEvKLac@)|C5cJ}rf&sj0&d^iW(lazNwM32{
zWp@lcbm1ZVf5reOT=m82T!_CH-;s@rs+vRvMvi9V5uIB*xxcF}FUHH(Kpur@bZghZ
zxU7n@lA?m#?9B9(D1TRb6I~P%LR61?u+;g(&<ssoLt`6fpYWud@`g#ZBv=S59_*B)
zqNbslos(x^RANSc>9`4PT|LvgCbc#-jH@gK`LiGPf9E8EV4jq*L|I-{T}yaV$uf&~
z2qbS7OqvGS6}7Migf>wMA}Tm~1`Ko_*p7TpKM1AWU7dJrJ7g2Iab;dw%Y@?_lU3PT
zl}+YsP|Jg`u^=}SMPbq5Awmc?&g0RWs{I^uXMv)((D=;anu${y3X_7Ip%6s1krXua
zO{}?Ye|}-{=>?VJ+o$y|>F=98qkD3Db0f%~{jfhHp&}*<f(#8bHBB9wAx%q7Ne=oY
zqO!<wvp8;k9=HInrHo5z`gCJ+KwRjkqn{cpA*-mSMWdTouxuT;POk1=K9D?nxV!OP
zsdgU$@`*?(OvNA{6!1$Z>zO*Z2gPQl#7Bn(fBLvPW8Z*b7o$YP&|y+hhYGQ*l(@(!
zVSU#pG&ZfEqM>c-gp!mH7fV9%qcK!OUR{rge!?p#CcUVpxx07i(BO)t3+K(5Ic*Ba
zpZ&ByBcY6>q@)BE44>7~H82ulBPnzP3hYb`wN>Q+2C1%1Lz-ujh`2H0vPv3ybW<yP
ze>R8f=;Y!msF(Tr`nhq*hC0T2u=)Oxfcv6rDTt%!L5GPAcTA|irweFZUmFnIq9no&
zehUTdC59z9e7~@`w7jy~iJdbimZgQcTItCNasF~@x(st$o||7(dU0LHti@}0ZQHbN
z&B|qq=J$g9*-!g35<vzmEd>I#9e}s!e;K0S&6G}p-AIHT+XV-9YVwj}#88x_h8#W9
z5czLKq?VClGRj&8CRPs40ztRl)5kwBBrGyOpk@q=XIk2$gwR-D18jgWdhl`m8R~bm
zs3iIw+B62zyb?YN6h<gMJrDWs_U_rOl^GF$1Q%lW<W#f`P0=UyibyLN-#veDfBUgR
z`*&~OyngMf<sg6d)BcQv44$QBK%|in*jA=FOV~9h;cqt31SMMINGotYbPcKYxd7gZ
ztUKS;*@?#mNADXH79F1)>E~*P<287d5;z1(mk>Re0xk-=Ee0A2;=)=D1<IznbGU%$
z>>Hk(Sx{Ef*fwRx+>Yw3D0gdpe|aiUA&`}pc09MRw6cjamu@<A`TXhQM-J@Sv2`QJ
zpZci(Y#9qefg33d3|2ebdEt#mwq8YE3f*Q|6>U1S<0$wWIZ&YmFlvC>j~OX0ucl`V
z&=)TcH&<szF58~t<m%z=9~1_-PAh!lg|8Ao<=<2M`@j&=d6?jjp{gJye>zH1NymtV
zlJuy=^xTr_hL+ChO~}-{SsN(gbQEe%qSEL!7#0v1ms8a`bJ5VAom)4pU$bgp$wH7n
z?V*1T0@qra!m(4m85vw^m!y0=!-Vv=VxuLASeft3$M82<Q4FGjh+9)#1yEGtW5=LG
zO5fB5#nV2X?yk-}j=ilte~0JHM?s#yJJ;HT20os86*ZVawPn5WjsH0!NZY^|5J3zY
ziX$Y)N@Ms1%GDy`QnHIG>Y6%g3lf7}tqqjNQX-|GMkJIz;mLW`6T9aQEM2sqZ%)tj
z?#Up3-c$b!L<vNmY`qFc)R5w(Bmq??jOw+?8CK}Z!UQV7ZOpNHf5`bJB~@i&T1t%6
zilaw>ua^hkiR)m?vaz#gb9pGv1MEG_v<!Qd0zl+YhKAdyC{Yx?GbI25z>{*Ic|8qf
zS#env97Q;J_=QF%W#%JuUzQac;LI{q#qD@VMPaN1GCmuq-X-Uj)wOg_nbh7gv9Y1H
z8syJ=>YssVg1iTJe-x9DLW(D0EI5ycsRe*p5xse&9aBi-V}k_{O={+i7?DKwLk>@=
zI(r5A`+(x@Z7i)=ENdIUQ&F+Phm&#?80$mDQ}Ak{8=$U8&QDp{@%TPLkqK$J#nt&K
z5x$Nrx`yDRi%?S-fU*mZOU)^+t}HJtF38Kt$^iNEp897XfBGN<tObQv!YLVedhqaw
zF1Vtys*18Afvr@6POC>yOlouh-yU~bDF;D#2;?FZ2sD;YNT9DL--%<#GB-CfGcz?c
zH4#Q<!j`{a0G3p|FNwla7^Zk?;2AUzB_IsS)Q2ts(yD)0Y;sm%dTg*K*NUM9r;l)6
zVls+q020Gpe`EiUh}firxY+2(h_I0V;_8h-P#Q)IqzT7d1b9UvyOWWVmxs;^cmP$J
zDa_a;3p=gDvVX_|QP2R_)0XG%7Z&8_31BxXQxm2UgKnTt6HYx+uaXd;0RZA*?%?Vh
z?#KT?@d6qIZ$5Aw&@`1))HJkUc8ZDYzngDJbYezAf4H9;+sa5sS&?vEV@8QeA-e!L
z5nFpUax-ooUO3Hgad!Og2$dBG%F#7&jy_EBWH*iiHDHaRBu`eu(30yF8lP24;i807
zE#&+;x}Gqy<{3bg=pAe<jE!K}K@+2L$r<_&xeort_zng8fnJZbDP2cZO<hAnQ%hS%
z7iAb6e^>9Ihy-XBI#@DkRERlptfZVW%0^8rDX~D7$JE4_X$11;J@wB(96(s6hERza
zHjpMO=z>9ymPn6qa~55FZKjQje{@>WxPs(R4?Cs?4AK(4+|i`%mQzrtF|F7x-a%2J
z0p4zo4%TKyhI-m~l2tgvOug#;xh<lW_krSJe*%W*85|oH;OUH^+8P?_YHF&;0}+)X
z<irD!HNdR^+)xr9J9>nOU^vzU=CxtG3cWx@Sy_pYKk1Qw9)jy?6zKhZ#lKHY03;U>
zi|La6@1}~(Xdmc*zc*_^P-AGpI0Z7mF-RNWl}+im)GsAA=3lYT;L07_&0kpEqlCbF
zfB8lv$AtR1Lr+jhxuB?_XJl#T=<Y|g|A;Ah;Ag;N`b61BO;JWd<UNeZUz-RA<Laro
z1wjEvMpht1l#XsGRgs5wS9pASetAxOu)Dpnwmfw1N8w3F5sZP>H83)>vg0_rdHcfz
zl`Bk8nKGblCo3T$MC(!jDIA2uufStcf3_$I{E!!kg#E;2Cr5|)xUyNsG%C%2CLM<P
zI`e=6$&^V<(5WI<51bq{Ln9LlDvF_~c>eN&83L%Gb4nRFkk`=C!G0N@uO!(7<+^zI
z1cXG!CZ=W*XeyM?5lOSi7^+js;W;_OP=uE^QNMApwT6Z*MDwGmNmy!jPF)pUe_eT5
z8LHg!3j=b3R$ftNVq~C)qpg_%avf^ugGnK4j>kJ}I1Y9;EK3Vy){Tr93_6`|Xh=6=
znwXheT3MSL8|rGP$V-V*R6KuqF#(~#5xA2LF&bH^HD_T)3+TnG=|MW>6%>U8j~qq^
zd%HN`@fdt(LH#l+I#5mX2=MpwfARM8aO1l;J2^S>xLgj~n!sd9z-%?aOhl4g7?AE4
z-B6UE832|?*o~!x#Asw!7No~Rwr5G#*Tu6QlElalGBPYHD@zLtlsB51TUc7Lpk_b`
z3mS(ou1itz{N=>}gc@G|puIs19?|u*RP~H)9KAwf(+YAjQxjt&Lj!%?f1MmynEM2S
zsiDxPsKvl@^2za0VS&D2<Q>=?F3-`)8CrQrnTpa7!V@t(RfV;&GB;+R8?25Di~)mb
zVr~fodrGpy8T2SrR#|RJbf5>%nrR5G91=qbfe$h^F*P+ax3I9Zwj)3QzTRGN_?#TM
z_AE0zs4YiP@%-flSMI24f4M;OV`WwKm@KZFe|V6On<EaDnYJ!|QK^MB&Gi*<>-^kE
zmJ@Vnptb|bmr^sZ;CO~6WS19YBt!<lyquMd9h>Kjl;DB_2?K2vSs@<+)z>E~JjQew
zk<~Xq23pv!qlCg?jw{JZ2=j5a!~81BFx3YmHppRU>**V!Q)6byf3k4^%ujH5Ok!+Q
zILuDDxe%I1cb?)*_{$5<YbfL^X1jpw4QJ~f!Lb=hxH)cZY6uwS%(BKQbGs*2<t2o8
zu+5BtiGZf7t%VXRtb184V|(||r2LxJy3(xp5Ko>J-PqjPp6BB314)>rk)FDOFrCA7
zIw;o!fo(vVAuBvOf7-aoC-vT(6oI_XtL>anm7f|F;O0Q+5vUe07!M$RnmYP)V>3$|
zdnb3_;3%l#=4YiQ5Lg}{?u1wT0(IhFREX}J06I<O*0NemJGbD3oXWz?gb3VGV|zzt
zR5Z_6yl&C72_*n-)6v$}(b3h_)6)apt3ii`X6YQ5m|r`oe|PfulC%gfwy8EI8sd8g
zg!#KW+L{__DG5nM1w0z5@lhdu9xfbP3!pE8DNO~6#D)@CT}zjcl#&V4+v-cR;zB&R
z;D0nBTpLM-qcXBe>X5Qq*#XFoK;_g|7U!hKM+AAh!I+<!k%10H#q*aJvL8Q;jRucV
z$m*E${9|*fe_I>M^DsA(m1{_9X>-rAEk`%?cUGkYunke*ucD@@t!F?d8Aw!>>GodH
zIpZev4bJPV&5ic9rz>NA1UJ9X*ic_S+tNr^1*i^zvj7uoSvmPerWWU<#YYCg^cB%;
zSCql5O7G7}$?CD(A~Gu5db^sc@{=QcoPbGJOGQp{f2?riLt0T?mku2n@6fo6qUxs3
zSzRs2FJ+*Ui!{%{nio>>{8xTa(1YfOXH>S%?wVL#kP6gS-f;zuGnQ{Var?-cp87OD
zYxR+1#AOInjDUPb)x<F<rJ}8O^`6ypnoHunZS|y;V59*LR0MfB*_i073%JODN{?yp
z5tdTae^6PJokG-QFv;m~@&roc^ldz2a_gonoY`Jqk{%P_YGc6!4g|^f84{p_$KeOW
zq!-n;PMf!)zo)aQsxUn+*xyT_`A?(b`419o>M`VkWV9h}8pUim9wkt*!Cp?b9ub+9
z?Y%2^oZQmiQ5xrNt}=qGiy0al(G6(0lE#1=e_u0c?ka#mMYvnj<cLX{fXJjs2*r){
zRE0NQN}XZD4@$_bF3CxagJCQo@|qf$!UEGBiux?q(3Ilw-7U4nnZVg2;La8y%Y||n
zib==J2OdE&X@%8IlV^9gH`btcj|~j~$S$7g7Vz)><poFe>T0SgO1SGP=#UDuQdNv@
ze`5$>A6<eH^Xoe2uH3qGYF$>S6H{q~FcCM@*VWckQ&G~lat=tyt7+?LAD5pH=)%$m
zDjXX}kAR4TaDO)rbk0a0k4-%cMMtNX<fg^LWWBwGv7xRyCP<=Ypztg&pnX+<JtHcu
zxV|DkJucMOl})lKA;p8?QG?#R8RYN4f1=_DCMKt}wt4)x@}ivd#2CmeT|n`S5A%h;
zs^GRPTv?<#b>X;zgtDF)+dVigGbIk1A6#1to_9oQas8y(OJ__diubWMR2YHjQZ#h*
z(W}V<pN@&WTX1}Kd1Ga6QkWOd(h$R&9X<TRVncmg9W3Ztijq`RFQKAu&h-jSf6Pmd
zj|jv;zX7hUVLEIPQE@35IRym;dFThr$w30(77&$^pOF|vjObzrKPX-lTX-9;n+wpF
z+psLmY<MWVi%rcfEy~NzNKK53jtEAI=Mn`85)S&87u=n*wYD@jK_RuUSq_adOE%v>
zA~e98u#)!t0GLo~A?FtK<gnsVe<V#~7?PDYISoS#wmVW!UK-{>LW<TmwzS9SC=xj<
zFHX4=7@xpznh**7R4Zd4W}f1|P{R#46%}P|L#)ow*a%FnZ)XKKf0)@386mEyjj8Km
zaC;5_IuPJ__=iR(B__nj#l}QOMTQWJ{Kgpn{a?j{c!9XPI&$o7EX{-zf5Dha%FLGM
z?urp}DAQtb-LV0!D-w*MllW*obYQ^HE6U5L5ctOcbi>@89juWXW|&zcCHMw|i?=bR
zN$nak001BWNkl<ZsmY2`83a-o)7r^9G$AsG;F=XC*%Vi`uC5Ns8nv{58ruSfvG7N=
zv$ix9kkgBdkW#0!oc)o@f5=Y4Z9N8tWZCmvJ^ca#gF-^XBO;?i{ekez!uWTJ>i-kk
zJecqeQv*T$jOx}<^kOVa3v&#g!4oN=u_@UJp`KiGeHBR|oE~3zXqYHK4c5Yz<ATnS
zt(7^-%S|k8InM6BLEz$9D4UZOp|%JJny^tp-q3D?@d;UKe1hDCf54#A4Gj$R^?{|>
z#LU9V8kuq{3o}z=BMkbL2gT#eGcKpPy|ySd(uZfM4_x{H2)AYPT-?0;0z)HW@G1sd
zln8vto%E~aKcuUh29gvT83HDqOHl+<<l-2Ur*B|jz(h9=-Q8g{MwsvpkZ&Y)u(YHM
zzNQ8Y-1))LgE`Bde+z{>aPgLOZ6zro?1DnkLuC-=>4a?M0(Nl%E@Lcgw;2dU7dQCD
zMMf$Z*m}a$?SiiHMYynpR}pAx>*=GU9*^pVlW~f<5lLwO2?d=0n1+4{2$VK4SVl+@
zJZj8XaX@4!D5(H_l@3foqg09GB*aGx7~|1}SI1UbTmm}zfAR`S%4$$Lr!!5=S++o!
z;O!4C-rPV_K|<J%r#1!gxi+SFivZQXA^N*U<~Fc>!V2(G#te+_R+d*(2C$vBE=GGp
zU3a81(=j-sX7a*Ki@WP`!ub}O6jlfi0Px)X!=QMF6)y~8extmP<oN$t^W#Crl^13s
z)rWjQ5j<kVf9SDf4@8${V8~z^o0?P0o<@<4F+E&ng>*nblr%y_Qcgt^xmv-nEatg0
zeIHvvAf|zy)&_UqhjAZL1g0FegM+=DtqrlqL|$?%iYO(BH7M{qAtD&5YVICe+CG2d
z$<6)krLpejs)CO`VuXZ}4%5cTJ1ik87}nC552I!He{C}2=kgoh=Q9w@5P;I2^yGw?
zNF0g-GbKTb0d-K47eP$}Zaq!kkd9+`z%$AcbMyLiGM&SecqCwpqQZqC3TIco8|KPE
zUPO|B>i`%-TppQv$hkeynb3dlObAw9V4btz$_tV=;vg!bgQyA6E5OszY<=QOn&%Ae
z8=N&Ee?P*-L|OPg5I@=w@uN2}pN&7bG>RlZe$8_tsF%p%u$Z_wF-$BWNlb4LI27_P
zBbebn_fvuS>eEt_lThXp6CDJdZe0~A41AU#VATqWN=nKqfNBLytDZiMM$<##wyusI
zItE6_1OVuTDB7ZQjOeOSE20F#1Kb_vuO{n*f1^i9D(hfsqS#<h9_xMXY;v`oDb_g(
z<(Rn04$hfM^iZU)si~o%p{{Pi3rHwxoVIw=!YMTw!91g37)s3Af)$SvMte&mJvCrn
zmWS~GkdJ>#0%jl*m?lS%_$ezZqxVx3z!8Wcj?n}Vjne<`rm5dq@D{45=$P2J`1piG
ze*yipkO2iH(ioQtgf^<`0BF{w5shyPVmijw&dwIa{^n*T#==UtVDJTx%5vEDb~cu#
zWQRmBsWC=I-N1t55geNt8{`gb`NJsisNfEe&<qsV(}=bj3dcpDJPW}aK848r&~5l3
zNriQjdRO#JEJ^eQhD=HXbUg_X0RY!Bf7F5=v?3)R|C9*!uhA&(k(9x`DfH{%_v_Jg
zb+j}AJBEqAByrQ3xcn41;Ag6OZxFx0knrfZ#1xoC;aM>><%wNiFf*qofV0ZOA%~l<
zg&|pX9GH41ga^JDtg@LASwAOfPo2Fy_|9Y<fbs`KVOUf|UYlvl4~WSq$%yuMe__#w
zuf3s)!A78TX#1+;&1e#$@aW-0A}6CmGh`T>TCiXuGQYaHd!CRRO(2xw!~J1vioQCw
zU)XwqeDp&S?Boy;$Fv;Udej0W(%5Ok!~i)sh)o)4_(%PG)5b^t!hd@xfl;<j?mj^g
zvB?>E$&r38w#K?j;+X0{*TBe>e`!d=8$V%~LRJno^+DKW3lnQWf&PBJJ}{gD6eq4O
z!??-GJ)#}~J_O+{Lmz4}+FF_#>Z-~lpS%*yg6k8RR@_jW8t&y_JbclWph94vk9++X
zjVVX0$Pyz{4eI=%d@v>@ue`1WvbkW)fkO$Qc(2g-a6jl2=&DHLUNVf}e}H`SLlU%p
z@yEiy*uoNwoxKCw&c+IzYS^-dQNOSPi)?oZ7(xGq-u}tAJ_sTI9jeDEw>5|F8ycIE
zQ-YV2Ky?Ol*f2~iZ7fa59ytnUB=InfiY}9d8*YhlF;Nj=!GZq%etv$wNPJw`R>U3?
z!T4infG3HzLta8t3z=sOe|*Nomnw$V$VeB|P7<uP6MiUaV*w7~@X(&Xivr{VQzYYb
zkL~R48yuOCmIHIr<(W}F93VKPgj9S8#%DsOK%58+X@*RYkA6thq!2Z#=^&tZ!Ie!b
za}y&X=0!I|iegO{YW|lL_fK{DOM3tPHIG*=GOW1nK{06s7~|k*e`TnS|3J)5Y%ka)
z2SfwF361#;am%|PCo>Il0lcL&9E6&A!&Ks2_dpDSw?g5(JXmMED2%2<RC^@UnGW93
zS(Pn4E#>Lqe3qV^@Iz6^Yc;SBU_hy*utdbEpacUnEEt`Nj7>;R&(13<t;~rJ;$u=g
zN`#7s4GJi85VA6Xe|+>q8X=ev!{tnKYqqnin}?^juYX{Wzqbc2B3qc?VIidXVUeDo
zO!*6%`{zm}zBF=}{981SNl!Ew*3SOX871|lnX!ScHb$C2KY{XSj*F)=PQ&CSMniBj
z8sK>Hx@H`o$n=u&wN+(Bd0DB6v2hR>CL|^%h5KShAtA`4f5H;bSzzO{fNTiVdg#V@
z{~$#FT4s)cNk!wQF6wT`kM*)M80MQt;wTEmaHQK56~RdqWKnoF4?MLN9hnsG;F#33
z^z@9(tn8fJ#85A;h5m55TuSkw-p;lrdMeVRV0Q)#)7gQ1^g|-Kz$AbKh8WdDIu8yF
zkBr7}V+t;cfBD~WkWXk{Km+@s=KlG0{y^>jru+cErHqy-$0s_oVq$e(VyL@=i4Mxb
z>894`4-E8lvPDVu@KJSH9Si3mq_M6^EseF6CHa{t=-L15(SupA=LbZkMh5_CwTk2j
z;4)FyF)+43(LJu8$>>_~!!xTo=B}CBR-O{TVGIkhe?KHCA2EvP&=WJ65NU#$@&v?1
zer|SFMp{Y|fNy~reOPutDIT^h;cvZ{XMXfUBBu=`Vbu=!MW5iXi0Ih(q?EMGl!WLo
zj24BBIMD7_7P15&%@eiA57hfF4EZnV{oj!t;PBSag6$R<osw6PmmKcn$YN+>x(QHx
zkk<!_f0v_KyGLc$OzN9Gt+S<}x->s4JFlRqw7jyq7AWg@mIkVVj%!+BMjXtn(>3LZ
zsa0E!Gtm}D!9R`V76#z1mAxG`Inmw@MuG@VRAA_}@th-`kdu+2R#V~ryquuWACsDr
zln7JrQNkty<&qV&j95tTaWM4b2(<#rN#ik#e~*4h_#i@5Ni0_M506hy&B)FxEH0}o
zEzCvAkBJEO_jbqYwxFX$C!$gmpr}IOfB+R>ZeO9u5`U0T1Vd{S*T147fk%*)k&-0P
zD<AIT`|X0CfFs|_KQt;KH7gMrYdce#DkcvEp0J<~Z@j#Ly$@`b7+Bmlb4u&@n$o<&
zf6|JY`o`wA&W4ioXkSN5eHGAqAPNJ<l@M=frjN-hTs<jHdksZ>YqzlUvWYX6&Yo0X
zoEGWBGbd2`N^%1BG&2@a8v+QoCQ1nKaI6w~9AHghXvf~kGcYnDJS;RMBseIL+I~VA
z3}h8}p!brj_)sq%i=ibC<hS~CrYWA8fBxu)M2wQ6j6e(b6C)C{P)t!#Q#Zb;wW$F-
zUOujTMj*5Ag~^`qIF}XC*TnrrsChw$SsR+nNVEZ>tfPxXj}YQNpg4ht3m6Q+3!z5=
zvtmTd@Ox>VjLagSiW40Z?CoL?OJyYSptFOESD>e(?a%6|3Jh*gdiCV~jcb-If0)(P
zHomI3yr!Y4ee$$feO--ZnNhwx3mV*a(0faLd0uJ^K=#;9=x~KZ#sQ7Bi#<$U8QQpo
zrj|5z_xDV0s>XB{1Xe;vo!t9t&*KAO7EHu}{skx@T&R^H7US6Q+)y&%<L&L`<pt{`
z(c6+_OkgrJQGn@*^j=(7Lc;u^f4!rpZ$LNtby58&OxGi?q@rhH=i(clm{V3&+tAcH
zscRbWD7CjB?N^tV6y|0j^~XjL%MiRh-Qf)II1a+XGr<aCY62raX6EKF^#g)A3as$>
zzJ*}=%K^$>#Oq8kG)X~*sC)fhgMURJd08ZbKio#Tj;b79!D?ym{7Z^ge=>Fs&ui>i
zwRhXPfqp<Rmyc^~ojjwrf7#&t?g?e-Q9fKVjI1fHZJjo0LQQdYQdFR)Z%}wNGAVh*
zSqWiaRrJ*uwr(NGMfII?k=<ylF3n39A~G;XnX`vqXi$J3F?Y$f!}|ouA`sMaa8cKc
z<v<pZh7Ws>7UbKO9FJh6fA`wfI-IC-ETNx^`^>*Cj~ARIb-E=iLCG#_nAq0YJ#)^y
zMavfV&!0Pc=CrO(r2g@BHI-$>g?Txd=_yI^vC)y?p(xpfQinSpBXs2PxLhuW!(nr{
zTprI6e`?s6;0i))YQs6Ar<aefaQu*CV_|Hdt)?iA!`%0_tAB!Ee^S@x+R~h)$N+at
zJFcy!AT3I+VfOq5#Veb+#g?}BZ$5Qs=Y|#Nx=yTVZ10}6VELNOyVmwkYbr~R^g<Cs
zOnOOO+w}RfyW1M8inEhr5>hgAi^{9(CsgKPSUf|MY3CM_lwZ@<GaWs(<_Y!VDhtx%
zLLjEVe3w91l$Mehe;*SSj`WWmC^537h1+t1^=PJ0H@3F19Zst>22aqUKoVDkOX^yu
zw~?ARH#NpBC9Pjqxf?xJTuKgQCECVz{J_}E(uPUhGkX^-8Cbo3^Tzc<t5>a9wghO@
zX3dz|)!ETHv9W$!RYhr0L0)!ddMe(BgK2jH1EAso7RtxRf7j2?KOitLC^$GcBqSs>
zG%OU{Rd8?+K!#AV5T8iB4m!#<upjOh_VIjztOP+e8rC~-%?P6+EiNLZq-BU%<UVYj
zsaSX=RZm^9@7meJJJ&CtH@&@~wR`r$l^b>*Jhf-2e`;e%YM49Hdoj}cf|X0=^>nq4
zuP(_gMAosff4O7Iw3h1pq!163IT+i!1t;WIPMkV*YB!RuuC9sI`6*C3#9Ki_V^ed>
zN(9#@D#YK*6*@h55gsyoP(1}z=;7k22|#QqkS@_1G-a{z=RkI6{<PNG;`Eq6H+ySA
z&)`?p>PL$bOuBmJT<`GIqPq5;zJ<#N*KOXp@9?1mfBW|A+OciRrVT@D23IU!x~P9%
z@2nZ!lR^Itbv0E2mMxTaB*w?Z#>T`%N5i-=G)Eu<PD)NrNl8sjOHEBlNls2mN<?8}
zRt}0AON$Hh;dRN0ktnr-jgU$ti_Gsc26Px?!@{n*>_|5&!J7B6qa_tJ4G4~c5Blh8
zEWh;ee{<HKe01~N;T>z2^ma|0JZs^qO?!@AxOIBZ+67Y@3KN6idAH42uyXUz%Ej|~
zy4#!T#y6uVpr?2K;+e2o%HPRC&(y&!C_bxfd}kNzpqbadc-frJhSKyXUq=fz-!D8d
ztGH(3goawedS)a>hx)lYIgrej(qfn^4Uf!Wf0zjPZJ45zBU(h5l7oZuig7Jd`v&Gs
zYaLgZ79BvYQb1-C<X0#jqYHHnST2Eaxm7JQ77nc0ux-zw<7Y2lx^VvNnNuf^A3GvA
z0vmuLd(i@L07(CB%}wL$$5mIBmmmv}o0F4`_v7H?H!r`Su&}78xTK`Cw5+tWq@=j0
zf2a^;3Kg(|tBH7B1v)p#4tj$f#1&5o(cfwC?>@+hjq_UaBb<%oi1rC`BjWOSE9);Q
zUfm`rr*+}Z%Wt0CICF5zVE+tc(N}ETd-BTtm$y&vTHV(<E;knL8R&iA&MoU!EnUz%
zV+tsJ(ehPmH!Pk3EYof_2IlBO#%7=ce==qIoCQl)t=ob;#nh(Kv<Oc-7r%(4-13I@
z8Q6D%=`71j$KZQAOI%2l78UTT6Qz4x6%<mV3W(LCQVXh^r}i!1Fwi%(xw;@F5~F=E
z#t7u2AJP~xDR~SexAO=?@o3ll!HwJZ9zJ#9+MP#_9zM8#@9v%3H*Z|Ka{1!<e{*L}
z9Y4H(_x3Ft)(#G!003Qo2mm@-n<tE~gS8t~l@%4`W#tu>)ivX4>*^aC#y2)jXlk0+
z)HI>7aePBVedENIc6=A8z@aoN0p}zb?t+5*-v=W6KOpxHuj(p^;Ajia64_;O{{>bO
z4g0X7$tw=s`}^C+*G}!(u&j6Tf1IW3cb~la@Xg;}-8#8r#q74qjH24s>7e&h#}4k^
zwsGysrTvRR>05T}Ik0|VcSByRj|0PkgDi1cK~+=d%=ycQw(UQ0>EOo2(;JJEf*ifW
zl5#62OrE_&KtS3tp}IILF#@8H50(m#5`h*y-eDjoEnvvR6j>3;`BhC_e{+|u-@ShM
z+^(j|yd(jG-mu>P#|SYjMx&c>ePeQJI_3;)-GkJA{od18-+c4#-Pdp5zIpT2>z7}?
zc=q(kqlb5IUcGqc#E}Dgc5d6eVQBSAVf@$8-qzaE(u|ZmvALzSy<<}6<SAX<-BYJc
zpD|;`^l8ZBbxoN(b;hjTe|fNiYxxp%Y|tOy*;t+(ALQ;}s)OM}pN13_%0ECJ9bY@M
zDxS}fq*nci%Bm5U2DVTzbWxlyCRjnI<q(?RzIgZ5x6f{0IJ#^7@;+orH|{}-fBn<T
z+h_I;EkprcecSYTD>m#tcHzv4!~1t_+c>mp&AN?Sw(r_|;Lw(Ve>ok{mT~g%N6}ke
z>A0pz)8{T8+_>}L$vx|r&T6kMObLt6D6DGinBKQ+$>N3m^XJX&o!#BqP>~O4eOI;>
zfWyRul`Oo&Kv6{<lU?B*2E>Qw!luF8it&?XE?Bv6c2`SXS#D}{Fwr|u!zGiCS3C@h
zY3LXj*?UH$!^rUZf9?B@p1E@S(U)Jp`x?nxByYZYg;f9i+0%#jZe7289;yD|ex&;?
z8`pvFU$SsPU+=7*o|!YIBMpP9XZFmRJ*RhWU*Ej>^B45@_b-?~e_mhT++O7KmJYz<
zTDN8uLA^V9LPaiyHUpmv(Dx5p0{DrFe|&Poocc6hOBHHRe_nuu^a0unQ@XCY5NM#|
z5S%k{-lo$}?p!`~VEfSW`RKuJ!j}HK7q`zH+`OW%Yr>=%^9R=NJap#TrE{l_9on~R
z+omnscI?`-@4%rWJJu|k*#=1H*rfEF!m{f6i5*jCFIcvE<Bt7X*DRge-C9?gU06{!
zaT0tL$x<wRe=|CpYD%(`qtK;i8R;lV3jcNi=2OVeN%5chhbLqf*ECI<*@dEvilUsf
z1X!fwXlKd9Io+>NJXuq+;iC++VM^bsjXMq;zi|Ejvo~)Ag6{lltokPp@7=y}<>I;1
zCys&ccWvLgdBeK3gDVHnF<!K2AyzTc^uk4pmn>bje|-7Cz>1YCSFKvLa^;GFf#u7V
z4Xj!{w0_f;?K`(`*|Z+LZ26KoQ<|#slEQs(!wiNC1zo>SSNxMRo99o=3+3ohSj1rq
zWI48$Fmong7ZcZW2+C~eS$*)vrPGJ^Y*{<dKWEY4=6xqF-+lfLY{1TS%X+(J^{-gJ
z<KW3lf46U3y?73Jy#ssq?%RLh;Gx4uj_%&La$!&Bges`=R@KxsOqkfwHFNI5WrORt
zqHtjT?5UHQ>nF5M>Y6cU-oj<eSFBt$xMuCp`V|Z3z@;xpj}7wRSQx0w2w~dPBo~GM
zK~cS$F5T1`o!j8(<m{p<lwhOuDnB~|^CWn?e>t&jKtB2*;iw+K=9bO@ak(f1U9x8L
z?!%`q-+uh+^;chg_2$hRta+^Zm(QO(x_{@^^(&XopFMr@_|d~i`MY-_?W4qR?b<bK
zR<9mhwQ6v1HTtWD)~#Q^VZ+9an>KCUym`~6jT<+tU%zg{=B?Xz?%uoq(Eh!<k-^xs
ze|~6bZ&x!OS$21{#p~Y5=Fg{T9xMLox$TQPN@LwjDHbGR(*Vln++28UTNds&ll@eE
zhrqOP-AlKhJ#lF7_Kj->7WFM%vvuEzi?^SA_4M|oll!-=T{dqCHr*F*K0t^6DtZE^
zjw3mSW&idKgG>5mPHt%yNIQ!B=JqdHf3bR{fK`0vtT}V%_b*zqY{ek5d7HQG*u8np
z(s?sFC)PsykZ%VZYt&!{h0~&TVpDtQ?Ht{Ff};}Bb5S^*lMMlNT2eeLIr8~+yJbfc
z_y7$ZGp=`JM%jdJ6qIb;cl`W~2VcH?1-i!<AL$>Rg4Zu!JbV1`{@vR*u3bgZe;x|?
zPaHc6Ss&8l&K*0pZ{N0U+tzK{w(r=nbLXzzyZ7wbvlrcpef#$A-Lq%+?p?d~?A?Fx
z@X=!@(93Y5uzSb4WqniekT7QTh6n|7z<y#04=eup#oYtlHE99%)SxmsLgff$M;@*`
zWBMxyoDJFj$>r^RLkGZu4-F11f9W4sheERpHy^%yeCNv9qx-gRSharJp2Mdv-MIhk
z$)g8%Z{NIj<q`^b&z?Db>g0*TDAHfQX2lZF@QkTbCbc!UK@dBqZvnP_%a$Y4w_*h{
zdjrdt3j)F;ClBu2uxe57w2sEI%xFJHOPbPf5I%It8D;?d(!#(mYPJh^e}z1JgTkX?
z;uBNSGjqs{Iw2O*c760iqVNGsJNJ;}g4*_3P`5aA`pVsBFOceABgGSr9t8RN7tfv`
z-QT-&`__%?SJB@NDc|XnC*bHGJ$mHG;loFc965UQ*s<fsPn<Y$63MAkr%s+k|3<gt
z#L3fV&Yr(;2_=f>&Yp%-f3thziUl(}##iLx?H<tUq$Z9(Rq-z_?;GrC%#Y$zE20FT
zcRqosa&i>59T{xDgyIP^mTlc4s9LSsuw(!6^Vjb^dwA#C#nTYT?>>0^?3G&&o}p0p
z`P0V_AKbfh3#B_3!T7;B-n$#+yXyv*FYN1?+S$_9(K)4i+Kisre|-y=u2{W(^L8R8
z0{=F+e#@@?M^WB%_UPWNLr@_e5A`8ix~g#HDGB9fK`f$!X&R{&2!I{4w6W*n382{I
zj9kc+3kve{a<Tjh#pC5?N~(q|mw>qJil*+~C9AjWJAU!jlc&#Ld<mL|81L(MAYZ@z
z3Y<L(^PW6@MC5w6e{SBme(ma&%a<=-x^(H{#fulv|Dk`O|Gs<$<SG*MKluBXuUx%$
z{l?8(ci=_nHJ2_N+q-#i(d;SBH3hJ*i^b5OHhh1w;=jClXkA}hd2#@?B1*7v$dwN(
z*WEolJPCdxWKEfDpV++G&c4CbgM&m!vS!oH1IN!@y@Pz)e@()tpE!Hz`rSv*U!$i$
zM7z%(Ke&s6-3w>Y**<j|3j_$z=4okdYwy6R&CGdA2RDI>ICK2a-W{9K$yl=q9jUV@
ze7<$*)S+FQRxj=Atjmx0<C<s)r*6rmMs|SQCkQo&?i|YRfum7Z-_XcJz{XioQCU@u
zGU#!&ARn)Ie*sl1%>vnaDBa=U_3+v24<A2$_TuF$Q2e`h-+c4Uw@7e?2Ob|ee2DYW
z2|z~xCHxPdW^xb7-FsN>lkxyW@V$HJKOa7N{20CDA^IXZH<wNxK<=||IwU;do@`Sc
zIf7l{(`vU^@h@*2-Mpx~E;s7E6;ZI&4KPN~F$fIyfAw%?<AD|vj(1dM`NU}paa4*7
zGt^!V96x*cCK$Jy*P)Pc_2%7&&tAUy=7%4?|NcAtC|<vK^5D*mD;Lh3yKwp1jobGg
zJidAE$j)`k`=+%*Zrj{EapJ_T-lao3kDR~t?EZ}lClBu0ymtB8t$UB2yLRu{+voSM
zpFh5TfBX8Cb6P9XLtHGhgzQ<vCvpriA11}tM~-wUD(jirxk4Ea1(S_U&8_W|e%;<V
zs?Vzp1ON#+6-`rSFJ8TM|EVkYkmjGGEC1%}ccAs}zW*Nn<nO;jg8por<$sOcKX`wT
zS0FE6zWP93f_#Ytf0Ou8I5;n!-@AI|@Xie=e<f`nmzNOeY(-NL@X36#;$Pi5xqV<(
zb6G02B1&+L9P#{9a9Ct444eYxq@EeaD=ek3wjH@;yz6^#?Z)kU4jw;q@!IV>ckbSW
z(|rHY(-*Jae)Ij`{ucz-h<<qY8hL{2m(E_ie&^w{S6_es_csr(9N)Qi(ahGCiA{~;
zf9q?<Rk!sl-EiRC-B<tf&5Qe2PaoQ`e)+l`2TxzU_u}3Ee)syx-79C1y<a+Qe15d2
zwXP(^o<$`fE)JGP80tA(2!M->rm=%(cxq8YQ%lF>sWWEJT>$bc6ptq<q%cO!FDk95
zwr%>nfsK2PUA%K2Y5v7atazmMAAbDtf2Y4;`8$@sy)Qor<VQ+=5Xko*2yuG8{r2U<
zn-`An+d8ytR%>Nuq^BKSiQ3Zs34#yWt<$^L&hKg{N(%RLbuc&3kQ1eNWnK6lUcUZ8
zAz^`dMF3M@3$vC(Eq>DEuBpf~_x3GXv3ApreMe4RAUqUuQFk6ZdGY$|?|vfne~uo(
zcW=IYeE0grGsjL}xON+9{@o8R?_NBzZFT>QmX;PW!k@NaaNCiK_g??~>*x2boI1E;
z{Qwm7F5Y_l`une*-o1L}2yzY!rqmb2`*TgyVICi94kJYIRxDedJIR16Dr8cj4hO}E
zT3y@FII*?8vkT<o6;G}#l~Sdde>u1X#vs!%5$^n!eJ4(zyLjdLtvmN0A#eXBIsk8}
zn=jPqdoo<Dgnaug<scw4Kp6s#7QP`o$9L$=-?;*R`<nUaZ;wNMPE7@P8pT9DDZ0A>
za%#`I#WPx}a^irL%nGgzBuGqiYX?VX7e2Iy{m^~q;X*mgzqkj-Wfa%ee>XHvnAqIX
z*87NUv;Y7g07*naRNg&n{!%Ew96NsU6bebtoxgbX*8L}6zWMe?G@bKXWna9<vu)e9
z?V3C_O}356wX-JMo^0E8P19r>ldXHc_qq4{0q5C2tiASHZ@fNL9(+%tezz|t?}TYH
zL>s{V;&WvyiICs@{cl1`6gvZc;8d#2fC6<(_{U~nm@sY4I`F)gRqDZ#CsgOR(TB!y
zZ&|zPhu|&h1C`jZLfi~M%ac2}BlVZP%CT!FHsqd;es%q}NMQz#?T6Ua+*0;mz`vrT
zg#@Gl<gS<06x*8(-}ccphqYL{rd6#pm_B|;jw<JNI~}Y}uX&IP4Y-pA`I>?CEq)eu
zVUGQk@OllS`xNr|AfiQ(Dgm?D@2t%w7xmcPu2K48!IeS-D+uUXp>r#^9gB)z%Zj_#
z=nGXA)$j=ugeg=jAvYJG!5{__2ZSx9pb|>x2Lez$+I>S=T5@_a#?I_bDHD^7r@AF~
zQ-&>|^J4caDp3&8-M$yx$$a&Qx2(<Q^=NIcuiNXyis`Gzxv=ledW!C2UQZYBJVc@=
zNt=byeVbq|3V`kr@2tH(_s2`an6C-De;qb|T<t?Av!Z#+_Un7NDjC}(=OG4Qx~Z<R
z@MI1N4MIq{w2{{Sf{Mklk)((Jh*qBMi>c3E7y4fZY)0MdhaV<ku8)|MiG`Y`>1fmD
zFl4YKVe7wjpR40>&LdX+?#Nn^eb3rn3J#w@RcH}FqKAMc<nj~aC;(N}?(vv?2HQZQ
zCV=bnJqs7rQD4O8yl+Oc`$97V{oj&zI;V{KT@qF=zL_R4TO4+iJ#+<TG9FA`*OjI7
zeXZUrHZ~TjFVyr)qP)$2tyxn&R|a~ogL~!;SDfmgLQJ)EVdsz23t7V+=;2S|#gD;9
z$19*-xDIm|GF&-ZGH^Rdx2;mp(#7<P>|ZyzgKsrkJ)cjZgs28xKG)kYH#Tca$W<$C
zrdyy*&T+ld{b4LusWNRUXKC}hW;#0Q;Zam+uG&M`PA49_5;pmW+!(!Wzc$;bo+n)+
zY#TFETZAOWm+@NlH$3!W??n08KZ1M${FOkNjV~h>9xgMfUaR%%zxVLCJQ3jerMmNx
z#Hf=MX~iFBGQn^#frvW(ez*7$rh({~Sam8w61MoM)lj6}gZS6@>)(5iKav2V<K0_t
zi^!W{1Du(%2P1gwU}dmTHQ4XP^@Y`?Ku9BUl<;YHr@9fl&kBUL21v#OiUIHgQ2rVV
z&Q2D4Pf#_i(dK^j2Vug42eA-%=y`1q?cE1t)VZN>j~~9J*;%VZBJe!Y%}D!buiac{
zx$u-aW8HYP*TXdPK`H{=o^Lk0(#7lY)_D&uZNQQ-1-y#1wRzaAg{(#uo-aPhZ7a4y
z&A16s79blb*U2jEKj;>cvL5%VR|2z=J9{s$FP^5Sd2w2Aa4vS)cv~%X2fI0zCtkm}
zk*A6EIelK!YKT_b>>mRuX=n92yx$M)=iVEQZ;1Axqvq=2dS2Ypg~BL~nkR8*^xNDH
zR%3KMR9A2Bo+m+CiQnsBr`9aVW>45aQT3#yGLtgG_6~6m!oqG|u_w_%6TqFw6ue2|
zYe|B{uhURve?hdYM}iv+U9J5MuN}fbQ&|(%5%2!th6-oY#E1~A`jA<#%g*ReW&v`R
zyEysl_}OLMuK4Aa@ALM^v#C#OjXqBVw(sFWjH|4>JzlzKcXy9D2h|m!2a%BX>G}a7
z+O230GU(e^5cuberzm>o7%0CTeXuh?_O2z>iY$HpiZS4)Mr2b)&WhK_2@w%SAcUgI
zIlVYC{}m3sq$SO({Z>EQ$Q5F2RFvA~U<S|G8d~ur!cqxhz$B65Mxo-A*4^##j#&rA
zk|0N&!ch9-gI!@0-Ya?^p8m@5lu)7_I67?pax|MQ<i2~if_Qu-26RQJ7uEKBJcK*;
zyxfexul;4o+35Cm@R9Z|Lqv0{80?5sQS&91Gj}$S!I%s6N}L_u-0TmGArBndgmwKY
zL=*IE8n6p@y6t%#g8hP`;OtBJ8)39r`7&^Tkqf)^T-g0%WH4=uVz>zLbEEaL$KU^P
z!WyfzmW&J+Ha4eU6L_a~Ep1t9cAGY6v)QW|w4BErRK6Jw&T0ea3L&CPwo5?YXf>Z+
zXqUS*|KG}|s?V3$fO=wLYLfu$bPu|v31dhuWi}IX7Cj!71|N2|m>G~bF@7QRjm?)<
z?C0Prr`si=Zs}Ay=%7Z>XGtU~{YGx)7sj+G7IU~GS)l<?mMy{Ue_Whtv1Idm-Ocp`
z%f{|F?TpwLKeHMF?-y}fJ%0CZ*YAS7pe0yWSII*=F^J%*F)r_)xa%s$VMGP?Fbe9p
zJ=d_P7X27-gApQ$4mLp%3IN}R|Di^2(vRxz@2QHW{(6->;B23xFZT#mNU-XEwfN_i
zwdZ;3GY~+GyZbmCzxX*?aK~r)k(eU+{T8%z^qQR>k`J9hJ6Brun%C1h=;IRa;S;gG
z)_sPnOtKn-rP6#|;Kp5DT_D6;n_t&fj3j;NHNLZ{hX|=zn2LZc5H+tL5o$4WjU+!-
zK;uIfg&h4?N)#>_|K{`Y_jAT6{jPM^{5n*SA>fg(K1E++W0k|v=y*NxL-&!0&-Z*$
zd+gc)HA%2-C~zMoV!N`{K~2<+7)M_c(&uOs^66Z!-TCNa?+A3%d)@Q6-0TXmxU9xp
zOP#f1iny((F3PCSEA(9E!oVgWK;pCD)he6Yg<uwdsKG-+D#{;0>Vgf?vj7uTNDux^
z3-Il!Mvzy7ye0fpbq(Hvc0U*1WZ7>H>#pYuIY*cAyYr!-lP#~Ia7Um?`wuyv#zSW9
z$ji5$YRSo?>2tdZ>kJ;?AQjTmfg&%Qd8bIWwKiAK-{|ilIT4p#fWb${!L4Nzuf{%g
zspS!Fn$}<L7L5ZMs9XD(#c}EM(@akj03DJYPGsqc=#Em{R~N5{O289Fk^I+$W2QU-
z`SM=C8=}K+f?Z}?^7-+V;Xg4K8?G;~VJ+wDuG?eqaJOOb3K@<dedP@21>kjt@Y{Qt
z?eTklD>sFGJYVi|z}$zERn8XA9Q`xv2)uo))oyjY+^i&pA3Y#Fv?SBO)<xO_r0SrJ
z2tX4xbTombR2SMpj=J@A!2xUc!I+7~?F-nwmT`nDqF*m!Qkzp?W~wd=J4uWkZ%nJ&
zh^@K$UEAyZ{B|AO$y=*0;^+C83fcmv{^g3;{)31f)akAQW}0RW!z$o&gx|rIx;()>
zu9WixyzaI)K`k~EWDH*awHZM8RV}19-|}Y2nfliOwkG%9>FH>i`Di#OS#LeTU~?)q
zS5zshmSNnWF*`T3zAp##bmJ?f^vO_igV%{Ef$OyPDnac}n6f)Vv4M$euQ9Swh#({R
z>uSw*QA1WP)odZ*dZZHxd0x#GNQR@57VdA)hP(x*-LdM|TMRDgzXEFMp3jHfqpUqI
zhqPwwpdKQq{a1wO^LNUg`3pQ1hL-e+ih36z&WF$LslvIhEX(?UhM3!zM8y~k4cCjD
zh~QGH;72Hzb%?^m-;FM}KjR<3jXQ150y%keg#w)Jo(j4Jd24lr>#RPq@%1-;m&|5!
zceXliW@mgAT+1XFQ3K2aggZU{Ew}ub1ZiYT$kqCd-cBGLQrv1%qOb{g4b%L2q<m?Y
zImg#i(BrPB^urh?{`(r<*(t)eOjV)gLH;<s<!>V|1UdqBG0FWyYd_eww|N;1#A3*4
z#L*hITk=6ZFr0y@q2`uQ<iJ$*2nSryhX-<E^W|jbP7t}fE8txMPG|mnicI&7EQ-9g
z_xU8;ajW|sq2mqrvHHT6Q3ZJ*+K(V=p)e1kcOV*Xe}GJ@%2j9{oN>$8Wjw+d4l+Ps
zb+DxivJlxxsqO~-#aKP6PYrpfKVjY(9uL=GWpSg`<o4rt+qYKJ-<g5Zh=<cTzd22H
z9+Ss&a*_lDR)MwE%q#-J(-DwY^=CW8VbcUjS*Oj&!0>Y2>-H6w*`UJ?3}asyHa1u2
zzhf)2BqPrKbhbcyx4DF$lC-=s0ve_3cSQv{6tgISgzu_-97X<8V=FIN*eVG-a(n;s
zSP<OL+iG$@$kndPwV{9gfl_pzSi~0|G{;F?`F_|%3k?0Wn9Y<1kIE9XoYD_53GK$4
zdEWoinb)V~e}6i8q{4DxC|KJ~!JSU_pan7f={~);!<-sJtgw=!2PcE+bb<EUm{4VI
z+{PLnTgCnr>Hfg=`v}($FKj*Loi<NZ(>?rOd$hB}%T1)K43d+^Lp9aa-QA`O#WUmk
z#qwT}Vn9UX7dVu;`eQpsY1kmJ1E%jn0UX8pu(5?g9|I>vu<rKSq7mzZZt3&P<4P*g
zDEE(U#=4E)ztwRrtw&32tG)}Q>+U+QjG2OSGqM%<pB6#{9+oq&^(_XU2nzc`3>NNv
z<#0iaQZRC9BjJRLb?k%qw2|Tfm;dbpVVDun0|^Y`Vw+QkJ|EoEO@gFASf`oER#HEC
zI+B$aToqPQyv!}N3>=93ewA-R#37NDZn62EMDxypyZu;nB#VJGaC8^Vbi-fSU0pQt
zVbg{ehq+2*N>Y~|?7FPyrio>}b6is$*L&^ZhFvd^54@nM$3A(L?XngZ_-YF2c!lX{
zSO9i@W=>Gx2k-UbG;>wJ<cf}0kh0fK@9XFy|2v<kfXnIby$@akQou&@<l*$ORTWd;
zpnsqIHe6<NW3w^57UkEUsMoMAJDtV|G+hVL47(o|!;(&-2!{vY3#>o+^^mVgXMecg
z-XJ4UALgn5=4cS(knjIY$U*+It5;~CC<m}aB23Oqk?|dw{6?%G5VU0ZUT-`ah^adA
z_Tos_A=yF_>Y;^Fqsvy#AUrg`iP3?<QJBZU9fyl;L_&TyZamzQM42|&0^7Y;ms8yK
z8_OZA(Bv?vDV@xtC6xSyhJ7N4HiOAnGeB2PU0mPtvhT1zRG9d_E|Gylgt@e>!2vLl
z{oS_Ewb;fc>fwl&3ten4?k~b_nwSf*ep4V0_LIfYDo@1mIa8?<C6^BXxp1<N><+hg
z*BfVP(>WXk6tp5d1h<z@o?dMU=<v-|W=%$QC_I=um2f}H_<9iyg~xTpTF$_70>}H2
zVic#U?auj+Z`ypkU!R^~VM4c!+kkB{w62<5yu{%7ePav~5*(Od*Z#JW<D)ao++^L&
z<&{RK`;raW8NJTOZ1uOvBQ4bG%RG%7^iwm`3%Uc9uU!xk3{ATM@?*$ayYbddR4txk
zNs5v-1QItA*36i(j&9$~Z`%s*tN<-ROk$H`-6SL4fqI-tZoc|Df6$Ha7AWPG+&r+j
zJSPr9aXol?xW@LPr(1SsYu&l=18MA>H?ADN)iK2O0{u7B#Wq%lMQMuc-u5>SY+KLy
zT@EL5Xbro)ug}LR#n^IJk^A0nqUGMkj27AEW?`udmGySQ^p0nYBZ#KchYw#vhKh&6
zU_AvOYtB9LnQ7Y;;L5`30Gb_&26=tO&u~<cX#PfHrc_AuWXLSI;`n&Z29_bSDo%@N
z*Wxg%ol@fU;Jrph8pRWRo~Dbw@kx)|Zs{mm`}gKS!WThXjzLvBbCdl0$f(h12{8~G
z$NVycwxeCOl9Lj{Kpw&c3XMkLx(=R+tGUTqAb*e27nE!hyM<W>Y`Y9d_~!TTtMq~r
ze-d;Ys@OA3M%<ZXS*gK=trj-Pk;*bcYAWVp<E12KB@p4l`y6`N=`wQC*b$mFbQd*L
z>iDpw-VaR6*7m$U5MoTa-ww3|<{mC2=WoZe85PDw#vq`Me1Z0I_e@Jz{B9#z(^Mfc
zFG9<+aAT(!y8NmFU<GYy)EVf5-Kmiui^97=x=>}wAc#YtQ0B@gnku9#Do)v3!TI2v
zAi)fd;b{L}uQ;@RXXO1S;=Qw!#0v~6li0oAQxf7T*Nh{01X4z0u|AA1@)Zs($jX{=
zgE$cpM={WZhBD};<)^9zNy7)|AUxn~de_UpP|h`*8_Zx}<6Lfyj7LW9gd&`aG*g|!
zSm&#@4ql#g5-EM9>B#L*>D8*4$XgGtzX^o&<p?Y8azY5R#*8^R&}!<C2;R_=K6y_A
zBPWy2&{vy39Y4~7ZiISWjO-o1$V1SWM@K0s%B#f2p^QpVVW}r$SSx*ZgZ?S+nn=O+
zca-*bYA-DW(23fmY8nEigprtzhXUeXqzGVmwxiHoB@AjLK!CfW@@wPcKf$O9&{yK=
zbizQw>s>AsT!hbTc)o>|`SnO^eT>X2{Va)|jB?u9UTQz5r8hv5l9888GoEni`v&Xn
zJfQ8mlZw`LqphPZF+H|PNzv^#u$Y<|l<ef_w#U8&(6L4ADZ;L%DR6y}?M<I*fH%ii
z6FKggj7B?$v#%J7dKlI{Dh~~I9z!vWHC13laBBe+ymYtbTHc=qfLGLt71z)68qLOl
zPOL5~JOwfn3d!KQpM5?*xs%O?LNwU-Z91EFnYy4SM?z%K?7zsrbrg&P)vpP~|3-~A
z;;7|-ogsODo;yc1_P~fQXcMF~az{==aNlU>gxloPmoj=y23gv!9Y2-nbK9;ZzOp{h
zXUi>*ny!aN@+0bD?%&1o`OcBeF_{h<6XD|wq7_<(y{YPb)0XsN>5!QL<-CK~l@<A@
zz`u|hi_^!)gB~f~?bkx)2#Nu^HT9k`8Yw9NT_H<CvMr6rI*~<3a4o@M+nHv!1)Q({
z@lol0`9}(T!Xu=CbR_?a<5Y9bOoc>92<Y8#-6%Uj9JRQVBrJ)4?6i56qRuT>tAzTB
zA{BIT&*A>G!%3N>HQrS|s@yN7bl2EwZ<-SKgHX2m5`BANT(yqVs7jKtSdg78<e>l0
z-F+2Qa|*ZYBdhqt8a8DhA$B3J<v#~I6lm}07csJO|7j9H&mM(DP7+<nS^Xb$<H>Cq
z@-UbeX@?fRO@*v$B*pbOutP!v`{2?hZaypMKIq4A`A%+nAC`sAwZn=St5z(xs)*`d
zjlu{E`ZK0QHOMAfPg#pI=z9Mk1qtvUS-h(6-LCGJ9ibaB!}R@PWl`f2WHpVAFfQ_4
z%@EHBHg`!c*y3~*vN(a|Dv&0c%b|!jK6D>`(@C8-beenIx%*mTay321LCBEbp`F!l
zUzQwp@Ug2KVU{lWJWmlo9j1@paS+2r+YaiQI&5il+rF8*_?MB9q0Nw}0JYe_$uT<A
z9%e#|QM5p@T&{nAg7XgBl)22`5fL#MxpHUjFG`YxX^oz$GeHHUkjP%_$xs9<9pO}m
zf3c`Sy_ea!i$jx^>sVlPt!>P{t}HT_6#Mm;e8sKV^uId4?<D=a{@v<mu={ekQ_c8F
z!;4iKRuC~mh#9pTV@jj=0yqh1yJ*%pp-pA1`!SR0{{?^6$6__3!Ya#E?{P9Ly!!dH
zii$bFV1(i;am!`<7*TZ^+g{0g$?x*;@{W7?p?v6CYiJ<Y<u5KPZ9^76BvJj5gf$Bh
z=r3rIJ=&+ko0q?n?M!8Y&H0BNWMQo6$?6r8tLn^=4$Z??RVD{+0O$f-0w9rUg1|1E
zZ5zYvGbMg~g@AnctL>0JAI~wAZTXn`nye<15N2gZ+GV@<?Pd{3!1aIpE$aXDa**W1
z%aX(2;u;YED-(tQml%kK?19NO)*X36dj17$J0KL_jrc0n1uj?IpL7^?kBpex21^=D
zVd8ivs96QyKTxR;C=Ff^njWEKXmwZk^rqWh_kFs_`Yd4J`-@VYwPlb?BsRkeWbS&A
z^-YGG+7NfNS0~AewV@4&B|uLw*q(ASpL?pVyY7Rl@XtKwLm&hZ<1jH4^h3}oC|<ck
zM6!c9S+qk9eTW`z3gkyszmoPC)GB#^euD)dq;98VygGn|j-^{y--&sS(<xXt{BS5}
zJ3A#JyK}~fAkuqZ>dZHlnbEzAdCd>s|8J?Ij$t?(8FL@=OLAiVQRv*9#H#ZAFU}<r
z7_^DWO^>{fm&?5~+Q*t4Q7@-inRssc@n}Q}=MMsfg?v0zg`hV+xoSM8Mu=xuP&})`
zHZd(Ni57r%<$^77ee^4*g{BWjwJiCJ+cAXp`Vwsh|MD-tn3JE+{O&wheX^8!v*VGi
zaIeRb=HsUS{`S;`LJy5WJw5Hc6eG^6=$jL?`4CiFL|ScbXpxqaV!WVS0){1Lb0iUe
zqNJDviGT<%ZCFYMxte^S1uK09XILtM^b-^}vman8Np(1VY6KTnh{FP2|6i3LwQPM5
zL)z&<0q)v`?Zj!I!{J)fwKd!8WmofZ4K?rM?X?$`m2|Vq>87tHhH+J=R0G?5B>W|i
zEh;Ri7)(@XKc94RT&aA1PbZK{&_oMTn}5MLu*@ueb~m3om-_jAETI8rQt2A@tovGU
z2Iy+am2~nxxI5_Vi1jdRzkQ;%-9@d+(#<0Nrzk`>^fNWhZ*i821R0YZK2%FW{_nyH
zE5<f+zS6`4VHj~eyd1+xCc6+sclMlvOfoJ#b%61*Ttmg4j7tHTBAk#Qcd@QQcQ+6|
zqcFKX&WeEk*>n)8BIDy{4b43?Ch6l>1^i<vyv)#RwVE|2EA{=jt?m7MzI<==(NZQ|
zZ#MQ?n<{yU3YwBCo++uO>Q}*$CBqa6YhmlNg-+k^bBYNroUd2>Mh?S%F?CG&b&d9A
zo|Z%=w`>kW{!B0tM^F~zC;4BFSE4dSUg`*z3UDr}!Fu)w35lk8hHySK@&49K1hxrz
zqlpd1YQ(QsuKOLnbs$t>Z0|u8qjM2=BQ2Ukt)#DJ_WsV6P`gTCZ>W>B6;!W-`t+EP
z=iwh1oma$kC>Is`j>D#h05{u@5DS)(Lr)9~%2kg7m$~a5PHwX$v0l|mn6$}<mQ>QP
z9V>VkmsB{_Hdpi#Fq*r5RV8)B0Q7;RRd_VN|5YNg{9H3uR#VT{@5&qN_u<NJhVA#;
z+Ly8xyT<O^<dj1OSy+%3ujNBU#A+e6l<MWAGb~h<fntrtZ^hAUwC><U;j0c|vS<$|
z0~!(}>Ck;d3Q`Jzid{MF1w3kx@C#kayS<o?$W4n=FsO~o=3vw)x325rK-qFAu9Bh8
z`3JxkF_b6TVM&?rRDQ2X-EIvJs!A<|*>D7%GiB0Y31+sd3<U10$IrGkbfo4clv6`w
zBT4kYFozX{k`fA*W+&Kch)b>=?pd-J^xhsn6Uqs-$6vPBIsTcNXc7~|`yMeM^lnMx
z5y6^Ga&Tr;$TBRLC@H4{NTYHU_^qE;N4sepP}l98@foP8_L$Chn6aZ3MB9d(v;-L2
zqX;C+^%1W?pjhsuX)Lgtx50X|DK=vdKaZ%l{E#{vmM#+r;Ap^1<9Sf7$!vYzA7sS<
zc8)idh;Ji@v~U7H%HuG-EE7u%uU?ZRW#W4aGrsZ}T_Ih4PoygZOpT&#R>jAc$)H_4
zLrUfP7dcxzPj@dsk%jh9zOc`?`;);nP^S4_=ybEu#9O+d7g;Uh3*7BHn@fB&OhiMp
zoFkzybU(Ws=4wFMSx&({b(G%^Bz?TmV4?nLJ9>v*I*bm4jFN)NF&!N0&HkB@1)}Z0
z#Uk*|{*BiO>KTdvu^W?NK%CjRz#(43-#X8umC>rmUca5?O(n*Np}#S#Q;Ja_IAs%8
z07yIEkK{%3nnF}Xh-bQUX0j$HXD%%(tobF#Rup82)=N1o!$Klrj+hs6b9m4>tKS12
zFGz6>_<N%+56eEzp6e{>Wggqgv+It;zGV<k3q_14uYv)Db92eu_li|$L#ylSQ&^Zu
zBhHll=r}v@B^)@<Cb+xcPr*3z*Ze`*O~S#(u${QnQdXvW(qh72N=CQ{V1jMQtuW$3
zDTpX7A=F1_zc*hFB-LW{I^H4_r{&}&fOU)w5mSXj8JF?v8~Q*Vn*Z7_JwZ_Mb6RT`
zhZ7>uYJ31x%I)-<NvU8+M1l8zx1xZ*@88$mlRa1u6??-44Wdi)nRSbIsa^C1$IaqG
zC{bNvdga9k#xk2z0=$Cp>LkSKCa4Y-Dr0EM?7Gj0ysVszny6?t-VXb_$RWYbw)Vlz
z#jP0ex0oijzjW>-QDK2;rEa-sD})E|1w1i!dT#(mRB4mFfbRkRm)*S09#6Zi)D(M8
zLh&~SmSiHriJypU)e+?cxiK1E9}*NWqAEdbTFK-c9pK}spFw?-Y)YN^cbWWWC>UO_
z70*5t*KprIuQsdqba9pGBd*!ZU5@-s=sRIe4yrQG*(@db&DiN!=$T=;32uy0V17~m
z!0P}O>eLTWked%@z}Yjd$#WlQbZ#i5d}xY_3mk1Ows-%dcj4eCp&edWi6;kjnU0O;
zMZr|f$x%?wVCL2Z)rTfj%R2e`Z&YCms;qlnoA1V5t~DTFpmOyZv04_NFwuwTozHc}
z9X7=dUV;&6tjH0hMGiQahzWS~l3n$e?g40-!p{9Arc#@ELcsgO;1M7Kc;4RNAJdR*
zEH1_49#}x33iZkRGhCOFyEEA{#60q79ZKc~1vDqBLI2Yhqgpkk3?}C(Ts=G_{~(cj
zHr;^=RjM=<({s@D$JphGaK0gDG{Urn<-O{ZQ=DW`l=j8d7T3@X<PSmkrZLDV4)`jV
zs%lN&4HI^DTy~i#z2*EIeRR|hrDA9V|0jFuDi?V*@R!{1oU0if-pQXFCaeHo>J*%n
zGRpm{XkJY}qMrY$pzNWNLDR(|B*kUzX4mV}ktkRA>1hG-?V0O94s!ex9_aC5LIdf9
zWqD^Rr{Nkjaf`!^&IfLEEXMlT0cM0iyO5QP$zt~8C#1CGmSz=2hx@yZfnp=euiVZk
zW(&o(CI{Of&J?Jft;b(aCA_eP^TaZ?Egs@)d=;+>hfps7VT<h&QJK*JMT@<pWZ35L
zSU-vC-76<lXRF0Zg<+VOg7@zTB2nlxio>Mp7ZLQ&W7{jf=p`YUR#qc(0LEUBS1z!+
z9eKk9AuILY=|8WRY$Dx{+vkn5BVp8kdcD3z8`-N<s<OiH`(V#W18@;3fEO^@G$2Q7
zkG_EHyMJYzjd4l`+dOQIzUdA#Ve<WNrrIG!-v4I_o(<_;GfyTh6N98M&aGf7Z9&Qa
zJ{EE&x{@n_g1RK2KfO52PxqTmoZR1_BKeF^Z$W^k$N9Or5vO*|tL$}Tuho#c$93&A
zw2`3lCuhIjSgIA?{1hrU<IAU*#T;w!JeI1Bkxol5nUka4F23PvgUOb3apf(@9qIiL
zcUu8jf(Glo-rs(B*LCxx)_-ISZ3XZ(SoaTQJJAZ$gKUG0#GYy`PX%<mFB!Izh>d&5
z2q+Ht!ZJqc*Za>g2Peig7B(hQyfPh8-}yYDWN6H>HI*2`@$;K|II3p)65%uWi_}<c
z31u|*5^Y8vdely+F+l~+Rrl?wd)a)bo0Ho-;8Qa1TFRNXxkgBonFXoksU&}xDrg>z
z3nJ*C$Qe*Y<9CY>;%5q2-v@RZkiAuF2qp+x0<AS0m)!LIpDP}x1=GKOzF$v+)}@01
zzlY6e!{H4TK&X_^Ki3CdJ#V^}+M-bs%7;H;ipG#zN71UC-gIA`4))^&J4OGGUJfp)
zkz*N^P-<}QQW3JDq6^OgvCM<2!cBwv7(x02DqzZDT%M(`L!u}z-riByVQny)I`jK)
zJXc>QYTl;>{&wFO534Xc1Kbl4h!DCh^ixreA8i41B!Z3gO#vphCAdU`?FZVwds>z?
zqtQCGrgf~bu$e8VV}=EB@yO4smy_VEwAmb?rh0o9wd~ek1m3oVxVI{AL)B=L7E8D^
zFn~Y?8@>$Ju+UKWlFG#^-VS9uANTri9W4VVB*N8M^b4{kHxhsH`AGI1akvkY!rB<P
zpKn|xXzo^e`zXkc)AG1Yb7djx*Ixs|T@jkOYBznKFk|e*Dg9=mOU8esn1}M`hX)~<
z_NOUAdF3D`^30nk!r@Ma{=#FkquQ99K?Zg~Po59cwD;Y7IPKUR*Ph?|o%^~kNwC6$
zHIUc;W^Pln(bko6V>%gnEXaLGh1#E8>TC!U=L8BPzJf0j+#O?y&&;G4qjahi#$CR?
znBi)27<tN;qy^Y!mxpnYCcZ$Et2Gr$YTrHHUrUhD&nZB*@*{+?%%~rz1(Ok40$RKt
zsH~2pqMzwiH(@K2JU?>Vo$i-{X<0#OY1EK<)JwfqD63H9AAeE~j`LwZ<#1Se<?C>g
zw2sqQBrW{;!sDYe3z~ldSVJm$E&WJM*@TlId`F?F$sSRv+De1(Xk=D*u?tDUQ|m10
zdKa7o`j<>bPKlfttn#9ZBaFCWK(*p$BZh=uHdBUKazb%!Ru6u4f-(diX#3b0wHyMS
zELdickA)JoB@`?&v^j~g(Xv>`#(vK)sOEUOMoaR{8{YjeeWpseXUg^+kPLbQS358|
zBF^~6VZGF#q@N!d%;w=py2a<?G@2>RS;DK5waV9SY~pAAO|Pyqrl5ifm@`M$O&$3j
zAA}z8@pN-4F+58zv&ceH5c!NI>bwHOKIX`Sk^yQwK8@fm&T4lb-4QFWC(t=>3U&Kn
zQw1)~<mqK&8P&3e%O){(Zi0V)Zz>*F`WXiYR?exx9B6`94PB$`q8j=3HeV`Fam4&j
zyuLp-IZTL}d=m8lb84s=xK~6A<>L}yUmu;9(2~#g;3BNc3b2asW_JKPQpJm@GONpV
zZQ_}4zD81E&f&7<A1?BG4wQSl=Y(ykws9mk5g3Yo`LDVP`xTiUUz<#<vJAopF~mpl
zrJnUU21eTDi*;saXbF}|(9?J{va+fNm7u*vx*B`4y`<%1cdHygi;l6iCRy>pY+F74
z^Bz<uJS=oVtBNFb`iGvyN<#r!r${tXvAMKtt*(9-Y-sjNSUAlnhu#zF5O&Getb(w#
zztqT?5htuR$?L75>324_`A0YtX}A_n@Sh2h+iO`AV1KevBIDaL#$DYtgz+=-Uhw-K
zo_1Sae)`-Vubu$P@sd(H4}FC)LD3GWs~j!}3Jf*`Tzj}f9+AK0(MzWMSg+YgwEcXJ
zrC?~R1(L<5N%%qzxB?9lRaO6XxSbUsqLN+GUFR7JcR8cH5#uOUk3~4O&czQ^-~UC6
z3w+bte~q0nc9I$@vq^WBh#QWCt?(1Y?Kiuh!mnddOTz{Zh+hS(aoIDWOwmKLi!0Cv
z1u2@#O5(>7+e@%!uh{c5?}t<4zUZut(DJmFH21)D*)$b<6*tSl;G}8<4&|-(wwpSb
z-Xg^lfA4W&=3AHcT05O!0IPtZdS(&}`8_{eO)9EK<O_-evp**tI)*%5+aa5%n$xDG
zC+0Tms@efGo_ZGo?$7AO@01E}lc4aUhrMB=6%9dhMp+bjk<GW@8`5#10WuK#o}+JR
zVR%U+Et&Z<7mOcNMhe6XN=`Z>bW^unJRE*rJsopG9qd4Gx9F(g7+5#;-%H?iL7A;^
z(C4nC5nTB5J@ArOSzhVO?AX-x^sSbd*5mOs;0TZy2)8@#HLPqKb)t<g*Jg5-)3NvU
ziy6pcGcQ-GE(mwOLj<vF6gys3>cAFj&1%O@alklbgfShvLx$m@SVe$tDuW+i|N6vl
zGe>Os`#V1_Tz3<Kwp;-b#*s8*v8HKH^}u;ONEP9)=#UGJC9=YyOj}4XnzDs?cjlTZ
zg}@Rr^F_7NeKXj{-&v7lt7paRV2s<yN0+*v<sDE*Y!j~oQz-R6hM?O<;6P+1C>or_
z=}e7O^n*b*b#&<RlV!WzO<1+3#=j(8Z+CAqp+%ib%(?vV?U*x;h7G<v66nkLbQR*T
zJI9B_cxc;~ah{{equ&7)q1vEGUAK>+PrxJBQ%5<{@Y~vgFjsqB%P;%;2g!}r+(dy<
zGRJk-y(w_2m#j3oY<mHhAivw0p?`iJhv+p1xX-wl&!aANQ>-$zykp)$&B1*={wOUt
zQh7Bsi>UA!sWCHO5bo>dDb@aT)lJOvLKm}C5K(KyoOca=MX;T4`Dz*z7;Jomp9*NO
z6<Cwt@cZ39y{rVU{bjQI)wP;g!u>bJYy`SPn~?0D9QBheUsqN(Bw$a`uvUnuPwPsB
zE<3^qPBgx>AEhdhvC7Qqk~B@wn85zzJ`(E%xGj0?|7F^_)P%{cq!7~DUEjg=Y2}T4
zPwnk$l%XXJzEq{NDlnx$2)k<L>I7V0VaDC^L46w)|1#}~MvNzKIB3(rm!8rao>Siu
z-N!c}AtSG!mOPu3h?er3Ou)7`LTVoq8=Hh^8)Q&@6|weSn_%t~h|BJ9x|k_C@?gzb
z?lm`GbE`YAltGa58dBPYrz0x#y+eppVl89(&q77G{^3!_wWyx>rqh;-1l%Ae#~UJE
zQ)UqC=jSoHJO6NPy=vkZN})1zAr%GVe!HK<ZSCyjVa?&M=c8VjT~XK44oaI6RDSy#
zuyOf19x=iUdxt>{qQX1I1Vb>xb86N)sfNYQ^oVRmj32JYX@Y(5vha-TpFPI06A6LU
zP`B@hxqw-S4E_<M<6r>!pCau1+{{I2(Pw>r7;b^AfiLed|0p|onew*uU|W7e^JiGg
zvlo2iS}~aoTp=k<(dWV0SY^N)_pnHbmeJ6%^vxSii{Ds6nsAW&(`g11;u$hjo1S5&
z8S*wlPGzuy61u0iqmN|aSGy}owzTfG8C!B1cmDab<g^3OUZ19XK+bW`_lwJyaXAEO
zJv|8s81e<blfC7c+DluaRepP`qoAPRU~$|`f?xIDG?y3Qa0BqJ5H4C1qbdl%Hk6p{
z@M3F3`qd&cpbxm^C|w30{K8d|5_|qSZuYn}*uJh(yJp&qrtO;t1;wHk4rLLZ+3I${
zQ15GUPaFPOQrFGK0Zz`W79RUd-e(Y2OWgm7qlu<giX%_Sr|`NVC|9Ppc;4np(#{(8
zel|UF(mOuhT@U^YZ(+(Y$RRuno`Rx+=9h^F`<gmLX~C!b?(uGx=tcz=>_Bcz0}Bl+
zl%H=@v7kkcB3VljA<G@;sqvd#8v=GjWmhs}01qDS=a+E?FyP<(99pugxM122cCH?q
z@o5*fE$<~*<sstVI)wcdN*~@@Jw96-C2fCi-?rZHU}`xod_scbvlVYyOHE~4!&cyZ
z;1GUAYNN7LzB1(z*-irf+O(Zu5v+`uP_aH`dcEl8^r!}$WVpFA&amtKb`O-5$%S0}
zci!^@ktU4_7T_l<Y^7r*EZp+Vc--N;-^%hz{1wBc*w&UlgiFM=^(&8`Zde30BI=96
z^w>Bn194pY<<pjtFZeynb8uu3hS*K|3k?}lRV`#RBW3CK;!QasFNuKn%{eH27|Ht<
z?BfAy5qdxGdhKh~=l?)p+5?~Yg_<OwWKUek@B>MK1+cc$-TF~r`yMnNdfmelK|A}m
z*O$7EZ%zthSGJKmB9IjFhwau-LkaboN_Nq@Fqp|R#*P2&{!Hu={7S;V5;iqj7dCf3
z>h~-2-EdT<>msW|p79N;*f>_&8>7GU@MuE%RLnyk<~t=0c1P$Kd5IZXdh7-M3a*;&
zO|2k69l(40xyO~%iUl1nAxXj_4&M~((jZ+@u1?|0@D1FW`<92K6Vc{Uu7tF!GQQ`M
ze4m@Y+ZaM^^<_EBY7EZEAlg>S{UW<L2RB!5xzohkE3x*Vb?E%N8_7qW*mgb!!V=+X
z^mcUSift#$7O2*F-Kga#N+BIUqFOD|;5EOP2uK0CO5bxqgA1Ay3BzucOa1!>*FJ)I
z^3P<AVWJHbiG3;58xR?7Xn1FBbXV0g&=1Y%W?%J1Gs<4~=U_-&kI`Lih6_3sw~_0L
z91Vp51U*E7RZEpD4QboTIr{<<D@(<*W!IXcUfI%CmNMmgZep{7Eb^G0Vz2V_;+`*N
zz(=z5n#u-%6wla4=Ii@2fcG!UkLn!VG~yi-I!1a1sPklWL&bABVw|OQRJz-zC1ds1
z@kcP&>q4!CIZkLs3T!vGAb-e)=3n2k{qvXd40SX`fBh2RotafOG%(taU|nIZ>Jj5(
zW7<j$(Ytk}TM5B#s^a&KJF&g4I3lqg1sbfmJtx1V9a2GcYE!C{3X8aasb44Bk2-Yq
ztTx+tTW5!(C+>@{w;NNzV^h&B%X9En3k6YG!-&Q~#-d|+O8=Q)E%)^v&HX2<k{d`q
z9zF|n1MheJ%vTqy4Q=H@{P6{dy|_k9^<!9^_M+LIRTD^2$a1alc$ZFBh-4C#Kmq^v
z)V@v`4Yc6WHfb}Xhtpk5ERv-3RDD&QjI4mR)B(F?<an$Ms(;x%(6}6^0Uq#$M}v2=
zv&=mv4<neUAMaVo*S^>1AaiMB6!l-|s2`3|&BET8SHL#k-GC}@olDR{jrC60yX9r>
z^6F+h;ZD^k)^X4p3UgFwsI*)RpiGrZ);Ako41|#-K*l6sU5$)|^3@&ITy!np5|!;-
ziu%6XGB=zGueLXuTG%@@rTv1zv<dHQ>4_X2P}#!9J5jW~if6;`8}HkY(=ftm`9`LX
z%HgI?r4pOvi+h}<BiP~n3>_*vN(;J7_^f8x&__6%=7U{e>?*S;e=oHL(gV%&ODn5?
z71UIi5T~pCZMnD`1U_HxKBgxvg;6i{d!1g!K8QixrqI1^K1TSrJ*B1uT$+uqXfa|x
zeg|o;q2#W0HZ(JEM-;^1Mp2a^MuDsLs|JaOf(Atp!R@?>g4fb4gel%~-A$iEBq|W#
z+Kc$!;}5KMe?FbeZsrOCX)3|@OY32Y$wro@=3;uz(S)2gs1}WzDxlYSKVj6HHr_2H
zT7sbj45S&DkPVVc+hR*Es1fna^u@DrH8%Wka(;Y3M7VWrlr@14!t?n|(GDTAErn$S
zmZxkfur>T3DO=MEN6b4rNIzeX!g&LzdcosfL9bsfXEg}Pi3}{JR)O$E^U-jCi?X);
z?{#Y1R?_diU{nQFiy6h^=zC?nIM>f?DLhd9ME0sv;!&c(c@yB!#g0s>YRF8)LPf4N
zI)d&Xa~i{IDbVtuGW>5R>gDEgz0v7Ym6fBeuoqr#L%~sG#`lsLlCPG=U~xj)K9F-M
z^bXXNjHpjAWB?vPmiB6VDt11e9a!wHkv2I3L$1Pqi2E0@wuVNJhDHwVId$l-WSOw?
zadNf@80FN+9|ZuLeFz==c6>BEr`Bcf-0kI2D{J0*_${(mB5N*hQGE%OApOAB^b+2<
zskp1SJvhdV0<EH5X1mczg%V;G!-Ax~Vsd|QBJqYh@ZBET>u81joEVSi8&1hcn75TQ
z`*H@+_xy=yB{2o*t%Y$~rtU81{uj^+zU%S(!mrr_X4(@od8jOH5?FHj?hXGa|3(Vw
zy`Pb3A;y+SCjG9t7$KvrwS~qNZFe4Th5#Ald6a}?rQV52;{>|f9Os>$*;ohQBp|sA
z1=r*Ny5D9+=-S;B))pI*jbQ!wNm6n0Eb1omcjrOlQwls?)<z`L@;+Y6)A%-6J(7C3
zE*i6FmB6ks9cjP#DM%Yn!7*jSy3;BBww$)r<>(e*ok=UG?bDc*U^F8Bstb1O<?Z$Q
z^x{Qnl@@dGnM<S8^NT;^plOPzYcMJ5)RJWlaDUnjZXMIa!uY&9T!GXTIOM@d`5MvL
zoLfj&2i051W0Y1blJVKYDn1)0XMhiFO`hanVQd}Rav!Ta!*UGKF&DeLkZeKp9u;47
z{~bFeVkVd?WcMq_qXM11PKT+st9D{sD1zuMp)sHE)xZ9JHkwlJkp|+?TG7wP!Aj8j
zAF}PSrFHI@&71$OrIwEJ#PZdinkFdqtHj_sGgEVj&Qh~to|<kt5+4daO?ec_91Ki&
zNLZ+>jP$5f7&K(fEd0i3aMpO>FdG4684_M!ao6eL!w(-7gR|!V;ZUENoaM%n>T(8J
z^2H?MVix9r9b7{IN%;-#knQoc2il<n7$Uov0AUBs3eBeRkbHb#VP(ZyJfEMSp+cX0
z1UUOrrQ~D3i3Y)dVTTk4wH6}$<`h#M6k!4;7B-F^S+aEScJisGIMxan74X0YAXshR
z7F(KMmAAaFFMY1?gy)-G2cx82bqHvZ4@+cVA!xoRhxo3EufU<9Zh;D!8Js#m2{(wu
zP~HPVj6!VdV@y3$b`wVbR8b5%P_g*t?IJ|C@k()$*ROQY7c_y_THI0YYUHbYu;|6u
zN-W^<%74O!JozM>hT^C%P{c#3HAVR<L6I(oQdJ-JgQABvqLt-jlSUdMBfjTX3NxQ_
zfM&wapN*WJk+4wV(!(_Q)1@guBnxLJp9~pBccMTN24mq-g2r0U+3wk-56+)bPyC4c
z+nD3DhbMYhEjvNCkAIbbHK8-r4=4dRwc=j>Prk#?-PAuGaLUY%It@k-cjZYoBWe4z
zvQGk;){GGxjwn3FUc_Ji_VLv*N}|Z5fM%p*(8EjpBljmW493A)&(e|s<xd$o8VjBO
z-6*{(8+cv4I8G$q4~D}Y3I3gRq*K20Xy%^(mb`6w)+%c$FO=Eq`$!o#v?RyL4Ngur
zT4{BRmmUwpy-kgD3S;z5>U9h@^u>pKpfC)B=V>sozTHSAf^Iz#riPt*n)sFF-<cId
zl+)X8WGQbVhRq1d2fH>l0IE^0wLL;zPksicyXb@J54meX<qc?55gXfSxJQJtx+1Ye
z)cN_dS8<b-x${ClZiX@|qSUuN1hUTPB7&B7`M4~1Q&efByEqzdqZCF}%5pUHzk$`z
zfE%C%rw)U|`mez=5(+}o;;Kl@_-v4Azu|VrK?gpZU^Y-};2nAcHVuVcdoMZDs=RYr
zDsLWCB+f?i&w7PFF@t{~d7iQL`NVUSvBLt|rkonBoo&GlUHk%j_5GwMcE|-}jLhUT
ziphU(!e)%MF?F=gPp!)39-1}Kj~jc@)0Vhg5h5mQ(m6a<Ui8B|9fkaFXV++{2z&f2
zOn=CKGYKIL0N0!cSU+K_(O4A{UQcZ*X*FVM=~$V^BjBCt7d5e@1!<eyFvRH$Yq$Ty
zNfh3|-^VZ`YYwl-Cj{m~-lh`Y^$FM|>U;x{!T<bDt67`>akF5rTJ!!?gBzkkl6|Jv
z8sNDN8DH30jP|(fkITk{sg?Y4$#UkU4PjP^^#{Fo1G*ik%M4D43+&0wt%+@6$xj3C
zva@$6E|tHt!=T_A#DHz?N{dej98i(aQ`89cOy7%~2(@ZnkRU?d|332MX=-~yfR(4a
z!sjtOq7SNz_PpL*h&uX+?b&=CeM^nR_7dBYWEf495Za3`AvqPL(!8tIgDm_V6&3aK
zi-N-B#6%;o8?;_UcI}Br{tGHzj91j`^QYcU=GlacST(pW*EB3+3?ynr%<fJSuM?aC
zd1mgtpn!Pa-O5_8X!oVx_1%2;+TS>PLy=(5ZQ&^-t`zeRm1>)=Uj-bW8<DtSWTojs
z>v|IEQJG;lL%(id@<<UeA&KpomHtC3VHd+i)|~+`Qk>x-g|94^_1o!C<y7>HrSgI(
zxYXfonSS%5MSt2lazPAsD|HdD^Z1A(X0&VH+ozHC!OkVtoZ$6c$PGKXGw&Mbd~?V$
zKV$Yn>^i3_q-cp66=-SdgT3(<S@~qFxv&o?Dj%lvY!Cq(ljCOvLCLHxkI%$Ywr86t
z84N)GQxwgUu+ruqdWtTm>u?QZa_qqt=~j>L&DG5P<k3T|Za_MxQ&|+^bzR)4$BqJa
zq+<8fWzL4YctC`pXuKjc6?U_toxa~7ggC=eblWD|esWq;iZ~?kHl1}DL%}$);!zLH
zFO%<qb?Hzxs9bjm*GShx;yb_SN2BJ&qIiK39v1)klVPfsJE*o9QNQ;c(va%)dUrKj
zNoi$?i12FiUlr0sOk=y~Vnlu8b<S>&$dM6BijxzpHR<sxnXyqpwNx(nG?UnbK9-O9
z>34;v<*un<-za&Pw)(Buq@w{cG(+fA7Xf#0dGoKJ!<Wm;+aFMfr2FIQF?`b!)e$i4
z7Rw6j%t}JTQ{St&q-DocDf#Gp<g#*yN3uAMOYi`5GTR$~6v_eXq*3WWj*1d%RAv}Z
zP($IG+Rb=oKE4p7n69uo4e~5QhP0C+N@x&f91qkvf*rL?1I^`YazEO+{k~oKIN!{v
zFP2`xsQmNBnQcBE0~ecOrMb@$HwBR5F6*gp=N?3#>e~D28pROd&aw^~IR>KAzg!b|
zzU*&fX6Qii{*8^zX!c^?T+)fK;m4zYtMB{zg);4`YIUXZF{jgI@BeTiukV$)pl))W
zK$jzP@*OymunWNmPX~Y>Qp<C*mFEd1tp;NvB%~s=tI(0k@}Xvp1rKN^$TifVmEn^e
z^6RX}o4L8GV;nHXkgulxES5jXP-Sr+k$DGms^6Us8R7rJ`;XsyXXag2#gME$Z!=s=
z<)SoyzrQAR)kQ9v=v$C!q_?Vuy0ScSA8(ipP01n70|_|#_6cGGqlkN$)Gjz9jL24r
zg~fJ@u`ztaec{W}uPmTrX*w&N^uo9#8p+sSBv6z9D__ZActEhG%z1vvl=o8j+zZCy
z8tC@<xH=sE`X7`>A^C0yvR0;sq{qyUbFXrUv!DMHFce3X0L{vlNKZ4Ja$;e35=W4q
zg8s>73(1W?5fb3pVqJvI<3cW)STgHh&Mhe)TX!2dUU8ba4Dhkf-5H2?kjAqp)z&&%
z>$d;8-9PYQHN5Zfd))aAGV#3Lw-K%^4i(GCq94rh7$Tzw`-kJ`(mKkH7pbF^nK1<)
z`#{sd73x^YKI$9r1ivj)Q7<VeMH4|LMbS)ax8LqBg(e|V=pz*;$oHvOT+#*|`$TR@
z8c6?UVXJmI1S<S+=Xh}5ZyfVBAsx<_xNixflHTyW8<@6#`4_6}_Tu`RBLo_kHwDV!
zSH#~Z4+Rk3LMqxJ=<mHTt0R&{z3(EP9xX7Ri}D$4_4@kGzEE@0VwZ+=WZ25N)IU+f
zMFz{}My6!c_n<5%pdCR6aNSmxK<uOHsaX9s4a-XE2AIiS?wai1F`{T#zi<7pzI1U3
z;HxvVoP+c+?vOfEve|NZdk{>Y0Y2W241w;GcE4+>qn76XTpY^vDGlitH5w!2CCOas
z%dN+u;25UtTL^WAj>FhZa_2g2OFY|&zzFvH5KwaDdX}n`9?vlt7BiqUI4CEv{$JXf
z7>x+f>+q2HLl8BmTbYVvl50!A!}I9$;pBeUIzi-tEreJ(Qx`oPjtuNif92fF0*G#$
zQHi(mDrK3>J>@{B6n9m(lpr~2pd@UHXcmS7Q|#Tcsg{wVWN$03K|wq4N9r0t9KwU~
zckH8VM3$p>H0Zg#+ju0ks#yQ&lcwVCp#KLD>84@b+}adVi$AGeyp;7zcm&^O<|bP_
z5!;9>1+3iS!@DzL8;ZdMe{;iry+2+B&pZAO7899IO)GA(wadfCU9<F$F9JgazGI`y
zyTWxc%1h}_D3ynle0Dg;9hovA3A(j8S}YlIC^%S!S%?fY^B0Rn94Z)rtP1iKXk>dp
zm=aGYFe5XefZ2MB4;Ihk1l^$JWla6CH1V-0u%+JAYdC&4PRq{!J-hjJR`@9(-9;?L
z$7S!l-<}bdixQ_TG?fVuO~f+vUHv=z`Y3#5rWv<=vHf$x$q9m8-4QfekyDt9VX)=<
zIu|i%9Mmg?)pY}GCVl!d;Yr!}7X$!M`@KUJ+2qB|iCev*O!HxS*(f_VPPEePKVCJy
z=bEG%3XdR^>`0j4<ABcpV>ikx{H22<9y>80sKC_7WqjYub}f^kv8c=bJ{}cXhxPP?
zFg{yMkoR+JKYJmwyxE+8JmJH^-RRi*CW-zn0WZuPO|H!Z)JZ6u4rdVw>gojmSPlN6
zWZSH7IUv5LwPhJY|Lt@$;^ruj^7Q%XNP%1cR}u0cWr}rtw08h*D!2fr9B+$skjgAq
zSZ%Zs=_rPzc~}~Vn(husg<t$7NczVMOm*CHhts|L9M9SN(cq+PD@zUy!fJ{=GYC{H
zl@Z8%TE6pIrnX*)pi_c0YXmI12QXolU4)uxv@mVdn<z`lD}b^LnGD3I0T^KIP=dJi
z{W?4UA10^`xLQuG8pw8Y;zGXqFpMOfxAO#n#|nL-lv%Sl9RJkpY59owQ|W6M_I7P-
zG+STaW%|xVa{o!h-FZ8_$}yMa%*Xs_-I5|Riwqwd<`p7m1gKdk=GH=ujiLVGngj`L
zGb0+oqdOlLCroiw9L*;LdwAomM<~(`nPdGiS`0r2dUiIasDGJX5a)1$OI>u#==p#^
zzrRe{EUv4@Q|fG8Id~VS^`XS=Fmt#m+lG_xs`SJ?X_p}@(9luvA&;)IJKE1s9JB?B
zXn2?63c~f8fn?CE;QbWv|Dov|gDdO8bsO8ZZQHhO+fD~Nwr$(&pwm&uwv&!+b)37u
zbMDFCU6snNnk#G0`99AWTjplLQm{LL!KA3DsL9F6s;a8<D8iY4h+*|S<E7QMhvtzr
zsV2F+^MAiyE><Fs>0b}GUW|G-WN8{F?Ly1*Qtmf>Vg;tmyu%z{9{b^9R!t<Uaf?ob
zcpi1hqV~?foRX1-K%|WJmj(MoV5;0f2hkZ=?w{zZ0TiN;sSu{jFhw5@Q8*G$YMGFr
zQ^sRw8tiQK+uJWVD2UGT9Splb94hufm0~NyOlU_Ekno&bF~y-6tQSf?`njZC^3sCk
zEe3%1c(a%Kt1CJgSjSLRhMtz4ClnUxOCJ5Qd7z!}Ym-5qz8EYbB!P61!WbMm*~D=Y
z<O}ln04qk_zW24^U(7;)5`t1cCc>?@re+RG7z6F}vM}F~Kzkzb+CegMdaRIXZ+}b(
zZfOETkpRWoiTHos9IBYuSY;JZX*$VhwhsteXgX&o*?!S-<Fw|UJ;@%_mT^5B!(l$}
za+JFyPpHx2P6Kije)I1)7k=0Y4#vEVs95Y{fF2FBX8{?s)I<y0#Fu<|vFRaN5~fNz
zX3{lGyFyj8w+7a7Mj(-yaC4rumP>eMc4ZKzdJE3Y{L{ar9^78ckP%(KBd98E>;6Yn
zq3mAbFbmu;Q<cprN}2|T>+^yqf3o{+VWCjZRlw`V_yP2`pZ0Xk44gMBh@bH-&bt;B
z;EWq)**)Oq77Fa@Kys<x9}Rd%<zysKPsTZBq#O}TF!jmmzedAF1QE9?3D3)tC_2br
z<S19|QUkJ8;De72&AT#I^pVi@o-8>edXcXn{)1YZ`1$Mo{y96q9-AYhau@<N@mEC;
z4Ynf<6<HL@V(Ayo2xYbD9#MuK*m74TApIx8jSFi)$ID$SHvz!qtA8Y1oqpkuOniVW
zX#=aln9S_f45kEs(GH5%QU^KTulw<Da#M)zZA?{8@5G5}C&OOT)dBU)8v);k)e~fA
zW5AV^$PW*|`vVR819SJ)4RJS2)h3knYKJ3lSABq*5{$Tb)T^;{%N<;)VLhk}Kpkuj
z6(4*To1bWOK@GMbfchZ-9PI`*S#_SH(fr{{U1I;+&-HWdC*}0PqIeI%lxK1jqM?+`
zE*n^87^iP@FBjj!R$ugK7xi-670>tEO7(WF)hdv~vR{q*8ZdPwNGMVOdzh4jA!qd2
zgNX7sAOMP<<x9+9az(hS&h~Bv0EEIAcR$?@|LPM42G7oMRgPpRUHK8IBtG#ArR5)=
z`K?~Pil9V9h6m|;JAqrGVRd=5o~iU#UG&69DbF|$)I_PE+zOI-DkM+~ma3X+C=SWB
zd{^}EG#ivU!n!Ck1H63YI^(eLEm;EDk9#O27TNp`hd<ZkCr-~jpEpN<Y4u;)GX**_
z7;OkelTD^}*%+E4316X^JcPf=*e6&|v*ZV+=!<(Shj(HlJ3OaqnB|aMe^QRW5Ri98
z?I05{2Su8~rK9C$<z&m-*qt2$Qxkdp#E1@Q#WwqR$?EL(m`Qt$+u@FvmB1Ui635Ge
zgp3#pAo+0L#T@#1h+7lT#Y$qki!2aw0Yihw8(x;d`BRZm9?^@%1XA>mIU{LhNqM8<
zj}(`9^7tlR-mWA7%X||(eK2EvXA}z9-zt0jEFF<=r5=cp<_WTWxHL1n=&Ec#2}|wS
zQxYTZ%EqmL=j%Rg?-?&Aoi&5>_~}vG3{KTO)s*Rw`|cGYQ(8&Dlcnw*sYqqYO7l}Z
zeN}F|+3Ln~#zx2?3$p!!Oe7C+xW#h(r9g-o`+yQ+Ppk~%43Ibc9p6vtNgOL@xA{t)
z!S&Nw0Osd-Q}h>nT8y(HulN1!2qYrp{B=dlCyneA;#+d%gAfQg-%tWp5_!zj?NUlK
zW%Z%MN9+)IU<wc5K$KcCDKnJH2nG&x9qPHw&8?+ncYz@^Xnkxz9p9e{QWtm}a^W#M
z8PJiD8)YKNe!1cXg3j*urz4q#Kcq3dBRlG|_{$0O<HZj^X0Y5A%G5cqBdbaZ(*0b-
zkVNhV2N0x32PsLYTH6BO260m-+|k4Yu59QEPzuZFrY-?!ZX+j^Qow9>woS1sE)3I`
zGlin`oIEf4e^rl_gna5?#P1j1o5;SM-##Ai_v_~Q{F$>=6EL1^N#=SD{^U9vU7*LG
z+-((TNm|@y{g{IW8pqz0;iwv-OEET8UF$ad)UdX+3-S75VrqCYm@G}j;@QuWrdZw#
z9|9@W<p2|fo@+Z`r2}yIID7%*(|o=B&5hJnx?FF9@CdBhM&P1ME2Z^An+WQRk!7Cw
z{}tSyo_9zwauW<yRpR2j+WO{)`0+cUF+WzWi)DdO)pkl<*kU6jtBs=iOW@ca3;Yfp
zcw)kdrn9tE6jWKwYcUBpd8OwzMf^S{2Xv+JcRO!x_##S12YT@>0L3ZJkRy6X95)vH
z)Fj1KwhmX9({(j2CZGWj5PJs%+-?le$ii&#5vZ_WH%a|4jsX)B`7iE4dMU@89`GW^
zgxR4kDGH37C8@ys&z8@nE_&`8^3<0OLGEtU+X^b?1ez_2PTjfOQx4L=j5s_sfWP9@
zxX8U#v1nlMKryN+>s|CX>#mJ+MI#fl^6$_x&S{{W3*Jr-YdX&$I}z%;AJ)yq&DAsV
zWe?)Zgv8N}f(ry>a^+|<(ePKm&3r0$<KKXf+JAS`@T1R;y1#8UUZ!Nc;XPfCp$>D2
z$M;9rSj+^1R8db=4OiC|kT@p#02;x=e|Yq)%^*HNMXG|us3ZLiAcyxDSbl;CdlORl
z#l>y;q8NpCJF;r#3s8qi4!2*4v)LWj@w=q=oQb9t+!O0AX%xVXDG*^1yO?yz$;(Ee
zGuz8kH}0M;rKKH~l@0wTlRu4W+7%HdR<^cKFQp)-VhVno#7sxH`ZRL~P`5E`box7~
zE2{qR2#QTD{hc45Ue3EA^=<MM$P(2Sw}(ZQ4l1rbS<zPi_iw;I)7eGfckKCoJgJ5L
zFDs+}yQ%#r3F+LtjBc55Vf>dcCYx~o=qI<Gn>3qAn#Q|o;(!inQ_#_Y89t{!0<|0N
zB6~0?V$FlEq(xeLd_4g`fFPMND~;t!Y+(9O-n^BSg;Gp`QGE3@WQ3AmN%+$D8O?<o
zh%MV24!zhOeRkGouVmt$n3hOR^xYhQ!7=?#YmZ|f1bUA4(NfeF*jn7~;n-43#QhC$
zv7q?y=*AW?zWJ&Q3I4XZh_9|_Ix>1v0xK>%P9|2RrC%4U-cbO2_>DuVpAAj%1?$b9
zr|HyLh~M|u!{)zM8J+SqQX&k9A>Qs5-h_p{LBRP5z&wIz5EdU%p##?^=C31~<RdKk
zZ3RKzfCUY<8i{|`lDSKF3E}7QTW78<spYg(hh(MBCU)8H6q-{8?z?)-NBwuR0ODXa
zX$+p-h>TOm;z1xsNv*av^#B0LNoZacMnsgFBqw*1W+jSVR7nAK&2g$!DnZruNSI_f
z)BSmZbYt+#W@M<PZVhE&MQndnype>}VN7>@SeVV;U;RB|rq$a>z2>p-+duHvdw<4J
zD|Ke*1^CNHpfGj}Z4zse$gSNa+5-3KeM}c{GgP5-hXeRS<@&9-*>j(|W75=IPMb1z
ztga=hyUfK?8@Tt(p4h9^L#mha<Zj0X=C}kgzMv*&q^U2jj&1^OXD3rOFdskWZ=mO7
z>nT2IFvn(fI)2a6$9~l!(iSW6)M<CK4xOodx!q}@B@WrD`mr3W2HxOwynC*-!D4+h
zWB%V>9Z+$w9<S~<Z^p|thdBnC2hP_95V?jCyb3V%ajj&5^SFSF#v|gh^&u^C&5y|l
z?yAS-#QDruRDq%tVj6-!4Ja`nAHc!I#WFd=&+}Pm7#rJynp1HELZg1XUGi3mpv(#m
zW)=vvd2jAlmq!a}*@^v;US#67n`5=Hv9UYH1!U{8<^g?T!4#?ny274ThN2<D;>HDq
zUA+?BGIDrtRt==I!-($M*8-poZPT7ZzetBR=f#vI<xouJp#s|jSx$N3t2#IRq|2~Y
zrgeHuzt8ZMg#+Fs825<{c-kz_5cY$RMH`R@2|_E4mjOlWM14uRee~-A$vnRMw3z7S
zfJgH%+7QAF!}m~e9i=~EGX4zbaA}mUvJLP$m_eLGw9?|#MiNqzn?obx@Z+jXI=>Ib
zhUtNnur6fHnNt1C8Ipk7mGe9rWOvPERp8h+5~1Xx*52-{h_Y5N*{3h*ltN{|0c4mG
zaNBRDCEljgl8x>lL_KZL(riBP7;*Osuw@n9u}NyA2{-Z(_txw~*cykMxX|eeO|=4s
zuCF3zw*!$j*Swv782|QneOM~se4xs#2v7E|0ND(xZ3l&_Ip=-z*(A}yTn2uevOxo*
z!%Q2_NLr=&+L>%XcArCqkVuk5*DSWPlO`UXhH$*PHxx^2m%HH?K16Oe`C$pL@3<27
zc|2T?)ppwQ^*I<E2M3~l40!y^T-~C8ibtFyaE6F-?ne|QnA6<ef&bLfltbjoJ%pcx
zahk(f*Mn_ncpzM6>@hX?*4)d9K=?26gS#gMAF_A}+(&t@yH2%^!tfd?NkK}Lknigb
zJf6kc!|B2vqr5)Vp@bhl8DN#U*Ht{vTWuZ9m@VT*j!(mdbcdgrm{^!<Hnq5<q(r1y
zU&=ielx~EaCW0p4A+8OB5|ZY`$hU$N!=J>Th}%c&MWO<8m6R~{nUb)U+hKd#_t4^B
zIsQo^5KR5c!)UT>A&$m4BDsc#+7_CQLQLTCe&Ss3M?^&sGPpQcA)qL#GYwb<IESK%
zB#3s4;yozCmapKNF=E-GB`{9QN5elqxi+yhb+QARc^lv5%T?wSk)1^sd`rRdxSt8m
zdcSU$t&<ul76QL#vVYqjb7N3L@`XJ9OhfsXMLP5D3{$tWs({k^1|%e?x=SI=MQIDb
zkOc2`5)>3BmJi1j1L|AYmo-z-(V>G8d{~p<1xtj7Km=_a>1V5^?i3vz;g4r?`TbU=
zR(xZ%fjqW<bK?&z{W?6Y&D}hLRCCjM#*%2Lyssh-+7K+58*fzo8xRIXk=`OY3TjeD
z<1gL&9*22UqhtZ%R7R^q3p4Z6{3H}?Ok``Tf5w3asb-cWz!db5ijnrpO8|uCIUNoU
z1O=vY&M&*k|M<IqeHY$OMmrk&xu1`hfool<aVpM*8jiY2_XJ09$oD&5i|w<<@`G?K
zdhQ480tFTk>Dn?*7EnrjVOzW{z~a!)k`iw$COaCphl)eEe-RNxz91t3?O&{MR4(x<
z3W`l_w9*e?<#F_XNU6uyX0|WfW-D)S+izQk*LGT)_x1nI(7a-gYUQD%oBKoLe-a%7
zmdxALyj8%6u#;!FEr60j#rdnqK}-}#=u3rEg)vWOPxEo%u5hr+6FuEnQpr4hP<1u4
zy#NV8ZV>-rs%!QllS5{EOq48WI?{6Wo|xC`@h|{*=#bfg{|K4&Zpanxq8XS~f(Bs#
z+d7a!a@?rC+KLilBwx{?%rS3y7B7MVcZUk$48<noRwZz8b8~ZPf|sK$9g+@z>Eb8z
zcBaHTj`2iv-N#fAFpI5ZS`R$mO=y?5Qp{G#$}v{`E8u*#;TS|!$l5FboW^3SWav$>
zvE&1wFA+C+oJe7POTs(AXo*;;fB-Z2geb+CF~*uP$WElSq8QkR2NF?vwmZA-krAng
z2&lKWA4TI5A)RGK$E}{vaA7$6@{>JTB2N&P%~<mjX$quzxJ8jWfcx{$LgF8Vg57=x
zZro^eGYH<Cswtw82OW3Vp{NSY7ma=N&}#tnB1s<ZG1?Y*8L=PR;7#L;S<MtLIjN@?
zp3_&V`-1s<;V};19#04evWNW4`11Sd4B!hFSPp_uy9>#VgzOh^=-{P`aJ6{Wj7F{I
z(_%O+Ly%p(8%Zqku&h6ZK=K%F1>L+{T|GU?(Qfw8aj^1}a?+J^Vm5JXC*&7n-r4}S
zEzQ9R6*(zAZDCO?(=X_0YV++NS~No{#`=6;IPny3!{tweqVtYN1=o|SHv<2&@$cK|
zBMUG`cD8e7kiBi4H=`T=bOu}GZ59TNO2gj(0zHE`^K#nyd90WuEi|8^hM*-PUi`Yq
z6d4T9892GaPBIeBF}_9(YT}|w0Js+k+utW=!=JVK^y=V_{{z{;vjTP)J3ZBOtrUXT
z%>+?rtH9JfLVxl-xK#ex6s92KqGimjTQSWV3i~3xg-P}yeEaz)SksjxaCw-36$++j
zf8n0$niWzV%ZWGe;;WQ;7FF{8b8?jB7f%l|L+7Aj+#&QI!N*@<nBHRmGNgXTHgaw%
zDVh0%(9+$mh({=3xzQ727?ml8gR%@KMu{lU$SI-UVL=OgRXQzzkRn+k@BZ*zTuXJB
zkHm3s3sKss_ctJcXEg{1ho~?^TZ`TIt;e>V5|@=Ew5YbeSSl@XA>|M!$j2{nA0tg=
ziKVIE&JV$Q-;x+NGcOwu3Bg<pol@~Kq!SHI0Z9?&vo8Iu6w97?h?c&}*8UG6IuhsH
z9v*b~RVL@$_MlK>rN?2E){XliL*`@i1UZW8dH0&JDGRP_F2Oi?Hj*mIg+JM(t#n?k
zRQjGGKIC**a*uIbJnufJaG9&64VsZ?;Q-cI1;uN7V`hGMAmbE}0-8}>Rwsr>5DZeo
zQB)^ceyG_NRYM@5U&*lLaXPV>Yt6^aXj{*$Bs)7n59EK<Wa}hn;vt%~AKGb4spXX$
zg%&oJ*_tRt;%?ahYbs}5wDY0`&&S<#LxOVkA=M}HxLDyqgt+np!XpXh2jUlkK+dy;
zqL`D4l)K7@;R&Eg!Dq1$((yj?w=*zmbupdFp9$pe{<b~3?Ec^(#&;YK)VnB#)rDCA
zTZ;+C7)suS^Loc3yM_m)X^6IlSsOoA@W$=-;o^ea<U^TYgU-j`(X_On1&~upLaM2V
zM>Eub4KB)2&FJ+W?REiwwU?9r^3DcJZ%<tt(0<ZH0swZBmRA2W)XJhd{*?^RCd7lh
zTq2?4q)S?o@-eVea<d4^t#8f<!SPK&XM~~QBpVRqwcQ0nVGd^HK#!9FQGG=1xM4P3
zP~l%>H3-?iui+cX*lC7DKWV(^5$QQi=6bCZ_9Ol)0PwguPQNlCKDTOMnt3*~e-VXI
zgb6?g&_~2(0&Fm{gz#;~VjuQ)sU6oPK@2M;7WOn~`3SV)tk>3cbhENSobpq6sI2{q
zyy+T6guYB9@lcYJba_lC>Ll%ddgn{<l*ZR$&E^VuntA%FD*XVqd<>`MQFXCVODwqO
z6e7mwNyRlaSC>}Sse#l_e%7tkb-1dJAw)QUHw{8=n1sYIWx0h-4;J4fii&>TieUtB
z(aGQNkK7IBZLCeO@?cshBN+yQZAN#Hs7@RFIZ3sDasAs7@%<`N(lGZ|VyN;c!(uqn
zhgE^&NV16lji`4YBSeZdXIzDht0B+`I9kV=+g@b0K@q@4qjJXP5%YKimOnpB)$Rc!
z$~i*bd-t5V-q*tZ=WBZnzxg|y41Trz_H@$}#Z%?r3z3b^3E|k~8MDsJkLSB(v5yS-
zH$GN%E*6WLoRyApeq~cpN-^_C{%-||QsUB?E2r`oo=uL-OfDe$1vNg({3KIljstiu
ziW#rBy#kqE$!zrdOXlcihbY`y%)tm4oRrcwYHFMsLkTAjvj&b2y~e{L&Scz2c61cc
z7(0qXj}JuyZ{a57+Tc=n_7_ngu{D8yy~q%QuHuT6wlE%DwEnzI==kk9HsFzVaD_aK
zW?S^a>JLzf0Yp--`tFKoYg?aueZK#5)ZzF|B}_WDj!}fs^d;#{Kn-1TtPZ#locc3N
zTI*zTc8HEwD69U;D}!pa&t%Dz_A)+B&fV>28o|)mk0u`AXj=dunZyyTEkkRY)nPSz
zrqAqmCiwBNxeeXnHRVW9NQiyvP(eG*eXd`2I)HFUMOcuesmAX4T(VpgPb9nL>t^wd
zWG24F!lbFOtX3&S0)}&{{0itDwV(=n<ta`_o4{E}FWVC74-<ntW(xAHb8<3neK<`P
zhyL|^v)O8YQV?JmnNYy99Z>9kwe_||<IS!IT^O4$n=q`R*zGw)T2o{>cngy-9fe5t
z2BJwDSma$I6;^}%qI)gO^I$qJ;zkQf1H-zsngNAXN{I#Y^H9OYs0~nkv1PdL3XsnE
zBmB$t$KSVbM$T8+;sP%yI<zfvC<?I;bX3^n6k}f8Gjs=|ErV>LW4oHT_kXN;A7pB9
z5|VPv2U*L|rxk1OWZP*X+VF(Fq`;xoXq?nKp6p1Ks=**k-T`i!hR$lY!{orjKPNDs
z9{1IiCw*^kLVowVk>7xV>-<>ild!ew#S%G^CiawL(66uZr4bwg{}j5nl?KrHpt9Bl
zO|%26^S_gmBIt5kKPj40m4h_SG_@8bP8q_!!~c6!-dV|IX;risvpIg5zbDHh|20j#
zrCv?A(c`Twrr;c0oj1@o3Ue2MLKjB2MUiZucC7U#0OqF-69DDYma<=5IXXo+4<zwz
znihQ?#bA3b|C-tsB3d3@kl_`zIE0Yog!y=w`kb(ESbAPgK3jijD-o`B1+D`T<w{1t
z%*i(3Ut(QO&XKi%NF~Z`Kb!kwyh1FHoV5xDDcB8IEz-?fa*_~<s!{NKe8)XQFr8DH
z_gAx%LTF?u11R=l*`@+YN>A?h>gDxnT1q|)>Y`RNKsFo4cw;c4dSap(;rY0(rHskQ
zx0W>^*9BS+XmRd_z<nR&f4B5v?Y%T+ZF-k{fQQ(*sTy9SDCS!`2?3&@uyVD48<P5m
zp>qCj?fq&cm^&vCa7ZoL(kwlFg;jnCT~Qxq2mi290P56&hLI|+9AAHe42;`%&XBo8
zS5aLC@2z(bVqf*eRD30IG7GOws&LO-d$FUOtsDwotn6+4yJfU;$QT`g@n4Qw$JdSL
ziV259KnQg^#kry=rp@eO0U|Ubx6%Bn30j}m^3${|VfVsY#!p~eHm}!f?=g8MKX;YE
z*|^IR@V6Phqp;_S`L3-F%{EQE&8(^TK%!+yroM3OLnlxc6PzAfxNK5Nfc_xs1zy}x
zk&{Kam;j?Gy1pVq$+&N-7+$T~eu{?+<?#VQp0-Z(E&s$x@I4CRY=G*+=wC0V`}u?U
z@8Wo<taE00e=-Q4NX=!%f)&1M4)TrF0f~5C0DXD#)x`b65(NSDuP9&_*V;!;)h;YE
zKRAU3`k_%uP2a!*osb|sD_L7r*l{L>Q~w{KV5jry3Sw<0bI#vp*V*}pjvsHz0*D_a
zhfBl5Hk&MQKSk*Btey)L(8g$x0$q7I4bk!8$)5C<2w<Oz=3Z;b=XSWaq@e{PK%-zm
z066<Kz$9ApI;Q$D!)DV7a1u4U7fem*e4V$k%vwsvJ_V_eQ5N1c`q45}TI{JN_X8{I
zPsRyFgfICKCPIqEgi(?We#X$B!9Nq6G(!Bn$aiqLus>d)q=(8ajx#lul^ucf^rJV&
zT<$u<F8}dC$fCJGoMel4Dx!NN+e&OGKnvWx1Tm}vUWFT54jBmM3TFJ3&uB?-N($CI
zGrI$WV*Fz}r$+FSq`-;Z2FWmj-c(IMz(OQbU}gEwa*ioO@s#LuulaLF!%84&9Tr7i
z?f&@q&QU&RA_e?X;c4lR5`stiyJcuJ5Vb6+I)q2~_|$BYM1X5lG*g>4*~2#+0Ism-
z!3dOQmF%Z1wA{x4p*mDbwmG0GwZrh6*i(i4K*^G1k(<}SUG5iJqKMtW8jp_+ysiZ<
zm9mGK&zTbc#vUW<^h}3zp!f=w#P!n?ViXGp2Zu;E&ye)koJx~1N)T&g&%^lnnvZ1g
z-NXdYEtdyAgh}zm7{t&>xV{A7swNrvgJ7$0-~6u<o|_|hEi`$LRP?LYh^PU<r6vOt
z+azHMr5wM8LM=87Gdh;eE3D?ZjePHsXD#fojWO@j9GLrqi;<_NtGzg<R>~15Z?@HQ
zcZWw^jJBap_+Vcpek%0QL%=1`KHq?<$jXqeaZ9=lq^0z!Hlglj?J)zu>5cqs0KLuS
zSx4`Z2Bjz5XvAA@b$ezW;W%Z@Te;U_Y5J#3Q<dB0|7QlAntA*B>=V691;rN-c+}tB
z#i?tm^y9oOd$KWxCTb3(2rD?cimp649x?Iu&uV55>`{7ed8;!-8yTH3&IpQUVRI?K
zK<Oy#9B<6`&tZh@)0jM<q@||&Cs0%{Qe6f%(H@AdyjhI8jk{3Q2JwdlU7YNn#5@)0
z6nrEHrMEK_hb$zl4xZ{x>W=B5;pOq=>8^=$Vx8t#IjKwK8TJS>OjzeWnH+{C`bkn-
z%LwGfPj8}&Uay<8Twz&{D<I30`P!J%-QO-Ko)3!pt0pQ4=FtPt!aA>-F6l(JyFEPk
z7dK*<s~2m9Ks=N|6jJ>TK7lcXEp|K`M49U@h#UB$R8vpK%rrJWHp5JZZ)~^<^2}cb
zCMO(SS%_KI23Fv~lrS{lXyp(jlMEtT3XK4Tg>wPwOcE3`BqbdY2?dQd{_oMn+uhUE
z3oI>2sJ{GT@)p4OlBNNibxbuBG&|sjYzX>)VcB+&#n-y~@#V<x|5m_s!GEeHBN!YV
zS0fI7fet$>59t5hjLd_iLWnR~m(%S3x&U0M$Fuo@_8VpAEgw^5S#wn{X>}hd6B`k_
z^4vjBD&Nu6^U-7+5=>)29-G$xWeH|Cx5Ifl0n>_2ixh$;EQ8=s8!G~wysSb@RCHtv
z96U7re^(>NgpX92Um080ilB$Z&`?*~o}=mWIGnG(N-LR-`c@zA1(RR?M*z3+aC$jI
z8ULC*^8>Aq!1V}x&n|LF6mouE=v<yzoZw|%`axs=m3Wk%?!R*ORhZTzT8M!H6Fh)X
zmBV}lVL#c~sbGbptGBy^LYQBWSFo}vP3C>+?5g$-x38O)x?dya%uMNhh6EEdiT=UG
z)5|327bE+0nptQLmzF6?lE#A6e)*nhF{x-UXke1NH}`%tbWl_>Q+?{4JyD(MpD&zg
zDZa-bBrr6gV%7{mz}Ia81~>ojfO){-mWwoDb8q_tN+deOCly9zT+RFYUDk<0gGv+~
z$Yq%*^5jHXijb)q9?y_Qd_LH|du@%q9nb}MenGvmtWZ_=UrUDu*Vs_lcO!27fuBp1
z7#TdT)V|^fN_UJT_<Wut3|&-IhR_aOK7!HVLopsPI|BMJ!bndPv+BX-1`vX>tcBzR
ztP2`F>hH{hX77`=1r;>>JZ8Q)h}J%dV!f2<lb&DW<p==Y(lC~Nk1ej=$hxvOO$v3j
z2dH^f{mvL`^v5g~nF8ByIR&3llq72JGM9dQ=%t2jGUz$XDM(EX&ByN{OXKuFVN*kW
zR~mGbXS_CJcfW^&i<O@o1!!w3X0a4^=X4@dfrVN*ZSGrMcoXq>;&LPs5$JEG5SvlX
z?wct=(^i(5gU~IN5o7T@6rE=hAfZhYwcBcHXi#p;T@pl>%z&Cch;$HivX5wFp-_jD
z`Klqfl2j31A_X<ZC%mj2@dM#j-GTq+TX&FY?*7*!_HqpQqzfQLP$Z4xeA5gzQFyZ$
zwYQO}rcF3Msj4Kb5V1qEvJs$&#>yb>H<aG|K}C_Ntfo5)EJO(KF|kq#(svYfI(&rX
zyhV%vk+bU$QFV<vh?Lh|7eHY`<d8BmOuK4X)DM3!4w$f1jUF`>)zyJu>js>2*cADi
z{@O+8ayxVM1wc|5#Bp#yj?i6Ms@jMaS6puju{c^3hG0xQz7I)_9+yy&Qm@^=Gk9(1
z?HgZFimye#GCS`Zu4+5$TBfM@B!DK3c$MvWlF=SkS;ux(pjN2Q2NONwwS>ey<e4f@
z{Isfeh4+uc!<nIx^*`J!9CS!X&%pn9wA1V-jCw8pM}Turd@kqhdEdv&BmP9=fR~$r
zJ5TnQ!7Ghj9cg(?wPP3)ZGxrkVF~9N5<8?ja+-$3prlR&SI4KgY1)^}0wjfV4-e0$
zAlc&ja@nnn2@`2c+l$kUMCyek2r%6Qvy<?ycB%#ItQM2)FDDoMA7E<Ig**Rx{oh^&
z@^huRrU2DS@J(7rH}+?oL!jN7DfdzMN4w{WGgoirc5Jg6SPY>waAOE;Iiu;=8Io;H
z?)SIw331BfwADF0j$T0bSbT1W)88(?o393QYo=|V+qQo|e$`X`aPqTVlRWu&1`bq6
zj_*0LYzk9v$1NBNNTG@N^<EswC*H(kd{qM|9RM$e<yuqB$dV|qc&Gp1tK@+orf8tu
zpG*dIhmxb-f1^YZn%U@==gmB@xfeaucM)(0VE~C-pDRc@Jgb)dFL#7p9JOEO5C~W2
z28NGeqHevLxinI-EQoiwAJSOM_!c0v%Ot9Tgh!$Ac!$I*oy}m${@GY3dUA-IER6pV
z1~72_ZnAT;e+RyqpFO<@c|2ZDo6)y3owlxg@17R{$jS~`x_DPrqA_J-W3*z~OoOwE
zOL(O;=?s?^(#<TRaVFQjAahp0DQzf-$q>FgT)(HxR}_zb|GgS8F|MUK8@G3iJ+rlk
zWSbOC434Kww;_d9pt`OJl!ys}ZCD8cfZf)235u_$2cp}3o+c|R%ap_}vKFuxOaVB8
zC>2ntoggdO_<Qf4bE>D<&>;g$BAMZPv;vtHoX|)db`lYUcaIo?3Ag81>6(k|t?p0o
z@)NZVd9D77z37!n`9khL-?v_YYj?-*<KyXF_}kA17~N~pv35jbDV8A+P_Y2O-8Lms
zfie`uLyvahRMnVvBV>S_V5u*$dOyM{$J~NX+S-av!3L#~uhB5Co$>yNr1~JE#XwVF
ziO`4PNSk&DIJm#CYFzW(AIXPKEbP4uP^K#IS^V3%&}Anj8IoBoaMBpE2*AB8k%RKu
z0a8EDRC|%eJ}Q~>oiL#Lvho0Q$wnL^p-bl6XU+<4P5BlMNLip)LatA5PZ!)yTb;hC
zJ@3)0Kz@g!JKg@a2PziPGJb&KxeOu|?0O(Wo$Yt`;312$e!v>iVyG~Xvx8I%?ry^d
zOoNEyS?rJA!^nr?WtpqEx&Ha!^qbupnW|0%wlw`bZdTgm-?iG(h5^1flP%_={l=IA
zC}uF7I2fr|=%;W!9zs^ulB!*S(lKKFmc>@bHhZLFky!0Mu(5?c1m>^h8X_mObiX?r
zEcbHqBWI0!JdV!Wfq(CIuiy7TX+h<zQBQYIUuiYTplYmzC=Kc|*jQBn>|6h)X-PMU
zo$!9xfPATINDE%YGhk<L9K#_#wBRu`DKd!#rcsW@4irHEn9O5Z+g=foO!3ATQZZDA
z6XZzHMgUdaB4?wk`00FFrg0S{@_@QF-_dA%zSCY;%`-AB6M+cZ^9`rNN$7?mfoWxX
zbB>OIyF0_OsG5OB5$`8SS(eo%Y0zl80g<`}Wuk}TFk~%&1lZSI3|V%xGhI&0j$LuM
z;Pb6`A1ntB*tWXGN3YCtk5^F{<MD)}3P5i2w=e>lIY*~+(>S7Ut$QEtoJLmHCsY)r
zve0*tB`o`=#XTYDfV)g9E~IOc64d=Go9aqB#;`Dvi1QpZyHtrMQ0}}~X9p+G$&_j6
znZJM!2-Zp=!1q(}zgJ)@nyHY#-ElrGga#Yj_Gj+~hB!e43JHVx&Eavh2bgG2!cM3?
zE9|w-v=06wOBA>f1{U9e3@I29Y4n$??~or(u?Ji&Rr?fBT?rhWj4I~~0?y8XUbO42
z?*9b3@Fz3ZK>S+9u4BX0q%wWgqi4%N=MOpG@gV?OsBkvftCmy|*&t_3+T_u(C|n+1
zw-naT=Y$!j|CCz6Ml)O_>!b;G8^zrfY5fc?El*Q~0->p~=hj{h{Qjs!FTlr3bl+Gw
zz~}w^by1p|F)IuEx{Ud05{82(Z_I^4JIm$`u7Rk`Ij<TMg%X3(bU*l$CMu^uvK9+<
z6JUco*bA<T<f@5&(jWG5K0e*Lkx{?XX8v~ocyDdc>2~lOP}0`r@_aZi{=Pa@JsIu7
znQ|iKq^$z<ap@K;4Eh3J{B$DQ9+usxMW1y!z_vzLaR%I0`5`SNyD{*2Q-fKYT^+?o
zbR~3{LYQ}kVbxeoGn;dP8Lz_C`}HObuu4<T?yx&)rcW#&o7futufF?q_!Td2ok(oJ
zUCG72B&vO{uSFz|Zi*t>9|9&wBO#xenG8nBFAg6O>s<sLs@G-as6OO}BI-ZlYOZ>*
zo~e)+%2klP!dK~dcfWkW?fiEWQcM~+I_L?w0ut>Fe|0n%vvg667Z%e`M|T1?z)U2>
z`sA}#QzERQChT$6(XiQhw?+pv)U~4GZPP$w-o|D^(-u$m_o?iZcAX?CMkT=x`^OXL
z6r%O9&7zoSr}OqPl2;|bj6rIRpqtLEW%|=(K7M5;-WUi1?&o^b>osvj5genNHFf}o
z`|PA><e@!9Z0f~6YAsEkt*(G)qy(jeTUQqGRqT%AmMHdF`HhvhAf!Y5TLvdwLn#q7
z)uo?6FR1@;4JYsc|LeEAmL0o#6C&_F|GBy9-NsG&+X@lZ>j&qbzutk-zb(TtJs1o_
z_gm;&%svQI0um&(_aY<_si?6r&48VUhT!Jx@~-OUVj-rV6?}X_z3KpM^7^xg)0Bm1
zXm?(C>%J;rJPSrIFw|!rTpA1Qdw-8{fAwaotFD}kvtLX8GE?=~MMG5OO;k?Ajk35J
z#PmK<W`YV&fY?vm@(`CLBdnh^0L^J3KIWsvCu(I?BUpN8j#t25%zjjkOiR>ZtNG@(
z8h-WWd8^mwKRiRZJD^hYXy|nlTot2-dDr%8c0B8molp=s4C&tyX6pFy!_z4)G5#k9
zH{aL1qcpf{atIsRY4L9TL8TTp*v+6AbQqeZ={<j%n{}HbL`P?fP|7>bU!D&0&4KI9
zDEe0-T`6I3!7y>ZpG-M=IfwT3q<n9=P%Yx9>?C&ZDIxj6XaLc3nF;!VbqtUV4)qCk
zG64q8^3I4o%6ORQ1>E6%PPvENHzJLmm<l3S$yA`V`+sef^cg^rm+xvY@Rohs&+S4$
zCwitpEgN3SY{M-i3O76s{f*Z3Y*a>Md<fIakY2C0mCPUx!>8Gw1p0?@KRoymk;w}h
z9@ceo_5mOlt6k${_?;i8R%NG>@6AS+TOu><jIPRV=r%!yw$9Pk`1iwD_%9A=EZ=bB
z3!#-R^FJXGB|gsbp2_nvHE-HffshaZsF9`~q2a9n4<M^Vp5Lpw7(wzC@}15AK>ni1
z@To?xukD=@cWN1n#h~BeC`>Y2uzfDn<oTL6jtS6p6}{%F>L8Le73h^Dpb(nl?lOCe
z`K;BWSO2WjfXo6&0>p91AIQ2HeflUe6%E@w@7nH{L?Rk6_~!#fcumX@S!uAFoaY=+
zjA#+nD}&bRV#Z+{JYI#=Yu|u?Xzdw-omR`I%kCZ=U_@fiSSn?B9!@ovIME|W8blHe
z7BI2y8&WQPkaxsx|8U^%;91tyJVZHeOJX$iU3qoH@`2(ecd_<Li6;G_x9<u>7;3%W
zdu~1x3i=%>&BkAg^Y_#D0PnJp(`a8$>%K+xvzkbuC}Rg8?<z~|unJha+FL~!X#Ru$
zNcs^BWQ!8xdv!-yGFDg}LqS=`R%ogLej<}dkvEo9bC)R#jHeTfM-s5@yVK*JFl@Hn
z9Hv804gGsgf}^RmT0WLikN#Flz8{FOi~1Ns;g)yg4Q3?(;u$1%fVB@G07Z^^MyIpF
z@%%-><r)0?7tDSB?mmSh^;^1jMTC|`J2e9oR*lOT*G#@JKvZAf?|m?we#IH^H<IOl
zJL66MM(&2Ha#g>|e<BcPd+d%49wH(jzM=L}F;&>hsF^Y@dLu!9ai7{kd;Lx->8)4R
zu<|vR)M{%4Ue(>|AQ_3O()wD)lL{WtP7wFiWrXX?m<2wy`4Z;xE6+zqPW4@P1)LoY
z7~*SB)s08HajT=^y>6dB0zxr$QDkh4#I>Rbnbm&}rrc92Q1;&5>!Qo64Q|zgY(!iV
z2Yo9cWxV?mFH*B!(UgF4nrX-x`F<W&PqSN1=K~GsM-f*Rp5E>6r@i&|8B_KJN$3~$
zwbY}Dg2|>rli((tk_&qor^sg`po|v&<S7qywu3~I$dymtd$!VnfPWtg=Tu0&$<@Oq
z681Hu<RZ`cDJ?PLdduq0)S$eC^ocAy`<q%$S9}3aq!a~*!+nAN7yHo-gctl)3S$W#
z?4QOIL<6;-+c;L;bZmn%F-Dtng9(pT_J(avtKH!y$5`j0KRiOR@w>}PrRRH=nS1FO
zA%j-$OsGIhT3;y-0QUGfR#T3~-9LaYx=5vhUv3w2|C$>NfxrVpmU0ftwRx3bz>CWU
zL;we0Y&ypyhfNMBlwS2k?s^yS+*c;`7E;MV9{}CqpXB*SC~Am%>Ul@HD^gA~A}+5W
zhipnbOQzOfe}$Q>CDy|Os{P~2lsVgn?4glpyT_%+{k_`~uvptdW}ui3DgEru8p&lq
zk`$X*(nix)71vS`+?5{-L?aG_**<7H2{t({ha8^|rolK{<D(N|;-VT+Xo^k{KQc;;
z8~>olki5UWzk`Q?gN=<EjN9VO_4)u@26!82jjnjcHJ6lxc-SuZCGYs%{QkECDWM`F
zA{hgDX{dn#P(_tcK@qJ6qXR36`fYl{N!1nDdXffR%4)O3{O{k+=j`ujh{BUPc6PrT
zUBCLb;{^5q>{(CmpG39w8^&;(`RNfj_-B5|*g1!ijY2w0O4hj?5%b&-dt^WrDL=5I
z?`CoRr;i{7Tus6Ju}apVhwRimYeSS<J1x->t;vf7AbRwckk9{vFNep*Os)A?Vlq26
z6vn4*wUqMv`0h&n_thv;`W4Mo)ZI@NZ1MhY5>f;{4brqjOA3<)oY(+!+Ond~wk=Nn
ziNv4W8W|2?(3K$C_Trw8Wt9n(U``!E=e%>w{FJeWS4W_!h|fP)8orWS+3&R+*MEPT
zC<ds%1I~87TDL1V^b?4IHF>=pG_yv=0VUho1RT~tI&PmqymcJ=0fVhHSA=G{-6d1U
z_C)5e)4o6iQPA(B%)8We7;ePQ)A0JxF|ghJ>-VSg@#tAsg*>wfsRU5hhs*|MrbfrB
zUI`H7vCs>(+31Lx-<$Bq(tf&C(aqm1kQjU{0CCJ%yt{|H5Zavony&DAxA1UWH@r*!
z74_F5<0FH^)ptPf`I{VCkGklI1;R7TXQy3dFZ&o7ZRm%shfYyP?eFyH0xE*`@mM(S
z8`?&({*HWs#Z|xK$MNT%8B{b=C3#O}YtnNA9P!!;s*^)A4+p3=yDRj;G<T!iJ|3U|
z!59Usz=&T4Hk6?fP|yZ!Ukbi|14cJ_Hx7d34KRNwW<%zB5mwn>ba_OAeJx)|b3=D;
zCJ=3@w+S##EE*`LIWCP6R|Jbe{iF(p;-ylG887}^-x#N7VPbSfWD@$NGOyZV=^3ac
z)>PXS!SNRZiaw&C5&u4a9D0(puE-G<uqkC$)`#ErKCBr>@M&$yu-Vg5Qgdu$Za;cf
zaLsPrtyC!F+0tq!kMy(2>Z1VmGKB41%B#SXrHzq2Gyv=m=(vtHp;92xPvsoO)P~`u
z6;?M4rDvTnG>=CaI%wQa!kdF?4Rf%Lu$*f9vCDA!j?0ER=bPXD(0d6%(1*l3fKC@E
z!L@Du%nGR=S=9^ufI{WR)q$e1<ZAVyyzubcX1bZzhM#XW0mJy(R>ROdr{nMnxS|}&
zs@b2cwiZ#)&_C9T<0chgTwCW-&@CV@PH($jb0_F(?GuTdoh~R~bplS_64Nm#-g71p
zo1=<lIAu)>xZc<$yQmfvh^&9$0ltj!AH}nn%1y74N|Yu2YVi1GjJn|a!S9)kg*tZt
z-LEo|u<2`G01%>oveEQdMxdm!Do`EX|KR0i^bI;;HCz*xHPW0mo7e!PU;}#Epxj!b
z7+!-CWWTBTkPjm(tZFI;KHBoNZ3aJhSw#!7lFM!Phze3#y|g{2lS>gt0l9h`8~l9y
z%3z?4wO0B@A(hFONtm+`^gB4|X*<t{m;|AYrp1j=61$|1hIWzP29gp_vjCB0i9`ZK
z&@Yn!zuEGEO65W!^8J~k9WNBN;d=Hz*<y8lbF;V7knPLpcts>VwpJJC=xC{F1-~7p
zkMk`lH@O;&GE&oH>UG;O0mBu_`60}`{Ut$3rTCLLW#w0h5*C`Ob|>h3@Pzk5*E3!j
zR0)Ce?(Qp>RaMSBULQW@=IZM1?(*{T_V)Vv`oklV%?(fsmvP9DY}BW59seNYsa$()
z4{*<QaeR4IjYbfO@^$<phQz2pgo92)FjYoa70ZH*+u+8A9k;<d07QZn`^VvnyNgXu
zx<Bd~^9w>qCM!2h|FT!04@^#uj*iYULviBZ_(%;=?N9i5iUxWe_mt_X77NX+Nae6o
zTD!Q;xm8X{gvOcLinAY3G*=a;T8o&71oidx6&4y>s2u~_ozKL!j*hlYb^?KyI^cxm
zwa3!_Yo4Qa*d)0Y0I_OA6S&OmRN`uP2IiS8k5TO>mo&hLckcC7(!apJsl7__e{aEV
zi)CX~XhVO_G<@MB6AAeT?E^>p+Ot1bOWmqj{FJgH(=!;pmxz&j#^&0H*826NT*jay
zB}b;@@={ZB@^G**u?|m;pt^WxYDG-5QbXfAzw+YdD>=}E0CfWc<RoCzE{p-_^7_==
z%zeAJr@F1`Eg5xEf4^lOs#6J*##dLDSC*Ctq+Czuo-N(QFWZJ9HMEodz=x7T()=+g
zzl6;r1C|ef%urQPBc)v*3enp2Iq`tD+M`j?j=Si9{a*=Hp9(z|Fg~_mFX{R~2u!YW
zexDce@b)in0Pt|(ey89*-CMHkd*-@ZL>X>7EsXWA@Q~D^TxQf$gVh!%YxCFZtGG!z
z1~am@v>(2_3dVv4C4^)c%%-dsN{r+{XGCQrQ<}ySN1MkZLR`I}{~xPq!=ygD#RNfU
zyTrm%zSYg;^_3+R?-vuxqhN5%qoempE(WE?QLI4^z+8+c#DVjmjhN)(05g@Nmq$}a
zYkM<SXviln9t`1OzTo>w;IaPj>=F`0e`xIM#PqKT`y8cb!!CDgm)qwn{SK!cfxqi2
zUG9Hc>dIJ{^}#J&iH?;On*2}t12^c{kIIH6AznN8k|OzbpeLi(HvbFL;Gmk1w6!Mo
zaB%nmD%|J2hT<k-egZ;XHBJ$#aW>v1Dl8fqHjFhl=e7d7{|%<YTPt^Pb1xBW8UJ*?
zDOlL1{N>d!75BuZjcH}-PZJ}_EZ2HU%Q`IiAE~csw^*9n#U^6RLrKl>^}s~u5=Hny
z@Z_L4CzL#pjTkEMc@Q(#3mA3jDgB2A_IV7L#$RnT{<X2GrLPsnOl8R6fg#aJ_35$&
zU<>OfkXK=)g!i>(=9b2X$B|SYk~7^sVqm0ign`#pSClovqelp0<E6F<TlBUWnhOUb
z>IV5LNbBq_*H$==**{08{srT7byfQ*m_Pj#IU$dFIK0_`=_pC5*qPhBprUP(qb?6%
zR<RPob;Q3pT`FHcbKf8Q*k(<PY2u)=R~3z5S{S_x=DhcQ_9@YrLAlKV`3S+pkDaVZ
z%s(aS3>NKkd%Z7j7^z7?lxIg1GcAL68V!7u`d(CCTN~9dPF>@x7AMk&sv3)iPYO*I
z5D-Wk>`QK-;+clI+Y{d36U`<S3uFUuB@N4IxYcx5M39tGkWS4~T3_<gd&_H{Jn45p
z-)#N*-rXzR%zRgBC#c_m_I}76aCLbq86uGeha4cquJfQ(K*mx<!>`R`oRP`KL_NK7
zuq<MtX3Mt}I5H(6d&2!BNui|^v*i!4-i0-mMerZHgbc^`zO1crzBuoLsxk(2^SBiE
zwf6eI*9_CDn&d@LIsp71pmJN+vhi(pFHqi+IyhG}`hTl+BTM=$6d4fmbIHZRfgG<z
ziUnbR2xq}u@rLX>RY>og!d(2Iz`C~obvFw5dim_|F1mTX9SLv9=m!5KifcVxg9$mA
zQ6Y(b*)SV0MH|8%-;WRoPg(&$M?bzXtF8=5SD;;)J2{lI<#*edxp{Z2$6z#p>q7u3
z%+hQ!eJ===)>q0WR}}^{G}c?qR#@WRiSG83GxL2y%dIV#f}7Sf7KcGT_Xp_fPWEpl
z6oiobPN!0}KEMUn$23#c$Bra|DeFEY!O`swkOuul5=2yMQr1~yb=w0}gi@<YxgS6U
z9?W`iEo)uwbX%=2P|Dvw+`GBEtC0!Rw^;!7$>UyvugFV^*p*pukH1^B4w-Wb1?btv
zMwd5a#EEiE=Nl-qUKDeZKWKjwa}q7BhEpDvxoog_qFw6;6jH~``gFqWqiTmdo10lV
z?eT>G#4=hCpP*w_8D#;9$gYuaxU03oL_Q8j<)unH{^~2YR^lf{qo%34OqbhL=3(^}
z{qkaA-Rr>!g)XIKuma~crK9K1Vs=_z?im=G-d<j9g*KtiM&2&QcNLoQ*yZJwvp~q~
zl%7bZy38F|c?iYroFwC2hjPJ03@&D%D(yP1J9?@3z`}x#Jr%&!^&fl|t;;<e5Xur0
z9UUt#<G=qE!HP(Rqz+2s8T2BgoYq-tn!}Q;E~B-*Hp6bAb^mcmS<B_H`MXfgrZ}V!
z8~@PBys8=kjxZIFmqJU(Q>&c;Z7PYRg%gFE0g(*xMwW+CAhL{>rX6Fq5n(j~)!C2=
z8m~Ht0H$CSd<1}U!tyuyiv0_(f}GDMK0d@r=6R-7wKcaY_1AB-f^*WH4b{K770t>#
zfB-5pg6wwAJMD2SkH_gSd}unC5TEGu$jo{Gg%g)IB05b7GZn%Txn6l|Eu}wJ-Sgs~
zM-`4J7c8~H0eHG4KKFN1R#;k`<ZHP7YMt(HyQ3J)a7e(qEiCGPo{^;yp5syQ#Bu8|
zGn@+x6vw?$0|`BfsAB=xT|lFwiJe6eyml#^4+*5-M04;He49JWu&FAnEi7x#MN8{U
z(rWu%LQdu{B!|-GN8K;7HXHLnu%1saahS<28)0a2h3!hba|G2NQ+xZz*n^9}$U$#$
z^PZ@X697qTN%ww$8e$t>F@POkYDXIS%9MnWmXs;NT1v&gkXo(<Y$U#LLUN`&$*Sss
z?YOeCvOHvcqs|53?dKlAS%(#IpF&(|As0yl@zUD-hc!L}iMtR*uXPtv(gvXz&IeAT
z|JFHusqHCVR8CG_ToG~&pDZh4R17CwvDC3o1Ynng9g7AtQH5Kw%_M|4b=35WV7s>i
zT;D`Rq4Hz{wmNjaw3!&xsp>!}p(qQXDko>1r`s3$2lyomi1fwDqPgd8fW5;7R{w2X
z5%@ibgAlB!Zk8-?(aQzKvYJyhyM=BSL;wLR?V+)e$|5#0GA=489?gRJ4hna__(T8A
z0uH`}>0Bxcazo=Whj?x)#)s)i7?k)mMBeicsY?Ph4CQC%M2wZSrS;Xt36X0-@l)8S
z4=E5!R6ZlWu`yr4(BSYQG7^k#13o`L{R4Wo4GA1n3W{p0|C3i-;}h`GuT9HotA^Px
zP(!iPm}3wMile*J>-$k_Lj%*+j8$|1D;2&aiMQy5)3g1Fw|=-;IRr0G^-R2pKV@hl
zb!G|a$t1mt<mJzd6K5Fk^0+vqv(EPB=8le*mbP^fd5ZrRCa^7Ol<aIoOTSH=U-@LY
zH5*Q|O{yi5KU0_|GWS47oPxR&Xz%U)kFwOjGy!8d92^*3t*{0brUoD?Hii($D+cS9
z(@|Sl8p63y-``V^oO-&0O?dX^TFt;d&Xc+25$Y5##RIppIzGl)lfe^Y)AX^3I|gNR
znxzB|XRtDQe75(2!U*x<;&@QL2WV&aE#-(8r9=;i?3&%*Xv$R4ax&u+vl<aj6)`a_
zEo>7>6e_+0e+FWX_5gsg30%U*F*GCdG#{2f$U#`Yz@)J?JrxNBA$maJ#Kj#oePv0F
z5@@r)6Fp;ZF|II<VTZzD__ef1dfQ9L;^+r^N+2vr=)!LP8U#ekdy^(a3nzP5D(KMQ
zs|&~^W4c50*Z9@RzG0wWkNPMj<{0cH&9%U0{r35!&iRgW4hH-+A`CE?^6#sRSx&41
zHU9CY4+5zQX09(>$8}%&gz|L`u5R_WEt|6;$n*t53$6-+40tUS9|sg0!s>G!H-~9*
z{6#>8LM%9qS3Vg%BI4e7gnbuwt3u=<I1qiu%OAg1SvKi<N7W!5Vp%B<Lkrl-zhqEE
zXj>NX)kNX`D+I9mP)uF&OqkHqUeW6fG>xn6T}pxvv>VY!<G@rYFiY6P7Tym=YXmu~
zmT1<1jkr%dng<AdL4R}v7JY+%0s6m@U%o-Vg?GQ9Kl;dpDtZy9r4op^UD@FIu}l9S
zP3OQKSsQKJq+{E5(y?vZb~@?UW<?#NJGO1xwr$&XPJQRz^9!o>uBWQ@yVjgz4CKL=
zk%IAGOuDL@isuXsVPTGIRBg62dOUYUl-VwTOnP7f>_{xl2;z-t;D1$xmORKw9LIQR
z3+4_Bw`{)+7fny=)ll<N*^kyeU>oWZv`%k>e;GyrUk|_rd<);RFGM%rlKa3nkqxs^
z4uuQRlr%8PNUiOU0*;ADggBes!H0o~h0&+KZmHV(+vr&lOGQvo@2{+u!T7|gAdi{y
zRviGtn!|2Wxz7F5WajhV&Wjl-UUdl{wVp1HC5<%9Z4C=uzBKMUI%XSo%76&!vl(^e
zqGDL3a>NlGaBfzl@Wou3@(9`xJD}uFYx{~b-&G!V3lKN5Mn#eNhy6g>;tBDmRvl!0
zL3&@X(C<chW-V@|S6xIdxFM9S*|CA6w*e4#dJ#BR7Scr6Td%^3bx3-`X|vizO13nt
zA{yw-#v1Am-a%>igc+hE!;<%b-#~#jnM1+H%(bE=9d0d+(?W!6qTb8*hZr<^NK`!m
zm$XR?HY^oNvyWzU)cHv4jXTwvCd>h#iL(D4(eqSp#X5QnZD6lQOfo5$nZl2qGz9ob
zZ&oeqAyE2Lw71&a$W#RHv-Tze*ihqs3#K9mdys`&wi!vs+cmCY=!L*yo^wFiKQytn
zE2%6a9z|xR40}_Fco9IdJ~4T<Vj{||HnS8|d0Pwl;`Bis^U|maV0KnLnox*}jSP>0
z2f;1P;}Xn*dQqN!F~kYp%|WqT7zOY|t9cxUEQ)U}#k}%@e%7jTK~P|@BUDyY(9zRP
zOioV7MMcBcJSLZsYH(z6q{>Q3EcT2o3dY?eNz9m2K(?op$T+ap=4Bf6&x{RD&#z9Q
zj(ngqLdZpPL6D=*5~bmsT9$%>3ZYw9!NRf(QBfNZiuPs|9m0*50h=mNLx2Kuim8Zj
zcsRrd=Xy>~|1Pv5l8+MekLoBUXhBW9UzMqGff@K*E32!kQnoXr@swYO<fH^fEN}-{
z?nuyz=nPLw!I31ESI``P_h4%>-$TU)$R|=c0lj;zvt5WGKT-MOS7sHeZLN+`fG5B+
z3u{Z0p1<V9K8PQkQtQ8F`T)dY)}hq7pPp_pM!SykSimF`2e9=KkqZSvU^@}*(yOP8
zy(h`P4=^hbCI{xUfn39gph|3u)mph;IyLMlycl{+v~>RlTpgWWl3e3mAUHzVqYJ^z
zYOb#VZdWa%HGQqs8;ba3KmMqQD%fM?gf>!9GCiOnBI{4QqLza{AhsAtVM6#N6OxkX
zf&?0umCVWm(_k@d{T5`rGfOg<a*zicYMZ${z{aGa5fR|-Tpk$#6K*-O&|0GZd!OJP
zoDKi+bFOWVs;MFpUH0_$G?Dk@8xzHV?SU&8i>UAzyQw&hbu2tYiB<e$pc;TVhEQmt
zs=W^%D?LkH<p#8SzxRXKsuX0`|Cy3=G{Bt09u)jZif%h3U~pBn6w;Gwj)O$z3|!V2
zJ7by`3`jzHaj!FBzYpMmBR;3_vm_?oDqAX4Ml{v3JNU$z^n-QkU(9&tmkMiZ_z9jZ
z0wfcJDy9`9ADx-hDB+is-4u^T=eb4*9`E-H)mk%>{s9`9csT)<y%7-MBvnOcaeQbr
zyTSkDdbt0D%hB=z&*T0AGhhYL1FuD(?zopDy^giB$|$?WB)tK4OORzi358sam)!|I
z3zhl)*hfX{Ey~x`lM+G1Z9Qu&B|4_$U(}`^&^&Rs)1P2?%I_g!(=mv>K}kR-XE0G=
z8|6T(7hvdnj0?8bem<^&nvs^?Qdzi8;)e+DYP@kO->v;oM5<>YdnF&4ijy?dtnDBz
zi%G|Xy0{h&9>i1SKP4rwe>{8~L<ff!mLRZD7lA5cZJ1MCyW!z5Ft@k$zXt~)+H_8+
zLEvJ=#U|$%Nxb{NGSfAV%qGHNir83EumQZb+yH5;fPAXD9;v^mD@wecFQF1t5|AD_
zniA?7V)#BNi)^j|e7|ot3C{qc38Ls2zbU|ZaLOZ8w?rG%*pFrqya&x0MyB+C>>QJa
zwx{Q+L@OiAQDo<arxr$M=EMCS?gx(r;i#!|enfzSlf?N^0L}Ipr!Ln|>7ekHR5nLX
z>kOFR>+eu5t(NA>Vr9*j@!1m#X$SpfgXc{^3&h&t=&o>k`y2a~Ag$2R$n+f)2iAqe
z1i%3A8}PPK2Wr(a(ZG*m_j`cL3=Xm2Z-n{|>9#-)QK%gCS~><W{KjSZQQ-m`lKD%B
zAEoMUlAf-z;+8HcBCnzV-(#CNjIYz@1MmSBB-%&ngD+H-;Jg=Rxu@es;+$*?m3n@G
zE+q!<db1IWOvHHoaLU)Se{?{6{npD7$wa57)aPSF=4|bag|yu>QzE1YbyI@^kypv*
zM@H<o!1|>rK{C+C{ZGJ6qAbp5qhN4G%jN)_khy>2OgMFfo{5d`Cnj9UK&TgQ3Lub!
z#8)NC`1dhJqVf`?yFD41sHiBIuRZ}if~z_Pg*t*wp!fw?mcUm4NtWA?hm{X_dT@-W
zmd<Wf{`>4WH(qDPiWs3gbZuy3OAB^Gc(}AQ2On#Bo|#ctI4N+=rr{P-MJVvq8|9UV
zG*Q<8KjDXtvcHX|jVz5q3`55_0T_&_bJ%D$|3O975YAG<l?|gd963^sr_=7`@%jcu
z^6*Q<N8eW7Ma>lFll^b=VE%p2za`LZb9PrHs>Qjb+1bU##rgU9Wlj#!)a>$NMsjzn
zIby6HhSGEN)Y`HKkHMzCiWd+`QBqDYbycYuM?VF{o<IlsH#nP3J^c&<0>H(kP30wi
z<|Z&602XLjNk@o1J;PBq;9E}g81<cqf)ubs1q6KOnEm1Br<@bkl+T7)W7`-TBE+n0
zu7+9-`ZZB8FZj!(q`0su@TcWS-Hi$-^W5^H1e|A48j#UV!IS7)+gwZhK9J2_@Ti}M
zRY7ab-O|ZKwl=OJs+X3s4`6Aot!rRnW#?cSNJD(-f_x$n#e2?aISKnKB*^HF1x019
z`JbT35WuM`A_#ZHYO3xkXQTK3KV{<Jc7n0QS&5N}FzMi<_ZqLqg`D`LsxB0n%?eur
zn@akLD1?Uw_5`In$oU+X`^C8h5FW3%ypx7MwyO(cy=xf{fcAI|wRsMfX4n6Gn96bB
zs@4iPqR%bjct{PJ*VIz64(uuiVB-$Hth!kCj7XfC4K<fGa#GpCGDHHNj=#?`UP@*_
z7&U<wC3hr6Mn-!oYT|*H6-v2u#P+~c{wwCM|NhJ1si0*=m1_30A~0wjlcFm5h3XR+
zk_FL%w}Js-1$<eJ0qZzC4GCsqw@j=_Rq3T>HnzY0lc@Dd?;=L1E%CIuJ-KzsE`Zzu
zp{N--cTuQ1QtZr_g9O&6D4-9+wXCSc$RI4JX@Jduc>^CYDh3`20THNsdU9c!S09?n
zv&s8ebcN!4&snXla8f9Aj}7AHUR?DytT3qIf(eMxD*A@v-vgLqiy~{X(V;<O2#p+R
zQ*-X_Lf6cO1d*Dc21~@X6~%PrhPM_*GZ!jB|D<U#x2~$LqGk?Ccm2U;my<}icW6{q
zPEIqzjN)^V*f7&>WSNuezjQoc*m5A>VSa31O)}}O4nABDH)<IWEwa{eP%xkkR1Fz#
zDFJ>!J2QVzP*#~m5>!9g=r;Q(TZsf`S2rgYhw`rBRKT<Mu(;1L#&elVmE&r)8gVxh
z-3Nac26UQPDz0SHP<%7irRA0T&EaaGSM4_<OjZ!+*+#LU*BHU@LrCfb#Sk&0U<EJ)
zI0{$ooKCTazu-JL2~|Uwnb^ep?P@UHX9DVNj%T~x^v920{AX@W1+;;qxRM15&qypY
zGmwvm1orfVE?pa7y=Cf`dSN<T{qZ0K#7lr(dL-IMA<IoNOBDuj*M0t{dni8m(YinR
zjMRzuemX;iLob`%g_cU42P*R@ml`}eD{;<RvBa-kQbR3x#gH;9i*vkOScF5hIlw>!
zRmITbBtm0?*&ukUcCVAB^7_CO*~sRSl)gq-q$fEV99C|TER~H&8KwZtD&APy*piue
zJBT`ATW}KL?kIO8%lz@AbW|keM!$GVs$`w4#X3YW9evqW^Dkdv4(cMZ75z+ti;7vq
zJgJR0e;I}hjBnn33gTdhBIg=~-vC_E69z@oEIy2O_CP)6mCm}LG=2()ft9D)7x)$4
zAo8Ctaj>wxrWJj1O8xUovx~qC?UDv1f4DRZA3Hyga|5zZso>(w`*S~w1(Mu_xPN|8
zO<WL+j@Xjk`a0#|P?whDm;z$WB0>=l$dM|vyTZ+AG5YE#cuBc20x(KC1Nbp4NglH6
zBTG}gC5jj}+nR_ut?fkAYMMjSSr1Z+B-W@GcgR!`M3Uk5zW0B9Lox5IQ@a^!64Jel
zOt2+A*{8@w5;)x6+`;wj%KsdQ{{2x%eyyymua4tfH6TtArqF^2KXZNpeD1USnY$QW
z?=_b#WgzZZl-EHgCo4NT2~ZpSV(IxI+$so{(q~KpdNMOIJ|-m{L5sKI6`(zoR9K2~
zRW#ODL6%Z1Tb9Q4I$TstJ`@a<+cvwTrYH|;7DBhrE%y6t{Z*{!fR6ZkW^-^B(P(+O
zi1B5ZDc_0+HT;^F@^jN<;O~ohcR1FMD-$Y3Q(V~*RR4J*@xHxmP{0R=KiWn_ixyS1
z*@0Jz2dnD*$k+z>!{6UyWGm`Ito;KSm}CDiF~Id(l^@h;Gz$D>St!iZ`wSt&P;&lo
zA%IWU2FJjY=rXHHYY$Isv2t;6!(}Mgx*kFle@#q~RoPeoZD?`G1ZUT~reFsQ5Hu)$
z53|&Xr7tTbH#;DP)PMuqO5iNrmCb)@a2Q4!e(>-uD<o)}Sz-ng{~Oy&u9d`YQW-P0
z^LJ0UKk^VrvbMIknX4_osH#4@rz7Z8={`<b`3_$YzeT4$L4`a*gZiyvUi>B|nm${C
zaGyVFJ_RcP^d(v=ZA^#=-O0HAZz3uhrvK3un~0zg@`d4HB0!DZ6@J1dgYQEmZ!ezA
z7e7j!!m~Ka^av%r)c%q%(9MZRlh+wcp2E}85oJcJpfQ9v*4@#~$pw58L)68Nni<@g
z8l&J_ffTY7QgX%hQ}lp&@%NbN?>U7SSRb!x<it2tmH9bVj^?IejDOhI4KPt}i11=H
z2_7~nf)+VvWq{qx^`%)1>vw+>T#F95UO8`_UpmMNjs#?IydpHp3Myh2As~VdfR$P<
z2B{OFX?+Zyv!@t_E7RvK@MRma)K+n2`kf0Rw7q?kNN-MbFn3(c3l)`R=HPGzXE<8x
znYCGEs7rQZP{OCucqwTjOpb&UAs;d>KdqUTSaNycF96H~8gUA74JB+7Dn38#c`TS%
znHm=R%;u1kzC@TaYLKNO`3edRcj~Ak&^g3G$@4aZ$pP)QO}xh<XZqHFlv17YIfanm
zO-5^OkA<gtvJUN_5-1H%<e=4#3yFzHcx7>$w|Fk6{uJ+Le1w`b&sO^k^dA2yP|^t_
zX&8v+0Kloq1a83m{P_?iM=$r#2op|e6w3-`n_P4*<ijiFC-=7{9#i)l`E{1-22s$D
zZ!$TT{@W#*{!*{wLo|O+l&|73aln)ETZHkICNM&RA(zGzPxELho8a`?%GSc_&lbuI
z)4hA+oMsZb(XYJBG7E894FMPmePGlM8h}OX3&5}xMkLMl$Lw)o2z+LbfmqZN8iRqk
zxg{2>BbGflHc4fQNR@>yskaLtY?-9%0sW-ZaF|V->IX}1$iUayk`We{&1Tj_pn%cn
zE$Zwg?joTSB&YryRxSWP5M1V=8o*Xoku)5g;aMR6V`c}r5^FhjqiDvpUmI4U`$~0W
z1EAXm7C8gSvDJZLjK+^1e%Q31*)LjXt%C{h7ae3XwYK))xSt2N)03fQIZT;;q{U%*
zRf)O|BBp<v>!V?07?nVwH;9XrCo)+5(@syo+NM!IY|#AP{-`wt&jmxYfRhd7HEV#w
zH@bv%{Cl4_F#C2D-i_Hz9?d$_BooYn1(>uGaYClCkztSB7cIqr-#L;PJoWtBhmhdG
zc5#x;Y5U0}@ay$R-C|JkrB!P$b88Y$SP`##t|=1PBefPzru0O&^+hreo}_3M88-r-
z4?aB=2Hyjml8FIzq*kE`bLb7cQ^TPxtRjM8Z2L$4k4<K?*lhHb{iD>WL5Z!;dB6g#
zy0|vi7B@DU7LoU%@9q3I$Kpb<A7kymeS+}t4%UD?Twl?QFC~`S4q_sBp1K#<%og_V
zJ4MlnDdQiX;6L0&v>p1?@-AlCo17egaBi6}#Qzn1&Glv#l+Bzhvb*2F$E3#pOHRed
z3uNqR2p8QNk6l79HlvyirHbPOfI6;7+35RZ&_S2Fg;3H^naP4IiR%1T@H>!?%0@!I
zLV2?|9_sQ4_y=>pNA|I~^V6TCbX<b@Qo61Xp8O-2y%${$)7;DqUf+j+P0S}K<86Oi
zPO8Z`HfWR@YIguNSA^R9-`@n25th84TEHJKX=TFLvdNv}U6AcRvxc{x0O(CehK&sK
zyelI~CDNO?J0ob@>mxjT3_{eC<ICMMQ=|Y-_?)?L?z`GtKMyXIfR>-b6#i{|8u5j`
zqTmqb6UeBjm?;0m{shJZlw!2ZqR4yKM22!O2T{UxIpYpJZ3)fA#)L<wV-TUC?8Pp1
zWflgt8(K_UEI@dYpD*bt0?-uCb3QY$ONbj^9VRJ3V<DSR9;jhL**%ovyCkIJKwpt*
z#~HuqvtK7oVZr~fg^@;t$&QZD{siNdp6;1}UZlZH($LUYTH@f~XwvAhnpsjB#NExP
zh#!jna9Zc#W21>mEJ{tqMs#ZMs=?PwSZjX+4-EzyH2z2$ahR&P09;(;;3-8zL%@y4
z+5(ayGJ@I4x*Br|SQnHM&<DCm5()|$;8)vneIlY7a?;0{&F~4GLwvHCKP2=EgN?w-
z8z#hCBUx1A(SDJMg7yDGJqDsf+MGTkIv`5JXL|*d@OM1*h8pP1#7KWEabMI#GwqWb
zg+S+5y~DadcSs`A1B$yImOmctDK~kk6iAO^%+S>wo(KBY2rpXrJAFUFb`5C2dy#Qr
zMTHgIG=6*B$%7nrbo;bYF=x^BFp$nuvPKtCcv?J?D&rUxueW(X1?LseQ}A0mDk#|5
z85sC|#fhJe)|37)?irNQ1gkR#C6VFMBcL?R6_;_f8fD0g0aRsGAu=!K$ib^B;L6Gp
zDi=ABoh&H{d2D1nh|6l&tu%Rlp7l|1!?CKc*5AWv*stj5wfWAU9df7Mqviwo`|yjV
z%DK|*IlG>|8g2{*gbFL8GXg>#z83Yuvcbu|XH3Umf<mDI-1>^Xa66~@nM^bpe2HL`
zKYr&qMJNd%0)VsRm`Fu%-{&a!l!vK_i7^?`NMh2J>2ZjZVALP@6Y!na#-WJd4Q#?d
z(bCpm(??5uw!=YAHa7(ch2A#Xb@vPEtyXKz(JOffxOhigD9xW9m|&DCgSpF+-a*Dr
zl?!qRU~T0lF|k%$zBHlcR`KaV>5G9YiE%Nl`HEUTfU~?ngh4Z)j^N!53(Hmqst%S{
zNF%q%i73U0VOsw$`jusVF?HYQUnq*uGOg93S~-#}%7jalWx;GFqVFFrAWlEK5bhrA
zi##X!Sji96U85?MlJG&+w(kAwv$`uB4$~f~IfQbmGn<(R(w4Rmi`HOG=8IXH&WuBG
zz;}sR0g2(WiTM<{xf#PF-XW9|(}wN&9aq#X!5kdy&G;LTSI0?bdwr}g;ckL0k$b1E
zeOMM7)OFd_W~R2*=GP|&duJiGF*lhog~_sN)J973^&L`?%DmGv0~3XK?wagb+@5)_
zTARB;hu+UbB|JU_Umo}2MQ>?AX6^1a`q}BIfGfMApF_eHVaD~^W9N>j+@ts7;2Kvr
zJL%TB)rhK|@xy*Jd}_zR)J0)lXX_w}Uq`cQDv~icEyCC2+`h&Jy=jl>O?KPrZPguh
zb$7%lX6+Fi6?$>}g4|=%vugSh4r>n3P5{`9A{1K=jUtj^%tTwJ{s?uoWlm?f%=k_>
zfTpXp;rE|_$Ych?R<F+eoudnEPB0pdq|*4^J3qduS*G?(%<d3j9Zv)@JkA}yHwbV@
z;M^h&Og$n9N433%iuu<!c{YXoI)eO|LFc*Bd225aC6zU-@rRy@M<GjK-ucOA`ZQ)~
zU(&j4jPsBZV%1980e;nY6iE<LmbJecU}loZKnj(Im-J@%OM>4-3L@I%!9u0@DmxNs
zzB?7y2vpaAKjk4x9LL74ca47VeGG}PtkT}bQqc1b9jEfYSVP1T#?e&dQzNGm>Mr*4
z;fS{=SIQ~1v=39V*f2J#Cx7uwt8BAov!U1B$6)!gy`S&%`t`D|O!5$S>3s~KKsNke
z!W;W0Q!dSOZ|?5DS!Tyum|Vi8|1x4!(b!IQU);*LoQb;FUEQiuGG9kvlgHPex98=@
zvDZjj9WotGoOWAn(FTA1OEL23jC1;D<UW|=)$AAFO82bp!m$^Ce^Ygzng<T~vJq@s
zi~_GI<0N}3R-iaF-`&Dz4bb<Ih}U$@t+c+f=Gb<;(UXW5<f3=VyQt#sa2<o-c~8nu
z&qm(b-AA{I()YtzX_cxMQNomg4fCO}JyB6_DQy$T?`?Oyf7;1xf4&ktDP2t?I4mLI
ziV-j>c@>9rj)W9|E-uK;HSVfPNg^8%Z9dXA{(w<`eF=edzckhj0ciLN`7A|TLZy4#
z16zSN3?c;d&JLM^Kza#Zibv+<sldGJ+szY&^9NnRtKjX@{3RLYSxC<v*>aq5ujq~_
z{WBwI7Tuzs@rDN)nW5UR5VRCozIjjOQIZ&GI<FSqxQx_plg%Z4Gj-Q!g5tL6C&Lb}
zUpE2-oD5G{HdW<o0L>07VUe89u0RknMnbm~QEY!4)Ss1Mx@olonA~w}P$bpX@R=e<
z2X-sE+l!m?-c3k^zWZB?_oA2V;*|5JZ1m#7<(MEhKi@Q}KtYWZclR$3!yA!q%xGr+
zAZL-#ky0|y(33$eKg{GRDbI2M8;p;L5h;mT-F}+tzm^N%0l)&4NC`0BvvZM$v2m=J
zMY<IjGM9L29K(Z}jeS_WF1hjJAd{Hi^P9Z~(6WdG=HaADwuu%l3o&6Myp6-dI>*K<
ztSv99JKeJ~(YLlrMgDv0V{m-gmS#)U@x(&)9j7nP`DRO$$IkHd+RnbL{O5XtAY<R8
zn)m}Pt}0hKV3^((nXb2MN2s`1*F%kjq15seC$-IVUZ4+z@M0UuW;7iwOhMBv1Rdh4
zHy1v(3%&Ob4oK|r7Vu1MV`F1`d65HQYHj(upt!OsPijI5C$&$A;G>lP>P1u}{a>?}
zwdG@NOqVs*`;<!)p`ePsW6qt?V{Sitqbq{Lnd@T~0MCMEJB8rbtZW2XH$TXZNZGZB
zP~Y5-<R*b1GyL;JxnKIc^gBXsW5zjPRUn6Eb`m~sN!V~>sy7u|EFvm9Jv9hOGxDY_
zi^aFU<a562`wz~DxaBR>J}ubgy+7Al^;DF<I7nOn`&GwUnptD5MifRQ%Vmt%W|s;r
z!%GApky4~E%~o?CEpOOnX|KKd^?7)RfdSQoVLI@-ax!u=C-2lpAC;M&JTsH$4^<a}
zq;uxh6G~#|8PpemcX(=JZf52GQzIqy83Dd80#|vK`L<qFg%$JYfKhV0yP~b@QEfXl
zCMhIstS-)0NhT@bYPG%k&cRpJdwDg$Ix`f|#3+kcH*dm)JkMO<8=QnmN7pv>p>lw;
zh@ZLmO>omtia9E_1Az&9GwE_O7xK!W2$Yo@euKCmhluyA7^~dzohaB+uQgr+V@kui
zJZ)anh9WiFUC!6L`^7GNEeN!r{Bd>Bo};}Akg$fc{{~EDnNcj<@9$=?_SEjPwFUrx
z%M_E<*xlg6WXB~74~C-gICHhK*m)-ie1e(Oo%@t#t#neM{WT?e-aT7+w9*3(=_r?m
z$juMUrxm5c0^vX!x`N1lp{dTXvS86o7-Xb3n`-w)s9P`DZ*(+zxUU9?x}JctEfWdU
zG<%G<VbE$;8noMehRCP8C#e(J^7{g=#*7sEv)EH-2`RH{unX5E$7CX4oqG1(S0Wu!
zdeBDV|9t#iubuvsOaR`APh^y2WwpgQ&e|GRr~3y4Xqeat*_2E+Dzhu#hUYaEuHN}B
ziGfkC)$aZ&u5PSf5>IzsF*Bh5^Qf}7{h5dl=p}_9hy_|ar=K%;l=VAE$pF9*GlpFr
zv@h4J(f@@BphF0Jh`=4M`vszzMsqQK&oFb#D{0Ixa1;MFP+-7tpg6WEEhddFb5poD
zihQNNaOjnZTKyI&#|q#i=h#41WodYhmY&o#RB7m9;&it25d-2FNq2`sbGB3(Iz5)J
zHVeIH(zZ04x_{VS079+L&texft~OiEY>f#!aJXAvWs?{TyYnv`=1cFFi3}|GYbZEr
zW)Q0`!X4>(!k$oM&~N!T*iwsBvHT6A_!g@^YXa^)cSe^M#QGNG(^BSLU7jImQdXVk
z>+9}kuxP0DTCEV?t_C!>>$~q}{)H4ZO|kO(equ596UHh503Qc9KZ{O0o=&zd&yyJ+
zTsbfFySaLa2%7so4A&yEIHGSN_jms&hv5W%_+WeU&=U!>qqT}1G)`prki{SENXhiM
zvug22)liw?P7*K8inuCC(wysXy|(p2M+XkT+ga+r=NxoQV#vED7K&RZF=Sw6$NlSc
zc^tYd#2_Du2JG~}VpbqLN}k6oEa40Ilz0htP<g_s%)&|!(USi<ealQ7R(Jx6wI59{
zt1#~KH*nPKc&G_DtkL*W_RtQDsek!0->}`0m5v=>2T(b`jt1|2%K?cq){=_fH7l$4
zM~oBtO-`4TE>5{Y*4JI1hfpVsfV_K`=hb_#<Z6RZ00ANWWAQZd8KW|S{P4JW?A3Ow
z=jENydav#HQ3wwna!8foL0mNB*sUuUFz9fBOffPoTZU%>fiCkK@H4igqbNMn+yVOM
zL||ax4=`W8jLXUicvtMYI%3X0tzHi<Yz#TBlbJQ75LqUub1u-+I);vWXRYzkp-*<Y
z?QFEo0EbVG5=~x3*T)Qg!0ue)xh4YwaCYtHo@r%oSy9ce)(W^rw2@3lM?=LWM1jF_
zyTuuAzTjQsc6h!ty1uPLh%|><o9oAOLB1ptwdT}S1vW#B2{kDBz~9)PZ^;CL{kWFd
z8cLVc5DjWB7K1}F?6od39xf>USabHq7&wCiYV52m$<|aleUBOSzz4!NKbiaRNxsh5
z`#vt$4RLOd4)3XuuD4r#^ULj?&nB>uR6}~slLx8K3KNnNh({t_+&B(M?u72g=SE>(
z$`PO3+&Gq!j($^0guDeqaNlDOWfFGYD`ld+J^6yimWj&wNzX5CXKfO4ZEESa6*}|-
zVtf}7sM?_Haw85$>yz!bP%hJWOB}Axd1gHd$|?#h^o129ob*z%)7wL-ra}Wzu_|dw
zFI!%21d*T=)a-%0S*PO4>C+r5*oq=>N^HXyy%0!W!O(w1gf|B%KWGUFXW5wQpP{4V
z<i-t}-0l(c)I{xh!XKk`gDgDr;jdN!&L*DSgG$}r&+Also9oMr*@C`zC%vyn5%Q5}
zpB0Pfh2qe%D~xbs2AOJk5Id{i@O-VdxZ2vf`?AL9OB%tk#O%d#vEwIY(+QZn`x#?w
zRxE!_CNn=sN0n_r5k2qyJ*{bZ?7nXAwi36{B#{zq1};mVSU(-I_<ip79;>^7s@1na
z)s9sC6aH1F2c_35TRaW=@llyff=K(5-9Q23-UwTa;b}eJk1t`sc)+Bmb0Bo9DfL{q
z3cAgCNeJ_PW{!$&7^7rX?TKJ4@7Js$Q?)ZLZ?b<%25=>vNEXQNcNGYH!gEi!sxISS
z3Y2N|6C&_3alcZLe3k7eM=Jq<U33HRp<jT{%>15Fcj;r5ucN()QQ`lpK?vnYX5+Ai
zdjNW9K?Q$GDk5T-wLRi?)^nD7q^ye)h`GE}`Q5Pa&}_|UNbHhsZT{<%EBMXQ-0pr|
zx&1!DV1A02-qdjZeG{vRRFy5bs8-+I($LXXbvtmn6)?VRzX@@$-3jo0+&kTe`mZSM
zQ&oPcB`<#JcBbQIV1?TfYfi7#VshTK_T<c&-PH}Ks<pWc)KcUchG}%oBNv9^_7yj+
zC=~OEi6wqEoZxte-8mKi=0-(aZ3bl^=5_e_AK<(qvm=lc?Fg9B5)wsRw2HQ+YM<8c
z07p3aFWkG;{(S>Lp>ub#2E23jbPM+g`hGa%R9daqdG0bm7LYDraTJ9;WdQ9Wff?|S
zS+Tu_zd#gE6HoDQQ&UoPb>=g%CE-j!-0+L!EYo@4hi8KRGCiAIqa}*aqNjNcm{zX3
zcU*4~RTFKtt@!S(<oaSPHa=2EU-GO2PS@mx2s!G^Eyw`vfakS8S206miQew$wyO<(
zmuDCdvdOu0*o!XZ8usg&^FBv&na9`HnYPsW|1gh0bedWAjKajM@Bc-};tn_m^h4Vb
zFi+y`!DY`W9SF*Hk>YWZBS}g8b6Ad+k<(9sYG&|8T3Fz05()Q1G2|5j`{)(*$rJqE
zk^1JlnMMKJV67uz<f~4hp)Jl9y=#|#EB+At6D_GEhnrRse8RWsl^^tbH{dRywzM@d
zI<qq<KfXSacHaSlM8>&!ra+uV*)N9jZ0<)zpwpU0*6!$PdD1b1R5}_K$^lw>Yu<~}
zdr!aM7GJCL<HqUI+gZQlSuA^Fo@o;lKTN|>cBdEMb?>#lSdmcB%i(-(RdrIb{0ZZF
zrPlYo$#xwWh@}<~61TlC{Q@#oy1Lg@H9N>hr$tbPE*-pN(Bri}aF=gk%V1Sb4vYi1
z9n>;SzE#5*ePoF4s=v}4q%$eyP8hNNsVISwEc=fPEQpV5*sioWUkCI*PM)w|bGw6H
zIvfBRhqH~SR9m|tzm_?xZ%6>O`!`jI+Kycsz;;KWfSgSz9G|fqP&Fx@+NY|`zi&RG
zFAP&eS}T<ZY~ZK27jy29S&v1f4pr?sC)+TLL?Jn`XrSz)sxmtwW_3Co##G{IU>%s7
z7f}Ee56Y_xzL%p!_Su`=p0^jJ8h38pjetw=wRgy>&9>*4!TaOsY<};j2O2N-DWLZg
zf33UT^f{}uq(ac4rmU>(ZDzNhoqd>uw&nid@txNk5sN;?bsh~33oH5W--Lv?zO8zt
z^_fiGx9uL=9~tdT-MycKJxIb|tYW%jl%P#`ILKZOw+QL%_PcY9;w=c?6@~e5R)7cE
zWyk0*<RFCu7e<<uQi$*69<=nm!l|`QYs^W+x?**xE=UhYZ3=O2%L-V*f{gsYr^7^b
zHPAmoa4IJwjKnKJOfkRS!qw&GIjB9Ql^>_n;m|%VUCdrh8!CNGj<L|yFShHghnJ$Q
zH8E}IxIHcw{`4Myd%l0{?lcC}7X$sLI=VV(xh|oq*K?1(P|z+9_*vZ^w|l+O0a;J`
zq2unCH032cddo8lYfEl>V&2}8lEXL;`R;_B9`AlPYKub?OuXs_0XU>4aWl}xuj&m^
z>@JC^x$#m@n4T{rUHC7?(d8--H>L?jJz`C|MrtY&TV?bVP>2NFOFnyQ?tpEOpW5LZ
z7ic~B^sXQTxnF5QnbezKxG&q(JB-<7>+J%Z+!KRqN}|FE$%&_Mp8hoycd)rC_s%`4
z^4=<Ja~ngwIBKVXB4ier{OJt=?-d#Ji^VY)2ALT9mPK@I^CR5k6!yQgML+-bFD4$E
zdCMT;TM8iPJ^=iDUk|=6tpQ-Z<_hjj&5dk?<<-1fZgs`qca!Z4P0w$W`@Q4lyy;f7
z4rui-6#iHYde}maw|j%-od1u4-2U9mi&T+`WkfkVCOvbu(4c-m{@Xz`;{9Io10*8!
z70Gn`J4s8?^`a&}FFaGyVPoUZ_!8E2wPhbK-$6}~d-&WX1aS<2m<rfI{3QVXtOyo6
zM?7*>l~KjG-ayU{OWWD!F^W}Jak}ez;|jGSn5!?xiftY?j@(6t$UsOeMj&pL>Y(ru
zuG&k=<ZCN_TzS=||L0!hqH%DMuJadG1w$!6u(<arW-Y6_U)wZq&bZa_{4fdF>h$}<
z=866E{om5%GGKXc<LjcQu{Ts{sl(^<Xn8wpyWQ<eO;-Q#Dxx2uviO~0)5kxsCC%pp
z{c~j99ayVcG-zTORW)?^K<9fbEV9naC#4#KIy{7q=#W8*_?r|@8i#BM#Rv0g_)W7|
znmNP>;YLSNKZjYp((!I2I;`E$ZP>pFdpGDOUvb?`1aL9$0)HvjA0BJQG@S^v#H2%h
zR_~FeDfvlbv{MY&ON&KsGZr`hv4-m~zq&Zj2n{}BiGgc}5o)oKfpfvrL2sTE`Y$R*
zlxgh^sVSG8hsi(5T5xQ`e$)Hm;<5ev<LPPf9_WSNYV$kQb2mS)C0jgPba?l<x|@85
zUc>@wUpn4D1*-|S@VVyQ9*isxXMqjX^Jl}4-mq`v+ppj2S!Rw?XB~pk=`v;dJ{x`c
zV7x_laI!`aDU2Q>0LW)&fJB|(5HMF@4GbPl%1rC(2h04e3Rri))rWig4G46xun&po
z1|uwfTyDdoZLZei;#~n#SJr}aSsB^?J;3z<U?a9y{nKfXp?4>G?(qO@WA>((R@i`a
zL^S1ckNoj^hX@IdPwTfR551B*JNu-nawk}ABt=L}uJei^p^=5z0rdj^Iy`T_7W9kp
zr&?nJUTT-yt={u!(@;KmI4aK~7Q=SS{mp5)`?aAUrKN<cd}e2|9luXN7kj{~H?aSD
zqhK{J?TgpY+*c9Ofc%U;L_K#fs?};ws>kNEd;SWj+Y=-bRE^R$pkbf6l7dhXUaXC&
z*S!AM^8L0x?C%%F2k}9bL&f7BOSIMPQRer(6gUAc$1K^Si2Egl@U)xe-y~RagnVNv
zENLz(sQd|ChbTg8AcgwUo%DD9*QyjCq~n84nm_pj;{3T;4Xav*X?$BjF$TN~BYVPQ
zl5FQ1rKj8b{_Hhg^7(qJFQY6bX@592*FtF4>+E6>?}W?a`{=Y24{WB>`s_piaQM`J
zJkt+;Aqc@oVT%L*W;tdtiFVK0<AIVh5#8!hbbkv6T$lvB2=v4bOt`3VL(>AzRy`z(
z=nOht-OUKrYV1I+22h@u)3U*`z8uepkguk9MiH{?t2f&nPd~U)_%sqFdXNA!V9hlt
z+T68soHW15NQPPr@};2<Jl@&d3^o$jYu%jO{v)nCJ4?FqonYki_?6X^wJL4|vbS^6
zPs_nO)-NP2yZ_js)qQ7FVg`VFc-Uw2c{v;di9|0s_O>=$PA^7-U2|hVS_m8`VTTt|
z>h~gCXR!|Ne@{*VFE=<3EGQ4z;8}!?@VBI?9UJKT;=+ig?gQd9iIf4=Q)i)|AJ`KT
zyL*XqOwI_YojEln4R=!W|2^CP+A!?&d@m*Vxc2|vQM{#?1frG!RnJXo$ra<W(NB(j
zq&3hl#?vasU>ulkeFKW>5YTWeb8%6+m~8~YaH;#Pp8TqK=H}*9drcv3VWyBMJlQ7=
zg>b4Sv$ek}N{x2#)9PHT&BkA21HSh;^AG~2Ue`IRNKd?Zbk7_O{SKG-L(|$A5Cs6h
zE2=C9^KBV-J--3~JqOYu*F@PMex%JQb-A7P3mw|ITQS5DV~k=nbhCA|PIGZ49`HNo
zXF%nU{ulLGd<KP!lIw$Biyzj0jNtch`*V%|u(@Ewvj6+eq;+z8j3#EaK?dq6#9dY#
z^vmwFrx6Xh!xhROb8xE}*gK~>7MI~5JINggMG&OA`y+r`6@vnz8%NSn(a_pY2b7pP
z3#-Y&5GW6xfaRP#638#1`&Pf7*gg9K;A6^0^KHKOYl)GYwyscawmP-g7PyP_oy*xB
zp%dR<JJI6~nsBE#`{s-WdsNRfJJMk31vVn=`reeI9<>e5jt$MVES$_NGmA>fAhq6A
z)#n=sR1$!k50o|u(m33sD`Oh&a~UOJr@G@lpdxt0y4i5s_ov}I@6PtW5>W(@=UXPL
znQi6f?h*%Km6Ub@84F<wlNU6B7Z@R9S_XPSRgAxyX{9AS4u1KeXknuaOO_Wt#8|&A
zc0o)`h&>8rX)t4iJ=tSx=%fbMdJE8J45IY<jz|GJ>2Fh_AUDa#H?hz=0PpAXb<US8
zn76y2=BY*^n)Sdg$&-u>*Fs4J1Fr`{7Jx_Vsdp5WVz}s?YLDVk@_gfc<T$ln1`jQ|
zm^1WHWPGTh#oz8&DvK>`obw)J49556hqF+K_b`-vD!E>I1y5h^Z*6VbCe0u0H+_q`
z&sl)K%XZfSsU-l+7k5!3$;YoqDx{scHy``+UH<f>fZq`8Vw^s0f>hv0m^ks8q>$T*
ziHSIqj6)v`VH<fiFrY}y@)`=j6&lBp1XH1$b=vHw54*>o9Re0N@gDn93m)&Q4elRr
zvddCCD0d+IeS}{3Gik-Re-VY9?>LAoPJ!pYtT98PnF*6NpKojr2L>k?`Kk5bL%B`m
zEV$G#b0H5sr@e|4U>MMlk&`##ALy;V0}=CYQVlg#C3a^x#nJ7C>aPdwGui$1B}LnZ
zY^bkI<c;kP&(Ak@rD`REh4n=bhvm1Ah87_zkqRtO#n{d-eWP*lxXJ2*G6@LZT1CM2
zFOyYKOb|i_bNPzQ&f0Q#y+5$G&h?5AUbMEf(TBG=-d20KG5(D97nkQ`Ipzzu(653Z
z@7r6(53_UKeq}`-s~gBr^zSnZeQ|GF?bkT%efQ>yM1o!~JA~?cfFCcVB0ql{FsIVW
za$wM<;i2tVK}2D7*R&)|Y>1~?($oXY#Y3UIS8%My!W|S}Bi>(je+VYwAPJONtiMf;
z=``NYgu|Vrw7u`#-fP;A0;RP(U)(!_dRNtk-CmCmy_tB)@6&Fs`1IRdUbh>-rWc&A
zYhQ*nmq8DggBP5<of7jv6$)-znvTxciX-~v78!Q@x<zlj9GU+za0LD6+7Fl#0`?k+
z9K0o!?D?~6f-~w$@gToErnRyAKn_AdzK^kifAZ|SOX`Yp!xUb#NoVuT+w*{j_3Dy9
z&nOn{z=HepVt;cK^=3q=u}!^J>-{?jBO>^y<dT9{?mu`-=nRwE(4-~3DVP)Nx9h(Y
z0MCx`+%IBg35_*O3rBbRoq*=6ElHxc!E?pm_x<36&v#NQhAN#`Q%h-`x*BvDXuHv=
zh$cRw`m0|IHxb8Z=%^@?v<VvP#F_q7F=kN#aziGA_!)XX@>p2edsfD$=EMwDA<D6x
zRD<2(f|&e%Ci}N^lvUA0@mJAvFSvCP8OJPrw06NIS8d^Ud|OEQDF7Hgw*l?*&Mxl8
zIw|>0U==U<{O%|H{zeBIZGI3Q#WNt@Z$GX<p078B?PZniGTH;sBWvnPz17z4Y@uX1
z=*H#XA8SZLkm6bIH8AM3JBWSlRQL|<AAP~A!x*Or`pC#DD{M{nv65T!TFM?jfQ-Hy
z4snq0bxEafZr^vN5`ftk>Qw)5Ea8UeC~}$eluH9X?~k9Gkt%s$dJ_n2y-dlrb2kCQ
z=6og5K?aox32fAwBV&PMpk>FDF@siGOrNZYNhOeERAx+0EbNe*Dv7CoVVqP`iIsma
zsbZ<-)?-zzXlyy0?F|ONd29e9yMUd$MZD%KCJx)Z8cE$ZBY>gm;L%5`aS5uw_NcRa
z_`75eHpcrhBT~jW1kX`VA#h>WtLt#u*Sf`89D89<&onhGqp!BQ!fo-Cr#`4oah_Z-
z48fIJhNH#f89FqnU&H3EEb)^g<*|-)>2K<7?1+2zu<yZ0xXs62gbroX)h<ozxr{Y0
z<kkK_u+@t32_RHt^5xrfN&%v`gn}8$uMbSGPyb@LWLYEen#Q2pXIIat6Wa=<XvPqP
zkHadzz)eg{Ok8|$U|>LcGADe0gvL|q3_9J?fOvn5ni{yfiQ^f>J!2bB8Uv<H@1!Td
zRb2nWSm-`2H9Q%#d%f-neYe*0Gu?S~0e1GxQXoYE=U<cdQ9(X8AK`0r2(P8erLTC;
z3$BP80LkVOJH6gtb5v=%>%5BDr~J#Y>AscGkL=(R0aL$mi>5#wPi2)TwCWh_4HBB8
z2|d(!-$RC{brWF52@c&)c9yBqqP8JC<T_FZXvl*SUr<@#nvys(x}vIg4Lbea&xg;`
zYx92rHfU<32MobNzbH~)^ENLzyq@3NW|t+bJ3AUY&fnBMQ=NfY;0yUi>9Wy*+5V|3
z2LdE-E-cO;7@5yMe*6dwB!@EJ^Judb&5QZcg0|~4J2d}UZsO|1?<)#fdund+kmKX8
zDh9fIe~($cq2g6?yYNR>srWhj{8C`k73l5&C@R`B=94I5Jp_Rj;~78eXtj7w?<QrH
zBl#6Jzku$+UZ3oeNyf4_HMnM6SlL$%mtp@`VJP&kD{Ays+UUTy_)4gllty%ARmIYr
z0iRwA8+d4SMl~Vp5BxAD{!o-i+{S%Yj`tiucclk@O;zugF3aq$S|MBVaUU*=%bWmr
zm{$Z_w_ypHJ$RyDcY~2w(*c%2(`+66K*UaUdDTk2^Eo6j`=o3bG|0NTnszwMU5#QY
zlh2OhaZD;Q4G~Mp#Hd1B{NH`Q0bLT1IrTmZLm}@#@F<eNp48Sv6s<PnpJc9ry;PNU
zXQ=4ewKktX$`LN_d)1yi?B#?$l_J2+&CSK7p?SX*c$&=a_Wg$HrS*tl+{UvWljTUp
zFR|u!-aX3aWFqv1`wRshRh!rsmo?Nc=*vqgL{2!_2)SGAkCPf9S}mA=bqD_SFIhe1
zrscrA7SyYw^mfb15BSj?oR?Hv=`EzD;$x$xRnr{{)nE0}4iYO_T6`y6iVNU<kpRv{
z=zf-i2~e}yIedN<=Y1?UvkgY{5ZIrqfgGiTJiL&waRZecbKA~!jlg6EW4M=&laJ|N
zED{hJ1Om<q6E?k_ROnM-E30&fg&nCwUOE`56X%;7SmYeI$gl`#IJjLm&V{lZi3^d)
z^v@_L(bVEPSQ?RM(tIQ9Qvhh4yf4w^64>#~lTRK()nP2>LUYCiK@TUhY3<iZuPyOG
zhDAlWz}m*D^X^*aaEZ|$+6-qP9{O(7uc6bHaZA0;9WdB?R<|c#-eZYfME7fd56C*1
z!=%@4b@V<BNQ+y9+xM+(B<CicWdZ|xmFmmez`43Kb#Q;>1yMZW1ONwTI!RZ+A1ZX4
zHT^{oFQXf7;rQ>z&EXnWDGjba;w`L@2T}_p+3jF8hHW$Rt7@SJ&y&`$cbHSa{TpTF
z;;aB8%oNgvDpfR*z|D&rBJFD69r+>0FjH0#o(BVA!aK}<VoVax2~4048(wu-6pL2N
zE6=_$Oaox<h|S~-fSEOWIv$C*KqxrtdUaO`HBGeL?rOUOJi(#S%Mat7Ss`f>Ral2`
z?%ke0zoa1=&TO?r+%?v35`#ECQu09M*aax~N;-?uXISYjRqNE-?H}b`0%b0`%XFhM
zF)()xz<oD-tgMXg-)5i1x5~l7L-j_A@;Vd#Nwr`~=YS7DX=^@bzO#8rxa&$6(oOIe
zX!?zuuXp##L*aQO?U}-ZLF;pz-3I*T{^&)tNLJq`HzIL%t_xz!Qz&&|fB3~~Jym)P
z{tA%x<>@xmv1$}`Wo2(>m6H#dXBNl=1KM$7Y~7$VZv<Znb*A3qz8Wk51&!N#71J8!
z;2NraTXYMMomdc#WuI8z5)*ZU90rdV8|kI5JV#E*=x)s`-Iv=Qv<^bb>hybZxOp3N
zHs{gB1>TY}#Eb2AJlo^qcY8jM*Q~CT;-C};A<6xG-etxo*GA4og1}$4KsB029(sPa
ziQ^gF4pE75Y_I3)pY&#o<cr=lDGFNFJUxWN;#dHLI7Y%6?|UFiKG?~Rvy!-jjB7AH
z(<XeApOjU(&Y0Dk%5E*UYTwbJDXj0sc^Sc(zZXzf=}oU|rW|`r053aVzFZW_ehbU<
zz{G~pbCUXNs`GDg$oDp0o%FMy4u3j2cR~fZOdLt1E5K#*OKM9@`T%lWT%ZmThBwQe
z!!RH~As*##->FU;#AIGO>4Q&4V7TN&VB>qtbASt+(!QVf;{hS_<hH|sl9~2t)upbR
z=2AA7_u&>?9tqMp0mez}E%f5`^@zk(kZ$s{dt?56N7Mp`Y-V<=e+rdB0{-E-Q657-
zW3lx)Ub3>13vm07Aw`K_N~`^k@T!Dt^a(KUq2UxaG>B3aFBgS9Z?G*uy~9nxM(1n7
z3fp{LF0i!a?Rf(;S@@b`<t_T#iTn=z$|5yBz_{D}`HCRn06s&|`$$a^EXC@$fGUT6
zYY@@?mW>9t&@{MfUl4*2X*AGe1)Xa*cWl^|P$}gU4=7xpBU?f95M)Y~Xh#{rX#<Fx
zRjztp&_R4$S7-BMcZ9gvVh4%MBwrV5Ctq)Odc7Il3>|)Lbvb{6CErZ!T=w$z@cH??
zv#f1f(YoaicJaBBXiW|bQc+S&%gaw^5bU0t7F7~QG$sV)xRAzvkDb}dss>&*3=|tD
zi>4IR0#9J<95eH?D%mvh8d=+XHvk*DhNj~2vo(mRDFa=vnySNv*K+sl&o|>qiAGZc
zwD5OX(m%uZD8T7MtMB>JWcq(lV!zL`#PQ^u*MKA*9^!y9r6#bmrxs*JEN@yget8T}
zX+Ch~%yeUB7P`Wi?swc@QU|MCN{NinCJn@rv;&HhJN>4(TAs|5^HSPk01iY;-{Y_-
z^}@%_&*W;PEvTLC9Mvz%1v092-Sn8nMK#rf;N_mCmS~oxDNOWAWZTSwiTO|KW1d*e
zYkHmd?99ZZ3=pdrP+^jsYz*GWCHbIqqvoBLS6Nt%fVbT${&!9Wg@<MhEHjFVQ2#Z=
zoFD{tHU9b4E}t?NF2yKRz%rT=lM7;cyQ!?_<?Yi|03Pt<`n}-*5ZACC2uHj;(dDc%
zkWku}SYF+_s+y<bU0>Hpk0q)Ur!_M?RZwgFJrRz0Cnm|OVVYbp$*qNa6^uG72t~*W
zIm|lC@ngQ$7KNHnHPh4T4S$F$%qd1g`R|M)yu7Zm%Chg*UgP>RfT-K!{asfrwO5@q
z;VzR3`RQ?zy1YB)Ro$bYs4}<hY$qV%?(_ZWs+23r=_?CPGq<K15<;H3AJ~!O*C)u(
zqs}(sRZYl^lruk!S6ixgLI_i^1{GfVzbt{Y;oo|6IGKvEh<grA3}cdq@!b9j#58Os
zOHpb@zfpX^t3y!J4zdczeOOjEa9tw0ekw3x9lEE`uaJir&MPpnOHRe}uUNqiwQ%}2
zBwx}BQC&UH$Ktq&9&f2?%+NmqTaZ&14cSZS?>}S|5N5E34ecUio}gf>9YK}VLi=B-
z&~FfUN%+HN0`fg(6GJKyA`8jpD2oknS(-|E>x-NK2Mmc{TVIG3C^-uTAz%9L%cY2M
z($Qxxnn?C~T$@e@LejTQmFY{>+UhMfw#uiy`L-o-*Vg5$a9C5qep`eLF*`J7CRTQ!
z&fZU$kS~E@&albI3Y(oGd1yi=#)oftW|sf1D91!pPmPu0JSiwi<;t6U?q5>>n*2_s
zEV4TUpmkeZX64^p8Q}Rm>}`go@<H%c&5goeZEfm8Kh@6^SXuEcIW$yZlKACx*i2t#
z@<wpjYt`2oN!z(CZ3K@9DN53k(X&%PDADg7@9vpGtx%y!^azoB!xEe$1>r)0Zo{hW
zD;C1~XlUqYM{0=VV=$Ne6Z=iVe(s-8xQ>(!;PKuCGM%7b1h<T327UGZ8FYE<`Va{~
z5c2sxp1)(cGd?#Eu6F;r96R*1kX1}2NKUISE2(H|JhsvleO)rU|K+PJVrNwlJ7s|(
zgP^S_Dvik2u1%?s&Opwtie`HI(QghE(O_0gr=soIlu&aeqFNz?S8c+}wS0JBKj;nu
zMwEkVs6rE(^o^Xpk|K`cH;);&JDv{?sW?1gpQvI{&fKrzw?b5F@Y2+vZ-St2I}0fW
z*GJ9%c6%60=*Xyu*raq9qbW%Pfrx{_5U(CBH3qY$I<1K*h~J@gYw(9YWi|N-LsUIk
zBG4xCsY9c*&|qcrHD3^kIC98pwGXm@F~+j1i_>nO{j2<Ad*)um5{{EY3B!9x8U6os
zqsY%^+1ppw71b-Xn9q56A|HGiKY)<FZ56i$wBJ8wyDqKsrPFZ(Z3LHUeu(4Qxi#4K
zuTxR7suE1}g5}2_9aoH`9okn@F*5z>MM_ZsY9?9#ho*Djj_ZB9aBMq`8{4)UJB`sa
zXly4ZPGdH9W81df*tX4ge(zfU`2=&$ta+XrdtaO5yvt)+)voqs()brNKMLUHn%xy-
zcrLsfoy-sZYH~EN(HTrU)UAD_2?AclkZ!gRg#^y98K$LofU(5HL<J50E$g5R&{xq3
z8a*1OB0zp+-s~+7aUJb`>D?U!iHPdkhK3wN?kJHr<=HR$*BD{>Td#u%AU+;9jT#>*
zS)n7w=3ry8goc`&l<e*r)qzW$<WPN4eQQuy4}-wpS%gN7#tgg4r3`L<5bszXD?Krn
z99h>%<S~Wt_DjrtlKWF|f4;^YW9B=#x%u8y#7*Nu(tdhm3~@$wZMA%+*ikd9=*TT4
zLTq1@AzUD~t&pV2<5Bs)f_uRf&#R|rt9Z+sZ6>9KV+r<$6Vp6h6+qvh^{>U)po^7e
zo=a983%iXihQrX-F1P@<nh(o&ENqcOK7O)s<dhgl(NVh}YU0qm__EyYQphUqfrg3(
zBl1SScitteG5-!-4X3*gGH<AR<dt21wH^LEj1l%V*OS<)LVd*P>h*cUtpA&AV~~b}
z{jE7EOfi|;F=A9=suF0*b^iBePpvPrGz$kUwB+u`EbU25js~uq<AZr3@2ld3Gl}9-
zHyyiUN@XY!D~cCN$zO7Q#lIE(AnXn~(~@U>A!v0ycw?6HXJf!?CA*WSVQr+6P<Ef$
zfJfSy-ql9G$#cNbIg1%=`r0Qi29^x%HzwaN2eK`p={EkxZ~{OwY}%J#sf!6D=~vVi
zZ(5QhpwlTqy1?`%pxY2`omZ)IGS$pLoi~q%IKu={?Y?+W?cop!snrPz&x~nBI}PXT
zNV7-OYciDiMTVq0iCF8Dcl`}3g__DId6|%o5`TuO*7S#|&o6G>rPFlv_&8yN7I3Kh
z5LJcco%$ySSqx}iT9ooPH@V-Qn1W1?Xw-a(ti((x3kQQi{`8|eHF3lF*ZFyLkSAT)
zTkkwn=>C7?Q{nnVPdLA1HO_(1Oz#q6W<S{Wz-d};l?+<Xr$;V2eFGfC#~f5o^s9zD
z|D|)gy(kefX}9t4Vi~5%Ue&Q))26xiGwC%|fts;A9zaP=QmA*-`>W1HVlq|<^($h5
zaEQ<~7I94~XaEmh1I~oQx3n2<28H~LgXA)6m%bA7MK}yPaf71%T>OA()8T2aVb$$Y
z*z*OXCK31hWWSkzm#}kty-o{X&yk2epTXS8>kEAESkF9%UlTopbTi}4ArrQoXBL0B
zuhQF_1K_z?ZJAKe8h$=G&lD!~UA5;7O+S-4NQGfFmHQ*wZr%AUu`V<XIY2|>+(P&l
zNykjDUhRz}6md^BE+pFOCHZ@?@yW(RSm#+D5YHS#ua-9APR2a2e8;BzW@yhXklD(b
zCE}hJ-J?3o(;cq}VrLAHNJ&$&t_EOWn|dSC0m%MObbg<7QtWY~#xP9Vzvx|wczq!E
z%(FYZ-%l^a{mkbt`wf5?s1KwhV5!~n_-n@aK+gWfVX%Eu4GZVI_(0^>rP4kHB)C7`
zMm$dq>1|j+6g0o#;CFn%sE>3ou9S>;lzboD=6>Mc)Pj}?EG7B2`^hP7C3yv1p8~%l
z0JlsbnUb??p}9FXVnoQbR+m?hl$qb{(%{6lWchnN9W|#A(kh%Ryq~Y^Pua;8eXIYV
zWDE^IyMJlnaa5WX^utGxmA4^O(VU0Bj7e}AsCaPtAW9?ps3wG7NKJ1BTznJrCh7%1
z^XJ&ouT;b{_+@S+f=;)SKJ{~utanYo=EjWt$aiMp;fo98jec>7BHb9x^YzYL7H)D*
z{5gL}Zvp+xGaH87@PEn!mVNoFuR%3`g|6`4N5%Ca#gMV(Zdf>^b(<Cbb{yGbe0prr
z?&ABDvX&YheTmc5@@~EzE%>N2=UYwt5=~el?tnE=bI=&AW<HrEy$Y6sqSpZftbA{H
zjo`*IGubnSlm9opz|O#GGV)S@A7*5oLqSExSd^2WR~g(ObA8Uvb5W!0>Vc<4W#m3u
za)fuPoFM4@nc>=&Tk;jn@V@)whD_PnQ!ZN@aq-;ds6}CcQQRL;V7D5ELgiS$+A_M0
z8=bANN@o08>~F-re`aAG<q?$zSdNC*R&PBw=l&Yycu+$8I2Edb4oVHn?TjvL4UZ2f
z=YSw7m0&{ov5zVZo=u!n)6~#QqOr<F%IekEX(HHEWpH$CT`8<?(ArjC?>v<y==GRH
zBB1A1{@g#puJuT)7;y10xB^+Qx`4~1C>MF<(#WAa5Y@c2H}0L2Xe=!Qm~txULz;k?
zREQ>m<7bp<^Y*+w0&((}L#%a<xzfy42w27pDhhW~)bdaSSb5=peZomQFT&m>BU3q|
zykI;p>#ZTpm^Yd~pO$-GS+1d>Ptmqy46xCTukxs=q^7cXm(_JUt>#AC$A7l++L%XO
zV!{iRaTr|nU%jND;=X+cd=1>-aH)CZrwaoNn4sSruT9R36<8C@Y_s8eLwW}VX}Xhh
z_)|P23Ya%l{<&`VO!!|kz65$TR*3s6By!m@GOYvEAYAS${HWE|&(oR}kh7!7PA4uQ
z3vc`I2ePZVfj_!~6KI9i_qI;NfZTBR4RGA~qNAp*NZ676Ha!3sk-Z&EkMV~5VxxzZ
zEyF)>1pTCq`G4^tH@z+5?Jq@Fv((`3&eJB1v7rR+`*`+BNYmUbW+3a3=Dx*{Fwq+r
z#%3n@$?*)IFpLmL*3ZxnV^r|P!#*?Gf6^jBWYe&~OCz<9$j%g=S^K?hHnetsGuRwt
zx8|Y~*tT(TLQn*B+I=r)<`%P%4bCk2>dj9_2O??=^+DdH9#DRUtcVS9e`WQe1bt`m
zdKYZR^-|Kx#LC9<`l>Ju146s0&)O%6gCUW!X=8`rMykV^rmR?N+kYS3SSc|ybqXjf
zQBhH-ngK9LNd^`)Nm}fFks5xg_S;pfE(Z{@^Kdj3xYPyK`PmT;eJ0x;!_5U_0vlT5
zQaO!v<aGX7(e5^Ux6rMP72L}X^PPRG;ezj<wwUx5D5t8xDxB!I!TEjZW3~%p+*h7&
zx`)x&UVSkA+&OgDx`^A2iKKH<0DRy<0{c59<T<D5nS{tmX`nq3&R__PC*Mlii6xy@
zMZh><4QKk*4C`Dg+!Mc5bIV;cW=Ts)PD4dg6{$-_&jd;L^Xo9@zy6H!yTm8F=G-JR
z96NsJ%E#>?>ZYptiz?a)8K{}l{QSMWM0vEGs^ywTd$rq|Bur}ERYf(mVo3hdg+o&+
ziKn(!V{m{(7%;R`;>KMXb#cB$L5hf08u<o9f;5|rTnjKQ#42J9(DEmu%fAht@vZ)d
zH6Qk*vk~<-i)s-u2t1Q@zUD^Zs$G|>Dom}h)781Z#zp%VrD(igFS2&4jsI$`_Y$X@
zGHkKGnVzgTy`6p})-B1aYGpE>D9(!v4hteHphjM$VlU86K#cp7MHsWe1niv8{!<19
zd+`E-w$>MALpKLAC~I3Ez^{2tLGn~hv;3ScI+yc<b!tA?Y`|B@y~Wi0xwN0OH;jgZ
zgMo^cyvbhN^xRf8>cavNW%&IIL8EoIyn`@Qb7*=QI}NkZF*J8}&Q(f&{TZiZ%P)s*
z-<*gDIYN0za{j?xEN%|gAe`%fnh8+;wyhB9A2{Gsjexpva(63Pd7iEDI%;C0diQ8R
zcZ%jYLik87i(;9z^})^<A0a_TiQ}SJ;@C!>rI(0+2Gch~7Nl3Nb*2>$A7L+$Nrv$s
z0_y=6Ix^Qu-N6^;`v%Qggn>PAYi}!&>~a?;O9E5ZLNPzFE|#FV|I?H~NfFR$<Y24F
zv78#A7iSjgBJ6Z+Mc?z-*~P@T;?IstH`Gzl@4{sfc72mLEtvy>mi>9W9+o~@CusU~
zaS%cpI)x0fE@;;5pP(eWr#ZW@I`RCcSD*p8Z<k@6oGd+Er2@t#`1L{}Td~A_n2zRp
z?ea-tqQX`y-qnqB>~3(fT7X7p+5QbF-%%K6_!Ka$pc}3!H+Afn&*EJ7_=V1TtFKBi
zCbVv&4}aUpS|a1Vw0;PpVr4M$z?wnA`KQ*moUWMUwuA~^d(WS_)#QpG+i!XX@#xH)
zR@#Kv+BI=52n6IX_QT{yl2j3{?egkMupjWAgPQaxx2;E3`J8v34wRm{adUz0`^hY;
zORxqptYjvAORE)q$hGC+-}b!~c)H&`Je%8$$!`;}%MiHt%!?S1AH+(!3d7s$X&5{i
zd%JeM^B{6>l=DouO&`Qi4A5zzS*o@M4zTO>pw(hL!Hi$Tn2dQ{#H|*q6Wb@QDgc=Z
z74JfW=bBzKiaV`G0NnkMbXV(pgRhON4%;?C#B#0#Z)hwov=X*w<8Wl5H;V)*VWk1_
zMJ38rZ`rUM#*l!$XE!#=YVS-ID6+Jg5gpptz$e^YJod7D9b!gX@yXX5$7bEF^mH;}
z%1#=#DwDs_ZFR6ViXg1tSI!2~xMkL#v0qsebq0CHVTY6txYR7jwNu1`5Eejc8|o{$
z+#$lkPflwvBT{c26Z6SnM4oPuNU<<k<l731jI$sQl?&R;;Cm~jD0`6}C4~v#El$`C
z#E)>VX6rJ?dRe#>c5%Ub*cSpz&)R?n#gosdtu$5TcyK$HU0M#&9xZEP9acdkX{%fK
zRTUA$+3#`#=$YCn%3&pFnDR#s?jqg(CnqO65CoUX^kF@xidw8w=}KRcrz7dpOyE~6
zwcbfi+|P``i241=&}VAs!^>vOrMofDzBqebF3n8)S$V8DL7%wm4L_=le$zLP$*cPI
zfxyG_JAs^yfoFPZwQpjXo`nrGsWosIXw^8rwjY@TzV0X1V{?l)B`%z&4vf+l7MuP!
z1@Dddx`qfZ>{Y#MRQQC>+Mm(WOJ7-TwS5*L<gKuKeBo8iEIm#lvj_ZbGQ*n=44<K~
z#-G{c6}xrw_+hPXk;B{^oK9dFScO1x_;rQ$H+=J_5v2dX5D_6^CMbo-XqLG?DCa7-
z4AYSu7_SQPAk`F3Y0y$1SYjpRa?;L82rI=qJd;V8qsh&wwEJOtxwB95MY0Xf`{T*c
zxtWBIV<<X~R_`?R7MCe7jCj43fVfjm;<7+3^ow$MLVSEiMq*-ON=h{Y9|YINS_art
z{h#N8Ka^H1o70mL$!sLOvo55^Jy#(`Nje7(0K%A6I`1=E<6r!FB&_kC%yvz)+W!_K
zAUGW5p7oON(^}Lk)I}G1_b02Fg5%jJKSS`3gt#8i34=Bki}yM{ACv;ud|@+zT#X7&
zow&fh&1M2~A&e)6m!6`nrL(%itCf0`i%a~DR=it%`ywxfFN2f`AwY<DnDYjQB~1tf
zCM;Nw%Aq)L#U7?}hc(}W(7E%@JuZ|qbiSs?je~QfD(|PdXVGrR>s!f8yGO0Akh(|f
zCLE_Fc<rvl>j<V+a}HbO?6$uoSW-6KVWb9&_OYy&YK?6)9i3x~0#iFf^77G8PuG6c
z!Es}Su|Ak2+D-*fs-A$D?s9WUFT4O+pbp=Z`47lYFkhTDP!AW@;ifg3c0)9A-a&pj
z+)*Mdq&6H=hx{3*?Cm8Gt}38~Cg4;1ytNUs!nj|sH5>=VcR9&bSdG#>6_%NZb|uVa
zdJ^q*pSI!|GH8~x;EEHnQkoOcJ1KU5CNctR>PY&`)I9;?GyM$0_mZmkqyTwUN?-Mc
zrUz&%b4)=TRkTzQtAx+!U{7MCO)3+6``;Wp?hj}9k%tD;h)~Rif=6B6C)3drlFL8K
zp{RVEE&Q6&A;GdcOD#<92YdgN)R1i`;=rVZ$5;R5_ZSpV-bhM>ofMeDU0qSrhQhu(
z^HwQMCEmBGpqyY2rI9g`g8~BZi{}zHmkGrjR}LmWKpj-NTZ7&a@RZe0<oEd&i~+F=
z%)Wk3Z5MVeyAc_0y0(`%xcN*RAO|n4lYs`wv&w3sVtxtkm}EU>Fsl#|t)qcxEx9GZ
zQK>>nzEq0X*1SB(gnI)&HL7*6Tm`i$axZb(3R{myXVThv+#qQe02!9&ZV7AWTFai1
zwE9x(*If2>IQOlU4#-L(q9C4E-FE423)5|P&!Sg>@nh0=LuCEV_CLo^h*KNnXiLtg
zbd9-kLX1q~ivy$I<Ww*OBk_e5)5k^`X~?Pum*R6>yp{&$#|6G7;abyvLv~_S-O83e
z(Vo)nhUkHs2Xc5V0Vf<p`XikdXvQ&}CKu)<6Ccvx=x@jC*U-vG|4Ndlw!2A)`*m>L
z)vC?ZWDGP<YVAQFEHxt1__-98;Gco_ZIT6hplng3iE?@8e^}FAeS;RZN~wLl|FPCg
zCHPTlMp9wTZ27!R?@r=I%Z=oqQKYs7tZIJe`B%6}+@wVTE}>Jdn?5qM$iT;2#miOG
zwb{=tM=KNkft9znfZpCbm(%du+BuB89Jy2f=(urC%Je2Xsbp_4P-sO`Rj*goG(C2l
z$_%(6e;lQ?iDVgXf%6#z=A!jRlZQ)tRYBS`#E<O9HX^z~<l2CgKHAcsMyJ?GDQ&*1
z{%YTCg`b}QdNSw!r>$M9O*C%-G2ml4VdydRrOM4UgLkIAsuKab>c6zE>SzW*9Cx#~
z6<IX8leDc9E7{TRTy^|93}a$`7Zk)IKc3jb{py<e$ZTG028h|Qhlb4-`y2uGVcD=(
z+_Y$JkU3msYDF+b_9E+3^zYjtx9Et$JNVn`_i8|fZ5|P0zs+*=nxm+;w$yUBl~cRK
zAx^Y7a;P(?AU*uMIfX<{GMKKS%EI@hZG-O$AE(?|;(C3v1c?^1Y<~sh6QK^+twYFC
zuHW#wnQ>I6>l{&|$CyI6eYemZ=40qBA1SYR`a(5C8BQTRH5aS&1l&r08NMeWL({9=
zpX>uh2#zS@&>KN;YG3a-unK}bcyZVY*X9n4P;IqMc`bDv!{UPhO9yc)rPayVdJ!w^
z&eXY)l*ZW5E2J2C=HfM{yxRx%ot#(HZh%Tx=d#*!u{)|#?ocV)LI|{)erRodc_C7M
zw<!xsXEN39pEV6D!Cx+0z}YVxGFfXlE(2is*AQigqnw87!irOxL#Qe$wKB2J@?&nM
zNoYbz;o-~#gjZZ~i0*IraG1d5*r$nr^1pU_jFje2*#Yq)y~DeZBzW3NBE6q5t*5%+
z&$Yd4^-CaxlX95Z+V%KTM=t#diW8a@-sg#*F(K-z_tqx*9{t)jREHWoi#x;#DsXj-
zacdQ=XIaT84_~c`n}(VXBj42bdSr%T&?8eeMj-M43v719<C}vi88|SLZ1!z%Oejc4
z6AC#WZS@Qe2Ew8e@p(U8#rfn|>&B*wu7sPJ9x@ohXtalM9XX<RPAjWzxuk-|t+WwP
zdn%4i97LBtEyNwAWm!glunL<A0K#9MmI(uD(+Z=7&`Uh_`O_vhn;+9KScrtb9sPPH
zu+)9PPij3r?{p-N-E6I~(2z~6`dibAMH2Uq9r5Yg?^rN9H)>ruOd7oH{A{^|qauHL
zXjt6R-7rx`f-x~;PPy!6jr51c^D4cBE%U#$gW)$6^440zlr#nTAX(4}0g$fW+ELO0
z+<m`EB0zNyu#6G9NN9W4Kl;7d;-dRLDF}<g*^psvV<YHNT2T>YpIS~cGNql`@Rgn1
z$HP+Ax|rD*kFHmVA7vtAs5V8~D6-~)dk5=Ul&LVqUZGrt&T@+`34baET%gUbzBMk>
z{M%Q$Unrt7w`mKNM4JFaEMR^%uGR~u)pI|2?w>_}UNP~So1C19wqu*iNDo%wDe%`A
z;fkL0w+4*WxAXj<<eUUzxt~T+k=pL&j;y$-+}yah)oj}vci|&+>%RHEkjFeh$7xqp
z?vg*1zgQe^5yu&*`&?O1^~L0OzBPBN(c_RqE?L%7G#QESKM01#WCN6Np5K+7>&cEz
z(eHw-zMGooQ!z!L{HGJ0SB^DeIG<hQ8s;2-$#VkziIFJvvW!W=i(>>ij6$fwIx}3+
z#@&6@RX@pb+SsZck+qUE^nA9|7fHE5fL*b!DCTu>;BU&~B|JPZeCgH3w-U{DQM#@d
z+d03eX($Sp0?Y0I5KTpHwqbo`W@n}ACNj*(7hPpWa)T|8kocWxyiQNcC^t1+|2Dd=
z5sWYAV|~9@4I#kHJet(KC#JSJzrxDHTZO?vB#d@E0f2pc@8z;<em#pL!nT^rQ*{I}
z#89$Z<?rl!mbRqKy;f@iy-u2v{x%5-(ozX@_knwrU$avI70PMiuP89X?gFon6S-8;
zhOpUiB(Rvs*2x?;;7z~VlfpW1q1dxFDZaRC8o-<CJ5gsB;0J4XDd~0I2XTB7t5vb>
zSnUBt1(xs%heN0&NcZ{!`fIJVYyC#lCZ_0`Yh~s*S1xGj2z8i!?dJ*{_KVS`lN|Bk
zd{!t_lskxkH?i+^@|FK(fR~o4;xFXlV$bC-G!fyyL9!7<czF15nUOM!qc5r2uS{c%
zn?5%u?~jR9-8|iMY0^zh4QF(QnOHkmSlM-*7gsF=LByltSbavS_9sW8q&bx#^^^Z5
z<&_jP6O#i~>{Z`JcukxPYNKTDrZT-V8J+)Bx34<`I6}0X<os1E{~a`6o>Cd)jF^lL
z{h6K>+M@OG@O)m*c*EX8?E;hsp4qO+aZ74ekxI{`F77bB2P%R=tF;5mJSbEXx&%LA
z1n`wb){p4H(&<8TLl!Fp1<gPry?_ZGlSywJP>)Nd@?T>0mMOj!_ksCfsOJiPiJY8_
zEFdfU?Kh|~9-%9>0Hd8Hw;Xs0Nuqb#^nF@T2>=%;@tF5|=uT?y7xA)b%lVneVbJ`Z
z#*yW9dw^ff$xMFqxRAJCFa*=4OJIdL*<(_8sZJ;T5y*P$&JdheYPT&>mZ8E>1NGmt
z!CYxg&rYvUQvA|bjxK!I2>q*+g`R~4nAD#Z<CTTp6$~HS1?AkfHBFJAY__9NZ$#Jr
z8S^gZ81Q!pK;<n2Lsf9T8w5G#>sSM``4(!Bsj*@O0(a|+Ia-JH-u}kd8nF=vRiI<-
z?eFjXG&MEd!?KqX`MycZVYB}5H~F&N1C&U;3qq)e&E&eCKSjsy_qe9@u4uIZ%}zlx
zv(I-gwBFau*>)Q(v?Q8;kvOnrPVP409RteEE2~0=YSnx}{ek>IKdn8!Ej;_~^Gx@Y
z)^Kp=PtOAsa0hz1CK4g9UGT8CRX2_`By|4Td6VHU0P*T!Pt_;Cz<#qQdIyeG0_3?f
z#q!ANIW_26HiRo)n{Z5}yjFpu;c)m~3K$KEy{zMQ=-V7Cu(h8ofA?_zXwFV^*SXxo
zM~52{<8EKtT9y9&$|%`!6NLI3g!6ZEmQuO*Z?tmz#YwP)=6&;Yee_ovl_c6I7Z-0<
z!28L<x&H8OL!qT+17|epLnhc|sEvem-Ui-M)zIxQtV1_v+qREp^AiI=zCjY96?ni3
zDa@j=J(>84t2y0tjiBUjG<f^hKfBY2wJRtf#J@mV>!vgcR4X}e9-R|(yUw^>qNq7M
zUUYaJCBz7LBGnq$Rd-SB4h|WDf$sFr6WWF1&ZYVe9>MVe<!CR~Rk2}%R7h{~{>FP=
zc+p9!(k;0jK44)!dkA14R_5namtG6n>V^rCf&q~}2~LZL^Z2vc_b8QZVZ%d-bFZ}}
z|L~_z8k~2|%pvlt2R7NF+}-~d#A%7U@X|az;|otPu?E~Z=j?v7agfRnEg72qv+t$!
zC^Y!B=&6!ilEbqq+qT>wnfUmZ=){jt4ELR6Y3_et6@}1$NdhkVQt2^V3<03$O_v}O
zwjm~{nKZs6_><ODbFoDmVCW1_hlMBC>y0rt(pI&3M9}PWNPM<ntfnyPZ6E0d4e`;J
ze)#t>6vnev@r+?C`hi0&M9hrLTA3C4=^V|pObwJ^0)i19eF;mt$%V|IP2@mkB+AUt
zhx@|7ON+A$3usVeEQpU0rbG5KP0^MvHNJkcBDg<hcFmxc)Q=|QbBCuZ@BN>cz-V)0
zz3od(ql?~o+imRKh&-}e#5cpH^KU~FMF3mmKPeKSW<;z0^P{#n;$z!V2}3CWy&eX$
z{-`@?Fol>y-mpJnL`?L*JL8TXwZq1jRofI`bTa=e_+<sk6LABBT8ic%5_G+@aB^}|
zn?rxv9P<C}-MGv4Trs3M8&p6ivzn6|+H%n61@}Dn%2fp(2UgAvGcH;P)`kAn&Z9m%
zFs>Ae8}>Ayh27sw+)5V>n~3fhgh6+38k`y$12z9M8w)G*B61oAvOy&O9)D6Y(^LSZ
z=Er9Q#9abbE@W4Sj~CJ;A8=?T>8OGBYCbj-Nuvt;rl)cX*qZYf{H_-#6LH@kZ8NM&
zVSA4EpN;ld?=o6Hq)DRv9&q_+F7<?T4AaoJErSL;Khv$T^_K%@P>#;O{KwJX9XDEm
zU_$p$FyUAF?=X7#lg);Bc|A;b3TR^lMEuch)LgV))okmijCK~S+Gy<PH0JiC3XRSZ
znA`BEcwAs<jG>-z-S9n@he;BT@sC~fZ7yV?ToO&rWf_j#^g`~xDFB}|ORI&jMm$XD
z=etSDkR3K))aoA{q|Pml;!f{{=vXIYISH?gzheA3z^-Vqel1)2B~O)>4B&U0Sn1uF
ztqq^h(47w5G-&#3?4|Ipur@xgq0Fh4ZB8|QCz(-!d8z%@Li#eWDaO{dk*K_Cc#39e
z4Xk7GJMf|S#IW8K$Rz1aDRd?@Q!PZkrcDC4_Y$89!(Qx*ex)NrZu_p&%F-rHyM0Fc
zDFPYlz5jWS_&|z*m7w{r32=Uh_%^I3j*1@<j(~`8vv{z-kNc;5v$@3bv@ygdRCi^C
zzY28?!WY6}AFvUVk#R$HBR26q<`vCNqwQu6)%vn!SyDzb%9$;yUeF)&`<Qd}hb=m<
zyGnpb(hxf44^t4fQq&-4XChrr<8{e|*?u`Z&qXpg82b6B?f@670-$HoHylLc3P*x8
z)#f%2&FSY%f6O}R{M<gapBhMjbyE*GvLpEYfl+Bf?gl&K)FS?5^VoRKGzRB1kKb4I
zG9~JK8QgEWkNro(;ZJTZDeW+H_pT3ab1W&SAvyZ)dvmvZTeIBAQ2LdyBOuS%i-MBV
z<@u(5zl3%~=08k1AX9Rz&0B{4hW(HK!MNlgOX!9@8YA5`8!zm`E_=hIU@>%2>AXY>
zf7cWx?am164pGf7CIoDZ%Dy;bWGSkkHA}5J50x6VE3wo+O4{Cf<>YRO)Qvs`XIsO?
zBHynN0tJPYliiJ{hMo%x*9bOee*c$yW597>9NC9KWwfg%aNvHpI?K%3)!1^oQN>Jx
zV$fxOGMr`OmA1lkk0MOgz;wBBad|<6N9cQXbR5CmG<m+7f^S(x#xn?UD-4sh2rn~?
z{-U+Q*SgdUNq-V3L68lK4(0l<VnW5b5eC6eCK}0RX1HoxA3~rpuyS)nuu=Q<uKhqM
zZ>C{z@9jXm!m4Xv<n14;Ak!iJ>a%XSe~IP-jJWsppL)AVobVr9U&2+tvhZg1Vr{*X
zW-#7s6hQGDe6TI_9hEWH21EzT86l18Xi^5Zq+5Nv-nw^I%?MDZHUb$U?!xr77nIvP
zd_)yJcT{t9r6q^@2M3su5&eRtK{B;wGLe7cGi(4$$6I%>7trqY@SoDoi5exFXK|VR
z?rI=Z*uTULK>K%j^o7FtCWM)=co?Zh1Dft!VD#=m7`mo0>cmY@pUuJ%MZG|;m`)=I
zbb`P7&mz2pVO_vTN<yoD^TVXZ|NrfVW)jZ90?4%5>2Iu3Q$mk8TG}Jiu&YY_u!;vr
z3}e3Fcd6VVm)3d+y4RF6vT*jRNdI}l8VBD1@6}U+#v`5^d?o-2BhCLs5-DfadP4qI
zZ%(IyXTmc#_w}!Cbun@jMfUXK#wq_6dFABl;PiNRbA5|?=P1hQm&8W(yN)qM(C4O$
zN8{`-;+M(wcWxU}ujiNUO!QcRpKl4k#Yb*jQI+tASN`qv!MKrCK^LWZC@*hqX{%Ww
z6=+xV;@QgQe2;qTuDMk4-=!SbJ*;+r>8&pJ`x~e-JtjC3KgzAjZcHj{O)IU%;Ca^K
zZ}xnVq^i_vGd`UZr&h}Ddsz122KVUAi##}2mXd$0BDT}Y#y^Jf`czJrHw6%`+=Yo(
zHsx|^2A&uaJGYSnq-aU=a%?X4P@NL7^TXFD*dU=YomAssIOGMsvM<=I)}wIcf_RmT
z6*kuAM9R>=o6>VZu{@Tp&}GwWC%YYX;6Kmkz22N(_b)nEAFJ1~j+_lULq0Z-ZkxJ3
zcpaaOBt3J21qc`-cd$o*ci@R!+%XePx5%>@Ykd_=w6X#|J+wY*Lm1)>73xaj9Wj8W
z`&mQv`}ZMvO${x7HK<p`Umm2*sXEKNuYM{haPAj^_Ql6bVW6G1;TgohP#iUL@!1D^
zht+dVn~RdiBpZo9=8j1=_r~jU*HAJw*{(_K&OfK>SVo$-Z@u+>)PR5xWZDZIh<;@X
zYZg~a+im%;=}|XLrI}8UjBdKGM{m<_nzW&sgUr+|*)Ic@IIDIf4fdxxVY>idWCbqD
zWyNX($v@trRc8HyCaV|QO(Kt%6C$upt*iq(s=D76l#1`rW2Yvj?<ISLak=e09tRLh
z8LB~x^3bdMvj!=jq`=j@5u((WrHK?cUS8g_1$r_j7_%(8FMKdmAbPC`?i3?KS@BwH
zMWyj$=imXRIbY?P2=&HM!XdNhXABf^#|=5R;$_vi96~LK-kyZ9r^gaS`%_L{l5AEo
zk6tXFF3}!}fH!J?2PX-}1OE|h&=F!ntKbjkGAPV&n3;}}3HUmVc<>r|Co89*q6rZ$
z%&-e#&RJGp<K}ohIoD+8ocHQur~fY`y!7N8*$ahG$n$MJ^&#TB2x|mK-n-M@tbeV#
zp$V<nGt3w<dZhpH5;$i1-x)cB1oLr|;0Qf<YsVROTwKh4xdgRO+ARm6J91jEh)``1
zlp|$bQWhx;4nTxQK)6B-^N*n~Wmd4s=(tkVx=4Ey)7!DxTz9`yfBi$yBtm{RaQfKg
zTPVOIEO^4}Wp8Dy6I9j^7edHG%NnTv9*7lG42H^Ykd#B9Qf(jLq|WdYx>QTf=yY?s
zw<ohEI>-iIi?>;X4St^ezya<X-EqpY7tC{OpKVA;JP;XKp3xqV7aw065D7JU)XaNi
z;~HML?p?8~e(O6<uNM^#xWjQ4FYx*v>@;-ddL^D62S&VdQ+N@CPl5Nadw2SM;13h+
zP%<|%C$%dZ1V9((Q`R`w#DWDjT%&>uOWor?BK1^!M-x?mGT!c+m}m&fu6O)5&)wi~
zbA=cc4FqG@4UB4d3}AEe6U1b8g6R`|GZ&t<`kpQIx)L*=WH#`^lXglI7KlR1?l;u6
zobn}%0;5Njmf#RWRK@sAB=CgGl~|NjyV|_)bwg_mhav#_soEQcK;Gf!bSPV0F4VuC
zbrA`5b$LR5S|e=LR*h<9`Bl~*1A~I3uWC`rWWX+Fk3T8Dxr>bHyZ`QC)dqzCrSRJe
zEQIgoi-eLO`U)qk8y2ffQ1FdhescR2h5$b|iAc0Br!CW9k}#F-*R53P#{7WX7o@rh
z$s=D1zO<k;LRfwR0{ZeN#%)Y2oM0r8q3N~3nTfd}VH%FFd|XU|O~u};7k$)IZU&bm
zp1{@KKiIUW$J1SZMCqEpX`xZMX;{Li@`ViQB1JBdFW{7)u*@*dt)}o{v`Np^dhHS*
zCrQ!_V`vZw2EGiuN_-sbuBVK)qz1BvBrh$i@du@rJ5HiGJR{?%H29cC4n{W>OM(6B
zjE~D7N|Ce&L^!0uELGNHEBX`E>`p$oT0kYp?Jf|}OKxY~1**mzL+=R53~f}To8Wsu
z&7NeFwa2t^e@_VRkqj-?L7jC;jz)HoJY`#^GlO2UQP11+?coG@;-|_W1q~f7tQ>=s
z8P0+$h>mU0N#mhY=PTDII_r1(+$L;cIhdfE9CKO8@J%|oJT<AD^p&9yEQV+Z4?vLa
zulmF!?#(Ne5LHk~Bm6g+fA;HW*S}R3dta0RkqCK9ZipGI(z%{(24bn!0FH!sbCP{<
z7<U|3o%Jz#f_Ag6qMZgd=HdLD=r_8u*2&6fM}>Z$ol)Q0;^xQG+u?h(;DAGAnztd|
zdTn(8;#Zqs1!j+#lTqt@GjegleIOe$i;sRp@exVq+>0xhF#LjAimw;s>0q%<{;vAZ
zFA9cT4)^mYacyUm2v&DeuhsYTW3K7Wcim12wkl)AP(Pjs_sq7SjC7neowi#k7;nss
zET7bTTspZmF`zXo<m<TT;Z#nGR-_$#oPZqtFN_y~Cr#4mx8Rg!$U=BD3gDO+5k2}^
zK6R|fXb(Y5)bzP{57srYvXU5G(fD}ao>BjiRR+2_4cM^T)W<r_pA`Q-_M|VZO+AI-
zYg{d)u3{c}0{?;b5_5pw)jw|t=XNghQy0XQ&&qvy!_?l3MxNcWLU8!$e-2Cdc9*P)
zePk3yzo@}QDB#;zXZLh%1;AWQg;Zwpsy8_xF9j3OML13xkIwEbcJY%53*vhkRzF!w
zIew@4+gKxfo4m}RCl()`2IIN%C5Qu4-uhrFjqBO&{CX0E?X0;qY;+VikkTtI%B+DY
z_4J$bS2Y$~EGR2YuVN|YN&L=(D?SB>n8CEY9|}*{^O<8_m0!8-45(&0FlZc;K!4al
zagLGSbmKoIY%LlZ42=qM7@J-vWrixcI)kP$;oxt#+oy-=Zkjk1S~y46rBpqF$+X>{
z5g6)ZzRbm&{(U0Rc$~)VWc?F-<|kb=j*w&|C@uE_nz2`jYBNyBlE0MA@-lu8%?TEE
zs@AVxGbplZs2%1s1Al1cL_4)~Rpbc?;qq0;jCP2+8XuPaNh4+9Xiv&a!up^D+#pId
z**B1Gg`H?gYoHiBrzV7ngPUq0esjne=~Bj_?^nb56YHOHRQKB1y);J&W_m;uw&9We
zJ3nKmG&595|Dx?PpvLWb$5u&tY$NTyMJD$GXPQ2{8W+4@2@sCrWXcQPR<32Dp$f)8
zi_O$(x0qBPj7(>F7J(+#B#(_v`W~N5iQ<W77SM9B=YZrur6P4#-~5imCCW>LxR}#y
zFRT@&n$&AHP<taww=``w6n3Eka~0E?)_5Sn`wL#R0sVnnsyPPS05^R;#|mBu4Dsi5
zetRE-<8+X(K5(}mA5evJQ6Yb~>|+?)osc9P6ig3G?JSk~o)_;V+ZgcrkZGQ{dUe^=
z)xPO=yyt(uZ<Q?Vscj|4oXYKhi-5J~9h@=YXA4H*kg?YOCiZnFp3U6^#jaerQ%H6T
z{VOFIRe&SUEVTs0Yc|MI4!y(?fE7Xf+J@d(cceC(1aReq!l%1_5Thi4<lILogYQf{
z4VmO@+-H0)yR48;LyJlmTFDp|6?rsDbTy3N!W=>TknTqWzi*ssdh>%Y&0m6K@k+_G
zuk!UoHFV_?;?c!fd!!#bM5!Po&O_rx%J+MjgBjWX#(%ddwh1!T9c-N_gMH5|Lg7}g
z&_&<10*dIzxPiPzcAi4CL1YTnRGi>E#=p5KeLGJ^yp&8ug%QDAGsh@81Boq|If_SN
z2`_n@D~kX<ZzwN1sjk(p{J&JpwButv<TO-)2zum*m!>V)uakE|#$ccImuDXnWl7Iy
z5Rq&cU$q($xXCQG;Fqa{m^4w;y+kG@etj~T0sA2?Iz;4h(}7&961-t_{ZJ(9kFey>
z6i6FrcWYlAPozLV6!fu0+m&TU2cyKh4CpOsTi48A8)|iA)I!rLhFJi(n%@vUcm6%k
zX#s)mkEL=%qB5-=wE6slA1UK!J%xYVz3bySa8($Xa|SL(>}Iw0;Qt|^*sfuv2oh-o
z0(YKuA#7;P$`j2gO6r=DtHZObJe0s`&+WSsez$O!&#}dO$e9+&_ouAlQ8@&7<C0d^
z?({DffgYhO!rX}?-B*ib(%s%`VzWmCW~T?Z(UPVNk@p;AR8L1#XzFc2(+eRtiJ3}z
z26NM+%U;I_Xwc^#!#NwBzRqp0Z`Y9bK#t(<k9?ZyE4voyx2A4=mf12Cj<bUwO4Ph7
z+u?r*Bjk05$qS*JYmea2)L?OW)K07^B`3qoPLT3I6*m1JuR%+4$=uu~y&f^-C0~(%
zWR1Bd`>V)Fh7y}r`&)cScZNhQwRmI*^nNyE7O!0JNw5v&?`j(0L@-C$!*r#Vz%hQ$
zbxz2$u_~P>D7y<Co?aCcVphG4-rt|KCK2$%;zOdvyGM>RHz(K5mL#RwY;Hqs9rfER
zC&b6Z(R37IMW~x-*T#f|_HW+qT^z`Y69rv%x*wYa50XLPXdFY@(0lca>jArUY^oSP
zdUzm1?A$ZtNTZ_^ewX^d<dS~`1EYeI=vxf9xO%<*$*D4ih)ihneUM4Vh^$>>3q<u~
z1UgKczLQ{CXtENij0L}#MI(iRpy0PSU62)`L{{`gHC5$x3?$X2mU;)_yH@ECW`D?n
zbkm|W$5x{VaL`0@0(99+3F%tOE=2oRFyZX>MEGrzW=R(oEsn_>4bQvh!0z_OSW(Qy
zszK}5l{tA0_Zg@Oim+~L-XGIp7)&=$8gm-3X;V9rbw#SWb?mqiC27r2)O+Pc6i>bF
zHOi2SZzXpkSs6+bKkrPB{kTY){8Etqc1fw>S_nh+st=AZ({)r^+MWK%#YNiM6eMol
z@O=}}mlV|$OG(bG;mdOB1*Q=ptHS+<T2Ezf*upI($k&F}M+OP+z)d+O^aFNk7c|aG
zHe!=&82aZX;ePy3vh8Ip+`MH^vflwsh?VOeltwofoR23#C@Ms7RIj~5CBl^!R%%ds
zqe75ruWdZq&oS2nXVvSF=>%3XO{D|HfjOzQy4|FOn){eB+Eo=@K*K<_n$F7OM{#Ha
z@fk6@Zy=zem~dy2fnsqBzN~;qu{v1U8~F?}&~sVEO!u$d2U*}iKA}#LthFvUl>VAp
z*!crbc9Fzq6bLaK1HQ4OZ1`PQO+ivdd}?WS;~zho-a;5{Zmu}ka2-dMP`inmsFb`P
z`@(0+Qi6#l+7+Y>fcv{g6~=ak_fHg2;3aKpyYPl!8-Pfm^k`#{uw_?4Hu|#L?;RLM
z`c#v+HxUUP0ZoC;8C4$1flcwWbf{q3o-CmHAG<`RY*ZEjF3hss3&~6=!52!D?25n~
zx=1ZgN%mpzI2W5w_Gw^?2J!$;_%Kgf;`ISBFeg}!BF6>5&q|DtO`Fpa3&l_c-)dvd
zuv}?t;NfSQUREcfTL1AQ@2C_vxtGz7QQu}3Pcr0-&vP?xyj^=EXd*s(66GTOg$vho
zAhcQ6<O5#KKHYtbJNt@l2RCloDNLuOr6v>n?eKA`N8$bKT3q)wA<(*^Uz^$jPXUTm
zDn-U(BL^4og)f9%O+PM0^A)PsCXWdU2fs_GW2arkl4chm&I;?SxUEJ--Fs@+1C18h
zGOR76Fn(<sS(=ne|HfkNB)oHrP}5Qz8=s;6O6&IC2+B3?DMAuNej6m;dbFk;jl8ep
zR{t8Va76x1W^8nXp01>rj-D2c5$2bpHUY;{7{389FucLdXJf<r`-n2HtoP{Cj`vX`
zwwu`~)Sxy??&re+q#@Yn&UWTT<%8NpgXEa_Sp(&g?~g%BrKl$Jt?pMgok2sb-UO>U
zfgcDuR3hau2&hz0t~1HvI=`*9kyPom`nr43URdOo4p~pcUt;<m^7zhlS5lA5O!j^)
z!g;0uRA357xSzumVytHiyGn^UFdp+w#LH^hvBc*#735=M%%l`wsmMTk$WRW4Cqz~{
zFGwO}?2N-P%T#{YL9_A-nsM1kk(My34@mi5n_f{cRB<TvZ>N0HW_B+SPpHhS{O4!a
zL<RWU(XH`knD@VjL?b7^aSPw|CO~83b%!|u%o&rh0eW9SauLutV{d<3md?_uZ20$0
z?Ns*190QjxEoZF5eKq?+qWC86U^A)~pOz$S0?EYr`rF$11pMAF3^4s2IU39Kr50Ez
zvMsBJCWB7i(WOe@^Xx+b%=hu2VT^ojf|H^e#6QokCsaPMSY^-49;c`Iio(w!f=CK1
z2q{xq?xV=P(_{3`$P`qNBJ@zKQ4!*~i|H6D%Ma0hJ*D3=6zp^>JCuDbSvS>=oAgv2
z2PbyBAjKSFw^bBfVy<<$b*c&a6DB!#9KHHO-C>LdLKu%)#VB69Sb8e+<5qxJ#)bV`
z`flJl1BdAoj$-bm-3lI9p?eYvF?bM=5*88=)Q#&*BfzRQ%~9`ogJO`Dp{rAO*<bJx
zS}heV%n$Qn`8v|q#<fow+KNYy$edRMKZ?MDiL*1rwY|idR%m9_1`V>zXtty55Mw1s
z>8`RW9Y({8f(d71(<p61d|uzr!HT~xGG(rI(VCZ(`Ev|Sje`+Cp2*xsNR|Vn<VHqe
z#GOSIpuUe}3@Khqnho}fDGfCLhM<c6as39OLn;_h<o`VTl2+^L<ZFrN3^E|h4fgHf
zsY7ksT7aweJH+|JdAJ`2(?C#k?B)1JL%DTL<3V}fZ|vf}5^rIv%J-A{3)dr#Zxh|^
zKiQ*1u3lPd4P_@|t_DFa1_Se~huix_f}=daKj)Du)*<phdBO4xSOd=~#inCHBHeQ+
z1uN0p+SspMU>_58p?%z5txSaFI@qVS6m{+yD#M?|?tIe9mfA}=eNyEmnQZ`+_#r`)
z_7YE6>Tf@=8LIuze-^}}C=vQ%LqZe{_N#F#dGl!eHVzBf$6}a)2!J}VR`nppJbavg
z*7Qr<U)rgk1x5L$Mz*IrM0n^C{&vSlEwHviBz`6+-Kcs44c4~pT@{U3i+3&MdT&42
zWYb=B=p5sF+at8AD;QlzG9_7@weTbdzuFpWyFk*L8{%52SN*=NG0_$}$Qg|6?iM#s
zFq9Z$^_Cp?xHa+nyaOJdX@>P(xik(3Jid%Enc;aVkzNT9bM1k!f`O`@K6kyL{ZOD3
z&tdr5seGICr(#|tg+iSmX*QONgLv+HMTTNQ*M<Tv;hwo~A(^>V@+xbbNN5+LD)7fG
z$8a8~UhIO?t7J)pzNyHLEi^?S{QbinYM4%l-)8Hf)@zjUEHD%X)p`0B7Km(0Z{oxF
zZ(zwctHp>lJB6X1aNdYD+7u;=Yuf;Wsuc_w*K038?#3F2*ce|!klpKPFt`6F(`|<-
zHBN$W7Q)Tx+FP#<im@U1=P7@0?U9b0jc8s|3E|<CFgv(A;hnM~1lR9%$gh~a0yg$e
zAr3Moh^MJ)0ALmiN6O-06;`V6KACi2rPP{URx4hPS39_;8j&9mB9w8f0uM?Y)g;@4
zTsxO{@#pG+&&;58T2?VA`t@^n>tr97u!cj37Z<glEfi$S74A8PP)8%+`pJCHriQNP
z_kp$KQ*zqp8-~YRPx$e*Fy}HUHZh)RLu|+FJspGZ6JXvoAH^4$6Jn_I{(N3F@uOyp
z@Kgkq{AX5$-S@Yf4xS&n;D@|C4GicBu<KNt)uuhRNqjeAM^QtBF4F?>Sk&c5ICV3X
zO6p)&ExIwT!fn3JW24OeMeZJ7q_`?SNJaEpso2SS7neuDI*P>PmhkP`Ya7FYD-zn2
zjI^BBfbf9u*1(G77wKWi-p$XqD~L2abDb|u<q9AwebQdHRn-rc@xS=r0rabNjPD|w
zjPfX<Hmp<~*9my6HC?7`aY#{Wlke$ZACnUs3q+@?U{779*Y8aMI#F9VgRKf#qB9?p
z2RgL@*6hE+jM^RkMnVv@Io6J+^9EC4+!QZ`0N__jUC>#IYtG|@KE1o_i?&+n_)5;t
zzcA@NU6#A&S9&J+Ng4QDu=!47vf8+u-G1IJmw~X2pSQbZGi}ws{?(eihg)!)KyPRx
zE;Xw-78Off#C|jo&elWg^*SDn9V(9Q)v%AKyU$Wkk;Z(o<d%_AFq!NTC+NCVn^-Vh
z0ae*i8(mIiO%1s`1B)3g-AV(=0*V*PoQEWea-utbOq8S9SlC!Oo9Y{D*yl9<NLDRb
zTMByK{IoS)t<GTFg;a95)Ihco6x(IL9YfRvWvVXW6@QyHu)iBjXl1%v>VW9Qs6_xx
zo3h%M+^n8@jzCCv-@y0fxZIBmxx+7Izzs&Ov0w-x)$`_lEDdA<sgktZk9)Kqy;C;-
zePgA&lcMnqCGUiZ_T2?*LGNv=EfUf#C9<ve6a;~P{Nhc8?X6@#-Le{`BzUFMqn7;&
zssLrJ#O%73=ytfn_K|vd!Q^-an%3PR$gGXVxWyWK<Y--$<`#QNz3Q^yM#6a@K!Qk#
z)sPZ|<v?aU)9ZjVNvbk1>}Y#F`<IpOMl9g*e6c935^(#}Gu<6U&v;*ux;+U~n<3B+
zVZ_8#L>hIh2{492{rIk9Hu=FyCN!yObO-LK;Em~HozNM5`L3(%D~Qc-6I5Wt7h2GB
z=?(tobSC(w1g6BL8TRj+`cw;`b6ocAs#gBf_{5vx*U#;}ZCniO<oML|Y#iutcW*@B
zk1zy9%G2pTpf`QFC{yfTxfPyQ>Z0wiWk5M8Y1fJw3@OVv075;3d1PPpGdPct@c3?z
zMy2<m8k-RJtB6dseXjYBaykg%l)BrAW4X8FL9Cr2en!wu%A=rDLh=LIG_%?2a&h=~
zCwBIqOnW0O0)nw_aEgp88{W^lTXvtA-D-{@psEwAA+kKvCu7_NYrq{x;nR3U%vMFx
zhF$tHwm=5IZ1IoXx=JJ4;HsE{`;3yXYK_(m9YtH|^9oOQ-**JH-_Y1dXYVF;;bn5Q
zNzh9KG;p=bjbTW9#LU1#SY5RnYPfc1sq<JRc@clBr`;xS^Rk=W1$cfQ49v~Md3}%m
zhs})YFzvsbkwPPX_~gN(bv3$X2(?5U2!C>dG<)C&AC9-X^&vRS{`00gQfF?OQlO*x
zL+kL0r@QH#LB^En>dred6PF}NE+gxIT7uIDVpl4o5TFLF>_fH`f7vsHNlsToPrK=J
zY1s#x{;%F0#d386BMJ(1+16QU;k6&N+mZRiLEOTB<`}vX#);_rx|z7`RB<@7fJ1i5
zjImA$D9T4EO~p?1qd&}5>x+&rI}IP(Gsz?}`)$cn*$<h^O~3Y4Q89V7)y;n+Ai}eJ
z^9InJfa#c~`zURH%kchUu%LDvP}{lubWH@H9vJoHAdVML2>h*E;3n*He3aWFA>Vra
zgx|U*9vIhFKF@0%4PniMd}wP7wo<kpT#0IUjd5GUyI3+zy0Bv!5^dvF`jOm(G+9cB
z>+ZQQ34#qA6D2fsqTnTHe%U!wqXHPk1YDQ^0<t)ZQ}8NpnSUs;uqXJ09@pDSqZ2#p
zSNnxF&obZ(AZtI*$o?nu4vmw?K5~Y*>VmCym#ZWOO+A6Q{88L{(bMMVSNXP#8Z=a?
zsB)j=ed{%syi=>K?X*`|zar~&9EJlB7>cEuRNVn5aE-l{2-!F#vnvcO_m#V4qJIQ<
zYO!|25&eXBr#pbBN%v%$*qE+HU&lJ23qTxvwxgLaMx@@n3c|J;D}5vsTK2s^csIYq
zSI?jpv}xlm|7{PhW=v}>E2_=#`42YV_oecr_U$acSj0{DcbEVx-I2*9e_YX1BRg(f
zgELwzqi)s4E6x*mAw<m3RbndIg2^91Wrhm|hS6*t{8OcGX+e;MPY}4B{CnPkx`C9j
z*2uRe^^0})u=d-R=Anvp*Rz$dmz{YqdJSJy@TmHR9_M!Gw;Pyfe~@LXETM00e4euT
z3WF9UZI9muoX6objo&?Vu;8#_v-S4M!FlX)t&<zUz$NGTCCPKwW<Jqhla>MK9)p%^
zVPVCy(~>E<t45$FXbYh_v|$?2jl&`W1&f1Uzbi%g^Zq$xVaw{;y-@u0I7@;qYhWRn
zW2tw12-8;4b2>VUg5|h9x#gA?RMTlJM~lFj%#Pq*Jgm0w+tQnw|7NHQiC0GYy~5n`
z{OF>M<8QU2(%68%w0<bJIkYw~8skxMUFhYw-B|d7Oh3b0N!!x{!@v;lqQG*Fdp%~M
z-5lI+PA&byoBC@6(F9qY@N5a@3(<nssp0DU20I0pjm9sHzMh1lDZ?WuNi4KxD6+E|
z>W-pE&2A!QrTB{=sOM%Gk3Us!yFd<mWNIDnfI1S<TT&tJUOuG+v<E++t8`XeWJF${
z>>CiD1k$gIUgSsdLJe9TC8X>Mzp$&T)m<GhWb-tWv_VVsiVc9mbVBokvEkov_X$CB
zF9O;e4S&JrFygfMLVy3xtgOoD=*GrMP*`zlK|@6#a2D?J?^V%tL+Jl#It!*a+HQ;D
z5Zv7f79bGZoe&5b+%32V*TxC%65QS0-QC^Y2Y0#iepUAmOifpHb@y}5-fK0QdX-cb
zGZO6{{^Bw-z8dLcL9-N)FvJ6lK5>OR>3`uhy_ZoHl){0wZ0t&A5t0c{!papx7=Un4
z)J63cOrFAsOKGjB>9}&`GvM&fKh=DQ$msA*1w-_~$z`1+!m}(Y5qL%!A4EhWhA33c
z{9v_T5}N}8+_<f&uDq+svVV1DcjOEu%f#c>Ee)juQjdk+ALmI}4r~DO&)8F$DX1F@
zEH^>MIPjBO`Shh+bEMq!Zb)$obX$5c>8&LF^uH^;ce;$i6lD^$SWKUp*@oZC8Y+Ba
zLU^{UO)agB4Qwn7_wdn^bXv~0jxCw$pRP!QewBr1B6(EOOo-So7grl}Ot3uWHI(Y-
zqrza~X1R;VpniAMwUGmyRSSYz2){TE`Ac$+wh_b~+WUp))B=tpBRQchZ%l_3&Q7{5
zOZM+;{~Fj1dB5s*{Hbp(v<4NZRBseT;)$xX<<*rDQ&d1>eGVdJz}c+JNtzLD!x}y*
zc1gt4qzXtb%Cc?D^v>)4SWGIydO)-Kg+50`@r@owz(Eo;WjqQv{0w=T!&<4hyr~2Y
zQOGio&I~!Q8Z(wcW`CxFTdzEJjR`W#CP>!cYjK0$fj2hJsVufS11C<kR9*Jgd}SGO
z)10W)@r7J?k&5tEU18uKUAe*F9-`A@^Z+$CjXxB@D9=-v*Cm>8<fWz-7KVm?Xj(?m
z@%<K-S?sM4>J7mC2FCDmU{NBUbHl00Wc?_Ksb&-jkKNJG^e*k{L`J7wp0kc-QCtTy
zEv|U9k}|AndJ_xwO*|CoZtt|0*llio1mn4m7r*wIuWvC2M$JB6Q=Gfytm%{pBc=gU
zQ073}<8SmC&%T1Q@PB(fDd9OOz6oKr;-WMoR^Xz)6WDIDQ7d0JN3lJ`%|7K&(?Pc!
ztkGq}p{Jvw@l7!DiOo9&u@=o+kscOB>?it;?AVXL$ab~gX5#Ub{u==fi#^DD*=zny
zPduU=m`wNEFtvBPl`JbOvtROY%Z7|zOV}4-hHQ_9QK&)ua-!)TM3eQpBMy5DPSf%n
z$Q%3ecL5x9Ul{<S@r(nu^tT7w2Oo^>_n?$>KIe05{<XE4mR3rx|A>$Alv`#5W~B_n
zzM-5(Byp$~zaW^rWCf_$82BvPZVM||W8hFf@n?H{gS8d*g}|fIpw{7~ReBKE9WR6>
zsQ&s`w(OvxZ-AnUV5#;olvnmzbD|S04_@pL0&XIys-Ew+r~#+&@HS~h%+lB2j4?0s
z@`_844@PI|ydCG=0~mkm)fzZ~-$d!YHTZ>1L4lIL-M16H5-fM#80P~9=ZxwMBfq$}
zkT0DaADsPtN`~VlL~vIx`>c%o2#b(}WJ!zm0YM1`Eg~!{7jHW4$VmY!z?WIHZ^@+$
z0(fVfn;$M}J>JwNNLKaqG}^LrFxfBf$x7&@kz6?A57T93ZtIXQ6R{PSeH{kts#zQB
zq@!v=Q@|dfNKz~9qS7x6L$c!dkkFId6;g@-v8$9X_D=4dAV7>I{^bY#)0kFQhqyW>
z^(@VT4WyzJ1JY>YxwVBg=!A^aOCn?ii0#T{eSZ#V5rt_Zrd29Z6GWFgXi*nF*JHo=
zxjfX(45$6ech{tnw)Hn^V|+ust(B@dlkpT^1BLv(_7gQWM}$EJ+Q{BTm5ODk?hgr#
zPX>bV#~3HvEMy$jHwa+!XoOA>RUb5s=&RL#(_4H#`2I@mPUFMug)jk^j5ac$9gKne
zA!e+oYNM-o{5{}pB7$f1wTz>G8R%w!5H6{PI`)L6UmEyaD2Tj8pFoP@#L<RRSwt>>
zBqJw$7}->IoX@o~YOvu+c_eCrm~i+e-ua-+`Zx6-U238Fp>u_A^pe#|$H&q!lA1^G
z_l)+Zh{WoF34Ryb2Pd*k-5)G~U9w0+xF0!zGaG?2u>{rO)Z8o6%j&(=Sg73`h90~5
z1g3^Sa96-u#8K*25%VCQZCWOhs*pQphJn;owc>X7yC7-BJ-1M_*pLI(zZbMx+^`aS
zKDa&;lbD!vh*tGb$}m$0DGR?OgYLv2F<Ry|eD_z}#NVr^bJ(W+G<hb#!5&UI^JZ)&
z;fW!F+~wZi^#$nr-MV?~yaDy_Usp&p8*~<!r}0zAxuW<xFB$kK{9A$v$_p?dxU9HS
zDL?f@KW62TmvuzD%qftFY-0R-j(^epc4p5wYWm})f7Bh91Tr6lf_X*ii7htv03XOd
z#rZ@9BiJ|2^m`JC5=Mss@;|1#Cb(#Ft1KQ0=08C;ti-0xYQXF{mI?I6srh<EG%#+8
z4)ABZ?>K~hIrjJRv{BW>ESmLb;a<{A>B%o)WS|hY8lKoqRN!#F4Aaek{1xQ(6nOe4
zHup7ctRCkQf(bm_ln^sQQuTJf84PRXn&1MJFltt_x}X?<qLVNIaDbk&=N{>pTOUzK
zSW*o3D|TqZ&#;DZ2;4%j@us??p@nth26BXsWD||Q{WCY-PXee^EpeE5;{Lv@M2x*)
zO<bk+Pdm)^*ytmkSzG=*U4A_|JBTOG2g1C1i5(@b4(snmro1AfkJGh_$Xg$ej|We=
z9!G(lo_V~?6TLqG=Wj}`cjoBR9jM<&a>7g@`$<NF{`P$FRIqZ_T|SAolmBT=&;hHR
zlcje~6eA8An|5Q|78>myW~wkWJH-0e;x0CAai(^9zH_8+3ca88?NJ%Yu6825;jfzX
zIDY<xlsUtn1?;TR))GhA@19MSwbUDKLySd5d~<&e3?IflfGE9>=U3g$`~@+TgwxQU
z($b5Sn+gr~POL@yKmX+Nkw#TIg<nOM<)k6=s{BYba3#BH#7L{%LgWa9<!!QS`nyH5
z^I}h1?|sPZHyeX)Hw`hZYfYk~(~=Vt#Dr!b0x7uSmyjecj1Kera~@roAQgjTl+Ryd
zdJW4(wfS>f8=%#-{p0>}Xcx~`kY&tV(C2BR>u%zHU}O!SOUxX#03s7Q2K~&UfTo+n
z1G5|sj?prztLSr^n_px06>ENdt;^rW+`x>`_$|X*6dGnubiqD9>;IKYlPc*;%Ncc2
z6OhZ`dCMZ|6S}|Hae@n`E!F9!2pQ)|s#fcf=VELCW-iopd#9(0)PAHt?WAto3ZLu8
z)hOB|XtQm>dj||N&?@1yEm=};=wPR%AM2Htl#3387O-z4sH`Z+DWhYi?N!+8@BN<q
zfI(QQ33u?F_arpSbNF4tllk(a#~T}lnLrwegFIq+X$9jsTn@ECe;76N+aCDlVjTIL
zpPvtWThmX>$uG4vv$C);Haa}PA|~kM_ZV*pJqd!AT>Yvo(IiBEWP-e<C)feBHlFqz
z3lP<U4`=yH|7zGZAYyQjjln=LHd9P$J)dg|Xf`dp%C5f<;30yTEHyoay+)h2)uEwf
z5XY435?KFyr3+>C?sNY)C9&B-SY}}5x&`p(Yb=}Wt*j194K9qT1oW%$o?D7~N=Vkm
znw~P_o85JIJl6(v>JyfxD2<%13RVOpHJKfqA2$3zZ>y$ga2}PyB8^0?1lEi?@O)7X
za9?)R+i`VCMDd9|zjkxjVr(b=uy7G6hY5x<+h^FC?(LB<GKA`qFVJK@0!d1Q<^sWb
z!ukpb2sPg&0^s1EpIlp7d6!o;w`)}4HM2A|HNi#R-+E&kp9JukZ?$`j;#mJ&{~T#t
z013?bUAJ^}x1xQ|>~MX1$Le*veG8Lxb-56SuI3vYz~15#E(=GhT1?na=ljIVou;7i
zrPP0_B%z#2JG&)|nyRuW2d<MRTM+mhS;IlhPj_V7F8Sj%G-o7;sxZ~rntohXM2*Wp
z79T4K8)wS`<(>BP>ij%fpVm*2w))-Ms{-iO2HRq~b>-Qr&f~r2QT-(rPcr-ld2n~-
zq}l!WwK%c&#d$&1nsHcCGP=<}q<}akV>?wvMjP{IvXo(5!Eqdc$^a(CCmsNY$xF#n
zNx%vAlZfN2C&p&fOOz7AB)ua260PGsYNVWuob1m@ar2)SBRI#Saaf7{;(@_1IV%JG
z(+dj&11l>dBQrAt6Vv_RfE6>#XGY2|CK{fGf4)zw>_n*4FD0CKn!0>Uy@8u%?XGVZ
zy_I#4>tvveDC9d`b6aaaFC4(b+Wz2dP)(#v_LCE*@)o!Q&gf|zAFrZ3Up2%3Tw-Bi
zA<P4#iX>duKT~f=DVoI{2>W;I7)oQJpMl>T98g&wH?2&!wpfbKB;5ra9bjRTrY0x0
z^D@y4jsKFD3YYvLEj2m>UOgdrhJYm@8`BQuwYygTIbk;Z$7v_E0iQDG$BnlbNpKxV
z5V$^izuXcA-Bm;?jl&a`j4qbTNgn>iQa^$~fu>c=f<;}8L%t>{9}Zr8bBtiL8o%vw
z_q^))r`HGLlNm})4xR%!P9@-Dr^nYfR24Nj_6~%OA8BjqX(Mo29&fpjVPI~Z?w+6T
z?;js;Z%2@$MTJHH^5Fj`{B&~C<`)~GvPyUf4<bi#twJL>!({OR?zismekN;XV4H!f
zizC<)$Y<<QNe9^fG;vI6!5i(J*t`=6>FgmXx~min0!GA7pOiyR{;jk;!bs*<XzH~$
z=D$6sTNa|(FbwPBZXI2ak*3Dh7c}t;QE{TvtwJRu*&YL17me%!!0g1q4HtSH*9(|)
z{O8jdo`)^!I*A+d70eE}*AoN(07Bh7Zgy;Oled;oocN<ZzMNvlFf1jO>kO{euy)28
z4Ow9QB(HKuY;}SwP}Tc7C2UDD=mPygrLf2COI2iU0m30nsy2C3Wx-mjHwAXHp*>u{
z9A`C^cpw4W4@2$O+RV{W(a_M}%gd^w<j1`IfB$u4vk+mmpGr=hFM2QhlPWik;17ps
ztN6v~xMr{JkK27qt0+6nRnjsOTIi6J*zN9xmG{2hR4d?UvV3yk56H*nP{z*NW<TsA
z;kjMFH<G}3m5bT&jb&xEtIa`DC6&gU>>3Ji1jvumnx;HJ{R)RyI{T%VP`){tY3X_m
z^#%?i-@`N8vuo&n?|ir87taUtLpE2O;M-<0A-W!_KCY2&1A&|8{jfuMjI&&XkHX0e
zApmmO%V@noRGBdV-j!RE(6@kuGd(RR;j?)DUgx(A4D6&J-JI&ch}^)AlrjXg(Xz}A
z;4cW8)J3;YNv@_L1KczUOg%y4>}G*9kvcb(Pv>^ev)`;}Ng0VicugQUJ~kgsDlEwR
zJR2x;*U&=3_i)B%SM(&l+8m~)AkD@ej_T&vtRy&#TsaL^o{?Uh@~_S`8HqWEH}d=?
z`zCgGf*>v{@+T`IXsAc3YPAau^tl`c5O5N-N+DF=wif%Q+rZN~<^#vJ=RPtra&)Gq
z0_*RL&$y4R80q`r&1pv(z~@KO#*)<GtT%cw@>+)=|C7o5(w&2~(cTV4o}q+iO_1gL
zm`zzPcN1eIE2p{cCdeFIXlo`H_y!kmea1u(+aLcciC4C&P=y;OJJ`y^8f0h;WYIA6
zHS4eRW#GW_CkGjU%w)xWK12tc#g=<$ebv4=8E75`Yg(RdwA<J08_M^GeAk`kQ#s2R
zD97*?$)`p+t=BiI8n5hTt9Q$-O($KlxJW$4F!df?$OPY{oz4W=e8@;nnq9$3fd;3o
zwlT)$TEf3Mb#`1v67Ozte^v2-mH?3zC{QxSs3c*K>VPI){h=<qB`f`~2}HT8fHjIE
zq?y0B)K2JYM7>8;Ml-A3M?Y<kG6ff8JBR@ZTrTI#6x7zsXyhLmoah>wn;F>nl+XQ5
zBfnm9<%KNnV)^5ptGfs$tesY$8#QuQIY;bh&6+`nppl(&X>)XZcwq)WcRu668+-he
z*&>Xu7>m2Tug<byB%_&O2VEulQO&&xELXme`5ubjF-fcPf9lUzn6T6*3*K182=D~Z
zTAEkn!SWol&50HZ<-e-&wu}s`RUO1B6COlS#J#~{FM*?5tq!LX0*`6TdYx|Y_eze1
ziI?&3Z;hL}k!xqX9*%(C`G87@2-wam^Q+<WNHbkcnvoTyMHkJ^S~0TqA0}f0*9tZu
zV|;?S7;2+C3&Xr9w}7tPW@n2ip{AtHeK8yBu=N=v==Nf_H8cFiaq1AKwwGX2vAVeC
zeD!<x{<{9STBM?=<K`ZgEwpc~=DFa$BbY3t`Q{LSb|ZBN{{gJZI$rxr^uI)PzFpC(
z%Yll=^hilD_dx!?jA=c{9rP#Jf7IID%bl8Un=>AT{*}%`mXK8%6!yPPG$y=6;(^VN
zuZys5qf%Oo2TOie*JoXBIG^*$A|n-~MgtwJX;aY7H`{YbTM-wP)fO+7>OuOm)p6FQ
z7$-@*ty4*XmjD(JTHze<0&j@@6H>kfL!fgtj8t!(2Hm$BJY6=9OtfYdZ%Qif$Uzun
zoJzv!daf~P83}mA5d4u%s49-fztEXHO8p;IfglkdrN7y?CG9_n4|ARcN*&wdn=gZD
z%Kcn?In#1o#N`rXOpB}pgHldzk$Oi%PIhzqA>=(UIZ%U_xxr6DaMJ3GamWkp|3E63
z&8i*EF3^z6NIQX+Kct&MdSO$+seYMaNij#^$?cGpbaa&NUA9kNn)TYaQ1&jl)eKWD
z^JPkb3&gwPi!R=e&;%_6cX+c3jqp5y1V$0?FF00Rb@~2>5niwL*z5)i1sXGWqJ%sD
z-~e%<HYn@;E-wsc6j97BB@$>&{-5RtX~h;NIL=GHOhlqNb0{V4WsG9qod_dv4QwD6
zMwhi@*Ox|s<L?D({+D!&H`_t9bv;1JBtx$CE=fXSW}vi%hY<<^q?;FCr?Cb#M}(x-
zq)s0xSc6Q5V!fi)Y5Y<#aIoNz)cR6g0*L&_Ez$T#U|tOF$$^`eDy@+Xn@yt%LXVf8
zd-r!?){+k-Joe!1`#T?1ILEOm7hj~6Rubv<$;kog#+~kl+|g@da31ulKlSef4!hz=
zl;7r0xL-*Hmuh2Uaqu*+^~MK3)V2bv$WrxWOr{swX}@klshOpv`5nJ`v9yDL!Tgct
zhB#U|Vq0@ayeHf$+_NTepAFz<VGG^hA3y2^F!enSBx&oM!K=_};PfY;G2J(o?vS|P
zXns|=OAbQ_z6Gj~X%9nX?wd=Wzvj28uIxe1*S?(|W}&a>%A9w9M;MoN8@}+2p=g_k
zJhKQkdYb-R<uPknCx&_pVjwXE<k-pjE@JeHIi|d>#;W~WEx2;OG?Ig5J6H?5_r%b~
z&V)Le9e;@GKt2(dh!+Cr-yuxbOWI%nOD%X>H5X>G4{~`KNW2}&@PRC=S><@j35Og9
z_;g2SmZBXjbV^fypxNY32vWW(`;N&cTD7LU`h4Zm3+i62Qm@_h{`MPyP46-lvN`Kj
zf3wq%L_%_YIGM?ZbjSzQ?gGKm=|8@ktp#aL@^9?pJC{YVtltCZ`!jFk`C3{gDJ-MP
zNVr)~m(dAM1AS1O-7}@`r|3^mg$ycEqR_#1S-UWPqh-fR-tUW)icGE1`~6OX&NgEq
zH}wnHhv<tnck&~pymk%1F6FcKR%G~0>Zuyv^x~W7VLegPn<vN_G#_Rv&&hRB5&Wf4
z-^tma{~kOHty~{i2n7k|AV*@CF^$t#i0+G+e)y}2cli5&g;Z|Nb$nH3X>09o{#36z
z*5M-aFj2U6#a&f2+3V~FV7+iHiyZki6m)8F*QlSif`aHoPzykH=uXP-E@N_FyV>R*
zQY8s`^Y`c#YqY(8xL<3sZ*P5BsV`k$thRncr$G_&3<#Ys{H^-nOQ9+|V2b(BXnUb2
z<k95(Hu4Ce#sCe27<V=(PJVP(OHlSYLr4?Z5#uel7m%S_ni_0;?EP+!SsL%_p{+YU
zT}ta5p9i~Y83KI8R&D(BgTBnVW6kMK^Wl!D-RsS9jaa-H>C1=@KRdh6NZP&9ASzgE
zXKZAzYd#<Ooh8Xbc+u+iRv;IG`&Hm!*b`poBg(e!f>8>odU1}e`j6esr%(ub!?m};
z<<^{%M8RJ4sI-Y`;*h7V0qFXT&b5Xo@CyYaMv1HL&nG}iQ>WwRDA0FMJ~OU83GIm5
z>*avMMxNjM7X4Yi*X05k1$Cn%f&yT<w!8f%{QlY>mPA)bh33{0Nd(P5@OF1Ouk=!a
zm?+t_jIJ6r_Kmo24RVD8YBf;31$J;6%Uy=Y2ZL5hEAtk*m3aFT-^_fQlS+JC;w7=S
z%4yd=H40EJ&@LcwdEODZ=^lTQRU73>3=-<_h;ShV>Dy2E^I3igWoX!@%(BI&hzf0x
z?*u!Occxor%JRX}G6^jrd6&)LhxBMZqH{rAwdf<uuay<1CZ{JvHQ1HC_?Gm4MS6+j
z8Bqq+sb)#-uwfH-*4WToaY+^3l6eC1_pDbOasY=*9xA3m0)n1QpO5|<)5#1$>XTVT
zxQ#ydArH?UrRLY<0u;FS*+DP>zs;LxZW$`Lv@@c*A@=h~)4yG&7vr%#bVc?mF>1t;
zCX@tm+)2X2PH>n!Y{Zt^<$;Y8(s33yc_7ycCtk4u^g6Xw4&ITRF1GNtzQBPZI$U2)
zfzIO$WNnSd0e99lkLGglXUcW?-AiN||K>8=->-g!%p&(toRkcI4BKVab5+Z1xi_3Z
zR7`e{M<>%M^WSXT)k|UJT`v*z^i&8v9*QkbQ`IpIxg9=A)H5Hka((qEOa7xuk;bT2
z83U{J9=2w~guF%WXKJq;t{?9thym<?AECiph%+gLL$mG9gPu?}XacNZx6v^x(Mck`
zkbZ1r8U5}#b|^w-#u*%4mPkN+hfNTMn&LrjzExeO{}MtpX;U{>7U`&Pn2y+Gnn>1U
zhxbRrv2gupU>D8~`W>gEF8U&inOAUeL#Tam)2U&<E3)Lg)z{9xoNVLqu#X4;P7QDF
zV{xWQW6T4!Q0tsFsYr3$Rmm>2&DP>^F0$z9LKd6a#SBHgbgZwABfVDV>KyH>Cpf&T
z<(T&jCt;(f0+YM^Q>xLaT`cIDk)*W)YgT?XX2wJ`Zg#?Ch-U<N^DlLFr@QY1FSn%K
zkrn|NZ14?x)UYeN&R-5GO&kF#e6rv=iIlJrR~!K(T3bmAp$ybtz^@<Lt6y1~v!3k+
z%N;zp^58YUT*+*!)}T(TI38UOA+$^e*=!OXbxJHjcD_B{&({u8xdhwXulM;MB-vIW
zbI1@|xR9FJSa;4iq_*t}b&~+~*1%m0P802v=lekWVx?}o`_uJ)7#_g<yyBct0H;~Z
z*v+?xdw=mxD#vL8r#Vuy6AP0rp`xQGo%}<w6D8$mQ$Uz@Mc`uvtwh^(z!=vv>T_Qi
zU9o^?#RG-(xp3-vWAz}%IU3Rs)lp}Fm*<-Kmi;PW#US`WlTaCc*8n4iOXEa!Bro<S
z;fUIrPl=?)R}7vh$Ot&aO4n7dFWldWDYq21jRG^}s86HMMyPah_ZY1aj2+Rk_`Tll
z`*&Z!5jsA;DxT-dq1i<;J$~Y|X2*jiUpVD*BeHqsnr9>?kH*9@#uaU#>h2dvz)`5D
z=wP}xrXqJ0Y#Ec?$Ab>j(pl^qnI7or6V}qvAWb{@t*<W52<(OHmCX{$u*PQ{1Qa-Y
zJlj8eI_rMZP|;CpteSf*@hVpfJ_do*y5c6fGFk3FpXhbJ@AJsN4a;j31@QiVn1GJS
zuln{pOm_9>43&1oe!lo)llptu(dfscdC}V=>}MBWMXXtod*?E+0x4FJUbPiurX%H^
z>|gO2c=@df4IhdhFasUmykq|yrk3kstJ~>#--ppO(fWMB?|yeYe`E)!y$n-f@9shN
z@;sg&=|4OrzO9a04s9^ciBml&DzYzaH1TkasOTYTUSifr6SfS%<>4ikAy4IZSNe%G
zywcD~a2V-uxRIlc4{CRV+?DhG3~Q^}P)JDtC|<!nV8$MbB=bf<<cmWU{|BYiOFE>0
zzI^Vx1na{S(bl#eSom>$Or`#hZ?%b9h<->HFKufH532vKxrm-GISM0;=C50;)JOs=
z^;bT9b3@JhlpK^)%ggvFQQ3ffQ#tC?s<o5Ej3n;hK$rOBLLo0i(Y$k7)bkSh&4xzc
zX)M7Pp4r>$W^aQj3x}B3xum`JaG_oC`Oohr3}Op&v81l@*4(K#;<mt=poR4O<Dz7&
zZckc$F2CE9*x1;d9Pnhe8)<u=l2>LI=QMupACZ}}$aJvrbLXjsEM<KtF4F;)#A0NE
zn&Wr-@oqLQI5<!wznkw&%4WyNB>foxykI(h?|ygF-U41AC{4w<JSsBQR5HTk+?_^f
z(~G-i%PabuW~r~4B)vG30(ggp$~0>mvg~y1kDqJS9)%^L-QHn^?7u7rRc<)vkm1nT
z>Wabo|HS7vBs!x>y*({)=E&?6e0^<nnfx9k?EFFJ_y_)i-xIE$xrfC4t}+u4#D`I(
z(2WeDb&H}jXP@7reby|wG}7@vSBC3I2xr8{rW?YH+y{e0;_btnj^vAo_#VX{y8~3e
z|HZ<R0AC|UmwdL2dW4Q>+@j5f!_G*=dpG!?*x~-+Bp_$>!3TZwZT`7YBp(i>CrHrj
zZx<ZE#nv;f9`kDQQ4hX9ae+D&2t-uuCB^qrfx2eT>!68j`X`kH9OL{`#D$6pDFk~C
zwWPg&FS~}+tA9CG<=3Bu6PRojDL-D1jc$AH^?2Q~8mweje*HiKZ`r-?UtaKXUEnM9
z1mTr69jAocBwR1hXPXx%B|*7pKX0o+Ywa$FoHE?Z4PyWBc5N63Wq~G@uZGmGved@d
z@mB}LLTH1OLHRR{BFFA|%t-fDzRfT;dY+55lt=SR@zCm@TAbfU|8UN6VT7Sa>9xIy
zg9N>B^wdsv^(P=m)6qb>Th5314iQ>6N`AzS`FOL4A2LX5N$Hb46V1u#34seV?|&9O
zJEBumgnxP8ZF}*RFaTh>ozkw~dYR*Xceq(~tj=%Ux8rQRW32Pu)kWqEL2uU&n}_Az
zyZ`bX0+h3h)_iM%mwQoBl^I!!&nNwCcr9C5z}1Dz3IL<VQgq*YyypM%Z~>lzwr)5T
zzsXqs6d0lHO7qp#f+FPZ{2SjEnixuaX2F;y5Xk~XQrnyl)Rnd8vMm_X2DE6M3?Uf`
z`~`Wa<Ga~mb_Zde;F87=NX)~_=^Z_H-dFIEmoS{dd*J0gY<Li@O!0}C9*Kb{brcx0
zpj2~;6cvraa;hIHsvmGY!fD_v$7nK(Q}<Ta&s-*(LF)uC>$b2)4QZXuaYc_49uFDU
z3~B5{scp`HeLyGwdd&xkH5BmpJp9JD_%$Q(TKq#OhxtRB1&JFy(_cac?;|3#9+2J-
zPw@DfzSL^p@98r|BLT4+S?6h$*E_%g_zGT(O6qw!Uri0<vsoa;iT)ZT#I+W5zFFS%
ze%nqhJO@A1Vl0-NQ13^Dn!TT1i*w;0cKm2YdSw8q=bKA8^LKLHa+IIVYJCV)B0nso
z-qsodX9sYpgWc4#u?J5D7`0<=Nw}3o2pAJ#yF&u&*^;dIU3b<78;2n8*C&6aFpzb=
zv2|m92sA%*B^iFKJkTV?#BHEIen9iNdpL1HBuc-NYU-fN{o{>FNr-a&z1~(2-%!tF
zZC(a2jZ5u3<h(iQ<RBI@F57Q!@6v)Nv2U^Q&S!D?MWZLL0X_Dh3W!Ao^PO!Zw-t)3
zJzLq271up69fzl$<yqQDZ%4b$l*CVg_8W;MO@N3RtW%Z{wCqY4rEQQi{ii;Z6>Joe
zxBli_B0L!~y+jhS9fMVpU__Z$ABw%db92DAs)vXPx=Jwa{{~zN29wlZwn{;HNDFd$
zu6X7bj-M~Fo)4Q=1zUiZ=LK^PH+%1Uch(JE0i3ZQQLIMwnnr|Ncnfv+?<OjoU)AB`
z3r}{+@`lJ^_RZ+g*z7Qj7sfQ3U-*j>rvy;6XC69TFOL?Mp;Hsb|7j<{M>@fOT?7AY
zLdc<7%;wuCCW47fe*2xrD<jD$Ql1)v?Y&rWAtI;SN)hey{Br-U^;s@*7~&8}V=2L4
zymO|1k1;jb>1yH@2C}nF*Pc=`Ra21v;wqo&$gjMf#T4@jijDdioV}Z@KdNsQF9hQu
z72ooET|Q3wYRxAnSt?(GZmTml+W^PwEx8%=abh(_j8Elt!3r@&8PPnu{w`>s98diY
z5ZDbjWKhteScUT7=u{`tky1_mx!JYn3}@B(Vx`vhqg#xOgvafEW?n0E19-f>*f(lj
zWO8$vIihz^v*vrgdi#ixdL75Z`rx{5`GBf?S&SM!cix`EcupnZ^ZsQq2tX-&Cifw<
zw|Tum`}v7UK<ZYYCJyjg=={iLA>cyqh8&JO!L$gO6_8T;!5@4##o}-u5sy93UdO0W
z>oz%kvk00le|->oe@eoWe^sN+=+$_;o_(0?q?w<_%UtIMtTq=K9u6nd6jBIIK~@DK
z<0-9Q4RzUgJv?jmnK~i^fi5A!v@{Eb(3)t(!{4?>=7>E~CVoE2V>4H55cD0_{v4J<
zYgI-c(`-q@2@i$LHfelq?bgF*^bU#EYc2-|{<c7z)ayEPhW9MR74VeM2Fjyyb^nMf
zJ&Yt<`=p)^9UPU;A0|G1s~|}t?w><~Yp0Y$Nkv5^`WgV$=jAU1oM%DS)T1f2|Kh3w
z?^{Bo#z(XIthprrE&}?0+J`|%Z-FJ*<sJ&!y!s({<tud$OMzN?{CnAcw>3%QucMgq
zNN#<ruTLKMi|e#1=it>Yr`Hs3tG%|ls%O#Eg2J83{HZ0J_2Ce0<xDY-NsppCZloy7
zHwdl+v)iz|g^KS$5S%Jhrh>h@!9Mu$6ESPTKKU=K^qx%(B<TJAcriZ<-&pAZkc#Pc
zQeHVSi<CUy^QldetT#HXUvm&OoC~_1jpd&PFV=^Te!@FKfv9U_;ilmb;tFFE#e#iq
z@<(&Sd%y-IB2<a|w-~gum&rpos#kU|){+n3ZKA?KcLYdJOc>Rh%{N0vGrcz^WNm>1
zn`a=|)#{$RX*_DO(|yJ}3cLO=@@%DQr=8#~H;lCpY{xd?4Hd|A(X~+5sK1e<DrI2=
z%NF}-`XylEXk|6&prOU6uwx2lZ(OA~(F4AeTGGyOH8z}cgOL)upl-B>x$JZzRqG|M
zJuuoBnDo{U5!#g4F(-PYrUc#Fn0o*lv1kjnLUw0s^L(_NOhuDX&TRO3hTxQDT(Uh!
z0=0zOzu3H{p(zO4Dba5;r+I(7-*q*ghf1!XFt~&wdzP4(v=nGyM85)wd3&ldY1MXj
z3;w3Io_|2^y|h)fXKh92yjR+?!EUsRYqR%&M`ML9x0mBX(Ed%KY|)VXHUSEG^%R?Q
z0V{xF^QZ6!Z@(`Iuf$%qNH-cg^{wsg75b7E)Hf9!B!S^0&AI<k4fr+Fl2c;Rg=(C?
zp9a57ta)D_9?F9@JN_IDZIr)6X0$p#OfL#ug?!*+Zf<<LY7=*QIsZv}Z)>f^>jGGe
z#?GfckaW7<ggNjfvWJIb5u#1AIDEoBRGypsCul4xtDzYAnF7vBJl!gP;&z5C1F&?u
zMe)@)MT-+elmA5mBkH&tGJJQyPX7<VO3rP)tl!dwq}7vz026SRD}4dd9y|ZQ3ospY
zdfi{vu99U6x#6dpDEK2otFVYM<pbFQlwPynhGKq58gA_fR*8A_;ik}-95~8Yh-I+2
zA|u1aQHrrKc=)n?8C_cSZHku{{^2U)DRN$Byehxu`Ft9+N#6f3Ct3Ja_v5y^))w$K
z-l_Hi)+P?xb_*t8FPNGsE6#MjwX=7xIwnU(xG0<S?=_buO=R-G(mf9Em;s7aWV}O!
zr{>OU2KtRTWd-@Ag@y8OvqMox3$xu>Ei@+sSVolGy()Dv?%2ecVThFQ6FMDm1B8<u
zkTN*l*B{(hc^MxG{;TBF&#xh3tz4Dvw$G0<jP`%|_Ctr!PB1Ed59;G+ANDj&c4bp3
zbKHe*3(_t&gL9HWC9(Xv`M{Vavf~dK<e|~5?;)W!x;r^7Pp3jH@B82vHwJF4ONCy?
z`|u%JCL~QT&AX~v$l*G|g*I=4$6HOaY&~_JzSXMJ>+a;o;lXyp!3eexk{fj4)?TdC
z&C0#4Mr~fcxu-<O=Wk!7vClx~+cV20eq7AP6aJwm?gex)(>fSNoWRD`vgX$;h~?Xj
zl=Q(#-4^E;-Iq8a$luR_^6x4kA5W9f54Pa8Fj{&pSGfka5kLJQw2tDul>OtDw}4;C
z1^!yhj};z;PXX2ByBlo0vH7rSquJg^Ro!nd$S}YcECvs}6Uy>zzb{|R#J@~brq>$|
z4zEr|tNnS1(^?ejHU+M*r1<IeNV~7M2WT^6fZNwWy=KIt#?FuHbv5!gyL)_kQ#=lN
z0;O0TUhfqTgdV|`;`(B3%$+jJU3zJHLO!Q%!suy;Mc?G1Dt+{2-=wIQou}nq@{6=Z
zHukAQ<R1p7;c#^5>+UAn2Ty}WAS`-<kb));v`lU!Hbh}!N`N!Bv>v6iWDt$Su2~Pb
zyPXDqKMJ;c!3J}ZG5+In(x(6WizoU<MTekj`+wK#n$Zh^DM2TKl%f(XLp3$9LKtfs
z8OEZ%on`U)>_>3{omHf;Z0|_-e3;zuigRX*8xho2!@~+074B){$6JUNS1cvM52MNB
zkRK?&$<~^IKhIMWcMN#rHYo}mtPJ$6zt7^A!_;h-I}OZl`eJYC$h021isIX@?PZx}
zX(O1J$7DiLQ7`XucJ62{jk^RhG%^`Vb0(rYltDvda0!?>3LBMLli{hKex1vRDKRv5
zsh6NtIp0h`4^C)z*sC?PnYxa2UZ(OQj&`rZkXLsCO)iHA^I%>vnNXAE&B}h5rin}i
zK|%uEHIDjKRIBaeq0R|T;EU{(pUFK2{|&Z^wiU(O)`UzA<77w|{=XjP96X;KIF%k0
zOw%OUzF&uP%ScI;P+{?0m@g5K$Uda-@l;z_U8YxyogbIC7YVx_oQ#=#4kM2xJK9p7
z;-cRHgZ{?>7rY8}7KdMVLu0NOcV+a9<rxY(n$iD=84zI>3ZPpMYfCZ?p?pR#v@e2|
zynKRD)JLK1d2y*ZTJOfV#o?M|z8CpXB|;ycu!czqb#A@Q6G@#+dW8o-{n{UfDr)x0
zqe3R)?v7d!>U23ANVqACB3Ws57;Q-iI~=zF%7~P7qiJxTT2@jnzgKv<3xbLuiY=|o
z<tKhB6DcV9k<jYiE7#AA4Ge$kHz6h=wh}JF%(y~`ZV?6P^3x7o$$R6KXn(SRj7!BU
z->5n5npVQA+;~UE7R&%XW|==<b-s#gHWWs!tdfDLm`7LOuiN3yC)8}U%?3T8W)ExN
zG~7P-CsK{D=%Nr<DpFfjR!~t)`mWLNnF2AH-JD?AEpYDx;%H(V%q|@18C#0}Vuuze
z0d4#d-8(dgon{vijvOI7&WN6NxU(%=GHGVdD*ix9NE-yrCJ(L!dHpnHo|(i`4!1S)
z2E8?mx1wdX*afzZ<BNpCF1C2Vm(l^hW}|K2#;(LX8m!H<s$F$&l(+m2^J%}s`ZE_~
zq}1fRvc!gO*g~kFq<#fwr(86N{o^?fFbLhTB`H7wNO0V!GfDnyd|$Gb3W~uWsrit%
z(Bv*?-Z_-^n!3ly&&N#VWL8UVM|1nD4+Nk)FoEm&CJKVo>lx;r8;z{4aTM4(VLsDq
z%&Ptphiw&;?b{*hqb|`$U3iF#gM+(wcyRFB)Fe;pS1jgke>FxnWL~vn6`U4f{CEJo
zcrZHg<|JVik18E^kSJ;-6jyd58NzXF0ai|`Jjz3@*K;on8th?rGZRH_mZ7%TO%D#4
z4Q=IXV3h4T^}mk;o_*Zco_=65*;%-(D_>IZ9tnJ7EH#^-UoSMX5z#~z(@n_AhoYcS
zGS}lUQeeA>qs-954V4~i`3Ra-Wfn5bP%UGJhvYVSxVj=p`DoVF^C%OaSZEi~*`If@
zu*BlH!b*fHb>Xx$WOUO=trQDf+B$wkVth1}C!=bP*1ppdpj{N&IRWtG%ViGpPtVHE
z)3pQhL|`_QrQMx%6{Xzlb!pODTKw@5kEkpy&5KGEgN-VvDbC@-PtpX|x523mu5av+
zV?jhu<$j{QxrY6FQ$9x*M*d?D1#i)m(x*8?NM_<bn%fHk!E$M>e@-``vl93vIhQ7h
z7jSu<N9x=jpJ>5nO@PPQZZffc6mKF!Q%y-V*<UlggA^HmM5k4tDaiDjT>w~~U7Gm=
zS^C08j`M)rFDHPG%=IsMB<1a`D5Q{Hd{@s{FSMEj?4&`O6vQN9RHS_D@4dY!#$sWy
zxY|ZiuSqmEeQ3#_%ZGYGxs*^Yg23hNsV(rl*$KGaZKWKf0O1Ok8ClQKS(X77KDv+J
z5+c&s3qEsE^DfN}r+FT^^QP>TiQFx1lauj9Yw6m9+rdkFBBLS}%x0`_eu@^16!a3Y
zkkfDQ@R|HvJ-Iq0^j5jC4x_@TIo4Pkw~a!BqHqtKK*82?rX2<3oOFO@+Oyf_eMG6I
zR)H+Bz?a}QVBUZ`B;c1c5;OCw<=t6PVw8w<%YocRw*B!MKkjAP-_z6HaJ_af&$m}-
zC@B9eb40t)+|vB)&<8XlIuupP5SKfYFtPm3cx5R~bFa~obXr@%?ZSml#XQJ(t~8pP
z2nH_n+S}bq0PwUwjD3DZpq-l7Ma8=AL`N|>IKvABepUsY^u*iQdQi!+CbrqnJqCpi
z%!JUaDESv@a`SX6Ok<iQ=z<x2JFeFb`O-6N3a!nwq$HLk-LTq|Lu)JDY|3do7FNA!
z)J4d4?<<Ap#8LMi(gnVc8KwJVYwg}wbKC1$3+3u8me=^=Fwd8nye^UVpwYVpfvedL
z{P52}DpN>Q4=h(4{J!s3@93N8DHcUngk~)rV(TW{$oT1*0y1m*ZcE5;CQ2&tY>L?H
z0acg{$Dn1$pq<NiRnV3TyZ+QxwiR@_{t<D!HorZplC|5sUa$s&m=_)PQcA8*9AKs_
zs;TJ?J|df_I-`nD^7B_z8qFH~O!7IZi{Sv24(cjI6*p3a{b>Kujw2xr@##7?e$5nY
zvz$%IzbTY2WBd_;>eyHh0aEGT_J6)R$Jn1qf<x>mwqpdOD2wTjD#8vYSzB%X481%(
z#YK?@W=(tf5H8<>diNTO$jpVsDKnp{<`l6^E7$CI26kzq$b`JbLwSn!GWj8sC#V2@
z%`_fov&n9wLS|j6z>nIfhw@}j=`({L)G@wCyu`Y0O%288<4q>|5z;+t<MWq)k6qut
z88Rn)K|_uT>q{uludav*B>8I{IM_p*Nhw7&OgqgcAHBN}EtIN#_r)TG=DugE={9Hj
zZuSry`?WdUh>}y9k4tR?U0w~Br!E4OVBzcg*)f$zW4T_v>&95n3abTzLH7ZNEhh++
zRm2dkv(^SJ;mX-UX^{X5`M1Y6MC>l=6l~#Az%sDi@GwIboirpZExk2n^$XE5W#dJE
zHRfm1SKT%rC`i*uf7GFQORkDEKDXZ=6-=jUK$F{HM=$M!ZcAj_DK|5l!VsYH3+hP`
zj#-(-4*3TiKA(~KO+YL-+CxOrMbWM=5TKJ*P!Youium#fAEuBE59hws?nBZ>M~QeR
zK1<<#@aYeV09(A@z*lT`t-(1Cj!C_mKkJFs;QGR{%hr@61a;5j(Pj6WhYM(>+Ilp&
zG_m91Vos`Fg{oFqO-A1u`36YAP|&?|gRXV^Cv2eblL+2a&CKalVJq_XxQy0>kXz<x
zGxFhIHVq5(v-Espa;!9K($mo1+V$1!0XWg(u?Q{=G+HmWnNt<Db{ld3(SK-&ux3d5
zYNb*2yFcYl%lSf}&Hm~(B@Lg9NER0vpRy9PMLSWNvBK-%DfJz!5&&Z2SH}G+#hO{;
z)8uuVB&HT)zmUGf;!?0C&Sjtr+#;ACJUIo&az5XDr#3!_Dk0WF+lyPE1~F?@pMYlr
ziw)E2weIsH`ygg8^u5YhRmto8Xr?Dk*`B*b@NTKCz$L0ErJX(P8%&;ri-W_1aQhD<
zjXzME6V1(tX(Z<&A;9Wb-`~<xGyi4D=bLXmh$gzkL|CwGs`%!N9>DO91kwiB;5pF-
zTmCZ%Y`1q7`0QnKa~sq`uP;>Xx(t7+X(~r28+Gr%$|{;|>*bY<t8iJ!e^kDw%)yu}
zSn-*SbE-cFX-{au5egK&RP>i+2IJXP6a#S%okN1$0N;UD1Q0e8m+?eHw#%c))faMX
zf)eV3Vk>NmG0wF-rI9r^OXIa37&|!7U^kga7OXEe7}<MnS(!HOf6|EU<8Cs9q^<Bx
zMo3<4TW;&zMcY1=jnJ$*gl$wk_D&=_Y3A-NjPy2ioQ4hGrBLv=OSFk#xzeSm#+0&P
zmC@KS#G+n?1*lbwP6=ri<%#w93R+5X+|L4}B4$m;{|R}R%<g?MDwUQ@R9y)7Kbtfs
zR!dAG#@RA48m|o4>gXj3hTKiH_*v6OA=>&?A5S5t1@^mGND4X$oB}`5Wu8AZl#IG?
z^z2&DtlHW8kSb?Y_95=A7>)?a(!w$c?%ed%u(^ivBH*rUuEFyO`}GSGfeL?35xEdE
zmr^V?+O52tyq=oazX&RK@zs-k-yO-a^Xb97FEeQ~8UnWLd-|3Vn}|p2l5`3;5$_0=
z#zJJ4NFq^;A@Qz;a^=lcO@9Za^Ycvc!oQH|@KxjXBeE%BDJgEZ<g_%@a7$v6VeFk#
z+U<m&E&&_HO%)Mp*h}D}H`BOClmnwTpTg&0i<D1LMr=Rr+p!4f96y^&c0Jl<c;`y!
z8u0lR>izirki<?RI#34l!mH!RY-j)QIpMpy_;2h&6M3i){clg74TmJ~Av4qkxlMQQ
zbgr?*$Gzlizk)#e3D7H5UmFVCtE%A`(z|D}AtM1H6q#}oYo2j^8}%LnE2q8ypXIh%
zmt<}#DjldnFGHwzrha>S`)Su#cck?;C{e22*jb0J8aSCP(WCNI#^Isi9}<f#!enw=
zdIKK%O_BAc0eP;kWy3~49hACG330}FyT1OG3sn4P-%Z-^{8|yi<@HOsAkf06X!Ye$
zuA3Mb&dVY*Qb%9nn+cmsM->#{FG6i9MZ%V8L+AFi*B<X9Pq1Y@{TxbBBMjdyt>;Ad
z2r(OQJz+A|;<Klt1|7r;ot&ymi({<(^9F<Z_aDlD;Se92_^)!1aY%*mBluzyUfo>_
z%k%!&I(wsk5zx@!`Fg6hgaI<<SdV0iUn4I7i!}q9dY^nHnNkXE2-cIm!&x?K6;jsd
z4{6$QOD%3|8G~PcQv-S1nTHeRgL$D*31LX@j#7}PYvp4fNX~x}99|=C3HzsLy(UJp
z(M>Nd&d$!yPfsrtRB)Jeht<-3hHG2?8VOryMSw-lGJ7U{6=ZGYn-)Qpt_CC3mfyt%
z@PA7^mLHISd7)&*N^I|iGfEYPKNjVJoYIcCQw*G%SwVH7Gy6VNRdpc~ozjwgpvEOz
zM$a?^k<1*YVUqp(f<TMgz-Bf34vFZ?OpkW>(|ORC*dB)#6@~i|MhD|>Vzc5?{Gd6P
zriavy0X>3Ta@PLr4Cn9USp%EOzsTV_fy8<apm+l&RqM;KX3L{w{$4Jhn4F2GG$nnF
z!nrmP&JSU8n=U&X60GQ^!$V-j9gbYc{BY}R=j-kD9U8KT5FT@OYglc0U`$2JOqFl@
zzEjznDCVCpG~lD^h;-zqySUK!-efev8j2Opjp=2T2&4t3G|hCPQ`1VWw9L&L08<?r
zj=Y>iC$1H~;q@aRt!$s&Dea(kf-kCL+v8QmcNNh-If_kPV8jV-(|`}&mHN?&gO|9R
zu_%kYS-gH6)_XcO=Y$?hs$MrfE`u=$EyI!CeTodqr!i_x7}jYC<~f@UTP|N}M^?lo
zel5`ELewtc_WrX@P}gwxM5qHFa8X!fx{phoq`}$|>X7SP#R*&QZkKh;=Ywq<`lV<N
zTqw99EJcbHU1O>bV$D>OyOqm8brq&T+CdT0<S{?^x7sx4b;)#?u6mc<=0o^J%PDJa
ziaMgq&+AZF5KhpY_WDRN(mai?6eDtPY!81%B?fInfrt+_z!_N;EPAvDu0BNs&gZhF
zQME+7GH7|w;?p7MpMrMI0@$E!t}0m^vuXs7=L}~;Eu?Q|c6y-$<o;gtm1JxWSgpTy
z#Wa^xP%_i6jdrVBvkdbs@uBrK#?G7!lfkvRbtQ1OADc3qlC7t@NJdwuc%`}l@DX#}
zD$;7Q-uleHKcJFRoQB%~D;{3O<B}1V{Q7?N1!vuDdI_VHgG$Xc)^Zzm^OBq?Thxi)
zPus-%-hV^C&Ig1e!P7hbK{_JD|Cb`R1G(7?*;d9_9Y@=rl}LzvU=g%X0jWsuh7<Tt
zN?+89>FrZCXiGH*(%@T&Igi91%#AbaqKc>ksC(;HCIV85c2ZFoV4P}tj}X!>c5-Fb
z@Ym2Vz2bUjYh#>JU7qMsSJMIaD0Sc9bEku4=()h0v$06h_dN!OI(7xnFTv<#Y`?JA
z<YFXM(4*b*TCp%3LNMCKn!?Y2tu|5sugrMY2&~|Eyn1w6BkPPPAH|VJmP=zD3ma@*
zJccmUa@7yr-k||tUWoEr7@J|cVITwq@~M=#h;m-_x0q`_gH)d|BRv*}u)m=cT)wUT
z%^fLibI&;CSIFVx$NC6ndg!+A$0I8PV%Psto7S}C__#`gC*ff*gjXbV(7@x%SF<ey
zSt?8Jf7|xCJx+&N?H&K>a{+BR<KEYKz9`U>-SOG;$XXwupoezdeOeNgC09B{i@M0t
zfCV>L&|Mg&^@-_#$~ZVq>^dfHDoFopdT7<*16Igvbit{OZ9!>S5yVPM9%8rJtZ!S2
zK{kbm@4L4kxi>s;fZnnaIDjagE;Ckso^MpN$VvQ4D<)TAbnP)LwfpdRqetk7m*eN!
z(a{b+QhEa5Yqf7l85x2|(f^4{__x$K8HA->*3GmeCagy3@imuPizqpHi?Y;42rF3^
zS>V)-rFr&XIgDJpVbfoxW@YY5gezlr=cq~G7~q4u8YQ0{Wvz9*M2%u=<8uA@NtkZu
z^9-3DvD~9G(KQ?|CY-0}xAO2T<(5(*t}(-<*Z6_r?WTee%a@6E)gh<Q&GRU4bA?h5
zZ)GmmfFMZvW-qDM-)O=65WInRh!H=rg5CMFj=rE>>^vMrk)+hJfU$ti2Q^HR6@pF9
zj;A&M>S3=SMQ`(nzcr!hE!-19{s>G4pD}#6Ael6ocy}p{1+0|WRiyfZ4%sx>OQ&8f
z6-I%=p~t!=7BHuXN;dEv+l~sJ%H4|rT$0`64^iJcG%*DOC_mqRu2T&>CCM8n(GKVk
z*`Pxz`SP&>75rf{-R;`VB~xNGEVQt;_XAK<ZaXF(8>XqPv7ZhSQ(FPdI_o;ijYSpP
z#*zaOr5{cU^K5r9Qlgb~xmxH1zV^*_-vY2?pYnR_OvYK>eu?~oNhTLPPsApGfRD-<
z>+VImRN1GMeB@3s8#)4A`=PDap*6?oq4g=9X4q3dt3|!Py(qO2P9XmKMSJ%|t``J#
z2NMxQnTBgoHXBznDU}P_yC5R|r^;%u@Ir)^*6QHQoPr-TE1Qc+Pvl{56z+|<asYst
zi^I+Pb2bYbGQrq7*m9#Am_bH4E0&+Ti2HHctf2+c%Xp=gcT1(dyuZ$wv8%^ml>Kf4
ziqz8eWw)5Im%wq4;<k>i?arS+qTKQ4r};LfNlY1p1637zKS1Cv87Uz#gp#s`nB+t~
zjIDfUR#nlQ-*nuDPMh<f<f`gKI-ngxata#q@TkqrNKuCsr#bww(_OwaAwxFx?P7zy
zSlOiwjhco)_^rIMu(Vu)7v+yC5`SCv4g2(QUcGesm0tkrQEqPz9<acM5Xpsx7x4I&
zc~3^Ca2(Z%o$m+U>^g{+=nXNjgmuI7F!HnV2DQR|N8U?+PZMbKe1-B810*mMhnM?v
zPYWh>YTT#CLX0(b^j$33AV4C$U9!b5x7h`xkQiP5BCcdn0pUq~L9RTp_gVvxpNbW$
ze$T>r3r<xjb~pDz#8`4eQ}S}9IigEVeCA+sZ70?VB(?bpbt`us2{+Lxu_Y2Z`jdR(
zmR66@F2}&n(8k2jDxemJ9LVl+{7hItWjt0N;D+)2CPw((%WXU$h`>oD@VL5L)T1k?
zba!4jm8Xbo*xJwlbU=iOP|CB#8(TA_hn*BFz)Ip=9AqJu_kRIk2%q;rKX-t2l4ett
z`I?e*NVDTM_7$^5W+bl!nKvhA1};e5aY6dVN49_LFoMtn=JU{jeS0z8J`)pvQ!_f?
zvf{$UFFY~3WXzQ2iKQ7)Uf@p1>OOQ!Pl}I=0g4JbCg-M-gTc+tRF}?Qh7aYi_i&p4
z^}x)=i4Nnz6WZS0KO9VGYe%JXOTq#2YI^4OZhk=TaKxd!nmXqv1$1nMp|Zdmh8VC$
zuUXkO5`}4#CV^@@fIPFnO$ge59T5<}|I_{h1<o#U?mc`^KRINADN>OWoThQPF)=c*
z^Cq?1_?e48T-aEVA#ycQlpCsH>g*q#Ry?-4d{kjhMoK~~3$zdb4KliHmPfqK06@jk
z;UFk!Si|cCf4VbZb_y#eOdDI28ZNZf8BUo!FnNy(@o|GefD$9>aXtfoGEh6_9c`ea
zuB^}jjK!9XK%UivDiW%RKv8r;Pl7G~H3J->NgoB@%8-82?VPF>DTi#xr3f5JK;5dE
zdq(GsnYnD+iLFZ-OB4O9RON>2*my!A-#l&7xT;b48Od=`Je0`Mmb^^xZ9>RHuz^b-
zQfPw)gRZ`Tk+G?{g_X5`TL7sFK{cwWI#=xHps&E$--!~klB2;AkMxx+7>$<&Q(*2w
zy&7RLq?H(moNg^$Lugq<aj7|oKl$As_8%!w_lrZb^p}1j@+#CGot(7}R#?Rwy2lhv
zS-kD+{j*ya)#ZvDHRLF}Wy0*0tu52)YfADYfV!XyH#%S^IijO~a(M;mo|nB_piN^y
zjuks$XhL>5XhzMQR4NH^F;;wMu&SZ0Fhrb*oFbU<ke3GmUwro=OQc}58_M77sQ4+v
zb|CMDx)#*`(W!Z5h(GzgpY<Oqd?7lQML3gj@(hfmKFp+4W=h38(Me9pOc;%N)|*x>
zoHeO>WQJJA(1e?R2=Q`oN-{8b?q7k32$}`@Mx_;xo4#<(&K0x9=0*w4l-?PtVdm@~
zlU_WoG&@n`?P#XeUHx7r>t$dogz3>lL^=1s<NMbQ{Nj0CB`FJznMgB_5;gm<oQVs!
zo^3zBy|q3soP6;kj0B;QvdQz;E?>~xFs=dyGNhxj&{vm#=PLL<eY!|IkJ_N?Uo%o~
z!Pu>`epYMS!l^X{2|+HVyqRA%I*hs58f(fk5EZM}NYH)zKpB+a3s^PYYe#B5sAuC5
zl8{?b*9`Htp7|dl@Q$ZL@)5(Nh%0QUfR!S!<J?D;EqxP98<%W7wS8$*d6J*Cs+_#O
zy`MO@dg{V|mKjsVk1om2NJ0$<cuyGSiBG(g_(BDf84A_b-q9;moKajexw*A@a&-}S
zoAT7w<ctIy>KJM&NN);$7^e=~8x>{Bnnj>uXW>U)>RMR|$kZc7C@APyxs$|K)}wp+
zTYKn#gutpEAdeu0FpOQ39Gsqq+70RV&In`Iu#Ad-#zpH}o2oLy-At6^c;xG}hB^*u
z6&vO+bh0+qQRP1HzIdT`Ogz(xvbF&ODF;U<&*12k+>$YqW;Tqk8kH*%M}RhQSa?KK
zbc`s#Q($YRuOTfFSPgFsTXn#zYHRY0;*RRooW?Xve#w7BtBzq>mTti$@snF1-s*#Y
z+kt(5ycF=jo-dysD#`%Jc3oSqsEpD&u&<Bywj+H!Y3H0sJ2!Gnb!Ay`VO~~xsyHIp
z2a7@GKujw+l6XaVZhg);S#A_t?(8HC2p6a2l~hlxt*I;>S&);Fnw%hx!5J|s1Q_Y4
zE6Ql{a9z6tm`?$)!US}bQ7EQE9RSPEeuIX8fyTKpFe|i?bERSK8kCS%HDw;e+kNtH
zIdF2PrKzbw&Q4Vd=?8WyFAZp*%Ue4QQ8Tvp5+&xA=Ou-EJD8{sWt1hL0_6an8s&(*
zTw&w_s&~T%_3PEUQ{tHuSWp1a+uU57J$yqV<5RN>%gV|tDk>|hs>rngf>z{fcK|tm
z3jkTM0|VxCASLPR<K^k$?&>0-BottIc97*n>_h*dBh<ja#MzFA=&6~xki?H}m=E!m
zAN|`6oYkwSs>0G+hMoFCb4>mTlS>Krb0ZCw;vgwTBRM%GHC2+9mYyt*3<ZiR?~vAy
zRB|sSo*&M{$3}@l0(?B(oWTh?GA>bnl9gXvHhS#DDN{*{P@D<4IT#%2QC%a3x50Kp
z7#JBY3Jne<?~b=SQX}=XrQ%2LFq1H@E5-Sr9+)($m<R&F^L+ut+kW<MHE=t}G0SwM
z)+DU<hPtUg$DSldw3VY2&<=xD{B}i=(Qye$scD(1iLntO0p1=0PF=LW9HWhYv&;eM
z!o@El3BcD;VW9j0L)74~h!~K{s;QseviO7fjWq>{A?}vsAkx4pPD+&Lnb@hxiScnU
zQ4u1Fx&bn@jLQPYWf;WaN<atL5_Ew3C>c8kCgj&lT?FyApZ!}6+{X_ZhQ=iNM*2K-
zwKKDVIaNuuY{{#j3z7<6enFytXmL_nR(eW8Y(!X)AGL{r!9cIRw1Ji;o?RKmx#@|~
zVg8;1S9c)_N(aYfj;ftHf5oQ#Yv)fcO$l~3;N0KIiNy5Ox%(xxFfS)7Lz0q|5EmKZ
z?+Hs}DW_WM0ISSNeE>AaR7ZJ8FGWMgfQ0-p(-uR#?Pve?gDDkv_8EMC)vw|dD9li=
z9#(UiX(<VD(cz&%to!FDC(^_Z)3R_4j?by62Srj8@#6*b+(J`|$G^LL>(Ofm*SuSk
z7G$diHx6AMaG*e`!+lLvc`2pfCdEQw;b3VjL!9P>W)acrD~ivI7x@S*^pppKz=D5#
z{+Q`YA>Qmm{a1+pCjLJsH2$0T|Iqkv;{QYAzlr~ki~j=_hl#n-Y<?2}0000<MNUMn
GLSTYEW_U0F

delta 103067
zcmV*bKvcigp$5C92Cz>9e^{L7wr*p%yUPr20|NsL3^usCySo$ILeP*vNJ5ObKoa8a
z?(Xi5Thk_$y-&~C_gmjDLz>=mnw_3D_qOYK&e>%G!|=bKEP2<vz6t{K^>B5xWt$u6
z>1e1Z$ViBc{Yt^uIV`7P+TgAg(;M?5TuhWkeMQ+gaTz5IT|;xWf4Lz;LrGSA+*hAu
zBC@IsQ)`|_peN79l%XaoD)^@{LSmA#3d-u*>M9Ddl48PR{>S&Raq<lI_wjIZ;@YxV
z#(LUn3esXDK97iuijJ`r*UitJ$6*=iswqf{i-`!28#{W`|NVXcD)u+<e^L7z_`j(A
z4g6o!{s#UpY9V0$e_pPR97|Ju9Sv1QIVpJJ@&??%)bX7IYv*-UrUyA1DhM9cSWzi?
zRV}8Gxe-%K6+h@vqsNRD5*85^6BQ8_8aoC(R(NF`eg(ZGEHrN1*fC?!>!Z-S2umwz
z8CY<fJe+JT47HS{MFgK^jF6~=w49=fx{9Klw1lY8NZ|21f1+Q4-VFZ+^gha3h8DKY
zUhZzrj$B)7D+^;i9d#vH3E|PBiQhy3Z)9oj>dWVGEDUv2<;0PRfr*0sQP22`v~aLM
zZ#PF93nK>LPCRj0CdWIjpmFMwzNW%>A6tF-5x|Q`$||X8F%6m8YD%(_qGP`rJyuv$
zLP|znQC?O`e?nAv+-UrYxRi{Xg1nrpv=rKBq5wZ&|CNxWyqc~Ni^H>J88OrqB!vZ^
zW%M{<F-aMOyrP_pq?ibLAKFHZ8He5x{Sqk&ar7I;!22kuGmKcaPCN&DTaL9A%fc8f
zTxEGFF}%!16N^k)n*iR?#@s+#MOI8)LQ+avMpjN9f9wx@%wL{Gfd%`xIa<S$p`W;n
zj+I+@Mn!9X`}oWVcQ!*-0Q?wXaVdF(ysnm-vb?mo@L2SUn3SBNs)mlHs*;=(yfPNO
zB8#@Jwx+tOGFlJP7<&ZBD0EZEDrqu|EUYbzn3~G6;zEMYGJ33#s5nAi0euQ_QK7LT
zfFC_he*~=wc_kGkc^N5j_=M<vlr@>g791OEww0wN%fj4PUsqFA5#3ZHPX!FTkDG(F
znZA~?jEK0Tw5)=Xs=6lFANZKRJc|Jf_2oOTO_>@Ba?<1zm)0`p1;ppobk^o41Ugx0
z%ZvaX4rF;HRSi`H0zMwl;V-A6p=)4bz|d5Ye@BP^sB!2OWepvDV?#ZLwuYKAdSfXG
zF=2RTxG7XLbq!1n;hUvIM_kFownosvr$F0!1n_umDXD7el51-W`92IoGs4Wx5%3Hx
zH6^qb#Q=Ih9s|#^b3wqfP4zUDr0{w~D^$-A><@g*U!KK-h5Pdztc-P4&_OCj9Hf%!
ze@3<*VabJcMJeH)cE%b~f}g+#JM0?;S!o0&+I&J1aw=MSCYH8Vruy2d@{+=%&@0MX
zOk<Xfm4&IHo{pxP;G5BbAuX@0rp3@wCl3tTMuDp!DkdQbpF#+Z2inGpO3Ep#>*$-n
zaV{q%I+lDNb#11ezJZ~Uv5Bdfu>k|Qe}#&&a2tr=9|zzK5%AuwT($`l-7Dk7q!qA#
z!Up@}p7IxHNnnuy?hclQT8ij_iwI$-S3*V4(#ba}ttdUp&)G^}RZ{R1)V`FV_oc9u
zqK2L++krSxm83;RiJ(`QCRX+?j&{~8Q{)W@{55oFi%TKjT~%2@77pzZ9vtDoe_<z)
zI;<&!P(oeT(2Q+nsH>qQgF*@Febh9xbQpRDMkWY&CR$?XZ3Q2U0N&Z#)!xd4p(Zae
zRt!ZXXhCv#V1L|G{sJu(EIQD`fn}f}C%7HO6m?8(-2$RgqXYPMrn*Yvf}cR}fO{<l
zw>EL2ipZ$yny?++{XKZrC_<AHe;X|-tEvlel&_~N&yEfLwgzlPh!=iwo{zl9;jOXn
zE=0IzBmPB5LS9wJ(A?U=#@tX>O<qz+@IDB8O&z9zk*TqPo~{<UfH3kRaIFx)BhTN(
z-qKiC6@4%<Sydf$_&a-n{ZY^Oi?j@|*kBI_3q4h7c-+_>{z_C%L!V{OfA^0J;Bzev
zG~~rb{KzN*ALKlbg>OgSEU#f;>EIq173AS)g$@q!(PHQomc5&QM6jQS3)hBaEXWX0
z3rAG2%077O(W4>qBR>Xh!ct0_`exRS?#MmY*Hn@grsF$N2^j?y4Q-|&0MF1t-lX7K
zp+i1rJC>1-imd1;F*$WTe{+trH;!Wei0AvoS{7J*h$q*ap)4VI*0ewDiuQ_*ij2sA
z^A>PCnAp1cqvIivZ*Rs>mK-giq+`Ny@ry``MWLdrBL|Wz>Pm7D`~7={Q!#7@zQ4Nz
z+eBAYcEsly2VNd>ElrIL^_jZb1oFfsLgWl=oxGfFS%%t>GZ2;6e>7w}dIraV{hOcV
z{}02?U8uK{wUL&>h;1!QUD90S4r?e%jfl319V;p(j?Q`!*tX-ORhi~?ZvK(UQ2~5=
zb0#`DBo(y`tsLC_BSHh=w70i`?2x*m%*eFb$LSh4y+y=Our4DbD=RA}CnpcyCUI!P
zjyJb)bmwu*4YXj#f71r>p$5aq+}hII#E`h<Asj?MHj2J+_>LS4108hM%d6{|ah$z^
zaWM10|8&1#!)Kniv#mK(U2a4w3mkQAeN!ttwwVFpsMG(1(<}%vWhE4#%RqKUUPIpk
zT?-M35&mv==1esiVM*lep@W(4=i}wUcXj4*?ZDwzk&_VmfACjm!LOkIX$%qlYv}0c
zG8jxf<gSx}`B)<DV`yQ60K}=+5x~<KA$uE^Df|S&9j}*hW5ngvnPxUT4@gjPAVEd0
zX0O0Vus`7Oe!1oe=45MO@L~28#XX8@D0)JVQb!$m(Lx_2aHJuj4#%pRijur6z7@E9
zzwr1lUspSGe?7E<q!bWL=GI(SWG-ON$gO66cyfrM>(;`~)i>PF&7P&NftDThOD(O~
z);1hlJ9{o#b~>u^XieY(0c>jo-!IZW)Oqgd>R>}8h45;G?6HKRrU8WUq5fWOJQQpi
z=opw<+Vb4I0>Z%lfXDmg8W+sk9G<V_h^I;9Wwn_`f9P&TakJRCQM9v#;n7BG!rX+6
zpVXNq==${wjScm2u`}1#P(UjK=T<B^c3ej%7gxTA;FhHgM~r?^X~pT$WW-3_FRpHU
z4^J;2U%vo%M{D9nA4^^21|Ow;sA^3RMC&+BBPl`NLQ-f+;u|N#-<$8u<ye}stnD0K
z+`avSf5HBc2mGa)IT&4Tp^Z9qwNZE^CyhL%Q8e;Gk_yDp%H>!QX;Ktt**NljL!yGw
zonc|1sR+qc6l3Wbnpm)H9h~`Ieu17&zpTYcYgsr2B^K6CX(&n#c4q0wAPCi%=C&^0
zq2UoxF|qN9Nhy*3uC^vRisDqdMvI~SQQ8Oje`^L-4jw@<A>J;w$mLgnI*h0&ATNWL
z1iH~9Lj(Q1_%4oU9Xh-6J-vMc!2Xyg{Dm3=?BgUFRWDIigxt8O@YvBb^0dS6;o@j(
zWrh|FB<*;7pP-09PXHcmY*{%4B^5}m7@#}c$=x^jSHeO$eLKIzl9v9JGh0iO{p}6p
ze~IY86-5W>nc2Ddg~g@iS#g1`)&^>lf^-dJf5ks2+=(e_6F(Il<lz9nLj(+$-yn8F
zmq}PcY*ct?kiU;7-`yRZC0^b>zF>dQBmOc?84PmtIPj$nr$@lSCa%kIIKC&4M~_o~
zM2v1>%tTkOy%XOjFw7s<T6I*xC6||1e^5|V1XtY2BOoFn%*WZrSX*%f@QS9cQH3qD
z*B)6nx3wsmZ>Ee6e|;+#|LBa;^2+MkhNcOv?NzyP{*LBa(ony_{=OXT=hG&x%&>6u
z3QNpOj_`G68*0do6GGuVLS9TvK@;)|0Z}PQ32`xz;i18S0sek|zCPYwe<yGxf57NW
zt-$E|H(5;=6uDuCaa{c)Q{p0meB2;LH?!h6IHQX*#1}<d*z1;)m64W$yH-KVh|Ti~
ziA|3Ua<@18AOomk<(pRDzxL>Z<LhQMX82pHeI=z~V(%T5QHBujn%vVjW9FRB>g*^F
zYlhs20zXyq7yJdP`fS&L*v#_mf4Cq%$4EnlOzen>ODJj^<KLc<mYR~3m=G6>(GT$R
z_H+mPgC6n!qXrv&G&T}=ab+fp=M#}qkd+K&Jp{FtjXm<1&_(V;Momgkm`6v*7_^C*
zEJyF~q`c&CA0A6jMMCh#>UQBplUE$N_n!}rtn4X?;A(y)t<7=?NhxXWf15FD?)*hd
z2Uo71(_Wq$z%^7D`9Vf~nA0QcNhyW(MXBLlxSm88dz5ueIWE58iP_m%S(%v`8R_ZC
z@zLSH*@1Aw^^bdq|8Ll+kw#|EY3j?Uuf&xY=3LK^gzVI~@BmL|J4<6TOXMp%^F4jt
zop4Z&B58_#A*W$r#q$b{f6s)o-Wr|t;`EkPW&39~&)InT>6y&~Em?s!>QD=I3QQ=d
z>z>j(b=r*nS##!2s>zA=v|-A@2?Tq796o+<o|rxZl=ZBf{bMrAn#(exe7L5X(&(SY
z!bK%6sjOpc?F8pdW>!vaUO{1TNkL9VN<vIT$luQ0FAZ<2F#;pXe`Vs5f?6k8zllzZ
zj|}q0nLR^O<h*m8-0+o;HoUT|tPD13X%$@)Yo2FNbV^JJ3iT`uw17w<0nXxAO6DGM
zWnGK5o!h>+yCT8ULIoZ5rd;3XtcupwwzjtR_KuE@n!NY`9^CO1yGyWcFlz%{`ii<}
z8HOe0RZbkA731e%e?|c>f<Ckoisl?W0;7`CvvLbc$}4N?YsQxr=4PfQ#)AFX&-zO;
z4EtE9z>FJ<4Z#j2SUCYlLk!*9)=oZQ2|(O&0OAhAz}Vc1W6uNPo(X0=Vs?{)qLQ+<
zfjP&?!#^DP1Kw^>S`*Ov6pVNwnKe@uZ(G`1mmTV4tcVV6e`8zE@RXv4`nuYh>Z;1|
z6*=)iZZ?L<t4Gczcz2S(=Mce+0nE&VfD=|>Oh$2Sdo}U|$yI{S0(8e~L8-_$EIuVO
zuXucIQ)^e-ga(9tQ64T){MirtOEU6k7Z~``vhs=$N0F5esumX=0_0T-LtQmZU40`n
zOLWnr=+00Ve{FbhzcIlHXMJ%v7aHKrcjDrtss@pPk!9F;#^lvaUf5HgAM4H0lS6Kr
zfwfyueD?VA(&ECroUDx0=m0l6Qyt_If>)1gu=M)F)C>(BeG?lO--zVgipEKGBv}Xp
z59gGGf`-1it+Q89bW&zPS?z?5p8jb)liFJvYb(pZfBx);{UsUY%#$Weq~%rAGzB}A
zOtbieLh!~iW~h@@QI<6@w24#@k-^c`Geqlw>%{l+2Ups|&6&rwMQ4H*&df_`nmYK!
zW>>aX<&ZHO<nq96EX>P7URX>-sDOfv<9PI@s=wsiv5@x`mXK9aGjU2|QF5>gB!cKN
zlDs<8f7II1-9J1bqp-56vv1DQg#)vu_fGC?Z3g?ZANChyw8unKkfDI4s-ex$XK1P_
z%0j(FL<U{lYzKFLPn>|)RKh7Wrh$nCFfO#!(02)o%P6R7G7L;vY>vI7vzv#vF9gq?
z9`1ZMy4**Qd?Hhe(lE&fIsD>Ex@Pts!Esrse+e<+LB1X?_-??oi!nl?XfY|PL55gH
zQcP%!AiwJy7MET)zOiHKgwoVdS4#r;v6w0%r>1L+zQa2>Hlw(vwRg_4b;B!`Et)rb
zM&A^$Kl^EaK}H{v5)$IrFnv~2N6$#WjilKP$gwli*HV!M8l;*Q10iorGUCRG$tbGp
ze;Sxs*>N2lot$0WsC=29pTD~!Sy0DZ4=z6-3V2@zn(|`Ed(bw<2X|~(fR`(v&eQ@1
zw+P9wgWp0^dx>TV2flxJe0qL)-Nf!06U)=X-K=zF1w4OQRUJbM4$s{`I-{h%Yvz)*
zySHy%zh>p~#q;NY{n=0Z3o<GKmXZX6f7}kxTXgl2^JZp1lHEv#9oH2*cB*m`<3y2{
zrHU?kV|{dgD<D`#i%KhL>X};EyHJL9y_au5P-u8mAO$lf#v5DOp@7hYsSam=z<cm@
z`vvQFtcV2q9a;=SguEhN3N%M3AtN8%-<`d)+AA|7fe9|)?#U`^>6@WX>K&P0f7;YL
ze|X37!w2^4*s@{mszI<n`)PkcMmo<@(qK}^2yUyf1zS)xC+TlCPy{7P;|LX4A3FMU
z`JAG+qSKx4=HkqAgp1xUI6NjHCCcB;7W-@PQ;KjQ&{%@^U<SM>sJ7^-%ZmweH8d%k
zuFl~EqKjWdN>*WcO>@VT>2tfPf3u@KteJAOqe3t%Ep2)3;pycQXDr)%_{xPdCypN6
zyK~znus`)t|Jf!CMw1(94h)7J_Pk)nqqAOFP7>{A8D%X4D96$CH?pBZ6KK>xwI4TH
zOioqT1gJ0Gp6+fgPL5nV2WK}=pMc<S;B{Kz9WVGPiYosB@E;OG$jieNe}4=Wc}bBm
z5{lYJY!swNCuQW7RyVeFPisL(y}PxZ0uDzZ=OiM9eg@M5g5q<l+Gi|Yw|CdJ%^TLN
z8d|yt>`!~>UxLB5mZEv=bZtf&r`jdxJKoTgyl+LvN)Wy>-;a;!Z!#j7L<JtVhMF?4
zsKkWFp+JghW`q1`UoQ_ge;1yE9mmdr=fX!$UVw+AwJ8Jcc>1U4#tgbF>w|avFBw5v
zdM3aKGGrh>LPA&y(=Sl078#$KQ#`)DrK_$mDa6fMPf3_INqJQwp!AJM$*-Q+J9lW=
z;spb<`=|9z2K)1#`WIk|U~**XRnVe_fR~g2R-M4B*CKmZp(+a<f2csWvB2<<{Y{E0
zN<_DmD2a-bXP}?AC*Rr8p2N1Wwc|SSke>(Kd+2Ey@l*0ZkwXz0E~BDAQSi>R0SN$i
z$^r7a>Pj+VGRoMCaQ5^Mi%HHZK*xP~c3hwfTVDm2;~^A<whoB+Y#@7=l2=~e);ncV
zXWPW)#=2^-Kkuo3e*vZe_5s~dR9q4PPtsU$91mR!K(iuh^9UU?2;<{ID2gWC^G38t
zqVq!*cd5E~2M729@OCzqR&2Jl4e+UGU*Y3UIhu_1G4M3Kny3b-DUkhBR<=C8Z*Wv%
zdR|F&L29I*6Wc(Y`shM*7Y4BGBI46>OR6g?%1R3JbF(wSfBw9u{skBljG(n3_e#(u
z15Xbg9#I8XP*PD=QXsgMico3w4315U3FO=1N-KRq5H|$6BIF4+mTzc~pBLZR!Io`d
zVQy}2W@ct8@XiD!f2skN1m2Hi;mMBxo^E&s<RJxwNtsNj5+GCq!sAl1i!$Ouyd13z
zHDUD;mzSute}XEI#BkL(AT%;AIWayiCMq&K^lx6h2^b2)h=w#ln+wHPB%(WMSvfhV
zyub~h!Z3p#n-oE%bwu<JQ6O>}V0&_S9{%CM{$4<KvobR^HZnBOV=@F?kMvK8Q)~dB
zcv#rG`9=8iKLVbTLGa;&#sNb^QCU@86MCnN(fRN0e-|2)l$jXe@6NR{(pFL+uC8%o
zL?qF<06Y<n9T#0Q?w;N_%y4ya`n!e73Jk^Q>NrLp0X$iaqe%@IQ559KsOwugdWR)s
zm(jc^L01dee~z{%w5)jr5-ECn8<vR?G&^WuRxa5?|FPD=f78B0(|(}VV{K-jt)i-?
zuCAe}f2FO1A`AyNpWw(uC>GjV8Z&4YbF{F8tP+YwO)Y6-p_9kV)Wp~b?9Y4ZUx3+z
zvCZ@$6EmVAO@^w1L5-FOk8p7oZGA0c8`prC^y1pWlrT?QV|8exC2F~2$+KHlUX5XF
z#dY-wjt&d-ad)z}HaF7O)xw>uf*xl2r#|f4e<E^u9{~>?Fg&l2xbQ$P7fjVwS65S2
zRY5lpktsr#cn~@ba47&6l*EL`juN7pV@;u78``VT4=5`sDH8T4J@PNZa9)ijy?+S&
zhtdQfasj@W4q5+h8J`v73-#|0dMyZP3{7aKKnHLvLIZAPGXtFJmlPfM->J{w%pKP~
ze?XAkqmAHu`9-G0hWUFyO;8}YprEd6WNGW<;ZK+Uh%R~1XTWXxMA}DHL0Vks1CGgG
zTL=g9>gm1(DuE*{LmA<vqg_g8<e}UZk&sbPk(&_WVP~Qx2bKFVxYJPxbD(wfjLfZU
z9bDXf0-%G+4LYdI454f%BQ7Lh>(T@%f9!<AufT0m925k8tcyg_e&TadVnTi0xNH*!
z9cDn04%2*Hcp!mfW=wR@si3PKG&vahMy4#<i=m}>{_=(u0_di58Vu~nt7~fGdl{au
z1X%=iboKNN42_CQO3NbHR4AS!f@Yy{bfwgR=j;Ma5#ByT{>GkT4Fy~9=Eu^Vf3S4#
zoSF*Sx^gnobh_nN8sw-_UU61ZRFJ0=$6OCx9jfSqNutvnw|Ce$*xTB$Em`PTH!?Cb
zG%zsG*EcXSHZ`}fw6eA^(brK|mXj2rrFj1GW(r1=BXA`fd^9psYr)2f7Ep^<)rD}%
zJ2)B{ZaItz@o}}s?J;=I0{qg-f7*~u^9&5|_x17ebmzOeI6FJ>9336F)&wU@9D1t>
zWg=4KLXW(D(GEocnjUa@1l3sDNRCEMWno4FM0=J7OdZ_oAwjhKphJdjWo5}?p}5h^
zf@NvNhMWOyY$zN;yDlxo^OrY0FuHmDqw)sPcx0ffsiJFQ<K!J0mtL5gf0dRL7Zn!d
z=izM6#@Z*CObv-X1x-WTC!dlK9UkNdhrB)4!I9_W>;k1c1g3%%xbTDzPiJ9mtSn3n
z(GFHa2aKMfv8jb6H0&wLj8xF0Sy|<IsWCyGJZocpxXK|g6sPVW6H_xYa|;&B(%P0F
z0r>fN!{T#xa<pTc<3?>+e_D#?FK;+=M`z0^<b`EabdA}L?g0_OzV1%gRW|0h`bVb~
z)wDK@hh69IPO6-sMFY7V0AEs7kLBPMmY7pfn3)(A2>o(aHnv=z3j)CvITCtW$}$2y
z1iG$IWOz&rphcFchYn~#y^c1Thgn;iofz)xVvF@vl%T5*T5QmTf1#zz)JLnv+>&i$
z51gNnh}fjK=m_YYa(5+=M|+;uO!&(ij%#S<D^|OJ=nY5fo*{9W$+$RfZKe+#=B)DO
zDRX-#R^=y#dUDN8K#72%qos)gD~!F2riq<LSaLy4dwp4ULZ}zd%D}|J+K%Vy;R`{S
zrID_hyda#zc{)hfe+0wXfRG_GJla~g$S3(hpA^Bo&adm9P*so?9q4XP;0RI+n2ZM$
zKMifBfr+`Ljh(ZHUr01$aSO835(zF32zSCy{EBqqU(|@|97Ua`V`~{rV_Wx-#N5iF
zti(uMQRDhVWsYy1zGVI4z6qs3ZqwG%($?0|(bd%j>{X#cf5WhJ2}&xcn>1%~Q)zmn
zH`h!H3k~so0>cA5oH%Csnu-EZQA$T6Eg?G8-_zBB!vgst=+cy@MQmu3(Xn(5O)Z@;
zt)rnVJ3iFQ5$+!iaMwnY?x?hkq8fzkR<=O0BUm{Nl_j}p36a4*?$G9EZltG8OY!{W
zjjYFyP@}<Ze-tv>7QBGC-0JqmihQh%WaSo`R@U0TeCx5z3%jdQ1G)Oh@mE&W(9+d2
zAQebdlnm^=V{&UJ4GhoguFH$@u`^J@`Uvj+VR2!8e6FRDjxtCcf@T30*0OT;i%KiW
zO;3mlhVCn(*sdUrRh2&MlakS8yGLeLcFgH%sVYc`fAn<*C0$KrSqWi5%ZHSLnvMZf
zWPHNnGmEQRx@Y#Zp?fJ4tz3k>14dpT#q+oQs-XwXkI1ZSpVc$5x-bo-uYBSQo2L(M
zKY8cqn*N4#e`~eT<HV#1RtzORqhjh5oI1W^&g#9ZXSbFn_;7Tk6rrU7H&g_BJKLBt
z)hI18e~{8Mw(|^6EpDtV&PgTmGFaqvq<8{tvP>JV*u45Fi)M5-lxD;Ry4kRdK?6bJ
zLxlvW;5qOEV>62D+WY3MSlHj)QdN`@9}?hAA^&+&JpUgVKI$>$f>g92b{eg0*@Ob9
zxDan=j%Q?6W#^ogJ5Oz0*i{zqVWB*V%!}!pe;66)F>of$5O#dcq`9kr1{LXH&5$KJ
zX#%5?qreq6(Nz)bcu6%w8-8$NUUg}1T0Ask37FS(!xT2??oeQ|-NI5!ntI#nO0qz+
zhtkd#BGZMkn2KqDl@C0FW7CVOTPDxy?QE_=?;aNx2$Wsi(@p8`{^bpO^=hgr%8Iz^
ze@azIDO9RT&~6L`?xSmPQbB$9+?CswO|8!kb2e5SB?!dznL1h;s>+H?E0@5;{F;vb
z&f0>+AXhdMq;PDUJOd*WBLdtVpmIju@%X5Rrs$ZA(!BHp=&ZM6nds}NVSyyN2MYJ{
z0^3&^)H9;fOB%)(WW<O0xp7G~B?LSie>|#Cn>UB}9aL1D;E2gBt7~nlttigTNQ#Bn
z(iMO={5W3ts~RrL!j?tQsR`N@#Fcc-xgH_$S*h_*{BY#3cs`M7B@L5iEt@`}IKkIW
zUw#ypOHtQmqMs%ceA=dV?jZ>|70s1-$>H8SOMOglcJd4ej|=m4vu7D-DoD^?f4#Ue
z)56g^EGa)DAu<R%{dzdNhUKt@M8qVeW##4N<)9ucD+>XDdth{GL1t1k(V~kf`~bWN
zKH+T~-CaSx+=k6Ex8Wi8E-o#vtT;a>Gc74TCNcy8??@yhNIK|W-f(q}V{K_+id<?z
zu^b9zmRx>7WLTgNaZ1|p1EE8$e~s*0(3Qo&BTt%Ps843zWYzUqTn_|Kemd4cLO?T3
zEbTBmie%2piP6^yv`^qSO^kwis+Ea=Gf(SZsNw>gva*tvK87<aE)t9D+gbt7A9{9#
zMu{nCVd=V9T%H4h4mfz80bw!8Nr?&Zaj`McQK5t)zX|4l{|!8tH<*W;f0KiqjitFj
zA{a|anR9p^ZkRELA}zL~2R=aSi>Wqrk{_*$77QHp3UboQ1phG*?Jy4)duwzJ8=6}q
zEgn_p001BWNkl<Z5d1>minlRgsLF`Y5d;Fv*xK1AEHNsW(3%wl*|b)*j*d2p8Z|XR
z8k+^pSoovbT3eb?;`BnJe<alm*e(I+%E(E^Wj#Yo$+F|QdHDwh1&4-3L`KDg1%Tig
z%j9pbo;ED_hNXcZe@0hpXn8TVCCdWSXK+VKSX^pOVwjhs1yfl<K&QtW9ttM%kb`A$
z99+>l;#gUrxZKo|<KW`q7YtWC8%1+cLi7{?st6k$>;vUCXrGXwf7>U>SqO#(2Kst>
zOeUxlo0_w%tkF?!#WFWDF~X!@IRG9<p7FWWopr@&QNBD&CTQscA)LeIxw?D%2Zcq(
z;!zA76bO8*o%C0=bg<;Os8Bf49chVRT3j4+@|b#hdd6tSp}jjo#s~}E0rQQN4wjOT
z#@kfS5LbS%_h7-ce{+Py9bEC223m@e0@?-5qKC{N^wSB{$|-emf-YksD7WbeOa~YE
z#e_!7>v6oG>vlm;Q!!2~;ZXz{TDnXW)Z<p&2+~inFd_x*|4!h1^;Ibtq&6{GMj#P9
zW}L7XFf!y7l|jBr8#<v;sN~=*;73!6@o2-V;Zs^n94h&8fAWe-s*pN2FgCSdb3ib`
zCjhQ^3q1{aaX~$vo)pA)v@yd|1nBw=QQtMPuz~3lR=}4sF~t0CB{>BpAlqr_V752p
zbw?{1JB4J{OkT8kNpF2_1fQirbA@mN0M8>J9Flh!c!3l1m$xRc+KQqa1byg762YTJ
zjTI(qAUX^^e|<w^6Eh2X*wYxYFs6&stPl=RLP?{9BxIE}&{az{%VM27vk$oy1Y;WH
zX>D-zeT4QQEnw=vwYRsk<=7B&Oynelkw+;`j6s3l2_C^{6$_8Jvd;ONPHkD(Sr+GE
zp+bH1QKQ5awT*3@eZmu?Ltre8#Ryr3zr10E02KCQf21VFMqyVRlqm^W49J6$x(KT3
zu<IF2eFN;n1D{cb=$mI6kl`Gb#3Kn?v=lBhQMkDA-LX~<x<w=ix(<*r#N^OX4>7kF
zS`*9<)&w>30%OjGEl&k+#6na+3sD28SHP!baC{R=TW1gNAD%g(Akx)TN$@`4KiUxf
zqYu%ae~mx56!Ij%KIuUbia-RXR9u2KO3GP?&=A3Y8llkig_FCmUVVC6N-~OkVq=1#
z(ygN`iHXnR1g%<LK~Ygj8Cb2rY1L&i7z|zHZtH04qGe!&jsPIN5J_7UjuBNgdPEdu
zJmB47{c18lIChMLk~WqmiVN}Lu|L$#CP&+ue__m#D`$+8>~J~LfgbYoH8j-K)z#EY
zd4Y+=&3#KYFPc)58NxFf!J)*eEf{zdFxpuf>8gVAvK+JrfPKPaB8BwF;Wh)b5on{{
zPmuZ~7r>Yi_<xBj08gQcj){$nPe@2iqU5Iq3MeR$#=KMzv{6w5LbDEoD15Vs?ih}(
ze=P_3{ubt@CW1^j)%b#2WgWP7wl<b#WQBz4)EFnNrpI#d42jE%3-$oD{1Fm(v~vec
zXhY=KGl;Spa>s=rJqykoUWMrTF|gr>CKuIDnzN#RVrh~eC}h$`$@L^g1_E76UlVH3
z3bg(9r$nfKjYX>t$YL7WSPv9CzGQhye}N*S&@?oXhByNJFKY?-fcXc7M#RJ?r9v+X
z&&p6kj+pfYJ#)GgomCE$Dq-hqVoH{+19ZI;#)G>UPGxf=GJj6Wp1OE@@?FS00DT`2
ziD3~TIW1!jKQK14G&3f^m2EI`><yg^HUiM0?5l<+qlt?kmlLl@bjfHl^bJkSe^_kj
zh%BgX?VTskMx%@dJ|Y07rZCm;`Grp}u;2caMv+$^4r3rzw4lrk586WEEZjS|(}+I%
z1snet=>LA`eTHF{t+R)3aAaIcW`0VPzbnT?M^Ow*9q8y8nHlRd@WfAOrjU_^NqyjU
zaiC)@I4Hp1&lj3gK;p#Bb%ZuKe>q3gGtie1yftJ(E=EgJLtRZpiPV!<WUw54qtZ(n
zOVT2|?M+4wx}q`!dQ4pF$81blVnmiG8EVk$55<GAsreQ4Z4k|cU=1AF2;jZL5+eMe
zQlO(Og=@*sf&=!O9}?X#B83hj9N0Qg2E;K_BYg(7@daInc-sqZ{g3hgf4|r-fA{~=
zLct-owFBQTEG{*-6b~tZ><rehF*LQbu{0%X<YMB`O@r`P)-h(|f?HC2Y;<IJNKin4
zzrVj9GG8~Y6*0$zYJUt1^di}I=$6pXM8`8GK4alaWqoUONEg;kqDI>ZJ`_E%0K0Iw
zX^(oIK)JvY$vE8Ox_J17e?%pw=R%)!MOL)00|*?_Mgkv-`I%5D5F;EzhQ2Y_Z+=L$
z%APFpIiW?@!;>^GXbHuMM5t4#{TlRyUf{on{r}=2fk!SHS~+?I$EFuzj)Rkxz8daD
z#oEMn)GRp=8URgbtZ#@*-i5hY=@1LxDWwr$bk7@>5_j|n!X$Vrf8@@~!Rd?#g)y{=
zY>&8_vAs`Bc4b?CTSZ0$pRFq^_)s+Snz%isCB^BXu!P4cuLun^Y-pW|ic3t%$jL7*
ztISOZ=3`Mj+62JE1O*g12vnKCe)B`3tA$vapX=rWEtjF#(3$4Sv$tlM8j+c6MS7|#
z>Fkdcs6~Dv^#7~(e*+dhQ8%=935dxoZ79o%3v#nD(g67h6h}L_db!{*Oip|(I5%T~
zjwh#M?%*4hQQA~jRbHH*ot6|A4~}7CQc`k+AHFEWsW>V;5v>I-UJK|9fn1M)37$U)
z-oK`~Q&4ho)6~VijRkSuwt6G(W)MxsGY!dpdl4asEJ2T&f9TDFn_ANV9VtE`v1#cU
z8JStxIk|aBVcw1`=193*8u&0D7mlf}veX!uodL~swqU>cAwjzV2t7)XvpB-gA50pi
z^g-1KSGgdSh1rMFkcOlNDvupY?fWsGg8cvO82Eh*zo)dOnS*ak*7%9l`AK0O_NLk>
z3O6vbMjI-~f6JMJg6xs4>N46amtcfg&!o2Iy2{dmtW@;ufAQ$ySg_*<MyJIDf@rm}
z#3;}*QPb8lVIl7x=g*{dtoRXGRb6w}%<ZU14RkOZ;bea-Qa)-7vFC`MOz<?}nDPR~
zML}Lpc4m5NGLUaU8GS@_Km!lcmheA)5NH1OheXK+e}H0#t6z9xRyJWdIcbS8VS(O!
zCz>QqrWd5?F&UIbC<*>=M*i1Nf#1!rd#lglx(CIi=8w-$iSTt|8){&=2>?FW`y=3G
z8P*=rIW?08X7zQqHCC4uWas1;7ME32R@Z@K9nVrvg{ruw7iGpn&$@wz9MQGPad08Z
z;>h`Df3V%df!wunPFGECjE}t$<-v(i2fY^VbHp8T($e&3DqNqJr4s$IX{pIc(DfcI
zC?e3ln7pPD8{r-gO+QYME1>4lTH!W}-~5oMg$VD*LqS4zX?c0Y`0<rhRn_H1xfxLE
z2=MX5JzUn77G{`!3~s3g<{t}UNSa?j^`}V_e}9<Ziw}*YN0BlQRFLrdECT!locP`W
zVbO_c*-7ZIwl!m@VDUiE2@C#s$II#2`NCw0p(O(|rnEQJl;szdjjw5FZtdu9EX|1V
zbFySA1MWe{4X7<8-qM_j#Vg#rXia-{1*Wxoct-ie>C0wKYA8vM^5t0&EPX{;N<Gb-
ze@*0ufWoZ-4-uMU7185^pc$HW?3}%Vq9P;0!$Lztf`jPkC-i}VP6Zy|UYeZ{=FMXp
zYT_WAX<%%Id!|1Dcw#ELq7M4Ig5xqv>l>SzCroT@YwwuYR9jhEn46K32qb=(f<dZ1
zL2?7?h{jN1)F<joOeU55f@(9=n~`aee+K`J`B2o0=u1dYQRBEp>Cpkd3-Y9A771CL
znAi{>S34LgBY_*8?OnZtyqq|{$fwF1ItFJ{PhPlb&GH2^dpeq`N-Am^TRJEA%^c`y
zF3*bg<FOd9-vRfwhKl^OSfK22ozdb7jfw{uYgaqyywbOE4@)a;?p@eFxuqJ*e_0S*
z32ilU?ynt>504N!;sAe&B}AU2crcDF&m9F5zCJ$Q-rg`y61^=c#)K0s$iehNxR=zI
zk~DuP@8~l142(X_t0yO}DJU~7c|K8TMYSDW-BWs|PMh98YkF@t0>8exqO=I;J_!Gq
zsEDwTK!0CvPds_e!QKwJ6C4i5f5yhf8rPgGSr#A=g7ra=k>jw<a6J=!DiafUjjBr;
z8t5|!%DEJ^@yGlw2Y)eGem1G#54+JoTSXR+V70V!`8D7bO<W@KoBLPo+rEBiAuyOL
zYMa|9PoJ}J`SASS3FR5lzK-UYSyNKi-ZyDNO-W92bdZ-{a6}9`Qu0f(e-p#uRAH(b
za@<2xiW|D;qI09Ux-36Kz{tQFWiFonVZnj^MBgQxq2_RhY7^;pNg}Ch&bB9$NF$rQ
zi8KH@Gb|lELlEwD?e#cNb+9D19TRrmr^NANh(xiHs)3DLP(of+YhVAYIRo<-Em=0W
zbm6=?{nL88JKI{C8fvS?f0q>(=I3N(pcMeYZWs!7eZ6543JU!o-NAQrb#-xea^gWl
zp#x5$KsJE*!rtD_&eqP(9=+!1$a91xgU$^raz45oF#$km<d~SxFf8iYT342v92Mw+
zWyiHt<)uW(G0a}S0=$yBdt6!P!Y!u{@7lNmZP$r4&7Hk77Ywf1f3kb+!oHUBj3{s9
zF~nw+)^|*sKdZN+xvC^5B{nfNE3desx?w_PKBmPRY8czPhb9-)bo5W_>F#Wu&`?`h
zm=PZeJ_Xjh1i7O0)TD&i=m><r4;eJUC`nwFqsF6|LEhNfhBH!DX&mmLMUEuS2$$Bk
zPwOC&NB1u-DQSLMf98%3FHP*dA~VXGd*?5P`59|BY~Hqc!`k5$gG(39pF6vM`qU}i
zo$V7RG}YHuk1t0@0E&0vFNleWiH?qnibR(Zx|PtS6ciZX?++)Hw~r6ZMe@VP7Fr-^
zb$EKA7k&Nw10XjB@{rK<Pf(*EF9?i<Um4T!>9!IKozXDff5Fil^BsZEB&n#Wk5%M8
zE}bc}ypyY^uGoM5+>u=y2Io!dY;5nHwP@wWU58HZUAJ&*b7@+*2g1Ds;l5zy(s}(o
z?M>CCc}3+_b<M3^Q~KJf3z9=UQRHA^=N^)nS2=O&)TzD5dU|>$Ru`l~>JU!_4U0?5
zBOYExYH~t!e`tWW8&rDmAUt_#aUn5D2oSY&@$`68pxDwRU7|K<#^&PBfzF-zeeHE6
z8L>g`kZU1o2A`CxAA?SMs0?YD^ZeuUYrAF-tzNfr>yF*~4jnywaNnMt+qZ7oux|CL
zq2)^!%^#ROb9!IT<gSjk361qN<I744^Kx@?a`1GNe~k3B)D+~45>&vL=%`32ln~v=
z=!*#OR>Hy~B5~~mvSXp>$n%BfRM2>%uKIDmHlXE`?FiV$MLqR7QSMgMn0Mi^5(*l6
zghs(fb#zs>e@4^n4W}O6I)7y6+NE=PCQhEYXw~Mu$1mPKvv=)+DUC%*!SK90rY~5z
zW!=gpfAji#J6r0TT9FsfKWF}u88BKZz?r3MX73)HkX_!?-2*dd<}F;Ze0FzZSw^&<
z6N}6Dk4VZcshK#Tv96}Fyrd{UGbtv_A3Xt5*-}arYo+0qIZP9QdmFkaWr-3ImgI22
zaYb$0)PbRSeeJbH=`n%iC<Rpo>@WZ3hXl1?e@p_fbPY)<YnirW<F;M<4jw&u_TrVx
z7tfzPed@%qBZv0y-MwSmmW}Jz46j(ebP?83#79|UZEa0;byZbm<@k#7veJ^`qQZiF
zv>38sAwUu$Gb<}An_d+uC_0AVI1ggfq=dNWNErD9BOz6YmQRs!V}1|dAAp_QG_S27
zf6~QBmMEW~cNUYwQ(1ovcr}~g-1bGguDpG6^X#Fm!waXQ6Me;|eW$KIcy;H@?$ra`
zwRv%{&j9!RySA=hwQRwh=~Dpo#e=KXZd@`QRHof+^eoVZjLSp|WXiPJ3zn@~zZKmS
zQ(MZ?BfV^0{Uej}DjGY{*69X8!iL)NfBXzgzPGi+i9{(8N?)Bw-Q%pFKpK@2R*z0A
ztZtn;Ft~AOU}|f1VQLg+`(Ta{*l!13gdBHh?Gc%U0@>Aj4jwyo?(+59_a8rc_~8EC
zJGXD%xON4W09pZX1Z>;9aoy^b%NH-0I}_dilO}a_c6M~Mx3x}enb6eOP+v#Pf8RjM
zptP(EMTRIeq}N1YQAufeMP(IQ@Z@<H8|LTk2;?*(<0&fiD`VC_-To8o!I4!xrI8L=
z0%8hTW%2kGRT6c(@Z!lU4&VRz+sD^W@7=h3&g9w4Htacd?d!KczrKBH=ZaY!m6^qL
z?b87F)5j0(*}iG*%4G|e0O(tHf9^fFVNq{mew?qpA<F@s#OZ}qE!{Kb53bvO;N;~)
zo0d##E=dk{@(EAQtDG=-)>2A<v};0jNp?~scp)DR6&@o5o)?~BAS*>FWMYY|$drPr
zmY%uGH|*IkIJc*zGC!G8&>K<v|82mNMI&3ExPs<sOSc?5d-2-s2TxvpfAjwRyYIez
z`}UhRuU|fY`uO4fySHy%yL{o?=@Ul}?FWtY)u4|*Fn7+J*|TQN?4L2cZ))$9$=#DW
z(JDZTpt-53xp~5bmX?Y53p|pgw!WchLQ89V2l5u{;kd(1>3q;}!$c=ZG2t;ZZ}Bti
zUto_;tesJnz&DhjNBxM%f2b0h2BuIL>L5R#%0G_Lv=1xjT(aldx6kfeJhpqo-~c*G
zH|<5hzxn5@J7@Q;TZA0EhK_0TR&3mJ{NmY@M-J@XzG>a6HS0IQrFii0)}h&5P?mA_
z3_#vnepzkHq-k@P3~$<X=+xc~%Vu`g6{UtJWENF5cTF2uzI4f=e}(hs&7Cu=x4m(E
z0kHMmxK=<86BQhFc!mL7z#w*oXBZG4-Vr7Z=8bQfG-JWaMYDR^>dW)eVnT@8fhx`c
zeFE@=TE4EeM^tw0<oRp&A3J^V+MS2bUcY<y-FM%8i=lt@;@Oi&Uq85qphuhk%qc99
zzIEf;;i197<;#~Xe_gs{@uGzb;IEmD*1+^>XbtpCnbOnK+dFk?AN~TZh^~&VNt37a
zPVJjM1J-*xx+&_==Z?jroN!x!ytEj>P5ewf7YzL4QyXVDr2APa(~a^J6O!;O40P1+
zL@^2If6}%O$(=ZF^O+}iube)(W8L6<^k6sRlm3Sncg`Q$f3jkrXTqfE^M^L<I(+u}
z<@2YHABMHRb^FfUd-oqae01lU#WOm92_2W5ky}(=-7v9h>Z}FJS8v*RVB4Byb9>wC
z%W{gw*H4@TuOeH9ZD4wLOHFA`N;KN^Y$I(&3Bff-aXtn5oV5OFKty6zNlnY789m6$
z7+;*5o(O|<f1GSBjd4u(+kwZ<or<~^%QZBuqHX5T_I*cAUATVt(aUcMdy9~N{ptnU
z^p9ZIqjUe#g|nxQA33;p*Y+(NH>_X3Zr$27YgP}hTDf9q5G?|74GhekJ1{VB-u(Fs
z@E3SR%$z-E?!deS3z3&Fhj0_oX_te056RhXQe;ocf9D~Ofq!y#%lwJ?VGg=97jeV@
zSqF|K^qf)ZVq&`X!C8&{s}J40eCEjBt!sxC&R#scW&g=5_nyClGho;H<#T#wE?lu;
z=b=-V@7%n0>HO&v=-%76|G>dRhmRaRwrA7IMg83qsvyf-Ra4(MVPaR$jJb=J4{z9p
z+=2PCf2K}qZJ5wLsb~7^d5e}0u2{Khc+J{%8&)iu4V%6&BQDs}fu*M=Eud-Bom@2k
z2YL0XItFIeXx)ayq~sJ=HGz9wT#%EAbrO8sow+t(zxg52;hY&SAhD>icj3BidykyH
zbo0TpH{X1NKz|F^BjjJcc=q%OhW_StwC&HGe?EEa@PU21ckSA_bH|SD+qZ4qx@Gew
zv<?KTV8sd)RIOSyjJ_b(#IogsL-1Eakbr!|dHA#g>7NL6`f@DD)AyHc<o_n{PtWgI
z(p47cZbqvh5t9Z`Jm>Dp<8s&p*MRUYnD#;GwY|%BoI81V-;Padh87PjTeEHd$xC;h
ze|+=w&gD}Fwy#}2Zz(>yFWq{0{|<TQAv=!kz>bZ>O9y65Zfm8i6M24f7cO0~dL^YQ
zK4a$Wx$_q;Ub=k6FuEePY~Q(O%bI2Lrgu-QgYqHY7BtrA#tfRLMbE^h=g`|ax%-Ah
zCuZd3<>uz%kSaYn0frp;e%fqVNZ??he{CaspXi+G&RIk2x9vN6_UfI-uU@}-^9=xx
zZhCalBj}&w10Okjw{Bd$eDU0wQzwodJ#ysm;lqax9Xxno|Gs^D_w3%iYbRa@TQ+am
zvUTg$ZQIZn@YN=?DmHG~v>Cm)ZTr^E8`iBEUOBXU-i*nu_`DCL@~XgH`&<bpe+>Nd
zOM8ZTYtjSl=tgC-h04jp9m|#3+7R+r;5h4Z15zqF2i6^g6MWt9(9(rN>yc}A@z&R`
z9^bus?%4hv8&_@EzW2zP%V<ZS^?&cqt?O4WU%GJq+}SgyPn|r1JpB!8RxAaCr%#<S
zsiU<60=n4)3l=Xy%N?)x6)RQ@e+{9lk#Y!+o;tK^<Eq7T`nsCSvts<6EE$R;N%+w8
zV`vV<FHKDRqI<h=RmjsXI3hYWAt^N@D;IRDp^z7cWxIa!L!yNn^{n|g&Aed6`tAEq
zT)6S@#Y+VJH*dfF4&C(#dW8LJz#bj>U*nCB5B&@0&!0bc?%dh4XU?2Pe^1~f+5^Xq
z!D={!3@wQxM~)I-9XxPg|Na994jzP;4;|RIXBP?(H=%tpeNszpX>M|)zniT&Q&mcm
zjJ-Y^_!n395BE10MDytpQ55N&PjIT7op|;(EK}$*Rx;%JCzec@zI+>zwLn4Zs*O7j
zpr>&E+1GckUpfPh{GLN6f6iUK{qWhFSK#at&*1viOBc`vI8Nm})(;OZ8t9+e-3E#8
z-oEMmvj!F|Td{h>mL2F~+`4J~+ToRh!yC5lK7bsl>*tQ`+qMoe#7&SN;uxq1MxK&X
zZZ`NL+E}KM9)SSdF-sdeN8AAvmy(%RQj8)d2&1w6cHpV1j}FT@e;^^Rreh`~tPY>K
zdhgkDgggNM{vG&v_{c}#zoO8`vEHp)w{9Z4apU^+>({Pby?W)!<;#~YUAzbj;=+Xs
z7cXAKU!gD2;y8O2t&wxERL-0_dHm>+LkD(jSh;xi)XwHA&}DbFHq@Xqc%KdY%WH?%
z4|G(d1kxj-sEI>ve|#9Z?&0a_Md%Zu)6|&j8<$_#JutkQyqSj9Y~FS7#QAG?A3eBx
zi@4KIp1XYG-lOMlAl4!7zQ+&m-M#_W9<npoz(IgAPg{FOXIIyx?#WYT%v(0R8Lo)4
zCl2r1xdp9^HJj0rI(PNXquZBHAKtxr^|FEP`ho<1M^kk{f7dNJ)W{Z?`-GqdffI`F
zL8DQJsc&RTsc}LnwYs{dwzdxJH$Nm|RFAl%w5BN+#$7k}%w96QW#5U5w;u!YZ{B|U
z?)&fG6Gq_wb_D)Uo;-Ps>=ClBzyA8+!-o$ZJh*@V-o3kb@7}qC{@*=h)R(sfcI(FV
zt5+^vJdfA>e;|5L?KK4{;a*%bonM)y{%->R>gKU6i+k(yqCXfB1ykKz+}u6Aef<K0
zLi{{kxVWLk)WIh@t72l`q9v3gvVQCCgD1{ixkcP;@Nuu*y7%?7S8w0{^wW<&{(zpz
zyLWG1Jb8Hc=GBX5&tJT9{pOwf4<Fw;e{|RS!GXT^fA%)G?<Y>2*fVF@x?M*v+<x}p
z=EYNoc5hibxOUsVW9P5mfA;P32RAO9fVDrneSAi!D@#+Lo;9)~M<459(rSHVOP7L@
zu9>YHr16k5*^FZ3&Pks(caBzAS2eV83rWncZtELZzHaB?b2q+z^6Uiy{=0YI|A6gB
zWbnV=e}4}vfm#IM^kaKN*lV;VUcGwx^5qNke=pI0zM{T<@!~nyGi1bdaqsS(D`$@E
z*}QuB{2A@lc?p3oR!q!R1p$lC2LAQ!Q#*!cww9;SBciBn<b?a5Lc*ispy3oGCw0vo
zyu(wA>N?T2jAwlhuidm`@1YZCFI~TbSMUA%fA=3edivt^x9@-akN*P0IijE5y+Jp@
zjmzgQ-MIVpv)AAK`19MZub$YocJT~IWi&T6)YVpZ^e@|Z@cg~k|M~vKgKK9F@7yrB
ze&?Yx*Y3Y~_g_D}d2;XSS+w?-^)(g5cv<U6(CS%q0OD$IX@sesA7DmMj5JK_y&}?z
ze;Zrcx+YJZK5Om*uulM<WOgVqSUkVzjFN_~84Fi!*^gq_M^B!;eErRL?-Arb{q&E2
z{Nta<{^KM2=PwL><3E0)489P&0t<&&HI%>j^{q?D$JsQzWO{2wW~8UBp%RQ`gVgQk
z01wOU)$KET*3R!~EKZK_ceA(9Q<oK?e|2SD`JUc>0l}f+L3l)fF;f$(mP0On(&V10
z)1f}Oc*WYyJNF+weev?;D_5^wzkcK9-G@(Jy!q~jf0DTW{PWL0ynXrj-i=FVkDs}C
z{SHF@-A}LXT{^mb^}^|GZEd87-?w0R`_W7H-~9aD^ZQp%AKJNL2oidiZa;qWf8#e#
z?_E246kQGrrZf~L1UQ<iK|em^97c)YsaPDI2dRK7BH%{Si-T4~t*&cqp4i^m-2?X9
zpAv<<ytWA@nwHeH_0CzgcE_PJ=PzBoar-V>>Cc|OcuD$s@7|;R|09L#k0XsZ7rrMB
z1Tgd;f`fv7^&R0cz=iSV^^?2Tf6gBVmwE1_+Pqj_uCWTysVMYWZasJ%H&5?fzhrt_
zRc<_pl3BsFfdI+a!rI=+#g*^j<?ZW__B#(J%Ax<oBP2evq^_Y6vrsxZduPsH1__ws
zCr+I{bN1Z%3zu*{^X>Qlz_|bXGg|WR-#mSA^YYna2acY;aP9WPr?0<ze|q!m!7VH2
zPbDgQVE_Oi07*naRGrw?-r3a+>8g1{8}~e&16SpLw1=~8Ta#_qWY<jAG}(5)jLDcL
z8<TBYlWp5}-SfX|-Fx0ar*+nM?+>1x)&8<IFR~%r;k`fCLwp*xkemOp*5IuS{^%@C
z{x^O8n`R2piJi4fLu)L~Z*uC>V=uvV^RKAlIGSd)S92>n3GBl`i(LM`1Xup;j0?U{
zMTXyQ3-E7Y7cZCUKCgh9L~~}nPPg5KOm9?@JV<}H#|=c$f^&gh@B(Fk2)<Vh-;oK8
zkNoC!>M<w%a^{%QyQ#u(ZdD@*uRG`~OC&axFzl0wQQHD@>hd?k1awHn!UeMzojACl
zfbIt}iPOO__*}aj5r{;_oZWE&PT{)01yFKG*x0*-D@AvchMj=fSe7s7r4@EMdm2&4
z?JksVBiM~G`WzDJ^nN~?n0FpRJzI0%n|m&6A{F$#-x?k1;0@bYh(L64o^oFEd^)>_
zWdS_ij~)sWVbQLvo6dw?PAGoi!mY(SFRtpSC*ho5TJ`mQ<F=fx`JvU@0|7&pF#hhr
zNNrI+g()X{V-6rC_0~E&Ip5y&2t#{7cv^}NAQp(pS<@<Iu>sGeTy#jw(bFRC*P{*T
zO_xGXFKmw2UBnSA&)E)LyQ@U2h*-$tyx|M*H7(nv0CEZV32nE5V5{@_+en{xBl1I2
za&d68mZ*7P6qLe`C#0<={|cQYk&+q;o-Y(}csDty^uVPVqqQ#c<j6<_Su?ML44bU0
z5JwE+0atN+vf=d90%%b#jn-{EC&Q?tWzlQ9J=$*qdz%D%RTk;=4PC2Cgmotewys8Q
zhtYkU2sK^JmF?_KPf}rW*G7>AP%@f5Znq=dSx7e8J$`&X+)G$<pRWOD{|eI|tFi@J
z-EU6{0q0j04fUi{w2RBowk*}V(BrlgHbq+U!H6+>%U+%8ZXFN0TG&=5q~>sG9O+eB
zFy)5{S$ecz(SBOG@S{i^usF=b1O`o>uazCJu|y)i7d!K(P2H}(FFV+l1+P+XkBAPp
zJ(INOUpgtkZ0~%BYrLg=-$S*StM3hkT7c6VpanDR0crBQd=$>TZjP!HBOCVZNA+^Y
zITRmCEJ|Pv#Gn~Y@&b`J>@SL+_V(ix)=+WF0r5^y_GDDNaC0$(Jg$n_BK21HEBEPW
zFKwuq?H;>5b@qN#4xjLz4^^a_ErFD^$g5<-T~4<j^Xjinl{#}GB_$PY_6t6Ge+H&M
zfyK#9;Wpr1&}nZ0wPK^$c2dj%(WncRC6%MWM11{e2Slo%n^I_k8h2Tv%!U4`#2bq>
za@-^-lvUa@lRw$FwF~nEz6!leD@cRk$);_pD1<49O3BF0Roq;gTiD@!62*s@Om@X%
zD<MEy-R1QD&=D4!ZM>Ml`nCynis>*9kotehFZETC3i&}8QA(41T%&)y@Wtc4z+8Z1
zp$P%O?LV+~FzVq^xn1JK!C&h9hYXaYIZD{W`Ff}JEJhj+mR6jlS-V)Bq3ynDW7`TE
z&xIMm@ZGZ3Acudy#bMGGuD2R*jZSwb<#XPjNaPVQ+5&%kWFg&XcktP4Y#0jz*t*SG
zHU$Bf$HCm^S$@J58VRs2%fnog;z`36c;U}o0nK<?&`vgWa<8k>?{Ib(vg_Sc4zosR
z#o)*TSp5hly^*+|-~aoo1YLy_q`3DFg8xI}Af#mh|C*9zaG;1m#%GibBUmpRDT9%(
z#3=TInu?yBcvK-8Q=nKOoy#7$-YRuXcRz#n8w_FKxmW}4!lTyWdjK&y)SS&WM0{d)
ztuE(B5^&)Lii>q=Lx?*KsjK%#dqZJgk3-P%OWVYhy`YD7Y8!<4S0(0!XzR3xUjBw7
zQH=RtBt_`Ta$=ImkBgv?)Ch11dE(O{|9<qnp6e}mhC7zf1!h0h<F7!&`Roh%bt{r_
zm+R&=B2Sm!y$}S6?GtGy@VI{KtvN|1;tMU-!t3$IIB~*~DbUc-Sz*4A_buIxG@!x>
zk#ATTw2+4&ik?n!2PT~dP|`lw>(|*Z_v(up%SHq@65D=d`%&3mnMGcq!<QE@h_A!g
z57l=d>&ha*n1B0a(h|T5HGp$Z>=lGgMy3jqpx9NfLy#*w(d92z$nG#Y9&NyTQ9oqX
z@A4&G?eLZiM${#e-fyzoX!g55nYRvJZ}yDrCaDSssgCRIzK_B;7;t+%U*A5$E<wNe
zo2`jxhQr75w0WI`BlEC06_}N6bMS1$<kU1gLik$~3zENe96&P1)^&3ux1@VqV<`P}
zu2Pkkb1ktPxcF=MT;58|1xc__+~0+BUYD1ZAl%E(QxW4n%C$}57U$#sPNu&9X3AD4
zVyTa#3&k>=@w6G8)dd0f=NEnHi9SmI`;VvNcS2lc;*B8x;<2aIMP|~YQ;YK9v0^;>
z=P&5OF^odkfa7rPE)sEt1@(XR;VT5ie}w|x77@t*79}^~qVb(RZ)cy;o>lXQw^M$N
z67av!Fk;;`KSO=(FZ_v})T6cAsa;*(1bQK!`n9o^(4L!g`yV<XW%+|&?yWVM4JmHw
z<4QYx_E%iHN%CWo`y0v@(eLm)DJY*l!m>Aj*OzBtH_Y8$U!dO4z}!;~><@<Rt39p2
zyE06oR301p+R)fKGJ8F*SY$HT-cjaM3u=X!2MQeDKQJx$DcB(@zmB73RW$+{H-qhx
zfqP6H0oq4#ub=xxAhmzf_X7nymiz7Xa>MWTh~{R0f%495Lbny;^7(yr0eK>QdR=XH
zfd>Hg7lO>~jjLb}x#MxYceXB=POE3D(XF0OVGorrZHTh76XMlDWgPv@weP?G;PoEj
zmBa_J&T~$2lSSYKlUR#0za$$8!>-UpAY^6Lh^p;xQqUnDoT{p`G1?HJKg3^KI&3z%
zDW}(2SRmiXqY#UDAM+O+RPFluzQ05h^4I|z?I1O99wbp`2*?*!OlAqWZ_n(PEvnLm
ztAe`x6i}h=m%#hea^#<?uIGpKN0E>JzA(cskC*#*0oq2!EVbVxXW>H)pURXSAGC;I
zw~x2vW~>xj&lhnSusSS6&quAa$7dd3NMNzFUO|M4w<M-H#~6c}Nzql6_EwgcK$@WS
z@d~j8qR9J2x~4_VMw=_~OpNX#3UkW5fbGE=bV^u**K_Z(?|Mb_H_X;oTnO#knXh_1
z0?sdUH4?9GwyX7aDNEPG#=`!BZ_w|Fv<u0}-=eB|?+R`*f>o3&p(SLszac|;xwu$<
z;b2TY!A?&4gMr=M#9{sy>?;nBGyqv6`i(@c=`%|eQc#l0KOXWCaT?6O%G+Z8;_NuP
z&0!hz@;6@cg`a~p2B|m*xIAqI_<FlvuC(|PF1h^_aJyKka~FZ?35&=(-w=ZB7HM}m
zKhV$k@OcHz#~)80t2`{ZOO95nGaG6DRLuWzCMUjt;%GJ3BCI?-rHI4708=UW^#lra
zlfO2VnH;!0oeXljd;usI0n=abmp!2e;CzjiDp%aD&c<nn89<GFecgpB5*<O{;VuiL
zJ9C&#q;as?KSY(R6Ej6OxDXg!l6cVQ^4PDyDTK`TF=!$oOK7{Hw(g>yEj3$>tXZy$
zDq7q8EvqPD9)&`;Cd4ELh^-3XE7S$lZijzcoAzof&V^bQgbsSNqR<e%t#(7+i1!wl
z=LQM*<v=oV@{=>Q#a-RKnEmkTNt-w7GrA(RUO)q8w`vRJ<Cvf+BZ<ovjm%&wJ#!&V
z9s}}=Cn!I!dUg-1y56(EKAsb+F0(f}%d41adw-`)MkZe#f+=$X&S;Q9Ype&AM=?W`
zo4L*-!z!ddztlHKDR{z45niPk?NQG4IXFay1m*wO!Bn4NuXAS5^@9Cw!{-2Vc;aSf
zz)|nfBY4Dn1X*hQSWsCx5fihJd@%U3ax!{DKr}V3Tk8bJ>xOuM{4Z6&{-bw5KADM>
zzyNh(rbhixnPjB{a9#g30E$z9D!=JdDjc1hJlMhU#7l$q(*(oyE);Az{<ca$q-fp!
zCRCQfc5WHTt-v%J#0k2dAQUY-H=j^(b$M-LO`$JF>$TZ<8vH`64UO?T2VE&~+&Sxd
zWtIKP9sWCqIr=8A$skB}g#pRdHJ3-nh?@56z}rsxBN!ma&zmjm<zRNWRUg_;Cp>&n
z)`QkH2~OgUT5D^gI(%^^N0-(;k0MMgybI5#t*8iobyq*hD$)tL3gYk-N7YfIRAvdQ
z3?fE`{~BSmP7`#uxDBY&kgdtj&#$Yqo&gCW`sbp5Dog1KqmG)P4#AirFDBgT$>7vb
z5Mos4x`5X$Hmo5-LFcoriweTf>3ddWFBkb+b`Wa}qFt|L!G?P}#!QLB;mC_PN@SPM
z>HHXFaFSR<G+J#yO>k>3AdXi1*8LG4E+SmEn;v=pcLOpeje^DC0u^h{cat<<S7f}T
zu+vjQ0zzpYd2dwykMH&yu5gP2*`0o2{+}1=i-5|KaFg{wR8b(FvS6Y7F-HIo7{{PC
zKP?!Ew@VC4;{qc`3Tzn+yI6BbB23mps(N0KAo21bzucaoz3tFIBFwr_b@8H*x5I3}
z`DvWU;<8q}q1La9&w~i!#FaGbA>2b#s-Z#J!UFUCJ>29i@3z*W$}9_^v9S?Gx+>e_
zG=SfGdyAm`fn~iJI9;I*tZs9AZa>x^{~gl2j2fCG0AAZjMA&G=$u9JS29RlX3M3o#
zBA>ZnJl4exncy3*v}QmEUmn9AZJ*;}5ug+Q3IC|q>vfXpd%x38ostyW6A+WfsXt(k
zh(TDh%}z|z{Vw3>^ck3YFYrJr;JrVQ5B%ADza6^l@_q$D7-CffdFl%3Hej8jjrfNS
z!D0!WYhuuJ>RErRE{{wHSi?4(G-W0=&!Q=W=J_7<-hqHd%<`n~l+&Yu0;!C~8vfoE
zPR72E8M&aB`K67(hzi?7g!=bAW}A-H*v;%t-@A}cjwen^V+iz=c0zh$W<0_QaAVvf
z8d+&^mcecBFb-Pnl?<94Ni&^OnmiBsvyjIQl~P%SdB~YzPA{LlW_?HQsuOB7Oo$a|
zoIoYF-k`rkez+Z8?1>!7^p6rOs9EMQ*v&R);l|dT;BcSe4!yl`#>+`@adBH)IXO9^
z)nwI1E228ojEsVU6km$VJM--UBZv2g_o2y6keZ?k^#-#3rnEnwb~d(`(T1Ie_v)_>
z!B|j>APRX^Fr0=`d0jYlFw_P2WD3denx=O`)Y6Jj{MeR%S2PUZvw_<+7OhA+H^JoE
z1^Y5J>dL68_U*NRCZGoKwSBw89ZZ5hs#?o)%bda@*nF!GbZ9#uT^^t{#SyhOrXVBP
zy9|P!tI&u1!7-LjGl{R)=6X3@r+FzHkbg0f5BqqrCfETR{)(O2Jj+Ow1!AwTAV_9{
z0zyvGiC!)zNSNGzE(r<J32@Pm_95D9C;u>!hN$yNYQe4{eX-Q`+<~`DOuW?k-rw#;
zfh%4hK}3G5=rXi|DFtwks#V3sB-cd(8dk=Nr*q)-hp1VtPLIdBU>~|~V9e)+98DdJ
zeBD`3^3)=T9_4gU5O_4--6lvpZbd66GT=O@;S$JIQY^NNF|@RFw8uD&hNNF^81G#g
z;J^h04M;er_gpIUf<WOO?*nDqFOyL>2WeJ2{wy8MDDNs>NkCQN?<@qINrYcY<6{aD
zv4MQIg3LpsYj{P-1kS#B&ZR4_u#2{<z|KyEbQSPK;sGxz-|Xu2ssdPRTRF&B@X}2W
z4Stml-pUOW)D;|)QA^2+V<_z9Y;s5arASLNb7S)lYVCEhR`pJwCAiR2G#C1`uKPOt
zhdeRkizQUvDqvJ_#qC_MJx6{`k0WoC(+t%Ftat{~9B4K?Flp5|K29SxkpL19pEu95
zW3UV3r$pH}XBI)L5e34C9yS#HQ_{39R{XckTa)Q3n1+LBwk1|^xrsf;=pYgx19w-h
z`*c)HT3`X1yY7vRs4=x|E!=bU+zh-g)Bf?c4>w4Eu8(Tun-oFG)y^a&-TIl!B%mV>
zop(M3wU^bW!LN%GuLcS4<>@Q18Ju@u8N>_e@j3ANnO7n>Sw*5JP+Y-@9?32>L|tfB
zK-Div_PO5o+w9j2O(A$tr4cL0Dxg|;dGa+KlSW_7ZhpAFqB35agjK()wn8jfrGPA@
zNl3r|FtlW`d2A9W=w`G|B=r=qK9LxSg&M6@Ky4@Q&%Esl*a^MQmvIqC7$gGMN)nwe
zBeic*e-kqc*;r8}g<(BZSuMZ15=(G23fV1%>15<A>akKo#rh;CplNc|-bVeSOJ{AU
zC}QR6g+2Eu=RrW&F$vd_){=}jKHeiu|0bLZ2y(38p1U8b&P7cnUoW%yd7KCeIk1vs
z_G#^=NPXS#CydjXW~;V)1#jQ(`oz^g^K`I#EKSYPCqToBdJtT>Q8n&)`qk|mIgefV
z6<R1{&!X)v!DcPuhsh|}>tw%3h2;%qH7$m>RzXBubvy+bYy)@PIm7@rYq>w8tUt;+
zV437n4beABO;?lC==?ClwThbAacu#0rHN9Jy$Rhd!4UUi4iS6l@vn`w!Dw{_UBW*z
z<DBeVtXf+J8%?Etpvc<p5*;I3e?u!itT7&ONHWaofbw_d^bB+!LU>$~*<gZGA!tN@
z*oZ$aRGbQr{R5#-84Gte*!ImKoeu9B5U)G&`g6hS@b=d5s@(OwZ75>0`7l0qxEO1z
zFWgbc8R8Jr{U*Ci&K)27`!|}(MGZ~1E?7<D)e@mx2%0tGJ4tt9fiQyS&K+&epQI!#
z89hpmukg8|)A3IxOJblz;qssO-{v{G>@9zDy*IXAGNmUrC8aySo`T?>MqQu;fw7e9
zmTr@G;!&m=fBf#43A+ncRJ>p76e*<&^%M*xXe;}J-%)%%uY!?MAP*vGk+x~!HSDz-
zc%4zWRCumR^B~vGSiKVoShNfiaJECB6>^KQCzFwH+?@SC?8Vy3@*=E%_5`Pvp=R;|
zAO{3@^W&-N^Xd9sclcvNsLB2v$lEwbm_tL;7^oZVsj6#$F;*39fLO<NIi-DP2#nW^
zRf001feij87Kw1?4sPHt2;G6^d5x%n&(=!ulmwnC;{_hBRvnu^*K67>ga9HMooF{Z
z&1RY4+KW7cJ-OZrFw9DLP+eua4iWkp6){?SlOXs*txwnY6WXbO#qb9MG6o;ys@LeO
zEDLL*h@0R!r&9mk)xRyyu(38kMVI-jfFNmzLauzOSWpCxg5rq?RtSlGg0|7~g;1GG
ztjsXkGcIFA38~&TmWP4klJY0%?w?PKBA~Gb;Sh+m(dA=psi9&b@^lgQ(7?#zHM5Ee
zBk4ur)b9DKCq|iq$oC4QVdyWyy!!eVLmkT;!T(WC8T_6Ep&!m}yp1mJpVUgToO@Kg
zIMAo3`;uI_(R$fmD4@Kd)=hbMLB{J2=V3*4cnCHL6XYfEJ!7a}$9;1vKFx8@yjRY3
z5NF{(URp*0xt0!^O2R}W6emvLPD4OVYRGS{3}T3|$X_1}<jue?9{JWW+fF44H{-C(
z@YJmIZ*2qFZyiD2ATQ_hX!+D0cslIzgMgSBA2W}=)RCTo%<rWO_=zUjHxK3NWS6%l
zeT(7G+sz(U3Yn!A^PRVOO{xp?id0lYXX{{EDtfG67|e2xK4<UzczT%syPfedb_HD?
zd2fsTCnV%&fF=<8IARVl30Vi*i2XFT0)s<jKG_Wi9uiO~IZ9{A8m-+ZD09JgEm}-8
z<P!&8sQ*o`EErs+K6U39YG>Df>z^BP)ENC_3@A1qs`h2QZrlH`pF%U?2wGCQ{?~7N
z9pBHDe3cxp+w<4#k6lBMX8Cb-P&ga*Xal94*!tPGrU}4m{C4$?<uGAbU)O7G+nrel
zeVE`V5CTKs-HhDxmh_jF)-yS7|G6&+AD($*lNf4TJuhVELoT*WEPD^Iu>XLR%@CPh
zl#vDDZpB9(yX9k*P^bWuGG}V)Y7UB4T2D1r>D^bCix+eWugs^jo+7IkT5qKH#0Bo5
z1vE7sU}2-ivw=g?W~t`0e;^&z@clmG_tSixHEe07M7^r~6O3s;RC9Qr=Hglp2Oa+u
za)C@lTWYN*F?W=Ns5l9~u%>iTVF|dZ?1;R)N&+9HWr?bvF+Bd)Z;C>ckortN5hmC_
z!6j6k%|BD3QowK#H7<-JD4M;ggFX1PTzHL%fyL(1<#Z*X(CE9@*zC`b$M;+B#n#x%
z)f&I&x8uZ81dbhXA2GbI`eJJSqy|oLO)!oo1yZrr&<Pt+`p7@`|AI$W3a8~FX)TmI
zo&5Wx@cU%%d<i0-I||@Ei=VOX30^Maz65FFGoa^NbtgBr3qZlh#ynQqi2&HIbNbGJ
z09m?r3?yPi+D$nBCV%eZl+CO9A5s2P5+FVe$<}eW8%JKd%op%s@4n&E7{ZZ|Kvwdr
zGmUt=FRosnD4D;DQuFP9S|w#Cwj>K)Cr%p}>WE>p-Aox|7qnx35}*sk7OX(GF?aet
ze)JVj0`KpWU6)yFoi0wFF|P85lNO}_!z64Wn5Hf!Qz(>3Ws0n|aW!kVH8i8J!t~M4
zaEuT-Bjl)YMsgUsO+#rQ$;$h9?&AYic7}kV#B?`ZWsVh<Y{(qU>1~m=zR)v}#y!r1
zN7CYDxt}tlCM>fAMIzAtF0RIbtPN_=bO5?V=Y+0W-ai&6PHM`<1@G?&T#PYLQf-7&
zL-#a{m$sVMK>S53@Y8)%mix=vYL5maNr{M<77nYPMccm~lo+toOxdxA%{NqHQ0T&?
z<MpZ%dF(*JK8>}KibaTxzI%SfpEq%-zO;t>w+BmENiJoL0{snn{ntSvk?RIzkEw<^
zJXt2)u7ZS-b3*qE5x>gA3rv3C`fiw8pck14EDmjIl8S^4vDcR<PVSx5B;@Bjx<ztQ
zb`ttQKeK<n>gu=)YX62<ir_!UIxw^t5QB~k)85)F=j)tr;97w-@7ra+Gb9sNSx+CM
z8$K9D&l{0D+GwM3h`J_e%Pk8E#q}Y|rH(RHXH8`(FD;jF@8a^P=r<|Afiay1VIaor
zJkf!y1q&)!O>K6Li?c@5KQa5?yg$sdJ#D9=woq73{b(}v>LZD;>-I}WvHkX602&l4
z;%F1lv99gapZ+cmWp*zT@7a&56nR&}^F48gc#xc`q8u@8YNOIEIdp~o24m^Um9vVD
zZnMTxzZO4SWss}aMYkKMXNM$~6Foqn(Cro(ypm_HS4^=3zfizr`s&!Jnq*5H@9BtM
zijos`)?XOio=|#VBj|S-+3ybW@r{@DD6W5p<Nzn<GQz1cbNHFUdvWM{RTc+WD0qTF
z;mERB(+fP}T<4R;)>}L7I?c+{!_<OpW(T7akkv7MS40?}p?`sMl8rWBuf<=f^Qp!*
z@@d&BZRv2TX(ZdG6=h_+7?$Fvn64d|^_`M6RiU|v`JQpK%_YbKibP7{n&h*bF4)>*
zdHt)aI=U&~Z)tP1<QJE0J_Q;y^035XxLXYPZ83bohe~?Q?{|<7{hLR^wcdSB6xR*d
z(D)ImVp%9SFINEAk1WBN6sbh0+EUU{DbgVdZj2eHU!zuO{F~+)SBQZgR?_5em9?qT
zE`IN84X?3TFis<p?trh)<I6rM%z3fF-(vE7N8hbETS+dGZzc$v7obLn2g%9lAEiqX
z-jJGF0#8E|eIOM`Y150zXv%nFgu4U&6ht$9&Bv?KTMkrv8<UBz;u%2A)frhT<Qua#
z;{?)N)6w;^ISq;w{#Y5ExR5ra?sVRDD)^K6hXl2CxF3^>c5p0h_P&wl=)PeFJ8Q+i
z$ytQ(X<DP{CW6IMulg9yLVnHr4TJgv8bX(3v(rV#R3|)NSnaR?^D`*b5Sp#HlXYWJ
zO7~YH6f#g|adbqOH&Q{VrlS^zwg*+nlTaOquCxU7FU~J$NNY=_5DIa7N$&gyC^ND@
zIUJ?>kkm=Be$pcO6&moTSnEAIeF9V5mOw|~br4rEt!sO$kk%#w@Y#KfpHO5tm`p+1
z(X1igTT(I5^EdsfFn?tt6OqDFAK+IEc5Ng^4Tu#OT)+Mt1H`62<P-Sa0>1oK2N~0@
z7c&j3s9Bv}pNrBvzrt`oR%$Bbb?DWCi0?R)!WX05&%!qWirY4_@Bhk2D3s9nezMUq
ze_?{<BDgU|fml@K!fQ=vUH_3#PE24mp81x~u@fkC7R9yyHT*##G`G_Hhw0_n#Xc?`
z0f0_Sda`e%loZ%bIWm%6%U?bvLqR!#nN=NF6OtS4t)gde{%?OZ_FnLn{aRXpry{r?
za-3xtb53?1iz*~8k~r<rxQfmPW-Dh@YNi=UjHXlm@B$Y~l{MZ5+P8QrDlL^bbN!E(
z-MaVP+x=^oFj)T1M<$_=bR(k`%qNMy6L2L)^+C&yY9#=&Qtap>ZrRm$B(ehi8xmB(
zFH!-h6%)!Jatv$HH$JtQ5JD{O3Y3o;zHgop;j3hWBVpzK(G_R<J(m>ZTxDd?*ZM!9
zRc)aU&_=aXktd<9nWDw?xzuqhFZrEsp}lwPvqW0WZ{?^|Q=uJ1)Ut!<?Bc6g0RAOQ
z90822H*shb2@aX&sBpR!MP!u;7^)8<cn7wfW-hnLe-+tnb|Tf6B*MPe2apZft*-B*
z+lq+ANi?ua$A8(3X5b?}-@`Bg;+?A(zTH_!+c~IX@+B=0WKDf<D}^9Jl)KmD7@CqP
zS!9m~kHK5|P(!(bNe0P@T^0Kf0AXCzRA$*4>)mgWqh&{!OEXk5aJEe$eQ-6J!{cCK
zP~k&zj2X4c;w1Fu`}sgyiksDITBbDfhVT8?odEsSX(g&oUJmkqi}+NYb6BSROY(eP
zLM!qou5cBeQ6z?<qeyFRc1Ayr=$uM(Z7#5Nc~hHr0#vT|_>M+=JZ^V@<GwZ?Z)Res
z;yWs!h+9P)YNY?u;UFGwazJRT2uqUJD#9olSqYviDd_q?7HH#(+z+kAwEHf#15Win
z6J*d4RD*f=<uUPvnbJlYPIlV1v*i(Y_6Ml{mn9hZLe5-dwL_F-nUaGh!t!nO9);P`
z);fEs{h9+8)PM2YQkV?lF{JjpOpbX-O)QIK@75a*wZNPHhYHT{@)1y;g9=&y1vKRT
zyPU3l2U%e*UrKaesefbWzl{ixRo>o=x5dUi(V||gHG83Z%nZQ$HTu)y<YRv3+hK<<
zyIb_N!!H}hw5r+MafD_?n`=3d<H@I5jv&Sk5MmLnu>-CD{QL7lTN1}ab0xk-9SzE8
zqC0y(`iJ-b>n(j@PXX6C{?c=zpar)U;VG-S7Jp`o_bpn!lwkQ;vItu_Wu?(+tW#a{
zGAQNWkTPg|3hUgB*r}{(WMtvu#%g?`v2cl)2CJv0SdASNlh+Z2{X$1Uj{G2cBa&Vo
zfFQGeRRcKaVHEy$#`DO#6m~|{;4(mu2#dZ80(n*L*(>vG*5U?HO~3iUcEPAWLSon5
zKh&lOglbJ4<xQ=s^8*B7a(CbvoYF&NbuWLxtNh7|*Zwv>4+7)5!60p`65Z8<jKsXe
z+PDkW+OQwPlsv+I+(ZN&!G$nlGdD(Z2&41R^MImrvh9lZ&1jHxf7d%Gya?1GI~e^;
zPfrG;zgQULr?1@-j8u@}Lz}@I3}<C|lY@y#uFuh6Z9v~}B}?E23&0Vej1Nw%jA*d;
z+3xqWZfhkK%#%&D&lYKQMy8`q2f1&5@VD(^(56_=-bUMR+Wbz<!l^k<5HKK!qP3AI
zfwUhaNCz1wIaeF6T)mo3s&Xk|zhm0BhS@Acr0avGW?CS_IQbdTdJLoP>(PwkNdr{T
zwX)m4ht}GT%i`9@4L@c4lI`x}bG#qSe2q11_xr?;wh^QBU8pY#T><q-ZKfo|LTk3<
zX#(3Ppk>o7*l>a)-^ZwV_Dy73|1bqM0;mcbe%!;pB)k%T(MbPMGj{wK%~OPB!SDn=
zZtis{uI;lnx;&rPu7fhI50sfH?UXQ0b6~p>SH_y|1pJtGE)=s>($M#%u^dPJb%IDT
zDTbF@S;vhYW$&&#Vkc|zkbH|YC3V=E2UGvb-{yYD`UyGp3&nJLa%dGZpg@!X8CW|Q
z40NVT@vkctv_kuNzPEF7aB&Pu_~2%0bJw_zgmx6SvKV%~9zCxre+IHX^qgDcjAwA0
z-t_BSz4}LzyirnX!Pq4jG0{{_(dvihbvwX~{3X!hd;Kboq_E@1PS?GG=Lhfb<a!@D
zS(V=6vpUTX_o&&g|G2$dHstQr3Mdnut<~C?=^Kmt2cZWQ!TL0;yjT9C{E~E%_wNbK
zUK$Yx-xTlov?z>oW_qu(iC1`isCWpbmy3Bq2+2}VqE1=PELDdsKz(O>e>XPiVLm+3
zgp;qx=<-zRT9*DeO~5M*8ukS|Z**gUfuf<=?;rtLKPMuH)!m-tE@Pw17MKsTR8KGF
zRI*Fq&8?*3;2s`d5&f!$z<u7GT*!vgzf8dt+N{yhWvIlK=jMniI~;_UGxx5ik_mS$
zS&FaI_VhInb2Dw7wn`{JSVh<eN;XFwwB3EhL7Ns<@O7Z1JO&3I<b1mT7M+>Bg29m2
zpgH`~WKp)L1qe+_zQ2-wApYw-Rs1bZQ+Em(s+>etLfn4X4ZPWoP7dx464FjEt39kU
zaVYO)I9s<JMX9FkT%;o;Z&<IzPnv-HjG*5&F%8mMa<cFL3bA4jp1moqHI>g-Yilb)
z!b{L*Pkznj`J5@v1^mj%`CCv9amIvp!FJKtxyO?R-p5pEd_a((2WX(OiI%*!<S>&f
zsX{YxNM~Q|<+b8R@|+8JCyu_hJ(6Z^b}jgpTd%ejR@_rtI+O$|K>sC;ep;z~@;315
z-;YE=g+L-qgpSW&)Gj~Qvp=MiOowg}LG|X)r~VSa1{e*K6{?amA&y!Y^stpYj~E+l
z@eOiB(QwWj<3A8011(j|3XqYk60V|nVTHeyG8%sLRR$F^Z~EO_Y%V@lWvn4Oh0BbQ
zXwAvQP0x$VODP59mX|7V_=)-aY0ktl9$h<SAYl)d{yvGqAAgz1N!*_kiE8I)>Pr|K
z7rf@Ee&A{&;`4!`HY^zdxIK=h2DNdQTUb|4$ium?EkYe(0;CCrJ=gE4(w&z_Brrh&
zQKg~yw0Bw=9yu$ov=k}jmH3zo)K;~5?fHJvbWAvf{&ag(-;h!bw!egMK}#n8<1Fy_
zz=JQ<Hne$eE@&w4FmbXfTEq_2j4tAgt*063?oFVNd&2bQW=<Ds@zgoZXbh3pEQbke
zlVKqw)`k|H2mGnB`NAFedD|Yg!y<!}r~8@stNbp!!m|}uM4{I0u4GD5$($cT7pPAk
zGny6ZYh^C))?Xi$?eztx8%*bWQ!(W^@_f&u=c^f}vwh7vfi67bxbkw_6t{;Mr3r2t
z{sJ3M0Ag>i`vuw#PED!OeM&ZyTbO5INJ=-<8*bqnP|Y>38jpTJ@=O@po{Kr%u4Kqp
zN5jxNAQN!Oqa=%r_}^WfgO;(}vbN6`&_9Hk^ir>@0Gr5WH9qQnRQo;z{^Q~01_`^~
zs7X@x@#^LtwO1v6qo?Y}e)e}EYbN%wqfH*pW-%JSBNbt7-UGE?v<imRUaLJIxO&IX
z^^+9v7_YyT!;K=Agq$?cWBMBtkMYc6;<7%??8^_)7e&{SrnvLO_!H8<e?MzWX;n{Q
zaNNGrc%@adoS2Zx)%y%81;Bguw-D{w<k9r-1>bbCHhi@qb_B1C7e|tK^$&hqNxd&T
zj^9v_Fm<@O1OMjXz{&6MGP-_9n!Rx*+-@=s^y1$SXFAxe<`q`JeKjhp`inpu98`vD
zyL&9*9d&;{cuSAhCbh5l^m!0HRQJ<(NB$~ytODlS*@VfVy|Ab9Xh)ad`^m;WrZ*3B
zx<Dbp_~dv!wtudi8rsKC<@ZNZlI;@^4^id;D>=y2m))yFY=?f?J#%Ssa$_*#Xm%bC
z@QsVJTs~UsM{~E<$M`mmSVkm-DSDDOswzG(wZOpGl2gP9x|{;Xi`Ph!8}YoGc3>1g
zyNFr|dKFy1y}JCg0114Sa?ZAnQa1i&zlFi-&)46@1MG;gNswYVIy!C~9LD^Ew0fYR
zB%XSHBL6f(gpzf2Zo1;!ano(|I}Dfw{#*(|=4}yD>G|ER<(?DYrVrs3NWnI1LvF;g
z6B0{NXZ0WE|LOcBNun?K-H^sV4P#kHZ%bs6h+InAe=0~2Fz+~}#&IDDD>buNoyLgr
zegD$Vx`2j~HLW(z`0sJtji(w$p3{k>`8i>hzfT@TU*H=E5>hvIvc0@n^Eed&Trlul
zigdVcFJ*zY6=mq+POAd{BzZw968~QEQR&I8l=`>P|3YZPMx;F6h>(h*dyBpIKjUas
zd^SA7Nfj1EVa)!H9JBlGhO65zgW3-<0rW><kCGcb;}_HOnBEIP<c4w@Ic>x#RoB-u
zT7x1b^{B-u_?z*bd<lOmz}5g8>hJc^LE@osbsAL;Dxon-ATWRU$iZCLH?R+Zi3O>>
z>(@gW6q5Q-aYk4&J}{2#;~i`~bbMmMKXFM3X$D|}u3N&MTf-3?xT5c!hyQXjPm?EX
zE1_49-_D}u`j5~=P&skf%`gTW+?JtKB*wx~?;UAg#J=%4g2!}Pp(Fqzr=*$S*ZIo?
z-DJ69?Qi08gA?SI`-<nM?k>ba&d&?^e^Nx8AA7fL8O)XG$HL*r#=fgcYDrvS_Ai6?
z;*9s7biL9gh@<2FiSri=cd>urZM;48jgk;3hCbIbQc*!W5?9aXd?-aoC)kESgEYu}
zH=`^23o*6czm|5qwL=VGrN+e@e&?_}IfEWsv^^K}xVc-tK+Jjb*zLWptSG}Bw<0i9
z%}57F;NJ+xum?q|O}8=dvUTS<-S^*;K_fZ3Fv?&D3**CNd5-CiQyjSe{ZmA_Q^_)l
zjQ6n<4SB5hnH|{M9WRehl`l8_+t=R(cqY&EyK+4B+@fL;fy6y}2<7;@tbD#_%Fbb*
zE)3;_7$~DDCaQlsIa~g?SXc_by&3wAkG}C<mkwf9SrO34Z*@P(`b4M%dxIOJAw#U?
z2Y}NZ2I9fZaxi`Y^md5~;MzWiOD-6G>UH`&-{r0oYSvHMlz$nQk3_t6ZfXLHprM5H
z71AfH`ulJ63t(&gP;P5a@(268Z*D$ke$*S5J2&`g+6LKZYf*HOnYE=HUe9_`pJHh4
za<kp%<>=#_P-w`lJVW0w1@Fkea79`@w2F!gx4DJ$xbZXfFy|h>Iot@{&|%8wkMKbb
z*6W%GJ2xA(kU&*+0fh7B@~1H<SCmZ+)T*dJ$dIT{0(ik8_w@merbl>5Y70x9AKlx#
z#gqE&av4`!D<0VGt8s<{?Z=5m_^1xdLMd~W7M7UbN+gWbu0dtTw$`?V32VCW+ZriB
z&(6+$hJb+pngleFtSWHC6?L(mC2B!yAM2qD4&15@u)L$T$;)G|w;shv9eR%5b%mZU
z5MShC04hTVZY6Gp+jqq`+zkDtex!FN3RJhuglgIa@|rb696Kwi#)}@u5&V{~`#J0S
zzp?jC3#rG)Up^hvMx=j1$mEiv{>g#5zQEVzTS@5=7nKWEUY-$YFq?Rx9NvaZ6$o&j
zZx}nA%;5EJrzut<`jbs`VQXH_OzH2*b`?FY19UXKsy<%qtB+y$GImm%4Hey9eL_RP
zPDk{cVrek8?;y}N*$evI9}tfYAHSa-#$K$SwYb<_rg<WRg??44>Ls%Y*KVwb!d%ca
zszLJOwt#FE#xm<Z8tC5;5acV%X)nt4Ja`*m`hA=m19Q{<{?Z}2P8-I3l_C+FqCSSY
z2K)fG_b2w<y=lSR(y1=5aFz4LuBZ@i3Te&BWHVi^tV8|m0Gw{yPq0W6!Esvg=em2l
z|8>ue9y7RQcc7vQ+H8M*T)p5QuQoV8yP)xLP(=%OrOf?MK}UmgJm?y-^BBgpBGIUT
zI|PGuoZ=A8#~;2rI-%my4c`>4;Pve@0TMZG<9-M%nJk}-AR-aIP`8Dw<HEs)?ytth
zS^J@cx?1>+mG(7hcjACj%olvXiNj(HGVvCZu6_bB>7G@HjPKpkurO<(&1fEzw3mUw
zg$`c>)xtWO><!t)8d{pK!U#&VX<|P44>XMD4zYI4yjn7(GyFsz<BJtvT#%puDD|>@
zc)h-DS1_#8qMU1UH09E5On#UC1w}yi=|L~^)zo8Gx@O|@QuYnaYcvKWMm1loCj?SV
z2MR5Bz8i^Ao)hebGQ3B9je?!k+w}nj2@-knEnzgnqrU-*BxoH>HePmT<XGQ<{1%K?
z9!-d}NTjvWalhg;cKv;FGrPnLIM}$CH;=j4M6P;O=+-M7NVo;-nLg@PnQd|jo~bPH
zUC)|rim|mzPr<w>7{fSv@uh2^!@<EJ=i-m|2m})$P58i%Qt*@RjzR_^a*wgcBY8+W
z--3q)x><z%{E#OErP~^-3|R}=Io8(o^fP?C!;{mPu9LP1Jm&!S9c_N#jA}YFJ$M96
zNs36r>pp7SmQk#g<mXX=OcoK3C(co-7GJZ+3ry(O0qVjb7|*3t3;J+J6D{m>j2$Uv
zrKMualz_sjvi!o5%A!0Q>^K7fU~I(u6Z{5F?WA+X_hu@~mXIH(@7^FO74GADP-hr1
z@#0IkR8Fhq#wgS4<b?si)<I7lR}zBzi|Q}pFL0F}qwbu*+`nY7;=G^U9$YOnB;(4e
zs)@f&%w5j}9nbMs20K5mu4l(yK_<z?a;?|&Tw1#a04qUlEVg11_uJ#;H|=fASlfre
zjn*YpLQI^h%m5Y{Q;HVSPu>EHigFqLUyuZNco;B6$=nYUb_Sh5KyGKHn@qVVn#pAs
zdnDS=%R8fe)O1heg22*Z`y&E<UeA#j4%Cm^=l4@{(2wsVV9^)%@At_xHO;I)2v+jn
zrF7&hq=G>W-s7jD<;d^$>rHMO{efC6@?1EUPDd6e{S<z}3Wdm`tkQ^B&T%`FTR-ao
zvGocfP9eT3vy)x`S>N2Q!v|T{qng3T?{Egb8(X=INdd*RZY~)U<d*k+{>0SGiLp$c
zm~f}bIykzi?yvayb02&Y3`&Ah9npHG76L2izDJd)8Y~ZkEVQ7!D7OMOm_b)ST2~&^
zR}q^PI?priF1(&bBtU0o%kYJEOe);sK#|erENFIU{AdZ7eZLaP-Vg*N=eQ1)D3otP
zoQo7Izf8y>JvukAiQi~V>8a@YKdmJ=Vy1zLb?r_P-s+w@;dM0_&p*0N!sfrnxo0uW
zPGnhvnvp+#=0twS3QC&wgNBCQ#ZaHAb3|1d#^*8h8;YUCz3h|{@L*;1b=p1au_d-;
z8kSG7ET9KQBwA%q3o$<hLWqiJc}3zpl8f-sqNrq40n=};YQN{9m|DqG(E3xA*>k9=
zWVdyKgGGUF%F^n@_2G0DXwwbOV<#)nG`|(nHnq}-epb;A^)2koa*E-@m|me9&andT
z0g<tvx7XN?ZH9spFEjNG+>5c0*O)BQh(dB-jrRZ*#tc|DIz@H!iPSC&D_6U6RCVQm
z!mN<*-77TQU<bEXcQ+5$_u^B`){&oFifMy}pp7J*AG&{Eg9}tUDmk5-4`X!Rk=9gp
zsnln6^%0YUS{aGJu%xvNJo`nzcV-8bK05QLcQdsfv%Pmf>1m2p5Tejp(h@?wvLL|A
z&j;kQOr^SE{VC$yJ)#6XNC5>&6cXRnk~o8ROj0}t|8O#GG#0dQPOs{vX5UA!W{`M1
zZW-Za>I>P9H0En&e;mF?hVew0bXp5IwoPVow=}ksAz$g!=pf+)Caa6D|Ks8;Own0r
zP{<7Zmpsm8s#QSmB}#~fMT9|sgN~0)j1CY$H497M5uvK7y<&2H1S)?YN~2R|Ac*fc
zc}`ar@hK3pAk`CWchx=(rbrv0@jq}akV7l1{@Qt?^}~7zl`6}swV&)9L|_8@nfG%%
zML|Ho+Z$H2(ZlQY1s3Mjlb5Y0Ihb6!j8qKLA3?lZdvutAzD^JlYW`gOmWFA=(++q{
z7z}jqRy1y0M$uR81w+H}0OQbG_q=(j`)r4doi@>5U1Tz3Ei>I|PRL053)cN}BqC1t
z3mqZ&&NgPEfZt$!@)yNKf|Jv5N;aY`eT?#F$_~P)XnC37ePt;#9W^4l^;fdvY%w`2
zAdtobj*fsIiW-;aUm<m*2)ZxhF&*IB<h}XQI9RJeMF}5reYNfKeZ&Y?(`3M%nxqP$
zK!yx+rm4Ojd<edp-pS3)34+D(GiPcz#iRbYo^!$26n98PGSB>P4P6aI>qW%ws#^x&
z>9`o!`cv7rMpzi#*FsYvt;1(5z`4obEvyk;T1`492u=1S{WjL@-Eghe>j4n^VyYN<
z(4b^nUR6RjA&>Cr$U3ivW-3Hv61?BQqzCb%rxJ+|?^`m~-cPC_%$BgHmbG_{sRQm_
zdb5k^mWB!jo^xTBmz(n;`;Q`Nyi2?6jRq$v*YC}d6(yCW8F-s`+y&4K_lBl^9$hS+
zcZ<?dfrtj8;VyGC3p_N)R^5O+-W+mwdH6#-D$@0>Wbjv481AROfSMK~<#3`}s(bys
z$n-z1S(kbX8h9611BA*98!~?hmS4Y`Co5a&*ZMuy$h#YU1f@9X>uR!Otm_$SgJJYn
zau1T_u%}VuGUwzB-9?U9WG$>MvE~+)wnJTgQ95GY%V`Mkq)w7Suet-u*Lo9vIqsg8
z$=~+O_mkqoBFV|UV$c1woWp}>A_N^v1G%Mc7a|+$&X32dT_S)u`~GCMfp9%MZvwcC
zp153!Ij7WoO>o0-|12fX1JB2&g5XYfuPidOQ2Dk-!<}L|+2&HtAkv#@w1VpT#Tm@#
zjTtCv3<L01a|H*M*8{<LcbPXZO1Y`KMJ$^&<%NPHL*m*wiyu;YI(>7~BHU#@GiTJ+
z?nHvKw_Eoj_9XE(PmTB>nVbW&*B2b=F9pM>06}N<cK8#WT>PumkY;RIDIE*L7p2$-
z_r@tztbw#GJqoz}Q+%w%lw^G^?n*nSvHnsJ8a*5=_VPZN7})ZEz^`E4Ln4x3-+EpC
z-*OgeR@dt*rY*m#nfp-rHJ>U;QO-9enr^NHV#VishEHgWV#A0-Q^dFmH4TY3H}kz-
zD;O!^7q8Y<{uLd5e*>7S+TW?Ul`zwFvf)!m@<ekt2I5jM2qK{KYpSL@rZHm`b~HK_
zb4IYfe2r{?@Wzr(Gt}U|hSGl})0lT!mJP4BgKQpDv0baa+7>3vQRjfpFpU`M5DzJJ
zGgiq@@{)4%;~mo9a@sFR$;qaJFs(Y`JJ+n@zrf=}H8-bH_R4#It&Geba4sgB1?fOK
z#o$FHJ+K^E>G*$dUv@wHH(*_`gKWy^xqoFw6|@E9gdxp(@<~<izw;!!-TYxs>wD=)
z!44%WP7z!KUCI4L0CUi<xeqNY@?WSM#0)$658N$-j$yobA7O?9|Mk&e!=#4|UQJf*
zyHb!F5wik<8{2zctaO<pmp&#)1nU<BTe98o1r^CpAG11M5ATDr>R{c=^#u`I24I_Y
z0QH4?y?pTMKp&(2@bBGg8<b@1U$}%gXuGPL%PDAum&WD(kjo)hNB|HCLI;gX7k<Xq
z_#xjgizl>a_)91RP_j_N!6(hTARB5`5UtT@eV3C4bYn9+mYfmcvwb|?-=;zP7fZ&m
z9+x-BO8A4#jd%>5etAENO1$zJ-HrtXgf~-yW&0Q871e-Xu@+!51HBM<p&2($wd@u-
zkVMq3*f42pJuVm4YV>|_Ap=w@-Z_<FQRd-1C<rXbd~Q)NYVB;bZV%CN3<R6(T_&gC
z9w(Q=e$OAI!4+$bUOjga`^(oo7u+yOr8tf?QQ~#(c?i%-p$$Rl$_^e1;#OvWfhr#a
zsqQJ8WRE6Bi)h0BuZGM-T_YDAW7pmCum1DH5!ZD*L|W*(moXw=-a_iLOe~}nPfLZ?
z{aL&rUT`lBg*)9r%R}N-uVzsQFMeQr9vBZd*z?ZeRFPKspAY3^;iD$Vac0xw0n~6j
z2x)Haf17>i)z&u!t2uw+I8O@zigK)d(LVYt?d+h9PRg0`+EA46FddE_sebS;t96M)
ze_j1l^uBAq-$DP^0Clf{C3@KKc{`kcXD9|8ef_4JL~v|tU(QU;OfOk$9k9tvA7P~;
zM9INW6jbJJ<e?|$X3RG|5#qUW#1Iiv=-0uYCA{n{U4Yfw*I>IGb=?9?8+=a-x8gJM
ze7ndAQN~hS%fBUD;U>Ary<-vnxW9M~($(}QY_oFn6)TTzz`6PN)$Y1uc7*>-yL9+n
zK`=0n4xEi-gO<cG6y-)NPiFX+WI{-Y=r|bL<n&G`Reqv+K^0uRe~oAZu598ASCQUX
z6w)i$){lX7$3$}`gRWD+;StoYj!!))2EgPxVd2zwoiGFS80+`q@c7E1dd4+1)M9cG
zwg@5r!$dm^xPyX#QGp0?`qzzIWEiMgY+vcr6cu}H`__7UnW##t?Ot|}^{o#Sh9jML
zpc_bqJHbbHP@$GThUrRmo<Vnv8P`2uF8YQh4ch%)L4hp$EMTL<?W4bEJssQ)^W;Z4
zRGx$d7f4_G;l027h3j3WojJHQAp=TPIO;7(J}TE7Ip3rju5L#2cdi0lxH}=Am47Ni
zOWb9feGU`v&`rch=@Fi;0Q>ZvfmkbQ9J69om33;@kJ*h~T>CzXw5MX=<uv<`I}5nI
z*+Kybxkp+A4oLjuM3G~agvHL}mz*xKv^m8=%S<zbe}L{c{Ek5wiOnq=;fiS@5$50D
z`=ZUzNV>VU5l6aR&JyiI>BsFWLL$S3MxxZVb$6m|{h$wO^3R|0kbPb+o(?rJXA94@
z;By{f_}DfS9w#;^SNt)kdLw=Rs+BHkq(yH_ruK{eC!imS=U_4nx*7%@Qt9}DQxYPt
zvpUGb+-1j|GL(JX@_9l^#B!-8*lK;AhkyTdguK%1Btg%{4_<s{E8g73(yE{s*^R2-
z7o2Roy;REoE#MU1*6W{(YkeKKWOik8w7rjUUweZ-t{V$AZ20EmB>okil1K(~yeH-=
z4{6);1<)53ODf{~^7edN2D{h>+AwO<nzuQ6^%asrUbk3l22g29I<Ac<5g*(AlWJp9
zgESo&Z5C3c2%LJU5^zgKt1D4o!e-i87Rtj(`XI7U5ZvSjTb1n~Q=i&b7QR|HlJffg
zmU%PGri*MiHAl6(v`E*{Od{B}g_T$mq0<1xc&7cqpa#%XmZzkZ4n{-WHdK!bF?DH^
z3qsm-Eyr?At3oT|`Pe03A=N;F(Qwpg(Q}!5l?`z=@x_nwClSaWaY9QncH|aZ<L7~D
z3`NYVo+l5ls4jvXuGdc+;|#0C(UXz4b+ejiZx3k6{HI+SgO{j<;>6<7D0A;1`Y@O|
z(5Im*tLFcpN+2QbHC?FmY2=e{0uG8wCTBpU`bDmG&ot|F*ow%#6}wOMZnU~R_ireU
z9ddrdrILQ)*FIwH_xw9g>9~3oF2i6+s4Ea3+6x9|ikGJ<HkN0(dy5H?p`a0e7TB6x
zNsg4&tuQ_Xh91X2I}S0ZuhU!{O;S4tyq6$-XsNK!&2+Igx;;d&{LQX2yKM2cU<n#_
z0?EuPU8ehQB@H1X<aRhZ>^_@=%0EJQ+#-TMs7Suu!)+*GZE-7jOQTWV-vofO5&Tw-
zBO;+Yl6fJz!1SiHxUPjw`;ziZXjL{Uj{vC7Zj(jSw{UUbl&{FqeDsykKHBR67s8$v
zhbvz>2<e?LR72nq1Q|6+?a=wSAmyd2(9#UZa*Vdn;xyJ+&F|leM$19AiT<8Mz(QoJ
z*9#Tx)WSRq#h12%B1a!fH?`pGvI@5T$^9$oK!Z`oU&AGYVzvqDLax^bLAazoPoim-
zoj#5rrGvO^rRN_Tk&=LHmOVFcAn_vS)5WVXgZ3*&&O(|~3cM3^IF7ZEGI|TDMJk2@
zc{cqp{r|%RP_p&d<pS3uLAj%EoLAtpc6(TK`WTF*vjP^r{m19Tx?-zm^CFkon_Yg}
zRH{>zDVu7V31Uh3OL-)Ow~mqY7HO<Pfhw!uwF)_nxETh^wSdR!jW%XvVvxSVw{J(H
z+*M?Mj84yvNpTW0wBs^iVv0TtJ}Iv$)4cvs=9b%^<6!;KUgor)uiVun^?iHFYTh>x
z1|IfSN;(K}6ss8IP#zsQXVuX5Pb8}1&td&CpFTRTx*|k}F<9d3l(l@!oNPS&|A(e?
z@T&Cv`f#>w+nj9MwrjE`cd~8UHMu6+6DHfXt@nI?YrXvg^{mr*?tAb3+1HiMKEI)?
z7fVV`H3q%~Xaq?^K@C^L8<#KkCR)VvED!IG#gNrk!IbnH`#Q$In$lOz(lk_A-4|Y{
zJzq$Dulo|K#HrQgZYo6kWgll5##<6i5@DZoZtcJtR$B@Wa}CihhDkO+T6~R>2IN{y
zeJMnjFCGlNu(>BzV~GI`sD`k80wYJ(DLaSwC<H_RA|g~&R9xK1m@WWaxDbXLA_Ymc
zO&HW-tZSO%7T{@QUWu4?KNC7V&Q6cDx`&$*XI8QVV$}bh@Qd!Y_nGhdpvF*k3WpYc
zDeaIV?46smAUXubZaFLt!WZKfiN0VD8AL<-8?JXd$yC%ANR+hOh%4}Nfyo-Z*T4=3
zA3p#{SZK6N8nw5dQ&13{=GqyuBHNYifhb0ohjd|FSmQ{C;jozSB;%dJxAFuPmp<f@
z-%&HJc*7m^$pTsFpe@y95HmR;3}PDiIYXHNodMkl+UZhY;1GLOsxM+9hkge%CK7&S
zj&4|q&ldQ+@u;~M$MHLuNS&Fk2Itru8YO`6PPQpGjakDpy%dk9q3>xl!Shio5hqt?
zPWELM5-X+3iO73b7EM&_vc#?wZPegqFDQN?O<wUM!y;7|S#3f`oJ|>A(z;(jgY%Xu
zEVs1Z(B#CC7k$FTaP@ar*QTHKc@$5w32X4A9Hq%jFa_(YblTIu?RE<FSi(=UrfPu8
zez5)|r^Rcy+)2B5dUMT+&*50B1^ejn@Y`AjK3S@sh(k@jQnS0ULPfES-ToPIW{#RB
z^esB=M?>FXeMMEJoHgB?x5%#~W}#r3Pc^b@E~P69-uMk`=Dj3xCOyz;)NNS``1rtE
z?N3ZeVLp9SPPyxsJz!CTtw^f`+YFdb-7VTR5DSan73uRNY)$~b$0@OaUn^trC(DDi
z2v|oiXLU|Ai$##jc#?&2c_xjH2wvjL40=#x+V*+OKju<&ug&9XXW-;b@Z97)kZ89S
zV=egv<?bNWFCan5Qbky74=+taxF$CBu64%zeIEB;>fg`D%uAQ1vVoa)2*8;Hj|;R<
zcLZ2N>$=dMt)k@`=nQ^K;(#iVQ*iw5mWgjrL<8I|k9ySQ2}!OUWMeDPD%@EO=)zf-
z`7I9D5J#YYzdutEg$86PC~IzhD2)DTUbKQ{Ukug$KCfr(8nD)?MxILnaT^)|c`+W{
z*Vd)$FMc-f@etnI9Gw4~4LD+0Z##xad;#PeKuSDCH5)7WuoF_4ilw8XwZ-&Nn*4|5
zZfWIsb|mX?h~&Kglk4O4331!Fa5(z9Q;GWOHF?2*A(97skdS~O>#&*Qj`I2C4@Fl`
zm=%*J#!|Bj4BGyBV$&T2_`WBlRBZMBQZ)G0)+7RmFlK;QdOjuueA<IrxX!c{jl=(7
zwquBU=1|+SQsSS(U;ME!OH%4;12+e=<2BHwBO!)q9aYb_PxVDNOQOB1hc?{S%raBK
zECjRO|8SfSxchs)K1ks{o#f63++caSnd+UX%Ux+Nau*fRuZKoIVKb|D493t9j(bVg
z(p;ZlC+n_3<w?l~RLHo{W_XXJw=e)##z~U=f%Qux>G%u5U|Yl%G68cyxJl7)0tOCw
z+0QeAB-J?<?7eXGm*T*8lCCfzZ@i@&=I!s#-N$+8Ki*7v0zLLJauO~eYI>s0*c*jf
z_4OVX9c7<#{A-7>z|IF$d5U5luuX<$UN?gj6t=7@s;C1{Q;L3|*abUxO)?Dq2#yRh
zhy-nI-Xjx*@hvVZb+kkJ<a36Jv9!9J8<fbZq%8V*M3ku^V87Zy_^xEw@py3A%j}T1
z)mhu!P6wYIKhI!WG=MI`5Stiu$E$f&I3DU=n(5pxT%NSldK)=am)q_;de)kD`zlj-
zrCE{~$8rtOCe`5-(|hjAK&V#(A0n$9#!Xha>qX@*$um=tu`dIl0by{0fXAax>bkA(
z&#z&zA^f#YZ$&?mnDkQKXyk`&voY*U;yVGaZ^VXT$b;tJ{KB@?+@U$KfJDzeB{UEI
zlM<Lamw!X>BBUhieR-OSikevZSr#ZP_&lCTQD=Y)%_|%&+0JTLX`A&o>Qs3LaI5b7
z<tqWeH?ztH;YK5Yx{!6Xvo%5As;FX?Th!4WipD~V8+_CZ6pbNH#;$cX-@BAm%7TJ@
zzNS$cA@PNYoZ4;33x$TUC3e0+B5>Dy=cjE=Bt)Z#(=CA9M%KZM(gHKqV6bUqM!)_1
zArw%}btUNi^)|Ten7`GzglKmYI#r*-Zc-fj@oZekw2G4P`N3eVKuPl3axBCV##O~x
zc>sov<fp_>3$<U0XmhG1_<uq|N;neejE%*hf0h*xnqzv{eefq$mV!Xp?(6YA5%79_
z0ilA-LLcvMTcu{5ON?2ppx2i5w1or;tH}UmuptdMT@^)LQT2%Y@Z{JgEn{<2Nq;X!
zyovS|9MWxTHz?XO-S$w5BMr`2=II#ZTOfHJ%-%w)F5sVaME>|L@^JlZ1jg?g(7l7?
zBOolS@c3vGTFI#A^=3I6cnQztuXi1cTIvXb5aJVf6?9_iHkBrx-JS_r;{k=0Km?G}
zQ!{I7ga*@XrY<C6!4V&ym=tq+z9iak#7LufhC@hNpQ>nH4FRAlC~XYI{(zWoNhN5m
zMpTx{x}i}jM7QQ~`vCa?=4jtXCNsDYU6tdplyF?Q4nH2KIus(dDT}KdPY=-35_01X
zrbNN~z=K2w$cy*Ul8X9gx73vs3;}t(!6lrNIpFAbwz^o8))kXt!GIlzRX*O%Hjxi;
zH~`Jc28`%DeL3UiuZy)yALRTl&(EuUqdH)GN}$>Hv#~vSVs2oQk)!mj3Hx)lL<|H6
zGmvP}Er!8Pr1(P)Q=31g*qthD0008RseE#f5MN!uy#XnCv~ye=YaXAN2;f5=uvJD<
z%cxS^LLVsFtB<bDY_Z%vwEc4>;2$S`O8cD%sbE1HFnM5d4+@=D-JBA_?X0cG?F}VS
zN<mFcNkv65G&(FE8V&{d;_bstnvEA0lhFn-$E;<Jv+_D}5M!$Y(wM**YNPTs9f$O*
z^*;-NHU4Mt>w7;58hel$K$xIX-gERwDWl1IMkk~fgg&g-aH|_)CE06g-(~OPGB#y+
z<e?Sf^jzvocw$^QnJ)=+Q2A6$APh|?Nec>ITd<aO@Ed15Qh&rP)<in^%-0YVmy(fO
z&c_4V9F~ewzEjz$>1A%k4naN=MeT1NB^}_d$eudV>WgQ`;p^)e@Y&R<HAAr4WU*MY
zmLsiYCLy?F22E}p@~Jk*(OmCw4<8$=B(JD2GdVpqBOL?x^vfjc+_svYTuJR?M&!~|
zLhZeyc8dP^&jOa!5QJxYP-blFTHWY|h_|E*yc{`W#Lj{X3UATMO+c@wfA#y#lZLpL
zu5e0u+RoceEGo<wAXDhp3*rPs1t;y49LaY@baZ5p^(W5lO48q7bKW2i<vgM0>Je9)
z$v-Oh2_YBp+vmX-Q1sO^5~5>z0xK<!a<v7`7JKMHI$r}~tOP+Ut{xJ*cf!(kENUlb
zid3sB6ZZjoIUJ(0CnmGEv?ULUGE^L(9SE#W<swRWuk{tyfU{Ndzd4*f(?lCR-u=z6
zrJ1b#CBH+5>cS00_M~p3#!;!$g00+`RMsx}z3wq(BEKG9>#VpGJ?ypl%|xAlhQxO#
zR$({-a9D+sJt68wm`J;Wnd-F9v1-htFymS6(4M!Or|5-F5uxz)vdALs1T8RA)a`=U
z&e#6rb$qw%0j5^>p_iwtWRUAMF^AAtfgLtH5PW3~2{J^aTJa+@HCNdx>efjpMzgY7
z2IO=0c_w)APHFXcdXM}bpWBtI$7oAM1}n~}UfzF!{8>gxg=Ut=s41#zE3GEquh7Io
z^S33(FSST}K}yi$-1d;DabGc}{%^6@mjlu3SGxsz0A}fhn}0HcfdM1#6IpMTKgt%2
zAD&r**l2DW`twL4NGCY$UT9K&{W!(X@tmt2`-cFx4n<=}ECjlWoe=1Gi*i#64P+Ir
zXeaI1SJFL<!mJ%iV)8<{yCWhRh^fhy6<3!NCs0Juo#R<89F4xSJ|4X4_}G?24=FFR
zyTd9n0kXJ#hM&xKWY_iOauDn9=QPsL#8;B?U)Fk<!<4%j(j&UcD!a9$eML4WE!N}o
z{U1*6%a!s4(QbG9hsE@;l&Z%iqCSWIvOhuBUHL<I?B51XKIv3mMGuXmJ62}UloFH6
zqh84Ojfu+#*e>J~47i+)i0_#xYZXw8E4di-0jfYS9%Xp&cyGzrCUi%Z;_{^9TntBB
z#eguo_FwnY0!5SUfX@Ni5=YK%UcR|ac?t6uYnLt3;FGtXcv)-K<3FV+wz#<IP}Nze
zgZ7e$J&hZfBy_1)7iPKWx|{t+iXQ(SDdcJ=rsENwghIVWsh$+z57j+JlVFBLFefX?
z0u~caakDb?Hs^X;@6Bh>lknexK<+;t(^m$3j;b<j*0tv>5p_-p_lla21a7b+PdeXy
zQ`lV0m9aqqk4dU4XhFKE@kEmRV|XL-&%z0G4dO9|6NRF(Q1|Cfd&3wl@)!KU`zS3I
zS4oR{PZvVI9}ll`0$kURZp7Pv{63?e086kyAig3MooNbkJx<!zKM7z7<M;*y{FEPI
zpx60YXvXAsCr07?HOSBu1@DuT_cnPvSSsrPsy*72jj*AM?pa8x`i#!_xI!#HU};aZ
zgHk*+<~K0)kn=iUFW6bkP#=y)#&hW{7Y@X|0?xrxWP)=%R9Dde7Bw@nJnBvW(k(t_
zVqzhxndHKnni}B-JxQl%P`VKTmMogw8bnS^N=TYxL$5Saj1S=#uZ<m@UBX0N^=LPS
zH7v$+?LS_}TVkU#`X@XdZr4kT(Z|gT=_OKcEQlIIJTiRBq^NtP>019SRIz~i0;kTN
z&|#xgR+uRoN(_G^o`6+NvMHeQa6Xz6-IiJj=L$m|!;*3|<fUU(FDp9}2mkE+^h1?4
zW!uKy#o2r$+W^w1a(W;8y6fcw_wMTr_}jGN?F<*x(+I4gd+3}3kzBR_V#beKmO&}p
zaLB~O%*;Qh$3XF%hk`-SVB-GnzP{4vPU5W5>7f&9@US|gj0FS->H|(tKS`lZWNdZ3
z#Vyr2rt*axuX0v@vs=jn1tP<1zdMl$ggi=I(KALhu}dX@4f9i4IWjV8;)u&AkTOxZ
z8655G0C>0OJLFhJ`5DIA3PxG6X&V>2IWF@(_ZUW?cDef_nA~(Q*mr}_ei9KC9(>$a
z7~zWR{V+XJkq>uo8$gt(vc>KAI<gr)W8iah#;m%0`8rRH{alyLN6Iyk`w;xLCS!vV
zq{;SKWj5rp7Q_)5Dh2b4^Q}zNQnDk<nhO~LpCyJsOy(D)IOKcVS|+|VTS;OB(u-*2
zm|WytFc_0!S51$Y@=uBTqN#JfwC0vp`KsSJETt`e!~fB#0avTdjrMrX*wj)`ZR>&8
z{3<FWY}8z7>rH|E*~!t>e4=wD=7;HVf0ZI1+&Pw2%<Vaq_xd+88Y`W_3T@HB35#Z=
zh9g5`1z-f*5S~2#zOhvjN%e!^O-45H{g`Ki3Fz{A{5ai*KIYY$u8>DG|EaRQ!nJ$@
zj*78(b)JSg58#PB%BcB7sSw$Yi8P+tkr-9LfX>OuAs{e=EKO0~BM<i5BuMDw3Jz;G
zH4w}3gfx53C>oK_Q@Yr0^`4!S*@siTn>f!pW-#owxV9v_MxSWh7xM+NsH~&_pZDt9
ziou;?q##vlzDXk)1|JD=AnFm>WtBPQ2|eTiSq8;d2ONw;2Bn$rX}HJ4ARr<<JU+OI
zBu7Fx%ZN-_?H}NRu&!(~p8q9a%>qx$oG5eH)mfZRcR1&DdE1M9pV|KWd<7?QBob=#
z`T3qiE8(117IlzRtdz1B0A*YHd}u|8MDAtcf;x>DCS?XmLE7!`wQ^u{K!fHxCG+qo
z?h%D{4cLW%rfdA1pIzi4NxbedTS1$tzfr2r{qs7s2_N3dkgcMZfP#{$tIX+jG@!Ar
z*)o_w`QzVWe^!W&Y{V=%)Q)-Jbz*W;Q*<<aqNm4oz`u%@lu7<M^^G-UF~cm)h4Rq!
z*oPKQb=<o-G_nDXeI;dColyeqLX_ao%{^w|_b)!$M*E$lV7@SUGu?5+=k9D!C?7~P
zeT*678*^E*3g<bAcs9f5%LtS>sTTr`NWs_cgc27kuB2mR$x4+&)xk6%yZu7#_s?UJ
zMwbikkefo-=|0@YiTh?thhy(@TFWB$`111gR~mhoNC64JwMtdWX>ht6IDIU|o&8db
z06dC!6~i&iBCefHZpI=p_wL+e3@tD;>V})3!M}y~k&ExOoQ@(vYYqp#SQt?i!1Woa
zEt5h>&-%<5bvqyL&Zl1sqDBJl?@kAO{?roda#iyVFN;QoH)(})atB87nWIAQ{8&rG
zM<*|_9_`<~F%<I#-^ip|i`s+1mdHX=1t3UBc^D~aK^y<DsSG+NpXm9KBu#e3aQBg*
zq279@M9Af{S!nARG3`FJ5h)SrZ{l3pUdoB6b}Wx;ltYBb2E(g+Rwe#Q!zdF68Zqy~
zU0x7U)X~GExo*ab8tx}<rz**W)&+5FW>ZqjWeK6V2bxO+OS;uSUT-(V!ldG*06^!9
z-{Fqv|8f8nL;V_*biRWVv9Y_tRr|*R80`PM?6?j1_`pnjQ34|WL?$kB|LRTb2`FH;
zc_+_6Qztm_B^oyu&#IS7-BH8_oeWFtrW}wuc=gTv#MNB^y-0uX9Kgp2G5FKy?$r8+
zw8K2Wf-MDj0+6E!r$VK0))aC^0;)Eagc>|m-^0MuiOi0KbuAr4WV|${Lhd$FXi%yt
zwL007_Us<<KK#SNdT<HLZVDi!C`%40q#(UlGEZsHXJ}gBpn~{So3zM~7(}e;!c_Jq
z&0hX_&-CwO>m(=KWS|ms4I~Aay1*zhI_fk&p0wy$1CLnr(6$bzqm`p4fDEKcZCGPY
zx88X_tPac^*h*3)#!#yH&<jt)`X7{AeHdY9g)Xr^#UEzvfBy|PCBm*O1uCcEo|N?c
zrX!bo#b?F9lommR(oug6FU3;IXm^|cYT{xh+U~TlSbvDb1cnudB|;%zy($|>1*exr
z)(|9`{jf1D-K{a|)p9Eq02RWj_$Cdst1_rxwLc`Y;7kt`6q`v379t>?4@v^zg$cmO
zZKyI>oFPweP~)$eP3Dvfo>ewVsDn#A<P9%t;x@}ku4fGUCGS3?r92W?-p_RQn?wm8
zOGZs021&x9Lh>3Y=U3;MDbVEeft<+USErlg)(f(LTC)T<Zp<hoKse@pCa@|wDjNyV
zE2oGX$0a_=ouE&s6Vgt63m>(@h{@(~?XUW6Fv^@K*y!>8mI{9?Z=a)VpqY`IhC(Q>
zb+0PEBoRWhipY^{;_No2tTM~m;CzP^B|9MtbMM>;K__$@Fn|s>NJWaO;w(dpi&qzV
z#Rjo}Rqmcn3iaLz7{-v9Xx^ga_}$gu_u1nx_8z_LDCGP0&X}b}ywzmy><YvJs>zvh
z{5IrS$w!Rr5CxzUrDv(-;9_5lb1Wrb_q#}YzC@|P;0;?-iAAooy?A$UuVjdMC=vf7
z@noWq7i4k@wLT{ezc|>4daG<&Pp@5T^PA~MmqPk=xz*zZ6iZ;*&*N9!sYMpeg38OP
zW0ru9iu81vjVfakSzapN2;u@`I@YWM=xqC9T?FzixO@Cv4`y(C57N0rTK5M+;0oyD
zUiZwXLPfMm1meBXg7%`Io`b~gA24=WTS55!z#6<Rho=iDDByX7bT&*oGi{z6E1=#2
z9Uy06W32{wZVXjQyN*mtYoIayDGWY35b?i(lSR;llp(mai3p0Wy8kQ|6|UV74kl$J
zRZN3mIX)CI<7$>)W@Q7K)Tm@V{&nd5F_~Yw<l}q3T3=pl@wm<t@Znu%1d(_WrbZ7?
z?R&!(aLyr!$WlVBXGO*iugU5OE%uPshf+8^CXoPSM3$$tDDP2I$8VgRhT=AMf$0Iy
zQMrA&cJTv_1P3Y2#gx_-lC`KuW<!B4=P&K$B~H0AO$*7kEIRo3mokI_5Ga>qr05xj
zS{plqf2?&4Wb6a;x>AZ!jbbDUN#>Xu9E%mG5g<!N<-Xg%HC(Z7LX%b60JbnUN?I-y
zmO_AXSUbEiQFZpZroE8wN#9qkKyK$RyY0nQ_njLc8@c1THqRg({nS$Se8)A`BGHUh
zJ=uZKIfBj_k9{-Pbj}PcQczeT3iwZ3fm(P5?%#TheC4ank)SUN0w)T5km&Y?Jz3$0
z7+Lz<l@7bV{a5RCeIx{+EQVXX-bXvr?f|)Cc?Lx|7Z!C4f*-~XqTf>D!W73EKS=z7
zV&Svt=Rr22-=7n6KFCzz#Cs)~_eXaAptz}B_$D~W21EnXUjK0BP{Uu$bamXh(^<fd
zJt3&D78H#dZNA%qD~!1U{!Z6%Q?^{!7rZWKzB8BMR;|o-V@RzU#S*E)##SrJc7U|E
z(1j6c0^g(wZ_OKf$ycC^ie;!x=k589?OzC5JR|JI0Xm6>cd9xk3NL)@2np~zQ!4o2
zBnWgcP;_bV+4rw)0((oI(GaI1ZwHWVsCK%ny%zevn&V^R2)6S-TcbY4r9){5w)uO_
z)&rZk;?N}|ZHn?+ps;90_@%`d0T!#w1{K2G8;l$+=B75f8J+D=e0aprrNo37*;;(A
zt50SEGhbfM2TP^s??6dMqvd)JaGEUO|9+h-Mvu?#7uz#W3&CG6Pv}dQFxW^%b29tc
z$vd@IB6$ENs%}yBdwEpa3I#m5=&+eFf}FWG>?Q3F5x>X&DK$cu*RSe#z)jpDd>OfT
zdwUqhGjVWtcsV&P*&km~_!S)DjjoDPI)TYyaokn@_+d4cz1U^4s8{C432(iSeIZl(
z?Z`W1m@1jjmU(RZ6`!*;wUv(R?RJyH0S&z;qE4s#Udp_^?<5DlizWPK2JlO7Y~qW4
z4oOdDJKvB|)S5YULlszA0B>KMG>pr4*3dl0II9y|5byKBs@&4Gp!e=llTtp-&eZ7W
zGO*g$Ut3eoL%p<hEFtTE^eY+7j}VWT42tR~%e+sob+E0P1ZucuQ%93^(H_eRz?!XG
zX2OEg@R_geSt8>NfO!_x@Y<4Ei!Wg4R?!eN>)%ssbz1B1l?5+A$@@q#mmg(bng&^{
zON1uE54RlYv@9sX_`P)Xr9W;@5*>DQs$j)BFYw@mh#D+Pg#;M&k-x*remV9|_D64@
z(zcXzWXegq%!UAQ^^Th!cXyk|G?&s(N4tk+otjhmY)sUGV^gwPDk;UiOgln^uqZo(
zni}gY7T$g`-YM~b?glzrO<D5w)PvGW8Sc0SG6Vtq`1*r^6elG;zc3o;^Y}0oJ$-X@
zLg1oCo}M9>#eeW!KIH4`jRr4+(K$%h^xxa=FB^~He$;tT{wgkKd#26yIFfoKn2KtA
z$XKxE+|a#kIaP5DEj)?d3P0mQyp>Nr7E!FP@vn-(i2?xJkkG#bc?|HcPcEuxFA1}G
zk@L^nzxU7~+q|Ejwt(@EwCM_Zc{s=>DxSiI-UI(|VX^a}gJyY_aK^uiFprJwvnf`A
z6grLA6l-COxAiymZm=<bqs}F|1}uZF1Vwfm#&NxdHQZz8kAa$=Uw=p3tF;&m76cua
z_FDLm$dUkd*F$1kq?Zo^!kQv^C**-LgQ|*>Y7hkj9R&@9*EAf;rq+16yllT2<q;m6
z7c>GEB)X2XbV-I0bSLS${pO&Vzi@y}<<OKEg2e8XsMt=V93TRF_pee3X8$@@on!u`
z?lKe@PJx){gM^vC#ugy@l}wsc4bmB+oDh~7s}+E7*3DG%xB394qiX=P%!&61$*?f7
zA+j>FJ+w5H6P=1v0lZ>y;HlNi)oSKvZUcz=v^P=L75y!M#OscgfWJAEp%EdOnwPc9
zt(@@kCN0DAc%8JDl-`x>+1vX~90v#Igxr8&PipL+a=j5s{Set56Irdp6;YzbVx#qZ
zEjVCw#-RKDg>!-rt-S*j49DH8fukI-P&e`J7$JazibZg8ynlK%&>XO5pdY}=Mn1uD
z*G1MMy^`2gVJcPe6Il;UG2SdDAt$Xp*~raX)4^Xw#>>Lpelz4POeQ(IjDzbB3TtY_
zkG%;Y_ecNj$nxO?P?u8RU?a%JtAq|3_8O3G+E*Bqdm?!kMQe?>+V(QoJ2V?NCIF{>
z`p<u{(J9>diE)Hu2NVsMgc4a_;;OCpf2!Jqj?eG*z5kl$qPb0YJ{FMr!(>w{4e#8{
zu@C|n9y3aNk&ZwC6kr8LR=_nz9uN6T)ucNjOa%>s9;W7nCM&hoHI??<VTZAbZvw0i
zcCfPzcxpSKTm&0TO#KM&k}7Jdh>WzhS=tlTh1=(>_H6h#iulCvQn8G2EbW93@@Up5
zLcDg9$_}m+0;wuVK;_6MeCFs=e4`9<2`(-#FA`ed+BqPuIQ)!1g8K9^z(<WCe(?F;
zZCK3V=Y?qtcKbSs;jd8W;#kv-M*;4hD<OEeQISx0cCGDeev?ISi1&3yD<~u;QCgaH
z%(Esz=>)-*?O_w>QCMK*3LtIS?Lj&^IOORI^;Ff+0DJoMb@k0f9l!Q~uPv?wI2u2S
zyrpg_S^6EB;|pQ^jWWr%2#m-SlM2w_9-a|<*ebmL&k}g|!9pk}0{&-3EdU{VCAxB`
zWtvuu3qbo<+>+AnkJ)2E^AAacU}AxhtuyT|toQ!rwAv$|Mx(P_iOKDEo<=h6UP1K1
z;TQ67HfSshjH6Uz<4X@!(i7<OfZz_dusaJ!@z1rn(rE0~8K6be{V3(_Y_2P2=p<pR
z?O?KNZ{il+ipdZFy(H#GO{8!HDQ+F1Ur?8x&F<`+jcZme<K1T*0gcf7+e=neRR2Ha
zwzaf!L2rIMS0Z4d*ziuzya5y$0j2?nJo%|#58JhSEhIW{fzkSOTkP7IuEeNt^F!<G
z@`7V*7jf4#dO-woc2V%7m)3l~PB5z+?fTo=PzBe*-V8u5)m)U0a$xGLV;xy!Ghc>U
z=j(c(QWn;6cDL76(_GSg5rjr|8LPI=dqcQAn{w$$HtfP$?q`+IiZ+(o-pRen-oKfa
zYAM?wq5)sc0wc}VoZxArZ7uC&9MH>rurZbDxerQWS|N6B{(;>;3>rr&CT;JshrR}!
zbt&pJzXD*|P!pm)Qnc>}->rTz*)x@~ru7~LtL(yv<cEx*18+iFBu6~S^cY2OAbd9e
z1{*mFOO&)zppaNG$)&!5FV?fHtlO3rcX#xkwj&y8=~&oD*jnn#obs7VydJAiC;KJ=
zuAYl2Il`{E6v_fYVNPT;FqLBVcKywPvAvOrpn$(O0UQ71=qo_ni7pLLiq__~7Dw0r
zCJP=CUL?{oiNc?pS=)Rx3g&V1h7YoUU(vmjKy>l^LPY!iSX@WW7gBZ!^H8!k+R(<?
zcVJlV)FV0&o&?*oodwUyZsh;zDusBt-$e~uuitf$m6x1s#=zw%K~rf<ZeBw6lpBJd
zq6^p<Ha9;$!p2BV(NNa_60vJLvOJNfK!PowK0@V|9JW37$gGJr1sXdjnO9WvriOEn
zR260Ph}A1(gc#hRLYJw+gT+t9?R7CRHKhNb=|LA&G)b-S|6_!{ib2I38B!>e-ogkb
zaI$JVH$wTd^|Ze%@*O104_npaXZklt<`t001hq3L&(sOfAi+r{TD`?fcqX}C+4a+Z
z#35&ZyMg-GDNLcO?GNHXXn(Ggpf4yxX#g*EU0<;ha?m`P^HDSx<;>}hPR-RK@0crQ
z?&md{^&9A1>-K(L<&*wc_&~9hh^iQ+77iBXmKmmpwBWQS@H~nt%p++p0qf!wPzF@}
zV535mkv1k;DL+MZ<fXN+vzfYJd}XBxWJz<Pm_T9fs?-zLe2h8dQ`{h_eJN)Kw!ieQ
zcDvWP;vIheZ!3v9bhWgXE`>y$vstFj$<+^0LWcAz(N8|+_QgOGZzD3|0Di962`{_2
zO#J#`Y&^H2CAm{cSlXOH|Dsd_(2|^GHK|i?eHs<VdH%y;tM&Qj6u86G<@tTRS!`Q|
z8MaC{*b<jUS2Bs(SIb>d9v!-+E-8+_r=_e<3`**Z3yG;<K+X>LRa%Xh%xh0bXke^r
zZSxZ)z7%amMQ4S(?O`{PgC;PgqW&gQOWxtKCbd&>lS1*W9HgBL?fY&Oun+CwV93{H
zVrZaQ5!~wOc%CML^zV04iqr;;4>q%l!zfor{9T@Vd7}xgxQei4!YqGnS3N^f`|sjT
zM`dw+mtW3{A+*2k|1Q;Rw|SjE<#s;15%Rbf9$x_yN)bbKxrP`f<#3%GNp_JYBo&$E
zU*s}#!!uZhAYeP_c4M2l0F)9VgGhgr#RU-wG1QT6gaeJEEBrl3{wrqQnSK~mUk8fx
z1PV9G)}C8P^rM<Hen0rOo&SIo0skI$=sX|$)>ik~gEF#;JSD%X!Eho!_U?YZ9X(xr
z?fOq^LRHO?q;AsRr?8gb%|U3Dh*ty%jY8w`49S&S*@2QhTU*5gtgs7QPfxC}@sc%_
z7g@UPhr<kVeL9~Afiu{#26k+32La!wqy66JxZl|-Xom%gQ6*zzw4#|z12YN>c!RX5
z3>W574J=YfrD=KME#u6EAV~&WFd&jS+NvP6Vs}zQhu~@Hd!|K=bRA{#z%y+L$wN6<
z{)0xjYnG$A8Cf1c=aBKoawZcfP!%MNdh;NT9O?y{3$CW}{=RGe(JY5rser%#-mdxu
zjvt{6D!CP4DHBQI&dvSqR6k(+v*7Q*4aSySW7a;|?4Tx+r+1X06nBS&1!~s<7YFMf
zm%Ex$VCX29+xc$cl-+8^-Kgi+d;fnevu~g4m$#)k9BKd~8cU(HKEJX#+)WIj2wA!?
zK60%qtFp$R8?LN_@d=9qtAsr}^mYA^lTm$68rDz_h%`@7>|Ic9Ua*$*2Xm4rX+K1<
zr?@!?!UaK+?vHkRgXMNUWFi5-FF3UCugCjQpsb?T#l&$mrpjE1UL%7!=GBGli{7+`
zMr?bDo*&S{%{Vtol==NoHjj9`gPp8KIu{LPo#4&y`Z7WRy)H9i9*p~M+AO`!uzPKO
z-t`hM0WP7LIPnFsX<}dB9R-VM5x-mEOd632c2$s}#^$>ts6-hVUtk4kK3FL0ln+Z2
z+VY=yJqJ59=h#sE5oR*EqAK5WM9c&&JlDl_&pRNYZO^FND<19pWb1dH+~$8rSNJR9
zHrK-@G_@zm5Q-Bk3oj>cAEf80Z_RX6k;`WsMrgoa(W%MV0kH(+mOM|`=xtAe;+N8t
z$KN@hWb;u!_13iPg*>l-ug~;4{mXWLpjrrs(GUcxzlMQKqpZ6@LJk=$h~Nq=VaZE5
zz(43MB=ki{m3LGw1Vg-7tP~BdhI=^hcTb>C9TmxY4q22)bbzFbuCM}@<);2+a)pEo
z-_Js3YZoQ>F;|rz&WV7`zplvxfqTY|*J<~wn7X8T=lHoUPq54N6UYqaOOvOKyyvjE
z3F;)cdwkU(g%(ja!TQCt6^Q!fAg^W*prjR<YtvE}jgrtCdc-SJsE@D*DGR;B+`i?G
z@K8Y4{6h<=k!;DJ-BNn~QpvsvbU>cA`zOcYOZ}Mqt$^IDatJ+(4$_mHUBuja)aI3(
zM1}fQ!%lo{>6^oVsBRI$BjO)afzakv5htvpAh*4+wuLCp5!6k<;HFU$nt(C}F!n3Q
zh&QbP37YSOrodR3(r$hHCs9-e%`cGn;=!E1(d+&=kh!mirN|WJ%%!p;qNvmxz!p9T
z0v2L%L0}+|phyz2JnQ46fYnb}jcXf71m)ZxJs@huK`Y)zPBF{H6we)wDqCJq)noDW
zpS6UL-+S+&9X{R1VWY>Z`*;FS3cqO;S>Dsmu?56Gr{`-bp9=I2wRy{Wjjw<O3+a%(
zriEo8?DgR)P8_=2lRzp0VZ*NYe9+Wdzz$b0h!VAJ!Uc*H=Ou|>X|p_6ctQ-E@i%BZ
zukGOzd3vBw@Fa&&ug~qvmAezt*2)i7p}9rrJoxK=fChKcu5c~O>;nK(UqE`B8(oe>
zmPU492>e|Ig;O9~laZ`-mMx?kx`<$pZbT9E)@u=Q3h4JbU8(==`VHI=?_hr1Xmd?&
z+fyVK@-?2R(rH=UWeU677`ud{bGbGAtES7XB&@HATanGgzy;Qvud&G4z{>tlJ*+&X
ztJL_2o@|OFl*P}ls~hkSWqE6D0pt^&6zY_In8*w`poOu?(wXIsr*W+12Q%OzQT6o)
zre-hm;LGC)wfV}Ks5e%H(~_Hiy@y63hr&Q$VVsDN>UMO+V9>1MX0YdWgKcCkOkmn(
zlWhgV?6IpZt2(j?Agy^?PCV!UrJ<MFozCZ1Oenk^-uGLv^ngujWRFXrxYbb1xrkUz
zDhu7Y1+p(9L{LuKhzfBEB~^>7+4(kF8t%T8Yc>HBZ;)*QM8dD=VINm+;?&jI8J&~b
z>CZs2DLIJCj<p<S*-$Y&tF#d<)jwc|Dl4c@%fX$!(&6a>(o_*RC3xCdO1%R5mVPxa
znTK(o$+&{#0(P{VD4orPDa0YXMJ>wYZ%fQg&MhwwFfmdwuIwG7dW<%*{*q1eGu!B~
zV-U?b_BFLxxMh4rN}wONGHN%wy-sM@E<0{^UyoJZ1Izzj=i7%}0lj4=TidF!2q$)}
zbR4-l%NL*LA1hN5+9K-%lq^50fyn7)qQzrvH%uaJz=E-hiwfz;zXm7Rh;f>6nMl~%
zU#+aOOHdEFS#&lET)>BsD8PkU+Bk~qK^>QHhtc)VZQB(YOp*S`E(Zpvd0pMMYY*qE
z=YlWXOGZqyChDT-U}S|NKq<7)vfxmtdT)jpiu&(c*_B1}K#Qee2<E7FsWvOdC>e-~
z9<O8oo(G6I18k!TMy){+sVL`v+==*S{T`PF@sHy@><zj-pHiwWw1xP@C{YjmKpkz3
zqBctc=OYqDquze`H;8BI=b~TQ|E8RZtpz-+%nIw56vRMy4{5|)e3~}W;Z?&2c$m~8
zibm40yb{W{H<4lx255=u%uH~SP<7W7x}P2aKCK*-Q+T95fG#?w9Dea+^A-T14Av+|
z#O-UOh@x;Li95w1ZWR>Ej0~mhZR3NJ2RtHeEvcsrxoDAFSd&-m+(+5golaL2)m1|%
zqCN2bYSh_bHhfna&G+MK#p7Zlbv?c|ADn)6NhX!Yn4*l5IwC`5jJxp92VQLk1kMP6
z3H85`o>JX3g2xZ_ozgy&q2Taocy1QuE6_0JLWe7?M3vd*f>K_t>6tC*%sauwC9%I9
zB5$(c#)NX1;CevHP)^lIJAEr5zelw1F6nJu{zkj^)ytPbHWo)3#oL;76yR^It&=Ga
z#3v<E{|S^Ub#KMP-dfs^3so`P(;EQ%lV;nO<PE*vOPRXj^@<!@Q947$AM>d+0jX*I
z_RoX2Ouy{(>7O(+F^FzWlHHFiaQmze_zqP=$PY4f`5nG96Q`-PJ!y1cH9|!~q%nzJ
zsGFLqf0vjmo1OcQLe(ZBu%o$EDiPZoFoyzs1G|G2`Foo?8oH5{kIGu>w!8ob1G(u1
zE74;~b%BI_=GU$6_jqvo-HeM8QeS6X-#-Y!2&zS(J!zCF5(@}!OQ?=C6!Kc!MqawK
z9r1xH&kW8oE_h*Ix7*6PEEQ!&qKd*9R47^>Klj{?ptx&VOKn@-fRKUHqrp6O=+;0j
zPhiW7nMUnyR_8mpl?9g_D;>c9qKvA+`5ZBgE981OnN<UWO%D0v(mAUy;8#b41tT!_
z2?u2bU#2dKTSz6C6>TK0YN#XTfwJvsM@<V%ik}@|=c=d*`R%S5NUfrOpUqCmk*s%m
zPltSHE5UQlq8iVC$)Ot|CZ?lY{4AuTW=RTGf)Ih81R^Qt<*4LsjR2gz4P@Xsp8@Yo
zLB7;&+%oHtf(G)m6Zkrpu+cJAxO{z-d|po*?e|ABj<&XrzzdY^abYrc?Pl>y{}e;7
zU5eYz*uViI0^%D=Z`u-zE36^|&dgozVBj~B_j;fKb4i8OK4GTHk9L0ciBZjbQvRr5
zEoF6CLm5AmYtJe?TmYvo3`8r~ej`0q3zN>a*9Dp0o{?ae^X);P0X`5e>sl-@sUacx
zFw%(RVlGNj=;f|c*x;FY2F$E*)jCf%0=1KhG>X7=c?fI#MgS05Q#n63>LtHD)Lb0F
zyAY8$-frQJ7&|p@Ii0uO>NMQ(+uYsbb#ixl3aG2HAxb9%?3)$Sk1A0Sf`Oa{6``?3
zl?*b@63@kg>yEvIP#<c}g$T47DV|>stjB#^y_PQiGtq;PY^FM612f;IrLE?U&=5G1
z4Z=?Hkv8?`Y*gp>KV?$<7)?j6HRA4Z9DO@v|Jhh9T>hSZO1dL_^HWHxtTL`Kh9#+J
znT_QSW%p_sfLXiAOo1<>J+v*EDA)KXtY;2&&}QolJ3rOI<d&Ccl0w++M_rrg$55F~
z(>j(%&Gwqt&daLYGIK7^T#K35s42zLrEx*uOrtna;FW1uM=h_|5q&{wR$k@rXeDVk
z*`aT<>+1c+iKW(fGio?#_#iir|NThQ&=rtH!2wnxGa{nMq6$n&3JZ#8qKu(`#q+R_
zI&Rutb`uleN?9#9PG9G~=RXrZ(#*JCwmcCo`x(^k&~AA6h10%*xPPz*b+FYehrS1Z
zr-p#=$%2ib`Mh6ijk0IMe5jsLO@lh?yhb<J-%;XuhJ4I#(I2nlo&B=cNlK3iWeMBM
z{sKsWIf4fo_)>(tvT8^h%q)zI3sDKwc`8UuHTbPBrA<yoz&f@5Skg?x+THI{#T5VA
zna@w{+74rBltxlyav(Ak^amB0l)Q>N>ky0mZN=-ulw)rz*a%cn!#ylOubxV~G$5#+
ziqx@cFDUNdv8OC@LF|&L$>~vZ>6!hsxB~102$H~M<;})+SoMSU;EW-`HuYgiYgzl&
z0wGeb`F>qNA(VtcqMLI}Z3dy2PXVb>Quby(&xBz71h3^DO!<7y5St#9Jdt0mysG?S
znUB}Rm{}-4W$)(dwPJ>Q8ZcC~=bGF9{<x|Va&GJJMdaoV^<2yTS?rkQhXb48;0H*G
zV57xhj6S$N2G2&Ej(f`MZ%B4MVx-;da8=qYG^19`Kj#HJEku?TKf3$;#WI)RK9_TC
zx<HevzYnIa$qa5Mx)_edpGK>OP$XQ^e0L)m`nVT^h4lSbfNItrkqw35y?aKY1A=hX
zV^o*-lPwN(+ZKg(MyKuPeC_9yl{Wz9#%a0N%cb+!?r)*M`b1P8gwFdn^i4)#lte5Q
z94wfG{QB73EI8py=CJLbOI!YV7blO`vm~BX+TE?Ek)Lxx7X_|t7(OBXy56dKwSyv9
zGK2?8`^BQ40=btAF>cNV?49P9tp+}*H;mLXZh1g&*Fj`jnVE9Na6A6EOA}zSSuzeg
zMSZr(>F(1F60hy?CyDIGDMY+Y5^`BmXufe8{BQGwv7JEDeWm8-O^=-E*@=$}^h0<1
zE^_Lbfh=@JD<5xAPnTt(o=v|pT)(E$`zABL8V<_#aNoeke}53|6u=N&<a1Kg#h;tH
zvyH6LItq;E4^4gM`_Jy?N<#r&mAiRMye^{q+BF$U*$Ut%%vhAoZ>>)DCBslYBArIE
zgOi-xJXX7Fml{6jXZTrqN**}^u4SJOr)t#ePia`j@U~0a>`{8+hU=5#TySZvht$sT
zraUB>_&*1;jRH8y&hWvtu{ztT2x5{2kg6aE*%GEj!hA=)_y+~iR5$>@kI~<LwY583
zF{e`=JU;GYUoU+P|M_~Y%iB=1IlVLtVi|d!|BYK4gNM#NYCJ=u5vzJn-_DVz*<-Y9
zG5Q|ZoQz2c8AT*HgrjQ`S~M;>f1U-K%Q|X{`Ku6n(>SY!u?um2azYzwOIL0Cv`lPo
zXHyKLor~QOmx65tr3;{YYKSP5;XGC8K^mz-^DV}uisx%W{EkQLGD66sOHPR;0N-nZ
z*9xNWS%0PXK)pGr1$>@ZE>{BSp0PXBXEwD=aRJ}^d5h1m8`PBKSRy7WxXMr-Prw0u
z?vHYF#ELzF%IYfZ-k;993=l!xg=`KrK{OW?vlQLLvJ<>_?0|W`#eEaN+Bh?tyQHLO
z2D7Dkp&m1Dwef3ORtNZFcs!BqPoCMerC(?;o|o?D$E9ysv=2PD<hw)J#YuD9nRxvT
zD_B29bcG?V=9fwxUuFpL)}YUz2^hfm@j(SyOx8X$U|z_7A2V%x-94N(S<U2=zg!!h
zTJvbPc)2?(x&Z2<8=`2AkF!#@sZXl&tOI1}ZE$sYyrIM?N|a`d+UzcmChi`q)0r5?
z{*ElwywN40uZ-^YV{2oe@N}5eNa&7i3G0iXm;M4x96zBX-nqOox823V5GplZ9jOuv
zk0%gV`{#6_ZDyvXrsnD@CnvXfwi=m7<+4#j3}>$Rj}Nf5iaj|{5;78RTxn%;fJq6V
zgU(OYa8f>~1f6^XIGXXTHvZLRv!k`o<SsSyTPj|L-^=BEPX>iYPd^}AD~gOD7{GD$
zohdA;tSl{Uy(cbH(cjm{PQ{9-we>&XS9uxrs01+DmVH_^A-}h8a-g^>9UNm}Gsv=U
zUSClN$aa*rmX>yMa#B`Sw)4LFz{K|;vUYIz$-m_EFqSwdug;K(dhfhUSZxqsmjwA6
z2*}pgbP!3IZb-tgohP)Ky`lR?@QFOVRCevXma#XZ0f6`_?>#`;w(i^xm1ZNKv;kU*
z5BsJO|H4~QgnMfjPDXx)kFNu_wYYv`qoqp$Brv)T28&J6ljhkP8`<ZR_KF6AKDtP_
zM!hg;5bp1s5^s;SV+4O^f|w%`ftom5hn?f<t?+huy+VdcjZ!O3Ph|HtyPaQw$CfL$
zic$eicD_0~I&?AKpX{}{DF<Yfs+8OzdAe|LR8J}<&`K{<y7uTKcQ!DwkB*_@q~>A(
zS%^&wqiwqWc=Gu98|d4qJ<}9GuSa>o5H9>O+kT|a^VL@O`nu2Ybg7!ub6Oo!Rwi&4
z-n@!bz{Ktw@#Y#=76q;`WuLby!%e;>w>v0^v|ctUKfb)WDVUK4?(`>`2&#%H-#Dyj
z1C6$sr{$@rf|#7tOyAfLkTk$T=QRPC9H6xQ;jrHQeRBH*f4sJ?8%HRp-C5q@{d`qr
z(D_TVHY*t`hA|}57~>pf-*Ew@sOS_eg~i^*x2?CcyIn9k=o1$Y27O^N@7)z}qzB2#
zCafa>lFw0G(d@h2Gst*}T}L+NxZUG^e-wLutcRqY#+b>@Njfw*BCHeC+baPGyCzss
z(i78AQIypek03U@Qyzsj=0UZB=EJ9lXK%Ee<bJ0a7iqErBQ}$otg69zr(RJColXd=
zbQQMK{Mu{G^?4?C;0IMDNuSOaob_yTn3<j+T>qnC$GF|-<?H|)^_Ss5pNALeJfI;&
z=V7Vk2rvqeqG9mIWI!PeqPzm1pZ2e5_($e-d+a0|zE~eL^&v*xI$U4=RTo&rcANj@
zS?b?QM4!J}jAJsx*#dV(U1o1O25-|QaDhB&1*Wo?cUS&~KzmAkR9R5!K;K^-MG+ZD
zq{iB`tRw<W91sf^!+$VVme%H{qPWZkAOIyz+4Uu6QjgiPtbOWr*gt@$>6Y5K))Ame
zQiEiS%U;{4)#37^a@O$Zr=LL}t|FtW@M_4xBjLw>6ZV19jC2&-(+EGEsg3+%5gCI)
ziPAo~t@#6m-^}NWU>`moB=vDx3g`_>`ap%z)oIVdqHeG4=ibMpc8v?w{h}_#Hd%c`
zd}#wyJMU>ae=EX2<rDx~85TJt1cY@cfL+YMDGL>IpqG*)fK3c|f$!RjPOZS;L}XMD
zk(SZ5Bg-#pcf8w}>$c}|*goy{`R*T-ioe!B<zH0q^?;r{S`OOZnOn6nsim4A^Mk69
z7YevNEuj37#*~wsoJxRxcI{wQ^qWR7&<x_zgk+kH>w|-)tOCIO`|_bUzynEUoKeR}
zCG-=6*<^*;-Qa|OSzWKeTX5oH!+y3rfkdxr`&R0Vxt=FFxg`>bD4d1aOUl*2VuRiy
zEM_Pf>hNr(K}As%hh<a{V<2efn>ERL1~sZP6%|;5us_Jqug<#M|M~uUI%hU6d3ysE
zDVGY+9*&R}IRFbeEKFv77RJ2#PQhohICK?GaP}8fva)0p%nK9SlE74Xo27rp2eP<)
zE-1oamnGLHi<2lpus2@OKkUGF)~A0Nbc5CBUvtm<n?H|~HvI>KRdDa$3I80e%wgRH
zF|nyBrb3=h+v)iYr0!o1+4ZxMLpzA9JpN$$W8pSS1t{v%TLMc7nkYyl@Ev$LWtmeM
z25eii8Z55>fdV;0GE?rJy}6Ejt-U|bJ1pR_C-aTC9k({xwcB)*3?snuelDwh2aYmT
zd>y4OfW%pB6%DMivn87ig7IHYxuw+OV(se&MQ*3yj$hBc_md0z8OZCZV3i|Ym^p2q
zz55o+15RmZs?82@6VtRBeJ@9%41ACYBvS1LX9piE3RVqu+3+RvERF3MhXj7D#6uJ5
zdEr{CB|EO+ZB<=w(VvRlnp{hQv)ge(8lvqzm_xJq?n#cLcs9>U#4lb-v|I$SAEhrg
zx3uJOZ_Nb>zV!?6{rIiZX#CWCI&XDwaC9_90l30w3=*y^fZ><7A8Zz4EheL3q8GA~
z)%B&&*3CMvOzPhxVNu(`#7G2M{8CbqlG3iM{0G;d5`%t&r0@#+<~ycjj^4{u;H?3v
z8gud?QMsu!xxlZ!&3>3gC{J?nrvby`0tQ8c<~s1Nw|YFIF%V;xR@9C_O$JSbutZS=
zgitHDKpdt<rS(*%w!_w*ABdQxWQYWeGt_VNO^L!x0&ZPF&k8S+7Z~`>8nJ7FClC*B
zX%e^FEOk^wn1YyDdED8SMopY;tih1yd(AbEN}R}<$$(|?(224TGBti$#kg?&{UE(j
z5?jGV;quH-8kk{JtzK`r7$ke;9#$e?o~7X#PC<HVY&uiGqrb>PE}g+pT|eZiqiZt6
ziI`f>Pj=2~A%Pf~pDG`9m}R=5c*F#l8PQ;rq=?w%hut??=4#+zXAPOHU^A5N&vVf?
z{sg-uDB@-yWu#_|bTn|7mXQrZ`f^q~t_4}QX7JPG#wT10sXzfzq+m>Wu!#W&reT%u
zZeqEVvhh6n1*jDz7S^XHxHy?OgjhR=rc%9ff^V=Q&!WpK6M?DEto@!iW8>0tGJCnG
z7~9AA7Hfk9;ufo^FmGj+^0}<JzjT8wLvk}~>13n<30Y60q@bW*b{jWYl2YMD(ipiq
zH$h~0VijOS(3mVARat304Lks8nbT8i8xS8cg83Wn2$faOktg_)GROv+z=*E2%4A3J
z367Vs7k2hbAL@EYs9`tDx?E{O1l*uc7$-484;OKYLy^UWiT=gELJP#k;gH;tLisw?
z;MaPtKzu=(6(o8BwDE5dn6&}<CqBPY4&snlJ!OP@-HTj(fv!(5u>ru~pr}|-a4;A+
zf3CfJzkezN`M0zeAC*&iUVb=y>Rz$rj$E>mvbKDbb49=t->{PSb46=jVr67QVt8%5
z_5X(n3Ils;<9--wg8!hfR~(bW6)${4{x(uSfS02`&sght3!Noi-CiFY20erg>ljIU
z0`~ub$(&uoUw|;WJzwC7GFcF-f)FFmSaV5`yUA~G!JzO^P?aHmuAKaUPN{}ovJe}H
znU{`rWmC$yo*phk9iTl-Ku<2>TV9uvc80?lhrcOmJXYaoZE9&_e{SsPXk}|_W{)@!
zRI+gxcC`GB4R7MB&1!+S<Q6{5s}?hgGcOL_B+{j153sYdGdH)kh7sz-_{Yqe3ac0J
zZ-RU>7)UAL5Y~c~o6yzGF!Ph`%Bc)$7X;k}vV4|+28gcW@%7tcT)a)3`o_lA)_%!3
ztPv1CZ+5mRaLJ84CEdRh^mMM%Q9P%+S7mg5_~7zK$XiEk`3SeUJ*vEF{U6HyZssNC
z2M3s`0AOdFV_~uIxf1SIPEp(Z%U~!=hOuVa4Q}ypu#P`S%nBxVhxN~MJfNmbK{+Gq
z%$9CE0`4wmkifG>FXWH)F6sz4szXf;2^A@0bwNoH*m+ca9gPBhMAY7EwiO9(NAz9H
z7xJSEL1btn>u@ldPQx$fqR!Ed6eU(?p%Eioz@^y$@OqBI<~2Rx|9?cCV|ZO%xV2;3
zPTIJ!Z8dHh8x0!USg~!hvE7)BZQHi-?f3f5xz7Hv|E{(7T64~4K68wFq<>Z@`sL_B
z{?ZxqYGz8pqNX=d%OCv`SgtQ3(gWWwfF1DB$MySTi@L~K1&_^A2fXhFJ`?K)y!cQ?
zYI>Ts#!rkU9jm@j)rKhwdmQBKy8&vs{(Tam*1NEG(#AO$qG<G7@2HL84Z{r-@e2}V
z&l-zwz-&!H5C&xXdZrbF++Cxvd(xg*j*rQ%3DWxRjhYJQr4|oe&#Xq%q;Jl~buzd-
zbYii_?!&*w4_5Q9cwZn;jxhsZ-$72N!(IKYuKko}rmIhg&;R?H1piPz8BOW${sLO7
z_qPsyC22H>s9Kf8)Hmh@`*Y7N0VthH<}HZ0O!^%*vkSzko{O`H?Bv`}68m$=^qt=g
z@f-yx_Ea>p_+AXpi`l<kH1~vortXQ%EPZBBobfTdUo=d3P7c(Eek9=cDMSFeIKch^
z{qgzx4e{+Wt-He`SIIzHKJ(2#9vV-?VEc+`+f%tunS~Z!X|0iXa)JwhZv#Ajytx1T
z>SZM3p?1-8tyzD_pwnP=-x{^Qsy-o&Xg`&3fX|*T`$H5StLl90Zq8T&np-n@8c8*c
z7G}WiN58kbmSrPpEr*9(FN>1G>I?(-7s7f$--kGyW9#B-cJnt^Y?hB4I4k+D3Un9&
zdQe@tp?FOQTSla8`9k+VfTI6&az9GdO#Duwz@Nr|@($=C3ArttxLJ1RJ9x#3i7WwJ
zk*+eQzWmI7Fl}P<jW1UGy)##-Zao!0<@JodqVA#$GXD;vhC#)prnF}uN4?DxRKvE~
zg@gYMr&pFgk5)_h^Lwuf9=w@}si|>4OpE3zZNx30j=uXD*8NZpXqC-9!eK%S#>QrI
z`<;<^O=J1R)p1c|E2f|PxFSqT)nt|gi2QSJjtD?pBKRpL8qFq)jYK?BOKWpMy#Apc
zvVqCI=@k`yWzAqpBe~F*x^HgTAskM%UL2`0%)c~E)uZrNnY+GHR7PXN6dRk!DacC8
zgvKC)6O@L}iMOKyR|TEBfH7L12-`?WLP9phb>DG*Q6o@R%WFkc9U2*4RhpHRRf!y#
zQAs~GE?0A(QiPz!Ng|Z2sG>L9Kf%VG>?TWobUV{4K+j(IEai}&p*J`;K|Zv&K8!Q;
zw)78WRya6nA3Ir1q|4j{M6ZU10)ww;`S!I}D<{|8ogNzS3{GY$%C$b(B2JG@%1+Ki
zBg7^l*xR?VhVZ7A6MItHycS~3C-dhg9=lgD5qER9vvYKGR9eaXH5mEfNbJm<27MFJ
zj2g&<fNJKsWAr_Z&mZ|uBCF0}4>}8cw`e+#nT%I%L}_qCH=d2t?5Q^Ib0El!&<Y~e
zrPfdko96%l20|kbcn!><zpY6r8#XT|&~ZpMp$zdkyk*TcYoJWQL1v$oF|;x4e`G#$
zu5U4s=4ho#;8kib-+mg<!DzV<?J^@`<Lq|l?@XpbkhcWxH%v_<{0Hri4vUKDd$p!f
zPQd)71Gj8{U~{~8PU@elD4oM9z|lL{qoHd7(F!2#u}501LFD&}L3nknWfE&>vl^?<
z@Iid{@VPjW&mH8}V-v!IA?f&|k)ab=!SW}v%4ZuMpY}TL%ZDE0gW3UAuy{AYj+FP{
z=MWgzqr|iFPQW_7Gxl{kF*n{)GL##;M%LTpTAbu{IJ)%l-0*RlrgDAp;U7a)dkG~y
z#&E!v*%+xgHZZ>^E-9g{8DOTLDjI}oS;Z@nfp_>ZK+|K-5_iOka&`OAzy&?QKM1Fh
zHd}igJ(3El=`jdT4llZm&JtYFxYvF?!0}O^N||ss5`gp{-Ph8Jktp|%cA}F}RVdb)
z&+jGbpM**%wRLt{MzIm4yet~9gJSX2F~DJ-&D>-{?~#MMmcmBGTT2jPIPSPy#8tyw
zGCH0-A~h1x@Prg6>v`r{`y1HH57Jo9cnkv6HFGEO5NgT+ko8<jA~3{bIpg;Tf@2ia
zyrZ(Kowc@3Di<Xs)zT_{nb>gHnw1!ygfP7!A`zLSaY`v`4Z5>|_}ktST>J0RN}xjY
zji%B0%m0729-+jbl$ZaAQ3(lXc@`984ax%e(X|DifyhHtG{u&Ogb5re4rsbv=1L0w
zpq(W*^=W>0C&$c)7TM2`#e=>#9cy{vZC{XJX?+T0;pAmt;NYO7<l&*EotqsH#@BEo
z+R<tfDSa*n1;^LB=_yExg7!agEb;f%Y{?07=-kHauoY0He#ng=Jf5+amQ^UKhcP>L
ze_=PSlI$Yd&@ZaZbJ=(OP+c?=O83${n}hV94UcH#s!=>n`d&MZW(Sj5s9U(klAhXm
zKciHUSt)Z$gOVx(63Sss2m4@Rm^o`4Zjd6w!zJ+{{=gx<$1DFF1ir8r)L<El9Vmd@
zic*E0jQ0Fyzr>QBj;85dAs1FSA9&TKr#L`X3x`MWzLz1e8qJ?^e&dQJ&cu+Ih@qtS
z#Gm8^NJi5YcQo|BkWSHQun7lt;C-(7HU`+(X`?t%vy-7PNN;*G<q(3ie)*c96nI&J
zFj=1M5rujAWCd^=02ka>{Vw0!t*NWf5>P;8EZsh_fI)j=wsL=B5TWn^uMi>zx-w+@
zBl$O|ObPHU``eoz^*%M@UxR~2EK*UeOsy}^wlVPt@NrL0AxJnegd0?Q6`}fLUZOwa
zlpt8>$T_=*<&-5ubL#L2f}=9K$qx|gJ$_K#AVft*%l1Qj18C!F@QS|D4z<$;*V!Y=
z43(b-v^9fL>q<c-UY=iebW>3<DCC>7=j)@dt+3ti!0Ln+&fniqx8gbN!qV013Vo<O
zxTy<5Ff)dBQNBFXB*(`Ga<VKZ?)CE36L?S+rP3Ign^+(2og8DL5lxh-2X)iSPWGQj
zp|N}{$HB&t0Fd$BGDco&G>C)VuW?&@fs9Hx*oSO>CvNtX*2<nBjYIbqbD)o?cy4h)
zkY6xFc^Ax0lbbUM_YBO|YegaQ$sdmw7?|MTU`a{Ifq?@xn%mnu<iI@zSrqV3%;v~E
zZL5j+@B-CS*&9%!R$kf$rmQG7clRxG*%x|S>kgG0@NoC|+t$j>$FsuM((3vA1REU4
zrlko{lbvU48u6+obi(%`C5ZsVoAYA;(JvYM;SO&cN}0o(3UAhq57vyk;loc*1vR>{
ze&P)uQJGx88d*hJEIBBqH7Kn%Pmp)0OzUL;r|GDYD6}I=;;+<aKe^c%x{#xU$!s<<
zCC{}F06MmCFpe2r=>eI;MrPK=uq=7E+>zJ#icmHTrLS4g^F(|$NEmG{vNr^MKKy$0
zqQ!*}AK0+ik?D!}|7S-20cWeRCXV}skZu5N{E_Koel|kEBY41+#<||r$-bevmX(WT
zW>HCOnDAR`2<x~yh)jKSH2gicnL%e1NOoa-SV-K=$^`8I(;5rfs_T>d(vecNZbvda
zsK@m>Ud6`;c8*!HO}~j6;M-X3A8lVA?(T=7G$jN;@3V17%Bn0+^)OL6D%iM4%IFw}
z<`fsFG&jdo=j3z|Vxh&0>J$tp`fB%#;6LI4|E(WLTyWZGrm6jt(b#nM$&o;%fN=et
zEYw}tM&RE_jEzRWV<b$ElN$`TG<?~`(v!4R?X+BS1aegodv^|RHa;#PSI<D=NSh$a
zFw<6Ck1yrE)5pSk5{{dR{xcoLC!E-dQFhZQBuq468GU)zwbOCic|Cp~os4S1Crg@%
zvZ8%6e5NwnD<%@+Owj3rU^PH9m%s_>Ky>Ail}@ePC+bW1XPP03h$K5vL1-eR@hwq9
z##PFZ9ZPV)PW+5GxwQ6T_tISN%0f@38Vytu)>?fd=xi)5yMK}gfq-jDCg#eqmQy7G
zeH$NH@WI04NVA4xT1oQI@s3Q867h35i11{BWJxCx+o<>ptG~925-MPYRU|(yrzoos
z854CAf*~9kb56}jD<E~|)@c16p1-_Y$|B<T;nOL&ab#M<HV;0!#u6}ar4T>MUVGc^
zU?X%u_z?p0kZi(l2;?Wdb&Xhrm%h+iYQW<So@Qt*x4H4A9@G5l!7xRPu7e?FF<@te
z5D-o1<?orzvPF7kI|8<S^$<t;{|<m+8k-QtH2$`9-cBV4;e(So@vD6o+3XQw^!#Y&
zslRmq^D!Mwmt9z*V*%uySK~UcEoE<tse*!_{A@Bp2;T~Bu43v25YRTC%iqmHzCs<W
z`cj}#Hcp7p>O>L=TH#S@@&{Y7K43$dP!Ju@YvAogNcTqnDFce4J8SXFrhX_;N-6Ny
z5^yzf&MObzkQTI1%ak8!b3X^V^QD{9h2GR66AkxB%{iq7%B##$k#PX3`?Vw3?$;#3
zNg<4Rm5NTLJ!fmj%50!IHq7%cx6rDvjk$O6RRu_L%g0$oKLPvC397BQEh0gnRYYZI
z4pzlsTle~s56~kJm2-nYJzR<tgGKQV3ZAz5wOp%BK7<{xPRCWUJPRRJVM!V>>D$5f
zJ>Q?HG?d0_;9I#rL3XEK9>}H9(#g9oEhoKYj(=<(2N9XmK3$o06zRG4BX-IUY4V3U
z<~zOt`0(-aE43VxmU^R$o<~VQ)!Y-fm4DsJLiV;f8(=0UuP7agg8Jat$T<^>De>uO
zBh6sBxpW47nv6R+o)J%gdvUqDkK$fbnAX|K|IC7?%NY6@qyMTO22MJ-`L<d^hl=B6
zOvn>?ZHqS=N_+(mr-0>`!c~aI6?m`svt0_?THblAB|biuD$5R1BbSSLYH@ZJ+MYCX
zUim{@1|T>o`zC`>!@y8pQWo}Z;tTEX0H81jLCzzX98J;C4vmh0c<bCRewvcU{%3f$
zvd%0M;9GY_+c=rr0(1;tc#-C#qLU@HMrhdf@OsFrEY7gdGE|upUgT|2)v0~gHFEH;
zP))?<1t~B#o#kIvVWA%sh3@YSpR|DjGN>^lU@)45CvphGca~JXH=P^)$r_CMFZQ*4
zH2%QIJ@d7s>f+eM3dCKkN)p-`L2jC%;W*s>kaVbT)tZ5+N;o#H!S|w~D(N>9Up(*~
zv|W)aBH5!M%w+^s4rfR@v(2miF@8H<XLc0~!_P2zVVU*0VIfXB4nfwX!Ex{*-I2N-
z0CPHKfUYdp8l{yk%zk?Ni{FnSz7@?hLc$}kEZI=pc)aw{djST?F7G>$Ec=s}ww2~o
z2eG09nutL?wk>htK$dP3H4zD2?aiUNnIE9NMQVyS)Tt)qA^w@j>E*GJt$BWR*e8WY
zJ)cp2L^SVryeXpwl>;UufIPW~l?5gU(D_S`WlZ@mFt3F#`|n_c?|XTCdQp{MSP}>k
z4AM8tm|B5+hDDKEfBmc=;>!lezT)75Vxf(#^UI4Us3}nhBA6flku1dydj4%|Y4tw{
zr}eNcsfBX<js547_A5*jFPUKIjq%%R*<h*#n+-jo3AGfgkEdch$Xz}=!^YYSG`S97
z$NkxQQmO#Zk;H2f-HcS2f9DcdY-g-a&(AK7_h8T!!S_0$vYq2cnWMW!5T-73qZf)x
z;M-42_hS0J_C@`(7ZHW^xJaK;a&@`CDr$Yi`D;ozG{EF%0sgoZ7z&0ewbqu0|97|6
zASJ!YkJidbYjtTkU?JL%^LKR^0FP)_ACpJn&HeX>5yW+r-SdS==_mRUEuk91a~H}x
zO=ujN1}3yNJPLl~{UG>P`PK-6lXKxy`q`lbC_bukGA=gSZ&Q6ez1>S|1K>R*c#^?}
zJD3ANzt_JrSD)OqjOCVBPi_s-P07?0966(B-@fH}kU{zSS~3JTOV70xs3T<y1rd&v
z7JHY4z71i<i}ENXPhkkP(*K}^cc3mQVP>swsAXQD#*ptO8Ql4aqzk*VDsrnE(391z
zP?7S<d1GviyLm+}$?Szca5FizwlLhs#Ldb`**mzQAunTaMs_e1gQ>)F8#<QW>~Xz@
zpYSU{;9Q>Je2NW%z_ceDa2YsZ?bV^zc9NV9SN^Uw4yv1JsFwOBG{h7FL-sJIAX2-f
z-1Pf+evCu^$qbDJBmEi=t+^vE7_Ahr{3Pzjdx{-?4{j3Zp1au(&!~k71!eGkbTnYU
zGLhphQqyz(N4$YOtp&Cxl+Q-j7w%0qnl<yKuw7CU^#?Atc5p~qpwP&aqVMR3GFCDz
zkAYATolsm><I?z-Dr?+KwHlAbgCs<=qP{Z!j)-D?J8`QSW4jptA{il}h2K8X9GKLf
zcy}u<oXRN}2Ct~(sFW1G;WFGQ1t7y6(s_HY@~^(Pbz=G-9)1YITQE^XV*`BfC23XQ
zHVeTgGng%s?`CL)fE9cp{O<u|5@sZU#vOwa6X6@lkB^R#4|V_PSBjcnZBip%FScYt
z{wNOX0WmxgAPY*sLlxBXA2=bits!^_a$+Rj=Rzf2?DTl$1{ZkVYZeibrl+FE8>uO_
zf$9cjw2#$mmR+2<dFG~DxM*m6QMvjr1mXM;^ciVN{BF=x06Q(tP_0cg=fOar9v!QQ
zjz}NczHOd;-bQL2R~*LDE^cb~pv3qG{z+0%I7It~y9S{~j>o?DFs`E1rq)RY_S|P)
zJ(rB2A<aDdFQo{wXFpVK5wwZ`Dg`W@bTZ*ff$kB+Ca2GpoPe_lLx^`8e?BD~qJIh_
zBw~(yxR>3bz~8y2P+?nKZ8pd==`#2zJuyqr>l7H7>E4nShVBMdq^Nh}BNAD}P0lwQ
zv{B0Kj+yKq(LYVzJwEV*nEgdF{ZD%X1Gon2v1SXjQbKAj7}*;6-Sl{r_qFIR<>I1#
zD9;`mU&_)^3AS_Ex<Y=6xBhekBB}^Shq`3OP84hez#;kkTI&98(?9!Vl@?#6<j<fe
zQ-dO`gru4R$%wWJc)8{8-xe0;r2z_LB?B|ojeh#o)gn%V&vuSJ6aoRn!V%r0G=5(i
zK-fAr9VhYMv0X968EjAoEN)LwXwhCq#%OxcDpUmyIuO6<3^_ChxWvVcXDF1ESnTcb
zRcP>10gV+uLSplk0A2ouFim#NtE^cT*9X{mA1_fO5MNYOPMn$dRiTM-LnID9q26qw
zK%;p<veiu+cKy&e&wu8ik0YbuiNYZ&(&Uf=#S(&lcW_&FzuXu9%ASGar~r2B@{8ic
zi||+H=o#X-0HYy$J~`yV(qKs{(}QF+P8XL4-~k0CT0vprVc`hvcX|0x!fT{;|5mjt
z%|F=$BMtL(Yl=~kXeCs&K#|gVGhsK--K(fLA(@H|{2LskrZ8Z2G8V@xs>nDpr;!=W
zJJCj!;Z-Ka#z#LR6qS`38CI?&Tqz!sLv3y0-wod7FdKT*9JRfpp@bmRkRQzwaCNZ(
zTuil$tc-s<iLNBB@gU+<>#*CJx@dX1fnmYAk)|{<$P&hd1~{W|>5yA-!>3YAl$~}$
z;2{vSCKAZCThcy#GTzL+dNSJ&??5Si-<IC_!#>TyTw7zG`~Fv~x3{;C{401+jW8)H
zO~^falF7fIC+@+!UsrEteq4w(6(>)%0J-KuQ=~^+Ab-s)vA~<ySEZR`4gp&Qk)HoF
zhuAn1IRXM~JXBaXI%7!AWS%%Ug+YCdhGIBcf1HYwPiF8zJ5PexN*L!lIUZ9ya(D5J
zUStC;u`koo__Jn{`t26)D@n4l2e@M+!xQ67^m+(8mxK%S4Bv*b(sBOF-7W(PeG%jb
zpph?69UX4r5w3dWgD=2u?(#+@MYL36Gu{y&KUkO{kJienz0e^)yStX9^Hh1Ztwa8G
z1x?M;7MGOe$)EiG4?C0k1abx6Rg}&aPm&>gvK5<93<kq`VE%{3Yu|lgar>xe_VrDs
z`khe6>JnZdm>rVW4h4D*fITd%YK<<#-2wTGW-Hpw)*?4GDurExKmM*P1_7D~9^R2$
zcvOocWhbf3y4aR)Wx>r%=T~y`?&Ojzrj2ASg{B7Q1ZYX|0&Q;dMI)^ps(ej0-?Zx?
zjIlZWD@1kgQJ8z^)A`s0Ru+4{OJt_(dwyq7wRea*#FNUVN<5Fd1{8%3b^!>adl(qI
zU{(>o&&N3GzXe_H6ce^V;{Ao@hKy2Nq_J)ali6`xo)hiR{~}Ttv^$Odji{EZRfC(6
za65IfC)JSVYziUL!R?Us3C7>!<4OOfIb^x%j4NeB05Nr(Ye;lN38OhU&o%Y=$wA>-
z+A>x%;xiYpu1X;Ny@HotFly1uvxn(_ft^)0B-wgP2XK>ezdP3a?wUtDENCN?s5gfi
z)<FFK3u|u28@JuKfJRRj>&+HV+jBoLDZ&_1K9%NThtmOPx<gY5Pi=FVn<+|^kL{3d
z4}Og=7$SR0nQsZ*T#Ex5giSXCD6d8aN06vOU&?EF66;C;(lN5n`_zQ49G|nE7(<BZ
z;6=`EWc04sqy9T{w!<WP%}$H+7y12Fmy0!3)y6iD$?T4;Yb9<+i*Mfwb8%V2LiShg
z{`t+a&zSb<eZTumMAr?$EYG*L+Q&|d>++J7cQjipyvQC^uhMIqZZ)&^@lOe($AvAU
z8xTeXb6_6>Y|Q$r4dc|L2giC4xno4W<ZZ`&R4oW6?HKmWF`h~D$I4`;R>$U@OBNk6
zt=CmQi(Zaf3gFhN8qQuXBot3NWp*vuw>Cci_Fmrrf&Ys_cuu5m??ilGwnI<~vG*k%
z6wuV6rjfOGtJw-nf7(Q%VB&esh`f6GJ3}6wMn?qzgoNI-mM8cHV~ev)+`P1o1DW|I
zmR#y8YUh;#ZciGrqt}>~m0SI5ZU&DFGljQirj&thO9Um~5o)6HMp~IZQ{KPmR3Od$
z5n7QoGNK7;kJH{ugeJ65(D8ZM*)LzT-RSt-UnG>rMvPx;fQDsV9v9YnPd86xPJF!@
zI$C-H0{<k=eJAz&r3(MS;vK)Q$T+%v_GW~TF3D+wt{G(M%B9!-x!R`Al+AzC;T*;9
zz-7DM89)+Lb!(>B2s)jCb_jOJvF?t>W1>L5j215WQb_&qD*cKK;}*#`d)K|Rdw62|
zbu!WZS6JkqZ;>oSdFtu0g5C<#;dSBbuxNYM!1*ELgHD?dvA4tBv?r5m!<>3bwXO~`
z$CssXQEdgdPzQ41!L89x6LmPkF_C=|vHf!IT7%#V7jS9!(IygZ@ArdBz7E&Rb>OO|
zz(&_6vpjmpt_Uah5m$4J0mdycPSg+$ANwfGoF^Yb&N3TwA?HU@3^n^-OVmZyhLc)%
zU@?2ZvDwGx;fO?(nAp)SOK8)5+i3A0V%8xaC{;^Hx6RN;GB)a^-v@)J$h%)e5zb1;
zp5fWc%WH>a=Pu2R_*tmW8|=0UPl3afc@rG4+xiQK03XrKkF<WR=E9i$@kQyHHf`37
zNu$nm_CWwPy0IZ!aCO=FnZTdO`}S%c(C>0<wFj_8eOx3PNK@M$f3@G686q3zms2rC
z%h0hFeu-g~s4^MgD6&DvcLW)`=6srV*Pa!gP6(Trve1t!=me9ci}7g@Bpbne7ePf1
z5<IwPI2a(LU*YW=Qus}VER~#7T23ea`!ij6NpGf}nAj#r<2x%vPv!2hJ;DHZ60R@_
z{dL@B6@3@qHO$my4|shWR^FrK)~R*#NG$hr5-NX_dkxi3O^0XAc+^MXw@zeBY&k7m
z<iCyoM7sJ*@Fx<>Zdj%W12afMcp~U!t*oSfo{h8F-s%1ZF+5aSZZuu6-k@hk)nuRV
zk%WK2;{=K2_)OvuIqzd@yFeQdAg+|MbhXxOW2WZmpO+Jh*TZ`c|A3k}v6m%;lC5Fl
zo;ps$A<+#TJz9AgFrCz|55VgVcz?Z^amUX)Iv*L2&+iuCPTXb@L)S?KzjC`=U;j7U
z(}S*sHo?dk-2{Tvy{r&?GY%OU!@xP&a?P?I6k_D@wwP!yJpE=7-QWV~T^{q#E_mGu
zBU5m3yjq+-kn*ohIYD^Iwq>|fPAJ~gs~Cn=vh*~>9P!SEma=uBxDh<Z)bqhS3PEM-
z-~}AXi<9{Lrpo=QtCU6qd5TYvk*=+^y27`UXR1O}mRh{OKS`(6?lt~$Gjh_}Ctqnc
z;p}p+)8SK&8}M;8y*mgfdQ|Iy)TBCX<zDXKBUaW<#?V)(JLU8Wb7aCAaFP`LI6m~-
za|!fuc?#v}+|d5`FApO!iF%?wEuGxdwB0?zda>s5n9a;KJ|g*f5|F6IQD9DUS-z-Z
zMeT;u-BCBru66YwD`rzj)^qn1b43X{ZcZ#K4;`PGnuy`_%?!Yh#1WFN=nM*w<JD2u
z)$AmZ8NLa(UUgMjB;j?*fA@O4Z}KL@F%9N9t|bahJDL^z`9lRl236V2hvVkXAX0Vu
z-TkLxqfc;mBcz8erUa$2?-GQYw#M=j4{u*|ioSu}Z|>PLtsm{i{|I6O${|P(c8hr#
z^-fwG-Ve?<UQq#3@FT@+LGOK?Ajtm}6S#buDjt>|kUOiI!1)VHiAl7QKVuh3rdrX`
zqMX}^fSZyj)SnE030Cp;k1T4j{65nlN<*Putrw6_W9yK}^#o^Fc2LrpKOf184ptrQ
zi+XeGRND+Oswr0?KQu-+mJ^pm01FA`)Y<oI+Xobvn8iZ}Osu)z;^HuA)miPIG@f!=
zSFUwjZcP_042U(hHQRpEt^Iu+UpT77(q;<Mqo%(c%;Hu(xDahL(Q|DK=95M!6?y*w
z1u#~?(RV6#z;%JS9gHcno#l<4zTcYg|59L!cS4aU+)LzeLK?3)E4`np2|WTaf-^;8
zCjlnCIxpwr;BjrGaV=%nE4V!;V%*khy)OIHiFHp@62q|^0pMfjU%|%d`k^AJE%#sR
z{k1VTS2hj}+ru|rF1oxJ2rKeh1Qe%>5H<w1BMu%tagoIt0SZHl9vxl-ibGHu@|Tb@
z7rnECh-Zd1W&=vghXB0$uUEkhQ23bA9Dwp+ci@>j-RP)V^4NG*JO(C4QfzV(TU(7W
zkAG`}RbLT<cC(t5n3;#v{=_$@+QkbYx1*ttlk*JOl$5am>H-6s?Kfx%Ixe*Z5<cha
zfdbj_Hrd0X5Z<1l><UX3krgQ#rT*8YDdTj|dfSEIF6ogu-*GzI`Al^6ROb7L43Iv#
z(UzVPXbM*b3p|2EsAyZ_KMjTveA|r%t4=@?<dA3*OTi777kZz@&A(B4=SLkMd$a>@
z|3*`UKHu(ovQF2pZdU8_+$1xr8*wP{J~<edcLy=MxW6+{hsk54(FE3-&D?-v?Ry~Q
z{Eg2}(l)U2fQi0y;?L+gG3?KF0`|~wYNOH10(fOWM^+3|h-$wR3#qS1@p(f`CbYGk
z&kUR;mg}JFi;9f*_n0(VJ;Wrip&i7OcWZ-#3#3p;1yz8n?vK`$6>ko{ZJsnPHctl+
z$%DkwWspjMUZbVuWoWpHkEtqat;=LOlG_$1DFa?BWPGC+uIWh}N@GL=0N~w=v^Nja
zKjk)~Ek==h8VQJ8YYv>R5s+e_bN*$fIU+>hFzB|J67%f+DCe584oB2oPpK9Kcg9`6
zvg;D^jQ;ZZJYTL}4O&MWNKM*w4_8l{D~A-}9i{>}X&y<r%|Au9M0m_`6)*--l)hp@
z(>jY~(j(=D<GX;vgHQ$l_2w0vN@@`36Abh8mv2=wjiAQZM8pQc@jmtZw~jEOR-2&&
zLi<08vQvz#=DL%oMnr1D2W$N<FOiM4_@vY3^Nr0Zm$HuCw|^kN@y1_|mbUWSwdDt8
z>U08>^G)>*$5%-!sf=tiP-ZnIB4k@)X8sKEz4iLz#zQurRtsP;vG+sD)I9)xMv%4f
z(V>BTDi#-WizqBb__LS4^U^KaX72MiH28u%%Jk5WwzdrCT025ED;y&gT721=(7BQ;
zM_B5OpLa*SYTN&r<Uyx6nnxJLx63DB=X|vpl$qe}_V{p5my%xopLgRg0s1e-CUpJm
zwpiaU^cSQ<!@zgUo39C3_s<y6xz8r22QY7{xMtMUu%6nMAo;EGFC&o&&P0yc09Z8C
zo}&qPHg8h0+~fEk8a`($Op}*RIW{GwGz^-J_EO{9yJl@lA2&zGw>g{b)aU7KC3crq
zLK}7lwkGb&;XD4BV}DA>m~2V)KYhEmVrv?fM@Jr;9{~QB%Yutt4#M!_L3g%=L%Asq
zXf<UB%*g6Dz29K3GdjaPpBQVJ>yy`Sz413?E1+$;UeENf^34G@gQjL{snKC}>kE0*
zo_zG>-76|--|)xtyk+F&<t}l*Czh>#hu6c?cE^VHIqYc4Sm-@#06hLi(94LT<Z8LH
zOamDIBXH?gTqVy2N{z%r6hqq-r<3(iOw&wZnNv%`#N9C=<#D5rHL6_qH!4-NG9;lq
zjV&E5*Rp+oxP31_j9$EC)b9X~ehp*OEin=q>hgNOSahwPh7}vq?{c~@R(|i))n2UD
zs;ePm@^tu`J=Nzc=zNfWuPQaE<9V}fs6Knq2;8#vhU&0l9sj|CUgd2y9P^;UD?<#g
zmdb|q&WSrC?bw2imk?K|g3q6kNry3QVifF|me%BfmKa@3w3e({Tp+dPjW!6m7(-U6
z%6mKK$+Y$*hJ3q7@YVPJ99>-$5&~Y1dOG-sLz);SWv?2)#!s9TbS@;N=j#rA%OhtB
z0;zTys9>ZK;h~Ws9r5axSEjf4atFKR6jjxrh&ql*+Cl<?I$J@Yok2K8bZBDP=G>=>
z#3KV+YP!47)Yrl>U47T0$7Xd@TOEGq3Y){Ca9s{q!sr~qMeqHAb5CD>pZpyDcHr?L
z3Gtkv#G-UbIL4&dR1xy=x;?EPr*gFx`0NkTo)qN&JN2W*%QT;<eK2f=msWZzA`>4y
z?X(nq(th9?cW1OKchi>r4gO*e6%{{P&ntFJY;bj^XPt>1XP?axhfy~myixb4fiy^=
zfB6=-ty-t}=E00<O9plgzW?$y;5|iuqvPWln)l}MGMlgN*M7wwn6?bF+uc+(U`4$9
zkXaL0QJ2pTqZj&5{};I8X%c=$x~7Ka>Iz#Ho;qP27$1>x(Ou<w0A{Mh_a#qzl+aNl
zECdE`jGN?h?iKgX=EWG>$4mhq<IU)@^!JclC!S5`=NATH5`ihEY$5NrtDcV2l!+!z
zqLPiy+6t%fk$qWQY$gp26&-&Ipp{Z~@v_8{-|_h(0`+l{v`T}WVFbFtX1|cE%cS9%
z<{u?7W<?@2QvrD339Bi6zq*&0H;*g*!#L?q^ygg7G?76kV{zn#o0MYg>gWU=jaI8^
z--50u2%rIlT*7w9%M`CoLU|FB199<`Tx|b|=SlDg`ZL_`Z=Nxo_V@?^m#XXpoHH=^
zCT5|-R6~6pG7^3VBHV{ARexk|rSQ>vd+JcbZND48mcquYfQ0&p@{G0+>eGD$(ZmJ4
zxa0cz9e*8$hW$Vu39S2buL8E)6s++q9tLt+Dta;&kjmYgx|Q_(;ox}P5?6tXw5(LX
z?g*|*pBJ>}5i|REF3$n<JKa9cjuQq^L4Z||>d5CZTAq;0W~s*K{iUvsH68UwW7oCH
zVpNC_@ORPPZuaetj1XI+x$l^kma6uNqiQ#aMJXr+@}V0Ry$q4jwDW+D;EM0)6Ux<%
zG?HkLJkUI%A57qv<cGfmoeY<?7%tJD%h)jy$AcAlh*!)zk5Awk+wYwi=E+;=8WEf%
znqTz8=$Gp8S|qcXH*!1C&;zb7^#IlgLR|wb|Ldie?^tCnjCP#HAB68B8=W5&af^SY
z%MP-ql@F>JCgkBA*%o8>66R{`P;>Y#uR*A_WDt(%P+@dTzqR&dY57zU@O*vism#7D
zPA@(7aU8u)FawyJ^t-%2hh<f&1c8sYWwyKD!nO{Z6XW~Z&DK&T?-v4|j|+2W<&~A?
zd<dF7>&4YVYXZvJoYd2!-+#);$Y^OvN~$Hx*u&QIx{g(Mvx9R!8v+5JmYrX5mLC+&
zlbsw%%UR6vG`PXe(ppN~G(#|3`=V$9LL{volPtN=P5{?Sp>9=P!3(T2s4c~A>aK@V
zPbccmK)&mrx7l^i%3s%`oK;NzLA4B`e992u6@ZuJ(}bvQza~05F?Y5jd?VmZMUo7R
z2E|%<huW_VecH*K@CKVqFqJR28>?B=B;@(Wws(39??mp@m1V6qS?M%;Jn!tx?|VjP
zw$8oobOE-dFe<XLj>^j9i5CoT&!b!(q+h3`%XjwsGS-Qy4Tx%^(xthY8;FE_cEl|$
zPx0@h{>ztq5W6hxD0eK#WcMqlqJ@N#2-g2Mq#>rBhRhqLVyk5EwSN)(Gbbkc-7Iw;
zvLtUpqf=TuM2P>3x1`8$8S+$+1XE*z8Sk}p2p~vbwemeF+K~G3naO(>a|w9A7g)Pz
z$`<fOf|gfuiA+fols9<e-P)GH!b(z8`6Rk>n$5<DYH+c$AG97E!V8v|q%_;f()pp~
zc+$O47;9A)QI_5!WjwB}>;5$5w~~IMo3uLOO8T20S^W;!oJKC0wWbE897GfGx;5+@
z15%4g+K(Fg{R#hhQ=`wTzg1UMRSE@TbjzkNIu{%rAGf%iUi1|+ZMJ{f6<a^vlzveQ
z&ukA=D=w_gqY(%rrD75C?nTXYPX#oayo7U}3WP*e?TP$<mX+kgM@nZT>o*v^HwoM$
zk5emTowqwdXE~P4m5LINIH~+eP=93tkXpMwUSx$te0GTiZzi*KY%14vSjjgQ<G=lu
ze1qP)zTv%U1b^~rh`@&-w~_F@42=jug;N?-(xPfF`0OjU!sPXKS6w}?hiy)ji?Ult
znW?*1%xr1!`YcVbrjF*~BDUn7ad`2XtbCo^czYBu7uG#K;rBcdpPd$awYLF)&Q}j6
zUWlU&&-a6MCZ-Ee8CP~}aKBKt(l=hr$F%7Yx`U1~<on+kISpB*{)SupN+27F7t-9>
zfW+YIM+3D_sh3K?>jh=06j+yumE!kA*Ii%p*|%BbPPi+v2TCx8{Ka;QkdkauDY&$T
zBA}tuAefh9J(FbEu>id5-T|<DFY}MKo9z&L)!A$MofJ&8-3uz}5~diWuc4D4#sE}%
zMO{%kD@<n$G8ks!6f(63b#6Jhnx?LRcR;VOZ>Cb-_M%<b;CR}`!^%DMu%PJ;4CD6E
z$>yy8#BWzF-YQ)hN6@`r-!5K$zK*W`*GFo1eI3|GG+RxNAu3r0RF#3V{xn2&8xh^x
zoB52#u8+t00_}A(t&>L;l|`;**LAn68!olCl>QKHhE*J@7E8nrGj6^@muqvx82GbP
zUL`pKx^6G%r7qY<$1bT_A<3)Iay*`wf5r#6)hiA4C<h2r`67icr_FPgZ{(4{EB8xJ
z_Py-obR6tnb&epQQETO5!JIhhkPksb-bLi5la3^{3A1ONZnr3_B-Lq4jB4fo#fc9B
zygJX;0@)$&LnIG<I`M$H^(2okp~LC$n(!R9i9Q|Y+%Vr@k07qm7YucZ{Hl8G@9oww
zA{4**Ea>+hdQ1k@Q)?5Pz=vVPugj|2Y<S<D&A*~n=iC8$=i*X6lRdwzK^94$yLc}V
zUoK2MH?B(f#LlRo1k^K&p|G2nUvNy!76Wxr%)j^gcAvZKcRneGp1kS#RM?vYm5xD$
z#BzK6!|&6UQP=MgFak4U#~<f0J24ad!zZQwk4leWE8`Kj@esAZB85+Moc_ekVEw+*
z(Mx~@3OJyM?Ig!}m(ytlA@`g^_O0L#{49*z-6JZJ>F{Vw3d>DsPK$$?@^}#?TS7oq
zBo^&f#&rgMjyu$wH^kd3SG69$_uJNn^;oHeHLu4VTzDT%x?<L4OPC%k=>~rrP*2$E
zI;aJTJM}o6M*NGmeR_iGrnSOKbT>OUokluZQ3$}z9t?Ov2m@&p-%zf-eM#A|5-^^i
zOd1&kdX*g`WJX=j`Mikc$y6d>I&wEYZ``*&(OiBm&zgAIn*F+e$v?jK;iTd{DjVzu
zXcq~aYnHg(YiGD9P01OC394(Cpo}}b`J2gXMzogNTG~9_+&tXf3wTNH!j*_88H>z`
zV*z%@<pmCj7ISpHrWct38!#_t$3;rs7ckUGhBE1kjZPo@uA>Y>6w;D)<;H^i7X#&2
zkgHPP+js19rO>fBeRV?bsAc-~h{wkC<HoY9o2@(<-3KcHw_LN2D!%UP(8YJ@sc-VV
z07Lobt?=dxsO2?`rM5mtNn`%<HjmwQ3Pd+F=>K&fv~Z@G-(>mwi28PDAT`{rU8V{g
zF5Ex#C0QaC4?Jn$|MKUD`9O>tgs~DnhsG3=M<xRqVux#r=;a(U&{+pPZ>RnP3y{we
z419@47C(a6DP~AbO2WH<^d`o99IAjmEz4l^L3p#B^7R*ldV%Ar{&K{(l8&+sbiVC4
zuRai4ULJd2gR_3FF1GTq4BtI;NKN__;dOT1L%xAddYwUCg1@E1pw2j1I0CzKeog;S
z70t--_Z<%3>gwGtv`26;$*5ibJEh^N4JMZnQ2Ym(v-#zj$iG8no$@a4K@US<oifG_
zczqZr^oX}FY;y+wjrd1?SEL7!zsYfuccQt`2l<o!c#0NA*3y(#63ISxW3GyN>c-p>
zaoGT+S@(^Dx_S#MGdlW^FxxZ{MUy&4PNCI=ZpdY+Mq8jB^-+ERG=E}2wUPkBqJW1-
z^+TpL&tK2cbO}e_BBfcUi-B3x*k_@m4tHqNM&|Z_FEEmwLR++u8FzqJetK0kf@>;?
zT?fBT7V%jH!<ZbBGkc=QS@vZ)pR0c4t76|k<RNZcqI#cz+;4RbFaV(rCv$L@E<$=v
zFz6L?>6OzF>~XPLPXKrWvn9y1T?^PKH8;F4g@XVcJ;6OqvVaL8ydeu4(oq)wDT5y>
zFD;{~&)iJSV~u5DmIYMc_Il;4b%f=RpsJC~|Mlk4RO6s2gOzGfFC{{(M5Ia0VL`ps
zPDG%Ay_rD0{es}9C<YTa9(~;CwA&FjF&`eg8;bPD#)ehU^LYb3zv7sS$N1LA9`dz`
zVBTt>tw9A9b|(ax5!$qbEF$=P9`Uq|acUY?dhTvq_RF&|WWXRyCid0R^Ikuk?`0(<
zW*0pk3|<IS#%1o*>4qrwIR~|&RNKWA+46o$bpvldi)%uvPmzsp6-hTboEu+72am=0
z7O@X0pxeOYwfw<83n>6SIB<H_qCb*2oAdMYn&Z{r74uCjZeh6KOw}kmeVy5iC<~@F
zg=4?dS>C-E0Jp#`eAY+t?i4j*K47~S8sdHb6zdAk-S>l-QCeTv^21^EqjJ<#5ycUH
zG_~2qRe3Q0l>$}RHbBaWdH=PzJHRnc%J`t2Lt|oYwFk@WYbWUJG0l7!9;BS~^U+ga
zCx_f=+Y+>+2=S%iB{+BhIl(}fWWFg6_;oc`Qhp!cA^v#!PPxw;#|6Q|L18N89*?qr
z9d{y|S5Ljx!pFtSHzBSmJo7Ia>gy2G=0$gMME5ha7^*x2@6;w+9IhwFtfL&g=MZ>z
zpWQ<5#yV-+NH>j&(^nxbzA0^dZFBk<*mY&Z$=^_KRl$x~Sx_Eo#4)iUlM%MeH#D2v
zs?FxWb|M_Ez~RtY{EgV*g}1>v0%*(RdGj>g_v#e3iF$RGPoU-l-x4vph&7oVkG5kD
z+aHo!@~2|0rQWgFL|!2}1{Ar<x~PhENc!k}`&13)v(>GnBU)0%UhvRmk<OPF?unqf
z*7lggZvFz<DYSI`8Q=T<Tp!Xg^(IVdcV7gUII{loI!6W_4iB%K;*!VAYX5c!2>;>|
zH8njyKdU?#U2ybZnyz24-R&4j-M#CVfa?Mysy-%V0r@wG*@+#tIyM!ApDo0N`S?BP
zuF(s&_X_s;dSqR#o04;6PGP=<F4Wd=aF>NTX4?ZtpvwGsi4i~1_ThUNJJp0Lozn(1
zd5oUqrx}o1LC4enmG7)D7U`i!6G0wqFzB4##B)4J*u61BiV4jus;!hl)46Ml$?mv*
zqntR}0A`X;ai4^IL36m&{b}|_mv>|Nu8M45&xkW6Tb>t$J%rMhfDg~rQ$J!@Xt-=1
zW?6-(i{{*M&j<5bkpHd%6z^X*w`U-ba!=K{dueodp;tp4vK;SbHqhmym9S@mn!fOl
z`EU4`_nRAU2HZBZgQGG1qjsm~l}bIQ>)IQ~lcUd<(e>z;O3K-MoqOG_{zca{6FnG;
zm=5n<3sgIt3jy&*s!ni{wXTTl<hmB=4XQdKqm$$f``Pe?OIrO#W-btTum^y%q5=hr
zA2QHp3z;3<(H_$_bj?4awG#*<n+?E!FEN;zo@=2SdE{{2pi*xB@VoS!xE_bqe9c&-
z(PO#3ofPU4zutlVQoPpg(|Ed8x{G}O>SD{pkd}qNdubJ1Df~1TLwvp8z3a+f5*4r#
zUQt0rqfSEu7J@kMSP+72Bn>!CVayxs{XvmbNIuBbF3UqmBNNJkkH_$hQWu9wFV>3#
z<)mwMonDn@XJ5#=tw+g`+irA@wl)lplIR|HWDS-r;3$NNCV`#F==}}3_2G{a{`V0@
z2O2@(#F%WjTeN?@j(c0!iczg}Iq7J-6|_ay*95F?&L}e;lB+X?1OVFp@t>pp3*$np
z<|AX!j%J6(zXCLc%p8fIrGX@`NB6B#t5C=MilVf7;BHTo^7PxOcAJMjs3q17tO11|
z852w3CaE$)zjlr>??Qz<p0<3LRP2t(Ax|3HT&*;_THVytywE(7lUP>cA?G*sk@z_+
z$X|7imLg=ofB!Bc1IQ?)ixh+?Yv07qq3=_QknN7pm75>q|K$D0yMi>D{BcIk9rQ=l
zPlY&vrTmYRr|xab>0;HLldCTGCOBOTZd<D#Fy%>_^~%*%qqo<A8*^E$p_)pZ5*42x
zV7-pI@yh|f?7_oU@1Nz&AYsNe{i`Uh%>xG)Q5iXNNUQd-2GANpA)gcTt>*^&wrJp)
z56@4k(bK?bWNx{|ZcWK2MDK0j&+t=+9jPElE-d{YvC?F~AH=t3&%9`nDmufudUrNB
zj?6;)0XK!JdA;?~vBB4DyWVcU|MO6j9;*iRhkkv#v5$=;(aU$L^dI>*jw0Fj|AvXH
zcDGRuccpUkfya)Tnwn5JS;a7$)gL7Vn>azG&HnLatyF5%T&$cX=ovo#lBgrFZvT?(
z1%#X@c4or8k&7AEgLykndt2&@S0^)hR;TSRh|0fAV%>GD7ZY~>GEf^{v2K(nUL@)K
zRrO@U+FM=?0^R<}D=oY?ehnKC)4={UzT@t$nOk9g2@of<H-e=L?T;d|l6O=z6#j_b
z-@+zJ;PjS}v@o(#(O{+z{G2GcSp`=#2X3WBM;X6(Zqy<3JvL{yz4*!;pkH}!ZgidY
z_09#&aoROquAiIyJJ$wL*n%7#FPtoAx8FTotyguLJT!;jc-*&o!6n;A`s)d`DnW0;
z#ab;oAkZinG19)4$OlitWp^E4EQ<H1yuaUYTv>)-0j<_FBnw}o)$rS{a9IyEMX~iU
zdUAZd-F`6fE5QnH^|mF#`q(+OVpC&d6DSDN?{Nc^C3^kPXUm=h7kr8A+%~e9d?~o(
zbTCsnXlo!q@U)v{rqD#h*8tegKY%199dPeJA2RI2*^*9cyBj0#*itpwR|=b(JAa9C
zuJZ~8{bV7Yxp{H-{D$!Cmj5z4JY<Y?RNU-*QAewuRv{HR6(tQ{(p*&}KW6IAIiwWz
z+Mh!Q-ccPmF|djc(0!w09-Vn@az}4wBsAE#?JGK)ELMwVup@9MGNOLY@BL-gS7!zc
zW6{xfiZRZZV0<XzNdC;-k>Tz0(Zh2S$g;r0Az)7Z>!<Ah*Dr40Mwl2Z615b7&h=_O
zh1Q*wrL6LAi3iW5@${PuDYwhbTN2-tMK43h!Y%`IGH4wBm>qUpcf%0FK8+#W$m8s_
zI3Zl+Wz_9@-Me=z(ePw+%IqDX=A#2RD_3jGW&b^7sd`#`6G5S{l|iu5q@pVPQ&gy!
zApl&TAcLypk(3D;_=d%VZ9nlo1f;!RVG=ETW9NJJuFrg6iXAxApf(H?)dY=}dhOwl
z(dBGDMZvGiwMPj^2Txrav4S`{CvbO`>Wq&+?4MahJ0R;-qvy^yn(W7C@_7KVp&UFy
zbP)EJ{OJ2O53lKjm;E7%;$clZ1C8R0jU)XoPiprqeD0beQjU|;oQNXYTz(7Gzy0~1
zUWO@RCfVHA1eW5sxu)x=96gmL&(G;yH~#CL;rG?WHl=g-_T#1LKG?^_g)^QdIK}>i
zMbfKO6h&eiJ+bGN=lWtkMkheOEocL>d)iOI+md|E4m(3eol&#N=zgURIb(!wfgvL&
zB@ON1bnlpyAPdTfMMBnhuE~<g4im$>l3kqTad>EDK4-oKvezH(^6w=Wf58mDH1;1T
z1T_`V6`f9Y8a*dBYSV4$mjF_B<?PM{Ceh7Zhzsvbp;ez7amU?hF@S@wIgwF2RR~g*
zy2j#c+sHOdUPm0d<=SQ`)KK4)Bq(45&EOnzCI%wgB<2f&&Q~L_7WM<y-XoOqov0j|
zN(<l41scE%kvK@2zu@;ba@4NOJXsCX?Uk}$|G~hj9tFKVUY-j!2nBE_8aOzJ1T~sn
zA1BVWr=5LTPY?QX03qf8cdCn@6|*TZmm5dN;_7QX-m0sd)!Hq_hCS`j5|nuF5{`8V
zecnXl4K;&Xb8i@5(V$ko??*W9@XIGrAwf!DC$6-?`m>Vzudnn9Cwx8hLfsnnRMMFp
zq4x2(4i>nVr=HaByubQ4I3~@n_v^uEWBL;wvKjX3NWWt(3W$Dsj%K65^*=fiuGzJ%
zG#wn6k2p9mrZB3MlG4^Hz&~NKgCh152>2+9epfUm;tHlTrbKKc`S!3^XfVLyvKI>;
z+l#;A5k$$Ri*72Ux7<r@cyL23#n)E7aRzO%##9ZeOx&L#dathhyyug!wK)?(ofxOL
zxA|8lX)%%sfbZK>&K7o={xQ+x_BkEys$%Y=J7uADvH-cOsh}iC9;lNW5!A%rs!0r)
zkgNW)!;I-W6%n^a=WmaOazCU+K+^7K20KX=vqBy7FKbWlTLbomKTc<m%D3uPLUUdx
zI*(CpiasAS5x?xHx7a>6Cnrm3DNDUq6dM~GvjVyQ0R-x{I+}?&Ss27e7seouZV6)W
z7)JhIZ%&W7?^^H&wh%GO*guf^5Z{07@A9J3jE(gt&*a>}KRUC|CWnl(4@3!1&(Sg3
zE<cbTnr{K#?{9{e&Q6wn1Qh;0i!Q;=&)e_iW~iRFddIwr@O=5|RqJ5hIc=#v9)}~(
z34;&nfVD(O$L|aJnsjLyY1zrlcmV-d2f5sgq|Q31g`X(0;|xXmpYt52xW++z$I?-Q
zj3G7By9Q<;^rMk+ZgJ%|Db?&vewZ0nW;bFyo3^HU-0HQ~WgDdP|J4$F-pya4@XhDu
z`7^HNOi51-FYJy-2@6Ol>FWbWXd48;x>~3G`iVq?MNMI89LWL#7B09nl+$bUljPro
z-bm5x2w2tUuVj*l4h?eKzFf22BN2^hlEYseu7~j32KVSoxt*#`N1`@h!QKe(lDmR7
z7R{zkPnOV~Y*m#tH!Tgc*IHKQrxybA@$(CjH*SvN-0j{)SEtj@O-2!JZ;MQTBC3Hv
z*q}fOUmrCi)`AR}1Ou9k+N$o>DD9^0eGE@)cm*Q?is~X*!2u5gvC6O1w5;1q^te7E
z%*|QxrJgD-(=Q!^O3M!C6MbgYuO!ZqyyfJNY<CuR$gYnE^RkDl+8Jm&ua=K$>gQ{<
z1$7wxQZD&y{$a-wxmLJ@;Wuu8vX}savk5f5^p;pcohNst@rNtN8STJUXk27eH<Xn&
zRQRBi8OVIr%r;v%!XqgmveM1ru#$fhV8-;I(*<gBjW&06N+jW*b&1k$x)2!xCFmyi
zT+Cev)pHF*Rm*K=uJct{d`-6HWfDBVV4>xYrA1$Vs>JO$5Hcd0(FP!CP^?Ly9GUt@
zZHF*iLb&W6+>iM$7u%8x>~d~XOF1*(C`BwplaQUoU^bYB$!DD8Ic0#19-)VtCf{_M
zIHlgraeX*CBE=9+(HD@S!NPMcgN=BU%ohAYNi<?Ik@=fnMZ2Ti;T1HKawb+XFfiyg
z+vsX)E~jJ5h9e%1WdbbNe>{6aq@<MPHAU29yfF521Wot2`m#8ph^3Ekbu=7^Ql62(
zQyK|n=@6#V7(xQ8l%M8y<X=}oBfQ6AOgpi=$CK!#>WDa4&#)3Jfnff+nq<A*sr9(~
z;zs^-u~M_eN_x0=-vrz9MEeKj=JL;G@A2Nuxf(Kn6jl8y5vZ$aX!7e2Fe9{p5r|gA
zD!%YRzb+-Kt#LD1?lrIde>9zCP#j&jZgD5LYl6GGyIXJw?h@Rs@!-LNySuwXkl;>m
zx8Q?2cfND(`9D+DyQ-^u?|z?WtxepMg0l&B+M%vQY&<g!{T^>)JYiMkr3R2nMzf#(
zTRO7Paid58D4YwREK`r)t2s-OLA%@v!&Txr!_Id>^@q39uGZ;jGdEFFb9Q!?lN<dq
z=Oq1Lpc9xHUEQ9a5(N|p4pHn<aet-g$+m(m{1Ir;u{_)`4=YSVL-904o6i=7e=Gkb
zvn<75b}1OOxpc~?!{?L~>b><L?JEdk8edidY7+`A7wkLOIp}5}nS}RnkZfHMbSt(}
zQXXU0`Ks04+rFpwvSV<HJ0*)EiP#PnEc?jV&Dj2t=C>`23j)OE9uW%~Dmuuc@rbjt
zehIXCS%^u8c9usW2rw;>Lw)O%OLKbq$9~)z)t^=`qNb=fjbSg~!Ci-`Z##3FA(nmi
zL(C4hKaRf5PTOZH+g`BEYNVvO4qs0vUWCDOIEH|auf4t9+k1YN2a|{ZbNl?z_`79Z
zbqNHa^MH7aHYCsjLG`Q;ka?z5u~(1rgtOt7E^>@=Fw=@7*5#Q=TEP`ca@vQ?A&=Xx
z@^{p%cRf4~t!)M5>tVNn{7w&J?X(@&+dmK<PK?(7*$QYhJ3p<b&$FxVehKzl7f*7H
zl4m$w^U<?w&GQwQ5ldy`f>{ZuGJ%caekK2aR3ZEGDG=cEvi5t936}C6?VpmCAK24*
z4MdbwPRJ~_`Iro0<K}Qdo&Hu=)A2O7nCWRu2ud+k?W65xIr(XF?b7>jkNRg{K!uC6
zsz#@`z-y|ii(tOQTv=I3US3&QPEJWlNg*Z@;oeM#1!-<~VA(xdmJ`drp?_-tj0x9!
zv|yz=9|00?iZ+CVe{q<HbBb4Jz1z4dJT&DD5jh}u*}jCxTaXHSL7bqmT)ADWOu8y@
z>1$2=uE`^viEaSJW@>fybo?_lTU_-{k|4gLq^grkpMin2pMl&kvz@eq31~)A1ZzF!
zSzFxE;<B^UdAX*x42{IB83XRNUz0;Z{E26sDS*Wbp2vUyGKs6N<!x1o=`wB}qb2^;
z+IfQh&%29j!V3iK9=`I^YsT9id$uatRx-Y>{#>_3mdcPn&)Sav)NRMd&&;UKl)(?N
z3MgPfXU)h324V#&PXPby-}QCS*~OR;0;*=G*Xcg;W!MX!WohQ=*P-Nbd+1!3G{_t;
z&wzf^ComUltrkTtdEHy&cOB*-NTn81!K)waH{R8wW)mBoU|{y7qeuFr4JPrOThc;f
z+DitZF_UhOtL2O!NcQ1cZ%Jm;(K4GY?aQ(dk?4wXpNJI|$wo#_Ya>4cHt}x~bg*0s
zxsO@J`e-d%XWc*kT5i@mjnuls-M>6s073Ce4@0h2%FU#koz?a>Cd&NIP(j><m6b#$
zb)C!lK@z!L$-KXY9@v%%=8ozBiscVI)y|zRq!Po(Mr}>5d4g{uenkAjf<5bc@Gs<$
zs=1>Gu0IL;3XKml`CF^*6*s~MIKh46O<~`t+$zR7n27wWfop!Pi_eksm>~vC0B14m
z@av4qry`08o+eb;oMxAi1Xod{aDnp;847LDH#Ux-PH4)IPxw54H63&e(~@wNnp4!|
z%`6=pOiXOR-ELcmvS_5leu6HWNFHOd#p+&&CwxquQMY}B6k3jPU}xG9-Ii9yoO)QC
zG%&Gqdjj_KA)E4~vm1?6Bp><|;8i;fmK^@)@E5dtec;MY0h-b9`SMrxGY)d@pe}<4
zr$$Xy_P$;D7s;v#nZms(p9O+2F})x5S%F0r`PEjkrV&44A@5sF@x}awVa*P?j@L@=
zB}SaCqh;yJPWi%%CWF>5E`E=M+KfP`?7bg<jupre%n{mH^|Rl)Zz#U81jx6`4;!dj
zgt%o)Op1_Uh>}Hq9u?^|Qbv;9whB5^qw}x@2$P~Wn?Jd+l_W`KKQ#jIjp4!=m{!TM
zAci!rZ8&B42of`YUt(=FQL`Gn%2o{jsnThO9LJhUoo-h62XmAK^U!w6rmFmn?tshi
zL;{fTQU+$|Eo|cs;_aXpfDMu)8RHEzR^V3xt5}WE^vzHSCR*~ikaTi`l_kjEG5Y}Q
zytALt`l(kj-gcFQEbD;e`6MetoWIcp%!B@47SC)EXvGjhxB2PpoTxyWr6aAwLrV?(
zlQM>#IKg16v$QSf5A91YT?7JEbPSyS0L!i`V~j`L_TNVdR)g$7GWPy~xv{$38Z+(p
zi;Rkr8s;>ftnVVRnN#TCbbH|0xF(E9@qqv3^-4(xiG>1EgL*+-Y+e=Xap|BX4&zos
z7dgZ5h?WZJxAZ~LQ-TMkK2wC#ngNQjwt`QDlT@_%MHcpZwq}Yp*Q^@|hw7{(auK}<
z&L3<sv2DZ|Z8Azgx)FFBau>)5*#*13gY)xdm1e(9y=@DQ?*8VyuVSV1`vxY8V8_X*
zaX)b$4;NR^Lnl;=l6jz2_MhI(`q`iv@u5Ua#z{6@hdlxo{;WZO)N9piIL#r3KGSDd
zkzqw+%>#$@xFyBx2O4pn&oucNIDkCy6n#txF&||2SL_9_{g_g!l)q-X=9VUG$N;yI
z)ol1~w3+|NL7PEy$=UEpS#Y6#7phXCr(lzd$W1M`Iv#PQWEI5WLYY84o+J&&`K{B!
zs(}&F8-7Iw13Fw_t-FfL)@m|w954RCYG@Amk;rU)#ig>N%r-WI*Hl{ICI<<*`FnDu
ze`*->vtu}5%wzdf>AX8V4)H<bV+UYzqif-o48wJ-655U-xghvqgbI<N@jb<GELa8W
z<~e&b{l-@Gzs(spYTlC_$+k=7UnN94U(>8_E+{ZvYx%y`dbz6ZdnLOE&Q{c22z4P&
z<ybHGcACD8CGySBnL=%RSK@T0U}ocJW#ysxj3NNYSJ?d{1Pkcb_*T=1Oe|TP<4e|Y
z<$Of4Vw2E{wdFHJF@`Z02eT{}tnL3qRPtJe!-4*yzZbHY5ZAIUr>#9w?dREENKO})
zPJfWO8j+-GE`<q9#hCHP|M*|fHIl3xL100kd^v4?1XKY&K5cLw0&Wx}Jw1ICRFnY+
zaB{n0cj-D6&7r&wdJA}{>FvRn;XTdj))s`cB#rlqA`lh!I!Ti`dYzF#%5rJBku-Z-
zN?as}_i}982b2F*4<{94V^L6>7Qro;mv8)_r1$v8kJ}p`i9P{K=t4GwW3YmEP;@RO
zbppXs)zSgL8Vn!k7ht4}UntTNu1^Psny;47HZFubcdv3zT(EI)^8A!w0^>Y$P7Au>
zm~(`@&zvOul@7ue(?@5;Nbik35Ag`I<5Lm{$#BljP3<bIB%hMteL693gRp%>!XEpf
zA|EI7Q(Xs=(}};m-VEs+x}fj|*B6<gL)+6Rnd-u63c0@Y4W7kz>UCW1qLBh+F1(<v
zUJ6(5#0QMWb+t{l0L+@l7KzRlhd&aLXhr64=T%_xkdjgiD(Y>M29DZynm!3(Nz((4
zolcO^@{p7m1T*6PKHWfOFr(R+FdF-ZN+C0AB9F&ath`)Qbhu_zs=k-4;n3z@D*TpU
zhwH(@tk9bL^>v-mMn<#U-|xWc*KXv6L$Na?U!|;}=+!r04=wQ4ipAH;)A7SN{+uRB
zUPd3;$q`~+Z>(Zd6l>lvte+w&<MU6Nn<755@0fq)tPhD(T;VtK^SjKdFFtPj-s!0)
z4{?6|Wo>f#q%=jhvEc*SWRM>|k(mh0*D;-z+mu@+6ZT>atPyt`YHbBbVOnUF1uHIQ
zW+TT0z&RK{Z`*Af5)mGQrB^o4$NACF*M(?I(?a8iFw**BM1H;G1B)v>D%Rn;JICOH
z@{$8{D`q*lm{=s}L(|C}4jxi}FP{gOm!OlxEKc-pWfh|?uhDC;0eW(=6C64DF1+OP
zD(Nd@CG!GWN-2~fL<25XJDkt27nW7Q$xM+C2-a$(Pqz;sD6N`K6=N?qzp=!srpG^`
zDl!TZ@uvyY6->ro(Z-}v4y>v)k({SP5Z|!T_ux$2ooaBy9vxNASWMZg?}!pw2o0ZK
zYUw^#vZ17<UiC|CmN$t67LXc%vT5Q-%GMfklm#UuhHo8A=K!AEpV82r{yWjJgn#`<
zm#OiBSD`}_YsYm|u#n9nu=+bi=t<uBET^=&Jx~=!={8iGEH#%fZ)^23dMz$N3eYS8
z{%lV5?w;O;dN9sk^y<j@*@PJIG=v_$+TJo%^XA<{t5i2TRsYBfeR;~`nEw46OM&GP
zMrx3<#H#9>dKQqGK#T>|(FQI_M8UvGE5-D_@0lMl;If8dfP$MR1r5b8Ab!cdx6<6>
zv879hiWAHTBd6>;+E`x__(>q9z1?CZd69o$pfkz6Y?D`<N5m4AJE-lxfvC_H6MwbS
z8OvA;$S`-T1WSjfoW5P|b)ZBHm4mS+2_|CiU~25rGT=@^QnZV&jyOrdR#~#fe<*A5
zDTDyZrK3d{dG5xgp)rL?*&w-&wb1Fb2P^fm`}J<o*zqVoBT<u%j;_4?F>`v1iMD*X
zrxDU6pz7S~Wprw4N*%pag`-voe+dpA_VK0QsviRz16kp3Fs2vVfJqbbdy49&1de?C
zHgGH`1{6Zz(vUm-soIHLPY+~WD^?v=>gV6ks(E)#ms<PeaVzYnz&R1{8j0;$l<lj0
zjNvp=I2~Xe(K0WvvQTF>{=7IozZ((c5eQ}uXC`&W!Ww_?6D7&wvi$kov}Zv?81YMh
ziJOL(wQ6APk47c}Y`J-ww{Jd*Q;Y{`-x)(JKLeg!Tfqwp$@@}rkzn;ox|AmBxwh)=
z7RpxFObg(Pn8mQ4!Lr%r{W8`4BjfWysd8>VwGaBn#s-PZgBG-hhlxFkGsEzb=6{5c
z<v)I-QC$T}uneh`l5lx*WbUm|bF52gKY(GrBIx{6RrY%5ekT*~c9gXiefSV3Mvp8L
z`1$O-S0WyU+@i3|e!kHBISLgm@+0Bb6W&7nJIHJz-FW>*nUibb`-VnZ2#;$d7^!kR
z05if&gpB3CFeMj-;FLh{wp{S8aCnn|zT?ye_xLn0oT?ZOVne(7#47W<#VN*WXfVCZ
z5$pwT<|+>2f$=*!#)_MA*?sBm@>+WblC&>+U+*@KMK9CMP`NCMEi8)aYAdSYn+K*E
z4VJ~vkYcH%X5G7-wW*%>XEU*WB$8pp8U^s)bwg3$QnO4NHr7<p!e{>w%egma1NTam
zzMVqQuuB)j>M>*V@)@Y>!)eA&svm+75G3k2F84}YBLCQB3Oq@d95__i@z3)BuW>14
zR)~c03wsBYQ8cu)Idap*jpvhjP>n%nyK2wsSf*FpOPgboDiVkR>PyC91Z4sJhQnx3
zpEMjDSy+s_Bx)(4a|P;6bA*1dGn=4mqYTfD%qkL9en^o0JSAk=>WjZBTYfCUIc+85
zT?xTdnc+IV-)yezA$_=hA^K?wTuQEw>`A}lR>!!(*3u(s_4SJ>$Ay8pHJ{5iK`01D
zX<haQt9Ll3*X-)siB^(hHZ}}mDGqdnYaH1MI?@|`V=L1Hp=vNkm?FM)d&Z5A7-M~$
zU}stv$+wP{f}Y~j;Xe<T<!Avh$*@DQzCqA#q{d`u`_te&u6q-y!*V;|?@9nW1TlWR
zd?o65b)CQe0sb-6_~8}uy|Xg9j^Fy^<i_vO;R#U%DP<@Q>Z?V>W{hF&`Kqqkk{dZY
z-HsnrOh(ujCdEv@Ma+nYImgQXW$5v>`M`!UN0@zs33$7*b-Th<w2mVwlyAFNw-T#)
zkrv_TAR_h?-IabA{e1`gI(1ED)a~LG5)xu!a#!<XWZm==BnJ^xBFWQ3>h!Qq{2dxg
z<0c!Jb0`1Te;r;b*V*_)rHT4oFj)7@8kZ9qR^Ue1f&XV)RDQO$*Y5RhweK49y3gRQ
z{M;^AUEHuLN)&7~R2#={v`aLIgYC1?B?|DoWmT=Q8gnkU7ik#4|J+mNONGSKn5Mhf
zTi^Ox8u<3Jz4%vbc6VW8bY{peL{$#SM;lH&rVXkPWysIu#y>}*C@vEI*>Tqz5Fgl-
z(6jpzx8G46Xt>2xXU&uor#4g+9jemEYJ5Ci92`r;*A$uT=1l;FAD{5gS%3TR!}&}X
z=;w2aV8Pe$Ct;Za6#cKmB7P?F(#ks}6-rMl(U7M0W)-E(ET0Ft?5^E*!@uwh2gEF$
z@1)h{*&UE3#wVcfO5-*wW=}6ceo_{!Ey{?bpm%eTuXxC9^YXogmwq&7Tuy{-hS7mT
zz<_?XecTZhnP?GEY2qXiUal+csexP9(JG>0M!5F?YdwYm`DPdy_MJ|^iSN}j$m;Gn
zocb_@`#z4@<1jq0<jdf1x#=^}PGn6N@+anS<;?Oee=ZsE5lG64pEFl?&Ra7O>bhLe
z3RbywVdXL0Pt#<g<yKx-%s9}{gZehN|6EOT;a~XZ4biqbysocmn1&e2bzo(mA7M>`
zgdO4^!4_Dc;oL&1GKE?S)@M2WU*$No%tL2+BVYX@n(<+nu!)_@F19v@Goo~{CHO;l
zVNk)0;#?wx0!*-h_ef-JKTWmC1txZ@o6<_01WD?pptp)E6&ug<beRuS{V*m=*(Q;<
z6WZ|yUgwls3z>)qwncAE17+?mWTu(ci6Wfuz^60-2iIK__Opwtod6AyI^)}fh~ua4
ze7pq2`xeQ~b$ZxjqBcjP>CcqaD*?AI3U{mKMjWurPDbg0LCeSHT!BekLbYxqJL=kM
zyO4H2Z1NPlV8DUDc(TSs8qnGMVq;xt$Th&F__S0rMoBBE-RmSeUtL)ymg*<%;#QX$
z5*7q>bm5DPEI3k!kOgK{e2S*}eSH5v#oQrghAR4Rgy$Cd)cl!qjX9J3i2Wr^U}q%?
zr0#cy3G~gcTW%Fr+yCIHFMCzX$$n=QB62ED!rMD-&#rfUx^m4^?AM?Gq<jR|?Ig=w
z`E4qD-yZ100W6(feLXQL_31zB!~J7AGJw)2x5l(3>s>`Kw_QpJ8hO|`LVt+c@9B^t
zRmL}izLDa5(Mvb>#!Z;co#Uu)dB|?;BrdDxc^jvi^EEPf%gEGuGw#;NUPMDX!AMnA
zM?<5_VdJ@(f#d6*vHa<exP${cZu=F;v-*Xwd|{m)q#HvMCh&020;b2~Sd^GS516bq
z-p9;ou^fMAs#Fg?cc<h<_<i2XtMLj>R7_(h&o|%2iPOBbbTe7iZUWrgmq2{$Qz((j
z1%&DF^*)F~k+`@|XN|iCR$B&5087bRE!pcN#0oy`o&gns$-v*~uDQf~TeV!#Aqlh<
zAY?;L)0@T^c%<oM!RId6n_-{>SpIu=1=kK3jYxO$KJx|z?M-6FYw<YW#ui0);`OrX
zzQ3u6jT^o}J+A#Fw|BdQlTE%XXC7J08XK$DA782-HO%N@g=z*Zr`DO@JTK~&gQ;&*
z%PvtZJp-Q}mWy}hS#q?srokeNp>PtSl#wT0n}|u16%>g&iht<%nY`+NsMg3Thag{P
z8)nZVcvACt6-4E5uPu&$ixcQvR%<A1)z9KP)%QwzsH^tQoYEn`+qEF~an57k|5rDm
zj@qQ4w1$>$LLilkJVOx6*NI^-o7GZb>TeygK}SzdcdePHoOD}Z%AY1W4aFP=Nx|QE
zmbh9M0s;RnZgPQq5zsV1NqUuf6%Bkmub9y~)VSrI3vRBr0S|E<9oYh&Vh`Ht?d7zN
zPl;0Xh|zN?mO)%*4wgt$ou6IrjDPq#^ZlA$2d#hA)7nIy)7Gw1{&~wP<xYFQ(noHM
z;6X_eOqpW@$0d*-sS9fy<PFC9nDMcNvp=;DVvSPRI=%$MecJ-gPA_;@Rn^t98aag*
z=9g486VYe_png5Fyyu(HBXt5gkox-DacU91pSCX%9+6ad#V)b&eW1FsTrHt;n5*+T
z$~)UztmV=QYX*rSxQyFK4hz@XU?odxnQKKncJ9@yUasNOowau9&NH=DNKG9x4hxl(
zz{NcWdqKCddJu56Lm?B#c@A%}Hn7e(rp4)NG1VY7Go()t%!~&L3(tX?w|LiBRLP<#
ziHhnBRh$@Q4c(1PG}ah44Y2>i2Lnkly`8$m=^3``pgtM2hv)EVRw08e{41~^O7JqO
zu8R6?Uxc{JGxhiuoXYu@-qE);Drt}b47(9dO?(nO7Y8C}^Hv1Fkh_0&^!;)o(U7w$
zaN0<QkP9P8QcOHY-{eSM^7gwXQAyzG{}zSw{u_77=SPGlIt+e_%m0<+=`=T%NFp=6
zK}bbIV*WlINu%nku`cNEbc)2k>*-jz((bz@=5%Q5=qTddP*@n!rl@64`$zXzx2d<*
zVMDP(A}#=f^;<PC`)i!UJVKa7%oY3Mh${la$%>CmNtI&cRl{(gop_<f-`@sjsUkwe
z8oT{AlcOU1Lh!_(DFXna=Ido$2~lO$i{eH3#}B%jI^w4*@Vwjc8<YyJ#`cb-hJv&)
z+;BZ~d;|&a2@ZPb*fv6OcJRKR+niY>N>u_Ii7P<;1QQAa3&!I9LVU0kHT4*cE6L8J
z9r=v%H}pe6M;`H4(7i)XefrQEk;;S6v*^EZfE&a)WLiX~AM5akX<eh;Y%8H<9wyI|
zKuQ~4{!Kag`O5^MqCPemSg3AXQ%XJZ<(&O5_a83B4|5kD1ySkW6DuP8OsiX?*uK^?
z`mO*T0=J+r0zIhg=YMBr$<)`B=fYCoOA|6al>Vnin{Aa`%AAvHkl&Ave`F$rJ+9~C
zGB@wp`aFk4(%$-~V~FgLR^uuLQ1u?(df(DmdU)wVS`ZmM;j5x02dC;Tegy}0e+<L~
zk9F%uH(&;5DQVO}b=9#}r1b5JgJq<%LVycsbqt#1Z}?Wy?4l86ju%}wahFerjkZ*P
z^m<~}-LE?tWkX7iZ>-#ba)<T9{sOJP3TuTqgxIJjrcwX2HxyJUF*F^r_JqK)8%_rd
z8JAgIMKORQQm{i{((*r(&Z5!(x)5+$HIV4`-hsE`*8H_FiWdWz>^I}5p9-Y|0POMk
z-WQW9g)^P=OcnO&I(nO;sd*YShdZ}so_voUax)VM){sAm*f@9gPvxiN+<4WB(VyI!
z=YFdTOr<B$(VpC<<0nV3T&+C*&Sm*hV!FQ_{i!MOM9EtL7mhd6MqWWq4qVfbm5msu
zEwlXxcE|9-naM@u<daz&1x-`t0JDM}^BWzde(pZWGLrIB!yOb>HCRH}I)<pha{Z$z
z!`)nESw9-4)`o^9Wpwq@QWu@(9!Jv`_DzTx2<CaiOw0~3&TOozf)370oT6xtb=qJg
zCF%|!*&~mFQYVDGcEEEJ+fO-d%OA&x-(^8EX4}1rZzJ{NSuzg}--MwpfoLiQyf9uO
zIz$Xvp-#*#sQS_dYj4xC&SLJEf8M+$D&~E*ftM6myX725PS;3Lva)t*ypNL1V`D4~
zIynt}TSMa7YGQ_lhRVuXT3TYd1`24R2?5Rw+)J&70aq7qxg=CNk)@nJ9kPUVnz@<Q
z7iXwU+2{$UxBMjtrn~)L0OFYpJhc;|3WdLx)Kbs<dW4vT)og2?9m$37KiFQtvFL*D
zwnzPCS+lhD8VD_dqST<eqIx$(o%nF(r|^~=o<6EHReBu|Uiyay2Q=`w)eD_D8ExFy
zZlUB;lv#fLN?=d*-j7b=?~G$NS18D&`?yda_q{Im4KB*cDJw+_Sm1WOLW)(G4xjW2
zB~$#hprNU$k^Jk|FO7vOSupC-Ol-J=gbDAmmuk6inuhcgv+%yS_50s&iv;H9#BiO?
z$2kPMq8rd;W5P?T>Pe=jBQu08w)4{7YwC&A0^G{rk_LRO1x)}~wu<!l7=<4ODPrRR
zWCJEkMV{HkK2bs!P|)V0GQMJ$%*1<q`q4Q7=w0~xg=A#g#Xn`Z6R;T`Rt%ge33`1r
zku7U=2@21Q%IZoJbF-CE6nK?1@>xz7ybH4yf{RHDGaSv%VCz+kf^G^i(aEX(_oCu5
zQQ&$IesFFH!)>#L?CjXTuWSW9gFK9O#8^USPSEGp%dH8hZ8ULp_?nkp;&1JvX0Wz6
z!sx!{xh>~)1fHIc5LFm6r*7N-UzlL~V$gQ$gBz#G`9%=&j1EOY$VPA>bC{ys@&p^b
z5hj=kEnfL`!CO4A&l#j2-%#K*Uh|tI)yI0nP%GZ(aa5;LEN1>`P&b+XaZ8w}-Lvt?
z&k8uLQCdL^SjsSM9QmTu!~02WA#aRrU|ycckFM=dF_$v{1r&I=FoWgH5FCVRRLA<I
zZyw|JZiYm~;M%3E1$s#0cH9b28=O1qI6ArOA?nK~Y;*1W0LzB{`uNs->v5}p8F;F3
z-O#2z3V*LDS-sOmeye)J)cVEaw&@itr3`>%LojFCoU3nU?%mUo-_DN>&bFv?W(l)K
zSiTgO)!0oC-mryB{rB*2+bwmvie|7T)a`s^k)0Z2skX7^`?S=L$9!iTKdOnslb;4-
z%=&C>z0Y<En^#Q@n93OiFA-_9(hzYUQNgdA%1>HQ(k){MkjrbcMBaIFYT=mq0F>fV
zqUlypS#+GYljj>kg53xxOY<u!obP#2q2oUI>1M8hMHYNRI9$G&s{#le6+}&0${4)g
z=XZufUWo25_!j177hP}O3sWDMbFIr5bqM#cGTUoP%1`e(hSQ*0n{?)Gh<$k2@t*bU
zSW{=WU<&(TDERA&>&<ad)ejNPfI3we7=2Rm?8ni@cF?8czg@iaHWWp;(+AVVZmi_;
z)krKF*44AK_8Mfo+HYa=#_<Ih!HtKxB|=-hKJ1)MSO%`KYe=|1_=x_vu#wYxND>UL
zT7W(~;I6s^cYPynEP<Q|pKT#0A+i1UIJ6IPqNYR$41{=tGx($lOQ*K51*m6twH35P
zc3U~2;1}aT2K4Oo<q!W>f+%tjp7S!^y2x%`)cln>uE67<qphLV`LUdyvqn)%W+>w7
zMuVWt!cYbiaftl<OhzTT4<8cpz|wNYI2v`=0+C0yB{HhTmMw&T9=KTpMkfbT)}T<k
zBFL7X=yyE+_lTVHf2AY`l)%EZ6a4}!n1+s1o)pLOfr4mM)R+77wN6hkY6%to6E~@i
zt*}Sep5osjQn%e%qj%(tfc}oRt=@opgf9HfOwX$E2p1l;$C#Z<yg|YU3q+F~?&mSn
zb=M=+OKggPSc_WTLucpcKjlQGmg;U^c|VD!IyxEz%Ql^NNN8e!i_nASNwv!44t7ta
zfZ6B7xhN+6$`<XvXJu=ey@fufhjXYyPnE6>saWMdCkB{0^*bC;dVT#~zRK)PMFm$4
zFx|;a8v>8d54j=%XwnK25Ag70g2YfOos}b}JeEUV8M|SS^&`)xeT{;&k_Xs9z81zP
zoxU2FQ)!PP#M_EM8h0l0%f(5?Dp*o7M9nfT{xjccW4K{5lYxG0e0)q^VSE;P#n3>%
z)ov@l<Uit*QTXLMv(_a|j@zY3-C2=Hmp=$5)X!JtY%{qVz552dGriJnpaYGQV-9qk
z$1WDb9@dS)>%B(U*S0^w6l-ith?VNND3+mB{Nu#2umSwLb=_K7$iWUG#KWl5sOTsp
zWTflAxY)S0`bG`)$yD}Dtc1xH{(>7{=9cVjc)i_^-x~XSkZA}#HESxXMCP2I?h9K7
z-R<}LAzbl<duQugIT=<rtWkPSYC{w+smy<Y1tZPQsrjYUEX)v*BMZ`P@#M?b5dQ`N
z9*bdOW<c2OPatY(cBP7whW%^70#!PLi&pK_`1s_gZnKMp@@59qt(ZOinx6ia)1%Hy
zI@nxZyA=rK6zY=y+PB~pBf37R{=#!a{j6X4lR&BA^X})q203-*P&>kCA_Wy8nu5VX
zb#++{F|eGGg`LjK-STT_EwopZX$60BMqYVf2%s%JGK{Bu7NvBGTn{}nXF=fAn%HTo
zt2Z}LHUr`4i3eXbQTClIQA>5r@88bPT2j<+0ziX}(PU2{$p#Nx?pM83w2RFSpF+Uc
z=@HbJ>Uf#8V$|XnRGIULk37+f(p8%jum1U2CtP}%x}(CyR7Z`Ba#eCB$R43T1^yB2
zU-3SX1DR+$*!vzH;9(OIgf!%4Bq|yo$<h5&U{A3++Q<36{?(wjx4im>k@bh;M!O87
zfOW6QpBEO9;G^rk{QgdZAFT(aFl%;=B@1T}=m;vl>fd46Hg&OR69k|CK4k<Xgh>`J
z#6ZZU)mvgMldhc5A+QND%fEWPK7z95I)HmIxk~v{yH<O~iBUxbh&K0ce*XhLbi+p{
ztya9<36ICc!&PlXn`RX8df#E$5+NgP3jj@kC`P8gope4NAG~Zqmh*51K0g>%;XGV#
zjP&2YGo;Ay*+D0J+{XZ3r-<K)?k=f^HCVTakL6)iwJSdT@5&z?>;GL>0er6B5&D%m
z3GL0@Shj(|0|;W|tPuLFBf~UQr4~EggPX%<=cQ8yZBXnv%lRlpiPn$9wIcp`x8GTs
z7jU8fpe|F~)OFws44@7yNh)B#9Eorc;;Rr%pXhYJj`Z_l#++1h4)O{APCtrYa7Quv
zrW4z<*&uG8xpT00N`{V&2@sNQhxtG_jV7WW9v!B!9dy<{HGkVfKAjePjAe`O_NxAP
zN023Lcuxi6=VAqYZ+2P}6n?Za{&ZV-)E0NgPlArIuwzizP9q{>+YmY1YY)_i{-5=Q
z_2r*G(LR2DY|ZPK8ypyhwAUdaIXk!iewGbBAVCHH{(?1^T5AXxsdc~1%yPRe8Wphr
zq{^bZf(u-VNPgO@q&ZPw$895!=a}RcfV$Z;c@-jq)34eJTjI%*4?tt1<C4(-sk&TP
zm8Y>ktcF8=oTVPcnP;-w%4%mYdFb93igaDq>gnd<;VwZ!ZpkCTF`Tt1X!{A-rL}d#
z3tn!xr{R0{QN$<Un6((vqaN?Abr$b3;MITM)J)2<Et{T)23{)R^*Pf~C-GF$K09Y;
z&dHG@*+s%%$bcjzcJqT0w4X%Nw;74;=i5^j8WW7vgl(x6dJ-3aua0adAcN=0vNdx&
zIK-fZm#VpVu?n6Dk4H{5ZfMtOk8>>V{-%Gxi_r>Axt9ZsMmE_sHqsf`_qaR;OKCpe
z*v;%Na@5B&(I)))-9kZ2O@$o(tOrqQWe#7W-ar><%jJ$L_XqrJRE$M_aBbw`q982t
zxYtTcgePWos1V=LYC#>Zu<qjJkC+RnIG2^!$btSb%fb71lD4_9b-Vvh<KHcbU9?|T
zp{ul$6orF;99CVIu+~zp<T-}!SoD?>1;6BT2Ut(C%$y1ucr~KV855j+-^0ViO-+<C
z2;*+!&TmBvz%}rvP<Lg}_oYGmr~{J%C$e7MiN+aPFPAr*%AN-QAq(-$*uA5WfVaaR
zj~o<;Pbiw6`Ywr3_H|VKB9pA|?#8rH4UFL%ej~suW{eXIPjf5|da?$LH^CMwE?<yX
zmFW(4+Ld#35WJe~dgA`_6~yvo;}`=piMs`#^}Fh(9LHF(H9?!#+r?DpUzYqAk~A|M
z{npE}tsgysElI~GohVVd$r$IC6dUjgsc`!A=ux~WLSG_+=$0X|G7H&mP`~ieRF^?=
zw?_d;*A48uX`FojfL~s#=9swv`5p?hFw{*wP?}f)$9ATvru9|KAUr21A+kL->4&52
z{(#g!0tS~Q>>8Qgr#Sx~`YP^5=Odm%xaXzWJ~<D$LK@A~TnlR2O7FR)DIfgU<V8cF
zUaY#ov+&RZh(3<65%D-=WGvHu=w58Ct~Y?iY(kwX4_+;VEVW6!lp&ipDg}}DM})D_
zwj~&93RY~FRg{AfuoZNp`BBR8sV0r{tmlWDHA&#4pVoK!Z_R{UA7$qodfmu4U8T&o
zn=opsUwrxfcKS+&f({!v#I8_sS@Hllmmg{SLvM(q$ct2EgP2KL$s=<^n5I=ej{uPD
z1qUg*<{*JQ5#8g%hDfK|H4>kYkOBvy)-kuSB})4bV*d>LLxx3BQ8%mJbn)aa)9!(*
z+4Jr;WZ^WLN8~jE6HWb;r`BlmAOBZ_Hp;mT*-YYzGqux!s}nXMOer3iuXM6(X5HUn
z+IBH8Q`N$S!_s}tnDpZjaIyNY0fO}`b2kgyJ##BLWe+Ew_{5ZygkP{2&{g7NF-V7F
z_#GoPwBX}dv-5+8Vz{ClG<@(czDgdA3&CZNmLqOZvfsxZ3oYn{r{pP9maT@*s~|$`
zOIDN77m4cuN-t6|3|YSCSr}LyMgXNnGSZ4+855oY3dvBzxJ&5A^Ju6PAjPI(@$UE8
z1gz_kvmxN~9_vM3A`Jj*e9Ml;|E#IDpU!|>%YG7s3LiojLq@ZU%xrvxZ*XJ`6%c@;
zt~OghvwjkJWHp|vR-`Q9J%VWW>cOYi)SY*seRRJ3LZkHg+SmZCH#w)J18He?WS}um
zP@$=bl_6R@N*s9`yQioO07sAw4&We^euRd}n5vWyPP|5C%kcZc{n5heD*Q3cV=oKh
z!W_wwbINaw7b}B;bo|DLI~q-}f|UO|A~iEw=WhFM_yfiw+H9Sd|4#BcGp(V341Imw
z%gd_`9K4M1-10NQaCrk~Owx38B+4&|z!oQiHt(Z+teD7YmSlNgwUg(8rTKibiJdL}
z8rC=|4A+C$jE*?r%YlAZC{Dsx)p@m2yHT4aAyikv8SU~8`Us1lK>p`nl0$vJ;?$^j
zAyV$XL70B6=wyT#B6en4>nqB-_UAen;`3_0JUNlp5}j4JVIW$qMPbLexRfw>S)%bh
zjrKEj8;{JTGz*&q;=AzOEga+<6R6^@dldw;pKC{K1JpC)R?OQXjkK;^Zx9QRASBl{
zIGfaltZeootLZ#R{s{sH!v?IJPhkmt{>!eWMnfkHPK>wUAQbg{mmIp;ETbfn%L%bt
zIYZgMjN|g(q;69haM43P3bLT8vi#b@;+ld8b&*`+C!I6^IhsL2HP2VRJq#qCMfG2%
zxx1>Y13QZylElK^UVMOp-}P;3i9w>R6t?Q{QsU?ym&W@TX?AUw?@?c7>*tK0np~am
z!kk6*31R@kn?&O%8e1f8JU#{UyNIkH%xPz7h9a%FoTB+>oq7%)_^WBQN$pQKfrn*8
zdWIV94?4h6*rxsiS@+{Rp7cS$quxkRHLOCR5XnRx?8HYOby7m^`zK2yQNNb_I)+;i
zd&(f230!GlXjey1I-*xm-ojq&WY)i=-j_n$Nh`CG_{0Q-;{r{cfyJ$%uPVB77_1~^
znCu|sKWm$Q>99lj-2<>XKGHhd*amrZK@^W@he$w?8)~WQNsQTNhL6+153_g~i8#$$
zQ^gY9O?qnP{=DMZWsi=R)r}pi%Yb~kmP@j`Y~gR+4&H*eez4_6EwALcC)@18={of2
z<c&VI*O<ZIIB5KpLWm;#co&u$+aIMU4>X%@-9Mjir-%t)0C96Ny<d|f;H@o2t@VAS
zDQf|}M2PF5@DuiiZPC04LdeR#<kwX*p`W_EOcmKlxzNwVFxSY^txnB(ga6e6({=MJ
z@Pwaiwh|;vxZ!2RFrj}SHVq-2h#f``@WB1I@yHEkF1Fr=s13}@saBat<ifd&HCE?+
z+-DlsB9_eYOgSu;GHzwTs42JaH}qG~04nHOxF}HIC1eO-aZnir`gt4=DVO3h)BN(_
zeiExkRS;H;?dQSA>MoxKJP;6G@~%eYV6^6rG1ZuwACdf*-AJZYg0<JGs;bBjL;dbS
z(#t<4|GLK@(|R!IjUS<YUlP2$g@53$xEot9huV>{jkhg{O3HM}nytr~PlneGd_%gj
zEf=rbdwOflGa$FVp=KtfQ*v#s15d0gJjjG+-R~|6|1us`(6n~0-D}}m4hPY<Ht<e!
zwovA*)~a?A%%BE-i8E5w!RA+`nn?Q{gR}Zq23k-Yd*8fB6;<Xbv=7Ml;wE0^+~tTu
zjlrM@A%hZ9G+qsRZeP*Ix6U&I^fVI1&%KA4x0;Uj?`Wz9McO?~|8`LA#Oa+>BxA1m
zT}eoITT$g$4<R@O*+7ca+(vG*oqXndI*cBtt+3f6$cSr6p+1ai_-_IDGFCb4jv3>K
zW$qtoL*AJc;M%O=1zEt5Q-z3dx9`(_e@1M3Ir1y9QXuX?%AgA6Fe-q}nvM}>%2=46
zp2<VBv1ZIiRasybQ~fgY?~~q8Q#+P<_Hm^8bHRl7)L<^v5i2#6x!pNajGM**RhW0!
zphYsLTP))l6yAZ>nTY!y9Rmm){S>aFa#TadFxE&+8feh;Lg=T=`NJKF-~=kzghujT
zP!ATYBqvKHpRob|6WGWiv9F#`Qz}61k2`uG3z-sRa}-zl@bflc_^G<^C*_jY>s5R!
z1doyZ?dGMsH1mW9uOa;4>=Dsv2-QNU#wWc$1%>-me1txQ7lx}`msg3q#&}+yl&p}F
zNi@SkI?$25uy^FyH^LUKmYfu6A}zQ>EPS)76cK`Xf#W(-7GSU=0JAY<nCX_6yzZAc
zw04cw&mC6UZ7jaV^K{XyR$OCZ>Qp!=?p9BtxQlM;Xv9|y&wNN%2TH+EVb2mBo?!)t
z6=j`+b)BD=L9g$2s@-J2zwIP_<?*<UmDhzNU*yof6fw)MGZH*LwcRl4wz=E2HWCo>
z?oYRu6U+(50GI`r>N3NGvNe+T>B@DOZJF4;YvN<VBfC9y{LX@};vt-Z8cU|N<oB~s
z)JqeWz7LRTT<9)H6N>+?`*u*gqqDOMf!@^+SDu{-yk4U=%S>+tOvwx4%2#dHxu}7U
z*`Mbj8nmI<rztnrb&QC{0r^q7N;{%7ojy=r2*?Bp(4+Rq>8~$mNTD1GHTg|};33%i
zLA<BKAY}d2E{WqV#=#)us6N@BlzJ&_LG`|fE<3oOU8A2PdMUMnjxRP|=EJcRQ>TUz
z)-A#f5xm2jniiGctzs_kv4mMy{|PPpfXMJDF~-acYpz*^Y78~EY<-Uqa>_*)^wgC$
zq4=Z+AOz--8rCVMM}rKedN9hp5ALe*E_;lrauBGr9L+h;53ce<dwdhU@B{VI@K0>Z
z>BhnLz3snDomFt1)scuel`H1Fn<A0@Z!#_>EQzoBe+ytiTvB%2rDUDpY`H;-9cQds
zUt99`O@?g(-j0>0)6HHylS6VzwK7ddjDR>9dGpywB<7RHMYbhP?Rd<?Odc~#4%h9Z
z*_P`l#mSYn9&)r8!@UCNjVe@)VwgikQ{3xhu^rB(Zz+iQH<0v^Tt;SjFEM0LY`;JV
zk>iZ3t1|&ln+jTBww24}YI{qFJ5Jb2`79)Ocv-fU`b#XGCUPXn5xJ_fCcmPn1~4;s
zINiU%+CSYfM}EGSe{M0v7+pDSxq#loi%+WyG#<Hm8OXzM*t5EhqB}_wcnU2&w675G
zxj5YWiGpwn#(|#VDRZA!Gfzd_)i0oD{!STGr&zO5ryU|HpfQswqiLiz;Un-vMx-Y+
zk-9Kw)*t7=KD1y_cyUd5;yULX1*(SA1KG#|H&8f#6-f<$H5qR5Pg_1bZY5{(q`;Uo
zrFwsh-Vx|Fp!s>R))Js;n_SzCv79G*I6|3ICUk0TbDXZT+UWxyJ`yNQzKL0&g}AT#
z$(?isR_bdS&C<tzSAR`zC5-N@vD2n6i+@PLI9cUMPVNMzGHw)CVnTlhnkjjgABH#*
zdDQv91pCE1CSu0h#eGR*vOd35aU`FXp4ori#Y9Mvl3a(RAuIE3j-sjvufy%UN6xCQ
zUv0UkM7?L!zR-uz+bQTUM_o^6Di62j;=@d4vy^_{!9-8W=`6ojrAe8|57QtAX;0@0
zuGW2j8{2nMu`}4V-qoZ5biKh)vA_}$Zt0+@KtGW1#R8tA^(9`nxx21``diBGSu(AW
zOpJ)h$c>rA_Jm4019^Fs-Qjk-u}%L{NcLZ8TKB<R;OXw=NTStuI@i90GK=!4K?*;p
zEnJx*gXcITm!-JM%=FaOtV-(D55H0RGanO2YT$9xukqAk8k#_Wwp(C%Wa236{tkwb
z>xaC!#vU@V({%QbOM?kMU$R{+SK4Q@gfC0u{ayO6iS%&X7iE2;8#=CKdt-eZDLHWO
zyK1`+nm8;fNX}e;$tS5NFu<5HO^rpaCUDzy;ZKBE!WX*`IDz)PX2x$JdmPu}Vv;PV
z1~|AMwQKDvP&7F}h|xDSFn@sQi~o7b8+Z<<8~L#_bhUB1QL0B&c|AWsd{=Z>FNG9l
z=%_Lj^u!LAlB+YI-lY4PAP?qQhI)8NGI*?vri(ZV6Uw#pt6sm4rN<%bkMwDd%tbPG
zfOt8&qp#Jio^5d`-Rc`fngt<!j}l3jIG*^fnuYJtsEi(<8p40FmU*uUezU^!6Vf3^
zN6E>K{uyPU=g61+1R!czIa(IUn7{u*wKsYDWu%szjT>!?<^Kxb^ZKuzh8&c1bQOsa
z5}n)?6`rL1gJ6;|*(+}>*G@!Fy@_%_*T{!dfAHs@lNgs92{OS&)gAx9?R7dvx<5~e
zQ5oSWaiowyz%tXw6kOWL+<>5KiQhPD=-W5OJI$Gw0lf!EdHBj><Ez5zN)-hBt-T2j
za)^?lRCS&v7ZB=CiO5>vnE=o|ZM4A$G_fZw$hUQ$CHk&;G3obn2m5stJw&;g{q!Hf
zbn#9bNXEb!VO5%wZRDbNoN~6~B6f|WaAS$7WlUZG)Hpj*SYqWs)<j!se;K*?nI3f9
z418RJwyQUhv#Z@tIdIzu$DA6ZE>Ki1@Yf-Fc>_-U#JkRd*x4MBIGle6w4WSbH`SMP
ze~s|Dt)knrLLNO`6<x`~{AX5LOUdr}#TDQfhNv5A^t?f)Fs1gfburED%fA{{RN7ka
zr(gxxO`IL%?%ww{T1owXHiz^9xR_h67lR2xnSYrLw=K%)_<pWXTUYce>xg5og_`}f
zhRX;##|?$?YG)-G<oUuDQr}`$I<<iZKSp6;YGJyU8{cj70LuAssT)~p(eL4?N4n=g
zIWmMBBq3N?YHWJbCMw-GK`P`2fDsq47*jwwc3oVOo^8KpQF;D<2<f5-5|_5aiL}=Z
zK}1vqN2h&S9l~?7V!Hk(5;H-?YO!L~&B=4oC|++9-Z4SH8%$1!#Ls-k`4s&;=3-yu
zU+=?Pka>R4Ez)Z3qFq9|dhw{^B<S9Fg$-UkW~bdD+?ZuvwG)?hQ1Ty&4-~}65CFZQ
ziE-_PDms2`xzWu*9aRXv&Gx;V<e=pi<@uQbtDS4iI?58Ad0G&7<Um5&w&J>@A@MZE
ztAgcS!_q*R5AAq_P?VBHWW=$=+>>9=nC&QNxpPP(fWgMbP7RBl9)TWiuaqemBs2Ag
z%hpCsN&P!$juuZsw+0l`#;!~+3U~(^$5NH4<AuNcK9&g!5>(I52@guI{?8|4kE!fu
zr@i(2ucIi}NolV0?u&0;G2OTlqC2++dK!DQf~bbH92jR0&L$duP5X(pwrZNX{tqVq
z_oyxVA!1@8-|{%d9YezxL7kMJBz$HnF+@Dh)TM!KT^Tv$G=Zvd1_7+p0LT%cdw)Hj
zi4SqJ1Zf?-DcKx>zC`wL5icv^ks|ZcFl}qiPKW|;c!yi$LN|Z28q=?WT;?PpHBbAR
z2EZqU>SoyTcU*Zv#&apc-5RU#Yf_BVjP%S%_RAw%GbkT{(0x`DxIYI48FN!Z^CFz|
z4jA%+DN#f)7p!TGiGv})h>!ojiXvlQ1A+_CJb1Efo@}z82R+C;mF~3`Y9$BJGQvH$
zOo7+cVd;b<N1yMyfV0V~kv3u4+j^Ed#A~V){cO-$Ry-u2$1FsCj~oB^)6qyU!0W-%
z@gbpNQLg7OO7N^P$|}XhrH}j`RX>gJ{4(?hN0ovezK-2Ib*vR|E6!GhWEOR3*EzVa
z!Sx{Az1rn&kqoV{l=h{$FB;{@@0+cbp)={2Us`Q+Gg}|^qe&u%GcH=i=`9mjNeeCh
z#SH5!?vX>Ec;5r<Vp(p$6G;}cC~SSe{8%Ui5tH+$+tRi$3k^4=4;*kj-KUij8$9P@
zQ_C?B)$|w+{pSj>#VJkm`8k_>zZvR(aI<(h9r0wzMcL}z@GLIo7lG|zXera)3%nuD
z(mNb4J4qMtft6v2x~=E=`b*SEaJVewFrkaFzF=?9@>}k4;2;=x-v0cmr?6O|f~B%|
zeao2DB_S5rVLk~?_#3OhbKGXjPaK)}gEGRkJf=u{#CCvL&?o~E!wKP9r(>*OK-A8n
z$jAHp!Jodn=u$m2%>$M=C${9Hp(bsx?pr(0%&XOWai%k@f2R-k<>Oy8naqWk7^1fu
zyU*~9jL^p1>*HHxY5Zf385b<wdy0J5$JpAN5v)=ydg#pz7;yvT=8Bp;)LzPl*`MZI
z7zstxc@4Nu<%Cab5{v6i&Bi}k0m_2vZ3HWA@Iyt6(_6{Cf?<kJ1Xv8=9rl*z$A?(>
zxFo%tXxVp6Rz8<P_yqp%&l^8-fsZSnTS(SwJ<spoFfPr?sRgq`^wDuVPx!GXw=WOk
z{~P*M<fTrhCXViz^p;-%pBnPA1v6zN;UsaO08H(ai_e;GX@frlxrf=@lHhdcZyJ7=
z2EtsdKnN*WA_(WR#b8sG5Qn%Vr-%J&uI?^OijJ*sVSgU9oVh)|?7>=ai<>Hc_S>~N
z>uzap+zL)UfISFf_=@|;2gY!*89Jft$^FH*JcBfuEnkts5g=0e*5+;H<aIlg`5*!b
z03Q3B)s-7opS~Mp;+~Yz2*DBGvqVTW1$_<e?G!>Vi@Fl6+ZpxuR*jD-cQtXbvfm@j
zY4LIWJ-Ye1BbzAcacNbY8gz!yRBUB+yaayj|2Uc2S_y7)jaf;<(s>S-mK+K5S|u!(
z2Ti}nC6wdMIOzpmJWf;Pn5>l%u6WB(1GTiPUz@)0&8<zT!up8Kjm-bi_SXFVzjm&x
zL5oTrhpjcgj?&Bfx1G}3o)3`c?YIa*&tKrp<#a0X0PIB_6IB^#b87}9s9@?Dpoyku
z2Fu7*;9Zv`{{MxyjoV~-$?ff9WvUtMLYV(wmO#h7?q>Ch48Q`wJ^dvTT9DS5-<IFa
zkd~wdYlx<JtmIo(Ye;!Y7VA-RE%ju7qO=ETz-Kk*QH7KvT=T)8FF6P!kB<Mg|2loE
zJGqIpdUSW;&tqsS@H*`{Gk2;@#4%30C~_7pE=JM9a0Y#fk{g$9eeG|#Nv+)?^@EA=
zMwjtszc$`p0@6c%*(HXYHL*EL5IVKbnFb5Geu&{%5!NP$rz@xuAE?;oy?>w>zK!I8
zS~%LI4w@CuyTx&<8Ow!SN*qJ4t0*Gj{o^yW*F3OGU<(F;@t(nXmvBE*mlHVLxE!#H
zP5Oi#VkbvXTUGCWl+g5_h3xJ+uD^F0^I4Zt@Ld9Hj>p^To)LQO&Fv2ykw3JN-iNYw
zZ7pl^<yKBx9sd|&AKI2sPsv#;8Q&`W2IV2ayy0&}3|e?oY5HdHcLr)*-%kG60^Sh@
z$O|dZSjZ~%dl|<j5j<Sk=jyVtaCQ#{liPa{>xAI278Z4rCn5E5*m@|fW8M@E5iR@~
z{sscaRCZ&Y&2AfhlI5HAn82x?z5#}O7ZMTH^x244D5(D8<jB+lB3Y;EnW5Hf=d9OQ
zO?vZrtOI&P8BBV@IkNaN{Wkg6?W%%LvS!`q)-|$}_jcY2fsT+L(k!pV+~R9QQbOX4
zqI6yo#?(PN3v43;{LgKj+tJj&7ml!cqu>roC0g<|$>(TC!(ZE0#RW0XC^mA4Y`77Z
zj;7Slr`aJ`+Icx&n3z#4HX(vhRW71KN|+~c!3outhv63Ym(`9JnGx#uyG9u_cn1Rk
zL+xZz>sZX`mI-yiD4Y5GH7WmJSat&N$DR@bmk~u(uMMlv-wpo4cYaE?3}3CdV!r?;
z_q;UDDH<6>GQBMSPB9hQk)@?|*MCA!`sC9fhz0(zhrkw7$%Ztzf+qZgs_#kSAmCr`
zXuJ;|JKFF>U&6DHR@2F%xh~}U4bm}wh(r`lU}|yQ#ONg&HM0Ry{IpH_;(kvdFV}hf
z+ODDdO^je1EuPuLMzMfO+5vI)<2!IVavC)sQl<I0lIFGA5*ebRp`|oE5m&YgA%f;0
z<EmqcXHWgbtyj&SI$&vQyxJ5Gq#apm){<7z3=E7Z-C5vEe|hJ_N37<vKf)kR>iE*@
z_R>Q^Gj+F4>hlhb9Cg!FYj#<JQ*UTYUlpiC$7^Xq`>}N*GzYxooEO6~%maG^{@I^<
z9^3t;)C-1(KMrzBDK!p*)5Em~3vilzS7fwy0k5}L(Ep?99JuRhqb(e#ahjyDZ5xdm
zyRn@#wy|S0wr$&PoW^c!+s-}TxcB~oGsfQU*k`?S%{iZbuXi6RS4<}4dbNF`spx@S
z4t^WTAES=AS`Y!41s7nSrRK+Hc!W8cwWe-=+wl7BUKQpmqa9|qjNiL5Z_K(!pE-s-
zV=i@i*vEyn6MqU7Slck;$B3!`3*!R(6eZ3eS%e5rm&YGx?$|)6JAnneZ+VfyUfA(q
zC2G}`$U?NmdviUsU`zG!fZ^B2PyNn>Xv$t%5H=LrFA)JRaLNa^Q5q!$)S2`cFbLWu
zYjltbd?)Q)N8X7b#Q#`QmgT1G?t^|@`W)KK5cy|RiQ8@hZPG-)Cb9#tE#^Bleaosy
z^upOv)Np%4b{=jxEj<iK<#L;(AJ@qlrTG%!gl>fZ>Io654A5x`2Qi}U>_ya+_PZAm
z-eYR;|HTs+*6<!Q49PwYig8G|*=hw(I{5JD*$RPn7*+v|IaK8)28ZZL^^Vsf1`uV^
zMo237{L3)71dDUrhkpS-?_xmZr?}qe;>!?v7Kj-KA%=n43Po}@Q^LiIN4piRUL@p(
z1}aoY>>ML{+wC!;jkI8UGHo1R7eQl1aAN(rO9~-C@XukMO>-DtmOHm?FtO3D`{U&j
z84NRp@JLMP2?Ce}bDl2XeE2o|Uv=yBqA7E|&%JXh&X$ns6c9aNEDrxm)d0noRdjBe
zi;azgg@u!qm7SfFqouyi&Dj+}POHxL^#eWSBKu#9gGFr<E7$2|k=^vzeZ_44JHFgJ
z9F}o6aixm^9k0_6JL6Xm1~!SDtAQg_#7OlxLlKBYum(ll?}4yna!N`)>~6aIlcF|<
zVGB?DbY1E-kHGp_kB{4yubb0(dr_RNxT7qXA%u#eMkPNbb`qiCJT1?&IHjQ*I-me@
zQT0|FccO!1pqWUqXOjj3?`<5ox5LbFo@h-3=t``$_u@H&FXii@F*&DQY&Y>U9sLb2
zd0y{JHSGPB8}V}I_|4dZiagP2%9OGN2Mr)VIHp=N5Nn5JE&<E;AMWC@b}xy71C)FL
z&9pSW2k<(TJ=pFwYS~{$=~5`gV7>nAB$%m#7a>ETaVz!^7eLUpgAt)%!CE;P3ouZE
z1W~ak9&?3vTTa0dY7l;$NmErv$u0L@b}C6`9jkpZn`gn{M0P1SBE_rv!G+y=QU?Bq
zPbUC%va|m!Nj0n$**f-v{_vwj$r*xd;B`<{;I<8r8K|~LCFCC!a5)1UALl;PFs1WT
zjN|K9mK&<=WqgcRk7c+n1byWYbUxr$+mux1^}C#%Usp^75h(PU$@Ne~M5SPg^_7|9
zLL#Tft?}1=Ao{NZ9cx(^mKyKEBu5#ltN>mo``dDE1!ErhMScEY^!bRaH`4xqH-23L
z8wR<^FI%drX{3AhQ@a^VBqilG1~B!Wi;Ezq=TmJHl$2&=GmLZtzF&Rtn5*g6ll|zd
zZpV(AN28^(2V*%{6t1SmtRrrp2XE!`_DE>LV(B*xRPgzM#`#1OioCRvZ&w;U_y8N%
z51=V8xU0e~HYbISm>H=X+j?}2PkxI$849Wijxy(+s?2;=zKh{VLCM;O;=d-|;J=H>
z3&Lbh=abWOErLoPg`fuW?ep7VlwR}9O4B6wf2?i&mfb@uW2~f|PQ|`_aIQr8v+!Ua
zjwoSa3V)?_u@81l@@i-cY|gt3fYFtSj+k?uoUi*v!~!w*SPDNj8!BmE$Us<3?#$lK
zX+$AVD?o0F-*?G)YEzDBAC?w~G(i+zgAiJ}F+OuNs(J7YkYERGy6~E=pxe&j+=3GM
z1!X?)#SJp7r{%J0L=M?;I;j|(8^$c89hWvY{~brw&QlqT%!icqI7|uv+O_ogd-EcA
zwWiRpMxmOesZEUK-vmvvIvMvA45_T4@3dB4872SHlzq3~86T{84~)_4Ecos&`Id1Q
z17oM(<Jeut$}}@98yCvZ;3++;q7AoeN+mlOOKH(7|02dPrqf_?KN~E}XqL|UR`2QR
zi@+99{-9nh1y{Eehd&nsu>9sB-nKYBI@rJ5J&wLkH52!wz89(_r-jKk8rc6c)6c{n
z`NN(2OKF+kEM!Yidbo_@TK${c>La_0PUrR3bH8@V6h2>VgX4SgZ|Uh?KKs^oZn7bC
zM6D6=sb0i}LP_&s0XNQMKH~tT;~9`yagK|f6?`bTy16)gECrqns2$^--VnkWc!vvC
zrFy7D|8QSwzy&)|kjsLRaaYV_8TvjysQpNI+V0&&^nfn-_Qqx^8Hx634Mk>;3;{7L
z>mSOn+D7~ZMce*xC-de0X>_jcp}CRrgkfu}51UIJegBya`WqraU_!l&rA6RI9i)fl
z0WEDwwH@56w<6^u0IgQV@j|?iv`;RkSR^s;U?6A-&4?@|B^6=+<GD{<GSLe+Co;JR
zlsqhtD*a>8Gwt?yj8foMlWAFALq9pGtF7}B1~=A}Cqe}O-qeHA03OWG5?_c{Uk-dZ
z6~J&aQ}MiWc44N|BOpgf0t<Z)=5=%4D0_wt!xuJig-`GVPRdKP1QriJkQmTdj481$
z@?~u@XI{ITjnwCkE4VHL*8Z(NI&Xj!#`n{b{zuyf*PMU)b+j9QRs1s;UZ1$Xz#DQ!
zK@TdmfJsX%_(%Jg{;RtXufH9d$gHv6!OTwAhM5;o{<b}(^w@hc3}wx!9KT~kDjN9K
zsq5gA8^Fp0NYXz3UzD-nkuT$SQtDL=;~<x)kRDMdB0|LpN)9*hzLDT26o1aG(e<5=
z#^v}XvuP0TZv}iZ;}t6`mAmEO#To1yw<rNeFbigZvutlKo{Nu2Nstr~K|Jc`qrxeQ
z`2$zbR<u~2&?@}xj2U`A=Rv_|2*+x%wokDakTXFCK~1WyzIk-F7S$9guh%Uz5D7eu
zq{b8s2;#JR6&&&-HJd2xC==&<6GXCXgh)cdhe?)Zv_VmseY08b@NltAZOG&Q#F)0W
zhnj$?tzl8YO#8bs%+{RvHj7(qK$Me_6^DFs_n~JCg527dS<?QMLQOui7Ur>$X^DLT
zOf`v3Tbpx#=}zu|N{+QfIXZ<RrbcQ?1?N3+PkY~PZu;&UCcJ1P)GNnyl@&V;jsG1q
z$>eucH5Rr~DMmvsUSVM$Tl58s^t+}_466t^&|2YWrsJMi8fBvhiKl@w@sVXySLe|k
zVyIihk}7x7;SqVeQ8_W8X(mF7HPA?a#sKx!GpW&)TKd1X=MeFdDlPq{mi@CPVbtp3
z{Ak&?s~eV)_iWqMCK9%l3CZLL2Cz)TuqOj91UZZDV=#x|dXib8;26m<9qo2csmPwL
zy!bDne(Xo;u*1b{nQdjxsGpj8zQURePASB-TJ(BR@0O5VYwJ3hxde?+IhHJdaJ6Q_
zAspS2M*eIT#jaL8B^?Fj@*JTPVKSE5Jb|r%%l656sKUnE)6Vw9g==m|&1P+$35XV3
zWSWa?P(j5McH!Pi;1WsCBB7=$2A4&fIKe!AdbRpxQuBs61_FXWhZ}+%O3GVmheix?
z9I6kE(eea0T`PCJqAJ`MTZ0$~kJz>di6u@x?fZu8!9h+zW20<ySHR;56*NrFP`svU
zhb#cmA8-j8@?WvYAWKICW>j>)$l~J2H<N#dCFPi16_=1V7~&R3LR4Jz-F7%m^$7Sk
zoho~LgoFL<$ksf+h<K9AoYSY>|B6HnXTHf1MM|T&xUspOT+oXpmoOEuZ)Bf`s_4bx
zH<E5oX8cfcZ_)o*X9L@eG9ev-0Q*RZ`d~dEG6^ZFpw|VkICo&JHMXeQtk<{YS2`O@
z$$K{D1cKBtDy9=(7B)R~2^nQwRk84*k%*M2+DkIr3J|dgQ@s5o*L(C<%3FA6+pbp!
zYMxyZvVC*>&+ORH;v@j>!p`}_8v3u9r~tT@-j6w~A3jZujZH3hh*5IaxdZ)7Cc@q)
z^=Atd+QQR!C+iXl*<B#aR49sI9opvwPAnxKvjyvbu(~Co_SQl$Dh>i9J|aetS#-fZ
zf}((l+3wjlR6@LiOIy3@nml5Xipb+(M!iW+4M@4LjED$?TQs2cIs73k+P}(N!${57
zLwEP0Jj&|Y{A_>skO%`Gmk`HK56dQZrEelK^z-<ydJWyC9#FeKoxLJ${VPsP>N&vi
zGpjF-ifK~-<jZux!ENtao+zy}A+RURnrJks^JMVm-g1QFOc^6?6MEalIcG$D^b-Yd
zyji$90pI;|`~(CH0Ivjjq<)c582izLdLZBc_*F%~c63ETR#RJASzTNncsZTRJrd|~
z$WAcOGgMXDOrQF<YjP<N_4IDl9z7)%(cQpvCaPXqSo3*7uiGKkrzSm71#&UvSO}<~
zv==w8pgH~fQ}_?6j+SS9K||Ahkvnzks-UE(9FDbu6F{E4un)!!OpYvXPmb=+3&&9+
ziWghJMJB<RRy+k$nMJW`o#%2PTvt|&?|{->A4q3r;SpdQU;X-~K;2m3CMheum>ysL
zGPJ9kWg~%d@lQ!rslKiKqgQEuM6U-f+1jrb8nU`O`vSzjh&kKT?P#(V1?LTbSg*WO
zmGq#b0iZKAHE*OwA-`ZuXUL|o=!?jU{p|%MqgH;X%gN)fv*!$?ko)dn+;Q388a^O2
zuS95K=d@B&*s!=~4gc`dh|H*nNGK3!Kb#ic+S@s~I6e=2e}RC+iHN7rtXRabsVpt8
z5}aQ1y%>w~BHI!`=B*a|ls<enPi7V|qkn*T22f;r1&7tb-Y5|GCl-E~#T4M`Q+Vqd
z7I`TCTUrWqm6XuNgS4r0u8K^AqLw!VCv|TJOTL8<lL|#EdUn1%F1XJ?U_}>Q_X!|1
zil1dwtDcmMysG1xiTxU{IL%yZfAu8{On-*}7Gy8?WOB)}wG;3x&8vGJC<7@m_nUa$
z05|)M``vK^mI4t967t<<t83Az$!*`u`TnVzdA`#CB)#($11&8-(RUvj1BLotw@6B&
z;#W|Nif1*|@1A>e%_3(2i`@6E(4yfW;3OxYI4&v!F!^Idt>&bDh=qB0SUn=Ck3mF0
zh>3=Oba}XYetdpPgpG#z^&0_t2Q%Y$fP}?DiS}i<Ay1Ei287;OZ6JC-DCzLBLq-&U
ze&vzHeoGpA1GS8MIJZ1nul8kSEYU}cSfFL~M|Yx^NELIG^v$ohZ<H4jQ_(4uA?Kp~
zUI~hylKqd$1e0#a!^5MhQo!v6EjGk3vD(^|cfZ#3+LaZ>fv3g@@AQZiC8Y(#cbWNk
z2&?_%7HS@3#@yJL;36k!Fu1)qXjoFR1aWCLyxKZWcap3OCRuNwc4!6?W`@c7MNwcb
zA6?uyFDt63CpBS8dCciVk-mvx&)+yqQ8d_!eM_$7B=OHh_~LXrUP<tsVw(A;UO)Ef
zQjui<8i4GbWwynCJyTUvy<7&GJ&dolVxu(@Q&F+-x`<i%X+23^I+d}QCPv1!x=FqE
z%kd>WECv_Pn>ro{x!2YfeNWZ~H}G2O^ms%uDAHh&GUGn}W`WXVo2|5<vSuh;3s(c?
z2Vs>xM)>-<<Spz_D#Cd96HhJ6-}dRZJK^{*mLR$M%SSQwwU7q|MJu2=1C8*@p6l}}
zxGW89mdE#Fs2>z%X*#Ewl0kwE``Af>E?PZl!7!r;X-&717NxKS9~RXyQ(EwUpnKHs
z@w|ScA%EJwBD2@`x<6Q71p!&Q-9u)-|GxP=dyc?j%4-)?X=(eyfxu$ud+~soYY{3(
zrKAc0oeVlFk>&%Y-vGnpUlxaZw&vvp^b`X;@2d(s2xur|^7QGY5qycNr?f3g7a$g@
zkg@XL?GRlKZRw>d-J~vlI^*M=#zd1hpPDgkZ4H%9WDcqf>Ax-3pu}m(wxTkK>{T+E
z-7_Dk_v8=TUofTLWg2IaHqe7B(&N25IT4)Z;2$0G`_{Y;8n9i0QqukMp!U;g7`#Oo
zqgzTG&T|&uAU&POpgAv+`BC<c&glkDw5p^mTv|n?AVAn_D?se8<8O=M-p&8quANl|
z4u#raL$e@3ZCzV4_UoygUG5jX8q8*S+YJ_+!3BxQ`<Sl{c&E%C7a&c)fSz(9Bd^<O
zq673yizxzt`lt#4nfg?<6C)lsX4=dFn6a6KzkRY2ngKb13;~<u*K-H<fv_RG8p=$Q
zs~SFAS2K$snPR%k<wHe#?&Bq$Awzl*3+td}^si=`%=fD?8&KTBh0pDLBT(=*p`@d|
zM(4c4<MqO4hda9KAkFm5j{KJTno0PuI?vNs?*#%N(r!LpQx(N9A9&IzkC-Hs5Z66I
zgRwvuXCd1o6L5l+fmB&Rn9TSf_+mT?{{8O|nsdbrG+2qQ^~&#n`jwDm<&@V@Jzwux
zpxabk)0XySL;aiTJSV#w;Xlo8S52MR2psC$4$r42SPHY0dCshk`yef`w+B37m4Nq2
zWoCQ;cG#i4go4+NgQ6lH-H1$7X?xGw4jJFeH@a+7K3nQJ!-`rO?$$B#9$Q&uVY*L@
z_9fI*Y?=HzX7wq2j>rA5fK+36On{xF!r->L*m#&vJ}_?5CGi=3BJ|;6O;0tPS26L!
z1&=eyt}FoEaB3su9Tu4ogsYN&YyjoP^m>4QfVn}eMr1eX0)JOzQPn)~_GFz~a%->a
z@lpR|pbi@IIBB%wT^XY-4>e?e2pEt~rVq?ga!soQ9P-+{ArdNiFkB(SL`gc45x6KG
zAVh1&tJnKXfnc09Yo1qk>jfVd=&<En2aI}k%_cJI9|zG0RLJMlQ&Y0Puf*SV4FKj;
zrNaH>3w}>z`vDAJ{N7upY($ieE~!efnNbwXZ_{%h$jNiBv<S(AKg5Dypr1TFS|L&)
zy^8&3Ijc8t1Q<G__AyZ@rLTBj#LYY#K}$K<v`tjUih9c2-#^uDmN%wd(OAPL&puAo
z8FLv6kGD3~@A*zGD_YcT*44U<4}puW*FJ@(UrvAP7Mphsnu9(KlPj3_2)?^}Qa$2D
zF}-g)0B)<@<t}vb%)y!gp#oaX5hO+snQ{w7?j?cUmDr0ZO)QJ#I}%UC!M51rHMJju
z^xzgqZ04UwjqblDC!AGH*e7%Y^HeI#Gy6hcm~wdAYued?{)6=O&}f(JO8`2qePXa8
zd0=>^j=m?q`&nm`IgH4ApdO3pe=~}Y5VS6yy}{=*%G@8!+7E;47segZPVs&9EWMjk
zOT(*h2zm(P!QX#CsUstXM{nO5Gr*d1OaNjHFDb$>%|W}R#Ryh+MOJ^^2!+)-9$<Q$
zE+3*cWDXT_tgUQ$XWfCL-VYzI$FsmpA{z?<-~4wIiNmrv^GR11{UHmg_rK^m|AaOK
zKO_&|mp1J;`tpogFRhhzE%ZDSiPS2#jAA@pxPwq4xcZs$Dy=S#&Q0=40=5^3`Nv>k
zx|pwml%86gf|3xkAa+^d>y>2AEz3&Ci!D#k>Z!d%@AHx$*sy?FiLI@w?COVtS2wlu
zWv9}%caX}^Qd{2lP*3{vhF(fRZj^DOy<~sw9+v(dl&=%N(Ce6<sqZ;{qPeobHk>KU
z*}SkU*G$1VJwM+?=&~2eh2`y0%{v17G~h9n6Id%ZhSU`A788<$a#w>Nt4F*2obi6!
zgytyAkrnP^xCVHeEcH*HWw_5F14RSf?C$UNy6qoShDRUJZK{T~(m^nB`OGtZmjinL
zc-_vRdStg(fx9|E+`*6QnmT}Ac$O`Em{Y}0ALYRrzmj$=^46`16RH5>;b?(R7_3Ax
z5lh!U;SgS`0`Fh@u2>b~uPtUPJS7pG)GMUW*5W4pHQ)w`Px=;9;??gjy(CfbUh<{w
zIk)H;EBHImb~=B3Abv&&H!1zD&Xmc!*jnN$^Ka-r*vd(r5!Q39+t|Ej-P|-KuBOue
zdVPV9jA`^bbR*on#g8!!_vGm8VEY^o6$5uWhGYv3V&E6z3a5_?v|~0PH`+S|iH9J(
z9IV(O3~;)$|MY4bZJbWI47gj9<HO~fhK=lg=*yERh>-+5@Bf2^b}(gL@`m0<#RY;S
zQ{A2ow_O;Ql)OsOshgX7yi8Z=FaCQ36ofjZy}s64Ms#2NgIKcJL7V&#J@EK+Ecrh)
zw?X?tl+?3RqdMBzzqU|ed40iJWph{^-j;5GLL{C1<f<CJ<_>x)o@FKdWc9^f9(JGe
z#lVt|=|Sk^BS<&fd)?Ev`SoDEppDa(3uNS&@13A)`5z0NdlmI$8)}`d`ck7||HXzb
zH;F#k$5)infC4Bh8OKFCvGcwn?SAgdCuT5bo98iieW4lmCm2u^EyR;gGoTz!k6{Z?
z^EZ|I(`qrE#upQj0g`TE`DzrN9QpJ-N%(>T6=cCv2ZbGQp0V9FP>7gSD5kRc32|^c
z-5y;(t1`Ik?sR)x&vB(6CbZw*TQ1ky_?Yc4H{OM*4Yq;+|8AG~`$u;FfWj>D4*}Qo
zvrEfzUD%Lrj1~Y*8;uFNZ+c4ThbAyJCg5%u2HTWp{>!DIj(+{gZpeWq^_jA!%J!J2
z+e%ctu)_aNM@qM(mO`v)0;}e(moLg!in5lA_X%``yt~r-yw8p}lqa-W>nJF<e%I2F
z@oQUSo@-!PeXJC5PxcYD1D%TMPpYL=H@lHA#|2oxHYYbsBbIb%4$$eU00M)FjKd^s
z4oAwdwFM_AExYtDI^_#!iYLDb=~kCzc|GNy-7mndFl|ya@$eRmAR66WkBnSRPxr&3
zz$o#%UtMiaPhVY)j1)-35cRw7;AA<)J12*;6ZCww)4U^G=LU6Qq&G5SmFg+Q2pE%U
zqX}XKI7Icjl01Bm8yyev1@NO#3JgwnyLXgGvTMoI(_8ijJ+LJwF&11A^_%m$yDr7`
z^*3Gae(%=!UTu#YDSOA}aiup8bPL+Q?;JM|tuK5#rL(k0w|7D{CFHOn0>r<q1t+)1
zp9T4~j_=rn^41`4JHjsc6sWJBz|RHPsJh5mXLbb^pfM8+%?inx0o~k)c6X5z!y07D
z%xXh0zdyD=lWlJ8ZspMg6(prqH~);93w!DvPZqN+g&I78xTKGYv0YxT2luBS<XD&Y
zvyp74_wx;i0Dko84!MRJe;v~(dC$}nDaA)l`28sbbYR*SyIZVHJ9EB`faV%c2yL`a
z3)$VIpIlDD!L%np>jX#PQ!#Lmv%O#+PFE^;yQOXoab!%*sVO-}LR#Vc`Ogc|?Zxgv
z_q5CL_+-LAw18>z{RxjP;rSaMRlfO+^VQH`biu3TSj+j^rt{8OQ8&2rSAH+(daT~C
z$`L8}+J`9Zy07Ft*c`E&M30RsHS%{XeOe6e{ekLF#Vfyoo!J{&-1d-SwTB;w=8Kyg
zr}{4=*iZB9lh5_I;+(er*C5vcXw7NY&}=DtQrfpVMemjt!#yP!EyY1M1;7)3>GAy7
zyS_?VY`>S;{`IY&b0=ZAf$bV2{ufM5P`y7A{yjT(1l*Xp`y{^yTaQ>UU$j2p+tuWG
zFT$m}Efotq?eD`*3A)<@TOcY0h#S-8aQMOO3i)x}kFt#Yh~ty>es#88MOUJY0&0ph
zW&&gMv_nF}(|y{x)cLhhXDHUW`6n{JJdi-0$TwyulxRBT`s3qSG^4?I?QU?NC5n%k
z)<L+oh3)%#-M`#ue-_(C=<l1AY(Y9_ZD45asGq<-%lmZ%NYgI%c9U7E8lE~@yiLeO
zS1`|(e<iy8D0$jXAp+cQcScA0y;`O?)vDy+T&KU5gi)OV7Fjy~ZzibwTH;Fre=LB|
zbPe(|6Z&7U3uePf13yI;rzyN0wjaoJ-6q+s#}HeVP}8YPx3SZ@p05Vu;__fif%3Mw
zm=kSl*td)Rj-{!bNd8v0SITTLL9ctvzlwp>Hs=?C6X%DFkLyJjPcinxkIwkZ-L}x#
zaO^o<AF=o;N1fgt4aN$iv-YqIwVaEv@IIwEp$SoG{>Al`{4fo_EKbYmB;(QUr}_-w
z4iOahw8B1ba*uY9FWha=`Ens`fl4dhRj!*LLe6)&gS1BA;MXVl@zo)S=EcGbl`b*t
zmXe>ap#`Z%1qdbhT1;zDD3T?Ess!uu_(EBRDak^1@iB5V8(s8IgNtpf;@=sPo~ge4
z;uf!}v~FjWKA*kF`;DAy%(Q86GgW1k(k>5f?QXyKF9T#0K)Q3u;dQMA2whG~I%*zJ
ze19(F*x4*M8>(I>XGNW_a}joXU*oHB#1KX4Sd`I7!ZoVxLt?a4*ChpZBxEI}fy{ce
z+zT{)^<k)v|GiOMS?R{y17K}(hWjs#hSPHIEDg28;7j_D@>0n3<?9L38*Yx&+Z4Td
zUd{*0<sQoW-P~hMHX;BH9IS1h?3g0(kE{4mhv3q2-ceTShaL<&DK0~+DU3sLtU6r5
zM`%ZC)vdfUxcI%ZB_BRV&-8jAFPxDrH^jNT{9P(QvDH8Q?$ZfZeM`r6Dd_uteSfr7
zMRy1)Zp61e=ihX4G5~Gc-)x<?PW~-I@{LuL{bPDr62Qgr3<EX_okE-UhCDK8d%x((
zYl==S&qm(-`Kbosj{c-t+nWr)Ybj04YDj$R6?I}(yAl+jjej0(@JBB;wmgMR=YW-e
z_@0|p1AOc|m6fng6=p5e0F>7QR#)e@phghSnf_JfxD>P<YwD>-^P!Ox7$@ie_FAZ4
zQY;-dl*Nw5*T7|MP-A40)o(7FvW;MZXqQ6n;ae+iu|r$CbFcdsP<kfb=cU1%_=o4k
z)3(6EJ_pi73F9M5&)fC+T&P>MOOB<0_t{>XjE@^(bT`CN-!RejUbG___u7g-qdy;t
zI1!TQKfm&jhjogy(zJv>v`c<AP{$ldNJuU&F5f5u5P+|k#OTFSKB$drX(+#3h37)c
zP(F_}<Hi-zp=TR#VGgoERe#NAe7?`$bIJ9)-nVzBfH8i$d*{VZmdh<#1PHra&2^0~
zM!Lx<HHma9#`eNuSrbv^v4K=2ETU{@sdt=$@1jETSSBVni+fm@RsV(h*|{jz)@4-_
zceQ<yfWy=Uj$ym5V8(+9t0Ti~K`Odt|BCro-}Oa-3G*7H?x*9|Lo!fT4UbzrG-ccS
zBar~?0Nfbg_{k<e9!PQA=O>YPMB&HwkPKvxR>LZUUbPB0()B`Bf$2<&XT`=Sr9mHd
z{L}pL=F~^2kZdblM^c-o8oxCxH1qjTYrjJUOr}!#b)D&eU<G5YEqeI5*H@LDz89UJ
z&p9_Abd!7<V*!O^FG9!*DK+vDcg*TopMvMQZU*zJ&6q{O=^I{`HnWsnEomk{DSduQ
zGLaX?3Vk6L5<$_`ij@~aJ(f&iBqp8}M*RV8O@|5P;C8?6baClsE<*a64Hwb9ewzXV
zNU}bzIgYt?D^xEm8xIvigdKks6?-<f{q~Km<!XL5B<*iX6N9JEMdhMJf<#hbgbzWK
zs@~kAEOpL?@6*99RqVjW-O)@8dF<!vl72j-*ZX+K>LO$Mh6c1@U&o=X&+TJQHwEAG
z>_2p{A_}Z`^-UJio=snmH-RYgYTeHx;9wXuIKAH?S9#?GvDau}(lF;+CZ%U0HWYyR
z-{dFL@J=npD3W;brcP(r>$%woU(htVp8_j_qa|Q|QlV@vXO@Q}ES~fDphv%Lc6;6Z
zm}#17@9rmh*W;dgCXfV`^I!HxBhej1*5G2o)*DYf`vKFnj1|R25yoJV8LhiOWm|eu
zOku9?LAkE>hRy<~o%rA7l^sN>TD-8Y$6l0b;~K!-gEHoTeNpJtg&u#m)qB2$afrDZ
z=vLQox))XR;ZMG%F6jNae0*y^1H6WT(l&x0B}7seq1$VRIX<3`2_PeLmY~zka^Cx?
z6n~uT4qWj#GPAuC9u=3NpCSO(fQu{IK`d^-IX^-}%vj?1wwA0^RKHBjC9@a!)PFh%
z$#SN{U%)n&aS*5A`feN@e4A1&@EK#?{l;heNo6(k{pDmbi7nKvtI2XZafae}=KBMT
zE-IMwQYsXF?36HBaYRqSF>1V?fXzPNVf4AL)u6ZEbn#H{0D;uAh7=H*Wc7jUv4)ic
z++6>Iv}4Fp)6c|BW6AfJS(T{vex523{5bDh&rqNI%X7wy_j<U*<a6~A-v8=m94NkH
zT*dh5<6hK=N?gO~SG?vVyRb$U@BYh{uhIxtQ><|3Xz2I7n5j6Lc}P(r`?yxy`;)5U
z9+CI^)$ni+zgK++>j7Y<R&A)i{r<5Vs12&2Sno7*pxaUxS*vn};Wsz`6o%9&hxZHS
zL&;=?`EK(nWMQ$}>GIm#%#=0vQ$H?c>{09$$3yxzRkMxYEslP8GLh!*_DTj>MnX0x
z=+cle0wA0FLl}0re|UHqrsi#xJotTUyzDWWz}M|FyUA$?lL~-Ywiu<taW;6v87h3I
zxiPcXmmh2Gr`n~Km>-8CnSG06Kq%&n9FhBd=WtRj0?p-TbB>D2RoNbFazizx+X;fB
zzt<3ce>pB0R#E%G;`>R^;}&_s9Mzw^CVHc|(e3W|iMjK5%@J&21YkK$9O_yC)2WET
zgsycMIj%!u#emfz7DA4LKqUV{GLlRbl7wP(_YkVGf|C4tWKu`%qvR|5mE`90-rQ78
z_DYMzzcV-v=v<AK%ke6^>{IuT?AMRe*Bo!>{kK;TXp+zKZe+VZR4A!{pJG`F_TCCr
z-Rf}fWcwV{tO56Xz^|ShYrEklIr2^NzZN#if>zsEAazVc(>th!PF~*Cc<`mSGA+BV
zEsB}8s?~QT9*JB{AW#vCUePE)wfft0<R`$D8oES4*in<$;`Kb=XJdsmmC0PMfr)is
zYFD{&>uY|p{US>ohBN20)N(t7V>X5dy>2ha<WDwPq3QmkI4O$a_$)f>lK5n6`dayr
zQRh7j(DCIG<gT~!Sp`*UP)+JK-Cu>eb*tLHC7Ba^P62Nsb+F`tT@Kgd;i7jL1U#N^
z1B*q>xEGb71y73M*Y_x20=5&Wl}dY3Dzbv}T(MNBBZrVA?fvQtnlq($oHJ$<lOUk`
zuK$tYYNyz)bstW99WwejV;QJW*Mc2>m_siCpAon6POZZ+fcw(~)*}cw9X_(stch+J
z(SSzwt}8`48!YbTR@E@X#g@E-x8fNz;2*E-qdLlr4+0!IG&Vr&kXyw}KV|+7v4~jN
z(CBu&UvaSm{nF#rC`z^O+w)S6&UYVsZLihJ@P?GGro+7?g`QW{zH2T)kE<5s=uhBv
zjtY$D`Jpchl%q;Obe&5k_+rHG0>bgJSvKTDg@nO~@l>{QhyOss!^1<zIXm@lwVN&>
znAVSj53bWYPF-n({V`;YVKUcelLuCbN4JP)#yPV4Yi|oZO-E6{_h7ck?tIzp>^*|(
z;%6+k5FB9x->3eMTl>Ga3a|Y&M6rPS#fg}s5oae$M_V;J8w>CB+=i?~+uxQ-FY!4Z
z$2|_dqCb(`W6Nq=2&6E%L-?VdaVjedYvc19WX#g(syge_?BF3%W>bS|eD+)doo=^(
zyk1+X0AoT?J;C8?>;*s}|D}UVQPbi5rU|d*Vg=X(4GXYKQ#c7!56l-RzW}?<u&2mB
zZ-$wm$(Vi`&v~T7y~MrghaVEJ$4h8<(RCduc1_W|%(@24N9mCk9K_&p@PZs0CcSox
z>D2Zdmf;IRg}YzP)A4mew9nG!E#FpUcCYE<I~NLSkMqmZMN6_MTD;g~YDq#--oX~(
zHpkg7{Ds;OwL{T@5$3aPB7jy}QZ6!_m`?5|b?TcKnuV5tLi|2vG9r<ZuC`!vnTx~g
zgQ&UP+AFw;k^=s6ox{SAZm7I|_w9bn`1PgRwtG5^um4ELddBKz>Qb|3f3uH-ynd^L
z^;|&lAMU(kFgZ3VqSArLvI<5fnvu97$|3P!7%;5gZ4Ol8h!1V{4?y4gDv{(`>|j+#
zr8@p*OL}p5pn4d20f+^d$YW#0dHMb3WB-)PZc{7_{Jc#tt^!hU_;N4qKSFWLCh&-w
zj)yhX?Pl_Mpjx9)28&$@8vdK=qEC9WA3Hi7ONv<&S*6*$qB`d6W#R7eAbt3C&J37U
zC5Sjc@#}deW^#T&8lOsTbYzl&zAC5rX+nUiW*vCD>dUe^R3Kc53f(?9+<Y70`w+E%
zTaS)Dq$$7D<8!mVOnxYGqDmO__L`b$Q2)-;)m7ETKC@&LTis-y%3f1$@h~55d@7!}
zHyXdIBr!iBRV~Z(x&N$d^~(qb(v?y1`1YD?Mi^V8MJ*C2@L6nden3tr+*h6Fn#y;A
z#;V!K&60#EGQLXR6ucx)2O`1vFfw^%{$!k6VaZX2XR-$@(0Izn{n$3XE23agbWlG*
zC90zyngxmRQUs&$fkVAQs5Gn%&2HEB5@tA9=EN!Rh*22BfkG$sHpA8Y?k`7nBM0_7
z>*(f8K6k)z%BFxv^)8D~k6X%u$FOWRLPN`5ICqcN;Qa4BunLlpXz=*VayX|Y`Yb^o
z(#Wk6sofu6`$~ze3@b~(jc3Q1wP&Umxmb(mq0CGk^Y5IYYOBNB>x}r>AiiXfK+$zn
z+U|@uKUBej;UX_&GJO+Doa#Mpm)_0(X|T_w@8b%b-n2b#tasf{6tNAfA5uv?b8BSa
z9mh{C)JPN;@Zm;m@xl&uCBcYp7=d5%cuay=8!k=rN!K){vqrt5Mi$bXft&;?7%wt1
zxQBM`kF!{x-Y;J?ahQDUk6+HGx}ju|UEUU|@uw=ju5n91WSBZpEu<v))%hIGpx6Ok
z`v!oQm5XO>RehQE53T1Xk8>}s8k-EZvb=J|<ap|@VnI+(?ykw@TZ1xbH@WH2b>DuP
zIA;u`QV|_U#E{qTtiUy-bhq7(c$rW3EA~SESYVlQ>TJxUS+ptL^-m&rW0WYKvZ-c}
zMt|l>!!}T@0xq%qT6d~{#7~0D5ZenkJO%+#WdHs-*5TiTifC)V-IdXw>++ZVZmg!P
zFU>_QYpnGh`h7lJ<MTY&HTB*9VWkKp0B(G~bmU}@aU^Z&tbM+0lSLOk=k$C$+&-H0
z=nDWRBYYWZOqN;P<@_!AhGVB+{(SGWxGt2|EbNFY$f52sx!Z%b@)w6xf!#4KAoBxE
zU^$80cAi)I!7j47IN3P1p(OK8BO#~_)-_oP+y4yIqBY2zQch~kcHd{c4?Ox&e<sFU
z;<#$Dq#fdN>Ck`ni~b<rXphdM%2;C1=e9Dkkt!x1m|OJAZp^msXX}5(FYby9LTk;{
ztn+h#;^pP`yS1tP3d4YJ5*0*9vyTA;XH(Z&EO&T5&(UhWR->y|`-87NVCwh^@_+0v
z>HEHX;L-op?QuC&{$=_hNhPT$aW{u|n8#G-I8!NQY$Y99bdVfF8PR5TyAdU$J)C4N
z1(7upn69VDV>X*vwkx5Id003ETAVy9-+0Od71G?t$>do5I#x_3F<XotNhAh@J$o+b
zaf$>_Af5Kb_t+31^FdBEtKlh)TF3R`bFU2nc}8~pnpRkDbl)T0nQT(*2(RTHO04y&
z(-c8|&{X@8m9f}eXK=9Rz%j_ovumB#@5KL@*zzsXLd+0AZ_Q2fh7$BR5!tXer?M9;
zWkYA8bGqv|P`ipbg3K+JNdxTri`vMBCN%185982FCcW*!u^HV->hwoNk;_g9g3DNV
zDPCeWeZB2(-)7u68FTo2j6~%$yXnLEvj$a>7}&6?HJE+8Z7k&D!z>pu)aRqCs;I}L
zWKe_q=8Ngd;TuY`bD+VJ$=uf_m|ww=@M4og`;MUs=g%hNuZ1@JnSoKrCB0b5bk&~%
zt~Z%te;2LSnqRjZuycPP;^r(0deFAy0KNfLvLG`B1pdcF1=i%&$a=_v&21jET6PK{
z);3iw%TSn?kb*%!A0DcqP)+z+$_Qs}6*7a6_Zf{<QBy$%Ui`!t9bE$dQnW>lZ>80W
z9axue^$e>1H7KccMS#hp!Y|!Mv)f6ZP=!qBa#+G*uq~zddV#063tge~%?`I5!(`vr
zi;JS&jzxxSPBRm+8Rpz+r0{5k2^vDBCGIk~?qoWd<X&<)Ggn(-mS-ZcbvS57`G@93
zOtVCPFJ^Y!qdSzg$jV9~xTUqV@jqiLn0+4E-@(aWA`^=JT>$gZCrp?kxQ1F6P`j#p
z>^GHp4!N|ZttbvKYJT67v**8idXRe@Vn&V+K}|J=*qQW$=NU^(v0&~~%`Gb_OzCVc
ze^M>$&xLJu4a|#|#I_cUVX_#Vy!^S4qW#Bk-@m{K#1{yCenrB7Jw9Nbo|&GXU7T5%
z<6<6|2D<^z?Eqi}l;YSMa*b5x=6NDOG(U%}0obLB+O{SA&+EI<f=i-!8{g0GTn@Q@
z6@m?J?*Yl__-M6*;=$0*5CgQ7EDUr#g@vu9%z}bUTvU9@$=T`g<Z@x*@D$-i6_6vd
z#5R-=Dz#twCSG~KIAJY@spA5BMUU`M5r4FkA48okrUMSyhkqpIr4C&Zs3vr3t<Q@I
zyif=!NIsHch^tC6@C1-UQlb4ejM1ESPX<xsAxxR>Ix>bAqkB8(F%G&7-%>h!^n-R^
z>8zBmZ^8@&nebywRGG!2cg}ZrcV!Tl^9iga#Y+kbCen&|Tj)alV(IDq*a=kfPBXBr
z(cm??p@B!bs=r2`;W`d}m81&q>L{bPF|98Y2^9_QC%opcSX@l{@x*`DR6Nyeg{3@9
zHHDF@+C3q0;IGK=j!sX0DF+*0n8Z3mO1oV6Zd^UA{?ka;gfJ78>{fM8)QBghNP*F4
zS&ZCmHc5i$-39x?-Vkg1oP@`DxrSi(Zd$s(E&(ixIF9!b&iF@!qVyCgZ>M~B|Lh9n
z<EpaA;Ia|%QMK?jt0!k2$=EMhPCEUED_|#G{(>p5*GEdj+=j4@Z-Y7-F)LpiCKSDM
zdU#2Mij9pCgrRZdp<&OkWefEce&CCrPJV*9fr-w~a(cd++Qv?JXgPV^_cvHyOk7@j
z@^~OP|Bg#Pp8Mh8o{q~Fc-hLh^nKZHVPZVfU)4$>fL=jw<|OxFht?z8gSeC6JpWoS
z978)5+A=i}g70iBD&V-Jh^Dj^qY4p>UN58*y&oe*Scq??rsz$2%%^w-`JPCRSLqcY
zF0hGr%3Xevj^CsOHAG>Yu8k$s^_&&x#|5-GAaHkfzZ2>p%a_<nxRcUKf|yfxThUQ;
zSWTQu>X>FF@%<N<#{uAM`DC$JO`$TbnMjIU^$rMO(&IZPb<vy$;-2Z;a>EEcFi<ei
zp+O-b!Tpw=2H%Dy_0=PEYD8h`;fnlE2Cnz&WMrk~rQ~F$VogEffo9Y0Q_`#8Ex@x{
z!#~$QRoZxSegBvIwCnvjW%Ik|%W78x-%2{X@tPlBOo!|UK!Q8IBA`J^S7wzxr=y+8
zh-}ZYlsk51GOV<X7tMn0g-xx1RA_1B7)Vk;5vlqT`eRT!VdOH2<nNgH#er=#t|@P!
zvS*;wEW%tywO;O6rn(-N@5)=VDsakm$!mXh7wT-Ci0^UtjDnFOsZ>Ud;=6OzorR~?
z`Mt}*M5-AB-_QE{-SgRI%b_y`A0Ft7mP8>{+c$9}?2#+O{DlAXJP@~MbC1F8NP`bo
zurDPz_7h%oV3gSUD<0YJ--)Wt^x|7`=BaY#)W|h<gC|CrvR@sFf2Vo5YXSdS|3x<K
zdk}?6eG&CO0<};CLdlcZqnMkKOzGGXYADbXVC~)?Ln2yvikqpGmM7@`WV0PR3;w&h
zx#6<)grha~da_D1eauV&VJp!%W&@@us?uWgkyw|zf&1l~ma*<V2~+OtMqlie+}t!g
zw2dhZ+>ZUq>CU0KW^jPcO&`!lvNuikmAxEAktD;n3kTp`AOrcLwgrf?SiILNlU8GN
zyWKuFy1a4{43J5mUmwAXZ+T?3GWi|B;0eu{;(I^;qC8fnvg(`|i_^pzPP<ThG{5#W
z$eG#=H8SRKRO?`#+Z3#eIi({#{4Bp=b}*uEwmzqibxqXi-+|)35dwtMAQ#b3hb10;
z5_21ovWJt#>B*;N_tI*Zbng$6bwMAayxX0n8vPE}^Apw7tZw&hQS4lalOZIl6sGKd
zIv^1Q6OR_dUWDQO7uG{C522K9R2|e;%0C@52uc;+^5fK-{s+@2`FtiF4FRmkOI;~$
zQSQGh=jTo2$0@s7ZGqX-iWWj%w+s4us7zjulXtem?@w2M2LB!^5D7NA9UdTAZFqYP
zX9o_;XLSTQErow9_QoQxPo<~JD$5V5pr-4vd#=nb*6pXU_(q51rrjgnpf+#|kU8Ob
znTNrNlKvQ(hJ+TF*6T&o!+@;Q)e^_lP^ctI$UEV(4}*uhi3DV&67lh~Vzae>TNgFx
z*=g6Atq--dfYU8@ToQuTi@pAkuaBTIT=Pf14%dt2PQ&xx6g5-X(f8$v%ZfzmV}q6a
zSJ2LzeAsA#D?IJ(Uc%+*kx*{r3g$d4yS;`#J|{aT5x@B#Z%S-%7|GIFWF8vN-O%9`
zkL!2)dk>wjHUqZBLCk!c{~9iP^L#p*yBE40kN@yU9=CZ6Uj*7Qq-mvK5gnBdM}mtc
z%IO3aK{q8}DcDg@B}2iDbMOX^KXWG{h>pJR#pptz`|#{Kph|r;LYF*)`|p(|((Y{w
zC~R|A)>hUdOUn4$`;5-+S1eZRc9>11gBFH0T1}3Fz>TS0XI-PsHi%oRHN+KJZn&eP
zqG3^srrP2zY{$pq22z#(!9J5RV1%MSIm;jIZ!_9ExH{K1G_xP>@WS6+FkE8R`+vLZ
zWGB&saH>L@sH$P|DzF*Nhpw5*;PcuD7_5ak<@B37Y?imN&HQqME|XlP*W>=O{JXN%
zfXfL~&cLRuTv~&7wWVOXr9E0<p~vhZ>me_aT4IAW0z@A3Ly)4Av&@i_A>IWNV~C43
zE*O0259ojsDKOeeXxHsE9P;W`0Lrsvk1xFg$&AtB=NfK)kI_mjE^x6!^l#rC&y~n!
zaJrn#(<oCaq|eW;@jHm@rnaYlH5Gg2u?XJ*m;oj%zS7uV<su^Y5G$9zF(EK9go3r0
z522gXJb+}T6OZ<OB?$qRm7{?0U13zOE=oJB{M~mlbd&l>_R@NC%EA%Hh%qa1E>4AC
zxJ!zH<;}(Au6N7V49e!?*@9Ltg>c3aODc(KDoat|ms7SlZ281;Xtd7GTiPO&1Gi7W
z4zs~q;meKTtfc-SqX8CbK79WU@ejG^2pH;YvR|{p=-{=iLL(NtBiD4U9VVdvDBol|
z`|?v_30RXD!#E=;5yy0TJ}R!B(pG18=(?Mm?Ez3{v5fTZh=|`ep{AyOK@r&4_lEh(
zIY|dzA`VA9A$t<^=Pbe{<w`lTS^`V}``(F(f*bVSMLDV~QqUTKk?kU-nFYyJTv0r#
zLb`1BoS{eXpeY5v8m$0b%P;r+kr#)1Nwri^sue_^S8cjqF!iV}OuYhAK;e>ajlS0n
z!R2|85o#{<a26x9%+yZj=Og9Om|7u2JQ~jJLdQFxBes?3eNyy&6VI_WG)Dn?wnmn1
zxhyauqm>{!QNr9p>Q4t32sHgf$w&*tSn3?E!+(~sv#_v`koeQ{z~W$tDK#;wecQd5
z>F)9(beWl99s{@4P@tkSKGF!Jq&7RyaYG)<DON*<po(T|Em)SnL|*zDHqikaYfZ@~
z1G6OH;PVshz16|}9p(xCU9<o&LZ7R@5vqszu|;}ZAu}mPh}xCF6a@WqfmiI@KQ9E?
zm!-Zw1P5JQeXk^Nk%KkW#ZQ-d4ZhzAjm$e!_DPSd<gK8*SQv6aQ7I-IURiX?DL^GK
zFI?Chm7&V!oFKDQ5*#(a>1~tw9BwyzFKQ6U7M6c)Z`?2sUs)qe%Zm_jo#`9v4FN|9
z=n7vnC{JE$so}3Tn1y}-gYqve5Ui<O=tw}stS^H89q+`A_NNg;h-(b5#BVfs71AQI
z?@*Z7;Phf&G@v@{C>6U-1JN;P=BpuzJv@J2h@`|uDE9KJ5Sg##T<LEJ&3tMN>e@r<
zh_t@e$V#KP&WB)S@oy!7IJ}Jus_2*D(BHzeNZ4Sw_oeBGiv<Xyf4}UsQFX|Q$z>H%
zG-l=585l^Gy+=hxB7$~jrN^hIr^5Hj%7Ghs^M?W``4_X?W`Ca)QLMVEvF$X57ek&1
zWO%1ocHH@5h#jOEcXMPx9BL60${>1C7^8F{gyU`z6l|v^W?@V~gb&vIEH4+Er2mg6
zn}}t(k(LIZg^sS65@)YB&IBu)3tmsqPMy;#inYeFP7=<!MV}lFZLv4MxfQ(z3Pau+
zwlX|x)!oII-YK>8eL&2;6QW4eHqR`zpt;j|U)S<lBz{<5i)og%Z;D-ju-<&s8NVnw
zNWvovNx?YSHz@{6Vu=y{O(MnmLm7}W8(70(JbJ%R^)6!^QC&;j-J`(6MMOp;!Y)GE
zDY#K-gG`3^B+@lBcq<5G3Ag1#G1c@q$(Tgn=&5KzS&0bQAcy^0*V2=rL0W3{gu)kB
z7Q(TmOAq+UkG;^_^3ioj?T=`dDdi<P4cL~)69a)yt2e+7vEv)Q`a!Q=R-aqyB)A@2
zbd_yrMG_7L2KiEkf0TrMW@Vx44mVm=%ARSF)ON2i(oY!z@XjtTnzEx*gQ&n5<GyRP
zUOwp!;k*_mY7v;HIQstniG_dJ%6u0UX4XuRlH)XXTj+Y~uyh6$u+Zt>--Ul`qo=Ky
zP!wn#G|dB;MdF^T$DTb*3t(W#Gg3c<Rc<YF-rZ;UOtrSa)9Guo^(XWkr`#L`ngk;@
zEmVlU9UX`1q>WfF7#4<bT43Ou@!?GxZ_ReZ6)_Pku)2vO=#27MT$5Lj!fuR%u|Ws9
zDn6w4mO{n%J%PDo_<!Zz*ccr|f`=!ZzJGLf>rV&3m!8ydDhmxoSeu^4<J*4+jFW@7
zM|k;sEam2{EL4WI+&m?X#dYPxsvrYXUKD)K^gy4k)bPnDO5g_nmh_rluj9i{XCJDp
zzG%k+riIIz+*saHX&ehWx~|_p>A;H3G+7n2qoq~EwZ?Sg!s}?P%mD}9dUvKYwju{|
zsd9k4lnd6a{zlPu&o>rio(F4nMjJF7S&;NLv^9l{hV@2)y{d2NmnAgPwfQpMG=bkK
zLU9bOI>=K-EOJKtiR5vYP>}-sa`u)s(}Qd4gIKeUoKH$D{c}r<8aX>ZQ&8`h9M%Wn
zMS`$$>gMxliDe5|_yW+TNm(|a62bY9I{;|f6vB8>{IdT-myn^kg&x37tenMo44EcN
zQaTco-f1xLz?CwQ#SepK%Y6k#ww=A&?W2bWf`VMaNw5ikOD8@Yyiik_Tps-?;Z*ZW
zW`+GlPKQZzn$G7LJWz<Z3uS@Nuz#C1E!R8bp!mz(g#F9+aFI@38`;LBGzd!MARxX0
zRJ|+8`~dwBkld)QzSwulbOigEgJ^Yc@147k@&|!;NU_9tPeQuf^8neNt>2NX-^IM3
zt-C3Zn4G1i-Qy#jKw{Iw>A14^<Snt3n8<M}%2{Q*s;F#CQ5V5{Gb~7!yZ!42Au{HR
zti%`&UFcrysVKyc@mK}7`P?ZKIe-a&?%uVIbC#IqlDKkVchQ+bn~j&2m7kuR8+GJz
z%N*ILy~a}L#KdBEi<C+9-J!4VGs@-oHN|&?5Bog$7esbGkh7GS*o4S%mNNq)qg*3L
z$rTj}>yEY8ibr!d<U34`*+?)EZvN{w_-SM42$c|Z6#BE#=jBdmro|Nj11Q;xq8?BD
z9nhIce{a|gcf*rCAgL^eq~)abrGI*Hago%_WIyP|w#_2mUeLcPyg9QH8W1JA!P~*e
zYUyAzf!8=5Ohqp*_3MPnL(3!jhQBm-7ILvhR4<q`HKX{LnREhEej=jH5>_a~Xs`KD
zqN?#(Iv{JI;@43XH`L464M+%%Pp|oiCPbW+Fs*un-Xa!?ohD~jH#N;X76*+gtP!i|
z+D1uDd0v+JFK;so0%k>UEC%lEF*qR!F@=a9*6xCkr|ii>VV@vlnV&X2Ck<$aGBVev
zqYw;BE*!owut7)8StUgI4H{IY6~!DIlo>Qse$`3FJv7Hf%@5`ZH2Nnya_tR&05}pj
z5Y}sxe{z5EG*I+ZK3mk0=#99~{Y#CPlXv->9|b$P`F?%T)^+YQluS;Os7a07WvB0g
zE|F8Z*Eu#6N1+)W=Uc5dbtJh(jkxeHL`F*&J(N@k>Z?%;2(H`eK99jjnC(|n%*;Tk
zXi!$s+w&LwJ9aN0Ku-QiSlEc~DV=mwWApxUU*OJBIg6|^$MHV_cnXL0wSy#^>U@xs
zp+Yr5<FIXrff~#7<kROQ(RcF>iipiDA3NiXxAvXBacVb;7>Xf>Hvt?G1uH-qMa96z
zJt!tEzmm$$x>y^Mpo_w3*)Cz&Rm#0a48&2qK-)IV&&|FgMKx|s67hc*X2hYuiUMIJ
zoBM}&)eYJNA-*2YHm3TTf1Kd(|BK15dMMS=H(ub4wG|kJR?)`)4~&QvM@57}6;)Bt
zky9<f*E2XgT9Q#Ta{SCC>$bnUb>-X%Rk;Zv0$$sjUG-GNDl|4NPnHB~#m?3+jo{`T
zK%Gi?$5BQKNH$6QZDM~6V(h>(1hU1Dd0HBF;##H3c;UXTsH4N|RsODL5S#!WE(rGT
zU+q7UA=4lKD87nXUXT*IBh3kke!;IWSsW4;5fv?l<|2IsRF!z8KX)I06vY<BNYZl4
zMmA2J+qQ-q?e=!C2JjwYgz;7(*UDf`BWqU}ERN@>EO;bjk}Q8A(qBN*Zv_P;9F2k|
zn*q${!x*-k<Ff@sF5wJ~>6~!VLL|A9v~k!);N|Nh^z=aO)NXR%{PhhlQfywr;7mVq
z(qJYUDkwJAmKNq%TY(`IsE&ZzdC3T-TEesl<9JIZe_mNPws~4(O>tI2gs+PwK(z2_
zpvfG>kK{G61^s`fFlf!etRo2vLqv(`xusdqOEuFTBnSF=Lz|i;n_5iQl-xsl9lNp5
zsVy=gd3Er-4Y@Nsu{(;BAupn2`xi2EFk_f}9<0>oRVI>bX=!Cm3taY`c?OG9a?3|e
zY#mpTE%66+LKgSs4a<@N9ZfupsO01omJH9&AQyws*~)*2)?bDW=CJqhnhe!IjYe9G
z2Tf=f?+`JF(AE#n;+lm0m9>p+T!g_e-XXxEytWRfCq>En6@fPlG2nn&vx;pb3ezS@
z0_Ao9dS(xyFK~Cnfc^Vd`%h$8yTrQp&;fl({xbmk9YAt|)ilmGmgc6;0VH;7n7-uA
zg)LRtA}@bSwO)gDt=vQ6GRsESRt_)D%T7;DV1^O`z(IzSHh|q2#Ni8QaS)g^9AG=a
zPj?5*PI2|vDWgj=#J&!OL+NA>RNf_#fkG$*sIx&mPG>*?YR9yrO$~K4R62mMShEp^
zXZ0Y8glHm66qnp%V9UR(fg-f4J$AYV^_6$$wDf;TIOIqkMVOHU)UB3{UtHd(>C3hr
z+tSujkrHgL)oZAsqaOtFty9L2sTp3Fot7vOM}>z3k^oTPMA4#HCs77t1()6=&_<##
z6H{{wD;ryT2Vp3Q3acBZp-gtLtBDFHe<w;FmKFz+cqFf6#>RM=F$U%?*sJ*|b|swY
z);E7LgOo*-n30G0lYjeS|CtPVzc@5YKlu?+2I*86FMo2^y4qtDZ{{6eGI7b)Q+G~n
zS=5*>66p4#(=B6Xtz0~3N>g2VfecU=wBbez?8Ha3RIaQd@AI;~1>7{I<XUm|jY`g~
z95-#@8W2~D^t4cWWuTUs6Y6sV%Q)0{$mV}Rz!#r=C<0X2*iBa)zrARIqA%SWI{Fa*
z$7K|ZK>W$Sd#(Re76jHgEW(<Mn_pNg)nO)~GGi*{2c3GUTl>Z#pY^6y3ulb4EzOoH
z2%2yeA&whLO_6Jz`y(Z8+)NN8$t)W)b>Z3_D`t$&m-t$%zcN_Y+C3ybt87d~Zi;^>
zKwzy8+vXW8&Hu?5t(U2jFI11l%>e6x<@@Uzo_IG$h>Mg5$fw|#h&0nE(RPW>8@F)F
zsk`rQTijG2CMRCi+#^ahV#54&%NMjZkEw!!49Tc$O>~sG2)=jkE*#IJHfa0Tn#5aB
zcB^ihvABKV<hr8da8D~uW7@dQMTLJcp_7H4G6PYuc#RnCw<{^21fRhp^IAJ{=|N*h
zpUC9=s>W8Zm$l}93d1*^7RgnI$PrhVP{CA+hcl->s$mzDQqj`3<;1q7lPXh#9khBW
zo45o^^J^zBoHK1=!^rZ&>{JPYb;&+qm?z%x^1&BMn2gkO7S@2j5co$)v&(<#CbTYY
zolsi>+NS(yYg%?P7In<@ROBxOUyRd%?hU{cVQLnx`7Wjmd8%s-br_~rRaH?jwD%^C
zuWmy1^q02ie+I*-9w3iEg)oFYlN_9$huRJ5`--ZCS9Eq&%c2d7C)MPLg_i2Qc;xGp
z=0*-_l@J}`>*io#sKp)dKKOs2cXT|X(Q<GE0V!93n_om+dVcw+@za|dYKG^_q%puv
z94(HK#Knt{*3m>)9wM+9-U7PnCdS4_26{X*Lr3;%^1Q=98&rPD-=R^*5Pdsg1abU?
zIbbh!;a|349xn$xu=VBLL#3Sn*>2?IFUhWGoYhpB6Bpo2@_3TYxsiW#u5?sw^@y_K
zf?-)1Fx-v7AX6ZwksNWnnljfuXERwsp_iv8kR$qrilvza<+bDL>#8eCi}JEF(vqd|
zSR)3mQX6wa9W@0`9xiKl1@S2WRsbYjPaA)LVjKWn(7pqP0LQrnOja1+%$2T<S9o$k
z&BS?NFL&i%vSH;;Ur&Eem)xCNbPN)vQ+aAYQzKs5X^^&si@zu(zp@}z9N=oHGnkDm
z6%`ib=K-D?>4^NjpyUFBe?tcJef70Yj%P|>;h`|zCiHao35twO%E&DqF`}}ns=B(S
zhCCaT1O@=uR%QUQViyL~>CEk1g+YP-em>q_o*r(<wgyw49cX_!5%tn<u&Oo)n7BLh
z5It>cPvZEI&GW%t^3uOt!&<$jmKHR<6{b_4Nsh@^7+s1uSV$5qwE=RBMp{~WMusdi
zGb>FR8wDe(h}_ap9@w`R%o28VJYSqiN|1;mLj(PU?w|x6o0tOAr6Wd;9yf6^NfF9&
z05=DPBV)>I#PEMMm~Qxn#foA2i0qvJZyb#@F_4oVy~2+&@>yvB$2KsypQfcpSaLya
z%L1^Mz4kBFusg?1mKn;WNm%R+aZ^8TdXikx_5wMe9SW;>c15vqiOH!MnK>CL2{DnO
z0X`layJ$c7Hk{*m&%w}zXK+j^fUhOdK#2@R)QISqc%XmEs%x4$XUXdMEp<gHk=}OX
zBGSbuPCh8lPhw}Jr6eWBOJYP6bpylD3MvcK@2?Hua3#^fJ_0*4ef7b;)h*n^Ko)cn
z*vnq~mulFLA20-kNleU5c<5^9$qJ@aCFimwTLD=e{e#0ranjVxVOi<P2{F;(!Bi#&
z3Inh9p$UJq-0|$mD9g`EiHi>L^YHTaMMCL_#GK*vljpD4w0GV72^HxP?xvjjJGqe<
zpE~!xWE2<V4a=6LrzR)HMuzx76Inh}&FVH3S3+(nfaX{kY7BZ+%}fxQTsUgV60n!O
z_AlQUQ*mdV!RP%ej)B4&`RbuHmy?;EoERsL3TJ=WKVLbKJAR11tye@+UR4t?lBPfc
z$P?%ZqteS7W-i}y_~QPxGwU+LowVV_q0Ivh6ew4?udAu7pd+}c2@qJg+F2+JPICk$
zi0B2e?<7&6hpn;3Kp<EMNh%yQbt%}3UDUsV{nzpTq4r<L|A*Rt9seI{|8@L-oc#}U
Xcp?$V#o9Rk00000NkvXXu0mjfT7n3q

diff --git a/docs/dev/articles/examples/vision/mnist_convnet.html b/docs/dev/articles/examples/vision/mnist_convnet.html
index b1284fa2c..73c198f82 100644
--- a/docs/dev/articles/examples/vision/mnist_convnet.html
+++ b/docs/dev/articles/examples/vision/mnist_convnet.html
@@ -31,7 +31,7 @@
 
     <a class="navbar-brand me-2" href="../../../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -183,35 +183,35 @@ <h2 id="train-the-model">Train the model<a class="anchor" aria-label="anchor" hr
 <span>  validation_split <span class="op">=</span> <span class="fl">0.1</span></span>
 <span><span class="op">)</span></span></code></pre></div>
 <pre><code><span><span class="co">## Epoch 1/15</span></span>
-<span><span class="co">## 422/422 - 5s - 13ms/step - accuracy: 0.8894 - loss: 0.3635 - val_accuracy: 0.9787 - val_loss: 0.0791</span></span>
+<span><span class="co">## 422/422 - 5s - 11ms/step - accuracy: 0.8896 - loss: 0.3635 - val_accuracy: 0.9783 - val_loss: 0.0791</span></span>
 <span><span class="co">## Epoch 2/15</span></span>
-<span><span class="co">## 422/422 - 1s - 3ms/step - accuracy: 0.9665 - loss: 0.1111 - val_accuracy: 0.9848 - val_loss: 0.0550</span></span>
+<span><span class="co">## 422/422 - 1s - 2ms/step - accuracy: 0.9666 - loss: 0.1111 - val_accuracy: 0.9850 - val_loss: 0.0548</span></span>
 <span><span class="co">## Epoch 3/15</span></span>
-<span><span class="co">## 422/422 - 1s - 2ms/step - accuracy: 0.9743 - loss: 0.0824 - val_accuracy: 0.9883 - val_loss: 0.0441</span></span>
+<span><span class="co">## 422/422 - 1s - 2ms/step - accuracy: 0.9744 - loss: 0.0824 - val_accuracy: 0.9883 - val_loss: 0.0441</span></span>
 <span><span class="co">## Epoch 4/15</span></span>
-<span><span class="co">## 422/422 - 1s - 2ms/step - accuracy: 0.9785 - loss: 0.0695 - val_accuracy: 0.9895 - val_loss: 0.0397</span></span>
+<span><span class="co">## 422/422 - 1s - 2ms/step - accuracy: 0.9786 - loss: 0.0694 - val_accuracy: 0.9895 - val_loss: 0.0401</span></span>
 <span><span class="co">## Epoch 5/15</span></span>
-<span><span class="co">## 422/422 - 1s - 2ms/step - accuracy: 0.9803 - loss: 0.0626 - val_accuracy: 0.9903 - val_loss: 0.0352</span></span>
+<span><span class="co">## 422/422 - 1s - 2ms/step - accuracy: 0.9804 - loss: 0.0625 - val_accuracy: 0.9903 - val_loss: 0.0353</span></span>
 <span><span class="co">## Epoch 6/15</span></span>
-<span><span class="co">## 422/422 - 1s - 2ms/step - accuracy: 0.9823 - loss: 0.0559 - val_accuracy: 0.9912 - val_loss: 0.0332</span></span>
+<span><span class="co">## 422/422 - 1s - 2ms/step - accuracy: 0.9824 - loss: 0.0557 - val_accuracy: 0.9907 - val_loss: 0.0332</span></span>
 <span><span class="co">## Epoch 7/15</span></span>
-<span><span class="co">## 422/422 - 1s - 2ms/step - accuracy: 0.9837 - loss: 0.0501 - val_accuracy: 0.9917 - val_loss: 0.0311</span></span>
+<span><span class="co">## 422/422 - 1s - 2ms/step - accuracy: 0.9838 - loss: 0.0500 - val_accuracy: 0.9917 - val_loss: 0.0312</span></span>
 <span><span class="co">## Epoch 8/15</span></span>
-<span><span class="co">## 422/422 - 1s - 2ms/step - accuracy: 0.9848 - loss: 0.0481 - val_accuracy: 0.9923 - val_loss: 0.0307</span></span>
+<span><span class="co">## 422/422 - 1s - 2ms/step - accuracy: 0.9850 - loss: 0.0480 - val_accuracy: 0.9918 - val_loss: 0.0304</span></span>
 <span><span class="co">## Epoch 9/15</span></span>
-<span><span class="co">## 422/422 - 1s - 2ms/step - accuracy: 0.9861 - loss: 0.0444 - val_accuracy: 0.9917 - val_loss: 0.0303</span></span>
+<span><span class="co">## 422/422 - 1s - 2ms/step - accuracy: 0.9860 - loss: 0.0445 - val_accuracy: 0.9918 - val_loss: 0.0303</span></span>
 <span><span class="co">## Epoch 10/15</span></span>
-<span><span class="co">## 422/422 - 1s - 2ms/step - accuracy: 0.9862 - loss: 0.0440 - val_accuracy: 0.9913 - val_loss: 0.0298</span></span>
+<span><span class="co">## 422/422 - 1s - 2ms/step - accuracy: 0.9861 - loss: 0.0442 - val_accuracy: 0.9912 - val_loss: 0.0293</span></span>
 <span><span class="co">## Epoch 11/15</span></span>
-<span><span class="co">## 422/422 - 1s - 2ms/step - accuracy: 0.9873 - loss: 0.0394 - val_accuracy: 0.9920 - val_loss: 0.0304</span></span>
+<span><span class="co">## 422/422 - 1s - 2ms/step - accuracy: 0.9876 - loss: 0.0390 - val_accuracy: 0.9915 - val_loss: 0.0307</span></span>
 <span><span class="co">## Epoch 12/15</span></span>
-<span><span class="co">## 422/422 - 1s - 2ms/step - accuracy: 0.9875 - loss: 0.0375 - val_accuracy: 0.9925 - val_loss: 0.0295</span></span>
+<span><span class="co">## 422/422 - 1s - 2ms/step - accuracy: 0.9876 - loss: 0.0374 - val_accuracy: 0.9925 - val_loss: 0.0288</span></span>
 <span><span class="co">## Epoch 13/15</span></span>
-<span><span class="co">## 422/422 - 1s - 2ms/step - accuracy: 0.9887 - loss: 0.0346 - val_accuracy: 0.9920 - val_loss: 0.0297</span></span>
+<span><span class="co">## 422/422 - 1s - 2ms/step - accuracy: 0.9887 - loss: 0.0346 - val_accuracy: 0.9920 - val_loss: 0.0291</span></span>
 <span><span class="co">## Epoch 14/15</span></span>
-<span><span class="co">## 422/422 - 1s - 2ms/step - accuracy: 0.9889 - loss: 0.0343 - val_accuracy: 0.9923 - val_loss: 0.0285</span></span>
+<span><span class="co">## 422/422 - 1s - 2ms/step - accuracy: 0.9887 - loss: 0.0343 - val_accuracy: 0.9923 - val_loss: 0.0276</span></span>
 <span><span class="co">## Epoch 15/15</span></span>
-<span><span class="co">## 422/422 - 1s - 3ms/step - accuracy: 0.9898 - loss: 0.0320 - val_accuracy: 0.9925 - val_loss: 0.0279</span></span></code></pre>
+<span><span class="co">## 422/422 - 1s - 2ms/step - accuracy: 0.9897 - loss: 0.0323 - val_accuracy: 0.9922 - val_loss: 0.0286</span></span></code></pre>
 </div>
 <div class="section level2">
 <h2 id="evaluate-the-trained-model">Evaluate the trained model<a class="anchor" aria-label="anchor" href="#evaluate-the-trained-model"></a>
@@ -220,10 +220,10 @@ <h2 id="evaluate-the-trained-model">Evaluate the trained model<a class="anchor"
 <code class="sourceCode R"><span><span class="va">score</span> <span class="op">&lt;-</span> <span class="va">model</span> <span class="op">|&gt;</span> <span class="fu"><a href="https://rdrr.io/pkg/tensorflow/man/evaluate.html" class="external-link">evaluate</a></span><span class="op">(</span><span class="va">x_test</span>, <span class="va">y_test</span>, verbose <span class="op">=</span> <span class="fl">0</span><span class="op">)</span></span>
 <span><span class="va">score</span></span></code></pre></div>
 <pre><code><span><span class="co">## $accuracy</span></span>
-<span><span class="co">## [1] 0.9913</span></span>
+<span><span class="co">## [1] 0.9915</span></span>
 <span><span class="co">##</span></span>
 <span><span class="co">## $loss</span></span>
-<span><span class="co">## [1] 0.02430356</span></span></code></pre>
+<span><span class="co">## [1] 0.02471612</span></span></code></pre>
 </div>
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
     </nav></aside>
@@ -236,7 +236,7 @@ <h2 id="evaluate-the-trained-model">Evaluate the trained model<a class="anchor"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer>
diff --git a/docs/dev/articles/examples/vision/mnist_siamese_graph.html b/docs/dev/articles/examples/vision/mnist_siamese_graph.html
index b2acce7ac..b12778678 100644
--- a/docs/dev/articles/examples/vision/mnist_siamese_graph.html
+++ b/docs/dev/articles/examples/vision/mnist_siamese_graph.html
@@ -29,7 +29,7 @@
 
     <a class="navbar-brand me-2" href="../../../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -345,7 +345,7 @@ <h2 id="plots">Plots<a class="anchor" aria-label="anchor" href="#plots"></a>
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer>
diff --git a/docs/dev/articles/examples/vision/oxford_pets_image_segmentation.html b/docs/dev/articles/examples/vision/oxford_pets_image_segmentation.html
index 820c0fd14..23395e626 100644
--- a/docs/dev/articles/examples/vision/oxford_pets_image_segmentation.html
+++ b/docs/dev/articles/examples/vision/oxford_pets_image_segmentation.html
@@ -31,7 +31,7 @@
 
     <a class="navbar-brand me-2" href="../../../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -712,7 +712,7 @@ <h2 id="visualize-predictions">Visualize predictions<a class="anchor" aria-label
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer>
diff --git a/docs/dev/articles/functional_api.html b/docs/dev/articles/functional_api.html
index 342439866..fa2ef5ff7 100644
--- a/docs/dev/articles/functional_api.html
+++ b/docs/dev/articles/functional_api.html
@@ -31,7 +31,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -224,17 +224,17 @@ <h2 id="training-evaluation-and-inference">Training, evaluation, and inference<a
 <span>    <span class="va">x_train</span>, <span class="va">y_train</span>, batch_size <span class="op">=</span> <span class="fl">64</span>, epochs <span class="op">=</span> <span class="fl">2</span>, validation_split <span class="op">=</span> <span class="fl">0.2</span></span>
 <span><span class="op">)</span></span></code></pre></div>
 <pre><code><span><span class="co">## Epoch 1/2</span></span>
-<span><span class="co">## 750/750 - 2s - 3ms/step - accuracy: 0.8979 - loss: 0.3540 - val_accuracy: 0.9448 - val_loss: 0.1903</span></span>
+<span><span class="co">## 750/750 - 2s - 3ms/step - accuracy: 0.8980 - loss: 0.3540 - val_accuracy: 0.9444 - val_loss: 0.1898</span></span>
 <span><span class="co">## Epoch 2/2</span></span>
-<span><span class="co">## 750/750 - 1s - 1ms/step - accuracy: 0.9511 - loss: 0.1634 - val_accuracy: 0.9605 - val_loss: 0.1386</span></span></code></pre>
+<span><span class="co">## 750/750 - 1s - 1ms/step - accuracy: 0.9512 - loss: 0.1633 - val_accuracy: 0.9605 - val_loss: 0.1387</span></span></code></pre>
 <div class="sourceCode" id="cb18"><pre class="downlit sourceCode r">
 <code class="sourceCode R"><span><span class="va">test_scores</span> <span class="op">&lt;-</span> <span class="va">model</span> <span class="op">|&gt;</span> <span class="fu"><a href="https://rdrr.io/pkg/tensorflow/man/evaluate.html" class="external-link">evaluate</a></span><span class="op">(</span><span class="va">x_test</span>, <span class="va">y_test</span>, verbose<span class="op">=</span><span class="fl">2</span><span class="op">)</span></span></code></pre></div>
-<pre><code><span><span class="co">## 313/313 - 1s - 2ms/step - accuracy: 0.9593 - loss: 0.1323</span></span></code></pre>
+<pre><code><span><span class="co">## 313/313 - 1s - 2ms/step - accuracy: 0.9598 - loss: 0.1322</span></span></code></pre>
 <div class="sourceCode" id="cb20"><pre class="downlit sourceCode r">
 <code class="sourceCode R"><span><span class="fu"><a href="https://rdrr.io/r/base/cat.html" class="external-link">cat</a></span><span class="op">(</span><span class="st">"Test loss:"</span>, <span class="va">test_scores</span><span class="op">$</span><span class="va">loss</span>, <span class="st">"\n"</span><span class="op">)</span></span>
 <span><span class="fu"><a href="https://rdrr.io/r/base/cat.html" class="external-link">cat</a></span><span class="op">(</span><span class="st">"Test accuracy:"</span>, <span class="va">test_scores</span><span class="op">$</span><span class="va">accuracy</span>, <span class="st">"\n"</span><span class="op">)</span></span></code></pre></div>
-<pre><code><span><span class="co">## Test loss: 0.1323339</span></span>
-<span><span class="co">## Test accuracy: 0.9593</span></span></code></pre>
+<pre><code><span><span class="co">## Test loss: 0.1321879</span></span>
+<span><span class="co">## Test accuracy: 0.9598</span></span></code></pre>
 <p>For further reading, see the <a href="training_with_built_in_methods.html">training and evaluation</a>
 guide.</p>
 </div>
@@ -592,9 +592,9 @@ <h3 id="models-with-multiple-inputs-and-outputs">Models with multiple inputs and
 <span>  batch_size <span class="op">=</span> <span class="fl">32</span></span>
 <span><span class="op">)</span></span></code></pre></div>
 <pre><code><span><span class="co">## Epoch 1/2</span></span>
-<span><span class="co">## 40/40 - 4s - 99ms/step - department_loss: 2.8465 - loss: 0.7669 - priority_loss: 0.1976</span></span>
+<span><span class="co">## 40/40 - 3s - 64ms/step - department_loss: 2.8465 - loss: 0.7669 - priority_loss: 0.1976</span></span>
 <span><span class="co">## Epoch 2/2</span></span>
-<span><span class="co">## 40/40 - 0s - 6ms/step - department_loss: 2.8554 - loss: 0.7538 - priority_loss: 0.1828</span></span></code></pre>
+<span><span class="co">## 40/40 - 0s - 5ms/step - department_loss: 2.8554 - loss: 0.7538 - priority_loss: 0.1828</span></span></code></pre>
 <p>When calling fit with a <code>Dataset</code> object, it should yield
 either a list of lists like
 <code>list(list(title_data, body_data, tags_data), list(priority_targets, dept_targets))</code>
@@ -707,7 +707,7 @@ <h3 id="a-toy-resnet-model">A toy ResNet model<a class="anchor" aria-label="anch
 <span>  epochs <span class="op">=</span> <span class="fl">1</span>,</span>
 <span>  validation_split <span class="op">=</span> <span class="fl">0.2</span></span>
 <span><span class="op">)</span></span></code></pre></div>
-<pre><code><span><span class="co">## 13/13 - 5s - 418ms/step - acc: 0.1225 - loss: 2.3000 - val_acc: 0.1150 - val_loss: 2.2942</span></span></code></pre>
+<pre><code><span><span class="co">## 13/13 - 6s - 430ms/step - acc: 0.1250 - loss: 2.2998 - val_acc: 0.1250 - val_loss: 2.2939</span></span></code></pre>
 </div>
 </div>
 <div class="section level2">
@@ -1114,7 +1114,7 @@ <h2 id="mix-and-match-api-styles">Mix-and-match API styles<a class="anchor" aria
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer>
diff --git a/docs/dev/articles/functional_api/unnamed-chunk-10-1.png b/docs/dev/articles/functional_api/unnamed-chunk-10-1.png
index 94a54466e96a38b0121a288b4101d6206d368bbb..754a2d2bc8d0ac882023373243f201bb53c5008e 100644
GIT binary patch
literal 20488
zcmd_S2T+u0yRD5m4WeT}LBJe{269$ZL?lVhBBBx+L~=G_Kwu1jN>p-|)T9O!R1_2>
zG>tSta<a)eG-ti@?Z4_f|33fOr%rvl&i?DvOw~Bjbie)f^W68kueF|jq^_#Sx}IY_
z3kwUY(z(+YSy-0Yv9PQVUAr28Be8e+a{OoYWo5<FEK9^cQAP33Sy*<nD4jm3<r?<8
z#l=-y`z>=+Oy;k%f9}37cz*dGx=(g*dU1UBKI6w19q(1TYJ6Chzt8B?o#+nnja7FU
zkI!$G+PQk`zKts%^xW!Oh<Ug=ZSR4Vx9_dIc<@{WbFuwCU)OK1Dkas5^=;MTVRD%?
zlkOLkpZeE^a~An4Ggq=4{c5@}=)RJLWq5Nd%cb3=oGgEwc3pAn_F6lZqnp}USyu0*
zEPF41PmE>9!;#f2D~Jb0$*VaxWm`VCEm@p;&AoxesA2Eky)Rz9>aQv0zCZml>snBF
zxN+*qWh`N5PM>zA2-O|@Uwq>KrC0yYdXV}hW48Ap2UWuoHgZWe2K!9?yf#N~W9s<r
zJ7uh-qC$;2dDo^gkVh`<jNf)@$3vDjs~hbyj(zI%vEB%efRVKLp`vLM5mHmVg}$wA
znu+J=ne--Q{2s+^+@{iHxK+iq>^dti#VpNZxJXw?NlCDBYPfCK$#n%Qm&3om^QL+_
znpdvGxK9tI>ZBSR5f#;$ot>3*?E8ys^Yg9G;-INWlS}7kf1!yMikm~Yte1hlzJ61t
zh2Q!O8?3YK^4NFoOiWHIIKOV)I;*Da#Mi=>Z~yvh)2cOVl4D{{-@kuf(xK-GUOwHf
zc=oIubH?15?dPFCw`>V|_Uvt(N+|EagJ<{b*;DcTySA!ofHR{hR^h=$YK8czQ>W?+
zUF@8GH)^Q77S=d<@^<mUnDSyL&r(7V)0(4w@dTHo<ByMzwvZ}mDf-buB?}jyIer(R
z^?g{wQ$NyPwvB_sz_KmtTJ^wpv9|hsfzofc98%3=NQsN<IrLWL2;fJ7c((_X&YwSD
ztH34`qY-`fh_rOGMH`Rj$Zc-dU&{i$HBKogOuk>ihEHhxyV!YUWe1zixKP`R)e$nJ
zN_Ih`gt~Y&)0x=@Lp$ww)o{TD8<*F=W;4sVJ-(3_`f8#LD6ScuGgb<P&Wwp43T%mZ
zV*@|l-0kS-kWMqy(TVe6jz_w8`io5r6%LzB7%VoIEYeFXC;Dp0jF{Ni{!qK((Y!YO
ztgA<Oc`GdRot?=h9t@GdCd;<sqN`P5iF*`$6TIf90_M5JZ^$R>q}JY9wZU<$ODV8v
zx@UTyw6qbM%*0=~!*O4}eDRy27kT6sI5P$`A}1rdZwJaT8=q+n^Yr)k2UMH4m);MU
ze89@tqqkHgHJrV$u#nQW#GKY~bIU&@6h2wTCZ`uJ<>D+gH8OIe+_}|#P(b;a({ziA
zAwJG%MKAVk+w?W#F4)suo*z`%Mc2>28A})4E@;%(c$Q8k2{rk2R|Ij(tWopCG7tLf
zu(UWW)$lXNI;2|8XK`+F{OkDF@85Gi2l5PO1~xY8Vby#j4~=zK%5dAwPcgX4M`t<%
z3rYJ99{lXy8}1^_^qR?k`*!c(Q$_BKg~dg$sri~TV=3z;<|28X`f+}!Qt|ZZd+|c1
z91g7}YuB#*LANVORpjx~VlpF&97jh-o8Fn-eH$H}+_H3+O*SFK)LZ|<hY#Egp6%P?
zKKZc^;<HH8DG9dFPL`!?;p9wqY4g&>jZ}}#)Qnf{4|i$JNcE@`$W2gsevo_8VK6mU
zGFZs8xH*HO7|7j}?+}0R!pmQy!)4W2`2C~hJhU}Dp6dN;^Lby3SjYKuib<MSFnW7+
z8QudMO6ePjPl)&hr=pn|SDD3$I_f~7&l1ypATgej*4x_~zr-9~(utJyN@^BsHx`R<
z?O4akDVkC{#ctyIfvg&R*57fSk?T><%(L&bsU)`RiZvS>gI)S-&+Oc})2gK)N6|}S
zr>M<K<{YhrclYin0|sUEGpAKuoJvl)`T8D8(T^nUWL{cPXMog%>C&Vr+2PgD{(}c~
z^6lj0<OGZhw3epqmULo;G}a9#W$-IM*vM7eTI7LEd2HME?E}AmW?9uIJjG^qV&u>Y
zWX^ly!Y67Jd5kBxe>deQ8|K<lS;d-vWwzEQYPOg>ZF-}y#@K6;O75-<7RVCt^Y_<w
zcP~Iex0y~%Nce%Zn(Z-di>0QDn#CdQ`VK|FXh`nHFZu70a&4o9?1xI`sOh^(#3a)9
zaJg-1)qW8gnw(=z8)(chC*JJYfL)32-W{bcDkvz3-0V7+yY$$7ubiBz&pf@&ac0C!
z#I{*a#$&49iDulZ7mvL&H|?|JQZ!MkY%;Cr>QR-+pK4sFV`*tQ*B2e=I8cA2e6i4J
zFs{2Y#B`#YQSF4=x?=SoKYUp^1)G?q8-yBtT|xwn>$9$@d0}aj7bUrzuoo<`+{xW2
zGEUMRtyt#xfTQqQ#@R(Gp`r#w?m21sJscsELnTYz0{WS!+q~yfNpxpH)*<uqfd1!d
zvNajz<y*IHizSojNxJDK_olmT`)gz87<N7$H5xW3CfW9#XUl!)1unxTj1g}vL1C{s
z$8B4;8fAO^dVhPa(v;aD-X`HSH^F<~<;<D;G^b`IW#u}2ZbA#5!$s4)qM{Am#VA-e
zrJqT=Pb5^A>uziFnhuwA8A|!~?Ly$-!g%CT{X4U=2}W*PlPc%JSV&3A*4AX*^oE*f
zMQWyna)?u$YWTp8R@LzqKHW4e4UMOjnfLCk#M}E$w$;%IUcGTw-&kNnd)Wh#D_=Hw
zEl$!NKYEm?#l2}$)B|q!bI%W{kx6ag&z}AH{pB$Q<q*NtiG0@e>ysKL%!)d{(q4y=
z#(ZSw`u0^;S65psq744b7{b~}S|Alz=r?@!=j1AR#Up(EFS@9t^KZSxh{7zDR?l&j
z-1MT{YoSJSjk$J_q!!$9w?+K~zbuuc@kV?K3_XUd7*Y9CHI0mH&YV44kV4BaS1hN+
zsYS|?JHuqX3d~!4mZp7N=8YoV2aeF^4jee3bN%}DAaA*f%F5JbD^}OfdgzBg5W5z7
zql*!Xy8Kgh-@@Xsk4uVS%L!F(?-{f5=wFkAG;1vG07dr1#6%OWo#!t%#c*(l*`^R3
zQrFli=(4#vwocr;ckj+C{!9xvU2R?S_I@QJy?8E(+pTMJ2yLe@R%vdFbEt{;Z1;H!
zRn3EIHtzb-kZMS(3~x9m#K-60A#-a#sg<J7COZ>Vnfc>~dTd-=!J_d^i_wlxb6wBX
zC~4PtQCuQ@nEA|JA7(k35)~yMuNFxTFL+nDqjZaegao(o&M7oXabK77=N||2>vw3k
zi5w%<{`KkO$3C=x`iGnM)#C%_4L|E*>5<me(=$*s-I^7CbtZ6WhApI>A#9Q9IeW8Q
zu0h6q;++5=KYu-H0jaXuVY=I6xV0#GYikHS^TWMd_OPB!)%bBX>+t*cNhQsuUXwv;
zQ{!EnZdxthUY(enZ(Axh@JSHrcHB6zL5OmRvz2D#;9r1EM>I!vGO^{jq?{e|ODyyQ
z+P==r&PsTDqk41M-D*$7l1m>9lXk1^Y<uc^XPi=W1A9DoK1|9b?d)7tW##x+Z`GxU
zX!i(@AqfGaydUu&1SBF`du&q!5;fyZCbH^Yd_5HD`F5XTKx43?qN47+k&T*GH%n))
zskf=%T00@v%tlQWm1C}w!@q~L5|X}|dW_lmKnNAt=I+@k>zQ};+e?i@UQ)){C(KIk
z`8AZ}cLy8VOgl;D7K%!Q%h?{1lPg($_F~B~QpYN_JLz$y^daE_<^1`J*}ObA)db7!
zS}FN7hUor4>9L1&uYg?=Z&M9(2cR1yJ*Ix2u|P*n96NJ~&|lEFj4{9O=)Njp;u^$K
zK_|EJ&`^ge(n&XAG^%-f7h{oHFZ31i_Fgufb2r(z)=t5(zxLCi`R<+`ou-~4wt(cI
z4MKm{>6Z96R#<eXAGU@jCMIbfL-~q+>_TK2mL1ZrH-gFvQ&KMYZ5OEBq9)y*xgf{v
zl$#hxPM1`DMVW5(;5~B0{&#sMKc2QJE%+s_>^YUIWPYg7D%W0OH-)t(MgPR)Pz#;0
z^U%JtPj2wbmBcWrw{6?@HItvdkK1KX*T`palFaVsTG3UvIb6^_#QSx?jofT2C)!%K
zNviJTNEurpWr2d$QPA_++Vi-@*lcg4^~^}SU?qo$RZKaLPgvG-t91^otKG)F9^zVx
z-F5WRt+^@2VB=>LOkPrQLc)!(>fy$WXvR*K+qZAK_q@8cYkuNho&VZ_`JKmyKIfvv
zl#iJdxgU|0EjBP0cN>e*NzpSiWY#JR{Y=X$vTKtKtcrQ&T@d_U^}>ZGz8kp)G^En*
z-M^oRYtl~h$=^d^txY$T6F}GRW|v%?pLX%MefsLvIL0c;`RU=@RQ4S^5@MA@bPNg)
zNJ<)@n+#BsL)JvQIQMrm<=8nn2VJ8d6zCWk{noS1wS6>E*J45}(8woc{eH^v(|waP
z)9JiGYO<~r^Nn%djW;wOn|=HCiCnAN#f6S&5c1j7_NC~tZ|OdZ)~sBT6L0S)x=-}A
zYvf6!S`)WQs*2v~Sf?^KMYqej`kkY@w>OzYClk%ITGh`%@{Gz{|3e`e8Dq5aBuB}f
z$fbpA<s&W!WMx}Nhon^niwyniY7?JLgtjf5C6*qa@gfZszX+-!AiYqGIa}>x1yyFU
zcuYV@h=$tSKg#_1^QS@f^{3FJ{h14o`O<z&G3b-TI5XIgniam$@5vJ_NEk;tDc7>=
z%Tq!zl$*a-TWhxh)keHE?RyYJi+GiFno<5m2x=FbeS7wtbZK!4(Oq;Os1z!xC*`(W
zHRRsB*{m>luH%zm(Z=DtygXi0^-N4Uex)8zesj$EVW&^qbn}|!pzk?FZGQB855;OS
zP0Y*F3zu}NB?hXHbE!#$Zb}|8$+Zm+;F9Xm<{9s;ra?TAgy<z>6?~HrY<zrt%Jf`2
zU9_>P2$|%*zU!E?w4h0J$^*EPy}J9m14cwYqD3cB;&LtY=f2x{cjg68qd!gH_1rvl
zDq}M<Gm~;!XQBL}R&Ch6qwq>^By-|WNXXkIqqbr%WA8&65dRAco+8!_DNUVvkjoKX
zQ~H7F7bmV(bWy^>cHwi?!eVohK2)fktV5l<@P4Py+#eZe`ysiXZesJPM9Lb6irG0A
z4%>N;a|VTkc(jy92g<s)_oQn({d&KwXo)#bwTc!LFrnntW~N5G<5!+;b_nczEzx>d
zK!E!8jGuIqn2?YfE+FxuT%pTw18$IX!N3lYyCb4sG}UBc`|IL8Jl8Qxn3F;+ZoT2Q
zforShJN$&uo5ah7VgyM(6)%pEd)Yk((+Y=@vTv0fsi>$BzWQxPdDbyO?TZ)RLLSd{
zHTOk&bO-55&pn}xS4nY~(9(zzV4x`{8Jal_w~05`p|#j+ewVKB$4_hR$`&BUh#@5j
zqMPQ-(S@F3ka77wAxyBL+^I~{r8!-v!3GtKE#%(8?OV5AhcX&y&d*X4O04?&RbzUn
zW%ApJYFA3Ive!j*^{5Yb*ARlfVY<!7uCVRZiE9&cgXzgo9421+eTf(dU9*Lgg<1=-
z?5xMSzFN`J!kHrv<oX8FOQvSR7V&`VlLOSbmSJzmCCNkxw19Z2j?1i8Rr8nx@~^MT
zQO=hNne;)Moj@yVB{x%735K|jeqtx71&vK9i%7kC_3E{_{oB&gQbD8Lb^kO!Q{1u?
z2j4aREObftW#dVE`Eti3gRU!JNY=vmq8cHcSoVO;8XG7`P?3qf)Z^--sy5uBucZ~U
zG?%_qi<vA%(6}H0?*-aQ1HZ#yH%c?1VXsRRhfqhyh4_TDH8kGnnz)@I8U}F>hgu4g
zAzL!}l@F<gsbhgNM+1GdB#)aEI3_Z(m9+)k-I%kzTz01yh?P9yHdPP3C*m?>^gUck
zS3^T%ptY#5qqB1x7gy?+FXw`SgOjkk2vp!Y618!ccocM<dowfNp?6nXHDvZohvmyp
z&kaRi-aN$hoM&ig>lg-Q>g*1;my<I+p>{TMYFSsmCykGeCM=Bfs7hW4WnonZhF~Q6
z#m2_EW3I7wJ3S&&%6a`3K*@EFjk6e`D6cc+7r~5JFJL!vRkPeIc*5OtUENe~c}S>P
znXj?lw?xeGm<OUegF>;5r~ATu>V58sU4oqTb)#IljU9_mQ@3p2DUw;!%DfBG7`mE@
zy*!NBR>+=mT@i8{3vHnN0Z-bX+4~1CMH3I@tz9N!h7#l6@#v7a$h&2GDT9w7BemVU
zERBk5>*|~qW=1#283xBf&W207U0fPoT9m_#qw6FtS--!O^Qaz196WB&eMiEIZ*Ko{
z+V!8soBluG+W+p$jb-sy-ps8d=-cz>nw*`>lDczfCUwvY8kX}YtS|ohOMT70<<h;w
z!|B@v3|un@O_$63<P#RwN;SwyYE8Ih@FPZPO9*|-mMs`3Dh;?<)g4}nF8c@Z``^8&
zf6os8Z+WKWgYGQ1r1JCgoAT_V73YUqi|E}`6BCY~K7TGq<>BB+!XV}$A{!+*qaSQR
zo!rfG=+mc9j{^gVJ?%g31e$@SkotLkSyI2yh0v(h4I4Hj=R~)Uc9zr5%gH-!ZarFh
zZ`~FvN?=05Vb-0;WAI4N+atc4Z43Gj2KJ;Rn}2(GY&|P$YK<(5=^l#Ny{M?&{@gMJ
zXXg|Aqdng1uC?0{eXvYKP*4?GDLrFqndv@?*%OrgB#X*mk=m)5(avp~HmUC;tq{Sp
z9s|>H!c5z6j(*|Qd*ORxPcClR<AL7ZS1_11?p7$TvX^4^!nUP=z-h81hmMbr#>U6T
zEkos1@U**3dAuCVHeNXGS|(on$8d6bu>o55l)i^N30D>t6SF@yCT1o^pg7@PBChKz
z`<p>f7W>Z6vzLVB8F+%2M_Ue6w3WE?v|(N*lvDA}P3o0+=#gW`G{l;3DC9I<s#tjt
zk4zc3XLspR)aNHVJIiWVa{NYoGi>v_w}X4?CJw#7y<!_XdvE9FWjPP<gUgpMI|B6S
zBnoOCJj&J18m)cc-ai<{7mgjQ>K=H021=rit81RF6VVP55)wE#HmghF2Lk;37s`BD
z$CEaOltu5^O)-<Ui&NdZZ(qLJ0gjcWoS!)%?aHTfcS$+(PbZ$fEALw7+S=UqEbGD#
z^-ZnUH%YJSh89XTE%6qS+Yz;$_$#R8WNb_EA<Qi+SFK*%Q2X%K>ocy=Cr75IT?$X_
z);xIq+*&)~9jfi$ym)JVjL3Bzz@VLg9#+7;I_dDyqv{qG7D43RuC4^g6Q`qlXNdJN
z5ihrxnD#&;eN$`5m#V6R6wav9(o?Ovce}g0sgs)qxZ+e;{$Ftazv2dsdSAfJLJ4{O
zI_c!y)e=LR=%KZ>wHpFYT&p?^l;q~-CV8{#VF0&G(%9IT0mm^GtY+31c(~hmk9M+u
zZS6${Jy#9fQ}gUuU)24{&Oo`j>-Si!IO-0DBw}k2BQ2qBb<&Io1=7dnGjkW6O^2qi
z;@7fL&Sx>$Wili0S2UIQlpIuftwk$a_G@*!zu##^#X-ylG`@7o+xPNUski1HkN7GB
zN>Ty&<8R%%RlvQTg<U+`>XvGl_!}4p#xoLtz%?~Bz<vIh5#C^a3~btY;_8#_{CZIk
z0@Ql-Wm=otTUfZ{yo~~7re0QN-nzwNjQI@L6MXi`jy-^Y<zxsbS&J)It}~i)s2rM?
zFTcHb@nU_pr6$z-;r;thqwZ9E{R#$K=-Bt;jaANY;`{gVAQDN&h0f>Com02B&+wjU
zUqkwtdFGH>M2a&Vk|g%o^XJ=mc+zo2_kDen5)#g>TD6Ms3D&P)@3=5y543*{09Dpa
z_2kKu=arN+LPbei0b`-Yv90(H9a02srJ~b_KTkkM7CM17)&xZ*YS)@#x;UWg4qV?7
zVK%lfUd^{}p91uR$s`qB{(G2&rh@Mu8U$CKmbvlk794<BXmr<IZvlZ{100ZMBV;_X
zt*E?87mbX<xIITsV&tyr9?#)dCdP^|35RTJnqc6Z+^{dF?T;3hw$wKY4|oCSJKcd*
z=CFw;@F7{xP4t^g?!+L7(qajzJn%Cs9u&x?&`wF|5!tqd7(`A&#hOgsb^lg+1_+d2
zX>!h<K;yCxaO2#Xb#C6wbefy6%uTQ4a%pb24?TA5ScB9s#hg(4i*p2XC%{4X@bEA>
zlHl^>e{gvbtNO!ya&m3aOl4)|RMc>f329Eea;jcN>0oc)W3k~S3_Ph&m%53OKq~~<
zPaCg(mm7FQGj68K$yUC0?Xa?uZM(o%4jK1sI<31-O>QzFm0QMLOG3|k2*NWP)rV_-
zY(FJlCG@JcmX<5k1x^7KZcTad;^JZ|IG}d#5tEP4X`+V@e<9Rxa*Hldg~;WP4<(m&
zZE2;Ray1>e`G=iS1G}kbd*f9=N?2D8<)Q_S{g(kUXMPvKvv{|B`SPS)tuUKR)=axW
zNCzk`0YKOVHRW`E^~kpQc|39wW7^e4f=`ZDA8$9du>9tx@GSd!&0Y*XjpXL)YBf+e
zcidIu;PTI(c`^2G-n^Mi($dnp4g}r=OQb4XDjr>dm0Ma5E|lOjkNAzKp6bZ-HQcW6
z3lEK#7?5rBW8I&<2n$Q$w4-Q1kUQwhIt^&i3y39G;FMg<oU_J|&4;-Y?UbO<gwzJE
z^N@*vKqAJ9fcY!eu34joql&(K`}VEyrS~j=i}}rByLZd;>7=NG!C@GpaI+2EE?|yp
zv>Bx-wxWu-%r|`t6{{vQv#MnW^;~+Z!m%iB#=4W5vNH%XjP8y)+@P&%tR1^=IFj$k
zkqp2f5@91WU~IOrxOS~DH4C>XWbyOIkBtp*S)8Qeo55$Sp;;TA9Z*i8P^2p}!vZOt
z9UUnh9hd2{l^~rG4n2ARC+t+T)hsZs_L-J4*fGGuzrWtyH+=r3XnuV+N9(f`ltgiI
z;gzp{hD`qX=btGMz;t)TfL$3toTPdGz`$J0OcsVq*bgrH)paAS8RieDKzn`zzW49z
zThr2<QgVT?Ll&7G>?Q9~vs_0nr<;t!I<$rb7BJ5bVrL%)1J5~S2~e}CL4c1ht0~vc
zv^Ubyg{`#knQ+>>kE;lb&2Lv5^Qe*$tr!Sin?Nc;C6*w2p^Ow^rYCoU*g4p=eZ|NQ
zE+W?d<>@|ll8~TaL)X{mTxvcSR8^hb?kAR%ET+RfG>S%r4lJKW(KZneui1tIsPgjY
z2lB%Fa0%^iRw5hW*0Ym7PFb;h`QP4)s9rtaJu$(<hY#Q2&)uLNpYy4*qT)qxa1VL-
zd#ISyROsv117ucRZjax`uuKa1rNEnt=LY!b1<RH#d;H{yqkgq2K7X*QIF#lDAPp;U
z!RfoqSE#6x<fnKhXXB|mca}4n^Xtf-?{6*B=H?mLXsXw-+=cyRn2bk5_Z0SKW5bbu
zK-iY4@Ru+9^jfhS7YR5`ud2qRX7%auwt)Fv5;spDIdX*Djg@Ls;RPf9A6bB)(xruM
z&meZ3^y9%D1{zZ7_alMDQ(*^KT)k=qzmxELorhafA3t87Tm7LwCnyHHhMsKcBM3Jk
zBM^PVw#sZrDJQ!yT<MI3iT=7EGDbs*2ETxSv?>36`;Y9|^I=dDyV!Oq^oGl$f|iQM
zc9c<HZm$WAz|S=p2QX+W7r9Rsyb}-;)3qyEB%8SRAMnB$NOo%B$`9laIS;#Q?D{U}
zJ9qDH<K#4CbT3mcd?|W+KOqEd*;6DvXKuuj(C%_F$D#`5_|x(@hGVh4b%WZkf4H-X
zSUEuh=jCJ~fdQ<*s9&GB7SQRP^M*Wkdw5gpU;9Dss~NIDuZgp_8*fPZGzT2oAm?`6
z$&5DTFV>nsOy-G9H?)a4!h<dxE-o}+Xz6O3nAQsrT^kU!JcIJ~?J2bW<ZKzRU!9)w
zEveppkKxfzCtA5G*Angtz@h&xiHkfw3-&oWL!oj$#g$8*Hys?3o8!ci#^&DI9_wX=
zO{j_<;M9B#&n4J<TA=bGTz=TbMh%r!RnC-~b(-*NjC1G->w+yTzrPkS$hzuJz)@fe
z6J1lU97(5v7{YHr$0qtU{l*g@0?e?J7;H_lTUw+iV2U!H(|Q1#skSSiU5mdt4b&^8
z7frpUFNRIcqHl`ec7|BsX={nSkh;%E^Tmr7htW)7EV7f>Hf+!YY;nSM5KNCeiC=9x
z-@Zry$f?;+?52JBy%EWCQD*#cDuU){xi!H8hRszjgjGFUJdZXzH#tbEgf13_YTyk#
z#?SxzZQalXEhD28v=71r4^aG?RxI2MOBF@WW^e%xW?DldaFZljT!H~9$`(TN6&Bv)
zaGNPWmImhR&e@V)cuHlX=RXnvwQ5=rMPx8*CmPnj0n{IiI~SbHzWlW(KurEd?Ay1c
z#~v!5KhI0TL_4}S-a5TVXK6Nash=voKM~%WL7{W1^Y6wtFcs4Wqle&-uzV^nH+5`3
ztEcx87f&gE5W8OnXwBB{c5rxjDpZ$c&$m~CmFamM>q%|J#wCjkyBymVNxIOVfVk9n
zHJ%n{I*CK#=9MDZy`G>gMBl?ue14FYP9w|+C_?du{KY9c6|N$EOj)R1Za6VMqV~lR
zO&im|Ymr_$sQxu@2`%T#KcK(CC@q-n33V#xa!DZYl=+!OMOQI-dwYA?{G(T?N!kSN
z=&dRKH_!?6w;`mPP++o1VxWCkLJ9%h)uL4qT!*mPKsWvdC<S{aRlt>?L1RGkl;$kC
zf-t}h{`zp2_wZqLn11NbnDnE7(tiq8yGOL}d?g%Acq;V^v*V<}Dwul&btK!hdDmRb
z&dA8H`k5I^xDJJ)tlu!9#nMbXRM3aVXN2GX6ZH&Ju%jjg8RS@XC!00!J$m%${Dliz
zSiMDQ^RTS?Fj|qj)nq1LJbU)c5pTP3Xv@~EF|uBB38+{m(`@4}9bVowcp4d*j;ppD
zs88hO;~P&J0{jxW{$qDwn$5>02>`31f<eLDd4g@a_s51<nVmkmtesV<mSBVvb!y3|
zDMU0QsksXGJPxJUYIdyK<Ut7^?pGbUdvc4aSC03h`{eJRbqvq^iMmsU&?e~+&2R3m
zVbdubMmHoVELJU3n?Y8eN7uX%6cmK;h8K8KA)X>WP<HN%GgaP>h}3|cH4seCvM$c@
z<HOCk?B|NQqv7KAL_MjCkMvGLWJ3?-Q+DU`bkw2?Fy8WR><MT|Z8rK^Sy|VVW1ZvJ
zf05LjS2I{Fw=kmM1s|RRCe;vST$PaY+~2Q-;N@v)#&hLz%s0d)L-=B4?uP|_wX{9O
zPfnjb>z;k^tnqmj6(=JF!@2(K?99yLxZ-|nLE-~9F^Ry$<Jk3OGnNf?;LoW4m<ax&
zRyhB&uHI$mxLXN2$U!=A0hK}*PifA{WHJfW)P%B$$p<N}!+ZChYHx2}ziyod#PZ|E
zk6&PC!OrbR51#HnqwMYNP23)0gusf$NO1q&y#$a*3=DH9a1#*a{aB>hCJ&R-R9+oF
zqa4hiIA~UX$Y&2l$c$iooDdHsM|=N=d9(kDBk8~Hnuy_6-s}0RSBbc-#4z9YKj3zE
zgzi0fKqDsh()m9H46+jj_e60Z)zxv+1I>gG#Ff#@tTz7tBWd?9CZGQj$VjCBSZ*AG
z%)wwbalNgj(50|BA|fIU>8IRDf(xPdrhi^jze=t$Go^rg<*yhlZ$y_cXC{pn6O;)P
zE!9;{US#pU<(7I6Gc&UsXJKjYg&BkXH3@>HoM&GiH9O@$y%{MIq{Ql7<!%XW0>x7i
zYw4;9b8>pJ^TgRVZ{CPBt@#D*`eYO8C$tIeI!}(hBiqdSc2@KHD|w$keM&?YNu)vc
zmo*+#1O1d^^E2~o6Stf<1*P9Ah7@%W&+1!JQes;)*}!2LdSuH>-w|I8tG!o<{$~uz
zA#2B;poHJ9O*IsA>ih8_o&-3dfgLPM+LB<2A9M~z+X;TV{Rp6|JfxlA=E>+kves+~
z0{H3k7cXL7yx1C<{fRy3=(cv&eY1NV7Z(-?7mDY}>I5%5MA*LL^z6Lua3>v#H-|)T
z5L>a_Y9}6g;NZal%;*U^oZyI8@JQ|O5!OHcsDF9P+*6z*${AR*K_IvJ?XO?Ic&S_%
z`yjA9rT0cn;XUcV__ak+ikE6pu`&*@FFWbp4a^n@80oLT;%5WGYw;ZazMhS3I;o0C
z8*baCt*~#o)f)T&b;z>uotX)>{idt%O1yCI_j;u(SN8uubPoMtYK9q%UfC@hbbS-v
zd3ET#;mzQ69P3LtS??Tr(+o`jHJaYDW<~vS{M9M+45D$Ddl6Bh_wU~u01ak=(5*%7
zwAEdErD7%Px_1B7Wk<pYD9FKlKsgP2L|ac!k5(qX4TLW#B?X}|^w=CLYSPBDD^0OJ
z?l3{@sFOzrwqcvY!oBCRLQ7BY%e>V)3WZX@eX6J|d)+?;asPEi=wJLyTlHuwZcSg*
z)Fj$_xfk|`mVtqLyM52Moy1%YmlHE-4c50dvJT4oR%12i-X<Q7{DoDHhJis6dW$8x
zY{i!^>d=th3!{oW{+r}g%$vH714hKaX6u9OtgQ&*bL#lGZn2%sM;*@9fcte8X?a(L
zXGB&DxIiA}KMhK0z+IfBc{j!WDLPXTb8gTb4(Qe$Us5>KO*gmRAOwj!LRectylD!{
zel35X!}<fv8PMYG2Dt|h9`GSE0Ewp&$St##n|nw@%KY<_8YDY~A#q=G@ZRFx3$WT*
z9ypj$&h4PT9g!+R_2YVKk=f}&X0WlrYq}-=$&){jWg{Dy6XFT;<!ez=5<Et<%EPET
zKvfV7AqXF9rB2Cg1E;F5*Fum^6&YS$jK*wo3sDdmeHf9cw(}Ky=~|#sg!RYGorb&-
zRy#<h{I#kuB9|M0oHFDRl6OB2RUbKg_#8$|0yz>%Cp2hqh3>AdK8zYo8Rl!Q{V5`U
zZ`vY`LNS7Y-9Xs%$)=0vkqwTcpSTd=!*l~`OvuizPcFv9mQKuS5bIWmyml57IVi*x
zvoC#E9;hJv?}EIK{ru{PNGI-QfoCx$BPb1G$U~;MTVtuW<whB60dguL$Wfm;bH>d`
z2pF-=bKIXGI55;(l4Vz{Sb>P3I3XI~UuppykfsHW9I1g=n?&62u#8MgvbTDme%?9g
z;sc2$azuu&y!RbJ1nK8hRmrN6vWzLGe55I*ZBP7LU^QiI>}I;@(mzF3Tc)<)lWl|>
zT4bS=dDfr90t?ln-pA51QN(+}H8A2DUO9CepN{L|F5}UjMII^V6elELcqg;W%Qq7B
z(TTZpYU-Fp#nwnl73(>UcCSa30+a*#ARx~31)g*L@!UC7{15M%%4)Bf_6It!Rn1K!
z?oDS~)}>&APoEPM63Rv1o6(qy>YW48IUxPX&vmS6^{Q3+Ld7$ebLJuW-a*Z3^hTPO
z78(Ev8V3Lg&LIP^i=L8@@b0yk-B9CM#2-_@CL)Uv0^g}4FE1ZdCdJ2BIrvLf=){S-
z(Y~ctkKxpdh0fA$ffxXLvY41PQcRg+Ax=#FgOp@`e+j#D&)`MGAbLv=o}ZmPA|j$a
zT(VSLIfiB!8z0ZNcW)`#AJ>dMp+p1~U@v13je)bzw}1b~ZYHuL#JEAkv|XpKFm(~W
zOV(0*26gBJ{SIa6Bv=VZWM!C(B?q`crJ4GqB76a#hL4x`JR(;*0Bhh<*6HOZJf>_^
zBBX!XNB;h8Lu!0?5>-@FOY6nzKXya`cZAHZUb*r*ND4uuVV~1FPs*s%d}IegD<{7c
zqKep>hJro)Y#{RdT!TXZHdqvCTImqm4G0=r8m6bDd@C48N=|a@T66#cY(O+972>|h
zeYQI!6)&3WMBJv#Hf&Y8B*?mL?~70o>#P1ea;*lLR~~`=UegaAf9e7@t&JEX{Tje4
zG1>ARJv#E%9euR{y?s}ShhDL#TewezjN6Sa<8?3|3XIAOpE;o<mOF<a?U!kC>eL;?
zG%gzs1BdC@*rbq3pf4Jbi?bOlLD3Q%bmM51a{m3!AlKGQqXqM?HDE%(<kV2hU}l@?
zLI<LG+l5SiQ?0rG{PQy0z9##~`y1FL101G){|uOn^c+*77hq1%!MHTIaDp*((pC4_
z=WW4xzmNnD=tiR7KHm#NK5WCBR580&>2fl0+yR5>C$eXAo_zpAN&>Kya2UBHZg#z8
zz_ENS?UvcuANcj_SHbk7lhe6NIr&J4ZQZ)HYjg%hyL?0ZiGmKedF!?uL|`A#cMdt$
z-;L!bs-?|6B3$I<Z{ZM$bM<OJ_xmHC;Ww-b=yqN<kWF^g*M9|zr+EtC@M^iYrlw{K
zxxK<XP=?6pgjm3|s0UQBFeGU0)2B}b2Y+Jdt^u$UHvjwwqHc!ZeAC`P<mz&)b<0JG
zXG*IyLHGn<y(yS&$I+;YPKB2C4a!EphL%4E&n#$ejaFo%OF}2Smhm9di(!LvReIp`
zMak0Quq}NyA0`rk>+K^EO+0+yz$daagx|%P$dXf;fOLX`j)=fuVHb{!Bc>AsH4je5
z_?e|>NVciLoQvE?`;>%u5&0Q*$`ub0Mm3yZ$*IkoHogBY@rQzI=N*jslEKJi71W9n
z7LPlq8MWB{-u?so_q!S&!p>#*Y#9>N$Tid-1^fDeXwg88IB~E8vp&H^*vW|h4Im!e
zpe$4r16bEUPAl9ujp&7CeS!wJf0^oy8}Cs6lZ&KTtk~-g{<WVFe~dfY58!QP3rI>9
z=;&}ePCtOH+gVhPEuD0Hmw?_od90S?i^l!&y(b)cE-+dObsc)X**B+)*6YZPa1oTv
zdvPwm<r;XTh}<Sl)!sn*=#}<EYH@<Un~IT8%YmWHOO+EsYw%g<<ROkI=q85`BvlOh
z6`Vn*cC40}77x}1{YSfpf!W(jt1wC|q_P=j@K-WqwOQ?yBoT-Pnu>qAPjLX?E3Kiz
z3+TjUHd4*j);5SfarSTE*F$*&V<!;|J7C_1PX68<mSCYz9AUw}taOK%3edH69{?jv
zcIFiMfLOVH)>#5k5rKWAp$8GC)Im{lq-yIX5;oa|3yz!tFn9*qN=jrGM#}uig`p3;
z!9Kmlns&j3X(ADNH6$NEy&n^sLt$i{ED`U8RXmhQJgqy{f3nj(>&L(jXrSP9oo^W!
z3y8=L;opHd){>d{Rl_3`xupe%a@mrZjtvC4(&iC<@+tePg$>#)0eR`ORVb-_7@~6o
zU>X;Z^#2ipH0VACY;Ww?6#fW}7A@l-vFk?<|D=ZaO1TVa;qVSA9QiC7m4l9aeLUKl
zb~0-{Yw7&)Ef6Wp#d&Lx%JXV!F49ZW6rgT#lSg^(lUC)kttcGsUZM(UQebD%kjDka
ze<ce3vGe%H(ba3yVAhDQw#Sx>B?VhZm~=VfsKJ2T(zsj^$AkR@G{&buSb<x&Z|`yP
zs+WmV;fR`<neh{BUJFQxi>FP$dl7?kQ&kLsADLNMIk1P|1J+s0Zqqe-1Dj<qw`4ID
zE&kg4naj8aSa;EM+d>uWy%d__y=u873VK}{oNfY`5cz6HsK9c`+&1q*_o?5OhTi=L
z7J6kS>os2=w|LAL9Q~bDorH?-j(93*NgR=7=piMsHqWe~yv~g{I#FYvlCZRPc=cgH
z!Ky>-pHo2pO6FUJrJG!s3$q;P^$1fUcMH{3kF+b4J3m1ZVS^zToPz8#72d1UP_sUf
zi#8IADSq_uA>uU8g)M9ME6JrsXZkF8BPptm0tYR85oHhN48q((^lts?toeOzIJ_Y<
z+x7f4*fXgGA4DJGr!|Od3K={W#n~_DJ}sbK5-slbtF?A{6HV6XmN<dO9l84D^XJiU
zJFQ@X1vvD9T3*DdBTtcU?gx;K0H-5XFU?7A=GO|0Q}4+O;Fhq8pAm1H?sezRoqFKU
z$^N?XL8ckE!$t>nJ_I*JaMdc|lG^AMCe#)fa;sT6x@v;_XXMtfao1yn>~=1J>!~SJ
zG;vYj$=ZXvDSMO%qd3BQ&Js1{Bs_$(LG)SN^LmtDB9l8f;$lHSHgxnHdSZG*x@kHr
z-GXqT5O0I)gU};X2xZ5SUmvWdjS*BtG>}-eB}pyYt#}BF9@Oh^2S3-22D6Gd&GIvi
zqc8Q}B7*LaQwW3#`|mh*3c=E`KHg#?C5U=52W<FvfE4BPkog!<ixS~dvkQLAC=3ea
zLvVng$-#ogN#OhLEz=Kpyc0kG&%ueMAxxME8Q&T``<r-q9GzL3DVKv<JD6%(67fs)
zWn}kPk&huy5xWa)_?Vy!S2Mu+y0Tn&unwZ6_!aaTRea#cfNKJ(5m|Dq8w2Cz!jL$?
z3%Z`36s(+WIH`x;LRjdCG7vZT_KGzWgP(%`tff2sH|{6q*MQ!qs1)OPN(Y;=wl-5j
zDdBBG`^P?fxR&5_NEg5}MIam`@F}sg5S{iy-<2)8{uO8%x91|p2I6%w2Ef<D>n6W@
zrw%%Kw|IU?iik5;Y_O;1z<9p?_1OU@nJ05yLieT*Tv%)O>!yOM?ihQ|zdH~5FCHHM
zqvNOlCti*iXcO#?pE!X&u0ag&uvV;z_#rQ;4HhbP^T4IkBlkx}Mu;h0!POLT-tA5@
zE437G{LaW$4FexUu(g}8{C7kg|KlUJ|0mw#f5rR$-(S=El)+^zLpb?jg<*`FQHWCm
zA@sVsI#`c`WN)w_qW4l~ZWa83BG;zfWeMAgZ6zyfg4t4}l}3=I?Cflg;WY_yB}Ykj
z+=cl0QyKz%%L92#`4(BIXI#tHZaH`%VN{j~1YG&@7alpX3gj-YKQ2`IbyixyG4s!d
zg@v=v&a-&lN7xSaQy3v04o0+Sz|b-xXZPgwAGEwYaoArapPSZr`Xf6>B)c#`AUg*_
zpo?+LKby6l2#F!tMN7?XHP}ZwVX?dpgzLP@z28LtLw>uem$6*r5x5<3>;_0+Ci^=^
z=n5RZiYF}>!PDA%dQE)4FL88pJ8S36aSGzZR_H%JuQ8V8;L7ab!@!H_cut;q`)3@&
zdHYT8rA0<Y;+$6RT?FF-M|__mz<bzdX~4^qP`Oafb64f%1&FlW$B(CpxtwZs7wqr}
zl5$B~ijh6wKuC#AaYa2)=H~*Os@b}0mr-m1bP4>yut3F?_1yRYa?dqTr+Tpk0DO+%
zY&t=kHnOr(G0|8PX+GSdY#dO~hnHYNCC=;MPFr0+EzdxPE*ldKPEvkLmNeQJ=zD9#
zca-u<^3jtgUH?Yp?9d;$+uCmaZp>(o|C4n&-npBDk8;)VlL{w^pxpi!pciRSqtt~<
zw+!U*SAF!{bfR(RRU-?xb;k}pW#GW=pmHDD+S(ZFZ(n#7wPz2d=92jR699n<WlMjB
z^dY8<Gt@*f_$@M1@LIH+BG6+IN}_&v_-g~!#}Awwh%ewiTZF{SUMZ>25!+k%4L4KP
z;E)hqnnHrhza%#MuI@x73kzE;@x1{5{qg(%jM_lNRYy#$tdfXUpQncK4^CDKNmqqQ
zL}TDWtcHka5IUSV(Pa6N1uI)r9S;X9(oGe`V~IY4IY<S_0eX@KW2`TF^YR4C3!z4F
zgv^LlW^aK6HNe3NA_^zB(9UMnkfLwQ#(HbbTJQ#}g9+e*>+lpY(_A!WTXTyO5sp96
zr2q$v(<x(*m#=uGy<&i`lv9upC*XAvAl?J+hgpn%{YY9SY2R@V(89637s0DDii*jQ
zdjmDm4@%3*WUQ_s4Quf3((S9?Uu$bBtWrM%uu5dwWTqO;1uOA1!~}q0K%e8-<$E}4
zy^CG|NBqUpr*8n@bgVBBh9~#}e9>)NwnU?CV$|XVs=~p~TBxriYH|GX6)S>}I0V>2
zs^TO@W5V~!wy4~W0D5K<$<(bo2;e~nO&?c;gYZd!Pav6nIB^H8^8=wUNT)qGhB((H
z0o8Eiz=1Oaas&oIv&IaQ0#eL{ycYn#uM2NiKSVs>MM#MCc+dAHgzUUQ?D}!2-qFMN
z>v*+XI#44LChk;SOm8BX+Yk{xO{ABar(xB?#tlN;@R$9=)oa#R1BsBk5wWNN#@nw1
zliGdc!x|9~{E%u|CO<)3Fz^vkLg|4e3a4$}NgM>p#R)7V1P=!e5*QI1x`rS)#Q7BJ
zH6nlO&lwK^k$|DlsyQ!%Q4Fj?<g!d$+gE`fc~IE~j%{FvMt--SI#|9q-8MY4iy(sb
zh5!i-c$$`CL6>Ix2jwJ$r1XX<;{o8rLz9E1T3T0R#=h(e1|Xw*<34iH-4RRXS87Pm
zh)%&m4V)7a5cvDpdd;{EUODITr}gQP(R{6xC7}JBjXbhOvEB#|n4wnYSIL#N%En!I
z)$yE`zf5T#MaVw?e(ruE=wxdeq9(UkA5BJa*XX`9n-BK-<1unUzbl`QM2M1J6VtqQ
z8@7cGLMvSJQ`6JAqQr@$c-PHpCUnG=1`)SNaW=&V*#Kobu;6gPRd++XS7PNcAVE{l
z(fj715N$g6^tkAKwXuHM>gvC;rV)uKTfcT~!cEs}78Y(?&AkZvNOCW>Axx@mMGH7I
zT!Uo3XM-u=$Rth_XI9Bo-g60xMLb2R7J?%iX)XFqzyl;b2XI$i^tqAGMCcsCu#pVm
z^0&i`IRa_8FmwoE*O5zr&w0&0IHF999>g~+3`cxhLMFf>Xh4Zjif~-cLS<H2%mNaT
z22tXO12A(qvJ(V1|JSceXswAj{zDx20heM-x?Zi0Nd5Z!kkrs~Sj4F3sHo;}HfwJ!
zTLJXZb+Zle5RjBpn7{xY$K0h8)(p&5D|$KPbGzl*wY4&=Pk=qq=c;ArQ_Nc+q7ASp
zf(9_%ex4hv6e10xChOzWu69rW#U~A4Oh6ppo`ldh#P>~P62~&hRk%$?Q=D72m>a^;
zqcOs6g2ks`iy6Xq&VXZ%qY|=X0e*g27@JA+`~a_*a?N4Nm}sHJR=+-G?%V0YRv99c
zK3<)5h%JD^XV%npF~r(cITIzZvvCMPJ);KYbLZaSn<{Jufx|vN_V>?Wh~dZ{f<7-u
zi`Wk>7BXsO9>A5vnLD8*&h6W;0{=H;v?9fBJT>oUx=>24BF@gae^)qlDg|zc)KJ!q
z_EY8dIOImpU=o26!(7lwkan+dpSTv8wZrFY^Gg3@50iAN$e<ufA(#tG!T`1*c@mYl
zw{Z&ZaS}snV{jm+SXGACA-p>BbuBR%oQ<-J;8-Hl;4Hm(OhV#k_qI4DbE-s?INXa0
zEzRAAqV3x5cvcq$P#7t6`WT?E3329^Yu*(nHf^fMH{!b+K$wPpULqppc~MbO?=zYO
z<-;4c;|#Q>fx&BfF~H&PMmw|yx~P%c?~L*nk&(lhzhPR;0kGH%g4P~Be!SK~Mhc%l
z@&%y?Grl2W{eZNsY;)I$ufu3OOFWa)&kgI>XFK$&(#LSlTg%NeLCVwY(S~N@OE;%}
z>xa-iCi=__<1xDsaYVWcY;bco?8eFIlBE`sl$<V1B;b1SAqS)*aKHe^iO-zTUcWT1
zdk$i!!6KYEt^oGT?v0pAHk3WV?BFn5hfXI#sF>56W!J>hbg)sL*`&uU<^z70%c580
zNjxxHYiINWBY*iAMo`_R9#AViArQ1Cv4Fo~{yqYI<?7XfMZ40L0|U-{d;|w`$J8pr
zp@?ySnESy<2207!%|=Ct6%k>Vzt8nIIJtDmA}fr9S4Fga<GPI-fq`vuLe|JLE;p@w
z76T!eXZ&se-ewRPEsj=Ead(LMF0aFi5`rylr0pAx-1K#<7VRXy#R*U13rUJRrme7K
zmr9BX*FDO#{&kco*0y!0+zvnC@5Koh4EH~b-SgJZ;JVf6A8%_nCBDc!e0_83so3o&
z*Pq<IZB<RjGCq|v2mTa@x6#P2k>@-oaJuIB_k<5_+iT{FgHCkPI$gQb+ing8N_^|M
zDcREdgg)L==2^tnvvJgSKU@Q9rv!t+SXz)bx-q?w$EW#o#*$Z)J#DB6pQ?zMSasiA
zB-R>^L>~Y*vC)}+mHNuFcto&+Rp}x;EP5v(tQCp0*f|Ng==O?lGJC%LN|<VVxx-Zs
zJEZ{zLmSHsGp@F-u2^#TJhtM&#gJ-v$4`bOxK7M^wi_dQ3<ppb0+sGyU}$(Ke2bH0
zILEGCFjX64Xp@tZ`NsJY+Q!DlS;ggYANn1Wl9JAAYipwhtE2W2nQ(mQ$nOvtV@>!=
z<>J1+zPthgDpV@9gfe8wJ%=!(!`RP>si~?t?w6(?J)RO@paJ<*KHeN&*U%96?b|n{
zz*Q?(_Lbh>kTvIP{djcLUPM$hDI;SvgeI`v50fMDr4~dkm?A4H%M5hQ&B@94s%955
zNk~bt>ACBh@lsU(BMvQ4o11^z7iSd~8X}MC<VJb(=0kS|j6a~oZ>~?a`*nwu4_y!z
z6^)CFI|ximW8`*yW!8=0i!wNN?6~3f9XjQz2$nwi$nM>jbf#~)Y#KiEnMTvaVQ9SV
zS|E=W*`;}$t$0cgdWa_BAl2r9Z_Tij`@U({t`uyt3fo+d=D)sj;#yjmETvZt9z59W
z>+0#b5J<y^LO-Jlw?Y!%aFP9a56GZWQfev{nARdgFV0Jb5@`9{2ZtR-Ot;(W@o&cg
z1J!vk{sW1qZk=jT_#pb>eM-8Gb;Xp^(oKazzsG^#Bor67cKhsyNa2GTg43w(>^yac
zc~%eS<FU09GBYo-va+H_Kb@POZ!vKg`E?Qjg92&h{GZH7Xi_*cy<_8fp0k-ycG2)W
zLwt65dU|e!ro=`+4E+YYYg_*KexU2p29W>nUrS$BRGbf&aJYKDH~7U1Vke>NhfU1j
z+eNf=b>kx<xPRJ+PEJhpAqkLakZMrunNNJd3!I_cd0&2E9E>1<GlHQpJo5?o;-|CD
zg~rGja&U14rKhJ8r%MRSik`*n8VKyd$|Q6h=x_*a`#;=U(NIQ6>h%x-!#JQ#BAwga
zDtvUy=;s*1LD$n0?=ELqvwAh5OM`>A=Q{MEr=zvLPBs%fj1c|fhY!^a44Sgjy+(fB
zE+{NC=7t*5c67`}SkovX3sE%GL3kGyO~Xj!jf<=qzwGL&2S<xIJ+o_Ld3KuD;#>}d
zm`E&OA5URR8vIIkX(4ba_zU3=Hh)<85xgIfzJ9c=zrO*jtyrdWUfN^o0(2ur)rRa&
zxmQ(HsyW%&-{xdDXe%h(#TU`YRD~$Zi8e`kFXSOoe<hcGZLzhpv$Mcwskqp4_V}QO
z8S@w^^PI%4jZV(a6<4zoi;GP}L_`E#Si{obKoRv$(uYaGkcHMC+t}E6B(-C3cv#ok
zIypkx?QKDUq?4oLx34dIn&A56!1G%Y{AfR^Y5RO92ggMcDb=<)?+s>rwfUmt<YaA4
z&1hINL<~Vf!tMO{&y5ch(oKqLQOK$F^)bG_t4m2od3v=PHVN%TPENCFOIausW4kjO
zslVLZ5|75eSQvF*zC1)R5Ed4W!G27PkH2P;%fgC$99%poQ;QsfZP+&80M<4(7s|@Y
zGU<$gp`nO|DrY!?gaE7TL~XS7>63qk3^!v+0<z>LmUDhlImAUaC_Kc14p<(J-{9Ev
zUzjg&t|T4h!#^M(DLq{+JUrYT>OM@+U~F`hm=kc&hCi9gi+_E5eEj|W`?34xM%wQd
z6ciBmoU%D2ml)*Wq_lE>J~hTZ%Jg41x_Ucj$Kr#Mm*d2DyAfacHu`_V9~R5$EA*~Q
zN=gbJJJ!xHW0{cy;ArJ8y`w<!*|KwIKVuKeEkzuDzIFQ{@x^z<_u>6d{=i)A_$fOn
Vhj#cQzS@sP>5S^>v{RSv{5PwYeqjIr

literal 17031
zcmeI4XH=8v+V6voZ5$LuWTXx%D59VOf^-WiDxlIt2oaIq!B7GOEF+2vDgr7^rME<C
zD1kUqB1VbS0D%BTq=XiFNqMi#DId<->+F3#oVDL`UY4w7fRH@TeP8AO|6Mmcykv4#
z_*aQv5eS5^(Ye!C5C|b}1Y+&6jqBkn$9Jz;1OKhRapCM~1Yhv)O?6f*0`UjJ==2Hm
zprrBMuv<Ik-tyVh`|_JM8U0ZwvSxO-sa(Y6O>6F~IWKY|d;jTkQ<fR?Q5eIFh{)yG
zwE9TOllYy_9$9S877==@=cBxT>*$Fc83#5#IO4MZfV^<gOOcIxZ(_q|KUj3W3j5@B
ztnUX`%b7pi6zW{_s=q3+tIImU(K;T1P_Pyg8`*_GtT{^*`e0BeiTLgGKY8F4Zi!jw
z4Mfj;)rbsHMtn<`-@Dgkl2MH<U|A9Ka&u?73w<<KxPRwOn>Jzb6jxEiVG;Iv#M*7%
z|4ARH=LX4Y2V9JI?9yw9mX;h*e3vriE%XGL7Nzl})TMbhxid2(!w?f?aQEatUYm(?
zrXOzVhRnC68^ykejZKS|*3LCfJT@Jq(@hE|F|aG1hwO+~<>lq=yA}o$P&lJl1=s!$
zx6hlJTIE?5m|we=izo4};j8CgTjpPV+tSkVeoS)yd2DQZb;v@huGX#LH=C8q4H_F8
zqn<pmxOnlgo12@togFSJ>eul;5=uL8qHTVhKKC=xd$hCA;TiA9o#*lKnMp}f$l$@a
z&rMC1=g&vMtBHw<rqxDn&CAcfs2wo&t>KAeDVB?o)(h42@E_?a&hjF01B9iu%#oG1
zF5B8B@s_CJqa+@SG&|dB8tF-z`#fD+iDn%%66X1ibsiz7I*fG|Ml;XG$c$_9=IP#q
zz1FU-t{tz{-9u#vxw!}2=f}FRUwiw)7KWQ^yDQOo)Ubqv1e=8({}?ebu|fa7a1K*e
zn$TVq=uKu9Tr2gHBHe6!KGVP%3g=}}Q{_Yo3k&7A1d*{c9IU5P_0-4Bmh>`D&jsJP
zb}MZ!htakiELB5Gt6K{*n>}8}2%Rxbr4P5~C6wN5+|FvtF~jnc;^N{gPn@_N_x$<x
z7QWt%8#nToW`=CCOjGA88M6FE@`J$7BhQ{b?VpshZFG%AAOpG_yDR*&yvhgGr?MLA
z>V}`4m1*WI&73Nlw6DGIx-vg5q3QF6iWc3zT`im!TD|!FQB>4-nn%>)Vvuo)j&;cV
zSaT`Oj)hMCnyBHvFwt8^U;OHgWiIe=)$`XN0J@Y>#Gu0QqY0FZk=87Ca<p{iV7yHm
zG1#B9ASo`MCHNvWMOQ-2<!K)?sSd8-i3a1F?m~~liZ^4>@6VekVIjW7hCj5mZ4!@p
zUV%U`U`|vF$voMmnrX5!(MKW))}hp+>*yQbSOw!2f2yukR+vq^MHYLbE&N0YX3ovg
z**OGRbN#P1^g@WQc^1E4ek3!|s|XpeIAvdetm<GDWoPdf+qpCM%^L%+z$s^xnVDH>
z@Ql+FS$(}>&m{I2RqF6-cU;a)z{L~9wTB%)?|>5Ui+aFBPX)SrVVuy1+j;03d_Mi7
z5zEcE9vO7@48&8sGIDgT{rd%#>b)^!2knkjt3KG_<w2Ej*Q(H^=|L^@=nI<Tx^?Sb
zAGW_k4O`r!rFHE>g0h9BrR6vJ@9J(Zuv`{u#Wu!@vGJ;tQ|9{Leitd^@~M1I%nb*J
zyAl_OYXg6L+E&e(rc}-}X{!fKT}O{~UN$l^@~i5Ndi2OF%OqLNt!=;V%2=^h@O1ik
z0A_A<a8B<aIYyR-;!TUweKmS-#WF`kj3kcxr86IGt1?sAb2hrI#I;q5fV_G&`>(qj
zloS<>4qYqxCZAyIH@|ZuqprT*wLRD31Stwnc_eDKZSoIub90xUUtW#&R20KyK>`OJ
zvMv8MR1V?O9^ALGFiG<y^kzc3Dk&(O8K)Ee{*=42pdIZ_FDsayp1#xV?CI%wiTV27
zR=q_72XeH9n_DiA+s8*f+IgrY{*cXVZ(v^zk!TB1-0@I!-|Pa3k7RWew7yU@y9Jp+
zulX1^X6mUk^F_6clvrM_*&90ZHQFDI86-N8w~TidGKoV-<pi0q*~{c{-cJqEWh}O)
z%%khw^t7Lag@x--Q?g+F&_l_A-O(&Ybrtpa3Nty7(Y(lJ#-*g-UcUStJ=$hedtbQK
z%v3i+utLEr!uR#P-Tg`%`RpK_4nuL|4>U)Ws>9R_m>X?}kia3U@_+oeiD%MFTph-{
zipjp54YHyA=zM&hab>{xXi(3citZmh<adUTqzKRA<1O+0ouey-ZtVx?6<FvsWlXzp
zE}HHOW$V4PxVZRG(Bm=ORj7x)IyeZCxus&S&i8)>a2VAgcX^YwMF-tG@~yY=ksX77
zwJX@qeptKJg@!}gB`j59OHM^@fh)p?bKU7yTl6_ocX>>GUac?3F#(xbUS1xtKv9~U
zQc_kfLw3}Jg~;#Q7lQnGOTM*U9F>pdE(aq6deJmRJv|1iXmr%=5UQq!RdhM+OxJ|z
z^AyaeB6;w62_zrgQ5Q3N2_i1ktJbT^REK@~l7sPS;#wbwQ1XSWsHi9@K4%WSrKK?|
z6`LEQ&+GI}F0diELF$)Ldc6qBRbxdF3!c7B+WQmiFgZ?NUuQT{;sy0*d7^6a2RrB%
zRKFRoU~Tt~gJjRH`jFO4<K^#<R6NLCT-nN23kC0*rOfdSE=eX_D1jq0RDA*x+hBzX
z8!Xp(+4?BnafHt$@x`Y!W4BErJbKD6x3{;KQ4Uf3h-*Gf#6{~f!uesOj+OqqTcr|E
zK?8pZO(q7d;H`_DD#y#{vtKp(hFeYb+nF)?I1VRtZk%ofkt}FsYbs@yA9obk(3Slw
zJ$$~`Jshf5+$AfEXW^=v0v@++exZ!4>{PDmzq46tvJv~U5Y4ird!FNC{Pi!%Vi(_1
z7s4bYze^*sww_7y{a&|;?uf3qI5HW|uR^hwmN$4T{JI65@bYE$9E-+M317N!iN9}T
zuYy8DK-VLcn(VZ*F&N}dcsCA*ld7xX^yPOt#bm3rRysVL-t<;k#egGd1XC~JVk7Mp
zgeBQqBJ?NCwX2WfAASCOp46d@<&-n0uj9=ETk@@opL(~RhFiK~Vq#J~Nhm66VYg3L
zWSJ!4js`CcCRmu6G11oW{X2=*Bf3jmMW<eRleouzQ+4fp5Qs>^k#+ajdQU8^M3HHd
z&!OK5P10$lgoh6w3V9b7RzSVoy*yI0Zu8Dg9-Y;iMeQ<mQ`oz=6FX-2JpQ2djjAsb
z?P1yL{I4m#45wx8VccINgs%{u=8Jzp<7)J}UOjuZZR(h0B8od|6-|ke(anvN!JHFB
z%63svcr*C~3VNR1=@8TI%I^!EdBR%==hFg~UYXiL9?g}>b}TP4W2F10^q>y*g$s2M
zGJ_0vu#eS=S=kQIFf|jd1LHeOYk2Rb&?!1QWc7~rK9xJ=k&vtte8tMv_fJ9TfI4yt
z-e{<LlWl_1vnHQPkHh-<C^nlNZS_jCdQ`7=WG5M#;Zhf0ZrzH{G6lhp43d0jnyRO#
zG9w8s6tghlSw+muY+y-h>+=@Vm;ox4!@KE@paWVC5qiFBGGw8?1=z|+OCyKE?p>h7
zACmU|y$qTV*{bBZ2cEp8?!K^ON#LX%p0XX9G4!A(zaRO?ng-zkool&=4D!Y^!O_X-
zfUa&YsqtOY1HuWrpn-Mt3k}x@UIosnVI964D8tn?Z-a-w$&r0YOJBW-cH{j4RdXLV
zs|?@Stn4E-xfO*n_*Uvp)0*648``VdKnq!acfIhhMnw4`%6i$5u_JW8Xy|w;jiZ%S
z?bWMAcf5T0vL4=wAPQ0VO4Zp7B+!W`-cj)0%5ZyO(~bN7ZudB(wz@aNcD%c^6_1@$
z;*NWV`-iq4YMk$QpNd`1VV39TAFQ71u&IP*CMGGF2N~r)JKT)na~IIeCmQRyGn^NA
zf42WnlLZOOy|aAJrCj>BU)}t#!Xv0Y%*_S6(@@2^n0gjo92MF;ucDYm5l+TgF69VI
z9@E5%3_%U)2wUhKqLrnGSEHsOP;%1KPs7UzF9uMIU+c@h8)7K%{eIH5l4MZWrHt{Q
ztPu0XkjeYkr9q4my%Uem_*C^;T3cIp&W}^8jYp?vAk4lB@ohGCO!QV&H}%m7l49cG
zHlq7XO1cT&-iBw-+JLMqbZSa+6<jHvYv)@phUEwLv+UG|mO(vomm4w+uvUU9Fvmji
zyFY*VQgW@#!?mK<iv@e}{-~;I0h>v-G1Xg2pCgdC<3!)hI}dK8hbJoNtf?Su>a*`U
z-Ginfb8~a2EAulm&#2TazN6AZ@3)qD(#=7@WJ#`^ZJ8dVNcSC_=q`0XhqBkUet-RT
zrb+VXLgYugP<DL;uH$M|K*3LHDn7?7s~e|!{X@h9#+nMn{*s4sf5*<^@^k`;V+h^F
z*LJ9JU+OK63n`Y*o8)hC>oGkZ=Kg3?O31`<$cuIA9Za@MeUzA7BBds5MT%;iN=ePT
z*o1~1bRYa2ClqsPSj7932ffUg#2Hjs%X4BweP1Hd-|XcwW%=KJG{#C(dsFht$}Cl#
zn||v`?h&lqm_}Br7W%wH!{f`a^ae~FnHYhrgmR{mKEqxJG3$e}P1cs^&M7EpuX`w}
z?%AzIPArrQh4?$>(fJ1Y*AlVn%`DzDsW)V@dbu5s9h^BzRJ98kSLec=wSmG3!&Q0r
z=^_2Usvp?Df6ATC+pnzLp-!_+Tf1S)J&(k>obvLkU0q$>MUIb9EYboaWRy<8qexVP
zYVJH8%as@Tk~g7xkkUd7o_P@$mm$crp@LmwQo0y(B`X}b7Q1g&d1{7EDc3zd&QJ8_
z39Vf}9HoT2WM`MMev4H0?CdR2$vGt@=BRLPso=t@r#~wlICMxjXaz)$8NsXf7Ud9O
zxRQl4v$TxAa^(u_ZQ5kbiV0|gOZN6}JMbvk_?7a|#evAtj)Fu#e}8oleRCtNXFcg<
z7AH^M*)~*5;?CO$%jjGM4Pz#FUtx)Kc;EVlhFq{l;0p9-86lGbUAH`(7W47gR!UI`
zsAY1<7g>F*&fY9NXy1Z@^Yyj*1%V9uaC>b#9;Kg=k&%ODb)3I=(HxB~J)opC$PO~U
za^;=C+EoRNw?Gn?hAd1dENyDkni=vmHp3k|veazOFdn)b9&_*BFD$ZbcnMcykV;@o
zW%K)9k*pGZx_@y=KXBlH2g&?IE?rUdv<!Q_M9$cl2YRBXi&X^8ta@dlYEE13-VXSN
zVita#O^M64zNa@ezYDAF2BCT>RWEE7bb%XKH9J~PetsKB0TfmBRJJ;YFe#*Yt;FTo
zL^*Q{i53pE>24)nTVkhO@LJ<@P{+>cZF}@-55quo0OUrivBo;Cy@cAqI!Tu`HH+Uj
zgHdc_2jC8G9wJIUfC{-BCa(D*0rGt~mCwo2Tlyhdy*!$S0YO?j{qXAPXv#Db1DQ>8
z|MBVB+!0xXOA(Z92!{!3Zl0#+8pQhD48-+6z6<|9{6GLy@NrOp%)zqA&IpbEs2ECV
zO*-!16v9i3(Ypdcu>0_y58<%3ZhbY?`gu6Nm10W|k9@6wv8x1Yp}98}d6ssmra%QE
zw@8l$A2V2l0Ql!X=3jv}|8<@NECw+(I5kj<0SCjm76_ftE)@lJ9Yh+_KNOa`;qflf
zZ=8s8KGyM$hM4UUu^|2+c}IYq*!Nj^dEqQ^w!7xJ2y{tQtAfkacP5iH-aZBOl;x;z
zfZXQ&yPS%O%Gk=e^XFT#FQ<L1k^N)G1C7T%AOn`sO_j<ZJhD1FJKs;N-6-~5c#j%1
zdJRU9!_|e$viOPHm*g0S&uB8c%RFt6=ynTxx}Jzzon*uBM{hoW)+!S3X_};Y87j+W
zO4)6nKj69M<}bV|M>5}gB$_E47G~eS{Bc`XQ?hm*iiDNxQx-9)lRPs5CAczhQl`c8
z=g<%ZD8a2ngL1geeQo%frR!TRo+jESU20f?I$(*nt3Fw#`=&+&zPo47sj%hQO^szi
zY1=b*GY((b|BmHXp=ST_p2(N=+2QcQpQ%(UbMs!_(iRD|bIDrhU%%dqz?>o43(BQ{
zEup*DXI;F0^t-S_yL^oo!G_uJWb=SW#IhmLKEJ2O7P|fh^<(??2}-Xp`*FnbDfl}0
zKw()uo$SJ-qqj$3-F9BQ=iuOg=FZU~?T;C7VR^b?Bif4`DG{(ys%OB!+rQp+Nnh|(
zP*u%oXE&(bkEu;N0*hn!N0tdF-ldfDYQMrFN`i0i>+?R2<pkAaZGW<ceV@H%33~x|
z^n}9+V)On*(ZRMhb0}PDoLHezA@-TH!#6*PlzR6ue3OYpor>NG%<F9;c6E|ZPQs#7
zwSV#Z{L7O~@U(#)&Vtty4nLK$M^)7f`unq#^>=;*RDlWbXUsS{ItsYkFHZL^eWFmz
zPMkQAp4I$DWYRwrx_w>Mw%qLO{sp)9;FM2eeih!5WsLaOK<mFPXMaS`{vwhD=t1Cf
zeUo3mmawq0%B!rr27CaRCVW6cZrA7wO$%F@7nsuSLWjuQygZicPsGXFK^iw86yQV9
ze{S4(cg4)C8fRr~okK%~BOi&YbU=czg5wc){&E%Fucv3Xw6xUy-u^Dgv`ij%<t*kr
zBJ>Z&;wyOU4jq4)I~lH$KoJ`mLA*af^i2UPJn}YDrX@;D=|_OSld#K!2O1uOKYzM4
zK9`4GL8v?k+=H$a*7!8&*Er~x6Fl?tys@zb@S^FNnNpvDliaD=tyoXzmUN>^Pz@h-
zY!J7<dw*AW1Bwy$5xRD30t%b+@#D#_FOQrG?D3713Fr8NG|-9;|Nd~_>@o|Vm!7`2
zKP)RFLlGj|$}TvgsJIw><vR%Mdf;_Xc?7Ql$g;$jd8PnGvWwg*yJem{o8wGPd9CL9
z+|SRit<d3YscY*QnV7QzB1HEl71Tcx-!P}z(9qzvI5i-l7g~;)AIkx5a^T33%ZF^r
z+5ik%0P+P$!v*gKTcG9BA94QtdEiayK%TP!^bB86muz7h>K0F_0(p;5U(eSJ7<<X2
z9)b2kDvOPabClKz`ZnYnJd%;H<~0gRLB9Nj_^UuErdK+xEG@4AdlR5y)XML=m=fl^
zfCpcA9v2_)RA61)FvJ74pWjvNyuwvI19E<=&Qt^_91d?+bufsnwLC73%=SDN(-^yt
zHuAg1Nu`Gi$6sF~%3A~60vwzxMX*95Ju|YhR(j$ogGHI9sa_CO8@Ss04;?B63HHMu
z#j9H>q9_-izi?q~Hy@~L*o3y6++W=QN5APaP1kLoT-K=(9yvjz7C-lAd^;-}mJal!
z+gF-E7gz&r99Z8B*g3HsJK9DL#6hlmGKgJ6ywJv`rjAVGL=XL>n?oSbn_nJrkfOcP
z53j*hj^<WgRJkrZzPP*SfyQ1Vfx9qC)*38t0?ErPCjtN!Rk*mZG4x$pTIsEkD-a^<
zs-GyDW$u@glglmy$5e{!NISc=7sDzz(R3i_f4&g(lfOMS&3F07;HPJM{+8nGKmYtw
zLMQ0`^XJdo-WhI2`ZdZ61lU2!bR`x5sb#@6v1jgIRaI5huYW+`DKLw0*Pi(H_LP9f
z2}ojneSPD2fYR~f*1eVJR=5i_z(0VEOn~tSea$MAM%^<=H|<eVGiE0(yds0<7@b)J
zK7!HRx&m%RGg^D*=UW0<I<RLzbH2mIdrqr~c6?IxGlf#HOV#PBxLwd&m=QRDi2aKB
z`N}yoI%Mjv4b>BYkXmI9>$l2`(F(m7#0vErbsJeQ#wde5oGAB7kM=%=@o8GoA=(Gy
zGNmSAbryB>cJ10V^n?P0PO2EzEp>$fDL%dkycr|SG$rcF#GsoE#e=0MmHP9n@bg*|
zfVtmGw^sNMJ>}!}phG2F?gphsY8+&TbC&v}3G{E#66&9QzW{L@^sfRKQ2h6@1Bh%K
zIz|lczrCTvou-VZgjO5T-=5ll7qqR;=%rd&-W#YZ%laqu)qXG^eUf8nDs@U;1{eG#
z036gSJwC@jmA*Kdhrh#!OojsO!v0=gMggWT)t7F3{`~YZ2i%(n*%qjYR?yTNAlx~S
z@rM3Czcj<Br!C7=7WQOGe<{@^n=JmUURz8v59U64H1_-%n*?Q~2aN=%%cJk`=F?72
zP9GeOdG!o0HX2fjfZ%zc6zO4I^)OLU3z_$zkDoC#R8&%85tWXEdm2g&$APyp6^w)!
zVGBr7wDBq3ng<8%f|>a!&UbM=+VP;Nekv~zhCx*C*-d}Gc<~~FJjSS|4{7Z)NqUcT
zT1fZ=cMj}2Gh&G8S8<D7FYXS^5R$1FV7Q<{YDT+$f{QjJ`gQ=`kxEF2i%SlyA%F%Q
zNNTa>aV&grup*1L`#M9SdcfTuP*&dJHnt0Fd^5~=q)_!5f+gL}c)AyoI?ubWCYlvg
zwS_a>oHp>fg$lAqr~k-4)0Ei?6X4m0TbtjWihPlrT%gY#F%?1;hvCl31lP^Kp1vO%
zy3M*y@_9Oh;38-1VgZ4nHCYk6&$Q#XT_&{7ZvZN?KyUO0&^@o!gjUQ?_Q|##AMWGL
z6YNsGKa<6qf@-%5rZN+iEM+npCL)3v2c818u?<E`cnUBWhpPO-{VFQ7=kgbtu)c<n
zYe$bBZL9Ei1|MY&G1&n{yB(km-P1N~4nXMZW%i$ebyW|mW?L?lI(?M@A)}2W+^CP5
zVdU#LfXXQoYW(uW9D}JG?`#+bFupuiOr7m=PE8JL8V_J)B-pipT&#{ZGBL3X4JFd7
zx};D+wXU%W2cX}zW|`6=WwFCEr{tlZVus&FdL-a&2u%UjpgN$viOotmHzkcM-cM14
zo0%Qb=Xd!^SFt`EubH|Zq#*>2RZ&rKv^gyjh6N?i(WEtf44`sHJ$%SM&I=^1Br`R>
zcmNJ(-`yS97ZzxM%7}mV>@8U)<RUoh*V=(z9|*UE2_Vk|22BDgJ(YA7n+s^Ji~{fH
zp+C^w7Old8t35L+pycPuiUDZDIE(5lO(~iMBs>cIceb);(9e}Y3>FM)Qo?z`VNfar
z90zVG1Dvjsyu4vGX09bdXsz0<;Y%=bio7{709q~0m)YRjQ|_w;tW_{|fZiq;9)ro%
z3Nadl<`MVw>6^!oe}k4j3NE~~_a`(#4}7Sr8(Ewu$9P`~LjlsHO?ANdwXdoAD46P$
zBAq`(2$dz&y@Cie*>mRcw+GwyjFw1eg~XgQUReAnlV$9a0*mh6lyqEhwQ!RhnZW-J
zV(gQwnI$0$UWcx}|1gA+LUDGm8>j*-1GuAYtgCo!ja>ym81S`%IZ!B8X)DhWAblkS
z8;57GjV$gg;+<VobOkE==f=insEQLalIWzX@fn~zAfGQ#^;YtuW%Xs+0-^!a#wl4H
z0P`@MrzlYbtz4HgRV#7$y2kZj{Qo!$5V1u{vwuM|TBWH!DR7UajZGem8K6<JAfW!6
z_5z#m6oPwpj)rm<3<O|0d$)!hnf&hEZsS)+4U3(=w!F8mRrBdL0K07pBU!iJ%2HT)
z3na1toKCP#O7Torse3}1S8r}_Rp170^PG<jB$j~rqsO}~ptlGxOIxYCRYqp!0XeyW
zu{3~FT4Bp2AZ*aH!<T{Vu>iP#gPsbDq5xaIPB*e4G;}~VK`d&;=-6(++|e5J4VU-|
zrYZT*R^(b*UCvqh{ueZ};6~rP?WzjIyScmX(G~t><R2k{dnciz?a}>nePmFoKvF+T
zSu?uI0{-z)i~l22w8{cjS->g__!rN0{o~A9WdW-!V3h@|vVc_<u*w2fS->g_SY-kK
zcUi!y|NsAL;Jxbq|DD^p>i<{$|Em9A_5Z790anifteyo}JqxgU7GU)(!0K6m)w2Nq
ziO&N3c4uA=ej9(U;O_#gvVec5dHl!ouz&W~5&kzn^}nzHgnly|az1DJ_AfcckV{_;
loMuEI^#0{T60>|E>c@c&k&98K@RubJMusM*aVM|e`49g4#<u_f

diff --git a/docs/dev/articles/functional_api/unnamed-chunk-11-1.png b/docs/dev/articles/functional_api/unnamed-chunk-11-1.png
index 2e9ecf2cfe0e2fb5f0e94e9831dbabf71a1ce7e0..42e568aa6f776d04ea8641d02d8a3e2758a53d45 100644
GIT binary patch
literal 92394
zcmeFZS5%eTwl%oaZ4PX+sDJ?^gAzot0TDz&at4(o8Oa&90TdM!36fQk<S0=<MMSbB
zi3?PcWF<<jKH2v^)I<GM54WmSYyaW4dkzO-t#8da<`{kS-p71!MfSqhP5U=dD3q<z
zQW6Rj%3rzo|DPMy;V1jponGR<)?JsmAVFCs|M#LW?iq!0h$1a<M#(X3nCkSrdTM$3
zciHo`&(aUHDzpZ@XVhmX`0U4ITIY55>|GTGzfI~U{(m!M1WIJc-1$tO;miNXjn?qE
z;fem43*r;uyf&Rip|oY1GwpVTsg+K}+5N4teS)c_mDkhN{=!@1>$%b~x%K~kONm~)
zk?Y@YpYQ)4{w~BfZVlzv7~i!Ie_eil{C&eUyw4%J01kD9YuBztY+LiL_;0);{r26v
zV{H0soP@UE-=m|3TQgmR{-P`^aLE7rxVPe;ep|iri2wPHLivAl4Q3sMu+H?#=e>8y
zr(ZAKQfW!ekI^o$7kzjC8|$i7s}i*GGN?kK0+uRg&wg0ry@5jcR(JOHX3DKC^)Jt@
zcUzk6_w4mpoC;!7PMD+|y`7n$lFa7e;Sp-!FxpY)Lno{?n5q=7pi%5zNEOPn?yY}(
zKtOA7FWz$b(EGI%%99V+d*uWr)=-1*-@mhI=TENj!Fy1q)7Fz;!q=t8fBi4NyWF>m
z;`s5;{CuL-_YW1%n0=2NJ-U1Q_LrJD7NSqmGxf_~-Cnzy?fCI40u~)#@@)Ed(a|lQ
z{^V(sVNe<S)@g#xs3xo;OfYj{VPV(Kol!r3-k`p9Qryd}GiiH&fXk+@sv=Y%)w;hX
z`PMB8HTUK@)fBC|H>S=0&!7J}-OZx&HrKkZD*p9r7CK@3OAce5Zdbbz|13xG+1{OF
zlZz19v}qF$2MrBP^S#Y`85f^54t5q6<l7Fs+RtyI)+8e(_2p}_Cci<&E>5+y(|aD4
zesAgQ>`c%u$i6lkBNJr%=hro9zsShQeEX3EF7134Mn;KXp2e=eZayr1#i^NnW3(eL
zM9eLJ$BrEbeMY*wqcq-_&<5(}S)X4|L!Z@t;rEYH<y@<4CFADp+3HMtDgxN!TxZ9b
zM5kwF1k8rU#!{a=`MYEsPe0OG*lyuUb2ddQuko#u4bCgi0Gs@)yI3T{9BOr_z#Gdh
zvBlqI3?ZT}_Vx2-4~vV7vC}4M?tN9X;UZ4;1J7mrSzeyKw~00od5po^*ViE-wLii|
zcBDN=b)@ah?t=$87$O^&+Yivv#!n43oAy=&ga|vNT-VbJA8Zd8^J3&Tu9sM|ew)(b
z+*I$-P^hg(ZG@<EU$9Zw{+$%csjI6fH2cI{bI<SC+kDh*E4@g|=*-oz&O-b4;$I&&
z+qcgQw>BIpuTQ&nuCg`LD12^oak||y5PRm8Zd4gqZ6S_q^F}H|WTJ{U@YgS^xKB$%
zsl^-tY|7TB>}$dVN2W7x)E)71bNyYq*M8n{yqr0CkXs|`bA11|M_9WU@BUV!NV^s;
z1k1qO?B<OdrGxr2EvP~=I5K8F-$zEw1#S8s$)w&Z9I6#x<{KXRSr;WcC7-GzS{Zve
zRLebL`}XaDI)%=H{GORzz8vZqYJJtAMryTnwG=KcULAu9|C_V(>nQ{Oh@xy9Pwy;r
zF|7{aw>M&EQZ06OOV%mWKv)>l7hEgxY|Ao!b}^7+KxuK)ZnhgGo?GpW7QTINcyVS8
zmRjy|2%ogF{aB}{+eEd1r+K1sqM^f{>(-^-yEBDDjlU+T+qb*8xR^G+mY*BR=^Cz~
z@F~2$a`<w%kaCyDav_!0wPyVH@4EUJnWj7&qY@z<9UWtOl#PWu<PaVsKO`@#BC|SN
zK1y=y*8O}em-T`;HC4oy7cwg!+&U;|eFMAArjhl!T~9Mi?fQesUW6Hg$AUaITt@k%
zc^hBBywucSQ)T=KA={r<)dHVCe~leG5a{CMWQ?7d`*w$>K8$bh%gZ}>TD>>^U+5y;
zri1qjScJujAr($YS*@a6RlqCQ6pVa(f2Uz`v#XU=8okT#iJZ2+&ri=a#K}3hnC;Lq
z-iaE6t;XZXAbov}^jHec3JMiPPTMbRZ<a9MoTA-;)Uc)9Zlq0FJ=4&xrMU2~=G5Y)
zg!}h5l#D+;dLcnEPS$>2?$V{km*>2)3@Z18i@CK~P;tmu`S=>$-`W4Qe!XyjgyA}@
z&43qKA>aFWd(-x8b0LS(u#)_L{`toQ`<4IL?fZwdh@;E%(<3$}JAZj@J&>TBsABQ>
z{OwA9)l{9Ps$kxx*vkiYb809FTK8NTsE<*OxwyM2MO(90Y(8nz9*);|Qj^b54(Geg
zXJ6ArY-Hy3e9_@yK|q>x6%|tV$po;)ARzVywrfW0+qEm2M(oeKw2tfPHx~Z<u80&b
zCTBl{-*m8?Ia2A3Nt5ljcdJV53teXHg&$~To3$$Ovg5oZB7}xMKRudn)9<6Xhytt^
z_G$y=R{03EK()|$s=g*%m}!xhJsS04ZnR)*xF#%KJu^0tQ!_9&o#x=F7h-O6HzxY3
zrDrmXYEurM^E{=+^)*#joKri$vG>FnwV+=mY5Jgu;}xfgzRS*9Z%mX+nz@c0lN7ce
zK0ZA3<>k2n`bgmvox%k2IwV$u{_b?Z%Asb+7C4N3wJcsVR2!S^k8ECAoKGGlPlcU5
z*)gAhHM9@f*A`l6*Zpbh;=)9&XK$3`<BWO1p@ul$SlRH9!6nopr8rqSWI9bbvDxVA
z>=G}U=DDe%t@HMuPzNTFwb(^OL<S;)f`Ss1;$PvP;^Hk)p&l8{<gWi^S<33Iwd~0r
zB`2Sy#u11=*0k>Vd@XbF)~Yqb+1hVR0=~ES$Uex#zdd7?(?G=-nV!hO|H8(_n)LtN
zRXWey*jk_^dj3SOUC-z3CD|8z4&I#2{PsX`SA#XGN7eBfgUY~GvxXu!mx9_<v(^mH
zZn{%e@6xC}<<GXZ?`FC3ZuI$3bE?B1EnTPH-34xctzDZpwU~ytnCWz>?H*&=+kB_!
z)cNhV;?g9xQz$#}yO9>1Z;dm}zJIu9-@b0$x>)03hb}kO=2Tq)j>zNJZu`)Sn&kMR
z66K0mT)+OR()=o?W#?N|a{maYK3WoehYufC2`Z%3Eq%B*$H}AIkI8<foTtsK2v2uX
zAhp=CuPWHnz~|)6$dl$FRBLICBG)|XIHFr<xFoR4ef}~zx|-&zD7S{xNFYgvpSpQ=
z=xZ`BN3qrS+ZJ6#=DD=4J_iLm=y5RG?SmdmFEhS!|MPwQ)&nQ2(qs{*SxC!4pR&`|
z2UVipBp}oh&fHlCoREdOJ7AAe*PG!ae)#Yi#n?-AujQl60*e>ME1mnp9d;c!U~1RZ
zIsWmS*ES~3!Z$G%)O>rrsqt|IDXIHZA-`RF_cDu%i-#KEGlrbH`Km-rVLz|oqe^om
znfZT{%mD!lTebdAj}LfmOhL}h6`d|}o5$bvr*)=i7YqmP<xu<5n=Wl7m}=6LuwUH$
z?e|W#(X^n0p;!$pLM%>o99w0$a8BDB(>TPuI@Mz!CTPoirhto#9}cU2j!3WC=<&A5
z$_Sv9R%#j0aI|<J%W5PzpTK45i~O3$3bo!k*2OyqacO60TJRm%VB8GYW!{#>s{KsU
zucupYVWy5<HRb30Sg`<yT;8R0lwh(t)%hDYZX^Yks^-VA<%xU;t={s|B1g<x%C}Vn
zoZOP89VraZk_<b=QXb!`>DT>NsBq@z;qd1Uk#*LY9uHCYO{Rxi>7D4#o;$a-a<DNz
zN+ISVlW4Ypiz#yNww>|D!S{Z0IQXRnaqHIHlER{i67c>mnqZNy<*(Hlk=5mboS!~@
z%H=yUG&GbZ3$UU+IwQ~HJb5F|PqV)!Tv0+oBH?oHojZ5J-F+3?3SDxLiLN`|dS{wi
z82kRtdYOyu;}X*LG=fsx!900OGvBaE2B#S*ly~$ywD6w><woD5V<M9O3W^fVURYeD
z^Et3<S5I0^6r0QRut|ON1F8_RQNyqAC3ZhPZVc_!%r=w9!&7|@BYhu#-;}H=P`JD_
z&!(CZcF~{ZT4qgCqAFd4g=2AqXw9%<|NYF&-3VQyl~#nyWo{r&`K{wkI&t?7i$ZK~
z8>hJAud4|z0J`bz$2^u6>?<GcW{<_nwIxrmY~HjfL)c}gS=U1ug=S=W`@Vh8WB0kw
z4KRm`dvx|4UX6HAMxX#3GnGu^F>Tn=Nm71eL2XN0Xj0W}idT#a+q@s0+kO|6JW5W7
zuAyPHTKaXwS^!lDOL5tbTTD#n#EBE0=4?u@UnHq#?hmBF!5MvClNO{^j<XUnUn}La
z<5p!LXQt*-&oj9M{JF6=0-hX~6C7hri9u~SmX>N_re8VYKB*VjCsVC2M2cS+4&S#C
zO#v4p(8w)K=c&P%2vL_dj!Hmb_l4gV3~It!(-s5RRVVu+J=!?t5qw+QXXob+97|uf
zb$>Jt8twX#Xh7+bjMZy5y}t7DtTFx3s8d2}@4hA_^hiG5AHa3<Qn*mluP6WDwFd^`
zBD2-A9maH!(3|5=WLr=Z5Yqbsi|0oRDuTEp72TLrqouZAju25VUYriCB>BW>CCBBK
zR#wU&&m;p%ya2!+oE&HvpP5lb;+SkRO=SiCwr`h<6x06w`}fJprG*K(09M6MX(KLU
zg>mD(m8MGzv)+#$H9nIQi$`folnoa$c9xxK6<*7Y_0h;S3y!@M{Hl0)Nw~=@HCZ!9
zNJO+g5Y<2Ljo>Ug3ESb86NzQhwnJYp>$=YxB<$S8YfuplTof-8#Kl#1wX(M|=#AC)
zvxUe=TyJa)ZD!|jIQS6$j~T=}4W01rZ?2WBvm0)apPw1aX%7O>M~RLN(sftoe*5b-
zS`ss^muQk`m`S)Fy1f6S*-xv~WPdDh3~)+YnkzunwwVVeb+7JeWZn2)pIg&b;E+l!
z48MKn4!_-?Qm6rG>m(%ZR_Y092;Jxb=KUIKSrk0!h+>aVt^~Yq_Lcj}i(5#)apljL
zO_3xB#Ha}5OdO>a#Lf3DynOkRiz)ZVj~{#<ORhnmo{KnTbh=Jl7M*HTJXwjzN)&aO
zzCq2mBY2Xm$oUHw*bf~#?KIv)`pL--o7(t`0c<pj<BAl@)kg#*S*dGjg;5_@(b0-H
zrcr|>AMWM;ieT!{&^acZo}O+WbU|nwPipe<{&tI;nP``pQR7dKXzgdDDiI8xGz=DM
z5npF!iU2~4fwY`Q^ZLUQid|=|y_k>2+Rlt>7osDW>-H7jGE+4^J+1QW*yV!}T!`nY
zhXh^Us@%h^Gd%q$3Oh)^nf=hmPoGS_yf|YqRqQ-vQvLiyNpkj#d27ab)Fc^2hKNbT
zCkZTzH{mDrJnct2f-TbTdU^_JHwW}SD5n#$mEzRMN-JQ%Cv(jDH&+0xMBrehA|&%0
zEs#e^*T=EvVF`d$P19ao*YDQL$r^8XI@&9DQwtn&Ttd1_ICyvzfa@A&C;GKb1oS>_
zDO{Qx8rWa?>ik1nU?8t4E}gfpk)R4hYtTC=zdXA(Sv~X9h?s)))YKXZCE^^Ka!M}K
z$&)I0q$}88x=}&9r3ELupI^>kdu$vlbgfB$m8w%{J5bM5FeMvy>S?1##pN#Q8>3o!
z(U}gLwsgG<`I9p*$Y*;0{{7t2{8+&A=Qnajk%}5)WcE&>$%>gNqs3_o;W6+oamJ@n
zK700T$ftbA->*VWnkS&pnby5J&&4dHcm4V#^1@K@@?z@k+qWBYZ+7#Q;yLLLPJ`Jw
zbSF&IMFsVhk3-yjA(uLh9GThP8LWN+SWwDNqtk8RO_(WmJw(tZmK2|3mqT8ozu&ZF
zOC!*r?ewrZKD{J3B@1TRqQz-Vw5W3U2Nj@HC3L)N|N84E4!DrpT$W0b+Do*`YHtL|
z8#Rw2tYse;qMZc*zg#Oa@dzJa7XWido(+lHV4l_`;g`ufcSS`-jeL5xr(EvJr2)JY
z9^zT2$Fdv0Wv3SS3i%?}S^ac1vXbHWB*<5dI6}#7BdwWb$6I@9=1<}Hx&3;--hO5Z
zXF9NQaewPur_7U<og83Cl#?}-v6ZS=ssLW2n&|2o%dVni^x@%xHj-t&49#dhval&}
ze%EJ4shRWc^kS~3s1ev;E1#g!VE)yVCpD6f_Pz8KoBIrk?f4T;&67BH2LfwD1!}*u
zdY!gLK~Y47H4hwVNk7+=pzQK)`>M|y&}Y`6l@Xy^cgy22!5Nf8B><h(X;swfUeuUT
z0u4az1PEB_pdbHesDA0vqau{Z^#_3{9UGqzQtq_peo^Nv9-|twOta3nPWd;xKdo>!
zA|@^R^dM{oj(}{zs|GO6&dm)zV2-%rwT%u9=h4uqojZ5(-~91$J?#;^n#9UkIi3Gn
zx438zhuSyW#b92e3|f8)MQZN~5BP4)Uhp`0US~oR{<^<H7~NVw2xM+T&3}`kTRb{&
z1u)>~Z6<Ggk^qt@x<5Vk?De8yaKBaeO+Mphw<PMl%mrWiDO(mWSKA45MTy=4q6f@}
zoY$&ndB__HA8(kMC4n}2<FV|HV|fwBclwGgeqUaa{60Pr2gjv<`Af9Io74ZFQd9qH
z;_5&D`hQ35{g13rby)-Of~ctT<Q4QmqI6*7-xb&UiZ8uJV?YRKtNVZLz2&9hOE0Qa
z=xm3eK&++O{`z(XMfcJ_B(%uaaeh*R|M&gwc?(*ep5ER7Thd-fY}-il1&vX1=kk2>
z-Hkg%==R^*1#qMEO5%Z1SM#@TSDB6-a~2Yz92aFEvV-AnI=W=Lp|8dOECix(=@uoS
znpf7=%7G(c0{}5;PDw^xO9Z?^(Vn*bOrbxF+}D8iT7l}Z@Vl%jPL6>_;xGLzpguuU
zn9h#(qQpj`dwhjX7fq$+#Y>lz&_MVfy|9f44nSYc6Sd;0`F2AU&CSgXFEvrO>e8;2
zGy?Z0&Ldkip{gSdbJLt(NAX=}XtnCrL-Z(7hz+%Sn|3Pzo%6ZP*%AT?G)o-{^=?o~
z-DSR}4Y8N!CL83GGYu~TP!a$$ZTk}sj9lKR>#-OxxYTZ0oD4wTX5OBnEgZ^QJgSQ~
za$xhsUmuv32j<U-*h?=9e2_gKe6!zqhU!4)Q{edfI(a$#W;1i7>nr5YftGYVaK9m-
zNp~|aq<;P^wYWIbr2(LR?9%fWK+^Fy>RzE1U(Bz5M7r0Kew(4MoL-BlOXuZ8r<4-r
z8QlN8R2*4F`Gl7C*5(xmjI~)_9P_}IjMx_FmHEVjnIhfNt5>ghj-PF|1A92&QL}k3
z*YJQ9eJ<c{``}*xfrdC2A;xzvb8}C*VK?GOL91R)Eu7W}neWOrKOpVo=wEuh?%h9Y
znzGF`?|FF@bS!<iw|Nls3jspNsIjbpSV7a4v}?wL-C~xOmQ!P0;!TUFoGg5NeA=HV
zw>;Q!#J)wNF2@7mf<e=&7iC~zup4ZYez=FT{&#no-Tbuqc}Yo-gXiY@LM;iOlPLBz
zMLQ9=Lvv~sB_#w14J6{TL3hMXz|jjnwzS<?ux1cl^)cM<l(9K)ivIj~$TyRT#V!PK
zVv$n}vI}pP^ykkXzA|fF<bhp(|NUfe35Z2&uA7$fj%Z|7AJF<Yf}*1qxd{*GfH=-_
znX#yj6xYS#DFf*CfaVMnv`M%W%ri(W$T?}z@%sG3J*>J#t}1SC%vu$0-n<#%$VR!u
zb>c*w)y`RZQRnFPQ;<CH#FF__>rb(tJb998cu88}#*NoCk>X=2?ch1``0rFjfW#*3
zd($FzzsQF2WF5M)cmD`GjK0R+smL58CkKo=I}6L_YxCLW?IvK*Rism}*hAqCo#js)
zSBbQt7mP*d#pUb9E5z7#%?w8Qiq(e;JJi3C`&I<Pr{yifo;|U^WBal}Zr7p0+D#2A
z6D0%)TQN~3dCL&<v12CSCArFGD!Geq01k2G*hpfP%{wFcjCU}zt-cjXLTnA*ruln`
z*E4#sBUjKLH0b$?_SrJv;5Xo4v2b&zM!V>`{=PRo))kqOzS?n;c4Q~eUx6*d(8BD*
zK*psleJJ~fAC~^8HyF!Qd?m555F2)~5^(ie3;GpnjY8))z-<1^a`XnZ5!wJZ4L?dB
zDw=P8(t!>p+#|z}AZ0YH8q|sL@ryanXib|EV$g_2=X;FzNGIzSi>*Z40Yux?GAz>+
zx@~pp{9s~gM_YF+LT|3tV@c1>vLpAVvG>?Df(o#{a@cc)2oa~r7Cqm1i3d9dzb1#e
zFHWh-dG%v&Z)&`COxFP~w|4DXRxkqNQ&Y-FO-e~>C*GL1tDuR>x9+_d@<|Z9FhOou
za~8lP^$dOar$^2==G&Pfmb#MFSIf)GhwvKt<U5R|0ClpHmztPRyn6KoK=kC?P^xKb
zhCzlJhp{&m+%8fRA<8TKS);(nu|k_lxbpJ0=N8&mch=Ly#-qPiP1G%R-?VL;%0H{u
z`U~0^_Ev|go;h=7ustUg88ayM4=;#e)hskVp$6!1Yo44a^Px*LZB8+e>SDz)HvRGO
zfk{K`<BLID$?x8so}Ha#1#js*7$2vUW1&sh6Ak=#h`3&MN1=;qrcte|&^>OVnt^f#
zQ4u32zSuaLbi@U%b&T}0W1zou6z~ZG#68@}Yay=^nJGlTGUdTeW(1Yuz``Sue&`?e
zJGJl`RAMh>d+binVRc0$1wdm_njt+lWQ2X$HU?2ns0E}=oud`8`gVHlx^+TSUr}fl
zo|!d$HQ}LLdMNciYbhdE|NQ#CM|xW-`T$418gTvUrHUE)<$P0n>FF(+B2!aSd$HJ_
zQ#L)HpDH=LyR}NzcjD6lOU5^q;BphjaZrS&=HlMgM2MQUl=(5qaG;U|;vt2H<B+5S
z(e4Yn((h!J%`(|bOFPgwQEc7otz3Z58<WrgIV~vC16d_(e^<oi(39$*n!YzW78Ml*
zBBf}g2sT}L1N6G$1<)xpX^Bjpc@NqClMkKg?}>@bb`J!2xHAK~$6lnpKviA0CO>Rs
z+;!-xeg=@hw#}1@tXNnE(K4$sLAlamSxihU(4o_*%<svQZylSLml5-wI}RQkZnSx=
znUm6sqQW&5Sr;o8X~FvygkfHz9FDp6+%P!&FNWfX0A4f2^y}y2y20W3#vqUF;pXO!
z(_;g%Xn<~3c4`@N80~cC)%U9(mX6mNMM*t9VqaDjz#e3P7FJd~6<eo)oNV-adfI+a
zaQ4q1H66Dp<x2Fn;q=|t-r(;_`hyVzGUzX^wWCJ8Pa7#1D>7~2>h<9D(?w#l6Po5O
z`tzZ$U-0C~uSBCNIq@ngO56D{q0mAXd#ZvPdLzwg<G})OExu(@jC~Y$Zrv~L(k)z+
zG(DUP6<@`gch<_Ftd$GwN21HHV{N%N^+M)cpA*6c{LW3aF)%Ii$d(Nv=gQ3xQ_(1>
zq~gjp`TF`cgLX%=A`$e7R>=0X=Qg^*Ux)61r+*C%&inCWyO#ck+(oa^vIr@w`c02g
zNewbZ+a=pn%+1DjkqA^RTwvPtK0T^K#vGoM_ULk*cnl8bH5ijq${uJ{66B*i?MK=S
zayqfCOe8fXLPM6F8uLLBY<;VnV?nh&Q1Ue?1c_}3U}fN>q&4Ua1@K#15wi}E5emmE
z*hx7Ik&DnwyZR5f{Ae7)wq(R%{2oq?8^A@|<`=q_m*R{f-BeIUt><_{o;-Ph=T5Nh
zt!T?O4^=d$SD7|snHlTSMLIgck&1q-T3w@F+~rH_oFIhm*I@q-OwS_7+8<0WA3SAs
zdaP(ZrKu43#snM|@{>&hjS=|DGLxa(^A;Vs-y4m#P_7c)z9rqb;aFlBNuM^6L`?A2
zzDyBGGOP}XK`hAl?dQAkg1{V{KJNz)>gu8{@Y@fop;oZRFM$BYYck8-n|}u|nSdrv
zrBWY+O%%ws^G3B1O-S8#pnJ|sNZf$FxgTWGcz^BK_$l#)Uw6co7i^@LIeOKhND#0I
z*g|@1<lzdOW`)JY`5Duy=O<)mbV2F5{rT>tk!Kx)^PSk9W62IlidinA5#^lFS4>Th
zeSB2VOxr=ViklznDlXo}AeMwe9}n57E#EE?DoZ@r&XJBhMMTrHoE*Ht+Rb}@Hpq$F
zwT!{w(3E4@6)kZGc?MuK7Fl@+3aj_ShXj4TLGDeAaYyc!$7YeT8y9&_vMue}o$7fU
zeVLpSC@TR<Gk}0o*hc~Cr(Db7mh{1{Vvi8n0PS3>mw?PMfBrb;w0YgTHxQtzBY*bn
zZN;hXA3r7|mlw)LcXWX()FAy24x8o){KXmUhl3F9e409J83YoB8bB2#HGw)!Xz612
zMYBLU=ycGs-<~vUc>#dUE+nJ@ZsrO=vBILjd=ccd1b7Lyb^APitW3CkPNRd+Xh9qf
zr4=TQ_f;pxMY^{cs;jA~<@>Iu_=?EM9bWTmOy85qn`jx{rM0!SI}RKO?Q{2-?cJN`
z4oUi&nm|+_%duv_^PE6IT27mVUnMkK+aVIDdGBU7R9^O2o{zk?_(<!OV;2~yq(S%j
zksR5_-xEkyeavg!Bjsfdo|1!rJOk7%Pdl?hn$FJ7Km_~#?ABtB;@<Q!dYOdfKi$5X
zKf1d|KGlQps)Q1I@KYFgF4~i3&y_`>><J`AT0i))IM#-@-O_&d-aYjU!)g=dWhkSb
zQza!OoJ&hf(o>svv0QD<vx%ejpFo$8*lUe=6i5;_u4hdAgGaC{a4Q)4_+Y1}ID5&5
z4_bqjojXcVYBh-3Fz60qnFIV{<{&et%r1n{NWGuL1!(<?TgozWB<iS@Kj;76v!`#8
z{1hw^Cp0^UfkL-^BFhUCp0&oHD`ls)L*a6QVWDARKQM6nj@`R^8=tRPw?(R{2)n3a
zS8wy+Rq6HgEGP<wx`a@!bTilm*5~+U((3$!{!3GoOX>BMd=5EB6!au|@r#v{V7=pk
z(^}Hrg0|Plw~cpob@dyiMw(B)#?jENFLrZt<A?6yW$x<chPc)(ocgNC^>_+g;Q{(H
z+|s1$LKpCi9tW%LyY+LNgV)9lvD?a~$H(ijfXOH2A^_554z!^pYM@&1a-^!JMl|&`
z0YupgWG-QI?D}iyq?g~u5>_p<R?Jku6L9$kKC`rMe|2aa<awjWvz~vz5pSskw;W*6
z?dasx8rBr_{P}Z}fxrUrJLC~dm>yu~zoO&YMNgm7(xRrFZz~OKQnyo1<XTHV^1I#S
zPx&0Z1k^+l<&abc2AdLz82Ft>a3C_~E2}!ko<cZDJY{y+^j5@d+<ELL{)!x2>?9u=
zl-*ofEryFk^Z#&n<EZFi8bQZjFTS7Ifa<Oy?lO!5jvTNU&8iqnR1mJRrvR;w@wq`J
zZQ8KGO3eccuxaboH~B#*hSgz$4ZZ2w^0Aj_LuY`jLtl2UeisFHg0e5yeKEgD&p}o;
ziwTfP#h0ORK%eF|Z#l`~1AwoWon2uS0%^^3w8WXEBj1jFpbO7n?{QQ1At1A0z+I_4
z5x&Bpvs>we_y)S*MiFxTlcqQl;rL@?NsM$kVRqT#c`Z9dLg&Y|kRjj(ajWl!V+H-f
zMcLAEY%g=69E#{1<o9rsAo)kKEfZ*`L9<7s=<(n9a{BRp{)=`VNZkV&mfw)i?TEu*
zoH#$whAE*AKa(98ZhMo64*eCtc#D}0`plf2eQ>PQAuAF%DLUPvM;u(8&O@AT@Zflw
zn{mF^L}LE#Fo2WWFjl;r^3R&}h}8%`<WiDkV2Zi{?BW?WanI9JV{yT;1uSsQ@tZi9
zMkV?YoB)H)i&M=+hEYyZs~fP#c4UEPh$HZ5kcLf?*bNA#VzU6m!44>bSFmx`a@fZ-
zhBhoRHl~2`%TfX~;Fjo)tW!`B2kI>=+|er6$LPwrKzmTi(3g%IjP2;|pXu_bz5j%s
z4IzJNZU$v$i@;{NM6_NNU0q$*3X$^JO}{2d`#$z#9)*}4f&=u7{=i7tGX><&@VR?@
z040M!5%t|DCZP}RfiF=XNuC_9<QCrT#n@XD-fkjyB5Ki%$S>*~fzg%0clLDL9PJPc
zotfyZG_ukAlSMdy>aA&APW|D0GlkbnH<e@n<|hKdovb8+vqRu(;VyPG)kvAuef=o_
zTAm~1a}IW5%eD^r9u%y%Vt7q_Ts4D_yuEp<uQB%5M#{B4JOOyZy_J^dW6aWrk`j~%
zx27cZhS4pb{AD9V;$d=1z)2($Zpj?rO(U3%q*0Wa+JVOd4Ifwp!;xPnc-K*m6GnAQ
z^d$+>M{Cbro(=kg`WPou9H9a-C2mM1S)++<HXXvkqYVQu>)~oj><2S%<Tl$I#AW_&
zbjt;qvC&F@)@RuK(8*JU-M^-%v)ihc5E}Lg;p~x7#8)SpJtcTC!XHj1d02xQ7G=K5
zc>6Mlk1DiJ8rmwn;N5uH0nVNFOTWyzinPTSCly2HbWjPesbe2Tvfa?6ByHO-1uDpf
zt5^<J1U{{~8<xiT$zzWS`<sY8<7V`Nh;aK@Ne_S>UzpvBWsY67W}t|){lsZAkT(X_
z2-Q6&FKn_1#fr4fDl<?s$LhZyFupc6ZyFoyLiu%^aN6s}d&>2u{(w$fi7h|oRaI53
zt(pK}iMu#8_y%rP6o-j>Z`#H30=5@sQ6F#bdQHp17_?*jCXG^&uHz4o{zLm{-BLK>
zq?aBBWX2V0zjc&HG|9)whARQcqcdVd=#8TtwnGwRjStV<$fqiHfpsQ4V@ePziEVxH
z&K*0zVQ@1rL`GM(w+j!TsZ^}DTNbPe;b)I8j0@V?u5p`{ReO3JACE~ROX#8p#0}X<
zp&Jhi7psU}q`-X5>eb-?Hs;J_$3WJ#uY7Dc)^@Q6LeJN<NahH~4}CK|OTe|9CD}k%
zSqHjH4gpiZV#UwlV9?hEkV!01CE^@*Ag@qsn!)lBd^gfntP8l7BiMbMiwk|ET}JSq
zEi^PaZKiT_M$0+Ck?b;o9I|Dhd!_>u3tToaQx-G@bs#k@E3ocdOQ4CCmR5*I@Fl@#
zOi?p4j#sp(Q5>^6`fw%;(AAtXp8Q#NQA+B{jT<-Wb+}0<pJNe@L&H1}87zl1uY&fW
zvCzdKNStAGbi^RbsKyPjydK(CD$+WOm{@1;(qlu0n4XG$mR@)JY^>@t1<6nk;iH9y
zH}flh7FK_DbB=JIGbVao2%oWE(;{4HkQvXxJ!h3y`~3O7g5q4^iC{Ug`QuEgG^18o
z%>J3;6A;k+w(&|?0vuF7ac^kI>2f!H{u${ipZb^1MY!3V65NiD9pJn1<dZLhrCln(
z*c?*D)W@B2En0(4D4~xJ3M!x($j)B|1s3OA3pp+7#fw96dY2Qy$L$Yf(v{psv3f)c
z8sre{3o6BbK9-iMhu`f_@vnRxrmZlYmN^w<<8qKmiT*q~%DEp8xwD|slUmU?3`T9#
z7@gvNG)EE6gRH1(Z81AH@P2?2w|mEqS7?6o0y_M9CT*;GZF;JVK2sg3blgX2s~IbP
zer!igHj6krQ+I(OFW}4X$)V?dQA`n%m4_88KKJ*U-EL6?mpmMyKR8%qj-DmMsjIB)
z0<4a_X3fPOq{dVlwxSJJV9GTwG_ow53Zw?xOS*iQUasQ0q%#9H$kT2tbB69>YN7b5
zZ(X-ZBO+g5Z=P6&x@$aYgI7}V9fF|OC#HC@l4G|+UjAL)S>3tE?uTy%F-H`$6FUXW
zW-7$&3nFQ9s#!M?O@93JDk6?|x{Qf#MW=l3>5|^hFE3R>KdxW5j+pZZTq3Ojfy?M2
z9z1xU5}M51rPigM;_9PNvnL&Csi^S}B(XH_+NR2e%o~qnH4A|MJ&&DP$>1tg$eY8*
z_#eT}+#VkMqiPF>%h<ibR&Cnyt}gMC>2oeEez#fOu%Dilym+8w6c-_?s*;jFNKd`Y
zVJOwI{5HITy1Ke0j~C$Np1Q&{#dq<xlamu|#J)|Gw3VI=KB=RLW#FnWDSBWBGutNH
znwwQoZTOQSU5y{H^U6O8wAYW0y5K9MEx^T9V`9k|r5LTiwbgo>|JTO}hYJ9&#I`au
z(ymJMY*54h#gWwxO5}X`awSf7b>#*D%g(GeFSb*s)Ib>3p{s7nFpvr0(nh1zs}Kkd
zy1K~PXdoUb8^<BB4CkiY<IJP`_dlmG>C>DY-#YdT?r1BMSSgJy`}w4B%H`~qK@r#_
zin~;<HV^j<?f_p!3<o)LH8&*njH*MF0Z%NLRe_x5Z$&;ZRp6q%WPx4bs;;^;WXA8J
zC;Eu@`=jYc>`)upWoi%@-Xt*U&;ukcUIkg_H<-nw>Iz4@ipPU7L<e=z&@VFWXT`ea
zfpq_RWrz@74R#>pj02Jq6k@!qomu6c+pa8$95-OEDwx}%g>jk0d8DJ?QX;?#>22Ku
zqzN?nL<M_IZ9CK&P+N!Gl1$$|caSOaUND+#v<gkV&*}H<W>Y%OA&e%*(<dBqZm3L2
zN4Vgg9Xqb8orLkCz}y42xDxdc2oo6vtK1|L0h+7lVg9jj^Fn`t3^l_i8IVpG$&$ac
z-rru^^u|;*XfD&?R|U$?FumGSbp1%91x-0|``jGZRg(VPI=YH7K@?wmeJq~a{Pdgr
z8L(ctcJ)-$b0U2ty5JOKQ%{x$F$#ue3EKXAB2zqkERrc8Fvbn2D^ob}NS^<J#r@US
zW%3v+Tbr-Ka4y^QOrds%i&~(7{m|EHld(vDubm<+tVZhsI5oe<+e22cHGK<+bZUxT
zjbl%BxUj&$&C{oERX5BaIbP?F>P}28vSv^5I)m5R`t0HQZ3jR9`t1S>5{`A8j4slP
z0EcFV{sV-*VdgL>WBZw_&e*hugPIy|;r#G4AT1F8&=<JrU-6rMeGR$~_Ktr_S~ZCl
z^4(ot(nUQc4y|oR+Hi7Cr}gaBKw_O*`pyhDRiyh|{Izz(a7l2KySuw*<~`AJmcvGq
zUJR65Cx|@p?#bbEfPCfAWu&rz8Q0PnBsP6e^qnTtUx^#E=JS(ifB?;@n-CGb^oYtJ
zTgFswGV$~C=g)j2z1>xkA0Sa@YYeIb>NmZXFEOM-F-z?Eh{o$@+A>I{pJ<+~kF<G2
zcZjqH1Y6UxDjpkJmN07U6!VGhU}v4JS2#)}DM%C@`^P{G&4C%F`#KMPz$oDJNPW!1
z>Ru7(KmIx}v`3eLSC9#~%D*!zSPpfy_E$RT;Y=D|+YS(9NE*^>PH0B?UH@3oiMwl|
z2If0YC4fOqL_0JD8RulXBTVBsO9MM13qaOdoSNvvkBL2>@5Yxyi|7uBN`<6t3yuWO
ze+fsR4r`$Vc7@-(O$qDd{I0G5j!!qw%p)jDDKN%Slif#{*a+VY4%ZaR2?|Ia)}U>0
z66bU}SB}o>!a#3O?pHe4d4Ky+EgdeCti2JFwL3od*wHaHg4Qw1<$hjgcKW}!21-LL
z;SvXf`hnVByD!ekGKB?lX(teu8Q{0{4Dm&QVDd6otXcOH{@5n#^d2y)>uBhcz$%qq
zD=y#7uEH_U1sf7pv{=g|vIBKy5_O6;FjCw@2kw9jjffj#d;LHTY0z;l<8&K?ZS<U?
zp%aRR_>#}J87`7{CfTPYnKlt39$w>*1Anz)9w0fwE4fvg=4=Jy@C61p{Xew+(O{Pc
z(N8&;E+@}}VN5`{?hAzOAdiD8=RDlsAGd1Fx_w^>VUg^f?l$tKnGd#-IMBl!n>KG&
zgbbNnR8sB+0R!mSd1iD(scZE3T<J3DMAA+?a{-dA@9k*@Rb>n@!*iag6jmp2+=O$7
zfg=f+VW2li_hg5wmfzgufZkNu2DxA8)1l~Yk<)n#VeFV}*lToWZ14pWNdsj@8U4;1
z;|4;d(ez6GlpL<^gvuwEV9-}XgfD7gR?+;ZFmbqmC8@w<4G}KzCT)vrE=#&|Qw=5y
zw{F;Q1&uF*6(8lQJ5jSXLR+ba(J9HWjzDsFj|LILC=#y#4_Q%4QA`@)iQ&We06?`I
zGJp15=_wW#DaZmgCasdN{UpiW)c}YmW+9-qL7-_2p{T<9)>j*u3XaEw7~vyC3#h%~
ztTADtO>n20x1^nc-oh2$VYlp`wEd{$Yj{}*5lEUs@CD>BK+aOVa^*=A&nJjtLuj?%
zv|El#c$Qw5M@;iw|G0rp)L989#v7;{oERT?2hsWAOz->muOOnseKa%Nw<1-=!pDEf
zV|g)0=D@tLHT+zJb95%McC_?PH;_1Oa0e0H3PfdTS=kVJ4}vD(>U)XLee6v&tZy(m
zKCwv8eFdopijJ)F0z~G4V55jS#G)+>uGn1iu?ZtdA`j#hVTKdn(AQ+XMKnl6waM3{
zVC-}hur*=y2$BVqj7OLeUR#Bi1?1^DROz-+o99p_>$~whQ=?Q(Ea_K)F4rhH$bO>p
zB7j+6KN_9jiop>SkTd%ACbVm?W%a#0Rh|N%ud@E{4&vYeqh|Ey4?1ylr<d|4o3e}*
z-KJZO2I^Je#D<Rjgv%{P8N})C{rd+8j`L_?&xi*+)WFBD#s_{i>PV|@b{3u=+b)~4
zRR6GUI0s7#QGaIDg*6mk%4Uk=JD7Yq{SNA&=>E2St&(loWx<L%ubAGRkI@mcyqn#g
zm3nitvw}^<U=9qx7Y7P-IvyR+fueWNM?P7DKW<LkW!g+36n(o!0H=D!$9lA#ak>w|
zu&5)|3^$s7fl3!QpNzuWi|tXFfhSErXBi(<u*rewJKoL6HTk(y!Q~<Ri3TJqEVel_
zguyKqA8ij$-SMOxyCD^1M)feCJSZlCAhKjIGj*`Y9($0d++GZ+T_NArIIfJi&6o}!
zejnd23~O#s#D$8LG&~=U{z18r<j#1<jwVP;p}*bWOO~CfgX&(=pJ&p<<~0Htclg&7
zyuPm#SB#2_yft4*+bRQ`NIfAoKg?S)y?GBuTKmb_{!_D1)eH+p3~R!S;TR5Wua)tO
zLB$LN9B8HMze6K-+-w;nPn&t@9)bnF=c2p6WU~x+d7!F}6Vv6?NJbgtAx_{TX)mJ%
znyA`4j54;iSq+OF1rD|2$XxBiMi4^=_E<XKXif>2Np%Y5F%Y5Sx#hu1XvodP)(@yj
z+@2thLmNh*kG(`$GDe1|czV>UQZGd<{loVGIY7W>lvU6Uyg_q8ORbZI@IG+m$P5rv
zIG8$07|stLJ`6`33huzrMI%13*Ay9?LrTwm8-Ei|IfbE>{KmdBuQAm4UwV%dpAQ*R
zLmnX_Ye_y*a3g|icvMkO4wx{@ebL#g(gq|bpY6{p)xo@FR)T65;iqgsvWKZrFRr^B
zYy_aoD@+B7@f}bilW2X5>(gKUaf9@eUU0W+H<6gguM;V7jsxQrME?NNk8P6~yY>aW
zC0v6SAY|DM7%5<)q6vr;4u5H2v@1wG?I%vnQ>~f+e5PQrZR3EbT~7v|L_BkHa!9US
zgcZdi{E<2t<bu~3tm&mULElQgEp6%ak8mBA<~M8kvS$0LTeC3{-P<T>>^LL)#awfV
zRU6N2f_943I$~wi#^fMS4_uZ?(2JGOGGRF`k&N5Gy&omD5h5+0)3|=JR-OX%m?m&-
zyY}sS4P=%GCJEeCqKtG-52n-NiN1njL6Rvc=Rm82a9DG3R+8~DFbk6anl~Wk%%L^4
zgPW5I^%$T~2`pkG0j@i?`1v!*?BUePy@H;N3>*2mp8(gR0<~NTqdn0~Mt-(%gCG*3
zNos{w_=>EBIm!n}4+UFW+k8&NFbrZ4wv4PVMgr=PN^K#qXcxO{qx%QL@r(>AT`%<>
zfS%bz^t>lep4iG9iB?97NXTK*rehnvfOsL3MX;5=zWV+)LH|1Lv%R(tW-&F?46vMK
zKcYc?7bprt4R;YPN^q?6TXd)`&(?YjTPRpW<6Sf|4IdMw5Q>*E-ayD{JQ~J_STy5g
zV1Z37?FE}k(reg76VQy&T)sY3gh(WweW9B_?lvt#UoZuL+D4NqMMKO8_?+-$FW3%n
zqsaoBuj=TS8_2Dj0&Z^r>5w2YqD%qYvf*?(|9HTx1dgpKRaX~V>3XAxLDG9KcM3%H
z8vqA$DA8F+Ph>udw7u9`TM)Fp<<BkvmFAL(DCkTW_cGkJ{Nia?9joh%Kw|9%Lf?p7
z&P5W0qd_VEF`;k>Iw1TF3J~E7=mf1Lu-e`b(J*ARdiCo6HfZ5oDnZSJZ_IBn&Bsf6
zyhOzht+TEt)3H_G<ogY7_6qE!NHHIke3Vu|JL`t9CDR)WrzNbKuxv>_Cv<_s&*hbE
zq_=>UnkjVS5G>UAIFnZ;`i>(?3x9t)5I;(EO<=p2^+#?qGcq!=U?w2m?=q?pk=T%D
zyI+33Du;wchU}mKs8ee(wIdrTHZt9}KpTTJ$W{F}C*<XAbVC0B;1rqDj77quO1Y{C
zkF|2N)KjGyKyl+!nd^KbfP7nAi-~XCdg$;;9Dk*9PH6~}zs?19tfufDChpp+JUC%<
zV;;+)gYI}+i14HT-2b+bV01f$%Qs8@+Y6A2>_&{X>tFrD3w&%e;a?577BkQzO`)s|
z?JGJ1_Yv!n(oEUo199MyXe92guU5G|1YBU}A}*=_78ZBnsfZuNkeBz&R!Th)6DYSf
zI=Q&CcUY2=LTpqbe{ZLc#)gr(MU@%n$)DeUE2>@_hQ9%1DlBA(H69VV-IQ-4vSd8}
zcsu}MzHL1#Jsm1dmQe2-NE`bx{j~n*?W2&}gzSf($CXfSUD&honD6uv(2~|WDYx{=
z4A{z-BI!D<DxvMLKP>gLB{MG#hu_}?3^$t+tibxmorBT%*2e!_0`NaI`oA*fpJoIK
zkR%~&4~6nal>o}Pa}-J(eBHmUghG}h{|jFtJ^$yeTKVD1ru|=k*S+gJg}x4G7N#sJ
zOzEdR)uBxjAZEu6%rl4aBavO(Df(|koW>njE_={HhyG*W5MGj=A5Ov$;zzOLMkdU6
znuzr}f(UT5O(&y_<SX`HEB61dzFT5&nW3XU{{|xrJN=$WY~Xx+gf2#vY5QMigKq6w
zYw@%2v$E1;;)eU**2Xw)Kk#b9hSl|nhYl%+Om6#deuMI*vpaUI9tiGYSny*G9j>a$
zX587C+Amo9tg4E&K)AqVnzuh}6=mQ4RsVHY$A7-VRVx2k4a&X{O&tN8gLf>LMPh}x
z+GuB{udnZ84*C<$AZ6{@0^0$?v1BK>;n(EYRukj85J7T~Eg4~_id`5O7_c5`<JT_B
zzESrfmfQdmA#sT$_b=rQ4T?YxgOkp{;c)C6mQ96!8?wPWQoHUE(_5hr1XE=ZQPFT6
z8jORSIdcm`67M#_2pa&eYTVbS2hQ*zd*8lGA;o5QM@L6Az@jWWftd9P%It9g__Hg2
zZzsbfX$qJ}3RAHMcbc1@FO=*`{_yj8`CUw@2Vlg%%IN({3iYQ4&G$C$d7zpdnDG2%
zjepw2BT2?0)yhjt&tpcFPR9bj$9Vksm+!rJ-Cefi0v?Llw@3HB%})*-<>KPn#*N2$
z@j}o02;So{>f+J8Tv}FYPS7`lpt=etiQwnVYmct{UC-yzK41|!EAr7K|9!2V&#d=q
zo{yu3{fho~IgVc7+%%SSehAMIv1{d*hQHl^s!G79Qi4+X(kDEE$jc4nIa^-B+4ZBZ
z&(MJuKTM-rIn~18`DM_0nAeQ;^&JDgbI`g&_Tu!vuccqRAqn;{Y0#r<G}qv5dk(Mc
zLLBT6_v7N?E;|KB;AyUkuk5frz~l3~ckil&*OEQ79r$r~58hZredYa6-Qql&jvpR@
z&gO8|l)MYa%DWtV0S;efed2TSQs4Id`?GidMnS^-pB+pWrM$M$g|(1XV_v`V*z3~`
zCBA~DHJSJ@i5EJ4ekB0Tz)Z#tA5p%KfiwB#?!U<fTm3B{ATUrD<yG1Z4=E=Iw2FHx
zk45|MV?Bv@stzjY$KYVd=43psBhSBo_5!H-V$8(=KHqCAuco&WE>?&3?%lg_>sDo|
zF78LzxMfRKDS5G7t`^SQ@ZtPiA&t~nV9!qf^EVuD`Lz%>Iw6K@nQg@r-6WHNA$w_f
zb6LD!sR<F|04U2JPdxDw<jZ*rEsq37k*@we6-$Pr@b&J?%47cLyZ;YYqhmZD8}hCt
z0ve|vz+`E+f`S5%qO0rMj~_ou!Nj(JrjevC3~yx|6mwMrJrlM$8G}8i&zvQp*t5yO
z!QrNZ!`|&r(VEIYaH^trfRe0^ZnO%6g0Q^Rc9<bXDO<c*$>^9al-x&eVf3-az$CE@
z`a^+94Mje@<?2JS*ykX)Cm6vp`a!?=%%1wzueln}cv?(y<^4>VF8V(~FI@rT9o~F-
z<zH3~MUM*-4-a@HS}|8g9RZ|J%Ix3d41Kd@k_p6B0((InZilmiRWmz@4{AOfs7z?!
z{d;=!wR>@Yz=hRo)`WLx;P3Wrbsz~kkL|<>`O(focBGR4bYJDT(SaFreNSN#tprQX
zH<4dZPzhuogZTl!kPs6)DrSJ6LUfXWIhA;@3B$w+5(1QdCE9c*G)=J8AA!v6hZJ58
zrq9gn`5s7y(y-+-@$o5vGh`*UndrDU8IXRs<3$2?sY5UnJe{1hO{|=rnhHR3FOz9l
z9Svqf0fqxsTmZ9+RY3wyEHaS_R~{p@dyLrif}q^7XU{QsmHo+5g4IB>PaH~CP7KU4
zX}So)^DruvKU_!BupTjrifW_zVaA|@_0KPdiQNt!A{o4FatadSJC5^SAO+cuoIUbX
zigCvS|D&St$L*O%((?bJtO9d%Kx3oQ!(O<a8WT=}{=+_Bl1CMsv~}BcUfv&%PYlZ!
z!PVI{+bsOK*)$G`BHT#}W6%MRB_En&jMHBMH{DRD*8lYv3s$gU>Y#~nYetN6ttHZ;
zKUf=e<ub6T0U+hHSz(=uh%CXANE~@OZn>XF|9(o*+ErXD|69tQ>B72P2??j6<Q$Gb
za<2)s7pjrYgZuZtH1wWe7Zg;d<u~z9$>&&`tEH_SKrW0>_5hoZa!3`lh(E46X<Imz
zS%2Ugxn3v3c0i$EP{_LHz5*OoBI3T%EAQG{8!hmzbU!NoA>CQn0f^T-qztZb8Av%s
z=}B<dRw5RN`?7C37m)&60XsG>BX7&ovdz8RI&WJB?~N}mEa*diLsCv~e&prlm84s&
z1MiBZ>Vf3%LqqJS<^k}ow<*hlGCK&89QUt`z6iva9rJb)+)u__K}zoNHO9>}eM{)o
zD=_kW_}(VL*n)`~;jzy|{jbr@Hd7D1a|YvI^<co77VoUv>W6-_rHwDq$<gsM;v@#a
zw2d_t>O}@vIASH2a>B_eKeuw`gwe6)I2x4#nl!mVY9I?z{p~9#9x^fh16CF=Y=IyQ
zE3pK<FcT*@70vb?aDryVAx>f5hrJmr0)=_rt_#|14E{!;$kplhuU~)TwyxhZGePKb
zF9Md3k75mhPg+K%ff~5H&@WyO0=sDu3@9k?NGT+RkGO>ZWXCa(c27Z%zU~7}66*@b
zQhqxQ?$xukB9dRf{Trdtfd>oMsY(i2eY<5C3>%D6wJ@?n^;_|!KT%k2larPpxBQ_F
zCiH1%-FRwJs9|bl@J6p}TMQ_ni`bLhn4F3MPjOLA?YUvZ-MfF4R##V>6_Pt$LYZ;Y
zuuNa4T=%l8T%tW?^&*4(B?2j-j0~1JP5yjpA`Am_DpJj7vrgE&j8j?zLP8QC&~et*
z8vh%ov&`wAMMwloMQcgdY_dvLOFx@g1(qtLS~cs2Jb39d{9Q;B9G!WVustTsDk>=W
zgC-BIj$?ku53xGaiCsA%`eRv{X&+8!O6yxBIYZxKbaf@VW5f)V4Q2Fw$ZTO@VF0Xd
zO$$IE)e|-$<SiK$mZUj&01&ru!bNZ60lu%w0#1fEj?T^ree+-{Ib}U?V^Q#+7=VQ_
z7-d6ueo%ztAQYcd9Tw;8$egv|5^*07+rKdmfHEJ1toH!d`q*u4`T}mX$qjTAsG}X#
z3pDg1Pe7N&kJ4=1n4X8LU%r+9o3~|<w@>zB(C=~I5>W&=x`9AL&K#GqBbI+3Mi|mZ
z#WBlcxP%H5N?pR~kM9e{-3Sur&zEClSs$id9*d4sux_hHLw*GEi%tE$9ErgKD5j_F
zmI0)h`1zZz3ob*oRhof0Q65aViE;rVCTXY%|JkzxKE3Ki8U``G0;+;l57A+BSnnh|
z&@8}@!avh4WMU512)fzkCRgxT_2|8gF~(Ou9UaXm2z%!OWZ4XO0B11;$v>0|&fD-?
zI*jwLSiB8sog0(X6`^yPI0%wY{eA;Ex7^CFubi=2hWXRnz5qH2Z-|})9V;tq$k8AP
z`YyLC4>hceVT;>yN3YM((NPNKI#+0;*1vz8Z^ck&BY5JeKOK%1Hkp_Mzhbol<Yr~D
zgFSiu5e*`e=68MD_{1(`@a0#9L0%&#O5_+5Q)%opu3)GDpiB@-xSR?poK*e6{d#%7
zJG@(_7gi&$9QZe{0N&m&C@APVffU4FAc|gw_A<}4c=J9+CMI=}W1^zez&=<=kes67
zUCqGoK|uwQavFzb_NBLlg*L;LkNZ_VX7{6{i?kACcgxiX<mVpW)*`op))81oY;0||
zbQ2RMSm=0-HwkZX2?@E4;vy<vQ}3VI*>cF*)@B}lk(yYrVw@D91J$Lm<s~^XeBo3D
ze62ioY9Gewf}3k=Q{SyO!Js-95qOBT<j6SgVL&#IMuAg+>SWxK_A(0tVs>@T+ewkG
zgqCORpUC1psBslm<);v-N^a7i?w@8R?anRND9m&5z?$)}tU5we)2<7JmwzJLxZ%^F
zJ*_qb9&)`FSmR?b-lmg57pVIjFx)Ft>sdTqMv^N>_$p0^JgDWwb2F?gfe83sK}_Qm
z;f^B-R33icLfp>8p^+tJw~XS;gu-OEj3kOoGk0bf2V*?dV^Ie8DqOScTwnowfN3)(
ze;KY^o&q;SWP%gC1&7hi<iv}T(?^_C7_c-`bVm!Lf0VQ^)&Zz0{xAb!9EOE(KS>z7
zhj^n8Kh>Uz*yw0@4bCz0^D9q1$4!QxaX*(5ZY*rMLWX~SL;q;f!PQN)U`N9s%B!Sk
zv0_NHE`VEraPTC{L+(x=oIGQb6?@y$%M0fd%E?8r&n86>8Hi|X*EIIy#}QoI@dQ<(
z0v;SD<lP;}{Mt@HZyDFOu3$9SYqqiWKNje`Qs{DB1!yY(btEz6L{ou-C2`&x;4oyE
zHgf@CdV}RXjSz!)djSWTji^y?F}PR{(J)lfwf7hfGZ`d$3m4!H%r_#Pl1s$M9WlL_
z&+r3%>-6i}KhZHU50E1oLDCZ_$X&dUVG<lC7kRQ^5%)_#vp?vBS>ftY6zW_sw3@gj
z?I@`KAPk>L0V=R~A3@LLhc?`}@%2+;hpC4Q77r=#Tm)$#^((sy33<I;e%JX<w;Ao~
zr|RmM@~ebn-vGlV3a_8ct{Q?9=?N_Rf*HZHw{g$oByLbZ1~X6<B$xaR7C{5L2*kjM
zdDCHcAlL!;cPAlwrJTeI|11R!NO_V<E%wkQGmLp@Zv`<;7;r-G)BRU0vL@>9r%GtV
zKcnI*&w!4Xf=1p8D^69Z3YPu#N^FeB0G1IDPEZT3kF!*y)peT+$O2RX?j8E}b+;De
zq(G?A4L@!olkb26mG?~?N5o>KJsuCh+E!q+ixr4G5Ld~y(1o=DID!bj06W7C8};(?
z@(fJxlK8Rx`Q?4ebdM^a?lV}+Nx|<sGOmtYtc1oEgW7=moc4rXs3fzn<f0K=Ig@PG
zlJ-892k`URV}7s&r&FXZUnW3PT3Y%k265o~dy29K^4$lsBjwm(-o26zlUX0|!(bu1
z1k4?s!<Yptp$YhK>m~Yj7Sce#Y2tnVGlGH%SQ*L<E4f!y3o{FtiJ&1VarA|pjw9Uc
z2H5rSD&ATt%z^d50Pqy{hKq36m$tVH^X>-h_Ve>&gqexpZP>c4p|7>V6mU?~xv{U`
z4M~oC`gn{FW<jaJwJmILqW+EeWL3|IhT~WXO)<|9a7_X!G}f4zB~uMZlO^wK!4Wvt
z=duVn^PqWt2DPl?25L3=I&|d7InOOmalX_<d`MSCzK-*)P!s?8?tlB$5V){s&)<I^
zLUyv&e~Q)cv;tI<hK+=(i`#!UZrJd7g~{1>4hAr|E6P%M4(|Wv3_7P5L+m&I{S%}!
zZj2CJP@19Hcl@+rZAAD(PS7;RJt03z<K$`^96$eGa)IYcV@_86f4lGc0-Lu|tSiy^
zkn$9~4C|J<_a4p=FV3?TaE(2~y_`tJ$J*C#*>?yhwmeE@)$x9GhdjuI2is9@Ibias
zy0d51XHL++7$L^V#`NG5;+OH&yxW14#W)KF=5n@^vvUQyq#jiM3cSjv-`~9$Dv(7l
zWs@YNlB%;P@z!gD)%OQCaVf<!?yAPdMiH~E*6lECumg(lxAqMTu%f8gbn<Pm@&o13
ziVHo$aGertG>mn1bpXKJV5$|+bg-h{iNIF{(VNZ~m8xjVq$(_CaQ0ij1*8>aCzTR6
zIP_dD*_8Y7qXfaDWKtE7ivbPx6J%!$V?9G*bi&L|G-O=`bO3KH*fyns1F1yPJueXr
z#>A;(e@`szY?Y`MOn{U6^SCc<R$}4!g{7A)-eI`Dl92AOP5HsRljpHq1QtaG{45{9
z33Se5$jB8#v3*}~>&Gk<FKe?%sA{h8c{zP~)uH@twN`7?66y7D^O_)y=3j?Wi8TR5
zE>e$>+cB_qg*a7fOINYM+h$nx{95`}NcT^HXG5s?y!-^EmGV4UHZH80evYfL{(|wG
z-UdDhcxt?p(5HcEu?-md{{1kTmOxNKGN3l->Xd?UNwgX6pSAOVg5HbSg&egSOtQtm
ziEym_`t|GjXdcF^1-e?=FX?XwJ;UHO`7{Y`D{@VRQ&M7LB8CVxaF43V@36lprf=c*
z61=TRPftJoem%pDzRid#03%a+UW~+%-naCmD|%znd}hw$uh_t`WU8%u#E!ZKk&kPl
zzcvV$Sv)L5)z!pp2#2x!dqIK~3>KkgZaH}Bip}4jP@AIBw}td#REKcnHd>4e!<n`E
zxIlz7>IH4VW2{1E)O_Xb+aHpPjh(%15`b1ivj&hpoT$;O^#DaLT3J~c!n1F$BT&vp
zE+Z|^+rD$>v*ycZxtzY=+iUHr1F$2N2d6~o+g5SK3})b&UM$R>N&@LjXvv`L#}Wy#
z7gIv*&&b7u>X_>>4Ty_yuIL-zyqhh$Dnhh<w5%BQGzKboh#k2`Ex*zQVQ13=5>k~I
zk}MK=aaXIc*fh8n;ndI@M>shz(?Ty-z(r-DHK9N#Ea*c&4|!et@Y8Mgg$-M`R=X}J
z(mfWD`fo2loNV<_pIS0JW5N9`nEeT8u0F#hKVOKteD$GIZMd+}?}-G!to*h*Ok_6k
z5Y#k=$Ib>PQi{88V#i8w8AGeGgR^r!0i+M7a3vvke#JaYA60z}Fav3i*k+@pPY|0%
zdico}_3rVv+Y_Kk#g83T=lObpGM)*Htc>A?Fg+LS8XZr*qWmseVf&;7+_P;Tw6^6s
z4q-h+r-Q16$xkC9xL|(R4yq#4RCfR?eiWYUv_!hr$F#i1%I@d-`|lzF%b}L`?%I^d
z1JHRfXlpWHqIos~>sJ7#e5BD_!Y3rTX)hNCp*qQ#B)JBIUNs`qxObA=x=;?}%m&FJ
zZ}IP{SE)w`O+%c1_wS3q6NRIEI+EOwHcS@2{ctZa*22ZX2vBDy<;0>UhRLsBT!ULZ
zi&>lDgk^G79WZr!3Qq>{-;s1Pj-S=)&y$Nw7+in-gQ<toYj(W?9p|^7{^umRu*~)g
zkM?~}gXN*+FbGs@HK$uT-mgA`hdsDQiQGa97}gp479yQL=qll8I$D5YE)>CvbT<Gt
zYmm}jU#A#>#+|2!jdyfd+t~%c`Yf}wG;gV%e*NAf3{~P&AA_;PAFM(JXa-J_e2|kQ
z!8ip5k0DrbvxC`$(;5nCDPY6uZK2<_i;>(vpe9N#QczQ8IR!DrP}QnZ2c1M9jrhWs
zmj?N{xrB$aHj@K(C1&`S_*@WwcBwJThCM+}>~GOQjv@rhxgm04wvRC1YM)mpH+7=|
zkFU@j6X1PuLd%i&^kKjuU1{}pI&eLRtztE55t;o&5iqwClflKx(bz@7>^QV`XmJnf
zbUnL{n<O&8O;s&ZJ391{)5lBq>fVQ^&7=s#R5^NXE;4d~mfOGr*E{k!<gdA!i3=%0
zsTRiCoTlV%<k+!OyR<M~B2OpEGXBSf8`tB5Gi1WSe@1Q}MvHnFjS;ciJ$U%=q4pH%
zI3mT|7zG7eQ>?27Fv*{r*Bh3WgS&u8E-x9X<bM8Y!;5#sZ&qb7hQ@}7l36Kuk<Y-`
zjJP)TW$y<ylvZZg6w-ydV7)hh-N$f-T=IgT(X;OY>{qIs#L^&QiS)peL+05eeGams
z?e|MdJDxd<flJHtmFK~}uwf%(m7S96a0^xD`y^N#E6W%>hJVdipK*XujTRI2<=MDx
zL6%&fK{X&95gZj<xWNSw|Hu(v;i~_Gz4r|2@^05fHMYc9f;A{;L`CTeg3?5zg3>$E
zRhm)-1Ox<3j13W_cd&rck=~0jN(brEQJQq6N<Y_4-Zisl&iQb@?LG5uX4cw!8A1Nf
z^DFm#wHtV&54MY7YTk<V>A(dc;?%S<wZl5X*?l56AByG+5Q;^}?|W}A0{*oo!ewbO
z=212PLtW>^8KVl}gYZ6TL+sB1-XS%5BV{|xLzFJilNJHO7K<xH&+h`r7s60<;A?+X
z7u@*c^=s!PCB1-264=!ZMX(q4JSG<SA#~zA@P>$+-#*(mS}rX*Q71=tI&yUmj*KjW
ztoG2QQ-(Mlfa^rI=sm&(4I-cn&|o+4jIqu)tL$O&{RiY70wR8c9Nc!Zu5E*f(?A^?
z+4!-5#93Hafccz*(k=q41-~UH{k0CuGyi1uDIm^os5OgkJpCK0h|j~rTttxImz2W0
zibQsgt=L)#M2{#lE7Txt;8mR<#+BGZP4%v=t&Oe6X!1+-e#EdpAS?w3_yGu|5f>1N
zkEDoNH5wMYV<6K3_jIRD5&uMN2p~wr8F0XKO+yI!A%z5{TJe&1`#DcscV!R|_Q0zG
zkN!b`3S)Ijau1RAN=Hl$l)Lcr|Iwya5%Q+2jH4)YiIX}=6k$;#jX{df(t6K;$dLI{
zlaUyCgx0ArZ?dQm+7cZdU4yI+rrKI4IS)`mmbSYeHo7Z%c4Q_5m;%l4$H77LEt6uE
z_h+2u$Fme$0SE*)Xd{U0SuWyxSTk6#BOLzNc%8;OP-0>haN4sH5>(Yt%Z!7G4a|D!
zU~iA$IR$+P5R(yzlm)Vt*f>c^S08w}2xY>etU=jB)Q|S0R{kw@+J0~^5|KikC6wIS
zP)!;THX~#39*7^9fo>QhmZ>~!KO&6aiH<~qE<F@N>W7T&)*6LLKw#$yj)lMyqN@~@
zlt5s}5y}Kv^%Ke)?>>5>84iyM_uD9Uvrq_P=@Y*Mbe8#~kaPv00DX`;R*=bOA)FPT
z?-8mRA9fW<2oeEhoTuWA#xO`OKqa4yR7gQEKQPiSA6*dM)j-&YY?v1nPX8dRw+!VD
z#ql8Ko%x}=yB`o&s4qP68;}^9uzEv(#$p~?7y#Ff<|JZ^(qpSos9<5i)AVBGGwja1
z9#MovNmV*)3&g1Vab_#!eYqBotrDm$4&%$gH3Ef`|8$(QgkGK1uB9Y%_Xj{>$3~xb
zV%EnzWWU+X`U&o+Z4F><a0S9`%uJ$6GJ$E^NCzC*moks-IZ*oIuxm1unBM7uIk8$q
zTLEJwk1p%*&l@%zD+%VOX&KoqPgz-5erJTElynN;T1~Jjlc6ZD7^@q3{-wB41F8&Q
z@&sX!&g9YXCipn=X9qZ_P`;9tWx2<I2Cq<c=n8nRSMM;E0x!A+QnC*Sg<OLU+OS0v
zkyt>!RqO=)MxF;O&#5phv#itJhU<CxdnEy#+oT<$ELQwWTM!%=I}Zz$+@upT;`hMZ
z6XQfxyz<};8ygjztkHnu2b{H_!B-^kPkVH?XOUg3Iuf~55)GUw2)Vwf38u%2T*%(#
zE9S7!V|#(?jw+ASybQY*TzLWeUBK@70!xeav#8MC#y2tA(P65NU6`89X;eCDr{L!9
zUTK+80YgZOmJZHCoGWIE?bMEXRE;|My79tX<V#U2xXo$tka@knu@3SS3KmHa=0jNh
zy9gWv(}T{Nuw{J$N0Gu-nj~hjci-T2I(F?q*C_JxMrPCEv~yIWf!DvQ8fH1?dKd~=
zc|c#W>S0+qhug6x`%#`4&QQw(DDhbipw>}UlnI7O;rm&jve5QHZPd@+*wm+Bi#Ys8
zipW-Y<naucUMZ5<;<-DT1{xDB((@>;Q2Sn3&S66t>@2n645825uGg+zE5%*s0|IW)
zTnP>hNTl~@_vlh=j8o`0M-k#FQN8vu`@sPdlT*2Y^w#7l8JrHpp5lcetOxC+YQ3i>
z<$BFJr}pJ2c(^RiNT35NQy2w~$zOFk&rYHII<C0-<gqU1XlBcu?+%?TG~oX74X<Tj
z@d#xPbEUI6GUEW3L}_W$_-yKqU@T4y!#x%i9zD>cYu?IFwmmlS5=8nJ^E-mD2tjib
zeZ?k^&#+vci|@G3AOdxR!xCZUBM3WXsP%*gFB80JY>2T^@py*Pg+P77ENXU2{J-L7
z-^eJ@KreE<r4m*i4Hfsal?k_R-!4C8uj>Beb&}Z&gzB<9_u){JrB3yokd8zD!FpbL
zHPm}GdaPtnfCwg*V}%V;8Hm0V5KRE6&jHA-c5=Q)0_S?rXd1TY)!s&-MEa{iIKiD`
zW+w7^&;pPIDyW+a<}53S%`0r1wsZ!e;TxEfsao@At5?S=ttMf2V0Eu^%7?<!$eAQ6
zP%y2MyiOHmZN%J_-_h)GtyRNWW8ft*>R|we8{LXg?;yw~_l2gp9~~4?X4y0*igTPw
z^aEJUU$8s5U=z^;6!;xG1ci71`Yn{8F<|R|SmAeS^SS(nhsF<|F*c4FL!`R;<gcx5
zZNB8s@N<L&!#iIC$Er<Q(GK*f3TA`gmu>v@%x+7a6}@d}aT^mgw5WuvA1*nM2b*!T
z`a5cA`3f=kTuB&k4awyRv#1VlY?KCwszP?NZS2a$14W1$jYP#!euX)U4M%=j9+X2?
z4huje1P7Rra`F6dbKldnmE8D@MD>}fQw;ozy?$*>44qe8uLyur^o3ac;e+!amQlDV
z(`^wJ7G{Mtrw(Qnhlt4dcQgi?a3SmmfK;ztgorFZK^vg=8a1$)M?;c;-{m}k5!g=@
z)U3Zp4C(Ep)yPHe)+Ps^nGRe6J$T50Oa53D(4j@bMiuyTOM8x;9Q+UaQ6`q$lA*4=
z@snGl5Y{20oDZ{jykaMnyw1%}Yod_^LRTmXI#Rh&(~N*Tsou~k;x!<u_k{LB*={b>
zW)Jz%aLDxFAqTb)$w0*)Za>0)2{U%zE5=Q67QI|uuncbQM?&EcBK3#<yvO2k|E^J<
zlj9W9`%<YbZ$Sj(x%>4vYJaAsT#}AUIuJS4(m}^cE>Avq0Mh-OHG{Mi?xy{C@?Own
zd=xonjU$p)!37PCpoA{GaW>+$a&bDx6cO5js81g7;sR;cu;QWAI{+ek2h8ZfrNvUv
z0DIyPC>o-?$%|Sp5m;=vFLyS9O?P)foTU#)figG`A*p8pO=dYoXDJ+2#_74BLN4I2
z;Y=6Cr<|P{4(`$lz-$X-7oE2L<fs{yXMt3v9t4RfBA4WTgPOcSLBBYtg<<X#kTzpr
z9P}3x`P&-Kk8~99;I#AyN>8*PQ5&RMgM5H0Gm^mIAT)BerI0LSOm+j4iRt)bO$Dxd
zN|Xv*%E`<;k25jVq)`@!4G!mRo^Z%zAr*iLq3J*f*ik-E0o!)m07Q;IW%*Mx2hz$5
zj3j-E1Gw3+Ai81F^aon5pTCL3FiCnvrTCLXb#qu(dOKjxeS&p(0j$(N9UZ#4xw!{$
z^&)D84L4L(uJ9195m3uIR9p^*$@Z3yM`6wZZtIxkb}9m305$s2h=a_`$|}K@*2lXG
zJnnlQ90fkjBUGO-&sUpf5x?9YtAgUJQ7`SThHeGK3{=`a5hC3W_!TSIK)i=h)_CLc
z5WW%k81L#ld>h?3*#!K*5uF3h(*UTz=TDz5+bjadBM8QQF**ok5tS1dVWMDj!4I?!
z2?2>c8JKz?qH`W8?D<>v>>;cmDSCs1Fy!b1`Pgm`in|)E0r_CS4#Q6<VpCI(eM3Wq
z4@al~YVY!wRdcjD@Vt$aF&)c+7sk-ovYIY*kjHYwKZ0>&#gns=47Hb(+ctDz*PEV^
zkp!Hs>_9QnnqvP3s#uC?G16nX@QC21IvQb(Of^Y-xp&pp)cD|#`glO@)TvXw6Tex9
z*sI(?*$eZiBfX{|Cg^Jwp>h{BtOa=aB4=aM0{SJ^NzknopMBkMP8`O@R6YqOWxbwN
zS6aT~bOIz+f;~`&sUXIU-$NA6pOVA;YHDf~pGHwPz5C~%0BOGKz>5%0`}09Au8N^=
zRL!j2@arYTBETTZWmSudsHwBw^XcaRdAzSZ^I_MUGyo<kYQ7|+!1ndF8e)8thRQWI
z(JwlZ>IQWCs_{}LtJM&E9iLEzb!2nnGb-V2;Ck+hBw&3pnrB}DCPWIS`l4fC7qxpR
z$CJ*aUK1UZ4OEPi>_o-m=O2IkAiAcg^~cw7YBS8De{!IPvtdsUt4_;0Sm4oFdNS=M
z)`+dYJ<v#RUQtm|m5xU<d5K;&-7@d(Oi{r4VyOn3HDZJL8im=vkI08L`s}3T3G3C3
zX#V*2s?p<4f&@*)X?{vpZLYCk(O@iFGiYGq>!&k>S*oE%)=Tg3+BZ1ZIWiJ9lVu6B
zc}h{Zuw7+$&OeaBToJp0oOvo2d;xr1+v6V+61UvrBx)8oFB}^`F(;z9?c5A<Or+g6
z!{Y`jmZ!c8N0mu3(|yPt25QzcNt~grkis?feY#+2_{r!`@eEq%Y)WysRE#)+kFfSJ
z82m8-`U5t&s8VBjy!j~yBaIt>FMXo^>{<Hp%7eRcMV;B>je6*eUy|itLmk`IYk{)F
zZTR&_;pb!JKoB^vA5)AE%%JDrE+;4FZ3RKK%*nk>qJIuz9?4NnRUl+>++|Gx&#&Ig
zTglC!0KOwW@DLkYiGJv>qzu{GqQ&f>^Awjsa5qe5i!U7F_=F!bI%7bKw;t3U&vW*e
z=!d#ec4q9K27qN^=CBxLWo1<f)y>JH#w2j}5(HTn>zyUSC7SLJ@nBHbl^JXf*0-iK
zPz(~DU{QgLAG82osV#&g+cK|+U(twt?+X~NeJFqEjlAY6Ep$HAa~hMLhz-cP3gHR_
zi<CmL!nNtK1ul7GCykW}ZWP$u17JrLZpfFM@b=j(xVb_8Adq5%mPv+BmugoK^6WI{
zB^kje20(j*LLcK9X-+oxrr{WiIk^MWf9y#qkej~P`E1Z5WalTQyg~_dm_|IR_-EZV
zL5n4Q(uJA16Ytw*Aqdv)kIG?u-P0@7rutuNh&_n+6ullKS{S~M$|pJNiLtTdUnlYq
zvQ{FbkEN=#u<3^3ZH`z#Y=(PcC-*n$sK3gahft>>w5s*kVjPMl6mh91hL6lS{6*w+
z?>9g0-wpk3*v+xk`2F>-PuXM6q#18d5;%n&;Yz4d&`AVHb1b2Vtk~lpyfOjq;eds>
zR9lUR$Q5deie8G4J2K7Tp!Wfa6X;`La@z!6(4cTK7^{E^cm{6Z94<?5UVs&m`XydL
z_L%c$md+i&LQUMt+FFSkA!-~@Jv`zyEdRCQWZlm^xtG!ta0OnQ;=<;6Asuo;p()ey
zMKT_F4?w!Q+?#4yRt6O-?9542ZKgN1!Y7g@eN=QjWdthgiypD3INXVmQDR*tMdWF;
zlQzi4zVhttvJH-_v`BWb{cb3*+zTh;_f|9w9)<=zVNE!~C#4h8eRy<rY!1jcj~>&&
z5zok4<@=)hy@}Z)qVBuT=sr%VdI7qLxbI?-5kS!m3%^pR!&uL&^tGDiYclcq^``hr
z(Lr61luVA{lv<ep3W@2_u16yI10j=IN&JP8^_S_h)7&V54p0Wi#1iX;Wc+^6YRk4M
zJIcZVt(x^8n%1Z55w&34@M4En8<YS8^(FH)*J`B9ES1vwq&@JTG&Y<tl%cU6WY5f*
zhp#Q*mh1zOd=^KjPQANA7&-@9vr?MIP%G*w*;hWHQoKXhU88{8<kY#dKDY*Ihb-x|
z6Fu^bpIsz&C_Dp3`a4Zs>osn33`Ug_z<RmI(1n@s(4%q)=)CZusCrNw*`)!i*I^({
zoI@O*#?WAx`mVcvDnMj-$D`xl_+gz4Ju)>q5T7<*p0tftGa_0aHZ3%{9BbLq39FuG
z@5WgArKX<O10dmVWKzxf?HXHQPfw45ZC)nmo#<Xj8jlEn&PNFwV^5OmlsBIt8;ryW
z88bs-_bJPsm*~k%PTp^L3Acgz(4C;&KQ#i2V=599Dx}F-^<!~aO&83spS>OXr0^do
zb8fUb<fDCJUDAa*!H${AP>2^aL@HZaG|>rq&n!Xr*NqQdr|qF&5c>08%uGuy3fY~Q
zwnyE<5>exZ$=rYKz?p;c*b{Jy(c%)S%6{Yf4~i-8F?p?fbM6xE;ERBWy#tX_5TCQj
zq&3&RcF2}w3aH%93<4_W&#~1200p4q&Bz$)kJxMrgn7l7T)l;CvKqLp>vI7Gr(|=%
zahVJ|B~yma)Wk%ch%S)VAU)0ZvzEPd|JQ&sNtvTIU}<r>pkjpREI86*{DQ-wbv{t!
zaGue18{3u1_+ZR0=@DC+@xVfRuRkm*g_e;Ofn;6;PwWx7`PAMNmni`U-PXUK)knc7
zc=AvvZCt~Q*u0L*130U_LPAWFWz6!nn;7sQQvLnDp5B8XouCiTOwhZ(f$>W6rbjQ#
zN5eV{h7*}j-ulR9Lm*Y-u2l3cKostjFEO{NjgOD>sc#)Xx@h*8V3$bg>;gqVmjU2Q
zjJ?I+>9R1W8|b2&=fH2xTptH)&a_5t4%tvQ{4E}X8-3{Jb}UD3c)WcN4r~4#oi2c6
zJ@_A6-v!*VeHXhp-YUt7qOInHGCh%Lj^J)wR!UrUX!T;MJc&V&hfv#7&k?}nLBySy
zhN03Uy-P_tD*jTO?&W(%dwQ0Ec=(=sKK2pO3)*vF*A>{QUr+oJQd!-hggn^~T4zj<
z07KA!fay3?w>bTUQki?N=7q1nve$Kmnt;R3rwOQHOO~31=QXHhp8E*ve!#p7io-Vh
z$Lw<^A6`go7R02V+Cc@qDibrQ83$nl_mdvU+=tm&3=4iu18UAOJ_>FsNR$?M4HP})
zh`O^-VRk@dMo`<x@bIwDEFtVatzQpf28MVmAkD#Lj~_qg&#FFcOCGE8#55Ol6sj3U
zorQnIh&@1o9UoNHI`~Xw2`wOhOEsd3>z&X+)RYJ7-a@=Z%p5v1R!}`P+6_1~xo~&+
zVb3(xU3UEb_1QeC3tzN*Cfu9EoMtDO-V6e{Exo_j`KnE5I^kwb!^Wd_0b^gk7F#ON
zrN7U>h^NT!uIG3#x>SLp>6BeAXi;%h^e)N$t}F9T>RwRK?8K@83=>Lqhb-FC#-`rJ
zw4t%A+9_?QXE5*kz?bpE$n)xE5l0!Fkj$q)9Q!a}G!bO|KM{#Le*N_-p2#&PTj3W4
zy%5KNzo<$Q1y@P=a%xQ)PC+QmueSR<1%F2Bqky(ohh2#;i%H+ct##4n({-$4Vt<X>
zhZ0LrAcS@Oef}O!7AZt(s+7@OwvG9Fz}j5y5KEVokx9ELji;jz>Xxn}x|Wt3Nhb!+
zA}3+%Rofxc*GJ~7pm{<PhCser*ikAOTd^i-^Q3-j@4NMQNwqdwQ(F&0=VZr>dnjq<
zBMe)El_tt0`+_p5JV%Ne4^kKsG>tS5((RM-dC2z{lM{1kvT}0%fYVE`D|^t1QU^#p
z?6-5C3nP{eVpo}*_81JuWo2jgW?l)JJk!+LL0VnjjC1GC^_vIt>F2G|SH-TBjfGG%
zP*5lc#2-n`fhnX!mt%cIw<o;@W4L3~w2l<xPQBtP>d$TLMB{UQ37mW5J$b8S1hT{i
z)B6pMU%?Fo*^l3Wu&orUqGH6_+S(ftr4n}P5S7uS?fRkxJaeM#B7pBeDvoWf<%!)A
zv@!SI-MbgC@{oznCzw#Y2PgrnT$pJxblX=*72hUyav0Ov<O18mEc-OBTHDx=h}=78
zW6}HfhV7SY_cY!43Hg4u{(+5gGwu~9PIft=8smaIw=8X1TnF|kJAGWIfp&i_etF@c
zk29IXu>|Xoz>(}8IRhVB)V3ZSPE@bu`48(rJ>fx&O~SGG8(NvtTUw0CQ{ExfYGo&M
z21T;r+__M22@M15&Of*&>yxKXk0%$rY*5>3spArlmqI>4aD5&nQ`k490I`qF$!Nf)
z@;qCvYRx2SrzpS6!`ct=#uEkKz71LoQq>*C$H(55sz{I0PL(@<{s~^ewY4oF6*WKh
zQusO2Jg$u0A~KFB4M-CBc7Jz-<dW*D+38V>WG57xg5px{$md1xKXh~TZ1iOK+4YA1
z9z((oO^JURE^Vp1QsFx@6#CDfM}~Ag5Epg<<nLh|VP&nhNLDJ3RZmDrAPVRjH0!vH
z&QSb}4RfVKyf*+IXq>%(N?kxwL;Y|~%QdSBd!I?yGk{aH_g@~HHNM@<<v6JopQHjh
zCjvJ#DUX9pJ92-Vs{J1=K$6~la9c_MQS@yBjo-GNo3NQpb0_H`E>cY!$04-I(Id6H
zk~|k&Rd74@3VJymd&Zv{mNHKaklF%MKgK?e#4fO~5M&mD?rA6=$6^T%AZh2rxf3-~
z6!1yz0pM?(5@C48W#BojrE?aL$9)ibZT%K7a1l`aXSWz94ow;rpETGp047fQ2Q84i
zX^!mjR#sN-91Uys6&h67#htG#x`mtLhi@WRNuPcT0@g9;_>J<#z`K)B9(OYfgT09B
z{;s)iz%-HugeA_p(3w{7@|CXh?8G(VK}l$!I@xtDq*iTf(ycuPE{T*@!k2mFg3eT9
z-K;T7<!~b8zi)@4Qh}rBB6Z}FQlN&@X3_8@jiQUf=M-WSS+8ij$i$A|1GVsfWl5Yf
zL7hmMa7|UQK)a==>JsZpZo#s<aC;EYPco)6dh21n3Kx3%)ck4z#X+%$+saRFQ`$Cu
zo8M%GftWzzed}IAxCCM7xCI8$@c!`2$jA~vSq3;1AWfyS#|mx-k*m=(CjO^QoBZ(W
zb-e8Kd$7o3gA%AgYdH0vX7}eWqHP|ec=%bTB2E#hK!RebV&6OOFomuaO?~O`#eM<Q
zB!6RI%GAt<Al^$A=hsNcT`plS&|Uq&)m3qL+rtNRc&hpmm0j-y@RQ$hdux5Y3ehw5
zvLmRa0`xSe8q3?-w2_Bp!Tb22iK6BineJ~t{uNiW32M5@k8yRCY7%az92E7#CFA(l
zXe82&c_WMiS(t*%e;<z1PW*OZIC10NlUg`{4#Kl(D5eA!7TJOPy!yqz*uy7d0FkKd
z!ln-<Kn!XSj?;wk0y0iJj3(;KnZp-Etn_td4p_{YXoR;_4kc1L3RSRrbA+zpC=l?y
zX5k{k@uj;sKj_DWv<;w>8(I_j7t~C;HKuIJ%SG=)xVwB7o~#2tuUxj5=<y_nm#>XO
z&){Ns_hW(xp7bhUMRf0R#IrQNl{C9t<}uPrOf$)S-LSX+1Iv2S(T{`0>efvNW0irc
z8el0Bqbg$jt!vHiDG7l<8Msl%`XOnTB-BtMJcR4E#J>&VDpTW>7gaCfS^%7z*xtCW
zWADU&%50^KkuDd-iBAkR@)Z3UM^Xx?YBaRN8hmv{)wJ~h{&aLTxO0D4@7JGo`OI3J
z!&J=<WN(@(9_)-ch?_(q)Q@4y78Y=Uwb~SzGVmoHZKx!GdzEoNl!D~r;*}J5i=JaY
z#dm3G?#Gtw^|VDB&T7vklUDK6_MrOkf_n(q7jBrwD(9&}+0i(b(v4+MtZ@HD&pF+9
zH8MMs9lhpE=(0pFTMRsrhKVF&(hzw^pmh7Y+j<d*AIUnFeRsI&uMJ+&u!2pFgzP2C
z1xS3p02y#3*2A)?5WOF4Apu&c6!s9OQ&3agr!cjId4@b7CQ@SD<LVlmOE;475tl^D
zn@dL1jikY8!fEM#s^|N+E}4eYwPo##K@CBc?nANtO%`%(4aI|t8VJkbpEikq9GNC3
z_e*n_r+<@w#DfP9UI$2+9MCyEbcUg`bQSgfqVfCZLO0f(yqYZDRo^gl^ibj}uYSs^
zjt0xp+bjZ-p5)w6(K+{C2?o=}ivTys34K*e1WuG<)5;{wLsC}?GWeLw0oMl=kDdt6
zieSJ++58+h6ly8nChUDam5Uln5@rjfg4eI416d%PD6hU;tD&s*xWQi*#w_H_TunIl
z%0}%eoQNmp-Mt#VqO=8w()V!$Bh>VQUh$8O7ki=>pP?~Us0A12RDE}X>TEt;XMIj~
zFbguFuy<7nt$quCi?7vuxpXo*S=R=DTqrKNqJ(LgsoVJtDi-@ERWNSja{V{;0wW-9
z6$$oU3&Q*bV(S_x;g9SeBpzb%5F1Asc8Cm{c=-D6`X?|76J673oDQ)f*C+{}FWeE;
zTOdGnw_lI`{dw~eE?-S;e_G;B{GLS}fqo6AOA{ftB(;U1%a`E6y(bjk&-!#YO^23D
z5z~&162~nUTkx`P(_@XIZi5xS$Bs1WILd$%G2?D&p6q9;sA-eHC8*r^XYH<Mv36fi
zoByxpX%<JXakAWA`9}gqdF%Za20V6{?Bp)zA!UI@?zej(LN<~psQt1;(b6O3DD03~
zGtnd+z<-FY*B1LF5#*qf6IK2RdVV1xL#dk1eQV>-ahR!Vwe2EP15`_*MYbmG{)`|`
zVjO%A18BYS+ZUPw)7A+0RBTJvdE=k63|9}=-7|}g#Iq{}@)Vz~?O?8nc4t5(6(>Qg
zOT^VRNi<m9a{c25BROZL<OJI^zB%nTPq{JC3CwEvj0@=DULSR-<&tAT5h0j#=MH=m
zABd&TF@5aVIfZU?Z-GmL{jm?fxNL&F#7t-EHy$WzB>|inV6LVWqHVo{)rjR+UGtr3
z`QDwag84`vWk$Am+Q%AQ>j`^Gbn?EPDr%zIyUuul-)I^LB<0dg<y88iC<>R1<zXE=
zaWfCvgWytIf%tRNIN;1}bmIXHj6voCwZ_%3b${ZwP85_;7)+Sbuv17gWUg<0-rxdY
zrL0yVr1JOM9Gl~G@-(KCz+RNjx8PctRLxUQ`~oS7<}@<{6iEp=Tuq50y^|jlJ&XQ`
z0MQ(1nuEywu+)Qt+rz=nX;6L?hi*e*06$gLdq103d4%Pv0${F><u+EzX1WF=;f|e&
zb#DRoSP#6xid1Qo$b=GwJkb0WcqYT<;iAiYs}lu2_qO>aBQDevZYD`IKKhtL6I4W%
zTN-$vY6d`<L266!%y|dLP6Rx)H3|XYQ*rG*4y#%w^-I^xTw8ZAGlzC<I?Fg_Ezc)L
z^Y>&H!!h^)=y&6Q6$B>Mt**|9qzkI8bgNT?daTxIT`St<z|~LdoSzsjowl19GtllC
zZOy8UIhT=ga^1|cMAosJsp1WH{c*~No|G&KAx!?7H#|lkd`D4V2((2uR`+?rzE*O$
z1aZB)W;YEQhj(OEt8svkc4Opwzt7;><TVJO4gsrCX%mBTqCQ^>+}`+va01BWn3cgm
z$4kIA+W6ZbT9)j!H!vJ;JkNzdn*3%g<ZF|9@?FT10NSYBmYDwj8hG2@!E04MmyHx0
z#b`!Pv|T_I$Lh?nLQokuY#K0zjc3UtPuSy8%*fKx(iIg?Cx+8c%C=qK{En-&RMNUc
zk<W^UC#5V-i+wyMHuc>^3}<6aId@Jjdj$u-HQVHDrk){pWUpV;ZI(y3e|vi5>*7MS
znzMV#+YN84ixzB!UCIUf7EUGx3YIg4Sy#>>-ZSsBCD;@h_x|hSJjnp6RGk8Gh$)OV
z{O#B3@Y9|6uJtD*5qqb^la~Fd_uEkfYsT&3>B5RI6gyCnH0@jmkrAl1#GARMO}wOl
zjWo6vfYz4OIn@TomxKlm!&D4@Ht}O*B7~yt-N`_6BvIXMs5JiwUAV~C!d0wG;t0Ww
z-}A9W0XU~j*la#7K9W^H^<4Sogf^zy`+@T;CLjci`x6+2-wl1=ew<i7h&?ag;I+HY
z==dFC1?hHkVF<)P)yPLXSA<UTJI~)#hd;Gl?7bwLIwmzXorYScYqB!+7Y4G9LWW&2
zP}16`fgpqS><pdHhHlXXnrdo|k{t~TvJ~Hb9?O(tPhU?@2@NW-iMsU`Lg)$Zq|F;Q
z`ugiQl$h?0XO7K<lUmiuA3%AFgNMxH*#6<+_~QcYDB?M%v~_g2$*9bgU^i%M9>rE<
z%))}Juaf&s=R&H0zEjuZ<Lr4dj-wLCJ%V4=_HX#zgzG53p)R@*6}_KT>cKY+S+Ob*
z#rN#~9^R~5Ff5T>Hc#&{^`{VpRooMR`rx?3`11k%krT&|NPtN2E64P?czBjOYLF$Q
zfg*Z0<<6^)C&9rcV;AGu?KbfD0)`S0H2(^?iF8j6MnS#!oFF`|jd|5Tm1N)f$GVB$
zckkXc-Iyn0;Lb_&DN=;Qw*cgMR8?Dd)Mx+u@;C;c?Erxy&i?cH80^y~+163O+578n
z024|Xq}$YCu&=jQBP|r1W1cOINT_f5g@M#bq`ge8KyzBJZHEnM2%3?#5gJ$(v~Al5
zVWbq-E^uz^o+VKW)ZfRW%323${9>L>0jN3eLDhHoRP_*=$LlN$hIO<_Fj+mxw4hT3
zUzbH5;QGMfX)qr#I0n~_zJUQNKBj|Vt{O+cWW&bIo7Di33CxL@`}zC7LhNQVdL=h*
z7rYZ?Fi{*YX*8$mv{@|LM8<^>8tpE#&^jAvH4e~`)q5D*K^fhnN_Z%OA2OPD?XmA>
zc;p2CYz#nM%qKOlY@_Vt?sT&3C2IqZEU_j3U{eC5c;YQ&Mc82yubq=_3^0U1ZA`OK
zLc9V{_DID<JwRMC=7%^q%1m2gvJs3aZSII#-gjggjcEG!U5HSK2GJ6Q!6y=H)~(aj
zS_1tWH#W|nAqK9frM1H(G0!kH5!HrTma5|bc4SkdEpX6EZUwf<zB%lJ7iz$)+Gsuq
zXF43-iY{|49TUw^BS%3lTz#b_$T^PWN_Wg?T!LZC{CN0MohC+fRvI?$h1|S-*3W%A
zQmxP334(-5KV1ejD@TJF%#KYq9D`U@1BlyJ0yL6jyx`@v1#yJIuGtvX7z=9GEx~bo
z?g-eESQARFW^cyGn4q|~;JYpM9#bjWNUh)-#}DXRya&~aoNo*oRX~oQeC=9vlfdzo
zG!y@lt_qObd3tH!lP<O;#^YyhIQb}Up(V2kGB9c95u6OwrY(pdyg09udz9w0>x+>9
z3pZ)Xz}yKyRY)#u8Olaa$hxG~gtMr`mPU{_^$DAPJmPIeROD-COUopcB0s#77_=zt
zD0eX0o5R9lR5H&U8Hh)l%pvwO4A(S}#tfICam*2t2L`Q3#3|4y!^Y6#7Gc`PywiJV
zPJqCq%EgOO5ZJg)rJ<Y6sWXv7wR&u(6$GZ4%HGvlvOi|h$fvVs@7}!|jYi;$Q<9&b
zUkvnBu_(6&((Dl%$=OgxgPsI7pZxw*2taGQ5N0<TgXv;N!SnEeF)WewiE0W8p6z?5
z+Uz>wp)!uUnPzI;<&h1nZ~%X#0?{GwY@`z|%}H2BzA#6jQHa9Be6b)jn|Bgg+jg;b
z9UUD=4%FGj#B?q4@LNU^L5z3A&Y%GH(qTb!_jN&IF&d$v6HHl@u(!klS#A3=%gD%R
zj5d@qWVf$ld->-R&83qCi?i@KC^w+0f9x!qg$0INRlQBlU=??yj|OQ2!$bZ43?#uH
ze3`Mc$;U(|KAe$!RU6o<YO_{Ll8ql4r5q6+6CDYoHCj;E31qcFN@jyrMsgY_Yy$Fz
zu4FVGfd}2R^AeZiI<1#K{0+$NRcTtYx9Sh{kDy$CBX$wc+5oII%1C=Sc2SIvPl*zH
zG5Jq?ZFqD#(^7r6o^SA-hr2tHSf0ZRbNRZdr<d3B<$)pDcyD#kG8o$KMz@!u6D(vi
z5QVhU^568ykx8ZI76#DOq6l8T<0|0y1Es2KTkJ+Tu54gnpqjU~A^Y3$5p249oA)Mw
zdWz@e<=xQ0!Na2lyeI}1EC+r#GT!TVxNpexQ8Y&L?2v}kPi+*NjB$4OKV9`!AY~!X
z>F?&-WeRf&R#wUD*RQWJX@opC4f9M@<mE5DcD``oAx)DCLXhye4_DcL%;M*C1#+$_
zrGmMAyB2b3E?m8n^1;q*5UjIhx{Fup?IN-+iL*sQ{c7YtdIjAe^z#DVEzy;?rPw0j
zGBM?S5Ld43{X+t<npiqbW(4W&2;432{m9oj3jycjDLPDW8V)HYiZWbhe0KM?3C;qC
zy;~r`j=^SN$8MXMo7<h6fC@~$si|p)2?egkD7|t&J%=~$i+KnXukGInKvM}=>EACK
z2OD_y;qy|&a6Rs>JY7cti+0oc!K-dbVl<g47O4d`Rie{TIBO}mJ5Toy3uz$T`6bS=
zU3t#%JZ&@{o0w1rW~mAtR1Dk=rEP6_SDJJe$9AScN0~PnzcG_3ZEyb#Awpys7j|!6
zzn*3T$KZsc)K_sA^_|g-EBzHE(U@0WbppT`z5MKI`frF=)#TVDZ!13>Sw|**K!kC#
zoC5qn3d%kw4%ahZ_x16Cs6||*-~QqJ%;rwiKIT(df*jxpt0iR28|xTfrDFntJoH>K
zV>F2Qj7@Jjqp*nH+*&OZ%;I9B+FDw5pmFd>^RBXWDh^bAofLm?V{qH9T`^EBDPt2W
zqZNM~7TEde)3Ze!%^y=Mf^NlGnY<>#InSA5kRi<7vwiDUy$NYW>!NkXcLQ}ftqFvT
z5~%9|Wao{l()&i#Y+jX%)XGLYK680p6nu1Bw2$uvw4gG%OK-qJ1n4mefoPCyF<_K-
ziSD7Wgr+VAiFI1I;&ZB<jv)6IPVHErm(jTH`neoO^QI@<<>(n_?WILrXRqhtfvEy;
zWqBYB{SHkSK)b+uG4bVUEv&8M**u_NqsGRpVrQcy3nP6LaUNt0IMyiA|9|;rALTST
z1ek_qFE$Pd#0YmPN=8FfA0RT4VSpiw1x+b;WATRrRUGo=q1P}b3g1<S<8xTmLv&RF
z>=%r~eK1HSYvxUzEj0LR#}5guty*}w+}`VakIUiY=AC^Mzy`*-PWVMh-2@=kSX5b1
z;HoxFI7CK7D2LeVoAYk3x%jqqZ(1}>#-@|ka4z0`dKOD^0G`YAm|hAB#&LS*&RLUH
z+!7-al5A-JRqyDo<B~x!Jb>JKe*3lOZPsbwY&fQ2gq;#U@`9b#nG=R6?$AAMf)i^|
zq~bf0I`|x(8k-YcL%Bc$r;@T)CG4iOhkCZJ8L}vb<#8mXSL?=E^tjy~L|x9h`G(W2
zTXBm_$p?6N<dzedcs!2q11W5M$+w#;$*6D=K{>w;zrvkk8{4Rpy5l-qq%x|_U053x
zbF%nlxc2Y`C@OiVm~Ujor5Y#1in*@oL^0M^6+}f9cS;NBc|)M1fq;KS$);`i{yNH^
zOLDQ^_xND%;`<62?d%&#H^oC=;B;SXU5DD9t6>nCD^y_lDEe-qO@|7kBQa~#OsN(%
zLM(Q@UTW3Arc6mGDcq3()k4$5Ek-JPRW@dpmX&Qxh&m#mzLowVtUYDWpJvXw00dsr
z80sV!qt#lH3-+v|;KEH8^RGjJfZmWltE#PhdGBiHIjH0TNZ-M8Cf6QQw-l*mZL4~c
zrdFt;s%mJJ=>r6@InB^9Nuu!GZ{xC5C<{7FL`jDV)qgovhSQ4_wj~gpQH+V|bPMt(
z=)c1bR{G`CXIN)^0V|UCmfdFecUUQBVASuSNfI9~Jwx5@h4Y~lsqNfQPqSLI4WS_<
z5honHXOL{ng)1cP%s|pP{238D1fTR{b5#Ga4~*<#N)91=U=7X^cC#vlInUzoI{{(B
zcCsgDzw9RkUrCNB3?}CRPR1b!9L=%;)E)d+KDaO(ow0k3Lr=1eKaO|pCZ@w!-pNwB
zxY&6S3q-T8a{+iQN2@?wTWy176M?dn7q(F%FsR9jX?ZFvg`Z3wAKX}6D1rFrGBE0r
zgIH|?SO_SiZKg<U@$!+^sYemvctofG4`j<PzsUDDR|JYWrkCUdJZ2Vd=ze_0w5P9+
zsB3R*4N|w>7;df~3KhxA&5gj%B@&_xX%_n6Ct24m*a;}v@dh2a{5!vzl_O1z;ndDn
z2Bg4F(ns7u!Sits9zR}oyy3-=<BgbPNi!vmB6hOEhA+)aU=SMzRY`9KKjUA91knrJ
z@t2MBW*6`z=vj~9B}u{Je!2kLlDrZ0S|=dgP)wut0g}9f_*;B}53E!uCy0_<wz5h;
z!LDa)4ft#buOsQ}XV4PqL-0TrgPcaaSYVT7q<N%bnj$bSWkHYS#w@G~7UVa`f8uM>
zvHo!>K$G}(Ghw_lgStz`V>v0rhvQHVL{K`R;8^RbdgdQK0r;i{FT1?1uON>xhr>~&
z6=$mz64BMM1vh{L06Qz4&;nkQY&zY5c)ZAJ1mjiP!0U`BPkxyhh)smgPQCqR*T#Oh
zq0luJeaS{GG}d`%b16_AvO`g>SL~FawcW)V-OTHu)!?AZlyZy4@a>X-o7@seI)L5c
z2geM#jVM1wwt(R_8<oNi>juyUw&t%YfArvyqMyYesZr0<d15PozL4^mwAaVRfk?Zc
zs;c^~i{7H>32@IO=2jLN!H|*6oE*(oS(Gqvlc|Wf15yHP^J35$3X(Bw0yUA!ii&#S
zr%I9+#KrGXIt%cdv*O}}P`um4$;QD!YpbNPl1QM37?}9|&=C#}DbN$D-#*q1+QgQZ
zOJ^ZXXWNb)18vdN)Y;A%U3gV{K~~niySw|mxjD&FU2(P%G5u0AShdK}DRzGTmZ0NQ
z9KHz#=GRyVg`JaJ+}t#jK}AGhvRTkXEc;nR#MzqK+U0-`2=PfEOTYQLdoNpOr;&kW
z9en4wrVJe@%uxW34C+HSS;Ws60;WwLn0XZ$$=KTe)v&$2-8X;Z4Samc89^r+@%ptq
zJXjEQoPzUh!hE|$JUJ<sdsM0O=Z`Szv}f<_;6&Yz>ekGf3o(ISzP<<X|0t@evZAjc
z1vZ$7q^v~fJ<ecc_r_jF#JN!_z@B2|dT818$EG?^@!0;HDV=rm+w<<;Udhv^|M2nm
ze}v#0a4Z*)V{>M6ary9OtQH45Q~o|6r8OwY?#$pggM;hgE<5=q$Su69G?zs-V;Ik)
zqIb^M8i^7m<>fwj<kkH*5MsgtlFD~|FlY3IB6fVz?RNFSu=^6Cm>B;2Im-pCgOCs=
zOmPl_eg6aZr)u(-LGVDb9s+&vJnuL;g1X$EcK5Hp&QrlPQjnUe?B(r^X_#{OYpRwK
z1t`h`a=We@Y}>X?nJIBQ-WhgqB+hMtbsAvhV=}GP@rjU{@K1~_P=%Q9MwPX-wF4_8
z*bXAOz&9?A6H%4^__rXz<G;C?MEvMySgtIlSvNS}IAd#@u5Vy)0Sg5QXp#urL&BK4
zleGNBi-$q@Cf>Z&wT?l8_;mUVo#qM9vcrpM{_W;cLy$%9;9#s>-$iRE1>>g~En{(>
zsLm!lB6MqLXvlTR0P&Las~Wz&tOS}vroA66hZX_>q|@{4#qw2vM42T>ys9k$7kA+J
zaW(uMgv@AyECaV6r`S;qQ-?o%NHv!hfmQ7zb#=vp$~%-lq^R%;4dsI6lUPtZ&LcBj
z$$0^e9%$VRCYS6eEC&7NujRUsHJ9Y13bjBMxE~G~aevGbTo(RkZ8eb{uc<lc%CBI+
z^LUl{duL_Az`#K9(W~14blASOY1NU|?j;|#k^fpglBrBM#g-a;b&q%_we8zd>`<g-
zVZ}!eb|YvX#l?>MbE<zHs*!)}_IMl<3O|3%(e5zk*2=szG4c?eO#a!Wz1&FH{F!wB
zcX#R>32>n>&^(8<t?YmDK_u|*k`7L&9+Xlxzw^Hq*U<?hf666iR7n(DRj>QYio(l;
zgH3E0)~veyc6bm@X)8?CxHALWS=2<JwOtHg4@uw&Z{|Vrstv5)%(buchwCklQ1ZL!
ztOssT?kdH9eU2S48ft+43TrG3y7o3P2Ctnu3crPG*{xUkr91vd3t&Elqp~qmYU|1s
zDqCT<z<}(IjlIB0q;U}q{^cs)Mhf{O7KOK0(j*q)XS;^8k!wI8(Asg(Q9N=Y-~QUl
z9;3sw?}yA`H1#Ebp*QrR20xERybC`dlx~o$tSpCdvTnu}b)TBd!B42x9p3Bk<(cH>
z?2Rj;nxHn}M9U)(04dWA=O?+VT6bN?sEl7X8eeM1Jx*1jMHmd&$X-np@3>y7!|CyQ
z<BE(wAiSYbF>h8X;#v<?W{4?^Unh+|-kZ{=qIX?<k(yd}^*eNb)`yf=t%%x$%H4}g
zGrbt3%tk+g>0Sm)oz_JCW&qgAVH||}TT6$7xJ}h4dg^$hkogV;rC+YE%-oMNcGs<s
zU6n^J6|O*q>&U(bbm7NOc7><w5ie1OdT?+Efll>t<UNH`(8&J}W84_5fES$J=~03)
zuCT2fH3_BdfeW^$5n+}YPjk%>o-{z;xZ=`T-~uz3f5v}P*$6~=(n}dl!y(mG@mIut
zSW#$+Yw7@6auyaC_C9qRKUEo$z&wT3TJ<!u0#fEs?X}0;7m%)0^vUYikE`YwR-D+p
zocD^KefuQ?B}n7zRPB$HPK~dp*H$CrlfFJ2`vF;H^mTRDO61zn*RjVGkx5@c;uN<=
zUo!hU|BW>MulRA+8Ae7pMTtanwEYRs>nF65hW>rs-Pe&|g4d;?q2W7sT|tsfii2b=
zH2`?nEjIl7s9ra~m27~CaXT8jwjj=ozW#qd+$_x$FH~P-uBlGLH?|5?BAy1vmcuwD
zp`%oljlp|9<w(MydIxSvww_MP_Z>Mw8M~w$bQs4C41F8N@?Y$ndUx<YX<GmHRsYwo
z>;JPiv9+Bo9aI`L_>z<jaT+8UUOP?+m}o=NVw})R^#BY=jij)BVZTR;l7czd4J)`Z
zQ3_IsQfX-^Q+_-^s8X00I=>Fn4%(DSQ$PwO0f|%Zc;BL^qCy74Vut?x&)mz0qUjNc
zknUS_VG5KtQa0{eU~EC>@=?9w(j9WVLArA$<0z(h-Je2a-Sr$?y$IqzjhF-20Lv~f
zG$LFBEMDj9MSKQxiRJTEz^dE875ZETh;8pr(zgk-EPq?|eh(2A)Rde0bA;Z1#}CSE
z!$pIb|6<Syq$Zqprsfv16WdW7%AK|nbPmh^<bI;o7`HLSSmjLPjJ}?;{6?exLJ@WW
zJ%#pM_t;rHn$qRpE^0NPGIG$h&Hnt;PoC@c+$=O-z`xRZVfiDomxLK<i6R~F^T>#K
z#v&ae2bO<r-#WxvAol9Dl&%)V`pet@`^VjojiSO*8F>YlDw4e}-KUdfM0)wZs$BBm
zOT-jPb|@-rqmZWFS=s&EN3oI{pYBu{iHV@(NxU(KB2R^i|DxsZ<qV;rM)Kz%ExH-m
zB4E<sMk5oRx-S2MF;~BT{p4#-<c6>+e#~{F<);_*91?}|5DZ2(H|srdpq0Q(SJ!Q3
zWy9AqP!mBXEyE;_$o|66yL6KkZo)uA7ISlRfed{5Q}xUD;m6b9!KkAjn4|z1pw*k=
z@L4-6>3Uvq=}>xzXFW*6pML(?`VwVc+HG3CD1{kCZoWn-t&zs<e+a+R$IZ+Cbm;2U
ztIuN=1!sa9vXB{%Tsp<CjEmyB(QnJ|d%sC#ImTRiU|@sVndP6h_TQi8(CsPlPfJV7
zkkImzzNEYSENx+5J_t$G@vK(lfkM%%_INwJx5ZZG`fn(EBA-4jRA@sYXWd0JghnUk
z&z62khP=8Mf~(jp7CefeeMk-(2p;?R_c^x6q4wAZ#Pp+o?R$`bAB3-`rxDD%g^}^v
zv~$+~49WR#z3u-uor-owRBxq{$%x{cY(s$~vvG072T-0qie~hQ`|*;z{QOu^`_(HS
z#ozw-V0l`D_EaC16O{oJxyk?)_W=MSZ5rV#S}1L0w_|N~$`;byUl@(9ECPry$?l6$
z<m;9c7e68C3#`Z~QdmZ}#p(39y^CT|C~ys2lMhN+!5k+1lUG;J3Xk$c(uqFIVabq)
zTG|;YW0R~1SS}BZQ^|b{l_8lZn<DY&Pw##1lgGb*1T_>nXIzVL)&XMs4dYU45TT8f
z@${H*aPB4R_TKspyVXiL8sGNe2yKeaBe3y#p~Km&-3#Oq$gC$e^ab1msKMTu>QW>V
zmGL0NW0jS#i75b?T1ub_5HBADS3{TrXmUTWA}Y}?CNs0Mk|^;h`vucvsSzcW41CBw
zfT7qYhq2i}H2Q*_!AOv8D7-Bppa1jEA1K7FyRWYdVCez$_2B+_zzW#f%L1>Xh#3Gj
zUNA}^ee%iQfB#-mTIvO>G5qj*?TWGYH!U9|A<yIDq$%GE*yt0S(j_4Dx!MISA<?4u
zh4-CyO%vUZ-%ndvDO@T7kiTK0sCBO!X{Jto?_32;A~j|S0l&+#$sS6ly0dPx>x1*b
z0%qM91eYTMR8A4Y7Vp?V!o~^)!k>6NaCLFb`41kzL%M-kNSSi-AX5qlaeA>{mhB)G
z7FD<82&5df0W;SJz_JKBC6OyYi35OLSsMv9Ld{tMw2i=10W(~Sg`7RV0j*PJ`5O^F
z_`0!Mjwf%ztEuY`wDX#6YPh6T(B8h&a|5ld(PjkD0Ybq<Wp(wyiXvVBwkj}RXTyM%
zmKo~2FdH~*0)+_K^zv}&t2P3DNC0+!xqdu%Fd?T7N1B<ldqy)7wq-__^f03t??E@(
z44`M#Kwr^+euX({d?1yRhDYz6?p?X!)Z^=rG%<o7csKdpo!Lai1;4KkDRzOl|5IN-
zbta6|d5FS0ND9R7v$M^ui-x!$#$Y2X-EEP3*@g@14S%3rlDOcTCMib>JIo4CP%T<q
z6A!T%GeRkK3|jrSl9SZ{L5Yti^Hya&wvsTz@}@aT7&KA%O*YuEgvJ{-l3)~gKxECQ
zvI-pShX9b7`Hp=44htLodZy-Tz`zJ@ZnB|REdrsq;uT5hirkc2zH^im1cXEiwumA5
z?I<0*kFKQ+s(|dQ1Ir<WJ-C7?coPl={%jGX6Cu!#2WS}U++jcicyMKS0qj`H6gdSc
znIA9}M(w3P*6n!$n6wyJtC=J3k{qT&MK(Wt;Kn4VsAJ!=y`-#+@8S+5Wg!$Vsw92Q
znl;H_U1z`(L_|fMhx3$@20-C0$6{k!1P|I{h5?mp$8suxT<BxvQr0^KX~HO`3oHBM
z1duGO;Oc#V97{rLRxycV;K73<Yj?schcBhbLyqomZ1ju~`t+{LZCM#eb9PhAXka$l
zXX6jt2sU6Ax%oj~I8^o(0K>=TwKczL9WN^_RYXX|UC_!eXMx1l<B@Tt(}axbVSo}J
z^n>9!+y2Whm!{I(F<KIEyb^vtB|MZJMV;&6W`&@EHA7uh^(8Qw=LjvCY$g%FPMw=?
zhQ903Pp+LdV*U54NX~uNokzDHp*F%-bdC4^A7Hp9gLKO%aIDAaummdM>|$*JAs2;5
z<J}UkTaZtvTgem@M10Ma&9_a^FoHRfN;CuRPc&+Iy5{Rke^d{j1pCO^0?@qmWupkr
zxJiU%4+#P6IjY$Q6Rv6cg79drUUIzbD<zyGwzDTI{@oMZA0N_QC}i%!kz?pT2QoWU
z?0V-Z;>JpBu}S$_rh&`Rqz+bpcnO1sJ3{aSN-?;ZUZ}+)G`y5rVb}tDY|WQj)~g!q
z5O@7@7HBS|L9qHG%_4wVf)g-3g%zZ<+ysA0Atd7dxV=i?fQaV+*3`L!YMQ6-ORB1l
z=d>yp*I{(Fx7iJ$&|Bjb$9H7pE$4~IoIn5i<Nc$!3bsW0R2wW!yDS-{$43jZ85$Y}
znNpPG@~9D_M999!>|MSKAh6-O=_T|D5pi+lrh9=u(U2GKGLQg^l8rd1Moc%*C<tRF
zSiV!rNZ)(I8CDyabhg8XpFr){^f?{D9~In?VWzx0HK0l=2?%UUyT^c_({i^r42o{l
z10i6TA&e&nQ_H1yU5|G1kk*?7=T@DCE1Te`2@!Q{a5cKLn6{naAvhk~U7ABht%XxK
zb+!Thd2|{>y9(o}rakp|E8-Syvyz0@L$nlNwiA~kUPY{x85DK`88gtgaV9817s1<U
zWp~`U*cKon$>%8?B1Q-UvkwAb9hHW?R<;BLJvV+Cf!XZ#OCv*E44xRnT?~b9T+G{T
zhy;Kek=N017jE~XS~NM+*avqAZeD?$N<zST0kJ&@4i0`%8gq&57T^1ec^T|%Y|+w!
zWDsWqJgcnfDW*FL%Ao`#SGmq-^}s%D#`-j7RJD%F?FD}+fG-ovP#j<t-|hy;2PVZq
z-hp2qH%Fiqg`bRlII~%O`}4r9yLPeTF!IH#Q;rUgi5a+f8SA3*Y$1vWV;|gjfl$Qg
zaN~qsI}Ynwticc~<mACE0lTyo4ShFa>J1QRE{fYHCl+~g>ZjcB4BX3u_{+W|1B;ey
zG<sB;n~&4OjT#DLH!`X|dc>C_3vYb&(I6j1LX&ZpinmZW3#M47us423SfX;)W{Ja`
zoIWHLIyuw56}k&gAoD~XMc}h9PsM;d46=TWp+20Dzo3=7^DUF}@uNxH70o?q@VUiW
z?WK=%X8Gg%1yxEJ?97}AM&_Z&w^85Y{QdHaG8z~M8due~(x=81pgkeObi-RXqX;9Z
zJtf<stU1=)0D&~*68aD=JB%ap#weww6{&^gp~4xGHYnDVclgtEu~&oWCt&5O{rp{u
z0`736H+PRA3L{txUd+Sbdg5<9h9txn9GyzGlgty;MC90l7prK<;r&_%G32Fy8VEGy
zyYkNi_WA}~VYIA#XYmm2xR)!jZMYJxa6^{hw$#V<k4zr@qzZJ_zePqyvic{gr--9B
zH@BJrOlz7i6%HPXl79%X!t;!fF=MVk|FY-S*VC3ZX~R#rg@s#&ZW`buTs!WgNP)ot
zrdtux3<rIB^d<&Rk_8Hww;skP_r|~2+=@9Mr~Z=KhWm6C%^k)g()pGNGLtBQ|D}}7
zPNaq(^<YJU^{2nrP(Blxy1$I&jATo}lFtDvdG0(AY}R#p(;&L=6H0JxqxISoon*9w
zfNQ)oN|hr56#_$FFou$sFFz|0_)kbNZIlul%n@^|bU_VZ0B0Vnt_KcC8j4O@TGpy5
zD@#jHl#upN%NRgoPm~j1)i2oiLUjHWUWSvwfJ*ienrA+Sm=BG#5#Z(BNkx0SAx8XF
zyB(4ss>dQ;zAObLOTQ0MNg!=0pORKTh`6R(sy=+@#KlXYoG=*cE+O6wwXHX0jN+1w
zc>44dR80|h>10TSsf*^SRO#zdy$AmFMH#nF?%~ngad*^QZ}sPJpPj~oI|MP8FY%oJ
z;fVkx5Jis{@vftzptK6-SEu3fF)pr&1OW=3G~|iS!S(1zcsc$x16%1tC<$N=PgM)#
zzlY~_3<fQLlv>eK%L!?GEoT=+9&sSG3rQ2#te62!P{I_Je!L=e#2+Pv#bka!<s^yU
z0ZIi)6mA?sp_MPwA;pWuga{?neWOnY%;k0qm_C6@KwZm%l*QLS{cRkD<L%$a{8XHZ
zb{q?+WUq90EG^na&e5St^TQ|{CwLXF9<@uvtXYZid&<7@rnQ44J$7ixcQo>Farpw^
zmjwpH19~lfYzoAK0l;}hWo6|aoG*o;50@9D5;RzXmY@GZLY5k3roe@CM=Qs+g4>Cd
z*H9$0EqyN%>&9SkUu5>t@H8y@cUM~6+%_PM!V4)<We}Hk;}WrrE(rrMwrn563ZOhl
zJjmepo1|oA{jh;ca7uAPgPU$Y2R<PW`A4VeNI|9MhJe*~G2hZ~mogzo6q1zvU}gMq
z25dFQo?*9C=v}KG^(PRK5=4MARpT*7?V~yrpUkk4jfZDvsO%x3P+4eZ4nQ@Dd;P$n
zLlRi!7@bl-@GWiV4f;U}<)T{n_U#o!64%ry;&SkT2#F|Wl3DQb@x_nuv9gK-{4utg
z!N%>z><h!Wn~-uHIC0|Rmt@?gGXTlC8WezyWTB`(fS^d=HxsOgFruGEgf-M1l*|ZD
znrrIX&p7_CA&K)CE(}W81fNW;9cnT~dHLU<HXFl3DaQUoLG@2U3(%(KCthBkpIFNO
z^dE*po{MSw_2t4CYwQ0q8vkAN&Hopm(;Axg3H>OzT&0jS41!znf_O(590jsy1i%=C
zaH@Le@aN56AyzNN{ScCD_G*n5O_)Q>osk(Xqr$_GX*?<uyax>Jh);O<XsVg@L%?nH
z_1{IBRkSs&S;4OW<pqNKo_@AO7jqDM{o(d!Lcad~Ww^J_vPB+8Y}PD|yE~+^=IVVM
zwK(;5m*e2~M<seldG*t59IR4+?9ZO169K*B=WcMuR6by;?(u*Qs5Gcs*~<PyDvB*<
zPbeD)D#nq82NS`GXcjYot8OQt%58<2VY^2oh8a<Bx|QR1n^|-WJ!BRMSVBm`mg{fW
z!1Er2v<9bDWM<~Y`T6<fL&(V~R69rkDuXff5v2e~CwuSTyH|=r$k?<kBN!fl|Cc1>
zGcJmun(OPnwB=+#VuP7GkooJw7A%|aV|@?WpJj;>;I)ZfSQRuNM;?lZx&DfHZ+$~U
zNeD71hZ{gt1W<GodJq-3HRJY>4nL{A%yUCZM48(mN<M2_t5-JSR_{j9w%y~!ix)JN
zQ#S6$+bLvT*Kr)?zcYh2h!?bUikSzqFsr({I;;P2D8JAMb2Z3oX>sBT=gTgK=mKI-
zqomDOJ4mqx+=7Oi*{60|{*M-bVjLl&8RW7<83TPYIcW~SnS<lp(0|#K4q@MAt1O6Y
zVAaqZ>O=MB10Ty^MH#?E3c_r6Sz2Ik{Hcpl2C1c|UJ_T|#;M&5p1A7BK}CTa046&B
zVqe)>Nj>A8>r!X+5&2{I_Ael4O(=7l@`1~7qU_pXLN-%i5Wi}__hq!ZhgU&0vnjk@
z&fh<+dr|a-!m^<XCW2$b9+8jcbMNr`oPz>is~zzV)c&*-LHlEXSL5aHuXa?9?I4Na
zNE7{jt+nhcCW=UqJS#1W+UZ8!%Xk5Z4D*8EZyz1<GT)8c8qNf>?X@ubglc=q!0DEl
zVG##YM$^G(LG!TL9G?wL!+g9)X6s-jnX={qytsYmPCOdCE!MeABm@tCJEm|4rE)Yd
zdk)-nDRZ^)N|5i70zEc$n1s662`Me`qh(|l0F>OFIt~3d>bY3OQ1_{idLDsJ8}~R4
z9X0a5+DYydtY}r@F@}$WRZ>87y`}BUT3nfv?)y>og=j0L3By13Zd^`M(#lp)AQrXk
zJ7)njbVmMnjL`wb<JR49(gX(Om%th-(nVpNcvlriQ<qeVaO6o(y|G*vj?c~2f};KR
zHk(Fm0jIqUi18*_Zp_oCPbFI@f{e?2kcNQ7B*LLvah0+ZvnBb3G|+0jYy1s%@_j9a
zPwY?p19YJqd|GUPfa%B1A+HsWl9Q9kYPoiN|Ni|3SPLp6VyHB(E){O1;R7hXsi!rc
zDBo?3UBihNb<H=Oe2mCLR`#9EOe@fYM!lHejtsL?%c?<y$x*Y#{$3>e8PGKX=qRQA
zAdV9)?a9px)J^~}8-wgd#i(QlarF*G3^cjh_CmcmR4Uu6TtNL)a<6+`3W?$&pjvCk
z;k&=;&nMdsVD66P!`r(_A_69AjLdcLFvyruO^DB71q+&_#S&h(xQ7$F34Ar5`}?mQ
zalwLK+cEupD<HsmF~lGUID9}ce8SQ$qiTB_q=R+B^}l9*!V~ld{4Ec6PiXQ2B;BR>
zAVbZ4=-_cQ{e5W>_^{iilN(yhu;h_+un)lY_Ho4SQWrr<<b>h_aV<l4!-+I^iqMC~
z%quD?%5_poQc|COWoX+3yVzD_%21*@#dRI&Kut$R5q~r6tzua1hHG}3^I)~typr4d
zYT?q0+I||6vR!X;^BK}L?`^W!AhF^MivQnS+~+P04ebP?!6;}I(U$HV9Lzz{7Md=Z
zqR<5G#RwpI8ivv|NwN1ZP!Noau?5L3-+v~DfJK}<`Q*fxoNov*8}T)t#>1qU+3=8Y
z^ir}4iPw-D<c~J$jsUtOZzMg6J1;HR<xOj`9i)M01e|IZue&}+@!XJcF_!PeBvw*~
znzxX6x9O;yr4f02FxsVlNQhFep#Q<3D2hXEntXYj%;4{icC&ZUyg{Mxy^(HIEHKQ#
zL~k}x9*CFHF>*XSK+I*4**<HdTW$Xh`c&TjTJVi(5;uYn2|_UeZR+X+8A2czCteH<
z%IwGW7Wo#=9qL|2H~i*^98{m`8Tf|yXFYz2qX&z*gkIs{x<?biOwbAZ!l<H+K)+>l
z7roLu5qvk3pqC5N%8$~}&A|uwm^-C!^IaAetjsZFn6#wHU2UPE&4m19<)Y0WgF%YM
zE_Wk6h{CpKaN!Zhv1yOUVe03<e*G$GLE%$lNO?@D8Ll`wSx^ko9Sh)=O-(QT+N$jo
zZv2I_2yFn-9nfh$AT{Zw+($9^-Mo<xPcQYa`}eK;B=#d=j}zPH3Fr|DrG@8`TE0O6
zrb!(mBcn1LmXzL2^K${V>=;eH%n|8B*B-!Kpyo7WP&>oZ-~SMpA>y}yN@Je0;dI|6
zi>2Ad0=LS&C@0!7Fqw%Hu$VWwl}4O`U--n_)P)e9vWuwqr+x&OwcxEe9InKVfic<w
zU<QQD78(RJ5gHiCMk+<LDfC-HP>cwNd_>tcMY>36V2*c*?mUEg1Y>U5xzp%|Jva&q
zWdQv_5#PWJOvb1BkRN!xfdUyYy1J#+A1c42R0`58z4rOsToHBHbHT0DakpgEl9b&U
zb9-EikQPVK28|+-S)h606wM4qX1k|Y(T91+mzivz>TLaTLx_aE_{onT_aR9zWS^nh
zs*j)7_ID%y)(d+JAC5>fgy;&4>h}8h_?(C90lW4*I0?u&n|e6Iacc^?EarkQzJTwi
z5^P^lDH5jWp$G~JZUOMdz`&r|)>gF(jUH|as)5iiViM}dG#81F7<P-4tLjV6M1~L4
zm5j7#Vq;Oz<4E=(ch0mK0zcjLjf~KXy5sR(z;mKGt++;F%;r#}`6nmyLU@|ivUD6r
z$o8TK7!2Z1oY=;V+kv{~OpKnFWTRNIk{(i*)9EjMe>bAV2n$?hu&9op*W5cj2F(&B
zYTQXu8n)r=f!Khmx#|`6U>#dG&HzRhI1J#Hyy+tOGdofM(RrabDur5lqM33xU}K|V
zLN)U5m+6*J^HSI$r%vugsPJPNS-}%fq*`L`n5)Vw^>&A!zk@Nq$e%9{MJ0~GSD!7H
z#e{FsBy<BYC+TJ03^Ab~Om+_+=j1H^Vz4b8N16`^vr{D1$DzBn><$V!T(cB<)Q=Az
zx%4fq(b4IQy?qu<m!c4Eiv4nPmja!!9~hc9JlGd-K5|Y4Vsf&bWAE<KZv?}G5F~Z~
z?pC2J5N>0*Wy^rXvv04{MhY3=sY8H0j(n6jcNUj7`SX1+6k7qkvFS*MC$@zF@1-xW
z6P-!Vf~}o~i$(&y+|<8x54bCFj>O*4+(sN<M=uuD)+*6R5;$sjZHF$BF-pY=vvmTP
z3R8frD8duSz(D5+m9@!m>gDm-B43uKTrcN$!)R*epo8T(Hrcqe*KwMe`P}45=lKhF
zxBdPT|FP&lR@6xS_+<0>DA!~CALZp4*nU|5=V6sK45v@8=BQY^B|_4hJ8$R8$c)uL
z{_^|Jmz`a{RxH}u+8G*ow;fnD+9>EK%j?+Jo|BW3Vi%TW5Jtm&0Z}2LypkI<!Gn2l
zo@A)NxvHIP`g#zlRE@Vr&>mE8FmH-M)*o`c-$W+Ej&leiVFfcJ4?h7SpELQ9=0IyB
zTb?q=R%g}}+inWN%YNBEl0v{`uz*bcsW41B+pt?eI>plk*uIoU0S>rmg%|gpCF6Oi
z6rF~=bod7;LZC2&iiW%x&!GgN?C?$FT)O$W+YZYYj=30X{*&%Z;gle*OK{Xa-__6$
zLkFE71uiZ$!Q@%vmAuurFx!}h<lLJAIUEqVE)l8P*w@iv8RqLWG9qa(K$H6x1{|=`
z(~n(g>J?Z)vS5rM1FB_X;1rm^_1oE(w^5|Oj$%buuPcPVCGh*xMx*5IGJMn&DbOa}
zcQ&7Vf~Aw~(+6y^ha49Ob18<hX_))i&)W|FMjSK<1R4sup(oIR(=|%03-@p!hG@{)
zw(VR}L{%W)E7-`z+Z{(m-xl482aQmMA8&x0+IDN&It*H)e~318KtRB;p3w`aeQYxq
zFk;lQ<+_6jR*g4W9{XGQ2Qdi-n|iEPPOJobl1cUya%~PmKJSnAsSKnRC%NoGmhgWu
zbF*(nv|tJS*1Xoe(qs&wQ7kbxaMk#u;4j0j*V<z~jVh=ZaO$!n3G4j6x1SpF-HDGR
zh7<FWvwU2{)w6Fz_x>PAqm)Q?{vqu#4qpoK<n~`8I}g^0N(=eJVX*LnpgJ=od_R(h
zNwWkR&}-lE7@4>GP59VdiEnQ+*f%#h*cbhLV=xZ57+Iy4($z8rm=>3F&=8y57ixJS
z?X@V7Lhkw9v{inO$Aw_wm=jztfLS8=Ni)Rn8}PyCDRM?kIKzr4>{Q^EyE7fkuEVeF
zx>e{ax8YWMYoHpVe4z;hxS=!SH+2rz=eJ<ZmIPm@t&u;40fUx7qree|Pj(?v`Y^6t
z*|G<2Zpt{_^uOn!y8VOBCd%3TU{VYcA^FmqA70JCgQS{vb{>?MUZ!lqEWmN*27Z*C
zk+;w&_SGaUVOX9}c^_(upgBw%Y`X50KR+gQ5Ndnw`Vvf5B5v0g7D(X=7o>It%|vkP
zir}BYhbX$1MK+vIjguJvzyVr|J1>FPp!6_$!ed(E;!SYE8-{A88TVAfNt7S29Cxug
zp%o4Hz@9tcXzjA%@V|x2%Q`y5+R7{)e@8WM<M9gn-N<;s)a3>GW*@j6BRHxksBSC-
z^AQc3DbrG>FM?i^%6s&j!!?+gmo?G0gMnt<#{56*y?Hd(d-p!9@pLLmGojKzgff&&
z4aQIslCdOZEM+E|PH2=lGL#{+$~>!7W<tnJ$V?fN(sOO+{;u`>_k5q<TF>*>eb&0y
zeNOTDyoc9ozxKZNbzNH`{#u;cz&?l3b2QD)M4#-{VdXzEo(j*ynj{tYi7@8>g=La9
zxSo~ObSh`)g4y_Gh+7M6EUEN%Z49hyYk>xZ2L_KH%W8fmD?59TCfeS?q3<i`Bav4T
za9#Uzk<n3w&Y!DECDF6W&uOz*^lDkQL;b>k0H7M-bo$Wx<Bt{kl#!T9;Ci}1s8YhP
zS3DnslQeYtv7Gs9OG((iecQHo271?~yVn;?54PpEMOIcLO{8a6kTWhdr&~Ynj58>X
zf?fBBA&#yNTQf9j*}gJ!a3s>08;tAixRg2+4zFkGS2FrKpqEG^iI3&qt6KpeD-u);
zMbvE1P`KhO6&}d7>N)1&_ZuDkMqe$As=}-()u|1?@ZY*TJqUAw^;zggMY-wgHCnXh
zH1zGUfw`WCPdS@z-y@{Kq!C0+!3^T;r!n4j4YvNdP;93dSG`TV;4PGXsm+gCV_0}H
zT^%i8rYs?BrzN-%x$qGrdy;~gdM8PZLEeG={7->2kasOL&fw2U!;#_~FX367UdOL%
zoUvTRxsICl6VH_r>=2bB(Sxv+G`ZN8$4Apfi4teoA}a$6CE_EzY_4KLV-K=eMFgwu
zXwAo_Kmq1SlNb$DHtCQv92`E$Mz{ZB22KO=o`&xOFC-@FHZf!ei4|zZ^tbW^Dug&c
zcQX6o6=>KFwvAmzORc`FBU_8cy$25_iat*RgI7X-fhN}RS4LW4h}kp}cN&<4{`52J
z+Hj0o5WM=sQo3}52iyMYzCD)_sQX%Tv&j#W$?Z@tC73iKEInQQ{Ko;$DzGt2+x~~J
z1AFd4pk@$a4t;z0f)_ki^ZKSy#8}nx;x2yVicD^G%@`j-3XI874x6;Xqku7LBqcuB
zCJHy{mz>`VxunBDLi#Sr#HH(Yag&1^xLb`a6OhKeAQh<QM47`UWRptpJ~cK0QsB$$
zua0S{$m_YNt!Em01BpN}qLZ%na~D*Z6kkf6&=@*`qsj*-Qgov++)NRfVy2tShXB?4
zAY+e9TMM68nma;WNeGs*TeB(02Uf_MXcvq49_MICqfF?HW0scc#PKW)#-Je^c2b+o
z5NrGaMp3q>-h^2I0tgfSh*e5&ZZ6WOr0HX%9jj%#`jEO)v1vY@X?oh`7+#yq5A~3x
z5(~yJ;i3Q@bo%|Wz-f03&8fkcQ?G=GQH6Ue1;?pI!wnU+5{6f^J-q*N0{xaFKsJN<
zPbOb%5rf0}T@n}a2lvn*5Gqt*$ynZ+5^zX$O}5p+SS)D`k8&)DQrwIYY73e`5!JTo
z0zqPzFsje=36>UFuq6~^%5qF5*7vhyAYDuMGjink`^ioZ`4x#+qCA{oNYA-}N~z=7
zJXg$z;F6{R#X6ILm4AuCgU=vy@Io4@4kzkZ#eC3kBGJ>dRW;1*&Ej&h%mwtMh=|bA
zb2`mjYrB|OV@D?q1yIl1ikdGrI~ud@sDH?w_Z*>=8<pfqKnmeMe*EADgCpd8-oK{|
zDn4HNNu0;=@!F}1G}%or5a$^;_-7|u95BD@W3JG-rdzxdqCBTwJ|!J)Eqe~47l9*+
zNN&_;sHQOajbk#ZvF8p~$Z^0y;@S0{j~_>4RLl|BS*Asl`|&-Yy0tvVB0m&Q{)%>J
zio&sDb%V}(ao+onRu8-(FAcY*0azFMt2jDZO4@%rNE5u4w|Uk$t>Ha6Cy_j_<tUi}
zKA$eX0F+KyJ&AylTKQg-rlrfDuO8TmG;2F0DvDsu`M{T3^lCot+55S;9bRy9wiUQG
zhiC|n>YCq%yZIzD*2t=cTpc~)HqFQ;y8Gxf^ob*HjGWRP@k@ObR@hOo5_oQj0d?Mz
z83d<cBpfe!uB0GysWzAY?DAK<lO<UrS09VU#6Oyhr_qQt%u8-UEhDcrPdi7>XiL=I
zBAJ(8y~fjn^OUr4*q7IDg7g3M@bmY`{?%VdqOV3am<6p`G^kR}gCk*$fW9s>GtHzD
zIz7N~oNVVozFLF2Hxdztplfv#Ok#*2Ck_n{BtD27JF{#?U?GQ@V+kqd4JW80$~Jir
zr;e*2MxSKs4_3G=iyyeua_c#=>Jk9CQHoCRu<i!Pak6`)rzdK#DP1M~K9(dm+NCFl
zrqI3?`*jgRPo!zoqPj_NYOVR{<km}$j-U@Vc4Q`D3J-sGc>OJMS5O2j#U|Ag?PiOY
z*N}Fo=I6JuF*O>W(UO$nXo;@gPMG*xW|o0RYU-D2H=^%0^ta|F#F}${eZ@9N#LFG?
z@Z$(Xnm&6BKX4;cOm}L<fk;!%sAMD=E59$amd<V#EyOyk&Pe{tKa86we%-rY0Eo#5
z@{ck)zk+|h?n#>~BCofPew!~n!il12K=IYt_q+NQELdRUHY>_gfZT@a#;;O1xs$~^
zR-hOV%H>+18-up-GsU=lpSDPXXcxg#SoCq9tiy&!+8LJ5zK;F7W2kNhmb*{A2j>Pi
zfB?T=81JA_o|1;q1Di_l_>70GFx#t}w!%;{4aXZtTBIJC01>HPdu1IG0F49jNaH>!
zKT(~Rh(yIWBnl$D{uBC+^AcgYXQD3*+vV8NBP!R;z)d-JjV4U`pkEw`A%n-VqYl6q
zB+#gDQAYeGsA(0E=q|T;^#^)yHa`f?grjEGVY|e@V)Vqb!;46ZDHNQ{a9Iv%sZW8}
z&zTu<#-vjR5ci{l4qR(J9zfP2%Px4jJ;=d#YgOBGzOKb#<TedUp{Sh)R7(2iCF01B
z?GNAA)fx>9H`c{aM#kz#G(9ic>KcuJ*je}stE}harqqF#hG$k{yQr@Z#3(L$pTadx
zJg<$0pqP9a!3QtX4nbWbNr_aTJFc!H0Mt}Kdd#`VP*P=C1B0bQ``VC)#8r!L)RscD
zY3VF#M9S_9m`rDsA9J-XI?7Lw1<0@zzE>mC&U$sJHbTlS6R0VCh>Ww(86X#tFJenh
z!C&Y6&-+U)8(c2Ncpx>&_VGw&sHaZ(Y#vgh7XHmo>KN*A-UJ;>cqDY;7{y?63-m$F
zGw_^|k+C+nt{T5CrRcXS1`wn?9EChCioxd}xlWo{Pm>hFS|XjC*pm-yOpZ41d4FG=
zxCzM*Q3@)e;%Cgw$7|wm`8<oNZfG<b%b@>P-!MZx4hi<J-6JC}W@h|YHiQ|UJ?%0b
zi9B3uY#Lxz?ZP}7<BLNqOI;4dP$cXg0xM_0V?lpHU9S*qfkf7%spzB-1<C(tt&PcU
zwHV_O22N;&{D_D4_mWoh>i|+xApPb0Q~6|HrOCc?W|)5Wh%3bY@Ypm|Fw;2gG@8&S
zjSpybboh$lk&2M$*Z86yi3tnb;DGiWVG6(RIz4XPos)8G6_{I}Uoow>-hg+YR3MQ8
z@ueEf9m1!tVw0Ui_yuIXAL{aCGQy)%5nP4(W+KF3TuZ-jy$KQ|0CHSJV}Sml6#KzO
zGx>@-UV3^L#QC3^qk!Wk!Ty9sB$_v-93g=z`Q+l=>{q^qVqsTuj(xvOOB#Nai{`*|
zTqURm9WOOThq`JgI<jSVv^oy(l~0xGrR1r|x#}O}v&nFX9p7aUQUV4mI5puq54E7!
zE&>c^!_{E{bfkaZFF+7;lfYyq1cPmK$Xm;GT^Hh*H|m0{6z-!K8$`)d{ZS!6{k4B6
zQNH*SEgdBcgkkhOjUn{4FW<mPFzeF}FcH&;WxcN)d|&DU`u%~PG`gXJ0L9o}<UcKD
z-TU-OJ|IMLg+37`^LOdqrbS$Ww@&Jmv<yY#RmQlEL^XmK_zR8Bi#rX0*auo90Ua+!
zy+Wq;Ak^+4hC7@qTh{)9qQEP$G8%uM1m~P-d;|6~-xZLhXo~UqpTLP@@@is$Z}(f?
zyz1#z&`(@AqS8>ViL!zdkzf?{oLS=T8T!9tgCU+NjjEhvUi-<pm0CYE(~jIFqD-(h
zP3s?G2}jX<^YCS?CeESsI>5agJC%7}`Z_N=#!T|T)8a~Ch3$5pru|IgP{x#IkK_#q
z>xD=-s2g5~J(CfhoNMVG-hutbfQlv8eJXXYY;;F$d}=lisYu5!vro~_Hf}q*7RI-M
zcBV8C8%_q<)))$-{=jb2?lae0ZhQI>OqP203(5!f=e!6oky5Y$IDx!KdOBNipil~R
z@i#NUSn$^7JRdC%z#2=3Kf>i_A6NPQbP_!f%NwHJ+DXi!ID+**%wvpF|A?A1S(IQh
zjVEkfOVbcSdlw&LXRjd^>DM4z3@7lyIi-Z()@V{Y?O(U_A7q49rPbwq-@%WuuKaUm
z2rWGVl(T))<58C#g%oQ=Z5r)Uq|Ae9j>c6>&Gpgf1%ND}KO4@8&f0Ls<$A&4v}t?O
zypa2KVA5vqzt1*TFr7VSrja54z(%J;txXmh22u;uSVITujpF;%teu^BEZ%vn;TvCu
zdQdYN6Bhr#?v9(S7kh};ZaOUvoFePcDl_0R-bi0{W-)Ix5_}+W<&Bt|YN|wLSUd%5
z?}mx$&eN5d$qD5E*mGq=JO`$eew{h<2_>dYkINI#t%<885_JyN6QmR$t<$+=GCYM$
z;<-s{nQrElV&vKyRs#6EQlL;gR8f3&d6>7)(c%SgU}aF@B1~U6I60LP&yX041<s%(
zox^D-r|&9{!Z<@DA7Y8#3Y)7K;6=)6qhp2YSiwifN|NOIoI_LcZh35-uiN(#k5>AC
zH4uo{J2`rp(L^HexM$v!Ek))Zx(~Iq1T>$ZZY9pVd`U4aE|}6E0l41%2M{bvV`T&T
zp@bA<Pf5tIvxU>V52wRb%W@V6du(^_K8wb#-FhU#K(U`BLBrNEmTMX>Iu`FFlT|}A
z+*=tp?m7F9lH@2`)iTp5nEBgOohe8AEnU6_h0BmjQE)N#2u+(|3QcBA(z3FSI{M<(
zepFK$uRjF>nSu{^Wrr0mpszpRD-*cou3Yx0W}~6c)5PH!<;0Ka0gtc3IE*P&9AIP)
z>^x&FK#Q_#On$F7?ihF8_GOhaa6~h^=8S<c)ZW?s*9P88mzDKiIzygW=Re|AcO$xn
z40N4;|36%S*pq$+Eq*`2hyiV`V#tUfns4t<-(I?Q*EOdv&1amY-x-VZU|Kp6@VPBe
zY0SZ%`QeipVcfCrIZ3@S_1o&CTrA?a+2ktJ#Tu!R&CZ?2WV&$Ub7Si(S=b}n?_x*A
z%)c5l`+Ms2G^)=U-*kQPSo|TAlSkxF9El<a3gDHNExYjk27m&X@ciy#Hs%z^Mm^<g
zO8W$9h{%G?Sk<XdxLdSgFD5L1U39=8fcbtacNwY*jL<#``+bY_*M13TnQ12fWd#z#
zpkonS_%z6WFaB-cLGh9eGp_`9$0Jd8t%2a8se1JXU-3z}w+aWPA*mwnC5FSUVJ;qj
zcABLFZ9^{!htX5Rf-BIuL~6!^{K5-ZPUg@t4Y{jzG2qvAAs_L2{Fobc$gAu}&=jTU
zT5r(vt?}(F+B<|m7ztQ7)&f~}diN#ZFf<H1_K>;W1^H__mr4hHPo)f>)cI>wOU#)!
zH(@A$Ap6-r<n*yq?fHg&)&yCBXp5%ok0;^(3eBNt&eL3E@ic1O&OI>!prWa*StSZ~
z^h3mQEUm|bt8`2RApx=5emJ*v34rU57V(aOp7#P^Mi>CAM`#63rLn`i2P-MMA-Mx+
zx<8(49C)sPYk?ttcFv=crK3Z#FLEk4>@WYzU__z6_oDyRRg9TU*Ug{-@Dur5{H{WQ
z0I-#M5|QAQU{COHu2|54M1!hyr&vd=w^o&s7xsq+Nja0BGbdV?zGz%zCdl868;XWc
zrhE$wp~+wh3;hZvh-8Nd)wQvQ4@ILU77IH28smfnXnK02TUPia0M9~=L+*g*H7*HW
z7~*bfZNx*W)io%u8LwyyF;!xO=!wQTXIs^w!s<qXsW7(Gw$DT!AyEn5hXQnRl{M8?
zDhP<asDDy8=@ef7Fcv=j7sVB2wgKm4l9%`eu!j!Go$Aq#hG#?0Met3K5`jnxf19`D
zUpq|{GW24s?ZdyLk`OffusmB;owAM=Yko4vh~4*`tLmYkD}$z(@dT^oyPt=p2l(B#
zCJOID$wre(2_ay1u>M-{5xRg~$WSv;wzVW#z@rV!*25sNcp<r`u{dZ99KlknmbhQw
z9Xf#u{2YD;!j}9Zpd!onrO6Bm$faalVMFDiJdq#XS@e_t4N|C@)P>gPSKO%Tueg}9
z<;gnyo*y#irWONK{y{m=WtmiHjwCvvFFTxvW-J|yBxEY+vFI_s^1-&~X#%==Zu9u%
z$Mk!1`89j0lOai%79JU(F|*!CDEdOUaDK!ty?93=rzs29jCFdypmQIvLy#|``*%*k
zvGcRuIgv}Px`02uD>My)TJ3mp3lur_Q7!4rwR+X6=2iET*u7@C7iIuG>Z&g1-<p5J
zBM_-LjXom|(`rw*{O@F_mji6$#B*g9(%xa65zGOm@`nDZi={OdUy&x3AO$8%E@DU&
z6ugnBdbmn;lRUPfC%SYouGbnS(@<Hv68$++Fy83Aj^fC#p_nvvW8-MBh)z*&qjoYQ
zLv$J)I3qZMnv`h1Na&1b_YIWeecf#;&I8v#J0_uoC+?h+7CUp3*6%vC83rwuF&FDV
zP!5fm$A6*T7B9gEj#?*0_W1$v!PLfs03yStb9K;v;CXEW{y-l(jyw;5aUw<1K`XlY
z`_lX2>_*GyC0c*$+IA%d3!~$N<AA{9@Jw$GS)g&=cn^4U9yD<!Npk1TBg9n)Y`e<S
z&F(TLp%C2Y`ZYZ1YfR-&3~WHYv%dW4O(8kAV^|aoArb8C^26o|k#?u_9kup}CO9PN
zqXHYn?KT`;6^MC9UpH%_`#tRO#Irq7-`_%0U=%fV)J)FAueFYFB*7fVkg7ZKk~@K&
z1~OJ+-<iO-wV>zjdI5f_SJ^l$QKxHMFP$7tKorRN_I6pIokc=rWEy^1V`kCf#jIJp
z3HzFl4md3-M9r#RSdK8GQBYR>h0Koc{l-#~Ts1FoJSAL)!y`@IROK^{rc#(%MD$l#
zLa9igXk`6isFfn|B5CMaRF8vMh^c%_#${=8m!Yu?fs<1*;)?ISUq0s?Vr6%bN2uw9
zW6vd(x;QB5N!!=h5)X^Eo8$s!Qld6uRorW<jxx;VR<T@@s>#l#wE-=DSN1l)#5V-?
zgi@zg=(SNcLj@p(!LKvqri_fpEG=?;{yhXX%?aWA?(UC(8=5^||BT&e&Ah~26}8L6
z2gX_UL@A!EdVX{xk(kS%S9F;f+NXtrv*idkfE$b@<^j;+i=w3%6l%fL?=IlxlarL5
zWG-g)UYq@O0)P++OYvA@T-s?s;dIc0inh*@%jcfdRb4*!Tbg;phTiIx3Od7qYz-~l
z14lm%F%h}tDo6;ndU<GE$7jbrTX#V(<pJ9QN4pX|_Vew7zYJ6Uy3T9mmBGVXCa?yL
zf(8a2xW^B-WZ`9-EzU?;w^0G6R_4YX%jKzS9yqgALZZ3a>XZkzY;xs-^J*|x*6_$7
z4I!F$!HE(mbrXN)V=ZzF6$h$T*F0S1aJBISHntz}>Pclm@GlbKQtalzMivkoy#V19
zk$eG_A=JQy%CX-I+Tzg%6bk}NE>ZUcpg5A@(5;_zl7KgC<b{`9jj9i;n>ggt);ELD
zz`tm@g~M4vyYO+x7xqbfYrzS@lzY<G#-z37I0%*VZXZb9U)G4LNohI(TrR7U25gUI
z(rrrFp-jlU_9t(6q@75O>6RX<Y4<LIs!?f@ec#q+&(}+t5?^$43k-wx<HYT?v;6L5
z%yx!Up_=DMUg!%@_HG#Wl^@>>sFEc<J_br8vVKj05n}1KuO`k1ss|qIYdyXw9U27k
zaDWmc(#eY;($bb7m;{*Gvp6&vR^Tm47C%eg|0pRdO!l5wvrBi$+Dr`YkUz<`%0YCY
zA%@Y$cBFEy7_ogGX_kvk1ytC$LX!*(qvq-5ZgJ`}v5#`YpHmmBWA_{Z2CZVN)?xiT
z2|f%gf2QsEDKPm%LhG2;_b}aoA2f;NQwQUDbpYklI6C<L2!I_yrheAlCfQ*YK*L08
z0%O=cLu`F8L)q^x7B?su3S`_!6d-ud9|dh=8(xwRBCWRTEF5}K$?0uV36ZcZh!!fr
zsjTrP0mQWuRJI`0$n9&d0f~1Sg|QwFZ*HNdg}~~j;xM!i_-L&v8J$}Td5hP+OZsk-
z4ib`U52E!+gAYd$_M_9#C(*OBG3Imhw%_Q$r}c`xJAM$K{d{2G=he&p&Iki!05WbQ
zOZ+i2o#MhmE@?Tl%)=u;lGyeCh;DuD`@X@}<C7I~6f|=0cVA0bD_|b}LoR!T(^(C6
z6`}<JuZPP8ne)q%=mF9cvKv&gG$yVDCZA#8n~`h%2%w$Vwauu=&qTkPK<OrIBP4}+
zO&xb4roajP*@pL+%ZHL1$x?*&o+9NG*9q>soG7P`OugJBNh=0X?9;_0%Y^H0LTGGw
zLR}<w^&+altispWCDdXJtaKW6wnUFvkV$isBFrp-=tP<f-@++<?sL%Jj^p#KD=IP_
zE5!%ChOoq63yoh36u35feg%y|O67isHF*vR+p(U!8ZRH8ZTM_8i<&3L8?x9tN>)ac
z8Ox*ak|pU6sGh<;twoofx@WfDN}<*ctmR~=O-}yhr01X$kcbBE66W4uHwo9w*QF|~
zo20Cz=z=i@*9T8n+Pw3f%y9oH%7b#G<#y80Wh1BNfTJ{oc5%jd<Jw^iE!YYSE>)*o
z96Y3>am8i{d-+CRvPY&jK6RhnHWA>Uc67uBXqqC7*G8^{+znH}I|$?&xAO<UMXv>^
zir7{Ck)XIKa2KZq6ITDTF2UBvT>*8z(90cK8~I9hRrneW0;x;5t8bR@)38ars5dDL
z(7ceYf4uGCZ(v6N1u4NjsW++h^pqH<c}vp31J_A%V~3`$t$S40Avmot2-%Y3#ZAog
z&@*+T;=_PG23E3*Sr>&HIXO5KXI$=V_B)Mkqe$ossuUZyv&7XN%S79ZHBWSWVZyyA
zkTd)>6G_=a=nXXlo(zL~92;4iu*oD-w0dr`9LEzx<xe@H$JfZB<cbP-(Rw*pD}{e$
z@>#v)Eac-39k<u3H<@$k*uLiqy&Bg4n+1U*`^wXaJp(0gxdUn+DhymYgJ&Tuz#8AS
zK-2HSJ}aGA*eeh&6}`?6N>ushWoiK2MM#;{{RcgU+T)hI+qcV($vp}I_SCPhs|}k(
z%|-ES&i=@>B3Q*zVdqe=y7Pjp_8S|O^l%v&8fcGx>t!W<$DwEtKw8FHa63dUWC{iK
zKFLh{LIn8X<;;J>xlS<W^Wj)vcuK%>tSb``wF8H^6Q*`1pt=g14${{-a%rLBN%JSh
zso|qwi^UZ9B(u)f)oMoaPNFK;>c(SUG&;wv&wQhq+(w<W(y;zlL)-l{vbyBu<>Pi3
zmBow4r0LZ^SiGV->um*@nMcoqRol^8y<ZeF*+My1uMQ6^v&8%tlchOEi@>p{AFO-A
zY^VMN?;x;}The^wjCK=%G7?aCSY7));Xm9Wo^Yo^=x*x(E6Vf8eK$fNK%{^bA#-OH
z^)jPYFS`b9GL7MLTao+0D-I}71iI8rp=xVn$v9~D?)^q3Yii#Ir@{q*I?5VNRM7xf
za^2+U{Q*pvjmON64$NSNPt%hY5PH?~F92_|>a!E8rGguBvlAA{daNyeHsW7UIM^24
zV#{=#h*jDr-Q4hS_s^zG01l;!H#r-JxS1!MUP2-*ZMC0F#v{)>o^X1k(WnZ&>?kb2
zW70B(dM~MKH>r}#K<WB1&J~q~XweD#Oae_)OTB{<y?=M`1kY<KA%P5J>%NFTJRpI#
zt$|zI*6b!in*6RpF0fr#_)~`!!OEym#Rp#zTcBkTjrIwRkzRA_)-4f_0^}@n5VG}8
zZ=^q^xOs~4Xu1XI9{{$mmdrkZNFOzwDoPN$DHB(a(1nM3h4Ve#dRFFUHk$XOmLcFM
z4Mk%CwX}{SOBN2=>pL;xEz+oJx?ENG2J|Au*cvumOYTEQO+AO%OHkjQxBG-h73cSy
zIvJ#1%d%T~4I%*lte?+Q{U67n=GN7A{E455_5CVn)|_DxY&ZD1!+LNWrKd*AcRgYQ
zl5x+~-xKv&eMH?bji%{UOm0jN>&cUnA^;tIcZr<jhWu;Sh_I+M0X{69sc3>C=RRiR
z#sx=l&~^q+%*wrg_pS(jlZNT!NX4<UvJQ`N7dq*CB^na^s4PJM*65jUM(7~2V+Dlf
zNQ~3Gg2PBF?`kG-xB}-<WKx7grds30cr5B#>RZE+!E6dsgr>UiApHpsgI5wCO`oD4
z2Fm*G{rh4d&V)I6p}dQl`90x7*alL`?P%`WR?;Y6LfIYqBK;!}9LNZQXyG_$OutSM
zgE+X7>u6vX1O}T9Bu;N!#lbN=n&Sj)wkGG<BdvlbWr;Y21RckXX>#YTCYp8$AlM79
zWn)PvN@)`GXndpVvlBDOwJ`kIbSdA)@dj}m#3eqYey-56M#+o_zB?yB6@C9wg3(7t
z+tABF{TJSNUfdy-`a4<`sYD<}YB<2ZlInf9CS6yWNOM_%u6Dnt!xpfsc|o|bG?{PS
zv{&zZN+erqI8C5#ai)Xp)?!9Qs{j&oV7(47iHPHn2mkBC=rwQzzaG8+$yTkO(H~?W
zicPu~2YOR)1F?%Rox+XOZ?(E-8e9na5g&*eiuGq@<j5`=lPq>(Z1E9IDKbjNXrNk$
z+D3p65kRD><1__mieE%TL=is02qcB9v6M)oiyC$_c$-G3OPHDt6O{nRB+);%fp3br
z&zw-Htb)dwwjVHFqPq>Q`%uDz5~wS+USF;JXaLcER<X0sWkjg*?*RCNsOlh+0ZwGg
zS010fF@F(JOk*w3>%E*gRD%{=c9nDtIubpjUv{VQo{b7)H<JGT2#cMUGi<sSTQ=(M
zUUNv@V#SU^kG=0Vj%|%T08VxVMz`JEbS4L2mfs-=pTM=us<7rt)XjCQ?T|u<QUkwg
z*9>?~JJGAIPf#Jx{%&#r5gNiOivmi|3-pse_cR)|FC!Abj2y+t#FlXrBB?||$xjWU
zD?JtLEhzxa)M*Zdbm+6F0|Q`e@FHgUS>qRv19(1AE<x<3?dWR98XlJM0jw{od;sDV
zx56rvufB*cO_3@`j;K=ms4?o`jrrK!n!*Ftm;pjtDb^WG+z8kX7Z;01aKl<mlXf+#
z3VEVc7YL`dRiYZ$U%oE^s)zP(60mGO26K=1mQpx(Y9x9fwO6)lBdPCHP{7Y<^pP)0
zQHTA?*D@1sfvq0cU)ss>xFM5a+=xVrBUneDJ6w<vT0NdcnyU!Bph@39dFne%5NevI
zK=7_@*~bS!Xb|1}MBV*SJ%_iVclcaO4(Eej#U<Pa(-@C#hGo`3|KF&e>YYP@j)pd9
zsDo<2f$87E07x|UVL-VN9NvhDKJ0?#v9A}eCVy=nl`u1rStweFF5ggzMwM2osc%6F
zfDP5wt7AuHWd<9W@zCmbbyB{XLMHoT`?~(1-Cvz_x_{&}h@c|ZN-@4Iktm!R`bcMN
zN=4xT!1Ej4yU&2iH5xR+-k-?z=}s%4R<|kr0SH*#h@>4KS{c?8GrkxetfP2+ym;bv
zfiCFKD55r99;}qm^c;Ymh;9^~y4ED`@u%0%iw#WrVJ(w1qC5p`SXCN!ImH4}SIP0;
zY98SjZpe*ZjrHSMBp0z_cibuHwA7n6>4QQON;Jh5=QZ6OQ#?q_iDlqJ^8kf{hen>C
zd{W9j$Ij>}MJHfGg&}&Nfjuo)|Kct6a8vtuDN;Xs7Tc8OUbc)r*~Lj!iuHiSZ|5Y*
zO#@Xw_0}@euu-O|zNYjSbFc9-V8OtdiQyKVsnkeWJbfCwm!6XdDq}034W@U*oo7i5
zshnsu0%_Vt$+^V@A^up~g@m@U@vy%%*VWUpfg=Q_zRFmsTmm`YZj#OqIL;Q70^|%p
zk`+?`&_8<<UAA#js|=7uUS!iZpo{|0UjP-zDk2|b0;|C8DZ&~w4QQT5ws1wv{Ear=
z7+=AB@{%Gi0IV%VAb=W@wgj13-_hg)y9vRH+Eq9YIdSaxfcKyPn&@_vXDF1mqlcG-
zRDxjN*cl7_TupAgqc#o~1+ZGS0>oPB={EM;2Zb^q#T~@b;*`BxgtgjD(`@lRN-umw
z%Zca_ghEA*QUi>{{p1x7J0IeQ!=QEx<t9YOMW0+_bueYnSduoHEz2;03@J9jm|f_%
zb8^lRfC2wR!kmoQTRTxqfVp`DkDtytq)No1N&+E7vrW*$6j(~Z+9g=R{OGz|`RRPc
zjgSPqp^)KcA=n|KIEjd_RU;Wednik9Sm*>Q5}FYV5USJK{u#5D#r|#_+-hhWx2O0=
zh9HG0LBLM#T6#B)cRUh9L<|?#cU95TLd5b%|5nFQ^f_Wa2??=<iI=a0qbK6{v%@5E
zSuddP1C|E4L3QDFV{`BZb|6g>5jJ=UaBn%QumV{LaISDVK!(3!&^97n_D5f#82gNj
zhXpTGuu*{uFu#Zd46u%H)z?A0JOanCUq}wkV}MFo5#6?J!ounzY{c{kTow%C=OG*-
z#G|ezpb4Xq$q0?`I7;GBW0s&4d7&N5gXMS3O#~Q)er@tNA|#&4c7h+Ni-d6Bh5Pqw
zHAM1&RDcMCu5Mls>>Y>FjL%9a7eF2<!(vV*1me=9Yt@9{ELPt3n`myx2FU7S6!ThX
zo++l4%Ps|#9K85Pa97$7{pPOIdAhT|##Czaw58jG4&zAgPDpQE<@x8@^1Y}#kp^rB
zk9$3keQM~(Bf}Le)IjTWSH5>*FE-sTq#1<d($|xPArKY9=Fz8H3w^@_^q32izTGM+
zXf4#1ym10tt>ZZAF0hy)j0%7X?~4+jCJW;ec}NR`RvPt)>B^|F4UVLtN%FbdYj4PM
zgjBLKKGOz1+mU&`CY(-uoM_VseKo!+XbDmfB)dfWRuMkz3>(qIo0Ey|1A&BL=B&aB
z0f860UY$~j1u&`wMiAO6qPpK+M`NN>;l+og@dxABH4?N%T4Wb;C=u0fygkXcj$>{R
zjglSF|9fA$J=}^CZEw&B68AW0Y_rE**pUW|CH^xZ9k<?#+0pZIt~@eRYbRx+cNd(I
z;`t8?<9oOP{Ax^G-P3q$0G-uiPTMj@8?<3pbR95xfUq+YD13pr?*x;G52Xlvg{|%~
zHizq&g31bhSyc+E<$X?H<9k$NvD4!mRkKEm&^sG_zy7o+-`1^8mXTuUh@U_TD4Yvd
zE1c~K)Nm}p58IR$_j2gQ5C!!IQ_4g*a?lcZFAbzQf_+sFgh*X-OZ+B8{>wVSEycSe
zGfu#+Va|L6TPO7Sh%A;r<3w{3UowonTro0cN)u7UmuTFndM|Ljy;H|ux)5O|HPmKB
zBEl*!w4An*W+@z4V#jH?Qm>>a<WG7k=M&_?H}t{O2{JTGGq<X{rKEa4-HKjMbVb^D
z-arNQtE<bKY9uqjCLrjQCGirnHeax^qFdF%B{P0b4#g2+6s*L(>L<|=dB!nxcl2Af
zFa&@al`fhbkdsuEiQ|zVqb%t#`-VZ$Y`4$yO5(-_&U7^vC}hu~qpyfKUGwX=JzUv?
z(nQv5H*~VxSyIr6A#<v+%v%zru!L!TOMjg}`2RdyFlYTwd*iGe1u-wn=AEawlcfm!
z_zk4U8&Su(Wv7sn<(}f4zkw9VBXUML*qxBUq>ke|lpk+~6IfNyxym|{uM$h>qKYes
zIy57rCZ$>jP1`Qg#stMzu_HDl1`aj{fUiKgHil{HDf)1CymDk=Ubk)sVrq5J_#O>N
zyhxPkw@b_GtHWyasYsecBf480_5$(PfeiU_*7BrO96Ln>JAv`z%JmT&J?B11wRqs+
zO-j4A=1w-6F{P8Jf+a1f3f=H++}uTc4z<y$(oom&;zjx6aM{qvh2=rhI!XW!Qw!r_
zj9WHlDir~_IR{5}tUIHE8Tu_6MNTn-LaN;{X=c@Ot+S=xx%pDFKMYA5LGuefE;S|E
z1v*egP@E?XuBy8B>$_V)ravsw^?ojYRt~}&ftfU-n!Hm!UR6ZTi$;x4;|!*WlKcV_
z!$U)SuU@?(p;KR)IE^Sr##jwSx&P>TW(T}glG3a^-8RlBT7+zf@LymiaWlEG!0!nn
z#g{R5>qH;lLS{ZSwN+GBP_vg3pGe1h?lgEWe3>%^e4O?j)lxY9<_7WrEtgQ+d5i43
z5xV{HMEW_l!a1`#QBS-f%O<Y$MXmO^a}n{U^Yw{<74QKJND>C)%?S#dSeyRF0kxH0
zV{l=)HeO@Or8a$HS+ZdA;WK9m;0I^t=q5wJE$uQ3bshWL6DO5WU7SN6M`v-Su-hr1
zIxMSRN&gQQU~WD;`kGM|56ar{WH7+X8#}Co<|o<D<^rpuqlA_ch54ws<>ya|^FVzK
zhlXvy|3s`GbPNpeV!iv(i3BJlsktARc8?@nN^@H#u@s}MR!BYM7}#<bT0TDj1GRq~
zku*J194Uzi1U{G)|IN$q$O&;siSZu1kv|jzG3mzgrZNNeykr$ctX|zfbIDqGH*Uf1
z^g_DTka0O2+D27q6{!xUv5++KHnKFEy3`~7?h5Fh0Xi2VM^k#d7+QXqnpM(K)nXE=
zT^%nI;elEVI?q;8r&B_PT9wqTR8r%$P?kl@jj}Aj0E~-P3ic;sBE(BN^nk)a3$e?S
zSZn|WwxPB_=Dq{80`fYH<(I%J;amMCd$bkqq$Nl@j37fN1~nNmz)$WZRzV~#d(MCi
z=~>`SV+3;Ah7OpxWrRX=6eUM7))g08CZ3>rDWLBC=l1O*NMvayhikvZ%<u6}3&&#6
z3D|-f#2a+M*tAd@jFg}qnH_jC7CYG3_TJvntweq<Jub_axLg)Ty4!Lbqm;Fff(>aT
zxeDV$d9MIOp_UZr4+_XnsK6zO1VZudplkR8^-xcd3ye+!ou30eJE=mUwMk+Otb<}O
z6KLcRmBRQxZ{Veni24$IM&b?NKKufq@u|^rpfwb7jm6`*M;})XPjfS9)-<{12z(PT
z7+nFhm494a6%n34!4eF6ZrE~*`#8o}e*9&wCwYPFxtP+Um@eI#ftH|JP}vl#hNRXs
zEQXRwybw|`QGbod1aKG<K035`j`oC8s2C|E?S7&dX2PElEf)o9IbzwxwyOYpRWJ1Q
zZ0p1X&my?Q@S|RuOaB0_Y(q;{BuHmu@l%@yN*yspRt2G&NN?V_1?|QqI1v3%qGbPI
zatD(|hO9x6*!e;HSpt8EZv2%OT9`Z0kE2Wn5B$QB)97@UI-Tj_6u|nK8-j&`jRs2*
z?U4k|)YHD07hz+67PUY(l2lro^xh~b1>8r=^~yxQ1|idEAQy)o$OoH8HN>%hC!rb7
za!}@9VoB;Lxul6il5fN_xvc_uY$-_y+h5=C+sO{yBUDqGNHrtSL!fC-wU-W}zs@(E
z{j=$Ks;Ndpu2VW4n`A#q-91v0gCY<KLd|HiRb}M#Vp9Yyx@a<Xp(*l6%#L8hIo_r_
z&OEBwVQd(QTo5RTMepKENF_~c6;zaWUrM<4XO@Y}kzl{&M;%tKSrakcWH|n_4>dPM
zq5fmjXHxV=5SHMy=!1%g(hpuzXHu}iZsn_IaTx9LM2A)h%YaY%tzH#oUb(3$it1d2
zRP=D*h;Op9EJt1_a*k9Qo<Q_SJcxVUx|YD|kO}w^=@oyn9jhH1siXRPc<n<Tl(&U$
z@tFR5l$^zhRDsP^f@e)iP$IA*(ex6xyGY>){0$R~11mh4ks?+9`AX!1BAA``T8dPI
znYgDVztdxz$M`a}vOO|;coi!@if;BDTKOK!t}M}OC&bI2JEmjNLJH6mIrkKvhS|k0
zWjUB)B%x(HuUV-M3T+Hz8=rIi(O23d5hNw_*utIjleruBd=t;QEq)%^03*dDE)6%O
z3na(6E?>jK()UgD2b8^n*{074x^Yu`1N&Q6?G4+Kcf$uf={h#5A|mKunt*P`aeRKj
z_qB(g6IZ$*ae`*ML4F{y+Vq@0Fmqe``tLXkcS{6uFr4AgUUmeUj-9v6Zk7?wNoR7)
zBmpc~*#3Wu2);m~ud`0))nv7jK%~j%na=AB>lsdwjaf??!U5}f4^Ns}XmJk)Xb5eR
zt4$5{A*i;?a8yT>jR3(f;P_UC>rn`y%JAbolkSXM16B<0m7|#j8}TJWuI{ez?!jaA
zJc6&AYr5$iJb3VV;+ZG4%y4<&CEX_)n4c|Bg=YbZ*|OU7pZn)Gwj+9wA2ysW<&%>@
z&Wic!a)gj<r*o~l84lL%h6PFX*e|frYf_BYmK9b#cnhlcMt%EvE3l`J0cW_7?hHGV
z6VSRTM|<v<sPiscaD6Ko&CnVLCvFH;sW;NF{obax?^{BOvLE@mX%psz=ml<OS>Yr(
z+>l&&@d`@vI7OVWLNO;9uHUE4#R7I^%lm1H5BM7XgzHRWuNW;Q$I66IwfQT@K5$6T
z&8$vjUyJcQaP>R3qqc}`iRHZO4*;$d;h6|!vlcA`Vxb|jgTP2MHcQ{J26&@19X5T+
za9zI1ZIjF6291L|(BpDlv+LByq^tz|C9#eJba|~<B^|BLYrgTpM>2f7Myx3`P&ChL
z!iMs-HCQVdNA4@jUgy16j8h|a>L3zDp;vo%$DuV7`*7>&(|P?2*9)cb_WmvRmkY=R
zl8+;a^f=k%8P2TJojAV-h5I<i!7Q7>&#J^`Y^X)*QLx8Tq}c%5E<kS)!}WWoKEU9Q
z{&_C<Ka*+x`)~h8Pvh$otwnzz_5~<+w~U7@&7MJ|-7l37kPeW2(bGme<;R66?t~Hs
zm6uuSVH73c_&}}qc;yiU%)n_-G&zoj2=Fe)yes<WdJOHZ|INQS?7SOh&)ErjRi@Ka
zB3BZO`-(<e9eUUalQPu-Wf;N$Zf{e3?xSfOIVgh*ho*mKNf!R`#@VQ1grN?Px?zRV
zU@<m1>4$Z8{;&nUn8V`#{k8mWulc_}_rHI?|9w{fPtR)lye<5R8Ok`JH6(C?dkAUD
zP#S=dB7lvILz_SYMMUJUg>iNJ<?cxCyIu{x_=VlKaMaMjEEo}w{vR=Igd%p~FOm)2
z;Xru(5K1L4q+CV{TF8aU9$lebv1lQ=VJ?Srju$@MDJKZ*8PbGF=rdvGhu_2j%y4}f
z$<x0JqGSK%JIx~F*gp4Ly2sJPAbp*DY44{PZ}~A8YTJKWNgLhLlgn)<(%0ZxOpasF
zm6@BdB8x!{#L)HV#vWy;|Mv$~0lgw~DjAw~baGNcGpRX5fWG-5ee*$9`?HVF5||1i
z?8Qp_NxP~%eb#2i5U1h&@QJrF3keBo!k@IO&ISI>jNor^A=R?gK^Mw}>f-cnT-P_z
zE#$rYNN#xWGELt`X<?(|g3It&Js;k?3?`qlY|OyJwpP+#u7cR2ApY+k`)rkGB9c$m
z82-`+W>X&+MR0%i5_9WV;V--K98-9PV8&bav%U0jTywwI_c6&x2^oM&*60~~Jp6+{
z=k8YFFLaAdK-CPKs)Wx1jcj}{_i?}e{p+al@z*p(8cnq9S92e?>EDk7GPD($qEb!X
zzgKtQ-#_j`UW;krL8UWLND^o+f&2#t>auw>v2CA%l1e-}2JSen|1w2%{$@Y-5ZV|2
zXC-8&sG#<cRNwf2`<-tXF%q+06FmX4AK-|mxk#P2=9a+|nuTN82pzG*%N=yD>*ucG
z;)(w(gJ&sNe500g(RS0^3TZAX+Wnr2Ul=<r{)AB8{{7(GQh0#1#EB+AmPvKe7d#(^
zl)2|=8IE{OIPupp0yUmVyDgu)fN+n0FQ9MYBhAf4ZM1EH8QoCt7;Hzr%X}bX0p!3A
zoMyO{T|@9JAyT$mB{8`HU<+o*;XAYnfc!=xr@vTPUq7^eHWY^pnaV~bVS((9BpV#m
z;ey#PL}v*2g1#_EvHzC;Oh(%+GIy`|oB#gX|F_Sett^qA@(WrCXw6vHNI=UCJeJP`
zg{acu!#B~DlU^fR{4@~z@ISjgLJRBI2=00&jr+i5Le$&}|C3PCJOBdTMHMC9^}hO5
z_+$?;SD2$;SMC7)j0pHJGE)HSU@O{4-l)n;5kDeP<_G9W;)fqEp`G-|2o)3oI7X;y
zlMD(L|Gi>G<sc6t+v|mt_g>!xoWPM3>ex0)=QSyOYq9<Nn?{u(dO#_+E?B(0XfPIk
zv3Z%H7tCeb_n@6Zy|&UnKyCh~5=3pdv7!e!n=h`H-0HYuF<X$qdn*MmiLVIhqxpY+
zS6(E5inJ@J-FBb{m__Ou6Mga<pt2TdOpV?P^ckoZOT{yS`VPVZAoT-FQ#)(SRu<qN
zK@4c(ZpU;Xs=Lv|w$_;i8cgaH1teN@(@5t4!|7dY&e#G0;v+O;GZNy}0f|zt1V%D+
z;e;IElOb#Nhq6E2AG>MuVcO>BZsNZQHzS`bLRB5`R*c|qN1V(|6<h+jq7X?bQ#S-b
zis&r_Kavw02Kdw&WTB)hrjF&Cn~R8USuhRjRYg<_!l8d4vFQ5q%C1dSRhMz&>rIb#
zXdGnTILVRC{O%^F->Kt<D?ZFNzIwOHP}px<-8)w~!x2`Iu(KoE@3M1-aoegMSB+5e
zoSrTE%k6e>L&M^DseG%iY>zW7Y%gGrKqlq@t$KZ}XV00}40s5yTUzwC=sG%zVV3PH
zD6#=(FSxn6h<^bCJU}=AGRR@|slvGW^XJd*u4@<0Lx;?MwCis`7R;=ycL896aC5-4
z`)>x&m1IHZ`3S>7*gbB*0MN|0po_F))remHvJgX9JU~{A30?sLvnTf<6<s|PH(2lJ
z7i>&dv44W@@j2(skQdl$12tQTHk*3g9s-ZWkHRn00MgjEFsG<ZECpLb1JK~@_^T+G
zxC81yU0cX-6S}BxzoxCUrbESS6koo2<>{M`bK@oaCAPpxerT+080*h8_YeBLE7LBu
z@khXy)K>*d;3Yb{Z%Rsd+Flr7CD7&Qv|faqwI()?)786dH8Zp7jhB%5mIdusGs-|I
zTB=z2=FPmnnAl!JuyO||Vzgce?37jk{do)E=+ncvGBk`YgS79%ucv*|9>XZ$Mze+%
z-|`Cv1~&lyLuSAJ`Sa&l+_Ae*R?8nhz7o510S2EgUUW)Z`z%n;4C}sCz`&wH=cA=u
zb>TLnE_5F1XF#Uk1lIDS<zuk)+{r%QUk`mBO9(13F;g{ebi>AthFDJUZ_GQ{M*~Jq
zWnS5S|3Ph>V`g@CobI73fXI#O;sf@+S8GC3f^O?`W233>F~CEYfs<ijo|NCdm6J29
za_z<pcX#*b*9u3Dyg}I5Ua*E=cLVGz!wQccJ^BhgY1*QK!Po?7S$E)4?J;uU+-Kwk
z;dsqn1`*`6(3yeu(wd<FBLjnZka`vW7-`EdL<jgg`s3|@2kpmunc>PO!mk1Zo8Fag
zM~IX&ttzml<gf4pR=xYCt!)#q@wyCUJOLT=cC;kI<`dS(Q$VbxacF-Fwr%u*lkjd#
z%KO~Z6b?Am(#q<VxA$sCFs(E1egQJ28lk}KYvkPt4M_%o&5~AAhW4(me}Gf(l>FM!
zv8C||WunF{St4v~YydMBz=38toKp^>H(+35atH7nEy2$Kb4nKHDX&|%jzFQVNAIA7
zeuiX%TQiXjV{%lJ&mYam$T)*eZ`fsgts6CCcoxu-Kj9eApwzx??B6xk)bq0Miq~{m
zIiPyjefTL)0Tw&XnhtNQx9Ba)c7j{J6g~3w<=KGYy;i4IV)>a!Xn2=hz?6|yaHA@X
zx0jQXJB*g*Kj=0W+at!S1&)9`B-=?AZj)vgO-*?SN_{NhkaR8soKqz~D`<&sVh2}e
z;8$+~eKc-HGh~^qt?jhSQE<hPTV8A)&%W&CfT#R(7t9tni-~CqD&sx)3}a4=5E^D%
zSTA9TWk7=5`yF8CSzz&Dh4K$r7h#tcC*46NCk?`#C010t^Ucj?!2Kg0&9+YR(d~0F
zGc&6aJyCcCrn1rp4>EwAw3+21USU{aYyLG4@ILM`$*);B4|`<2<i!gY+=vp%FNEi(
zYWts8--8|FRrvm^8T^S8&gB3Pi!Pf1FFmaA?mE7^=!V29OS53H!8ngcG7Ug>aRyu*
z6{nOK`GQ2u)}gUrv-d{!y?e{Zi^uUQ;CL@-iFxf<xEb)<>Aj33{z0icmRjh(i8|{Y
zcn~EE{QUg93TvvXeTJvLeOvhek9s4z{fj_I^Hk2Z|Nh0RP$$Ls1|iJ5_e);=u@Xe*
zHIf`GEO&reR!yg1LW|FEzRiH2aUjCtsWfvIz>Kaq1>rH%37SWS(-c2ss`v2p={k}s
zp8;_9Dl{}S)Nq|Q`=t5weK)ti?%%(^jVqE1*K@-4-)PV-Dk`eLfwM(3D<3vjVVBXL
zbOrm>Sc9Jd>oKVDGRmGP!7}ioWH)B)!lZ%W6--PY&}K)TxTrYA!omW0@GV{MF$|Sj
z4w&pcT1#L8tt}P?VPLz%Rm7pkFq1vEK?8Si96Jd8c4J>AybWg;Q5jFfhox({xVU^7
zbt^*;RHjyb_^=o=>wfZ_BxoX*y$5L~PXW$==-|Z7OGs~B`_RAm{rg4OGw-q6E<^Q$
z&EoD_+m$2YsX4t8kuK^#k?v^`%U5hNKeNzN0$h5@>FJu{4?j2SG&ukIq3ZqR4C3<-
zpyH1(EKT*7G7$Cm=t0Q18L#WA_s3<q|NDTR;(&`@ZiU1~cLsgu(R0H)3=JpZRIadw
z;&>6>$;5C50IKolms_Y`C3YZy4LNlIVpIJ1<JdcYY-GLv{WL2}T7WlHVGKzdO0W#)
z;z|yV5VLDIIMK7G;1c;A|H9117H`86xx=<n1q~1ZC^Qy1IXR`6)-FZRlSd+dJlE02
zcXiK&dI-1xJbz^sWPyBN@R~*(cM0?Q^-(sgSa{>p1`F2X8`mzu#{8UV#YL~az@i~Z
zcKkD-sVXo~BMZ9{M=$i^rd}d5!^y^F7V7JZvn~2MjNzWa#S(Uy2h7dQeLeul-i=7)
zALrkUU*0~3LUktyE>C~|jpxm5a^3kgkT32wUw%DSECE?~^ldAEnW+IDMHpyq2lyS!
z@XA2?sLi;~Amik=)*1rUC4c|@2E0u)H}9d<so<(v*ExXG@DM1v#z9zhW*Tz5yu6Pe
ze{0~SFZp-D`dl)%u=ozY-Yue{8{p&3Q&0(V!xjE$Xh;=4kW;;IU?=P*<wVh-?){}c
z{l2Vhd0JYUX(9GlSXabkiGDbs^5VrZ5cofEEP~y;0sJcLfBks%X76U4ld7|whJx52
z)f*9JZp*E}C18s@g_6qB&Q2b!-7^*ze7p(RB4;3@@G_W%@EUz+_lizH&%Ahwe_>^6
z?2q!v&6Nc4_YQ*3o0b-KhL?Z`wT#Mn5l1qv{gFZk{~o5YECZ=k7LA+|d-UgIpB=s}
z<B@-L`b6I>9s~<g)stfPKNS@%!R{+W<F_4};iTobYw%My(ZT>t-nRpf-aX>wjeGdN
z@#i?~L_vkB>q8&&89qGeBrE1|@u0oFObEXHg`m~m1q^No#}Y~pLJnMexMGvk>(*8d
zgbP_T5<9+rbp<bU!@EbBpp_LA7XF1~RzXFj5OLjdX3_$m+krL&YG#Z%S%rXb9C}yc
zLtYUzK!36K+%M$G@9_GnKzWKRM0}uw9`%3c>l+J}Ffz(2D=#lAD{JrVbw^egjyHvK
zR?X~-6P`Jz#!{Dr2e=O@GkCi6f?-?2p&0<I`vLHpxkF-gosN!9I9SKMU>)+`3mj2a
zR&MX_UkYdV2Q{C8DI~64le`fCC<60cM524HF>vas8w<t?Vvf1>pG}yG5{2C1tc!vk
zV?7%}$*Q2FRHIfbbrg&B1E_Y}3($6WimvODrAte~Pt5)HhsFH3L>^1{H!nwk2a93}
zRxxYjD~Q(616@PThDQPx9y*Bn{8%-QJ2Jr=;81Y(3yPu?-^_4C=8!rdE2Gb7XTb>A
z0KiU?>w^+zcJ}hcM`*=u_%~fTxManOLZnMoAXQKnjGVAqtQ^NAskM4*qE;Zyu6_7t
za!3d3?hMFd+#0cKT)+fQD2-4;*K7$oW^5;-6)i1y!wF&`9!;uaN#yhAXQA)_OIHxI
zl$*#HQb)^+ie97JcMollFl64Awzen58F3+D%0ahiH3?5587AVspt77CdE|SBggk_o
zHG04uI6&>EhS_07=!y;A3E+ToH~o^w_z|qzv*0ChYo~A{ov8veBPuRlfF*`~i03SQ
z_ADoY&bJ@m4WUe22@Nf-_}1nnJY)0`T>~pwSYiYp;C}B;u5=pv`k;3fp{L;U=M9+2
zNY@XWkbkgr3O;<0GZ5OlR}LSC#O+&G@oKRCL4?zP2RvTg?`R1QsokJMgJZ@F+`=_G
zTO6g*Vgyqpcm>VPY<PRLjL)6h#vx&U9d>$D$(_8k2akYr|NVvgQ};6BJyBlzQy~z~
zWA2~j;BQtQM^iaCs&va;N+;=`b00sy2IuDAe-=NOJ38k6<G+6~_Za^BNa!*A|FamB
zM|lk~U;tFsz4=LVxt=K!=nSi#hrQo{H9o}&?Qn|LIA;M)?;ALkyOmm}8()in2KTEZ
z1gl&j22wg?WbyO^%-tA<_xKW{E?XG9=l}iZzps;?!~fyyjChBd%;4NPpH!SZ0{Qo3
z9C}fyE*-h{H4Tu&FMd|fLpX2AWW%aO8muE})LAa;J={mHr>5G}<sQJf@{6CTdyk4r
zrEc52ii!#aMa7blFlLs2U_RZI<rW_wkJskeCd)e)lzhq&p!b0Dt~(PP9J~jU3a@fA
zJ0~KXUj_uM94xk$@=By^l$<|)@}!`+_-{~_-k~H$Z9hDJ>eicg?-l|&t4uwBoC_#$
zx!WpdDf)fr+rY411W}GGFsCzSX3Lb6lu-Y#Lxbckuz{5;R~~Oq+{pP;nDQc!sEbqo
zfF<Y_8EGU18R|9AzbN&@$lpzLpU9_@`%QtpQ$hODVeC9+Gn`DYh7}tV%%E>O0}!M-
zl>gqJKIU=EAl6{moQ169*lDfOuP|6VzT?1w{9pX5)~-Z7rk1hy*YDqB*$O`LpeTor
z3z>o4!_#StF_EKImF49}PqF}hlC6n`uUuM`!XnrzyV6&_!4Tz9$C5lP9i3&UWmjXD
zq*VL?RnMy|4!*=m)kjoR)|WtB!Rvy8_8-7{38gVAI1T-;GG;+}-X>eNed|`~&oG>%
z?oUJgQGhG|Kr*~!lhoCskuNn|MZk#$v^R>|n!Bub8&$wmhw92%)H+LXdtR5N4<A-{
zxyZi%ip<IVpxw#ToXMT7@%HlK1Y>LI?Ck7YkZJxwHnbE2mu_O5QLoK$&?FN^f$fIB
z1ol=ek#q@_=3BB%ud}-dFwN(=4Fo?4C5e|H;5aeuDB(1zEZ~s**ZRKq=)bJVQX9P^
z$#N6NI{BbwP9neR#DMeAx;{|6Sg-(U#n^vZ!-kj@-$;2Uj{F_s`iBqCM;;NDGWARh
zgP<(o!IDs21Va552*`HS&msprgRYQR@vYg_?CcK!(@-V}5rrrF2<T&YjZle3-vH;X
z_r-PA&)^vzmi&UQ#I#<7guk;rIdhL?i^?R$>!dRwC96oi1|Jb_5CiM_RIuW1ZZ-~P
za@GbCbqfe@7}n%tL8GOpNUlRT5Yu9&n24e|<zfpf&_4fQN>q`ExdLM8z}8Fpa(~kS
z{XQ3?+^)U@1PJ`R2K+&-F?u)P%+C!CecL?XHuwebf!L9IQhZ434q_!-Mg@4a_yF_+
zJuR((d6#e8_(XT+8cZqBVIL5FQCW+0qkVYzA#Cg94f0WJV{pM8+~DRNI~F6;u>aOR
z4+5fFDAjqFI}vCFUFlr_>OD5!@AR(r3Jkmp29-#?i<+9+eW=B_9LPF{*3q$Q&T1Ll
z0?~KL`6v7b)Cq?{hxrZ&!PfxSzHG$`58WA+AW_c&r_k8~%2OFI9VE=V2p50q!Zgf`
zSK!BLv>eAxDC0p9@D0Qn+|Ksxi`uS^F-l$i@wCR;!rc5LZbEkkUMTjPdJp0q({}=3
zxG-d8YwMbye*if6J)q~|$V9nML@e3G$EW)0tSlx_GJ=5r8V^?$D=yw1-i+vYs^#h3
zk~#OocDOmldj244yw?p4S{J6F{AS182QeiN1$Oid3INQdD0ZS^*dTQ^BmU*td}*@!
zNGSOhbCycj>u~*!ls*HV<UIg0{lXZTdj9g23nz9%W{mvt_;|*XCmRVF##-GXCB=b#
z_8QUZJNUnba8`bqo?g%M2g>|;cxu%=jC;^%IG%+OJ|4V0rY7hcq8f9l`-3nl>NL5F
zMAw2B(WbMU9MnScXEcOCMbbdxJpKHh`;V@win2hIQk=Tyyu!F2FbjAf2do2kc(T#Z
zj-c}fkOfeUbLEO4)Oi2Uw<35e7NE>s57<mC-~ukfXVP`Wk|piZJGWELh{|zweXf&T
zaL>eOw^EhjW@i_e;CMh>$1QpO?Ae=;>~Zeiy%OcE3}Q5jPezRQ1>(^T^5h4krp!Qy
zpaOo1Ky2D`4aN_;_2HNuE!PvKjEAclCUZx3W|O32gtaf9R}U&W%qy{~`}5($hi_9;
z@+eu2q6`cTZ-IXo25z&~4Cv^t;OnYI8Ep?!X+v}cW6;MeKQ&bZRH&%2UoIdoA^-1_
z&=XV!P6K0JF(`8G<53u*ZJS})F&|!5j3}CeyM-Y7tBR2Wz3(Q@lc(5yFJHbS-s2wV
zbuf!pQ&`eS+R-X0B}E3$#%ib%<p{K$UF&Cs-Yn*4cliEg(fNv?I|PfPXr|wVbmJpT
zd59{_r4oZs*$(Ugi{fon6*E9FDh@!p^#ESn+rDeL4zNoqvr&OR17FwD((-yt%ub+C
zNv=kuF{`t`u=A(=i_R}!UK*9IgToL5W|A)gbbIFfdB4=0K^%ZMz@AP-1m4`-j?PkV
zZQOd0Wto5r5JC*N2f)@UG>B3aXRw=~WspUO`8b>=af*iHD38{NTZc)lcMxVxHC&tl
zs4$Wv%=qJ=kd=X_Fqyn{M_w6R%biZ;*zbWmuNmNQM%?OdA)#Br!R$C?Fz9auozP@@
zQ2H)`mNr6Z6<YCh%HsKR>*t749D)3)9dVU4at2k^kLhVD{i9XG7elYWgt{M{tc6%e
za-a{q1XVlIyJg)rFlyIHBIDrH()(OqUcLtg(Zwk~4l|Q{@Zozj;2a9!xptqz9znsw
zDCT@Qn$V}J7@cn31yW<j9~#Ft{=uuXs%lUDw?O~`*SxG2?byEE0HKl)aJqQRY)9c2
z4mRK^=y~cUZmGOMJl<ngw+rgnXHXBkOiH>L5WqqgiG=k&yZAa>c5Qz)=awz=Q2)sx
z#&vx8G9L?`3dn+j1t1H{pu^c{@;oZ46O8W-Lg$w^H#aBc^$ZRTsY2)gtMg?z4X^=|
zK6F=Ly47u9dT-0i8Nnn6a<K^dNdlYBU~6bI20LMiK{9w3)ho%BUZ$qrj*Q$+KLdjK
zS-^Vk4<FY3WE(#;c<3l%ruEN{+tjSBg#kD}9kYHvs-5p53H_cscE*L=5iCG%wdc#>
z{Ob--!vN3Bao`z-MzP~WTuk*RUTJ(n0>w*8c&Mu9E*~GaMiX`=G%APjKgt-;OL>dh
z6xS;Qs^K2!IcYrP>uX$~|E__o@eMQ$yD?^u;0GjL8ur)26F~U~6u>Qc66bH#hgb+Y
z-TgX&egI)p4X{Ht<U8D!X0dU-UeOT&Goht7ftc`Q?&?CLYTP%~95j%nMliPVUVaV2
zu1KXaehlZq8uXTE!=C*cZaXKmb{ILJ+4VPGpJwK#=P*@`mslsDAHPxn<^^>F6Lb*0
z(8TIGC!i9N$%sSe<m?Gzo|E71Nf)%K-hvf>8`oJiKKdnh-NHjuj_k%j_&1m@b+Q46
z6E7g5IAuFzX_>8^Fol1H9<x{9lgE#j(8ykZwEQEt|6D^3$qY1rYk*BcLgF8k?5WlD
z>3Fz2d}io{e^QJBo7+@x24}Zs;4utjG6wk7U&OBh%KJyS=)3{+VjDmLL!w)s{0T>j
zwMQ?YJ9%^SFFOm&HI)-o0Y1=yrIl34i?hBLGX*HY(&{+}n1NXR_S2)0gA$`9IARt7
zT(6puM7sB4nhcRN^C;cOS-er$_^GWF=~b#zKp732>^Z7WAJipM3()r}UmFhDHN$53
z?F7FjD51lTXIY!vwrYlK!T1LOpW^{3=&Ie806p}|?N>=jCkCw6T&Ey)TR5p+R<Lya
zh7Hl9$whD89HRUn-s)9ya_QZDwu7e{UcAAsPx`W7Ozf*zNb>9!JEz%SnS+AdltBV3
zE}7L$fG)uDTl=A6;pBY(fEQKR$hRT6QhGi#IJk>OK=Dk0QvVN_w?_crM8{td7Z+cK
zG{(m_Eoqd-l~WdWjsq=|H5gpZ&+DNK%Mu_<s~2Ls`}Uv+;3bED)5y^E?}lGs!a(V<
zt?;=_7-+@ZV;ZUw=!b5DaZZ_IE)SO0Dg?-wQOU|YT=G&lC$9vP!Ki!7<i@AKpl@zo
zau_@L8(R7*MtS7m7Tkkk<~9JUr2vDHD&s)c(y|ek-a46k0&gj`dK!&n+0&a4Hj6Vr
zi(2&nq8D{v2RNbJtLv%#om0Gwl3<6UCTf>F-*V_`2O<%L;5>fRq#T#p(cOI;;Be{~
z0pBF?5Nt-O-QL$X-5_}*2WK0cBocNPC;FtM2$P!8bjtc;p;Jq$nI=h?CT#%S--jwL
zCf8I>SC<Wa=NLhKp!=AI=|NU@liq;{!fzotGD){K!i)aP&d!c7C&6aGeQW*P{l%HX
zcaY7Ub$*hDlCY6^&`jqfdU}-4*!aoYVy9Z#*l1+;oIZ7G1NOHYSQRF@B`u)sf5e#}
zacx#tQ1@$B9qc7}YVg656pTp5kSAADQ}c>YPe`WQPI9lY8tpn#r5_5YdTns}q4g+{
z7Jzwm+ZhjGy`K4@jVK8<9r@8|^`FS|4hzcVYr=Gcs&Ai^kvMEXJbIg_5_7>GJ!gp9
z)LqtLSu+794sB_cy1Dr&_TV>!li!)F>(;sA+BhX8^=zepVasEueebX3Ew;qydE=Vr
zjL3_k1k1704UifqIf-EG+A~;Iu^wIh(s9C+##E^_EZZe4yarPUGhn40Z;b{axP3Pe
z6ZGm1Z-N~A@ot0pkWa>s&gA0AU1}49*r8=~q8_!i?8pl!DtfoAFq())dy|^)Tzx&(
z$n2ooT~E3;ZOK2adCvEcQNBSrY4A|yD3bYtnwr&M#Hg}NLKiKEE|G@oz5DlNAa&Y}
zEf&e^cFzI1Jeuvd;5+S{HYSX>pfP?5r1R~aitPDt`6|g_XJ@B2Ze-y*m%^)wpI=Ab
z2H4MQ=6*dreIKR@x+7iODk8ELA6@|NNqc+yqqS1owk<?Cpvmijtm`Z?B8KZ&f?GCk
zW`JB<4jCmTW_X~G<W}Zq18agr@a==k0S+vt+dsbL2ZVx`VIYWAz($FJ+loyD@8A_1
zAS*<okkL5ZJCmEJ9y3=AAyjq3>PJCYxhycq5-07_f@2V>e8=pm&2S5_{R!ihH>hh3
z%+2qEXGS?FL_Me;3K4E8?>c;YsnPAWSB7r!TNDv7J|Ms+^Snc-r_F;r=Pp|KpiQiW
zrH`oNc=X%j`6vpXqCQ*n{K;?VYJ7+7Y;0<^IEYZ6Du6=d!0&PIiqWDJa4B$P%EcM1
z!kJ6Che+Z()IF^DLGRz4T=+=rY%jJi+32W{-(=^O9cV{?8z1*++3;jN8Jjr9qnqdc
z;K6%5+mD!U3)$dOBuhBoWx=4J6cYL0aug}1Jx+gSov9o&2t^`-Q{NRSa$Il-4|PNE
zvpyU()&WD6S{1cVFug7Hm{5<$M-`*#?}!rP!z~+8*MTPT1}W-I2;4Yf(={-&v!{6n
z0{jOg9-}eu94_H(6NM2<-5=O?yoOo}3H3jK-!ZnZ6OT8;W?&s&^CC19Qa85Yy}9Fj
z$A)3YQ~rRW_1j>B5HVn|AwMTy@ObNy4#?sDhCPREe=;#Kp<)sxc_K=*t$X*bLsf+c
z`wp+`BhsgM!DTnEV^*=^?5T3fGAUWQ{-@s@G_gWp*x;A^m1F;XKHeRUd6A^Ih40@#
z1D}W|iW7hm`LGa=CEghg>Kmb<n~>)+Z`$Mmt_F30Fs)cEx&Q}&{!rW6pPYL`&wgF0
zs%H%DVfCgTRBC2z*%^gO_%FN&S!kC&m^5eEEW=D<o{e?yum1YMj9{)du;n<~_vFUl
zD1jmwmJBNsrb=F6=U;_DkF{{$xrULNAi(|GSTy^k9x3h<f7P}<UcUROzl*`!o;@<m
zf7((nW7~yY#zDCzOI8dgm5GUoY4^N?kM~*x8V5nT{tS8D(hh7l0ak>&HA0duzfNr#
z1ZZ%s`&lA+b#w_iO`7{&KB1;|8)eiI%8D7i*N>#)=t0?cd#n7z=v`I-hjwGE((O@n
zill|FR)F=m5NUM>m=a-^;kTRcaZyuQnKJqqs~N9h6w#aNYC+zscUEox2g&G>kV0gL
zYqRQ^U}7q4UNK_@u7a0|Sw9NtuuwI?TPJz>Vz%7HhFu3$&LgvYY&1)_`P{ZnVF?3L
zpqlEsEw&M)1KBYz*0AWl9kvGu2blcGqM-wj<u)|GF%zx*O6cDC`c>gNzzzsFV`vyY
zdKC%hZA7<Xj(-2ZKpqHucWeG%<$Y;5)ouH=l9Gr9DV0)YNrSl%LMXE>G8P${NoB4?
z#*`sLBEu$ZgeEd8Wr)ZScZE`7E8{k0c-PhY<vot~)Bp4T%X2(+XWM)KuIn7uTIV_y
z@TnWvQZX%x^a>)^Al?dLSz22!iEWna4T>xAD6H+=Z=Nv&>fNH6PZ`*N6iw;2$Z4`-
zTlKhGz;p2jyOYsxR^gZ*{@yotFxxmY;x1mF*UDOLeIdQO+?R6P?RG<UDyEUlW-vr(
z7#JLExQVzwLI2{{*MFzLR)y?zJp3IB3pOBrhOAgTaIV&c#Ax)P{&yoHQVJS~4gJQ|
z>CQ*)6^x;AHpjU%<Xu4Ix`_%@bsCMG*3Z4TTmB;$OpiDigNP&ADhNtYV3-3b-+$sw
zz_=3~^N1$1W-yv`eb#c1H{}<KU$O&h(xD(-ZM}8LwR?28kEgNcIk(Sy$JO|?MGvyG
zvmLFFaTMwuz|lFEzL?Q6Ra;LD;L7Hm^uTke?HC8rAvSIe7mt>gAhkS_T2w8!q&QZP
z?Y2b|zcD|RJQ8&f^KQR;H$td8jX@~&@EAlPB*EK3L0au)36V`cSp|}3E;u^}<Bn#9
zp2Giubmb(ZJ<dO<;Zq)XC<T=JXuqleo^lI0TR|&izOD;fCiDi_w#1R@;)6YwKbsV+
zT7=vnLHb@g+FNXWrWJVpftH1k<o5IBU;zMCzf)u>odzPJ9-djq8z)9UiKM5e?d)ve
zaHLR+q{>W0R$B{f&Y!;nc@q(o5YZN*2q`;>z#@GH23?K}yH>PegtP`Y1k1C0bd$BU
zr#ZLC-93}EofXOW=C2Q%P(trAf659JR~0=tMw*E7J_QY*cD1titg&SEh!nq!g#aQs
zarcrwDusRzwcjBKxb9-F(Y`sve;1B5=i9Ql$#{O=2VP?C)iMtusK@&$qlsC@8UMvM
z9&a<N8yj_#g<bG*5+!aWUqEs4>D5(9^q!Kur}4?y6nnOT*+X^mQ+;&Ia<aIU*n==G
z=b1kfLl%Dg;qM7<l!`QLW|F#-lXEV2V5f2E3cQmFBs1I7r_=W>l4cPfz-%!IJZSr#
zcSiwqSE2fl=&El+n?}s=s^(wfOGYSFb(%&;;?Mno@cz)I=p=?fKZdR6oB&ms7C9;<
zS-Txhx`|r0r(R$5Sx2_XbS$7ed)&5DuMRn6$4qAjNOR1HvJh%AB%~TABkI1W5tv-e
z+@)P|2i~?9&bBl@KQaN_Fbw>ssB9;b4cPn$MQ`|PUPl1uM<yTz02$<*dF23cEiYrc
ztTEYdtri%95w=a%ZI|R%?G9>7CA{gc=AF@_$~<R#T5gt>meRJR%im_tI&e@!_ml|&
z3ir<@qAViDoD`)1k-yppkl{Vds4VYZO_@gn_HmfsiYf!yuP(SBy*flw(Y=6tQuH@<
zqBm)osl=3aKKd-$BHBYDt*R5N2Q*mtH3LO7*@;9s=)LBIZ4)>KU~u5M5i<ZFxaH0x
z9=6wBj0H6H4GntATZFkV)D5Z76`5sU#$WNEmRCiPjjEYJ6IH|UlW2JGmQWxrA+ikx
z1#+7rbasf-)4k!=*X&taSH}j&w6{QW49dLRW8>pfP&C5kvJ#K0x34cv>MKeHDf3L7
z4k=3&!#LF4sra=K9NqY-rnV1*P%WV89&js@>?z-NyRHOUjqKj1Nt;!;Tr6dwv{BGH
z;J`9x<kU0x<<*P26?>RStM2da1Ia~3(@E})lIr8oEB{|DK&SGmoSYoS{U$*AQH`ll
z4>K|{RGA;-(hXq5t1qYNM!A}}_tbI-a0IH4A3rW~FpBhwj^AsrFQ`B%4-!iICXN$W
zG6<>OF8U&hiT=N0&cN6WimiuKB4u6&*kkeclz9e<Q3>~MGS>a6%=-W+z-6FmpvA4I
zuV01i69Cz(V+qO?XX8W^btI_~d>;3Q8ORHA1{Wr$*W|odQb#v5bjw3l_>S;`daJ=)
zJs&6-vKh0_+^8n?5{^?3h?oVfO(G^n)sJU6vIbu!WIDVaG`K!J9M`mF^UJ3&K;Xb{
zkM2bio<8za^WmxwKxc>uj$5XGi!!mZdl;T927{#bZV)0K59}l*zn+nY(AU=4Ii6Ae
zdGCJ!tix0}mG|PdLj`i7eG!f(+Tt<5$c$r8Xg$oIMq$L}8vG+cN9ZGPIeu2vb|Hsk
zmScg6W)c)yY+dk6A#`MUPx)1NxC>Dp2l3HcsH=z<+Qe~y2m+o#&Mww*v9UJbq_p#1
zArywxc!EO7_3rEssC7c8f+d5JQGDXq_~e+^h~VgzbV?ZRwf*(y5pK9pP*FUFfW}S3
z&(Br*dJdZAf|fM#lN>bw<+Z!78t#Tlqr)vVTO%l8fPkq2U*^)!u0D(bWKln;`cBXL
zLXG>dJC(%%Aq(|jYrLFQE)J%c*Hj1+g5forbh<}WQIY&V3FqRE_EacpQ1VHd*`m!j
z<L{|OQ-U3KGp!s;La8H4k0VMoeE?ix955;X!@-otIs>znTSvOD#(i?|R~mD@dewAe
z^W6a?8uP`~P+H&m#ms|U<jpfqM5v4H#my_!K_yh+F1GG&!qbn|ZMksnTo|HbU`_qZ
z!&x+S&1XuHxg*?$kUzv5rV=b+`Quok7na30mgVDMzc`;Bns7VUaLULhJW%uqS|S#C
zJ?0yWyd02R;=w9J`pLr1=*lM)qoSV=x=`%)T!z@qP_F+%+01gZ4JLHy`#Z^5fLc0j
z&#l8lL1wyx7MftC*cQcK-6@dO#xr|F<_tcrc(mu};Ydj@%e9-A3T3Dmt;3Tv>Z)*s
zEO8F2cg(30GPU9(-KG3gWU`cM4XcuC$&-Jv0bEdFA|Im66tz83U-ldJgVN!wb?GZf
z{E_~Er;q*jUKtr84jLWlRZB}x-ytT(MkXhQ26sPCX$9Tsk`pThcL_`ZndG>f(H=|`
zV=WzE2c&49dq%VIgR#6z8E0JeBlfs^_ts!@;;sW;lTD#$Oo>i;+Z!?c0=?*qIA=tL
zJ;`?6r8Y-%iT*G5Y`TYW;{={D01?5zKET2CAM=Q2YnCGR*d77Y7-mO(DB6GVzd4$Y
z76-M%)Y2I)EZXI28Asz%+9$TkYqhN+;Bu{%90&wM$Aj;+B7?rH;NTVTIO_b)t?@4s
z#Qv1hIha*p&kKM2eVi+lp^qj$q32+@gl{;Um1}PjvM+Nmb9cWAH{D?D$qk)VkTgeb
zmHEq(*~n(&;(1}oqN1f3cEYqTcG2c+MIiCL_biKJr2FGfG!d0xlu-cGU=Sl1$lRIv
z_2%8fZpOyz(fIhIk9~w478hMyRDu0*-@nrMxj-^cVoFE-Azh^3Yh_(-Dq_^>`G<*#
zs}VS`7TN9`8ne1C*qatUZ*T>7c#1MN$G0@~EUOTc*|@voAg06(;unCMJZ_70&CHNR
zu@(~wxNT=qRY<X3fbsmJ^?!GmQkL8`zwr)q?|^qik=A>od)?#Qc3));Pd6oofr;aW
z5R&i(0H^Qps|6fc{o5O>T65iVVAhwihqiHg*j2?jQPDDT*7ngrU2GC+s8kD%gks*o
z`27!uIE~q(3kwU)Iow>zn3bi$08oA<6FEOPrDV8|VY`<IQu-|bYmU<Fy50W(ihtt4
z0QMUYD&Y%ULiC5lnm;Eo$#4JWqoyL8e#spKqZs-t2~S0~={&1Px{6sS7T89|3wH}r
z!k12%#{8?3W|A?2=8t6Pl#naK5HYcB&C!gDi_=OMmbQ@TX9$^u1IK!3r&tFx^s>kW
zR8P5QW{jxLVoaHT4>ewrH{y80C&-Cl{b;~TZ_t>l^`YrbPwJFw-1G&MLK|U)qxzx;
zgmJ<__q71h)!-N1+tb4Yp&7UVF8mvV1DxKa9l!-i%V}X@u?PDFhdvjrLk(Yr=qMn(
zuR*|Jsq>*yJYh~@`woM~;8x&z+|MCWrAc^pT(iI2Hh#$H<ZzBs|1z?VylQS!9Kni4
z6C|-`Hp2S6i;sofuy<@^@l;rryKEoDBQi3waKru&&j2x(re&|r^!$i7c=^vkf&>b`
zk@vKwS(huBcHyo1MMp>TQrqsdI>diUd!J^}_}7f5`X#A^@Ed50VWFYWCO&p6o;miw
zWBrJGD-|E$BItFCFaU)IE73X<p174}whpX^CTtVi;$MhGD$yGtxxWobI4oO;riLGY
z2~aNJE5@6ww=qNxG2EgG`J(ULOTMJY_O81)s?{^QsTeO)80B-7m|@$Wy16maL&d5i
z6-REjJnh)S8nt6Sf8UvK4H~)=n7kn4LDq2aP<LM4yO~g1MQuKri`*9nb!Q)RM;`sT
z0t$&;W(CVoB5&Mh7uX`)xJ}vTHQ>Cx@VOvkIMADa%vx2r8x0r41{(m-8Fl8Egno9_
z?^lR$<!~~$IyU}MI6v|5Z0;^}w3Y)OU!ZYxYJC*hop(w%tJtO9(p@L1UvTPnn3Dal
z+XEx$Qv^K|k8sS?e*L)a-Mm6NmD-D4D(k(?#oHbXV+?@J7PXuZugrl=x_Q#N6TZPN
zt9hE=L@mR3fc{I}vgvQDYU@gpZ4|W=!TRmkwTm65__xq|wRHR@Xa|fCa7@GSLfOxj
zCYKQ~&<~qHY7}E948oThTWc26G;3@1lG2pQ8D{9MvZAU)?i>Yr_y!&u$uGRdgL4NB
zW+~*`Q+GDe&o{~6|AQ7|?{MLQWU+|h75s4Z^4z7mUu;)&zA_Iv#`-w(rcyJpy2(HN
zo);zAa!62-&{&|taCy&+ee8jR&PDuc%7|ESfM%F&osxpQye$OopUQoeiS{0)B0n&o
zj*kc2g!td5HR6fCDl4Fi{5A6C;o<Qg%Y5}?Dso$X?K22B7+tL9wBviJ>mF%pXgEuq
zZ2&wzHV~GOan*3xztu>~8@y=ie_QxidT-M{m29{^I1`?Bhkh!;@m!F}0ughY`lWm(
z#<L^ktu4bYKN7YHIq((V-Ns-&H)v4NZxXXlR`xBKAwotutf@)-J;@9OMytjxO4XrZ
z4aMTZKKf@pLfiwNC|m#qM~-to{9C}q>=Y9dJ7W+q#8_?OYABXUr;9iYze6NFV-mxw
z=<&3xKTRB+Z%Xx#G@n*_6{wS_uc;_LvNb+NI{wF;y}dn^i4nk>oxQ!4Iu>mF@HfiF
zh)YyH)+5vh4~=)4TkYClC-SMnq$0D}tU%}3F_VoeB%ZgZHfjfd&Nz})Oxds=<=x{G
zFLo0G70KfpstV9qH*f-~ykr4aR^yC}Q$qoju#M~EoO2?1fm2<AG-d#sKqB$1s@l<E
zq;@Po=dtV7A9GBUD_iXs_wue;ll+3;&RykfL?hU@XuVTU8R(Qncim3ZI;(ZbBrSj?
z%iLKJ9VyxfUTV&TlkPm2S^}=bySV+Gyq<~AUEb4=vs`w&Su{pglnBWz9zhEt#aA{_
zgajR3b+sB(GU}$cqLC2PdJn8*BJR_?N}h_U4--XZ48jQ#Lcg7xFnp-ui52rPXv-`9
z`)@O(ca^BXM<D8j{}~6^l?JwP{NZTq9{%pS*m>K#U=Oi^@FuX~Dky1SFRucA(g1|J
zh4>EoD?SzxE6vHt$)drp@|zi}R4>S#>)Y}*UaMRD6*lNDQ=S`9%bka8I@-d_JQO*B
zDWY7Q?Ig(}d-WgGw4uH0>y7J^L<ny_l|`|O9(0m0;c3tv^jZ6Ag|g<^9<d&Rs~#8v
zD!E?F?4+ztW5$BZE4OZ}xV~b>7B5*gy_7r?$a1+Sv01ra$DMSHdn7C2xoxK?=rhuD
zLnD$k%XKTZ*oe*L4JkN-a%k^HJxr?2E_*Q%Y#EdAIC{48oMry==QVKYbN$`%v_H0h
znzsKO`2Kx=kdp<w3Ll|7i%@Hy$bn#4dkTXiXD)NXipG4P@99h>yy-5d%%BIa-S!iZ
zACIC?TbNRL;1@l7UA5mMHkM(yB?UrzvqfJ(kYveJ&A@M%uG{**2LY1`xpzu$VaLwa
zO}zJUkI>V3=V{gQwYRjam3D5`=WV=e#ccTm?+*5-egFzn2>bLGC`Y_x<Yr+(!Cru8
zW{Z&1TTD7=!Ah)ahOEFmFET&U>!*4uu)xC@F_W8DFC3o6SK`>#<`+km(G8dgn2rU5
zeO-B4mfBf-WO;9GQ7I(J+dBjKqo0>){UBTApIK)EH9nena1->KZ3T7#WF9+ZJOv1z
z)@6eT0MfO~J5c@B)>#kv6g+>ffeOu@Ix26&Y`WwsVnD4GcPCcPoAvr4B~p7`uGmJM
z=<LFtD%!<Ef2>ZjP<4E8`MGOo^_G0*9WImXYqJmaKgH-%*w}vr<exgkwG3mb(sbX=
zn>SO%UlOO(<f#gLZ_;9>D*7-)ZLWEPc8spL2>e-W?)V3yHaHxPC51b;Z$C7UkNH}$
zbB^^B7&>&alsR`~{sSK>37T*>)D_@c9lOgEF(5}Rm}<u~!QuXlF2LPg{oB(q8kN!?
zbN~KwWo2dRTd-!r=J%c#6tvWj>^m1U|2XPzU&Q(!TfPmU@ocuR0A^b+ZHLD@8ZrLL
zE>E+>gTV&=gc7G`ZXW;3;qE?U4a{<=Q!tsxio+>HUql=XlsV+!%hA`Kipn{)opWI*
zPemnDv$L}^8KNSy#f<dyZvY6~t?neKb>m-=vu%)<f(oP4Z4Y=e%nI1eOH#I7KzY&L
zjTG#x7`n^xwbI-16h4eC-asN95XeEj7bs(MSVQ9*kThctU^yIU^tra?nxgCU^432N
z<w&#bMd;zlz!WVx0Oi9pqr;l$8i?q^rDYs-WN>@xnCp?8e?F4t4)xH?pd*RqLs^yR
z*whTczvCA<EX<j=P&83?@DgONAQra!8h{Rf{|lfk0@7Oz+L3CDUEmoX<nlo@M-Nz+
zdT<lII<#d$xc~UvoJE=3s|4BdvY(R%g8dd+F84oqPY%V^=>5V~!yTeLW`E%`@8Mtf
z0#q(s2=Lu#0+1?PP)4zlyPTr7W71B%U;FZ`4M0s8YN`u6;Rb5iP9M?5SZ)Q-Jvj9-
zlNL%rffsC{I*11KBn4F5tWZI;Xw68q!6dC&auEe8Eb6v(drxg8q9|-iBGd?CK1L*N
zLylH+XKM)Hpk!DZ$U(ot7i0B|h#z!q<}f1Oi?8`LAnLpK?*||noI+}kP+MAt^(7Wl
zzX2q*@w#kh*ifPBWV8J-XX86RD-F?B+C=PePQ{3Jj9Fvy%Wi&g^t&dIOZ7>$y*=g0
zwUCAxIzDp7xEIXQcW;vH=M<D@u;OSw3osZg-~-a7YgQfNgUt&50cj94h6KD{;3abj
zm<2CrG1Z*xfF4k&o9At9)cClV@}OaSYibLI0vo+WJ5c|d=*A~TMBD`XE=pen7?bR~
zxS(C_1L%e=?*~9hZr&R9pilkR;PGdkq!m91s*My<IUkAdbDuX+*Ak&69)=niL+EQS
zVh8y(4-f6!-3r*ZkyD>8kzWJPu&0h|pqqJumFk2#l9%6v{03!%AdHHFQP!pcM!b0G
zQYCKUM6uW@6<Pdjb4?nOPYT`;v>U6jX@(4NcmZIF;CNy<tfPzjO({-dJl?WQ@sgVx
zH&ftwJk^h;?FBP4Gon$oSlYR7j--z&<cIpkY$QvwQOF1d4}>?ILkpU630*ip(TniU
zf!Y?V|A^9<_5{sI6&73_u7)oe8+iXl6Csm4Q%G+Ms)!#-KwB%Oc>T#4gI4E0_X75z
z8p_=UhA7IBF+8OJ2?zf?+RT+eTvD7MbJT}ELL+y|1_w~PnvAuX9tX(P3!Es$op?q*
zdHS@0?ZWBPLU?ATQeLpAx3RTN=CT@MG>9Uh%fUYN!}D_|_<`*^#U@PkLnLNL%?UyQ
zB3?QSi}W0d0>I04^9^K~sAEZJD0?!lT?E+06F52o^+YL4R45d4m*<dPg#46-VP*3*
z6k&IFIUo*1)$Es*HRWBvrB=l_*ySr$Synj5a|Tj>(4Gi>cZPm@je@)+5ani1?X9R$
zQryW1^Y01F<>yFAAUKm*b^6$#IYj@0*8TXY{;O+PRJd+oRx}wIi_Ko#AmtWWH5Yy9
z!k6M`x+~V9Ye1fQgXWb(qZNE3dy%d|HsUyqpt0)o>C;74Z(&l}EGJ(s%!9q{1r~8Y
zn(vC|0g+-uOSC2|a@?~JuN=``4^@*PYd7Ywnn#e^i7dn=y0*F5VC*1;c08GD0qV6e
zD<)~8+eE-Gy@-~YA788eG@<ISJI2f&&|jKk$xz2TV(1y;1aOh9<NiLp*v-gDBZ?eQ
zDSmjPO1;NYQ)+iJV$u4r$kMsE6X1$(ft;lF8W|eif{Ib>)^JB&QeV*|y+e=R3t>fI
zm^uW;!XJ}$X0b4@bKLlkB0`yFPt94GKg_`4tox@`nQ-lZ8ercaXg}<HO|{$T|1U{|
zXd)D!1chbGHJq~OgWagr0)N6L#e#Z?a59DCh75&7_@GLC#ZePzI<`*USJv?r7PIS1
z<-u(Imq0ku$2+BRgrt!6VP<A#x6!qOtvK4oiHo3RCn97$(bs`BV!94Py%fOB7-igp
zcYY6n0gPA=0y-H3U7vUd7yBeECL;z;TT@KAmR<hyc@yRmUjur)2RSjGH*5~LLG0l6
zaA1%lc`|E%r{`T3OpPJW!GR%2H%8+Q2xDCHo}r;oB6@?mbsEkQTu{y-y+QLuhW$u?
z1AO!rf~Lb8f}sxWZ7QauYH+jybEGdKSR{js!}#4QJ`mU_MM``Z`5;0e`bcc~@B_;6
z2r59{9GRcLe~00}!&a9Mxvv6p7%q^?c$NU_nNffF1B<S*8zra$SV4g%TU@4p`S&EG
zyr7qt5K%R%48nE~3=9MY1yzFGd4e(p&dIA$Ua3K*fnA6BOY>*7AsvWbt%mW)3LIoc
zVmV+p0{likMZ0qw)%Ku34)h^5;P@1@bWVWq18-{P!=rYA2zJnuI!6Zp00P2Iz1(|q
zPDn;C3AE#{i(H29F{Dg3UBr6sgb@ZF6~OS<DE{C99e^SY3l6`d34p-%c6c}^EdTQM
zgr6WD3y@uzjb8)uF4Ayn3XmzlAo2l8;6)DLf>0h`0g-P0!Z0ov>9#CQ;~aAnTWS&X
zoi_M5*XC!RnO8G_od=-d?Lkk3`GwV3xf_By1#~henaR^lkh=jyO?1Lo8ss59t1Cxj
zLoVuNHbo6Mzxd$}en(??ZE-=7{8q3x(8`cO%xPeww$N|MM;**aP#nDt;Vup&vDtFy
zen?7noI|J}4Dc-xttM_<6^`1P&6^>QVTW*Lr-%qEm>!~nUGFh>hgVwr>542R{GjNB
zO=xhiD*k>C?g=Ith`<vR7fB?L5r%@2=z-QIz|W7Flz>-!kgENl#~>;aoI4vN5>$aK
z_(q7@8Ne1fZfxVK8?dPL$JJk8!3+Q`(a0{oK;#u$%tWFbU<LC;?g63#DY_u+IF7Mp
z^6cU4XaoIDK>%yNKR`+Ifd#bn{aOnoxySCL7!k23u~nUyj43SKPmE#}TIKUlGiVl^
z*Bb*|y5d;8RI^wZp2JB<s0*r-4!Ro-^!pHfGNv7K>c(oDbqeDu_$=g|y8P<l<2(KH
z!3Db)%Q|RgM&udN_PL>?xKa$2_j&?8x6|E_7tEYO7^)G--<qi##_@Jn0TbgnCM)@s
z4l@gXKK-ocXcJg*k=gX#^-cD8-~#KT)$Md=z|$NfLose;nEp4WA9Oi?Qg7Wj?CQ6R
z_<rL&6bmKqkd)N#%IHi)6ESN^6UFd-WP6KK_{yRa=wC=`s69AVu);JQN^REm_IBPR
z9;Re;N>ymR1;%UXd=?N3pzcz|&6HNifbTdf6iOe>%r1BO(edTo1lZZ~G|eOWr?A>U
zW%K~@p(>u5beS+AWAVhZqWi&p`y*x4HGW4zYiNtq%fcIEtwK1H6#q^1?yu_?+^sTn
zxAp?_(f0fvcn*kK41<^K-J9&Efxo=?>jS?nB>QH6jTt>_m$fkl8jhXMOG-%l<7Pq<
zfPiQEufl&XiV)$C)}(%S9)Qn_*nZ$&ocfv10@sV1m~3~tcYgZx?vKeRE@5SLwc7V7
zl`NmA7{!;bHIi{Xk4y$M=sqByIj@6td!Ci=&xTF@UxNqFBPOjSIOM*hlA%gL)|(hr
z6@`}B$;*oA78&6oTcW(Im4E#Ex<E^*;xDAMUqKV-JoS9IIrVnSQ*0DSd0(F{J5;Go
zj8t!-I>==2z}`7XI(P}H9{q>N`M^9x_rgv;1O!#)yg(`{T$Ym!?z*rOsldR5K+K`0
zRTu<kWo5k%=(Fw#4rC>Ah>HZ*_833~xaX;3PJTf_D*;-D;>$gvyC=aC4IY7m5biau
zxC1|LMkTXF*~jAY%E=NLFc;LnZt}@5#h^m#=#0DwD%S!dKv?3ZFaDk!bv#gbmzd_t
zUu3(9QM2v1?l>YQ^!>G9!V!jR@W#j^B1>Do?|WA&+iHL2Ob9sCQ-CTkE)qzSBb;4(
z&SqGSeJFBuD2>uN5m)N$DhA^d2nolM4=3@VrHAGC8lp;qw#ELfIL1s!JVp+AmzZ`S
zl6*KgQn-BG3<1s@7&hMb(sFQDM=s3TDw%=O@jEI)5m-Rm$w7yJmX(``L!!^)f61F8
zeis4akj)dN#FH_q#b8;6du3JCQssr|o!C-0_RJv|aR>oaP{IAm6>xHuh~>{_8#V#5
zhhrZHWUjRrfd^C|`eImKNZb-{nLyzr^dXnn$7CQ_#mc*;H~t)Lr6Bckd!*V2hle*6
zn(V;++$FC`Rlt7t8}OgPM!ka_2^S2vY~4CaIu=B-$A;8k-MSPVGLYgkCr3`}GgFIP
zC>F3jDG;3aN%Xe7Q!XD(6~!bmrS={ERmaP`it2{1AI0}ygR1!=T*jYy{3k3VB=%6P
z9SO;$@nw@q<u6IC2xw!-lo%o|8ydkH2mnVh<Lxy)vV_Du5D#R71BID+#G1t(O@y5H
ztP~Du6k{jZ)<Wd%80anxNLZpTpP#;ud|>>ed{~1M%>w4MG+<@W7$Ax+f)|aK2K1qL
z(p<=jfaOb|5ryP#kBm$}PL6~ag%*euOGdZAz^j*C&M<`+|0+1iyoE-9JSZ4AlaU7K
z>r_ybjAHfhHz;}9wBj(uNG2C?aJN8dR*eXS9JU1bH$cec=enL&LgXPJEXsqjWGfaf
zv7;ZyHK{=HG77B&W(tWq;yT`g_WfN!NH3%``32<ahSR4y7zT!+!Tp9@%JvsbMi9Ew
zpxrlebb#oqqxpT}{rAelH0X>JiEJMgb~SoD@@`0r2YgNa?AcvxJiFOftXKhFn7lGd
zgIh5%rZ<X|UmIr`uLMU{2`z8{=6NcsPkv%zV%<!90k(rJAifDW_9F}Po<t#c=gu8W
z<kx`A3W5%hz*Ys~+pkY+OUFgF(L+#}AWGgw^V*ACPCg&LhY<*4V6RdEl>>en<>zv+
z6#*cF&`Gd)h==dCvt|-JU~S#4mlAP8f8(Z2z4%7S#~LACN_d@@)4spISUTQkY+f;P
z=A2T`DjmUX$#?k!gE!`|)BK{|MW(Of=i*+|`dQ`Xl|4teR1@wlTCG-aQjum+Ut{c{
z#<|vdWd?<PZSvvt6h{S_6Ae7+w8$;*`#Y5#{`)gOF@9s~knGFhGfg#<o@G<sUphFI
zU!wvf)=HO9+-#IRZviQ$(&>iSW<`U`NaCE{jlsdejsOwx;c5_gzoRt0gYiZYm}KI=
zJkH7rj*pM`<+$&$A7_}j!+gi6W-x-KI>zX*+DueIL1Ffw$p}u*9Wq{rS+q^a4dJ9r
z#;-1(VnwhE2Y#VEqF&{RHgp}TuQ2#tOJQ<oGZ4ZPL`FR}<3)VAaRE1CAwdvICC0$Z
zDPVE(uw97x1nKqAmshd~8Qc;Q5=1rW<7AW|<zxu&gj~;UZLO{A;YtvSwCV`X98YGW
zyo+7m_zb!bj}k2T$hWDyxuRH3cB7DP!p5+$Fkmu&VA<j@hv~9QD1}7C#Y527DnMPv
z3-MkkZtaw4+OPMB7OBO>3cS3$-yU2j3&JPS1ZRjBz6KliQ!rS9I_sNK`KTm*p2$h@
zJ)eT&7$V^tKwS@8yt*RM)z!5lrv43jty2&mj8?6Ze2uv>zPysLy1Rcj&o&+yo^Z4-
za{zf{U}>o+B3%MYno*dnOxE``dns*WFc=1=rabs&-C$29<%2yzj!sSkK1pQW_+9__
zUf|lXik??WMr$_5apZgGV}G21k<mIZJ)uZ!Z*Y0Y7*8|u2Qd<YMFgUBa7Lns7}D0~
zQcbT9A7oOdyp4Cqgy1<Z%u@E5n49yW#1BTUtAR42c<OBV;)3T>fU=Q#N}EuLklD=|
zkQ^z+#l@QRc|X*9rp-&%BUb%>p*0j9wWLUrCLXroP4chur!|#|Q!hlGKYM0xX}I=}
z@1x<;a1<^Mmb`wCQO{%}CG|`_g8G(@t6>%s69s@hMS_aR_GcDze+9eqOf#Q}w0xYG
z2p-wP_x#P+r7I|kQo^@ylsS96oc$30RudBw^+jKJQ}NSVBKAWpkdpU9|7)bP!Lp4@
zzk1MfpqEpvoJB#N3bz=);CEFJd<8F*U|I0*Sm`sA<=j*{D79T4ZNS83>gwzFKW4pt
zBclGtNpuY?bOb^JYwIKxb-QGgshArud}3v0CNHqfdViEO#C6^h5OSW<>sqyhr=r`X
z$f9KGt0xN`s$2(RE4az{@CInpte;Q}4Aww-SNACX@)s;3_|;dBBm4%DV;}i#&mo|V
zmAiDZ(!<-^8zNqDzu=8$p-mKj`O<_!*}%cUAtXFya3S(IeQ<DjQN(4|mrM)9c6=0z
z3%F^%6U~~JaC+FqPCnsY^-^hZZv3`SK;w7#Zk@wZw|G(p2r5soVy75RA&k_IKk<Q9
zV|u8tY2V|(O3K-73^+8^V5lQWMS4I1N?o*esW#ro7~URyUEeh-SVlOhRO)E0aMF4j
z29`U2Gs17+hB=-+JCsk}dP7ET`cFg-bzx7u-ARS~_1874aa%U!mCZU=&LXmYLmb^?
z{Mx2ixd)SU=FO{DE|lZ!pf}MCe}*j;uUBmIx@(~Bv`2kSj}rQ!KQRNwY5h16is>av
zaFHLyAc(_pgLt@+eMf6x7PI7#23m%@ZY%8`;i<y^zjJ;DTWZ$etMX3VrTrQTpOYX<
z%wq0sS;E6Kmy?yH{c<JK2p68Or-jI^%5-&j$}(L~EAl%N=#B^VMX|z{a^l41XV0GD
z#hLu@jg-|FFXLM>rFVnb=9lXV*#0sxX|03NCagH6@S6WC{wQ8Q_?iSuw1PBUPA2_U
z)MJYb`s5xaCV!Asv7JRac#Or%zg0+<P8gzE@Y=qb_$nW|iNQ23Y(4!o^&J-xf3w|b
zDX{})iVz~!%>Vx%?U~(rfbzUXnpiP7LjI)ya@1aCm(LL}Gt6BgPduNQ*q2KD^THK4
zM!ZOw4z(;JzR67gvp?qS+OZJhMxH;@$diSoJtkHT>mGF`lRs;z>mJTkwYm9!0B7Xg
A1ONa4

literal 66248
zcmeFa2T+w+w=KHOt)jFFDh5z7ppuj%*$5&6A~}l)NX|i`ZNh+{s32KIHW`$hQBhD4
z5H>-whzOe~VUxoh3(tS+ovKrH>%4pG)~k08fBn_;2EMSuoMVnL=IT4=6lK?}+Omp5
zp{$WReeyhovNRk2@BV8!KG}Tm%5(f<`9*ozlazV#zh?z8k0_KK6uFbfR9(aRTHHM<
z7#;INT_z?aC1I7K(c3upP`Q4d{k&hb;>E7fPdlF|t^5?nyq@FM<G(({#D&S*pE0YL
zJQB#7c1XC`Z?KIcGG$h`iBagRSD)NJ{L8MWr8P^ZewyL6J-OeL$9;g|R^Z80%)b;5
zfiKO+DOoH2d`r2sV$r5Q-*Vr-y86#|fiiEF{rOIw^ZVk3@2*C#vt6|C-J9ya@M7fa
zg4t5tKi|C)T+I9DJIyWs^K#{tO-m@YE+L1EZkc>}+TL%7-*|I9UM+!|q?6_5zL_#`
z9Dkp`i<ei?dzKND<wLH|ylU;*v9Z4?&C!}K{`zyt|D9*}Km1DnRW2v`v!^!UFjMhp
zcTL=*ET7xD?`JU=4;&U|{_s-~n}}lmc-6S`t?aw-x%ILC^m385Va6Z2%G}=~RFj~=
zmvOo3cxPFVM%9Z`tXsDx2$)t}?kKsho$vZuF;Xg=(^JH}=5&D@<NW2zv7aCBk9v1=
z^}z!N&g@f)sA<fwV0_qQz;)oj0okcrIlE5Xjn&RD7naSj?^MlqwPW3X{^X)Xi#`YO
zMV>!@o@LXfn28D3a8U=D>G8q()&h4{cJ|~1jTE<OtL$jJk9pOLo9ym=hiX#{3_jdt
zRe8AU)MGvasSoepbBT#*uUogSG5<<hU5Y`<)2C0jZI#E<3to6bk(`_8h(8~Fckz-X
zPVMiO70k~~he*0R`!QBi-spXxQ1~`Wd*$!EbG^e)Dp|jv=eNVxIF)@qjHWD`!B=xv
z7cF6ulZiVY?VD0Kb)c-W<^EO?->f#TrnuMQP2;oUZ8m-1UUQZecKrC^`}~B@mcqh9
zMaF!)4~Lj#ih7dv+TKmPy59_E_s9o+)1Ci(Fy+!DyY9Nb|NcHS{WhF45BD>_aq}h{
zek|A@rIlu~wrR7l<x^S*V|-kM*Pzg{QQg(mHPEUiQH!RTdg;!?$B%!GPM<w}nxCo|
zZq9O0{YzA&zsKY-TbfDv;qxjg{rcm$wcBhL%L<<ue=HL5*@h()rHgm44mOVcE@rlv
zpPO}?qQ|7=3mX)AxOIiQ2wK!}`;WHdUvcPf$<uiL{JEC?Ru-0*cos8yb@VW!Gm=AH
z`PtD$jNW=BA7&2q-nyiI{WPau_3dJgAJgyuvt)%>ivd1Vi#@lKHvHk<rqIC2iHUBG
z`3aw0yLQ#&x!81;ht|`*mMmGq@b{iML9T(t6Q104y3k{Clu>VB(f#!~bv)nS-(PkL
z(X8+7GRjqBGQ~QXmdbb5@6nAHaqK?l)ctj*j;HkSN6uBu%o=a6Ez^sy$I|4{e6>Tz
zQ$E0ObA$Dgii6tev4u0k^7!jE>&Y-thqJwP@oHspf2#)MI`x`pZ``mU%Bi=mfoFul
zs2%Ps6LRWNot(@mS|W@4q~oQW=+z(kg{`N$qx@@9jms>vmQW~RC$N96o?62$p?U1s
zG3(m!!^USar`Y)~TpjMJ98_=dnw=6A>iDv5adPvf&6{JDq$b}FBt|Jo&#JzD{hI0&
z9{hpz@4uh<Z5Iy-^q!mSI&<bss*W&j>@?UaStb?MJ~)_yzd7mDQ)8ERKhvry-g|yd
zOTUx}AzQ)+G5>1bIYv)y?JL6*DUA7{QjUW=cD(7ITB@6z<46yXn(h-=!*=+jLWrR9
z<ZvgMg~`du{rvprSP!bJHf34~dCz(5*}FF+Qzgwr0r%KXo3yA;Hp5n?GB75;Z0XN;
zXPD7S3@=z@O1xGG72YCh|FSqUh|eH-uR=)dx#!0}M@kp<)FmZc8Tl35D*Mkr|6m6{
zX-a#Wx|JlKdXji9*bNO0Z4okmd_W^P@#<BIT7tTQLsvz4s{Y8Y^0b!mU;)#bEE|I-
zJUZ%lNR@_E<68VHHhPI+LBJmQcv*jT>+#UQz+cWSd7mo6#cQzxpKvK3uaut8|5Y9u
z)8CYJ@bKaDc6N4W6clQ)6d9!)-raeFg|zQ)uh%AO3DAPi2TUG*=Pxz&addh;W}z#>
zy`o)nd-<EYw{M?2dGaK6L`q7^ZLnZ`Tc)_}_oIxy1{G{@4PD(Z>*nm|v(w`iA0O=8
za`@71vjX$lcw<o$Q;+%C@t*4FdnBS^i^X01aM!#wKZmLmnJgc~Yn<mvq5SsF&nuO@
zn#Zc7{4;y2Uzpe37lxZ<EG-k2^PH{n7}8VSoIXxPGs9(;4A0-+k42q+xXZbB6OVSy
z5s!&i#Z%%g1A_h12{Bv<6t(U36}2&3CK7nvu2)_NIQFw%AIptu<JG7gzP`SKN#l|8
zGo|VJCr{omD|qwfsE+|h+P*`Xm#f~j*N;9@lKM1UVBMJhDSfUgO15RrWAiOT-|gac
zy>0Sq{hFm*@9v`57%m5r*eL8adU?Bq>sRZBzJ|2zbvzo$x|vf0ZAC?NrpNa-@tJ>p
zc5H30^vv(#RZL8d2{tCSorxW#{-J?n?oQohd9CxSdu2GY#P6$H1TSC6O4Q49K5*{Y
z)t1zsKQDP=bvu_eW?ES^M($RKlxi`|+0CPq`6az?ZhBAy0XxmIII-o^qkVa5c#>>C
zf@kjKViD5|PX;*leLLhMU!K~y;nPFU=`ng&q_?wcId+ZX#BW2Iy|lEnBp%9|Ve@K5
zP#Gdy%xj}kb1bCBJeQPRoSY;ty?nWw?Utbfd2d4<lSTD`JUWg)Kdz&>nLLk)irII|
z$LHYQy(QMcQO227RaIiP->;fYD$LIgdT$TROM6_;&r~q^E0nh$TYiAxsy8o$ZCVrY
zUNrk+{%ukX6_zEw0$x+}eKE;RISw(!!)Klx$f=cH6yz}c^HWrz=X6g^oJwPsO(gcs
z%LqyLp65R7=kbJU*|vrx26j51IDY(W*{?6bM%rV2X_Y38rwfn%v)H^oIoxc!>`v<^
z2h`4%-QC0&u|GS_1zDvyeGT24`I2pIvR+<lv(H*2<c$QqZA2)1&YhCuHJfV|jHub<
zNuq+HUQjLHq=~>k4?N$#eXCvMRj`pq`=|4R%X<;%sm55cwMgOpfvjBTj!L?ZrHj@-
zKe3{-GO}%e(N!5)KgH-(D18wApiT?(<d9Q=<m+%@HxS`E0v)+O2k(0&8q)q``^uFo
z74=;D8-;v2*-S@V&qPREv45#|PzxbJJN@#*-r2pocOMCJdSVy+0z>XK)?j)@PVTji
zI>umpdH0JKr|#wIc46g-@~@{*zW=rGK2MOYJB36u*z(c@Oj8?AVyCty&CSltYzthm
zVntr<K==!tENe`5bp2GMfT<t7VEp@a=6J;2fVoY!nbs|h>EXzrPXmL?*N+Ig1ls1h
z8(<R07a(y^i$kR(drn2L*5$FMJZ5?H!s!8rgck7+tNHT3znVc3aO+LVQccm9T()9G
zS$(|*?oKa#e`iBl?R=a{%xHgBTSy@Dwj<{PMr%}*_9ONV2uaV+q!s(H?B2UqO!5(b
z*3e@m3dNp-yo+r^l)qF@jT7?d+A^g`DXVc0^)y;)V_sSKzP1**5Me8G?ZBr`V`Vu!
zRS}FLkRn+)I3mqxbA1wqBUEJF;6M_amaSs3ofthSo0esLIL~`G{daq@@bmCIpC~pH
z4BOZOqSGD2DD)uTJG&|Xd6BTC9T5<4F_&%p$w7^6nb?|Ev{Znx1ed|qwY{c$goJ7h
zbBc<*GaD=AmaX2XiX^H~mSx@2V{7twx6tLUJDNJarJFtSW0y3~;^yKyeSHP9dV+dV
zUtn9d-HEawzRFbp@=%e`K(!QoJ>$=h?W|v6eGC)Oc~$4#yLT+>*I&%y-o5+eNLQtc
zruSfL0X6h%q_lUdVGB?~qh~-QzVLel`1vr>>=`|3KAp%E<?7R-pDaJMDQNW@)MeWl
z`7hP2o*(;hSitVbN!l~p{>BW&{@08am$u-*$*C!G><lgDYQM$HR$Z=%Rd$>1&uWPW
zTK%{2m&dO&dQ*zx@y-YIa?^)@l_yk3%lF>gr2mz#Xy%9QSm178hBxV!ZH1m|d+Xv<
zWVZ@io{N&X&w?cV;KAP+Hf{RSv!jn9uKZ%lYOPF4ZQN>!6hPu)U}(2{akPBU*+>a4
zIe^Q<EnAc9J3lLyhuPQR9tx%%4Xe3q>_=HjbgUZF<yNfS#?RwD*%>6{F<~3i7%-RR
z(u8&8G&O3b5PbO5o->babFsI6^OkzQ!*aEKc0T%yV*iuJ%Brf5i+6N>eSYHZojb~}
zFMc?iV%C@*=RD9%w{NTXXuh)T1heEz?BQ=b?ufXI2Ghu#=B^MS3*BVSS)0VHTX)RN
z%%&#0A{EA^4{E-8Mxt0&tD8K@caDzf0NW`@u@+yR9-^iVH``$vc(hVa0^Y{coDfO3
z^GfDFG2FaG#5(L<adCg(^u)wv#H^myg1n8qy0PKnE<M=4xpp15jvqg+ny7gQ`B!Dl
zR^gfsf9Z%hM_Ck}lBDq5QC~1xEl+;W?8=!lcLRC#;vaI#{`wI+SJaZ{a&XU{J!PF9
zSqVt-&F#zQu?`sBXL)3CrzpeXTc1ixOXGTjc=dX!RsO0^(upB>^xG>#YQ3Q4w{y6g
zW3B?Z5w|jGX5Q7SS37JzCz^V>YL_RYRzpi)#Hvy7LrF<)ZRrXQkMHUQp%*V+Ot)^y
zwajQ&SiO<^D{@D3{8pq4zQT!jMn<A`KW-Ekq#m|6mHB>q9cQ|I=b511MU<=J3&po>
z!$3=3E&haC#HjKz0y;9^1C{y+aTmqeDTYIKq^b-CoS>h*N)dh4*?nyIwrp_F8EF0G
z+U;VfO&X)!fCbH4ax)aej<i<|eF+gNi|Zx1WV^IibLSYXx_T5D!TjgPhk%FnIZ_nL
zD=89qzXt-5#T~o8oXjUcyOfkZnJ9hN_GQsjs#r@Y3M@RD>?|q~Eq(VN?^kA6Hk?<9
zQFO}nl=b6~PDT#w!4Qw0cF*ab80v^OC@RdfX&b0v02<P;j(*mR)6}+#SUZ}_1dEes
z!*l}Woocw4lOTX2ZJ(n_Cqkrlj(xP-Xm|UV%H~CsqXTKbN=Wq1v>okM7Q%c0cE#ey
z&%b^>_hi3{%*|D-=dp$rBP8N7Eb7gXQ10HnYoFjhSU9aX-s+y>Ha?K#+}{|d7%p}|
zJ@F~2{MgvyZr}ck$DmNRw?0L~-ab{ftqp0<5iv3zaMipiGl|{f_pvA3TIYem1HVP~
zVhx&Czq}J7Vw21+=@x6(QF1_q>brRHVwUaOlYl+V@PY)W(y5aX7dGD_>#8BmG*sUK
zxyzqhK*G82>2hZFL`-WufbFU^Yt*hUU!xkYy8q^l8(c?@Xny+i>CEZV=K$tV<0d`Y
zr_=+e+4$PX2Q&J!n*9O^!{3cI-)iqaM#-L@8Ht>)5g*L&0FA<5Rp{wX+eeeDe)G?g
z4BPg1y4%(z@*<u-eE1ObNv2MrBmF#MphY}LcG(+Ryhchgr>vjr0^zc-oR9k5^Y_;{
zd)C+NigVyTK<vK>wX>s?ltc*ZnL5shAhRgua>REaY{>2;F#M>W`$)NnLK)!WKId{2
z?+pt%;_9}C(%xQ@6W_0E=Q_rTIrUr`54mw;>6iloPJ{Eb^c>-t0uR}=N)$}|xTk0!
zjg6z1>$rb~EcVsF@-O7lS2ebYJD=t7`2EiU2Q)xgpp`y4VMU`X2*ou0>VJ-36Q?Y2
zOZK(c=A#oG{=InpR(3FM*(YADKyLhasU_EG7=^+ZzpYO;&<n3__A>+TSv6GnbwG4a
z!ovFc4av4!fCUr@li%OpZjo?(_2;&S7X}B9`~QOL`maI^VRR_}r+LEv{MdQ5vQsEW
zs*Ak6v0N*fit$j{2a~ij&dB<0!{k$T$YTXrTcGIDK;F%lWWt>0Rks)WtjwlpE|zB@
zMC1RGzWmQW*21#Ih5th~^#5a*<F;%}H^W|$lLiaa*S~Y;Cab2*@?X>5BTvECX8A0C
z<0s`gb!jSpBj7yZN-O&oie^M4p^ep@P%wES7&N9D->Zm_Ojx~%mkW90)V+;rU^yJy
zJjdgZRLp7A`)2~V>HY1Cvx{F+sXetC2IY>0VA)6&yh&#1&oGpO2P7nPrROH1fC=M)
zdR4G{bDjG`5N1_hT{uR9y^wj$x#C-E7y!p{D3e>+o%h-lFanWxsE;2%9(AZs)}zHL
zM~xzXx!YZ$T<|0@WmA7^fsViA*i(SkXcVj#V1&fnMq_#P^J7uP0b?i<2JrKfgM=p_
zr6;gx-Q-ADfutkF<pEeGLSzM36pb}n5PG6G*WuTf*Wx3+zt(Nrme}2WIi^H8Syz<i
z{8ZW1R5RQ5dGYjz7Tja2*4TDd*66p_S0vy?dz!LRHuC8w0*>?!7R{@Bc;tgad-CAH
zv)EX!*|8=Y?M%zyTWhz+uVOtAivmi_YbIN^a5>8M$3U{h<7q$?+)XIe@4Aik1=C3_
zkzrcN?(fS=8BoV+`~I?~MzsLtzj;mULz3)+ddefE4FGj&5$+spZwzRG(<_($ssvEb
zzq2zwZ+5mgBp@I_(D2=jvt0lxu>j5{d9xrz^ZJ+M(2;@Mx}W=y%pA(5R53MN#>e=2
zg4qltOT!jfa2XNq1ERnMCJEu8GQOJ)$7ghG3OXN-E&6VK3LKVVHl{AZYs#!VT)h65
zM}~PVZnSh3S3!P0fAQiAvRXj?o1?xwaQ@}#sSH$m1ozD`QoK9IelKb^1biPwWo{zY
z7Nf+$dO-Cfr{@>EQjB8w@M)c^2|qeI$U-o7w~5crKH@ac+%VAfK?}G$l@P^AcFpk^
zGIK0#>PV=tl|7yLGKHc;M(PG$6l+alHUlFI7?pU!XDz*4q)inFOYOzUm6Oxc3A#D<
zZa?nqBZTEF2n9FPmbBjJ=xDQv-|bnQXL%`g4*~;Ca_@LxX-L*7oj)JY>M^3=pRpqQ
z;y`l_`|2k@e*6$LtKOZ}<2CeQ(-iQQ#9=?3U{CcJh0tF^g_Q3LM7fqGXzlWn6fZ9?
z?bjD?miV$o<zE>QOVmNtLhuP+!DvAFBp$cNUwZE2!=c65X*Yd*n)J?~MAfTJ(y5Q~
zOwxRPyl8eTRJLj9?l=IQRe$}(&okeXkaD(c73Orj&D(%*<Nu<ufSc6`%I{6oN-OC+
z1V|rAvR)Y@tSZOBblI|H#$p`o>=j?+D7G7rVUPSx3dR&)xkr2BfOB{=fY-JOkwIag
zz-KgCq+Ge=OBvwBwoi92#Ij`{EaPXo&o0^6(f>|+In(AH#r(Qrg!Mwb5u7<W)te&T
zH}dn-BdRgh5v{u_N-f9Uc-hjWvBz(&A`haTsA;I(g*p;==;KuRTx5E>V8QQqtAp7e
zKLGn4(`36FbR*(ZM$_Pr5BDUiS5TU*P!+!U2SsK#_AouA7kmv+L2k1j_=;bZkv1dD
z9eCs;Gb{4d*ls-srylP<J`k^8kV^&#RTv0B*_5KW+_fB1h<<K*fMLHX!V`cP<G{Kc
zP>p>+3Old~@!-ySXfghH$o85ICEFUnzvz1modRj1LvZN1a}Ndz#`PJ4ZEgLY`ecEF
zi|h$DjZg;N!UCcjY$b^=DCepXTGMUX3N7#Jd~TTj$j_8yP^6D_s0L|6HP_J$qpL(z
z2U1~H%kvq#tMr7e2iU|sHu3a=`|%siF{ACZYDnE9+gNdVdaQp7gaxo7&kv}@bD=H<
z&Q!xht0rj+fx<baLOolBb!7i?9#ho~0Iz{Ks+M|5j*LAPd(YSB*Abrj>SVeC(h*k*
zfSnO9pWv0_A*6h4YcoI~)WDMyjRrN``(N`S>f<N}gQX<1F?l^zQ9i{}CergBo}P>^
z0+ru*EI?qQYSXo*MUYhph@2)`|LoWjf3FG>7xKD-cy#s_^Gnn*Q<{aCD8H`_b(C%m
zNR=S9XRRH^xXMJLERd&6-Cfdc^p%+ncHY_)@1D4r>J-oK|EvfbFa=*JJiAT8m7gX=
zIE)^5+uk91OeD2FQp_n{K@0O}H#c*oqe_;;>Dw*r^+TW|IY})zZYN&k?J_&vHZP2D
z`%~2$WaA-|iZHX?d-eopnhs7sc<|uDXkUXs*%(s8V9oejekLOfn9RM6JWL!M95E$x
zc*V_|`1HS|wlrt^TQe}CTS1&V^hgR?HP#$K@RslIO24^!Q{#3ERB?Ldrel2#ye-q1
zKauL{>b^|1OiL{qDby72ca`xK2?E{rRzJlA8WJ`xZcHb`+_<qK$DymbXfJ5+wbR(J
zR6Bt2m?=S~;)!3OHomg#SKiadRl#<3fUpo+zhOi8K!@{V#3e9HEu)?lVWI+M!e4r(
zr>3?bxKbIbHt||!>4S;6k6aN`QsgqI&uP<|AKVjp#O_^pZ9))z8hVGzl*OgZx7Kag
zU^FGb2r}@pi%SZtiI(3KBo$hlpfNu7U{Q1QQ|KmpNy&$J@UjTl`W4;aez*_>u+Vx0
z+t2FmS&tBuC5PT`(5hepD_8Y+zsl&$NTqw<!(R160RC&X2!5n%Z9&ytjVkzX`7Ybt
zhYl$tA!Ig)RGih+e0qzCEuhm0bfE@lq4X;wCNZY09w@lYo3mfPc<~}Eca9pC?=Lks
zVFbnS9bGaSY6cZWAxSZ7Q$j)Ad3bnK`-e5A$p(Oi6VL1FmwZ^RS6CQ4e`NxsP0u~P
z!rz)4B6>w$dP|lwo$IOyXI;0h{d@~(rv${i5J9u^nWlM6+r?C|FKba;Zqs)G+7Uuj
zr=@22lGJJWjUS&Yx{vkQxov}rQP*TslmxY)*>UU<Yobn;BD97KosH|)2hitlj-^BS
z7??yrqs`_BTQy#qDlkKl-r_DEg(?Y)M;>LM*z$H#e=T?cgu6^DBa*P6FKM;05=2!V
zBGexk>{+*n@{5GL_hbD{R{B1-ZizMjUC)NNmf7$WS%MQR_xa4md_AVE!e{XSPf}7+
zmap0J5)>UB0rIZDKT$h{Aaiws9-o~;sL7jEJ$O*;%J55|mY_K?khoi<JhHuJhGb94
z${rLI)dCa~18)%sp=Oi5>oI^Mmz+p|XbS|M;qmeMSBAyu7WKTKL3%*$iaB;akn!7Q
zHsY!I>cT(dGBHgyfF)FgP->*tlpL8<AONID$M0_zfxOp4c0XM-hA>f&I!<qkw_=Md
zA(7?z3LdlH0-vo>;67f5E3*Pn0kc7by^LfnEdSs|R20!~@Im&dl>KLB>6r`TGo1bs
zKVD1ax(r&oL_Qa3^AMo%%{Dv?45R`O?C0kG(k~3?C1l-vq<~SUW1TfLkT+<K&7x@F
zJ=Uk?GaY~|0o<XVW?Jd&em2j!|2&PVz>{~Ws_GHNz-!W&m0Od)Oc*a1hdkB9vlS3l
zYU2A6&=_a#+_@v>K9-2AlLAt{v7oiBt&Q$}388Ch(0ks(AsPGv$sj5WJ>#(kl&BVd
zqbn#bf_#-bme>ZRzFYrc=ZRb2e2r%&hS<%<->%?DH**1$q}KoXxxOG20ou|bDNR20
zFfY%eOB2)6TiZNF?wf^U*KEO*WaZ3Z1VswwoW16gL--A^m9Gy4)RZa{!!A2~i<!gP
z61aYJi+h_7Qm43XYrv4C3pB{~BNdU-HjX_tpCA?41{(R@yC(qHWJD{B8V}hzUIp61
zGBDx^uU_Lf=pV)7uWj3phen_9oEkk;AlzYJw8_Au%CKpGtRH&iXFN=9O<>fXyHT?K
z!UPy*Sz>I18|_{0{P_0TNKbq0aqVk2hB>Wp8;`PM*I&RZyn4BpvQk{yJnM~SA<a}p
z5BBz7sEm+YTV$=QtgM)oVO+M6lZ%T>$KZsww|56Nu5Ek~Qu?KM0|4*P!Hj)DTw87#
zdQ7zYI5&ay5Z2om)&$SUn*$4R3gPL(NG`}zcC#$N5aE=ls3^(mf&T9Z+F8@8v7r&J
zKUR9;kLf9o0s<`c+df&hu!su^ssePm209^!2|D~bi+%r|?uD2_s*42Z0o3|YI!zXI
z)a9w(-@k2r1RPM}#Nn}GVP()7<3PR#`i`?~NTqkpLHH3DothsS^(1(EG};vTxfarX
zf)Bgf$M(%r<AXLMNADPy1@1q5xVj<h0q59Yn~k<J>ZdE|Dd3ypppBvgnI6c?#n@#w
zNn@44Nn@z34;Bd=^HJ~l*(_uLBDyMs9Z^QgpgQ3>I>+|@4iE(04jJy8h;=jU1~H>Q
z?(v1_yLXhV$&>$587>}!Fc4aopd57zN%uJj^@IEO%b8Y$RlSsZbk=kQ3_0~VnQx0?
z&W4NCrY8cl98i97gJy*{JqsHIQTRJMIy#0mBRq!Xh^~)iT{heZxM-{`M(^!SfI3~%
zUczVqwsFeqf_VW;EpA)yCd$A`!0`*aI;xMk7xm?Eb_Co*73lU-&nDsA^J}PuLOPug
znO{vBdmx3LIeYfYr{^sjt=B9I(~9UIWMe#nqL#i(?}jgUbA|CTJK290F9Q$E7hU9R
zoFb~!CX1j>yUTj8+JJeB;Ax2QPUhS;`9m^O=L;aUCF$kW#kq-gVj7(*rza=lP^Pi>
zCtR+20e)4LIs!MvSNtrkW$sI`fP(Zm?p6yjn`Aj`2JN}IIrv@zT3tKVcw&gC(_DxE
zpZh)E?LGVS8MY(RmNqe)Ab`KbY{x-pq&j)wH_vuq=AMIlY$_-;DGzB(GxbNTN`^Ll
z0NFSfYO<hnpXS0f7gd17Z*fy<wZ*(90kT_bK?EBG?#^j8>HPd8!*%Fl0jkI0kr9V@
zf_HDN*-FJKB6O#gzRl)apg+Uu^QE9Z?0fvzHQ4oXEIE95`pXhZA5k4QAg&9iNV1-}
zWVw%*w}P|ZO$Xa9e5$h;+Ek_Fxb9Rh%E1_s+04|##-Cr#kAr_3tvB%QJS|-*HJzwZ
zPsl!8VD?^3lE8cL-lSd=Q>&znSAiR}GGR!}&x|PEKM`=3W&aEEXhjv-d~FIT`NMaa
zicd=u`Pq>pnLp1IO5{MzuZCojcFI9$V%P=BZJlZnyrEy#>DU;D(A`u}GAN8|1vEE3
z&2wY)Y!P=(Y&k3&8B;VrCr-=3>;~%>A@$Mb{Us)UKB9)k&x05YY`ABEx{Au|+1;Fz
zRu-BbgCY*mEnf5Wi`FY$+ha_-5$?hG6V06nJZa71)*x_$uT3D|^a5BpndJ}IvM9Q{
z*Ktua*PtkR`f=jk>Ep+*$u^c}=kGgm<Qo|`Vlyce78I=c#XxqIlniT9gi(byKaG@Y
zUY8hLm#Fn9X^-02rHs)5s&Cm1iP_bTC1W<8yW%(_3jM9uM9<935MeN;DabO2g#fbX
z+=#$KjzyoUQVaV2Vq&TRqu%dl^zOz=>#S+SdDF)C)9QYU6B9p|3%9$xgajqb%$xm0
z>+4?Qxi4YT-uf7pycP>^4*@U^a&^plsF<6_n`OI+=eQ2NAgM&#+D7IXCqM^r9_VA6
zb4O~M{Q}8PatG*a<Ea~8maJIYeP5*Ym8b@6IYe_8X1@7pzn@f5gyvc%ipvJ*2K>tD
zmz6{72j-TDNBQ=%>_7jWj0Xz-utMPuuVdJPBcry~;)nw2M9i3*Dw>~7A&M6<T_9G_
zV4X=vMP?}-o$NaF1KRRX(1A2#HMZ@d_9uyMDdyZq1jjgROkj;5tn%`Qb_Q&wO9ZjQ
zy@4*KyklOz<U(DDI+`wv4N}c)V9++&@oO4BCOJ1x8@_$QI@Mi+gONs_h^uV_bNWO!
zbobsNn|2=a6Vxf$$HgV)G7v^*+9g25HlEe1RvqS%hkNFVColocLMcKbm@W`7<VuL|
zUL8X|Dxi7Dt<wODF4=s~Bw<?+(HqdDo@L|FPCrdNi~)}x5eHb3Zceo$UoP^nez6a8
zrQ;kRRu9AhYGoawir$pg<LzS+nr@HMq=Obw=85}f@8P+C&bU3gY^JG4ww`Vt&~1p~
z?vT2KQXzu9!V1N_gm#v7I3zd|2Pt7;BAQl1zN?K@=49p7)kJAr3w=;vc5y$GY!SEU
z#m%>hhoLCM1OCOMglKi^<pGD#i`7Kj5Gv2Z8v3GqB|JHdnWBM;to+i`(?>h%jBky(
zns1EB`%;)S(7dnW^P_$Ka2h<Odq6J18+56M9y)Z0YFsfR!3|Z#OiNo^n>xE8?K~mE
z4r{zC4)eHhJ1V82|2uv;1X$cGCm}5hU`>JJ_}}Z-zre%9kwgL&g{*tf9Zvm?Y6NPb
zj_#||0gHy}!rH+a;YbYrmGT?}G_&t-{^7kZA`UviP&-iCdro&Ns{fS)3ZE<~5_Z4H
zrUxEoExW|ixkgBOokiwt$+}nGsi_-&4v)vmA{-@D##(q0^8Wb%;694o9fyrW412`|
zckRpOW~Yhu$rE9~ItzgDSB(=kXR0`7MeUq@K*OAZ==OZckuL(m?hji=VKuxAU01Zx
z#o9Lg<QGf&MQaD)9b<vSljUfT3BOZ3A~A0UL~k(~+dVMQB5Pg~FoCKe&-j*McS1_x
zXGh*BJcwABS(atPVVa#w_t&?-8g)2q4NENRLLa9w?_P9Od>x57m!UtE_kyf6nKttl
zckX)?Y^98-epY^`<lEXkViDgzudnbetCg7j%h@L1TLavJKID!JFuI8xUK7W#0@Q)k
z1<Fs~_4QSSt~A;iWZ-7@a$cv7INL(mb|anFfS4gzEBuJvaaeVffzivExMbbAMSswz
zyXz`7Fgcbb-97F7T7o7@I`{6}u@gJ$SdY@~aJT@|38NBUyBr=_fmdO{%R0NKzXKFx
z7`KUvZmn->kLQGJNli=IU2ke5mpXakoQzh#@vqB~1qd+N5`kgPDAZ|(WtB9SQZ5n4
z@m_d?O-3|+eEX(Jn>`xFx^bh)$keBOk&dpjNo7X5!FPW2tlzF=yXWf>XlvybX)&j@
z=(Bm8J9Zoe-CPTDg#~WTr%yM7AUOpNMZTy6jvKI}YlB|@e2HIa;BKeGhlFU5a>%G8
z7V)1ff@{~W3-EN>yEMeb3`SuW#mu{0y7aQRVSSc(+L}t0^XIqC1@x^X_BXH@0^w~i
zFqFA>i)ydm%*GaM{IT<LohE#Y1;V|btESzCKkZ9nP4)t=%$ZQs9fA2U)XVxHzyPWc
zWN&AaMA49CZsHX;9Mbpc5hIMv4vOuH9Z1V7fPKtI+>oWLE7w9tdt82Yn}LDB!t?hw
zPu)1BA)Gx{Vh{)8PWxyEc1`LPBGEM!RGb-X+dvFLfEB1MorH^6WKNyR8D%hD>O`JF
z08F9NV#HZ@WKT=}9`zjkX5=3|@1m^@r%2<^x@_weK#zxb1l-*s_YvdjFr$1%Nmj$u
zpHS~aOErU^u55c9a(bmg8N2t4T~NsnId~L=TYh?AKtS#l>6-Hi38~G`jTZ@B8NQ^%
z#8<k28Z99<TQBqTGig6{^5l`kY1GL%waHmeJ)H^h421M^_%h{9;{uTf0lMJ^chTHb
z0-=I)YU{v6TbZ;2G|B-_J85>+b*@BNKy?pmK7i7+y}e!0|6s?Ltu9OQQOb&|!CU1C
zx|W`D{o1v&L1CKoS$8;;W-awiJ&qi+&T%q12Q(Spy3`B!$(qqcpN_?4g$2a6S>5o=
zOVG}Ur`xxp;_q~w$Ih&Sp{MI%rgeDq(<)d1E!>K18d5(PzA?55)Uq~jD=cUkk&}}H
z!ycp}a68$cC>bnS0w&lEtS^dHV>c_<5$O8LNQd^*o<~sr={3U7)shOdVLT=JE75%*
zr!?KN{oPG!TEs;oBkV%%tVYJhw9co@QWMWwj<QiM_(g_Lz8|%<eTDkj#K9U~3hg}S
zB>DdQsh$L30a-9uFEOpAT6`VLN0ALar&d6()(hGHHw(+9k;W&%=d-3j`;LMJGrrOS
z8q;#_Vhi>>Ev6WI#$CI_SlCbaU4>kh)Q4Uaw2C}Wbdg`01fr1?)Z$g;khmSEdXeZX
zu(qx5_d8z@K4FHqoKro2W95csuk&3=q}Xq)z^k)v*%JTi6@NJ32O%>Ye)$r9o!C(;
zHiJ`7)hTP2FnfP{9aX6i3nmm=#d8!S2X^f`4kQY@r5ZdV(7a!gQAA)S`2i>_=-_i>
zXVH)X20YE#vknQ^zf@z?w8R50HTiP`k#P3!-=7Oyqn@IlgtCW@(#<xvg-|h_oiD(t
z9=4K}kZAtZXd3A$TsEHTK5m`Y73uop4tMWjiY2^TM>j5CyDb*pi<lxsMMZVQY(cnY
z!8II7+zKBrFgY*H;PEDI&F?j>Ta+Y*ZmeSAdNAh(G`4oUSF^w^qr`7}sBvD_=j^4s
z#EZE6w+Nkq)rX&_4R2Y#A$ewYmH|#aMBgT3eDE}OY&w8y47=(B$X4`CrHR7o75lSz
z@5!G&{S|<JyM7V4U7sR~STT?^1{roC+zjb2MB2Mh#pn>z2?zjcz?TZ9)j$!YCcnO5
zlZ2Z$Z8s^JAgEWgt}BTU*PYeorf{w6bISPg^GE~WfRcZPz*Yu67o2qUR-62GTjq@Y
zDpW3%9u4d|&X2)QlggdP-`5wGr>K9K)ha8MU?Mt)SPE-^G>=|xP4o&LR~k552Jwno
zYhAeTO#ezJYgSV-5GN3dm9`g<69XkO>|4sM`B&6n-$9laq7`~%!H5vSe&y%7tkTCD
zfC`-8?Z`9sYn7(FAx;$A<KX|w`2w+oh|`7Wb|B=;kZkYz`Kcqv&rS96k%&ji1&}vV
zp|W>Da`qH6ORHjoh4+ZS00DYJ&YK8us5=gDbDt#zGB6n{W)qg4SPURB6oAF6Ni$W7
zX>0NvZ!sNtnL}=bV4Q2)eoelNL7?aR_a`%~n$%C+VyXhC0|#9SJOn+cf@+b#<6(UK
zHgjzXd~6SNq9V>#Z*`JXPy-~KytCeIAg3!GP9a3SZg4Jy2nA2)I5S~KU?RxidelI)
zUM%8ARQKth>824^N=!DExDOV@ASPc1<xJ8wFc$`i0e4^z2zN5?teZA<jhzh%%l5|>
zG90<;15%85#>o}2=ob_%09)Ko87w&V9~J7jIcO+S_7EEjhM;+)QmQ-X^q*Z_L39t~
zIXd6x2w|{dF+uTShlw9<k7qJ*W&tPuc$k^mi~DCzMtux7pBkl8)9BzVBK!9(tqc)5
zNMq7*fof&}>1(aXv+-L97EQA`=L<cq!0Gp9HRoc=Jpsy_)g}HMHm5%ie?tJw@(JM;
z6>Zo|+)W*sE-YN<-e|)2Q#Pw#=+Tgk?FM?kIoqxTFu<~cScnZduqvJ7`w@)<3o0Oh
zSbYYqWYmmlRrF5-%_)OGYur|p@HvpDx6V@~?);t(c##CHL^{TI3Jcf84TXujR8U(n
zI+kwz1bT0AfVyxWR~2JqlqC@j#^_VP9+{&@uQo-tgS}|5asg*AT)$vexnHV>rJp8V
zi1i`bB2#jx&CQoxQmkz1(Ow1n9*e-*SywFWC}3J^ZzW`4qkJSx8Kl-?FH};U!{fF5
zOdY`7R0b)mwD}Gls@i<$;*EA-q%rrhP?3}&#J0^~9Qgb9VO-L^!1ZqTQb_GUM3$MG
znd~3y?dmcOlk#ljob78ZXyQ8m@^+KK2{{>TyCF=B+a-xxhMP&C>pzG<7o7>^+&n8B
z*v8c|KjJH)?+btZJNjM`oB?fSawNm3bnQ)W$a;LJEJ%SyTw|kOl5_|siy&QQUwi?q
z8)B{kDaOh?eL1A{@pKM3tw|*Hkje&Y+SxT63JXGSG@*OqVl1>eXS;5IFM#D(?|?yJ
zzDDUGu#31%d0eLfXq~UvK~^q-4+ZRdqyk=IzgUb9v=HAdtz(G6KuS(Zx9gBWkort_
z6E%Btb&+#rihhAueF;A<14Od7ssCqXqyZF}YNTJrn<dQIs#(BsQ}8E)+s$sWHt{Pe
zDk4-1mSf(`rBb9^l6$bHpPGPJ$9~KcTJ92|-<>Gx@T(hn#<`3%Db%d@FqJ^-2k{7y
z<U{{whZYzzf7FMGFi#z6rj_ZeU%#Fd55{7H;I~jUxEr2?+hAa}cLQbMBK)`u-Ys^C
zSewF08FZFZq2MFMpk9HS+P|z3aQr8LvmkCW@P@%jQiC~N>?BUml%#7RyAZx;|K7ba
zFjx~xs3y}&2b#%gfEI^@S<IUSNNIutNXCPm4ZM;8G*er%V11Yb(mki0t(voy4Gj&6
z^&C#i&aSQ)ptl5+Bt7*hk)R~#)f0)sbK4t=yo<b0@Qij~Q;`%3HumK~jbtuh>76@w
z5}ow;i4(O2?k=$6ws3m7zq3dwpvFXcX`_Z-u)Poel-sj=cdX{?iwATv6G?QzY`S0i
zBI?8;SG|N;;u(Ss7g9Ws+7ukb)kT{?M4khlA>9ru*j<le`KUwHw+_9~OGT`Thw4hm
z<nl@4&4(Z$JWcI<j92bOrip=+@$=yrnGA4oz1Yq*2%!uN8XcROl(d)|78o-i>|%yI
z-TGdaV(gl9uS7PG<_*GHqAiGs2^kJuiiBez8WfNtZQMT|ydyHsOO(3Zq}UMqa>#vg
z<V(x`e>y)Rv<kTl=+cH!c3nv?TEFKE7qSX1MyVd9csjWh7*+^2YH)U7*;AqQLQao!
zq*tT#&cw!gMriKo`2tZhXg!3*rH&w^lnt*eqfn+jm^y^Nom>1yuZz4Rh^cVy6k_}*
z`ENDnwFDxaV8ckxIw77z$LssB?(N-d|7HO^QeJ2QGo+sRsrsUP{#~xcn8*I*XRzqA
z8cLb7BJrn<o=4te3~FA=ZQ%3$sEZw`7qU|ubSz$fH?iN1y*{}P{_2Z>2NzuEJ=hu~
z!`xi=b}Q1*W4bPQ)$|K*7xVcpuRxE>tF6}1+<s)P&Z4|I1I!K|SP1}yP0_3~bp$|8
z8|C{ZnH$Z-d=9+_T|h(t##SDFcl&!Hn}EqTNola)y6h_{&02r{J{FPFS;Qn#*eZm+
z{Cl<oL{LuJ_~h$S_Ov`l)+~S2ujZ&ef-A<mt4?4&jlR20`ChSLJG^l0%J9$Ed2MxI
zWJa<1u@^rou#JBsUL9CStn>Drc~=Jz^%!<_2VSs5#9bI9vy|epHiGn(UQk9*Vy#@A
zD{jn`-Re26*Mjt`zhzBW#-&f|iP8;I*;VoND8li(hhR)f%ksfkS>4=o7oXs3!Q*vR
zXKu*f`y;>aXq0XHyH@t~+_$Gm(`v*~RI>|T|J}Q~Fx41~|3$O#e`Jpv85zTvKSXU|
z_3hQ$f3x{v^}cU9L7~LJw3?lC^XAPf?hAdwSDUu|r;U9jDlCg!Pa*3dcEA}5kE<5&
z2mpYQ11}=shJA{o)wd}-x>(Usg?rit4Ig4E{ynw>uQMPB5~Bu*+NuI=9lgL^`di3u
z3cvU7R>;C4_<!_7|3|O+ztLj--<Ko)VSN4n#<Kg-+=k;{UH}q&@b!PgZ@aM7D3rvD
zkXs0i5uQ6X|JtQ(NLh7yzO*gwM!?tOvM*y#tFWZ7q{OnUV|nTGqhQC53&)lmcy`-u
z(d)0=x-84yUQ_+Bn(MEnr=suNJg`gc*p8)t?Os~_ac1tfY@uapns{D=f^$Iq$j}l0
z72*yR!D)Gqet&}EC<oJ5Mn5uryh@B;^ceKCYM3S@wqO{JDxbQB7QQzxyyr$D#XP6d
zk^Q^WrV|snNp*0(1#+O96EXL`xUEA>z1+xCUm@@gj(32cltEMH__Q5HaMkMd<e&of
z?v}i)Hvg&&3k_hdiP<Cg?B=qS=z^dy<CX+>4)-1!=Df3<LOJ*J0a;qVkAiN1P&be@
z2Y8g2H;`LqGYe}wi_C740wstss)(^2^(CN35d&>P=Qb}xk08tF4*a<9JwFrii2-|z
zU1<vyu~7$p^K9kAn;Djn_FD=w-~s?&N!M&0eh*DF36;}G9bA2<i6!6lEywD&<EW<5
zk{~sR<?!rP$Hr6mGd|K3k>@fvrsYleb2Jus)GdtH+l8z0P1?VnM4A8a=SB=p(4T94
z3d&-c>Ijr2!*y*bE_OA@BZDiyXk!vQy<7(te(C2wd?xxW&P`TO4NO{B;6=}!d`QOO
z<8{EhgMxyBmFP;Pym4Q6`{8Fm*SRAV;sw?m;DEq@O<Ettk5Bzg#`LPh=Q5^5(ryg>
zU25AJymZwGGMzgXluy9dltz2^UHG{5&n=>0ibZo&YlkxwW~jxFIFPR?=hFX7#3CD$
zi=ZkA^*B}xS10A}#)V0hLa&Xi|0A{L2&94UTNfV8V+W8Y;73vC6nKa#$?@in(sdY5
zmhFFj{QeP}7Ssrao1q&nbYvHdO_cAPXsJ?;KJ$dTwsY(xK5pCj=VKsyVo_~!Lp5Ee
z=hE_o8Ko_SYs<obO+0uKtOZqM)ykEhe5&wQENlMUA{LOJNH1k#e;!Qd&&Mz{oF*d5
zG)TcDcxQz;NFT(xiJRJSqu@u)AVSWuV~mbICvLP{KAbuL8s}pvx%4*;G~w&&OXPE-
zY%SJ{CPo8*Whu7J_#wV7e85=(g6?CMS&Zyc$VKEUQ-+NAe}DIXaSS$Uf5pC`jJZJF
zKs!;jarl=nXN@mi;^O2y5f>K+OM@!$L3MOEp<}zht;hhZR4um2W4P><II$Ox8XhH+
z_VXT$Z79--ibd?xAgIFY<UuBZXqPSz7LdgfD(_OpT72WTFyBiblTLT^gleam+$TMG
zC;~Q)KESj5{hch<t5<8&%~SxTjzPCpId|^dF=pJ2vgj@e27g^f-AB}Q(t`K(oVo^n
za@3kE+N(UHfZyOoc)AVo;SS2eZuP1My+DE&7BmV@LNTEtk3rACWhn0?R!4hHjrAu$
z%aQg+6rwm6OL*f5k(9#%0>l6h_Zu<56K<0jL&*sPfLp3)J}(<1c5a9X^K?j@>>PP8
z@sM5uYg8KW<D5iiH4Xs?gTsT0HnjaacOC;_L4$$~y8QrgV>D&a)nODP#(tbTAfut@
z^n|n|g3{YA<)KM>fgnETyZHyOtXub-#3#t7q-`xZ=hv1~r%n+^2q`J?1)}yPUVJ}H
zia1yyhLF_oq?v=WAYIsp4<DA0vfklYh`4<F5I3Q+cHTMu{2Uj{st;zO-gEh7nlJFz
zfm1o{x3>BkmHOq}SCB%-zJCdE#2SG&lW}v)ee$&wyZN-7+#wX8mrlyc>WGzqmh4I`
zf+oTi;0t8AHPUk|xTzzCO5Fl>D_?d{0Bp}@6p0U?KD|8YDK-Lc&+_Ij`Gs3YUk+Tf
zWVz;Dp3}C<c#@mPKRFDkU6T)}!rXJBLD$%ujg3tPF4~qcR%ithu+z)P$lQv4^D!8<
zkB-bHYX^7kgDaE0XF51!P}a(pdPDx^@wCSBECE%L?-i+$8uI;fWo6L9yPm1yR^OnT
zNMhw;&~Q4VFmB`|Za_W&v7VQ>0aZDCF&5P?<*v9*Az!d&KpM>|b%+6}cK6;r6?`IT
zE&_Sf2_`VB0+>HkTw0uhvWiDEpDhgSWzL0bS3swH5)m<wT7e3`-+mrGC6)~vIvx$(
zfUtd?S?cQ=k>>AC9K=oyT2sTyz>}dCG!2Wl8?k056c(^vjzr?+YW%YmHh|KFM*7w`
z9J?fBzyhicd~5Bggk`BsHI^r4A7XySR$)|#4<h2mFytR|J;!HVW|;w5WU)hV3mzDR
z!-QUlwc8Dvqt2;bFNUxLFp`0^p=H$TCxF}_cJL}p<P%h*S0rxsk6^HFUVSpXS?l_}
zg+aWV@S{v9n>T3XhP~PMmEm-?&md-mg_e$fUcQNeL?*Vw?~R+lI*D6#S+7Yh!|L<}
zAJYgOD^M9Edk5@6LUS-T=UR!FeA!Gp1dmsJx*ph09?w>6P#NF^UkD`MT*3aH?Paa`
zpUY+A#^4yTW^V@AeBfA_j5B6(?<*6UJV6hxUfs+|G=Vp>yMz6s_K){9bfn*ZvXnSE
zt6)Be>2KMSBQ-m!LL1^G=P(ey)N@L2)4#hqf=njqD9myqoR@_TglOH&;KRmuo4mTp
zLrrrBp&@GE@-1WSUm!jZS|w(>nJ3WzTm{V&dH{7AhtUv%NNek}W&g}fE(U_Rw-B^W
z0~qONSin(BFGNJ?hl|>ZgLq)-OZm~T!-g<f>NpSRqQ{$P0Ea!z6V`5#78mevz3?SA
z7JBAw<kox!K_dZNL@rKeAiW?K*Ncun^CH{&vJtBi!si3TkHJj8w3}6M;Tvfe!TL_U
z7v}y#zrc+>OACA)Lc!Vbt+$>d#?TtnD@+pfGST8I3DJ}~ef|1%r{XmtKi$qeH{47(
zLUMi(_OUMUaW+-%J11>GP|l(2i1|xjg8Ru%0uyq`rye`QA^5N1C+IfvZ|VbUWEBpp
zT*!cfP?Qy5IyTRU>sbsy?oi2A2J{~cgA<8M2woSpr62sc>_V_!ME4UPu+5Q1h_7!{
z0VEpgdXTCWtQSZ_uR?R3oltREIN{`A78uDb7#;6RgwyMbqp;=<?}btLvjY_oG@QzK
zva%CLAiDf)bcWR9z2?&~=r*NR@V7VDefs=)*yKpSOYDYNY;U)Y`yyejx%4~pGhNak
zAGMPE|84f}FU;N%Vh?KCU8-T=`NTp8Zjy*c;?99#VTni@YDwB^L<}WTAHw#x1gF=q
z8h+Akx;xc{fm*!j9{g-quU5Y}wOSIqEpP876O$egf=;=E$QtI~UOlII0w4>wXgr0#
z-weSvMBF8LcO3MYc%7_d5Y$OX842hFCuA%<eHR;LK?H<ZcahEc&+UZ=+$_@K_=p^U
z1M^H?Viq_a-PK<VGM)RiX}je~(?*858ZGl?EiC_|@bh<r!Tr>_5upzTQF30DeIpfZ
z76;ITmCUB&fuY>-x7|G;UM^dlCV@64$!M@pU~bRw_EM=-a_|6k8XgR@Tt4&%S4xcd
z)*EfZVL9~0ea?@9gX4%<);JvD<!58b!XT_8IQu~gs9ILEi9m_hCzfenHUVPEL-Vp6
z>3(h+Bep%{u1mSz*kxn|(&xzW5HJ@7mXf9-(ne40+xiHZ5bUb#n`Ww`WSL3R{&<T^
zWK)BU2H6@4<Hi_cxb$)|H+`rCW>_j^O}gW~vY=~5jtZd1moWXt+l%=H5rWXm)J&gy
z_G||}Q~>4$V;TshDHe1(7wiJyq>1oLI8ww>I%2FUWEGFY&fesqk!Xl<n+}3y3SuAO
z-HH56&SC-8`xt8kmikjNGNkArT`0uOOOtXz7s+Ux->IR6IrJxU`06H+mA&WZ0(+9t
zH3e>-N=ytiF$@I`EGqX9%zdMbJx?tCgqc;5#?}&O(RfU@&)-k~0IPKP(#>GcCdVb1
zm{2>N;p(b~s6KNf^&L_7L!>-&%fuEY_`yQ8vd98gA%}@Yc+Xv7yVb7TgqoGO{Q>6B
zA?ZiTXih-U(IEy4>>S;?j)f<_v9Rj6AZrs`1S?1sXzpGg4zJ(2d>B-8P_)<<uL%)#
zpu*n#T}>Lyb1fc&oPm`Rxqfpuk^=GhlHzS@q*9tEqkZAB*MtD$juw9W{P{dGJz|yP
z(2r9%vgrVD%|692>=z&6$+v3BRU=F%e!;`7j{qIC>AC$1Hm!kWot)4{%2RUw63(r{
zacgIQBZ$@r2k4k9`gh{M&|b!1C~neaW{LC|zEe0q=dR~aE|iNNN~ive`koMrWMu?3
zR`_Q05Y17^@$TQV=M;Gh9Bwm;^?Lwk0sc9IiP1B0b9UkB`98klZ}Ud%f%C%C%a=e_
zd_CB`W;6d$;?~#kU#R`@*#(P*BJ01uBLnnbdNiIF_9Y^$k~)De4<YK+P3+iSC>+%3
zvL^>)u=qMVZexu4cEa@z<EBQ$)NiwE7BDPrfk69je?t6Jz)#d^a)OfucH`s`#BPea
zCXOM(q93kVk;R4T3tw;Q7RZ4AqTfCL_*CLb>a}avQp+HDu68^9Q5HUTmx(#PAozSX
zj((*asVefG*Og8oSr5(FtKBYBW^QOEUFK9@GMpS+s)TLJ;zks?4N4LZE!=^e88CWm
zdUKmHe*~2zY0N}T6AME<S`$M6k6DmoF~4yr4M}DqLhxl66vXU*o<7FhPy=WOQ&P{X
zexwhaHg{>!{n+141@2k;?)^d}&_mWE2i1@g5?+INv|1ptS2gE2uySx1d_A%3+{eEY
zc0M47Qb6b~MDKN6sCBL!LXZl&D_LMKfu0|Sf{u8mp%$ooJy9Nw*SN|17FXx>u{1`7
zh;g8-`bKQd;AnK45D?{9`ShEOn}{2)AjCE2<f&6AUf)(9fiK3oky#^dwGZNUj)Xxz
z4`>2j|E;J{5Siogj&8r-t`I^Ku2pNkR;k~16?j94VEgzn%ZJj^UZ9RDbPyAZDGb3U
z&}?3haOyF}W(hGE>=kcYCEm1o4At^>SabH(kCB25D_d$S-%BWly{N-UXDAx&xsvn?
zw9(mCg@;73l?;b7>GmQAZ;{hcz&I(8zPLLZ_N{X3rmR|dQE2Hkbnl-aGNdZt`WhT#
z1&msUs+t_P79A5q8p(+CiYFiq4&;cD`H52LQFvND@%u__51sfVvP$pT>Nz56`f$90
zV|Eibi>zuIO~`9zLbpjCi30sHT$Qu?_U#kz=lIBn_V^l}tmYTMt_K7JXdPpa8{@2V
zIya)XL%7mN?#!750UFt*Glx%(*oA2Qm@J0e{HYe?oKYOTNKRy#f^`R#C%Sw{lF~+_
zJcn$z=tWkkCEu7|Z&~~n&c6h}kRI6R#<;W?_y>Tabt9Hhed%7W;q71_S7l(q_X5j*
z$tVECIDq4zl>83iiiDot=ea22w4!1i_~^(kCsyR<nj{^O80mu%uCV2iBi1AscG0j(
z_X3=M%<u#tVc&D<K@-3v>AWm|dZ^fNn$(l1PKe~KieemCfgCu8b}L0pm<kR#P3@(^
z1`o>sX>svl<~X|Zc;c;VWA9e;UB-EXXXND%KIqsWJ-HT0@us@EdLaGXO9)q^4R#n{
zMzcfZWTNz&c&M`z+&geEP&JHb1=dCA_41q<DwSd@d-BCg9q#kK5hK<;%#x!I&<q$c
zD~9Tpa5U&VkA_P@k$7WCUWfY@WpQtE-avJzO<`P{*R%pa@Mr6wCr_fkefxGb^wY98
z?6+&reI@=3Y&zC&Lw6!QP|00AZULQF(Wd+|`aI!2>Z2Kv_9h%`cu#2>E$&d?!R*`S
z4yIK~30P^M`>9~~^Fd;AAm=%d1qc~BlQ_+rxM5OD_|xJp<`}DC)j~%l3oC?!6&f5x
zH986ev5!1q(``sOapFV<^ldb*YokEsZBmatcZ}E@NGIW|S)8p`&EY-M9yGL=D<2R2
z09`y9l3+`y)0~`~<iss93m6+!;5l-F5B}90tTm}~BK(wQeiDwH1yo$QeQpZ}b0k00
z30Sv>uN1us0<rUnpUM9WS~HNLEE~H}w^-#4!t_9$ChaL`MmU2&q7E{Qm?aSe%+bF~
zwIb&oVb5FGjLnyQ{#=cgib36bgBuUxkXElRkI`6x6H{&7nqh32___I%oV~q$a)#($
z#jsd_<SR9mK<)@(34QehaN$5l`lCaOxx~n;kt1*Q@?ERwg&;%BfwM!<OFfX=%Zt-K
zk6~;`9wi%W4TqF2IDu-&FtsQUmaT9P)ZXbe%)DNgcLx4oa;^t)Qow6S4%4jHcTa-B
zvip&{#+u@<SQ*5IhaMh6uo2UUb$0u5$=Olb746vMSEtXOtwG^Zg98ajYd-d=I_gTo
z+ri354lLD1S83Be-<%Hr!722+!oC~=yHh-z(p<O~6a$hPhz|t`rWfT;1uc8ee1z;R
z!^?Q-EjaVmI+wmzE$;j&tW*_Jc;G=vGxfgJD$_VN@fsSpIvowrxH4=11%44Ov_|oH
zc9Ft^j-%bq;k8u}KS?J)1_YMG1*(T|J2=*)7XO^FpCM;BJ<-rt#+>67_dsKc#q0Kb
zdmRN@zoYgdy6j=JS9r>Q43!2=s9rY;v3)-dK!NM(+`loFI5|Uubk7WlA2@>L_nJ=Y
zCTjo`Mrfwcbjv2tX)v0IwbTQX<6-JLq{{QL7szpp#hDNkYGb5LalVpWt|#1=SG>%X
zl}B3L2SsOQnjL8h1l*L)E?u^a=nQNBnZQKp?bYkQ>snqgbG!a_?U(7j>S@VbAI#Se
zg&f3bZuM}V^g9}WR3#^tA^MkFNdx8Apyp65^78a2%Y5IhWl+WhcjxfnxPjNz;RgN9
zIW)tw;36EfyitO#Wo+26;Xama6qsr2HnPg^gAmU$vUm0cPq3MA1A1f7kIa3y2z{<T
zGywXh&sn6^HLA1HV(7<*wWY|*rpaQ!4(X$_23mwcEi>XDK~=Z~t0dDgAA`sST}=QH
zO<}ssf3Oj*JGe9t1F;$5UA4Bjqbj)%(@tK$h;&8NLDFPbq9D?8i7we}SGv6f$x@=H
ztqpLm(=ic?Wh6Edl}=fdLlHO(8N^IlA-0W4BQN}W)vko?ibC*WzfTUiDZ%RGX5b_-
zaheyJS0ZMEu@B#M9@>htOSj?t;_5?diunsCeyol%qfrCfR6u+A;*Z(T7!l~d?sjTw
zsxOvEaA75WoMUNICydTA!UbDtpsOgRA3)VuH`npg@C3}s<RItH9wQt+B7;gup{*X)
zE*_3zye-T>)FKtq73DuQyw9b}*3#1Q7;_am9H_-lQ9=gA%%LaHYEp8!TjKHjU)v2z
zEZ&>HH(&ocg!R3}6%llb((JrppZAg!9(v?}!y*Jd2T2oV&5D)D+veAyHRN43*m|D%
zuTrC5Pgs{e9@&hW`IJdtwhzS0%$sFSdA01UsI)+WF|t_T^k}JzC+w~%C1j|#qQv~&
zEihDi#jyc9+!sQrf9nZ!mH3jb;XV;l%Hc~N|Ca9<lEFKM`K4i*uroRiTt-i?bFL?B
zkc!iv<-a<@NoNEzw0^ij0J%g*bddhxHQl@poA@Qiz8MAjU#V5o$niA5z?PsfJ>Sa!
z$fk1b&pvXV9(6DY9l0`qO=V-m6>ViR=I)AeoN7h-#jo_CH(dl-D_7T(7^^|Y$1DoL
zIlD-);bTA{uKv@*K}VHr$)PZ7JK&r4wZp=cV6ee1_0<&O>)GBv11oA!`mce_(=6w|
zUGLEzpO9=XRtN4^XKHQ!shLX^n^y8#D}RAIRULi4;b=`-pzg4KAq>@ks~`yS{^mo+
z!POmXRvr467odUB<~}G%nv~eLcfjyePi)Bu1~_Smm;y;38;s5wPQB{bpkLZIBav0o
z9wqht!hYK#U~+b>7u@@Pa#yeEBtoG)eE1yguta|%IS;wShQQN~j*00!YaHT1Ed0df
ziOD+$M=Q`vIy|G~SV_RdkIAH&gN_tAe0gS0%wKBe_X)zn5!qxnn(fabn9;Cjh;Mpy
z)DiLH47fG>r^iE~{168p7A$)Kvp1l-O_;R}&YrCF!`TxZqf(OuPk>mE5?92=Bgh*F
z%e|!(JDqS!_sNOFQou|x0H`|2-oQdZ7J8d+J@=mNH6X{V=CflkLLE!DyDS}Lan;b3
z{PYVg5D5Fxx|tk^esCXlIJSYlOhk54Si#J$ikI`-CQ860k<Ju*eSCb}9lhS;Py-0a
zF5l{q=ZJt{&k?~t-Jpji2?IZk%@Al@)~4v-;EpNEk_79#eqo=RbW%s+#Ob|Cp37JL
z5UL;M)aO)p@Y5{MeX%h??!ARNsJxRnpv*2e1%lJ6RjaD;9*mEi{@}c2NIt|`AP3{L
z*q8C#y?<Xw(z^w@Rp6l3s~3XT<>%*nZYAt@2dQlrJ2-GRE6oVH$5sQ6p_^u*aFnuP
z<+#MoV^)Ie#=BVGmf08Dcq=%t<9wK6wIHY<jaE1vqhd2SzY@406?%VSi{dcHC<s&~
z5NcZdMECFigx0o#6Z6Tf7G7bpXwGzc`dK&n3~QFg6p4-ej0~h<s?$hs{CAuxbJkl4
z`?n&hP*;z28?-25^o^Ou;nfidd^Rboq|h;^Z*a&#J&;@Lq8qn{#?pfQNYW)3=P-5)
zxV$mLuzyNS=`VQ;)kxRN8Xm&HsXI7^0BCJwbP|;7R&v6S#2DuOR}TRWYuhqXQjQj)
z5kPSois*3f6LqDfQ%uPq$i5j$@a6x$rZh!PSl&Jb&us?H?H~4pzf-Q1a13pEz2$Yt
z$;7=mezi>3U3pUM=M)Q4nYJDI3T*HsdYlG}C<=t_f&TI{Keq^E&+LGxNf;~~2Xaw6
z25&vpB|RTJ>jsd@Zm`I;^NB{7RPFavL=1T%Fpc$Mk^9u&Y<RMo2j^uUcWJ#6(<BUn
z6M8ERr@YITl@67<=m8Rt7K-X!*CC~m|F%K#kDEBLGWkQ*e0hCW6^!D<E-dG5um(MA
zd<#tbwqk7G?Hjow@LyaYr{*$vCO+T*5QTAa4pdAjHn#7XAl`6`>1?|WM}$g=ak3>8
z(6jI8O{!?5emaU_lmRpZ3Ua6rA;76t=-r_+Uc3kpbp3q+`C0`9r<m*SXE=W4qP7Ug
zS&O>FPZ_g#A~&2atI$NMHgc*d>b}RQ3c;mV{CIzdnx=r63$(dcGWD698Zb^8!-)_a
zP3-T{Smbf2!LZPWUAK`A)w(2|TC`orGyHLe3^1^$<?6L->%c_+v>%$vC*2{me)Lt{
zoU*GgbcDayryj9zOs<HuQMrbLSbgNRWYJJa&H_}i=TsljRramj{X^(~u=gflJ?H)Z
zuQ6k0Zo3(~l)(%pd$v*-h7dx?(qbpHY157wJ7p|IQiwtnqJ3c^YqnI{kxDzQD%E+s
zzTEf!_dDl0*ZH06oa=Xf*L7~!b>DY)%lG^Fem?K_>-~B?U(e_BrT<8|tltcvz);O7
zaVo?9>w=upVK#ewE93+#i)}EY6P#cgqw2fiD3%~t+b*5}ZGE)V%wg5SE`Hy}g@lP4
z72>266<IN{r(IaV+o>*2sbxp71a;S}m|nASUGvprAFAyYN)Dz1G*qL%NyblDXO%sN
zXGnnSy49=SfxCvicx(*b{?^(N@({T*<;+ph<7m4*y3t4*(+hupp9Tf(^@%fU>IR&5
zp?M>`nzv<3xJX}+EG?o!S93P_z=Ol%b8>Kk8YfYrx4Ky2E)bWez(qr1Am2o?hGu1a
zs!Gz02$d&ab@dT1_LN6Mf7Fi*LIN@^mK!3!7Ps_Kdn#E7_SHVE|CZMlQBngE)vC9L
zbsb(9lbQ<DAiXO*OK-gRg}p%EFq||ZpqE+rt@*v;hrd7#bP`#@iUu<AA&r(Nx&`^H
zUR|TsL0jFtw{^{TzGF9M|8k=8(-qZ!#lhb%<)8c<2g`7^uD=uvOmLB&TZG3~;Xi!P
zYx?Qd(J4^<2Pa3#1~zisyqR+4?sJtsP~jbLe!6B%`7cn%Ri4j9)>lp3Skv^gC<Y%6
zyL+FcZ?9^wh7SjgH0saU#!hgDP*rSk?wxUSz2`a0>GP?TUtp(HlDanTl?wGv57&vA
zlM&>ew9CHg)K>S{Nj0+Y;qCfY^hG%GIyy!?nIqQD+9H-g!?VWPmb^4j8!%MSYFw&w
zNb+I#yYtcRi!^vnZA8625`)D-rmBlB4ZkeoIlP*#3VmM)O@*<Pr(>3<*jlWkKup^h
z@K%rj%AC8Kjh8H0;sts})p$WNVUOUCJ$H(IjUF|~3)Ygk$uAW|k<E>ksSvp#%|EW#
z5|dZd!DqYAp5@(Wj<yeWK6>O{nz+&F)(=^Kdt*?ZcJ#DM`MVb<I!^t0(c#^xm&V~6
zqT8TD5P%>wYt^7P0iUL9SN+aT&uQ30`{LaLCY}w;vOSR(i9E?-OAIN*L(|eCoc``N
z)k3d+adDw4ack6*Bx|b&+~^Zc9aJW4x^Oc1iJ$7%C*i1MQCWC2cq*CC6Ip?_sq>Hm
z6-&<uuhu@@P>($o8+2<Wkfhqk9HMOA(myL^25s+a*ATO$dx&wIOZ$DVjqVTLa(Hv|
zBX2aw;%k=_3DfI-^{$z7)SnbQ1nPQQQMBZg0t&?2J~Ah5;aWA41fgXKc_jFc8=MsD
zY0!x7q{ejEV>)S5r0W$Q3MHn9TGw*%>P%yU#!vNcx*k5$r$D7IJ(dZDV+#sYx^&4q
z?9!&^2d|BFoq{_Bs`p(~X*6$-#1+*?6MR+4cBbS1bB+W^x+abES%sQr(g;w6*_nMr
z=#LZghW%imzTo`aTk~VQ2kcFIF?$`!q{c%VtL#63uq?To7YT%s_zr~_Be5+nO1}|E
zlj}SwoXDz}DP!g1>-46)yj}!mu+~YB2Ea745eW@h0X&Vjb2n#qGaWs!vf921ENFG}
zcr`UW(dMmRzup#~ulqyUCuw;V4adgYnb|gMkJY-f$&1M@;W=*VOrr?cIPH<QdQx#i
zhN0oY3#dCXtX?jBWy!@Q_4s`)RkK;-Fg$3ijlnYF5mu*6cb#}twcyG0#}!~i$3w&7
zm$p<}`)K&={;)UF^OmZqi$9-#m<CnzLcvHyoCnA09nnp|!ou`lM2L>I^<<e23Qyk&
z&YK%INq5jMXR7_{l0l0-?JHa@<}E@kwktI@a{I{;U&^?;vpL(RMdvSdxlJ1FwmZ4P
ze;`%L2d^$=X3v5-SYqq)dx?EI=ci2FpkJu)<)JY3FWsD7uk-0h^K^^6S4H1g+!@~X
zp^es(%Dx4WU1Ot-$IZobc|DbXgo{1lX9MZGKmr06;Bj&uOtT)q*{X4J=-<XU)nqpc
zUue!KVMK`mmWWcQ4Lr7MhNv!8udCO5(BqvHW~W!LS+mAdH6UnwZs3xyvPU+U#i;9E
zf(>X{zrK9`{<gdBo_G8^c~$h~ZyMn(`8QDTZN7XrcMiqJ>{|OD4%S<JRW!-ID&jci
zY@Y6AHLGhDfkhOa36<=DOtwnq5O5%feKaN03$jsherHu$q?>^>(eF17DiyU6NnGzx
z3^A7oOqcf|e<A2nVJt0=f{bw`gA2;r8D}LNwHmW8`Rwnf`WaabnfPY(%F)`lH`wcS
zUO9Ttmc-U}6Hl<F?HsNW@$cnnT;N6TS9^UN<oegEUt9QS<~=xkd&!FVrzf|%bLWnT
z!UfkzYW`*z9?LKzAyBypf%joUPUS;V6p`IHlROyJ{&n=|nhQQei|EI@E339NtqxV*
z8uv$ppct3TEz^MLtBG?B)i3LAC3jw@suxrFNAg6C)(<dNUmlsF4FSi7Y<pSRMjoHP
z{!L8p*L&TTd|CRq0S#^ApK@5{TRYD2b#01xs2cXzztA`ZPUx;k{h?l)5|T>~H#zm0
zvEq=G?WtjAemc6ksZ<~##^1GSWk0ITh=ZMiBi7yBa$s>{{fAF`7ALyyPR)!$V{KSw
zXXAK1(3!o}K2Bw~Zw1ttmapy)`AR)tQbT>GGug8o6LqCK&N=H(?okBd`3w%gCES{Z
zr7jFYVtA2g=b`+pXxQ=N$HfUYzRMmmjdyr3EW$}u1f#+&8!{yE^+m71h8jajE%Gi)
z#Ifk^7jW)W@PlJoL*5b#JpPz(9>X<7gyLEQyyckb85roWDClBDy)tc=)+J;5x;`!&
z_-<>~QQFL&?wx8^GlXE**~xRtQtnvXFnb6*rP)*4#j<+3xxViAe>a<;xQj35Fv+ho
zaYeFham;caYI%dO$@vR^yYuVqjVWnAQ@r%!^hmz+>!S}|QPqtGmwztYkdxQr$hNcR
z@6fSny>&qI3>IDL>l<l$9l-~S?kD#)_mAoNC(Pa~aFw1~4%cPt2uY8&F0F2JTc`m7
zzejrSEYu~(v^=&Q5O<PkPOKsKOI%F^V}pa2=hd^M)U++9xVSAfmh;odM>XGRgtQu^
zrlw}tbf|U*_x&RKHX<|_hos!mOAKa@n22~U<L3S3qxUcu==^e_rha)KiKl(*mjt6~
zC#L~EOYqw*d}4}Kv<QIR>mNQ$*3}Pb-Maqb#UA>_-XsWW&EKwEm1<~rGX7Dluz5yj
zUh>j%42>_QrQwxwW0~M55tG~j%UXVSH9e>0=tYNH+eI&;ObXCVj((J!wAHW_Jk0yt
zfPnmfGhHXj=HU)sF|eaBZ6Dl>UTEB9yEVV09iH0n>*|>?<Eiu6-$%t>^u6{9A>8=}
z4>PPgdl-8$%I!PPU1w&ye8-)D|5#Jvx^B&yJclEfPi8^{@NanJ^KGt<I4dP-8#p>l
z4d9ie+djMYiSNXmH4O?^k{mLe!e@(}#Mjin&B<Wnxa?f*MA%R>r1&tx+s?S{6LTRS
zMw+9PRQ~R2P*9N>rg3YQt#kJcdv5;Q$cD5b3ojm#bMJ}Y)nGU?>d_ePTuTl41LKu<
z!+*pNX|Y)Jn>a_w>ap*(o|0?nt3HbwmYqu2F#olCswh(<%Q?BKHu$C$EIhb(>8Ibf
z2e!c4H)sBElkPjQbj&SDC9eC2oitF78P0$f>jR#u94YGwHC?BD3iH+vX}Koy6B&$>
z(K2W2kYGdm8yCm}J-T%@<iydx%hC7(9B$W=-a%deZf~7xV0ECd^e1R<6%8MH0ro<0
zx;env&^~#500`W1pK^!>L;v_?o&M0uBq3Z{J979zS>0&T^%>Q)B*92Z7b5|m$5S!M
z0<=uJ(Vk=DU3J412&uEv!!7-ioC)g@F78WSZ)n_2`|h~WRUL2~f<fWJxK{T|JRb37
zTA|7y3p2dt+t2V2D@0l~JN2F0_11W?qU#*MUYo?xsmD^n6l-hQwvJ>Wt=g}=ann8d
z?BBhE$>r}^=g5=^>-<YAx>GysHHn(^@$u$gH+;~0?WjHGJ$LmT_)TzzgPg=u5;!4a
z{dJY@N1T$5cUDa^j7<9@_D|{-r)_tMe!oWD{D$R?D`zGe)^*+sWWChOA#UZJ*1^HC
z9BY6!9)jXDI#375nHbY^S)buiMMo<Oe`y*?0TS%nWPPqvziG#6tf2G>D}>;eB84G<
zk~G;RyiwTC3Z6^?o{^jJ+Q}3Ny1%a35+G_+sKJ1$GCE-<tuZS5BOpRh@lDWmi$qnd
zXKD4m)#>A{AGFq7pgp6;Z1LPEj0L|CR((B3qLg4?$Vk$-;~X!0f8JiTQ8bjwIWA8*
z6_ZvChJbY1jG`hlHU}_pe3lFQI>K~c^N)0J5%-SY)Jx03I!OCORORg-b;o^bUI;_b
z^S8Lwaf^3sNepZG6Ctq8;M*Iv^$wh|W4uumG`rd6fjCj6vZF4LbPS@m3Eqeop09N5
z1P(Dd0nC;+)9Z(cU0TF%=UnW0DANPPaB<<|VDY#eF07`AOB3-E-Bpiqq=%aoA7z)V
z^sn+(O&>{{bnT_3Hy&Kan%Zd<SaRPe7qYmkxbr5}Qyfo2tsYWL2kid1&2M*`G_p{8
zA+e1}5M6817#MH1=f`7WkrZkF49lZjh_S<Qc*;tdt)nCUj=95w000wB3|(DQ6@#gY
z5Jy-IZ4khuQ<pABi^qQ?J`X;Sqf#Il#Q|dZuKM!mN;m9tmw?E1>eg)?rk{_;VGNq2
zAwKN!A9sNoTrf9}=H#2>sw;wFKbOsJbq_|;Pxds`#tpL2E+DGX5*oQj!`1oXLgsLZ
z-r@1k@mS`XM3Gw^4~0f`GHDFcQqJh+q(TE&p%F{^lw`~DA8uOEhNeJNCq#NxFG+fb
z;|rB%lf#9(>rRZPHGdy!Q^oy;G7i~7OU|6KdE$RPinGzXhQj|B-~FlGXc!e$R312R
zV92|l){2W!_hmGgK0uqLlISrs12yxwMzd~=w5mDwe*zpO>PxtLKVE&<8kiyzdg#sD
z?>kg%s|p=e7}j@g!m(QmThNd?3pY*tmoX;261pkv3^=QpqbP>{z*7?_V#Vuwf^?Ga
z$C&sM_KI@iTx-O4kFsx;>?r&<Wk+#%7ShX@Pft#;6B_D)pp+hW0U`}|1%%DQ^-+Ks
zR<pB4BR8^CHj&g8Lg$olz_H~|>qC`t0gy-Nq1mp@&cdS?`2b<wG^1ryZ3O?Py?(y|
zm>80~74weI$uo{}0kCQ%jiY<#iLHe)M#}ciHVG;P^MH7JcWpa|bV`iKs7=L^5K5&d
z*;N3s5`NhRCNxPkKpC)**?^P|ONEuaZDNn9hiLP-1nkpfvdRMZ2KPNj&>BdnDD4mw
z4H?m2@3z)x;9dq8W({!8OC~>*iJRM%cjm~ESk<}|Sgf+@b_AP5Ep<~nbHG9|{8QFS
znGkmghsVP5qk%z@Gil>&{A&NbCY1lB!<P+1fB^=xK&TElT{P0MlmRS6eW_lYVYvgl
zE~IhQTX}^t&^iyt5Ep6->|o_o3rZWI$cY}3CC2=%6ilZ`hmiEB^0EUdxj42sjruib
z(4h~9j}Gpx7QXr6!?uEy^^#|tx5<Kg@wx9MUa^*V8ZRre!__B&5Zya8BC_xL?mI`Y
zVRW2N7^Yo2!@*{>CUGMZw?5%K>fe}u*G>!ipL>E6IcG2K*ww1Eyi|KUn@0$5BJdBN
zt|6-0b@mIN(hdUkv(bnpCNC}PH%z}39VnW>>kHMa>?Rn6Tin_Ln<S=b4v^1&;hUd4
zdDvI~n|8V7f%RPO*qF%G=VW523{S8Gg-goq{Gx3T=R{9|m>n`H_;Y#U_7f67gbq-&
z**DhBaDCFh0nD<pX=m`OhpGqVuBAW0=-pW(Kl)ly+oACv8wyr(02=v@l|J4mtKym4
zl>P%ao1QoMQa&}fIpYg*zHif}vdCEwfax=5R-`T7yYiED54G@37m?N1!xgv5Fb#{i
z)C~#{j_*Z`L%UY>rzfq<jVwvCZm-&O=#WYo8uMUe95t%sR(DH3UB)bg@OL;Uvgl&V
z-rDHtrNM!+ILzidY?7(A>>CFzaXqPhNU$y8*+8}p%lq)nci#naJRzIEQJDLz@<iI3
zzjQnX-wzu>gCNqW?bU_clvFHWgxd5jp_h5pc~i1jnd!EHhx^so@dIe@vWb(~8Cot&
z=hAq_lQ#%*WzprZY%#uz?)k?=m0McQ_12#|@PGpKZ$CVuR!-(i#>;Ls*ryZqBmQ}u
z9EB_<n+SjX-^5ggrjnIu2Ov~7L!0Hl@6ZNq#LrxKm~u{fWJt}+O-~1vId^=S{xv;G
zUL6$)MyJ3$5ILh2?y2DF0@d1ARW9(G)=u!nW>f@DV@_my?%nX|ZI=1Qw@U;{rV%b4
z29gYeg`FlN{}mdDh-(C16*4BF`u+JFV#iXVN)ov4S@_)xgi%VoK8Lu^kl9({IWJ?8
z2q=!D%UJvy=@{0t6-9W@b<7qUMFuuGjVYCIR!=%{?IAFw3^kLvy`)Qe@FGx}1ajpU
z_?$w@g=XTkd6V%Lcpb^Yo#9iE-3dWNW<SA=W*J<dO(M}zH=!iBa(cbv>pKj(nkkn`
zAGMnEcC~9aLZnEo*xEL5*NYIt=<$YM|FSsp5)V=8+xAN1UR`N~UywpkWia7Px;{3h
zLAW8$V}MA_3|GR39(_#33Xl{ExJSmN$fW~0$&1sETuAiJ!heICF(o$oNwh3dfy|+9
zvE>&^VkvAB`6Drx;GBZnvXI9!+C19;2v^3g4jEwm$fr-g#+Z9i6Ahxb%oBJfYa#?V
z-`;`bAo862XB|I7inW0fTKvbP&Bak<7;f<!ofz>LLhy>+Q@<=aXM&}PC0mH)$9OO^
zX(kwC;QeCcsax`g4aneAI^818a7G=B^dO17Cm_8XZ1`x%@J$<<y~V(W^ALDcSZBW+
zwCE=zwRHv5)<Y~h2Br<5Qd$>nyM$dz^fN9L+8Rt^naRvzTAPz?h&#T}{>jz!*+eg>
z;e!w7)SbSUpu5Q5KCQ6pok?#?tv9}!>8+hja$AyD0T|b<>ghpu;q?U+{Dwyo)ItkC
zcJrLLC`f+%aP!rSA$u)uH;3F)ef(2-$H3Pk-;B<=sz*eDyt%a(&D+fms<)x(q4iqO
zqTl4-ENXVw3TVhM1l6t2Zi+apkhWUljTp8hRyAwB+W&ovSKw3O`?qjnY^IKBGwF0!
zxKAvK38^*QVODfti)nv;@}Skg8O%5JjqIn2JZ)>ITNmL(YGh3S7<9r4<EHWM<UXB(
z>*mjM^Myk?uu5ol0sEYx`lMNZWzkug+VNhs7CBp~1FW$&>wmxUhV2eL+$E@#D6$By
zN$cgvRb78t^fmFbbM?nP!lxR>GzejY1(n`)S4rji4f$*?VjUANMZrc`i(l0p>HEnS
zav8Zn(u!qF?h%XOE>GdbHch_tEw^c4(FeU(LmN7@m>>;RY%aDdf7^P}<E<vN@1)FZ
zvwqZ-4WE@2X1{g#y2U}=-N}2$eE!29A)HM*T|=j}b5%!(v9DTw=;^fWKQgr|ZXtm9
zIez#x9Y_(a9j0IT>2bdnEk5mPUiX*2_mmcMul)a+Q;S&nKeLIObUdT~$p!d-#5SJn
zoVJ8|*0wym^x;t6X47ld!_Zn^b*r7@{sEPF51TPR!krl|ZS9T%@17$eET~(qxSkeU
z7mA4!PJ>w~pM~gr`(;m2p~%1gi?5cRKRQ!v;V7LOwM3XcSrQ2`{fQ5S(II{BR-=Sr
z-M|)cJLm8l=2k8WG4x-RPZ>f4+}S<@@~<^ZuBDi5OnQB1$Up5^Mk~bN{PC54fA#Ns
zKpdG^Q#M}n;>gD4j{|zQ{6|c;sScCA$(?z#RhL_rzW@IDo~TpeUJ<#qZ&8PiZNBfC
z(xuC9fAw9yWYUG-=9-+D-sP`fuUv0&>(4_jF0SEkA0EWt`0(+=(Z;FO2KxEl0musP
zpKKV_!sB1+`u|a1buB7qC%0I9hmN~2Ts@1-7#Wmvje;7oYx=nwI^WL$(lU=Lvx%;D
zp^A82K12TE7RLdO%%*v7-+pwuByC8sy?k<)I5DtwxWX=KDu#0Uupyl$%$YuYHw>M^
zN=yE9Z)?Q{G<!!4{^V{reQA#J(Z23tA+bQAiL#hN5erASENKJ`l!BxtUUYBI&wn)#
z#1uumbBZgu9i;305XDdXe*S4i-kQC=$J;L8@O4g^$R*F?%q3LK7W^IlL+RnzHo3~^
zwOUMVr-(&kMHxoR&8U1VIwj~w6%KLo)82}ocKG;ame{pkSbhJHk-}Buyh7mkY8&-8
zjMrO%&YSItLtED=PM|G)=%16B)h=}<Uz$%+8l&*qu$-*8h$-2cxo{VZ=+IOzb>&^|
z{NwYxh<_y2eAP^21BEq(y5hqqLu*?^E!ZYgm-H|h7U!@>7+0LyqA;PD+ah~hrL<Cn
zHSzZ*xvv2diCuLOn%sPV>O|%=xo3dz8IomGzS%Fo((Umpemw9La#Z`wP{7UpVSKx_
zLc=)eE3Nl8zg7;LJ$?H7H?!oo2Pm$im3VwIy2ar|O0TgO5LFE<`}}FQ4H$dE9bLAh
z)Wdsm7vAg;Rod2WRizpFfYoMUJhw@=6uQ#)+ezcanRRGthkyKI9?#c#9B25}H-9Mp
z#53G#+9Ckh1*`;>$C#>|`q)R|sqr%ej0M12IyMoS#POSspWk{0J$fi(<%${ftb#?~
ztI6k5{`}LwRM)LVBR-GpZpfc$e3H%`6mqZVcTD=x!=|=(%IgX;p%&$z{XC`$tb)mX
zVLhb>)#Qba>revCt=dt4F_)4~8ZY`;EYEK)VduG?GsQNw5$|EKVrXqYjn~;G7Ls|p
zD#H7T-$rrFOi6nwdYzw*=BNA{{0{nqAf~Aec|T13{Jq8hIt~B*6#?bH<8CZHA2)!T
ze@TptDPfMpYow*6JwWUTUI`}{(ft|d$3jAc@Am=;z1W>gsR{RD$0#fQwDZrhlD6vN
zwjo6=wfq7Kb>ZfRWmDjb*{wEE&kI(28+#A_;L6CMqI~`8G(2yqDJ4&08(d&*Hu*5Q
zxW!<F!_N2PDAryO?PTr{W*K#k8vN{g3wgImljssZgPc_jeEjREBT>Kdqnk<+J32lU
z;v?OJI>(;!`^yzq_r$t5Wj1neLc@G2Lay!-1j0aqg+swfJ8Q()p7nSs%|)8;zK-zN
zBCz#H+{&ijr4oi>5Hc*19r)FoFE^>MUf={7LT0GKY-A=7t%uSjLtf&qa?^yWPj}6J
zR4VKHB9*u}a0rffb>&pyGEqQ^(X&uaO*6)bYKZaesgV0Ld*1DW*jIoxA|)XC5YZ4&
z`fA|lzbMGy0>9%>H9@Qtph!vQ{Y(8WRk{cV1`N=q3da^Y8T$Ka9vp^mzf(~@hiH5Q
z3ujE7ba`-Q?!BgGkVJkH3Z!?Q`=4F@^8H^l1$!P+xL%y@{CYENx?LY8K?yM|lJVXN
z9BwBQx`D3RI=Z1*83C(t4YKh_rMnc+&yI~UIm9sZsh$b%XmrIl(W_lOeIyNu4NgsU
z1#|A|>gtXaIj?8VDngfFb&Qp}lUKX`2@Ij`78xF2>@a?X<2h(<^va#zblai$mixZ2
z9}Eudmuo95ENtI^c4H(bM%jE<o@Rn*KOhm<D?RbXyCl3cG{f61Z|3&mVXk%iEExb3
zmf#An)!PkM<25qBF}HThX#F1zc0`yC5_gZ&oiD!${H9y2oB&^~fc+-pOnRsM1HOQ}
zvX*yE7l2H}P9`w%TRF5`1(I`PZDLQ)1nd5AfoUjQ6oc+kYI}ch+wqqw=@p3<d+@!g
z%pcav4W9Jomr4qct+RYIm(a*=Uz|y^Y)->*?iQT#(xd8Nuf>3H^QlBN%T^(45^Co8
zQgEORkwH*fdjgH`3aOZN%5)ivlIoNd-G>g*bKvn;SLP7Sdk4h<hL2i#`}<usLNg-*
zUz~M(B)a*t!q5PT)8BbN#13Al47+Z5^6J|QtbEm>Lm!t#LL=}B9ZwA*&e9kU%dA=<
zOVBl3h(tjM5hA3QE<KF1NKMZ(p{8H8nAGXb;WjGm%5F^+&y<;#1GA&QOFcT0VWk0<
z)7bi1q4H>euijoMJ=!ZaLzOO|<x5*Jq|H!|`YyE#OEh6(bDgdq(AG3^7TV{|xk5Ed
zvSlK|)$wIKs#zJ=O*2x?mG)+i!R@<udkLfT12nx8DV3?A=M<J3^RKra`^A0Z2a7x1
z6z&@u2h?aXt`9S1|MU)RpdI7w76*2aR3or&Mgr~|xE@=APo<ld558ZGk}l0OjF~`5
zW5#A${GmT_qpRsMSs8N&P7Sm;Y|oiD*jvjn#pEQ4$uz@kp+=N4`Nbx{`H9$L7$Qru
z2D%wv8t=p7G6C?iO;wMI5I5q|p}O>qOx4yr=nRMYPW3-DI^B~+vGuc1otvZ_7K7kx
zR_f0K25f{vJE!3z*r%p){3DoSGC?b;T8PA3tQz6jyfln;ZFWf-BWat=5o9ZG1ZR>V
zmN3$-0pO7hJ+Hpstb58#*2;`GW<D&vQ$npW=^fI`FBR77_1SoI%`&$xb7KueWvvC1
zB-W2KWbP4I->fI$Bgd~|0K^&xHtp<&c4S*D%nGe@0OQz$OaUWG0(p|T@fulUwnefS
zTJ%@sPha`dlbTbkG|@#f=Nf&lq~8yY6*q*^F{Ch?mll8$`WHWXydeV^88%p^k1nG>
zMN}jM55;ncLOFn-dOW4WTT|k1pU>7C`KT-H!YAPGTow}`KGlJP(k$}YhHS*Z{;^4K
z?R8t!D@4AEq(k{ejDWm?hR%NSZJYo2DrW&ZBlA3#DW%lGEi4(zAe87~{zr5<2r^6&
z?0cu&PfV;{ts&JkVY(V(y_ZpB;l<|K<|bce?o|x=PZ#x;6D=UA|6SURB}1RwyEo8c
z``wF(3y)yl${|<G5V{x0ZUnh>FWZOcO*e1ZBJZ#voP6J!jl|`y<z#MTHEUDj(&3XQ
zFDvXq8WF0!z-1_^7;wQfGU)ttf;js?=IPFuk0}q!`aBfU-(;E6j=Lxa0I;XOvvd^8
zD&EA(&Rnj@#Xki}r<WTc1;dqWoNhMMep{)3$tWg$)U)SsZl%q*qTx(FM@Bsz5}^<k
z3~AG@T{7TAGNXf|Ohr4!iq9LrYM(rgmOVU0tsOUr3_(bP;*<>G)^hP!EiFTsIyQ^f
zv1<H_83!VxOt)x00paw@{xfUlwU7`YQmbG35#s{$9*R?Eq$JU&=KhI27ho$2t-r+V
z3gWjm(O+h$;Vt$0dLw$pK#n-|OCfAvai_6vT;;Ci53$6vAd{gp|H*kH$RI5*bjN8B
zUul}rM|@jV5*C={RGH{s%r)46**QJjml29TNy^{iGJ12JD2sxtNKKT4`D>!}#d!n&
z+Om3a-Z-hTP#i&E4~QT0J32Bmb#?Vgvmp>ql_ns$L1eDm_|Cn06(6l@m?_m4Q_1N3
z7R!zO-#Kq^+6#M{GE!`^pn)*cC_o0lDV#S<YWKL^p_|yE#YTm5Udwb+4VjzfSv<@z
za)I=U6)qhbyr+2A{9V#gn?*s{C#vgHE=bZ=MFxe-&hPk9QEPs4*4ta@0M6qAblTd6
zE;B_*9K_7ukL%Y?<yemrM~q!IFGJaE9Y623b%0A7umk$yyJ6dqnQ0*72*h_|?OOj1
zYZq7j(|3biqm;6W>)kp@QBHpICkMm^D4oJK5HucpsMC`gp<Dq+iKT|Pd<YkH-bD05
zsoPt%c>&G9-M4f1dvw^V@!b&j3K4i*Xs1{sJ<Funar$-aSbY)}g4A<>@_%Ndn)dT0
zsoD1RccKXt>kWR#NX3=XuqVP*;c90g{*}*S7E`@QaB<FV5iUX@6ecQTD&8hauomeb
zk!ww@fAO~ElrhqBV0UN<X<vly6mkoYl?bcq=$8bG8MZIbfI}3SAd^!O{t`M5k@jiz
zj2mdq16V%fx19qD8l(E~iLkA#nVL#C=HHc5U$QNua?k(ewyP+^_~~+!e}DDwdm!uJ
zbW_idD$O5viOwWghTE2i^py19B(jO1l7c&HYfGuIQ?6iI5B1{XnoV&(hfBzYi1u;g
z7Jd{(&+BB+fSvsHJ?0A~o3FCk)0gk_Z?M8Y>i@4_Ex${98&3EUeMY#Jmk&DKQJ6>h
zm2vZF_`NB=@Ri31y2eg9GJajuKTSS8w@#ikc%f4#?XOyCE9Ldz?6NpYXw_l@Nu7s2
zp#^=*Zk4rvz9~ZTBdjklzG1@cf{QwrGeNqK5Hls2{_&StPw7j4(pnhB^qWA~04HQB
z&vTHF#!-H6R^fc%fY*;G|Cq!@{xpwLQRWFqCQk_m>vIW5+mX0{)~9y=<Tkm_)y;#a
zD&-wtdGeKzM`lmEPN!aK=!qw=ISMt;@iDIxqbyo-8k@!w@1wK_2Oq%_jf695lsl$U
zUN!Oa-k&K&t(Th9++;__q4&j_TT0X36I~1_TSNiH=wcDOpdnmDg<Aj6xao}8uwmA_
z<rAkPo_#M~$3}}BrV)LHK7GoEVNRR}!q4AUu;&x5_-Fagd8E3C$B#%c3y$@UgyX@Z
zM=zjT$N&cMEfC+bFT(vI9#xZ%#@4>u$F<&e)Awp;%OkE`iofaRw2u2Q#`q14>juY}
zTlKg^(qO09Jo>dMG|dcNH6<}IyrzS}2T~6?YwMBcWVXH{gO#Dh7{c@#`t&1Zh=^m3
z96qdf$|a~fRu@@$<TBMWn?LH!nd2t9e%g*Rb6lHamX)H-3nGm)d3I*%1n<)gyZgQT
z_KNbGf#0-rO_+asUXwO(X45(`@1%(wn@6K(E{(b9kqHQK#Huv7YR8#8r0`D<{xk(@
z((5pFP1t$j8<tuxW8a0J>hzG+l((qwxA`2DGV`oZ<TXpO@w=3B9H#~;n9h+w^oIDu
zT6l(5cR{REU34U8(gF@Y$y~eERDIR?r`z-1k9EEK&nxSnm!I#I{Ry9{ZAQHYz-D4n
zsdsAqlxOd$R<Vt+bU7JMmUvMpuf>gC&aUt~N~*8!^+O!ESSTBgy&;T{`D52t`kVn@
z0b(*2WFVu&#HY6`x^j+8hOTF%2Kl3Jv~5I01d`mg;c5{r7JqHo`BP#JPjA6;K`;-A
zsIj)fonkeR;@YUUlrfTq#-tCpr0`==I66!Ke1=jZvm&gaF9>0g#Uc|4r86HM9^UfF
zYxWK2{rYVUR|6=ND4idKIl=K$+A-`DNQrb;?Zv2Kv}=<S*u5B4w9i<?fG#TiB&zNW
zA)mx{k}h-E;kuRsBX|3~(rt1(x7)AfcOXdzt4QQu8kX(brHe+E;?6$CcySU=CPAPz
z8HTAy6HC)Vv=Qz<6H4P3tr%HKs`#tOsP8Y-0l}kQEw639RFI!vL}1sBJyAy4yQRt{
z2sf__P$ANL+mXN-_3#_VR(oh??#6z&$R{$ZR%6qOsJXnFltW4h`!#MY_I7wP@Z;f$
zXpzp}oV@*^PN-SdMKCC#&@5opiN_AG*+zNV5ff5|w`HQ{e8~d29${%hteiu6nnVfy
zq-D<H@PTONpN%h%_9H12OE{S!#u$hNs1-(=rZ14DCBB|G+8CX(bo#JFg~cbN9IRP_
z!I{O(Otdt#MMUf*_Zg>`Ay1=!&M6ZE39tlFi;$I{hN&?KZ|N?oi(W7x^}N{VVJJKF
zz2o~GH2%f%P60{OuYM%_P$abjoQ;cdl1V}ZCDc0UJBXWxT%4rf02b7NDS#yx>2Q2p
z{RysKv3N+J5P<rA2D88Oz*I$;j0&}<q-2wOeHLJzj2os~L7dhV_62|x$mdxiE@$a_
zLW>N4i~i&my-qkZY^jo<z-3`FK3PK7xY7V(m>6o%P{{O&R}GFn134yba1<LLCbHbF
zI3=;H0vnMndH3D#JcHh39%3HGaX=Pr5#=j|h3|x{Uph9E^NMj?@FTT@9d*>T2j_LS
zJmNf6%k9wabGy$y>ZIMOd%X72&`l=OM-GAO8rw5{4c<N1kLz(r^iJ^wL7LqNz?Vw{
zdK&TrrC9wL36L8O@F@eVu8mEA)p#I|d4TO3I<#vyS_WG`?+$igY!oG&&%9ERMV^5F
zmU%pQ4hnB?vwa&MvcU{}{y2`>;K@K1fle-HW%e=bO#V84+$7z2g;NS$WfAE?g*3+<
zB4Si6pD#xwa!D?2R!xjd7ve4ByZ$~`YXOwjiyLSa=xW>m>$v3c4KBj0x{{JU>PGq9
zgoK_13EPH?Owl%p;*)shSAxd$fWTt6aSrIlSgbzF=QpEl_HRVn*4O;R@6)DfD1|_Q
zPM}ye=$Z7{s$z5f*!-gT&|g*&$i$8J;i}FasE>|6ZlJEMt^0@;X9P=TQvyP~>sK?^
zU*1Q_(_+T&zbybz>4%GywHYzaa-%n;;ISs31`Gn6tI9M}kqhl=SO5_CNo#5@)VZ+Z
zNl;x{zc2=EFs--h9MH-^>9@goJys67Yc;@ar2c~3f$@DjBIoZuSJKbznmJ6#Ikgu^
z`9H4S^n-zZ;{B{|Qstp=3Wk5$_nOFKs}ZpA?nm~~0YPIl#hBt`)4U-YeoMQN@hJHk
zRj+45)tz-BHFwX?f@_q^foNv>qj8D0el$(OY4i}6l}o0Qz~WS@aWp-sh8u)ue(oNM
z(2SGoSQnW^cjqnds-(2Zt`3{4lA47FZ<bKV$WqZtkL%xqrc{E?u$a|7M;W~yJkSy2
zB17d65aBsF`rOJ9NPOLMgiutm{m!I$0Mb!b=szwdl^V~L(OWY8%DjYF%P{#Zqlc@t
zjbD#`I4hR9W#xHxi!8SJys8J%+LW)y^T`DSYE`7hzV?4YqIxW)UDdeliyt`Op6^yL
zM=5XdynTLc|5*IsQb(Oel{W*enr2y8?pgU2inAhOfF8S4Dquk#2n~@=bqWV1$n*pI
z_wOG}*%lWJ4?N}a2>C4u%xL<0YHqnAu9;$T1lL>+55U^BY6;@_1&cK}=I$gr_TYgY
zYQL1*kgG8jU53WVluZ6f-b2R&-k4@%u1fP<j+QLtyQ#i5+>cQV>vYBmS1==r;UivN
z5`*4x0P544tSo{R8|=U&(=4yJ;2J+PzzU5Au0s81a16m%j{B#)hvQgA$8&Q`=OoQ$
zlp@8s4feZX*}H;;j$2m2!*wv-PyUw?)3rIGz(<eXOlks{6_skts8tDV-5c)RzyFLG
zuHq8r-q4*9bjaWGH#b5@(TMhC)7C>v3<}F}$;efgJIGg#H@(-oGDVPF;zd~`Xp*?=
zLzsHX!FZAJ2lJ$2=BA6NV3}FORAP{SLy)T6AsA6<gNuPL7=HB1(%0>nvoBE-H|f3C
zP{lmXCmdeJ3V2V;YJ1qGX_?2Ggsujr+4{x0Q++%(?_c~t{msC;d*>_t+(}1$^5AM$
zj@%=<&(}`-Fb=bZ@!Q|LyMB{JmHxuc_C;!Tce0EGG}vp>3Ff1P#BW;E%<c8BcD8@<
zVeiJAu;4l8#H=Cluaj7mCYYQ+8rFmB702ya#hnw@Im5N{gQYEI4JRL~QxD0&yiN6X
zRYpWD@kLE6d=7We%3;(8NdH&VV+5`Cw`o&+FBgzO+!|;eT#i-hZU0ZVhNxohut)GP
zU3X1!Yfx4>8lq1tbhU<upP!r05ssejzzZScQ!p2Z9y5n5KeOJzIWIlug1#p&*^@<L
zRoHm@&Yd2NCbrmE0(rl$@kCnQ*VCG^CscgoEPq+5s-clyur4q{_34vW4sZX`-jrX7
zH`$hRIiid4&*30XzufJe<u_N4IPN}YaALe1AVb_nwp}?$f2O57iHuTi+oAjDc|8|<
zzFxgHE?vjxsxLNf_NsN@KANel5kH2K3F@!+k^txl=`Fh${`ra2scFe$H*W$!DGSeS
zA7BbyHb9UZI)6jmbN4-oB;Bd!1T%>IYsG>ENASI0Z}rD*IXLd5HQ&!@P#7{4KL2@!
ztFq-Mp&es=JkDDT8BEQ#YXZ5kSso>az4{TOq_Kvnpn1V=LRyzGbvYGu;l;eRRarBj
zn}~J_>*=Hw5h+t?LkY7UzGx99HUf;AoB(TdaLS!OL$<qksbVN0qs3@DjJSLP%LTHD
zyk#y#QJDwqU&^^Xij43iC=!)}wAO^-?J(^jM`@etgzIbJ51K@pk<1TO`AYz6PT>mz
z(w?gq!`D;4{jte|u|+}ylx#%DdCM6t*wzz^`HEv3$DTynpI<G8>PKC~@w!q2_*%8`
z+m<T_S>2u=^po4Q;DLAM{cyGO)gzUQ_m2DZ3M#j)M7U*~d3H;KVJFy6#0z0Pgn*&$
z^Wc+EUOh62;Y3UO)QMf;Mz}FhzI!pKY1JaKIo|_%-F<JUf$KkK=X%wqXR8cT%g%OZ
zDB)>x#)24S?uux9o*Vcu<u_XIQOLDxGRr#h#bbT8{c1aVsL$;3qU5U#Kt09H%wOh0
z#vMZlYgOh-@83VCfw_e;TT|jaPr@I0ofwUYN?a5v_0!+V10ip(UOxe`daAgk0cMMd
z7bS}Wc?Z8fAZZLL4!zhlq|76ntE^BUU$?JZTV0m5Ow1cV<3;(c#cmRR5A-J`p^k;;
zJT)D3vzbKp=0@yLMsxr9=buiVL52^9x5Ze-7`XSwk_nVto=q2qe6OuLd}M?2-?1Bq
z^}b%)!FK=+gN>E->5aE9?mnl~&ttROxCaLF_WJc-Ifx@$>XBFd(3~*#)bc_$6y{Bp
z=_h+%5&lTKb~a1JMY@CL`$^fw2vS7uFw}k3LQ=*j4ff>G_2sV3#fL3h=Z4tcN!A|y
z#9Q5ulRo^#W0pQW08_W$8}U80Gf_T7fo6~BH!o&PLqFl%KKxkuTlbM|r-**=;i~ZX
zO}<#V9WpsJzatTAOhaiGF-3_D8JQ_O-{V2O#@lX+4N(gUr%@#A`lL{LepZe{(VwQM
zJ<(icchaw1X3AC$Sr%!1!K6`K>Z??)@nGM<CcYX~MzF4KXzV9wM{*NO2M?+a>Y5oV
zc&>F#r}rH=fC)!#v_>~arAEKpa<z_^eu3MOe<SSB52-bWy}XK!x-=eG{9we&K?PEO
zN1W-bJ>&NL)V?0qrYd#$INhSl){vy#!RKj&CT_HI>6FR%o1v*7vNsZoY2Sxurd({Q
z35&~X{Djc%Jj@XZ8m+B+UeVg|VTt4OX38YP4F#o*?~0ZSc|c4QB3$p+j66ndAC;Vv
z5!9&tp|y5^SGV8U=|)>#w~DIlPNcuvNW9m~IWROhf8e{#q1lH|m?H_!$=?~Juhyph
zxlcLA#ZC^s{|#ypChIhHicGoDP}xOYU470I+ltX4rB*lOj1!$b__Ix2#Ttb<0gRW=
z<^(9}Y*VG>B}+W<8Yz~zppO~bt+tDH&LDHcrSlC88r<)}Z-{wbe#3VP(6WAg^;p)`
z?sM(On@&+$6LkJIoBRmsu8Ezrmwb%?`aI8Br#dLBN8+3gHv-ttLsMgFG-~m2liS2u
z<u5aB%Bx9Z{7Gkw3Ga-mj1in3jm(9cy80ig*EPaASkDk;A?JwwHyd*xDhZm2WAb1i
z8@8*Dj!_Y>ErQ6pb>bw5iK#t>P5)rz7zIaF%U5w$+m|*v|6_le$vckL3B9bS6m83O
zMj)eb4?b;Y{nnp@c24aI?OTZyq7^>x-x3~{9uL~Lckf6;PfCHx^pgAIp8M)HMb?bp
zY-*3TX;wx<gG;N&LseyFHQL2MrgUpFb6hwP(o!SdT;%EdmzEd?9U>&VG}h;3<j}}2
zN-5v9{Yq_BplW7a)b5GH4o7%}xh*%&(4}sSc^(``f-r4xo~B1%NwD7x+_QM^;-Y4Q
zUOL=U+-X``TtBS7vir`5HdztD2{&pVzt10if7`F{w;`}4*0%W-hiu#}UI3IxaViFc
z96T>ri{9m)v8qnhOH{^LOcEn;+=|X18{5FCC_1ml%*vYI$HewU$x>ozP&<Dzw@ow6
z%*>1;>Z`UmM%(0X3wShPnQqOG`Upi{Iy-G4Crk88&wBlC-s_-GdcuWBvZ27KE33<$
zMcOhuHR8vr`ifstn_QwJL(21D5SIMXogi05QGYx%gR(Sh4Y*Y3k`yLvX?=PhKlE|v
zD{<vZWe(1@r}^_b8#J7(N%`3z)#v`34U_vm{Xg892S48Vnfr9D{SO9TJudNgsR=(k
zPIab~w#+&fKYEJg(1u+#r2m{~$C*Yf^}9xNZ0wayP~zw;i9krJtl!wNsplN{*sUEp
zr)GU{x?ke+b!_|Yz$7jw9M>CsfO4~e3~xg3sNv%@4x}GVy5HuHz<;bXj==O%Dag)R
z-QCXyZ<RWiu`e5ZME};_{iT;-(ZCt<{udmp6AvXC*e^Y%6&rokW}bDa-n(ea9)@ez
z`g3$2uO$t=_d2QU(x2prvYY&<gzMU0Jz}+^%a>=~qTunG?^1nMmX_W<66v_qbOnd(
zVNRh-Sg~y+trFKIPMZKH?dL!%!pw-O>b}_c{^Q5u4v{#|o!Ud;BS6<id<3L8K*Tsk
zyG$vo+s#hq<=Xnj%6go;C)6XCeR6(#w`5O$5_C)TZSO9Sg^tb{Nu$uAYJFJ4X6qME
zkM^^3o3M+jKF;vs+lOk?KQ-!Z3u$;ZF0%PAKO23w`tN#byI0s&t?u(S6+@d7sbts#
z-zPOu+nQS?Or14rR;<oXZ5|j)4p*!w9KS5{2S47Pzhl?92c!FkxAY$G`X^QYMz-CI
zHu}ELiyn9CZ>Jkr?MasTj&eGcEJ+kVB6)$863Ke5ECRE-tL*sb^#z~U>r|2_k_<3^
zI4`ULq@-BAu=6elzI*r1DL}>1;liI`wSK@ed5umoJwv;5%adudH9|r{dN=*_VW&_^
z07mOPf)Ur2HpKGO8-d2>e66OfJw9+@V18J?9d6nBwB%Y>>E4pKy1TqTtc#0j&Of~9
z;^wowV7(#R+0Y?fe}$E``C+{3W8eQW|C><&IDlJ^9Z$kC6ZPsO=f@cL=G*W}vx=7v
zaidmHxq4&>#o*MRKjR`zx_tz14f^R;msf%hY|i2!Opk5?k6sy;ot(D8&TLi$UVEO!
z-yVV%&#TWN5=fm?;o%EK=yGZCZ2gHdekNpc&PM9ER%tf2C1Q%uwY(rRGt<fa;f+It
zzvDq_j&Jsl6{}oZ6~EH5d43%0frKspV&<+CTxleuo}DUI4Q7dVPLLC=t(|u5&~+Kl
zqq5HT@L@k5g0A@dj`i1H4!j;$7zrionV-$tjE6aoVs(M47a@0QV?QQE=J%XP<-n@>
zSaOOyD&2dJ4E~X)cRe=w&G#!WY*#J&eE9w#hi_0{j|vdL)Pt{DTYzL*4O%|B?G&-d
z&Mt<mnKZ!Oz`1oU8$e*wfJQtIObBX2AAQtKRUXpjh+c(?!@HkOqy}mBOz##n;fyGw
zJk5<Fp*p)KRMxoAd@#3~xEGAeepK+Jx%5Jlbmr0`&8z$%<$(}ljS7p4`W2_jr1Qci
zpCd=!4j&j|XMKIRs1D02M84X;f4Ft?E^yL~;F`3BBQQrh188_WCGqj&F)_ZA=`?G(
zsMNF-eSQ`|;Og2<9k90`{+tM{tSwlRU(_((uoUbqwBgA=lzNfE&OVd}L@nOPh4YM|
z1>8_?6F^BP&Sxghfp<FHo}ax?2}qQ+GDy~n(oY^0!?;UVer}mb=n4{(!#Tyyl;mTU
zWFgE=3h**;OnQl7v_@$&HNUNDlxwNq#>U3)b5(<4W}NpsF}D7BvTmdpC+P&#(uIGK
z+s}!pNaEy2PLUS_WB3Nv3#;7oyMF)3?P8i^&y&L#^_`s?pkru?83eGEmg6GYOEOeM
z<X*Vl6uy5QuLw{G?UMSP6f#HRGY3~`J`^UcfN~xH3<<q%1|l^*+!Z-?WL^hHr?}1v
zY`_N&!qWTv-0h}zVoU4qxn0<qcA1vD1W_Pa6)Pcb)qFCa>(JGrO#AljTk>ta{6>an
zZ!&>nj~!VNV{96#3UjMs%+jy()F}UwOqRXv29XeuDr2yy`(w3FjIX3~lK#n5d^_V$
z8DBFJ#LMu`2+o<!mlPQeWlSSw%!A2oRm2(Pr~B8%qHM!@PqDIHI|uH!-D0t_9aF70
zf4)J_0!*^Uls1tZM``Zt->+}qzZb_kMVOpmr}l(>W?vm2AAf+((6JQV*|SGeWggX!
z<G&M7RL?HPF6Zg@9XdQ28_ZGS6`F>f`0Ok$?O4i?-NgEy;C6rg#h9z>W?)y(Vx1jN
z*f+AIY>VsbLz+_Kn%3wsGTg`NWjW>S0f~LSw*>O^iW^qr$A$I+$G&o*nI5Z#lZyV^
z{QRC`8i3wMq`Xl!*)9#HC+a#dOX^&#mUNcV8nr~49abM<`&i$T4uSfn@xsH4Q!~s`
zp%YiSd1BBJ7MPbCBn|WQGCsP0DLD4J^r2XEn4LiQ_$;uzn7P@bP8yRxm_!LI7FziN
z$)>5xfNp2^E2%%<SP30;c<*&mu2+kYvmYm<nQ^bNW>x9lG3R)j_F+~Ekn)ZuAb;rb
ze<|}pu-a$P>C0%`-)->XSJsavx%b<BE_2G@qZ8%v=~4;_-j+OhaNc4|H@`8PUcRy(
zufl}j0jsRO)!zz<Wgqo|C<(=$R9LX$pbjq-+i6_~QH#nQKA|@OF99L(nD~<LL0F_@
zVyla74YJCoQm|uT4z?;5QUp;?5zP`LpjSM}vnTjH&Hu*;lHeCK8RYW?`>c$k7bxW&
z8F0$ij6;X)I9If$*qI)MFhzaOIi@QV6LQNNyk2xD?%nS*s*@$sCU$y)G+WA_XHc}n
zN}r7<b`z5AV;SZb*H~%}N4EOoJ9i#D8}7u3AJ(m1D-&A~u_ue2D|v*Dxb#9D64D7b
zR$RozWfp`lSiCEN!QVF|>Gv;$Kp*1R?>+~l4B6Q+a@UAe3E@umJNgZasXA0>@~%vb
zCLOBYy*?V-x`;NHK8?CYt#-cW(P`0DG<ucv5D+AoOWtC#D4S2SAu6XD3QZww((3sA
zd)+XYmk)E_@g(<=^86yP*=wBT8EN~_H<8YmIG9M*4L6G6rcd)b<%5381SxUq30U>{
zDE&{c+md<n1R%<Mb@=-6TRjZ)RqypMfFLx64#(Tg2@#+W(;y8csUuO%4kxiQ_$;^w
zl`-{D*C_fTxc96dGrrCJE_Axz4^%Uz%e`w!?%y9TDe`)6DqOQwf?}9P{hjXKHhwtA
z>KdzWo@cGJwZ1sqT@%bWlfmJ(1!I9?JsSqz`%621_<kym$m@yTrS3pxBSp~ZKJhRo
z`6_CjX=Y|Y@s_PC&-<<TWS@VNP&(UvdBl9M{rzP0nG&fB$L(+i6Qr-$1DI7gCYV-g
zcN)D$C*8-oYl7G$nN0)~tts-3xifQ}MUMpEh7yb@wVI5M{JQYLIkm8$uJ^v+Y2LoK
z=&cK2AiHGVPad0RvrESP(Z(NhLqCAVX=3~r??8RUn2fVxlkoh<)$%j&PxYAl#(<~N
zN1gXegLt6RIn6XH*hpx50opIz?zqSOm?6ZuFJ|pYk*H~(gFtDy7+emn={G}Sy)?~1
zhBET%sNGa2OfV9&aOZCvSExn(mWPjcYt#0C2JKaBN-m1(me#kpi*qQ9jYtlaVw>a;
zRPC5k?UD~+pDmH5(s8qGF!^1YaNNKfcI8ImJt0m}+E(xL`o(f6_<{08R2x#Tg-Y97
ztQG6ZQhyvM^8z60=PeV12SCjF71O5b0v^dPv6}<=+(X0KK+NZkVt4D}`8YGu>nI7#
zpmJ3+NnpMl%_SVoBrvsGYH~)lUI^~0(8q%}dyQDNQ}prdr9q$pbl;z&oWIO*H=Q%l
zGB5`aWR3&^qP9?XO(;WzNe{)s2Evz26%h;&Kghyo(e!l0XGUoHa#x@ek7;BsX0p`m
zwi_CW!wTNxQ9-VFJu!jCN2$Y6Mga))4ag$`BSEJ~F>4;Y%zjU$b^$y=sBGeBNaFz4
zfqC4?1vEdc@yEG<+pW>@5FTIQlM~}*WR_ekkLEfFRP>X+kR@?E5<-ViUu9G<ZiX9g
zezRI1cZR9lzGUv|CJ%E6ZGsAa?7qo-5Ik-prw>!xDQmt|Z4~kX2URNgeIQs=F)%?Y
z5DMH9b-5H?To6g2iVD|)+p(i!LK`W(CE<rydYOLbMsPfF20#fayW5C;BEGZy#$~Gf
z3p{&;vVOp!<0Bkr9_X*dP04oXWqarLZ3*DgsJ{VU8ZQ%}$Jl%Jmv%DOqEMY<A;q<%
z<=(biu>=|>gSHNu*C7Tkem}$6!kDLZ!v%9BJae%!A+8>>rU3z7x|CkhT<jc#YDX-Y
z?%I%%Bj=9LrgJ}B>a5*S=Q#LXJIf=hrjGbr2z+uM_h9w}^<j^$B>cR0+r6Zuh;4ya
zRqXTAd!-l+65U-~z|P5{$xh4aAj243brTPzhnjA*a}0v-lcT4$=7vt^75n104IEhW
z)02+cuUoT9udY<O12?o|3z09>Zr%sl`YX+X4r<#czI6Q4i8g5!$V$VCF*2iNjQEMy
zd4yr9$UP%cw`dVNS!svr?sL(9jC<e=o9~sYIFth&7A*6D`U$h;W(T#&D<0mv_mp$W
z*St=8$MAVRu9%iG29|i#L-gjx_td^>@v*06XJATmj%gLB2;mRtO02J4dxl_<!Xs=P
z-mkMiJFcL?8+_VNDg|A&Ye6t~)<e0e^Qc8_Mmg_ttW%4dsk{9q4bJO*d!COFOZBnj
zdqxg@+`NuJTLp@lA(tc;9f`#()D58CU)6T*uPP`Tho7kjYQQ2UaaJbD=bNU~irdu3
ztzy=z321FTy(PeSuIS$6^X;cs>8BW(V=1#>eb=rsg*W~j`tm2Ob}c6T)X(kL?9aDF
zO-PO(ePwuPjvgGeD}z@K`uL-QT+IKw`Fy%Xt7GJ@OLwk#dlY|zddHbp`qcDmG0E~y
z&ZN$r8#=aFEZTU*zyB6_)PIKr>;J3wK>GD9o7?>c24=49;H5851p!kEGQPTU2R)M3
zCS$nIDVAWPaR*AnzDcElCox0utFQi|(4fnDHz!G++d>89SRTuNS<Bm1gtPnHFl4$6
zqsiEs5$7G>!w@*f*_}36l~%n=U0%0|iRe3THcQb%5)fS5tEgJJIRN9F*do+jeO@6d
z{Wa&>t5?f8MILqtf$lgk{l9)KH<i^k{*wy;%l-e>_5aUQFaP_h{_897f6~4Bm+G9+
z3^p8!o<%+kL&&)X(NwCfp*O%Eu5+%xOZmhq-5dD7bj_`0SUGgK$vuhzW>Zj4v|clq
z-wDiBsA{{3KaI@ZI=w#`czJRU)XJv@epV8n0k*6z%}6cC6(lY!bMwB>V)b7Yn&EJ$
zA-4)vRZ*rDtl|Yf>pWXzJ{8j&a?A<6Whl*rQWakCyh60_>K9`p|L(ndJxuzMjhOjG
zXATvVdFcnA)T5x>`dO?Tet}FQ6184*wz-AH6$-e#Y8U=wuf}JY_xn3*_Hw2!1UHY#
z9@gaQXr~;Ym(^_fNo!TtwIvNklWY2xwViqHuWM6nBhB~DJ@c2D&id15cXtgNe{|rU
zjxGbEriV_qz2Ea%yPZA$Sn{Z@@tLV|(D7n}41-jI)E>bFxgVT0^EC>bJc|tqyy_km
zI6YA(3Iu%7u?rOfHfx}1eMPnrpIg0$;>79DS#{AY5-g*Q!$?hq<FV9{pN9ST`I0O0
z_Xkph7);&U`NMzufoMMgV4|OQ^&Eh_V~d+Y-MtICM}Y7JB2J=~BGG-&rXN2I*HS1k
z?zJ?36fJ*Pk)K$+;~8(0>7#fitAE?qb;yEni`Rp1+{)X-_m@l&9etlR)Icx%XZLb1
z?)K|~>Eb@tO&f`=MT6qG9Gtr3__k#7sU16Z*jtX0t5#85^^zA@&oQh!-7iuhvp;B~
z&>&nJNbgg|VMQ6C443(UAH^l7MOZh*Um3l+KTGB)k!l)D=uX_eS^7ogp6S}$D7oY^
z8xi36DK6aTvjF`!mcX(Q1<{VjU1y<qJ!W6=vPRD_boUH%#tPDT9Qxm_8Eq7IeAoq;
zMrGon6iUh%;;xO}+@j(0(w!P);J+Abk__?ZUen|Uj$eL2vb&J?iZ|7?e<O_dN1w&^
zXE-xt6q<iABc-_)cQ+{^o}ctsh`TLrK2%D4x`FX&bc|0N@M9(3tJ_Y&>h#P)a#qj?
z<56-xJrw{1h$5=Eb=Se?@7=q%Nm*WG`O8R~KtQUzYTGH%*>9S?$J@;jC$3;GHrti6
z0MOh<ohD`30LEwi{p-&fDiqLM-PUu8%lr3N+=2hDyRqY=Y}}iXK?V_L?%kWdD}zTO
zjY6T7xim&1_!C<#l+H!)i-n>sKnrcdfz&)FrnToXy%n>D-(Mglq5C&b*GKb#@5DH~
z%*^bffM6PPnj3wg2l-QeuBjZza$3H|;0%dO%gJ=HW$+p`(*~e>NtWYwu~A#>Q>@;W
za|o6<R>XfxJvoz>Zj&<a)}rNSUHd^+u)se>bo3Wa|K4t)+%SbNMxRq+m<jo{wstt*
z?08eL<$O(;%}s-`bLY-yQ5tj^W@JE;qzAifG~6|e%x;~tm8LWJmx?>G)E0{rUMBl?
z`kgr=vMR-rZ+V+b?cAe>Nn|Vmd8ycS))TNq6^2xI9zF5iH8zd)Hgs4Ffg-EdU;3i|
z7vzRHEBYuA7ry>`hm1F*Kc|Xrn#%Sm{gUgO(nbon<mY051Dy@U@9Zz->)y_T;CGMX
zj?V&Qd&&R@O$T3+&qXv1gjFK%C1xVf#*dCQH_8gPoB~O7FV?b@MqPMxFi?tViHU<<
zO<7hT1wN1WPN~b+COHV!ZrWlR%cpwW(5}~n`Ef4JQ+`nBeUB&^up-_JS6)xx)hWfg
zUn_wwig!XcU`z|g{S>CdK|0L(qn%K?3&`CQh(K*`*C9poRiHXpzcixD)K+0KdjR6i
z_|j%b*a;hhWAZB;+qP+=Ci7w7dqsRwF0}`lYiX7<Um>=?za$Ug$dbMKj}E>%qYyT~
zvHY-TyUrnnESlG7pP|E0GqGx!?<mS@s>YSHbhjXSN141H<TWYQV1|WI6o)_d(JEfl
zEI4>lY-QfVbzem9s)*jxBFsOF-rrvO52AP2B=yt)z0JfV5lm#fsVc!=syApM*Q&k|
zi%oGE99msCoqm<TldPo7%MK7Uq~upJt7Jx0;a}Ry_aD4R_1?fr`lsrhW{8mt*$A^Q
zxq8<-R1F^oX%VTeYTc%dLPrcq;BM|15q~<>RoE|pCX_SUP@lQ*)$g9g$Ir9><!Q|j
z7Wrp|;X&Dv5D6{AXR}Ls2f4o4GoT6q-uTS5Fd8I!x9&Xfl%{>fdB7GCe5^jE={Nws
zIc&hh=lg%wWjD?Sp|(vG583P46!%rs=EcDMGAwfP<jEYYD@4@*!}U3gf$OMQFk+3=
z^zaTfM_k+fKg!-8l+KN|?+|n%4=2jaJ=eW`;+tcZMLv})CX<V6^V{W=9-ZJ^KkiZn
zG~%dg@2o$H{o$PFqOu8_2eZDPre>zFW<DS$zH1T#h?7xj2RQ9-BO^SYH?@bfjr3uy
z6XJ5k6OTL@S2N`ELiJu4eqZzRms?e1r&~x4XW~5%8}%p~KR>_TDPkvjvL=o(D@pYY
zwdG;U?xK5#6t5ugR-gPi0?c}RRJG(EOA8=fRTQl&4R+QMIe+V)*L6{dl+EBx7+6Hk
zCSqmf2`PJVIs7}j^cp5P%uW?arA%r{8dK7P!Ih#}wmTVeqS%wEie(kjUeiGAG#6hy
z1C>)hZj0vrwUDT^b?Xnz<FJ&L-wSI)oJ6Iwc>Mr9`1srjkHm_RO|E9>3bwXuZ0fMN
zr?zJ0(G4DXS^7|P-t@$a5vcVt{UAvBW-`A=)WY?zng`6SmBrHEa;z>fa4u2gjdd?{
z@A}Blug_uTk3#oOS31IQ9f1%N&=I10C*03kXK*}mvGAwF3o*%-48aD4%xZ_3r&n|r
zVU_gd=;Lf4>zh+M`&1bs&lY=nkh}D`c?5=h;TQn#2iFJlNJn*gJDZSa6U9b!^ft39
zO&^_Pvo<7Ml5{|_C@mk>NEWEK7(1|R{RmD%s`StJt|At}>p{g$Y9W?w>INz>Hfyu>
z(Ck(vAKVl^GJdchDqD@LglBDsX^RY7Cgu7?iqL)@!FBA(c3tl&_MgJTr{#3A4q~O&
zt_*08!)#&vS>NwA<w1Xv+fofWTXVu@=;&0g9w!K@kOAnjJgs^@TVJMtiI(QN=0U|B
z{j&XBu{ND^Sc#@?Sd=Db<->m{2Z~e4tXVQf>+(7u`)N5&l?Uj<<v(-23W9dTB&}2)
z{MOM5BZZC~1()n*wB^?8F(C4ykDiq(ViF;}Mw!tS@h{%Impy_$!jQwvv%k;>Nvk43
zeHG{`{R0_K#kba^PlG*JGBhqruJHR!lD2Nm5lSNWK_56hzBIP6W?9tOV{s-tYyUkS
zJogKLtnG^nM&UP=EF3F*A%{3GfkQ77b)ku5T-GB1fOoC+x_U<*#<u+;)7_eb?m$M9
zKI2GVaOU@a?8-2Tb1{VOXC0nMUN<{c$e=H;tWb8h!Z6l<?252M3?z|swtd*1-MjBy
z-|tM0*-}wP92_EX=KA`1vVj;i`V2<D4d)RM^BlNXbBf~_RiVKshImtv$dSw*4=Tqa
z2Rz0=*@{!I3dBtgH{nz+jNW4DS-h70p&_^*{+j2K(#*pYGp$n3`TG~qtvT+IV<u~2
zaT^85=DP&pihlZW<bjZ0-&v1`%fjrrAx6<JKa1Y+#HOA696Qb+uBnKYrQ}4g%+7+L
zUj(0FWjvP=fTRu%Yd*#Pz`@QQmMe{PsJ;j%BN$+EBXc(8_@zB}6pwoNM;y(%!t&%|
z@u(hlJTwA3nkZfC=qw5GSGr0i(C3s4a25Y!`v$QDhR-TaYqP@k;QR{9xZAXD<0z$M
z&;)B{uukbC#`0Kal}1W*ex;~)J03L443{w<fwUm$@!15H!8w+31S$sx7o>-*6ZX$r
zi(c}k-~TSlX41_xI-9aIWvJoK72$>xhYXobe>p`4lF(mM*+0A>4T`uZnD`*OGOo(o
z_4&NVCSw@HvOI^TQshVsUDokPS8F>(`k9!+`_Y_!N6%9kviI|!alX^$nSb~=W81dD
zbaUWS>=a8zp1)_&6?xTDua&tEHv_~?;<NJ-_jUGX&3_UHmf+>RCpax8@+EUv1EIgE
zZt9>k?i6EKH?Rc7<}?{Hbryw~IKaqrqphk+c1xoO;{Y9%q_V6iNihYyE1RAA^(`bj
ze!i|vOC+VD@gY+bc~CQGljx%@e6GXFvdK+o$4Ln1P)8=+@y-DRvyHfi$l}m$EO;v`
z{L;_&-mlJ%Yr{wSSN=$#Y|%yor+PZ>1=Y3@;({mt{>w1o|0Iy+-`Dd0{&Va4CfJRF
zL-w?*jRJ#)SS8*HN!8L+B3xXv6s#b{0uF}wbPRKD7=5c6dBn#+DdVEZ@QX*?n%Wi&
z;9wdN`jf}2H~;pNljg`xH@Fw7$a-RA*jCPz$zzr>fi<zfHDuPY;kR<8d_9;>xbpGX
zDe!JY#uT^f+djQI>SV?4_S<@?pIb#+={()x6p(<uW~tG&)iY$Up-cBpX8K}51%2f-
zr~O43{tC%xeEd<ImBq145@@*0=M)0%^R@?+)aL^xj(Hel6mt=73+?AHcPJHN6Rk(7
zv4Hxo%aElDH~@bI=9Eq(uW3RyEO!5KyIS|z^^@D*rTN4VgbY4nvI4$zp5wfAhYtFT
zBO|jPA|<eN73n#?S`<aQE8_Bx24ebje>Xz-?3^hrUx~j-5(mB75EXO`6>sn7UMT}F
zE%CV;y`-YxNid>j&xCm>)sk$-webmQq}3_2OE1ufx`@)iYrI{Mbf<;f0-J<R8dD38
zp=LfKqzq84`nwaGM04?Q`2NAQam=Lz92lqR;W>E8o9ic`qFjWpcHHBM*JvFX4^}bm
zuQW8!)iR`vBShxEGsoyO_e{L_WWozruuLYAV?%7XgjztHFCqbNQr6h2get!2AmlI%
z=1eHDB_pUV33-B`a@wy;ub(fpJUGJstq{Uw@G!~xI)@*`Wl95@1ZFhIfdU~ToE;D+
z#t}4>myr?<y`hVpK!(PxRyoJ%awiTu2-B|4ar#z0>bRaz!u<1S&m>bCd*;<pF_wgO
z$gBc(zDQ0c<l(HDi%4c?lP4!3R||RJ3VBKrRciEpl+>e`R~7XTmzTn$H6OyDFr7zj
zvEk{dZ7xl9>%xlUIyr@&qt0_0H@UWx_LUCHwNSEqGSucgsmBHYjh*d#3;)H0+S}}j
z*XD(szv70(eTmGxOT-ur0ZUFlY3J+MKydl2%%^nV6x!mSM|LrHTgz6_&+lTRVa>eP
z;OjfQ^F+4Cm_36DAq8zHE|W+gVp5Fw7}X@=YnH}RNrn9|Vz^y((L!-1r41Xp#D)7U
zTw1!yC&=~#$qktGJsDm8JhGpKfV5SnOm7$Ai?9yUw47R~e{b2@>^sRn8Pal^9A?V`
zSMV5-b{Tjbn68vJfi&n;TtJT#S)cGK^)V_Qz(PYf{C=t9y&f`jfoEH1)0WKjGE)OL
z28V5|Y3xC#HnkzLx6GnJIAd8~3@jzHO(olzw!iynGPR3aKD?pm*2APph9)aAD<MU)
z83Iz7ZK($l`DWtVmdo4So)>Pop9$K<sliX?i+)!RRjOgS$G8VtWJ{BTp;BtpB(kJr
zCKLraC;DhDlc|&X8GqUcDNE*Q71mSEn;<F>V=s9NvM4dmm#l?JAL!GM=8eEAEHiJ?
zLql$|9(l&dA;!^LMyz`sf`HUSMAy%acQJth#*R4pOrA2Ocgm1OmySuUB_+h#I&mRj
zJVMfd+*SvzI%&VQy&^4n@*n?V$EtUyCA?6&VaKpr{Q0Q5qB79H7*o|+rw_0^B8`Zx
zXR)xzEZc0X>Gb#Tw(TJH6_@_)C2R|_uEerP5z+G~y@s#YE(UoJJ!V1n`PCr&3MpLx
zR1L)-D@cfdgr^h=|A2Gn3Nf<N_dij#lTp}JE+b{sB1m6pjo5DlS6PmyWV&Z<?gmLr
z<S=G8)R<<3nN<JAYAC_<VfqB`?{u^D5B7FvQftbVLim9%VAPCeGSOw@I3r|*dKBfU
z^*tD`Sm->{5jq_^;sOEa!O8JqqT%AjWi;yu({drUe)+ie_sQGD-HU=qr&a6DbGhq6
zxTuB{rJ8z|$*FumNNIBrdw9`!2%Nm5b)k^x;+h-F2j?svH()mwuGA}xR_rAur@RL2
zAA##YDlT%CcriC2$T-gDoC`hyzHA0bgT;V^;Kw*lC%Ha}Vibq#A={<HAM{8VQdct$
zJrqEcp6nY)WW<1(auyK8`26y`e*<ZHP=QYXAsVdhQCTGXmXBwj2#!h?I>BVTA5}-O
zY-5x-GQx+87YY%zc(PE4=r<MezpU9Pr{~RpiK3JhBQ7R7E3S%>82pY2i%*z==d*a1
zcQRx|jwgByU4|MghA6`;ACq1*{9ieqM1|~`^H*&Nj25xy$g<%Uz936p=s&ARKq*O$
z9y7=1afMmR@vvnW;-hh5jc^q8-5a5hq#I7a<u6#X?t^U=bfz?=4Dc}-gjixb9;8o{
zkFzIRK|B<ZqJBb;>ATN)!-LsB?m?`#vQWXdDe263j!DO{!a;ixt8L{z=iivXIybk{
z*}9Bu{^6)4)FpkAEH@FBWx!SXSj!{5+NxXtd4{Tc`}XaR7dLvAX2)c#oyAVH3U37V
z+Z2k2u^LL>1Kh6vt-UjiYWm#PxV3h6Xiu$`ia1v-iZTz1R%=BCL6CV6X@nF66ofDa
zdi1o`ieQyVf<%i11qC4@vuIIh8I>@Fv6wPMAfSW<gTQ?@c+c%UXRW)|UH9wtldD)D
z|6#v-Kl|CgZ4jDyAp&B?XQ<gt%a3G6!O6LpepadMD=fhQcOtL897|#^9U7w5IYHs5
z;hK20a49icJ_nIdeD60~6xdxa7FUI$B;JkS>-a1aD9SR_mgf)|1hQ-qUuSM9`G7xK
zy3}i$D<Dfv3a&s1Jt<)EMS2Ad7;F0$>Q8x!d>9c~2`pN_bhh(2^t-{x9N;d!b4D<>
zU+6zTl*2t(MBQo)5{5}XChzjKl4vR!Z`)l5#!r!vI%IUU7>A~*>D7xtMe+D*xR70b
z8~SkvO3-O80GG60tPFo<L9^2YfDx3a=f;9s_8EAO8Za(7N>&w0pEXqawLx%@2TdWg
zX(LE^DWH^P`t&_sk&OSsV>n=t(P1JHGPhLTh8rGATmd}wr^_u7S|<#5i2z`^qa?tQ
z&aXcN$D?P<f8z=;z!iu)IKvgNo2?(~TQ(5j0hPMVgNgTOeo0*9*6{&8&mW18RVv$I
zV<c@q5ip2Ev`x&J)3lM{tNVb4$wgq9_`UqJ6j8I+&qQf_3}}RNq=mH^;io1}npeB;
zuq$Ktc-ZCprc`<<p7J4H-Z2topathc!|1Xzn<A4rID$O?#8Y)ke&D99Is?tI`K{H9
z|HM@+PkA2r3j!LeT!{Il+!mjO@sLGYowlJ+FZ6O<v$+uJz!SJpqartnhqjN=&NlbQ
zffxB#1|EIt?3Z7Q?uI_wnGDy60)}DEm^IzAV9}djgSR&jiRqRLrt)xWr+nW>d03UE
zIZ6wi!1m7kJP&<382;RINfHcsdD+$J%Gx|K@^PLo++h1Sk2Po?+u{g+(|!)!)pni7
zA{Y|on4VI>yfj09W?q{Aw;!{jH9Phl*Edi>bo%qQ1c7dUKcY;1%H*wuMlUrVz!_%u
zA;`W6G^&7_6BY&K2DOTi){*s>eEkSW8PdV+;{%%1w)Y+dOlsC(XNfcHxX;=PbIH&N
z0Znzq3OWp}@JYQC@!CgY;kpv#$QSp`U$*Tgq~O$&`ABRbtc5AWYrz?(?i8h)=ta>1
zNTUoD7uQC$PK+t|-bkN-Tgapm9%D-8m}&wQkeD&WSeHyuS8=xl3*ZpR^>fRH(nWo=
zpbSKn^I;o<5xyy#hN)8w3nb;7NMCc%nc=K|mX(!7Eod4XZ5$tt!FOzNg7C%89DRb6
zrcka$R5GCNiW<!@sDM=XI=UsnZ7{d$lB9&BzfPn>h8_Sr@{1;5H&gB~!!4T_;ZD&6
z-oW#Cx*RZft~<;V(rJO|>f41GILl~a9p7gZ!|p^U#*ucOm@g<mkl~quKgBE0P=zE$
z+nk#>^kXX$`Kx_6L?YPAu96LI5D7RW>7bTm-l)be0SG}D5}5)x7h_xIE(CG{4*+mp
zwx-kbz(%t-##WeOA9}i&CfoIII^bZ8YAOE2C;DvY1szTAC?!Wlr_w?r!R*xi95;S3
zpMOuLz^*OLdsCAk0s|y{>Zm6e{0U@L-UY>*Y7IJ~pF@-R0-cOxubP2Z=&_jg6Dj$}
zGiTJHCo>E7vqBqj!_;x)Fb1#+Rva%9iO#J}3jCGFHX7325}6D?U$Y6Om(rpI{4=<}
z`PI@E?0k#&<~T7`Xpgue!~<c>JEyy^N(8V}1X7X~Ctcu*O}v7FdwfWf<bq3pDudY)
zm&^y302gElU{QBko35KD<Idp?ZU>iOjcBOe!5Gi*Imlyq(lAIk5qO#N8{~+28aYJo
z;7kk+Dmg9-9sYa!T@*`0BSH_!`%)!Gq)<iGfrP}(yc5#NmJ9C`!cSJ%Is->h`DT4n
znW?g?v*`@nBj#`GvZ7Usv`?vmJi{Aj;B1@UFKNlC?3&z)j9ozQn)<2AUuZBABiYw+
z(DGW5WPe4Y*ouCLQibeT7i=hJaTAXVRz+?71+L2F6aK!w+134t1C2}c5#0|!$-XTx
zfT6SwXAI_>8FAy!A2FSv-GVT7KZUci7iuKhLOJ<DDcpw7p7u@k42r(R&mj%-T1l<m
zBsgv~T2!hoMS9fp`ENM&>M<DO#QtDg7Hy04her0?!x@As8=Pnw8<zoUhQNHGTV{sA
zdzks-Gd4kh+lEL$<2i_Hl#0>_2~_R)MkUyk%XQUat`XG@vwfta(TXWV-FWX1G*rQC
z(h$6W48A3F_I@M~+95Jen(sp$MpY*Q2E!oP6*MwoFvA|dT%x*@E)vAJ+TUpZ8e&uv
z$j`lXGSIkah<Jurfd7l$C1bBZ+L6bKD=#k}Az}ewx;>-Vncf}4QnkE(c*Fqge$~Oh
z5DPY~C1OEz!#6huE+i%n40nBori7nur%(}LmVI|-Ofmb9ta24%YzhWx3F_pDnfb~g
z-(}E#Z4l}%dpCiM$JxFTzpf>|SUF+aa4WouJTCk&<Nz$B3l^J-hIcBChJ+VY1&-@c
zsfuRL<w*po4zi;gGlDP{FG!6~y>{(e)Vyj9BQX#$TAWnSy@Y)=I45)E*FVwTwo9>G
zxG?ZGhQ+obw)`TN-`n%^N;)2qC#PVZdp%<KCtp}TCVCXI#Pxd3liX7f_4{aG{@bni
zt^``N#C*^9r%zKZOfAKFK)(bAGM>9x{N^2~-1voTwCYLiMB=uWFj$DFM9cuQ!30(v
z3pi2QgEO|q3Y*YpDr&@RZxir%xh8QLuJAun>_tJ+ykQw02ji)Fx9^ivcO<YExHy*?
zzcV;d+IuGAjleR|U-<<y@hbh1h47JZ%_W%{*PmIe8W;LEqx>J{*errSQ<yS{&dN4U
zlSKJDyuw!KZwLm7%Zu|UwFGNN9^>fk@!f6>6qs1b)kO$K+^bwWaip9%_>)7eM-n{Y
z@F54>NXm^$hO#0|JVPpoxK3~N4Ww<*|M!J^$rii>xP6^y=p7gAcVJ@}NKM`~pU>dg
zW_`o^`#x(Vmd(n!x^x6xa&yfiqq?q>uJ50}nea8er&VfX34#JxWBfxlpdb;T-#U<S
zLdu4)W65)ccVE4I61PTU;`B`%2*K#}4rGi&X?PVj^{!1#WB#i3>-W-ZeyMOiii|4h
z0w;8*alo(3jC`bGG}Sj2^7hFF%(^uf$*e6_GPb}WaLB!>a|Q&Vf>Gj2m3w`ncWsC<
zzfx`?T%)f3q$`+Nvdp`<JjF1h>dV!UzVqJ8*l=9u+!zlfZ)*{&_tPE7{Tw^)0w<`G
ziP^emzE0`E;3WydQAlOe{5al;Bkn*O2nlU436GIuTJNB!ukkPqq~ihc$!8PV0Fp|+
zRa2|RWOCre<QqM9U_4$$*SKAxw;MBKK84gw^eY{9(0{)m@s|A0-95;qbMw0-x2jHl
zwSI(1aheLWVh6n)sHr*iu50x3z<Xkxm!`}uFB625`RFa>NYSHNhaLkSQUMardW<p%
z1{W+fxP~r5i&HM82k12ERDugn>-+TP6%{>r5>Vj~7u%+$F}dpLhGCXfS?!AO$yfip
zYz94d=7Ru{I6Z{whHSJ#ttOJm%w3MWftA-0^RX2MbA7_Tn9ZXyq>iPuU^icURn`XN
zz1lI(a|1b|S)pT78U&Mc;v1k@a(1;Cnf+rb5vgD^FtA7VU`^0nT;L58Z}#@r2IJUH
z*dBh5+`z;&C_qsF;CbR?wx4$2h|EWdRuyk5F(DyA%wF<Cm$aAjiqHmHQtYSXj$&fW
zY(L&r%s$_?+Xj#QJY~Jd$>U-)J?T(OEx(=oM-S510w2*2#_wI?0I6`e0GyFfUcbiM
zv)s&-q9?;%#0l>N*^JHb4O9;m<<X=SB)Zg1goOk8<~q7Rf)Kxrj(M8}7eB;~@C?*V
zZU!C5XZnCQxOrU5u=8@s*|BkrOAbLtbq<%YybWbHF^=_5s>aMUO&Si^5aX2i$g@U`
z8F05F#$n!vc&(Kj?fw|Bx7U<S>Rs~bz4w{Y(B-b`@xnQ2j`_Yc+3Qhv?M(Fx7-pz>
z2KCSegG83M5AUFLha`v!<`x*<Q`|(pK)YH<N275i8knca7d+f*h7tD%Zrbhc*cPer
z_Ca0mw%h9g%My1til55@kG=*SM1E{T<)lPZiZM=4l?KUtqp0@-y1750Etrr238~Vr
zb{Uh1D+M)|uQbK!=F8-whRf_t5Mx|)AAATC5;55tTfyLJcLm!p$F31N!t?^=NS%J=
z_C#&isDwI^iOWR+Ee>(VT@(Kea}als01y=)gVC66jOi-zJo%d5aMeV%EaO$=is?6Y
zR<G&q_RA0LOf68K&}I!Z%&k$`4@_RgQ<WC0=9Gs17wveQpP5;!2BMxN8jE3LUGGG-
zqHH|jzUA05BgwY+w9+SDUJynfc77+q%y2Zi+EB}zhN;L3$dz4Qs=Pm$)jN=~DL-Mk
zEM|)&ue2-JC~id0{?jE(IO2d)#X~|oAE3-0v9D6|kG_T+R_Gbh5AZ=-=xgwSY4r2;
zejRxcY7qEPmz#z!g)d^*J-`~%n95nvjECkuKPHEsw*BI^Uho&!lh-ng>+me^Z^Dta
zPLLvy(UUR9=DQQU*{327-YlNTlS57o%rQK66kTB4XL237qoSwvCOEY!O_+7S8j<f-
zjES-l!Jet4$PV}}IeZmF&Bug)L~e~#`YRGOP~m#kwCNXY`fe4T3h&~rJSciP48Pp5
z@v11{$9t(L@SXkSAdUF(TtRt+n&X|@-FNe<DpR7&o)7UNtkClJxPRdx=++Vs;Q=>!
z?VBt32w3p{A~}oY%5u?=<S2cL-=NlS{SoVzsx<n)G%K2o%iz0@FkwNzc4k?!eEsUm
zg+nnBae{&t$*mug9ltuMdOv@}NKf=>IoJrfrtbc|b1xJMJSNs4zYOlXdRoh`?*O{n
zwnQC=YS)0rRq8dj^Z+3fO-H3EDA-yY)o4tT#*z@4tzax5x@31Pce_B!FxwO}Ec3C%
ze4Lbv2UI*9vGa;%;a;YPN1A}IZb!)_@!<q7@Qn)Cou3*Bn(3cuZ9d_5mk;GAY|FvE
zJ~lX-TE<fgL$N$qIA*G$ij=;6?T+?8``%+kHbPr!PDy3E>Ke&)@}b-Ek1f}~Pofoh
zgi9hJx*-YAuJ3{V;dKb-b-ka%^t`)d6+w7h)l#QHGoLVS3|OK(6ZlgZQ(6$|W<Lc-
z*zp_TuTcRYPaF?w#iAWyk<NH_TX0Y3<s}Cx_3rk1b1za?@&XejgnW0-VX^_cTv?`W
zyJB?=0x6y2Y7<XJq?))GQgkd3aWT<^&m63k@#S%tW#VUe3E%@aN*H(A4g2lDrh=JD
z=r;imnsVb5Bwdpx;+lA$d&D6Nhryaq9NFI6YbNm|D3G0uLJJ9G1lq}nh?0THIgU}k
z^8Tyxwfv!jXhLIq&LdBibE6mx?(7+b!eN-SlmvXvB|WOHVUiHbtj}V?Ia|V_(blT&
zXuW+aV|}fzAX%a<y}Si2&B*`UEwq6k>{`=~yxjMnR|h(u?=(mOL&E-wemw4%w*4+@
zTps}$z5#k+;$1Ep@>G9pep=$X)nU=PgD<#jncZuUg12EZX?_4H7{F@~rNF^NEltk2
zxk-Z7@5w%Ioy14Jbm(L@Hjr`U&gfSeLLfMV0Y=S!C8UrQ(m-cu6s1?HI)qOKD9tRa
zym;{<9F)3olBwgc5#h$|NA_rqHidG>2P-g^CcQ`28R;0iS%AD+;O`vvGNoOp@`Fns
zvIE^)y+gH?Rk!2;1r0cWnR4#OdX0qv$2hAi0lr8huTSz_IPZ>Xet?Qrhvq+DU%sM#
zR>bt>yEc9KRv*BFMbT6+5szCA0!o9V0J%*<jWQD1jA?(iVl9Gi$jbZE9kbf-Ns(V6
za2~&Yp5BVzAE1RJMKyRG^G#c)PzJYGNSL91w@Q||gdQy8{KKFiD0x%6;@Vl%WBFRe
z;W*J$xLvq@#GfSy^~5OLqEf)Hg0${_d6;Eg-67};bwRDBvfb&KH6b?Kx6u3`Z6-(Z
zNJSFdNmL|t1obY3<k})N=|yqldg5-!vbX0ef2;6|huf6&aw1qzM-P5Aa*;yupq>di
zcj=Hoe=%4lWb)TEv&zw_JQ}>d86kKSl_6rRD_NgV#L%i?Y6(FB$a28G>Nf8yQ}`E!
zY~c>1Fec^2P#id3IA;bxTIeVeN?sVL1U+<jIueUsDX`8RZbHOdg0bed1c7$uIeB;C
zx&!xJW?Q{xE#xu?@oj5*hPY*_Wfp)oJ-*8O{TiBs={QCSR)lCx#t<|xc3~`;=N{s=
zS3Ly5@yOU7G_&UI<-xmAX5&=oi9%C{L8G9hgXQOzF!^kF0*xmU0j%S7z~1E?H-7%e
zKTe-s0xe$i79YwP_zWxA;|Kr@)Nj02)Qd~dEEc!Mtuy~s&*$ks#4_A9ZD=zZ=ctP1
z`J)?uls||jk4bLzPtYBf&_n5idN#H7^EGRf;PBNE%_Oh={j~}Zh@+}m1jcQaAZLt#
zRfVaTOKNf57~4mA5UR*;Xc<(?X!}p8+=qx5S#Y2x(?TjFvIE|v`nE=tU=T$uXYdSm
z4%JJBs9#&qU|Oc$X%Els)7ugf)Q1Cj>E*OC)yca&Mk7GtD7Z)VO8X%sf4tH8DqFv`
z&%}3S^%p%aTXKl)JUzTWMoV$}8?Kv|-bqSDdH&7F7JXG$TE)J^hVJgT9qGGXR6oD_
z{`=Xj$Y?l4I&R0@u>_-Zq76cC_yTn&<nx|oi$43zqX<lx-B>?V8jL56wOs^Au->5N
zj2`)+u`7|M+b&*wN)%b&oH-U7y8D`+UyE3xTXQE99nip(DClK*uO@FUG6KW%iFkZF
zd^KhYH?SIx0=R~!pdPSwYS9zul$HW@SA0VJi(oug3!$u^I8gy!D)m%&UOiAi&<ouY
z&PJq1BgCCLq(IB#6JvdK5Ma!XI0!7!F9z(MracO_N`R7T-ns&}N1YlyXzcYitd~&=
zg<DTvGzw|Nce?`s5B2-dF(6$77xy~C!A_cQ#;aU6t*RcArv*K`k?L6!cn{oUs_*WE
zzVsy~BVJyjFd#x*!H?+24YolSxttju>Z1%hV|x$pAw-a8Sd~ep#i>bo3B9z$288Pl
zdcPQkuLNKcp%3~Y`FRF=asy7%H=l}C!8gCw&hh2qr@$6l8>>Ed*|xUsN>$|`jPmYB
zb#ioUxXS}K&!(j!4R)~V?vRDH`=KLgVfF3?vrSQvk$^H3cc+*KHQ|#Cgf&%yi-}sM
zbHS<#eV*y%G{|==C;V}bX-zvGFj{1fHdi4T;cy*sG#f)Lwt}>ai2Dm3VP4Y*SVB)b
zX5iw}8x(jDeSq=MqUIwb{a8E;XKde(1fCXIyLN@e5{OPu)^bCizK}gM_a5HcQg!(4
zrd!z>r~1)c+Kr3rR5K?)T=xjOgZvG6$N-fH`cgRvS-CA9?uBX%Bj(b1rJL!suxNGT
zpLaCJp%takgRJs+ymn1wZ7}xOd7pa;B}Wr%tf9Jll^G%w_zX0=bePNAH@Y!Xvq!3u
zT^8?#x-k%W3?EM|_vaYRDD+7%5a8p^)!(85y~$%MK~cj$hkCEk5!}CV6`+DH+l4`m
zPc}~`>2HP)K%i-x5skS75#)a@+-PQt8H?**M2m`dLqsteBAfuhS8U2f)k=;SN4niw
zgZH=C!0lL8-CVWOvGI5Wh}`ndvl!xtbMtQh;{3F-51950xdBqiG`ZouhqhmA0qfxh
z7w7?tNgFisg8W(P*OQ`@{>Z%QkCQiMzx93}v}C`8HS=fi1=^tM#w9I3c;(ACaLUz)
z_fI0oX~97d<@Qz)z?f8YwT_0~Lc$3B_9D64ntUL7eEVwbInPX#hX=J79PE{a%yP%B
zbkUv*B=w*-)Q>BBB4{fT7W6DG!SCUscvm*y#^MjbSK->a7!R!_beRpguo$S&ws<nU
zRO)6F-GGXY_8Aw_`mhP%91xVLutja-3qK9@Iq}^4!#B-_CP95lBL?9d@Os|Fi<~#=
zgjPTiq4TH4OQtIQN|L_(8T+cxb9S>QoC7+13YT5_W{`{MA3?vm4xvMh9`HD<s-q8@
zuXUo|+WIGCSVcpuvj|p+GZKnx)F-41ViW_q!#<?X7EtD#(&!VAm6-ey`=Zx%`8q!|
zmpmURaXpewgXTDy>@jFeFAZcWG1@Iy>1dAUMcT}W5~CYxg-5>4Lms(B@)#~7!W<)y
z380q=)kh|dihg{Fdh3{LJO;8~W3o^0?-z~GfT>6EMM*jF8)*&;oy=mLoG>)b2}WI4
z3A+wqY>1c;x5?QWnF&lwD`-3bAKMb)ns%;cN(bkmlD0q|stw(MYDw@>tl>Ao$VVC$
zvh^XbSRwKiYq&WtHLB~>tvG}ZCZ!X|5(z{n=01WBOP(l<*x3_g5O5V*f-%SBOV&Xs
z@3apRZq8=6m*~t>$L4g@16eF482W(6KLA5?aor(cL+U|Qpl%W*+ZZy+P=ZbiAMmS?
zYYD;$S)zkcTQtBq;~B)YYd|Vb3#yPGG~h7211l1FGiY-WZumm%av(D=a$NG>APKsO
zbIUDw<s~J4YW<!%2zzLKyVyW=&OuBD9po78)PjtB3-^@gd;u%vR?tU6i|zn!H<1;i
zKJFB{B1tcRQ-;?~7|*e5+wwB;2({dHIYs0NNU?{&9%)1$h^ENV$@?0!lIG;z!oUYn
zCWsT}u!FUG`Vp)8$T<H8y^5^F#nk7vjzGmvH3|r*XnsxAp~p@X3S4DlO3-n|t?;T6
zy92k;3i%6yp(>}c#tFI8=iI6fG;e|>zWjxsf$nkD$W^sRid+V(HPD+2UocPDhvbeQ
zqP$hnmprMGum<Z!n<bNuu*#L3cB}A+Tledt@Hc<9QvNClCba#)j4>YR{RCt9@>$dX
zmUJ(ItLz}+)Uw@k!f(speTirU&Mq@-@wps&010`^>qg>36Ds&OvxYx;x~^^4G?GC=
zBd1dzFIvRIVY5*{a}#WA-h;y~+ktXR{N1DyM{qiPykjoA8g;N^5&o$tmCMzu{v8tW
zh<LDoL=b=%Pt&W-E03h&fj**0MAv2q9>rR-RoNNnSRhxmyA+Fq;%kh<;D`MyBqH}W
zy!6{ps+m;Mcs?Dm^eYK)<ar*VcMK+xd3mYJ?H-`H6_b$Q!QX%}o1tQP5|=)o_S1DA
z8V9}O6V04G{=|6JAMY$Ml%B^jRr%&W-cdT04GYr^G<(B7JD2m?=|4TTCXCQ~CIjm8
zKS>q$r)X+nQUz17-?@Hd{DAw{v3}M2MF8_@%w1JXJ>Jf*PQK-%{KhPAn^*I8S^La=
z<2&vD+f-%R$b)e>_LtI*0Q`LX`Tsdhx(45mrHkkgLwFjtk>j$k5LRl%!QU*#SF4~e
ziF5Crnwgu0s&>~?`1C9fgFpClcn4Yvv^%e;L+!7$kD0F$_v3&1+s@F2VJs=Q1_60I
z4LdpMkFCkoo!^Jnd7pP#{DYZ2Mr-L$_8b5EZ~j03=>Pia&eg^JLpx@E&a5Ow_`RP`
W7k-~~cMJWOy<eN|%Gi1M<bMEKba=x6

diff --git a/docs/dev/articles/functional_api/unnamed-chunk-20-1.png b/docs/dev/articles/functional_api/unnamed-chunk-20-1.png
index 9b1a6efeb080d13ccd367ef5fe668bd22248b0ea..b5579e26b447ad5b20b115a17719554e823ef559 100644
GIT binary patch
literal 197539
zcmeFacTkmS);&sh4{eW(Iv_!mV1tT?fCLGG(A|OvC`b^IXoHAI5>Rrm?Xe{ul%OI>
ziJ}6c<ftt{K|q3nL?s#s5+q2vYaeIko4WT_-CMu<RsH@Mrs|W1a}MwOyw9`u+H0-7
zzo5f98f#W>T+PD5vPScOnm+z9U}5=w_YdFWC()+a-uUNxD{T!mmKFNn&9d}J7M9H{
znrc5Add7`(dqf@i{BqZuy?;{|WqkYXw==<kNr7d*{wQR8;dapNw?{S~5YpSFm9+nU
z<rX>4ob`ug=kEN)>Sy8V_P-z2`-b(KpHu!SY<~Rs@ge_!fk^qMWga_Z9{DBYHJB9H
zuGE&xcszAYENOhEwS^-HSMfhz?T?RSjQr(){u|%zxNvgEfBs;z@Ts(K|MLgZ|0iBx
zn2F15Hwzco+ZTP*B4b3Bg{A$Q`b5==2Uug4<%;E{g_`xc%6hNuSXl0V_vKx3xK;Sx
zncMU_mgBLhF{_VndU*DIv+@1oKii~#L$~wDDA%fLb>49nmi*wbP4Y(W^x*Cvx~u>C
zuk`DGe;NLz-tgc5fp31=|KcF)fBxXae_y~{j{mO1*In`dy$=7l6CIC@WEbZ<`Rq!3
ze7yAX{BUAT#8&McyM%XWhj5E0S(JH;_D>BrhG&<5x$~_I^ZeC@nCwzL%tq0Kr8WeI
z^mxg1hkdqsrb(WDb-<UF#Om96$@qle<ez{4-sRn`OKG9H@q2c=y18lR4w-v&bB%Tu
z=<)`<`E6An-X`93gFfg{{?Fo`3x;LhpB3KzW4(SgkA&F<nUl}AH{Ce>n1270R}*&o
zPZt%JrdzD%KD`y~S6!KJ>@ONlxcexSUcj>5`P|1>VF4=(y`ud%a>mo|nv(x{#dE4h
zb|^6q8}pU;Kj(vQ#?U$Y_t(=OZe;xXKk&`@(ErB^j)>ti&X1}a85zkswC}nh=eY6W
z#fuZ&9=%niUQ;_9#=Sp(P><YpC<*`O9DjVlIQNtuCpWjz(`8O^B_*W~FYoev{@50N
zW4HbO&mUe2i;0PquPiSZ=bb(>Ff>%%k!LTaRPTDUz^VI?#;?Ery7SnB6#X9i{_3z~
z&B$$dI{Z^Z^&zZm+=lO#=|V@0-8&r%PIctkwcb1CW+3kQ<c~F64`wAwMmP{f7+B}s
z+>#c2Y&;U7_^U^s_Hc^hFS>W1YrAb(;oq8JAac6(Hv5sgMqw}S9<8}^<gV502R1pr
z^OLnY54#-2^4`S7#kD(EEG3<+zi}ltcGJU?^{$x|q0e));7OcFG0HSikKc1HqWV>~
zr64Y|bDE+ZBg?#`TAjz76Az=?VdT$?Z2f{Oa~g{i;Q^uZQ^P#=<4d!>sxw_~jZqVS
zUB6Au!oosLjnBRP*QPxCw!(o3?IgYA{jp)D4^C)?ZjciCel6byOWz6goEeoj=f6v`
zYe^IB7xbIfyLas2Hk;DZFYmZCYl@_XN}A+uFJGLjJCtqz@(x#~X+h8Qmlyli9g6L}
zbxB}jdc3<y?cBh#NNe1pUEA#+MB>i;lIzsHXYHPMHxFsFq#a7uif&!!FTQhU=&^?<
ztI`wwHzw~73%(+(QC)oQqdr4*`o%ul%$n9gYC=6e<7HNso3n8Lah`oaqWx^#I|E~6
zY`N_-_niAsX>-k-8z10JXR7J`uTN6u58aRz?eU$e-yr8;d@w<|$k<*ZYGgbaN2@7a
zH?eT!L0$fJH=Mj0ZOJmj5kK0e?NZSxJ*~NRE$?1snw}l&Qk>}xSgEhhv?$wC9m=~t
zV0kLh`p<v-wW_0pJ|#&#Bx>S)OL|Q~um2`?_B9)21Mg-P@7Uup##bLJ-|NtYo9N4H
zvj`DM@D#G}8O^gAHLvhrQW&}aM8Dd$A@10~njnVCmv^_#OWd2=@nWmRxrx|#rHRV*
z$tHQH+1X87n*;8-^jF*2KXw~#2z&j|Hr+f~i0{m!Z<^8$#nzvGyf9i|X_#%XceL~(
z8&7qLW~9~g>rxXV4;%M8IXTtl(1!ed?Vjy6r77noa7P*l71%fTj;*I>ne*y(N8X!=
zML)y4M&en;AMU*CDf8j;8n_Z0R)PbP>@nU`)x$%3UPwsjn@4h~2I)FMfoBx>)cNin
z&DL2cIQy8DCnbJmqDC-1PAOn{QB7?Nc1H9DWzRE7_wV0-TG74R{?!(V;&(|_gZ_<s
zXH(2Sp8sy`0mDp_-4oX;7Psc$)`Vso8XI-HTx$4(b{x6GSu}h8o9~*l&b9w~DK)<0
zV^&F8yz1F^FY2oWu~pTFEIDiw{bslo%O-2Z?1ulgVb{$r)c_n8_n6+gU@kEZgfXoX
zHa1(klYM5#wk{t_+0(Fhwky~E+*nu9)7_`rvf}4j5|^WlbF9>ZdTepp`!4cW92mj#
z8=RSCXAhETP!PtUi*+^ATlrprFNPpnC&NCwEWdeW)8n`80*nW*E`%63*Vfi*nR#@d
z3&P>yo-Z&pHojiCFjf*0?Q>}@e{<GL{1}UYjbJuX+!@F&%z;oK={n9PQMh?@PsQ@*
z>=3j5hm9)QiUAFAil(8KICVKa`&<^^>jrdKE!Ey;o&DIB{e0X8u_Z`Dz-x<Mx=wtG
z51;R+$9CEKwk+VjKD^3|eTtPUdPS&bU?bNNHi@y`8XS`N>k2MbwJxUpvtwsQeyHgx
za?_rc{~K;3X4~h?sc%@Mjeq}sO&_8`xL9_n^n(Wvg83{FWHzYydSaz-J@;w5ttY&n
z!N*&%H0ms@q*U^B;^mzqA+b)yq6z+A-skG>x9W_Dj9kA?`O{CK0(-q%x=mI(Wg3;&
z*Ikp;Rje>AaH{J07@;kp$!m%bySOwSpsQ4&9>P6=eNq!9-?KqcQL(B=UPuUG5eJu(
zwql&uP|VgFv9&R_3B1U7r$@>c#_VPZWt~qtVQX7spu3lAw%fyUq;#r5ozYlZ8~ic5
zLS=bzSoK=na>dHhct74L>fW7cUBj4oW$LWH^yPD%O+bD*;_H53MyttzXPSu-a(-W?
zk=+i~&YKsX)il|slZMyZ)vV=}-ZHxAn91FR6Xc1%(|<dep;aP@KhF{s<=pu#%i>cV
zrBP1j(uw`K4zC}mO+9${uphT0Il)>j=reFdL+fSK%Obbo!r305x?B&TE8AOqwZ?D^
z85)9qXG5m!+p@wQ|Gp|c&ZEa`Se-(gSpb4#uv_C^d7DM#(vZ^9(x?eUwyHvx{=PTA
zuaj`iK8r`ZDZ66XyTdHl8mA|8WvO>1Y~$|JQFi=vWr2a7b8j#8Pu59FPL$GeF!s*0
zOON_haUHFU_vks+=Ou2E8`x&yBYp=T{GsgS1+gBLg^~5yrIT8%caGSHwEw<N<+`f{
z_E<lbRqT}jkEO3mZF?xMOdo!8fVcP6>Bp~Mzn<s|k*r4aI+Avd{_JL6g2Sr%*%oCA
zeSyMSyFHNf11%T(cx*aW1|oGOK6mvl4#~UJil#?-m$SV~J!)*cfufhFL0V;dj<q#n
z+{EBD%QQ#BFOLD`c`0l3E#Cb3cf^cc@8j^qs#g{pS2&B@OWa4Tu9>@Q^6GksG$Y_u
z*GX2yP4_JaP?kD)eb<jF^8=B_1y1I*F1Su9hxYpl?{5Ac{9Gw|m4N4FET8Fsf`>5e
zqMHgjs!P-Q^>_7CwYog}!)!WAcILfF4p-UpWoFd$!~E&THV-`~`mXOecbY!8n!&#*
zQbd=*C1%8_IR5F$y6}X(K29$FfHo5^k6Nb2wUznI3bkb+p(9spv+?!0%ewO6Xhm>8
zLn-U2%U9$S^#>r@d7**9!72nT?nK{7O<duvrxP}gX<wW-$vFf*ixjDuo9MTG`{%{4
zn;MsWvvr~9yoI!&@28)qM?2+B3Y_@QeP|6To9$6rtGf6};eASE-&}uqIM6{AVGhTR
z+@OLUZ?RF6XXSIdkW@V92d^14ya%kg!V}!m9d!1~^m{4)#kr`7Z@&A!Qov{Aiu>|0
z+juU@R=8XVBpNP)M<}4u9lou5_AE#w(O25B;eGC@CaZ^SUS3`k1Wqpr`fmU*o#Z9>
zd9!sj8=L&@r}xZ?9z31k-*a|50AR5F(DN86r>?^5if$)!s(X=noA6ZiM^hO_(lWLU
zp*~|pBJ>z;AI-LqJ>9zfhP<;t4Znw3+}UPLmr8bxN~e-h?o0gc#>xORb^+x$tgrhr
ztdRP@U$ZgFwSlgBVdSB12cKl=r$3`~#l(Ocetq&sYFx?Q)(dQGY}Uxp%nOjQ)p6)5
z%xSg%4j22=q}E0j7QMrV|Bj##LW=-IC@ho&Tp@*kcd9MRyw*$HG(Qvok<g%a84hru
z(s-%N7@`nh;>)xtk@&OvbX83NoVg2$xZsK9h4E{iy%#QA&>xJg#asJu=3Y-1<5Qae
zlA`+455;m96%~p0*IgACYg^Q^A9#AjrG9r?Wt6BvFkcB$pC__+KTvEKJwKhjpIcM<
zJMtYjVJn+1t|b6AoEYHWtVlUN&ZU}nVkp*mTi7^`YSXQKm!gVNOV+=?q9MS`ujGD^
zH&M>fB<{?ge>LrLJXe2XcT=`yg>jLqO_xWnpXP7^AdyRlU3yigb49%CQqoSd!d=++
zn<UHU`cCSu+(F1bJJOzGTz?76#q#NQ{9P6A*)e_sSb>4@?j52|<?};@@6%%S8!`|}
z_}rdLO?>&7ea&thpizCir>r0Gr{}B_a&r(CIOX1&{c5eqBE=r#0v=s18bPx=$4$<t
z(eW1+5fKp*x^pyJphhsjw>GEOf58!$McBr7X5{r?f~DO-pte0{-~ByVBV5AhyAH%8
z_M?`5=X?>CUFyUOw(z$A+$nsxo!s|iood&};^iOr_04oThvUgPONk(-6uXaZl<)O*
zt1Xi(8i<I>k?--eZn0gc$^UHfi2zwv$yhMuR6N|0L&Np}<F@hVUZpki(alHwgAOgH
z_sj@yxa#uU$zfDa$-q5QZbQ#cbrrfadmXuB7{o1ZQu&X+uGcC9L@Kj>bp9LDxPX<V
z+5T1o-7sC1`7JZ3ucD?TD;6j9cN?S@|9-;GZ|QRzJpxDXrH6@2JO5ZGnBcnTSkTRl
zz;0GFbboi$(0fa&YAV<3rZ}YC0U|`Ek=~{gSNM89VCkdfL=}%^io?y1fM~TR@63#L
zij;j4m`Ie=PQf;i44ALW)h#Rj^5Jelq33e1fajav=3Zv|b1jTl1T@d>ZZk9Z!1nH;
zZ6gDTboWqWVud==n2k+wrzqWYj4cYgCgiXB8#V}<JhXbU&OOQ!g>NoIxrHOlOYb%a
z<5QGIy3SR1bO&Z>%CWB5u=})O91gYUu%-X}VCw0#9rTbLUd_Ek9Xf|+V!-fwpL#7x
zBV1rGVV~C~Mb{G<L&#CqqOlo5%Pl9S`%m797CTzqqN~ck*JtM7p5a=t>`?!uFI)YD
z3<#q{8R%#hSr%`ycdW~MgBVq#I2^ZrWqGzDvf9FHFj_y8jjo<qVxD(w!saSCmGb2=
zHbO}0Pi+)>#+PW-9A6zPK5N%ea%;aD6%MH#KK5omHEtnL<2TzazgB7Fz5=d_`Se<J
zlpI@{aQ(Sc>lEIcPZ<{Ues?R%twD8Vu^lgttd_u;iJDrTZwyHD^ET+rcQmLz`TTm$
z>-9Aeq6kM#$r|aQ1;`0^3`M&G7+{>X+6i;fWkw<A-tD_#pQSn9dW%lj=;DLc3-V4~
z8`{j=4%HSr0F8Eds#X5Hia+Xs%5{dxnb$zFRM+;Qq6))vsS!<&dvHvA?()91no1S2
zd2g_m!Z<3IWnX0&q(5zF(Aj8O?(4OP!tX%QSiTusIpTYSoIVvcKOOUu=3@DYg9`MH
zx#ux5E!A$Wu3BZYpWZ$(uqa%443OIQOe8TB_+m>K-84bYIRrWFTo1j42SwfPo#KS0
zPxY4j<(_`IW9GGOVieE4kP-`N0Em#1v0`S8D{h6VHeABIxDtV_?Dagdi~7RMsKMB7
zgxaW_vY8IC8T=+{!nu6@<U<igYQ%>r-igFnZ;JmY-%JMftc(JE1OiEAtGUs_ZH`6^
zkh&^ienYx^xBX*6`#X*AZyvplkJv_;8u`k#R;q^2?O-Vmz}5~^@xl{WYi`HR{IIqB
zdn$uO5;m;jJM$BP$5ac+BjTPY-YJ?^12o3S_5grjE4OkFGvkHN8bK9hl-7x|`QmLC
zrY0w?-aPtk>kY~zELGqUSvCiuTu^fKtqSI{>sK1hHOQD<Mgbno7kTX(4?;zlaJaIx
zoiZX3)-omLv{(j$<l>k|MF`HW1}gVj$L<sL`t0oXukHm=EznfrF`hEv)Lrb@bnqoY
z`)$<1%@ar)A=fOtMQsF7$7;q7Q}#okGU|}3%4t+th?@J>yIAQq5ko&|I)2Q16Bvkq
zrQj%Y1eAbwX@SX^33KoFsV>9u9@Rj722S>AcUU#ll;__3j^o#WAGM>bu|i!G+c26_
zBfzZZ&77ZJ{!!Gr(^Y(Kw4j%P!+F)qnQr_*+Sx|^RV?kl4Fb&|2R1V_&+*hu-aplp
zWEEyw>gm$qc}K}qUk=cIqA_4u#;NS{D_6&q;K17^yFZt$MgHbQaMQE~z&Q@Ybn>Cw
z`!TgI1r<x;qqc=O@#-f`RJ0y80w?8jNz;DaY*3G+;AkeU@UR=9l}g&H_QFThzL<Qv
zn_a#Up-g8ZfjP5tFOS5=m%K{~Nz#fQuw0s3oNDav4_DPGx@NMAt~^FdhKuD<_&t-n
zJ5QxpSy@dTR~8o+t-{1U2>IKM1>T5`j+R3$|CuAZD4AFG)ZsYq4;ftL<0G#h_5<l1
z;vIjL*_pOtj>EtN3OV~Otvyt7^w?OMDM#vjgocyTn9)-vJ9F?zYoxPu9yi}Dfi9v{
zkORbgd;k3N&kgw0+BrhH$n3EeU{$I_YwoZzp^0mTv&UapF5i-K>M|a<Zm9dij&&s?
z544Ly=^zKq64%k;*<E~2yL5;#fTAS0Azq0`Q9@Eu68I13^j`Z&9e2rSzA2yo+}qTi
z0Blv4v7+H^VdqBLrHJ^_z9tbi)u1FfsQ!uYcV*D}ma&J3{rbbk92sL*Z#3XV1;1>g
zl{#@J>BHj~6E?=v5$`iVJ$<?XUktAFuU@^n580j3)z#IG%$qiK?AWod;C_WRuM5a9
zVj%MacRO~-BKGo7eIbkQbwDVl*aL@2C7`;?UAWD><;xke2H&taSg%!?dou@Q9kN>@
z{shm9^+9SKieL*4ZgK{#a>Z-(X><2mnLbsI+7Sd4M5P_?yR8syH?5%N(=J=NE@f?n
zQ%q1x+zN-6S%A3o`pu@+yI~Kjk3Br`EFzbp%r>A>gMmHjF-jgzF>Q3C=SPp|?e6Jg
zbZ*Rh^X2nL4T@hnPOHvU^V+ImC+JtpJGB3OX<;^;P=eh)eUOGBzDdS=u-t)wo9d`1
zBdT#`1?W!A&NiuwCQ8eBProy!bKuu{f%flHL!veFr6>jwMCEV6l@frcovV!&vwnPm
z)ilns#NDyuTu8BkKdQpiu<>^YmeF#{ASh(9IKM|t5bRaFbV<N_JAEgM6i`6i%Hn{C
zs^-St`@il}1{_u&BWwaHARLcUB&-5Ydm_YgF=<Yd8L~_jrk=|X=jqsQ;O_2D<;v(D
z+JoG~wI*yTD~nSt)gEWnUtU=F{E;UBxx=8kE}2{T%$rA1^@hW7ZaT7d&08c0NKJ*f
zS&Nfg!onhb<XK|sS|}(lk$BpHF<Tg*!E`(3P|ZYIoVNt>?C`1)@NuYZf4oZ2Y#=fw
zI3h7IQO*JtopfeJUh2Yy#}91kn&tG>7l)nFY(7u7=%z336>xp}qsw^N?0!_a5!HYV
z`~cAEV>#}l9iD(!2GyN;_SYu+yj_qOsrb?VFi%%=UzkE!v)mIkwMFa>q_}p#NuXdU
zugL-NhHI(ZW+=G}-~O?wxk0j)VaCP?q#~(>d7~(<JU?a@L{y`sPYzgFc6lyUvn5RO
z*n=$=exF|o&3@(EHyhYN18c7y$!*b*vZ@UHiUn{t8;@Fe*>vo-2w3`@GCce|Cepnp
z+yC?3c2xES3ca8cHURa+NZi=3=8HS9>&GLMX!&Al`t#2}*Ba~S=zv3(v00=94-zu>
zx#B%AjidbP*<j<Ma{9W0rdLUpP3zaTkCluIR4jCl2baBC2g;@oH8}$ZS`9}+TUn4+
z!CiKp>f#wbzv(7TYhoQJ(%TJJ%n!wCe!Pn3tBX{XqFdN&mmY70gY5F+=T)ikmQhKk
zA8YWRGc?S2FzRpqqq(?(P!dWLQELQ%;9g*@!3d4w-jyRq&AD*yHLV+!rw&pTYwYvS
z;~aFy$Q?<@;r&zL@K7LL=>$Nm4XXaW_smP=W<W<3A~yHkd)T0*<T2WjM-^8}PSARh
zwD?XSG@`1yD89c>PpocG8s`I3YIXMGt4y;ofn%)(SBRA)5Mqt;B*&^MxIQ84Gcr^4
z^!pa=Qd<zn{dgc^6QzNHz@tv?kG*iPny|A{tRI4*L5ULH=~UtvVhnz(58+(gO{Qa=
z0Km;V1eKt2hs)sCsm_?0BZihrBJ4b^jTN6?nF?PyPzf&o$bkC%(oCl$cH4u(zF;vc
zJVi<OwajgojL_F@g)=?zT&A&dlXEG%_SKz1h$uCbtFKx5dl_clSdLgmof0?g3qtmJ
zqN*y?iyN{D(a~I3D_3Q}PAS00?c28pqNY$oDO#8MM*#u`F_a_4AG0+3v5!M|EIfbQ
zAG(1(gh#UHo|h=X|F(nI5BKe@E8&i7_}pfhn0rMWbyY&Z^1QTs^3og7D-1A4dkAWA
z)Qdd}%ll_kQe#FXJrZniy*ZUr-~)2#Z#B0gK3Rb(aQXBYJXHm1qZ;T3H5}une1vI`
zGoucvBY_ch{;nmxkvj*3mIf|ZyL1>F>IPimuR+5&&bB#-yoa)Ty+39WR;-<%NJ>Nl
z-2f?C&^HcQdIMtb&u~3(5@hrQs-@KW!5S@Bk><b^*}-nSb)wxmd~2930PxE45|O!d
z>S(t>ve}RF9EIF|BCaSW(uc2A{`AOdoLLXVq=AHPO4Yvht~pgZXy{O!f)sdxNNE*>
zXkxf)Jt?`hRV<%d{L)q-(Ntdb8CW*e(dh#!2-LwYAye2c!QZV_xLLUn?IErZ4{}px
z+o2nr09c;SZ|oK$_V2niBM6X99d(LU^gd}S-H^>{u;At+ip2Gh;?Umz@@qEuy6#f%
zZ-fO)1NIdTUNvoIcr8xU^y7qxzx*Ow%RTTfF94WdNdtoyf2`R=l!EBwdy~#P07i0O
zX2%5qEOZppP#iWv?n%jcu;>)vK=AZ*n%3vs2WJHW7AH=juIZ38qp%G<>_}f86Hc{6
zr&?T3OcbPW1rI=!HN7-9uPiAlPe2(vP+wK_&I7;6XCf0#8iMLLfCdb^o>DKY>UWyf
zgU8qekawoXdsR>4dEJ}%N^l*{-9a)YOKb^Xw>418)`SAr!P>dkjsD!YNlhiE7dL;F
zLV)GSuE0`};G(5GULyd7K{{LS7^!3db58`pAP&|>yF6JVA7v^*6zjo6l`>;%5uLaV
z25ARHU#g%sB$_Asv;-Zy%N<lUTLvth3mV?a5mY#I!#>#e%L^bI>rZbUr`C>MHFXNY
z;d)@`7O)G^M*n4T0T<GGFt#v9qxKdKlxX3ub=LEfgLG~=?H&C6{TD}_1H@jce7&y|
z9K>*TUPoS&?<jB<1g|YQ6K7L*wMy52GI9>=TL@nXK<hh*d7x76w|^YQrbHy}ub1!L
zITL3{<T=&#F>M(;XL|zzNT*P9qx&T|0M;d`Yx2MrjtPW)F8dJ>7+PhzNC;*wRa~@h
znN?R=LjY(Q^B|R8W~>o50k|M+N@N_5j}BfwJ@Q(Ho}yJ)#aQvjJ}`u0Ghc{Sw9hKq
zHdffruxiw_0k!A?=`q?XzE&C}U?H?RYtUd-2@kQNbQ+4}x31w+xNqEVfoj1Rb#lhc
zKT}>L**@C82a-3Xkh($2N{y3~GllQkfh~PRSrY>Z5+t~68K9!?AM14j2|n1>3=R+1
zP)QHQ&@)J)kZ15gokR$jr@cnm7UoX&Mk!ErBqYkFo99r1H8VPaV`%w9sBBVM{Pe`N
z(Hb&D4i0gyHA$RLF0GTCve4c8*>D6}+C*pXa=xKQT|79xzipD+N%c4htBW5kS9ZAR
zFc+&VR^Hiy50T9Uq!G_C$t9F$Nr-t9_?O#fX}n@Z1rIW@5=&-!!r(>8c=;ycA#<Gj
z4Ee%UzML|1eYRD~xu?|C^r6sRAS9Q*AdzZ7n-D%F_mOr5a5ViWm3Pi66MUMRtynQ1
zC`8?M)l2f9@aX54wYZX-vLsoydkx1MRA1Y9EW&*mo7BMK`ZAJvaF?pQygYDUJWAn;
zrTnM6gQY#jx@b$0#w`JYlVn25#d<*3#)W5oJs2asm0~DZx$Q5%fL=XhcIGiFAz+HL
zQ~>BfTx>6Nv8FtO0uPLK6%R?Tu|tHyM;Df%6@=t@VDBuxT^A6D`&McIrJ?|0(;Bdg
zfyhi@xmgidPS@MNHcC{FDELuReQc`(yqlm9;kwwG@*O)*A@`2p1Fk>_!ube?zVk2n
zwqqnKmZYGCDX}|>A8YW``B5~5p~y<*n`+jUbevJz8G_oS5^Tcy*)BKTkqW@BKmRJC
zhLV@m*&O6-bzlWy<T-jbP`x>Y-ExGPIVcRlOIvbLBBw6#q8g2^^PvEDeBtmblQYLY
z9-5z@w;R@tl@lLb>@x<y?mF{ijSB!?HMZ;~(w}9Wx)fYpU9Beis-zI5h{OWrm|^&F
z7^N+8sUVKMN3L_P3IqJ#srLJC(@dsF5&@GGiofb8H;(d>lz}bgZYWi%dRA~E!`X%R
zSMu7%u@k9DG99f8%9+r6_wI>Tr{D29Y9#L2R~h7jkY59;Vm$?dq54=2L`}L)B!5n#
z$VvG|=_y~YxrM_{RX%mzd@Y~CE)+VE(#wQ<P>5Odor}x}H2H+&R`BH<H*NR>Tg9#%
z*s??EuPn=rWC1{gv4Y9TL7i3?P)WpQ@zHndLcdx`tbE+z06|8S(XVdA)At<BYlCUz
zxM>=dv7HP$o!|d$!(kYyO!c5<_5n=jFw0;guV8kiw2_gKIRGnCqz-Oesk2<>AcB?E
zmet|zP(ys|WG^M*%}H7|BKSATqF5<M0(5f^M3>dinbwu_nLnZ&A)#1W1@SppoB7jx
zH-aX|m0p2{jLJz*@ZN#KbH}bt_z=`le?Tl%BFR%0m2%;Akln1dbh#{@k`$}&$L|@q
zypM=RO>5Nnbk2>=?5E$eL=~W_szM@-kybhN;^sfEAwPg=Dq*>KifAJO;Q-a7H3gx_
z(r8TBn~cg^v!(*OT?)r?bpr=|*7*mdFVUG~XU{>Mr2(9fgedLu^v6qE_*L;j&$kzO
zc1Ww!i`&PjWH>+ih+^R?LEkeu)kBDkhvd6+qrdujZkB!?>sIogue0R#S;Wsdv3E_S
zgqd%iFl5^D&QG43s+X_9mJm}~L%(3jPXZghJHc<}b*&Dwz{^)+F5D_<tYzVFg3peB
zzff0oNj3|h6{#jVo@u&q@c4?>W0<fDcsnH&yZ*>=xsmN;iFwzSZAoZcB7r{nk=*h3
z&8e^+5Jk^=@(Cm9=FQE2{Ee3bx=axKD@WY-&;{K$Z-y=9@MB8ro{&QE_w!{0{pWt&
zX_9-Y<s%3X<1+7a&tv5!tHU<&koF1eebnj?dhervP4Z4=aqb;<(edAR#_-CVba?d-
z<QE|v2@vnrI5ASL15dB32KaA1`2)S9Q<eF=`3Z~fQ2J`a%YOj0Dv6+4{?8?GKlQ)E
zc0|kiKVSbZzNG)YfR^X~hk@(A3-!OaP>1s#tY^7-mI_mFTud|T?JY!rgUMFHk-_PU
z{tOuCjuc6I|GX9ud}1YJax?P5vOHR=F2nW6{f=`lC29lu*<no-sOjXvlm3a~oY&M4
z^Mj%X@Xl7Iv32m?*WgVT{q^fGJCSz(@x}WySk?ddy8Pc4FjwRM*gD+bSB%&XOO1UC
z@g%i6L93?`-yLQaR^AqlmoMy4+I@V}OB5I4%k{Srya5s6oTf0}e3-4l;q>F5G&7j|
zEISTs`)Y2uK!0Dgj-eC3r;1&Jj~!TAg9P0VZP|jbwuOU{tAZTH1L4ocCVlnDd18s>
zE9UFDhp|&RNwD2|;x7*CoQejo-ZshmaB}!nmiZ?6t~2|#+8}S=A2ne6en+IquEQId
zCD%`+y4I4oo&>@=2Y0d`%z|}gAUjcb<mVhf>ICAk{?M91%CknIieD1Yn9ER%Z8gb8
zuQR?=LLv1>B`CLig<g;pgu7cBMggdsORg`U;X|cVRr?m=fmYFe08Q=B<?V8QaR1q<
z%7EFT59P7Xtdl6&IY=JFj@t2mkYSSQ6!C_2kc5bF`719@fkR2_h2I>1Gug$Ji6Rgv
z&2M2w{M8=lz9(vTXF=i0T$K1)LJ8s29M++fxr+bnVdc-Q271QE#v@)U;2o0i5M27h
z<UwY%2lHO-BE0_o&nmbWjb6Vuq;kT^ePRfgduJr-<^doB7x)02UCOcYO&}4avsF>f
z%`Nn<SPY$ex)YbsJ+!RK7=XB4oHFc;WnbNZg+AVy%3azIlVromyf@Iv9y~C(Ae6!S
zP~00d53$fV`&Bo&$sLAj@8)x-ilv!Cc<ZQ|k8MDrmIQ~u<y`jJh!QweW1yfIBs9UR
zk*@%DKtAjcDp_*#k?j%J89l*$p1<<V<KL~JEt$rR<Eae{3`ErlRKTW{eDmf_vsXeg
zB=Ud7Fuz-!LWM$=t!bPftXZYNA<>uF$(XDzzF{r^eG%6UU%qDf<RQwIwE2f00?&W1
z&B?_Xx_5i<hQPpI4sx#9ZoEFlH<&Bu$K&5b-_F^5b;H5Rzdx2py1@V4_6w}6hJT4#
zt97i_H(<c!$$g~~nK}u!k?~pODP^C=y~BIQ<BOEo#^c>=O>c2O&p!pGzo{Z%rCH~9
z*{}2_x4}2})dhg(3zxrqjO?IlX%13UAHW7<y^g|<<fn!3$^h-Nm+UO$Fd;*zctID(
zVBNJwetEzbqRU>X{N?y6@;h#jK5?LY0rZuT!;~+!pOo+1C#w*R*DtayRRuXZbyB=5
zP`WGSyKciX>EzBkew=w}BcNr%0O)daTQ;v@o|(h>#}2aobcJLxJfrB=S5PECy^16O
zxw0}cLF$>oaY}xIE#_(!kW>3X=XwMUu#L2nXi6tHVFoq4I+0!HvOot);hjhBNxjag
zkC71}I7RCKLQD$&u^Z`rpkS$B(tipdc5@ohqLU-<c)krW4%nu}|Mv^~{H{qD)mwZ?
zH>j#8k8ckSf&{L0lWzgh-CE$m!s`R~+LYrs4ror}ZdBWas62xAN+{)%MF(b1Eu}T9
z^-{H>Pk~uUn1PkG8XC3MkHE+0o71`f=Ahkr;q|M%I5raqbW!dbIe!t~fHkX?3&~*F
z9zkjsN_MvbvfLG7Q-yGTymL2;XuZuIKFf>q6R0GeTo|1<)P&ec$XSH1Ae?^d{@Ugm
zmBD<ZMv$3~$Y!M~{c9-c=;Zc=2P|4sjmNB0p#odCnR|o*+-lS?`1=!l#)9C*b!t*3
z16hbf@pe+W+RwF_C!^%aK`7o2r+HsyLC+5T=Sbypf2`%-XdCCUpXyR9^hSuT4C0XW
zH2ArWnx8uY9?>gglCk~u)OJwPSUk?X-qV5m*6o#r-oq2U<47<SIFMQRpcdPO0#Jv9
z-`YYJA7XTTXU~C5Q5z~9&F`d=cPodF+j<e5P->Ua6p7~~hcIMnHQWZ3JY>hI%1BSx
zNWmY+yijdu<?SUwD-d}u!1nv0UMGR#qX=D1FB9G$8VF+skLVam6l;<dC?Nl~N?-$-
z)=(<mMm<kF0r8a^_MAQSJYjFi_3+zgQ5<Zi5Rxf<6e<X8q<bjPKyW)j&M@XW;LdPN
zs^N#=Bpr&VNFrr8p8bmA)ddf!>c>lb9Ef-JAQmN>V4Nhvw-maodT<dUvnPa5R0&yu
z)(pjM<jA>=8#IJ31EhO6kt=;rs@+frZ<BLqKRnQd(>3ke3F`J43MO?1%GmpR_1PqD
z!23>;AEg^ToE6F|*1%g~WOBg9H|cmEy(jO@pBb#fx-cHtKgJme0u7){);-7x!CKKf
z$gkgrzZz!hT(<kFxT#S}p-m<?AG^z7^iij1FZWh5N5QKjSkUFFO%ffsE84+x3ZZ`i
zu4R7G&B=}~0C~q4((sjGW`;R0z4PC`?CsMzr0do{gxOSco~R*H&T%}61Ho;eh@$+`
z55`2adceVG*Vuw2+2sP?*z<=gu<7W2obR2c_50R(r+Kj~wNO;T5ny^eZZ+4AEm@_L
zwUyYAY<n3n$B28Ff%$1sSw_;`W7mMS8GQc`VAOVzPZ_bsZuoL7NUrrL$ec~4j@&br
za)ASA8|aNFX$$}CCs^Ys1sqwnuwEpwT#HL;LRUzOTTcc%G6ZLKH7fLnR`ykeu!WWE
zZb&{<hQJ#{yFJsI5``7iT;3{4JJPz!i;$31lNIPj-gS+4mw1%BoNS1TaXq-L;vorD
z^vHMzVb&Ab+T*`4n%hUl9Gf__Lx^w$sgVlC^qec$!eqlimV$-CG~L7mBH@zIkYUGf
zaB(#k2OJa@R^Y<z8gudvf>=s!OUENEY8)Lx#*1?`D-QSwK8XV-MyC}Cb2JX}z<PR9
zU#xSvrac2M9^7OL-AEV2J9!V+^O&9a`PLMuM5!F?`rM%|P@t4jbcp&J?e75F&mpYA
zI4HI}L@PJxw*(1kr~dPZ7f4Q_4HZ9ZMVxEDI{o3LSY|*?05E^*jdYVwD*~}?7eo%-
zh!Uv)*}0XN(O<0dz+_DT$?$mdk#%8X0W3+mi~#V$n*ER%uBPjfChNZzNA2pXMOHmF
z3FUyiH?6G~R%V~3eAR(n{IhN`9|PGq6l}4lIY_+fsL2z4c>%&bb*k`|?T0DFq1=Vk
zwJ2-&=VFW`lunB)fS4j#3ickpD>E)f(5v9ma!7tF;9%?wpK7PY>Bp?`G8qEz%NuY{
zhmGa|1cL`)meA}NgY$yBO6lXhhp=|+DC#5`3N)jm-g8GpD=}i6w@*u4gIp30DXIqi
zUWs~dTF%}ZP>HC?gtmgw3RZ%NRO#k@7|>6u;iBT8ZY8!+CFEXeeOWzPfIUx?Cc>x0
zXh{qFb}PXYHi#9;{jzdzOJg#Qr>f4L_bJz<ZM005$lXf3X+J%g`<AeW(C(d`fCDg<
zt75wa1a{v}(|m_x4sKCm*JcwN$i40h6JV8GO(T(-*sb9Vex+*cHJOb{AhqT`X9w`Z
z3Hp({cXtk$ERrdxN9ngE>`4K#TC^04y$4CoLAGw>r&K=%49F@(2@@nohL_Bg>b$Vn
z5CdpgwMM?{u+Wqr*rN&D%pN+$M5A={6ZZ*VP5oijY?7t-)fM65l|~NXTDN!l@GZYY
zm^H{Vz&0w6mJTAjlWiLlCLnF3#MeR8<RudlxgK($>Eza^s}Xy4L)I>0yxeeH9uHAz
zN8X#~WPDpjvgGb?hP$%~b}aGbq14eZlQVDgYfL^_E>B)dG5G{eHO^=;=`h=JV&6U1
za5xMpXcFJ_+Y$_48$kn}E=4T4SiMsl4yVP&SZEt*YFYE6x0!uXBq=394I?vddVC#J
ziayeOLC~sj?=-$VQ}_dukl$XnGws{W)&z7=?hZPQBVV`xeK_kq$A#@je^+G3>XO5S
zW%PM{;OoMW<Elv2M7nm`X<4pplx4<?AQ6mq2Q>&sZVPm+WOCAf?)k9;lshzQaY$p4
zo$hdvRKlIo2gM@{3~(i^!Q`@<fHA2CRH^_H<1^0$q{S)(O0-ufk+e)r2i9QQC>{{S
zour$nqM7{!agTHTwu1+hmI)1zleQFU470}#)b>5&93hqEx&A`L-(<)+MSiLXt?+EJ
zN!M1SG$d0JqWYyP#0zwFDCH}H59ny}10t_P_flW#HH%<a<OUd~5Ewup{u(=fvkp-3
zit_>vte%4#^^ZMo$w$>e2sp%%5xBCfL){vLTZL*6LiZ8OB)_px4LOAS;6>H$SVA4q
zNreR4h#MI?0C}tcuGMSpfV^lo<f@M7bpSCZjH|c5eR8P=7C*gvZcC6*Nox}Sa{e(=
zKr+gj31}8Pc8dsANz|+1ce&3N|Av7mPwT3{E>pk)7%Vf=`+<N$sh^_XC8I~RI$Y3l
zzzmx)U;gyt7_aFj4L6=oWd4*w)Wwq*q^luIZ6%3`;8vcFs)E*3K(Nsg>(0~>SJ;cd
zWWc6l#FzJGqQ8czg&SA+`<X1kHra<~84l!>He!Gl3HJF!c?c%qSng8e;n#ZhkKtG+
zt0eWEU2zQ#q=%af`H-6YLaUi&_D{S=?(~MR{v-fkprxjED;&R$Heo<;SB-GBL!Qg2
z9K~nqcafIY>kl1N^MgF}>)hc}%_*8=g@_#7?F9$|MSe<1LNk1@B>E(@iD5g@MoZ_D
zL3LP(o|UN973wp3y#k*^-DoGUMXsCAm-p$BlQ1Hf)RqIl9r&OG15xhqGjfMT4i(`c
zuI7)#7j?6&r?~lAq@wK>wm=ZF?ykY>_2ET`#jHcL57nR$aNtTMWrs-;@t=zA-BJZJ
zOq)z?MPXxdqw0z>Vfe0PthYAlERXuBYh^u3?jmbAwQoFd25OF+6ogIYx15j<RG-gV
zDB|ivtsE1k(!$C5K*WIZC`4jyU+5>A%p@<Z^J~RJ;u%Z=RyF&PL2S?ry%3_Wigp~$
ziZe!oNV!{>psI>qx$nGHOgffIbog}#nn<onHwrTpgve1iR3ce1dkvPl^w1+OdW-Qy
zw=oolu3P!=-#v2o>h<0BMz!;3N;AASTu<rCmF#$2oyz<E_}jC0d1WY9^i@=Wx1Vn}
z(tO`r5m4O-KIQ%>c}q-I7}N`+C9aY8<^uI0_O_9mLtF7ALO~E}1(DAIz|;COg22T3
zz2IfiqC3&E8Ckt%lXL&u_J4T+<aXNhqU%rOvme5B^9^Bx(~sW|SD=;W5>j{63dA16
z<0DBxsdIXV#XaE!sM>Q>Z3{E|40n9piG*3BiW(C3Dv@L{Z08ItPrG09YpQmkRPDOV
zU&QT;yo|d;lW^m^bKL^}Ow6P++NG2B&?(!Cg0mVr<E4~Q>i4PQ@W9#D>99cT$f@k+
z{@X^`?Py?#bzS+ZNZjAi9d~rh9pJushB1PO+^|#xV#)<zk7rLBeFtE;A5BEf31r$|
zM)fsZ^l>ze-23Dl8=IkCA<z(_3vK9`MH3ej1mBuEita(`LnG1GsgqimPR^_x=^AcA
zhT^+xlo=ejqLPjVkzdfilJtlVxP}`i`&qX?U=mRh?*)Ddnu9>3Kh_DUBg8QQ5*@kf
zzLuCUIvv^R<_oSx>crE_Q44TFY@)o#q(4AmQ4T?Us)o8tpYgu4z`4VNv%Zm11<*qX
z#e0D5zN8Qdt=ArK?o41q57`0uO$4mW1{8Ueb(;Y`S;3)5rSwxY+fd(T@A6DRJ7U+=
z=n4c#o;hVlZTJonr^_cFV&7IlJUYD7O&JZ?6F}?fUA>6ho}dog#vdu$)J9ROYV9(a
z6H(OhC?;aZ68+K-^ch4K1h7Bpx7L3KIdXWuKwYhph?sB37hj_!Bk|7)2`vW@jus^e
z#e?WVFErB}!inMi=+}H&;0L@#*PGjqZQ#(@dO@bG+;5=})-r=J1b%<AA?n@w<ziPp
zlAfP^iqVJG3oYr~ajNJp&15Y=9|OsMb7+{ehGI$<u-k?i`nAjG*Ll`rbk+VbKwAB5
z*$2G4JIUF!qm^O8mH@EwH)x-Oh2O1H-jx7;h(<S1-ilLn>qCoQ7<`n&odMMEO8y}~
zzOSr^9I+zRS0T6O(h23(?gHW#ag?+uEi~@CnvXm}{eY`Sse>ye#E)1zIMwpE7<6u1
zgtMba^(fNnBQ`hFKAc}hq|Tm<7yg8J#u}-Kj~JMCS1JMfp=8&0H#j$G?x^w;z#s}$
zy2h?iK&o9x`B#l3$ksph5lFO{ECTVb3L~U_W=4EF#SCx%9XBlgsz?VV(<yB6gc*u*
z)T%7bpl;WwidoW*!CZ(?2MrEu2ukU8m<jsPKM*8kU9DE+I=|B5Pv$AG5y=PUyYS`-
zh`iK3z)7S4HH;Aet9A}p=!~*P<MPUYoe$(mE6DouGrd&BqLZ1t=k&qJIFWidJ@3#i
zT8g-z1JpIFGzK<9(|e`;>fr3-kibCni#{;!TZdylSWdlj=uX){LN-=d1JOAd<rdQf
zMUFJru1H-KUWy*oMNyKLDpUl#!+DWC^<pzM`VxpGUq9Y^&DG3*xE_7K`ZL59Dnqp<
zi14p+eXZiaWip=-&X{K$gIR6Aph^4mce=6Qa4F&@>#;SM?O{s>2xS&)V>2jDH=<EW
zM>8$&&|x;6aFa8qPe0yzl+Ed%5Z@%<ffo&4ZUkL{xQ-kTL&nd>p570y#boX}>^oHH
z`Bf7httOv9D^t9jpK(U&C3<oKtg8x$==FrE;8m@HvydNqIS7%&w}$KNn3K*|CTbg)
zsHg4Nr;+&{;47?xO1$Fm9!~A9W_MKhBHcE!(S$U?Dn^!+;kziFIQK2YwU0GYB*XgT
zs140O5nR4(RPki3(oVaQLdggRZ;f))L8*Sf-9d&0rMR{G)+BhyLt^29fz+nKP@a6+
zH42aAP^U9UdsJYqg73?}kNJWUbJ8&zc)2#D*3=4asd<YyF~&tHWNF`8KWdqEYa2?8
z2TXFo<)!nPM)lq$Ml1`?3*T9;X_E$!z`C7?RM#G?kYG);cKHHCfyBwRRfs^_udWk!
zSVQhp(saE#dU5Q<{TpL(mVXNnBT6yKV|WBAGdX$Y$2tL=QJ0AE4!3{}AS5@HOW2&q
zqNsj3Rc9){d?oK_EY5l?WCtwS*BTSS3chNf#2Iv?XpLjm&2?|0X3VdHm(S-$-9hc^
zL}I1BoF$sBd4}4wgi;-YO+GP9+2|C+8L~;-c7M3F^@4U9w;7_AC#dkLz2k^#26+0u
z&NI0Pecx7iiSkqZn)GIB2!Ts?emtC7aGZf~d}j}#NA#+sWrFlr8q-*L)8WDO)#pPU
zE$%PF$QrzLY{(-T56~p1QrwXY1|5E|6}?Lt#Xt!zk3F4nXxWz9VRY|$31oa*@(jk$
zurki(ooW)w^pEYGzh5}juDpym*Szs45?dNi8O03vz)y8d2n)UcWejClOjCPb{`@6x
ze<Q#vj+k+H$?9jnog@;8sE;fC^sK-DghVZ{75`+>z6_U!aKCTy+4oFh+Qae=+dTXH
zN<%4Nz39RU+Ec`Kh1WL@pY3<DNAKbF@hPvDM~+qjs)X<r+#G&G1`?pBv@T058P$gd
zW0JLAG(I7{>)JD~DH3Ibt~@Pprfwhq?VE*?(49fR4Xc%)tT;OR`9jaW{A)^Y$t*02
zCv`&q$pTu`q78Y_xen*uv(^hCS@&8m=%U7k?dDhJz7mAvbbic$t)Ul{%8l_qQ_*f_
z8^S#_g8C05IV!=FTy^fnMGhaED=sn|u19>=mtg6RV>W+89Rh!UIS1i*o1>&@{ypP9
zCjr}nZg;V(g{k7*N>}!)k@ZNk6RuRrv9tM#6lh5S^6){zYr>2JvUw*XR!bd!blwwO
zUH;ryE?F?|_73qJBy#&zAv_C}#P$O=g_BVj21#CA^cD1#OrwG9yNK#kpWn)HP|1%w
zI}EZ^1K~O8my4^5!Y2pohE)(At}V>XD-gBUH34}=N2ws<KG3#rr-At!@wG$=mQQ3u
zS4t@<J(+Yf<TkCrt>H62^KcF~CJKUof4V&}nXzZbAa%Q^b%CdjOdmxqJ@mQ)RiT%^
zuAHBP_p07{;1RLGa>x2_R`LBEL7={685_F>VfxrWA#1$+Me`F9_jJ>EW?-FW3&k@v
zc%dkRD(~CI7=lPJYLr`}UG4hx^3ls6WKz4XjrBr-K0g|7a^|FT##<X(TMZ~@t_3eI
zyu1*Q#vOOV<WqoSJ*dqO&RY%H8Z6O%KxxXW1BB#yxPm2JyB-*`EdwI0&SbR&UKFiK
z_vv8@FF(CLA;rbQ+qo#z0Ae|W4wTKo*$cz@PT<-Pl{n)`Mdn|VFgNLX?+{6}6nR8(
zm0kJpu+gQO;p1pV29c=fQfCBWJ@!J?ygtFlYW{=hfjiFo=U=0I;DXOA1m0dXSm&%r
z;Xo-w#lt<*f~*&F8Dl9$MSUvaaux5&u#_n^*lBKpDDP&kB=!{xyXd8QJ5H5cHZ}7{
z%S9#7cvSk*RHJG*Dq0@VA;2{KG40`{`-lXhvyQOf5iHSDq32iY7vrn&D{dnPszB(F
z&e5M6U+2H8-sBug?Y6o;xEy`|zo+cl>eX}Ni-jqTe<;y*mWZ56+de+^UTd}#x%N=4
z-+UQ^7#E?h=PBesR(}R7TG0=RE9}mDv(j+`q{;M0@6(T0T41a4tq}bHUN;0=mQys%
z<jfE=cAdXqOo)kU@!X}vl|>P9SXH?YkC@wyTM%ayiMLNDo%wZjiE&Btox`_&>4KO+
zwqCN2mU|?1g9CsXvZ|@TxtE}XH42O>*B-|p#Jug4i^voe%qQG!y?tzIdwj@3m7=Al
z*14n4Bg*gh0zUeVXCY0<Uu|dY&$OFTnsT~B+-r^Kvnk>3bvGTvD7~DW9ZY1l(1Li6
z{$+pF-J5RPgdR?%ZAta%;hgq~`{1_<Vy8wjcU;WD)1K_9mPf^)l3hdre+ZmB=TUQA
z@wou9c;VW-I;g*gN=MNXhNWNEfd98DLzWW`)V~ZoFzA#GN{qay;=%JMU50{OFrWm)
z<07>_v|DqZM1^iQYYrm5*_|l<-PgcF93;VrJiS#Y<g9_Rn5`_s4e`XyXfB~-NiT$h
zk;bZ|M%fXio{X^wnjU%fz!d~_X+q2#=nk?bQ_o9EiI+b(3iTV=2p<SyL(3qFpQP}~
z_(jd|{|MeQ(Rl$6NNai@H-Zuat`BA1g~j=)eu%)e3O!eQxE|bWBd=ZWx-^G1sAEN^
z|2(Z1J#$GuxDQfT?W~6@T9?T#5ZWI-;AL=lB*G}dV~Z{}7oc{nUq+7yHkZ3B7?fuz
zfACx}t1)0RVq1=3qNx=K%YF)jQ3k}rK5bEfhujmSw=@H|Uy;eW-Q0}mdw2^nyH1Bc
zirxI}J?K7;KHHX!Vk32Df~~Esh@-9R(W<T5S2@KWi(h`m+CqGRT?rub{7mPxV*9J`
zitUJxJk$dL>Pi-MQbkppWdJ`FTg1hIpLbp81|T7yfc`TY&JYIub=b)R02?Vcx$wiC
zyCk)fyg`66Vd@-cwy#TiNcAr=@a&C#8H>TT^(Y@D=l!Q|ja?wv>VlUQ(Os|(-+R^N
z9=IHt!E&4+Zf~jFo4psYMWR2%Z#LlXq^Te%5yd7i$*1YW7e|Mms3VHIS1DY=qEvw}
zZ)iW)*&bMTQ(sIV#<EsP*s<2{MLH4x@<JEfNptP5s7Xa<QHNQ?2*73d$X#1p!7rxu
z6L?zcyKCsyN$2Lfvg~$~+swE1LIFPxGnRgdO4jnxD!(2|g+M0Gdg2P*RgC{Ag04X8
z5w=^Pre4sPSqF0PDjuS9PJ?i))1S%4ax~9?W^1`I4A>A3&)$zye<^r@NEz?mOXhdG
z1@Li=1-)sFKNr^sveN*U^llGRH?$5_VNKSPQ&xWg<G!@FCK9}u9Y&c#Y-CIEPsdu@
zZerUW8lPc~I{5tZXh>9B{u}CDE*b|VCDv3#n3ybw2{TDutmkNo2(@DmE`!_SR#+Yn
zSn;X#BR_5RJZJjiSUE2{v7<suYr{e7fF>xucq}@BHUU~NCUbcPOGV%1#jE&v`-1Hg
zokd_#A(}vlx#COU<!u(t#6jgU2OprPs>52pEZ|?tVdxOzM|(mBYr16Q#vWA65#0Mz
z4cHh5Rum62(Mji8#8Z8?9)I##H=*lIc@z)A0Q^NeScIwWS$Nq`E;=*TY?7-4C$ZtX
zRcjCWg#Hf-IlP4m9uyTrPAe!gU7l^d#wq%W$VUuYd)l3akr<@!*{q<0QUmD2t)QiB
z^E@!GSaPXcS;zozjr7Kn45B@Et3U@P%Aji$`M&zsOZ<R9h$h3!abv5#qlXNF^pJ4>
zO2?wa4bw?J&U7%pTg-1Kz%idOkn`SL0{dK$bW@Xl>SO1&D@Tfh$CydVbz+o&WiTGq
zaSDb^s67sdAUZKmLpRz1xPrGDK{C=GE1`K13FAN}Ccp8#fJk4Gn&EhbE=+WmOlgh$
zXQv|3D?rUiaEw=@4cqu#`~hmudy4r<vpL{rLc?z<9LX|MbSj;wwhGDv@uLQYMSpAs
zO^}=@xa&Q7iPnKO*WiJ!!oj3Nv2s-r)_NqgG1=pCD3H+T&DbGBy-=Wm733bkXopS2
zh=PmRHp)k&pYsCNiF>eiK3f(tllyJfbWm44A#lc|y__X{-_bLz*wdPZgc;$r7Slx1
zxJyS5BiuDKqLH2vSP0%dvfM8FHdCwpFtrWsR%Cb)JDRmqwi-n&nU%<k&un+Yankki
z|EfL2E@lg!7XnH%HpGp3_yA;3rVO=OL@`1-yhvU*!jG3kkJBAMX6nlubRr^O)8o_g
zMv&o@vv>!M(6rQ*C^J8kUe8Q~xP0!8am@eFES@?>Ne1topXpxwZiuz}(3mZHqw3y`
zImQYER9@Y?s>*IEgqspTv@+_l^sdAJka)Q&H&;+La?NKYg25QYa=0Kx4LgNH18ZER
zkS~%kc7s|`uFZ8(p=;_?lxl4C8E*2y@g+o!e$Yh>qV6xSvRQ$|kizVl4GL$DlXZSD
zerEt7BJ|fE+d0p51p9?X$jy&YCCD5)2ZPr2UBCXa6H$^}MbtuoXHkt<LTVE=xL1-+
zM-_}wL-`4klioYOzDnU-f6Rvlr@$o>LIeWpFae@w2yGBsBHw-1pw<QZ)E#Q-ed9^7
zJH(+6mXkS*#--qOr=G_t%8&?$&>aFrh#0oNBQl23bc~I14jTY6&B;O0Ki`W-10ddQ
zjyY(@H2`YMi3E9}lIZ`%j^^ZfEPx>kBr>G8oC>t&U#_1${g|frD5FNQJF#HN)De2}
zZv&S(J5xg*!KR!_VrM}?>3C;%w}PqTC568WA~gXkn%shG|B+mH)JjHwhzd7i2I7!Y
z)J=>tn4T@0NWd1(*JS<xDXn?B1Qkd!7K2*wzOBhZuPI^UcqI>=;wPl_+&HTz8GUH{
zupHTK&=0FWLl}lRTn5LCbEAg=IPiug<kDvsEm9;8ps%E#?Egh8_;o0=G=jpQM3nIZ
zCdo@A%GeAYQ9zkPuLD3gxFFQqMop#F2<o}=2xY^ly<yk!pE`KntwIm;*@l!aGg%St
zFw74uFG6Xn0`0SUWCb+<x97F)vz)VDXCYaN`28?+flVS1i?TV>3|f6CLcWgTU{sXH
z>_=qgX8z5IVB2!7c&eBMd8Y`>c~)*tAJEVh3d!8TB@Q>PhT$v*V~N;uXUBW?;*9d3
zFv|hCRfB;j>rtO;DV_ugqUmu)Vh!{NB79Sn{^|@seuOy~2p|tLJH$|^ZvdswtnM0M
zP9T%L6___u#g7q)T}2ccLbkr5SrU`q9c5=Hoe=SkW=$r8)1jb1v(;*s@wsC8&J7(L
zWcT@v=dLO`bQM5TSwSod0Vk%00;v^_1_lPM5CkD2Ontet?9AIg=g=TM)$M_qTAZRV
zhCP}A!uO-j)lVxgX*?x4Gwc{*S%pG2E4a}b!2H;fC2C8>c8z$j0+$QPXS{wThPaAk
zfXkvzX35Q#&AKuu8K8NGm3qSMsw6NFH}B1;9=a|gvp#aY1QSbBu6_%-f*7+2dR!QJ
zF_3q-xpo)^lI@M`O$z9PqIqZ}V{zFzeGPmi5M7UlP%o7W^dONun*2i440B!(=D-C4
zl}L~p4KD)0H`*qAXJ7nG8>j~P%&lgLx|HCAOxr!q%42r<?AvkTeM#h_{&*6{@1b3q
zoRefNqK-%E!#)MC22J#{6YNam5vHy?6uZpukXfSRbO-l9;F?@zbZsOZ*8pSo18IcP
zaDgTy8=A5eA#oo4zvm{<;UaCr)QMq9bt?d2u%IycRG4O=k{yZ0kCB<l0wZvU7U`!g
zh$56+E-ei{uzk?r(IF@5hb4DF6*RQVO&90>q$Vu@_2S!TAR)^Sdib)a6Wt1~OzIw@
z*hDKtHc16oM<d4)jv|iEVV<6)L(z%8P??4U#p^#1_W<9mGc(NsqSD|@TG5F=*(56k
zmp@%1j1Gv;rCu&}LNiW-;!Lx8F0QZ;<Ph1vU2u&x@P=|w5T+?R;9&cyilu3nOcn&)
z^?mRWhJtod#9_R`$WTC)YDbj6@+cjAFZO#1Ht5NR?0}PiBORf#C3h=%VaZE_V5bje
zEf{rgdTvX58ks$r1(i!25es|0&k`MI%6AuaAJN&uR82d$-0Rx~&-=H%HCnX*?~{Jc
z_DF<_n6lQvZ2Na$LzX5TA~29b8I!Ij8Xd=<kV6<OD$--{$tz&1Si?0`WKA9i^5y*I
zB5y8B$OM{b$R?-4ABiR-Ee2_>l2QTg(+ld65cM+wRtwE5ZcuV}pb`zA^H80wP7Xzm
z0F*tZ=gYIx2=ex~JdLUL8d<DpO#PfXsgnIQJ{x~~-cbw1I|sYoJL<$sjXxbh^BnXt
z^iy+Pdl!i&B39fALtA+3y^wcO_W}fsT)7UgmIS)6$Pw<Qox7H4*R;bNygIBKi;?0&
z=%#sD(QRl8^$(@Re?-f9GzrwVYAR?#A&8h}2I|SSZ3Kc-@{xVH6l&nBi7^^MuY1xb
zo}CwImdN|k*BA)EK?ca$!q|$nuwNe%OC*OBa|Wov42HX%+D{G5&w?O+$Yu^Z*~e>u
z0ihnNZID{cQrdnL;sArzDDOl{?iZR9mFXKy6RhTkJg}kD_^qen<o!yq$QK!^nApeW
zSAGNH^Qk!0Wn1ChEETHQ2D7YgmusE)#qw-i>$d*3-Jq?RW-hZ>Imgv;Zh%*IWd+>R
zbZ+}llS`aQAcM}}90W9O1-bR82(=X($+Ylpj)ap;Ym^3OZ{^^LMbezY;KkCh{7$DJ
z1n;aDUiwRAut6-b@y()MPmS||aw=LwA`XqgacojFKz67E@y~&sBtAWdP2cBHv0z^A
zLIVkMtDVI`Up3AZp*+s#M{lv#$KDA_)OP7N)Ic<VPl_5|87B<VQ6z`3Gy#`T%4a@E
zJ3o#VQWIhysBrDm1PXtfpun4JMFnf7?^H7E+OL@EZtFOyh+>IQcb9vGyG=el`t3V|
z>h1U35ssI9Yk=l59=Rx)FdbGP9b;UxkfeAN+;P-{Xs3HW2Pd{uHk?0?yA<(`k|6<n
zb(MUWr?%1P%=D$%b3PfYJkxP$-r(irQ{&}G-)r$jZ%0)UdhVHn|EO<rur_sNc#uK>
zMMx>nPj3jppMt3DQ;jJOlV#tbOEUKqJGG{ee=~WG2nj?~<@bz8^&IGM)VU}?RQPX8
zJHTaABB7BBwbvvQjk+*zOom1Rf4cw<5uLbQ>_mF*_tH?KmUG-p^h9CAD{#p)0DuHF
z!zmKkX=tn-=!;(#+>zg53N?5)e?v<%UW1(-<2{4O5T%0&k-1Z%*rpU>X=oUQ+m3!M
z{n@eZDMvv#KaA>bZ6CLpr(@6ud>919RUMm2W@9I^9AQ+_ovI?VT|S~TqaIu`fzz-p
zA2TrTed_||_8r8C0jl*W)P%(C+ZK?n)j@_G8jN_=m<S?>F|<(9U&@}tEly4x#KmMB
z5&B@7zeh3%Q@=wxN^EMU)&uh5J0HD)P4@NwN5C9FQWD4`yLdM^RSoDod3Dr*9Rc{R
zJLL~1mj8;$na4EIfr^mTqqCj8%W`nSFsBBHjXk|0+)tbM))L(T3CJ>cO4QWiTx&&L
zK=s?mp0tNnc1T=7GVhIJh{>4;R9%AVJGwK6iXMQIzhO=)7*HN6%J*B-#AhH_25~kh
zPp`h}J^D5AcLp0hW<Bpq8-e=MsVJbf;Rz_FaNPK^(`amPK5B$S&|pHtnn?xCwQrN4
zVCK{mM^g$2mC`_;Dr6sSvV>7FNBt)xIuQycl_*zoi_;jeQ8>}?<c*`TVq~i)FD#~h
z9NO=KuF*~8c*bXL#n?Ro!GJ}V=kZGNv<+znCW>1{8t+ENn*A=+y+TGq+MG1D=(@6(
zE2VcrJoF%`kR`RBINbd%0r>!4Rq(8Aq^We)K)(dv=OA?TL4phh8NewzG(`Oab13%s
zVB-lQUpfK<H)d%DV-YHmY?IB3T&c;M#sg8>p&6MZh84uH{SMc_3J*TZo9xddbXO5+
zP0yd8DJ=)C)QZZZwNM+x6Yj$I6&Q$$mu96hCwE{l3TWk0c5*IMh49n>jU?en0U`RR
zVV^SxS-=A3Aj8uT@CxaauEG}(OmmbB?9}Q(0xU=4Eile5ua4PPp^aeHDyR*g`wkWj
z=D1Ys31*K7`UXI8bI|3{OlcZ)j#4;)56(4e8qai%Gd)eT0<A7hD4<!q<nLp4`M^c8
z(=>l4%-DY?^_#rkFlq^q86?-SQ;w#t=k_^+MF-V*WxSj^f=3lh>kt-cI!x7Mxd~zc
zdCrJooj_$<MRgS*6m#wh*$D|!0Ugs6Y$`_ygH++zb5mU<#i#k~8RiO;eou1`iIVPv
zkR47$*vZBOeww6300DMA4jPU_Vh_26Xx;{?j=YW3qDLl&soVl=WwLPhDPXAN`OyS0
zrUaKOVB(xeS6vdN9YLYRDU1Fz&Fpc(nLLqueX9`J>zO0e=-|?{zJEFXwp*6_l6X!7
z!~X@{*e80|1=^mbsb47FYrs}=W4M7oO!i9TSj7F-DH!=P%@}682-qh&I_JI7mMGdn
zqfM_PL2%blA<DjDzP+8o9q6;t>1OQ-WaUcK3|7w~w$e$ujcyX!Z&Vx8v>+Os!T{4`
zTYA~}tJIx;4^Hmwqgz<!5_^}%JnXs^odI_!@;hA&ecDPf{ivCX`q&KZchGpL*V#LT
zhxM9hKq+%cbQfrVJ28r)C)Dh&%i8G^)T?cwDV;S`z7T1AnX5Ad99lI*-N{@UZA9MI
zVoqtT(LXjh1yT!3vXmBb7))63*OE?gab^KNj^Nzm7b0e{n`lQRl%Rp)oz=qkr)xp#
zCF!wk`$hb^_`|&~FMt7k%d}5!L^kXtUQug1+D%y6mmQUQ?bfoenBGQKoVVDp5q1RF
zM3OkAlVoY;;Cw#u1QH1bH^B@J`AjUPMCx=5eQl$TvN`=&?KZ0cXaRiGc@*p{BfDsV
z3>4GGmU299GQ?^BD6I8_U~Vo=Zo`~S!_}|v@8zXtp;4sADI@CUVmW`|HtYIZ+y|S!
z_UeIOXSu&m5hWyTXRnTN94TtH7Clk<Di@1PgnUE?9UQXqX^RTy^H}<QJ`xIqrn^8h
zim2a0^aOqKPd~R_sFTY(9<0TDo`LtTuSPyO2qmH)(X63G1zuq4qQQ0Mr%^UTGK}n?
zU344aQq-n$rVk7#1LcEA-jNLbt<>A0UrR09bj{ZsPyb)+y$Lked;2b|QM*YhN{O8&
zQv*fFRFOz#A!DYJp^zz4a}t%1gfx&b8A^#rnMF~VhcZ@32$|pO{<Zh>-{)QDS?7J%
zIcL4^S?l?&=h=Ike!uVc^Zg9>eO=dm-MLGH8EEJtR`t7&m*CqiRlv4TX%QVUJK6+$
zj$%DgagOEUF+r(^-1`w;764W`SVwiNOxLD8`INeE<4puZpd6WaC`3V<mkd<3@C58_
zlZH+uqf+G$`qfjW4;Zoc%NukL|9J5(K|*@U)ENr24UK<1qDdIBxPEmfYRwS!)FXZg
z(0bkV?(S}@m^|mbjA39h=}P<WyvaY_l&NGUL>~Wmof6J9<^RXmOm~$3d)M*L-}uj4
z{MTswpF0-(LGVFaT81c(ln<EIOia;jfH`=3mSR#q{Dh@)8H*;w8D`W$*J1q~^1I>C
z>y(>wO?HUDo1#($Di-|iVk+kjX8aUi?g(;-CYPoEdcCYN0XhDUm)d$zxc~T?Dd-Q{
z8vgOJI79o+f4yWw)o2y|sRs_d1s+=bJ=>Vl^#}kUn}1bzh&V{G98HCoOtKy1$BbBs
zmwwBNLXwb}hxl;?!Uc_@*(PI%nf-rXeY%w%l0)@6Mf}-w%Ab3g{V5A4JnQ#P1G`d6
z9^O|`4-Rle`&KhNK1>wUlW1S~$E#)q7VJO1{?A(b$6)-QyHNh~Q2&06j1%eaFX-R@
zj~?o`soKneKLrx3?Q0UkF4z@Va%szMxhI=9b5ZW4KljY!kuzT`v{-(Tdf&;`pZoZO
zkG%5<_L*_?rC9#7OAc#P)P?4XG2LT5&J>V)XR)05b-v@-GT&o#qTh;3Ri23KjH}Zw
zuC?xY(eyqdzT8Xix1-+Qm-s(!Z1tZ`XpFJ?9}m<yM!ek1gHmemsY!~rqx{EEkGcdh
z>)Or%VlQAzcX#L{v@@h%kl)A%Q%)_G0Mu><tNio6R4g+*A}AvxvjPsovEw=8?VHIO
zfafE)Y$hfxFY$Im{2N`KVAoIbZPU(UAeE3`T@(F|J->1<<IGS~K{U*YAa0}cB=Sa6
z<b+nO;C+cU+B76hZzbCvY}kZfmX*sxu8u;tgg3(+zy(8l8HbE)t{0k*?J+Vk3hmtr
z5~ZlP&qI&?&cv9UU&Q_->;;7V!3XZ<78c}+gcq*@GUZD+KyHjkGaTNe=)0$U6)Cb~
zX88Sm9i>WVs1$;~a?uZ&q7GCY&ZSNbfPnt`o-D(X`_zmAfOS+RXpx@%lwo+z6qJKE
z--AI6eJBGm4QCF2+8|x;Dc}9SKLdBW2No!$<~=~emtzb=6B!;$vY*N28hK@@%>)R}
zklH%>aM^rvO*u8A1Vd0_-Ut#;0ZdPcxh?&W$@f8G`I)N_J5f&j*ZXKKWZeE8s*C9F
z?^p6{OnolILcc9r0dgoond_7YO7S#|8OL6%0eP$(mqiAlf{ODPt?D<WrI#aN_8a{i
zln-9{IslwV_Ys#A3IkCPmK#5cFz%>CfpJISm0p*%z}&2nk;&1jrm&vrj=9p9;4=VB
z709PG41a%;2ji1o^7VAYRA#UKeMQB=&I5sICXBz?^84q-_<T@wFSM%S|GuUw#<RQg
z6VEaD;0!G&8A(PQoytysr<Os__R}e?gWKPM>|2V~A%FdZDcTzNZFH&%^b&+>|KI-L
z1WGePO8k2(kCy!P{;SXrgjx(Ne$*;RMg2^?g5>+ed5u&tU_N*&RzL~h5B%|lhCKb8
zDd_xl(OrK%w|#Mlv2gFH)G+Pkw1lYb?=HfGg9z~Gqx2Id-g%gFSHV7*b`GP<O8>yX
zz7(uQ{EhHme?t|R$cRllLmqQ5Y(-C{RwH4dMgf;2djGnpxr~cSCSe&$I`T`Ts9zZ%
z<7flMYGtyx`)q+a&2zv+4}%gItA&3n<B#6?y3g<ghXe*dyW}H%G?N-TS%tttUA}w_
zji6Ql;@o`Jmws}pgj~#-m>-kO6Tz5TTNe2Z?i8UP;?^NTm&efhhU+viJ1{;abRXjn
zvhIoQS>^OWogy{icuJjMVy5<?AeLn7SG|7T|LdEviN7EDN~~%;TQ)w%f;;7F0cas0
zBt+-SFVN?5DCXffz?fEAdl*yu)04kec5F>zBI60+1K8BaB><K~=dW;s_vAPLjXt)c
z1N0ZMi+~g!h<!$~9YE|Q>5PZVR3dli?HN?dasUPtXk|R2Qzd)p4%lANj<aDyjG8^Y
z3zL>ILC2WO5B#Yzm<(@Zw0>Wb=<g3$vE7c7s>w@k{$ADZCixi7VW%`gLPD0!m_2{*
z%Ff(_^j@;|G+*z$+Uw$&1;~gbViQ`wUDhmCH|qT$wE!U88r9#AC)gjg$xQry=<p(s
zy{^(tm4~r`#{NPo$wnsq5;9oM6S|DIP^)cQy8a7kOh9m?C0e6{;bX$M0THtdjYvX1
zLUKQ{wh`CD+f2n^jS>Uhzv;fNG)jGl(<;e?F^dm?k9^$l+;H-Q2i$f@Mh7?Cj0);Z
z0xKB5<hz<204rxnfq^84-U;>JtNOesFe$Rja|=-yBm@gcUyut2-8~a?mZpmMI-(c*
zH^Dl}v}jW=WA!n9NdFjOrkVO*um3bY{XG`{8HfM5aR`0~fgNC*wX0A(w6S+Fix~Wg
zlt&@}q$a!>$_D<Z`MnYPy{HVYD*B{r8Bha+^n3h39w|V-zE1q=#yBts(GBhZ?G+s-
z5^2x7^OFo~7rK(%#qDhJ+SWT~7xMR(GIpO&Q%@n|7Z#t#1?0}nm~lKXG0*v*p5N!f
z`#0gEd9z7;y9xb%Xer@!N_ZR9_aWx!6{fH6Xi(2!t+_b%FGJIGFRPTk%XDWxymz;N
z=G1GlV0=UEv_*&uZYEfsJgUPT%P&zgE^2dD^TOT5rYTFD=xr!qJ;~v~%8p1GyaPD|
zU_152trx&mw3QmDkamXTD#Gq9=6EwxNPQW_S?DrLY;y|6UKjoZ3=47)lSQSa0)5hS
zsFIZ_{4zO+ra|PHk2f@f_=DlX4?4Gn9B%MCa*&wq%W9@g5<TW@1<{g)_AP`z3Z%+F
z?!S^W*60-zpXa`oEJp~nrqGpSob+oRz^wbgP2`b&2#g(H+KbYX?T@TbN`|BV3e}bi
z!iS&A>`60qsfwzZ$a#RFZJ@&$gfqDixB{--XW>aB$h)7q!H|h1>6#gz0=$nfZ0gSl
z#S9mpBUBVCh>AhgZ2?#bQ-B>*|LFQHE{5FPsbE$(VeD_a77X~$hmP-2g<A;rij#_Y
z`?!dpAPXIcRL6*_CyxLkF7VdA<Sjwflhg){bW7A<XmPL3?amZZ5#ih}Ae^4C7CZtf
zurUM5-iKph4~X4+7%q!+Z@pjNZV`9p3YniNVEh#5G{~l%O#K=3Yx$;08Ij_$GHNSG
z@HnHNBZB_|dOs+f`(VePhY)tn_U%uXK4KhzbFcB91Cu1d%6j3@a{RAdnEXqIdV_Je
z+$e4$WpG9>$^t`teB?;7K+H)=4LHhm<WS%_M0OWTh~}xYHh?~yq?qU9X#w(89|bAT
z!?PGi^L~cpvI(_ddw+g!%+zf@gwbWV{F*{OV>LWdf)Qsp#@>S}UcTuOmhmO3%}Bus
z9ORB-j}1boX_bTGdkJd*4PnVIuvO8Gq&uTFc6q@6D~($5i%UpZE+^LlT5&%|J@$*m
zQ;9h{Fn+<p4Bs0$IpVo40jyr{{JkM`=oVGj?DAA?<^&B+MU-nrZR>)ZCor_@vWH*v
z2LIZzb^hxu^I!9J3E|Xy>|M_p3pPvRdvhdXyh5eY?HX_O_Xa~@8dRK(+sAn6bXtJr
zQzirJePb>kQLB0lD`~8X3uK`Z^)<8o4$yB2Dv<fK8d1EqFQZE5a1D^|eN<80(*&n#
z6`CaIGi=z_n+zcB5-Dge;Z36&I|V>D?5B63_>HGmhpRKCnUhq5ARl;onn-4N56a<7
zK2tA3u9aYkDUGLXIzq}tt0p3w9<@@VpFBL^N~8B+!$PTtO@Oy0#Yfb3*s9tX>V@j5
zmAu6+mNX|Ws8_X8*R8Rq2x6GmAH#;qdzW$$q8{ZjyEO{5QGads{Obequ+N@MMD|KD
zC~C-7`JJjvXgPqdOV1NgkTU?`P7Y*wP$-6r4nb?-5AIWxLDdhYp^XKq9j;RMJ7T1H
zq&(9QmQsZ{Ns!)4=MRQa9!mN)><14#19h<7VXm*I&KyS|SiU>F_W^P~hJXoOhRMLn
zsueZ193WTuD+hay5n~IDOdd7g!RfMZr3_$JV!nY_t~h^C26GBYAUkaapj@K-p^<Q>
zjf`5??0U!up`yRiRWqbeFyDEcLbbreuaPSgD=Y5PA7Yt8SU(4dKE+JxlbITjp>nas
z)tsy(o}#MdyG@hS`GmM+;=ANej=dvsE5&0|!&8&S5dG!6bHl6z<$XAhSW9ew8&4Ry
zMPh^7A<9sI{e0KqIau+x7GU@L0+Pv4)jV~O+-K37s)po{EYw;EXvQuO<4@LVSulLJ
z>b>mo9*$<F1RI=OSxdHARQGHEr?c}$jOHP}yAW*4{y{u48Tb^Hm2EsxPFB(~6JJ%S
z{Q}8-NYOYiq8=-p*zE9hbrkb#&K5%zux^wSH6=lH`c+b>`<FrFKr$zibwgOi=NtiD
z=@r^S0gqQw1LCbls0xBJ+3tWL%9u&&c>ZXIVmlOZzKrUr80_nPVD<}9J3X<u$B>=$
zn~>&FpKEr$UTVKU3+22o`O~+8?cG>8N|_d!9I7qwp|e2kAo1g59FJypS?|P&NoLfA
zA*CSaL@w~a{3q-%G0doy5SEXr)|fYOy5yrb_v7)O`{`+qBNyY>>c%j)qPX)4W)53S
zp3kr5t{0$yTOWkk@oRaw-7dUBjxb=ocp@I5vZWC7@v;Zka~Wh)`N#rRQZL7Vewxpq
z;*O#GeVuJLZl4S-<+C@CCsID(p8=kajGW^9U0@A&W};Nx@XOO59`8xz2MO5mfLqDg
z4n$kyN{rgFj`4qAvW3<W=_0W)KJd%Lfoev%XkBY2PDpn8u)38E6fwZbFGbg#G)U0z
z>LoV#_)xcF+>Ul)1NJEHYun#Zj~Jfq4?Nqceec*+CA6F{j@nS>MTM~~hh@g}QW)9g
z9WgWtSwjB)0RyB5p&=$&PEiK<zpH409be?ud;-3EY+p|sHA4pyOB#@pn*bA}CSh^H
z{&nMlIGTO-)cDAc@9jVhE}lIj;9D4(R+;u19)P%?iv~Z%0So8WkgYYTK1lGBKUw~}
zp)`ejvi!*TPAMHiJcDr3cN;a)bV#6eNNr4(?Z}ag%!3Jh4_OrimJdU_PDfdgwIuR%
zhT}ZoA!?tE#lMu?bzsmny_Gsbl1RMrslmjLi!y$wux6X&QBue3Z-s(PAmjy$7N7*h
zk=beJiA83OLuT?c-?Z4FmT(Ptdj_SRqYeq%vAaOEt<gPB(Go&1rIdU3Lg5~Tj<9TE
z)yOu{rIA@fRD-;=Xywb{HmL76K{bqCGq5RDbfoZy-0J3r3-k`ukP5%(*hqYLrAT@s
zL|XxokvWm^G!~~`RUpJvn@fUM%8}9Tg8GV+W{32RyV}3wY_&sLNpHr3<^y=<%Mj`)
zg}%qK(r~NnB*h4|Y$pve@TwNzi&x30DFvZ(kQ=pe!GYO5;CA4^0n$YAIYRj}PSR$y
z6WjnvIdw<yp`DUUc7l%ak2VcwFN47h@A2=9`J`2f2O|$nc}{i}6lwX4MoH<m`XyN>
zfC{iv2F-yBy7dc1DO^-8gW*Muh&-9QK&@|h@5M0u;#g_7M${mW=W!pjx6q+i=csB<
zvjg7{_}YL)^*%e}D3AoY3^Dk^)6J4fnbal}bLKw>m5)$yL8SMj8>5E;07{(Nml3cO
zlJi^ymFQ4!8#*SahWQ!9xo8Up^#w8Wm8oXfGpNs+SC=1E-uoj=j()=^<69}3_%ef<
zz+0GAP>B>FnAmp&smK9V3s)$hrl66bFupC+(a}*r-8nxQU<aWcoiZKa4m=4zV|Y!H
z#EC8p-&Mf4k$uU!I(Kh3I>B)6t3!86^gE>hJ-tPc)!h8K^9aJCD+p1d-ZVe4)*)#c
z^)LYzM>;DpMCfLpU!Bi?L6)yfdpE1kfvWQa4-X^o*-~PGBBYEh-l<D#;qG|LxtGM(
zhMnj<paS`abz}n>hN(GGlgLOsoQ6D<`UeU3qP}+T!C~y5mETG5yaDQlK(9^!8v8@1
zGMgvtBC|+jWl@fOFl=atHwNx{2Xz=4%Oh(SiuHrW$`R#|Q_%Eb(%H3O1e8M~_pY5g
z&Y>Vjt(*3T!bgOmYn}VX7O4UCJK2}%f{@vFBFzr*=V>=uVwuN~ScU`BnAJf>W_$w#
zcR%jH#EX7o2t631xAvB&@J}EM`{;)|KK?84A^Fz%QT;C2$Ecogn3`k^Nl|yecY{>1
zB&_JiW0?NidL)2TDPk8r6xjg5S*n^79qF^aOCiJya>K3cKS&o<`8^Z;q~#Ft&3nUe
z+(AsvH(-o-$g`h9)S7Hh_sVdnUN>NuzR|xU2W@PwcOd@QfeC-v)1hQQ(i!ged;{sn
zns1ZC%jpahFIah(gIyJ}s*Ez7N_-1oDJlb5nsVkaPWpLoY%weArVKIbQ(iV864+$X
zgc-er5~9-?xZC~YKy1~eG8n<Z$N+){NgG}{`C>ZL%hSbSfajEspgH%f?BcH|W>5@+
zvs43RO~WG|$p8AVCSu-5p#j2N<2ILZoS#=nD>d}JQUB98RlL&WR@1y58lg66`$8RY
zs6{{QAw#gr3@mtgq^Q}0dQe^>E7@*UPHY2e6jM;}Grf~a*~uTe?)h8|YyxEESppli
z0*3WuIK=dCTTl_gTh%nia7ClGI;R~`JGXb_Ml3B4EdJq<ELyLSo(7FwK@k(xe@<r@
zPg8TM5)a0rpGV~x5Z-kUj9~l%e@b=<AO^pJ>e9zAGjODdS=cF-pIO7>g?+3LhqlqD
z25e4roqPkCsKJ5A>sQrbnbC@?)TF?woK9w(FiIJA@oArcI6`<d{8XB@YJk%Z!gFvQ
zw?f@6aNG_LR|r%KHE2m0-n<W$V}7uQjeSXvCLckSM#qM+Rr`k{$S#4@q|-YgYh4As
zvfb%nW4iV`Q3TQky(IHP6!Y*Ow?n3^xY9LS7{*soj*tPVlr&)5|2b<tLrRJ>Rii*T
zLPEqUO<OLH#p}me&s#T<dk1fWJL(|AOX#63vH-r$oJ>Z=i0~v_hWC%M%*u9v5?vlE
zNx$C3si#f^TI1^{CJ9wQ3>Rb2O15!$0y#zQy)1FJnjw!qJ$1Y-Z0_i)9y~Xg^sFtT
z`t-V$1GKx4zn}$UzX?h|!{s*^jA<Y9#uHXvfwwhpwCnxAj5mcZ9G@AW2I^S3Um`Lv
zFimxdCWh2!Z{42*`!RErYL5>#N?~1yV)Bp(E=Y(hrb<SDr|<v(F>1r{fX*=+G5-Tj
zc){zIZ(!c&sxsur+~c2^t%mno(3t?AmLMS-O*Kx!i<PKXnu-K^PM8AE;2Q`OF_j<h
z_5Y|vlIrKrmXd@))DnM4a6;4o`Vmq+&oR#`Ox~zI$}gaMbEaI*!cc+JKJV$IJJNN{
zf*AeHJT2Ju7`u%BSW;!p4xSdIA$<K!7Clf<x*#%M$!AtM;kE4N4n*7+AK(x@tWsBy
zjuQ?Z1d1FC94VT>5WZ$<Mu)ODGS}XoSyLIQ?lmm9?V}znV=>tCoVW}Sokdv(<>Nbj
z<!1WOF{M@kYAB=(P`K9Jn}7|C@@1WZ&#(=SNnYy-BQtht(NR~Y(Rh&&JhUiH?sa84
zH8%>zgjUDILe8I1UD=r7c(^1(Yk}m~6-o&2HOgaBF?)99Q`k(PM$`bpy{XX^=+v?A
zgl0}M)PcVqP8orS2<ns1s5-H|H78-)u=Kkq_bx5DJhZIVEXU0r#O@d*em=v^C`gQ+
z1AoZr8O#N^`{A+db>h^U5tzsI@>RTtbK^XQ4SF*>A6wiNe0%0{yl~_!+8{BlZwMFY
zF}k}3`)_9~2|Nvr^4_{9IRBwA4(MO+rlL|i_~KZ7`a?B)vbruA!Kvas5s8PRfA7}Z
z1c3O8*1UV4Tk{@mIGyUEHwM?d_XXWeFEWd|s-Sdi_c-Leecnn2lNx<@?crQjRye&W
zOXyV4<^aCvvZ3Z7@p(-$I1`p4<y7SOO;nvK*~7@BqB3{hE<y69Q6AD}b^q(qp#>V)
zIrfU@KsNJeYGCr@R$c7YfmLePPE@OXcs2tG+>P&w&skZKYi;a*{xv&|@|~j*6MgL4
zXq0{_(XONs$*l7jYvN~_MRcy-kIByr9_o~G6tyLu@dO8?R!jEC&^_RQH%gcAXi8Ox
z-20g{C|jR-PiZ(}75oX#AB;gX7p=~SfV7TU*u#am4@y<G%m*zF>;@@f41L`v*$0ew
zglyB^%8VElB)a=0a!~6-?v&YswPb2kuo-T_m>5cu6nF3H<zOYx(#O?xSRtm3*EagU
zSi^&gOe{=+cqnJ1!Z&xjPy?ipJ>qmrYH@5Gf?_-q84WIPcfF7BC#oXva4sSh`MNa0
zw`PdH!uos6iSGpl7iAceEOKBr$hG|=RS9^C)O@Eh$q{{{f-D(q^CD_pz$9Er!bS=O
zI2@5($U8gt%Pd=lX!5Q9JL<Y^HI5n)GECik@b38^kkmg2=LYpl1;rL`(8-CLY9yM-
zAe_#nr~%6G;R43jzJlsDpixgMa@WzKLSraQH$=#c|9Jg=^6VKZz*`9VuW}QE{o#+G
zJ%^gP1qqSUD`=Os5mUtP<1cZ3#L+CmV$%M_GcQi^tr<>Q4w9~y+(hA(OuGm2lKEVc
z&*-CYxcE|Wy~P`(@EYaYQBG2h+R{ov1}n+KNZj`v1q?^kqlzni82VsnQZ<vPmdY&h
zEu#Ra577gu{Mq8lACc=7!?1(`aL*2^iU?ADnuvI<0q~#V%GXrbOlsYfZ?N;Lm!9(Z
zlu2rBa^*SgNSgT_-=E#r#kHYw=_*o7DT_DMVCnI+B*azJryxY$it&W`0FrI|>ZT8n
zsRa@T<<J~hMpCezGgu1*Pcu92D;ad-*yO23CuTNs9jCN1GNWJmswccJG|Q)2mgV?R
z&My`c5kd1&p^n*GTNEjainup96yeA+{A7T|L_$@SF%!n7!n=^5eU5Jk8?InO5TXf6
zl^>)^r>%>OI2flG)bQ)e@Pv-R+1WPoZ6tK0uDF4~4geO6CUv{2NuFB>De*Eov8h}P
z>mf_^4Z-3_gZZk(sp=2CG8Mqucj{N4EEB2D11>@>#d%NcZyqGUcaI2xa$ZAekz}_|
zATJ#wV(>MmT0M~63QOI4eZK+yP!dO=Kc)*YJY>54UA>7IQVJUgM|x~i>mZ{&mxh+j
z7|;+T>P<dS%y#M@4uW5_A4;JfcnE0A<sQC6L8@E13#<?dsZR${$y)w&T#=b6C;%#T
zQ9P)LtOKc(2oCVd5o|crXsQ*x94x3tn@SRB%&1(i5B+8fkXc=R;UX^rn570IRyl&f
zD<DT%;QWKt&3bss_x^?hdmg_ZGJ8T-YJUBm%-p$5pO_vakv46IL7A<6h5WK$2NAYH
z>da2{3jqVvH5LGOzH}t4xv8fk5SZJ%TQ-w5B$9Bd#WJPdA0+~hDCt;AsV3HsJPvJg
zi6z|&g`O=JWyTNP{g#1zp9=C7n;)~WGJ38-mT&xQp5PfmrM!kSo8=J(UN69a3R5Li
z$Y>MWM?5}?VJZHO5~5n}257p+Qo#A*kUE2!mn>a33Qp>`X@x!k`+_v)Hcux$coRWp
z(>tl$Q4Ba%p!7{^tQ6V{$t#*lU&#5jCe>o!ha|%t5bG>LyvArA3MaxA1({LxyG`cW
z;0MUV5pff>JPj;htWwyi=23;SMsp6&K`6T!Yz;8XC?P7aq8Oh_NGEg|jBcn3G3N5A
z*x<?4`g-|W7(T66z0HXgU|)F?0F68nMx}=ju$|m*Ska2qf(wuX`2lqgV~-~G2qE`%
z(MAdhn3yk6e2-{K{Zj)F2?iSlU5Wxg*waE6IUV*Qw@mr)jSrzN!GHjjII!*VhfE0E
z`(w#^sy8DYO4dF`eDR3Q<K7=kOtT>P7J|PJsnkoJ;n%3S!my_HR2LqsAIejf2@GQ^
zxr8fDsd1lzCAL2*<uY|Mkl>XTZQnu`Ru7azH4+;z?(AKB&*?&CKR05IQ^(k7zY!D!
ztbk>*j~q`*WITWKKn+FLz*>AgU%>fGKV>*67GaThK5o=^B7$1Y{w3BR@~~(x{V6sr
zHone;J@LqWHrpa5aHocpjE)1?h<x=e18l%sV)n72-!GpAHF;b{g-|$DLXJ$~xfdc*
z){P3s$t8!y+)BM3Dloywt+nkzEPjoW`BuuRIi7;{SggemW(2>+a%MPA2Mhpp$;pEm
z!aTMesElUV>>!xr>zB@_c|u;|#DmEnU8uBnF#|A10=Vz7i~Jxkn?jso#%;YeaCfv?
z*%{90>hn9;6@G_JhdS)^J<XYTsMbdz;5hKOO_gEiumv!hLJ~M(kGg>Z6v65x9?SuB
zL{%ZwYLjXogHp(E9rBGl?8WM_VX2gek-&`1VU{JNAFE3{tV;PeIh^F-n0{Y?)VYh2
zGFKSBDrLFIKfx_i*F^&$z~_%zQm;<3KA{3?0y9c#u=U<J5K2#0?GVOLkPuly@Q_@u
zvT6kkusFoPL6NE$0BCD|@uXylpyY83nxM1hM?55gx`{2K!fDNd8^dnh7;P?@$-b*(
z*si=1hicugVpXrj<Q5UvZKzHM6+KbVa;hsc0YZJj**1<7^aP`rKsL09kJJ&o45tdh
zF$pq+n8XKJ-X3Hp_YhD*%ELz0Wn^kWTobWZHqF^DbjQdwlFF7CZbSfXDllEXARyfa
zK!M^s0<Lig6%reX6`6vf5Mul2jTk8rrdeg+y6u42Fb4j>0N0gEbS3`6<#D23`S^#2
z`r8R$;`aQhK92hEtN7-YCUBsphMK|%DE*-}FQn=?^0*-PTAXzx!%6{sM(u1{$u^00
zMik3xM#UUIZ~!O+mf2$D*HnLB03z`|0wGe!)MU9ik_$7DKIEKehH87`P7y>U?a#?S
zsjHIK`<-31-5H<(3H3NC0Gz-<4sFN*6lllBNE2^M_efS8T&TODI{K8M5rpBbpql*Y
zoF%t?tN{h;<VvpKbQn>b#>0bgVyCPfG%TZF8ySNU)!~pM^)3+)&*-eTW^|O(tU&YK
zKWDRWaiWgekEApNP|{5^SXW|6u_k;^7zWqUnN7b$1~Q~3#84CV2jYFWdU!3Py&+i$
z0}5gNEACDW2<FYZ^cn4;iQuc{9>kqTiJ(%3(2LWKNDD3@9k0RFdL`im^v3zik4)u2
zlf}|f{KkDej&?(=eOWZF2a|JkYO;4qMS_}FxuT5<167{t1RZw?q**#i2JS3REzQJ-
zoVrKN3dV6TDoe<CPkQ+O*6FYrkRAYP!>{82eh4ib`>kr@pGV=4?0I4Pmb{{Jl}T?4
z$q`2Qc&BA0D%(j*@O2#bN>uZI`zINUwgf-}49QMetq<z7KAf?uDFlrYqWBMb#Ty4M
z;i4c!Q^Uzc9j*OQx%2#;IBD#tp1|S_P!f&GO8T85wcoPTYoSz3M#m2vDugKpb*0h8
z>Dz}mjzME%xdYzjro)m9GpQ<%Zo5Qe1I{!A;*F|0^o~rj5=hIbPvml#h%=TN5@>36
zK<kIr1%e?eiAE%~k=k|krA1B%Qg_!N>umIep@X9*Mu6VwPAXyOOhF0ou@Xbb+k&W9
zphAMWmJ`_&t%dUyujn3v#^JVcPtF?B($(cadNtx1#ZJmW45chn&<C7b48IJln_w3{
z#@_`6{Cy+<ycFowpl`;$Bv6Nmtqx`rY{2XNVSSVM`_N>%<44Fk?KEH%PP&DVJWF;~
zh$d-qnU*s4IjmUw(v45^8ZsP9Gf`}aS4k1RApbdr+ykfAUq9fV#+U>dzaCAEJ|ayR
zwRXye7qkLZdmuPd2kT$-_U%>7Ssc@}In+G1(nRI|h7?$kPyot^WdI)aXEP*sOqH(P
zDTNH40I9B*6zSXW22-Pw&oJDiSzx>YDj^vajU9maHp6)fRi24in;Bt&5@pLwYMVgN
zGxDD-Z1e4gj0l&>!W<z4qmCCcm2sTe8+#q-{{H;E2IxP3zfMj3ryC(-B>vNt@%M!I
ze>)B@rwW@q3+3&{r)GGs`qRhvPo<QVmxOM_*7(%;oA`^|$YZzmy;N{@MAvlNsJfY=
zAm6R!B9}I<$Sg3r=Oc17*#+&R<kWmCJBBaHd@pV6`QEdtTFmHeVJjDlor|;OkX=TE
z=i6>U*7kW>p}s!<?$tl5Nm*X*-@QhrAA7*_UoX`iyKeoh1^BP;(tq9aIK5qy_20eJ
zvtNeA{ktph;WhqLruFZxKrU$W4@>_4_QU@1Rmrjj!C-Zum7iN)xu?-bbg!0>SYc&l
zmD;s0h<6&w6joL-KDo%PG23%xadGib(uGE?L|s-cF0P@Z*|=o&Z1cp8-&$MGB0;T7
z-ov<(@-FDC_U_%A>>@&cT{KtFdFw||MxT0mY~#8a-<nAwK5_?(YL-<JOHu3p*zi2#
zlfrconI)qZ+g<0i9SBN4<lVKIE9iYkuBW6QLGfF%YoF=At_xzkMTk}tUL@?K-#PG7
zLGj-WB(C*;-$=S{W&5|E`@e7!{p0EVe>!pgbG!d~+W$X&y2sO8X8G700<kFO&^CwZ
z6Y_<8z$D!UYf_`pRemn6U!rs_Fi{W!_u&xi9M8ro9{=Ub5iry5OxzF^qCt?%jt@RQ
zHqpqm_U_+5iwR=h9Rw!Rn*{s$&Hve5yTsn!UZNEGi_>s1Xi|B?!{cCQ<FX{+$rkqT
z9FV@#P!}7K+Yb(k9L$_!Sy{zX@t5>Y*dgBOadC3`LGD!(cR<d3p5Ls_W1KRwa_0JY
z3tV8p{gyJVpv~b@?gv)Rf1k5uO{ep>4}36mnFYjGq7)5P3S@a%717A{%%1zX*mRt3
zM2b7khOf5*xTGFUp$)_1BaYq8Io!-FTzNn$fBg8o+uYpz<=eL`)22;B>bmgN>(}au
zx<c+c!Qr~*=4-D922NuFB$DcATXX}#&s6|39D=Bcc(&^3ZaMQ?pgdcVSxNN7e)yo}
z>h4ZHAoMoSzFc|vQr2BZo4LMG&U_>8#F&qd5AB4OBc#8dmW4EC&ccQ1nh$-70ucJH
z786^JUGIY)7c1*&AkB9Wcn!tf3^WtG$vMp|cVt;1%OZ878#-pWe~*Mid{~?W5VX8J
zIi^!jpKb?<VcxldHQ>bpZs~?U7k&Hu`80?!UOSi8f(5YPF%~KVk9Q6hb;}>V2GWq?
zK5SR)1Rcgs2qgSEcDkKyc=NQ@7mg-K^ItVKHXbnoWA+*V*xG6kjI-o@mkMRxHg7<G
zA;6wnBP>*JOXdPJJdbR-LOR$%1GeHX;pvd<HE}X3DvI}klamt#>su^JmO_luyhmg$
z&*5d<+{fOz&B*fL*G8PW^l5uW<eSwZ1GXxwSFZ+`bD_fHd((@{psy5@S~bP`FJktg
zSYjGjn{9Ph9PI6PB6UF1s^g&Ur>*c`uVaB0TS)i1uNlHv)B>u%mq(;+Tr};ruq(@r
zB3D0jd<`2D-XZr(#^%mrzCkR~znCD2O{$QvX?)HGO9uZo3F*OC89z?GbMrum3*F<D
z++#0)e12_@dG+ksvzj=|@{1`ZJPtE44b$^<x9It9%{x>3eLd~F3T;~3HBnqL2WGaj
znaYi<4~cK~L*SikS+-(FwNKGATpDU`X3Q>yvxHO-WT^J_ZEfc0)yw3E-cNK_2d}~`
zzNCX~*|NhY+)H-g?yY}53zhMfhU9bWeYaKO4l{wY1RI0|1ROdrvG)Zs&wEI^#q3+o
zmG^@an+5rcI9zZ{B7MaJP`(2hDi?fqHda<vYS{*ch7uxS2C|;`xMY*u=@(Zhr<xTm
zRF0H39P@(96tCg)$E&Q(-I3VSa$sUQbm$PQ*TW;M<;)w9xb(FZFFD%|TzXqwZl0q|
zXe+F#eGALL_Pql}bYE2!+dy#HMb*#8?LtFCfirF1a=4r`DkdhOZoPoOJR~KrfrM{?
z=myyNyY7AzYF$K@!ZBa61gk+&+vxYxvXSzerD?Qu*|Lk+7gh?g%$+MNp*a1{SQt+`
zpcRM<?^ndIHRO03r~4Pg7>O-2Vv%4ucI+6^<T=D3>3NYD?%<QTxXarSM>`~!&vRCx
zcg*XXVQ!88uoKr5yxC(%56wX?D>_mu11bP~u?UL|H&>G>As3HnveO~sR{1^=tfX^D
zd-d0DE|GyFv@de`y37?(uh6=7{*CCE7&BOFI>-7ux=oIYgB!bo#O`3sK-S&>9v~ge
zv`IkuYl1W!l|ih~{VxgP*H+p9UD>Z0*f`~FY!V+PZ0>^SyKYEBPHv`#re=@fy09H5
zf5tjbp~uaU-fxRIsPisdpJ&dVeGe2CJKxz(FX7-t)3F@WT||m@Mn8B!ZGM1GJ@{7c
z=&7;&+O1Su0FrDw-tNzL-X73;NbqL=J+RSA3qK)8zbS9|G#kFE#DsTw@Opn9G-(pD
zsoC=E`STTmk!ZOhymf1j+M3Q>-Q)pN{3Xv7?mrb?>u%w*FVNkB{R^+QBXq=zN9RmU
zt}MLDy<L}U@!~Vk5a}jz&Q34V!mdvjgQ$!DL{A-Cc6Rpc(vXe%7siHr>+kL<-#O<r
zrtKl@D0LP|Nl6f#i2>cXh@kH?V1`}T4BkORhLdIy$haA%rlvn&4|fk}MP1M2#IP|i
zHmyM$9XYvCpiInbH*8ob*b7-`Lu2DcjoEJ}%&*WiN2ZAoRDAjBnNaMV!ARFWTnKgb
zH0-x#6_T`O?=>hXE9XJ+W|N!+lnL<Qqx_OkTz1R*1fQRMC%5$S>LhU@?X^fA2RRA?
z;OwfSZJf*Jt>;mA)i<Lp7XQUpzG9wbW*Eh$oC_$PkJ*M9U)&C9-P0rDbgaNV(4s=J
zcmy4o@ueh!-l91yEG*=aG`sZu`}aWIBzJr$636ml2C7k!?c(V88q-|J<-38an_ID6
zXG_Z&Aa*S<`}#33@P1Q{#|V{=EJ>&e^$C9uAY&b~!Sox)w)QuucEEXY+qp$_-`Bu8
z90J^u+C@blSaoDV`3sh9$yo5H!I%lVTR!sEGzgi3SpN`U_>6J|ZhE>w`q-#nfsUJh
z!{ytDKcuCl*#$NT39;f%p5*u6A3t4v*6pH!;3?JUt=wk0a&3FXyiRd)nB_*j)icXw
z`fGmZGUi9b&6~@yw|+FqNJtO>cf0Cx$ad#$-OJKdx5AB;l;*-#&%F2OwYk`Ef()`@
zlqGH1G{2&v0_Kc~<*C@sI8a2Q9-#m(FvYP`Wht+kY^DurE3W5^$;l}Ji@u2R4-Uac
zkj28d(552Q_4@|La7nktkZChkhC!zD0w7X>hRxWZ8jz#_-_HdU_zq$%)UnJ3(UcFo
zoA%unh?XJCT4`XQ%)-1!L4kT7XVlJ;rxRA)4yTE%Lq@~My6u2(F9Idgmo#=T%Hn#n
z16sVR5KFBTM4zWcFkn1<aYehW->fA8Yd-&)=5PZ0jY>J!zb470uonWKV!33z{J@`o
z0<7b6>C)+#m<^a<7n75>0JmF#6_K&26$&Vr2CoD1O{?a0VIlL*ojc*Zxde0L@aY#t
z_2(9e9sk6{vu2GwB;Jy4{WHIQ{mKhtTL5ot=gyt`Q&4a|PU&K8@v2`*m}mB%UQXkb
zI=OV&CWDu)g;zg!bX-`xPU#MC=_9B?+O=!for)t4+Ad5K4$_u~TU;Y8%|&q#YJb2M
zT|n#LHvV3?eh^^#46_iN!8rtLzeYwr!&XZHTO2DVXL#4B{%f>BS$DX{3eH}>%QoBf
z^_NG=T!lhQ*Q1JG9D9S9>yJOLa7p>%#^^JUQPPDg1@WH7YP*XQft`D1_{KGBW&jZ`
zK-^{zCq;8hOP)XfWCGy23)A{@SC{pH>_7haLkxqykXzahN=zQS6*f#+#CEL9m;1bX
zrv`xYRrmO+G6c3iBB#cgQQVw*QV&Oym}AF$8oAG}X5|$Yu35c$+U?u7e+&-F!;t(P
zij?5WC2KEQ)UX#D1fC2l8{2uzWLH<$I}l`$r7P`E5#?%{G^J>jgM;lu@JNqgh5u}D
z{k&i~DC&dK3^C;E#PQv`cYp5fy?T*_`<viZfB$@Jcu)mfr}dXTf)|<p<Hy^2#(w-b
zPT_T%$C<NdQLw@brI&W54Jc_8c|^#6`+6kNEy-neTwENqhC84b<u=IiT6O8trQ=Rc
z=N~=V3{BS=+%>u8R`(2p_aITL&y_2CY-|LfFuQ;|A!gO`Y2%V5OHiF-WZSr8@nSU)
z(?PuI(-G+HfjoJm!|lDQs)d;4tQ;JdaMzOFlbld-?mBXWm#T{rLhIgFRvw4uyt^@T
z8EnrBa0-3;{(T|f0I(LzfbH&t$CTp>ma1e(rHk89I-IKHmM%FwWAmf!YcS~jZHjsM
z`KzU+jmC0V7cDw<?b;#;%I9EyJXRWU8*J!30tNa-I4a(pa0}&Gw{8}m_iOBc#}T^H
z)8ga%qp7Kho))l)idgkE;9%>I4raIwtUr*bdm5ZROlz5c^!4q;Wz&N?cj3aV;Fv|z
zF>Mjvuf-a@h%>7d6bdgK0}-PA06%dTGM-onH|3Km9<~GNeQ%}EPO|B7*f;JOKx;y2
z2$ucNotqh0QeIRfi{m9AC@3mz9KMb^+DR*HTmoMRv3D=uWbo`1FyGIh-F+@!-i<wu
zf@q*06sXOaH*YsoqmJ5pjIPeo9}x4sbSW1w>ks5cEhv$21>cJGas{hr4OTDI(<zE!
zd0GlrICH`FFSxOcUu(|i&!5>h_W~MOF5@MMS|-!JtY|T7!C&}}nAf-%CN=A-RsO$z
zxk|bZ&BxY_BV+gR<6B<6dKJF@du~^5mc0~1cYu8c>vgN{Q;!h`#D9&69s>deXUbmd
zCJXEPa~GASn4D^5juIRDHSMs}aOcy$8R>qrcD(vEt>8?%xck4NC*ED>kxk8G@_mGX
z5z*7r#@c%_$Y+%Gtl}qRd8{~g+35f1ov?J*oPb-m_$iXbJ*>`rgt@r{k3mO!LmcfS
zYhfqBzioxT4F||VcHlS~+Izt1J+^=xhb2Cpb(aq+k&Zrvo7dUgQ4pS1)yy+z9_i}%
z{yoWX)8NTpKkj~$DAIds7eQT}{J31mL>3O}ti!#{I&Y{$TAeKffPCkY$6Yp_TLk-1
zR&Z+pm_;H7hoO4#1yvqjLM5fMo9(r<f@mt(Os??XgQ#*ZmbIMIL>fG=ja$IzZh6jf
z_wL;mgm>#ZkJ*B;3QFw7@vYrykHGW(Smb~<;BC6uP!NMJ%%9dHa`+~(O6sH*s&V9z
zN5sazhqK*kJU*Vv_wTjax!Kv*kH|E}sB+uh!m;23x2fVAUjc_z_Fh}G`D!~#K2)D&
zWN<WY(9zK$d+W<cg#t^GgF*M^MzwZ;Uk@ltz?-Mm$4%rQ{0&8?(;L`=65~gg3hJt%
zVn7a=sJ!+h64Vf>Wp8eY?3=S%?%=^&6r;<*^x3?0!Q8ov5uj^oTc+cQQA5*&y3O3&
z+>j=(4Y+yphz3V3Hlj75#~}85Q}Xs8Hk*6D#K~6D`Ju6gM^>)=@s_+l(gyY^DJ@1V
zy~GI89c9r8@Qd^7%#R|;Htwv+_L_LU^H%r32`pTvZ<_EnuAK_NqfzWQ^w4nvbKI%q
zs-K_M1P4(*bv^5qZ<-`~VV=|tiq~sTPEMwdO@WGv+hnYQOTQ`M6wiC`aco8NX)te>
z!EIc^-3fH)VL$5;MZ~^r*BsU*OYXnF5p)tQ)0?71f50GkBe<8SnuTRvPG8>b{OGi^
zJl(y#*A<^3Ve2-bgOMQeP;Sqj3*ScPt<FPOkOwXJhVh9Bwyxfm7M_#P@*y5oE@`f7
zX=yo^kdR;=F@=rz<A_*YhRoKjnkbY#2u4%YDFOvZ_eUkejmsyhsvz4Byb^}oSY0?0
zp-G8dpl4%30R#_uNG|0Ac;vM{Zus+0SwI)^*x?dNypfEh7M6kOIlpL^r0Wk^G^5_%
zcQ6JpKnobx_+f!{>*T4qB@Zv}-j7}{K$McP<N7JCx?QHG_qvCdEMLwGu;SstNfbFP
zC_$&H6r9+p-JN<dDysrZiv@bV+h7^`7Jh)WcekN@8m52wm#H^cW8H@P3(iFFvU?!s
z)$Vj#wrm-?ZmpzUJAb_Yq+ktup}Go^5H>gir&oBw!!ip+$Dzd|`t`V%bzP8k=m+)}
zkA&Up(3N!T@H3Z202<V|X-q<M!D0Vd&6zH$eq2%qfBpIuTSsk!E@5xDsdeEYl5X)w
z%*!?#368D&vX}G4#uD8vX1U_VH+*KzJ2cWa!%duBaAqOP%@U467O{bh=FkEbv4es&
zb*;?H=as)0Up%jz?g{+8n&bnVxIKpQl-~(!W@m>g<!FkCi|aXUX(}z<r~OVR$B&}M
znL5EG2zGobgtK0}cv63}l4Cn;IP%|iVZ_!OemI=tutUuOoZI@4ukG#8t^wAUdrxA4
z#P1MIGseDv1z&iiX*9P1GcAGZ3knrQK2B>yIMfG2%3F_FX3br?2-#6)Mb?fJU+yg7
zbsPVc5-$<@4d;HUlZb^$VqC78taQz-dXpoV%-}hr^ro)9EWPvOmTaKGq8!9!yXLH>
z-SVS_%B^k1N8C6#I834CUq98%v<T4(8Nlx;O%~DI@ow74!kh-@JKQo}1_0kH63q%;
zzupZ2^V=gFI6DH$sCax4!UPfs#Fu{PRSW9>)n`_5cQ<p|5IMB-@bi}(;TVM^b8V=x
ztn4WixeJTSSlj;$i<=xDP>mQRqtI%%g<O)l8{R^jy-U;<NAi=e%3rl$$m_4df$1Q!
zH*i4m;ZU>;Q(0Z-XlyKRgspN$=u^aVbv@YGk0wu|2vW#x>g<^_S|OX)ZQNLPBqses
z0=638k#F8CwBK`z-5AhD2LSDyw;<|8GMaFF;Suf<GM-2uLmZWdK>g8kKdeU`r<e*1
zyFBbXmFq47Si67L2nCm$cKE!lL<q22Pt~lNZIttC1NX|60yz4VlT?SS;cAl_F;?4O
zn$cXR+#|~}XU-y61#K-VT)gTF&J&R?$=O+1>e^*5tFV<6cR;bBWDgNbeydDp0;0Ya
zgbjO2ajGU(qz5!40L<dqv`OV-=V1Y4-oXza-LihvfME1mZdnYjy9K$4Z{Z#BRbSkq
z8zagpn3Fp$q%|q$>n58OoB@WjQBq23KmYsueEGh#lCBgJeNKg1uZck&^*gfD0*6|W
z1q;_%rfUo8&RGqNZmpiG=InyGhw93-PK}ox%y<!6f>di)md*RN@daDk2NByAex5DS
zfAYtaQTYieSLpj|B{v&9GjZ^$YWZTTUU0_A7?9&8cENK_1uSjty2)A-19iWq>0_bW
zkDp4rJx2dplYhom$m>+fr-staDW9?`zn2&MW|qQ3R_?Wi*?^Cq-%KVH0<#l>iauS}
z0s>A$-g5|<<umu;O-!Ev*P#0o_q}`fluzb40D!O`>9=5#MN|p3kd9WY+B9%PK&xir
z<U51l-?|ho(IP65(*6i}^j(jwdXF$~i`TeokGub#v&cdyiNS#=2SsJy&##(DmzWVi
z7hns-k>rDErzm~H@QO%w7s3yeWlm$lM$O93egVnTeH@2BhK34|$AapfWz+DCb%RPc
z6Vsg4TTJujpjvYl#E=@=!}v>rC=?YHkGr@iU`9ZfvhUZ}SSvz$Xd*Ymb@vj$T&B(3
ztDCV8QVL~fS2EQPaES{M;X}_}WKsy?kYlhDUVs{^<49L9V`d?ppT-3A>pmou6o+AJ
zv>1>^!h-ZK(R1_g-fs<cFWjHDIjE|tQVh#uL6-$7)-D5s@bZ34!$X*~2xL?CioYyX
zID%lC3n7v4j;omb_X+qzu7+MFM-Ffy*viG|XvYwIgGF~9L3QyUU~n>wO}1~{llCp}
z)-7LPsjEdr*-^o_MSw*g<!yf~<P<YW_CPI64cb%7mqP2s0u|HYp{ZJ&eOs_0Qi=f#
z@u8kuRCM%bXs&KF9L^BNN$P3v?D#YU;@7RXq&+Cxl2cF!(<alKyYK<1K`A1jrr47^
zG9IU-q)@_z@XP9eXMb4tfL%J89qg!fOSi6Gh|}Riwv6i!lca$YGpi)4cU*>QdN$tW
zWA<i5=Y*)?y@U8PXJbc|_3)5Pf-6!{(O$|vuqC>V{*3!3eEs@$%IF(1U0AR+enD9z
z<zA;wF-dxkE(KGdo8_7^<o@jV=jo`w^+%gg-j0>r+>a*4C~rkIx_YLwEuQ)7_wSbg
zy)D4LeK}Y_hn75(5!Vt@QQ}}!Lla5uK^c!<Tad6VV4^3ADbBih@pP08<>lp>0BC?R
zC`@z$i57CbV?RA2uU$KZovQ^odz}Tg1wsU&7CnV6E6cTSI(EHRC>C49#x{!y4*b(_
zZV}Rqo82Sa3v3ub+mU93s(K%@Z|zPh`vpJVS06uq<guvfZ}1-eZiqCMNk~Y@vJ+~K
zUGzAh3_E}E;>F_n9q%)mZ%17TtEtgNus08C-eCFCTas?V6jKciVF!}OZNmWM;E8V^
ziW4O<na)F|G9QmBzOoSaf`Vg-P9{&}wsNkn&xVF;wIDv*4w&Hh&oAmo7hfex3id#Y
zmD&aKDmcR7`9fEE8tA~9O`8^i|K(M`Hxu&z(>PV`poVLG$QTyzPn4S@{gx>89{su@
zNI+*HDgw5OtFA-#^iYW1#v2Je+5m)<A9F^|f^5?do8L=7A8#XN6p+MT;g-QlbEf8a
zIHPuHXk7p1(C;9)Jh{n!8#Gh%S4n6c)`J=iMV33L3ofrl)ErYYGfaTf*hP-yz$;_)
z4+N(~F4#1s`_5{9?zFG7ZtaGB=~GdR{xQ3Oz@n4I(G-Vp5y%7qC%;08f!zU5CKKon
z5zNlUXEXcnF>{tqf7P|q;);jV@}*19VWt&Z2O*NZ3-bcYj~j9h3F=A}obgY~|8+Wx
zYhL-~R%XuC?BO~HUAi(NTbYeg4+z$<7O-q#PgL=`G%KxfKOVK*!rf)7bVfcGYn_nV
zvgmIuKyBJ))c+bdKZpdiqmGa{Vgk+s55Ebga_{V+M9Og2t~0rU)`04LEp)9#D}hh@
zBFn!-|0nx|_cwUecV!xU8wA(&(diw!q`TOPV+m{<eMWN(0=wzznArihZ*K}KXKx<F
z<dp|9OC~XFUDqQ-JN*(+-K8V7^}vw5R<?&*t=_54#@RMt+rS&Kb!`Jq8SxV=;W9kS
z!2NJk@KV1g_eVr4;20y(7n5&pQwElJHA8yxm)Yev=k9Gw)7rb&hq#rt=P<|MXbTPt
zjaH8My6>xYv=SR_^R=mMmTR0`T(&1<LCRiy1D1MoE_Vka*pAmrw2`+Gt%QWI>&=Cm
z15e;0e>Uq)g#S;Z0Iv!Q3r!2lfW=vqYTOEV)6i-Ou=l#f3)j9~q(Wk^jlsIJ$GZ90
zY>f!NTd)?T8)^|(pNNkJ#zBgHUo#R)-cE_PD3kosMjbP22jWByuaP?b=@e~`Kvy5D
z)L}46(?hr<UGD^5eDZd%D%}4k&ZD}CxpU^cdjM`Mx_govsh-owC<<xh5S2e^*NWFF
z{*qXafuKfyY><p;DTfTW7ujvCiqljOox&OQ?Xf*hj``X8o|KjXhYUkH|IG9`a3Mm1
zB6=TU&8B`Gj9h!olM`#aw0wGwQ`Z6YShdwXq6kNkvhY_WpqfXbo#?%EPe@nSq1m>3
zgkE>k$*zsHS%dY)wOjA$??$e(&RYP1p_PX8mE+%46%|=?b933n6{atO-2eJUft>KM
zhx;+e$c0g%)ZN~`4r|D%zqrgM3fRiy8X-+&ShP>orUqHA5gC3KAkd2gcfE}EuJ0eA
zCB5i2x)V>>e3hZ0VLqOewv=r&PJ}&Xf}=>EY_F-7hu+&LwbIJS%H^=*&zNa#d%RqQ
z+01gUu+F<J<DFfe4I0l-$n(H~8-TeoL7!fMVDqWXb1bCveEXjO?7xIxoH1hto0$zx
zL~4{;^`QF5HeXfcWNkrLRE=KGQOAe~EsTFqE^dr?fqjL4fr%R~X{U6Cn{aSi!5OfV
zydxu_11n^_vPL(5EOC{(IR>@dBj)zoKS`{3E9k0-_(}bs6qAB88`<WuzdG5<jEzP&
zXVMLU(nhn~@@-x(_1;eG66^gnO~GeYy~8+@S#Cy%v1#tEHh*UqQE7%OsK~9)GM;lx
z4ME-gt9{u>z{*Wt60ik%IQ)b(@@;X$o6yc8A&p2h^Y{7*Go)gm#>1jTCuAp%2mG23
z{L;fg7U>Q;Yh2AJFUA7J9xIQ*`6ls5Hw9dY(yBki&)!9R5M=!Pgrd+CFw?m>ka|QB
zkWg52-Kw><m2%9-R_O0x@pCKNE+3G4-#qpZpl@?f-{}s@;AQ)_^LNT;uP8Wk7Z~ay
z^fuU{+xhh?AClb@{leRqk7C$)C%>O`bGt;1F{reXT$W7B!^QU;djJc&Gd6AxtG<XE
z1OH=1IL!tgb`epkI3@%4cXf2EcY7l*FVE+EI|R%ph)E-xy2px;AMEyk_MO@?gzTR}
z#lZ4ak4*&%B;HZ<`a|~w(><IWJjkmzY<Ap19VrujtwQ@9IfCQ68!|M<p^P^Lt6HbR
z9vAZT=~EkpiyDihI>{@7+XoZ)5^`)`m^JX;pmv^0H=>&l-JhDfa3MSNq^y9rmmx-D
z1#=n=ASihMvBV`x#s~}74#1}0M{z_q@P4y0+w$eHn%vO~w`;c9Fi+d^#h}QtY~fEE
zv~meksY5a!zp1dU^Q5)&ole>vZ4XAPJZpFchl}!T!U6`<3ueFn>diu1=y(~19zb1}
z%>YnCfd{I}%36lTtIxpzG+FeRi;GJi(R?RHQ2@CJV8}8A%f2IzV2N}%45w((15|Iu
zsMJB09#_XaQJ#h%aNmP8ZRV3oO1{VK?E}Ui8X(M3bQ6D-;SCX~Z_X4L15-HjkOtTk
z$30nL-Gf5gx1LbT>lW=XjyWQHPIqAJXvp9<8c|q&FYs!{sxO!~ubP{Uo%JdGhd>*p
z0Ny@`gM2Y(w9_@Wj(v<$TvFQ~B=IiIc%4$vB8U~1(oj+^w956?#EPw!e&%|Bel;4B
zpowllNf{eL*Tt|<V@1}cAt{XlKJ%s!n?vHvvF#oi&!6*G8Gp;vu3sPY{Bsg=TZv8$
z>l1f6&tokqaKF5@a^$lm|EUy{bvU#;Ib$`ECVs59u(T{O8k-X^Y?hlYc4gP@-A}#(
zY~=^paO?y(AjPO9IXWwN%@cHgfZAE-eZS<!Mr#KnrL#Z4vE~Ef706JTTEE@#*aL?u
z-QyU<Hl~X1keHa5UqBr55trqGe8~ego8)>0@&&e#&Bmf5V^3xqu_$pfmmin<g(6GE
zjsalDq5s-0Ec|AL6&(D%S;SAe*<!Q6+dt@=QGW8fOGn>~p({W$>icFO8#D64t<e4b
z774axxo*|I;)%9*vGESHWA}&_U@tE{@*y@LUijyXB}&uw9{wi#R(y9Su;q<69&OCF
zc7(sd?l5E*6Bh5kGVg|2-9GJ+E@u8VOCPwB_^L>lrJ1$=1Y~CQbl%*#*Y_VzSa>31
zVIxxbS5Y2*vv!|fM3zz4y9Rf}y;*R9f7TPg9c$1W&@Q=*kT39p8#5!|vDw}GBxT`G
z@W(4`9NSCSA~qcSk`p!C_RHI#umNqT{EM%gM~amH_VtI;Z>vlbXBf^|J$oT1Z-$Y#
znjcc&g4u?CE4QmSWFjQt^O|y)vDt4{g_M-ky2E@`+_FIf?FeHaQ8-Zv;k$w8kvXy+
zsLz`{W6`-qJnPr*iP^fjv-IOfZ2~YIa-xbRjQOMv<yUb|Sn^NQ76644y#8p?#iOr9
zzW9XO!x@d0ohLC)GS1yM%d^MPtA@Q)f6AS3iV9micPEM9{XM5dIsmszM63ZD7@)kt
zY3oPSd}sv+3>446!7((P0xqEL#Oc$g<0amnU&JFNW!PB*zneSe64)%!ebQ*?Y2^nF
zfgVpfJu5<cw7*FPU%zge+=vw-2VBPDX<(+%dYOo^hsC<gClPAI8f2vAKA%2ot`C4)
zTj>txI|2)yA9r+If;=a<p<GwtS;NbDX7z;QwE*;+UEj_r8#Q(PdEWpMEp5TWAXKgo
z)~6l5+MO7CMfk_AGLyraMp$!Ng1v}GXQzF94(h&PzP10tpIPs|mv?-bq3gINr-ASW
zN&WmY^?)Kb#$X+?mu`Plc;=xe=PF58wGEBP!V*q$qe^F4gl+rn29pHc<0vOv>wO<<
zX@7ZXO_>(B*z2}L9CbZj_U1S2fMDgu{@&KsHSW?Q%~O*bG9#iNWtQc4?Q9I$YRFO!
zXjzyWds%U6Pi{s1J-=B3W`Z@_s#eHc3jt)&HUTCf3-ATdyRHo5Ih|ZGCcV1J4+{|r
zhr5sPv4=0KYpne(9}F09m}Uh5r<(S2m}eiEx^n5#POQ>z!Z3cBq5ZYVV_-kc|1mEF
zp7Etcr5+w0kG&CkeK^`)(~^1MKDPh59O#nbj;;in07@u9X@|*6CIIPH&Am7~gc}jH
zJn^muX@y_a4whj;Q&Gb)mlG$p0i@dyY6p1{g?&TqU0n&!=Rd(&eFeawVv>M%d{BOA
z7gcj3rA_ugjEB^6S5iNacMdo_Y|QjR2@UZO^KWEkX0~FO&(hhHjq(wp3~5B_zfhB9
zoty=npP0RFXAHAMZ(v{`RIm$+A0vfD-#;nrWk-{fsJs?g_b0584C)RY6yK``+I%K3
zgR||ZZIeBC@L*EE5r~iO=fJj2phmggMK!P^6PBeTD-ZdiYvN*1NkY1l2y8&1$~uGr
zW&MO(lqdzfwg?bes5JCum}yc52J^$V!K^QrK<^AgOUv~z_A;|s_PEV!w_S7gLy4>7
zg6r1hfBJP}EJ8TgOB$n_vNwCk+~4uV*5X>DnTgewKf%j^X!Cy{0M|9BbE7@I!B1bV
zdJRz>g=*P4^_KJJ&V8kS>_s%cZMpT*)#uutD43e^Oiqp+>lwrBzYpn5d_QCm??(h#
zIK}=9{SKFf3aiZsoV2tjA83sBr0)r<xYar7<1+_DBAy3u0$2#F*K;PDM3f>Ib*$vz
z;HytM(4ySw<?i6*L>=aBlYzOdLXKYQG^u1;0QuWeD&#QOZuw^UHoFF0XMM+dKWqp)
z@i-UU@P_^?wf>9oVJ5Y4FSIyZg4A`NLnE&GJVj#;jbN#MASNp6u$~9)5}aYq%MXN$
z+289vW`E!SJE)SgK$arfLA6(33ecTb-)O$iZ&tl=dAg^TmR3^zp5KY<xP8z-c+dCm
zWbYH3v=1}csqRofK*05{Hmvhzx4mVlI#R8ZsH@(eNrXER!SIBv2xO+bPb&}-oWuAk
zCanY$2*P+>^+V$vE{Fh9M<f;-v2>Jm?aXMthWvr}11ISbMA#1w3D$_|`*9a_0pAnz
zdU+tl<nRiaPl=tA<I-L)Z>N~J`*E8H>O#x0Ipudsnw|#?IEP2xckgVGC?xDEoi8Tk
z&tJ!A$E$=n{%8gFdwgQl>0^C|HdIa@CXqoLRlY6BfrS0(5sviHr2S=6@L!5gcSGaK
z`QO$RHaBx0%j&x@G=Zr;zmnh?<f~QL$Lb!(#b>3#iBP+90te07_agA#gec{GHe@~f
z`4#+;{nk1mcW2+MdcVR1<2$=_>(;F(C^v>E0W+a?;b$zTyEtM}>&Y++DQuwOF%{-f
zuo4#4#ZIm>E;K93$(^&=VDIWGhEPBtEF>o<$L7u?*kX1Te!YE|Jfwm;+m5QdJSf{7
zgP&Hq>;N)#WZR0`L8TdvB<^rXE@^CBMLAlV1a=kR6Ge6j-VMAxI>3Smy{#QY1RT(9
zz<df-W+iqps^2e6i*8;uDg#Vpc41`n<l(`AfUa%D`vg6cTtuNrID@Mgnt&4k%$#P=
zSuJ=_W$)fO28M>((Pbl^de2|PM&{Is-sh`c1;>MTBiz9fS3vS2yNmXW3=?fdoVc!@
zI%ZZ`ZAM{KbCmFhRHv{Cqs~dlM)x`^S-L~5payTcVy9hmBEmX998Y)5dExZ6au!6E
zvrEx>Kvhm&J~ZMOR-j6_*y_xt8WY>-!48ANMCOC$yxs+skW+2*p|~lt=zjkN_jFb)
zwdrkLzh=#D?dJ&24}tO}un$nfdBBT&=~Lha1TsA62}N|_4agkeqmIO~n(yoNG5>_x
z{AK}Qx&!bwb;g&n;A*qnQ|(vL;ciY(GiJtY2o)YkcpDoVL+p06xqd7A{Gv$i#EK+l
zqlLVIC1g&T@qDBl@Heu=;8lJ=_Ur*+DW-5aY#)IIx4x6G^u87d+OoK`Zb@F~(P&*h
zPjrD<Zb7Z-xGY%r&V0mTE9RwlFo<o(@l%7Qxu?whG3{%_3eHTi{FX173)mF|wy_`r
zNS)^FIqE+W6}48oo((um%>9a+K=Zn?av(PV_8NO~EbEKR)Cda(|Iw;g@k6g>l9CH3
zm7K2bav-2SfVMm1=j(e^W1ApAnc!}u4Lnd(A<CY;EE}*u^<>2D+gA`n71y3|sFc~{
z7iaY1x#-KqOO~8{^F}%1&pf0L0imIBp7>PDa0w@+G3m~i=5aZLGxUL1ipF`}i288I
z5P(HOP%T6yFh}G>a`s*?8~$UQ(jK~^%gx_lrdU*?DbdR>LYymB_#!)0OhQP0OVOzU
z$@5N!i4Eu4qMAeP5#ZH6#-iK<@M~!o&-T~)N00I&C`?*@(`hZ?%s5a?w^+9T{edec
zQ*rH+4{8#24$h*AXoNT)p~8k8_m7=8@mR4l8OIR+!h)0!_G#wDi-Ewar)+XK{Td(z
zpIwM1xYo5Hcw!0VYh)ZDV1bqUZE4SV!y7>Mf#GyY9GfUF!QRm+q3WLYGoz^%%Mca9
zi?QX*{*mfg6}@j6m6{;r+5^4XqQ#4^cbVM<=pX;S!)9x|?dIvSKn!sj`vYPA1Uj44
z6&e~EC+rSRKe40iqchh{2TnIgv_ZN;gEj3(|MT6EUWF5Ic_iswkGq*|0f}{Nrofd2
zN}KIVRS2tKeLL~8)H&LK>p?n4L{CHS92h-Yy<50!mkAJbOPQ25L=+O$-r^&pqtx}!
zw_d=od2iXGqsqv>EUGh}J$nfT(!u2VhdX@Ht$VC2HDyPh);AUU<K7K$_LeUYhthKf
zP?itkL+(GnoW|RgF1#QO$eHr+*-<!PrjssQQ3o#ba42rDw4IlpS$qHr*ZHVd&Bet(
z*{(*qdIXm8*ylh(pGCBA2e4$u=mKJztT`3L>xU5LRc7ezOEQcc@GN&slfokOsBb#h
za}p{Z407cs4@mqy(Nwd#ds-^i%YmcrmX?+)Ja6@5yv?ie50$>SxGU!%1?)Ykmb8Cv
zvSIw}e;hEXDWAH3a>{%%Xk`u`Y>133V9{N}K3}Yg1SI}ug1@Gc|2kh@;e3(Noc?U|
zv}^R`o8cawp8Mkiy_j@!?&ou0vNFj`jc>iMa8-Q7r9S9&ej#t|Lrg!zRXkG}eh;MX
zL8ZB#(4sZ@x0;>sz+qkL1CJJX4O@T+uUPoAflC3XvP5Z1YpYWgym7UK?wI;X9cH$7
z44hKb+;R4@IF#Y5#r3PO10VnKX=hr$-{s4Jh8u=st)EySi8R+RoDQt%@HzEcBdXr-
z&TG`-cnIRmdU&JqaK{nEobXf7g?*mVVK)N<PRxf(SOz!451%~wT%|Kg`bnX`!Ilpn
zT02l+p>1$1u66z4PX9Pu%<6M$;n93y4c#x5w%)y-go)n8d7q$CCj)-PefzeGT@~NW
z{5<;YVEGpq$P{7&=!1ImXow;B7=ooCQ((&SIHQdqwE~uV=Pq6h8=pFA7p7kJBo_s|
zbXDE(Bp!SsC;G&8qvMtP-C=Q2NjrW{Rkv9#(-9C7s<7LxS$n_ih`*!v(2#>oqvZ*l
z3~P$Azc+%zDDNi?E!ePFwYz7E-S}5+JvzA|xDsR*#B#Y&)+ZdVuADGikIblVSg3P}
zn6gE0*L?pyGXOl*PP{*|?IvWjZT1uJG&KbV3*7udTRL_mB21q2Ql5I&xfj*RKY;kO
znX)LqcHUk)7YN{8XnawQzO@O>VSCDkAq+&(`Mh+S+UcL#U)}-_bng85yC6B%M^5#_
z)^|!wY2Uv0an@KnNtR9N!^#y7w(iyiX8>+q0@PppiSH<U=(mjqsfu6ukn~hrPz;Bj
z>eC8k5<<C1@m8;&RkKO^a*D|y4vF%BLZU{Gt8mJgxRoOwI)JbUFy9IUNgOgSb(6o3
zjxU-w8^;(43+mOkuUn&cE;G|Tsj@RzL9Me2VFc*}F=h4|34$eB1S}*U>I$_7${Y~*
zA8=T|SpTxG?=A#2eR!f_AKezl&a2TI6%)275EM~fFkSz;#+@A^A|H>oS0(7M!29h=
zZa;FiL)bek+v83sCy37unjZAPiycvrZ`c0N;DTFI32wI6DnM(*ZyiugN1@D@H(H4|
zQr_Mu^uG%!0@!M0?SmN;PxEYUjgbHRiKa*~d!A~y+v1LUHWVF9G7!(7CgV1%-md_A
zlL?*V7;eb8*M@fBkR!o5sax;d`Ewjbz@kafk0H{#gHs-O%0<1N^-or1Wr$yiy0SW*
z!|`opr`IXlJ({5o-(8jhq73m^{C~0cmSI({UDxok?=30@1}I{n0xF1vgoLpz5J6Hw
zP`VM2E;|50L68Q67LaZ*XbI_90!m6tH+<t<y6@+C-uM0Wecy4se{POD_QvH}*SfCr
zoadZljyc9h^jX#jJ+Cbq;Fd*Zl>hP3gphH-ywEAVhxH(nhM=PN=yMgh(ne~Lp>eT7
z6v)>xF0KoqwhkPTstGo5NWF5vsNBt#w1{|aWU2-tB$BZMLYU|f|Ni0y1JICFI1|oD
zOOs|{>{Jqya&dJ0b&1pm{f)7!ND6IOzm@&+a%AGil!9=$V0ZX=9iWbfncwI4W49L3
zU2HGd2)^6E<>YO23hqW{EJk|Kw!yPTip+%nE_6HY8cJp~Vx(I+GHBY~U%2MKCCHx+
z$TmKUWcHfjx$h%%&x5;5#gKXdxB5MVQ&?G9uLGPRJY7O}h5K&r@hzf)f{eI7g@7D>
zfzhx%dn&~n$L(u8YN14#nW?D|@Zdo|+j8M0=UU`JV4eOfbxZ=-Zx0!x{a#u<?dKNf
zA*XT&d|}U$BKXc<&^GZ`2JUbVy+6&uJIOlFreQ@IkGAi<kO}fi$isZ@n0ukUUY5u1
zWleiUqOZ`Yz<6J#i@uGmeu?dkcE27*l3_0pL~fwbFZ4zB^5r8SetC|Lj*OhVvvh@~
z&d_;DN#3UTqV_XE-%7877VQ1bP1XRsB1>&kih%vp`$isq9bxckPB40FyH9MSy7r>n
zTL)eAQ<9QvpegBsrY((t%%xfrX@%ddT(j*q+95yCU`tVUxD+682N*cPdNbdhe6U)#
zF=1bnVjLNpVt_#cvS%S?ZLkgUao;344jwCfUc>!1*AQ?&5N4i8*8oGo#6Z!|*1@zi
zf#oY!TnSZ?2%cS%!dUjazR8s&mgqe^#mUfnNX%x0mGD|TN52ekd;0i9SgXFTH*N3i
zTmjf3(E3iOz$flam^%Z3Rw8zEa#Hb|f?js~M;Vk>+7%ssg!O3z0-L;1q2)2bO(}h~
z2hr-xnV(R$2SE^b;t0aS&5oiC__A`{b1%HS9)KeF9XKMPGRr_FC$r`lSK;04?Ce*S
zx=DqE7BCDTrT*<vvyw8tJNuR|U;Z77dgYfGX1bk6HUEV%4Hv|co<l&Ma;^mbx0UOY
z>b5;J{1DFGbs8Ug<I5scRkw>hQ&j&*DWq4TE2NKT|5vx;*&97d?-SifVHcE`cXvrj
z<`YaS6j@sHc&%;Dbz(t~jnY|J<s_Hz?SXE2qoCQ!cGvjs_L|xeFlfmr22K|Wmrem$
z#ZVb^0XC+mQ$koAgr<wH)xtSG2g$`-+NYV1T&FnB4WG5PwdFA0x_Psf83>upN7)_9
z+arKufNlG%+;)DQ39t*8X_R)@3NA$rg+%EGqRq;_cOn`_pzM<QpnFT2K;}CuDd}1~
z&AwtKShd%0*@f~>tPx`Kng_w^HGmlHZlky*yy;{Pk@UFn`uZz582(M%sXU18`;j-&
zDM<Z7XtYQqFKtLo6Z)9P)27igkYCmQHX<<n@}PMP4!B0G?qf6E-e_VH?UQsr5xHER
z6M88zSv9H&W9d7O1eBFkiY=4%f}%CPuPAg_94Lf}fRV)qxE(j}1e<Q5mne>m3)3d*
zFKtmu$;>ty@G{Lo@O&GZ36<+!BRu=4c(WHiUw<+@hSj3^RjW1zG2Zsee%ujQ^;89d
zainclTaq%CE+&@3Xk;;H#fp$ofH4TYBR|^-e@c)mv|<9c9dlOlS#I(|+Q>g7gZ^st
z0PkgA+$;bNi}Dp=X7@G1(9l;(*7xcF1pMq^3>R9mvAtSrX+iW*!>z%=!KN#&cV6D}
z0axceiUb&<pPL=?1|OGLZ9jH)KC1UDD74||7Q>!)NXM@QccG~<MJ>zvxz6uYu_^vL
ziPiv&!L|r59|5G~Q+xI~3w0dyjr|x*)3$5@ki>1LZ1-b4M9A$XIc>+s$CIfS8EsIH
z^`hOMk#*GZgA8Uy3h6c5r<PdQC~q&M|Cyk9Xxq&e0+CCYd;c?Uq$vT=@FB>urZZ!D
zgkLE?gFIMw(bB@gD=@I&%NM%g;bF=eoeh^?iy5e^N~07)Mzr_%@f~PXox*>UXS2|7
zBc#2zXCEq;05wf^0<c1y>tEQ<rQ|^9#$QE2PLAQ|(WAGt`Mmbt|A2fmdFbA~duyu*
zEeEp{mE|Wlf=Yn;rwU8g8MXhpm8agkc@v{LK}H|35_W@TsFTjV__7f!k6oFAcyuQ)
z;H7@E7gJ$J51JUztzGMN&1$skV&c+ateShM{ZFlQa>Y!<M<`Z<Pq8Yn6Reh%i7z56
zLryGvcP%qnbq+!=yC~;RpO8~9YbITJou@P?Sfd7Yk24@3UlkUKzt*nJCsP;o`&e0h
z?7tyVYe!n}3G;^X^WICJ&0Sk^8{nazoTZ}zI0R54+k`V`<u9{`QNaC}gY8G2T%j>`
z!(%$YJ3jIlZU##f&zf*unNgwtu<QG~37&HoE>!s%iNq1&J)ucqpytrE3<l`VR>u+^
z&43w?ysfPeDy|rZ)>IR<TAmTVYVG6bV)rJ2=u+kTT&h=~D)%9xPf_AASueQCV^w+Z
za9qtKN$4U9f7W_Ju_nac{uJ;P!8emDrK;}u@onfFHt!>l!-xu$(3i6G@_%8dH#b)d
z)Fa*ZoIE_6a5GN<J=03Hs_9$vz0))iK-9g?#$%6tZQpd@=qGf(L@I}O4a#GCj81s;
zp}7;*V4P+>{Ms-F;XLu&{GlfU#%BA`nhA1op#vY_9`Mxf@G9W)<JG&1qh|?))b7}l
zgv-w#_Y!^JIsrF?u18>gEU~sRns3Xe!*T7Sa_lU;8#WhrqR)JL{imp7R14_-5*Agt
z&N=|yvN)?T`LK&oG0m)VHApqCE%zgaQ1lWQIcUdMKz9~9e%L-Us`{ouBa)a`Y$&8}
z5ZY2YysRWZkH*;R>?j;w=Iy#-S-A)}d)H54;s`vK4Zh{p^ek6uP<hLRN=Ql7^}L9=
zk{H%edinNXskoW@Ozk9kjHE3RRkbvOM@jIL#?w156N{kPc6L@2ML}=ks^Vfnb7^cD
zd5a(&EiK~qvj62cXc<O2sobm9?fCkat;C9ts~Olrc<iWTI0viQv#*n_W2H1EvFm!Z
zR;^qK)3gm+ilae9OGt$3izaacPoP32ZUf@B*EN=%JA?XWe4|m7CY<mCR&pI^UDPO;
zMC(C;BJ7>~#d!c91MdoE;_wmvS@hl$$Uupt4c->{Q9>UuzC+txFJM#!d|j<nqn51n
zYpZr19H@t#h@92UFl+$8v!{_CkP%_p3@RrEgfRW;siNb4LfNJf19cWHfKOC~6VuqS
z?SDacvggot1nagn@`6+e1Z3`?X&#9fJgXfC6Q3`&OtZ&a<hlUWr-{Fvq#;aS+aDVh
zV7RXL*3X|66r+QnPn|<Wl4DZ!_DR$LAyonF@KtHSi<1(|w!dlp!D_Li7WzdkEiFxC
zLg$#Cg7VqdmAcX(pcPxB=<+o6esjVULE{qL4`1Bo`h9K3!CWeh?G~^Lm<8I-PSu_R
zj4#=;N&*AJ0p*J!>6n>NUp0+EY%O;ebhOC{mOyP96<Q`OYW$F%#SJO4{4^O%LMfCN
z9hXCpANDY)#$VEM-!4(!bq&nr#w8SBQD{13cVqxLxPdawqYoG))UwQ7YAZLQ=+Vhj
zdf$nGM8>?&{k18nT3}rPyiPp)?8OUK;rE<e&xFM=iRsU`zL5bFP>>mWe@+#^L7gsd
zLMMnh8URS5MxEW<3eZU=YV+SwFD3MBqjP*xf3K_X?vm*S3iT}jY9(e+F22YV{Cl};
z(U$uFx8ttdyt{Yi1RXzIjsK1=`j?%@u-^CAs+|XeYT2JTcn!#%@K2xx5aCy;M`Fbt
zV(a-OjBJ95A#B|?5K_*JE`kgo*(*>q&DIVaGP0$->4Ntmy&(>YwctHj{^TXosB?To
zsFh&i-70y(V>6-|IB8ZWQ||j7>tv0Pz{tjQw<55F&XwmF>tSL(p<&Hg6r7R$8MrVq
zI4E>UuBfy$4-t8@6BoYieQbp*ZymcXH^reqypFz=$C7DGXajq_fffs}hn0BMTb^kD
z9lMt!e`Hk)JD&X6?-1WFGVq&Fs0IuUMCLV12?_KJ5Z29NbfH6~TYx}5Miz2D)^4QZ
zFTyN@b&Z7RjSl=DT30LuZp<!7uU_Fj5NVvxcXkK7;yQqyetu(Bku;*Xh4FRhEQsTu
z<X0snKHbK>R7OUoX%gvu0?#e2FO913^}bvip9Y@Qz`Jh!4nwXb=>|MF6x80H+kiA%
zuQi%6@1>WQ*2vH%s_V#D?xkd-0*FBR$g5tx%0$pH-e%l&GQX?eIy*gm-tr`R{WQkL
zB;kB#MSiL*PG;<M96elixqAV}=w>iBysyF4hs=mecQ<ii+zY6qOh3a;(i%VUu}<CH
zd=c-dNp&@2E=lRYBw*Eb*?kv`GO^cQp{28O8H4){!Mo^Dlq{rxX25`+K=Qfy(%Ls?
z=^GF5!_#67$x)yl^Nz=dok|V|An#yJv7kIC07Y}3JC_4)Lqg$GcJ8fwK%Y1=2k`~?
z)A%zGu0cyYr0kASy{Va*-b0Yo=LG#Ld?u9lTsKtVtItDH{_>mIBGd+Z$x7bPG=6(P
z<D<8z8xZj~=AN|(SBGsky+2$Fd>u5r7b-d~DGoTv4_PL9fSLr=iv5{`kj#~Km1s#b
zT?U@X_N%GJPZ2AB;6y(v6nPfJfl=G9Lh+>fRRQP6EHKpg{ueG%yvASuVs@YZO$;rm
zVl(xo5f4@wiaG8Q%fB0FF?U-;{UZ*gXClTmdD@cCS=9d#kodfw1Hy*aDAM*wAR>xO
zqA-XS8*>Sez|$}j^eFiqfLqAFOvEu7=#Yy;GwNu98i;HRU<*BK1|)Kl3>x`khWB+F
zn9~V8Fx;RIoO=Z5PH7RXp?_h~1sb_Js3J0aJ?{jDbpBXZn$=JcL5eVDDd#16`0&N$
zt5zBG$$?(>+}nG<Z(UQXg_4qwl!$RlR-})Krsi8xL41tIRL@@D9I(1q>LFo`Q+l{y
zRfnO?{4t<ALOlR&Igu!CPOWSWM@m*4JA{!jcuPf2gx-?uK7(P4Un3(PbAeYu3Fx&i
z4-z>zYqxdd#@A^BAW%PD(ALg{!Ne~AOJ-&XY-sehO(0o*jqXMpJnd?x*S@gT<7GnN
z*{vLeYP-V{JYj`zt!TiMcnd=N@!lB5Ghbt<;*6L-3iBs`D_nqE$)E!xgDsONS~xa#
ztC2Kv!_rK`W>Cs#eqjQyuq}m<WgNbdQ8Rw1lyq%t&OzluL@0^p*~vGRY+*)~uOAp*
zAtN2lgBa<M{?-a8rr6jHwdTDso$`PU#>U3Z78a(a8=&A41x?vI5sGm^1i=Hf{aBxQ
zDF$*)J*f3apKB*laCMvWE4pzaBv@_6n2}1hZS8lf3_t(`k2#Zi5yU{um%aLELPg{2
zsquC)OSSOVx^)VM?$^M@K!1kG-VKq(tGNq!kdPjTUFu1yjfSSNEI31@Y&E*`VI9R(
z6N%#085D%-GyW6k<jP1I#G*Ckj2q_+j7q)-^rM*S!PWHiFE9F!+CdB;tyjna1g@rm
zrAwE>W-v6gr)IB!fKtkUpAka(%GMLo($6%Tuy41&JuDq^n}AV<3#iU%5DJx0P59@*
zaJ#kO)C@xtOkKT7qo3)F0b6c|(5-W3^QiMInp}XF%LJxI0+`{LwVTH9E@rqsr7z+j
z1;<+6)DDUggl-E7J?21zi;;@h3R+*xyUZ7HceGM@ET0-hA%!gQPIj2JY#cym!Bv$-
zGuQRnBzVx3x;ydpwxIhJ)ien6z(vh3U={oITLRjbd#8g=<G$3<U%#G7g7ISi#^SKj
z`w&c7XXrM(e?<ll$+|NHce@)W$*D#&oEUG@Vj>Bhl=NgKV=!E>hnH6lWy#<1fn+QV
zRh-$ia6o1y-sF0LZ2U9e)mr#J2E3~N8Ss)13g5wpd64ypq+9Ca)D#03AV*snSr1~&
zzUt0#rT2&3-hiIl)6oj1*M_ZI!_!y<|2}(`NDCJcC>4__2up{jg|+mzOR=Z;VcaqE
z8U`mTr>;&&^axdx0hND271ioZ@KbkpcNiC?5Gqr|)#RwQ1$GCw>BCz}`%Zi*yxSSO
z7LQ48QmyHvs-DZyEr*^0k9{ZTxC<D`c%Mm3dy!xJDnmv(%ik=j)bXi1b6e<AI?=#$
zTk$2I^WywrnZ06pcSkQ|3RCfA8FJ~-WW^b}$ILLx_d@62GE(5%yacBnf;@DT*PvIf
z49rINa5G8BPZahdO?;;mjON=tw@^?8w=EX|WWD!{+YWYQKo8Em*#I=}^1RJJbEe{f
zMNx;D$6!tG4Su1Qz?Fp`qKcf4T~35NPGsOnH=X)*qY^5MfkMz#KgVxk;8YI;xB5JW
z3s)4M$N^uUSdY9Gzq54d?Q$zhM5b>P#8ZV@?=9TEhy2Y2M3ue{W#FZzU8Riq4%wKQ
z{Adczc%xweWzN`djBhVVNNfT!vk};T^~zAMyV^50t&l^B7fnZYLtoq)3A3wn7CY1b
zdI&t^B<5~D{v6sLvLkO(o0g6%2yRhXuGtWSeXZnuSorIuK%srxllZ$k*Bq0*mUi=_
z?l0ll(3)X?@P}!N4fe5!qVjWU$y)2d>q6RX7_U*#(Cs4&9$)E9a6)pH8IRZclW;O^
z3|Ur){+1+MpXS6S21QCsGNdR|R>msJ&#YXo`QgOr=Nn8yR9MAR#l_jZ+V&)*`@fio
znEt8ZpW)x3t(P*OAF1b&-jkA^q!k<HZxp88GO&juOlc3t9Tmw?8P1`FkIT3NczARk
zn%^j&ojzVMTbT8-bykF@J799s(J6OGm@Dho$$|y!0;!zw7?XVfRE;bpK{_2tj&nHN
zJvu0~+Xz)%F;rn241>|Lttf^D{7aNR2c9%o9{x7%Q@8Rf4#x(=sscC)JSQe6;SJ8F
zF>viH^k=R{(9%Q@V7!+G?#Qj9V34xz@j!KQY<rfj0cCzPc+o!5oVd{yBlR$wVeZhM
zwa~!;PciVO;5mri7d0vB>FXoYyyC%4v7#EKCO25_UiH<J^*jbjnh$wvxQp=jE&qyB
z-nNjUEH5vgIC(lYUykbDgOQw(P%a1dcHtq&z(}IW)zI&=Y@I+zndkyko1b&D@qk&`
zac9-eBB5sssFV)mPUUlX&~c{iVuZW=#>$b~jj{7_DKon=HOP$ry3^>e3<#zs9#EOZ
z$LB1J$8~o$Dap!u5}(z>LE!;$^MN-gBIi9i;O1J|lbmAFie7P32Ml;iz6@C)m5JGs
zw#A(G3Ml1Ux3uusUFMeeU@a!LJ8#?*ge7uj;-`0t4;`8x91!}lyv|KpvNk~F<_)ut
z=Mx6(a-yO*d?8qB21B8JsDmmpTJaZOaPOYNCtM(|oED4aL3Er2vk{Sz?Co||;P@8j
zF7)MUm#r|u%$N<7r#+Bs9{`emDJm-J>OkC9o~?6bCXPIPlX4mw8ktQ!y0q&du2<E3
zX)&{<W@Rf(5MUb}1_n7hr6oHZf5m+Xx+#{HNVLxTh9Q1c+%u)#Bk3TM{$0}SWBS34
zxjW?nbS-=f<pNG2KIz|C+2f{{b=nSab<N#L>b&bB=oAuQ{xSbh*aeS~iWOGgj+2(o
zkx*-WsUEe%#M!d0a>>|UH-SCr11VHLX7EjC_k(>TLCgx?-BW~ci;1bUwNC5&pcRi>
zOl0JFH-WatnvnP|1`)*1XtvvFp}V6$v7&r>G8uEfuG{Nz@Eky~aRCDG-})m@4A^+g
zmS4>nW`Yx+S;zCU`_T}+fcd6C*t8031V4QEP$4+r%FkR2Sy@>#oToZ@wcze)b`(7?
zMh?p3v4DcG^qgbIWl0Nc^nF0W3@yJ+gbzV6?0Da%vpLgcCS7B;PUF#P`ygjiAwZrP
zqo}>LSt)lRD}eips4j!X=QF|?WSg;(uCz+v&@Jy7eZiXfa?M2#)^yl7nqN>!^1{Cu
zwSEMJR`9?n%WE7Q93I3asce*ite+GS?qgKXg{DmVj=d~;<LnOuwP>r#m8ZO@hcA`y
zS0;*nP1H!4<95vahD!dP?aqOxFL@Z{w$}w6@%vJRE<*=2rYm}SYyk70DcF>i>J-~#
zJ8~=DB1}Ds^}Gi$ygR)A_Nq18p6#&dcwUTr!4o&<ou`DpzW&w2&c$61wyG8hvG<Ur
zhxw>yXn>m)vZg*ZI$uGw)5|7|jBrUuS9M^ac@z{BXu?yVY(|0@Y5FrOzB6!0Ic;KO
zWONjbWKA&I%W_^C=<9m~&-d#T17omyXMpGcXMrz|<B;kUqZ#U#y?_F(o56&?F;>{i
z%c~f`pC@KMg?t4})y-fXr0vkw2CPzffR4fT_urk_`3;iM9&g-)I0RvW#WSO>c+EmS
zai~<qC6ZX(?nA{e<ra+MBjJsjpOJV}N(#RX;+DB>Hu2y}tSFh#ii4fO@cmqz5eEPO
zIn@n)6?x^ufj$0-j;nNrxh80`JIHL{k_YQFWY+`6Cj<McuwXh?iA|k|?q=&YaMEEE
zZv`T76z~qenmOlWHg>Mb1as?iE$|GCME4~$Ec*$7+|dk{d_#48&JFxe`cY{c^W~^-
zmEBjI5<W)1?MTSyxIV}=8@8wycT+&swKA#q-TyuXo2IMdZtCIb@)ZZpPYg|?Y#;h5
zOH98Eme`ZgBB+^1<XZ~cai%E^y#WNY4^?6D45Nvz?m2MWJt5e&TM*WOmJGKBvijNc
z<&bq6#pe=NP|<En%W$7w^eD<BEBwAVXs37G#Z;Xyy0R&s;|#hhPD?C$ah|X8rIXq?
zc`xShKrt@ii(uWvT>UHcpN5DB*VDhdar_Pk<y)L|{KcgERc@x2v9Q$}IK_9-zvbSl
z@(!UPfR57=LFa`Dr@(okOSyK!`G9UaCAj5<jX6`PflJ0%8z<iSs00b5qIgjV9?XjV
z!kPK5j31U5p7^Ij=RMwDTlVvO!r_lEjH)_fB_!^BFwz?+h)&kRiZc@JiQeYa2N8ck
zvJ0iSo%%IcB>0WL`EdX8G2pJnBW=}Rsy|zW0-8VgWJWYH(EZ2|3?^HDK=fK-C&aS0
zP!D%eW`NFnpw*`RbTqzj=a<N*%@-d6s^V+cOKkL3@pyIKgVkxSoy8N*5UI06W*?TU
zVkNd*Qr&D56k88NyHDRqOdXXKNaeyARSZ{;hF&}4%yQn_IlvkMf<*oMWO>P2d#je~
zI55+!ps0AjY0}E+Wp%C;*WVk-p;Di_VBvBp#>A)xBf}$5?N7Mo$*pXgF^skLF@VbE
zU9KSM#Y$~~u6y2rrg`Qgc-H)@5|l?ggXdV=pENgm(>qOlE+%$wq=nKrlTx0Ep@9I8
zCRh&}B(^(S#I(Yctw8ol;^E}vWZ_tr0Y-5w>~xrDG#Tb5pZK1VWZA5z7a0+8PINr-
z(mo<LA8oYNmIS2iRR#tIRM+&VDE&f<q`moa#2U^63{nxnw#R#$j@QHS_G+N3ciAa=
zkr&mwTFNCZUc5Mx>NJ~rc*4d-<aSHd>0zwi1Ng0Aa(wtTJeb7=tju+KlRBE9=N3cw
z9v4B5*x#WpcX2kCYCn(;b4{p%6~j2Wx-DB=LIRnkCy`PQFW#TOQ@nK$OWCo&=|HI#
zBkYYv^0^$j`!;!7$ve}coEW?wiq9Vitck6cS5pffo!5yS5xm{vJJU?HNcrpCz$>39
z!SwHS6_u4LH$9T(L|U7wzJ7g1+{a^5sEs(r>{<Bu`1TQVIw_~Vz?Y~44y1lFlW-Z>
z!m~Ai)pv3F#_Wh!oc+FTerzTZHDHrzG&<?sN83-wV7yJt0R)>j4P^k|;~>$WI_G{s
zuY5%&i($?!)~$1h(mF=bBrstp*~~b6n~;W@9OqPA$6a&9qUmLUf?ER)FO>5;g*2pp
zHwzGxj;&u%FJ(DrfZeoOjCR(BvXQg4bcI#eKWD9?mh>%h&BtL%Jq-7R<KO_a7%Jni
zfvSxNpwgVQo-qH{IJ&wmp;TT)3|fip3^8)jrlMK55vbAAvpa2AIq|3K@*Ut(Q-Dy+
zgvjTH^4cQeX&uquH%{;L762zh9S=}OS()0FgO3=lba&)+#9k@HQeZ$nTw-^8fDdw%
zG>LueRGS?)KP+LS;y7q~>nY`o^~yuCzhRInP%xz-F=%rW9wdx}GssgD2jDwm72r72
zGhE73kq3)~#@LiqYnf>N!o(+rW+siZB5@F;^;C$Y8KQ5;hEXvl;@N=+`08^UdvH{A
zbUVo#GT=%#-|DjQ6y93JG=H;=feEI1O)Equ^Guv(sl?c(ov7ZIl=Q3pKx>X05P}jW
zL21SXM4sQov_lzcFKu?Mv#P{(+7wAw0X#vk&KZIYeGp=ZOvF<vYoz3$Df|RF$AXlh
zyXitDLj3CrRsV$>;#^_h#baYIgzIrrRJGYh?4yDBXxG%%YBk4&h3zkJXJH}UA0V?G
z&lwFIN}6F-(Z}P^oa<H0bp}>^ocNt1GunGKOBKWFM?@#$-SfegqB$j*IP@uVYZg5V
zsHm;7D!{U;f7PI8{59mnFx=aP$&Dc`G6<h)s#K>4W39Ji4Z7o~6mUc+$$Rx^XVWWl
z;!{dqO7d=6>NJb*NOQg>&kIIPm4g$Ln1x|(t%X)ezqIRy2bpzPWg>%+sU<EZ?eLr^
z)lEf7X^PGgaXJ^apP#CFnZp1&`4hAe&mt5&YSJx3snNc@b?S_L+os^hD6X}}?5yg(
zeqB#&k%^Nhan?x5Fo`Gr@Wj$6pBSG3rzmb{fxg*PO}OZ0V$4j8Vlpja5yA^lGHg+U
zt!$#BHWL$58`ON<up2muzlvqD2)rSz!DA=B7~H52hyhC7d~a`W8<b<*4dSx-A(pX$
zB%m57Q<!Z<rukvHKpc(=W&qjcL00I>i)|wg17a{vPzPv1o^t-(6?9j#y8spMA&wE&
z@S|!+-SHTJlmJ;U+_Cd9<BX_In|?IV5p0{7wqzv%8nw$mmNWcx3gyDKm)9(>piE?k
zGKilAL9y}y54HIj#6Cf{Bd=R@GJQ)N)$T69iUL6B?Zg7AG*~P*-;Vm&XdaoNz+Tw*
z#*zt(w%~=HV1khw1rbxuFBuE4H8+kQ3Zy3~4zZUW{(g_1*p)P@OTEQYBPP*##NrbN
z2umB(iaC`eGbM8U&Gv8omOwm87KT|Cxrtr8!cZ`R$xfQbID`_1ktO9}Wvt$K$&P_%
z>kgT;Ng3oN?XV}ZC99WsSQ8VRx&RFvi@f~&1(*+GCS_(j3P5)_%s8Z~%8Dl6%4SpG
z6xh>7>ctk(gPneK5d3o^$3LVd7;&47h=B?oTkyNtx|*6iq^s@tXjd#xA!7VS<hxvQ
z9}QFm&{>KprK$XaQ+?&D<Gag>SENdErfCH_EQoWPf9#@9irvvo|DrEBw`O!zi6AaG
zXLObGb2{T52U%rh|N7j@*s2$FbwevDAq*TItmm7WdWg>l@SUWL3Sy9oGqGIxIy234
z0x82Db?VdRRieM{C`>xR%S0W_o(&t&1+EZ4K58yhfj($EvRvx$D6Uuo=YL5CQN<!2
zZf=j^CoHfBu2-$RRB)2RtT7FsjzMHZ((P;4t`%&4o2-S}Xt+>pfg9-Sdg4qCD|Rl8
zlQk&L(lpvGapuMa?mNFRE^FihndJtzsh|O(gl<NKJIp0rJmZ@BFeKxKIi7rCY}IVl
z`V^?yX^^j|fpdonT5T1f8}E#pdl;%B3^-*?YU=BCBn_>x<Jyq^Hw>Fg<YuBxSAY{^
z`v&UDhk#IfwW)vuw&9=j-@y|M*6A#`SeuJVx#S9<U4(|213=n+D2F#eYfP{H!(4fA
z#pB0lfH2%TIf|<N&8t_tk)kyiI^Zy;g)C%R$+mOkQ!X7JK3!*o^n{qf1+-0dayc<#
zc-Po;mc*sD+osAW%L1pWgBKWxLldNO=Ff*N{KEzK8*iu)>@e{Jc$P&(HvEC!H|p4A
zBJDbet`D`%G&9HD$`Z-jX($h7wt-b?4ozGRVxB_$a^7Ie#6$5R4-5h7jzbmNKsJVv
zh#l3KNBP-il{<T_Jh(?Ci@2#Es%SSOPCc_QD~l!9M0{P8n{|+s%3QnFJUHdSN=#An
z@qLLimNV+qv%#Vc>d^K!8s&Lj26>upYm@|u#?f(c?no?Bt^2`%WO16Q4o)!ME@FEt
z(=#6JOjl6U3jw8RU@6EbDsBa}pzl}r?mc8kv9h7TCw>7HnlsQSi=XpK=ZIg~am+6q
zHdX@}+WrTf8k?#(GnsbG$5PZl<CoZ#K#xWZenBC-o>Ze-k%!LU4m5wXYwGF>AOm`I
z<_6*-G2^i;tU!*br+5hPksDbjEifR{uD^$JZ`+aUAKZc^EMn9E!VypiN4Gl87#XBl
z!!3?T7%cN}TXR84)GQysBnvw+z9P^qFCSn2gt9yCM<J>{OJPhb;oL63=CT}PB;HRL
zmvIK(muk0XmShl^panYuzbHgR^{F-m_h6_Uj0pF~Mw=APnRiKfu+psa(A|HXks;LE
z*QWut9;4H}z+4{XnP!pc4o~7qg`o5EtXRo5wkuel#6+pGriOtW@MsBSZ9d6*ej_QA
z$u<&>ZN*JCP|S6xUr<zbcxJELqiTiKOp=5oo4Fvz0r!RpL8sBOV;Xp!%#|yQIE|hF
z^Rm+Yo`?jNm}V|cxi1zeiNQQ-s$y}jVqJ2h2ks|7BEP2v75$AMvT8_ZBStYURH}Jv
zW;D_jF|<^Y+eNdG4&p>29l?Q&Ro{%kj2Onih)%O*(e6iN|0*A!fU)5dNZ@g4IB02U
zVY(>2Ii7g*;1f1&$AHBj2ObS9w}G#D-7i&b*@<n9>p1e+;TqXX`8OUh*A4Pp6=Ho_
z+3&|@+klQ>neDZT@Z7mhE@Bp>)eOpQoIr2H8#Ddz#PQ5j*lAsz8w(dRGMzmjI{aj6
zA~p9~-s6#OL}erMPrW$eskK~(mZyKu<#?%b695aCI{Y9NZ8garSAbaCVe*||oNiLj
z7m?U~S7KbWX%NS0<8!eK9<1cg(AG0hTo;2(_wi%&%XG3iMG!^>o1F%kM#};41XPdf
zl4&-S&naoyB9p$E9Q7*j2M%5s3n#`}<OTq~u<W=iJkq>p>$iZ}kKZ6ZvG3s;MvvhS
z_wf<dn2plFB@Er9K9J`8Q^Xe4(=aW!pL3(DE<C4``aE~XEPJBcZsE4dFSSm)1#y(d
zds8;c=p2m5`0;Ri)3vu5nL_5AY2=;=x4ZJRk#gR27USS7C|VYk^B%+h>16^(SktCt
zzF&W(>mFphU>Q97t-CEMI<`c~3GiM%cr6m^H-sL+3Q0py_m7E=h<Hq5f+>z?ul;O7
z4`6tCfVdvIj+62Yqv#>aHzdcZI7e2MBnLHnNpNp&B<7tv0mN$e0Y4%EA!gmbWK}U`
zybXa}KLztCj>9NOEUZVCb`Ct;Zv6NXq6=eR?ym=2d5~thuwJISwh^_E7obQ(Aqno<
zpmUr)_P<1XePAkffE@R(<KcmpSHHo`a^K^}m!Hvvy@9<pOCFe*JaxgyyyG$BxAAmM
zhS{J?*d3K4)88FxxVz}r@7^jon>=<TB6r$9_X)WtU#R!!szb;PlZEq?bk0QS+03OK
zAI0WAi)lJUxoVo(Vgr)20B>ticV%UTUM-Ob@>Vu3R%&f%*al#b%Tb)dP`VWuuB*j)
zk37D571_JGY!7sTZ1!vXI=-h3Bj<e9QZA&rakW4;Ypr5dyl>y*u+gfamAljs0XC3R
zv0!F#CT_9Bx{a^%Jf~0du*u?t31c5AdC1L|SihF?Kz)H?_VL!~6qTD$ol?gUusX8o
zeC~G2ZIP$y93h6|1Jq-jIa6(}<2We{Niv07qdp6ibrZs-UP>%Zw;U9OFWB}7xfsNz
zYZq>sx16YrbL8VVXc0^)DJJU#IlO(r%#eMzF#NGiUhLZAJfI^sd!-;lR!VRmv8TJM
z`*RHtUs?^5XaRkr;lLsL16bS_805R(wyqTyo&S<MCsB}r6EI&fccymZ=~MB=E2?Fn
z-+!Hy2_;I4z7_WSvGgzBeN16v7+r;@w>Kf%BnR8OaLKA2(sy-*xaN&1l!?>Y`zTri
zu$_MlMG~!di=VdUtUvETE{a|Ga$lIgamh>`cf%k~v67kP89+dYp{Wvi*CaGbg-}}g
z$IO0FTi8enpC@}>`&bU})YaDJBS@>g#`g$<vnt675-kY)M@TRGoFmN1V`&AlxSp)A
zf*ELkauez8ji*m@lkb5Vr9|l%W~nX}u40n~oaC=^WFQ&m94S`)VrD<A5?v2pS*a&`
zdb7qdmrxou$){%1iFelm`b-0?OkBu|8zyk_(cD1SZ_ZCJI65{2@43W<?!Ouuk-X77
zNegC9trL8^ypkfl=IPU?!zcm@G0$3Sn4P0OjsC1G0RzhFt~rd?#^YVR@lfww8{%Qw
zzCCf>@@gU6ztulT^#WeT5smCcYo&#byFDcOk_~orW)vq3@;T+#5fnDo(N$FvoIY+1
z7X8HH&lCqpW})E$+9Qy2j=Y@!sl~65I?#>$9I~7<vp*pXLZnu6_BAqRy^b&mxxj7m
z01D%@%a4|bKZoRVRLySbrQ~#6oDN=W95zRSjq)M#{)S4%0J9QyQy|j)Q#Lqqi7VyC
zGc|%eR?DRSg$g~Ei+$f#$+L3u^3&y%bGva_*a~jslS!jP*l`dg(Ha)ftos`*q-L5a
zl>SHt*(K5`s7lp(A;Hve4;f_{W#<4tAkS^w>Zc<=Qz+LDn0=&BlFpFthdJr_$)j+Q
z)Nqbm-A|m4iNoriJ+Mx_(x&_K9dVQ;zKN$#6Z7Z1N(yDV>uSZI5bm_U2;<P#Q}l;#
z@KPu}tXm~Ttyz|c+j!FLXUTFWrdO81>QPJ4EBJhYd8yWi+%2nakC4BL+~(y0j-`|r
zKC~FLoYHxb7RZQ`T!VvGD9r}^w`mssH)I>W&%gf-{}37C^V^H<&o9rdAQ<I;{fKk0
zg^{=V_s``M+`ldI{`}_IYSQEQub<m;(!^lzzkc4{Ri_#3{_~q%oBwkONZS7|+=Y`C
zIZRg}MGPMv9<{P1l&|Y(FekMhi4zIv&tEU0+#Z?0G2$(B-{WeIy-nmP3gui(%FJuz
zRnOoXfBSeN(pd7_5b+TONj{YF_IxWS;z8FCzDz;ZeJQjCf3_yzR@M2?b$GWe#KF>)
z1zBq7x@BYqTz6Nv()-U+`2GLR|KC?slzC$$YI)*2e6kSB_3ug4C<Une{9^i$n*o1}
zI=um3xDhbMyPDm%8dU8D_HA855&uY(g@;kH3u!o@E`r+n`{j6{cp)D@zqb%95QQT4
z0EMqP0b9mMRIdjB*HsxdqglEYR?ciPr)giIQxAmUc+dxMf;83VY}u-E^Tc0!XxpYL
zBBADgY&Z(Tz5nb7n5h0IDq!jS=Pr=_^Pju$|I=NNdntr>DoQ2o3(|(@P8x!9v1dHV
zU!BooXA5t4KR#0A13&J}Hd;7lJ3!jTK(k`Y<_QJlT^};NK_&D#q`HNWW2yrP9&m>9
z6C$_HW^@8W*CDKp^Z+7*-_aR{P8HG<A}xx)1-Ft{eF*xAUWF?G;#Y`WBHaRlSM=&v
zJf$TCLb0@PD4xS?|LY^cx$XuNg;-MnNL%NQlQ*E)CCFY9NoMQz6|qY>Ny{9rr?Mjl
z2|2V`<JUsorc17{*3JChO^d1xo)?3T3&Q+a<<}@fv@Xr@n9YP_By18WoY2eoWNC#-
zFtx7}wATyW1A*B`*vj}sMJ0fCmZ=Y`R@>52q+QG;Es^#nAPPh!Rv6?0i9qQj^Zw7@
z^gOv(AAvUgS+q}1RM6h-EiIInDC}M}q*La%86(nMO0%>t5m+p9mU%O>YP`?a{Hx@5
z;<hA{3Nb7T5)nqxb@^G!+HbUHz<Ughmb*z*)Ow_LJJAHcXbPq3EXlL9qR~a*hIx2a
zDmB+>aq9ajF`r4x-(MvWY+;7C&DB!EF}nt6hfU90By+bD6do_`;={B;L~Lxi9Wj%p
zPy!o)AZ<I#Y$4Y|#jw6{3l(P@J_MySK4bu}H+YY>9m!=U8FXpy{^Fi}_a(4tSyt_|
zzOMS`)fZ{6j={{v@DF+?0h58Xj7M3K8waOOkat6YhfXjPl5ra&+LGB7OuMHOBZ3W%
zO?u(Daeuco{aJSef6}XZo0TLF6p3ZFp>U*X3we(ZG|mj09Wf@yBt=)~;Y%lIOGo!w
z&PW8y%-2V+b8<wvjA&PB7*6}bin6ksz=UiXuq0#}fw8ZrM%Qm3ZzewNCt_y_NLDMA
zI(+lD0rYxuP0v<OZ@YEuChaP}ra&zVZplr?$)sEm=zJkN!a)1$AJoXg+qKzEWTIgk
zdaRG!;{a6vMeO9>L*D-ONOC59xX_RLn_eUPuyLk6xbJblKL9)q(m`9t;)H4>4_&&^
zZGWx_rstldMZqlUXyp=5-iUJbHTfbn4S^r6mQB)rlTJlLrfo0F^tnfF0{}uv6QBaA
z34RRU&|bWaM_ZEAD$Rws<N3fP=>GZAy+2PyNeOno_=&c2&>Py{X*ffkUF1=KE|x4j
zn8X8$DZT^=AAZ=>Knrle%}$w3bK7!*_PdUgwAWWXKVkjnf(FQ4W|{wamTDo1eXWjB
zzZK5J5t|}TUdqlxN+Mvl<+E16yQE-Awe$>akxAz?&AyL9WB)7)usgp$lGF&|tD&yS
z)$M?r#7G*QOMYJ|n;IynZU8;IvlcPfO}A|=`Hb5nX}bI~sandVq~u%1rvV53-&KVZ
zSwHC6mg%*E!~@~UlXv$1&b*{(-#}JPjRm8DhUU3VO-&d;KeOR?V#cPHE}EpjFgFcm
zzsz@C+GivYghrgB^Qg5LX^-LVZTRO7k*$LsPD;93?B9yy#gxobwD*U<R{$}CkDlF@
zwu=0|$Q^9$q>SwNc<yzuE8FuoZNb0wrG0_OKihdrDx@96M5jK|Z~6C9!bGe>YOrOK
z-rvr&`*tp~_wuUunGh9c2Cs(8VjG2W>*QG$hg;tOMPEr}w(E~allu~bAl70@BS29`
z5`ZkW{Y@M}3K2x*dyk+7{OpR>92kh+7Q~RD9j{|IL)zM3Jb_6+sMuYP{M>Mu^d7+2
zOs~0t<G28b|7)kk1*1X5=d9;1-?v`<Pd6Ze{5gJ-^ockUnuC`g$uN+92rn=13zGUH
z`QJ4APYO)l810N<@DE8{aJz=I-qoSqjyJw?Ux~c>-#-^<PwU_RiVgaovXb(jyFgnI
z|1Z`;lms*WYEtOy*L%Q9-1&&vQi<?-2_{B1D{w%kCxDO(NN0%DZXgR`JX${TGwrDP
znFN7B22d-VA;Xr3zXPwHvBzOyVGSS=k=7q+U(nisO7VLDRbK7ANeh}yv;&e6DMY($
z@Qhsatcp#i7n&_{@3#S^Zac*?BTc|Ef-XbWTv4hJ3T>i<YLGiWVA77>h#N?4Y^Y6>
zUFX5a-iWwWYSmzr0j`AxIOAn?7eQ(xY>EK4Tr|AZF@iH7I)q^{!X_!vYXt$e5QDK)
z2U<oGGS3Nq6ZyI>RMI&r1kY$S(`Qf+HLwREYV?ileyKvzl`^jo*;;XYR1(?Y-eI6n
z`J^9r-h-Dq^kEsjD@io=^eZ$23m>kM;Nnw$DKTi53A9-_$hz%`WpHY2EN3~$5djnP
zjTMetNEd~64YyA=Y#O}O$b(+J7#>4Q+XTWF?;!HGJ!h6bb0RyuNE-1OfTzAiFaXtz
z>-ln+G<6|g1n6@8phh2hBS^<csvv-Tbk81j%rlxo_~IDV1SimNdW8cg%a4>VfINMZ
zeWGGoCoR!4bwgh-tGPQHH?IBuCaEGlSgd9%l6Xy_Fcj~5sj?9dRM&LY1TE;h7Zo7g
zsHkFrArJJi3GpV=Z2-L+*wP4rKFdsaAg51%yZL$TaL*;Dj7CS25|vL^*xQ2Zkhn#*
zlj#eFL^N<Uco=AIn2VkTahint>+Uiov$8_5IO|P+JlL9&HV?M8D;gWo*NmV_D<%b&
z(w0;;^D;hQpi-B}_1V2ZCp#Yj{^h9C>yp1g6lrPVNKPylI%sf0!C>9-UmwUBw&cyy
zDW{3S<54AM1MO4vJ2ouW*>Up|Yh}QvB|HNHq25=!xQka_O#HZeKxtk0(K6Y{$k2#L
z0j)S@zVHCPi{=GOFUsurkazR?zZg!)J)2)JRp5LsvSd5|``qB<se7{-kLai$vxDbV
zCTT|(Wwk$s&3ZszlEJev;@6SM*g}N}1VpVkjvxQl%Uxbx&IAN?r#5Z#h-+i)NShuL
zBiB!)xI|JuY!ILdlkM@{giN_YOz0wAq_ZE!8zYc3H8SzoPyI3s=AfnSI6i)I$h^Lv
zV$y0{eKP8`?$oazjlHc99InL&nt0H58SX0&NwFv)AIPv+2{@-iysNWNjwm(Pmz7-v
zDW^13lRRqtFbLtAJ%fXTv#Z&DVQ2BBXJ#rliwV~o|9P*5OQv=Q$Scoz#Kpx0*$oON
zI-o%t4#NygU|ql@eumv-EVS4=e*Aa^M5&~{UJ-(v(=X!ZlML#m(1WN2pAQTZq-*=>
z>D}#%R+D!_R&?WgN5@;p8eZ+kRug=1|Nc+x%?ohEb#ZZlh@t$OGiz}Rqd=oKv*c{*
zPR&S9??JHyG%oSb1T?!|!N1vu{>q>0EILd(5T$nky{k5~hlH~&l>8W&nT=9{F~ZTr
zm=O%z`L`}Ws!3dUknIJyXv(3f>N-J9Hh2lSP9u3^#I&Qz&p@u(|KQ=nt9*)3rIMdN
z1pa6mx+UuJV!tn)GNp{E0OGylIgBc*k7!lhy!pP-1_Lm)kl2EKmMU_MdJI7tC#gY1
z5t3q>^2QjP$s%S6M;&1fg~oPyR63?^-u3Db20O3agdyXam5KA#9JFmnp-5ZPkmB1*
zKeo2EYW5=(_g06y%jaLhNDLW-8~aQzMiziR7!m;Xo`<$#z6$r8GumuB#CIFjqY%OM
zp%0``rgab^3<=;lL_=iL(}9d+s;RZwm;Uo8PUs9?eaDLSl_IE@nfW%D&Pu`<HzxdK
ziF6|jpE^4`y+UbUMJhm7-WU}Zx3Z)K9L%D-6THAXhz5PQ3G(yBqEMc(h0EC}H?5pt
zV}jtqKB%PG*w{ouI?ma}r5bG_3%Fun&b_0j2}B6|Cq@n>H3+y;_->eTt3i4T34rY5
z*S34x*xE*c@ia0vCI_(-IpDJpfxLhSsECO}7@53y_ACU-K7XG%(>2|<FBem$WC=o$
zT^g1<FP=Yt1#`dryu6=?>+r8FeR-Uy5T+PW$(s_nc!>2U!`m?sw$W;Jt45a5@O`%R
z&<43!KrDu$gZGRdkq8gpqQJ??$pq=~9LRBbIuO2teJL%!Cu{lo;8jETYsf#4D}b@c
z3Alz9>0MpX1Q4Kfi|krBL4mY1?L#YY7>;8eCJ86QU!KFr`dh_#rvMjx#r4%<l3!3z
zP^*WSt>B4WM$uM<a`aH_KzsWeGT85G0#!ha$xV(zEhmY1v*BdM>oIRew0c0Al85AP
ze)4u+-H_Y2DW1bvX1y5E)$X>&nQ&3`?-@L+8HgS6B!jth26fJJR`{p;24-T)U7LQt
zAgmi|S6%K7#Ry{`Hrq>#qCqpIV{p)TyaD6011;HP=~qKm!^_)SJ4p>Q6k}jpLBI_J
z5+R3~q}sk5a1Zv=|0MGLckkYP=NAsO2PW(wY}W_`O5c9ialgWh=V7%R6%0`%^)9Yu
zw_9|Aex{<D{qO_ODc-c@^StyUu6txwT>oOok>6dI6?fB*eZVwgMWN*CATX(Vp+Hbp
zB?{M_G9T{GKZ`lgeX32Y5H+}!gISFT%!2Y)Z2G-r#XpiI4t3!ug_tO;)lQ8wD;`A}
z#ABN(f>px6$mqMTtH1vpG&7h2-3@3}=Z@mmzb}N%ffX8zG4?SKrXaASs@)AMKM#;O
z$pY#DQBDTWy>S8B95a9kBnu#LvU$svFmUmDiT9zfBJw9bXy&e1zB~f+3|r`4Veb68
z<p7ptv~#4SNTeXK6XTW#lcTJB`WUv#G^`d}F=m!5&LB%TD%y08rKT1Z@-Rn)lnYnj
zahuDC;aO@s#zIi|7+{2g14iVPHpCbK37Jx|Hog(Bd6%Ro(EwBqATkiESFgT|m>6@l
zy0P&Jlu@OTQ}E)dvcc`T3LS%d)4I9E4NOc?;9Kc|Y1YW0xokC$&Y#e~DExOm1GKde
z%`ZZ}zNpZ3wHp=;qr*`}90D~MqI;u(raZLz-cjeV8RtPs;jf^rc5IFx*c|W1a{a0A
zvUf*E2txA;5*`UjSn7IcL7piDEs}@aS?>v`?;eKJd6yumP{nV^gC0B3n%k<ryY?Oq
z#%QD=-L?ZDvl>AEVhdg5PPu4>!z4%{ikH+jt-QBevHiISqFx;aX>0H(55?y$%t)Y)
z%Px)whXwD~`R-~rG;n#hFQV!yD*#*g?f383qw!u{aIFKpp^t3pOz@!hHSqPS@tsPy
zTy9in4%zhzm;!~e;%@vV5tcEyC9rXDT)+}kK$L<CZ?b(!=>k*=lQHl8CLTwvrrtrs
z7Np6VNlw?Vhg>xOohneI)0BQ^2S}sz@W`2tT)&MCwxd^@QMO(hm&2;4Sbu1sAhz;C
z6=+Mh#(U-cyJ)z{gGLNNj}h{XoeFrA<480+^#{o0BI!+~6xvu@>w*ZDE_Ca@deodE
zJY{08Why}X?uvTXV4#_)X(hPv>a|QDCh;($jT_w9#|8VpuehC^ZEfncSHlGGKrl}=
zgGRSG-z8|MV~~>r!IndDl=5KhpSTFU8l(72_!C~Z%pkf*VKSr=x;>HDdt_c(pPCYb
zkv(5}-=;^A$B6u;0(i|wle58k=72n%cMce4^&na=cAj_qbDXYzgg0nZXMtPiILuf`
zfI(pNLSTnT#CiN#dXU4!6@0;tQ(+TCnI=C!UrTTD*RQY0iTxr65C-I-j^JAU`J)1P
z%V$@R`I{e~0IjjCwuw)_rf66R8d|UvKTz1t9#-NDb}+}GLy|ams-UXKYtbu*4!RDo
zuoNQ4h10@^8JB89NHh|&pc^H_rN^_KD5$%nPs523l+69@!~~Dg0ACkTuCzQjAmAOj
z1>nm2A?s?i-=_z@pDi-x(gQ{XNJ`%cHGF%%j(SIf&3K50U-USI--1BSHTfvHcw1N4
z24Fh2CezB~{AGRpC?LiM$OJZMel@U<8|7%W--_Ld8O<IrEQ&U6Cn*W$$$IQGFk*_D
z%eeE3+dsE62bZRYf~dFh&7)ckWL4n^=fv#B<1Hr0iLokeOSHZD)QdkrXs9Cf&!Kow
z4Gn<!_~sn5+$#fZEf__=0|~5KFP6Nsv{ImmodRxM9dN1IcUO3ziu`wZd06m<wxOXo
zyzU^1xO#&{kXZ09FR>fsF~yHe;XLwM?Lx@35#e8VSQL2;jI)I$sN!paQ2h!$M_tIw
zXb;VsSz0O(#Y<Q$c|OVk13(Ew>-G3Fo@9vRMByAcQ%I!8VQngeC{mnUCgQ~b2`PY4
z+>F7naD^rIjiZ>Vxv&=2R!~IpZ6aopnDqMiQ?@D~Ab<hhBv_FgA|lNl-@;%s#9loK
zd3gxwJcIoe7ZQ<+IL167l$0;WW=8Ami{E+X=NAP6o*H6p#qbU!0fbV35#&$|r|V+8
z;cD8=S2!OGeIn<{jQ`;<vZoj3hnx<wv!7=cx+RTWnK)KlRP+^VZDjAoKTm1JH}HPP
z-JRx6p&k%q8tU&q<mBWuJv(cFCA50Yni`BR$Kvrf2*}P7Co>YQH**C|dpko#IuZ*3
z-gEBAllee|dDCw5J8RrV&Q=nB(jSGAj4aLc4K|~mnIZO+xxA#-k!`#v?65lyR%f3k
zbqSyTi?B<<#wkjg$RuqsXOfU@tokjL7t#z~7y<X_=;{i?0GxmSc1S+pN%*_62;BV8
zHQkv9S@#Gad2FJhEl~^5;DYLc7x}Ca$3z16;izazH`fW9z<GU8Na!j8(V4<*a45B1
zB312&|MrZB${(AcV54q7=-w|;ffUE1GF1SD@#&WD1C5DiA*uorY_O}__{2mVV%KXh
ze0&31J&-f}By|7)*@Qe)H*rX5x~+{vb{)O;iBw4)iRjTt0VGW+R>6r;f2`I14FXjZ
zL?-xst6QMS7Y*{FwjNQ;2b*`_d1CJ;WQjYJ0XZqvFd9mU>gkyP(iv6To_+qui(BA4
zNmNy^DJt&;t$ssnDH`5?)a@TKc#pGK4g|ZzJopA7{+Aptqrv8@qY*f^i>$GR`pgWT
zU#Xize|(`?@&rVOksqsRu^X3QO`vJcr&Yp`*bbT2z&92{&6zLXn086xpe4e^D4h8m
zW^N@qI3azTb$IH4Q|Sx2`>31ValW<v3a(OV!Bb`x=wTaAjDuDyh5~N;Qxr~q4u|0Y
zP@~z8IX-?yo^5N$vu-@#i+7NK<_ACHF^iurd~gShf*#BE!*4y_LNXq)fwypU*@_Ln
zqrEgpSR{|c&MVtLYBA6xk405xc31bt3vcf*G<&u`#qO2Fu?ZA5+qFROuRpcRhkHaW
z23dA57CFB!{?HbyAshyGH$12N%WC*)@ta7Lc2NBGshk2TU;z_^EJ+%cy*d<rsAU0y
z)tdGsE);fm9qELixJ&+JauhV4I*Z(#BtwRAMzI1byW}IJPaq=LPItb<ERi&k!tevk
z9j5-?X}(WQyZeYmAOTe~fZtat!RJMCizvw<JFdbvpeffWx6&suF%ikBu9+c<IEckx
zELbFxNItH~(BQy+-_m_V8_cu}Sbw_}EPWvVBlish*4f|XIFOUHzHA}E2BW;*?(Uaj
zyNKa7;e79rkxoJ!bAUC=m7NBI$lp=kM8QqyN<je5hYtaJ{$vCp--&Y>T-6!YuMe?m
zu{lc~ajiXKQ7w8FwJ^EfuB9tR<jJ3GI#7dw3dsTF4b_P1PgPS>ZYA)Mcp{L1*Rvz6
zY0hEuX&2hzth640QWzTqat(u_z(6v06ui8R-H6YrQQtdH(2M;=*9FaJaa_9s_>47}
z(^IJ+vI}npoW!o!f?|B89W;IJnVpLyWlFDuF^55dbvfjzG&wy;^t>1#P^md*;^^q8
z-3=w12gP?eWl>L!qcYVrSR>8wr+{Tgj3(L(sH?}2T#%Ef7q<yRu%2&R_L2$M=WJmx
zcdw)uo4Y<Xwh3|YjRvTWpV^O*FNYu3qubWZV7NG%m}y19;tU<a*EmTcp<gsIGEy>O
z54C~1z3#_fL#y+65x>)ll`A>e*w`d8k>i7$BYo@EE!A)u{_>y-$zf>$8a;aas#(_M
zCUy`weOo(Ux>*({F?*rlhvU%RL5P$fd5f*jEh+3Me^<Pr-=Q#l^Z>3V>AnJ&Wx!Dx
zfgN5a21&UeuYV=?#w-RJhdyPA=zx(IHj36ofVG`c2=C2$!&iHy<ph5tJs&(!{*YXn
zR3FH*Y5|*)&^QBX5yPfU8nru+Umcy+PI+G}C8LD$!kbGi=uzYETR~yt;8e|ufXFbZ
zC@@ulRp^Zn7*2vQ@TMq19CS6P29KBXI=w>*0Xc7lNv_qaYza~#uKEOGmD_XeCXIt)
zj0Wt4CFV;-uKY_-5vt2%iQsw1>+SX$%k5(byZ`WEFI30&x04Bgu%X%UKAnUiU_WJ|
z*AC~d9yUvmQc_adskxNF{x{qt&An~SYC1Yu$juNW30w3UzF=za#cQZzM|OYC+PY)M
zI~ZcsM&%&bCR&@(;v_B#9Nu6jeVn{Z-0Jrc5h8$>`~(j+Ngy^t#RV8&Et1rjUOT`I
z$c4Dvc}RJ1`_q@I`GYukY*JD_UVk4Ioc_tbzjSn$j?S0VKFF8sr;cvfhf_cS-1Pp0
zl=OQtB$lJtZC=bl9kzd#M8`0jPFn01*$?1>{^*j`IRRorATm<eC^*uSAXNh#s25Qf
z@B?pSDw~MD@R1{5+ct%-S$PjSv;LPD|G-ady;5+_*{W)N0y1ZE80V6lIC!C6y}Qgm
z9LJ|}*7Vcx%jBiEZM;`MA4%|uLfx(~eh_Zpjn-1mQsm%_zl2;nLBvEx20GK27yG21
z3)KL2R@T#y$%AlrX6qmdR2;hkg?({I?cuMb0Ip|39>BMnTqh?~PSRMmo+#NX1p$mC
zInLWA_Q4;$TcEK2z1(`;Qj$pS+7$Ag^hXiDPC|kO=zJ7Lb(=MV1G26hBFi-t@Qi=9
zPXerUkScMqB+|NtNdIa8*yv)l;sAH9N^XGTl|DALEhw9k6A(MDfgt6`Z_r2F5?NZq
z{(0BUc}g5h?S2)XDImc4$gf}T15`GQg8&h6`pL%q?YQRz(t=U8F7IGln@6;I6x5!|
zg2hOSlb?ymB5mPRiNM5;ZvbgS0377Yl)_avP5~gosl)l@FXfwvQz{ip1b#Qfq<1Ch
z*Iqzm2gZ?WcKj^NJ{zr}ckb5BsXN$`9Sz~BUg8$r_{dZG&!anDN~J2Js{M+Z!~FQ8
zz1Lj47OY}wRk&-x(>>I}ON$!sgNb+%>W+T9WB9RvW>Nof05MmZAhxAvDTxxWv|!oN
zrKJUj<Hz#i?hc|R;oHsvkf!(~$9a;v#+pyf?tBLoDXk~Tx|Ju9OA!l{i*9fsjLhc)
z+#QGW*YlDX(9dctxR`QA@j+jwbs~6D&n{t>Do$*1P8JSzh%7cKg)f1Dk#J7!#z7!>
zF;3hMV#kp2dRl-33^t!B_T~RzQzz=7Nc7;*Ba|veq#l6_M7&|eaD~%CwP9AoHJ4_%
zMWPNd;`K)ENSgQNXYiSn+x<|%s>d8o=>(&&<vT!|7YiU1vKdH8D48_U?&ebrAs}rw
zD1zFB%4Xf=%!$JQcslhLvb5QAQDQV3W&y%ZFNcS+0&1<JjKRWIvB;ivcrBsT!z^f)
z*T#esz8b)&;n9{D<zw=(Dvf3d`ysblt?5Q_{k=|CJj5+&v2I^Wwjw&-ez0mIXuwge
z5+7zo%ow3dqzd?D|LohaFb)BMhB|g?E%g7Bjhltu#3L#)%y{aMZ-k$YwQ$ftnRXH6
zXc?#_!xZQ(h#e8wjoO9KD^`GPX)Q#NN++n9R&i?+4{yUUq><!+g`NSa<c0mTL%S%2
zl-<ClAa>bB?G6kK<TL9yH;;L9eql!(-)E@s^NB;L<=Kv;3rF^TSQE0#W&%JbbsY*>
zrRh!EUcPvt$-e;1sLS?g)(iNd@wb3H>d6fuEnss6`)yE7_>ml>KobI1S_*Nb^7V|z
z`ECFQ@kY@xU{FHvZ4M&RO}B0felK$VeY`M8;b5x9W=|9u1oqdIVG*702x8M?0std%
z>_y`q@?nZFD&Dwp;|m`jwtF8#uzRC$q53dn%|GUWwPx-%a`>m{{Fp43j*ebx*Kp<U
zTS!@*gNzW#`ypyO&iWWz^~=ad_~83PU`~%)FqR_-6&eE<&Yk;;tAJhIEA+MGE{RUR
z&Ho$+lNO7gUtbw^D}N;jAkN%kvr#G*)W8cGV6;m(ATW?5$a)Y({hx(oanyks;<2G?
z2j~~dr?n9>dp$J{AV<adp-L)KWC1leOtC#>2u?LzF`A0Ua1E;XKmm?0O^%J#Ad5;u
zY$K2`?!%-8jz<*!mMQgPlauw>c;+vLkf&fKDQ32HyIoWLJ)-=y_pBRf6#6z1YdDg~
zL5p9or2PhQwM3O{m~ylC23g7Ax9eCeA~KP@cfU&zMATz&3k)D4vp@LR;q$0R6bX3K
z8y(lnvN(0t+h8;09gQeKF(X<if=tI?rUG1WCw3_}Xf+zPq-aPAV<kMULJhhEEYKcU
z$xKN7Ngu^`_5e_ED*z5x5~UvX2#uh#sJV+P{X7V+5gHoWX>ET3c6+b!bQ^_f3AQ*V
zj9mJl&sSRKgD>NkEe3iPAL2S|Bv}GLz*6dx6?AnYiq#RjkY11}sZbG|h-dTVP)aG_
z<|K+p&9r(3vm6Q|us@E&;=3Ve|BwyZl%VB;C?t<>fIQ{`f>BxJOcz$OcC?Q^sv)RC
zI@ivxU3m|MY$!5>Xlv7UtsOXQ;O-lK3x_;<L2L*aEAO2EjQ<SX8%#2p0&%GB<y+!I
zBkE9`v(N!Jh|S)j_h(s(Ki>k_^-7}lpD#%{%01b!Ujj>%_#r1ckgbl%_M-g!ebiB6
zWDVg#l8?KMc)mpu6M-hA;7*fEV7pw+woPvQtcf!@5-Vi;Q=;C72bX%D6F@^H(aVJd
zZ1fD5a>^m9K|eM15{B5kiO^Syn(CI;B~vVaVAGCJ07$)njZ>_c{#>N=Tl6V<QPtLA
zRSX3F?!Wlj1AM|RHApFJPBT3P@b<F3eO99*T<&HdIFKmf1f+zlq@O>38oE1t#S<|T
zc_e~~-K)p}$|ur*68IcSD0u?-6o+SPR1VfON`DQvx+@8qPY^nvN?!W8^x5JcK;u2i
zT$ER`&>&ALwnu2P1<#!zPod%AR86Trm&(=;`G64Auzg5v6N~NfG>G^&3U$<s`ksQc
z)AM~OK68AqSs9c2@wTPNI>#WCOP)*A0x?l1*uwZ6?&%Chm0#fzUll2Mfj$DXBP1?d
zxKJ{#GI4WyM2l3GII|Vp({J^7kUjzOWB;xM)K55^>eX{mBB-KCSUfC*tV03Nb&}KK
z!j6hzvbmWBZi=IJ{4%a`d+#5R_7hSVKHwaX3{}KATaV^;IN6gpDitbV+#L)53w><K
zDDXOS7G?}cC+bGsKaVI4DU$)Z)g%<^*T=@7ji3ib03kH!Qewl;oId>%GuF=#%ZckK
z9@t*`L>fT5%ZelzNKE=L^;k>Jv=Nb;c#x!Dgy?e)w7{}DCKmf&cQEY|NsdR&U;uSJ
zXxh|5m%w{qx{S16PzjE;E-utZsfXO{!cHF<9j%70ZmFsUWag_dDjEhsdurCR3m)(|
zN`h>+%RDF&etpR#Am%vOB^x$g#@m+FRp*nW0-%W=NUCU7WFVM^161e`I`H8+L#h?Y
z>)5PgRnkdh=q&zLr?Q+>qA-n*h5Gtc57KIb_Ocq#yC`U4BFmIRkpNTOvlzv#A)=5A
zb2bkiJm~Drqn9PO5uv19^LVOBDLu20=0xi(RI}3k7H<I6sl(luET9|J$o-vj-Tu@C
z0)-;(s*$J(^0|CbqIiw~JJd=cbtS@6Gyno5-7oz8{gHYAxWck|3lspjc@NOI3eUDk
z?Ao6w4EDFDXR{YoC7UR#xvZt5<H>A+(t}~fPVziE9)?bagMr2^?1(jTK}w3B@zS|-
zM}%Q2(%5guro7qNusK5>=qZ}HwM6Tj6fNjrq8OEdekwB4h_SJ;9Tmh!43gPba4(Sb
zJ|zGI0Co}>@V3>+@Nt1hpZX#PLw`yC7LmtS&bCD77l|7e@<p_CdXN!%K>`&kYwv*r
z;fLlg&L5iw6U}Jibl)TNGrnl{<G~V#Z34Q0Ul9$(3v%O$3+q4=Um$!`A=v6$psA_p
z<Is#Q<8G=T<XDw4<L_KNg-(yr)uM6eZs=jw+JhC&LVPAID$s|JL(bj_7!N#~eFqL4
z<m8mXRwGSTY>k+p#c%26+KW#|UbC%bTNAK`evzM3lZ49Mrye%^J}qz?j~oPAPHGJ+
zrs+1UY|=LG52USnB#I{uAt*HijUuffUbFn>W)bCMS<8xikyq&=Z`w~V1azTYM!5Gf
zSVBeL8(XKcA$MV%3zKt-(*YYO1}j#q=s~mWeM5yHi`k#FoIJiiKWtxtb9O>!pVu3(
z6TVUhQq+;9|8o+H_Sz3WNCN?V;-gGdV4n#7qk?-WHg^cy<iCDG?O!|m!v&!I_WM^D
zY4Jg}+H?5BhY$9pE5%hg{-E^#^|#1BjYtrn<n`Zl=RJ^=;xCOO-1*yc)B-vm2WHfv
z<4hS@|9`r$8$ggwk%L0sft-AhohxKqQgo?&;-7f?-##d07s+u+UNOW=Y4|_A9fh7m
zcF~!id+oxPd?@&n`2DxHz(1O`;n6~h-=o-CS65e5``+s|07`zxZ4!XG-x!jkbS(Lw
z975V(e?TNpc}?3++lAjbsf{K=22i^BAzU-o;pXkzqY3fjxJ{K?<B}o+THXIYKb7(b
zG6H0;G~#4_zv<6W_`ke{GX1|ikw1UF?mxf&U+)y<KlkK+yCwg*C;$5``M+{cUZrJk
zDl!1IW%asszmr_?Sa2{;pg!PXG(l^0%ORy_?gS(IZh}LHU`73bXI_nhG4c~>Pckhp
z7LP@xeid$?``f`LxeO%~oD>A#y8)zw)VJBy@3&7p!B;Nm92bo*+A1q3EKCl^-!)JX
z>ACnm&CJM1m;_WK{_67o!`_#;bGdeHE0t!6gh<GcDP_o%B100H$`DGKhfEO}i!!DN
zWtPlSW(^`q8A2puk|J};lp(z5wfFNJ-+O%j!uK7=`|M}G&))I--S>5^b*}R~*IJ;@
z=JN5c^Xx<=+4vS>PtzAP-@FO{{_Bxi|Kd_2sF8~XMw7IS7aIRlM*Z&x>cjqjHQ)W`
zAO8QF_`jdhzkmP#_A?>ppMtp;&MdgqFsKnC{$95M8Yy$Xzcb*k;Su3gd4ZipSWRc^
zA3~92p>pO-2heF@G3+je;H+F~Hy-zrg8o3{XXS8!U8oDjLm~GPJH07yDFHwxKYm<P
zd<=~Yycpcba(M0605aYG-o1NEeaAo#5HAvou$&i>67ewz3@y>YR%p!9&eu9ep<eu3
zhQi)vywYL3^AM?3w9cb5VG!%S(nEY+`!-jtp)(z)t&*Xsm6!|sp>cYBryghwum$YO
zOxPGmP(ArFgxe!(KkjgJcdUqIGIaT#1AG*<2`K+SiXK3<j&d3V?>#&MfVGME7!4HY
zU>@%mg91D~0ZS-chXywqUt~ZDS4UX{Z03wrIA94hzyWE%nlm2H#QX|H14cQ}D{BHe
zc3qR$G1q4=g3^<a-B?gczI!;QN_}<p37C}b<9Ri9fcHrNmrsfV;M;7B9dUGW(gFq?
z7#WG4SP;BWk%QgNCYFi7RA9iYgCo|wN``gEY|)TV<VP_R$>sgV1ybWN9X2@`HHTtF
z^Yr?p4^N%PqVOqq?cU7|f((PzvH^wGmkI^`S+o8GY<A?I0TAKyWP$d~(o78^{)5Rj
z^*05pE;gt-G(dVr;G1W9g39{5KiK0wXc$PxhLv81c+B^+v$F}X9UL^BuJpxYI*T>H
zaB@CEfsCym|8Pu3!Oc}+texDijXJp{#pVL-MaF5Tch_hy?BnEAfVpXGc@j2+955+e
z2%77wASV}nu05cPIxph>{rhze)T?TT(m=tSTW;A3&JUbt3j?<}od6^XO#_3tI#<AY
zZA{<@p8gMeY7vs|E*=X1+%UTBurRiV4ysK^NXTgiP^Y9GL=}&nSj%mwHDh!3u(0G}
z7xn$$FwfUFH#dKiMhtI7b4_HX02h}cc>H!4qoc>9jXy*ANkxfh1?{IB<SO7~(adkS
zI<fMu4a>2}dw@0bA<N<Q>e>aXin8`gbG8cd^2A_Z@frMSgG`M+ww5izE>L3Lymvhn
z)fn(BIH#x<QCO&EFuI8NgPTA3Z$+v72|RPjzZ!-b9<6Bi6Hs<Ul_Sm#5STD+efstp
z;{hY!{zw9qZeq2Hsf!|TQ)K_3$Z-ZMfr@Ko<qry79Xz9Pm^a-qTlu!k-(S`I6X46V
zVEt+g9h78t{05Z_0u53gCie5{<k>hRAnA<a(<xqt@8LRzO*_VMY`zP+(F%5J;_-&>
z<^iV>rbmjx&q9pI$B+B$^_<bb6Q!p&n|c|#AAFY<Br#zrs!7W6Aw0wud{wg6@^MFh
zFqVF!LL@}PS^Poxb0_X``Xt+6PmbP|djfC+qLegu`CL6Y!un%$Gz;xb`jOSBQJ}K3
zJTBl)^w6vcc8WFBaQgms8x9Ad!&}HdhEC3ZhC6EI8f8{lSs9j@KEZfIHE?vzb3+et
z`@5kPV5h`mB20k}Y_K*8$I(y4Z_7pzP290CuUwvG^4K0){z`_%=Vieed0Bh>5>mZ_
zX;S<SVqOPV+ezfelDw2Kog);D4a!`VSTJurNw<yrOv3vJEkJ<rc+D_JYzC;tjJd7u
zbK{5NWIVgSjF`k`XOAs+yoFhIB7XanGUCW1kmGt@=5RbP;S<4SF@G(n;0b7d-N<*S
z$ZQ-#@dkUoDJiUJftFtMBpS9lCLBJ2CMWP0_nnB<h1F|krxq@sF-qqR=H}*@7r6JY
z71pX6gRXq052s?+*YDpiyRg8WnLR{=l&IfWc_d;`?s`ba?MDeHqUhx#WM<n${3Uus
z!rNGC%pff-T{!z39>+(<_{MW;$Aa71+Lqj#@wyp6yIh)#m@;&AonKB|Mfc;J8g~zI
zxvjWJ?3?S$_s7j)g2HG`7Pmgav%QVCum4#I#V>!)SdP)l?hhr_IwvrKsA#rqZt@8t
z90PQK!jc~dyW7YMYq?JL33_0r^=*i1!DD2zqT-M%O!0-wlM$@gj>Nz67v`rqHQ_#N
z#i`5ccmo5#czhO{kdTm;8g{wQsb$vkj4uu0;zm|>Y`6M5|69@9JlEh^>o?uJYBp}%
zR*O03X(A+LxxM$iQg5!V&<8GL8yUJ;Iya=Irgj_0r^z1sAq;XoVR}D-<kOf(sJZMP
zyf%}Rdrii>t9>uv49uw=3Pxpl0?)f>_W1$yI2a>l-hTR2+&w)}!MM`Ja?7oSl}&Sf
zq1O0U6utY(F&c~hoy^~amHvZxW7VMbbLt_<FmVk*jC?<FH*oC$xaix+woK6f3TKb}
znmh;#Yqrd4h0{=|G0X?I;bI=LU_n@`#tYbnjb7(gZEnpv{en{QW}vbfOz(KSQC`8B
z;$w4x<Fp#$1!w$@Vo7v^5G*;>7>BnNB<>L`@6`XhrxVn*O>UJ;%k>{~fu)@nB)0Gj
zU>yrgs8mopWHER!`sK$pIfrrBda<xsWeHtL?-PRTa>3ig=J;#~SpXrni-EynQbgT{
zP@2TWip)|d9atZZY-^31`AmNmtwF-qgRHLnaPP!Bo=sf+`_9bzt2ei<G<?EL*l<P~
z0S^55`BM;ImDpDhLA0>(_HYdvD|2y|Mu{hK&RXyw6v>H|FGa*h2D<*GR{b>$^nbQG
z{b!=5<I4!eIpfuB?hl2{Lt%j>q`D;xD2wQ4m>RI_2C~V4wS?hld_5DcrUuk~j3;8t
zUc|lChZ6_!?R;qa3j}5iX3HS~1j$agrwq0aD>y`>=5Td}p<~y!EK*GEUNWasgjl5b
z<UWabWP0oDUjzRWYH0I#`;FGAzZ2`k0a%_~@LJ$J{~baDv1m?o7I3_??gJNz(U%=~
z4NlVUd|g@ZdH-lmd;$@B1cFdg8}xF1@L+@yh1=pR<n_ZfA6r|A3lP4kQ9>g&M+kiD
z=KOOe$)0$IqQr0*?l4On?r<_UG12v~TEpB7cNBiW;ncJ=tS^UTmf-e-QNXCY)20%=
z6vRZ2^VrZL35)`=L<^M<FoE?a1{o@Uf&4T4F`eAWLMS-g8DMn!f<>sHx-6mlf?zCI
zU>c|0vK<`EzC^OeF-F*2Ey{(On))S15Q^z>AbpaxaCHXu%)}Ruk!@FSI4Eqpi<J$Z
zU)2B_GHmb5H}-}7)JD=A$>^UR4NpjgK|w)<Bl@AEuvHPZ5^n3CW#}vTr+Wq<F%mm`
zu?GT9r4vB`0qXcc#1n__X&%EZB&2?@Vm`ShHh0Y{y170dQv_}3g)eG6)v*bRI4E#%
z1Ujs*bQLNfs1<eE!8fQljd4^59ph0;y~QlvYKupI(d`Boe{he<HWDca?3$!gKTh0G
zE=TWKK<SbM=v5sW@I3w{R@@<(IHpc_7~<(=qnU0PO!Xq}1L+cBIp$>mp8)jA4!=#b
z1EZtYD_4Oc2>QTlZ{B-stq=Pp98OfAO|4&2Rd&PC^oq+Gw618+`gcB@M3uuAg(W{~
z$Qn{6$y4W8GHQmt)(W7g+`M@%&<7CM=`JY3T5}9&`~ZQgu(%Qe1w7LQU6ObQ>HtcO
zKTfnfc=)iEDy$Z9_H~w)0`@9$5x4%*QZ`KUVs7>R`wd@SKT8}mbq3V%vbLk;DJu2o
z(WCJnAK89|ab^^@>$Llq@CVc2xv~P_ypF*RgX3`vxdEUPja>WJldU;@9@BRq$GRLV
zb(zqI#p+OHE0y}2{MFl!dC!;T5PuAs@rhM~eu<ltlkw=CQCVapVh?*im#JHjj1qzk
z?LGWw5<EJwA<p#o>ni{WEvq_}OFyM*-eEV(7K-C8sVON8p3hy-+8JV`RC|EDCS>;e
z-_w1Qhl!U{1`=?QC%i=3PmGS@OOC-erUTGzTe<@9A__Gmw(-xAFH{4$(1I<X8Af;G
zEwYfNF!+Vi0E6O8Ps2hbg4V>PGkh!O;bU4^X!afjFmp>d;*)F%X<e}gmUokn&OE36
zIpMUk!VWId7=sJfI^0+75A{a}Cl$F2ulyYD7k)n<{C-uJR)Ih)!4J?c{gx1S^z~%-
zx#{Q0&b({s&0hud2@X6aMiL?wzmH;ltI`Upo6x(lO@YY=al}zgfUe-#2gAfCIA<QG
zA5zgoY62D$$4VbrVrjRlTu!U@ryj!{3c7o@1JuDMz&GaI=g_3JjX`K%vkWsUe7-Gn
z$hUi5j_hrR2p1=J;>4u)<j6=SQhWwN4*fexTstU$?Wi7vU%HTIs^+`qp6z(0#)k-4
zI3QuX*VMM*gU<N@BWaUxEJh)Pqp>Ff89fC)$arxFIXu(DQY#3%+R#es=0jPePQOJ-
znM-6bgw90-53HE6#JZ_5V)m5a!lV3{P{n}#+yqtYb;IvuJ7abYobY9c0g8$5d(kck
z^}IUG+JR-e1V&D~;%kBNUHc6p3~8J@{nolIgSbHs$jBIBeVZgUmLrakegyJ_5b{ZZ
z=2Y$l1cs@S54oQx<0!;KcrulEWdm)E5I1N2o7%F@sz|wvtD_S7S%yk+1j7E{A6g$9
zDVS?vhDhh1NL$%Hpa|zB0Im5)35|Ie?1!;ii+F9}Vxrr-H;+6O7#V`@-1!9iLK~K3
z6*<{pZtVFnRru#gp@rvyXTJ-T3fgM0@a8N$9->k*JUq2o&hyO75~y&G%f6POS$_vD
zKqwZbY`o?I=d?b5*|g7`sCFpIU<k;7MHp6{Qq?nUQ7MQM2jE+VluBeMl(nR|`zX5k
z1x}V>xvK{3<-q(}c%EmA%v9kk>$161zGxSZ5)ZG9lz8C9NO|6PY-`lioHGi+JX8g|
z&|k27jTCn`yyJn4v~b$$TA_B$o=$h_tjD-;5vw*HC}2GvSDfI_phGVhG7W&=Y0}h&
zBnDxwOZr03MWN5$IaiU~2hh2Kc+qY06A}-;1`ivXEYwmB*fIu3Mh+mO#J<0<1LfT)
zSy@Ju7-vJ@D*M&GIUF?vISn+*X``7jIK8V!4pDw3@#XaP;BT(q=N|~3)RL!Bs9LL;
z6R=+0i#H>nC8>aIP7VXZm_+AvFOB!=UOI9~L7mQBhrcsl(T0w1ZR<MA=_OzPhKb&$
zFFzfw(fyg0UXm_*Aw2p1OX*~})U=wM+}bad-qcP`PHlOHdlVHFc?*_*e^=Vi$;rma
zxgP(2esQs1WaO^P?eVSk@Oax}T1LAkgdcI|Q98&Oub)?f@EiV^PK{16UsQT*hYxGo
zFQ5vJux22%ifLHz_{2p2_;?8Z;h|FRmA{NIYHi%OQC(T0j|pBn#}?j|l_fiIaq*3X
zv^#eA;H{iuH>abcy9hUbj+X0#-LRDY`q8+j-2Kn?A3uKN7Zk|iS*kqOPu;9KJbClx
z+Q29w5fQ&TcecUrscrKDy^XH3qq}fgld`k<TKclu^O9H0rgDcTv6R>2?>r5XXl-4c
zEI0{uY!M`T=r{&}G&sb>=%EZElr!1b*l1rW!+0U!txG85^2^HneSImJnVBzITW`U(
z9ELr6IE00@OilIl^)F$RFJ?0GoP9@zf1_rShkR;3+IFB_1n>g3brEL@Z9M0a?I`2r
zQL8ib;2_>aq1TT$4uWL8wTw)Bt^3UI4KzPJo4^Hq1<woJkKHr}Hlc{xv3s{ah#g&h
z{nt%RYJ-#qPzZ<MVwnJEcX#UxRJ<%y)70FF@av1qz`)4(2t^>T`<Y#j@|>12Mxfkt
zi5|vGp(96>AggNs_|AKC$4zVwX2(L|>)3Ta`*_W*4vUr-pWnP$M;tK=hcaUD;R>!y
zqG6@JVdt`aGOHTn&%VmL-YW}Ks6?0q%2Ilm(UODh;QXlFH1HHVySg$5%F(se4+wZt
zO-&KYwBI*28e3Wh!XS~A)IMsC4~Wh0roA1=8t9jdxE?w_KK={9;yd8sTUZ++t$L_g
zWel2hvQMq%c5YXvMvAG%I3&(7w8$kXE~Ob;Sp}hStP1x3)=~s@ek`hglwa+pr#OD>
z(ZhomD!!t)#P5jo;3zFAxs{hU`kX7$ba`7w84I?&a{lhatb^*-)}7G*)g^}?adlRB
zN{!8dt;$e{Rf`oPg<40348P@aZqo{F;xTW<OPAQue4CV-%GuIKu6v-T^CdgGyYNH6
zk#uKzFOumdvhx|O746%5k3aI378brg9czTjN634{<LX3rWQnwZfB@QHzX3(rJ3E^$
z=dGc@x0J_$KuMeVYyQpTS9UoX=j5pXNsU9vYT6+gw9)w6wm{Cnva%!4`*;_cU_!f#
zM9sm=yBX~%yaFotr+394W<@xH7#SEEqNKpj3Ivz3h60>iAky<Upe!8nu#_k2kJf@Y
zAY)#BRcKh6`rW!^48tHfmK?l3(b0^5mzNKrf>PEOBxh~cE>68`4i5dx8iU9HJ7sMx
zEjOW{d%J=Ir_3EUK5kPthmZdii8x0L^W<AF9@J>iDJ(JZE}$qW(4TrOOFs6X_|VdF
zqo_!j{B)Dr9SAPxGaa?v@`9gwugc8*><tED=tp%5SNO$o)Cf#qO@4zQNJ4c}Zj-en
zdFea8LFY^NBM{YtajxIW!0uRV*S8mqA2(pC)tviO2N0)M=KZJlniWkt5H;GmHV`_*
zv8Q#1+z#xz2SoUWRPVaHSYp70*B~w_nG762E-S~*O}i~EE!$2Gwzs#3w|rrh@puI)
zA|GUDx%X;<#ub?M4!Ml$lC~#%juH4nYEYj#z)trWD9G`p%6CLGq<*pu5kq40_ARA`
z0Rdk|mzn+5tJ>BQU0q$BCDyxfIvfV;cVW47LuyM$hd&|`c^fdjldY`fHt%X``a#0v
zECJf>XRSuND)AM<#&q{kp%I2nfUtTf<0*+|v~?^xSFQ-&^!25or@xK(ny)x_1hX(;
z!N=U5YHCDOo)Q*G{7{A7GVfKd(c`>&5L^bo3<d>YZ#4ka+fFkJivU!z$w;O*Z5)0t
z|LJQG-R|>xXtu0VTzx%$ro53+f=O>*A1mJF+4^?8<O2Mtx^WYDOy0mw1OoDW`wWt>
z>&o9}CcPg&-hVJHaOB8t1p3Gv7~d{59h&`GMTzWj=*yww{q8r@juE|wLH?WxB~Kea
za32-L=NH$g@vGjT^a;bIRDP14PQ}W~iv2HZzI-tkr3?A57vRh0yOb1QX4n^2Q%j=R
z;jh^r-!(fk^R>UfsxvVmAr$zXlH&WZ`Ndhj+4(g|02wtdQx>PZ$--_xQ{3(s1JY}5
z>0=}6E-<k2DD>8Mc})M<hT_T}X!1M$J7jtb1(La6dOA0DW^VHK_BOh%-q!b;=gHz8
z9*?xYct%fMf3xqEcHP%eLAm+yy)w&>8}Ps>aj@O2`+=|ux`huP4j-P6i(@ALHhO3&
z@o&F6NxZO_-A@5pnD4Zz>UtI(x7I#bo!WIJ(4#aUT6<bai3$(G@BV#SQB_1zXVwnS
z?;|61&kfI<p(YV;d)hZA!^OqL*RNhtU_!X54R)(D;(ph?@I1ht%$rg|^J>O@ued!0
z3bp&OXNmGQ+*QO4AWw-MP^k^%!JzRc1h4!X=)@3dQ0a=RE6az_;bhd@*P53YIS7}v
zE=owiH5yjdgJpfMU#~^%uL1W&7_HQvfh@@m1tjCIpt8wMZrsL(Md{MAG9?%%G3R7I
z>Jh_7s+PwjfAvXtt*#tJ>(L$5Y~S#FL!|M--=XPV2s`XkCHIg-gyZlQsBs`qg97(N
z9{vp>xEmtZ38-^ojrZdXks7H7<(OKixflB*jPXr|dtV%xnVtQlA1c|y3=3XP2OGuH
z7cW#o1*D~!(f(8eria1S5cmZVc^h;Ddgk}+5)xz;#asB!9I#%ZeIFG9z7K*ajFas1
z&)ew0EzKk6=>7riybBwN`<D$%@g{FJVC@zMs5pimW)_x1z<p%t)6Izjwo*#%W{j5`
z^x#DMg63$vowao^e!B36QUSd}zoPp#mVXR`mboLoWBvr@+q2+cz3Zl`!puH|YuwD^
z$olJ+f`WLqES{&G5ZA*p+<Z*6@bmNAclhv5pw*D&bLY;%)4GLlG~shmya#xAk`^@}
zrsPE(#u)qypjFqy_NpDc%-)t+sq%dF4U8fA=o<lYU5|>~aS1mAUPQk6g@u09Cj2Eg
z3Six<mCD6nK=5!1ET`AgeF?Vl&D^?mYd(Y)tO*R)Isc}<eltqySD1kyuS$!#1^M^+
z_jyt6rn&v!ER?LquO}bRGx{dn3|?xhgD2VoQ#We9?E-DK=D6YM)2B~84;p~YE<|#6
zyi3bC0nqnU>)ky^LU+j3;>-j_MbQuawz9Mg!{M9CdSJtdK%D}=VBFlA9hhNT=<r?f
z)*R@PhT3Z&n2l!H{dOSVG_6{r9^X86-~|Q|Jh-+X^6lHVlh86;A7nmxas~}4^5~Pz
z5EFICL+9BkP+WY&t1x%d)ipF|S)A|f2onLo;llA7^4yBe3M|ShD&LEv!@^`?!aJx3
zT8=G4O|G`FkrQK*a-4fBy^q;R6*ki;1mol4!EVnM$aAXslW^2~Mf!Wf;0sqbw=hiH
z)_JsCTYEC=4!V8hHnJhtaRC2o>ZUz;a<K0G`vGSIRCb)8+%``tj40^nJiv<i--0LZ
zV+M-WhlP<ZFQu~MA3kKTN&rDWo!1NsIRJ*>B*~G{(QvfKhofG&HHTRhCrfCVnNM$I
zGXY|fS5i_keue_?o2LQd9XGPo&CqQ0U<_bFRC&V(I+PZV+Mgk!TM0<<@dZD9`m{Fk
z1>SFy^#?41w%3D^*0dxQ4~<mU_E&2z!|ff{p@k2xQtSn&{O0}p5RG(@fx8-hV93Sm
z{0@}hR^5n+Lak{jDZh9*4jl>$vxCCoXbCgqy51g&S`0|Z5DQ9(#%1QhG{VicAnC#%
zh}~aV4KC(ng_}yC&fM6Tt!xz#0!Wi0@b29qLn^VhN0nVoCoC*@a9%YmDiPB5A3369
z{1Tm-$`>#0d*XDy81!v8-W+eSVf*Y6Ha{HhTaT)6d3ZC7JPtfrl){{2V5!X&tO$83
z$gOUX#uqQXvZ}7Cx`FeC>#y7QsZa$KdM$3ETcqBXnKJg2Nz)M&<4PQZ9T^(Fi$tb-
z(&hn1KkbK<>#u#H7yP%gWGu`VZ`s6_0(6kGkggibH9!BU1|e*hQ5nz<|A>`U2*_`v
zKT`lDD$lb_@Hj#w3r1uuO-hJTqYSiru+hTIOc8x4Y58h4rigV>55%%G9&ZQ?3DMS7
zwCrfE&&M3i<R?#>U%W?I5sp`;F=6@5fL3$(3^twboxHw>i76Zd>3NFz#+1$|E3<(Y
z;ev~LbAtuwGwk@LoLsF)O7eLhd*m(y6VsIBZQ%Sc(Db)lXm{>p=nk87KfwsI;2V7%
zR5P+J^2a`V2M_yyr%qhbyslSpX?-?J-ej7`^12ucr>moFfmTDnyZgb1b-Xme8cKkf
zx7qKp2__sQI4uVUhwM9IV4a#8R53EO2nR^f8Y@TBWl7ZX%!t;rRkA<^Sm?ntQyeJ6
zuZ8t;QpcGTR#I{49#|}!ln@py*mM$vkz+R>7uW5WisB`~nibmDK%!eS^-Fjs4+D}B
z43~*zODpa2Qfe^()@rPGXOgqCRWQu7YD9`#zG#^n*Ht$6$-UA5n^>ckOSZPEv)qo$
zi5@-gN89^2KY4#c6_tV+05?M$K7WqGikt8U%dJglhs}JahxSRMzF+gNKhdvyb$&If
zWb!M{NZQ_M=%c5eU$0SY3rtoUko)Ja`ttF$2J-PImD&PF{l{LmLMteFaBx)Kc*_OP
zvwtPFDu}#BJUe9fMY?X|94H}zH|$+p%tB7?_qzz5Rovm5-`B%9{EYz@&tZ)@I-fS9
z95I*Q+f-jKr>?#Oo0N1zIC$MbYph2otp++}=xJ<hB<r?u?vvr639;Pn<mgDwN&Cre
zkv4}s-0nbRrbDHmhbVD4Z@qXSg(`GFv2+DA({NwqvB~SZALWgab_g^GnadkfY<@fj
zxtg?R=NA@U8kf30!vl527ZqeyFL2r;K)Y~LbZM23KHiEP{<!^?SoLjn1oDAW8aW|A
zgUHSYrqDNj&CQvyq~-*`<4ph*E(z~BmbuBkyP#LW4Ja6QLum%}5Tw%!9^fV}i!i%C
z5BtG`)ZpIy{QarGmbeb_S#~DtpyudD`NMk<t3tP+ZhHgzC=87*Ja$G#d(rc@8KWko
zVRqJ?#a2Lmv!=MZu?1=P6Kr})?+B}^scqk|mB9x&Mk`%2m4=qq7mTM<)7rB;?G^er
zZQUC2x2d^V>GbKX=yGs?qWY$$ra&=s*K2fk53=WZpgbu5dErqp*~)x<7AP%F>Y5tn
zClP-*BtY%D>#)!M!pMaux?-R|-{?NL!=)VsTtw})vS6+pA%G8EY}2(9ES_VL1o-|<
zfNyj&M!aKynu*eGJAnp$P2Q$rveF|=VX=TtX+T$nN*-t8#G^(ayo+GHL-e}A!EQw_
ztE;PP5?(bZi--mS_c9H;c&q_7L+Ckkt1Me;gH_zx`nc8Njw9Wnh+z|-#2fepp{TC$
zFHZ#Mg;#*x&S{(Uj4C`ff$@IEIB2raK6A~|N8pqdjX{wgwgARHw?l(oPIDiEOg}21
z95<*@{DC*qHg<<l)PJUCXV2mSY=_Ue8eoiq0zur~(eWnOxX(6>5KM8Lf<K3-2G!Qr
z--JI{eh)aZs=B&Oh)xl~SqmUzNC1(nR&yhZ&mM|c?sRl?+~0$?UUh8uAtxWtR1PE>
zgVzTt>b?8$U9T~YN_L=S!Q}nUbJ}(=VdI%<3Qn1vJ4cUbi2oD&l_y0Z?YT$?IVJ4<
z8*ESsf-FRi9!Q^%V#wXdT$BfMqhe!&k^cxd4C&Ivkk<a;;f>%{k@<wW`(Rvys=-RP
zXaK|R`wyWQ6S3Q|_GZ#T_gM+PJBWrq&d*jKxH<7$FaP3=uX7x%r%uEo)=@%18<?F)
z8kY2)8rWq19MvvR=4A-FA<-@KkW3|>K6@4^nMl{@ot>N87_cukIyw-D@H>vR>#whC
z#<pG|a6$p-dT<5J74wv;SP&RxEB404$13dQ6%)@Vas~PMK8cB}m(N?i&<XPG^u8Ip
z^k?!hFsx(qF5vZIi<Y(Ux&sya;k%k#0QW#_NMB4NoWV!9d<N%@dYQV;hFPJryE_>j
z<`y7bPn-bXXpAu-P_!7*4OQn7lSb^2?CQqTYHq12y#(aO3Dgesv^*8Oa!zSA5Dd>_
z1d^v16PNVd18f}ROeJLB$GJxI&rrHMllg3?%wG*o^{)NAa^CVEdSZujvJk@KoebJt
z9Ha7A<m2aAlhx?L5I(Ksj$VEx@~%us_$<zA0N_(gYb!g-94=6MA;l=G_hGEf#Rv_Y
zi}1w6<`-K>0$MRk!jYyOhMhq(xep&dasbzBFO4!Es=PL7^lflZ0o5)8O%X_H-Y&rx
zm{jshF!2tAD4|ePcK)Bs-R?Wf&mphWK~YvlB;7>OG&~z4-T~oS9=id7fWr`53qzaK
zR8%;y>fwRGOf{mOL-)Bq&hi`CfEdH~Y&(h;xhd4D&x+9x<4E%aqLoUCfV(%ng1!zF
zz=OG`8c{&+vE<dx<Jj-VsWk3=NAFN19`^fI=;w3+P-g=uklP%eD!xZ~Y@lA;pF9YX
zrfmvTS+lj-W(6`Y>(p64k360P;49PKpP+xu)5U2Vwrt*f5}!ibu@{7VFnG;SeR>^v
z15m?NgP<@2r#;Z@qt5v*bvQa(8H<mptd1V@m_Lr%O9f@8-sA%$ej1djjvyJU2GIjk
z+j!F0)RZ$iSK*Ve7L6QfhSZ(F2~<AdKAKY8e429)4O#G6QyQI^c<@?0E@qHut|Z48
z##%)ZYzs%q;nwb+9<A9S$b^hyDUaF7m)~jJ@aa<|0byZcqN0)fM&-JKSxdO9;&&Ac
zHC`mhc=Af*A?_)+f>_hK+t<(#+;Lh(q%E~Fwb>VZEZFePYmS52sRJlrv@9wttVNz;
zq-n#g-PKe~P%gzi^vs9hP!=ibH`LX!fh6Wcy=umsp3KMySimJO^G&it>ZN$Zk!4Rf
zSH1`z<<W<a^_3pItr;(_%_Vbl?x*FralZQJ+&=XM@zh0UR(LI+;?>SLzI<?3BbN~9
zCo3b=6or_dv~%xXHRP}}TxxN(ZaHYV8Zev$f;uOKam(dqH6Pb~d#`=EA~iZQi2Y}D
zObkIi_t8Bn+=?qji=d^LfKH5!ypM-pA}pSPBSa<q{(Z&~j4x1zcZ`t?{c2dS;h$L4
z+ZY`ky;T52@@Z7Wjn_|uY}zj^9XG$UM9ad$au|$TmZ55#wcJJrd;5K4Ejyyy0072*
z4vrIosvaKF(b8^*FVAVAMrw+VG2v*YH&jB6Xo4y^gi#LlZXHU7lfa>jZgiGXavK?s
zc}a8J&CM0)f}iQLLrQx3eb9hAL7FM+Q&Lho!V#He+qG20b2BZguCX!IDz|cV<@`N?
z7V1BJ?2G;L)uO_?CY~F+d+jne<5ew{cxslW`Qv&2R3N1GsTe(-7Vs}}dyjd&pY*e$
z(BFR<4+LNR1ZH2V!C9kg-U#F?wrTTb_A?36GzLG>BQ-D%*+k{9XjVdW7Jo5i%7I4V
z?cgeD@!1tUBc5$A(Su*7`Uj|n3`><4lbYkl*tUWQdiy(sMJlS?Ygu2=5UC&Vc|Qp!
zDAe9mSNmEO<&4>>7JuJ(AkOl*<lV3^nrL(}$C)*tmV>U~Sj1=VY5u`vbrUGlc5_n<
z``RaEuuM;2nX4e3fSH-u070as5%CVk?=PPJxXcatCOP~&8HtxPtsnE$WrFFDsA%fm
z9uU%J^&w$D$S@rl9{v>*og7d&!(tE@eX1=`VT6>#@?!t|^*Di#{BEbNoAGLJZ$3Xi
z%I6>yxq;{B#jIpglv#0Z6{{=miq}g+&j;0Ga66NE0y(W=2@{kTXh!MWTIrl0t~-v-
zB*ex?BG&DvHPuk=xvI9r;qhXD%WH{FW!j<Zc<(tZtX)v1FPG0xOiW08%wCd2v5ZMt
zZz;V&ZN-gDPfypu`^IScLDIrfbD%DvEV%~57w@qs#7lr!3@j}z{VW5h^{nY#y~~e~
zprA`vdCkthfB!xpHg=C?dk&+y($S=RSryty1NJlLwu;zv-b2>Ykn%={--hc*)CDCa
zgYVA+Q?~-Lwt{SEqW<jjt?5IIUp6h=vFn__p4Qc6s8(~3bN~$3>~||JE#(3k#1$GE
z>PTs2vRiOX{inba*CC6uLoSwha(4tp!es#r2UyX3SO)I$O>b|!xRg}Umqdl&<hLbU
znHXPe&4TIs5>)zif{67TD6t36{vc!wQ<lEI{(g3Lxo|}m-%R*Y^d?Zx9xy!<A2C;@
zT{wFlJvpbwVVR6F{sX|7Gt$<@SR4Qde5iG^gH4g@Ax!9$3XhLhZFE9@f9u}Z))oQg
zQsZNGLkoW+j^%z!sIFCbD;X-@Me+Wq^PGN!y)1ny`FW@2yzw2E_<H>5Mth#I;o*Cz
z4y_go^C~yBe7=}9l~O4zEF4%;B03~>{LGoVn7{B4a3@?l&mdYjmrx8?V^<=raQ(Rj
z=&S9R@E7@d3eAm;`f6_NB!pFIWByR;ia!qh01!;v0!rXhFfh3qauh>S^kbHGRabur
zg+i_58gl=ozZ_X7m`~RXoM!^d<FYcuv?Dww<`kLrscdIr@@7H9H+QR{p`jye$5sh|
zz|UkW!1^Sm9VCGY1ekFID#+J-7p`6v*|cq2>d%fh&$v>AgcN{5o=5`Yh7XU7D39M|
zICL}SoQu#d^<no%>ofNG-9z51eVN!CCvig68<Mx8K2GDarY(h}E^G<RK+<h1j$t}Y
z?AW7Dqo6%*+5~!u&ne7QJ5;ASYiVi*VClqX7XYwla}kBUnIgSEzA~NWe2I>>NTA4|
z-v+6Xf4<dhX)lDjSRJi{>@H&Q!3WTc&nA)l-N3*A&$aNgClN18LNP(h@y7Y5nC#S9
z5K>iQKt_N5tSHRS--q}dmZC9xF{|*y_Iqg}*TN$|4RLXC<;Vv6IpW>lHWWPO`7ebx
z4^!~#`E%g?j!oD9ffCxZJ;v0~+d7$r*iD9i8KK$umaiwwyo{3@@RCQpQ|eY|3)j31
ziURoGR9O8<*#$j?COE1&OK2V3XRgwvRv6RP7c+j0hJ~f^=}K1=Dcg~8!yzc}jqld%
zUdwb6_<+R`2FQQ@gY9cO>FE{G7bXkFMZCEKyiD@c^S>C@(>4XUd|el>N?%n?l{Ui8
zDeh5u4emlV^4GP=YU||Vbz385F8&jV-c)JNpZOdqlErYEOhmZlmLTo+x^^13Z8Ps<
z)Q9DM_OP8XEI{_Ij>(e_l$Lv#yiEEDqe*6Zgc?qG94cRQ%sCKhHr2$THRQY%!bV8I
z2ug>`s_DvM=5}_?i*ZaCNwH5VNYZJi+}$QdX6E<#sc>AQOw*cXWMH_5gnn3tCk-hH
z)Pj@D6U$49cWV)?IG`kiOS2yH+P{iv4H+YrMKt{9OG9Fmz?}hXJf!JoJ9aG5rmrHt
zh{MEkVc(Ms+INKG;n0?pk`k7D!q*eX0seyN04>k>fIS|YVnS!9?zgeA2DhJ<@ow{1
zzma(R-2;8&4~qJ$>3i8B&*|#vX-orb%l%xS-?OK|je66a<VS)jeeN}eIXOAH=B8nT
zm=%7%fp3gQ>)pjHuod~;S_3wGn@%lV%mP~>Jc_fin(-#jCj9{vpQ1FriFjF4s^1S|
zVy2kn+1>BJ;vN@MS}YZFD@kx!z<n?Lj&v?e^$iV|uks@IY3X@xWE+4=y7ed_6Rf)b
zSJZ$#<V@4hDugVCW`JPnLJU>OQd|G@ST~Pj4Dgw-<Sy^!#Yku+H-xEsLNH~}n=d(K
zgBDuFotgyGhQB9@5CDZDRb#mAf#)CS&vlJ&8vXI3tkkI$C@+ZI5`P#}S0?t*3i)Jn
zLm11tSF?{Ia<|IJyn<o>?Thx-*6=LnQ!l*eZ_X4zO3qBFrecec=%^P0)gNdJG!Y&d
zSrZd#E3_-|IquhuFn^UEWva>yAU*S6RF<$pz2WgqJ0mLk&u?fOqU*cND>qYJV7;}4
z`rWpUi&;52{DrVhw|f@;@%#72tBX9WM@)l1+n;8Bzqt6r*)Iq5JI%sgcrJ;h8?lI3
z?6|yd2ivVRnie<GECwx@y6p064AJwOIT_Aw;k#(kQ<$~)^jZ4uHLVdp)CvnbR3UGk
zy`(euq6BD2)b>C3n!}wR0+mi+FCMpOOK7a9HGTC8=FhmYRD%j+Xx8e!wn23^(RFD~
zU{=o!21?mdI_L04NT{*SYk0;x_DQMSJ{)f%np-K_<$ZIa$F}8?si^`?I1JYXa0dSp
zQc>E)#RZh+gATV(B68+Rx#Xdwdz(IU|08JrZUD0)HWlhbRrm(`X5yMgs1e*?0zUoU
zYg-&3Suj9CKQ!Q+7Sr7oi>LOmr|yN>fQ+9revav-NvM0nXOS8<@tULj{m|N~`S1$d
z2O+ay^0x4%=V+>{?^}qEj?RoaEaIHEtAA*S3hYJQB9Nf#(AimM(qU0&hp>e`#L+rO
zAq5D*{ad#PEh8*8mhV9xCa?UOnhNKK{PflI0*~L0H*en-{zQA%?fM2tDtqxK21=VB
z9`c>_-tih$k^|;LjIIDl?twVrhvo{CSZopj3F(L7aM3Y#2ooVoAqRm0-XeW6u0Dp8
z4hfUd`5qS=+ZyOFjy*`ossiS4bj*X%GBz>U-<6h|dj}qFV{p7VZfLkCifDzuW^x<0
ze}iUv(6ch!!hsn7xN{E7N-zfXw0mNkia2;hXsgYgMwmkj%8B@;v|nIIghrMOYSY*7
z@-Pe)p}Rh=gQBO%1(3-VdX@DeczCF`w4|E8?B3!aYT4>1dcoda2vOtRkWibVtn8ZB
z)>dsy#9aDABpMLunU<t9Cnko1T&!Z%Ew%`&AC>i;V^%wLO0!3*vYdU%Mil;#Ktnw#
zxJ`Hq&bJc1<<T<>EdJaCZapF=c=3D*-U1`^y*Nt{Ctrt#xb?h2CEyF6J?Br@Bv1q4
zvr{>hYhD5PKC0TayjriGFmr?g+nS@E+d*j5Ei&5%lLaRsUUo1r)Gm}i#V`*|TV;fq
zxTEe-EiSDc9b}Qh2aIO1k|`~$FkLp>NeBZh-X<QIF)w#__tS|o@nuJW=S8s78z<#Q
z{JW~EHK;63z%JrCbU;o?$$#-nO7A2)drUH%QNulDPhMXo2_R>S+x9VRpduqOKVn)C
zBeeOj9BwYIZ{Xg_8xCZjw0`&bvjLKHMk+H>ip*AwC!CVv;P5S5M)G?MW^?n$T_eTp
z;GS=z%xXx535EuKjQx76u+F(w?#8aH5Tr}xc8CH^F?kTOwoX_e<up{N8&!HWKJNu%
z{NY6?H5@84)d#8p+P=bNT;_?c7FXnhf{6!dVpQs`S3nU{*-os=n8HxK8`W|04`V)S
zXvF*c^3x`vnWBu@j$-od92AT1P*y^$&&41^{+3u*0v}PbraNq5RlwE~AED{6$;c!W
z*uoG$kAVSa^>^!m<Qd)JS@xP{>CA_)<+N<URa?D|zN{<nXyhQ3Y^b}AGJx@Ud@!!?
zD**CBwU2!<qsL@F_W6u1mw`ukh)D&X=Rmz*!mX|;z0f^VmbG}M{QB&vpFOrTFrK`E
zU*PNHWb+4WREoMNo6Z)YhTn_<1aj1wZHMf?9+!5X+Y0`Jw`6H$*?ng8Q}CiHItO;3
zsqLmsP+yh8{Jv57{s76a&<t9IcwO+^C|_vY<l{4pfD%k8z)QlS5v!v=LVZ9^h61H;
zGQw4P>vcD`Ly|A$6&0rh)A_Z);nYonDf<e}i1*+L+M$?(t*k7QfDe3kvz~CsZ4@+k
zx=(KscLde<s|GI(Mg<d7(|#mPM@x8p?E7D@uPekP`!931el1L!dmxoLUmzAbBfzQ>
zhIqYT;)X&F(dO?lU{R30^!UGC0DB}+obZ#qt4ouWTFu6%PHj$3PF~kx4wW$H5(xIq
zXm4oqfaU#C4h9JNE%rBl`I68g6qbr^qx_N*)81>~N~|hc-y+R0#?Pjffe_f;AlBh5
zY||<9mO{f4gC(%zTiWUiq8i7+gJ;wYQ5XrCH!&kqZKJa^dBR8$N4W(P<lYU_?Z1Pi
zUf3ko{QTgtW2{ivQoBE`14Y`VwxO`Iv-4GNFAF;UzFwsVR=)@iF=q)>Ne~>thf86X
zT?3cC@x;SCtEUOorl}`XEoJ8n)3c_zmwtk7)eP(nPs3*}Z$&F!SWEwJ*d*qs27{kx
zowfSlw}Sk9X>0g5Iel+@{vsDHf$g}b_x}nuFYcir!Znk@r+BR!Q>m4hbp~~LUArf!
zUF^14BRUE6b{D8q^AF$Sv)rmNesH_g+jg|J$^iprEXXe#7Gk<AJI)Ns=$xg-_wVhY
z$6EPo=$%I^!VS3JfJdvCGH$*_FyBE(w+@lx?ey(~#Y;MEHtd|L@VvQtN|%+Enq8IO
z?KpAW-F@22yngh(U6nUqE(-z@tj(KCA|3}iHTsi#fQwt$q;qV_*0(mTvAOvj7^1;e
zQ1hV$U(kN`WsVp<&^V!2d3~1GS`uW}ZI{0!COlwU*S(Kr%a$!A5%37eAB$5jE-5L%
z44r4B`n$Q|O_4@-H!~)Kw3R;pFn0S~1<EaAiYYNQH&^mBlOc!0+j4Wo4Hz4Xzy3=d
zx4XoNfd6~^g_?Np@WdH=;92}MR}raTLzS_i<r?4Cg4K0LkljN<9Z-+y@LHs?gPc0&
zEcFv<ZR@dTJC?TzEti3}%^5nawajg?xJUi&YAN0$EMJj5D*%iL9#l1jhA}folAf>7
z8Z5d)g0XjUdi)cOkei(^xc%w_45!5@*2TP|`UFo%;=ZsfBpmZ@dR_^2D@1L|gQn6q
z_?K*3;OiaE>_<SrI$v^taSS*5Aq->7!%<h)%hP?t)Uq#@smJIyMQbGE&wcnbcum5>
z(y|I_@Zq)5?Xr30Ef`y0SNj_!E8O7gK;WN1E_9>z(e&eLa;DMK(*wzH0*WSf2C#Gc
zq&zNj+(c)o?a2Z)JnlEUn3$NE{K$i!!_9N&U9_)Zq+hq<8@`*?DS5?7=Qv?QuEr_D
zl;-+Q8Av!=nM5`sW}WI9wCb!=bh=mb`7<-%X%!GhX6das4g_5YBI<u2%~E2?n0YU^
zd!NAzmpfK4MkyF89aB+JX|LUQ(5U<syb$VGOVRW<bqz302<tJUpWj{Kq67Z%MNiaO
z%buENS5O^Yw6Lfh)rH_+4ch1f<}Q74{yjZ?c_L>m9ejC;iaEY?oe3od9jbsF5>NKz
z4ZQ}*Cqc!PwMT|!NPwSz2g;WKt6$hUj6C4aWNC^tfz2&kmp6o?at0e+8&6gjRpay3
zaqQU8;WU3fxLDmFC0^afk?Gd9u4~bf8;mTN&}bmrEC|Yw83vu>#^z6;gV?*A<E6Hc
z^TwX>>52pssUML4H;w9ol)a0W0ZT9xtP@SoI43?sj+6KDl0jY<w+&j1Z7NHyJdZ)A
z_O}zEzV(wr+V<nuQNPAr+fKLjaVjY)Zoqa*SzCYjJlXj9Lp~}S$pf|!OD+JD)<>9h
zSit@eLlCI}v%Y`-_U%K#WhKlMor+;g-Ppa%XC)REu9$J=iG_OUj+gIee=E}i5fGaz
z#F1BJ{go3CU!Y*|2ih-Tstx%F8#iy3!>u_hMw>>?4U`wxKR;Q3j>6*O1$NK8fmLEE
zWN@r<mm^DE8VGqo;z#Lpfg`8R4rBxt>D+i^y=z2i#Fq1@(2Y4kjR!7t|9VZa)KdzI
zXnmuZ$FHxlApEsP=5*X_oyuK*-(2)4E-E?!wGVt`#+(I>+l|*_X|NKsEsx&5yNYf0
zyMT`skRA8;U<PO)dP}Y3=h5Pngx;zUTX;m@U=|!>55j#u+Ez!GbFGL(L$dddr8D;E
z&f-M(l<YTaahGuhxaLE#t>F=VqGLyFp@+(1>=`LFee{CMn6s^%ABo-uDE%8s^UfFM
zq7#*$T3cVk&;4~`qOsFRUEK%V&NuwBQ$2#Ji{}v)h~M*rcXoREJ0^)$sB&=&^_wUA
zD&IYI_-+pzjd;Cr^J(q|py^=D64da$3tF7IzrX*K<-y-lPdBn6X>Zlwc7y_>!A-v#
z|6xZC2m2T!$9($ih9rfLmrpD{782NeWRIK_M?0+kQ&Lh=dr-mm?|S)^Ul{Ea@D)-+
zY@yh!WQ5<a3n^X!#S(tS8XQyz0nqpYJ8G}#wRz@YwaE?eH~m1y1Sw8xtQKM(QdUK%
ziBcInXvAPeuYFsDQ^HWl_PPQ^`5s*&aJ?d;YqBObZPNgPRU_F(7B2NyQ;ovmJPC_q
zM*RF!n<BQOzD#=z*hz7JtqUNgCPr}vrL<|9@DTod0W3k!0@gw+1jOQqC~Xjp4M-kS
zgj;TIZce)@FLdm7F0KE#D5X8C4)Fi_Pbx8Gc743@I%wEqz`!7%>E%MyInPFnL6g!j
zF$Dqq70LX;!6Z&h#dTdM>1#!2JUu-JWj<un<_!sTOrMwEt7jfmT%9Uj+eD$gmpK=Y
zes|j=E&U^Fc8F{_sh27;X6KW*;qIL~HSl&h4Ajy{U!OUtz6TBlRpVv(WmToNzQ3&-
zKWQM=g@n2m1}wgq`3?36%RJZP!<qGhhVw&-B*fn2P=G!nxHIu?pl_>QiFytx&lp$L
zQ6>OjnOFuP)!ag%^kmBN&y$QU`BO2iKgi*dL$3b~?*&EQZ5M2Jd<$c>5E`WGN|8%(
zlao=r5vK@7gTW%X22|X`=;#AyLnWJ?5ewk(BVJDQ$cFv#zjplS%6S>v9hS=vA3X2}
zMLJb_P(|5F$!CS!!gcb;nD-I?9l8_1Ib-7DO@Jak(;|@)35uU%if3&@VNG#t%E=B0
zued_@@*NC%4gd>)K0>9yS({LE$wCR$n<mXEO*#N>0dVjtK1V@TplEw;_a&+0q5&1>
zjRTgs>55R?f$BRo6#_|85O{wp<Pp%|TOqp1qXO(79ITOGhzhn-_MWzUSGQ>AjhUTa
z=DOf97c6iu-T$`?L;ZbV2sXHI3$XM%^sDma%Y_l5%c>^=q5yKDlCCRjwMquvp7|Q^
z_pb-$b?rkF&7(7%?CfigEzi)BnY-H0z-0PCoE_uOKHh)&GscMdVi%<5Tn)_G9{;MJ
z84McSyih!^qIM5J&INXsDBi_QbHI{LfkS6K=){|?u`w}++f8fi+){W7jltOtNLsxG
z1Vof%z$kZM*o)@o{*1zAwF+iBe$Shn_4S#Eyhj9ceAQ|Q{p(Z{>*o&L-*)zbQr3!3
zLVSF#rn}r%t96-0M)-E-o_eT`BD8WGs+^7Hugdoa39%JB-A&$p%^-{;_<ZROi<@a8
zm#nOKJJvApT#dr|KJ^2!0H$$kvvZ7&;BP|o{%R{*J~ex~;sh=S-eq6s3!o|`q}ZiB
z=I<2<?#Y#d`4oL(mridJc=;ju^V(ZO@{$t(XPK6%{V{44<x&OOcgCfLK;-cjV5Z#V
zCdYdA<dt)$_1gRVc0j+P+P-}|I#ss=&2VHCiM%@F8@w1Bno+rWB2tnJW{9KQt*~tP
z+;w4!6S`3bWhM$(2VnDOwbLT-^yOo&>p!3slf&;VJw!4OKo})U^>P-VH<8$2G)Q^1
zmy6`9wBjYZ%JmZ)>0IWvZ`u^LM3D$p9!(Y(VP8SlA|Jni04Nadz>R$Coa4u!89au&
zoLFRWhq;F-k0$M16-X2d;GMnvpR;fN5`c2pUuy&W1npIq;l(?Sp&y!JDq69WUn{P>
z5mRCdVt2!ZHoCI!EDds6n#;Wfl2n1xI=S{JJi|x|0_qm+h-cI4CAMxbS|%QKf9uAq
z5HLaUi2xYI*cCiZ-e~6}>BTE+TD%*m;MPKgRj=>~6C2(EUFRb}%uM#NqLa@L$_%<n
z_H`;P3XIqtcaTz>>7$1>LSL|WBps7bBVJBrRYo3&J&44AGjV;)MTCf*cSu(NDEQqd
zhh77&VtJxL)AqNJTJEpgKqm*G1~kc;Pf_q(yaD#fd>%^u?V3G}3_v8qzYPdP<(EuW
z`%A6C?ezlxb5Q`e3<{sGchw6=wr$<2trxGUg7z7bLE&<t#E@;I3GN*%ECNTI&0%GR
zK<k4DO4dN4HR#NyPfveUE1}M?2hIJ=lQi)Z7T(OvJODZ=DR;SVz8<!glafnyds)fB
zFHG)09=cX+B^GhVBx$qB9>vyL#ThM5fVYQ7-Nn#h%g_VQH7VCpZ5&m>m@<9cDJ_?N
z>Fmrn&EXRY4rzuu>mK~#xkH-lu_$vFp$a-HFr1Ni2TRk3D+RJURNX4g;H1h3+YinD
z9b^2fmKGXJ;keYBZjpo?AcBi?6AhKgny9No4&7_0s*>FqeQQhJP<~-cedSqby`-F}
zTg2N6vK59l-V%5MJ>HMQ>^|Mfs9eyXpFKhW8IvBj`hL$99Af;E$M9#Png61+N*|b|
zVfi;o|GgTIMMim&{#s{-|3InyYj!qu^cntr4g@GL=Ra(xQ98K7wmw#Yb4#p_F@NCj
zP1*iN((A7{@yTUx|F;V3TU+CXMJL@f6tD^OkBtQb^Yo+efP+2RK%^Jhh548lGwy*B
z0^DyDB@q@TpHaS6ef+4EmjE>_1GYS5l?@P`kxPZXl%S#spU&C-&owFb5`ocuY#*Cn
zxL4H2dYyEVD9CLI1&zusEo~skJB|sN*pqHiHFxU&+!)Niy@7gp_Ae1^VxT|nMjQNN
zMX_@Ol~8{MS?<Y3fmyqZgb_XqXRS?VzO1oK0ZS4S3UDkX^aG>4hNq1hw@$E~-<AR`
zDmA*ep9iB}w%#1B9P~$*32AzPFmMT^wm8~?h_j>>qbb?YFN5>{4pRr#EpTY7T7K#)
z9N9Hum-Nevwsk~<KCg;x^!6BIp4zrA5HtZnc2`wZg-jf|^7$3S^>sabXaZ2LjN;{M
zRz|jdA16VJ-3qYTr6)W=8`2}VxXLGfb{0?qV=KYQHs%Ef2S`Ef5AXSG0{%m~7=v_y
zS!Uqh%t3rJf5lL={h*=3joKk|SWoO|&qiG~Y8}0{UB>EXvGu7RFXj#C(?Bp{-$9c-
zMFotB;5Z6bPMUe1I&((p_;CtiCSzqy+?hFsu>=g1nW$HP1D999AXDUY8TDL&U7w7s
zY38CNCgE(p?1@lclW?;*1-3JJFfO`Eo3q`o*Mqw>M$4pQ3kAM&0Djk&Qf_ynlzu0E
z1IwU~dSy@TBW?d}J>s*C9XkdC71%zwlCh4P;<CGoi<nt+f<Px|-@WK?$PBaoQsJ4M
z|54Mj{0i{(MKiOvk_tt5_4;L3a*HR>p0q37ikHx`PoaN~$tc{R&s$C{s5<eY0{?oD
z^Q3n$`rd9AfaP}fSe(h!ZLWgF@OE`3A#&4w3Kxs^7n?gc)8=XzM6FP=MQ|>ABGA1(
z&a0)RbsBIM&TBO4nXePNXNI{h3|O#|RQO+L20-TF2u2Y83oVu;<S5uR*V)+GuJ4Mx
zqkixX`?A%)jeVHW`)s^k?nMIF!J3}i0f2tulgoChQs+S2;Fv1mn96Rff@ENiDWC=w
z9*nT+U3^dP1^|zr{T+^=nkn<${#cb2`1{6}ckif>wlQav>K`VaZ!}4d350O@V~S}@
z-;<S2!Ddkc*&o=&xzKOP%}oU|*y@1JO!e=_<xBm3--d?+aP=<(Nry}Twv(QPbv;Y;
zJ<LFix1i{^mZ~g1CtXFs!CS2NDSYku-4`@L(T|6y*s4rV0e^IGuktyU(ldIj1tO-(
z-e^(qYk{w|?sx+XZU&WJEQHoXDrb}_gfV@7etw^%BokaFoSHhSV|Jz=`C9uZ@-aXX
zr*aeT<KkDlWS4={;d(uHz6+4>9d60M&urB4T2lP2Xqh>$(M7h|S$*)JF_4u~;Ugq6
zJf@lPnU<8Fx(IPO#J?L_g_6C!eZk6ic498`pmApR{<pL#{5S2Gq&;FAeKQr9L924x
zZYhg4?ANdU6S5<$=)qZl1XV3*n>IeI#iqr7%KL-h#KYKFG5}{7Gbp@Pm#<-5@MVB$
z(tAabl$1mPLV~ZLDGo-+$iI28S)+cgVN89wQ6X%oQ+?KXe@}smq^DjA{^y){36#Zc
z`gx(D+gIklQ^U4@zu+irk_dt~CopR%@%bd|epqC`)!ZxZu*wwW{tgG2fVW}VGL6ms
z9XHWlih_A}x)rA!PDugARG`=CZb7`*Q$-p$f#2cL>HR@XwGCF%wsL6F{pDxKmO$-z
zZCR9IgUKTlx3siaB^>=arq4yeu$wxS$&{zY6ZGYy!4lkXSy$uR6rgmE&EaEcQ5_A)
z$ErP1;u{2dU8g?$6q{3blX11hjX)T}WafvWYIe>F|E7uPhzLK>E7LCP>PPqKQ@M_;
z9tUsmxU$T)hif4>J>4Hv{O_P7&QZHh(*nbj7?|t3<Ml$=AbwGsXYz`mH>n)qdrk3@
z#>Lo$lZ9C|Y@K2qRL}6Oo5~>y=TK(Sbw&Qbv8b)Bz1XV*3$&4#ynOv(xi3pFKZUq#
zXPwv>ab-D&ABl#zP0{yso0>Hgn4I{7v9~zf<7)*114bTihp+Wc$3ZmYB0ELqpvsHY
z@ySN3FZ>ys1Si3{#mO*O&KJw3lyO3BK>S75Zp_!~ksB;lIOR41c6}Q4M&s&Vo@Os`
z(|NRI6#%L{V+@aYcBS4rwz{clyQHL~qp<6mTfX)YEEM|sd9h&rwWzcrx&LWB6(I7)
z`wgWas|ko6k*!9>V|LXCwwA+0dK_s8Cx)d*(qO|#Lo(ca6ciYp^%~Oc{=RLPs!v{j
zs>iXK)crF@F3cBy@BCY4T5>-pKAuJNSLpNQGGs+rn8LwcRd)*RVGceBE%)O?lq;IH
z!(b^$9XrdYNB^}|*~6@UpWj8S0whpIRF(Xp-1F&$t0(_%7kmvGU<!`)dXCXIt6&9L
z{$T~)*lp3^j^giYZwXR&L(kak+k2eo;$WN7FSV(QQF9TQ1DWR9aZBL&zB}PT45zB!
zzTE)k^EeP9XAgLFLeAgm7{seY=^trA=B1U({DOB&>%upK45SpJmy%eUxla%IF5RVq
zcMGOAMK~`Z(tR>=K}#3uLyGL2nwv}e38J<dHrLFh=6ET8#F%zZynPFjpT^Sp(cGxb
z{R+mhbw?lg$W`^=k6|0r?HEME^}ya24J}shr$Oa>ji`Fo?C#j}u3}5tA(3~;gegCr
z92^c51nj5CV3~!L07>JzZ0TWP;X_!3aP{ZsXyIOTr(|^8In46g5{^&}j%--UV!AJK
z_<co*sQm(=NlSTIdsCyb2Gf{6{XDp3hB4e^4Sgz%RND@KbU$Yg=r10gfEXjaWxav8
z@4mbx+#OPK^p@}9QJt3*1<ZKA`ZIqce2H3H))JADfh8JVN4IS$Yd5F+?+(hSe0JdA
zVh>D_OgBbwA;sau7qxSAJd8QkCwrSyTiR5WhwbDxo>jtOQ&D$I4MSLeM5wGPWkiV@
zSkNPxPOi7VMo-e^(Dcph$Gm+_F$Szr^1WPB0{e2Q7E&Uwm}hr;pN}eejGzSyknM<5
z@DAI&Q}DJo=q8Fgus`uY3jq!DCR+81ZEN#^tt-kwLu|~`vb-^7n0#fK2i8B0gk}4B
zb?goy?QXmUXN`K8;giD#sr6$O7@mMUVR-K)2ug6D7n|ZF+C*Y(obdPRe;l@Dxi2q@
z{-eY-fBszFwoQF!&zh`MxLHEBJCo`Irsc>&g^gqo9%B#GyBbE8XZ#DK4J50&v2h!q
zs5$d0$j!68q4(q2QuxR0z&CdG^c4O4=(mcsB+)`>6$;F?Q}w2Ng;=z{i{0-cUJcUx
z5cAcP1Iq$e#daj+7fC;Ot2yxfJ(~Ifjm)VC!*dZ3fOkTkAI)t2!5Op*NG+e>l3B5Y
z{Dty;3yt%ok*SY_MvIdEc9EBc;uduZn73G+ZuCDMT6Y|^fbK{12S(Cti+SQ289LBJ
zH4?UvYTB(yx>jaV6e8YGo^N0C>d*S;pjE0^Iw)ohl#tpU{8c-7@)dCW`i|`4<4YEQ
z2I^?rYXh>(K&J2(e22Sv!sE|i!*1|AKB97O(bIJZ3lXw<p?XCO&pZZy)He1D0S)Ug
zUc5el6@^nfcJ5pYUnFR99*>{B6`p%guwOQO`X(%G7$c<%(kBGvT+B9DVr90#(8&Ye
za984JlyDX_#-Hz^%KMIW8l@&tbb@$Gcd#gH8&r&kQ1U_UzmE0-v}hZ<ZN1~Z`d0LE
zV_#jnr{4}@4k*zmGizjed)&wprkJmzqxGHb*jodN@VAjgvu4sSnIEy|n7ZHdc}xcj
zEx>TkK%)-5+#jaNEv5Gp5(JOfW8qhTbN3RviIn@VPl@=RLxu;tGSq8Y5(|s2v^;g2
z=E5*d&N0Y|DQS@_)CfD`@dp*+u&m^smVR5`=abg<6^l}Wg4;2qoZcYnKeq`+k?9JE
zTO!s<PYGMbc-T)e|HGR%De_WcQGP4v1W6ywfJc|dlRxza`ML;7=`w%((P99W433rq
zhPP)fs^~$-z6Ij8=%<RB+FRHUvmy*FQ~GYT!o~9%71o1oG+~AB5da-l{<;}inmTh$
z9Mx=}DdPP-HFcBN9A2idKjj^(9PEj@H;4+W{N{E-EQyQj7GAl0nG)3bv#ZcvL#{Qz
z@EkcRt<FSBD~S3v&(F{bmj-!>*F=~fiz_6s6_UGPY5Y|84~osp+9E*7HGKJGkij8~
z5BqQxIN*p84)EdXVgC8^XHK&j1<^>y_EjSr!b}ioRRk8-k#fv${>sbp@(h*)iX=Fy
z%RkbSrd86;h>p_tonjC&rQN3F^K()33nkow1%dG364Mc!eW{(^awXUuCM0R%$+vj^
z(}Xzm6~&=0r2DTBI!=Ac6*)fLX9d|pU*EBU_LjQQ>@{SH%;h5(zi<O@{5#-6oEQc{
z5M~qZ&C8hEN9tAc>#8G@;LlNB(jwvf_;h3x)fj_}hd74O%d1wWs(_OPLE}%{`ae5b
zTB=Zi9KtB0n#kP&^DG8_2T?cO&7eJt`+MsFF7GS|Z#=j73`=luk{79kxu#SL(JVzc
zQ&Ll=q0#UiY##f^V4Q3VZHK9w=x@>)p?i-8!6wpFwG%#qx2Ubxe>(sCMDC6o{C-f~
zlhITsEO!&WBpB9%va(a}fu}mu{x~^V2+DesG&uU9@qv_$peAU!PLnYK`&mV?=f(%R
z;_t;*?5dbP`gwH?onlf%=mrZcDky!h##Y`1hFOqclwj6NBF@X>iIJ}gq`Ipi^wA?W
zyp|A*CCvYM47L@bZbMRq3XdcUL2i$3rPon&KUU}RU$>^9cyquxUZUZJ=>u3fO^psJ
zc=M<rCv1X_vY(5K9bOvPXlg*_goTG|t`*f`#a0eETy8fB-lrg>e9@tbIa#`Y4(jGp
z0Li@dSwo2-KmsYLsf6IiNuVmB4TWP}lTt)G5Hkv%Uz-fY=ICdh=+<Qrx8Dct_Xg%0
z?S;isNLcuUnHiUi*WUmvt|Z|G4KT6=s~4OPVZ#ju@@>J-aRlYr;}2MRvGyJt0Z0(+
zLzyLR;-jliV2GB6VC$U^7K*5F)75s5fBpJOVr<PGw1sM$Z`IkqjRxRMvh$e}H#~6r
z|4hOilJ`lp=f7S6bqvQL?!X6kuy2G+*1$auw}s&T3gJ*<8e^jvHf-1suDAeRTL`|#
zrj2Cg(Q!D7o?Ks3uXO9ucihQN^mm!>?}wr%F?6_N+yG_EJ!xy;l3+{<fJx&)%nHt`
zV^QD%XzgS;3^%vG<cpBPh_lPv=&XJvTbOuAnhp*PT?RHfC6&_<v$BCqu3<+*Ya#Aw
z-jwx5o@40hYO5zXuJ^&cdp>xr-@)7Xp@k3j2p@#COL*0Btk2*iI1_(MQ!C=yGHfw>
zMnj5<*LW&~jdx+NgdjQy21?O`%8a3)kJ=xDci6afE1~Un=`h`DUO+2!t?d)EnPM$R
zW^*Yp1NnAj9uJD!u>1FY5qK}-*AQP@Nja*V-&q}C=Jd+04S!hE-F>N?C(O>5++ln=
zN6LtlO<T5HE`EtWbjJ2Xr0DYd^$+GQavV4i_(<uLrqn6gUAq)mK4`tc6EA>?FSTAp
zRkemlHfUyVx(epiWF0rdp-bz=ib_Ft&0|Jc0Xjx0ZnfA4&!p@CI(vlnSB9_W{Kl8s
zFuEpRY%uulUFtEZ$hqyBqRFhGu~R=mzQQc<7yH0#7Spukq57f6a5KsiQ_t1?`}Ze-
z`$XCkhR{n|%@EN4hrPE9t1@fb$G47-j$<nZ5+*8wNGTv?kcx<uAR!<PA|NF>ItDEQ
z(x8HLNQ;z?(j~bC2_>baq~UiibmsAS-*5lp{qX-V`#7F?r1rk=b+2{Bd7amJ?cwF!
zNz^+4#SwxDy8nEIE2atNNoX(@ZHUM>?%e52rqH&>eRySYWFjE)`SSu=n=8@z6WG&{
zzc%z^xiGbbxi}Q6GXk)C?|>NXUvL3e(c0FQ<n(eahsyB1PE1w<=L!&6VL(#+QXF$r
z&~afc%7CX*cZpNGu2h55^U)NTr{B~i2p?~Ppb@;t?`xi54x<zHEBs68VC~ICX9St0
z#`&flzY@L|two4C08~w6fop!@JNzuCZ+Hz2^Hub1`XAF+-oJpr@H0A(ztqQV$8#p4
zI$(6-;3JaVPUI<wO!13-#uCD--{9ezLKX~!?Kf}V{(>m?4)8G+7M81;nzwO37<&9G
z8lG1F_S-3F21Q9Vr3$gKu0ure78ew4Q{=wuK0MG!gDT3!;LXVUUqF_zckA~o1Er5*
zMwx>(OQeHyfdJ-`f4#8)Vo8h*WOAZ!Rf1!1y^s?j)n4(Px&VJyP%Ln$5BD-ZLiPJt
z9}yP5YXhOuLlQ!$Mz(5$pDHnmn{&6?37*OdN^)xIKI9LYQjA-+oB=w!qlg1CI+Yxl
zq+2E4201}OxDWh>y&8dFas{Qjwzgy5aotXivyU=P-(9mGNe*W&$Gdm$9zqK0A;5kw
z$Oj!ckDfjWP4u|MC}Nw(=-1mPKCcP@F2!pzx}SRx-seGXZf-L#Ic`Zaas7PCDc646
zmx~x5{N&H&XIAe)db=NGbySA`AS~>7{cFS0t5&VLYw3E1v|5G9V}EuTpUJ4KIR~Rl
z(_P=t&~0~p8!HTCWD@|k-}ON*WIJj1n3SWHfG5=1fvgo=^pW)1x1vzqz>h1FZ8SYR
z!-(RaRK*E^sFhk<Bmro~XhU}SM)x3|eUet!t%gmfi?O|&VPs1eKBALPrVJym_bf3M
zbp>2oMN(Y<_+y#j3~qc2k}qF+zh{f407U9?R;{P|R35h$R!X776uOK~vCID4h;?Rt
zi^ujtAn<WK0L|q7*!B0;)0a*QS%G005qId1LBRnKx@=Id(S=lNWr_Yupu=X13-)SV
z(D4cKLVl4|mj?H?59rm}mS3D86?!jqEdW!ACiru2#3^bP$k4nykJLzk%l0Ji2toyC
zO!W@G4w_D|HadU(y;>U@9+Sr1NPBPyKwDY2%)hDXufJ}m0!6nxR!VZGq>k;k`6e>Q
zZ7z<N`!XRaGHph$k%rX4gj(0cQ%&5D&`})YIbgAak+CW;3T;BCA2i#JAtYa8eB_)>
zHOf<?kW7v33{e?ZyC5a?{P6*MAqh5@rgJKSfk3cRx~n4=D$yC$ZEk7V&3{}Q>`He@
zPh-L`>x#bQx_md%)jBOc>#t8F%U}Vv5y^)MM<npx^}i}+3)iB8cJVVi8uea8#Ko{f
zu)<n(6##hF-q_NXB$x3|Fi_S)w%5EI*d{?NvDzszY>$!wNG0Wg2lwv1{geiTq+;|N
z@$}0EPNZ%`Lkkgrf<c!3@U6VKcsFvwE{pZpyzo-z@Vona!Kvf<cpm9BD3?rbhRZ2C
z(FAIEOG#DHM;EpOQ-`FF+Ji=dUe{&&q`~GC$C*zDmTbQlUubsrVtgxf1@7E-jD8=*
zVcvcIQkI!^ER#>J5Y6;2u4lx>SNEHjLLUM_$ymm|na>MXP#Jc7>4EsrH4x3*uBiBa
z<x>-Bd&1S&{iQBu3pbnMF0Zl^kl#Jnz>mVdo363x_QXY*TI2PnSNB3rOwz*;<FA-w
zHYjF;^5A@X(D$yT6PL)aep3oS-baah_e8uucwn|o0bP(r5YLs314YIQ{)sfriME28
zHO8FzT}NP!NRXJ`yty9*vQKp`e{KZ%Sr{F8W&KLZ4HKD>O{OV>0q7teow@hO$LB1n
zXu_o4!zblEe-QXgOkOQ=+g8h$b~nIp`&^!l`IN20r4W0MrI+e6K_e|;%jjc_a;e9;
z>c!~8^A!_9hf;4$Sbd(!dosX*FjcKcsUhzUj%d)Wavb|3y?r#h{^4h^elMT7@Oc@-
zT+Ti7Sa_r&=%D==KxBz8Nknh-o=v=)l($(oa`Q3@Dfc1c{0oQkeY_GUDMcWC07dOA
zo*4Ql?mI$y4a}@Z{E#uM3($8VK>(xU+#lfhQC?$xKDW2u#-#bz1qCYsu%N95Mq>rS
z-ZiIJV<VYP4yji5dOaWc_U-X`rNEGobr9kOBY87`a5tHcAWyr+u-qMmSR4jAZpSGX
zfms8I;17fsPar>vEc_H=K9o5<5FUL(mB%e7t#=3a8}wBj*u;qD@Ar7*HwQj!0+ewE
z-sC{fCxiv7H*63C^#B^Ft6?pdnwFIMUU}e6|AfRN2^A?1{rw*x8+eW8a_nyr@U*XL
zX#4?8khCTvJS>7+BJJ0RRF_e7_pg?!fowQLQOb;kQ2a7{CnOKd_V(-7!VO-tr%~K^
zhaPE2vm2I_|6bZYG*r1$M75ShJ&z7j_3(LzGmyNqHi)m&CjG>pzJ7jN_h6OX;aiD7
zh-hqIeSxa+UD}pQwGf@zj<zVGGj?s8XZyvPYGrMd?bk90+C?8T;NA|19K+doP`;GA
z%vmq&`tl>LY8^FVj)cHCE#N`#d}NHs?|yEs0C0=@c!UIG1FRwrW*68?pktWAhTbDg
zMwE=ULesuoeh7)E3rvQD&K!`yZAAGD4g=lZd%@!&{ksT3Q2ts;nL-N+`KZ>S?!iIU
z07(H9y3n9}3MZT3IHLQJ4aJd4Vf{UL|M-LiXofBWhmt%Yfm`FkctRt<AesRtAk@<>
z@6^!IiM$bLjl5MfVhDW_;iwGl;5FNiLZ?i!rU<xNr#Ox%U<=>~y~pyBo*e2(rXh7b
zax7@LQ7DAyhvY^Cy^9Dn{(8Vsk6(K}Jf9MqFXm2r7x>q?g)n&JJv=-+pdcy=jTZdg
zUfxxNu7x*tb9W!X5So$D+=SIKjz=PMzL-2YX4ZxvyerY_DZ~#^Va%!fA}8bM$BK!4
z&UoYlfJi=M;2{vzHnmM71W(92K$yJmLH*ODikyvLIi~T~3iAB10DAAAECciOPjn}|
zhSVY15x@U_#@d=^UlkId>PV?w7qYcQgpNQR5s@qrygJ6sErP1``y3l{l6N6dzlVgH
zWIiw*D^{)I+$#y#W-B(Xd~9?3B^38j0y@`v2WaqVa45igA;JW8@k(28bgdyQfOe7n
zJy)T}O4wznqwR~%K|jJ$!H*nfWLzYb1qqaZo`eSm@6`cZ<<U3}HOUE4_B)@hj}rnA
z4>}NAQ1+)dA%NM-#kG|jJ;+mC)zZ2H5vIKY0`%ytdWV{oDJoT2-_#{K`ul%JsmwW&
zbuTwJ4H8CqYoVCs=*y9^nS6Wr%-zLQt0)--dELL9+G1~Sj%NPra=D(|)@33a<sLp=
zxsnEnopXp<^#}ICcs<PlrZc-f#(3}0s$pkm)%T0@lhx#s`ID!^>Fj0Z9q()uHo6?!
zcfmRzGb^@ow=fUAP8j1iY+)R*trTRls7kms$qVrrsi2q8b?%5yt!(PgQ||NEg5alV
z1vj<s<dy5PgN8j7$mcpBeBn;S{j)tWF7zE*M8e)gbG+r*ZHt-nn>_`-T@JPzEy>oA
zC)0ZE1`X9+%Z_HVRPuDV@nFVvI5yV^vXCMihmPbT23I(l0?<<p#wV&4=c<q-G^cr7
z5vLv%5UPHOap8L^1tyz$(1oQ1(NELyYO!JE`sawTKszbKa8i+G(;|8l{#{gO<>ll$
zjhQA?dfFOki(2f)D%j0Bi&cBxrB(^<!uIw+q#<~L=kuRD<L}RO{H{KA8Ye~oVJstx
zyLK$9ueczr?3`Q4$r*#t5sJT4)}Q`ZWsrLU`lvj8(cvi!Z+Uxyw+JmxS9L(6w<XR&
z*<e8xbRh)Qt{omH6ANA5RLBEi0B=5CjOPB@otyI)CX&YJMMIibMe%kzhj||?-O}Ye
zeLy-sh~MkH*TJDZu6M8UYG}s>24-{AXXZ^L)IonXfEh7DoU#e}LG&wF0wkYb+8{7~
zF*j9tc$yD2Fj<lauA}2QHW)@lLNB|D{^rGS`x9xeE@{yQ_9iA%wf$5ZuC_HU#0x?3
zzE0v`h>v9s@CwI1%nyu~(GMOfm22epv>f>0HPbqoM;&PVrKdbVRyRLOEub$Uvo&^X
z5w(K2Te>^-Sh`!(-7DPkIq%)f^X@JG^cB4oQDg;m9EI{`>XVN1#OW~m&ick&im9r~
z#6ylREZ-z$AA)6!LG9?(o2fP0+OcxXxk?hXsx4YXKU(KGianSKd~Uu^G2dW(M^^AA
zt0>e^1u-7^lTH{XBFpphod<>V$4`3o2W+!^1>N*oH6^V{Djb7sQtmh9!uI0MQaKC>
zJ>{OSm7R~B^Ela`Gv=W}WK-|sE7v73h3C0an=6dvbG2K&&zTs<M$6cwXLl@qGg*8b
z3s+s;@S%L>h(f%0ze#Us>iWT~y_cYUkgzkR&lWo<A{2W!ZDEnr_P9Ci?2~ZRqtq~m
zvF=3Ph`Xw>dqtD4<Z~fIb^Ev@+)P2!RvC(slFtJz*JmK#=Ehc0*7|r;A(Hvs+)3Zk
z!BRO^)<rxtPg(3_jnsU`eWGsBR!sRVT?p#rn60KWePe2Dxzjl0nSCYgA@&?IR(aAS
zNoNub8|V7i9H-AU36-5=%T23vh>eX69QC>u;u~X-!Xe*6c8sc3OMQgz9o=$&nV!-!
z7LAHyJ&}4t@ij<U4R745R<Ucq*jhCM@Sdbkc5jj9XP>}?&H9%PUf2S_2c@M#__znP
zxW6?HT?g?x1~q=}e)B_YY)Mv;O)81!ai5nkP1Q1l-C_1)+lA~6)XxjUU3@At_-uLR
z2^&EHm~YX=w%+DDrS<;%{Eo;Djz<a2$23x-_X;v@Xu-i;$FPs_i0t!AVO7h8H2(Ti
zXFfk~u8S=S{r!)UFLPvZ$Uk-a#ZvLmk@W-60(w=a9~IINZyp7-Kjui{%V#A;&3;L$
zw;cA`y9-3gw#5N@dSe&;er#C9qXXwQjIYZFVL9*C?c1rdyqz=C&WrPHiwT^CD1k?|
z6t%|lJj5Mjwpd<K7RtP*Q`&ZT`?Qr{Xc)S*&NRh+X_x@^E8l0RKC2u)xqOYjb}aDj
zI=}q6t~#JpwEPNr3Ur6w&3;mde6)0KN8in;kG@;9E!on4P`W($aSn(4xw0+0ZaoiD
zr&%@Uh$s7?h{2Y{gMYO6R_0-5W*69<lW=TXeSLlPCr$znc!@diyMhDku>W0uKl*Yv
zC|MW7O9kjbzIP^85SV>y#C+g0+`sATWvjUK!+@H<`(*?FMJ?`oc6zwlW(%t7NP>xd
z+?&!CWHbI8+LHVeb+S?m`9nvq^vFQ!T*b72CK+`i9vuGKrYQJ91SNHsU*6DAvFY;F
zAM`c`T;(o%f%mz>EY4_Cr|;`;O!rLeI(F^sCCdM%kmf5e&TQF+dWW=azV`*b!wld<
zzWjV|SKwSo)m?n*)SC?s8xaI}npg53B%^v798gC0be#5cbVjaJh|(@2A|~S-v(gx7
z-a7d$y^)nItMk#JBS*@_)lma-1yoif?Ko2lB}psgS#TqmDq&i!G5n6hazUpFNT?=K
zJmLy=@LS(#ngu}4k_eGFGMlA6yKVk@c$>3b$O)l^5l>=OOfkZR>hBdLNNnL8)p1^t
zMLF%zdj7Ak;|tPH;9WyMKqNKc8-j7JUd{*&#6vDrHCxua$=qQPa3N}&bkX@?TE?3B
z4$PhQhiBI7kFrC3YPbl?P5or`jQ9gw%*(#mxWMMc*MI!dF`2(;kCJ9q$?VY1w{R>B
zy1}NRciPaEDO!N{&;$gjKk*SE5Bd;g+;*Kvd^3m;d#mGozk0`$JL?TDHE-IKKV4yx
z!uQLKJw0uk!ou#HXufd<XvWhwgC2Z+(bl-9+DeoDy^Gf`pDWA9lcA@om`ed<<pE4h
z2c3A;hMin`7n?D=dgOzarwHOqj?L$oO7ed{haOH9s}_r;ZGhcj7|&emkK<J!vR!p6
z`qs_PV`9SK{B>+YzK_I3MYXI-ThTVxeWIh`iitaZvp!o2F;+WfuzLiV<ndw-bxmL9
z#nGO70|<ch6K-b8k{^F;Td*W<GjFPZO@}q{Cde6_*za{;h6}*CEI`?~z14Qu9gUhg
z^l^yJuIElPfQEYEj-v^n=v|YowkE;Iq4XyOZhU%e-5sFYUV0?yyzPM12aWp$@`|c7
z!#QIY?!>8g)!+$U(Vb;n_<CQJbGvR({|$G;Uh3z-SQJ3Dv|c>s>wfTHs>pG!=rVHU
zxi4ZuE~xcIr9-bzBG|G?f22ofkv9~dM2^dyk3IdtG11WvW{0xb@P<MW_N>N;EYSCz
z^RhQG=owHv!{otr9Zmxuv;HD#LRz3x>f85}m6o<2QJ<E~nHwz|$sM*&Dk)-#F!T!z
zi&odx)&{Bkz~CT8RD{RIGGhQr08ej(cgi+JKC=Kvv1=xDRA_p=#Gse<?jREnG}1m#
zmj*SVJ*aI?CkFwS%_5y;AfJUnNKXxC6~DLF&Y@ZeK#xN}>aysJLCmxjq~6-eo@=&h
zvp2;^j3|pFp7ycit^<B5n~7Icfu@ps_tmV@=j(YsIq4)lYXU`gD#~#t${M57J0gRw
z8`PvfP%&c+&OIP)<j<e%F#2x1WXmolX`h2bi5C-Vw=K8PG;TOs_6KxOq4+kC=(vH7
z=S8(WR9bQkW2){w3KyU^E8dgsQ6C)Si3uLg@C?U^uY8Gx>_FV$g7TOcR)P-Jf*lr3
zr$vQ<Q|CNkXJF(EBZ*;h;M_)F<U|pSYoZ`!if7(1_4Wi(8P&xLJw2E-v3NPp&Ia_3
zV;J>SZQc9akKpl3f5YsHQ7kACa`O&_5Zana&WI>@iF*(LY(u8+iiJ5rf;l)gPZ&@U
zg!gg^3WaDt)ts9e(LcC1YbyRDxC%J<#ws_fMY137Oh`=hAbn_n3&I^27s!g!3WdL2
zf6_>CQEI+XyPw}>YHs^CNLnq0yqNaJ1M!#5_4W112rmw?vbv*Vqu}X%23nXqcRb<a
zw?1oD4eJ673ZhYzRk8kj*AyPO!61L6a96GJJ@YDms}omWJEdt(EllP!!i1XKuxYrt
z>Nv(eAd+?iEH0Y!t;ui%0YecIFM3c_U19sq!mdNy?J)C^BOU%)9?tk(qS9N_cd~tK
z)&Bn3#p9N;HzvnkCOYim#M@5|FD?vEVUP#|88ZcoVND&rjbu*@&3I&=uIKH1$TRe^
zLlcfrigrJ%TF%SLZ^S-W<&A};%}Nx(z-b|KvB>kQY;(m?Dxy(T*#v{4nER2d%jz;%
zoMZt_Ki}P6h(_ji1oNIGP{$~^{34(`#D7&|`=Ke>gwCqjl!W2Y(b3}Bix^hWFxz;-
z(u>$f85GLm1VXAZz^bYgoE_kq9Z2o)RaA2ZfL?@Cu!XEMf_x-`VlcJ5sDClQDR}}-
zI5BNyWZzTad@yv!!BA^W?-_$w`251R8-(b9)_4$KYF;I<9nISAWcjca$QwYha3d_T
zN4N(c^Wc5fB|kgika3^kmY(-aQ<mLy=+T_7e=!w_yagAVQBlQIM8&dEJNa74Bh~3o
zKn%3-I3O$E6CEG#2Gpyl<X~8i;@y^$Ay%z6?B?C&Q)6L^bkh7IXLI1^WdIUkKRKgI
zxDeQk7V-HHsu;I6&F4t=*xDe}E5ZbzEhKk{3cbbL3N!uzy6x`TS+%?eVO?n*=Q<cW
zkXw0PG0EjtgYULR97k$RrteR!5|Njer!y~Sms(f^5$hzwtuIflVHLC>JTroDw+Lu_
zcuES#dcmnr=#$d{l;ueMhOqGC1gCh<=`(nOb0eOP&AuH;ko^I5Y5P(<y=tx1f?E)C
z(DWn@DFa-N`-8~&LC8bIn}%{d)^(VOQ+EekP=FiWiDIS;EI6AW;tcR2Z0T;iUM7R9
z>R4`zt!Xgc<`bs6r>kw#rXqSP6^tU$r{+&~?=|OBR$idrWIB*sR7l<hj~IhHrQe}w
zq6K_}9IL5LZ|P?^Pt^z|i1&A65a3&Xa^OYUtjCny+Wgt!)DYTTY=broC$!rh%yqTl
zUPi(er^ECoy%S*`>SxFFJZ2aqTgFbNXVSi!&di3_e}@!u3o#6+_3?+=4rN(mt~z1D
zqhi)T;wm{gxwGJ~Xdw)MH2y;#%nhy0YfP=hm_{0C^~Ki<;#{HbbJ~06_RPSvy)g@8
zJhpVQy{yB1Mp&xs>ua9RT3xdONI@#d?~ozWBN6_P*dLGbsvFq^z}^TaQfi1e(wPA|
z6JqyAFi^43l`U@vIq5K<!><I>Ig1!z_KA;9(DYTYZ$RIi3xZO%k#f$ecNlNB%MsBy
zF^i}fZ6fau>Rr)J-C&sxjmefIb*?U5Omm2PIa+$812d;Qu3XPLuyo51-3&)v2Q&V{
z$erK=l9hOsVH1F=9K^{Ven;*WEnaR^bw`1v!wieJgM@Y%i&h9m1<Anm3vzoVQ+c;s
z=PE!SJsHG79$=lGV^4@x`<qpa=le}O`)@23S-c$Q_Njet?mn1Sd%-oyuItDM<T$9=
zQnHmnMxMMLJnavrAd%m1m-MtM|Eb{}F*?(;;Y9$!3{}iZ4rCYVA$Gh*oCmS#AeXse
zmx`gG`5DBfz|MI9B<z|uLvithzm_!E7VfY&g@9Nz-`rj~g1DuSoL3Y;M((oN?ZLnX
zS}}SHIn*sOBzn%sa&a9sDt+rwxGaksX|Odg1!C@!ufDlWybo$6OtEos;V?@uxG)$D
zpwEASPc%wRnjVx}LMCJ~VV(u>TOHWNORjL?TV{R%Xb6_7`SWOT08wEtDm+^V?m%Mb
zgxR;O?28`7tvrbcT0qgJCqXEoeC+e&PwBU7vtdnuxdOlB3D3-zaAM~5xsr+}#<q~;
z0LJXZ(fqkl<>40C34I#36F1i#4f&1H)x~8gNIQ2lE8@){ZD5YIRbyrT*!0{xOajvc
zJrV4WrX|CGa5@?34oQ9NzP?#i4QEAWh|!t8+r(~A7Y*{baMOl5(axibd5CX1U_(4J
zTW=vtzPt(h-wkaq?@&0>0&^F2<w#<ZeSCe1k0n9hdI1kp<%5570mk3&KwP?(Q7}h>
zdSchB{P!xo&0-de!2moOq(d*CuzAMMF5Brm`mUvG#-`+6;>Zax)v$1WQ^rRX+UaIb
zlePC8VP+nsZx=MI>Jod(8!T4Ai2#AmQ5T2Ue0se*9WA@&*^r($^j6JY!|y+c88@aJ
z0I=Dw+;bkAxv%0xTe+3>$}ar>d0!u&x*4T!{ZzKW=gDKAfod5Xn+(C`&pQ*b47hgF
zgwi?&exA0^7oT*tRqR`2Ldpw*n{1EGjfLr+o|9a(HxNBhr?Xma)=EvUPk4+6#^oO;
za@GL(@-$}0v^vZViaIiE^5W}{PaUQ26VxXlhr<QpEc4Dhqs${APB3LKMx4^NhvrW>
zruykw1Bg+P`!J*T=swn69DJH~)j%S0&o|AJ_jX+`wtB$z+Tz5w$EGS{*ou*blO{cJ
z*n}~jf-{HOK6~vO9WFmyH~Yy?#Y`fXx30?JTdLjtOlqg|U8&IG)<*}Ad-8k9Nrw7M
zu(C^+pLOt^?pjXKt3z<?Sl8$<m`?5H**vOEOdmo4<v~19R~W3$J$u)@3HKCK?ny~s
z%BtXNs?MCf+mCcuo5;E-%z8SEhchKTro7IRUXVJwl$FK}I{mV~xMb}JmoyNRW*~0q
zO{jb^s?XH;jc(jTkvN87EX0-?rZ4tNJGz``v*Vmx=CwWXZ8cpXU4RB1<sl1_V^gd(
z)j%CkZ8MhB=D`EIhmN`)v!5twTSw3C?Q{;<55P>5*@>UT)S8IK)0;3qt(*A6v5lLu
zz7_BDe1}w4IXnSXl94#|d1mQNGOHP|&y|f~aov-Ub=f?Kyn&N+*kLlet=@o4=g`7~
zwt2AH$;jI=6>f**9|q}Q#GWWEEycz6z8reh-;RhPzTTlx+wZ{4?W2@Ns{$IuI)jhN
zTK6^QR+!RzL~?W&({)gsI<o*%g~Ms3Hzj3t2lAgiBUqfz`PTjhMVZ!S@94YohuHe;
zG{T`Owu9+1Dk@{tuVRe$TNPC;<U{xF_3M)yb(k(w7e(~U@a*Zu0h`t|<B!KtH`@j0
zX5$ilOcj4d&YkjLK5{LCd9`1kbUk!D_>lxW7@Vo(o<UM^a8|;i9_W=L1Z&@(KP8gz
zjli=AKCns@2@U`d{`@v(3E$datCA5y^PfGNn6+L^cBlEZd!`g&4&VfbkLHeJSCDu*
zd+61X4V%0SvE6Pk(m6dA4Sm4R&e`caVdy(dp*%7GiD={w<J@i%sM8LTv~tO?fH}-B
zB6@Tnpfd!i7noxM{32)epl&M_kHLGQAHnHmz#O+5p?jInBh7dKMVA7Y6T`F(J301y
z2rW$Ui>wqrTAc4~Tr~%RTB|8Mb<bhinvvJPY<LGBWSCB!xO#wc?Je2(#GFN3ilH6E
zRKf?Tm3&8}WxRg>X?Ll>DXH;e112N9r5*I-9~BSTLjT_E?}x)|qiI&Mh4W14hhqhD
z6dO6kdVbb6*g1kxSEDxk;K3pAY#`1iSS6F3WINsEXShEN(7p#jZ#*~AL_T`7nbVuH
z-+%68mLOn0+N?Tcz0VM&*|<F@JRibWUp_|ZtjEJBu*QhvQTktPKJn32Xq0>HEL=t@
zLEsVh638+6{_e%fwUcj_>BV$9mmH8TUu&cJcB||%VUKi^w$?j4xEfK_(?Sd7^!i^Y
zPj=4_S>=~mt)@`6uBLw0Td%dA;?(Ws_zUIYcALFfPyCtRAZw#X^zHfJMJIhlH~(^q
zc-7w_-D;aH6$cl8b-IT5C7S)C8hq;Lj&*lqQOn}43|B~K-n^U=?b&~QfVG2%q8HHV
zJiuzR_jO)bNa~6!FRq@-JM(1a4Tfcuh@RNE#8Ru(X_QeDiuk=V8wBc%YzNX&k$Z=!
zu(%_N@TnUu?z!4@t0=7A{9Z^qSA4Ld={ZUH`Q@Z?@%+l4zp?)|rcD0y+Pw*@@YB~6
z=O_Mae*Ri~1)khbFX6q1BA5U4HD&WDOgsGP^>8)jiT?C*I{WV>P$>VuSOq;;mJ}q9
z#_Wi;?e62b+ZrP)SeKWtSG_ljjUpUshgfZT{nOnon1-4R((R95!k?_mTO?^Osa16z
zqNO<fh4eCHc;l#K=xWNvOrsuW);#|{AhNC`Q*G0rP=p)sNm4@?^m(cgiX>n*uz&w{
zf?DpY>fcf+6mbTdy%b@Gt^fCX_RnW>%)X^o8*NF5N_y*-?t?;E*$<$`h2UK2O(;H&
zd~jTxAKNHXR);8*q$zoCi=9y+c?Y{eP8em@AJnoU_98mjxcU1+a~i>L1Mq1SCwnEx
zTZIG4YSS3-T}soCT-)8Renu9~^v%lcbp}u>bwGjm(1uORDARBfQGjO+Z)V@lH&<6s
zc--K)m7J02D|nxNqhaLSP+mD$RQ-OlkB<)z=43`chxw@dh8dj2jLRr`UhpTr!yA=z
zP*tZkwkBLT(Xp2DM6F<Y{ZB^?5#siLk26-`U!#Uy@$V{-qw()5{C`}9v{G0hyo9?k
z&6b;#debuOX9qL;jnu&kfH_aDo(CSUg*a8pWdntx$NQK~odG?-`r#iFFv7O-8$pai
zF=02&!=iZ4{j-R*z1uOMdrcY2>ROo9p>I4}xXYxjapAAu+0LRq8|Mj-Q52|4)5tQ)
z-K0al*CUs%V5=L^MFE^orfe$-HB60kbN~q{pD~4BwXZ!K6}%boY4MCHdamgSJ_#GC
zJAWCMzF>zc4{WO^cqH}>p4`B*FTc8}-+&=|16N$e=$GF9p$!cU4aT}f1k)KJXW`|A
zNqu30itAOd_0}Mk*}V*7c1hjgV38UA@{z>%fG6!DAcoQzcKlEa8K;sE`McpjqyQwl
z=r1ZM*s+R2`Ea6h?;_jLqaKLduFdlxy%YQU&ygUzz`kFb*l{q-&I~*y_BhNXp?ovO
z^_cd>MCnZI0AAxtkdW^`pa&@B3akh}0GjO#R8V(f2j20Y3&P`3M<)|+T+uO3ehOuy
z*L?CRV1}3-Onwo+84hp+>C5AKt(Qdzke`2#^^{*2wXVjTde7PmCs_^W@r>?v_{d%;
z=_`E+z^z=ljr{51ph=#;Z?&J1@*(r^QXJ^C4O_vq8bN#fxT?H74X(IM5`Q49(TMqr
z?QWcv#gc9CscztU>pVB1P_DLFl(QF+nXoA}lW4>UM^dTOyNmK+^6xi{hb(WTbXcfw
zUMlat{S?YY>lAKLOoY7U^Z4<Pw>Xabk00-6SVh@?MFv}^>m-78JSp3S8ESqZB$POs
zbP;%K!l#27*YEc2cf|6ERf4lD1T+Hopj0oPAaQ_cbkd|Llqbh>kYSplFs7YWMT%=c
znn<N8e119Q?%UF(2>AV9^tJ=Zra*Ue)2fgx9*#g9Y;XvV7GL1c2#6Og6VA~Dq~;<I
z#O)aYPF3oK;Q?S5?f(4TGYc<=_NYMkJXq>S!sii!v7U)6S_2^m4~TBMeftg9`S`pi
zXLARI;&LfQj=AmB(k5-3z#(~u!YxOQA}W#XKwsS2HO;y42I-`3G=kFJwTL97>3pCv
z2YTh<X#6!^d2;w2I3<h$)}2|%)JO1dG4v(>y7w2#8sDg;?Z0~&0e^XruyfW>fro6p
zgtwdsM5RP@hIi`I9UAgA!8A!lgQJFuFok^UdyK#PF#4|2r6;gs(#{i9f}^G?KsPkN
z%Yl4+4<64$36Yrd@iw?E3T0r7z&R+N{i11d^f!#85W>9qyK6;}proA@W3agf(wbe;
zjP*#t_Kx>wHdI_Zi*2+g4n-v3&ILFlH3Nd6!R^GpvP6`0?HC8;$wsAsrF3QU-?(LN
zpBCn!?sP%)rdRUKGGh|h!_t$o>;)bdeB-a*Z0n1l(M4`z)6?(!f3vm;qaG4VoqKz6
zv&t#xC=r7hN}X)`n2e3{eso}v=MB#<ln*h70AP*MuON5xa2LS!YCO0pz>z^Y!=Qf9
zlX}p$XV@0X&tP?mKnbbtoOf|`U8>)a<GKgQ*OQ9`Q6tqBQWWr>6u>Hzx=hJwIoxgv
z$5J-6UmSIFEfm%&ny7B>SO-}ytF`Pp<vZ?TFAB5%uvj{Swt!W1cxWk49{xBzf9hBM
z;!f&aKx0_Wk_RJ4FesqACv1ea@Z!QJ^oOj;B%z%iD=3GqiC~ZU4(}#+<kSp$vH491
z*FzdM^qU~s2t5HM9CMYys<2l7u(RT}vv@PYE?YXeZdm?9atNkPA-A!QutvTK5()yU
z3p(t~lqc^m|9C)=$l8gOWg65f6vZm=2J#us;ruUcunA@q`@q0q8tlNPd2#T2&iPPS
z%H~NjDPMux#v;0OGdNAvz(CS6Au(jREu1{B+mX#3<QNhQ%Z{bk_+u$n5dI1&b?6)0
zqIe?)vT+LCCOps0gol*Nah(*|fu5I{k&X5gAh-%aTN8|n+;(n(L94{~g`diNPoRIR
zgWZgdjg>U~7+L-m*5*aj&7!Ti8W}aU&@Mf!-f2Y@6*{c-Im0WHJQSx_-}7XrKP7v9
z)dY|me5)^p_W{|KF<>hAG#F$um|-K;_*PCmMU?~tN`2_?ytdN*xc7@>`B5A^@v?%H
zHqgf+hD?IODuV+Yexw7yIy-a^9)>?4A8qx+q|!EbM|fXwfdlM0v=^J=?`vQN%mORL
z{n9<)^D>Ilc|9W|QjJB4wL*pqH-GLJ4C=3KR&a08(mxwip>1qxbRb^FkL5yzizG(R
zaIw4|c<OrkC@Q8Bt)qD5^w`Qtbg%D#k*P2xJ9LjNImIvDu8$kb{mGsdVJqDg{_X%*
z_5h820P+=An?~9WbR{_~0im{}ng}g&k~d-3VbwKJyh(35Cn@<1un|eWD2Hy6(jzJu
z5el}*@WN4`uz7HZ7nY9}L5Ee|TZ)s8K7rBjdJ+RR)Mi4|%6qFf0_M>+{6(1++@bLA
znOa5bwcsdshhxWI5BAa~PaU|=d!v67E3$U~qDB8ltMGR)jkil%+tt-o4d@m5{e$^$
zAKA#3gtDXusR6;Vj?pua52gGAvpumPu>?H(t+}^Cb7$x*>SSbaMtUq^xk*|3a)yyo
zGk^@!DuGFIK+Wh_|Hm6R!?G^Xy3qc>kuXTAT0nc<Kzut!EbCN-iNh8W4H+h8F%6*t
zD*pVEl83*&4DfPrv!Aq1LNaG3PF#h?*$9;F(%uX&;B(pmHCJlDbi$%?{e$_5m+iSv
zw|Q|)XmJZF)6HrVvT|};Q5LqN1Ku){v^bMA;vF`(|IDUK;wYkTM-j|0ceo0{(Dm%W
z3_8)Xd>Avil>o`P!;2{Sxs%e|ePRkI!nOZd5*gytg08<hOz3gd4p7YHY+_uxYpw{M
zHuBr#^}^&NN?IN0tE{r#^Btt@4D>}rzx0ZxVqjA{HGjckD-F%_+$wl9Ez%^g_#u1j
zMQq>1;v$>Q#udv}oi?lF-M6*ia8I3C^tfJRH-A?$S8ZuuoQ^@aPd$4@vZoklk(s!(
za$kv(d1$GP_*oeUMR-ljcBssFf7oXv+)`C!G1&iT!XbZnR?Sv(@W7-R<dD{^U$5Y`
zS7hss9saKJe*>ylcLn}^b>tG6e2=OsBRo!z@q*mEEJjW~YDiR#BaNXu*{)>7_feCQ
zkU&JE>}6(QQQ`jRpBsOmgFxf&my-lm%R4_mzkv6ih8`-)v_cL!;MQqbwnE0p0z+&n
z!i3b}up3Zuu7l<n9rX*^dH0IApvyrVgX<(OTxft4*VbLT;-G6!S5)Qk`nB-gyLV%R
z|5%RNZ{Sh29KDgelB1ADS-EP}5&Y5JH8eFSY+FFpis)h9)9B>M^Yimtv;Am1<jyc7
ztE{Z-9JCoF#l<gbYCc03gKuc)kdr{kNvMrYPGY3-In2@{?+ux=fG7(h1i{562lB_C
z2$Q$c(<fl)$^CbhrlvAaPv4aT`;rD+Pc*BUSy@q0SnzrH@Cp1mqJqjjS)(pNK|z?n
zKto3tj|)F=-~b5)@#e~jIT_y<>8o}4_b}!Amk&kc<b04#8e6C#tOx7ZE?X1vh5|TI
z_b{q622BguhH)34-UcE68KhQ`P$)zCrIMrQ*jjpCv$G(h1QC%-)Fs~x@uLJ&58k=H
zr5$2ANeKzB{bcP>59n*oN-cfN8l(?AFEb6Q`BMJJP9Z2s$@W-PpQw5TyHOTBBXX$R
z+1$FN6R?GmF$siI{l*k+<vhDA6r41;hmIuWJcF>V?kk$Zm)^ApV!9{cc%v><4bn$C
zgsfbpRPyrjT-@Ap7R`}$vlb?)WTl`#NzZZ){j~~{7{@9FsJ8xhwwx#AUNEbbsV79p
zQ4I@NhY~u%TOlR4>B%YZCr+GX*s>)Yk;%Ckb#GZX(g~0b`U|$Jmydp+hYB26D>Y!7
zUpL=(HOs6ETP^&&=am2s?el_c3^yTMMvoy*=v{jc+MfkFc3sEd+K*t{NT8AJVL=vT
z*taoe!W4Z0XJVU{_$8Fpmp%amMOmyySATzy_5}ljSh(v%Fpu=Gi(Q<ZlWDhqe{MI}
z=|(J;k~NqMhCmK&7m^IMX_KFNrlXA-FzkhGJ$CaG)LR|yCq|?9#z6LfO!5JNW*?*o
zvKtt9ZU`gZ!H~i@BQ;J;fN>2I2))&JI7HB18TniTYH3{^9a`+68at5cr(iIkj`_Mt
zB0t9HFr^Wxv;PY8ntH+NtOmJ5FZoo_CP%&-DEzJ4aZDHs%EhaOR(^WuK@?$LPh-k+
zM9d~;<q+0JI981zJaHV_r5|Xlu<girHjJ`2vPcXI3q$u_g$EmCnN(jD@-u%z$U>1g
zdrI!63>J5M*m3a+MvZ6Cj$vI5@TO1?(6Jc8>T^L0Aa0CBLRuPq2&vqdNuN5q8OQk3
zr%!ZUS-M~_+(Kkj?%{V>Ry53R)ER_d$KsZjZ^Nt6kkbI}Dw{ilfHHWV9*2nibXtiX
z${IwMW7%sdl#~MQe56PJy!chnp^h+lm#aC{sPJK5`D4F0D!y(?)%Ap(HniI3yZ`R}
z`~C0+b$;p&`V+AN8R!?U&1`dsLyi2ham1qq>L+j*hxKq_$9td(S&yrp0R6YAMcl$<
z{^6c8TWM(xCY5c`mUMyaBUE{P`m<OtgAb$G;<(VxnY}A!|6vY}OJH-H%Lr^e@9g4I
zgUsB>AmUNP<^z&?D7|h=a_JU_U_oD7em=~nk1lV8ep_BH64E5_4IAsl4@j(Bv-LfU
zVVeD{C6oi1t=251e3aK(fBbA%g_%%2@y1ld0$${S@ZFiC5usoU8A37WI5&&ZSe4^~
zELezsafX$P*^u|^?Q2SpYw2&b>7fTQQP!8UeC3)iSPB^sHdAEP@}@ydOtr}8ISr#m
zKqRa-vzRH`-zOS%!y4<SbR@xxNQdkbn#dc|6Vdxm5_J!=k8yG_mz2@FTTgYkzVp~s
zC(_yP=(tE%5Ce@c%i-1paHjd9IlezC{_A=_m+-D#Lh(4x9QNuyfArZ~D8f$kMuqWl
z>A<RSVb~2_kv<Tfj0Q|t?gQ^Cov#PoE}1dC%$aCV>;)dsl<$~N6oo-H???O6D~wsK
zPlsBd!XL|5jy@?}41*RRh%QfIOO%8CFj@rtS%@Z7UpdUbD#!xmP!UCtR8?6lCe6l;
z4-?Yaa=zZi<9wWuj!t?gsQV*>cn%ZQ3ws4Y<;Z#qtX}OdlN_&vVBQf3IQ4_#^@490
zMA9a32eA8@-X1HjJaQ@GfkL86@55iu|G067|8e7BBe`}UpK8jsGD)P$JP}@I57S^~
zYMRV9i91<(Gm!0MQ>(D`#=#}ydv;#plBj6me77&u1mni8q#J0D7>MA`GpG9`;n4Un
z5t|q%2r7I*KyjQp{{$8pB`;;977Eah?d<Q5gHFU@%<zLfRYWh791zKszy2Bv)5K>#
zlh$}y#AR#A*bxK#SQ;&o94ABJ%uvHReC*gowDh^SxLm4!CNcpMma-Ao4EWA+(HHmJ
zud{})g^4@|ZN?Y+4nRwd0<E9{KqjKXcPreu@dBNAa?d4cfWyWBP`L%<0MSP@7TN%<
zo7;AI!B#S(cQ0s^?{;QpCJ+KG_8vG&d62$w2qZyYkRgLho}K+I;d1g{1A$;DE0DbE
zx(|#2UQS{vtswx+{(13l#(WMI5A|0!B@k6qPVk`>F#@e*O2K1j<|JLwP;;G{H--Y~
z47h4&QM#Ji4_%mXFka(Gnwe-02XG6hCU}~T5;N-N#W1ST8w96b575GEka?|O9aeFm
zBQyb>0ZJ9Nb5o}7J2^$Mb)n91DKZ)S*;wqHKFsfd0LfME1_;lO!(Sa@Vk+!whCUA1
zxtG!T5Qo0uSXj@XU@BOm_BaHSju>oI1NQUKq6&HpFr&7_C={OWiqYC^ux95SXQ!t0
zK(p29VLEt_^d*4QsvbFr?!+0uY!=Yn1a2;m>BdS8GqbbgW{(Rnbz`=a0Z5Q=-_!Ft
z`D6tV&B?=|S0ti4B(dNnL?h&I^-E6I=;5m#@DgZopaFSAf}m8~8I1KKn+FL*UOw*3
zcrxilS8s0&9#l{TrtKvNEzZ>rOa*Z2a?JBzeNAy4=0(Rz2%ptZu=XNS_y~aShOi>Q
z^Zm{z+N0rv3-ix|Xz9Mkj-U8d)?TTM{1HfQ@Pb{|L(jytDu4^h*AMs<KWOmiTh2iN
zLlL{vEY}13R>_D9bm#@J8<Y<V2()xg{ct_5yy*F*p{F-CGlH76f{?Ah;rPP{81F#i
z(T5SJn@U4ms5k`^o4+Z)x*1cD1Aa<45`YM_q`Qv6?$tv*1I2`f0f+8)04M8kPQA6z
z#R}DI9K(AhBZ?SQPwubr0|KQQL_Ef^Z<XWmH1q*SZPP+)w<4N<kMUycT^!~V=-33$
z&&~Px`|Gh^hk_BJ>TsOYYde46uz?E7F>4~mZl=e`lrn%LO+`dp?vW3FL>zrdGG^ww
zRz1hU#)8X!oN5O3)6{j2(EE`CCyiB&Hy$>a&%E2M@yLhV>F%I_z!`E7ctml;(-6eZ
z9sPykWc2wvmi+yQIGt~9VO>--Q&aJc1zT*}t{4su4lX>jRCl>Y<Dfr7o#Q%GJ$xE@
z(U4gWm8TSWEes#QF}jvft&Ov>Ehxri8>@ktE!-Mww}8!Pgp|8n&EilLAj~Tvx*mue
zaOz=mj;O1vlNgeA^4quL*~fS97KW8c?ln;4tq11k+JRC=3au7EX`P@LMCnH9BPGFj
z_Wnk2o5Yr(g;Nfq9#S6I^&je9OjtqBbIrPS^1{MSb{Hzq(XmfakviLP{rdH7RKFvN
zm5FwZ@-P&xiWkU8(gIOt39<trhx81IY*VK-G7|kCPDq%E>t_lIM{+}iZ1lyaDoLLa
z^`-LUL3AfXz_y`17<&4UED}c}*s+sQFo=!UBO&e>kf@%2-Es~X**5B`6)XIkNms0E
zKd>pzAljIea4am@>vhjWeuuQ3JVH!fm+i>94pzuPKk3*>_!e$oHuYAcw>1Jd7pXkt
z@3Vtq*1mlYKAo0nCA4B$8@6rScr`ts^zGX-Scx3S13X7rAr@nzM@L4I;~G}Mfa};M
z2*e9eJ32ZND!o|pbs;w*j8$5n|LR9{%Aoo`ZnSRqehFgiCurljN6^{*F%nA710^yH
z=SbXV+T28AQLb8FqAUe(na&eRk83BX8<LJky=z2yeYd+?J?ff`7x&tW@Kx5s{PL^6
z0(hYwS`TRtmdcW$vCdF*V@MHDPQkvrv%W(i$J*4j1MqAuf*UiReb@%b)Goo~dbEGv
zlAlISasEW-+3iiLGjwdqY0;8Chcwb9_8%m>!<3(ouPKtJp<Ag;s|2#m`lkHc-1S5L
zNZz&u>B8_DwOU;vtBvqp_VQ)_GaZtTvTcVJl}>{`4EV#h-zH^DCHWhjSIQ)RmM)GV
zEtu#T)FAKeo<TmT)UaHezdiM90gqX}!{0BC{P(pMQTJBpfw1;&MZ*`C+@ZjSgW4|#
z^&l!YmvY?B$Y^3Z2i(-MC`~~cXS{_yq3a4<SR&V@E*irs8nNHju1&LbPw#o)liTCG
ze**M=McSCrnzwTwewe_Kf`2dq@`(XPXxQotMsI}yu0BiQE!-)K$L#Fv$E^o<?|$7*
z@*?BJ-eCy4v3lrUyY_*849Et@+B10gOhv!bFz<&`jmt9T#=(Z^tL)1`5e_?A0Y3VX
z#*WEM6snXDNc!9mXKkd}zFl`j(0r?+;%t6W!{egEC#$x=(UHzog`SpNTSfA8!AY?I
znSS=8DqLl=c~xTwwK_hxlK#L+&+ipY3WNH8@3d*STMGcj@`);q1Hqt@#GoqDgoB@0
zw{h1lL(`l~&n1r+U_`N|$7p9dTnqEjqZgpo!&HPG5*m!eL`uoCUmZ<SEFKsgCx!}4
z24JhlG0uIV@P)nTNVLM$BZp5!SOmeF;!bNjyBth(f7ro~zbZ9dD+cJ_8A*7BM_ZP;
z7)r7YaOuV=8YR$SkBN!NQhFv8pSu8=&}d*`-8|ATTOL|u`6-nDxhGxGBVqu5f<z9X
z!I}*l0KBVYjBN!1;Oy+IlmRv2ckQnLvKJ$>1;sI)CkY72wj$Cukz34fiLo6xo396~
z5p~lwoTu5eNyC!m_;CeTKUrLLBQFv~3vqg03yeI$T32;m#&?l0`)97~N<mO}g^m$A
z+jz#hmU9}nnppyl8wY~a=qFbH{d!+3-Jm!B{!j8>;cYOk95y0Bt9%d1oJI~PCocY=
zW#pXCp{MXLgip=In}=Zkc|?T2>LSSg#OiM|m&5Ua>qz0zMb)>#wL%Mh@s2DWR{af9
z*nk<lp@^!{u$yAEVa=LH_;;fP8Y&%Gk6%KZ9t#I}oH}BK&~{r99G;QK1a?3W{SldG
ze?rID*hAYTD^9J;kHlhhAO@!wj9|I0oy^?Fn29WgnU}Y5z#6dy`9-8R%`MZR8sBf|
zn&^L-2UlpZ*rOt<&;#uiJ<RWdwsVk*YQp07n;0Si8h!#i`WYZWexM4JkSP)k$e39Q
zXArbcVCx3ZK?|f7z1X;JmE2qM1pH7oSDQrusJ-J}qbW|#JaVGGn&+;Kv(k?+<^{dO
zS1|oVoU|`IHZ@e7fEP;VA@kFG56c#|)15o7>NRi*a#P2PV5NNvMj$rPI6gO*jf!p`
zd`}IUzy{QUg3LxgOhb7b0bFCH@t~>eg9k6B^B|Fw{$Vv0H3uVA`EracL~=rfD`k8o
z%JcpHJB9KbP}TqYb+raSMLkm9bfd<a_8i0?hG-F*L5mF;J382!dxc0&n3<(ufP)-~
zB9qiqk><!E;4`iivK+Xau3uX?U=6#1e=xDCzLLXu7H_~|UBkmk<ecKqnJ>S1QZR);
z>Ef+huPP=wePPe_;rshQ8*k)=$yI)F^^aD^Mf-Wg&_t>UEySH29pNNUL?R5Ad93lQ
zg*5gsNV?kJBu`^;3_#y-K&r!{#g;q%2rUWcX03*Zo+6eKlhPnCZGgv*TuUeN*Hq}%
z%;SMfItt>=6_V5vkm&}w)k7N<yo~dR-Q<mo;tO*A_~Vbh_fPMrInG<u$H<C38V4`D
ztf8fJ3tA`r3P#YxC5FvkUUL&k-V?}ML<4AQthC2?jFudmS|e}RAm217Te7O>sUYtQ
zFc6oJD5-gkX%9QnjXySA(|UqE<S;iPhSA!pXqre<z5<kvd>l-h46K#q(9ZAAC3t1U
z4-jbSYwxvhFT%#AK`{j<s=jXy@Ey$l6=Y<V;tR;#4DmP&fB-b|LcV}3YeC^Y*~&1X
zBxuv{MIGo4M8+YEb3qbXmdZH5XV4@G9^{Ae2**b!j1Y`OvZ4=dlbONHwws721kkjF
z+2$ejzn;882Q&`naCCf}E65R{G%epvy~oK=>%C?Z>b`^hG#Up_>>37^t^hMJ4#v42
zST3(Hy#nI434kkUj;dTn@*#tJMc?aAS4A${9sUM4Y%d~&vVaPu{aD2bk~iXP$RUxR
z=&col{i-lsK-hLUN3)LLgNkZ42g<&oQip3d3&_BEl*Zg01r$Wj7miL&QjwY8{PcPm
z%#9>EA>>t5f+6@#;8Ls<NwQE3hl1T@4??3LugcG#1!sRc2E~e)r9dud0$3rklf6}A
zlYRB)kS$)q|LC@jVaFKMM!F+1pxahk4K=O=B<X!f<{kzGT}kW;;5CnI)M1vuBFTXm
zf}yuU-W`tvpT@(|d!T&x3^W66!}&IXts<O%kKS<;Z$;im3sxdY$Z4XHL#;0fP)BML
zg@&y|92_+R!HAnN-d3Nf$ZYhbL=CYz$7LJi(A^XZ+_qrS|LfNq0HaQ_fUQAZ!G4Ag
zKjO5dudQEVkmVOdpah-@>dC^ecwmA@LU>I7+A~7L))Q4v!VZCXVO9&%0`;U!72_G@
zY|8{BLB0v;$m7aK>T~JrEneVp^uo&5LDq?mc2Z*;DLcyD6+;&HG@_A8@Rnl8pdCbx
zPl~ZK-0W{)1?nS-HV>=IKVXa3bpKTqPym1(gGwj?`qncF4m|Wn(|9&cvtryBpt&eo
zZUa3S6yD11n>IZ~aK<ZUI$DF&Rvs!z^w2f^$c~+5oIKK%PF1VLgBf(mf>1m+CXgZ^
zDYaC6l!ozCNQ-&I*cXQ4lJcPOGeNpz@4w?*dj^#tQ*jotrY!R)0O+y6HiD=?O2l9V
zDn<-@0YuYsTod=uoKwAdGYvgAB)QbSc8zDh4s3)hSHt%Z>$Tv2#V1A?FqQ&i<(%Y}
z0A0l*FVAj(Uy?`HrwqbXa_Poxc`cC^fFf_fZoy|ig}P;B;;+uw%>7_b$UuV@u$z`*
z?uI)skTEcxWXB!9-mGUHUIIY`Zm1#=7z37o<0RYR!xG3|i8F{73`IgY11}>Fl2}0@
zI@=h#ku->)B3@#u)o_QvXF!PwbW8xDmVU_RCL%8Ahu+IYOgf!FqQ(<{Tnq||&TxNm
zc+AQn*>LFf30D+>3tEZ~fte7Eiz27`OuNJ-W#vF%EwSiJmBjPO!fqN=4)gMQ3VhiB
zTNeFdUI>2%JiafMQ&E_!#tu=3?ccwod?0KyVnrPM$ajPvMMXv7$Zf@v5MYUg<veu0
zeDF=oj`9U`1oKVK4K`#QCTq}})dv7|#Owg&SO5-)>Yt(_vvS3XL%h67IK}UAXX&<+
z*HM`?d4=&o8D=N}3M2mGI@y|Ku8Xx-*D1Pl`*s>;0i|{j3>Jv^VB!&^5`dubuIX|}
zG6<e^@#*PT_OlrnOhW=GJU&&-O!CQf+<Xyu71`r3n(}&j(a?_vaK$YmCT55%c4oRK
z%=htQOGgJpXLad@im4+c-5v8SU$c!R8szd*+y-?Q;r`%7c0<w4+Z;$@9LQx&IqFDz
z6iB@UX7KKxe;xu}hVdw`U_waEX#4gkJd8fzm3)vZgp9LBtvzyva9l+IeP$OlSOznR
zCr|BY2-Tg0fR!xC^kU^=AeFouvDp{sy7M8y2BH`c<Mke)B_L+k#>?dJRfJ_@(oh{d
zA}QM-{=JC2c+J*BySH!mV=-UE+-!@X=ElgElbVWg3Q`0}L#Bz*ebKP*=&8dPNeM)*
zE*>5Y*kw0?+6}g3D?*6#co8wO5ACs9DZQ=;?yYxJ|LEs%5?lY1lM`Neiu%>pL*^C3
zKjCjnhK&4YH{1LFZUg!Llh=p;{q{d=^y44@uF3yg5#-$e?vwwuDF3d>|5}p&t9|ma
z$Lfcx6TaR{O60yr{Jz(4u@&9WVy}-;Lpnup7I6svIyFRaB`-;(A`l#5;uM|U{$nMJ
z6yWcH{kanOb?K#paiG-(ob`aHM-zABTg_IO8$lv0&jZxCqu+n&zyCYB{3lcUpC1MP
z`|bZM#*csfyC(l@Nhm+9$-@l*B@TdSkoEHAA%1@T{IXL=*Zlr__TuF$*+}$03{3&s
zsoZ>U|NgV!2xYxEK{3+=punU)i-fm)xMm;&Nv!uQI08O%N7q^J7p<fj{hJFA%x57<
z%1vr#OmZg_9Q+=$(s@#NYmw!$B<%tG53$seL>_#ZF`bI(X|9%-7<A5d#qK22(2^`u
zmS6e#7^2M2P$>L;&12dB+{HryoHJy0K`BO#P`HqPBK(J4^y(Wqm}_L*nxh23rUp{H
z7Km9uIzEiaHJB8#X7%ce2<&1NlM)fTeL-&fI5X2KV@wrOb<m&D;J7%ePA}x30>&s|
z{~%87f}|_vU){ZTFBUN@DRB~R31F0_t=Ry-X-`}a#$@XN{aI9i>rvCAs}J$RydT4q
z&-<`czx=wk_`b`h^Q4A{aIP1%DJmpEZesLtK~Yf|LNyRz)~sDy2LTR%49qbQSs_6-
z;In~B^T{Dq5^iDEKsEUzz@(~qc1no8B&DQ$!J2}Wusq79Fc?JTR~8f+V-NL26<!3b
z^62v~O9OMGmlPNgD{5L*OyBL;#B--vkYV~uxM$jp&!?Y9MYXj(4i1hd%m7qwvy#u<
zcme>_0Pi+~wYiBA7plm05CIZW7=A<k>FJdKlr8~Yk%RQ+XtoN)bEiiErHBZHA!AEN
zNraGmQ0je;pc`bT-4bNY37?FCArV_F2H6HPK>LajM!%(Jy6t<3mhx#6r_JcKGmq*&
zqJA-&Ef)?xQGfv=;u{Wrf8KwGI=c0u4N*erwBjumK@}{sSK!>EO>}fd=FI@wNi{`D
zb1QZ#W)Ww+{07m;^|l)ZVWb;(5J<+x@vEM=loU6fgaO1wY6v$T43b+r5c9GV^jB~L
za0)Cs<bnvhy%0J?T4Vh^tqI5$3Q_BHM@E@#rj}(^@P(Qb_IYUNx>Y4AG&`X#7A0ka
zmVHBxLZEmClk-061Qal1WV#TnXEE63>Av6qVq;3oJunXf-QNt?7gYgmcLO}V6Q?D4
zc4Ft&t=HTbg<1hCB?lfv0#3xvN16<ib|Oo#?1@#+&m#jQgC_e?OS%^qXZV;0*jaH&
zi*FPU!zl-Gaq(jie>a49*83?(G{f#BNfFT0bvRq8Rd)cOeh4|3<S9lFDeq*QE`4)1
z)Y%*5dw_Ke<f)sT0jY<zVH(h9kCiji;D_Y5(TYi-6_c5ur;#~C`!E3{|6GyQgR%3K
z&RXo9_j7noTpP+cbHSura68bYTK|()iPfm|yu~ZgA7@<@xsN+eV$kHU%*hisk!GdN
zs-t>5cXR`4jplzBa`<mUA&d>V$ZRH(*e)vx;m|-%T@Pq22WiiurvzxBcgGY;`Jrpy
z7Ms%XtF_vVNy*1t5`pSK3MA$PfKf@K-JOa`2Yp}^!1P<th9KZmHWk+;3%-;b(r-+s
zH&QkSx=xM?X&kyKvib@^^N_K<2V9;|X%Bsj^J-qCu}bp%`LvT(YG7{GA_-41nj9OG
zMPrK@og+#c%IGlj==e0kfmt*JBh#%6PPQF{X&gssJr1=C!VE0P0<{MDR05^}KAUah
zyY7kEM^a$X)}bh@T{{Ou5(8_Xl!1I*4dx{Eq@%+p>lM>`0E&_v5}z{xShF8&7Ai!h
z(NWU*p5MlvpW|BbM(>Y8Ghg;)kL|M}MHe7xayYEy*kw0V1oILbPzx($mV9~^+5XX5
z-p2rmd#ogE5`oK1P6}c6M1I*aqrZ}OjGj>o*;RTDSVbaQS7LAreZXie6LOK2LZ0#o
ztWo0}?~shpi4~OB9k>6(E*gCUl`!ocqLEhsoJbkg*3@u;^3U>7fS-RHU2{AUc?2zh
zV!?Xnm7Vv|<Pd|Mjx+tZzy^i{=xqXGBcE_P*fD}c9#wgH)ie|oOmK<Av}f@k0r;fR
z)j!-$cpheqsQDy1IZ^I+VwB0Bu&n(LFTWOmaleG{$xJ{R$TQ(?;K?^~IfBZ-l<7&w
zTjQnOwZ=n?(r+Zd;1vNp4ONxhW#7Qd`_TM?6O~MsFm`A~U$bgC#s+YKY$npyrEiNu
zGB;3NVs!!Z!pzpg&`dxE!T_RmDDA%zI)RKB$IEWN$2^5UhiBU_3`YsV2v=l*)_^#3
z#)&1XcJ%KmA&NwsQZYu`+ntduZ|WJ)6S~Je&|4u;%iH#4HaKirOi9nipMdZhG{-bN
zfYbaQRFL?}ekk;#e3W9e2ttidU|`=vWwSG&7GMIrrdd1?@%3e6Agt*DTUzhBs(FdH
z0x=_-M}*Pw%B;(!-Uc@;2Lw6QXpvZRusfO$RcP=qEHlz78z~m(b_lx=$W`KXV9zS}
ziEzLxZHtX9EvlfB5GOSb-H2F#Rm55o&~@9kNR0XDg&}?qP-7ey)vE3XYzWYe&`rhY
z<zK&iy1dqUYCjVb43@)j)#^bRfu+6k=YRGN13Q70O0d3GR(heR;A&47TI?3Wz-<Gx
zSTyC@W<tq}V0R(B=A!C$BX{eCHqSzm$Tv9H$kGnbgig`i+}t==U<57!W3?K?Zzh2J
z#DQ_9LnVg2Z^4H}=@KBc(i8zECQ+ct8LtHM<H2(RsEqGnH1g-qpVwD#W2lF#=dNH`
zK<9N&&#Xd??Hr-FU{iy;8ID??0nQ;Qu7NwqQi$%%n2Kpkat3yGf>kNyEZ*%BinxHt
zNPz^-uCl!eHKz)H_P3Gzcn+Ld^BRkyW*%anjLgQYJ44|sVxTjHLyt>=*(kvPZZ~XO
z3^fp$bhHFJ7&V#+*t8PZ(Do*9oWFo{S!R}w{=%6rAdqEUVsk3mRix=;kJeRvFqJ=Y
zWkR8jkdV^B&{fT`mO;4^q|<6ttw`z6bPxwR9bs9J8NW-<JzzA!|3Ot=@9QMN2&g5g
zOBN3NSh`RXJOmJ*Z&<7F2O<OkZR7w7915h?geJ6cj5O{=m2q@p!ooTa_oI&mBF{|J
zZO2oLO##jKEMhEX)mLfcNP<fjloMMf01Sx)6XhN-qD=!Ipb;`Ylsz1KK7a$nWCiEV
z2kJ0)#nrLS476JS$Vx4lL;#N|6F%u<U(=jsVg}Ijll)X5r4<XNyMv`swvW6J5)1EQ
z4`biglAQ^rSmdWqSF^ISm(V^iLBU^&M37PpHn&kcDt&Ja@<<5`^?brnfYV7LAHoc(
zIzT>B(gF+*j8{95$MjhN&jz)ucbm<JSG74V=7Qg9?BYcwjur(8MbyYpLpq54B}pj7
zXbwsQ1ExzdFNcRAD+;vpTM$P~F~ZnY2-T07r7;!<05ir=oc*;Ku!Bm*iSyY>c{7r*
zw|R?Wa9`Xvzg)auxB9is4VHUzKHGcl-I^R?eS2)+6>#^Dns;WwhluY6hwL7KpHV0=
zgf2U<*ccGlI>D|&0vQX4o50@agg)SEYBr<vRYUN<58kIuBz_>YJ!zr&wB^f}%R$dY
z00OaintmWi26jx{)A#~<v;z2!aIXiZr>6<2kB}Mh`8Sb<t{Nx^XGQHDS3pH)R0Nz1
zGmurddJ-Dq7wdCvtq5g}@aZf)FkC4?DOEHnXVj(CkBtoz6G03;jI&259;lt9&{#n$
z6MAQ!z%w~%*wj>7TU#-FNftxj>QMeCyn<JlR+Z78cMujDQ=1e~gd&7QZ2TjLtl>ZF
z(W6I*g>(gKs564a6Azzt7*qeS32(JKA0*8aV5W&<@<Y{#ApZco+Ep`KJT;bu5o=4}
z=|mO8+*-vv3(;E~VnIv=G(geT7aT)She(hg3yVj~bK?nm&4LXT<9WgOT#7(|m7@EA
zTrc(vu?Rs6lm;KrfXGeb7G+@oP#QZtJv&SK2|$m5gz0+-$V-z-7ApT1Um18Cz|AH>
zJ*kcc9@Dr@+%T!Ef;F?g0>cDiK%ygzdO|oY$ie`Z8Bp3;E0gQ^HHHK}VoVTO-I8R4
zcL0>%5BYB(zIDlCw4oH8(bjKX3$sMZz1~_43l9!u6Xp>z9RWqn#TF1HB4OlX(R)Wu
z0aA&s&!3;77(-b01}(4B$>?G@?=Hy&#8>?gSXm@b$``g3%?zly`D5=y5sC!}28}N8
zsY-;KNb0O8LB^xROZXF{+k&Jv+tux45a87#7@^M*^OyjXPP`NrFqQ8Fq8<{ko@4;U
z#In4oNDSSle?U`d0~Fdmbk=JykZ$2<l!$eTjP;uzMW7}aOZWj9Q4tZBb8e|Qv|I0=
z^bi9r3ZWcfpT=VP6!XO=uji|@p;ucCmlKfi(zAA?804FDt6{Yzt`;Igm~hl4G-_@C
z6c`<~?mR`}fRaIw{CzZb94~O{AT9wFm`O(mXu(aO?C_Dk2z+=FK#n63=#q3@x8NDu
z%rEryB@gPHkY2K1^qjB<>kJ2z6uFt1C6Vzy24^ZboVpP15ZVjGh!CYve;Qi6a8xNJ
zLd|YOgoqcQfxXOf0DRALNNH<H#{*7SX<M5*2;ZcrK*(@VZI?&$32*ED=C3@At{BFn
zk9ZFRr5AWVxs+=XKUz)J{Pvp+IguFAg_y8*PZhf?;&mm&dnr^>pj1T?(|MU#URbhJ
zLrjn&jU<E<K==@d`5Uuvu1O$XQ01DzFezZL<UIz^nMhu*_vyd}-P<3Ri9^_B6(fph
zjHCW0!fV$tB!2$-4k*)lp$(cS+A2(DB2*DYm{uQnzUw8z5>UzoTt`}3THMhQNsXk#
zF3-&$LUVTV^a9~p*3Xtvsnp1##Wx?pD>O=c6AaC_!qux+zc{RpUJNp6Wgl`Rqa181
z_A>jM=n~quckf@CMkXdEsEqAAULPD7XiyIC5iJ+<NXo+2)^-=!QDp2+T(VSsK3Xd~
zP#H_R)kZR=$Qed1t-`g)*5CCBzRb$niJ~G0;UidiG+#o&$U7Zmp<`D@%3F%s?8e+H
z*iqRuq8diS)Qh5$SaKP0_VMiNco4_yw`|#>l>^1!dl*$y!HtfiT}VK<2<xo$b61x(
z#|4~+wQrBPfkYaPE*pih41^8c!LSG1z%g`u4$1u0Jx?n5BcQW4Ze7&S8k%pcs9>d<
zgT1u_GzOhvDsuH8Rl^~H8Z>!}OG<KNCn7;*qQYV~jtd&>xHm$YzTMEzygz&RPV;e0
zM)FRF`D#w&aTGe9V!4as+Jh3FgVLc~*@dB6o^YzW5Dr<^*CBf&nncG7FsCG^k!Mn^
zA0&d5nORQQmMKtZwZ<IKM^lGs(Yc?yyLD;>;ca&j2J2DvJR@_D`$VU#V#pANk3Nw2
z*=@rgJ3FrhwL@vBxN`>OH{LWgvCr0>jwftA#E>v(9KV2qI#~@E_g&aY=`*+Jfu>YU
zrgGAIqPwDKK#&`EK^^Pq#M#8Yzekul|Bt=*j>_`hwnoX;XtE<38wyhFy#vw&G%8pC
z3r)I!NRuuv2m)rCV4<p@bfWa$rGwcB2ntB=AimO@bm`w*sM$H+H|}5e-f_nrV`q$W
zvLpDG=lMOW%r)m+>(R%gWKz!cj?qW9Vidk%#biDCv)?V4365l*3_{i7ZBSg=*|3ys
z?PRl1pZfYwPqgyGOZiB98pKthmRK#M^F(Nc{fpGG8n8-|4w?7Pyu8z?-D1<b&$nj1
zqvkT)^Ge(CRwbL(_#2s+j+XFyJ;Je~Z_}Zjf)K=ESQDMFe&nKf+35SuB3rj1uz;iI
z)MK%<w0Y141M9<>q#9sjp-S^oEc>B#DvbQHpWK)USfAr_2s2+MT#Vz>x1T?M9+(Sv
z0cM2Pi~mf2*7Q`P(-Cyry*UDS;H%q&cxJ@ib6bhrY~d%_*7Ocx`|W@s63K8T6vBxv
zd%lrWS0tcvPSGp>1vxHJ*b0pS5wx3yR)HZ9j-A_zsb?X#fk7yym$i528V_+9fvg6A
z$RyO{*^7^eo*0qe1FJ#H!WBU-9GVkQ#{=_;9i;$d+sH-+R(`)M0Cfh^M43YKQjOXf
zbM(68jsR4AQOGAWOwPYaT@mU-e!Hm~z8^54rFG1Tc##tOq^>*Y<Dx(`2}i@MkSm&o
z0tV8CbO)6z>jY{3_Aop(UfFIK;dZ%m#>Ohm_ijWDfF)Swr)P2-vu-8<+o0FQ?6WRw
z0Lm~9QJZ9VoO<@};cI&A<rR)^g1e)n(229D3NA#v@Bv|AWe5(+VW}K~0AHv22J=h`
ze^eB~;6LG}e-d_>UjzZJ-VD`Bk~XSGzI4$oTk`tWYGPg4mGQI3iLf*&d^iLy?%Px2
zR8mVr$8Hv>N_UKJ5|L^)p~?u|2-g#40G@B6@t`)TME~BW3|hVfd6)t%RCSH}c^FSO
zA$<xC4h8x71O?FezS;}@kUR>@WGt)qb*l$!li#G}sQf%+pMzcD4;<Z?xuZXCT>Qhc
zpLU)~c(>-~#alOO>0R=CetYAJ;u9;^tmOVV_{YT;F05R8Ve#=0#g#uaUy%EGP50(~
zzx?u#XWM`6o4$M3E~%)c#5l2Nd;MUipXiQ>(kAbEb6wpXvZ$Dm8~Zc1?+}4ZM6(Y5
z5N%aiD5rGRceJ%>PL#4pOk4tQ4kzib+>7X6Y(LHmC5aBB{q(Z>C9xfS25G`jmHQa`
zBugeLUTfI0YwbQ|xfS9w@zwEubrPmaQH7<}ac$nY^R?+xSzBD5#t+uu4|<<Hu%jq)
z?$@k;Up*XwbKnEQO|wx7PCz)+i8LZ_p8Pm&9}0&QDb&pd8Le+1*S!n&ehRoDnH4TT
zWN|Y1kQFi+Yh2yjY9WZheyF$`jdM_Mu+2)1k810jRKM;-EhQS_Ds(Y1*h!EXcTP$Q
zrRInnC>-F(?3dN~!(^uwHj27vHbb}B#@CuzdZjn@kL-T2T~=eP-?6yY9{eW7eElbN
zIH+1)gNg@`_X0pBklTnOGYcSyQZo9dw?mLTSP(tr*-==K|AGIKMQz$s8RoWm%NCLu
z0^^hhsY({l#lB%IjChEu>wR*kSEWBTZxfxIXL@o{AJ{`fj$I~cQ9V3@fjW1i@>Rxf
zt0Rovom1L9JMA`)eykfU4pnAY6A_J`trwNq16&ZaPpAg=I2i(Ai37tovtnXwj0laR
zD8qU|B9fJoawZ}Iln6*><nM{<>9)`u`p7ZZ8P%!Hp|Q<|4NZm_(~-yyRQ7>`K-8i&
zP+dY_cwAFcldX1UB6mK;FtOfYZiEMma}+z6Xk$r=f@(gHULLIw@rz-KguTJlZ-?yX
zCUxOEdd{2sV?gEg`o#R6Nh(1>xRWU&&4$8KlTvt8SKj$ewh<jp031*6E-4OX0N?{q
zwzK(H8wm;}_=b)A$J+t`VGxe(*oIJVgZBh3R<q3W%#_vpI`k|aZN1G8W6DS>v@CDQ
z3T_hoMrmS)f)8CVl&8BkZ8}agW+We~ouY<N%JD;|nE6IEOol6}6LBJf2tbuZ%OwJJ
zgF4+;lKH8R1&}OIn>;Fq-RJyw)*`5s;4|mxHvyo-Jv8l_mDvYf3{{uSceGF=jKLD8
zj`Q9WiHQiukzR6!N=0-D5Ue>k#+Lm^*ToJ)4TU@)FC;Yy>#)K)-Mn~0<wVmfy3|Q+
z)_mtq(og*llMgp?lb8_Q3@Nbd?Mh%XGzt<kyJID{nTzua$xFPEt3kx!+`sy;334V|
zA*lm%a~(n}qBU}JfN#7l#Bx*xY&Q<Ac>7~wqQl0x%V?<(Lbb)HNTut8;NFXppPon}
z&Je#3SC~3cOp7>+x1>(+lFk+ngZozb=*t%o%&H+txiLL%M3P<V;BX2{;E=_n(}+n%
z{k05eRH^{92O%Hl---n;N{>I4{`6oVhgKlh(cl0F+m~4a$qbyS5~qc(hGf;SWulR%
z%N}ux3t>JQ(+5LzUhEkGInSC796j1}5Z}$7&GJ!F-PFD=XNHDM--ds%Y%DYDFSz`r
zj=jx9z%qE`F`x~9{Mh}G6n@xrgl}}Ymwd(+$}=NJk>;-v?T0naTfijX)C@G4f_`_w
zJJv^NcdR{m@}w6DoLA1jez?VP2s<XUT~%U~`F-2ZbdwTsdhG0M7IvH)TZy^CMLb$)
z1#*2Qgj*Ha+m9TnA)!De$m+|Oz==1POn(By+6z$ivbb*S(mORns;X;3U^i+~THXZ=
zzWFeZ2;;3B#DNwK=nD>Fh9n|4hep8%sanUn1Yu#{2gT^i<)kB8?*OhH+fX1R!j*_T
z+Uv!4?J63@(YvfV0z8JJ;Jlr~evA<g8ej9+uhw9|RSl&7ZY4=lqC=}8HMv_MglnPM
ztc1f~iN+<yQ7K9T__l#p4Zbz&8`HU&o3eP^`XkHwZ_34~H{SqbHV^pRjNg;m?AoH!
zn6y<669@goV9cKOvcT3Ji%-MD#lu;tMvd1#bJ*D0`T`2RIx+<q#20bJD`KJ;rT-|n
zs={&uz_!<i>h@-eXCfF@#QQ;ExIDx&V?cWHa;mXE<fF>^rM5OUK49kSJuJnjkG1kl
zT>lVyMZpxY%b>hzsz>Bb`wbm5U`pUdAs+iZl0?@gg;H{Pc53V+bSMM{>7(imMUkkI
ztRyQAD64IB25jK#UagovaGoV#QK9n36)UW4(s}QU-y6zPu-yt8XZP5jNF+v|o0<|t
z69KN|b-oKe^LTHWcvAmE4=i6;6xF?W$r2w-5V>~O;+8La1GPeTSLNGP5e|hR_I!pl
zZ$sK)?p(XKcm!~m@U~`fHI;*>At_dN5kPU{WQ`!k2;qT7&8!(*ZdT?Ta85QeN-3Fv
zl8n(6YoKP}v;wNjcp8JF-qA~fD_-Mbv8M>!jsACoV9TIY8isoK$h6Rakom`Z5g>dX
zaTw_sx8H7<yCV>!-AY{gn1Woe;iwE@8=}J4SBGGsQ!@}Uc_ZO3V`_kx@L>Y`Z&OR=
z3agABAba|@7lmh4v7M8VIf7%;s&DiJgfU;s6?P|-%GTPQ032}WRzJ{E>2(LxZ`cH2
ziDhcN8x=O0mBp-fCqPZJbsPZi!CaW}Rcys#evtGe;0zPqq3H)arYj;W;4ly2r65->
zZy!Ns4O3zl3_*FP61~ya%d6|%gT(ZD+`AHo+9*W=w{B$AVg`5_wU{D;L4WBEz(I(y
zG`nPFl1(*Rl-_=o7jLimV90RM+ByxRzsizaBphlzZ>BJz+H6!+<LAediV9E)jJbo(
zyw~dv5>4*bEhy{qI!)0?LeMbJ%Y#SlA}OFQuNCA+_gR9TQ;WY0vaHUC<_OUC1fn{*
zl=`c<ktbh-D-I~(P-0Nm>v0s{K-Z$W3w4trU`|S_gd?L$jMB;I#CU}*(@TipHH5pv
zag1yl{!R9yCF&Vi31|NJbnRm(@smM<r|~Qm;soc=G_!yQrEqi9oy<Lti-O_wwZRda
zqlwd(<Sl{65v}YWh&aMoJ(*jH)=K3)RE;0sS%OIZhh`nBRnGmA3Wp+Xw|3Tu0|Zb|
zv^@{KJwQOeM}<IdFq`4a`vkdq7i1f?Xj#b50;y9UBU21|4`y8hk(@EOUq>26Su(~_
zG+8$}zv<aGE&#?VAj4!JNGxN<DW~i%5yA(>va>GKp9*oiRMX<o35QZL#>Bq7yqqKi
zV3lg~&jE}z8nHAF0ax=iAo7_{Lqk-qBLxJb=}>C+1Z<!PF#=6b{FL#?y;vX1HwYvK
z`Y2(KfQ$a_k9!u~A3v?|LK49&jJVGTV|497oYc1tJX_O?AK%*}7$o|r;=~Hkt7Axv
zSqQA{dPDAKA{rqja>$jfke;0ug2rrDcZN^W^wMD;k69_g;Y#dXZHbbJDc|Kj&cJ}C
zmNG&sA)Kf(ov;FffIuF$30hxoiMSaxguD<7_}|-21gcVkDZvDA)WKTZa(p^4Fn4yh
zGW2BhwseNPsKYyxy#~7@kZdL~xxdz${jIs+B99$*M#mCIrj~i_&rrjXy@cjv9IUmg
zR(%*fii<@w)+kM26tWmp^~>x@9i@79-J$=~vxq*Sbmcc3XlL<2xi^cn$LOS;a1jpk
zbKM2KfZ6lK+d+&7$4@gyiegCJI49E1vRW6I!s(1fdQg&(=*!z}Vz70v>M-EK$`2y|
zy(Kc^?BKaI1|oCV1`~9g!=p3vvEAtZF1w^6B*4I;;nq>J4<|?7uT=pYPnF%s5O9t$
z^cTUlYQ_(T`Og&9Y;1H+Jo4=LBeLp~3gZIQ^10<)1tfQ4ps<Id{3W6RQeHKX3}}Ap
zqOly=&+FN+_05}<uE9$Vho#9L+zL^;in;dLXP@lVw{AvV_Nb!?)}5?O05NK4953QY
zAwaaVGYotEq(G&pL`4;?PA(@XX@uv3n_eT;(r6`06Q~TsVgEqxygkjLaq1r-PO8<y
z7e)_zJeAW-pK;sv9m;%j%MFs(wqvgV@x5sqO?+@s315|zs9_x>SfQs;zZ6ZS4ARk?
z8&im20@ribpob#e4Q2vllz{B0h=}wK<OY<_yPXE(Ja9fk@bDiw(S8#kC5@tR`A(w`
zRYb_6v8S)){gez{z%q{g|H2f>5bXvzjL3n>G5PuPPQ*u?tZF=ctV^HJ@aj32#|FUG
zhIo$2GJqT*!a}(KrNgU&L#AN>))Z}5f7k+n*b5u69o0)$^*@X@aD!oy1k^h=xX<#z
zr&oft9gcj0Y5DJM`~WvnYyvB|i|k7c+?urIcuot&$(Sxh8fGg5AZC6x0}>gw!WY}E
zI<}V%@Vqwc(oHNQ!x6xL4bDsqXJCfu;|2|vp<u-1ofTpiAY0m61fUofDK(0d{DG+B
z7#YHIGB|w7cKED{)Rfh)2ezQhtpvqk4Mvx!_3^EufUi$sH_qOvxl1azHGoK%rb=BK
ziv8(20t{1O9T^_3Knv*?(1}2KJ*N=w@f2`gb9eiboJ}k&!<)B%gIk2^FbmaF%%a52
zDKdQi1&WzEe!V8(-ZcN?(a2<!bl=%){v6rGdmbYZKGn7(Nr#8k0V2~JO=h2DakWAW
z>ym@R7;!ePjY9)O@~25_kS7$Z%3*uK^VVT02<y0%jHe;ci?NB=U__T^kOh^}>#K*o
zdB$^=#+u{dMkT1iYB7FwNq6h<JxFb5EbO6h#UZx~t!POz3&IpD#*tKh2$1m{9blIO
ze_$Wryn&2H%8|C=(VprzJ+T%9HzT(!i_0`^`&a%D#^0=izQD4(p`YQGGQ_sz-H6fE
z@o2R+sCN#)b4XvX4)OQ|o)TG9R~6DUuK}dETO%h6Xsq_c{uA3DpW(gs_5uyp@o3+*
za`bQ>Y(;+{<fou;Xlei%&8uy3&|5FKNL5cau`e2jmz3f9Bj~*HxYnSEd5#OaUfuVh
zO?hLZ3btN&S2Mg~27{|C-L_)hQgCq2b}%~(Hl42wE<^A>1{I4ktCO3XRq^5@ibV9M
zxd&nASS9RMS&<ER%IAntBx;aG18cDIuP<N5|7b(cqJ}L>wcOrN)K~wSfye6-vSy6>
zW6+~8Lss@fR+#mx=0+fz^?2N~*y^E8=a0J5x6k;yUYzx56wE}=bf&%x;)4fL{iX)z
zj0bcvCnf8mD~e4BPD)MZfU~7?9MBP$MONS2a(DE%-+#xkz`PbsPIQ42+=9AviyyqW
zsBYi{Cye?&l{rkHJ`N6=@L#9Pj_X++4@i|{hs&SdMVF%wNrq!97eM=3IHtX?ngOan
zy;a^^AQCybODqrb-C+OIKmKQ^?^7yLyyV06JtzC;3pgN|OhBMzAFMHtV`%uQ0CQ@Z
znZWH`%fS(eCql$B{v1_LOId=ScuzExk7C7{Nv1;M*r!kL$UDl4dzZK*^>GN;m9NeS
zeMg)DeDxF&JA_)NU&Gni-}<Z$;ePq_(y?R5o_rKSpog*`L`WP})yRWVcFkmi+|j{f
z;U16ME=6He2zD|5nD+@Kldx(4<YD;--|A`o?lC^sn5g5_CR&L_=6T1DRpbUXVp^xf
zI*p>(-#)JMdqDr?sMwbw+vQcml;TF^nX_jvLE<}#)s{4OVc-3R#;uSxiozX|`sWk1
z9L>2x`jQ)irglX21fY_*%r!%jBV=v9yl!>)gP~jIU6#tPv^)EgbCoE09iT1O<&Os2
z%(2z3re~4Lm;rvr=*bq#>C%=P7Ea)rbwMguh5z$=cHA?9fE-|qsSWlIgVDU}<Nh4!
zy|pDhUCdq2E%>9JKlYM`1HagSUcD%t4p_-w2dttJQ)0hI#yt}i0nJ_(8PQz`j+e%H
zX;OporV0_+&qvaDP}c^_5&Cm@P;ZkSX(S;fr<km&fNyp%!9H&E92bqqy{lGV_Hn;a
z0?O-1Tsp_LmX;Ra^p7<K?M@7|p%n5#kE{6^JVt+?R`7LbB9J#7k5feZDIF_HS#!Vt
zj^(wJ*Ehk&$17A<?osb?c!7BLX5c6$7uT(tAaA?ODTxDYXJlNBL8KF6yDJy%4PCQp
z)pNwzZqQ!{ZsbkgP$W8@Q+BDKr5EQKZzPRz9U16B4K9RvaS<MO)Szr`@CqpTAZ$CQ
zse*iYcZps7h!FW%Ob!wP*Vx!NgSPUpIG~b!+!FzXVA7P-!mzh~AhG=_uo*{9SE(Zw
zuTatjqeKoC0!`@nJ}jEd>uyk%=O2@?buBQvk*@%<oNqTKXzvd09W!PF4#e*)0-grn
zgh~(XHjGyH1qmf>`o0L+2u+RmM^SlQ;Elm8u%zpXHpHZY(YP@`*=_iM5Jx&q7)FEL
zdIn{bnvYFi0PY(0d>v=B+q#usv6IYsljL)8_OUzy8SZqc!4%-4*Pnj7rZRQ{?pYPH
zVlU(s;Z81;_4pT?MJ_Ik?D!XtU1O#?bEXH9lqQ0k_H9h@xFchc3Cp<@<7Q=yMoLUM
zP+&ySY!+-oWmW{aX>dUvUb~Q(ZEj(=v79?Wz{n6-xz5NKBHD;xrsyM|n4()k5i+Lh
zP!gvImUb<ksk#&<i!2@%+<R83T-?2O8JBRCR;M@6QyRmPOpypfL}LJK^un;>;X{<W
zA5Uw9T%8np;Nl`Z((E-gZ3u5B#u#t2?ed5nq3MZ%_g{F~nRRV2UqN<WZT~s~IR22I
z*OQfKIYSW0w7oP?o2Gy&6}5Ur|HJ<?Y+YylBQgKsck56xB#Ic5E1MbSjLo$j3K&R}
z#S}2%eWjFJfxXbAfd=@yVfhDW(ns7Or}}v5i>sq{PbwVZ1dJMjVbCQ<YzQiP5f^&V
zVnbCHsdgvFD8|~>ZA;*$U9)B7cy#wIx$7mh9c;VW8Bj})h)ZpMoS(;`8^bYR@Iq{d
zt9=J2xgR4^9w?baByZq+r43}+S<_4OP7b<D);!?0T)WRfk4N@w>jz$6*(y20_${XP
zt$$OFoeGX{6#&N=M_W{kxDGbhT-0}#<`sAlk~kFLAb3hbA`<73{8p%DjJ|w`bj2vQ
z97X;~m9?<_Oq$*K)AM3G45OQ^VzuPvuj({gfE&|Gj02w$6kyfh?;Q5Hrcfga_BL*j
zW%yCzp0XR8%vu1nzH4<dNl_-+7B9aqXBvu9GBPB=DsVK@#Gui^_p*k8du*}I@eExv
zLT&i7aEbs++hz<#?fWa7`My*dJ1#NII4KjoTJ_KR{*^&Toj(Q(|EU%nyLt|4v#?-a
z^OZjvPJWc2O3w+%hZyBE{q0?Tu(89WJ{=qPo<V+l8I^}3He9mqP^kMw{Be8d&asbS
zh!UAo-p*-snCj+(OkQV$B%^{gj1M)_c&$Q+`$=GF-tTJg<_bq$c+KD!ZR5}TeqcGK
z{_5&i<H?sLH?6JfxCD5kK6G3|1giv0o&r4~QhssI%ioFk1w60^5FEGl`|rr(=lg9b
zd>D`;Gpgdemzr|m8g>_$edOI5GA~s8cE1LD(Sg*14bM&%H~e~(s!`;j3cq<ALsDQ~
zRm!Q~HnzJdLg&UQCEX<5mrCkZ;D4i0d~3RH7r_R=HOarRSw9xljOlwo!r9AqND=G4
z$8OWCO6Ohx76XUZOaCjuvP@SXDDmpp#((3aDXV>V3p>%2j@;ZSA`~@OR6Lm8bYbH2
z)YNVi^_cM0po}HIVyM!a56+<?7)8%6;!z075P8nz)dnFL(_jJ0)BXz?5$zpQ$DWHm
zVo#PdcZ=3@yR6-7yMIUZj{}>>O4-mgyt&~6{wBerlXb|k5h8kN52yo<M3<&2>B?D-
z0VpUHTv2f<n2Qs|N;lx91X&@2bKUc=N6OCwi*jZXR~aQlA6MBsF$to^vUM5Z*tvrO
z`bqauJXcGWbLv_A>lTP&$yo6tHMs2Q{~iWREuiyF5E(#yvd<McuCFSwF6mz}dx(?{
z!fgeJ)r2}o`w3no_$KKe10WlbJenHuQ`<bj)h9eaIV%+8j!JyuHQ?D`*4c$AA~@`c
zmn|oV6eM^+QtZT~slXt#4Nk}Zxd8KXg>&BAtCaYvt<z&QRHVxOavRS8-FjP*br)8P
zWA-#R2a3Zbb08^dX`}M}T7?FseR~uiS=AnQobUq#4pX0Z=tiO)VFsj!Uc3hc&lBCD
zoG*Qp@nMYARgk;$;z6EmP;&&Uys&5TtxU1$_MpU9uU^&huRk0FI69Oz>6J_Zfg^K^
z<zY`fE!|+T@nFra52oIJn>KFT*!-shyXS^pZa@Nd*fDWrsXxkv#(_}~Tw_rdam0@D
zpdGq*?+481s@f7mpDa|q^KRq!ri&uS@9|5F{pyi~`nH_RikR4%u`5p`p()9*cE8i$
z6VwjE?U`G*DUK#k#hYYqF8&y8EoDQDclUsAiTbAs>urO;rwe8G$c0`P_@D+lanp<X
z7FOc2J+_9RS9<s8P60vd{bXUz)}S@-_QGo#R<1{ybX3htW94wn$bj_PTQAM^otTx<
zK&HLhds4G5y)AVk3(L)wkH4AG3IN_Zl*tTL|33BP{Z$Ig5YExI*FACQEg6ND+c0;h
zZWu^lB{Je#jH(dnqmbr01ksb=DX&EBtahn@hKn?@=P5%AkFoOw)WY(q#%iAXR20xk
zKSf8MZwWYddFL3$UV?mjo3qOv(q&(Fz<7PDKbbW3OcI+_KL@R>f}&<NsB4Z4%{{WA
z&YE@LSXO4wVIEU*p+Tucp#}WfvZ^aFpxcRU2s~4Y9m!<oavrHVv-%Tgi~uio?%n&g
zEtxkdGAX9IqF;If1tQpRiPZ;#rkeo(gkKq(!KCi`@^OYEA=kHdG>ud%Y6$r~iLo7p
z=)HS+U6T9+c(4^AQJxz_Q$gtpVSm^2nh9vJen!e~i^JRu(M&YcMDOrQZ+{F(#fZpO
zrth?X8l`~Q6efL*cS@}6%-F9dZ9j51Y#G?Lr*TV7T}Gg}vj6+#ROlsW>m&WP4O|^x
zAPNnUUShq#*eed(X(iS2!Vi6OzuFGlh|5CyQH7eF3}k_9R3cBNKzbsIwGJJEPY-BN
zkWy7B=<Cxp501N(IKzpsCs-mX3pP{U=k%3E2@d6K7Ni_F9K8cy{11dIo!zrE1O4iN
zV94JDh8?+Ff`G#7StgnuVuC$bS@(**I&IwcZG<;mv)Ps{TlTpIqPd6~0=QG+uwiz~
zI}S<4@ar=e4ZyuKlAybpq;bremS^XbQWzO&1%)KR)^gN2Yw7mKyGnHPo_Lro0(}y>
zg3pYDPw&dyG;SlK%fjO+cK&KBl)>R=>zfe*Nyt98<CX*?gf>32<^hkqX5l#jObQGd
zJNC08QT4^Ps0o(mDj?ohqv{?*%WyUEYPuq3$>g7msC1b%4z(7##}I(DS9jW}i2N8L
zaQ{`!`~9|7I}<Gj6`~YMV)+v9Tx{N=V>c@cmq-`*XqQZ)Re_oau!HBaymkAOF<~JJ
z)xoz27F+p^o9cQdVdSHu7jK1;zAQt*%o9nEqh`Y+@@w?1B^xC2vVhoxm5FVxsyNxw
zyjKKyotKPv1XGUKT+VGKpw!HrP|#&hqhd;wVeN~i#|B$lKL}zTn3u_o#=2N*haqru
z)KFyy?Tf|OistOu5IL@%n!o+gepZU0268_^XhBU=erpbCKi=z|hks&HAQ@^h&KGB*
zd(SuFHMH}@&^FO4JQYOeyKg)mr=3vga}%F?2juJ8dV7qWhzU#=wVdsZ4Q6+TmTiD-
zS16lsI+}e#h!O9i+5kY1qfa$<hAiuRsBGL#1Hl7YU-nvnsNcbcYgNm#5r{O53cTDf
zrVb}?$3+E5?no<weHF5Mg)|0fXE0QF5%g;QY<BT=iKT~x8ewLlWd<Wv{ihyV+GRFL
z+<AB@6E32(cH-Gg&fJt?<_A2nGSG4z;=weewP)i;)E&O_IcO%tus3*_=bpGjJ}>o>
zf63t`_zj6&=1deo8nSWYU<3ZPa*M=eR_q$6FpSH-mUiMaN*5y;kJko(g5b!M675Ll
zLqs>!te2g_spPYtu?7eA3@%4PdR{Z?+>fAgj19HU&k5v}ZQDU2Bj6wgX+F1rlaW74
z1O~9|rFk`_{6xKPD{eu^BO6!83Te6@3T3d^3l_NowycQb;4Ib`Ybnl+N1|mX=QaMl
zB0WglQ<s-*E^+qAJRLd?E<3xuiAhUVeaF?^y8R~~QA*}pwgG%vwWvW~sqHYKCf)!!
zW@W7YCk-m|%C4Yv&MW)#G~gsWk9#w%Ze?QP4EDqo-W&us3{FsXdZM~8;U}OC>pc>%
z{nSM9W*-?_60_03t9-)YZ2g#h-EQ;|4&~g>L_aK(F!~L5vbgf`GqKorVdr_AeiS8>
z1Tlo6Wg*q@$8@so0pgunUvs9uaC5!5kPe8qjHgB3c6(8%y6@FkWE(gSgoDdR_0cf@
z>s@wIXhR5Hcw2~Ul>w~mT=66SSt<O-6pGru2bstKn0++gY9i*<5HS}<T9+>8G_C(2
zS?aFXphXWTt~+ZUBrm#x>2|>NK<~|sc7z$Ca#m}7$y%d?zCIq)6fo;)ue3rQjjBdV
zL?BUG4FO|7+Xo?7^wAm$S^%=FQ&ald6iEKZI^N}dby`SEM@~_p4Ja8yBv3o_LkI#V
zk#0Ez#>AbCV%FG{q3yvuV)6#FSr(jdC#GMb1!cB;Gx$WDD}H!DLu>?tmsud%SjXOA
z>VCxrWWwtfPoLNlGVFBx4CW7s9B~K{Y<EIHMGnS@V#^7gH|Lxg-1nWFDlwN$Iy#ST
zUqt6;30S?h*&aB9>5e=gPAoVT18{H-k3%h<1;U#Eu&}EOU>TOw(|t*fD|n38L#D$A
zg0&)biLMx#AaIc>1?CX%u(pO+V2H(ML6r!c1Nnr8<QvWfB)vr*2cA%kbRXk**U=Ho
zXd|{LC*4;t?blsTmYV>ny2(NY=0zm;K=)<{maPr8u^83Fyr_`YNAICfhN)uNCy^R?
zXN6ia@_N$Q59Ac=7olx2$h#YIULb=h;0o~RmU7Obx+WPU;0JzSTxiG}OurvO6VR%q
zdsY!ju;1G@Mxk&c+aeTsHOT3fFdlD4w51Gy3zdR1!^$am0lTK<aQ=CZCJ0B(VCYG{
zGSv0KoLwJ{cLRVh!WJ;4rl}cqtu)ueWP<4Gl#MlQz)lBY(7QSwDhX&;Ibx><fTKgL
zpbqD#4A}scv#9eRA%!hW@950oJL5m}ue5ro*1yuwxo2_E^a`Z|p#KBKW$UCu4j)L+
zqZGvhjVMRV-~mm1E8kkM3iF=F{TU$k1e`$SK6_7Y-Ek@mk0ti$PT_L&_wZPwovm<r
ze5ts0BU}ehi1c6m-l|`hyuEKQ3s`;4Fkt4Rc~1fqVAwol%s+sM&OY@wD3_YGRw)=i
z!lUx2pHYd@wd#-Ki!kU`$JLUD*C@?biDgIXz&1;uEaiaaIz5JU)}X`CHdKt9$BAH%
zMXfWbOZ64l-iaLcKw?O?f(J<5u_ASw!2d?<x=r>sz&X}CngEUhnFS0in6Y8V?l^?&
z9gPW!uN&Yj!?eu1TOk8_)pmUUild)y=bI_8b{OLxQ)Eh8WpEaOK!5I;VY02F0(Ns1
zzI8AGyGbkl2|vK!vsNrm_hHr~k5g&1JRIbL)VgM(%oBmXIt|8YWc)t)8PFZwL<R(#
ziZ|&(p$Vm}zz1c>?@+-2lAi=M{mcq&H_<28<cDo(Ia2#1I1vVLDVI630M<jirTH06
zLl9n!#4$*0w06lTXCBu?fRoSDA3xBwaPP&_M7=YhHzkrbfFTAfT^{`$3d74Z6a$U<
z$dkH5)NQvJ6iO^=CYiIfbkBDY6Io+5Dnty@6aveLiio@$vf|F${~;fK;ZLm{Syj+;
zF6b)&j3J#}L>NGSu4nNy5UH71rjBj9@U)in+*HvqUmNtJR{h0^)0c3+Mga}z3j2fL
z(gRDQgUj4N_QM!R9>pOn@`&b7w8OSzHr=g*7lN9N5}gpb5Dd4&ekZDpKxwGNp!wVa
zD)t-9i%^_L!XpjeI1I)|>z8x9*@eqW3JLm>;9u=V!A|Zt=*odVMv(gB;FV(;dxj^l
z1GVJ#kS&5=G|oDQx-EE3bT0-n3NidZR7&ZcWUpQd`ex$#EWg$7DC-zRG!9L0zhIaV
zXSL0ao(Prf?U@Z*B48=hi%ue$<H_pA9|k_6c3+qjh{4#IzW0Lhap2nS)*mw|Kqh5&
zBcI0tqHWAN0zeDPygVkYaeIhC#>(UoyLB0TjFp~M2qtF8$66iCw+jv0v9p;!=>iG!
z1RaP!w&-dF-2WqBQz0ZbfU)(&b|b+?qQL?Tz~Zm`Sbu#AD`<8rZmD9Hz1RDtzFXgb
zuCVQRG}`_vX&cT%6HhuDRIy%Vm?>AsyA0Q9yMZyJewc>`8x#<4Q|mRnD^TU5&J=Vi
zQi;H&SE-ubWi?_R3c^V5d8^98Z(IO?&m~2>i2n#l7pBLUb;O}hYg8571k$;_h^~%=
zx`~sHV(SG1ut#ebC`gzrUNw-c!lbXY<5dc=lu-{XpJhu{H145tE>815<s9~8#9&yB
zu-Q$HVt~uZ0*@F8l=}fTgyXCzFWWqak|+%O#NHh&p$iDOtb>V}Pb9ETil$Jg{s9~p
zu4{VeBndJnBWMr-ExRl72h0^CF6C>zXM$D@&_av644!>=qI(h~rYcCHVE06>nS=);
zNMC8GZbRHBzzzvHiY6G*m*ydA565gkW{=Yr;O^Jv8bO5x`IVIWpo>Hygm2rrl@?DB
z83Ga8Q2oajkkO+XeSc>K*%C}4%fLX8pI{FH`*0f8BF6z!M#HXP11s&fG?4DXywV-)
zHO&k_dvYk&-WHP40cO^;8-c`IWQd5pCK33$<a&jpLl=OAWhBu;i(mFQQ;D>WcvSCn
z`_Wk<Wo&5WE?|=cevj~8iKAr(7<16tksHdT@mGsh`C4u@e?m{3Oth&7K7iQd$JLoj
zN(dAo0GZ>O;pXrtsG$maL=o(S^1f>=MXmY_15YGcB*rjyci)S^#xEf5`8`s&hKv@<
z{s-$hgs@svLuzWvqMVm(?er*9G)R`LcrN-6-TGwx$JJ-pa24e-`&YZLIhazk)0qx1
ziHHfnee>f=YDoz!CtfW`<6tNnVrGi2&Yg%H;I{b0jf{>~;b2^uF!|*1Qxjld4A>>e
zA{G;AaO}e2DN9;mbXTF=D6>%Emp&6?JHxh`gWM<~lO>tza@MXgFr10C)c0EwNCVjR
zc>(WK39t&<?qJkT-aGz87?cN>zyPy0$ADaN|5d@ckp~4uZFOMZLyRFY1>S{KGeC|i
zk&4FvjC6y4Pw5{lGh%oU9R!c29D)$EZ#p?SV1+GbeP<Pl?%JRwE1g7bXfjFWHid~Z
z@lfEWxf|3|AZc>Ipr`;D;U`QKH{*J!2nulcf;QjELW@Sr0I3k6B@Y-+5<Lfy3Dr*!
zb36mln;w2e)KJ8jrKg~+^YL0t;@=`XqPjY7btys=jW&_{dQj7p;b<IFc{@oI%rwI*
zfIP47f8;LP{$gRs&hgif-PjmIWE)^zkN`&mdzQmhkiT-}527Ix@QA^3{AEpme5FVQ
z)*0QX5!G>pczLkpS{;#E0bvr$S)!0#BS5#A!dvSc_U6-i^@DHJ&tV<y?IB2fzHZuD
zT2(NS`zl0ai#isn+AkPHq%43HDAUgYM>;sz<BocmG9(qfbw8pG`sWeFvn2)?Rw_U(
zv(O9ph}ynDj3ATKBFMMsZh?DzRo%=6EghxvyqnPKkm|QMP&}99&S)!oL9+&}bCR4`
z(e_A@wj}Ku1n@8extO63#(N=K)C1W08{)wtY>fCOU{q?5eF)TExOdfXpx<{n{ul}y
z%g;AIvLTEoR*!(rCZOuwSntRG>=|SW4e^?Hz<Tj+<Kwm)y@;GEjVHwNthe7Yvg$@%
ziH6W8;s!bmB6=Zk3PoIv4+W6W01g2_#zzr1pdKZMJ=Em2v7gCVLwIKr4TX@~LyI4+
zg}|NGQ7E&jz$xK{$0j~Lz7wso5ol_9K<SVLyu|D0h+2;ra0yOYGM~`&7{aOnFv79W
ziXY**i^{J*_)_~as4d=sf`Ozk*z-Kt4DvFJO7@oAba0qo#(J!Dd9vi*a_(PwL;N`1
zFQJ=g?=CvXqxc5b6wFYagim+35*?X>2AX3`9yene4I^F&1jQNv4dIxXIq@y@GKGUx
zNN;~2#kSnKg|tp<8qhdD1hDzfwY!56?DcR%5t*Nm+sn1izR}VC#@d-w<WmrF$Pxv`
zi1KJcaG%qeeNj_OfqyInn;#WHgS-eb1#N+%dIkIrTC+mrZcuN~5u92s{f+?e{N+g$
zi16~!I8qQ!R({}*w-VT*LWL9JX4miQ)Gq)5S+i!%R+Kj6VF-&6s(teJ@4{nOi|(A)
z-$X)3S6mjqM%!l2R3JoEp`fGhjEa(mN0eL=tYyHFLeCU=W5*?|1EZA27MP0KRu6P<
zIR20V_am53@La%4oh8M@AII-t2b3dc>bVQxCLp8fA$l<eKV6pwxCc+pV$PJuD~t|x
zl1(5A<QPY2bV=S>ue!POwUHg1dp`1yCVnAZvlx?JjgA^HE)(7WK&w=8qbpIy7>^M`
z_LNSZcVdQ6fP$WhW72sceOotz9XT@Q)#{`GLf;JnZu^~GfVP6s7l5;dX)kGs(Mg0a
z>y%HrCT-M|c5z|A6$kz<rA^@bx{4VsEiJ?_Cn+^DV}n`%kWq3nc3ia5A_?K7No!lX
zExjbZ_SJLj0k?+FIQ@Fdn>2O3h`E%JQHAMFSOP;PpRQz(1<1O5myTH|ks|RDG|b~_
zc@~R=PCPssI0&>@Z^_HlsgqS?Y5-6y$CxXEm<?^UjestRzJU<1kFo1?88}|i0S-`3
za22<PODX-S_ddGZqUfLK8O)0fW(dCmx|@QGio{#;jFg)xSOf$kl7H)nEUK#rl3YPK
z05zjU*GM03BI#~L7=7<h@s8p8TuWOfq#y5qjEDoUo3zTvII;U(<s~LAps7K*{!A(?
zaP8pI1>EEePyfn73~^5&h@7`*0osDFXMX(t6&Ktza?l2GETq2VH5-m3nfn6dfBo)i
zJuXeXr5bWWGNg8Q2OA0IvRYm29Nohcj_upv_V^BL{sH5?TAatjd_z|NKd}9@sC_Za
zZaCoafy~)I+`J+wDUddiO(B?6`2V1R-Bd*C%ML#U&L=>DU?f+}x#|Xn*Pt>%N-n)H
zucPqV&?o`tk%WOX3?zD8X50vl61sUy_z2%+Jb$x4Ip2>QI4TgeqaATSigsoCrvIWa
z4oc+h(RSzA_OBjqX>yRrCPOVs6H2Os$O9gUhG+>?Sdc3+==5EnX)$&-<88ZfL`b8Z
z0<Y*<{2N{j7m=!nZA!+@Xd#-vJ0gJt6ow#VfHX_Stkk-r^ytM&2eH7|u|f(N(!LTy
zgAz9YCB#w1yq2u>$)*7G(cWiwxyku<6nmF3D&;mm+v1H0gct>X60=ZYi96GcClPSO
zC42`G42XjKL}ulP&xie6LEMSJHR}O|y_%Xsj3jz-D(TY_mP@~E)9$G#2hiber7>Ju
zuq7)BK+;7FN8SPcvxFE03JxVqQBWr-;9wH*1Z7zd?L)&2<(>62k3_<WKsl;kN@Ep7
z;o>KUrBI;)I%O1Rvw;i(cvBy^mvXv!2o({)cus*IjJ3p%VG@gU^BkK^$gvv*%_W2!
z7^}YjP4|%{0&Ep9yzHr-sD=#7D6(NDkPJ*9{8ci_M+^4{1}T)O0X{1ta5Ud}x%DfC
zIW-lkw2$Wng)z|oz-@^{;TnrU*BMlo72r5iDb)<~Ae={XI7hyWAhH0;UIqLHnaJK9
zCSPkZPUyG=8!Vv0>oO}2+6-<}>g(#1!nvveAL|byVEE1BQ;}GP#sf-^OwQsyDgu8X
z&Z^+HZ1e(A<lcZGES+i8k5n+y{l*EQ9Y~~xLL@<v{VNO~65DTHQ^2_)$CEuUQ@~i{
z5~c;@uuL5#4jOS&AOnb`U<7&TI}+d1PCzR?Q1xpNW?~%CwQQ>t$qu2>FhRz^2s7>H
zXRS!x+q66r<s<^&c5w06?!0h}FnX~1Myb(2H;n4^cbeIUd2ZW-YM3-$3VD#iT~*|$
zpR^A?Y5UvwYtdv$KV8&;SfXX)GlDFB0iPHZc-<ukzY%)Q(Pc(Jq78r~2Ez!T0(VIs
zr4Q~&7ga(4zhF1hVVZe}MtCHIBKriGCxy@z!a;qA3IYQbLh&NVI2($AV22%@LIwaw
z2}H@6>e&Onyd?EA_%<D%Vvj0Si@yV-iTGGVK!IE+3_TPP!>E9wiwa&SjE9ZcTSAiX
zFw>x$aC_WM8M;Ie=G++Rkd6|$D@g7GbO6%b=!vKGF;GO6^q1j@BWnt`qr>-zwTOyd
zD5R`_Jux$|8y!AEb_}1@fGQq@vxKPP?|`5EoQ@V|C1hvB8F2Lj_C+Dp>JxPs8)N{d
zB*U-(H+K)bH&E>D#^x~UP5@>vJZg8Au$ngodq55~GBi|9(nchk6DS|cQ999V8-OF(
z-vKfO_(4`3i2MYkQ6mEg+Wg<VMZ&Syoz5uLio5%xjtjxS{?)-I&&+WhN8T@ZvbC4r
z(76>~()Vi#vM0dIuEHf1_m>2!8_s%U{)XjwF{v$wc|gC}1&QbH+kP>w!n!J?2!%(O
zPVjRHn!e26R-6the-NTwm{2RkMs}dBv7hBdCxLd}!RLnEty?#+B3+90-HCw;fD*X$
z`%5Ni85887H;{PTz;Rau-ly&};+kbbgAL&;;ph@Un~@EDqX)i^SVD#^uH7OsFS@3b
z^I8c{SQz0wLOi(E5Jg6z?ApImet<mpqeNmHfequdF+$oRCjf#D0JS-aXj6f|R_WJ@
z6|Q6e^o$SLED`})h1JQ~`G#IRL!lcA83bCx^)28KTyNC>-ZjX9r4%*ncO#2&CYJVq
zrfh0<-(0LgE3Hs|cF}82-BDJZrvS%K(DA5_r>YW32_+I7-t#J}v1<OyKQFmxY(M<z
zQ@r)!rAr$vW6&={mGrRV)}>32Q1!L<J>Ps6fjv>U?gfrNeZHU0F1Fpeb*mn(oqp*$
zxbQ$al`P_q5E3f%RNTSd4$64%-5NH;A_tZhDkP5E2F9?ie#>j2XuG*!hulW5)9ven
z&fKe6Qq68wJ$>AE@v8ony27UKZ&%kQLSDf*AKZiX-=}rG#A?jm(>E9$YGOvteRtI_
z>B}McOB%~^TNUPU>SVI*GlCRZK_m7Ho3ExKMAJazgCD8~T)I6VD6I4RXSWh)*jUr*
z)j>Nyp?lC%<N#-{0nM1t2=MBhYMVd3(YT`j3g_#D3L+FXTAR;}NR(<CS&mTrgh|Q%
zozerAU@~Kyop8w{3XUjfml$EpV!Ee#Zh~cKXQKn4ox6VI>0%alm3M4+_DnBzH!JM|
zWW?fK_>CF_Jrem*N_!vPRx$}GC6S&jkKcE1j%E#TfPRCL+&YgX<7tSpaQE}-xP>VY
zr}9lAY*T-bpN?1<j;`B#e~>8IxuDDr5#r(I4hU6!ZA(uV(oE3@Rg}%NNiH+|WO^-x
z2jp&LR*I^EoEXB|dcJ<@LI{bV=6}xD3CT+|UO(fHlPtHJZogQs+APjeC=`V`H_KJ-
zzqk-AuS1~ciNiS!69VmTko3R%cdks7?fLT!f=^0@_Rr?7RA{46J8dC`{6Y4DVUMS>
z%bp$xM6<bN#R@m6r{&xz#A)kYr62It3Y_o6kY%R+XufuC+Vtm#CTgBABc~5vNuLm=
zR(zcrw*CAx)5puLZdn1ofXnzj3ui771=7|FuESmp{3`wmKL5X7c-dN5&a-fKMpLiv
zezPU^Ki;-C>(R<5^3C46cW<vWj}Rqu++$s_aR4zWebLGH{`=A<QH-ZAU}?@_f0>>?
zFVO&i+MDhhT;;%ey`@60m#h{#(Nm^@-U}LKfpwnSCv4vTN=AD4CG%n&?^mK2VF#c`
zzJ#JfC$w@B!JoG8UClX$1N=MqD8)a{E;4ogdb<A-$^g22iZx9)iZ|Z+C<DHwrRB=^
z1(3(SmDwfq!{AcY`{TGLw++k1@i4oa1^n%seS5y~@2gAy61_FOB}f(}SY>tovHhvn
zN;nkPeJMZF*5myhQF>=);3Z)OyiDYq@ZfJR#bWxC)Hea$-~aeM+g~)@zyI;eegAn*
z|Mus%RjzzPP5<_%EG(%n|MO)0?FYuce&bgB?GNz@|M8!r__rUVeg7Aj^S3|!?+5?y
z761EjSoq)nmq+D4A;Mjr%qfwIGBdu~5gm+on78+BgQZOs7(hv32k}zzKXY;Q_aL>p
zZ~OKf+&*duPs-T+2>A5R;Wp4)IC!r)M>&0Y-iCiZulV;hD_?#M|37ij|C`6!Iy2+z
zJKYW<Uq7;Zsm@23Q6nH9Hg-=sh!08YQP+7a1OoTR(WoFXIy`3)Fn6Bin2aLEVH7oP
zFB>`NEwx>7-Pf;(7db$HAMyG-W`NNCftLfBb>YV!w~{p$@?wRb?&XCA^;fz6Ke+(^
zzY0jV=d}lAf}(eq&1kmDA6t7N|ANQHjXx~Dc;Vz?-{VWxnqFTjRrLOy*=gU?Rf5|c
zLlPb{PCGlTuXuh@isSCR%@Qgud8;Dtzqoh%!AibY-<{lS@ZGk@Z~Z1t<eFK|)d<O3
z_N@87uUFa7p}uUSX<BTl7$^6*iOHAaaeA@lSN`&s|K7Orw{INCg+I7`^5_4V2UZ>T
z`sde^weYY1z5ol$f3LxRuK_&_|M$<qZ&`A?Jq^0He|f^j*$|GMf?S1n1I3Z+ZvrG=
z2uvb}k%Q)hc<~0c>(;K#2be>I@+fs=oVr<5=j%EHY>izyK>ObgJYphcr1f)wfI+f7
z9CKiESEW3LB|Gjn01SRfQMa?Hi66yYP;x7_bcy_jg1k_tpT%N|1mlMt99Rk|VOYA5
z?UKZZF|6pK<M3DA<tAC1zkJe{Egb+<TlUqmi`&l_r!U8!rdL8Rzn=G+RJKMKIz2ON
z->#Xqq*vts_g4yS%8iOMJ5FI&T()`;M#C~khhEuaFTt;jX<HrrjS2U97w}vEkH1v~
zM~t0&_C(bhq7yg*`{bT64*aDmH*lvU(Nec))*t)E1;CFodt-|5<7tksk?!{1stIXB
zK$CH`w6M4;5aYx3aL~%GHo+9k&%Aiom;y4_6^-p*6>09j1K^Sbf1)RZ0R&jPE?#r&
z(gxJy%|T*3{~jtavFF8nOW9-$*JNDh+m};%T~-#C8s;2mQ7VM~A;>0z6&{BJ<zR_+
zq0Y#ziXBYd2CIs`H{T*|qMU|6T>36tyeq92c;<6Js#D{%YBCdw!xlRCHGjQuPiw9N
zr}^tmLNPbr9bbA1c6a8=tMPFh%?8ptCD3Kb7-C#0MpLCqAAG|zo4<aM3rmS3uAe<p
zDsmuf3_sBm#6<NzaD|TK&{TGOm@1cpUp&;0aeWk?@#r;W!*8+fdD<2>^@A?H-q!#8
zob$%J<K?*y^L|yrJ>fwg+)peoI<pVIJ-!TY;zk3O;F_DA{_7yh6p8>0;V$>>3s`zY
zg*kT)>$l^#``Ce^tSjWQQk5~d>3aw~!{#5q_9Ip0^}2yvKuscLUEu2gr<?1p!YDGX
zoPXqWR~LBZ4ge4ScZ~5WkEFjQr9WK309ucyBSzNc>rFGST8ZDVeUzA(NTVo!1M7vi
zG5+D}+k8BxW@2C=%ZZzl69xR;3|u|kY%vTU@ExA!xzfv@`}*=5mln8kZ2*NwDTXZW
zJUl(y)3)Fj#<yDDyjM^z$-{*i$s0`p{87=iZ>MJw+=tEn{o{_SDdWM9Wgb*iRB&1?
z#=8Z4qytV<#gv*xXIxx!$!~8XnV6kkB5j73na%;1g!Tj&X2*gcXjY@7TYW;3UVrPi
z*WUuf-4)Ia$}w<zdQd4N5)<(CGL_`<0bE@R7KelA4cuLKsc*MHr~(dU%1eVv1)0vu
zOXKl+bs{>eorjNT;~X`u`kL}zKWk|Dcmx~kZF=y5Z;O)oh8D)n`~j_~@Yajk+z@5p
z2kuAzdf9Iev`fbm506(siZ7RO;hV+ph$uNBC<AL?J@8*e8T48H^OwFo&*e>+ElMC1
ziGI00S261=2Uy2Z64T%4hL;Hs_;xy7t4FU12;|d=JdXp9EeAObH$ek-22^)b;0ArD
zy?=eEBey<(BGGDr5Wog!z@Yu1bXV&hSAA^b(9zMMNZR}kPv^13zYfJ^%r>1j2<&J2
z;ql!uJTj|DKVQXV|GBLk*7433StaNb{;>bdBaD~A#j&~2ENx=^0N?EN3#s}2*S|iy
znF$mqKI%2m>$Ak`W&lCaCf##0;#^-f0JrBjn8hRTNx5<qR0!60YgwwAaKD*e+OGOt
zM{sD%pQqt3ZOC9{IZZa0`h|4F#XtYKf}xFr7H@*tuc4}0{#v}Gt=E6@N8Jc4vA07=
zU;oaCKH<CDpwX@Y*ll?t^`R%GzpoIVIq<6ZhvOZ<MQx1B7hYq=k)Eml{By<Xzct{0
z>tGz?|L1#wXJ#iTSmM#jOM|N~6K2R|f;&9V<^!F}6-M#0&ePK~0ZJ9H4^Q2Kb43EI
z>9G~W_!iVJnGp+1r|$_O|7ejzOX+m(v<fH(&c(&W;nC5~m^o}EDn3m1^+DY}fcbpJ
zh73G-96nzIi9U9-pF+d|0|KsVm<0v_o|P{J1J-N?G+#H+7z;2HVd(nOhjdrK#QZr1
z11ndSqYe{%e|u4+Un;;@YfK!N&?S)du$9#d^wWJ3f7M3Qj-Ub+mf4RdUB<w)>DEZe
z#mkrL-t9y1VmanQ>m6dVhKn!bEu{j*MDcKc`Pta9FWKO>A9j<wg2FEb8Oa(9Xbrj;
z^kM!`0OOuk%Frbdgw4x_4D3tXSopP?2ww*>o--bIe(>E}-o?;gITi!<xG)%HLutds
zdp2Ys)o;A}?e_&RKzaaa<#tdSYQOzHE>Av}Rntlg{LN0JBKdy2_w~xj6a8=Vfdhk4
zL72>&0R}CAo6`qL@u`bo6oyGy!box@v^3Vqz^CmuP*rz(UJ?tA6{l4|D{=zeLIRbG
z7AVC-pb!*bu4Evmi#g7rD=%-`4503)<KhsYtc8F2xUklAJ>6NnqTpPo!`u$z%>EZ$
zu)dg?s?BvpOxIC`9IN&-R)%vwRW(1OP>Y?fcFr99yFoN(gUO&9_In<{KxZ}BM+q2k
zgolUU!w1ht?Ee6Th8qT`x|aSvbr_*(gPd>(df_gdKbmHu_+En{+XFZYy#mjT6SMk~
zNCt3Xt<kP>DuKE%-A#t-0Au1Hl)4E>)R*Ca5sf($ggQI`=k@n$<Q#>4+a2PqYhj9s
z1;ZD64LdqJ3o-Q4`uwTw5Z<W^Dk6_3Pb5Glgd`f-e76>OUWecoU=2Ao6T`leo+LUp
ze96dluW$slP_Yidn54kOe#95nIwxj6?hwB}c&k%{z4i!Pj_oHhTTkIaU?j<vlEc88
zI<G&pnJx*W0<L(-B*8C(aF-3QuK`PdQ&|{`9Rk0|8nYc{_C(8@acu>X3z-$L>r%pa
zju}ijo{BmwDwUIydj-)D8y>T(Z~cUL@tDf@HejBi;m2suU-&|}s<5=!Mi&MXLrm0%
zt6OStz=EDX-z)}d&1L)%4g{=P?#b<z@9pKb>3U|?WLB6D22>Ee6{;USj0XD}Quctr
z!3CEVrMLbf0vK4Wg*ML>m!WD3#ed(4OSmi-eorbD-m`Y)^1FF$rh?;bOzb);`D}Xy
z`IbMj5;b@gz`3Ydy9C_9Dhz~GqqtXy^u@!GqLS8cMGXylK`9bqoI5*S1sQ;1kRg_+
z>PAnzJFy~U^JaN?15DMI**^g}z%WJhC?@X_D7!cuN}>D>#V3+ymU+n-<urnFTxFFr
z;$MZ^X=v#JYJ9MaP`k&H`RQI$LC)oOwI?_cM<Pc|JRFB{^pAX==v?^oAbU6TdXWvS
zi&6B%UW|caSR@5^&Y;(By0JA@-i?TL%{MBgBf#aW`fT+UBj`(@Dob%O>NNRtO3?>a
z)~H~tt~fH+t}(Z54E3{oV;&{}-aR;>E^(MQPlpZQu5#Gj_2z_rAZ5z_yl{6gaf|Rl
z!5a<T58G5Z1jIY@TnzfGc^BskQ5*D*PXA^nYfsN2%Q3FLrgUD4S=QKwaJckMS()3O
z*tF*o{O~yXfJ^Q|&lR329#}C%n2Donb;B_^@?@{lDu9C#Lm5(_eQdyx#C)bpQv3lf
zmqBNsWsTAy%$N00grr#HPL`MyphDGxK~1_|iUbrK+~m>=8!%BbkTz=#`Y%dyt)*QX
z%+%ajBbOZ7(|tSpPjm%LAH&!PMhhMmHKxxFKhs|zuHI^#*5*j($fcG7zLJts)?|k-
z%(8#uh<g)!F-XD;!=3_=D}Dx%Ac|p+Cx!{~`2$e96+}|uk&u9CR2rOWgYdiQda25o
zeGlPO<m2<K0~av7x?$0B!Hol1Bc;~ZRuF`nHdeI>UB}_CJ!j8e<TR_?>E_QM$nutM
zKwF$5AO}8NA$?+GJ!+VAo5{`Oc~;iZQsLg{Xv32yF<2r~8!ROLL!lSM^7R7EH&Dic
zI&69rc@25w@>4d$b{ryCB7@<<QVbJbr*`9_$J3{h7?PS-(SrD6c|J#Pn(w_Fzme&8
zVudIlv_7&%F>em!9u2S^NY}&3ubJgTuo%ebjfGj#r5gUta{_*rAUYKyDyT+T;@r8Q
zi4bSnt3y|ZE`4=5#%aQ1mp9PE{x(otyc9On!qw~$HNgA+;Bujl?wpoOSYEdL@-2r^
zxJB5w=10z?g2S)3@Rp_{B2%ap>W%<<0p{r(NAvqfrMtrXR)Jt!YS6z{QT5?a_Y3ja
z?(<fL?mZh?KT90Np80ajn*FNC94^J7gASzaW7uz}GzihnE*Ro172IM!7Nv2@-rm@_
zJPTwVrpEl3Mv{6fx{K&Lk`UkAL&=@YT&HS!+@(d`a}P4Y5dBIy7({+my@;X<UlC!^
zuWAv?mBEc)I;~@+M<JN2E@c-qz0#0?MnfnTd=TeeH1ynH0Jx!(JB4fh9K?|!-`1J!
zmm=gmN1%Y^b`u@H?B8PU8G~jlKQ(78W_KZY9INJM%I5tT{m?tEf*suTWH1!bvUAeo
z0z=qWc<PwXo%TQ2B^&yR0!eila*yAT{%eq9od@j(ZK*YapxYT?Hy`vUT1IfZfuEqC
zhB89*<5O>9XN(qZFMH?JhE^=aQad{O?gaZCIncS3(Yv0a(}qir5inb=nw1*#ahv*m
zpt9#h^1C>@G<+!?rG^AZuzCCI2xW?n-V0g^CoY&qzX0_@hm4!j(xB3z-hrYj$Hed;
zc6ykh9tz`X#cqiv^Edkl@}Sn6d^#muIoaoDTj%{5(%@!m!M&b;03(L%z^`S65HBXX
zpr1<?2mvIC8@`3^j!Gu>T5Pn)H!BtDe1(I{iTR})r9j;HwXhz$_v7mA!D?2B426Ik
z6la&d-iYT7<W4TIpj;4|PeBdBjEstXfT(krnE{szg`0D?7cD<M{i<<zcvuk5P0aHS
zQ?l6SBn+tlJ8qjRSmAt-UGdCM2F|lTRbwcP_SN$)Kl<rq8XlAnVAGY>6|(F4Jsgq<
zOl`5*<0|;!UfT3PrZ!k>33y`Y#ABu{2#x8_gIOao4Jmj;weS2^O-I(;2<#9FD$W)a
z{U)bQg1<)mv`W|WzDM>sE@NNi@HvkM+c!2td&Lc!yPL+$hjnHUe1$d}?#4YwIhb>e
zXY(Co=6sk0(@t-CET{~m_h+x=w0L_+f5N^Z+ml>J4r#r*hZmgtna&PAn_WsPIbmQA
zorEg5OoyyLy{G7?YWnB*6oj4sjHD9?H{SrpEwr4J$`L>}(3hr^`w9gFg((6Tu=r!J
zGIq8mHD~Qot{WE`09uUi``UVcd<!sA8`U0rCtl%M@d)@f&Y)0)s7<_FwBi*ASWg)<
zh(Y)9HA1*@aaj+K%lkvlB%U*oJ(y277sBTIuu<4cPjvqG>r~y%)H2FQn3`-jH=deX
zofbzGq5Y7P#z~B`4wp{<xx(I^aUHk&5DMwbnG7ROCU7(tDv9)KA&wJ8Z(vYE!@QT=
zv|WR}u(ID|?Yeb^+n%aEVl*Sn@wCnkSXC}@Y_Vp#(Alnx6^F+@6Q|{Moyzvy*?toy
zh?{OWHZTlccxb@qEfO3^V`||V>eHS%J`|Yp?Uyq|V{>cc?xUVAgv%!nSiHBgW`^JB
zcdEe+nNWvhrlOjqJgxSScSYve1Dp1r(eTK-3P8IUo8IBWG4>Op==pC^G;+U`JgTnt
zNm&`8ZOjaGBJ#HVQs<w7lcoy|Qc44dyf;<?6ObV9iXAPyBmqx0goCtfyEM1A@%CMN
z_Bg>`N)k_1zi-ShXn3@tv(YbvXRr6?UX6L5gaRneS3!(-v0;cpT<;tm*YU0O116vY
zZzdi;quF7m(BT#0_~Dw;WZVKXrXZXzd1KG(0^BjH*Lf;0^a2EEYaqUdV+R4ecy?Cy
zo5AT;Xs*9?ZlCt&h!=<Zc@*U2f}3-vGeL=d1v;T8qiK$f<(T<b1jb#qe#vbF3vWg=
zo`SiVPH{z*pp0V|LJhAz+V$G=Y={<24%lnaeDxT1is9|+P?UWMx95)zoutwfs-*Q+
z&6bLUjS&8J68uQRSuP$Rs)WZTGre27#vF8}?2~4Vsi+?38%xMtdofAW3o<j^WvV)v
zOL@Wdt^?n!VF=t=5$06nc<Sc42_u#6g)$pe*ej4nw9!=@GW+m~9zG^u(ECW`wb4WK
z8UVJz=oje!43i`+6yVi0MYvp$`gfrrjG6ieQj1)sMi5<KmJ2Sx1)Z6w#v_Msg>G0&
z6;gy;q>SCElW<UssxcG;ch(wQRNm5}4*h(3ET|m;!sf+L!y8ejgU4}fMdzM9dzg4~
zU2!A|DYt;uc(b5a(Q5T;m}zRE!D8=}3j+iAqN*w<Nl#_bH3B<U;F(rgx$@kTy+s*e
zxo@sH?AiPQ<ztCpK!0Nzl?m%1gK4m&&J8GgA;h%vZ8PYPDDV%3iO=qW^68;97?r&0
zxkPq6VGQHEqWZZu)G8M`fsysqXMJ&nz3qp_CUX25PS@l|!%j#8DqGwio})@4rdDpE
zCgR>bq(EKH^HxEH`%;@O_1bUV44CW#xbkiw`xJs%$wVckR#5h&q_ebOAY(Z`|1mq^
zPcIkox+<pRu1&BVmq<)6AP$#ZknM|U6>7>;jz>qrue0D#A?Ip}!>H6msIlsT4`;*V
zcq$9a>uq0aHS=rOR6r-h*&@;=k+fUAQ6UZ&)ZRtyMkI-0l;V|>^mkxMLOWp=Ny^h!
zD0J_`lJOump^g#K=xG5hlzf<BmE4im1Ne9^1A@{1&?ZVVnGAt-{jQ>3*T*#o-T(#b
zt>BKv*%(Cdd;!$G$tJy#5!$Xo+rdZpnIeWK^M8K24qS*hMkz?4xr_mv@o+arPADQl
zih2@W!=RH|C11wo%~}Ac4}slg4eRTq*p)-M=cRJsB^Sww+oR@hQ2W<M`v`txEW<4*
zayI5(bJEhJ9Rnt`JAtycbx%ZCk6ajP1pE2vR!Q)la*kYc_z6^|uMqLZ9pkMI^J!hr
zoer0R6B>S)5B?@K!s6Ks6RZg8!%ABXT6trfGuc6?&2`w2)>zV<N3btGFNa|d2~@fs
zW*0110tN608FXfS`u9)r0q2Lg-y(Oe*Yz`#xPh2Wq0kyWv}hZj*I3fC(K>~U!w@TK
z-G@n2ZQ3)Stn?~_An>lk;ijCg<2L~jX+CJ>QJi>QlrZ4REKY!}MKD+6Qz}|1xr@&Y
zxo!F2(0ZgAC^|L#Xz905r49Xx6IMc&o!xLq>m*u;k36Nwe0!Oeh0g4&(R4*mV~7Tj
z!buHo>sV%+>udq%0)K0q<0#H3;}xMuv(vYB>LD8sz0sIEY(M!%1Ag%o6HrFL?w*6v
zB`2O5x=A}BmQ&HD`5YonT(xw_B~XzS+kL3`(|M})By+_-SYC>KopRh3)<@Ya6>wBY
zMx;O4c?1RET9CqJM8uH0lYfoC4&GA;H%vTX5wdZ%zL43!g2&TbTN4<pS|F$E-!WVY
zK7MVf=hLT!Xd){d;}L3iEtI)i8)#?)gSWCMe=E#og1LP84HXKGm7&aFV!}}pnVVe=
zNbG~BqWXQPP1>U@s6BY9ZnF!_M5T5g#7OGY;TWQT=$Bf|kEH)rIuA(wFH@3=3nY<a
z{ODBHS#-R;Pjzf{DR{~TGh=ZM(js%{@Itt#ql6CQTD3L+i<wV%#ho{pWF3kcv-N>%
zQX#ao6|N9r&lq;@My!Jb_grQNo@XbrkE<@jccjQg@ie-I;3K}^h29J@F>4fv)WNZL
zV*A8a6cJAtnEf3^CVRq|LG$BO;9@wxY=-867&aY6my)8Mxg!#V9N_X`t~dik#gREp
zh|Q=Hj<Tm(m@-Vt3MT~f)=gjg1Q)0zPa)Z8qvjbx)pc-gG)z1nB^x)jzl?sOW*e>V
zL2u<u*IB>t0G>BO<P4hzH{Io2mD*%}4&H2g3F6X2)$UR|upg@KTTP~UtmOc?IO28z
z5d;%hu>Ug#vkL8#c>HyuClMGVDaTcvKpmnD0mKlb9194|j=)>yAKTlj_3*T?96PZK
z6(oXK;X_Gob^^;V(%FJRD_bi{Drh`VGt(1IDNDbTngW6hl9_PAy^890k$$V-+&tLu
zMyiN_2`D28{jAKX&fe6_Xf!edz*QhVTPa>`;6RjB{X!JD&6c%l37D-V_u2!=C`c&e
zRtf6u1hxp6XCatp0hyq79Q{mmxiWIGeTu^?O3ExO_f21|lsviHGt$BzrNT+{?s5tW
z*ZiKI4fy~qj2m`iXA}?NmZdRWMYQl-khgd~(lQe(52#o@2{f$_>D#PM^KL#QXCDw0
zTp)v~O`n)zygR-$A31cPul@`Af{nEn<edr7)e($`pmhj_0&5^@xJS7U-(?*R_IVWx
z;7alVB%;&?1<b7KqNKbKdZN39IBFXx2!#dM>?YI$CEdjf7k<WKJ%Y~Gqf5rVubnbL
zUP#a^lS8NU_LN+6fSj}%qh99_Y_tqIo6+@{p@EaYGvLX@eP%4fw$A~0bb5)}GU&r5
z0QMxcA3=uTd{ZTN=;~{CEHpl;V3OG2kKKRF#@0E8Dyq#P;4zYb628LV9H2nk<2n|g
zlGnnSxu~qnA<1cWQp)S#CK}Tv^P?(`QB6OLt-z~lRV}1FnFJ4`=JgV?xJEcNrMl19
z?7hYF>?u?&T0jfjP}(nC;onxvyq=QoN7A1i@6)JXxQ%_)l<si1Y_ee3$nQhy^jC2X
zv%Q~zvj;Ltac8vf1=0=7+)$vpVFD)UZ-5ecySjDT@5xNi5oD;B#|Kqzx5gz=z1i-W
zVzY?IdM|SbUE!})eLUnMJ70@S#Rk!(J)P>|P1aH<ur{)sIeA=eWF*SZf`Dqm;#FUy
zOVY!I(V-Zeel)^_!EP{ZNZ^_OTuA-SdUW}oOk6RzaC#Aq!QY1F@(N=IEEm2{co518
z0a&zh;_1B<Q=?{CEq@3#k4=w3>p0dph?9(mF@q<^32bjston-IkV(<6PK3NQFDfXo
zp%9gfsUg>Rvub(8_Kz`xXLv!+xC~im!w_*j^uzqwkg^Ai`~Jyt+wR-wT?lR|1!>)(
zz8_w1aeLbn>IhX<KIq;|I3Kzq7FVCRI~^dmd#(6P+X`N>l!-yG;x|wq47*z+fg6Ox
z$4}G)zopx_Ty*pu!FY+0Ln!}+Wdz%PyNuGsHM7x*q($tlm*=h4u3bx@ycakQvjX2E
z51skek1Px=(7IAOqpGSvV5>Y$Z)jv-H#K@6b>og4bUe{LXx`V+<)Kbn8)_P*iKHF2
zFnG`cnZgV?v(~zV5<VvPE>uWi)ZtNWDH|b279c~FfpdKD0^9(6rZkeuy_W$5d>^_w
z{E_O2517$LSo{nTv=Yr`KM%P%_*ur_N}R&&qabAlnn`s;plCI{1sx5V>;ied6vR(6
z1$slmB?wMTvj3?Lx$;A7=D-7FihSfyZ+TV(B(cBfGUKWda=X!7u7m>e9^NG%$xIu~
z*S9ehg>DFdISH^tW4J>omo{E8AD!EOA4#o%@+bjz-z{2l5^X9GUzA^5XU8bR4dC-B
z%%kZA7G?#wE=)?Z5H(VoHg8dkrGnwx^Rd9dpl1aGVC<>GLT$a3&_&$?C{4H!oglO#
zOcRG`m6`Sf4)-CcE5KRgBA^JyLjYN%fHoivF3d+jmnb9#1C5QawQJ|DUH2gkt3&HZ
zDFEQsT8bGIcZPuRQiKgkOgvoJN>J0Wfmu5TS^+?#51PCoOBGNy0lu%a7p~VtO2TBV
zMA+kTDLyR|BMa9?<IH`oIArUA<SpYqDu|^ys{m=a;?3{fqxZ$4B{8fiT1oG_@C9g^
zg|hep^Y+CPX(&;(QR2?W=VC?h1U~NM|6=bwqpCc&u3=M-rdVPPpnwHbKtQU}O)PYf
zvgx3LpmafsfIt!xuu!Bofhe$%E?t@>AV?FXh?EUcDbkfHAn?v*a>n?c_uu#L{qZ<X
z#<9TWF4wiLwdR^@P8(m+VARsp-GUT15MwGQ(9octPUT|_LA4kWYFh9DwS%g6Jhl=2
zMSt2E$Q9^*@!)c1+t7A-gTwVIhKCdg1Gt2V#Wxu2tov{XNm$QL;fRq_R27eIhJ*m6
zf_CpD@yAu?$yVh$#GguUcRj5`^PNtk&wYMSXwF0GM+BN>ehydU)O`MJdeV2%uX{58
z<sJO<=T+DKg&hC=<0_ut{)NZ>{fnF5{FmJD&tDwb{l6dZ|L!wb;r5#2<Uamk#px60
z&)@$b{q-?vwvEZxjvYCs^3MFo56Mb3Li<%-2&*62c>D+5BU)eo^uthd@^McQrHvbt
zRgS3aym_+aRWG-!^Fx#ISrzA~Ba@!;TyxL6;^SlL$Hsi3J7WGxl>WO|&VIM{e^Pe;
z`G=c)2mcA%|NXjumk#sM>Hon1{`a3_zP$23h`|5;_5XhO|ND>SrbLCn=ht=rAtW96
zRksZc<Bu_<9<Fig@au!>Xc2~^Dico!Gww(o=HIFDJY*W^q5~2F<UbTDGO(m>4uD0N
z4LysNB^t)Ya9Q)c6Zjy$w<^7yH@E!z9MbY2;zBCr|9lUV)VN{w{^U$7{xXAx=5_ZV
zyA9ti<(v$FDx#-N=@1T}U(tjL>me*$=^{-oF2ntyWr^m@KZAZ-ktr$d(q9<=B{^Ce
z)V%-u*OC8Swg0=B|L;YCCHwy~#{K+qI{Tkb`K1_U1iHxR;4*pl<WG?QQ5xjfB?!#>
zXH++NosnYaG)Tz`TSq0;=!<*jQ1OxDk6jUT00+~N9!a-;bHwN)3`t4DuznGs%YMxS
z*&%~H$teJ5NElFbZKcg3A$9<0QgpI@X_bOr7dWi;`4wm}LmA$SKI6oZhVYV&VqeT{
zF#l=i6ivp1K*OtS1A}a;xm>b0HmhKK5re+T$fykM;dYjeNx@qhg<op#lb1I`Y1$6j
zMOIP!U9F>#ZBoPK_B{SK#@{?c23(Pk#cQ`O-pr*c7CHF~OuuIqJw!)*ZSt4eBw3CW
zEeai=WVAH}U@;26w9sUdG=Z8=I&@{okqaqoX@ol@_mG&|Ud$qCVp2}j^!}RX;-`;J
z7@0MqNh74LFev21xIK9{LCl9{#lh+Ad;2!5hjD!MRxU0FiT6_S^75RvR|+IA{Cy<O
z_A(%<8C35~%|BLc*bjdi7^c}h8^iQ5nLk0T`Q(D`+DG1j(R>&8SbZrgfSV@oUXsy3
z{V;?^vJRBuF<2jd$kB|{20(1U3o6tF@Cm(~(p!gePcOh6x`-~=o>j^iY)WjW3xy9C
zjAi#j&JZozX!JVLzx>a4$?lAhU!h}n9M02N91t6b$CXZlUqE6@l3j~Hb=MbuA-f0Q
zTe?<~AL5sFp*Ij1n-*pujZw>Q{i^(z$Db+hzby8Un<}tiWq0lYGbG2-?cW^n9KpkD
zJd(ac%K1E?=k>ud9+Sxc(}oT<0!v0P<xO=ZpHCFf=iGbIw3`NEE8Y0jTq4?e?||=<
z)6apNKCw-E$r}hIy~#5#w6+%TWwH#$4J70I0At<e;K=kd`i?Ci4i!KXB#OHeQgP{I
z-wUEcsrV23pm+1x+B+x)5kDQ7u~5YKgQR1MX2J;G2q`^qGC5%zN}BPgC)LR#?26VG
zm((DJ0jR?~AEts2Xef3Em#yQ>i)30V7)h4Ns>DZ-`)%2xL7OBwR^LLoW>KmDJy1aK
z?y-a~&H%wOwl8z;zmCcDGw!?^JGCZ9x<DsmpjdOs-1y7%iH6^&onb5iB5y;wH<@cl
zj%8f~Nb%-nl{5gFlTe;cO&3rmEi0^JJTAQZ0%&bPr(YK!Fu`Xgb@>3T$SoOD)qRuK
zf2sP-dXT2}pnHV6Efc+8X*CQ4XW+Q928fW3`kJHdha)F`WM{W-N!`U7<HjVp)1^_U
z>X6lO+&FYda<&PXobPlG3r||)&*-!=0UVOZ$aF&=-4w&=4xME9hDP-Tffdw!=TQJh
z4qn&<UF%aQnso5sv`yUqHv8+JUHZryBBAyq%+4c{pXkr)9i54dw}2njQgY&#i&$qb
zP*2$xp;K;%W$S)v6t&I{`SZJ#vHaU`l*hk(i<(KYvEJ^&y75zR!)^g4GVt=s&hZ92
z0cF5F7d?HLzjs_Y2<pmnG-z(#k=+2f7zCgZ^gr8yHD><u%c|6xa<n(-O@T^hU+9fW
z9UKBqPBE?vki5Z8R9^$#;rVD3RkrvyQam9u0IX84T)*)V=ov!xQFv(lyK_{)Nw@F+
zY}gB|H?Q0mjUp8A8F;6E$D*QavNFaqAV(!Bxmk>oJ6}@hhPEZeI3RC_8wT;A%=vri
z_5+)t=1U@t;zFTk7mlbn!~EbRwl-QjR)tG|SYMJ!fT(3%GBf?@qtH{mo6Z#}4wE7g
zo<e?kBIFWP!uN*+NqlK?wBh%HB4|yJ#1L8?k!RS1xT8MpGCoI=G2oV!8q4{UXH;1i
zqOLyaT0s5ZVbB9@%s?(VBT|MZmv1PUrq=xMy=O0is5q>r$h}y|HwP;OGLiAP?o@K)
zM7A^6`uCUSTRT5q@a2G`L$*nYdA>J_LXCYfG7!RSSxz9|8<0~j7;R9TW9St&UP7NE
z1HI$2ZVg_Rsdb$xS_wL6a@(!`9K^|c1_1P^Tf(D?t3BsFA7iJ70jQ!K8Ahrv*dxjA
zuB)OTwcb7MrZ2t21c;_6*@o}&*!A|X>u`9@xmI;YNHr?Ca$#=(?g%vDI709fUT&k+
zA~w4O)|hTPZ+;i%JwFR;-Ao7Fm39h%&NU=PgxgJ=#(xNXC-lBT8aJUp5;(XBYg1Q$
z(wM!*pue5Y)`$hNT+~112)7rkLDdQVFEsG#RL}w|=vc=1kmuR|Y^yvd4l;u7;FX5A
zX!McqALsgH=co$*?zeB=h<i?-503cakt+gr@%0#-ixEAHFK_L(`q}e79Cw$O=1vu_
z)@F_1;fCJ%RabcltNF9i?O12Owx{_|6D7ES1oT6_fzs;r+9t?)v;qYoxfzm7f?sDE
zsiHvzr`WJ|0?jvF=%C0p&kpfb@fUn}y}Wtj#xw3u9z&+ZP~9hj11v8F4jRQtu#T`M
z%Oq~{OY~fxQZL(lG>Ah&1xvtMd(ERRwtw0Jn4`78B3u9yz`ke=RABKSxdi<E)mOD6
za6L^i%;zP|8Mk6Mc=m)b>Rdqm$OaB(;99<V^=cb@D_p!EA||B(3Q7mf)Dww^1P)A+
zJBLpS$=~)zlcPF12_&!Xd3o(B_G8T*MPYbzi8Xa(Zwwgn@7zx}bE&~cR#%EsmoYP>
z3yJQyF@;a>zxwLb7psJJS$#P!2xwFX|H0hit+-(BpYCjaw~-ML+#G2S!5GXSJ*0Qt
zn5b}yd`hAJM8Yg2FA4n=a=?A?@y+-BjdAVH-4J-XjZmB1QTc9_XaZVVuT0e;bunI-
zAXlbXhc~y$dykSn373ODWmmKU0tjA*GLoV9vaA1{(%D%Vec0&^YYgbr^m!@ZJ8fX{
zym=w}>-0$o{6xvjB1h1LKwAYG33?f96^LAQVf%tBEe0E6k`NZ91mEHq(#NT2Uqp(s
zAP~8yCERlZPon}YM{sZQ?J{6m4y}_U9H7#&#9$yxR5K`0WQ>s#r)uxMRXiXAa*?R5
zl@|oCGFLl#{)JY!SZlz-`ydEq4+C7Bz4QQ{$j`~4FJBJoI;1dOQ~%Yfvj?iwep!U-
z(~xr~0@OPs5N!a73P9*amqLvc1tF2QgNvL;NSa7vYTvf$ARvaM=oL8!icA|iU&Pac
zj(;3i6tC(wy35X73$kf3@9=@&M+K@cvn;V+0*27uA+83?J7GNd7M>hJ@x%++FWjT!
zTYyrjyp3Je6Y}3Oan-6zKjVdwU2Q%KR-i}==Pr5Rp^>yuIPdpUTVJn`DC7pY>6@^)
z5L_Vu+Z6|w?jQ>mIFRQm_3ozo?d+HZuqQ3&4<=X^JkiLhlw7?Um+@T8$jA&c<Jm|m
z;|B5}HYbM$o5l#Fg7=}Z(_pm-DkNE!rq)z;g8b$R(}Vg&Ahrf1M};KB^qB?N3#5sU
zvv$8S38Rp*w>$5VRTrv8e+t9$d<vWIBui3gg2w_$&&kCKiw2Pj8FaCxjh5-~S-dxj
z$4=tl!HH9=ehIh|{q2w-m;iL4ba(b9Xx)(0pKRy~PQuaV?=+R^0KqEGWqFfu6?!e4
zT#0!8WUfjycoIH}TY^323v@9^83dKWEAd2}%_D$sos3SI|I{RnTpx;rHih^yYn-F;
z^&Zq<TtKkdx@F7R7#Cv54L`qq1<Zl|7YkoLeTBVG{t|Ixr0oR-HN|=|7o-4yCX%Q4
zsqPi>iD~=Mn$3?i^Mj+v!?6EZe3un3I+{g8SeWVBDc5W2zi!_2Z?UfQ2C{FTHG`2W
zzkh!LilJ%$=H$o=<(eOp+6{1HO+305P&~sGe0`t+XSQqCE>E<%<lgg*JNlhDkFO=A
z#r53snaDu*W0BKi0r>}ak_>^g)Qmq5zYUy_#y=pLdP%}=pwW}C8&)v`l@s)M?qZjb
ziyAWjwDdBbu~)fA;nUUvm3!0+X-Gei_a3K=4vux&5p!tR&ER++MwST8u9GN$NXLy<
z2QF711%<Y$uVXaC^jI-2*xNbKyYwh8hx(-|V#?&l0~f8(9Zm+kY=ff*$L1|~T9TGA
z68yW6a-^yTQF7u#Q0%b}A3hj8xLiWT0UaPJN=heTT0okOWbg>-w$s-qF@}+jhs?xQ
z83B3vdO(0K&=r#PBtcjlOz7!rU7ykvpo|O|)}J7gc8*T9N@)XOff{UgmIaxSO5fy1
zjp&K<M?JVPl=lXil#ypMcKk&H4fG(P)3X~2%nY0*ZY}shH7U~mzyJP<Q*8z;$af$H
zk=U#1{E-EuSkvH#8uY+R<2-O_hu))*y_yCQh)xb%UpNm~o7P}8kQJ=r$YtTHpc`T}
zklXMdkl8}OptPN8q(bz=FP?11@-2iSD+v*!JZ(ef94??QRKIi=`g`eErRjB7VKNHI
z9@2`8hs_#t1ue+dO=GMv4Xunim>682Bo-&K_d6KD7{S*f(CR?nRySRj4=%z-kiOka
zOgxo7EX1mKNDB`s<w?XMl9NJuosRW_g3--`al{MM4^mYBw_YPR8}g;Z<0K;}{J4sJ
zO)Y4m#AUC7D(jmFQSef{am^Z7mQyOr%A~vpzH2kyn#*LLGQ97GQ6S(N6G`&-(W4`B
zpcdojG24eP4!81=c^Vj3!K7RcstxJ<+H~ZlKqWiwj_kV1=V%1eljFouLT*E-GU(xK
zg-8}fNs?*9%kk>#m9pkVd5Q|64Vhe4(cHTiM{!{ligi%${;a4+Cb0Go5!5JEL|X5a
zyyFp<BTF%Ca-c#eX{VKQP;~7!k+3-nCK5~=ED328hj<nw{tBtkqaa6SvXsx7glL&F
zvGquY4ku||%;lP}43C#-IyGY!7)+9D@kV-?URe}`o3=&UgzmIb6!2UM;+hr|S)0MF
z8e=*Ydfza3o=?uUPpYepdf;fNi|&e>3`y#f#UwchSayN(CpvnX-*I-2<<}UvZpOV9
zqM)Su60KPh`NudG6|Q8l0a<z7^hLkm9a48d-xxusl43Y%DdMTqK+=<?zwlioDUvb{
zn1q2?&!D@5p#_Llq@kz7aI7sGOGW~YoH!0i%%7+k37J4qsSjXMH$zVlTjghP>ojK0
zz4yakcd^E^4D3nd8C^8F$@o(8H=(FSCp>@?K4`s8c6|eHZ;W-&<}O~$FhQp?1!~UL
zFa!8*!v_7J9a6eTL?AhDKu5Gwz$KhXv;axC58`5X5&fP*{~vxSTA1l+gS8oAwI+`L
z_wV;jX5wnn;T|TLA?o+XA08>?paG`i)6u32X?y_o-g)$g%=ms(5~G&vs#R_nA?mr=
zH|+{i)O2>b7W@uWy%AFS#FsE56o3qh4cp})0Y+?oXSy+H3?{)ZDE-&BG(qap@-1>u
z_cV>!I*42y1Z7XEvZkh{AOJ6XJJitK(gJ{i7}iBYJXW;^AD|X1Uh8ZEaLaBMO8>9&
z6n5@>q+N^{V_ekbg#ox1{nF8fxKu5m6sd!zcqxJC@3T|&7){1m2Ws%t8xO8+@@@A!
z!$y1M15``dFzw=EG>6=lour3V=Oy9$J0R18D4<p0>DuxMQ7Hg+K5X#Yb1GTBhVJZ-
zL@RP#U?&B7OqR5aeX2lNVB<C-1E-;=sgFnpO-pBaA0TzIrn`19!_c~joCBV?P&=7!
zoWQ;g07dJ*ZYroXt<kXk+xX*;gWAiFA3t{UZ2^HfKRE`KU-$1a5C6*r_y7WObYu+l
zO^dUS`Z>*+;QsPMz;sGYtVI$5`hmX_V65W0`}%!2Ql~F^7$a{|TgfITq)LmVq$HM5
z4)_=Ab2zg}B#b4CN!uZ}`0{=nyBbX_j8`#jIsMnNn==w=XY)TqePr0*4*9{B0sv!$
zrA*LEp7c8-<e!mpzyO%!BB~n4j>KsnGi0<ahf|P5>3Q%3x`)1Wgz~B02hB-5vyuQ%
z-Pv$NX(-enJF#O_6QZ&33Hzo{+5wzcN$X7(jUeo0ETFT0l<E7G`?$Xq*6Mi>+bUCR
z-Lbf6BI$6$1X|m<(tAaG=2683g54iCV2UMYf&C%WzKD~d8c0(B{$D1`9p@jK?PuTO
zB-fZy_Y~UeQ-e#WhClYXkMr{}ky;qAv_L^G)UNOX#|F`BWCARB+0mdEAt;9<DH-`{
z<f|-nKs`(s)OkHob(mbgw&@f(La{Ex$<fAB?af0@lMgt^IB-@R!eqc(ZEvW82Y@!f
z22)a@eg&vpw)D;*aadb<FaWT_cH{>a*n|-eo*#y8W_=<_(^X?u@pbdWiBK&lXj>qu
za`28CvIamQnRpuuxEdQqP2sIYu-t%~6fKQ7vICV{xvgyt`W-XT8v}_$rWIQjQPY}8
z`x7dP&Dc0IhNtZk-P7a!^q{O4xB3k)qKZ4#D&($f+=_XzI0g&7l`~;`b4__+GLi0s
z*CJU_gf#Lc_R%h@yb}$5T;|O1Cu)=i4^1rK!D8E-fJ4Ox5Rzgdhek$h%Tbk8Bc(81
zHer{uQsK8Y_L_{6SsBp0hQuW{G9DQ8Tf8V{4KvwWz++KyQXff&>U_6w?rFsQ9X>tq
z@=CT{LTVFeQsQjb2A<b}C-VeAy`P&au5&~^#~hlNF?iCoq%`Nb{gg^obv0G75|aas
zBh#g(kTrgbMz^!4qf+OZ=MQ~-D-W*S1FP28g*v5OW+NIa!{8GHLOeIsDTpTtXRgqU
zRO)dVYi+3ORwKoC*aGjzGfT+r&A~=*`)0!icL86=GL+`8ad*YfK+bTJN<;y8e~S#y
z1BePb#_FStQVYIV`0y}90Fhe#`T^YfGLuv!zicp05$a!oZhOpf_U5h~5H<}YtY{!M
zahici+NgRpnkY%aOBM2F;*j1B07Dc<V>ka>v-fyAaX#6&-c<exfRryC9v~dgs`bk~
z)#^1x%=gsaFt_vH>T<O1>6Nt54fjf%-ul(AI}iCnA=hRe;XS)||HTxm(qczj;DXT!
z=;+&V444x8LF$v2l+F2z(<=5Bv)CowLM^?$y=V4#8(oI8Q;wU*P1QH>g{h&EwWU1F
z{ec}om;N>6Z#e3_Mh(g92aJK0&9EcQvAPVXP-DsgU@^=8N7*q1V5X~U>eOGX$GV~O
zr?XnuGojln0ZMB0o!%rD_rI?TY-oseEk$RK0}F@%qml;17AZktdHEhT;LTkRw=4TW
zWL^Z@CDpRqw%PVz8zvNYEc>xM{_JFct`u5%v6+FKq%f`q!FA$;bZYC;!D#;Z%7aM3
zRT(~j!-^Khtm-2hanSp*fEACsE|x~!MY<6AmQI$qrCRR2uKn~DGGs$*WI!7iK@Vco
z*+;ni{ckO6{sR@u^R59Iob4MoZq%#k*q%8&`qNr3VRx>Nxq~S4PN0p_Ei?y2S+cRE
z=gQ;p?nOX{;{`p)r6eoX%Xq?w_{G)0K-XyZ$pys|Cpc)Mm#ocSqc=7W{3H<BSjh}K
z)blMcwSBn39dXZ%+w8p`ZnHD%9BzVDSHoI!1(u4V>#e4%t~>@>r}1K*m*u_F7R8H3
z{YSJf)TW}LVCDo7W&aJ0Jl+wowLA<+q(u~GnP%z+-Pj7`JZDW{Ig#@4lxGJLkH#yS
z$Y2f;&g4IMTxGRRW3))+bh(S=s-KO_54J5Gisly_PqKE(@d8(w^<bH!n)U0*vt={B
zk`pWMD>n$WKjt#uJ2>-@+ky#f&G?DhwXb3;evPhO=(lNT%`_Bikwt(#2WoO}%9<u#
zb|gPeLyzAFaEb9oC0t1Z#G+P@^D;`#3{iAyhmY^GY_DtMQ3mhLxYZ(L6Ee_NN}p2m
z$@0c9qgqSPNbdFG#gOqCNqKkVa?HD?jW3e5{mv}P?Z9#W`fFF^Q`C4>9rHtt;#$^2
zcLvz7?_h7i`F-jxAKn~Kg3_>qw;vuhJK6_M(Gvh8GR+H<!~Wyh3Hs02%ND~#SLgBd
z%+Q84K1{cRZD<+<AO|fT3HwNr_ATFj2Q2lB=O;3%z;H1;84RiR_dgA(Oi?~4bn1Z}
z25<KsPfWvza5cc^_(&(=4+E#u5s_jByiqZ3@cy#t^uev&5sc3X(phJeUCkh>@+c!6
z(*-e~!Bk_1;H)j?HDl$yOJQpupbRcrEO3Erq^+fezki~B8+zSJp_G$jI8Flu>ldVg
z4YyiU7ely_NKt|%aOh;<Lb#R-=7_sZCr0EPs#EH~!MFmV-%=>9jnNAu;EcdI<az?&
zZ6m6N0X&XE#k+t+m<+3U(g2iErwg<&5{pMGToYf`a{H)%0YuzUi7AJ!t$dCFC&X?3
zG+c*J+Y-K)jzyQH1y{?a1XB`kM#@3f9aK4!LOR~M3z{Fw&8K{qQV=1}Bl)R|ZFAau
z62F09>X-G4U`!6v*e^mM)mWC8N(Lg90~*zBbR?BRL2)AY1}aqW48<G0{_QmMd)wJ%
zswTFZ;;H7Bv6`ARG2Jq)#E!~98$A`GRu?YeY|zG|jsPkU9KJ^xDS!>Dw2+kd-^kJf
zLDohR`3Q3+*C#>ix(daSKIa8QC%Q!lHRYA9V1-jF6H*MZcNUP&sUn$sfrUfhf6D&g
z2K21tMObrSF5iW&0hh2CKYyJ|qRbp{vSDj91|_@NvyAsu9;621odwIiB=e2UBsc*F
z945GIOn!1ZAe0*COJZ`_+;0eUX~@*l>gLSup;Gn6-V9AfpYI(A*t$smQAVN*nT_KC
z-^7TVMu5-(%*iS}(KA8P#^|9M^hE8~`Wb*adEn^6!d}_EsSFE7KwP|a)7&vd#WIs>
zwCW_z{Pndmxi7W>RVLZmIw&C!i2_MYPiPL$I<|vHu=GgRUVSRyA2oI!R^a#Ov1IN`
zxb!nDcu)TN<$t!oP#74LHL??uu;<ODTtpY^k6>$|_7yy#7sy5qD0}vWC=%R6Kw2H(
z0|I2Iu7JN%<TP1-ljuIG`Il!NkRK+*-eJOI-JyrD4M656Xdi$6aqwh^mV0A@oH2E+
zc<2>EyZi;2pmJ=W5oqdwm;2&}n=Y4+i{5C*N*&A<O@OhcJzO|#y72nMaqHw4VbQPD
zjO1AYk~>?teK8O~rbJ`|q@<*Deexqp6p7Q1R+-+v@Hk}}vc!Z;6Ydymg>`4sS&-J>
zyLG%1rZDIm7F-<EAlKgG9+#+}3wq!kpxYT^cr++2Q(S<P21lNA^Eq|@!cPPnK!IG=
zh)eX4W4B`1)XNH?J-e)&$9)1~h>?;JJp}MU$k)t$qnP2;u?Di5sQ@uma?V!_a1n8a
znc(xbi>sdqjv42rYKCm5+)tq`B@T=|IBMvwcLv&qLUZdje(!n0JW=BgX=Su|6u~C4
z_wrO={}R%Co%q3uM{!F^>`%bBH=l8QgEVXieLH=;N;1{rN1Fv$Oe+|igdPxbk4@z8
z#0GOR`gj5}zy@lu_h|0}s(~@VpYPtC2V$t)>$8~E?{ygB@t_=M0wt4aI7Aif3Td*T
z%S?u{=?FT|1R*cv88k0wl%7IWa@pW9>~eAoAmk}!!7C%4X3EIylM?TaNbF}ExLV*I
z6MjR>4;Za2ijGc{Z9L@oK(7V}(PuX^haPmo?_F;LSZcbE$#-U3h8!lsQS{F!vCo=I
zD09eOy7%+ZqjK6O?%>!>$034RTGNS(o)VD?0QT-jzrul=8~7U@V)gjBqJ6OVCU0N_
z$Z*tTW!6+(1K}M8THw&GSlTs!nG!k3zQ0Uz`nZ%1Z2r4%zPZ=6x&&Sz@YLbswE52^
zCFu<VpT|qOlSJ$5bb<Ti?>HwZ2APU*EzE&tG*zGv9991`RwBbPk|+NyJd$k0#I(^i
zsT3_Q6s1<qMh*^}-~IfI8s8ZLfT(x#$^oKOX5!yRO(Rt)>@9KJ87K20(jpOfgeGA{
zoPb@mb!oxJ?7_ykbFhT3hssMyWhoDH28O7yZFMgWSSp{(Q<V)?i4w9eZLUTA5rFQw
zQJXI&G}ez0pr0{K)dCbztYVijsUSwax2|T#PMnBkYA!5>k3ai*WF5gbEZgw1{2Cdf
z7yzeFts+9I+JQIQO{z53>5gY_8F)vC%N}Xz&VmE^h8}qK33nzxCk=#A%TAKqxoNEH
zxLF97fDZ7wPG^?VGJF%Ar0cy^p!3d#vq*HQ2QhTO*$fLttZ@)%D_4$}O81C>&Jr?K
zJy*nWrsCundgqJ>V2<R*J&VhKnSjsQ<O|augJx)sREcXqY&(^~ReR1_?Nf!geP^r7
z)+YeSIFR(^EeJ}*dUwSM=x{>0|C@E|3S$Hp<K25imbOz?k}^3qkHp<Rd*+$db{5)Q
zFK!=Q^VWA|G0{4<PwWR|c8%|6W+c$>H+PR5N8^x<dQWX4Xc?DQdftLf+J3D&zkfh$
zg&l~}`j%z%eDnmAKQ1gte=6h6@qOcg?&9e(r!(;R#MssT3=0&PJF>ey`s=nPp&r-s
zO~NP*`)9Nj)9WoHyOLGMCg)jTy<B_lu{Q|oJ3jrPJ&styY_CIS=Sf!ij;y?UKxxaw
z6c!g3x1*bSuWp}2#(~3eX+81!gqnax5fj;e-ZDMB+SaNc7};NYeKdzMa3nc2KE0(k
zn&lJYQf;G|qXzzo<Ba2@WMO^1>&mK`YOX6eYW;zW=Z(5Y%1b&cQQHPSe}2*es%U!F
zXlq$;h{Y@T_Vi2hC^qyNElt!!^UY6O#HL@r{4<|oL-=HL?Tb2HzH2cs?yxCcIB9h?
zW>h&RA`O7hIU@(SWt3|(liXO?pCz}NmNtFggSY(IZG6TM2SoamEMf;Ef!6$z;Qu$7
zM8;IJ->m$8$}KEo4%7;6mm1J32*BRupp-RQXlU-{EADwJYX??P78<cs^Q;<&#n*d-
zNgaR?z()94r3>J^YAoQ)2gN9pvD9z4X``+OKx-7yf*hsgK?c2Hkf(}_7@aTbq-z&v
z5py_S@=B1+URiq`gbPa}f|DZB5EJcO$CMY4fJCU9@tMX9<6)swA|bn}MM)GVhq+NC
zzD&FGfdsZ&nI$|I*)S&jyP~`f0$u>NO6d%X`J-dDnkAdx(6A*{F>7olq%zh1<1X?K
z(eAXq6}wu*tGh@m9sFxR)u~E<T!hrZA!1i826BojEr=yS`+qnZKz1B+?&az9WNJ_;
z8zTdZs=v?<Qu%qlK`c(!`TLk8mFiC1*6WI>!&;-?08yLvbd4bC<<Jd}-cL<kT-Z5{
zF3(!<rs_q28A@cIdgyS-9-w!~K20vwPK0lnr8pFTcC6C2VMns1h-Ip0Jk~`F&-Dsa
zSvZ|6aBCeCPC+8*nOWpF;TOvB;yQtpD`{>MJ!9`nM<ym5%b{0LjdUphsqA*-IWC>A
z<t~4`&@$PuT8Gygi0WT-Yq^k*r(r@7>NCblf`8)HRD$Qgf$W%BLuU$=okT96S^v*K
z0I8EOhfs)(xU70xv0HAnjuglJmuME|gwLbula_VlTvx6pgxpR%2LSX7ff@Jq3i)Up
zGuC>pPOWFggZjSu9m9oN%ZAG5|2(dZ<g6&!wbd~0Rz=Vg;BC<B?7ncR8(o#mAnVq<
zA=XBY>zOJ9>}M48c>Zu1xNoFr-r3qWFOOCSt)AZNdrvPEGup@e=jqksfGkcH*1dbP
z(%^uk<jebLtJTlo__^Cwxf5AWk@3s~d^lP}!Tw*DVS!PMuUZOFB!Q{`v&YnV7}?k0
zsQsfq8V%Xl>MPa~(ZfI~Br5c{6nwf75#-p)(bcs8>BDV~*Y~Py%7)dLgk$Z8Y3@+2
zVcuC0T9KasbW|g7*<41mm$_j(&jU)RS|)d_W2#(6-t&qLj9IMC?`XJ;xuKiq0dyhE
zi!%Yg_}74rQvDH*IW%a&p|J8PAJe+rEswiG(b}}SKg05jt5{Q7A0YrqtI6rp{Mc4l
zF<p%qQ}sz$IM4O?bo!$2!H7eji`A)tuj;M=y|j*)MdpS&X1)2|_7UHQPt<T@*Q5FR
zMZ%IE>t4&D$By2<2=TV)nol6%+QB>pyLVQesAt33y?D!%*gZwkYhChe^_FBH?)!EW
zMZsMAS=Og<|IdXg8BS*`9FAT4j8l2>6;H?7z3MJGHT*r}ilyycAG@89)nzp0p_YjX
zcHNS}?l(4f65wL0{jZb(wTU1axJCDL{4&=^yu(^hXGSqq`#a0U`h`ZUY&8<2VAaH3
z9&t~Rt&A~Mu_G`kDvp(4K4(PUsihH7`HyuG{PKoxs9#2Ft<CDMU&7ZjiIY<~;)~2Y
z5P}k~!1GMWg=^(opbdx6s`&eTG!pkt!TJ}c)2sCH1Zd#3)&TH0e8E)S!lTBS?I@)|
zZuKhiA?GKlBN>3Z=KvfeiQ?<brwQK`F?5>F$M)!W3-k<(J=O(ub1!c0=mxpIUpu|l
zUrGBr`@rY)WoH;N(nP4VIr&>d4{MhkG#nyfLcUyPY)Sez)Pt%NCyoII`ZJ5v6XjW1
z&g1%kPKrMH!Z^f4UI<M^hX~Z8op6f8CKYPG4KNu!yY&7g>~<4*HxQ4$gWW_l6BRF+
zW*?XUqD`(lW!4YjaYT^e<T8Y#oQ52Yka!UsGr0^KAXm?TVw6D*<Foi2&w^V(t84)*
zOLjTJgaqj0rQ`tQH?iYhn6xeQUGWKoZ6=z>HgE}r4$7Z63Ws9OB>=ofH|F~F>nd8_
zn7zvZ&mawy4x;K%CmYo_QPlI|aOt0!osF~j=%)wQmenA>s-l_kl3J<&EAFDK^u&C7
zA<T>l?K+yrMM$n67o{^DBU%cYB>^-pD}dzX%m&aDY6GU(vV<vIEtFX%&x8R-PXZY&
z!^{}cOkpt<PsfO4q`O``y?yh@zdU(*y&gktSS`o?8aT13m?}5yf$#e;HcC&nQK_n@
ze1C12B8;ttxNT>Ri&9a|ug~FaW_dY59xij(4FRcT8kuMq@VI(c_~8A9pkU!Kp^+I0
zZYpubG+F{VW%N)%6_u}mJmE6<j5E%Ktm{!v3QpjuWTEyx+}MjY6VQOXs=I)~9lX~&
z9=y>8$_p82c@|^>Jn0II!BL!zAf7C}LR7LJh0O}egb}P4&;%yqNaDFr9!~}`2phww
z=0zJXqT2m#-8ywGZ}4jW1jY6|P;k`25d`xQX%eqYLe2^hxnvia1|ty1a$)9a2Q|b%
z#?eLkOztos$eh5>(n9xOj7!XI)Bvzd*~;zmzk&9V<y&SEn71~l-yKX-wf`FJ<*5>X
zFaEhwn?{=xmD)OyFe&rB6HnBzTo!OdOrtA7!gEkAO9hLJtkPPiqZ`=7ShmD0Vm<Kg
z8h?NzJ`G(hE<s^X4amF&csOebZ6BNF<ZH$ro4#k7O&0CzyuV}0MM3*SXn93z{IZh!
z3X5pk(sKGpFYMa(4Q48cVj(!Rok2oghw6ZbPf}739WLj21I^+vAB^xtt_bU8Mg!D*
zr!7Y-G(yO<b5g!yaSt|{^}+;)=A5Elx^69hf~E_ok?0*CYMbu`wgKkJAy)5wCktX-
zlLIg&{nAkKoFLXUh`VV(RAaaX?*_MQR8UZ$auB>>YEu)Jihv4Kb&*3Nz+V}tGTMq@
z+3f$J53T=96zqg15cvUDO2jxgH1-agsA8*m#N65Nh&}(}CoG|uH@%Bb@Bd3ddpZp~
zV@$ggyh?**1sL@pHoBAqP$)}rp?s+ANE}v1sY&(H0c-!i0M%6JX#msCMk!g+zcM%4
z@wQn6n{qw(RW%~CN3P$bOazuOrZFtz`&Zy1+pjk`1x(p-euY>!2wM4Z4p{$KVuj@h
zfA5^xvr9q3g8|v-xEJ^avSwyLaLCFlZVXI~W+ahdd*C7<8!6qgMC2(Lvf1Wisg07w
zCfX+FVkeNova&U?^m@GKOykCYY@rvKr53EZeDD$CbO92UGV7&JpU648jPqCIu&tm9
zN1#<&R)dayuhxuM78rCx*wb!v6t>er{Zg#I0cHtx`F#D+XjOW#N+`c~0w)UVnSdDC
zA-8$uk6bAW9lVHYA-FHD2jqbqG#JXlC9O@)v7KOEZ4a%s0X|5!*%jIv(6=(-NKO#f
z3nL!AR{@u@MInx_(oaViIDbqy4iCe45zpAT2k!D+OL6cFGg^wo9a}~X?_$O8+-Y5&
z2QRy_A@sa$w`Np8E=?%v_4<5UcMvGlxcSxj;3-9rkYpFQk<(z~rAn`005WFW3lr>W
zYku3nE?V_k?y$Apcn~&H8fsO^t~G1fPJ0GatT}|Sg>x;_R5%SC9l|qviNQr1_g-QF
z$p3!pB?^}$&{#UUxBIW#Zgo-a!)kMm<cNa%!JZ3dQ7?WFv+RyW-le4{qC>1vp-Rpw
zwH{O&;N$@7GNanNGXd=?|3Ou&8!_V&SZJ-7-n1<B1&09Jb=R(+kUL9MtS75jSVTDt
zsvd}z##iYAhHR7Uyx-uF#Lc%yS^j>l@lpu-mAWWM<C*nHEwJz$JSK(=vS=6q<!FCJ
z#!^Uv!@fY6O5tq?PGI{CPo9mzEf+2r$C@S~3&rE}u-z@M*zVvstq5*r-T3co@J@<L
zBuj0v<B*Ppvb=HRGy_dJT|7!IhaekCHoD_`ncl~P==5%ZNReg!5-f8M#wYIvUV-|$
zu(JpVHYGg{3q+7W)3Pq~iQBs>r22OO(RGhIp_W{4@u=ai(H+46`~b4I+?V`i@ZV9q
zn`t;cYR3Qk#tR3B<&6CN6@#GrV#a5hR}X{v72NQ(XATmGg#MKuVNL_kDDK(e7;Wnr
z)3UgS85>@bnw#W|s16X~3`2(dSf^#!UTJ9~3?vknI|A%e_1H%UVP+3EYHbIs;Nc;}
zy`pm2rN8}o`~`$OEFwMNmWG%A%*qQDahqKbUoP$^&aicp9E2VS+#SkAMxl-Msy3H;
z_FzO(Og1<VcBR7e?yzs$+Y%HHxIp?=ZVvKc=2uIYdAE_)v9ls(6;@K*_}p|l7}D{u
zEuB%9@#ID}Lqjy+k2(z6zq}pW%y=mp*LURyxfx^Sgy4?THH(2Vxkgw~nOtUt?um`(
zv7tg+A9#DADcUsBAFW((pzgeMCNjMGQ|QtZNtWqNjLDt+TdDa{;{iPZ)c%ajS?DsB
zF#IwavROSw8(|IWM!<8&GA5si*TMJf5HR~Q5%+$U`!5&Z%o*lly!xHj=Y3Vo3sFK6
ze+yxON|?D*UcLTFlkw^*ug~9HLg#r0SKKzNvU5|97=-CT{Y+TBZXfd}Bjp>5cT*=^
z+1H8aR0qZY0=L^LwiWHrEK!r=4SkpK(M;&i#I1jDA=SIyaDL9K!6j!N8N8idh%$x0
zjJA{HMQThffC8kyWs3+iDKtX|3W>v@I<^N=zI*IP)V4=SY0mAj6i3u6DD~ffNQL%W
z$Qd>+!F_;*`e)xK_eP~!6!y)Rb#f04xi#aYHpnsxi(P(-S@F&>1%#YpMy4JhvEsT5
z45K;4NMPGUaBZ9gv?{gZ!svI3B53Z~k7{&>-Nit6x<T$8mbc#-gLMfPkK=&1*W0WK
z_&o`0Hij$ilZ?}J3B>u1$331=rjl{|h`==ib<tpJo;Muby?N4_pR?K{)s*>zMBf*B
z=m*L#Q{Mhdmn$(zMQLGBp#VlYINyVlALf#Pf*FGuRV2>bz%6frq^6bTX3$$AehqRE
z4F<dQ2U9>aI@A{d-M|%-7MS)P!pU$wIJj5^Tr{cr71%}&;hDOfn^WPR_js*_ZpMKP
za&VOtK({%4Y5@2a=zaIfG51o^_LaP{Oz`NE@8qu{R-9edk6+ZC!ND=D5<`b9p|Jo5
z3%wEA$X0deXx6hGa5i;01+4Z~LM0gU#O56xN3S^A*cGU*)2p<I%iC&_L*1>Ga3V6T
z=O%h{pnuVM#<YR!DX~{+81vp9-u!u>RO<TIzhAqS?s6nIkGy*UU>KXWgz}$yP!8?m
z=r>!3kSkcaFJiZ9ho7<v`!q40_AIk6egXS)Qt81^WBj5WI@FX9^5bN0W*G=3Wtf!Y
zfo*gf4^1_cpjX7G{oq_S$xa23DHIr1E!MJc12Q%+Q^fVk(72*54j6kQ7$czh6W}sd
zo7;(WGhC7JklX#)9xoAJXJu|@*k~8Vw7oJckDOkPOUR8q%a!9HfP?PC+nTN_QBXty
zQt_E-jiUn86LHa|01gIg<y+1tS7-I)FE9aqF&9yl?FPV6w!xyq>-|%$@`3Xa0<|9;
z@vT?Z2hS8C#R=RY)of#oj79<k-7V{8HNM3uwAHa!k+-RLM3lpPuQ{`Iw^gZtpv7MI
z&f?^N?Z^%{ZQ3M`O21_(!6afB>xss&C@#4_L=H~v(=7_on%GSULdgB~fxlQ910r@t
z3vIYu%zS%P>SBl}2jb$o9fzNbzJ3MX9@xM#05COBm{4O4V&UP#7azy}8m0VJ!u`!8
z+8gH)49MA)Y`FpL4B_qRU5VrwIb6FvH&$RQxps{c4UJF+U+%swgXmSmA2BqWdj3Mj
zzGOw;J}|tdkx6xBCLBWYj24nAP#<-z2SE#aOk>J#1Egbm-0$KxM5OQ+X!lH4a7Gbo
zRO*seSsI<SN-fiht3|0w;{L;b(W;wiB`vQ*%Re0OxKhCG&b?c+M=7z~NqkSbPHGS}
z$uPRrn8)9Qo0>+2OH)5lF7A=#5Uf$5)+!ELPRU!HZyP5Tz&GIl5CLEAv3t8%YIssD
zYpk`sxJ)ZN+iZ*hNVP2^tVr7(cHb*^?iRX{<<|t}k4S1R0NxmLyMyrbv|x%A1I8Wl
z5+r=LxF+_iqhdF9<wHtcX;{6Zu{GxMUPVRAL}^!k#)dYjg*Zz|zccR;uO(P3h>)>v
zUzdy^An3rMd&rRn#t<h}8L7+;e)}M+^;IjXKHLYlPY-1G&7b{!je2r8>)dAUO;6Y6
z#_~{43-{BCoW!A{2!qP?0r`0EG^U5GkPM7q1bffDY^^?Cq(vdpYxM+KL(FMCM?uz|
z9hKn}Br_NyAmZD6Nf31jL+nrH+~}WHA7T$#6ju4lz%Z@OW)WMNh^Y~LI@M#coKHBj
zyQg-fTm!on+m5y1u0dX&<A-jvht}9QIXSIw3h5K@xxc5vq7cfNzLFwOqE7jw5#QC2
z^Q!ety>eG>^}3e1(txYcwJ3-#R;fK}xg>4noa!yGr$%iqmPqB;Rc>?mc_)x!lq=RD
zN(E2{8-%<1@nu<$$@2(Op9#qIJGZG-fC_E6j!B$nRr~%RP<}e6L(cv#DMW|JF({J;
zf#!GztfXgoWBB0MgJ*t&4CPMDZRsHkK8c5pL4x7g35k@YknE~{(EeM{rYS_@rTERI
zTAAF~zVdkSc}IJchz|XUCQx211%eg4{1kPxEflaMzq%$2>NO&REr#j%(1jfiR<mA=
z>@)K{YooU=2-IYpT_YA{+%ZqYH#w~vcg$9=S^K%+uYCYnf0_z6W>9K}yYqqgWB_&~
zet_Hd<HG0=;Vn@e;t8}t!sXq`vivK|Li!3E;<#v|>#MbuvnB3%Eyy~b^OMoP-(~$)
zIY^or+7vpf43y_SF_7}TvDm@)_P9Rj;-XRJy#Da;$qRFiuO}uZE<BVG)AgJxjy2?Z
zdhrPWG<EJ}0>1IW2SS`Ngm4$zJkITbJcM${FdPKMm~`icXNH;$vT^Ka*mewigWsh%
z$;v+$TnUIKt)$5+(jW_bbNX@+W-@`GcaCa;VUDjOBm@PC00J~FN!35rs0<+(RU<I+
z)hX|8pRS$p-_<P@Y;tX0jy)&kT;|@GQ)LzG8OqII>5OKNN}YZBL!7^$Nk}KLRqfVu
zmn!9465LZ`Q>%bKc8FV4uO%qS?G(``8%J>V=~;t*cRXtNh}xBas{gU(+*2$YjxD~d
zsC{wvCqhIk{X0yF!w)`hdx2mUOnA|xwMfVFt@<tD??B0HAO{9=Tg=dk^WS`fCHyYf
zMF?5R0+g)_0bvri+N}lWr7i}mY5Nj$i1bMVZLNDjS(rw!umyI^cipgjV+OD0#F&A4
z-L0h%>S42%BLSo_FKv=o*u&b45naSr58jEzKawUQ?ysH`)B8v4;!%CKy@9H9WkTPj
zp@ncI<u#X#(s2Ai<BwA4VZ@IXysbj?lxaK!t%pb#V93w$=svT`yYIiS-jh*`R(=Lj
zAH5AReHeJb73m-p6N6(bWM?S<8vff9&|u~IGhVXZT)e(3;8yb;KKv?6G&FDSmy<?(
zC;$A#x$o(j;;`zj{8ADFqZqBeuN73u&*apxPdgev4%TJ<ie)Ro8XOM@eP*S9@!YJ0
z<1?$6o(<A0)g7k#fXe=M+jm!vs$Uu?YkBB}S{H<mzL1f;j<vr5Z%TV6BZfKhQjiw}
z=Yqk79Ki%@n$fjANNlVIo+AxPO;4CjR?G~IU9_{h-j;W+DH{q+N9l>qA`w9&36p@P
zXm7Ry7t3lM@2N1xc7KireGX22rcPiA=tPEkzug>{kM4l|EzUq=Bwi!~5stzx!?q9K
zr=vTFw~>YgEtKNQJD^!Q_Ujt1=V%bIvUToECvdioc8-l1jDg?F0R2V6Yt}Af3|smI
zq?P;Je9{8jz}DNbT-_pIHie@Ysb$u3V?W=VQRcQEe~fy=*h5_pS36`5sx=#1cJBEA
ztVrT&jO>l*At7^S!;fMmEe|4M%$d^#sFPb5s+3Ry3QH{wpt%eHjk$(`yjmC_AwVGN
zt(@Z3|LEj^E@+f%{zv~3qW?^@s&EQ_6A;Z;;jgXdj29!rOWqU$QAz_OiLP}!GJbOO
z9_udA@m*OeW>i#GRth>blfPvaaVmycWyMH14s@Z*k_5KPUF5|4IM+a^%s`LGVBJkF
z8r;q~HKAvS8dd#)J&5sSwr*_66A>w(zHjhdr#B1|Fl5ss(2qix%4rzdS7a7Za*IEx
zXeCT5mI~+B*aqn>?aqF@6!6a|`Kn7~X}o3o_L0e>qeR&AbbR?P%4HifIdJgV{pMA%
zJ(Gh?IcOGO8dIrM=0+>6el)<BKtzi)Oo6l$AqKXnSckNm00S5y!^$nEz6Zf;7z~n<
zbwIh~5H*9tvou0}$b()x=I82dd?$mx16)PH#G!q}>~E9;m^l~%CMvlY>Z8ze5?(Es
zs;yye9>WJ~W1ORFUdqN!9{EwrTh0{aCO<nx%ZFfJ3`6Mq`cyW)`Fe1p5?+52mSx!@
zJv<zWQf-!Nw@a!4DidM#VsaeuqGiv9L|z`A=x2Ur=mF~vCbFZoJ6yXju@%u-unS2_
z7LelP!%ZMB8MsDAps)d_=AfWLnwOlO)1bU9n_t_~lL#sz9WN+U*?jhp$E#Kc7Jrk6
zW+zbHwjz%&0CuhHu5b|!C9Pof<AZiLj<C^?vC&y@y68cs3Jj+3=!K0ZC}zt)f34Dx
zC~ihfSDGSvsg^mnDgR+TR-x<sq-Bo$yMqAe@VGQ3znV~(@7r8F<-Dck7JeLtC!9Yz
zexaCa1h$2o=zFOYYStcdn+1&n^U>P398W~E<O)MK+)_HGc4aXH6{dEzLvJWrxx1`H
z6&qZB6ww(Ncz4SS%E0!WB`g{Xx{LFXnIQl~OdXL=58Ox{hQ@C5$E~V8j*x3K_eKt!
zw!q@ymhHscT1&y4PUf2-AD*FQ+qW0>X^bIBal;JeR6@645Q~6fOj+G4&DKnmH$`Y<
z&=5^Z0w{F_pT<DuC$$Tt$W22#!w{xRO(Rpgl*8r}F-YI-!no%ZpTsr`=nW<Kq)`Vd
zi}v5Y-E60)`)PKivux^P^t0s;D-*Rp-~9U8&Ks}7{9Y+n>6z{?o_#4d?&Es&#db#N
zbM^BX)`D3=M%T0D8Etn;d)mrugwvfi8#8Ra{pse;-Fc=D6+Vdus}0<W5erk^TU(wG
zT;m}tdamS|>t3<J@<EgKMQPQ@!)<%tUK<O+P~d+3I|ZpVww@yDV19M2e()cY#|!L@
zpR<ieDY(jp_&ot1krFrhdU<(`$g`N!>S=jNH64>^@Ufc}r=vF;hPHDZGG<{?3&9A>
zR34O!t8Q0shg?nBM1yZJ7(s~DXAgy_9ODrbY<%sVlALTHx1<!TtwQtAD1B9b%y5N1
z#Q}!}X&GrR#zru1q*hRKZGA8SIk+@HK5ds3A4KsYw75i3oIX^NTm?7px4Io$vv6{h
z)+1-z4Fa0`rMkw(ttg@@lMLHkilk%Vi1_kN`Osa6X0uD!gf|{bCt?|7z;&Vm-^52o
zGQAu?1CLjoef=#p(QmY)%6>U^;A|D<<}E&87~w_fNTj{2n5RZVZEdbscE&fj%Y)_P
zxM!Yy`)WQJjYtc;`Q&wy(?TM|HvU`WEW&(ze4^k|Ont)Md5T1~TRXKmH@5a{C<H${
z#<m=IW0JY|&8xw?E)BwWmZXL;SusEH>C<KE#@$NSXZ-nN#x>o`<=dbVf=RXf*a@}F
zml$M(#BJ_C9&V$I^L&O)=H5HpIT0>yZsL0IXSkBgWU3(F6qr{^=gRtt``A_MlX4{s
zM*H3v&J`cK8iW7Y?XoKMi>|#$G?O>Cbd6A-S;8x+pWhcN+!?yI?|QbEK+^s!3VR{d
z<W0M_x%_ee{^2307R<Y9Xarq`rm03$v3-~1jS*F#SC*iImX7GB8a!zmQ`+4&!F9?5
zHh{ITJnDnmXBhPi%{$PQ`a4n@Ly%9A$2I0~gMEi~Y{2c*nZuiNN*3g;hYZw{J&P5A
z^bX5E&bnhE0Vz6V4UMAMQw9d_TIGC=MIy2c)cTC>^Z{dv=P$d{VH4c&?3vqPy%8KV
zqd24ShIYxzCoFy9Fz0C1<@ski`sBCNVmx1gg3?%go7-@!vU~EA<~<&mPJ4cGJ|-q+
z5_B*&BmFAps*?1{cMH&TDgi6BF-O?s{>j&x+|4=*wMUURZxa??EH=W3o?yZz^Hx+E
z{lF_k3&wE@9^gnTa4=e2STG7G`Uq^a4sE_2+4_vz?XOQm4;Kp;Wn~>&WhUBsC=py0
zZwnnb5Keqs_b${yU84s<qkH7s-eibbflUZ3C8M~ea5SpwE*N4+G?F$IieNjRxf|E7
zyKGnB^7(kiKlnDkjC0@B%ZOO9_2^3R5=WtQlolE!CU+sjd|~1s8q<f$ISN$}n@zcd
z%0!v1XiPF1XF^??>W5Gj+}Nj>lA2nNMpQii2vF`&A{=}wMM*sgSl8HnDeqd)35y7{
zV>&LRZrIrVgr6GfXDuz+jpo1}zIQ3B{jiB!D<f&*r=huyguJA$^#8cjcvjR<Vl;|V
zq%bc$s15CmWE>-YMJ*%c=Yn5EpQ@U?*f%yNh<Z<N1nw=5-ne>5-zYH#j#>qJ-$g^q
zD%&Qxxm&Gh$anWgtL@ELQMJZM+aZPH#hD1-wWyt3i(Qf8?}mRy@jIb7OcL}B3~=!8
z6A(~E7m^p8z!1EkE)#dmtCW5)a%FqmeHCe7Jz!Ufs2A28Z6O&suDc^ebUZyvO&(nC
zI5`igVc)gwe?-?xnlsa{L|=<mzOk=LjTcGB9u>YuOAMgzZ~s#Ty}fV{c8AvYghfa1
zS6KdNQL>tcA47qzB!+RLov1Nq&z?POqH*TTZCO}2ZJRm~%w<1yjiZfE&&gFLRB5KH
zW>95=?E7D6p6yeoX>KQaD7*>^y`M@!KkzG(0nq{*Sh`$6zjt3C2Jf$uG23v_Ponkp
zXGN7MQH)}o$b^<~u`<}krrMEN(Ky-V^1z&C^+V0KG}i>XmgO0k?`lGdVXu$1|MuIL
zAT5=kxUEACoAT$MN8DMMmmIFPAz-tz<|JdtyZV~-Nmj_e8qK_yz;#_0_-GbwMq}|R
zXk?hUvzqsOpiECxOl&Mt0m*9Ip~X7>rM^1=FW7ss@02*7eLA_Hb<eioi!pR4-8^0L
zkfs<Gi<A^l3++u|&)|8KxC0Nr|Ni?d4xWt771dX!r(Lix)&7`R0rc+Hu{jjA^`B*S
z2tt$KM=86x6?vFxX@x+gvUOno4$j!-l8fh}TAK`uf&@fFv<RhZ{Qbe3H|DtS5F~sm
zAm^sw%eEyQ9M)2)MV>FQPw|9a=6d3)u&QftE}cSdcZ+`xOUN0m@sXgPF+eXk4}A@j
zNHqhaAuTP{8|XxE&To8KUS8e@`Rx4YIXs1Wq)?pv=fEA-<5XvA#M?}j4^B)7WAaGy
zp=79%iHV6F<CK@l)(p57)uUz`&5nwQu>A0QHgZhX>Q03GtGRQ3qvbdWGE%2$&at7l
zKc-24Fhb=*tSDC7y{mD~7cZ`RqP1oNS3!s2qe!IPY{LhjW_w~D8@MqC({gt_97cL7
z%8q|Ignn<|^GDw>t_24NSApU$)FLxjrj0{eEk^VSuVu$|Yi%8!eyH{OGim~Gk#(5x
z&s^kOw{~qfw9Oy!&Ro<pH@^dR<*|zx-RMe2w+19MjvMPucmaU#L%T4PrpN!Pti(V7
z-?k6w3E|;eSeF_@db5fklB@~X_}rzunWf<C*YgbKXjZPwx1y;S`tpr0LVr!KYipO@
zSRb8s|Ng6!YyPI57gX3SAO5wXV7|BOYD_vfH0FB+BQ}aL{)7HUvs}(4FRyx(Xz~2T
z_wO6fu}Rp?xHJM<IL^tEPq)+#ypzVV&R079^x3mW0Ho|1rw@pLu7)=+^3VYdZ6be2
z7N<GnUeD+XN$--A*U$KFEW~mp@kH02RIoN%mrcAATf4f9NrK7;;xb>vU29>)n|YnJ
ztg_kt=DoYBCQ$ve=)jfMD+Y4#Le4+$s)Vz%^I1&NNN)a}`WolozqsBEo)$aU`vZAU
zIagLvx(Q~qzS!#e`uZ@y+_kU++5-}TD8z2&0c?+riy<Z*(rfXG)o~4rixU7B`?<+L
z`qv7NOl22iK3g=d(hUVj`V<*XKTgXQ`<tXE0|POEf%ad2{dE$U^>Cy|i~6CM#4Xwe
z2KxaL+`DgW`VuSB5{=09ZDrG&ahcA2s255#qnz|kobbbr(FM!P;(TijHvf&(SFIWP
zO+&|iousZ+a!Sgzr-fRpLBik%df#s9i&l{+#z|A>id{$alvPwhF(s(w9*-trE$F4E
zrm*U~u+2@y+bhP3x_Cf5!J^r_dpC}na2!-76eZ>5s~xrz8JhsJuIG|;wrteKumc|q
z)BNh(j-g)Sg?Lzgh&^7UGbmDJmlIQ<d-fEMNf;@@rILDr*GSX2G%4}{8oGHXB&?B-
zVT!UAt?xdR&ZYe^YI|d#vRyL5FhdhP%Du-!k$BpLnRo9hArK0Jt{4e${Myy4KLDn#
zZD<G&2v}QNTkCW*kTV}Tz^v8Dxa{-KggrbXBe9QkH*MZLHhxGzpngCnY1DP~Jxj1}
zkf?2A_2a^cYh{SRyVs(s?!V_-BG-wXvX>bDsSC7RJ``Ue0NJiTEsTB*s3QWu=kQeN
z!VFFy6q56lI>$j(YTVZ!L3-2;-&SG5FA~Bm%>(m9u_&7VT}}`z|2%LeH}b8#y9X9h
z1e!G?dlk2C+g1na6A!kfMALQ<Y#r==+_-TUSi1wO+c7Ho6sbfqn%S`<gMEEZDa``s
z)^DU66$_=uC7X{Q54^bu4?iUU_;7aUdcJ^rE_f%)wsk9G*Gyij%eh4Sv|4I9Pl;Uu
z4~Oi=>f3fXp=(WTbyl?0o;^p2Uv<L+dGiB|UQP!4emZ(A+k3TR*_h-!s^hHCnQ0SD
z7vRq2D3OD{({me-j*)iA_G_lf@O6rTP24!Oi2R8Etgj_0c~3r*nRr)ia02v~dcdUp
zBuQoR09*#^uA(fUOR5Nm&|~>oY?*Hib!+BLwvLT}n8J|uvr^ZO1Db+J2tUk+o0;m{
z)4)o=*=-2g%pP`jc0cKnk}FAHefNsv_F*iuLvAfeKek{XsB!pPzay@%hiy#}yq{p3
ze){@197xLG3*OmoLE=Jp#F!f`6IWHh|L8l9H-aKzPHlo#FjdJroQja34oCMBM=fQp
zM=NuS=sa1(8R9yz^_AzFY7}eYi8hfpZ=RmnA}>dy&n6X~jEfymH5UNJD6Xqv$nEp-
zcB953u9$>CDWzT;B?ThcQK!ARe~zF$uLETx1t`&g<S$_UKSguL67aNJ*HsKN?<R8Z
z{a|aPaEP!W2tZxkMWs*pvmF#SrTNT#oCZiB>^#uG4u3PR%!Jj<eh{^`&DF5BX24}h
zmnOzFxIO{VumpAI4J6yj8sQXi=KQK3<gy=VE}-D92Sv1@x0i|_{OUWkvJiE90kdCq
z*#EmAmi=7qgrL_FW-<=6f9Mj$>x>bx@N4^o(n{6J%Ic5A-p!jf^#R3nzw}645Gy+`
z)3i+Q;=Ko`tm^=H^rxjJgknSkmNubCpoir{RQftkZ)BYL`&nP#StMM`N%Jn{*syf~
z8qXqCqu2^cK*(Qz{Y)Dd_iwYeN?*d#H!l2M#8soo=>vj@l;QZuxW#U$6%0%$V2_tz
z3fW$t?e6fF&y^|%TmT$+ce|yIj@r5FGybz?P29-l9XssYmv$ZXi*ASIQrPPysE9Yb
zd4;B<<9TjOVWtE3QaPpz8WD*gJShb|OcxFuT8QX{`+b2*NsbV&<%Y$N8;K(g%L2<>
zGvtn*5Qh&hE-n_0QA@{$^6Sl6fnfi(eoU$?4dd0i_3+i~kakhJb`=vDCqb&v$3E00
zK*}fwCmjhVnnFNrn=2YF;SL>JmmVfSxa^51p8ZL%D|9dmw{3(|!N){BM)Wv1V*DF(
zKXPBaN^&_?^Vk_;Ih(g{xBcL&q^g>4Foy$R7_Jq$`!S=hKh{&xHIU0pNip;e!qi_n
zCU&xgXiyD{@c}BZC>0G2J%HU>yYJwnNHX^6$P*J4wR@Q=i-0K6qywduyRSuONBf=6
zz@X?KkARLOuLn2KKrR_W)xy9DxfVhouxQWEMDs5V#3?rZ`kra5%C2sG06`MPH-dtS
zN?2HBE^{%rZe^A<0zK8myu4~qeb1u7$&-JUnW7%D=ZX{S@maKw;*7d_7r^QE&Iw|1
z8CN4l)Vr#|L?lWzSh?!e*gMqY)V{EGcnE@$GiR7U3X|C>*VEdn>#ncE=PGDLkv4PK
zy>i{5yST&*JZ*!_Qzj;vB>9F}``c=J!=6Z@$Qni>>^BR^HMbk~5aUZCX+um03BFuJ
z?%E2*YWlHbq-^nrzkghD$w+pY-S}IOTkgH8LFULnPRc5UC{Lw%S?Guz?z9akPS%?=
zymnK<_RQmel6C~%z<Fd}*ocnEftP&y_n)L09~$r5RN}@dRM7~=sQpE5q7dc~-F&NU
z>0c|0zAKv7t`0Y++#p_{u8PWaOtaiGn!wSg=aiL`lf&rNzB9V0Gwa<R77}tMmqpNd
zNe|}G4#-oRok5c@rI$Y3dE$v-i<2O)CJ@_iQ*KZj!5l|tgy(lak}93(H*fOI@CHcN
z4KwkAMeLxdv);ywTZcZ81hA!t-ABc0i^&R#LmK`YM$kWW%6-QCUS7Mfv6HC>NOh;C
zOc0-t*QTzGvKUz)m7Q1JZs275W$VxS$YAJ1gGTIl+pE#0F<kQ5FDNBt=-x%(N9MVq
zDeI1=n+cyY(i+P??@^TT<3*^*C5yB6^FGupWP)oVWnu>Qjg{f7xr@7!ib}KGEE4_F
zGEWU-lr*=SMzFZLO&J+eU3dm|*{5E#Yh`)swiaejFKyqb1?2K1%E~;0K^%}aKoLn)
z>6$@}T2W6<celxnUN#yg8yXr~-6L4@gIHU4blT`#OOM2T*{d<PZp1j0^lxL?+p~fT
zT*n^vM;f~nXQYXQWqBn%Qv*#_q@efKtsSIm!gAKu*1jdbSHCxZ^Tv$c8P9ap18kc&
zTTERF3k~fyJ)RH5bw5Di8w;fKio^0Q6flgkd=a~6@K7_5F&Nodh{o)=?v&x2sm(s7
z#F*zw+@sPdj@r2uhkYxc;`EO7kx@~pG^~n#rh`^08V<verVA?v(a^7S<-if_Q3jO#
zUA8-_XM@&$lr8P$Yd#`Di?&$h=;(N(i9Ea|>a#N!lM$hdO4~2PjwT)Wq-}KrvO8_k
zB#D*v$RiD@-3uZH9V1Z{vssg{U(yDeA3InN!p(M}nwK-yC<-KtSr{RmoSl_)Y0R7J
zdQB|v+QJtIJp(gN_&KFron_pHlQOEd)Xp1_1RUX1>%{KCjS@DajD?7Ahiz}dax1o|
zSzoqz8Gd-;#T5wUraqVZEi=ZRnHej{Zm#{T47*C5N5}A@WDFYg8l>{Gtqc2E7g<CG
zcpK;UigqIhPe-xg5D(Y1&5dyp^j0wrt&P&B$Y2&q#OYw_u=_lOpIVmm-)(6+impS=
z$8qPH6n+PWuzme{>zk)JGLIfT$^hAcf!dn7G7M-OX1=@z1uNuYZGvTYR@z~k!H+Sr
zbTcR+kP2)%{zWg)1umBv4F9k9)as&Y(M4|}?^QMsVSbSLK)o@PICAyX_z#RuylPz_
zc@R6Aoz+7^NB<7LT-bx{Z)~z<C^M-mD50Q=L&@)frb>ixj0#`M=dC-a;Z>`v!Z$-(
zZ1mV^N1Qn}C{5hvbYkf53(;k7gD*sCaK_n%L4_4-vD=lKL(^d8VD})ZU;XX{ebERr
z|8@kXh>=SsS<<`|FWKFotr9ykJvAkEN(YC+IgKxcWk~Zg9<x)omx3^}ntlr?E}`DA
zfjswteHQXXgaApUDOLF>-oX7EntpW!6X4KoIDX3+_Ckhs`bA-nXl4{{?dN?=QD09L
znh3D(0yly5HOD`|+sN8Y1*}(6?n*~tu}ztY`yHlHwkM^tp=p#-W=sP3$_6dZ@&i{x
ztVI2?f||eHl|s-|t_KC_(KiC>O~(_iLr!;l%Fxo~OW$|5{WFJ2?x+nIDdN8{aZ~!=
zzO^uW32B)^B(^tZ0^Qd3U_m}v5ySPk!jv0PQ(Y_)S60zzP;?BqjVBtD5)Y$5l%_L1
z9cdvo)l1v|NWZC;60S`+eP%DxK*#$5lP+t4<mtj6Hl8(A9MZcU^!uX2r=-MKK&FPl
z4w0PG>8RC3!)O@U_<?q(YbGOy=kVKCnWxk>nRQQ}K9?1aRom+dE9dW|zZ?uvJ1!a#
zLh)-HsbK6^*Wr7sM|_t;phdDr0<Yc$68JfI-=51DU3H+Rr-vV8OFDCQTano4+KOPK
zt1&hX4Z5@_Hi|{W2UMv=#H%rMKmy{`LAj?{9DvV=9Et9ZP-DcKKOPRJn?mU;y6-?l
zJm~;{43UXRcwNv}tnz-vnbVleCdiP|OUraiPWKFM+Gu3YVP<BQjv$^6@^v}@7k&&1
z@t@J`GwMpna${cCVA6Q`z`1+C%F+Qg(wFItT?U0Y@2Y0tUL8iOw{}Xn5$dWZ?kQ*S
z71>bnU?4@*10j(?iKnc;-XCZAQ^GyLFPCWgDd@6%eHdLQ3G=%&icWUDhIZhK%MSxJ
zunUKgEfkal*J&`dhdaG84_!!n0VA<a9DK^L-}>^(?ut<n*87MqI)#?wqFZ<1KS~vk
zAiTV+!{@dZhQtR1Y{JO*{<tK&NC}Vu$}BxSJV@|#e@ggKJW5m5lP^khP2HMRzsak?
zI?PJLZM!KBWCX0Z8X3i6TtifM5aqjZ|G<Ee#49#+W?fTLu}9_ovq&rQ7tiMBHMg|b
zyGcKfBSmTGK4ls^SnkyU9DWyVy+%)muA;`1?EUIYej%YEju=B*(NMZyk}gFEa{jnN
zx$z9T0vw%+miw@G?T<Z&Kh!C||I^-;Mm2eE-8gX_QEU-KPz#}G6uugb2nwx;K?adV
z5}AYw1WRxXauIQ$tyQmn3<(4T6hWDhDah>9D;5cZ7K~g3)GC9DgbFGmpm!g%YxVoq
z{ds@fbyxk-YJiaUeV^w%=j^@DKGyY%<1SsktQ>`hJ8kz_vSm5%*T4x^anN36o7NDb
zXa!E|tnwD8K8vyEKSH%#Vq08ng1k7JIB$|F{W?OQI_!Qn2gWJ!x7@<_UUan0#1_80
zb5OWyN^}o1AC+?;>O_Q2!wMm?qE*osPQ}FdJ~0ObL<(3Qj_q0fJqDv!glxLm=rGpw
z(Qfw`$(!@fozN!3IsEQGLWh7`^$=6#0)b!7$qIm;8FRP_Pi(}ltXC*_7&vsN=MH~A
z^)+-hI`5#;o4;MS3)~Zryxdd+CGx0Dg++6b4>eN%%2qTkbzYQ;Nz;ofdJ}$mppHHt
z0bqfqu=JN{@C3_1g{yS@`(fNM8O`I$;r_aUjZdFV6=1;LUb$#LTPBcaAJ0H#RPepX
z(}KB4VlUh~Uo$Xgh-%!#d$}AM+{Q9Aw9auZ_<Qm|H?r(bLsoPA%iJOEhm-*y>3Cf3
zhVTzN?OfUod)4|h3?jZbIPV>PX@?nYJ$|}#emMo*9CI3FsQ$on9<HG6)S8X_Fg`0=
zH#6)0<A5VMLr?cdRr~Ou8Y>R*ym&4uGrIisoT*c%UUb`rBRpVk*s^LT&ET_6i&7zc
z{C1&V#mjb7c9;)-sO|L~?0^RUxp1QU4nc956E`?26RVE%$#aDR7lJIe5(gx9^}qJ6
zC*~GW;CW1MIPEf98=GGa@=|iD(Gf*<2}frp`Qygij|vz!ZX6$kPfbMLrS)mwpT9JI
zw8U~c0iUg#XvuDeq9KwRCYJQScp*WGSB*nN2!Du2Jv}>0Taly_2rGosDGyQkq!3MU
z(#lG!W&~hNSE|e04;8)@c}Y?U%Y?^TCVTs&>C<D$<`E4g3xT&p;rM9_gphS1)d2@U
z<;sDH;+ukk59%n<2}0N}A<d`@ghj2Wd<Y{zw~Yi6VCRuuUAQYP#u+2Sc~FD#hn3(a
zmVjBfLUVv2{Fq1B#abb{3~B&EErhrz3ha$7sxMAcKln`WUWACD#BhN{@XKe)ny21k
z1C*U};QiexC#4kc**EMsrm1POumfFU?qvQv_~89JLrCQ!ag$EO#=bVX3{0vF22Da3
z_VuCdm*^aK!Su~jQ0`)7+hfR*5BfVBN_Kg+^>R@nff)Hp8Wq)nI=YXx{o7jbG~LE?
zSME+~49sIi^Vo1OYU@DHVP@sM(t2P|?SNB2mO;9F?3VxS7!VW?yT{U-gL$=>q^^F~
zJr4=B82#=FlnsVBt74EGwzBHKI0wFm0N_D@jVU<v1WmbcNsB`*KwQMZ`_x==!MVT^
z57Q75K?m7qdNZ}PS#-0I_t?RFDkla`L<&MbQ8PrrisE`lOyy+aL(nPaY6Q>r93)xR
z$dglWW~(=~pk)i*D2s6>ND-@DE?4D#L2#~vLkvFKW^(tgn2r_y6oe+0$_6BQOscK7
zcxYLZ+6Dqo{j-POW>(&~;pbuf`0?Wo3^bQf0cN7crXxujAJ0O%C__f<Y}9+ND0_)>
z{#S^*OP~?{YsPc<ra|Ln-5zFgKN5TY>*3+y<xOaDiNx|GvP1_(H<{Mm87R@CmC^ug
zEqn0Gk(W{6cbsf>RTx_S6j91fLC4o6z-Lf4X%Iewxt^P0VVhI76Hk#Qy!WAVO;O-o
z<%6tQv+rP_48Owkz?=LF?qxiBG6@gDfN0Etj`RTB7{$<P2gAJj228V57EnPan_k87
z*?=-5z1xHTWJE1)<sL7Crt!OTPL$uBHgB~vp#H!-O=My+<mWYLib{bvNxi88@NMUV
zyYLB9-?eKzP_0V)VvS`;2?UMaY5#NOjU`~ytTF_JonnB?d~ulj56<~s2#z)AUSx7O
ze2f}Pff-+U*9m+)S_|9nUvjzf{#{62hy~Xv);aauJxNxhxrI3nC@E>uQDA=tM4=$K
zEdshlT(JUVf)K?ZUynN*!Hy`}pTl(*16LK)%w7KN0z|nO5K#)$k2OVkC(1Xn#763^
zu+^ZoMt8Vvs1dMOG(wdE6uds@9Ux(YR;Y?jo6Xw{XQRul2m0#-{e^{vF!SNac*o$|
z8;nWt?g9_MeFnG+0LgB1EDn?|$Ic9b`+0=};0}E<rNaa@k8|hGZ^16?hQA>k4%R{8
zZH(+Z3ftl7>+5h}kkOG2a9{}##y%VfEeGb?{zDX@18+&%g)hD0v53q9FOSR8j^f#t
zRm`OL7@CD>rjiiUtJ>e(+{{C`wat)RLH1?0X3clU-Tj$QQ0msbLeH2!iVZlYL!|H&
z)K2mUY%sadUTr(1tEFX#)Kuv{0Pi0e6sQhXR*1ArC`{zeDBKqqI-HKg0=Izx3|mZF
zw6e(1#DrWq_3$g9di3@~$0+~$OCaa9)AG6Z^BSUgw75)Rs1LT(cF&%IU>B6uJM~M?
z0~lh$U@?|BQY<1;x6wNsi%MrbKD-4+sp;(iOh7O#z~+@9lx1^c0IQqs`z>j(=za)M
zVK{kbN&Leupk=bqIT1`pRm6OHKRet{1DLU-PP0d$B&)5puU06e2-Y^BtME`-lYOo>
zJlLoq!SCdMd!CGynQS%>WSCK>2~2R=Z1&z&x1&p;_5_T%HoyhfOa6I|=BNS5wzs^-
z+uJ)8ndVUiYcY;#hqD$&n1f0m7Mi_CbU}ipIlid_4TL{Es&GGt5I9pyE4^Y%^Q}F8
z@X~CspPiqYpYMogZh)T?!`th{z(6eh9L9F+92;i@)F}q@T`NV0P}d3}tvGtMkxO>2
z2ukhpK8!Lj!s2V(k{BCnNDHfz9aYW5(X}{e+j%6lw!liu6brKtzkjzPmP7^E1O6OJ
zkK_;>#okT5gx*pAlKyofQ5IO?D^gLrPHOb&FAsZwYwZBuG;_v`1@MlL;3Jv7YAfsl
z7$i+hG)5shbKX2-l;(n=W*D6hzSV-a@9hxaY6~z)S$gJ-5x#}jFrJ4o3}6aP@&-_~
z*n#$`%>!y@fGu0G4p;0VhL>kG_$xeDI#+EBuONl0!9%YEU50q#8~Dz6;lfBP-vm#H
zKJF&Zrjb|_@5-R*0^^y`g|^@F3tKDl^Ka{9V@TuQzyyw0Q=?O~$HU(aUZthvU@}~e
z!2AgHSWpEtE&#eu4hBLqf8@s>=iKWeMFptB80a+YP(CwZt(f)80D2VU(3$jiLw;W8
za0Q^O1Uh`huo(!Ob8BThTYB68aPY4B^bmR$Jo;a}clFJikvPX}JUkNVb3j1sdraG0
znCFSRHB(cwLvALD!$=OA;UVbvJot*gw)$Fe?naQV%l8iB?MTt|)HtZUWUUK|H!(bq
z`*+156M^wrGCXk1KDCZriCz{^X@16l@6Xh!cOsC=0IRsx4fOR%kpY{!#{z*S92@^5
z2pkPMe&{gX^ux|!FuL$scFP{bVJe5}mQK0r!g9;z-ot*-uQtH8YHf$k%H3Z<2X&zn
z_gsLk7st<<{?SKHZlKZE1~g+82q93Ex$8H*0m8vty7U*-eqgdvYz4k$*h`uDRS4Ct
z`&OkWO2f;$U19aZ#Ig{>c4`uMU??8l+DvNj@vubWJzGMqD##Cs=od1ddoi@7x4d_%
zSDL@bvexPKkYBI%^U0O=;mwyrN@!^Y#N`Z0igQO0%NyY5b*!6r#`|-(0qz+MK<y>r
zpgAcx;pV^!;v)Lr+yi&NGG8U}*DWlP`q?^_IOn6(lS4>2Dq3J;9;zZcG77FkEJ!US
z&{?cUC4L0zOmu~?QqBkQPw-on;ZA(j3z;<{m(R9s36G1w6HJ;oF%sQ?$D;TUEH4Do
zq9-=m+VUjS1}&WFBqFrRofhf0_czPb=wT?=q;YYJ(4#wZ)-2=&%M0Gle+~?jmXO}t
z-ZQIuqhfUIK%bsA54qdAf}FB%kK6{>8PemH`x}5^1&Z>VPHHNn9Iv9KrQBj2mn%|#
z=psOpJdE>UMbpx^Eo*8J#M;ql-}!(|_Xm+QVchv=Wq@Vj6F)z`=6PGhk`YtmY^)sh
zCW_j`VPPXm5jN?(BNNdU7vyKg?*|UQ-^Mu%eU%&zBI$~YH<UYqAlGUP+XN+03ff=<
z)g!ndF?tEaReqX330k%QiP_o0TCMdvcBWhum&KCD+E19hHBV#eJ!nJqs1lF&)HJI~
z?jsc`xp^(U<-@MVk>zEQ{1+9x`2J$EVMEcWE+o%TBrF-Y*_A8UB+XnclQ@hGHGgfU
zw${zpHw_k}4LV&h)B#QVJF~tp;}B+$E+FziISgfs9gioSD{6Nosut~?ZguZLI|%_g
z?aYuIWNvy9^7WJ^v(m9=^}A*)Ta(gRWzXRUo}A!s%a)~|)1B0D;~=k#$U>ZmMx9`O
zP2o27<8;B1EPZ&(v6il8GOh5Xu9w&1`P3=0XrMJ9LOGnZlHoN;n)xszi&Oh?`lF)*
zRMi(dC%P`z+pmtL0FFY?k95LLkFQybZjEd!mWBk9hf;+BZw%u!0;wKNxl#@zMNuFI
zwK?jSUPo<{(@9=ni3eZ1nB4ifNM2aj0iytUJen}{!DxGSY@Wz)G}6GO_~KGD3l=c;
zg_HV4l*qFst-8q_HB~-pjyl8bSNiK$DZz+YQ3{$Lk4AN)U<|Zwb)1{z75V#66m5f*
zE1eUAJLhB%A_t@rrXrb}u^l;p6bI?aU?>gx;J8|r$^C*x&KJaM&6siRW(mV_91%uv
z&fCuU5`cR&Z)b)EYZLYj(!F}GM5sm%0OSHE*q!EDRu3x+zgM4XXh^`loQ&0~>t`??
zLrCtnAtFx1@>;|AO$QAOl!Up~kyEFEzW()T982L4+V*og+xArHior9KC$GpY+In4$
zqPlXJCOQG7RvpzIqxr)`-O;L%!KsJFBNa)FpdS>N!?sVRCn@(8z#gj3FeX`qY2ufM
zfmsXs&%#|{(zIzg!;{A_SWcubgup=1zlq250S!Ty(ul;;04IRgee@k0Ad&=b3UluQ
z-WI>HQWs7F7zSQBjWd?vw5`%UG3WE&pG6|}3rS{fTmmt)UM!X;h>j?nip}K^l=LAL
zFV9?&D~pcy;pD;SYPD=KCp=>0gS#dp^jRv(6gifPDOusj5$R6=i<~R^QNb&I@d&D(
zyE?iv_}0*KE5EJL(*yb}Ub}<AkZnb=KN{<;1|R|~&*Ww-5so%f)v0SMzeYOf+w~gn
zT$a3o5gH!~QKqu}RA!pqtnrLnpI}v?2O$=mr*mz(uDlc7FOi53&XC~j@(4TLh!DVB
zw5Xc(dg_!ZR41D#Cuhw5D0^fiamkaPc4(V2bS|A4w2=C{k^htxtoY&_GC88rgEo(2
zIR5Pt#7?;PK|!2{YM8uz431}~JfXjKMJ3bovv!#SPUI9MFsfWNMv~fRzJ2Z-Q{Sx2
z;}gbQTLvioMjJnL?x&iX)sW$OAL@a)59LUpcbI!sOH0BjwRu1I{60mMV+*lb+m;Df
z5l>20<=yS&-MIll9Q-qDx!32EJJ;O06>&sANo_&gWLh?%lOs=`UV^M2<38RD4ONt_
z$jGJWixYLGh6JJCq)82T4$f|_q)IxZ3(^12_mawO?d_wfr`Yq8fpH9xc5dMA2^Q82
zhLg@ciK9oW;RMD$w)Ddc*9-<3=mu#zjzj6z)P@J1#06dZ&7JkpZU~eqsH5nz8Jb7^
zb=|+ag|VQT$NtKU)*7%G!$_>HX&c27>7)MPPpIOn4;bSgUZcN8{Q2_#yCZ+zlYg~-
z{)5)Z!^f)8jmXF<$tt`tg~1rR(#C3y6OM%ASFc{JpkcU<nJ`y<E)B$^&H5e(Q_$w0
za?tF%2xnH{=-y_v;pp07e5<yW>41iJ<=7ta?oU7DC!u13Q7W?WvyKMSbj|M224sYe
z0jwvDj&4_p@GmD*@`Wcqw$Y8qcqsNzeH=}Sum0hWliJ5U{fF1^*L~yvEicax+C2Kc
zjvVBfHb}EZFP>g90htvTq<MOlj!x}}(daiZrfVS*saQ#q*^5%6D;9rDL_U1&<9{{l
t{@bF*Lw!9zRqh9r`2Sog!|CGt>Bp}vUzM;Sz@ENhv(CXP*Ycb1{|T(lBlQ3P

literal 121898
zcmeFac{tU3`!>Axz3*!8>}ofPLR+PzNRlB_C8dEfBqfSO#t<@hYqvv&A}N_lG9;3W
z8Ma+0ie$>%CS=aE3=8WyuciC9_wRlGd57mXp7(i=d)>!zr)BkB-|y#heXij=&+FRW
zN7VK$n9n<(!C)*<-LI^HKQtMPxspH3#XG?#lb!L$T*HI=lo@RLUsP^v5QDLXp{l$~
z>rz;6lgphd;|b{*7n?I@WFl`}x$UFt!|TcG8ULezr25Sz8y}heL-6YT(K)K`K2-gr
zF7VLv@|yXBT9b<pELr#CB4M+*8@xaKux`z3!3}40@3aU^KRkBsoQb1JopI-AcG8%=
z_(12y^8T0E%>!8iLrc7s1#vHbz09&?tNH%=D}MStFX|3G$bY?F|L7C{Uw^?5m&5*7
zd|-lZGiCw5ob<W3caAolVK9zxJvot<ki7@DG~=N-GZ<hrH9lC)hb2#YKQ%ENZP}2{
z`5U9^I3B?4&LYLBtuv#^9@>s8Igg`t-*1&v$(-*QN7f%Wa3EplKk;r{MYonC=ePg$
ze=JYKP5<>5{IF)j#z+77`y00Z`2c4*{<99VR`LIE9he7Crd!~6-W=+#eyn4E?X`lT
z7+p}QZ#}zsL+>2=dCkdXn%^;=mizMU+|l>4e}`qgmZKN_f}s_tx0U`p5+wXT-8y%>
z+s|V#o+?UDeu-}Ta8S%ILf)n5must+Y&n_k`{2QY<9V)5a?>M~;^#j+Jf3c$8=UAi
z^kzeGOo%>jQ?~tehrar~spsAq_SL5i_P-b?TB<I$7-#W9;;9$u2Wo_}R%|)(>gGOv
z-EP_OPgkPtdul2k?7bdmkng^1*)nBSRaJ}n=fXRzn^%{Hhy~Odc*u8@M+#KNst4fP
zwZ<Cao?;c0%Om95N)2XgFLr%ce<ICn`IZyUy*xcVkLSC)C7#R>U9tK2&#5Nm{&yuz
ziY7;U9&f%K9iSjP_otOV$40t4U;Og^m-$>=ZC4g4cty^P#18)c{UWWpY+BALHQRzI
zsjSmkr|EBI@&9B<9vEoOUAp*B+pLnzbc?z`=72`Be(nc<#etmR5B>sMPb&!uF0rhC
zt~&MSB9D#31O3dthNX)5i-eDSWm^8%t77XSJZ2`g)iysruW9(@{e5p)`|ev=%;~X~
zNQWdNe;J{^@pJD&WB3%EhCXXd=MRSu+~V<Ys!BFc>~Xm-sAFtd_iVkmP3x;yyXNuT
zatN0wlU~~=SF7)0StMTjSVt_yrmg6f-Qh@u+_y5a4!w8o$T`_Q$jp>(8_UMhzKS~<
zy##Kvq`M^m>+|YMMfB1a4t)pe2WD6U$;?PiQJup%aPgr=AD^D7o?-tWT03*$%^dbM
zy{CE?ghKS5mbUwNz4BQuvTcUNg%!z)Q)4i0RC#ZRSecwZ<Lo-!V<47ly=&L5sqtb(
z|BC2+CdPgI^3JcwAD_&yGG<yfX4?4j%ga7Gba#n;_xPxl$NQ&eUXIsf<SoCxVvA2|
zxx!B4d>OlrSASgP(oA2pT6$}nWVr&X*R=lm`4XmTkWh?8X=&;80S`7S<#d7PmzI2w
zswXF<+U&Wofvgs``{onRH)cByL=0wLoIihl>zfUczW4WBd(=4W<m9BQOZR-aUS0G~
zf*`k^jBNq!s@8IzV~oq|vM+WmDVG~QwEEJgf3@>U!6>BJQ^O)@%#gg1m^%t?&fCqZ
z;`G`?b4Na`RCITZsZm?EB$!uy;<-iLHg`6wWnYNsDGer%#JNKy>9mc}i*%KhN4hFE
z7&pM=jEa1DR2#ah;+IxFI(+?Tw8~z0cXt)@^>&~BHLoa4#_oaRK=bm5wxx>fiSo9$
z66J{_a&&KNxZmcj;noxX_nj7MF&L^pAqetHS?S$Racr{JiK*d-(QP+<AGO;kfAop9
z@kM8|P5e;<E?2K!9lX~vCDh(^H-Ea%Iyx?HWBd3awRKN$wDc|%%n@!7aQ#}aQB+i|
zeg_`yOGAeBUGuXK^mCm#|IB}9f9Cvhn=+ZsSK2zKw9hHV-lVX1<HxMi%;j01hHi3E
zauc<Q{LSlU(f+cVvK{(PSoW}oLh^m_#5$&l^S-pFo_YD(c+-G8cdee?;V|d!c$?S1
zTvNsxtW{K0q(k?9&mup&GMNoUeyg~5I`r<BQ7`uAi`GfjS8YhQY*>+xh^^`0hx5MO
zvVOZ|n89?nPNn;rvmYL+3KyU9m>xaf8L`9i1YNn}`T0t$!mf+mRl-I7eD`d{makY*
zgoA$K&!EF1!{05Y1$`})^{Md}?#<jk;V(*k50L7t@q2eqs`%@lFWbJpToiyiFNQbb
z$saEiD_LC|y)WQEa<1#dQ07pHNVFS^dGyGUpZgjz0z{H?w%ryph;V4cIb0T^XRCU(
zrZLyW@oG_fD7XHlUz|s~m(SzBbXw1$&RE=`_v0NYYyHF%X`A4!mf^(t1Srl(H@ghv
zUh0mIHT?Riz@~`D<e}W4L~F~UKQt#M$~_{6Tr`g#U)eI%Ayrv)uYY^R9Ih3wJm+vR
zvm--7xD_YfsDCKidEwLVUuu&L-rkcEqYGmTIW96Ot0OeCBh*~Of9KHawWYo?OFbq^
z4amkSl1{zYxN)P`o<;K8okzYb9&5ByxA^$<{)gf3u^=~ge)d#V5gl*&?D-x38fQu2
z=HXMglrPP>F21~yn{ID8xdPkdRd}7dI0I(2P2hoJe`A)0xtLz&RysmK8`S;BT09he
zu3vA1PYZYxw4wON4PigC;Sk^-r7xVmTGNzfu5-00W*aiO*QKQmcF4m-E!o)a+uiIg
za~88t*Dv`24lzG_%!93ylw8Xtmh-c^!o=AUrWhVXxPk6M;@sQ)brJonQ&^<Va7X<T
zrq+g=%r6yd`xGL!86P-wXX`J<tW29WLuPNCaKk>|l{cSlS*kc*u&C8nqRgv>Ju{Yq
zJ%aiBk2Sl9b-ca1l-_IGr#N{urUUVgvBvTB9GwWguO6@Ggb2;r%i)J)FIN|@Re5#K
zI`8iDzJJZ*-@lXj>B^#{s(PJM_Sz+t%M_=+zFfK8^ho{ShKNgs%T}zo?Q$yi=CLPO
zu~-Z9mOQuTrWJz5EO=hcbhp9l2EIXRvcl&ZvN|JHEm*bVP<^^xXW6B%1@nT^!*MQz
zYp;7{c1E~-NZ+S?C|p+3xPd$G>s9HI_j^ji)>h{@HanSIEuwSzd+eJzL6)`!>|g$V
ztKPWeL%A4n!PaocX7AS4)?0mt@9lW<wLUqQc_L?OsPsHb4qh(BwBo_7zKMwmn^x})
zw_W;_1&{WNXZ4S_1q9q&yRWD!Uejo1Y9hs~I$`6cO+L8H{&h=YlJKiX;UZ1nkC;R_
z3H1p-k5|yDh*gigW%NirlE0@mS-rjn$LCgC&6AVd);WVJB~10H;YjwOh6pP^EmR_n
z7b_39^V3o@-n^a@B%JlBYS!*Pwor<_pLQ+<t8uFjX_1eQk4Ac|?6mmOa``bmm1+G;
ze_nmwca6{CcvdZwAulm+^#vt8r}p3!qyU%t(rR1bzr5V1Ckx5==)fYV?nOFMDsMg9
z@vbsGfJZ{exS^2SU^Q8veKP!%O@CJBqt!wEdj$DpFYd3aLBPyaSbd|{d1j)sGUuKb
z`8Z3LZDM)jhMNuYavIy{&e!l{o&Io>>-MZSbkI07hjD{~S84C?V6i|fMXSeO>j!ej
z^i`%yoW1V!pFOxK32{u^uH)@Es*zLo7YRSqJaXhlx?9ODVT#@lEwO37xb@&{M||_X
zD{~hFG7VDC9ex`oBVoKcc;e^%kC5YN_lpmAmM44er@MHqZCmhAkULC%Hc(V;r;;rY
zF?`Qmw-4dRPVgZ3zKd(Y5&MwtF?qJcLi*F?@1H4j&U94l6QF$VSZO^za@fW5OVq9_
z2Xg!2NlNnaSxo~E9zI-m{v`6ZO)E~zq*>HGYcx8Rq^CRQCVlOV$Dcieg7ENVLf<kZ
zUQUGS{g@K{R^f%?KvH0ug#4UmPoHkJWOcl|SDNd9uvgC%Ivje-rL6v;dZdC)$6b?S
z=071ms%DMJI`*$Vm3`6M%fsVxYj0|~noRlUH$j@|F|vc(8@gdf{o>2(YB}HRqj$m9
zw!pTaZrwL~kH*oA`fZ}uMDbp*``xg($@2?%x9@M5fh)~xw5t&cF>rUPXJWI9-buBr
zE*9``-G2VVI*Q213|x{Ltn+g}h;KfXy&V~#VvGfODrIHYK68WX2niedhKEYUijkOV
zl%*II`}?XFHq|C{UlE~ud7AuYPOYA8LAAOlRhE|%$V){1CewaPb6a=qlam?G`$}(b
zY3R|{jZI9);kR@N;Fm91n7CqR-}6e1^uCWNYt``K@i*%3OSo2S_Ch$?CwOyXb?o`d
zN4nldBVQ_nDTUxBvrurGeApY*zwU$jx(3C_<Ihcx#xI(yHFCi51gA)?a<DCs4AJ|V
z{|0p}s)3Ufo#W2Xx{^1lSD=KPvJ+z$WY%vZ6+Mz!0&XKm>zxARnICUiz6+JSJ~PdV
z<ukZ_>V-p$WikSBOh@toKVC^66+hk>i%1C*wfdzBYX#M0yB;RK+Gbvpc&X<}Mqs+6
zRipIWMf|Hx!tIZJK+c;{jBs3{XT<To{<7V1rnO0@&J}PfXH~4%p}$;}Wely?+)!y7
z!|m41a_3LtfEpoBJYQTMsi+8>O|PAnG_QHcCxvi_;PtTwM&y1dOj`UzhE=$Ot1ji?
z>C;|o{o|^pwBvDd&NmDX4_Brm-pQLUdcRf>$yd;`K4ycp`ox)Eu@HhgtQwDZOsUI{
zYL%oT|9<uQ=GxMQ+q{2Uu&Rg>foqCWXRa2}3MyG5Q5=3v88D~5N(%-WaGCeZN235g
z9RoMX$*}<e+@K$Pg$fduHdsl3j=t0SKGlbd7q=A$s5`UMYV{8ymumCBPRX4ON~3+#
zlcUj2!yPuwPMw8z<?`G}7|vX6Dc`We;^U3<>7*C^J`@y?9*N{0OSk;^^cMgW8tIX)
zLvIG_%H4OIZ}{_?Kt#olQSIW!CAV(HmI|?cezk!PRmBues*hCjrM1Q^3ePKdSR7Ze
z+pFw4GaMNJ6PKsRNm(OhxjX?>R9>z>AbMOhbMvg{xZ&0naOfY5sQT3oAJ2%R1hEY`
z`NF}n<XRdRx2W%EPwl?P$DU~TE5dSHuX5{e9Lv5-Mp(~WswVx5zH?X1L(7m;4!TqQ
zj`<>Ep-6HrWfdz=?W|$XME%j<^rEdj#K7OGDSLfi3o4O2@-7aIuEX!dt($X>XIPzD
zZTtEcn~J^Mg$tw_Zwy_m^QnJkeDF|&oJ<}Z(K)$u)d||-F3euvNued)nS({Ej3!6A
zR2znehFbGj{nzu{SVh17xRh?oD_BIQ2u4VYT#=yt_{+x>ql<m@l4^?6BVRtm94xXS
zJ5&<A$t8MNG;{l`?|Q93zANqRjbHW{A_q}zz@i(%C)-Sq)(*7A&GcD&uuk5(bxYZO
zW@f_t3#!FCL*F?ozgoO+_03VueU+5@>$&u$-3$o{Im#B+eDn$rED_-W;}b`!mC39(
z*JO6pr<v28w8>C{b-DKjSCXu4flJVA_}CGz*@X9x)W7mxs;R&bD^2#8?iqh6Fmofl
zyvebJZ|~l{%R|{FiY(o$MQ(<$B60Z51MiN87X=6ipa507TC`&M^1_#k6vL`7-QKLb
zge<&hTEsTH&%X8)cYxf`{;Xp9Vn)}j%c38TyqI+oGt7+q!a3Y}KLdvluFZ_sj4eVe
zDCx~Wn#3v4)~{dBQ=aQG*1uB0^-`l#XL!4h@ZCROUJGI6Rxc|oEVMziD#k%5kxW9Z
zQHHAR__MRBHAcMh&KFgtBV78kj^|vm$?9xDnylwKSV#-I&&I|ktC&t2<IU_XeT!gL
z6v51*bW_h2s%{8hCYC>Cd)G}0FhI+6za!_E!9f{sXA_#ht#@Hp4ts`W%shoJ40GzZ
zD~^K3w?nkFweUJsQ@+bZj-55mbF&<5Ev$$+7(&Uo%5-O0_^J(I7j~<(l=@tn?246;
z{UuFs!%cU=oBLn6jWux>^EvAIA~}EuI(FX)E@LGvpq}lU_2RalU(cVt>$IzS7v8P^
zuYw}jfHv?(<4Nn><+wpVO*NVIqFJ8<zaV%$H?I{9-C@aVUYq3S*y1J?ajDh&(wF<!
z2pMUQV8_Qs_>{Z#+g0Ryu%(fgw~@;@_T<D~cxtK~2*jY`kwA>H@SUlX&H7lx@y1RC
z?w0W)sn#&NGH<+JO!WxDod48RC|jhvBv_=aN>kmBs_^w8dP|yKI0g_R1LHA5FrXax
z&Q3=wIbb`3z;?9HZ$`2OlU=$<?yFLX0({!_ef&-jKbYqZm+ch4^!3w98QZf6No%pu
zD>j5*)PS283YD_vKODN<JA9{O5gwFEO>#5?c`U1(iH>`NQ^V!O$cc=``kR9NIh$>^
zH0x97K{y^x#H<$<ySaZ|!p<KUGi|F{##Q8)AB}+B9=Te!b0CLm?-J&}zMn^@n|VgL
zcL2rjYPcaEc)WF%kx2GO23kA>_RXmx<ehQEP7fZ=hCMx|<G503c_5ci!)MQb9eQ#+
zMR1|a=NsJ%=?2UaXL0W%GWVpc{o&VQEh6QdhVnS)ka2m_Y!=O^yN}SsVA@eqf>R}T
zzTug1X=|NHWK+g>oO5);{+r+ABYQoq=vaAJD#l{fP6vy1OT67yE+E5ZXz$`YCga&2
z@#PFg?={}-raxn;{NW>%0UiZ1>$2_5QLvm5*he)rLn-|=Cwwb0+jR(yRyv1zb``>n
zebF@5w4d_2<tl_%<D;{YUa5LW47HllmJux3t8#Z{>g;pkQDze)MtdwhirHNNc3)6V
zTQsCU$=<_xhA5wJUsp(J_vI6x1=!>J!rUjj;9{+{95>SUFtirWddaxVsqTeHdu!j`
z5nnqrG?cLOCo^v<W)*<mT^g;{X-g^Qzh2Dmve}`(QMc&~eS=b25SP%s+1!4O3)Pgr
zyHm|e=JbEOP3^~Ogf{-|^;GOXfIBV6e~5=!qxj!s9X`|1)q0Cm=I1$c=5Uy(>d){T
zepz;%$B2SbTE1%A+5OLv4Q0OskQSPQ1VP+wa^#zXT8?al6~D?V^B2U76CA?pu=_}~
z00ryItGAwgeS@+Z`)fG#l6SYBnI)lKKl|T=&LeO6*Xw@`rTG7md$@d}3$%!TPEL+|
zEl|+MJJ&Jq!$xjT&Un~10U@mB<2|zHluhAsm+7~raQ(;yz<N>;`*4G+43LkK(`*b_
zj0cMQ2f^r()4e7(HM%B+sP6dy_xk|rr}~Sa(Me|CDgx=!%xjN1#wl$<R+2Q1bp1u`
zG$6!RlC^r83l@LJx&nYpW@vc$U=vb+lRpV4<b(<$X^xvfEbT@}(fXBB5dDk*d3p0>
zL(9wxgPEa_wmU}t-UN?Tsw?tY=Z+q89sl#%Q9ZrYk9Cs$iM-)=`h4@>P%63q<G?lS
zuPv++3=%r*gAceRvszfqV-I#+nR~w29~e!S@S{jQ$}PZN7%wX>m?scs%OBfT!_Pd_
zR{>-n@MnwuSnT0@E5vd~G$fS~6_40qJtY67x0Mwrujw)y+h}cGdm<w%q_RkQ;&V_N
z5I2!p0E9(_$lw>F{yn~2<+wYRMs)O(;hVtq)qEcmI?aEdi359L^`b95w_nPvYUA96
ztB-e0AQ26_=)S)&qAs734b_1blrb#nB}nN*<9Bp;uKm2*ZhAV^BarLSqZAXS7$-@%
zjM8zFF_v=>xR1O$u)kZ`9!oDS>w$0QXaPRo+*f3o(X=8tXXpUfu^oVo7mMg5`4S)U
z{{Eimz1LT`Ou}DC(jKx|`Ro_TFXx%v^<KK62$T#n+oN!~4q@x5&x#26?D`O+d-r<d
z>GEDh)VRxH!yW4N$4uUZY9}rF5nEZM0@FDx$bIPWS$twPI>YB^mHhs%@2RqK(eLc+
zECz1KL#5w6t88JycL~$30?)J%@qWBSCvOQLKL3ms_no47<E`>-4g<}6FMnI`_-1V2
zKUmk0Y!>H^HQ#&Qm+&~Lq+XVa1SCktmI`KyF6Y>R-JOLfYjOkDB1?$Vg=K~A7QGW%
zf>q`0nv#{~p*c!dvHQaOpep2(QPsEIzsB!FUJ-E6JqL@PnkPn}hQ#3xHOA&GT)mQz
zw%}J>RaQh1u0MFY*FIk}R;>cnOk&a~$c92fc@xW2`_AE^MYF#=Fm22ngaN5X_G{^4
zBTUG4y!Ez9-jz>f-=K8xa^oY|`24yYCr)A+)8AxM%tvKvuz3XPuU9-KVJV%XIJs>u
zEiC<GhX1rJMWPzmJb<UKsXef6iMZh_uf+0$ea84kqCAkO*Y(VxI5Yx_QT;P`38m3M
zuDNEsS*OjmVOeH#&#c2p7ME{xpBf+k5*@(IQJ5KTYn$vz78{cTU?pzVm`PBG-2hO=
z0IDWLY--jab&o*aMpVvf`>LZAsIg6(bDUN?HrfobXIj|TVL{rCUe#2ek^NEj`C!kr
zg+!gi0skW|%U`DBPG+l*NGmNga2y@8aIlM8lJX|IrID6L1C*+Ob#dwrg1qHWsw@w+
zYcC15X>l7XMD@KA@TnKpg9CGbG9k#LnBpLT&M~e2arJJu_OCMp<StGu8=i}_8bqv+
zt1=3Z-|^c|JWIgWCXKkn&BqQ?$s=Uh16n*dQHBsFV1@~2`_+-U>o?k#ZmGa$!KCgX
z^ay6LiY~J@KyF)^B6}*cC+S72pHy>P;HV-p)bTEL4_+$Ft}K+@*j<|}Fi*hsLKp~H
zBCj_J3&+3DAwM$pd4r;NYMHEvzjUh?!129g@<5yT_6fLM)C0^x(1->*w=S};RDS81
z7OitF#C@tedjL>GDYHoq-2Lw}`$aRSK6{o5aqI2K`t<p;XZr|qo^1iEJdZ=3U0!{g
z5v9@vGZ7Rl$stJUi4DK=m-{iTcV+F(Vme-Y4m^0^ofPPqTW<NxwB)|7u|hF-@is|*
zw~-1BbE>RdM$aQ6s2H%>dG4%+mfN95RdGjS>aY9wEWtXdhS`<~pYMEAUas!DQe5z;
zhK5FmOMI8X6`%3yWcOg=R8@M?fJoBLtp{VHSr*dEdJ0dgobNte62_)e`-OD_Xuk&l
zq^0sM%tN2s+l`m8u*-MGL6?<=IU4jvvza0JQ|oP}28&Hb_|<@FzCztyD(RtpZ|j-e
zACAKs4U(MrfjHUrO^cq&UI!-;elZF7OzdhMEv;Ke<?`9n153kY?KkNhqVFI|g6EwF
z0$5C#$!K@AOP&zO%=5K?m5mJRo*DBzxl6Y+)h4jiVDdxAHg|R4mIoW1d!Bgn?Q}Ss
zQ<SGs$vh-@rsb?rOQJ!(LVNirFu9n<VeR0PR#`E6EuB9IBYeiG0`I$BF)#@KmRKYG
zSb{rje^swuoZf{(P(FP5dKb(9Qf7S~VLk~>H(Bx{OK27AoX49v0RoD7xQ5sM#N#Os
zvR{&m;C+`5PnESrKSvs*U|y#^V7S&lxWp2x!oHvT)5cNfnnc~;&}5-Hkp8aiA_l_4
zw@L#813AX|tW@+mEV3{9i`j242J&{!UBG*5@Qux?9Tr}&UD1xR@Oy*2#JLeUPJVD3
zOAXj3aeUiYPjpj-%4B;I2}&!9t(=DFzzd$G7%-DWH`_a*abNyayI8bIdBosYK3hI|
z27HPiLP#O_%fR%OJK{!yVq(_ik_g4V9m!s?5zd>srYqHUZZS8niccL&E=$gvG?N3R
zrF*h%ses3n%}*=0t|7O1{ORdkSw&?L@;lY!N2>r;?A+vL)`~-Dgp-(p^Za}o!j5IQ
zpqWKurnqH3SR;bK+8XW3{X+Deh1;YFQ6XhS+eoE)007>za!J{<$I11CG4n7m^JqZr
zstp>`ITRoKaFX_kW?I-57zc5wsf&&)ejYJfnDuGev1K+|^n-j+$^v?xT-Ad|B#1*q
z(gc`8H2&tI&u$Z7djUh&GAj#htv65h%V>5MA^aZ%ix0PftGEI%kdN^4fW;MmTgb-?
zO6m8A8G>8@-$wJUZO&*0EJVQLSdJ(-R8U5R2JFdNf-ZX-vm{8Qa`5(MZ+OT;;GF#6
zA2yE3A{t%?47oSgd6e_5^i3eMTH68?{jYLy7md~$7<99?lArG}V22J)ZHRQWZNwK1
zVx<a0CCv>%h^OG1m)iWKT7jZ@m%DaIxJ>3vGaJMKh}#w#XtPX#Q&J3+U(^79xoVm<
z7C<l&tRNcMOHHzYfP_hzFR1-Ko(wF;7r0U!j6A)8=ma@dZ>s8mN7q4vl6qcvYd_?Q
zx&1BR=7@Vse3A;lj{pPU?J0lQ_133}Smg{BWy1N0$Q3`yXyye3`6iPMoXO~w*J9y@
zrQXWKd2>WG(vRqA$s_nGT4**gt0TOhT7QWDlBYU4Y*ZE|v)msD^l6W^+<FSO1%M<?
zR79oavp#*m#?H<9^sM@`+gc90Ba@Ilqq><9*z-4EX(_I6_f9g#FsuFc$!FWF0p=Zi
z$ATqo=o?E*cC%EE6hnmLDQ6>i-VSAh^Ep)Kh<F<_DT+hdW)C(at#<4&mc*O1gTlpc
z?@DZ_0N|-H-Qv#byESb-SsNusd7nH^g64)nxVV_N;y9K2e_FL;vp87qTW*1F0z^Sv
zY7f@Y@ShlQ!O3&1m&RESq7^no+%PW6!PdRf$)Bv&7>;#GW*G_)kmmizAAiVCBNe=N
zN72K*eoXw#AHOQ=>o-S;>*(3+Ey*MBEq)wlulu>3P8Gbuy&;o!$V5scYh#UO2F$V2
z-sBw4H{b*_Lmy+`ggZBF+H`>DC8!LE$pNKDwFsoQZJ$sOdu(0}`A`b%^`Utayrk~Q
zPERh;jkRNGo2q}658D4xN;}26Kc7AGaF{r;6zsU|J8f+%EUm2vWS6sYAf9Rd{0o2n
zVRk(w+pfS{`CZx7k~tv(dKWqdOXqCZz#TRdSo!#b<eBnyOPGhXd+$yE!Yv1*#s!Wg
z9*1BP1&fkjhry$NL^Tprq>&5_*w@6jE+5Za3wF+hQ6|1UxpWHL8%GXe%+kpl3eE(m
zi?fE;X!Z^l*gtbB3u%fCz}gU*N5r5|$^qasbWj#a(IAL;6^;xiixQYynrd~!&d6lX
ztN$eTCF)8mrAp9)WWOC<1q3lbVbb)ub&Gp>E{RHbRM^&a&HFN%98Lln+afsQHouF+
zVA6m7y7WWQg=%e`LY&n6&QT)6K}6imySix*=~%1<YpPQ*vO0}U1tMyf{ut_>nvY;E
zVmb}7Y>gXH!U;DtJU_n_bWi|EBPgvu7$WLNd<^nIO@AKX=^z%gq#vQ|DMaDLLqZPV
z)J4@g2I0>J5GjpdR4z>hxz6?kybBleAq5zubQTVR<Xwn5{K)e5l>{XbO+*<|3Hw~a
zF71K4Kd>GpFy9^B#}_;@^XTwBDhqhZu^UengSbQ$Ii$M;wl7TYX_)?DA?_S@|0)jD
z_xdnuri~f!n75q6dkd8IZbwQ{)k88`k>O(*<b^1I+s0b3a%)u8%RC|s{P}h!Y<tf8
zXTHE1NkH%=H8R6*B`EqBGbx*4PC3Du4c45rHs10Zk!|2wm^o_9-!}$Z3`FC|6hoKt
zymJ)C@By-Rp2DFk0#j?C;}R4!!~q|CwN(A=_N0W(C(}1~O;1ftnZh|At2IMPfCAMO
zRHp0KUGSRn!4Yi0qz5O)@{CJw4<Zh~y8PYu4=Bk*!YkeT^kwTj#qnKor~;41stp<R
z!De_#l6bW>t1rF55Zt59Tk&FL=-4<k7&c5A?23lJ6Y$M32vc&$ia;QTJ@>%Vv_a?S
zL);VL?ns^d2z(`i;1bf_wPIpoU75r!?;bdB@VJ|kD^}-C3@nm!JCP8%6In=_pWcOu
z*K=aA1w5S1Xdf}#_P3~J#ffT`%{f_BJr?pKQzLTsc375_<W!T6>t<x6!MS{5Fs1S*
z-pMBuK>}%9%VTW9eIykwGgGbC#ftKwqS~a}i66`249HXFmNEAPvK9I+B1Wbd7W()W
z8M_F)i;#a&rFw<<I~5ycWzwlY(v{sgn&UJa$Yhl{=E1icmP=wed}}ubXt_MX;|dEb
zJXQn=-+{gcby|9t@ESdRO3gIWn@|(Ggx61bgL|NaQp4X3Vy6<mQrP11wfp=J)!*(P
zsALV~=hx<fJbT!yNNgsXZz7UC5jozilOOuL4+NOG>qx~u;m#+j3AW|RAEhF1KXD7{
z|6aP<ZN3Dl)xpos29MG?I=4ACyn%bln$_?v$8H1}f7zCG@egMNM%}w(`RXygUZw~H
z$2y)^!CjYUS|cJJ$c=y>PJbD1L$sB=#|)wdM<;}B&Z&CYWI=9VTel<Bdbgd!`{rGY
zjPRHkYUlAQWhzV#Tx%xj45hzv9?G|z*JPZ16KL!AE?wOtu5&3eSM6u|XU?3VN=3CT
z69%?PZU9d$z=q?{OpkSWuX~Jz-0o!mC1R6F$nqgwYth-Y%$c1?BSSB-`Z9u;eedtT
z)E780@mzhNQ=Ou4`!y-E(1d!4#mvK->OM+!x&P)m>9_*U8hmc1>u1l(ATIrg3hAs*
z?10>z0f7l}!}DgAH9hw$vI^e2Z21o=KOYIav+_ptMlY^4i}&u3x%Ol0w&KjwzOqI_
z*I#^pu2JRM#*zx@yGJG3ym_x&juu)cz3GZ4Z`(H0t(i}`=5IfquQ(M`{OCpY<AlfA
z*(`U3&YaGXT>Al+$OQLQgIUwLXCHAlj`XMwbOnd%f9<TV&#vIq(r1@E`o|g*kRM5@
zxrj%eZ(sCK0;0aFV@ukeg)-}Uj6l7fG;u-Je^oTLng=q+)qsrB8ayT|BU_(Ws$EB<
zC?uK}Y(CZZ%7+ghPED!=nw99^F#6jY2zo=SohWP7LwZ2UJL%6i_B=o;bM)xZ#qnAR
zWh9>J2@C04Fu6N98Um_VW)G@{SHR%-+uPgIq7xGn9pKmQJ4}tbL0Ff}oWjM=ojNH_
zlu5z#^5x6-zUa_33D+L5^Y^*~Jwy&NF{gKRA1fCuf~-cU_JWbrGUz7wb~^UIjgR<c
z&u?d4Eg@QZJ6;b*-eUm~fJfP-UND5(114u{2g${?O-#7g#2oRJTzaNynO#$FdV>{@
zK`Aqz_#?}3{R>tjo~vTH6S*7poC_i=E6tlmJR6$-+mgQ6(MD>yj5WjklUhX<)|b*9
zs;jGY1zw<duddUF5c+=UdHa%*l2+npz}nW<;Ydyr|JaITzzeteO5de#d9;e353B!0
za<kuUi|r0;ABuNf%&w?yLDsMmG%jWTGk*%q{6ZxtC8UPn5OteI%173{JGI=Iug|&C
z1^&gp$RyJ3*v=1K7yC^2-;dZImVBtC!L>#;oxVT2C+TrU{-Z}iNjXC$IURaH0E=p$
zO$<LpOyn_uF_hCqWvcCd|8Cw{*DWCiY6O7TW?PB!vfiG!XP9{RhaMq;x}<wh)|_*F
zf;e$?6Bwf??MkjPKm7}BBc7w4_YR7U+c`Twxu<`j;iQyv0haN`p+B@3pL<Cqk~$4S
zQp%-_R4PO5rREu$@mje(&$Z*VHuV_Y%hqksw+BT}<qPEqR75Xm{(1UykUG?H5njmW
z_C_kWQ8ppAE5;sr;Ddx@l9(%eY;~j?Q}<XgmO^r@D=AAq`3}^>Vv?EpAElV;$b(nP
zUAmAv&7K)x^W$;1bf=yCBT2nu`oXhj&sHR^OSSO$E+)-%z+C8o+mm}xJg#dqzdO^2
z>c`Z2#M3UCyK#H&r$=3LOdfITt?rADtv&I4Omb;OUX7nVsZ-VkZC%cg+1YW&sM`W4
zSAGB@o}_shl!vuWwx`)M)6qBv??UZ#V40i0xYpH!N31CZl9W=$gckMG`x6I7dCr=H
z*&BRpFue(Cl%LfVr!VS3W-AV@rn7pj9&~549)l3uKT{4?K}Ei{Om6tCta{oisF3NC
zq$RvG{7%ZKt;mn+Kw0(SsmjQiTeYL}bN%zihwb0*npYh!UVF5A!*0kmqYI{yR=k<y
zUN&S8d#Y&&QHyQ!5lMROlvETk|L(gWo>eS`hZ^x8VNte8@jv@aobA$J52+~3&pIcV
zc+d6h3#(=i_Ox%Z@uVGDLGJm-xu3i~jO-ruOpWCZyQ_cU-0YEAHR73C$@|||3Z9`g
z>dL}|CORKeogOnLsU2Zp!^8WaZ9FHJ4z85hqZbjE38&M@C30EM>Q3IwWRHD1q;1Wg
zY%KdEUTYg08&g4$1kCIb0+kV<2wDO<EqypVvIIhX1i!0lriYRlCNGlKKq9Fe*bP8^
zomUVXMMDU8R9joS(`EVKMEOAltQ6c^etojelM~fnG{;W7fc#b}lhO^pW`@YGq~JmU
zqixNGAlMMTQs$jK=d@TNGd<k=lV$Ju+lU$lZzBF#_mRa5#y-{EsjcIlwN-(m#>%UD
z5{qjNi`Tw`GG<#|4WN0w79iJ3_q4k8FDN&3U+tdLsV0`Gko)QFh^Jy~Z0(LOb4+3-
z6IbP)hLP<TI<tu}WLq$YyQ&VtL5F7bY*&0&kDT^1Ac>>_F_FtlNJwDzklS<%hwe@$
zc{0=-(hFSy7L--@%b1j`%ndL%kKAP!92{Iok}L8`KQ0xj>SrT~x1Dt5>gtM(Oq$Mq
z@IcTiyX#S+)%JFl{KgSkSrqB!`7`7t3OwgHRaf(_b|3!}egoD0@sXaE3ojRd7MRXI
z);YPZ3Ib$Blo82iU8lM8$M;5evU~Dp%!_LMA$AKh6MhKCEe0&T;gdNj|8pP&mU)+f
zGcpd9)e2%Gekl?0-u;9e&dG_ydxt#ST~R0dK!o@Cq8mbOdg3h@TY5EDHJ6n_?&;i5
zMPKH4sHrFSz2~wzVSfDYj;9@{Z@k^oF(^)b0Ya@$&LVAttokjv(JIY0JagG4SI7d*
z@@MdKE3^g{L%fobl7n!4=T227so!`z7V^EILcY_@cN4HGci?fgRs-NLh)l>WFlwkb
zIkF#RdU9&6ZF@;z)nH+)9;u-d9zRZ%9RO|ew#WIZ{ytIx;xyIwQSp(Sec}paOwx4t
z=Ng<$G2B~`?16f$5H$|hwp#AU%omRzCuMJHb2fm+D_UyzD(6Y#gS^l=U14Qdx9~&v
zd(IVU&*8E`Y*rDF{#8^;aUh;W3gc_qQV6!v+Eo2%CDfG71JOrj4{g!sXI2hr5Y^8e
zgTBeY`YZPu7*_pb4W{SX`eR~RVwq1l@lG!UsY75*WhK{);<tl)V->r0k&-^_i`2pP
zOyb1`U#|`LyuExc_bIJTVJ}2I;us{I%RuWKHDLk9HXD+qx(=9812WVei_W1}<)^3&
zA1r52l^^DJM0Bol8mhZ%62XIl|Kye8^tAG*8+n&1r34MfnjFPc`2bEPOiyB45{D12
zxl8&KC&kg!_dRiN1cb%c_Ysx@)Fwp7^2x48u|f)f#kbf_KTgh*Z7VkyB@K}8Fp)Z>
zg(Srd36_p(YAz*$rzbgIfm#*@Rc{lE)+1M#vFuz~m~N*@$Mp!vnW)X>ZK$LOiE0Rh
zBeaRos^i!)sxQUSaP$#gkZ7ANLkAacxqrl5=#OmSnrvMNqOPz40APi6g7YygJy6Hx
zZ+bHKNbaXUK5$uu>0dbIT!26stbd{MKrYe@#F{mCe+60A0Ir&E=?1mLrlk0IZye26
ziMpwTMfpX%KvsGO5H%sL_RJPY%PQbRRLbR{ax?^y7S{ye;xSdpJOLgc0NVy;`UiN!
z?R3Cy$EP+m-n&3G2ElA?J#_H&pyf_$Kw?F#im_bY7WcV#sYwVaS>8vmk3q#&EqU_z
zQf(81M;p>tko-}fHHy_BM9p$!oLVEWd3A~AE1daGxs6sI)z#H~;l`{pDF$frEMa}P
zTCt|O{BqJO6~(fe0PxsH{KE<IOJhk#1vQ0<&vVBVk$E7o<L0~geiwN$DVK=v7s+kK
zHryQ?rP9=wN5U|Fnh9*(=L-slsp%HB8%pgBJmpz{1PQ8qm-Ix2nkq<YTvQ9+y_}wt
zqalXDZWI#>pyc(;HNF)@P9e(@Lx_Fvxt!bJe$Am@?KOSvhe6mDnkVPhd@`qd)wjc*
zJvWZ&p0n?Hu6`URmenE6IV5&mJ^4ummsKby?jM<t)Mk5WW0qY{V`c(&t5o#GW)PJt
zYfm|zB1tm`+01iuArChQjI5Y+?}YV%&5L5@Eaa1YMWSevFdLPiTB_pOtY<o`%`%8F
zp^W8i_)aO<Wj-a_umT}!!CNYH$cwCkzzLv_y7|c{f>n_GnP()w=+7R{@L**SMPP)t
zbjz16rc|6Zle$P|nAYqaBnTWwi-8g3zJpYy;s7rrU3T~?9_rJ0fcmztL&Eg!L9yIr
zgu>F|!85fH28z1A5a=croi7Dd=zn3OYyG7KSBUBzcj*TX7H!><H|OzsZZMxG!Z+L}
zypY31p&N~ilTIIrpQk2k^Zg;o=Lij{xMp>^7are7AS!tH8Y^5yo}4wPptY^|fz*Vr
zz;AKBoN{RB7~9}seXon<NmJj?S`}ZfiQ&$Dvf*Vt^2<LE<BZKu&hVfUVePrRK3GSe
z2Y%{ZL?_|UAQ6{EJt#NlMO)df4znbInXlLE-Mf7LpuqF!Js#)dr$|d8Ua1t?qQ3~S
z+Wd&mzhkwOnwa!X=K{Y<E~`5c`aP(-wsQh?RAJ0_rrJ=3M*4dR6$5QKaLVjSXlUq&
zOlMe-cGChs1i`h_h&CM;-vQ)#2xaZvQAtY5NA@g|*D`5AYP>492uOBBbt7;qkMLnS
zf}Nuf*FIv6_Jr6!0a>m#0ImOFkGWY$<@Vv*;Salrj%>?$;VuB+IZ|65z9Gr38AS5k
z!H*u3u^uJy9i$eX#<Lrb)}>X+9*C{%0+<&gA|f(8!vjdx{>pH(LxZKg|4w+()PxNN
zAepV6<`LrABjfO7;yMXHt%s)QjmPDQ@uTDD^Lt3WlAovLg+)XT8eprt$`x3N6+K~4
zS;}}ARCZmQ1P%E5QZsV+JA*mMOk+%Nw$)-Zoboe(TRoy~hey=C<69daAOGgk3@{_!
z=|i^FCu$DKS~hKWY;vst*IV&PjxgF|of)?TroZfy@j%V3)MGPP$Zaw@HV)Icm0fir
z=F6Pe$|o7z3EW}GnsPtwuHc%<zc~^Sq5srAs62b(LHY+SdK}xC&Acq@4Tyd>)L0>1
zs!@sm)&j&cOw%E+qxy0)0BedGiRD<J>(ngQ3}huSa`&VTjA#)0)>okHSIH+KS2|Z7
z!AGrjYQy`Kr@rh{^g*T9>DruxG@vbirYD~|PW||G^!544H8mn;96fgIc6E+ryTr!l
z)JqE6Q+$QCrsE%EiP0xs=_0cHH%2CJ?<@w2oy_ZMcy5PJn!lhRz{C~AchF|`ILluA
z@-t-@gsN3QC5*d$1XO2xCL7+`DutJ>MltH1yn3u!IETo%Cg3hjY*%I0%WI1Qkl^ni
z>)Rir@h&Om2GH!*SjFk{o6%Ut$=AyIpxmY&7=2V31fx8Fa>c3oMql~fgThsz-cPM~
z1Ix341s;?GP&mqTL|ZRWV_``+Fz=ghJ>Xtl;oyAVx!~$-UEt<i@Z_5Q1^jWik|X;(
zq2bP_$1we!;Hm4}V*c%=A;n;6_(@Si()~d^ND;bNsEVpU@kyO`VHZEBlGykoY<@UT
zAAQJ1xC0WO7E`O*@$8G|$zj6%KL83!pOo$lvunNl{leGgGiu?og-GCA(QoKYYr)aX
zLcf46oAppn-mTO_oC#=wFnbWiXA%56M?bIq_}E^c-6Y|+`Qyh8uh3*_^!w%SWZpI0
zR=9GAMEK)`pw$neh9Dx4qw|H%B`z`17dhI_h4LfsrMu&z4O?G%+o1km@wxu$*W3X&
zG59oYv?X~%B~VBLW@<<WU#pC~6cUETlx28U=_T4PTP%)7OJXhviRC~6K;+Hyxq;35
z;ZZrM2s$x2^ghztlkfb7^Fd)+P&;fhk_K58A+Dt9rsREZ@_iKXgGk_|(QO*$`PLCv
zD@I1ij}vS``|m#XqNR`+CIAB|aQ$%!2|n~Bq=O+w9x)|hZ?lQr0s#fN-xdk*lYj<&
zllk5x_89DboLmI8vLTqb3N-If1HftodE(2YQ7apMeti?z(9P;8beNZq*E_-y?9TP`
z+&DruHFQeKV-#Vm7#rWyCxEJMH7MAt&IH|Bo;Wy$j8zR8tF%P!r{1nP@Wsc~2a-oT
zpUf6ZYYgHq%7%;DFJ??d(&>7<p_AG<a2iiW3EE1+<&L9HH+YeYF!B*`|G@FZmN)Hh
zMIv8YONFB|ODyN>ADN)I4jfq=vlW$cX797o?mhu=LuD+?(|<0nxJ}h(1aci`RwPnI
zdZGhIj-De@t8IhggsNJnDRMfc9%?uv*6yC1RiV#vr#V6!73AgS7s!?b=x<%@T(DjJ
zWcM7fDNQ!WnmR9*A@w>r<WWaWvnNs)#)#t2`Nq(NFKAwwjV!_l0eO9#mj=<mxo$4K
zDf=3c2Ck3$E&VL)e?haHB4iyWo8H;s-(6l;$tAQ~?9NmdN7am6h4Fk9vb{kxHau;f
z34nOGA8`E=E#-rxHKX2<#~JM3R#|zW>J5n!tPdn2W<h^GlFE~rs8ksY-e_)L{jKlr
zbJ)QjASiyH<l4Ej#T)!db%P=;E3%w}M;326kFs+#ubhm8R37USywY1iTKX7FwQ`R#
z)zMpmrX@X3AG)Ziul|K+U2}rOIN@{6Zs?^iJ3LBE+)PB6Gm9WS>bLxH(}VR}u=ZsR
z@MDH(e!Y1h=782>%R-Pf01|uSQms)l1b9ppdu%?smoB_rMNGPG#PdxECtLe@@lpM5
z<aJT+5tCN{UaK-b@{bqIO8MOhk6T3yCJuNf+~0B+(iQ?AtBmOaH@?sP^!WpqkO2qD
zQu-Az597W66!1RB_=EHTM$~Ig>dr9rpWQ3Yo3p=JkM&Xd*D1J#n`*J<#Q^!Vy8}(H
zB(9}jn>LMn-T2R*AL!?(_8+m5e|veaK{w;yUh6#Ri~oAv_<{EFua}Y_ZN=YS|G5Bz
z@&AW4I6?;dx|}pzR3FEe!dR3^X{Hf^-F|+V+!b~IcW+}bHr@nrc{`iB6HFaZNY=+e
zt#`j_J;yALoF$U-o6Ryzb!2a4|IJ+BDsKP3HaX6{#+m<H6d3Z|doxQ2P?EB{6P#Hn
z7giw*i0tB=%dDM*y<pnMW_upSllrl|v0P^s9ZF>OTXsn@EQ<*h?Iw>CrF}34DP;V0
z+@l5bWeEia+W|}6YYjXoDm;5N#<D$|f)g?Z@QydslJWnn<~&;8-fJ4%|Lq~)FTt|>
z?R8oCXK1?r_FA%wS~dQ9?f)krx4*sCEdI|0{<8+C6#ioe|5*e0g*A&c012P!7Gy9!
zLG<E{PO-E%ZNOU=QwLt|H1(QwcXuZyLF@T|lu1eJIsWGTA=F$B1VHQZu~013Ht3u>
z#fMLqbx#3qScB9kOcOXa(nppKCudnB*m|;M@)!sIjLb(L;c+wv2NhGN{SA)Gic%)!
z9Eu^w=1yz4U1<D1ezt)U!Oz;YZIzP&r?ty8f7b7NdC_ztz|jv3Ou)$R&F6H#F&0o?
zJTYQlYVV>a{8ord-e$UqBaxJTJlRd;8zA4!-I5ydw`MaEr60kb6as>{?jDY=x@FV6
zoV#vVhXh9HF^P2_lqrJLrH4P7%*j!d+I*V0`s-3nFis~ku9iN}s23$id%_QjdKDmu
z6&+*A8k`T`YR0vrv$L~Jhbd>&h}Wv%%m`uB@0k4;q`8DFT<2C620qXry1jQ-(G@TE
zY9BjBYJ8s^>|4T`K0Bo7-S<q&wB9G6w>fL$u>WeAZA|FkAH3zX3o2RpRCDzM@g?oa
zx67CckjLksIRaoz=0s*3m+Ii$+3O~+#cO?r7>SVItZ>CNCd4YY#fT?S+VQB8g47kV
zqU?_09-y7&k|m^p!I&T%0odKLi)C7ymZS4x)0Z%MUx~GrvtF8CJoa{zA;D-o%YNki
zYIs>Tw;X6rf<AY3a(WNz1vt+hzWZ9_e78OP^s&nwA{WP1#W(}@UjN3~E;SHjuFxin
zzShGY5&p(~mZ}NV6B~jLEZ@w)0<8d;xbd|fy{W_!;@pE)+^R1mm;z($`IeJ~DxEnT
zE}OPyn54|KT;JY)Zp|qu4iIv<%gU8|)2-Q0|H%g_wGG$|Nf&-jLXYWbv!4XZPYN<A
zbN_$uT9glcHDlv%9dqw*5)q-{19K8M#%MagS@^V2m+#$BE(0t%7G?JB!_>aHDY%3+
z1>|d=y}kW+BAioIFTq)sCumMlf!t_FAwTO3$9@?;$%Fx*?93^zoozI={e|<e!3Scf
z7<US1wTpSk_{$Bw{)f(Nd#Cv4RbHs5eK0X<c_zoGqvV%zcDHnF?j$Tq5NkO53x>}b
zx<h^ah+!u}JkjhE|HE>Qk@N*~wxS-oD#K5||4SxP3Tk&jE$Sg+D2Y=)x(epw3_DFI
z+~A`3_RpS!T)ittNgVs#_j8xUbH1G6^Goe%Ep!qYjeq@f@3$pN<}A@Dx^b6*L62?Y
ze5L70LK9%Cx&n%rRtEf2%#f3GDY?uZ%38EsNMkxts_c&71K$dd=euAc)GVJEKyAH?
zA9F5MFG34i|0_lbnb<c{BMeHh!Y!Qrx%}XDI(H-V{{&Ot{Py{^|MmHOz)6nHcD%pe
zfgWgCD3Ftd0(L@`wl?@cn$ovB;+Xxt_rP?h#bK5FqnsQ$0H-b6=5TJ{$OSxy4YW%;
zoWkarbKaVOj)f8R&a}H@oPcHkA6v9$n|%x4gTL}%&yY5E#kUo*=B$teb?M_uJO*Ry
zG-#PGr&|M7Nr96-=jw*t+H#z831C$wcJI-ITTJL*3cm2N5Z{I0zZ(RBThm|w?2b@3
z%Ce)dF5`+<<}LDv$GV;En?t0M4QTY+*_dzTMf4_VE$14Lc9I${w&<atuYwvhpE}GB
zp28YB?We?6bKW%JCFCyFch7EMC5`a`)V0N)#|eXVbjWgIL^(DSRlO5>C_~HyqBkzf
zUEFBT`8V`XoY$8BWU!3?WbusuNox?rN1XG+$}V3TqZcEZQe6k-N>O<a+E7fP;?5;k
zaQX8Gfw$otgoQyx<ntfV9%MBlbSBRTM=QqNoMV_%9vldT!B}>0HglM@S3oTPx3P@f
zPVwX{hsH1qX78rpO{gL1we05XJuLIb7rYmx0TR^os1d?BAxev9K`_R9J-!2{;C)?4
zmPYIr;+NbpKujS0N$pR_aU+QtAT{yrbO4r3{&bPT_~}HwtZf)cu>wq$k5r4R1W5r1
zE!D5uptg`_+d3<dl=QF37jU>SOh;>_;S!jlwv__sIfG~@yv6Oit2pYluYYi*Kr-+V
zttq5@rlE8+#gk^rJyHwj#pr|c%uzyB6<KxV(26ck0TD;~W(b~FktFvO#Bj@r&dO6U
z&vU;&I?$q+l*_EYgmGG}0KpG(t_tLzrh(|$S8Mwv4OoD1+C0;4FtZ-QWqsD&T$&O=
z7&nbUBR$}E&78Djd%(X1T@Pl*JiUoY{fe)Axb}~&KfT=gQ}kJ45B+A`+A-r|ONsr4
z!=Y7Q>_JuB^!FMe(ftUKsb5l{r&c$>;EvIxK{N+!$pykL#8zWM1{TX2=>i)@O$-VB
zKw)FJ#q=9Lr%74X-^%;s$@KxfY_BIIRzuw6j83{PNXDd`!t-e`4fXVN0UrPCnoA;z
zT;$e?EtLl44P=`}2H>=hxlCK+*G@##f({uWJaGGjd%9Iq=nT4a?$nm+Vv=YjW_Vc-
z@xoNqNVlsVs}b+%UQUnpa_f9~c;M#GEtV%96Rk8gEJ5$Dt=l7ZTui#*J~!kMPtUCU
z?aWG*(g1KU_uu3<Ahnd82tleCdLE=r3UsJMQbZTq_kI*784b}UW@F({aHU)tY^5=E
zLA4V_QhAF(QCgD<#meE`V)XC&dUZy6C=lr+6wpL`FeKZ4L|9`r`jc-U%+nl9GLu91
zc5I<JqxP0Sj9sWTZm>ionc7uT6$YqQ$Qf_FF77)1RVZM($Hf@}+6M@(J?76ldbsOJ
z2Xv_`CWI8%hxuPLSiLAzcOMGM9IZKNmt9oL7H}EkGuCzz^=I=WcwY8euxmkJ+!|-?
z&2iq##d#$iw0!5h@_b<GUb|$}Ipny};qAEZ&^7-w91-E=^2t3_>i3U=xer#J7yRMC
zwQ1JNLcN6&vG2m}+MO%wOs&y3cW`ML3me|b`cr1OKes1|X49&mw0Oqf0J2~l(|%Tw
zQHVs)p^o~w%JXsT3O)8TcjyVuAgK07+ePd#=bNJ&jSaIsK)sWlb_CjE=eZ4G9K?9%
zlwrJ=(Z5sH{ecgK_V{}V<b9c*9`F!=?89BZ6T?Tui1|UI*y~Gr$i`d2Gx>u|ues#A
zcrlQq^EAtl27q}Wjo!xJ05bFvKzeGLbQ<Ll+UWi@lXx5WCI7uMvzLdoE;N-Td6))r
zQ4$UE$5`G0Dy;yF5)`e^u#VIbpy@WG4M--4o)NX<|MY;bgJ4Hye<r4u`Ctl7yBjgM
zJoxUip3-4ScgS&zh~dg9KiL53#oe)xPIHW5QS50Fw$7XAIN{-xrvl=vm3VP99t-!0
z(`>P8{LUK(#FA^Jslr25D9LA7!Aj-=C8pMF5RuGj4wioe_eZT99n<X-ytK4$a!u$S
zx9gh~RBLi2waoN*e%plCXE!mTj*{+ft!iM-?N!z}DW+{rhS#?kmTt*N9Gs}@pL?Qq
z2BRm7z&h#l<$>7VPDG?{X8>df<P51TJ~4l0GG(}IXR-WPV-R!UW9lCTQKXW3u3GW<
zD5+ylIn=Gs=RT@b*q>i@v8QGWSpML_JJhYg@6vlzeJv&aQ7K2bk_^lWVyZpjk8;xc
zsQcUQVzL=SNG)t7%@d_3t^jLF{kfZuKV1`RM;Z~Sri+?W)NcwSC_)u3pe{G4Tx7Qp
zT4LzbnemNa&Dv;YCFsJv)D#D@jHZPW?<WzFQFrbh2^X}CT`(Pr=1=jI^FeJDM#IAB
zB+}d+w0sO=z;7{mS6F~m{cw7dLxVI1)FoJ0qw=PSvPO}M7gKl{>8=*p@$vK(>aFO-
zKcMdA=LYqi2_L6x(})?QDMAevsfQEMKbHc3QiN3uAI+hL+P*v#0k5bzG>udZub`!0
z3ewP<1`E*7WIg6nw3kQrn-wTxz1TB7-wu@V`Zop8=mabOT4<PZeDjKf6}OQXi-tU3
zFNS$$Mm6qOmRE^=Am8S_nY+s3NP8oezkRzu#?|S_WPlWW1A)%A2+q8{HU-f5smz@y
zJkCb@?MPrwLE^FwYK5{(GdOhm?KxnUOyoLA7B?AzeolqH3C%}`;%%oe);e%Bjl!s<
z2lDDN=NSk;f|yf=W08HS3Md=QSFeT3R(EK$DK;x-(`c^2Mhr+=*LMet_bJiM0P3Uj
z<@u0#rba||$*1B;t`{1{_r#zi&>arekVNJ6Am4-DOgUcSa+E9Sz?SiAF(1=D8jdBe
z(34L)&H2sdqx1DpoXpMzDAa?lD@fAdJ4V!Yr)|qI<%j0rgED)usUZ?5d}1f*tvC~p
z>hciU=sZY7G(f!4I8WCP)!QrN37UO*&{z~ARtFDGVawz(Lo)bT-_w^1OD~my5(~B0
zqn?Zx*5oEg*&y6NCg|Iji)R6};rOu01DQk-L@syi^}3e5U}GFWOh^$%l|629#Ba{Y
zNol&Cb18?aDtx8vMbr9ek{?Y$Vy9lTgZ4b?eU&FW;KYULF@umQ&VbpCQbNE%s{SBV
z587xF99jTKt-xrfl1noXg;w^nYV&7;2O9x284H}|rQZG;u{8!TasrO(r^%(zqZl{P
zFl^>Hp124N-8bwlNF`S^-NQ_Q{M}g29Zk}@theoVF8VZioI<W)O?>wGF%K+M3Ur9i
zSrUaN>0`Yh?Tixjb7lKx)~U849+Z9L6q(BcPdq(wA%&GJVpc#8a%=6ZYlJjL?S+_X
zAdM3%8m@Gk;9pEbseL;U<Yk9HZM%eaAv@WA^o^Bg=0ol++ye)2*riwKp2ZT?z_X`7
z#ijYWIl!N^t*i-tlIzu*ED>ZlW}rjTDAM-TPpbyf?Ci?^@@0{9=8EzpreBnFa`2Un
z%hFD#eWH(lTg-j$eB_Y_@8-;Dho1KFn7bmx7{08KEhlf=w7^|d_Nxc3Ga{uO&zQ;3
zL}z5@MZHIZdzCGoSEvLlX25ig8FY7=p;%PYh3d(z-Y96DH|CuRn%^^U9o$Px6T1Bk
zTmO6b!K6bN%g`f^QdISYn5JtIpH%e(RJqTX@=*B&x|GvG#krUjOdRD>i&>!;{LN$U
zPy3M(wSt;gabGk)xdjS{ZSJbgiMmbiT#&L3*vVRfwB;GgKo}V61n2bhm{E!LwS{~d
zH4<*;^Kp~Mp%7B*J2$yanzLF6H}?{Y&t{)IFLbv#mD=VYY<`x0bDiqn0*6@v-E-XR
zSWN#zsNUnQ9x}LLl8Sk;zASUQj0y|eTys;{-nM>HySA5O6lYxZ9ouSjKiwKV<nhyc
zjQ$E6IF<zgc&(`%UF~dkP(N?hjLq-abQYhpY~9MK2eV(EisYz44X$Hcq7H2Py<0d#
z&TP-p(dDT}B;vAdtanV*L;6(aD6{1uv|_PMI;aCqTBd2&>_|taP+0yK(X^RW4=e_B
zp`S}lPeDo)QtiSqN6Gsu!VYZ7bdR5Xm;^0=cO~!D(IFG&-+=60qMwb8*p`%RIrRAr
z%8R2t?VXBjr!qGcWxTJTt8Z*o{7|zOR|jTzt}${;zaR%t+r5B;&hl<a`!p4x8vAit
zO8J|;Zz6jyd5^)Ayp-h?8IfX`1Yh;jBwwZRAI2oD4PlOxNN>eSi<>P<c3&|*>};bV
zaF9gH&$fr+nJqsXyOd+<{tX)s#Hycox5wSsLy_~cis}|=w$Xt!_HX-V*Gj{f*6J=V
z8Y9u7nOE<IygLB>LdJ6JzKj-s&MajY05Tu=TOO(gWnz%=uz;yDW+};%&~`kTJ(;|T
zMI(ZJJNt2W)SvO3HE#j$GE&j`6=~1MvjKB_otY248V@_}6Eaaww7bd=ITuom1GC}D
z<G(FH6=n*rBuG8Xq5eEcNocqR)KTKp0)GiEa9gwYcEMKcgl4cM<y{B~Yx-pH3;mxE
zQ98w~keb;v+1D15o1?gIft<XWDT5Q^+ef+wYRoBy<KYJs<&Q~voS&g4Fd%h>OhBwG
z%<h*^OjU3NfXNfUgYllE|4l*Wtes~9ccMeR`Zkz<TvNH93eKoMHOUHBq@AY6I-s3u
z2r*RD&!pMm4VMBgo@lU4>Ku)?Bfx)PnDk?40ZMo#uNp{GKqxb>jv(8_GP;cPHAvy^
z@<{PWr8o^+Hie(CKq;8=a^X&zCt?WgQvezR&heGvKV;EHdP!)BB>TkSb!fC9X_@=Q
zw(X-QIS{yx2BZ3V$Q{8qd_QmD+e<neWUK^X;$7UFTKyE3wbwWIrcJTpMg}EaQ1Co%
z8l%#a{zCwnRHsF>KxRnrenHVN{yP8)G*S1XamcCpHRtwp=l_z8w5OHkAVTyQ{c(2g
z-G?JwqEBX434U9fqd~Mbmw#)A4+_Y~L<*jOtS=9)8=@U%bo&D@8Yzk@nwN$}R7@aL
z`da5trx(DekuGkc9(NK%`IWIt4Q4bP1~_0q)rPcBAA0hC(U?sKm)!*}apX5EIIKH>
zH=8^Cfc%q0EwT6sT&o%qoJ>?pM(8Y*<S}d}KoyqphIO!)GK@0mHs84|9(zVm9;l#n
z9G+R?h9ZyFG%hY~zsp-JxA{&v+Go?Rvypc2o4+$7f46ZDM><X|120;pMQ@9q3LPD%
zgbw-IN}AI^LndPt0@ry{ZvZSjmZrx-e3+?XbP6&bq*p>T^_NbY1V0NEVIkV%S3sb&
zyT%~|!$wK=yxL=GkQz#P{hm}!T6+<g@DU@848e%#225jR+Z5TuiX_bCU=F9REcn2G
zvhqb3xdA1k2P+a+=xzG8-|IQyjNxO1HhKchNxalXBgwyCD7^;M$hD79{Tjg%13=;w
zQwl*nBydW_y>+SPRiyb@Muml*`$W49nvM!_Gkl~}vcURRpuRGN+6W^&1oIGeXzyH6
zRB=!dkWQRoY)I8|`ZrHNTS|(^mE=1}??*YsUTA+FAn7Dk)tw!dCx{WF9w}1x`{`>6
zkG1@aLF-$Pw=G4hrw8YsLCy9s#{}8CA6eJG|DyoP%H#wp9;!~d?73u3XZF)PvjUV6
zyT7@@qX=YyeNkv^lqKcf<Vskn(lMHRPEyPBV>GJ4rp`E|x()v+m>G4nc@a2Y$gko@
zqjLeKAMHJZt1ADB2z;cT?uaz^DKt24!NYtvx|+8va2$#LNH`XGKMwJ5Is3s1K*$WG
zQ(IBOFiekNq(cbRE%nm`K4Yu1`1j3H6!*WyAssR@h7ZjXqHzY&a759nL5(x!7;y69
z6{ZoOq3$`%6dE&b>V2!)N)OA3oAd4b9p?BZ#zu8&1VA^_tMBHm{+Z;6>@oI?K@Kw9
z!E%oo2}*TIbw>5}j^;~kU$FbDv8^P9R8BqD#a@+4?N>-Z<@0docy0SujaA*j@nbA&
zdUVf<7rw?9BSmX6{YctRjUR+`e|pXN2b5XS4+nNxm8&BrZajlDj3(p4=Ls{9a6S{a
zb~ZFD^V~q1rv>$xbg`HO^eEQve{(QV{@>2uMsi(@1s|}7yb#Y>lVp!n>b-w*u61PI
zH<RB^arW|qam*E<DlY|tSPCiTKwo2Ik@{{&%AdcjQo*PgYGtn=l!6WvPOBd>h+>dI
ztsXO-9yIO9i-5tvb;8V2&KIot{@Vljy@6m$slpYB=^*54g&5Ozs`vivP3-&kxjRg0
z<lkPGCr=|9|Lt{T1!t@OdZ)ybuKU+ZNta_af4#Ahi?afMy<uE{%KdMzsLh<s`Rkoq
zb7(F8dR>0@pPQInjsL8~zpdi`!b<E;-7WbXSjXNVp^kI^gT42R>gwFOM~R7v(HJ#B
zQBdpxDj)(XB7y}F#731KL8S=@3P@L@Mx`l8Zz{cu(h;!G1O%i?6QziNfOMqZxi-mr
z&dGZ|+&k{LW85+R;Y*_OD|_$fS<hN?&NbIJEUYhaPSE*kFp7y3`H8PK`Bzccyx%{L
z4wM#zYW|UJx6k?;rw|&m%od6J86I`M4BB_lj&++I1j2ghHALueV?IuC_4p6-cldUZ
zq!dN!(r6J70_A{v>3pxBqRfv`X-+MK!je6aI^0RBtFtIXpA#n~4zX3Y#eQPB{P-R(
zn;2=45xIwVtwX?lO4F&R>w*c?tG;ma{B_mz-+#8tUw7mGzjuOrL29G<VMYA#p_Jqv
zxf3Z()^bH1IcLdb<hGd5{eyu(Z<HMLsBp<fTBWmSVdVOOyn#*H-*NwJz4JI*+erHn
zlSOTZkIrwGDLt9p=ERb>w(onE`;<uv8IlJ!NS#MTBk11RQwsEQo%1~Xc_ekz2mV}?
zf;y=HaGiYU5@9m7bfs1TW&jgRv2;=8yAtGxYHbNXt_zghlSv!3d?fxcUugS)<o&f1
z5YM+zLQG8_Ob&nzL`QFZiZKV34b*T&Y!5B?#6Q${t#`6q-;nx);r^p)GgViw=-lO>
zvrzLu{13&M->`5?x}4o!je<?SSK~YYw57I(Bz{~DDLYYBN-40?%5eo)BHm+7BMOqO
zLe-um88_@82eiuNm|+o+vdT)tNF(=1>l^eIlZj`{xz)mkice7-OD;drIt*YZpX^tt
zG-ap=)VujO7j;ml5zgZya^Wm}7G-M4rUl7lhM91M9*x^#7w_MVG{|25q5jzGn;i2$
z43>$zQ9J7!pi0ZZ|IpAqnkJGDdDT)HMn}0bVY6fkKy9)#v53m!NnrbK8)OWtM}akR
z0;4KkUO<@cQR7#cQ9Q}N-QR8ayo$@L@=~Epx0~9w<LB_lon{9f2;H!G^pu0;vV4{P
zCWRBJ^epbpd}ohJFOTz=)6_3ZS&e58$GJ!Qep<Qfk+LzrT2f?v%GzJeDj#WFd?Q(H
zLK^Ze>kTfIKtNia4wgFM$!yUAB>A)y&Q&Lwscb~`SnPT+`O5q04?=q^eEFr)Mic~F
zZk+2yOT}BH^bBdIrF|_2kht0B#DaXxU@>YO1(4iIq9C-NRHwOEyvlqqRM>5R>WoLY
zCH&M~j~d_x`LApx4umX|<N0N0+e3ls-`ksjma%(8U6wLJWJ8aXwqhNYe1tTT7{jJ%
z_{Qi!sGDr^$G|!hc2^J#A^^a1Ha8TYYps?oTQM0d6$59X>O5MPO=vTirB?jnElT=<
z$mpO9rF4_@9}J9D;D7@34ce1w?3haP%OBY9W|Xq6*O-V}VRm3QnWzUK0w)4SdN|}r
z12|BNE0^!UH)RE@5O%0hXr*~YM#!?C?#)mBP1ZAQtw+A{A}=e938gBEJ&f{2R5Z#x
z*(8ezb)zmJ^z*6SN#S0N`yewq(c-0U64h~6h>wy?P%ldHJB@0#nxuWW;Ui=qPeY2x
zu<|X^c5-y3<wetYTHw!TPj<O9gGo7vrY<apKlH06PgecqUF)@-&-X-}&0ScX4ydWs
z&nTWpZHAhsTZv%}OycePo+un1rP)WZ5UY$ea;`#A?!r!?nm?D7&i19{+?Wi&w->`2
zH#eY{q2n#E+66Ub*e|V)un$>K!`Z$MHvsnS=fEI^7meWSryn@E&;^<+PPFP24Wk8+
zbat)eT8Neyy**YQM|RwTJSmU7)0l~Udtu+Wdj)^_rD~PGc2ms=jC{6C=gjoY9L~%^
zWyFm6fQ^d|T;Q0U2?&*ZRLmNTQzd~S7*@|vLQVBsTz%oecMBzqnaeryaxE9{9$cLw
z{@9hfu-mX?_&3BHmayE@o9~~<06G)h<Up|{DHs?oCK~YS`wKa=gn!US^Cc<jlv42a
zyP<bFaczn4IqL0l;>+MZ(}2LUnyy=`hp-ki5S7aD{2*x@aFf|iJ4<|xOt2GpksR!|
zWM^y-h+{IJQOq<&>$o$``oNM{M%F#&9>3c329V)f{SX%&z&%d%?kH1Sd{9+Vb&N{S
zq5T=HA_h%Zw)IfmJFH{&oFHteZzA#JWPQ^NcD03G55pS#1%C`oJEiH8&bgHW|2-c$
z&wLcsY^2=$4D5skx)D{wFh~UTqvKqydjz;Geq#qUNTFxSY2l8yYe4(021=p8_*{_k
zkR{>aDSy<xWHxA4bwXMaR9Fa6RQUR2r%F>EXkrRkS%4`_Ql6^H(-T7tl!gbzI6dRD
zr65Dlf#I1GHM5a{4D}UKpK&s~)xoCx6Yd+BhU53of7ivo+(M(J6gVvA@L6TmD1i1q
zR@_33IPC+P%p3|V&FWrJbf?XK*F4tW+6s`G|59cr`U1C@p=yx^@DYN(Kc;?1RA-PU
z3Jc`nOo$rsLlzY?X!~vf{79CEc}VA$VMdYG#ok5{@^hsV2O^5cQa#x?I0!6brkOG$
zguxcwMxB~V5Ir^DK@pILHt@yp1T?;)eZs6tukogDeO?Lrv^9H1;1VL`z9HE#vvCr>
zva(<I&<RNp&Y(#InlaHz;3fYEP@6PKQ0A`C<GCw%R%NwFh@F6a!$4kR`bh0`{pz)P
zv4>rRP1p~V(9uM0Wr~>zxWs<IRRl9UPN@WMwj2tWi+%6J5~e9}CDL3zr3Zk=+!=eG
zzHpzat4_ywcZsIk4_94n*I&3#q<whFLQxZ}qL5_);v%}1=J_RUPSr-Ali%NL(VpQL
zNY1%;Pob3i?Dj2x%!UW|yKV2WB4yXDYkdq7L&Gmx5Z(WBE6VU8{=)%e2i_ilI+Lr}
zamL;pGEP1FEKt^mXc}0^UW5ftGVgo+)&ZVX?r7N-irDLbp@YgnqpN57SBqq6%Yb?)
za6Qs_0V+i$R|71|no*smcGUD2N_7u)0G&C4W<_rWM-Jnp6Xsx{xNyiGYtXWXagh54
z!YnGFg$c!djYwR5p+w*nmg(aZ_0?ucDKu~qcV5%nW+7gujV~A6R_&mP!+0Wkl@act
zxd{Q0C0lx_$r&_8Mm>#vx;OF^7u2yQoZz)Ogu+2~6|650;5)n#Ba1LBB&i1tr%wj#
z4G=u?_gW!eZv%o}{MAAX$;T_uG`@pI^-$?}7{#Y=!W=5qQu``zc_$751s|WUsNyT9
zwTLt`tU9Yn6JYX3KOmpt8(HkZ5Nx{S%o0~G+8OF0Ur)mT>U9r`X}b&7$gRB~_4T2h
zz`7he0U3>nTCJ{u`rr-bQf3O{BD$KpK|@x$P5}C}4IIH%H06Tl>9NiWzc1Wzh(dcP
z8s`lPFpw+l>4YJ-^(Z@mh|3kw&(1%x*ytSiUe%e`U~i%#brly(f7oMfKioEAJ~1!(
z{kb;%WEDd*g<1iR4ed*IF5JjvJTE;`?aK_W;dAr+{T<E&d6PmlYUkYxSJZAb&#!p3
zB2Vtry#r!t589d-_Fq?SxU(|i?Sbzw|4c4itvdkodp448$b7o28coWBy{$;A%Ek*y
zjrhsnphibocgv0ViJ<ljXRxpxJGxt*1R+Ut$<M-tabYPHb&Ewg?NV!{1OS5&{A!Kf
zp7uE6PLzYOfA_MR<kuFEFRwtMJGqq*u?r^a6;8yUpTF9KlEY-<>Xb;zBTl17R`_<p
z%|rAy<l1iFWMXuB%eg@G1&WN4jp$i>W5lMz)DhTPi?Lm{O3NB$C8Kg(e0c0i5Di1>
zcYuac3mNI!A9XBz99X16{u7jjn*fPLDfyR)&)CWOi++A|;LqpyF`*oLJu&U8j(e49
z>B*aaJ<OKbhw0(iD3tG02V1z9RozHkgNCOPT(D&M>d^ZhGb9R*63E}xW8t#yH||;`
zyjeVbIXS%O%yv(;^Q|S_F-8g|!9k7L<Nfhjk#qI5i-e`Oc&0@q_{g{q-15J1$HU8p
zFM3%(M&#V$6@!|O+nHyV!aTc5Ui18+T)CIGd^hj}+nhg8m|07B&HaAEUliG~3@O+G
zbaf(Nok5cbUUJ`fSO-go8r`LF-fVH3fvXcUmr2W>#>vtsGV;|6u&7amFw8U;kBEbg
zS9(;`Nl;M$JMtECFn?>b)q-HqE%9j&&9ewa1-yIdYrD5Dx+w`grjcUAa4`MSXip%N
zw=6R_V>)&e`dw;=Z2@}aLc^gb1^n>hdOxWATymo&3aveHwZqscq{!P6`fHU|0yI59
z4a+BCUUl#wnS>Kse+lnT)9EPtAi%8^61t<-c7xPq9U0<!8}p2JC)A!$p&?6#spPaX
z*B}SD^&5QR<_2^a7ZF3dZ~UY#{Fj(MY%tbFjl?~IH;Jc;AUfZeOz6zOgSXpF0zapG
z<VdEtj63;nFQsNumx3X>S1=?|8F2eeJ0!bBsw7m!-v?==?u%<b3UeMyw{E|t{H5nF
zEdZQmYjC>uorKG%Vf9B+RgmMc#tfM9rI<o+B(8qzQUv?1^?LXCm5*inh}KS*Jl338
zf7@}2|9dkJ)u;|<i&K;|;~+O2U4`?K6o@pv8H5}+=^BX9A@j!_csmBXFTAtXz_H_u
zAf6H#b||E*LLNgB`r61Wjl#^RV<1-}mIZL6+n+Uj9@VBF_MmZK{AT{ro#qww3JXfz
z0t5tc0uAJquq*KHdU!E@#-+1fduzLw;BDW=dkSJF-z{X3kSRQquav&WcXP`hnQf<S
zG(MxP0dwX+4-NEEn*0jVWZ#hhbS-Qk*;G+KJ>vS-?txNyqu7az?yfhV0X!c`Ur>yS
z_|9+uL`a)fU@Z^emWPRj@7>!?p-6g_h2nN%y>zwG&Yv<z^LuzwaA$Ij^h!FVdE6>`
zq1LJqm4c|ip2tm-qG^m+{NBE;!ca>l<wR_f^SN_D)VFsrNgp<`X<j^$Jkbk7yUoVc
zAAjCQ1c*!DLxrXRSbp#tYx|<7+2ah7qr*f|8%|_kB=7v1GPL{q4tA*3%C0rZm21i^
zK^}C%SRIgVTxF(sdq`EVoCTzlHLosNf-1ac2}ij)L$@u%b~2?t!5M`s+MV)Vvw4J0
zBOtdsRl8XC<R#}=)=qEWnI6uGzv-Xh<e$G>uv_boS}mel$t6G}D+qbeMsi0jogyz7
zrTR};=0&#JY{$Y4=i!71n-;oU&+W~nAQ=hciaUej8}2rA{32~WQPQdjjY_F8PwGlu
z8QJwFU4m!bL+x6^*S3X*1vy2})!U?3g6Er>Sj-kv`_}Ne%sRc2Gj|2Ue-T``<!n(D
zXXE05Ug14Va2YrqoeVG}7bZ$8$>rU)gj)Q?r#g>qsV#8f@E%>Yk34kAp0%P^oGKzf
z-;o;Hm@Z#rqdS5Pt~gDJWdY!b;NbxH&J=(P=0rJUT1uH|7Ac`%@5_$7t*qXW@=~J5
zz;PED6X@<5GPCbNt3KJ2<21NA_90%Lh7wbfUcBoDrOqf%HPZLOx^)cpA{Xp`v}l2W
z*s)Ag^t^64DNe2*G%TfPgbsUR{I3qtCL&E`MK28iX(i^Duyv{~i4~$#hUO0@b<or}
zV!YVhI!RCmxqd!KOd4H5{q_-JmXK%(z%F@n158O)>}))Lpnj0>1Y~}$6z1T?Lgr31
zGA|;_=mXhO&~Y=(A3m>ZH!Y()IoOy+H)X@2nLb&|*fR!WLqcOsK^x39NM+O~L~$I-
z43jAJ(XwyDc@W*^^z}kdLS3@>L!tPYJe1%jr6<`am9uO{FuRa_xP#1yAx6<dOUimx
zw?G`t0WcZhY48&=U}gf)NNkZ|6@>wlE)TWy{ZVj^Gtq{KNkoljOTp8<56yC20+L3m
zt`2hP$sH^8;rOaE%me|Ugc+9)_1bdnJ_rIOSSU<&!|EfK>;-qVgUG_3a9UN!a4Ahc
zrF#i*snq^la<wus`LP9o^d0j2k$&eR8k_e<JW$JA{^<Gl9962~!~LpymEcB5+}duF
zA-*p5D4seEn*bG7H2eg9*p{e#1M&wj96fyCP3sor*qg{6wnhwg6%iKDGeR{k&SR}d
z4rFRVv%Nigh=>A@oxH}r5y9GErituD$cbbZIUFOgr2~_!B^cic5_2QLny#Hd8QnrA
zwl4XC=szPymC7o$lyfTxJ;SZ#`Wc=xNgYqIo|H(HXH2SHL$_u?!;Dh_h{}AbU=Zcq
zaKP>6%b6Uw1k<$;m=o3yLXwfeo)tc>Tx+)eY!~0zn&|;U%}$5DJ6(8kGO7u^PFMWi
zB0*KC>3Hq0a>vr?NU4u1^Y9f1^8C#H*TaXillG$&8XXB|U?*mhsQ}0?gNQ)ojS(5J
za5f+RJ-Oc9N+Ec*;@D|EF+>$dSS%CrfBk#NA7Y;N5JG-CJ#7m4pa0`&V}7GRL@N{N
zhV6bCLzEYRrM1^^8Yu~RA_C;ovq9&LHP1{n2TT}5ro7d0Dhh5!Jv|1zM3ZZ(sz5K$
zDY;K$81;JrAcMP=*4JG_NaosJDnNcU=Iqi3bEP<pSz*8*t&z|x{sGcvhYd3F>11>z
zk{X<FQQ!%9fKc9%sw((V%j;OW(1R}095rk~At9I9hcOt@=09W$|GfLteL@{8gC<a$
zwTpR?*aadmt)6VTnIIx2H#g8w$D{&rq#Oo0^F+{mLlf@fESrn@b5>FjS#Nn*Ml;{V
zM@Fg8=he`4*~Gr++T&Z*lDN-g9oM#rZzu)KwxGf!@zn~!7`K2E**?+ngLyZ{i&u0D
z5(MN{I8(p+1bgj_&M!L`3hKo!AZS_N7HKOpw>Rs;2x6jEyIV2|fyoGHU%_41r~c!i
z%QWne!%x)9FYpTS%rJk|y`EjAu~5bcRiil7hR$^;Idp7XqL!^Zp;kMSf3b%@#hmhc
zlmB>J=iD}G>%v~|h-w+ki%F?pT0poVPdC&MmZcp#6v-)Pi{muP66&mX1`D60>~=H$
zD9Sit14qO#kwBGCyVNpo)&X=Sg@(C__&WLdTOnjL(CS!LpP^kId$&`dO&BFtOd9TK
z%&M0FBaEs?2zMn3L;aNRe3iP99Msv?ZXM(a3{9<aCcowt-iKW|z7#IurgO3<5m-jt
zz}(!78kY`XNSHnQO?HQc#v{X0OgJv!(P>p!ZzZ`Q%fhLKO4@z(nxj*ukOb_j=iMQq
zdv^G)U=f?bD>kFjWbqs$$Ag_3Gfpy%LL~u-e4)0EH7S^Qkx!7pNhA_mRUdC4*(3F{
zRdUYYT+2sSSj;){vTPw~QBewsdFbTBHZF7gBD|(Ad~eiTwqUg{Ljk284G~D6>~mJe
zr4?;VaMRZ?ThA3I1{GG}3PBISqh&U(b^=-X@(WzKv$`F1Ge+3d)7zYNHtlRzoCPN5
zqtq$}V&PtC*1c)4lXI0uAmFGS9lEs+xi-0r+!RbdyW-eKpNiP#S~6x~Y;W)sT?o~u
z{Dll#E8hvMSFxB<oxvE2+i08EL4AmC5s*TVSW$E%m5C}MtCzV}d|u7|f;ZSmT3-j@
z%M$`j_zfOsQ5Zl&&%@MF>Mfkm3eiO59>*qW@!HgjZrrALkHDdwzdgIWxzFri?w{DN
zu*uF!O=SZ2g>MKhy$QR$6{P<(+CyTddV8pcWim}c*zt4=o-H|K`x*2@%p`}ZRXS>v
z${=n3(ah{>NnkP=F1{c$!kfI0aUL&4*uL{vD>BnCuAvlLU-55ouimVuYzo_GH@4BQ
zI~Vd}8({=%ahfCD62uSz(peCwMiOVjYA*o=Z^3Govweqq==C@ox+jKym~<u3axf#j
z%IiNj+4Pl_aY>#Bv{t`KYmqvtMmk7fKS-8>u&OGJwgQH!$ei98Zsf?r{YSuwlD6=G
zuIB?L&>UQOrD*i(UZ|e?2AR@=vlGY4>UW?I=-x0P6?P^Es&Swo#BfDV{$Ps!J5rGl
z^!*v-DyakV$z5=f*wFE0M?3E$n)WqRXOY<r$+L2mR5X{ume=VRoP{YR4Va(S+3TDX
zk7t=7S%QZcww27<G~72tX+$eTW)@c~W|XCDGIOUL#!M<XGl5<Ms&69AGPND2Eb8sk
zSqGudZcUn?<|QgXv|2lnT-MuvW%Y-2(O)I!YvE9kr(ecR4grrGn1><LF&K@5<o^2d
zqamp#D;<yk%{8!U%G~~O{JLJ0hu~Rd1|svcOSIIO7@TOmCZ9gd=AUM83D`&B$iyV9
zeG*0%&D#QlFcdu&^@C(7wYi(j=;|ziW9AXiu-v!@Y|CtAvNbgU+{Eg_H?ry?ZNp1k
zwg4AP#*SN!ZEl|bnee!O-(KEV%?tk)_>@T4X})(yF#ksm;N5i_f&dyUCv$SM+$Gh^
z(w6aSt)yAYj4%+~6nXzUZL3i1%b9T?K|SJxV-=${f#UW-Yh;R5S5azCMiNpM#mp?e
z1<3^~FUa8+V~R#{;ML!c0<T&%XPBq5HTi;d$tC;3;YG(~C3W5W^Op(2nhdJrroL73
z!=LK}&rBiU$6NRlA%(0`IUW<06gwjxxC@@GT}&apcd9x}r_E_e{8ZTf(vxcJ`x{wB
z&!*s(IZs#tY-jKBhh(WpI4q<2HO;`w&rG9!bmIx>m1YNv5=pT?w!7wzRpr~l?k7e!
zt?jQh?hT&qa#uUwbSGqKP@2KXe+8>2cy^n#<&N>D^$KBSN9lh?Qev)f^iPET+39GT
z6hMmsZknTad!Xp#|8#P|zRlcT6jGs98qPsJp3s*sr~H?k!IXNvhH$<z;4aNxzAH8U
zvbBf22hF|RS6RMYOQgyrn02;7$l`(%Xa{0|m$_9qJf+g4rk3t@MI<=>bx>K|*mHZn
zowrVNR2ViywaE+${tMIJUiV9bhfDOi!|ogSGCx0NPxkJ_M2toQuF4Jw5>NC58g?Df
ze7Et@+q^M9?$d}elkL9>0+f=Sz5hJ`7<o{QPTRl-Emqk$8{Ne`R*OEPxu8fmZt-nT
z{#Pq5)E+lyu)G(^;woY0Q~XO?Hf83o6!gtzal1(hG?$UM^xrQ=-DU1DetTeiUV2ue
z7Vj<<7ZK^0AFb_8P$&M+dhKm<uJ+&$uRUq_e2L5Gxc2<oX?c`f@rea;B7f3tb3CHD
zRQKj3z`LCLj6DK-!bn>3YO8=U&8u{`6PdmZq#}7ZdzVo(Af$uVGdffpPFTyiZ=kvK
zXY9T^uSt%SMXhc>K#D*sndbTsF(O4o0qRvH&LbF=Fv&iQ1HB7{b9<S#!`CzUnU41|
z&=;qeX^)|BD`wVfsZ)(M%e_#q1zD<E+4^#tP2<^~<uy5zA$S7f3MWgwV7Y{D9GUS-
z0@GCc6{x_h+8JPKs=y$AHN0swWc3txu(x6gCCYCNW8yZ~cEF01T-xL6S3)dYG>$Sc
zVC3x$FWmY;l`ewcF0yW+%&AN%N@M2m^|V*<0tIsTI9(^+ltY=0z>b9LQ{PB&NaDE~
z3a4b?WKO(W+mTQ;hzDWMb7BF7@>6$eJHlE270ExMmXXc&(bA2YD=qY{+mT0O;_$5i
zfy{hP!GAlHqxERwU1>o!@GzckqK1<T;_UTqp%0_Iahe(x1{1?Vab(g>&8JD`G>*J|
z1k`57@bt0@Y=x6wFHjd1#Fnq4&L6KVyI*cL^dxvHcA$e8BoHJ<Lh<sCpc4yg8ig*j
z#-N#+Wk53y*BTz_?R^X_b}qHlB~3`Xu%q#VYzBL>$rrn?pH-G_1B%-i%VzReCCm6h
zLsR5kW;Edh%KPItKP;7efP|?nJY2w|ceDVFH;a)~y^e`9-+@Z(-Z0Hm%eTr1K*?F8
zxzQh0A!%OfTRGISH|WJ?$rl5z2n+G~s&mcAkx#aLTw6Bx$XOS$vovKeX@YVl_MXF|
z4om%?`sYh&Y1@W*)k+uk9KNOYWl@{{!attw&6K~%A!2YU5`|^yN!#{+9iREOQ~cKv
zWN@r1|16-u_isUF2W%Dt=I5bqAvfisVQN1@p`_N6Xe3^53CvFbMRsgY0<@oKs`Y+(
z)A3R7s=J12pZSu|O%xFeH})*<){_(QlCXUyT~GVPcbKzeSf%}-_0ydLKi`4{aM3U-
zuNbM;&2uRUq%yd=7nd&|IFdz~AyqDqMryKc>oaIjTT=ftC_}xi=aj6tLdGwG<yFbF
zM^^I&o%jTH*6g5?i~25!11C(IR2cQTD)AaCAcw5x6SEnL?C<8jx1UM}{$QV|Ql{^Q
zJVce%OTul)QNGAK4d}56<DYjzov72IRU6&V!#Rf2CtzfZ*%S+6SP_qOI=RmhHygY~
zM0c~9ho-KpuVC38!Jcd^%18C0e{jkCg4D;Te(?5`b=Z1lN3bdL;OsP?kbk&rtF*fJ
z)9D`;ikw|{TYHh0$Fdu0=Mk8$E4j`drHZhaep>5@w+8Hohg}r*XL(<^r@IbGBd6oo
z%hiuLtw7XAG+ucCx`IGeL$@9*{GxFLk|!y|ntZg_>54}e8ryN<hnS2c%O(8l9+LOv
ziULG$?XQF2_~hr0caqzuaJG{Ij#Z!>X29`O;>F=^K;x0B_0NT^Y0^B;m`?WD0{bzT
ziTl_Mt=G|p8t_LtSm>^#xkvI!t<5xl8Al2S`OC=JsysAKsoT`<dM@ugjH$yW{Z$)i
zpAaudbQmKBL0+tb9YgcLh~1}7PePYbrt&^S%h#}cTl}Kyw-Ljv1O|aAC}&Q=`!n){
zZF@8Vs2RExL(k#lDW<6$#W8`@qTSA34_?(d@59X--iTs<J}?8G*$3t%pEm8y4jdPc
zY-Y@D<|cZN)9US!q`K0sCQ~+AP-M(W)9Oe(K|peKP$l8kj&~BlF-p5{u;xX5e^~S*
z`2Dbzw#X+x4ym;_3GEmdf3f-c)4lsmv<(>|(FHqDrvMnvwyLWaxb(KhnwMZl`wDxh
zoj=%`HO)3ruYJ4C*;jbPhsXO*9`VlqJsz>6sum;H6iUgs-{->WC3i_xWDQWcGH4Lb
zXA_yM?a3zO7XqFNX7pGnoSr=kOjGL4Qm<bHr#O<|Q2d5czA!rdk~&E5-2N855WGE!
z$z=Vqks2BwBUz6v;G>yQcydK))Ej8%7S~-izQ}RxK#>_hjx+fT${I>%+TP!(f)#Pe
z;cL-H|I91omfq%*)EU}u5t|5AMjI&UV?To&JPnPN(nvF67>_QwhhH#E4X0Yz(D9fB
zsOF9xHLC{ei_zES*WZXxzN@>hL_@b9!inPaI1MkfkA(MhBbvzAzQc{E3>qa5W3fK`
zP1HyLO$g!p)SZFm{7P3-tW1Nl2Bh}evuD0;zMi&3Fxk!5;x#7LMWHB52}I0a-(lH>
zS#(X?vp+%skn#3FY!x&x_Y#FccYH-Nvp*@_Tde^B9ziRyN_(;k<org#VL^0IDzN=G
zo@%~*&#_-bbS-j^d`vTl1tlo69p%`)?n}mAhxyV&*XTy`@vA=}gM+>icG-7X)IGgD
zyX@b2##Um4Bv*x~J;Ivey<}`Y5uFo33#*%NV7xw0&RytTnyUJNS27&(AgeSy0|Wp@
zL{k(I2NDsm!8~7sO{8^Z38Bd~`k#m2-*5*JhK>j4J~MpUZ87LE4biCFRR>Ug1)(4)
zbGDKa?h+V(8o<{NC_fKJF9$f{Q$xK?qDa*>tV_v(8%iVl`W-Zx4Z3Wb_Hncgkh2Um
ztB--ErFhU!s1hRMuXHd6oPY5nu|-t*UEmx=Wy}V5fJhCLGTpjB_2^Uy;9L<L1KFlH
z69gdfVI)(~m5MXbf<g|GBI=tNqj@r<tV6F%K9DFC>t5t8E0L0GKF@b|qY)qN@yZ|}
zjaS0Rh;exazp#SPN+^aH=wTNkUn{kCZG|0azLgP((k|qlG_2d4P8X849fT!uEA=%}
zyB@XOoy1H!vfd{ifJ6n<nM8Bt^Jsn<vFw1xLTVKcF>^zKr6{F1f|(<BZ+G7e5nH5l
z4UNOW!`HLEXwZU2n2__;ofN}va+rPyjttEQGCYdNmtO$j?1-LfN&#h;Pp#6>+}M%X
z^8w6xGq=YbHhZ<vn+%4D7<2<n&kwa92Ng49hrrmv1C!H1lHvrkS2?Jb+P!eVwt$B!
ztJgbF>`l=p7eE=ewuZNnBXkbYlwpXRrD4@HdvntfLzN;L97(;=bJ>+ny~;FjZX3pn
zD9rJv9t&F*ps8E>#!yx(STm7Qyi119>j1x@nBmeciG7RQB|v};?>igk3_j0h8XMPp
zuTTk+^NikMwLWmKz0G6X+YURcn;u7Z^Qht;*9h=JJyeJAalvF1_w5i3czDcP85oAj
zp4<}x>W^AUzt$j|CMl~o1RW26fCx2C1y@BKI0*uyuuJv@^nJ8X(>@PMnpaSw+%fl#
zvh(ih-K5Y=#PEi%l(hGZ#bXf$lqS4hbrtW<{NY`KhK?B;DPO^}{6uRRrJ1h)68KXS
zr>^+$ROl5DuN#TNs?U7g_Lyi<5eT<Dh&AYmc+jZ2S!O=lPMvT|fhX~x8?x(Ds6+~A
zh+MR=^Od5^z{{N1z|v##gU?{F+J%e-o%OOJ&@Oz8;G_J)Y%p&_)J#BP3b^mIk&c1%
zO){?&roaxxx~B2C1SBrET7SfS!wT^871$H(S@ATlSrt&)^L+VFTnfVXp4z(H7!tZ{
zXT4o%bMpI=!%^+SGvc;ZoTrtnKQ5Z=3^({BwXE)-<?s8{SlKQfTwZRtc<D}!liw*U
zoA1ner)W8sZ%OFmPYGwYZ&uEqcbntjue`+@H=8v+b6$|PE#&^Q+~utI^bcMwe0cf&
z)YRSWH!e0eq-OT@G}xDQC=5@3ur9MM`*!8~4_OX<GGRj_HWLXqvfd93>ATcw1o?;C
zoRyDiu{W7kJKQOwuEuZ3|5oc!kU-{L<FZKY*c_dD^t%o@``!pjv<&j~8Z4QN(`i~{
z!ec+`{gimHOV)x)zjh<ZVTb~FZJged0FU8pnAhhbR;|LhS7g{k;w`e@aPIK*_AWpa
zYQAq8I0oO4<uF<q&gN?d6S)|jxjGP_M%=fs>sD{RjY@PcjsVS$bQ&)~z-x`EM3Gc>
zP{_@jB*-CW@a1sft{J}P<&{t58$lLCM|5`&7yy|~=5$0M;@39aJYJJZe&r&nF_7Ez
zLP!c0@rM>4UDv4EWPAZ;G%am*DH$1=qg}5&;Z#p)6U~++pRAg0N?{PATE+8;K_Ugn
zLwIB@j9%N&%ywj>N|}pZc=(OIf{=mU23PVlzrUiQf*Q-ZmA?G}4s1>|O@%lws76(_
zYk>rlbaxU3N?Disp&7NqUEV{^>9s3uq^2G^^qYRo%rUXa_uI46cj<(nAoyLP`m?5g
z>LsHE_=l4L53$<s42^!W-<c<?9c=shL4sR(LwC=j)BI^BdGA7muVrVO9GPnTBr{k*
zb2joD53AAhRDUp$H&0A=$B?sd=L*5YF9a2wC!UW3sPpr<VQ(WHLbs2K=S#S3A}nuW
zY>*hTR#w>Fy@ACU-EgX#iMX<fYoL4|-;J5?i%K^iXEi(7tizvxMkG?ZCAE~1%iix@
z7}I`}+P+A6Ne_S$O?81szyGBL=*KjrwG@sgM*B?5{Tz*;MCgY=V(JW`lKpVc=Rqer
zI|hQvkdFFi7!;6mI_0|vqYtB_H*<}aUN?oU2n{?bAg2hJcPEk;d9slT*}~B@3!)%t
z`EUt$bc@`+N!T&*BF;#~^B@Xo$(fnRnHeymsj-Y|5TF!Qccij|rYF1Z+%<<zP%VrP
z0aA}eyQXot=3}vP9=P$-tL`bn@ubDaPe5rUsAoh?yJj#@p(Byn#9T<*MU%*|d|Xg^
zN<x9tf;;C*!gDFtl`DW?6i>JbvcXdw293LG;x#!)Y)$D7Fn7k19%-E<f)t}0L{dcO
zH_b^mLyIFB=__~Vpg=e107vK%vs&tC+jx3^iO^@?uSPV67qfPr<o6S8NJO2+<oRDr
zu_!4IV1|xR&mjV79%(Y+;^PYUOd3J^kQ|nvIU$ufGf%W|?H@v84!lv?6@uZl?zU1>
zI)f)ZIS9OHo*I6jH+j2blW5*#$x5tK#s&%S`!Yf`D2=DZMDL8es(>;CaOciMli`9}
zF;vRn%gnQ8#i1rvL3?`}-gH{kJGrj+_VmovOYpHx8n39;ONbizeK_>19{(;eF<l#*
z{Li1*1h-twS^hvTc9B@C%8+v%&xl_5Tl^B_+Cp3Eo@RFejY}A;egu4~cvL+}E%Zh<
zC+D<ZN8~9Y2)lz-ni0JgXtE#qgyj}DEJ6~|2ZuxEE=NF*D=VXEmY*sNvDk{<#!O+N
z#MX<qPQUN7VJ{Y}+BPnk9jX_ay+<&;+I_s2-vzRI-5Gnx_m*bley@&cfPuEpVRX-P
z(f74`kh8{@mBP;Nf)r)lq*{|ggOYq~-)#6|CFat;!f^T2I^0cp?I7PqLm7luI-i`U
zMiE*UXrhH?p&~nrq#?o@ZA10~HjMK_IN+C%WPL$xZ>_K{?P<ZF#Vr!TNgJq@0;FMl
zen0t^K&UkT_!5@%aP79kHd0;1f*zvS^5*+Yk6q(*xl!pchgJMvpZx~f@bX4DpG(HT
z*~sC`@2jhJWrGkVmF$A!OO6;L10z*u)vjre;K>a?Lm)N>R`=#F7w)XGnS_$tphcFY
zkZo$+5`0vdpojYGN4asj<d@HW(!c!!iyz0S&G#dzx0E&+djY_^@iDYH;QJmiK+Y%Z
z5GajZfqIbdxMUJ*cWr_jmS$>xKUE+Qq$*#&&3xFuU6zb#@HrbHcG+X|@2nh`9CEJm
z*4tz?L`y;Xg`&g>BSalWGwjkz3X(<$9p*;f3{OH1`|WfloexOJxOsSZ=<|qj{M?Y*
zcUhM#6Ctf#7P*dPKr-iM(B1KYk~Y_Dj>{Sq4P5@XjTIhqHvHkRL~Q+VyYH)2_Rw&@
zKB4+x0ZU=3!cmcpuvmRie4l07Qnwr7!W9&H_uS|kUZ;`KkbGmm_cs#1zTw~G|8MVG
z*}&!f=leQ22-N-ULReViQ0ax0Z$t-IN#}|v{<5D~BnCsIzh$}4$J`8<Hsr-6m7_)U
z-+qMVo{HX+iLMIxY)+k_T&;ZLn>p65EcesCW(MUC{DhveAZC?*BVjhktV?QTE?h_!
zl>hqq$ElfIipd4@UqAn6+MXAgCzsQSy9)!!&68KMuna#!qUAhDgQzebF6CA<>^@sk
zpflR<ef&7KFaO0KF+@N~g+BvFxb1T|6|P5A-a>!dCR6%C5j9S6HB1xQ-oi+QW+GlE
zk52A6gu^~KjN9%)rdyn-A+cxZ?w7}UF8iD7e)#`;*L|<oG6a@V_0yb;9%Bz%D!_5-
z{;+Zj2A`p1i<f{IZF$8)ObajHD0K47nJ}{D9XJcE4<#+tyWY{+4B-TVCSnh{1=IJx
ze?=CtAi5lGND>d^0(J?|p!#oJp)=+pfNPEt(}jK<2y{KUB}AFop)J4+ex=Lq;}F30
z-;!ORN&oXdKN1T6`0f9nx%vJHK>q~c|4$Ag?kyJ)TkjZAMb1bi11BKsuG%$%LMfLV
zPxLINP$_UgqY$rR&qSiUHUvEv`25sZze87)q1ZI^IeBQQdy-KLX5%4@iH4(L&H+D=
z1W$P6D?V_7H-v@_jaJ5Nts~|M)ZGhoxqibar5qUj!%XHjf!_FOa|FsIviy@)_3d+E
zQX2MnCbyA@Uj@km4Xf{%+Ka--rN5Q_kKTagbsYex#toys+fH3w!#F;a>yu?VYgqW$
zmub?i;){O#pI=6nOyANv=a|Cv6ng#T7RI~%gdZ@gdtM(?n*#AaTD6R;l>U#hhY#xl
z4YtmbMkTHb|269@=3nmoH2WEZgF-bmnBqVs9;VurG>p@`vhd1eH?s=^fJBn{s#}cy
z>hfs{;rI312VYJs;7eNrb;dvna}g4Uw?k)1l}E$)*Pmz2x?eFa>u>yEZ*tGuyW@=U
z>PB+tVp*wArt7MyD1>kPw8Gsxna;<{ZH#-qK0){d;szu%pr1c7KECwp?5W9Hhn8x0
zZws^)`FS(G*OzIkJ<I*b_SS{+&_3P9=z@TR1@sB!?yFF{(;T#b1X0+@1SH_t#-MI*
zG9Q*ShAR&B?*+vBx?NsI{~#eyI6U16X;ceFj+J%#F@N`%+|iF$p*f0;NWIb;WChw}
zA)?pK9O$ebl-rY?U*#~pfW=&fMpr%bE?(4tFUh`f<3<s6W+mPC8A5fAs@c~g8IPm&
zp%Y)xD>ddd+RP<<B4zCxSf~ZWwyPg!-hF#<_1-S@zJxG%5OY=U4>M0Ji!^PpMJfMe
zLGm&hDQ8YcH!h?9&wIVoW|g)Hv;5kAr%S$d$7xZM{dwlg98V?NTxu|siX&IdT|gf%
zC^D~qrb3bW&^0Oq06|p0?>;Oms{=MhBo;mj$O$e9j}%9dAu@7~q7%aOcP!fU+}13i
zWroMPFnbyRyDQNFe<1I`4{HMYnfXb>`Mq}Y+ikP=dpBLl1}WXWcB9Xkl_fC(_=zSM
z8<!`O=@?2`(Sq4GnDk{Xf%<dvAu;N<*F@E*19Hte@~>|-bU2zEU_r_hn$3PIM`lYr
zq<+hHwPc}0%+a1%>;2<T>FNpj%}OxRpU2$WH5|-0l0C5Z+F##jDw%xEFbB>?vl8)w
zK>MK0%$BE3u;vu?Py$4=tLaEZibw-L3vSZ|v)op(&xtj}7o-`F5r)j)F*o);S?SyN
z(u7}3EGyBS{i=CJ07b}eI<}15_c>#j+s1sx6>@?O3#&_hF+5c>YxR;XucpS1)n3%q
zND^h}#Rth;9#QZja|K6<c<2L<)x9Ty27`a})!XssprpJl4g7ITkhpiREuEcSw5F5!
zVxi~uKFvYb5wy(@e?%sXA8swYnx+cwE-T(~ky%US%b8E800ka`lDk!8ZyFOx3Tbf1
z*?<IKTx$>95Sm>V67IBp?i>Oz;|iSwZ;u_!4``IfkF(<%xV1>&o&Obm0y|D@q)(Kn
z@LZa4Ilum?^@UjNob2*5*vE4MLH_h3LRas$D&mRAh3L^g4f*BmItuHc=&$pI$b4?f
z+&PV*Y^lskN~oX40hYZryLZ9^?-ClPn3woHfykf$W2x0XZD&72Y!Da|%FXXAM?d9F
z%%rHb){@!mDJ#5B%=!ZQ8ENwHe6s4tSCaxVa_jl;Kd@yuZz7eLBx31;*?%j)3T-Jr
zP0-x${e~ON)J;5FRty5ToJ%bVIzU1!f>i3iOQtvapC8r$TGapg`G4+%{?AZ2mVXB0
zA&BlL@P_5G9EBt5PXB4uZXr4ssMUx}5{rt9TM<p2j)_kR?J#|}lKkswQZS*kw5ka>
zPdQg7|7n2P_iEqSCepH$kf))7$72Xs1!7lG8Vt0(AID=y&ornAY6>qvuUJJ;C0V**
zJFGM>TuUmDS|3N`O4LB^h6!|QL8wpnT~7Dbvf&{YyMWrjSER{D`5a^RG(NbSKaL3U
zGU?qE^?|Pu(A>3$NO^cK?+ST0%Ip#Le64f6VGT|$k5=Pi?;L$FV=p`SL6yb*K9gNN
z(<Zcxyq#gMSATy|^h`Z9rV}57aWEevGhUFTCD{NBXZNBr?RD4hw)DPhL_X)8Ju8kg
z>9QBuN8D1}>}d&VA*6ic93fGPc0wFD#uL#B;UjFMeDB^p){KC&256TY-klQ~x`CkU
z*#Ljpt5EBl<>vxAA0*6}L%oi;6HH%f=DxV%bv?{H2nbgVb_D#Lhi`Xx_O!U{g-iR-
z7pl_4SW**|pAoX9ToGK8p1p&8jtCPnxJ^cMp6Jn+MjZ5@F!K5Zu}a`Bj)~6Tz86qK
zCMgWTC`f`2x##K0_~t?ILo87~L^sUELe_x+3WhJb{Gq6LLK+{7AkvAG!1h-~roK;N
z9ntA7aA4->um(Tg&xX%j;yBGB9!U1IWQ}+U2aI8fJ%qRYlynp255v9B+OAgc*Cj<j
z8kso)Y!HqoD->l5no~dmH0KUGSUk8jL;j{3=e+<*E*DA^2r+T)cpW4rDtp-8kvVR^
zEVWG>*8TH>-DWSiHhUyF=>k~!$l;e0azv^#^fZ_t{c{GdE3ozj9WFH^$)m2UUzKMs
z<~v+W8C{G-?<zq4WI9NvI=DP8qOGaP;T*=q2HA8)k(eh?JLAo9u3$TJxxn4RG>L&I
z0P<Uanr}HFyrk+wgLL*7w#A^iX+<|a7VWQJF4JjFjpt;2K#o0R#*GWiqoM5(|0Tjw
zmCSTwF;apAXrflVTZa~f?4u#}=xn=14OZyAV2Bfl4oDJ@Zmf-TmBeC^eLVHM&@hO1
zh9YQwV0zT((PDNsCU|Y9eJ=j>)v*>m@d?#;y$g=;RllGfr>L9E_=cC;%=gfs<HSWA
zP?{d?qwe0YtXkNBD?JMS1IJ`-od=O$))VRPp6DH`PY-e3y}8GaI`QIaA+r^2b!&(F
z4)@p-J(`1X<3y}3>~<_hyz*K=rSguIrOwVrIJ{y+XJ(I|4u48~u<1l1{pzXZuZ?KH
zYAI)8x)Gwrb4SjF>l@y@bqpqht`!MTm)LAxClB4-A?hEtn4Yf7vbE~^L40hVa<ZOn
zA8bg&Xm&+M7Er4oKWM0<lai83ZD+^?Ax$~l@tD-XXlRXBBxpaqWaeUo9jegb_Fylb
zf`6sng9jmrbm1Am#pM53X@02QR#8PoPoot*X=iJOy}Z36Fyh0@G7csqRq6R~&rl%Q
zcMlSb?$sSwP4eWe9gTe~GBriXa}`YgeCl%?9pgHt@C|s!f)#GueNE^7<MFJvRf7m1
z3&0W{6ScwyDS%S+-L-0=)r9BoLE4>MN?M{Gi_VX<iUO*bEGtQFM5poz6XvvOur~<@
zD7oETW=wt~Xq|M0-@X9XC@07tT0vNuS{>;WKI!kNt%RF*IHw_%70ne+@POP(+#FdX
z#&zsL(_A9j?2?S&uDZT^A`pi5WT=}o2(_!MsEWM&FDjZAO5{jNQVD&=EFzAg5qy^D
zZ6K8_9M6zL|KzX4la&OD?(7lNmX~+;u-$jz!!axHv)7RhMFffhXKg|AwiC34($FpY
zKZ^diwyFCrKYW5ep_#;`mH16$RJFUNQT^ruL^ao1O(Ct^{H@dL5@we=&{;-?a0qP)
zwH6RjxnK-?#EH?pux`+K<-U>pMXS3rkVEF9^QI=!7IQqLG>=VLbXIKSM2j^UH^mK-
zKS{k2DKcZ{SS?vX_8Ow(Tt7CP;HcTHVPIfDX1-()EQ#rJ%A?UokgB%Aobnb-WFns%
z*%|!V@gbG)E$^+___HP9sk_-Uq>e6_6v0E<HrmicdQ+nyIS3^gR!858-s((VR>T-T
z$nOVdoQFOV!<E5Uw{hKju%8p51Pquh)st3^=m^d}_|;6)EArEQwbKQTlhPr*<nE2~
zM!!6nj9!#YD`cg8qPJ1aL_(xEQ-utOy;~s|IcjUX!!*JWSakqIKn5MuLH{<>$~5i*
zEnTvoD78IJ={qyO1|lE!r!*(81+BLUgL_QFAHJn?zkKUtmr71i*Hh>N=wgT@W(M5g
zBF{Bvi;>=uMrbqREP}Qe+f0n~I6-odUXzXozWR${JI<c4sU1NVn<MAbQ{C`K*4q;1
zcoY(+VPlB1SDR*_%yh+u+wn>5RP=a5hJ_J2W^!&)CmDep&;!%r<c2N*<LJo1m_%<d
z@LUc%5t+<VGPgk~FYM|F`Jx$s_wXzel%Vt)XW<}J)GJ<Ip>=KB1lgIdSq~oaPAM1m
zrO3~nAtk1nW34#!WhS=Lut3N-o8PBu2yfatkC&-9U%wzDWjM~DBn7KR=4EN<ByXwS
z|D<OaxdV&L!F`IbL%=Z#PnipJPYaZmk*B{}xA-)P_x124w5KOb#Y<)?(7@G#CH2b4
zg>Hw$pdNpU$%rmet^qxq2>pS5c3e2p5+w29bTRRZ4DEL(aA1pv^kw0=pks_WT4?8?
zqw+RnMR}kVw`*B};ieAi#{0GoD0LQQ@ZD05J-LKzuxZTR&g^~{#<+p<F{D?_5d^Q=
zeiWg3sN|Xc4;VtM1Fc<|{mtf<vgeWYq*LojID*uH`lnYIy&gInMY03vK@3;&^YF}C
z!{*b#pCy>za0}Yvy(y;6oH+GdNgq$DUZfNT2^^S+5h$$JO1*gGltxWmbehq`5-B&%
zRTv`VgwlG4jlKR+Dn#{n;3hADI9MyBmc}SjKqCz?@!LOd-tv@sD=1W|b&y<b$r2QR
z2HNSLXCafz{B{2mIB50}yr!1Wnum-=G!ZBYBTa+~-QxRbS9Ie^HQD0OB8=!HP73k>
zN(8EQ3F2s7ke!UP1fJP;EpeR4-#1y|p5fkl&~BHNm7U&>W$4pC2|eA=<-r+CIutvE
zcK}^m-vUWwRBc%u8n(7O7?c27GZLGI3r;R^sY$hG#)sn(Pzm4YvCBx6r|EdD7wqsw
z^UzTgWa)s;2+}Dg4K`2pT2+>wY1PKSriYMNP(VXt>mo{~C~Q#X)q)<`+hl6vqPChS
zYQ$d`#NPsZD*0oGs3stYFm3S|+;FsJOVY*F&)mTLfB*B!<<z298Yo)pGeeR{+}2Vn
zYi4aOIQ7@sH1C6DHMs+DB$rkX=dl8jGpWkrM}$i<t@%nuB!YUSC6}g<3@pbM8O~1F
zsnFhxHeOH5ct|>vrzgbLksp+tZIeL3%3mEvtsGUl8)<G8&ZZcPUh?`wl!~ZHCqZ<U
z_Q71J(zcgMADkTU&XE({2_*+@`#_^m^-*d?=D5h4q{yV@TR3)_I1}qN-5EU1!<Fzo
zAEE>_$$1D07mjP5nxgp^PlqZF3%h#XyO#@%>CAvjq+>VvqC?Kop2-+?-=CYv=I=Xo
z68V2)J;G1@&XI!$pQ1yLmO|1X0>!y6Su|SY5{RwsK~JDQp~<WDUja3e3bfn4PjRzS
zpV7@-AmdwXn<upBi#wBNIg*1Ggkx9SGO5j;yMMFmvu!R;Q_k2{4MF(!?7LevpHZ+f
zNKnOiE@XFMHR*3cq$9F=VWn`JYO<b5Z->#$<}(3h7u_xbhqi+-Ghx)Cyn>lw`WaWm
zMS{am*l?7eGI^xsXZ<fWR58#LLy_1==PT?Nl+kX4m58t_4oC9a&Ms!MYPB`CiN1Ai
z$T<oFX?O!RaggO+PuV<RS^<`OoBS_7Ci%o^i+8k=OIO3gZ8$r1o6-mn&!yPIPGm4c
z9*1M8juWY56+lAWH)ID-UX|$4e%DP))}VtXU?p83BTVGKtIvaOwKS_y)eFnor10B$
zxC0q<V52Fo-=N%Z5~dUN!t2%ncG_NA`w?Ny857H@CU)yHIF){iW#7A1=rK4(XB%S>
ziIRe*oVuccV%s~%nJJq%3ltaCV+F>CFJdbf(2Za(s5EAstaY4H0j{*S`3zI%^7>@G
zt^f1#g^g;etV>Zmds)VZhK4>LeV1mYtK3n&lv`%>#(H+b(eSX3<)TTX?tZ!2ChCqA
zwW2MH!^<!b>7b^Z@Ame~x`Y#>53tH+nm(8MMJsB`eI$NMp(*WsM)2(N{X_QtS7wr^
z<l9tJ?WfS*nL&|~E6Juwo$@sEGnY)%(03I-IF4nu9EIzzihkNCze#Vrts4VRE?}SI
z3^D$k=a4sB&YWw?NG`pwJu5w007-^twKcVM=YO@cvpxi{)I9s_?A<^&jo)N8%K4(P
z9Z|l5dYoUH0VT@X_VPRQJJI)vy8vX@@Z^=@Dl7>b?Lk0Y?GJv@pJsw{;yYL{&<Y|D
z91{yPJHAptxEeiRilC)x%!z=N;{;3l=pScs(%Y)?qdM!EQ@^bUJVV!>4k!L;aZDFm
zt>ig2GZVOY|H8$7LJK5S{SKREJUq07^}C0W@8@5g*LC%o1KYk6%EylzWIlU1?~p2w
z&K2RC+>(n8bmp$(R=WEA{vUst7y7Pw^<*K}fsgOf1%3Uj#a6pbT-r9;Xj|%Aa_XZP
zm&oYHu~6eotM4|eJ`O!&ZD<(XU@R>usp{p;qmbXTdx1YB+cAiN-xxXKtP0mnz>#WL
z6YCqg#i<kOY0gR;gm`_mvzDV<Oa(LjQ<fgf7e2oH5i=tG`U`BoWV2SOV%;Jy-tLHZ
zW6H>Zq4WFJy8wsE;TxyM9!S$O!p{rIU9a<GOJ<RQQ|Dp-YSX?8t0zPG)-M~k7K5tq
zwhX-fTGvhNtf~X*#VqvK8&-tgD!RuLQdoFM^5DTXlZ5PX!z=aTc4Lq5Rdv<&>X)yP
zsX)S#(30mQD)tegKIQ%UuNtiC?CebGyYtaBTP=6qccROX$YJ`=7()Bj;+e75MdHJ`
zqv!%OP!00)i$<{g8U^(a^YcRH-@asICL~nGt7=n>)r$u(aiOiDqsvtthMTEdbC6|f
zTE*VMk`FXcwXw-SQfNt&ipEjf-1GDE>(klvE-kVc(5YN8kk1hLFxe=?((r<lO`ctw
z^;$ckuU-4@kGVQQCdyW&bKc*~BUF3dd^q&FqnmkQgUVI2kC3FNZWXs5Pb#qZ!5?)E
z$+CrcF&*Y^X|Q--0VlJ6@v6Bcc^PhB5xnvL#>j1yknC-BGk~EF`;K$^YC)c!VHD~c
z`ns;Jb72?W&WoWb0gyuKS!b`RtE=;BfAi+eT$7h7vfg}gAOrlpFgg3pfa@{$Dun1c
zqq?YmXl7gGV;`U83s-DE2q?@_&Ef078D902k~17Y0Cw^s+yx;mI-Br><>}QwF9vrC
z?f(6XfCV2y(4Ij@h<0i`dquXrEzacP0rZ%SuqOHM%lm8?dE?Y4^u4$R+cL};vP#x%
zY~3q9Qew!mCh2vsxOQS<;?ZIqU0q?l(v`)7Rz3A$D0uBAjs@&GW&IB3pNf!N+D*Wg
zM96X4PAOXP)qtHSpGMkms#+kS=A?eu0=_bJtK03{kGJNaE`9pgN7fxB;2E_M2)z9G
zWYs}nAr0twJPm}fDrxKX?efTY%$j7lWIZ>B!C@4ZOIib;?AB-`GHZWd*rB`Y(C?55
z8HnaaE-oY7Ccj-~4_jJV6muNxm7?S&(O;l&@nTZ6a$Gdt*$BdraY*G90jX5lEaKb~
ziB5!xe&4ZoP#Ob*iL#pjv6cXx<q!A&+BEy;UP$acwU@!T_N!wu$`F#HXaHEzXitJu
zWHj!17bt*A81Cy@t8A755)fZG)oec+*lvv$x)?xEg2@JzQn+}PbLV2Q1@_{oQsLuQ
z)~Q8>!dpyT^vQrdQRxrK+P=Fu@>v!6uCEuyALS4A)+ZZQKemwO&P3bFIhzOpy#+1t
zYL9-q5;u-PXAvkz{k_CdO9NF*i_6#$g~QGU?{3jnxFic*DbaAb?78sGp_;?O^c;W1
zx*PiR6bs+mW08uav~&y(l&<mzze^jp(`fiPwM))bR91fdg(g4Xht<3!@B1JATmwVZ
zv8oCfesXc*DNay7zWO@U!bgMm#^F@<3j)F=iu3+;k7bYSp+k?hn!bCQ4uGRG=`*s;
z7<xjf^@ATZB#$0-&7K}NN;Yn&=@da|ufaALM!T0GdUwc?5t)QdUS6I7u+MNSj=@~$
zz?05@*e{0ms+2Mjhsm^>1Z~sqlP|XoAtm76yZ4=e^}&PJFeft}DyKt8<129z-t+N^
zL<`B59XpgUI8zZLtgDbyH393RxZ2||r_hbrA>2dm?(RZ3L?s~Moc)<E{85M5*4N)a
zBpQcz5#=o|Dyl$y6E<%<BtF_T15>F3D9>--p8q=R8z_{8HRU$SVEOcSt%{5-LC^)H
ztbi{S*NRc1u>k=q8;0RHs|NduauB<6Zg9R#hD_f$^C(Nut)DJ3%$0V=rG8qt5auuY
z6#wUUk$8}&!2q6XPGbKxUlh2vrSIt=LF(&^b;IR+9|8q->vBXO-?N)P0T?sszvsqt
z9Y;+PgJZfb<br~{{1A>#>p1=&7I9WVOKv!E=~(Z@9Xh%50C&IW8DzP{=7f5OWiczL
z{qT|!xr7>D+&xsNDiF(#8fN|ZT^2|CI5DWuHGd?dkx(~hod0lDy@OZv@j<vrgci5b
zt^Dm73)XI3yeJo!H1-PEevig34`1xv^r^|Kl2TGPkYI)5tV_4hl9jzpr`xJMm&#+N
zd<ArG#I?dfXUnQx=b!bAUobZx0+wmiR(Q9lu<-nZ9m~7tI+SdDcO9o=C!nQD<h1b^
zqvFV^9VhAT?%^?vXdQ!MAGF;QHd}^f$L~LINoV4+#4g<OK-extK`>k2o){GsWdq-d
zVK90;qM{ATQ^-!_fwl~zRV#DP2$FZi1CLF|g3iU5i2l5C*L}dPolq8Q#Z(5<9}&0=
zuw+wQoAGge7&P2He6bdpt4a4!)u2Y4FfX?W*^ggrLe1zWsOMT>HNF=XCn$&?z-#m%
z;#z;2xWum8^O?tgIYwSyo|ZYUM%wWuOO_ax-@k(TIt0{r6h0Rr#;eGiKLJNW47{Oa
zj~!D%5`lmg0fXsJz<$fI=E*`nHGYW6sBzsNuKji0baLR@<8XFq!b9OE0LU01LQzQd
zcUupHqVXUUB{fNh>5mP=gkfZ$^1zUA@g}T;40z-!!RZhWwiEs4biPujW}IebO}CYc
zS16r}fKO^b=0^gUjK=fH0QSfF>#re^k?*y4+`)3f{E`EpdQ#Dcv4r4X1Z2P4Z2a+9
z3D6D8K;2%0`<{SvZ9I1MjZ<V^?<aC0yi)*>m>NF&3bY&@!NR8t7o%%Yir$H?0)SjJ
zj)vPVE=uT<y65W~1&nm_#*LB?K)(UqlB8G0Y21{dgvF|K^ytwm6W?45nfG#=vxL>d
zHPG2a0K%P{!SUw;|IGl^EN=Y1`<WluZr!>?BOm58&5m?uVOe_)%Yocl?`NJ^10~7{
zLqo&+ncuM5U^Z89OKa<qVhk4xMPL8dMCo}J0s;cN5vq_lwI<o2hVQCP2oAQL`^*&>
zdKXI6^VS-ntH=L(RDXT}{Aclyyn=$o_>e*60OH<wf*_0C>{0%&Ut#{w9S&LHsD4Cv
zMa|o8LipI${PQ<ob~9<#-+uKFyQpQXK^-dLjW4%NzU5%ae#Y#h|NOPZDYEytAD~h+
zq;LBs<c~^SKCP#bbUo>MEOu)*m#_NI7to)+MBscoj3JpR_LFGQrGsnr$h^xr=k=JX
z;6GpRD}duCWP^=Rd)a$7dHv7dcy<fnY`C{c9nbtq?|c^iQv&}%Of$dz41;GoxhRs?
z1C76*Ip>+v!#PcV=3)Q#etXcUYlw{Xy9^HQ4e+7}ZEHJi4=Q~eu*DFXD7Xa$1<RY}
zyg#!Q3W~q~oH~Olm3&a~gx9dF`}V)QKd%0V|8~1s<}LVd@5r)v*-4r`^!J}NTmRdg
zV39ca+kd%smdlU++p}X~Idz3O&Hn!8FXk|j{qKK$ZSFr`{O|YVpF94~Qu$wNV$Ifr
zb2GlpR*qKW>1j+yxS-#d{~1L{1$rw_t2v-m+W_MW96pomDV{vXNSZETwZ07i`C$Ox
zi!M@GdwC8t0iKAb`fAf4H@5KcmAZ)`>dadR5V{8S`?pOlm)COPZ+3Nb2b}!Grhr5#
z8i&)BUef#={|yqMCtLZIqFL*R^<EpMk#2sPk{Z_eckkZyYKP!p{RF-GM3j7(gw@*n
zg}3mJrC)b*sg-^F-=UF3V(n}yg<txQpMS#0|2_)-3EclBTbMm@BrYHE-7bl2HJt`c
z&uQ8?_6kYWDy@kaI4V@mX%c4Pt{NZh3r6#qVw61cp-cBOzkjx<u=#__Pd{J#c{|7D
z-y|%5GCeKL@<lxR2bK>mldIkpHcLFe8^*$_+?M-A>gad#T?6#_DEGL0{$5~U9Q8yu
zPj6O={)9ON^CjF^o{3?<Wr7Zlf#q5kB$={DkJdn=q7-{lKq>P0K0vmL$m?0z*dmZh
z&&|mE8D`#+RI;FcOP>IxC}dd0%?lO7Ww9-jzx?vc5R@bODBon|<P^=#A0r!6da!@4
z5+)pgTT?<?C(uYXO_=&#>E*k;|He<Q=g$wc$;_g#y;aa3U3~xSo3PMOZed|1U^<)O
z3Wnr89K}Ku0HChAClSbcd(CWN6mi1N?kPq-9Y*?m47{5HGJ1R;MKC~~e41+5(+V(X
zsiv|OU~&d(dE@FMXTN3n?w7hH65P7I%Yt;Kmr2~ele6Ry8=pows%U@=n|j`5siV+S
zL~TV$54v-s!Qe!n2)WT!6BmQI7LkyJN2@*wmp1-tGwro?t~TP%JXZ5{FXoB}Ty|c(
z)2dhFfP3e%Uw{4e6CXRsO|wr58X6k+Gq0_IYxe0s`tQ2C|MH3N`J$3{nW&9E4>SV1
z&9G}eeDr8awHsJU|C*O4w`wfk`Na#YIy+dhNGy<3BH;rjAcJ;-oj$Ohiz`CH*t995
znp-~f<{9<={{ECM8~7^*1m!@IGH3WrEo`fQ`#LFV(Z%|SZ+Ei&gk>wtL{r7i-e%>|
z!KC`Phh$~vU#vYb?{3aZ?}!xvE9V{yi-_R(@y8$Yp~`fDW5#Rp$8|wb#J+U1;-kr*
zF4lt4aKS`?6^_$`!N!g?T21&(J;J=D03E6u2kkO@0T;zb;PKim{msnG@<A=>!~`BM
zK8NDL;_D<TLyk`-<FCQ!o^!bcZ2Y^7n3CoftaSo*0*(tDbsL{?_+&183Vg0O_&GS(
zZ<{0|d~zfot#SN=e_-b%FBezA7!06XV8guv3nNAR+XNHT(L%t;&&2RLmNnln^kUy>
z63|(83Wdg%Uac#4f4y>bE!Pcx^L29`Jem&5+4k|EX<N*pZ`PUs;{G``Q1sH5-|lCA
zcM|1sIp*jL^zdUs3#507@5dDs6as4&ojld}%oWs#I+kuVsLCxbx19?f#h@^5KE9*4
zS)8f%)05U>p#F<WOGnaNpMs|RaB&7w@=u=QW|EecDIkqk5Of9J$y2=p3K40b9B)8y
zyno4=;r97e8OwuL-gN&33GpD51sWaDh-&bprIPgG$C4EZDV6CL^r&hB(;3px5hQAL
z6pAnfAfy+ub<{g>{z+qPP|L&0c#Si)C}$%PM~txg`P?=>`R0M!=bOiKCwZ?fY5hKa
zo7IA2+Z@%$u1yR|>#uWMzZiQK91j<nw=+1wBrY}z2&@fAcn!|(zOU+!S)X1BIQi!*
zxYiw?kH36%tCS$b$#alX?6&-TGJ6V`p9)&vvnulOJuZ%YIh9a#{mb*TE1sYO4Zgg8
zj+%SXp$WL7IKv#VC3zgrDi+VKy&;Fn?oHpG;^XB#0yMP>m$Pc586zDafWHchO^(1l
z*788Z5mBnZatp(b(34+#@90~Us$I|kSLDo0nL1jnmRQ@`{nJs`{gG%kNJYt>0Hp0C
zr#Ikn6vo;ZhsA%~7FKsE8X5tWGv3yB>sVN>Ji}dC+WP%jpUmrAK5A|%`G08va^^1H
zTG%{K2~mz+NdFE*l`VrQpG%u%PxG*@6j>tiXbsn#_`AO@a7EqlURp#%geGECKE^y$
zo47z{=bzA$U1K!7cF#7G-_C-kCKteG^^Rb}hX7yNWNAGaZNkW-OIdN}g+Ku06%=&r
zb-jI?(4Hi#J}taj8Qg;c(w3#GR>iRk8-9ZMS_P2!B*z&CF(_2hJKEdY^wCAX&#HYh
z%Y&Dzv52m}cn^5W+<G_}sHkm^+lI1W@dU`ZGxoTae9!WQ>!azClI1(k-$+-#bo;A1
z%k#^VF2}t+JVIzVY_N9yV?jsTrK<Di?mPP3;-3m{UAAw_y-Ee-7{%GjhPE9hjVE0=
zc8(aT1JbL7K~G=x?h|7ICg<ZoP`xsrvuo2PDQJqy@r*nJ0%CC!;C0UEO>vk3J?x&7
z><A4j(DzUY3!hs~P6~2za@kXeUK$pmmoW_TKETY6%xwUeENnBRzm!+4YM>GLgaH<o
z4M3GuBa7ID4uL6+IX1rF{`o<b@42McLf-dgKUkG$LPY^$ZxIvI{A1DT8r&_#FxjQW
z7jRTWB_#<}4j`BoCdDG0+n}o&Vu7{r50{~Ldk4=|oOo}B0zmg>*zG;NgXOZgfTPoI
z1!?=VN}66Mlr+zox2(<dyWeq27y-Rhdvex=&_2AaX11NGv>RtNpe4}3hl>aChDND-
zzqJ=(ZfDRs91qLd4qjVtuJap}s*^PF{OG9omuKH<nHF#R39`xyS_y2|Ia<H39dBr!
z4hJ48!b|xs7}f}a>hS16y?7Glj)kv9TOHbjBt&(%y-6?HJ9dBF`X;9!EJ8X2NT>M6
z6kz+ov9E1PAfzpN>Z=Vh5m3I!a(tj;6%XnB+nr`D2Z0i&D-owm=Wb&GtY#}A;j~H^
zS|V|&pYH7kHG==aWfK2Ntqp$Sj@j4?4}k@EZdNMl?jfKVXYV+W!(l|{7%)-Mic#(l
zLqe2bL$GkwZbcYFXBpPU8><!I@g|*m`F-U8Aah?Y^jN9e0KP<e!MhZOQZOw29W6u;
z!MOMbd0+uyd#ml`-_F7EESSy#-NrTiA5A{|_|f#6?|vQFIcFot4Scf-0A}~@+_{gb
z7UW$=$Gp<RgFgbamV>2N2*G6l)sN6!I%5y9gH!R2DS_E1gDpI7Q`^Ds<WEhC%_z0y
zPM(xodKE(nL>F@IISG<U9;<FCFgR$TQ~?uK!8}{({rJL{<p3!etQ=Ue0_<9ra&W`|
z9Z~{KERL`8{<_|hdZ3X)wj<|;KfO7C#7Y_Oq=d)!S4IT^3T6?SW9g(z^o+G6A4O<`
zauxJHe|5a3I>4{Zm>C`oA7LO<`T+9B5jEylnQFVPeYq_O{4?FyBYr|BkmN9#TI|Ua
z_`(&0w-0yMvPI5hn|D?55Wr}l3TG(0ijUZT=j<hH0Jy3PnKbUQ_*A}+z8LrvqF_Kk
zbU{9K0`6x+c$7vstoIs-i$HUdHR4CmC%meT690lvsfttI3121&ghtN)i@o;_>iUYl
zMM;b?QG-crAOb21Dos#8q!~rAASxm#RRKl1bO8ZN6hVqu0F`0|>C!u>NRuYgyV8q*
z^bU9JuQAE*y}2`Q=Dz=Km`tJ&;8V`oXYaMwUaKP2qC<gthTV|(z_I%jJC}FI4pl4;
zMM}QVEh9h{WNl}kz@w+eE$X+Ny<%TD4e2Rzi}mxS+eP>Y!M*my1b_u)fE9=4P=;D>
zCZu2qiO$NvHQS9yq0|dd7!yS5dHwW-2><iK$h4fJ8=mc73J>@Xy>{!?t=opanzPI+
z=yN@9<1};LJeX+MYFd@|v3Nq_>YdDdMcWy)@G8+8BfPc?l0a1;*;)eB3d07|G!Dk<
z1$88hn!jioy&J5iOd02WoMb!RxhVfu`@r<c+y%v?Jp;_tDz!K*&*-NOg$+M))Rc<E
zJhiyyFzfx9rM{+=-Hk2B{cSye(>%%C=?MG0rbw?$4K61Mxmj+x`SfT;00M^s9Tf27
zM-m4Y1gioaEE@TOX>q5AD^D)$TgQ{)?AN(Uyi~;eiAAfdx%Q*wIZZ<ydlb^H|JIE5
zgyBBn25oKa?;G~$U`&YF!S&`%3)^B))7gumC?rJ1GL}^C;KTu8Foe$Oo;hP%o`ciC
z4JdX)%<PL78(TmpjIAKDT%dhp;M}yHnv5W@T}1Y#$3-}aTP1=}b_h7Fs4}50f$`Yu
zLVU0BU4FGnWqWK>G4zZ~bai!oD?2=^YPT$=(S)t2;ZMoR%2s#VqhsR+O3{hRE>EGm
zRnP=Ky-okaTjjaBctgF_Xi-E=Iv7(ZoKgi5F|qqT&}y}o3=jtQqQg~|f)!z6M=g$C
zX$OdX?nTPe(5-@r0ynVCYC|)@APpK>zhOh$toi|OKK+`r$e!HL;jVVngj$aaJVSWv
zN<vWR=KykdA0D=>Gaf(q(w!|w{);CoZWDamH*x*Y4DdQ!*9g<Jqu8I$q6i_ztI<z^
zPkWSpCSd6XgAP2#pg?!%a=Uv5_;*lhEgpnM8&d04prhW|9SFXT;-g6lHO`Y@@S4~e
zz0^BAn5c)hK6Us<u@3je97f{rw&W?QB%Uf!R4!~hc5~S#<P&JoxK&r5C=2WXMN+;b
z0{!PQi*;OF?g)Jk0w7!}9^8RarZ%s**t(|UNIoNz7B~5KCOxCY0bkWesokT~{1%D2
za#(}9xx82SWp1-a8NCi*XBBf}vk|gklWdwd=<4DUz4{oQ+b@<Y??Za5tthb7wB%N%
z##rFAvB&g`6Px(vaikS2z(j&dY|snVG%hHcs7u?$e-H<M@v!(|kdxjw@NO|*CtiAB
zi;UA@uci2AzjGdb<M?Kw@jtu!Cx$l5i<cV2K8!jc-?pIfguGSr+n>o{^JedNTafA`
z){SG|9h%#iQ!@&M(4^^G0!DxI*9P8iDC#~KE2G(iQ4G(?83-@sZ#QfHvo~W(d$EQH
zLg`bqs!c{Ru~Bmn^7(qwMYkYMS1UV7{<d;uM_eFCwPs-)VYl=|ZDY|~3AS3kt@Ziy
z=d6g_!{;vUSNWc79Sv6PiZWk53=iMT8j~SJI-7}lv}@hb>ad$mL&N9}+s><dUSjjg
zce>Fk0Ue`J-ziKfq<K(e;|*k#EaXU>VZD;?#3`2RpDMKS*sh=6W>o!NUt=k^{FlM4
zyh*3D<VP?Tvwb+279S41g-RU=Y^%GC(39$lejOl{-_*zDl-lR?;>>xccGIjc3O_FK
zUVVl|cT+6?iP*dMb}GcIx<2t_<9;IZ%9pT-n?LKaJ=;Hleq?f&I0Th0^-4SMI;$ge
z@9B9}`RscBn&>rAA~jwUYA&`(VQSm;<Z^y#$<aVmDjB<!c4fv>_kN~TefaD?@9ExE
z6LnT|`AwC1O&)mLNuV|44aUa~IS==Jv78-MeSG)%33-GVJN1`s>&>4`<+)%5E5z+p
zu&mE>dA&i&sl+Mb&F!TTqNx*`6@U5}&HpL{=Vi!I!E4nAarnUg{WSZ^NMcuuEADY!
zmM;Y+O!6dPy9NWD>X)afnT9n!0-~Y;zF`tgNRUlJQQYf8`-<5ZdQI}6br08CEVjmO
z|3F?;Q%NuiqXs>d`(QO4xPSltfkTHXz&S_XUsww4-_1v#{8`sKJ{fOP7UWKyU1Wp@
zcoh7OpgINBBno|gLmW`llQQ>g$aKm?p{JW_qUHV=$660&Zj@23tq(y9pIS`>h@eHk
znTu;Tl#ePgH_UF*Vv8e{4s?7bVP^G3q-KW{#!^$Ho=RZQWvz4FdMy!DLTApNwToci
z|1LlODB58nAaZ#2U;|g;J{}<mmbUD`cxe>ET#dA}kkC;$clzlkpJthaWke{T0K58N
z;2B|f9Ba)ED&{-2To=+2V9Ax&)_(H)(xv|1R-dbn0uZ1X2oSgRMGVUKfmPz_;h~H(
zw;c19zi>*|4?C3HgG5;m)Gp1c$I!3|b()*ysOWI5nfikC+~I@f>4Y-{E1YwY8XZT!
zYLQY0IQKyi;%%3n#j`*os^<F&R#ZS(D5JMtM@;O%`Khs}if{R6G^ynafQ1;UilUp3
ze7RV^2Gb_bk!|fp+O3S|jSD4)3-m>Dy9Q9Jo4D^0X1l^_k$73oI?nm?xri-AO}E*L
zu4!BA)}L6EYnH`0C$py`99m6rVB@r;^_h3{HLrw(8XCPq{Qbav(7)}k5S8-JOkZ(w
zp|#i2UM?P0E}q)mjO8h-AgkFb%lV~Z^A@L{eI$-`&1uaTT1~}44uAXV{_gIK;@TS?
z9-`6{txL-+P|2LTyYc9oPva<BA9C~;93N8Am`Zc7D#%9~xjp>AEOHa|01LF^NIfqH
z#q=T1fm}legff6K9D6~~Dv%D8acaci@y*>cXG<zLbvltDSSLm3A|=y^yb>s88j!{R
z6RFDCg=Z=ks%bnwixbSG;!M2uEY97n1Qo;?p-Vf_y*bZy0+Lx5(n9NNM$uXGO*j*8
z(0C^T8cjK<<opvvfc!<pm!VyksHvv4@VZ%+@6QoJ7QK3AjmF@Y@5MGUmI}s<5*t|C
z2@QQfW`-)j82H6GFf%dAWD@+50(}kiTnD<2t4sa%*(L~6QNgPxTZ)$phJ(CO{<slN
zSoK|IUt`l+r@DTpjfhQ}(eKn)C0Wexp1$?`v(f9j&((i!T^@#kB%{f#ZmzDvpbYSr
zG`&21wIJ+VN-c7Jcc_n+L^vWiwG=CPyy+@d`rRV&Y@Ua%oStvy8IEFuMOB~kl1|qr
zT}`h0ILUEK<z<uEr3p2QR@1RtnY}oMdRI@>*>TCuZ^Wi_C;r8Wr1Os<LA$7|<2Y)$
z7NQRrr#>1=w7sV2-CV5H`Zg^02Q>vou5>PQu2r@p&!rM4(U)BlW7D?Fz5UL`o9Tt_
z;$F9I7fyDRBt`O`T}_#pH2+Yai_+61>XrkF!E%W8l?_6U{2)1~n{(>v>TZ`5C<H~-
zf2toiOgFg3mbMDWmWd~WUVN6dM8<`PsMNk45$ef`$hmE59OdQRKwEV<9;GQ|(?pq4
z>{uh`ZX63oR8=25<K=Cu3P1T2Ye{YNG~ODu!uwi4S?WZ7VjXzpFeSx7XZpf_OlY~b
zQu7E(hM`_g8OJmbz&_p+r9EX_iy{^&={X9hNZalO1VkbP_rS>UI3{Qgl|T&iE_kMr
zePvL$3qcCkiF|wqsD?0)G_w1QoDFmEDHo^ifM}xsVSU*UVqFwI+k-K1LcxIXK?Yj3
zEG+sW;J;eD*{Q?aOc$k0ShhYEJF5VinyM%QTAj*Tya;Es&z~=mnSS3EjfB$<1liT2
zXuEU+j)?-t%{acp1(30L1MCfK*`+U5>-DbKo1dT0?1|d8?VcUxRlB~71A5X75o5WP
zS>%=3%XL+vBWkh2Ecd}{zk{9xTidd@X>16T@%Q+8FqoQ-LAC|3>^O!9XfpN~U?<t#
z+Ku(~R@&-{{;;*(WS-~ebDTA$$#1qf9kk~2tMM|+O3J^Ptd_BWFb}LU#_#zn4}S<O
zax=;X8{XrgcM0CbxEv9H6-uBp;evG@PRGCD%$J;LV@=&n)q4~ymx!0H7B7|lnv$>c
z<KmfQz8SJh7WBOG$GaE!<Vs<JFL_Sk@Zl<y)JBQK2PV5b0Z$SLUoCwXsp1pLMSK4Y
zP2=>yNjZ8_V)jCW#wJ_=;(1qnZn5;Q{sW^d7)<zpmuJB#1F1!W`lRTH3>q^_PW)6@
zSXd57+{nL!7c3346q(JN4}dE;V4w!5_B}Gm!QR6F@$qDV#UXb+G}oj8Bp<^lW$;IM
zjBnkH3XOQupzKG%=HrBySFwy8GpBSoZQrM}^1fUCfPm&vK$28k5@`)incl>^E3+W|
z<BL<^IWTYvsSGsPYLhL3=>#NHf3s@b3k(_TCAbnZn9*IFzmcgIhZK*~&`8qh!ao~*
zu~-IFHP8uA#LVDO!zyVM9pzxk04>TnJ%{TzIuyA~wlUw`@P4{eRx--I9g^0fT_0WG
z{U<J&VeR)P>N=6ZL@C|#7PId@wwlH$ZM?rt|8~ZB^P=E0h;q?><HhJ%s1nH!f~_0&
z?AcSZg9jl1H&g~d-x9J~wS7a!|5-6hYD(2lKgir5v7(p}pIWo)dzW8aa-MT2anvg_
z0!wqP(?c$&J5P7jRAhHjC18@=wj!#sH)5hLEKl*}O}}%T;-#q7ILx1jBRT%9*lVeS
zbGFjjW>k3zR_e$U*f~<SKAIZt(&<@piqfCXr=IQ73dmtlZkeX~A2|-u+IlBUcxWIr
zJv>~C<HlI7@%RL#A6Ta90WG~kagH-~&d1BivtFap;W+q4=x+4X=bp{{fsi9^%z`F@
zP*Pyc*0+UC5cz=cFVgd&#eLpc$i4#WVMyc9+oe<->DK_G_U(VSPNe~rY<F3(XIC;+
z;dKrns6pUt+7egTJy!<9k^s{t^dt^>VgfU|%O?$(S11ChLIEXWGf=n;8u@XlmIqso
z+tVW~TLB;q88`r{UA5)ZrLN<eh?N1R+K-Xz0RK!=nfl~b;nTZQ%WOmWv!I|IlB1Qa
z)$t>KI7WF#enetQx#_I;F2So)yiNyai$1pY^X4hOHpbYe9n?(yyg^1e2XVF(jId}_
z-T`n41hq5hp~(ki+eTB}dSE@_(lMa$5VInJ$qsTHe`g(t1V^=JE0^lZ$z9tdA#B1T
zrXplxKfh^SJ*xAsS;wHwA6}XHM;{yEE*?nQ##GH!*Q?>6@p^u0C1b~n;KfFxGu$fM
zDyJ~azv9@npHvdn(n7H;nLQh9Et0{=HR8lJQ*+Keo48Ft9^h6$z0(|?k^w>&nS4J|
z<KEGNTKUc-K$-5r)ux`0(_Zso{{qX2*2rwV5cUydpcJdfM2)HELz~0Y%vZ=Fu)~LX
zd+>}ew0ACd&m!(qdEx7=HeXz&4GI9}sWQM5j(M%y<L7;UzQU%tx%<eU&jDjn@B{*6
zogt`Fm0~Tgb<f}5*mh6uspc@`4`u?@xo$Nj8>28prw^!khtq5t^}FEtx)-UknA9xz
zd#XSWGiH!eR3cLJzl;JJpqgSC0>G%-v7<qEMDw$Iov8C2!^~f?JW6>-G^cie#gy~y
zZBD(27|Q$DK7Idnev`_r=U1w5+2N=Y?t33b+Y{KG`N_j)&jxW^`1sun91OmUFWFv(
ze&XpjKR`I>Bsfe7C0wy<eNnw$(9M0CP|PLqfR}FwMQ+E!gUB4Wm_%y#jV{bskmNa`
zHpR#p<&6F1W<}*@tu+)p7Me&{e?9`RpME|eZLdab4N>AS;&_~=-#K@`3+0uzt+D`D
zs@l*k-K8)rg_n8*cyrB;zS`Hu*1S@BPzN~Qs3vzn#boKpYC3s{FYYjnJ9&YjC?W^r
z^ohU<Clu5~>*R!LhGPWye>@CO49sc@6a@Pr4R*b8<9s!CQ^)xL|218`Lc=&++Gizr
zc@H8Bz^I8*pYpSPUiJLK!ooJGldEYe&ED~m=~5%>)Z)(0bAj>V&_KB;qqm>vkO<eG
z&C->_JOImnGGDAVKL~~M89lw!ia=n7zTVky-}-u=p)>~I;x?&K%%D%qda9>VV`6yv
z%5>UH(Mr@7OCRXuR2bKR)f|qVPYCulP=cc<2@v(SA|xF^F(M|_U(0KWM|cu7cK(>O
zI9Qp9(kcs^yL&_Z&3F!edT0D%Xt46Qc~nF&rmAqBzyDe|aGU4j_50)X>xP`7_A?}9
zIsr5VcL`yImSeEZxIzqT8!6NW{gXi4hLL7<qTzN7%L2(|ymt1&;5otdU5Hlip^jr5
z&+Aq5YdzY8k}+I0(T^4uOr!Sw!l-xmVl8a`_gZM}zp~Cr>kp5G4zM|XTp)8jp}BXW
z&hY=WomFlv{X=1xleTl%ABNdm7T4vbJh)I2hT@&s6Xai`K(<h{O`iaq?U>rI8-Nrg
zc?7_sX?--=yp3{4e$BM)O$%JuGwdQTD)5ZBzrU>7_8!;$Vto7p2)R1}UdAt1C{n)I
z(bg+uX}qNXm`=4hJ`&Rsf3?a{Qz`s%x2C5%pCvjBKnF<p+8Hyba)S~t#<dUtb;dg5
z1;ztP<M==f#t*X!lR@pbh2MflHV^_%5!MZ-N472!h4Ggg>)3p3@R5K<%(4eB)j+N^
z3E0!bt`U6KAt{A@`_5NaJ(XL63(RwGT;Wzx@dnhzPl=Apeyo}e9IfR45S4D{+29g=
zN^Uk`@2s_d<_6)U<xYC%2gOcqu2#n~ATS+Kxg|~#hbqptGRz_wRoI|B#u_TpKZ!IM
zBczcSK2xVIypqygB(2ZT^*jPXbQMgIJ`Q6xu)Pa(gM>g&R$;d9-IMEHaY@`}UljVX
z5T#+mN&`mb=H;83?%MxdpU!NhG@bck6RD5#5KEq3$)%y~DJO0`3~Jt^?}q**XBWl@
zV$--fr~u+WkeF3Blz7(IckRKog62;gT4fChM*YKUY1A5?>tg0loY28cM*3&O*;6by
zx#$ixnvf#6dU~n=2lsP4AN>4z7a;ZHP!(6gs2~YreAgNB6nV{25}gBuR5B9H_}n(3
zMvmR{)4#u*+B@+fK5=)(yQ)vI2oDB;&@yJw^b4S_LSRC#pmO5FZK)z5nKd&{mhml|
zLPAt+0rI)3w_1a6T_`y-G6!xdZVXh3P(5RN&lPdOx3UR%yR(jupq9ym<TKvDhO>o@
znV-~A%B<$FZxJ0~Wo4~J)3~qY#*G^ndnq^iig~wx@AZ0-y~pi5#mCKPrvv-A)1~nY
zqENKS&SAPWp+`O4F21YcBf-Ndr#6G?kGr=}J<D>P`0;AJcWQp^%kzlzV%I?A#VZiY
z^T>Fe^mE!yPqucGsE_uSrhGHwJsFr1(lPPDMAKVO)3~rCkgh8NrJ-InziQk*9Er!k
zE|xF3kD^%3cpL{YYs|9G>v(GA6fE5t7&mfiu{@W@SN%Gj_=0u`*qD^q_4aPm(a}k-
zU1?+bN2e)~gr#T4b--^{z?nylQcFyTw5iR9<FgrB)`!T<{F0qrTY^8-Ijl~^*FE}U
z9*))hP%3#3oM;5Aaif$}!|93%5EGiM)$qz}P1jyTg1&vYq=El^m^bEs`Yh)_eld1F
zcbH~BCPrDWKSoVtuM}7jY9qm05}|n$Px(^AhZ-LhnV*M!79YMq9gX^xi*<6keY$V+
z@!X7f7pd8`-q#Owh`7SfGi=2TaQ0yKgsj-vNx3R(kf|#^E!;{4W8P1VzrOgqe>TAr
z=<8w=fg`Mb)5eESmndEyM)BB1kSVpD<I+G_pG03S7}ZJYannXEJIyd)Pu?A(fYalQ
zfCeYj^2DyK_hRqGd;0s39J^+KLo!L(H}@BdM3pa6uN(@SYO$l~x6wZi^Tjt?!m>gg
znE5vTCe(AiDHlbYdKP~JSMYYWG5$%r;>Qw@=W-dGw^s9xt1|?2Tu?_j9-P~jJ`*Sr
z7TAzJ|Ma6}XyRcMxD1jz%Q-n2!xUIgDX1wdJcSnTrp|vG|G-7eZdhTyCC@?f{KuU$
zsCidO*~L-29;I9uMlUB!DmwFx&!kITCyM)2KS+RTk$?VTje^B8MTVERU~qRn8%A`Q
z(bp7dn#HRJ|2@BVGTdas7E8hBbcw-pM}(bu2_G7|Z2^}>@6-~8A*9p9!2Ju(V!6NQ
zaBs_%r^Z=?sz}-3P)CWqk}_shQ6(&YVcKYopHUX`aKSP;Xd!;p1u#j;cVgf41D)i~
zsfrjX_0`;z6$vv@2}yX}KkbTs?EMwe?U4IXY~>3SunEC0^+I;m0A(bSJq9ax@WUzX
z32>)QOScroTVa<^YMd@HN!JzqQzGW+RG!C#7kkliZ%{3C#&zX#+zqeU)^GWDemdrr
zImAUJ2pwqrh}>KFB`n(AB4oadwoOND{g%^oN}FptQbzNZH_zphvmeWL?50TSmq=R!
zBx3edi#zveI4Mm<jJ~^fvRCuo_UdJu6lHPBS0OX0K(1Fm!*e*mvZtd5vE2CGj17jr
zpj>CrbG7_V)vG$4POQs=eVJ%oFvcaOmhnL?Q@Jb#NZM4|446ZQyu2ow2Bl6$XyN%b
zK<PZF4Nb>bNhfqlgU;AA+6LFn40xjo4FwdTP(W2SoD_>9CSzh_N5E1t#NKY~)ZBT&
z5IgJh&pAS6=!%xbATiB(DjXYla?SnpB`6+M(Y+kCFx}OJ#D<*%31xN*?>d4RK6UR#
z|H2Lf_41xe;<3L3lFpYj{mxo{V9nO;6Ls<D^Au+j&^xCg3NTlE%LWn04KTWiiH>`6
zBWfC-K7&B>QgRHek+Bzz)Znwx&AmdWQ^YcQMNg%!Z9oYTzI^l_;8F_N2I|Aih+COa
zYvitZBLv;26l^>DF<a2YDBjiop%CZncj7$GpeM`Co0l^(&c%Eyj;*c2h@#Q_Q$RdW
zVuJvI1Pq6&KgYu{WmCKhqR|yhf^CmoW%@K%MLo4v;)IFT4onI;=K_uVWuMGoWKN`J
zdJ5)O4SUki`c7)_PRu=mLWVt3Q_8TCFA=W^2Y`-Um}cy~w~c7za}jGvGfc4m-r7il
z7t01R(3&rJ)DdyKEFc9Qcz%5)d3i@rp&M%m&4AlL#WYcN5K4Q{WDLqZd*;lhb?e@A
zRlYpE+cVHPkr4?C=MVYX0_l)108_4Yc(WL4GW_D8j`@=!78o;@QWz5h-q`j`Y|tL3
zDLeNR@B}sL%8tH>M0U7c-S!1;393IHT?345)wtYaLT%kuwV!o=<Z-GyRvmj+CIX{i
zXk2VbUw9gjD(2us8^9LgNHoJq`oJ6Xh?3#%;|K~5FcBp;(!S^H?j$Y=yOlMit7=M*
zzF<Jl%M(!Nn?-D)lWc<H<EC~~RQUEQ88zr{6H~4`odl*N<o3@NC?53YD@BXKBsp(d
zm8c9rKF+H48F-Wd8z{c@{3EmZ6RB3&{&j>!?^ws9`G>;d;$&4c#^wp81@QCDpvN=`
zAgg6C2}?is^d4<IQbJXNb43j+3qK8DmG#HW5(hAThX^-!&ZlUJupx&{rrW)Ud-&rs
z&J73d=FQk(4MeCt#!Ie!TG5503(-&skJEw^ZYY{)7h0!ky?$e~XB;Ws%!gU9+2rS<
zQfT%o$ck7v!1*htCx)mOhyDa<kv;%ol48?ul5!vuPzkfj!=L_yk<Rw%z<fiD08$4Y
z?8I;!s()b43-q>12=h=^hdTdMZX#Q!S+&eRZJfST60p<mJ24bEXJi*Pg2oBqrKGlc
z51k#rT{JgB)J+P2!+vlDYnyt50Ti1Q2Q@s7sRyIt7(@J2jeS?$HdK$JsB8vUl23t!
zmlhI-ZD-ysb9MmmUWuqtAKC%A6RSKfi2lS+>nzzc9K$nFm-<G?>7ed;mG7z6j9fm-
z*Cq(U%Mm=Gd*0sdacS6(L?r+<Me&g}^ue@M;bD0aOGg<jI<(zy12rYCox_*c3k=Xe
za-11a@$4?TS$O@tob|9k%u_V?pB{M#gy^~oH7$4xTR@}w3M!&q(YS|^Da)d{KT%8V
zFASn#8gv&P(g)Nn)T5e#<?;>$UHU~BJKn&}ZGf_yxMeh^hNkTO${Hl)Wx`HHO*?jb
zUy;aj8NIbM9mo<gq#J(w6!?ttT>g0TlNtV3k@fdYz!ACJ_XJuf9M00j<B`;ZwkvZ7
zPN!-EUL1&@(lAb2X3cin4Ud}@sHTXN9-_)#zzG(fo@j-DU8HA+HPPXYmTDV^6C{&Z
zF|-ww3qtVJUQ0{M3`fw5a)<qB$Xh2E%ihlD4;b21fN$kxhFkE-4FYEr8}WQ+6;cy_
zj-H;XNf``4XPE(*jSn9KXds3Gt~>oA<NZ!^Ju&?mEW5#)+>7dx-NOkW9A9A8aj=$h
z&>eNd!ev9d?z~+goc!1)Wx&l;v$Ai-sIc{<^l|hu61$TVN^BG1Fj72N4^=hQ484@C
zD-bM<Qw1dIu#HS>l@aj{l|biKSgIejMNkKDgqki`a&^e0Po_bX>}1(2mj8o*D&&R*
zftjgSPUDQ(Wmu?g1-_zl$>0?uj9d;L2%H@4HLTqW+MAjc2xSnb`AwmxHp3xRWwPfG
z1V&#@vr@71UZZdXs!`&@5$GC1871ws?%A`d0raKdr~y!F)Du`g*W8)z<ruV?o$d{v
z%V3eqDEv5E*mV7q|6-2C3`&4?;;DCv$~w7sjPd>y>Ozh@zM@j6kCFT?ixk{zTCFlv
z9o#Si>myU}7Xx6z@81jn_p4vIv?e7@!ZtnCf-yRdx^nHNMU6wg&Qme~yoiaygPIxL
zMj>yR4lvDsOu3otn`tFGYrSOR;dJjWt)@f#t1a%H9v3{AJgK(ZCQ<mzI~L#>Ta8Ns
zsnw$Tl~`*L06mFEsVb*XNxk;FljqX6mgn};^vizdGQ5@!nNA1H%_W>KbWwiTa539I
z<W?AXzcgqGKz*eA*kOIV9O~I(ERP1jyRnOpqrbaac_%<cWv9uqK=~dw9DTs8EI0!}
z`J@sH>yySwdADvoMnEX&>p>*-02KnXD~>>5&S)$ZX+-Fe$GiGkKyEPrCJF}oXm{^E
zaE|1VW*<g}vJyHhp$NZna38vwb$mZfpVD8k0n0$>vnc6B8VCHpCq5&I><gKBN40F*
zmP2`LPoTr-eGg8VNR0JT-x*Dd(yS#i;gBqm1DAB{+(?XI&Jn~z^m(W2b5V<7=<Nm)
ziXiO1V|V}+C>9@whtGMY+k1Yn&lqBwy>EE?zPv8?>sIb%YrQHpWFPz~duQKUW})PY
z$Sm!XwO2RW8+0l9!E~xNl<y>;N33zVYwWV<<7OJp+*j5p9oXlxS77hD+mcgqOYRku
z!ndT4ZTpy=|IqX&n~7QJt(j4U9b=LuQ)6|j$PXh8B&va%TGQ)DzOfo$F9|!7tt+WT
zj+NXob{Z0MTH4muR^%N8r8vm(1^BEI0F>^=xx3CAMRrVuoOMDK)LT}QfSqzqcQ_eY
zfsTK?Q#j?WQ4yO((Dq)G-9i{wSCGjdT-$>*s16=KQ-Xf7>NJ~>Zr-i!pI>GzJaTkQ
z2#<+9U>ifR3QaB+)E*2yI<l-f0m2|jBVy9M;R1+^&E$9mAweFJ!4n+i<>z+~oNnbQ
zt*)+a7rGfSR_+uWJXh?-A(_5w?BisAg4LL2XzE1QVW-;9eb?B*++`&n=h#%{1lf`w
z?Q#=3*|$Emk&k+xuJh;Er`jhrM)z&+XxapbHKsn_n&$K{U{*qE#^&sBQ&(4}T9D^v
z0f>c~l^qhxhOXK|jRRZlHtjeN=o=s{B;}xecG#t>YpGh$;q_|6E>0VGetRi7s2j>|
z)!r8bBu$=7%~^RyNl_GZGQ)==Sjqz6Ex&imB}I^9<O^wY!dpNj@W69LvJc#=ljuEN
z{1yrjJMJ3QLq6m~<K^|79mJ|MRVgA_7k;gbw4x@BjP@f?HKx>6Vtq&5lmY)q|9**<
z!w*FxvL_=`x<lP^JLR<@{PUdg7e~GOh1nP_UvCAyWy|q2h=)FpOlo!&?UxNGDZjXR
zq(ef>4;vvOC~B+iaV4c^FuU|C2Tt?8+sw5sUnxXswB23&n50E}6!i&0AhgCjW0cEo
zqi%r8?qt}8hc;ryMn*<veom9YPVZFHYn0^SsQ7$VQm+JZqRewnM>RD3OZ`JL!g<F6
zCi_QAo%F1$LbZEd4>dRiEBg1pPCdYR??Bh{59uk|f)yJ5Cq}TMpLKV4@33-%ZcsUy
zH4gO0oUju;98~KCPs5J%x?#a|owHHvM?6CAj+vIFr!DdEYb;HhSTq!LrZKg5hijM1
zJo-vz{qC;JI=Q)sx#ATwV_KT~`=7#c&|diXp+kqdG0$&2bm5v^^UmlI#Q^!|hd+lp
z+?I}Z7<QR_YgEcJ+lJ4m_E%W9Rvee^m{YguURstedEeE#PCwYcnFP?lE#~gE#<pSj
ztg+kF|AyIaR(Gfu&5DvXymcy6?nq&7hh;}B4!<d(({KK??)NctcXuB^Ca`jvc|D@L
ziMmCR$kC%mOWPBI{jF7@3G^~s*GA1;#ChF11#lHe7EBxU5Np*jQVNIa5>lk?-90^9
z8wGtxS$~C(AcS&FMMy(Q2<ULQ8(_*Pcp8s;o_KlckIy*EduweVUp|0BI~Tjyuhc(p
zC<L;iAQh*OFe>?__0}eWA3~P(+2ooAChqRfP+DUTU0XIMHn>4;D92Af@<e+fnE5)$
zA9^QkE+jTs-vor0jb-5z&%>gNfr(?1P&6y@_JJ{C6ZE}SK6=s@3>`f*yjNlk>aK=o
z0&BmDvhoVjuOuBZ)`pPo0@o01WIe49ll?zQUEp5WaJ471&ihZzPzf@&<kQNKRFE1e
zhls9GRR$h0XcBIu-({T?duE57==j!jozX6KwO|{gNS3C|3JK9Fe|xzh?({xs(=?ld
za<+2`R+Igo%bXGw<{JEFQ+qlWIw;9jb-zX<ZYlClO}dQI_wSE%nyWtZc@jJ95>du;
zHc&d+YVvLSu#1MXboBPFr4M_mGg~dyMn*=+HtPb&h&q^hU>K_+A$@{-1aJ9w>{3_o
zU_?;DW?H*TkQN=<C-p#o%)s$^3;NHm1W^IY<+;wxz6x{u?hr-hLTA4IaSFs_mP1)!
z-!!DpZ5Z0ekdh@Q@@*m_XL^O60hf0t4LPkeN2t)`2F?x#lXw|vVA+j+W{r=JC(GUK
z%{RD6tdIU2zFLBfVTg(L6DMi&Vc>Ebc<x2#3=CY6uH_=wXyYs}Lm6h1(ym_g_{I&i
zmSjk?mvP}vb2B8??qJ423ba7}T>!G?zz>5H@4)5y_3Hv<^$z4)b`vMdR{F{8-kI=P
zW;_>?^!2NHPixV~X~Ksr{9;7gv1_!^p<}E?oDHZG18(4W&OxE@G$dpbG=`SqPNJbX
z{1*HI9TbkX6DK7Mv3zO!T!4peOI7Bql_b)Mfp@%7Z}EL3?1K&T8{h{|ghhT*B1BPm
z(-dOOtWJ;5gGCV$DFH37m=PV_JqaBMXMPrEfqv2%nB5OmHZV+1Gt@g#{|j<VX{f2O
zfima@*e@4l6iL3bJGQLh#C*c9Bq|)+hAjGK`e<z!({d{<xz%Q}l4#JAEZ=zG+(h67
zz|bww7IOhcz6OA0ODFVjU`{R&jklSD?wB_2DjIBR>+uWO<@Vi~YB<CEzMWx(bcRX5
z$4^5WD~!C~a$ttdMjoRKH*(ZW)sQp0h<8Tvi!VY#XpriyO6Ft*ov_UiHZvM)x*)|i
zeXTLt$}YFCi@l83vY+LXLy~+KyNzz6ZjeL?TYtj`*2YZZ`?IOo9EDw(Wo1r1Pc!(m
z(vi3N{jLQDObaG-C!0~dgr0y55a}pGGlr8K6fu3Z$I8lzL?4m0uK@eRlLUpyF`;y2
z@1p&JdlEY7U^McV7P6RpJ4_K(kT+q>AXV5wdy)CHZiJs<Pid-%h{!d8=z{Lfr-Gj>
z+K=Q;f7X*R9B2wNg{{{$0e8u%0Z*s;W_4e*LEwpaQ&=y?5VoHDv(or_(~6WhWq&Z3
z^zEB4khVNU_QQJ#uV31a=61e@)vRK6sIOfDX_OmBt@teje+YShro9NevZ1}7Yn>uA
z%?utN?7cl*#KTtj)T315_JJ<;E|wu~d)KZV9}A#L@nJLr*_*}VgU^;va>u-u>}Ht}
z3oJW$=CIJHB5>F-wcf{(SOB|xs`Mw{Hn;Ul)STk5a+CHd=~@~aoI2uC(;0PO-leo1
zy>Vr#7x{Wg!5bj-Mz|)vgGKKLqmy`m_1G`-UtHHdRjgZrI=>ma)HGq5k|&HyEr;!v
z`Tive2!kwLHB1|SNay;^PHsm|T0V2tcTD+D=GranLO#Dg_}zPr&ZQH6(n1o$F8mOp
z%ZsH`2hHz?{7qOU^*}{nGq`IU2Tlw2+PQ>jXxE-bYQ=$1M;80#M`9sHX1B?Vzg{w^
z{xj3WqXlw;$Nh$ln3i2>{4Je}{dA+QXAhiWN}ZNI_IbyhzKp5inzYwX4%(u~e&EV+
zgeln1`~aOYMc!IC%A$cw!<aEc3R|M6cm$ic1!7TfQM9ybmmeiDMl%5UVrtSD$=Qx6
zCbB8RFFh=pt~5{@``wX!cDcG`5+8nRKV+T6%(RbtM(nTqZ0KBC%5xUs*b0h~^N-*V
znVXE7t2LScXvt8`{ciKRCO}ScSx#e6R)xbRC(^h1p^Uo>;GF)rAR0h1yEo)&v*G03
z0?V_V(w(pG91=W!k#6^PBg_sxEdR8<!yY-0>Gu9Pcg@Wb!-`CnVw1ANE;Wx28)?`=
zntxv?tbXJ+-`i}-x-3si%ohAg+t1jcuvKKdNsuU4;AB*LT)^~*Y4~IQbDz!9cA`uR
z5B2-1;|0F?>jgfaQ@|ZtcRpw8^cZ%@4?0}P?4aEPT|SsKLSjGa{e~mDg{(}MJjxEj
z(qa?iMKCd0`f0+F#=CQAe*6c9c>6!2=eL%;QTo~X_578xo9|6FW9p)5myfKMkTVmL
zaInofwn8qZeLp%4<^MSJ6VsZlj5YoJrM}!I8!YbcJ5xvy@Sl(WY|WrIAWN1l|N8j-
z4~Aaa-=Ep{!~X^5lp6&tze}zxFJFd%G58)p&RZV`x<cwKM!d@*m#{1k<A$o8zq@{j
zavTY}fcZRzjf$lza>3dW1YUi0pm?^SKf-JXLuNtmH!Qm={TtJxGs%5RQyqMX{Y@Ug
z3{OSzs`Bz@WAotv6^>d7R92NvVf0@*l%q5o+JA|KlM#o*I;Nec@BTl17cc%Fzp0J;
zpDTv!T3O19${-pOsegGq?&(1$=Xp>d`B}{%*r>VaGb};{3()h@MnOSAOIF)xpsE+p
z-DMcdXcQt8ktH%BA%vFA@HAGEdtypVZEY=iQ9mx|yh~mk=!|QkbdD!2aWhbaMAc?r
z&0F#oWC)6eNA}@xK-JXHZoOwN-vNKwE3=W7-Fx~uD}0u)Y*|A_TO9_S%n0^{5EXg=
zs9WF67C`EVC!)kb%)xB7{JB9BU29eK3`F>pVkRcl<EowiV?{F!{@=Ky|Bn~Te?Q>A
z*!bW1KmqBJES(SZr=6n?V2hIh?U}e<%CnmXDg8zeuWFyg2$tE$%L@|1&`y6ebt1)5
zpY)B3dNE^pb%hUMjsY+ua`O;Dm4SLjpR65sn$^VVAA<`QFRSrXrJ8Ac5{`*BuLIA8
zSG7F4%It~}6t|vvOs-n{H<u|eE@9&;(hvs$Sp?*_yF)sMU~-Fe@a|9xu1}*3yqXs9
z$O|XT;Bu&qvE5R%&HTuhW1}AkeQz-ZstxLbxJ>0*=heTXTvsploX2Fz|NNQ&zW?Yc
z|ApVU08rS@fHW>i_-lG#EbVjJYN$_kJ{Yt*v<31We?Yfw-z5a*bLY<G6+~amLAt*F
zt87_AIdl=s9o_In(53d$FJQ9NAk#VW0_e!i&lO=tm=|e%IwuII6O_WX>FrEAtGK>>
zJErKLfFCR_l|%aLD3ryeAa3+NzM{PREmHc2Wjad8Zh<%Wl<Cl2Gs;&C8sHl+N2U!4
z2Kt8gBBVKvXDn8Mj@ouqs1|mMZLZ@c)|opQzkmP!SI`xTyM*w9BcEcteH9xU{)a#_
zrk&>BzR|m2$@hS^*&&x%QZtSpHQ<ZJfI^^~1fGYVI|TnM4<%2#Jq5}H3jnX!f$uIw
z9{QR;r9@-EM%KZ>K_SQF+_@XT`25`P|1a}_0m|E-dHinOi8eGe)FGQfR>%r=rjW-2
z*mO1uIRQ|;AMU9V<k6YOq!IYlam`YPFd=Z&{oMQuBzEC|%jQw3`Uz8~^q3hx+R_+I
zsdxEF(k-ZCc)uR?j3v`rytPDfDV{-bw=baw;Hd~=b;RtzD?gHkQ<`J<^~Emq*{~LA
z444UlXcX=NQOsq|7z8H|p@0gZu+Y$EpeCMZO6XasBhR@Y%uw93v;a_d;d6YI=1KP0
z=;))juz;%n0Ht-CsOVX%LPAo2WQ=_Md7c2##75Cv4)>^e|9;KB`Ak3ae0}LZZ^6BV
zdx)W23irS(vRqP=q3jF85g@r|tmLtyl60qbWa+JD+DUGA%TpL8ifa-9<8vTs%X@c%
z?g0?XUi}bz(yuLoo(~@+G`;Rc)3zfA?HFMw^B(a7_~%a+E4nSbC0JVnxiYgtuq25+
z!pd|c{9+yBk^O!O(@+ByQH~<KI;3SncXbX#(byQ_Vom$r4Y+cqw}8vF;Yf3mmHc|H
zC^237`l{AvWBI3)Z^v>O8!j?++{)YHXZd0ONmks}Mn}9IrkIfR_uTa~8Ci-(BjN}<
zV03%ScK+A&e5f;#9g0=i*Li6k^9?jvtqt7hML>cykdGX!m&KwfP8fPK=8${7un>wv
zy7pKJF(+^rZj$$Ly-ZBMYZG`0e8<MX4b8B#A#hK8B+>J^p>c_L>#wwfKQ5B!PqSDJ
zNfiH8I+(S=UF?)4D-~S)6U)&>?!aD(6*u5-Hlj9t_3%{f`ZDXEZjJvTC9qhNCkY9c
zw90LO`_}{*x0B~dmxh+44tOTLjbksfm~WC4(}LspmWEQQlc_!*b40Wxf0?yF1^uql
z-D*RbJm*DJlqV^~Hca(lMMa~byIN^J(=x8Zj3VXI4>fpq)YoCJtzu_`_xWgNr^s9E
z`5#c&T{;UyLmNG2_S&&$cNE^4i@+h3MC<XTKfCf)t~z+|;M3@6p#>HAOv7<d?m$G2
zvzHMw$KO@UN_(zh1Whmbccw?ae_QC{D}#FI036i<C<Kzwg+R34a)9WWpTO#W&wdDF
z!ekfVmKgQo#SI|Ra>zv)Upo>GO55+`(Fpp^*|d^x;g&h<dnQ+Rp0AJi4Pm6<SyV<k
z)JH=YuJgudqlhD70^S4D&M>UXr;i^$)|eW1o5wV62tWWIw|4eU?4oEu#s%Wx2#lSP
zt5&Ug0$<veUUVKGW*N}hc_bBR27zF*O*2373)aNf+IJ1ZoD(F0qH}dR9^%~SK$w#g
z=X-VpSKVpddITz8xLJ9S-=%8<;erUPti}>1^+^ytI!~W>b^mLj6YBsMxb~@WorSX;
zeru+8m&l&oyLS)wHAv*rW6qBJ8d$g1R?^lz_}4`|7M@^=hB=4_=2;o`rj{ze8Vfr0
z6Ej+|v2{1!>2XF2hE5L-7<>Nx3(LvGGb2N~SCmc@m|%5;tfM6z3e<ckKZ(F$%ri+c
z_DCwm0<+Qn@V9=5=?Pi9mqGZ&9+Cfc3D_)8uo`+r`xe7tR(H2r)r!BOspN4wuofVi
z9!PDZ%QpPmR35i<jk+4Su)-^CP9by|F?%H(U2G$<<vLCN6-}eEM?9rQPq=Z)NSDy+
z&;E?*HFpfFCMHGVWr15gL)B6MEr`ospXi<r@)m;oRrvcemTW+8ICVfNE%H`TQ6V2>
zhS&Xj_6T6%1h}3pfY|htjEoGX<@DqGAgX%<h;n$%AYqpp8XDhiKHzbcdB;ZZ>rbG_
ze6;)BU*`64Rb-+=u4k644Voxo0lqJX`6tt(hxmOD4Pi5~CXA$FcW~umgQ|C8Q)C_Z
z>zip2(#KBeuYuz0-|%mjHUCq#^gp9ZI7t0xc>4e08Qfi&@@*Bl9>dh!YP4p|TD|0Q
zkjkz`!%q)NZrc&vpMV5Y3YZ2UGX>jR&<V*<Z2;=ux#;I9cYeS|Kwi4>G|o20N9`Ge
zFEgB+xrF#59G|mFm3Ekv1)+lSM6=Gv&W_R3&}r~QmynlCgU0d0hK>S~!3X3$^vS1`
z@$qOo<Y~C5s|;mgF@|qW?V95|a5^yj>L{&%M}oC?*s~M$mSu;K#H%_~R8_4^kyZUv
zPuF#%@UL?jky4b>lRf~dEJE6{l{x4)>M?RCHip|ZQ|-EMTWO~%h&rxh6k}j$xJUy*
z1|OCO{&iVBwWgUfbO7R`Q}mx-u>$(Y5;U3Rc0~zFe(o3$e~UKruLu!FII<Ry^Cg8v
zZL|kHb}y0ImgB+BM&_fiJC;=XdL2byGhzJX_fN=_k5UlPM0>oNX29JZAYg7&OT_Fe
z*~q1#;ZLlGPTUKW?Rf=~K3-wL=&!&#oKd)C<$r!wAt0s}wWdkshjkFEmnoJUDy#;5
zU~h{_W12WkEGG>WOSF)OJm|WKsR`8YA}M<oSR1WA|LQv`YF>ne-5eYo97iXCDjLca
z$P!Bu3G4v$nG&a1*oCy;-WYEe%Vb4I&~h=W@On^SRuWev90r7B6`2HOrWSN9y32*&
zB-ZlcsfR0<$M$v1V9&6?1gyNGf-IAD&{Q0NOvfe&K&*hIjwRZbkFqbR9mL_O4V?#x
zG{nf&lt|Gx(3yy3N+>+Ow%e#aNeBaRtkl;;A6$S`U&qm+mKcd`JK*3jvvQ~nss$c6
zN#qc*X@xOt`M?n2I~lVcmEAIA&rvO);uN0buL<BPR&y+7$dET2{8u7;&J*aD0Chph
z!<nhGlG--gF-2Qu5{phQxvgs>vfGAq-P+CUf)4E#h3%jIghyS$g-p7tNQ|yh8$GUr
zYY?OiQMyXyDhx_?nHBISQW;Z^MQ?`UH<-O2gCg!^tw^4MZieIan4fC_9l+Ypx>o<F
zn%ctwM|w!pUz}y!z-W@`EorQYV#W!R?KD&6ls%R)mQH?UYW3}Y7R4@b+kJd2cOnKy
zRyIwHYh;2Woo$mj)j-)lGBn+7?Qk>woft0DopLhoHu@dw6u<2`w8E>&!pCz6Ys<0`
zqqJmGw+Ybg`?X(7S;mr!gzLyQBBxkC1TiJ2X{Mp6ZNt9fXsS>@LlQm_#IcE7b6oZ<
z<n>l%j%N!tV}Qj_<=0JR(V$+}p54d2j)IOEI+fXnLEW-uZ$WP9s-kJTA!DkcKOuFN
z+NVg{S5lZt9W{)?qS%hMZ411RJdk*_4xcavZ0WK{AZZP`U6C_u&Sk$FR1so>EF#Y&
z4>bFAp9fql^ztxoqzm7-?(YC*eI8pi7iIeg6AS9DW8lX^^PsIQ!&XTEzLXpJd+X)u
zY2U8Yys#EpA9wH7G6Qp<{;Upc7xk+gn<gt0slCf;Y)`IDMB+eOl6{j<KWzwdrdo}F
zFkjpeWg7`TkxjB3euZ-`HE5fi=Ej^_&~W{%BsJm7Rr4{8g4wP0U(-?i@$)GxTVoRo
z#F6&Lw{HZ_eAXkA+Hm(o`0nJsxwhyj?zK$EAKr%S%;K<plQFP(NZcwS;V`>v?L)81
z47^!6ME*ukGCUzKz8GO;B)Ro4ZyJb?a)`*%X|?mvy+iN&FPCThfr;tmRZM9M_D(8o
zUQWH^;vOvPrOhy5&YMhHz;1a&-L~!CNzOzmGQy=oHLM5nh@9j`$-z4ri_|;^Eo@dj
ze=B8jX(V<7xF*N$`uND<qCR6S(4F{-nQC4H%pKtoi^WL^E(fpY_I6*LW(1nUsYOVn
z7_E4J>)K5ShRYYMKELzZ5m29jT+eZ$K+C#Ye?NOg<?0FH4r94%lBRU^mJa`Vh^W;L
zrq1GLx#P0h<E}&DH3_1lb@hEiw<rt|Ur)%qH7^Nh5rriEF$tKt=)i^GWQY}*01SeT
zO2*(l3czG`MW>nQ*h6%d!+L~mH<x02wGdxL2UJxmXyAQUiUt~;K(uwuuqkt)03kfd
zN?yXib#7=_*KC*_HJuYBGc<~><gqGhRQ)Sr$xTd6WOL4balU{B+5$F{8vu{D2$8Dn
zDCO_7vP-aGvcK7tH*#NU5jj1B&g)L?1$5K3artByxV<Mlwi#jGz$z2dvf2B2{L4co
z&4uLF`6a3cGF+Zgc>dW5FUu4po@)g5&~^nezwY9$N*F!N?a9*yEioD#9R_cLxCuI7
z66puNDns=;4kyxFg{ow5C&@Kg9m{I~KwCB%I;&vBd;=A9YSGKWKNhf;HxHs|Eessy
zo^=y>*)dCgc{T3rH12!>Spoz6z|qM^`~b*8dV)`!3kYhZ*2FuNW^jz?s0wB`1t4IC
zaji3}GhD|lZs|dPH6oG|^SR8B?wlJnJ*ucUm`&b7`njB2PQ7BTn;YpVxSWitGGTJc
z?Z!9^087aEQdL6+G((~-h4l!gbUulT54v+xd@%W<U*e5=|J2Bfj5yJm@wJ-4QmXy~
zgVrA+sT*Ew=nDj?b|iD6$h4V=byhQQq-lY1QeU}{uvzJzmDxy~RMl+|44yUEAPwRd
z2n5VmI|CqdC0SzvZHO<LB;&O0@T_ef%bas~SZDnfL|;`6vNy!pTaa%yR|~YUERxjg
z>Q=Yl5g1rCV8^f#m|<VwdZWJ3tN5>Y7Bl#FA9p7Wfq2LD09W%LTE(s4=WFetc6iw3
z#`r{u69;sI%wSj^mqe{1H#CCieBYte0s>46UYoJi4Dq(q5{z6>zpbcB9R1-X5n%K^
z?Po@+a-mIl&-QVzC16+v0|iq7<k`vs;V=}7a_BKp$!{2r3@1%8pReH}u*XA`w-|d2
z;6WwYy5X%RhDE9ud3kR)jo^Hm-a)1wm-_w!^_ZB(oksg4cDUNj&5iT=d;j@9b?3(K
zo1Y&jUF2%Hr|ZDc<Yn`d*L*zu!*5?)FJ07ky?8Wo&AWR~e!4JU_SwE<#UGw8+GqOX
z?`w8$3`xzD>07HM3O}y~quaL9#WfbfLm_wWJ8Zou99Y7H>4BtATcji!RgZ-zZgB2G
z`evn?3ze7<Fylfoy=h%GZ{9qToA%^6W?cR951cU9yOhI@L+6Ez;~IwHOkCTbm27^-
zxmN&l+#?Dp8#iv;2mA9U$ntuIVR;H!yr_z{eHeN=tqPv{nRwOMdaQksdV!WO!&Z*s
zhZX^Te;KW{8q5E?NnE|%el$F57-|4#y4)3t|Gaih+)vhgtWP65CnrUz3KJfSd5-N@
zL`GKM*CYkzv-yJ#xVm1~A8qtqzG6ktmoMkDgmJ|?;74M}Zg~{;N5H%y5jyPRHM=r%
zTMUaMSDFsAst#m}m@nxl5;-~Hfr=a-YfKjo+W<!U`$N0{$_;&6q}Ht2iEuP>atfSl
zYhTByQDdJTb8j(W;SX&o!mu%XP$ucoctF=PRwvVTC`fH2SsQdI>HD|tX!y+(r#%O&
z-NfhHZ~NcgHYvfBNdhz;JvoJ8?(@EV#dSu0ZB~`uTsXNDhEDLUNbF8?oJq+VzH#HH
z#BPfqtZ_`n#;f!cniM<gH9~;6*_u^Lo$L3GBKH6NuZh+36wpX%-EtUwax&8ij0^Pj
z?fl}dh{dS_`#ts0f|Hz4>dDOtRTl-|h=WOk>s@QtuC+rTPOoevQRXd(h&|-yLRv4l
znE5~cxOvT*H8WeGc@s5~O*NKDdIX=~*RvdtGeGAnocs03zPg^RT?(?!>h_<RnJcj;
zA3u7uCUE!@;40>_ha!gUaULMfZ``q?c1weHbSH*xO~#6<;6Bv0B^_&3(=ZIt__{uR
z3+c3ysgynrFw8>UsxXMpU&_kb6`!Gd?P*UNMn*)Jx=7B%VYG?OGf${kt?%o}i0%Wc
z@3$I=E-FtKbU(1_t$c&AX~{TEZfWS1D_6E*d?FpXE7oOrC7my)fr|Pi2tj^t{trK}
z1obzhs$fqj^yHhs_|C>x8WJH=(4#$R5)2~!W~CRmPEP`VBN1n3D@iEYa3ls|jAo0}
zwr%BYDLCm)`6^=~;~umg+EUDSI0cI3+jkVpGok`6N_@ece=|z%1NhyGGspS?>J!IF
zFbGM?$*235MT?F>AuS$5MMVKpYLJSHeh5)>E9M^k(kItQN6~i{1jH`bvY$WU94k6~
z27G<yGPvUnBkz9!^&y2GQhmb#9*q1??*Pu9z^s_puU|8lp`0n@JM=Eus>eIaQYC8B
ze$`a#h|{X4PCWuau-9B;j2FcUh`J#_S8bIVS57-LuJ8Ak*y{gnZ?FEJdz%zE*kv5+
z^)ql7^q^O-3Oy7gHP1;+2ST)1XB1(ea~pv=niPIXP=lsUL%=9t?oo$xR#{>Xj=|j+
zIQoEMg7*8hYu7@+q$izgM@%c=udLpOp!<Ud=Ew0+#;7svBq`QrNz%YQsR4^yw7-CI
zDc$w1UAvb3_8Y1BghCBKl?)Z(W(D)EO<T4MwS0$7Vlh4%&3)(rOU8F$2ujfMaIs%-
zHrPRq#PP^MYyQGz=T)n)BdnmfpEZoZK0}md0%H-luLm#zBx5?H>Mvwnei!h;fJ0!S
zy~Fvp6&89yjC2zRfM3>tx$M`y!T5I>*76bj=6G1ZLJ4X|A=~QJ;pptpNS~&Dop$s&
zkoZ&~#gTOp+Eg^Z)(|;|;krsBKvKN{C5W>JNYKQW2L&^YMj>x+J<7J?<{o+s4@8X(
z#L}@qNtX)oIt=uSO%P&-ew7NU67nIy46ki@pHwk-fylY{JK>EQ^KcX<cb$;ykZ(#*
zjaaqytQ)4?&WJhf+<6J^nIWk16p(e5;o(|ArV3M8XaOA{(PL;|lp|}90-{L6a2@dQ
zo0T@bJyz_0|7)em#AMFQ%>b<bdIF0{<^kzW6~;f5lFolJGPcJLnglK?a_SXRROJ*}
z8v0lPnh|AnWWkh<I)DW057(W<V*2!K_5M)B=dLP^y1FHJl{7i`@l?NL%1hWVVc{Qy
zXtGFQ87>KHv6{Dc;*Lzo!`6m>zm2u;n?MKo`=iweY72xQTcHp!GMxg>T7iTiUAYNo
z(J)fePW&2}v8j+AAsQi*aqg&4l`!`+^uma&6<{UjS(jnn<_Z!{ubHhqwqzt(zlB+~
z%g##TS(Nyn{QFt__&=V7ADv<pWwC%Z3JI0+T5)r8cOyMz9)k2wDDK4$#2MY_6EWqD
z;LSi`!9gSKg3!Sc-C*>_Leb`gu#xxSZU|W^<EVuFM2B)Rlr(5!4^v`!ic^>z8Saig
z$+toZ_53O!?LVOxlCEs3(9%D|Z97(fzl~9k;c(PD^*>_x2F$oYPyH^2%R_N-5b!(5
zQDK3($aH`!m?OG^a<qAy6BOHMiQ(w)g8Z|Yy<!DCZ_A;;Cxz;Qj!rp;(O_UlFR+{*
zV9u0M6XhDH?Sx?VjhQC`!9pF=NAO*(QxCPk05loWIZm{#Oq|p%w?-ETqiM-}RaTxA
z1{D~18v(p#h|h3jUi*w4NrN9~&n|%zK&|1-&<dx3MEhR*p?~h`)!dbte|u`}>@)xL
zW<Kp7QtHC}JwsFs1A0J93c)d$qRsW|-FV|EIO?GrO!DWwO_^HcDL`5>M^n+fCH-e5
z<WFG!9w3#e5ueOyh65<Xg_dD2;|<^z520A38;i3YI!hyh1(}yWqYWlbk_oh+11=8k
zhgO{uE{7qZy?ps|l)JnTyM&}I!&MIB*BEF~Ct2u&BhXRQ!~$nX)*)DQLdSz)MhA_5
z3^s0psyq_gQl!UmhE)Pl9Ca8MQ;Ih@PMQzcZ7R?aQUcNbnzJ+S_UfVBRV2%fHDrKA
z-<26lfsW)Tpg{tqkGr>TKf$Kpm6q0|akBh7JXT*m{xO4k6|%)gyqm7K8`8oyE%s;q
zwy>vpUm3)!&NQI$OCP1xiIv~D0I%4|kP;$3ef=h1ze@iOqUmb59hEgTFfqO2x;)$W
z&)5F>lkS55v0uJ@VaxpgKK<=keEaWz&*a;a_~*a>J(GW)#kW8G_e{P$iGTk4|IahI
zcFM`-ub=tl9uAcd)Y`KQXEl6w@kLPjYhh%3+74AS)X@Ch$(rqz6g+@Ps4O!TjD#wo
zuiOhJCR=6X{1p(%bNXf&{3AJmTMAd8RdC0F9pAb$1!E;!@eQ|l=1-=6t#SXkCHm7X
z_Ww}@ef#38W&eHp+q3xg-~XP;w<qz>fB*eV9JfGK7gg~yr${Xv<N~kdu!M~*BK;j;
z>)Wb4gWMxptpt;67cFI(BrA|M$Tw90N9smOGu!u|?uOLP%C>vHJ0+Q7iq!@_qSCPJ
znd1QdlHuv`(@tw0zx?_n&UxmZX@he+L7Tbq&sJh$y1)&~BRFiR2F)Nx`G6vQ`aRdH
zDFLmFaHyD&5-p|CG=7f6Ys32W8@Fv!g0{`(V?k~t+)F7Q#t5N{9hlpzk#2{|FyaI1
zQ?f26oh%?ADHa_bP>~JZb!}k?w3W!Wj6__(JyJ&5WE0W{X;A~{D+59})MV39%nmCJ
zMaYSaCY98<SI={z0vh%HBV0}o=u11G$PZ5hE1+7-Hy+t}ZQ*i72;eE9%?bME5lBHC
z!F@=YlvEAy2)_NH`Mbg^7ntNoSQhlBa{%M1D26ofJ@~bNEfk8%G8D2kC`y7+C5J%g
zY8Iw&lGmq%UZp1W&trNe`S*Sv&x1riMA&+Vdw~LVA}1fv@nmWhA)hRC3Agq-0qMLa
z<K&oS4yd6LB(%k4Sy@?DFvx<rOuarQCBe9{+Cw)X?*{zYzw&ZD7Yt5NO}wWWeWdL*
z7oo3WkSKrT$gQGz>mmuyBe`Bs&1%zg`T^UNBHtW=`jr7HSHfBrt`rCdbA^m?7XI>!
zRdNv!2*c^|0dmPXO<JxINH%-${3$%5Y55EaH)K*n@}j6ex?j3O){#Ud;VTeVlc}5k
zJNlD`2IVij3T8%l4(oWzPG6o1R}P~$0REROrZ1Es1k-R4Z<b1J2u?$xIs$rTiVU9?
z=rLm;E(DUi0`Wfnf4zh0wStYFt=BG`+ACqxR}Phm$E1g0U8bY0O$O7TKK3AmRmM%j
zYA}jq5%3P*wKsmB7skT2%euhBo|*{yuP%Mp*lnHyfv8&l@wX0~Oib;|I{TUb+EkZ#
zr01qB7~2Rc(VzMSzN^`chYRiNytqu1>?a+o*Q{Op+N9wXdfZs2{OBL;Cwp`3Br-C}
zay}zj5p;-+3Z75%`1WEiMq<<|4E>LkixV*i&oFaHz5p6P%GV|`=lJ*INvmMcAr7KN
z!M2c{H?f%m5ZPc-#d(p{etq*2VF{>SlW`52@hl3@%yj|%89>xodg}95pFusF>HH$0
z%NLlO^25fslEpcE524^gzdD=%6NP8Oo5{xSI8f$X{)8-+MVa$j72JM#%^B09$tPjf
zw1DZ7_}>Q_6O#-u8Sijq!=MuI1zHs+07&)#{JTFo)>S5l{d%&m34g^MhwxR+8!td{
zQVvCPAYBEgGa#-ag@!Af;CFe~rlK?>ahfrE+e{dYka_g|QGjNR5zb~a*;UbU>t<g~
z!BLU$1b`p>YPA+3v-^@IxPM*3)Da`W0fJdw-<;4#DO6)eAxTYkI1gtabyfz(8d-Uh
zh9}lBZA0*w7|Ri2lU=s7R0#^90cC4Asc~dg|9Z>9w{7k+6I1Jkf1t26KG7%msN8LU
z6!E0(b9eYarK8oT8M5e$4+yLcT?cB><|>7TUiybd1Pu|Wj0aD+Wds9;7&;(xaf5_5
zEoxJM!ZVrh9gB&OQfV%cN*1D9sA?Oza7q+xl=ncjJ|tefdQ}9JYDATNpmHQo#QS~2
z0g8dJMPo==1M=*Gl#$)+L`PWxpa5cz<fcDfxU=syu7Bt}UewOMH@e?gFHGk5vGlsY
zmb*V%6>^NM_y$xZupcWx-%VHM{1}2JoB?>0ljuG#gQ*h~RY~*%ls85A!mjq7LD$6+
zK{9KYR#gz>w)at>R|t>{!M^AQU1bs+22yv*5{3?hrJW=U#*SXNa3N?0q4`~Dn4_M`
zXRjMFFQb6^*`c>P0`4#^6$;Z3c(;NA5tKwvU=sXB)*xYC#Q{=1$`3Q#WX4)^aK>C>
z{cXw1Q?I`#xlsp+?b}b##iPx?3=sNDBlIicDTV;|6`fkJnC0VW<{_pvLZUw;g(1As
zF>%l~nZoMozr!liPJykajgqNTh^*Ze)`&G^g%(xmQqRxFCl7(q_vi$ML2btX^T^~m
zf#`i4jEJbjD%g0U%7vE2;LS$69)3Ci0OP5gJNJUbJ`w58A<=^`g=lyS+8nh>W=~H~
zqTv3ej64|9F$x+A@mxBa7B1g>4s;|^7CwC+b~p_g4t1Z*=zKxAEs1LaYzWn7=7Py%
z*4Z2`<5%aFvaq}WkU=I@q;1bF@B4_356lR>3zW%x^XAPkx@3eol4K8Xr1qB?KsmCW
zTgP(E*56^%`F?<D_U_7lxxRe&fmeKJKWqd3{m=me{EEO4>qGV6enE%g<R}oF5kN8a
z&pnw6uSKsNKWJTlbELmXOF66};xHUJvRo8Q(*~}WSu-B1`rnVW^>-5S)`W&k1^lfL
z0v0Y=qJn)`0h#le4w<=p^BJ_I2zm*y?-R!x+)YO*FbdFbPNzeYfbMIh$2uDvR%Gi~
z9<3v{M^29SFQge7>!i01e+@Y-fa5@OylV%tlclk(3Dm^1ArG!kpFYt$hbUYaqS`qe
zQj^f#i77YDPpt9+>;&olM|2>e2fP`^mw0Z_mldK5CNn*pIk(z$^17Ovn+a~OKGXw^
zoT@stOvm(tOPDSc0RPzr-8e{FDU!6YtE=L^eK*hw4Z_fP2=KiJ!e?Bev-9!JC@94v
z0R6)NwG&28uN-(Z{Xe8Bhj1npYJ*gV0Ksw`0f@SL&<i+<g#@S9NPwwug3_D>MuT?#
zFM^z2He?4y_6H-l9RqajyW!QuVgFf(4*IZOnDHEunVXI!=SVEzywi=)x9S4QZ0*~D
zBjURLUZJVRADHb4r2<}@tkE%pSgUI@hm1rwUe({b#H5DZ{@G`DnYRIcZ1x6~;UP$s
z!f>c64E;ENzCL~-^U&MeTnYJ-;&F@2`ejQ8?IdsA_*q<>d!Ox)W2<BGCJx_xG|~-*
z8sHThNk$*mQr~^|9c2u7$REZ??)cLWbM>xf+#M`><{oC;Q!H%FnuUQlUJ5m|94DS5
z#nBauK+@J-`qcUK$-XHjrY)Ozzn-<X3giQ&8#d>6>4D$n*xLcz-gvSMyD|6-=`<eW
zhLoBeT78B{Qv5kgL1hbvf*VJ?S=WaRJ!m|%znuN}@l+2os4hh|V!hcVlOB!0Oa^wp
z6&iq|#vGCtwj2w<(VSw5dE{w0)0ihO**l9)v?>=Dmxe(U4vk8H$f^)w|D~tE!@@XT
zH3FDsEXds-R{i?xN5v-O8b(0oLd)!8TXv*00a(@r*C_H}?t56&^x$Z32yOB~qv1?q
zAi&1K7a2I~Y9gW_qaaW<00c03gq>jb3hn6Xh%kKidK>aijlMFho(RNkvoI~p?ZnMu
zRR~lxVbx^WVHbRG5~QX^^;v-T;mFjjCz8+}(=n!mDi@Ia=^ixU4bfpSPj>fvvr&4%
z{P`1U#y7LI-C<@joep2p4=;^?T7LHadGX@7M#Q0)>PQuPy?N0R6XEAC(ho#~QJdxA
z*yoy-(*D;zX}{TrSNg4Vx*?S;GpY+XJN)wH%hz*Lpwl0f*r7k2tYzZ*v#kiniHNs<
zrv19qyk{nF=HhQl+$Oi-VDs!AN`Uy?jx=sG2YdSkR6)ErS4JLmXF@UngjSb}8C7q|
zN)vZ*yJp5+vZc5+ZO&~gtBz6UoOJ!oHon=Em9Z!IuD3|s4hKbh*v?*jB%@>{sutY~
z;kb!Gn=vM)FD$0tder7Gahx#}@1Pd=0D@!_)M2SJu1~W)c35)Y&$4^xKzSB{)?PYP
zJPsT@Xp!n)Z-l`CJLqm`Xf7W{nobr>q)BVema)AGu~HG;8cM-@RASVlwXbo2M2Hll
z=K1v4_}~sY3&|0)w;{Fa`84V%QnidH!MX`3K}KA<bZNK*HHAiKQxoi#3~ijyh)Gv;
zI(4dDhJ6}Q7spssnxfNQh<zQ0C|Cs@I5-d@V!(4ff-@NoCeQeyDtGAS{R&S}S>!xs
zzWr#OV#nU6T5gLLa%D)ZghKJ*0M0k_WKL*tRADY+7+1y?8FaJo`|rP3f;WBEFm%S)
zf<Q1T?sL4E{RF3b^&yIMpG087tCrgI)QwCM(Q&Lfv~@CI>Q0o7=iT%H=V~2Z!R!E#
z%nGnf4=q}_kPIo)YBiOcOJ>ZE$>$jk>|DpKd)G&Mm$k5t<d&O|vh4xb-&Kcxyy(TN
zAz{e4O=HuMO7T_r+`$W8-6CJymS7h2nL|e#ulc;-kNf`@d+!-m_0@fgk~hV~M2!VS
z5V3-Ypn!m&fKgFY5T%0@r3wfH5Rh(SVxdcsDyUSc(mQCBUR0_yi69^#5ITYz?wH&1
zFZZ7NJm=o~;XLQQen=vL&ED&`)|zvUIp!Fe&}hcS)Dz=JXR4v3K%<m$Z-=ixw^*-i
z0F~oBh2}$t50f`8w9F2pxts4alY+ev4N*sM9Ck=9;zmlq5zKEYLkdnMBD&v$NOYP3
zm|w!piu&>bA}Z;Z$#anEB_x&$?L5Uh5wt}x$2B?p@Zlb0SaR5{Dv;HOY9je=lGOmn
z#*Sm9>$Dg22N=#SDPem}?n4M-fV2Q{lR=VjTPHTzlI0toqg4)m#-Edgh|W!*?{y58
zq^3k+IR??4l6Tw0y=~i(-9?jr<nuzwD*R&d(SV}ALdjRK2u|AJ5N(Y|c9m<pA@4Ct
z;NcNVoNu*3H;{VQtX+HCWD$L~<bZ%RshL(jrFv)ZnLe6u$(}q(Lgjnm(d9FA8K4!p
z$^}hO`vKyzvFGy<6Un5DTw;5qXc>{f+P{A-Aq(NWYbIb;i^*}I(S)k+qL?`dW^*7_
zaZ!T!DW!%fV<NglZM9tN?1Rl&7bc53VJKH^Yp_nYd?W9C_lmS~bqgO!&<WWgd3@dx
z8M<<`?8?-fA(Ic4J92PEtXZ*5@As{FlH9XbTf2YLxY(XuYgB-xL>$+>J?ISXBI~ME
ze#HzBNQZD7H4=p5&V^SBmW60N+2^nCxvaHY=O-L8VL1*saU-vKGbKuy$)*+mwD6^U
zD%BL^|8K%a=)gV%U5ieTEzXCaz4h|VXj@idyZIm2u4%%(NMGFuFHfp`?Qf(VC*#jA
z&=iy8sP4k}y0~(eXK2z~oTHY?lfv~r^YAqi$m#J00WLacB^%oX2Ka8_xAV|Ev&EKe
z#s$vSSLtozg1}^ku46LFIJ>rfsLW`RMWj+mlzy%E&|Adp=6`*=G`gSMxq&=%F89$#
zf4!OR4<X1B3D(RbUbIdw2CC^;sppVQy`k7qBwN+P`N+&o*d}OX!&-8^4`bmOqMivv
zf^$1F!3H!=U`7^OxG(R2SB^qL+r2-<mGXus*|_(5nz9jXjEnGoJF_85r=6OG-ruua
zXB|fJxyQfT`E1^?qoUv3vNgT*AA7b98!kfI%U0bLFh}!0|NOJ-P9k)rhQJx?RV|T*
zp=g0E{7cmN*Pc{$-LBJ{{8B%w&vW+hyUMd^{pdb71Aov!+G1k{=KM0B2M{W41~t&3
zh>=tlIp2FN*tezox26-gQv^4Qb@_BWYae?QuM6IjU4ek^TO>eF%pRg6<~rO36bbfR
zNX=3k&%%i+%#nl;3*KALvHjR1H*!9_a%!PJ7FP_re!Av*4(sXHIyEI9tafLkxpuvu
z{GxO$g?YMW-v0%_<tLvU%nL1yY-C~C*i*|Kv;sW4eC^La-$kyUh>Po*0~oYX<r|%L
zxj13XEmD6bf=41osjdxBRG~57$zha<VZ(**aN7?^L;qJ9rbD?6LeOmrh3O|Uyguvw
z2q;jyZv;$3#=&vJ*RNkULo%`xabvESJm&?b7fK3UIOSnZgscA6WLMztpkq6s(f{Z?
z7kZ{~xap!TcSCO;h5)}#{cK$bXoDj*FQfZ`FchoKz^F8q0THGY<h#9rqT~s<Uf75&
zHfX2H3G%){E};O;m;tc&i2bm1&5m>GMxe?88k@>h0LGF|7{t@3N9e?-7LL20e}%u$
z)KS9%OzY(%4zxbhLLW2h(tslQO5^Gxy;r2(2D#}BC_lu(Lo?rP54T>H<`m>{*<$U{
z!NfA!s~-a=B-64>ia-%0ig30<o!SN$a|rZ~sOF)a`?3?!5o2|>Rl}~cCr}4f&LI5w
z&+9*<%Z`qM0%{}`0C7|lpcu_Q8;j4`mTu^QyJNJe_%ceL-D01vm_s^9SN~k!^IOQ%
zva+^|=*qV8O{pd*R~qdX5UAsu>idTBjPOcZPvO?`mp{+wJruDWRDP0K!WgzB?ulA^
zPw@(tD}`4pzif$S3@V@&s5ha^C3-PDARrnY`rMAAC~Wp3LdU(QSU0icJ<=PH+$wwR
ztP|(CmhV7f^{fq-z_1+N{}dvAtYy0s+oKyfsZm%|e(n6dF7pUds`%(ygVkU%#vlqE
zSA+`(2}cK2^?rvuFS|e*wJ>0-an1dB{U-2pgok~+yqXYh;DLpZyG&me1zi@9CMw^6
z=&63HQ(>2M62#Oy|N44((V+*=p0bVugU<k%>;(iQIr;qXQt*`U+LIUzd5arNbk_v%
zGNU#z{j#4ri=0|47LXS}uvMFBW=);wKmK_8+memXaiL6sx4iyTcL?^=v2ug8IIhi<
z35)SVIQ#Y2ax*X)NhjlTlt#S!_PvY|!qaD$xD-!VAyW)ls>^2a>YwY^yL2%&7wGxC
zna7PGsP@hT5s%1$V9^i{N+-4J>=?X;eqKGUNTrC8#}CB@WtY!!<6UL6W{c%LT8$8!
z-CB%<Ty1bltSn>$o~{TpnY`g7_#qM~vI#&Hy9Q9l$p6`?NS<nriv9ml(Vl}tVBty9
zTQbfVuH$Nj?A$>8;U46>gBaJFvx6n#^hoEB_5M6+-q*Xm0xb|8#BcrW(tETDx))e)
zB6YrxD=kmCuD#_2kd}+8w_M=sq&0Yb8!IH}aXL<Qy+R(-=?uNZE-)_}q{*wQ+I-gI
z#grOo)=#d>9sta+Cv0dj0tl0_W=Q912(k*wabN%@R;6v!0lH{SFr9dk4(!~vU^n3x
zcnZU@;&T?jh$2I#AsybfXA#i-GOpWXv5p{WEM|_<CEu#&+W8`*X@nBu=-4$NHOvIk
zLbSM3Lxg&XnuqRGO^T=SLxEI1R6HWw%8#->a{(Z-4&3JqE`OW(T3_#tmnLRR`)8`_
zUT%Fay705p#onox8Mt9_gqwI)6gu+#=hJx90Ri7#1IKaOwrwvjI+^`TuFMfLKLde|
z3@<D0ayH|tupYTq@v^6M7H6fwIv_{E&P*g%#5PPECP?z&t1n2`rX^-zo1A<PTbJ|n
zuvBTfCy3%;qY=oLodQ9%$&@iJOkmLoi=>LQcge7JAZI_ksh)+4_b7^6>K>bs4qgS;
z2OyMut&(kF*ZB$CRc#I5)pOW+X3dV%kA}<%ra@Yp(|#Rc-r&IqLYNZKI?m@uEzt{)
z0b9Fuc%cxIWJ{MWO%F4g{M$R={k%OQKBuH(qs6PYZ{NzTcwiVXR$!%okMuhFpzu5<
z&+j4ju*WKdiOQpoVW+8p;||VE0$GeAg`bJvjR^1pN0jh_&ODp)UTg2-f|*qHQOnl#
zP9sf29ylgxyHm>}@Z;oHHi#q!_*#)-Tu!&b(Y(#vZvroGTDjWu%~bH;=0@I2jQqy#
z%>b)+ih1SPFNDL%H#Nu6Bj+txcd0DKnt4tt93*D+`hJ}nRDlBeqdlFk!kW^5pNFW$
z_{J}k4RLiqaN1QtV=F9BwAki(0Sn0sr%5o;cI1$>?0_1g!AaTXGLINY3jJhOZ#()*
zo4OVNZaQ5#jvp<SYC&i#06|oe=tks&LG=gn2_Wn*ozWmTuT<A`{Q|<_9Z4rqMt@_Y
zGCc@T9|t1u-n~1iaPHi>x&BvH&e-(>y)fb)wC=J-BEGF)9%CA?e+4QabZ7T*f+UEa
zYE|T|s=}|965cR+HEIcvExn)AqD3$+u=+_R@scx{0UTBt9DS0gTw_PimQmm_SC`@|
z87-tP7KZH`(ZJOYHpWj89qZ(j7nigS^7PCw)mXXp(D60SzdUvyRRbZbL?auSIpjw?
zIMft1F?QQ|ckQa#Q+vm)+J(PB-{(yj-E{WvG{6>YSR!<1%nm6s#jjPABDd-vQeC;C
zh8W7C?GdLQJ=>u#3Cw8hMA1SSCBnjjbpa}#%NXls^n%fXe*L>B+9%OKkVgaUsIjzZ
zRPQq>%=fDJ+~LrzK=gVG0n`cf!>|5ftsVGp)rK~SAo!Bo0EFcpwsOJL1RQ;n3g6At
zNucHr5;rpd1R)DIMwaT{8N&SzU)2ajzz<Yq;9n<*DuudI%W+Tng{*c@JTXJ(J=*~#
z^y#Om*GsD=iZJwja#WW+`@u71wL*%x+Bblhu!JGganaEia1(88aq6vNeX>_=<1DV9
z;Pav2Amar)1{KhG`>CPBfejqUI@mwGers>U4RRY;cVT1?FGlolA}CQ}o3I;1N&Mwu
z*jli$IQQGVh0<aepsXlhLBP$l(2ZyyVim|=1-+=M;&7-7oDUow?|(L`$l1F2^CPj+
zJ>@A0q)un9WFEXs@Wtu@=48RBi<FI*{_um~t^#X}@B!T3m^Cy!tKWF-R`{yb<0mLC
z!9+&zG1rYHeg1;n;ZiPbd{aiwhkX+Lb*&|sA?c=7Y!d3;lH7H|ImoTR3U1%idDb$f
zo0)BRea<$nRN@%qhPT`~uIX@`KyRy_qqNrO6g-@N{uj=Crroe+5iT9T&*akd;knXT
zIPt3Cb|%E#N$Yy4o)vE-a*U*uNAE10@?=nxz8jh8y?OvYtcq_F9$rKqKt4S0lRDo8
zSZyDf|MuH&4fPc~^X49D=-4slY~iHZ_7A@z&fFY^J_$k&Vj3#FPI!f0lW_PRjK9a=
z9&3}&&`j0@I&yNDBl+&G^=@5^$6RU0s^%$yvqiw)2`Vy-3GU+{GL6o#saRL8Jg;tq
zD#s$H$75Ds!Q#dbPrvTMwYB`|*EUtNMji;c#(8x(m7IX;T=s9;^vKNe`9BEk;Rx*F
zWg(t|qHj0=5((%!kNpYEemJ%X4G)z6b=DjmGY}P=`O?4IlLroXLHG=2-$mPzlIskg
z#7(R=G?7i}Zj>8QGmiSvP+)Mxyq9qD6(^7lZQnls1zorqC50GVKhV^vHY%;BR0!Wj
zXg^}bJ&5u^9HY{xw37h%sA+6}R({RBjDFOnFDGAgD(GltA!|&xo54^<Y3~xh1A_4H
zl%t*}n3uh0T)b6R*cpDSd#ryiHZz^@cwLR&>cBS)#>3ksV2qw2mLfnc(SiXWf6q%W
zPDHNL$&RCLS45(4@rv&PT%r9W6ZcbY;{p=DpBg-iM;HE1eYw>^41u{Lh(liijl{ir
zPrrXMiKm~h5}DfVBl#sp9Js}y|GOA2n?!V*MB5Q0#KL_?6EJns+Q%=1zINt&w})c0
zW`xnT2-X>RT5+Qgr<<qGFiriZE+GMj{$yoZ^K;9;G}7ei?p!CLqEUu6;HQH3fNmPx
z?2LO`khkQ8w?q;~jzdU|Wvpmm3Nj}|YPwraJkv+!0d_yA%5yP3cyul3KsV&fqX0s^
zLqCDm#h>tErafenvW*vTFmvGp1+*m?{pd9B)4JCnhaQcAy_>0N!V~F2WAa6A1QaPZ
zkLa3;ygO$I7kBviHe`+AQhR$YH1&WcJ8xck?pQgByu@9SNmH0XH|V+bW*2}Q6<`9`
z4Vw!Fz<0)fhrQE@sxvTDF+Ro@Q=mDD7x}3Hch_5sAm=jOv}5BIA2v@1^Gw6Uwt5e`
zPKg5227}cfce%s%jj~U_5ib44a?{4(vp|L~zEWq0?al1%rD~Xo?lLMsWpvHKrkz>L
z>WA|RGxz$WZih!m3u-K5wyo5CLc_`Ufh!X>U-2b6bCV%OLAX7j2GMo|oy@S^poB8|
z(N#x!lV~|!VS@CRnGPXK?DNS2=wdfC10N(Y%TtTu=sN=Gvme;}FNJ5n;p(6P<N@ik
zyVjxvd==%|D}sIsfH(}ky0lFLph<o7sK!n8AA_bI428ra6Sy`2RLx4=6*ty36RLp8
z2MOz%h;3@u09f@t_zC$cx;T9VwVUD`!3QZNQ@?yUne|B6kA!zlf_XIn1_H(}ybdIE
zy*SYX5k(tyumFlSHaSU0jhf>oTf@mYj63^rB-UC7Yn{uz`)LHKF}7HcWZR1(gvM<@
zr15+SritT+<SKS{f3=Mwn`k>8;KXe5F@RI)4o^;FVghaPi^0-4UTJ&k0#FeO*1h>&
z|L|gIQLqDLvwO@TrmPVvTh62ed`Uqcy&)cT{nrQ4=R!xraL(@X<y<Nr#DJF=Hn<g|
ztB-r>PIDwExp<~|&wCcDa_f`hRtP$fCyaSEWuZZ?9VM&G@dK<Fj#12`tbw_gvrHPw
z`jU4eXPS=Jb&!YC?HX2AL$TTmS<03Vv<tGy5jtybMECsuC~S(Be+n`XBCfBfxooDQ
zt{@337P}c99v(b+9f6{>x7aPzi9&xq3KPsSc<iDzVW8h%xx!Kqd3~c}MNdz~58%Kj
zV-|}&g<Ci76sI|SdT|maa=n&py!>d7Ij-t!iC>qOm!lrhz>_F}jVmV2fWO4~OQuhB
z%nMJ;zCuT7qrSM{^e$i&q+?7S`Jn1|7{)<l-Xmw<J0HS1R=!aI4a!G0Prt#pa?c*N
z@L2NC9n5978XbCu&!0es+fn&s**9T#4_64yy<L0%@j$#PMe<{`!C3F&4W~5E@$n+{
zqnNRLj({>83ru^LQq83$GtgHHUa?pYo*n(+wucW6gG6T=oXP8=cIQoC*ox}v!JUQZ
ze&)35WM2$vn3IMfV2*0vPb*gBwAvXP>gnb$KK|+W&eM0V=+51ec3YT?M3>KeW1~<y
zrWJhD`lhCBfnaS1puy|n1Vl>{Htxq<s?9uATRP#`{j5C#GcT`aUTH(6YERM-Hil_i
z<^#y~rXpMC-M(E0idQc%E<k<&Y$=!F#c3F(QouM5{VCC|l6h<i8C<=MsMoWAn-Db$
zIVcE+pmF<BK@I)6<A@|@u#+-+?@flA0$srK^#^@KU#tii0U;w;xxg4ZDglGhKNca%
zIx}4A=>_2FV|NS!E<A&_@FlE~&tOp(^7G*lH^`Q2?Tozh0X&n@QZ7A(5qC*s`%O5N
zDX4164wzl^VlXj>z=+Vs9#g?JOWl18n7Ck?n9PykzH}dVH%D_TI=k1+jnr@;2>_<-
z_K=0g53OF8ugn84yZ{_hbRMxV1{8?qWJjoMiG(fMA3J)~6u8^+>NnYfYs*3kJy0dq
zNYpXwL!;ie$;wngOM2g8hMiMi!p>c4<RjC&HUL!Z1=;`&IVcCFAl$dUk3N<%xCN8L
zDCF3RI}iI25b6a!PO{XoW5?QIuzdbE>ZBloD8n$Rp%4I9$iiK-%!Kq*NoY9_NMo_j
zq8)H4*bNB~9;w>xw#jK*^7BOlb+?NXC`43noODUkOvdN(M+s_{Ydr+ogt@9IhG*`c
zw4D!ovCRu<%3R?v(E>1ptMeccjeTl7op%J^i8xBQGx=~E<J8v!5Hg%SeRZ2W423a+
zr85v2NBwTHKH9@Tna?h69sPOOZ^@EGBAnrRfW1#5qLM`TH-B5MFA<w@v*;=|&ymB$
zdp^W-Mf9r)nEC7w5ovlz>Q;&d2GQq>)=M8<5ek>}<KW$hT<iw|POjOMK>PasjI$+2
z2{g!bKUFtEyB6&?A-NJB<lJy*PYdgdS6JwNhKyr3G7f9wcekPuNruo>rx7Sze*E+T
zc>6?O!1Uc+T4prEY%V`rixDq6{O9+#vbqUX$qQ~yefssyVBfR)hja?KP!W}Wov|MK
z2{eMd$ze1-sFVbnV|zJSwjR1~jer8>EXBxtie@IS8`|_x<SawmS=)l$xN1a@yX;!-
zx{0`V`Rl6R0<fRMRI|BQ^1<K!H3w6+kui*I{};B?o;>3ShofkCkfk*US<#pW;oZFX
z7b*nkdLjcr>de01Iz`ANd5Ju%LkWqjjK^gjc@0_F(>W)g06-vhcR*Q;!yU}@3&Ok)
z)<k>69ves{^TLbCVKCZEU>)f0<)wlL<W%y_lIhW|R65{rMmvqbHwE<-8l)7pJiO2K
zLoPAAz_ZXrmr!<hdx8p|3M7QhT&#|v*fv+~b|#uZiuQhNvQq$xPXk&YiUSdGfDox+
zgx?lOtB;z+RH>~pkZ4k3aoG<p8!|{`?CU8gwGZLw5MTuE0y$j*x*q@}2qU~F%2;F)
zKL_*d4wBeo<pD}W?^?Qid0i&KPS@O=XJVmiu>&)>e6~=jJOey1Q*Dq2(|~Xc2Ii+@
zfviMVXlyaYM5$`5%3DcRcJ>4e{uyXWzKTHF^3qEI$EgKF`Q3pFm;FmnD;H6%9EWg4
z4z)Ys6jM^5r$(J;UdpE493V!*3v5|)Zb3BTE&$YqlF)VZPq%@)W$EtWp@dE;z`A(K
z^3j?hQYIkC03CrndrosX|MMG>XfkU%M9cnSR@ZsgaNm4zkuEC=w(0P2=mN~TtV<RW
z0McP>5od{Zb;a#J)fYt5Lb@1$$`v{v_fcr@!6X*+6q<<w6Us1Sr#{B?Co#z`o@@{P
zl_MtA>vMK5V5(fQDq*~kYgSk-IAl*p*;m|br9TGUcz%Qb#>#Kj_@4Wk^L%cyp=V=r
zFnghx;h&}ty?BFmWqOny*}61<?Iydwz`;Gs*q3E#yKg$qyGcNVd+k?wo>%TRNV}F8
zl-9WPg^L>-xk;G~OLqj>OJ_W*3>)rQ7?~K-Ee9p5$3OCe0s4t$uzXpITg{Vt93-WO
z#yP29aJSqE3R(v|lfjTK!6YB4P`<^T_olz!?VY?Y_+F;txl`gO3zKw;BvL@^oUJjI
znn=?<{I;vaP0MLGYYY|Ug6krhC@$~5^2%<;QPs5k0nm(#9MnrsNi!yQ?AnDWK0e#g
zFSQO`$qUo~BpJ$F$l7RLD!%06gVn>`ku>`<?5eKgK+J4?Q1KcBBG*ep&HO;zNTh{q
zVd4`K+yPybg8lj8eo2%h1gw2<%{S%^1JstAfA550slVH8!?ucviMjrBmFuZUi&jIo
zAbxi@EGfNlSH?!iDpyoi-X!dbWG}e?G#ykrefno~{({I2g*VAt+72DqGCW-p`blPw
z%gx;fj!h*AY4}f5cp21^#*A4-ZEf-5A3+0oJX>rheDdPXb6Snye4jjkNlRKpxgf>V
z>AgIcqTg@&EBtw|_dXtjg6X|n+s@$9A?J@Q*MVPglu1*GP4uEViIN~!C1q4Wbg9*|
zFKTSq_v_#K`pyV}mjA4~w4m-ZnD$m!#7j!)$z%xo?SXm089gtX-M2N7I(y*rMAw<W
zfB)WUqE;-&O6I|9U=2dmi|v#j9sG+tW)~P;i;`rC_1#SMvu8fI)+9-~2ud*W?Dn{(
z1uy^e_jfaVcvfiUlmqeaOv>N7WdY0?DU1-dzw4=oQMkz%nQ--#IPW0=3xL^c)~wl^
zKS~=7Rb9eFa<=@fl+iOso}v%ViRgS2sESx#L`O$Q@SgLB4&VZRf&IbF)uQv^_18LX
z7r^~7w_X5g1{5Zl6J_8AxR1JwI+1=ksd#1}Gqk%%Qa-d!s*=He%Q0O{;vz4htgD5s
zQY(?2l(dT;<;y%9;|`?BR&6G6S@G9m7c4z*pr$M*QTN&IbtP>0$yy0fXr4oPu?G`@
zz2&YkcQ-dRHKn1&q&Z|`W7C5a!o2zYP$Dm<kmSsfXp05Gna1Xo{Sl<rG7uZ42hf>W
z*#{}5UaI?{?`o_d5B>i?rZ;uYd>?(()z!6w{ANJsDM#}r%vnoEM~B4Y(*Rd<GqyBw
znJP%Gb%||kFiz0*o0khI=Caw{yl+PD!30TzSi$61A_skKNz%;9=HKxx5}$*8S?{`3
zDaXCCJO&Bc80^GKA=1Q&Vd`_}tn-7lm|TGeimypx1eQzpL-)EK*dCcR=aNGaWQs-T
zIZJ<dy@LEu+_6MrpvAz-rtSuUQmJi=_B!-A&Chh=5u~Aoa~Zd<K8bFj5j%^~Z%!=H
z*z4<D*%xZ0r^&Ad!vP=Wdm-WJfl|F3Jm78wk{1}geG3BWPwablU?P1z6m~GEe>Wg}
zKPw!EZ<Dkxb8V0bKLxwsJ_I5m^3hG#6N&SXMpZy#ICJmd)%1~UGk-q}R5nP@w?1=S
z)Yu^HdH`=6gY#R+d5FYBFblBVY@@~T>(Yx8(mvpL?g!^_B5exX3q@KRl9+FvfP!Nw
zgn%@L%-<!Sfu9<ramIjkm7&D%28+?55Cfl9pl-&eq<G^=Gd~6N=}CVnc&~|K+hUzE
zX_I+!?iaOt_i`l?UZxro>Hd__r-#s+2TE=TZh0uU3@R}-SV**Lyn45#^<86}&nb)f
z@t~6M1nI3StE#I>!`TyEVyG@1Sh>6KpF<bBiY3W;XrJ{P*+SS>GS?D!2?7#N9WspB
z4C(od+`)vz`$)rUC0dZ{_W%yfYIYm3iS)3k6OTKtZveU2$&S^x=H8cg_;=bo!_~K%
z#F~0=5*dtf1f^1-c-<ftS-pLJ4;jW~)S5E5jG)0Q16~qI4bjcRCh2&|f8%s*W-~VU
z*E3ta26osSS{2>B@waSY=o4KA6t9m@G)8(iWArdlT&!Q5FInWkb?JpCdlWayPaF2_
zf%pp>iVo9hm!41Cer`Zm<PNScAs6XBouq*dkE}YRntIU0o+4Se_V$ONjg+T_1cAY&
z<Ug>5cKu!P+XC9`IUUv~qvnz92;gO3e#e@%8z6&syG6n9%%U^W>$gb0zT>+uIWB3M
zk0j$M(sr|C>CwW8ZhU<6v0NnXJ|srgJcrmJhyu8|+sZ^2$$@hJ;jI8lN$<$;?v_wJ
zL=f5=&6<!B>0^U;mw#DBqp{`4=D#}pO_T4eu32IbD!$oP(aPs^(46ao9*(ZrqaTL5
z7R)`$<pZH!BMJeu(WU|yj8CM2^)H%qqwhJ_CRS%3%`yht%8}Bx#7mtsF;8%>4KfTX
z=zhYRY0KThBJ0&5w~jtA3(Ij}CEkcq53=@(oD)HhXdmuPvt&0CD#G)7AP}^R<a9NM
zNP?-Li(&Cy!aZI%Eo$~Z5hC%cjE%P|KfmV_4}w3vXAwm~E+{-D)VY<j+|{?1KKh4K
z+Vl>UqRlLKZ9fSO-R0bg>mPiXMol+lSC^5$2%TPOJ`Ae)7))^so9B?9^MIrf^qczy
zcc75Xd9LUi$Zy>0A6b;Q#74n1%kI=3E?<+rKN^phVtlT$+x4dzy<vYem@1ZkQp0+F
zN*fkCf*9P^99p+qX9IAo>Vq<wF^KoY&Hc`m+k>7i6YKE3=YF9EyGf6z2+aUdgh4=k
z@y1Gs(HiFD$KTLZ;=PU4=1{Ekiixj67M9zsd60jMWs^gP&-O5UEJAS%R(ukdx)6s1
zfFxnEniruLcoRlmZk6;K@yuP`MS8fqchi9k4-fxI<%mAyRM`-S=U(O`pPr3_BcPF&
z|G=|X9sU<X78Y!`VLgQy4#6ubChZ?a$u`l4>~Y=Iip?wypl*}X8{<!Emv1XDQ_qS7
z>O}i2QVDO!twWh$dah@QQ86&La%eY(-7~!auCXdpdr9(?sPlAR#G9&Xqq@g#xr=Y+
zkXkYEOS<6``-fZcd;S7{TCXl#JbCx`SL{j|>ehMd%-j@haa=MZKI+S|9Na(I?XlY^
z{~MvF#tPi;7Yx@$E(z(||2>PZ_(&;KkOjmVplto0{$zgpafxJuQGkjdD|E%wS2KUZ
z_7}_FZzLVKxv!&tA0+|d<}XrI%pd=^c>==m|BmK5>02H{Mtl>?@gu(||4$pMR6P4=
zq~I%nUGI+a*?{CKSy%?g5DCR{T~HBQK}IlJUXX>wxduSJkoLb=qW}E3dpVnNXA9#C
z_>=q)S};@{2x*fd;Pc<!7=4kDrT@=zCHw!i8nJKW0P}a<%@aV2jjRG=t68ob#Vm+0
zDQ)5ggp+hBxShBG@f%y;Ais4-VcDXYy7<Gv>>)rOu_sxstgQws{Xyj=`X2n@Bu0oH
zYQHR$ktI@?>G1u{9bd2DrU<qR)W2p&pD7%zXa$Sd&)T!hpJci4;}^L=<`)ib`L_b|
zzt;fO+rRh0zt;da!vBMPAPyl|k&utx4J>zyz+qj{4IL>@sMsEzkp#8l1M;B<S;Y#G
zQfHKv8{dwBI69`gIHh|=UH#>KO<YqKoT!Kg$ym&!i;6vz5T?n}5cUkc^Z|_K9?Cv5
z2BB>roB!c+`uaSA@FwU6>mzsL-x!s0MMWC`57SoTdEJMRZ#N(>Kjl$=8NQ&QnO&IA
z(R9WD?v?1O+K=&rg=M%FWXJY52hdQu;6H&6?<)apLl6K&W@H?|DnhO@BcT;|H)`dl
z!H^5Fkl8$n5Yyo34HTAKW9l7guD3plU*|7Dhw(SmTbXyjS0va!WrkObcJA1517&Dt
zgg*KZ_|(D8HuT#zlAATTv?TXmNYs@_5PEyza|WcjsdGhgZLr|h<Q{x(Uny~M@iQta
zf!Sl2&0t5tgLDrs!5Q*7Df`irU1TSq`moz+d^>k``<)axUi}i^k?D1*_>Z353*Ebk
zPH4h`afrKHKRQ}q-G2jQl38Tvj6U40prD}JY4k8S$OZ+2QgMdwE-h+8#r)tgxL4w}
z5F-x_I9{$jY=!^iJS?%mNP(R%JZk;Nt?+|>hd$Z<-Tf2cX)Q)mWe0*`CVNWj*T>)T
za<0bW*O0pTn>TNwFV%ofd<00wiE@~c$gcXlRI23HaH2d+*!a|ma(wVjHg&MWWs`Eg
ze4>2PtPH<5`IE&D%k!lVmP^M-gyWqZG>Q@W5a9hQyjc#`e~xFFe?b~}H7fj*mFkc+
zKK~Pn?d|$_pT3+ZKv&{P^R72Ts(TA_S$=7FWxo@j-+!Ru)a^0l;^Yl@%kNI>=xhWf
z`go<fYb_gAZ|ch<GJ!@dmaQd0(Avg^9xZPIoF}ea^m7j~CaG}rg2R6|rC0u}{dqSK
z*i&~#B;#UY8I^sL!|h@+<h<Z3PevW_@$oA9vm*L<k;c#7X|lgHOEvc0gB8pb??gyZ
zlX(1Eu1J6X)xbV@lkf{S<$ZofiswG=ub21rF<)~ZmO`Z;UtDuyl^5DsN5N!Mgp<b?
zdm>~A2Z1)}qmJa4utP#YM!9R$@JJ4dhCl%J5&#;R^0|-i=<^wny>(lv<I_7zrDKBL
zp%GV5WF(iGwyg^LP9!|C+@Bxl8t_|BICrm(ywEI29`UfFjwQnlmh|H%bz(^qTLz4~
z`qT8Lgg&dgN6Vr+nQ`?@B!=(9I3M8!@4A>C-&2*k9$O;%_UApnRAgpz4)#&>gH`l`
z@_@nQVD4Tyk78CF!hL>aloa_L`}9~h{YhW}Sil$Si$Mt@OgB33<6E)$kF%>Uf}g4U
zyvckIfyo3kr#0FI)ai`?_e@fl&)6CgxH4GI&`e7kHwN~_m9j7IUR{aHZsIgH&%tr*
z^{0@-VDC)*^yoz6>4|B4d14pAK_La@?`9sr{e=t^y=Vb9h3)O_We0>rH|BR_gR;%z
z^T&4Y?dv;=hktdRB=W6o7;#9ZuWXZmw3|Laie)1`EuPp?^-tXHVY%=}_G;Ni$B%C)
zU1W0fPdKp?1|v$wFPp^2g|+FMS_c5Wc?0_}qICK~3)u7<mte-<hJh&qeej*v$1i0G
zwLlY3&~fsuT+oXx;0k$>!+L`u%fTW{PbrUPW3n}OIJ>>=A>Ps3V8Rz?tJI49281ev
z{v%3Uc_)l3mK?c}3**nGt7!kOA`{`*>I=8XewJ>!0j7OI#obTf1#)x^?*CC`Nv}eG
zHm&@;t{?yYzh&|CKV2|=zDYi=3CsU4-C6&q*WlBsU||Wd0$2xLp2@F7{4d3PUk|7j
zR=6?ufmpTyj6!nlpe%!M`!P(pMG~aE56oW<6{h0HqR&P^ZqKzo6<&>e($do76BD@Y
zcM#BASXfxcAnS-@(sO4HV1yzKGi;(MbfvKj#+4(yeYp1-v^zMjrl`0I1T=Rr+{%FI
z>4D=BBEXc}3zhLfspt1;dt3<s3*hnr+zDd<*Cnm3t%bCi?`Cfpy_-v)-%T_%51fb1
zbeWd-cw55%<=f;wz3J#$80x_ugEd5ed8cT~(Jzx&`{|k=+>2C+I$nUcFh6+%P#D|Z
z3#aLRB6l^=NiFO5A+&e{?u^OPq>mOaPp>em?`n^zjr;2;hE{fhe6x-kO^fIUd6Nqm
zc|^NSskyN(&iX8}fpKW@&n$^`k414Yc3^~}{r&`2eZN4#yY$>QekaLF;|#1BBpoMj
zfFW9n31$v@JzQ&q!;{GS7<zQsZx6_i4Jf!tV6>K#u*`5!t10x+BPE=k>CXgxlE(Wb
z8HjLJ80sy@V9mzJg~_-cvYY^bKl_wNY9(oaY>;x^2icwtklu6$lL>SPOghhuc;Vfm
zrL|uVb0t=wrx`_xg{AQ0qaKtnOfL_Vmkho%m9u3s7ndy$B?)(zQ}~oW0<{keSr0s_
z204zSkSiGR)4y!Jkbx}ImHjQF6!&qMb`Lh4S_Ym-C^|HED`y1JJ`urWR01y_pIN;)
zw?ah6-<R>$lW*P#MZctl8irmGiy$qK3ebh|E_M8|Dd)6r#1?*7>87tef^JSyK{oyJ
zH1)Q{bO;Qoc19W0NBKyFjtv=Os$;FkcE-u<nOwi;UGC3&(wT46uc#Ve@mX89L*)TX
zPZ8v0r0oAdDXt8uGJA(5@<R}1UqlPqSX+0aK|qX^GHTo*vdX7W<I|#Osu^kiF)in@
zQhBMA%|Jul*fBJALhg=4)(e;Ywu5WX-BRm-O_tY4I{;yIw0+T_l6--eQVwW~zgUn>
zd#H6&lVr-pp~umQC9L;wa1PFY3gZV88IbTlMt!?wBmSQfl_-70(J~O8^)T#oD>AaH
zJCwR6FljOIuWR)sGXe5EM6j%t7_nUFv+PD!>f*1Fco$;$Obwyqq1)R(K2Xu)Dn5Me
zj(Y~)S<1XcD*SxmQxfXCiW&&?gdV$~Yu5F4A0ClAtSkZ=mDSbRKu`M>iDGoA{)Z{u
zMO!o+V`R!*C&Idt?AtN^*Yl(9?4chMO+!9^K5o|k_W3@!Wga|%XSaV0u&ikAGdvEC
z4%bpvwWqaWT^)Z*fn;|RbBQw8hIHjDyir)#R*c!SV^FcWGrw-#x^n0oG$-UVW;bFQ
z5_UH$Z7ssE>59r3(>Wzx;A%*LN5lnnQ7jS{oh+=pSPBeBX9v|v*f9>|fl80NrT)dI
z-JO|T-t@w8&$7?EJMvRdmJI>0C6-%+PAcxbp^n?gVev%lua+<N>Mj`4WkJlGeP97l
z6ApqPfw?$~Js7b{oymo6RXM<E_QkK+e?`|I7o8lYyANn)$EdM{{*Kp1=y&Hzw5-^$
z@8K(bta3M5%DijK;=?gq%RH*Mu0j9glzv~S(2ZYq6Rk-ZN`R62l0x-9$H@ytJ%;mH
zRPI!8^AB4YI<4StVkCbrDZs+g_Sr~6u@8nnrk7v<HP@Ku3JQu9+~gabIzTg7wc@}(
z2aDRFbhALsWBgZckM~it#AoF3M$JU<8R_EUfRxq^(1A+Pg?Rq<Yn~~FJeD?oCOB5B
z{KgH;MwAi6hw>y1-aC#n)qQ5TEm?fwu+UT0w^2@F?(9lOEtGc(pD*YdED%R0<{+Z$
zrl=wIPcC|YdnKlUcMf%mg4VDP!!aD3Q3CcIR>#l^EUGidRJZQca@@H&byJ{TJ(vl`
z**c1JJXGi0*7Wyn$U|8S!fsFJrspm3_g+x_gCfVU|ED~Z7j;uZw|==7huUa1TDzqX
zIbrAQaa-tj({%Qp1Fv8cbXV><=AdscOp~pSIrb<g>ZZ_|PnijRhB4JK5NgW#QVUmg
zEiTOHqsJo%$^$vt+|SL;9VlINYjre7Cd-n3=A@HWYunZ1YHoao5H_w%^GbheGCa4M
z1SRf;3j*knmK)|~TAXT#gaSx{KV`@sq^rhv`k3Wf!=n(9WbcGMPxsxcoOjJLC$(tr
zQXgu+1^!UQhEA^HkTcz1*sA6J3R3H-E6iRb`agUlmAgNxTKft`&NardclJYOB_%5b
zslqa;fLl9ZKT|8!N8Q0ok!ZIy6(kpntp}-hdjy<pj9RdSt0C>;MeRb@xrWHH6zu{%
zP%o!OB5PYHE^gw;iY^bV{ZFytKTZ?k1K@6OxvX#=G=~bHK9}Mr70YYMx}XpFfD4x*
zuf2NL?n{kIysYM>ccbALEE<EHL(BYke_t&AcxT3M$DK0rI6C9M-<c=ibMf@gbviHq
zmV)Pl-kqT?Y~f1*$|ryM<(CW}akv0P%@fClpO?oK{M<4Sx~Tm{cI@{BUHi$-1VhG+
znx8*OduHaxPcr=}>bZTs7jB}%n*8GWGN|aB?w<pFD0A)?kLf{h8wA1sKv8_dEG3V7
z0av7m4wUQ2_;^~KknUvL;~zgoZBzY5wR)#L%p<|`J_Lsqf38ehVr9l+cDp=y?1FTF
zmDmYGoyFicu)&A&PCNSSXGXKVsoe>u33qky(T*H}d^UL{k*AD%Mx6v78FP`<6WZi)
zuLKd%&-Twn>z<|8kG){-b8^K|=8pWFcd5Pyb7?gg>z9E~iRN}8e6oX%YFE@wv)rR!
zufY5S$|U#id`2&je<Ray2D!xoBmcvP59;9VVvZ@QP9H_qN)*gluj+@G2XXH%R!ITL
z;5JGaDvsuT`G^lNs73)4zEiF%!A3UywF}co<t@e2S>MfQV4S0<9q2?)qAtVKW2g|f
zrasuc{1Q@O;mEhB2eCI-0o8W0?Z?}@Q?a$D{wg#Vu!KUNV2dk9d;9y3X=rFr!yTGQ
z>#5I;@xra>Y_CDuW;!`rGG9>Q0b2{iA8?w;aT3k;*Vj1o!1b0-aR8ZhY{lX~E7Gv^
zOgA+|Re7mwew6F{kTc=HeP=Q0Rqpoo_a6n~?c0$&=;2P!YUUYce)h)gosucrrq3~x
zC7&;;dFudtU}Y=!fqd+3=`wWIYeY>+X*CA#e7F{8%WW;5At9x&qnu==Yq=0oEn{%U
zECXFA3K<$@is%&Sc(Ms@q4o?hq)rSf1YxVl;NU8&s+M6ifr}ij55VDS3~RX4Zlq&9
zWt5l@K);&jJ{Z`7=N?@li)-gzKHUfQHMma8VHT%s=2FjG(nq72hCX~ZCE1uEWb3Dl
zxxNoB)E>K?U;iloYLQ3!{c*}}tY8Y*T=4A1%B>WJdE*6~Xfpz1)cfm?AClRPLGzrl
zI626SP8X8<CUgBduD<B`7eLpF&!Fa|0~F_WY31&r_}BPgQGc}VNz|^<qTd$(TE6AR
z6&b6j+%+8!8xz#Uj#!ziD0et!x}WsWylZKdSb9iC@?zFMgZFOCnzF7#oX_5Q4rHCP
zVvk~$YRa^3{(aLQhOeXxYtGAZhaBe)G3cAfOUX+)(C)Gr5yDeBlONfkAZmDtbPIZJ
z9ozWeUe$OD7+d-XkE|m-MHEqGbydo9Js&II5yb8?{X7tF(3F^343Uq?8(j<2Ik#lD
z2}!v^1Kx@Va>z*TBcy$|vql=^F%P9v;@jz#c<uU%c>pJsjXkT`<eJ!M5wbtcUmwg$
z)-Di52D|~s!5zfm10aHDnPvsaZX>T{GFUG|9;^=%vWh9?pBo+C+}cMh4$3rC<!xUe
zX7-Sa>XYUBcS(U2k?B45ih^Ol?l66=vUByuo`^wh%{yA?L&22M9u)iSFx+7yUv&_B
zB|i1!-HGD{o7oMtC&>=#6&&zh!j+(TevwdeB$S?%UwQ8z$pzEq7}M_#NXii~scFuG
zcv6Iu5TPP7mM(Ibn(2X-p9X$nQcK`6VrS93a2lAzrI>)_#3r&Q6Ua39T$Lia>du+s
zPKGyzI2|+Oa73{xqp}E~_YGiYPr<WDY;=<<$)Qhjwz`*1t=Q)=b3LVYo-mkOG)4g4
z@jaAF?1%={Vo2ZSdzcI*A@bFm@3|9;cZ>*ydCbc2$_5FPutW!{wO!vRxXhXfF3}G$
z{~R>m^I6;Y;}>>(|7GOp;}^bt{2vm8^p(4tzKqp>{Dw8(d{*=M`1dQn{};~!JdU5g
zaQHud$I{<F4?}+Z`@vuS{U-jGS&8opk1R)?!{h>>Hvadc`S@i1KUs-Cj^<G&bY;#Q
zlc`?cEjy6-HOr1&;QHS{)~lTRj4U!~Xoeqd*Qaf9=X!uF-=FUp&h$vX&S_U+yPkz5
z%L0^Tn9liGJE8e>AE|zp`bw6AP0==^%d?G`qv1_LNZc>rLQSa#vZF#*HD&}#{&I=9
z+<o&U(jQY`{#OzTAY|EtU#gu_T@ivNAN3=wLB2ScOae`XGq0R@b7n@nq%AN_)5iBb
z%U$>SJo_bQC*J=0X;GN}`uC0Sc{To*UWqG(B4BdvK*P(06^=Cd+e}}e-Rdv3XsHXL
zTAi)QBT?ds3w1#;D@U6$mX?^WL}XR@E~c};myP|(^!FO&mj>MNx^I5jaUfmuynj=L
zn$WU8KMhTe*J0$B_|=9XQy68xL9HV3`0odEi1CV;zk$+h|5FmY(1=dGfP9n&I5e{_
z!c)@D^YC)<Z3JMD$6*o(G*Q!*pmvk!XFY~DnoH4D&TwGAXo%#M3Fs1={4I7Ud8$C#
zgdJ$(Nusc!-^zhaYe_b{EiwOckX7BPTe+9Tz`y33R+WaHvn$8X2_TeO%cyRRtlYb1
zDKl`zw?J?pY;tcn6&Eerzp{M<?YM4~jRF}9Y9(_jdm%B~i@@*2H}ivaS20GKS#dH?
zCSKFF?-ViIA0Zd)MvP2Q=e>|x9AH`<1N!W~vDNzq2<<V2%d?AqI1_>xU+~vZMnbBx
z8{7D(+v0w(ShI=1N?tZ~k~1A-Wl2<d;=EJYK`z6TBw(sf2C^-SH;^TlW6Fn%T#0|;
zl-P^kS}&0UyoJ81Ys3Hyt`8#g`XU8orY$|t520yH@#SL4doX6?g{9f8nxsx;Od7-|
zE<@&kGQ*oh7=8gxF4Kh+1HO3zoE2MDYE4&;vx_i2Z!yBBa@m|lg95X+Hk{@x%e`V=
zDc304bGC`2%iVX0nZKaGrWOMc>wBW(nt#Y0e*H0?%MKs~l4rj#59MKN?N!PND2;%k
zs2aIo0ejM`K^SWX8pfMrp4%5{rUx>xhj87#t*BJ<iGf`Rr&KBfwSrU)QObBDGb`>0
z`z;WX5#Eq~eY*Mj2ES<~U6j`9BXeOE`Mf;~yk@xSUX-{k?!!zU@!qT6=F+ku;besg
z=7Yye6JwueoVXV$J*6aXl=Z!XR}B<0Bm_{2yo9u;YMZ~w-H}@%@jeFJ){kdZ(-S-`
zWh$<U@N<>t8nBLjHRY3ZG^No<$KPbfHy_g^Up4AsLV?R9WQx$FOwz3y_TiG2!XW4A
zVY|CO6B0;J_ACZwc6R2VNoR#3JxD_uJvYyVJ-?@83#=o`eg_cb<3lreP*nFYpE#Wz
zEEqW6QXZczonOMB^yCbXDsyU9j{wbT6X2=<#ky(B8xf%fe0H*;&S;L1`}&}0h1_X-
z1EFHp%`2L>>1!R--kf;mdjs~RzkSTV-gtqn_Y{e4b*J4xOpsBGkDs64Y^|j$9y8uJ
z=7LP5e-Uh&89?Qqp`rLV?_GK|p&>EV7!h<Q@+eLY2U(KdNA*T2t_+Yay%ueIFF=^0
z3$x|Y%z_PX-D1FsC?#Kg{as^T+|0kGj+q<K>QB<_<rvub;zPq#ds&pP*L-8Ez|B0U
z@`7dGHh!DV{^ya{&0Qf(zQe&+G*3Vnhha11$M|tK+|F_(Q0p@0Uox6EzNz|Z$%dc7
zKRIdEMRy8~ATT>Y0CI_!O?KEHFk9eB-O<aKyU?eiBXIf!ddV;0j9Q%(DPfzyvqzQW
zBZ(foOWOjM5H*3UzzdY2OW*)uYeX!e0{>R#0=BUS4XlbN?)L36<x7;QFjF2OJ8j@-
z=||gco3O+$5BWhzsiiDu%bUlc=RALm=Q9vldQ<LjF}Z{w8z{5_@E26D0Dz;)RQl$T
zEi`YHlzp7hj2;hJe0_MsBTWR5a#XlIrR#J(0SlJdwVMY8N}*a<ho6WdCp^9nhkbPy
z=UM`%+wplW!*jWZrZqr--@IJ>jqe2@Q*`M<#AgkJ{cH;n{AuJMz}y+`0~qV`JTXtg
z(cu_-GHfPH4C-&n17q9HVvvxK29r+UGEra+t<Mq$u1l801J-~lJ^*&EI^lKj{Db7A
zwVPAVq|vT|L$BKr{LgywVl;|z3Hk2Ipl?V8YtPx%^>k%q)UvHR9aLfZM$!CDR&wRu
zH~X2F@i{(7QXkc_DrL2L^wp9h00i}MOZi~dVm~1Qm$6L~I$gKix<S&7blpR&?k4(f
zM^T}5lf5i00HIX7!etoQ&@EuP&SQYohj3}UToi8d7eGc|0_y1jK<|wa`5j<w>L$CI
zc4sVL2!L6W_JtT9#he!NmxIM51G;(-T+W!5>{K+GHgTTGXgPy70-X^F8e`c9L{%T?
zg*QI$QRHRaSH50;gQju>A@gV&d03u?yXb!Jwb+k;X1wac)2%Odh*(Bw9)X>}yIIkr
z@&-D31UoGXXFLrO$4?19&FOM=qMJmdei0j@C%fIkgNk9oi;kfE$UJj6fBbl4Yp@gv
ziUri&BN&%r<x7kx=%KendVWWMmPA3RgCO#!SXT}bIJo_6M*&?WBrMv48Byk(7)`>=
zzVLBfL&CKt+9YGBkLb{A(GEXxnl(D<@D6t{v;7G(I<i~WJTFbb9~XGyUoM0(SN-UH
z19y5<#}olR+g9DoZ2Ox7#tO1?uf%D=Fn+Q=M@d<E4KOl=xvwW^P7wpPk%1CJt2C0W
z(=~eyvc{Y2R6XK53%L`UhLf_B0Z#G*^a{xwxi0<TylSdU$TcmY;y@=MrCx*B@k%)%
zC4u!_Rj+1M|2PmVUzRJxld6&;Iv$#IJKtP`-z8mhDns*+LRC(oWe)s=W%rU{^{SO;
z1`>fh#M#z`2=hFMKqEofu?D!0pSK3w+Vp^v--P}~N&N%~^Q>mBFc(@$r^voO;UNP7
zayy;c%(>Z01y@DK4tP%oG|vahZd*m8kBX5P4dl&uMrXFUzhr6%Dw7HY*Obw{l-^F$
zwyJJG9Q0}D-NY3fFx*sTd<Q@K?7)N^4gcJ{fB$}_sWch}t1&MbpV`@#>zzG~VP{Dz
z_`}B8E(gAKsT)~VO*|vij&75)q)FOfp^rOmz4i>)H3|@_iRs9O{A}=eE})DU?FO2D
zJc7|@AyNy_xrss7m?l?@Br9cSAUmj^iGk7p>vSIM(YH&XTbcQ6h$LJ#;JP)5Y*@bk
zoiEMm>4DH?GIK^PdM*UcmQUD+?Gf8=l5noGLOGadQwuaKpI$=zwWd0ZP%AAQU_d18
zR#RxAK9G|@og8#UYfZe+X);!nq+OELh74!IuZ;s1&wtLh^d@adRPjnubzr1xnzmM$
z-aOJ-ZI<39mO2;EMDR1675_opSBBldc26+LZO^O_rq)j>=?c~6&@|Y1$1zxd4~-|h
z1Ke_Ih#dkBhf+?~UiPe>y<M|SI8ec8?pn=c$?g!|{hpRC6DRDQ`EJEaJ3PEUBerD~
z=y#R2H^n)75{5KnbP_`^_K9f7t9tl0?etGfx-DB_wzBD&+0^!iZL3(j3icyb*l2IG
z^<AWpb?|Zf!}Ev3QFM&~UsBB<r5l`v&&phqkau`XPQN@6DU=O@0keNby-vc0dRzxH
zKo2Ve3>`7u5J7`{7~tMDv3Jb9@u_hUI#6%%NTH-IIn1-0G2^yqzrpJ1$~1I{Gp83-
z<#moxdv<cLiI2w1GN&WOZHz#g*t&qw7u{hD01zxD<+*&A&NrAABmm?q2`pGZl2&(O
zei?fJ=ZSf|#r4z-Rc-5P22RF_KD`Id#8brlz!zyC!TUscOYDTDrp_?W@7nNRw?ecj
zM?BuyfxFkis`1P*8C?E<&Ibkr&^YeuxNz$e80^_UEP;XTo}3bf`39RBLEgwCC(|)i
zErdL=$T|nYDBLt=)U*IR{yzFJyg6b>zLV{swZMTepqtmO=n|>a43ZNY#1)SZjAIx6
zjFEn$thSEn&=6q0S#60Y4YH~#O#>4xHTdt_O=!v+XWeM4HA~k_pYv;?e0rVMufI@k
zZNfnOkHY5D7(N+$FMEAg0*66tRizxOf5<{m&5<JvPG+1cWm7YHlU3DpNyooQ(Z8vF
zYhui83q1Xn^y9IcBbkZG!N;^!$#b0Y2Fhi-A?JG&r>v9~55K|-*qjVeA7>)5^9c7|
zKV?-b(()9ad)+ejy*Dsi$!xIBVrGH}v}9dg4$$sSV>zYuV3la>FVCmKpD^j$0_a`r
z{q85WpJ|S8*V*h{1a?wm8Mn<2LRuHcxE61{kgv9EnlvWI0UAtY&VdpFN}WtV*JIFe
zeTs=iJzN;OJ7)k8_*4rN?24K*8<fb0)-0xHXJrOzz0p7@mszL12E+cBsIo%cLDPFo
zP6i}JbsRgOoZL+@7HEa>rHvxJs7H3&GY}6#1E&#gI52X>pRpiT@?Jhfx-2P$@&n?6
zqO+7_%>jWTZ|b+vH1EzFnh;|Xv(Rdrr{<v*XgskU5whF5>1bdSGrLNp2-}uY=s1_C
z3|}}t3H4^9)p`uqPR6T#*dQocw4;mNV8PQgGmOr7kd^$OgGid>G)I8w?!_W$ns3Bu
zd68??Q{PKtu{d5D$)ZC`HIntxH60DI+?y^T-V`C;#HhApv{CN~S%q4CK24fZiOfEr
z5PMG)K}Ir!S5io+)F9SCX!VqXGvA?&HQ(uLh1_b{smkFXugVvhbXy9q8fb`IH<`O)
z@0;X~4~KUl=nUplv6<{<qXS9s28sTf5PxMPRxUdwjnHT?tcya9$-404INIUchuxYr
zj14=EA+=tmAUUO6rB7Y?BPet*m<~l2rIAI9(?(zFy1HB<zJ6k%nFu1vWW1p`QZVe`
zhNo{b<9`$BFA`ng_fIJ+IN_s#5>pR-pqsEli0z0Il^Vj<;Gx0&r-cp^JChx-LvB*R
z{(#|LmnE7K1#Ej0(nF7W25gr4pQ0MmdF{}FRDhyWy*@3B^$dxbDh{h@S;_o4tXdA$
z2Bn2?l9sKvFm4acan%O!e-p#Aw+qa13mLWz;s{n{1D#hn&xxbrqA7i~!3A<C+YNP5
z@zt5?QZAnao!O3J4F85^xw@h*ieu#UwIP(Mjflf8<6+rcI7rNYMv}RC`YSA@u+nrZ
zw)ImaN(~l6E8J!e=M86?HtqZf((AAi3ujV-xJz||^U$h_P_BmKR(kyi0xBcg4njh$
z0rSlib_|cIo-A2FoLP0Vt=g=$%Cxtsa4Iny$%N|rL;<5LWGhK>caZIv)z4Et6@4HM
ztumtyW0#7(UC^SJt+L?)t2m`PF`bB+$!2!w8|4Xz^au*{oE7p%#KDbY_G)gsu*$;D
zH#Gt7W63q;Ije5Y-C9xrA?DY%p~-?V2~|^UYTD0sj%ym3Zb_{)P1pP*Y3&*Nbj>ZB
zQ_bhAj_M??2;sgJKl`ea4&2aq{M8S}JF&->>VW2R07=#hEM>qvk$H%v6@STrVx#o?
z0ZJm#QWx)?W>#zeT$c_I2nq<;#^0ZTCmc691To~Ng5?{O28e=x6BWYKGsids?W9G<
zG~D|4cip694Z0tiOwoFm1CaS56rtYM4W0A)qy>)FVsBhEYpJV$!!qJuGc6?xA|%kD
zto{V0IXe)co!YMG{^)^eFc5LbivS5Fi|R>}d#rh|k%<EIW7xa`NC(pE#H7hFl*Xl~
zR+TJ9F)J<?yk|4EoCg?Yhj*zz+(fk=(iWL2eQFR>hGs2KZDQ;P9_062U|HSZOUd2R
zTTGMOh`50oCHVf_sphjv8%gEzG6rq+klzAXT}9F5CoZW2y-G(T5xj{6-^Zi03B#zj
ztlOWtPG`JCQl8$Xi~Jvh&{#pb8rpI!Bki0}knq|w(6OSvZpdvlUkoybW`5`#brs_q
zwu7XUi33s<4U-%JUuHbvNeoQQPS-qA+b^P#>}r&Sq>FgWe$CJZ-!uZ+$^~eec}?*(
zoDTD}?2V`ikP3O=jJ<OteNKuBd!ws@DK(X551Ni5KvnrQ3GpPV@KY5kA``n=_fve@
z|N4*Blt>1S(Af=iEy8VZ8Co*+Bl@`febDdAuSbj@3flW}4pkI^4)r5(#chtybI{om
zX;v_IZVqe(V)Zd5SSS4UYTRI60><sk;y9+7R+<pFWKE0Q*VhQcK5~SD>_*7xW_l;`
z!lceBZwQt)Yfv9lXi7tf$O|c9kVXkeR>Ztlog@6KYIGj8Ss9|?21+?FzAp~3sa5D{
zyHgoU=rz;@KD;glUQVc@R*WO<Fcc3y2w&2K1=9B}&ZRE8D=A2d^a_$)Ty+u$#@ngN
z?HwsWI$H{cN)$R|V|PZKdEh9x!RcMvF)gH$W0=i9f>V=M-f3OUuOboWoCB@=%kR?4
zt2#*_XSONTHE>D_F4fJm5E#-KVP0LAdlC+u?Rwj(nlNhb2QgwHle|KXRFyOoXxsQx
z)zh}{4fo;35Y*t8L<XA`*tEq|3g>ugcjER{H*??K_o(_%#}593rrSO>lqFe}v#KMN
zas#(Q6#YXCxEIu<m>^W^IZr;}am`ZGt@znWgRF#GO$oPw`DBG|Y4A;DcG&LzeFu@j
zWO#-#>LzU>?*Gd@)GX2vU{@Y+r6dFX5DzaoP<DZ4YHX*1OBUjNfGY!S&!?Ef*4fHv
zMBU+?94VRu)v8%U^acq5$X$bJ;UDqybVKCA-eid_TkJZl%J;o{_~b+opvh99c`e%}
z6?BPD&<N3Dpxu1P3ENr{gzvrUs<I429s{9_X57bB2exPIjBp%$5YSGcUPz(lLo=|f
zb)@)GrN4+I)KClEdu{6a?zTTG)1^SjXNWDtZw;y#qrguC#00?OnB6xJ?O)Y7jgh9O
z*mfc5orgp~&LI3QuO0Iit8uf}nYw}9Lo}YW4s-O(1))m&61SX?X~umHpFGz&1D7<E
z_0hc-8CH+7hbb5T@{pu6G>;7TX0_Frt`rGtF~~BQ+MMXMUO2Xf%NzPsl88f^JucOI
z8v#@0@d#JSHEqW~KBPLcG1JAz%yiM6c^dsHO;vaj4I}>hQ)2wRL|WFY<c?h%FP_p@
z8C$*gZ5eifjLoe~vW|&!x#(gG#hKf#3={w~+|&bZ!bx_mXV=<hlM?F}kTLr;Tj6%i
zE*^M}l_lt*!9seA-*@$px{LDkuFW>LjCW1MRiZIU{}eQ+@U=E0STU<@0iXvnpQ##V
z?YtX*aUOdl(n^v>pD3Dkm_M+HJ`%u)(16SfsZ#Wgv;!m(5?hv&H9Ykp89c1GLI>4&
z!jWn0DX*xIXT56qsOwAtCBBLlT;kRS;UH73&pG>IW~5nR#r7<7%kM#W66G7Vm|w_2
zApX46!~U7xNU9TSlcvoJL~dcN$D4YSrYS8YEr3iy$UPv!wFhTZtB3Jr&0X;6g_KND
z@V6Vr{^h%yi5H5p^;CHls;0G7AlDbtNFFrEszInMRP{GtA6MyZxQ=3f(=!s>OJ-zq
zZaOhF`nnpo_Iz{^Qr(fT%t8c%OrPXbns1tYy~2%+tWA#`A^#|M@J@wsrL89Z0u??K
z#EyWbm^ih>QiP#F3QA*Li;S)@^fpR~qJk>@2GONSE$V<kC&xBCtOcCPAlD*99-)2|
z_d6MLXYp;a1=*pvS*}^Ph+&T&ohtyopigoXKp(dyb4j~86ZBR*-#M!Kpwm5~BMbA<
zs*cOlF0DR?U`OVb0b<*#Y>?!|b!3B06xX4O&S2&T#LAcs@<~Fe7hosf4volP0)1+a
z;r=&Hf|pf?Iq)E@OW4qS3@SN^Qc4(19`RKPmQgV+!CU3cF&v-I@g<R3lRN+<c!uqD
zphpA;+S$mfCI%;QC#tBKqr?iW#IpBUhcr>Fr6V0OWC=K9y-SJeb-BalOn}C}Mp;{^
zLCk=A-0=UDE1$#*`6S(TrOL|G58&u{<O3uzJ1(;>`=A_M_)CW};Uv?=``<l0Tt7}0
zi&ELerU?EA(3K#3J$j19&jEz&WukGwmG#juy+H{M80N*+OJw<!lp?2)lJ*2Bow|vo
z+D+FG6H&(Mr9t`dgr&}rXji}1>~t4SbEk&$NU@KnqF4ziUUl=xK;nT`M|K01n7NCI
z)_0<#@XQ3AXa5lz0>_mN<SYHac6k0cRGANSTp{z@Dsuw7p<-vG{0tq9Q0ueYV%LFG
z-ab<j68~0`Z!e`G03%~MB9h~#p;9bW2l2N6w+nF6TSLkb)b7x9iLR$?Jm;)zHX~#V
z$eeLZr+MVL&=IaBjkf8Ur3OeE@=f<A$~4L${0jwDCyAo-8!Tq2Q=d=LKt#mXO#J0@
zxpLenK_Fy+LrCd-b=+98<@|A(jdDK7A1cgP6TMLIZ1-<UIHp>C@qTEbs*M4A%6FTi
zSy(v4NEcu{kl8di{iBdW)fsNLr+x~^cesePLez4kL#iWJCV>ESL!0}o)}<FRge_l&
z>aX=_lXSqB`%$YuoW*I<>=lJvtY5P$&P07)o?q>PsUnKEYHeMAF;yB-qS#WEM0E>F
z-B<Bxu4WC{3g35Vaw`NT9v^tcvg@10(M3jydj2JoymtJ?wLIa!9?F{Efws`<Q~Nf<
zOfStkN6fTw2kMRpwtc^z%bGhrHTn8RW7p`m_G#W)5P}@D+IR8t-UNduKma;HhQlV$
z^RNL=sV0Ue-PRO9I<`RRm|m~;43jwsF>GftQ{ZjiKb+a%H|>I4acA-`=uA||1^S2V
z<PnbH7H;|}<bi7?*7bXbJsh4^hRDpdTxxJE<`_-eG?I-PuM<2DsGE-ob>CeJGFe>F
zD8;bM4xq+K&FMJKN!dZ>kWe%|fMn#1mevM1S2%Y-?8ypCdD|hs6*_!9nGL5r$o5V=
zVXq#tLn=MCN!~Q21{!r`Xq=Kv6Pz(7L2ZxdxCaoLUQAGQ!;%9_IXgFnBN3sjADgNr
z?Ujf~*_)z^z1t>sBcU54rQI0<w`89}AxbS~dZ%OPSqTND#J4u+9J_Pccch)LRZ)|d
zY7%Y{dp1aN57a<XvUpae<Rp#hmVM(DTd7B}%iamp`fg$$cO&AY*y$2nn#bQXv%k_?
zVlX}$q0LpCxMRtp#<ih=lL57e5!?)c%G+87tW2O5m5ym%mFdb>LT~z(N+x3II6`<`
zm{-3kT_mx81zLW*{2N;YQ?v|O9r#_QwtuRTZVNah_o3$gyxImr=NNRU^4miRse_Td
zkTV@}pverAbS9%=Cy^C-pm3I8HXqSt+NE23fYy<Ks{tvrSa3n=zAgz&J&pzUz;g-|
zAk8`^W16(IFvjRfEe2ddeo~1yZUFpIc~Eju_ZW3+2U?&zn)J_2MO$r%%v5R9&AN1n
zed17jl(#!>sJxmve5Dk{wUC7++Gp~4<|v5*Up<X~`@<!Afx3vlRbM!8BAIT=cbEq5
zeQ)iv=n+RG35pAzZy_X$85r><Q=kD|&?0nv3Z7qe=PNAeWt@R3hHJb#so-^FqZ5ZN
zpK}hi6|e<#@(a--2~f~F)dFDB|2Ab;OJ`CQG-On(!;@5R<1D1$EZ9wi;+{7=!QAd4
z{vlO<l$sWA7G_OxX7+zGHn-?&P4OpIn1v(`0r*JVZ;)l#*hJgCX<NuOXG10_rSW%y
zn2Scy81u_?uGoA%pg<SPb)rNEJm0qA60|o=M$ZNz3)usImi3hP)yp@a2Wl|6VAOy#
zMR(4Y`itnunJoKRZA!Re+V&F2Zns^FP^1v!GjK5B$K?n9c@)r2OkVckc`9+gTUNC^
zL#3DZ60ro{q2s}uqY!Z*@2?(cLp}~@_ZII9_8h9bO|`}}n)uc^wU8ZxA@qc2TRU>f
z51310_BuIwd;cVtrVde*Af}y#ju@R~Hu*^evtyX*4t%~+>Vou2Wd{z<XX9PbC73kj
zKT0tvx)+$P>ie4yQHshs&Y7bAz3;ZTz`9W>GTl8%)g>K8!YwVMtQ+roMW|M2Tl6!j
z5ED|GqMn9~L9Atk+@E}1%r@f1O7H?*s(}eiHDys_gcE@Hpy`+3Da7n7nmz`0*pQ#*
z5Bv=Dew%@D9BwXxf2up}hungS0T|OQ8Z&H(QVsr_p+`%*UQY8GsT7F^ng?lFAFm&$
zT+?47+qFs^4F#&mO*m&Nw)!YD_cnFq3H%{?gvx`I{By{e+|j-YDBy4MKbp}?Eie=v
zd(4K-8JSPam!XvJosvSQajE4wgNd67D$Gl*pcuRW0gM@_qtO`iu^Wk8h`15d2qaiL
zVBVrFlNYFPmCG<nVihT${I>+L;tt8{Fh_TWZ$emjHSHKey?d38fPB9edUe+5=SiuI
zL>-|1i#>w?9&5|Bip2!C4)}1AQh6Efv@<c;L_dk<NsYkZ8Id<tF3Y)8?M&CSt2AZx
zZ}KBVu)^Xa0$la+sn{`*V!ukcF5i%GVPTm!{mhONf8K_k1WWV@s#EfPk<sQKR!yAO
z;6J33$Sbt$eN_`pjb#0eXJG%KYjNI<uX7Y=-f_o-XWy>V`k8Q2`!47h@bFV2ADAr-
zM!aW4ZYHu79)9ETr+J~7Q+J(?9KqFUt#}Oi$;lEOJiz*8xyqzgxl`JfzrBUbv%8`5
zw?i9!EJqr$Qb|Y_ke$lRK*zS_au?yy?~6joX+7qm8D<E9i~0obv(noIV`c)93H_ju
zGX~blzTSp@%09Uze-T^<;qKr`$&4&OhDVX&=mV=9BZ)a0+QOSjLysvKu$o-xUS5Ix
zxA$Kf25FP>!Gpjs#i`>m3d2PZvrG6OYz>PLtFraRPZ7MUI$t4q3r2?m!0){i!)kAU
zdh^>eQ1OIK6_0D+g3!azz;UF#Bl(v<jFk(U_7lsfel!L9qa4-GPRSH0Lqv=MAh)QD
zO}8e>80@C{nkSJHJqF@0J8F~SL3<eP_4;CBBgBXxFd&u`7{t0)%2gV0hwygw_WWNz
zKpEBlqrGbjYQjvTaq6_&X-iEN9kqy1)EX<UP`nn<inf4=6v*PeMTG?f1VyC_;>w5$
z7#0aosES2VBoM2Vi;?20K)6*Z5~6^jTtq-Yuv`oxdrm;R5Bsn)ec6Z21470G^54Gi
zJLi1om(~70fm$(a>NNBfN<chJZ|=od6oKJzRMY|c?{A(HlHfl?JPk+B<;rG6-$F!e
z#|}$0<+FRrG%c_;q-3Mx8&?@wI5~{aHhr^UC%j{`nd(-zXASP(a?$kA?k{eg&$;bG
zhz-vD%D`5fcP!EV3U`RsZj3E)4|`dhm7$Q5c(P2*7SUb+R;}&phsm9sv{j2*Py4H{
zfN?OZ@wMWiKLw5+6*lWoTdiYh(cC+NIJuI%CPB>Q6E@lB%ja*65(-H?f9g}WU{;T_
zGf8k%IChtU*%}1nP!^u25b17W8i<H#IdAmL8!_WXJ|-^=H_d#~li%ZoE!sRKlhulH
z+kkS^xgcbBU9@j(_VGd;LkPjy;w;H~cH2s8lK^B;zgW8Cc5iIk^u}yt&f|G=%ca^?
z$c)mC*(}9zUPhgBN*Pt&h<`Hx&pAA3nJl~j^dc9ZxG^f^#pX&4d<6=y$<}z(Gemiw
zEAecV*$<OL#;Ems3Fo5dH)gJoFw<-#!-y%D+7{(poUg^$*2#-<+ct}%)nCb?5Z_`{
z*~}I<&XOqt?Ow4!*w)}R{V+iWt8hzrKhBmZB;i>HLT{l<BuXog;}PW{=q4Q!)cLo1
z+dB_=EK@zy+6@f5;m7Mb5ol#7!ZSjCG_Wd4c0SU&gK?e!B`d6&0D>YEGgLOSD}UMh
z=HmdM96^ch-aFmtJ|t|C2xG;^yI_r~9S%0w;mSBvoi0KEq~Wm0SCUIZFNHlT!87(C
za~`9}&__Y*TkTm=kl=nI;kR)Lh77+->lxYxOr$cA>~<#Ub3SCy>lmiW51B|2Fi+VF
zCwXG71fcLFkUO7jFh-J%aBo8M0&<e&#~07fhpdaOco5NxC=u9$5KbcE0ybt50U@&9
zPzs2SJI))V@&)8Nu-fepu1Vj>yR#ED!3rSCnU&SEbLu`|rYx>e04-+oVX4hJRS$X$
z=e35-SD^w(mD-P@96x4nn0f+O%Q=1Sr2rMiv+rofYKOu3ehmeuzt7_!Q(gmBUjO>R
z_qDAwTyONn0hOgh>Y)$!ltbV;=1+k=Vr3&PA089Ze_g);@nw~)UL$qZ9zLznDp6c}
z`uJ#)N6|XSe2h$;SGw6ml{?KPYU{YSOv&=#$LUS3k+OW`s~N66bD6K!WRtb@x-;$8
zKh*3rUo9a)BoxNxgD>5&wgqg?(Ka+Z*{3Mk9n<<9c3p}}jg0ZA<TB32%8o|lk=EMv
z)NX#{Pyxhb@f_t8Gl)HB^aA|)sg`%MkEI@=l5Uw!x0R3weJ`%<s}BiT+qpTA+JN4L
z+Xg`oI5wbow7cU2*cZ{dB=sQ+t1hs#DLwa8)d5my6R5LHM#AH)`OwgiazYbtbal@2
zNFvCdu{MoW_RO4oEX=b3FS5;i2NjSZ&NAN%D%0cVO|9!AFp0J=9=lu0y4@>*fD}Lz
zIZ%LM7d<|UuB3^<@_G#vH^Am0V38`lv8`wZO?Wo8A_ED>VVPVZqA9R+kVtcL7a8J|
z&X6-A#jdpONic3s35E{g)0-%u{iATru0sRvQI}UQB$y#<*+~jf)d`$rLq`{K?hRQt
zX4FE@wyt$y1y-qgYg`Q)L+IvuzPlyo?ylNtY)gq8$dEBRj!chBS~EL3budqS7aD-7
z<75l>stVKN<nI{gPTfiy`(IKsG)F!1%>cehTP&~|*T?^5D_Qlm>L=(S^55WDgzl9b
zP(Z~i6Tk!wO*7SXb2G_zem?7NSB6#c%ciUN`<?dO7#1K{`_C<Mj?WCwi%UoZ#+8Ws
zyKKd_8QGa6f>H)-W}H{WuZ;Q9DX6pK*Jbbj5)fFm^rfjT)>74eMoRAC1;|%Lb?4E2
z%+os;=pJDVkF3Q=k0^+mRd*hY<6Ovu`vF!ZIouTr={*EPPPd@-c8g)i+{mm8gE9MT
z0R4Mt`HQnazOx=i;Ds=@b?6vdR_WlVh8Hv`<Ma6%&tu!Kz3rjC1?G{W1-d%lgw*_}
zD48`N{#m;Rr86BpQ-+`nEC5(>G&VbNzV8&CNNjJ>S@h^+q~r<m7INmKzE@!W#cd^V
z2vjUQ>GI=OK3BPA=Qe1rTw&)+;H-emJ}6Y@G=saH5ekUa{JMJ7o8m7&!TEU@THDNB
z@09Kb=TpQECzJNc*v|0e_s|iQ>QmZC$tXH|Xh80vug2oc#hKMeb;*F`SKb-<EKmow
zWL0g+x=6*JHr?1-;yGjs#-(KC9W5FVJIz$3XX4n~q)jVaVmN!q^Yh|@0{{a2zT*vU
z;X<a~T5kGcmxDTb4HWoM-9}+4EwiY^28KV50h?p#N3T?f72_Q)lefqNf#rw7Pf0ls
zlxg$H&EEt-M!dVBDaKyHWOo<jiy{>+vVO)oN$XaQgk(>9Mot4xw<rz;^xLwkhuQb=
z)|6A)ZX#nwn^O22G~zuXse?n<V2d3yM%>%$i`_SER<QeZx@`kDHFbpz2bF~1S%&OQ
z2jq(!W&#D?jS}3=ii|M5YQ1)CKpr)Y$2|Xdx_Z%H*w!)v0f_~)D=Z)mgU$X}f(1(#
z-1hP#5x#?vlS-&|280(=eFbM(7m;EnOQ!~@6JeJECn~er4ODtE`1CGv1Yj?;$_G>j
zMpV)OoI3)>nIFZKZ5WhEGhk1$tk_0PBl?$3F%}ajcc28B1{FmZgHwP%NfsfMi9nrb
zu4DP)>4LV@Qw@;YMqxRgR=p(W?sFMQ?!|yf3!h=dDN$7>_zyt8?sXyYvDSJCLYb}A
zN&clXS!Oh!Onz~T**9)WAW=zrgk)n7v=2PmC>l-%SEg+r>x0Sg0-!>ToYc;smc^>J
zOzl!cwFq5Xp+6+R7n2~vUX$?|Y7N94J2#soh8}A-jX2-^OpsR5;<d~>%1?DR-SB*h
zI0==#cy}>#WiW;+M4!%ZV4a5Q<%~gn>`(SEH$R6@4)k@yJG!bl=(E&nEK|xbQQ>;K
zZW|obb<n>gNqdONh=dR$eoY94EvTxZL&J+%F-r&Dnn19bdZo8?qa}n{Htu<Fz@#1b
z<V^fRFThDhWW!5{Vl55_KoaJx_+jNQ&lbmR+pY)Q*9xnb?pYeCM}7bnrIu&#&%quQ
zp_AOHOn0ji)As?kddAKx^p6Ts>@j$DS)zR@G(0cyp~XqvU!z!QMCEULDHyjz(JWuR
zap{(V&NXReW=>`oC)Jns<%yF_?5qm=I&^Sr?jB5%*CoR?#hcoG9W=b3+jM2`1!VNG
zYb;Tky5Oo}9g*P#fOcXxoQB>LAZaXKYIeF2J_Mq^u$8>Dvgy5vg0vOlj|{J5CJt+{
zt(tbbmAZFNUGg$U)i1VaF-r*he56+{Fa-y17gCq(at#wkEU@FD;i=y24Z_KEu3SUe
zBc!|1(C{yL+jrRpOJtgB<)il{8p)zN*VvsIHFYwOb#v2Z)4(xPhe|xHnnM0vsJ<Jl
zby6WVo|D@k#Xx|UZ!BG2bp|0k_W7}>&$^3wwSz>IY{KS9*y>;+!4(&Nb5JiXG+dBs
zjh2tHmh|bj;EhK7xTRcL(X@2mH3M061(M(&Ru8uYF1hCcP#z4S-qvbdXZ>-9qFw%>
zla+TOO3_Mx6p6HUASZDuJhdDrEAdO~b`Z%z81)P#oceEm)2F2YRBj4ZkCj+x9w8UG
zWE}e8?cRrlw2()UaKiyJfn?<f)Kv)Ng4GdQvaZEeXve6QpkYPA%%qY~!GYyyZh3x8
z4MqHcVMY9sje$@c`O}YfW-&})fbBx8GcWMHwkxle0)H_dC+7`+V6*!}FOr7F9=1yU
zEeniqlftOYKrIwhy?bSI$(#7*`Hjv+7q1447MLlu>uvNBgJIqG@Y~?)|649!4aI*P
ziI*V*y*{V@JZ8l=JI4!;3UBWF$p6v)qjd)sFY~_^W+dNNTzl11|9<_W6ETv|$&w<?
zt2g7nQ}>T}&5b;HXzGY><jBv4he$HLe(d{`E0r^^AOGVo-zI|>rr_zlBEM~y@5;Y@
zQ}EAZpPX99{80jzP>;m48B#@k=aX-L7FP%aE_IDv{k=hn)BB7B_h<LdQTiU{C-0Ac
z_I$lLZg-3|j<@ZB&EVdXodeZ<!9DK^_NgX@{qWxP)#mV*I^pAJ{(4aO&DU@L!GWr6
z;|@-DR0llSxyTuRAU4#m54NYCc<MedcyEAK8%uH<yIjowuB-X3B~Y&QCFlRrKAw7h
zuYLWip9KQjF~hbWubvSWj9l(M{KS&cbD0UhevDZ<VxT$ljMX1}`R(j#3)vU1-=5#(
a)l5(%NVb1+?qCCdv|hf-O8)uR`~L-L;2Wm^

diff --git a/docs/dev/articles/functional_api/unnamed-chunk-25-1.png b/docs/dev/articles/functional_api/unnamed-chunk-25-1.png
index 2b7ae2386ddc011e7e5b3072ebabf50abaf17121..2d9d87e20c415f5571ae425e1478802eb94b6768 100644
GIT binary patch
literal 400451
zcmeFZ2RPS%`#${X+t$vikO~=v3dxp=kdb6%6ot&}JsJ{1$;c*22npFVNcNr?Wn^UU
z?K$7N@85C!kLNj_aU8#A9sk>X9QRird_M2@>-D;>>paivyxt!2vJ$(<4v~>aq+ODi
zFDj5oTfdM<<ow(I!7Kaa9^fCOE!xr&7fEZx|Dy8aUXVyfNRk)NE82z)ci5;?G|!Ds
zXoPS-Pkwcge8&UHlUEgu^xtQ`y-LoL8l<my(qqto`!iMU!Z*zaOsPz4MJ*m=o*Ely
zTIBa#wKrW_PA#l3j0_LHLrt}J%5IIPJ>TN@POrCmqY;UU@ruM%t$!UUew|1D?|=O4
zyL6E5-~Ukcuifyk@1Om*zpUFAj`y~;I;r~McBvmzU-)Jc$(5tTmp(<S(AiFJ1L^H$
zmY9tsQjDaGjMMVG8SxS+N?KD>(=LugdV5oUh*(A9%k^a0z~Apkr0=hu?*7;J|9Sae
z3-SN<nxsw25?5k7m^agOLsnL{dqTqZ@FfX}ASdB9)zQ&WtD%N{zHxDJT3Pp^ho`X1
zpIpfASmG~4Z-|PD`W_Z4>M0N6oSU2DTykXNHZ?UhG&D^5Tv1UmK0bb|nvPq)=G7&y
zq8{ZGP2={gRI^DO8j`2j`Y{$|ZOgGrba!`e`k8bi^ANYbJjViYEstOE@)Q3hzCNrU
zv-NL}c%Av*ewkeym*x6F$JkgSO;?7QIqvoA*SLw;xHy{AS+TLPp`oFPiH7FplRYKA
z^>uYm4=gw88XEo?9yT&H)rwbO+Ofmy-Me?`>FNId{&8}Vfi?y?Z|kisE#E1|*G=_T
zlkeZ}_u_@ObAVI;TVv9V=FU!G9-f)1aDmp=R`J43SP^NT^(&KC!1|fy&ji{oHznOL
zGczM|`_k5CKR^EK!sDG26BGM1E32#b?AWo?7o=r9+!C!2-}@jmB;=#<4<*6F^z<%0
z-*&l)-MV$_wB<m}(F>2~7Z*cbE_e0ITxYwkudjde=Fi2MKT}gveGfu|gEQ{diwj0P
zd$#xBK~_7djvOn)rgyA|MI2_|$=@_?c^~cN<+a$-)}|)d+Sb<A+^j4Z5fWlQSjUuY
z<X!yvb2OG`X6<^4rqGQWH%fz9Sy`KtH3TevSKzuEyNuaQOiX6}G*kO(<yy0!K7G~Z
zP-Q4@W@>8fp^ER{&$+DT`J`Quky)A^tj~9t!{vK&uKfJ`Fh;_6v?G`I+&PuGMz7^;
z1r`d89Lr#SD}xdry2{GRMus3xEt9tN_|fm!&XgmR*t6#qAv}qBYj-yuDEa(wE0fFj
zAod%_`3m)_BmAatiAnM-%gf8!IyzR%t$fzQDt%=EeSWDK83BSPPMwmEmghEZo5mIM
zKhVmz&oJxxX8w6(WQ3`@E#D!towYoW{e@tx^+?;7u#Q}tKbIxl)AXthG!4H!J?K}T
zte$;r>G$C0=H|=;*>pno>9XPcYSDs%f@C{(6u;=Nj!Y5QDj`wQ-mX63@uw;2g}|kF
zMISnWv&WB%M;d+o`jv@Zahv6ux~iI*$&ZgR%LdyEr#?v^J9bPgQbdA8+BV=#<T?LW
z6FntsD|20b<(Zr51gr%u2cE^p$47r&SzgBHF=Zc88=w7Q8YyU3_v6Qpnx8e3qq$?6
zxz<V2K^!3uGfg`9&3m8NPLBWj>PFe_hs|f^yH)>2#&3#_j&5$Ot5CNxgtyc7LZneB
zuc`BJOKP?7(v`?kHN@WP@>Dez)Jsc&EueVYHWEp`_Nw?+T#{5dHfn9IaBaJ{lI8fs
zMDSVTkn^AO^YT{bd-jwDvZ^I0#+&27s%01q(yKFd-Tqxzfg_C5v5V$x1Fd?hwwtYP
zW5RWZxzQwshEMl57C+21?b^L-7o{(5A(GGH4fbx!9YbyH5BGk3*}Q}5fMuXq^tNr=
z2HLaJqKP7r_vLRy$Hrc8-@Paoaa!l`=uEj<2>0D%qwm<`5MzIuQ<_EQU6P&>M;42}
zgPJ2u*lDr5*lWmV){jj&Iq%*tvC8K^fBqa8(4VMC)hQ|4Y-Vj8xZL_o-4iFbz-Fv-
z%dSJ0AD5Mt<v1=(W?S@Y=Gh+Q`!!W9LeH)LO#QgFObB=F#6|Z;RHmPW9)E+E>ad{g
zz1iVb6uEMldsbE>Z5f6W4@!!Q!-X7X@x7g$ow)^$<KpySjNYd;v4blMQ~jZ5jfa|3
zw35^^uyA!z=O6yA49ztD@liER7gvI>NT*p<aq-9VMF<VUQI<r7c%>9g_K2qVYs{6-
z6CsOpW2U;gZ+KKxRObJr6nZlI((uZD_>DW~H)%h&b*uXunb40$t#yr!GY|I&UzU}n
zs#eDV$<NQfeOw#WHY76giMO|q&1gF!l#7dN<|#ILd?<Ro&iW^xP)@dWaKP^6UQbj)
zGT*p$cSTv5Usx;h#f66w7cZt8|7fbLOu|pS5Am2=TSxGkMyT&|ta~kK|CpVH<rFu$
z%=72ZkFNRUwIW6CWewucU@NtnnO<I)k`rDi>gaIl{r-GucEn+RJla-PUA-N<)ORS&
z@=ue@RJly(S>#~D%BySvLqfX6j6v@7!j3A-{>5cwLy0K``{k-CDu!#KzhZGM%*`7Z
z+}zv>S7vYEGm3l_hKGljS5}y~YH$7cXuRtXx9vzqv)V_a!NI}1hK4^-Q$~Jd(j3B>
z{z(~?{||vST=#Y4+0{2U^D;4AT8Wj4#oGpjhW_|)r@ZI|4UfTX`Ei`u4AU-QWV8do
zEh~t<YYGZf>}re=Ylys@>}(;s$=Bb%<2tt7d3C4ho$8kfEUS>9VES=E$N3s03&tQ8
zk0->}HYeS~BF=*R{5d8aIch!F@<yW(KR>?@R)3ndlONZYq2V*{9d1oSzUJAgp`r03
z!;rZk@U}z{PO@T(riNev4sof(?q9!t-M@dILHudWp~!P)-C=wd(`(N@D{DDX?%f;3
z_evrD3$o4EuYdQI?G_JF0p&f9%Y-mC)NC$XTg~mgt*os4@PW%jF?9%!<!3)8X_Kx3
zb~{gBUti(p(kkL9Pr|~(aSI`5joUsN-_Y8Hs?}ZE{be!XdSX$1f^rH+C-pRP6fX}?
z<Dn~3Qj3U^;(a-mgU5nLa;*OJn)D)Lbf4GL>;L9WQy%`RuO(GSjRye0b?uxI0Rc&|
zOvjJ^9;l7!p2*6|dir$d#Fpsm35vBMw&LR7=v|hNF8JjXmSlE0pKy-9<VBThGsbad
z-X)qhM=W}Lb{1hVmpfMQBkKIdm0b4|6l7bsZcTbl!)rn-WZzI-EobOh@$$5q=?knJ
zGU&DuRc`E{rr^tG&*Tf8SCA8>e=B7eTrV}#(&_>ZkuG?B=ktTji&t=A^J7te!ypi4
zGCK%0y!ei#w6yffL!zwl@+yxuBDGxh$<59lBwr5TjB#*qSevewGj2;~?Hu&unWH?R
z^rU=1BiGsqXYbMMpT<PxpuD%YC8*shmioH8&AW@Xe?6nb(mN~A=ob-z41D^+Lws&t
z+h6qSPnBK8M&>4FKA3twLf9#xr*VGXw$yCm%+MMV#g=e>tArLFOH_Q_<EckjSy%?J
z7HSWVA3q)=?!CW&_S8*R50BZkFQ_Vv^bU%sSPHRH#j*-sX=!P~^~qX=tMo>L__!#V
zbvPE4{v`L+4l*(_sRQaI3YKn;!J5^Z$E0EvMuBg3I33GN&&$jEXwso}$mXhwN|ybM
z!NmOc=L~MT14s%KaxHmwX}3im3Wv)x3PnZ<I?Pr)p$6u+<*8}Fe@HCs3kwV8oB@uX
zQ^LbFPHW0YN!2lDWoKu%OZ)ivEMBy|TNx>$C9!J-6%UxOu3afz^2PC-!p4xZChhY$
zUWOTe;*k_l)OH^hQ~^wzpi048;_Um?B5V}d0(>M6ZP}OdAjm%=!o<Ivs1#CvD+Q&W
zySv0`%|^DE@)^<0j~@fQmYZ&v4nFUYE~~0aMy>SYMEDsaTu+}qt=1DA9gTEc$K?#b
zmr}UO@4RfAG=U;Pr1wv*8@fYm@7;T^zBK}=+~1$VjD8lV4|#GR*zYGY#^VSgmGG={
z7E!n9Z#1S2NjG@;`PC+`PiO`iAQ^^e>FVlA$tEJ@_u66;v8+2Kr}`#wO_Cg!M9h1?
z*F;}nk~6ci;u+YFH>L=x#hKT{ref1Q`pb<u5~%j=J39I+!q2b!&mTj1C1QQI+potG
zW#iL>T#O}u8sh8Y<mgL-zI^$@U06|GUS3wVU(R7^Bs1yVyP}>h=cS}*NvYX;nHd>8
z+&)*o3mrap@26|8Eov72RHugN)vH%N^diYT`UVF4dR4T5<w+AjHCs1t9(-awKYnZC
zr(sHdQ<j;|!w>s8?;?x3uKsJW8xh|G^x_#99?mmtPDVlY=x1hRRL`@$YB;{SveX^i
z)YR1NxsZOh{_!pvU46g1ckgz89%@W%etY$6jXV=`W?CW;ckg_j-IS+U01-r6CA<$Y
zjSlQTaNuQ7P#i-8-i4`znp5+gUW$dO=@=?GYV;7-K`eyd-`G)9l!`S6s(rX+*CI|2
zy`b$)W8)5F{nI+%&hha{R)L1*XX!gkbYGH_lbh(}EL@(f<n3xNaB@&kP{1;5-E+7d
zWlx)NVYroRBFe*~^!~=J&I^4(OLJp}9Lhl9<)x*pXWk?wC9P`69AjVr%5OmF7YPjG
zGHQ9>R~{sSM9uZpZtC~0gJ&*QC+hRx`}qlQSJP!>=9u2l=)fy^mSoL*`=6ig<Nxn^
z7cPzFVxJESSPz3UFv(foyLSMaQbO<YuCu{%W~gzokHCc6*IB!(Czya-GI`FM>1oSS
zH3Q;A-WbRGc<jh}oMYZs<}jAeeAu+^b!iBXVfvPijt+Vud&S!o{8mF@=gh>aH*MK@
z;Jw3~$;<>U^o5@vEiG;OZ<Li4oM4f)<tc!7t}|zFG9T}u*Bq`?L<VqKUFb*p9gMIB
z3vl|lvWyZ;WO<XpP9QkI(Upl3dc;m@YAPzk?%lhqs;Yqbj9Ne3!O^X#m>)ZgQ(`tg
z#)i9y5-Y2$3}|P+Rz)1wX6-TH{-Z}f7Z+cu#_sU>(dqQk@|oWTnp&*pyuNw!X1vj}
zVb-zk?lXo>CnvI^yYw>f<`uzQCEzF}VW_fy$k~n^(=K|l4=JHL;?&2FAAt`89LR`p
z)QMgvgz`s6I&zgwbgra+`t&Kouvx0W|HTU~<2G)g9$J1&{jZOA*~=>J*tM%suhJ@b
z0k?x=6V=~~4+;};DL9}kDJ4a=b7#ZpqrAK}z%2ISKMI_ddP`qe*<3{ghz|ZmO^TOs
zy&JdeB(j!QST^eUudnWjifq~2ab+MC^~|!7!cLx7E9itBZIzS)qxAy(Ksb@SMF{i|
zj9WdxYoeaYnQq=EMVB(#Rk&8{MZG9-17!+u@l{NW*K-f4*gB&n;yhgVi_6&(!otFk
zAG{9>sU=6wzhIUFg5}0ReDL6(j_5bo9uP*Nd?@Q^Yd4s2XKrll!q(sTXms%KVa=x&
z(iapJ6<-{eu^w$#RZ(H#`_-B60MdC-$lmgK7cU)Zh$Z*E=Cl~EH{;{?<m9UNh%BEv
zdGbo+0;gskPU)vd+sO?Y-kvt?6uiJm@PrNd>*A6zuq|>7rEhuCW@p6O<;zcO1vxlm
zNB0A_bZ`E$ef##uuNPd-;zIA%zX8{>8mx;2ElEyJE-fv^QE~zOK}ne5+>~-76I97K
zT)<`&`LQyD2dQxx011J*Y4c{c$B)VCu{2DkX9WaqNQ@)#Wa!tv!kZQp6cCg;CPoMB
zPv%N`&`fV{ufFp^8k!S!><=$HD@OvJ|8l2yo@rDTvY*!P%*zJ%m5L<*khk})-MfWt
zM#D}`P4<<e+^EI<S<ukPHoxP(o35w_3?2VH(|deX&e?etgdQ+@5V*6ar)OkDBT0=1
zSVTY|7h&}D=~HYm;A3%7kveb(P(Hu#19^G*BMbTYW<BDk_oEQoX0Tc<=R&#rbOcN@
zf=f>ps#u<Zk+CsEhoT<bypCyUP*{BYAks*CM@PmtY7X^tAl>GY+g-(?0V?%d-rxEw
zuKi0tixrDkY-?$GQsPxxOO)ZhCx0ane^2B#etv#~U60+|YR}<EQQCh`FX@|}j?UEA
zJt9v^Ah~GFmg29X7+bU{@L$r?C_pBV66g}(M)ma;z*abBH6c^rn7}y!64v-F7U92&
z-OWuPn?Pb=qkLK}D^~cs)?VBuqA`94@oxTZQBhH-ya6^;_}8fe>oA1GVg;HHHt7r1
z!j74U-iNj85W+T|l7iyVqeo9lOxv@r&Xz_U09XMi<#t+}ZfIzjtRil(`24yuWE+3y
z&YgwD#XuV>mNo!=x2=cS0@zNR_|uxEhq}>@j0Lr!jF*?Vj7@((_|yX?=|Jd}uk+II
zs_R58&Hpn}>HnQF?0?=HTS+4QpV7DfrC87K2?)rAol9Lm1Sua-C|<sNIa&1&Y2094
ze3lCL@+wE#DHp&Q9%n;gi~&htVdAy|;#~g{rTTlGh=1Jvo1KYo|MN1j3i$fFCZe00
z()8qF72*}ykRsGOvzMtWgSjLY)^cqpwC$KTM_uXb?{7|2mYbRrMO}idRxZGa3{?7V
zy`90r4kUg4{CP-&Uf$kLv%^r*LSDRhk(dZQSOP2o)Dyg}ak6UP+fNVxktyHu)d64G
zblSz`P2S%JVF8e68r2)>hL?|ztdtZ|+6TafjEp4Jv^RK{so#|jGea|re^UM&JjKA!
z1Gy6Fu8*f@j@^{5n3$M>L4sKHTZLFpFE8nyZ{9c3^-t^998p$ouFGTBn4FrraN&Y4
zPz<2k+qZ9#bzaBD;wq7jL33?wZS$ZXKv*<wdKat^-@ttH{Q1vrJE-x#D7Jv#5rTGp
zAtAUY9Z=_=KWpCK`ijChGuG8rUY?k4D&zkB`}e}t#Xk^C_U+pz7bz?wDS4WQCnGi0
z?B2b%$|(_Xae8s6`<4Upkf`a-L)9z5C*Jz{csqk~vbrj&5Rkg`8bAcg2?CQb1oIdM
z_7?IA3JH~;d-4ALd*s%ERb?oSgIQ)h-QC?~=Xeh&gPb*(>I2c02?VW$L1oroN#%#4
z+kTswnYmQNgG;Zf!OR1o$3rgo%pED#_FbT63Gd&Jqe6ggr@epA((U30RvSIg5dXr5
zS0h$|iJ2K0y|%i#y1hN`yZ-Orzv-Q(uR1z926v>WUA$Nn7wJ6o^Ji2wyMwLmGBg+d
zSAD;K6Krh!=Vxnc>%~n*KR%AcN(Yq<>o$TXFZTF~6ij@hS-d?kFmPb+-e&;;u?#nz
zoeP?{D5<G~Ak&}8WynCa{$n9Q=0+^`Q;UrxMa%UFPVs;6;DHA@^GAcmA?)|1Ds@mz
z941JD8fUXvDLeuHcSPEOJvaqxG(Q16POc(!)V}Z?W!DrH8=&!WOF$?2@lKWNLeD#u
z)NY`cZ{<6f_8&ah3X+1#jr$woS_5?k%WTUsi)nup%%$f;%Qp!L1RCKbaj><enPx2^
zA))FpJ{{3IrY68lC~QnjOoLnt(5pde4k2dFpTFPx2;gOUxYYu1bf|Ck67ao$X678E
znVsLWijK>M%{MA*b>urJN0xeccxX8<#HOT}iitgFr@eNqzVrpNp${V?Bd+q*YY$|&
zx^=xnlP)+o*lB5YWuZUP?Ku>Z>TrR%FK#=~fB-ZHCIicf<i6_{6BA=E28jk*L?EY@
zrY!45%0|df>Gt&4`Pc(yA3l8O75@H!?5K!{3vdF?;u)n+MMXu3O+bZ8l?~E_a0gOp
z`c5Cv3PIkMe)mP{KY@=e`m1))@hg@*x*8*X9Y@vT_C$9vSYm7#HHmcR5dmRrtw0=p
zLq8tU;o?ejb#(>$F)dj59d6SJ;+*mR{ac6y0B1_?RLl^^dxYmjg>d$2y1EKaHox9T
zi8Yx6yNKb#xhB#=_oW}8o8C0MPoGnlN!+^i3vd7Qz$tciG@Z=Y0?MA8_BI)9&ju(J
z-#IWgwl+O@jbRIS&R6$6YuGCD6s2T!JBR}z43d(P0Prj;c?hc=MEU)6LK`sXN2aMV
zJ=!iK?b%aaoCVMReEZYg{XwV12U_1Y2AywD?Z$teS7jtqbUdN-u4YL8^zH9^okFyv
z*;rWSC(Aj#pHn~mR$l%m)U;sx?%mEH;Ls4Pt!G!3<^W|zM@HEBL<9x%zH3Dz=MhTX
za|Q`yJ*cAHJv}xfZH?)7ua%YpRPNfoopsRoz$=5s1oGoo=0`HSa91FJenCOh=pYT$
zM1w>~R#lglwidb+j=n^Effg-+%3!<GK70T<mJK^sTj=6aT8dT#ShKXG>DRqr^iYA+
zk$0hSMT)pY1O)|!g*jq{(Nln8EH&wmJs;e5&T;-;((`~5$_7qGn@O%@p6h!_iOeVy
z&VPM=5^UE$@tQum2iK9=8%wa>$hn+J%dX0oO-x21$0YE~uP%)RG5(HpSv`2@kec7N
zQ)J}iFMvzKMO=i>oH57!AYLtOY%)K5fTpg1C=*^D|5{TcpF~;m0Q*qx2C1i&;P%?u
z&WnT0A|iz#$!gHFa&ijRR+q3!PGk9VjMe*vH5A)Icuk3Z0MWVrOV*LvGSbo@m0LjK
zuZt~5BI!Hif{&Eu(1ivb$f2=_Kn5Y*bKq1I@^4^ZAQllUgb;Dk^`G~u-m$)0GgI<!
zFF-kuVUv1}<y}X|Wz)iyv-Z;ikUD?h#w#k`K|5hrOIOLdckA-y%dt`cMa)B4cihBk
zA?iafG4yH2DK7FL7q2=@a-9?<kZ&Ft@<WiH-z#Agkbg)l(|pqLipi>J;%_O2e`InV
zKmI0Nf%`Fv0j|(ys9_MNH#0r`ol=q-IwjQXYOi&EWO^QBWgYB~baBQq3^l}Gi<P3Z
zu&_X1l();#0?i%poQkTdCJ{x%3A#i}Gk=tk4^NypfsPkMUYFTcJ+zmaxxOGN6Ez2`
z&<Jg(uCA`B=_5!U(bf}jSp`Ymna>lGXNY`5RI+DBFQE2>hlQOyc@nyECxmyb$j9Ly
z(VMnzEwr5wvnXn8%pelsPNLPgU%bXYG&C1_yo<|PAp4CbWIZS!0c_V3l#<#IDST*y
zVsR;Yk{b3?Qu5eM_F`pov8E{U=t>_1>x8T~?RyVL1~LsnJ&~IrM^aHyad2>;I|2sW
z3SDztAn7DB^GIu&M2S-Bn>TNEP@R0{C~58`%gdXIbwmdmO8OFh(7hem9r0f?6rc1a
zSKII+#2)3Y0w<>WYoUU6<~WN*J${)+ckaw!-MezE%DcL>CJQLYiji*AGELYIbEe(U
z{Ad(hSXc;Qsr?anqBO_X=KOu=`wZ8|`8M6e*M#3D($BtEphl<mXFL6aIw8GO2RVa^
z1D&I@535$jzwU9fSXo&qCdX-HyK)&?>prQpQ_zK^DuZyqqN@ul#i6JP$0EUk7o9v=
z(b1tXc~U|`qQgc5yFF9EdhN}?s>zkGb617xOG_ouw_2E5>`jqpA)2XLVn$HN#2%aT
zW?2r_5k1aqwXWEgFAroN-iPCkq~PVlZEkMfl%h2@I2g}BE-hV-(9Ma866JWZuZ!n1
z)HcK=Pqsh0z>xL=A|jIISYq~?a-Fjs|9Pw$e_A);eSHa?gp4XIOSo;wD1A?Rdpr7p
z`s^N*NaP1jYZoE;T0PpI@@-~(Ts~e=RaF&f^to4iq08F4cPBf#yumKfn;EF8>Fh+x
zC`3+UopiYI4Ny|%3S*NF9iRE@mea)<%5Gx6Mn;T_(xu-*G3wq)97XM|4a=dIuIcK&
z*;miu!DsRNNxfl5PR3|jy0wh_8e@Nde~=8p_nTHqr1tLHcl9jm29oB%cSMVnvJ<aA
za{D$rI}5zSH85QpBeA1I!iV;>wzjzC!sH#~o0b-pQe|>-a`&AFCvlTOW=i6zwiDer
zjYFEYp)X&a!|^;u!_<}U&|Ff|Ul%Lwm!hku2XRqTKZ^4s(h=VCk7j##c=!sK`cB$&
z@{uNmOfJonlaoLs!ceT~X!cN1P(W5ehk6cuMn7_NaRS!Z7PdS_Gi|JJt?AJ8%nZ9G
zcap&{nnkgV<{k3*OQYuG4v1<%dzRG|m6aVR8>q_<9zNWab5TuA4LO<V&ys<|?C`;J
z_o96M$VLceL+kCY3a@vHhO9DHuvCT5yYs_(68cehq|3Ey*Rq*a7VlYDknbtW=;xsJ
zi@a{onP-=}QL6KULcG6GtO5sAo8pI;I?Z#fNBpH?e_zJ#tEjv@&9#vfB`E)QMdPo<
zJ<_44r$_VUp{yx3;ysTx^eS!bSEpZaJ%;jy`hJ9gAqL@(VxJ2^MM<d%D(W6O{)T)9
zn_F#wkZRuqCIi&3T&ZZ*a!C?SR*d&6EL?-u`Qq8LXQ)<zfmC#K`6$S~BC9Hq25L|%
zfRIr=HPqFqIMfeNQU)MRg@i0ti>yr#HA=rcd$*3ULuNa26(CYvcE6^!b~gwD7y<{k
zXbS`!q%X^U9e*1FK|p7_N-x0A@4Pb842=|646P49DJX)d8488=zrMKn3NK!cRT!*F
z=mjP%8qxPuS7Gl5mTg)-gR+L~TwHM_+7?U)Sg$%lC~=Jy+S@=!Zsz=F*D#<~h~m_$
z<bn|6078p;u+1^P!fM&(v8x9m@}TQgpnfRtvfqzAAMq`+T=DJ`tvV%Nzs9s|@HRx=
z#P&BFvb3<kC*4TZk?J(hx1V8DUJS}dlWQ3PwBA}G8jVh`($c1AO!K07Y~kF}Ilh%7
zn$6PRI8L-u6*Ofugco|rf2pPER<en2s=VayzK6cazL7;nqX7~p^A#1u^XR+u#ZOte
zxj5j;g6v0*h`zj*X`)eD#mr<nbnL)cqnGw#SQbn7x{~&6i^ke3SFQkb9OZ?Tq_(R|
zgpu*$vR{);JaTq*q)7Agk7Cg(83qBp*9p}KTYzT6_gJ%~S>tG@XGuxM1_lPt7yNrQ
z>Wx_1Q8=+dp8dUjecrTu3NJr`$xYB!ny^92Gq;J0gVf8i>H=)F$bI;^(UAFArw|Pd
zP14v)T?vYn&>RXXDyg%fP3{Q1-XDs7R-aLw_iiN?$BQ}%Vfi>u`diTR=Vn+!lm%%H
z@nof>kkQ$5IbfU8vwOR8ZBq7`ewK$mj=cwy2x?botf)A7?pzk*1XFI&l?cIBl?Cbx
znwp(BKVX8m+~W)53f0HYb_?ks>QviG6A3X)-d@vlYX>+R(*(K`zf|(J$<h-E_TI9&
zN%LKJCjd1n1vhIsPPo-{3i0#Xk9FqX303nz>tDv{XHQQ|%;_f&AAUt*L0dUaAO!n~
z?gnFd2IOh-CckngFPIe4K*aU*^s?tcV1K0P`Q~Vyx3|yfHP?}0vTKI)2DD8xV+P>!
z@WBIATs9vaUEPnmm4|Mfo(EgF9-DV$1t3L@fw(X?w@VYRP`N8n?<^S4o^6UMZhHVk
z7|v%QF4*}_H4V8M0x2v4=)w9P;*k<$=Ha<M${NK8pB1t`y62|^1eQ>Q&|KYD%?C;L
zcgk@$v3=CkaSQVofA;ldfHYuhQGPp70DvPYTwd$0t*&e_)-<5#a-8z6+5P?7w{Ks+
z>HxrW8}El-WdHvC$1=CN3R7`vh6M(~Ly@SI6d~k5C4eJ=Zb{Z1q#GNXIb^c2whU>*
zki^740G+CSu>;IFj?n6xh_=xDM2{C;nl1Rg+l{km3|LticNMIljSl|99S1Re8K)9i
zsh;_bt-A+A>jUgMBvRRSLL`#^b@$_~+Wc#=v}hgwnup9U{=KwCc>d=@P|DK$oSUEj
ziQeOCn>t>8Qb<U{=}1GCv^I9eFGCkyUsF@ghCB|f0`*~O_?U=Zx0Cv9<`us@L+H$u
z{Ht3xZE9(4EiFFH#f8?=SVPIxt5+eK<fO43t&~%I+lW<X+<*jVkt9LAVZ#O>&d1*X
z%L;63P;-F#piN}rD?Qz)J!=9v&doIF4b)nXT<gncczAY^kuAa`gwE25>j@2bFO*Gb
zbBTH{YEA=ZRt_093Qu(RA_QzqYNF0#=XrQ}Q?4gA1!*~FyR11Ujn<L5bwhNoVb1^1
zfVLcHtU3&Wfbqq<iG8s%L|>M&%{X5WZG}|rVo#up%BCxip|#MQ)_yea(fnHSg{7&P
zI>$+NcHG1itURT<!hZPeXr4)^IzS~laa%~D!Rx9Wn_%Ugr^Sii3?c$TLJS^-J(Wpx
zOQ*82v56M>1okbJC8?vj|7<JNJO`w-?J-W{pSSXBQmV}{GBE|bu1|kIt&(_|P07z@
z1igo$jS%&5D*hDIz<F>YoxZl6O;jeWTn60B(w)t3-k9yy6#)T(XpsUn1qHv}1Y<g|
zOvXr`Wh_{CKnKx%tqGHmS5dJAk#kDsF=${(l9Pqo&VXm?YQ2H86S&|+7%YvRE`~q?
zl6iFtICyq;{I3&%Elg@(U2<8TkVs{C<mQIhJi*BcmqVF9D>ys5s&y=(I1oa1uC+1j
zgcHY#N=lX%7KCiaMGf=N!1MEcP^JOu2=46y_Op_iL%Al)a`L4A@a8=v(ar15N=4g{
zOK;J$(7w30=^{K8bmh+FUJCI>jR^sxe8R%QLk~e9AZpdgm7UA69!U_sEGaqIlG+2E
z0&(C)T?WXgPSe;sG{i<5^qQ8n2<-O0&u%s*$L2={w$RGFjG%cbFx}#@w6shgEoEXf
zDYymq<(HQrr8V|rVC@qFwG1l^M6*(Gps$a~=yL)HkrA4u-DYMlc1ea*4s7;Ia^QBU
zfZ}66^;;%sJ65W;{pI-=<7r1(SXlOoFtFNy%8j6X2Jjad9&S94{rU4}XJ_Z`$hWxA
zRX8=orE2EUUqh#M8oHpweBP5)kU|28uTpA&?}1*^<cL@YOUlY>7C7303C}n-FfIt3
zpBhsok=TwAeU7Lv^}}u&Pc`?d>GARNA3Dk$gJh`oyx3{ao;|<Lad0I2bDasn=h-_r
zB>JNAcBE*DKqExHO;*bof|A_LlfVx}a3CT|wRPWI(_R#r$B!Oei4^Xru1>*Q0Z0;t
zfn8q$Kf+kVPj-P=#Vrhkp0X4)Sa^l|ZQ6b|``C_v%X!-}TfA#|7`BC0Cfw1}qiJ}j
z{^2)=a&(}BWaO88{Bdu#pTKPc>r`02DCM4bcw8;p%im*|gGTW4gMma0?mR5m5_RpY
zX=gifX+_1AqzNQM(4%@ZSZrGZJfS%PTpXGf9YyWHL0C6TnC4A<!<plWoXQW|Pfm{P
zLx-MX@0HSgJFDfU&1m+n>q>~h#rQ0<G+}OPN-9-vc85i7$7Z9k2!DUYkii2lATxqX
zqWjJA6hTLr=U^fOVfv7niR!fkP%?8j;e=||bB0XMWiV`_sh5&6hD3`inuWRuGX|ZE
zdC`u$n^B+Y>g(MMGNfYJ=9PaHDjrJLrW)ILaFN_4(z|Zb_WQG=Y9~rEjapx=^|M8Y
zxNsb9n9x%-XHa$eNUdaZirSZY;pAr)2Wsg~-N9fO+7|V$tP~9Z@8i~1kj<G2Hv)sz
z7v`3uAW1wuaCCK`)OEu~s2$fPw{jK#?(b(_E-kC5aGV~f1*R89jxF`}LXSIc?P~=3
zPK%3+1|$55@+{~KqN^Uc-{h41e(doV_Ws^p2y$TvawuLQdloi9aYMMAIP<1gf*jTH
z=`#*A@!Sd3<n4H;T|eEgd0fKdh=|Zohm}REnN%prPiOu1nx18O6O;R0x~jOC@LO%&
zO*_z-7^LKJmWyi&_4v?PBNaGL02-wPULenalQ%{Mp2G=&MW<JTOaEX&#A)wK?caZ}
zxH&+hZ#+P7w>9=%<xkbOhJH{LO6p0KiHA4z?Af!>L(pIw0|6JiaDhqWZYpy4u?z8v
z380Xd3(|n7JF+eG_?l`mR8&<V1H|AJ@aHF<o(Ax9+g1VNOt*SzgIYk2-&^PeNHq`b
zp@_yjkQ(_I2m}rfpgKZwMSHWG58`$9y<ZPT&voxP$nrGZm;2vs&9~z?G|aT_j~^e(
znfigW{-`e<B;GG^_l%;HgTqjLym(rs>%)f^Qd6#L{&<1yvH(Cp-D8%E*a?SeYg*iI
zsS)rQBvnHo?ARbt&;`_Yw;yAW?bcT33ZaSv&c2C@>+kE^zT+%f2Z?f@P;n8XVyd$M
z>q-Z;=SDlWw150n#8ilWd(6erwv5F0{RwS==V%Db@HN*o_W70PFg-lrbyfKHq@N%=
zoY$>TYKnT0t{R=Rp>v@yW@#6bO-KVo14w&&lu9qIT|?HmQy%D10#XINJbI}e^2%;_
zOO~U(fvo98obwo6di>ceET$mBNjDXEImQ%+K4%M%y0x|i+3Ic3bR;V<<6<JLUvTfd
z5k}9WEz>mee4J^^`}5~Up~u@k_&s#Fqx=!L>8_mg2;^3@AQBv^a7G^WKr_6X&(<#X
zygW>LkZSt&-Mg*GYn%o1-Oyoy#6{q+CfWm>*nCLnR5~;lgpxFJmGnDu(T9gZg4#O4
ze&eGhw7Jw&z!GxZp;oRq#<38E_;+wa*;bW{<^{bu?{I(UBalzhbNudr#jD$LPXP+c
zD+?_L{UXjky>!MFsc)yf_%YQ|c4`22iaiC;hD!Rmpf928g9bt2U2oXpA$sXzkH1yc
z3^1CiMo-=KmWmAll_!?ET15Qmfe9S5Ku3p8g*!Sr*AkW3Cv2b7c@`CtLwhoj?_wvB
zG{x2h+{By6L|+h)UNc$>PIIFgx4hE;{2;s)TX6rN9_TEbz`?WEY(BECrCgb#z4ag?
zBLk$eDe((CMfTcl)()}LFGE6L!I*#<5=yC}FRC)(cJ}A6ne3IEIqyZqPV^FV-b8f0
z+FxJ@oOPX^07U6FyLi6ij*@r?E)tV2q0B-7YC+f{cqA<S@9?0>?YWhKfdSQu?J!y>
zT`;X;m16B@hPrj@j~+b=)qrEN9ia||Xv?Ne{;~$eN3%EOf-0Mrm%^z4MbrhDC&}!@
zi9}R@r*r~M^Q?zL+Y>CZHTI&K<Gd4A(+O!Ur<q2mdpz`2`C(_cxZLu6cqp?czT#>!
z#M7;XQ3=sWK{rWrGa{Ultr#6I^#aH22xb_`G*x~scRGw=`axMxBF)QfbpH?ll$DmO
z#jV`)BiBHSmFZH1Ax&XXhkB6fy%M>;un$y;u(3(0uhtxE7YojCIS9e#-n;CB@65Jq
zY%AJxlpCyyP~Yvpoh;35$}m)(K|_$xmR|K!mWu*%XNVJMWI46Aiy-w+Cttx<cKgo|
z&ZcLsU}d;~avpW6hleEm!aMVm&)d%q$Mg7wh6+G>r=zQe*bN=M0j)0#21uT*F0HOs
z_RL%O9a>y|4^|8iVU`b5Gw{#^om4$`I7mJFojePq3@1m&BKqC;EG-4$C4tXpXJr~}
z%-y*hovVpph=l%+Oi;WOZeTyr?FOCCRvpt?k#2SV&!5*ODyJaQ47?!$z0=HJL>f57
zU5&PexMO+o=+6f1;M&TVOU{iE2vBknf~w{|?KJFrKho@b;$lj$iFLUo<sZOPP>CH{
z&hYdSNb$LzOz@dLbex8IAdW5r`1sWmet9Em*6b(O*@?!%?+%FaR6e`ZwDT<bW$EbH
z3ZU#qzZC*Bo!E4%s{uXQi{j!#Pp;iQ{)HX&c`#cc>k;?d(8>J6+mw`)h91I(Gw|}P
z!>rNDQYXmv5}`r5(<9N>Oe6=SW_H?3w_jKWoB!sig`*1nin^x=sY6WA_IE`v`em!w
zNQ{x7-obW4l!Cb~7tRPi+wq&F_HXyCcLz?B!~J1vXNL+gP!@0k4Ivn<;2=W#6TT?a
zw3l#e<1QJF9h2fYeI61J<Wr0`tipsjzqB;glFGowHiVTyx9Kkh2OflGco(qt_Mf%6
zP6s(bTMQl0(?>$9H}Ec;Z8HQ6nw^+n4G^fk0kIoh70|D1!1*{(wxGV~Ucn{=WwErp
zT>A3ma|Dp=a_S!(WNqE{7L~DEWH%BcKDj6Qf_wTmc6N3&V|Z?Tc~qBkF#poUi&tKr
zg6S<5xB_%X$gH~v?pGM^z^>6~L7n@6FojoJ-&&Ad<;D%zi&Q1XQ6)UOOp=WE00TVR
zY48o2W7JoI%&q1B8XQzTyM5;&?#o$kv$GdK!+C5TH~2q$b_!>@k`Y<MhVHrBZW^iB
zNl1WHa!;N<Z5VROgDzYu|3&s8x(Q8A0%AArWuO7mQRt#2Xbi6LB@9{hcW>{Lep&(R
zD=+J#FKnF%gRc}d*-*0zNdHVC>dg)d4vn1GOH`f^+~$zDZ|mum%X}{_otT;;PqtkL
z=wEE&p*QEGWayOaK3audTe>D4Ov>X_gscC|7XX<I=Z`u%I(?I@3%MU5;@$GWhC=(Z
zY$pf{5C&K%v$Ledp08P1=jhH)bK`U*jeP|r^AeOcKT!<XccKD6f@Cvs`?EjmwIV%A
zX!*Iv4fXV>$;lafSVw;k|N3>*&ISSvx%;kd+iF6T{C5BJvtiHvaUSU2%TZvuaEA~M
zhmhmEJZ+lG+G?>+{Ii8I1N=cT7c8EhX64ugR!MsRK=KlGIKd>fBS2DUr!aj%Ro#v(
zblOBUR>6aptSJ)FJ^HiTy%M2SRawcdneezyqq~{nf7FXpd{D!Pc?YMkprARR!9wI1
z5lC66?qI^^78a=Njwli?E-ria_@Xkz>xI00dGf}`D{o&(g$UYxSQL#_m>I0sL$O0M
z5T7I?BeQGIp2*WW;=m7wh3pwQI6b!q6rk>G+O!G!c;R<r5gJU(LEuE#awEo%9Qlc)
zi~EK$38{JzW?a2Ur#BIA6xrb5OKIg5$0BcyuKtSLiCRZZOB;qSTrYq)(BHp|@91->
zUB5}UaDu`U0gkaG=mgi9+C=5#h?quvLP5d%@S3#S_LWiI8vFw`OW3=s1jY!#MJW4P
zoNV`K-M2IFEX>@OiXCcA>j(V8_yW;VeDQ*CnZb~QgOJcH@MrbvKL2G6LJg?lT0}3K
zK%w<<)qv?h+Cjm=@tSN73zL0VUB~HK2}lsQ0+rU2YSh%?Kq+d&ksDHhXpmVNYtsCs
z`T%7Nnv;zOJkZ9$6v+<=oVNx%ipM)_y6y^(INV_gC`13exT7OKkDT=y;xtIzb&bpr
z;_h>yMYn3(?e=q{#wI4329g^{@=9fyo`_FProMtha5(h!^&L#V5$^5nJ!&=Z;8-~m
z9lvEwJG;CIPtIsn8_X2@bq$i!0ApT_&SaChX=6Z&e^LY@*X<;wn<gfQDJU3?KCopH
zZv|JQm;~*_KgM*^S`-WLrbs7S@iZA&21v!q`ThpH(NAt^Og+@WGNuL@5xRFmf9B<*
zX-aJaiM414)6&spaL^q(l(2MiML8x>F`=PB(JW=>{oWj-3<#2cRu>w&2L{lEIiIaB
zuH~J)I!dJ8wCh)&qa7f!GF?w(pP2leiM3F&PG+;*7zt4KPF7?i4(*pOVx=!rh76{0
z<RIn^_^89+h&E3WIdpi%JVCGfn3r;``)UB?Yg`KW?*@)1V7jN#*iG2XfsP>J2vJib
zU0x`uK*HPtw|VllL`+g$|KO2kf;uXMfi|VC1NQ;qZ{CcO-w@P0GD3&dNp?=Sj!%M+
zid4GDn1<AQEb}E-@`jCDAd&>ksx@QyLqB7&_V%4S$leKPRLHKmqGJLb1FidPz3Rgq
zG?5AQ81resL70|HX@RgWyg0y+^jUk%nk6227$Ty~Y1wxWv5=CWJaZ=Z-AzR&3T;FL
zxaabw$7WFbU?A*8P@vy?Nfz-C0Y?T(IzEGS`XQTNeEeMxg@(}Z@D2!U=%{NwzSMgB
z_?-!t8$j(KP8#|4R-oHkgm$p*x-%<|xWbs;8ZhG*rA~9$hS`mp-zh=Zsr;r+A}Ol>
z^{ly0EX>ctdGQ<R@qe&HL9Si5L>a;`0+|M5jf68~h#q#*Nt9O+=jD0CQbWK$U=+~8
zQ1?LN6X(V;q*=qPi0NE(F2RzCmO7RdgRqpX&6lF()8UOlwjhFJc4VCzp(X-z;8ycc
zL>VM}8sdsTf{}8k(A|Ow7Vjg#$JdJPEgG!!j^m$^1&oc2XCyVF<r`A8)^OpF4-N0#
zy9c`u`ZnmUUoRdSGy@SUUxPd`3RMxx0*0-i_R-POqD>5(1s6;&3f|M7k0G_fn7!Ql
zLJsMK02FA~gP|ejlW^PUVs~N*F+{b?TXoV#CR%<PBZ81(0BwoJAiPs6Pz-@r-zu_U
zWDW_A_io*5g1P2coiOjq-6s<vXon6-h(pDeEn7HL-hZm7&|fQ@$LGQM;tXJZL5`c3
z7kK_Rwsp^8!2?L#C_Hc*VKyWvh+we8!|mX{06MdS^(B~QRcSqxfKQ+kn7BlZnY+Kw
z1+a$Zh9OUROzW<<x3ta8KcG<rd7hZIz~13*2tjb}Yc}q{YwNDXO%LDfV-N?WIo_>>
ztR@Bf39`SWWNB5^N306WVAj?c^(lZE3~m^(J+0YFesn-%UcE}+g5-a4?x}nx>bo%N
zT~;4p&v9nv#4&y$p$=4d=;fWA1?b+x<ExLHM9^Kuc<rZ4r=HGxC&Q187rf|3uzN!{
z3atJFE@qlYNlC$orx2eG0t6!+b^vD&ShFrhV$X_A7aEE$n!ZE2Eqm7vwJ~0SYD`2B
zoK*X-;n~Nwy}tgM`i$svfboMA6tCr34)GXRV(BMkx7sg^<r$)eHMB|JgCazzohgp1
zp=j|=qiP!Z<U!^BV*O9bYaAd1j-e0y9O9QQNm}o+=jrY5#{rCe^M;H8*aw>ped$EW
zyCP6Vw3P|5XrF__64rwv`*Ft;j5T0}^_G*a_{EEahdv!+>v>XtE!CfHB5Bnb#H3;{
zbW-ZKh_s=A1Px7S_Y7}}Xpw*lCOQ@=lcP}q;ghdCgc-G$r*3^ISGoY+M)(5W5kES}
zrR#rZ7ahq};qP-7Tt5T6&clOl-~)t<L7CLN$%Q=-QTdz}(|>;lFoJ)2X0j~KNIxlq
zNC-HzTAHQ)yOM4sBOT%A<tGmE0Spt~AzHMu2<M$$-$s(_zovHOX(8~#qZ7h_E(v?e
z?bU}>VSFzF0+<|IG1Au|&{Xgb4h!@fH%^QO6TWYs1-$#2MVyT3uCiAZo3?B*7`c7Q
zv@<X0lpN;oz1MBzuHE-+YzP+n`#bdM(g)LLVpKIUK743<f9s(_{K*o>vQJ2r=;uD-
zBu2MuWNRs}!Y~F8wi_qf{*A;BD<l*^+Tgw7mVl=Xq-kKXY%8k7x|@Y0dfI8O!zN*^
zOJFSTTBqcSvVLGKVYL9e4bVAr=fYH=!}I_#?rwhNjS<!keZ6>J!-lsL^>QKsrow*k
z7eKm+S#bCX|Gt_7Gcm{<9AHDZ%Hb<_b9WE2A-iLY<|}%fC&x~fDMGv<eCNcVvC8@k
zHt~zDg6o5dd9R7lvrg7eH}eef>Zio@`CsA}{^x!&h3i^Lr&Ec<b*ANw5u(p@2QKk*
zusp~jd&}a<&FV?U7R3^nk+OLykt6wM@}j3xapU#Wn;SH)?n~W}d~6$#=jE$|QhV6z
z#Dcr%7xJe{t17~;i1~Re3c7sG?vHR7Wem=q%I}Z3aOCvghc|56zdq<j68-OZA!=ID
z27{g>Kj5x~{Yar>2kB>~5gu3)Kl%@;P*vNK6!jhb`w3_lGFF_}=cprB{`F*%`$Z7{
z@wNBz`j9Az^uK<g?pntNyk86o7+gtGR-CR~4=~l}jpqQ^lnTwYU&}V{Q=Hj4&(7ET
z44=K3<b0>@%ddM0OL*IHt-}t(sW(jXd2C9FqTQSlnXrnN9ds37g2dN0M#39`!zqfZ
zB)<Oh!hhD||Jrfzr`qj)tpls}bM8*DCneaJto$u7T^L=mr3pPHBI2-OA$o=o3!I~8
zKxoXj=k3Ip4a+guGs?bye~CuLWbQll>@%O6$~kg}T?O#NiC|(}=kk5RkD_!9Dh=Tz
zyp;h7`s#Y@y#9pqjIyrC(j_5#?$;w|!*sR+2&AA|M3g8Qp7`?f^$S@x|2Jf2H}QJW
zdcenDWtp&nLl0&{o1Y{2y}d*}cy@?@0oUGCY&cs7clCwce6QsfF=J%r0q3o217QM&
zy0`oNKCT?3{stmsMayV!Zqvjw5z?xKbKmp-*Dw3?Ee)rgP<$)@vx9J2z-=iuj644O
z>8vlH&H<zRGiD-=@bkQX0A9<>vHwW?+IrygMv|!C`a84Dpq&TlWOK_zm_-$<R)J&m
z99H+$6XIeN`3IsNf+K3U(p3)pLlT|XOXM+DC0EfrH2YzRWc94V)lsb9mNxc+09Xaq
zgW%ux_V)W`@J8n3fByuwR*;n6hSckKW%OS;W4RL&tuk}iJu$xkfQ>9b6u8}wjqE3p
zdQTl9t~eh6^ssfyElH(<pAUB2#&S^55FaRi43L<!G(^{O11aj<`UbM>gYBsWt$D#j
z9@9=o48<4M(4)}Z;$HuKZqv@?ff(Pav^I3UJfpd=e@BUgB;_fnXDBHxje<Sp8Hbmj
zeSM9dm8fJYC=}#l=6OH!FWZ@~gq}@0my2iLh!eNObP3`MtSCWI4~j2sAh9?TDJ^P~
z<X(U5UJg1x#jTj_-L}4|`(yEMJjKC`7ffm4*@fM#zmXSRi49>nZ_xD4ZYVzPTgPo;
zn-s)|)a-xKtY;EttO%#s*5WR_&y%(HjG{%LAb_u-Lz*N~Ap?{Tye0B<rA7<<?3kYA
zy{_>O$vF~uq!zSEaqdBb^bYs0@uM5%Ux9*a12V*q5(8(VW;BAf<9}<1LN#%#`1;?e
z3IE4c<A2WIU=eD>4wCj0M`tvvhfo6u_mfNr_@M}_eV8Lgr=knz7iMkfy1aT-VPLDo
z_-SMK3MHY#z(9!LgVh~=`_QJhSJ4%TfkFpkDC~C&gK?3FGae|kXvvO33vR?S$)Z$Z
z-M;=(oZG-stCp^xibu{U&SBeq!EYTpteRH_<y;aVN}ZV@KqriYD|nYqOub*78_A5u
zB_V8~l_{fHh=)4_RdAp&;D8VWkqAQ))$2p+06Bqc^bMQ@W5B)mMQ|*_*-YGmNjnju
z(D|w&gqH9?6AU1gbz$K0(oIaZz<Vrd#mTba3u!yzPq5Kcku;ur4S5SR#}veI^Zalp
zQXZ&J<Pg`qRo38}P%`WL=@1dPs?IS~&z5-8D8HWto)v(sEX|SK8})EE+vDa9q-|#)
z?>uk{nWE7#L71_q?iN`dN0UJpMjf;>I6sYoZy911BYMuW9Ralf^H79E72l->XXM{p
z%fU$uB&?wgPOV{ibsrsU{ES3kMs2!_F&xayU=Hf}kk2#s>FMcaO@|~b!6kctaP_>&
zvJ+suN$})9KfI0Px`{aG@(Q$ez0ZkxAMnd-9sB$g3LF<C4%@vxu-o!?upYQ4AsH0k
zI|+`s{wWvF!5sw$&B&&Mi)bNr+4gv2o+1m09$XW;vk^EuoRkA6lCy|ma5M(ksybJX
z%YfsXx-8v5>yRitkbKZV$TVy=K}#9wetTQryR@nVJRsxr^vHJxUs_166SO->uKMv2
zVRS5mCr^`+ihF7$=w-$|1R<Mz-7RqPGqFy&y?)ZzO6uM2kd#KX|IDVSRqc<#W!Qnw
zh4L6uDJDU&MJC_1SUWxPgIB*Kib*`uw%a3&3>78;4iNg)ad((3Iw8)Z2`J@w>#T0Y
z%P%gJB-fjP2bkQ6n)u8f<u>QnW10sa3&y|ffT*{K`Riv<Zg9|fA2{9pQpYD7KlP&x
zj);ip#CgJ47OdN5Xj!5Re6Ok^4jc>{XlATJvw&!mxmWKcD;~0f<sb~hkhCF$5j96W
zdlFr2JPqQexVZQ&Iwmdf!o#*d?_by>G>y)xR-Ubyd`3#a;!w_?<Q;AH1JUkOSsQV-
zyTIV^2_o1W43OZ!=7B{*ej!WfLQ@MWP~j(c7<ye5!h1F{h2y5=ckbVLX|3CfQwk|Q
z%RNy(`il&2M>?Edf+?Gb=PWQs*uKJHw)?n%{U|byMpQ@358&8EVlcYqYE!bt54>~B
z*--AgAE48>EW*59nI_WZvgS1G0||ihQ#mD>8{R!1f0-37cS^`UBBr({0R){|%W>bd
z8V<|58W(z!>k;vo7hm-e3YSxH05D-EGWrHNU(<1-7%KmXqOTdy!0>bsqUVD;HT4cp
zu_}f<JOTHYz+Ah<4#%Q$&O*Y0B|-0GDengTh|tF1{`O+i!s8F2q*L8k`?7=c<aM@M
z8=~3uEqo&pxN^zmwZvv%P(3U3Lt5I7Cf#S)2OMiojU4XLCX~xa1fHA2MI=U|_KSMr
zv2DaKzZq{nJ^{hXS)>zgKNL^s1N$UspQq%b!N-lFRk+;8x%pi8b!816$NBLycgh%(
zm5{3d#R@RSa<$Lw;TSkHv?SK^hIm%N6Lw;h1Su#)*=6JPxAI3#?0&7fnj0Ytz;bo|
z#zPYVKV2ju_L(+zM!V(>5!-wBG+^JM^@H9Q+|0$=uWWw#atQ}T?t?DQH=>^6`aJRn
z%q`!Pd}#dpFU!e^02yeP6Cuo~=k0qPYZ=9)8FwXE;r&_irJoP?K)xflsgI9~2(dkd
zuCiNs21@zsmgf?O4`vF|?v_JJc<<vE7<d+5KLj!TWd`&SwQD{th?#xfOpQmeaM+Z>
z520n(7tj&{25C~W;);JaV<PWk-Y2e{M`F=l962nb`qN0roGS05K2;uvY?zhRcktjr
zr4+yqSpt4sPL1L#BOZG3k8gxeBOY5B)mn-(O+3nn_-!PA%$5w}=|dRZ=x#a*u=n|m
zHSx5|&D84hIQbA9h&VUX6Wrq_27%ZFRY~y|Iv6^Jy0>pnBQKjUm~BlU5-BY9WqK)V
zg%Hw2#y@Q|<6RjsbvdCDT6($W7b+k{huH?wlVd<vt`s{C9o2kt(4K`Nr!VkEci<Y@
ztdN(Tkt*StdmbK+hY5Ux2AhJ+fL!Z+W3B8&iXeh=$_n-tJ<Mm>H@v-yq|kz1GIBN4
zlYQJelEC~iQdB2Z_Sz^qw`U2WZsJZ;`(Sl<JV#rloN1w$q&5Q;70;|O7=DBO_C3h;
zhGcvk)zF!&P4RU#fZX0o1lVF#LWRoeL(O%wdIF)H4kXr$SHrBB5})D4){g3TCrINX
z^!b)Wm{~?(PMUP@#H&D&2#mm7!sgqm7wLTi&8fvyH?w}7L1@kJfxRN@k)8d33JSfG
zv8(?H(m1-ON6)rmdApzz;hBh<u4i9R+bGP_;+Zig>n?00JqsmZp~6X+F-`28w>mM`
zKyu3RQgDW-1atJIb`2OApzBv=h4Q1@vK&qH-Llf$t3!D?O;)mWrigZXiS8n>Aqu!J
z+f#vyPIyjUY9`qvVGMxP;Ec6jusjPC@dV^>@eUdmOAww*Z8*1*1Z@%ZET)^LBXa}e
zQBrX~H(hg~CvfYX0gucr8C1<^?;pq%)PHN$Zq@|f+W|cA#K=xjUlY585HD6b@u~Y2
zGRiceDWgQU$cVs5hKx<@bl2IbL<I2pP?-5O8zi=pj;v!{Rw|0`n7R=e^>5!SkG@hC
zQPAbyE7vqdHFbxGxHz{@BFlec-*e~z9WR;4Y>STXx$Y<5T{yFTK4>{V2mKIorI>=#
ziU|58Vcd5W+80WYpWD!k!NU=H6F)!LuI6x-f-FfhKj#2=faq?5J-k+F=aUnQmdAt$
zFc?|i-t&nsj~{NJP3L)=k@bK|ssoeWlir9s96^_<O5W5y0y^||ndw>D{c@w89qWo*
zzK_QG6dt)!B2)rjf4JjhY4M9_cd_W*bo`r)**;=H_pUE!8M;91`q$dkvq%qqdrRgi
z+*!Xj^fqkTCmk>S;OpLfxe!ItV#|B9E+#=yeA;rL^#^I%vo2gvN1i00PWvS-W<X1@
zC<JkDT={0qe^?`kzIEV%b|1lbDdl+oML3T1=~jhhLwL^$#U<8b+c%07Av5;`GMfnM
z8_FAAGR7n9QaASgt9Ml6vPar?9-oWZ!Z^ksg}?<cLQ~nWH^Bsv2#T?g<8sfnivQ0$
z^j^Moa08wF)BXr_=hNTpiegF+gwig9Lr0iQG;*dl9M;BDbj-@-`1<@eIczrGWsvk1
zmpRh4wrXRvaEV&OKQOQfuJ;xhNe+&Xh=^6x8khiBMw>A09gZA!{_BvJ7%)Tx5>iwW
zDwlZb%qEiSGa@>q9xz8b!d#SZ#v5*4#)^4OBR@$z=`%JZcQglWhqDRGV9H(s&O{yF
zSyqOk>kQ{%amRGz+A?!+6@?pId@#D?0M~*#y`990Is*LIeN>GWbAuaV?n60;R|f?U
z5-<a7uT=d5V+dZP&Eo@CT@k8;a9`X6VkSo9U6$?WsjN0!liO^ZzdTuvga4Hv0{R5p
zMx}ZQkJ2?cA$0pP-X;Ov!O^ALPKQnoo_jg+_Dfi8t65MhQOwX}*gyUHGLFj#>a-Wz
zJB8RX+z8E$g<@*0ojH8_uHr4w&W<qqb}%)RFawq{1re!o^Qf^iF+_;_!IUB2;OiF;
zhT84=!|fs{P>rOCEXVW(>K*ZrlugDXO-Vl>AERkb9A$!ABiD0WdvgUH24DW6x6+*X
zmY6s|L9hogAYO^0B9y(XK6g6n7T7kDvww%+fCR!WiGaq;%_876o;yI9xeC}684wWA
zm3HDHX5D<*(TcdGD0z<78K1PDw-K5pGSFlZx!lu^X&{g=yWhcpW(N0{f}eu-;}&QE
z&<M1M@!djz3~WXws{i?!E{8E8LOXaZdf-Sj;A`BXHyO#9b9KH%#H95@-zSj*0?oF3
zL!%JY*us?^E!h|Xb_lEyV61~lfA7jsciR1#gpV+AetqB)1GSUGLul7eIes3)if^>%
zgO+)Ymb*fPf(&Ml(N8~oUm^oxX(t6)Nl|-u9Tv>mD^j^IbB2&6q&zX}ISXr2*M~dO
zSGf=Gc1Iu-qE2kQ+lvL!!WjN7tN5F+WKF{^(5uQ`T71cc9aotPen43}26@TjFAWLC
z6Jv1}&=*pq@a9y{dXN4dCRR_jj1qJ62(4e%@i!-*e&;o68HcV*DW~W88D$9aLnbyr
z(FVhnHd5=mT#k*!Q<r$qUqC&jyg7&eXIK5esQ(8xHK4)y<!{qqe1Bj(*tkbf-gTA2
z4JC6K!zWkdZ-IP3RVZ$Xe3b^}1Be69WYENKOIvA?NX~`p<fnTW&s)Ro2oAsh`gk`T
z(W7OYTSOMbU|=1#w@Bw397bI@|2Iss?|^RDeawd7g@6U14@4Q3{XkQEgm!m{hCcVP
zi6%c&_*vH$s$Dqpr|a-Xzxh<rT)V{nms)o(p<IvVO;r)tkVqd;Kv0wsLu#nC=sZ5$
z&94t)iEAVp^XMvJu1^4D2WNq!G|&d-MPQXZG-uxdSpc<)ATl5lsvwL2!*D`$fN;>3
zT!SeAj#@#uoWOCZ2k%l<;BYSEnX&m-g3qLe9+!*A!?0ubm_os$ufGt^5Y6<!Q$d<A
zY&%870?;I)ov0Rwna7bEN&d8Vah4jLNK)z8*OTg5ryBDKp@qeSFrJ;ceL2D5{Ls2=
zg%|%J&H=uTtmn7?yzu{{^@xqj+4w2tI!YCUhhRdRi%-Q#M@cCP3_?N!%`ff(W!spF
zr^@3=ThQhFkdY~BteQ1=Z-0MAs5h6r@w_zx{ipV!xTwFsd42*a>`&43|B~&Pf+RJF
z1M>g56SDT`Y~a7W0RKJ5m{?HKatPsDtrncRQdJ09$OgkraUVqrc`@ymJN%&>Y5H|9
zn0NucS7>31n(qLdMg{X&&IBFDoQ0Ab86|$ea@Yo%800lTJ%l=HZp6^&@+eWh*HAgk
zg9L#m!A4}SySR3_q6EY0poh<R?uRKBAs}nk;r*NT#cX`G2h|Xs0EMbTgnejR)$;BR
z`#(v<h*NJNk{q6VdwVG7!YwlQugF7?`-l!VwAx%$185A&cwPzcHp-_8JiY?3OT;V0
z_R<j40?#Mla#>x*z##0i_DE!S4U@hK(3X4!C%^0X3VlQhhd{!S!zJ|%qa#R?7<S1c
z=ot_eY$`Z0#8ca&-4C~+HiMi)eizZ&#eoMvBTu1Atd1-NLINpL6U>!h+pvldAmc}7
zb+e%g1qcO_-!dMoK}@5<eF~9I59_xZQ+?ofyx@#kJw8i4^lLBm7;rK95Ed<hcGB@r
zqZYA$&|Kq3QiMUk5$F-igw*T=e=RYQ(3n7s?A6N(5RzK_=n=P@oq2YA%}290QhO3c
zk}bPw=h61QB;^D73x65=UYg<+KEVlM3fle}2n5h@ycxQVY~T=BTdeNNgs+PrGI?G7
zRQM6NlP$hH+76#vA3FcIO=a9I(NqDBgeCX#!mgKMh0|qtD&H#HpK>ox{Q|ncxKzhq
z^PC3A2`&WEZ7MSTB{PoF6ZeBOfn!l34k1auz6~S2UbvMKevj<l1NqlQ1EQq?4xh@s
z@fdPEGy;e6nHnB=-3s^+n%fKN`VPujdl_=PPy6_HLP>+jvx^`0{WV8mAaTk=6Uq(5
z9zx^5q9kb*3ZKtF8GDPz{}IS@J9F5xs5Sr`40Q-WR)b4W@6H{jx2m`R2!!1(tNkvX
zYpH%nF+!LjgfLMBmp^&4(f`8Ud&hIx#_z-KyP+fv$w)}qp^R)&G9#jdq7XvL$_gzR
z*<{l&OC=fE8upBoktkav+1u|p>;68^U(bKfAJ6OceEo3`<8xitd7bBb9Pi_Oyg#PY
zKS#KZIQKp;ojBbE&`;sjiEOMp-U}KwR6gkArIHs4Ftz7|4%T1+_lr)bU09fI=%d_<
zA)wulr|CBMrXL;O`cK1fxOh?a-8ycFk2CUm{0Kf`{J`@V0of3$0jv;s%c_MVL%EsB
z0T(BfEA#GeVkXdK$rql|<aTJ3nO(aqPqqVIPi8+__Z{K$Z`1sKOUWvYP+7|h0CzqV
z3?{pcLeU1?1u)<WLJ<Uk!+15@TM!1T!rkM_EeJ47^?_#hg5tQ0n;-4W7rthtAqC_)
zf2vvS-s<K2tAfSs*!Pn_4UvdB0R74|U<DkgdY|{4+s_=VvD2-fIrfw91dDA7hEuRc
z(zCG@c%9;a=hwR5XxXr4AAyMSezD*s3j8>V@L|7Q9<~$R^mo-19Bqd=IW#z_IXb<|
zFa2WORV4B0>%6;f!7V{_mEy49=6&EcmT$Xihfov^$FUEKJ+O9d=g1y^EU^csYYsZ3
z4Zq62pa*64=NJ0dIV3Ox)?K4(g%=QmV9+*U0_5C;Zoy-CgsQqa|1bJD2(Ss0pY&1o
zBp2wB?kD-JQ^kF}(#B|EF8Nqe^K22&HFUX{8Lt%x;7rzL5Y%kHtX(95#EuTVej!L^
zrVV^@dV_A&4sYjw2+uS`dk$gN$*=KOf%V&agv-YIwIif?JeE%c1BTvnGt(drlo{(j
z51<-Wge6<I)9F6Y#7v6@Uxdpx02)JoJ&>r0;9o=_HP}9iVpL>b>ZC7MLLMLyWATqZ
z?~14VhzbtL-A(n8kaZz3y*FrTzr-t0Y*Re_IjuN!UY*Z@=eXS5sjDWG4;2vHwkEBH
z&7GDjz08Ug#P8cjLdIa^!_h7ESER6tfl_1KMIN>ffo(-~pG!dg{l%t`e=LPmS|C7{
zo&(G1@T+T?lqs%=QQ?HB`+A#H5;Kmo)uXZRKENkr?$>j25uv8PDENV8XlOO0B%NKp
z-gD(RHxD32_7cJHLn=glWVhN!7Ga~o9P^*2_sk(j04a~0P}~6-nPPmG*lg&y%vW<g
zC|dZD4yd(JOKCUV6FKB%2$hc&$M1~qe&am{z;$oxRwreiQd+>&9}sMY7McRLT*ts&
zfRLUb$>R&<H*0=5RsjeUEE<1@Qz0l1ttb~j;<73pk|$7(L{y72RjVsTqww9(wf_vD
zZrnN~al(&V5#$9P2xcRu+h#{wMo+U9K3^+rGmLgIK;qign-9|vc7D_ShJ)zOn~!r(
z^|=MAsIuV9t{L0*WD02-5cbzE^&!Q~bm06O+{vc&w~r5?0ikN4!mB4G)<$(=#O5gq
z7}Eo>S7<t707;b7>Glb<!;8Unot|3em1n`rLw#)Tl4rUA$k!%-_-VN`$mkUm##`Z+
zz<BM%JrQ^nNy-+f(U)<SC`AGFSph7o-D*2`za6%{bY(iDlLOASaIWXxrq3Nmx5$J?
zN9!>xvtJ|$Dc(=n*X;9eiKqnS<;@+*Sf&uM$xxoR$*o8V8pV74QTF2M)<=^?w0T0b
zvuboW-+)>McKS8YA)Ts0FJ%>d#E}%2-}KMJBosnEgLG#f*@Ddq+q}+Z9Q!Q_qZ>Bf
zQVR!f)B&YjxXTP24crrsMG4YXsP(vE*hc34_ut2{?ObuXxNNdu=izk7TnAEeg8~X<
zo5?P+r4%PvDrrXpFfsT}BDzlpkpes)+y+zwE}=9?QVmqxP&Xv?Jlo3Sq<V69myL2K
zF$Y-Ygsj7-L7pQrP-s;s-)%)kiJ&c&(3`#Ng{n!-1i2+PTq`m#$QuA0bH)FnMhH0>
z)0P8`%&;3Pa7jS?cD~L*+BYQVIj=SrVhk)N51AB^q|)Ki484xb2KlJp(kx!K##Q=9
z*zfSjgjm}kK84qn0Ne&*dUd}{rN$RU;Na5A4-xvnx7RUv&9cCWQRcF^;1F=+BZ~B}
z&e=$W;r8ixb=nZ4;Yoo>FwVZZfZKb3<IHFtx9wC3JU1L4`^?)sgxH-|-*3jsw4)%3
z?9B)UJ&g)+pUYVOAk~4dH8;7Rsb!#)L<9`9yx_)l&)OctDX0zS5k}=8HNlnl$88T7
z0(kFUkXxyK`%mb=6Rb?7%DKZz=c^IC34(n4oa{9BpOj3d9_Kr6Gu47A40?UXoB^$C
zf8A}Ln5x(k^%1V^qq-PJ7l(z-ln*{{?=?aDMEX+16*I6>2x~US*iWJ+;I~MN=)%Oh
zKTgmD1qQr<R+wfMg9|u>gK_p6o$F7?Ot&17eN6#hMM5T6eijj#G3XBHAxvyzfg+OR
z%qhR);Y36WH4J`W!?|xCUUY<n%r`_!bSsP%QDRpNVV#MgB$>zamoOrov_XA4X9pRY
z5Iv?X;37IraSF^lal70BorN^@Khm^-{#*5E4q^$prSDrpifsyi<~|-XS6SRDc!Qvi
zc2vN!jy^3k!4gz04@Q_k6wlXbW6UlYCB&YHYPS_e<~Q9l!mdClhbcKrc?Refazs(1
zY<UkFJ@@6VfLU&0g#jS$ps3}iFHb&L^?!aEI4`ss=&My$Uc)m0g1zkO8u$B&mgin<
zD2zKfYn`)JZ71DhquM`**M#0E9%1RtI5&^zUGZi9ij*F)Xi=WT0-T3C#9>KCngv(@
zyDi??eLN4RMi?<2f~O4N|JVQs4KG6|2r&KfE{2iU@Jcw8^{8RvRW#JpJk@+ez{Ws{
zfhcUtB?Opty4pBg9)s~tC|(hHa-iv~c2#>)<Lqkm%;QfXBHd>|kUba)fDVQDMEno*
z$cqsvNNkF^FK8rNnwpEH#i?|MiajD@UhGsX6@eXb8n5#g=i#=1=V{>lV$0+HK~PG=
z!Nkr$=5V@BZkdGJu#{R}$W3J5n2{i>L1GYnLIV_Ee6g!45l&5`jvxRi0^6F1+rb*8
zgJO%oJM4rk(ekGBcm_<JXhe}}7YqRy(ROVYyyFpI;j(IWvscBX@$Rr%q$L(IaY#~d
zCe!q?28hZu*eNXbbq9GE6IS8QI937*#tKGA@fYJz_C1vzlK+OJL+*DOEe?5LVa^dR
zD)8wM7rBf%2v*`YLdT~6VTAiYyWE3^M_`Na=P!WN@;$wd^aYk4JJz2U{lULs(pz9j
zAFf{=^jtwWCXu<OgcMd0unf-b&Y*zneg*k1*s=Aeuucv3jWSL>zFZYHs!L^9)Z)Ye
z3j$dZ82DeO>DC^&OW)3!^mBzXSc>R6LnKWe2-^1sO)QNVt&-_m8&?kthM&j|y#%rl
z8e|?|mjNgTaHNH!2P@AE+!OEMIjnq14C#-D8WQlWQe<<emI04@3EzM`!zac}tP1rM
znDTqbGf<;rLq`B&0*IC&AH;+LZS2xb<4rGd|JV_vCm%66E~1K38n&%_r)_0_dI*Xx
z?n}QbNr4Ap9U3}`svHazQpp5B-ClF;@&2f%9{YMR16vj$JWWzPPSWahhUaeRi`{hm
zS!xYl2@1Q2>48Aj!(^aLOR`sc+b^(9O9DxZq>NVS&zJ?k3a<>blSEt=Fs>zBPG}+4
zk@zr3-5H`ZNdEq>Dj_}$wwlZcqUhG9Du=6!{dV~!K7%MqqIh6D_%3kq_Y{EqvYlWZ
zi_h}c!9=3=7zmFAo7Qj+aeD=}?9TTedU5KR4jL<ruI(Wv5U0j263$=&j;%_Dji+*x
zVA0m_4tjF_b#c;A9sZ3_U{K&XKF*ly^__4P$G|P&4jk7-7T`%FmmXm6dcXe;KpKz<
zd+z&-HR1U`bH8BJ$}WezwiFoq0POq9L#Kf6qdp}fuT3z5q$kiPSUZ@9L#kYl$ck$=
z1st#}By8}%0nD?gRQ}r9i^N)WNvMJ#j>(aROirB&fhI5%-Kb=Ax|=8^$|1-@501~}
zB-q@Hq6yWW1XDA!E+iTlxwP?ziiS8T_CMT7ZlY6g8mqfH5m{QM-i7MR6wVjt{6Z&w
zB!9D&Bt*!~;7K_@>OAs^r+ta-vC(7Mu1|tVX2+El%eRV&8PI1%b0|UmX_-lL1?qwX
zZQY(diP*bx<t`LbQomm%V=>@}0t&-P;)q_J=sa&^xz0Gt+m})xWs5%<xV}xbQ%4m7
z8sfR{;Z%K}=i|9enMg-WZ^?#oVXFO%WR_DWOGQZ{#N4<hXT^aW<ZVWGbcsa+TQ`n9
z{jAs$p!yRpTLDcNJL$ns{+^QXN^Cj-nq<{?7BFS5|0E5ORV7%~fak$axBNb`q(EBv
zG^F6+TzYhS5kLnq<5(o;i%r8@;P>+1SAx<ZJF|moTuB6W02iuU?%(Kj1pZU<XZA;c
z0CY~>X-z`_$&0`u>hUmT#g&B3>XbXfNHo*vxnlpJNJT<v9%<nA?BL1UfkiE-A&cW(
z^>Y3K@?n&1p(T{jh(xbJda~}J0N&sbglE-L;@{cz28bz6f1mnzqR~F1EV0Sn;{#&T
z%PofMb_F0bm;%3n>$hvF(r(8VpKW{LYQi4Iqu*U-Mob!Ko|y&`xTJwPvz`cRl2mW<
zZ3;rqZrjS;zKLIu(vl`;my(w}O{h_}(R2ey()t8)$S${EB-Fxf`dpr5U7b_1jx~?7
zRm_aj><TTNvV_OAuE(j9$~EhJ@`;@2xz<;0;lXMdZBhr}yd;AsYXlRURF3=8^gX&$
zIh3FJ!VQ}PI6}Phod%Tnw>spdrA#n)kh<H$vY){0gLKrSC#w+w-)cBGaqM8Gbm<-?
zJv4~6I0v)YzV}LFSO$Ec-dRP8P%UBLd9<a@^hg48=#k}T`1Ak|$-zJyvbS-V21tc%
z)gJG@_GH=<(g(+MZ(S8%LX|+F#8qZtDA9LIq(6`bp&vul*WJ}Q6-lkX<-YAd##2aE
zIjSF+aPNSfGrX1Iv$|Nix~s4jCJF6o51dVn;V(&5@}b!vC`(haJJS&zqqDnivY+3~
zmU8PkyK0VgHU})&8y?DK_t4Ki8DBHcLgzvup<3<R1hyK(Yi}2`9^@H89sS7W1R?}d
z_bVDfxvlRE<!*-$u(6XlizMJ`FUCD((w&?JS>b_2h*q#m7iIl&wsy<yNtKe<gJR9`
zs4@#RMxz!)%w%TaU|<tmRRUWeiuqt+Anv99Q4QM?F78|(5)m;215YU$S}2FtsVv<m
z0cPKrO8*QnOHSdor9b=g)DTK?w2piW6y*&=p>+&_(k4@yMI+Hm6+<Y)xES6Wz@{i%
z`?@PJLaZ(md!7WDZ+jKMDMGDL{pv0cFg>f~9=g=>zYc#RwO||8LxQ%a9mrCQ)8SP2
z+r(-eZL-(FUa304DWPKMdKqZN*yXG|`#gvY$&Eb>108Q?B~O3yowKgx4MMC`UZ7NH
zezWO|WQZ>V^?B(Uq!kBYmpV}oM|??&?l_qx08`-icH_XoV1{X%qnu2(DsUZIF;N8C
zarbKouNC{8=zH$nYl2(_S)=cMHz8195R%CtTsY=>(rb(q${;1%P>R$DCU_7j5AUE#
zF{*D!+GerwVo@#S-a+Pg^`s=ATo~8?h=ggA*j+w*1G#lybi(o&B1(wcdF0Ut5VBej
z?&c-lz{n_qRZw}=L<)A}TeKSDgpdVKM}(B&yt$4?0aa%0d*(Qad}ihQ7Eho=l-p+q
zVUrtt*MMAU+jNYcjO4(f1kcSCzK1<g&L;2sU1Xz#nrmwig_}c@8bZyG$!_poz?XZC
zE0BwT8iKVWpgzt%SOK{9^w?J-5E`_*WflKYT5rT`Tei)2G0q=~$_n@69GX?uS^Wst
z)|kgGVP8E5?+!UcCZ|00&=m-|bR*iSZe0ZlOUm&FAx!mgCF=pu#UqDHiQXeUq6EXh
z%sij3B5;7ntpJV+SHE8#lw19e=-4Z360#i;S7z1%xDAL|>GJ9k0#xiI4Wdz8R2#O_
zcoQ_UZQ_Vrza}?`H;Mng*2_cY?ODE)#i`TQMRvfmS%vG^^UtCvvl$!m=SKA#+7Z8z
zxh0kh`uL3~;{p?+@*)%j?2)|5Pbr^odOp5?91&_hSPpxCRRf8V1aloovmp#R39o>e
zBIE#E`5Uk!$Q7(kU%z&Oo&PG%=q@s;Wq;uY3)0h4>^4{zJvo}yNTxkL(&WZ?j8HX!
zXc@O_ono)Sj_W&X)`Bz^*A2P29?=ta^8uKqxO>r8#Sv18A-t2esfhW4dO9iSAwGda
zM8{4(_FEUCBfK$)BZbg%VHO8`kqO#J<$p=z1{}szqBSIklMv#u;H6QbBNUr}ym5kf
z6}gg11?Y#J?cQW4QkA1}kKUtDyd1wagHbQX-4obT8Zja~W6Fc3e+XE@4Gg43zVLXw
z<TMy9@u+Bqny6v}BSs=<27EV1nlOkKm8iTU^a=Vpg%GRj+Lz1Kv*VeSETr-XpL5Rc
z^UVjK%AwkjNL72EkAAUEp|kyak1PkXP6>7#;{H@ik652R!D2hJLqUX5pxP-f0m6?G
zv#mm(!O1a2zY8AuWvUIyx`;!ODeZRbCr!3}m1T#UR4@5#S)y9<Q%oGZu2cZP!V|H0
znkAxp|FizM0inheyH`!vXZ_Mu(O<A82uZ!-Eb-|QA4?T@!f#p=Ob1-T2)7SImtTLK
z(evNOy0w9DmX)n=2x(lIuV97oti@O~6S004KHZe#(ZEm%=4#n|A34{c#?=={UGkKW
z;He`<O@KEbJNEfFKoo_NzHlVRW6=X+CmK%t1VH2CDq@GEB0u$I8ury_kSH?^iY4>F
z3<Kgi;F`7-K5l>q7C@Y?tnInF*lqHqdfAN-peto^`pTF8jdKq5X0}yIBy>pRPW9(P
zLqam~h*gI3OaUw+k0nk3Rg43l)A}}ki%V|xighe-5Op(=v(G>`xPK?}Jn}}2#Io*;
zJ(wDQt>b1k!Tl?8T>u`?pKjb_4I>G=DULA4xDZaSIL|X*#ZT%TxyZ%=f8Il&KN0N4
zgf<LxWJI)m`9#Z{HOB<<SALJad~XR*Y?PUsiZHO@IlL)<P$<wnMEkGT@)BZNMonxE
z=L3B7KxHoi7R>B`7m1zQnlA%bFXS@|TAA}4al}?Z4F)yS48@a7gT0BR_AeITgJlum
zrAgvAlH%81N+t|b>Rf~1;s-V<$|Tkw{Hc(;`%%YTfo<q)^iuY;^DssH93ayPZV;ge
zDSLtwz%+Xm{7%Qdy}!`y;45|B9!Y6PvLJd_jT>c_y6Gv%hqyiKTVCv5rCEkxW40;h
z)_aTduxu@9vjwo716_PZiA@<e3T-vjm~iCMsqN#Iy03a97~Z2?wn-Gn`d0FH&*c%L
zu$66ej{y3v5^SMx!nD)-bfVds#@{e|5^2kMN3*~kDSj-SD&05R%r`rDOabZb?GCwK
ze_`Vhl4u=s(Qx5L)hC4GI6H=u8WH!`G_;RXk#b_3&#&p|BTb}4xAhB$9^YgCWSna3
zfB7k%SY9o9&~PRR;YcPORsSyMz|L9<f=O(QQLO-F(mn=-Hl2iSw_cC8KKcw?Lk9<u
zrzt5u#4DE>`*afJ8B*MsB?-1qXSPr5eTC^k(2h1){y-AsqqF#>du6A=p4rEe8O1aD
zno^*a1f8`w4Z`?Lg8KK0q0IDMh~h~FAm5~uVHu9XZ}+9pMOmeZg9|{DPr~Q7)xf$I
zz~ll|`vXWLH_dE)E1>~SR@<T8iXjR3VyN~9BQ3)aF3GXTY#2l!9*Uw2A_L1{m%GA&
zqc43GQ6S+%+O}?FzyL5=&(oBa1VXADKLL{ncEj3$;GeL~J@YZNoAgJ7Y>ZE}kV6}%
zjk%oexD(LcE<@Y>rgS581a{BfoC4vRt|x!rCD4pQSf>ye>mp0ZON|Q&+2&;S>5MJw
zV~jU;)Ow&S?s#wk9*<Sx^xG93`_ceX71lENpZ&f)anN}<Ig*`@R9MQM_g?~@C7}+N
z&2C!{t5WXq?(V8vZX66G$=hw62p8NmaDW8xHu3OT{m}z`3IAFEp7UL#rSiIU+N&MF
z49x4UCp(=yva0;r5$-L341B8~8?a85js{|&d(SoV4!Vwp-bRbLvRP%`!39JrCafFl
z-Fx@UGsujEv%xW8_6n@FC*p58#J{A-G<2_ygnM7iZc)+zPLu$2K*~PwMAG<;-p>R~
zwtt-qNIxxelLUem&w9~TQk(-ULF{g?z-N8Xp$QZ+Y3}&Yj-{r~7F|iG<?b6}FNV!U
zrCYairv{{q+ZKQ&^!Z5rSi-08f2Hkz?OUWFDSTP?0|${9@<Et*$(opoJnlvz{p!*w
z6vAR3&q$)av8gj`X?tQ=6vOVS_Zi%Zf@oAj>5{ocf6V-ik@I)2GIeY_U{;(%27QuJ
z{QN)B*d5fNZ)`GE^*-vZrOW?t0ZOrQyk|D$m-d0vDD-x=u@uDAkeKSS5_v#{Q6H<2
zR>$PY;kQ&@-s;DJXaDGcHaa%TBp1|s``CpVKLtSLmYLCDhZL7JHaq}wxx|$zn2=CC
zN+E7VC=*|4se6t<$9L&QXiyU?N-=FQfPCSa<#p;GUR{z~oj(yO3RI2h{>F+1dRD*u
z)M+@4vdjE4EJO+orJ&vTqw=Djk8tZf*ThMn#})6;jMA1Z+nO{)kc7V2R&iTVX=jeK
z`Fabe-k~7cn{ui8Uck*GZ{IuL>i~an`vW^fG$byRrf&%FS={R0$(2lX?y2!q=j`CE
z^-NsXK+{vucLrJWac;yNhULSuP476w6@xmxWk;T6QGGr#*0umhOieQjo+KnP?R;g;
zwrK0`Cp;Ildsg;UHN5*i4tShhl$3$eo)embb-n$kPEvb4DKS(0_^QcNo5Vuz3+}Ca
z+=`&@tF)Ng0IgwX&8w~)B9#Yg0wyUv4=)6lGjAjZWNyGV98c>XG86+g%U6maVu5Li
zfqu_-I-cM&_o(8A?SinbM~m(1%<Izf&qc6(p)T>8v_|*e(jt2kcssN`WYWCy1ZzRP
zpx2Qb)Bks41Ut6_rN{97#T2tL#8pfVWhjD3=6b-(`Hk-A<(`i;l*qa}aP5OUWw}Q)
z-L`|0zWv21RBtlMFQvxsbo!ta*c&J%AT;0r4JY4NOO)RGzUZC}v3ht_JT60mQJXJx
z$CWR+{IhEX*V(sUjO1`WIZdd9OHz&W(l;UD(zj%SEE#WHRWl0(8HC}RJ7OSi4LrJU
z$qhM(R_ZDs;LCXU!EP7Xy@d3sw4S{`p+~q9-{0-hvJAJX(y5aG#DwwGNyHm{gA%@{
zaPHFd24<1zo4=svn-yl<Cf0fN=xA#PvLE5G?N>wSHcg12cPDLIVP>KZqBx`^cB^7z
zNfBtkav#0UiKeT3NOL7pMe-Z$6lvx^1^ot~u%XWPv&sN&h@F3Pqs^!T5Y?bYjcPbN
zM>q(jhf{or{?F*{7(_%t3<n>>^HF{-WY;2bB^@7U*2IOtJl5-(S9X_jguEWn<PKkY
z=bqo+Wd;A{^&$1$HMR#99D03{WF{rPDBC&!6epaw!gsI~A*W7IvDxk3J{xh$#&*ps
zALK`SLaQnbB^709x1jM8&9VD`-)w-*sxnX*yTC3n1~xqcTY%lPV)s+q#KP9nf2p2(
z&$_a8`Ut;a(VQ2(VU_M)(CBT2yFFJXAB1dNuw!=kATL-vb#ipqF_hhK=E5XS0}dr@
zjn0oFk)WN9<roXK?g4}-H&L}Wk!p3JwG@Mlc%O~~yR8=P<2OU(P+L{yvq)_e2(+&4
z-CEt}wqpJH;@Y+~lg%TqooXI!zdB5}$Qg!WF5s|#XDO*d7<SM7RK9Lij9nfg!ENWc
zCTN6Y%x53LTZdTP1LKr;ST^O%SY#TCgs9B&Jo{`cYTb5V;+M41_`WYH%$0-l+O_Uh
z`+lVE)bod$aQW`^T^z6WHv@jzmGcBdo!zUA@Z1UJ+P>ui4d}GBy>Xc|oWZBmpPN8B
ziS{@tqhh~vtUNp7*vj}%OpJ^Wx8l+u(}S$l$Nhu&j~u@*9-0H&r*%96kzj~Kv+Rro
zJ&e?J5Wm5Shi%##is*O}n0r8ce=cBk_yz(*?{UluCS9`-7O{p!LIuO{#@E$Z&sDJr
z1vMMOYnnLAf8Z42u^IHH5c(<3qH2mqXa2?8pQjftSH)Y5b`=*=@b`6%9JW!Xi(fOZ
zz*o(#O9M|Wi3o;PkTt#ynK84-YsJ7Lr8L!(*N#odk>RBZpWPa_<<m&~NK4$_DPOJi
zykphmm+YFyc{AQYjVITJq;UHg|DtpI<;(9Y`uU8p+VN+t3?|hePJi8MawW5yf$#H~
z+rI1gu4U5iI2PQc&><^0voaW6sC_2l6?b<d9~lhEc&%MxF;=)#Tlwr8O<5YOb%8jI
z*74GaJ?#f+qCuY4+Dl4|AD#!I*r>!}44PA(anB_&F|qkL*SCGPRag#zpm()LzrR~u
zc@AmhKC#}fot-Vs_0{Hkh=Qe@XID7OKon$wk@KPPU=?mq`S|g0T}}UOsYyv`qG@?N
z!ePf=oh>+;G5agr>+FH4n%ir36&=;NvhUwFQe0mT1Ust)-n;kl#gnKS^_ueX@{b=U
zK|H<1Chfk^(juU*52_2|SWu`h$Z!4k-+wWV<|FcG1PMs!$>{VRNqx@dmD_5RhdVsS
zv$I7oT?1l4h@ZryrKJxYIxTthC!*P&#*W@p2P4r->k@U<muxUn6S##_=Glx^zLl%1
zEAH<zFpso@vVPzEYt@{j0LE7+eFLBOEyegZC^kf|oGnM>0GXr!s!epfNpn|Uux*$S
z&tr?tP6qaSXyOQl-@NZ_Yd(^A#i7S)U$S_>IZc;)T^R6!<&nlcVgEpA*pNCgiBCxA
z6+BaO(m(6*pmG?y?7&~riSgZO*pcAWoxU!|t*WL5ST4?ZpDW5Xi0*5eY0Qc-It#;#
z@lT|=rwUW@ZrstqjI^04KHcDki^9L$a;84MFY@xqGBA5#k>3tFjCp;`MW2Mlxj78C
zO+y5Sl!xG+FjsW62h(}b>oIe5S4z-Ph!vrnl>zFmobxyvrh;NXE7K1u8k$2(!Etd*
z1NFOd-a;4Cpx{BZVn$3}U;!JWHw0n)zB|veDJ<z--%@(7H<<mHAKGy)0C+G~xq5XR
zG^&rwd9*wO4<Y01+hIT^vGyz+evAi<vD;rh5@;-WX;@srHYptXvNPFMnwr5y<EVZ6
zHBRbv0t|p&DQbtNQ<S=A{dIctY0DQM4#=2<e{H#I`0F>1!PTMrl$3{@y|*e-j<0ZQ
zKD=>lo$ZP<l!vaVYqN-9Sds9ZCi}t3Z<b*9Eg?7a_5B7W`1z=EGpLLsI!1^DJD!Ee
zIYlISnKDv#W7-U*Cn|;TH6APT=Cp%}m%P=Q-9IH|ZpV%}P9Fvx+(!~s?J~kgR{EY*
zP}6@zQ{V&=g+Jr&LrJv@DuA}nx$<AWJmB)E;xb*iI#eMcxo<hb<csp(!!rU84)X>G
z?c?I&D-mg~q-joej%mmA;sw+^?>@i?E*viwz!+#U0VYsoGzFzQ=uO5%F;tEp8-ia`
z?8nd!++yw0-qq+#hvAd&`7WN(7{?|e9XinQUFvtkkI_iQB75DUW^*Yt*wAiKliIcE
z)&W!`35{D#@=5|?Z$@KN*J4?S!-|W~fgJ~V=wQyKon&qh_DHDQs9^G3OR31}n?<FC
z3j%k=g%t{P$EXv{2mB&pF&+~NM5VC(m}7E_4N{=HNtxyC?OSZ(;^G)YkPWB+aKxd0
zW7CDS6Pm2vkmW(8+XQ};Q^O07aC&irknIrn%6|7x)c0QO@}R+J$4$M#fbAe8aZ+;0
z;N|(7*G@C@lyAyb%w;TBZ_e2)!X0>Y#{=UK{Bu{3pI=RVZ!pS3zP-VKpcE7o2=x=S
z4nK)8jM00A2su%u((n8E@@y`qSUaJVxcNt|QcgAR3jfFWY)D*l<6rHB=d>eHvkLji
zI|R<o0jmNv#jwEPiG#oiOUoamZ_22o2xb6)KBU`I3s;VFLN5rQS|y|Fj&qAR4jD1o
zx0=!;#@0|-HWzb*sZ1QW>EP;NRu?Pw3o4||E8xJppiY@GYPAqnv>rpJB;z#}rklUP
z9u41Z`ozN0nvk8=_TInddsVnbDw!{&^&N@DEH(r}=o;PQ$Z>yRXC;R*(yd#&)?u`*
z>1yK%49@<El$f(ixK`HhpB^a3L(rShVHEpu#~5I~h?kM`#oyN#jdQVc<!4MUUoHlm
z2~xEA<5IncaN~+k+}6p@A$}oVEyHuUcBp%ws3<DRo?kJKQ1Q0kgN?(A4BeNS9jN>^
zreqIKyI#@|F7Kn>85zx|FVG6!AX@PofjA=7ZH{h%1S=Vc0elUmbnu<DF)bb=qOE^c
z1FVJd4u4c8#MB#1y?M9Ss{1FATQ<Y4N|k%C6HsH{CF7cXX8Z2lDEC_*yp)jZk5`=7
z#hKvKBfjq|7gtxgtul(q{k_3}Gn-mlU*jE3tn(`>E)IA%<R|u~<)nUUN(!&6nigR1
zu);XKXja#YL*bte_4n~?`t|Mu-xeC{ahg5o=Ab%0FT3}}`Y;=2&D+v1qt=juGf)p$
zR5FhlOOC++_=29mRz9M@g>V-pnEAl@!!wSAerZho6~oIAlE8ECM+6$eer)vT&+=bK
z@D~R%cVV6jG^9SKv7LKUIdqEFlf-a{+gLv~vKN{w?%>yNN!yp6!U+biJ<@6{-A=Wo
z<z-;%HZ>)F_G2QPWyh41_)0#hN3VUF`zB=Oof2j@K)tfIZ^5yBckBTz7;Z@3AzQez
zqoiqM8iaw@*BHoz0Gz~$j1pvS63|A}-_CN+-oXJy>qLl;O4DX$WMs@Z<zv-=1*w#b
z-!3j*#zVA(HtWaTzyArG2SoP87}}Dk4#mQbMcxLNa8HM#6G?hBOG`^a1&crj%WCC)
zw{!G4i5`!1r~h4dwn?ary*C)GF@iw#;h(T-h;tfI{AhR}CupdEq!Lrahet*hK>|yS
z?X0lIlmM*QHj^QY^#%o#fQk4-Dmq?>#Bw#flSO|G3eU}#4;($J^xe})l%Agcx7j2q
zE!*}ed_0k)9KxQWd*~?g7{HG_{MAowt5EIC$9$SgX7_<0Mmdi$?)K*9tQ7mLvZMUB
zVnMo~S+$8kzA1Av9v1MDZ!Ockf7h%DFs?y35tgwtJ;IUqzyVE<TgZ-qQ8ZWw3D-|O
z(j^&6VX5mh@4<sw2bV*7jE-i8Fh``{HFUXp{<hPG3l0>TTNxPA&dF!A-czCOUvqJI
z2c3(t0#B_#L=PlB`};3e?Ic?XMOUob7}CUBY)L68gmeMBSe@cUNo)$npS*9xE+Sg7
zpz+XI#gt54gqr0w5rqPg;HGp=k4>95ZL-4C-fhk@<C|yX;N-mP?QLG&ng!Sv!)4A%
z?y-UXNXlrMc(D6KpRay%+l?C(E|1dFt#SpP^o@<#gkKVY!Ecs6G<@>xw<>lNn9yc@
zY@U8sxz5wah;%_&P?Alo#6U?{5*8K~z0TVG7ZDS|@%krBfT;_jCR=0R$(QHf;<rs_
zX%9uuB8IJcKZFNp8JT9<vujvgGsJgeOqUjLT}L}RAq9RiM;+*fGQFE}<kZZg9%vG<
zCq(nbut}`@xw2{JZtuR@$i|n{DQp{nuV6=h#qP%isd=2<y?;+h&2Sa(#-aEA03I@6
z&IKB<AMoa$T;xAs%e<YLIW9ar$^D$R+b3&l>$6aD)D8j~#`l25N`7MY(ahk>VZ?N#
zS`Yi@$8q<pXUb?gLS1HS3+tNCX>rEAh)*$$=EZP;>FgTDl#j~82{y;)!A`-wOk_{{
zt?D63ujU8~RA+X@$kFzurlz4MpE30U=q;MRyDhq*=QscH=8>@tiJS5-4RNrx?cmkn
zp1ikfjouNfQ#&cDvdUV%TaI-V^U*~0ov2U|oNSI(vg3SV6TSviUMW<V;9vZLzJeoL
zQMQBCOx$58_&a611}W*~Fch$DfKEHA+=x+!vsMviMRkxNj3`BLvK*vD`}HY9i=Lji
zVUo=@S;}=>(ba-=!1SE}DnzB9ZnY2}P7jK4S+-4#O38~SduqBn)}m~wv$SHu<63Me
zn|Suab%q|~Tq8(Vr0;eiZUs52e}bXlX3rg^0PPY)6u>$9jQUg~_Pe0%uaosCy^cnE
zMTMf2Kev%(V?{*@Qf0mNth3OCG{X#yo(#mXqVC;oZEc8weM@-<j2kxpu@Gykd6E{N
zB&b#F`P;j&i1Ii@dAS;n6?-ffl$LrP+O5UxjaGddFe{j%J=K}L5f|;^WVhIAB#Mbo
z==o3;gL_B6Mc>o?G848{Ake(}vF9%0`d#NN@$1|)dyuT8MLgyjtlWdFHcY7*VM#Yc
zU$iUsjE#+5xNyO;_I;mpj?z|tfj_6dvD7ypKn8+}*Nxsk-k=B+!q|p2E6Vq%AcXG$
z9-c8&U%C@=Qo0|T!hBYkBc+z(sNnpYiZX{2hiTY|2Z4`!6d$NraNY0v4uRd&@XjX^
z;}NS_%qmE72~e{9Y#DZFz9cjux-rX7oc=wOET#CodJgC`AzR7e^T~+G$#DcLF*z2{
z1lP*5_`BlX@CVu}$J3g6KSlpWT_QH~MTkaf*h;wqiW<ROYcSWk9X_D{KK$*2rd~5;
zlt=r#;B`M5DhXgH99$!)Wn`k7u@^j9Rq0raOg!?w=m|YJJ20ZOs5hs{>=#0d9lU?7
z`6D#s3?ejXXlP*k(T9FZPPu++{!mULu>#^;k^zzO^wZjsqYWv33xqEHC;n;0;LgTI
z!Nl$o_^bZZC6xEg(zb<Q@(MOkbHKgwvNCRs_bq2c8V|4BM9cZ+a`WnLE<`Q#89@kN
zpey>)R|>V9Q`a8=k$i|FJt6sj83r$m{lbupfNOy)GEdzfGP8J#OGvn5wn0yZ!~vR1
zXaT#P@Qj<EzkbRoawdHzE2|*QCCp4=W@b(@C{MRMd$tob{o~TcG{N0kfk8oH?tZrq
z#cPOpu3Q(sh7O>DCA;9zP<Ho8A8?n_#gDQZRPI3x*5v_}S+oadJCOk`SP$?-9g!^0
z<Gkz4_p7)X96XAVhbWGQjh#$)E%v}$@Qd{~T!)>BjG;Qcn)MnrHMP#!dv*v6Cf=Qu
z1tv2IDr&&k#^j_VG%X8G#P}5szI_xKe*JuyTS_j)Efxw99*w(nYcC+YZ+EvCUV4Qp
zI`8KYjn;|Zzng3{kFUkV9oC;IYvrS$=&dgWO?%^tW+YWb?u(z(ZcVuqgX{G)hnAUn
ze}FkpLZSmp8C+)ei34hwo$DL452CYe7U>Sm%#4grngYK+vc3&^9@AwyZ+{FjgpIQd
zKu@%;<Qii%moP1FTl*c4Z3$^-N7ZMalhx>5*8RSz3fXjMj0b4fUYLkmLn=-MZlv4g
zRvk7^zbKYtZw3#X97ng><pvMnf*8Tgc6xGhQag@rWOS67g~g^Nr6;Tb`$k(@em<(J
zp+Q4`4|2)hAwO`S$Iu`()O~3Gerq&Fz{Tw3+Q*xaW0;{Er#HDUZB<?~EEIPPWm8a9
zJfRLg?_CX(jVT?E#2MRbYJMQiQJf|Tc*23L_<=^Rr}}MU9b*7!1MzmK0<M!!gM<t~
zF;KU!9PPUP1X~M#lP*)R2s*e@*8pPO9jU6w+D<N-sMdo5-v1n>xSBZH&m*9;@@=>U
zt$E0JA*H5P(NDr82LKFD4~>ii(fD9vwDPSSv0~kbr6Qw-Iz;bzC<A59>KK0ez<Wad
z=7duu$O`xV(FP7+WMous8ByS4{+;eJR@>MMelab_#36A3sn&!f4&ydH493t~-0|v;
zh$~9eiiyvA%1-2*xD+zwTL2Ok!)3ZF-$Cv_nivAEjoPZh#EEmm3OY~|hSz)|It|W<
zpJ*{c;M@||X;tcHfI$yoPR8P>-EoZeUdru6viR|<*1^S2#F>4fA2*u^VZlmTLC@O3
z2jvdMrM?p?DhMd=8HWbro@&476#c`Bot44!fS+%}5g3^y$Cd;hkjQFLAqil*j=3$c
z3~04uGyF~0C@CSK*L5D=`jp`v2Zn}s4Gk|4`;i(iO3VQj8=*lAnT1Gw#Yh0FHD91z
z@7v?viTztqUe5Rf`>=KB67_A^1fCh;*)7F8N(EnAeo}pg4J1%+w9n4TDahA1#7Sri
zO|b%Ij!#4YRgaH8BwTi@X-g4p6;xSkkfIL)EU2ygh>;OwFR{8T;?(uKpm9V7^XMD^
z>Bw(Ix%@TWbM40#TzDG5@K<wPh(K@J!F@hTHDX>u4~RzwW-QFmUL!As-VB|8Cpm-C
zQqUrBmlZ(?E=;5=T;h=nUV}pa?@hO3a262g<l;9K`!PPEImb5XEqxP+xoAk*E58+E
z8=xez1l-Kk#pMz+A7(GYI}sfR2+d@7R#qy-<#gcaqpPK2YK0GQR#?*%?IjA1NkR(&
z>0%U_Cu2uCSgnYX5h6X%ng<;dU5RHvCBwL3qEgK-JdXZC-WOXUyr_0?9{`IEKmpTz
zXfS1gtbtiXa0*v#ZGD4+wrt)kdR?XA^NNKyG@cNmgJfZ9Wo3YDNmo}FlE+9N(E`K!
z!q4o(6KyV+&n*Z{xf_--_pcOcYB=QCZm+=Yh|}vB!IP8zM+~LWNS&OT!lY2>$z$|d
zJ@B?9r*Hn4=UGet)D=}olyP<#l>mNlpwDHJc78eYPEU_RTKYPi9|&Aq*FHi(4~-2+
zTU%@u7tF0f^sE1`RlEilXU9oQt6&RiL9%-=y&yi8|7_kk=)QMFn6TIgqT2!60T65X
zL0E4fO7&d-fvSGmvQl5@J!b%4h`XVCEfhCO-pI<F$^4mqv~S?TYOi-SyG0=6Y%rzi
zMmO9O&`3Kjb)#iu%kM$!-QEl-I;=9TA_oo-PVs~!c#@1rO@}WlDk))*_cv6kCQ4_(
z>nV%<4EKp%)qE{4SYS9VS9oFIm6!*Ka0nZz$&j)i+YZ!hQtb4v!TJi&>>@=~BqK2%
zRCaewEflvH#z{I7iHYb-r!7~O=1~}K96IHUgcPIuLV^szb|kG!$lKe$d?B!$wDj4M
z42@LnSi^WPh{2MA-F(~#4jc)4Agm6dI~p1qxDB8>e#>=?$m?W)EnZ-H1)2HO)YSC!
z^!xWu>FBh;HTR!j^ds|KTrPks4>!&k^sgY4i?B)vg0XAFIzJ0N4VLp<fw#W6V14|)
z*PkE>nNqGr)4QQ0oW2L3jiY{E|M~MFjBiaaKe6==JcelHNsm&#qk?jBa*0<wF&qpn
z6pyaO9&X4qrDlCo$abYcF+(fyF53nu<AdE4(WO;-ecldRsN52#<X%vau5^Th21t|r
zm@Ld=qpG1Hd-dC4SQ=P&1mXpQ(4*XRbRFhY;|f3RTg;;THG$?qcl~U!t0>G;fhe{a
zGN&F>ZUciaSX8SxT{E*poDA)_{_o$uLBU#j_$>9&I~@n^pJ8hGm=Iv&{0TrY>D5h2
z+IQ3oLyTdrmEUg#UV3U|5^@&hmbh7>{eSrI4Xya#;H{9!2du`VSUe%2+C2LctW}W}
z1<p`8m!}_xMFt8PqhxoT%V<TFPTx>3=}@nrv~*|DQZ4u%$5m89#G-Ptnl|rX6*zQE
zqd`&ZC;>MnA5E<;&IsS03R4Z=t^K&p9@7_GZb#SICOPxh2DT3kM2p)#zScY8&kRfe
zV9S7`Xu1Z!xmn`K^a9iml;ygfKYK<xuOM4Ge5VGO*xcM4gUshia5Q;6sC@U^ZENh{
zi*bl>z@AKM+=!43a#fvI1+`tN{IZ%b)uim<%g9JwV6YWDN$Co{+aVQ>JpBxQ9U2q<
zxkMbEDJuO;=Hn&Cm}fBo0>!;&KrI3oqSCz`T@GSpt#?*;dt1~g&Q48<>{GF7d--U}
zD-c$k=EYihGMm@f8xzHnUH+B9tRXnAyNn1)pgVUK+)?z{VbnR9CUHecl0viau2~i2
zB)nEnfrIBTE98Etz%b(%)Q^i^Dcc(68fKJY_M75Kx6K$v5oUZ_Pl_zI(y?QVE0OiV
zCRt7OACpvQ6Mt=M)=(`Kh*n;?C?8(yZMc3oWh6v<>4;@8rMKZ@SPX_wL^`Wp%69U=
zuH78~1eKM#s1p%8{K930mg11-BA^53waucXe{^)e=RWQ5s3?o6)XDL23zSk3IhCnU
zLT<15Q(8+y)6f?+cuSbejvuT0&%8Xm_$=99Q45xsbe7}SfpkGy_I&s*2PEzjJ74ko
zzi1r^NZEswI0N1W>C<yikOS&HStHQbQU~RWa><ShQ*f)ko(AUTc{@KqX6f&YfI;NU
zpCLK<>C-9KaNMJ}m)DbN2@Z~92}aHJ^<>@wYJ%ulIgP@mf9}9x2CPpl?Z23Yrsx()
z?bU=K3Lr+aXIy!>?WyzBP{^JL(*i*4Rjiz%Re=XwF|hP{vr$3k*RSY&&v5SS96v`o
z=n*1mTaUYMPW<xa3t1(<%_Wr=reNl*yj`bVmjNAK&Td8cR5y~!xSvI_bz~YV7%q3@
zdr!|v)brtXpsXvOV0a!8_md}2I_mZN*b{Pevr$85JjMX`MtaupH1`^7VWSNtrwtC0
zC0Ya+{y8%@7o##QiJ8zAzcw1RQLx36(S0btpn!B+5V(LzR7ijQn}&ul02_Z+Hz6vD
zn`YByOgq`!D>H05kLDT$*Hpc|{$rRFjAo!rGpr8Lf&mW>4Pe7svWug_0ZaNgwcA^n
z=zy0zMWlcCpU@^dP6=8?jPi_Z%s3$yaPscEAtdS%tVfb|l7~b&<B%)2W~@f)N98Y3
z{TCdH4Ouac#Hg6<oS~tipVL=66^3F`AZodO5)fg}ZUycAhDB!*bC5G^S4r(nA~V4(
zXZK9G*bQ-nQnF$LnRfZcdLNL?&Mf1m;I4Wgw+i1*Eg=&c8cO&-a21FEeCSlexo!KZ
zw$1DAhUnW@P~Xd1IBv16rKEg-021qni``<=d8#uEIBNLCZmtdQfb=9=JzTeb{lfhG
z@}y_BuH46ns3>PpkFd)y5sOSby53TJ_~yTWzE{Dwskdq0fg<LQ&it%7tCJ@&!668k
zAZl=kan{OX<?jMqEG#Va-F*7VNum`eay;W62AkXU`vH5LdRvD?dJ&T@rP(p+19|@f
z_N2<l9dB<V*cB9CaJZZy<A25wfjb<lnUmS}H7-}XD=WD+eR-Xmd+68j06N9afn|g}
zqj=%pV?YnF2c)jgT~&<CtD}a2M<TESXu)sGXzhJ&En}k5$RZ=;_V$JKeRwa#SOYL6
z9gN;xT$1tN?~kpltaRAZbqnJJErT?S7&<v!ss>&5)wF#54X<rxevgljPfo@nz?-_w
zfl3(WA|Yymr3yFOmv;Jj2Mj+v^V3XJR>}o?qDvpcVP1lWMYGBWWMKHyLifeoJ=<Fm
zX~B(HlvYwhM44MaG$ba*6{!k}tv;y_^Qa#9K7RZdNw*_1OT<*lS@xhQ5bY$h;})VA
zF8&p<R<eo-rUp~@7DG;gK^5~b@=+gj=;nWnmRxx2vT-hVI7WcZHo!#B8Df<1cZB<=
z>+-EN(YAwzB@*e!FwMc1pgtr-9h_9ld-|!4itJ@!DL_|5B0<qhPl|>IF(-Uj9qE4y
zM1moDFpTu$p#$-uGsM8cxwU%mve3GC+xPYZG;yJR0AJ15tVC;vV-bW{7W~ZPZWN;!
z8$=Gbc4~^0Ed0w7;x34dL?(Z&Q%%`Y46~0H&i%mNqzrU?xM(z%6pCNwex3R4!w7nz
z6@$J@N=flGZ2ot@*8xciZrIs8P+L)t^Jl5p4>17L7O?l4cYh^C3+!;1RAE(XeVk*2
zl9-!w>ax7VUW6QQ`bLntfEgDlag8=c>Vp0K6JQN!9{uWi1_ls<oCHr0MXQ_Ai4GAU
zCVfE<2R;>IMix`Ma}avN9zmJ+)wKnkeB?(!DXlOwMA1qOIfoO<r|1b5<=~bBM+LFn
z3K9=^!uuyoz%U)f04ii(sEf;@pb=25tv7^yCKhwY8S!j@^j$NQsg!`<)VDRgeR}~e
z1C$AbFJ@rG_jO0c#l-=59S6ih{u*v0`uY|!uU;0rx&!eN$iFFbDNmu%iV_UE!WbAB
z=6-puerambdd@TaQ{*<1G;<5yxE7?_V2rl|aRh(+se3Huz&tsVX^gWj1_A>>MZ~Xq
zQU*2=o2aOScrha8(66uT8{R=aB?}CFyg1l?Bc^Y{Cm$>L!Z$d(xgi8SkGvfZFZ}E3
zK*)nvY#RXG31{ShcL~v0I}li5=C?<tMi2>g(mV+-0TtH8#YHEGjDC%P)wBtx&qLP7
z7(+4VK@1z>dW<$Jf|3vMKXR{uW}+oC8?OtM7j!IhaUQbzeb0Fy!vfwGUN2&asu>|S
z5^co1Mz+Jp^07M%A&`Qo8EJURtWgslm-KF<EUTRP1kZ`Y(h~0-C$#pa)2r_sD+r0o
z`V}_=EQS#(Kn{TZFf$SJDkS%axCv)&I1rl%(HfEY1yJ{~_{s%y^VhgZ^!{KDP(GeO
zam~K>9|`MrVtN|ZGmtUJAX<r~zcc%i_GcDjG_(wG_~O&1aEdoyX@cvjzT&n6bt2e$
zSZ5RfN<`YV0%4+~@#-}-KOmL{MTK_W>kUdwD9;k2Hv$K&tI+v36gjbCevzpTtrn-W
z@j6gEBt%HaLWW#tT(1RxP+U|5J>GN_9b{Z4OmPHL$~>|1a!5sRH3+S%v)ssi{hBn#
zLm<EjC1MAoa3_nQX+-f5Dp`f^UIw+r|8sTrt5=xG`x~RA$WZJH@7Dedx+N_>B^2!W
zl6u#nx-vDoU||Q>TK=?k-{Q$&FBeNQ`K>9~pXf%hmYG5xSmwPMXD4lLgQ2zcGkAgT
zgK)AI3fybu^H8k;I>EL9;Os=mTV%QA?fsY_-TX<5eeYfz7`SXmF8tokp&GS~@;aUi
zyZ71i=VX$3lJd4pX+-zO^&ejgBz_c^#4I&B-G8RX#}ByjwkrSzK<a}p47dm6j0_E_
z%Y7x*KCTJO5r5x^RcmEX<l^9fqhs_yKodF>UK|w_1=|W53(0B`^)@#(!-C?!&(Gx}
z@ltHM1IwZwC*Cw}Z)%#v!1G~b8p#zVY+-|dD=Z>48ikqD03%sgaPvc>qo^DHc1V4o
z*8Li=N#oQ0CP<y37x?<6YCta{69-6B1P~oXFBqrhAy2${P__C=S%2c0kR?1kxKMO>
z2v$~%iz32RFtFNYQdGCb(SiN`))k^-q)q*2PJ?9miLGk^HVucL*bI^Z<Y@!_{8ZJ|
z9qsM$0-}74zGA4NML10tLiz6QQuI5ejK0A5Nq*cr{2IQ%jY<7TCYJA96k6a+kzbe*
zK8hn9@b2<0w($LW%|j!dI8Al2hqzIW!zeEk@e!0m{#^&eLG#V~_eXm*?uG!noo9+o
z;nPm>zdl_B=?TzES3@GuxT71x#x9JJT}VDRo5y3N;g8AR3Y<N&xR9I%3IAq)=38uV
zCKpi;)oVVX4XK|oU~R#XWKdJ#jM)qZS-Arnz$~n@XX7xr6`@9Z(@w31vt?Po{32pv
z7T_;IwtzSZ7=3E1rLYA2FWd=-)FF`$<OCczALO#qVw~LPV$>16Gr&k-b$9Q+eB}yU
z#9J_#liZ5)TqaX49ISB2P{-e3Ow!#4z$ntTNpD~K9p!6$h_Z%;0N@uWZ6wcD-M)Rh
z+<QG*r;Qz6&k_DF$~bHr5bGg8rme-aHPQ<6;}ak53f&>mE74~A1O&$LL3h_OL6!>B
z;gV*2H>^sLL?`AExZ*v|b>Cp9nC>m0=6^cjW)e0oQ`X`Yh#KzP$tg$@DhyC%+kmM~
z*nHUDV3i;cg5Qr=X|yZ^dZ?gqN9zJCZofTeQlhpNx<<<4f;)~JIl|7)jxYnJ4j~4{
zUS6otcE8n}YJ<8F+QUP}3g7L(vAJ7od;IwEvuA-w&g0d7`|(2`Pf*u*6Kl5x$>!$G
zIayhJ`q^nD(zDOBU;Ut_jsOsEMOj(d((-9xp&OjJdMSS-GOIu3P6SB$uB~Mi6$5Z2
ze)qG2)+Mq!i=8=?N`0eK@RuoNt!9j4-}aP2OV3s>jOson2954`n=VsR{&^uQ)IDB0
zq`rz=x~G6_N@V+D39ElXmB7g)pLU5eS&c9;{w007Pjz^_<!J4W+Dy%=bmc!TK5@M9
z)5;6d4i>QT5s=80VU2!?W-qF>@*jpB{O-jRonUb74|x=6n9-eYpW5V0I${P3L!2J7
z9@s*p;G8LU@SHRSbG<nftQ$mbXP(Nrgf-QK;h;i6=xMTKn6SS_K4N3=EV7i%5}47|
zuRhGX;TGF|gWNj*Cq3<#v8;;1GUjzr7ka{2DT1}&vxI8Q^DyhwC3mwaYq8<4kr6)r
zd9<)XFzfQ8vA(USD_<D8I1X_{WzTx_)gc4g|B3PNX4IJ}x;6s}7jhMG18V0Qn$;hU
zt`5||{7O68+CX;SGugN_U<cMk@Cu+R=EQyPez7KMqx$FuAJ6}ZQ55w{*n!{(QgZ<m
z6!BxrjiU2-QO!o!qsc*<Y=ywj)MGL-2#`6v(j`Y-e5dH(6gEA)vazJ*ZA623ksT+O
z|7iiqisIYXZ#_LIBw1e~qz*6<_J4X%RMFinCnR)o$+4%*7a?`dc~85;;LboSXYUG;
zPeZI@@KR3V=)!akh%|~#Kh7HC6v15eWJt!_p!V%@;VBcvCei{51V5Pr7XOg`5JP(<
zcLCa7^e1;-boT~lXYWzLKW;`~ne-1HKc+Nh+WZtMtZZTok`5Md@5#a+0z$?2E-IIn
zfVB6?B5&2O0Pe=YA^W>IFix0UtNoj2J#dg^!A6-I0XfZnu(S=01Tb-YzjZP55Accp
zOF4=OAH=gX04DS&x{T=UwhW7-qhk*uj@ReS?baX7C^~>8+9N9Jh%!!d>uVsRa5Xze
zw?Dj#n3dueH7xR_QEG`N$Tf`h+fcgu(%!C@D<?r%H$CrLP>Vro`O5n5Iw6^$mNPIg
zfd2ctSIb)3h-B29{yA)@1KTt!%jon%6s0_ZRd~G6%}BT?6()>b?DZ&;Z3B$@^_Sny
z$rJrvHh1J8kUsEBb_+=a;2t40Xl%}k0eB@<|4`UHLK0?But_4|5&ex$a+6ECh~CR%
zSJ9&V(%Ss|WAtGAWC>yF8~+z5@YYq({X8cEs2Ih;)89DOP_S(wrO(y=zP_@3WlkD+
zXK=`uh9S%ttarAl>liA`*$&WWKTPAYO3(L4R1yDpV}0kZ1SkGm^BB3yzh9^R_uGFK
z;{TmBxo9#>L*cATQfuTK5o8?AOMjFqcq3AilMf!jm43wmOwljcRP_l6*MX1N9dKk9
z$lJTo$ssFS0zHjCe?}2}U-`eWX7~~BJIMV1{YqmeLb!jv{vW@+OU;Smunop-0w=Tn
z^Vvll;1we8Yz{@fNLU}RA`}!ZVSu5N3vZm2Pi-CHCu65Dd5yKJtBaQNcyI*bq_|(1
z_}0q}p-?9N`}NR|9IL-yPw@!<{pbIF`+wJjg5v*w8=|J7`Hv|L9z=*{4D^7B4r!*T
zj*!#XSuQTFoCP9DiJH@^6nkutH>=HHZmP`^E#$*p$S!(@hZ(nT4?VRR8Zu<WwB9Es
z{D|$?-+_V>^&t2F+0bh{51cXUr6~Ec|IXj0ApioWMf-<`5wF)iv)qiI9XRsGuHRTk
zIKPnl)N9tk-#vdeXNhBaWAxn_TKv+DbwdsX&uYp-y=cFX`;drWAVagxGXe`5pO~mI
z-HSZ8#BJ6lVv5)U%`$7mJ`@xYS^yrAIhwxOmhu=M$5yzF%od|Ke=xdn<nMLc1OCq1
z+CKmg8uiWtiv}u~xPrkc;W?PxMM2>YU?2awQ4X5X3nbNMzZ!JNhxbwZHE;W?M>X?M
z5`%R8t6a<}qqxCL0xw*gOq>E2Lin0B770YfS4DE<$J_qr?=S)!(T>@uGMO<((MSE?
z_v`#`<UN0W8CcD$#6O?+KY#EHtBO(pV@EOwR|o;xTQEoM|6W`BYuCm|iSyc?#7+3V
z9rb^=5&7t)^UZ(Pk>-`r5o-K|?XM@j(T~Ar?P!%Ks{w)twrAbv75tQf{lEKY#LQlR
zdz_~^|9;-~zn_<X%lGYX^u3^pCr~IwOsUUbi&{k|i&jMdD=7w~AqH`5?ae<Qxc;xV
z?xRA^h>Gc@f=}=>v1{VVi&XgUXFG=n9C;v)3!;O4%AcqH=fBGx1DOoUNr_tD(SVcn
zs>gslg58can)nw*zL!So$j8(m#dWlid&rJ>6|9lbYEXcIdt~$_DLtGk%<V}5(AbSu
z4(~;BzkgroVF7S0Pobf;T>bblY>FGlNPLQTq>!YYnQ?zS#{PHYsGMg+B;)Jn_sW5c
z?%Y9Gbg9A(a%8q2a9-d{dQWxn117L)e}OeQY{*s^ylZz=B0#vS!0Xf?CQZJ+xxYkJ
z1z?Ur0ezLgK1cHW<m)iq-$y0?t@C$-{_nT{zq2Oy1ClX4k_5;=gkWX_z`aETg8U~w
z{v2pNNPE$!0%Qo_lg%(q2^T>&kUfK!449$~d3INqEVe!`FE2Ou5OyMx?}#xreB%Ba
ze>PdLTYg?186}RoKbRZ)(T1X*<KyE4936r%WNJI0Dv9>@vxi|s5D5_>TvXc;f<cf6
zEfTq?0KhPzo)gtP0c;m#6pnIrRL5{>d?9TM@bmNY_ov>p$!%fklCrXo#2~JI!#bHV
zm@DT>cULnj4AWsXkSR!WFY3~d)UKn#kE85IFK9D7R=Nw=qa#u-B#+d~At=G7s>hmH
zS;g9rQP*{vwP*qXI)UB?AQO-U$k#}%y(x!HU$q?uq<+Ye_19LrKT%hH1KLUmC}`6u
zj%<ng=Mb_%2>he)WTsu&MdTnaVSJoo;wKc#(TLp(OfmWQ^&9P2(+|i5$(#*j8b}DC
zuy+bj!>c83JWAEpKQ`;IfH5jWNRhCLc?AXf1%e+4(2ElM#gRL|;c<Y3P5_G~e65m1
z{2kl5Kj!B~=aJNp7|&}9^f_W|!|@5CK=Y<-6}T0y=89<<?e1*&^|TAVTI8=@M<HFo
zgJh48In&4jf%pL8fJr{)x8n&gDWo^{Dc-+|5jCS*YXzPKm<yf=DK7w-97X;6L=zGm
z+S)P0_Kprr=mZt)5*E1K9Runps{R@QJ1=CR;MSeGY^AT?ftmrRbM{b0o&OBP34@>9
zpvR0PthPb6<TW6{k=|Y+YKjg#F7VC^s^7miK~4+PuHBi5#12KC$=TV)58ny(1|>iF
zt?1kYUPA>r7Qho&mM?7<?(&l}Gag7vfdB|Qkb5y@`D2{kOaMbgPVgeaLW-C7DKKn6
zo;qKaA*+XXekSkg1(ZBdql$WT6qVgw16lOEI#uQ69C&WPM<OCqz}C<j$0)U#mZ7Hs
z^=^<%05;NW*gybX4E%{;tx~`?2QULKU<)kD*z4Q3Z|K2nsjN)J!SJJl9}jOjumCb1
z0Y`L(zJ2=umJZr|TA(j(D+mfS4gnx46fez(u99Wk_h(n!=qCbZAi?;>La_n>S1dMV
z@)sX~H~=SrEvZL`*<87T4!Sp})rwktKmk}IGY6RWD5^bpm3W5XLv6rZ@uxMlk^A+G
zeKd(?E;bCme)fR^gRMJv=Ay=gy$W{NYsE!C%_k3X&0&`|#!IDseE{uKD57+v5hDRa
zo!(OhH31Zgnu`sBF@Fp~MPT;wbq#WpR=P1xRb}Nlka!8B8@({dfA|`XJ2^VWV<!QW
zeAVZJXFG78<!}P#4!Nm$F5xJzHnz6v{(K5O1tatM?{}#gCx9K&dT;ji^+jUj_vqRa
zK+rs6hL4V{)75Rofw;t~TWarty-cB^Ag|MpqYD5+@p4B;$K5-3s`z#ZJ+`>&=%|hs
z%wGUD&9YW40BQQk%+Ac5{?ZSN<X4y!+f#(CZt0m89v+Uf7301W9~b-VT;XZA?AuN;
z_mjKN<3<k~>Z0!g56f%giW(XUZdv1J@D9nKKa`nJu(IqCz|jDm2$hCs*X>^O`0e1a
zgPqFp=Y9;l8Kb@4ryY=xAa8D>h7f^fn8yKd^HeMYD~&buJoV(Fdv#!o82F;#qdn>2
zw6d})XIvl~K=yBRvB|3aRdP;FPEQIFSJMk!SFu+hF%8@#Ci4oKH6Wd52&}|^EDtM#
zZ5DNs$PFS26y^o9B)%vGe_NE&f)n58g%Xey$lp9NS~}?IFQ0(fj{TZ)7jX>_Fn!yw
z4%bw2WzPSu3@9q0BL$ZH;`f`VaCCtNnB_#|0zo)#`7D1bai{xfk!z!85<>wuqAaRv
zsNHLJ7%&WaV)h6&*6B>)U}9p0sv!$2YewdjmD!!UcgLW~Ypabz3l|f<n-X!)jlV(L
z4KFC18fKzjbq1Ndp5qR*Pl5H9e%T$i1qcpG$AutD1@s@sVp+}T3;|zyj)ESlBskHh
zVvV~olhox0L;1K9U`Lq0ls07`YW<ljAmP>RQ+xoWW#f5ZTs^(L^){!!K)Uq#nHL|2
z$&d~8Klc&$uICq?pTaIwKjs$>@cz0XA;%v&ITw$3g1Clqc1lu`4+o(n7C5E@)GwOA
z)&mTPHVrrzaL*yNd}Kcq?%$(4JqbOqj85$NH*XGKs>X`pV6iuH3FD*!&Z;sP>MkuU
zx(cKcmZU-R^)Xb^estud7fFL0j2iwx&Luy@>}i)z#xWrvykY(BPgd~e9ZD@bYN7p>
z^<yjQ`G{5hVxgBJ0&TWS)aSO|!6tZeQT>TtR!WD+RZ%EAgj9T3Coc|D59dwCS4;bZ
zi<1*R_mrWw?PnMAAi5a>8P?YnBgs4g7>o(KVnaEgs~<jSPR59ORrDkSX>WHri*;cj
zJ!A;>Pr_xU$B>z@AoBG0!+XHnCH4s5g~)iW6!oNFQrm~Aqm=8{qiL|V7PD{!%NP98
zqj=Z91#s35xH0iadHJyi(k_p$iC;$DOs~|V`2PKOvngmZM%2*le@Gq0mc(2Vd>CvY
zi5AaaYv{$XZy$Gr`FEjEkSKNbLs!ZfHMR~zOSQ94b5bl&)>G8}x*0`;lr!D9f}!(Y
zW@b%oKuR4zMrh;f8j=nlv(B(6@DKpx?JyZa415kgKyg<cce$LGJX%n%v68Imf3~I9
zojcqSZqCkCVG&W4U%#HqT*BFBazmTo(aa*!mc^=nS+!7quB$@P<sYu{jl;ms!y9T>
zP)`J_&i@CtAJ+42Qi<vUO4jSk%g_;p{>Urfy)s&UkoMcMl|kXyvHrs(fIoluPeS?f
zCAKtA1gb#Qv|h1F{A|0nZF`idd*;lUyZYLu*@kJ{JUrD?ifiRj&y)j0!k-F+2~4GB
zUwInU)DNB}h@vU1RVc4xgMk}-x>FKm(VTJDh(BkVGpCdn7kjnluX}ulu7*#CNKRz1
z<$3&%eQE>m((5HpsHxRzN}=R};}#Zq&*dC;9)yMkGJ+qKA43EDb$55Xc9Z!ORK$=M
zfCAS&^;zf%vO>Rp*Wwj$1|&X{SNAtF**}W(&hlVCh&;Rl0}b3cF8v5OnB1OlnCn`p
zhv+BCCSGum=V0r~G*Nm#ItMa@6v}d!H&anbqNUIGC+HgspkN^S3B3XzGlZE@fM+A#
z3dF)78T&b{TPwd|!v-IJ|GLZsT=Z{ndj0}}8*n2}cIqJ9!pua|>ODJZQ7aaiIr(J)
zZ?Vp)ni`)R%4PpLGxOknKCIqFm`;SBDI$iTMWJaO$EH@y%gZYV-O^ZC#$T`IVG9M+
z7X|uzd&9zHNFwxKlmUs4C%I(2C{BCsuzG*v1EU13WrSFM_c425u&*!avko%j9krkI
zYyUwJC5#>flMi3NE@OzA2rt$f^B#+N>Y-Ng0~{`tPb06m!j~Y2B=uOuEA@w9JRmip
zYO9T+lP?#sozMOv0WXW-?kU%<<+iB-sQ_jZBfIX~?EPyYP;$qFw%Dc=ynHm#)jHW;
zy$XZIy2AiBVE`ruomjlME&JC7(=Y@jC1}n{!$!Re`V3hWUOCXsc+_l(XeM?g2+JW?
zNM!=G83e;uu6YZqrhWt|zy;J<OF9oUt{}SVJL8FUVf*{;?QUa>AfQN3-iA<MdV2cn
z%L_-5f9~LOc6EjS`akTwcTiRPwl#=)%o#zFf+We3ksyd5NY1E;pyVhhAUPNi1wnES
zN)8GLn20J0l7oOm0VN7Z4wApIJoojh>aObUs;>U)d(^Fa@8JM@@3nqwhB3z&a|7WL
zR<f{&2`jnxx3YQovy7Pk!{Ns^nbHAw2;Oi>mp5^yEDvuUJ48-S99EZQ=r@KQq;a0Y
z5vc-Op`)|2n)oD17$2gd<4b;Odh_N@N*G0>^BHc4x4Lm7`!WkVJ7EJO(n8=5kNiih
z@INFkA7QQp<)#-XE`GXTlI{SL3FH`1+fdQ*3J3_$(<h=Y!rUIEC<la`MZd8j!wk|F
zznPx|`wou87{%hTLn%TG+T|GeJyslARN2K)R5)T!!mo>ackdPf8`&`)WgCcrS8r)T
zyMs~7Nu{Wu$-DYx?uDor9G=Syf)fS{fqN%L9YmpbxVL>TO6k|2T+HrXK<M?vi{f>t
zL{YMk@CE+ch3UHEx<miU%NdoK_P2f0;z^!?`~#&z=Jt<J1Ynz5hz%nIu87it1EV^l
zIAS;$*%+EQz6OWvBnyjGd-ESfOvP2A`6|LT{;aN=3HTFM(Ll&e%qh+t0+o#BC<RX4
zf5Fp*GN|ogvVbfUulyS>wf3kmOs6(bW1+~&G)btI_uCtw@M~ekVZ1~-BsBCr48w4)
zb&DAB-d)K4J~t;QK!7Jkf~fo;f{q^w0Hqh@O%}6w`|#9MA08(3lbloGR3HYTTuV{^
z0YhF-m|;9wV9|vp989mD9!135WxC-%8A=r783^fT0pp-n3r#$3WN1i`udF+sUHpOD
z0}ohYYn!=whVKp3#*0wWLTIIP{rYtu(NK@_@^Vxl>FOWDDrD^6<1`aqU+C@%!MaUd
zMQa$8{+}i%IVdOy2RjnHJG5V5@hZy6HD!QtS&A|ppTX7DRRs%z@R7R$kmn&U{0bR_
zFlxoFS2m^kUZ`on+Xkx(1}|kUtBcvoETQ`awhm!J59UgBb%SxDxA#AYLa2bD+Q|n^
z7NFqQ1Sg^-IQD^*=mRAu$*HIc{~GvQysWEBFlKN7(53t<p2j<R<m<cpz=1MEW<1jT
zU3P|sKS7Sd=QRQc0K#t!=>Z=8pF19?2h&c2vX%Glov=&c71}7sByi+wYB4xBsQCdQ
zAfeV9sf|=EBev6?+_CQA;Q?Z;QfB54Fk38RiBJF<C5H{1JQVR}R8%}y7jI~3X&D<2
zqVfdhNDU;mnt<6()c&&jwDlZ`oj!CoU{jeABNXDR<RH8T3AQ6}sLlufqMgL||KH4H
z{)GbnzyI{~W=C=3JJy^fuRqe^2NaGRI(In5ob%4EB+tC_dCBUn9aaK)pN}fGQW_~w
zG@n0mzw$EUo=4B}9+Pn#)*?Hcx>s)Jxvt)W!yENh8}_EAk>67(C8jR$xD8K4->@H!
zvCko3y}w^42zB*;f1Ug9VGq8!>)$~CpMNd+FKF;zKYZ`sK>MG6Jmme~6C<3H6aSX;
z|M{2y{QHRf=b!%PCI5H1uo|`#J4hs3_|OR;e3s?u5{1j=aY&|xZ}yWEmLduQE7;#h
zOPmZDXJ_ZI@Nhv#j%SI)dD%ROjsv-<(uTmP*PNnaVrFe@o0GP?5{2?~v%0p7YvAy%
zssCe8`9H@F*X94OhijqI27g_k4C5>mC||umF0KCjIptP&OHb>!Z{R}7F_D@oqP7F0
zfhJF$WSTArc{3rQp%+tw)XdDz%fGjn%RyG?;^IOk1B1iD!a_ku`<Qqb4uLsb@K#t<
zl)z>4y0(#QZfR?4!vw*=Ht*FdhnqLeUo1-U|BWwd0K2AVW)e3gV6%WKf%>g@eLWmX
zctMa^j6m1x-MncuJ39+>PghF|77%IApLb)gExyRf$z{e~WM$O=;e~KAGm{r?Td-1Y
zPI{0}{o)<gLx4t%EF2R<4Xxs~ZTFG+LAGMmFawJoJTpKp^76i@((%So(Gb5gHa3PL
z6CYr9)(#(=kU(;9;Fk+I!sM9XroMgqc4lT~7#R8Z_#8cY6p5K{G9ftl<Jy}ReAE@V
zI97d6*o*>(grW#c4_q)x8k*SXX#B3AW3r?LYA~S4GJZ)A&ClWq5+W`b*lfqndd4kZ
zQdET7_@#gjS~bz_jcO&9mX;ctnvWkp#)rlATbQ2*&|X(xAKqqKF*Hnk9i^bqBs0N>
zLM8uaIvGy*&=tTnq^s)+-n+E4G=L!Owv&;fNm*I!xHc>-=_MsSO-(N$q}Z`z$Nruj
zsHD*Z_n^G#G_<h+`Lbi0>#pfuq!4p+ylGUJDVw0Yz9smqizEU$pur*OzkKWPyQ{6S
zG4Sm)iUG(MtAArMfXSScs=B_uCw40}$oe-}k$@G4-U+BDc>nIJi(Dh>xT+T}d_ht3
zgo?isZSbO{qJjdXe@NnldJS*Wv^Xov5(H@&Uh|^(#p;5gW75ZQvmZe3fHE4TrHDwh
z_?LT0ta$S%v6iqK6Z%EmLR>Jui|Ej4C@G!6_QGT;DT!fG8bB-Rr6@iV1yoAmj+D36
zwY6ccPN*#bhw`H+&(9B8&%lHLE24{wSE28~8WP4@lU~T~b5Pr$xxxP{14jue!MHug
zj~&DK4j9-xFap0zZEtO_VPb${kw+ItOaL{Wsl*8ZZtf{GtMj@S_4OM(tI+Af+6g38
zI$*dHPgP0!ANl!--un5ETP+G+PA)E&y%%I<^BYa}fg%cVc4B%}O^xf!&+TSLVDDj!
zrna<r??srik0IVkFDfX^VB(^pszd@b$V`ll0RX>6XX=j)7(W`X;bNio!|0DNmjc^;
zc5Y5&JFsp536KwA0~UZy9Y$;19h9Z}_U{jP{FsA{?Z=s{)KoA}CdS9bfX))C;vYO@
z)1Xwo-`o&8b7l?gG+@9wNBntk!N5^PagU3C;Ykb*C{~1gSS1{ZA2gJM_Bq=9mV3??
zZo<Iayw+weh){z7wVOmhk$?%OY|F8;d~7unP|BAx#LCOD)n74bc+6=wS@`)2@?sea
zy*4H0(&El~mF2~@l)Sj|+HX$q3@B%6KY)-;d)h<jR$E*9VA0U#YTj+k?>!IfVj}e&
z1XPF@$eg%e5B~f7?2w|4r34yJbl%Q4@XQkT6JJxGjNLGt1xxVN>(?e{jPY2Y#H`h@
zS5oqC)s?_80!GaZA36LWE{sm8>+6Z+LH~dNTzW2zAGrFc{Fzo|-USA#LgnSm2<>xu
zrxlKzf(MmxzUR7Yuk=kjJIgB5u*gV^JK6w)jafn2?+&PIhmCIN7-bdG<Xsyt)Uify
zWf+3xu(Mj>eXMYw;2Cai?nMd(Jw0(QF6E4uXq~6pr!n_E;^K#9HlW7@)We0F0w;o7
z8>ti+f`K7d0{E^Jz3t$DBDeec_gJ)M%GepYC~pfW{yj8At(Le{6r7h&3=E#Zz4P+g
z?CI&*A*`?A>bipaU`$2P2Ll}{vD?<Lt{}^LnUZqe_zPOEyuDVImT2T<?`{hZZ896c
zw8rXBrR05(<(@@LQI_5!EJ)<z%@51ka*BwE#C7I^O{cA`Ej9F@e3VRv@YgZt_HL$0
zs1jrT+|ls|9qZ3FoZG95d$8o7DJa-s(YDwh0T(sC(~KpsA8t3=D4aVJGqclmGYU*0
zP+u0&^2w+J_&$>L5*YIBY-LqdZZIWF;^S_uP*YRSpTFzVN>)CK5#R0Gw}%8kI{kor
z(9+V<(D0<rmsq?<glh|Z&2h>q1At`?t7y!ri@8X5myZUpaB}+3X~*yw+>>)j<Pnuo
zN9TE@Yp=olj*E0=8PJQ=@D{nFQ{LZF8ZG8zqt`8~-il}b{1_hC>=t2<7|cJB1hyUH
z(}j5&Cif<`gU~;WX`MAI-}BI!avfk(+$^9MjftB>cX?drS_i*!p=1)EBajL%{CM<f
z+k^>5j}AZ+8ka9;q&S{qtg5Z0IqP}G%u}En4c3@|$pVyTN?Ok?vNAJae!BsX!F!MQ
zg9ib}#Uh;Y&k`HL*PvZV=8q84fu;73fQ_p3SFc*z+fPHvm{NyLTooH{5mWK@YAio|
zVGt4NY-tJAQPP!>F~OV*kpgKA0rxRz?+OZ-(C*r|&$v;k52HI+D&S$D&TJ#0OGTQv
z`G!zrmSw$+_;f*eZ69v6Afyd^L-AUeS-~%@xIFtgH|818?Z-7G<B@{>K&DkZP?Q(~
zHvgR!i{aG1W5O>1DI|=+If3}}J;kfb%W!=6DIEe5^V5T##N;m<^Vu!bj?m6NDwR=I
zz5)KjY}-Zf{P#&4(LE;8DDEW?1HdBWIv#jPe*r}=IS4^6wmydU;33Dh;1NWuIa5s$
z&$V}I4;GXma|4+d&jRueQ{fF0lOYV(KCh}782rFsHS78F>W7Ga2$Ie_c}_Ah5gvMz
zrW{hr%GoGu0onkOMZ6%F|5bo_5Y;%YuoD8@{JeGdn{2gc0I0B!yOI)uS*N`OE!ucI
zi1Ge6{v6DB@#2`6BPV}PcX>GjAKz@?*^TMNMbk#@_HW-HdY=cB04HsVlgw7vulK{Y
zu6_E(jz4lTGBUR!jBYtNpo$60eS`D@!5>Gm-ZS?-35l>&82I_qnT}5g6$IfPKzc>k
z1H+(_@Ap^IGBYhenny)8ILH(-W{iMQlFWeMN<3!0dJYZ_!~{KC_jID57NI{!u~yqY
zQG|ORBYewHMrQj|GHM%KL4028_MCVK>pCs1tZ>ZgbjH$Dw6wKFE-9fNu~T~dPtdl%
z{c*qmaQ{L7Sz;W)@OTmiCESdR(mx#Ue#MY>71$2yh`QQZ$kLIKRNy;>OAa{W!O08P
z-rc<0!x}6H(5gcJoA2<z01|U7Kmue+_(3ZroQdvnLez|-I=D~+BpSFXY6(twp#YKa
z_lFQjACLgxd$n23Ug<PeOCW3T`i%=fD1uoUKq34$zJXaLQ?3!llkK=GY@VEae7w9k
zeZ8o+_U_$FxVj^CjgAuREHE3O3iMvuGlv`VbGHE||E~d=z?l`rBw~WBj10VKKEnuO
zs_*GrcYuQ`Do@mIpq>K8WoVn3k<s7tY1LL2(mx12N8i`i*8}XJCvDYKS0`4eU`L4R
zaA`LpH<C7WQX&c6dtB+8Z>q#4?hqcs)tkjv1O-l+#3v$FD%rGt{fau_yp7EyijSw_
zZV?Fy#9a%mu(P+1zP7OAS!SjPwj%&TLfwifK}JRfSW+gYrgkUYc;V;lV!H`nzIdm5
z-!aXyv$MnC3ib?Pv9TNiC3lv{HH%R6f>aQqcL6aC5GNo!GgDIu=jl88@*BW|fqX-p
zm92Tk19t|&V$Yk<xSK8j&VohaNop7xU)>_PBdZ0&(3qGjSFc)IT6O_z(fM}fFydKM
zM8rjPbps2FOu32t-8Fdhs;V$_z*C630x@z#F?+?eX=GKr-H3<?YAUMe*x1L$AHtB?
z;2!`(9t(4Gq-oj7tUY`8LSG1v93a13oSbA9Aq)^P3hT-S(q_wHel!Fyh=O_~X_xPL
zsrQCR_$a~wcv^yN&~%7i1gADNJ^ei_fN&pS#0BGnCknV4cr5np*@I_7Nm&`b2=)Ff
zH=Kd%tw6O4S(}0cIu0S@szZ<up>YK(w;>q~!{8}I9+ZD<X33zv-Mq;Y8j%OI7{S+N
zV@<>@bDLihmfJyEC2l{gw0uxj2o)&APN>iNOT#dqvobdy1-e;H&W+E5CAJQ?Z+Ft5
z&Ihbt!<Lzq<vQ6TVp^z<!VD2bR9xHxb?by7sjo3;fru-b^Fb+aaD;Ki6ylb1`qs-=
zu1t31CF3msFl44vR#qNL;Y^Q?i2<#GSYR{w_8AI6a2II#jH{q2_~erD=FK8z5+!*%
zy1)#8wyL_<4th7@X$ars##3Vus79>!FsYsv!vr4J4DNMmWZJsA#fBAo8c4d+GK-2t
z(av*l;FeZ(b;ZoRUg?1R7i1jboZ?|X3^{r7<ZPR!t}Z5O-7pkJ+XdezQRdB5?Ms({
z7>aUptLo`}hYxQ}jlv@HF=s2yn&Jq|ph81|02(4pV8b}yT;>KD85t-I_Co9(Z(T4!
zM}TEfj!BVYE~%-hA@1Sa9NILjp0O}AWLSFWcL_d<h-S*l0qE@9^=5Fuk;{fhMk3LU
z;fUd+Mn|*dJ-D6P!>{G)D!NQ&)lMVeMUGzkFBc#OJGHjX<@)-{OG-+j<HEp0n3=}3
z|4~}nFmQB;ITUgxF?TAn?UO*Y3Pdzm41TA+S5H1%3IY^Tm5_`WTnLyJp*8e-l#j?D
z(IvTtkjdb+H=wOCe)|RzZfL0SMJVu|Jc&r%!1%jhPNMm}mfgNT2r96hF>ne*aJql*
z-uVcw?P^bC>+24X1f=<L^Ws5sAK2tk41FqMpN7(B;u`{oB_z;j*%(wWT+~)q&%?J-
zO65I=Vig7KZFTivIc<1FC}}f1mPBcVOP<DLc8<Vfky>7UqC$C4e9n1etx#@V3ge2r
z>mRq~r`uRtPmGLEK#2e9#K9@j<k`~xSGasg?Tt~qpDHVTS|~wcf|3EN1^p%xVq$7s
zGO^zX`PszArXh!LSj9ol&)tj&4xT}Ke6!cpFDWiA4zO|P)2G}<;tCkc@15Ptx{)tu
zeEFS*vhr`RJ>p6S^$lN@l&oP?hKI(yEsNaD38xyfd^DekagQzmwUF{GkG_2Zx>p!n
zJ1j0l;qD;Q5WbeLT}XXX^7A2BOhyDg$<4Wc-#++X&0we|G{_fwJOAbnsaX#~2d0R@
zQ3B^OMrT3HB8eI?>FH+cbOB1<`49Ns&-sJS63!_E;b?vXxMQ=632)vw0?@vHER+oD
z$C8I@u;mOT_z)bgJ5fLlWZy)g%HS3r5fQRf3JKBkro1*8##6k6gl6~`T65#Ryo`)T
z$2GA<^{7r-p6HJGQ=sp@L$6cHj7*M_W<pF9RV}eM9)<fK1Hdp%+NJKx8YRX!+j_HL
z$xQrKx?lx{*t}+JR8-LK_t34$$UOL?IgN2z7o_I_{{F&rv_KtD|M<E4S2N3^oy72~
zrpyW5`_JKF$Js%7ohu;;31v`J!C<x_o*z#kh(B;NK-Tn|$ocFP@EgTC6lDGTcB*S?
zf)Ry$@a)eQjbG48SGmNBqI$i}otaDpN!(fYKOZ_3Cz)kYkbwTw0S~m<9s?5-Vi6a5
zPleyQFz<dmH)AT{2N0m+z1P_h+01|dVr3wQXW~axyWLQQgJV(c#tskxe1P!0z?7>B
zXQcy<iG*-guPW9ac)^d{7G{6FoG`Olc$k-!<>$on4R^b`98sD<(Wc|&T~2w=a8^G^
z?(*yoD1{Xli!Ha}+NvoGAu0ixn1~5HCQ`n-)sLH3LAStkzZo~&#Zq5s%3d~n<pNav
zS+wO(0|ynYw<#&!g-02vsHt&TcZ(lS&#J3a>YYgxfVbi3sCi;mhu+hB_qGNb0f-?N
zl?~ozfK6~|uP8#svC-mdYb#mq8^qeWirA{>DOGz<1<S!m`D>!|&Y1{vD=S=p%NaWs
zMg355Bb72lUdDt)N$H~8jT^twD7#$n=fe&b7li*rYNYOHj#L!ZkIu$(!tbckg4i9`
zTAJe1%;rZeV#j30y<9$upg|#TQLlFC62VNV3a7WNbQCqs!=VV?G63-Ii0uyy%h9f+
zzG(uRS%uzr$0wT0Ou5CV>z#L!!=3Qai-MOggPeGrQAZIc8yD?FcUo$yOz#*apYbjo
zJ0zcyHkft=aMsbJ!{n!}tt8YLO5>rN(4~9!=%atSVS`+30kOn;$VOUE@FHbtQBhY@
zQ;<s$l5)Gn65<2Tswb)z<GYn(lsC72Mv@@ci6t!A8Jvzpw_mVan}K#OMUk3@rr>Sp
zz|jYU72nzIFr#_EIj{$LwYqzMe~AKes1*G{za-2$PmtbHQB;f-un1}0{sHb2GM?7q
ze{!0E;Zz3FaVIE#M2P`5Qicx-KOPSnJ{7K=d@k+4GEyg*Yg0a;ZMd}<!swp>Q^PQQ
z|4$7KrvwDFQrKrJnLIw7;N=ZImh;gZ*%sZs+-u^l<rzNUw4cSWAJzh#3$JfD=RZvZ
z{q$VK8Mj4FosiRd{u-_W3LnMD>%M=F=cGD%R5oG_)5xk21Xm_SjDcU41wJ^f-2i`x
zFzvHuiLKx%%Ksz5pYPoaes*Q2A|BF+<Kf#4=7)C^i{3MzXJug%?PpZUY<?P-E)ku^
z?5riMs5GBtWSB!rm;Yi8Am${@fq^4xkfj1=L-+FI$HxPo00`~F5}?Js<!J*5zliM5
zU%W`!)YjC*2^vR$MX-Ve-G}q+0|Ns{FQ1=ltZnZDs;Z$uTFi`22lC#D?T=8e!Mv+l
ztUqi3bT!1J>e||5)|w9=o}sKje=wnLfl3b9o`QN-p03$@_Yoivy88Mmw(5Ape`_m5
zv?Kh7i=RE~2ZX_RB)bXY@u@j4VJfCU`(Iy~8;&+~cy-=7$_R^msIEf_bjW1XwOm}5
z(L@urT8=#z<N)#k5B@PW#xC%Wd=>I9@gka<j9UHp;1sT6>>M0uP0m0%fd@5H6`)To
zVpnMA6&gJyrG16LYm`hUPlnj*;oRz7y;}8s|Ni~pQCzUGts+Zea6SlAu-KB=+36qB
z`6*4xYP~$7w)^)_flFYl;pXA7jt;U`kHxlOXQnS(Rp3SY%kQ{_gd%i4zkpE%*0gPo
zy%G~!unbiW^n`K~QsS@Af5QW`>(i%X1hNUrKG1sM@5HdYt3XjTR_Hfe73<!kGCg(b
z6j*n5w{PEQN!wo_E-ZZ6JW;9=1JAQqKo}B2>X(G@;)aS0CT)wC2<sVuKiN%_z|far
zY7xb}KUzEaESwG&kFVf_!&U;Yp`y4zMc->=DqGtrygIK5>Pq5gfGEM4C#?gZR%uDe
zXb!gp;8LYN*%UhFWIw+>vsOtj%7}%`#x4djSHT}d5y`hPiT7s-ClpMIvsyHnn3xFb
z3YXdQ8IriM$jc8d+9)Y1&dtttq3`EhjtAI?gVXzuSEMl_H_hoMJ?|h!QE8keE+y@6
zsB2(q&)TG960#lywp-O<b~F}0jjh-?InTKN;Yq&?+<@??ysVg7>@s(<FjziQ3z-LZ
z6I~e?BdKX=A>V0kAsDTKtrX=20MKRU_u2+P8KG*=k?iqo`nB#pp~v`eX3Mi+_w&ik
zzy3VBd-o$!W~cpbB&Uwh27+_*n%&izsN|5l8yos>flZ9-lNJReM@NTBCtPDkd^Vg6
zQ9a=yP}+j(rpy~S+T+$n^@|rP{Z9f}>MnNTPIrN91lTr8O+-?R!4G?dlevFnCY48l
z4y2w~Jd0NJ>sO6MX;fWM3tsoRW?^MD_k`IXP&U2qHHO{Tb8wG}9ViMpC1q8aA1xd_
z)Z9AwS5PN!+q$(Qo1KjfCVi**yWV1O4=)aGo!zVUxZ76~M?wH$lZyy!;6p?fF;1D!
zBL*ZCRNyR^UW%7Z2C#tncLXQ5HfUF<$Ex<YTu5FTYg3F6MVVZ^Y7%B!nh!XwdT~R*
z8xL?O7wq~@T>FIOXpcE9ZrtF~;<tUKc4pwhC!o|ny=BZ%^~8mgR7<n@fkjMiTMuvj
z<2q4J61TV2V&;Hl##L{?qLi8xMK$mYx#Js~#7&LzNc9Wa+OOZeyF3}uDsL1ZYu2|g
zzqFJ)J}D?DIFV<HBQs29{^H3N{+vP#Bph~rXlo0Pw@O8PL&q?2q#oT*O*c;kc3DHx
zXhR?BtgNXSYVC!%JTAa>l8-OMq6D{1P3@4`I|uCXL6|uA-ljzJNz|okbR-Rp{McWF
zRKvR+kqnO3EW<(xMMXtTat5M9DyZ7<pIZ4@H60xn01P;wJlW(iY8juMvN1ZsiBrl0
z1^il}-2?F$2`;Vxm-noG`ETEjtmhXzygQfhaeNRbi|m~Hv|#7H9i4A~%wHcDpeEwQ
zPfgMXyW+*Gn@7>omb}fA2TSiH2PZo_DOKgE^VhCDRT^r4H0J6PBjF*c(@{}jf@qAB
zgixzf<iJLia#2%rpnm=9WgO@Wv=!hLJMX0vvGeA%HZ%mZyaIp|y*|RvjOka~s3$#;
zJIG4hMtT672_>fR{aoDQUn3(+fMlvP>}6$ltj6YGN;Y^h#7&^+JOjO(j#RF@#|Lv{
z8Uu$rrelnIl3Q3TdaF#;)erN3(7ABo;pyyu(%BJrF(3)4Kz!{K`KN4sQ8(XTBPYTF
zckSX&Um+$Y?@VPTF5mMEzK<Ud)H7v#rnt%wJke7aTIW#bxIAlOI`IMMXZ6+z0$0VA
zFrJoT2sSo0?&#{`tH{wWRfsUiH5>i?n=7o17a{f6?&Uv!g0_dz$c`MRIH{Om*vsSU
z>@3_lzW=}h)5R=JN)OyFV}s#<X&7b#?~0M?eedO!FNJ_D4MKN)z4G8!-V-P8X#U`S
zzs8{T4i^aN5plM}LcrUmyY)F{Vx~1~060)}6o5h58(f9P@b2Ba#`{Z-xVf<GZE-%5
zIp^6t)<~c>(q$m0LDT}?5TN;*)jYU5?k*G)ne?x8Q<Y%|lKYNoDj{%oGdM1;uA~`d
z8o~4eRN;If1J6BSCuw(<uyCppMe~vJty+oHQ1jv&%}i<z|AvIK8;D^w>e67ZL{Wp5
z_rc{Ant`}8Jg{*ovG{^Z-0;sTjFEv(RPp&eY^JK~_dts~lV$wTxD`3~aN~|6Xo@K)
zF4&BeC;(snsq9?udBo)*wH#AOMh5HbceCdi8I-~|Pw4pqmceweb&+*8CvRs1HVw$J
z2A_75<gnc^8S-ha-nfrU?p)@aPvwuYR)t3$_y4<G=ciGunw4Vf2WbtX%AY@eoXgoF
z4FC&UQH6BU|FDlECQfx0q^re4*JJzueQe1CL^6!`kYYN-^@T9yLw7D(qPmcoW0nET
zuJ!ggIdFS=KXV5#i;KsfKAR{v_6BvBeEp${w4|hbIF$K_99z74G$t+Wq^)&)Jse)$
zUDA;(F==|~r1RR6d@h5*z`<_j2L?v$e;BF`5XI8pBRsGtnOGW=A>=3-a9W}vQ-R6N
zT3?^GCGfPMAhZ2>p-vvr79PAeol}?O6x=_WyuxOaB2sbjQp|Z9lH+)as*dE1(9W7p
z8Zn#zXi?Ih<PZ8guXeCbdN_@&S)+#8qtk&W$z)ZPlm_RkP)0<(%F2J(jM1g@Tq0Bo
zi@<c0c!MXB2J80)^Q@o{j{ab7ZOv(W+7&Fpi7s3TSXC$=9dAzFxvHlqFHePnu(gX?
zQd>K`=|=%J!OZDrC^Chh00<p&v0ZEe94Zr$UZ5&Osbfk5&zD{MU!{a9+yb_@OZ@oM
z9S^_`u3{3jgQAz1l$GlVWdo@+H%55Jf3qq^%#Dp%&T3p6<$U1~KrU;>ZQ!)_=az`&
z^)GMGi<G;(V)JfxG0WrWuorZ`L}8w9;4oZDgZJYu`0`pRAcNgQ8#ofChA$YLSW{gw
zwY49y#su`|Z}mYv4o*&E)5ZCzzH21|Up}+CxV@>_p9~J;!@2F))mOkNHtRzG?48SW
z<`V~^CJg2QafUFn`4Ri(qjGN`ra*AwJmq0$`ls2ud=$9Y#A{h(9jji$xN|EwC4;3c
z81uEFh)KJwrA0<Hlwus)&_OaNAq{qQB9OwLIvZ?$d^|i+SV7Pl{+ot!YM%-c_)IBq
z&Gq&53foFGGLX$Bur_UmIWz2WX^dR_E1U0Ej72UsgvZCm(yaCd5~oNasfj8};)H+i
z8T!+rBm%cFY3M(GRo;lj_ejlJRa0v#thiE}k^vzS?|t+=P<DcD`6WV#`Xs*|vTrD@
zo(;~DOkJEv=tlCZS6U^sn6UR&?68<Wb8BS^u@eYbJ{@4fUwBr2y@zUsL4flSF(Ez~
zNMHC*P<S{wKbPK&wLc(>AJ)y^z8$qq5Qf`?$}~uo1W7jAN*py3!(1Y!_&}(`#Z5pT
zegoDfuqWgF6RZ1wLR9j`qUEIN@odB?q@GC5%4O#`36Ge%T3a(SCsf#u_pFN^7_<Mz
z(8$!pq`ITrCowhkHIPCW4%uy&qYmwiRyGzl=zZwSVR0Tre|aUPI5DSFQ^}jv(u>p6
zj{*X)-s`Qq?{-$dtehP1bNR$N6zp%_3}Swio0cZqdrtvYthu~&dtLSR$GA6dS=9Gq
z3CuYNyNw^Eg>uM5R7a==OMkQ)^LoU>O}N5J7oAj-?IVyfj`YK5o<b*2I<ByA8C2oX
z_bx55FWF{u2n~-?#59UXznyf^=hlJZ4R)PHJ!-fjq!F1J={M0mE&IjzG*lH0>?bfN
zP*GBPVI>Y=lfUp5Ec;PB7=7ykZh`iYbjpdy(|n80z(sJ8DhIopTU%kFz0=l`Tjc@x
zk->*trZMjSq@(a18`A)a4sK1sbS@~>WkDgBKMW#AI3~5)+(camz_o!X&~gvtN;9au
z?1f<BD`i<@3se&qe{Vg{bs({*FlKwjG7n-@ZkH=L4d@f%r>1e$o3KHPTBmAiYZJ%V
zpC!Gpr;0ARq~uI$oFIueVGW}E9j>3}a}J?dkEVUuI@$rX&F5nCiPtu;P#^zVSNVA!
zc|5CwabPeR$<{kLnem7s>9w^Wr!5S<TTxOWK^mR?tZ(e&bnJAjkI#-!hNtl%;lMqe
z<~DgKQtcSGeJqjA8#BfXB~D$7z{{^xD(ss2o{WKZS9iC)kfpi#^huh+TLU`^cE#=j
zeZ+2Um#tkjp1QnE=19|2R!hWzK@AlkfD^C&5I`VW1^FBL(pec9gf}gmk(;u`dfkjl
zbgaxvLLZ|TJmIu={L@pfjb9TJ#H^-wAv4UbYF?5>hC1^q{w_?mbn#rfw`U-EfD6vh
z{zynb01g&ssI-5`q(3yZvUCa$d$L{~DLNTQA=b!UpmIBu!84+@t<A(n2Fi`g5vhA)
zccs}vk;xb~DlZF(X`DCvF{Kx8l9KM(pDj~~^48P1bo=6<p3#bsLz-ZwvSFoSh(~OH
z|MfK6<f>A4>dv?HUiM?>5sCQ|vmu#`DE6@iCd=I34NwJ6LPvC*!OES`-b0DqJb|2z
zV9@d{=8KcbV9<m4G;l)?ntV^p)q7K$iN?~?8C&4IzmZu`+C5Fg{Q}=P7RC5v*Xzd~
z?V;@`4A8@j_xFD&-e>x$yTI%qe_;f%hU%xqB;o95Wwi>N6s`nPYiwZufNb1^c}lPD
zIfV-cNhsK^11Z9=V|292E%OhWBZL-vp@_&x)i~kc@44M*7hH7cPxsPAZ@TyGpuTim
zgKrCofFGWK#}wvRFX~axC@Az%pFT7a=Q<%l-@<}TMm;Yt@3w`h(HJc3(Cv+?+u^^9
zP&dP`6OKPDeIFQ@%lcmdfdhpEy{YI<)&f0L2?ykZ+V!k8`UD;Cc63~JEKh<j*vdhK
zc)$>GamQA!rBTJ9Q5#>i^ghN+zeJ)f3p@&iQe7FRlBz1k%XGF1hv-Y30y{Jhbzu2!
zMs{|+F9$gV#aWM)u+k3dT=dQ%n(26MF`f~g+1l_0HMHLg<x7_$mty>!1QgDn|0h~5
z4*<l}d#F=3VAoL9Ls7ynD>EEFEnDZ(zG!_FZJZvM-QdUL5`57KR$_bjA~%=&wvY2R
z4sLE@#0`K2^J(@01&3E4PJxD9aLM)AuhCHq9MCk^UOu=iHusb+ucBuu)mXyp*2{E3
zR$4PHQPF||XDTNrC)W~V0MqX7XKce>?xx0+H{J^TrLh*(v+<Ayfs>65J3|>Y=mXbE
zjrtyOojMiaYE+C0M^NxJviYl5VICt<zf1>*hp{qay7xyinP+=cSwLzkGyg>VL<f)8
z`1p8>^!c-uW%W2C0J=JpSJ5xPoJxFg&-0(oVnP@oAEItO0}D-rry7mjzDtw4rh?oT
z8=En=s7kmAfE22L>e#SACrVqa>i|q+Y?1P{+QTO+J9}Yi3BA>s<sb~qva^39mg4XV
zuh)T;f!VkKE32Gwj^(z?umrdLT8^^@Y3&HZ-?Qu_`~-|E=+{tR0yY{sRZBpToR^7!
z2Z`djkEV|Y^k^eYt*x=OK;CrwqtBnuV<(1PL?zxmAe`E=t%M!}SX7i3a$S9_o0?6@
zS-#EXvi$<5z%_oB$+$DxjwjtR^Apz%YwOQ@iK}JNG7KcDbyMs}i7(NJ<F^S{UN2x+
z_yXJk)*HMp-4@1%&{(V*JEoTOBMt*MM0ZsP-%IDPxR409b7RGoFI*K4Pf_zzCKeUN
zlq$t(Er{Dct&7x4f9>N90p3r*W)lrd$gif230!Xbm3G}x9wn7St@ZKa56qw9es^|u
zYCL@uf9C;l9*@Y5)a8!cpch-u(~6Fc&ZoE-2Vf=CDFdD4yLW#uN0!1VEn7tbazv6f
zupL~V^>bf!PCro<KcOK3Em`u}uZZA)ssR>Az0Uw;4rt8?(}$rOH*mgg-xi!YoyvC1
z4mWf3^cB#4Ff!oac&u;pm@_&q?hKS+xOIM4bMrwu+(*SX2q{(EZ!qP+GO9~gJUaNC
z#b|N^0o3`5n%apA`zP|Mm3jPa3=HTOA_}W?^4Wmn#E96RER{2rk*Pr2i&SD|YWl6%
z$rg(e@Xy8ANFaPwa`DN2=9Tf7kPt{^g3*Oy8m{o$9xTMY;<*e{G3OZ!RPv#DIaITw
za-8vZH;54d1-+FpL$x2M+)w2+@;mg=XGa&hkW?cW>9+$Fw-d81yY1NL^LBP{(74xN
zLW=sG?I*RteG3?ph4Rj0q5)|Jj~D_Rnhk8>6r3ouA~FSwswXC`Q+5&y2-$G-?F2?l
zPuGRD4S)K>5smWiJb8j$w{AeSWWBC2k;;QPf?28XHDbRK8cm{tZQge18?^DzJx*`D
zJ1%_lj&GtS*-(G~Au=*!TieFvZx|OM1U`(`a|Ota`6o}{6&;<KQx?y5{a1XY0Q;e-
zb$7r}@ckyf6V`Q6$}*-4LM0F&{u&!Qb@}B9M#lTS8W%4*q9w%B_!p(+h5QOFsz$_J
z8o&4bc$Q<KnxfJpXOQ@E73Y%HwAQ74u*~p<)v2%|hLN}ILsV!WF)#b1-<b96ndG9J
ztGxVf+xMDG$Es<6RR=LEF**z}Ca}Xu`TWepW*o1?0uD?7zeX91pP_=qkOaDA<ERB_
zL#Qw-6DuzlI~Ag2z`YUlQ|JDew^HyXfCi!j2UIl3<$dSQ9fZV!f-iesFJk&%(&-&r
zP7hH|sLt$N+y)9#(tHdT_%2Q0*@M~aD~!@I+rSw(NG*WLzHh;IO!xt%N2Ffutcqje
zGrhm;oY%64g^8(3eVZHQ@!~fplT?n=j2Vr09}{)JB;FKBLFp_#Om}#Qub?DWY3WZ0
z!-V}zs6=dv^A}l$V4QCQ{9HaumFwn4sT!qTuv2zMhTP=DZ#s*<%Gz2h97oi}brFDC
z2yNOsG22ToB}9dYUQY8_%J7|wHTBs$CcB%+M5NokeGAzwi2~x&r%$N!s))*49~(i=
zo!gRr)6;X~<Ncjjz@l66<XHm{2EwMnj^06ZcO#JJ&rX1z#Fp2;BX58>*P)FmAfDIY
zOE*yHX0{VRhpepZpa+P<rT7T;La-PlN`&P{D4ff^n?+iLfeRi++b+R2HD!tX#s9UZ
zwpMZU8>4Oh>%STM?}_rSBE4f&Qx$ps^<lii^2!Rv={U@bzzio0oN)dD$KeY}PXgN2
z(kJPPN=ox!6rC-``~&d^1v4g<IN4EF{+L;<v>jFb#Im4gMA2VjRTo|Wm(<ol9_t11
zC*d!iIw{kN)Iqa`gE(PQhV4#BI2^TQQbtDOF-p=A@;y?0+%ufV_@4BVw~uxP;Kjj*
zMx!<{HnxJjmv*$^M*T*Y*CFlcJjdyq;xQP%`LX!3j<jWa&UduMkm8ID4PiWkz3lZi
ziAWeA4n!NDMf5uD-z#viv!lZqEfR24B(f7M!+=DtSW=1$34LpCk1Toi`sK@Au#Ar(
zyLW!0L0~#+-3Rt)6`WL3Ik~KuavwfIkC|H-cd}e*mgX0qm5vo5lw7DvVMaWO@hv&h
z*^wikHWpaf+4t<)C5*U*q6zzmuH?$c!L{L~*QV#<Q2?%c_Y$NsxFX)8I`Gl;2;qtN
zmkSUb9$vwDhwbZIG`~iDq%U8+dZFoOkhw%drsd@ki}nj*^!$N9h^(1ZpOrYf9Z*g;
zlJ<DzK?4FexbbyObh2PNbnrS<Qe1OCSK0HayqsFtKOr&_g&ohy=cFjI8L{@`Oyph}
zby8pNtJ^OiokXkF4UdoGdF;j-t;I@ATT0<{tZ(!g@<gV-7Cd&-2UZ;I>=-_cG7Ov%
zvgxTR&LQPq)cYrKl8$+raqj6<x8wk!^ocpFQ+bl#q>A@^{<TCliR%|7;4^`09Q+|K
zj3#k$z~_cL1;EeGgCE~H{30#YGF{iZfho;X-^7CubznS?#_Q+rdZLGpkk7|){JD*!
z)0SgI&nP1Pa83O8%do`|(FIZkpm0dhtdXETAVN3>mY@PxdRwoE=lR2jJ4$qMAYC|&
zKX4+wwO*eGN-U3nz@YIK;`S9)QqxSujmyqdP+Y^Kepg8o>q3kiDdSJuU4mmRGzU60
z{z)0=8AZ9scNR?R4UF}ct`aR{aEf(YH3vQvfWATh2;`&Lv~8}lgS)5f*}h#K;^E`F
z^6n<Rh?Me}dpu4B-hFWULtMoYhhHI>ARG&zfB$mg^WjbTi7*OKlE49b-V*Z=*;k3D
zwZ1-JU#g(%muE!08kp#gijL+23J7&O27+jeQJ8|mH4IP8U-V-j=ioVj8O9&1bQsUv
zg6Xc(zV9DCT!%dg&NhIK!J;pSvg7!bl#3Zv#c8>@jmCPTZO_}$eHn{*`xxA~QD4oT
z^FoOcgIMqglKIz~z$pYl;Z$u|$u*Z?AHkbnWHanQh)1yO<gn;33;1cqYPU~}*_{k4
zadGbB1gl-FGR}gb1PT*F@nuj`(Q6nvb$btvjD!|efKlT;%mIPxt@2Tf!+<*kdz>Q^
zc~@8n4t8vT`smTaQ;R#wbQC=e1O)^*IXS8G_c!$ry=na}+l&X<(C`5W--w;qnt$Q(
zBUQ~I<-VREP?+Pu*@=%wdd7qid)3gsxr>?-apC35mnh=DeVK#j!PR2NfDJb|hTtY0
zWruq_#{kneOzs}#I?>V5HLlgWv`2XXWx<H@K}4vfGvpa@T%^8Cc*UMm4+&oZbl?*v
zUR4M34gn?M`)C4uXi7H-@{3=6En#+KQrIF(2NrrS+9qsbVHhBd!Hh=PEKeCh%gD%R
zcI2MDo+V_1KYz~7|I$>+!(a($l${VU1V0uc+y2b|B*$H7H41}E$YAqva$7y1*1_4A
z`1O)x|M8gVl#OXxJCF%6AffU}j+gLo1ufOMH6Qa-0Qj>*&oRtOp=dI-!Y0^b+8*cs
zo#7E9Cch*+9f8NiDplN=$;BRXOc~Y&^zwrH;0C#76OVXIl}?D^@q4XnGy%PDEx_0$
zYJ4mz0EY-n7X*6fF+*q*vBGW<*}1*NxEOm%SFn>R0Q(Fd?<*ju@~KYiFE%wYk`8n+
zi?rBNRdCjp<(veV_}C`=&+0!cMtwd`A3Q;DKhN-^^x@~?5SP*&ndn!s^tZBlp&8C0
zonO2JZuZHtnX&eE+evYJ!HvCXI4ULB8j2i-tQdM&*WP{(KG9UR-R-}E9?)h@B@Z8u
zVwfLjERKKMsOc5O;pBtILTzM^B-`}N);p%&zSDkW*DK>)oSkpdnjFkLTAd-OPUcZ@
zDoH<`);=RMGk{h+M%?tFN|ZjGp^;Jad(c$aHD-}IjVpKPYv|T`>WWNJhsFjbWnf#7
zPC|~swBx%@DrfRU;Ke4jN0%3m^IRt>oWVG<B(w#Er<@H80z#e5CC?2yKSj7$T#=Ci
z_KLiD8m{}j7og?v2t8pW>CI%Mv4J^4Z=Tj5`K~GNK0KL<l>xBNJYhC)l#DFsq0%sI
zaIpWwtGc?iG=G9-g0^kjRxNi8>Y?b#PVmbS8w%7Jf`cA%UPwB41$DHGcH?6wwljbs
z5^GCv>#)K7ZiyDHkaY*fGJ_4vy<uM_+Mg3d2rt4uKA&YL(gF&v_FwNYErhIGF~VZc
z@ok-rA&@u?1{@ha^toGAFt042tZV}x6}d@fa}`xRM9i+PIcS)gz4lS^NS@K&ZpQg$
zmrSo<n(f@$vVEb~tMP3jx~u@$<9?o`!^}YETj&XR@<g8520AT#kM45DJ$>qh{YEIg
zr{r{?{=k^|W9Og#kK6cMcCo>b7lUkKytf==r|Pl~EL7+F*#vz~Ja4vFy+4~CvzjxO
z+bQC379EkSt0-NO;*my-&jtVx3OO4BS;^p5b}|`qGP3E0D~y3x3dK8!R^I**RMgVF
zq@i+5NWctM=KY@AJKtIyI8I}3OiE%NLF~Kk?enJMH*H<gEQn0N$)>ZVa%7nTYvU(5
zpZq)wCMupMv5w)gF=*#$%E5!gj@IPb@%5GQT!w|0Q-gD~jAEh0?7rfl8XMwal*oN3
z8V1i~y`YE_?-ODM9X<p~X&z>R*5ES#h0e>EmI!pLtcSAF!TX8c2XViirsgM4sCNS8
zPrzVFO<SAU(|o^gUFu5!t(k5CRQx`FzHMw;>=cIaSyhNNRQ&Wyq<$gjE_^g+LE<Ut
zJcH!`U)ZW?S}-qXn%aV<;g$lB!h3pm_{rjt9*OL`r_>x%9@L%8>monIEq%}DV|8?X
z7nC--87j9B&VjTMMl~iTscO;i#!qx#hbH?6SwmgjV7&qr2F9n3qOcJYB%Vy;ZG>Dq
zf>@V1BsPfdgP1^p3Wla$$^_>hWA6sHs==K~48hY+o`G^rU#z5}a$suWA;dDiFM^h&
zy5MtCCcS}OcwKL8H!+;@Tl<DWnHF8q>8rgkZ|~KEFGck#X+4H#0LjYIz6wXj{{`av
z*G#|BCAMlHYb~s!tZth)SPkKmN$2x-@1n#v5W!MQDyW|ueMr#whQ*x0ArnWxiu4Ib
zB?P9XVpDRqnmC?%!!gVNhq>lZ05oj}6}O^0mviF80go?{0WRU*p37876vkuna&rA=
zY6cBrB|M7cm~qY~>a2-=`M3JCo>6WNTW`vaN}B!t{{CS_jJ6Z6OG@H2U!Y|g^S+NZ
zFL)tTQBZ@d4J{!^-UW}fFJ7Eg5e1KyMfHa9upc10S=!@n8hedEmzm&0XV8Cxe(G23
zQOE#IYRHPHN-kvgXwoYCnq?4yBMVRV2e#<Q$S>V=loIa3I$TrdU-k4{Ej+l4p0VV+
zo|DrOc#cKiD#fx5p!@<pIAMs3GuR{MmxOm%)vi-0et9?OMD=}rPwnq@00`810V?T)
zA=)($lMk^iUop*R3yO?nX}vOAw$=9F%Y<Fo6zko#?*%W@Jo`bPxtBzM8&d{YF`TmX
zVigvSD&+gso=YdP`xVT>lq71<$>Fe1A1D!Zoo6-EvkRFRXg5|@Q<F3m8$#wWT4}jm
zvrm_0!W5HBN!Ia6;O(%RJ4iJSnC%Ld!Qh#7rUPKX#oup;4m<89D683J@LoV|tw9zZ
z74-)=eZkbvU2Gg2Ev>DC-i-Wwd{MmDFzZFR!}sP|6?)H3r{k=I^PbjA6_m4YkO}0@
zN7^zMLT0zJlF!|20d<?-qeocPYKM9k9BA}?4)eeuCGLXI1lJRC%eMj;M`1beJ`93w
z0YS#W`wc(}{3M2c{xmc<zkKmxL}=(7fFN{#m0!NJFWgqcW7~$h3Zq^lbMp>_FZ_b2
zurS!aI4|I5mVteYx>rCT%*iyTuu#Nh@8|MzL*(5-lSPPLFn)sz+6kmrv=M7q$v$W@
zhv$jry54!;^Q0izVmEAJA_1@g_b-0;4kAyqO_7n2X2!<E92iGHDl-l=HA$Nud~8&z
z!Zd(XR`Ge@sscyUvyfIG#SiXcgIuxqv1BT&4S>=CXQ+h%55}#4EGbY7p?1cQ8geb9
zU_jQUrRNUxN+$?f9UU>d`wamE&${oyH-db8gzJ;smRX84ug#5mgP#R4F<q$hJ0VS6
zVHpPx*3{H=(DJr504AVzb;gIme@`swR!%g!i82jAQVIC6JK(C2_{gVDk{)29kM^xW
zPl}^7b4q@FW8H1>nI}EE+j%+^(SEs$^oo5x>r!o2eLf8|k|)KDfEd;#i0dx8@AWIO
zLc!vA`*tdgJzVxcv*i0o;k&eeQ*YcrjiLOWG7|(iV&GPnkE1a^JFAg$#V0tJ*f|pc
z_pbKictGin9<A(*N=WDhcFSG%ryOmvqT)EvKlm*w$;q8?f)@0aNA6`|RIx-9o2yV&
zRh?>1{sQC<9Bgwlv-IR-aEzbE$9sTWilf2KYw*NB%*_XwoS{UeHRd0X1H8d%6<Eze
z3j}Q@Du}AsmkCP=5ABYNyE+ied6H$^KWIP`>fU^rfs;&NfeEk%<7<;}^TSCc*3Mu?
zhYyc76QrFec^+#_j!n4vuBswny1@HXMIjpOZyXGzrxM@0x{97XJHf;>2qM^E87{8L
z(|_P3Nel%u8#~gJY3xcnu^t^T=FIVHW$yAPBZ8S##qUT_i})WT2}p5yBAfouer06^
zLmhiR2x7;_#=cgdP{TsA`qit*8Ag_?Xh5pBB8bH8deOxD#;X0vss7>NlYASBM=@Zb
z=D98j!Gam*HjFI8jt4MY(=aLR2WFr&l8I*>rOn_Z9-Bnl@BGPDVBWM;T+G$4r}Dxy
ztpEuyEea3Il-dJgB$T`i$GS$C_KUs(B^zZ9AzndSKX>R~mQry01zEHo1MaIRGE$%H
z3-=)CCx_qX9How5B5D<Git3MdNKFD|BjuPNIj|dfZ64J;I;dXtWJwTuYY-6ZQu2iO
z_!!RC2@E}v#rUOhNW$38bDt4J0Crl=7;pA=g-KBBO-e(e%+1BcJRY`sUs;`SPp+U*
zIwi+3Z}9wJfG~A@<uW}rxU5rzVv-QUh}5f-rDkWZqst7Ap-DCR;BS?t8jLn_u&_8I
zqYA=im#<(QOkb>MU?E&S!E0y3yjCW+wdUBpCi8hPMEJBZkUp2Po)4PnItI1?HXi8v
z0mU4?q^_qIe~c>YMf2!{<Ra#Ts&80MpN{lAwOGu_N3A#G6i{YX-Ac@D6KOHdxdDX%
zejkAbF?*x_u&7S!?r0|yAYK|vA-9N2#>UOfLSbPxa6FRhFy(cQy`Wa0YtNU$HNlm`
zB@Pwi()i>rdufHr@3hww|1K2s7ca_w%j{_$Eh)9>3?@fS$=6~$4oSps7}hb5Yps|&
zpya?nPUH%@Al7+lrYCQ^uWD#8+`cediwn<uVcVcj^Jx7M+2n1K{Idx381M6A|7LUZ
zEpR#k)ikxY!)t0UBK`jTCVpI(U*0J*Wa7G3b_N)p{&sUf+JrYDY?SbH0aFH=4F;R7
z=)h+b;}6or`F8B1lzh61evJq68Vm+dstzYXVRg@gidzMD38(`wQvhF}#9-cyXaN~O
zI3$WHDh&uPAcD*1ax187p_HuJl2Bs^McK&6g`JO3;{P5U6}I@w#Lb<H=L!m3fRC7I
zKk@X%Y}g*^d%Mvs7I5DNtn$6Y1+(mFqz?42Sm_Gg!wi-<@jf@lL=))89jC0^yu4eG
zn!SB{y7Wyi_&k_1fTq|CWE@j@@aEB;A<skT0kZOpDg6r;#8<rQiMT`v_~3IZ^;BHM
zy6e?16tR56&&;rmcLua5#E9&1c3BwNMT@&R#ybWFg2*N(M`TP);jXoNjJruj-vd}k
zKTkb2Itq|!alAb+e)$~05PhtRDNW0|1Y^eU=QX<NL>-_c%~m-g=m_38vX+TJb{P^9
zYV(-FTiras5QO<hQRYG^^*!i@`PKO=VIY%ot=mGMg(W<JfsF{607F2GMaDtYp&}#u
zEO8ir0+r$qZ00!(Ku2r`5Ef*1KqG)N5O9AXsbDSxh;$Q!DWqo}X<__@2R3V1JkP7?
zP8f|sfq)9c*~JBDxT%?0%l$NnouC~U{q@TQnIU*__0OM-tgNq$rx8PoAk{@SMIPn<
zcn-0g?+@!BVlo05zQFDUng&l+v2_)Qhp?4F=KT3J2$bp!cSDwD#8<+GdpQTUuhT2<
zaKfBm!DZ2s25Vn^5bivEL2E(1mzJJRY_y;8JpiJtAkGIs{icz}$gq%=;I4dWY-|K7
zj51DIIvI_wxcEAHQC#MvhhJ-ZpKE!)e}9&R#0g;*)16+6XLe1nT0n9q#Oy@_U?stv
zY7^DH=x-CKW~r&Wc%D09N91oXV^pk8@W(j!$r9_)puzuKoc8z~CJ)CzEuVCDV9bB>
z22+;PVg;!aNbB5XLib5M7TPX>*L*l;Pu+9Wg}{6BH_JJ;i`0I`7Akrgl6ozpFJQXf
z5@>859cg?T>FV^oukYHK;y(B$p3wgUB17kR(XQhIn3bF_q8=o<aX3G4+iMUQk9(zm
zrZ~gM2>Fg|2^B8E+VqcY{PU9T0<M6+K_@(^pV;)c&KW3Lbn1mi2??|!FZF&UH#3t~
zLt`tVT54QPI-@rxK&{6hx~Aan(2J3F^sQhGfTusv*&NMMzz<tsgr40pL};Q>eKm*`
zDS!P(HhFw>)EX06%X(<NBY6{vsiU)Omn>ZXlqsm{$a<TNi4tYPyPN@Dn4S>D24}^T
zn>%1TA{2MK+jP8=n*p4DU^{Es%DQ~usziVUb-d%x&$~M{y?|nsVrm{?NmFL@fz}LE
z9}Zgjw)XZwKR;3nFH_SrHBupB4L2TGf?B?EEwOo&)Z*;+zgK26=m3q7|I%AM9ehr{
zk_w~(T=0pf+k=}&6<pexNUJe^N<Arl9bR)Gf~<ct8GZZnUAuQ5?vq)>AM%AzFb$+@
zsfTRDg$dRIAC*n+_pe{_Dh=21^P*g^6Hxi(JDPZ~L|btE?{8po1>4cv)IPS*_6Uq9
zeH}!;C06F7>;A&&k)orNU-y?yCy2F`b+%wvGkDNs4$=f{pc<Ps89e^-nimgssmsdL
zi1a@s*LC6FisBM#)9Qy9m=i{O$j^_>;-Pk_M^EhX#REjLHB<*B=cGd(Kdwc6O!g~;
z51?Smld7iXW?-B0@rKE>uxLT4Mn=6qD>=CU+uP7%)FgOm!dwz@5-q^jZ{JSRp_cy*
zB<lUj7{`nPJd~kheg#$ZOtU5WKukmA6%|8^e8M$?cqh@*#@2nvcm%Mc_Vj2QBw-k!
z!t{|bde=Q=MMW$07LZ~oIj4RU784^xjRdKgTFJ>C?1T1ePy(VRJSA}A$&#h!<PSXz
zFY>+0fepNe5K*-DO=~&eT>rqpzBe~wt4!ViF2QaV6k`=F&Fo7-NUtVzx$P({acRC(
zR^n}HjoC)#<k&!hiDEK7F3!6q<N0%J_#DO^{@Jn^?pB3bxg?!bA7C!!8D~O?2h{9X
zwF{XOY!n+J0Qm~K^HYDv9337BufdX^vA>JU2@Ng67t4U&^UB|pxY95L@CUqgnhc|8
z4`N)uHJxcC2j-0H&U8dJgE(R8N?FP^&t(~%0=QxxbB?@d3Mmd0QrKq~l)po4FDwD;
z;ZIq~<TV1vj(|sx`jt7j&!d1y%76<7_M{JD&D){ERk*u!0aXQ|8#u1+;xOC2wd2Ho
zCeras+S-VV1;@ld?EyEJUsSZKu~GGma0fCO`a=RdWC_YVi16)>+u+pa!6Ffg_WK5i
z|Cp!Y87AMez`hiANfj0D;#40U6_p6)7QKRBJArLRa_M64Ei2!M{~1e6#&^&%aJ3;6
zVyZw4&F4eQaj2qs45R?zR~e_RxFv%r>A1P!0k}7K^%(G7v>uQB{jL9$)k8TUyt@Zw
zFOBsF2>~-=h0gx^nwl)|H|qLIOy(CBqWLs-nm`b7f6@xuO>4j2!6Cw3gHC@Cj01Fa
z=&+HW&|V-vX;2g-)nllIIuRcOq6rtktU#y01&EAnCm7RCtzfpM&@#%XLtrR}LT3Md
zV(N{i0Hn?+0qWZ^ZpBC-=!s$b!hks{4hyCOwT+GN1wnTyi1qdb1<02)0_H(-!rF!0
z*nrW7UIi!20Yek?Yvtuq;Jc!00Th5V!q1p~MqK<?vx~;Ti+Xy5m>Qx00-TMHkwhKQ
z)wKfB$Z`BN{}~0$57BIMVeuu76jbU!YZ_{6LFvVW6eFmKuc?pW$s?!O@l>~>ogx+U
zJ{LDPEM$Aw84?kJVY7B(H3OLpJoQW5mYx)-GmR39S4Cg%T~-5D_c=B_H}^e!070)`
zK{<$Xg^B`*00%oeo^2bf>Od<_a43d`m<0~qC$Qn+R3nT(W{#Rlknam#x}>QIf*;Xe
zqhmFJLjmexsL`QSg<%M6Qczj_h86@hl8%o+$n(wp$Ar};W&A+;zJk&izySgYh(H)h
zP~fhSlZS_ef!%?s0AH8~!Z-?YGi4^xI;f)T;oy!w2@=^dJZhL|;M`!ojgAHI5ej{1
zNsN~@f*x8!j|;3xRM2MzYEG=-I0$?Q`O^hZCw}GDEnbgSFf72%!xs(;q6UE#W9gn=
zdlek7!SPDQ$%Vup2?C;G*bTz_8>bxOPPkq&)aIUcsRFwUpDkNckrJl{t6$FJxVgKp
zA;m(i2mg%G<eZF*CWt88+5~xdVUP+Jm?TYOx$FX@FVusxuP+#Tpmv<fpSuRG0k)yf
zip}+zUcWB+RA36zc5ID=PI_g)7(v<8bQq=i_&CZ;-F8cww?@}dQ&3Vy#l?Y8Qw=dU
z>Wv7A)j&A;V~1qK;#-hoFmJ_9IAkuko+Gy+kRj4gLRSK=pk8HfZ-T@>!6bl`Jz+F~
z`8_@?F3>E_*wR!V|J#S-WOYR_ufZDywh8*OM@^cSBHj@+iSl0An9?$m9z-$2AK$1v
zwhScF-^Iy$@;9&&t8(w%=tJ-@*Hu-aqO*kR9&{pH(kR=f4<2mS3~arxy`~587<L)L
zCbJVeXfZPe!z?s9T3$7sa3@EffJzI-N*s2d0@q`c`&W|u5t3~&5A+<eK!w8%1;uj5
z^v1kzNdXE&)VHp7c6`f0txH81z~JR;>+4HlFM;F-9Vs|Jp!A^thm0ElqwCTnSFKAS
zB3fLJla!`t$J0HT4}f0^3hE&u;}eX!-v=N=Jl~w0tr$-7@>;;w16~k7by|fpfrV?8
z#wW7+&Cbu4fb)PLg8+}jhYgO%UT`?Tu|nw&E*<>3@IgRIVe55aMeEkW4;j9SrK<KW
z+gx2XOiK?Q7d^l8*(Tw8+4vONYPf{KacyjToQ9ejAXU>-*-U<wu<rXqqnL6cYeK*y
zB`pmF6OJWeR)ACx5P*gS@#^>3Sk|8_L&hUdlUOs4S70dAaCF&@VN0uy(ZPp1?^HHF
zlGmfY9RD?3^2kS(F%7U_I_fhoOr)$4M&kmG$uaRD^fgdoUQmiKBj8v(Mj_ImZf^L;
zs}ea=_~BaHrE4%Qquv^Qb_@mo{vMx`A+1Aar57ETn;Zn5I^94Y6=z%2I_UPyp?*5d
z3E`lD-U(C<pznFz+JP}<2QU6LDe^gDhKNPcSbK#os5R9^irAENN0#4Vv<ZW*{I~yX
z-(HH99>Fm!O6VLP^D{dl1ZilldA0gIe%$$%5{g$b#~1fRBx5C1M4f?|$y5$`a3bb_
zG6_REw6GUWxsi}idCb+Hj-!8t(V4nUgJ@6|Rl=<B!xxE(SZ(elSHFKj-O5M&au-VN
zU*V-xm~!apnTeKfe(CktDXTr?v5n4Eo;C5b%%czN49axj+s_?U?^rKC_g`E9ElJ7k
z>gwkxbWFIc`fgup@7?S8kNT}(@(S_@26k;S^*nmLmrYi3tiEhhbS{{6m=62QC%Q{`
z1=}}>FFCOo@GUpQ+p`|y6(#xn6*vFsYAx!l^aNM-ixtPBB@nWDu7U8h@ba1v!DvD(
z6uAbX+-j%U(*OhU)ay9m#LAD!h)en|sW-)T9=np@Hd#pUNo@+3dfnDDcZbEn*`>#!
zMY&KHsLeV`y)Rh_j7`DI9NZH_JplI;mJInG-esI7w`L`SLjpCwsnTQwnq;U&SuZZ!
zOgW5hpObJxK^I-9tMB?N6$!8se?x8ZtQ2-5&em|`4(3IF{Xsu5m1CCRa?<?u%BM<P
z+qhNzs^>73J_vWyedBD$0q1}}=r9?9oVI>%@2N-H;vWwGeA8{Sw>xPKZ@apt#!p-P
zLJZ^drCtDRcHI6DiI2Zu|82|rBHiKG-Q4TaiA|cS;^c3XHt{F)=9#<bpB@-deMj<v
z{s(SmqyAQjbi%y|jdXy4{Dmi=*m>cJ61mokB4%gLvk9J{)u|lZXkdE6J{4h4dLnyL
z<U_yEx^d#4Vo+IuOE2W4csc8U8fn(W?3FtHMsM*S#8z4S$vpO7kL>>0s97CN4z$yk
zi{!X8i3fozN6-G?l?#}l<GvBj97}IX!HGOO;LmS(ne+lRabqu|WJC3He4PD>8b~2^
zE?m8MSsI&n^luiz?jgjV#u`yf{u0$PATR*1Y50PJeLu<<k-kx}$#=w(j=b!2hpRU~
zZl4t2X_$l?VL<mLx%JrH_p_i&hS*mF7!k!HD<fT1HsP(;KKe){LOBEHY>{u;H_soz
zj+qNNpmQKQvwYJI%L0Z%ch|Ptn=5%Y$;0Qw<%Y=;$Z}NU><Vp%|Li$>V27bYKhGs_
zU*b-WjYF>6ZCI6_{6v&}r1&Ce2ihdln=ofO3EM-mJ;!&`{Wxp=CuhrtL=~Ye$A)3A
zG8C{P-2l`AkSpalJ4yfbc^KAL7X7oQ&S^i%f!mG+gw-d>9ZAD~J!y9_{oOG2Ki?QK
z(j&Qf(*^s-&#uq6T)wo0<d_63B&lC}Af4iYa%ZF1Mxu5efz0>z?$389kN+WJ13=cC
zzdySHt}@7}M{&tX`af{~XXE7E&;PEQJR}Fo2J!WO{YH(kiAhD#juShGX$-1s;;$Y0
zj?g|;y42gl1<sfykjEEeTS%&OaV7vB8R~6kPIZ5=m&7oc@Dn+;h2(@9-s7h$MA3w^
z9YwZ{<ip{Hq*H%i@&D$f|DWI7|NI1T4gYt&gf(A^@ag*>g~r8|Af%nyv4*+60T;6L
zcUXemuOv!d9NG%aV=xc4H6g}0dT_nIY1~<8Z-%u2a8P(`Rc_xPGJ-jc?XCJ9SdORD
ziul8N1ItCg$>6NqjWBiDzhDcttP%m~#Ew&#Xa0=IQV3@U8G|FI6W{YEc4s`gi|QJ)
zM2cE1w1qjKU*sv_F9j1PRB|8I-@Qg9S;42x+uMOB?cZ-(lPRw0A^ZasAoqX)@-7A<
z))@}GC}_ySXyq^q6=pA3Sm2k6fB8s(ctk(TVn^Dkcm4MmR19TM5h=v{eeq}ppZ@*S
z0;2#jdE4L=-|UMWz;~iBNms-OXOG7pEQA5RcgcSN?}iC!hvX1HBwgSzUr#*vHVL~A
zQ=ciJ27W=TC3tcb5W|o>OT5;jZJ?(h(rd$!z>s@4R@B%-`bKg{5xvlv$#(;LZuJ8R
zHo->>+xqv98d2yusM9}Tud(ETUA@;)Vqc%3HE?-ABdksIh)24>jCQx-E)*6LJ0c(D
z83l^J0S>WU^z(2M{tl57Ne;Pf_p7Ld7UJFQ>m(%44*d;RpL?-H=ZVD1D01oAK8(|9
zf!`_=F=2Diunlo<h&4p5XsF!}y!rcy<cSM*jBJ+b2@ZnihdW!C^{@wV@DQ<8ZRdl%
z+$Asn7klp>)^i%QeP@ha83qv|`xKH*q)23+T?r{diYO{6NhO0sC?Y8-%BB=0LP^n-
zHYkyj5~Y+<X)o>G&o%e+9Pj(@`^WpwdpvX8$9>N|W9s+&ey?k-bDis4=fb_d$FLmg
z8j`5DmIfbr<^TLQ)O8u^mEm3cP^{}H60aoG#EX3Py!dcs>{_%F1i~A7{4`UMKP7sc
z`k$9yHuWlR8R52a^g%w*f2pgF9EEumd0eh*UCdvK>_QwPzM}QtX=4;3Ko!iAXx|^q
z%m43{9Zd(k@$RcajJkI39l{PW#aw=YID214uDPb=N|R|}Rh0OCQ%8x@(r2>n*E-Ba
zZGq6)7QA$wmJKoz!}{D@b>-me8M#TZk`iBj>hg(sY6VJExbzOstkaNqR#Zmc;*4Fd
zVbhYz=S}@UiHS<a#HgEu^(@9s`Rk+|{#L_I{=Y~S|0f!AdnEo25xI{bGN0sYi7wN$
z6@n4X?cf|pz6061OBVIn%!m<3M*DjF(Z#;gwu_|@!~v*N?&xGi#a(p5uz{hQe%fvq
zZ(5!={pBh@>M{wvM<Z^DcvdfJ@xP5q1-#v-;`OUn{;F1+|EZk)tEBxz=|g(@93mhX
zZ4ks3Vb`T*5BA^xaTlW`CwD`^hHY^u^^Saj0dQNXvdhU~1&lRtY<-qG-_rv$q8GNM
zX$Lz!-rrh_%|vKY|9wHf!wfAeJd`7yhWgH)La75LJy(Uok>UdpJX(`>@ymsnnzCPv
zEb0Ftp_3?1#xPqsr=QRFWooE~s4|T#yytSl)xOMOeF%?08=m+`cR2CI-;TQl#f`WZ
zBKCIhN%6WqEvs7nQ|d%o{&yUck_YMwjtgYr;CcGgsY;X+)HyYA^%;EL&Gw({1U%|e
zCW?KE?YXCxX>0!m|3jxh_80ThCK*0RZLi@jzxXIV)wcADO0Bmv+Z8p(fSr7o96xl7
zk7CFh==`0Vz{RnFxFUR=uv>H?;Id`xcPg4WTaU|q<2SN9Z5L93Q;#kYdLCGP?nsnU
zb}Biy1XrtH6d|9eEV|wKOXBK6akpnR^d7FRqS+kFOm&pByIQt43%TGSK0DRFrX{q^
zNU8_IkjJpXO-%EatjYo^C3G}KaIv2-5%}_ABTL?`b$$l@DRFqgY5<k;(}uTkDR|X1
zf!S><)Or&w1mQ`ww>?~{XsbT*5Q=u3ST2+sfXn2Z8?rp-{tn^WjXOA6;&5B!OahVH
z^~$?Xs&DvgPIo-K(Lm}t>x9J*PIQ&h58u$Q(r}xQw(yA0?EZ{4<hzVsIxmKnN+ils
zSP;>|uaLWz$8p)$O2N24f6q9hHP9J|^WLGMvHbWT^m=ADiL*UK)N{K&VAZT^Fk-NQ
zCqK%{$dtXP&{23mc~g+@hC32AR4<l{aJGH(i#2U?ewPQ2i#1J|U+pTh`9P7Y%r>J%
zp@!(G8BA%(;*U|fbMo@eaB{^-sfj5IxcdJ5<hY-ST1n+lfS=$Y6{IK7pw!aRBKN@6
z1~!jEg6}`oWHAjyW*0WTdDlfkTIRpccAGk<dvZCpMWHO8nHmtrf)ZgGOZ?{iPh?FH
z4rov)_)K!?XbsAIJ?WmyTe5sybUg9}=z8wYXI%-~Cl03aQAD>G%IaIaG(-9kf(4}d
z$qu=P5`y$8EMOn^&!mTy%~Rr)u$nGCH5JWJ3hXqz-nvQWef)R|D*QKWMdEDcv1o3;
z$EKbv<l7DTh+-S!5mgy3`)e=Uo;Gb7hs8yM$4>g*Zd$`7!QZ5oe(I|$vsKN#3K1Og
zF7FsUc{p!;d;aH+rb5S>Qp;VScP<b#@4*}cGtQ)Zg)X4Uxhc<#e`a`fAJ6B%jOBKl
zh$Uy2t(!nI%rw_7*GNApjS;@j(|oAJ7Z1Pa=l2^Tb<*azUw0lA`b{@mmGL^xc$ND1
z>rmUNP58~s_0);xB$d8g;*i{da1RkkNsW5Y2Au1<$!wa!PE$PP7Bz_(b*?M*^1;IM
z57<iecDZsD>0a@c{~G>i0k%5^u+Z%^$Ui&QJmp&r3`*qr8G%~hf<#0pdHaFkZuG77
zn!mmreN=dGA`S1q>&r$>omv|b?BTw9%Y&V|A7Ur-UMs;!C8@9h%B+!<mK2wg<jx^b
zYcALa;eKiQR>GW!1px;uLfaDZ-hdZ0J8C`40vs)?-P9cepz6&M%EWaA%gn6%z}TK8
zjSAy4o$WO)R!^!PcD-9z&*aNoJ__aAmh3`ApE7jUA8^_+wG-E-n(ck`hu}?z*nRNU
z->inoAAO^NwIqJm3*B@~QCyvFJU>Lj&6_Z{xc!VUk&L|9L+-%Hx}TbDMSYF`g;2Kt
z@u8nu5v{wU<_by5NgQ0(c#f&q%_Pa*>;-cX>s_5~?e>@KpvJSpFhH(<!wHSXY`yt1
z2VPB0`O_$s*Eh^|t%5XcY$H$B2hKb&Ma(aBGaM~HYLqdjj&qonS%OOZF)^urL*ndw
zhq5a&jX%Tg#!j&L$paW<XKkuAoQ{oM37e@FT+<<6$gt#eNv#OXMB>1}9wi)*bdvy5
z)b@noHvj&mx^TF4p+i||bC#z(Llx!O4o+sY@q*-R7(6N4dzS+=NA5w0MNZdwx+S%;
zgAzaN>qQWJF)<>pe~t~w$nBHla7ovWwlYn(=s08}^b-}Ir)yo)QLz1SFD1iohKO<Q
zKk)5Nq@6xGKXcaC)2WN}`2&2qVEHe6AFS(^u79Rc(=Cb>6mY0Zs%~=#-{?Z_7$F>^
zad7l)=M<tyYo<%bbS`DVBa<F#r5E&AuxRtiE_dA>!x@?)F#aSowL=4u)Gr0r-uRZS
zx0)YNyX)4Kiz%0aJ?G-p8lzj|lqvSC6OMPXq|4zn5$6T(zoX74K{_g@YYUm~NUVF^
zxf54=amBD~d=c`L<adyEw0rYu35gbgEx(+die+M?-95p=Je-y`HD%ZUhgLzNYsWqA
zEDP*xE8Fg4G$?N_Lrmb}#Sb-!PlTk5Pix3)w8NH6{^R_)b9>Xq`INZ4uj?vlxb`}Q
zjc&fx)5aV!$DfVgiefA8I+k6rPrvxOGQv~PM3+?~XmfSD5(l$@IrpnEzm>Nqw0^1)
zb80bzR$hltj6?D{|0#FtGI^bwP`~f3pwV{TYaK1NH|pdfT0GTm%Xf+8T^&8BOvK)%
zIYwgMp%WrcnhK!70x07%@J_GVOQUl)%MQxrciR#J7tOI_`4@?Y#~y_MM2vZG>Csot
zKPtjq>VG=|R2i8}Es3Qlw2G@0C6kR|HTgP3by{bAU}x`-Yoh<2us-|)!p#sP-SfZ6
zKN<~h7P!v+#9?mc-F0uAX;mn?CM_{08n;b6)OPUq-V@eM*lm{_O10ARxv<Hf!m&1K
zK=A%gHZPS}{h8kB0P6AYeBNR&R@Uw5WYrzfS*O35Q>nN9k^QS#hPQE<e?N&YhyQ~^
z+JcM);uR=?wo-dt;W((1RwEZkI!`=kZeBt8b2ql)_E(SD2x+Q6n0Qt(wa|3ak95p(
zt;am$+;D>d4qN&5V_x=j)e=3gP1f?#l4w}OXnRISbK3a8w&sLR4VzaGM8h7uGbv(>
zRdZ+6h13JY)!RIGEjg)}aQZ5OG6l@Y)l4op@b%36=JlHOucm`v`>0!kKg0kSODlQ&
z15|`&fnHY6iZk9&nYaTf)I52}cxqiLD+MarzK-Zj{Td?P1V8f3@2~o&!B>u7&!{f0
za^GYe`WS@AteB>`KC?N?%db;A{au-y_s_roC@t=yuz9}h4!M1=re4$8$()RFm591g
zedEaO|Hh(K8iHk=q$|yRIW=Rsji(FA0bEqgt)Uc{rwV>yxM}h6!LIpkEVY<x<HW84
z?%aCvDhYF8lVxF?T1{z!RBZB2@)~3oK@e>)l<|z+ZF?Sl8l-XnpqDb-ep5y&C>tBR
z^Yf41TH)K@;1!9dzse!bGX(aIRZd%2H7gRp{Cs`TjbeptSM=&TA9Ara+h#3N%MXMm
zJH$uQU2=#ddTx6Ghid+Go=A$gCpL?LOcg}hSHWIEyQdx``Y*Y%kC*5WQKk1-X>t9&
zmL8cE8WO|0*Zt}yY25Z$B6;5FKJ#kRXD=2L%^Mt1nvnXnuDv03$mTRM8)BAe@Mf?^
zCvvUfYW1LrBNzYh^z_7tY<tNl6vu}MkYn<h0wZ=CJYKRJgP~70YP{cw$Hnf9T2EI}
zQmXjHv!FCwv|diS3r@Yk^l8sE7Qm#;uU`^B;wPYY;qVylNl1;UE!qTrTE!@1sZz6v
z?Ss)OJS8ql$qLQ%)#v9PwJd4hOE2C})8$afTUDOEKVU#-s;ftq_>B~zslAxR@B8?%
zc0ZE_uuV6kRUZ~pCP-t(Hge4cpE%0nDVI@LQ8$s-z2Aw#mr)irLqFm(#w?*tIhXXF
zY*$ETp!mgFoxe&wra_<&Q+tfio9rm0dZzl?+UbzZBK1T@$H8Xs*O(Lm3Pw2noL^)v
z2g5SoJ|K>E$iEj~i)DP-iPNS}XO+x3WK4N>qT>Us$g^@h_8qCPtX15(C#5j|rw!AF
z+ZXg}?9#^b7YKg2@>7+lO8$*g<()@C8SDK54iujhlz(__WrP^ppx#mX_hZ=F(EhA`
zK>+wpfaC(r!WbjRs%YqZ<1_rP+|7<vh*`FjBz7k7g$0Ju{VW+<bX?Yb@0pGeY-IXn
zuRAf^vWtX!#{X0=kXiv?wa6Fz@y|4JjL%RePAo3YB1}A~eIbi9SG+bbD`>zOqntpe
zy0jmgg1*I&92xfC`H>N$0=__tBGI&)u8+3zDD*przTfak9C~!7?<Pb-T!-NOn4y=2
zKNxV6FcXs?^ogXo;cB}nf2jHP>s%3V?U20ZNzM`dS~YVap$U5EwrAPFbj4c>hc}_j
zapA1%KEVz9H{ehD`l~o`zC=eOySHdPdA5Dl|8Zn}RbN*vF;BfLaJa-sFX*kq3o<Pt
zo?dLN{-mRjO{GMS#_g|(4c~%ji2&3io;H1?01D)J(;1veX1+Q-JKSKpXCbnUCTg(P
zj>m~qTbyc>9d>^-h~xKv6h;?MULL*Mj%z@s$wPEWOZN1=th7^XSUN<iU*tbWUUbs<
z*Go9s7W&yuk$qMEG)ksUu0ufn*~4jq{@nszy7Yr5-^X~)se6>rey7|_?JDQBo1JTW
zN)HKz;exg4qk%#i{ohN+SDa?>_qxX|t7jD*qjLXi$h5ILo$K39#O^#`2^~p^V5DP(
z3~ZqOCUIy6#UB2{fj5S}KZ;^2^)B&sp@u0x%wfggY}uI`i{!-`mTbE!I=w-6@9Fs=
zARm(Ljcg@w$dS1zJgVb!JkneHz4WAfS?#Ct-P5JbB)jlQb!;V#wTZ=wb&Ab$w_{N6
z-VsY8&0*36|MhItcStBM@PzckBiG$t>&vKxU=cSO9uh^=H8(lQigqlnpqLEb&LE)2
zH|A^voQUb}c<_CQ$&n-9Qy0b7zmt9V(5^^6v$O36R%41G69Y#~>bX%xb4j71Q@`ey
zE0-?4G{a(jw$i-&GS3r0&F<4&{_)@YVu)_OPy%N@8Z9B=?n7RXm^BuM&AOv~=gz7O
zzaP|uc|bzbM+?`>4oN5afh>0G)L_;k{u>IS^I57arLL$b>>-rCSKm(u<dJJ|yO4`n
z<M2;7{V`&5%tmHD&-2H<yr{hzIgcC<8Az}UN3W~xa-5HV6kX#==dzwpik@1ZN4PW-
zS#gvG2KGCPF02_~m-&9TvE!rey<(5WjW{1C(=UA^j|dEuMFEm}mmVu^L<Es(t^sjZ
zS;k~tus3B<)$bqoGqvyZmN;DS9|E{dpc*J0>?x~y_l<?Z7Ltc<J}MWdcPmw$>(a&>
ztHubxW?!&($^6VUP_U|n2ZDL04yPR|LT?F<fx|t}wr&`|S%e5l=L4cgvuID>boKk)
z?!4?LPKXu#_Uy*}L=N4@o+F2?4VrvsfTSy%($A&n&hPuH%W(1%w%xNTLV`y>A;;z+
z3xgz)w)~cuX0rMd)zF0<<DfqU4fyPuISo`;o0GdU*)>%iQzogghh#chR_L_8`I-p>
zi)lCRfSX5U*yQhd)a_}Hv%!EciWe3#nUDIWyqLnxIlD<zEaX!u^$Fz_+(*CGmqe#U
zMgcb%PZ7PWe&aP4b~9SUD?KXr6G$pL%zu|ri@&3!3*J#KN>GZ{<!TE?psC@)28&+T
zat?2j`+pH*9VsIFaT`YlBj!a#-@^I%z`)V@Wv?&G*uAb~LbCgvul{b(NS~;5T823A
zf(Di<{j;qW(DVaf6ZMowO(b*3Ua*S$TAwjn**w>6#Gy;AFCqA`V4PBs3f-k)?g%av
zl}DLb-GXcwADPDc(ws01@~p#}P2Avt|MLaNUT~Zv_+aWTIi6uX*f58M<Wa(lnXO;%
z5FqKiQ9FevPe0KH*QtRgdjvW3HoS+#NVosK(g(PKG3TR+cmoe6UR^m-c80lhv@^Sg
zMD6-U5z3nJA)R=qU%%af=7Jm<ld@%R@<5->2Y3CmN;eKQo+j@u#Ar1&AN7S?&Id|X
zK*a32Kj@u{=Ae?HbdDHDUvi}_>g8mSKK>mF@6kY8rH9718<C0@itKJ6dG-z3FNlWm
zS8Ug8Fk$pZFndP=N^ghVJxGNcT%D@UkPV1kG-SG6x&LO>?5YPGr8)oQvm{AZEeftw
z6V1&MhrM>}5mC|@!nrR6vaeZy@4G012=%GV_C$m(sZ|qK&uuAz%@7DGMi#)|rnRsM
z>4163#p~dHAEW(w<Q8#b9^teV-V9r`0e+SIBYE&{u?<arc8-mYdddl;bQk3|12hjQ
zsmK~a=?4LVg6))grYL`7x|U%q?gS1L3vX=N2CSYl6;s$=1LUDk#EN4W)gjQeI>+Zi
zTtY<MJgmR}>_O_(9=!+uofmNSY@~HjBDGZ}H42gxPQq-61kvrbyYXv|$h4y9%f1WK
zE9fGQ?d&gO#d*$N=Ur!=*lWD0)m1~7JO1+=N__%9z&y=SQOs813kPQ+UA{>|B3V)D
zYO4Gy|G|~xr%Uvi2q-`ETej?yiHBl?zu2SXZ%t-CH%9OzA!e&ru5>26DaVv}ks=W;
zlfQR4bRPO4sFD)Z0_z3Y&($pA7&Vv*7PNT=)}KJ&f3xIA=&gVW4Pyp<gQCc2J=j-F
zY3SBC_i@z%sqE=Do`DYWC9~5qXV=7qxNWj~zwX@85sp9Zg&g(NX!;O&4h_EU5*!KU
zmQNJ{wqDIaLHc*YyVX+7I>%1N#;$6b(7C9A+NPV)F61RyWtn>^rMq3}Dd8S8*k$A1
z5BEc$(%Y0_B-dm&kw6hUJE74OLPClv(YE;{XXo}Rgc*TfDEO8<WIN<9As6Xr7I!z-
z$bONBVAJJ7w<`|jwdv0MxnkKe4%h1J89AJKu37U7Su5#-bx;NLONPyS6l5lg*cOlG
zj+C;_vY&r@ojW&`0va$eGsjgzVu`vS3)_?hxpQ=VrALD>U;CLT)iQ*o8I7>F(3n%h
zosow5Y_eHaePflNT_fL{L>a+N(=&W&W}IxvXdgv&A$73S-rZ}~yv4W=zuf#eFT@k<
zY{_(yoa%s_k3XVD<klMII0Z%{Z$|E>^DG<;d=4#j;&-3dqpCRfRqw-mdJmVFHCWWn
zBMa!q;i+Z)%oq??4Kp>@XE*rON5I~EYT*}=i!$%)9}(;BOTGw27jtBqw<h+Y$eaEB
z`gzHN`M-qPj!Dj|8q@b-s(*#jPuMa2Q+>OTpeedjC;pn*w-+enNyY83(fa56-iINY
zo-u0-Dc-yWFu4AY8{XjnPJ-oQaQJ-3{68cvOcPJ-nDE~kV&=aW?Ep!9$#AI42I!Rg
zfJ%<h%gkfQ^hVH>G;Jj=YO*snwMkGhMlrKtpO6f~q-JI=Tb{@%W^FfH65P0$!=?l>
zS!K+57lDG(b828cYVofiU`NP!lA3RGDz)rpmN^kCHRI0bz_JQUMp2@EyzKAa_N;4P
z8Xmj&Qrx=J>N_AU#u;qgi&dpB|A287?rb^HURVm>yW3OZOQ(BYr4>!PI#O^lzOxUP
zeCnqXD2fsa6xSKw_I}+*{rS1Gqj@}q0MtV($K^9{6(nZGCE_;`<l2d(HXfqg$3+}t
zmKrnrF=&L5h(jfp&`*aLI6Op!F7B{yv!Vt+oGmsNo{TH+w~_rUZ`u~!<H<9;xP?P=
z%ikuQoo=lVOtJh7I-#Pa$BZ)Prj7{{<|kaJKwK|5fixd|+Mmza$^$MCr17>D`&Vdr
z!}lR_{lGsKX~CtVEpp?pdO`Gue7*|hy?EwbgZuIoUsBh7VsJOeW7EUk20G(}XsN6O
zo8PKOups7?ALt`Jxc%JDDzG}gPt+W=llFON05gK*;yD$mFQ_|){giL+YIuLgN*%Fi
zA=q;KWsbM6Z;NOn(M#ej<p0;*nXO9HCBySBCc)B<854YBT^ET9$ELMxvgkXqFHbok
zxX`$V#Jult$=i{nBo1x<k$3*=S!;#6m48Uwm^WFw<G;rKmp3^OVX3_}q$OYbPG54o
zkA#k`K<S5%>cYU+O%DgGWX^5foqZRMYoNs<uDgObxX+%K6I7rCR#R<C+ZDC$MlZ`*
zo(lmq<_Rp%6!_Zy;e8P!7@yMp@8{9e0<ViE8eovhwHdV=^llzV70!GMo4;6NM8Pbo
zx9u+%C(N!8uaD=oJZ6W);eSQpL(gwLg!g^XioTzb-Z=`@v>Xb&H19tptQX!Bq4Ak<
z=^52RAuJ9LlaQF@B!YlMp8>o@CJ}yO4@z5r&Jyk3fB)0l^azH_QBzb51$wp&X`?73
z^}IwFzk1cG$=?%c{;T{iorHC2tDue`(uCs2dayd(nALM?#WY?n@R*m&858vvvH>Km
z1Tk3c1XwA-$ib@KyDKfxXR3{EUFIj+Y!W>p{+EkT0B0mz3QTWfAOryX^@r$=@z=<+
z;yd%#|M>_1A3P6lPrEM9iXue!i4}?T_Ma7rso4L}JPi+L8m8I_a;OZN2}*2pNsg85
zW4z7wYEt9(cVz%O_y7KruY}CM+_Naws@Ev@keKyBgB!5T);;n6?`p0KCnz+-Uv=IW
zf;yjlNxp3N^zE!VIX>Woo$j+yHSQDdrnV>bmyv0va(@x4pfC>Tlc(7tz*V?~T7N1S
zjpq8*Fn4qvWh;XY$I;cE{A0D37&GFJ$JO0K52vsLKr6s(^+f^=0|TSiH|m^*cfBvj
z8L2^HN&$4oaE9;?rnz!v=!}Y-JNd*By;W-4X(fO+Uv~P_AtMU<5r7EfxPsKtaRfQC
zb4%DAaBE?sL~Ea>I7^@mq1XC9iJHgn7G*HGC;?s`PfjYJzsMr1SZIkxi6$yAVKogs
zL@2{R<K6s(O`(7*gMeLR5po}k4q*gdBqvjF-Jvxn2vw41rdpXlaNFmNUFl~UZWNgv
zlt3N$j|gA`ivdkrksHA|Y$YSXQj<e|-&p?&(A4<tvK7r%GnW|7?nYV@TRMO1hYqy1
z)?U3@%$wrzDSX)o*z^sr;}QuG%YQ5+N@f20I)1uoqQ07uvpsQ?U_J@j8J|RL+G1t*
zVJZ1tW!GDDMB)pOmC!Bl(cj~fK=Vxu6u-NbDR{1YNeEA>h9>?WB|LG!afDa}3|9C%
zlo|z;goJMU@;9R%4*bTCZQ(`nCF4oZd@3FdV%TwG$5NIQUq8R9;?om*9~Nv~RJ!44
zuKeUj6n(s{)F5g~E-f#&@N3r*_TMT>xEl<bcW$Nstf`)5zF;{YTONGxR5dV%v;Ny#
zV>AV{^tVbUrI;9HoD;o&<IcZE@DHy(kLheJ*$++dZLfPl#PNL(AO`-tdmM5u9Js+E
z*oUN&mR?a4*M?Z<ZTHy&Ij%W9fg+bQsQ$$tW{r0*`U4m^EZegCA;XnqGIn{*3Y@>6
zj~)*v-}Wj_AivBx(>3?44_#AwaigVpll<|s0I}<oBn~nLG@5uBN}J;z|9##EaZ80i
z8kDrJ&nRV%o#})Rakut%&E~CyBs1mkK;>oXt5?)Ws-9}!F2yCRB;yM~1-0uIzi8bb
zs2HLWu89l6or(gU3PjM{8OlNb!Q4~D7Xs7da+hMp<LhU=9OtvCze6gcL(9MU{YrrF
zsiB)iUG=S2+K|tM`h_mM()va56uRVd2hGgp@~p2QPtSF(d15lR*3+eAj$SvGWu`IJ
z978?+$4XA9SW&&?%1>pNmgRCifLJqkw=L+avG#$KQXE~CML8qFw#|}m+1thapTX^0
zf5Mb~o70oq;BKe3t>MzLl`H>xVmux;$#d8gg~gDaXjfLf(cRXtW~S4xpK}6s1XH=@
zaNFxHap>;<UD==O;KI$gVV#EGl)}zEpkS7QXLq@E2gS!fH#Jb=k(Ko&ef$M*+7I1d
zmLPf<GCMoCjgc!RJc=`E0B#*SBf2`L_Q$yHosv`z-6&vZN5V=#@fMmtq2Bh9Pp2px
zLuRxi*M+8FQtZX>NmCF05ao?==VfTayH}q(B$acP5FGVZ*EzlUH!SJp&dZOHR&-J$
zUbJyVAn_CwP;X$2H;G`z*W8_wLh+P?1kvf@P)XN5zq}Mz$hk}_86M0v*Zg6!4{R<j
zPuAo|XOqgM&qZf-`4Tw|E=@oc`<Jkxpo}wGE9@wlu%wdKGrI7Q+ueX=5#RT(p+5_#
z;%H-cyCT@$r!>sm*BdfUu$m<(rq$(BUQy0cEqLKOe03L0-hQMc-zJ%?ktD8C2nMEl
zdP@hxFbfLFb+*_r!T-GeG>z0Kn6M>AZeA1>_c+fn%A)Hj+LiD%kpOr!qTxhF;tGLX
z2u(6Z5kdlxVx87LzF-ZVN-?(a<OaNf%kOuAl2hMPOVtz52I(0|zlU<6#HnRapmbjw
zidB~UHBn7o*tsr7Eac*{b=FieP_$)#-ga1S!;|X=z6Veek>iPW{ry~CU79KmPmu=)
z;jO3+pKJ#+3j%3xcDEaX%XRaFHb}j*kf#6o$9bYO<++ZQS@dcw2CO6I&cQ?>YGUj3
zY}d<8P%n9r#tX8vC(aj1>rLP}dwTMso;#r?wv})Tg5Ix_pdT)L?9yIqvgQ+>LuUK&
zR5Ndp92Ex0*?<EHU*7dG{^T!q@j76rle=l(>%A$%^--U2vgPIF`wX74m12%uvnZqL
zB2827ZnLLm$KWR)i~g?9*=XguXS);Z;-K=o`32(C5q*k7P6fp_yj@PuQ}5~mij=9i
z<{XJ1zImsZ-Gjo@zq)|NO+?H$KbVz^I?HrLvx9J{5Hs8!LeeP7**hYxE*-&5iKPR5
zx#d-@J(Gp(-OUqbinpk4znnr$kp1Pogp|n3QFm1l^>M?^%a($Sp~uM)iX}0K;;4Lj
z7TtwBQMcubVv{4Y=pl1VdpJ~-vdn}EvX|jBSvC3T)9W=mTUPgr2%3kcK$PQR5g6yD
zkzfJ?Eyh0@ii$}J0zaRtqGKSEuc;$2n!V>gK@?<rdzf65EEd_$kK|VeY)T4Cy+s+@
z!B6toZACu@_NomAp1rPPl$7>RhKWRg%I6U$i%1|Y(~p~xT0HxCA5wzoY}zbb^sk_C
zR%yTGZlV+Nr`Df;lutjKu90Yrw!Y-o;?o;qT~q$7pF-VHJ^f947hN+G(k5CXTdp5b
z=p#InVDU{`Y10d&LETEmi-Vs(xAA5tY0x-umBJ|g3+M7{Zll?RRns9DR-(0<ngji|
zbIQ+%?k?}w>3eIr=0*8jIq9@7goKHbIO_@F95(I;4i{hufDc>HcY_QhADD>tF7chG
zx8}lVlmrJ~3^6OQ%p;Wp5!PK$8zm3LZvE>3EK#DN5KITHcuY#u3U8g?OAV%@5FU`1
zI)_(>H_1+JFk_>Y5Q5LU$vAO5<|`J`fFCUzT-f;@w1vl<P}p)l5SwYD`Y5Dz1s2w@
zF%9i@liK(Y^dtO2F0`T%S~b!M$B_BegFGti8!kBZJzfyBxsJOj9W!OQ=p}PAnCu~y
z=#x9`X$|uNod|J@<u<FcGtghO0Tw;%3t7h}6E7a79g=s#+7KNc0S7{epG2dAaV$F2
zUV$y73kFZUT8Cdaa#uR=4!H;bwCckvhe@~gT@s7nIY<HRHPASip|l)sZCD?+eh(!o
zz~^;`4Kv7sW=mV&%gvRuK5e(=ak{H);d`R6Pqf;3bCV5?q}OG;zP!sEFd<#6q@}wn
z^aE6y{|P{3n*f)T?fP#At^&8Hsi}o?NIf_t%+Bh;l(TdN`@pGudw-*0HuwmOX)UGi
zlJ_ADOnAe6E$Gm6@EYMZKY+t{c59g<gS!Qp#4=7M)cyGMHi9-z;zW|!1{i9FG6|^P
zUbK1UF|BI|yQhSYtIHfT>(|Mwf5v=do&5|FD^E$3tNb7_rVkZ3YS00CT)(MySS4@(
z`X3%g7{Aa*d{kib5|p|;=*Qa&7cVv;rsy+d+I7$^DD}3^H7!tglx9c=-ClObN6))I
zu(nGlS1$mX1Mw9ugFM(fPzjuixm6F|S;A(tMTA`n-VkpAGN;`#n9M|^=G|^UOaOWj
zM)j**e(PnoZeE^EGB-m_lXEHMFRzw{)G?k|U_3o>zGNnhj0*}24(w7Ujr;tG?&lqE
zljJGpmmR`OudD`aI&jfekr2^y15jgVsen#}2%tXUVI%Qv3GMN$MKYrE1GyGKg9N5*
z<Aa@h3Aqqx=Uq4J#ws9aRe{qGY`d7oYjd0?Ex9g4$C#9(+q}|{ri$rTzUvs-)`sr`
znsRjsP6(9(GygrMFZv-uI*Td(aaqd~2aT@jl#}v0V8BM?1#~XQdaC_iikxiD;KtXt
z);uoJQBWbC3#9S_e2WEwdU@Ym;<QNc2@GZ&`sTtnvT;!_a_J~@k!+XF;#@>t{Nemw
zQZ@hDxaL+>XkMd?J>ML{h{Dy29*OsG)T#a?FHD1G1p|}j00_~r<uMD!mrymk_n`kq
z?=+eV&bRL{MJg=re{O^em{|~os3sNPc<x0$Ald;0mtjOxzL`0Tbm5alY;`J(@eK%=
z%bz5=XNuoKL`PP1>Fn7z^obd9i|4$$_LMwT`s5)XShPqH4Y4psU<Tm62wP|_HF)L(
z-HAJwEoy*^J{IjyVsY(L7X{9$I=4hlF}?S>0l>^Ap**JCzCWS`oiG!36_gUO1lrFB
zEANax6sbIMVx?$dFtWh$V?@!f-K&uw_^LB~6!L8F%TqA=A#%?MG&aPs2;d_RMzkkk
z-JeqP-DtB#<SN&=zG0(f%((CyU|pek+3=MDQ$$R!3DupxGxr#pBIqf+bpO*Lv@EyJ
zP^Q|Lp0u>~BJn8Iy&EVLAFw>xGE4KQ`I<SvC1rZUQC_DyFF(Xt6Y{b|Z+s&^N7@;Q
z6Gs3VXgTwQ?_f(S$ze0O`Il1aUom2~Joa$Gh^kKRUyw&y`x2|B!Yw#C^|>utHhiP`
zx1=r$z7fdfp}rcpDa4j@Zl)VdxyGY&*kV#GbILe(u#AlT)$y%53epGp)OB+gUB{W>
z(ax;Vo@)0W9l-i3zsralL36%{n@O0W&iY^xF$K~N$nE&1ck(Zzms3&B5Y)<a*YN6@
zhw%<{i`}rW@Ld)HvaWIiA52?pNpIHVgRa`dKg#w$8}Sw7G+uM!D+;o&6b@-70dRN|
z>q3YJ>{KFp+~%?A-r9rEko#DF*BXWCd^I}{`T;_iPgt1s#bo2}pu5lOG&}1x&y5M5
zi=?Qv>f9K?q^6!jZ?H6N%gXqLO3Kob%$BI`E+NT}al3*-_1wPWpIR<eanKZ9U!Hu`
zfZ{|f0o>;6M1YUcOhKBN07P3pIc99#Hh}`32^x4%i+`ICWh@VE$r8ob=ihMa7t0+)
z&&^+oBK-%Ez`xT3-q(*6x_TAl9vL<H^~$|%`ov&T&BQtKBO`{DLYA=3RNd3cwz_jJ
zHU}B4{aIvoMZ<>qo0$Bgb9eR^3eTFBF|6lA2tTc6e9Pzd5lz)|rrZp}ePe8t;gO|U
zA=mF^ui2<OD!{P+vA->L^<1R1PA;fnzLL>y-RLmG0Utl!&iQcigT)qeiw8R&EM0nJ
zca-)amsd40pB=tUp6)xNxc#>UONe6veP{cs)yI@Mt*ZLS8$9RqI48^&j)Xpv_RKW|
zUf#FDLNDwk`0BU;DJ~TMkZNe6%C_70<(E=H;Do^5n@vlR;9*FXo9EFBo#(oV$H{z(
znk7V7$%xyA2xQ-X>b}w|&@J5xDs=>Vqt{e#mmyIX5j+6k7c7?8ltdl0J5PvwS-PO=
z#}6Rioi_70{91V{ctWK2%L&mn;Vh@G^nC;2ujKLk-8(&-GKy|@do}@u#T>m5?u-A)
zp4tGf1!gb@mjeT>sz;8~R|lG$UN&^~vSnFAcS2k2NPF3p50QK8IiJo#!x~#H`CjW>
zyv9tQG_339U8fiK_^F<rkpWDk2O(R!AAjJp!`6>Py)9heVrsJ;92`X1MoY^5*NN%N
z@KF`%$MoUavzIToQqH`*aqrK?@n2{cunG~s>u&B{lTf_T9a^5Y!gtVl|7hydM;5uO
zrhSs1b*SxeV4zr;Gp@#&M(m!Bmt&K=BwO@na`j{6%?IWS02a;g;i^+6d28QTvwg=7
zu_6Mz&i}lR*@rU&p)5P0x0KxlE+(C3L(5mWyrGTJ=Hm?h-lQ~bS`?)a=yKiMMm{*z
z*&d2@baGMy1G+J&We}%ZSN0Uk#sey!flSdC9n&*2x!6yB716AiHhP7gms2`hFN)2|
zii$=}-YU44JEecFo#0(ESoO|O`ES<nc6+Db3FDP{FA_2xUxgW!d+{87z#r|dy5X8u
zw^2PNOsPx{S#<21k1iN3No{U7+$srFgE}q@J8d8>qz@k=>-Nsx_P@0N#x@JC(IEd`
zQ{#dIo79r+ahSxB6Yy_Y+p=Kp%$W~qh^dbkYU42e&gQAdv17-0FiSH|>s|&eKyP^W
zum`L(fV48-n!gB)2C4xY=<;kau3L$E6z#PzA!+I6%p#R*CVEfy?{Q9Czn^yw$X{Y%
zLhk9kQ#;J09*0qS>FIACcO(GRpcTp?KyuZrMyjca#lcODjY>b#&*SyrL}#DrHYEqE
z&6^7$RfNGt^G|*+wm?{)r9FtgNh!xCYQ_!t;j5^>yy^|;I8BVXU($q7_|h!-e(gsu
z6kPK13Tm64J{B%Xp>SuLq=Gpy?c|P57cM56ynm+t<?eDRuFt3L($lYJb&2zM;W6Bi
zeL$?$rCsSymBP!VPX*^TSk=;h%Di1rkmFbRzsbu+*?p+9-G1Z7e6(<ZET>`qNlB@S
zNftkoE5JV$EL5k@9WfoZ#u3v@PyC#H8$t*7lZtNUsOaS89EgNqr2DhN^6!g-^qzhJ
zCtl)va|MCjQTOjNF3RR@K$w&9iIux||3F`5`@*EnpcPR&1M;CA!!50>G-;SH+s^8?
zS>wl#S5*9t`cUi2F?u~5zl}TkHvoqpJS;buZ196zo!vHl_Uu+XnWls*dTV=gRp28|
zoaos;qK;fZUAp4(%$1#P7=gA^3+WYtb31&oNGvhp$;$F2I;NMik()J!-VwnY3b`Rk
zvg5*LK8cCJDBaNFD>r4fT?Kvpd;DBT{VSR-@qB6SboNK3v16}u=5D&@E%jO*gEGzQ
z<uO)n7+EZ*ql4WG83m+y!PnOTq)K1g-A6hMx_C#^yA@*xjN|Te-$0u<2yfvPa7+nn
zn*hjU69G1Vo0;<J!eaIz-8o|ryn1C@FPhLMrk585esjSW&&7DeUEeM$S5NqND<IK=
zD<^mhK@<io!~}a}aP)#j`l0%|-7HwMK7D5Hd8#b6f$E3;Z}uY%4GlCvj?_}fMB?J&
z_==l3=HNK>Tk)Y{3n7zed`Jdke+jY~VGE1KcoPuBl0H5R#M^qLVUsnSt?2^i@^INY
zK;t?oB?WJtX2e4}ZrC6Hr0KnDko-l{HUme@AbA>G|LYy~3^5q8>Xc{v?(JK+2Q~(v
zE*jtPn^pG-E#G9XM0nMD$Z6b;US!np;_yxKW{_Q}^61e)k<0e1Wc*07cW-RSayv%>
zDlDiYekC6E=Q^>jm-nMkGZTBeG@E@02`sRcdb}@0FvJiVZ(LlCE<3d3<-l;C0=W?*
zFjHJGz^yT%q3@c>Hc}<OlQrA~e;BG)h9Z*=z-=6DV-43Lexgw@n~IaqTj_8^$O!!w
zT;-2GwEezjOz=Q^S+r2=dP}+%HEw%-?WctMDgC=4r?hX_*xUQ~7u=4~`bb=eSgnVD
zCluen8qCDU^!~%n+M2}{18YpYV)Bbq&nS-#@|C}Y*>-^?r5|Th?CSe9dBA_^Lhu_{
zwrm>Ax(PoMAsf0qUCt!qBjQZgs8Im{0gQ>9J#7v^Ok7>dB+yCwH%_02YMxtflbp(?
zxGQ}xOsa^YRYgg-)AC+)G=!bs&k(JFb}jh4CeN<iMS))2ecY+oatM^~Xgm!6dXbc*
zkXh^VmEB&JuML@y8E1Bz^25lc^T&_fdmSvUFaq_Il_^r1Gr9p82lfb^a;1gwvaIn-
zZiZ68Nd{`lGk~Q~n?*r3N-bj3>ztg2C<rAR!?USbuC;`K`xYGdJakj`4Sd7rEX_fO
z{Kviz@59$+i_NHs_r@kzUeIvq*mPf>`Ra=~eR3)W?|lEyE;8!lY3zAg&%1YZ#>!>E
z-r4es6{G*I=!JFX1l1d!IP=(9z#*(!y7X+m5-9<?(JT7XgUvUfF;3buqnkBplQqFT
zE!c_c#<PBzk#SMVg!7UZ?nf?IUTt#w{(2o9VV7zg2!*<=d(*F4lWSI6&CJFh(%ZFO
zkIA-+1$yrSXWht0i+07+Yf?pVW#to#+{t&&7;3C;KNZWO^+>qO;wz0=A-ZE~^IxQv
zDiCZH`n)|{>0}O~S9oPG?O4e;Yj5pw21zR{|0p0@z8YEr2QzMNIrt*kG2Lh#i^6Ui
zjJYsO`7>n^Lw4GmyXvq%%<G4d1!g_(hHYc@==imv({6^GD)C>t{NmAxS3{&uzsT=)
zxL2IbE?ZgI%Y=fL(ef8w?YFgkm6oP!seR)^E|ZxzBJ)r58<R9!I#1;{b#V5|y!`xA
zE?(dNJ*ND4E^iy`8)SKlBa>b`*dgb4bB%J`EJ|YC6XtY269W)bc;-B1amw=bg>P<M
zk1`!9jGv3uJ^iAuHLeZ@nSAzI{trSZkz+Wnz@~Uo1VXT}7JmBJt*NT&hplu~`Het?
zk~F~ETj&%aGB?<;F`>iYeso)#Q`*pG@X!`y)Kua4G2t{w#EdY=-Y|uAq(?I@eP5Qb
zn&lzue?0=4Y+>2y&xGJ`9y)bpzra2d-A|tESC)h=A+nJhZ3=h77TU131U?rhy`8?d
zf-|+cdBr_t-A}uH%nkzfb>C|4#`_f=L(ebv@n3$ZEMxfBEt1X~CZ4(5vJLaq&#8lw
z0^cqgb!PJo?SV02+sUPgc31k3(-x-g7H&cQi&;{r|7ge0vWMTovI)Kn&udrTZ0UA;
z%Aon<Uk5pO<u<Og%YirRJL<de=XX6z>imKNxv-vECo49$WLE>^@6Pt4a91nJf=k|(
zcN^-w>N!e|V?8PY7yDc`dqe5RafalkS15<0spXIWvfiEU1K*Yx78Zg{glR9+kkJ-C
zrux^pNQQrz8)*!*tuAQ6jRv6+#Tul_<&p<unF*2V-S4c^=;Gu9lt`*E|Mc<dx^2bM
zy7r&cjpla8I0sn1jUCg5(8wgGXQM&)c^QmamH0VRF>?KH#njVKQYx@mG1Mq(Yc^Ex
zil2t%XH?WrFVvNhhzJ&d6f|312w8jE%GAt^I=Lo&C5h#v)R>;^`krt>k9zdz`mCa$
zNg5hP&i@t`Zl2b|P(j}5dFcfh1MI&m^3}s$58amo`{k<sjnk~ZdCET2!4>+-s;c&6
zvDkoPx=Hm8q)>!+rq)Msuq7<!7KnHUnwNQa1;a!4&n*KdtQ98uz|0B?;4fdkeEm8d
zS4TE@*M9INE-P*cd8cCX;xvIHCrm#Rc%G7CMn*(K*sn%@g`qy9<AIN7E~n0cpQ}X%
z`uo?f5Bd2Z2G`@qNvk<n!&zj~Eb7^*yxx{;LJ#-kBjrtxb@r`WMi`nYMtg|sOt*y;
z93tHd@yp)gg{;0bF`03W{DmM%=0rt3na^-vT;v9r4Jgo$So@VW0TQ0>%nCw+Ga;hO
zPAx*!^dJ9#{%i2nD@p1E1VfsbU)+0Xit!O+Z`b6p%FmYO3{sehhnY&Q{@JjSuyAgD
zVpWiVOevMsg%v6xYbT5!e~S_*#3}*u8Od4X)=e+^&lRwFkCO#==$vc~I9oYebL!2e
z-0^`<qmNC9Ietx^yd$*y_4Vr}ct3smwC;#1_q^X^-A-E4#jCfz6M^aDXkW+ofd{f9
zH-)NpL!jzqS3yX71jP$l@@D>wS!p2psgozq{g80OTiqd?bnq1?B?c>4LBLb|0fIMm
zse?Q)H0qZ&`8jA1WzyH0ni^V}<>enB-syg=@}0KG;t_f1>r4Ccu`vq2!L024;*NRZ
z<PRc&r=zz#-*W0vHhbJ(tgv=xwrx?O>(#4QDb}GBxJkp*Y_5~SY_>Rk^LX>c@~HL=
zKj1k1iE%Z48jcUY#g*{Z+&_;!zSq`Px;fO{{!y3YzPoHW;B3lqb)1oV^<B%;!iX)J
zuGht)t0^{WjLe33>#iRezO%VFS?X)<p{YeucjpbRJAdKA^*ArDsb)&*>V_NkuP|#_
zs;{35Tnw{1CAGq0Rmj>yCmTjh{`-4%_1!7KdLxSKWV*L~EO=rcao{TiT+$f(KmPcG
z;&gnWL8kpy%CMyOCS$j3xD%SB6}o999XT1F&bw=Fq{A@ZjGj7U#{0uzQa9}w&SIaz
z-g_^pHA<s|M{rYfGwSwII#f5EPH^_^Q1pogpi@iKy&}Ivt_|CY#&2$6FP#3SyiIy=
zc?4>@dWta~AvYTu8pw(3A~riyb-y?sHo~>h%8s{wnX`AGl@~B0V$#fAZx?PFS-90p
zrh*u^^>)*#t8~J4%G>sS`NDjz_T8Yn1J!2DQcxb9d*5tzkz)CM_Z3;A8up+#4BgjF
zcRQghvH8g*y!WiVKV9-T6CC^%6dx9U$F9qUQ|HX8I1ClC+;sI^l|eZhS7~dzpE>hB
z4v>9ND`j!z<gWC$Pj=2fJ!HhyHKA>7Z4aUzF~~4+_mXD5MzMJYY-In4P!lV0@@;Cv
zLwVgNWG-A(v9I2Q8}@6Zm3}IWl^)NrWdDL+*e5psZD>2b*WO;_m|uVMttW1YkIQ)h
zEUO$4FctCmNVPS>;DJkRgApZ8rBk*ZE|?HM!8<**bK4c+PZ-nx6}1&yaS>P_>`OmH
zQF@-L{jnFevTG;2r(=?Ie>p{HCoxLOCHC=S$o#^{bT_Z0&cx&=`G<?$RMZdW>hik>
zR=*#%=k~t8=L1~4?SIzS53t<0^~w;vUOjvMy*Lu@aw~|D7pHGc;mn!ME-qT`cUsLt
zYi2=4#;U0PCkBl&2rsJ{cNFXAU3<=jpIf6qAJk+0(BGVQ0?xYxSGa;@cR!Ewo$Ql;
zx2$3VC7E<{-MCwM=1*TIO}rGkc0$axbibOvN(W8a0Uf|C9+1$5X6#g|JJ%j^&_2-S
zu96*8$HM!;-F@0vop&*g&6a#zk%*(85(20l@h-OM>%{sInwvQ2-RV@$N5LvSJ{_gw
zZegmffx`R#F&G=L#|eb1j~<k+*GlAlkhAnL50FyYRdc!bglKl)x#sn99oI5SPg>sW
z(rRACjMaLJ7a#9(W@r3DE$7og!4P@rx*+RbTU;V^QWoD>!<pB0{)#?`0I$aM>GPF}
z_Pf^(b?!gLEm*Qj!G1SS4dfDQUlZ_qKcCU|`1O`#^PKoIzZ_R?caL9r;K1V?w<_iU
zvX9m_j%|5LalJb2x-}H#F^f@t4d3%&nwJi0eVg}4SHJr0$7>KQUY|X&R4&#s0~pL~
zUTUtatbE7)l<T;Rj>{LZ@Fag8n{mGP*6N-EWB(G!hu_Lo+-GAW96@l|uPyawiTND^
zDda{hLGyFI&d^l{eQy=xSH&*0LHEq7T`+X`v&y9k9*WEA^O-f&i&0EVHdE<UI)Cki
zbxuxka#;g=Q7@3^bU!0bLJSb@=C*{wjE!=3OhhDl1WbP~b%=TZ%UK?ABc&RBn<JRl
zpXvU3TAy?y7n2OXDZhoII}(=to$HDdc6TlxE;@bLt(P6Y#Xq)o=MUGurb(w4c7;N`
zJn+bcVO#5<(^H?0Q@JJIlT=>%t=`-`E6DXga7l0bJM&Eio;)yu=OL@;Ww*-p&ts)j
znpk;SA})IKQS2q5_L|rhuf23>dEWh7vw964I`sD4N<VNTG9u`Wo$b}>lrH&?B4U)K
z6dx1FDBcdN`~+K^oRoATthy%&i-yaX_`C0DkpU>@I{#LIXP7+M5<{5c-Py<s?sJaL
zCq>M=tF@FT!emYkz7cG1k&*EsU%>I4>^#YjnUT_$#6e<NU4Pz~T%xYoF1cu$+4XwI
zdT+ME1sixJN1i+E=V`+#;rScnZ)9Fx`0L`@&aao$MxV!{`N-M#>TF1(Lk+lMvZVty
zBSdEHn*_yDk2BiAR>5r}xWK1)0N|%=Z1Uf|yN_RtR_FYc3Kly|Qx=bN$x*r0_=@Li
zma=&ItsR_6Q>BGpTW2Dg&6?0tjGFx=JG1=r5!26$qq+uL`^+6<eOYDK=oTltrafb4
z&tzYs*Kii&$nH(M3e2u`Qa&%sO(VhdKJ>WTTuq#mVY<xjL_0N_aTX;{VN%P)T@rTv
zg@rf+R}>W$VU%&>z#{^nMbo{M(?f#up3?yeH2vp*$qC0k|98AP*O44=*KzZ$fy}!4
z=7&?e{g`CMxxh46tGf~pUtRqmO;iXGQa*;R5cPLT{;1vdBH;_g0uq~gR4N(>nE5@z
z>mLqnpikuAOFg#sAG|L|#bs)I_Sw!YiMBiL%=}CCL5ODUsoh@LiwVLH234>!?V;&1
z9bv~*d`2>_N80FYg1PXRa+s8MFjU>e%529D_kC{&8Y)U#`A~v&-15F0_5vE_LSQRW
zg^7pOrGQ$n>9sS`XA=lt9XZy9IU9aJ%gRscy+#3)$j9p^(*>83YM7o27T+MNnqe6Y
zWm$uR$3QJ72SO@8Ki|>O(aqI$IGEk3+ct)@e{af|dc0uCJDOX!?WX?w`t?d1qgk)B
zqQSIJDE5Dv(YU8r+2J>@bGgl(DJI>kpY~VLeBL`G_ne>KgtkiDn|DrDHC$M(_p^Oq
zRX(d9kf$!4E;UR><`JJ18sI5Xa5_pX*SMO|zr<Yr`?5nYk#Oc5e~%L<X6^g70%am$
zvY=z}{@rWq%vRnvVU16OOUmNy-{Q}d@HpdNgYfS(E}H=f*seci+Op4!11CQy@bSxt
zKNHZtZ~MjX`A}@m)Z#YFZEvrg0EI}A-Aiwo$`FI1C97e71vRlzY5g5KRdn$}4?Kw(
zbD=fGDkt1`))yE(cMC&q;@as}azlnF7Bn?A39@OW^!Ltw&BNTCL0Y{*T2tH)9P%Hu
zPdVOnF8<Q!Gh_LJ%=AZh*P{_ohXNjX*U#EoV%Dl_I}(R};pJHr#<fs)G`O6wS^C0%
zhHJso=X?B)Jzf|QdZ=I(+@M9lJHFZ;r;5CivEvU`PWraft3Js8(-b$bG(c&u(gAO6
z(ft;rPurQNc9rYzDDlg^jwZ6JXiR{2oJmTbePtJtMRy!iyxOp;%1TGpBczp1o^diT
z3J?i{^efsNp$3fAgPEu!F|)XLc9e_<?z|=Y)9U(p$<DAyHGldDu&JolQ`e^V4r}Yk
z$fJ=VQI|@}4lf&IH7Q*sY{X!lw<WaH6GKB$ENJ~T+9IoEVCX^b)|oVE0Z2yiPB9S?
z5joSb(1!CHL75`R0{+kC*<QR0L1%Vm75vW&e*MMcjMioOQ7y*Xyy8~^gN~{Bn2RRD
zk{555+6>&iVnc}GKgqvm(hU{^zG5m2u@JoZ<Jj1bd3m(4pXcV<lb{8?2m_XcElWWG
zf3w54CoOeRn9(eC#>IgT$1;T=3Isfw8QUPYi`zgDUDh#zHOqc8rZi?RtKFX_r$qx?
zK+D3(`p&FdXoZy=U@tpXIZaUVAB>snlbm{rpc0>dc-wImb&0`?%8psSUiSEfbn1Pl
z7ut6?=rm+ktym$P4B;~9GdzupJLX8kjj57-X|*ypy?lIRh7O%@`UD;yj|gM5ndov|
zf&7HbGCq63^e@Y0oDt)at3E#Z^NlY#>Uhp_yL91#sn^6|?1C_?G)Mf4VZ+kEltKId
zyjT)u3S${>aE~%YCH~EA*Xi07Q4?4&)9-U5Vg{50Ys%h6!POOSKM=Mq?55g!-K|zu
z%A2=M)X?~pcw;vMsN}B+a&CjArKJZ92+llPNc23Vn!fU1pG_@aXlx!_Xg+JNo#Bi$
zuM{uY*H8U?FCIJgXYJxExzjh>q2}msnUa(w-Bje)?c9vZ(+>NpcmKWiw!@9qnZOGi
zCT2m`GgUz6S&Y>%pt5W8s@eQA5ZG(OSkd?8dG{+GGRyi4#RN)bzq4nrZho=8d`_y;
zxS-@q_iwhe05J?g;$J7AB5)4eSA@U)i7+39NwW+Tn3;Whuo7pO<(WaQ^5659&fGRZ
z{Zyat|9UxJh77@CZ1%Kib<k*ViMZtxfZ@&m>f4v!^Bb;Mtf#-@R=%k`wS=LL7}VLc
zIB?pcRzt{8fb{?epklR6tyVOD69x=0SYWevZ@&oDFt06k!8@D&pJ5>*`_8w+^0lw#
z7MWenxHj*%bo7)FgdoW9GrVG;0cTt2q3MXbp8SV56azBw?c@CGr}4GP&yn81dprWK
zwzRg=Hv7b8#(MjRj{5X7Y8IfR3Ry;x71oT2O0GACbsKwrQ+GNiLhdl*k<gR>o}*(k
zw~oY7a%his{5e(*yjK6IJ?uKm3M?(}b~qg4T>2@`QNo^pTCM->Qc3L#3imjrw2ibg
z7fagdq}^LJNcNVX8j%E29iMc0E2f_brw2FpR9yRPy+gX!%Qs-|v{xP_i(TLT{5H}c
zKlg#fxq<VkeiHM)5(E(=W_R8rxg{PMLL&g-L9ugq;nX)}eh+>s8N@p|+-Vu+=Q5}^
z{!CaHAk>Jr;ml3sd4m#qo)Y6^IC~d5xec1U2#seGg#hr6B@KJvX85kSpU+Tst;l;c
z^;>wa;j-^Ln_S%gs4Y7=cU*EK_K0eCh8XdtsUqK(-Qx(L-Ab)e2KC<6+A%imh1gUa
zVlkEsuDa!zYZQw_YF~hBHq7m>Mo1W>ocER%hV#vngO^2>jkKSiTA|?<9zLm7+JD>U
zJ@cl5W+hIKzWX4o6lx%G&)y|V21{Svw>~0f$^1duXhwvU&4Jo&#xD_hyt18S{3YP%
z(SJz?1K)=JPFVQDKN@$6)y`uyHO1njgK}Aa3{}}VvXFI47G5vlIqs~hUXZ!BTF4&#
z=fz^ZoYp>0fA4OZ26I!U3H~MeA;SNPmWv(`6_a%L$n&7^IEp?X5q|%EJvagQMvdha
z+}ioOQaNqh4_0%JP1(C~$F3y0Dg>t2S7z%w<x;0G{hsVPw{&UuN%p-KP8+qY`qn1U
zJ-6@FBX-4|D+r016V>g6rm;2(aVFH3zu4-4%%rcR<eJZiAlrbf4K6<*3S3W)MF0FB
zCuP#;@f;B9ol*iC?bSI_a=5Jb-CIA2v9;<wa@J5|a_|wH8V>&G@a0*T=}JzJ5vS<I
zhgo?Jo*G1Iket2zP@8u3j-Li$Oz2M8CM-#R56jq=<SeUax0|YzYa{e9F^e*Rn2&_+
z%Ip1mwv&x9Xip1wmQ8y>lP^4T0CS<>+C&(r6(cUrXM^A5fG!3|_mNe;nNWLzQWM$y
z`Z#Aq9fM{agu6<X9p~5+E)tk|D`#<!9zB-txHqoT#apxMKSBS6`0d0fNj)zqNl8{T
zMZ_E3Trp&CqlEFG%XOY<@H(8PiA#diB2%Z`zs`KH!q{kh9A4Q<8_%&Lr(Mv0^D%1s
zsenDuhJSSJ>UwP)I=3qUOCB6Xnsve~=TzCVp}T+mEb^*fY*)d)r~_ZYYsodfcqT2q
zndN%g+jr@XDE2i-cQ0kH?l9J-DJUqgIDT^9)(DsO<(qWgUivtfoLx)leaoJm73`x!
zfPB+MFC!zPc*tjh9^BHP8Z`d>Yps{^^=cdZVE5evqf8gsXuvrA*`tTpOl+7BIn9-u
zNjTPumN#&40xT4Fxi28orqlBav}|o3{8E(FE8UuUc2w~h;$-&FlYwwZY9R(&t6c^u
z75}+~W|Uoy^n~szm77ElJX7+!z_{W6G*KpNZyx;cV9H3wMO#{JAwb|gEApu;a$<8o
zefk8~>by}ci^YuluRQO$`uVism&u;-aI_Wq6w1`T(^7O*UV16yItk*mahTJd5(BB;
zz1Mzze6aC$#<lJj$HuQRN%S~#{``MP>RH#>w&f$^(z-FeL;E%B<0VuN2v$$rTJt_c
zcY)2>e?r;l<Q{g+h={D&+2W}cN>)1tOZq<CKp=hZP4hpNY;k~$=`AI7%G1*)Uo0hM
zxkPbZAlhl~Ng-E!Jf<A{_kU{ve8zj|Y`V0{g-cnTbLVp6+M{&;sFX^{J%;(B;yE2R
z)#4afW?<j(XgJ$=2M=aWCXH`OxPc7i;-r<6t7k&to?o02cH@`*zAn$gt3M~Ou4P(`
zBNx^+tePsDlxjG8XVYF6He|)@$8zEJzK@T0v@C7c51)2?$K9k~6>n{hz0VJv<1jk!
zmOf*;o*gFtl0Uen6jeH(`R@mf{}sM;&aZMJ>rL~5!oou^a^2@xv!A`#{*a9grRo+I
zTmZ#czK5LdzyXl}lD}RT-ALL5U61Ch2CXD(KoS3fM#c4_nw^#eU6}L=-8szD^puhc
zgOlSOC)-VVUB+W$X-2V430NQSNzFQO_tD~UD(fb^=QH%c^Omv|bq!isMsQ9ekQ49Y
zKqe8*&v94M_3L9@TwH1o(}%xBp(nO*y6(}bQv~S8g6p_(2Rhmu&@v#mVMCc09PO!n
z`x)xLH*Z`HQNZ1WoaFSFj0(}gomjPed47V63(bin?{h*7hh0KX7q&7k4&2qsBQBn@
zzot;>$KNcHm#KnBiGFyZY`^q!QCV$%#E;<L8qG;sy_}6S0OYfS&}@emjNZNIn#mOB
z-?lNX!`New{#&ZC3jEQV1vnNxsM<%3qxmp7CU!D7GH|+?N5;})x3>CvoT+e;QuhtD
z8e=OW9AdT7{VX$iZ+K~;?^LrrypzNMOg~&dYyC*Qc1b!H*|y`!cPk_o-A$)}b;l|@
z2p29Dm9_1|{Q?3Gv0Kl@rL(+zXKet$Xxi9<ME2`O9r(%=il3@gJlO)wgS>dUs;byb
zg!F^SD`UO}%aPFa4s?&ZyoruIgBD~A$fkHmmk%F4pxfu=@V=RoVEOn}OmB)OE`t{J
znizC>0e6B$q+ESUb%F)%tjV<Szp+DT{CEzc4FH`gSbcBtv7BI9UOYle-=0-fRWRd8
z`Gl%L{rc6Cl!&CHZT1-$)CFaMBxZDG-^!0<_0QhfqI0tU?4-nC`RbX!T3SxJ$V)8p
zE4*<}V&28a+7UO&O_ygMhHmZU_T*4|Y{^_vx<_X6w4@E+4I6L|TD9~hQXqI+&O_1@
zc%~J<x_tZc<tMH1Q$mxVgemC3K=foC)EULjNc&~(BnEqB>MD%9yC#&ew5vb=Y^)V-
zI{s?Y6&05mk0=`Nt9=9ctSgPjgU@Hwf1=Xx-d)7!HnjNGhBJbsXjNiXra1QR;lqX{
zWg6+RV~?gOIbcBl{tv3(I(^1<Y}vJb8bE06#snG;e_io>+j~KzOXs1qvbPU*>D6W&
z!75u@F7Z;}DaLzGJ)Un|;D7ViuLEeuG#|2<K6?Lbo01dHOLI#DqD|9TJfxe_a(0x1
zU&DCIpOlj`Zkm<>MUwWo_b3iE&A2+3cwwG2eLX8~*M>c(17RY0__(X<$)gwI4yw`}
z|2H7YuTKB=5kT7z>seQJ*{X>(gbA+j5hn9`A3rzobSran24BiW<9~oH2s1b+sANBf
z#P6G%8H3U=gTZg2)~R<-oc|st2Np+#Mz7w#|Dp9LgBxp{5pQ^{+_mE|QxB60ix6)b
ze$V2iEt~nCrcz@cT4Jg6gbzukR~J4F+k9@8rsfvQK$iaYbWH@tO)Ato;5}CJZxv^X
zy~JhNu@m|Czb2t6e+%YK>I+q#`}+g1KK6j>fPfjV`wwA!;eylGUSco<RSJ{=q~?N^
z7VmH0cypv9>1AJr;1{=y^76KJa&{hP@4>dfn($?Ht<^B&zbEZc66UeRjruAtdr9KB
zX_9tnoaUqC5eY9g<z8^gk?%}DKe6@}j==p1X<imOb+MmF8hfmdNscKUc+Aah<D^7N
zcYpt3X>UE$sW2xv$)#ucF4i9zu6ZrfV!M2?>#21mi*yI0<C6%sWh9=)n}R|)omc4J
zO?bmG7Xi^o;->%`zd}n0aiP+Z`zq_*Jgu{#g^3-w0M8RA4m6aXJLvkGXV4_<Faqb2
zEh#nhJg!q}X}KL@jzw@kbtNaCpTCO6h>vSX*AR6Ix0pge`{@K{OUr6{nV8iM95!qY
zL98)@I#Y<ajC;Hd1)*zgb|gvghoK6;@S9k_g9fO%Dqp65|F_U4biCSACdoSFAWf0&
z*Y8^VmGkGP<AH<1W73xRHnIfpiu3^3dcxtNfrX@~6)k!9KmP>LM9w=j=uRke87sq9
zHSptw`b#g`%EW~E!Ow$#J|ihF8?-8G7c^C3+rj&F&COe)Oiy;}v(}b5Os&*9X`#M2
zm62}pd*9x@S(~!1=@7koUv9NJ44E_7+8nEYskEY|Zv%!8zteb6I0-`f8%*myaf05>
zSJe2fTCg>>WadVGa<ut6c5q>0f^P@&3DIx_Z{ED&D@b{pzx1AnnU9>SWZn9j!Kksa
zo<A?5`9tL*R<iO4-ydCobLbn0TH{YW+5X1d+0uto=&TlPtB2gjTE@Dgt%cr`nBaZ!
z0{5YxOxw+6aD{MPc(78bbnCIundDN&(0=o8q`%!zWHS|x&4FkiA3d_)u;BwwPc#&v
z3q&vJ!2|9F=S|FfYUKz&;T?u@Z&>iT^e7whFszsjKMF8jv3hl8YO3(@pe05{^@Jrj
zbo)@JIDb^>2xot99Ou0vpA$r--b8nvViot-KxDJNzC%WgaDh#MXvCT}>)pG3_|9BZ
z+@n0X-HB1z0S`wE8#c+nX@LPXGJoKj-HhcUfWGeT?r&Yz^)+F*VqMgT)(W~&h`ye7
z+i`q|2~sxPq#l&?BN6b(bUsw7)3{}%YRja^p0~en2a79I7xx*^CH&b(2PcW-=aqAY
zYD}KI%B(N8nDLw?8#l*}wiwgo%_{}G^UqUi9_BrJPjQ!zl4Db1mfGy!Ki9VWY+aaX
ziBvpWyRI0saBBYaE#Y&9e)=WUG?C;&t%1>hdwS1o+1YU9$RB%0)sgHcYA)=|DmS+o
zGjCnp*C-G_i-DO?+uFY6F4<hL+?J<Lr!5H%6ZXrI-9{fha)ievpJ&)_#F@=3Z)6et
zGf5*73@U9`)1!-=RAN6uxJu2gf0mL`M$^daj<1QzK~g5!0HI?;@64Q>CLtXBwV}J`
zVwvN;q_{Mu)`Po8R`0ZpMucoJ=xtUS|1IT~`sHZtlzIF78!;d8_S{SvQtI2IcwCog
z1hQ&8o58*cszciuM3)@#(yQx(Qt8|MmkJQ})m3VR%Fw6mA6HY$!{kJ*J1R_0t5oN;
z^x&+OHaXd08!9^|oJ`TaY@!l-pz0rnHqb{*oO4|#IE!a+sJ&6&^_HUA<c(F-V2KUg
zoU<dgbb_<IR{pOq!`3)v?ixIGZmgl+U?=pvx`7Y*0;;O2F;@Hkv0D-h<qY!qoY+s2
zX_Qf$A1p=;VmH>TpyJ;zS>?s<egresN!?AR3?06Bo1oQhFgix>&s%nov6-1Fo-xq1
z%SKrUS9{RF%XK?!t*udY1_aCq6{brdMCT?IZ=Rh$fQJ~fLn9rZEurJv{Ce3GUcS~}
zRZ}~{L)YF00-%prKk8OE3%+Fs4?e&NZh!Ld#hVFxL)L_zHu$3ZmhOwmvo{0H%*wFI
zqC}~<?{PLEp%Y47EYL`sI&Fftm>)rc^c{qrQM(;jjZ^@@Pj}pk=-a5d@SLIZJ+gxs
z=mE$!Y$^}O|BTWqkGs6Xc#E^gww;G}pjbRX<^Nsn{ZmQ>Xk!|Qei_Gn<i7?+ojLRR
z|6=dW!+Kob_utH9CQ0TL4GJkyGL&dSr8$+%YAH%8BnufcHL8SUXq1pfQxuJq%2bJ^
zq@<z@A^Y`YeZI#&_CEHXzdaoL(;sV5dcU9d^W67!U)On_*LjgSqcT{ALa9bGT-6Dw
z-m0V%pLYBC^-D<Z^4G?VGKYH%cNsHg44bwCu+@F-O(N7+f@Vd!yMdRUo@$&&KR|(7
zc)O<=WV82=96Y$8Dl<*1FQkW{srX?>4kAeG+;6Ri9<HqXg=?G2t9FHz5Tfj>=ZG(@
z9`wyfPaiYObwOtD^Jk&35Ut3Yb}jvOt@icn!J|iy9y8`+ZSBP!@uW|JV02%?NfS#F
z^I`jP<0!UA&-Q%0cRy+oUx#J!gbg<|SafXH{j`oZip&xmt&vVgOrL5f*tGM~*2uw%
zT!see3&sP2e+(3c0M&PNYdrEjuL1K;8pYn#)Lh662lasQB<Yx7*IjpP&H|waiwsrv
z_dm!<P@56`g{ik>rqiosC%%^x7zjactm+u(>crvNL1(9uUFQ1lp`vfjNaK-kZBxHp
zEt98(Vq2BM(~4k3A_*M~73a(L=IXy0)s1$O!cy>tKInAkAj3E90e&D&g#Hs)QmhE6
znb2_&7z8#Ap0RGkK`xYlCa5@g2EcH0c6^&UVZyyD=}LUYcI;Wa37o|7`#Cv?oi)AN
zTM0#I&*@emx23??K>f>qSA{NkhajBBI{~)?A;7OdjTb5`OyF|C;GoCjNkkn+64F`f
zWRKu?%{1@O^fw2gBUB6h@HBaFkwPC8(r1nnDw;~=K3kHBWWw|~Zg^HOEqwNLPhSZ;
z=nqa&3XD(Fi)8aN5;h)S0~H@^6S)LRi|tIug|-1%2OA>2YD0|KqeqX>sV}l(vBRcS
zQU#Gbb3SR)Sbl5AEWF!E{+Q{*0??w<L}<bE`PE=gP=t9CZDcy=;YpX_-j{wiHW#{B
z*siCs=|Gc55eLcOjsGnf5=A9Br!e+`R%2BTSw^I??(#yq-0vn*=9CmReTTLR-<~+J
z776VO=og+j(MKpYI5*J{v*LtNsW_mcEoRER8I_Q0$C96cWb8aWXaAX$|Dy;IU_}D!
zhYKF<5@5(sFI+0Wu<Z80{+71o()2;ZCifN$_*I>JZ_govV9dQBkK<_|0BZu%;WtUA
zbMy1vJUoPEJx+WwL?krV^DGsvZF{f+fD?ua-86ax4qu=4))?4uq6tj#69hfpCVz2c
zl19A4TGod$7{F>0LqT}-JQ87Iow~=zc%K*1#7)u>40=(of8*V;jB>WL4dFRrup6T>
zgS+V6Tz!{pvFz1+9>AyO=4ie*Z{9p=YCo$5^L>B|xb2m!X#~|m!_ocwD8n*Xr0fI@
z1m{$0zUeDs*8a4j+6CMw;Vh$hWX#Z^ZX`Hpv-x!-H4rt}BgTOx6YaO*7B<=IprB&c
zx-cOby%gtAc=I3<6qe_VvyVQ!eod|R2^AP}s|KDD@FR(siN{G`EzQ}p$<u|7WjebJ
z?E06$dQ68A13Bcl{lLLQFRIY#z-plh21WM<HMJD_ZR7Ar>FDTC56C8IrxRrq%NhNc
zEBOr1RB#><J|iPTSXmhv8QgLr0k8Os1p@?C|K#l31$(h{dew6g=L5GG(Bd6fgExpQ
zpYkR&J#tW11k&u;vu9(bG+07H7_-a=w^q=a&$n^e$>8vy|HlboDzAXlPY*{9xF(=4
z;r1a|AUSCGzB0^w`SJ=<YL%4KjEsiKi+hUcl#dxO@a)Vh+l+_zeR6wwpF>BVzKcJ2
z$?)ii={Yav-`olSkmxSXI|n|cm9nO`)-j>gY1JyXpWik+Jy>JMuCkuA4X?`b0eM5{
zV7btCU(x6CVJ+j3Hw%8d%E+qQ1zZb`h)76&91T6@S+n(|P4Vx%a1D>k=6i`K<pb({
z#D4WI6b4SqO1X?VlzvB7XZ_k~A|g|(h>gh$?_)$V196viOXlhOZQ1~c=FiH`*eQZ7
zJlz%!@nLf6FU2KP3`<YVj7+;glF7oqYCc<>gTO3dQWZhxjrZj)mM73@l)o5B`atMO
zF`u1y$>NQ|(zvf=g>t!CRpzFx6IAv?kR@b~TyAR%b`U>o;EH}Fcv6Wepk=q7Fo#9}
z5X)5wo_0%#O<_KaQ)xi@n_@mSII=B}5BIJhbXCHN$%3#i%y%*SAA$cc^Ubm|gaSW#
z4NR)}!R~jQR)TL@JDF0sR9yKqXR0B$14;HA#%k9F)3+k#xM8T9Li^ybTp_&9-5H;@
z1``D9WfFf|^ne`}ITVg^(kbEXEP`G^fuU3O=k(ly0>ProcP{HVJ80{nu?Z^e1Z7VC
zja-j*gB<s!1tYg4^Qq!)5bZ><4bFmmr07t($clbowuSXO<b&!-?P$}!!p(_1_d7_K
zN?v&DBhpu)cvr}oU@qJu;`YA%`<<MrXdHOBwlv>}$T|yD57b1VBHuRjxp9(?zP0Pt
z_4VQSOJ#bNx!;4XG({Di)>Qa8YxTq9<aRdG#^gUPki$LGV)@r=mSK!qv9mkfEC|c8
zgY3H8c`KX)bea_^l1O-w?IyqRsucT`IJ>Yn9L)qNDPmx}#4*d_V~<A(Bk75x>zzZ%
z*eEEH-*oX?y{-|9uwC!zGE6^v_Rb6A%!?{M6W!TeKEFH6`{gL(HhDTcJnojzjL3Om
zqc*0m?&8HaH@)87euI(0XpUJi?%G%VI8({die4Ctc#bAeS!(QkTB^_zbm{qzFbpne
zLr9SDJAZC$<Sd4r(l;@`{AkyyBZP$`auJPP#nX*DoEeYXu=P2lxK64olhvku(ce)M
zG1sT+#rQg-)(PFu4%M+049$eE<Js%Ch{#^2zliK%&b6&#vZIP5nW%<5Ibn2H(IeM-
zitLh^22=F!mzD4U{(K$R`R}(o@5KMod$P5BfXKWS=8Lyg-|gnKh?{@t80U+=^KEh_
zOa#_Ex5RHxVVs7>%NOIT%hoP4GFs6nA(AOfS^adk*H}vp(LayN`^0eH4%vSvY_e1*
zcdnVvpFNa1!c4l(um8($YdG~1IXiQGX>>6ZV8ZAgB6qZa17Wgb(gX~f#EvM(R6Gd=
zu!qwNRvSzM9yOX%b~ll*zmQ7Dk7i<=h@A}S0Fs3#6|Q6<=sPUs7=vguiULhz)~pTA
zLemV#(8dH$k$M07W%}Q>_UHNhFI{W@-P@fv`TzZaZB6HCBVu$N;7w>DRZuW+-Y;mt
zH`4cbh8(t{gEWx_ugr+inoETbl_<Tlfd5td%+wGp>9<2Sx6hTHGso_A73X_^HZvPV
zn(NOA=A9zf+Zta!-J9UMckjt7s{CE#%GAzS(dkl34jFb{Zs5RyU=#r=J%k^d&+j}4
zUO=iqvN-!+gl&ti>&j26gBEuFXo#WNQ{O{42@(L@J6GAb@#Dsw<)~gb<4t2@CBl>d
zsvN$`wf_79=?tUpH?E!liIkF*G)<4#A|kRv@y{<8eoI#}-69-BM*WThIg9MN4($b_
zBk{ADCZAGzO!l@2zp%^f&yQ88pE2^jQ@uaj@q~%+lPgM{pXvCTP4X2F#X4VIZ1kU3
z&#%s!Of{AxwpVzEUFY_I;4lMJV#c_LKcC@H|IXhv8X_s_xN6n1=>5;0KNox|)*cdW
zqsU~xKfmC)efy90cJcc0B%VkGRhIqv3=W;|a#q_x;kr?=a0BOU{_~;LVQhJ&_i5Jx
z0+x|C6-o{iUe0oekcWxXRlj~+OopQA!H`h{a_#@ak1zT}h|8x7zaYC-R~Y{}FNNxM
zmWD<{MPYXK1u8f31Ai`0PyvXDY<V($?&WniI!|`H@qeyqxOZ$a7hySWi10dBr2jnj
z!$ry3)GI%K{ZbVFb45M>Tv2vzZVaW9sK%cQx0=jASbnWDjTLzu;%dw5-4ecaM*)RB
z|8FfLKNfrrs%ijA8ycKmdv^Y~>z@m@m(dSIB%1Qp_RmvK+Vtm+b?Ll1qu#DTHY}~l
zqZs+hZP`~MX_S<d1kiBZQkHLc%Ag9TbIZ|rvzy!D0|!VfMMV1k09;0yv$2tXa%I#7
z6~-?LV<jhx3&*0JG1YJD$S&1w6gj%Ayk_gj&MhFkfymX)tLpsYxWk>-|NozFR~O9L
z@gFU~{|9cwhj?KfC-#F>Cmssc(Nt!X{Rqv_5njJ`&44e4#~wZ`n5hAr3>Y=D`EBk{
zr(Qo|uO=rm8{#xV0l*-^JQxO(+)5a^2jU6U1JZk&zCh>#3Y*O&I|#d-9te?AM7Bd%
zBWG;@<$x(FK`tSi4>iF*Q2BTI401Wqrm=7vAU_KiEMSz@cv7{Jq+4OseauiPA=&4W
z0B&?Y{!=S_nPYwf&lzhPO9R4o3KTrki}LfIJl_dN4<;zxl$W0mtuA~q?*c}Qm!@*t
z0ir}%f)D6xsxd0hoY|ex`%c_ZVe#sNkYzFifWrLcvuAbR2ie%z%oV<F^ZVcsn8LkW
zd*q7e0YASFR3MRK-qh5nQuKA{){`*`y>c=i0t)bj<^UiMnAZ9$wue9)P7rxis6EFD
z5nrTPKT4S2d6ljPmx|P4Bo@Ei>ca3Js{8oS`BU`S7!VF_l{fQ<3QBkKanNabAQ<8^
zw4+>oFYIJuZf6KBS?DK`#GQw7u+bP)SN;Ayh@-iN#^<7>&uwi2un3gYKjUKO8+*E+
z;3Sx*1tcNdHyfK<viAMaa50KwZ`v`Fa|;e)$?=QyvUbVy&ZFj#*j4_&KF<Dj2iS&K
z&(U+?3bDMwx93YsiyW0YR|ob<Uy(4-4#W@s(_o>^Pt<(^Y=uJ@ODzT`L@GCM7SJ-_
z?WpExAbIPFncQcD>qW{BI<HQwPmfO+%DFUBs9jM(Mu}ETVaHqqGtu-8iYlT@*}=31
z4BEgNf8$d%S^T@gMqZM?qSqfy>rZ%ic!e_x)Z-dc5G20oq16M;k^iy!s+$?xdcn6s
z(D5*!F$GIL;+kmF-ym{;FJT>K9L2+uVB%}0=-&~cOVzsRIxtbii}_vd{C+?MDAKV=
zSN9=pU<zIqP&q&$FfBiRes}W~$1QQlm@#%J>ex0c&w4i_`cs5y5j=M|gYT#gn41L(
zOYaHmMX*MP&Q;q`a{u_G2~nK!G{o?^1TYIj3jkZzh0qblhX`$!^>yWe0Vh9=HFE-M
z<GNil5I0LnNojpj;?#z`cv3Bb_Kyx_guz&@3!U#=L8qcKzfab^d$fA=n=?vSH`Qk6
zj;~zDYOhx1Ny?<T<UqED{@ertQHc=@6@QVzE1(>F6jF-X6hSN$zc;f+;!Ne?keZvD
zL1faRupXbnXBAibfn~@ckD|Q1lC<tl>(@uW{Ts(Q?|XJh@Td7bUR^L4p0?=ys3S@-
zs1bp*{WU5;hgUyaCR_-2^<=lG;x1xq1WvsAH+qyyT(ZAMc217qwK#FM`N1JF7nh#d
z2}ZzL36V9{de<R1SPKv~F>1P-;sY2Aa6=F#xW?lz-vR&7<fLSE^?dmk3a-N*G@fHw
znDS{p4n9+=nB$y^Do0Z*c+=<F#c3Iac0nDi2$WI}>A}xb1L!2KUDT8kZq?b|LeY2U
zx3NSKV<6wq`?PE{|4d0<-Vtd2#tq*eO5jk0F;}Qh(e3xygXp4FoXQQ~ou-`xx46eI
zgh@iro>{gO;s3yFynFYeV4dGI0$@D3R&nGc&`5>t=FOY;CC5~QGIF<rGJ&3os;9LH
zSFp>Xw~$73ET!i_I_2HFcS5|_+Y8~c71X1{F>m;C6dCus<pi+__`FLWMF=9Bf8uf5
zoKSU*3|CYpi$vCmKK&}C=Ygj&1|6N-h-6LBz4UiBzl*f<ZB80Js&_{%>dr$*@;KAQ
zY?Ra3J6|6U9&8kL=$X!hL$W&aq{R;?>c=PjeKSYmgUpQuQ3rddP7AJf5*xOi$zQVP
z&RHHj(6y`l&9B`){BUTh{7{+WDmv)%Z|~ijh>`1y=4gCo<hn?yTyG)k`k4U5p7|1x
z`gBOhrPwH@D{tY3=9NAM=yjKNytKj5k>=(rGf=~w8M!z(cQyZhSnPhaD1-Iu>yWAs
zbAIVJV!CGvDzB5$dcJYP=tya&_OIH1Dk7q6^8(@4iHJlm8zkht-@=oVRhz!Bq~q!b
zNlHq>&+T~}1xN~Uq%va$J*Fwi$+juZqdj6|@_1P>-9|0{w>UF%@|;UdblI@s{kwM-
zrl$98uA;SwSEQcL>{IBqs;2vVgi<qS&fJvp`Jz?GdTg>dmEJf{k(cMg1Z0(&%c-aY
zdSt}JY=GM@^;rJzRrURIRRiJCA~RnSXkyj#AjQV3)OzFiL?GzXw=YK*6W_f!9|8WG
zYHG4G?q4{4TJSH3-M+=%{uNu6(0(1EoT{n-hQT9NE9PdD%DZ&0KxI%|d?F)4ZR57B
zTfYDx#yze3eKjRzu4nVOCT&EVKbv3gSy`rk;?$|htwx;6W&fv31GJ3XFLYSD2QuVs
zx5p0-U{c$(m#i(up<5=z?(NQc>)G?_jMFnWOUcOi+6We%98U2^ut;Ep;XCo#g#~Mr
zpF<L)%Ai~F&3Ss5>!hvssiC2vXTwkS>9FW!>A<Ct`}_6lSGsNezqiNz&l1`HSlPe(
zd{}3Dvqi5b8bluGz8mB&H@xj-Rn=*=_{pp3U#mDGt#`|_Wp_&Ei4%iUnu!o;GWc6M
zm7AQ?GB_rGBQ+ZHaa~<q9!u&#RG;3+NkmJn6qQ2_EBF6-yItyi9oS&Ie#<H<)+|1w
zHE01;fSJ2kRkiZ`L6AA(FKVF8(I3<}KYjdokbZ4Z5Vn$x=B?Vlcc1m=qt~h!t*xWe
z!zl{Mgwy)jE?KSEA@sOfa12o+WNxO1GH9nNdV&)_#>Qriyn9JZTs-y%C!FwQl<{XR
z2;QomDIWEudNOSd_4S`}E*_?dy`*HpRzqFg>PlxvNBNe=TG~Z1QJ<aPjovMs1tdJK
zDbqiv*VWZwHox0u;osRhEb*}t6440>h2F9=XL<sBY8|mqJn?dvV`1*U;FWuqEaAO<
z4j(;QTUo_9L@6k1f1B#D>5Kd#mA=EOYi$QhUDc0Wt`uU~$5(AyKAP<Jor}Ao<p;+R
zc}dVtu~0){3KK_0%finB@7ZsJiK(gdO|3PY(ap{Xvn(8{$c=@s=<t6(BTS@c3P%Js
zk=Eba-Q3)~QY_L8FBo-+6iwUvsccYpV=F7Uq#=16E0|q8dJ&zNI2SIj-sR(=sP6Me
z(YLm(W9Y_>jp&#V*)-3oZ)%zgXH&oM_ptKwdP9NCtM<cv8cpB$=Z$w+@107Pb>|hD
zn{lm=;gBC#>n!VnwNz&0YJ`hi!HLl%E_XAS4X2$uQSR)yb1Ug6u4puEt$6m#6Uk-8
zq6_P`#E|@8;edWA-}``cE`KaShUHOcrfQYCX}pX1h~49aj+<45|Gw>u-GAQp9J1_l
z=RC(cVwy>7ZQ$GmJ9f-&`T|)c+K50hfCSPl#lqZt;>3yaucUi$NQG$Z>+E{EdR3By
z?DA!$hDZbky%jg*tz5fyYeTsr4XPitUDI};5=G2v5?^>|v%5R|qj-~cO-&8|Ha<7!
z&sEGznAbUm#OJT7Osl#`5`4oj6}^?31P2H;_pMvkZe%oZPnYkstcjYb(ur5~vl`g(
z`Sa&PUPcKMW!T5-HU@Tv-qC8#qf&ly;pa`Un~7^~-c;33p0&V{HWQnft_fx*Z{GaH
z;f?8=r{HlaTRyxK^$_`{_U~m7F8Uk^FMs@4m~F%9eSi2D6;)MXx-#LM6fS>pPpQD3
z=Sky06)GBe4NGorlTvcH@m@8(d~!)>Gpmt@4;?zc_qL!Uy?5{B^$qvDWo2Zt@7>!o
zGSy;iw{Ck_7sxRlWo7M7nWv|hOBL-HuleVO7#$8QY@wcQJ{S`pk7C4i!GgW`Wx8}B
zyC<NcL9>IXR##Wo)QphMrGZD4<P1yzpcRV!?ON4HTb%vj`=K7+&bhuOUz}`+G#JIf
zb*%Xb!dQ5np6fHmWg`*@vg?ejrOUkd?Zr!%s^Ku@{8xGWh?KXanz748lQ~GJgJ!6x
zyq(iZa!;c~ADMf`!!Y3_f5WYeUQJR}>N7}gqw_xjlu9e5|9p^W1zmwOPOfBa?|SjN
zp}~f2;~E?&aKIIL$SGoX6heMxD&lK1$K^X8OjPqZIY-OKERCtBgh5FrVfFR()-`<v
zgPCmm!@q~<g(GA-yW^M2iQz-Z%U+F}In&a2lDP;&l=?IT;4)S6!dXn@j?cIlp`bek
z@bD;f`u>bV7xmF79#WfdXbbrSw=XUBMbuQeM^fCy*|RfyN-kR^%nw7!^TPShau6vU
zwO8OqzBQuMf&4aGJHfbsLrv^-ThQ_2Q;`~;J112(L$3?h4i|gFbA*hGM5)}cVZ#KY
zZajdg>YOqZ|9zrvoq=W0mfC_56btZM@;A0K{A=4b_0H{pa=eDrE>5HR7Xmo&^zFYp
zZmwKxGOI<_i;k!jD^?s)U6?=;O_52CuIIm+v4?+N+EJ`CO0Fakr%C8zU7du4gzRRK
znR@;xLlDWJsPQs1^fO9A?*?iGUU+8aW-a}xn##(|8#}>0Cn1i_8ESM5`X1Q6ucpvv
zjNm=&uVZ<`#Kn8|=po>T488?eOs?$A%ZZddtdlFbfL%Vj9zN7Xh=)oCi0dUaY*Eo=
zQaSV(m#M-pUR=v2L94%V<r_*QHk?qBA?|&zN@Gpdix)4bHP|ksk)@^AIhRpRu&>E9
zfP#<#p<6<7;OKqy5QY0JZaUt4o@4v>?j8O5i|pB}S51^c@@qcfast4}7qUZ8L(W1p
zEX`Ccu4$P!UiSWTqmuO|Do-Oz$!AR@=9d5YrpUa}!$*y(;2HAVcR1=B7__yvp0q)V
z+)h1MX1-lS<mB<ve{Z$$EAs|*-tK>YTV!|VbN$cX26aa9|NL#&xXuOipTBk=j}83)
ze2wcZoaO)XHNP{H_|IQ^YyCT;sPh~DeM^!5-dK_UCzj6!xf|i=PWJWGYGx(=mL|7U
zR8aWGSxr?{b^7$U?_c=tJX>=ss}VzoE_<24-=c?#^^3wt{}oXod9i@ECkqn_@a~=R
z@e9C|pn%T_7w{T)Yx+Kuk~VKl0Z=)+<jw)8?f?1u?{@gVXVaI&Huk(THx~i|2xZZr
z1!QVSz8o`V{Z>c&C)&UNh2LdQo?z~2DqyKedLrXKqSzAT8P0-1C>7_%<X&fM+vprl
zO+7j80Li`&WqxwyDW4E99)xY`6DI7bDj|6njK)?5h@92O+3y?T@sl>_w@`Ul*fa{u
zL^P|;s6WW;rL0t11d=O*eIkS$5gjuaQ`IMaL26%zTy9>TXy3klVXJ<3c;j?<@!~~a
zUmY4Cp)DGbd0sI4G50B1Ge;kX2UrEZbNE&PbBT+K19Q^j1w544*t0{CaT&m>q>-Gj
z0B_QIVJA;Azo`7~-AFP!)$iFGXkIm(*tbug$duVMgj1mimfA@(O4slPoJ*8=pj&zd
z1`1QAkTqSqa;1*Wx{#2Ng9r6Cstqz7H+gdWl`F@`9bnW{=~kN#cNz!2a_f{+3Yai|
zL$X`_aYhG%cW`VAlc(0LL+)#7X?YzHD<2PmC3(~C=+Bkm9L86!ToKe<3@kwC1}`EA
z9UGpe0Kl<@K#jm-lI17M%BBJeZrRd`iW9{Leb~KsPenO1eR@O5g~Om5NFrn7@1u|x
zj`+Tpc`V951a8Pk!EvtA1f(_N!I`_yml!c30ynQ4;oXe#Ye{(hBB}JisrII<-m&8+
z>PoM%FYs|7f!jMEiJB2YCTRv#kTCA4%w>w{J?=5G`AwMQT)nFNF-m|j4j#-!X+YVR
zc<GYG=fVlSr6x^!@aolXK!bGyHmoiwE*?Q9O@c|6-ST@ISCQ5;>eSuRQsVQEtko-4
zU@&={oS7Cxi<D17Ps6Ppp&6AqIcrHfw6-%jtR0Z((r?layJW{<PJz#F8*!~@0+(Vq
zDj*{BmMy!+ljU07U0q{e-{`VKlG}%*w6KEpxz5GKr7DX=x9Y~mksjMYL2#~%jEb7D
zB@jI@1HZ1#u)}3^Nu|v%X@TEF_(;|9&U*`vIKakMAntS}IHfvVPp`O{TUaQJG^C%3
zlx?qizd+>iOW9jMgFSopl*%Sw0m>CdkWsOI`s8}TpX?kFsnDvpMmf}I1hXfB4v&8S
z)xUp#P1@b*$kuV}$%9Is;M=2DufB~z0Ra=PM(IDE(1vdE{Kr*gR^0NhAfC<55_5KV
zZcOo&)^nvQEDc}D7JIqzyT|LDBLq)P({DTktydLH#yT)+4PwU%TkLof_D$;UBlKee
z1qPmJpy&asI9F@q`Sqi%larI2qKb;jY}dECg(=7VPoM6f4?=&#q#7>1EcQ2_O;Yk_
z-$O!&%no}>+qGU)zVv}*pNSo^*ZDm3OKi>yBV*$vnnh1UkIDyU0k3h2_o6fy^E>F&
zspYF4m_<@R7&jv7-Pn2U{zo<G2T!^2@>7%0v|gUF0a#hfmyrn;kxggmHhE|qHe+L6
zrHqD#P5-66JE9mn2XivSM~e*wvTpPFNPK=}IGLcihaJ{OWo*2?@<8)zk{u`X&Z?}`
zl#$Une}au4pV61llo%atbg#j<#|f*4T$SppG2h<e_j_f|22jp~Mux-x>=i7L@PutQ
z3|LN5Qh$x~eji2Wrk{dg^4Zg{9mv|^&gq<$oXf;&?17@=@0uMtc8o{j`SV*+x=pqi
z%}8uZq^fehpFe-b2<RGVs(Z>Ig*kJo3JV+A93>vPB+up!G`6&Kagu^$I{7=V<`Z6s
z#!0aq?-83WJUZ$-d1>*Up*-%nkXtA{e?gj{73kNy_cwZ$@c-3naz%JLaNse9<b-W-
z;MeMW^nN;*s}dDWh3@;r7tVc4!&+Onl*bP@FvurLTmKzPiN8qZ;;!_wnVEFWL?SUg
zZrQ<hgq0!cnmEN-)_Z=v7#*#<UMleT+qbKc3`BlB-gR9xp$Rmr@8VC~*V0nK%&|a`
zW1e{rM(zW<vu5SVis#vS0D=+elvAicEYyrA9MPE3-!Y$(K6r|~U4kt!ie@@PnatQO
z@Lr$4e5s8%ta$$FRaYVvHV;Pjaj68dipolSaT9K7W2B=wYnIh(!S$z{c~Z5(KrK{u
z@|Bl;B>bfp`AN)Q=U84^dWucOy9zS^zztknnxMZJ81<e0!^I3w;#fg-H!o-h&j`u~
zptUi1boJO;etf~C<O!1A3y-=w<S~t$C2v})WF$;XW8a!4*rLv4YSP)$waw8#Do19&
zt*yQ4d`LkKv%<_Q;DxI<s>3NfI#IJ|AjDr*HhWFZ7;FYVa6Iy?hpL={0*!BDJTfj^
zs6k%KYi@xqp=KXBVnmW?CSjbmW7U#{D<OX>oUCnaD}~WXRPX|Mmg2ne$;JM0^K^8w
z0cT=|?XVC)8Ucx_bjEFqPequa$o<Z-073exrDd9$TB7qW_J#&t`O1o=Wo5Kks<<5n
zI-3*R9Pk>C_}5uI>RoVS1ixlvMA_rV$8AiswO6sLh_E!}M%_R-jb*=@>QnPwK38d6
zq*70sA^*gy`Ri8(0~lX6KHHTII=rbw5^o~jHywHc5P?Y>4;Rnh%Etqqd^)f1@4=NL
z!}jfCl7l8H4`EaS0HJWTP*|sCYz#A6`<rvZV%BB*i2?47FWjlwu)U(}orQHC=!&mA
zT>~rV_ct=5vjvP}wX>k`G@2oxc+q{N`q&>Id9$3ZMw2O8f`3bfqV&Ovy5#ZGrkzL@
zIYC8AJHSIFskCA_0UueMaG;`2RZdp+zz;{-FTPP77OX`6|5j;(Oh7JntZ^U3Z~6Q_
zo$)Pb-@Y6U-Oy|@8_#zOKfFv?TB&#*$EaP~w=7<~Sbk&gl+5Kyu#VU1@@G@YU+AE<
z%{0x(E-0;7&!0y9RgG&L9E?dq8Yjh{vMxB%QCm2&R`(fgZgSf-F;29x39o7Z0H}AK
zG+n_Mg=+f1O9SXVy7WRo581VPQfk{a&IBB9(7@<zc#G`<*QjOmkKWCg;2QPxJVm&h
zeWRcZFY?G-mqQrG4lp*u*ZuZ(ZU492RK(m@Hy=p4<Mbabz&|-`nor^ow%X(dp0|6G
z*i?Kf{jrh*m!zj^=pqyvEE_KcG5h<w!@FLtwJ6YX-V*Kb5VpxZH2ASz%NYeyDHulV
z=XCY;OQ0xj-1ySk`T^w<_u4(~T<~%q_CC#GloBjm5V6L(y6B^8pGx}+6~$<AakS8J
z0#&yhn4kn=k6eg`K>2P9C+pv;c1`;Ul}+q|lqQ`6ARz0*B9;sYUAwqdmfb_(TExNO
z5N1I^++hbNCk<McSX`!Y{v7FcX|6ILmztV4Gczj+3gT(N;`qLnl9K4WJ#az7?uQj`
zK7L&CYPCrL;%<5y5(}bLF8lfU5j*f2AT~^+k0Cz3fb|bZ!HGcsDHot$Zf|EN(0$}_
zBo-tYWN)Y-SCK}=FBl<(?w(%Zkz>XPPEFu$D8Rw$=nY9vPp5ClW}9ZL*0J0**4Nh$
zo-U&@e~>Wg4Nhl<OI1OIuK!ztDnmvum5bY@Z#mfZd_wOx!jQ5jmj=2Wj*NUw%t&?G
zF7#EYPWRmbQdV&UpQ-(iCoLzU)6VwtrA^-37FH_Q_cXVIh9{>AjRdA=Bj^kv9VY2*
z4d#3x8)8W>{rg47K?J5+kBpl?eF~#TE$}A2@WU0Y7x=s&k(>AGal$oISIO8tG}OS3
zkDcu~2%|sr&oMDf^Sg&h4ZReCav}X@JWqk%A3Ws*-95LA6ph@uB*T)_k_~oN5*~3C
zh?b)xh}TmAI>S&led%JQkTV({N|Pr~W^Od@MmP;|srKG&K_Ee``gwUm<MnnvC@;y;
z1tUdF0@vi;K2e+D*iY?HePd(7>IlSHthQy<HNysD+RWOb_kw4MQBJ)!YXeKs3eZm&
zUj-{4a491rBkTc8O#Cljb{1S_THjA!aYv-KaMorxZlM=~qm(5_o46qp_NLP!XT{-$
z8^9KK_n^NV+<}EpWN?p<#II99Rlr9fi#?DwZO$Cw@I<fTxY41qscKO7V&RFU&ODJK
zFx||Fp*VRx=l<+tPe`qBMZJP^9-5Xa(Le3bxo~>$cvr@(yt#OCRaF(j`W4Qzunnd2
zEqP}Ifo|TK1zI26eN@^}sL)=)7oF>d{;M}{ZV~8%mooLtULoM>40sSLuc_$`gqb;Q
zIY=$NGzU#pH&xGzjg{Z8PWB!o53shP_6t8SeyKQla`c}1x;oEhDx0BmyFZ1LWB!6%
zYmuacgfJ&!{ra9xD(T8o02l=~a!l8Czm+iQg-lyYwYA~tvF|2}badR`m-yNoGn^Q%
zyynpTzl}-zhfD6%8nhQ%CBW>!ZKl~pMMul_fqUxg2$5eDHMDy%*y@`%E6a}_I<k<)
zaS8P;vL5YIk-0^nvzQW+uT$UB+Pd01lQ^<ela9&p<z>UOK^gLPy+O($*jjGrQ>Dt1
zJ~h=rtFYg$b1efzcec06rAWIB8AA^1p0bH9@rr+(U*0%kqda%+nXkS_4j*3TF#($T
zuI9%O`<*ed?N(1Ov$q)?R53w+#E~Wb(oP9ecaI|tFe@`X=;)-fjd)7jsk^7MaH0OI
zl(T1dg?_7H1y(2%Io6%NuY+>#@m%HjW3ui`9IC!xNBZWp7waYOh0gaUMt}V^+1AI>
z^5nN=SWkB~6GTY`jm-3jyz%7S8Ymv;A7q6S?B-fnSv_a9)S0#|hqe;PK)~&$kMYHI
z)SFsgwcx=_*F2~t@n&U6wIl+}9&UY7rzHN%1UrtsADxA<TQ2MZlZ$>#mg5sImURf^
z^kR?{$98P|A~SNSR1Z}ssByO}f})E1{Oxc>)NdhK@XW0rEQ352<6`dviF0Bb=W_0^
zz34zfrE@8lKz673Y+Vn>DxMhZN~aq(?k8$Tj2xM0b!{RYO?k1Vz8<3u=)bSZ;=Fk$
zj`YUZV?x8uuM5u|IwV?E)JXY}=1XOIzVvyV#S~{>Hu;<FRyH=*w#Fp&@e%V|xL<N#
zah-@!$ggYn>I$Nr9ULM(*TH8Ry-km>GTXK-{adw_w)Q;={yt_plYYdxe)>e*@j5fX
zNmFld*-d$yjI9ZQj?_(Fi6hOv?WP_%+{;c*Ike!_EBojnwLxB((SDoRPTp|H{uI^v
zqm$F3@+NL@auQrKwSt(3dh*us{DK0pS*OG@-=B`l91=Wa;iPX_EoJ?qLV|)??9TUr
z$ihHSEcVFAp~R-~dY>o)h9+*MmqW`vI3$GjIjy`&N5=K`_Wp$z&%xU_<nM3qcJS+f
z;K!~J$N-F-`cgUPX|6Xi8sy?oWt0tSQuad2U)n2R&HjD+V$XkqcT(0ihW{cJK78az
z%^2gA5}M=2?MjXU?dWfqv@SMbYC8@o?^l_HF5i7PtpE?F{L~&$n_8OIYl^0=4g0~4
zn?Cl<T8Zjm5)uz9b#k8&=EQ8{^TI>6#8!Z_PfPh+R8Z|G7<B!b3`8<yod^Sf&!qIM
zjbkF7Le<N2LKBIhbXSxGCUT0?X3P+8AH8C062;<8T##B~N#7L5=q)uc_-FLhyp<a^
zd>4pjNh;gaug8Km3;VO!x82>)B&)9bEAROFeZ^2kajN44a-m}6??9g4=6t9nO%@Hv
zhLk%zk8;!gB%^#={lDy3yMk22-p0)9%ZCvuT};;<JnW&EhmIyD;{@5v1mlGZd%Ao^
zp)~ArfZC*`K@P1%)ugQ>aHGlEXe0udbPr^HSNex1diBcK0n2~>z^ExfODy=YfBFQr
z^r7lz_#vIzckdAOF4WO6n|0#Jg9rO1U7k@ia)Xj-3dY+KmW0Za30>!7g~{P`B|7a4
zzO2kyWN<7V%8_*OYisLYWwAkj|2?0~jzb?_8DneCmYqB2<m%PlIrp|NZAHthtn>!$
z<qSk%#@|RxkNEjf2gAHRIyGzPh!J{&y6gEvmLM=;KI@bP!Z|q|eIZ?#qen-OPo|SB
zadeXdag6pB6_r;~!XTKEfomXQ4w4L-O)5zGZtUZ<VFNC-0!3L~Ue58wI?CAex~uA9
zGo=tjkHrNAe<1<MoZ`B6?Zmc}PX?12X>%l5$KL)n34y3-8DrTvV{2uE>0Sb6)O_kO
zHw{@7hHGZ+O7uh`CZa5$pEl^et>7$&kgr9UW&E8>N7!QZl9|60Uf4H{IQW1o9m7mj
zM@L~y%zfKHV{7Z!f`z-sF_b~}m%4qIGRJ;RmjtCiyt}H|M>>4t6@`3lTWfTDJenT@
zo!YBRFClMOzkWSu_}jO%{Mz#la;o%NW0)2^>-#n;D?Tid$~ZNdn;)-;4-Cz@|144e
z-a2*JJ-1~dW*ewA$dY=seFbkZEuUOwoVy{~`9<pH2@#9VQQMQZaI7a^y;@RU4l9OH
zdjGzCIo%iH=z!SF(+nXA$Pkk__T0MFL6KNV2*ZC1Ll0ay$U;pLO+eq!QgG)BZMxk|
zwI_GpPAeK>Ap$%B0W6ObtXGd7*bfcuej*^?GlQ9LdbViIR?zkI{6@+SWJf~X(z1E$
zR!CPveSLLW*bo{-D^s5G{I%AdL%4`{LHozBq@)&@M%tw(D=Lybpm=r1WT}3P9xm^g
zCINb+&Ykb;>x*$z>8reV^Co;#A4prN#kiF5^9P|R60$V6e6Ck;8{hPj3X+Kd{(*tg
z-XXA<6V%n)NtA|6j}8n9@<!wqzxEpE$&!;($E&M{WjtPEZ5<XD*MNPh%na~j5AHfS
zL`_Fl-16k2{LGi|!+hwpN7lae#DA%eQa8KLpQ)}c9}sz1Eqr-UNPzH|$_4-&F7!X&
zHD(Gp@H{K2@n3u%uiqkX^paUHAFHD(DwgAZoL&QXH7M~=dTLo&SxqqGSscF9WAWn^
zXR=de8}?e9NTf144~|laxwH%pn^))ryJ@-+jjh%-D==f8NF~lERX<0{@3@N&WYnlp
z<XDpe>n6y`>e}0%i9OFk%Z!Sy>*8FRQO>HPCcTnJvN`p$C+^xuziui}SdAW;l5SHz
z$Nb5#u#zTSb93>Oe$$=OjGRZk$F?QAy<TrbYo3N=B_!snS7(wp0}mFAj#HJi&3ou=
z$JVZnXwSPjd+p_gvPCanhTBB0EZ)QwpBU`O8Bum$Bbr#KnYs8EXAwujwu7rKB2#uV
zxV7_FSc<Qf_A;{z<QyX<>M<Iivfk@lw%*ZEn1P|+b!>U{lx+v1m7|JL6Td&B`hsl~
zv~5^y%#DRnOpHD<ryR65HbZ0B$P0Dv-*0}@TYkimovcpo8%E71l{Ph3&;J;zhCZrF
z$L!MhAisr64Gqs*>t2$5|1xDt%C9p0#SlGui_Xzlte_HAhtCC-5YX9_rl{7q1@2?W
zSaK5HalE(oF*Pta{9S4~c2sUy(rQ`#L-tNJw&1ouiSj?e4o4SWs=^h0%Eu2$l>Q|F
zf2{`eycmJi=b_6R-`qb~am{AUni{IU_x7<s9-9;#PAB^=6t~D{SVK<xZb87EchhrG
z6eZ`$4%g!7py}&N7DdriCzqMiuW4IaW>oP7k1=$;=fz&3mtN>;P<d2R;-9?tqB>)@
z+y*AZ#2`w%qe#G%sJi&6HrpgJ{ngDF6D_UXXU>dO(@nT=!5zhcZ(KrjboIq`<9n~9
z$>xoQ>!N;xx(~*Mc9TJD->BlqeGX_-OV^yym53PE8-K#c)x*ghTYD5x4#&kG8g~FW
zMzgqroZRM3o6a7tyyYEkv+(;=^&<mT;Sp_`DKA&o#fYL~$P~RJ3*VW%?u*3GVvT&a
z+W6H~SNghj1&qCAq*RKXt$Ly}aVK^7NQm)P!=Io<y*0i4rLmXPo|}yS#6g>R$qrjS
z&fYBd<fd;{9jcD_!751H3Nq)F+^f_2yor2BFh8TbVY~&cK9RddE1a34vpUd<(`_Ia
z1|SQReOPGd=DzDUY)GmqXV^>Ce09EOm$Wh)<_<_Clj??t;+Bsf-(>cY4E^S!wb)gN
z#aXUdbK$Xwn!AUr*SkGzkW(6IcCv-L5r$dYd?QS59vFM+(u0>06T!8CN)MHe5)->W
z`4!`!)YUH<N<>KOIluXUoLlaPmg%C6zQv(&ms_eX#Sd}&wi-EEgK^N_VmX#-R9}k8
zLr*WpJ~^(RRmbR@?2%sa;<c@ZIr9z<7tb3|up<Z~Ii<jlt2pqsB(XR@2cPiYAGV~+
z`KYLX)2EBJ1qP{MX>$W!Cgn8Iu_T+Wq(a3_HF?9*?=V+lLD+Tbu=LfQo--Sa%k~wc
z;Yp1t5Q_DhqgoEFSl5;JH*gGHux-(Mhb>i4MCT78?x6EN!%Fl73E(gWg=iMXlzd{G
zR@GE>9RmX=ZaV1(@}YCAttPSKY|Lx&VqshHIc7o&=0rPy`j}2|TFs?T)5eiy{`obT
z24YY;0cOgmz1C0b*0N@QpvBh=waFB>Tg+uJ?2D9}yl>w=Ue>gB9>;8LLa@j30YjT4
zzDPa{U!*ND+ED_Tf>DUKx`SZWaky0QMvS#&t;NR2LefZ?P0RR@IXv2ZhToE7yDM-2
zU8x$~+#@6!vA3xvLXfv}@-mf$5kxxm4t941m?3^qsN6qs%SQx_sk|3%XU=kSet1$%
zZt7tpo?q^rJM+rFP_h?m=nstgY&Ss=aFjPFAXF@@TCslp!(aUei$+~@O~uYKOs&6K
z!)fuREi1^)&3j*jIQ*KXdvK=gdz{Gs@xBE}uB51VArGSsw{_BwV>8EIJWnic9`lrd
zS=6)~B7y#hNJA^|l2B_Ki7{i8THl!Jft`g7Eo{S4dF|B+gw(aq405Nce;s*OL*2~V
zQNgch)Yz~SCnD$dFBp7OKq(uF(+4fIs8umtcdri2=h8@FY+1Iv!!3Qz=i%}g9T>hU
zF`ua8X>BbzdV1w;pScaj<`UpW4N_*s0bS9H0>N7yV?Ts!D=jH8dE0Zgp1*6_sh!JG
z<Emq4_;q}$G59!qdER7&$f@dj{w%h&8`lj`J0Y<U(e1fQt<%~->!PC%SSin%73#6j
zn7D7{F6vv{*4AeAL?-R~n?AcAmz6=cN9ChxX;zw{u_~uu?Z$5Y(zv*!uDT2bpEh<#
z(8xJEW~TUp;mmcf7qgu>etaa4H#4!{_}4p6`g82=Tf@%QQjEl46d}~2)$U{EPzrhV
z+TD9`z#50k*A{>0*#>P`nmo>LfNDQ9Dbe}T37KDy-D!L?y*O$c?}cvYe%u?|4!i`n
zd#K;ANx1(w`RIUtItv!a+J}qUZyL-!iRuuH-gKvh8s;qDb<_zox>ZNj?VCgmGWpW6
z<cbB)duG;E%p`;T)%?0!gW#J|GW4BSF9kgduorgb3)x%BHbH`LF)s`By92GLqnZDi
zGmi>0^atIiT0=*kTY*kdU`?Dwiy%k1{|2*yFI{>EVj60L#Z7+3v(eL!Y#>2oA6oTj
zZEJ)3%a%NK;nF2R66D<dHx!*o*rk3DELFR*03BH;)LMw9);l}1*l+hOJ6oV;aEQ0t
z=H<0Hh4G#j2}_{aRH>w1B<%tDHOVDefR(;qgYs(x5w+b2KmoxHAk7;GXp{1Rfd_sJ
z!(!dktj68k*m&^t?YCBL@Cw~7_qjaop#P;v^}}8Udf$t*5$}dt8CfzRsP4Ta7=g(*
z*|lrewn|C8iP#Sr^TgwqSMg}7LMz)_z)iqHCO){CF4^I3h4ildpqtxPEa&|{&LViv
zJ3mIw)5j$6!)!p5eL=?^K^DqvMJA%@&RDw&zye!Gs~~zU0@W%P!KkL#-*4d4u2Y<*
zj=7ujfz0dI1Ji=HlUx&S<NOdvvaGGT{Y;3D2b{~1w=y=qn`v-qBANI<&aljYGlC@?
zmxbMZDPjDCs&uKh<K-Wf;q}TG-AtW&Jk?G!k&gzv=Q{T$-9-f|E2TOjBG9<+Qm-4@
zCkpb^e#gNXYvt8XXw-{s^E@%A{T!C>f{B{-NT;!y9-oli2|4K38HrnBz3*Nqy%&D-
z?{fTIs{Y!e;^y`lhc08E=?}^WcRvvl!W4;R^)66e^5ORlXSIKLKD|7O<nqk#kk${U
zGCePCa$OW<df0H{%PS7u+zg;isF!2S^wzJJRr~7&P$QMi=Dzq;tS4QmXdH5{jyPE3
ze16eN<!Sbp!5%)=+~jB`e5PuIg@?N}ML?`9T3@4JHcdr^aqmN$w2_z+MUMITofoed
zdahh1OBU43VZek7xivzuiPZC$ol!_gx{alsol@&@!->$`6_oHic*#M7?r6BpR#0#f
zV8kQL<zCS<4WZv{+Mdi*Pmb&D9Dn3%c*nuwva&T3txP`TXIB`!(ch!07y87=wwxAH
zlI;2>{1S~Zcf4GPdTvLgLA$Ey#dq^`Kh8FN{;HueL+w}bnww5Q<E8rV7pQ6s1h<GE
zaPsfJyHvWZxU3W<;l5=Ie5$~QDW_mCe_-!?ua?taVdM(JwWBE03x@TOmDVdsQCh&p
z#Aq%c*?3|=OHsDNt>u}Dtt~CDBQ}@a_VICE-bTS-He;zCVhcg&?=9qsl#85?XXl1y
zvPe1>ZE=Z}nKNf6>mTBLu;ltFO5?}ht*ngL?sNr4iE`6F^-I(3Yo7r*je506i~gK+
z_4;+Ou@mlm*ue~d1A*8G4xa9-FolM<_9l{}$m9?8VI=RE_#CRYNnD;*^6Hfj7(C!B
zYryRC;78}OhU4%gOp4^Az(Sx>RFsunu`h>dRl2{%hvbu~q{V$UKG}3YXrps1-(g~6
zf>OS6BbH<_(a~o+1P^OMc=YY}x?LIuuH9iidtbD4)!msUpXMAU4%zb<h4RB?7NiLP
zQ(Ro!eM!#%Ym^$jy`_1U-MV(Q92xq3tM9|r%2m(0yKRv^G9swjw&fFtz!{&8SxtYZ
zKApi-Pf%g4_ft%>pFUNaF=H!2&_j(%ftu#N_>~GaPt@h+9;M@TDiy`jN;-I2Cx_-x
z7|n#5Uwd&?9)g3Di9V}Wza_N2_Zg|OUP$T>hNKSJyle<k&e$s*>3YkSJ$d|i>Xa$R
z+9@GJzJBqWxpsL_=$yEtMx2EqxvQ^|!7uwqCT4IV#C>~xTp}5gFmtz3ZMWKwUPCLr
z1~ttZ)SV`lZS;$=7|>f1i_yZatvi17^x;Keu{)l2;_dN3jlFl%(y(IA{eF)H5AQ8Y
zHbxd{bDrAHK`)aRqUGRNKO5{|HD_PJ!Ey9lz)$%bb#+#%D%xYT^{?IIsBBX+bl=aW
zjk@Kuw^+mM53g?j#vmuN*Bhm~Qi1&zC*uwlaq^^W&C#xjl6A97e|<dC6~Y^&H-HVB
zP&|p3=J|~Eo~6z3+ctoN3y*Amd2AIro9oXmmG-l%=`ritC51T)we|P!^QyG$bt$+n
z58*k~1h8UE<b4V`oUDYA$er+w;1Y^aW;0cLi{08Q(QkV!UA!1nq>U~?=_OSKg@waJ
zqb$DK=I7={{-&SvT*`x-oKJSO2UYa8y{>;_JHTM|w}dTo@i}(Nd<hn{p!WCGO_SCP
zJvTJRYplQWl$Vq#9$~Ya=9$)t=vC!m#%no$(B{mg%W8;yFDaE>ZhfDK3_RM9{ey6p
z%gKLHq_M%erYBa@Kr+OX!`-J)^#ITn7G@@X6ep>m$Sc%OkBy8>!nogAf+(v58#*Sn
zoz|A=>#R3@7Lz-s5uh!5?dPP62r;%u{9@adEuLm<Zl3cq;qo4bEo+u&yH`MduUxsZ
zqv>}CtA}Y&oyd&K#NOv942bcd(I8+Zb8<^G21?l`6ZR$}abYdKshoA_+Xm}584G=j
zt$u0N_A*)jeB%l8m??gjFBa_q#6W;1kWDVRf0*`@arp=0l8UsE%y_0KlzW|BWT87h
zVOVZPR+{Mhx;mskUi8uOp@w}IfA-=9rgg`Dyp9ho(uQeyEYKi}4ae(UWykX=C>H0!
zo6ygY(hNsIXB|qPDHm#)N5~&OCLSjo@UUasCNw8D&SqJZs-Dp`89vP4|G*xrg_WUW
zO`U)&(TQ=^ue3E_T3*D=%`3@pIQ1!j%_hD&ynvt5Nekoce_F)&t&~xyM?&6s7pe2h
z?<x4HRg9a{-&~5Ey?lrnAO!vXr|wsJ%^f4<(xu5aE^v~RN}ro~<4?(d7kT<@v@%j^
z{t^D2<@64zp1Vy%1Xc6;)vnr<>Uq#&bDkTskdGb}I^rm63)2O@6Zhcj;Ju$OFNLcX
zEgZG-!u<<)&$7D2D;f*#rxq+ddgI2{dGqeP8<qJ;W#YtrN(}UzP_p*pB>TTNKmWa;
zV7Vb8>$1vUlnPr0-F@-GdiFr~gl~H_2BLJ~U2rw(>QS6U2N@nP9F>?Ng{7&)wX+#t
z6&JUlNEq_y_M6xLINo%MPx)Z~%<JUzbpsUC!fxN5>v1ySxR3U`$jr>;(vI<3=)DDg
z3qXuS#WH0(Ni6x7Fz%MDb@Qb@!ZgvGoQplJb$<>`R|^CUGr80RFxQ_XhlVJihwZF6
za*|Y!1@iLpsT*3ncpB%>gj7gu0|24~yBEIJ=>F{+ou|js9a2%nX1e-cL&lga>TNhU
zNH?^5lxPFQ7?BdBR<MzWfX3k)@LGM9Hkl)?Sqx>^NyDYg^bLnFzZuC)gPTVKSfn(*
z6@$ZFgp;Au!NMYBjDb6ox+A+EHq6U(D!7fgSnd1waxGSay5C(~5u*i!3DX<ju+^;p
zi4H~#U{rn2Z^~qw8-7JE6^?p*JN@v$+$ymYlYI`NN8ZJ{o(j>=@r+e1xXTv1DzuyD
zhtNcM+voQj>_BMy07EG#@RPq|n59i7&`64SB-w~&f~Ue)3j8L}14Ha423T2IKK_0-
z+Dmb=)U`IVtdf$eVB;0z21MkAW<+RAm@g|nF(hAGTU+MF`Lh#NxssBN6VY2R-}LM*
zrO9I=tB-DddwJVeJtlJB=DD|d2z-f2)pA11D)S452BlkSf4$Wh2(rO5S;6QWAOd`#
z5xNN?-*g!$DpJFAQh0(@>C=69ltLjs@p=Z&T?}f=`3ge~5?@zeUlIvuRIX?Bca(LL
zWP{?Wj_n_wO>2K=EgH1c&Ed^2J;K3y7z%&YhWGCi=pz*Bir{nS&Z#<;>sE}8=oRGj
z@<t=cI>!v?=i<eM`T0yTVpv`e_o!)8rZmw35O0f(iRU|oXgy*>@8ZXgzkPY>LOXx1
zZKJuqdISoP9gMZ97`TTzg6kZXn$`aF&*>n!xA?g*&62LCIr9#t%}`HywD{Q8$i&vY
zrElI$g`6SW@zs4k<r%Z(f1uq%rc$bVicie5qtvloyyz8^4qaW9+rE-y3Uh-a|57xS
zd@c`Ed^u0KH3Qw}gMz6KGh54^J{_)q#`ZHn7#*Mrx&3~n%f%2?Y2a|k(SU^0E3`=*
z;_^TfeNpu6CT+*YU!TqA2%`y^xShB_|Kh5k=_)FJy}G68w93%1pNGj-0i9nHqyu$;
zWx<t;1++;3Z(q@`Z)i}sZEhFDonP2%-3G0Hc6g;cZGOACw%S9qN#gs`fuAq6QMIA}
ziY`C!iO2ld=d)6%QNubrj)42Wj|L><N9+gcTzqos-`20EDkwA%;Q8RVA)Xski6bkx
zGw@g0mR3l}WmArlbOSu1h}+vWie4WR`S?$^6ZHJ&G+u4rUQnb>%A2O$=G5244{LDF
zTC4AQ^t*u}{iJ^0mJj7GGcn}b{;gx+e#{bHzf4xP$f>9}wf*-uYPkyc#pfsQj~f6Z
zbT5w7u3~CCV95PtU*j#4cB?%vri7q>{?a_<V7SL^9|kwtmv|B3qn(bI`l2^Ls$-hj
zPjkHOvSk4pyF$jSyfexp#~@EjPjw#h6WVt;T$#>rKID)(ba{})C(#Vg7OA#HNTX@-
z85k*{&e6Pivx&SP?(An$7yG|OSrC2O%1s9o)_L}yb;s_oo}TJvOQT*~Y@S7eb#~Yw
z>zYZykYrhv-{=8$9)8BP<EPibplH!1*BBJ972QW?Qg6dF+21QR8!xS=WO-$m#<%O~
zd%VgViGCI9OL17$x%DEp1k25j3t<3Qe<2~CJKmamtW5S#jhnLL%ky=r&v(6egGO)N
zx<$Wi!8tX&=RbrbdjI6|i4(=(afOP@{_{L9Mh@~3h1SrTdqXuI+6z<hHXvtenWZl0
zmqo{&TCQ|^Yolaa%CTdQ5E|-aSq{+{$W0NOf^n<Dwk**GTU;y8|GvB|Vn6w{+dK8>
z*w1H4G{DhngVDAg9I?vt(T_~c?z_ifvq)2}wD~#t%!vJP7;>7u{;+@avX#HJbQ#R^
z7fL>-sNmpo8s(iXXfFSK;A8)$QAU)!kz>zrYjE*s&UwsfCv=ZmYJEmh@logb?el-)
z{W!m(f4D(X$B=uM7_Y!Z;bY$WR*?%QPfDEFh9Wx6P0_0E^#;}HI;G|1TahZ33KM8z
zCBGHu9NV_l&F!DsjpxnG&2j!Th#^OAVrfhD;D}iKV6^4a6`WqJ&rYHCaZJ#PaR$|h
zS0uZzFslg;4u-fBMNPYE(Z{R0^fw|0cqcvw$>D+p-90j(nY-=}j)n|wJ$K?dKXbD<
zxjKy)z+S<j-D~bo+Qupr-r>S2Ur_|O0783Wb$n1zP+#%M)kkj#Hpq;icRcIceZJ5n
zIlHZqSjCRZtE4blUiArv==BVf*TsS#Jc9jI8{sb|dN^jo5ulg18j0k}>C;Y9mU8xo
z<;XPq&r8b?oW8$p5pOwu+67lHukU2l?|m3Wm!mQI^qkKOx^jE9{vu{P2gaV5HFbG=
zu0`mRK`pnhN!`~NDAYTyX%8ho&L;VPoYoZjPoSBCs%p>}>->z4E{!ljlzYcb-hoPG
zhulR`An{9$W#a03&SKUa&*ro33&)s-7*1?>k*wepC8{wH7088^2ZID7{Es=pqn+(}
zJi{CR+KPZER5;c?Q@w1yk?<UTwH^3{Od&Zr83IqqW3(<DqUN^ynhkmdiG>Py8nHyB
zdT_EcFg>p^muCIIuSH3DHkJ=E)QBk+l&C>{P2bkkv@9tYW<5M5HT7Fu;w{F8PjSj1
zP=N|?x{KLu5B~8hD0j8{__qEu<=!4FXDGcp^k2nwz`BvDj?$)Up%P6;u)tbgAmy&h
zS@wLkZ>&`BtcMS2^VErs9lP_}-%4q;UOIQ#c6NVousK|rktQAO=NkrZJIV+UMYFTC
zQodPgWOlNf>=4rYW3lx;T*_jHwLK<$>c~6vcQnw~UpZ{WfQwr<Z>AAcEH;)9w<Q;N
z^4gd+moLjzotMm63~jssUiF(@riI|eP~i_%X_hmHe_2?$nGzw>7hlnUh=|2Oa#N-Z
z?APzDZI>ACV1;^`o9D*A{_;FijlB47f}m$FPzIBD8(dskisy_23G=npASI;G07v&v
zS!Ml*_nLBybK&9d8M^a)2cr~%(`jY95nS6NQMW-YL{9EPVo;r6RHF!7fYCpR3-<6R
zy%>zz`h^?@UcuhpzDj%vJ-*ReReO<OzZXq89vU4&Mw5d5M0x>c)1FU0b>Qt%5X))r
zZ04)><_XA8`W=FpZ&+UcYmV#JA2>C!^luMMwcX>GByr)RHrkA7QQP4(QIN&ECQ>Ht
ztIE#G(yHqvz)?zD>r3S#U({t(2eQE2@`Ywlr9q6*teR~Wu}|0k{4xtFe5+^y?F65H
z($;U}48IghIOUkGFfT(~ZpXHQ^s2AK+#>Z@Nn!~AP<BIPuXEOxmWZ@!pS--~ouMmJ
zux6)FYHbTt?dna1qSdx~&6?s%x>UO$#O8kkQWUrSg&gN+OxbxH#FK7SLkUK^CK7dn
zsx2||PZ6kx6zRT|BUZeLn0r1hE(1c49*S5)Y+T@Jy&8UkQ(6sEwpe_2@tHXhU;BJg
z8stq?v35jljYuK-Ym8pv*O@L~zJo-WR_dw4GitrdTPKTNB)RW<dzjhT_Gh$~a1wKJ
z73^oL(D38=#2b&bV_!veM=QqvQ#mqr<a20cqS5I&-eLMBS%Kl<7>v(1?@)h1YLMm*
zPWh>5?c?I&J`7!pkD$XiI4QQVG$z2`-Mnv;R*%5y>guq+`usH5<dVgtt?M^N_7G_v
z+dJx7+&SScZoj#}e>K>f=BFg83G@sV%7mqh4TJ0V2yZw`)TVBh@btHYQWzYZyJ$0_
z89{)}`Js~s6<zEW+`nJHIs?hDSi4QzMl8j%!THMRA>!gWlJBRCSvaXTCN;w{&8FEe
z1mdVtE;tog>M7MSNrFQ`5W^6HRvi2N=9j0b0iA0e6*Z)Z*<bznMLkzAh>9%3%q}&o
zoXES7)#?bd-=m&*Y+|b%IvGIK=91&RI*^*y`jF+!KHtn8bay9HQ*@Dh64o#3Nx%O6
z>6bzdj(@*`|9~~1Hd@1C<@}adTu4ZO0^Dw4VL{7uRkX?leoHuchsDnJI~^RHvt3OW
z!87J2P~3<QQX(ic)>KIxKM)?|JMIARlHT(WjaB!Gi<8<PS;{3#aJ|YiW(0o^F9%KM
z!0tN?w-=mndw%YJ)3xT1(Q7O>+T$HvTpn%j=|cfIy;`UD-eOv{mJWY*+voPT=_ao?
ztjGRi%ZSKVUbP2yYR&XR0N@ilkJ*>mFYPZ74QPcA3JEzJs0S7N=qbg0MOXPgf()v7
zQBq3EyRF8ZDCeyOX8VeX$xws^35%3M(CeR6x`EY<b10p@?VsaeQ;|`&teX^)Kad4a
zQ%U(34({}KQ+NLKY|%eo9nsyqX0!&Jn!j{Rw)=wSo`K913%#a!5+sTIU7(gDRxH>6
z@RzbAXn*k;&FvEdgLJgDqqs$28OXqypI&XN6EpE%oeF{sN)N`uGt+9x>O=3?&%W}e
z%OoPqpQfu-IG<<1B_t%!Btk<JsodT<7rM_EV7c%bft^+qOjO4I?YSpy^d<Zg<iAu5
z+cQ<YbSk~PXO*-iB5XO8hp8Go&CQjYJ^PdrGc{7cRMxKWfFaHllrC#~4*Ess#?5+s
zrNGjh93AskUxl$fF!up>^l_UX{pWdiG{>9PJ$U?B7;FaOxw3y90^@!I2d=xUp|U1V
z%vmOO+g)ar%%A(>oKD|znQhcr?EaV!f43SeIIyqyec3pUfk&fG2An)OlkU9qp0wN$
zDOzUC<ZgVwjBLIQ-7BDUfimJQoZYQ%@6>d^_V-uyBUZu*l5H;|mm0T{l62XQ5J%r)
z%BSq%a&P*}21t9+T9>VCpJ<yAoT3E>s^`Ds)7w5SpL)qT%J?by0+l{2oa(;6_yW|r
zSbMOssjzWy4hoP;>So7=;hZ(Cw#?gGoE9^C@2vs#%QA$=`Y1x<LBD_zl9J3dqA=cP
zHBnJ<Bl&ef&=9CJ6h7HbYpVtzrCPmY5moogq*cA#oA%F=I<4ksXKK2j*F(CyA+B@=
z#AZDi<QAielm^-S+VdPg4+zklyi^5C;$*}Y+9$2-VOvI3cv5?Q+~M|{$a#=i(Z}Zy
zA<l^kQ&W_=&Wmv7EW=t3kj^dmf)(T459d!vf9KW6fd_YwjET{^J^GEqme`Y*FMnkD
zowQ9xYg)VhIfqRj_d7x<Z&)e}?tmO9j}Q=0(n7+!aY}2d_Oj4*8zYC>CnGH0<l;`x
zyXlTZ?lDR)9jzmK=gkY1*2@k~d|F<9i}7M9Gg#1i{%<GN9KURyRcCTw?2<}{Epm=m
zx8HktBv4m&=i*=AWDU<JbFNoxpTQwz{2@)DrvnmvTE^$u?whK<)U06E4^mK$e<1Lc
z+24er%gP=<$Sezl4tt;}=&RPijl0L~WXhGm2hV*#X1;caBc&Fmmbhh<>=}N3UAO9e
zJ0+nd0OQaJvxlylM_s*fqv+woK|VqE2TtnP9YRmkN;mSlCZV+YSQ)xmpiUZ5(sfu!
z`w1X9`T1j+nAci>!>iVe=ajsp8;$`6h<j)aXb+}3?<ju|aB-fOuRFUotFGey37|Bo
zGn!Y!9JYM&2^hbrkFU@!F<`)qY10Nj57C(zaM19Kg_l4x3wOJEnOskJ2@S0q<@f3+
zz~4k&a?<p3UbX68Ug*6GM+LGKd8PlYE8(|p?F_o6zE^hC!r$LoCMc9o3CP`h@t^3W
z3BnD}bbB|zW5yl?mE_UHnD~SQtJkv-aa2`LHlT!VsIsHr!d-}rM>nyafpLPgDqX}y
zBI4((A0c63VMH?aSTFBd&|i65ljKX|&G(%H*bq*QSqzI@#l0t0zBKei-_D00`e0*e
z^oeWNrdM_#S2q2=vc&Rs3J4pMz!(?9OYKicP;{zpK0AMp=hCIIqETRm*Q!<m*c1C_
z&73)P=_`iG^c_8Q*uW^kO=`(5R+M&|w&h_ToWp6x>m{|ZD%okc)AKvHpBqm=3ES8i
zK8D}!nN|0vmi%UKcB`pOG%CPv)V|EU!go8E23MBeLVWlI@84`L+`%@J<^eN?N22X|
z_wIp4ND?=igK=o$75<=FanpFF-^X|FCaI_{@^)yY<h?4I;#j7iH0d%yGIF0_bKpKA
ztL~@SMj(*YfraQQ?oXN-Af2|hP%^SoRYe62y|(}AAR#PF!SCj3tcKdv$$QmW@5Pas
zYVBz<{kHo-b~d2wlJg@_u)~+H{IVtcULBDvbGWV;EMesHZidXE%Is^n*&A@a!t42Y
z$*elu9zeR{qYg{kPmCM8Ffl%U_0jfqgN02Vdi?mE@cKX|Oq-*F85E;GT6lMCiv$_#
zob^Y~#l*~_#;-~VI`?UuJ0}28x95`w*J#P%z`UI|{z854s9xQ>_m<IdfZ?>Hqr~aF
z{1nyBl}elRgq2bV%D`!2nvKN<L?O%xNVY(PGR%2jR=t+6bU6(G#`M0gM9yw6`>YU!
zyv)b+SJ$W9{&^KmSK6Ueqe6-uault$snfT`d{H5M+z$^6Ejc^CTfN2gys7HBAqF8i
zM<yRWF5%*zyBae3pTY~;u}XeR{NI90(#SThv>%Li{qk|s{FYqMxpD2<(AX>R*YVSu
zGKEoXpl>yC=MxfWHS%;`w~$-#Q7R4~i+*OQ3{y3~)3|IY%Pp1`UXWZ|@Q|0cpSF7(
zo|NAy_9M5IiU~P~y!;^zp92<D44ho@u1ig8=GUW>L%wNXl~|IUEkgpHCu&xerCxQ@
zc$^<g@+m<1!P1ypQ`J4J>zQ<W_Syf&-kXPGy|?Y+S`Er7DnywYh!AN|8B<Znm?3kK
zL?R+HEhQOJ<|Ii{+?j{WtcqmLJd>%+Q|9q^e%AB6zhm#?J@&EppL=-w>v<OL`}_S2
z*Kl6vbzbzEAgKJCqYKu-Ckg~FM|yj|<emLqTT6(x_#;jU58QVMH2t*S1sDrRSB;SJ
zA+m2UsmBeAHg`@lfZc?|f)r?AYS2|t>a@ko(C{c|vxH})hQi-V&QiQPKCZU>iE4UA
zhUeDZ)|lx-@tr;*l0il#qfA&K|0m)Nq$NK`R(V17Ak)MutR~4x1LF^8h7dGw8o*GO
zuIbGMgXhZcKLtK`;2VQ`lSf4QN#E7aioy2mV|LD3KtIu@=>TwncU&wEfu92mB`Oan
z)?_y=gSCS%L`jNCspCul+Ns6nvVa}*IH?~xaG;kcsG-AWPP7y8H!3;Qn>KluNYr-#
z#I42cB<fkB=>|$l<Cu5RoMZdpTWSS=JLg$I5$djDoV<d9l5k0A$jK6Xhn`p1XtZb~
z_U*%oJf`{R?c=phiY(if=*vKAlkir6U95Yh(8+`c*q@p45E?@0noE6mB|A5_ZujB0
z4`C+BykemFa{>qVvC6#v$m6hw5*s^TxVBV?kmks(o6X5lP+B!K+^;Sp#~k7K4hdLa
zr&p2&v-qJygMgYMP|?fJ7ayP3y>)9Cvr{GnUdUa(MlJ+6l=%P22ajXKG@7FcqM^3?
zpAZmU8q-=7fBh8|*@Vhav1ZUUeCa8)*B~>_9(G5Ek-dWh#wA&v&Huv|Px&<`DV=b@
zf|4xo$+A%yXS^uHw!yEUY5N*tth_(N(k;8t4>E{GvYy~rL2L_#%%H@F+Gf_=GTfdR
zBx6wD;2q}}Z_9k@)Q)*TkH^(x{#D$~L0XsBH3{E&>{n&I+@Hhda|aI{GDUufb2d9M
z@k39?qZWOr7+MGHM|0scXTvU;YOqul76H~XYMbe}Y$u%!klU&^J{rL3QRRbC81FRW
z3}MEic0T@44b&tqiW)8>FPdqKI~f`aZno5R_*%O=cecj?06PrQ?P6PLsh6nsZFE$D
z6UDjtGdgduaMSm)bxD^QplCQC7_HwRm?k3CMmrzc?AzP(fSy-exn^!Yf&y4^<bfxg
zrO2%?UYUWB5rCH}iykVipDr9iqy*(p`~z%+9P4?uhBSC#CQw&GSeUOXkeQxtUi)^-
z@3W*ph=Rjfb*BsUMfDDs^^wLtd-D%$-;n|(aMcZ4U%xrCp_nG5`Bh#>5wpDSO=qE$
zfY^N+TV<YbHD1wIxx~MjSC7J=az<LDJayW*ZK>pE*rgjkY13?X%{#9eR*TQIcWq4*
zwJW$gmd&;1=aLMbNej?59)=to-UF7wz?7?Sgl~7|W#iR6*&T1X4=#J55Jps^pdg{8
zr3JYLv$}Vc`o9z(8DvH&$Kr)!@_9}rB9c5uCvJni>kh$ckqI#a!n$ztk8?b<c^D|*
zF!y=?okGkXpq=Jfgpvp6eRmjxfrxvHO)agiw!{YDqJuN>qOSlT{g%RHc~<D^lIE|`
z-3ye>qGag3`1JRDb})9YqL7bg!Wl|HLlgdv_H^Gf&Ljc1p}6<^n@-VskZ?&S`KnY;
zsNIyDYY7zJ(IG=ZF_g(v6T&WUWPEfs3P%A7v=(>_{4X+g3m+5^rp!8vk!q|Ceg6K3
zk=YnHPf8QBO!>SMyl-$x(D(HPZy+c4L_0OE0uceD!=Gh4QFb8}B=>wM9qkD=;J}~5
zMjsvx80)tiNhyKW1LsFFgWRsu!Gw{C9$hV(Q)}uNL;@fm!8-S`CqjS0tU^%A9GbbP
z&|&4hpjeC!Ba|KGR#6C;%V)m{019-zPHfz~AQ_4N<!%Gt_-!6h(Irm~SIqmeL;eL~
z*X{+?uoJg5N~KWn-Q!pP0MtJ+g9>>lmz6qp1RLF_ml24DI07{U1bSj$5R`-9D-tGo
z8T?mpzOL~wpqpi8#|2jZ*S`SdA;_So@xVz2e<9+v3EJXdgR!E|0fFPkc!>)V((-6s
zKzkKW=L6_nkdle~>I$0ZWjc9}adVphfkbuE0v*_R%z&^_mds@M5rDCAeR_l!<|JO|
z?c7A`*vt2zxo#eTFxPyeGisYD2K4NX15qk<Z2J)cVIEkE)k$b*om}wvn3|SGXjo_f
zK{=Tyx51TsIzjDEUIi1PVoA^L*ZfYQWy)h4Op@ik8xNq<0wU9R92{8aUSPdVr{14L
zF>(@4fVt%;I$e*(Te5Qku0|gccn3t1#37Auw5z%-G5JMgCmLNuPo4w^JONlYUMLaE
ziStId7RxP`JPf2T2k5?B_Z+yuy&okP`U3qkTX85g^&R>zD~YTBTpz3eCQW^4-8DF&
zOH-+87Xbz6*Z*pVPwv2kpEwHjk3!DJ&6h=4^1Znvt6vNqSlD8*l%r#j%{63q2d*$q
zVCX)*t=?_)qO^;?icy|PI&$w?py+tsu?A^%kBN*6*ZhyY_@J8c;z<F=sXLvIW?53w
zYUYeoyqxhD^`p*lf2c57gEXCfwo&NAnTEp8LvD_jW#1$QV0e*lo`-F)5K9E_4Pg=G
z2=y7UQ?IvLWcP=}${QYdFRr=~J*%iH^nMEg)GKmxR#{GMjpkgGa{V9iC2V%O?u;+z
zSiseCK;7hk4q_y?ZV#q+v8+GNST*U{Nv~ef)m{E2xprE;nDzNC$TK1q{}ytA!zEp<
z?j@hjOwygmQ~7aQF^*QMUV&Bs0ebTZ?sfMmoHCv+XJCphJlVt8l8RzHW_|Ig&#>Od
z!DDjB9*gp17eTiULnHl%Ma?T}YM8S1cH$}YTC2r=UkYo|ZS`DUsCE<3az8Ng)NR?Q
zLGafftTz=!N7L&RMGvR;hnc+6f>s7)=rk3XT<kAN%OSB0$V^FH$Vz>U=EK^c0+*Ne
z=QnuTL%L4esCc@K#70;w@+QB+0!V)hihtx!ail%25_t?@0LNTn;UTpZzGd|80M7a>
zxeCE*#XSAB)qMcahFbk`v^_$1tdCWLlYOEa?F^`6uEWlNX882?^93i+AUpvQsnRte
zt5F@sN0k0S1>@%!PKYsPJ6o(#T`kEUK0&1%9FV_2_e`6b&E`8bmJ5o}(!yUm?|aDY
zTsSlaiRPy-x-{8FbQt0KU150+@xLP!;6jYQ^FLg|?qp8!UPpTzlx~nRZ{Rd$6`P(5
zh9!<XYAiQTnjvHo&c3q7$@pGy$MN!R%o1aCSZBb)Am`j<O!eCAam}?zGHyYB|GTnr
z*SXw@91$30`P=_27tlx79xv{Y<y_D$@PNu@!*P26?w#Hqa%UqY0h#c-oQLbxKo7u)
zy*xDzVq>79$f1mxBFZY`qgju-@UlBausl{pJK_@mpn7~zhPpWyMlmMmVSel?Bm3P7
zp%T`+mna{o?o$Zi-)!}S8E!zj;LA=bDr#Dx#X_ux3#75{!q~w>P;Twh!c7B`z18C^
zB5_6A!&Izy^0tlM^^;f}m{8q<vh}6gm~!^F#ChLUOJx0Wwp)U<XY%G5?yrBC8Repg
zcl)q5tpjDPJrw_c4+(m`xRWRvqseBk5Cj?VZ^M80A6Ohq{(oJ7|MbKETmQpr-_%i}
z8VMq+n)xBBGck~D^l2+V9trl0x|I;snG>1N3s4KCqH2)dZ8VKtRX1m&IzthINqH){
zYL86?r`4?N%|GK%sJ37Hx60pNMgJ1c{m=h_=fJ0rh&!ece3r?rqoPFNo8TwJmyNKI
zz!aW+cBHx=bPEv%MV{(DZ!nL{c*BP#RJK0EGhjc5I&$aWg+EK8q+tqb(xG@TIH)Mk
z|2+h3@3Vhv2maypuhCxr^&iMA2`Z}pS`VVLM)kjN9F%1!5KmP#MJE<GuZrKh!4-eH
zIEgu?3f~Jsj)G=r;`1JS*6Tgnqu2kv4q9q=F~h!ej`Sb)tb#MbZ>NGED?hmc=}g+2
z(B36%H0dXehz*gYXx85(KV^Gdv+l?JkXNr<KW$`EtG~&ldq)cGdeTpb&{e}<IL5R#
zpsd8lMhWCieD>Z$CcIF#z;v+=0Fgk%dLhK^CES@UaF*p>{P0Z1eu$`_1y|E`ff}vA
z*AY7`l;ExJn@QtVJ`~6qcr!&u0;VZCGR2|c3BF@9nb=L}-n*j^0h4lEtRsH*dJ(Ti
z6{VbQd0hLixkFF@{r88Q?bm4$gG|m@4NXF3HzW7`;n}F~aOkKbkEeofq7?asRy6)>
zIr-+Hv-tV4P8idTls8yZswE-3z=S1vMpV&E=b$ZgYOeh*Uo>H)#?r!Q{9TARXFBMz
z;-OJ}EG4vfwT$bD-Tr4u=ARzruQrrs!vCMOG_;}mA&F72`iM{amLQ}5yW&`Nd>(oJ
zz5V2)+%O<<&i7c=#ey}Y0bny|IuNb?Gi9(1_t#V@(%SK4u+!hJO5`a+Aqoa0ac|fw
z1hHTbD;qbF?~UWas;-tx!7(C6;;#FRx$Pg2bp1C!eZ!Tl=EJCmUp}=RyL$gWw!B9I
zb(sO^K?Vo?^GTfeq&H~CFu3z+qt_DPjQtR0F&iQa#Xx+K)rjq<U3@wCoICf)`dh(B
z{{D90>DJa(IAIiWG>yga!)LsB^SuS21W4Yj0}lu|9wn6-Ke)(<)!p7D&7{YWZVE2H
zx)zx&^p=>!$4M~b2&VmRj{{QxLQE+jOMcPIcS9SbPg1X!2JrH2q>8m55+H7?%dua<
zJpwLqm<ub`Acf=wEAniJm+JKpr#I%HM52uNgZxD^_kZgORBzzcpdcY<%Sfec%l~iJ
z`#*LD|8sZo|LAWVRmE`{F*3tUa<JY(`DC~Lwhw{h9T}Og_<yJcWBPU*6afLOz6CSj
zJhEZGQ$3Hr&`IVYUPXcnHS6liN(>4qM`T5|6cEe%oHN*TS_Yh-i*msCF*2Yha=DKi
z@lyG>&CctIa5fT}Ibcg@6fZ#vt*+JW`Wi7Arf56D(0~pL`M=+<El{}yLgb9up#~29
z!H7$_k4@dq><B@1uuUJlb<Dg#vwLJ@A^rm81^grVJF;5jVa9GmAbUi~`+>sCQ$h_5
z>&^_>jx<juWOnlM^Iw{HxBC_f*Z|L5K*3klTP^5ry_KYwVkjN=svE@<<e;yd#<EL&
z3vr@rX$w%XLiA{dNk<phY}Je<ovE5LFTFgKl5QEFteFzZ;<x7nZ-ex@<?xJ}b$gY;
z0Rrh;5`FT=s>RY|p48@>VmTo%l|o|1x2?0s-bdWi#?r{=YJ%*b?=+f$RZl)DhdCmH
z&-R@g%b)8+qwWyKi-H<^$C#Cv#vx)qCQm><D;kqzDlJ{nZ24+P`(hs^j4v0i&8=w+
z?ZdtLjoI7zzpF%1@rbWqk6W+24Tfm19};0i+i@_&aRLwDF3)BblDyfNqaUUVTw+Bi
zh4C(tjZ@>Z5Tc=~0h6oCzV``QfVc)OtEo*FGa2xs0TbyK3iJ3RQ>5(Z$RIR2W81MV
zC@P5oKZ5mxu+&KmStXm8YezTqGA!&gMD!w8r{4$bse>@VjX#EPBlO!eW>fHeu?JyA
zw~yj9jN`MuH=<%MKL6vvQ}Nkm5QW|dHvp!E?+Dd*9E=IoZ$;%sPT=kuqO*6TVE~T>
z{W%M;!~5PL5nCCxbQ=bO+0IsdsuzWAU!WMOY-7Qi`dSzV&BFj!066s;-U5Z27xS;)
zkTmJx7J5HquNXD|QH_iqy2=4-e1FZl`I;xt1Kyj%5UwFJagoy(cW5JzL=6tuSm(XA
z+b28ijohtzfPpEX#6{2~JT3GG&;)_qC-Zys0XeC?)^$bNiGvNDn9+Tntf@%yaKscb
zy&Uijqf?OT0(t+O;Y22V`)erKT8TPO7Bhj06AAR<nkU<?+zcE+oV|LKsO4?H<B1k<
zl)cRWRSYs}>RNT!Hf5MK@Q&#NzHr7u0g(aRs7v{R6Ob$jyZ^V_!c3Bz9_g<N2Oxzc
zybHZ$`LwB6TVNVXEhdp*8@1p11t3cw1V!A%#iy>*(OGN+H!&^?Lb#4&>Eu*-8|bA)
z?&!znSy%>SyDY%{P?iPQk~8u7PXaHsM1k~0j6{Ek4#~xHm$*W8YvEmDzv4^~Jw>l!
zQ>+B`b5ENDv^N1#KXhZofE>i@5N@P@`Vrd;Y<>;f>64p*;63I^O)o?wdLeAd?lh8u
zym#vJpMt(^?|~$e#eia?-YR)n<qk$_3{y3vd{}><7dw^!vr*!rQ&C<HeOCEm_OWN@
zXcyep#@D#)1^uSLczI>t<K<Nn5A)ORVqo9xq0Dm}C^c-!2G4KR)fk0!ek7ULiS||0
zAmwzSWn*MG2V%TP&*KASvu_Ey6g!X2**MBL-R0P*Wx^!^Fud2Db`Ardn_{~yMyXgI
zL!a;Y51T*Ah?1&GQaUi5m>iz_ZogaTs8)>tlH)M=<&yL5Evk@Pp(wG?hZWt%CUSg1
z5828QObchR$ci$*kT=_UYi;tW8>fCcrd0Ie^h2#oNo@&kIx;krscjALg5nXxWUX-7
zMs2uj(?BjcRjgTYAZc5JA|muY3ejhc<qV^Jhu*G^B-B%-+?w+{uynB&xQP=S2qibS
zP9q{v`|%gl0EHmf4=!@>y%L;jfWQMxh4a#;cF>;df+2bl67sR>6$Yx?1805nQwyg)
z`?|t&9t_?3R#CBa50Tlm(xmLhthJq7B~0nl3Mwi!XeKQeGk=DPKB>Rl?ijv)4~AfN
zJv_$zQ{3X>+tpQ7v@Zg5HzBmK2f6_b6!o7w(777Eck<~6_?^=016+Qme?CV?kj#m%
zl?7mZK(O$~Ypb3rb^NU>BoqVVIU51t?9L=(UV!Ny{++x95^ehi-lf8(C-@$~O;=7J
z@v)@g_!})l*v?|7+)R_Hv8-M{u<nXF)RdJ8=z`HL_YitR9mY{v;3cG4-%G9>TDT%s
z3j^2XIGlh&voF3M(nL}nhME`G`T~`D5{q`GC(7eLRiFO#vU&E`D<^@m#UrJ$i0r?=
z5GIqrTE1TT3PMNkM~qt9hX`%T525QB+r6+nZXa+yzVq?{)=i9!PRE=%x7A;6Jr>%j
z*JS}vDxK~R!fEJ6t{^_1_nQ{Vzv$`~wZM-=iN0No_##Sf_>KXT3&{gE>E5k`-(s}6
zFo{RLkio#jBso)mIYk^1nl@_EeQmw80lsKw0?&g3plV9^yGvm^?JC2(ciJ32iz)R0
ziTTE}h;N`KTa>xsLp&k^wR)8*0Qm9o@y)X^CD;ng5<itH1mwB>ZuJUIVh^n?(vF~^
z?`W^tji?u;Hx3Jw7)p#o^+M37-?s4F$j>XnWCil&U+;mn^3&m01e54zSc{!)^JoaW
zApVsTLJKSXf^J?;uTfNx7tBxpLy+mQ)8p3)UY0FLSN?WA1Hb5``PYuoxc9I2WUh33
za`=h<%76X(2TbbI-11WHNL`Ap-WVe4=JE9a>$lHUy#yf=@wbneE?ysa6Mft1dwQt*
zve4~0&hycv#X<lo4GS<ZMG?igS9ZA7E;xq9%)$;^H1wdubt8s-4}5YEJq&9hfL^|v
z1Z?IV;vvO}wpUjb0x&xDDm;A3D0(+&wJXFmKrZA<%G+soxo>&;VyR&1R4>O`7PwXZ
zhe=X~L`5wVtJtLPbU4oX04RXJfLG<ipc<VV#pFD)qF5z4(Ho}4F1&KUn{!j2V%EVK
zGkEtOU7%EA5EHZAWf(l(zU`16gqQWYEy&NWn|@Qp{j>W=GV`|doFIx0H|1>GeI*!Z
zP?;(cUi}@p=KG`r(LKrBzJ+AXvAMwWz{7>9Wv`JG1eUfGG#8_jpj=0w3XGOho`(Cz
z^mA?rML^OiuW>G@R7nKAmV%a=tWiSMXqY-AwMW>TIroP>4+Gl9b=o(UeZeCzWQt4W
zvGl{?{MnigG8HHF7LT*2E@KFsjk(ix3qqcWgh5Vmk;{D}4<G$bF`|`lRu!OtvsS9)
zy3=N@frcsbkunc%XEN*(B@UMBzS<7r;Wl(sq`^%AU`)vC9kU&snev>Mx?BYy8u3Bg
zBS!?1Ms}tMdrsZ*6J;bt1H>cpt2`y(=OUP3w`+Joc8mGQ4_G!i*Yqj0yda=7Xjed#
zhOjaI#ZkQLKMBxs3A#<(h4t(@Z2LJRp`K$gq>XSWi!b)!#=}+Y$K}N}+}{?Y3oW9s
zf(dvl&9m|X*>C$etc<2r6{N}aBQ*dx#}W%|M4gCwu^JJ}H~VdH)-s^tJob+FN(fMu
zf1NQm1e;c%om!pLgNq`>>|gVwmGuF^N#U_KjjH4Py-tww$=3Xs6*dnhI|4Wq>q!bP
zbGsFH4p%3r?+227XW+^RkXdpEQQSB+Swv6p-`ZWeH&D+5(}cEl5#<>TEiSz3TW@4_
zt46Ras21wBE8sW4rLYDqA&?^hX4sR}`l4DdtWEDdQp$VqbvOEz+Am>(uINpnxiVq6
z4*Zt%2}Fm>K-GOj8CT$O6_Y8K-VzH2_f-KFX2*23+iDsd1u8E*=RO(F0`;SIaUl&_
z8$t!~7|C4(194lxjd>0)XIBTP_cyw|5Sch8JwS8(@W(a_9RzqIwVfb^Uu)NMT_`vB
zh<NBD@seY?<9@m2*yOw>=>|SoDXC8+ux=wG1xb>eGFlox5P&gWo|4PmfEc`Uw*%#u
zBxQcY4xt6jM=3g)W4OE4{lS`>Y_kvq$>*|@IU`H^kqR$|f=k_r>_AEfq||f_1{593
ziUtoQMF+PEnCJB0>*?a5osNVe4oA~Gl7P^N@lr^QaqY(gI>cfeEzJlVars9d(knjJ
zsld0ThEiV+d*wJ3*SviVw(AyZQfj^MoF)~|{q57?iwhC#7Ou`D^;)ca;aI(c=mA1&
zq1mmf7)L{b!$tJZk_#*4f+EO{IUk}EJuB%oc4HG2^*batzv?e5;RP+*$xg1UjJvu0
zim-#mtJ7ppG5i@nnokkP!1m)9?Y0s8o{VTp90wOqd3)p?)C&Z!Bwq!K0~&;+1q+u|
z5yIS{MHRD|tS&ED=1t0&a%gYl)vE1-OYFQ=ef`}VDB(&td)D&cQ$bVHbca>I4PQ)h
zTe+3gvHuCP&>H|mQOZ0Ju|a|f8#dUQF$~`oY@dXP3>}3pys_~A02gP{5!=JVfq1#l
zZ1i;75V64MR6t!LdTVc&ZIj-wmv<F$!HrJB(!I1UJM{jgwOr2*RNXDJx*w?}4D1vJ
z-68+$03an-6r^TF>Jbj@NFR)i_LDn!j?UvOvEimh(^>N0IX@yyMvjchUa9G#h#Zoi
zref-G5qf>!0c`_@k#KaA0lFgr0Rb*Uu;M!N!9d+JT&Dn+3q3bpOL-`&qd={GS&8uq
zybh%DB-WOHt1;oj&NRS$opHvZvmUwn_X9eKE4y#mK>^@5ax-Kr=S=M?s;dhTX>+(t
z*U$*YZG}+WE9`8MOMq65vB6}O=r}7<Y2?b6Z1-W8)ehlYZUD-ji!TR&KEH+jV9>pd
z!N_oCAF)^v={t83XXM@>Uc=+s1#sL2c8FWOR2R9bMv>!H?+J27eutuaT&f{&H6)>g
z5nUn3MTg+Qh75d#azi}SiJgRy@)PLbR?`?hRHE1SSPt?k4AQE1K*&Z$bLv>;&%h;v
zb9Ip20iF{14rvZWY%GZ^5%uws(T_NP!~x*6;V&%8n$A6iAbk%qM5qFZjy1YC@{-Yo
zsMAY8_?^V!c(ynJ0u2U{g;RyR`D$+0%9vf@k{;4|H1ani-#!Nh0B)C}l2R8OkcGt<
z!jHnMpWNYdgYWghj!l_JE@|;Fh>RqE5rM;lK}V8|vx3+L9Y7>`!vHVWsXxiMv0vn~
z&6hw~bH5jM<l1MH#tqw$<Grld;r2Vc>m}}~)k$&Kz*67sG#;;97YD(?I#OCCh2y;^
zeGo}G8r8jT3xY|_%Uf1W{(*m`sYw1+;lnf%b;BU3Uqx`mX34i)&+BK=gaDRz7Vif*
zI4VjuS_NnkoiwN?7nYWn6VH6T95#bs1!@iJx0km5k7^;rHIsk;1(5x-4F9=U^uNmj
z@kE>FqTwVTm<yv04V7{n8l^v?&!TzuR#H9^zodjkRIP^0fs>L{l%Su7gM?%xe?B)N
zFic7e6;<%i>j$*nZJmHeAQ-LoPi+kv(&0lpsb1$K!(x$t|9wgF|M34%cF}g-h&)&W
z&m$t(XgpA?yZd<y_uHJ8>#IOUDO?)M{T70PBAVfksn1Y3Wx!zL>K{$*htIeT+@CV^
z7iA8y1Pm1TKd?s}q{1xl>;5sA1pJ$LiVS}QxE9Uuy4j&{^gO|u?bjiU7j!95upv7Q
zZRD;286>ughheBa`!~Q#RlTaexC8|)(ZYsaM7m5?yLPKprddh&TjC}V_+cm;0f}JM
zhGyeT5f4BX6q8jt;YtD0Kw}Mhp`&U{GX^9kM(P7xjCAzFyTE|mMw_<7rgm64w0+_8
zUVuD=@RJ-EL_OfKg}FdU8#VU=ka_Nu12h8yapjfctR$Q{krT&n($LU=Th)q(g~vJ7
zAgyu@obNm!Q_4IxD>F?<dxabNS+5YV2w@?d?VWNsRc@KCH8j0K(OC$*O*{>Ox)7T@
z+0!umq8F4A`jGJOTzmnZ5;O!U1k@7K-_Hd)mn5A*Ld$;S>Y3~1Ku6cYyO!U1ms3%_
zQR`wRV=iG4;c`K*!ZZw7+tbesLG4n${)nB%gQRX*?`OasdkzLuyA(3@K#W(ld5qqx
zcuX~<AiDMI_uVV|71Q_F;Rs27{mFeJ@`~<RV1Q^G$7PuI=1YXfk)q*)fQSpw3XDFE
z0PsYpJJvvdBGAw1*kT;YhLAaq*Andq$aDZi((U5Ek_-<On*cCS1mFN9fUQA>MN7(!
zq-aJaCTty<di@G?3?Nb27%db_i^y)gF`*5~|J@pCrUo<;e^SPOE9`#60yZ?8g#+dk
z6~&tx{f90<m&;rS!N_GLvsM~4p~zl_2oQlLPJ*{{i|5%WBfrx(ptnOp8srL>)!9d=
zNrWMNV*tl9Z>CYFcUBih0nU-FjT{^#`rN0AU<ME>yx9Yb5$><zB56Ewk^~*d^OXQS
zLcz*^6mKRD#tSN35-8dsn@|g8YGqQilp{{>3GhuI^(2A8KFU}S-X0Jv0%`zaPBMRA
z1n8|FFs)&m+jKMsP%MEl0mKn32xH(*KOaWft(u^=C`uk>Kw$E`=jxVs<+oc>xjy+Z
zzCo_WRbmFWZ_M$b73fAO{sx4<;{cM?np6Q_vR1jkGrt_trsc<(M;LC=H+~f_1I@sm
zdxUoS-Exe8P(OH>^lVf{1aLcV^ot!_#a+2gm{t0c{lccTeSiQi$$H>8kRw&kcSPgH
zjh}$Uh}I&nz&L(Nx{7;R_xt4id~plh8200>@ZNU-BH?O@l&%}Dk0TW_jOslYn|$5z
zYM(yhW~Ds&ev(VJQy)dlN#QpF+9SEtQXR0x4ksk_PWNX@{A~n0fvZgbW<E@PXEeq&
z;hI6;?x_ZX!7zCmtdU@ba_e;RGD*W1J#ruh^gu*%>e~S;H|ksUu_*{wHe4`8C^e#w
z*fDtG{`^yLUn6a5Wt~>&TM;2CNi1-3F>0*Uh3~}sk)Q_Vsi28F4WvJCtMp8(@4_}R
zr&IMRoa_j?T7r;vFJ(<#Q%}91L|evRvtBw3U@4kl*LIyS;AWs9fcHOz4CZ|8-050*
z`Bx9|NijJgJ%&P#L+#e(hgAw6Y>}|$d|ZQj|LVtxfv^4yf)|H+l?xkJ+=~Bsg93>-
zP6XBLkT9EKguO}gt_z$9m!2XhJIqB18$o_~M{h<c4<OSLggC5npP=icPpqCkvfDBT
ziP<+IMFYUpeImZ2;+lr_1F<w8M^)w44zzpbBZ5`e=tOaaUSb+csApV~aHd;cfIt}J
z;PA^RMIeun!FZ6En8ScPV(%*Z!jNEJ0)FrqPE$?-b4H?q2bPvqQgTBsM9KPx?q-O~
zOd$9vTQ~z%m$Y1zG2o6jZVNsogV@cl6;sIWqHxsjRAjtQ=DoL9&0j<DjtMWQi^_mH
zPS;#GpaWOzR>~C`bS)MaUJ?d`QH`lTI3hwE#s|ijtFRpCu;Kcavddt-A>N*ju<Mj{
z4TU8lIZ6uCye1!r?s3iUCy~G9Bj~5MA$aW@ZP&G>E}R<8tJ{{mVH(DfRM8rA>cesY
zMHNISkBf(NCmcg;06GvEH{@@3cI5$BK%zgd<hFQui1Ns#8cYL8K$hP^fY=e*lri%V
zu6Q_R;BIV?c7UMl@0WM0yz~nAFM^B(q;3|rNX<;$p8~?07ZO;gUiXc=tvU6x^_(nR
zP;hTK?8G?p5<IdlB*FaxG*mpILLzE*IGJQ>`T-p<^|b&?r9PR3F_lSJxJYToh#S~A
zb#o`MaY(`jgeOaL9ZpAYO4Fcb0N~Owy!?~f)!)7ipwt4-)>E}168R0*N^7uq<4xQM
z5agCX*nx51VE4!TQr8-LMu7P<pqTLCb2xKwr&3MEFr6S09{+UM?WxF6Z0v~wD9C77
zm;78`248_E%f_z-aKNPGqBspC@b^IUQ@;_!1!*4=bIgzllPfK!4Su6#&fvRdzV^TN
z5+kBkj-rdo&zknO<QF^`3D*}O|3>jod9XZtis+TZW}A(SECo<1vZ*}L6t9z+C}-AA
z4V&72=gYQ#01ASn5W*c!H9&))<)Ss7H4p;X;p?p#nD?OQP`_mjPLabd%=RB<U6eK-
z505=)!*BMuE!IVbIQ$Lcg`$sMKc6>EGM=44%pX_$M8}$8^fBV(21ePz3*#+#KUS0b
zZcY7D4(#GPAY>MOBrT!<K*@1I2!LRKR6&}TzKKu#y&iJGi6b;$XhjBI->o|R3mc(_
ze3hg#V9!FZ-W;ofyHD7B?h;f=fYc|GQK@qK8D@f|N1GA_y&4e01Ss~){{p+0_COxg
zn`#X3f*Ybem<DP1e!U-dQ3MQw8pW4mJlMr!xQ`pkzoQoS$qmR&V)oA$0UYv)PbLoU
z<ut<AsThbN0vnG{tK$8AA7vsGCQp{>G4;<h6|SDR7h(r9Yk(puDXHN^A(utW^)&zV
zT_jgMAKvlG%}C}Sngkq7Ft-#Up5O*F7m|Ju#F{3`vR9J#g8U1Y4+iNiU=x8HZlt28
zcTelIps3GE2BBITtWJea`EH>KDp}o4bdZTy4!Quj-b?iFrT}dQrWp<)WfYmvRH~xF
zr~v`;ja@kDI7>(ZbFfDQPI8EylI+G$hzJOdFD~Z_H0DBnvl4N2d^PzD1g_;`YFHrL
zW7KJm!5&QcURES01SgZe1@n8#5z_{nRG<!KP0Y~;z5-M!;&rp9aMhn<OYs<d&IJ|N
zUs4u8&<F4)`1V6wUjk&~1H<r3_!!jq`96N`%ZKBE|4?~Ct{yJ;(gRwN*uD}ZrV;=M
zxnJ2Es%0X+O;r0cA*hl-rAuO@QjnB4C5I42&NW^o`YR5KOP#_=X<M9BW(*iK*1XcA
zS-9AMw;?Z!fI(3AiWdMeU6+{#slb9lXy3jTsumfJM==#r1%s*O0zRxH_AQ(*7R0=k
z?V?SB3^e)pwJk$p!|`cvW*2cgNjSVAcf<tlJ(!F(Th4{X1FGd_*!kI<CX^jXaW=pW
zJpQr2sM%wScJe8D<ZV+|!l)r#C<Nnr+^BlF0RIoA!yt!Pbkjh}=1t$qe&;7=<47cc
z14rUPNScAoQ?4uPg<3j4^ct}C{kzRzcQ7diRlu+K9ydIUT0+M(pm~{$t-FtMF8@>&
z&g{6S1G2n>%;r>7@dCO+aZ(OL4+%jD<0*Tv<Po!z!v^dGaeeAGY(ogCE@4yf_KP^D
zMops$M+wquP70n?=+u~a>dAJm8DI*dlhFi0;9%^7b-KXA1q0u^x8oiU1C1Wm6%WoP
zY<{KBc>$dnHfT{M5DW+xMo+~*Hi;&N&PT2-MB<)2aue%&cCJig2z6h!hN;ZLl?3y1
z>!=v_e##kaSlbm1QIzL|FD)C0{ZOx+Ld4!>Q>WlRyp^c4-TAylFn?OL=Swl|38YDY
zmXA-#<p7fey9sxg+RJkhEsye#T89%#Q6tc({SCy`$Gj`Y_Ll4tk}ZAf0mONiyiE{H
zdeQCpfa`-&-5^Lz^&5jxn}LmL2aHG9C$C+MEQPdxa_1Gn0!_e!0C+q(<Xwz4M1^{A
zZN5ZuXP3oI)c91CX0SfCt)}nc^*uZ(28FBeUK*6PqI^JaCK0w8^;*_AB(Qd~037)i
ztayuHfudcP5_5(qMY)`5n|n#s)Q<o6VHqa=D--1gtE4)f(qahrCF9FHiIy%*J37dI
z9jK>@;5dwIkU3B`)T2w@jHT5h4(?i=%|r)))8VA7EPG7hKG|VYw}t!VGazE>*HCT)
z=>3SO3x-+oDo0AV7)Fl5lM8+{X>U;lp@}j=Lm;~5-cH0>DdN5QxdczdomIc*FH2Gn
zLq5V-XKy33Md3%f^{K@W&Kpq!MVHId9C2I3X(^k3R>td*$qbu#YyOTK3Os>bR`~sH
zHk2|75koagKqAH0k)AFU{$YY63|8ra1+RmWmu_B<2b2C|mtT6B7JBsp{cbFza$~=Y
z(Ep(8W_VxzK?0e}d_+1h$2E#bPVn?JDZ1&b!IIHG<p7*jg{oBt;?mf{-{2!GqEs!0
zC}e}%6YQc0q7Fc~H;j!vO5#xL1JeaJG3=5M+)yr37$YBoT!UCf$R?c4C!M_+T`36s
zRuFsJK()*|7B;ILtP9zKV#LpA0KJRrD!DrOG`cYWie$y`$tmC&bC$$y2dgM0bESRQ
zOUqry_fc_vzC(+J5FGL}iUIo8rChj8SLzE;CxaIrfqP_>*tR(!Je*~TW(|6ravw2{
zU<?=0627GG8)B=4VS}}trn><<2?UEpa3ysW_2*$>4z;q}9as?ciNAPiVQP(-_~ez8
zo;uCVH`9MZCXJ=iyyxGrHVqN(Q)B%%iNy{O^v(BvA^RP%OIB0l=*+HY)N+6fJYD(2
zJgIOGWBFm3I76^-)J&G{`LYKOeT{I~A&g1#D4274I>O`_g&x|xB)Rr$4JK^EHg&k?
z0E_OYK2omO?GhJXSP>c#n;@8cE{XF&kUcLT&7mbYB{=|udw(`D&Oc1j&KM$OfeFsg
zfws(XTnhJ1Udz`D;E=Ko-SGmqzj$SY$r!ef`{kXpJpss$_na&`Gzoj*A^8opz|on<
zUj67@PP{OQ6+76;7sf0Oy#d%Y715A=GHY09f+sY6@i-erKnuhPpOjNLb`p9$7*?=(
z(lIRlq<USnW&Xv9TgX7!cbgUm`}<opsB4ZkryQ8PtMeBaC>+HhJLcr$fe;`6Xo7ab
z1%s<z#raQ0|0&|6p#g7fob*Ac#xWj5E2e8of5+WwabtGU5@Tg>(aYd$icjA(^5KdT
zK?)SZ>2GrxxEb8@e0!Qu+~??Y!SAK8Rf+ws2uSF*ba+507Tr-KY05q{W-XpVfc9p+
zvISv6`?*t{@9IFvasjVMDt>*50|w~&kR_Z#&!iD)6Hy6uEFv_nKLzu)+s(x?laOrl
zV)#qcm*Bh(#`YnqBMgqbrV@JuhZ`c^OBXPZFadcy=P5x%PYv);vy00lske&mO9p^J
ztcujJ^&~T0TCs9C-=URW1d(+5@!(6oo$wybh1SN8Ye4p3zyR7tEW#arc2X}&$qlLy
zwnw=Sg*E2mBWPczU(1I4i(}tOH!1Wo9|2Haq-4HP8)Bd4+sKt@ud>pd7=|4-X3|<w
zvP9U6;h(#l28z1==CXu6Bjmuo124e1YOi~A-EbNkNxC%5O4=-P6VA#h1TaVNqBDb%
zwql|ZN^B3}35!zO*RU1in5RJ&e8GPn+210VJ>eC~h%~xL#22Fbz8x3=k~Tb&(9zx{
z*xo55N9m`enTA4IaIYW;yWN3452Sv0HNqz*jYacLK5*Y+2wI);Eizv!&7)E=i?aPg
z!A+Jx<WMk0aCx`gszkc#(i(j9A{eIn@+bd_TwDHyl*q8OkT?~DQUvtrrW*oq0$1>%
zN6~2urmD#yI>&WjvoM9{2ND2A^zPHagUCTn555g_mlFXMn0h&hyj#Xd5dr$0USHq&
z@QR9oJ<L<++P!PIdamGw?R4(iWeT-Ph?B9MFjI@8O+he-%TFqhQfwOWS0)1^mpKQf
zhR|kehgIeblyiO!zzwH56(m^okKbh4<;#2ZtvE8Rkf6dh$Ihqj=QJQw^ZMdzlluyH
z#H-=iU!YuJ?$mBBE^GC>5xk6VN>!z-RtQr2*UqIV%_Cv_yhiG@fq6U<{MN<`_&QBB
zb;b|NmggGkpjAdVbKs9X)Iq;s3oyR`mW4+KzfqQQMyx+8(h9Fk8n%L$6(p7?z9@-J
z76Dg#m_!dnmepcGSO2RnZ^?ec7k6{WTO5KQ_c+iQktxGyM3DPGY3{j(hui~3W2u7*
zAefXAcFaSr{?`8#u#yXk^W3YKg|Yv**rL{K&k4aBKOh;W+YWtXE~Fm{L7&wHNCOzp
zS44@;No?_P`2&Wtuu1T2kN2X08!#APfrp)pQA1?NLBy(viY4e{Ty>yOm4bb_R3f>m
zLOGX*co!I`d~(GPXi9#W(|bIFUQ&Xq_#0SkM0u$^3Gf<XX?v`93?<n-z|DZxOi<RF
zFDrgc<6k%9z?yH+DP($Ix=6pGDN3FQzThfMk48%Ehh;Rdsgd@dyH7}Sb0VIHQknxj
zIji&6-S-aNYuM#^BYJJ%f}|NDEL5OSRXvcjnRM~5rxV-0dL2e^$8oC@Vrr<<%fAam
zx&E+qLNL&1RH*6vw~PZp=%H0#Z5e=F9|iOe2k^Pc*aUNFeYihe8odJ8O(|<0`8}-h
zF*Cz@WCK`>F}P3A=(Ffc`6jlwp<WD1J#OS0-u!~z5V8DdMdi+828x|CRmRs#yHhds
zb6T(<Is?#z`cRY*2S7EHwttC858S^ZHX?%WpB5sJJCYi!<&C|R=nuO5nGLbh4!f7%
zW^Mz7V98XP8pC(T8ssB89Jfu%N~=LSLim)iUt<})<xa4;;DoBkPjVi*Npol9eH;Qt
z;M&4rLIiniUI#5PgVtC91{!ku)Q@rJ_nqV^!(laY+SIY-SlFNZWe2RI8j`baS#3x#
zVzC>2&HE9cO_tyYYQaG!&0TA&b16oz5V3mAd51seE|81<9AhTz*jHbOL-{5>*ON&d
zxHIXk9aY|!Rzr>XWmwRE&aLgvr7D)}98Kx`ZR+35rwO|I_VYw!-Ho-#d{H{z!6D2n
zGf55bh(90O>i$^!!RE7DYAG2Z&oq{9*J#VB>y(`v?9`U~Fr%h(Tl{85y7y}{g9puK
zrop@^51&=;+#k5S`uvw&WLs`!<@EsHzLiD3(|5<h0>diW?@G?StlmPaG3y3y;NoaG
znaSe9a_ok$xrQkUnFQN<vZsG|n1zikq1An!7u^i?kTtzTgr$L%DaMm&D}z&^NVaaq
zXfO*IK5Q{3=oErWotEh#WF7FPYL2@?5?Qv(8uc{VRd#fcT|nq<7y)*itldozBK;a2
zx<Lco`9VGY3b7)!m)DR7Bj|pQywKZL&C{<o?W65(jUCY0ke+`JnfUMvaaA%LC(MFQ
z?aH92#ns|Kzop(~CY<AcdRz_-g_QDeHIrz}R8XC%N+g8PpxZ;1yAt{lRGgUo$>22<
z|A(ZcWJ4K{cjZw^)rhO-oz?CqL*?^;3CSS`t05qQOX1*Uz{H5Q4=!ak?GJy8)6F#8
zprLXjn$K?TM0ll#UcEX)h0KL5M#3UgCK@uI2Q&a%#1Gjvkzq+9=(iEBRYQWOR_6!0
zqE?;==7Hq9uH>XD)_fV*Z?&p-^Jc8;Wtqo;{HVet1_oj1;v7mO$}U`N^X~8oWZ0NU
ze*WEGe)fZ7Jv~HT)R?X1p>hJFAK?&Rc@L9Cu_Zx<L=zyf_u3_fDPp+dhL@26UT(B9
zXU<1doB9+K_Lr5>>7NH6$r2pnf4k<5$Fzp`<dH?7C~tqyyv2P7&$<>#lbf4c1I@ue
zQ{Yr(!ZVo?X^NxuW|wzspc7f;oleFSrsKgy0m^x684}cZo25jxvN0h?Sa(>Cj3X*;
z{Q^u=YT+$*PouFuXXhB`Wht}DnAWRz$rn>m7okQ9W<~)Z(K0{vI8`!I<esF8t80PT
zk=BbL%*JkVdk<x+nR?U!Ejug`S~=Y>hc^hmu|ebDANMXYcOdwdHryD#WOi!kc;B-l
zIo}*bb+<o6GB|M3_%cR8+uvG7X7N4FVP;4@VQ^!r<{*me8VLu*`%OpBrD6JEgPj6q
zb2|u8Sa$7t7r!@Ldu(V3Oag7=49q4Q%>M<+Bc5Sop6ZFZE!t6X!Msr9R~}WxsLE5v
zUaQ8)l{pX7(na$HhB3E1k~;DC3L;Wh@EcM;qla6?@io>I#xO`U=yB4e>({R%-!K-P
zdkj^+pxcy3HPR@`BggUc^{wA?6?&FxjX!{>2A7s3AmG-9<Hxw<1a?a2M=Sn>26+_K
z!0#_pjTLdW-TRB96Rh|v3vV&(fpU`SyuQ9ZNa@+GWPW(CUpht-fD?JsC6?#fG<dpv
zPsvB^Fre<;93A~|hYa<PiZVrJ@%{VH@&)5BUolVl`1+=|1SKUUDN;?Nn8S1nd>kFE
zPo`f63`<5j4Zi&)n6(2l_j)oiZMS9lPM75&pA%W~N1LK-d|btr0*Bu>CRJ8amS6;m
z7HSY1scGCXXBJ1t!)y()AO9RTpigBDJ+<N`zMeHGJn0!LNNpz?J^KAnU^<M~IWgy9
z#vnV)dlG+_(=-4#7^|)DaqJq-SeXcPu1H34D(<^sit|i^W*IwV;27p~6aILz*A_Y7
z^)+E2JGqT79kup;=?+SZ+#q!%yfT$96PTeH*E3Rc#o4<73O=Rd8KI}Wt-Ct<dMoc1
zcPOSCU8iUxYBQ!blRoPsivw3o|7`DO!z)qBZ!a4Lo8F7q=c^x7_pI(wO$=L_rDd(V
zTY(&zcuu)D)hC<Mis{YfmZ57_Mwa-H)C<PfY$CXGaI@_d>7b(M=YSZE_hWP+V-X^s
z;)xXNfq~{Di_rDi-~k8MET+uo`zkz~FG5*xv%9CK=k&3gV!pWUcE$=AdqP50r1p&>
zAo<&a2^Q3_AtQV@hCcyco(3<g(G0<gy6YT`8Q(H;N+IJ6pm*R4sbX{lUKY+nODh4j
z6cMLCPJ5d@BWfq9?|0E(It0cF<%-k5@)wpto5;^Ej$CusNJF<kYTiVx;O{pNYamk*
zG#81i=>T=!uCLo+*ADp7{v8$^(LMnD&AV}@BB}yHt4nQ`g)J3};?DDTOa;utK>2yh
z^l%Mu!N`#LAeii#t9KAhyvh9eH<PY>Ft0jr_iHN-KnE9tZ9nqUZ;*0K_~6_G(4Kyn
z4nMBVu}_<tnAmRVYTlkTiB(9Y!Eo@N&khf6d1oN)MAwC)Mr@&1m;#7ikNCC4A*t7I
za4pb`c@j|00&<<AI0sxTC@xbX{c#@U?oJQK0-%K5MsjJq(g2r(vp{Kjmb&RzW%&De
z01Pk5DpmB|5)MbCyofZrjH1OK^5&OohAAqyWuFo5oB)*&7z2T%VPV(M(}0q`44<(a
zGQ*VFBf`QOr&r)ZB>%-;18_q=1~6z-p2OXQ4XMQ)Up~`p+eU`}kpcp!dCeW)K%i7x
z&jV1jM!Re8@PqjC3qpcU{JVUP=f@v^Yw>cL=hr1319I$kUZ8b9u!-(Q+CoN?ew|}F
zIBM_(X{+WFtWiTX7dRL%SS>!FhOisRI%eWS<KvQ;{WDw*m;o8sGYM@_R<C~gW4Ia)
zqM_?n6#l&DUON=2QZMA^<rVoqurEk!LwiI5KY!VePnDGlA%<j9XW6#E+afND>1K^f
z$Q{v~(A~p&2ZPNJSSp=h5XIz$LCLp%n9vLwtIF~R3|>SC_3q^j*Y3|7P4|x@EKAhS
z*VZVhtNRV;cC7CG(^KWmZEZ`y^fcdTfDsTsG-^_!$|eROlPHw!35gtBXy_JCB4--n
zY!Mv0oc5c}(k9?%+i~bEPd*SlhFURnCnuW#Asu20ZiSYZw1crW`zAWHCq^CJDRVOH
zQmXTTqdoNrfL#aLRWkP+xq81U8dk1Hqzfd}i*wJctWoDpS)G8Llb@qWdthX6hs|*R
z3v|8?sr{arfp$u`;zrXq4lV65n9pi8o_m4a;ry+H#PmbMbSQ?F4RqszeOz_Z-hFW^
zWc%}=^&GG4_9;$*PHf`7J<H!iK+f?hT2XsWyDn#0b0|U;)O5;41s)uCFM)pz2g6sa
z3Ee-}Wsh7s7-*yVegH?Rr<#m>)7N|G;o-GzM;s>F^3)YFkLatMmzMUj^Li^_P#^|i
z6k}%)VCDf&5#iu~BvH@E>mL}7zGor_kRgyM(*FcM4r;<HNC$lI=o!0q4TH>6U?gCc
z>j%>ca3?T0{Qx)j1r3RfO&1mag9~8Zd?cZAy>SpnH>wzIO<_xqZTBS2e{kKzeYjT8
zq+dsQ$lH34^$nQXozb4yTw&&-`kgy=?5hgW<f3Kwp3mMM`fY%Y9h$0d+}?%IZZ;k#
z^xvW=HC!3<d$O!zD2`p2J=#c7muEc<;NhfjcrCQ{DE$0AKdxb9EqX4Vk#V!<Z3n|?
zTU<7Oc>MS=VmspnfPBHdDZaj22PGy%B<6pt*ZQ6ZyaJ<#%3joddSDi`hEh7<3bl|A
zV|Ds}Jn$q@hixLn`KQ1~hC5#bvDXdka*@kdyqD@nVDVP;&WVeOA<XJTlw+m;xB}T&
za9gY^_$PRICO6E6<tuw>7c+I*9#sD=r!{@QSfKV-SGYakomWLwRb$w(o0TK+Yc?3J
z(Wu_N%cmc$5SO3q6o~>WQ|G-MK>PwQevlKgp|f;#W5O=#f*%_AH6#!y!Qp$+7(`!Q
ze5*b^{kTu@n_$&6fc1+;7J;RhV*KYY!x60!Qf2Is4$92TJat!L>*6<8u)^t|VqnMF
z8g4&Rm54c$uEbZWgrS9CV_(GJF~FQah#Z}qSj?qPP?6bPEdtlyAZS0x#kJe?yzI=Y
zkfC^yLFpRY?O;YsbtqcmB7<jB2c2L7vmxqw@#15D#nw?kMV58%o(xq;9q_h~{Q4Mt
z2~a1Dfq%@h>Hi_l9}kQGburLy2BGz{o!fH{P<P4Vk`lhAu-TyodEFYrz^MZx&-$el
z6Tu_H*mdH-3?O|pI^Xro_15X`LJbwgb5&rMCgIdJ`pP1{!#DJ!2!bFthVg1`6Xv;w
zE19fFNmg-{m6np41wtjrwI6<>;In2<5i<5Svsa_Q40LvOVmTQY##`YmiX#v2@~Qmv
z;MBTVWPKp@6wgWOvOb^=ES*N>TGmdW5~?B~9z<2qQhvD%{`{f}5;Z<G|EPKs7RuZ&
z7k3+qbP4=)+qQ{Y_dSmP{_{r^;#2Jh(6c|;;z-I}19J%j7z>Q#G8dZGlaYDJazB>p
zN<B}C9J}!fT?;4!X%}{Z00Z-HQrWXVAuu>N1KT!uHl%QM@fLb|dXkWoz^SNLn%&I&
z_ZbWJKayj83>)RRt$(XQ9A?;rXP<hzWnIwk9WNWPFoaSDY=p+0%=3huS40~q5F|K9
zoI%KS;W~)?SsTMT;~ojq2R)oc_!>mliE9O$-JeB7B+TT=SZsPc-irHA*73&dv#Ot0
zvhGBgjh#JQ{Nl=)6HRh8CIcHbY{)|rj$~z*Jj}NUT4->Pcbj*koK|+WK~`UcE>3uV
z@95~^@)ENcN8|kw_A#yV_~D-!t)vK9%qEkh8dC;4P9QLVYAY`g<ERNZrz{BZiwyha
zqCWq&GBh#Wk;h47(9j>u9~^-=S3f*c2NGbIVd0zK|GwHwyaCE0!M~RfF`&c4x|idU
zv4UtSZ^u)F$3=8E5Z|+@>>R9@=H433)PKCM58%)_p_R|L1E8-TLU1x{UJ37n0Rg>R
z&iT@@PkU~_1QV2*^<j}9_cp`*ZP>&!c_9%AEGlkfB&NGz9ygnqFY9U5LbfnGTF(-P
zz`IZ95UpU~ag5a{v=CAM9yCewPc@o&y)~L;x7f|wP1m}9NWJhYJ=jn*P1}U8C#qE-
zO$i4E1~taD;9zKb72~=UEKkW>mH9z>6I!%@?(O<xqznz;9QS?o?Acxg<=<$i{ib2b
zpqI^tSw%n0%P^`aLALDP_tr6)vZf}@<!qka@^WZIXu{F%?d^?8nUX{m20M+Co7(yF
z=Mm7Mfogxm?Sz;CL3@;Ok?fFH1yf}O7_6|-fcxHozUo6|xf(&rS=RcL^>H#+Sf_ju
zCDPB`&v=1}h?8QHl%(g1sYVU5J~CsH8#isDbT+BHJDCw}75?_md><u1S(qD%uX%GK
zBMk`<DOesg_MY0e6cWGyDX}Kd8_>kYdBY>X%GQ?c==H}ohtp!OJMth^^LWm@56Xdv
zC163x6vZb~pga%s-3$psRTHK<NN5^U2XG^InK5lJ77keA3ei4aJI*?&7_dpNz*)=J
z(-9Kt*O>FS(y@R3@&#;*0})xbQ?5XEa}9tJ9mb;Ls^U5TsXSuMF1Yw(ADY7O6IioK
zf-}B6ZFH>E*Y#)q+5B6nnenW&?63;rEwWlCQoKTTvWP(wJ1!*j?;qW`xiS))rE!MP
zCKm-Sl)%CX$49tgDD=iV$6i;gtP;zs86snar$mQPThwR_>R^ACm>vs54lcKb2WTbd
zS`;y@`-+rghwguD>Nj&6>Z7SLs@f$5tVuD&kedVW3c@=ax+NfVx9I_u!L!A<68XU4
zkud6H-RVXflurgX%zGHzO8VQg9yHMYgUmbTa`mWx0x_%jT8z0ujG9X4iNVLG#l&=n
zhve0FoLsKw-j?+^{i6|Y5OzfoX*y&R3`>C<qr-aJ1J4fvj~X2QR1-Y(g#Sp>;++=V
zW7m#-%HP2i;+zeujqv&_&Q=ae$*_5E@$db6%~)k(4B-e!?J+MuUww`Sb9!!<PAhMJ
z_#U^;i~AL%-!dUGi#ox<z&K>gIF*k0Wl;DM+vC)?Z)h-{;VP}GJ0>8Ih*)KEauTUv
z#7GVNN!!aEFC!x*&FXm?R*H}qilcBYbUR(D>}j&g((O(=R0a8d77=}Ov!A+b5IbaY
zgHsMpi|Jr0Tl^6P3<@|!8>$#ELr`MlY;*3x%3)@~b3{HM;aeU%tgq6J$HSpbwfTi}
zmWUJxNiTA)c?Zxa-_CR<SACP>>B~LnRQ~n>Q6^~JS|fK_xb1=Okf!&!O?1F{Ar<i*
zIBih?Rs@!fSuAgPE4$p!Iz3dsfM*~FU4o+`7&N{Bns9vT6gA}XAcz!y{raNvMn7#1
z;?EknPrvRcisr=`6c(6e5B-sWLKyF#;LalJQmZ}>gEQQoo%ZP!zd?cSOOGQ^L`|ra
z_plZt;YiD=4v(90$ByP-)V`2C@%^ek!>8sM9jncBbZ$UOv78;iNr9$eo5<Kn*t|w9
zJv<V#n0<0#p_{q6xsdHuL?(t?JDQs8plJc->y|(Nyu9fxD5n#EyxPo@lgD4!uzTCo
zy~3(8-ZQ;78K)d|_UhHw=8Eajjekm_0)YrHMDfYOT;Uo47Y>wA#?;=`2{m}q*yg@T
zR%4+<1f>=bvftvH#ud0|0tr;^3=a)8seI2*dd6V0f-W|#mKPVtkVkFH)O4{ygjM!A
zVC(eok+4g+5-N{K4hlaP_1n@7D*_c{GHS5Qv(s_-aBpBsOb`dPX-M+Xczi%l(<*9t
z@023-W52|3tZQ)ltSb3?SO0NNhIeac2e+|<&S}1lF6Qmq5wLLZS;&jFS06x$%>=rc
zKwZ@5ByAppq<F?R%G1LGT+l<v=w@7J!C(&JaEKxHV)E5AYYN<WL?0*DrJ4FW5c#!Q
zYJ6jlNS2IRS)qF==8$S5&%!o0_#+B6P>(WxgU=@yFo+C6fo#LoX?O<Ap1khC<)(k{
zb2sVjL-aqz9zL`}*Kii5hC1<~@SAp8NSE9Zepu6wS!J&f9=q+BM~3YsNhp1(7m7>y
zg+ssZQHoZ%#H*Ru4xkfgredUoI2V^ODh+UMyEg2Tlw3oe$X9S=5w)#&$*k;bXfEiP
zOug{mdd%I>Du#Db53RnBY)1H8sAI<CkHrFW<@a?35ayXa($h1A0`%N)lM;Wn-T13d
zw<U)gKD_70fwq*A`pb;-qHVQ6!`okeZ(6?~VsZuRLWUMp$}HRAkV$b(pEd-eQM#s`
zP*G7G^raw90S`$=WB&U&x&}O|+F~>VD{H)V;<0l$>}kkKP(}$2&dtej1-BW6s*EJ=
zve%D(=gAdQ5Fjz~swE3dok9~37B78@i4XoqUSyrd3D(O)CY^6yK6>b;_`r7c(}NRL
zWJamfcDMrMz5_TBTil<8gb1Tc{y|szz3rO$kjt@KTV0{t`WVl8!v0%T)h@|2?5X+0
z53y_0KnB7pZ)oMPT-e4Arj0n#Opv)R@~bGQN6F39BgY%VA^Fg#$rc#2{NoyE4ft`a
z{}<E)N4vXKKG&$T5pXK?d)y|vm=k%EGc()h=p>Q;#p}*|P|rBR!xO7KC*i|$>{wk+
zO^Vv--T(IR7B(e|$U|rNpr=r}cn-oM*4qyVqy}0@#*{g>#6@#@-b9c4JP05dJ9Tws
z2KKBtG+X_pLL;!}m5o1ew`KM~03+z=DX!p<kb_?Zfg2un;o;;wuUkw+xx>9Q<O7aY
z*z^=_xp1Nr;tQz7G+SN0bjjB`wE^Y|8;}rBn#<wVc*=8a1`SItvoO%pM_<x}FtXuF
z{?B-f;p}*)z|tW27t&$lRC&`xug^I+8LplBQrltHPMFqmYDEF)M=zV)YftxtZHpMX
zP+GbhdP`Fej#;#yYFGrUqI$2kAR=|FGym4_jp?9(`Iu#E*ss7;gA%Cx^*ewFoB)$t
zqk&yM(z~tA(9iFZ&|v_q^lehOO{h^gA%kOl(r`5*-qi6se3KR7xE)>ELJ<x=)_ZoY
zW{VlPP29mKnDEH509LW{GBO7b9`s(xJnf#UgQ6eey#z@*7U4_Wk1SHIVBEly^h312
zY<M4E)-WE|Mif&70CmCSfRg4>UX6In6FJCd^W#T1*n^?|FQ?=vzgBRi<6Lhb*C0K+
zWi$#01I<z`FQ`OFOd-N(6H5k}2yGg~kmjEV$2fAa;X`X8z3>1tX^dOkCxCkJF<nPi
zhHeCXuJ7jn;cjIYyP}nanv-!iLb_lNF6SPN{BBLff6xq;&jY+cEUEG^jpN`XHAbR_
zNr~UPsgaKL)azpvVCX!kJe7M&JmI>=o$yNsO+_Rh1ZI!y74UD2oI$p<fOKh(DS0Sl
zb4_u{H*5A=)mXt1U4QD$#0a7Ec{INo_3f6I0akULi(n69cW-6_+o|@8#z;-IwcpUU
zoPpbvmq)%hhYXKzyaeGH!!72Hwl<vw0fTJo7`%#RqOOWK6iO{olLH!!-O*MN_9?EF
zm9iZ_^Z-VTFonSl2L2tZ(TubXXK$(oY{OWK`N0Wx8u#x&36NnT^){e91xp|YSu>pW
zvtZ9hk0|=lH10=kR4^NJKWXn)hF0;GZQIx+UGw-gQ!C&#)J;SY67q!7V}Zdhz921q
zsOcS|IA{l-77(cNOVnP$=9GDTEqm&qb;5PoUdU}4MI>n4@Ey3ocG|iR{*uv6{oP{5
z&gwiQ<|1`Q=#RNABNjvd<w>NokNqARTH0Nje-BVE7WeCl$k|V9-@bkCCL3?dcX`Lx
z`K<o9hGcA0p(-qJg6-2vIg!0V5%U!E!f$FMfdHUIW{0l<QOg=uKMZqUzkUsL^q<>z
z>CBn?@e_Cf6tf?W_>du~C<|ZyNT@l63K*Vvj@oz$yeW>87B3sWM&ZeTv01@wl4cPM
zrfd*_RB>5aCjg((F%wLcMDPJ#7IlN@_gy~5kcm0>Xse3;12T`qR+7$h?)!r0NhR@_
zAnT651Gwai#Kq##rmdby8*qDkhtd71pCNlUBLHhky2_<*@9pg3l5sR|J=J|fX$(sx
zokKY`ab{b2AT;|=?pmv`M;cKbEKS7T&7c`T%OT&aq91Pq;pog3dLcVAXuVljBwD*C
zSfO5noI5kDQ)N$!U_twL=#e*SIf?u(=HFEgl9{h36=6k!yd0Z2s{kqL;e^b1$)&}`
zuO%hrym=tp1LgL&?WsU90J=OZ5h?L#(mfZp+l^3P)D;kX`b>pvF{6)$x+yKY81wGk
zT0u9YVN>i;cf{}LG_S|xvv-$F9J)Vyfcz(h8Z}b5nd#|e+{iVwA4iHgs5;D}wC(L=
z@)i_g$Sla*+m?-A!bzDOZW0s`$wAPI`UCJ(T%2Y&T_CsV0<+%gW@bak<p>8Dz70H?
z0x*sW3JS1a$j`4p4`9g5_x0<&2LgLpxxn<r#vr685i{-tGQ#DR6^u?>MdFE|oP4i?
zLCN)_pi^pcrIl#IYN2olb|&dC#_A%bohHmu^a8h-S5;Ijqx*+YIS`)5vM-^p?9j<Y
zQH+a5coSS+AucXXL^!a09~3Xqqf)CD@>`p*NK*qaU8Of$VayH~vJ;{gLWv(vLOBET
zwXK0t+R`pyX>x6c55C$=LYIVm*%+%YpoOXqvv-4YX#-loimP1LCqQK^N{uQ&s9Ve}
zkl)z~km=OuK3DtVf%kuoMvcl>kKNqCvJ*DGaYKWmR{C@L-lUheZQtHi=O2HsYq<X1
zeec_M?~oane+_2?c~@06V5l<yCMlKn*9Iwj5SxN`OJeM=xq^@l9t|O&WlO5GTe+ds
zcIcU-jCv?s&JE9FrGFcIWFOaChlPb_VTS!fIJvmA_;+3CXK?`IfRq;!+G@3V)nX^m
zR`9rs>J5_O6Z`kW(L*(U$1Wc(WTt^i5-JKhl2EKz61{yNQ$tDrBgkrHWmwW&<U|1X
zGGR!;ycZM{JbZW*NH6vhn)bEc$xI&rFHa-<z)%0_w&gv3y#C9VUL-=etDx>at29j>
z0+1tXvDVx7B^(GQ__cR?{Ma;bM=*?>s2bYXr03=?!61^dON(Eb`(Cv;Rjz&eq_p2i
z9!XwAY#w~l6Vsqi0S*r@;iaYnm?JPk`wLlt5KX!l6P+9|FA-{yho_9V_U=9N!S8u|
zGJ2a96J+ZsXcSX@f9Tdn8fd2x#9o6N!)IS4Cn3zDRCDQ_UINJ#jmP>L-I!0?$W~kA
z;Gpf+3-I8uaE9ZI*-D{G7hL3SW@bz!sz+7vRs$OP9xyMcy2!H4xXW8g!Z;kj_-c)P
zhYwdS>h{QUZa$0bH~t0lA87@*kUfha9-5*Ht!g46v$`mMW&a)qhIH$Gxdh2kFs-oZ
zq=^wEy7$g%B6UM}3hgJ%a|j$uC0t-h>m|?I)@q@al)*bpR}{>#;R8rNiobZ+w(U9o
zt*lH{IMz1@8MMJnoO+WFTK7a#&EQEgl04nN8YGB}5^)<x%SQsD`*m-mk&bZfZ6bM|
zs7c<tWao$*XS>JO&rgFLf($n<ba|)twPWw#cguuhPP>gJW7=N?EdYk|8iX2gZ@F8#
z1`YBJuSf^}hF9WvZO2<J#l^-T^&!(zPRK^R#q@kGYTT3)34j9O_HQH%j*YDXlPTkc
z*)GZ9k#f{t00%!vLH5Bx(3%5Q4f_7B`wsHe{zg)tgt0jA;aO17^$x|$%k(%#?^}aT
z$-+I=70`v=gf#<YN*8q<wEL*@p$!HKQRhY$aKM6Xi!^0tR1R6$E62z&_DA=W4~&gL
zN^X;}2QInJxVw*_3i!y^ZQpwBjUU%WzZ>e=@vX3SMHb<g-XY-laPNwv!FvE17yMXj
z7VF@|3(ygUM-bHccxm?8bbNws4_^?V>;ABgob#%lLz$6{bJ_f>u|Xv68G^04*L6w_
zvSyW$=<eIxAi-Do*G8%4G?S<4Rmi`5tXa;@Bx&effkGS4O6ECF8{gZ?NG`1z^qj@>
z{y+BKJete4Z5wtqOQR?nh>$TwGS6c_$vj3f&kdrYP)QOphm4U-KQjqMh>V%1CS@oy
ziOhW4S=aMC@Av1o*0<hwt#{r2xbL`r!+9R(aqPpkZ~M0M-WS9Y5-^I%633Pbu=_3E
zwWU5Xa)>o014TF{It8LuUgFMcZMDCOpxP8XMp;KW_@NQ|JnULB#n-Y3|6^U~P{#p7
zAtVz-Hk7?^Q<spu!$QYIM=zsgz-Q^1ZkpC<t~tW-0(8dy<doh;m|{YXhf#uP(nHk<
z(M`@-XV8~*KzvV3NSM(rxQ?Rtr=zyOg@DL_Ib0MF5Fo_MjrRKa&?e+#jRZ&EkDh4P
zadL7(KS#P&YF~f<RqR}x8bF+@s5b8AS_89}ubX^`;CW&gnZa5n=*!2>y%+<tIStMP
zpE}D*kn%Hbiai1#>FmoA31$N<B`+9L2yhmVo0^)Lq3lLHAHXzdOVtxB*t3BU9j>!c
z(*c`B$aN;g*}DVXNeC?Ndwb7eM+46N=bwLI@c=fU#ISiM!>3m+c>>Cpawe{_?iaco
zZ_k5>q7R0CAm#TFW@ZeqG;izln@4FCe0JX9HNV{`8UPn=skIrE0t|5W^l5!C<4ks*
zj6B2IR#^dBh#`QoBLQL>9}czr0w<yrr!na>)d6fZLP`NE96k_Yz`b|yhXv}ZtUcZu
zR(z4*BkpEoyd*Af!2KA}I>rZOgP;OWzJ-$V%Ju82Dk`Wk6La2)!9on(=pTOjRxZZh
z1A24k-4`rIPqu=%La2DXdxx?h4mM^A!RZpxL#(Vx0L|dRa!-z>6S^oEiP+>C@}1C%
z!@}YJ4kE03Ki~k>3)G;^6=_#ah6V;sVfD|g5d&~gibAPDUQRCJ%yVFK*jMje1;v?|
zzCrV^uJPwH6unFvR~+FcFu(-jAmGLsM_)F2DzmX;dD`GCq(ab;3YCyzY(<1l5GGVd
z1ely`MbHt=-VveOiAz~{<eksA>^Ol_VfV#+^c<1iBgaOyx^0mf*H&I!ej0!d;=-(A
z9&aKf5(1?SR15*@<_p-5_~hxIKamsa7ube5eH_3()U%RgM<?Pl?B>qGeq1e~Z33DN
z!ZxCJKUE_M`X&In5xHR)5`nUq>csV7mST6&RV#h_lP6B(U^hK-eTds{jO71;4dW$}
zhQ&k{^XTEj)?OtkVwN>An&Rn>oB%j4z%U7kiQ_L-Mh6FzJ_%RTx0CgW5u|l$<X`Mz
z-~0;;=Rg9ZIdI_S_y>W0MWO@f+GK!f9BLb+ru%mf1FgV7g+i7k2n4Ju9({XW!YWY^
zDJBn(ipRR=0Rc_OMx0|O5HS<=*0WTITQTf;1_$1L)j#d3m7IvCviVkL@hQORDX040
z>{~|5P5UH<fMZuD%$)anZ$E=2M<ROeXa(~YnHx6<#Mx=lK1UBSI0F5md8uD$ug}{P
zWr^D2fsv84Xu+e}DLq4-$<0Odi2?^8@pV{GP`Bu}zM_^^sN<FUHQ(}DsJ3oJKzrcQ
zEA^E658d6}n9$tu!+G931rO_E1I1yezd(2<+}*S4gi-yxL_WUq6rXrnyHz5Q<-9j5
z;4JMKs(W3J${Oy^>K)^Gnk3N!pX*F>2+C;n)$`cZqYbj*r_vjQj}5|{!@2=+IM1yf
z<9|N>C+iUd)OhvUa*~Nc`f%_5Njzh8p&@W2m)SNF<7(6!dF_*;<1=K}k^-lH|L%P9
zcnCLzYVjyEz1#&M#_IBWO#jN*Gsy!h&@9T2WU+#X07lb^xIxN9kYSZSs<BSD_V2xt
z+txiJNNOHRPc0Mo(a;cs)<ry0M_$>dx3{;;j0a4kg8zMBAm!BHRL_tOXeH>>6Z}HV
z(Aj-N!P<ZEphTDGi4!gJ7YY!5U^BK}YX0`e^y^?~!M?)I<{uLi5&&-ow;Fhk#rh*%
zbm~zDkL*R(&TF98k?8VT{t@EOO-tRfZQCczAVEX4_U2T0bs}BCV>wF&h#Y30=vvSw
zdM6Zdr|DOHxMl&>i<nnSdu^rF5^M_r#hTz8B%g`Yn6YXX1X~xj*p_uS!r3YW`CALZ
z@M-OEXha|QEsT;B2lR>0MOxTtQ*M8%3@AN*ZT0v9h+P7Wq$u$doo@cun=TUjb8JzJ
zm$#{gr9?*D>xN>o8cY<?AqfkPTN<*u@AKe+9=bR21!%iAxg<Dlaw-ZJgQklhoVejj
z+rEyDio)z+`PN|h`_8-YrRiRgKRrD$Iq88GXA}bO-@7-Ym9Clmo@ntv`~lQ@L~w}6
ztO@x5jHy(VlN)NM+_EJC`Z4HS(-QOwQNPxTnN_87BLLV^;B3mR@76N-O}GCS7r^|X
zpI<bf4&vo=H9QyV0^8tpN24|O`9=Zh0n7~fBVi0%joJC!*B6h_2$DHKk2F@M^1H_n
z?4X@bxMM>zijtVWkz<+rVi;UYH{|$KOA|CJCuhL@CGICpIYvxO3=aUngh&VX7rxlo
zSa_BzHLkvxI2<+nzR~h3!hYz{So2pK$sq?rRiz&>j&|?<yq-^RrWk~JB=JxSd@eyg
zF+6{GiwqJ0qMmJDOT7}>hkTO&Tyo8QR_DxN+u@miVSCferJK(JOs@a(5*5e2th*n#
zLKzACbDMW}0N4{!6{zUjT}=$feId;tvPXs0@k1$&gLO!IdJ+7AMY+oH9SIrH_lSi8
zcphKyE2C1i5#x8^xm1*t)GkgkfgTG3N$`eno3roes#+?SOmCRcGBCIUddHW1?Cmv9
zM^(@cs0%0y$S3`K-&_tQNaQ-xi1@=VGeU8w#)bDQW&%P3b;D<A4R`?}_n}kD+Kt(C
zd-j;E^wcar0KkBX#4>;sG;|W$oSK@N7y}6zfi|GGp2|<L5(F&)A=ptiHt<6@fx!w1
z5pz$_1!mhEf2ZIq#2m=4TNe(v-MR&bjQ&_3xIe_vL=1&p+Udl8_b^?6xEE%fiQha0
zm9r_hJLr7?7)R3Y>PnDEQ5?VP>50$*s5{wX_fYq~WjuZVfB-;7gGiLYO}v4pifsZm
zhbK<;-pXA#uBM`=*1FO52bXV+(@GRKQA!6;;pyo~lvK(**5Epa8M&Z3NIG-o6{HSO
zr<=pnVNA#j3i<|~!Q>=C*TrAoy?r~hOy#pyc=s^P6%X|m_)7MV8=b%x75Pa*^7d1H
z&+ue0ao&-&rDO^g(o11s8WCL}eZ7Xz8z&Uc*~;7;z;_lN`kgy>;5s3^@Vaw@UMl(z
zYOaNL7_>62WM%8?@47cNHAS>2gZB>i3#Xjpz6;3T>({R_z__xa0v-|Pz~<MA#(P)(
z$TN7hnXBrHSfpNvfe_8VxL;t*;^;y^IFGeCuxr<amFgjIk%G2g^3C!_sSg)=5Qr5{
zJYIDLY%Ew>a2t_Ta{yjOMTO9gKoYg>qH|bC$W=_V)c*!AfCo7?bk5p(2wN02cB1VF
zkCQ;#kSYSG=;`UfL3aXq3|R^Q0AvPe(P;+Ol0oN9RvGjNzhY3h375o0l~__#)N_3)
zt${EZ0v~H};(&LAJ$V9nqyrBf`;SvQ4dyHXR<O<?de-H}4MIg5$xSOBHsF~X@c9@U
zju=4#ViSG-<M5?Fc{(X&;I+oVS6G8}#<-hXX#c`T`1A&*G3ll!#NCG&?FI3E5KcH)
zSqG6TK7IPMbL?DS;T3c9XR}sU42p?1Ml4S*EFMl90s7(YI?YG#9@g7KKG2RR6)`YI
zsG-7u=FUq*Sige-c^S?eg0U9#o8p|0Zmx$W-X#?~v#V3P_U^TGyPhv&0E<MlJc6!8
z$W6c{BJF`2ga0L>F7TH6;yPgs>_in-y9r(%9v|@GFPIrVE06`=B>)02A<U*Eo9$1*
zjtNOftipZ~R`F3M*(wrihTsyfE4cXCVDzSjYiFl%y>&Fe2PDEr&n*ymLkvAdTCcmv
z9%mJ{Hb4#qFIL*is4PxJ0VNfJq&0G5^dlY-7G6NX53bfSucx<H66YF=(J7}Y;=koC
zyh!?3Ugv4e<*dMh_>LPy)d<cJj}5i}Gw}%Be7qC&f!~;?yLBsitiADJ`xqE5@@J-r
z3FsnFe9*H1$Z9X|`CF*h5zR-vWotw&zorJ|I%EXp<>f?hi7h*Fs*1If0JtB(xC3*W
z0pv1<oENze@Uxe$vuRbPKY#tgUPWi57HR{2Wf2Bi?=nyQ(89dR({rDwH!z8bz0!IP
zjk?=wt&yqci%%PoB7QotUS_E7^68xn<$s}dT~RIv2Qqlfh)&$q$>#b9V|1A}L3f<Z
zM(W5K*ZpjOV-L3&(v76nJyuF18jnNc+SDSMlt(l^eCANV4hZ_pzLk=)2<q52bLpSd
zk6j=%TuCLGYhRbeomRi8oN$4(fVsy!q4Y8R`EV)~iSD!!g%E#mF~!_JgZFO`iVCSS
z&%{Dlce<erc1U1Po%%!xLvc}&8&J7HrW~KK^`qavp8{OcI<oEd^kE^P8xWmn4TfRi
z+9MaQXJKWfJo#zM-DX%eAO&GM9-+90ik0K9ICcPQ#00aT$2f{o9&OxEzWaosz;k!7
zO+iVJCYT7pFaWJ1s=`RRuVCh}O)d6aX2|Y@0b!4w8>Q!Q!jsBSJyjP7a-z?}UO-H0
zQwwuV6;?`5F-G8EHGC?KvEn89BAODQU&%i?!F%}dSD=c=s#e+YA(Ot}RIWsTb6|gV
z0-^2GGHpzHV83{;<<3$ih+pg~>5!xPjOLwWnkCII(5}<}6e6|j9s(C<SWud!#DQh;
zS7;AAdBPQeDCvTrnbmOwa>H40%DL1O>^s0=$h{N=|3TDslRDxYm9G7|0cX3ov@Klh
z$K+FxKOf41^^U!h#7bbg;MOarO9Jgm8{<vdTz4PdLIsLKIAf-@NS|Tf9L(ztX<}wX
z^02vaFu1GUr#P-M5QE*~<KHai`TEwQCVH@7dltVaFf>#&oy>C3Y$9^_a6C=PhWKSV
zm{|C^ETa;(#~2-v1o$ASJIPVr)z%#>26UDhK8#U!_LyxJ_7<!n1l_UcgC5r;j9{*a
z)-Hw369*bD@W_MjX3FA(CM$r!Ngo+me@uQuo>Ik7T~);)_?{;O1w`VBv`%ci?O4~*
zF-xrG`1w#x7-N)~GBlDN9^L{F5j+<5aR#v8c6u$D+|WD-k*O=+W7kuBy@$PU+kwcs
zY_6AY!l(bm95g??#$~X~FD_!TT~hwdr|%YSvVYoHc>vqTeV4*o7DABTS1y{n?g8w1
ze9icR7LGCy=nY>4D5Pw@o^<L3A7i)Gi%pw0J-%o~XtsS)+>*GquSDQW04h|vA27Ee
z)HwS7gec71(I#r*6`~Q!CBy)bKR-ge6@TRb@tZaCzXE_b#BBfh^Plzk|9?$%mv-2F
zE-vUfvxR};_EIeou-Bt?tqdZz)Dzo4C2+)~_@vL6<Phx@V={yAxiV)3*OD{{`wln2
zHGcljm3@5{q5fB|@&Ehd(v5is?MuuxU7#L?H1_R~D4aZ$E}pTX3NZEi_pyvsY(P%r
zwxkkB0fw@_K3<G;92^i5nG4^3_C4rrI@J63r*RWvu@s@KEY5`ZB=IA)$HeLHzbGh9
zqIKopfB)yt|1QLT*5vO({NKDGiXLAh_I<QG&_Wz=Cr(3<A$1(*r%aH^WGo^#Knd&7
z@Im-Rw+ZBr5DaK>Jqe6TCm6Hrqoc`f7Z6{f*m$T-n4;N&6BVj)34EZC7czIuIYHH?
zSA9EFUai6Meq_5H?9<Z6fq{^BqTYP!27DEKsc^P5MYD_shEv^MLxI}>;p4$>M8oS3
zbwiQ<jT?`I#Fsl6@(K#nvYXXR;hrccmIKyjO@OP5X8nU5#EVukVR3*xKeBDik#?bI
zR)oqo#aE9_+vvU`cME)u0>cj=27%9UZFLlaf`zEJi9X}V>k+%~hQj;gMhXQuSPF`H
zMh}lQFt5E&R&C?JOX(H22cqMYX8-;Xy!FkNKfY$|YhZc_g7m7lZ)slRYi{R0L&ys<
z4ZEx8nAG}sq74oo<qgvKO&1s1oFFg-(NHGz4ABLJ#r+<81`NsCeE`7oWLH+4?BaTg
zO@H69nU^;U;vM7%{(&v9s#4l5-lvi9Bb3=Oyp4iRj<_eO%m=#H5%ytNO;Vx=mO}0E
zNq||1Y1w&>8~-C!i%u#)EnP@`QHw3v$q1fot-0d`58}py6U<K`7aM6w^j<Hz{@;5J
zyow9?SeW(MJMRZA?&eWKgYXc_2`z;=|J<(l<7wk7CaNmrjgw*Y)n3rDu4kY(q6ppk
zXE8Xz5Rj3T9Ty5z7R6UYjUC_vwHM>zMPBh!T=;8YX;BqJHsd8;XU^Ai1)_;kF8ml6
z20?)e27+6K$I&;mlVa%4rN_U2bXo%_EisbdiW7#q#dD+K3LQlpW9r;X>(vbYY=9KB
z6>b9$u${^91ME4SR`(rE7=0wezN!2a@d22aO>z4<k|z4Bv5^s^UcG&zxOyr5RtQ(_
z+_|%V7BuHZ%XRt`q|Lp=hJCVC;54Deh*YDhi=KF>Z_z@4KU9zAjqQ};_`43N_(art
z`1K-*yS*)QLgE&%4QWExuK-```R8jHJaw-FEFlzDuU!9g1H(Oe$!N9+B07PNO}wYL
zz2nc?xkA0ql!Yq!r>_*QK#xA|fm+HTJBl~o{@mhF>ff_quhZX(NBU-mg3c-qK_`%w
z4S4W>*P;8uuaWn-P-OV5ZGSJjhHc#3=^wR`KOfQ!*-DD**RRKH1y25JYKlDv{*oe;
z9&i_kY#pf=PDO$3YHYgl4~6zUV)CQ?)ZAQvzd!!V<pUacDf{rwq`xP?z%=H+U({Aq
zO!fDn)V?P^kM-{Zd)}GYpiA*}+uyJ51Tq9G(%}t-i-EpAjS74T1v~#4wl?5lT8xg^
zcGzZ=bDJqb|GojenzP^<q)NOc4&QC&*8OTwky%(;3cqYcA?5RDahcNoYx`T&{#{%`
z%Hqb2m$}gkTPdRNAr_C%U%I^n_H+f}O63xmmC$I!2gqoamv7lTMI|IF>ynK#M!dy0
zfSRFj?b?g{x8{6a_wJzr(>g+Z+i?)B(alB9KMaD6r1ep?y#~wx@iZcvF+{n5ix5x~
zcP))d5b#M({Kn1;q$}wcJF>Wcf13Xs*T280{8`#R|NhUP|6QN|6Kj&0^!ZQlcf9W7
z#}df()J+4oK|o3KdIe}xQE^>XFr=M=qN5FGs`0|haewoSHz|1Rn>TE*7I#G{3@Rk1
z<5FOYqUk?@o0w(=0O#Qsf+P%xiiPXU*|R$SpJAEfqYwW{Q{L<Xr~?!VyTwL30%UNI
zZjFMCfIv`?D2o1!TcrbJ6)^O+d80HP5K13sSPcsH6MuINohd(w0b0(qwAy?)fq2ul
zzj5meoD5W5D=h`so)KR+#`O2=Ud+$mz>$yY0GnfIcvycOaqj--UG&kG=K+=iW8Hh*
z3jg7D@ZX(!dpoNB=yf}(SM<QesKN58wY4>vb<ibPw|>2EYiQb=H*YX3iJgy+5A+ip
z%x~lW>$h&osF_DuBh(8klMeFxp(b?j5o)jC!b54!1o$Ap+_TEcB@n^~)P(BQB}GpS
z_-v>jg7=|?2~UvHLV|o9fH)Qz6jCVJRLa{U_@J;b{O3YAkfq1JC5iDFq`N)#z=iKR
zO5Ffj38rZ3BU)ojQ$r~PX*ftaP@Yd0n1?KK3|N1l5Ftso^+JF%k}nBZ3piU$`8did
z2q8cyZD`Pva7Dnu{Q53F`?4#<#M<eP@g0D>rMMk?%&7R@{rjBtkl952;|1ax4n|I`
z^I%nnUxGL<sL|-SGH^!uh0sLAoD(!<_Vb(*2gVGnysc?x8@vz5G#><g7>GMz`1|x+
zwaogkBYNGJ956=*jFhkd%)5Dsn(?NO#HK=9jek2aJOL!<ZKv!yz<Nf(1$4|Pfy&BC
zq*r!|5&!GwJps24Xka3DFI+#s?G4g;;MR@Y*>v<75NimYP*pFrf&AAW2HY50a2Hta
zU+h2!F)Vcdp9lHxZk5smm6=N`C7#>(g?;3#H_Ahhbu-X60XaqFw1FIImmQQ>q5bCq
z;1;zA2tUMrJGRyo%ApC+xqtzJ->44a=l%QlFA}fdjzby$Ze(kX7%dQFOV?I!{_KLR
zgpU5v<Hx6<r`dAp6`T>!4t2L7utFyRVuCY~kdOcnj+%yM2E!oGo=rDdgeD+NmQPw#
z!|VjK1s*~znNS+!IR{h$xcNyguKFAO_(pbPrx%PHod{W_{52QsezjMir^Up@WgXl~
z&_)e103@XysYa=Eq;L%dgj5D>O8E}`0g%eu_s1Z^)YNgo4W|j9>Ce_OVra-3t(2}c
zh~%~|)ZqiaO5~v|k@3`0TKXT<3-O&PH!*cgn?Lgby%8Bzc^81>76s_1Mc3O<elMhJ
zJEj36p{-6^f3<uwBT-6m6+hKH4|o*&Ba<7m6S^=7Z~H0u`rD3PgtMcdpk)ggMhB|a
z<6b;b)RsbN>(-6xV7~!OF&GDz5e;op`dC!Vi^z0+P-VIRjj|aLo)+|;b~t8;tIskL
z!#G?Nh|rPZ_CuVmE;PyNgG^{J7#oeH<5E*L%nXZxwgq7Hw6o4@!050IpE`&BT;@dO
z#LMB%01w`)Cja*%wsthiZ}j&8CnydZtuXf$YWN7eG#uVFHK`Ve2?%H>>*Bz-xU^d?
z0UiL>XyA?Zp5Bt{O&5r|JG-Rezo1U^Sya;yd}MJqVXOyeFz6cz+X==Yw}^;=z655~
zK<`>!e`0tzb+LW%&KS_-5iw2_sLyv-Z<MZSY;?YQm46=5BnA7?Kk)-a;3yiw@ss<h
zqr(|y$jcUYU;?NNyl<V@$jDR!<5?GZ;AWfR1Vu$lP<qLgK`e;w%Zv-fzP`TU;o;@K
ziQVEJhzZ14uV@{GeWLg7-Gf*SXFb?a+Avzi#>oOJioV_MdU_G%gqKju{F6FROzlMJ
z0O$PL1<y3;1++^*6_BGIdKphp-|f&sps-D17$ugiAMgJBf4`4l>h_G)aCsqPAecis
z5tslP8v4v&4YFbyJm_~BMZ{iyRQvO^)~<UXo1yr@UXLm(j-NyJpCiWg_lPZfuIDe-
zb8&UW#3w2J%rHw3NlZ}i&8_(Q^((r$(k~uNBc=ZYMCXpC<y<$p&!0X)7Y^-!xZM;}
zUv>N+#abJ-0%A7ydPw3E@Nbm&()|3%OJ5B2S219R$MB_^4gmoZT9zpCs&&Rmdbp!D
zR*K(Y9U!ff18)cJ)33QX!QYBo?jAlUaf_(RL6JyZ;v{5}aAJw?M?D@*^+?Os<~-Jt
z7j@xIDDEEKvhlVNA(ZlA?G=8%{0detea9LKZ&m|SFxkLxH4D6uHgb$Yk@p=_VjIjl
z=K%~m50~8!<1VFctb_Ln6{6ZcvTe)hKeKedmVNsCd5!Vr+@W1nEo(cMhxFK)jW*tt
z_TT81EvLS#YMreTH+y+E?M>A&+MAWsRNZ#m>N#f8QugXv$M?!<g??F5pO|~+7Pm5<
zmse6~*xx)inq_wB+Qo|jQBezHp|B=763)=;U=oI*LJA!@n+g(WnAl4ctdRMAW2Nz}
zs@P+UhC8raD%!#^XCW+%QWeqy6coLb|3-&j*KXXnk$o(B8h2b%qjV=STEzL`>`QFu
z$B*~+Ux1eY$p(<>y#b7gJq{}AovJ`sNDVMM-@HW>gTC){H6pyHm=c%zb6P=j+1d3y
z0j}e*63aS%czs9~9dr5g<x45H&cIJl^C2TBdgO?w<j0vAqEH7Hy_NgE78+ntd)LyU
z@yNvwn9GQMwAil_pFMx}Z&X7;!5pMyZe!B}Rzzp=hRh*mZHPmobEz}=WyK6=Zcdjk
z3%N|Q*kzZJ$zy2HM~Z-*yk}E4mEC`F0nTqD+1jRsgtWnmgT|u22f@+Xo*w<LeUpD~
z<4|kH7!fAnC@KuWE6B-ha1p~GrG%UIu=t&+n%-itg_((omBp(!P~b)d;zvW|FO;WI
z^#4#@{X9Q^21gc+0{{(ApFR7St-ijVYWsFpRgbk*aaEVMra$3$L4GF2x1#pfm8$vf
z&6~q$c|LY*0$rU@=}yo9w+)KzAy*br5j8KhAg*Tqz71tu6s1%bRu=hy)-?hv`1gl~
zQ1Sx{$Wh!^{=bs}{5PS$(b>PRdPDg?*dzXQ_OAi^_di~i`S<Dl`Th2u|3SF%hv>i7
z=HHioz4zBq{QK+Cf49=#U;p!!|5;dy{}bCM<f6qsic$fDU%zH%$YTVK0086qH&8@T
z+C3VOXhxiBGGN0h(<|~#*zTcZKxha@(*OI#OniL7T;NS3Pp1gPv^{P>dcMWP2mSl=
zpHuR`&lbdQ04b|wF$<kXCAq%8#t#TPK6}g!yUO=Vzx0-HKN(D8#qEQeDT>f4i1{S<
zRtp@EqJZLO5w6Ta?E`l1(8jzMFOGpuGA8p6MI10zbdf9yUWb?)UlU15DX?b;08w4p
zg?~7SA+t$qBS;HV#&T~3G!Iau5dAkb)|}BO80S)Hn-qr;mNn%d0uQJVh_5_IPEJlt
zG_wfX96A9VEMRyLD+h8)i2vnux3;#{)ukaa0h*0rdd+QZ>7S_4lDdE_6tBZw;B_dg
zVqhs0mk6fo)D#4SvwVC`4?b2)L$b)nH;wKS1S(+QaBy%yiJ6U!?b@|#j~+b&sc8i4
zR?r<fH0g|tJ|k;ISgxUn?1g9;MPJw~&Fy-GC@5RYO7cP3sK98iS^@YW$csNA5Cag6
zWU98Nrn#jhCpR|@4YLSEy}cDtKjkX%(~fTqfgm@~wjz8J$R_CX#g81bIxZC6pXld;
zlo!w^8@qIyd`*;%LCXPqL+VfuG73QU;De;buDf-Uq^wNva3P~Oax^hNzrCWOqOPtE
zUkY9m;J}mv|IplFdsoQA%R7mMAy`Ap^H*$bZLur}@R5*X`e`?uy{f7zg6BFOlQ3(V
z(>7WmmG9nJ;fW!>OMKLTl0V$v{rmSvIBq*CQ4F*K1MFN&Av#<alMoB1gi~$VqBdK9
zXBYHCo=b#c(xUFQE2VB2t{kZU8l}bDgh(;><p{1TFdq8G7I;cxyg8S8C8k3Wh14&+
zkj@6H9Q`^mYm&rh9X#qV?Fx|f2p+p`_~f-xAD>#pim)pdv>MDoq2b}+{ODPly+%DB
zJ4#7ODaCpfwSvJqenG+TQ8G3MLfz&)=k)ZZp6Ra7(Q9B2E-G?CT4&c6&TH)WdK7r;
z^sB9Q9o~$G1ccB71-`+2*+x9~hmRg*lBsRola7W(MS-E%ipKM)eIVNuW65g)%37}?
zt+i;xgAMvYMkmgR;${>hmU(9uep#&+9;Wue=GSbQuFs$rWpjf5G!me641Zb+J0Tzd
zR-c=?dXPx{CX%9CP)n{h|4HFe6QC|gEKUgs45rFBJ2`;>6Z-hEj=op>vO-WODe53L
zGHL@RjhDgS6g#?U<Hi%Mpd0Aq2y#juIdY`4FdHiwn}?}sQeiZVjH)X4dH}y$?Ax3c
z3nF8T(3QLH&j5LY4v}Il0lMqj^q}u_-9}P22yE!<AvDvLYw<sVASXe;P>F?lmc|bx
z{#GcUfn*J?+%!x5`}eT|c8~M&t^!#7P^S<c9E=n<oS2sYw0D^UefJo0%4yEJp%~E<
z_<DMe?hX$PoryZ-@6IA1;EZkWXr9CkQXC#!JYysRT%v*ENfQPYwy*7^6hoj0@*6h!
zmoks0I6f8@f`N`yLRYnX=ww-Ih%1U$0IiS_$Hv9s1;~KV=w_}PVk{-~2AKRHyB-b+
zG@6DK0VwT-%#@-|Zyq>Y;J*aWonV%bh?TSeGX(T~5FnATCp}K^ky{fDlR)3TLsq&R
z2nykkJ&;v}8SPCuW9@h=9X%Kj#=$hzIx?9%)*RgdNLIm<U~LkorcI^xXXelr6s`_n
z<Tc*)=HthY7@O#Bj{FEV$I)C{p9tvY_vYL81Ye2D2S-C}7#r`s-kG9=-BPJIrC7bX
zE6gc1SY+HtC1dRKk3a(eU27Sz+<))@;tT=p?OJz+@r2Q^(9U=o`29HAoltDh3iSXF
zrLge&>C-;9k^|`Z@1MABin|Z%*hK4Fg`NGJu?0ihpe}GwUmUHyh_OkW``FXS<c5zQ
zqpmcCDcabif*6k$c`XCZO^sPI&j9k^TY`gw@!4n2G+>nmBw`FdMXtSjjh*kO;|(pQ
zPg$$mANcwb<LRNQ;*5~q1@;8dG(^py%!KO`NAn?wuqifrFw-&@*C;3sJvrE$1~?WA
zF7CFdbzu<#E-=?(9;=2)NeSuyE+U{v$QN_<79suzR`ZMz!;X4iI7IQ)C9x4P^sVa>
zx8~o>59VcOPtDAP@)lwd(0Ef;$G(r96Fqsb!nERCfSsWKlg7T2Lsax8U>?i~p%od0
zfkCgUhI)|R;P>yvB_(L}NS$DX-I)E}c%VpCkdKcL-WyWn<L3v2Mtdw=3^-3@V4##_
zD^8J}o!x@tIf5$Bt)oyRrH-DQ^r08uaIA5i;I(BueE2qgAuJZOG#I7ukr!oL$BzB}
z{X5gTgVdF%xU-sO7Do?P!eN*{DTDqX0H&!`=QA=gn&9TJe1HyKShrNPx97tVz#_wi
zh=4>3Ii(bVt6<w=M{6yfTR-J59nybJ0EQ}3_7y$AT4<@jp=fKdY-+qjb2_!ScK3k;
z>0e9hkiJJDv0Ndv0^5HJbM8P!a(nVUAfo!Zx*<g<9y85V+hH#gPM@!C<HINXRy%}6
z3y|!B!P?l+fW9QqWr>;CE`1o!b8qKP-M3(V)b2~eXkb94rMDKC_xeiSn$L#Bl4Iix
zDl@pl)Tj3}K++<sEC99z0Q%bH7u0AAQ&NV|D1r}$6*7fu;|yJcse=wr;sqvji&qIz
z@7nby!XkkIA7OEjsh-X7)~#inUmDI`?6S7DoTDFBz-mIV8Xt*s1>Px(U<)TF=YGtP
zF{snq$SEGAg!$UXp(V+uk(iMo4j2L<f!peG5%l&k*9WH>Ohas!%pu-s^gpcjRZ(rM
zKi&IMFU{6}N4gxYSu6DIhYyP|6JUh)kv`!(;+PXcrHfX8JF%SyVk0+9y~g@S5tFNs
zF9yF`Szbnnf)f)R9PBvM%%rTWtZ)~`_Jrlp6eW02AFbX2b{?KXn2&V+d@Qq9eqrHc
zPu|fLC8V<mWoT<PGeTdS61`=WU1dF(khS?;**|w(FZikp91(oYw9ok)X3CwJ*teg7
z;l0P+uDyw&um{uJ_9xQLl1JC?r>6%eHz;TtEHq|K45L@*VCClG!r3@%{Dgc`FMI*_
zQkbU9PP>0UskAsVlaOXCI?KVy`2hxoc%%by5fMsdJ0QwV9@mXLcONoLAG^DczX!Q7
zwYx;S8=H^L0{wJsYn?89UAZ6zG`)Y%rux?1ctq~fMfHP+4xxi}s1^;>W8>qsp>vbH
zy3UjWFw)TH7hV`axFuHh3O%fN65^M?-VR&_;v~)3VR7NYo5k|l2JdJ+A|8v5K6vL+
zh++GnXpd=VDW~s?7uP_}4a|e@9+KQ(JJbTM>TOp@ihN039Mrm#W{bV6-eVD@gf+8}
z!fC)Elt1q<9W@Jod+5+11wOFXGq+oYDg8PiuaS1q5u<PdWBKhuJUx8XQ^A7df4>Nt
zuUDChMfu)D3O*2?EI?lXGzl+p^5jWqxPL*1+#^QGWAD0O!&qY{7r+S-tS6vza6Z7|
z^(W(XcuaHVK`<nw*<cQEWY7+E=FFLAd1`=s&$nDTxApuKnxsI#_S-K2&<#`kEEg9S
z;YUGrk3~-fOg{_lr(FunK|tcnP<(=tA<@2wMFDZx8&9eF?OQGO5PY!;t^qmJsZ*z5
z8oz)4PJ99epdsM#s?+p=YHM(8>{&@u05)j9+L#xEQ2MU$aY#b)7!~JZL&Boq3Lb-#
zjPM5qZI~0(Pm|}fOJJ07XV~Yph!>livti~DNrNH0tMttT7?!#j21IoNH8SNjc^+<V
zcuhP20yvz_8gxUvIsFyD)%e&L-Ddr+rq8gI@?|=-vjeqza`!`al2X|AQpG(=MAUws
zIJTihm!dR4kMt5lEyS`~#@z|`gMDTx9wf~igmrr8eh7X2r*~xTmQhx=L)pg96JS=@
z5ia>aU$d7bC3!E>yl~3SK{|k0`1pgWKE#-4U`>QH6U?oGf&xCTv$GRRxF3W}l)*~f
zSJJ*Jzd|E2H1APz@<iZ|fcAiZ?ZD_L*R8Gc0I@G$p2C9!GX;tSZV+sP12}i?BWxw6
zP7VUL_v4VWePfV>3D+Sf^xxLhSOax7Cv|>IsY`;*N1)_eHK3;G=C%SH7FNqK#=^#?
z$bH3?NtPS9G7Ox&#}Ra=1ZAaw{*3+$;T2+39iHz%{rFN28f-^?qQH`3@_iYF446#I
zktUR*Qr1MzGa)AdLj>IV+=2pulP3uS0{S#OcIlmsClDhMd_6eT4r~Z8`LVrl>hS6(
z=n}$K=h9Lhz~mFy0oY!oG_A(;R#rby3EXn9NZ-J~6?P24AA-u-)>aK;V@O;Zpilz=
zmLViARA68%c7{%;y5WR?DT^1Q*C;3_=L+(n@$qrgqw-Ln!Krz;_q4QjLTwNn4hlLb
zfUclTr<KbX!8E*?^H6hnQ`6)?ZOGQl7feA)5H^Bn73%tVNV}A877!)cQpX3+1%#2r
z=&$D(NFc^gV%nM6K}>*wgiV>2_RF<a1i?zPDvze7E-wb&!VxbLw4fKdeAvt=99D7G
zX(m*vqs47r$7bHz$?@_GC_zw3*RNlWTU$OVEF3vHNT(kb8agGXAto;FlAEBY2S@Mn
z<%oDi#f<<|!S+ox?)9hVSEzPo!Kr~X&Ap?l!>c^l9DG^9a(AeX1u0>fXK`;o1$`hW
zc7T?)Ds260Wej$W3=N$@T_yK>VtTqL!YnLhWEUDDMR6DX9Knc~J}Atw^X+#-Q0aFS
z(XB-+>(!IRN-V{fzpd-%UQkGk6Te9mANh<kaa75&Updg1$kc>Zl~5%Xlu!}YS~b4F
zQX^*m%KHsSKYEepeXVS5i=i#h*Jm2prK1nsGnKt&2%pSJIeO%bMn}dKggl%)-_T}L
zyYG<1Eg?J8&dD)kPud1DK62AP`xU4A^6{;&UKw&M@UtDkd{~V9c?oMb5EUl8--tOV
z@Z(a%lP6nKn?qA@?|yuczzScZiIlBGeu^!S)T<*ReE+^rRUa71L6yJhXJ%)KIdxz@
z|1_6h<wTkD408~MkWSm{4-XD%>*>{mIG}F?w^~!fD~mu?HCm|L!z3&nL5KMXK>xDO
z&0h!EquY{vY5P8eqb$^vqcIt{E)K3WC^)oTjTO4WTp%MD6WB6D?PCstncp7FSvj1|
z()+`D22ecbuK0|MMq}^ep<!V-65*U9y&h{S&!0|&y|8J!FDd9sGmO=nNluY&TSiVP
z6U*#CAlH~JWJ|Uxu<s$9vSAW2HziEt`|&TRWRNsnU0hlM+Rn|2Apv|FCz*;tS@`0C
zuR6t8%gviN=f%E%Ndg>#4lC0}gb+_Wp2$Gn19B-SP`i*!9UR<HJBU3KI-9DFs199x
z4%XH++AL;f{V1f=HYoa>d{qlp0?={u%=)vbMkT)yS-X7L1U<q5ee_}nDzRX0!@xuD
zk&%(QXp5vsl}A|OAdBpr921pSX!GF90YZ{SmAc^1?DH_{Ah_UucYXu?Hy<CPqW8zn
zh2l!15TtVaNgQdH>HK%RLH9XI(T)AZKUaa;;o(w#YJ1bwHHWs00oUKKc!;)a&pBxN
zAVERtTd(ANC~^WgAWHg+sb4oTirtt(v2<QC2r`*)7+&7q_1s`aoW=Gl(tv}*G^+RS
z-g&QOE+OVXlf3-owm|U6kVea#JrxRhU++$jpPzix1$^WjR~DVC%aIGzgdaG<3CHZg
zE_CB*cJ^horXWt1c_9FFWEicc`LFe!>PCXCgx;hgg}dl315hB&#pOGzpQ9&+>Jv;B
z!qLDrZ4@=22i%IVyTc*){TqW{fo8?>%t5iS40e0s+$GP@-U|;LeTm43+6T6%hMu2@
zP5mMQRh^t>kcXy7rfa8P13Z!``bdTVBh<5z_;r2>e<-J@i1QtJMXi+6OFa(7Aa2MS
zd~sZf!}D;S{DN?E0aUhS4s@wtc@QnX6Z!B!5j6~7U#h0ViNJSu(DU6J?Lq%JNFzi>
z@|$TGt@x<RNSIV?D9Cuqy-Zc_dR$aAs!aW<5Ke#`^X`QxE^439j%`3^dQm${l{}Yk
zEzS6qdov%6JJM;+{L%2m(_JmW5?E?M`u_f>85t`uQ|cbJW8_Y-1s-8ifE~LN)RT}X
zG$;gZ3RYsl!osdBJ9MGqxf8|2?^;!$MQo!xei){5wg(D@zyW@Zah0ON1+v@J&I#yI
z?uQL2;uGSkk}SXtz>i@2{v0!k4hiAE`P+W@E6;QfD(PvHl@Y_tjN30c5{ikiFL0~D
z!j}$<i^Gi!Vvo>%cHv`VBP5Oy&y$<5TUPPZ(I(p`Vq_L|P&>gh9ve9b8yTV2;{4Ae
zv&mUmlZY{BYXj{aF*I5`Xj{i76Q}v#@N*=Qhvm?O6cQ3*WtHdN>#;aP%$z`=)q0W_
z3W!h|Y9JHsV|bcDO6bXUg=i)s9&Br(EDxc)55~vX*epRa;k+rxseb;v1T>@4f(Th+
zh+nH0FJTg>I>=_Q8mK*03RK*c*5CIEM+qW)#Xtm5DF21VCn~D`zCOaBG(K)@Y@Gf5
z`*D#|H3TJBSLTcz9L6!fT2Vp45=-{|y9DqriRA(4MvH~S<GB;Z-sZ?TXsh7-5eZgp
zt!a)BG-B}Y_Fx(ePe4ARv9?X|8|aU<3bz~XbQnTzT6ZxHg)DNV|3*$;UeG|5AcHZb
zi4YD2-<J?z`<9GCttl|4Wg7{P)rMjPe>rB*3QUal2Ph-FhEsCgd&A_AU1UgzECV+O
zhlxX9dhvG7SN9Cxbu^PK$D>ioJc|4os|ml4F5ySOY~cC<Y+&L{KA^9LhBHjU;K|&)
zd9$q~5rs081l6A9BRCxxiU7RC6>Q~&sLK0_oXC2kM`UCRsZSqh;8SSFO<EzSC?qN>
zd1Fdyo&&HJ;!<+VE*+$!*qMBj=M15XfE60}2}A*=0Xs?e1|_oO(f<A;Vy~I+d6*QH
zRz1;7($YyfYMQQrIHC*PIap0@M9|>yv?d)uBsKo^>v1&cA-vqW?Houb*8s4z$00_-
zvV;h44QA|N0HZnO;ckrrsIMh5l)QYYHRI&EL9M2F*X&K$M?68eI@nEUWgwtJazBHO
zfS_2j%ovL9h?D&`t{3P9{^{xInH1n)Zx0=rHB>_}um<T1=vL4SUPgN&A_r>6Ju8Sa
zkPUG2@DL4SptN9(lR<I@c@Z&?TA3YETe7lvGeQcLXrZ65bAVlexb5|=f>}W(f&tYj
z`E>Anps+yZw696=r)mb;lNuYxp|BIQ+Jg(eefu_~Bw)0$$8oxmas`dh0V|{&R^&Px
z5_J$bH=>Imm!{YWq~M_O4EX<Z3?Gj_BG&w-`U1?CMK)U+m5Yw-U!;}JoLL4;80<S+
zcCt*JMZ+SbAE0xqP=ppX0$2MHOZij6P}EJYmN7|VWZT*MoHIYgSBa&yrN!hkKM|?4
z&j<oLsou1ktnVX-XS6q{7LX!bsaz7})~(EK>=x?#I0i4CAWg2vhP}%bsmJ{VfiXO9
zO>*dsa#reEWZuw|uG99Lo_2KY1`wNMzQ1}#711`@;X-QBx~1YQgjf$ix5{&Pwe1Cq
z-v0iLu<82B1?Z1Ui{UX!3ca)H!4(slyrpQLGm~d~adENCUIr7R_CY*MzsL$iGl{C|
z<mQB9fCfY(_2Th70_k&)jW7p3eQJWF;z%XTU`=?`e#`wNmSCLc5MFVh|9q^vXPzNu
z<K$s_y&!LcHq*%5moFpilu}8`>`zrVlZQq}3F#QS);J}WG>ni$LE7S25+9n`7WJ_k
zl4FB!s_*lWlgZ@fy|sH&RnSU0ta!bcuk-x^6wX0%pxcb%i`Ak}g_Ry~rC;QAwN2gK
zLYVB`pc&Mmi#V`m+u)Ys)2SxTM||X3^KCSzlYqUSeLk=?o@`J+N@8taCIvW886X}-
zM9}boDID#^W-IE6dwi>a1jok#@!2%2`H*`A1^<*QOf+{w$Yo9PMh0*)uRB3_;dFJ?
z(b?;!Lg$z)mxNboKjl%N&kzNz#&L65LIN`i2AqwU221*$7%FeXUiaCwy`w|>NmNFL
zXxMYU|Kb8TXlPLD=3+krB#4_m8ha%S38f+%HurO0!(E3CH7;_ZJ_zLVsH&%!kWj?p
zcE@YiEL9T8swzkefozicF}l$p8?A_nii$S-jf{+@;~uV_<272ZBEA3Pz&ti-U<a>0
z67|Q+h3c0LzM;ThD`V>z;%Vpc1g;>^2bQe<NrV-ZN>VdYiIFD<VGd{9BQrCP6_CQ1
zeT%c-OYF=25=vcLm_k)CbT~PUR>AnyjT5>TXF3Y-P>%9*#a`BPaynU1B?seQvtj=M
z1@N|#PjaEuweY)0r*GrF2P3*r3oRs*{eE&E@q7R4SD95417{B`S?z{)QXjh95g_$*
zw8BE9h64Aq6ia6^aT@+gQw*$si*QhZvmS!v?u>IR)EV$iBFZ=DRsgG$nK<q2MqzT@
zRu&5o<ziF_HZ0lhiJ*OV;47}mJqm~qmvI_=A0iv*!zAK2Prk!?%zQ%fTPIT2R9^lK
z*%yvCJd1>cJJMgWuM+8pTRP2X2OtHuwF4dtgZnxQ2|<jo$Cyd-c-FaAd)CEr8|0zV
z^8+2P-$RWL!RyQ1*BFTBCKnYJrcEMkoh37yI8Co=f{hh+nd#mFtQuXG1cW;yt**@4
zu-e=85doQq`i;2XxEwO0CR%nejv^ZU<9d4YV?F$?&u$ZOVnYa*ACX-lk2vkUgo71f
z$F-yS`Qe!Yp`pdON$9j?z@nn)26vk`F>h{ff9A^%O_9#@&$wU(3#bEk#Y?a6oa2M=
zbJ%0GUtHo$cGHfLf|nRn-}~{Frl)fi45i&QadNEx2@>j7;h#&}65BlZ;EGc&>@aUo
z0tbW0fop3`RCM(8lo}tvew|iQy1>#raJ(4xz>|WV|7gfGXCas4?_mr|;0<lyH$2Ve
zsu=7%6p$-XfLKC9x)35uQ#2e*OywwlK5+y@;Aa)apwQLLwz*5p6ImG6;y4Dq0F6ZA
zVA$yS=8crXeK8ZKn>UvTup6D*n8Ek*@}1JMbza-#tldLAXZ;DtA!rO?pJ5bz_M=Em
zW}v*ODYfdTU}T&VbnSzM)okLVR<JTUPAXn0`2hBgt+rRi5Q-{&lqHCf)i@A?RmX)r
zFk__L>K1as@t3Us_C(FXGbj(d@7OSq{Fdt2F1I>r;apD!g~GhN{{6p^Fd&Wq;h9av
z#@zfZ-R7CuQQeWw2XS$X9zjr5NIguokZOY|T9uWaS4^>^y`pPq6#xxd+CR`r*WnrI
zJ$HjHMuI;^qvV={;~8rur7fd}W>G=n({S`^0e!GmNf><#eDsOpC#>Ed<Ws@Puhshj
z?tKl%n{`=v%tLFLP5`V#P)-G!f%((Gc=XTR(IBs{4D&rW%0uk~ufBWNt|_3pNK2qC
zN+@QR?oLIeFfISy@;U#X2aaE$u%ks26REdPLs@wg`V2E09^PxIJ-6e-hYz+NKRzCj
z)EsiKJ0|94;3#4XXPUCdYtBah`aK8)Q57Y2A$_X}#Y@fQ`~>_64ieqnCs9%L#fcpK
zvR2zzA5YKDvi4HB6l$=HiEuq&2ulQBfvy#9Zf+2@h>Qh|^>%!_-@Uoa+PX{(=Dcah
z%kvf$VZFTV+qWmJ?ZBwPD26SX$Ng48XySj5a4$V@Pg&)0q?^JkM~^^a@6nl|5Pg`S
z@*6tXpv&ik7(}o6k&1w1Rp&i9sE}N(09u7k>A@jgAFItR@Q$`CA;^;J&z(Pwpp_`I
zFCd5VEE9Wp%N8-vV~M2X<aZ(mBQBy$()t#0n%^NBVuLoYDH0Gnn;_~Vrq94-!%1Y;
z1YJ2I2po;`n60e9AK|Qd%)!Ay;ef`ksDlK+f5`=K8lbG8%xA73!^R3&>TkZ!74%*3
zp;r3<=3TOi>mw%~{&kdwG7%gBN8>>_GtO@YTki8H{L>8%W9|;)Fo^XEcRVyTQuNCN
zen)O|VV_A#pe5I|*LHU5ETW<y_0#_lq5u~mVPQmxrx`W5=5nAx^qsQS#?DT&zY&X0
zAN2r2HflGw)*2pwwxypzVNB`SN9_cj_9WaY+(7$CxIefFIxMmP_7rY%K}or{w=7|S
z;PHtOk@B=V3`BZip<txm*RqR*X>Tb(aC1bHkBv@)%DNTMNyJDiA;43!a33+K=IqvC
z7)|6atN~?+FHm86{`@&$N2H)n=VIv0C*qTYI=#k_7eK7j*3K>+`u8AEBHH>DDwpI1
zR6yH2>JwyYI>-O1esoCL_!u!B+mK&lP|Dp46t<%0?>KeA<SSZDysJ$Dvos|;sGR`w
z7Ar#|B5I0{%m8Dosbxm37#WO2{{<HpLbUhZt4=g?(=A1H09o@`TQw&wdIv{GpILQ;
zvCUo;!Q8h9L#uvZ3o;LrA7Io(Dj?~Bp8zBY*%BZiz-!Rew6Dn^E-h07wbuR*83yrq
zYQZOdBL8J%WL!L311D<sn)WE;G*@(BU_7~@>;x(@GjZE;#Sd>-;m)LW<_an;l~`kQ
zlw6tFGrKoUtyfWQbd^v{-Kw(@0Ik-(+n9>ific~~X7Y;?ST(JCkIE(!zb2zaLtuD4
z*!3fvSBB9!HzSRf0nqz~Rbk5>jw~II4fd=)C4BMAtJ9g^?HwFcE|y?F7ea)=F-67R
zp1bRo%6-@a{)skG5s^Qem;CBgbm>QjFl)z#D8T^d><R$7QP6rvgAis?_V-s7_+}~1
zN)A2gW!I1o=|>a~7n;0!^3*8_sNXTmKB7yr4p~nmnmJ$GlB}`e1k24g6`3E)Uj(S=
z{ct?ZWDlhUrd1?81RDApa?J95ZlqUV#h~7ovRxT1Xf7Ic%k4;IWW+6R*=n6M)YT1l
zwm-6-_7`n8okkfZ1y=8ly~(gn1%<}l_db-m-Sl!_hn-SaoW9fUW%TIFH{w@OYpCsY
z@2tH(v(^O3qTKb(5G=!u<dtq+i6fzu>-VZC!3nMZ%m>;kg2pYA4xl9vks`BjN`9c&
zxD2vSLiq@>CVW2!2XtuMuv^UuV*QCnh(o+fS1s}G(2_m-Ms<aFc~yOQ%e}~|a&;*h
zAaKz$z6BsMBrO|RV^kx@9OUUKi$uMqAMrDUf|&xS$QlIh$PlWZq_jDkpZW&Q^)X5m
zc%(8Um0cERQqv|Ia8dzT8wc@lb*Lp--$;6=nFKLwJM<y6j@RK#qFX_#S_jZC;^;;p
zA=C_jrl@<MZg~wyi^R>pK&zwbra`L>4#Vnh_{jPo%9ZJ0PX{g|#03k=7S`5E1%c_<
zX)M%5B_+xPI7P^-JhV@$@15K2BL{?`b@SHFk6IGTZjAC01@ID(Q#j4I_Cs<lNS|eK
zGKN$lshwW?6G?y0CNE^N4WOtig;aKTM-))5IGhwl@ImLBs}q{$(JWA={%nR#5p-S5
z9T{p_111(Et4gCTA9Zj`Vr5}~!NMb|*>TiK22^V$X^dE?het;>e{JW=*S<cOo0IeO
zal`xCT8u5ZpOlmo74;FBD>5|omZvA`iV0D61p7b0bHy+UGaDQI3<H6THTlnvzONaD
zFQlsb7C(5PI*!Dqbu-=#RJdz}H)gGMmDeT+7zevbQ^8NSB-EWP51Pby>oFnG_3Js0
znlln&+Sv7W4MYjwa7~$9J_`JXssBN(wy`|(?$KMv7Mp`nh7O@bJt3>=PU&U}uaU<o
zXW^1i)F<({=&QG9aCEe^wu)e|(WjEKk>TN5!B47(hr6O@Q;`>u&%Z7Pgp2SN?wZgD
zLP@^X95Ly$OA*MTqfpXP_t;NIcTmzjx}cX?u+#lXLmYq7;n6)od}*PW`f^<72b6m@
zHr+O*whOCpZu1@GV47P;;P*{oStQDB+Xm7!hLlH1Vd3GLWol{t0|QRbYp;IcT#@>e
z)G~zxC*3?1=I{|ce>Ko|Rn;sM>QFWzmqo#RCamQXirt8S+v5OPAy`d!tUsf}F_Nm9
z#>~URv$&$Qps=pBNuyHtd4l<i1o`~?PDuHLxpmD@t_gT(3%1T{_$SE4(jm@9?FF%_
z{QW~N?CF!581oEN0gT{l!>yRp@f?H#t$nf&Km3+0FAaFGq?x#uOXOjez1|F(nl^Kr
zRdyhM>6}|?JSmKZqB;>9?ZX1?ar?ipYq*A(!KKdJH2d=9i&*+VZSuoY!n+i_FEMt-
zUSNsb&K6tjBU6wJD!|6lOg2bFj{`kO{TPg6&B9_1eJUjIC;0inApekUvVL@;iKUA7
z#`*JK0106GBEN7#FB9Mhs3zx5NHbN5h}~iSrI~pB4qe>-*TQ-Vnybd`hVlxr)BZ{t
z>!;h)N#0hFIQBD%Z9V(oO&i3mYCOf<s2%AAP-#akJ1B_hSApL3?4X{l`{qIVn04SZ
z0#JfxqB@bGSPj(C7In7MZ2VEZRSUH^u;Jj8?pvj3AuK^>os4_4EjyE)|I_4TK{SU|
zOIul4bu!7%>~=ts%FWImzY6G%INo7et!tGAf}DJM3PPP9&U0wjA%mvBWP+YevXQk8
zDOKa%t<elJIi<+ZsaGSc@7uTXyh~}Jh|@<`_l~4$oGfbXx4pBBlg`9Dj6AaD;|<6a
z&78GYQ`<GGr>aVHBRS6fC<@em=sFU@`S5&BP0o46vuAg7IAOmB-wd$pNQGt5N-CB_
z`3%jW@rrj~*r6RJh!_+og@L~Q!s24^PdOdZu>C}LPUZXe6Od`eo(1}IhC~XU5WtMA
zfA_+i8_0fMb8d2+6RYvH<sZCuyPZ>cgcF1ox;u8hXM>cUfxCeG$(EBc#`Vd;C!g3g
zY(gr(d~ruCMdBP@)BR&+)@xA+FL-G^EvI?{VMGku&Op3}b7%31gNKXjCZvLZ?xQI`
z8$NZb%GD5ATsk6Lj6@)jNE>1+j9DV%_BScAz1mjsi)=WcbUx{%OF@1<B2r~Z4ILeY
zW73(FddVJ~S9}B+mfaXw`A(dGWZzd{dYD)84ZXg}XAjBBR-#G11A~V!unPPva9E5G
zH`eR{B+;%(xE)kph~!oD+8E}?Tn31V1wn<JoRNWcW)wNU!{ngzBv9p=Kxp6-b-Au*
zr}uQmOOA?NTrBqYOS=HUJ@R^w8Wk$|9gG(bJIul1lm2GrxVU_h+3y8oc@Aw70N@m1
z4GmZtE%{$Y(uQrdWjUNl0GUAi1kMEzXRP_D=gfBimpPuEl|Zg!Wn~FWZSiXA3r=}6
zE^$#&(N2Xeb{!_BrbyWKObh9Tfq&VGv@%!1q;UYK+EZ%u#zZGKn1?0AE<SshkmY{(
zT`(Kk|8~wszOd~e6<rMr?K?2B^gu2Fh~6tyeGwx9=Jvae?1Nydf&PM8V42G=@H6PK
zgOCY(1$@`Zo;=dG+CDI`G&PV)1De@Ul^wu<!E%TJ(nv@MLKhuCU}ok;*kLRxF=7XP
zAAo+a=Nv%h!Psye_5B|SyWu$jpFnu^5Hk=7%==*E)G67{M%QS5=ln46_0zST6iPCd
zb{ZjLH2N!~An^&BpHQwH!}CP)#(VzNOGJwZ5Fj^~!YYKY6%TF>sUxua)XZ%#;EC6y
zbI@xsJ4?u;!T;M@SZKpFp-MjT<0<NeGA~*{DoW6yHkXgvS33_qN@?sk+lcz-n~+wU
zjJ<l7r>*RT;ynmIA*B*u5eaX6l%6*9a$Dy=yTLw5i{d(ZwK#K^?SvIbsF!iJ4fBrZ
z3#tVd1>JUxjEI;=tJ%Urjs$kL`gHt$($LV*Jh&d!(wTRDk>7P>YjPc;H|c6^H64>m
zVum;YSO9*pVYz9cnE6iBEG6U!yksDAJvme(DcdrETRVp|{GouWX<r~&<BW>Rp@Rpt
z3{%#g-U*8Z!VN1{D_)YO6PECHMY&8Ifp>DJ$4j6u5T@aD5^<9r02;)lK}|b*h_iie
zw|4nNz@0ngz<3DddTdk%hE;TTLYoa#R<+7#gd@!$Sl}1--0kbj@T#%ivN+Brz?w)Y
z5O!N2ufk|0;P)9T9%hV+AXlJT5lPQEo!SyL0oD>h<$81dsjanRO-7x7FrrIByx-V;
z!E3e$X9)g;e@9vo)sfa3h22jM+IqGOQP93ArecoK<9>cOV<7*~Vkp<@xfPkt=W2&=
zE>V6ZbcIp(!?^)_fEb>gsxOJ%aMbKfU2sf{3J>oWWEcSJ5Gj3WYs<q~MGTI<;Ag$2
zv_hA(z9%a%npNOAt7*x|tQ!q%;`EMYO@2F`UF*MR=OGc60l7~7hmdtJbWrPDlR9wV
zK*#oX7kHjVMMXVh4lTW#@d+Tp8Wf?cr2!2mWeFxcz#@liMz@;QN-=pf|FXmGa@Xj!
z(wNpDV-vjn+nxqmxr-;CkvXf{+e1$zvWknZzzxW}w^*s*n>gH>qr+jCmmGFi;1^;d
z-4qf%MrV$Y^q4@kxP?xj{&;+I;O?Lib)T>=f@(?Z*Clf<`2iG~^>FX6-rkuiZwRxT
zs*v@dfAgae(VBxt$O)yM4LbP`Is-E?)zQq2@|zo@y2W)5R3pG1K)=wQ{rjK9#=0ZD
zCQ8I{Bvn<Sgg<iri7n%khzK$0u3`(cR#7L(K1iV}R&sG;+&!19P_{kmzR5i8T{iDq
zr8#WxU|dc!V1E`aDpw>?ACN}~s;;FaceXs0T4e`XeK-#@GvAdg$Jxf<M19|u$@!m6
zA8!8a^D^A1<pZV$W;4Y|tXl!l$?2MbR!hx9NsuZk`Jb_Z%KCK^4WDEI8qG#d^lQRH
zM9oMeO+-B7^(y-Q5&_gHf=d}sUWtbS(>x%Ewty{$>uNeXA3GZm;;?#cd5_)>qcJhC
zD;SK<?QUEKsm;x5OozUsx(Xw-xU`i1+<H1fz;17k;}cL!U7vex-AzN|w6^M2Ese5J
znrgZT$~prEcVKYM9PA`qrtgU<t){26BdMcTq&q;HLZhD)I2iO1`;VUL$`-PQo}DL3
z7r^*J-htv%OTnefR#)+40X2EASLFy|un!S#t1MInD<gp;ym|k?wRK@B)LgHT&iaVe
z7&x0-S%rAkz*Kd7N@7`WHA2=%iv}+SiJ|J^OETFWhr6Ol9F1!N{drLPSifo8;Wjn5
z&5`wPDJhdv(}T*HyZp4&)pygL(0Vo*5*-a*AM>RXVPu4L2+L5gY~L~iI##N6e+g74
zz)U*L2ogik+URe*m&pZZmY|O8*8)S(nTa>~ly*w(m5EQnogE#>bo6ySWI|-)QOF6d
z3<gt-m>Nss*qQ41=h>Gpu4MfC;IVl&n>)*1hxZuCK`hg+zeq#7T0Mcn$h@))qd<Ht
zN8n6j35dzsGn_Ln;>*8cF-V5zcoVG4lp!<f2^oh>0Nr(qy+$t8?$(c7YRfHo6@}lI
zk!OW<Z^F#E?euqP5@7fT$`OAAuGyof&J@7liXJV~QlV+i4FrE;b{nBgzyM`;xB?In
zzCO9O2%<}HeFI0XKViV2dVh&>E;t;7vcxVe4KxNADtB?bzDGd$#m-!+*JqH+po+^J
zkaw}`2(=GFeYwX|=s%dIZ30btnGF6Qq7u5zaLcu&YQX$xq9cse&6A%vvp)=V%%UB&
z3*vVew6wN<36iPs+57^6h{xmd2?6I=C&7CAM5lJ{TnTQ{@m0PQP?;Ilu%z>)w}b%%
zXYyt#u^`|vc<t(pYk+5YQ(RzgW5cnqK=|zmN4BF!F%l`zO3R2us)J}r&GsbF2UC}t
zIeDHc^>c}%vf*go1>6z^F8a+_AJW<9Q2#Cw>7vA;<1ae6K{ae0bI^m*KxJsMzoBjA
z=i?iav_;fm{yJ55_RxG{15_!Ifb@CDFD&jb&50<hB`5-~UWpBx<xZNJm`IOTk}m)u
zxk)uLyo9Z<mMHtyG$$~OWl-u8<Mf={L~j}I<*9<Zar<X~?W3gyYbaR6#xf)kQNn;_
zKZ<rJ+M6Fl^%gvO>IJ(Fun0T9$<-w!uqgs`T{tI<8Rtdy3+TGX3W=XM5yVvvAQ03j
z#k0E@(TY8~bMSgMdzyjDj$Eq5<Ye#TB5wq(afODeL<&dDQj&n4^~R!DLbWM>^+1Tl
zp(VCicn&ZbruS9W)I4zIz9qU}zS)Y#ar|<E&|+tUm|YNx@Tcc*5F=!PD}{{qK&qAV
zff&X)RFk$4ITWS|&RoIu_3X@Y1T#e?vvh9io2<oPFIqm3NY!2GL#|L`!45A0!U#1T
zT|{W87V1ZgjFb2Q4EQM4x;y(@Ad~C>cm8oKH%s9TOx<<9d>JbIvMAD`YZUslXl?<Y
z13U~sC~&bjs~K<EN_Qbt1`YTkJ^<B7B*bCR1Htf^kdR-Xy`Xp_A|w>}?E=UGlapDY
zVSBukNF-vwDJqH(0zn`*hng)a5jg!IOQD*JXuGnjD?FO%1iF(ugIv>&gABoA`ehIZ
zGHUU3`}XPL<AZ~90Tp&GWJpLz%&ZLneW8rxtS{bi0XSdSZ9~>GoY*R;>RrhOQ4dYZ
zTMrupsSH_jzv=~1#=5Q;z3dDBxEyC@CKPfBpyd5~P;`hkCnPY4%B7#r{QmtSFE6Cu
z*=q@Bw8j&--URfUVM)-*uZy~phQ>6^AeI4&omCK-c>f;HR$h!NlQTw+xd=cgusy_?
zFTjFAs*GSA6>P-o+1c3$qo6`H1`OLT4}a0yXQ8osdlvpRgSm-k85o5wU%(HJ8UPGK
z!%yfXDc@givP^JDbS>Q~7NOPP^o)AylPE@@{L}vc6;%ME8(w!zY9lKlnCNbXIy%rg
zI2`GG0jZ_8_h#C?AbhOvQaR*0<(oW7jLy318vV)58>Z^A!+@Tl;VVdt4V&S(8XFrc
zplx7%+Kw`?>AH%K;mP~OjOeRhk`{MkY+0M=NKKdscvGs1YxC0iqC21o2ypLgHS)|e
z8$WTySe_CFAp6dbUY4O3sWLzv21$2AbpMBkBeCLv3<w4)^Re3b$`_@{g(Rk=SYw$G
z-6Ee6L0d6=88B${BS9$Z^5yJ+096!PYei23>NLEWz#aoUJoMwhu2wHJQZq(25rw}E
zWZx;JAeBc@9!YU9B6)2oI!VEjIhzb1En_77`E_qtM7_3V#k=T-i<OkMT}L#AFXRQA
zZudpB=H-4sn4SH>Y(dIWAVOyZX-D@URzKd!+wU5nHP_VE7P7Xz?j$;*f$jn&`uXhX
zhjYw9Ltu89i=~sP820Ru9qb<&!St3hv8$^qH>)rBW8A~>R{QSkJFyUf`3l~NIWbu%
z>BCzfNEk-O8wvJqb#rqdNw)NQF{*&rg0o<UARqflv$A##^j@*ZEz-{2>>#bu+{gr!
z-v0@8wmOut^O`FvWcVzwIxy}vZ$k;rQ|Ucs{~EBSeH#LukK34<GDij%P-5;QqXL0S
z)*8uVUT-WGRYC8#Uz2ClL9U0$9DrzEwMdzvJ{xiTX#y+!z+h^8%})vDM;?^l^t>zU
ziISlEoL3;d!Veh*ZEW7RrYCe=wGFv)rw|$ako0mkC=LnPW7h|iXxjK%Cu0oiC3m7c
zgZ%vc(Ke_0{q&15Y>a9H%g(c2Wo42_BAErn^m+yC9+2j<*85ng(e0d$Nz1$!oTfJH
zQ)TMNvyKVD!Hv3Fc}u#O%|$=>7c^_UoX^<FQ}Gs(#v+thfV>}ETn&Ls0_HM6T842!
zdZ^~(@8<`?my6@`4_W}rPy=RbBg8`dXlv0Yc$zJ-Fb7d<!`dUd?ZW>Ldv6}r<NEde
zx6K(cWhPT8MUg2}nMwmnqtc+rtTH5$+8gZ<5)CL)2~nXkg(8JgD3YP16b+Il4H|y0
z)AzoA&%gJd&+#00$9LGfKErjL*Lkk>Uhnl@YtiK=HqDIZbaWalCJUyt$+WIpF2`&B
zIOZ1F8UKq|Cdz#I%0IcJZ-L7Wl3Zv0W))%L9Ge>qAKn=1v~uN2RXJz>J0EsQUnyB_
zZ=bsB3bgppxf{1^IZJWNxpRSm-_Xw|E#mo%IXS-CC|t!i>*gl2w;pqJA>hfIas3%;
z%i5#AHQ$X=$lSb5Y{Hx2vWrjz<Gu$E$u=7v;Z#2|%{o+DmhS&+WQ~YOogWyH>TL7b
z-!<3n9EAA3zN0MuTMMuQe|lH7MSYm&Mu%<BdWP4I$}KOxYpS>)>CqB0sg$ry8am>9
z2avYrm@Kvg0W~^ve{{wqw;7Q^cuN`Hney8hU?+ej1^#6n4=qyVII@2p?EP&zVc@so
z;yJE$>(R*|azG_Ns^eK%#d%`%s)t8A9h6S5Y;e+9vcxiez3<JN&ahTgtqqltIp^mm
zJJ6fkqPk^UVRC2shitrm&eJoMBUrL$^l8NJ91r=uZXjw_c_g$Vfirb`SRF_T3LUo4
z44%{~EiQBmI(m80t&|cdIDH5>0fq?q?%%)vN77MO2452P)l-Q6LAFtv-Fx=*Q%CCq
zh{Ir_d2-?ZFg0k9>EDPw2jk6ULJ<C&v8TF@#}&S~UhWh=TKSKR(Qm~nqN$kF8I94_
z4=fT3p9pLal@>FGZqUo~H9YLM?%b)qwW!m~MvLtr?r+@3wg&y;IOZp~o?6&#?%W5E
zH7m`{BQI-5JySn&>5}}+yDK7|IS+=P4ozvHU|_|%<QjND(z5r5E6ph{Ev19S6?oqP
zbC*kDZ){uhBeU0LoO(3HQNRD?B)s=)xSti8N;lAKiaowDCOAz|bFRiUle50QVs0}U
zeA(71O!MHSjKCsJAmBuiq&#6l3u{Hzn_xJLw8+K?3nQb~WX#GJy$wcAFSnP&mq)`L
z>=kbbpqolW1T5`dxG3@xeZ5mve@nqMhPO!)tZm1i`T0qoz8a1psi>&1VU!Mgh=^QB
zn|3&r0y^j+s9zbd#@6-*>if6Dg*fHtVt8w254;2ccIT%@4-`fZ*kOcxNR*yZTZ=Ne
z#WS>7o!q35y0^8=tgz_6fC4b)6|gB%A?_#wrFsIP<`$kL71IPRfG1KbA8&42teHuR
zD&so)H`&=?$>;G<P(8F&@uA9<JmAZhqVSk+W7oYnH+K`8puAjBRv-N(@v_mTaK;M3
zO}TM<AcCb!mXNwdtN<*8{^E4f`SJFR8|D<Pxdz-K%j*>#UCEjbafv%{{no9-UDaW3
z6}F~tZs!)i%FSi?twP0F$IQ?+t>q>g{v2}4)_;2%tyic|X)OF$yZ4xu1`j?%Ya`&^
z$k1??^|G((xlCuipr29f>ca(;q#YQ3K68Klxan762J`Q1UHH|c$YW+(51AoDbZ*|d
z%jhJP(WCzv>hn-z=v;Mm>$Tyw$S!eFbL`c3{XCyxnB}|h+_ZI5UAJz%ar*R-FLfC1
z+7LOnsR@h5Ak7#Zcr5hik@t)&_?*x=Rd?~?bxeJ>ba~;gKGe#!o|+N!P_2K@o%=-M
z_s$j1^btpYR90FU8>fWWmK@V^Bx-0MtZ-<f)(Sn@6NHjtj@_!6OP4N<xH{$e;>4^{
zq=&_JDp6A$6|USdEe@V23Tp2z^6TEK*GX;JV)wWmb2Mfl7=qYSRi87I3B>Him^?qe
ze2lN`xN*PAbM$seA~L>vw@G9?t~r%R-RB69$m#F#FW30MP;qV~^4OT1)7qT)RcA-b
zAIJ;ODhv0I=-I26n9h;-qhZcaE5(^JiSk}_(6IB>enT13sZ)bbzo*yG`;Q-M*WDD_
zykgn6Lj8U~<BS;9pvYOMr_80%CLcIHyjgd#gXA+T!7vJdk1i=5di30h6UlK7Diso2
zlm}QyrsU_J9j-EAdG`7phZa#YM5R;5HdRz=boV~FvAxzu%Pjp;09j-A$lb>u3^O^R
zm0LXSDKfB5$Er_k%uf(yZ!6*qQ+a!A&~X3RLm_QiK*cRX+(RgYy*X`WvCPqB$_XDM
zjpUos&v!eb)z!d@*$LGtf77kiiDS@~Go6Teex{D>(cv>|1da1!?Y(>VuG-2?o3(iH
z)9cqKAuaPy)CD4fkQO-{n-5i2ft586Ulv8sOic0E`H}^^HyU;(TiE8mKU7pv@%!_m
zjxo!IY9=V<i(GiSLu3f8pnQCKe>p#9PtOw|0)kK3r-#@yj47D4{p*qA`I?BD;PP&>
zs?MKg=H}+I2-}5PC8nV#@(3aex7x2u=gphvIj}IoQPW^KRnEB#150jipV03RMKat^
zCA@pu?G9(h$XVX7ONAo5V=+;h;vx7#XodqfUii;bz4rGKSV>&lC#DxiMvm84B_5$0
zEKtAKa5aNRW>E@d8KawhG3;jX`(+IqEhw4!=SIjnj0KQ=K-$SrwuEZcytvrgqRS40
zgQ7ELnKKU~H+KeQn$GuD=iJ~#LtU~H4y5b@$=t#)0|#Ex>#Xte3{7<#6l84K(|W5n
zZP?Ijh-xwP{8=iwhhGc(>~a39nX4h-jfNbIarmOL5!)$R?NG#3-+P$s$__~_HQv~j
zGv3RcDI3ySw1}QhE7rdG_~y+H(rIJ~iwX=4(^FDpr%v_Tl(rUaCJ!NoyqC5l@o#xi
z(HL2|eCT(m@#oXdq@YK`d$!z6A%#U|<N37WTkH|o6;+ip+Oo0Nud7j3T7GN&m?1-l
zB4_@IW<ha;%I39ecLK?`_1D(LbJE6ad1!%4s>U=uz2{(b4-b!XzLQB`m|~#%H|k#=
zg@TpfY^<W<MuyQ8TkH0OhGm{l8j?@@7Y_}Tk*G6Bc^6_}RhNsjWBov?8*JuTm#SUh
z7Os6oar#XBz@@LM&z6p;T`Z9(N7ft9eJLcQfXJqb*~VB?6p%b^cIpvY6s$dEBHNE^
z7BiZ-LDX(1y3EIkljY52?snO}(a;5(^apnVgb<jGWDB7SqpF8~Xa0^>f$(a}s5oPY
z9mM<|C?pU*WcJ#%7jNCV1vQP6;VqNM#UP5lEN}fDq`|gAMFHp0gcb}K!s~f4gil>{
zwQu-fYT@t(q#c-dMUGW-!Ge9H6j(n{740(CZ!=dmMYtlW)tFaBMIsExFruDw=d6v5
zLDic#ZTd(lEPOC+p0!VQEcL_bb?(&Z)r@igRw;rezjr#P=u`o%452vc>0Ll6KRjtv
zP?A)y<3XQ2d-n3>jOo*1Q%w;W@g{WA8clEa8g+*$<lEprVAUudYzG#QpyNKsCUXXH
zniwC!x&#Fc^SBoWb`5^<wX$;QvO#@f_!xXEQv?}yyMzdr(4XWD5hPpH(^He1<bd-a
zl$?g`F!6hpoxM$nmQAC&NlA&b0^VT&oZs2ARwOe-v76FWZj<NMKoN@c9W}|BFD~;k
znmTL?JRU}-66}RY1m9l5h9qyIt)r8ckr73%8NW>uf%~G#ZN$0G^R(YweCr23shJUI
zvLhN$G=JjkdH$J?A1{2gxeVv8VwE7fz4b3)fT3>zPebtWLx)n*OS<FK>C@(hhIa55
z4i2AKdq@h*#499GhdK|W#pWg^@xWu=BaOz|j2Syth%8AJojhq2w|43F?X|pz%wJAw
zs^;34N1ZeBaf#C7t^eeBs@f@jW^ess`~9ai3NVFpaVtK4Gr1#xFzdDQn)Te#cO%0N
zoUtieE-DhSacp(b;EbJ2E{!fJBxJI~X7Mw7wwY<M#Teh=vvwTZ>O|76Tt&zKVnjsA
zmoHP`!TBNiyJ*oOF-!5n1sYoDdVHo9;5RM1DQ8OZ?zx|K4{SXGMnDcVg|Df?g)BuC
z$!L?KBQ>lWGrtNzK~9rvSLadlYB*HJ_<Fv%%0kfixLfz`3G{d*4+#f{zMcq`67W8b
zj*g>8-{R8SDd>R0;sFK3HyGnupwGorMflI;$6D|~?)Bv+1_lO9JHo-ia2K!4jo{hi
z@v{^4*;mvF(KRjprg^wlFYFC(znhRCih0(p@#)7rAHmp)LrcC3hRx8($l2K$BhQDa
z6IjDErya>SA>ICFhqE(!7uvhUW6|@Ot;vR>#Dk|ttWDVphIbPZ9sWbl#EX6pL*|{y
zE505dPc+a#EXD*&<_rBIm<RCc{C>kbP`3a|K+TZt<@bjzx|BW6D>1Q+EmEQ!M70Ih
z3SNz30Blqpza_F~r1D18C7c)w3+m#Y&}%R=bLr=6#SQNj!q#s)IhIjF_{F-hu)hle
zroELeA3C9T!O*3rW=F~lu!!k2wbS0^f#gEcm5<eokG<?IJyWiTN)JWFiB7xgO8b5>
z;)06{zMExEO*#Orh3~8D#}_jC%XH^m<7FQTLzA|G=f8j7e9p+&CHk<pH^u0%NyQ(x
z43NHoC3-Rl!u)1zk9rM0flV0sj3T`uj(Mr65esInS{~?l=x2%UyzOrCAw$m$9~uTO
zKhnIDev|%IWbvAoN@c2q9-d$xyJ%tCZn~jJJZw-JH%_a<Lv!)s+6&AiAHVvr=wM)M
zskLg1*#NHm(wT<r;!90(ZX_7V{kJ}Y0HgTd+Pq|W5FrMHZL=>nW>Q#j&nx##g|5aV
zmmqOx)u-rPFAPQ>QQ1uNydpO$YLQ9Ifiam^>x>U<?r5;*c^s{eR>jYv^3BA2o`i&N
zH!jV<(2!Z#bDMp+f-Acpn4i94d2(S#z02J>h(|Jqv8<44e6w#y(HO2E9Q@B<t6+Cr
zYaJP*$fz2(5O!m!2SU+SG1u8-uqodP;e-W^j!M3~)ZF|%MyzZxTGTbu=b5-hb0{Cb
z8@&aST$6dTYm@Y#L0(>7uYH7i^y@9>@(WFJ(k7m|`U7x&7^X!UML&MLxEa|~VqVFD
zli}HPCC~1<y|lbs6p>gSSw44aQd($}oUt!ffibDCeT2HsL-zQ^`Si8>JiDZ->g9@4
zjs)!fCa0<lNW_EPA~h(Fe>U;CQsCD>gGDL#xxS|>%8P?`u;5RgbVPu5qUi}&O5&#M
z>IMo$sL|4BZ!t1&+P?%?KveT41fW4~`f5teaW;iXShkTTzT)+5y{w!Zf{oR30bM2j
z?s{gp|FNtKFlX)@cg3rdqW<e*J!4~4K?u#hv;#7Z4lNxcTInUOB)$sL@KA4khRku&
zD61_T6Lw>y2AXLwx5=TB(Hq<)`Bsv+^IB(88b&f-cY=>A<)x@v4!euyG%Q1jhF-5+
za)?(k0d-?}4igm=41Qdd7~~pqiwrNPu>Ac1OEtD7K<mm+lI_Hl**;6oOx5}Pqe4Q$
zzr)gF>6-2HU7Vao1t!GoI8#vks-@qRWNaeOq--&AM%p#6U{m;r-fI#Pir*pOk1V_9
zMxA5WKb7aIIz6a<gZJbOrgpX~xM(se?9dB7%<zlDW2aB_aV2eje@{b9*t*DX7QXkS
z_kDPkbRBOfirM7`{%ew>7`oQf^$PmJu~)YBiNWjiJG5v($zKwE#MKFcoB?oUtH~gr
zKi~fJ)7~!H%S-lj{9&uLN>*uk-xR0GtLF8URnYZ2ziL!Bg|GKFJ^Vh@Y)w^|-HN!x
z(y*e~aeX=u(45z|Z>mH$g?)c_?>SQ{V^XX8sRvseUCi{S4B2~5-w73Ld)<MUwwy?#
z_G?r&+pjddpI_k@?q<58A6<gQXzpuKR}~Z|&yw)_C#k))s`}+R6|-~?Pr~8AFBhI|
zs=qgb&NH#fp%NuY$u;eZdrPF7@`!tEt-k5!-)?lQ#%A>V34dCKN=W>Zq2<+qk#osC
z6Z2odP6%ut=*)<O>p|}c1Iz;-D%V<dROq}`Or)U8?kK)h{B-K~zrPUcasEiG^4|qP
zVATTWTq}DkTI&Avx=y|5flBpL5)zVRryum$b0};4+W(;0o#tk=(`^ZDTGc^9dhCDo
z`uiWO>iKvlHOsftuCH_z4`_n9;ad}ld6BM65-r$M^`JBWIUe)%?nagWCGojo9gG57
z`e2JFW)HO7rZePiM~^;M2|L6XfmfzH!g*JIBvi!Qt4f@mU=JF^gs5C~rRQg>80d;C
z7~2k(t3b{1{JPAjHOs|elR!Q5r}sLWCgnd<s`_aE?=Qr{{O@{*Rgm}}YzKw#C?;4b
zs2~=7P1Ghkg8H37ueZHfzS!h)A2pRwctw1I^0^{DPGPJa&6?WP=#T%n@F$rAqXPT1
z6LLDYthZ-Q(~PEP=HIbv*_wa1YwPLF{{219ds95{2{bt{cRla-8^A_?=qOyB<w9ps
zNB_Vzb=pw=Vz0@*{TZ~78S@9n9#aSiJ-pRU40f>>`}wEMVYxPrQVp{-t?NYNX@%{7
zE<_bMHSGHsDN5>nDBYCSn=8I$qWBh%4TKG_;ofQo2Z*+f?lZ&@xp8=~XoCc3{gG<I
zWEKf~J-(n1{qNj~?mR!W9+~oj#{3hWaP%Y;7K(3K0dy{dW)F3_9mYyMtU3go81>JB
z#f6$jK4Uo)9K{DqO1s(y;6Df1&mYWN(}#&8-NDm+$>sli`#SNWLj%V2fwitit!C3N
zm6sWq{`s}YTq6d#7yq1WKiN&)%c8}8`T_ApH$c~jNXqWTd;jN?Jj7f6**Sio#`o)w
zkEs<MJfXzghQC!u)6cEj8z^Hsn?$_ZEymDW{JBG6Bl;-7qf_u1S+|6uBN9tGxp<Fx
zW&QCPvL}lYlu0l<;BR#X{hwcNXEx!b6Fz(HD2TiE7Z-Dy)$TC1hG$AV=C|z6YO9Jn
z{j0m%&Qj&Mhv<v`cfL0-Jukt~hSI)^hx=aG|DWrADz3Y;UOztyt(pjXR5I<v8ZQ-V
zypK3?5{l=-i|_}Gk`MQ*U}ew0IS3f){+54dQ_AP9(ZuTgT`)s@p>Dq?CD=$&zf|_;
znj_P+W4DlY)}LE%9O>Ud!r<H%rpPhZ9d0}8D4k{QGuZn16<G;|{(;u6Wuv`@|1sEQ
zIs?O1%wvl~y<`+`(lK`DY#|Jv;b<+qLc|py6W7;f@c%n!aTEOf4-VG<TH-$|^^a<L
zf`r7oEyg_c3n!Ot{CKM}Yw^f@LO;yF_q*TDaqf_Aq78qHFd?z&YmiD}qaDN;*jur}
z&c%#g@9S}ryoL~T=d1Z$M>swFi;|!yVqK;)i;g|iwxb)(;O`){k#0Zd!}FNu{yD56
z?8WXoGbTT6Dbx2TXD%#ryoV&+<h>{dKOiX8pH+6jK!MbVU{SvDSDN38ikA(!341<l
zy7qvcD}v2JO-h}|T-y-RQ!s<S&+nI-RpMu~AoLZK%gEiuz6`un1B5@UCQWJ?q3eGj
zXd4N5T99qqMqzWVIA3Twl+37k@eu<?7;V~?V`fuQTr6Fp$Dv#0BCpl;?=`;|3&B8n
z*amsdL#h?i%$Zy)7AVi9U@y_8&Gl9D51BUp{@LDE&|U$5r(BuM0YMhi9=!bp1xhx{
zt&3rm`5|Nz1RwR7Yb%J>DCwHb7<Ss14ruPfv-$ou>Zp{&Kg;IUzafUKLpfgGbFKyC
zlxSq~-K1~j<?v$&=xW+qUbg$aYGjD`5axpLMH;LW3+Vq!^zB!8$&onz&OXrU*ViD&
zw>0yL>Jch|oMu;jj@cAd-Cr{Ic<o-MR5avm*Ds8T<UQAagH`8sK&s%1jZrfGsdaY`
zhHR|zz3;In)OC5PmA4Dpu6eXxy)bqUuTLLPCLxhBi&vb(63U)DCZ0fPxX&-_Cbz?t
zG0j&O#^pcs%5eyi%6B)yr(SJseUCweT)6h&*9UlZX3w4nzr)W}W`&F2pZq=z(lUGe
zHFQXj7j9+4*)@M}8$En8;sJZB5fXm4x(`sK*mX+U?=N^KV^~7jvD8vG1gvzRF<E!7
z$G|eBo9Ql_#8eiw;npB`a($3fBi>ZTK3DG1oJgaqRE<$n?Jz8o4dsLc?p{0`a*xBe
zJsZso`CqGj=H6n}Wi>h$IyIiz6f?dJ+q1Jj!Zz(TElKHAM3FMB!*bbpu`%7c1@X|@
zBOr*Lj&QL|UW$1eWVIt}TCQarw3O^viaItTH|okU2=`GxbMoZL^!em7cRcDm2qt(N
zyP2vRI4k<3{-Qdbf#!KL3f#5}vAoQl&9;fBBnV7?*M_vW@U`?wr5nZ`$REO&rl%Th
zS?oo-IqAwxxMJnI#V7bz^`8$razLcYe-)q9SLY#1&LDSInbh9qsXU65iIUPL%rXu-
zu_j$fTuHujE^~sr){mUNc?J5n?t9gE#5?76#_VLINn^+6B5##_pN7CuTYUFIY@SN-
zAV;ak{kpka1b#=l(d{|gY+PPz>L8n*o(s_43`ayq=)V70O_fKx7?M~0xQ$ylQWXq|
z`g!aM6l14;&RciZ*|gC5#_D4;$6aoOYwS{K@_BRJfn#%Oo*`P6E1mktk%Kr&LTAx@
zV<#N4VqTzx#J$;ZS)q$gAC)RO{=W3TAj+xL$8_#&;I5byyl%1#h4!r&j;8N%FUhK&
zgrQ#5{^i(ghYYU#>9L8E?n+UZa2DX7?^c`1)%Gmk{?>dQ{meGJoZE0@g9k&Ii)J%{
zBa%&+)_WFvwDAdzkJ-5w4YqZFkL&fp`1{ptkJ`)`v{V!Xet(Bbnysn-{%&-CF#|@=
z59d5RbUd+SooLS|3Vti*VO9kf%Ww=O<99b3ikM(i|6C<WSP_@{gBZVNTA|S~*nHJX
z&Sz`=HIJhkf-y7DzF%gCY=1FR=U*|HjtejPh^cCl4!n2B42JpqQu6lW^G@lwVXA!8
zo7X2V70BtF5mO}|IxGTfByE5ZN&bbL04@04!M*Dku0|802Ts3#uqn7G?~qC7EFpc!
z1_juCyLv5gmixK-xR>g~3gVzGbOO3*03G2=i$3<3@(;ru%WoFkHflkeQrnYvQ2y-y
zde?=18?wi`0b=fs!EhF&L@gLxjoBdma^udiZBK@*v)q;@r#YmMhc*6?=ZCQ*Jz4I&
zQ%?q2?;5SX6ux>G`mk4Ol4d&?xSXEs*`tTmN6c8j^;SgeQeKZeWu_m(U?bI-Qg~6=
zyOh2H?wMileg*;8<Tkf6c-_0A4<70co@3vwV={A_Q~K+p5`_O;i>dfmYQ&kdXU|^$
z#^;(IjIc^B{MlcB&epoD#p~9t6>bX;!(wepK1W{U$Ody1p~;;y`>!ntHMq=vAs>6t
zQU6Yc>){00Q3M5zzwjfU@LHZSx8@mLT%z;N8afFhADFjQTw8B(Z6n1NvvApiVM=xO
zBXiItb9WNKC3#N4?*fGoBfms2sN&ItEwJjRUKC5mAT!IYsWCEp5r~U_f_LD;AMksD
z<6t=S-c2bCH|;;}GtOx4k2_y@ILbkZ__SQ>JzH`Qw`qWyJeaQ1-yxy!-(jK<Ah}o2
zVNyR`w}@GVX(I%7tC`3CXkU)CJ}-QlVodlBs%ve3Lu;a_=v#nfCZ5DoRkp!=RA+Ok
zWIFZ4%x14M6l(R^1{MpKcIpWxe<Fq{al4z{Ir_d61T>!GxOP8d6Owo1smyOW?=o&k
zsqmDKM%%8@PnP5Z4E9Yo;aeB=W)$eW)1cmm9-ZcDsY-XJUgYMTWBrm==qAd}7Z1nR
zDW^(oxg-Aj=U~~Yj}|)ZVhc$A&V_$UtYwD%paN0FJlUsTFVa+nGRmTBsW=E#4;qdY
zkZ`)-t(<VV&*~p10Lv%T#r)rr08mKVYLc>&f%=l33-(mrG}}5t&i2mm?w%Ce4UOku
zw!}hFH2%Wdbei~8#*uWe7>1p^biPq^{}*vG2ztn7uO13*)_=-0^}>B+v=vuKQLOFQ
zRmzRtcKxl>#EiEMm(B(hH!t3*bL#WAkFoZ8IwME0?6Gp<gtP#<B|rM$Xzhk1;V&iQ
z$~bP^zhc@k&h&rwl63D8Q%mNwmr8!@zwA*P2jP?Z@5;tsrOsB0rcv+6YBj2R?z&To
zi{k{`#AT#7o|$C3fDFU_!z$J~%4UA0XTEdGeqF)MjhW;|`rQ)nNbniWjiU43JL}I!
zzMhw=u88TRQ{(pr!Hjg*v_L1@fY-}rPps*OrI_p9)`nwg!zM+z`>~2_39`bilCN9H
zN;H(`h_<GYy;n#BhxFf_gU>xu@fFK$pnfRpug^0uksFf5LI&^s^<HzN2W*Pu#~yA=
z><zrm|MjJFv|0b|o`?kvhszcS?v4_Jr&Bqu=XQTR-PC@}qiDT?v@6@!$fkXb&)i>w
znXn2A7vfZb1e4KrcdnLsgjQm$IGoydHV`&P81dcziByw@P`UEz;-;9p<GzkR<9VDT
zG)b4@|0pc!i4hpE_5+ppN@Y78&+FKASkWi9zg_w!3f#G~mbY)rc0^j-EC-n2u?8oN
zA1+nk^te~9m5WvKx|o3D!yN96F1(@|YETNvH1Q@)Y$Ozv{(GJ2-*?vOHF3>0D5ycL
zeGn8bd*@zsNWM#D>`V3E!*_lDr@9ifz>%Nms^j@HZ~M#Cm$n*Z7fy-}`#f1)f5p_z
z&E;3G?UQjD8V=wkVMNmjy_SnSBQiz#!tPwX>qIf=_S6ApvUz?%LBBbZ{%a?T2QEoi
z))LrNX4k0ouQP!Z0FDc4%YY6eT}wkTpQ-U4KOc3K{Ee$EnZMAA|CP+`N_EOFl3#{V
zmv&7wk6l9qI{7HFGB)*arq#N&<L_qq+xAxb{5JwkdCU3sDEdKAbcG4*v3yJDG{;Xu
zz;%s|V#(3%&ACZ)Y=}TV`}{F8_9WTo*Is0(oVeGyji2|Yjr?woy$o_l*ubzqVOU!1
zM$LczH$%S)xsjuej^uiK&B#0M6c}r@T()q`L}r1NNXiHId3^`WXj~*L15UaX^nbV@
z9waI91zj8DF0pokspWDlU>6RI0P@pp^q1jPM)LlxYW!xJ^<SS|#xdg{xra*d<keba
zA^PvU^XtPVA@&0BEDKiKifn}RGUf(D@(Ems1br`$50E?7-Q?s!;keP55NGUaxd^<w
zPB|eXyi)jHkq7PAd(KY)hzuRqlCy3(7Kk`wQ1kN%3BU+a22j`X30t`&?<&#-r#g)E
zq}k!q8J~8AgBPoV5QOhaAAhOS#D9$}Y2Lk&fQyCVZbO`sCF!&Fkx1x7T@oR*UV<!7
zJKfvqx@7-xe$LJk2GhIrmI)3F3@o-%)x_s`k8FWNQD}a~a0TrZyD?<_e+kF1Kk=P>
zh?{DmvZVaKi-$3qOEH2R-^GW6S_p`gTQ*gDoR|=lh~O~HXQXd&YNx+7_VEqUUxq35
z8N2qyQX$7k7&B@!^pf^x&q3CWuS43T+odz@s%gY3Z`oJQS!}E=xARRc>!mFeZBj8k
zB4O1%)$fc)`|pBpXI=ip^pGBX2|nyN+V4)UJ7@lRqpYXc&eF{>*W~m9f_Hw{z+s*E
z=|p#1_6XJ0OFvyYIH+g3KPhctkt&I>gpNO#Su+3msPk<0I9xt4hDG5{kP2=5O_P_>
zGwsqj*K(fay{r8{iAvP(H=eeU81qz$d?%-$CKRPOI76k2i!$jGj6ZbA+qzh7xN?Ps
zIM*heYwfjE!GT)eZhbmJl=Vu>|Lyg-aLAOkmOHXE5wchl3mV+!!i15=kh>Cc8A-lo
zE=CTYjy)CqYVZB*>~~%4b%4hFl_H&hx$#>&H`M_>3a}h8zNo_@*Khu|gl{(J@}G#k
z7x8oj2j@Xuw%%+`2x-CaW3#T{d<pK}ai9=KGQ%n!JrC*(F5or}D;0|x$GLd3_y;Nm
zoHFr+Te775aOXG6iC;dkjo4nWIkyo45VE>sOh)dl%K|d#)@_;)i~?>DAEWQlY?K_0
z)!IBh*QgstYv!LHw1Yew?xVt#o@`=vqW8ePG$!E5+mmZH&lPacTvu<~>Kcf0G2HJ=
z4tFF`l1GWevY91*u`wOoZ-WM=bCRBUu){?wd)26}-OT??-jk3x(dU1YC@k3jApP-w
zUD_#KA<uzB6m2lSQ*ViTC)pZs0AmyB3-$p9cYLcW@N<HSisPewokdc}uZQQ6ED>l3
z-q>+GgPv!>rzZaXaj)^0(zW|FG53Bb)^CFg*aCjw#7wAAlsIoWDO`+&m5}HF(h&bY
z>Nv<w{QST03j;0lK@Wcv!)yqCc44UC+UoEr0nRf$)ChIIq1PGH9t{ys>+3`!o&Z~&
zdD}$8QO3N6uY5sb9JNWH$y|eGjwtrxmVv?!B3&!z920x17b7W-bU^F_|2|^FOhyk2
zT&OxhQsEULV6%@$%h-exI0yCe%i-x)WF$s~3_ff1lZfu?(>qkNs8m?AIVA#5NUqIb
zdTe$Ly%Ws;ZQB#e1ICXm#Vy0B5f5!ap#_We3yef%Lp<$o#)f}yx1^H;{n$k-3(<E*
zaEagFEytR`yoljBz#>Fo6DvA+5VxIa__W7S5Pzo8bR2`^Azq#ZfS|)|7J0qwmeiTV
z258RJn@hvBpW^^tlS~w*{>?^(mpFe7elMaA0vGWanVHde5-#c(UI)DkeP*D^E(~V;
z%oeOLfR2}A_}iRhJX)%Ws3s<q0K}B9+9^l^%}+xXac>}9aixrVMzUP>cSosG3@KNe
zo{(U$-bL-fp+*u56ni())kt)XhRE?(H2J`S5orr%j_}_2vVrUEt(d<iP{*;hht9Z+
z#UXm&*RbJd@1&H%7s9A62o*Jt5QZz?mdqP+k5vz!QU}Hn1cu}uSfNJznEsuzL-Oa;
z>Rv6}FEtql0ld|?Sp8{F6*IW<V{)G>uRxbo(Dv&#eESB(jMXcvpnJJ4RGy5%7)@-g
z+*$*JD(e1@46X*SEY{y}#1y$I@Olp<7LG=t<b$0X@v;l75oJ(e9yVbjkYf*ld-g)9
zxj6q!YOQpnHsOe)<A?Y%c#Z+_Am)7MY#n*cwtFlXK55F@fcGO`k+G_}^W~U$Cmv67
zHm&VkfPUhV8MH7yw@XrQ$lV8|VZ&AmR*2++{P&h8!yM;8*j85(!+5Y<<Z%b@@PmkA
z;!f&R#xW#XTCHrg5`2mm!8HNkft8N;tDnFSuV;1wg4>BD$z#&n%a$hMs@3VxfY4LW
zxCSBIdGDc9OG*5G%&86Luc9R|KEFDMzqkN$<@2r*XXRRMUs_jH#qqP=Tb~p6V%Da>
zV^R{*aepvar^=gx_?tb@``MBQVhdjIF>&BJqWEG0B#nxwxb%boC$-@}2r=4bl)&mL
zveQ-+c?^N^Y&riz%z}R9=m2g#G2EWP2G#et_+fHCJ68t@_)TLMe~nz&DyE4oD>G|k
zF!0okZxsBy(~SC8y3#d2ch4q-X@W*JF3)ISTm8sx9O>R@`)Wx%%;xl_L_uTxNFr9G
z)7PD!HXHkPveLeXW<oFeUx8g*E24D_?+&Kq`5kRSBmHOCCLj3KzK&&x*qLYd2HK;g
zSwQ007`fq6URRAHPqathXG=a$2+#-s%a8why9_|st-OUp)TLGG7}J!}mW$)xXvG0B
z>tNHT-S_O-y(Ru;7aQf{If%oiZ&qj>^ISX;aN7t-v+;?iI4blO%~C$<0aWaAbUIs9
zb~ea4)zV6>`1nG4k(tM<+@5qI&vOnJn0>8lo0?tQEN2i_lv0OenaLJ0nN&@s^N=ns
zNwe_{)`*(@<IdQNC4$gRQ3IjVKf`sa@uKQGRf)d^ciLjIzehYEhNmRr896JHcxZ%D
zSX66t0g7kEfKa#%g2aFwCe;H~tn)Cbw?DZm)r;G@t#?+?S?7oR=$t?t5#Anz5GFUN
z?WWVh${$!D9{!$!y$zb<<cYm*lHbG}H4bB6!dVHoXNFaVK<xj%gdm?@wXNhC^)4mD
z9W80n7~kd)Fav_&*|l~Q2}%_f3R4J8Cc8WkPXqG9bMQuXHVvsJTue~7LhaC$7iwRi
zUi5CXlh~nf{|-jEU$mlDlg5Lz28KhQ+mWq9z`OKDJ9oSP$9diLXupJ!4b|<D1DQ|i
z;*i5>4UaZGl?TRmvLybt8?EJ4^MkBT3zjB)C04_^*mv<jhxo?38~pq+xV5)FI(&?<
zB2d}>VZ)^om0c#?$qif*NA~`p+1l3DpJ0dQz!;yns`v?pOkb_=0Ai|uywS=a+5;|Y
zE|q3Xd6@?w4O__(tH-4cyc?6hgmVINT@&$@B&YC8aG1N<Im#%_`D|#oRzf&okr>SW
z3#S#1|7e=aU!QYl^c}qWeWO=?hM?Y)WcQ{ZzcfnfA^U*~Xm4vs8f0W8okPIdj%w(8
zwlcPR77T>$#o$hV%mcz)B8Eu0h8MATG%AFN_u^iSwFFmBp6K5Huk)J^Wrf}!#?YQ6
zdats-MK)qGT{b*XmRo~)6)b@p+U$Cm^Jbw+B3QefzjGyq+-{aw#Tj`{3nSkQPT==n
zG+vutF(xC{{F*H>84xU3J`vTH30|Uy@|{(I$qi!?KJV+{4!&;ug->W}dg&np83M9?
zdURL=+Z4|GE@5$gPBrBYO@et>7$80t$Z*Vo${3x!%7s-=piP)T7Sdit(u0D|Irs}=
zLhmZ`&;_0XXwlqEI7j#AsN7=`%CBT!9k6rV`}gnl)tlBVS-pIj428J3gu`J2G$%K*
zmHMe$gcBw5=VGCjA((~1(QucH<!Dr}mz6!cyLi72GEj)e^g~;aVM}qG_4U+hNXuhf
zz<suzNC#2Kr9ZneenHeH_Z$<mR+dt8d<`%0HqW+&@SA_Nu%kr8=(K$*Z?#mmqr`pT
z(O1|MVTEo@cQt3s{RUL<b&fWqt9ZO8zn@sMrOr)+=gAQRqK{gifx#TnzHtT`TzvPj
zAAe$w*yO&@8}|~Xt{W{L$z+W^Fuf$H$%7pJaf`gwfd(zoG_e<yG8$fbv`uL%rG%^L
z`M)U*FVoU?e>pnC;@0LS%-luQHjakD=lwO^Er-n=E~Otded5|{Y(-{>{iqpaIo1Bo
z^k!bsXKsIyQCw<&4Tf<WG4H|e)7)DcPsB`dH3=jBO1#k9tl!WX4#76<5C7?iMRj`j
zq13*RqjRKbga9!otNz?ppw&+E<PqYo#OY}?yT=ooT;(EJFeGT|h7*B-X4wM*+Fu8`
z4YG8a?{*f%(Lnik|6iD$R*r&5?m~x5_pfdzf|hId>^>lVx@%&i+-KePL*{R_ykhOP
z>`?Gh?e4a}ppsDVD>+!ke!ZZ^l3^(1jNU-88ZDbif0eyvUSdcR)#9zfI1u{-Z3Iz^
zqGNB~)~Ei9-bX;2>C=G;q(^^N8~Sw-8)5^+*h)YUH0q$igHsiXdYofG0HFc#G)hrN
zm?{z(1gQ?72%KCVuWgjRHcyBzxEY9YY9~Li#C%~CB#_9sD)*}4G$#=>^cXmS(%1h0
zy%&OngG~`NaAp<b!&)wl(z+cCk>cyU#B&PRiws1lLR@wc1(Bj}#W{8*sW|cRqyYpr
z<ewH37Q;Nf0`!bI77`RRo5Y5254dCq=XGF7k%@*(Jmsl+Vg>6g$Q4?xk*nE_Pois%
zkf8yb3!qZL!57q*Z9SBcc62A3NJ~X>M0ZcA+25}}C*}H2SDAwjJ-r?acY@AEFFqNp
zFaLF&igfq0SR;{{a1d1}{OwCfWfNA@@5RJoFpQ6*8>y>(?#bk8L~|*W$;J*cFEcYo
zz}?l25R}U&&Zf=5SyY=D0kr|xnih!SnqcTdSRv2BTU_cs@9(YkV|i(I#B-eZsdQ`d
z2ua|Nu^N>gzdsg?8MNbhwB7Duie5Os={R0ch@iG`;o8jL!+cB4t$PKq4c=9B=$>o9
z%~g-~LE72VK|u6&;~NPL_Wb&Ox_SItjF>tgzQvBriw!Ay7I5VDjLB>jf$d%D^`57t
ziAGo~IqC8M(jPDo{L5FOp=$tz^8HAbzL4Z>;+g}Lt<JzaW0?Tw;0DqEz_-V+h;izg
zj5TuU`&8t|uHE~Qu;BiEoffW`>-Y^gG6iSxcw0tZ5rGG(N;<~~oLkh$Ee)pfg;IQ6
zizy(N+}$;B1E1!Ts1}$96)Peak*3Cg0$}My!~BC6P%kc|*TqN+SbfbXC$k5<v-+rb
zLQRKX%={LNQEK#7Nj&Iad4E^6aL{xi3Hx(kv67h9M!lsh)LHTeLa^f$NJwl+%qDb-
zir(=PY!X|@WP7*V#!v}ncG^bc(YoAZ1HqY0e$CsJjx+^#4{V$Dmqh9~ofPY~b>aBu
zOu#3KqgI~kU@0%IZfEAST-iC{bHUQ!&RSV7?;(mxN)_y<C8r9{aXdJ!`l`m{eQLZE
zovw&<yI(<V{Oe`j4gf26_dV?Sn%)+y_^(4`xfPp67h%RXflMWvU$9;!C-jt?IF9=h
z<69h>;<CJJ_Q96n^Q|EUb&8(DUX^3Om#QX@zZBfB)sUQp5D^?3iPEAn%v)J^tKFNG
zjt>Ux=)mE?cuYa&VEkmf^5PNB#!Gq$4)J7_AqivFOh9HTH`hK*&eRIw^bT1U+}Go`
zQ~X%;SlsV3e4pwv8ZNAMzyFf;Z&|AHpw2Mk(ls|snYR6cYMwFUfsXt3-IX(m<T2?r
zrJIrlP5qX2Z^NQ}*)z`az$L18<g5yDXfhsoJV3R?Zye-)iPzZ!t91r7J?c6#Wn{w}
zBmIdn)vML-iS@$<9n`Q>jLUjl=#&wcOk@T8@V@#++COJ=9me7Z;g|1Y?3_~lUQ{Z>
zAN+g{%{;K+8&JyIy}y1DruuM;@1dr?nI8Iw62IqSe;Zvs;V)Iz28Jc|+DrR2-|9<C
zg92Lg)W9t@dp<!-W>w<Eb=~(A@4bt|t*&Qm%c!QtYNBfq6h1SouNH0-mXM^seBRu(
z)0_F&eKDS5qd)8)gb}|Es^a;G{7>3$i(+Y|8p9;C0=}=K*Ml&=!E7H-pE#D?<Uw|O
z|G*<x3__EsK;?{@epXAhC(Nxjn?|ASX}4<~)`fd@xcF$G-SlScJl!*lKAW86vgl+x
zGffR>bF^L3<MTgVNyMUtad=Swwp3$wUy>H9YHVwdj+p)HK>uahhSr&92Dlcy`fn}3
zGywY#A}VP$<E=W47fJx5a^L3=A2hyz1aZ(AWZtIln7q~TM^lZ*w!;oQclonB4C|hO
zzCKQQIj6rzit*ZTwjwYy_4|sAGg?Zf3>%St_<gB?ROP41t~;*WLu6Sje~(kGS96yF
zI+FjzC)bJi3YkThUIWa(Uwar7$TFuNuvwi^D|_-?j=ueP^$CewiKNDifnRzsrRPBL
zqfLn<%TNz>>ea6`%?&L=+P5r@eqowMFaut+UJjRxB?_Cv;hVo-5Z%7Lx8cilw%sw7
zwfu4Sp?-2?N@?h#tt~N6N~w_7ZX7*P8-fB$H*~6n+c<_eymnG^6Q_9OFMycTP-Yo7
z5Kicq$WN5+PVr76RB@WkM?1ft9SoC|R?PRQXkq{A#wQaVjBPs)*}UxCOk3mXn&&Ek
z>pcsd)C%q16D^Goc%;!(e6m#G3d{@rnP0;sjPyv6UR}rboW52F8vT<e{>n;M)bdGu
z@Ze#pT#=^6+Dh!s(W+N?z?SIFF*64~$k?0lps8A^r!3*=<wmQ+jnk)M__c3$P_foh
zH#$ptFRN0rD&wFF&wt!V4Izgzyx&m_2bElW(A}T5k0C!}*QeNejR=ta=RrxpO2fzt
zT+n=vu3`E&KJEH)bg|5C*(L+7X=jb>ULF=fnDX@tMCu~FA-PzvFJG>W9F$57BfY)u
zd~L$|KkJ!$W9Kw2Az)BmdN8rE$oq>*%eUXSlrO(+T+1S``fAlF{ZAe-$Ksb(*iNn)
z>2lg`H|#zNm7O)Yx9(s{XH<7C4w;au*Ek`)y(y%9eA%DiRc*Kf;ZMeVLgH9(xZMyp
zIFqcgv@uncm%pSixtT*^w^OuJyy16d%;Z@@e(Sd|DMZss`OfEkE&l6p*q`97po=tk
zl0-YW`%5)(&!ACzK}0}+b^i2hH%IX6%NgJ4*+ahMF6s+M$P&I-<^IoPV9ObeH`r5w
zg96*E5;UuUPK(CbPzJ^&MRVf14tJVIY)z@i9XfQ7#?wDF32V@coz?y6#IJqA52`e>
zh74$ib%NWcsxl7?v9}>#3=ZT~XRuX4j)^nbss;+uQeFbQ;~V0kYIF{pgM7qg5fL07
zp!AD|&LC61nMPunB*`XG3W}xnTo4L^q)2xQVdwm~6Byq~1P(18>(;EP1H4n@OPG}Y
z>{%lRPDCpnK^v)EV?d4#0vM2uyqhl#8&0-Yg=yGTvtB+MD6pJEP|<zz8>FKk0g^A}
z+Rp*H$~@x*1^sEt|B<y7#<SjvMARp`I_a~yq3yJ{268`5yifPjHxAMdfCwz1-3Qj-
z#jJ`bqT4B{tGVngurZ*R`*=B{ya)xdde(kV3>50Pu1jq*lltt8Jas@rOxW3-v3;6%
zJ_xYiYePN>e8KF|y;w_e`h|ssAp&YZU|qx10C<E%OmQq)<gI)~R)d2ETaXoK<{jOw
z#5b@0gVV_4rWEX``Q87)RJWFX^E#J9A<6C+VW)R=RFTmH$Au-YiKE6=XbUALgcDsx
zmhx;Hko@3q+;JEJ_5hbqaDl8&xrsrObla>YtY!u2IN8pAu>=#b;YEotAM@P9pTwif
z>!mZGl#C=7(^7a#n}*cP;^P8(A?JzY$kKBH3OA9kn@eJh${gwLfRHwhRR}ayjmdnL
zyp8YucR?ulEEqpR1aJiu%hvkDi4qVu=_UF>HtDdsW$`Q!08=X_Ml`7h&2YBqVc$kf
zPustCXs%+LvF11IGzkL)3EikMAMF&nA>Fnl@5C+^3R!87IsP^WZ}x&W7qlpWMX6w?
zHqZsbGp3Rd5I0oMw;L9euwj`2@`?R<yr{b(rdEv`U7ddH1}g|J28*z}x_5%*&5b5(
ze`Je(cDva*Mm!dQ=^K-NraA<E3haN2gnR?`<6!$0@8vKp;w(@pvU=!j=W@T(r|<Hp
z>DRgkC`14F0>0XMPh}1}l<h$s|LC+zwJfoKf~Kk|AnFk2TeBsofQ2|Ew_eDMsBk$L
zq4L+?VY>ZpRB6R!)$acQ5jWetCJQW0C^{sv1r6&`b>p@}%K+3xdqm&==BOBsy`-07
z^X5A7RL_Aci_?7na6%{hOD6m?E0l_WuRf+iO?J;5b5#S3p;mQNVe_!m#(e<51hO{>
z#d=tjt|p035o*)^sdOJMoN(n?r`s)67W|Di0R3ODBbk#`v5W9#YGGlR#%S?b?D>OL
zwP=jtm1ftBAdkJpEu_@s`Tn<0bn*w7S@x~H8kl||_ds0MN=6IVr5rp3!FzwQ>r<;U
zN6)2av>p%v1=szg_u%$wwdM^6(&~TY&vt*998oAnJ*>SvApF3f!hXxjcSASz4`@zm
z|D7afVO!kyX_iAm*~P|l^tV~E!uHjhH$|~nrN&j4LHIyP$pmAhf$l^Fmty~>H_5eU
z(pg~{T7*kSTg<(8VR;YNDSi)&J#*BZM6uSfX&6ka{J9##GKK7}(dB^M>7(lGD!NV<
z?0CJ@S4!xlTV_{b_Qu?k&i!a`BSVWC(hcsOzH++kd1;$=<%vrfojgb<tT%Db`Pqj>
zU&oaWcOx*7lDD<2BU(;+35M1^SOn)aUCz)Xwsp+nd+Z6yo%_dc%o7+1GWeO&Od(2C
z&teB8ZK7)FP|}|7fd6B8Yfds+H9q?VW>aQJC5(oThYKBI_Q2JR@rZ;79IT;TcW>vb
zj3TRT?`*wNTlQU2Et^)jxlZu5HO~k5wih2c_-0+WBU=4eFHh+--p}K~7-{U@RT}DL
z*CmgQ^5<C9l;>U>S=%l1#B~VM#)XWP6w_)GYpIv$7jVYOD$33n4A##*`9aymAc1qw
zr^#XMl8}?vUK%2IgQIXt$(9J$vh5;^?h`V(Pj^o`7aP18G{$~y_(Jo^AhOT`og*^c
zJ%3@m&ZjE#*FH@W+rjqWJ_YG!(oLtYucYpugx8pSA=@grR65ypk?fNXNHb3M_CNfr
z(`BWYJ7adgJpVey&g~KQZpCNAE?F*e@<*(~93Q^S$mnU&!<m8^UuS15u}r(|Y9e*t
z+0`PhmtB`e@0YE`dXQSbi)KqZ9+B|~*`;`UPuJ8^Jh;#9Csv4*>JsKfJ}sT!P-uG9
zwDR9!4xo}5j8&S%Ha`DQdF|W+H=VVoaetK|?JeJ6>KNc?jR$L#g4&o|@Rp)D`QQt$
z!37Lr+)Fe|+Qbbv(HrC1s7@rMlXzs$4eebs7Gg^55^y4&mpa`83W4DcH0<Y&)Y`r_
zpSK@~HpMDzfICW~>aHXNw<X(!<y>CAnorcxqJ(3|j`ogHYNJjZ8z86E+v{<4k99Ue
zjJbF8PEGUFQd)=j3|A_2PVb<pAHtjMGv3-A%BrR9ifVQtw2HpE#nuPC2Iw7;>3ir#
zzi~eIWSf6^6J^O(1^?`~p)tuOg7z0(CpJ{0PS8x4OD0kB_k~y~=QWb&IShib-+a`3
zhKSg)(K>1`APEMRy(MqgD8<@3TzITG$ja}+g;Gr!oP<o=&0VYOF!xVX><6aRoalBq
z>sw~fpzV+NSRbtEx;0PVtgbuYs6>+PVaMJr1?{Z`!m8pr1-pt@>9_MYsh`zdmJAO6
z_WaqiJ6KU06n?x&sH2vujS&9QXUL?|Tl?{AK7%LQp)IF*Zu@hawi8bICk_N|Icx?+
zl{_;GJ<p;gFWI<VGtVt+dlwp})P8A(_+`s{qnW!!8bTzfWygxu%xVn}ku&fcJh!la
zmY!|?{cisU;0<oYVJ%U$fEBL2$2oQHEQK+=C%aHHBv!uVZwrUjb<ZX|6B;6?vC?09
zoC_Q4Ge^}U_z#b|sZPE&sij0k=gFbAPQMAxQ#K#9R!w{;AFj2wLDtgx7OU0xOsz2N
z;pBaGbxvw?gh=xjjPa=0)#a02lV55RDIM8uU4ZNCAM?q81!_zhJ%6%m#?T+aGT!Ow
zk})m%!0nE=84x>e-+Lq-$=BrkQ$i26Ra33q;#x&E4C>%p149FiT1H6r?QGp!DU1pQ
z**OlVh{smUuFAVxi@}ky7+-Pyv6NC)cf@Nlbw$$a;WwWp$<O)(&XS%xaw7B3X8lsM
zJ0jCQ%-3n>#<cNQ;aA*C!FxR>FR1JJv~g7UoRm^c$fetpn?)o%{yRd<33;Qku4N}t
z#e3cXkLzujt(@Mk5L9(|%|#&Zf{eGP2JaQ6QvV*#lRgy?z+A%QQ_*0m^a7@Iw-&!P
zrs`u%fnVdL&jUHcNfw$4(z+$eSI-ZIr?uHt=sCGE>-AE@BUun)L+?hYh^Msr?FAa8
z>`tD2co~YWQ<L}G)d{8RN4YP#-n~5+h%D}n)r&9aV||jnGOc&Mx6cafMGT8iFZyPQ
zHGJBBT>2h~L8ZiG;mBbTeAP)DZ$LGbY&Om>x&HbXMYa;>)#g=bk2in-l)3W53DR#&
zo*Q1Za-!}L87Z&ukhrr|lwE9^e1BkMU#ueko#~f-vSDYB>PwATW8z$4rfy+giWgcD
zy6D^z*=pa-E`bC8Ro?EqO5}7jM@A2}`6(;JHZ30HEe7c%Pz|7|@wf7)&VwxbG#QMT
zyLUnX$(D2NKBDVx!It9&YqkXA?^}rqC$D5NhTYw&UbrUW4Sc~mkzpO@p}O)Pl$V3!
zy4UtlxjzZZZ#8rcxDp8|@Q#Q!3Q<zlL{SGYbK9Dhe!(JWn7YAG@o70b1wCJ%{B1}N
z<=x&{)u~<-xJ_-V-4ZWWm`w28wcAWl<B`D_67pHl4v~Il(ju%+o>zzl1bXfl9>@b~
znYqmk{A2H-p$bXnk)jAvbb&2sqwEC`pD3!vxF{kR6_R_w!NIDLPW+4Kry1Yc<6OS|
zwcMoj%hRgshq@mevv1yoFxf-B75DA?8qnjSxkb#~3%43h(TYIMNN<YLbJzH%Z(J?)
zUdv}1XXH6OlFQtyl4msE^e@l9zBW9aQ}D6Uq37eCQ_6qW=Q`o)WY%{1aW`bok8gvI
z0v}DRhtPGdCoOR-)Hh_y>b>n9f4oMYDTL2&ZOq$H^6i^m+_#m10cHtiyVYH1kN>ve
z*#;fQ@%|n^4QL>;VW9L7nhRHNJqr20Vy`rf3AXm_+xloBZAW+I+W4!DntFOk>}VTG
zN5F^~*Wdy#CnZGD#k|p5Y5Ptpjm(D66#XqOjnKQ?rZ@iEXIpBCVh<P#vui%+Zym-?
zD8I5Z1u7yLOCQ=>O3HL;!;MY3$|ZNSb4Xc@U1=9jdNIAv?)mfo*)g$!dB3b`dim5X
zAL<}5k1ze%TT-$Km|;JSnZXe*W#89;57WgcXi7<$_#_BbvM5kOugw=Gf4a2x+sIFT
z-+FH27f4f-o)htmexSN->sG|3hu&@<V7hBkR|*%@LVwF1`9X`Eq(v*&ufN=A@$Fc`
zX|)I1-n+8=>0@%@gq``w5&CXBYw%72bYq7$IUC*D`*`Ap#nbleSC})$(sfY;Yz(07
zU5mx6+spbKU<W{WoUswS>>Fhrqhm}-TXO3rL~CW6Ps#h4CXwpNk-<j|$8})_UvCY+
z0NP&=8iy`9N=KUUk1x_-ALiH0l*U&T^a6{`OP<vH%y+E6XY9wCwb{3o#*f$8y_-Rg
z-$>nQ-(8KJ^=nurI)B`(@sHlMXO9P*@RcAkyAN1V@>gdD<-NAcl}pZ0CzL3gYUrD_
zeh%eeBsEhzJDa?}=fyxCi=due9+L>7L{mi6aHm()XEIVnRIEPlPTAVR4ab#e!UC~>
z!^Eree1j(+jN7xTreL5>j5Ck@;iH!j=&97p^g>PWzrp6zckwvv-RIJ6v3AMst1HFm
zh_(fRMSE79^WDFy`g7=;YHPBUDc54Y8_{+!aCgLx&?Ec3;DMiaS;ayPSS*#DV;{BJ
z=g~qZr=Oto(G%$h9JXw*v48pQByvDko<CHF63-scIb+GVle;Pw&_CoUJcgpu)WJQk
zye!f#P|lG-Q?a7{kZ7T#c=Fq3faaSY{zjITdAPf431v_Zq$*L(^hr8><@?@^JKGA6
zzWDs;`SVTKMnSSOXqH^3aqWG2^(~zql(I&3E(BWky+vSprdB=Z^}A)26!p4J3rZ|z
z<n8X%rHkUzSi|f`V!ZaUr8FDe(A~_09-KuLD6%(3+ke|kVo)TnPCb1qh(#RFFXA7N
zy_kD;<?*5s)7>7^+CxZ_*xW1zb`bhbckykcr!$f!4Lu4&vOMO+x^b<G;umRYoe4RU
zw5>#!?u5JKJm74KitK7DFw)ExN@=k;BK^!-`%7H&<dkiOXOFDbDc9Od`zsu*Xyvn_
zhV&Ed#-BcYy7Jw=jfUO~@r=H<$jz;Vau|EhY`1$BJ-h7RxOsu|UkV$H7dWRrXG`DO
zl)J2cUit<fE`!#PS-Kv!N;J2e$pmuB7NbAr{~H$^4+pm%vkGOeSP`{w%AN*;Z@PO(
zt7dn7!V~{MQ{4?0SI1$pwF}zWoP_?GEm=>Wu7g@?UY-<dcu|(NC2_XDU!>*a<b3=%
zn`>-Yy>sWS^2#B5v+rD)x;NmY!pjZMP7mq%uTWpz|9)lnlTCb^DAxq(iJ-ZDzd>f(
zXgMjFy2VoaM&+RghIE+8I<u`g7H303RC<j#oVC4G+bA-GvztTr8utB|Zpq96i3qyz
zSF@-%%7~1fwo7~I(g78?*}!Jg6`9NSdtqV_d)Ul6>H^6M-|<@aVf2MryI!Rqo7ioE
zrDgE8L3&+!)6o0Mx{(!Kdh^_orA3$j<EM9hz1+gMXk-QDSmc1b<pw5y?e8i-^3(cW
zGRV5<y|irWfV=|-4sZ)}C(;dZnYt3v0R*2ilAOu*vME|L$I9N}JH7h9_FRL`;0JGd
zY%9|?^Emde+#$6Kxkr~ZuD=^V3tA5F1AxHS$P@x-f)>;*zN5NvpBrffOQlOz+&cYe
z&GLxNHUEnKf+$x&3^#rjX43yCdXlTT#!aT{RX;s)r00zms=}@W?w(~8dw-jGq_S7U
zl>JY(;UoMqU53rx8TFzzNj2RowsU^hCnht@Yb<v%Go*ZV&6WPkBl^pl3`^q7PUse<
zU!NO)7hQv=msdN*dCa}OWqOuW(;Nb>*x;ajqb{pY-Lf5*32&y4PSAPZAOwRn=`maT
z>9$Mnn=uC%1hYoR%*xV|+<$-j!E@#0N?69ClpNTc#myHyRrQxQjSne)l|I2Eq`J1Y
z|Kn0ljLN3Nhv@ZO*1l1`XL#VZsl~;>l<jO2D7Ysrt7)+h;{(qt*KGj?>%*-rf}3S)
zzdzv^Te83F+0YN7Ov7g+gL2lk76zYg+wdV-KXbTN;L(N)Hrmnlx1C+>lO`pc?b4gR
zj=*~Q*GH|`hUFltdsgFQ>)A_&;Psk?L;mOa!B)1C+m$!x40y5S+2hBnfeSDhK}_g)
zVKaZtnl+eFKbseahs;r{{i(I-rzl%EG<k<msEUp}Gh1zD4bkj#fX#R4n&5nC1soLe
ze&oAH)StZgs;*ORw=h7q==j?kE?>KTbOn1ccRcU5qs{x9i#Bd72lgUQ8`!U(DguC-
z;ypKBQC1TkChF_f?e4JXtoOqjw~Z0YcE7ewn|43P_;IghpGOu7^ryPDC<iJU$<eyP
zA6IO8pY@qnpPX~TW>LlHI>Sp&ABuZe+PXZW2EdJqJRa{fM26GDHRjB5!=jN3F*-X)
z<=(rv5K&A3@knBp&d0~|)4O-;wyP{65dB<Cf_n6WJ(uf?k57<`F$YQW#_$P&<DU4t
zZ_mrmAEa#gZrS|$8I+39Pk!2_E80z&x9zNdJ<vYmT}Fn>KEIXxyj`EsQ>Pkh^uD<;
z@9F-owC;nNc0~=7uGxcGH1uFkhQ{^s^wbGldozRs{xl_J^N0Af<l1ZQjdzxWocT_r
zao66|EF3S<LLT)p<x&^D`cXY)4j(?eBkala;vYZMA)wjwf{l&+_#h@G1}rQD2uzl_
z;I`yC@<^XKHQJr)44jGzkG}mpzg|*1r|S^=O=ET1XEJN7o>Yjbl#HKFai{srA&pp3
zi1H<;thVFqT~toK&y$bK+Vbo~;Wf30Wv_NEdb6NJXVS)-FyPTEZj7OZ7GNxxaZ;+-
zhLj_{5XB2}mn0Za%=E>QGqb2P$6<_)`BGd_@ezDQXXj2`yHd6mPFlFg@(OX;dbt&T
ziZ3sX+dJ1bxl`G~GZBX4e#D2eGpIh**_rQfb$m8;!;j#jBdR^AzBsshPkrRFg4d_3
zFP$lRdt$>Wi`u)oXTRB>oOy0%ukFu_uUYS<QXdk5!~6@v7j3_yS<LW${ferp9AicW
zHOO{b0N4u)4_AuvqF*6y0_~?Zy|)NIs2}LnWg!zPPW4W2vGhLj;&t=%zJmt~y4SwX
z;w2?f6U@|o&kyJ^W$01d%)no=GiI1YMU4zKc(Kzs#?{5f6`G7pzN7ou;NYkvzuVbB
znTVzSm*jWOs{Pck-wVD~Fd^vL;EVIQI>;n3jpSD`zrtvL1j^wQzUO-3fM{iAneo^Z
zOUpXbs*=`~&5e!5EKAQEq{aCs2B@XiMW(5%mibV{^`NopL%-=B8JfEFHtBE&wC75H
zAIqR*YAHBm;{q5@jRYkpIe*BUcb1ryuP@Fjbz8tGBuXGWBu3+_M${gC0!oA@Sb9ru
z^X91X+U`4xf*%iyp7fyn^Zc9F&!2y|e(IGMUG??!;@o~Zfni~d&e$+q+u+jI|JpWv
zv;yzZn9kNVw5^j|*@x2f$+y)$YSQ7i=wYfY115xeMN}pAIY6=pIoWnT1*!A2mnmX$
z8qY4Nv8c!F5n}vD(dDgV1yIaorN6aXHg@F|UgvMIxaAEE&m)F(i3m(YvHMmx#HzYy
zmxZ9@328?f6a9_2JL_GAoj2V7LKWxNZ*TP?W;tm<K%4H(S-NvX>(%LPkP&nj$gR<u
zkhWv<_D`QarB_@gHKI5&-Qj&~#6nZ^x}QHU+ax7^_uU<yb;mBXyLl#0D9|JNz9F6Q
zydRnSzMW%}Q*(IteJk0sEe;oFEpBTY+nU=9hRQfQ`}UG4>9k)OlW2N&Qt8hrj=PsF
zOfLAAHmmk?AsH4|<ZGGZrP!F$i5qv6Wy9>Ax$TaO7wPE_3TMwAQ)d`SZP4mTgWFoG
zJd$8<EdI$t_>pm<i(cy+0V_ZFW9s*NaVnH9mCruB@kkp_0(Yw5@r4b28a)u4Z0ZRN
z&WGgUehsCcZ6WaO&m>kJUuO>hjyUowX2p#KcViC8Jum#Q->8=(edhPNxm{?OmTBBo
z@`@*;HwK(Ginw!UMerV3*Lht;Kr-w^!m}Yqp8&VYbNi0ec5!w#S@wB={8pu>>r?|K
z_!<AjL*WMI0J<pYF|kzK2(K7nowibxD|{WqQ(k#})jy^^t@<3G#kk&3Mn7F|+y8Og
zNzuyYYi_KL0`2RHrumfAE>`@v#@wt#b+Ks>(~`db^U5&k##(yArzfNJQ1EJ8OX}gI
z=h1hE%eO@WWQ^5UUscr5D4?m_zVNFSOUAuMRPT3E|M{%bxiN^Fx7$w@^}e$_KHNIN
zB;E<eF+<q-%a{9>Cs~>w>#)ZtCh;e#3)3}Shcgapx}suw%PbCdeRb!F^wl1uFS6jM
zmO^5WJ$`w=WM(f(4=NGbm5RT9-Pjrzy$S&ZPQ8Hi1A^EI0}qdXv*pG5scmovUvoLe
zC_Mt5#|>JkwyTkZWW5r7kG0%ZY5s~#O-VU;?d-;%23VJpIH_Ah=YWtX&|hd_NIO~E
zCxdfp6U)2J?{k0(B=~yscL$>{OUa;Xo-kylL0y3Se`^8aH|CV%Z6;5iJYZ(-%EhFi
z2nkh`c5L5HN!y);&Edrl3y(fpRf=#k*3$XsD61taWR)WgC)5AcZ;*ZOXP+q>S+{Vo
zRVpT6$1S?D9$NnS@@0jR^s=H(5gdy0co_AIZf^~>g3p;w_Bn_tk{UYoSqtq<KbrOr
z@lP>-Nu+1^b&Iq=BLT*Zv&ddWq7vWgtoS|euF;X%&te|oGgI%*B(0?ITG8*<Td<vZ
zTJ@KwTXJ@-TJ;%s5cTc`;*#{6S@<@l@m-F(Peup>KK{9h%BszBW5z7l?MZLGaV6bP
zs**u3!)jnajgwmF1evb%4gX`251F08Az30nWE0J1#TJN*-oJl8Dm<&d=1O<6B|qF8
z>4;{g9S3i1uD-2UJEs}%YOpiA{QXUc_+a34loJG7`eLO-lFIb5G`9s1Db={^(9v{4
z=#}%o*10FSlKf81MIISDDPr65<yj)QYirEbQ;8fmy#LGl)B%{dzcK_H$F=p1=>K@4
zMpMAIsj@Atk=<G!J;pieS}Diw>)M;GF>rf<-na1|-(rDNBX3@*7<mvy-;SabNxcu`
zo4wmq2mdkb7VMQLPY(AOpuF-{hWnJ>>FKK`F=S?m>H5cy9+QCFU}YD<8x?vmMktX?
zyL}U8&tLv2CgEfK$cr0$Q%D!-QI)xVDzOA+hY;I0-FZ~!`D!~#kJm&skh@)(|FXEa
z7_(7ojAoEG25yt3_m)b|?5ESKLTh_F8av8gZX10yrrF?P&*LSRT7FL(+;hO`_t=Gl
z!FEp$#0jA^8RW?NJRt{Ev<_ojl!5tZ&e9ymADnL_rWR$YLMW7zb<0mpFTrl$gnpeu
zdCCKvq__S!<bHiMvkU`XhMbwc#R7`(=GTilmx>iMZW189ciVK~^5s%8zZ0gtv#|9u
zyTTgNk3L7Y)l$sRJv+nk+<>OH6l_c;8(>}tg=U3|C(r$IPVbmIcN(F>_8^${yT_hM
zfS7w*|Gq{h5z%pMG7ea@NVu76eNtU=CH+Q;Z(&C$=3k-lYL)6~KO?E<C!-V3pI^lv
zfLg=B3Q(#{HA=asORr6KOAvJz06G8JFF3o-i9wlr6uKp&90u&X!NZ~VCBV@QUge;A
zny;G0$fL4F##!*4@n3Zqn7#|qS8TFj`;d**_4Okrt@?X&ht0I1=C{l$8a*zA4vV!2
zmi_KM|2mCduADaP+8dCAU;9FAc2aY>QthUa2PVF1L(Gedio#~KS0@TZJZY7ns*#YD
zWHwYLPhS45QWepv->H11QKSAp_TD_2%eH+VRZoL7Cz47e$t*)=QZgp<SduBxD1=O<
zRECf-Q^-6ea|lsXGEbopC1c2(4EwnBzVF(<z1DB7z4zLCegAnsYdt+>_<ZjBy07aT
zj`KK=^E`}9O3T~iQ3%cg%NlLZPPEL06GBJ7%<{v=h)rgqWXnP%hK82*f@B}dp+gAk
z9;7^i(kOIivYVd@n~t9^r6>%wX-!%B+5;|C?u!}(97r$Zj~45r&?|zKn!E(CPR|?3
zvMr<re`u|Oz&fFtqQm!1`L~YGsPiFCJZj-F5vAgQ_igfYJ&UVVk(Wo)ML^mDSZgm)
zwFM?}t=9iB=%#2sXrUVgQ2U&J_Sph9DbW$KdhV@<DipvI1Uev5Fh>4x1@RMdY8Zd+
zfY2TlqI!#?{QMdYmfR}6FHv`f*1Hr=0WJyxB!L2tfMH`MK`K~e2@yLHn9`)n%>3Q;
zU?qNm>VNyB5zrVq^rWK`)Trcywx;G3_{f}RU#b=qaVkXfy(#h^vQz=pb^93g`n78<
zex@$`!Sou=y48CSNZ&eiMU?su<`&r8%-L{ceSB0@ly)R`V(75U$&-Wmgp1}8Da7L$
z-x8>Ym>Qb<fi0MV^)}QOZWsmLDpAB^>20;-&O4%TNaZuu<haNUwoD+To^q9Qh89Xc
z$hnZ`25tSt3st8J1#t4%omo_Qx877J>_u!>jEvWEBl=7I-3gR-AeH%}0w3VmmV~%#
zCMYFn4$E8L59_BMTz10`=@;50XT2f=u$JNNV^3kCN6;j<PP%@4X^17~OJL(EK@f0F
zPvgwJP($B$4xzvP8)qn(W7FqcZdbpQ%zd}a53@!tYlKowK9*xp`#Fgt5m-ach(1y<
zvEHE&Mc|20$cAt<*ruw=CMol0Qt#iwHjP`h8t|*BoSvCM!PyEI2jOnp#+a$GZo^dR
z@Y#A(J_YOY!lELCoxGBgCCKnnyedLTxgKsiRl3Di4t;QGqed8uG~|8bZrL^j7W?<_
z-?8IaN8Q)2Uol7|i{W%=5E55RnNXV1lFZyrO|3E0{qk&J^0VqgDo@X@l7l^aHRY-g
z+#V=hf|iJsl;N%YwdZ*WM_xZD11tEHP$5!ot&;+E73jiLzNE{-%d8G4?p06$Uuj)_
zblILBHHnb7qWa7I@%-~^rR_mZ)=Wi%PRrRCY=HSio)#prgscs1{76wKb<@^;LKLX2
zy{$VQIsVZ8)7AZCq}OLnUERYBB$usX@(dMu&V>gOpykJM!^+MMv$z7X_Qf`|#ma5X
z>_Ge+WQ%2Ea*4vciIT;cGmG5_N4O13oQR>BK<H)lofk{BD9af$zfBiwv~-T>6a+dF
zj&4d&jo0u1^aeFYZO6Y#4>mEVBl_`r^yra~kGAA_(;xl)iI+7B=Uxnu-0>$7A4yDm
z1^s7>NLXSEGc$x~$snshO&$X%kT`4VsHrceSD4DE>+hlr_VdYHD@{pBN#xV;Blh+$
zYp>h~_M}Lj>-bZ;uFF6)r=Od{e1l8V3b~MUVKOrzO*mqxZ5kdR$JDUz^<a8_SJtzw
z+D)NGeovn0K9*R;Akf?7askfM%Ep)C31#fZaH<U(j?_s%eDp~9^l4fSI5#JN&yBA~
zVNErCZToBYSx{I()~``*8Yv98!OA*1I(N$$Prq-O-v>9rp2rV<0V04!LyZR_Q{R66
zl)%g+lHMY)X-tRZ1Vr0t`+(#bX3RKc*wY+War}Hnc96a$&=>L?)ctL0vrf|)MDUR}
z!^&rtBgrHpDw%8T>}IiG$v1^rSWeC-<6bWVn?x!d+{(tuNl<Xi?5q~rqnXhZ_vmb?
zi@~x%*1a2A)##enat+~gZ-AAU+Qp{6`EdBD1OQTJLVn)@t0rZ0$5C(MX!eRc3tK4L
zv1LQ-nFHM1r~5BFDSk5$jKK!Y^Bnxy4s@NSe2cTC%Ur`dPvX?V<LOzYP26h;N*2#q
z+B@<t)5s>^1G-g2EuK_={_F_!1E)QxRY6%fQp&Z&#xuU`JjR&@yL@c?+=GWq<{Z6b
zV`Vi*I`_j-s?ljziqX`<g42;>v@|rf(2I|}+c7S1(7-YS>OCmA2FiN6qB|d6bEBIo
zT3nB$2WziXFpV`}@wvFUTV023o-B9Yzjx0YP!o(|R#Pu2J*4Zi6q^^27>D`jhS=~^
zhb;043tw@)1=0vo1IGbdBYoZ0mPTsJB5z@VJ~k36>fHNM_KDl87MflJ9-Fi7Ja~FR
zvb}`FiHNm`+7+E?q^E#EqQgmP4ADY>-~owCmm=q?;*bG3q&ni!%sU3LdAp!%JN@>{
zP|IWOrhIz&PQ^t<q@yt?y<MEFI*Oc^IBs41;m2FZGPTuw4F6+yu{Yrh&B>}8q};r@
zrQlEMn<3s?r%(G#Jgvdqrw`n1p^W{<=Q!QI>&NdkXodM-u-LR|(`C3Vpm1hKN>=@|
z4o>&tzN<*38mj<T0KtX5M0{Y}Lt0vZ^$aijpiI(uuo<Z^`a~>Xl-}wo#n*D}U^^j&
zKmi(>AK+?n;P~6`dDYj~19})NRhoa$<oX6hBMz?`7;oe-Ng$*&zEptktPtPec9qT<
zQ6h@VZn{n42BeyFwB~#F@>2fkDCoDDeNaDZV1U7h-+=6!nBXsF-FG}U#Bxx};gJnE
z&7Lybgr}1L|HMA*l~X=&D1hzAkxVSO=$|UOP4|HpTI61s<SqPQicyIE%96P<pjc^U
zJl@M7>k4#W+FV6U+<OUh2TXgHE|qq?*Pwi;3BW3sVOu)3P`qSIOvMl{3MHS+GF|SL
zMNbR+EDEnY15$2s3S(kqj7Fdb?-COelQkZJsX3O{wSR!$o#p5VZ}ZUC_gZt(<KHD|
zjFhmq=Q54mf#Z2!o&U3_q@(i*6RBaBz0A81nR^LdaioVH&6Augj4Xsv11vJ<THV>{
zj5g_B!60qAMMNHr-{yb+&b`=hd}J8}AVp{XlUd987&{5L=L+HwV6T{xWJdh}0paso
zx3aWWeksfMB~V_#=lGbArzL1^Vm@*{Wxp&D@mdD;;EAE5z5SDSOj#9DPEJl+_6kQg
z^#elpFyze!G><w9hO?#sDF!<Ec(F(47&+B!P5Xo93qsh)ew^Bs+Dt{B1ISj<r-USn
zU%f)FM|ww~MESsv{NuJ+1wMcXL9$ew){}kkpaG<X=iSk7-VAKn-Sy;D0%8#Keb-F5
zU|4O+xs&?kSaFy(Ge=m=x<X0aN4%n64?`TDRff)>I2qip<L1FND2GQc-_HJE+E^uZ
zn9|M5>myc2`H)8PibVk~)ZP~w$d{FNQgA9Z1*@YUFm`$FeqvnQy$275ZN8y1tfsCm
zWQ*wZn*jtcDZAt^U|55`%2s)U;#*p_ci7^QGO$H`Jlf9WeN})yXu~j~xr}Tjy*C4}
z$)cfEwq&ynqOBSwrQla41a-hDveOzflw={Ao15XEhgHJ(t!-_8V$}1{JIf1`sV_02
z6l|XN{>PIr#meB7y;ZV?;!mr{0*pIKk2|c6_zC1+J2QSQt@f3bCE?k_Cq4u~7I$bI
ziq+jGyG+dKXTsanwtHh)yG>Ro^c2w*z|LWDeS!qOh3Y296-?q`ZJ$Td78)MTVQ}(n
zDjWjOrF7o|lx_mCAy-sRU|2}jPHS~Uem|C<#os6U9!eoO9cr?{wn<B+T6}by(7EAn
z$-OV&C%IJ%7z+HbekY8cVAMv=O|f7Z{Xj>~PU>Nfwd>b^{P8LkFp&;07$xTr<l*Z#
z)5q<tZe#X-e%7?*(<hwwBGIBiL2nmgrXMugth_0#n*Ol)d(vry5ybjG+g9!xh-8|R
zi+oT>$S0J(ldM&me@GAE`bqPy{zqPUB;#O@r8cj-()GsC*q%8IzM}QG_JO-)^#C?e
zwMq3gwxu)ptvdwUWo|2!b^>;9XZgTw>a1_aC1y#V070*j83&(ZdRb7wa_}ICe>`X+
zIXvCFj4)YQQ4y{vz(&&R!2>x|3vuz6e$VTW0u71^ux9C}Rg7r&?tO*kCxlNB^O2q2
zkpr_+Cq7w?ZsV7z>$huiIwqaTA4RDF5kJm@ghc_&IGXf{k^!D{5S#f!hr(Vg8c3@L
zi*u4d(S^7eKm+b`5yKdW(}iWhE&`E_J95{xAA^It_v{e^p%kF=WmI2{Y1aXJp}v;?
zP;-}UPFUIbdj0e1nxH&iH#Y7CHJAtspiB3JNGa#x!%1k{Lx77!BjU9Edl$^RfgsXn
z*nAN9X3@T^;gbBSxEL2AfZ8t5PcVwl^_#$prDTz|SnXL1%K3(hX$f#dq{_d+>*83h
zTwGX~pB*d3*YGpo&A)6ObVNqONQnllW&AYm8s79;U$cvfN*$nsYV;jpd7fZMq2)7&
zP`Jkrir~1POL#}0L-qhX>lsW<Lgz@V6aR_h$4Q&*ze)|X_pe=6`fYxx;@kJxM+mL~
zInE;ng;1A|hlg6_T?Qy>kf}dNNU>9cGi-=xC{_xnyUfhP7Lo+Qpu-U8?P%S;hTx-o
z%V#*6vOf1oZPt2z)&b9!L}!3o;<ym6nczR<VlNg-2T<8XJkXM$yeCrZS1<-w&Y{;_
z5F=#Jl$26iw|Vns(q9iE?i2{YLAI+iXwc4h3ng?k<DHy2b~htq3~do;2x97+_%u&u
z_kfi^tk{g;#?71eNx3Yb`I*xr9^)?O0jS_9t>7XCe)u2qLI;$N5wK%QmIsT_p^q83
zNyHF=!H{48{9b$i0d`8>Ev**Vt4NT0Q1=0NscG;&4xI+*e4POD04q`zcBc9em_cCY
zIDz+0*gx#-iXtN!Ao)OYq6kGIK#wRF8JdgGwrFI(RP>`kUvv!iX(XC4P$7l;s4D%1
z(Zl#%qCP?5(kU65)|C@TN^bb&WdQD^Mn0n(UUvSWx3?2Gny5Q0f{52)nx=jTz0H7T
zVsqk<{RW?HfOw49kte4PrC&iXu|p@>AF{H1tlRR5G}@Cy28@PGtv$DMO%Fujke<3R
z+9QB~#BgFR%qxhzp@#1&E7{Fcp_}Inv*X5yG#8bB8BQVkM}T&UtiIicCk0&3B7Gnf
z$rzF)A_oSL6;I#)bb$n-Ns(u9lVYBLauGt{SO!H!8Dh>8x3`y<Mr}{-TaX3N;fWVi
z^FX0zugOh$bN|}QW}q1OfC~cj7;}8r%`NUz56Z1lsM7t;od}tpm_P-edW{hxo5pxP
z5E_ui`9FTFhDTdpubO(75`jO4zJIK)Ch)&nY$f3QvqL|re>nm#c5-^9Y)pmwMMME3
zZR~;lgZLvixCd_<z!h_d(hc*w1ejkm!lHV4dIt7^VUaQ{l!6Z2ViamGTu5lRh@ma%
zaM;PzPBMVXcb2wYfKsr>qeoh;fi4^}a8}rPT+@dU&;hnU7FPY?!!O(+eg%e=%c;-_
z0$`1M-?=f~b9cvXJGDG~MP#zu{?@HqU?*XX0K`_^D!VYoK;?EFdjWC?K&e5@t?BMA
zMW_*c>M-phI$aR~`32)KVyYS$+m$yIqldf=Q?&5HOt6B~tUzV$BI|x1n+=Nm*RG8r
zO^z(aerFC}hbf46Nts<R-to5ty0H;6Xa}o+P>-!h^JTXhuH*M4J3ITt?Ce%b$``y{
z>8K(g$>~Et1wTbPKc>D!1_$4q_^vQHj0s-qTLZ(wZsGQjZ2gilm65q8tkda|7(>Jk
zbTUvt6A~5{K73eyUW;^6)~uhNoefdeQ0Jq6gwa}%fcO#m-i;Ab-g<>8d3}f`0gzYU
zeSmx#*Ns_sN=wSbCn1jCcF)hx4<X}lvp9}#l<N{3r5ngOOU0G}A8fF$n56}^?C9v|
z+qWqO&)9%Q0dS1UvUCJAEfZto>I}5oFdMK#d}I<)5c-rhW6~+kxM^O7<vVm4xM1Mk
zq+TUfy7^boR9d=iZ?@YFx+qi}CgTPPVyE)?-u?Sk31af-h(|Azsx1*Y<5I~90tX*|
ztFAtU8Zpuxv|B6nhyIiNuR@$Q-OinH^RHs(()}QU7s~Rks;*8&Ss7D;s_W{I8yA+A
zmiD~~hc*L}{ImLv*zg<^;g>M|^QFz8I<{TedETgzs`=%nE{RvH_k}Kil!jK1Q_9Ln
zw4Q^c2Jph&utGiA#{jyBOcv!T@gyTJY|dNp7Z_z0qg-fOKqwe%b*QPSxtus@Zqopg
zsZdL0q(sgQR&2mOfBrePZB4)Oka`)Xiq7hOqu<uOvxU&}t6ExzDPCU(PD@-jt!ts*
zw62AwZtwfwj!CC3L*jO)oL$O+!%=LH53Ng<KkeV#SMuv?#Qo7dS#jJxqdv|b4os}O
zE1eg?!+v{2nj3Ip`1d2(O#Lf}^oYz%pf}=rn+?NMC)9VnR+ArEEzIO!NHKexd@O@U
z&}m1RwY;IH0MP~nqfwCq`@mY&PE*)*>so0)mh+1ckB$i_`(5%Ftx{nAS}4fKyhZT7
zw|O7MR!>5-nUDQoIuE>OUrXrvCX_1(RgVkZ-adU>e``_nkM&3#d6oG9UiW0)`!ays
zK_aZHD2|;x;JxL`Kd@_HXz*}}ATU5Ev9Bd+eLX@Fj8b^J9|MN~$q&9>IxOP46{CmH
zMa8`|@Z-l!)bc`Ngh%+;2JFrPk+ly7;H#g{*>F=Z2p~%d?(<!O0)sV%9m-AjugWp|
zNIU2YlRVOt=o>2<TlL#<VByUkG13YiCw0rwQH*yMgDiNz@iDv<LT;-cEY%RP`HkA5
zL;d^b{CFs;wOX!MYJWT0LpTMS@7FuXRVYarE-x~PiJd{4BIX~S2>!fwoFWW#5COU0
zxP-KGntQeVo`&IM5+zpA7yK!s=?qk-MX(NFM#3b?E1`a)o1m4kisH6}ZsFWkig%Ea
zZX{X{m?&*w1pCl>JoW|Gpcd)uH=FzuQC8gY1_>7mmSQjXX(~~r(94$9pFl-UG=(2`
z4)6}F&iA#ov`TbGHR(`~1t;m(I|LLTv;4#&CTCS&?z%5Dvu*EQPUqVy+tw^<h2SYv
zFyblj4}mSGC?gk&z~+e#tEgy4W~5r*f&}tFb8w=5+hVR85Q4p*qde6}cOH{F2Ss8F
zNZxXGJ*%|No?QSRBGeEQq$NQJ^7-lO|L+e|MGXL^M63`J!pD4gF=XjE(r2I1<7^S7
zL~#jGNKC9F*$m@TQ#KQylWMz9^qiyXg7oJ4^+r?6p~yB+p;{w7j@jxqhUb$pmmaCK
z^fnN__IB=~c>T|Vrh18f0H@nESdi$6=J69rw^62h4a<qJ*gm7HtExrzORl95XB6Jt
zX^j{$(siDT{0!^&-hnOTFXZ+072t)xf2D}_z;*uX753jRudKxXH*4~trmz#<@mpV?
z+2`kY-y(L*G~)tC0$G9ST06`_5ia=h<*b|?sk9oI(UHm6c|W1UdOeUtf?)XD*zl}e
z&vo^Kwks-r_}4LNBb${k;tzrC7^U*}^}qXaKzQv@jJXA=PE%J0`;1x?9}dFjTr9Lu
zAbYvDZ9n$IbuhxPc_=7o6nVTcWA`H=X#f3rPha2eBD;(^6U6Djc=nN>$BUg#!Qk&F
zQ&7~){7YL9Ur;7v;r_n<`{jQv#Q(^eNZ4l@{+kQ%-$z28#Q(&8{QhIb>dGksAyias
zOuqhm!mQmFN2w{f8wfSBl8yxHm2D!ticG5%uLl-`D8RyD5$X2er!l}}1!8N-Pu!+v
zO2*8OtvP^}*RNaGk&=tSnn)wtQv_GuAYv6U3=9{+_ahuPtxNb5GgR@Hzwf^?4O4so
z2*9<}1W4?kPfb|)#`PWK$lBK<*m*`&sIIYM?B$_D2@5}0zWr5m#VN!9k?6Y?hg^8w
z-I)LNTRrGv<Zxw%=fj_WKWh)Umt%0>4?R8ERW_nu&jPq=`uvL1xjhLdj4^;QUp7AK
zxE_gFMtDU~k*jB<?E856Dlgq-e@rwXFMx>n3nem}`}Zrs{m#Ee67J??lcv#??|<ba
zS9g0kQ}oEj-G2?z>=E+eUtJNYe;XY=*IJC8i>J;rdX7_nZKe4!@|Fv}rKJS(!1qDp
z)I_!6|ML9`&nZq*n-yF3rF-Q=N{J8%P>JAAvr|(}XoDU3^w$>dC9a-g&MQCP4$u=<
zMmI`sC@lV&{`-Lsn91!fVF-f$`vKTk3W}FwD~8*35;GX)=I8a*o8G-+WnwDpT3q?O
z*O`AmFB#Kf5Xo~Uw*!?yyNl!O-#2`RP?Cb;JLjpE<Fbe}P)ojhn_Ph(8GMVd-6itx
z&d`##Bj2UCesJY+)M6qmbZ+Kt5u#vnu>O&MecGLs2TT)x=rU6*HW=1!6&XZIk%SF#
z5~R7Dy!-C%pe1RVx{~iw5Nr-l*8roAbB<1rb$6fV)igE#KYy!})c<lx!H=e$dgU8?
zACe1H-G!))h9;i#8rss}eAr8*R=z8IFT6GnS|X5K&{^?1*%r_M04QeWj<N=*I=1ro
z=-d3*Pvg_mw!|wY3EwiBB0C)6dwC&Nb#wzFZUmn$g9GRArVkWw;^gJ8|NYtj*Y5NG
znX~=xPp>@A|2uKX|7NKI!ecQ_@3ll_Q&TpeGXQ|;z{o*2#~~)I5KzYiAqXW>h`5Lr
z-ktRz^<WTNkgp<y>FVl&^#g1L0H739E$kqmzrzDtR!$wmPGaa@$OAh>=FkQ>4giLb
zP!06}`YfRO0gzM*jg;x&+Z}MU-+vJ#1BfW}ej~4cR<F)Up`oGKCw%=Y2u*tE1x66X
z!i7GlGR!2`Ls<nJHE-G2hKraG3(*I>PzDgv0C(udjkIOum4%t54#<1^Rz_|X$M>C=
zxA$3a{Ll?ktDwQg!eR#e3B|gbHnU@!i9$O-i_k73fDkaOXuWCmpTWGGj~({l=7Lm+
zEOI{s!__gv{S>zym5hyz^;FU*ILYXbLBq4S?7f}L6o+oWHb#E{6b!_Q6pw*s_w}mh
z1zf-qpmMu!-@Yq}C7tGgna={_${MAiqkH+ccGMoLw6wHDRQu6<dmTLp0EX>AP+*W-
zu~5QN!~`S(EBg%UGFr5M4h;<rlpG^E6+oyBiWr^&iUiKI0EiJ)l#yGEWC3nd6U%km
z6KDp3#PM`}MkzTbpaF^T=52ot<3pAKKB9K`)A{c&Xc&25cGq{56^W!0x~Yeof$xX_
zo&^evtohU-4ah_R(}Tu?9$+r2;r>M~)YL>oej~jGEQehI3M5|oK^W4FrUxne@$-(e
z?BF}0nt>v<;_1_Qvzg@TC{DW(BXxrS_Sgokt*9Q<VSk}|jn(nLf@VAqPfz+4N#GM;
zEdTsc^QQ^B2(m+k5R3tc9ugQRF9aqqCfn=gef3X*<w9c!U^_F>F93(YbYiA-3lAql
zlYzy9)0R@2a1omuJGdLY9Y8#T2X)Dn+;-ea+~E?M_eIgE@u}k$*z2rM?R)VqDQAEU
zpyq<n@I7F@lN3I2pB5=301XGefSPU3=qP&O@UjA18i(b>>Z(38X3_5750Sx6EKGl2
zpFydMGuR0jot`J57)tckpcZ3@RRlxZ(jE0GU{j+2j|q9<#wA=a%DG^QYbD$TN7xuv
zH|w~o>oQR5M^Ld97S^@R008bv`s{!GIuww6>cNVy|KdfQ%ec_B;^|=UD2EaN5)uM|
zm{?J>EauOn){+M|)7#st&LJl&3xYxx${850>1XK<pbOYSE#Nl9OYxOgB*ev`AP~(d
zg=S#fE+ggfW5-^T2_&27zK)IYq=T222Mz$f%Pi+}*ls_x{m>4YHV)tw42>8USET)7
zm}zh3fQk#@<LpPE<p4TMY(r9a2OcCSK3E3hmUN7pHrpJV3eYzk?c89Mf>4)=qI*)S
zMSE5s016`4FtHz}7|wF!RL2f#>V2KCTkM*k1mKivsdxcRL)4S`vxig=aKn+-nbqh&
zz>Q9Qp2AF*PCGP^xiKQ70CK3&zfCqa6YjF=0@{0%^-GZn&M$_FSSQ@bKs5mV`pj$s
zkZ`5cN$Bp?cPI^;IFow#RseS5>2E;3=&EpU=~zXv^%=}dWY3!0H~iy|j5Nl3#X7Ee
z{Wce2it8XAJKjP|h?Fc-nsU;K&Hxrl9D-KCV`MTPkQ$g=J-fFk?_w(2VunlOO+cS0
zMdpL4s7-1~Zf>j176B;^<y_#ULnN8OiHHNd;^nK-AzJ7Qx_CuVMJ4~a@;(q~=WvoW
zg<4>ov6+ra9UzPKx_uim)~`GO>OXZ%p@IA3{q>%Y9+ACBFzUisl-xiRod=$kohF6=
z`6d!*1f0!bD4G+7NKJO_`}e&4_a0+@U=(0JZ4K;1jT+7D^JS|y7N(u%m{tbM7oc4F
zRc)_^ixD7-x3;RYX0L3<(SL{om{YQ`$+2{}b*r#7(3AsRkVK^shBvR!9r(|%%z~Et
zy?eSlP6%B>lL6W+Ox;s5uSrRjB38G|jTO~Fv%WKI4(*K0U~=M9buM(anuEI0QZoc`
z%i`8&S=dQ=@3ybl$<yw{qPVI(e`WzRPg29;lA}JEUWyHt5)Ow!n2tc|03HHy)#u<I
z85){Lty_<QHlz^71CI<_R0RD1;0<e}hVh-4N7IKUf9AzL+h2%5L2C)@Lob|K%~lf+
z7;cNc9D+mIUeP~*rrX?Cc%1NuGh*B;YrlJC?Oj2z%E?ND>#5O}Lp9u?$9<Qk5G0>P
z+1ZECJ-Z)B^GmC5kjXSa><Oh0wcHz;nrNUYuDUgaS{2mBnSn0uMI{0hXL<&P=ef`9
zPo45kL%%5(sE>M+&`C<uP7mwrOL?SQex8?8_XQ&wo(e2rZBj(cFP|LNTeE_wGyWl?
zws3+rt=I@DQ9W&ls9v!Rgfn-rw?F(bJ3AXCP8+amER3LCOSINemTO7CV7W61Cze4_
z#`P0+oD%9iPa!UagPiu|h0%#V@nLWg**F|;-I^KgJkw4@_yXOr(2o~=Xi+LhtE!HK
zD&7(W+{<$>Zx1o$93A9ory^JLaw}P;Xq%&(8emD(au-n3R*jrIr>QAm)hp9tAZjz9
zn(&GnYH(^@ThVT*Wf0l~GV8l{a@zK&tzh#i=;?)is;H~uVPcZjYbvZb3C9CDf&@{o
z(&*@~aEK)!UDDL(pJp8{giSOcqrMPT3lo>RfN^JbIY%K;@<v;ZRDifXeNZ4!5Cmq<
z=#7GA7gTNVVTH1-ZX00ZQW)6BrU!b@B!oSbl&~<vqodrZL=n`oh25)=qo=Pg7d9HU
z>#z`LyU@|j2G@l9Qn*?Cy!I3DVLx@rtt1t~v?~q=jEjR@T*|f@*RQ`&-eSZ{huT6i
zCl5Ebq{Bqq>32DK^MC%Fu^&L7en~l5`%%M2x352bBycX^Sp*r`-7O=XfC@^QpY|C*
zCBns}<Yecf6nM@=6aq&nFdVfJ<Yi<4B%k2LeJv_AoE#jft8qri3NWs$Kp#rZs$4|6
zEwnN5VlpZjd_6B2DM5nMR#xRkxDjkwaN1ty2UJG&!$ElDx-7Hp9;_s~BxUuuAB96|
z7T0@PFEMSm0ocd>0RgeGu|Z_WDdwys<SwPuB+sCI0$gnzs%+!=zn{I5(=434bVLP5
z9i`%qv6t~b&hfZzGrUw)z(@%eiJ9dPB0sg!9ox4vRfcILKHD#DKaM{4;QjF^#9?!(
z)##%oq7?=a7C^sC-vN{+9PpH@1ad<mt-@x-3ayZ1DCUR?ckWR*&u0riF=H=9)&QM0
zn*2!*bNsU})u`(<dNhZCegTI9kvq1J;Afb+#L&y~G-*!328?222qWXM0>q6hXmdw%
zVQ{|oX~>D#sH*x6B~ZzKTyakS+Egrve|UIQY@1NWz<pP>mwYf)-|Peg8(kZgKl{KC
z&(-1xEh<i~eEVl5IqZN49LC<hKT71;;4o;9c7dw_2pj$S975=eJj~5Kied=Qr3dLh
zdU}rF4%5@oc9{I)>B@TU{cHF>)YTcH5<YYYY(lh9Bh9XvXa$%Jr^v-2e*Noh%RU@-
zi`*s*eQJ4f12$Qk&(0CloJEvuP2E?mT7~1F9qCzuUaQD7K6T`b`Xx>UsL$PK0-5Ct
zY;tq6CWq+dTp*_E!!tygC}whQCEWj7LbBM~rG)$dTBN2BNC4MSXl7z!ia)9jr78qZ
zr0XWM!x5CNl$BWmWo2bmH(;E=s7)Tt0{>AxAmy_Hx;d~Mfqn2}a9kN#U;Mqfp@jeu
z(s^{?YEhT*7!*#y>Q&d&a5<W%fB*TDj3We-<dPQTdbd`EZCyEH*Lg_v`??$CH6wKP
z>YqV;Oj=LDNQ0J}MnMbuH;{#Z1ylf72RsHq9Vuy~;}>1QW)3(;B*A_7kkY~k<SdBg
zx1eFBr8ezsK~|t)<x@<?O4`&50OE6Zx1qNE!rU(qK!@{xjE=(I^pYZu?J(%z0*HW+
zrPyBkcn~o#avUTAKT#^uj>M0rcFC_izAvs*+Wh}ed@U0IdmY;iWCwlQ)cpKeFm!~>
zssh2qc`9nl&c#(@%7>~q(X#-_?Gl%20~{$XWDa^lY(ghxWcG^P<YZ$TL|Opa7Wj#%
z3nD&LyV8%Z5S0c!SHV4ei2|h8D_M{$)UyPB+{#@&6a<yd0&6)|>4L)|A|SWb#A^qj
zIoS{)E{L4g){}oL=4it~SUNphwW81`r2+T;ANZmGJ00!7hgeuhL0aDyeh-NSUg;Mh
zjL}VonXX>FIxrlWB~gcgbjsOMcf@CKnj-A6UKgR*2BH=6$D+&aS?b{_==*C;7$3i`
zWIW);h>=sNX+}0JI|b$u#YTZeOrjH8IkI}MB|PKcag#8KO&g?nc%VzZl_(M-8}&tb
zU$?V-`H$;z-EP|av)LDvvFBYPL7jsQ&_3nL{u)Um@{UsTItDpAGLn%()rz>sXl^d9
zbl@G>4q0-eK5)b~m_H~en42a>|ELmgT^Gv#?uS9vacSu?SjOpLUH^u@bH7ljLV`R3
zr3_?R<lUl=6X@xf27U~cCS*5xu*F#yR3TT<ix`mTZ)+K$Hj3btjLzk}HVKjo4#FZj
zjjNieiyD3Xqr~ipbSzEl?mz7qG=L-JI5qGKx`CpFE9HvYp0ZbwyeL|sH->0^b8=n=
zhlMbUI&zXzQ&YA3AQg#Hm=QuZXxy-Y*L&?ZVqAo_AfXUt99Rf>`G=Od$N-T2V@0lk
zFtVlY0k<`d3rO2@bM`Iivs0QN15BZ*&=!|~(xW8UQn5uVHugFlaV5In|FW0xOB(pd
zy}Nho{Y`&#lGQ6(TBc!vffXQtkc6j)&93U0K!|=0L~QA8_#J8*8Z9h0l2wq-fHzz)
zHH`rOA3HagH5hd42$T=A7SLm-Uz7#~MkJ?I6JWC``OJZ+X@}Uswh3l~(*n*0zUcyt
z46Sx67(HpZgLk(|j*pM427`O(@4o}bQGgy=Q2~%Y2n}^Wwu8-t(WfEVEAZIuKX;|g
z)jz{2fvKebxOgS9q&UqDqYlilp_pRy3{pK-m8Uyh5%PabSgz}eT>m|AY5AL!onwbz
zm49Z)j%4{gn-z^ms&{U_u&zDJ*O@o3?7Vm9+w<PeoSWG<wX@UcbGq+Pj4#f)Haq4!
zxfs#y5cRX9Xsl>-Pu)91+tJ~({e_>bBR>^9|3pKd@^*%T;?5T=@PA&bNzC-0*Vh%}
z|DS*Ikf{4VFMB2eUjF;_dLs!t{_{Gw0!9A&pW`4v|NHOAiFc7_`k&XU|9<3u7xsT*
z`zW2r>m2Pa&A_XTR8WrHr3V%R`_uNBBx`6(ue^o^NQ$H^9Bo!sR<nGBBga_Y)5(s3
zafFj<98Ths&5CEvgj*MRcqqUo(HD@(q|v=d2?INc)z`X1T*!Z3|K02q|37VZfxu3R
zxhLW-3&K1+zKoR4&Wi|0*@{G>BEeb&yi%VpEFv;EFmO0kf`U=)>eW$DSO?CerKh9Q
zG^mB)b|oqKbadz=B&ex@cTXGn;dB#m2B>~XuC_KdaAARM3>1C{V?-n*B1=GqK8bM%
zdOwLt+)+NDx)pw*=tduv^tQabJkrg6wYEP#D+@!w`B_-pI~q|-dbKg?DH0qMNYFfk
zk{&^fG&Hn|EK^ca5cyQ2!#yR1Pf$>hpC3{c8aHEK1_cE{l&=5#_lFg|h#vvvJ^y70
zYF7<{8>E5iXP|-t@r%ayHL6xxV6U&d@CJtt9fF^O?^nASd&~&*5nSp9w<owS1c8B_
z5p_!H>i)5@Cd$g2b6FS}DU7C6RL-3{2Te#2c)-Yq`||Mc2n(~U5z$^8-5igGaWOG5
zwBMaM^SP~!jCR=OA@k_VebD4kIEpGEQ*w#=-}eh-tKS{E-+*uqA;JbXsM6X%O5mn7
zMc`!swNrGW(%X<<^o+EI1P52XeS2NIcW5XjIr)R#Ly%;^_UAv^(bh)BTP*!_pLYOW
z363CwuE&0UsyaGyS3cF(BWxrT9E5)8jniPsdx7p@Bu)U!aqLo(lA@)W(I1Lz85;Zr
zKq0Fv89EUaV2B*ujvW)d6&vSoa&d8GC4;XBsWral2;6W)M0M9yG^>C8`ntgD@#6-_
z50Nh9D;qX%COH7%GF^Pl`s4RxKNyi)wrm0WImUHlgWEty?l8ayqygxad6xELmD6gJ
z3*>xbh6e^X4GIIZzaveD(&C4%=ZT4*5IWsC0lp8$welS;EGV$Jm^7+Y$;vD)F3!t4
zj<!jlxKlr&1cGWE!j(WUP74bQv%=(@Hg)b%*V0N)yoTV9kwQU6TC)Y$g8p}q%^>eF
zuyr>*Jpxom^!_632uN32K)44|$_&xXyE)YCVZ942X@u79?yhzQtH+C}ap+dB!ps4L
zrgTT7?)2P1Xun#F{a$9~5eH5alW+GWgR4;Q-n5atydFMOy?N%Xapn`xKW!Zy>_uEr
zA$Wffap&#otEy6-KljNV2gLomqyTqJD_n)PhPEQ4NHIoFsTZUO%nhmUgXj7H6ZayF
zRkScNFz7_&wma(BQEX`7JkE3D(rk|o0(S=>xG_I~k@B&;0CZ-d0nwZ~0V9mdsHNJs
zckh~;2Oq3&{CGxBucfuMYWA!dNZA#?cM{tBlbZMM-xH#{Kv^WyiTkhzFOPf<4h`Ks
z<BDN9^po|TJ9h6@$Xf<NSasD)my=c_Y8i@lE4KsF&sU-fNTi?R6*>uO$D+f+^udyQ
ztp;G`O;;+oqpO~7TUds&0stD0U#vNR_F(kY6*p>?I-W>LO$ExOiEcKQk|%g<NP4G#
z_Tvhk@T~d<b+ffMU%4!b#=Zid)2U8Nc51inCZw>67CT;>lTj{H*T_iYtTSsUj9bD+
z*KYzhd#D?^`qCnXUnP#@-R8r{z&;cwrnrzs1sL5Q1cZXx(#}uZexE=h$GRZsmpQbB
z)-ODKcx+6{q;gZH?ib-J%v##om{4boJjh#x@r<qQG`PCJ;MZ{H(>EjWhGj!Vjw>@x
zmvajhmGSx~_n<Tub>x+SA^?j;2u%*00WJ;>Fxv^OsQZx2n8kc(US7YYOd+H#xw)9j
zl3!3DBC$&vSkv9RZxDLZr#pZ~y|`S5tw8_tCDr$D-%jL_vFKG-Hy&<+5XHc<r_3{~
zduzUbVWDWrs*z|;`~t*AVNil}CxH7Bgc2%#%sHN?r;CK*hK4L5Y`1@Z6i%fA=h<`T
zN?;0+M=BY*nwj;Xc(B`gF7A6zPXiK}`CtG-0Y_G75A?ecY^HEy3M<U}6c=C;(rmp+
z3(Wv7$d0HicfIGpH^;x0O?y@&8X$6Zn?D^~m#Hg`8D%gY3Z+V#n$6g-?bHi<I-vH8
zj+r;|o1PuRu0$_ipu;Tq#FfFY%<L?t(KY}c{#myvIw5~}co<NtC~{_`<8TMy@k1m5
zb`=YF?bGvpdI_-kb#-Itz0f$9x=K<K_mA$sDJ~czQYno|`TF&|#i;7>Zx4NZ82B%o
zLVYo?@6x49$PdV%jdfROB8Tj*$phDcB#cG<=N#DI?N1J}Z~rEk{QNngZ+^eJ6KDJV
z7DVP+cF5eaviRm{4Gj$q3=G;War1ka`<V}aL~8R?%C)xZISB6q0|PLCA{*_IEl=;0
z<;q64$1pmCMt;)M)5ER7KWNTy6{2w*H1~wEJ#yB#5+FP`BoFTJ0Q~|fo$0k}Jl3)e
z@Mb9NRW%#~a_9E=+aKs(efO-13c}<2JwFeP@PK95KS&RN4SN4^N*A1TcQ-ke%^$gG
zy&%B#{%Ow^3MYW=wQUo~@;{#4(|S~MJ6s!{A6M)dc}ocS09Ve<6?C*}z-hxwEm#3T
z#XVAPqnj;IE(12OzyAaVrnR=VqUeCMRY*V}FntO8<|lMZeAexN{rBFo3^uWFQyQ|(
zW8Xk1ITqj2+zfiigG@Yr4?dkapg$;Ye)x1)^)~zy*2Alq3|$B7|B<KXEfk#~5m;I(
zRpgC73RYX+g)1ZZcH3w0L}8GNP=tm>g2oGS{>)F0NtqR*EA)l#aR5SZlqh1JbvxVw
z%8RHko0^g-R|pFX(E@~W%;$~{MZ;2E&Zp=gS<p!bXkvW%@#N5aEW(Cw!^29CqTyRN
zqZL|6sGp-7bC!q#2Yk`@((hj=zpP#3SBcM}%a7c&c{A!!Y}W4BVVEuuQe?oG0T3)x
z&S7|P5RE3#PvAdrpm(q-5k=XNicr985F@Xr<kE!m2h)52`+wvooGJ_i$p5$Ir!4I4
z?UCIBeE=AO!f8jL?WMtCNXuiUQ&vP%ZJXg|IAdJSiKiISL~)EE>H-}U%(zj~WoP#s
znSZaL1*1>oDY{Fe0s}Sm^;42)Q{2MB!mx+XHUNeQDqs%k7d|e#eQIugg}N-9*H-ZA
z7w2;$oCnYbfCKoY)KyAQQ0qqu8XHhVh7S17efzjnByNxd8<JYiO5hzptm}L7B<`Rw
z?slLw4j~HfSY=#EfWQAv$#aKbSdJc@K^p+Rm(`+r?%WbU4D9ifva$r>PfC){<O=LU
zL8fiDJZ88M{e&x5u1pxB2L%sRL`;m|6CEH4=RCAcsTmk*T_T}y<pnnq9v0@mjQs#L
zFK09$cpGE)lG8sx>kkY3&SgKY3^*0Wf;0dCoS%XD##v(pAdBj-_x5@xpc&G>vP51a
z?<Oj${zs0F1DrrHJga5qh^X_tjg5`K?|einxHhz&u#Dz{Iz5ypr>)(J57_g<!!y>e
z$<GoquYhXnp{$2rz%El(epHQMQ44#_Q`}J1dgoAlPR<n09!Na^^(I)ecsMzo7Urx_
z4HDNvJ76g^HSvOYl7xhTUO_mnrMwO0^KC=l4}_zvs8nsu6^@X0;tw1J7WIG?o6oK7
z&};||h1tVfq6IliOI=r&r~<>r;Q7LIe+(YdXF^@PqKgPe;uxwg<G3%Ow{yh33}Oa<
z@WAh1>3&yV%tU#d6XHv+o)iPc9K==P@g@3(=xAub0{nrcQ`gYIh!19~3fo3ZLq+i^
z+0|@R92NDj%s#+8Sz3g6N5%MJ;i9}eW%oJ6IhzY7l65%~Zmgx-vu83$0beFT&PFc}
z52k!b%iYBDniBu#z~dwsEP}{L?k!4bYilF2&*qKTL0fk7ea9{-*FxzfBy9+bP|jlb
z%3eAcVhIe+j#gGj3OYZ2^b)s$3}*RD)ZHleyQL>2{J`uI(3@CHj>i`k-a=C}ViwRs
zE|oY#qg0VLFE^JZ|NVkn0I~+C`2P6uqw+~+5>(#$`fR~ofUCUI34G~|eKJ-<7%C6*
z9pP|kNM(ZO&ebnMO9h2s(<%Sy>1nhty8P@A92{1ogvqbWuLGEDzstg(cYc08%PxY~
zXm!zd&}9xD<zr$Z(<fowM;zY58sTxF2;^#c>5@A~f%BiE-@fgmp*bn3si#)}EVEKn
ziJ*O;%|Kx5?GKNNUR0a$K2TUuGBAj2C`UIJ!Uri45xuZmJKwjo1T9*wU%x(E>4r}*
zy}14L6ZzLMp?c&Sg7GH?{??;60nDMnYER+gf`WpPHF60F3CLlnlBFaipEB4twzLq0
zw_T;RtV}wrqwU%}f@_+_(}UmzWqwxK_7+ryJ@fU58=8AumKW^??Ytj8jJ_!VQ9y3)
z3n2~*b4R`%tYsAw3!i@s%N8Rg^HgKt1IH333eAKHad3ELdMzZzK(tbI8aWoFlskw-
zS>H)!_!;$ZrK>{x{LEW__VkopU4|ETAMV#wRV9;cVAuIw`jNq*$dZpp;*OySs>xa7
zU>JPhf{g`P1&&8kLR3@~&e_AqCph5<VgQ5_W<r(K)t4d1j}P717BXb4#$$W(WCg;o
z%a<<?Xd#TaiZR6MPmSLiTj}YwBU{5ROH58y$Vdd2pxbrX8Hroq^kErTAVi3C-Bup^
zvJo#>EeY3y({#I-`H$`z>76@wQtn5^m>3fH%PUB{cK`8|Ie_p1T{bCuUMD6cJzU%X
zDL0Yd7hx?B)(i~3k~yG;WB0+9$NkcwPLz(JlrY>FzcJPwRvCt_6Ui3y*=-HTNsmaL
zUn8BAn(8_?e#hWR=beL;#z+7+cgnYnbs<wl&v5ouiSw^1*re!o9Fg6jtEXpGD|H6B
zz}pl=tZhQC54s5MR60L6hnB6%%Nt(zC;Tq~aIdeJckRQW8$Pe1!K&2O(NWTCL2}e?
zL3J1(lbW2|lZwen4>@HGxF;ZnN84}h?cvcQCID|AHqJ!XRpLwpU3)}xb8^6!f4jR>
z`|HN1!NCwWVI4n!tmw;{a-%7Z$&?iK+QYmGIyxbf9lkxVe=vwUt~NrtN%*dC9AbO5
zT{m#EJA4nP6&8j?`u&1+E_YO*mfht+^bCwg4xgY~gbxE-ZO2x*Z89i~qD6~SmFCZZ
zvc^VrPFl#&x^3@_9&XEwy}m}AnVEFAPWjQYqIVUP8~jMmC+$1XfBkrMUB7Y~V&yC)
zOACt+%d4m>3%q}#ov4jvcxXtt^dpiC%X(VR3gep-hHi|olF$8cXg=<aDuz!+=dQoA
z4BMN5k%J-#bf_#+rUGeX6Qe!>RA1wAQl$O!XN@8djk7jFLCei4u<+x@endE$THsJ6
zNXbNM+(TcFcd=JQVoY0`Zc5zS@Q`M~78Df*KD#bX>sWMjqo~(2-;TyYDa$QX`(fh2
z$qTo(lcgh(Q12gQS`QhL^IcRxP=J!(LCJTB#>S_c2KHAF0Nm7+a>@%S6!q@iY`>;a
zpg3TDIP_UeOeQp>OCGO6<#XiEyNUB@oQk)Wmx{;FHQsB$cBqk5tJ)^Jb>l{pd+X-V
z%&1{xmG}O={0#T^i+523fI^yghW@o{*LwKjCuRrjmgb<@ZiDQroF{E{d|cd%&-EqF
zj<w|xyBEmJq5D_k5H7zp{sjx%{(!dRAJ$@6F6bH6N2y)D%-PT0hmWl=w!*pJEk-)+
zn3RJ6>rhmIOLm)T-p0T1Z1F8yv=@IDS4FfWCnx9V=s3^v3w=!xTw+t!Lfi5ae)BCN
zDr5>0{1g$VcsEBlDM*Msg9Oxc&;@ITOZD~gip|5dHrLkH*3_gU&}eq{0s;hxLNOwr
zqkp3dqHnUoVVfW=VZCS19!ORow5fj=+*M>pruS6EbDW3Z2S{+N%P1hn1_WpT{7Bk}
z4%izw-?c_Ofk1i?d7`j=yTcgq&eyN?5Wt4I8%)sj&(9$l3Ly3jEW4uS#d7qFu8ho@
zKgSwnwzbg-i6g~d_~`C_|2{Gyjjt|y44_wxBLS>HJL3fO{}8rII85Bjc#U-V{+4}J
zLd8giKZ8kf&_um_J;EIkl!H@KFU7L7e~r+u75Vnm+62J{^61|1lg!Q=PafFrtjTtA
zUd)XTb$nVAuduKV<ZmBJlkaWCj>Muozt#r^U4ko_0WI4%82qTHNX!ITSm1)x&i!lW
z#RP9nU1Vde&=?5q0}VPoNt-QrA(5$N=l=c6Xa#R5mBvcmM5qJKd{)-8XD5N!!2ky6
zenwfHaMmV-3c#hq2Z22q`(p!q7|#BK`}Ze)E`cY9a^-{he5AhcDmYF+h^9Ws;`gav
z#!IvD@XXH49Qtz-77=L$o^nj0l*{aBBr^AL6=JQSw60wDPZM??5Mw}QL$daWb78|l
zdt6^jk3c2)>0R+lQL1e`(XcIV_b^4!%}$`%8L=Ld<3|wWA0j@e0M!%0c;ilR5b}^x
zBU2u-eF)0kCWyu|)|^LC?e*Aim_%HcQfWa2RKc!euei9lV5O%mV=l)@@l0M}Wv-yQ
z+PbX1IQj^o&a$^3(9YyPxp1`b@ra_Jaoc$IZoQU{4%n}EO4f}axMcA$+EF<wF5f=8
z4k1$oHD#u&Ol$*^#m~=9ZggV=P&L5s2GawJq-daLaq;5#=CkoBd_qF34+6x5h3l=;
zchb-RsBmZK=|pgDt~KE+tD>lQ!;ZWXf0$jeuqd!j*6DQv1wr)@`1jsEJ{LB5!z*Bv
zi8DfLO1H0j<)6RaMewUN(ruqVlX}{bmg1C@A(TU>1SwV)QZXq>LeSU-wQ(fb&#0jK
zL{u>ohZ>rjK^Fg-#x5urBxy+|p?Rx$B5R-%zRsxi`rW$+*$lT_!qE*`%hw2Yg2cB}
z^Q%|6t$1o12JkSU6ijKt%zQ$vST<Jf<jKqc3waZO77;XA83~v_)ERSPy~PkJ?aW2x
zb8~a^_lGFm5M9cB^HyLjmEMF!9r2IQ@ZIqA!!(tQN{(VrbwydoyI_7hKjY9RPk6Uj
znwjD8*nWS%9b#0KbiY9k`2FV(>!Cx68rBFvonK!Dg#4lVjDu{fwQRM!xy9j#EZNwm
z2TO#kP*S9!(DGsSvB5E>*oO668Ae7&PpgC>V}CC>*_iFfHY5<|*DwGVbl{BqVg&YQ
z)@Mi@*!RruhHh(e-r6;5cx8jMwX~4b)P@OugisJ-v+Q+4T0_tWlt=n5gGlVw<d4l-
zUty!B26;fK?Nv$#Au}T(;V5{zWd8Pjon5<jjeXAdb)ikD+Y?=H()+zAAwE7AcY^Rg
z<rMM}XZK3P^uHo^A;OcVxde40klzLqm%uB=p}sl#C6xV5yKP)l?gtCq8x!N>j7uR9
z%zIlbEEfxouGf&4Kc=>W<1;AMdP$&Az@)2PKu}O5QOL?icC00JSAV2I9%}bDBM~qA
z$GC&N^?0=A@y7lnH^AdowC8dzZa3FhD?7_^57u7Tuo8|4LWft{Ufi(?-svHOC>Lhy
zg44)v-gU|FR(?8XCcI_MGe%Pr612M&3?^P*w4pF7!yhzL9`_m)@cuooM(4qROyc9v
zPV-AhfwQqr*avC$L$<b-7cV{u33<oO>`E7T{88zpri4e=wHl+>>0#S$JQ$yxk}{8;
zb`e8`wIyG1-e9A5T)cAf<XYQ7#AZt%zunwjxyGqvw+{#;aaAZzTLoKLTgUp?BEPQX
z1GI)nG$ExBL3ok3@tXk>+dEAI%x4*`*d+@g#q)8GZLeRC?pAy(>3vRLUvYni9708l
zOP4;w*jG0C?0hV511Yqdk>yHJk57pG9x{E<g1Di5P+I|!DeN2YoGBqMpRv40#?u20
zcbzZR7W6pjlmkAqzMFR)S;5F%l6M;Tk14(c61lnKcx})HSZ&$T4et=C(?YsV+*};t
zHG>E?P_z3paU+{d-OE?6*0$wu_xEpN<M8eRV{AxwM^i{PV8}-P)!EdmI1YNB7Im8A
z)jHN&Nbf1SnGOGDy7qv2%@JsX_oN&|du8<1N8mxoMVsEdd9-_cb!Do})PP#8@stKb
zx2OO-?N_Hyhhh8UpB2*5(vGb5j_CtH1wU-G3l)@S-Hzu9Zm<-c!tEhm{ID9O5Cpd0
zE~~q^NKVHh9(h}v|9qTRR4mj(;3Mz#buXdjoCUUjWhF56Cc8jMZxFHmnW187=4b*o
z14mz3C9xBXlbewdMmC<v+yEvbKCJmv_C&y9M{N1Mduwh^SgGR#AwMO;swK2Hy7JUG
zOWKtX_K%c1=~0OkWl+V<cZS3BWNsZA9Q@$$&+TW4iFQDcdcrV_?&O#SJbkI-tdEoY
zQ{G!QZtyvFzj(nbVz_<~h4yID9+ecW!s>k$dR)6`Q74Ew?t<ViST_j3;V}vuFj+Mf
zYB#-N-y@qye`s?4IF(abJX+sR8aM`6zurBd^8*HMyRL5zo+FgkPmcq8QTL^q?c`u*
zPx{q>jM{vc>Sel4z;(Y`Hd0dhz1(F;XTV9DZSK|dX>Wj}H*^>e-BfU6#8N67^tCoO
zgA(M;|B2Upr^gukC;VDT>j0`IB3?&+NPu?#$UyeQ2@@E!=IHW{jt=bI7iW^5KkuqH
z$$cv>4F>lae#xWoKu#t@yYFmrH}Q2jKzSU<gjJumx_WuYu~x(Z(q|6Wcsoi08(Ym0
zy#+(Q&OKQ&;WnpR1dq~m<m0XlR~W0PVl`M9G0EY@@#6YxmtQoEq#UO+c-rJwb-!V7
zbHJ4=>c5st6tN4waRfhkB2Pnx`<Y#tSpL#h<T%UyO-+{0;OjBK6U`!>Z{ObQ?AOb+
zq|x#nSKo&6m*(7)PFqckNM~lgdtL;jl(+kSmk+>%Ppz6>BYpAGrKZxmjSq-1uI5fo
zPj?2yij*R&8mTwXifV4z>z&oxC}W}4FMx_6C&-*NOFvm&UYI-LvY-Hc9Y|@53tlIP
zA8<0}B3+oYHLYvZch)lIVus?kdQCUBJk%W4U5SA$8+hY9glRS>omyyj?&O^)Z*r~f
z+Yk$A?8TTAGIFI=8SXY<QmE2`q@h`#7_f)**EneigVLHjH-^Ft%Z0_KN3YOJ>^k}e
zkZ;n)a)g*?5-Qo{3=H=Dmb0)(UdUw|U)yhT;ld1}sK>5o7FK!r;<dlo8tM8B+xD>z
z)Zsoz!8p%xcgf?4HrM!~>Hts1PxZ2}P@TFzK0>2f4oQ#0<o!%F`|bQ51mt?CO+;80
zIS(59Dl=2|VpL{Z?xkIV0xBr})YbV;iroPF>6VYnDg<1`P>zX>W&WK3@+ATe;S8qT
zG(bH942VDxd2q?6N&xf$1)T@dvAY5gMPJ<_`(t0<t&S#8a4s%h1h?k+i4*w92@s|b
z_x*}L#m3H#+zApYSr3*!wX`@mI@W~?@c`3=UqY0elb_$e(Ok%8<JyTXMMA}z|Gonj
zcLoGEkNJEYa-5Uk-^VtMWbppEv^0q49G@5x|DZWuu3Jfvy6U4fy}F)e-5ph#eraXp
zE6ClQeBljq5odqzR)5MYym_B!;84W|I|dBO22NTvdlA(eQ1z6TA_D@TjjuHyYn*A1
zD$n<B6#a}@c@l;@ayfGM+FifS`%Mj~LQ4O!$T8CyfJ8I04x(G(*}ZpKBXk^D8^D5|
zmmk%)Tv5FM(NWFhHE<u4x~q44c4nJzUHpgbb$k17RGyl%j~+P!rHX@&Pe7L!m3VAt
zY019$E-EfIb`qR_A9_Jgdb6~Q474lI-n5(s6SMYnoo7Y6!%GNJ1q3vaF_A#2z~xiM
z)EveSE+?YGQ1N0`={Ewt+Po{Oth+l(QsFANRwD}WS!y~u=7ymzTOtXSXvL2r=4bAv
z{Ks~G#<OR~zxs#{LhF{+UQ<QIH>`ip{{1S3CuA$2a_6z|4e=CTTAo~D+pe<|K@Wc*
zx>{WU^cvE3YpCj}Yxpa<Jt~YXe-@#(H(t{cuXA!q?cBLqABJ^n*LKcJIqjB@gOn=p
zh7WCr#d|cLLGQ_}aPu)m1%=;e3LDb;09dwt_&OYSeZZIo+Em)%DQ9L-6Hu^DLkaST
z|M`b92j-(sD<UkMwlg+9@MnlrFsL5ZJ7bjT$P|zIQ56QH4$TG!3Q^-gs)r5b3UPm&
zg<W&=ygOVuB1%dm&0nWJd)5o+YQM;hIOwSX7y>$2rgP#%zN&o>&trMiD!|q)uZ=wp
z9G48OrP~AG6pTyx#njkXYlwvRc%PUqU!nq?kUix_(8(>|;+KMcPRA_^BiG=F)}vyn
zvyczMRNXtC3@G0jR52df4+mlyC(fEO$=E;Z7IuL7iv!4M+axub#O`_9{Y0j!sbh%(
z@s4mATLuWc?B+3n$w_+p2Gl#2Om-uHY9o(s++-j+m%!}igJCtSPdih;ez{dT3qrn=
zlP1>)S-*WT6a6FeqT{PE-c?7o{uNdM+0cmpRP8_Jfv%{Vz%gC8hB1M~+eNp6N7J)+
zcyhew1TYWuC6k%=D5tMJp-VsCoHpONr>FKV!m{FGXTao1R!o}+O2!jQCQ%{bQC-&8
z)qT(ZBl73uh&Z-WQuO_w2?Aj-EgKCf8=7QjckP<&lR31+lbt=@1GlVPOO*jr4B*c^
zP<1WV)n{)P;Rk|1igf7{*!^>E!M>{vud83Vg0T7;=pD#CU{#WSHVa|`_vtHFxC|AQ
zk-7Cq=F1<vas4{ugWE@pyCJ*Pd@>MER`A9ns(Id}_a8pQ%qXj<gt&gC+h1$tnw37J
zkdGn-swxd~^Jzfj&>aG(RlW7~Zi*W=Hl}IEWp_gj99gVD&<m@srry0;a;1koA3i+4
z>jjX4(GXss;=K3AfLxefxB%rx#W2~b5zrcNO%crN<C*%M=`E^Zxi~pTfa+!atUO0&
z^=~dfS3TFxmxFy@zg|a=1^f0g&`<#m)G?GV;Gi{zY<ZrNLNDbaE+JEd6K**u?|$CZ
zorX_BzCzHVQASA#F_Y&)J~Wi>*R5OBN!>2FFuk(7h-@uUEz2-ur6e6|z{kR1G9gz!
zxB66hA^Hz~UH`tn+=^S#!QnR+Q_5*hRa;xuS6O5K(b^|`?)>74kSLGO$bh)l@fTgW
zD1K<Terj!191&Oj=R7DWK-~9?@_*SDx&E<y4n(x(tKtEwVgsaqYQHo2O2yFW$#YmX
zm}3?xdvC>qGet<l(4tt?)dfor_GqQ7N91}mYL(R9O-MLoBH4{HpZn)SQW6ryIO%8`
zQQto?Z+Wj<g=Vaa(uM)rHFL-l14^%<6oWfCnEkT;BNEoN>(({0X1zk0gOTy}Jn9OQ
z(721x+iW)Da1(>pC^YL?cgULaB)mPQl2OLcqx307i^HQp7<1P*4qcnEUlSw<$tipw
z>%oKD);takEkKJEl+A7qHiDj-i<&m1U2#HIQ#TIooOXkU{DFQ0^tF!Rcig*oxiNZr
zdHwd`sHveCGDrp2hJz!P&#_+48cj_%esF(piuPD5Y@PmI>iE01Ht>5Zkcvd-fmn(G
zaaX{QQEwezT6CO|t(88vdaIRod2~ib-TQZny|4IUw!Z2G+35G=bl8<h6pHGNZ20-J
zVXJ)U1&3e|%2~n%EXtqG9r$(Wu+)u;PC!+PPJ@As$4kC20>MTdHahVzU?up<U1EQh
z0jvy^t^<81My?8Z$l$I@s<I`jkA7N%Ce09_9rDssRb@O8cu&$sZS9)@jO5cK1D%tm
zHUbPoE(0W&Rqhh$87wMNisF@-uf2Ocgc0pZ;NBvNMNnJ*&v*Ci6Y@@<WDcMp;GnL)
zqikrbL2N55!OrIon9$PEBl%`?vjg(^=)^=;iRWNFIAI8U&*h^wd^QYiMm83rTb(}I
zzHfDiPBig<$bY1`94c;OC@N7OI_38yLPjrgJ@&%5Ed%iOrG-fq%lAG|B!GB1h*@nR
z>rlU+;w_Ad+AkqpgHkPBw$S(4_P=|ehSk5>{hD9arn%sv7=8hPxTRoL69r~4wU7}7
zXJqh5XfL6@g-92q@{g6339tnKD&u_felRa@L{%FCV<vz<kf;LlQ}B+VUc=-kCHK@$
z`KhY?$&>EYe-z*DHd1@P9n*m2(pM>=q?$VA;38~z<c{gPh`Y4$1<TC<VQT+~j!Y%+
zX>xCOerETEjLHB%0f+PV{hRc=ccc5x60Ig94@|=q)eRB#fZ#2LTq}siQyTF&3J+5S
z)q(e~GIe_$1wdcPXShb%iFh_TMWB<&I;MG?4?jbp3dQ8BnJ*Hxx2^fr^Jaa@G4$FU
z3vX|imkicTi5MsqnF=}>k4XRu6(^9VYVmq6A;bZ`gsSS`Kl!LLKvE4_y=3YWNd5>N
z$GS@m!zN!YL_R>t6U9DJ5fO++vO*r)*nf9Vwd3@yPjuJX&D$K>UODi8i>nT8yQdOs
z%Yf#8;e4Uv{V{QIGGRP!3)pqY!_ag)chwJ|1kor2-{QPfvvA~kt<kP=L3FD4F7BCz
znrNmH-_@=4=K4AC?l^)YwQtZa);=8<AFsA{jr#ZPQC9t=GTq{IAhv-<K1{@aijS%~
z^fM*6^f%$d&;YT7>>c|O&1q8}o@m}{2M&tm1m}tLjjp?*fWB2YmVq;X8qc6i3$zwx
znV}R$e*S<VZN49xjUOhTH#k<~=sW2cMtnazu0bd9^?1)2pX75;(#X0sZ3(#6QVE@o
z*lTfLBCYw-UY~jGpukqM#=3+j<gEzx4(cyTu?_w_LKcmtfGffAAY(FN<sg9%@D|px
zrltl;@PJAaaOw-RpLlsGearD<B!XK+gd4cS_xSTr{gXe4x~R2{&0Z+hg0ABjukFSt
zoY=yl3dnpg`S@YDEVEs$_4ODe!|y07)&1xVI5)Mnjlvi$;HKDq&B8*#$Dr>mj(c9$
z54m?074Gzds0F+U7122clGt=U=!DQc3o1+ObQpktwy_~$&FKhX=vi&9?4>5PdjzWR
z#mL1uj&MB{M_n2op-$e8D6RruMhcIRUc=%GROmi9zF94?{@K$p`fpsZb`0qGIC%f!
z;Lh-$ckWEGo@9z#4-XAC5R8iY8T5zXsj&8qMyd42Se0}l{KhDnHoPN|?L%k{Rv~b!
zJ-c_4A}aD<P;B7wM`&!o*LoY`rqZLC=ZKT}p&&AQ@;s<yT?ip{H8kEh93p_2gkHo8
z5{blagM906S{@e3Z?&L{+)hy|*r*mCaKV0i@g~RKm_P?L%E9N|^lrJi({N5lF_cbM
zmkzD@3=F6kPQ$9-(ENnd5Doui_#u>8*IB;CgG{C?jEU)iyO4WCZ)Rj9Z9t(hg9#R3
zZ+XzP&Kkt}o`~V;;D^nfKV5TG7)%$;fjaN)q227%gKlUeAiz^GN1%f+fT=D5e{MY2
zJXFtH`(Vhn=vj#{NB)Z!Njm<tBD+jec|eH;nZ^2oM|yHH(<EC{?v>BM5?>3Y(aJGP
zLZR-G2w#NJnL<aWf?fdje)MRw;|A%HZR`4HWPL7Vm2R2hG>UowL|R?y4|**ylp<<4
zS8S2v^=~H4`B0swP21F>W~i%%#W(F0_S>`-Lzsa8FXQN=gdShZ%wt%Bf)qh#&n~Mn
zk;rlA&s8I%Omuof5~cP}JLovzRtD=D6?X~By862KK+b-r{OQK`!YZ3FOul&{eFBZS
zi@?GMqF$p3tv)1Y`}pIV414$O8xGPdd3<dGN6-={H==9+rF!T+VxZ-#RVwObqmMw8
zls{<PZ4yut5Do0@b!dI#++{Js?oL@}WRvjf)hkS%?D2Yys7b>WOY!vAVZkRSdw1#&
zHeD3(!{fwABxw9g6J$(ZpU}Vv*1t^#Su9Nc9&T@!xpAY3t$_LwRZwm^7jWp!rn0cJ
zA3Sb-LVZtZ>`!C2^t@Oq6*y9FDPv2W|7bG_f*uQAs}%zB=3pE(wva`1eQb362KcqY
z_=x$zht@URwQ>EP<?ZEoZp%<kZXAIZLD&>;Xq_8_-o1wsO-?8&Dq3txL`I2D&CtwD
zxE7T2E)Sf*0m*(d;3SnE@fj3nnr=>)k!V@wSBfdGp>u(nXT<kkNVZV%D_N+xjE6O~
z>P18kU3`=SMqq7U7TcsfEw<G(;u5ZKjWn1~&=3x<HVX_0$VRI(o2!M{v6@@N`QYgH
zM{xN+6Wt0%bhBM>L$`?ygC?tY7r=4^S+a$DPTDev-W(Yi8!H8=(sHAD3f1X|#2+B|
zpz2~B+&>Z^;L>k|j&``qpbN7gsUb3#^?6R!mSwilF;>x*A+V+Rh=4#&UY^VR6q|`y
zzY0gcEQ*k~I{yS}7#T4Y)F0_IZ*;)3R27VpziLx4kDT`2MqK#ag*OD$@rXIp-VhSi
zdobd7NK*f}fCzT}4DKDLl6lz~0Nfk1tMTRM7#JAJsb7mUZObMWZ(qc-kBISmcab8G
z)c?WWn}=iBzHOr_l_-@+84I^DR3tJLArvx;D5;Q$M26_8Bx7co=aOVhWGqAGAq`X_
zB1A|c^S7U#_x-l@&)T+kz3W?R{jomV@7MD@;l8i?x{mW0_G3Tx1Dc&e*4qa`^Pupa
zx_o?e?^_V781-iHDj@M#biVoWB~G^U2aYp9UUTM_*4AZkC^|obW@#b$EDTyuAL-U&
zPv`OLPo=b>ff^dRZSek0!#x-HN>T@-)3fpXdrQlxj}X}c49xlmz(1jIvbwXKA!Pdy
zH%)SBsXGdaQPCE(IwTt*z?;V^@LSCY71~9+(yN2?Zr_m9!P_;%@U7myF5U9<>C^Mx
zv7<kKvOZm<W?^F^gJm#3%;p4Z8PJ`9X6u86&u9SrV%11)DMjH|S#Y3j<qZF-c>CK+
zdNcSXL4TjxG_@$A8Hv;Lcv^k*NPQ*{Npw~g$i|4(`=)>b!J(;P<TGv?|KUwW=Swf?
za`4^S1&?cHMc*Nb7-xR=?5oc7mn~K3z44sTaFJ;F?V+W$WzcyVOb)mN3mc&FvGwIl
zURVFtfmT5o<KONFTNg!X(9$o3fyB{#j`@$O*{Dt=jkg}z0z*HqSqDkWDwH8CNnRN}
zWxb;^n5vE``-P}88W7;#_r0`tCdzRoao>bq3jD~_@893t!~uHm-u=&l8xHd5YMc_D
zH55dpudE5`TdGkqQC@$1+-nsecH5YTVH13q@8N;|{t4kqZa!<@XA|km@C&9D9;c&e
zYs`OvEw(m~noZ9ALA2xOz(CD)$Lf1qEkwS-#Q|4~xoVrKEqUnB{z)`Uc@h2vtux=^
z17m=Eq$Kj6e_mmZz4P~#i_OvRF%~dKRfg^eHZup_kbnL1W#W;lAykf_<NN0<AwWV<
z+@3T9G81v0NYMZnOl^J`Y1dTIhR5iXxBDv$ghfVTX>%cp1H}FOw~AY-31WkEo}h)#
zk?s3leWmGDh`ay;w(Zz%0B}FH_dCdTn4TqV;@mRCtk{lp>Lb+D)DY_zlfXCYB9sv5
zYZ@ASb1lxJ4-98MydfGFOK26{gy7_%nT{vIaNm%iFjYTBM$iNd@}z6Z_W-xVEio4(
z+ts%{SimN{f|38p2r#Ws`lvvmV~>3{nrL{_lgoKJiy`T)7criT3~g4s2Sg2Z|Lg7R
z@2Y!FhW{3~z!ID6Y|X%fjPz4FZ(mT0I}E;pgYWK^+q1{tvMzXXih56FoHt_hv8!*x
zcHP5=f}Kvc@>0N3o{M=ia0y2W;6h!@B5{)?PQpEhngUwiya~=#xV@V1Gn#XIu&Nk^
z)SoXn$X_};AMBee%FG;v9!{C#FvFr-WN`34P`jNrp%T5^mMVf<E4FXn+2iN2^#kk4
zy76eGLfueE64#3Zvs#D`wQvx7Vgk0Ph_7<hKv{u|;1q_HVYQvAvq+GUk}d+1%2WIa
z0sJ$MG;Qu}&+wSJfR-g07Y58|2kQ7=5H*Gs4Uy?dMbew^8Q*=vEV}&DwwkW4Sl9B=
z5;Mjhm7(UAmUYh2BqLdec|C6gAmm)P#MbHR=`n5Da(GeQEH$`qXb7_p3R^IHpbTa1
z!oq_k$Bm3WqU_;$e#h5*aB$E-PcL_Ur&>0Mm04LD=B_HP78Sh?5YPDF&sM;<*EI9_
zbBqx@Zuh1lg6U|8IkZflP*?w)b9T*aA(CX&KPfL!XhK{_i9Qzl;>sa1dDtvMTTAO@
zxdhGT&5vHcF9sL(cf?kcH7V#qamu=4f0~_3M%vU!haC<y+Bc8K14(<%`wfaJWYS0n
z#xPZpk`fvg=J@sHsusFNB)<oy8jc|j|AsT(*%5?8iKOl~=qk__LN0tjTpScf@aHgq
zf+x{f9(#!z{tfro5JVZAJHl2#PfAg-3V9>3hZhcBS0(NO?{6}AzMo%IlnfQ4oFDG%
zYp?M1WMz%P(*@>+)Pu0kkojCN@nWFf&ESL9Si7ZEErwRrc_R~Jpw|xmE=q(tvlv1^
zXD<Oy2>h2+yN|i&ysN7ye0=$QOU9>8InW!PJ5vY~Hk|9?<t%~f5XMeDumYyP0*#7G
zpPmilh^%!|Ra1i;MzRJqbd36qW5PTR_K{I}lt&kH!MH<)Mq3!_u`G!pqB#q86x-lX
zv$3(&<pv=JHPqMlZe*6jHaZl548{twYkizYD1v+TfEyPgSsHgp>-IETdIh8pUOHDv
z3ki*2Y&xv`6K*e5j_k99cQP^_bsD6(wPWuYs(HT(jP38Q(Zc8e>MiWVhGXY)2kKvp
zp^VpBb6_%An_LDi-(OLHeF?K*QM@UHD!jl12U+WmX4(AG;=Jq6v?otW5a50=-<$-K
z?IQ+>=I1d^^zOq4WhEty5k!hzF58$p?+VhW5grz-Y3FB8JPrWQC?#5Oh^nhof%AY`
z0mX4VM^Noz)srNi!I7ZKz&VDh0rDP<{VN+AON)w*BXPp#Kc`v7Mbn3-kxkQpr*5g+
zTU1-1SLg^i1)&DPiO0O(Fs6{w(*A&T00*&8UzeArvZ={`Uzb`m{YyCej|aWWQS%Xw
zPFLg?f<NpUWZ#OM!}5x`$Ge&}#!y>&8AyB9#^yLjBOFNf#_-Ix4<Gc4BwSCu2b1}q
zjJjYS#2+9kXTs2uNah>;Ju&fBX8VorWGo7-qelK9l+#&3%A0ogx&<PQOUo#B91fYd
z$~cV_6Am}?RDqzBR8)2oLe=9s=0Ww0bmmx(c`|*CGSGiD+8zD5Ln$^OLhq<S92y!w
z1$Z}J?nZ-0-_$P_<OhQ`du1b^`1sVL%{d!?3Qt-~s}00xoU;~>u(>V<f?d@c_)GzO
zef4+xGWKO$XKp!i5h+lw(4MZX7d^Kg17gM8seTwn59qX$DK8}ph8XMpaP#I=B6aI;
zj)xWg)Qh?`|NC*7c5LuVI;a@5ajOVMMF56O=%r5(9^1~l(!xf0M-vs$KbjW7l~0P3
z)r@~G^BmcNnl7T(F!pyBu$|P8&VVfZ*^!rvDzf&UC38O3nUE6M8fc1Xu!@4rtSm?k
z+ya~ON(5np%O+n&A!ih#tD}Rbd?_y#y(E}WEx*>ulSdw&d6=NjA@9+Y<;x*%1Ulk`
znD$pIOSlQ$C^Qtn(%nl7vkh*v($+kSygXKqNM=~9LJ=obVFIlm%x#g1wXyNr9t@)u
zf(V{5fX{nAGRc(EqB5L5_75hF=1=}O|E|5g7-(Y2VG|QJ7v@-3G8|FOuc%3Vr2JBM
zcQ>RK{LL}^qnoe9^#;lqlB~xovd>+-$jr&vI`a;v9u*)8ZfJ|EzE+&dm+0}yHicK~
zyy)cMn@Z*S`YLVDxD@wNoj#mv_mj-b<KJ1mV=E$JXWuR^E=qlFT3X9X!-a!x`Y<m~
zJO1;I*w*N)<^xbByVFCzYD(^T7{cAFF{Ohhbn?Dq&B)<^?-}$sF;LrsoU)U?_M>z7
zb>H3CVI$_WfEZ<AdAo^v(jB{>#4o5*$sP0`c!h>c*cUFUE$74!>3|snSe6}o-XU2=
z5PlBXVuB`|Kaa-(xjLExjpE>7SkJv(G*F4b8W`%eYqwJW0xJp`pJt3iGiE6@HOEo-
z^lQz)nV`hM&?A5}8u5oH*WVpF3Ex7Cbpwc7d-o>3_L7p8R+sM=J9rS$hRj6%e3Rl0
z${BzK$mU}z2ZCRSJ{XO|o(*NDCCp=_X6mmKyU>JGkS77??Guf786w?|Dvo8xjv&!D
z7*?q}zyUu{|Kcb^6=v7K$c@1KBeo1cHE?*?%G^98GSV5nJEYv9exI<*1d}ox>y&UO
zPbw&^eo=|$K?}+O<lRJM46qOjD=V;W@Dt##<Z+@9JS-x**PR~poy4ZQz26<ytzR#8
zx2>THWhy|ceKsdpz<<Iv8>U4u?oB9gK%=GWf2Nm+;`~usS{y=ZxirS6UR#~Lfk9R_
zc@Ouxpxq6&_kt+u5#;OzxVJ&i<Kd%6>P;>uPz}f{zfQfYZ*1JSR1OCzet|VigNFgu
z*3!`tDWim}6n!IE%pF4hiAZuF?Ldzb@gD$P3nswg<28O-+SzeJcs*w#ySR7`A`Hjq
zxu0faz(6O_YMh=%TM@p@6#{mc*a9L291IE^Dz2T8KQSUYgmTuU!h#IMQ7Ay^_DV=R
zRGvHq>h@K|dQTQ4HJ`zyn$1LcaoNTO5_#6=&T%vNfXKEDxk~qmncu(t0s_v92C5A9
z_8$4}@i;yGinDWq#O3I;G|GXmr%=R-RmdcuIE6FEv;U0n|K?3DGGg3O_*&5XF||g>
z?CGVn7^-yl<KnNzA*d^@?>Ck~IgjX2c&Q`M-3OvwY9d$KF*^p+YdJ5q<9llYiu7sF
zlVJxM(c<+8NiSdS!q_N3|Gjb>hr@@dq~`wliD(FlVE^Z>tgOiBu-@Hs^Jd65sdICu
z;1)r{51th>u3aV0_Zxn^ZG)}syYlt0vW>W;<bwuTN(HQL2^te<F8%q0#bh%;fzVs5
zq}!15qNs=s+dE84>Xg;hM-TvjTI{!2*YWo4xoox+T7t*3ur$!uWYDKM+F|ws_1e(z
zFxfh_XbO-pDf{;u8U_Z2p^J*Qc7Z^3o>dmH3L_9(%$H?UQZQG9?4D-xOYyyXQHXDm
zG(r3Pu&c?+@H0*h3KPmNDTlzgB$|!FXVW|WyhOy56gv5pt{f}AdL=N!LA`$c+i^V_
ze`TT~f{OO=9j{$9moAZgU3AZ0AX!8!ygj>VMaYjr%NekQbtTH7;^N|+-@Xt1IKUDo
zbLVCAj`BA{=D9^^KgrG*-OSGyb8}dy)SE(hxBvn0o<FV`VM1vZ&BK;&g6?~Dfn!#0
zq+-ON21sDNw5;Yu5mO^=1Z#`#J-KNg3kWMf`mT!Tzp?@lwEA1LSG^2NT$vN5nN>jt
zwP_BDa!^SL1<XhmxZh|I!fb?9cQ)p}om%V<L74@73sm;n#-OA5xp-Lc!+2BF1gFIv
z0b=$V6KqCQRP@nZ-d_RA_!s87x6K@h!iWE+CBE(G7-Wq|Pk%E6#s!Yyc3WkfX2SO@
z51p`_LLPYRcnDTY_aIB7c}Dh)&P^mlN3+c9Y5%nZY3evVU1EM=A#-<{Df#@rU>el%
z0)s_1+6X#jA5^0DpGTKX>Y^U*r3M&MX9j7vB?34cA8+j#(fMmQj&Z?Z=_dd&fR$A;
z36^?sD$U%Tni|#-FktCuZwCy;>bCA_m<2~bi>bpoXq4s^9J#O~0MsJ#$E_f}07cBI
zjCGeugI?RBkpX#2VE!1eIlKu?1=`J<IgtturU0eE>|}YE`boGY&E4GeIP=4o<R^3U
z^GlE+qY*J6pPU1D4!x^M;j>^KUx5N$%d<)M*tj?{yYfOI@fgn>W`VKa7{Jt8Yb_le
zR1u0;7CK>%ZWLj{;s}#fYIEP~N8v}#m?VJba!jl)%hWq@Z0MaI+7n3r&}sl=N{QWx
z(@y4Ukh5Ds?lZGF3<C?Fa3OHq;6n<Q?!Afo1XKj#syNLT?Ce^N?*10ki02sxc^|7*
zK=rVK^P~0!1;Gl*02H#X+1qc;(&`oFq3HvtfaxhTu(tA_v#}wo$FS+MMX7g3X8@vn
zh<tD{>UpGO1l2Ch`6njkGYS)o2MDr_+0cvzy~RM1?$sf4MU3oy#smc<E8#WPmOnrh
zjfaKfB5&V!M|?aIv&P8AKp&IZcG%e17<1aeQJIuzY*9f&;>Y}0ohJ*+-ZL)|?j#=_
z0>_WASa^8ui1{779T?bx_XGuB<i@wz2hq`L$DOl4+JsQ-N~6iXQB_fqOdumw8F0S@
z1H_0RL~$r5?7SJaY`KCgqb2PTZhamm9TDE>9x>W|$oz-~CRvKMAZ>Nw0+@aiNXPGb
zgdZcCy`}c}fFWn^6x_r<uwvl5`$sJ=TyR1K4VI{?$<TQK{V0pceC+pg*@sc&%L2{9
zO)4uZ^IG!(D^hDPxdTmI<T<T{lTRx(n~@cv&jJSXyG?=;!N;)dnt_GI6}wP+<&+hl
zO{5HU6Z&!jR+S*rM=1Rr{-&)BBGMQd??oA#TJcX3KfLV)8|>at<}O5Mhl)SLZPhsR
zv2}Hi(l~{nU2Tds^s!*ms(0_gPAQ#^{<Rh049WvHTsKe?tk8{u+=8WLC(a%)3N<OF
zPhDLLAVs5xg4zQvmLK)gfdd#H&Bul-q`>w2VLQ8xDHCj)Hf}@@0z+u|kPyJYSpWC$
z5#iy%!&)@Ag5;*(0YS!Xhk!0f`S?P1j^KbVBs-w^z%)>1LXMNxj9PGdZNmQZz8J<u
zl8zd-2lzQqW~_36$=?SVlUDJW^?Tn8mA=q|{Cvd6fP44;@*8tPW#nI703wS}>Jple
zU}tUp2}BIsF0e$7TsQO0&%b)*$`z<3!t8;|KsEp>ehB$k1!s%VN2EH?9GuviELTW<
zr0zvI|9MQ&`s*D=61Qhd_Pe)lQ%uEuCj?#4e8zsw!B{(yeK-61`yKI9v9YmWN`YDo
zrX>I;T?2!Im?Q(A2E-18(E(N}b&`*~_RVwGIW&|&>*KZIR?hS1OUNxf(4$IEPse1(
zq@R|F30O!yXma8ULl02!P<5~9OzA$#^#*tF&J4o^YwI~u$93lH#=I(uN(A_N$Yx|_
zic3iFdhP>|pK3Jmy!TZ4)2GDbFc^s%cob<Sc$k0zkca>Q!DELK7FouQ+YqLH8Y7Oi
zwY6drS&%M$`SNhQvNpy%Nec-N^3I(*JlK2-5vPcVkQ1QWA_a+*XAZ?I>-;yqeYf4u
zpQm8i;^W5xpMO5lR7Sk0x8Z|8T5oSJ-Vd{E0PXO8xG_+T0H!y3KGqAT7bI#s%ot)t
zHck?Ti2+;KbYy*jlten~XZ-v-Y5pyDPJwWNJ;6k{f|(grbP3Gm2Xly6zgoz$IM%?+
zPx^a8?E#Mvv$hC=K_3g#&AtJo3-uIx&Ho0H>Ia}H2nPm`1bCba!15@YxLq{+UloVl
z-OL$}NFNcgot1SS9tn;pG|n3V2~({g-vQdC=a4a!H2i?gi~FAjREs8hd99rtoA7#u
z$qL+>TsT6Ybx>j6#m5IUB?Z0%N&h0+6ptTa(naY5cXQ~UU%x=WZUQ3e1|IU{0q|T4
zuricr)GknCR3v~?z8gC#arkh@EbIbg8bN{qv<K*PdibsPxO0fdhwbr={Icxqamew+
zE-bOjxf`CLJ`6CeQ{Hy<nro<^@cCvV78%Rw(jGpXMu2rkl*O<g@HO~`vmlQ`(ge;G
zsXM~W{M4`_uPa0Pon16251e9VHKN}=x^;4m*#ME;fYLBn(a^BQXM~;L>EIcH)r&4P
z68PTl-!Fh3v@fec?m!F*<*2w;L>_P#Q0c(iVFt|wTOsgnXthGqYegIZ+=)JuhedBr
zrGx3@`~3Md4l>B0?g%r4^!?}&EDKeF1p0fpfk^%UEru@D^jDz-lYEGwTi^2U+mCEP
zvfkcW28<6Q89N?lb)|{~9G`QfBpyml;9JT)cfV8Z*fA}q6gg=SZqJp{A%)R*N?G#$
zQ8)oV>%aXI%+SL;^mVO9&ICH+0%??tpmzg0le4jR<#zg~iZs}2tikM+>^J3;2!y~y
z!oOs-(z~r50gjb2s=T=Qy&d78veZGY`Q`OKM3u`67j`|SyO~G+pjFUjzImU@?MNvO
z4SyOXK}>nF4c<z~nkOvT-BB2x4_mh6<-MzU{Y5hoyitFJz28CgmkqJ11%0J*3L55d
zyH>O4Vs3V^Wz&%@uxeJZ3;ubZsvgb>^T^#DJD~{1l@!!{Zm7fWdNDUcdg<ofyL%74
zR5^oSgS(buhi_lk*N@2i^m``~r0Y{dTW>THoPNK8xZHcY-AA?%Nr1&3#5D)rkfrJ_
z<d>)bri95%p}`!KYj*CzRBA3wGp;@S{exB=^lxR)#e?~VqZi)EA@Sa>SLIEa6%9NJ
zYLP@csL;htXcN`fwl;8Y|4QOm+>Y1B)X^mN7o0SAlRF5zNd3UGi^d$-DOnI{?$)VW
z;e`>Lsp=v0^Do_b@2~d?u-@_ILKvFAF!ZF70K)m3#UM-|X@p!+!tU42-%z&EW{Y+G
z#V@1dVMPP%2B%8;hB{zLIIAa_kaDA<8h)Fgwt2!TCqEw(x&uhi9t)GGB_tGjq245j
zbMesIt}b^_Hq~r4^5Me(RI4`MH(Q9>4mX0SU}Lr*xCj{Bp7N?GpC6M3(}E(gfl7`O
z&^nq-8sJ10P@n=xdX6^j7ri{JIiMs1=MmnyLxa<RPA}i>?H2UhZ7*F?@2R|(K`?1;
z><F27^O8oNf)Tf5t|S;JxHG!8EB68JQI`%m!AH75j}Sm4j@YqBWVYmVzy$!rdq2<k
zR}@BhZPBs<vH~@~A3p%V7&!Soz>TQJXt)&(@{yWOvU-K-4m6r?S^)FXcG||HwYYs9
zIvLc|{3T5IX27eqw!M4y9RK)qOY*-KMIM{G+!p0!&;Q5_wj4z+5dLL*kUl}<8XGs1
zV&O{SM#)V#9FfRi>j@g~TDIr#YmlbozO{F5`SyDjd1b&j4YsgG_wp+&b3=xwPI0v|
zto%&f;S(RpoBZdO*6IKMisCKFPmhW1%@DjyHL>N`ORZGrM%wf<tJZF}xq0lQfG<Dg
z=FlhB>AWZDMbpK7r?#?YB_?D&F*m9yaMsm5qqHh*hQ<2$tVjRn*$WrHYw*^mQnmgT
zlA7td+7qQZ7CYa4wUwlMt5qSdiLb-Etp1fEq~QtpSE2Bq)PIEwcLj<6t58fG@~_O{
z&BN{gN)c{3#`3R1;eYEzpwD0o%)Pxm_3NrLD*M3{DjL8}?yp#ut=001iYnX$d99K+
z{zYlS+aaVSs+ZZQ4l5ZUyu&>U9U^+w>8bwwgfC6?U%4h-;aATw5xU2uD5pzd6;*ma
zRC;h=OGNW5FtrI9L`M2`D1?dwk4sjhy6F(kA7Bc}UD@f*lz)F7-e{X%JLUj774Vi$
zyuAOeCHt?u_y1cr^?%8e`QI<{|B%W5pWfI+YbKne!{>7W1|X18(sQ7?n;wERkT68k
zJz(*hWhP#}qh3$-Qll4P(g?d#ZqcIR4_LT{uuo)QZ_SZKQ)t>+-&(U;7y2{lLp!jP
z!3&Xl3`?_to<Fv}0^M133N*-8tWoN*vYBqw1u|Zn$|ox69>qGS(XvvTN4;B1^>ta?
zu6v=5`~LB>w}D-pIJ@%m7+j8yh{*Fd0e%b(hscZ%H+PYbv}}iR^09RWDvw%rXOoOm
zYpBi)w&9#W)*>yVV4H^IZVZ3xJzaH#xRj+H<m=c{tlj<&L~V2~<3k=&QFTV8UKt`1
zZyC|68G_(w{(2G$D!3tP(Q3W#(KM0{)zOfmP`(M1ZLs*Y_QExcs&GJGD^htDzkuO^
zd?T_Px6w%sw`&9~m*LM+Snq(YsCo~*77zT}&1Hu^A3h9$=Z6JsWIu$GFhjs+J=M)P
zO|`qW58kY82!N}5>G>GT;-M9;_VFHBs&(w^ooyQFSm<JPs`K1^*+9v@d%87iczEF^
zw>%9~oq!ZIAP@mPxU8Jt^-WD|l2^`HovMSn8x}`r^sQnD#izOFG@4s00ra%G<o=F*
zl@x?HpKynvyS1E2GQ<NPZLF)498L{YE1d_;UuVZgeBLxr79y?>y#{#!v3364>oq*P
z$eXJ8sLZfA)a}MoP`|tv57HftTTQvy{*a&25Po~N?msS01;vjsfc9JLvZv{&4u`09
zU7PyL9VSL@ge3Z1cQ@|T?kd9~Qtex1cmN03uf02o0joV@rUctTerHs@hB}yBd^Yv%
z+kuCEq?b*9da~!%Dm-13KuBV8`o;!+sqlLeBlrcHpbNI&2tzl@3NS)MpVKaY+Zs{>
zVf$&<(QuI~PdB`M8)>XlyDDG!>6I&|mz`=e7uM=Q?oHanr)3VCkQe*M_kd(NxOK?2
zL`u%!;sls3*|Xo5SMVaHLGuiR4o=(wRedMbHB^T;K%#-GJ{k}nmc*Dxty@z~!Mg<=
z3#tEwX%m=VQs!-P(WHZkF<XjGK03#+F42+eHx0?#`H&Y7emW%b7JWUkCxreNd6=sI
zKA_}dctAYuurTiXGI@2@HK+U`5ypiXrDt_XtbmrrP^QmTuQ^KY6~97oiU;D8GoE*Z
z=E`RTfc*y5Wc4)-EEK{8P5P|AM{bAK9kkCdSILvU7aULgPNXr0HAdZi*^199GpDaa
zS*3JbD$1U1x>v|26^L_a5ZIu6fy5{PNCP`JZy`q@(Q*UB0A?9SP{q2eTSN7-<q^&z
z8ju@4xi$zqLctVyQReW<2}q8+-sgi61^*H&_ySn9@Zfda-&vj=#l62&(pPSW(vyD=
zK%cDxc~iC6E0r4%9({#-{=@z!@2{sJR((vZ%h3h350X&brT?S-+qmE)W3^9%Jjbp~
zHB^L3wJ+(d0H1<n=AncwN&}A)6VCwbhF1#qBF!j3U`k%p@NO5t&PQG6BWTlAl<V^*
zDf7#<j?)`B4rJ-aC*_D?((ru<MO_flKm806VSF*ML#IeOAIxKYN1B}9pFtD7_4I#-
z=KoIm_HP)XI-Cc`_#Sv(=OoZ5;9l8NKRI3cgWqQ;nNgsu3W1Z3G&?q`o9tmEH4MtU
zO@D~AFpn5lEKYxhMnRYr6%5{Eyllq2ah&W&%qmKr`icEX*U%ZhFSr1><J<SKRaBiy
z<Y-Z;KGuKw3Axy3Kpcs*WdF|7!tPV?yRj-*YVSev4q?E&<c`qGGSZSZX%}43O$Tlq
zAu67<x;?W<;@mpVdWVzare#ZF%*|x7c?VrkC4I47Sc4FEpIoTcGZbtWEi4#~E`vFP
zD8laz+fVFTxd<h7hX*bwE<p-5`~X`%?6Q*X`nCbrAw+4T7h%7PL4zlYA^{1bOtixR
zvWoV7a<=L&Srdjjh<IBVXjtK996=S59P=w^U0A@X1YhA%`~rNc`jxOX2<2E}Y?NzW
zg9!_=>lSeyJOyizS)AcP0qz;xTE~6BvhC}`7+Y*QN<fGq-259@fB-pwZm?xAXy6KU
zvWC~ViS<l}eSm;JN?yLcii&?TKuP>i;Lj~BP&LKE=)lhgrFaj&Yxm%KyMgn7IB+Dc
z73d>CPEc5ZbMChF!tIwnGLKk)y1-iRR?AOy(K532MvMO$Qj*T6h^C_NrD>=RryV?7
zd58LoQVO}tM@cG9^5IFso1b{K9z#e0Uvx8FPG{&poof<nHF5h{0OhqK*ASrqde7m<
z4B&ERBk1CsyIn{oA|v0V$`?0<In_v;EwVVsvZM~#6<j3s3&5{{&cpt}!3tz2xyG9h
z4J9gWmKCp`CME?rV2c$!v3R#J)x3`i0HR!R0Z$OkESfP(S_Tb9%CWa=sB+#CPervF
z_Iwc(#@qqys2K+Kt&r4)J&nxHwLI%n3F-yrLVYb+Xx1-u(590-AXQ#()1fxS%}{N;
zF!>X&<VS)J)oS8{OMX5#wwVR*h0gQuk0;sRTjE%acMAy#z2Xr*3oP05miP=XD>+o-
z@OVp@#8p4kzw8WGcEGkv(sBeq4H1dNEpbKmMzO?MI_*tNbPNAj;w13c<=KSg4a<kH
zY@+iJ4wMWOakd@wcz03-J1phK9%8nXaF%g2<wtX>J^`u3(fMp!f==A7**dsSmrquY
zYQ8+f6UIny87vfM59NyvggNkg<S+VJ*HfM82stx*@>n8zH%yDQ$Pqq%!md5En9r5m
z!p<^%ivk$`>!16U7RNm`yO!Ynw>#IOZ(hx=<k5VfgZhsKDWK9DfcM#7`VK;8+|KX!
zxC84zaV%+sN&3yUVCCTLsJ&+uv<&hS6Nyd))vj_Z;yW<rFVws9H0|Pi&+-!LF!Lu+
zM6Q4j+3-moE0(@RAfx>Q>WjZ{#Yk2O8p{d%eLO6lV=o5f7AC4XCz0-0fZD4nft4a6
zO5~*7OHm6Kd=&3$Z8M9DHpQ}#^vd*V*CcceaO&3=n4zl$j+*>)88Do@m`(4MsGFL0
z17S<xAEF_1%f=kfc&+<RAObh6xDZK_@$@W?Ip+?L=QJ7}3wss0chKIYZL~j0D;*h<
zy>~GL7l>kwWp@9dffX`Kd*i(>SR-n`QdZknpEeO^+&ke8&>CiEW#a;*d$Z)%n?*m7
z!HN>mvc_#k3(RYLsg3<?ps47H+~Y_I1|g@3$>iNYGRyv65OzV8*pAdvW*|(*yb+Uo
z0LZ!^A=!BX7?VujJ#O<xJk=ZT^$cnv$P&b{q-}h8FpBsymr^l%-Y3C9>-8_cFH{El
z>LoNQXY3PreON`qZfuHHB@p0z!xI5)Qj_;R7tR54*PR8^<lBe)Ze2Mr9Uhl1<YEAg
z4`$@x(33M=KLvk+Pb^kH4JUN%0m=rdd)viR)Hgt7DicR4cK~rcz}OQ(LA~u%-_tZ3
zPk6Tzc^JTS5+G^UOHZpe$u;|G*+MJVed8S<dJ-_q&Pia(m{QuEj_pUT6qhzI-kA%%
zR_ryB<V{oykr48R=bBXO#R}Oj(2Y`Pv5b)jNEZDe%fpKXP}bXFh9y+SvWudrnAEnG
z*+dJ#K_^yJ_yb_rTTdKk+KS^}rg(`JGiIDjd5yZnyz%x!Ba9#Sg8;#)*9kp0X4C7t
z$10QT*B<Cb`Uq~Y@>A~U_;@+YXoVKzz(!J(;R%;#V8{^Do4GIC+S0Q0J;u(rs|;?)
zo5R)8X4dLaoCMrvdL`NJ_QBZh%jLo^<|Fc0k3&|;pjiojggGNZWcgrp5ur=|1@l%~
zF9mb~{sIKX-?Nlhfhvtd94%?zhx!Fsdpp{)@(aD2?ai<O`Hn)8)~c{$;Oh#`-Djq{
z`OXc_@SD>z=4KL_#vmyP?BDmZ0{%1~=|LnB7>?`4eaj6<KyD{sMYM6#ZWsd;>B5r1
zphQJWZ7e7Ix{T~}Bdf%ov4?QPJZ~#89At`fSrnro`{!L}B~GR-`dIZU56}~FA22X=
z8Nk6hR_+1w!Pn%U{~6(zFTT9F%AQZd2&phUOoRnD&|CMeNjyir$J&&PE9f&I_CP=H
z-2Rp2SMbKFcTWzL{%(npx_TibT(oSe9Su>YC*(-1T``{J4W~?|@P7RK{KxWw4|ZSY
z#yNcMj^&|0FaU2e3;%<@MLr%vbu;vGzoU3t?DG>1*_hG>A_&lEDTuySzX#<+1S6^I
zIj$!f2K-MakK36J=Sj-ht`n*)L)w-|OUYi+t`kSB<}e_ijJ6GQZhT@gL2PpyPKzX!
z3|1oX&+e@x%ZX5Rv)Yx~12(o7I-Vehmg1PI#uD*X>8F$kWTi_Bbb6^jRMWf>*t~Jq
z0SPFybgl$bjHTfv4vZ~)L)UNzG{Bh*OzbD?`8XZhjv&i#Y;K0X8b+Irw1hz#VAMyC
z{4H{jEhY1)QW&c7IK*tbirnBXQM*baL=eIi>>o4d(3xC@_$R1VPDS6z+8^Za>0bt$
zqICMz*HN9})2GZ{6x*$hiWBk?KYarM(!%>ZgIQd`&Y*Rd50Ew7`1)Ur29igLum5|H
z;{Se;{|ms?O0q8<qew4hfpPVfDk?|)`t-+(n51Pk9;>9Lx;OItvn}R%1jbC;hf+It
zd$T!w3pl9ni0b#HNG}!=(y67C>|I4={soJ%D92J_NC?2!e+v%&|MuCLH!~~$Cl}zq
zYTLWkKGaMw9gtu8b>CChQBt}B^ap)`2AoSjn~NBvI_XB#F9x>AYnh{Uh1R~$LZNJb
zkRtLcp+i|({EQG0yJ&F+ZVf~Qb*1%ddG>>igM6iKxeElIqPjBlJWVX*`{B8Sz5ga9
zW1i~Cgyv$1y2#@pLnozTwwx*#EMiYCrbW<mP+G5Nvyv+azJ?bWgX^pd0Xu?pos;`^
z58ztmkOIO*S;A){LW4hNg{uTflwNu?XVsc@pb~RK0>GTx1YQ~!@b`@hXERp1z~!LK
z23~@_<@Nup=a4+}^5(ZdX<O8C*_||W1K+=Uz-qAMiMSJ@IgL^r$RDZodXQohN&w(G
zlo8k>1xN%b4lDi+tj{V60^D^hA8W;C#>ox4>d2dRyNxSFnB4gDeqj2!)(70Hp6lNN
zNLhqL_H>>F5k`Z@`h)@q>~1WOy^6>}qO4MF(VEyPTHIzU{Eu7>QYfqz0j0^{u0j%$
zmI7pKfEmE4l>CvN(%#-+^-Kr;8FZ*WaKm@iIb4U5VAS(B*U<AP07b+@BeeqD_|~O{
zy5{D~=-}aQVJweaGw0Q&eJuf6EX8cnME9@^$p<D&dkxHPkWK>H(%5R359MR*oNTqP
zWj}c3>r2_0u6%RuuHA=}yae?kiLNR3(v2VzvWdJvg?3+htcdKd(4=~Nl8Vv}UQUtc
zNzcDkoDFkrC>am`y+Zyp;v{-z?+to455SFQ4P9~afGQ>b<kS+@Tr(C1XP3_jraPb<
z>f;zdWukHA9%hxQ#~gmuFh@?D4}FHWD+O-sv#fyIBnFG{^}Pb!1@MM5^@ZR3T);zH
z#4~8FuYa7r3zFt|0tqhPqqw9;@VScnL)3{;c1_zMt4ovs(e3s$h!~4s#yyygULc%r
z_z`&<rDb{$i-b)P!c9>)SzVoIu!xk1S5mD@CqXpMfvnL$grh8pV*a)R2S!M+vca=m
zC=3hEt2P}Zn74g8pvY2`7ZP^ecsp`$iZ;QL_PsJid;RJ=$`ml|n7@a?zEfLZK)6Q0
z@?-V%(fjJ9nhK6k#o|vpcqQjP=ZCOG6q|#n3L3;Dv_%%~XI?aJag=g0M@5e5LK~Rs
zwpaxil`DSgDQNP^w=x&vk`>N>8Zg9)I=n-l#C4S3e4W3rh5B9w84!;g^NJua__uPf
zo_TpX8&gGc6p3zK)#gf4tTD0EOUL7QgJ~m`*PoseNQy7}&pV+m2>42ynOWdUtLzLL
zKov5M+-FTb6Y^Ej(uqoxy^|aE*x=~^WAMbuLcYIE7i7Px^O2`NAdw&di?L0xewK*~
zZf%0=kH=_vA2+!Qbis4B(JUlt!|ryZ^J2JL4AlB~YWvQ;J{my)1}HyLECuz5pa*dO
zqY<&VFQk&S|NP>6^reR7%^}p7Q;H4`2R4#U`Vd)CI8RVmmqFWbV4Gnoc}P<=%w|Y8
zF2Hv{<w?WjDx@_cO<k=<(=M2kCIwj#f9oZ<?a(s5Sr<Y~97`Pqj_8!+fy?GR^5HTv
zSyaE5`l2yiQSl?=36AtZ<YNvSgc;|?RfY+WVq1G63Ehd3y-UQYI_}ZsIUN}Hp44Fg
zdKPC}Juy>w<CZlv$Uv@`Shb!l=qrcn&X441H5516<E};S*$@;S{vLj!E~I3vSP1$V
z)F<>9tsHO0y~59LxfV0^a?R%BTAsvHhYBzsTw`bnZ*Z{disA;RRDW!8l||O2)Hmgr
zip**vGrvXfZBQ~+9rD3}7y={-`6G|BjXWyCllKOusl3^pI>DMYRsM<XtrQFg!T$(-
z9J?56ul$b67x+p^@?0XYm%=UwbqiGx%1~mM-Fiqmki=L8eEr&qJTb-wiwC0*W7iT&
z`gr)C{Q-h%1M<U6(TA>CgD#0tD!c_$A4f8tWC{Kn%MF$0TXMi#p4ZJzL6%98o6|T0
z4-e6>zwpOwBE#Kf^KcC^xWn*Dmq8TA#lq3{!J;bfpC!8EQJ>`xuiRibOYr*sfDgoD
z<O#`fEG7Jl>byZb<Cyf(N9kHd&Qip9+SNtFZF+!MCm_WPrXZNS03xrm)j$dmc;^6o
zndr$L_W+gyEa-*QpK?QxA`rqOE)g|i3oj-cg#1c6pccb-B3gI*uj)^@6cnR4q57F+
zWMZgMq$JZ{!50s09>EVJevgmreR+){F=7`re4H_#FfH&Xmcb|wM;k8^aVRU~0jemV
zY250lT03B#Aq5h!e|RF@Y9*^LaKJUEC1m6=33j!05r|p?e=2|Ya;kwPNS&y`L`nw6
zArKuO9<JA=b14;F;-j+31eXAYmhvv?Cj)q5j$QV(UVLsi5W_e;3+T#_PY+5QFe6(q
z)!;y|S4sx+5y0ZS2b)e~!UJ8ThN+TuPWurhp1_i~OI1rZz})lm^6+-1%Da^wuXvg-
z1K2WUE6|}Qw>b3{7#4%kWJkqPL6=-@JJ74n0e?<;IWVX5EO2xvrq_i5VURLiurK*p
zR#BgC=OdjNEY2pU5Y7yhN(_!X?y0$RDI9hPCx<BujL~iM9~7Acj{<~)aZmD?fIj)f
zi7mpfT}QZX++C@_Y;xsDSC%_!k_2=Lu+Z&46mzf$hZJFe-i(0@nini1xX59(Xxhff
zXKyw91YN%V1LcYR0-VGLC99dAQ|~5-AMmqq!Cl&9RNa1XZ2V*XAeLeR8jHYk4|pFw
zr>#$M3*!)B1{l84UumcFQovSBBKl<WkXKmvGKMugm*;z!K_kJ>0fQjg(3)tZz+Qak
z0~TtLenFf_^n7eE9W(Q%TxD;)L^_e8U9l9T0mJx?Cpv!X^f0usi2vw8tD({i%ey7f
zykd}#qAp>9Uul<Lo^2%;^L*hJoycFrc3>yvNF7kF7o!S1$praE_CJmaD-I?pZ`=Vy
zRk9oi9ye`+1IoffxDnjy`k99NfY`M^Gs@DII4E7RL)~9Lvu^e)gXm2yxjp+%wU{%P
z`NX}W)PX8>u7hq7pJdxO)3a*JPTUiQ5+DhUI$g7?fSQzHIx9VzFV0chrrFr8tRRyc
zYtAkqVHiz1r^#afXm$*MtdWZ05LRj4lXZoU=Rod}cp4{Y8asDQ9qK9d&nnPEuA4~$
zEM`BStt7HDIw7G4IZ1jMSfTo*2FBKSjr+by2*f^-EkX_=ji@_q!4Y**8MzVusBTQ(
z5NH(%c6HWZ-F02cE;GLbfAr`)`&r_TSPt5njJ^Q@I^=7NGkgUp|8cmZj?W`EaOzLb
zf`6!8{OW7NH~lK_i__=Jur<~4rcTw<PQesq?u+MD)zwLI6!P&BP~>1U>VYdtzeyst
zc?PIl<e%OKg}=IDXlVKdrX{{)NMSk>dG6u?bjmuNrKjab5M~}Dr0Es_m02geH%1}6
zLPs|}vxE)(gUc3;_9bjSkvoK^uo+zC=e7P|72f6jIyDR_T6bXlT}UZWeWOIShm)Fy
zi`fQ~B8`eyon<78OZF4_fj%s<C~N@#i2DUZp|<RH@i`>TvjpjZN8>;>4qpQq4kV{v
z=n@pzF<WF`dpUe^Qx9+>1Rf3PpU(jHlidgCO~Q8Nh7xbLNDao~5}zhMpsU2jfl_o#
zfPBEVF9U@mjY9x8W-bHaAW#bGa|4C_pu~LIv2M3FLDi7O5;T~+3^E4fvrIzqv{SEH
z?eJWA^HO1}h(d+}F=ZI0QvawBhHQW6E~_oFEWqwhNsBQ)p0p3hgdXx{=PrAdVec<b
zq5Oz}8Z2nMg!~qv{|yc(3IZ8)&+9dF<`M9_&^I7s+;A_w36@~RkeH2U1Hd(7VQ|s|
zb)h*ONMyx>@ovgYujAxraEe67D)j8J$3=EpydC$j^Q7Bg?9~={RSvzsEz;wRFKms~
zx{%uEdd+mZ*}XgJ5uAZK0TL;-U-qGC_zJMu_E|X-*Wt9>+m}L4rm2M!DnZ5sT%GU{
zowTjeQm>H&vKAV|mEr^TxD3XnS?z|pV+;JVdp#x6@#m*&IGh=PYG7f&zFgs>Fn>IW
z!iKcC1vc+PP=u>wbDwyH0PTh+0F#1J`cOY-5;A@Qnuo)#tGB9*A{!wHIv`14Rht3!
z07jgE=4XKGopiQp*)(#QCpY_HH`qK%FTbZdXC$nx^_D_S`57@*?8E-!!OA%NJMKV9
zF3W5*0Lb;$Z_LFO<mQzRs$h#}9pY6?#0ub}gDD6r1Rgid`{2Zm)`7v_L4sV$G|A)w
zAnihO+iNzCtDaou6pg?nLl2G^J}C+C#Hg`@8g?g(cq0Ak&JJ8N&aeqQ_o?SLgq#Z;
z6f*v}Lw|sY5#ZAAS_Bcn0LJe$^+tuXSiE}ngMLBq_Yk%h7?PTA^<26NV2->AP7bKP
zK$J)k0k9UC2BApS+^`aueHXFE$bqm%G1Z1Unt__@!iDeX>b-|sCBQ5mM$M+jVNUl%
zYDo@(FoW7|5q?X=eURi5wH$7aKi&<)7bB{x30(XPi0=U+Fbo%FT7edN<&#fnP>Ih*
zDGC}#avwZ+(3Xqh)czb;2I=LT^g?2}oPZum9_^UCpywlg-RaYT7{rrwIOYKhRxjmm
z+`t#R<v?~L9Iv(<v#WsChmSX$^uyi7T}QtTNspI`VNS*t6zs)ugG|G%jz#GY3FV$T
zDhw4huchGrDw29oXYboz#qfuPaAB%Fore6yTKM~J`2T`=q1NO*65};eNXBmKoF6AL
z+Fgf5W6}YibgKGvGFe;nU>|q7SDkt^(lTMQazeN9|EvLDo>e~g1N%0(dJ=Fh#)A~;
zRkvt^%h7n*QPWB}1>+3+t$Wb_;+he35O?KgbR7scNS+q_XgP%h;4z$u5=R!u0ruW6
zy|K06pi?%YEfXUnQu8$cg6(5uzusI!cBxlyg=Xal4uUG1Dej70@$f@b9L%%b4;wg_
zCcHcMa@?4D(;1DF)ym3#EX5NH@$)RbVy!aU+}hAmseQ^e<2j8B6n;zd@9L{9^WO<>
zk3I(Fr;4mU<VEK3f`yiExTb5iOR<!pIpyS%0DsMURY5%oygF{dyLi@neu*g2Z$jY5
z)p;#H4OZJLzK4%*Fi?8z9{Y1n3ODyR?3p*i+W1=!%`3u*1)xMi(#30-=KvNnj>(BO
z8}xF}-QkRx4QR(3pwGzXaZcw_!i{O4P3NH6k$d1<_c4Na>-;g>mO7a*g&~f<?vE^3
z5ts3a6AX5XMtCmU>vzYF)%z<$=(cDm*)Q2?6ua$0f=8oM-e3V3rv^3pzZ5y|qZpnb
z>KKymJQu!k=3c9uQ9~)2P*hO2ryb?UmKOptmWOgt``3(VcV3Lyd$!WH<mYzh>^1a~
zX8^PxIf?sH*c%OXQt~KhFJWdtGEXMYAmK2aG(#chCCs+N=bs$?`r#Py1t=ZAjdA=D
zfqyUt$IIvQF(e+_<R9h$Xh09K98HrATqNDW)D(kwNQS1c4SoPDG2#erDc}IM0D91|
zkQUIbFJS|wEQ!A4?e6Iu;zzDjZQ3xm4mZkZ)D-R+t4xvU(cPL%LdGUEDe$D><A5RV
zz0F3zKz_!v#UojMPh|jR<k9C%I{VT#6gB4J{nP)MCTO@@ZbFmg6dn&40sPjrq^V3v
z;%&8bq;W)Z>h!hudPfd+la4p(?7VpR3Lp?679>6eYC(#d#>>x}8IeofhbSrgOuNu1
zNn3}KyhC={8Z{hrNB@>I{`xd6khZIUuSD=`i@oq{GWez2)>!@2jZ`1uNzJ`u5z}&G
zhztEShcgMU6q&&N8=X0AX3KLOg3G~QHI5qz@$>&3lwaPd*NJ9YNWu2-@V{>pmUJ6l
z(ZQ%>e>aIir8>4L?*i^7oSX4k&s;;57cdNw;-Tck;9YQ|kmUYM`ee4iOVGtX4q`bf
zE5B+LBm5iSfF8VXo-vQCT9=sS=z}pW{;>FXjzT|97x~ORJe+5>T#Qt9(0U;_ZlAAi
zLR@1%V4iGop(WM@?bSP;q)&wY_KQU6z<Qz%r+WsD?Y`R|J!SmJ3lI^~D?{S6lUw+Z
zGQarsnpK^pV;K|rXHj_&AA)mF$%+aMC0P{*=u-A7tdyX*ILI8gQs4GBMZf|9iTd(s
zjGLkfcNGXxZP{Iaae$CG-U(M*qy{9s9VVA^?CH=E4=U0*>57txnbym^;;LejRz~qE
zs6!H9q^bgWr-Z}Js~f5xRxba^_WXJHiAQ)vrOPu~Pm-wG1d9HHaT_>!*@a7KH!}Vp
zYD<q~=HI%jfI9XG<V2L5d<BzWHW)mRhnw=5&mued0Eqax>;6In9`31FhJ09`E#$gf
z(Bxvs6!Jhb{fpz<6F%#mp@`%T24b$*?OMG#O(->;CvgksnIZFZDRLsD1Ep6c^hZBn
zu<360jpTSofhe`^f$NA-(vCwDC{dVZyIx{o5K*}UjhntK87ageus%Fd?%Nl_td@o~
zJxOkJ%PITW=|J|r)k!dZh4WnllH;Ynp^~y?=^Efu_+H@?w{uSil5VQ_rU+ImM<)-`
z>FVXbi@*Wy*)`yiaHs%m$Rs)jFz80m&)!TDvEUPEs_6g~GT~=D^!>p(1k8eek<>q=
z4*A_8T{Y5(LwE@v_58GT!%+{9rJWaJw~q0|Iku~k1JwIt8`4N)$YH@Nc+@Z%PzU7P
z3&}WGiU8Ab+W8BROfrB3)&NF|t^p$;0s->EF>!HmD17_h?L<RtcBW!EE+WDS2nxZT
zSS4)g(9cAz$K{=H9AMTf#kN+N-%SL&=&aY!)Px|-6sb_q!$z@<<5mGD)hi%?s^S_d
zuZZPsf%T+?XTz6R_s{zyq;XhhhDm8A=OX(MV2j=xckaj1$bM(|)Dk2uC^=#lL=NJ~
zl5<*p4InS0b6_-V1K?s!9E)fj`iBGsn~eHO88d7|R{eTA83(s7`FsA$`V(@MiIWaP
zl-e@M;m?KRsA$juMIKU%D3pxs87Ul2*$u+P!|;7J_8Uh<kVbiUIFpN%E3#sYwr7dI
z0k?6llVelG(w}Tpb9+E!GSyY>-8MwA%)sPCZ_+(4z#xWQLQTiOQ5Xfl0^lcpK?Fv#
z6NpEbUr-U+ia-NQ`fvD=dIb7JxdU(^BX3e_;n5S^@7zcD1HcoKZN^f+T(xeI*&tp=
zO(||l;g^Ph+i0!|9;7FZ7Lu<q*oSCraT3T8031<LynT2)D#YSbAF+z5cJ-=?3l<g&
zKkYn)92hOka~J;sa5{Q#XS>61<RM<3V?ZJ8o$ohM@rsG3E;Qo^An!nT6E;E0XbiYw
zXWny*`jCt9F1!ni1voEA{g0m5VK^&`@(jtF@WKmZj08f3H!gv*M{a^TdZpVfGkFX-
zWylL3Iq>%ZaNwwLAekmf4tZsi(TGvZX72kI%An-W?wsL$3^kKH;KT&arCI#n1&kCN
z2n@Km!C#_+IJE;4WUtWi`~`TM*Y?p=!Y8zy5f6wvWnL-<BrFFP0pHuu(n4DD$m<Ap
z;Vrt6WRezHDeBq#ou%Y}6CnTC?$RpPPKG;SF`cYyj^L#*o7aPL#hc~yv!Fv+Bze1j
zR0&&J#bHCZL4wV)eeiq&P)5thNs9R1N*TkX8`B>j85&gE*Ep@bF8jIUoIFn6))uSv
z(eEgXw0fjj8@&NQq~hOSg>dG|!dtc$KGx!QgpaDx$APp*V6VIukL&9^BoOs3P0KoR
zM?mSiqTz=zw=$rYCIcz}K7-llL&ox6$hYC9xFOxTU0DZ>i{mlf2-3ONk~x-}>gulJ
zKAd~9IAXR;Yj%Y^d7ucn*D$oXEq*Zt&&lk@AHA-04%qw^`I0LU?fk)>g<hKn!j*ma
zuKAqMCiJT)AC*3hA3%R9j>-y@A_F@rmu#4+dV1+^t^8|!jdrwQc6>h5rFuF<7oil7
zTKT~Bw<ivUXBIeIZ_i0g{EDhQP<D!hD`PQ!zIJqX_6O|En`e5Chx7Ua@;<k=p!WR%
zIlKCjCUa0};Ha&aokHY*F@OM7#Pvw~1)?PP3{lXyu7fKBUHx91k=(564mHu~xeSC~
zr7@<N4I_|wDg|12(hG^YLw{hs{Mk4xzB#kKt{SzNnCH?$r-ZuLyfx7eWZiS5{i&8z
zlPy^{NG^n94n+(Kv9QB02&5=ZS`2>D*MH_z(Mux%`equ${0S)UH;>0ksL+OyW%Zkq
zI;;0#`|%%8;M&5$%n;G9_@U%`stAtACe9JmfI?$qBnHQB<8?}_4#Ur-8>;&&{~03*
zOUbPs1j2RnY&){90w8$(JTyf0bftDl^$^0VPA3_r1bWM%`x9x)C_Z^t_TO~pVIa%C
zPoLvh)%nmNtnZxR2bbX+fSm`4igNH*6V6&l{=a=teg6o`HCJBmym)8C<jSkH_G^rD
z&6VIU=|*(%P^a$;PL`Bkp?{{Gfd~bVgGX%sya~~}T+A03IOm1N)HR!+@E46+pUE}D
z&dr555q0c^_zV@qD5#OrM~m4AD&K#NaBM5{b;B=cM;2)&@QL;4(uUh;Gg~>TS=Oqh
z+-H}TNH3`SSU`waTrp}v^u)bi_wm6;f32VX%xVg7c}~t_^f5!8UX80fD5Kgt;)`<!
zewfnb0JRjpzjL0bFst5^q_mT^H>w&8eL@_i-hCQnNpCb7F20{mskRS!-PYkEg;W&#
zw`qFhG%N+DocS47BB2{4L79{UU5t82&iIZ$$bon`ij7vV8zPKteWV}FqNljc&4Bj^
z%OcEm3CSQyQeZL2Yxr#A;IQTL3J~mHvkw64ECIn~k6F14>JI2|kwp8O2A9#_F6SuU
zdD{-G@|({76{e%%Gamw_IFpB4ksBNwvp;w7z2-#qOEA|P#p@W%X#SubiOV_CT#gqA
z=Kh3vj6A@I6Cc`B-<%tlFMw+6TDzkT^vFijQxdLyFoKago-p2Gr?CJ)o~!iSWC!Wb
zxp&<M`BsC#r|@t8p&X}tftCyDyx{ZZL~&Qt)9lV&CndYaAwxrZ%9q>vWKi7;NiO)-
zaWwG0pi;)CC9Q=bAvrJ~`KIN}8ftD7;^Nt^<zXk~+i{PXlA`Q^2BGXdZ)t><RYjwu
zTyC2nLEmsHaIRJ!`A21FR>b}Qj*|Yr5LOHQXy0)>MaAR%9>bi(FyZR&{(_1uWnc#*
z_x3lwjL_E0mkiQpF|-uZX~U%gx)IzeKlITDsVVoDyd&OsBv>JM8-;I2mTqbtU44<F
zr}m%6KqoYNQ$K@v#c<~teie8N;|Cd}RLRdD$^@8Nm9QOY-HG*W3vfx0xc>bguu0<B
zD>M&MSdIrN_W+iJE@CP4!I9Tgz*hv12;HgS<r4Ia<Q}7YfdHA3LEx?<emkq##UfW=
zLgmwe;8>;v9umx(1XNdPtRAX-U;3<W@8tM^GRxr9?%Dt30>n8KtVw$_0^n@d&;1DR
zSDF#N<?VHnZeHW5<YM0GcX}bG{o$Lc!sD25xm|qICv9SrrglHKdM>v#*V`e)kBe<N
zGss@>elYRuoC}EIDX&)vpDq@6ztj(i;}}~za=J>@h)&I?A*pZHZ``q^;2R=$TvOl?
zt0|y;M}_YZChkOn^Q6^O9pBq33|w!K=5++}Vn3cvdQc}1+-@K8T>71mVogMlpP$n8
z?DG}My7%<z%T?>EzaJt*o=%mUq3#Ll+>@;bv=NL5KN7}*_`HZB&F?ea9!BRg*R5*;
z|Im9WU_8JbNEsHd%%RvO!|CM3zm*3mfZvO1f1Bum2ty?2-{KVvo}OZQQGFg%fB39`
zI@&bVl-h2GW9sTcg$0+ZI$F;L-x%4oA<OC<rwZhzRIb<dXIp*CqPc;_Wo_2bHN>*o
z569Kj{hII9oJhfo9k8ac#xQvo(}@(hsr003FRTV+pFd?ju75YgEkRvizYz0dAUr5n
zKN2$H<xf}%`3#4#_UU%o9XsL`P@v4<sIOx`w3Fo2T^9(+u5SDlD&3>Jm1<GS^1w7x
zre(pI4Ebo!d^`mrj<9M-36@g_9F=RD1Bjq<`RPWhscLn}B_vgJ9i;;i#>vn1zl*zG
zx2cANqiNoDuAVe%p?eVA^d6Mo3GhJmn{X!*Z-8C2pJx%7qb1|p-L2tbJbE9(4;EGZ
z!+jc%u2(qH=^X{P%p2Ygh37ZsA{rL?b(uJQJb&^4bS03UG8yM1FI#z@$m0BetgAg8
zCItPYjLk-~C2T9LnMfI|zD8Syu_g5o+UV3*n_jb%4-WJjW-X6SM(}$wqf|G@rxr?(
zjC-dIW&sACc;yx>uI{hkj)7`_Y4?IQz>J(@N;PpRd$i#Izsa?1+b*T&D4=OE@EEOM
zFL`ti(x&L7Q(@4Fc<!KlddYiV9P<OqS?GwG2(C(|TLzRn>#c79M0<5G3Od*pp%(TY
zynH{F?@=45c{}8O^BSf-&#cj4B|W)zaYK-B)0aNLs{s*t-uL`BY^!!VWFsV|J(dU&
z6nVI@RK5Hc5=;WS*(GcYZJN{9Z>-v8sq6~+S!Ja7S@@E`8{J5X<@b#2Y>$ENJihTd
z0?ICSzz1;q(86~Gg2J60m(2L$mi}-}T%89(FF{oB4+I;6szFSj_g?Dp+f6MkBF+0>
zz}*Iv$h?fY7{62M^miccj8jLgH&idt!9Lc@a>Seh119&ke8gm!15DTZt+H3Q=kj8o
z$G(+;Dc?2#!djzMTqq6*7ixu$N=}_c;5EeSqfhPhapnNW)_m|^&_bJN>wNQjILM+N
zdi+K}+{JGN9j5BMrR`({j2=K;=OjtDk*ItI0su2XU6oCJ^+v=Wfiz%OxAdQAwBt8e
zW42*BXb~^OWuOobl69A{J!RXxC<?$n-2>&JD1_Df4U}y>ckK*zTicv&{13+N7(}_8
zF@NP{R!Q&pbDM9`!Yo>7CmVELq26RFr@p5Lm>u9G<C01Ar3ApU-eQhE**DZ?ACsWy
zCC>)hORuv)9pHVlzg74?DX!{HfIfNV?<YQMoejF5L}~$`Zj+87q3~Sf7d1^yLIFAF
z{iz8-bI7HcAD`{wyBl+O?1*iLGEQO6x8J%@`e@`gw6uMt4bOt70EEm82s7cU2e2?b
z^R-JKt!NWMnzE;u4ZA)Eh}*9C@K|RMS*UPxPzt}aWuziYT%1`~N%mQcK=j6(QR02!
z8*BLo1?>&VLSuTrW2p@=v2FB3xD&=NdZJv^%kKJW%s_1tH6&zw-gKk>M^5DVdIrwD
ztNb{DJ72^8c?Tu-_lZ0)-KD(z{1i2YuHVPn2|GCa4hu^7-15~(K3Sag5{iZ=6DN5(
zns&;%I*vMfc6pC;Ez6O7&uvv3h+B#nT^KvbwVX~n0HS=iiRue<u%w=g#g%q?Ty5Bd
zEt<eG)2xz?-#*?Z3YRz!Bnm^ArKqX_?qk<z5)kLy_qL_KUZ;nsaNZDiSZYRdX>Wm$
zea#|sH$;+XH*e*cc4ttl5f}h(&u?y57FZp1sb_=D4R!-vMc}yvl8lWD(mRa)F&M}9
z3k+=b{HthlHuzCk<*Lz>OeA%6R!b*OeK29uJSaO2b1f#PPJP1ojcGJRsIfOpyI#|E
zgN6Vgd^bFJx=0}?nT;LjFxeJkT;JSWMtmi0E1o~JR_}VQKxms-!YUt}TlQ)kbkE~%
zfIR5JJ<!)pZYgv1ba%db&NP;In_i7g54laHY(M+ZCq716c2vM#^6uya5$UjI?P6FU
zC_ZQ)(s}X^xJY+t^j%7ThlN`r+<@Lz0@V9$6quaPIxgzzlqph+nG(i>0z|B$C?z~}
z*a68spDnZN;E=;8ih`4&dD{()Jd+HTG)^{@u)%Y1>@z=VF}<AR&Ed=Ls?GAB#&rFA
z$*$hzZUQ!EMP26y*p3z2H1c+anQ6sB`lvxGbuK`T{ud~DgiD7(J-ZhJ5u{8mQaq>q
z=Pr<`1x%8Vb~Zc`a-~uXY3R+`^?rWiQ_<nK@UAkn_>Q?;!CJcQj7vc=F%U>G`YE5W
zr1Hs=@$S?+9ZjYWn@{~Mue=P+!3>_UtM_U`pb7csSqJ>1ZBg&te&c&FR|)r{tw+17
z{}0-BAv;yrq(?6CGy2ZU81%P=AVsUue{j&wG@L5<64f0qQS}B5v^B{b&_q~d4O966
zJtvUK_al@rWG{i(ZkpHYp(?m=@O}x3Q`=%ZP{2S5h+=3L#iC8Wj3UweiLhbLE{9hj
z2Yx5vtJe(vX$jPdnlf8~+@cOEC6BnYqocDtVZJ(QKr&J;3=+P?wvGIOOfz8&j&Fr5
zb@tD!7TdqSZ_GGPrvb^_Z4H57ossPgt*x6q6~p6D_sO6}JK6YwlFg<e5WbMg?EJW2
z0VvlGU|w%<`f)|Tm`GpQzT%)6u(G=B=YZU?YmKrng3Nb5vSNW!u&*{=yMY)U8OH5j
zd>0Zx?ceo;ZRx!4lR8sJ{$y@K((gR`BK!m3X{HwI#or0bqn`>!NP0}TfS`&HrnJqc
zegEj5o<~R=Kr<Bhml&ZPA?G9@xx3Fv{?btu0odrPgj4+o8$q8j=`YZ<(0B@Kbd&;N
zaL3G_G4a!Z0FL>`6rZRHdjGyi8D~r@ZQO@oUgvfW12h{h)C)ziKWp3_(9$Amc_PP9
z$>{+@Y)ob4;xJ+w#igs^D`|b8Y_2M#AX@;;-~@O99iBo27Ibl504wUtT4H-LW%n1~
z1VWw<z%m}r7RdZ8Za>M@oHzL@*n9KcVVZrcWV%}EhIR0lSEV$w%UeD4_hmGP1RN7w
zpPQheCURZbf5+AXP5BdCJoR7DOky{F@HLvpiKPGtc0xi0ZW`BpCg-&=TyDLpOTYNG
z&zyf~g2+oYycFFYtD5ef_xxBgPW3n=-+m!KN8LwcSJ^STMD2}&Jp6|I$<(Pn$;s5W
zDT3R@l>>I%Hf-yxIjAh)d&XyzfW%S8$L6cnGBO1{+*MYU_i|~eb4I$t#rb<n%Lk*W
z-?Nuyy4+fCT(~gPA_b^4c3rUgfjdkSA3LeTv0MY$Sve~IzK!9BFhs<O_SMkl;vHQ6
zE;YzVK>{0bV2rnay>p*8qf(%y8{joak8NA}LpZH~i|7Q}mzjQlG7NgP-^b6TKr~Ff
z0ibF4Hwv0O%lAHd1wPgi5%5{)AQLtVdkj*|+9_!uT-kp((d>jMUv1Kpd|gC<GMmi&
zZuM*x=hBwN2-4Y^ljuhE%O#x)balgWY(&I#C7o!0i99FIT!^+zM`)F9zZZq`H`$WT
z-Q|eop7k><9ZGzfHxk;;zF&N7klhXE#hw2WNFy5tQ_><$KcNx=QFP;uy}J%2?9fSa
zlt=(C4e4VXFf}wB;^C5!h~;-g`@XokYF;kstEO<?_*Kk2R^vao#+ED{sz)pDco>}s
z1nh7kR&t-EPm316F8lx{HMth}Ra5({9K!mQPX<d!0iszR_gpRs;#yp0e|Uo4%CYYD
zkhfX#fYpRTeQ<R29APJ$6ion5YPEkfg^_*;TvfnrfB;Gm_QBzOh3=hzgx~-n$`7sN
zsIR+!DK`Ak@%tT2;ZW?m#%nNjtXsiq@?mOf7m})A3c4=-@wZBO?#=$axLf1bc9jJ=
z&D|M>xqU!uP|5Y6D9S|M1Z+eA^5+<$V#S&fpnAX=2^R>cuI<!6s0#$*Y1G37R}uik
zE7Cb40#exIgH<*W5yXAxw;UQFqnMz1qMk{K8LioJ_ODO!G16s=nz}#B=G<QzASJRE
z6tO_52|m|1xBl(mbe{!ue`J-)!{I7l$0m=3e8vV!8sPoji>Y~}<;@>GzLs%Ha#nOS
zq#cPbKv%4f9sy5-kWHtXZ__dfg3d9D>N^(n14MJi-y?aQFrMjT9JaCt0YMKU1JUr1
z!2XG1nW2!7s3<}^k(+a6$PWQRAX8d}J{L3J_aSw)UnZLLI>Gsz{pZv|$+s4wo&)Q?
zlAtHZSn;)7lb_N0OYt9Qq!OYin&|!4(-4P9Y=xD$V>;X&sg@FZJ()3qKcUD)m)yLO
zpf&L*V!T{4_7I&HZ*6vJJ!-f6gq46^%q(1F7d{lYf{(EO`LT8y=+A|4#Dt`cna`t}
zyR*TfC{Ivte8TXVUvP?7B^5cAUWTqfalJjd5=X}ukiC057lHcOSMB~lLTYv7mo>?h
z$!8b-?k>62<e?XIV2rlUxnlkckOxA+Cd~~rafvh?!J24W(BDDC_cQ}2L{bySeO(TX
z!o7N0;E%|Gb&3JQKM&3YK1R!cjMzm8vo5)&BgN-x90HIhEE9U>cyO!I_rLEdt8O9L
zG1%Ce$8;OmL%*SmXp6S63+`4Va^Ryan<Yt!RjW5}%|y9r)m8|kDa~Y>WZfU(O@1&R
zMQEGLApcX#vhTdQD$RtMf!_q<IBY9RZ*H60vLioqJiG5`C(=KH@2H$@dpY}Hpe9HA
zQtHq~H>^+$WJ>ov=0(ed@z|?VnY1Gz4oLV())cJRPt-Yv^wygThN-JHsnIs}URSk3
zp%&};5)U0oI#2`s>LZJsl!;`6Bn2-(BP!V+uU$W_O>4i{ZWd3vZ8FPMn1cEr?7eqb
z&;9@R-{~~7sqCzxj1mnEGD?&*%(O_zR-vKcL`G<$B%vatk{t~qNlR9C+L_Uyl=|JC
zo!`%2zw0`V?{QqmbzJ{^{qgzwI_v#@y`JMS?vMNZ@eo{ioguQC!)x$nHUYDJ&s7J+
zq?%Ms_|kHCyyZ_~?py4H=s63(LF$)_$Ad&tygZM;Qw|ZmH6$;0o9Q9=e*)I|VdYxv
zBluHH)}<gdcELGQe%t&^i_4c%++#D0l8U9vr%%21@N5$!23ndcbw@Xh@eLbgkUC_W
zv7b1L7%Gt-=fv@q7U1K4>2}}JcN^MA%?^%wNo!TNUe40@XVbN+=V;5Qp>Otkl#OYL
zYtgGXtDwcJYZdSAD}q~5!r`d#7?iiH8_u7E#yCPNbHNhGm9h~w&8P#dP)dn14wTNp
z0udOpbpcJjMr=wP)$rCK9(Ub8c<{-klcmKp_RyJAdukV$WJ#Aey)5WDTBN0cf3KK&
z<|hB`l)6Js+8w@3`rYP(p}iT&t~pj&(zD|m4;l{Cg2IJGT_PUvnOZb^x(;=`0<l0!
zpK2e3$H+VE-xpf`qxy~Xtt#(gK0>}_hudHJ%Bl;NgdB#QSMD0Ql9d*M+Yy-$n)l{9
zc2Y6VF*1Do^r_E?Rpo*k6``aE*JjME&b4I#6qGyN{L+Oeed37Au5v#je|5bXE%nXa
zE^Pu8M#<I_DDBlMqV%VWJZ@v%o_47p(mnTDWp?goRy4eD5n<18Po%`<p|&G_U?5(-
zJIP)CRaH<**d&;v;F-4UF#l)uCgE#DB`#8X>rKxkNY+K85klzfh;b_0JDW}3D?28c
z&uEx`Ai2A3w|n1f^@2xwdo4HiYh~99d{&>>y6A^NAG!U9BJag!C4x6q8Eo!xU=(8K
zGQox^t^UrT-1BB%!Q#Y++Fn0SzHXw{SR=hG5F3^OE3da^J5QR~>u-tY`tLUI7JM<s
zJy*xkr{r!21*bpHg!H;aGc#rziszNB%u6{<xkoqf>5<LYk7}pG!q)g)lYNE{Yodm#
zd|B?U;tnButxdcPUq-Q_Ma7`xDyr&0=W-Kl_lS_lTy<Qvo>mTi$OCMFrjItNr^_>q
zHQq_RbGijn-nTOGu^v4pf$KFjB@wgq|MpbogkTNlOE<7Vnd{dHNlbxHf%h2wwWupJ
z4V}g<FpZ8>RxJ2Y_J)QQAuC4swO3YDN$5!5=|+5KRCdq7lY5<5*8M{CrTZ6~seR2W
z>mXRGpxJIXy{@_!Jzg=z3;!CG9Tn1dcBGt5oj}m5(s308fd)NCj3v-haZ!xwyE?;d
zj^-E9B!uhxN_QxuYRGI1n_t2S(QLlInyfk>8e)~FI2JKY3X!Q|y7f-;39n$L_RH%p
z6J1>k2a^T|!_$__To!Fl02-hT^zfhN&BGjUD84;q`_CMx<4u*Ak1;K2Z}@;d=dTPq
zxs(_@eU~o~ON!pS#p`=Sjmi`X8s}|7e3%jbXHM|Q=>p-GJ-UF?r2FTi$(;S8mpTp7
z-@7<?WJPQB@`AE}S$lOv%2HQ8>IsiAw51Vd)x$)!n%+km4!sjfPcON^sE)psy&pcz
zxQ!0NLT~kckND9o&L#=THHJI4rh@u9zK?_>_5E#cfGbTI6M>W7`*lbJO6~KP1nK|U
z++*-$(|sErgOMw1bvU+<>{`Dc<I>0d_WvqawZ~@l{I%l1>@BU^hzy7<9K(~^QHK$X
z)d)1Rg5|tZR$G+pqhgMYeK>I{Lg8Xk`eV{M<>izrhq|44#G@shI{#U6O8V)Ud#A!P
zb-WdYi*ANCIK*1kIVpk?ArezPg9ZS)a^+-#Iz8_fL`W!l9-2+&6jr4^?rkvG>`tfJ
z+A+804{=s7yq@lvU>7m=_nbE#M1*fegzEn{+w7U%Wwc&M&w;86w=B1_M+VB>mJra9
z#ApwVnP)0acFcl25gG`$HN7?Yw5bC@jIb{3m?+wY8fD8eq2qV$1`KLyX)_iYdgOrB
z3*J#i7qr!?4rdtHD4p8VI=c5B!)E;iMCS!7ggRdrGd|ddn)Z^qd(Ws#yF2Ffy@fff
z`Z%n_%Dl4=$bXXPv2%C8lN;+_^=Uk_4Z`Sy<nw*MY2VME$h2lLS)gS=ZK2Cc+QB!E
z`h}_nqkH6!^Om9*EN1c6(5^3kb<t1WUGlMM0W@nJEqk_Qe~#R-o`GNQHF%!HI>=3K
zxV7Wt(uU8XM{JVctTFTCuVgtQsMc3jPi1={yOb$F9!!QD8<iFZe|fO;WV7Xpk)tvR
zXCl+a!WDh(kx2@(Nfca3XS%)?Dn~|(udDmJMDVF{lponHkE;5<)=AH0IW3f;3yS+y
zA?WV@x{jnzhr{toSAf#)Q?FsWRLWb+?N|7DW{a{DWu+-z{W*-qC;Qs^ogFsfq(Zl)
z#nB%|gsk$L)US~+r0>w`D<Dp7tsh=%kr;07(}*k0Q|UWH)PsM0JMpJhJZMw#-V>r9
ziu=zLjB)g(PePZ6!HDqxTX6WQ)+nHvdJ958&_)-OsSL1$b#iZS&uadmu4;c#%g-Bd
zd2?4k-Y-Z)Q6W7}(*2oeSR4Xy8!;IOGeeYU1(GNZ5Fr(7fPuQy>v+R{K?fs{_HKGp
zB-mDnWA1XO!x`X9P%bib=sm)%Z_%18CE0!G)MV3D^XA?sA=w0W#b*8~SN3)jLQ*1K
z90G>qB8~<41S>|n^l5c{`|xiEHO&(gnjsSz?|1)J16_FVMEi$?#7G!mBTc}zh34N7
z<&GL<H;n#%oiT%)W$)m0_qgnfkRSI(h^Q|v35>b~DaUR5-44Ax<AHG=q!on=0nbBQ
zml6SzDTVV@&~Jb81?zbh$cBO~1Ve3a_dzl#tASV9nGpZlQQyeE{gpgmlF!2fl39Nv
z)hJ>X6SX+#4ytT1ar{-biCp70`pA-;n@g)03?XQ?0In1=D=6!TVkE(m6Ovo%Rr%qX
zwLFG4rUAlJ>0kr88f1hYKELnq_X9TIZrwHB_OgDD;brH9HD)Ve+6z)#PPch}2|~@O
zQ22RHj`tm~5FeL)&ea}7ZkmSpa~G&fqwCDQ<LD0J4Yg}^s|Nrh(0I9fcjz07v1Rwx
zh~aOY-aXqt*DK?rwz!~$nWZ)FT>68%Kc3+SX<b0IKl2HPW!V_NA&&^9Q}>vNjL{~9
zP`Z4{`Kf2W8!1E(PPWEBx6pfcvVGH3447M(TiURbuZ7F>DlW8gHlAD@v4!g>+T$SZ
z8F#ChDve;uK@@okqW=xd1*=>2GO<(~@m~jD*(qA>)el|)VhK$}kLZW|ww}cU8QOwE
zjL=X-|1chZ|El&TjnONba{(vyI84DOzD&bc@<vwR?0&0gIaN?EPEq&jUe0|Y))yH2
zaWQ1xBp>b&^w9F}f}Is7g+|Hjq}F9^Rru_oO?rOqMZc0}$$LFN*ivTy5&KAH()b!K
z+^qMrWTE)wIQvx3Bv(zwbQWmSvqCPjdFAv3mA8Lj_ZT$y3ysAKG;%&(fyst@rhkny
zRj51J+B`b*MY{Vvnq2gWk0BRfaSQf>59(wtJ0w;lb1EdoWo13^lA>H06)R|v=^IMW
zHG$P>6g2To(N;-L#h#J2+oL(uo(Kh?GX3OYIj{F^zWpcLwxv%@klt?X^~*V0Cf{Fz
z3zbj)2{xHke=hWZ5B>2p-17mGhB)5uU7)@99Y#*Y#4KP|8R6?d>*YCjqp^7h-cMT9
zx0!{`OMHU_qISsg$FY7%TDjyG&W$=d%U8?_WD6Lxp>QsJhW~eE??xk=kz#avc-IAW
zt|v|JoYX6^>#;X~A!$uy>#x3n>jqAly=9U7lqsHA?CDMYGPX+(X(o**8QP~dv5t)N
z{#!R%x%?>l8EUHq0myK(<0S{_F$<VH=C<WT!T2U9>FTR;RJ5frVfl%c(>J=__js-{
zf5ColZM_4N6vp{gEnd2%_o^X1`lSFD8|3Q;>=U25f70yuHFc9C3qAZ=#&qlBJa?VV
zuKXKv+vaj7k%sk7YO%*DayzB2)<TYW4p#g<5Bp5IcwoNsP2+`dU#NuYRo4=cR)hw6
z;1vsQ>t9b4jMat&1%sLK;LiodsSsSgF9~LoV|4#0u38v4ARe4<iu+!1KW_G?CHLr6
z3ybANYob1t$CY|Bpa!MIXA-m0<^yrox7BsbTt!j%SGP2y=^K(SZO)FFb!ecbI@ApX
z9rvqvQM~)1sX~!ON4NbR2Z$1+QY?pitqq)S&zNSVilH_y0sOuZy$e3F=A)5mpLeAy
z)n2eyr@8Zw%+Mic^h66cuBGOfgW-;<<Ic7Ae>b+9&kXul<tAF5*3h)ebJ3zXVZ%^^
z!PA4?9+;6dfr)fj*zkG1l(cj3$U;^3!3EA+AGjs>&Mxa-F&nk;LFdB-RRNu#|KZ*_
zj#D6XzNh2`7_idbp7{I=H(ao=?)^DRCe?BLIzfnC-6}+x7&HPRFOE6axQJ35QxZg$
ziZw#pr><2nV0`$nP^5GWr~u(P_W8+_>HIerVkNY}wH!__+Uv10*Er(rIROM;<j=@A
zxOm%LlWnjvN;g03Y(Al|m;_b($Nj!nyju_$%+vZPL^81Lbw-X{0W!@B{CQK<@&e=>
zx`i0P6u9ogeeSV8y9zC&!{Q_w=KMP&zy?veh#yJ3n4nZocvQp<f;P9RA?XtO<kFj^
zeg_6Tfzy1&`#sIaK$L-pV?|FQB(gBXb)il38iMPBs6>o=pm}-~=-$unLFul#zIuLp
zTot6k7<o9nqmYH1%F-ygjg_g@c<YJ|PP4+E+8qYD;LnfI*}&jG&|&KLOjXi^3ssA(
zrsB4Dkgm{(+lP(pO$|&0!<1Eg6{15Un<lVR`@C-6oUNaKZ!*x?jEiZQU6qr&4=Kyb
zuO!843@uKCL?dt&;o2+e?|8VvkhAlu9{d6-<KW7ql%P`$%I<tH%9j~7%65E;8%p#8
zjU|IU=O{)b6yCE&)=Kf`DcAdpZDE{cAq&aoPX*E7uDpYg$Nq?FAfjn2nwJQLo9o%v
zB*6l=_gIz_CNPFCZSVkEGB-H2IEeYUSPr;IDfk3Z6c;Q>($67&a45eIEyi5_B7+kC
zvGA2Ry5<lhNfU2iN&yVtLJ$m{GFh~W5#`nNSb5N>ub_VWE$#t&^_KN_0=T066dt^h
zfcF(6L4YU(dQ1i^9u_kp2)QMl`hII#|30~qW&gt$uE<jh+{ejJVL$2N4u_xxfBhX$
z&x4aK_8T<>ho#@RB@^H$i-0g(4XDzz_n%4lkKIpC!6eh=Y%)qL6#^!Q?<-VN!QF))
zMPd6VMM;joC{FKxX7SzR>-m@n$^1j=JS6eOx3G2bNQhiGQ8&C>TX3TB0G~i)f)Ghf
zl2Ec9k!Y0b+3U*3jQt4UC?XEee0NcgJLHVTZ9Lv{Qi7M8u^%o*2uv$ZR0l;hkcb%t
zm%>mx81ZM6ma*9kX%oMbwh^|K4s~o=+3VQ|j2a29O2U;2x5mSt?$x+#)#G&@JW0ud
zK2xS?`-N&Gt{6k%i-Ya#8B%gmsb$bqIOUtyY0n1z#EcD+_3!1aDyH5zZ&0<`QlL-I
zy?uRaL3cyeEeMR<s}mAlP=$YlrJ3j00Pa^Z^hkq49qAm)tY<Q$ugvNkI^b{Lxp8{N
z0`?o0&FHE;x6}T!^1XhE0;!JLM1u2p!ueC#GZ<a;tBvX6`D5(dZw_@rL)Oy&+{MTN
z%b`7A<3NdiWOpZzU3lT}6m930vQ4VL;h{&3-oC%F%y@}e^5TO0^=D#YGChtej5C>e
z5BCvSc>Iu5rO|4_YGCH9@_PqN0hOSTwST^-Dyjf=udr4?2cu+{D@a{H*GCH7r*_D^
zZ+BuL!XwR|?Bz3O=%I+i2W+lq9O&06H%}@K5{1gi03*FrgD3;PMR_83#DmDYzPGJ)
zs9R6gSc~R4=!3yvNgtr~n;aOcSa=eaBypvS)$AVzJlq>>=Qk@K0r|<?0R-=^`41gm
z5Sngjz7i7?W6;y{kW`A?vL5g`ZIYZ;y4u@4b`boc`^Ro*R7iXs<`&i^S~#>)pDU7U
zRNWUXdcpA*gcaRA9FSyWX~jRin+vLZYR><mDOEx1Ge^B&8-49m-REp{@(_30@5iHU
z_!}V^Rf!$rTTo@;^R8?*OO6#pRJr%Z*ygS3zL%vfgGUxc6-14GHw7|IZNHh*0sJzS
zQ~z*unkE#2!Zqe!7uU?^<eY=sQ1T_iJ=p&?_Z#<8B_~?uJU19f{R@+{10%{&H$>>T
zIn|Z)7gHUDL_Fu>{J8+J0{YqYZ2>6td*ETJ#|M%w&wq&~n-}Z%Y);o~n<E=Bo%_8v
zNvH9ObtDr8%csBSfADBTgcrW1tn?1(vHyTc6@s2&zcmn3CLHR_J>3-Zv@4PN5(yY!
znPi-wHw}jU=1x|F3%nuH_os5nfkwT{o)cTS19oP%ZiiXLRb{l~Z*+VCX#=Vx`xHae
z+lb>kwa91*&u-v1xRGLOd1&RB=}D=7CtWZ-oz7wiQ-q1kcoj!?Lt+4nOUtGsmg(yH
zAo-D(lfva_sNrs3hxtf0+gyWCk?Af9Kpv3RCCyj&5N(~~T~4haIBucSh5+%c60&&q
z0n9~$qrF1O#cB&GS52%rnY&h?)pzo@CZWdT4pXw2>+#HT`qU!e=6TrY!t;qtAYvSB
z6AABI^cRhjjs9wFXVjvvYzc_jPuWJyh-)}4%hI5Ns3k&O)O@dFlw~`C_maFE&^&l=
zF)0$N4{?kt!|LPdWd@VsJD^FXt@%Vteh7|wZ`GB~jPDXnS;YMg><qKXTQmR(oYF?7
zLWMB*u|p6+fKZsWEG$9idXq!CYtLI)B{au;m6W;$o;wLA*{Mgt_yQ?(u1@UTF%C<h
zUq9|wjp&(`(%F6qb_r%uSWNY=o!)CH2%IE<B=#shcra@bVIaZEs0M>2)8vz~3h+SW
z$Tlz~3)U%?n}SB;@M>{ZS@m5GrY}rCJ{U{vMe^lree5jFa71N<=%UZZ1Nu!F8orl@
z_!;|&eyE`<9^NQ%^=j`mhjC}E)Kt0=3&1~WB#<dY&P!&D$%dQX7Qn0J4wUo}C@tQF
zc4D4fmREBExDGg3e&+hCzI$i^cS>!!6aB(m$<E;gH)V}JDXsvAA*U!=Nys)h$G4G|
zLpK;l8U;TG;m|7vYEZ-(v*6rg+%|@b%<wi=F(WuZ6uQ#@$cTr8tOMANfoPWQ?(Upb
zp~@$C(;d28xD(tCUTO$2hG>8Tbi?=(0_q4ju7YnCadhliilC_kGe}>;iR?tKjAn#H
zg`qwPGkhd5OAM@7DSF`5e_9ySw`w_v$sYzkkLdRatxPi~#B|?@-CU>NaF1FEE^8du
zvl0mEYWW8SVr~F)5Re$F%^Oga*bK-ms3b8bwk^f&)M`{)o+iTtr5HjQRwLRw9Gg^$
zzkScIkQU-IxWr3=CFnF>4mFj@YtGHvD%t+Xp{sG$wsocL_G=Sr3V}AJGa11OcF5}H
z=3=rig2IJBnu4yyrtmEy8t-e>F-2hnD$w7gx1#F+USjSEt$GNpOGs)7X2=I@i7<%j
z3Fg1yu}g9DVIRum3Q7-xQN1Pk_AuzryjXR#@2H4U2{-6ODKt5v1vmK|lHy40gNSrd
zhp-^?b^7km>#DqVh+qh0Y*Ol1uw4k}zhC{k_@$WJacAvFa2XF&DM@*{y<qZv1;a`K
z3yPV0Y$B*5IJ1Z)$Og$7;CAA{){+*P@bS3a7sSQA{+?&WXmiSeZdCL<h*yMcp5<zb
zmn?65BauhXO4FH7aIT*68`#c@*UkqtE53nsj-TzymE-f@2pd6|BiMTc$^38L!-wbK
zYG5HmnGzITRM;-^G$XD_>+hCtjqJCGJBJ*ghG#xOH$<-A6^I#Z2f@F#_A_?y1V#R=
zM^>v>t~^8y$DkNA>ws?1y`@y#h#Zws?~&bzu#H57Mlyd0XF!7k{&WySPI=aCRF+%D
zOw?prls8x+o<oRJ1owjMX5l&0(}Hnvds_-@<GyLHteCn?&7l>8=t00Nqh!Su))g}W
z13!Nv!YsfbP*s@=8=QQA8;6KooZ%u$)_jnBklk%AmxxrwodQmqzODK^#bjDuFfn-<
zC`TWYLG?h-oKyga910KO42bk{lT8SDd4g5a#M{+ZHW#P=dx_+4UlWpb!5W7CBSh>3
zqv3&5e9lqs$4HBp;hS1Tje<p<Yivv!`V%P6rP!z6&ri?C7UrzKiXl9I{=BeW*yUE?
z2e@oR$ovAds&$y=F^l6pNW&sPx5P;or(?#PX8Ch+8k*E$TVOA}BQz%|6O~)6e<~5q
zm#IPl2V3YpZgW!JWby;q3^xr0bC?E?4l6n%Wv<Y^&UNg8HROax@T+v$6(|v&%Zn33
z4OV1T09l;K+axR;*7j-g-X{)4UJ80!wuC@T)YC1@xl?*))sz%{*Im*{Q<X1%l~ocL
z^U+XY_=wS4h0VKsPtl{pv1js@bt8L&OwQ`9Zxq1vxXK18o?g~FvB+*w%}C(li>9Y%
z*ze|RS^C-Atn`&`-|O3$PPBLxBRP4m)Vu9R0s{j}yPPJnYmU2aIK3GL@S`6+-@xIw
z?^N%7xVbalF<~F;Hzl)cHTEZB?S%8$%5zGd9x-)lHk_(v?AIrf9A7j31~{+LHq%yv
ztW6ZIIv;pQmt>0>Ev|Dvwfp}08WG>&)^*$x$DggJR*#kTt|v@S-nT+7rpls2Yo&Kf
zOlaYgg9&8@lSySnsC?AQIaf`<@s~>!)miHI;C}X5oYU};MiDa$QWg^iWT|`>asp!D
zh2O)nf(N%=G8pWrm!mm0xa!h@2@7hJ-1X&db?;?kIrS{3tHNO`^06p|zuR~AQPniu
z!}<ta{pB+HOe$0~u5#Y`Qccrv*>3&e2<S6Qx)Z{u7VjApByBR4W~<Eu)fg**e%5y!
z0|y^)CL%7ndYupvRO3317a(l1eTTkJm-iF=3K6Q<cbLB(5={e@T)m@0=2F<vAux4u
zi1juLKj|kCP#T)2tTSn$PiYQ?=*wl$obuV-k_9y*y9q!jQVh~S4Q_ChlOpe6Heg5D
zld{>jI5V?)RLnLu9s;1ZGgw8&Kd7iop|Q72ps3>TTz=!=@mZGmw;!{HX24!&#Nf9T
zcBw+pY;m_)lCkS9)c%TX1t}-R-p%PqbLW|6nh8Py7y7BJ-x?UKnliFCfr*_wi}ioo
z98sJ<%1Iq0@G4hOltz6P>L!JOJ(CaLON*d*2`SIePN%@oYxmv^{R9Qk-ytyjO!V2a
zQ|lhjoO7_#upY*K(b)yjA`}+FJ=SUde;t!_(iL(@S&JuxllRssxHFgM!{S&<9yH%t
zXP*VVuKR3dbvzaNvCgf(MT7x|Y5xzsv5md4Q*G6LTjkqdE{mG7N<@hx(1ZxO+P3v3
zV0h`!h25$bvO`+rdu`1f=;RZ#Ie$jL!>JpfSa6H2bz*mpjxRG<Qlspyznp(#CZ_+m
zBx)^vtvrVeQ&iOq8e7vf1;bk~-+Hw5e;)m&V<~}wj75!T6f}0RlEsfQleL%smKc^V
zTGvKBJ+3Pdk1p$fH5=-wN8mF=Iv}SK8p=X(As=!8TaqxA@bNY~l`6{OEn7GQweAak
zkhsN4nXezwx*RP31^rm$I`(~9EWJD{xk}mxh|Nd2$9`>4R?;=!5wbh<9Gv8`1z3Tt
z`EnBCkQyZz(gHb+Xa{POQTDTzNGX9%h{hAI%%nu?d%3%UT2NL0%II*=(B*^}<&J&I
zQhsL~*Y)<gJoBM`mj9(P+d7G1Nd#8`{>KwCE|zI%^5E@N{?kNt;=mzJ1sJ1|4L!7l
zD-?J0ivq28CtSfI*zcVd(Ma>Ke+vGmJF4i8&Y%B(+gSy*SATssx<Xhh!gOuMLVZ{)
z&Yw#5*0~ZpLZQ6ZII>wj>;H{=XKxejOB2HUx=5Jr!?3n}>VIQcf&wkw#=cV7yhJ2<
zB<IJ(HZ`ma!;W#ZMi_~Po0O+Ujg^&<@OlNW>>|?T_8X3fmhpz+2INBt-ha4^<WYWf
z<+0CTOh>}8Io568|9@-7*8kh4s{gCq`ac>O_CK%l-{|jq#-8D-+8wzDg)51ahyxB@
zxksJ>>3*<<IWZxr*{MIki|Vwhn{l8gqiou%&-_b5!gn@KS7<RY`1gA<XKibE9Ld|b
z&8GBaJ}yq`4mr^lMKbon;lqc)FfLtM`?ti7!EKjF?k)#g$axk11PbJqB65cWj=mih
zju}cjvE0|OXr>p&-2dk*pJ&|m&wexJcmq}5BH3;G>78vI*|Qu0Ok&QQk-ij!JtnKW
zVp>9v_0YDq{AK3|@JZGsm+TkQnH1KFLow}%)PycJ<gz%kev23b(koclivU$Z@0Q$6
zUcuG(Zg2f8xQa6l@=%d&-TCJP(+rDD&12}sS#&%1{ip3$<~Gjr={KPR%hIK707%ez
zau0z-X8waW7cKj{PT`6_oZj{VX9aFV#5}Z9SV0Gg9VNrthLTuJ?Nh?ie+B_HYtMia
z&9>K?^McfqoWAm703gEIvu9<@nD^5*z(->CWpW-t-*wZve~_ufQpXRA2^X1IhvwIw
zqLhpNd8c+^?BY$t&w__Ika>yrliD646|_PXrs&-hbRrtPw}-^EL@|m;!f-d=GwNnc
z+d9o_8;N7Mn>x!`oN&^&7Zh4W&7;3q+x4@=AQt%<|3U)CK5q=Ma+;$eo_%@S`pyab
zynZ}iD<o(?#6Q1k+x5dzgz_7aU4_qk1lphj%v;i}i-g3%wy*aklfQ=OHm`?FzYY?o
z8roJ}es;K?8<Z$;e9Le#XQj;|D3}2$9cPfP^L(;%!5+wA1k(<&Bc?lv`@TXfJ(!dw
zmPXE}?O!k7pBoZ_`U*8*#!$v(-S=A3XIN58%da#51?5*}5)xU7z~UTs&aj@0z%}Tl
zV7O>DhvfuS9PuWXqdqJi^RHc(C|dxDlFCk-2az}xP3rZ5Qcm!Y4r%89Gfn8f;UNF_
zmi=(p7tZkxS95c7FY->(-c$fwDpR6uA_hk!a@3+k4TZZ#nN?s!WUcTVj(=I!bA=c&
z{O0;snpBE`3?#flC`zYl;5@J`!F1tP34k2(s+fTXHk#E#Mq-EEk)R_-j(q7a;Us`m
zVp}#-Le2v4s#EuYb)-;9n3DZpTA`GuFk+^avjcf&!L!6a6CuW42hbuw|H8>3h3=zi
z&%xx&o5b(&MOH2hZZIS{x7$ciY)S)A{>+;7fTsa}yhW~y(TIoGLf8RVvHyB6iLwb4
z0Kw8mYfE~?U6wG6K*M5hJLIon$KX|L2LON&IeMtA=;A7#C-{$KkDf5k^K30wk{F8S
zO!4(5V6fixL%{{m|3YL@5S+i{o)Vjv2&|wbb;;=uIl>0XXM?-c!p5b)N;EKJ<K$$P
zi*6}Ob?5diAUCF%6Hbh;zrdTzsJ-=59r7Js`XYBme~XNSM4z^imGZN})qu+dB5B`g
zh7wq1rvK0)ns1JnVg!UEDs`xC1ST@uq`kzn!NOXYnv3ay<gS|Z*$x`gxr-u5^30x7
z7IYi^i2@gJ0Q^hqZk6pMhJ}f{e7Lg+j|93gdaoD67?@U#Cf>iFf<tcy!aKrafhh97
zf|tWC+@`Qsa`0!rHE>;6n9tA@@LbufH}D(lPixK*$g!&eU}OqWLw~z<)T}VZ2P&;3
zB+3H*J&rVBr<v7lQA9iyCvr_*Dir|vVY&dOfUpEB0xchaf~X}!>PP}j(I5In`4Qd}
zs9qBe67p#E%c7QyzoVAy34n5T-id+EOOJpCaN^R)f84Z61zigO?q7`sQKBrP_JEQ&
zDjP_uN}9Wi1%p+K(0p-3UI&S3cL#_~%vuaM#WN~7e8v<3jfRcf8*vY#&hX;x@cpc4
zSJB09u8Fbn*JtOSvJ5fC$A2P%5sfVYy7+WjL}b9=6`e<5Ql|A<9jXWsAC?9APd%Km
zjVvK(F;-ku<8cSSIg{yHMiJCjY9SP?dwHo?@Wlpe*3A~RZ`Nq%b<rKC=r1TBqG2-a
z#9Tke-9S+&ln%gy%-R^`(8^!Pq<tEv;{6*BcQCDyi&gF}kyQg8-N|#jsG^XmDZPFA
zr$Qrb0Q^)S9j8C$02GDLJwyk>%B7VY8L2CKxbJ&soEDQ#FbmzZvO!w8l>@9-1I!D8
zb4n}FaKJhpin!Kw|1kP0FY}GGq>o{UQVQoJ^ls)cSx=xPm7Z>ZRVCrEAf<E!a(ukK
zUNlVqUG%raU>PBNtozehKUWMWYV*a`lqFR(hTsImNhHDU`r}ndr{(|Q0&D@C6Qfx~
zX$CJ>eybLmrX#(}1xtrc`;Aje>DDa<l#|_d{A$XxioOz9KFvqqy`mk(tZG5a3K=7v
zpp4f&4iQB!MD+Gc3~<)stbpyW*!UVYWXq>+m$5ehUU5qjM^wB($}TRh&vphDl!)0@
zJLJU3LawykMt6`sD#jN3Dz$*_=~?If)-kjE<2m#u0;?9pv8j2kfNXwrk^uK@*~l))
zYC73t;hYX<&h?<xMGPfv$F;X<930sdg&xZvVwwDpSMBkl`{R_~_EdPo8^v=_PQ^{v
z5CPZ-apeQ-MV;yKSs3~KO7=rx_zlX7Bx|bo-Xf5U#?a8FP2NqGJGU=7XS!T)86-u|
z4qmJ3#EMamt}uk&2Ch{>+oYwPR~{H>i=jq6HudzG0T1Gl?TH^kea!wJO!_Dy0p0#X
zC3d8r`O3VBIWIWhq2(y(C_*S+yb=vFizJ=$-?-WGLUJ{5vaJGVMikN?2Epn8-I?m1
zPvM~byPlI_Y{o8(m9dVf9$<~aluNV1Z<g()Wp~-F6V+==k7MI=YPb^uqbiv5m&86X
z_fukyjH*GS;1yICJS`w{8KkP+C@QUVZ4=@TFeefI(c2#IZJ={LRm93oYeUfSQi)Cw
z=OsHtSIlzCb7eBqm_-+bLS(QHhs5{&Bpe$WAeFKEHBCJ!hzO8l#k1DWu)&13&2A&b
z-z6=d=IVv-7^Jp#!IsJn62m%)c}}x0pXy!W6FQSMqkY!=oEl-%qJ!swk;UNWDA=@M
za!`9EVs@bX`X;ps%?}mKijU7I2nlk8=n039_y%5#VW33RB`mRw&v;~3BdT+Qj|`n;
zjd)5h602$eyH>4SS+4kHy=cV;$mGw!tSV!{q7^)7f33b(g>b@H|My(ne_ISZEeK<-
zzN!(+*!8m%1{+VgHAl26Y!npIsPD|*BwQ06gz8$jz35&&OJb(tHswRCE>lu@m=4`X
zpusA?bO|1!x~|B}bb{i$Fu`Vu!Sb9e8|a~Uh~us~Vz#9x`m?A>O)#AqttT+Ush{Rr
z<;*F*A{cP`?rTYXL`*s@=xb4$?kYE%PM(<sfC!-n%#|!ImtnO1+#cfm1Xljw?K7hH
z`zC*=8a)V~BFJDWzx*}TTPcC;&-+M?*#?WDDY>GD*-6zk=5=v9tr+09j51_Fa1Nz8
zm}Fv3duGZ%GeUxb%%ki7mI!DYxjp-G=&n@S=@HhmBvdgDTF4%r>7S-{?s}-;Q-zjK
zpBcBp?}}J+ibnmdBSx;V!4(L7QuR^)41Ed;nyd4|SNLg;a(~L1e90T|?&a6=l7;B?
z@ma9Se)X^Zl4u**xkIk2;X~2bQzfP>Z!gF0!8lvk0O3(PuXlVd^OP3qD~>XCS7Vji
z<6d7v&ws}giatTbwxkyQrc`-xt~pm;GZI^Lq3a6ba86?;JyNN-tQHM)Or$GJ9>2bF
z+O&$JKV}ab|M0(VjB_3wxx+z0Pw>z)H1!{kIs2oNTBi-&FN{BX%Eugk-8Vr_N)lo5
z2VQfZ664^!8!AI5-8v{c<u|Tv|4A_oN-*13q&^U34bF8&usH3a!~Wy5b_Ymf+R|gW
zclhF6>KQp*D`!4VR{62`Lbulg7g)}PQKFJvd-!#pjPtbLL`LCLc8K~2aYasIJtZRK
zzUxb3Zel0fHr%RiD;ipAhp)+9Yx+2~oQaO!QvH5P#aRMX%BNRd)A0svQPLdIeLW7U
zb;a4*JnaO^@f~zCuo!knGZS<i)|qb5|H}6rlpO{^_0Q9J+T-iI_nz6C!q1Wt8((U{
zoUf*{yGo%Q6tl%V#}}P<x(cTVUzo%)qT(#UWO?Z*h-i)fuH`le&u!Y3v!pj1OFFgO
z+7Sc8VyN9wp_g;?lN4`k_J;p9HX*r>wV<PsvkU#P`}~~;4v>8M@#z?1l^aA$B`8w6
z4P0a^Mfp?cIXi4<8#Ote6mv#fK-UMrxvya9Bz>eT>P-{yzB$n(?(U?1g5B#^1O7Up
zqM@AI(y9CrR7a<~9s1pWt4euRZ>OT9%F`0~LGM{5Ya-^sq7~kDcoNQWwHj%Sxsd&Z
z15Qz-F@uolU}|i`qta-Z@f$9F$0?Gc(JAq&lqH${Zh-=vk`r|vaE1X_>lq6mK8w24
z;FvvBjzsfw%$X}mGXZ$`iqH5fR`g0}C!F2|L9LX$Vz4^XY85?x@eczWM<|*Tx-na3
zj$b;#foPdrpbhI|R8@uE60Rjd)i&f0dYfooFc{JV0-uxB4z&c9=2UQ(!iMKM?);hb
z_4*qpPx4MFC1H*gdJ9C5yE4_`v&HHYVx~ee1w^4F!T4hHLy@3j@v$;3V9G-nxdDQL
zw{;;l>h%0O)zkaq1hXH1<wf&UB17fz<Kx5wZvP2ueH&TMif_73We-VOE9#$1N1}Gr
z@ArZqGnfN;hk&tkF=jD^{(mQ#SXpo!WBeREgbb=&J5LWkDPm*F5u%bwVV>>OVn`i%
zSNpHJjwO1NZjozGmzkCYOcL=Wb|h%GP{rdsq??b|iSE4dIxq*Rk9N4|#>2q5^c^6=
ze5u-iQ`74(cn6bO8-A3QrWs-A-{06|8-E(0Uae508FIAZ<2QjMXGf$}b9AL7VgH@^
zP``J)P4Z1lR#K{hM8g@u%O38PI3ogf5cy1o;OG(_itpDK?w)=w1q9+<&9jKKM9P&e
zspt&^R70(@?3N{`TvkbRqQ*N&7Zibq{b;lGT_$2IF+p6Js|sUUIbfpd#JW^ee@7uA
zVh5i=<}GDmBSaT>yxM`$p>`9Hf$}>o9R92x?E#n#1@K~cRFGCFO^I$vJW7cFy<Zqb
zvaM5bgmg`4zyk_Z|A1kMZtX&7H1vm*kC?!OUw7QrAK_TbpQZ!jBO<R56Yy6OKxe#H
zmLC(?3ys>oBAg;4HP{X0wPM1;iv-(~AIwOY2N8X9`bzu<*#6sIa3j(W&In;M<Ok6S
z5%^ied|mdKb%U~Ir=&Vi#(Q0s3#Nnl6t@35`n;`|nSkrNR8(BR>MS|PG@RMSb-ICE
z0o9cV3Lqhf0zVwG5XJdSLYTsHFX0;h*YH{*M!Y5XA5UK^=FRq3iR7UL>4J|F@>;Gc
zIge;WDIOgCwXV0IB2hB`L!Kq`F+1m^&{N=9=g|u4y()nmdDZoXnU8`2!Km=?hrmr_
zON0yL;M{AWAE58*cTlJPUd+;VivAb&+!0Q#<r4SVo+;C$mQsWhgUR9q)bl7>iGKoe
z``<$qZRVUG`9H!t{^vvfZ%=?T4;~^hO;*f9!(3!BqwwJ_!q$yQ{%<8iVF6ki)8Z<z
zqt8a6#c`>gw@vE1HyG^RU}5g)9`x_85GcJzi&;?4&;N#mazIyBqM`NW?fg3m7p@nV
zURm>OcTm7%?e;UeobE8AZ}hGm?RGv<RZ#EMJ7s#nS%rP7T}Lbolj^$U++X_chqgo;
zbneyeV@iJ0@{`-&C&jGPxPAOkO{?eYaNv%wdEfG0t;-AFH6or$>GI{v^ZeR14E;~>
z32017c-epdRJDybZTtKG{PTa-#L0OZe=S>q{Qxyk(|u2Z0Bm()P=4vC5hF&ZsxFYy
zKW~rN=k)2*^SX7}kr*BxP823DCpTN3>GK46PIj}W4JQ-*g*6`4c?Ncbgz)2+&9d59
z6(Q;@lPA4|-JNu$<gb8);~JX?ke1riAiyJRcx3TILcbSoeVSRn0R!AV-1)1vG(3E8
zWyjk7VjwGc!tg)H*ne;cwUh_Hv;YSr6YhzGgGlG)%a<1~j<dD3)zwYIiEIOJQLwt!
zNB;+KA4tqC*B^H4*RN+jFSUg~&CP=I#q2l8e%ji1T1})fzJLFYJ=LEzi%Fmph7Z3B
zVw#^%W4Cks>tBEU<s-BI_;E{9)5n<9GeMZUI&w@5ppVQ<vr)b!m}jyGjoLs4WBlY#
zeSQ6}U)5M;=wc^^4Eyo{4roQ*u4eZ>k_ITw$lTn1eX~9<H(|mA1%-$EhySI&h_?X%
z=}8+#qGX5$XWKl%S$q~KNEN|0jLoj&gNPhL=Z<o_(TbXbN2~w&Q|W5|9V7gJ_?ii)
zR?QypMid{&Me`1I+8mF&+w|tTu}uwB`p@!8<wJ)iefaPJMWf!PeozpFpNI^&yUS<T
z@ZrnJwnSgsxpNP$>d>>NzbM2&zkr!GrtD>ole07Mq-X|<UvOA7;n-XPbSpTrY`dhU
zx(|^K*qoqyg1x$Lg=nWrWuKfgNpF9jMQW+Qt^D`3EyUv1?r{QW;1><&&OK}>dPo2K
zV$m-o)7?CiiAmrUVo(F!(fJsf&O@0cWC)#mHvM4thQbDj$Yq*aruSmhV*IavJ{6^<
zH6J&GBbFe>r@7x2%tMl*3ksq^1piN*f4|1Z9@paf$jDR_7cWn6eWt1%yyzIhXAaF*
zX@4KUIZXm+B6`K~<BhKn(HK$q;-QL|5WLQ04x9WgW&4d=Z;Oh&i74DHnbuefN*>U1
z(BGfb>;dn%=6U=--jjMu_{uV<CO6?CxX5JpLQcuyZ7#*>9So}D9RSZ4$F$x;KXBz<
z@A4tP8X8<z@2zzu={VFpsj8&JiM%!CQH{TUc$WQdP{jJ{x4iz`au2?9XIiGt@aB_<
zfuNeb3rdZQiK!$3=&GBJ&4ojk71Y_<%8EvIC+L2e-t)<Nz}l%(rykusrtJd8t<}-e
zdYzwdl5nl8%zoDfh+^l?wE|zh{%O@!4CV^^fztERgnKV9uRM^Dkl<r*0Z0IBWc<~a
zk&%(nld47dw;^qt>gv3~&dm^LoVJaNRPbfVRE8SPQ}Fr1e*0=4gW?^0{CMnQ+fAp|
z5zfz_KfnI^(nLES_h9a*c7^7YZ>qK%VxwbXv{h&1=0z_gPA-4kS(*z;*UQ$Mw{Kha
ze-G)yeqj@U%I)_HYl*D-Z-Ufo<mTz(@@wC|eLHm*4k9(Qu2l?OeY|~mNuunbb`p<8
zFPy(yk3>X#`?C3iDwPTyujZS{(cmy0ZtBmJ$^bWirL6Sw6A1v9sFpB(+<EL)Gq9qV
zwc+Pys5EBG1)}pX4UKa&jmxFCEs$+qcDD1DWoxlHKcFbr&#H^f$jr2S?&aX%uw~2D
zddDt1hb5z7j*pM838K))cutv%*Sv3WcDoPJS}<_;w~C6NoYcINy55*mpi~QzQds6p
zavJem*Yo$5*|YbRXg_v@I1xQ*(}D#HioZNd<@VV573qxBloc#*qd9W|L&TSKxY@Hp
zgs}aTuiIKpGYe!cHdqZ`wQ}`p%bWWbp4i*a&>-fX-JGw=!0g78uU@@^D7yS^Si2U_
zwxZ~UPi}H-_54d`f5LD?IQMFB?=K(NSRvAV^ysojFlR^LL$Cn-4j3wFX#9d6d&{%w
z`a;of^TPL#$Kep00QZL!w$RoXhPc^>TtP65GU(}+MZzHURCUTILF!?>_qSS)NrHLK
zA$ms{is&8z<tqYajGG5VM-WrWQ?>-wInV7@tbktV)Y9x$*`Y`CJ-Jb%gn+Fkh_Dx!
z_VThgxXa4w>Mvv+|I+tS@D=oGFsv^(twqCXUyCCbC$bbEf7wO^?X5jiQoFt!A`-(>
z!*z6;0W@lYAl3^1aHsnt=OvpXAqJmYeR@1#AbAIV;yQK-Z3qPK#b(19aigmG>_XZ0
zf9P(hrK4kIEpic|v>=**H)Krq_DmYcVO!~g?wOq$yyv0OWiPiTVKhhr36%v?_|m0I
zLK=EpJq`o?B5&<)0ZNmYsCN`f^D&09_X<vM4FhbyhBCDTCM<=TIEzMt>~1}KhE!ak
zm*Y^aE$7}}ye1(fI56Si<CM`6OvCzgeX-Qet^`8jdjFdr=BB2m_Bhf1os!^FPu5;~
zY1uy&hs)vZ4y1z>Il5R$U@63Y+bC5~-@A1EkI1EvRNs(KhZnARRGzs5Hka-9kKfR)
zb>KJ^Xg#V5vx2+qoZ$L>Sw{Qbi@Nme*-FMh-NDY%>QWZogxX!S!WKuzSJ+vgfi*2x
z2(}0aNXZs=OHY9SNOnzQlMV)*l9t_Vx3!o<&m;o*KaId2piKC)=QUa=KtbAUuB<u~
zHBzcQ3-1ClLx+0bnmc#y;NLs8Z+}jskp~sLg_YSdMz2xa2!+RnlZ#@(_jyS}W8))+
z)Q6nlYVwtDUe@UPG|rx*kta?W`SOF~uI}#e?rw8Y$fNqSFQ)JUk!a7M?9YGscd38k
z0|bBV%w3vzqsw}qo}u$$7)I??ns=JKnix7rV-f|)uIjP=)N;<-E#EC1!Ijnik(Y2`
zKF&*-d9kO$cs(BisIlL!+s#IGKN#cO_@eZm0K1~Y!E4zVO<A`rq70T`Wp>ZMJ`@T5
z!%G9WpkRLfEC(uU?jD&NcPbI&3bJV8_4B)T?*>Mu?c&sl6Q8_$mjX2$*1woeQZO_#
zFoEzubS)tk!G{~on@4IG?eA{_icfg(3`KNo*yTSRbV>1uzJY;%VZ*+BS-)vhHJRe^
z5QDL1k-M*%<SH4AS5a{U{pIt#zJHi*WfdM1vzoP~*7}%ay`6vAn9zd<yVcJ=%Cv%)
zFTY>_7*jhSyKw%a#e^|Tg=7Qz^zPleUq6Ajq6p+FHZ?Vg;xo&Bz4|lIgwPi1>FL?o
zMe>NIrUL$dzUx~3zDOFVW{BlWc2!YVSD!jn$n%gSpJNpW65!6P`r3Krps#O}=6jml
zfjTQ+{y0}>$`k=fBlk2k6A_W$yeNSn@LIm<qW(f@(Kns1-th9$bd_mv6w_J>J0-to
ztc%0L&q+CDL1s(H7=6QV8?l*aR17MUAM_Jaz7tmZV?G!Mko-ps?u!Q^+N8!NWUn>p
zpvLE7*8jc%QW?8NOuwDVI9a3XW9FaD;j%?HZ4!f?2n1zneqBN#bc<dDXGxdz^z}t;
z6;K9a>+t--kt?N_cOggE>*Hg6bMx>~qZq)Lo1HCkEPhNB!il&pt{tPUE;<pW5Vyj+
z-QPzZzO>-c@Hk6D2QigFlsa*(h}#Db8uXcw5T&K1)zw7=Sp<0PJWjqTKSwmn1!xk=
zoQChqot!S;xZy$;CNu$wD-5(FHjj5i>XXrL1;Bm1OqZ+F*DEM5B0X4V>CifI)Tj#1
z4V7qrUtiw?2N;?P89hA7;PvU5oxHuh2ffK>P9MHfEDyDe%8H8Ljg8mXdYrGX-1&b1
z6og!b!$iE|=;+8;KSGN*nBJTMt|WS>H7j<%5D*_{t|m@KM&@kZX>Pn2@#`|zs@0a)
z7DZJL;e5H(ORQ}ij2eVo;nScdFvgV=+MVKEL``zWj2WqiD$B}fV{Vl?@h-e6(@TjR
zh7TW(jeN3p#Li)Q1_mJRH*iZG9J)6wcC7e)R^rr5<xZ4tB@L?GDTqJ$ret;Vl*MXX
z?}?VNnwl0;8B~y(8XC_1O+;^VW8;at-=vJ0BSG$~KWVRJ)U58#Ek^m=iQKOkw$}}s
zE$9uN%wz0}eCuL?;B5UQ*ZSu%%RO%l4&=dY)9p-^;n6-+oZa4dov?u?KT&bLMrFs3
zA3q|yE8kS`O-xM0Fvic1@w1Vu&@r;HiLmLqO=oRQ;cCdvocE{Q&SPQ)N#4ZdMzrE>
z)BK1p<>g0$gI6RUX)cYbq_4(;^=I<*C3p7f-`^oQfqn5Gs+-F1bZlY}%yA%1ZuxX=
zid^x(hsX{ZG}gBSq{{ND!|9DLo<Far(317=;RL@4$W{0(Ro9p<JH=iJR(ezV>jE0V
zpY2gfyOQ~Z3*H`4HuG5X@<he5uSDESuUxM9YiWx6wV5MTAq(uXOB)sZ5#zsYjRzK{
z#v|O{KOg_0_SC1sIUteRiQ@GOW>I^rORl~}YoS^+kNdTyF8<6%i}usOu2L7iq5YS|
z>z_Cov&9YEWa&SDP3h95i+L`$;8Cj@ygKewhfbZ0no!-P{@yN#3ckkClUm^5gLI!e
zcJJ%s)Av9{yS4lLG{!&aJl)~a8=fyoc1=uLMKo4cPI9Ue($v^^%;i&Y@oB#_PAkqs
z^}Ww)P_*YK)%Lr!JM;bf=9HLK@vEkuCUzD2tgy;g96d_8k0uA<38=%Z>$v7IZVSY?
z|J7xgQKOPJ7w`P-9uv)vktE)T9<qIau2=GiJ<#SVZy&jF<Hq~>0X^zaK}cItY@T85
z%+@|=((-C<VC>rpor$-Iq4?z7ygYS>7`QsVH-4ee%4z+LK_L0MxbZqCu+i(0o>?+C
z4PNX)<=0b0>7b`<&W-cwg=89AuQD}i^ERv2BGTXIy=hgGy!mHDXz00bH|f`+qNes-
z=_>(>@p~t7<7cIP+E1rtk&k@bq5di}5{refyiZJdol?5bo;@;7ayxdbYVC+Tw=bqZ
z+TtBXf9<TX7T4Dwd;hJ>I3pvYyR6R3n~GZ{&7Y)sdpn=-sP(Tgd*EA1UnIp+iJXgn
z7H+G|8<Jw-Agx*}waw%1iWMuyd-$Oc%zE_b$@=?#2D!qmkTK<G2=sTVd)@3SJ{(CS
z-Le<zv7`UP1rX_V={t)O?9HMt?}BX8k46`e$4h0MZ5LfY__n&(QL6oEfcuVV>e5{&
zgz4++qoKB_zOcsvXVU#iQ9mUshnJU5`}E$1Mggl^6jG=Zfiv4G0Y&7=MJp;3TgwHs
zKz?#J9aX>Ys&aNLb}U)-gVVizBO9s;M%H_gVXO~%z)MIEsD+n<k!TlFpjLSx&7<+M
zb>y{xE3_skD0q^pqoZ@txK!v=pHyImK1tRr+>pE2t^Q%rB;N&Lhrg<gN}Hu*(o5I9
z?xE}Vi{x~~Q+4fvH@<tQ(b@pG9=gtb`}XW?n#`@7(~S9=HEUq?4ABwj=-a^+B_)xK
z?aLF)*RH)VUF&`Gm;qKD-&|W4B0t}pG?{AKM>V_5s8!e9>(?4}Jf6l0bzdEe2ud-2
z=$qL~BNmTHYmORSq<ZVr^Dc?%hZxIrWK$<|B{{h=LIwA@^V>VOw6sJji18W^a%riV
z{o#zLdW}mv{)xO%DJT1TV^Gm_c}Eu)4c7r(fnoNn^4^Zs^yzzt0@bpwmg!CqcLIwR
z&Zkg-0b2Jnas1pU^i$@Er<p5rcfv@6VTb%i?1SujXOuviOiJItd?LvgWqqOo(Ycyf
zhC`6MmBHq5&@!L6nqC;z85e$D^X`ydy(Z8_X2%W*!VWG3>8lXMRcPVbV*wy&%J}tR
zrM)+HKs|rL*v>O>sG=9Qloag44%2(t+1c@;$oai$vyU7;Z2$J~A!XM^i;lYaedr@^
zDCV)EHy4CdAayF;gcQ{TLx}B0Dl41x+KrcH{-}15PuzOlwO~l!$b7GQSz{4I0|&l@
zG!8U`k!<ux&C?0K$If=_*8e+E`?*f_w!{|H-zzd}22JR?p8fO<A{Catcpk`1R>#%!
z_arSXci3$1nrcQu?;ZA8>r)bHLgGYJf1R>NpLqo}3D#8D@#E`ES7Ta1)Sg-WCAxSJ
z?BDQM^cl#1%H+ungkE94mb_fNYSmdEpPj=TloZ@i+-0QC|G9GE?y&Nsqw6KVP`~<=
z&dKPqb2E<uq7y#ZafFtZmWD=QW@ZdQ3~#lzC^hBo*U3&ovglGBQCCT^({5wZ9WwsX
z%F59u3SD>hl95@}cqGDGsZlHI?%lf;wSMIYBv(v4re`MAUSt4Ng~_sRIPX&*r0Epx
zmMml(lnj=z%M@3|QqhrQ%oxL7y8(a*anM%BV?OhGypY7?gJ=Y&&fc|ir<l}^jF4KR
zI?5+4t>0J<P!JYm8=lgm&)>#mA{81lc^AQa<<m*^@ux~upPL;Clw~OZ(kE86?^8au
zBK4nygiScNr8hQO6M+aB%IMMbeQd;?P5d=Y&n%oCMSQuM^5aLZyq?i?YwXALJ;OS~
zZN#IP2r*HJ(vTrTJc%BtrMzhV#Ke(jPoD~*vY7@&!@!p>m(;F!vpC>I&Z}3kwW4AI
zD8$&+^ULB0B<kRcj~ka$V+;!Vnivx5g~jAvdcAg|rK-HWr<O}W&_6t_N$)bk3nV(3
zHJveI{d@QRK|xcd=F4Vx)qVl4-;yf9=odBIby%9_mExXv46KP3b>}v>o9tA!@77}v
zUrIsE<lp(U$>`g6J`P~$(32vK{rDs!eU5T)P!Nkcs`{p=){&oGzj!;o<@+oH1Hg-^
z6DC}`Lvq+LFR%3P9ecbVJ8W@raap{$5NA5W(UE!S(9qMt*=jdD3SWMVIn^cVPu{+@
zYD{-$;6T{OJzpjc`~yTo3!0Gpo;Rrs{29ai_6hHO-Xc7fzox$Ijzr3_IowAL;#mZq
zwttkhQ$LuQnUQrhqD}16x3Ah*(IxLmwfy46i*s*Z`0Dzsq_h+^mi&}!94f-jF`pP2
zdMu>>#3qPu&%Zt_y5T<Z!1$PB`&tPkYu}kVtNsT5Id}8+tEW%7UQ1s;x=>K{sa$2e
zsLgUbE)04$ciWQ5Q7PbGzmNKG^DQP`ninVF-3b@hGZ6ffwMm%j9P3&ZHp~(2X7=gc
zUVulRPIT<lY2Bh1gHKsU>UG&WOTjm?Lhgd7k#opx;*&|s3H0+_`iIzk4@pZ)+mhh&
zk3rOdNQK-V<K5V`^`6ofi!T;b)!dEY_-y-YX@K(X%|z-M-d|ZY0A8_b-bsSO@M%&)
zvxnTFb?RwF<cdEb)hYZ84G0*PCh;`RQX}`ydeP_Nu)Uw@Z>{<tKg=?rBD8o^rU!{e
zpV1#M9<dgqm6bPf?Ld=X0t)<2dlGA(cX0dm?a*AOnT=LaNhFkh;`dBBN{r%WfN>-7
zUzKrPBD0~^eDEM|P)*BW?P(&Ycp`?<`Z>+#?}5|8G&XIF=c+%;w{9+4E_N=g>9S}b
z8q=`j_*`1BM0;pF(^*5RyC87LqVUca2Obit>I-8nBRLIWCex=3d{=XAGZymH#_wFb
zF-z1vVBH=aGBop+bEV(!y44ep)p^a8P9;r-m3Pke4jH@D(r%x!M}xkRooL)#d=`j<
zwaULy4r-izBj{}Z%PUpg=gsr`?KNzdrPmOoT?#_=py1mweGS-g-Np~OIXQ<r>hb#P
zR+&f{52)(`hA?`e=*aL)`b*F(jo_HXc-yp%Z<nS<Ukm`=No{n=EgD1of{+af`@t1A
zfoD}id{({A@G~$@effbaRQH<sieijW(o(7nba0T8{&IX)5um}Rd!yebyG<(UuJ*Kw
zdM&GaEYj5L#Jw6-cSA#8j@_-xBv=Rw)nS#Aqfj}BowR2a26@4aH4|1{f=)o&?F{ZP
zpjw<p-74~a-;VVUlA*eZNzVeGqR<O~8hvcNb&}JWBbkrZ;*#fA-@uH1`fTYwb~(Y-
zi_^Vn;$faLKRAnwKwjTCSL1>O{$b`W`u^>6k2!j9{<p$}lhe&V!sV;6m_J2Fr!08t
zo1j^X^Cy|6d3kCl_mXj5G|-ds_LEe}pN&CzaXW{>!_)s^dpo)ymm*QLWaH0mQnh!g
z#!E6pff46M=ff)M7F~}ikbko97Fb+=jg?&#$4d3;wajXghQ?ygrBbcG7CO1O=(^QJ
zS5)wJM^?>nrKyYRXE}NK^6lN{PVMkmIjz}lu>T>0@p5wan{AbqmFr!i%7C3QdTNE}
z(Qn_{>G`zk)0J7UBgVZlopcB;MSN=Q9IxWLHLABN*68<YZ(@?qk#wO9C6@oHL*V=F
zTA1*~tnOD!pgYp;8~AwWg$rtX^alvc&{n4$;H*_oNFaJyQmg9GwJR>=&?>_M5+x9a
zMY3a?m^-{v>pDEg$=dO))v!nw87x8K;NjixW#H8vyPLn4Z;uIMYe{jsg)Qu+a^Cv(
z^=D5n$sbbkwZlD0y^%>zc5{j@Fl&#P#8m1JGp2|^)f-mU6LA@;sa*-EIm^NC_F~3d
zL&Km^9_Vty+;f})jiZcy_1^C0CVjC3=2FHTYm|G4Qf>BRSxdjAr21EPFCaEJH1n2w
z{RP#zbEasdutc=MHS?&OJ1SV-ZB9zt1+nY11bhBb!Xjc;;oQG&CnbG-nK;wTnYz!X
zu${x234MIS+}G3r|2#AZgPfqT=82xQvE$Y<K!s=PjXe+9^DUnO*Rk>wZJy?AqczF1
z?Ws?$79=GlVfzY?J82XhK6;2jovO-58=9K3Lqh3h6tL>a-HQ@q2Q1aEb~zzWf_&EP
z@Rci9%A@Q$JP%L5&QWq+p5^B1dM<yYU+G2FTg{PS@29O;Q8Q>VB&-V!o6KfM0WrNl
zRY7PK9fpUVzgXe&nk^qonAJBwJInSA?Q8Fysu1{O{Ho=z7yGOJNr&&|FxY>*!B+yM
zC*HNzYA)PS^Of~d&wm<6#fJD6SeZ>TRzAcKBjb$yPl}3cf45xKITWAdI}0|R^w?Fj
zKU*7L`03-vV*vrD9Of$@if?$I3R%2j%t+tZkpy823KhX`@JP~4V$Aw8wKJzP^Md%{
z#y4@Z<91!pEU|Pxc%Zbu@~su?>Esc%f%>Xt>ae)jSkcn(OzZ2=S&O|#M0nK95!Y-a
z5wb$-5pxNNjx%P??EdZU`Zw3(>KcF&9~hHy-9%^L-dbN*Cwk21`V$S6PdX#zXkMhG
zsF?Nh($|KagQ^e@H%<q66A8burQd5Ht-gKw<oMU7yu*0?ezL=%asBC4oWh`QVKn19
z;5ffgSlr0_Z{EDAxqalhm9@3>m7LY9SCfQ1f4vPlxxv<Az_w>;0i#VKE;epo1FnAb
z%Q^arK*0{I45UI4?{<)anqupGuS13tfiyAkijsoU4hC**M&xo@!y(Ux>R@!^@iMHv
zAWwn@Q4L8*sHTlM(|^<~11f&F^a{?3PSSgWAtXKf1h7Sr2MW5i>hay=RXau*bQ&;p
zsO^Uh*uhx0W2;%uny8{PM*U((WvQJ_Eu;V4nTQcTiwu7INw>dS<I_}DX8FqL?T||F
z4V8vXd}Q?7_=SCEYM*%KHrcu2;Mb?y0ApW&NHeb-4&r)rQ_u6Yy<}xKT0K1)HR{Ae
zk~ucvkc;Z(57X8-E@(9WYZY9hE6>~%Y*Uk2Hggx*Zc5vhMiR;U_WOisfK_+-LLddV
zdMD+b!=S6{g)Pj_xp%LW%CDKX8DS>+sr(>MT&{7Vogt+V)xuuaqiZiQ$Pk0AC<z@-
zMoKDLfi4iS?-g%v_5jNVKl4pTh1Ngevt=lyhtpoIA!8OArry~RMMu+*<Lm796Wd;B
zYISpSqd8K*$&WblQLSqBBkK*8{88y7s{yO;&Zq}QqhYLByN|0barpH5TC8J-HCGQ0
z>r6%B(y&`2rnmGF{qlefey%mnz0mJ%`nH2le_S7|w#LkKdM~D|gl<|uJ-nh5Rh-4m
zvC!d6nnNmP+r(138Dm(tu~*_-P>iV1-cHoLAb};08M8*(LNmpF_o^hHwB5{^I@2;d
zu0?I^Ih(P$KkJ`pgz3gnVbL%eaW#V?Ig0QzO8KChl+48CmabHl9@lTp5j6Hj?l;C-
zVZ9kwfUCRV*wwGc#L=S-s%;^`QHZ{aa}*8lG4qS3&N)y<7P!#rq&CPWx#1aIj8PhG
zo3!Hx#thi=JGiBK<j~)XFE#skjJ538N6W7~<}`JTIoY25JfD07ba%qV`Kmd^ncv#7
zrc~|6m-lWDT^{ckZ(#gM2NB@?4>M9NFdxV=9K0@m)1jx#yiYnWof6;X^ggi{+*fhn
zF|~C^7lvrj@#NGz6+w#!7yb0;{rmTyRO)wToY)I=8{4(^s}5f7e8MT>kk7fjV`Q}g
z!ou8}8yt;BY}pMf3nOb~9iOys-xUWX@81v@UwPF(bm};z)AU|6oK1tU`ERYM=OFdD
zSN=W~p|J*qOwC*O*|Q&ld~XVI$bEKVZ|3PriT>HE2ab~1p^|(1i+oL<0WpW;eEiJy
zWZok>GKyW~2ol|f4C{<=qKBlUWRD(RP}&9h3V0h^7_YJ_p`VVjay1zvmp|F5+^~&K
zaoOV%t%`2dy}X&`kZl(1wd(e*TcQkTo}}O#)-zgu=%Fj&*XOrRyJr8S(u1m40vOwq
zlaUJZ!c?<r3Ia*K9_8jPlsc+xYolU)P%lht(L9wFBNGDF=RP&tSILt&*66B*ls_A6
zIld}czkZZ;B6QAyj2|zjjg#U_iF}VM>es1f+$V(q13{>m(r*2?l9G}t)mt{usS9CQ
z?yg((u`r>CGz2sv!S#G_aIg!&x842DwI}!Q-%rQWkhB!udt<Zp>hZZ29nfEWy?G#V
zPQ(bsm;T;v=GT_RcG5l!aCcbldDP_YbFv-!IdZ`o$To6c)lx*nX}0OfNS;WA@JDA}
z<%~DO8LiQ~J}O(gC*2%HKT1K6uuTJo3k?byVR8tF>ex{G)vIx<pX)cT*I8>`SximV
zT+8O2bAYT-b+NZyC9#C$zs+6FF6bjIeGPPk66TVZJ1e4P_~Pf=jVKi<9CepzUrYeo
z`0b(n#4K&iAP?=HVegKMJZy5eBr@a+U6pN;t1g}{A~{)5G`OO=TG@6<dJ?9Z_IK*~
zr;i_hL$Sd+bU8J%TQ_f(m6xB#ZafU22sP(?Ey&DOvSiPN73@B3Sa67wTHSuvt!Xs4
zmz}+i_g&<vQDOAX&2_MQj~yz|td{l7oqnQEqiMU`jlou_e~&2B{cxZnuH2n4tB}Xx
z5FBgOVPo8^uE)l%!vTQ$6WkNPj5QLatNffCQK6@)<Few%>ZLgcxOoHp1my9&zrBHf
zdo$&2_BuNs%pgGi^6vI8_Wn$ez-Z`lMX$1>>t*})jZ#vbt+uuWevkI>;nQn1;;f(}
zJ@*Ow+pN6H0g7R{Ge@R<9cCM5^7I>6g)_-IgqOTwY9FZFqtb(DZeylo{Hxj&$e;m2
z24i)Wdz?N)UIk=oe((#8+meT5I<8~%X_}}yxiIuOE&>aLIZ{~rttidbFiP3_RvkO~
zB#i-^aY;El#;aEUDm>oOQlzI+?@YS4X<_@PRY$v>Y2Ld{mM9O-TnLq|Pb4urhrzTI
z!VGYXox=`~vzRC+7foUcZT+JB>RzQ0yOvp04hJg2^WlZBY*yGpfz4*%yQ5L7Y!u%$
zPo{X|Sfw);JO>I_%~UZz<KXIyz^6K~19Q})8kHMgcXSs?oLiAe>WNOd4~PtWmsL(T
zmdXH_j?ES|6@!!e#^2anJfptS9cx(g##Hj3EPG@+3%t7-k9x`nN+}l-4#+6mjI*6|
zZ#{KANGhtjVq|&wXqS2&{p<5xWM%h8{JrcTP~WL@4}Wc2>ej(>RN+_;2rASZTi9aK
zrP0dyxw%uQz4DPjF+&Cq_9%PU+rXgDph2R&k7(p_v`BVi^T@|XqmEX0mvpa5ky|lx
z>x1c2b&6z2(ia|V4bySE+?*NxSH{gvlXevry?bX}J33rGU%z&0okynzhglABGmM;=
z3|vEfl30Z$IWVFS^hyJc5~16<m@WO<PoFY2c2~pi-%w~P0V0N~Is}aHv;ZL>RuZjf
zARB)g&|h-+jFiCWu$jGK|907W+T&D8$q8pVHjTdOaCRy8Uw+|)B@d_9ZaBGV;}6?u
zQ%~%yxM=!7&+M>?-dRn}UCtI0duQt2s1-#akWX|%W<R<I96ib?l~hoFh6syMi_Zs#
zud~H>cInow8OG%Jlk@JMs$G_CHa~ybh9qx?%F`roJWX(jpi4j+mf5&*L*?{IpRGSc
z?E0SYSw}$(S3}z7E133k3iqE-QdeK(DR0@Uea5(-J~OY>nlg%M>kQlKijFCHyKdiZ
zBC32orBkloTf(g~lIG?%!M{(mPOcj)u>y_xj>r+3ny*|#SDTv$)lDuczS?nI&eGQ*
zd-rCt`4}C<!ATT$hKd4pK~zoZ2hb}@U!tcn2U%O`;N!=pYj%?2{Sx>Rm?HL@GOq$O
z^tvlZRkzFRS&r*yxd?QNc5YqcmCbdtd+F-x>hDjPrKz^&i9J;_&Umn4`n`J$Hn7XO
z9(M0nw+Amk@z<t4A3wn4!ZGiB7po)K(Y<^39x^J{)XE)R`R&_Oq@#GYl`B`iOL86!
ziNt<lV`C#DLGM{juMLU$@w@%DLQFEo<3MD898KT8!GVFrU%y&Yw5+|9$(G03CeGMr
zsysg-Y2rVtPTyUq`tiXKqrns+-PeDA|8(x%Sly4UQ)?`GfZwhCT0!n}IQ1T~47==U
zx}8qAjh}O&B}%&|XPsePtrST`y&tF$m~V)#m(}lz{Os%|GAyE*Uo@s4S)}<wa-5>#
zdFtO!{y+BKJgVpZZTpSO5Sb+|L{da#sEidwG@wC>q70dmCPk%W%1|Vc6b(w0%wq#(
zN<tDPG@waBG)WnHUhn+wy`Q!A+JEdno@YI4?Ynia-+f)K)c5=OzTf9?9LIT_d5z$K
z-+7*}gdkze_Ikloo-}zyz<V|@G=bB=B9eWXw&Mr;^9ZeiJD)W^vAsN4>UMVaSL&$R
zUYp$pA^Euif86wCKDxeO-|V6{Pj^(5{b*l3x!+-{JonVjSL)<WH|_FaveZ7(?yWUK
z{^-<Z&cK+pecs-cmX>zKBvAK6ao8+#w?S0qqtm>HC@SinIXh83IyAZNTk8p-ri_Y9
zL@HEc<l6M6vP$cQ+8bTXJJH)W^xe$?=1VPGMz&{9s<{BjIFQ|WJ$4or7NVSUtuB0>
zdZ?}8yhyFzfK5jSQw}ua+ZuSPuh~VfkA_wOVf8yE3>$U|mB9D#tY<P+s}Hl8aln9e
zAvT`E3J)N-Y{e(Fd~eeipA%=#oLQMQZF5uYO1EW5Wn`OQJk8Iy7^m`FBkM-_=rzT?
zs@Dt}G>D3O!JL(c!ozQQ)UDFZqPrJIvd4^{(}o?2=l)i`x2!cB-z%%l;_rE&7g@DW
zp3G@jd^@bvM{2dhz`dSFDEl54q&C~+<lpqzfZT7x%UVtlVD`nTFGGBGtaVUoDW0I!
zG^O&%fIfYoR#yJH_I*sHhP%7_(T?*vZx0L!*+k5SI`Tj!S21D}3Me7JsCe>k$GP&C
zyqa7-`fIAG)$=@ZkDW`u`*g?X*;!dx=RjF1PGsN|7&b9wl7DjD(hCFH7Zw$P6=^aU
zAAes%J5sU0ZsQ=e&5|8jT-H<>U#X#0Kp~P*I}3e~UD}>eeFhDJn%|;(A|xaq0>Ki%
zm_?gEP=@A}3*XPq)B0)m0Gf)gUi7g@!AxyxcbTX6C4cz~A0_ENO##ynzYRP;NNR-Q
z%(9_J)fQbgHhmwrAmJqykJzw4bsB2OiY){}Wl=>`ME8E(wjB?XqCaN0isW7Jl~MMh
zo0{7(GC0_c8ds`up(JCLaVl4xoS%BplKkF3u{bcK>}P;_0{pk<(s>RJ=dFW;dTK#m
z0bX4mE2VeqeEj6c7bef%t;2keiuKL7kjT+q$jd^#?mArQIds;-lu;K;vA#6*R^E}8
z<}>7E&OiVBqwbiM%5+4sv1NDT^MR*r7mT^ww?_}-qirVnQ}_T3DE)3?bZlYjdKZ^k
zs^HjZ>K~RLvGE)GXXJ)wp%bT0E%1_kv(>e~i<{f}=O_1;q_reI-2o<uf|VunJJ7)~
zgZyN%JAn|oxZ@X-Dh(K-&CvB7-folGIOuO_hYn8H?#Ljw4@2bT?^;hBCl;Cb)y4Rd
zee3=ZGaH6AD*t*q19BlRv>yH!YTN3$$G3)t{d@Q3EPj!mqqbXHAW_!V*1m)HoubT$
z%b3rNd9x@#5+~Mpl~zU4Kx5ZITch`)GLr5CBzVi&7pnj1uiJTf3pC}-`oHl)cU-z4
zmobtRH-|QTOV&v=%%x_WT^RRa;nk3b5B2JHKqepoDTq2%e{8bN^?^dmqJKbrb+*$H
z#;4<E=Z>t-y`B#9(qO(uZRz;XvI|FTInh-fx;S}WvCZt|R!%Esm12`XKA5p&(w1ND
zr5{~Ur+mmdX2I(v2-unTo9J-7IpoUdp)*8{ceAhrfc^*G<MEF{bT2IOp~S3`JZCv*
z?lMx$gVHHt$17UAgtRD>xfdqO=cV>cPD%olHdvyPog4<40t=|UWPV-N=SfxBjyXw7
zOV>bw2hOwO=a!<e8&&AUe7fnMSU^JotL7if4my##z{UXt;X#V>x6u3_YKOzZMuz9K
z{pTNWviLYT&Gw}~L{<aU>pxpw)Aw$u&0SW=!QMQN^nK>BCjWf+a*yswH*ab@e!GMY
z`D4LQAWP!cMW+M4(|?}}3JbGqN^G|R9Rx_!h}*k(xu$PRj~-%qSDhYI*6^%lUAukJ
zK+o;=6MYOPdVKLW+AV}?p5*uPE1UI^UFWUVY+9*1;5QR-RW~BZaMshb4>3IILYvPC
zG!3y*mo7@}Il(ShY0{*HCiE7UapBQ@k0LK`=<g*<mOxoAzPP0C^u>z{$<uv&4;X!D
z^XwjDI>u%gfY=d1!q!gG?!l~UJsjoH!HpkZ%+|cMcIncnCT$Me!iBHsrO&;7N|-B`
zI<+>)NO{@g<>p7~zJI^&ZTitme?~Nb_)L@Bl%GF=KEl?HnaiOeG_wZjuEtgeWL8Wo
zTY^T9F1#=UT#$pUJbvH%_wONSuN(eEWWzA;bC!&pEpkGrSfyV^vjRu)mH%Qm;}XOu
z?=AJh<_G_!1yE|R%76U$vAd#TfmrFx=ju1k0sLaki&I$u7eEIi()_0(prYdwK1;lo
zC;pT44+SYRCNk@mgyY7I8@H7%CZVEsg0$6uO$NHr-=VB)((100?w|K@sdf)ApN3gO
zbZ5?toBuql&e^JxpGUJO7E2H`UftTfsfpdF2M<#ChoM+$_FOi7`c~_J+toc&`K<yH
z9O#SUsp;p>12`-lH=3iT$2i(qPd^FxH_oJiPbv;1NA~R-py7VeGx|MjiotH94W0Sb
zJki{HW#xx<wWQ>+i~$}i<iQE*hi15AzlQU9xJIO4();MwtMqi;F+Rh?J=$x(qJOW@
zo1xtU9{C%Sko~@1y1lxXhsY`!6UXO!aa1;ll+IG0#tF~*C-UCCdz|i*6TQEK!-aj_
z6i_eS`Sb@lNgb%r&dbJ_ez@bl?=I1BblUpIM`9-Mj2JICdg19jgsqgT?Gwj;k>o=j
zMIBdicO&@?uK4rwEkd!m>FXdE;V-+myT_=^qn)_Uur|MdMW1+IA^Fo+u6;V|xfn1>
z?by)+^cqtuHW8tZ6~ab}BE!bPM|tv$X%1%gx|5WYYQN88{GyxBuD<ZPs=f9?AbihZ
zaYTIR_SdzuU;z&Pge8PWKBaWl;zC5Rk;2@K2fm4vV!k%$Cel38?V<C5Fh%R>W~y!p
zDfyOY=K3yGnEi1HOVG^2hnBjbBBN?uLa&C=S}H0kRRt>BG6EzebPp8o6doK-_2)_S
zN?)p}u)^HR8pl4!$QW>S*x;{8>_aF$)lM5D>?(2Z^63A@4A#5_aqkvRP+c2nMw1KE
zlKfPNrF-{Wqcz*3(Scas!RT2p&d(ID9~Yfvo7|5O2N%}ZY`*ml(~E=kZ_zJWU-6~0
zWzK#2pI&sO#d3oN90>E$38H9^#3%UO>%Gx1fn(_x5)&usHo>+}{hbe=xT<iEB76Ik
ze_f#9E>A%oHuN<oBzcvkWmrx3a|{+_Ey0c5#<91LPfK86k6UxD%1L63yDeEe%TNv;
zuPUr+s#SY!iqPHBVTV2ED_h@m=Mx2G5k@uq`h+oMm743a)Kxi0hqG&*4Yx6QgjkiC
zAQUf51bKQa)b2q@KpPsJA60Qt>y}ZdVp;Dsh*W@b2?ICm@*XW!IfWFNqksOHytJ;f
zm`FHy7*z*H1iUE!*iX@XkGWl~)_1?&ym?PI-J{PRd7ZeZdf|@MntJO5h&P$L#h7T6
zHC00}VPjnw^@k!YR(EK7?W0rN&S;mzW*tjF1@cJ4*H;H}WO?)n!Cx;KJ!Z@W)Q1(F
z`M2mA)hc9SH}42Rc0@gUxa5i~0Ydr=7?5!OigW3#_;6T-85;zLbm+icB!Mr_4oiC#
z<zadC5i#GL`Hj*kqQ?F1USDaF?h|XF|21KOeB|6}!nx@$i%08u0abjSFYVtWO!u=l
zv$Z{Gt#N&@dXv)|V%&rK_xnw|1I;9C&8uo@I`ydxWEUBFE9!rpt9Ux8)&OiKc~z?F
zI~0A^w-zA&u@2!9ixz=0E63`0y6t$pRHhfLBC<)V|Ni}3J+cV4I5!Ii?%5_^L}$6#
zR2V*>lml1zr)^uc(p;~XVe!_6%`Qjw`1>D9y-;=`xVd!26Gnq3XT`e{IV%b!>uPK3
zRgQyiK){_8q^+Y1${Pj(7N$(iN?k`18ui+|$Qd{?xFP%1FwcqGcm7YhvC5h`F-%)6
z?6H$>lwUnrS^4qeAycPrrl+gj`)yY@*)un$HjNQw_>#0OOPVcAO`kJk@MQA%roh!%
zADG}Cw$g`?1+us9+rT<?d~vjUY~Q|Jti9ni0H8f?88BeLRG&XyA2Uv;q@*;jzH-a@
z8{t<^_60wPv^8<W2>oV9l=C#jSu^gH-3!?FqVoD+Vb7$OC-`)fb#z#ZOooG~@c9Dn
zjrCE>)wgc?xsgQNVa!3_+}`8E=A1%e*UxdKNB<?#x@+b34ZC%AHYjN7Jw5e?&#zSh
zql==7UQ|Xa$+$mBng6`<NF3{wfm6EncU{xJs8@A$_1i5!*DpgSu$)k^JTOX8)NMcd
zUp3waec9xZ3^ln`!v1}sTC$QK0Wrj9cuHcTux0@BCfl2a<B<aDH^Q_S8A<N1s=$ZY
zlH5j&2h!><eE$3yNyRlf-n%Uk9uUpZN(MJ1534M(43|D~XkMdWOApNCDPQwCbn>J^
zQ__MSJ2{$`SK{mIT|JgNu$`v14m52OnY6O1%5Hq<N!1wZ{=ezT554(!-13<mxI6P7
zT5ybEhBnt%a?1QTuOt6#CMwX*>7`H!NBlAld#$`*B~tz@002ZwP@1Cr(4oWUhLu%U
ztNe2Oa7S*x!%vgP4<81emK6TZ08Htj${@=?30!;f_vwU$gjGQnTPoF!yk1%+yT*S@
z8`ZSWDq>b&-5Tth=35+DGwd*#n(_HNGT)DCZoa;1$3~-=;R*^zQ!qjLv9=!hX2Jlx
zEz+)inWVD+{N$Ydk&!m&)Lsk~HS(Rf_L`oKx+~xr^YVvumf0~{FD6EkiO?t|PmLFq
zbmxa65<@~Ps91z9_GpKbJkt&<0?u3CI`iY8F>G;QU?4Fy)aPBN?daq;HTAFY%y!1-
z!BiGy4Dkp!$GZPz+tmHJIu&ZkCjo6y)bYV}?bHcf+4<#RIS|dcKNeiBY43xXRar{i
zw{K!~vYwtE6@=P%vE;V)=fwT3hm;Y>1~qAz0}3hHf+ZZtJmr67bi^TS_`DlpV`B0B
zGk@9KcWKA9UzSorR$M(?4jyx2;`f(5*Rs@QRMy@r`<t4Zc@LJVdLK^Zi0F{<>e1wh
z`>Q{G6w^970nbJ;_jo4u!xSWrW8>Z~6o=BS#Xc?Tq2)XyWGn71f4m!>bYViFlcXeG
zabiw9Z=KFQ>NE6WiLaRyNLOBrn44y;L;LpntA+E)qmaY7hB3;ffxxJeojM(SdXvX`
z$nH_^`?ZT60zmc~II!@fesA4?uVinF)uD<4>9N?jX;XKQ1HL~yz2;8G!FmZZKE>v2
zaB(U7`ka=%UHkTIayRkpj9mBXyXn6Nrw>RK4RVUd_8mL=4E6VCoTBpC5I3ps1~IE`
zZIx167b5`%(Gr6(8;<-8Q2eX*qIUlxmDyhyyR3Tb4(#Db086rnIr=fy>3Jne?5Y$e
zlEJE*!3);bgS~z%svKH*I!D)}5QQz;!1@Vhs-bh3`FOc5t9q@gi8K)^!Ma(=X4|%H
zVHQOLLuO?^eE4j!ZNZO{$YB|=IUR+>Ws&#2ge&qIK7Vc&AD<UijCcF?FLAfm*2l#u
zr8}xYdTyE;P#QREcF&ppWv83eyPwLTb7Ek@cC5w;QX<;Y2Q~CrFO~+S02Pk}j!C^3
zT)g2X#I1JYyzoU6qV!@uVLz5j<WsY{SSl59@L&xt@&E430j}d+WKIo#u+}wYpt^}P
z{!;614e%?+ENj}zN74=J-v#JFg;#ziW!DTdkL9h3`l}%#-udNFb#)&3rK-V~?8qP@
zU26}!+jFh!6_WITCu4_NP&K0dd?q*8uF=Msi_BI6yVPw23p?QTBMNq3B2H?-{44%*
z3-B1FOKn!|RCyhJ_8Z^~u_Ey0I8~#&#~kOqoPOtdWz|$e;PB4JAC0*f9musGhZ;TR
zfEFr(M~&zD^yu*o8#~KapTQKmtnz_U`2){W-n)h)f{H-B?|AhAn7WwFKPV><=5po3
zBZc9_gMX@YO^!p$6<1K&L3{Vo9ft^3$q!N}M5Q1hUiEPYwYT5wI0)0+zeTMQhlfK?
zUSIot&|D$*Gc-Ijd=8f~S~I|sab2Eg7i%YdT=A%fVjS+Yo&hq%*;gqQUgJHN>D50c
z1_un7IPYs{|C8cjLdrGBj5Us95dgG6*D#lvT;g;Bw~*l1$|_1S8S#u_rD2|OfMLu!
ztr)(mZ`6p6P=v%D>#nbuO8zl>W@`qSIdkWonZBY5WWe{_?ad>c`Y-Wbtxx&gdEl&V
z2mh}V^y(D9SqR0M?dvn*GqJ%3E?K6+hD^YnR7&4kN$1DyNnd;?NIVYuA<>;zG%w)<
z;IgJkWyyk>>v*G8W&R=bGs8F@v>w6Lr)x_y`WD*r1o$Y->o4xuu>&tY<+`j0ad<Y1
zGX~ezceG_L)O^F9hs1+!G;B0vo{bNu%I}Y$Fm&&ml$_ju;IN!#@yjV>OdRDG1~&B2
z9XE0$6C#V4A2DtTL+QYvX$G%jSmce_hf2kht)S^IN){{PLJllEjJQPLtJF136$bpp
z_*d6O@8z&^l&WeS1zJDRW6fE_%#lc4gk-PNH3#EnzePl|D0ut}CZ0}+IX#|=--T|J
zH;DJxc~kUmz*#5x!?dL<XfyE#G5MTM$we1IeQuI^sPBh_rE6Uq?s@_?TpSnFR!N>K
z0l^1zdlN#rYK)bQ&0A(O1u61!OKMP1kl^GaC(i1z6R<Hf|85`MoYtofvPWQ)l9V)#
zROu?-9aOd$`aS&iAhSI$udFu8S~*#b@-B)yTq~sml!=CZG8|gcc100<s078?dwuXq
zRqsab2g4E9U_--x2NE!moPGrI8?Sf$%{U=Fva_oq(Uu!VNN#^uT`i1@82(IOoN-1i
zdCh?s2HLsc!;B(OGoP%?JL;(DC3Y^g{N8ly{GOp)&rG$pZQJgfh<N<ZxA!w{*>onN
zTmeakQa%s+T{TCH0E5acGE8kV@bIjOh6V=Bz>Iu0pGr$(UOl5p0EcCx&r@*FwF8%N
zmOaz1`?P#0FJD`1u%^jv_T<^Fj<6=#KNLbIIWN!IYwUtsK6_Afy|;{^A0*O>JD0S5
zdwJ*Mvn(33;oVkS$&cvw9HHNw=szV-CEkq!yFfBgRm*IFxc}XX8t~ye`8Jt}R?`cS
zLnU==ZOe#WSz>r;jLn`oLvTi-*{R4Nnpli+63b>06XgGzq@`v3$kH#mA4*@)vxyUF
zTh1&RKW`Cpi;Xav<Hn8SSBbj>zfmvtH8Vh#`D5}UO~eO)^KAHFw6Q{nY%SfCqY-}$
zCjjxKrB%%Q1hZe9wW^uBw#(W$3FSDm&bhf}Km>F%4{df+ovjUD&~V7D(Bx2rE~CPf
zoYXQDC_Lo4a2=CBsa+){!3jsdI^Ty6ZIsP;5O#<S!LbL8RUm7q?`Z*&qX;(+AwF1Y
zChUtqUBGd<pPr5@pr?lipxk++I2~esl`+9MxNqOnN@QP??kY#plofyw98tt8PBqZe
z6U`07RzR~1lYwwHT$bo8e(;=*U#qJ*l;@wml3D2LT91_hPPE*a`DpG34<0OKqp6Qb
zjCnNWF-?$K=fd9>PUkxqZbOgHpcjLGOop{LtkzX}BB_WAGI40n-o4HnHYih?lSj^W
zU62ilfFi$-FF|4Kq=jHfJoQ9G>1HJr6&rsyySZ84&S77C{N&;!`V=UAm9nJ@e4B_2
z2y4+jz~UpfT(CiFjo-d~EbH9{|IF*+x61`1buxV6EBKGMCth-NAZ9i(pw4JO(n%i9
z^IUo8*rH1-cFReU;;ET}zlsqNUg)>l+B4R-l7HnoiUlR`!EnxZ^7C(y18X;Z<veA5
zRYYf()icZ`S_$3P)L~-mkM=Wr@hhsO?cM_!1cwPIEh$0pjE>40#R%nI#ZPrNW?7Mu
z-`U>#5!h2~<1z}QuGLk&+r6i50la+t^r<kw<t&sB_m7K<qXrKfoDg}oG$1Ur>2bzQ
z(>0VT1m0D@X13E7f^fA}i|X!Mj1`ggxC*W0pn(JJ_4^4E@G~z8gZSNaqH(Qf5?Q=d
z=^TCiy#WCMOm97XijMy5fddDAsjmAB6!v^T!C6K&!{%Rp__j|FNRjgS<d5QefyVXp
zoFM#aYU0h}gM0@^bG!s3Rx9i)j-x7nyUF(wW<zYRQ0_c;m~f-sZ{f3G!{WvYrWT;$
z{O(c#kC?T4(m0)KID0^lfV3cXdQ?!5f<&srSD(jgzc10$)TBUPb}^41FMTMO5AG}J
zaAT*;k)XDcQd09ncJJDCp855_9c}&`n1ASpL{!$^m>j9l^3t;=t{Pgizd%s>C5++_
zT`cW4aG{R3%@-JxaM|1zW1|7=yji_tuWM(UoCD9a%NhrTX3kdYdrDrt|8JYywazb<
zrVNUVxy~(1AMX9?;<BgRsxA$+Dc9@K?og(JjYN`?QghL<=)os;=rs<~Kbze1bwEE9
zeQ_5r48uXs$4|_vG9g?WMC@T;FmF+fMg+6@$}LxjpSTS>n~x2X?s<gji4XQ=fWd}1
z@v}x*Ue)^p$thN;AMev;YsO6K0_k;Qd@$^*Em;0;4QT`HtTqT6uLe9SoJ=RZ`Qu!4
zOupZVp$ZBLpf<Ct#rhHu?PC4+61v@5N9EVCrqXZ3@+ZALrzl5c%(b>IMkyFOt*Ee&
zVetWmGKZDt5V;SvG*JsGpX$CbnU^wjAhRlAWsg3CGhTIiZ6$Fr9ld(F&nmj{uVp)8
z!ZkOWYh_v^lkKyie)5u>uf@{8C1uMP95iHZJH!ifJKt&HQpUD5xV=gw6DR7}Zk1_>
z@bn6bl-22ViOimJ)<XV>J)g9h)%V?VL45=+^ZPYrIiX<6xIx#%V^OLwsi~`5p)_Ud
z-JZRAjRzp`$b>!xgZp~ceb7Wd{|vCZi%ye|=*+KAnW#HWXSUv)Ijmj3bn_&a`1wvV
ziFVr0mc(|&C%wncXV0FInKNfc;kTHxw?@OfENv$?DRC3;T<smM;W2*PxSU^!ln>l9
z6}uTn{I={22t0Od>WT+lng*QWY?2Bv$GN?t9EjNcV{Ay7;(>dc9D5G=sv(@|JbTZX
zYVx0_0S00<&U~1h-0ekoBy@X=&tj<v$TWga14_m<z;_s`K6!}*g8Th+|Lk+5iRVkQ
z=>!QVn;bOk#5-|va~o~<;$2(!vw&s+jzd->JbQ@+&+09OSbTrz5EEay$x*z>ccrEG
zGGxZi5{73{d+hKOq@+DA>w6IVL#8IBH{T;6@ydjb{kW`-mfx1fQ<bwNZ;wDVzf~2K
z{UpUyg|HLe=rHzTkJ()%-pypZT&zZI_f)$@X`s4<#Gj9?-we(`hPA_e>E-*lHWJ6<
z5H>vR*2cV)fB9yVF>>&K{Dvg{+l1-wllWijKs9kSOA5>rX#n`}j^oWPaq%w1i`rw%
zKqsHWk?tDB#_HV?N(^hcZtNlv1?Cxi{P<_Fd4;d(CO!N2PbNXGp1mzY`Ve-7f4`#q
zg{k=8ujE#fk=B3yfBWOdS$Lyv-NnhE-k8$5KSm9R8STiy5W0D^-jl>mmv^rC7dUT}
zboctttKo-!o5r#?{GVoe=u<4fB!u^2Z-m64)&Fhn)%wW)*B+n$_DpMi$p1TH)eoP~
zMXRSvTNHe&Su{mnUT*tVlkF4z_S}k^Y4%zDNIyfJQTm@R4a<48|B}LkHTEW#%3m&A
ze)dxP^x}*DaXr;G9+tng=XN#Dcp;x}sb2Z~T&AJr=gHMiCvR9k;9c02e2f~mj{3hp
z&4>S!1K$6~KUQ5qMB%QUtAlGIs8m%t)s*O-JA&=T>o;#6HEeQn`ay;7KWvitE}s!*
zBZDuk5P##_Mv7|2h7auBJK{_y;mSt7j1D93N<5os!gd%ONSVv}%o0zfyPc>xtae@z
z{*qyT<IC9XkRpBZ%)rr6pzEp+;6FvR?kSOw7`g_ifY{(~dj`#9*wJsN#ZUE<N?Rv(
zPmOKst7BpoywJ~S*;M{j?tx4U|4(AyPz{Zn^vS-*Pm=OQwSVxwu;|oVuHG8kN&#v%
zlFK?z?qrA`u7q*ne_lw~-F&y!B{W-0X8nELvmmKjSfO(6wokB{{i|;!>HfL{aQ5W|
z=C!&G?yFz{#(HpsZM&~XlY%7_TlcO=NURv?G_vsiec@-XJAf4{5^}968nHH69v$pi
ziGL_^-&a(am$9rxLc;Ptui+GVoB8U?692?k@*cMaweHH1*m|&!+z<x`2jOI0B8efZ
z#4{a{^7#+m-|loYH#hIl^&@iSn3xsg*p4M3vF7hVxc!jQIUie)9lMNKvcv1!^*6wm
zsw$0kSg}HU&?Qo>KM!P5{Ja`BtZtt^otOO-&OkKfmmY|pSJwJqByRtp!z}h`@z$I=
z|L@NeniT$;sdjc*;*K3Nz>oc>_Tx|Fx{Jz3LLyd2)E?pCbzfelY;u&6^6#}roHOI0
z!kJbg%51EvDq>NK)S~9Q-C`~NOAEjuX?sWfb&tQ#MQL>HxlbikRs9k8jBou9cEZaT
zlTsNlcxscpoD<~2VZ(m);ni;Kl|Cj){A~j%U!9!OMuTN#>kD^D^9M>gor4WlvetyJ
zcP)z<tOXeg+Kmg#AH<;{PAnCYkVyFZljTWTICvb?xNj=nNx%00c|(TicNGg-7xLUz
zda{qFcJl}`a>&c~i#+T&%+b-&S109<Ey57ERp;u$$t!f*OWZzoPEcth7mVL1J$;1k
zdRGZ0Nn8ytUL3ok)oxW{dd@<Y;nFGbV~h<<a>d8h`nBlqTNVEtsN4D}{r4aLAM{LC
zJrF1PT|+5|-j^?5_}|LPH7i%fn#Lpd;8fr$!5|mBRU}S&E%xAh&Tejbc68uPA5Ta6
zsa1I`T)1#fgvO#mfR<G7wxE6Y<<JnIXbw&E#>9uAcE!QnB_?-RVheuW-1v1&^-VB9
zupMbbuh<>?_s<3D01Ge&XtA=gQdUl2`X0oIi4Nw;6lrZ<wP`Ey?MQ1#i0aFK2R7OS
z-~d_&ik6#G#{~rKndDiI5nd%TEKI|?W@|7hfO>z8I&+X$%U)DsPt)T@3!G}K#(pgv
z>70K8de$4o(7OqfahKV`QVpnLyYKEBa{=Oc2d9m)fx#L6=L-cGZuyFeim{0+RHy7v
z8Ba?U6rT6GZ28`7cf=zMAqlA%ea`1;ElmdkBB&x?(-8xlv*+bZ_7hZ2jkUEw8i$V^
zI~Lyhg$@&E_M=Dp4IbPAasfcQU4vLw?Am%^YE3{}cr&0xZ;FepBmdw;B-SB4i@HN2
z9UK^Vk*1ud2H%yCSUU60tJkl817{LTT@b4eMBKk5Xvq5pggUC3WfvSq23Ns3)3OTB
zpootqH~(ee^b!;N0DvQ<J9{@X<?7AO0{Xmg2yFUhSPM7Se5d&!M()~qgr??i9to`}
z5C1D8T_jNLdu=4#=bH>5hD;9u%)$@k>CtgBvIi#1xOKoNCrNiuKR*NBzH8YtBpVFz
zCM9|7P2X`NQ0D}^*8@o0pS*EUK}$=^{!`@-2a@r?J@m2%dJZx)e2iXX0^WBI9-MkO
z<MO3THI0oe3~7@a-GEww(iSaTDBZM3Lc;w^px8PVxN7-w;Yk2YJZ0+C0aIn9V;B*w
ztQH8~jF1g&+pk!)>O0WDPG8^2Q>OHvDsy-vNUNBHJ~6_52U;P(pfPZn0(z(kXQx|!
z+t!%bsk5&<))m5eXrr8xHD~ou;R&<Hh@SkX$-HYpbO_6PlC1ZjTf5jL9856G#f7VD
zzJJ#u2BTxLEF9ieB1$c00=Z`EP6SJ3$Bto@G>s!d(?gIwVKx%S2EL$DzkK|-XuNjx
zx7sCs=S7<hD04p}qwgk-t(H(4TnbjG+pg%iTn&wv;EQggdHnRZ!?$J%&@cmly?ggw
zv3Bjp($W_wg=JSRz><`S&^ILkj|vODU~8{mzYd(HQoW&#gz=QtWpU4B@GfDGoUE*{
z4&+547}0n5N*s8_@B`OJSQX;#lP)u?v*8J2I-`#-UKADz`UXwni0Z&^r~LNrUBoVl
zVxPHl=dynN83!Lw?P1q{r)~L#ya7p&xqO==Q+UaoEwzo(io=za1#Jw5K@)W!x>&tC
zit@E!w;ZpKfj@9&=cvC!G1DPjWz?w8h*P%9aYH!euL}y8Ov7C#|H&K+zjr)ePhfar
zLf0u&YmOaS%gL5|_wU>C(}oRue(kGus@G(q0YpI6(f7>4a)3cdZ|vWn1>P`h-tmPR
z2mgrWKm-&wS65djr?C?!8bNzA0P5%0xqEkuk-@r~&*Q`aog=2^kQnIe>w{8i@lgo7
zn?sXOj(B=<>MY`n@4S5ZlF<3&$tb{70tT`pdHE84q3)ohg|JL$@uOFN;A9@YtCkG%
z%NG<CU#9Vxw^D{qMl=Ha+%qB_V)gCYw=Z5?;ps#D5`%C(3;N9D7G8xPDLDyD@7^tC
zm|*SNP)IG9oUpKFt`x?>{HN`{@2je?Ett>fIM<20`<sf88Obqk8uy+%w;qKAF0BL$
z`}OPjyUeFyt_pK`-ZZ|mY8~?diid<m8exubwM*a48)DRv;cYwbiizXK#V<%W5)u;9
zoCVZ$EHF@%msm>?Fw({zOb9O1J@tdDk-^+9lP~dR57N_peSL9%lUW)85(DjzRu;3+
z*?i5M$=;Wg+a|Q($fv=v(n%UYvUskr)IeB8^2yA-JOeFim$@+*fLQ<9H3~cupqX7l
z!UG`Fq_gjGt_PL_j$evoC2kk&6NTOF7BO$kLoGW&Acs+8@<=B;9znR+A4sqfGOZ<B
zTgW~yne%0QQEWuX>;};PH~W-t`=?{)&~}(O@y-0kkI2r<Gb_OGg{M7n0+{{TxSOyU
z;8rlfd<i?c>J71seeBxh00v9B6=e9wf2XZ(2x$U8u{{YA$Y|0)9q=eD83Igv<ZS<6
z(0s(Ld7HZMgL<8=1qVrM`LkIz1}6(KGJ=&Cfnd*pk-?8hoWB~=T9&Y-6A~c6<3CB&
z;VlLl$*#KqV!5@v+}Na94PdH+5O=O>LFgD~^IC8>McW_s{`cP`o+z)SGO`#?XyK*_
zcXoByfq;NNVhsnX-r<UhIBv39lHIrGE^!c>Rz-77r=e)(D6=do1B9k<h?dqLuI98~
z-|lg{>h~=&!s~{qO)mf8vK$_G*#IwEDt2fOqV!Gm^2qE)OM&796GLZr_rw$LgTLv{
znxE=8Z|>j1qEE(mbiJLnFI~G<ELjtiWr$iVy%At)qVeD;eEc_$1CD~$kV5s`cTPGi
zq2i=TlaPC)r?2FRmt5kc^E-0vG~l{W@SvwW@$;rZtP9$2enN1eaV`SY>>q9jenf~v
zP=<|8!sHDVdgtc|Q8&eO3(lv~ng5dD!s8?)c99eIUb^sB!q0Hcnz+K%KEn9U`2o{X
zd$SwLGJDI;;L%=1So!;c_*6L$d9t(6Z7OEqOt0T#C&`LS`f-vjGRJDWT7qEiw09b)
zja@#;=WStzJ#!mr<;8FJ_4fAET<QPJ(9!Yo@>2M~4Dq%yGm0|!yplgvVV$NNW``Ku
zSIM@P-(P*ocQG3l!G*|P{gqk)hT&Py>9A=E9k*aexLpncc?hO~US^92F3NUS<GLZN
zwwj&3TAjHfJh?^$8_3Sr^r*d$7juuD2(W5$cuyaXzRk^bMf4c-wcWT_Q+6J%O0vK9
z1~1+blfP30OqqpAq_)f9>M8H=)8Sn-&pa=N#R#Dvy;55i)b{;%bn@HdmadC+y}i6>
zQDUe5-~j5KI;}o=(xl2CAEP!hrwN_Sz?s3I-$KpdRa^d0vqwWwTYZ7*4-9hLTSs4=
zsrdw<5s<r%mW)K~3k0riHkDg%EYdR6rDiQ2Id9%PoMQqt#Eh4z>5HLb$A07G@e$|w
z?13A?6^>8e;-aeX#o@&GSMNJb(1*fO$X_b6Ys$JpE`D^ld5@jQ2-c2Hu)x{*F(rXw
zUs)`cGlC3*seuqCSU7K{3v|8Ei;24%OtmNCWVP_JsM+q_y<49a{HeM+m7Iy>vHY04
zwAeUEHd0KRXuS3Hef%a_3z)n*dr;<Svn-Mlzn<^S^PUA|3zdEor_Wsv-a)bT?>}Wj
z#hlDd+mGs;pjNV>tcI!Rsu|)C^wy<EJMGJ_{NFIvLILpc)2D=%MJS@YsI!R-UYn2n
z(mi!lZT*{iz>kvjQp;3OeGu-CmDp}NF-7VXIR!l}svuDq4UF4Tb|$qf<>)O5(W|2p
z&YZzoprE&DZ6w}hwKn?iY<LVIhE+mJJ$(4-ouxh36(-gIVz?^gmmNKJtOXi?k>~NQ
zOF<XV^ClP>`RWMi9sXxCj7}U&N@{|-V-(1Lqh+R_3OifS`%AZ5hMJ%x29lZAedzdl
z9!1Hl`SV}!Nq^k)Ubsh1EKb@m`||yat!@4hhM()#pIf0b&R0kH85dbq_>uaBas2pn
zo2wXjjwMef#Xu_Mk;dyoGaft;4m2fxDpOv)d9!Qh&WA-4UfjX^hQpz|VAhPRXH$C%
zznW-W|0h{lT{TU}X)JBYk24u6>uvBye*PSC#P_dnFZ?X{$Ab~f6DxchXc3SKpU&10
zqe!sorycin2vR<`fmBm63GBo><lw;}9(oRG^Er-7b43;UBP2k+q9p9PVm8r;+qP?%
zsXE(jexPxIQGDUEC$Z#WH?kFVJ1WT|hN*GH%aN=q=NlL?(3?9qD>L)bwQJ{0WBo9u
zE7O^`dGk*`%Rb9It68LUaPfN<WwvSPa1_Z!_(aEu#CsH?q#-4T1N-(-8=m|jBLq9V
z%d+Kb7^O~J;xZx`pQplJ4Y0t71QB&@*ic7Q)dz;Fc*1=b`dpU|mxS)}a#eCz^>N2J
zbDnSjMAJt}5fS0AY#D>&>0F!%D86e4<+001s8ukUv%7}oVFNgqXNOUe+d**tgd_UL
zw!P4_n%In;4VeV8n@Sj*O+R4=&LS9?W5V+|I{DhLzWVm|XA_0XlZ)d>?}b)YUrE^9
zMiK5}W+gq8?yZ?-8?}mTKmmtO^enG66no6d<ob-<aXz#;fo+{gwT3z{^fu<y>ZubZ
zVC)zkwCN{en~(3_$=#STNMh^6UZRfQ>f_<*X>gr=Z8A+Sx3?D+Z<6{&L=dEUJg)I_
zW{lXL`5D{E-8hs)NBT;ml|NQ(WgMR*Cb;_0@hkDzWdXpb*I7*6z*mtI`wnJ3)Gu5J
z+3z6RzkhRWxuJ7M;wyRbg5@<_ap!${MwQx3>v8hbss8=1pNxu<%Ut}8w3FL{!Pp%d
zp!?S%$T`2McM~RF&FwXC;HY^qtb7)O<4h;@YHVS>&i!#WJsS)zk8K9k8<kX#2<#l1
zU|7Q^>(=I-$wQWP&E*j%i$P*)PR~=`>KiIaMudECpirc6_!Q$#sSue==`Bz{PA3=U
z(D4^pS?l>rC6-@~9617?ABjHQ{3oHL{K!4*YX>skNgN#+e1<*(<T0pF%49cWr?5gT
z{6p18k;IVum<#FLKrU5D*FFe2amK^_-0_)*V*7UOsO6{E2D<k(CrYr_ENM&@Pl792
zn6fFGPJ*8+;<(o;vZxv!s_1t2bpN-JD`_?|4;dOc`K;MUhUR$7%CE(sM~PZ8;-r9r
z$>>=4Y?>)_P@w$!{kviDCF`-|*TeQqta^KTdX5XP2b2n0>*}lXUc}8^1t4CX_F09=
zN?6agAJ_S2e8AvoDoyd+p2GtuyT{M(E(G((w6jOgDk>@}pU`xZ$t6OS`qZgYrXUru
zs+Je4ww(L9q_o8I7)_K{z9dfEzdtj{{eB%Eno&mE)pMCN2zp^4V>aTIObjhm&+c>#
zIg58rMUayj$}1i@`E+pb0v(<0iD_kLHfr|(g1aAnV|~G}>eq;OGxZP&8fEU=NZwKS
z2}EI>aJo<>z>bfr_uGCKZ`|nnR(mZ9yRsFz%N-m9_~&oric|Et2<*q8--G-O)_oBu
zF$I0uAm1{X7@Do#jZZQ&O&k}TNxy%;4B=hLANrs^r>}G^TCO&Btg(xa<uy*$;)E82
zX-fGXL@Uc1M%<$jO*<3J&QSCE^_@H56CaAzzB7E(cHX)(GB#$EI1Z|>F0wjb$0@ja
zek~;kt!>CBo1>#=VYoSB@@kxZs%xjVFJ5jo*L&(rQ$|j<4eqwp)PMA>HnPB@)3s^d
zi_OHG@~>P`+dJ&xlhw-KPZ#|#D{6IrZ?b!kYJ#DWUWZ{Jmd(_r>A&4h<}kgqYPtNS
z)AMLa%665doQ~{9Srl@7zRS2(ZJt+Bf&~wZzSiX?Nr^jN;!twjF@hpw&rjS<^^Sk0
z%}KTGg5IR3Q3_6t&PK7OBO1#)Y4>;<p{YDb{`{|~q5B8P%8qVI9eI+L(LWosCidF4
zF9;PBVwg}3^jay5$exFA5@|4OjWWHL=am&#l{~L<SzAw!wXWFWCuoyckjTJt=2JTk
zvGkZOcwwKm{XlPJsA7Wqs0-?!vOp$=*0@c0NyqfOl2}>t^|0jlMiaA<3uqUwc~U?v
zxcq_oS|M$;fP*vtoY6TcguC=z$`Qu#W*g~_8ei$0%6aSYK{jc(*!?X1`Mrcv7LRv|
zu|Q(qHOPf7?bB`<a9m09^X+mphqRRCl<p-nPz`nV)rsAd=?mI0GT6X*GkveEwv%=b
z;{C>--`C4-m@;XSJ5qm7&o{K+=zgrMKCG-#od12hob&>73zrQh+A-1omiG3jA!nz5
zI?JK0Dlb=TYVS*P>TDMep~CfM<}z_rjBWT%G@A5AcIshAkGAX3!EY+2<C(!mJ=?u!
z7`5c{GNk0^&YhzHDxbAB3@5X)pCfHpunoEh<OTo@62w=|oP*RzHX8Tu-n~m+sdEKL
z4?5JbW5>6*Hd_>K!S?}djAu44QnLK}-rzzD^Di9vpfB{ZFRz^Uy>g`v_p|zsp{rRi
zlN~uN-yZ+r;IkjcNZ#e!Hf-3C{zVM5V7rNx@%`O2(O2-~J|~lzY50yCz|uKSfN|%|
zAdbLM_rEZCV>#NzF`ZzwdgaQ!1>EEb02}gWtD}oGrq`YkC<ik_GI@IySSI;UcKn*z
zHk1q!tx(?|#T(;JO}b%PTV{6<8TEXJhe&`}R<eHm`fG1Mg^&~}zBss?X-rhF2dB0c
zlWLySAY5u?hYKKSq?b2vAy`;dZrC`tTeohzy7jzr_3CG4hY~gE(Xbb@+z<G7WRwMi
zJdu0Wc}Q({qWxzcWsglf=_pGrIX-?Zy=wYB^7+ttQ<2~4YdIMj8YWq0ZkN-0`R4l7
ztHRTkGg3CIeDD7KMa;>i7O8Pp8>kaN54lUY!0gjr(O%oh$?4ggT!U2wlmaNz3cFBF
z?~YjOnsnhXR_xX;<(DltZoHnTacgnFwQs2Zr?AH5+ymjM!4>YiY2KVUxtB79@vq{G
z%94p^E~$)9pY0N!A}0wL8f2gUCio7u1gJvcXka8hYM0@^sUH1C?$=_fG;@g&BUae;
zZf`H;S2k+Oo*VHGhbE;7Ze#3UYzVaKp#3Ii=W*4K419HlELy06W|H#DCFZa}xv2ji
zYfha!`D5t&=eKU0DKv4u3ZetDR!6~KzH#=Gkg_eicA0+dIg+N-JH`EH7OC(YM#77e
zhbSqn;iv?C-7OBjN-dmWKUOSTzl^8A;pjfE0;Cz5WD~pieJLG!AGrm?wJw|l<NcKM
zZ@eYfhRquBF%RdYUbSBAyw^@WsdMReIcT0G*&8x7V?9^BSO3`L)qkYCyjZgb3S{}k
z<zT`ltKeg!*$Z-c$tV2@CB%@2d8I)GXBV!UZJyNlp7TyQrMSs7X?UR=*9a%vlHXl&
zD>(Bc+Pjg8Hd7{C4J=&yFJ;ov`o61RrHbddEEo^*cKX^a>rPj0+z5H1MbU(a>Qg_U
z>M^g+e{hK0E{AkW`d!@1wpHL_Ojd0{8r=Tr390RJN#UnX)Ss=4K63oH7nQ2>Pf?#)
zSgjy;mwaB0?oZ9IpP{+yYxKR)f%m^Or<pcC+i-o)9?9Pygl~?3Tn^3XJ9EtWvc_pB
z9bFF}6&YpJ<ss%fjn?L*P9c0%R0w~67%El(yl8Pvo_3NJqonbqoFpiDSi-~o?gvK<
zqcos$3)=0fa7R7UyZcAN_>@_hnnRi~sbH2YrQD?wL@CHB+PX|3EkF7@mn3K}f)LZ~
zt?Xz^GVa-hC#d=jRWd=l0b8HjhccVIEGWH0=B904@8@SH8x5J5_A(AE=#x^2$@0aE
z7lR><abO1l;QU<Par$nBO+9IRaM(bC+vmpV*jQl`1q_#O*S&LRQ^(|mET{lG-_sG>
z)(-=G-tYBy)C)4y@*@wNLmIdE%gI26w{A_T1%`iq%6n=>HkrnofumCWvrsjB@vWXk
z_uMuqJ#mX_$1R!ZkeMVW8FWmu(x`ER>BGhoF)^j+<`fOPvkVA0#_Y1S`M926lg^wn
zjd%4i>BNWLN_1}ZK}E(7HgKqzfJHE&#IJ6sL9A)Q%m(70pT>US4<)kAby60mXiAmi
z!@S=4KDSO%@dw1XB&e&`T@-iM@n+};F<d#uCV)=cs>Y6jib7lSp6bKdgS{1+o*xYj
z6$<vKc&5xIV|JCPn<|c^ny09V%;I_s9{d(eobJF**1NWBZ<*G99h0;6c_&vtnBQ_H
z1H0#g%GVEZvH)NhK&2Tk_jY0qkJuLR^(r-1iD!S%5>|HrkOs}5Caf>4yC@lH`djrO
zJq6fVuD!{a^G5ULZQ&<?z15F6IHYe<b@zCF)c*bZ6Ms|mhwWW8)Y{f>Z-DD{5ww2l
zCvSSJ^kn=ms?%Q#J@Xxu{FqdHyVu<<*LI8x4Q){4uXfsrZl8|)nMi%<uG&gS{IDP;
zVxpj!|Ca123=$PW+Su`!;bosbeQNoIt{d(vNV(to<;zW(WXzlqq&QRoXjJB1+^P=e
z;9`Ojf>$hGPOMNYPRo7sCi&>|N|!!MI^TZqpn~86*tO%wGtM=5qBRCFS+i7Cuh69<
ziJfyf`Z4ed;zUIwX*Mdn#>^uWyrEzJ@8qbive8!5l8OZu=PcC)8TP2)%^S-|HBY)R
z+eKC2vtYIQ-Eutdx*4gw)*?Lb7<pG}dlYK;?8r~AdYK!!9Q@hbT=nYa32OJCFP44V
z@Wc#86J)@9DTP(~UTCp0VRzS8mc1~Zk-i&yWR@z)H&%=p1{D{WqE7dFK4@P%iLK9D
zMF+QAeqf_tmij<!T!EpG$^7oz-U*#CorYh57@rS$Z%4^%o3oc%@!^7PeVM$fZEWm2
zJS9)faLYPm8wSN`iWcUbx3j>gUFO<WoZY@dJmY`WR&96nAG>s`ZT}@bc0$pOT9Ry*
zV6r4XmMYQm3&0@FxXbFjetztJlfJNJzrVk-m*w@qIdkVK8b;5aHH*cjI^#3rKKO4g
zf7A!vxN}OtoRW@%y`r+85T6sBP5^3_(0|?dlx7rfGE-M(SX{<`X#sSh0QNRMymsS;
znB=kz9-`fYqd(&H7Iv^X`<tkYk2Ux5`Xxlij3?y@Q3Ie9?nId;&1rqqudT^hPMd}|
zK*<c6{5518^k!{NjlLKa6$PrOKM1S3Jm<`|67S}ww+cvF=w4*$8G|&;h%DH&>6t%M
zY2#vefcOKuJIrCCsw#8(Uln$<95pwp>K3kA;kGq?#JL-5a&%lC63puUjJ7D5uCj83
z#jL16Kc^ANmW<P4*fP=sH-ovPe9QfmRuSwi3Jv|PbA&l<*<%ev7=u%@F;bOFvbLIs
zLRczED()z}hHd+kIK1gqJOkW(bA5e6_7j&I>t_WFT!a^$a|(DhWoSd_q8>Z>Iho&$
zSOb{-Z#!*cC20gT&5dJv5ea!qhcv{px6JLjT-rfcS(o*wQB2>{H`LUy`rIUhu9*qT
zVH$}&t1B0AU%00}`W4S04_-n>1wA>TEbRx1A{^At-y_#kv)C+0yJBqc_o@!&w}5B`
z8A=3?vC&dcuqAlA?Cq6#cKJap3`8q_{|lc2qHi%|Z4XKGiUJ6iImOewr(@cu9SG|n
zC}*x!T&41?(o{_s<pF5s&6Y=fuAE<V<y2w~h|zCENL@P{4JL@f2#<L)xld<ZW8=WG
z6ENPmg3Kt?A}BhqNIm8+%Kk!_crggzedZfC)s5zwmTmO(jYbnQpMTr_L|Iu9r{21<
zsx|BQ)QqgpdCle3(KywA`}WMfzNQ8mMbu<ks@UNJ1@!DW#9^?y-C+h#Q|}>a(gL^V
zqdCjMk8D+$A(<0*$Vr0e!aStBe1WZSmiR>SO31C;zZ^R?s*Bv9Uvg60pFs)LEV#sx
zYQON5CN2zs7fW0wmJ(pvwQI*9Vd1mMQXIeWVpkm>7!Q2-ruEF#u`_ddCVHmn@j9&<
zrS*KW!fuLDTm_HZzltNqxI8c~;~misl6ZIq0N0J2Su8j7#t$%c7o|C-$F_rB78Fcn
zry7j|&45os6?R*zdF162w%i0%q=4+IxtWF&NQi00dJ;Uv3br0IXWw<kGAXPjIN=K@
z;xF1Ux<aK3v1Dx$YoGkHN<xCAf4+zfRCoAP!@fE(Ad|Uz%k8agC|ZsT6bMa-Ev5;*
z>wq4rs;_YCI3s?C-q8S00(OT^Qb?O7_%{v#cNMUwDy{b^Z6-x{S9n22=T(3C0`x9c
zOO_V}h>cBng1UWHODJbL8_1=SanK*eRQbRi-e9=Auo2RJ;M*N(n(}{qVr*26jsQ&(
zkPU7c&IGZNx#pusjS?yWpgEpEyv6;uH$7fN9g^8gbiy)s!DS5-;t4vZpIi{QR^XRH
zW<of!F4<3rJQ0W$^41jk&-r+DbE89MvdY~S>pomuv<<h=G7Ol~0^E2_<){Y~CtTem
z_J(^6XTE|W^P2XDhPt}>7QUuR(p(izVsUXX5cZ4TCgeP3Z?=5@9IE|+P?a@Z=zyl*
zWe`V1mC*FZ%$UeqxF#Nb5}_&F#7i_K+H0e&TE}}zZNHi<mLY)r<x;dZ&7&$)nXWNc
z---iQyxB<SVfpv&4P`EmMu42As;&+wa|xfNHf>+*UaT9!?R~T$_8AjB&$%Pcsmv?R
zKkxL5Mn;S!h?XhBgL@qG_`G_&*%=i4yLRpJQ(a6uBZv=o_aD?Xh+r0s3?`OcxgxM%
z#oj);)pP@btGkgMs``QMLksYp`j5CLLZC&#?5D^Z3Rp#$dCU}84RQ}b#D+FIWOnk@
zckkRO>{%0Tfxm+wM)Dt#Q~=gM{?PUP?fD>6ZRTux*d^LD9{8<M1T+<y>({R3ogUlB
ziRDl$dpt}e0xGep4z|W+-IIo!Bdd%_=}9wC#{Xv7TYYF5nBuNeIm|~wSN7hqOLF^I
z%#SAYZ7*^2TB|xu&ln|#wl*DAbKk`(ZD<U>zC2d|n?x5bHQ^o&T)CZc-%h~>4d{JV
z+1jp|z&6@%LT^@>n{)TB&f-p*Ec49EyCxaq;^b6`i(%OiO+zE4;77!>6}a7%%n(aQ
z6C!Q`W)crfGOmscwgXUWZs$~6mPvg;hBvQWBQ!(o9L(0xbs+0w&%D>xAito%Dxu4s
zr4Ri#bFEQMJc*oUldT6wY;xUx(Qj3jpf(~<xLS7EVd~TruI$IjMwwDwyUL;mV%E6x
z1PuxzE)Jz%&}_pN-BV>G08US_K#;GrfKDqKW~RNwMh+*<6CWBn3$XZ_5YU~(y{QZG
z1R!3XjPG$0i-oXR35sJO*E0%aV2~LdX^MzNTVVQjbz^Qz8$wn~bHT_wtP!}8L$t)|
zOXv2T^!X9Gsp)H2@HA@FsQX_K5m0PA-_R>*=T8B)g*b`2Qc-#Chsl(soR5YsOF8+%
zLhR+}&M|BYJhgbyfJ@}us#iCR?b*a4UfJo@z?+=-^x%cAdiwfw@3+bduGL$y%mu`=
z?)RV_3#or$1BE47j8UeItwEY_wBoHa*E8tEw2YH;#{9b^nmppUOQpot8v|PvFH0ua
zzcPlF;N;^lF0=FpsS5LFb3u-~+-tu9M2_0B!phj{1?@6QP(a=2w#TcPQ-U$_qHeEt
zU|@>Z{~DWoc@LoE)5M<;<)DZUKtv!;Jb@9@k575Hz;)c3zW2dv^@=+w@rR-)-HqPG
zf0J|W;jb0ZmcsN4%#Bj`Up0ds?P5(mnB2y^W=6-V6|fHE9q&|K4NL;ZV}re8mHEhE
zkoJ$@m}fWF5(5Eh`2H6`6AfLxT#%v2^y~Kn9A5gJ`66m_5c;mN3wE;ARk}wH@Q|**
zcYns9fmlXn%et|;lXzr!*%E5YQXc)`2T_vHWCYEY`aa`+BC|ZxxC&GoRLlQPl|j7Q
zDm$321&%EIZD>U3rr=z>LjHlC0`T_2O_c&vj1Q$zg|$8VDfbJBjt{%8p<xL*Euh^d
zrqAiwfHkW=ecDXWXd79L91HLZqPv_=B7HWqhr%M4N;IkVjvY`3$Mg>0qd5Nv<^ya+
z4Pb}N!4J=F-MTeE-A=*dVU$|3K`falrEEx&+N~&6lZe1T<E`uaky=<>^b9c;y50|)
z9Yq)=7K)Wdj3C0+F~HkZGdu#1wOmF5xi9|R$;E5}RT?|i>B-S?2C>hdg$5gx6D4qN
zK|eHP5U8mpIrVQ3fkaDh%V;?jm3fUK%Q9|A{gKN%cJrx`hmL@o6ZRw1r%#8UU{2J&
z^4<tbdJG}@K#9imSwiW&>5OiYdp_mH3&Ah)M9tx2-jkt10t1C-DYg~D(6&Svy7a>m
z1kjY8>GPK_imRXfHXo(Q6eFPSPhvD1Jn<4n!DHHZ!;Uu1(708*HU^l0VJrrkd;!78
zSP2UfDypj;)~#E&Y}rA|T}ZH?pjtkDLJ7+c??#1^qeh>X3FnW$-expb;O0Gcl0JoE
z3wLDF6a^kXmI9245&}7v&rSSbIO4qpc}e-u!Z=@vP{QCgbv#1Gt5>cV&Y2?`TX<{o
zK*MBR8k~u5y;LcP_yB(V_<_p7cGW5^pFbKSM^+KnJ&Fz+#M&wVO&v%a2Oz_8h(bQu
zs6V?A7?#=d%Q+6Y!#``#TO6KaPGcMv7RLYjb(aN}V!kaX;p9+x`TDa*hO%}?(e1q?
z^N}29<Z*8b3ed<*0ER`yq@n`vLAzb_xY2DdJc*Y}XD8L6YT_#wjs%TTUS1xX82LUv
zS_r4PkW!yE5+<PUYrb)rJ|6wW<0Ck@ZrHGnv0W5HwC<ErLd!|9cXLx>p~HFl;-(If
zIL~?sL(32=OG_xEG-L(*i^cF67!9z$1hB-FmjY{7_H)~m&EL<K&2Q%Hd)-qyN{xZd
z*t7fh@8{<WGd7R~q?7Sx!`RW0k%3E{;TETQvmoRN9d1qTddqLt%$d^f!Y`P{i{^(g
zNCu*-O~%luRhPsP&l#!$K6A@_ALOW;WVBP;wNUmo6*tFFq5Fv7M=-{GAu6YO=eH&%
zCJNRfFRP=(yV0lJS6nylxHYKk!-tFzt2Po!nrJeG0L#CCWHZEKoupDX^8Vq%cPB`U
zLCUY&MAiR4CDgY+QK!#w__>WJ(yLnsM^)4itsOCGXHGX+M2h`eL$*Zh!E4vtFly1#
z6f3zh4f><YXmYTZDtpXhW5v*02T)8P#__6$l)?oAxO|Tk0`!4@r!NeK4j6EZW^r+{
ztNyKw3$~G*)fR*qV8n+sQ1j%Lbm`&f&(RR;JE3R^90&=6yn$XQ7~Uc>tvP`Y5VdTd
z4f_zryI)Edqgq>^Cy0r@{%Q+JN`x-1wT!H6I7@-nu3dzU_+NW%M1^|YnCx!i;TTS`
zycs`d&K&Nw_)v}?Kh7{JAH+~V7Ik&pk}oqyK#wA1Qv;@ghnEzrAn+dU^7nmM{{v6J
zm*>WAYzYRS@mC*oGo0Jz<t@G%KBq0a*WpO?KN8<MM8)y`$@h;ab@&~T$B)~8p8oB+
z6+|~By@MDvo8D8dRp<Y%>p{%S*n)#UncbS(iFgNLI3|{9alTh%FzMhYp}T~iCUJNc
zd5>d9Df7LiW(1?j0TCpBbZI<#(F#!Q8IIDqsVHhst^@KNb2d_AiatBLva*s-;4D5I
zL`B%v`||B1B+~c)oltnLG<4_++<YDcv=v+-GHOTh3b&5?Z>7C>?h+FJ2TAJxHw`?j
zT=f}dZ!Wz_Z;TCFPs6e>mYqdJs5nHbuj5?ZGSFgv(Jz~+9Mape&q`bKI}t6EmJb^?
z3<qe=5`!TR?$2yaNQO85W!q8W<}z;FKH0Y4!cMeakJd+f`R^2q_-DEQX~OdV-+$aX
zIwRq@hsu;Xud}c$v8MA{vAgbN63S?La&;@TBIJgBeY$V;5}gWAaM)-XpDRqwm}^Y$
z#MJpMIELo8VgLfCZ0AxBDP*1M@YP?_5b_h|X6@_fc%6@g%4^;A#N=dXxqq2S-6e)0
zrsq^rcHzG#CkZZ&qVm$EdmbsaAV%v8zLT##MiYg^HlUs}+8QKL@MBtl8zU1$j*d2z
zTS3`2)99UO{li96n?(Z6blPq52<a-dDqofh7n+HpDBX{e;X${~x}^Xr9)q`0njye1
zO02oLUPLi^Q$!+V*N|5AJ(=^?%pkTFL>h{gLYW)^(9M_Pz#XBXubK5nY!Q)P^7-??
zfB#)XA=WFP-Sj&ka9i_@JUyHFdwDGzsd{Q^#yMzg&0%MUQP^a5qY3*AW2O6=>eI9=
z5qY>NoCaKzWuHRc!;(A8<(uq32`-%)cFbM(`^bic2<YyU9xU`)zI<o1AHr#9;YPeE
z7-dl}03snBP+&vWyd}}~U0Oj6_v44VGx%<>U|V-zrJ>=$fKEMLM`!*NK^o0{=N`8Q
zQ~)-Swl$pI-KKO=FF^4wdTTU1M0eqU`k-@n_kZvoGgsa1uyQ3<@d}v{;KTd<{CeFe
zc=bvR3@c*<Pcj5SL7tcy2@O5yH)06252(yk^gzN-Wc+wFA&8GBq2ejNj3y58b=UC8
zcut|G585pj+}YWI$I=7Atv~!JBwM)W2;<_)*KPXk?kur1k)rY85)1GI|DT_!jUP^3
zLP+rY=^FczSBpMRY<H*bp>qi~h=o=+NmJq=joUcNF@%S~EKb&e6R6Wo;}3s6+f<Wt
z+5Yu-+0{)#i8O;}w?N~-$MwU-qX$<_mt6+wyG&lX-4HtN#2vXVz}s2ViprIw{eTLl
zvA-xP)3Gpdh{cSxc=}E0FbKo?gcgursmAG2^FA2_4WBeAQQ=-9UgJxbE=|2~7zrUN
zvlD-`wD%@q<la0SW!P%N+MBv@X1{BrJgh<nD=B^kntR?n2O;yE^L0EI#c8{UY!p6I
zp5nbSY*6sgqYJxqmwdzHg#C!g*7-^;i5TzIvnE6cV<s*`ydSU)<v>_CXM)nvcuz-5
zCow71F%1jAn0Ei%lzFvWTH;%B6&1o?3juXdXj1-)W_F=brK?*#2#{4yl3sFymj}Do
zXUXE}BZQ`s;W)B+S;72RQ#!oa$Cp0G$|r%yfDUBT^rgT{H4P1U_KT=HDU77X<{ewc
z;iJW8+3*@`e|m?Ml+7U0I6Keh4&MAup-I_EgYki0NKvMknm2En2UYo!<-<o?{;HL@
zMwh+rkHraNlnoD_>C+k%w%ypz!qxR^WaL|*HjHg*zx?8m@Ja`$wSY*nf_d@9S6txJ
zO>gOF@O|6V{#fVfom!_GtydsNT*y+uRsP}OOF&WEw2pZsIxXv&-u|#WYWROKG&7AK
zb*c$-#<<eWgPH)-4p;afnia0vl~LA)+#EI&BzoSk)9%5(Dm3p6ylk$c4YPt&+u$->
zK`=+p;>O76m>AJRMrf?P8MiGz8(;`+hnf}eI&01Mx1Hk#o9ta}9IZesy+I?z=bWzA
zLZAhdw|@5?;1zq9eEZkH1^l6uY~QWRXA7M1<^z6j?1GP%QG`(gw?DQXd<ijyZqL5^
z^v}z!jW=EHn<E4pLOd{|voC(yFS4^Q0H3y3nLVgiugefpI_;ELjB4>oH3iev%<-Qb
zuM2RC%jeM|3mgUtvb)Q_C1qtNU!SA#2RWyHH7}mEhe-n5YwytL7^p3N=O)DMYp->I
z3A}yf(m0p+py1#JoCF`PjQ#TYvt9oAA|fOYQQCgR`dHI{rKJxaKfd^#B;Ax&Q(5mU
z81;ckceN!9`h34eF20aJVxTixr*8Lq<?=wL=J;Uyjvv0~E-u(FPy5|4hz(l2Q=sd8
zXBfA{U>4LoQ%UkJN#VYaJ-dc3E%muIpb9(&9JS94*6hs<9Un`iqIvb$&*WE&V5!l(
z`Wy_T_k@-|V|M0lI}x>M+8keBR=iNoZwK{eB*|sjDeqevm^<=!vVV%DX>4FQ8`A>!
zUk7k0G7{?T?Q&R&v~S;@N0unjaCYkru{Qnc#+Al4zMVCe;pt6SglSk*`OLT-!AqfO
z{#jlR+?XRwXYu;a(8>{$H>`q#;yUL0{2`1*pj8ai?$&SY^iSOyIZ<Mz5g@;`%PNU<
zKT7ADZuD<R^bav-hmaNuSK7VkyqqLuI!bDNEP;TkSsFcY;+VUW_A3t`{(&sP<wNKM
zfK8VHmui3K)W~1G+21I1G>GJBl^Dj*bFhd_llRC&qCm!xLC${YP-$72?OBUywBl@Z
zy^@}nr|#&x>W*+7<Z<J9@9<vd5RDuxcVuATQVh@Dy)%XdOZweKYC02_%F`BFS|+EY
zAT*R&8bsFu>kj;cQzk<)7(sz?M*BUJ<LV(qHApQ2^%y=chIP|o&?LQgej1u^K;{o#
zDd`y*X>4dM`QFPadD7<Z_RhWDvV%jwV$>Ptf^?(Dpf6`ja2w5OIRa^{kgp^<_OYMq
z3PFHVE1MSKsvH;e7h{3s@+u#4$ikau=4VN@q$)rX^VZ^nkq^jwv1F>IgQaxwJDWko
zTY_oe@#AMxpTb+^JvqD8$C=>B6;)DHOs63N^^kaLvlr!DK)^ghLuK|B0j=NJqEZG)
z#Ims(`XpT9rG9PJ!<Du<q@$FS6buX>Q%vCxZuky!L*OBDqLk_fK3t)lm2XiY<e9cC
zo@Zci`TF&Z<RfS25w>xbb$$PY?507dJBOJ@21>O8-EgDSwG+l+2i+Il*t<8ANuc|E
zTBz1o_5&LCiL9%g9e$4x7}s=dNP|ts_OBLY04Zc-e1Obijf%cLV`MKctsn81zrJ}1
z^f~tYg{r@0pmq<ior_4M$+AMVTa*Dj0%S%SLbWn}%^6LA3ygR<P#*q0xcpq5aU5H>
zj{4|&NcW~KDMYOjg@99V;b-Rm!>9ulyN6Jv3HC!y(xc!tI!=tG7q3gC^GC_z`ee)q
zm;hRKJTyN+BB|ea3>(7gUB;F&(8yCyM1QbvzvXV8P(RXOlX#$q=3j}s#KWgSWsQMQ
z&C1^MuUhcpGL#nlxGz$@>Z(Od3NUGa=PXy*Jln(XK9bj~mcktPVOpjQMJpIVTgL&h
zy0NCR;L0@bd_D~3ZL1i<6aVt12-Jj|C5}*r0#OmHH^<J^&8?xfwpY6$!-lOwfD9<%
z{pXiQ@e_Vl6RztWx($poswgKY&{30P5}7=+Hd&?4?V<kIKX5XGT=Ng?wt6lmXq-!l
z`>2*IYVYW=FGOsRHDAHAXjq(St;fgs1A+F{{`5guvso8~ycJ}Ds+j!ZbRTN8adiBO
z=nj2Zvcw7MqMYZ?$20kKK@nj-fh&1he@Dvag;@!&sTUibg!0+AfuDIa4v-nOP%d;9
zg|sWHeWPkhr_VDp6U)QluTYgWHtJ~KcKI(YK=wMLza>8(Uw$apy2i*FndOA_VIifP
zH*fLKGcd43atP9Tb&ECi*F;93s2E_JUbG}%g-jOF{zw5wyD~2!AGDn-;&uRL3Z?C9
z(j>D<<|b13hhujykY*%jK?5Qhb);Ha;-5Zx1T1NN&29nPH?40F2))?SM2V7Hr@eC3
zDpYu$6wUk}-fd6Up^p0q+)C)CA#wAr7}1NX^=oMAx3|k!$N7{%OTIIG=#h&=GKq~#
zytZ&H?<r!bSs+t>U_Ky2e2cutu?a@bEwxTyfgC3BJtIdNTmkSXB*pTRK%14Dj@2-Q
zkk>pH14?R`PsQpPNID}2+(5%49Os)|&WJ%3m~3hL9nB=4b*FMr7G7P8Jyq1z`E4Y+
zoGvvQWo1=RqLorH&w+R!^y$crmRDpU^yAR~WzYT5#lzA}dCG!lGL}|z<<O+vj_cRo
zf?QAX;LV5SZ$M3%Ja|>HO`7EaXDF8DUt8<y>vJ{JUJ)p(UY<+XYzfy2p8=U!wdIJJ
z@8-L#nCac4N8r~L6glpE{Dlg-4QI-ktuyCsx$Vv+)gC@9o%x<Co$&{+go|EAAy{JD
z*9`ua2FG4)(rmvo&RB_z8m(?z+bQ?=1TA)1uIe>=H<Sc{e6~Xbk7)1iE-1GsBj2XI
z1<rI3#_Xq9&$nmnlqt!?@TOvSsL}ybT;r<a4I+|{_{#7)LCII_{_aC0R;cr)(pL(V
zP_R%Fb33}@n+TWZnu<2?Z9EHu^A%!YuvTwT4X%4fH1$_7`<>~ntw2S-Fgu}X!)V;V
zqAt;(gtgxeM|t17cKo8M!A4W|4V-LR^dcw6<jQZR0D`jjy&_fD&AO$SrXiM%(>Nxd
zc5`t-5qZe{Y(?)B?l(h&vNjtCH{|)V0ebLEfnVo3uUq#&*n9J+9@qE%yKN(6rVtv~
z(MTvm$Pl5TL31imDw!!!$~JC8hBOEj8Z<~sg$N;3XctAL5TVJ~K&apA;`2OzJZt^d
zdVXs?&syI<KCQL)wkz-V{l4$(I<NCQ&f`3eI&M|ogOk$6WUB|B?_+t03hBd_3pAlC
z5}$y9GDW5O?Rap)K)+4@mf4tVj2NNcqP(Tn5^u-7{w4;x&BXShZ`nSVOAd5dS5QuS
z7P4ja8<lI0a;Qf-Qu_D~S9iSjgG=?CmNJ{8e$J0eC&I%K>EG+@m(-z^kblbHjnV=p
za5iSZz)3QNOxW68edhmF5Znyh|A%+hd63Pk(uBY9oU6eGN?mSV`EaeADO<wQ)zuLs
zNv2V4@;)2o20qG~*4Az$lNva?uVVV@J=TE=eTO}^?4R^B!N4%rV$rjG<sr=(AsMz7
z6+x$yG9DM1eK%X~xXsqgPGh#7**5iAvqHZ89CN+kTZB(WSgz4n#e%zzM_R5H_*8~9
zoY)lpWK-5%c12F$DR5<t*a@QbKuo6y9&l?7Y!;FjDH(O;zx_`;S1^-Hl1B*bZO2$4
z_#sKPxa3c&Pe`8Wj@X609k%pUzczxTfuV(Yvcpt<Y5e#MS`$TeJFkDOk_wO85!8@e
zMyi7(WRCCj8NJ!zQ7nh3s2H_=94Nw@0`a&>HwjXcf5_YAh3*qZqrM7;3#KXOE^*b5
zDr`PBSR_H549=VMZN<2Zv~&|Gu}xENIus~8TcIIpxtnbIASOC*XTa)PwRc|4G?N)K
z*pN~zr1cGD+KJ1oPn&OFGWzPrZ3|AjsC*|&39fUdQ}8^b+f_@`Wo0^(ChZAZ1&RVZ
zJc&9Te*@v5oT&e>KxneQjY<4|jpS&Vv0m-MW$#v3UM&szLte4SYB_SdmMiP<q3m(h
zVByjHpu0^kUa}u(Fr_v2GiRKFu%C|gyq1>vl;R-rL74XMdIeKnv5zR-K!F(KW7R;o
zOVSJp4lZVvW6VbD8i+sijZW8OaL$rlkWvm}w%~!y!v_yU`9Mwdpv{A4#?QtL723nQ
z2%>Z57(RMQ|DzIR@1G0Ri3APsxoi4L9wDE^00}v!nN#{Wn3!l%wR3nS9$4J6U9!5+
ziesR(pJIaZ+s}~5n?kiT)ySI^6`hE|A!oh#3m=QE95MO~pS1o?!km;(FJ58=v<Ih!
zh-XECVPT>N%~8E?fJ;ak`D}GNViO87<|uwlqBO+_fig7u$uyfKbjlp@aQphUpYar@
zS5}rrM*Gi37gKwQnMu1l9Zjl+47w$`bxuiCaDW&h-=4i#b0CX#3j+};VqiGhYN7!7
z&!0E(kHe(uQ2@<?pGXWZ*)8HyKP`xZ?BU>7PnAxspd&;C;;YK#6%FK1-8f^4<Q<oh
zQB^%N+Lw{rH;Rxk0VdeO*`Ysv_*QrE^1X=|LimXj<W|DN%x0H4tQ0^6MGc2<JdyaG
zOOmY~xM^{O*&M6*Lc$^7tfHQTRN<ZaRFHP_W<-A|CpS#cJ!HX5pyKz&ojZ0MtuR2$
zn+)G3=cc*WbZ3k4nM&>+20*A_N@fRQkC~Jpb13yB?iu@T-F?O`WunIVP_^XGUC<io
z-xjYv!$Wf*SNrx|{Ll7oQc@O=W(rFnv*X*xMvlQ}jfeT%_X`nbrL)p;A*6Ejs5fPi
zp?*lXD|o<>8mUJiD;n7@`QtaNTUV5s8M9;JUxTyGpa1vn-8sJMQjQ(s0bLxQ`Ldw%
zdS<vrK`?okwz@~RZeu+WwB5F@V6eQrKU-pL;^#FV6*OGNlk%_w=YH6Nts&2zTVRi7
zJ}nF7388mR&acTLl{hxebpct`(~(RR+je;$6Y5ZO%!q?7>*`hRPzEUgV8px+2J_~H
z;ji;Go@TFAR!0LJ`A0(AWhdM!EDz_@ujegDrBduz?CH%?)9|c*N}qWtyPL17EbI`)
z_ZofH(HG2{QSbNhUlcZ8`oo`nT5p+frHR62Uu2gf<a*CDJq<HYRhPbFF_R=M^ctT?
z^<6a5(!8%^FQIaAr#n4!mwk@%$9<WH&YR;1l0tpNFDXsVci?|0Y#H=J6Zzlzdvs|-
z<$PfN?~z=Q^H6_+YyeLnc%@MJJ08(N?QlR(>wTJ7nB9$%)}LUL>bz(91b2sM^;L1&
zK_xqzi=^L*qNsjxsN>CHDf*(9l{_s->9)=**xEY5W&mDaDfQ!aU2pih17#{BnKL2Z
zki@<n)l3&Gt~upN;xa8+Uc#gHMSVSEOWlffp%n-dn2C?cd0d9pw;5W86_qF1O}Jp^
zzO#^>;SxNZ2poOJOubhWb~ol8hf}yK*imxqn|U;8`1|`;CwmlgL9VWEDX7y{l$v=w
z7~^8C!SFG$EFqjkj6hG#la&7ODaA;VU%bH$@}{8mK5xQYhfGawe~bN=`9mR*BZf@F
zyHZFgkNo%2+E+9!-LpbF&RDZ<o$O?<6%Gy&5AU)51O2Q)XbD@ZGV?<;Gmp+zOGl5k
zJhEH<q`&sJkx=gG^}3-U$)arLZY>|=)D23rn>91%8ijB`XO`}n3Vo5<yz1i5Y3<*P
z!y_O!m<RO9H@i&t4oc0bdZMb*vLGLCZ@1!4M_G}Spbxd6d8jVC4o&xFy)QWJ_*Kv1
z>f!V2)m(ndsi^ur`plVqZ+GL*!hsN_69azxkD~##f@PbB*EDYvnV(0I;4e)wzx7TG
z`Ub5n7bYwd(x&V`cO(~7Kb<;UBm~1&P|lh6bZgnZo4}aV>2EFy0=zKf@br!f{^I1A
z@I<Sz?uoJg&{xFhn$VG|s;YtETQ)VMKe_F*vTGE#&^p7(*Y|3G@ivPtQGb>U+uu_n
z=hMBNg=FyY#SMUSOD+PZh7DCEyu6*`M`3+CBCoc{r8)GxX<rxb4_PgHCu|!UxyDdf
z%MEcw$fWg4O0~t2&Tf_PD8f~y;u1h{6dNJ3;0lix$D6U*snKAR9$p+tnT2lV4jOH@
zdQJOhu&tcFKW^GVQ$<a*IJ?7`S2Zj%f|aFLzsFU5bgu6}lYTAd?ZS(k!@@nJbWT`w
zJFutfmCKhdUBzE`i5ZQ}mrxXj@7vaO8_V|W!aG?T(DJae4YrGR)V>tNF=adH7K73k
zACFj^E8Su%$g&7Y>7W*VhhsyrLH%d{r+xKDD=V8s__vatHhkG&eMix0LyejRbg3o#
z6)KX1J4Wd#^&N%s`P(IHZi_jF{e7IB>fpG!HwBgKFzrF<Q+0b!><qLpg5=ki`ffO2
zMuR{jVkIKvCJ{b9%>U=CH6NYUS4aT=s8<`X`uvA%J<#8ii*q+i+d4Ip1?7#Jl3<;3
zy76>Jsax-$h*>1<b^v$eiT_hUfsl~7|GYvMaHr4!cb-3akm?l+sz46GC4RZI0-&%)
z*IVM`Rctt^A@QkU4m<+>o~?(qO`pr;^K<(>Wd&SsDzQuLk$%m`u#3}%8@7y$Z@nAI
zR(@Ff1z;e7y1XVMZR{nP@ZfQsW*ll;-pY}b6c)Q(T3R}66~M*Qi9uT;*JyyG^xhZv
z0Z!-j3mVB~6xT)*P#!re{Y^Wb4>52X)t}?%FJEF>;E3(DdVo5-f&a~Ag97ewvgXM)
z4Hz@d17cL_iL>mFyf&!|f+J?z<i|7|9Nzw4PH`vM;7#O93roKc%LY6TXob}T;yJwS
z2`nfiKemspuB8wJz=g2ESXo{nWmOcbch5_AL<7KY4f3}h>n|fyX8pa`i*xUCn(D=Q
zY)|q8y*E+FWmJ-s=E}lS@=SP87QUdu?t(zHvM$Y19W%!C*jqt`_6gp+(aS4jRNCUx
zKaNiGi<UPYQ@K*Q#uA__<<<#<8%sBRRQP!=<L8T06tKd8DD$G=kFK=qDR+BA1qV{i
zLB!s7BAI1T=YW6jr-YO66i7M@K4{AZEcw<%sjO_Jdx%Q=@Icv0N|kzt?{ZUsxMr8S
z<!3Xlo_O~5A_s@c!K<%2HB1DSvMSNzWE0$OVW5}+!Eq_n#e}*nJBRt_yd`@9QaYf_
zclIDQo&ZS^I1YU3t3**Qj^xW1uF&yIT^Cv*?Q_@F_M=`?;}_b3XN<yzwNdR}CUpO7
z^vYLx?!#@(_?h2vdafcN;^1}CqcJT8!ACQLLqcMwnbW=ldr5dS!*-%Y`;wy6SU{`T
z!N@VCc%@GrG+Fm$tC6R9rx_C|Cl^exbI^_|R6e)m3E60phi8l)E9uRfJih>82>PV`
z8)K?XcgLuW8@KTCcykW=6@xBtLRPs;9HA`V!b-C2F-%J7=#JgH)soAlJHDhyu+)tq
zN*Bc}@&w(MOqE{;7}x`G^X8yY8#a-uxVWW~%O$v&K~zXqZ%+kXH;lEK2`YHf_J?`<
z5M!0&SR|#ch1Z-T6)1DC*5LZE?b*F$Wo2>A>f85pnG;-h%GN{Lqc~}4j8sL!BKb=1
zX+>nh+;4p{J{f*KtngFyQNp>C#;w|cn6U4jR9gOO^G6eQ;F!%5X*@_3Y+YVcnRVr1
z*e*`5J=$g~zAX~nw?(T=+h3fbZ%cnwXDAmqx-ukdIYmj|UNL?9_PzhmV2bxd!f3sF
z+K24zwH=bi%T2+PZW?D*nds~k8$RC<CpV?LyzU~StG!9qu?n^?-2`+3v>h_V@>5Sa
z^MxIE0e)ya8*OHPCaZr>xyFsY*REZQtzK!7c<b$(H>yVl!)cc6A^Vs8Fi#-cya6#!
z&7b>JJKdF>lDX*q#y=#cRME;m{d|Uczj*&fl`b=JqgY{~Gl+X6E$w13A~hy@dJP2O
zNEHooh_>aGSHE#23&2di$CD*@M?U)t@QEK}vI9W*Vy2DGxwW}Ca_2BRhjbF7z0*#!
z!EdK+j4F)Vtar>1)lBX3es|O2Vr&07_4g7V%vcz*Il)nzVMG>68VUVS=D<l=iNIxj
z48DE+3XU(-=JICT*Y!lX#^X_JKq@CrEv&2{ZG-3TP-RV4;q5yrV)S8*nGVH#z<xpn
zep-gPRlHbm!6P$~Q9%E!*Y5jsYM`Z>#MX;jc~tY1)|pf5i_~BulJYN6+{Rw*DXTR%
zEa!*tTng>kMl_-?tJYRiTbs##aVeFzQ=WFST2aL~^IF`_$(;t(ScUjKdt@s2WV<9r
zC!i(AuHE9jo}!5XRMo_^dh0RUhk@I7!>r1#YMH06uWxHxMhKs8IQE|g*h8_8G+HHl
zCA>6^=hSECaud7mwVJ<q(gG)^Q>EJ_$LyI17)_(|b>$tUwI8W!h?Po7;r8a{jVRPX
z6e|2Y<E)PNJG^J{eycxWL*W6k)ns+Ocm0-+y;<O%A79kgs$Pj$El%+D7s5*pFf>@h
z!Hn9%*?3<Qjz<7>fMC@+A+1B8oad2>HTx$}zp6+vBEi0FnCMj8;;`xIr#}vky7^Ax
z<JX0wV*eUG!bSO!kBfR&wS9?;Ne5m)CQVS<pMR3io@AYZZ#ttkNP!4Be(FF7(SAH)
zAk#tcNo8bWkBk`Ml00&Xib{>ZHJHSso^M+JEP70TB?5`N{(bOJ3n_UImBO+P3F(%N
zx3oe&wzNc&@;BC2kSSn+y*nX&xJ3o*xze~}uazoex<@7U95&oxpF)&u2?XfJ^1hs!
zmi8;dj~%N$B$5(ASxwynL0!ggieWZcP)r}SBe1Y4@0`{vDa!LrHaT^ZY4qr(2V|JY
z^hc^Dw&Qqe*t$}a4+s#yw?%Bikge{MtLHo|OE5S%%>DLxZ|SZ^r!1=7T<$LQ{iw-_
zL9uyUSC;mpKU@VUlCS|t-yn-JoTkzKqP+UWskpdzysraesw6{AQLnc>knGiYyW(Ge
zy#|lF*<LcDZ(GZYm%&yJh4OO{zL~7lGJe#l?Yl)c9XX6lj0fBwiNBdxIyc@yGM>j=
zU;pgnj6Yo-Uuo<m7hQeyzL|#6YDe42*We4nJlp5r+S)4giMXoJUH9fwa{aZdkRF7p
z%=q83W0pfZ+s$msvXb-iy%8k}kST3J`!b}DmwfgxBUtxc4Mkg5_4*EY3me(S6uUgI
zeTwIQ(M#97Jihu`pN@iID9DjmDAmfEm}N|RNq>vd?K`(py*~MDKMj}s#ASxcKdAk3
zDofOhTY~N7*3yO%r*`ArN}h<uX!oSn{WU?KP3GGM<qS7>&+7G{q%O^Q&4uUzn?qwo
zHvzt2>K)o!ME*(6x&~~yuMF)t2Xx*(!!tL%Z%lBn@2`&-=aqDk?kMDLNi>37Zd6xS
z+e2NUI*gqblN$&*8c-`{Y**87wOzIy3Je_9d=@9$b!!(NPF2W_N|j8Ry=KjMfP{vP
zU``R?;Tk_@X6(@F3p%4cxu8H!?$!3)0Ihu3YuM{lRh^)QBDbD|$4LASbH>=vB5M=2
z;7L!He$T!Iw}a{{djGlQDX+4(H!eD0GOCvxH_OHL^}-dD0&P)?{?aQZG*}EjbFeg9
z_<W-Aw$~~%Z>H^J-L)&75AWNT57ZcJI0zj8gR2@>W-X={7rSEcpyEfi#0}|j55(|D
z{=F`;E|=tjS5G_XD{}tx+tMbURN7+DXkGdqE{dSq()2*^(hK%7{js7*J<=Eenl~dO
zeN!2O`~$tfXIi{B!bb{;*0rx*Erc~h!_MNYwdvX0j1<}|Yz-XSWRVwQHNQ1cS6%&n
zy-LUUz7Mi2dW@U5&}APb_N(jdKQWv<{dC=j%BjUV(Fx|Gl}k2tHTU0Qv+nQ9CA;%i
zm}Qi9XbD+k{<FH5?$_z1O!sv3(P(zL0s?0K?U*in9388_f(}6l#TaMCw@MGPrql~q
zs}TEx3MW(tZXCg72bcI1USu3I<@9FW?b64SfuDyjI#&F8XbegaR1EO)WjbcNX|-_<
zV48D&dB>XN-5?BkMMO@WQYqO>kRagIv!~!FWGubgv$U!ohy*$BJG_#qTU|0<?9u?#
zeNNA#rc)30r>3T+vhqDFoB*AxDqr$MVqMD*NcKt%HQM~K_u#=E#a3xs3}@%Y`Nce~
zet{t462Kji!#)j!jOwJjqr~r9$`o(qLgS5W3yX4#jlQ%s^YoLlXtmMJx%k8FJWsQ#
zAuUa-x<lEBgb!Jt&%gRp&VC2OB9`a>%9>j?Cq#OQMR6xXzw5y}Sb^wzQLI(w-@h+Z
z=!n7o1Z<ke3vGglSDa|~em6jrwEh<NdjF{j=v}5iaj4_WSUz_%mq%TuIr*eU^05sK
zQD7c+KJH#Zs5}{ADD{zL4o{>Ycl%P%7F{jtBqal5>1N<c3#)xaVMlB2EyK^952<}Y
z&U8Lu?!Ex2RB6GdK;3Gz2V;SQV1L&b>sw%5D>ybQtn61LMO(y~C?X?pdB;i1E_&%%
zK~Z6O2e(_ld-aLdwrjU2scrPvI(yuD7mHA&NnX#8559(O1$J^h<&eI0)43QF7`To+
z0Tc)eMvG?$#aJwmC2QHXP^&!abDr9g?pb_8bo0p;mQOiHQWj#~e!kY?h(+|__roz%
zux@?ZCtW!?r%y*kZEcH1-z#K*-oFx~qCecA`(vG2j}4~>xo?ot2cXUL+VY^o+jp}d
zY%FXln_DNfru8gVf1W+bH~DRCZEak-tAzD70_KfwMTSGue^5!q4mO_hX-9oG7xxkS
zN)6pC@@IvN0Jf(F*ichcO<ubI@M^5f-grwrQ`15+D~EMXjl$A;Llor&ais8arsGH#
zeKwgeKFQN{nFWgF;?zjTHGkCb$&)9Gt~<Q@C5C=}*J?jI8UV*zjhCCaW--GO2|q1%
zOF!{b@0u`9xEU@#eKD}m4NaaD-v<_x&B9-*C10&NlpBW>VRF#lXTZpco1PiWpjsNc
z|6EgXLYdsafo|>Y?)UEB*GjTbwjg6QJHGW#j#5LTqIUKYlf3px-6RY|*jbuRc8j~G
zMRT8x*AA3@2}#_(dt`)n6$U(r%S9@5;lqe|Dxk(rGtgHMTxpFTvFRnCif}`mGU*xk
zhcB!LS(CARtNA^Fv5zy+(MMMq9+c^rldV_URY(ez)2GX*+@IA`PQa2!kBVotS0lmm
zoH@qT?F_&j(dM%&u&heI!CFdKe-5Y7oFnPSYO`-~FuA{Mt%PWjOvO61&M0B;fdeug
z&huy$FWh8S^3rGOjNZ4655*CVt1Qk)<SFPzN=T^T)a0<~#>ik;u5Eqsbqc9ad$2cf
z)U=q^>8)yxT`gnJl)mFbTBkbCBUdo;Cv2Eoc7E)@;3dqf6`4${b_pN2nMCgMfKXM^
zVWc`f89WtYn8_P#^Mdl7RYLcwW-wLukB5<s@V&{qLvt!&R8zB94+tNYxM>cIpx|9H
zC2c2(LvWu=ni>!5M`CbB0g7d?@cYZV0tV-$qTGIE$rWe-FfY_Sn0@&Ec3Qjp%P#s+
zrjztiPNXyd{!XJ8>2jqy{+(CkspYv}AqFT97HmdF&26F%9yE5>PqA3)1&A0k{twB`
zAU3xrz9J_k`S5tTOF!xpZsNV$hrqmAzAWz2m*VGNx@}wB?Q5-aWyHm2a#Mi()`2`%
z{jj(`rJ;(su&M?GKA-DM75Yi@{BYM^vdH1g_JPSkS<6Yf$&g%gZv{i1?^XS|yVn$N
zZ%p>o<5iNQH-G#Hq>vy}X@ewX>KIH4UZ3+smQZ`k?KFw4x>1Fq!7k+C(zoW4fCNw0
zm3lNQh#rweq(Dd11_$T+bcwn4$n^c!yj+J~-W$MEIR*>e77@A1S6@#me#!ln@sFSu
zvG7{Gc=1WJWmTS=v|nrmG+gO5cFAVjnnl?wH3t%Jl=V;2`%jk{o4Qjg*rNUIzrdav
zdwjZ$zrLIk8Worm3SC&9g8M`hdP(l~kZo+&+`a0IfMo!aJ6v7>M{DoFgBQILp*n(-
z+MIh%o#g2Kawj|1aJ?$wMT`UvQ5-eu+^ppM=FgwOI_g-%F#+T?@6aiMR1KgQbi9B6
z{%294!&&LyJ%7MHTAlY?bc{1zPsmVWz*?``<^-WPVS%-UQ*FpJRDTpC=FOU6?sO{E
zQXhVTM(&aH+tfeG5r6iBYXKYK(ak>fy0;+f;`$_rV}nQAXKJmkoZ|j>EbTUigH$Jw
z`zAJa@HN!!31kj+H&mP1vVSH8>eTI14PAxuL7?TS^1J)97av6tkSlX*{gQOCvX;S(
z<rNNLP(${;R5Ugg&UFc1i3`shP*D>z2dql9148ZW!rJDlhE4epHW;JDkdQogXW;#y
zi$7jCL7kI1unC1;23l)_1_t&l#znKnp7K8cCXazZ-;Pc~$uxGpTH!4B#5BCrH<%Ak
zeU;+{W(3*b2lBEW4ZDbt-2XsrkpZl}`ibGyanS*yd6E??v=mBo<19dy2K4T&Ahiaj
znG2aHzi*bbQxN88-OwR4&KMkMi&il8kV^p6Ou^9n^kx!gEs+205+nikgV7lrEEk*(
z%Ean7qK)S*ST!N_)UTbRfg;u94%9BPkM{jQE|ph5yu%s=$|#9CrjWny9N=K0p>vo7
zRIZy<f;mWX#1T)h?!9~Wj{W&Y&}2ZKmtXqG`-@rC(y)K#+*;<}d^{KrM00cJjH^+N
ze;dS&PfN=bOsn)IB~X-F5K~c?FA1?)2?tipDYE3|Uxb32)UQiF7iZ^tsxkP7{<A5a
z8Z@JQfl23;C{eLdOOWy&JZHQ+OPWsuy6skMiTbm^YDXwg9ZrQ?36+vP3A0>9#hqqb
zLa^(;PhZMyNhzD@&Nxq6Mb>eS<P4ug%02!K5XT;3A_tGlHg=6cXp-6%3J+U48)pj%
z?Al-Jlv#h4<MC{a*IMWpOwW6n>L^8R8e!zuVx>NLGC@hH8WT@5neYh%yQiM8EWSN8
zjE;(manBxb@|toFn?5`Rk*NA@S+k<vE|_q6jXRHki~&2sBagQ<?G3zcuvuG3SS5H`
zqA6`j4JjiM08dh3PS=Iaf=_?Ft!eayFa2AtQpRoi`tekCj;mb0sNnCj^^gDDzpt11
zx<rF}J>}A5q@SX(W<Bv?Us4TY0|PUAyLA9`u@iz7x*#qQ{W(hqC1H;9uG<bpMovRe
zpQXIn+xy+LMV$HE2<Im+gKuZw;`CAzk_D-+CVdqllWE8c(XCg_xvDFkbt80RPE$Nk
z>b0;^7@*Xr#3x;467}@eh?7Kb-u<F`<(o^FFK0o=!ix1!x=@v6SaBi1Hai)~<!;@&
z@eJ7C4BNjaIV_GbF*N7d@LpnmcqK{-zO!cC1_?l>;RpR0(krp{2Yr8!67_~LMWk=0
zQ?A5gr1zTGlzN9B9#5LM5%P6)e0rdoY}XsQjAPZQBu|vt<l__C7-cwa02o(VCSq>K
zTKA)m=6?g?#KHnJz<!=G3fsd`CkCp|7BHR|?Y^B1!DJ}t(xxSmd{iumr<W+IsN|-d
zhy@sJYyT=LDIGo{Lj!f~2U{V%nUJ2*5%>bl!gh=3%e*Pm@+dJX&2})p^i|6;&O~@c
z<_Ru>W~bDgGjRZyI$u(LS{!*Z6`{zp0Ooa$KBz9R|H1#h9`ygA=M~zxg&|<Qq&{bh
zAT5+iu3_!99*-LlPLJHfv*civP)fR7%x((q<f@MPfkO4xgr-Mye=Xkrz-?J=vOL`G
z5}|pSbNtlFPmXFwu6LX<5c+akzLR3{LMd)*`tn1fYpg)=k&)khethg$;uwU#^<TU1
z3q6gV?r51;Ddzp`h3v9j>Pj^7R_HC9vArnTB0vPin2+zkZ;M7^u(=Vg0%=oT1T{8`
zAXNWngn!$n@WV0MYNWhEm?T#xIwmHpWQ6oaHOS`)dh?9kgfL0_w9WOs<j8{uq6N$d
zEfFozOIcM_5X2^~uYUP=+VUf{R*6AhXNkS#WkvsD-~%>rv*;;5^93c1-FV46IG1(e
z`DQ7UCA*7sy6$DJth{RDyS7K?TLeV&k?Qm71OEAEB1*9(*OB*Y;53+kIS_N`hbmm?
zY+&tL9Z|92)UKZHIqM7k+rG#b3rF0`sTa&)XRJ#khmRlC)uU_iYQrn?Q!r*LT_>y8
zbUlK!IrS}kx<<#ITLeCL$2OKQ1_eXq&aJR(o17UT_3+ZGmb(B<m~(WihWWu{p+904
z(fMN+Mdol9?*w@9qX8<!qnJ)7X9iM*5l$!PC#eBRx)vJWk2Szy6Nj)uwj*~K4yUN=
zw)3C|h|MCS2u3})`!0GMt6+6BUSK9a8(>9@xjb~HYIV-^{SL$divaL}4%(@s<Er1k
zKes+ZuZFwDsU?sw{sTAW5j_<)1-w95)<CA{ySQ+1OI+PGY_A9_Z<TEB^KOEen8=`c
ziBhKf=Dn==<q4*h1bU*b5Gh+GnfBA-II}hD)}^}dET;Lu{=$M{D#bLr6^7twYW4f~
zh3q}&u^-SY46#!?2ITe^Pua4|60WD!#VM(;oSd%r$Hg{f$83oGV82)e6ef2%i@FmT
z=90wmDjC29v}f5vWo136iWlUz1)>p|C>ux`u+DvfUSzFOv+#tTVs-H~ww>?TW6L}o
z4B7JwEXTx0D~%j^yjg0PmDQ8LbEYOG=J94iz?cv`PI_@JGxtVDCt&KMvcvQVM8aa4
zX0b6bVt$>Pxk&XEAf<hp+u?`zgb2J(ivzNF3SA(2qCAm{A_6!F7!^YxLZ6-99SRG$
zQfZDDflkG>!a)iA^{Q?G{2MH7g|l-sqbCDz1?<^l8nV%Qe=j*P!%Z|;N+CsEvy$CO
z_Jq(ZEo^dov`w0X2H@<lt65owqZ}EP{#xo9sl6pe49gRqIkAu=Ll!>w!2@#adU%|i
zrK(+Sa%E(6aYhjqs7>AkV1NhrjDSwkmO)lfUOrh#$%O+1$%WomXa|;T=&A1xTNS_}
zHuM!4H4|NR46;y;$&O!PP2vuFjeW<xYUQ=rVHeIXxzYgnwg@My()i~rGQvY?oudB!
zfuArK1^y~bNHUW21mHNI@$6r5)D_FF^|GRGMOng9j2$Qy$>6%o@HG1=i1@G?<#;Db
zn4QNbPM%EpbTiP*){9>pVUy3KlEL~85_@kIuJx7h{wuhAMI*qkka%ua%>kF-AbB_C
z8Q6-sq4Hk9Bo;_HM&du)$4r=Db>?yG@@*M1iGIC|5NWHw5Z%6})!5t5uWh06)R<|z
zfnWt$#%O3!f&N9@c&)94r<$EZL>0irkpPpvrSAm%Gd~x3{jdpVw0j)39G_XExS{f#
zPVRQ{pNq9RR<;hP7%=Zze9B<h;0iY_*it~Z6$z2)9K(WtW-fp6b!wr-(su$H@;VWi
zFUYlNpQB8<v&_JZ9ea!YVN<<iD2@Rse+4YCNV!lLY;hrS)0?oW2iY<Q9_UYiff-WW
z)W~Q$;m1AqJW7DJJ9UjXSZQ+#pP)yPwx#d%14a)rMSu0_my~JV=dV7z6dL*jZg=qT
zr)uKDL=%tFv-J@u{(9IAN2er423%YAK=#7yb)CBpWFY3`xVJp#4a~ybbOq;__f4I=
zVY`aPbmY9v{QD)Q76$0QVvyalZDH}%6-Nj;G~CK>x}F83`lan*qO{@Tp~Htyot2Kg
zm6L<j(HJ@;O=}IgF~U<)W>gC-f?`&?^!LB_h6bQ&y5-YgvkkPmml#KR-_w^oYvGX4
zqk}}f5(H|tD)Gds(nWJ180R%|CZ?6c`-n=FSBPJxCZuRdq)D$VHEyfm_axYTutKbL
zH)#-2c&D-D8d#N}^B=}kMZ>{3F3N?~%|2Tguo_mK5`zs&N;YGi4mTtpJ96X^xvwbS
zZOyG-zkFFp$C7B61ppFaocxrAs^a%8kj)m)+jY&Nm*Sn^vwi#bPXi~WIy|0K-1c~(
zu@J)orcQgka`hYt?$p{s5b2*j!y5;5j$^;DeGU$pMDs9%=u@nlp!i~mW8^3}74(@8
z1~u5KB<dKr?Yh|>+&57BfUI`jP5F<f1CKMaCLGWiJczk|eg>g}<L$h2Zr=;i_6@6O
zE2f4J4uU@q$?M|p-_G^0w-dQ=o!K!{rcQO-Toc1Eq+rm8bRRrqh;X?9f;DTZ1V|X7
zv*F&&*$U7h_bpkZ{Lxx&3IevrG&bVH(s+Md0j-<+^<CEuCWl=KYdf!#?R*k*2cL#?
zk_ubTJ@?7u#BhG`vG1IWq}C3Z8{+-ut))pNMi)8X07}l-q2A6fwsUp_`h^&^HaDpG
zykx&x?(gm~loV$*J^C@@R}`Mkp>aIz>x(~K!qwpYU*z<|#eKks-Z?T%3c%)img(t1
zKHqpjjc4)cd3#$62Nj>~iS4(H*IvHlJVK`fRBywA>$LRn@PaaVxlt_;@{TU-*fLzd
zA2E=`2`%9v#?0lWG6Pd}^V5=&U^7&mYi7D%sMbboYO(CU1qB6xT_Kh;kMIM5C{nyD
z9P{jG001!_nd~@CYMdMA%|`fo1V9LO!>JrNi@}>FNdMd0|0zgC(FVJ>{wp5G74pgG
zWu%i7GiEh>+_*>Pqb+)OEPmEogI~W6myh-sg)n*}5LD9Hglthda(B@=012iJqU30u
z)&Zs!V@(gZ48aabJkMYkhBrc{gZK_4yRY3AuB7`A!#108N9`HCAvn-~*}zLWLSf{1
zC#9p!Ev;()@yizvaDBA*An%mOE-7kB@-vIJ7-4i&!LCyy*-6ReOdFjZrq<R+AL<Ed
zWoV-bPmb3~nvgvZEi@P3*h(`U1Bn*u3hD<-aw%vZD6ui_i0Ke%#J26*Jv-B@VNfe?
zl1GkS&c>Z9Iwob~F(GWq59+om(~}d(cCVSJ=Tc?@{tKxuvKHtiM+2@ZU#J9;STXwi
zdbT;|TYlwB1}AJ!4I7#_NVfYf5b1*0Z(V1uTJ@gg=2CDud$F>9)3?5idQ(G+<v<I9
z0LZX}Hw6e6PH83CvE!`Svl%BUci5+b?Zvm|af46HH!uj4`9P6I=JL7m+g&ph3@H!r
z3{PiZm9n%F!G@_*vrs~Ts>wAlZAZ@48eeLFGCgViL|D08HV}SfSlFO}1FgxQ7%<zL
zSPO3Wys=yDjueh@|JIt3j~4u30Fq^#>EC`W{{B9NE;<)B|L7^l>H-3c0CeWFKId#<
ziZOOO6t50sYyh3784xs{+7$FK^eTD8*TZ859~SzHDl+lu1qlvWt(0o)sSJ)r{_<2-
z7$)unTlx(`G}(KzI)kKBr`q@K=x(uD16a>!?p$lGf(+WIU=yJd5;P&)rgUbNt$4yc
zWE-EYFQ5gqoSZebWxxAq1_OG_$yKu0pyDST(Wg)R;?0|8k--3W6y2dYZ#(kOlYEZ6
z(HBWg-I^$%i3ugP;}*5+X!0q<0zaxLRI%9yB4t5B32QrCHBj+qOBZqn0f{fTE;np^
zaf;Oiij8F18*XGzRDh6rIed66M+V_#>(;IOGMw-6qAz5zKX0fb*V!Qb7$f>xDm8Z5
zE21n}!VAdKj0T15+qaLk$p-=@fs*0jRc&|ScCVnIQfZI*F)`6W`4>AV?;CQWM=yzk
zBksz0s%!}fRhv<2>gS;ClDUyoH)^ECdpex=9Ciw~hr%PR^2u4v$5!m(!>?_OZktmY
z>C|-)V}SMYj^GV?%gki<W=>g<SFq|`ij&3RusLOfOlKu+_C%6xf7y&HJ}u`ijX1xD
zBO2Nsyu`j61wMKriz&_4ro#qZkl*+fqa>!3@kvK`giYRfj#pg;Y;CxG^fm=pd7~8-
zvrWI4DkWArptU-A#OCtZT*W-VZ%5*AS`4Q`Yw^WDZs@aSxy-{E6lp4nhNosVAo_3l
z*ahy>GQ%I!I*VD%($8mBXElYCcJz{P`NIi!iH~kRK(hq6kt=OzCa|a);=Tpx5t@Qz
zvV?P8{e^-d6XjzV!zVV)vt!3bFRIA@JFl&~;XMkUwinrdy62;jvuA{avO8M{`H;gI
z9I$sykEldK0T}gjoUZP*DQEW{4h~Ky-*-1;<O{;bySQ%>=ezy7gnP&N&}nCvJ2+I*
zy05$OEoefj<4u81wVC}L^MS%s+GB#$Hcj*zS^GL`bPZhmcPTNJTu0^{l7S-KIRe3f
zW4*S1zQwd4xmZXB5q1eFz=31Nlm43O&48^V8Gfm4?ZLm{(JpgwnP;cVAZ&Q2m_LG4
z)W>wg0(%n~Dkr%Z(tRo@yi8h`Ut8B-UGb;VGuF3g4zC%T5IqjTmoU)SKSxF-)EAup
zOPKHRMn^L4ovxKIS?92+I+!)j8D_Yml()K2pj&rCC0L$ReIn*kU*7^5u5f&lS+Az9
zK4opyjt$yRnBj*SIB|`E#Mzkxnfsm6(|MAMvNlz?ASZR}*0*DJX{k__JP>hsFSX0w
z>$Z-Z&Gb+wdHuc>r}+4<BmP{|pJC2l%l3lu8DiAv_f4dZ499(V`Yos@bC~D6CFI;i
z=TE}KmmpjQ-W28PwCik7_d({(u3aRyX2Fl)6v;2V0Qk}yTbQE*cu545yUg5l>DlL$
z?8MFzYZScn4i3S@YRrsFT*HBIe0E)W@F2HtB1Q9R?7vvsL0e+$pDBYki#q90M<)r1
zwDUtWe@SxiuVe5>|MTaQd+2ig=j*@U{(U3<zrQDohD^l{>=`6X;UKvg)u9{Y#FG*l
z@TGY}-Dt3BECN7r;3sp8|M0#o9{XfyOhTe6%9XWJKO8A74UGq4QAlhZE;6eBe9iy0
zmVW)uzq7>B{r&a7`*!E3!guc`N!*C~HHx}v(}4(H>0lCwogl%K(b4WdWJo=6LE__6
zWaSvKT-T_+b!=WvVLyqjbIfS1M4jQ{U)^U~pu`OPm5JotM#4t9ddP1XBk}(KZ?5G3
z?7IK^+rOUce{bxs+aMwFzgyq`bJoeI0}<j{x2}6jdP|?}NzEE4RyO%!q71pnq&8;W
zIJ7PlycWOJ?xHCha3DfN<*lEUAeR9`jUHX#@F$aQTPUKNWrhFyq0_&G$BN$6)s@pc
zo7{8a;u{i^4~YFYw~w$8lXxLfFtz1G)~<CMH~KTT(x5HNHx=}LMTV_^Vz*NcFg;4I
zi4F{M>!b7tE)oX6@7lzzyLZnao30-lG_+s2_<@pWX-P;7xi#wLYPb$$(xtk;%*%KF
zJ~ZwE114e+7O<V_9>Jf&QD4MgGyl(}3}mcHz~gIst7h`;lhObAzUKJO*|gozH1nDK
zKtwW=Ilr&(7H1zrQvH2|BM^Rb<!g2M>DIm?MUjx`i9~`H#n6(~3ua^2DiPFSd!gHg
z#v$ToL5?)5?BHnaO>qKs9KGlg%*uLQC4Mx0?}}f4{Xl<+@UpVD`+Gp1oz-20%9yRc
zKZj%;)}v!GQe$K}NF^3e+AA(sV$FHbGBF#NFdx*C$+k~@kM-jF3bTb9uY`moSh0h-
zd632N^Ow+5lQ0k`Fkc5;Ev%m{*6g3$Q@ol<_x_&Zro8Y5N<zR=SzSFXdi|nB&qRxH
zvjIQZ+O_oeZN~CzIpvOwSL|t_FK5v2_v^WA+nK$V8~{0ozzFe@3W5g4{WOA(f&7K^
zeXlyobBaNg`(g=b3fi~$@yDhnGKs{}Rb=(*J9GJ-#IWB_G_TtQPkXRNG+^y@j)==M
z`kfLetMRRd>gt$Brvx0QYN49WSs)?N;Xjvx5aJxrB2OLbOWjrY{^B{m*G`b4&fn{P
zFOlFk&f;(U{p;3l9a|lLCIYQ}t1LA|#Z#bHF;nFE>F*oAb?j|)E|~mo8ue?1+)(__
zRoI!|suoZPwPk#PSRuWCouFGik&cFLTJsL;W|LGVaM5;%(LmA>b<EhPH;99){Ih@m
zBHS%FHm7&uqP=g^WXA`kLL8kZ)1VUw1K(fAo{}=!j)e;)P98WtN?ft{a{GO2#UHnh
z_<gVb=Y9LM)GeaHZ!UniI==qrUhMDd`~Tqr*R5?t<WNH5=?nlRgo9EABHX_}nwfmj
zZd4$O@&yOGk-VCm{`Bv~b#OaaCzi{$PbX2nz^)PPlf%R3Fh3F+6lBiV=|>_C%h~@Y
zS&}#lm((muzsH<{4iYABgz?;G&CL6>K7@8K01{pJ(v6m+Me8uTr3XREBoTM=SCBBN
zB%TRS2RwwDe8ak`2?*y&e4I9MmI&GkUADcbsrd#w9g!Cbr-H)uVl(FT{=M|tBDpB~
zk79)GK!o@ZCYI-Ki&*n<;$)ER_^Qhk+sVnva4m2GXd3}(4+BL*^oc?H3|$WJo72Pe
ze+5_X6Z7eFP%egKbd;1tv&aij3hYvQVD|(&;XO~8M4#eldat)Y<sW{1)zuvbXaXzY
zi<16+DYK{D5&Q;;p$!4Tg&IfB;JXzSQ`t$JD~}!tp$;1x8ww}Bm`Wc1@ZbD=VUP`_
z3?)Iq5Mio_QP6;;+=m;czaL4pAtZX>4hR;~g9d$r$}U|0(#s7;XEh?Z42Ae^;24Hk
zg_9!_YKWK6(lGdtm!CfwGzeXV&#=)2Y25z=7M0XRHcKgFxfB=@X>Ok=4_lb5v%N7D
zW&jZpVWEdSz55G_xkegFSOu>D>-8v3h{=pz4zm%Y5W<%)cq9DHWV?%_h7Z3Lco-;!
z%CrR?ylT;iECy5IwA@UOZP{)<%=FSF9VMoif&Px3^MM1xMGykaHx4P7@@4-+ONJmA
zZ1}gV2Ic7T`d3WcyV4b8V0GC>Cn^B}ov7a%g*>)t^{&_#m(HAd3%0Uv@7@h}KESza
zTR}tD|IPhD7ZwH!vF^6>owqwN00VH7>YT>d^zos+B}{bbY1>^n(!oT>dP#pCz4CQZ
zk0(%OVV)NIAUpa58Tar9w3U1~%-%-$Jj}`Y;du@~89YD-5p44Lv*E6b7ry|~qX!vl
zvGa~rTQ9h-4A*&hdNbBZkCH)~<qc7FU%q@{-B&PM4}WjzJfR8p9_$$+esj23+6Fy;
z9T_JVegFR9I=z&5&*TMuEr6#yWQ%w+r9`=Nbf>7Lg}+7hafkq*NJ(H&%X~Qb1>;o=
zSyoE-@83w@TbQHoy=;5q?p;^lah((&2&fCVX>EYa^^*tjenG$Y%=ArIk(@>!rT9u(
z-a$dlWYnrq_9K`kBzCP}W7EpPy?dXh7sDi$)S0&*J_rR`smYR(lK4EBm+G#2J(pny
z7as^646FCVIOW(!p0tLq!i?Swm2)aqVEbC^Bj_~<5(WiDYrg@p#Zzzbw_9x9|9ef_
zvjjP6EIvT{F&mF#IqGrN>4)+`8^LAS(b9tl(>{A8tG!%M7fMC{gk`L8$DEfIrVNAx
zXaz~S2~MPKSJVcJ-T`(HEIlj~kBwLaTY~=ux)Vmga57D&_E5lu2E+U9sZ-&Tju}%3
z$!6@>R2miFQX~QNip`q^T4J&NyI3D~L+nq`Js%n}u#-fbYGvQ3(*Qm89#8rW9H?A3
z1qOF<1eBG9C8NL>*?y*84|xGgM^pPHk1--$%!6lH$42ntG{4OnQ0#v9@}=V7Nic9r
z8X%SE1eeU`gKDErRNUbcjiS6U0s2dMa<phu{n_qJHpkVsm)NbYDfmOe#E;l!vP_Ws
z*iGD-T_yHVzfwbBnYdi!+N~Tm<;2Z>K3+B;_WYHzSrBDJAJ2DJ^3~6IU^?6+*H+Ny
z<K&!3D_S%&;BMCWTF%!i_Z~cWP@fk^KRJv(H_PVS)kN|D(B-^lM(f&Hw|UJ%d;p$R
zUNfAl;{oIn5;szQho|232%<bg&;GNCIZGO~arGHc)QdBfaPXnC2*{k*j3$uH#4Mt&
zSh=ZpK&}uTn~{Ef>~;s6lL*$Vrq2zIjRLXE0=)#_vV;cOSC7kPd1fHinBv@I&WE0O
zDD1V!{3NRXPVKgLuI0`}9)*Rh<)kYeXYd&;#^1SncX0R4E~hbK>?CDgPMUkN?40@l
zT&)&?OWKH`9}q+{l2Sy+nZ7;&hzJl!2tXR}9ewP^GyUb|lUEoS8?VNm038JL30?6L
zTM2sglaZ1pd2#Q1{st1ZzIglgjOP<lrWk(|VeQk)p(fDub1vSF*<9(}yCVsUD+)M6
z;xjy|udDUDFE$6Ev-ly8m?MR~xr{3^(!jzB(JKZZ9Z^@p=%Ch;7Ujb6XsXxaS(tvk
zr;D#h3SGMWh7h`No6S}-L@&nbGvE^B960Ea>&9`yd6Te_u@9TC(*<x3%&lI(h*^9N
z1rGHOZ1RMiFS4?-;J>@QbX!*1qC)bxwc>Z(AuP>jr2nL=nrh<tNp+70DvH|WBM}~{
zJBJyClzbdLZiWRC_U){VTXgv|(i+U9se6q6IZ_<T|G6v5ACPd?)6=u~AUfb@XS<Fm
zTC!%{*!Cmri0O6bPM;o@FR!Mpt&PmOV|(ipVu<>4gJ0rA`L*Y00$H0CRm=>T(v??S
zqp<uJ4p&_Piy_|g2|LC`gP3a8Kr3@zl0mi3hOrL}XVOETBfmw9L<5*obRM82s;d(-
zHn&RKVtn@TrqpP9?)EEGm}9_-81%R4eqLVc)eZZVCrv6Xo6lRB{0G~sE+Yby;wNpk
z;t(FY->>$uO7TNswIi6aZ*zmYM-_S#cmSHi+SotnJ_+U@?K#3O(@sIhp$h1itYhob
z63COKselryV~2%={H?>+eo_$c<;U}@1s5*KY!RA%xw(h6G5P&K*DUX$ZHl)B*~F+(
zF2j<}FfqkvCMA!Uym1vLFm2#-_5J$y|H}hOs^h74ypB*SEsj!KSjHnTDG|UJ=}{Cz
z_4J_%j!!5kF1AQ#8oQQ>i8Ct-^79u4D74fT=ii*Q0+K4dKtbWq8h(#B60!kAjf;&;
zwYRzEOOco`9I|HlP20B|dqQX^%~BLsd9h;;{}j*r<PI1?#|=WRT)KqR1FKL8xdYq?
zJm#Aofr1AmkxbCIsgmJ55`*%^nsW!klLE#UPT{P6w;p1KmtUiWcY}cq>-PDdvJiQN
zTIGN)B*y?6P*KGX$dA~bdClk>+2nJ@Ec)lT&Di+Bgc7VjQq$$qhg+!So?mZ2q^L*x
zsMCzEP}7pW8;iSLOOX27{`5-0R^@XUFJJ8|>OZT`?D{c#D;5tJv%G$&omR%vD=Dut
z?zpL&rsT*4WSzeAG+Q=jl-vHN(zBzsZQD9*_TMpC^3B)YhP$4r?;w@<sr|dwiK*+~
z)YqrFwj3MxgnQAxEbMsw6ofuVht#HSMlXY)&{AKzF!@<_wi7EuK<m(F`7T3on7^Q$
zVi2TkiqRmato!ny6JqN4w`&0{)+q`K3d+jLKpZ+cI={r(i{U^#c+dcsTf-(bdc$*C
z&4yC2Vn_<ruu{nS42_s`y^0nlL8NEbuJeqH=zOX(N<@uHTMH(S?x7XqxgVN+pI7o9
zi68#d+aM=)>C1yRr^CpBKPYC|igqU;J6ZVS#iaW=IfA2q=ZW#$xgVRGLqagvb;YHL
zCgZ=yk0uwo;R7M8tcjlLo+6);@4W8Ue*KyZy_#<C!2PE@w1l|=Svt;2Z>A^TUq$|d
zs(t7dH=1O0g65G;z8u9ZR2ca?2%SVM$L{>%8C<sTX&%0@tCZAwS|iLH*xK5TI4}@d
zqnC5USU-^{$<JO;qJ<C%l7u75a4LsgU$Ei>7jsDO_u&+IjtiL9{$oAqJU96m>W)~q
z?DBfDJRJt%Mo&+BVlIOq*{$%sk;GVZ&36+~918FL(j0U2Q)T<~aq|5MXPgG6z;KRU
z{>F|^V!>1BLy`(`|G{2^7XS%G9WgxM>ir+2xq|yQvEitAwnak1dHUSBa}j;HUm+!N
zL-H-*g8+6o<P2XiF;K(A*I$3lLWy$D91sMd4l2Y5{oDQEA<ok|#3_YX$xW0U0SFk(
z;0qf@3U#U*DtxS){16}_e|<c>iK8<K9U2GGP$yP8*G*Vi)(8VQFl02eG-w}1pRs^o
z9+)q3(vrf$G0)qdmX_kBM=KgfzRsOGvH1mJ5_VkOk;x@&5%~X1@_Lrz#ECE=uKTo%
zLPTTZLvbw9UGw|r4~b|F7g*(s-#bVgnn6XSJN7s7@xqY~%E#<s@xKmkZi~d;-w#Io
z@gzcP;=khSzu*3SAO3q!e&2`x(*>dapaO?kmEt(lTaZznWl2c*&ZJpCgj`(%QJ@u&
zE7x6{#&`RZps<5ivQjl=D8jy4#hXUmAq{*WwkI3$uURVlA0EO!GmxXVufC-2u)lsm
zl;XOk_xV+}{kp{dzeCo4zy1HbCoiJJxb(jENk~0hzVan^5|quH_H>A1>o<05Uq+S+
z@?CXt5%~*%<^77$3UAX8!D)E|UXXeD`)8Hq_}F|MelI6Sm@>V2-~Z4bJix;5pS{r3
z<t1+HepvKO0MuXqiHVIR<+EH=RaG^X@w4?!lBMwIa-I{@QkEjL(4FY@&p(9Udw2s9
zf+@Ff<oZHFEbORVb(XrjPh#XBu0vNPPH7H_TM$my$NnjQiHDCH=4(zk@EkhtK+oju
zNO(e85@wy$bbI@&EP1cnK|Td21NmhAMPmu<L%vJQ$`Aku>|HT!#`(`#1@>?^C&vbV
z_kzglNdH50Sq^EQV-OYj2Neqg^rO)1od!>c<PvHZ=0SXE$p(-yFqsJ~gZbURT;8I~
zuSXvA``&PoQVtoR!zN>a-Z^c36^wh_kW?Ki@zIn{{N=net+2%|T=^5aTK}HR4CmGB
z4%si#*si1VD0c5xgd69RT{gA#7yx0y{4xheYWro1P}SYN+wb%PaB#VG;qU#(82~(R
z92@*JOwY>tN!>Y9Pj6hAP&RNLeAXf{B+r2d%G^hvC)8VOI5=e@+}`vUGSQ@21>)J*
z-9A+xKD3Jgx9euG4FtWGubwSEVvEn&PuA`JU}j=*?Ut|oRn6b=uv3FMQ<54ZNwUWT
zM^V_@WlB9?lUhKQffyQ){NriUJiGN*)}(x5s-tN<<eZg89-E<YtTNh-*C7)$PRvtQ
z+V1JA6f(*lP1DJx`rMvqD;Rm-2I+3Jfn`6`{bC(sqPHp}7XKAUGvkmGDW>#$i3;S8
zFJJldLgDgsQicncZdX?N;S4iq<wymE#XQH@x&cNbh7A)ZF_&2zs;MnIAA%N24vnkk
zYr27Al9cApy1VY}!>P;e4_rHs2``4(#uh%lE|PF+i`pm2{-3I5Z`Wqv$PRT<7S0=^
z=UVN1O{1VgX10YYo8El3<?l%hRD6o`Z|f7ff8dy5$Y~E+>53>WkRq3m8VwW(>U4`0
z91#D5?22(ys_eJma-|hxY^dYcdU}SFW3z-GdtkcgN8w(y=JEpeSL3P$wrOsrV8XdN
z-Cj8E76<V35xkx0=*+U~{_RwCuBYRfA;LlVSe&C~z>60zSSFCwoELrv6bRRHH;dgU
zujOWfobGKUT0m`D{_Xt5Yur~lJ~9S*p`%3x;|TAL3!<=P`v7k;xHroB{K<1SxhT5&
z#|2k4KL5s?d5OhjuU3ne;OKzukCB!q>Um(4{hFkf39(}~8fE?4k84o`W6jCR$wf59
ziWIJw{HRfzezt1A>L4NgE7uqc3t)a`&^rqEu}^H_dQwANc&2hxm`CzL=x7AaZy2eh
z#MV@U5d28vb$%%5<*KSc{DF+W*lRi8Lq1%-A_K!}*idjAd2dUA19&OusP<JdfNtpw
z%@wUNCeL=CTLvu4ept6|fzHR2awLGDJYI^X;3sJh^H_-rlx<Q;|DW47iAv~0#MOE)
z5^gM}j;)zmj_Ls0;2%TazF_9$6Po|{k^ifFyVspQkubTR$8~Z;BWgOkqMC5iao4@m
z@TjKx{jj!WmtF%M1m60kynP#x77T7R@_(d;gO;~5ae`yErOf0~Mvb!qK?<2Ba7igW
zm=F9zp&VbHk(Tz1wTE{B!rabk_@#NXBCB#QP>%_fp#>ZIfBj0vYm_{zRmJ%#IsV?&
z!?Gn$gX8d=(Ii+-&5K|qbSnIBPHjjLN<x#Gi$%bOTomN6IhS{(@PoxH-?wid8u5b%
zf2TenY$D5J<F4L22rve~q$-fZ7#>*pZF|X&t3~Q4p)fAG>jA+NKCAWhbG4!Sjt<M#
zvU&N4b<$J;Afk7kKli5;VGTh+yqJ=r0!Rdu0;GarkC-N=^-y%Q8T&TlPaZ3mE<JMD
z4{m%Y<8#uX%x%8rv=8%WSPC1d<!8%tDgTE|ez_|;rm(TJSC*My{K=*Q?{T$i-|BxD
z7<ICyrpclut5+k7Sa&v!^XwS~!PzFM5s%2ap#bunxHIYeLluWL%P2f&St5V_`6mFO
zT7{CL;!DnvJ6fWUUmEhE4kDzP6nh5~+`k>14iB3R7pff~m>0f0YfoRYdlY&Sk!E4$
z0BHevrk2o3^zrctK2THfB<Aq<ZITi?V$n}}jk6105N8=RrOA_&v@$nPX|iW+8c&I#
zlB-~STnCp2k=hxlLR@p<AQRliP0kEIl4jcg>Hy$vM~x3;W@}?p&GCdx2#eSr`Fx6q
zUAFmET@MBXRM3tl1d1jE^%+$suf}Q07zJdPkx}{E7*u16eHK^^Q?j5;QBkiLq8qce
z>NZ<`S=?djEio~Q02a?7orgww$}xyv^(OF=?oA`>$jm5w&RIPr{-}??f;b1&zH3UK
z6jnM&kr%TeR6v>aGE7#DMysFgL>0W5zpZ}mW605?mpH1R|FN>l?%au;dw3Y7VKv8n
z<+Uvy&Qh#~0{To`j1Zw2HhZtNo!tOslN%VU@^LvxitX-=x+@&aZ7-3lWc`(}ga;wC
zR4Pmhm<kfcCc0lZcJmia!_<Uf={T!PAVK%jo5$EPwMrGa>I$Du2p#S(&A)kb38(hW
z;(Z{?_bNAieCf^YqvR6d7l#i-$Z1+bZ)IFaIqKC-$#*O`MsdYRAZHg%rn5;&`}(U?
zy?>8%^;7tL%ijH|3dspiVXl>>U%=*LAdZKt%ROd`%9h@w6y*<<=rPeB!Fe9DLBVt2
z@gAGz-v-1uuJtj;mpDr+D)vdME%WfO0}>@4LR&&x&)o=yGK#LBL9ax0n4PVNh%dK(
z6?qdSA)1(D%VN|69Tu&iNQ_PHPz!}wW6>o|HH~*1nzV2RB7}pz3Qy`wb-wG;r5wp-
zJY*(zuwNVBZ*)uqeWHLt<K8ekO-^<_D{Bjlp6^X{uFlFfQH2ayVcZ7$JO%VsOmg51
z91t)c43i-+AuWrj?^AWD7;)TjJ-k*DA#z10ZYAg4JzculLE-k_M<-dTWo^vn^K-X&
zB?F&HlP8m&nEZ^`L%vvB?Rj}aP2i`<!4kR}cNX>PDB(M?v1HuiakR_o-``-7?7WqP
zmu1F7*y>7;Q97QEhZ5>cU^gz`>7Q2+vs8vINRDS$1#5~*-e6(XM@uBO^Vav59MKrD
zl+(}rZgD#-@<V@rQC`c~sFjsf@OaNeZ(4(crWn@POnk|4PJ|q+xp-@9uU7<n;fff)
zlg+Wm*b{eTrbSfB4`Aj<{{)Mf?h%n}bVTv=Vho}1Qt}_}<65Q_JuY}JGjwZ4FMX3M
z&iTzg3_H!W1=64Y)0)-;vp)ZRcTu+B7P33y(_U3qD^-x}+4Ju7olKAUHoSZ1U3I%_
zlN{}s7GkFWlMGsLfS_sfk@&m)ZOIPU7$It*ZHta@nSd5y(LedEeReb`F4Y?faDAyQ
z2T)|?fll4}YrgT%+W~)8`TJ*F1;<|L^^<Y2uDh{LdHC-&4u}5Dm<XRp+jd#Mhtq(H
z>z8536iu^bB^O7gpRf8v;$B>Ehg@lcHaOYs>Z(t(?U@N}Rcs`)50IN``?KFkq5z_A
zsaNX4V;Dkwa@J;HMjuq_F+820HIg<hBQAjk#w%8?oZ)sH(#dZwfItgD5o7e(Ter^$
zm`=bjD-&qHay!O(`yn~OtF89%NPMg$DY={Ak+XR6h`+YTqph)EOQ;wcJz)_}JV8Uc
zrs3(8jjP?~S4<e9WSiv0)o-5gZZ7$KuYsB`2v4f>=Uwk0gkKoULm-+g(m+F0MX)(y
zN->mS|AvxOXYJ<PeAFXV{uoWj(nc4cr@$mg)9!1vhr33p9nP9Lg^c!Wno1l}O%C}S
z8RO^WcNS{Dtyg&zq2uOA9x}ZT?CX5xSMxz@o_lLCX#+ghRj-%m@)@ZVo(IXH!@Gu>
zW!jz6O`P?#0C%oXo{x<P4jRjszR6tnX$MCS-NgsB$@f@XbaCkV&)y!sCm>*DQS=_O
z2SGtWNa6VMYPIW=IcFK-T%YBW+e8cHA^Jz)x~T`DyysQ<o1eQq@B<HVvl1`#aM8le
zeag8Y*QpD?u&+qFUGH!CQ)Ey;-b^$1W9OdItA~ffx+haiJ*Yot&c=8DW<P)K=-xuM
zK$YMF)Kxor3+}z7FSMbhaRxZ`-X(`qB6`9~B>W%0(C)hQtmE}IN}ijM6f1$F%#SOI
z^0MwJPtD@tkL$VQpF6s4XL;hIZMe%z&<4)`%&sixKXBkvK1+;$8t5Md$M>c8M}<d3
zAPp|~@Cp7MrclLlj8f#(CNV^Hn@>wZoO+Nk2JnSNb5D*=Cv~Tr4oS+fx;NtRoV_G8
zSH^7I0GPdsM~^5&>OpqbMUpnMb5nQ5JI*p7L1%tnp6YGp9a(r!?6)3HGGWCyy^!kF
zYr@l&o;;Gn*Lqz?wZ=#*W<qTifFM2hu!bl8j2B%c8RP)yi8N$#e6g@9UWBG9pK{r|
z-ju4`IeW(J%yx3=c*@z=6w<u`-(}3wa?8y9`cYkb_QY=VEfr<pkt5?C->P5c<fM39
z(Y|Ky;RuR8THLko*QTYW(m^>y0m?bc_;g%bC+oM2O%%;M;v4no-j<z5{2u%%(PN&y
zghXirt->b{<~K8k+tkDy^uNo~fqKz}p3Ua)^kmjJ3oYF3^H-Lp&TbiZiRKreDRnhB
z2e~C7YNR>&4Nc0buDj%Sas`C#uG50AY^YiH^`ldT>O5{N)2}pg!!(^{k}~age_%fH
zDg`#{dNPxT)c73jjNnHgn78FS)<YZZcak3qJ%XD4L3BXhknC{s%i`w%BLpi{Xl5e7
zN9?%d@nlqRv<y9#!PqN|dQ36GxXg8)%5^Me5{va`&mM{&%-op1g9eqd>`q)R_zCS|
zvywr}jKPoT@O+A-&=CRzZ;ln&;J}5@^VpyXi-Cw4xz%bLa0Pw{Y^~y@b_5F)s8SWr
z@Xnqz84bpai09EzY9u($8FPxGeV!DcymY^Q7l?A4!`MAN!!s~D`z2M?2dqP=D6s|z
zaR0`Tbk(d%jJGN)EA=}o1c@4&-+7|e_~lE!!EyP0v{1X0!(UMS;^m?ZWyE@ma2w`C
zALH`Y2(^KDfC=^%X~A<~YfQKCPCEJhnFmY*lLiJI0G>|_`X1I5$RDy?@k)?_4NT;7
zs!%_%A$TPQLQ7D>C}9gNJ!kKpUM3puo0}5I-FWw_RE(re%qI&k+f#ayr^-l2i`$Jk
zNO`iU;4VZe?k*_!b;1ux3-^3LUp4f13tn`SflNb}pwLZ=C@Ch#X?!3(;7E?bSti0s
z?d>~vyy#o1?0mxe2sR6=u&zzGQSf6v+hF3$p+jP(okUs6u`5u3^Zf~NXOE1^eDUsG
zoZo=(Kxl<9E@m%EWk*7|SeCh(RViZEzr`iwpKzh_^q0ASG?670zmPVdJ|kb8mk%<-
z-c+lZASq3G`5nQPfaJ(=&dFxvwb@6vfI`x+^W%i4yd-?`Qm`oGbvC$tew`}@->^Wj
z%dkJV!A%IU(wo3PA(8;G3Bio9PawTmfDaH^Sk&X&>;f9qo{3byacZ|4XKr3S^70Rf
zK~059p4=Ny6>`Q4X=M~hJCC1a4v!t(V|+}Z8;fd0l8mgZs>4@-SQY7vxIEQy&WU8o
z<|ftogY<@UJ;t1fnRb%p_un!A=B<ai8bco1OrlPn6#F=UjYx|}>h7^IOLwImm>9oL
zKPn|70~*LbX~!ZXBYRIz%g7K_Oxgm*cDW2s&%H7lLRToaS@@mLOq(6*X?yVNmdVVc
z$GE!wRjQQ-7|?lR(EQqUG%dLIdkg`mSVbC6$=#Jau}JWa9XX%(w=Fxl!fe2+dCw92
z@~d?p5=2_x3gj_p)2XrZMqi66bR$7#^7im(N8OlfrRC)X$FvQn2Jf{@OhS3ZJDKyO
zMlR35ER?YUce6GsI@cz3=N*>=Gr?45P7JAcJv^gIOMxEtW1q*1qINmR9Su}6)ZO%X
zmhJo>*D+(8L=u^Fq$zsTHkN6P>cAs6@=g*#YiwzsTYGz7GE0Xt!)VWf=4pl>)QjU6
z&KX}5_V3C7{R>#%ltvA~tO*_w@=^vB-sq|v(!Ez%sXKnWrR|)74xUs2pW1%<Wa$vr
zX!y<yC*bDEts)@I3@5VFj+s}-J~9|Mkwna8{V~=<*M_+lVx5DuqRU~}Z1>3>;Tdn0
zTuJM$u^Yv*$EO2#MtaUOG+a0@An9Rt_KfC8HMOqKT^_CN;vwt$n2)yKZ%=A?eYK9E
zVg5h=bSZ7!JkP#+>cxxXjL%(;2WCA=<70|yXM|I~AXSUyY@Rz)(iv)>yK%yZZ)PQj
zA9wnCNfSAVLk+?)=1PV!y|<lv!Y~WG<E+SP{=xc<sb-N*qw^1j^4>U0^S$JI28J_x
zV9<#0yRyZjYJGW<zwFKzyl;{Y6*EX1bBAJz8EcKR%K=mLw{N^8j*Tg6&Hjas<zyRa
znHU^iB|7DROb$_n);2an5gr@eO=Tdt_4%}u7cUCjWpiliJF<hKnZsR|&J#?V2TDeU
zhR135jW}k&A2>5IUw8J$V2<T1*{KsZSfCixzNwdO$mYXOecQ^O*Ts1T9~X?K$P!#$
z^&}q1w1-7R=&~%`WhDV|l?+3ua%XWS;Ma|r1?Bo?Pmvzy{1~YpxNI879wEy#*QcMn
z{51YV8FkI4WkLKy4xrof#GwCp0Nb-ImQjUs*X!i#cWm1N(7R)0O_1+W;28a<ZJPbI
z5$-&w7OVWaVMsUZ*%s;F<I*P@Y(u*I`J61r;(faCwT0W5EwvhsGpvR_a{@YnHuQTF
zs1|;j&C5XkC6iM#4I1$AfnQ*}apzcFT?ens6l*l5T7(*SA;I$v@p3?AA}0vLssAQ^
zAw09k>N1Er*TgaoUT`8j9O~EW7cWj0Vo_qY?!^UDosTY?A3t)_V&QJy7hBKhExBlz
zTP<Q+hq89Y5hppO_n=xecW#&N-TOP9=*>GlJUZsq*QLPjLd7L>a(BT5EcJ9G$PH_A
zD{|9fGX-Vo#SvPmU=QrsLl2u&JGOqpMbv;~H(gxlWoT?1`XK8ONfo8o@z=c}3{kO!
zWX+yFUhVMp>+M{G#kHNf0KG&rkp*Py5g$x#zwX%|EirZnP^W^=BpscEjvaHlTJO(&
zda|!Pl8<>{EE~$fseVmAXSy%oWq&Q$vqoEs#x!)b8jI0vg}zv0x}{7IixQ-!kbF<-
z`Ylk7roO4WckUE}j2EYZ&~pumL)<_lGB1RlIDtyajNuWp-s~9}QfqB%n>V7nwiz`(
zn_qF1uzVSjszte*x4@$nJ_vijp6>H+XoZ5AR<uoKPdj?RpF&R(DCrHv18J6*mJ$qD
ztE&Kuik8oG=nziUXYkw`yG5D{m<N2w4^oNB4=v*vZrv&YdYN^Mkkz}*F|~kfc-+M(
zk`V%|vwsmy5eMIt+qt>!oM0p7Wt=N?qnJEr^%)kB@yk^k*2>B_m3i*!uWT~?sb$`I
zCkJXvlT6>ZLN{T~cK<dVjHL@a^E&SEcUHU;F3s6(wKoS$<s|86X2aS2n<>gykR+@s
zUp(R`D7A9M)g>MtYWF+Ooj)G{A<<zcY9%$gdVr9Rt@KoApNpaSaw@gmjU5u)yVmvY
z5<Ot@xLL>c-8s7E%wxdmn4RP^qJ$w=`@2IYN<8q7Yv^Z|?r>)jcRmtahp5_j8r_sq
zs<0(M4}f8i`WkWNbbRHFk&`DjLyZ9z;Hbrq6NE<)QV1Qq>71<P<l<H3HTUoRFZSLu
zsLQO|_Qmd2+ni7dDs3PLDp?d2On`s^5F{xkk~x5=L~Uau7(hURi6lWpMT`U+MFByA
zfP#XEf`ADL5)?SUN8f$!saticZk-ReZk^hG*t=iDpZ~L-wdR_0j4{W2^29LY#bfY~
zW4A8XliQtIB6cQVQ$KPVAyai7OXqI=A+`FTIU;}2PLO3I#Bh_rvE*j`pCu3UwBEIQ
zHyYU#gWfxi*ORq4M26O)9#F0jq;E-S;ivj7yX_50)NkgYR?le-^^|hkgZ7zn7(k69
zULHSo?Afs$Vffx=rEO7K;K+$Nl=7*1_DoHjD9>(JJeMP%tn2qwY6_zebGBqYvcNge
z#z=icuR{dzMFRi;CoLU%d3nm79c7ZUeBApAs3|q<mw->;);DTN_%glvmz~TMwj9~d
zh9OZJ4jfVY-Gl4k`CHY8lyr1<A~{-;D2O{Ls}p9Vq<MC^ZumIsS!B?5Si2T#D!=yb
zeYra8Gu;Hi<%MO@(YIXwKF38feM8a0TOL3_ms55*ol$CJY+*rF|L{K9p6!1gES`f3
zywCGyqw$O9tdMHu4o~l>l?>>QGH+Pro}w@ELP5VpFt(auXmD^=Dq5m(c-yzd%y_b7
z#k7`<Y9V~XdhJD{UbZwg4*k6O*-S#VZhCLrh_h|RTBf=tJ6{uK-3`dz`v;N^u6MOs
z3zU3C)+QyDl{SiZP9AT6+eSrNq_LP>)-K&!5QkZO)T?^?cKzbE>m}rkQR6+C;6@yj
z%zXvw+k?m&ql-7(WMsVHTwHm;4-f=_2|cxWlJz6I7O$=>?ZBmZ(3N%a9@-9V3BDp)
zx&lC3W@smIRyP7PJV@#LgQ^qO2kt0x))XGPMCq4kThQV(;x3}VMDPHIn%!x_w!m>=
zz<@vlhnRoF-Aj-I9o&*?zU1g(Ybz_v6&|cfJebnAA>P#PtlFI;IZ*>9!gw!G9PZ4O
zL8)gKGpv3wD$-}@49jv590f%iP{|)sc1R?>F1n$_^&BK4Rp1ty{p~8n>2<wf=^Dk>
zO5+1shEW=Y&@Rs*-etTGAo19Hl>w2rQcsXA&3;Lb{l5lp4&<Kl_+E4WhDXwfzW<sP
z)yE7yFTKjsXNWgLwa6(;KK$7uLm*8DV*R!|N9|zoT9jq%pmYSG(V@(9^?Rh7`LHe7
z)~;$qIXpvfPVpB&<k<8e4)UR4OSToe0en-|I=>_%RDU78<L3FzW7}<)TiJP3H8(XC
zea||J8`i&zcdWZUaXycXuqm=8FfdSf(m=^OO_x4S7<#>5(8Fy>U}cM53tGDCh7(@C
zcu|)S5MJmg7+VLlZ7F5h4R39{a*oy5U0<YQ_ZW4(FlY_;V)Au%OB`?u`F%iu8I>mB
z#w}`ZNU1XA=PzGcCoZ&eba8fm1zWharUQZu2i-+qS{ToCDphvw;!e+Q(Zj|f;yQ2{
zCsPN?EEY|>DKlwO%yZN8-UL{7{2_g1%$gtVD)>RNEdN6bltqBV7-2?>7WSiVR0uvx
zPpXylL(kL$fP)EjJ)`13ehiacHttvtD2n;rkcRZn1JgEEzl(ECEC(KTmZT!%B`5zl
z|EkCTgO~HF^6PfTu6Vl0Iusb<BI;=C^xVFfI<KW%*r}>&%a*&;p58M|N5Fd#1WA+~
z9T!z_&RO@W&Bjyot=8L6lUzSJG1O(nQDwUk6|<Iny9lM7e|DCTj6K)s)n~}KDWO_*
z{9g6JI4+jaJ4b9#3*qK?GVS-1Bd=#)80Z<^J>aO*T8FJy!uKe?DXKbx?M>#I?W=1h
zwk#k>Oj=v|35_a9G#8dCmnS!S<-OgC3p(_Dq*~SDdPWN#oKMe6YJ0i^O;H-Jx-hqw
zp`qbdS--21Tdb@?4BYdoVlE9!-FP?q&ip${4g7o*%wl6&A3h&)eOHzLaaW1u1BFKN
zBEvsDDKR$uX{K{ncTUt1A8{e_*tG@ZRZjF=VyBnNeVTLJ70V=mj-vRmz(5(#_3t79
z>L{|GL@zwAvF+Sd8Q<YJK5+9Jj_x<K1#@7FB)yM>8rMyC;v+w4qVR)yBwb3ld^sm@
z>IZ<<&@YvUs=7SK)-URuJTO8;#rtOxB7Xkk%c^|)^(L?T`!CQwIWpKsjid$QzyC7_
zI2nO`T@qT#o`jPrjW-v3X>_M~1+|VCM@|VIC}bgG%EDF!iuU&L0m?AlsApH9<2x)d
zWRF&2WxUMb{*m61T_?8RyK55~92&H7cD?>|{c#QS^Q>It9FK_9);;a*!GnL4op+a4
zmMN2N_1SxAa@0e!6LAP&>PGqGdB?=WOu4z{<j2s`oE(FYnslmTGLTa%MlDOIKn{9)
z80P^Q{fO3<blaXeW)nvDNL07E)Lu9JXIr##iTqxf(-)1GH5(6dXB%nai8rJF4>&O5
zijZo7x$1a}*3-M8c1})V!6t9HLGG+P$F7{|2}ni!VEkb$W;2XZx-MF!(XPg9#0gjS
z!=n?fA)Cjw+b+AV-zN<asouk42?2ciMnyxHg;~dq6QO-Fd=ISalPG=0_EYs3vGz#Z
zcJPI{$wb}Moy2uAc*yf1GBb1Vp|N|?`uU1c=dC#GlUuyMQ-MSVJz#Wigy-YK&Xc<S
z=Rs~PRI;C$_R(gYLMVwGqpLrXilG=U4ILI75+cYwz+VF<FSi#|26xM5j==FI4df0C
zQASo)^3E~WAP%d%V(?cFrL-mfC!gM#ZuVmxm%Z!qmlTzDwskXQU;i*@Yl?s6lAvrn
zq(t%MnS-07Aq~vS_WpZ)!|8)=>S{f0mPiGkQ)y~;8GXCCFTJPRIz||C9WeMw+JeQ`
zcfRerG_32d5&xvHiwL{BUb3N!R`$1VQykZ-CLtt30`AGr;|tf%d6t!x1&)#o98Bp4
z)>2mCFZ^!?ATF4^!?aJ@<&=%V<62G2)%um?kMJMpPPlw-DAok>B_O(~cF%HN#W={%
zRBEi>i#a*P#}LL+ns42(;SPprWs~0pFGpNi_G7whkd*W81pWBz!T$n{2aPq$T#$77
z|F*id=RC&$d(_!i=56q%fSc$Y*KwgxJOeff)h*R%X8ont<omQgooT#u>@w@!Z_`?M
z9hyPDBHCeFc_AH`KmYGN@?rl*<vR<(F-*xdG34a6i^w(W#ZOf<-;6xA0JtgmU4K0;
z|JKryfddAps`On1a$)DNG<NYJ|2f9j{aY7L;G(z8j|%;L-h#62aVA>_l{H;f3_dfS
zo8MKMMFm{Q%ho!h1LS(}6X}$s_5m}2kjii_Cx1A${GqMsbYBT^iB>hKH}|2zR0$67
z#GENy_`2AAfYZa5ePM_eE~)cw+II>4vuwvhmtLbLA!}~`{$1GqK%_HgT8decdN?dT
zV7a&C;kFENYdgm(q|#q}Jf4cbud4bh&nA4_X|+nq$CgTAh_Jd!#425fp!bb2my`>)
zOq=PR{E=4&ReSnMsWA7A?@(X!CR?2Rmpjyds(GB^R8y&tsEM)t%zx|Km?*IDjfDbe
z;o2JzB=Uswjva+=razvn^vHM)V5d?t?^zZ*n_W+#cLxp>1%f~jgXm;b#8PbBCRv<5
zR+H1qglHCt;*MOk;Mp?))I_8JM_wH}SJ!@sQhVG#4iXKi8|RNvQd3j=kUe+tqPvk6
zn8&sBwW$4psrO`)$fLJy$(eLND}-*6hKawq6VVMdkI&|Zrjfqyv;FnP#HcI^bD15Q
zRlC91yFM%MSddz7ucC)(!fP^$K6b*Ctm4onUoy|4fBYLBiu(%{Si+ib#uI1HrWe_`
zW}@jt*ZzGj%?#aLrGQF>{xWJ$D6Y27)=d}&G1IwjoWK}iN|^MefG;zmXf5WS9bn-!
z#sD=dQfe^wWQO{|Slv0kb!JY^&Z6mg56#neu95IOyy*E3b(yH!$|#pE1dV@35M<qy
z?OzY{b(Qed{U&KT_yqr9PeCVsSGS1x|MA1cbFA5(J$uLo^%7Gk<UzqWoa&$6;M`-^
z>{stGtjM;xAneQ;xmBy?dVE?Za$e}j-FM;<bZxQ154ytUN6zisVm(_`$3vRzf2&>G
zfwFM2Edjnl_HX1@Xmsl9wAE=-a`VEt-f|v3LUuCI+(xB6uH*an0WVz>tw8X0NfLCK
z?RrW1$P;l&r9(FwxdCN1yh0mA_^KZM_Wd0%#@!TUkPqjq=mt%SYF-9WHO)=oMl6b%
zg^81}2g<s2gR1lC;=h^vd^+091NQP`eiJ;T{(#JQty+<KxBR@ksj{=y=vE(W6=*I`
zgI4lDP`gVnp9UyAhYyV0m8-Q;m^Y|B-mc%srrJE~J3H-C^$uNHv%M>Oh(lG<ty^Ch
z-p%75%)Uewk)M{_jkRU`%NQu)PFbH*m+JVqhSA#G2Zv@PR>q))kY71qCLT9Dn{v_q
zeA9??x)I`YT%8e7WJ>D}cE27`NoGkbm^^ivzJ3Qqy0GVLC5c0~l$T+!X^)_tAEn`O
zOOz8U44NRoxz-?ugJ3oRVe;7~xNkm6Spa7?(R$igU10AC2{N<(c>M6;B8PrR5aJJJ
z+TH7{2cqy1`urOU|75Du+1Uy&=<&E(L4lpBm)T|?*CSTR%v0B851k(GtkrENVE60L
z$lO#N9Jog-S#8Y3n-2NoCcb=?yeTMWvca-tudl5?vUL{i9mEn*ZJ|+MZTELF3ZL2Q
z2FsaW32Tu!;M}kxdJs*0-)ij+kmAZtoxQzk--#0^Y~B#iTK{g<+){oc`{54X=N#+}
zZ(#JyYF9BG`}~X@;cP@w=;a~aT3gE1He?7ejQtgVZuC>%RA<iIOHgq=MB?Tj{f@c3
zBV;ABd(HoQ+y0N2Q@x_VEBl|S7H%&0Oyde6<V`+_8J1tyiSI*oeYv#aXVXIIN;7AL
z3yFASkk`RW7b<&T`o#zC>9=gWgKjD?048$J?`{FNCf(euc~Yq{_Jk|A4e~qH=y?e%
z;<XkE(-FYCZcXjFQpD3&a9wK*v^nxrzi8RW5yxE#Kmaj0509hgibr;O!yRV5Xv#9A
zF-gu-tD%+py`McxbgGqUh}=iE4c*D6`Xz!>?C#Gi@o^=xWG^jcsKu${uBe9c49o?6
zn#Q(ismf7YuC!rrvfK|_L7lu1R50+b0r`Q#5s6MfLlZJ1{^D_D1J0TqT%|y7NJ-$z
zH*^OgAc=ja88kRz#flXakUxDZ+!xC%J9oLI0C5%3`|`^x(AzW$*c_rc@=s;Uvie1u
zK$s(=<>ZGs{HZ=8Gk?;~n$DgDE~QC^3d0;qM-{zY1Y_t9?x%aRgWDVG$*f+O8s9HH
zQRsFGIvh?DHearE%Xu&d_uz*gBb%Q9My!i_L^2EM=>N}H^rwH>&L2wbdkK_D5K6Y%
z0!}K(DV`?^<E_lGYkb8z<t{$qTZ*=`{!0<W7Jz3rQGBL=$~{j+$Kbp?o9Wd`tP8k5
zxkh=GNMNjwl(x;?kTQ(qTR3EZ@RPu}sKy9D*EPz$pq;CCH&JN}s=TV`w*M+3r1`xA
zl+d6(BuKRMk4oz9Nf<L`%tI+XWI=-N_}^3}b1J;5^5YhT$k(}9eq^a>h2MW^0hBh<
zXlyOLAd~jpzy3%T=EWua_3#T6m=~T80WJf-E!@+e>!K|6Me^}>FQvYHv;vHmMHJ?S
z&3x9LqCpw7!pc(>*Fah;ing%T@8MhTI<mX@e9xcD+-W_Yf4|~m<av2MqQPHMC6Yf4
zKK9$PU=<y!Jpdzg509JWA~@mk5nxXwl0e8eZ`h;Ij2veDj5<8-#&)6*N<XB}d+B?|
zCu`@<CQhaxlBRJNLoNZOr!z}lzjDb|>LcD?piIH0@*DLqLXrI<d`^XhrGUkfkZ1tI
z@o7*{C%nC>Q~$xH8u$^}1Ltb$_q8fWA2(cNsy3%Qss_qZ0QOj2ej@8D!2NFe_8<WA
ztH@Xi6uve#wu|nJ=#_PKV9mSVM9!RT(KWn=sMvmY--pf0Fx-=^^=8j$2B&Tq)c@8N
zVF*{ll6ux|xz#5Uk?O^w1_PY>>Cb0k8B(9|3ENro)G2`m`NkHiInSOg@9HO^wl?D+
zr5y62*y3THyJ#d5#2sIZm!I9f<75<hKbTxmg_O4EWeq!_SJQPD!mmkJ)~sKTMdOYQ
z%SlK^PM8QcN?I6HdIys`C`Ti`&R3a$S3hlW&ElE-n0n;lmzQfF9FA36H>uvrN!sAF
z(rt5=MSVD-#NT5YDuBQhqYy)9kE>>FirN!TrG5WAYr_FLdy7I4`S{Jb>pW)~Xke-)
zu=S{|30ZPc4qfbF1BD){?2U?t$4}Jt$yzvM?3gi;+q4f1JX~nD=)pbmk!Kr|#l0Q{
zYalKvODZJm3#-2`4166^oRhPW<kX;F8~zN!`}zEi&0%Ua{YlKl-PZbSA9@+&6J)w0
zrxM6N$_n%E4rwk4Q9D%%{qyX}Y_xupUMhU;UC4op*Y3RwO&wTX0(9M;vW;^f(dUOZ
zl7{JT>hGXmzkc3T$nT5qi}pv8BV-u5(_2|Ksy6?8Vp}DhWQFnT9UNq@6d-)0T1?es
z*)x*Y0x&^(mpy3x^wK=$8AI6WuRR2)^Mu0W%NR-xSG15IA`~4x6?wi9J&w_99nEa>
zh{ED{^JO=iD5;|YGHuh}U<0V{uv9D?ZX5Ai(e$XV&-#y)&-y0tlSTHaDJd;+b~Q!O
zACm%CeY!x&*r0gC`|{;$!sgs?Ab4CI+V;yWy{?pOv`cCbH%+ve6R<24wy$3FkNRV0
zme_u2FzS@*wZ=uz$GEFuL8~s2hw}5+fhr^yM{NdrLwJY+UY|cE4oLX-J_Ao~S&I|)
zU0;L@1UJS&HS>k75<rT-f$rn-EpQZ$j*c9}fiAl3d9`ofBC2X_sb79z!0pL2tRYhl
zVR9E?9`eV9sLMu6SP>c)_5*&KY+x6J^_lr*x3{m|aX;$@!(~x!45|=bBdYd0pm<~y
z=#*1EzL%nN#bA}TXlngU1sC@3kyaZCro~&}3LdMx?jcz?y$>jt1l%740gLUGXWG=b
zx`Co|T=Xsa7{!;MQ%NCY3ZKrzXwU)!0|WX4kw~+#xER|6KW`A%(G4n@9Y)VE(@y=-
zkA1?~(J=c?xietYl&puxfoB2-xl6VPZTObgCykh8Zwjj{P_V!E%?U>y6tg<}<aJtJ
zlN?Ssuf4HZde!v%u;4j4^0MF8xIkElHVuR5U&hc2(oiDSJ*otH-B-kWut-Gr$G(zS
zbN2>VSe3^OEa;DqhUch!i7`gK#!av7_AM7-+mj~>^hYWye+8i>Gj+gI<e-}#!DRa8
z7k@pKoal&RN$EiwubL`Llg@6yDLMw}O6G#H1TqnnT`;86rB83}VNW0qK&ZhP4!twv
zC4@Jn1u}sf1X(JZ3SvDeKFz%Z4py4p3^u0V!Td^DmH)SR^;oqS+6R8_7ZL@uEI-&L
zcNN71$pcv!V=A|%C3RZsaE9P^K7g-cor?-api0?R_=diGd5@Yr!(=Gx;T4@(tCH+}
zPsY5XoWv7E7_i5C&B(~OVI>R~v9Nf--wIS*UCEI|=SeJ^1A!PXoAGDSyJ7+EUi%%*
z-_U6*#q%>Vs)@nD7J~#E7E-S7_?w(79ILvPYWn)>ofqgxso~*^ewb>KZB$lPiu<<H
zH1k6k8rm^Xm_<Wk*V5na{TvaIq{3cR6PqXJ>uQrF{%fQIsK$>q9V{DqaGp?$lfx*r
zk11Kcc#*7Qd_ky0tkAY_-Kl2ro(#Gq50$+(FbJF?jus);VEK99ZsD|OkHva5*v~J#
z=smAm>|?+g=kl}J4iT3O?R~)c!7^jNtoyOq`7FmsevV#izcZ9BCw*flGg2UQ+5{Q9
zrm{Lsa;_EGCi3x|2i~mTmZ`JsBS@@Buia66>$)V)-TxbvRPL+&Dsv=pI5BfgO#PTA
zRhIKHsgTXdzF_-oe0L1F1iHZD_ZJ83^ei5lpTu!WZ;hU5h;-H<sz*?5^{8FFeq9sg
z!_};~qRm7H=hohRN@SNDvy&UKh_uM=Qgq11>4+y>xjP2AtS95I-uqZSsgVE4O<~nq
z$`-~^BeCc`*_!^d==hUkpD)&uT;+OT4b2p`+YSu1brk-6CuD%xx#%2?Fs>AP$6mR>
zfXuR#PqycQk!^w<v-365j^Ok_VaROaQ*$2l<doGx$iw-a`}a!a9jC|blT-yITvy^~
z|JrfRyLacm+K~H)eZ=olqxnA~S~5iYu^8GEMUZp1H8+pg8WzNDH!Yx_9Su>!R7d5n
zkAWSYM8D0|Lcx1?`<wZo)7M+&$Vq@Digd#Mz)$BkLzmU8;wF*a36AiDsE-lg@sMx$
zm)+~kPoPj?`bf;f)4QAp96CFDuZolYY*m902i0ertmK(^HLtdI&GB(_H2?bR0%5h!
zZebuP@0&R!6Q5Ru{mu|!DHM8_mPsAx*-8k8fJ9!ELDx6MjQx-CDjs@w?&DN}g#;HP
z@;;;3K&}P4Ec*26BQe59b!tVPLbV?rHd7)_g?RNv{J>LU51zfo^$4{!)VA7;885Q3
zGi%4H3mRjNNJEY9Fq&Oq_ZJ-9_XK3{Ge55Xs3~&a9bp6bvN-ozuh&ds%W*~}mp_{X
zdm+$=bdN;E#ztQkmhnRw>QYO(t7jyK%p>Tp!gZ^t<Tc6I_l}M`_wL2tV1B7P{sK5K
z;pQAG6%TL;X_5PCr-Zl}qn1-!XCs~_KU$lypqQrjszVBv)V66nJq6Gp&t6w=-u(PL
z@wHIbRPJAEMbinVe<h<KltSm}pa~!$Y5}jn&!e*aShb~8>*7VbUtc_R%Yhm5!t55(
z9ATc-zd1Bda4$Ti#4bLcNRQXSej`{v`|T=6TAqY%-svJz;HXyi-4{ElXV-7s_yHe5
zkwWHv7h?0s!Gn`zWJW2h$fYTYFxcn8WeML$rKK)JR1ga`S&7aC`k^r5(TB`fl`)h~
zw;L*F#_`zTD;Qh1vLr&I23<O}O?QQXUntK83>1C|^f@I4M;yVMQ%z1yRh1F;4iu}z
zvkrzM<)qy!&%^_vn>vzp5c>59!SrV3QGju0z?e7lb?w6?Sskd`8N3tjJANuwcN;fs
zKzNnSW}_Z4a?xKbc)e4<kX}5X<A^$m2u3Qhi&2{g4lLl~=sy!|!36<ezY8DGNeLU3
z%w>Kyhz=hbaOdIK>-5i8a%Yf@+pl{~fd;`(Pc?k14I38)BDzA<4LBJH;(fya`4UJ2
z;C?Rd*$$QFCe+R^(E^bvaFbv(mG~t6E6;>*8R!==8w{Ng*w}k!-Xi@YM~vGwz~4UA
zpZHBTgd{f8__KJpyYqHNB4Zmnb_;zP@7|e0?oo754X8U5{kEvz2l)ZIC&qQn(=pk3
z=E#x9>?`z{9@H(s?z~c=uzLJ>7B$$cS={f}AmjLfc}CS^lTJwRtV8<u?*!-(-U5OV
z(wm9`VO1yr;#SWrWlC0uAO%RY0p=-0Lc{ZvmX+P%O6U02(fsnY?a*n()q*CAt67`h
zW3|X1K{(Ez`<gP1HE`=}iQQU!4?&QlV;vg+fVP=lK7R2$9R2AeGvvnd&z+aZ@tP>J
zAw(Xsj_40tSUiOJEEdO}&X}d(FgXh~m>XUM`oL1tM9@yU!H4c}ak)ZfM0VxY+F-YM
zO>t4tP8#jd?a(@ve*XMxs&le%s6fgGvLFj4hfvYpBvYhnnAxGI%}8oq>M=uMwh02q
z5#us&`hxVanDu-QFNN3K%~b+8z>XmJQz(xbI~HO4$_sP%bMA3%QHNV5(Ex~0U3<jW
zVeMLFs4$E%!qCc-!!XB4I3DC{n>SykU4m~wi(R5*NJ-*#$9I5@y3O=O*e!dk`XDcl
zw89RaEERVkzC5?YfxrJIX&`Sw7|Z<z0BJ<^$R=W5dkpVXJV%GYXiSrfV@D!RtfR^y
zoU(Pu5uTtAp%OuIT>>$}mZCf8^_t}zbH3W;+mVE#NAK!?Yus=>>tWkCk$vL0xlxoF
z)EBZ7`u`KzGnP2A;?mNw72o&`6%D<9GdGh|%rC0MxxF<wMJk!){gC(jScYM7H|CX`
zU$z|N*WZ6Rk7~4T?b;n&1H_u(YvwcmVq747vw%GF^$~uMxYkxxoqRAy{FhIEWDO~V
z+NZ@O%NC;PVzvcrFUOU)v-b1n`*PpDeY+b!nN`N_A^6Tc|5XqqMEYq@*V)?Hjif>C
z!KV3ZpxL;$h6f=&_(+3EZA1R!O*qPDbS}{3K;s4BdVE(E3gwLcqy7ft;aYoVAtfSq
zB61qN$iZsN%SQa*E$oCT$KOxryE4_)>`NG5rD?U)g6EEELIV7_fP9nqx>M`$;Hqb9
zw`@tcHA;snn?04w(+5^o7a`-hZdrbD+T_Xl+oTV0^;3kDuI%d7_-3MDTzOKTqSm#a
z6L$mY?wTvz?jbF;uLGv7a)o)SZCHF<f5{SBw+{V$J3rd{o~)q9$drbo9ohU4pig0y
zjcDB?`^=Okf<JdNUNaDko8$VAA^t@VOB_Yye|j|7Rw9jJ5r4LMqUBOD_tKi@y0(P?
zAbI{n$J=Dqll)t6r3zR7vN!+yuB1qQ3o`bPA*mrf${J3s??;eEOMiXtL-dJDR2^uE
z?FhAVwV#Sz&UST)eVRY5{DOmMj7znxBeQw+=G=Nj%Cb+TG@(Yg!_HjOo;eX`ItsCB
zKQe222VMxvp<mBE!25bAJcJj6_U?2x#|~RP`>DTdK9Mjl*|c463wwdxT<yoh(hU_7
zuPvh0Kd|;Q@9X~K1Q^ZBWm~6#-WV=jTDnEZeCfGS-sZJa7bIS4j;k#m0qxGX&;{Qc
zaDLhjF%PyQt?Tw5b05d^69qVKx1xx<?a$S!T3<q1aAmECh}KxvQhoPmk*X0oCV~~~
z&tIa-XeS%zF%V_?@+I1u)RAYD*8Y9Rt(S<D^y<~Cop$#UA0;KV?K7Kc-^a{frmVt0
z{BN6qU+ee(ne~EY$UWT6miHex5V1tG@PgV8GL0rDkO?ej#^X#S4J}#w`L}%ouQK64
zV>jMh;zNfIH;FTy?9?XMSBk7z2QxO{&p(ye9a(?<Z(E=Iw6vJ}h*Y#2Q@YyEn01wX
zAZHassX;$59fWJpF!vG>$?C<S6E<8{(tmv;9Cdq`)SFELm`VHTcmdfG{;Ejh<ca^=
zU+DiR=a#CTYrEg~GBz6|_1HmVp3xsVy)_~Wa;NUh?rrpH`-9%<Ls5mRY}p%WlBD9M
zw)}~7P|hmDyp>K*0$v$8X-qL+X>{@7Alo31O6^*Y&$GU~a{D}RpwgPx`Kz~oVW4U7
z{+X)jZl7lko$%9=OV|{1kHYT!=hygkbfx|Jb?~S#?cnFH|NhG&|A($d!qQ7gE|S7a
z{*grA{Vj4*`QCJCj2J$A;4SKi^0Kmjax^v;*9kqJ*S`rHqR&C8pYG!q&J-tDKQAlm
z0F+VcImuX>fK(9WtGA8$OZ!x0ep=J4=?zsT9~TEigsXsYiQq6Qw)WTEvtsDz(dIxS
z6FRht$d5@#t!xCJ?h8i?fBI@)zChbLrJ9b3*|5Qz1p~QvzqX`>-l>PL;mH631D^|)
zzek7u^!K`Y{N%}#a5t6~`bc&VD0)YN4~guX@M}|-8UvY7+F4ulQ^i|D<ihP?zy6uT
zfCbxWs{{}jX||p}5Yc(~>kCA~j$o~FNNY9@eRO00`BM04S^xFVNcidho-g&(N653}
zdj02L#TFRfJ~V7LzqLl@KR=o7U%^Rj{+Ji|9-!5quuXW-H3xqlynX8Qh|qyEx_Yj?
zGb=nd`>&s8bB*1hKOe?3xr!RN+EMVh`>%fxASkW@=6Y*RQd28_`t-_~a^Z`K{W{ZY
z?0_SPti3f4iqrolVzlmt=Gmv^Tli0r#$~(V#sLw}SiZpZm({RJxM>k87&6x>O9)|L
zU#n0+vljNFJU=Br&nn;>8Lj)Z@TdE;<o1EKx<75)C;fWGM)Ks7$+U*Z9HU@pq-d5d
z7{>?AJ@M-`8XO|S>TGX6Gi5r7=<uHp^4IvEPiSOyCM-b*7<kN#tF{utBh)GXyaV<X
zO_)Bt6sdEKp#h-v(sQ#d{gwzfnAlH~(F-@|>Y$m=Z1Vl;aDG>A@2{md7#48hRcYyT
zHMJX)B!!Q2{LlB6=X;aBdf%hIU~Zjp#wTGX|K|@UfKFj6cC32Z&*k4|_G>rEtrC5^
z^VhZra9__~7pV!A?j^F~8rd1g#PS|qY8IA^<vM$I)z+<7-aO*77oXKEvEqwfXc42J
z6sB;0S|QRi#p>)*hfkQ?bw5Y`&5$}fepmQqkO+*)q=1U@1d(pbGym6gkpB+D|3AN!
zK3=X6USO4!17H;bHSc>(&CO2eh?#?R{kor@9{^1gDGrT_jKqR0JgeoQsji-U&|0vY
z`|;pc2#GrC>+6e=03Z%UHj-hVZ?!cwV`uKPMGyx^mjCqWO@jf(euOfJl$w|5Of^wi
z`U%nFzV--zA&QCE9mZUPQh_mpN5CTni>o{=&l9$H_pqNEeBX6~p(dyVGnE_&LYFH>
z@rPC7dOxQPoHk<%V$-k?*+fFcP~lH3b$|WIT7awLM~`mbwJX_W0e`U1VU#kaJR(L`
zQ!K8|jW5}lY(;B33Zy-5ZuHp*6)CK;yq@~jSM3<R$i+?<PHfMNpJ#J!EQ9ZQFbu!~
z6y=IQwSZGVDfu2dHX?wo%5BWilXjrr`UBOVAcy8xaPVG6Zcu{+G;w-?VwgBActioZ
zQ3Rbv*2LVK5hI+4hENZO4<4jXq0CQ4oaV71g9lgAvH^a?oA5h#j)2w*O+<ry0PPPK
zk{O15TuiwdEFI$6n_dj?*W6s66DMf4X#)O%@p%r35x#LI&v34{rpVk_V4OcU?H+*r
znm|lu9$!OUE3XGmBD8%K|MM!%<7H{-im&%!hvmkZK>GRWE=S4%g`U&ar*B_{Qt*tH
zO1CeVkw3Rh2aBE#HchD*>OQ~NT8t|~Qc^)vbM<Du|Iz|H;S0m+EYYq<ixsf}VuovS
zr<2o?>D|CDM0l%&6{#6$qcR`ReX*)=JV4^fpmeAA$JYkK)P-)o7>>t!QSK$|?_HQf
zcx!vbAN%I7SNr)Arz5AuP#!u$?la?p8W0%F`_R(VB>1_oVcy*n9r2E$J51Zl6NrnK
zA)^2SJ|FuBNB|5Nj0z6xYcZ5lASu~j;lH8a{XymV*hf6V=0rk*fMN(bECkzyt$h{<
zUeA1c#1+N|*Ea}vitev7V5hrooxou0xS@40y-+@8?ATvq*eca0NZ@ODQbBXM|EUQs
zX+SB=HkIwIX|N5mTR>^GnUhw=mrO%7S66T`{5N)q&>BpKmG6>&(|WbiHBTE^B08A;
z$LZwKJbUHB5#Hy8;$Jsqjvrfb_3iERc9?Ut`iqVLG<XjTH}aEu|3ui~+jdVzVSU;B
zK|>AdVSa99_ZP38m$`mc3G~#ak6bmp9Y`1nyk$Ys2fUI$y}I)(J3Dr+-p?nSZTRc>
zsn`=DykD|e`9w$0IpJmbHGn*{>eCRLo49f-*AMeWu15CRg@+qZ-pOs~kbAJJ*`S{X
zPzI;&x_|J`3v*B4udB&!G32Bp7DKz|ZF-WQznV@NxFR&l4eT(+mgCIhMHdq&sY!4_
zff49`=7TRZ!||=%_U@fIV}>cl7)(nBN%oc=J9g{<4d#|b=s|i}5+DG)(2K1>>W{$A
zTN*HPOUpoPh1^-3@DQXL0J3h#f5~ImAT+5kOL6(NcXXEbS4*ZRHm&23@P5lJ_x`h7
zWZ!;<lv>=g#_0VZG|FU`KLABE-Zu(U?27|@Xj!xI1V9OIf<B_tMtScefL?`h99J5Z
zc1L6gZ`9?Z(G-S8S^{+cmdrGs<fh<_OcMzX42-t$B<dWNM&CI;t^?pV=$2q}6&xJw
zT{E(8BuSORD9^XP8Vi@z5+L5ye8CXh;>Cjp3$Aq|wm#a>S7dhcx6c&{4HY7yBXoae
zgLcde{@c+Unxem{8M~cuQ~U&#ZWIk+-uEi^ycHy)Vq2hlVY$(vTu(_Me6#$y+bW@B
z5rb^$$odRxk%y|}>(Gc!6yDJmAK?MumHoGAa{^fQkwb==Mg7Wi^N0UT5-<2DTm48X
zAooh9C&fQgaPhglv!267M_k;VMf2xp03MCpP3o8ZW0Wvv9Kg!^o?vn%W(Ew6qwl=V
zWPYyq5e!uqISVk_htu}cx54%STQ9lMo-VqHy+2o6=T~@FhP)3K%vNfjCxK==6;(1>
zVCjx~atq8~E5t1+gg(%@M~BmP+7milSjq>777r+FXFE(T-Hd-ueSK`{%HMwm_y73_
z#{HbwXAW%g<hj143=G(D&rx$*YifA8s4yIG=%pBwDDQ4Qt8b>|R&dKsmSukN$_tN3
z)~bT%{pX9(niIat`$3Ej=y2}j+Un}?+p9d@-E&Ngxj8=!E2afI?qMm_-50gneA9T7
z3v+!(UrSC-P8rVQou2e-HK))G#m%<jRRHrsAs3qChcB76b>WU{M!vB%i*?*|l1EmH
zlG}T~jCyQ#J1|~UDg7&ZEqoQ!q)O7yPa6K~;<7A)>|~?MeP-0czUsKp31r4U+OJi+
zC@LxjL@eZ^CPT$-9sdp0RD)L*R{LYjKiAb+q`K}32${Mq1&qGhMOIdr6d7tM_=g+x
z`}G}jV;IE=Wv1iCxMHI*vgj&qR)6F2i60|f&MTmGG~H3qGSgNg`}$l!z(`=YCTDry
zL`E-QFXrw>bp#;&!`VAvQFD@HJzc-PgI;IwBXY#F?h849-2$Qu-Tu=g<UHUKz|}`S
z-V1+odB#?t_($PN<v{b;K6u|lwq@HL+tZrpWkzf2j^~owQZAm+6T};b3%?<(9XZ77
zL&B|vvL5{38&(PKZt!1}Y(CSyqxWvW7PI;rY)j$Zk+{51-pZF#G@8TDzxv0ol%c%r
z#S3m5Vl9N|GSk|K6IwaQYL6$$%E;h7wsgmHR3Y@+_)e$d6daCrs6VCmk{Qs=%RXT$
z=U6ylZ%hTo^vkq57JfZRGoe^F<6%8B*CEk#rBl4-C1k$r;xiZa99?IIJj8s8FnLm#
zmEriduax^xF|im6n@RZa?YLLH$enK{I^x$$C}yJooicrECCnhN+LHjoI<ansrk*h5
znQ9ToD)9|SxGhW`rIA%i)!ygQB?pKrAZTVaf9Ea`^hVcy2vot!=cb6(P72~{68n-8
zdi!Wl>|`V-ri+o)@)tmD;epj&T?eQ|HFs=WF?xv~U0qxdNU`AM1dR-R@Enr8Z&(9C
z(0Pk%I_pTxCN&9yE+37T&josIei?q|v9ZYiHkGcQ-bf=$ZtmiaZ}4fUOqW1Ab0ax<
zs^eK6SWamy7awHQ!Q^|rMds?iNZw>V%teZeZ|AOEW-}|m6h()qc`Zy@5udvK&EJPU
z<z{8Uyz1_^vM?e&3o?fU^Lnxgnf|%DY{BQfrWNCDJ<}e_)7;DN5DVI;U7+ya0zVVD
zdyYu|gFoxPx#BKp%Ai|hq@_oGuyod}fL2`Cyf1VEq~76UUZd_{Fq9em7$D#-Q=Sbs
z3PNq!B_Wf_@6qOMD6dg3Aca;K<cci*&#fItCW!2Sj;zE#p6=p&(3#UJXw7E1qk1N9
zR;)874krXl@6xQ8^(Dm#6<=m5W+30d?+3WLq>>}BP>exXV7i7=!5@SMCx0qS(o@U!
z$wGK(bfHODu${jGK-maM$yvNo$inEz$kQ+MX(lOv=|dwvvMvS#4NwPKh()V?xV(o>
zmEO4M{6$WvuIXRUwrm2l={g+L{Su^&SKtkP@PP7m)C)%jNy=DlDlUMmMNP1}AYctn
zU{^7@`)w?l^S<!^1Zr`14EGDxM}(ACPt2nHw%sSO6`UHFKty|N$ds^i=LG(pFvUNQ
zDsbRs+H*~t{r!=Ao0S!2n53RdNlAfBW2IebiK#tU&#oCTXwWBaT$tqS3~rH@wl-}N
zsF=BU8j+NtOkz^N;!>G55+65cC^D#_kNI-<ce<2rWsR7C7xAqAXWjF^6h2U|EHMf3
zET$>2($X`3tTbkRH&m(Mut})Fo#H8DlW4m@JqHEO$Zh;42u!n`YI9K)c_O1mG{E)F
zyVhN}kft^Ts(R!+Z0RWsIw!UZJ_1N_P>F<Um*lbw;S<r);cdvsocC|vE=L?fb@HOD
z40ohSsIs^}InT3DkQm5ZsY5~n9@N0rhA^RN`@`qYSGEZq5Vu%D5<f%xK<Y2g_h)zG
zFVxQ1dS)-(#QTSDD_}xbgj%V0xZT3vMNW?+9!>ddm&2YpgeF-PSN)D}BzUxICA|YP
zn!Dj9`v8K@`rgR<+5>*oZNlpPweaVaC;aEWp+6&~@Sh^{e{Q^=|M};yELr&P|JQ%{
z>UHk!O#!b`_C70LBETTOfDYj^hEDu%*!O=~jsG)2@Sotsz74;ud~UcXuq}NsR1zFo
z(<%_tq3Zj&D}zs-KQDz412n3AHCxW8PkToP(<)vAM~Pfe!r?<@SA%JX3mhyr)@gS)
zK4=8s@Tyg-5-R$O^c)lHwSN8jZ-(Iii6O{(TF%fvksL4!-a$f72dH3UTidOzZrvNG
z<ryh8PTO<Jmc8{+LBtuyOqpK2TM_T`n=3n$(P04TxKL%R^WW!38;KN7rY1k-%G*GU
zO1D1MekwR|O|e*Wy9RHOt(7TvlFNS=E#^u|W3c9fsa#2oU%y^k;U;prj}TQ(n@~1D
z6Ly{-^kC&_ACikM6jz6>@!m`*5fi<GB^3H&`u?#d6=XJZ=A`l^NRZe!Q@L@dWq)Ew
zk9wk>)ng9pMe>N38O$H|JE9y!rD*zdU%yT#b&!9I)=ki~Z-2rStxoB(ZtKnsm-vUM
zB#=;bEoWkVLv8H}?lr^!XmTE)Q7Ip^aM<lVU&GUeEHN;+?qc1&Lq$b}rit$)VfxmX
z-~bTWXUbgsvjXECqAi@Zkjs*cnk--b>C-35YV>kt;W`7nyn4uutnc~u$byzY{c|0?
zd2ocIP*m@XKpB|HlbaA%ipZU-W-2Ldr`6{0w!i45qzi8HKf1rmB?!R+^u-Cjo`3nl
zYw(|c{$=~G_DVntx*7-GN@*!Fgvj?kRI~nb3%AV;ZVv2l3oW^f#zEBI*ujdSmYuTw
z*xJMBTtK{p_B_O`2u9?k(c8>lu%L(Tj2PKC($*mZ1}H2Pgc>UotknH8v9(EK?ic?)
zt7qvX@HnAkLC;sBrmjxzHS^HuF75juv5QPssB(e9mLIj({+(sOnHh`A>%1249Kjr#
zya>0~;*H7DWz(yIzmhSD+t*3>)-pVBMArhcR}H#3(*#u%e9iE)LI*b=I=#HT0WljL
zoxQgy<_g1uXw^i5B?w5O&H$P*a*k<LV}=gBVDkos=new6zo$=>lXJ{C*su2VY;3bx
zB(B`R^XC&SF0he!{IS}eDb?!Z5J8ck<~KfDgBLm5)yAW><nNo}_mdX%6_Hcd^%1++
zZc~9GvRbQ^N;c}*AX47j%gi@o;I70Rn!o<4;vijhv35oP6JT?iyV>&I%$_O8<kUJf
z1V$pJQD|D|yr7PLEg-3wblx+IJ>vbQ*F22Zj7Z;q^H_QXQ74zZX#<ADF8k$AGsrf0
zb0dtF1(=Ge!etutV$KPw?3hM!;r`77t*W$pI!>%h0?V=R%X@*~pZqh=-2m;r_}H<l
z)@?a<;>3{lL_mmu7kiHtusvHp30;T@6{F5By)>YI|DyEMh*V3rKGA+cK=7H?hu!kN
zrskg<wAYinhfuAn&Y82yxtLm_<_MFFM(3~pTbRacpSHIF7dz$kkC;XtQ`|v9GCF2y
zDgw}u&7b%r+NHL5p4IMdkBB+Pzm*KcJZ7|6_nBh{a=O$g+NWj*iWJm4Pf_kKa=&LL
zaUBa~3X;yblCxh58&yp*ni;Wg`ivDsRMOH>N6UL3!<bo{2Kl*1Xl}#|B@~&he6&m~
zNY5iSaNhYm{hEutd-qOdlGS1BG-^fMw2zgW_W?^*m@;Lt#KEl&4$n!<KtwkqUL?;V
z1MXsCWytg`@&Q7knV9$md7eS^ULc!OJyD+V3uBq|sP6H-Rn_mU6ri;w7}Q~FPXDI$
z%d6hKA%O=!QoQT(?*!yFaz=a4F1<xhKK1Db=ez4R`el&~cK!JAF6gA$(M};r7vS*i
zg0hx$&%yY+$Z5Nuxm8a-n}Z3NTW{A0QwoDabutkKa%#*Le)@ZtFqn^HtB|b<H08wE
z(J{ks-Gd`G@YcK9S|K^rG4l#m0x-iS!eGNvjtf;jb%7GPE`|(I?l6IH4+lm-6oHWF
zL`zf*Jb6+OBZ8AYa~;$-=L|0r6iRICWK01$$pg$^Qs5$o^A2Bi^*Dze+tRp6`R&v{
zlyib<W{mt8^^K<sM`=Cp)nohXIb>B;pUM4(4z?(#*C8c@gV^|3m&%Ez!oy+z<k42K
zTj=!TFatDzQY2WY$ja)e$?ETz0i@}0-b3c~*7fV3v4MC`6bkCUdfPZC-`d(V9J)WF
z_j^&+l`egXQF2}Rmb<@_&mRtwuJr2Z5|{%NfZm5l_T2%)p%v@A_|y$hy_k^ctBJ~e
z2cJ3>60-5OG+JgD)a}XiS>&Hlj*u%q6mGk18xranyk#&~@(wR*Zlp`2E4DXs1A0fx
zC=P~DZU>{UdCA6^!*lQ=Ka;Tw$|ywp_+`#~=APP=qrhrSf3*{#HBnB~@>s9alZA~g
z(&$C!k3&Sw7>;laH;m}r3?CmK-T}44i|Xi$jyD)iML9NP&>*2@>Bcp#Xre!;z2CB9
zOp-!Y<sT4$x_8ZW+ImA#N<Oz$F6;~k?8T3s(O2r_pWnW1ezr^-1b2~V7R3s#m?^+%
z0ne~Y{Ns;5FxxXK_X`YU0O#G^m3trCE`OrQwde02(X)%bH5t8Xl%voef(uS(?6xa5
zsh#J0e(0s08832&8c$-&<X5L1{l27qpGHGCZ2^7;My6;YTLGR0y|-h+DXt~tkwT}3
zu0}IHBWwVYgd(b|uuj2`YQ~+wa8QM#F1Qb_&<Z*RG>>8BBm9J4)YH8N?)$#B_A~b+
zl}m#IwgaKS=Whb};n<q%bjgY23y|8rC=wW8D$u|L08qgYidu+!+qEXZyuWb?b>;4l
zFNV>oa^_6k=g*al90d7COC7#-r#_n5dQL8L(=f?o=;3C}y~!d3UbwT92^m<0-kh5(
z@Ln7wDy1d*`mu;#2{&|e=6MA&It@9zGuk%-83umBl&E|6zIS#C-%#{ZY*uzQ<koxY
zMi4_4RaJIWJ(JcEH2*ee&XcE4yOCOR9tEfWy26QWZr@o!;d<n43;vFd{!?x+STtfP
z7x}2tJ9MB8(Q2O?GC5->?^!s&d@#?;tbjA8Prn1%19@Sjm$r5~xferEt@n6yl&Ku`
zbae2M0bl1Lz62DD7*(xvR2*FPHx4lul?Rg$I9jM7s7;9uLPwLfwzkuR`U%_tl(v^y
zG~@k15REv=fn;cup9>5`m)lijs2Dt32OyVjSEIU7h@!hFXLTh*)i&&8BuU|>20Fb2
zAOv`zK^^Q!@)7sDJy=%HK`{?rTG8?ba5Q=)oR06@vBMH4=QV3CbU!BmEryFLn7GkU
zck$v}idJHXxRlhFhK7bup8%v*-cGxHgUiwY<uw;Ejg4WaP8oxTv!{eHD%c+ExM)8T
zvs``%J>(`#@E{ffkgZuGfGc?)kHU+xR|PkGQZ8sGDsnSQZ}txjo%ppU&%AY5Y_gyE
z)M?WMv3EKb2^iP%<;!728%h%-X|sTFXnuX|*h99NZ{@#XVPSV&K5f)LYiVi8a$=;L
zY`5gQ7df*#ZWWO8Obi29w0CfzJ?0KC8yajUoi0=EY`+o|8hVdcL&Xk;prNRk&k#DY
zRWcX44V04(gSZhaX@dR8jR7V>x=i*<WD+hTmds2`j1$Xg^Wa7c|9p?1M%QvHkVnlu
z^5JdvK9yYIWS;XiOuo9l=PK4|69v@l%{%Ga7pZ)_spOb3n?)mfkj~LA7nQPH#AwRp
zl!uc=?%zK;Y@+Dvc%dif)ul92VCTDx<2dTQ>r3UFUpWt~P+>i$7dr&bdCO;91z0Jw
zr;es(z_Cvh)2AfrX~%zIO%}HD^(I!OlYB*YD}gjcD|XOKC%*MH=1`vZS}KHjw7pak
zQ@@3{cI_-HD+2^dr0F7N;uRDpq%Hiac}$wVLY_@N{p`}+bgyQ<iD3syvdjn`Tn)Jq
zDo~adwz(sdXl!g`?v3#v?oCzKI3h*W=g-p;b8ztE*fxABe3FyHe1`pX(N@F=?rU1S
zf<42JWpVA+tj)nmvG_!1ZRoaziHYyn#sN|>Vj{6WlDXBfI^t|Icv-JzC2WWn888R?
zib+2IqsHqE@3N%cnFj!2{fDNh6nn~57A4eOf`C*=yLZ4S!UYtiIzu&W&2;2u&H4^l
zD056lS6BWfz1;I{gh7D1nVP#lZ`Pk!lh0K@htx!RR|6cHVfO|0<d(g8wL|l~pTQ_s
zV`?x5vsu3J(*4Ij0l7PW<k(3oUftVlU;kQ`My{whRwDKt`)AN@kCs9`_>99&{e>*4
z$t#}@VezrUhYnR@VDZ_dZR{Ifv0U1%Tik2}VfjJih$<hCndRa&y81f%?C{H|EAtu#
zGt{u=(~v*^H0m+EP)Bf(-zL);e4HJp=<3+Hj+9dK>}MY9OKRJ>b-$69zGWB3-Fhb+
z=+xFggMkfi-zEW=2S|7~n)$E9sF*k#E>n_VjoWE*ZSitbDnQ@z0w!4c3!L^p<Nj`$
z-$T=puHMwuX)A41X3t*PwGv|3B-nFdB3%ZQh)=V#jbmk%lnUzd-7pjHTtx(khnh&S
z8a;Zne~%sCh^S9SZkq2v*xF9BxWC!R1UP$tJIJ2~ikNMUvgh5yExJj<G~_ol`KR=c
zFwArnz&+<fhu^g=a0lAgF-g*wSG_6_*tX>&bLRt@MJ=j7QK(=Vb3t*$xN!jX>X~2q
z`3VLEfBb#e>?}SI#Bg~tga{TbtzeJ_^3nMrW{dnKCFba=ekpDH_H9Mt>qSU+>QoM#
zA~IzS(eN$=_6yg#ym(H{Hd>NrPKDE`=U_)!958lUJh)*Dp02KbwDT-*vZ`TNvVlH6
zO7C8;nR*RVqKG6yLTJjkHA{W)QKR79;=E?w`WD`Rw~TY3)AsEGPrT5KCNmSyEZ#WV
zlq7~>jq5I4rWPrniA>en0$i3U=&kQ>z9W1o+$_z#`nD0(*RL4?Vik<gCdXoD<0T?=
zf-}>i{pqYPmsj3V(bOz|7wc<30$2HK>y78!U>`}E!I^}I$E`e@ysm`{d(nGD5sHpJ
zo7xXoefXfDNUAtHc3)?2r3a&*%NeL9g%&s@>K<utZ=cPO29L(eNlDGbTAe{o80?Hz
z4c0bjn?|#__*RXN62&81Lh2IJyW_;N#c`!HcHeh6-^A7RZfEiwEIk80m~ly)4Y6s3
z<Oc9v7$Ev4KKd3=tW%%R^wv`AaGT5QY7?mtEWJH7owBywRZ00Vah;0pbGgi~y?qOz
z_7Gn<G%XZ7%}^B@Z)y6VOhX^Vi?F#}lZm)yd*x}jn$Tq!(gFg0M%Cux!O7igL!~0<
zxDm$C59B%;IPl}MOZl8wTmqPtc`7i_zNwf*>dKE|@DH7)+t|cyHg%$3`c(1e;mVpu
zaY|^<E-2AWOK22AE?f&f5kydujZ2Y{3w3q5O`14l6wajAZ0ytdej6rDp1gsPKh!Do
za?mFLiyh021Yb*(gDp=?Y~T_=4mM*~!|6i#y-}~=e>r>!J!RS1lhN;C`AL3HU^T>8
z79fLcgJk7QXlN)Ms5T`K*8&E>od@ob_N{yqS0{`Oru#yh5}PJYQAj{3L+Qhw=dDiH
z+eS5sgcTp&zkg2>M6M@*;4fa#H`RjxfWoz4nS;<m%d^3+wPsb}71Y^r2&Lh4`g@34
zY-XbK7Or9#u}&4z5D=D5>W7h=78w{oT=;+PcG<OS?2PSez|H5(o{i**isvR~r(NoP
z<e!9;Rzg(MP$f!PiCLRFR=r@!fN#{x1*PZTPwz(H-SVBHLt<<2j2+c)9vm4^ag)^j
zkKVoQ92`O|wG<RSoZVf?^-GvN9(;sClvLtnOj*-Lt}>UJoRJ;oU4R=x^XYNd70wVm
zWMhzKs$W{CiHFa+mlw9Wjw*g=rW|CN;LA_#sq^^Uj2-jSjJ+(Yh1NrY46hQ~v6DeT
z=A>i48fP%e3GKO3_x`<u+96j`9_wwXcC~Np3VA7!=c%7o73_`w`|rPDK$l*^XqZ9-
z*$}So?!z<+*C7M&J$}5@QH?zy&h!*Mn5J27FK)_&beYp3A?5s3p+<lAKI^(LrYgO=
zRk6!K1ls~iKWIV}lm#Uv>Z+>EY#tKE3b<P|D$UK!&x(qQH%Y?Yb)UG%ACP%BH8sJZ
z!qsKa?AaPWjKOGWco!QR8+-rjY+~Z)>HCnb6b9VhBm*LyhPWV^AtNed?fFv7@LPJx
zjR4D*c#&YO(fZAzI}VI5YrBaItBJG~kp`LzZz}DhK0ZvBIon-k&CNk<5N^E0cNnl5
zE?QL3v)S|ggq^&Djxf_ho^Bcq3z#uH0!nA-P@~)BfSk0j(et;qd%Wz@Fh^ueM3%gC
z*RDc6seiPZ7jZtz`SF<}6Ijhz_*pkww{st`{e-sgX%9Eg&(ym%M7<}^0Z$R{mkMTa
zY5+^PcE>FC>4tk1-x9m>&h_+kF9vSYMhCm(jMRcFm}s<ncN<T>i!!16Uz3OV`Covm
z)Lu0Kl{PrvHEDhKyIj8&$G`Rx>6267=-kzo3q6;0q=D<vSi*u$x4Wm?qcyHY+=xs-
zan)U0QyD_-cH3vvz%{L+*^VuoD<ZO#mb=AW*;M-5j@XfCMA@WY{PDxRUU;FcI_pEw
z?(xpw(g2vXTYT`>XWDOYkMTe8xPvM;a>!Q<GKM8zA}=7Iq~HfqUCAd(Jhi#IF0+5@
zx4PBIoUNS4iGq`J$PoO!y+7xF!k^IeCE|%1jR72UVSteL)sgRWlz!YR8S+QcN%yB=
z1d`y{20PR`Io*Jln{MpjFdyz4qAk(I^>FCi-%il7P5npeMZA3ZGSu=t$J<w)@#89C
za~`;3WlVj)d!9{Hb<~WEYe<dZAe%Cp3hQJJ9X{+Gz0_o7g)qEVm{=jF;Pv^=dl0<n
z7t`->q9n)St2XG^xhYAOo7C>Usd`2+bz!Kw;II+*hB8Rk?U=1v=XtO0y^POzAl`ve
zP4UaVfesxfAE|%nQNk>T(7w+35ijt8^#b&7BPH!S99zg5JO)I4yv<&2`X(&+j!u<R
zz}Nb|LLxo#$-DROk)GV*cRFW3IodL%ewLAjhDOKjVoEc%fWEAswy!IpY#aNJZKgBA
zgSjj=5;InFrZbf@blT(2GAPsa9x?`_6myyB({BnDxe5}y9gDBLQEF^&-;?PSybWIo
z-JUAO=Y<8&71fKtb1Fmbf>6V}G43tltHp4|y=$G+kaK(;pQRTO`g%Xgzb6V%p>A(}
zD`Ve>wsoF9$k$MqZlb1t%3bZQCL}Sif&7X9tUaY#s=j<l$D^?D&``9MR7h;RCt*iq
z^6S(kd}p==TOKd0n;ZKp1rqlzWXvOfK_~WKuA``RctERJ22gy}jrhh#MjmP<@=qFa
z>ewXRpha)(y3uyTn%fnjDT_KxSkzCwS{<YoF{@@|`tJHc;wlT}WzK7x=D(X=%>5j&
z!btGP&i)2^Jc%fNEsi&4DNh<MyLUezjbU%3P=f+i2r4J}y-aC}NY#;yaIm$_dR)bD
z(V}<z!^OJzR=TSjY+UQ3Qc`Bu&uw1cLP3#Zh=l?-7w5^Ds7&)AP7;ypUNlr!Q(-%r
zUR>vMg_Em6^R2K+i2FqAv~_3s#>#O<YJWhj+jG8OVkp4D&|<VnX7C|0oRy71@xqza
zz}c-b1EzR#pyEOyA46L)viTt(T+V<{)gO0wxLR8?LD3+^{GK_6=ham$6m$XJ{hvnd
z*;O#<!LhY`C$)0xNy>N6y?O0d9i_NHxA{KtS@)WSy}dnzLDV{L|H;<lA`BxJUS@AK
zm<2BgnST+&qd^)kU{P6jCHFv{UDm2*f~KaGYXt61>Fa`pymxf`z)K<~i#tien;Bg4
zg>l_Q)G<0sc-m8HFIIf@w=H~B*L3LkhvL}vPxpH(cT$(t?0zn8vqwo=SKF!9CnZuh
zg680JN^3JH%n+~5qF@I}ksqaMwM1{3UBE{YTE476|C4~ONcsH{vcfvW>4I)5gjwE?
zL2qg7g$}7krE9eEatY{*ZN^Hc6cFB~QJozf|KyBT)1@zzHUjUS)zB*eqo-f`eb<g1
zWw?f9ABmok)6LnxUe}>AL`a*^B+i{|a&uh^8|#?&9FI=ZkPj=>)+etuJMY;UYw&X5
z0n@#=jW(`t;ZY0+eFttEV0@s^(K%q)y~86$pBeGpdr|WPM>UC#+q)vfy6^>&H`9MG
zJQ<Bq!V_X<U7PEkJu!(-g8cm_rpg^&{`-fKTYNk_n(R8f&gaao8gA@!<yk$m`ZnCC
zP|{RZ)g7%P(zLma*y#QGQ_bXka=N>noz*06$eqXUv5?vIY!O2{K%$YOL&u`zd1Lqc
z*`?GgvgvmO`L4&j;^7ld?B{NT$#N{M%E%OGl7w58e$S)DPMw(y+~F~AHsx^lkG&oH
z124a~*V>d>hZktkorS;qmWW5s7CZvWttHkudwF>U_#N)4_wKTa^({oAcrh{0QD=Fx
zxR&9Ul;-0G*@U4F$AZOq{-AI|{*}schBnF6!-j54u7unJTZz5;<I3%D-@*m+=lgX3
zJJMYA<cQ2VGCmcFYE(48*ZV&GA-3wPrZW{BHZTK^+1c4mzVf``<HsrKhZc!ONapqS
zLPwd|de`^L@j8cW)oOeN)0Z^kk{lE?#V>j76tWAgj(x5;%mmm(4^_U)*4nvK#r2c3
z=Gp4Yj`lbJg;kH8svEg>jvpnTW#{GK{bFO~%oZn}{!L`xN~CfFWowj7@CbqYvFvFE
zoh^<p+L(;5rpaU_<kA~vUw$K3Yw~n^|Kmc-goQQKo9QTL4=}^XOIhzmTnPu)E!w_0
z^Wl=}PoJjN4ECn^KcQ#drDx)Kx*?1ys4E!Xm5+jY#p+e1PoFOBH1y9e>w4vsum7yK
za3G_`IU#?^(TGco*i;1Q<{Oyx9eSa^X77YIb)NbFi0_skWeg9VUHzIe4v$_`?9o|&
zN!QS4u!P#8ai<*Y?CNDjOON^ZsI7crECd!Igi2HiBVut45(~(E^5i8Nj^L?Nv|hV6
z`@YecHHWZF{iQHCT%(90n+(VeB9?hsPe@k;3WQu%p^qHU8zPd<;8TOpo<we+%4xzZ
zHa_L6u4vDhkv$*Jiyy8(&@rG*TxE8S-{_e3McZzAlif_TyVpBJth^yoOHE!rr)k5z
z8gsQl`o~t0j@#MVigr#R7ku7xp>qGmDQ+c^ZB7Ey*f=1YZs<sjwGR|Ny8>IC>@;$!
ztSQcZy?DyUxk3bukd-HEY4p@{ij0oVF`vU^9^~Q`PWf{jZf90j3Vt#{>BjsS&V={F
zoO)wVvg=>aO!JisUN{aLmN{-aPgqnnBpda9DAenX9n|?y!?icAwj;G}0<Hmp(S1pd
zvz|Tge?%~5Aj;A^FYsd|(FCfr_UZxk@dr-<zvF$7>lh*ZC7;hA=Qq`lZ++*LpYPJQ
z|J-3{U@j<GK~toC4#zB~yRZLwgVlT9^p3q6PA;p#2-KdWH$%UeTp_g*%HW<KZ;nVc
z=;e&QJH1usF1bUbQ?Pr68H0}Ibxugw<92zW#L{W_mdb9xRWu#~LAlFn3qBxWH5ri|
zIon>{J{26ip}1;rkKK&)EpOy1uPI2moBl;LATi73@|!@24Y8*kn+9hpHVS(>OwUaD
zEdH5AT!2f?WmQ(%q*Y-OWtLJ>(O7S?`3)q_$k87@Fv`HaDU4kkvf1VuzTeS>3Q@Nh
zG1gl08vy8RW?}-K;Pg_Dp9o{UY^ko00Fs1XMW_zXvns^%Un3ivb8qB~Q(wJWS!juF
z3-k4RV3Yu+bZ5`|TK$Y7vQ?qqF}ZQlRwk`wYi6udW0cKMb+K_#D!vCs2D@tnPTQ;H
z+9D?-6WwlYsupRZNd6Bg=Xvp|JA87>*RRHt*R+p+=%-~Gr<eMBiN&`#{Q;X;;Tv?&
zHMF95&O|2-Gws#A;l^`Bn+*%{&X-#+?6zOGwr^aaBgT)LO|v<31Ds<gs;HQZQ)b#>
z63^9?fv~tqxm9AGl<eqn<EFiqJCUK@({o2~p?~d!S^TImV=nI*)lt=nQPGnln=Gm+
z42fXz?oqFWW&yv8uJokAc%ppp1TUxl)*^d`Rl9%xKCWoo>`bbp6JYVz508Mie7nue
zG>sBg%-Ykq#$#?eNpt41Vk$@B@^d?B;8B-UvHv%t$P}w#A|ef<vJpy?JW5Y{#Z3>N
zG2&W^k7rXL@&(e9)<4ouGa?7Cl4ptHX95FD5f@N?q#+SNV!DwGuX<{(ymw<i2Cp?h
znvmoSXW&*yNZk1d%W|Pq@0r(iZ-2IVUS?(#^*k6Y7Y89;chE`Ca~Q9wUAkF3FhYvS
zlyU*f@GR&s6%DD@aB`9<HEp93;(>CvhZR#<==q3&uS=_LBEB4UkEqKrV0D<*STiva
z5=r{d<O!gC-WPqU8=?;fsfn78v<CmKwta<6!LwHtwP3>*1ugP*h%7+`>HZv7%*cPH
zh%%4iEwIh-K{JU)Y;oMsmhrfQ`hNM4-8b0#uwgm~B?8z$n?oqG4<0;7Ei5^xa0wGp
z&~Cpm=%*|Cl8Hmp<a-7xjag%TJLQ2=?Bu%dF8#JUIjMGdZE1J$V1jw!Mz^TElj9dq
z=SMHg7c_IgK_qIzxL;l>0)i<8k413Y>4=)O;BfsJSE-Owr-UI6LWu;Wkqu%D<@;up
zyL(Fa5l|1lkwk^}q-(&IodqO)20Kzxeh=^r>^@ABYVop-<M&E!+}qa*RD_ae2}YLX
zycca+Ymf>*%#{$GRY3Uh?w;iC;Xz<pd!ulBvg9zFSZAo*x9uwd9~BhjVEUJ$;xgy_
z*8qOfuLU1WUSRNM+a2vxXyjHh=_?93vT#MQiJ`#6D_dXw1=@(ToEi6fB0fPAk${e&
z;=iF&;~60+O<|k@J;!Rf?_oU853O__J8N|H^hsY9oc}af^k+xv+_hp)H(y!Q`D%*q
ztWnDEwbngooWSHeQG0Q3$T@UbsbFSZi43#@;wGKHq8ky*W}G{B?)L55!T(0;eUzB&
zC+kiy##&T4Y}v{~uJ3GX<D*X$`hLElK1N9)q}4jgEOT^(!j-64m9*@|@EB6KK?%KC
zTlPYfa9)M5>@+w(JNwj?En+5jEYm3Un#3x$Z?r$fHlMuJe9Xa3Vx@8`Jed>g?{K<q
z^gOekv{hrSCnX&jD1MznrQ=FsT4zq9*>c;bje@ZaI7vfdjJP1>6`CSfpOx#~jv;{W
z-)GEi_SPgnxA{)M_A%fTSV-%dr+j!F)^`5-6UT=4^NZK)F`fL;wSW2kz^8jRh{DDQ
zBNo6*N>}UWJ$+h9a=|ttF#|nKOxy;$#L@-@1!c$OewnLYt<?fz!a{I~`vxz#$Y0V?
zzAK$*1<G5zj~a7yoz}VPaWfUZESSn)m&oIALDg>FGb@8-D0awH#*%7%hi>a&)3m~{
z!NtVBugA}4?7AS>d2Z=jYvg`Lx(e6;jw$8G)+by)TV7oK*S2Tf)1C}K*Gmkv>-iwq
zYv%q_n3|e8NY&X_m+6AED2>_cGY;BQQZjz$fy*gFJZ{c167d##J=fUPzS7%l5xrfb
z*7}gIZypr`z?t;<(M=o!4CU&Wy>shU8x2gWUro)<&qsHj%T1q~Tc94D)-|fmOw~1e
z@$n$@wM}t8(xMT8926^?=JcMfFLa`cZ4E}K9enXz;q2+d1A^v$HM}=-$U^7e)3W!U
z57!M*q&8MpFUc_IFOSxP6WTPe0&grJ0U1^2W_?vfNfc7L2Nvl3ZaPwLhA_x5%~-3p
z;$-#fec1=EEmjR%`Uby;-g8!xso4xN2@Cs|T4}IpL?MPf0^G?-R=v9qoOK8KFf<ab
zZf-{oAC@2dbgC8j6!8oCUfD}?Sb$cspLw2NTlTt7bJY#D$><dAM=`L2-lN0TC=E;W
z21%NHJa_KeVAJ)!3bqgmfV*RuNHTi?5I{SW{M7R@iwMO#!L3ei)42QNuZFq4fo_GA
zVJ0cRnHOlalzpDtOKu%KZ})R@JC-`mkTC!H%|g{uW!7Fny239QTjl^Y3NLZ#*}F`L
zEcW9@pld0N#w)>Z`}tw)ET~pZJ6i+HZN9d(dA&I_nqCDOrv&fRYkSa^)%t9ZFNn8q
z@e}X9OQsA+5!^hq#b?wX3RX>r)YxvM^=yl&m+#PCNGhzpa=^5S6W2qD){gn`@#9K{
z`gMrK;0;ySaM|1Gdx+}E6>4Iy-u?A;f_Rkmt6?WjfvUn4Ezq(h*=_<q7`?>6dPcbA
z>i@ysdxvxV|No=frIJV_(y(V_C97--g+xXvMUq)I4J9ktDSNLHQnnH$L{=yxD<fGU
z+qpgZe7@(Ob6w~BbFS;0bN!C%bA3Law>MtT=VRQ*?S8x8?tZ1gHS~{~g3n4HE;zl+
z7~1<psZwU~>VqEMO#s_ChkC&859<1e2QvM%s3ABbU5}wlF#e$AKufd_;1D(mOlwJi
z)582BGBOxd@R?v}U`ZdPt)NqfA}4B{i49+1J%Ba1K;jFsOFAflqnv5J3khh^U~R;u
zoD*tBB1ofPSUvdA5|@^r{7U(b^Y;_h4h~|4+Q!;)iHM*-V<q?2W#*ur;|NPpujAFw
zJM;FjKT*Hm1BEo;QUgWkDLfSX%aA_AgymZE9i)C_u(Y)7@$9T-E3v1xnpzY8-BZHd
zJ=cza_*7Wnl~hj5v}*7&e}7SW>Jyv)IUBxMT!L)`*?f7C(Vrbw6vP(zQBZao#(_MT
z!<b-AjhRwV#R~knBMnq=fD$^pc}1RD>Q(e-BsUEw?!2pLymNcmCy2buNo#lO?>dr0
zqImd$FkK+6S!Co44trx5xP(ugj#&ePhm%H;AYQ!e6au*3lLM6trW)B(sIo&1>g|%M
zyzT~{_aLhv-n#`RV8U<w=E^|0akK6;-}-s)X}>c>nw5Jus-;01f|=r(1n=VhtjTm0
zN=6K(y4W1_iOi&#UcY`F?5(3UjA|%LA6m|IjGbrJ#P)(!h+61Kg7o$j$DmEO8s#`Y
zGpFL@x$imYuZR|07CJhjj)Vws=uj?vJd7q-3v^AHn!q)J3v+Sq$t>mgcKvDJ{3n+`
zWuS--Ir?-K=g82|Bu8DJysE`7@r9>japy>51G;u!vGMR&K#l7sQ%0MVVS%lJwvmnQ
zCx=*sl3bb$V8Z>5w+P&VMl(p*LjRzw;T9^uNUw^)D^ze-td!Pv*Gnb1om0#<?;G#9
z%vhd=hjSPY&Hd-<LM-BT=<*I1hkXn>PJg$$uDwXRd3^}q?J92HoXzXkRwntR{_t6=
z|J>(W<?nCb3Jrhxe4WX=d)}g@w8dvZ-_>^){Vr<iHI;^I^g>RZUFXU6*mR7g0F8i0
z?AU{;pX`sRH~w^_-l@*&O3=|G-A;9OB-rf#<>mRlCqQ~3b#!p9{b|G126lrJ5+=Y=
zC&zC$#Ulho?SeETcdyL8QNDC}%MABl<WWj>vu;1`YzmV1glK{43WFjVrT05|;oTJ2
zXDTEu00;Xn8N?;Y{LRBIf6!<#z!;LyZDbC`lK@Z;E4iX4qODaPdo2Gl-f~RinJAxo
z8~6xrh5HnYjEr3S@Jo->-l2a+8d)w}5Szg>cVmoLBj4(k3B%;LD_|_8T^&N=y8vzT
z2-SMn-Vb6RW(ogvodEd8F3IPHHjYXEhVT)8<(yOt&W5B}fFYp3hFlI_XL#kxPquIa
zn)#au57Mm8Yid$2(?nVG{cPSLn`RLIk_NrfK-`-8-4157H)rboNlkd<Qy^kXU+p&8
zt>h3F5A8OaTEUMUPJ>avLBLPY)$1cTu~UrZ0Qlr*X63Mkt5e}+>rIE^G|k{|(zfM?
z*X#E)NC3vVmZz$LdwEG8KH#(thhQWaQ-a3A$#b{Zde+#Z`gz=HGi+64?|z*m#FCRL
z3(z$~eL?H!1IX@O2R3^=Tr|xV@SP+!vG2$<YyctX(p7%}6g8~f73Kj_WP=v<K`Jn8
zo`j(Tz0yI=kpjdM^|`ydb61RBdFtN%jS`rq$B&PyepIqg5x@Qh`lH&|gMKRC;`&Fc
zth7>gmr*`b`4Ay$sb8R5<WpPoT_)wDk$r<gp!Ja+FHi2%x1#EWwzNL<dszEOpcNIk
zAc2ZKC0+Klxg@Ko3XP2htH?G*c1#RK5)kIR%}=_C%-dpg(m`i$@8&4<I8XYML(r#%
z8gBd(k;^qq2;3Hh6V=c4AaxIybl~CU2CNzm*w6+l9qh#m1IDWht#efu1j?M9K#|Z8
z@9~_|$_45)iG!&-bZ8O*dEL>$NLqFAi(oaX6rPF{7hE}WB~E^aK6bs{W983Wbi>6P
z5ui{Z?~yJYRN;sUNer5k`3XR=qTX&buwbn4*Fk|%1bpyYT3Q_w;A=cuikkk_D$wJ5
zb=tnFe^abjm*b)Nv3QQxx<eGx)jV0O1vk-i6SylMHLFRBL-J!MrWV9Nl0?<2KXi0(
zof%3YRs{oLRauKY3`AIsQDm)*bU0djS6cJLa4@=E$%&aUUJdQt_xIl1A9Ez}fxWt0
z2db;8@=?y^&b|nUxUaqgZA6b&`n(;n9MW???#<1j7IfC6l?y-jkYyD?@58dgH~rAX
zyS&g^MmpE=M9dx*i0_ux2rpPvszSuCl3XN?Gms!0!e~9<(wFZmJ`=YwK_I$1_fdxZ
zXp<?P=D5g57CC|ZA9v>teIqISvS@xSxHrYHt4}_luE-JLzt4l>Gm5MEX2&M_*cCVJ
zUmgRCn)Od8R`5ktz06{Jy~zzrlxr^!kFdsEz~nHkd5Yb7XHCisf4{^dE;`dU1XwP8
zeDDzR$}VFC9F=zAyWReW3m}N{njPwPn>sP+SO(ZuPRx#Dqc5KXNPt1c1WHwVJq&>J
zX!gf59mDR_EOxrTw|%TmqJYhMv}y*T$%|lh33G7r0*T{_w)wDUi^i^e%!J~}?rzSr
z>Rx%(r0#28i-;4dod{|7<u+fxc(Dhmsa4@{GO{Tmi`zpZY>I0tctpK?L#fs@ng}qD
z@1Xl+69}WVPen3pIVynxxgoIwSmJ@E{Y$&`;T#5Twi-IT+d$9SaoG9J-G^8<N%ypv
zlxXJ>4UYw_{;es<gAG7kp5)oV<UAC|5$$Labf$0h3D4h>x;sg5CS@pN93r$P(Urv%
zM-aQF@PtOQc2Ly5yJ83TzGwdEr!{5>RR)YyC7eOpX1=|GaXlOn4f7apq+~SPw<9(#
zwP*r4H{1N6Gwd~udn?^<B`1pQ3e0lBOtNxuUx-<6t~dJj?CS6Gtpa}Qf4%8)wqT_H
z?%e>ybdEZgG9Zp4()z?e@2IEJTk=G$26m_VC$!I;nZiPFEV1lC-*kVmms|f+ix=Kj
zOfv4S<(C%F;u`HX+YYOPPKaS|j_A!#eW|Iu@H_+2s!u9lC*k<$R~I5M_a1CB%ID#9
z*|0y)ZWN{=<RJMIOi6tG3r>FimDaMA7CeimuGbh~^`$gz1Qmpgdx?=8j<0%G6+F^b
zzn-^ZTMAZo)O_@CRc)*x;bWJ9<yvvg`)g7~<FjNLETNgO=-k>6%u5VI?2SOS%`Zx3
zzxA(Etn+|1cAAQ-?r{F~d`ppn7d_enRNDr9cwRi5b^>}!Gdus=%42UyQS72)HCIZ=
zP4V7`hl0A_*_QmNJ4W;x0DlBY1MTv2Wq9NfmLT&REJLpOv4g&v0gNKtKa+>ja_~Z?
z<)621-HP7Zjv+5IC_1Q&Y`K4e)Z#(4?)Bp4=9Yl7In;Rr24YSciLP%Z>*I)s`2dfl
zL-QrX1R@d#ZI!l(M8%g@Ki;P|dgA(6n>F$Q$OsSm&$AhD^*-2pL%FS>!$SAx;NVBA
z8J5*amgCEyByOZ)YD0Dzd-KjeTe${zd~f$3`m#BN@~5;L-y!MtANvE!+=f$*St#wm
zih*&{2I?8G0XBo3;yoqm;^z+!c@!HUus!ufPWZyx+m~wo0Dk_kjc}Uh;-LSF9@TMF
zp4g3rX7c+UZB764`uh#Fe3e_bW|7I^2GYJtoI)@3)BdsXw(LXFu`!qnJ+QCVhq6!$
zXT5WWivTaLUtJL%m)JDc<yD+yrBy)=*PXsYNWk>o000Z;ySOtUHg-Sz1xB1R*Eb-r
zR9&15QsY-~4=^dslbEca1>J<X+Y&1kc<u;-HAIn<ZYz@k9`YZyNXU|%ao&9k-jnv0
zue6Z|DC3{TPY`4O0;MUAaoUo*6~KsD>@50Ek3gMv%Eky6v`XBxl_Fqw#4-TOWgs4+
zJ)1$B_K7%x>CEY|&a*3ei_7C4EPJkk8AhTZ_f6rLIRN0f@(uCDf#La{$I?^hbZ~1^
z<JpQo{6uo!>k?M9*1g;z1u(W5l8o{d!uJNr5sP*A#h-}4KtZG}bAl+2(Kh9x+`eVJ
z(|BbneC{HcL0wO7pjXrAOVl+mA7Obw@9BVmisw!C;?^x?e@9@L3pA7-$67rNkesTr
z-4VNmXK7f&gYk$q@I9z4JP8kP1;`VtW)~iT%&+_)IE;_26`jCjM@)i*^6r>|x_TX;
znBb1(xeeX!e~Y?x7T#q(_6#rKJAxa%thJ6BG99UkRCNp`pcb3T7&!LaZ-|Ki<-!SR
zlTteR5x#<QA7-73JV>b!(a|^nCWta?^nV4-2h$KFcTvKI^V*tNeR1<MqVCK7xQ_>V
ztL|dKoQpq5qxi2EMK>UMfxNZh4HSDz!F_ModW_tB!9&_QenticOeH_2kWuVoUJr<P
z5JU=r^QFZg*n?)k^(#zLQNZOZ+&6ku^WmY9PWzQ!HiJ_GSu*hDk5(fWzJ>P_IQCLl
zX~Z5w()cI<4|AWb+_h%22P^4~{r0dfjEfG@UcZmt=L(NDcf=t<RyuJE%jgyIF)BG=
zR4$9A?pN_CYT?L*KLfyEk+~51aj3_7F3V%tf{mLum&Ihd#2{u6x50PM;fsFJJ_v0w
zyO1}v4hfx-vCMu2#09?df4z5>g;)}{M9M}T&Jsaqai$dn9r5&!a3wz=Wk>3%rLWI8
zoC~LhsJv3XYcSGIeacFX#qq}-<w0b0N?ykR7lRSYf4+dn`CY0f`6G`anc#fy&1s73
zvp%+juV`n~gPRKIx(q@S_zD;7sb>0vHQ1T1Ve1>-b--!SFW2V-5+j}gJA~f&hbODH
z`!sYT(ybDwH*&oA#n;0JjnHU4bAq>99T4#FUiBb6>uIqeoG>4E?iy^+KDTOZWJD&V
z`=Q#>vR$l0bCy|WI&y$jaG^`a0<(e}GLk?|p8`tu$SF;Lj}N)(Fz8)UNMynPVwaRx
zl$qHQu#0Y{f>y(G#UpJgAZ1^I*ZUq+bt@bmZFY7bfi$b{H;&V{3NIIP2M}H(jgySQ
z=|v939Sg|^=sxaqxxRFJQqbAwZkKVW6G-F1yrm(U$c;%=i1iXCvhb~J#o4*4i_?v+
zWNjW3``g$0fA}?kE&Q#9zlb;IZ;oCDAjwA&*3)<l)^@{Mxo`~WfpLTakWz0S5C)t$
zz8q?&FK$a{A&dV@cxaGIsJi|X)km56l1ID$yh{Uys>Ea?k4iOxPs{d^L4$pVZm0V+
zyziMneTrV|B#b`9hCoZLuQH#5HE_pGKiDnI8qlx``^5x$ZQyp;SZNsFCj&}oR>y#t
zn8cr8l?q*P9zEd2)QG4DJPM)@0E;Tf;yEklw~xvTK={I!)vS)ek`w<9M5cMLX1VN9
zCMslx#D=TqTr36(0z*f?Md>@epmV$2PGMaMW$fgOqOn5Hof82<ibx{flhR**(hYRo
zpU{1E3~2|va<pr$@$!hS;Q7#F0|k^xQi?mg-s6D!8%Ge50FE2v&@aHD^tC4BkECZ*
zFAL?pe0lh1wGV<HCj2WXC>$E>cm-zZLQv3gLFbd}ognUo4gm!!S==U%#H;qQOeL!B
zi1LvA1W1geKJgnT5tvMc{L-Mu%2oS8?7+-u`w>W#<Xq6_+*Js-OeC0cV(yttBf=j_
zkR`nvf5)UCY*1y$`L{OOOmI^OfX8rVX8g@drK8X%uLCsNtCX6~P=d#^_2k1qoqL!(
zw^NB}r#TbwH!Te*x_J>5KO?sKZ?H|hWmXorKkx^jrs1D=7i?6CSf)+DfjqiG{Stjz
z{pWQ!6+xDB3Q9N`<d6a&49G6kx5Gvd2k;l`Vi4B<XC@p0q~JPgl#~mhh2rGg^X~y7
zimM1X&fiEAD4sYm(t0lqm8ODm9>Oho^86cS{uFp+L$F1YG;F2^-V;^qhE@GC^)5~`
z6-ATdv^fWFjdkoP6Sk8-d-f~AajCp&{K+vRalWuR5U-%{oqRo;QXG+z=^EH5lpiU|
zI7~Lqq4TD0(C|^9DpsZQSH!ssl=me-zRVmKrr{TB1+Xg>>1R2$SgJ?FudtAjJ%%v+
zlW7&D4;#wndVQaikc<aQ_=%m)zX1;(by?p))6eJ%s?rU*BHMzXbB1#UoqiFsNVi?&
zHiNq;Ub#6R2Ul=Vu_E=S-~xI=tPvG$d`?H51<py?T%`aa@i5gqFj;Vxf7L9j_db@R
zD!@k9_&d`^Uzvo|n9f*WrsoJX`HSSb?iJ%lk+6d!14KqxM0J}sICv$FIziMlmaq`=
z0>}34N}#8sV=JJvpSG|)DZ4b35Ws%)<3LUgznXNMG!s>~%}3_r&j5(^JvuhF+3Wql
z2V&OYWde+kaUNU|k@TLcIl<nHM~f}%|IbARLUt(QFDoJK;UN@R{szMpq&t44+`5&d
zlB_!%<q><uAkCTEnJ!6|TySf}{S$GXzdo%G-MRDMda&v)DzA)9gYD;Ue2YB}c0Ihr
zYFmzg$+We&fwuOvm(0?E(AHE<&Sn?`0)ZdTT5XY0T>V?N>IQ3^B5g;qGce)L`m`YX
zJtFk|kT(kUVT+Zw#F_SqT+^<70IcEa{e2}R2Nb=oL3u;dpXrsLvtErwR>`eK6%GZB
zjaJR64J1v*F<=iZf>QZZ?U8ySOyTSa-3<er6U4i`zk*eXeKAIqj7+G)L-@I-9|?&d
z$zY7biilJF$$bdS%!ZdO4S*zjPciiygA@~0J(#W)4|fV4;m3xq;(ac{PMEr#3BVz>
zM)rj$k<pIP7Wa!JVXSLz_(5fm<ODEQk5)ayqN1e0-UTaiy<2fN>=m3`nR*}ctH?wT
zg-ci_*Z14SDy+`1qZ~oy1$B=3W-X2Fh&bF$3{gAVxQ)RYut9jy;mK0ph8lDqtRj0S
zUcgno86C3dGbO-}AP;lp;(WcwzFbSp5V#+dZ>F;^A)Je!pV;LaB{N@RmijriXgxr$
z9g{-`3UdA#@~U~Ln4@uJo`6MrauGfwnP3#VVNhUkFx&}e+C*^(USGkteQyh^qChTU
z1|=4^Y53nqG45!l$@S$g90KyrCHBsb<U^q3ia2*Z!Cs@w3fev|uUiIdlY;>A<QM|g
zGMw(ot+E{6<a%R;1qz4ahsFnia)CQM3kr0KBmI@SNGGzZEYIQLrGQ$2>=$}dnT?!2
ze0BqrlJ`b<vphesT8xWQIuLeVU23M;vvuclbjBknTZH}=$CmT|u@Hz>^MPez-&hP4
zv6gk>eSQSJF?Hj{;mzpP+v@cmN27T1!(KcgGp!OJTN1F4-6Qxx$g45me-iG38fU@o
z39>rVz~F|UXUpo_6I{i|wtKxN5i@D-LNm$7P5Jl;2{4GxIBjTC5_FEhlAMBKJ4)q~
zHLk6QF~JN+N!=~TA%JA)JIlHeH1f}%Rnc~W&4Jkh0b_Iny}=i7fIYRU0Q~{EqA$VI
zsi@Z}r{?F+A&nA89M!w34=RQmvh|a2=Aky=WzyKaOod-S*1@A)+eMuqmlw2iC59(p
z-Q!!hUv57muJ(Q>S~X>vmM!1V+{Vf(HTv@XK^r?i)a61$LOf5z0bN6^DfsC&N<ex=
zw=6k#rNc23eA5fZ0!iSZU6h6{zw;2KD<`6&9EEXZ3kZi0{d|R4(N5N@Y(rlG=paZP
zkr=_;(k%;`V`hozu~W7zhX7Gx2vsRil$?kvn168b(a%EnoMRUyD3);41HgjdPFZ|>
zeMx-)j>=YppP!R+4z{k^9ci*sxi7l{Zh!|q9De{$j;pBFXa@o<a*j?6vcTq?$H)FD
zFhg?_EDqKE6L8&rNBx&cKplBP+PpDwNX--(XBK*RL9_?418kR1SrofjnW#K~-5@0*
z!+LNNr~uvAUOl9^+PTlR>OiRYUWqsYqS<D*3*lp)fqS62idRSRL`={hd+oo)ARgR&
z@<}Hho6v`?CAkN2OBPN&@;dxCj;ImF4?wJ8XUsPCSNIj=79WJc0_X1Wt3lvH<^>D-
z19wz0{m_AYh9o*MG0=^ygFwN1s$;yN?bD`UBjqCA9SS3%)D)eeg2HN^_6dZpLB=Al
zYT=Du0)Pe+9ytOaYW(68-MDJPEXR7{-9jQF+R+1w`u!X=hc|-LlYkow2OuM~SnocY
zTaw9rlW?cX=j*Wf%Lox?&+idLtw|J~5~({H8weEzOHZf&=Rh2m52@S>6sPe`?-eZp
zZf+lg-vx*t21@hkcI_f13*;$4f4D5X-A|#qJO|-{6LFf@GMV}Q;Gq)SII@dq7$i%W
z*&bR5UxY~zZ_v1Jj{AuHh+nNG>O@#;oKl>sN@Gg_MI;>P-N6gJ!tx-q&iR0J{_N1W
zX#C5U)>v;60^&p?Onnv=)rlOopFU#q{S#o#%%eIf7pI9GhDO-4#6j=~4tjWzu*Ti;
zJ@B08=#+xg28si?&ps=?JuqFy#f~=Z^^*>_x}_{ktBu&IU+j8{KMUx#-+^@(<>!ba
z970!*lCD}c+YRV2c#De=fBqjdU=<Tz^N>O4vT~4a9RXmF&QJaxC_Uy+s5D@6fhG>e
zPPAFS$n7|_JVBI56;@u>tS{N(y^`_J3_*D9n;XZ!W*M*IWDyl9L>XTDoXt=d(puth
z25&AQ>+mkh`xwPS^25+Y-JPsGO(F4!89!#F(Pic5J0fMr`6KhT5hlaJLxJeeuEC!i
zr_Y?p!KQ))f$N=s5ZH_P);XXBXS$ctQlb}-D(&eM0<H@L%vm5qwG#^p3_o%r83sPt
z@$Lpq9W=YQC^KG~UKZd~O!=ltldw96Mtnu%LeN?-p`h_eV<hDoYQUUVJ`Dp8TtV|l
zT1U9>rd*^9uc6i|u|nXIg7CuU`YlH=DQ?4t9PO){>T^Yrw<C%bDIB=FmH(j;RrK-&
zXkaw(ID^Sd>Qf(?9Sjz@B{Z3DHlJ>&S;h-m0QzreRcLD^!RF}iW(JDeEHTIkcotf%
z#g|fKmbnZonIgMzw1$jPr(Fg4q2>!u4}6Ai#0f_nxbgaJ%;Kq$w$)sMx>b(A1!~ky
zFRUVz0+@ZnzDB@fUTu#vTk-Pjcfs)5g8m@2*{ctzw8cmd_PPIbH<;PiK(}KNMB0>$
z=zCuXDCkiFsk<QFAi2l5C!~Vaz=un|SIXJGaAXbDMv~G_XCV*XAvqyi=yP;435pt8
zFPWjbLFx$~oT&`{cGzi5XJL=XM>lD2)#M}ECp3=zflBeMoZVPRU@Xt@lVA+6f!E--
zMBw_27lqvZGfesWocaTYGJ?TUlmxWSKC^C78z#LrFD=G-BrQkn=S!9f*B#4<dDw>H
zGg%3Y1IaY`HyyS$d|L#Zw=qecDZEWb@%CLZ{2~+{x-pV4k*zYrA?I&A%REeq^q<-z
zE*_zE_nOR8qHY1&R&;+}&NI1|RS;T?BquGbteJDCWlk`*{lq4CwKXsvrUyBn(wqK)
zfuszE*r%W=Q?P!8v9=Glek=jF@?eZ1Zv9B_i_ovg+3dq9nEb$N;rkH9kI3cYDX~uN
z**}6GnS#4as0_fn!_mD|m_d<FPjtFk)R23@!Um3UQ~Y>)1lcEkf1-~SP`?!VS@h4G
zp`pAF=NUDoudIR+{6ShzlwtzAOem=IAuchFfSHBzK4x}3#t(#E-A+Mej}30@<B*wG
zDO8I!f!N#7ddu6dj*X-yG#aj@`>D6?_PVGHB3uXw&;9?E3R_R}PqOrX*>6;zbXY?X
zVuef^`VOYpoZHoeaaul$aPnB;8rjEiBPm7ef2=}8jfn#AC(*gE*Ih%QV~)z>BM8_r
zQS*)vm1+u#JOA`Fppd5gUrDzA<2Qtwpm(*Zf%pa)a+Bc(io|YY<7MDW4ut_-1XOg)
zaE3y5rUBd{9EVL&e%9iVV~qIYD>9^|-^ye0`%oBs85H&<@RJnQhjUYj9ykTX|LQyW
zzxYA^zvK1)T?UG`$01@P0=AJx<3)tTf_$(xR&c%d7SI>dPtxI*T^OaYTJ}F@^Y%S8
z)M^lppjFjaXWdFKeVHSiL&h~>=muUnjY$s|Jka7uYN2#gn9V)J%>+HOcyIWI+rkV-
zuAJ1_-U9oKTvr?OxhSaR$n{A)4&#qlG5vBX!s*u3o&@~PqR-!_FkXz19nx-5<<$|-
zW~QL%EvVqc?67??ZnJs_*8bW)M#~4U{;O0~h=kwvBomTh77>|~@|6*l2{(-I;!`p=
zL-|>WaVwF?d|UeOG51@N*|Xpus(Nq_e}6NM0cGj0S@9AVMe#e2MPUn|tL>2_CPv3h
zaeG<z-*ajuB}_ob!jDlT{hBX6&=FRcVt*76_TY#JJ@4%SQnPJvB-~kejDJHgZrc-W
zF@PYt-rJuKt)+Oo@=uw&F_Y$wBT4;`TDhG@*LDSQHx%Zg06nleXU&4O)Q=`1vt3NZ
zY>bB}3FEbR(d8PVrJnYM{Ve8wTqye|^*F2>cdi(rHQEZh9V)S2!o=7=c83y<8K_V`
zQT394YDuEh<cudJonRD+#)xd|k#z1lcD`W)-c5ykS9>ehgFb-%M~4%GXKxB(YyLfH
z`h8jr$!w&%1GkRB<-^^Uu#lw|B5xk65BXwAq<+=EMX}Y(ns^g=5(eBJiSM{+h@e;m
zc;T`Zz8&)KX<JCe9K5aY#P`q&Fiv=&fe#wh?Ya%4+5&-v4^YhtvR9rIN`xZ#eWCIm
z>p~(zJy%A+1Y~Ax#c4>X;7h;N!yevlheDmKNF&f*(8@T9KWXEc8vnhcwg61GdK*6$
zZa?2YLrjOCiUvi8vr~uSt#mWu6F<MU84gn1K_e2VSz?AdYu=WAMF%(t42eaQ``1{&
z4O~E2i+O0@i5sAhrK|b;dHKh^9gT(?9^Y1gdB@lO$N!y&w3Zz5{}+CP+ju_Zf4Be?
z|8IRrinXT6MOeI8$_}Ki*qujKz*fL<LKKQGR60OW#}{H`0K)Uwky(!@xWu>QT)ho=
z2RPCINHV2m4p1bp#cHFa76E}^T<a%kZk#_V``wr@bo*ZOMI;j=X(?T(7l3jEj?*kx
zN@UQO4?bGrpIwauin_emy&5g)WQ#nq(GV_ITYwz*AFGVW6af(t&7`pt_zlK?r+QTb
znH}h~(V!Rq`*xwVH>MiR3AzxKMm~LHK|qrrR6w$LM3l&(IokVyzB)k}Y;k&+jCLJM
z;a-EkNBi$>+QG9l2Bz`k1JdhJX!(?YZ2|0p=0<2#)G`$NG4|vB`<~{%B+&)=o(TX2
z3Mhp341<OY3n{%6V06Sax+~HZ!+_-wZ6Mb&)YmV>`rtZa@OfhVJ>&u;f(F9jh{O}B
zW5(ys<s;cd(o6mWJdPB32)|g@h~O&I!*)Y)UZQ|2(6>z92B8ZQ{$7hM6t~xfkb6KS
zfmEy*xye44(QLxgK%GIO?J5)x!VPS}oRicVGlO^=C~SpbTA@~BU3I{#!htAtA?F5@
zeHBDW4wex@9Vq)SjK|oAm(%3b2HO_N!w8V&V^dX!E-koX?MOZV)S&wL?88Hk@}YH5
zFYzlAzrsFVfGn2OYzWT=5I30or^l+$=}mq}1JE35HLRmVI~f3N4APw+$Tq8)w^Q7{
z#=7n88dIn%3`^TnXfgvik?OhA2juo-X-}|*^d9>!$oZ|VHFFNyszMZK3EBP`Ngr5{
zEMRpiii_Zk^u5uCmOJMzl#mwz!iBmn0iP7c^wm#uVx9I)FjZQaqP?i^tF7!uyKFzp
z>YNg2McfGlrU3_o@9&)rKlX<_dv?Cz-maG(j=wMkYf~PIrOseYZm}9^%kJC45lQ{(
zA{+>YFHQh>4LSDwSUhk%fDTfylsx+V=!=BCBZ`Rqoi6~JL!Vu6=qnP>U(qgOX#i0a
z{ze<zxG{zXcS+2M*%dqqkd9O*SC_k2qe4S<@<JiU9zcxwkJz)?+V<Zq-)!i7dRb^O
z2BZR>tm2&bB8#?{Ju<8F^{m6L#i-4pnIZ{?E6u+2_v`PE!0?!x#r5j4Z#tDWF07!0
zoEUg~)B^pMop4q}<PfX}y#dTf(j3SnbDB045f~CVGXZVN%s~a(A~xSS?(TqM2S>3v
z;uL@s=K%rY@VGBD#X!xAEm;oVCO|oEL8g8lcjy#Jl2qS7iMF|**N3N<_783Zdu^r+
zO5UiVhx487f{y{#;(!<ifKWulfXCTu5N|<|hfIK(d#0+UcD44&Br5G=?YXE)^QT1G
zssf-eS%`G(^~RM@6LQ0h*pY*s8Pyb;f?tut#i?fb8~r4j*X5smeSCVG2Jr!ep;2n5
zqVkD=SVI&NhYt0D-J_CkHb;uWb3!yP{P&&K*Iz(-xBSs#>7(2Zsc9&-Zn~-Fc=QYM
zb>7lJ^uW+cGore644@S16f3B|c-?DN^uQ&g**$@u{%Ej%21gD;`4?U5QK-YUWOKVh
zA^z~(UgmT@L5a6(Fr7#Cgv*`;V-y^YdqvObjS!@?Ji2VLbDW0C9qf&d4ruinLo?dR
zdX$xM5!|POW8@tm2>bn>1k0_qi-L!&z|BprN3W2U(`5t`A=c4E2o>!IbY;K534<XJ
zF7<$0h7u`L_bSkEwWMtnR2xm~Q(ynvT2JyX?6qp+`%x~#P-5@#*I?kl?1=YlMozSf
zJ8_8-$6-D<W|_Bg3R^(p6JT%(Vgwoio^4*f0&5mv?p?NW6?)n*M%A4&P+O$om$sX2
zzuF2#h|*=kmYKWo;{2O2;6y%FEGX2J+fY!XN|R(w7U+fLaj{CzVw5rwmFc;@AsVF6
z4KO1Lp5mp;Kx52(20(PQ0Zcpj8<@Fz`3eeof_$bKF-oX{jY0Lms_6G?Pz4D}Bz;xm
z1~^fZ{%})7GVBXG!8xoj`uhgr+55M)Q@nNmNt1w&Cc^l~#prhLR$P4wp_Z5Dk>lS$
z+fVFqniqCKxamN*Vke}&q1YYsQp)Uk?p)Ck)LiC#qc8+W{u&zb4B*IMOlCmUfEw=N
zy&cjI3;HH2KZ@<%wQK$J>$H^j@3NQeU=$w04CPZf0*HgqwMs$ZNhLM?CH5JC;+HMT
z(?Fsf%FL~UmCWUS)r3F{sHw1fv|}+$;Qos)Pzfua9jb{8o3CG8s)t~z_$PiSs=1(!
zK&l+3We{juEL&YHi@{9*AXDK`2iD@;Lx`|+-}-K;G8;r_;XH+yTCo^R!lT<?2qN;b
z1YjSjfAXoU{i~{M`U*Jk($bQr7&P=bDw$8kNH3#=pj`42{9D4|1w)MxZ~I#Lp{V`>
zr(h(3O*d}`7X^imI(ey0MQ97X6d)zqbBZML+`ccg^^vpG{H7Hi03!yleR63QoaXc=
zj+kv;@6)URu8=9=Xav?V3-(bRIF4}0pPUKnm&X7U-3$8-Lv`z94S>^h>sB1mfmszA
zS6xn06|T>6pUD7DO%SEyP2Xt|{x`$8CY*<`O~%Cc``Y_34X?548~JOR7>au<XrLJL
z2zE8myb77$Dp33q)hwSeH(X-6&dUOP2Km-NhH&uau`*3yK%m?Vy%9oHBwiRM{qAc<
zG&cZ#JXH4JJ#p(qkv91|>KV}}>-dCedw+zV`cKflDG&C$7>0g}6(-!YgK~L^5hEvJ
zUj9Kkvof6`qhn*Uc+%W_82<>>XKM3LwY8(jj3CKM_(+u+J5Yyp2U8m9!b5Y&*LDF3
z)r&idc?Q}Z{|T71v9iR#!i;XbnGe{omzI>99bpJ3qflKhr~FN{h46kEU)UG5ZEPkH
z5ZI!4C<)T}<eLx18>T^s_8I>QN8O};jBh3im@(1Mqf4j7YiRu2&?wl;Zp_L-L?(@E
zcXy%onV3LNk_akt@Ze0{?jVUT{S+%xF&#$>@DlP;65haLr5H?gF@ih^+jC~P0C~{d
z2dre}E0EO8!SkJzVgb-NS1V<NGCr}xXk;LC(KKz0v|!XD6ZG6`aWp!?<kI9MMF<2m
zFvPFGfCJ#KUaA9;7%bM^-5fA#F8n3ajdKP4Fw|dA5c+33HFWE%s)ibqY9g*iJE1NH
zIoiCOzsXblXqCeSnnPR~3F1G-O5{3D_Ih@rMsl;=G`#W+x{!pDP2s)Jci4|cx4lh}
z&?Y)NMA?{72bL;NgP($7@3yF@MS^wPa!p8@f<on=Oa6Er$sG<9N>121P>~fhz2NJ+
z5UpWje9MO6YUW$A+(lqTV48#=(1SuM4ijN7Ax<Xs)CSe`v*hy;4iQXy7bO|G4!9aF
z?{j^<04FCWq^eNYL7ojNBSYC?XU<&<SbFAeaa3JU;e~Cgc!lDe4G3Z|$WPJ!jB+CC
z^aypy8o{c=MdT@NA6&nVg6RUeaBo*hBLX2P4N1iBBp5O*A?mg9bI(xpCfOp8fzfAc
zGZgU}cK^P#hKup9Jb$%86L=C-V^mzu;!O5gSt1<33(daPq@YM&^JhPqE@6RIFjT6z
z#N-6*u*^uuFD(`(zQTluRt0)PZB5O`6JIrCmQnhSBayB93(EhwN}gAsw+U`)BrX|3
z;!`%kC{>Y&6NtP<&XU?=oGb8|Ebx+<Lk||)(MkYHLq4n^>+m0_ilMUq3y9r$pd$(&
zjUg0^aF*K`GnE(b6HEAGe;%T4l9)Ud`Bk>^s|?l5GX=-d7A<L79fCF*gIn#U=mA&H
zp>Y1s9U{ZX@z;x?fwp*5&I9#4Qhr2<1F<WHy|ayz`ZXYHjY#oHB+~+1TwD$l|D{8(
zE_$qv6NW!2k)Xsv!~fFg3J`u0L!d~HSb71FvhG|oV!UqC3gct@ZjW3ikr!&i4s;aS
z#~^oXhpp@T4mJV$Yte*v`ZUB5$=E`C!P~xp&=daqtmSr?OPnbZlw;fR1-4}_io3A!
z_%wHzHuNS!P*`JX=ZpMaAEeWj!r+Yk#syMuD+BKlQM8}cDhSGXAo<6Z2;*X-ZUMKu
zeN_pMjGKr-N3-B7Lxk564R%Nc2Rb`tLEt9&3NAiScw7|JaS7EQg-JXas=4U5jB7l!
zZlW*j*|SCL&%ccRzX1`HeiK;1+o&N{8(;q?AnyM)%=`cU#Q&2^$a$o51BGTN0n%V&
zj{w|}HC^i2;W3APraj%#C^;bsy|#+Ezl(L?r4X(KpqxZB>r5mNMGn}glp;!sR6iuS
z9M|EQN)z@1h`UUXOnhxdiH7{|ZCL^U;Oqa(-w25qgAPlHj;1E*W(ZyG07eF9CtNba
zbe&{9pP$lEc^%?Ip-LRLx#=8@wlYi+>_x(PKzS|jP4+G87MqC70c;rvH30MAvyzJ<
z@71AMfe8N(g+ZPagPz3<XC@NfAoKxg`Cm<~<v-PqO=N5h(*#<nU7<W2zAOnS0ieJU
z@(9sRKO^AnB?|`t+|VVgR$;~^5%eTjM2U+72t=5;hS(nr>%w2Ev9Dec-8Nzq|HH6=
z{sWp2{`5b)eC0|P5Ld8}AHIS$afQu59e?rl*LCuUgow|8-j=2&DTIO{fD*TOr^0E|
z8=y&XX(gkUk66$hkoGYD;Ex=bSkn9g#~2s}O4YYqS#BcD4IhS-AA_z55eT5E%4BRY
z{#MoZpF$o>eJob_JrCTp*e4RubQ)`dR0GDlJ7l>7#vADH^w)L6c>r0?-zZ=!OafGM
zHR|Ts@;Zcs12}|z-vRU>5Z?957aj+nB9PpW`N@qBJP@!@0IZBbZI{plf&En41LJK#
zAZdq=3<K+T8(<Vwr3tqZsFQXBBVsb>7e6y0Ln;xvs3G5tLZS@(DZ%O*m@OK}^ni|7
zfeVCuStWm-8KumB#Zto3H+1NndAn|h_N$*P!FUe%8Jp@5E-Wh<1vY&O-(tZ^LW8&t
z5HrHl<ct&IU!;~M-Z$()q)>?Tk@i6x3ZdvYJYIerX(Pr+K94mdzztBm0CW41t`%+I
zO*kcZzRFjqq06&4oDoLILI~3E<fI|wh7|yGQpD|Y8br03cYZS%$)iYpW#H$PYjB0#
z0pYwM>ZO&ap&h2!6!PSapCAd<+zj7BK<-e{UK#hL(k{apOp$sqym|?KZJ3{fY5AQk
zr25G&+cupNt6bWv4$4XDAGEdvPsqH%6QLkM`Hg$3Qs)&;#MrZA@`PtfAPRyUxE>>e
z$-vSJ4+&x2LmEF$*}O6X!laz^1Cb-)PKHMG!5zDzAb$759$_<0`#~gJ9w_P4xzEkk
zC50P3eL00pa;#nmq1ntwnU(AgD<zOn;I9!D6vtR(oZS?wN=8Z5j1Ue`lcAb86IINu
zZ|0`;$*nERq=6AD;<w^8f<Fu|U0w$siPM*|9?vbbL8JdD3f&dYlH5j+R(fDtNw*Go
z70fmjJ4LRsma-e;^z#Uh;*`J*8;;dXp~r}DGXcK=k$luXg6@_|qmI7nMTD}w<0l~0
zqVKb&dH^Y{=0z;Ru@#Sys4aEx%n_?K81sUI2_Rs->3|ABVy-YXn&IRVDnV^|n#y%x
zTcGXiLj?$sh%_H>96~ZGhQmL?&fjPUv#B5op~)g4Mhtx={Ii8tAhR6^(R4I>?~u?1
zu~8{FXILCYbLbA@m46`(gHABDRKhC2F-FYW6=^j9y}!_-dT}Sy?G<$3g|!f|!Q`Ud
zjl*bbL`eJa#=P<BFAp7)>`Z)?1sWN2bm?6zz^o<gx)`KPAX8m}fRWDBR6ZEU;Gm-=
zHtf3{;$YCifUliu90Z455{35o^Tc8yqZqXeeH#!3^fNYSADYmf1dNp}v}6e#;mI}Y
z0tuoB(Fk|CuP`BRc>H<RJvtbLqLr|Kz&L0b89@ys9e;9k4sA^awMnb))2K^HTmN`l
zrSR;CkPts89ms$O1_udZM&dW1FbKf_1+*YkmVb>Vo+ZF`;bG&ae^8JZ6P!-@H>4#I
zt!SAQnz(8NE=o`3Pfq|;vkx~MK#NNDDy*tHQV#-ssKgLn#zHG1UYGHbH&bkDH*JEQ
zmNiAH+6NA+oxK{43L!>T`(rlC_@DE6p@D%7urpX6Qp$$GPyLQlH#-<Qi2>zOC}YJw
zDFWyL+`!(*s4GxuAGmwJwKrzsd$laqa4W}eMpxio)uEZI%CiFf9CZ;#29ZROR=DZ-
z^#t@n0!ffKEVuk_Me_Pqt1`bkoecbkS7j!SS$;DQS51uHz}0v{u3F;ZmC^TAqG181
z6WQApO+JW*U6fb+y75qsKzn06HRb8ow^SF?C8D}n{`T3ej*XW@HJK02Ed8>N`BhqX
zbIeYn_2~GbTNxx<7(Hq!__SqZDK61@;dW=FrdE9=4oEzF-yMPX@P!C9+rDq#whe~-
z<f&8b4s`VN8c`B?g@sSE>VbX=c3F{L9(wCd6tQyu3PZBHF&zkc94S_l6BDU5@kOWr
zgh@A1GJYgMHWCr&fhHy;VLFiva5yL`AaA{P?b_%E5fKseIfUd~g|@nbif1dmAqpRe
z_uIdN!LawA63t&^{8d#|m>G`7Pw)oeU~H+L-GG_)<HPS>_AD&DfT58AzP^X7po<S(
z$?xs$5n0)hkwjHEfa&OomRL|A`jDvYy%bBqXK?f83^s*fuc`EoQ`GV{HrcY3+-CaK
z)zx3Wes!g5WP@XB#wluZ+X(4h_*pzO8`}WPkA#GTpD5G>VDEqYI;*8+hx~M!kB*Mc
zZ+bHY#YR1T382`I^?-BZ-A<nl#DwU#@p6qwGL_ZTu*!#pgwoYUbM`=@7M-wYSv;$w
zgK3x|f`Wks8HFqpXb#2ag@uH`YkzqA_VzN5g0o4sIJAc`q!YObl%IdTproXfc3(ng
zR!UqP$yN%trSZ4<Dx&|ike@j$JREa&L=PW+$sc7+<6GIklZTU26TkZPtEjMWVynZ8
zg!J_E>FH_g2dRr73Bf-L*#|*E0ZtSniSUn@Q($JMJTNcnjf0E{8LzR~e!r0jkvW&o
zpFj8V+Emnq`K!7~qus5o4mia^I|M{TLZdjkkQ$@w?v{=x=xQ#G<O_<K^RQuSJN5PT
zzqPeVdaOi#JEN&-4fGM0xJt4)>>wN4eRpMyD14ro$vP&2-d*fAIN>zwPitxAqI`iF
z6C0UKr7&{<NL*EQ^|hvCe~4LC{NdqK#$KXO1zVAko(}S;1%@`hc)@>O7w4*<OBh>&
zBZQM(efSA-tbw0Dxj8w9Mn*7S!wAXV>vz=aPoF&-hJAwI(S-{au3zUpuZtZ)C<SUN
z{*SY>v(R5HCFEfBwY7;>f<i(gIGlzH+cs}L2RIs`#)zme6~=EMm4o6@XlN+1OZ*<b
zo_|~+iYsT%$R7-Z7UFf|`RbaQPkJk0a^W0d2v}q0;|^<(M3IYgpxc?}-xo?~$R5)&
zGSKc_{pCBNnD)KEFXd&%e+>-W#6?0DefTMKZ(RUlxOTf^YaqCCvfCEn4bE)!VP%Rr
z)%^r-ngRRp2{*5Wg@xKm8yReaGd~pJ{O#K}VAFb_VL{|%+RzQS7`6v;{L@Ll;b;y+
ziEVY|ZyAP=xf&*6#F6{*)8gMhf3j=7Elf}677_|7w!C=pct8UOD=QiMGdFSP+Z3ED
z5QD?cn1mK3JW2l`F)@*pA^`8fVNd_Y>%zA^144verlcLb)wkZNV1J+s4&eesv+N0%
zJ(H3K@Grx8^%@aFmv%zg8F8*$7yFJGs6dj?f%Nz+%>`cjIW#1W7XT)pwx<J|(g7EX
zxhz8hQpc$gOJeT)G^Vce%K6D%!99db^M&h|SKTw7L%JD(Ljj_ro1<RZr%&5}(+DIE
zb2LGQYZ%8q*;-pOfIf=I^JvzCWN7~(><`Os`1VcT*f?g+IK358wP~KFVEREp8G&g*
zQIQ(cMI9X-Xiq^g9W%gg-4b2qKXvLekgX4pria7zXVMr)(E*AeUS3|Zdu^}d#F$*y
zICaXhDsI^WM)HbouMyOltW7UM&CGkvUEH{a_e1NCqh)>6VsirOEB6YQy1##S!Tz2Z
z%g2>Pd>fcVf$=Zz-@hL=ZBjWUvL>*IdVSoB7Z|XtOkDwPJ}?3TS4~aX=J><irKAkT
zEBK<fmvch+(7}UMeV49Z$KG7M69!?i{`(?86(D5OeCyhYl<0CGtL=Rpo6FnV3tkV-
z%O7k2S%ru>LzdRZhw@qxboc!I{h`4GpW22|Yg{fkyWjEGpo`%&z4`9r;9w%|53mM`
z^&B>Tfnwu{aQ0^Lhnj8JH=n=)VZbt&XI))gYqy%C9XMcVJelgInb{DIHS`A$?DqX6
z3Uh&fk6@a{oo|SYoPwK1_S2|4qphpU@*kQnCYvUWrWj$Z{IUm7cMH8}bGop)+Jb6v
z|Nn3Ss$r%eNF2`}^@WX<b^XST_z1BB2kzK3z5?Ka|IEOi&(7{%p~BRJx;9-Moje$d
z{@XA|f6-9$__1u~cf_yQdZKppV=LbP{veRTd?Za^6zw^37ps9TS*$1CD#<Bl#sZdY
zrUgHhg{5cMD<Uim(Cu^E9HLEqJw23Vd`bmz`gwUFr0JXc2k{Lj4-cf-BEAJd4R`Ne
zwNhDd@G2-MzQk|-XogJg>z8%09KUeopXF|sDt3gL32I;naVP!tU^T${9`9i=gx|;s
zmg54UC*aYepT9=1JF)uU+HaLqa&|iP!ETr1LHp9u(jY3<CYWZnuL7@gb1`o4PDO%t
z_&WaGzW46lWnJObJ$?G9<HUaB9bkaRkzemFa?sd+v7^0x9xq$IDJ41i17r{LFJD8r
z00q`LXJrkIvSk^lHPX0?n=c{$^jmZ8{CTww3n(d-ZwlN05D==+tg5>D6|rAkB}EgL
z$Vg!f3hI5&#mR{lisvt1KF(sP^ZAWALnhg6_~ZNEmSFHu&rUdB70CoSYR~)kH|&wT
zc4Mxw5GSqOzrB?oBRO@>p2cuQ1xWXP_`uG=0hJ}Sd8ov<j0P*hv_katriT%7<;&BV
zsC_NBjhG>}jZ|u>smVCb@&<7X><)r^{om@whPQevyhk_U8l5hB9K(PJj^^PHjAhEj
zPgz}Pp47vHmWhVggJv*e)4xZx0$f3P_hh+y08>qV6s>`gF-P)^9Ubl5qcn_jz{tqx
zy6ndyQ<|RcW5o&!h{l7U{4cLi0fT`s=4h^5gE~cOW+rmYgT^xG=!f;C38;rPstV>@
zOG=XQ@dr!yVK!sx(P_8~8kW=Nbad#BDY}j{bM=%AWnYm-Q!Ox9Kjx!5et`F|VIL10
zn<-#r8kR@<H{03$!JneEuarJOAk^Ky!9hXSEiLB@e`B~3BION*pKEJ(N$-(^8-5~U
zv3oGHd9F>jq^hO{+Xl2@bWj8`wRInWI+}=K5cvz)RooNPSH=alNq+U3?0(JEkRj8$
z^k@x5q89-dzq~}yfa!!bs8!$H4?rXH`Ev}M918D1H#UIWkZ7ZiDaFOEc))&2cUT+5
zS%5gE4;z3Cg#<HkHLlAQbHR8)*T~5UTGB!=-O*7@q%uIpfGQDP+DT<)am=5w*1ak7
z18{hJ4*RD-fYH9$ceR|H3NSGbwwhHsFI>QweBQ~Srsn2XZ{9e-FgGP{Z`lv6*Z1%5
z1@e4$RKz)6MSU?f5>4FqN%)kX-}dd>8Qlal11!$Xy}TEih<e!UY7DND8q;g}(QnL+
ztj2aG^0X)9<l4P;^UfF<LDKvQ(j?W~E`7uwDDD3K@nfR5(pO)q8pTtO`2qS{CxlJm
z^IQd#Oic%%5eu_lou3+`WotWy?v$fh!#DIJZyrXkcw(k`uFKyW5R}B>hO89K)A@7f
z%*@RZP<)P&i3!|Rx0{8fKD9=Jx&k$tRIO7uP_V~--rh)qwoBEkGeXMR@8AxE9!MM0
zbvXFy@QP>y`t%*c4~PuI^IIvcyUkt?G!PIyW_8J{mQzev*d7B^^YW5D&drW@lkSFx
ztKs7vXF@_lXU4l@N(-&9${4SMX;Vmm6r(B-o8kXuyj9vgj6fJ-b(c$D-??)fxHTF~
zR93I<prYE%`El3Ioi^xW8p+vcX=!=>{CQp7Rscy2_3M@7<uNJ+;h2i+EyCY$UlMmB
zW1_IZ*)Lzl-c1=87(jWLa`R^Dp#zxeR#`dF^WLj@x0{>WrhgNnFm?M{A9{AsH&P>(
z2%~bwF7>S|UlOLn%&yDM>Uq4e&(rH1?+WT)Jd0JE@9uvH13Zh-orZqTn&3CctwI*t
zJaT}BmKJX%-j|XHhC3CY1$V;@oyZNqML!<Bh1xw9Q@(EBrof{kOxE{*IUoB%C8_V2
zC1xlInC{p$8!pCW?2I5w#D)Qdk85d_rU}8_X1M?Ob4Hq)vU%PC0T@R^-!h7_Wk?%c
z#Os8_#8&{*AgG#t>iYFeU*D}QwH+OiN5+$lg4W>3)}8xX-`E&ba`*=-gc}yFa`N#d
z<t#O~w4kt5$$&t$0Yk>1SB5pxeXh^u8ZzSew$daUJ6>40fG8em5xRti91w<*j;rci
zt<$G_YDbf=CIn|ajf?9+a-X{EHSQ=X_n7yl^%)4<N3m$E<{4OmTHF|U7SvEDAt005
z_y&0p>NOV~PIz?JJr(gC&ra0V(WzAT-?435NdBR#q6?@<?m3T~c*v!Ndv8f{@=x^Z
zU3wE<)!YoZUEj_FS4AaCH1FL;)NQ+Rr26%v4mkcK7dOT&i}w$0E^lZE%oXh3vq2i(
z9Wgc<AON-II2?7F>PHgifmm4a9UJU7Vk~IeR#&TX#h_=PlOak#FGoJ?lD5mKP=9rq
zB2mOR-^Tc9yoY*<;jjf6g<ICk59mR)JDAr0-aR>a`SmTa|7k~>-a%Zw4OQk?`w*Jo
zSe48!T#yyrZy}q#<?^-M`Nc)8dw-{={rvq+6C=x@(TTd-Wd-^=dD{`rh$l}Tn+_pz
z!m_D}a0&>Z2ZQIP>iOfx?+9o;$;|D5sX)+)`HfBE>6w|u7$t|qHmzaGFkeyOciQy<
zq&S$O8g}h*P|%50jnmrNegOf4qa!$O(>%Ipion1hS*wYO9y#YVJf#Is+*S4ro;=M1
zym$YK0HiX#U2Lu-VC3+27`5y`n>AX`a~hy(?kV8b*Ec?1-+tiO-p;D3R8*oGrwcG)
z@s<~iCSAV%T@5VRQ%MfP*VpaqejS?Uf7s7UlLW6~9d$}sd2@mr;;Qv8YEvDm6gQda
zQ3L)ryQpu#5Vijy*)>H`n2)seR}tFQ<<R)OfJFzEQ*q<0o}S7oa~_B#x1KqBn>@zd
zR}z67(rsDYMsbf7$(CobdNkSRm{?dm#yXD9&N@r}Fv)fV{IquRSoS$rjHKe$y9pZc
z=*H>8633|lagW9Zo#RDPclgMWgr}%^Wq*fG$-Qeo(8eSqDEk-cqa(Veh&<c3kBIrH
zYq%r=1W@h>JmTk$aW^b1EFV66`J#C(OaFcYDzxuzoGhV+F~o)p6m9E(!N*x-h>FJv
z#Z)b8(*?wF49qLQ{#^ZGZ{eXJ7$~mxc`E9)ISuY_B-`lxL9V>*3AF1;UJb&5&Fs%#
zziMu*f^M=!fX?j76?V;8pBn&F{oF7i&s6$sUZeI!`-MI<P%1pwH&pNnvn4NdeAlL>
zPM|u8$3}40_8A6HVZ)-ZhzJV3?Up0^6jL-mEyB6bkEt>uDrR;0g<7TT%gme{TkuK_
z20nD7@8bkrBr12$vO>RP65np$g!uTjENQeGZqT`J0QZAjyYs<L%YGigY;1b^_P&0e
z%Fmzuoih<>+rHlEEf33@R9J|KVMV8wA2eV%vci%vhI0ny(;G8LNlQlynu?2yf6+)%
zW>JKk#sH59y6zzS<A70oFx+(;5nezkc~ts-#!|2@!^BKG0)0g**q><ub$b8-uiv~G
zZ%Wx=yhm!!$o8)S?d3y}Y(e54TOU^T!k*bmBkJE)x{~%wU>Jkq`2I9y@_9C_(cis`
zuv0g_!6-rJ1v&*St>D6KuQM}wF>cO8mqZ&W%sbb|rKUP!DB4K49PlZf{Ymo)XZv7)
zOjlngq67+Q@WcwLUyg6-`Sj(J`z9x+rZ7Qojngr4@h9W)m#Wc-DDE+XYAre+PkUf^
zoNp%r)_~%;jEudEjH&`gCtFbuLXI8?YhAjG1J-k(W#`pp+Tm8GOW!dau+OU(nzL-X
zSuQ|hCXVCE8Q@<ixj<%TCmr1kd^`qH+nseQLDpKt|N1skZiqq3AZ9*rpa)RebRPAv
zTjMHN?29;O?Vq70!3+i9pdgw(d**>g!BlO>9IJ-p;^IhW8Q}uJ9PI4uLZa8q%!oi3
zf<8zQA<LiEapLN~7Pbw9+ZXBSG(!{UkwiRk%w;yS!s|*|#0H-ou6t5LS-0GMoP1ek
z@A5$kXRFU%IO(wU;`f~K(<xWHkbi)LC^q(G#Pg|K+3JeRofgJjbaY6v585bLK(rsn
z&sy}EjED40SHAnQ#NZfiPq}XsHxCbslyit5hmZV0Z?A8kK6ytKoze&W0!RAO_nkX;
zj(U$nZJ=4c4CNr8wb8jTz${4WQ5}%X-t^#owA?SDJDBe%Shs|9d<gN6kbNGaoVqiY
z5MN^Ws5|N&ct&h!n(kwD-rZ^jZzIr28kAW?$IJ|3W~E;9F`z=uF2T@JU0hjqQ9OcZ
z4aFFYW&HrUw2~60p={c)AzO)I*Dgd+tjn99-OI@lN|&c$zPUX<<a^pv!|Rw!GFtC1
z;9k9sBH{o+M7C!j^Z<<CSL6Uyfo99PZN?H%`UT(cPiMyS=V04NPNDWz2Snl8(MEO*
z+m!1{h1+zQF59;wpMrbO>%>I)d=@1orMe?5tWyhaQ?o!GfX5B}m4^9~l#DDcahx<M
zwdW&q7t+*P;caU3-@y537f>L)yW^htZe?c%%&?^7MTEvcl2WXkR+b931!kf**2{1Q
zt~eVn?{yG~n|EtzHNm*m(G3onFP4Ky(K|k_(XwW|Az{;?H}!|_eOT4e!X;?Oqqvq7
zF3hG>#qZ9@!~{uFwt>xi>u|Ed_zkPORuF}3&P-P0fb&AkijW#tfAeV2)2G$*aUc2p
z_>=I4I5n!0^%c^G2u)94pZoj0*PEx`iggc5T%0Z$F_J7iP#bwP*l6psi%4~}J22t#
zlD|l*v+U=%nW^1yGJCphyP<EQ-Gvcvceh=YP(co+ooZ;Z_8oXR2VeU2qjHw|eN@*H
z@nZjWKECIu7HpC9!H(VB{tm2%HjaJ+L&LPA+P$_z<rOA2s;ZQi4h`y^8wMF(ux`h5
zqZ0kOR|h%job`~y0yRngY9AOL9wTf{Gq|cOFYg)Bor(O;4u=aqDcFOXiwjrW-|&?0
zh0k%3e@3D3osB81*CcMVq_?06dSzwNO7TdbGOKlo{+|l^m7gxOQ<n6b6KZN|a%C9W
zt-x!iDgT$)JfLFcp4j>Y%0>&O2L&ZWt$)b&D6>g5VR}TN==<xcSpI&0;9?6ReK~0#
z*YU@KSEN1pB|l&MPMGVk)#%29ft;eEVea&9I&yWkoc2h5w7r+WC;U)?_{U-8Dp0&~
zLzN1u^=f-0>0aLYt*<!Nh5(!G%EA%rXsM~08K~e~xUQ|O-CsM397pu(Px^Zh;zv-<
zuy;7nnevd;(-L8x<J9i%?$aYBtN^E+^<0CLS<`h&UF9A=e8?c}DxUX)&lGMqGVkor
zw(Z*o|3eE#(1<BeLqkKr*W5uS!c#`MX-Hr?!lAu^`#1N5ZXp!`Oks3!ajCm~!pUhK
zEOM3}8(X7*Mke(O!b}S%GRqYh=Ag*)wD7aLi}0rMm=A!xo4M*?w8BC|MXF;*)`)na
z?;HSj5x*g@CPB*L!<Y{CSPVHY!=RhReQx<fl6UY+Wb{bvnvyS)eHH;v&T=>t3;37A
zShwFF7}jjWYvSdpp4S|`bay|lTidUFTaDLo1g52N-DJEyji4E}q%}Pl&{b}05BN_%
zia%HoXkLes(>0vyF!D;zQps({so~a|^ETK|9~B1T&fJyv>Q!Qp@(?DJxXj+aeTFJ)
z{)ZZWPf@1lxbv=_q}T0MWdbgx>uwkRNIe1$on`>ATF<@E4hUP~<7Q$wxcP4B3HB63
z3h6*=Lqe$6H+(Ub8VLl@_xiPHL`gm<IzX;b*V@i-SoXDN7Q&%KFhZdW(A#U06VIP#
zsG}3QjV}9AwLOioi1R%Nvb#7ro%TSzA*MJkHg;mNPl;Rl#P)kO_ZEVe)$-l5?e>|T
z5A>`1@&zQK`_nPMpdXCO@UzzW=X<BwhwfX}Mqxe(P`X21y3l`FnzyTqm$gSYd&&54
zSMsU7i02fi{U=4i-ZuG*<nTiQFT>Rv!>|->p+$Q?!#bcfp<cQ~5Mx-mCxg-L)s94q
zBnx__#jrBaY>PPNz+XfgsB91I`wx3Ils-I%ND}8n)MLd}C<KZLGC;DBe;#?NATQqm
zVPnLjt_cIKG9K;iL(TIFCr0WOH#)z*#N6k%{Z8g6Z$~ozF3t;JK~@?7bk>Snp2ea5
z|ERVNA=+=SI6>;f6DaX>>YqozSf0ha>}ibBN(F5yu9k=huE)#7L{IkzK~L+jz48Zl
z8jsd@bw%eM7ADGc@7~=hrtqE28$`Fi_WHAihEb|6gU@7#b%lU!Yk4=$Jo*!yIA%8l
zpAV>5c~y+`*T6!|rMtegQIEyaB&%y5_JfKiJG<gj^)~T(bDQM;CYeX#9Gb{)>$ZgK
z-wfdPL!}*pgiqDgCcJW;k&}GM)$~S#Tf;4`U89twT>SmlwBCV269#4-h3tD1jtC*m
zcP2gBJMZpiUEvOcu%)Oddn^65=p^%_0RADiWe4}z9|PJWQ1wV4@wvHnK-gd*k+>!b
z2x$szDu%1a&jD6Il=s3qB$B#VJM3|<PQqC!wsTpGq1&|$4Nn=DYQ#Tyk@~h&+d#bW
zKoV^3hV|>eXw?%Y`=t-thFyNtR=j%s`sa0C0Rby>^ZDsvtq-^AOggqrujRGJt@JxA
zkubJqaC1`HoJJ3?!FO>$1u_utzs2T-<UB^=rs|v&(TY+Y^+1r1+jli)eIabrA-}r6
zcMe0c&$$Li12`Vi{U?Ura}YgRd(pmb@Sxm%&)uH;BtxHHp^_3S*z=$P)J2ZT=FGP#
zHg>Nsyt|7q7Wg^s`oPf8JW%N2pn-|x7Z%QrwA3K*1|t9CAfLoAhvo@uD$GUSLP;rs
zV+W*+&HW7*H+Qd#fw}rd%`AQNRO#vM(P_UIt?mNNeXwekM|~P&?KlTHR5}MW5o%yf
zz()4@hWb|tTn0S9YWupuj*nt@raIX^uTMwGNpEYko9)xNGCR(zgHW}{7Y$1Y<f+>4
zmJhiGGk2HKb&I%`cgO$l=0cXxs?5%<{pHrO+gP2FIn?S*c*gU+-QWhFIB{agTN!ye
zY99bt&}o$)9Zfni+Ga|Rwk+Q+!+dMx9jI^g@Lu}N?)tl|Lf1d~V1;AL<rOyL@A{m1
z3YA<Ve|I8V89qR9cR!elC=Rg>jDegVDYLZRU8?bjHCLl&J)jHmH+ISTS`c*kWwab)
zck{C%5hsB)%Dyy&3%X;=mb6RR7lPn|2M@kS!0g8tDZq4`8q*>!UOx>Jda{e&NX9>x
zeaoWdDqEZYPqmTcS;lEQx*WbU-(s(c+@aySM#Uy@b&rg@q}18y$VjGLyFh9&KvW4B
z;e}Dj(~=wO&v$ZYW}hoYXE!9aKnB87s`$mx-nBC`G|DUVEZ=KSHL%2*7On)s_<=rS
zD<HCst@AySCffz?0EMxQZrvFzn>1<=pZrwL^bqEf&(lmSANylTUux|sHQ?6%hnb>z
zLk_xkVk;lBQz+#HKF_zLotmB9Cv>?A(ori=gMfW(Y&wAAAV2A-n;L+&JI1vaAOFJy
zRFr@i0sI^sDgS0?#3EkK%sja~>-*>tdw^(3<JM<b{jS@S>8E#Yhwb#Pyf}3E{DGJJ
z%Rrl3)#{<w&|YB2vCKV+jFZd?eCjq&>9)CeedK|C=KL=P7guT2=R?j#>N5Hp-z-c{
zzJkI4?#ZQYaNUb>wWsLJL)R3A@V%p(d6|>b+d*b!j{4(onOg0l;<7D`*}Kr4ZeEa|
zuc@YmVGOaai|8S4N$PRW6BGM?{LqZKm$4)7sjPRHs2A?pEc)qyc}atofZ{`6U*9mT
zma3{xS=lOF6cv;BVIU?OnLh!bgc`0~H(r*M+K}!6Y~dajLG_V~%e$EGu-DfCorD;g
zmoN;+{(ee(wAG!mk2mYIfE_|%!odN9PQem9+#Z7sp0tD8T3p#R<16dpZeoVqM<eg>
ztXoH~PI+Od>cmo|`{IB1=wC2pq8zNRkayAeAm4yUr>DJ)!QK<ta7<*tv|IV?bCFM;
zFwoIq#&8MV;!J@y90OZ;b;wFV^H)@VQW*+S#zI@7p`qbX4|N74b9R%qm?vOu!oF;3
zT70cJUc)sn_nD8k_m)d)Ao1X>I<<lzuLPHyN-4AcIASPer&EP^@8o3mqD(tv^WACm
zb!)lud|o&E;M-wgyAvn@)C3_dnU1nH+S+Z6I0emT*0ZB}vxX}xE1+s+c6;?@eympa
zwX?VH?}$1gacEhQa?u}IYunusimF&-mCe6Vk@i@aPCIL{&#1umKIH>c4sztoPf4$7
zxwe)~|0mNmOUq@Hs{lz_Ay37;Xp~~14D-<28|!vpb38gpP&-7W7KQ2U!naXsIxT3A
zvKb5^qA>*MIrXoQZjFQZ9MpEv62XJJ@bY#^aLO>kLU}aZ1{yl3atqZ_ebvq5$H-{>
zFh(@_e50YJe*5>%r?>M00*7BJM$67iym|F1v&GAfS=Y|em5PdL^zZ`A=+V7)MrI1n
zhibf*&ouHFtfBbwFMr~DU`@C7?ta9uL{9hnXrj9F?tqDb_l?GDJ7f_|0`|ujyg6Vw
zfaTC3mHUv1Dkah&Nl0}*aO^;;1}rw{&HvWknMd`QzkR%#3?o^xRE)KVp(0VCk~O5U
zCsJ8LiBeH$v7|*BQX{E}eA`hfkyNskXiqA}QqiVJn?yXXoB7Q-&%e((&w2iNzJJWo
zIJ)oeeP5qzdB5M+by19d#>uxNUy6zGYpJTj=T$s?+FM;;U;q5BsixB!3Q(S$7^&N$
z+*;b1o0liQo}z%6rU!2J8D|)hLcU>lf7S{or<8!t>(;JiJ-1iF6o|0avFHQzmoBxi
zi4WX%JNwQZVT*Dbm`WX1f9H|ga&A4`+E|qwlI4gwI$!w%)O=&_-<<K25fk!fH(wf+
z?OF@sMz)&ZzZOfV(EX~VTiDj#-or!B<>U$`{VY4N=wsf=AK*6_-7klsa=3*$IM(o&
zf0;U+wO{^n+_{6y!_al^WFgP^!UeBN&Yaio;oYur*10c=eh5Q8W%O>IJ$+i(L<}bJ
zL+!xa3=b*M$Qce^X_%B8tHQ2CUQ{2vwbz7<j{aAT^6uSp4?Se))}ub&+TMKEuKZpi
zd&324Y-@grz(wE(lyOumR6bj7+<5zi6PtPy*5>Z(>)l$Ns`q=F-5<8py-<68TOAYD
zno#F;pS_sp9vm#l4CIG>GDT*i-?6~R$mq~lsouT8l;wso32jyI6p2iu*Gs-mr^VPz
z?zcM=3lAQf2zEQ=T}0HX)(z5&$0$sY7^)&4!hR1|Xmt|VsB6%6(+W2Wkw3K?X`a;8
z<32tY<Kj+7{_ZV`xOM5c(N3YtZNl1?wA%U}nVRICeoSPI6@|xp=1->1>c}Fiw;v+i
zt5+{`o2$vmgZo}B9We9YSQ&3Hfr-1bu|dXC^#M(3eF3cUe50S*vinNtC(LfSHp->4
zSDIGHh!bH0B@-b1nYKkMfnkcVfk7_0V8yeF!E<=2T4ryF5Q!Rgra7ZS{>twAVo`3*
zXNgIsX=hT?EglW2_c;RMi<esedgq=!d-gB=WwHI4gW=!QpOlnrUg#GcHENdYg`m7J
zFdWLL$RRQ4AvWIq)8)Gssn4&P%SL$^_Mc~9a9n*jY-K`$Wzn10ua}Ls%aVMG?~Q#}
z0<fPG@MY`abp1UpOV3N{9Uq~+&b)hz_`7%S8Y9LD(m0cp%v~XYfgizH60h64s+Gw)
zIy!dD7X(;?P3!*x&ncxy%kL-7n+~0A5xui?wnx8_`5Bru{O-2+OW_=&7l5h>%Xv94
zof6%TPG|NNGZPRw;-(X*65gT5(85;n^t3d&kHtqM?&!8a2fHtvQ!vpw<YE~aS4?u$
zP?e$+t~*m~C;jOfeYBgoToO}!@RhG#nYUcS=Zd_NqN0>m=WpU+t@!!futOOcIx}mp
zk0|LDW=`L7bCcg29`$PZ2=E}}&y%OWxMA-()omx%qBbk8O><!irLUS%KnbpsK4w;n
zFVxl5%}DNK>wEdRqf;X%MF1?$m^m}aY^An#I?acRj4y|d8R7#8ow~QFi}_X{8v#j6
z9UR)|L)84;Nf>|D=$o-|ZElH}n0(5jd3vidG8%n>*3G}%)OPM{CR{e21|O4^oBY`u
z)PDW>?!epG*`BF`6n^Y(DQ;YCVq$P2xBwQ!zVdd>Wbt3zc0o6jn)+X_7&u<<KyQe+
zLE~#(K(@Vi?`@c%o-g^El2UDDrCzsj_CRZiXsuOF64WskJwI!D->=HdoPFqLCgoh%
znk5V|eK%aoPH~@twstpgO7W~s)#h%YEUHhGwLM+I+g7=23X>f2{tI<kvFs$m0`V6x
zXN%kQqQb)Mg_BcLQ)hf${T21^13!gN#~N1~8F{+9JMG`kU$nTic+DHrAt^kVC#`iB
z!Z;GRh%46_sp)C@^HN0_=B9b37k1H#<nHrg*|`IQxX+9p`qa5X!`Ql0!hF1Y_iR-X
z2Mry%AGEzin9g1vn1S+fgRyatiA)2!76_%+23|9rJFnb(>lRTkF<^nKtIo_KkI@<$
zfn7TbZTBeG-*xJ29&Kj%K$zf3KDMI6m{cB7mvFdWe8v`*4e+G{Xmad=?K^hRg>`@w
zNm36WLl`B+Ymph5aP{h{MT@*=yq(S33VlPvaW=!`<&D>^Tdd`(Lx4mA7;aOU7O-|J
zQROB(&YD{;TO+d=s<x$9`-$1?rn-fD04?JSQu&ga?&pwpSpK4{tUO_^ry6P%IR?4l
zr(HHf)(S(3c*;yad6<>80NLKvtLigneuiZyye`t$k8au>Eh5ITh3U$pAxtfL`ts#1
z(#Dh(ZY*-u{3(VKxG!;~cW*(x3%kve-^wo^;`0)Kot}G=2-`@X8CqJ)y{|e5N+uA!
z^v-^I{ZSHhbadbluO}qLT)K4J%d7mRqw35RTGr*0^Y7~~A|y51?~yj@Df)!N*JT;i
z&6_hn#qZ|ZxVUr!j!HG6Vj}hJZ?(G!!iHVmp-C3_t2=k^LjB&BKM6$@BA~rW+vd2C
zXXO-}hW;blvI{i;^A#zzw}ySy1IKGfiM>ZTNBOE4_f}tbne+_LQ#sX*RqF=%>3^>r
z3r4cudk<A=NW>icFx#Yao>8Y#S8P)srL?rAv6^5AcdwB5clM{h6j6<J$sk%;k$rS_
z`8g$c1jGx;>z0_zrrM?#n<n>&r|)n9PoV&(PLGU=;*Gf=bU`Z$+aN49X2sv8O=Avx
z#4M+`6$c2@eerMd_Ek`6RLQjW%MrH-$Kb@|4%9!S0whjFvc5LzZK@jvbp*+%4G2vr
zf_P1k5cbNQ7(O#gKb74xuGH3GNZWSp3T5c<t4#3@&#1$P?{Hu4o)5pa)Kpe}u#lbd
z)4>@7DBpd?nCRGnHA-uX?f=yt`wZDx7kC)6U6V5Rx9ZJ^l<sWslYQ1#UQJO|^+Zu4
z7ow8hGF40KMsu@mes&iUnV1)JhZgLztt%Y8K=Sl^Eup2$`c+-s+LbF6zWjQWRtrn2
z)v&NI`0@~Gf!d7`HO^gU)^pa>ss6_r9U99&;IbQ^Bgvh=W?y8SOGkUV#nrGVE&seV
zaVEx8{x%9S-qA0=N$SbT$-S<rNgOFG#Aw_8l_UY}=!1`sO?mK~J&qjvey6;<aLb0Y
zeypu?_#hdXJDu;5-^qB-yK`W7%_e`Jfr%*89UHR(Xd|+J*WU7GNswEX{~I$o&55~^
zJ$u^g?B~eJD?>-i@8tJh7Ks#|_4*Xrjuy1s7yKvh^Wt8oYOH+^;sU15Xlrftzl>ZZ
z0x@gGZovG6CH?ioV`ICj?n@ntD#$m6F-o^yxZ-Do|JA=<3k)=>?sM>i2PqL%d)V4Z
z8T5T<Z(crs{u?BK651L$lJk_q^)gmN66b{~(3c5O9@BHUrtPS0Ms111AOXm!kN=FK
z^Y!A(k5?8J6;(DiAT?ZN1X<T#?$$hx6%G@*_`dG0X=cIX3d?si*tvQvTj>3MpvRmI
zI`jL_!}O|6yXrm39Fok~!Tf7WLS4Ka38_fF-NnPXlvN3xz-=7^qgeW2U7o3B-sNy9
zDoU1z`SCGD1lh3mtXYB%EF<u1SYGHxi$c`oRp01hx!9}{66ow#8>}@<YT=wu8I$rt
zwbz@giZkPW!lX%C93Pk#R6KlmtTjw4Ubf<@<v>vdh}f24qi?rfs*wL%QV%?}SGwG(
z#}-2w!YXz1E^I?iVJV$TihWDV%g-dp);flH+%;b9@!nTjY-wL%no!{;hDizg4FL2w
zd5^%m;HfT$h51ne;q5<E&5x}hW6m*av8E=~wy)_E)E9NeS#o#7Gm~%U=3csRp^9Qe
z0PbXUzo2YG&aX1-3<%39T_*wYrT3IxYGk&dT^s)<)u~tQHQ47qs0SY$1<s!h<c)vd
zNV<0IT1bcqyFZA;M~)ncd9<ggsp;s&DUatP#7av~&k_4Tg2udmHOpc}>h8^(t8q}!
zz^uh>>7N&()k>=X@P*80VR!&F)iw0#?8QZ}2R`>GCx=MDFp<rx##{~w5oG+y$;lNr
z-`3UP2w%XNnus-=bNi}=Z>Prxafu)xj*MJmViLj-8ieQQOET=_r8lBo;=Tct9-js|
zCk|zN4DEy8BVw(8wqiPcTa+~KL2|6tD{lnUr-LZER3&kWh6ade8mW=l(IMi}f7O*;
zLT7~_an3016820Z2Exw$Gvp4ZrG4R^<xBnL%5%;@=>M7cO*IqfYyOxb<h41v6j9cw
zg78;B7{O=GAQ2*$R8&$rM<dSHLn4vL9!z%cmXmRoN~4fLC@6@CG)~lu`}X;w4M*us
zNlb_$GCj=ARSZT)^r6%hHP6I76LLtqni?CK09{{I#rN+3dR8k<zk2mCYLJQw;}rF!
zaDR;Vop0(og;riNYe0mRbLSqJae`HixqWq7Ko3h$hM<&wiD{#rA=ShCAdC$-eHzrF
zi*M03wrBa^xP~I3IT=6OvX6PM<1{s$(5f<fq`0tfRKFtKZ(rudeb5AEtdttAKTyd~
z82p6^us>0YE;git<bMD+q=3T#a^O*|YfxXk{Ps04$&;Oe_sGsxCjkZ1@ea5lWP02Q
zJcrl33$*KUe=H6sc=?1B4R!D{c!+Jt-&Y46nv^zv$z`NqhG*65NvM*a@5Z}R&c88<
zoO2MAj&L(<TaQLOZ&XA?LcyCQJo~aUI~>Y{9qD7q+scLxepOxF%8ZrX*U<oRr%^U2
zKCJ5+_KFoIPTTXVAQ!6Lk-tC}$rPB3du=jjL@hB?UeWY9J236`SFXya?PNp2p^={&
zC3lM7*)g(+Y{|A}{*!8Pp##4#>`3!uM6Z~X!JFYy+eEc*-hArp?B)2t@#F+7akEi<
zr3lE<&^rF3wF3o#ilIPAR&KER(^qAPfzQDFv(C-0#Q%){%`Y*r1-s8FTb+(dC{))?
zB~5bVTo@lTlPC=GE7ngfSn!RnVs$g}LQIS$0%moQwo8}#zC&K~JUl#jjI!_0gdKnA
z9>KcvdDj#s92r-zK<O_myQ_HQAAkHoP(2m-EbH9Cx_wWm74OtKqJ#xyk<2<56qIAs
z=(EMz`U?g_c6-XL#X{sWcBwO8TBq^O00|%)&6zRre1$=(!Wcem-$2)B^hw_gz+p}#
zrP!gqKn*=qis^9}<F}o(s@P3(_qgUK2d3{A$U}Xsym5zgJzv~V$363wcV=J&rp#dB
zMf*~E9k$)OcdJ>5Ud;EWG)cyr3tH`&b_#@vN?dc~_4!*w{%hh*m01<BR0+zFqFN?9
z=Z_?Rp8)l7TSPxyoizRv6ETzpG!3ssr%btHOl&Mtk9N}WJ1?)G(@I9>kRJl&@SMCq
z|NQ7+*4fg!vL$Y*UKw%<tI<n!U|vyb#;vwqH63mP5U6q5rRa@pim$?{6_?Hb)8AjJ
zZ(q}dO+B_Xt%DC1h8EDHn8M=RVD_;>oHo?g|DyhuL=&AbiJ0cVXKzVsP}N!Hw>(i}
zj*#-q$O+P@WO3D~RF1kNt8L(c(s49k<jCKTrVsT^PTqqI#@zh;J@ZtNs2YxuBvN3G
zZ>UJ<^$%Hou*N@C2ZKc~!OWRB%1)Uh#za*nU>Shcs2&@inVqflV~B!;xCm}x5<6>j
zbWAsSw@S&X6w2;I^o5vwjnb@)J|u|3E>)7Eqq_E77VK}puwfJ5*hwv&DPG^ufEL__
zk}$+(j)ul{5(HTOLnFW(W##3y%rnLE4=}BMFQ|r9(3*{I-R(x{;m)zq(Px5!7Fi?D
zbF0phZWwdKBMC$Wtlu{7kJ<g*j~uxRi({3%AORMIeiS@lxg#<EQ&$(yyy5LzskXw{
z)f3&=LHIzwGMhJzA{?X4hjoe=8Jdg!gOY9W%M}|obW@c*c|Z5T^khf|(iClErO%$N
zPf<I5`EnEDBJync<g{Vypu{j9A$m~Q^iu6}J|;$3ElZ&SyOa*r6|(c}C&|SKP6=qW
z_4N}b`@?L!CA|WSDh}DIxvxcW^CO((s<P!0nX=NtXod58S?KZp-8;kQHP)iAFui2I
zGlgY5Ep(BGM*+M=$Y+w=!SdSE(9zu7%&sAU2vktu5;itA-cox`p}3#QObdb|yfBq3
zfrvPQ9d&kgp4>g2*ydf}H;OR0fKIIqX--qdjjJKM2DD!#FI_(|wZCNL`}gl_Yd_-2
zfq0P0wUAZP^#Vzcxfi4V)6gKS|H^Qx*_z3&{RHV3ZeXaatRra#i&YOFmXMZ~CNfcj
z5tGg?_Oai)cU;oAu{y@2?ZWhf_17Ay%U`!Aaee_+m@vf1wo__j;^X2n;8TTA&0PUk
z;h53eUyhD0<g#?-pe+v$4$joCd;csoUUjjZgM)*;y^ojI`D=U`lx5uJhbEdPTI)-c
zvKAaVe=96(wDHrUE`j&nzQZx0uSlf3mPUoYskzvUi~U`>vW#Z}q4D@JTFg>pC!5A3
z37m7?c<1HA8aC(+aB=}(W`{Z->P|v%byXFShkK{z2Ok&~wvq2oRxz<zSAV1Gq#E9J
zYwVdeN1r~L3dqT}l5B$aH*q-ejHVgRL6k0Jyen3Wfs}wI5{XI(ru-S(DBHVx>fE_^
zX%U4-XvK5Vz$7euZmahvu3fvf7c!8zXkucb7<y5GCS9+Dh&9GRW-`>hEqAr1Pw$|}
zA?PD4MkN3w-PP4ak}0pY#Y<h>#3S;rgA`3PTC^hVdG+cQNE|kDGAS20g}Vxuzm<^i
z0q=trjbIrP5CG49K0aR1lwfzu-#!6|J^8!dC>h$)c4KO?s-`9|F4WD)xP|J{aPDHW
zB4I@-mFJ^^0ynBhAmHMX61t@L(3@ZS_%V>25VU8_La^Rr21i!mt;eD-CZ%O%QKrK)
zzQvI9gJ<a|DpFIbVMwvD53%RmTwzkoD)lrT<|<v8bC=T7yKyz#YCaNbP%4_`pm?@z
zqmFsG$ZhL$`WZ;ApwE#>9>Xu=4+;u+?8Q%>TuV%3p<C3=n>RC~+%I07n4GzJc$sVj
z8M9YH?>bJoq?Hd9MS^b7(OWa(*pSREg(Md#x*7|eVF1`D+i3;D{>TYYW#CX?`W*J`
zA%}UvPM%5$<8U&XDOQ(0Y$(=qGt##<=<lAK+|H7yZ!FM2F;G{xy=5fmuRc@<c@%CP
z>nkc+UiJ2V=xlO6W`$`~rn{F~3>~baR78C&bicBj7I7NAEU;TNmQcW{O`a@_>=9b7
zzODVk+dEA46Yh}oH`7Q{4DRsty>E_xZiA_;;R=!&D}E22-b6a;s}<_efgm+jy_t81
zbg=YlwYa5F?sJA;H+}iyoROBMnu+j9E}|j(xctgnpBzRl2cJRopnc8|FUdcXc{EUb
zF67gV8(QhbsYj2_C#w#SEh>|OivXnwPJjOB5m9T0@ATuUS=?ccz6Hrv%$xD&gH{?D
zDXI3R>Ebp#;@SP0($PG(wpa&AJu*8QJExU8Wf+p#ZZYXO6&`KI5RDxJB7Z-Z-&y29
zixgWd!u&ix2({Qg9nKATtwxJJW01ywA89>GwDGP7Yp*;r-oTzS`qt(`rE>%WzaFKG
z1DtYO*k{AfoX1U<)}#{4vyO-TlxZaB_>U`_byyy$sV641>7to#ICJd97)}UFEJHGQ
z6e`84Yb5`{RK>)$)FRdTwo*}Mbnn6kR10-(sW)$muEq4BI+fqIj{y&e(00Ur5X+Rc
z5q5!gwTo=DT)%VzA)mFHWfn#@f`F*&G<ieC5DQCVfDjbjdM~SLYduS{#6}Ge=rl30
z#9@=o4_?2n;j#p91Y=w2TRU-HKM{SVv^gy{$r@M=u%K6}Gp<RfvsOP97qhaUskW^k
zsNkILk{g98k9|m@ivky%o3hy>fXQ#$@snUF;WqVU@#jcjSNir8Ti1Wml-^uynG(q%
z$2KZn!=7x5KD23-Co@K1ZO*J0Z5K8<D5}WxvW9EnSho$Xq<Mgc|I&S~#KHJOoJWEN
z2Y<=VMT-~X4gBp^Z;4ju+P9s04hx=MX+AU$|DWpk-tgreuh^f&n)SZ-6`wbxv=-t>
z<cRx5$hy3He;qn(n5s$270c|BIXRS4+6S)1EEE?r6PdYnC&zd;w*zv8*A$}l-)?0!
zy~$~)=ZO>j><YwZPZ#7DjHz29><Hy+s<)5l$8{sJPRY?M)@LK5Pqsf%S67c#mnUb7
z9NtGi2exH@>7YT)7C(udw=A!yz}xqiJhW+{vOb$35R3hiX2q|_>*HD>AvQZ{tA)kU
zRV72VfdNs!?QB3r=G=5;-R30@Jg~?SO#c<L-9GjPbL0N=$t>$FT=w6O>5p6d=jY}*
z-ygw0KOL0)@z($OF-;I{|ND7$bTR{x|NW>6*#5sCr|;z5e}2rA`u|J*?>yKgVKKrh
zX`euJLLMksrPwGR^v(HAr1tbY`Od=94*uLu^(>gfV+rAnngtNO`9zKYcAxF?8=c+;
z2G)Y{Fi5OS>Yw?`Ki=N|{7nXitev#0VG37JrVa-6{eC&&w?3W<EC7Bi{14mN|J~w_
krT)0?_y7K1m}8gQdzpa>7j(QY2)D3kzTv#MIp&A|3!HufMF0Q*

literal 270993
zcmeFac|4X~+ctbQDixKfGBqhBQ)SB3gd#&BGa+QooVlx#k|7cmA(F~e#>^o@h-g6O
znL?(_!@DhA&+qrW|2^OLJl}t}`|7@Qa-PR=thKLw-?nYv*1GO0DadT1-a$>FP&S=C
zbNT{>LRCbe&>dU160dAkyn(+^)*8skoTe<1|3&4+J)lr_QqG<}p>i#Ju*Lb3%C`l{
z`SrKne5CQYz9CxaJ?q;;)YR*I<W?|fzkWxtlB(7Gnr$7mrj|NU$>@NOdSPtmjTFv<
zQBjOrR4QIaC2ZYmv3}RgLSWo@mCzBV!;R~9CD-c-^+mY$+S#7+{Vif=b-~fmaUJf0
zLb(&GeU@qYI|>yK|FPxoR&R}2MZP({^MCMyEYHA7%JrLNced=2_T6^APL)FWaKv?T
z@P&L>h&=-(c%OkiC92$_Eq@^{(&OO3K?<d_F~`aJY$)aU0Yl}#x3q4rG!?m#>&NlS
z<m*2#{J$EHa?3J**2cFJ{k1XA<+!w(G37L?$xudZ^J5b~W=yYJ;IR<Ivk6hT`@cWZ
zA31a|6~XJdHPhaWRdD6l57>U#BJ=f?@_qpUhfhxg*xg17r$@Wgl67^;3!5@5U3(s?
zWti6|ow&JH!2EmjJDaYGhg@xL<DWbie(5m_Sse<Rr9GT(UT?MI+4&ghKR>_4>*P4b
z%04`JG2zkU$64-klWA#bc&FRKpZX0<JmG?tKm1umWwU=SDK5<ZA(!ucY?5L2?X^`)
zHa;dv{cUFmx0axFYwnfW=*LG)6e1<uD?_-wJfAUgYfk*C|DBLpZC3aE?8Aeq_^62w
zbc((#A}J$#R#AcvH&Q54-Gx-_hXy1qqpl1#yxr4F>m0x)ZZKKUeq)2d<nQ`qJx==9
zdIgKWk_*@-XiaJ&#A(Gwj)xm`(k_Oz%^MFlXPvld-Inj({{EkyiZveV{ya8`xa-_8
zTE@<v>WR@YoQ&ktKY2M*a=wH11V_)S3$e0m^0(Zrv{n^$9NdOK`rfRJ;wo_pPYrZ!
ze<fy?Chy|lFv2j_GAV=!a`y+}*%Th}?b@?vk4L(w(=dymg?dZ2LszcLd}O>21DC|y
zVCvlDP^H5pE_`;7c_URChg`_sk>6~TMNdJ86_o1+U$Y%>_lw!M`4g-d<>`@PT1mWZ
z&9+>-zG@7yMSYTvk~c<F9Iw*tSMYuC{K1{qs`0u`Ee>*_JmOQ$c6Lvnh`UY(amYpU
z4zv|`&hyvc)g1z6r!xcCB_*=?4;&b%7X7WAcI7<Af=lJ)X3y?}s;`@!`<~(hEb-cW
z1ixy6`e<TitIy2Za~CdLSh03Bg>z`0AJJG6s~UeX)%eq~iO;9T2Su*7zu&-{%cpm9
zX)?3mDdudKH%pL74OepR*i2t!O?`V~W8=(EJ!ynsUwXZ+`M1}qNs|aADJO}=>8`Ha
z;w{u(E>pv-E$@sfLo(YuMAZzlY<u=8Mu=H7yy@<Cv#50Lf972sAwCx*$8&Z*h{bcR
zsV&=a=t)~V7K{N|Fq}L>t=A0G&Ax3NobP_1*lRxGu`sKeq%BfdU9aP$6V5PN`=S5D
zYn4}HHJ*z{>Z1pl^xXIb1O&3}!q*HS(6alZ??pH1dCc_q`ub}772n?U>r0&Cir4aI
z&(8EoE;)DJ-M`P{?9^}z=4Cc2^&%3m(SvQQxcs_LVM5klUdWI7Ej8QKNLh)p>21GT
zn`zZDTPMfcV*f>+^_abI=bb%fUtb1>hihZq!~{rWyG{)ominGmR1C;}jfc(>G-<=q
zz|xC$og8iv?aknRrqrK1UVU$dQEtMYr|+p#E2~rU7Q6c&AMAJP|NKnaZ^uI}HHn$-
z2QmR{tnOWSvcpCd!5>23_Wc@Sq+CCI(w7GF#Kp~RInrKO#$tC47k(mWIXhLbB>vK*
zulnh@==2i-^Q64VAGbF#UrbPct6zM(Ue|R9FB(jW4ZbPnDVQHy?ft2uqQaUZwlddw
z+-0KQ2J@in@J%&FCSVc6{5?@C6U!g}47KLXd9!<5`<*vsl{;Fh@bs8e@+R7n9|1kq
zo^x6gLWxKh1xpLR4H-AvIm|>$-NRC+Zwv|!o@g)(C}DZ#xwNRJZBw#8(k%s_hR7w!
zM2HNZPz!xas_B;(CyYLfc6hVv+J5BNwBzVYB#~Mf_LeNWzPJb%vtQp6XIQBy?K^iO
zO<p%hvuw(k9BQIv7PLUR8W<R;4CO883DLR{G@-00E}(alS0m*)LYQvfxoJe`k;8{Q
z7H9g_)9nj){9)SWF=tt|=qxdD7{$+h=JnoCgeu2E%_Dw8gGrJ3k>Upijj1Mz`ntzX
z3^ipgVL+eu9qs!RHuYW0S}HTrx%Y9F=EY=P*1R}p&URN_jQa+jcRRDiT@JWl*bNs|
z&V>t)dNJwVv0^`BTvhe>NKDi3>|aU!rNWEzeUeT;zYbuHXcc(4OG}s1O3w44CX4>g
z9e?J{R#7hus6s`GK7CBiaQ3e->->-3gl)UaA3bFjFbhsDB+=Rbtq?O;d40tygGp;-
z9z^%n?K4q&ei83_Dnn(21hJ?bhkj)3h;*N9?2p+0)V^*YE;1s1bbh=hSu@=%sWi`Z
zDj%PawIj#rH(Tm!4Di91?PnkClM-6&6)F&!?T@+1bC*pYVYuFLEroN#hkgo0Njt-;
zWk+7VFg-Q5R>q;;i@Q$DT`8ORrl!ZO{$^+ID@+uRPWI7kaS@R$hhN|CZq${Yl&1AM
z85eF_fy{KowC=gr2BsvfOdhOC<WA8dyh=A-)W6OBdx8ePB-Q4HB<YC=zGq5Tex#Xd
zWmtUtw#_aw5{r*lH&^(IG-AusFk}BaBPB=2N%yIi&@+mZ>nW#}SAkJ))nhC;KdYMR
zY6W?DKdS)UXDnraH9PtS5!VJ8LRQVffrYfQ+w$CI#>gT-F3q3)8Nch)Z3HDl+*3zQ
zWCU~E$g`7Pr=?#abnLrco{x!U=FJ&QDUZ8*F;VjmhA&{4ahx@DpZt^4Lf_-nctsnU
zv+N8Rn{kb(tkX{<Jv}~8&Qd5wAtZ75R3bUwP{6_wCE57EXrbqHhqtiF&7jZy0$+`k
z5V&!QkrGzT?^d^|+<9eC;#2E~RNtIosg+bpay@RWgO@jVqzKiz?`JBO@4%NhnS#^#
ztr~ZeQEKnnYrVUwe?D(F<^74j<G|t4HTLFF{MT_K1fos#@Q-vZW7qDK{*|11`EL3=
zlE=?IJ{*hmU&23D{Z!PD=U_l{eJt*wmQ~kfAhxQiO4|=*{!r16pI=`+IFpHl#n^Fg
zC546MZvfPN`EpZ%VR3uJX#D{yaoMBX+}weM4<0-)%mCC753HoUjKIQ0%}4ztkq!I9
zTz|fb11>m7PmKW`LiY5_H!HfezSggHL(_&mg`do}&aYoV;Z$D!_=BttU(c@*W;Xf9
z?RmeBjO|g}3HeMlJ$HwyNi6sl&!q)KlPm_V@*!8~gk*r^;<fxzJrsqU-}%fftp%P-
zN}LlpZ3{`L=~)WPak)qT%a<=*p1o1R^oyI=fQJLf+{nfrsTST--H2sUVxgDk;?Ny>
z;#bej45i<gP=jx*?~VZqv^<K4@F?AGA~xAzh;$|^CwI@S$jontz}Vc@8i_dz`z{7G
z?en;LYo4o2q)!&TxJ=UAL&Jf(*t0WCb_`+^%H7Q<BgbFalVPxGIoA6M*JJgZAFUmF
z3|z&wa|S4&l%?Yqox&Zf1dUW<q@!`IzI^vNgN(T6&%2#o7KwJ>KSd+beZPMaDXC1G
zmW|p=L2_YYKEJ|u`;q*);T&y06x*E9Qsx|oUm@}CKtm7qE7I&R!-}Z&BbDoO)CrIH
zm6WLW%Y&@x$10kEV9wfsA}wHC_2|(86OZ<zU$rqvrFR-rk)|2AF8W%WRZw6Pb+pKH
z`7@@~4X|gB!R|Wvc82x!NVJg3$1Q10ZYT*GHph6MB#==`h}8L7Yqs2(-Y}cV<DRGz
z(ae}6CxDW@>Su9_RJ3gK?^n_8ZD6Ndzp`A=6BUFzZt)h(rvR*x%Hl=MKqEOn5}%wu
zdyhB&4u+GST6}4KJWG?T)JNiqVxu3X+B^#o3CMU@Am7(8X^$YYO#-k|q}C3Vl8P1K
zF*|S~;3W4xtxT&=RaL%L)<A9(R%4_>hC6hQ%7+W{yTp1XIqkFc1x~6KMv7L?y=G%S
z$DP@H_{!&~X#oV@AkZSh;?Dy`34%;|RS#1r{}?1^0d~uJmHIK4HEM{~;omW*I)Lpg
z8dGAi9%~l>9)*E=i`2HIX;hfTJ)K1D=`+jy`8uAgFR?PWE<d5Io`1@P7*bMF7!Rp$
z&a;^ZlsatJTQ&E)U<p@iL@L=L!(ua9!qkQ&S9@=PF3qsaAC;@wvB`2Y2lbc-*g+qF
zYpT`tcbe%J+w}b$9O4TTU)*OWoIAXjdOw9_BFs=>zhaDVTgcWHUtWCDFnNk;h`+%=
zOF7Q7JOOl8`7^!vy>2tC{BU8LP5_YeZpm|vCXq(Q#%qV3q%7s3O7KS(xD386W4t!x
zd@e@XFZ<7Ttp`CtofD6RZ4546yf|;u>M=KDf?9*}=qK?F^b!D({<SPP%_Tx3IUkn#
zvof>ZquP!vOhEhi*O885Wo2bLmOft38Mzzc#TgcjW0OtReN(SH$B|`uQC|wHU%h&@
zG+SRlJ$aJrV#1xQROjcbCC^@+8_Fy|IIQ7XMY&!?x4gV2aKk$GwNfW&M|_W>%&b_o
zcJ)&o5G#jz9VWZ7kuw?^8aEMt$Lz1<Z`kCumX<C2%+om;R9_V7ch2LzVy-V<yohie
z&ccN_N6ZQ4wy8!O#Q{|)UJg%4NT}q=8T1z&E-x?VyvmQ_{Af9e#BV=h9FE`Hz<6lg
z&-)=En4wG^_dnmTH0~X#M(uX4m10`J1D#+6{;EYfMGj|3s794`8v604jNS84Lywfz
z%vevNX8J6G7lV4lq#A+4*ykE{hgM>GXLZHY>d@%E^6KULAxnVYuyhc4y)Zt*vT5>X
ze6lVeE$Dz^XYv|MELFu?FX#p<+@@NbEPtelbom}3Fb@GCfY=781l`jH_QuW4RB$C*
zG{55yv_T-{b@_`XVIlTKdM@7GAE6F-i=aRq2gkfJ*8LG!xeelg{?Bkp&qbRuR;Pj3
zP{-e`3ts`Z03{Jv7jH1XNY=^8IOR?ARMd&Ui>6HLwx|;~0Sql$bDhBy5$2B%8_7;|
zDQ?&)D(d?6<@r4_{=O1~GOir_l7eN#Y@Iubi5Nf@&T$+fnQr#auX=pC%b%YpEkJ`N
zW@bOJ-qn({XFdtFRpC+}gE*Xr^Jpl?BZ5*=Q{}SP0ws=qK1C}IB*&q%hQ)&CUHc8-
zOR3DM-)&2n@`j~9_j$1vkRfzXOYG7W!-e1C**Pw3eRKIEp?81v5yI8LJ4{Up8o6s#
zQ}53M)_sVev*R|V-(G*?FJpNSZ-b9ucb_<aHW;b?#GP8a$EOlYm#IbAW#USX`ASOg
zZnE62@4kp}$8b4xez+^{GEweeh9WJPPDSC=36jf3X1RXz19SrGcOAE<-EY&f=VSC2
zlxe|-i_FOT7cMuZ7<tZA>d9t9*cy`|$mfmZ^A|6`CtSf7Idq^d%<LgD0;eM|?hsad
z`nA!{AfDBf>u#w+xNVltWUW4b{yZq@#*GFPYb;hWQ?Hc4c8@vyZq4J+*+QYL*|q#3
zJHWV*uv*oLbfomb;wMqhkOxU2!Pn+9<PPxlpBI*g;eU5LmP4YwT@4FS%PK4OIU&Nv
zym=FK0E2P}02&1ZRa663!fy}<BHvxV**HHlUeZsYrwb$D^gomj|2HJa|KC1C$!8P0
zgxG*5a^S_fo?sDVmb0@3%QC0bin%6%dTqKH*C^;G;Eh}Z(Ud4oj!=87%n~D@GjOeH
zpsnvoIyp!SkhYFF{A6+Ie+F%c@Gz(*xy3Z13D3=MT_6N8v&B*CvK-Qvb@srEpx55J
zNkK-Q_L!)ZxyQt-`_@1TIcc~hCtl^1ZTT+mqkz7^a;Vh#BSjmKZ^PTK<NxRTAqxP`
z{LY^v86VXRi3PMk!0anB^NIM2iNKr<sU|hZmkhshJ2`JE(=fCo02VBQnV}%}x@93z
z7T%;H<gY|>6{SHhWd+5j{3fl$w(;M7sQbVBWP58OJpoxiEOEUA(-CUET8sobWoJpB
zh|=Tr+tkd=>};rVh8V$BiyE7l1rHh51n!Y3oi|ZO4MOQIalQ+=Kw7Uynkf>64>c$X
z;GWPIPp@9#zHmE69_?)=??DqbvL2gw{y;(YavBr`2_#$4@Hxx;MkCvb#uC0?gM)*B
za>wdb2@73(te}n*MfX^NlU_6B@<)%PP1ozHU+{pr7?B}vwv_`+L%RPW=IX!wNG~)4
z<fEU9uej-{<)O9BKb`_C9n7rfK1o$b&$)H`_U!=@vsW%x0zK6M_PcEg3JD1@-z@vR
zE>;eeV|MmevdQJkd+R()eYXQJvZCrKc>@f!*wu*Zc`i8Yj=`%#&>1aSvi|_c(Ivk7
zT+tqU3OBCAX%CRmXO^|JIJZG}H-+UA65jhlL)<^W()?W85}TFjkF)_y=WGd+9x0*X
z;_gE~(lK{z!dEZhJG*0p%~`1+RzzTxdwH2N5fEsuD?m~@L)-3?0H(bBw=m=(jdU~B
z+HWO`A&(z7;wcm*W}d%Xs|_@Ybc|p)ud<3U>#%T-(#ljK1QB_%@i5*d)>gOndL0pf
zSj1+heoLazhj(>%cRxGj&7+ah4O#$j{|#y3Nl;K7SZpfT3NX^b;@odSXTu{1AX>ZW
zN2Gb(v$V>J3LlT){y+74de@&FK74o>Tq>H=XX}H94;`RNF?L)hwPeL2zt`V8@n7WN
zbDJLNOjt*mWJfK$dViawhugHN@yCxJp$As9IJN13L_d1^>s!2#^H?{KhDbA+BX*Bz
zeNTGo*fs)MEeGtzK0*u{`?!XA=%JeK3{*R1dWZ9u0XJ-beDgqUB*7`LMkfawp+r6j
zg?2abMKPe!v+%|m=S6BuOdLiu|5D)%gl%kG9Hfn(c*;b*0*|U{0E^~yb2a_AKz5_6
z70;q0{$Q%K5#r?`So1UQ*8i1RuJWA$t80GfP=0?G1Vft$HPE})sD>hCh{w4u_9b=g
zKZS)oKn>;JAWVhQUrR$^Lz;=d%5eTDa&djUS~7GENFPWC?Bd8D{tyH412t%?FJt_(
zu~?xU5xi8II|+r%`tuuHX%v7t1Gg@@G*g32k%ToSD=XVo>PJPxG99@zABQzx8#)MP
z2~9#u@a-k?6b4~L>&b>@(-`#dp@7Hy^zI}!itG)5QQ>W5NkbemuO2-Bk_?2l*H-ny
zg*u2@M8+TQ0knE++f&hb@Ga+M6BEyoLMjjj)7IP>ETId?K6C(=ik`Dyuz=*U(~4x}
z<rfg@tq=kscL1uzob;l0`1w@?laid(`$SFG*$8BFX2%d#d3MKbX2RwmKdXS~zw7eP
zS)Bbv*A@+!o31FiV4tj;_vEPQ1&lwkn*j7FtUf19Gr0Z3wpR=vbt<c?BP85s-~Y2p
zQKTmrYIrX65C9-?kYhqqkZ*f4pP&2zdNeWc#ZF)nNQuv|`VoIj`Kk|CH>71Sh+BWK
zPaauc6#)o|oirDUHt-_I_9IVgmyZhl?KU6)d*2k(AYWeqq2YnTGTK#UVr;xmxQO6i
zBp+OP2>(h%9|Y3^B1pm#T@<4G&x=}mn$<>^m?!5!OpIIO0C8m=)NT8?3bL!cAxaK_
zohZhtZE-S_KbiC@kad{Gzor;fZsp8+^KYJc6Cv)}YxWq!f?nVXpVik_$q7=9Z*j)V
z7%t}3;z>I_NRl}=Z!5A?7V)l<w?NEo2!J|Tbbi>Wpi;!O?(F^&cC(7$10ByWL}BI1
zK;iP;)~Q7LcKi57zqn`!Wd_7v7&O2@_hERr8?<u5?HDdzP(Jfna&a_ZI^abD*4P(C
z$)|yVTih*^e%ImVlizjvMA;-Eo`l<C5%7d}LpM5m_N<N1z_<8%(DtoIPi|>Ve**z?
zBv@JDuJQ$Rx<YyNBFZzf&z$C!vTk^zKQAZk&$2ykkp=S(mAS$JHE2@|AzP)MSy17)
z16hr;?FSA!4L1vS*@J7$z6RRwt9pqb;rF&|Oz8pA?^2saMP6-`pieqb-GMStp}J_(
zmyC>z9&#}mihqX?LM?U)mx!if5xaJ^_M$FXERAcP{sEGY_Vn@qm!6C3MJ6d-GGWL7
z!e3C{N}<{mU%TS^1#6UI$bKJ1l2Eb{rDG1xAdW?4Pz5-PihMppWJdBFjQaBxiVcY~
zRW+S-aD@eJ^$tJ~xA)gqj54*Nu>1jkjYO+!{u%C%v|>gWXKIo4T_JF<dD`+u{}hm%
zadr4(R;P|`u;5mxJbuhVrM&x8PCE@{^+o(b^v)+1_f<Xqf?MUypWfZu_*a((1b<}G
z=<EVf#Wvj@Eip=0w`@Pb?WDY{;pL*I_BshAcep;e4Oxyf-NZy|!bMU+#ZKMcc-ONm
zW)q7rdN}`_Xj13Z$xg%T$Zq$X1^4Xf0%<2AizNO`C#PUpQn6iYM}O`I?l`}K%L?hO
z8dLX8J>P8x<R$&ms(l<BCCxm!$R%r--;_TrJm5`z>=%l4nnmMbHNy$sO%zTCRLtYA
zrjX!o1vDWIqsQi^L!Nl#!I^@E$tGkxMk5=kZd`l;^#<8e=+gV^Gvj?rsM~-=$fvmB
zOPCwG*>e$M&dAOjS|%Qe+F4`Zs-HwihY<^bh)$5Wez)WZp$$^v!vN`vXhn3sA!|47
zn84*hb(&p%C64#iD9X#Xfbc+)9Y%T@aT<U$vozfm&=v3nAhM&_+Y@{rOJ$E%1<QYV
z0nR8Y0-VfsmfVYU`T1&3fY{-<Vypyo+m^(B9hlb%gJuE}8vY<v?WNK)L<M2V?`#aA
zEFkn=C`Pii${LOLegfNTNHLO&P-SS9VRsLX=cw`e#{XyF%l_Qh3;rl{)*xPMcn)!K
z#XJ3O1-O)-4?>Tm1*{xb^Xv>28!O*~MCH?Ff^^F7<O-ELADYrujWBjW&&7Gjb0Otx
zdHK<-_zm)W*XlR4Lj2|iu|a^p80Ihm^NZ~(?kQ~#w|#jv)ql~8bR2faZ=}e|u1~&A
zMWNZvk^V}qd$yB`-R%u!&*`MZx%Qj1wokc(SkK#@mN4b6(GlLANwwbVBw!#4oPvEO
zF5s=)8yp-Acq);tcvgOwBEe(TN<9@1=g`c&nT_)709XrPeplk*uN)Putw%Xjr$gLP
zL_fGa1oA{voM@#di~zD10J}E?(51HT&NWA;c!M>wEwpl8?RR{}s!+gBUfbYho|LsO
z`_$ztuEm>#R)69-!5_&^o?}|AWTQW6uaxg!)wd1xr`v1h$5#&HpP^A;wbHcP?36=S
ztG-q-MOw~UOY#S`;14M#o#4nEl?g{Qi=jF|T}y*Jo1Kf{A~CUsO{|wnKIp*%)C-kc
z%H_#GK7+fT{5eu1_TX0$a*~X7qY4Z7@bS^NJ6I{4e~92EA<_HDkecr#YIt!ObXT4m
zPkahdRBI7+{gZ%TqTw2sFJC_0$3&GfV#tPN;q+a@WQ{{SgiK`L7J#?u7!D2&BDo=%
z`A9f?5~#m5#ThARsaeHQ<IU>WfD&Gd&RWzc-Xn-+;MF&C0iN@FJj9<BwWK4mj+!*x
zr)D^qi;y}&yDAqYxVYx)Q%GW{g<}b?`SS5ff9XozXL-;L?JVytmG22aI*Fm2B8Be5
zoh`LP$iJTGl#xFZ6BDQ>;#gz+Q%Bj6Svv#PZt^<0YVF49-Y{!|%1~*Ut={d*E=DD@
zwcMJHjx3%+dddRT5xwCQq<b`5302sBc<<~U0qeX!7Xu}42q;3F(z_-a1zkrFTI&<@
zBn>n&lB`IpC|tyj0H_;zvS{-_#G!BT0CYmLCHayeqdE!UZN@FAjoRFhmmvMC2C8nP
zaGU@Z4t5D_dh@~ag3*2Ok=p^DGd-cZU%q<D>f9Z;ii&3Av8H6bf~)=_5D4*P3#w}R
zUaP)9TYo+h<ABA2>UP4UxF$$mQE>?=iF$G?9zG9M)y{vs`D_87lO-4+@U3k1C>n2_
zl6>PptajDIoRFp5$3F>;KRNv@K{xN1nw+hzZIw<0-FaYmkTsbM3vh*3aS;dKD-k3#
zq38i-1F{AmLjiD(7zanJO;@Ss;&@~mGF3m-{v^%xgQHIyp()iu>tjK?0MrL8V-%2w
z7gAQg6_Bt|cPLyvkZ_!mBp~q4Lu8zr7_e*5_f~zW)Io43+(xbB1HF%uG`Sva_jxq$
z-G|ca^QSxSLHJ?3b|16lfpwO(1H;QU&h5QbZEYM&#ZrQy$DO^iYE+b9A{0k`m6Ma(
zz|M5Y&o9>u(x#M>^7))jA1<tqo7Ev)YQt!dx>wt1YHH4xoo7?parDZ4J3gePNRQc9
z_hWP%8;xwLrL9_&A06MH`<nPbfa?vzBfPTAp>=NL^(U#UGgJkv4a=yTMMmP_&NBAJ
z5$_R`^Q86d`BF@(?FIBdskpkk%tx;%O_%K~LnUIE+45j@9@u)VAIXy4Ci1rS!_6~3
zTuFC^!`qcNV+pe5xt^<v%mbto8LK3=gY23Qcj;n0&7ay;iv%`N^-N>2(}uIw*|{O!
zflauoM1{Z{HcGPYI&|QG>HPGl46B1m$7?P|0>9^Lj`hx&l`NdtPT{m#9+-O@pCN^4
zT;H-=DDb6jp39c6DqnwOTC*3N!qXI$)v}jLUU~{$S%iupa|8qt)Z8o+0_8qu>=&UC
zpZGjtqu*IxZUc}{7iXeBtSP)nQ9(hXCs=tb;UEiRvijTIHPv2QW|CC>AGto?n*?B#
z#TE~#nX&b#<Ox!=J(uR%`X?XRI4ikM#3&p7#HyHH|5m?N-@Sin@sVZ2n+|jqYQ8-K
zp9uUteA;Kev7Bg~{ZnUTo54XG0gm)QY>-Jm{!S-g4t|gwEfQdOcpe5Rhmz=x3L?QZ
z8fEiOmEL@MLTP`$<=wHhl&G|qU`f;0fFssy+cowYO+Tg+@>1M20d+D~3VFx`20U@)
z6N8Q8;B0ztE}h_{pZ)E*)Pz7o2igI7A0ULZUBNsKs=T~EePuM0!T+3N^wjq|6GNG8
zBvnCHp9W<m3mGEKBiC2nV?JFQ?CMF#hTnudU0#uzntGGQmVPSfbS*PFX3yzutAMRM
zkjh9E$pkd)1uj#ZKTpNalr3oXm9t-FxA8S{dnm9{GTm7)5(s9$#VytPK~w1Hf)*|9
zq@HDKG||g#!}2Cau3++qIp{L_?k+bLriPlFD!IwW?wkF3f7huZ*9G=+0_WG)95V@c
zrQNyoea-b%k>3dVVR{_QV3F!yE>OZTb}Lh?C#%o3dvX|fyhK|e@Bk=7lXXE0O4L<3
zs7JvEl-JU*GysVpcR^QohejM|bD4!dHSCeCRE<8DS`y)R(4n1|7N!)@VF5ORe!Z#U
z^QpHX#xe)1bncxio{}n`jU18S&MH0BKy`R4(}KLh$nb~rKyK)BpUxNrPIBqGCyZ{(
z)|PF3ZMG=UG0dqcDZAOzHa`c}to3uw4^`S*!{aaVsdF6sr)S<5Fng%stsyQ)tOw}!
zvyOjCgfP6PP+k%>zz)e-nyLcC0O{yXE1!wyL84m~Cm1HFC!^6&8@z$Sae6sBqo`4{
zGhLE=fSQ2r!@fRYzzN&5RWcEB7jN2jtRnq3@O%NN8N_U)b~*=g7YbD}!d3ps(YsmW
z4z)zrX-YF4M}~%52i>ReROAtxl)04w8hfLK#(09-^FgIMhFa|Af9Wj&3no)b+6Las
z(wB-bZ@71QKGV`i5mqNq!~pTh2GlL&-KB6LYu&0($-37%yu`eJp#Ki3rwuDrdkC5a
z>Uj<z>%AA9C%OJgav-$~E$Vw~0@q&OO8k+sc{HrYZUunlGvy7Oc7ua!6K|X)Ds?v_
z{|q4uEA^Q5BEr6%9_ArLdu4&rsQj*cct+Z>)^NEdQo8N?KNA;;ZR{k9HKC;7M~Tza
z<yBSCRJe-c-7)tf_$8t&0{zzOd7@I$CNKJDQ8}H%7<5Dy6M!lz>El{xi&SQp4q(Hv
z*tCAV3+V5jf+}D}-~zf_!QP)^!W-TjdV~fA1^HMj#_gVYRJmfzs*$vv?C63(cMdIj
z5h)J#we5Muu8<rS;l0p<a3!UxAGkaS2H#1$*vfbD+X_nTJ?Iadf8afG0lMmCiF@%v
zns@e|HIPL|7ZUCM>{HirZi_Lk+n=)MbYb`Q7U^Lw(w);bp}lorZ!}u<tfGztv&6}V
zWxy(hCwvP0Argmfj&)W~$V#y<yeCMPHrw8`dK5a<68@0pDC#J25u*RHCv)(OTTOUT
zb2_Q0eU}i}{GpB^!Tlzt#>U2_=RS?aBi*#*I0aZd<~NkS!Msq`e8mXAb>gf`mBTRU
zK96!A4AAHh=?j05Jar}l9v_2)yak#W7Ta5<LA*aWpXxy#wXRQU0fVqjuTRu+LiA@@
zEv$Hn{HPM8;k`0K^(rJFCRJ66((Yx&@DdnTM;EBY{`}`Ih`0`cYQW<h*p#B@Zb^QN
z=qqN1ED_%RtG6Egt}drX@Q+NtgY=M9&9)AxbF8P5n1zrHAuc;9R6?d4pO>I-ARNN(
zeo)XDnr|g{mFbok8G4hR;+93Ikf|_IXGR#NZUTDpXOH5hd4mcguXWzF?OCi@dbamn
zbslV>kcYJG=ej>WFq|amHHNw7*sp}sy78At2floNMeifjX26x;zI0UlxiZN`e#Bnq
zfyA98i26%`{Ccw8^}(wU-V5+yJc`c)ib3cv3|V_>O>i)#6SZ*B<msMSrXnKJK0o^)
zI$<MaQj|!BeB1ag9Tz^ZI9AC^TD<$Nb%5h;(`|2t+GH3YE|{%29n%lZygAKO`Lhxb
z7CeWV^P%^&zlJ9dDZ@9&66XncD94)@Y)3>rzySd@?^XKLYaCk~>b(N?u&>ohEBZyG
zk8Mk}QE(KP-7sYAO~Tl6=cdm>%-ZT!ve1qU=hrI0>5^wz7=E>AJ+ZHCaI~Dhxgn4Z
zps_#lF+#D#xAvZ^!w9Sf_v1Ohm(eu!P3@Y&jFbhidtzy|T`k^(TlF2j4~2lf2Z(}K
z-3u~QHVQ`pkk)hyNin|L_;+V70GIhRE|CcZ*`L_2qX0nx{b!`Sqk2*yVA8T56N-xh
znAeN``KsKt|2O4o`?%g=O8X@x<VFONcL7-^#JMY$oIRA$RP@Wu{Oik65(k5fGoNb@
zD8E?kCkVTk7*oi__>0_b;eACTXE3G3M2d<^MK0TryqJT8VkiFci2bE=&))B5|GIsA
zz+WmFd|reKg~cb;p$etx>GY2gtS7|W)L`S0zH4aR!klz&L~810ziJa8mGk*ci)d*`
zFH!b)qBVXee*_AxDHaH-g%A`zP}zT*1kv|+d~~}#s`$%pNlq!9vw{LJ5}?rlqz_e%
zAHNE-U~FtG=&LozvRufa?GThfyaD#EVmcYV<<L9-0E@b{o=!8<${nAx5g-#yX(CKR
zi3)<322NU(ZW5deb5q3fQRTkmI?=zSMHnh6(!8|wIha%9&?q|n`*?aJQ7wE-z6VAN
zDH7l9y#yBu=Yg3YzO<B`{$Rm;y=}}87TP4&Ky2JUAI)%3Wx7&tF$K@`NL<Le)eTah
zG~*1a%i2vjKoTG_z4oH$gQKU!++qSnF0eA~)U(e*ZEkce+_`g(ob)^R4NlJwY(&Gw
z_B~9(utbYzlf+L?6RLdS0=e;3Ce+S9;j$!>Fd7Z39UwpaMx0DDpanFi&APil4(P!~
zm&-PE2;iQlzOg)a4sX0CQiNOHy!iQ)KN0>E#JzIO!DbSnZoaDlXXvWp;o>6JHsre|
zs5(i?1mf1)fk%JXzW2n^beZJIB{k>^oqAg}PBrK_wJ@aGRmJU)rp~UtD;ae32f#@*
z)gY{(_BEh~GE>_5ZqD{K(7bwnPNL7w@UE!IDu;X4!BAo-3i*4~TIN1v)cp57s>!+w
zFgf_LrSj2Q4ufbO2o2HjMtwgpmRG}q<iV@{YUUf}ZS{Wig#I70+|4osQ4RWV;a!+_
z8a6Qk)Y%VTE{9&6js*h*0d_&eQ3Qe#x&x4?gTJH#Z`6p5SH*_v?l@`zUse)y>hLb8
zY#F9sj8TpVUWTHc3AYJwCF}$c%~}EbU`&;or-n?zsGxp(<8~~MWRSK4fr;mNU>gQo
z#Ba92sY;?=Q<He#m+cK!If)qO4x<3j**0sB@vv2P)co$SwGx9XjMD^~2Ou>733M<7
z!b`%AE))O&Q31cqgxgi$ZPm$sIq~xwSGc((JZ8iLjC@^yk;PM6L7K<s<_)G)Rzpt2
z>e!!eoORzR5v!%6ug&=K<z_$>BtMtotiF&#8e;&yD0UE6d9*SH2~azF<@3J|S`Z{j
zQPG^wZN6eeI+jyV@VC%_`=LractbA~HNS#nh5MK(F_*63)q0~aVIO$!HS-64t<(8U
zPM^=n%I<)M1z*znI<G-V!i7E<i$mJcF7P5`9r8uqBe_-AJJsq$Bq&!$I7h$r!=Qw`
z&5AtV>A68QQB$D$*aT7id$O1ZAwMu@>B&~bHsiP8XjcnAv0F_@Cdphk)mXmjxj<b{
zs9wb5$IizgD|N^>jemV-)AdMh=RGq47i4`RRhw*wVBuGFuPa%acYe8f089Thd6(VM
z^|DMeW)qOC`~AwjH*Vh==b3*o#>B*g77|1?OILG0-l*d`lz!JUry`Vh2_*msw}d6J
zT#fuzTwEMUDz%jkT~PFj2n|JZNh-i`=b;Np7ZZO#f!T7j_e9nXcS*%8m8xThX2l24
zFUz^)z<3>~pPrqF$~R?gZO71e>8*a+1Sxk~98)g|eG@ePgN8TTMJzSsq1ECw+}Kwz
zJ<<Urs=k{+V!Hh%5RL3}lcbfTn+oN*Oe`!sT<qFNJJ$7Ev~z8wIU5hqmm<E8fnRs(
zPK%&(jIsdrao5XJfK-I=Yj1el#W}CD!RsW#aA7dDW~}Nq=~F>-MKkx((qL{glm~5Y
z&^PX)QL0TlzCuY|sSVE4g9<MH^w=LXbt_Wsz*dQH01;A1@0sJS*yqnt!wiV6P^(HN
zUMq8$ury#ybdQM@wlMV05}oR;*J{ZazszOg`YGj}76hCA>*0UUQfR{`m4~3p=y)#t
zDbZl14DP?XC<&9ofjSHl+`}GFQU34NJa8e|QGgt?@X6Z44slK_N9f$hxmCsZA43yH
z3Oacw8ZncIck~NlArlbBj-^etaocCq`PF5wAR2?PwW4UE0Ptw0^}`qZ14g3VVbNcc
z^@!<i2688)`W%>LfDkzL?`OC9umy3F2M0DFpmV$8t{(v03z17`*nT6HMf@86o(}Aa
z7tG*pB0Xo+3J6mxH|kP6=eyX!s72tfBzPTp2^}H42XtH?-vLr-T|Z3iS%31`=d=5t
zPLRGVN+E2|#7I;vWQN8Xj2*CU7UIWWAwjc=yA0OIFzLeO1}dZhe)_HleK*bAYoo-i
zwYQ7-p_UEg`{j1)xDitC1`DI77#1l}oj|L=7KwcdXqhyJ;eUXdnTjPUrE0mt9)4fM
z1G3puG~w_N`F}-vtEP-Z()L3i4iYq~Y7-PA0_p4NHrPTNvyNt78u6B-Pk2q{!1qYC
z`j{ki);R!lc)!RFo#DpRILtY;G{m13c2Pi<@tABhaT;vkuWnK(c}~!?i?KMmWti>L
z8R0<wWo)h%g+I{`lfB?LY8|nT61;^L`4rHer3FJhZe6h8xKywF1O1nIbS#ytZP0ha
z6Ut?<Z9VTe{G)AYaSEOpj+7qKUjPl%yznPj9-N37BIac|Oh9n}oiHcC0J1d+=&Da+
zsAMu=)Q<y8m`l7KPTRA$&JMn^_4$G8DI&|pr*0Vb7i=w4j{r{~7l*2d8+}1o?Jh`w
zcu<&;$0J2_fY{65;S{cT_{?db&dAfapQ4mN+)~H$v3$fv-mePeMt6e<+GoEhuSL(A
zG#vuD(Rn=$XeReMVj=3^@i6#sN1K~S>UULEQc8%tNumFMZ3AJy&PEh_)3!nQ!$*9F
z|4nEt#~s_RdVd)AwJSruJIIl$i?s}FzekH${~)RbtB6ySx|eU{L@=+OM7?qUuli(9
ztS8ym5h?p@%?FVn0$4=_4;*;4`d1_EzfVFgkBTx~3AgQh53lEeLKkcp{A;8+Zu;%5
zVHwhFP*Sw7&p9=A0A^w#Tz#)m^53^(@5Liz<+ExlC?(5QF{Rl<>TjZ+%Qdu--!O9c
zPPrQO_SJ@EtJ3n0N0z8z^Z^E!USZLk4Lx<TS%5+jo?f=pp(S&lcrNj^@-Fe%{pSTG
zTe5GIWgh$N=8+;FVo&G$LJKfV4xN=0DsJ#Br9on>MLQ-GKZAvD^RlDyzqgxnd5_`M
zk$%8FqzpLD+Ar~$REKA_x^SiR{6oct+$w$c+bWiGKOtn-aZr>h_8OU(gxX{C7-fm<
zIb8&<l+HkJIfHaDRpDbadMvL)Chk8|`JYH8VetQbAuqDKI<am0k_Da^<a=m$$eF=S
zJIvpy6iZn!>`F}BDr$M)HFs=nYY1cP_iUBl+@aaeuf%`-A*ZF47P3`k=<U~#ttof*
z2HhDPpEJyN^04a*H!(ik^!U-#ujJA2KC!;>s=MKRZe!tn2-WTvwO4}uNy52qJ~Lh;
zSzKJ~r*aLb4UX~&xf?fbd^Jmd>2c;vIY5jBBRhUCN=A6&8uEOiuk;G?BO}{@ZK~RO
zfajz$;B71Y8AX2Rd}!HDdwd15d8Vqem6a6=RH~{uesp~WBXJ&6qON0iDIiMQzNtOf
zV|XVZATyt&z~hXm%NdZf9c?ADQ%C+uO2-FB@)Mt<aQR}THSAp&>geFVT)z2N6Zxpq
zXW*1a1}JSWyotS_myGswbaVi^q@2Yr70UHPb<39C;4Ry>HN?o!axv>mk^SEd%%(6N
z!_WNo@cQAN9tH*mqxehY1&!x_Z(!fSgRkM!Vf@{SolGFXKls@2)Bn1QlIfWlbh*nq
zTQq9PfY2OAvlLegI;7yG_4o$aNL_XM-<JTMx_<w*F7&Uht(B0F5DXwc<@>KUh>42>
z&s{3?Pv|Ee<g%T9#Cd#NpK<SpRjXEk=j#Vt!V8pD|K8UjF|k}2nvH%K!8u@-{zGAu
z{B-u;L3ostGly}gOxwUke6sKLn~(1BtAnkdLb>X&Um+aZ-fljAc=N`MxCdJA5dzES
z0&JzDYiVvC+_4qE6;)!fJT4XHP$^KjKP8A6d{}2o#_;O4ze5>bg!VvH!bcw-^5WjV
z6WUJCqnFRJb*n*&CWH<+F}kl`!;dN4#>=yFl^U7@HuR-`G=tGMXw?lebM}94`YPi-
zHFFy~JLY5dFAbM}kf#2X4C?jNckXkJaP#>#KZiuOOG2VGa2I)F74dLW+7BT_i;9YT
zRoZ)^@%O7@|K8u7z|&dY-etH;mhBtygQ(rhWBUFTdbOt$FIHMn)6yoZ#)sVOA|Lke
zP3->XpMSP&*^&~bzC58mf9K(SR{4v{yxiRA=x7#p@>x5V-t49HAJR<Y<lvAR;KWUC
zIB@8YO8mEyl9E|3Me#ph0ZkKB&S2~Xi1{QiFi=)zMNgLL(H*gdl=iPDPoBg;CD_yB
zv+(7QCUF8B2VAUcg3F@GfG&Tn3tWCe^8N8Uf8(BffAVh(kZ(8teJ1kl_04}nnS6ii
z@6(WPcm9n(@-1c8-;p8T{^z6rM+0(SbAY^`vaPGPm$<BnT6Og3QDNbC=qY-5c<Act
zVn=@k_G&69<X`P5!rYS1+Ml1Qu{XKPAn9yoWc&`QfbA-<%5v`9`M##6rm0DE|Ni}4
zTwK`LPV@}maInhpv9VwpMqG=g=Wn9EuDoIW`YgAZl)CeXejlXABQ~8Mh;D>EFuh-w
zU>PO*W0AEcChT3yCNGbSjs2{RDZ`et1|%^wU!J7OL1h5n6Gzx#+t8zg@UXC7zkYG$
zyn6TU-AnStDJdy=Wcf=#;LqP)K2deCfn7^QMJ1ye_apiU>L)xB5B*mv9A8C|7O-yB
z9?(;vP~Nfd@+L#W>gnmxxH^QV7==+Uc=*K3W;!~<M#=$h?nG$n`<iM#eM(PQXl`zX
zRp#{R)57g?_*ChtCjylz@gfm?3Cqh>nCwLjmvDDM{zTej92_1RYQeUeFqt*(SgYVR
zcl`Y`l;`sE^NC#m#w9o%p|e`Y#o*K4pZmL3_Sb=8c9jKq-@3)gy6&7({5MF<^2jNJ
zF2N{U?BXu=)!`y2QI6u9@iyPT%U+ml@rFY+2Lwo#YGvy5^z<Q(ln<>kqa!1aAaKCn
z8Fw*$$o&xBQvP=(3VWe}4H4+m8JfIz@2-Ik5=RC81w#u4L=~C?@FvmBV77oT=kGgt
z*B&E;PqT+d3T@lg)>fE#)^FH=cG%;<O!vv(Z3SpX(A*)r;6Xj@-v5Kk?L0M5(zrrc
zRF|Kca^27kBuhG}9IV^C`Q6*Vmb|<?8NVG49Yt&Wq>mVuZuIujNix;rfAa7l*)D^)
z^LdS~-6DjQXyqf<MlRL*N2~eLY~wqwbt3U1)!Ma4M-~78q;mnUpC4_Nfq{X=5;v~P
zj|={r%q!wy`$W^@w3eZvA@PV{2i6|$?!G>VFN^4`!S@;T_{bMj893Ba5u0eZH0QgE
z3JD1zxI{!mcI_gadrxSKr%s)!Y^IArRwV9zbcXS94D|G9cVF3m9a8(}464J*3JTRX
z)@;}zXfX{vPbbesRb>cX2T1nxmg)PZJ3Bk)rhe-|H$b_<iyNt_T_87(jgL$H!T!&e
zU!71(+UtU=yH;IwB$ea+3g2^5l5W#)yuH0GY3srKQJa{Un8H}rhl@GeVz7I2+sMw%
z6mDN;A%$4dQ8mX1N5|k1J<$iN6TErzCX6Gof0(#6FCUJ8>El#(SIIpl?&)lmgAcvE
zvH$M2#fWr>jF1Sk4w@B-5Lx%nCZX|fUP;RGq4yT{SNNF;Vo)L#8+!(ggcz8Z3LwwI
zCY5n41mz1o1O;s9#3b3JjKjs#1bF~f8g_GCw;xD6=toION!>ko&Dq&~ezXkwZPtky
z=)u@B$7H$y<tjUeIVaS#EkAE?hcK)Og1nbc;j<y81h%NCt6$Lw+uc_a8Q|k{kdyO%
z%Q55)_#7K1@8F*J{Fh5k)SJ(rKg0Y`X+`?R1_sfIG|w-_!_bP5J=Z02<rQKUno2e_
zT}Bq<zzH~<k<<7z?9dTIkRRa;mUkoC9OhK|2{ws_KKEL)X3h6=J@7b2TV!ogykC8n
zZa?gV`*O}EUZmY|w7gelHR{=xzf0H40ow%924J#(+wsrhpJdNJ!!1oIq;>Mfu213o
z!Nxo2)fXduR#nt})5M;F(e}bsrt*!avqCfzu_N+jnj2&t<D?zDJ;-n_C6H_u%{0PF
zxiwNwAFatlFMU#YHy-z207-3L8|`J0B^@K;IQS^he=94ixX1i8bIogzbTCrZ@z*XR
z(d9x$ndQJ^=(+!SnAVF&c6W2T%zW3^cjLy5!690?(f4FyUC?(lTDo-=iXwC;jpl?c
z=x*#+j0kRl;N|Mdf_-R|-Tyu$KI&r#G>`hu#xHA|TUua}-SZ<GawX;%o0-3rmBpd)
zfQACTLKA07nXT0`(x7rzV+Ae2fa2=vDr8!B84l6tClIpvI3Y0jr9Fv^jO>?snnL#t
zYtjR#FXh(!64BkT!5}5f*Dd}#-5Qq7o0YI?=2!_HUi%j~P{&`0^|8oGwCz9m@ZjOY
z7moHVK)K(}Z}0&T#U|nQ2%mfZJ|8w}pxk}T@2qQTa>cH9BXJiP7Lei?d3D{8ssPMj
zaKL`#^TB65F@Bq5v58=>D3_j~&TwDfROdZj*-DImV(pcR9a*xmuP-J(>E8t5)r#!n
z;J)(Z984F{^3f#|0BDzJ{~al3XXlS0ecjy+KQ{warSy=!La_Pl<Ca8M5JuL@ufbcO
z^yOXcxN+iHKf@Dn`v;5_No94Aa0kNVjF?;x96YF5;E|Wr@pJvUb*LBxM^lq8yxO{T
zD-=@m6Jfm%bm(;}>+9>2yw@TOC6+<c^`g_(K?y3af)Z(V=^T)%EblxB$4c<i{FWE;
z%#E)acO#eMSs#YL<Dr_-m0~xJCffW+$Elkq(>2B2rmx;S*?8b5*v>h5c>$r$B{L|W
zwJ|bCe6eA$Xu?(-38^V1Lt`H)LY9PO8UL*)oadhG{601{mHBNKav!Ib$MlGPEe~n!
z*mT`aL#XJaJ<3Ty-HwjlZHl=4SXC7qQdyl^U0FE?XHaa<+Oy+W)2J>wnxAfj-orNi
zS@fPDvIbh=3>PbW5&&uqE`1S^!DNXeQUdb=2>W3DuP0yM2-W$aX=~=4(9lo|t65Z7
zh<wtH=gyrA=B8b@j&G?SZGL)7m-*?pX-Y@eXbCd3BA>$QGO#ED-r(d^>>7^Nl-<(e
zf^o^L9x;zYX1!d`Uf?`#4Au7X()&wx(u>+Qzgxwru=YV1{=GH{7_H&nUPeYnWA{0T
ztPi<`L}D~lPE07MXh`b@9~70-=_b0SVq{1N35D#M6t3(KrKQ*`^~4jp|JtF(RGEwb
zlz@J0STZCL{TdOL(u1wCnH||PxE9~te-qvG^tIZ3jk#Af?lSuE&peSmzvleS6l!Yf
z(_t0CEf?4Cdmg|dazXT;Jgh|IGr<PU3m10h9L_1%g6$KHtdIE<e?C7urIT%svjuqM
z%0uE2hsZrfLQln<>-Y)%I)3lpOW`LFOv|lEcMmWm!2%^w&rvrkAV3k2(07pYn7xZj
z_QPEoz35awbbO8^i_bm3@H@@^DVloN5dXMw5v72nO&vz**bC?la2!6Ia?VroO&$6W
zYR^utX_QCGbx_}i>ni<a67*`Cnn)_rjq+ycF38PnY-|_VIsr&hde97D7`lM9+unZ3
z)Y;T@01|MbZ`on=uRz)UjxC|^-n5}shsS%O&C-tv3OWLmrD%oXBrJ;tzCB-lTJjt^
zWPwDG66SO7UZ!PJ?;9|n{~}((H)|x9a^JmEcT|S^NG(y*$ar5%aj_xFeci^ve^yal
z{#5|QJ^BS28S;&p?hzt(`;Hv>5`Jm$gdn#B`asy5j&ocpqQpc*X3*g{+-Qt+M`U*6
zs%-=N{G`XRLk3>@tb|`R>M>|)0R>J?Oa!;gVs1t~0G26KdzLXJCoez04g6lwiLjzC
z;Cuyu@XOEF0JVnV6az#P+F4UmQ$h^~HP`MH6LUkiyN(<uQUI@3N)BgE^cgodx05Gt
zoX`g*z=Lok&ZD2_vnc!w44<8esUE)Fffpc)&;{LLpkS2Ix(Iwqj(qCu>H_9IptVa<
z(i3%o<{qK2nr5-fhv7wCc%?%6BREWK2hJ?ZNk?+ThLcw8sv!q~zz+bIU~aCMr{@yx
z_}1;)#NF+=v<Ms=U5a%acK`n8XsHbYFK_}1)}JGkQnUiN4jtM^LxZe6jXoaN+>=j8
z%V^Z8H_L>nqp^(KX{@x3p8m2@bnO)bLqk6yAg?sEjj?>s3ib5%(x4)@dL990#`4>=
zZrv52*-##x+cc~P7wGQf!%?Y}CIRu2GI-|yVgWv=Tj)G$c+N!d8{SSE?eD+xODm)|
z*<p5ah%2WO9naG_=o?9CwskX~SXfvXA0L;$oo(GF@j!1Hk)5z0ee?;I$=x4>C*-?7
zg`=+YasV>2B|0IMx7F-|_5LbM!uL=Y@%W-#T$OBoKx{2lQ<gs944M;L6=W*P%Nz3D
zb3U3~bo}0kZk`cb`QYLJgVBGBokBO#p}xL8Fwt9?E@=Aj9A_HM?c28x&|8w1w<cqk
zM(Jgr`Nf4f_!c}_=zQ}Y1_zUU`6#IS>YXuz*Pq^BX6oUUIQRhWy;N*anV)yt;D~Bn
zfHXb$0`L?|7zi^NQ+UpD5t8+@XU{Oo={wn6h0y?o@4!z*9Y-eV>pv^*d>cu~ikBB0
z(lR^zqzBY7yxL651poa2@F)C{i2w)W%$Bof`R$U?>_A6Nl#}xc&~NlHAUuF=vg-hA
z`_tJMk?@g0CxUULro5a_$Yt)!6&wMQz<;4|`eNS{927u|(x*>j=gla(Dze^)#`d3I
zWl3yOj_YEN9rly`+p%@RzA@}Uw8VZnICkde=b>DIqQc3BV{&kg(KUo5+Oiyb_daSI
zL(2_kQLxa{f5pyGq;j$c0bIAOO)v2xcB7%guZ^~1dEg$9(4%9E=Zr`T4IcnE%6+S2
zsOP9itn3#a2g42wZH|EfSWNgYE#=_o!Lz|YH4Rz{uZfO|%5!^P-0!rcl^FB7_wVlC
zzmH`MMTI;B<$4T)#vFz=BE!6W`L`v5*Z-rXU=aQzE|C!VUy;O~UxYl}qk8=K@k^JY
zh(JRzx@=^GhPzy4OG^vPXFsQjAew;1>sM5CGB)P=IAGrUlVGgj;;tEg-+|V3FFK(Q
zLgr>?b+he1j*j9j>U;L??S^3z0o+&eAC2(8t!zeZgmtKMQV+&aS??}e7@4~#3qp8w
zUfU~CSS(QtW(%m?+}&~DPcRMj)zziDDhyawzGvIk5S6RMMYuVi+}X1qfx;5?h_@Mr
zH|1B?E}hEC%CfPsLEaLBQ3c5xS&3cJL;I~k2?QXdtz>M%23<#D)<fIllLa4iG)VH8
z=6CS9z!7%%$PqhRTVzkNNf)>aEE^OWWZv1;wJBY%H0x5b<+mQ}3btx^^9-nIt|m7(
z7qZ|IZYAB5HeB=M$s&RZV8I2gcpe+u)7?$>pg~du_HoBKNyP1qq<4^YkCHz>jq{Dh
zd7lJxN7M=AT-a;e;L1mf2gfd601QC2$5SL@M?Y>pQ#qjO?k6^#i&^T)mI4z7mu}jg
z`nR@EW|kp52i^OzgR!4R=xY0MS9Y3SoG<lu`1xgH<AbdJ<{Eh}>4l!fP(tn<#6d^r
zCOwlPBlR>jp9IEyP1I7$X6pN)fTPABh%&LV4h;;LR=-JvyAh_zu)wu+bacS#4cQLn
zcAD|n?+zVZTS=+?&7v6@kX9qWJeb$E4LZzE+mzYM$jr)}twfbA4E#`522rAMI`d~o
z$0$tnkWxj31sfQ<Hm||Mk2Q3SJ;!eGG5AKd)i7b}x($pLGcz-j!!IP}95lLQ^D@dl
zS8iV`xI-$Hos$XBfbXB9*(nfr<L9=5guk#CZq7gWA}VSpC+C@^rAs&y5hq-o{v(x{
zeDEPLX-m_57>eIBuoQCSfo7t=pC5xj1FB|<U<0~F!NK${I{DcNwO61r8orUk9I$@h
z!)<aV27;5~Bfi3U-tYDGPT0{wv!SMur<3t&1`i+2dFrnF`-8-J6nFUTgJ>ou)?Bq#
zE8XnjlP5nPEt(B}H}bpp=Lg#<UXyJNSI)!~(|-pBiDJ3q=a=U`m*MddWgVeqa6%bT
zo9#UQ!msm23{jWh+LiHVd8>L8at19eEkReRHz(U~<Ow$Rm_#_gKnb4ZfR;u~=tJyJ
z=ss3f-t+O%d~4v91hsD;6eoCD$VTuJCkg?AJg~6{FVD=)nRAKZJRpFgB^(0DK)sza
zd7Jm~%I=;6yu7~=jpUfhPZlFM!36{cI&Hq+$~1AiwY!oUe;wu*Sq`#)2s^4yN=s9j
zrA~(XLi4+(zJ3QxC<!#wKG2O9n%WkPi`SyBbku2Bn_2J*El-2_1D%k?t?DV4?~3oS
zJqD8wk;T#?znT3h^4dTSR)k7{wFSU@&BcZ6vI8lCOg4k_3*eV|k*J}mxr2@_)#2C0
zo|RM;+e>$SHkltgiZ`Y7>f%BZL~3CTWqq8xJI6S=vZDL*y?-`^vlA#5-hwM7=@lgD
zja9*FG+uRtQRtoNmwf{_A1EoIUyA!CM{I$H^g#;m>)XF?A90S6EzL&1BE_7?z=w{u
z8e=sQ)$=d92R0}EA&e<{mqCfS0IUG~<~ir3@uW?T&RFeeFU)e_O}0-(A|N?@%X2%I
zc+ncxD*Qu$QzFZ@Z2;HO3q5)`22$XGqyi;vo444gg@ZG6fUhvF@^BNu8~Fnu6I#{`
zc+<PAtnDhm$#cjs*6uP@CdFJS=98Lbw@>jhu{T7yDY>X_r&Rl=*FG$H^8E9aqkwn<
z)3i7WPOK|uo5lkbxwIcU&4Z3zfP`yrpB5M@j7URnRG-{9+?1)Psh6YD1CdOM*GN`D
zfrf@A#PTYbiQunC=kZu!P8ueNnzesSsvhZkvz04fwS>LVV&U7~#l@+Duv9fu(^rc{
z=``Ckg`1?BS5|*tQ@YFN63f{0l$jn=3`pG*6w@OwC$Ss)Onka!)A!Pr2ZjM$DH|q#
zc%0%jq7vNkUE2EqG~bcc$8aJ95dcFA=CObeVkm)paSUL>nho)$N1x!SHnAQ{L+^r|
zzX$$;r30}!@7Ar=uD6EH!2XW&(}=VG>eZ|G4Q%TowsmlDf7sVRi;z!@%sy4{&VwxC
zP#6t+d;6<b-#{$O%yfb`8O;h|c!MGKgr=JdikR>4Ge#yL2Rh)rC>H22u3WitaV)d|
zzM@?`JV^<=GzLIBV6@}sWZ}DX{o$wKwt^)L<VC^g-+;}?+dg;htfQfUb5j$i3*y~4
zgrW{&A1t&_B|Td!Cco6xk&{(W_P!=)d}(URfno>dua)CyX=C&DF*i=dGe9PUlm^`a
zG7}gxB+QNL*KZfG?MAN%yzz?BEyi{0*Ed4`g_<G7ASNb;zR2F314oW1Yc^mFU{3=|
zS2{*7#g(Y!;GHR9NxFGv(&d84D;dtA4j{uDWuX#8)m}NYF2#+Mh~Xs*;GtIMK32O)
zTRS_F*&r`XMq-o@zPIyEreU6q4Ok4+d`}^Ox82u`Kr_N&gXKUbHeb5kEJFK5h>!1c
z)CjD7jJfMm6D&NW96#wcS@-v(tYQvkt}q%|5lpKf!Q67`LLKfAhbQft^%H_b+}UYR
z+p+NpxD(D{OSKS%8iF<&)M?lX(}|DJGB*J2ms$Z=v``l0!QFf3caWX?e(WcI!hDxG
zs&*0yk7$}+US9hOILz~}`bnpP&S5qN==^7+V`8wMc&Az6Z4>!EvTNYW3%|ho7|!1E
z?aqXGt&75l?VCC*vf*6p1+Uw;Ay@4fyi$0G7>FN_Uv@fu`UB)azcjk-+YhvGiHK-4
zJ%y*fLa+(Dqj7)RmjpfLM`da`_2SSQf(7ybCnw)b%Q1B<m7JWaTyIc4t>#O}A<sAE
z!&{%>*SrHOtui}2B7$du0Z)6N>_bAk#RN_f=%<OxD->)SNtT)g+WG4R=qPrWEB)}{
z3>T-v7891}V&PjhcUF|{qCUJdmWI<#P1AAr9|c7ryqNYt0Qfw-z<=s-Qh8FN`4Q%{
zOZ~Bf@1=_^>>m}uMWE=9oQ9;Mn|a~2iiO1>gnUy4WyngX0!*I_5AyKnrd_GU4kYah
z7Yy_hDl}k}pC0Y<Q@r-Ey!;VPgoaY+D`eEe6cH3ePDNs3isL_4a0#6l;w6Tf%H~@k
zZDNU-L*Kr%{+F12zmmFo1E}ZaO!U~BhYBplI-_tg=HP9)C9Q62Z*~DhEkFxN<va9H
zHueJ1_i<u8c2JACP9}V;0IY);rt{{~Kch!@1hAPJ)$K0hg>z@l^rH=O@nXo}ci0VZ
zi1pOKmp*i$N&|3I%^z|!4h|{AhCd_f8gg8Qs-~uyQa`W=49c3{qr$?%25&q>2Qx%q
zxoTBrYa?}F9c`nR^)waJ$1)0e^yuY+Hs-9qzyI(Mj?EQr>BBrW@$oSmuoMY@IKx`H
z>zMB;miP0OUvT-@oO{N{D?^qb@-miOF*M311?rQ2<L62NDnU*eAed{mA3l6w5wZR7
z<%=E4%joDRGFSk+WImdJ*m}^_(eW%X71Az5y)X+i{077(M<*xXSm;~xcohJZR4fd@
zz>^<Xl<2<=Hn5B0yHZ$z(<70v&>5t`DIrjHvBbl}!+{~Kv6CAI9`PSZz8ue+3yBRI
zUS<F$unPoYDwKknNXaFrZfn=9G39~+)B~<_WP89L6LQeb6F&VLlZzSf<ibS$1BX-t
zXh58Cahb&)7yAG{98KYgBS~-)*S?sPR%}1tBzsQb%GT{@^jHb3O2m1U^b8Cd-!-9N
z!ubI`>r)$q*2@qop@<yy3`N)K5TfE5kjwD7r^i$^lWMQvfJq~}eIEu0g#7hgK9DNj
z2nrAW69ZIrV`znYdV1aqlUGm(VKfWPvR{jJWS)I!!D|Fwsis4R<LtZ3x3u)VdBbOI
z{rm7xEB+e=En#PX9D)M_^()&guyRBa`DygY47OI-JI^jmwZ#hHtn0l^wy0aNUMbwd
z>Ke(*Yez$H!nW|AO%-$;aYd^*X)9`@dC-92<5QY^%)7vJ8Ha|Khra-2z>awt({!L1
z4JW6*fdS{93I<cwFtKyoe$BH~*HBsGEh}1D7La(hs8*Jj-?@7i8!PYIKiE_-f)g!p
zdg$K0d+!vHBR4@4?1#$NS`?=`!kp03l?`UXn0LzrYDxC~P1v2zofzerI)T)8&&6b4
zqk}7ulgR_oiq=&4=p7)8uJwY-Qr}iP@BPY1rpfq6wnKc{iAL{NpSp*JaNdso=XP{3
z51A)IrZf_f%OGMCFc=QWM!-APjUapdANJlnn(Ov$7uFm~lOaP%Wk`n1B2=Obm3hio
zlvx?0q=Cv<GG)q?GS72GsYIr}<{=>|5g}BBeO$WlXFq$r|LnEbde>g-eSg-op5NWX
z=X-sw>%7kMIF9o;UFK%&{8=Mn`%-n%aa-ATtfDPJs^2t7H0G<Nb+Qcg%T8RE;VDe<
zU-)ynR&m#?)5t@?lbXXj8zhh1y0CU&;|d&#y1d$uZ|q`#gE=m(rC3iZ8%6M!|2+{C
z?^Iq~3%sddWJ-gyjVm-sv<#;)xUkz<Mcc#OJ<`xlGg*V*s7$LRbYb&F_;cPYxZeDs
zeE7rj50I1B8szjd*$2IUZ`jjAY+VE2oEg-=1RAs86m(`F_n3F76ECl_D3l*U9oF65
z?SRfz7JtHwrTtS-Vs|@a7&jQK(8d-V5|Y$Ssq1b>O313p(~QcX4PF~?chOZ-I}gDW
z{9$UdIA_7<F)=VO7g&9oL76jwdZ-#WBC-;sE&`{%Aapf5z;ydk-VbK^!-t9d`QAMZ
zJ-z$)@0XaqSl&6uK6+*$=wWn%l_4%zat}|aquxR*kbs@e%wog}vb-M#(Sf7v2FZ8X
zGwDQifk0W5ay_|i*(MGav`xvavAYAYyr`bm4{Qund@1>gF5~F|Aj8#dO!F_kK{Jsz
z4I_~ab2FNErghERzZLG!VD{J}nM1jYTno5tUkwHC8P_32r)TeLoO{G#+o!4V@$n>_
z76pM)|GAI&_8%GG<_?XHPM=px)!-aCmEX@K@Nr~B+N^Tr=(&)+9R^Nqe{D)0&L2Ra
z9uQ2M1$J8KW{*l4ar{~82|WS?uI98ZG;obQBtkl}aKhRJl~h#bu#mM~4TK$>uR&=%
zi;@EEe!9qgPM&PU&nA8Lhy$VJIYZF6d=|VVg?9*)8=b{XwYBMR_dt_@0wi$Bt?Rb?
zf`IvojCuVq?kkWMTfhNlf@VeD1%>b;vFCbUsyQo=gtr*Y{NQDQL+=&jODOG89{Rox
z<TETG4TM5G8R*wQZIbEVU~HJvZZVo?`&MZ-eLyL(Vx9klgtnhnrh)d)-w18uw`<zA
z-nxH->~X7s0Qqlkehn%WYz+KnA{`Ja-FRJ`S7hnVb<z|08#gwAHG9AYyg?*Sdy(2c
zB%&XaaZ*~XDLQU$Zao1BF)?fd+1DI-Q3g=1Ky6uZ>1S40X6HDq#9@hOJ{cJ!(Hru$
zQTdEaOuB}K8S|ji1X(xK^!8TZqjojUs!UlwgEw-SoQ@X~rK)IQC6B1(D^?`C-LQao
zFlkQd9of1#!*5%v0Zs7m<LIBo)2yd~fDz>JS}O|?6ULUO?oMAdyYu8wi1jotenI}Q
zBl8E^Ta=t6&=uI?Wot5mb9TZ-B}pR{+K1eW6Q^y2xlkO3oH#@hMZ7*_u>WoOWTBX<
zd30!h1U_HSx!Szk*-KDrckkdjb%yOUT@?p!7006gh1*8V6$ikLA1gAPiaY!>h-O8Z
zT+#j&AxpbgNr&#4;d$U3-l9@hxo-`H4H-bh@Xk|wKkq2?rxP@|gUT9DOuD)_5u1Xp
z4I9{H*KOF)n|tQg*n|4z;(sdZZMlVnD94W5G(?RFXl3dYTQygiYn4S8;6{wjuw6fI
z8yx;Ezs5oX5D|6_Gn+!Oapr!|xIY1L9ILf(&{P;~(haXcp&B|?Z!6UlZQcrh3tV2f
zB45WJMj>%UM<H&Z@urTFlCP*lLJ9Wb2AD{}gNW1}JwGA*dT_u!ui!HJIv_GFy8U;5
z4d;FPzVdMyNAdpo-Jv`_W-$vtH9`2ZzB0E7kgt0odE~X^CodtoT!~!~+>{MBhZypZ
z)8-9|yKB9^8DJ7@l;2*ZFZa*>>Nlxf7XO}l-8xZhFb_V@S4~%Ij&@A-?h}}!ar}at
zce=e|VHz9)Z`d$)0Hj%@GJ&`Yj33SdD+>#6)J~3C9g|;#qQU_|(m|XbE{^Zu<~}Mq
zC3=O`wqAb@H)|m~D8%Vq?Utg68w_TQ*B_jHo`>GG>3Z}gy24xo)L3g6=}f^Lg#Mp`
z{ZDEe)}-oYKNvcCc;=B1Ml?Kpc*MiTIv)8)(_yfyRtXESHMsR3GGHD7ho@!NZ4}y3
zCY`}x%5)42c%K+&X2351Qj}rA+Vp%l7rK3LfAfga>5n?HPM}tn!+ug&Hw*E_=;wYg
zP1hNxeiB;#kZMtt<SVuSd948XseN=8wszKrSo-o)(liqtYr}{0Bq$7>Vwd<&rzb?v
z(%Gs7?PJ~cJVd^rBPR64w%egK2>}avX-Lb?Zr8m4Iu@PG@)s|bWbFnzNOnBe@nNu2
zD5`K+=5a=Rxs<o&EYN~fQj7z^h3;Y`|0G>~;>y2km%Q`$w0qH;lN+P*A9*nuBtD+1
zaBe)xT42?A<k%0T3UiUhAo4?CJMvBtDhS~pf%3>NaPAE(cgJqpxN!$RA3uMqNuS<B
zREbT8Lw=5fC70;TQ_AZph{OG3<yD1?#L3Bt*kGgFgGHi(cq-gt=|tFz>KhwT^W;p0
zSJl_s;SDvdY28-0>XO|Lf!YT85YOzME%J-SffoKQLyddHANXn03~Y2VIK!5Bnl16x
zh1+4cUWl}#e6RKtt@6~Mv~Ys${L;$R5)6W^Bc6=|<gKvk2?n9a+inKnJYgi+0qZqf
zVe9MzCnqP-Iwu2XS`gTs4BU1ja2$Z9{1+0<y58i1f|hC>9L?ZMXZ^bisK1IlC@M1x
z3JPk=%t#~LJrqn`T!Gzpzd6NneF^IHo^bY^2PX1+c9wXm@bX$C;=ukd#K`$HQX{Tt
z7{=k|ZpUU^_ZCnS;`)AzxAfgdY-y1mo^shQD0tFf$o9^yZXHX@l(GC(Ba!rsjK(GB
z56yMf*B`GC`|^U6RFPY6b+8U>{2_R9w(-;u<Zjj5ImN`Z4d7HOHg6HiRux51n69gK
zw~PfOa}Mo43@y3&{maJ;7gIkVbJD*YkDAE;XQ3jlDBN6;Ff@g34Q8FVpLM-%YQYBt
zt3#w16;wD*fc;XVEt*k6;;+ESwGVD!h+}*_Jn!nu?itN#go~X!UVP-Bn%dn*eBWf$
zkIkmdB6%4op7?yKz%qD2PK>1<m}Xt4x~on^u&-`j?>!_^^KBYAHB%bgH5$NPbY_G)
z5SLhcClJ|iVFqVq06Bb35J6PnMcUp3Xe=tJ&HFjrWU}q#J!FByvh32Xs#5_^t5NZ~
zG+QC1nToi*XYZWRE9P7KkT{=SIk534D$m_>x=NJA9rK)?0mXe6|JFOxotK9OucO-E
z1$ofHn>DAn=7sJ~S$r(xmO0+QFZ{(!V{T$*W~?m}tZXwhh+!Cb0mukFfR+*fd~n(g
z#&wuQMtdvdI56(T9X-@YpOH2hT)cP@C=bfYY4$L&MST~W1_MAeh3tqC=I0LbD9qBO
z6BuD#MBXsULTERqRlpZJG_X9$+aL*jhVPiE1$RU%;@*{zQuAF9%ASjK+(LTMztJ1(
z7g|yd8~lP&ZI^EV0-(2#kI~bMl=uTXcJ!fE7MJ|qP=lj$C_?+X*mX$VOw&_h)DR3%
zN*Ba|YDjMG4e$>PZ1p7FhXvO~440Pb!sY;ZzJaKLyLkO7=|n`8XZKcCr6tQq+8<c!
zkEJ4G+og5?b<<H*!@)gef(DggSz}Yeov~|@A>Ta-H_(}sPrG5mo$>K;VrmQSxa(d^
zi&jYQjo0W^Pnqxg`t>)_O_T--E$+?xoB_Kdg}H=+0%xM|*-lx8xXj5914X~gu0^>1
zzwcQ*x6ibJJ|6(2sQr6AgK{E5nMEUVX2Cd|Aw~8c@yAo1nhh95wLkXZnT=yF)|s@I
zgJvi=X&DaEB(LBdfd0PsOJy2dL+~M24?;Q&&^hJX{DUoyFN`A!YYm!A)&iQNp`$CZ
zFmE`A>b<NH^{0@R@-*0Y7pEA|a3EoLG6&W%U06yD8m?WqZDHudwwZx})5v5F;kHRG
zkRF+?z<YSssEzC_Wd6Hy{L<^689JHXn7j^{y1Jv}LQ?Q0NAM9aNVJrj`~90tRanj7
zydgFjE$Ye1zFi2h*cuTN$ELo>@Bdsg0<*Ir*^vtG`Xc8a_lMguZT=)n;Sv(hN77(m
zZvN0-sTul#X!j#)EKw{>S$Td9*x)~pzC?GF8+;w=5@u?>NU0vmi6HddedbyCv!xG^
zElVh3OAy=srfTIRRQJZfh6x;vnypsASl`g_8y5xMjI}%8LxznCn}9z9tso-uoxf^W
zk*Nh1Ba6tX0iQH6KO#6b(q!Ug+a_t0(>|D3D^+Gu?(zU|=0N06?&rW}Tp~Xb2qg|C
zzLC=9n9Nc}-2l@(^3VlAOMB5$h)$htV55I&pX6?ouU#WeO-%$RiXTTaa<J@|O)M;^
z6m4!C=Kh}dR{K(B-pi|+ZwPE5`0LP0DrOX!Uyw%#Qrggptnu^cBxHvyPj{$HSL}Tc
zp_4*O10*r34N4O~el#ORdtApgWpbBuPKPu4-7{Mme*Sv2^5Q1F@!KgcW4{gdNPwST
z)0H59|2Y_1)E!7!aqOH*tmK8to1=}f(z>%EwoQOTcJ+)P=SH5Fo0|(5hsbJueX;$%
z^;T3kpY23berB|TCAa1yRm%MRuWVAzJiB+BPkek0>Z-n`2LFOcaX@+@)se<Egslz&
zpz3DlvVrY#SDpe9Liiz5OoW7lApd^q{Iv<}8=^9FFPO>mgIElJnn89GYUJPsh@+_z
zwYZ{!f)dlpwHr3bkpg~j%^Rl|#@$J^vk(2laozQtrAVivvlX+zj{Ix{$mj|0EUup(
zz-(OFLjhqEVR&9^Og|}=kpYXCvYlpbXS<~uEq**|%n-^uG&w?-p@s4jdQD(vKo#FK
zc_zD`i_ULm=tKx=Pl&KV+F~DSbwPkiAO%T4UWE+ELW}S6Ba0j!h^2Whkb7pNheSqN
zJrtW?GlGKM>gk>^WpEya@C&yp#K#=A9|mk8Xj0CX84?>iIx>=;sR{5Jwb!u`P#g+g
zHzSmj%OKe-h_dfL5vE*-t+{KP=uw1K5U2pq1=G~j)HZI~lpy^P#LYV1j$zvNipCfS
zi)?8D1&UlW`GbB2yWQLOHi8(vUR+$9)qoChvxjc*UBPL&9S{IDsZa;8a5Xvr2B0ih
zv2F@}ZKtLyaFL&L6rA7Z{OCk1?<xHmpTo1}jfIilmxUBs)G}J#05;wf1)r>@D7@N`
zHdu3rrifpfvGw0|84I2KCd|m91b92w$y*uVPEWBJ<I;c%iQld{DuUA$fRgwLrf-Em
z-1NEm`H#zeNnv1TchFBM8Yb%8Gt1w3Z7c`cMPeOjHgg-yc=GyG9Xxvo_89st85$%?
zu~cyc$gkXb%QDQrwi($qzvX#ski5(=e}N~a*`q>zEgG(-N$R3=*Bq~P4pHq$!cK&u
z7!^LT9Ki4oAP*tY(R0Lm1AXr}>1A?HF%2#Wg-BD+{GUJ0f}F6A9ff)k2rd4ZAI?j2
zGtpF;BlgSRBmritQ02wlpq5+LYhtiZ4t85XfvA66nk+I8Y$I&>35*N$gy<P3qP6}=
z3HCekB4!pA7GSoS?}6ogO$#c5qB`?Q91O`mzypnV=pgL@bVvwC6pyGDFX8aR^{zf{
zulb-pT3sm5+_G|V#^Il|cKv!Iz=b$tKpgc94P)S-0eLUt`kPm;-~|#7yb5`8tEH%p
z7c<vWSQ7yhsMx({kCN-?VgKWCvOT|cjpl)D6ZE{@TKAH`Y1g;a4#Q-{@xdUMU3Go^
zD4Jw0VwP1X-KzxD=AWJS3kWpy6~BZ15~*xj8xr>WcXV`sT0b;bnvyaEq`?6u&cH><
z1W;UlK!$;{Y(H>u9~ajnED`*f-nBWqxh)cdDTw&h^Bc|+M12X+rLW>p7P{hY0P{eg
z1aDhRK?A{1iT0HDSwXI)%lgv!3BZBNEb@O*#l+b(D4vAk_Ss8{1hJ<l90n+8RRK$B
z&spdBqnQ@xt|6ttA!e~UA|>>8ZO`dcE;x~^xei|g@CVnkI`gf-oSQ06ipbK|&E}(>
z2*}CfXxg)MU)eaL8<#^B*0ky%6Q7>x<p|+=@=H_-`<jkZN+Bk`G~KFOzgPzH2T%!g
zq`cix^-DhxISE{uj-4yqst!|f(msuC-W4UHN{8l)ZWPEV_1rGvWj8L3{xNC`{(K!K
zXlM0f7}YL|mnkOUiey1_{B4qvot+)*J|yKwQkfyt?ObHf$Ba*gV*1(wt059l4$qXt
z_b#2S9(EO-LvQJ9MeF|3dQ{A4n8CI)IpRdqVm(b1@j#m|UA(x}svkQ#QTGLyUyKbf
z2ha6+r3mPk@$vIc)?II*#~2tG7?qjc!NqkJMRSvG_Xv7D?2!Sk-Vz^>vyhRhl?Dyx
zHasEf&8j<tV`JNynM<Kje8czT008<M_n=ton%Rh$4oqdwh%o?uc$X4HDp~O8J#-kN
z!(<#-d9AaE6(>L^l)Y82Uq?Ef*|KrtJ{a5rBZHX@A84#P`K^n-c?pQ-CnJRik(qNb
z)w)t4aBOk?jl>wcs!i|$z<H3BGD%KSpsY|8V#BYer4=!5hK?3UA!4P3@CKu!6aJC|
z8>={~?z~<K`)~IMo<vTByXI8+<<g3ZiePIzCp%DNY!++yCOtY9lT*v9J6dQblymKe
zh?^=ERn{XE?kE^Kp=Cu(0Hy_n3|1E>fv8)M$^ZqcG^d4VL=OBUg6}g>SD-dgq(ezr
z3biBR2yVX1*m89NfY#69c#yOKYXjDkfIoP8{h0&P=pJtP(~eV&^6_qP@B~!zfYMRl
zpwZ$01jXQn6Xhzq-vjq#Wnch5k_N7Z%=pE!EL*k=-W{W8Z$k*;Tl*5bS|b%6TF($z
zpgF{-z7f3TE<s*7xeB;+;3Zjsh{b7Jy>4q?^-o43@syPO1yr!?9ueWod6fhID}veK
zokpbWwW6t-=WxdnatI=RL&NvDI-&ry;xU0_<7MfBBE~r<*kI3-;`K@V>ByP!<?Zba
zFpCXjCJe4XTj}fTgPdQBFB%$B(&H+@vRUP5<Sd&%YVHhQ8~)lv51cXFiI*oTko}TA
zi`dwCKw*%Qva_?BnDkz~dKKMpqTuYw@|C%W+#AbfXZq&i{&f<oin@GS1w=<P+T&_J
z^0gxt5nj5#(ucfid<X<;prW<)^_eK|LH80xB&7fNR!>y-)4mq`2ps+DQdO!~v+4~8
zkOIikh!Yr6jc1S=KxPZ;L#Q?(8A?VDfJ%4-P&P_T)f3hNM<Exo&bGTv0#@?lmuR!f
z`k@!{`wsO>_mT$vW-sQbHe6j{bZ-a$z-&=Mk``E;niS6yWHb>G0<fHH-O&J?albIt
zIlY!arH@j&1>C+PF=AC54e?J6AJ|r}dXjOW)z~J+)HLZr^0zg7FNq2+bOj;*FFqyX
zNkA$pzaEIfJI!7^X<}kh8y?MPf-sI87FQCwFmkRa9XaAXwXlWC_Byg59;OKHZpRk_
z>4t^v>ul8oh=@0yZG38qOtP6GHpe&{Nm>U4QsU$Lu)|vg!K9x=7e<QKfuR2D&jbXG
zAkRxZR0F7i!FT1=y{2r5C6kWsc^pjNmeN`xoFc^n)gVc{t;bhqVwN6YY`xTrhV|IJ
z(-ou0zEsg5^Wq}g%T<?*1j;Vs<3N)z9I=jFmUI~@zh3d20a_rP(vkz6;k2&$gK5vs
z{!@cc4c<V97~IrO&fmju6l_LbhCAuL{{GFLGZPcOfq^bgPMZ6+zd4T;_7^~^TMQl9
z&qff*%I)f_sJel?WH?@M&*T65xb@Fh5Be-Mwq7j-RlU@0TB-20x0Uo)fbyeRJ+M!p
zW5kg%h$Dx%cLI&iC2U}YdH>gSG+q)z?;FM2R=yQaLKv+pGgaRo!|};a*OxZQwaM!W
zMj)G)zLt8J7r!`o%|(B%faoI4S#lrkx%E`z=6!xT+b^`CrF+}o-XHSU|I_hIR^k6`
z+49ZeJ)!?<0m#q8x@$lTd3ACaaujN6>LmevC~4Q}oNQ{+P+W^HkN40Zs7bAq6axvt
zSAh0oG#T&^`Sp1L^g>(1yJ|_@N>%jg|5UxY)O9WS1bYuRjQ&#Fy$xDzPy!cSQ37Uz
z5+lqHqV*DDroCs|@|f$<ib?A5a>my$cJ$!EQq^r*1p@PzOFg^vFaA%T<o|oU_I-=F
zZgI=hx6QqL!=$%NL2D!5u^dxTzucAU{c;sKU6<u@QBo*Jl3sjL3>3?~B<7Lr!GDNP
zXof{>#-9Gtq1+>{Z)&Vsxksh@%c{jEjUQyb@O3i{&bb#hK)`yI$qlV9`$foWpgQ*7
z91C_YHF^E#xBq{*CZ&ba<h{qga$orK8-~LLS|!e3?<v**-inEhO&)$yRtAc=K{S>$
zMlD1HYNQsplBP&Zrvyvz;K2hxmXEB@!v-CSynTO7Wu*RkkeFIImOe%L$7P#?rdKXo
z;@6~h^sfsG!_@cQUajNq9v*0e?9$3a8xR%C-!`C}<z7o|Rreo~j^@s;$yO3KJ|E7Y
z#{WF0v9+~TE!^x+&0R#E=&&vN?(y4FJ9q7Zl*8P77p_F)yID%J<d1Kk7RRMeMKxjx
zJ3QiCV(XncgkSHI_`BmXs(!f`E&bT%@BcA5X+3iEC?U!E95_*9Kf_{JJt3WtmZpEd
zReCDK0<dwm9RI4SKw6Gg{cSliJ-+&JrJbD}>HY5_whaH?4SJf$w+9_dmvO@z^br7A
zuJ5EtD}G61OGRZvbl;%l)=LI6kgM2~`R^U87rBgj0<M364<eJuk-fV5pJ&d%nu0O$
zh;2f`j^G}Z3U5QIt5W~|F=$HAYi%R+I%E#v)z|+#pQlIi>)4nWBot@Mm!5C<pY<+B
z+6lpt@te1(5O_}&W$4e5wM+T;-{EHwIyhJOuZPXOzyFMf(OESJy8|jd%KKRI{<Qx6
z11X@jfCqM_`*?eIba$VMC(n2AAJY@RRi%d8+1L!t$qWzu_@z%q<RJKZtq?p?2Dl&H
zNnOHA-;(BUyQG>S;G#ST>J50;ZQb|0E!=_haOUtVS6ljLK7oP9LA(@dy}Q7wf!}%V
zZ5yhbTsTMmkZSP@Y6>KBj51n)*-*V1l#s`v`fF@Nzl*3XCnh$@ywW1ipttIuM^QmC
zeWX^2xaB<O{nR#wJye1=cWdjk!b|wDI}(w(i;G8%oGeE#BJi|;+qZLcvhf`X#%E@-
z(4UOI=458)9PdrDcC92b-lczrKOYxgOG^th;{MWKP$$jaOJBSHo5#%NzYjrv|3AO|
zcX<x}UtW`%%pS7Gsk(qVk(NEw*ifT>`0xQCu%V#=QfkzeXvPL+4MgoJ_BKpCASgOF
z+ED%ikX--w@4XN(f`%Z$rM|u%H4=q_agH)#Vq#oepHac04WJj;9E^1frAc3$f1WyT
zSlieNs=X*y|7!~s!1H7)BoP2rF%VwcXc$OyFZPpOk*&Ul#=X+H8CTF{#Kjn8Z6|&l
z%Ji>}$}kfKlmW<R2CYjw_hNT<)zm0Wslgxz-Ng99WH`*9!xton&dyW7ict|cgOq^a
zYJN42K?W}%0EE%Du~l`7)bU4%9av%S6-1cbYu4N%>BZTzEaQ9P!opy3AZK%@uYMy7
zi?tq47@A>Gr0neWnx8E+8^urd0JsQ7IOll?KBekTKEAq~M$f04^p+erc#QFloq#4n
zo)jiv)OEFa+qP|3G`mUTL$NUD08t5`tFEOyB%~SRaX_~?F8pyPE;&Y|!%*sPT0_4J
zl5#946xij3(${xZ4^9%15<<w?XMQkB0|^L_!5)SwZM1l=2Wgjr=fMm80|QQd&zGSa
z#Kvu<)E=}Q08*^;WPqhHur-2^3q(ZE8Q=$PR9L#nco^S?^FIOc&cA;OeL(<E2xo#z
z$OZD%^XCCTfQ_Mm@IkT+S&VJXhw`w7I5|1!-vP@)#Lz>b5X0Stg$1x<l)Ub9$kCx^
zH@#Ml=m}55w&hGPBD=cQMzDeIxm3q#daA1*KLgICOm=$08V1^H-P*4R(YW16bBRVO
zOjGh+m(45geScb1Ow0w>I-Vvh?K6xX;QyQ&!L@t$k>Mk${EC+#G}JpJ5b@;66Vgx(
zh3g^)4TXe;K9upMUB6yzKA*gvYy676_Qa<JJ0KhBJlNxKDVFWqKL9}kR%~O%f%)2i
zK!|={#sjodGsLK{w$1}Cf!8L34+>}syB#M(U0odkcmaH<r>E!ic3&O*NgT|Jz=ES;
zH9?IB-MKW}SKA?wFJNj?ht?F(4-VUJ7_z>9;&w8v3x5;ukkh5*egk?rIJvl>zH<Xw
z3yms-2eTN~0a<JEaATy%8-qqTQ6BU?J9<EtYGmp1e(e{YI)v9=l%Fp$I*2I}xC`6x
z(srJ?iRrpGy}gI>DlBkG967ZW@mZA%<{NL(K#-huFE*C6+L%S-yyHHuoP|ViZ=M`5
zwBqqsL<K(XTw^YbJ@V<3Ii<lyfBUoQ&Y?u$6Td#brcM1qGz&M^W1HP)$fx|*V%D{_
z6@eZ|yoCoR(|Q3eX%c^Fdmu+p8bat#;gs%0j~5Ql0&HWeY%h1=SAa!Vu3oLIO%6F4
zdEobup}qiY3JB5K5y+6<wdr)bG2=W?#LfeFavsThgyvu@yWjS=H8nyR$M=9nfoKDR
z{H82%7=a_JUQ)o@wfvi!<p=}cf*{gwdl%?ONC=L*aFiOB1Gm8euxva;i`m`tU}^w*
ziu5HS211QNI_HALfFeR_=@xjH^2!b1AVS^a<Wxw4wDeNccmg8d8a+~FrZ{dXKgf%Z
zNBV?@@QpuU)O(1@Cj5MZzxhT+1}GejTUQq6-L@DH5ljU=?(3wB@!Ap4f|WUKCLq7r
zFD*Tyr(a@Fi*R6EetA#Eo+uu{-4Srw7~RW9#9%?`RkW}+QnH(v1z^P7_Q(|w`rW86
z4a1l5vltbZ4sVVy;5>?BvW7ccbTo}P(YB2vRaF4NF*~up(wg6d6Z#A|dwK*9y&q`u
z<9Z_S76WCN7`Srf%91BfXlvISQxv#DFnRb~q8tZlqb7W~M7jc$+aO33Hb4^=IpBvM
zdY7n1khTa53JOvt5u8{LTr#+knK=riFgeJ+s~{EI5!fUT;x0}z&QDluEWg1IAXdV*
zRuM3V2lj(yuSLXw<seZdrTxanunC3yN?8v#ioSKy35xG9`h6r_hC^kI78X&7<IVAV
z9IMT@K9EZHq_h;^RKw+Vgvx*N$*W;u$o&f5YwdTARAq$pNbC#I<g@NGeU6xpEC&y<
zje$W7^L6n<eeExN9U78!{B+DS`$1Gx7wBIg(sB^<C8)gzVTNvWjmRmue;-Wxgmp;F
zpHyfb#Qlf_l0F=L^q?a>FA0Ry7#bQrbk*wtVu0jxy`|<+0QVZV9x^tQjC%h+JE88y
zi$ewY1U*_2xB)lv%fVg^ni=mwxWh&QAT~a3B?MP9F!$E<%618mqVM6OH@cGyvx3T4
z6$PeCHTCsI#rCmTJ=(B`bA>$I?5qh_2nc0dKP^|qUAW75E_0wiG2fU4FXo?oScj`H
z#2oen$6cuVmG}4*@MYy){_RNl+6};WQ*0DJ=l=cf09-340sZ>Y89Li9Ahg46CRv>a
z`|s!XY;_(f1y8`7M9eYss{)t*g_&>qfI<5c@GdX`P7J(EulO`H@n_m*5EkNhE!{p>
z!>G5s@-Z|rI_}~NMA$E!2+M{;al!$4tLfR>*z7I5|0D$g`19w_G__&M7Rt(gIfl*;
zTrja4Nv_Wzi;@$xBDhzm)0uNhEfEwEzH}d-*w&wb?N)5lK8ALM*`m~Jf;?`6A;Z3(
zn5sh+|L?|3_09lmVEm&zSyEE+_UA_!*aSD8f|y^wCeBbC(fZhveFbw(Qsn`nR~}B?
z(~+1u4ihty0;hSww(NPE&JXDZjK;*RoTzECn2sz75dUT)uLDw2Whkc-<@Vi=36I;0
zXm56n4r?C&7r=S`)*d{q0q$cMcQ{<0!ztLvFX*V4ueB;wMqR<GlSkV*yeyRDPO$aS
zTY&LF>URIEq0QU38>^|s&-r-3_H4ZG5Jd>o(hu;6d8%|qTp@*4&58vFpa%6G`HT!E
z{@CBrYOnw8+jpgQM=`ZUv?K~C*lX_5K9qJy9T9Q(u5qVb<SXwf5DkncOks<D2#Y8h
zgv--Et4?4LJu}!D{zH$liJBN^(bhIY)<`@C9%HzK6gCj;kgz8{*FV+H)i9&-%b+e#
zOsNs%z9q%Xd=6J(;*Ylg!Hl4@=I2&6l7eWqT7da10O!hUq;nX=N8Vy=c$OmwkNfwx
zZtIRivjl2$@7l_0^E|BSs+B8;rsu8&6WKIML|?dT4c&cRhF3|&yi__oABNBydi32M
z#{?p>PZi??HdL!XpmVI?DT0i~V^moeFI}RorKO?4Fl1|UuXQ2e;YH}ZH2Y0_NY0#T
z#$oQ~*e1agRyzm-S4f^V%ambq1`=DVNF4`9<SUr%K`#>;5mAin$A6mC$in|*Q90;q
zLANAD9#<C}!jpp{opCvWz@rw)!F91ElkkEnV$=Bl{$n%lU<}ZhYb14sF41ZN7y6o&
zD`Czp%`%9T0S8JjuO8h;-*BIR0<V-;PM(BNmrdksK>0}H)+D5fU}_4Ai<xA~pc0s2
zLTJm`_xC_ela?YVAT^HOqT5!6%NF0m#_Gk37yU%!fpY>%$jHb@EJ&Gu1u-VXg-25>
zH#?LBt<|9^`jIT2&7kNR7?RcXz^Ec}((J|SF&KVB&7iB>j)O=<1s8cFB;3gS(Ju{R
zCB?HKD`vql<KL-6+)JUs>xNP#SI-OUqjt=Gl<az{yEW!{p#1OzMDxfX(}iE)Q`llq
zJn`m%&>H=lC<M~eZO|Txu6XiLNUw0=sE85&Fm%J<>grEGK7`lo3=bVRa8%dS5RDx=
z6VW^>SXTe&a004=$QNfseX*tZ4GNAWy{oNNNjj3Mb@SFO8;7^GwIkqVG5htw_&cbh
z5zy?$de?n&bcd84c95in5t#rMhQ@2Ty2OQ#M^dNV@${yO%2b@BRX$^W2FFSi+%~%E
zTYqhsMR=iQ-9SSlTy9~uV33T|DDg1{(WnlGA_1WpVJh8~f{%n!BiM*|aRs0J1xh2(
zQ!(@{npac`7iXt!2D){BqoVOAu53Bs=;qmSWh|-0Y^hL=FM)i7@OUW0_p`w5_;Jvj
zopPzibXp5nOn7Ka%u~FCo$1W`J3gKTq=Fiz$W;&ztjzOwV+uLqo(H;whlxs>AQozp
z=bz^|q0fqpgQ}}jZFa5Mr|_s6z3-%_qSm?UJ*abZ!%{F~VCP?lnaIDFlV1eh7uJVx
z*?7gPV;(F~gz^-5z+!+IaAeOOI&{t04sExM?d?BtORk=J!*=4qf!`1t@-vWf0EBw(
z!u>}45nOk$oM~FTnLPjmuHO6WPNnt@<I~ZV-cNL~7#s(oMbJ4Jc+|jj2}vW`j6<|y
zk=!9Mga5HzZ@E!~`7DHM7|+>^&x2&t=#s+~<zz<$Dg$0iIfFkI=H?hTlL|TuyP@Rb
zL}K9+yg;?4f2-S%=prLt`gqmgX|c0N?-gpu8K_TPU5ko~8?i_jV~^H$gb}3tXgOoU
z`8b2uXF^9!JpqSPaH^nUp`QPQYif9yH=-8(W@sIOeI{{Q!tG$)Dc4+vhJI-@cO!gJ
zZ-_;F!?_Rte$N<j3#|qC!rcSaEWS_^VCPRE#U(NGnTME!M0=Mf!p-x=-2M*-<IVpA
zu>4=fM*f3bk(f?(@ZS5bZB%McSFTz`Bu-t=@Out-;8BOExvF}`I26c&^~wqQST+9R
z2l@eXTm?<7kdyqmgIUTjwbgrq)89ARKQ<;{FxJ<1umv~z2g#x$zVGUrC(>|Im%euV
zwaHuh?|;Bp^56gY?WGl=qFP!LbCi2Izq6s0wsCv>YCp=M`bC41^9hC0tNH>ZAOM?|
zVC3e0jP%OJjV5Q~Y1WaQMaxX-UWw-jEKwo{eZ%A8h|_Xf1d7{1H2f9mc$pdiPm+Z(
zBks9fyLJ(o4D=ykVcL&WBT)%?z}Es9543!^0J8P-AC=cbfOa04A4+ai14MRdQhph#
zL^7cWg%7kAh2Uq;vIGlwo~c%(+b=7-0J0wCF|g(a05XW2kpGR2jv~H(M;bOV(v5uk
z`SXKEj~*QzRgDz30F;Ic$7c~UYT1wi5UiVJ^b`j#?=TXH0|4hCzCjL;?*e{C4{{M;
z9>lE<-yiu*kt_Q(iLJTyr|Pug)P_+=mx5b>{Duq?_}cs!6%{&S04`w@PmhnMfW`sx
zfinDegY@_N4u%Gy&pB+6>%E3v7e(hq5FL=)JbXw1L6jKcmhWJvE%?`rn{i}a?)nVX
ze}KcGOG~cxo?lVCr3o;xvDuE*9%6#7MfVvG6ufO?Cu%k`GCrE<JrDd!BP@bXf9I?m
zP$YO2AF9<xQo?n>zrE`+ZVPe)!8FgoAzS4@8xlBN5?`AfuUuKc0~9<Cs14mCg3cV+
zSU7Fa+$Jx6r5eiY2O{TAAg_0}A|r?3hnbJ*NFQpjV1@&)@C?yYR_|)irf}q_dbY6)
z?a$picYZ`o(X{ZstFto@;}yll+bTV8SZ(Uo3m(f|f2}^ck}Xb6Iaydhz#c@HW9f+>
z(y$~(yT1eNx*|1dw!C-1Q84FJla-Zqbe>G<M!yQ4eaiX|04}6MiR-Z&bZdQI?=HIF
z>dH8J8BL^v(kuMYX%njUkdkSGf-}wi<pzg;wE*1x;Q!mk=uE^i6%*88I-I)5&IhLi
zBt}}EsSeS?k+yJDG7xxGe4JL(E2%lyPyiM)?H|4&&4W@X`^*g@#>SY3x{G1!I_c#}
zA8`44;vR`&MTHw}MUO5hBUP^-H>XdKd`N+_oFbZ5=A1@8eIkR5{avz4R;^ymV=up>
zD3|fsY1ZY@5<dH76F|cO(w?>^1M(nRGpi2IbsSV)vt~`t6OSLCkN2>8!N4^9$yC8(
ziocSrJj*ll({|upGPn*L_=62M%t>%(lk%s_j$@RSu#{BSk@@Qkp9Tk))T|9L!D$$M
z*@?}G^&IIi#gKd8EogZ>-g2n<Om&riKmhUD!)Hw9MSDXDD$Zd}JjQPvU)9!Y5(<E*
zO-=i7RJAyvK*NC6Dx^!09x~9p#yJUMm&4(<L~37212i9fRr44)ZkH^u!hu)wwfIVL
z2dU9GH5>{;hq-Bwz(Ei_PJwbc9|!DMCb3~)Yyl9nu4MrlIOHtm9i@mTJ4x=TcNEFs
z`6rp^s$gj|fL$<-dEgj*2KsRO*cX4@YM~&cj6`37W6%s7<Sm>l>TYweeg}ktA*=5+
z4*kl?Lr3uz*S^4U=qs^VipO&e1>(?3kAQ<IrQb$IPNAHH`!KYPC@c7lpZ>z%0`Ryu
zLmz@|7M3#1ZWwW4yZG|{z9NnscUsHe-WWI*X{O(Tj{!KGwOvreL;5B$y1|Qx&YmZ;
z-htFs6&$o(VMAKJuDW2X^G_W0VNPUzSl-+h>xnk{V#uzHbkv2&?4X%mMEF2s41~i-
ztDgd~Lc4)l0ny{bX=f*21+Z?I1N#i(MF;u$p?%9QD2TwDKK{BSEI|WTwc&`{y3Rh1
zbRfT-Ks*JEpWeTFH(regW*z`T43m`^C~pCl6N6oAUP(#4fpRjtF+2q}JZ%~3cCuU%
zwcub5m{od{;#dbx3cW~`XD0>8H2dIermKe^1=_8(_2Uk?e0UVC!g{EZ$fklb=f1sr
zNAM;PZrz-mw&--DYb03yovts-BznmeyuQ1+y27RA8;p0iY~GAg4<XYX1tuxO;JTkZ
z1IWhuPr#bfta^ckx?#&qw`By{8)dN5Bx`iSBr$ODD&APj6~A!KBeY7H>#8>=G7>*U
zIILU9)c}dUdv_y+3A)Rdm9wZ)E29wIn^VMLv1C?=xF2=#;_0n<cgy7k3~!3zb6Y>o
z=;+XMaB!fk#^ltb<aBJF))YDSNpsIPHtY6g1wIH5e^egzih!s_4KNm0GYRco-miRt
zcHKI)_cssPo?)1+BHNlK#Otu{7#rSpg^iXp+ITFZanXs-?^x8OW^+s+fKoP$qM5dA
z+2!&U6bCxPpcMDBJpixVGct^n!S=%JK;!D+-1Rm|5vs)Cr>X>00Y)=@07eeNNm^RE
z0<HO|JH$*ki2`p0%!;y-xIkgQ2?z)v8vW_N#g6Y<N3m0ZAB;ZeEY#XhpFJa`1vv7*
z_lYUm>TqQM&S*;)+Z8uXz)cMQ;Nhw6?(Rk%251xvJAMoEz$&4So!%CuWGsdsh<^v2
zAFR0_0=5BL2gVEX<OO*5Brn(bp`@yc*4KDNGyKQ^+Zp^WVxLIfCKrM<q6ZD|0iMLj
z&=6ilaAGCa+BjW}(Gr5f!cEmP*b8I=BcLbHc+yf*NCcqN^hD6c-7Hmvxf4FXQdh5B
ziQcO2*49?g2t7Sy#sK831Tn@k1bj;{L-%?Gx@7UDm9=y7^YJ+!3?%>=3A~3Zwbg@u
zm>;0DO42x96WYkzd7h#dYgx3}us;XwS26QQRfr_>5yk+e^)xn~oXSAo3GxlYsdqRX
z1<qpV|4A~(j)K)n?H-3fl-8|t2LbG0u&cCDLmudvUIRNIZ0JN3)nz(<twZtXEAg`y
zK^;|}t_22kRPEqe^xnMirLeGYJq=A0&mT~XSP|#_J)e~O?pg?UA}F9sJVAYEbkq|Z
z8l^XjM$Gb^DI|)Zc?9$3al#(JDQ3EnG$8dA-@eDz{X8lo;>8-%<FBDXuUeUijt+QY
z9D8;L_^c>8H-{Zrj$@DuF<(s_1%-!WXJeIyckV-cJ05xlpvc!0OV70^MO%e{LjwYT
z`xWX0%uCdsn>n79l|?4xp(SWiU-&b%E=B>+Z5zhKo%6}7Tn-&Ngmf&_`h|CP)LV2P
z0_#XjR)Hy0?x%01KufHg<FSzx9%z4h$RDZt8Q~Q;<Aua!Aw|fk5^9Y)KwwJ3rY77K
zUQgcVURPBy(mh5(mRLXg>lXpq=JF@dK&b4QtDcJSo~XzfXcU4QGY4_lJG%T-ErFrm
zR;(wOQu-21*@^CswR_PbLisj8Stj2mY~I64Kkq67aXbGx_{dWChTws1_1M>$zU~|X
z0jy?}HBMVJHVedAN<vgrOYeojwPm@1G`aQB+q?CIbJRC?-;XqoQv;RSoO0I2COSBn
zBSRl{)PYe*=g=f9v3S*QW@f!RINi#ASSB>@4m5#fwx>cNE=XCV$bcU!_YMoQ_TXUW
zQ&hMxJBVkDp5`@o+~q`rRxKXf*|=ESl2dVmAVw(F?+2oUUHbtLBC=-(Q<KbdFt~?*
zO}p%V0-K|$bGTyBUwD%Gj=w)GNNt1~jS<7EkgFg93(LwHUpi;b<<S}rujK^!HKVmQ
z*f4K<dlwNx5|5(#xBg+m)%koJ5^5Gs7B|V^6T6h$kWS;O+S-BF#1#UcV}ueuD9OQg
zE9j8x9QzMDCUc&Fi3KnM#Nx?}&bWE3!C4|(8lu1}2YG-qpg4FzbY<U*iX!f8=+#Ak
zz}Hq8+KW>|>bSfW6Jld~F#R}Q9SN^O|5uHWc!gTH%b2UN9}nPQ+Pp0<2-2h@OlM$V
zQV^cctCKM_Fp#9qbD7Cw4$KIdhT?MvMkPC9rXI(7i1#DLaDBT^y{UU+w`J;&n^odm
z8*I&Ai8Cg&b8DgRK};F>p?CT%_Wh1WegzQo>O?v9Qy;n|J|T!&g`y^T{cFy`LReo>
zu?oupz(gC8X1-tphNDcJ;zie782pR3Q@z`XIS_!@W@cu9wfNq?eGC?wZ{HH$6T~s5
z(CX@O^YMMd?o<#L!w6AeS{+?oNU0AX<wf{Ju7g#_x)r3QBAH_NQs1jUfID<^`rO&Z
z#&%d<9!8b1(BY&1_`!n(Nc{2FJi&l>h4Aq{@~A|&NG>|~D;8$=BC$+RgFU8#q9U2b
z4&i|h13FCSkk2r7vJaR*OE71LhAW?51(h&fM2>i*2vu+1V8p;~PR?xyE~+uGaKQ^2
zBsPXeRvR;aM-Iipsc=H}UK4sF0<~9Rq(jE`hr`M6Ho;S&Tq=vZ7Z)dom=A<(=*nAs
zOnjQOXAKCFRd?70`4AG|4i_o#qg`v~&Yi&1&Sb7KJ_3h8pnrENCsAkV>gonoqAz%@
z4e+W{1%rfc*W>BB?jvN*K;ak>5Es|c(<|{#XRLI)Pui{D-;#KZP_K0dh#qtc(^ABv
zKF!?;_#q|`iCcV8_3luNXuECqA)}+JU@5F#cV6@+#I?6m7M5uQGezQH8bRap5lI<-
zGW~}mr%~LrkDRu5XucbV)DuI}11nKLK#CG0?PiZ9!%bH1@9+QYnJ2nSz-yz^i}#O-
zASYfv{zzwD1Pzk702N(J%&Ww6gl$gvtTB~wc$X+399f6C1yc)r)AbpkP%>3Cu~Q&m
zc%0LP{i@yj>f7ij%GbRaXYR4a`N&8tob4vYPU03eHkUrWqE59&Hy-S>aUE{nF5JIm
zJ1o~R-GN`a4(~a&@%6^h9BPw7z{;uinHV3P)^heLP!NqC1{iIdnys1^6vwEbyDYIm
zN7_3GR3FUkcclVtNPmDKiqVjFbtQ4^r>y_kdFKM8uVyzAC1gjcA)A2@Tha@psjO`V
zkV{Otht;9*7yyJwgd%MjkvX@}iR3;zl{(HxdEX0;_=q4!VTuhJ5+m1e2XTx@ns4)M
zC<C>jKsRHuI2%!=k0`PwE5wwm&7eiW^U4I0`E{0u=F6A$SLL+Pth;Teg;zG!QMLS9
z;JuD2jj4HWDU2m)$=j-d6~dJ7)BP8$Cc@Y2&wCnM0G=PdWKT<)W-zkjQY|ic4Y}9N
z&6?8{&-)aNLfeAu^jeb5Ax*scv7xGp@IIi?E9J4>b@=z90yj1~n+jX=VwAF#lL)M-
zvNvG>=FML7I+EAt6NE%C08F(;)`aT~VXSbbE#NZQq)%<y1+ey#($W}9e|&oI3u%@3
zz#e%ay9*t#G!B{k+G#DQ8<QD*w!)^%y0;Xu63%4woqNxqjYnQj?~}O(KN?#e=6aBr
zfO0?6K^jcvDF7!qPm%keDHl$jYCQDVp=cZ3Z1mF!|FFy<jx(g@<_30x4t)Fe4Yp~p
zyIH#z=(P>s!;P%0(&rl3%<2)5p^O3>Ccw@81=|a!knyt_GS@Scc5j`9khZoO{lF<f
zaKdTNIbRb@CQcz3W8?{ir?IAHQ;(LiWP;MA(iWpnpP%2IM#v3S9)!YRH{S+o>P=($
zX@^R7+ZY%a!1>Zl#$6NUS}EJxL6~1<L~~WDN6h{Eme*w?m*uX{4IJY9Fw{|{wR8eC
z*8TFfM%{=s1vCB%ze~k8{RI-pNWuG8qTY2&m{TmSSH65%cydD4YvCybL^yvBc`@E1
zP4FJ>)+Z1(QN#wPwS~{H>JfYR*G4!KVC8nWXcf*!98Lh96EyP(8NMQ%yj+}d<)-O*
zAr}MZxY%HDYgk1Aykek3z4yrydcabkCKzP}91Hc&7#BithLoF07Hvo+Gsly;PQ>V4
zTyw$Rehe9?Ka00I3eAV=J0ofpB6M2RA@+Y0$okm$-rc+7u<{$~ZXkrY*Kf1{96;9*
zbVz9Iuy$}T!P`UD1J{D1B}dL+y~ffb8Zby4#{k`!(3YgY`Wh)WZVIl9N4cVJa5n;x
z7aaylX^7o`wpyev0yt$zFVNKV261U^MIy;h-;C*N^t;_b+TNb3dGw;z@D;m2)(`%C
zf;vAvMa*B~HX>$yEI#DK)mK<<j4j=BS<s_4Met>!=VRqP!*Wf?YnxVj3X7d)HQLzy
z^g?c6>T*`*+`#0mGt|h;G9efuhH(NBHmM*gQ6JyOHWtNjs+FE%!Nm^Mz>jXm-VX~C
zgOS#>Pm31z2<DHp$Q>jrGx$dRS!i9y>Nd!?4qiZ?o*e(t*6&3Ele?VI$!qYD;nb;a
z?5KkA3mE$`!yoO(07kuX4d^_W<w@j;ZISIaNI8222^o}4S}Mo!f<U#Tacv|5kgg}6
zhKIXQ22{SO<QWH9tlm=tfRASMID^tJ;nF8hSW7Rc!G~m6FyS2<>Jw$RXfiED+w(Bb
z8U7(eOGZakCbL>dIegOu^whttEz#=|IdO|$P%e49zywgj>_IY$gv!hqVz!^5k?sSu
zkoa(>t|W7=EK6G!X<IcKSk^_qI>)uHlL|?JIzK}u#0GffQ4sDhxtCcA*S^o!%oKp(
z%ydVNZN{uVl#5(LA*q|Xhpu-h&a6f@LcNrYn00V#!Cn|5IMP-OJ0P35$!slNQicOS
zBQ37^%H`D0!sNw2i_7{PCSyR&bTNkO;X-X98cAQMY^4tSuhr&ud2F44D3ZFJiTMx`
zK8;JOVCmO-roaBz>oz2n8o&CGbD+@fMcHjE5of)xe@oUGX#jR@8qN20k531x3q+6-
zuG%=YTn$86eD&KZwZ?Dso1fj$6GRgyMbA?bpQ*4Y1DL_Zgz_JE{DyRuQysTSuwcH+
zU27&WhJGf4&c%gDR29%F=_Zi|b&u6YVrP356Cl8%L<;M;0def>>%(P2%Ony!+kJQT
zL|913E(7~}%&}ar?oLl)749&LDC(6Xv4IV>k-<h!9sgFr^`GMxBz8>y2yh)f3F>!H
z^&Pk7t|H2++`w<^1MC9+W_M&gN$L6Fnb)uWH9eZez*n(@A#dR^Y^nD9(8fW#cvYKJ
z%|L%2CnqDHg@T|I(yB@+Agn~i?xAFE{s|{Yd(qj^5oVA{E#KCx_@7()vF9`0>BbZN
zqjy>So`xBFE;~vwYWIKq{P}Z~T{+_uy<Usk=;@I^Cc}G$4+2N{K+iUq>}%a}1>z~>
zE;C4x5xWpEK&lt3CzaoeoypFqdTM?0!GS%O??y!ELdW&nt8+eL)Qe%p8W6V6biL=7
z^6}570Y42yqcBRT+W8RoQH!VNK}snr6tWs#n-fjOu?OL?;Qv$y_m;O$BI|U@xnoFZ
zxX=dNFV9L1{D`qluz{9qtRq+c*D7OcPMXx$%n#qLKSim0ztu4qRBXlU@Lo!*F3IN&
zew7?cROzMvUHbb`ItG{x=bD9lYiV?JaS=NdryAIy$n;7)PeP<fNH`~W6DlF0I5kK#
znmb%IxI)}7Ud(_C9f()B$D@l1JT-$rHko(#?w#6zlzmC@rD~09nxNjn*8&+rqI?dB
zJA`q-?ES~2)2_4|VQ?elr!BqAunsk^YjQozDT1sURG8X(C4P^+54Cs;b$Nu76U$eo
zy|8(y4|<rWom%80Vf+XQN8omuG6>xubQx+&uEm^rE-^FB8xtpgp;|)1Fg!3If(?zr
z9grGk^y$}1823+cy5Kbo_ip{sO8+Nf*!XOq`1<IA6zTqujDhr|So{Q_$*n5jPkkg}
zl$Df~H#1r*$-q4cLkwwXT9|f)e<TtqzR9^FrJ0{wwms6Xhxy2OV+KG-+<mZ^BDg=Q
zRskvIa{vNZMy@~yw6>T&M5*GalFVXw-5wEh&s%()fOC*hB7n0Dii?ZGnu9)Td;GIK
zW@doGTpaq3KU|<Y+i`l^u#=zv`!}t;x9yy*@Iq$JR*5UF78gFv>LmPm3$Kunddc^6
ze}5HN>_W_pbe$uj7N#%&mynF(suZ=(g6}-?m8uPgOdRI#CVz;+5R;4v`MkiudZKX1
zaEB}K^c$#3)4Wg?jNrl(RX@3SvL?XJ)Cja%=ne&YUt{x~<sFW0ZgQyGai&52XHAG<
zp3x6fO;FiQ&de|xZEGq3oJm38#;F^*qKtAcJPAv^I(xU}SjEbIgDsC{9a(|W8LTRN
z2Re^fIo_)}er#&HD!<#ksJk;QH}yL^Iwbe+hs}`mzJ2FWTcU>BVzlctsCO35N_QpU
zNIoJ1;sp&<c^LBI>DJYJH#tUR%@bN$T5#N`&t=sNP9pr>*(R-%9pw9AgYo;eXF~fS
z8*4qYqD=&eb}43b^`8NSFx}?2;WX=M)@{!cQ0#B+s8-gqJ@bhh!|dj~iv`zhlhV2`
z>w4kBCV{v_EneLEsb1NE8k=wO6{bBID$`5!ft^a4n_S?}7_yr`m?q@XsU{$qNasxv
z*Jk961Wg0|!PU(Vt69rhzdwhNS?Jf{viHBN=<u<+mz-dSfstY|UNbPw=}uK>_wb#F
z>hiY~;scpQ&>FCnot?;*NaHmiW8i8UJoIVI8i^xXe!`!%HG3Mq4-O``+$!asURxY?
zQdzlaA?%bJ@<+|Z$Ivjtd{%(lui<%<u#)nWKCmgF-m*1Zku(OEg_TJm`t0{l#4Q%i
z#2V?;#UlWXw=gn>OtH)BY@*YD2?_@$)|`f8MF3<-a)PpSdenF5&c@20iN&e_1hns<
z375NY;Z7)$(6mR*aN9KT^xf(iDY$P(5l4~x$QjlfD@e}azIBm0dDcfGw#9Kg@e<V$
zMMcIQ_2xYs^5I>Ihn){!wKVIO5#kyPFSy6t{k#kfFg%C+{&*xk@ZV(rC$nsCezp5S
zvU0NE(10yP0+hraL<2w4$2RCHh?>a}n#tAQ>>npVPn9y(gR#p)K+>SXcvtRu3H|rz
zmQU?%OPli?#Ev97p37u>>va9?7V4kT09-<%tkSc=JG#O8=^2W_U1b@!_Iijos;UOG
z6vHyY6#qu7_tjp?cANn9>;sh*F!T{|X{3VnKxHMsRqM+u;NP=F=Lu}0ETa=O?W4tP
z&Z@0Beg~)qoqi8qEs9wTi%wR@{5MyOdk9?YliF;wqXd%N)uPj@kp%=<Cz?Qvc%CuA
z73VnZ4a{@sb$IB2N3G92E@W2TQPuZcQPK9yYOdejO*XSnhk97UW3HMw-h|7tHi#)R
zcTA-S@&RJ!So(fSCPeMuJeso_RdAv1?c2B2WEdZrYRQ!{l^zZ8`Eu0K?A*C_Y<NK4
z$kHF@?3R@+um1AF6c_;CWm5--MFa!O-)|G<G|lVM+dvpru7PqJTo7f#kMUiOnY>v)
zKG^Zr1#k{_c8gzj8Ate%M*HvOb6Gi;aZ@BtgW}-fgCr~SgpuemoMRswPg0HG)=+p5
z4RRRR7rH9n9551j2LQ|42cR^Ipb5s!ocA3Cy-FRj7TLYC=M6k1OS@)>S+$S1ybdrJ
zOBj`aBoTWHw!$6CPzw82XO7Fa_Mr0$7p__6vqLBnGCiJx@acK1g?lIcvt)NyR}hPs
zvfg9Km8)MNghqdSk&zo{wp6~fs7_OJN<D#iY9uj?Fn4EYKdUv}1S`6mk3hshK8gO5
zktclv?MxMxq9QtUn<nS=Q?-Z^-{%f44j+lmRQrwo7<zZ?*j;OG2-=XLbX`i6b*m9}
zQ`&$z-P3xWFGZU*>i465h?*JAN84W7nP}`t7(b+SvY}LM#lot=60|>dx>8AC^(FMB
z5Xy7K39aK}ni6nE>F*_mR!1tVS;SB35F7qx9z-=}$@B5y%u-O4VmYkfSPj<T2onWH
zCtCzsI^qwlxFKB%|5w5CUx%l<eacYaQ%gkO^W|6l3Y7%gMI@)lg~VtOZuiE}eFZ@d
z(}lpp%|THdDiqIZA24X`fD-!)dO?lHNxNks*4~`f;c}J!4KQy<*juk?H252a?J37O
zt5}?O*mm*{ys5Od#E|v~d>pgbj~lmc?55vYlq;2ACBAwMxxlu(78ee=a@hCET3k`e
zlBJTIoED(L%L>$%-{KQ_nR07AD;FPZonos(;<b3aFay@?xYOb}irQ(^hnu{Nsg&<x
zqr!`CF<7e1W7@iPLl>xkmX12*nToxm_Nd;|VgcvT@ArDu*{MHWQvKV6E8&Vm{1DwS
zu7s|wGK2*mZAHG;?70vslx+Z}0hKE)JH|;&7583go=9Ze5!3mwyztos$|H_HtO89y
zuDB13fbN8H6K53NDad{P;lhETcyD<DVk9|=)shR<h4y@8D)xFBC<+8q(z1gubkv+a
z{Ska4cm@&6cRNrR0kgh8uIslGX=)~%D4~3S@SkS=`X1mX$m1b)h{PxXDC3Zr;A+J}
zM1Y&wXtcc;HZGNw)LXdLRxqH+#TEKVKx8^R^mr5k0D+6=JD_{tLwx^D+j|_)QgDI2
zC{dQjM?M)V#c3EMFId^y0ebG7>J0}!ygGipoP*Pi_=aRmbaj6D{LHD=FIvz=Duk@c
z`?%@v8{9)z$I`u(v7`F7o$9NCI5iwwMn*1f?zy=+h-3lgiDN<vmIw^F110e_5>jW|
z7X?e~hCa$2Z~yzR-@L&%#Ti^EfJTAQf$u>t#hW|q?J1x8`r;RybU{hD=-;y>*ttFU
zQeEu?z$hBe_;YcsDrIvPQ#}<!5ZOKZU>|BFCliDDqXQt3DN<Nx-B!$5r8Gs&u3sS4
z^M_DXE~oS7ftAJQbm79)HhE6qxc8;IGC~afG9v%g0)&Tu#>64L$5;hmHCx&+A0m3z
zmM>D(6^y%?SnE!r1^@+RusX#bJzW|2q|(x-(ykv)08({h*bisDHN`ZI+i+gGACRiT
zDbj0J6(ma7lqU9FCWR22A_{58q9m|nHo}R+F$U`aVV)=*ncmwP2e`-5>c~;xm+9ZU
z(A3N59_}HbMKD$aedjasE{WNNW8JRQ$a{}>m>xWMHRmjZ(>b59C49tIEMG2~20ONs
ztACDTq!NR%uCszEl82opbjT>pR{zX_mH>DzgPm*LeWBH&9Y7c<ZK~4NWM{DE->^V9
zE0_|;ktlg*4ObDfGZN2<c;kr?D(G70gPy^0pn>Oq#4Xm<*E3{B!rXr%3^Mjije3K<
zxbDG|Jmt>~T(z8)Gq>K0c}w^GKU;?3jwe*|0{9dB2ltsf1sJ4iDIyTdGEE7CkXa$1
zYtCZi*LsF^n*B~ITzd60Hsl;2^vEGF43Gf8gk}=rW|NZh95ONlSAtK%Z_Edfb{>k4
zve}1T3P|QpUMvGkZp}UeOa(0_7=K-HabhP__82=4SuPr<aIu|)RseNI!Ow><u1l-7
z1^Nqy<;1aL`+m_4frVekBASAa7l#2isO`bpW<_mshYkmsB!-p&8!E_gBsz{_OCz|6
z4nOGy{sNT$GG0AUdx(62|KT<K9BKcK&kR?ZW>4cg!Py2-4rpI`g;7?QVTA*D5*xAt
zz#N!LC{LlvM7~&zU;iEGa~U}nxHS#lFhx0Lg&|mBR?$IrFO+D*!)gnzpvysy38)QU
zfBkDI<1FmA)iEjANE;aAEV7T+0s;kW_5JCOpdkuwd!VTr*{kt>o&Y(~bgnntnrFc#
z<DoTRHwqx|Hw+T_JJ*cB(Mh*{02VsiB<u}Iw@l*$Pe4)lG~jd4>;SQWu&5|HiObL%
z$&+DaV>6fSHNO|VZ)u%N0Ua?wS=IK8p6RY#MSH^GoW~dD)6($a!xaE=iP7g7ogCiY
z#f^kQUfDB35if)wSM@EN>X3BMjhvlWrl(R96SzP+BOG=OgN2KWh~Rc9h2Si8m7O*L
zQLmgT7z*G|*4PhZn?G<Zw~$qWK}W$cEChY*a;r-2%alX~1=l2Ws|@$17BqGUFKBD>
zhT($KuJ3O0IWHSrbIun1GX1Axk)*a=@iPu70+fYFTDVSgxUiwF?hghxJA0J@aXs4*
zvq$QvqSa02Q($)!)B;Emk-Y^4?Cj@g2k5`NGZ{iUhmfJ5wru*~2yAPNHTL(fUu(Z?
zkFsi6*ST9>yw}#*Z0AJ`jlH@4{XU(zz$`6Zuu8W!uxHI_4uAUe&|-U;pZ4+<r&(QI
ziA(W5$PKh0CnivpYwuaje7FoEFPiQV&qvA)%|4S|=1RIwwj`ZSC@;?EyY0cnbWBA>
z<lEuZjf%w{wl8g`?)kzYPx&sWy*SAWhpp|C3<cPcav_Tu{r0V(;5=-+QteJ5;=u<k
ze&G8!ax$CV@@E2K$1&3qlB~-2nmaJInfxi|FnaAe;AJJswb#n*PdZmf4d$shqc1eo
z<c^lDogHEfNG;n>EU%<CkM-w+OK3I*HP-U-rF?I`a)baH--mHu(Mnj+6CfDpj{A6&
zd5zT}&><*qiL<EywT^^DF(M!7Sc6B|*w~nq@!>g>P%8kT_Fy&v&ulajJY|IagG&q%
zKx4G`1-|z@@FHA<*hp_ZreB>-I$Ir%i4+&WkWX(0*a$_CK?$o4cSJ2>1W5-0iH(Y3
z34o`APg}wN<DGWKcHCd!2`k!8G5&LZnF~i+oMAPB?z;#01R5h)Y88Fa`$$VihkUch
zSj9WyIY+Ik=rFEuBMjNDcH4@vx_DHgwYhn`0II+B84B$uJ*X~96M-W(H(oUz?sSmw
zzP`b=gE){U0Aiz+6SHzP&1n%)kO<7;{b07k1)Meb1>!T2203JZfGsews3n&hZK1Ll
z7)s&;Y{d~qa2U^-n{Rdf{sU!A*1Rs7ixEsf*rR7^5LCg(j~WQnQ03#f5nmdwR;?dL
z1HhWqtKnWY2_^`7gu3eL^G;5{e!I~6aR9l;in71288UNcgfd)J=(YZd+so&q7W8yj
zZ1gqa4_iQGqlZoZv&z{71Vi+^zz0rE+YyhCDaM#Q5H-9RPrLvR-KnUnshioz*JVUr
zfVvBa;7E|bNHkPe^XO*90INXOiqf(*S%qoKjE)Jm41jP?Ev*)mxTq-Ike<N*2^Bp;
zGSVxkKj1H%NC*vS*Omn%g?UEiFfu;3FUS+d{TMM!$z1gF^@VvNrqpyHxZ(|BG08Xi
z{CQ@v<H=9S=#f7g`p#`#J#b>}UJW}1CiTOc6Yi<-L^yXh1?r7D*cAJBy)!C2WtXXK
zbBbe+@}5DBds@z|3Z)OLW6i4U-hcl2=78(=e5Zz=_va#OfBbI$by=>r@#nnExw_wV
zHs$zF?=2@Jzdj5BpVfiA_|4?ICZJjTm$6H$HgKsKRt)06N|3>3_;{#;(oYqS`Eoz#
z062vuWt-TX_D9C%aZCmN&)9u14B4E8eJMO2??{i8+^65-gj?ST9(8yV0HDE&Jw@RQ
z-IzNNCM;!|=Hmu@H;hPRLN^WnQLj6`zh6Cn|7hd=Zgb5;jdqj5xzv%}pUmtv527*r
z2W-;hW;zJ!{|nw=L|P4k43HZ52Mhrt=<-o4>z>Ckj&1tqj6A+3Q@n&DDmCdC#`WVa
zPu5^l6FX93)(!_%<gx;usaa&7;N}Fb5R23dGNGBp>-X=}+1-xRGojn?N^u>N%O3O2
zisXpIq^YMik?e-y_Rufo3Z%8Te+oXD_mmcp*_8}`-t)=h87lpI9!q2XP=3rWfaWx%
zHxqMGRFpJ4WI0vtHT}qn?51CpJiaTN{(+-`mCQ&%qEK;eipp1Ww71-j#3njd%HW9*
zjJhZ{?XhzdbgG}dKze_+?;QD<O#i@T&w)WTo+(h~SJ0Rdc?x4P*cNKJ4_*-}-b16@
zf7F=#xs{BK6YJ??0d!PCn&p>&Um7drc2ZGRe(TgI!<6o!OaLc9!TX$Q<u}N*z>`3)
zRZ~*AUmvS<;wO0Tb5_h4N~g-ZOqtp#`u<)YcK77*Q2FHP4EfLEuR^zOqdNW7tzyn7
z{IO9Wb~WlmJZXV`fu;Iq-6yEVAx{W2lkrLV_hB)nDh>3zxjhQKKJ@KvOZT{{QTd*$
z=zo1&6+CXP7It36{HR0ZDTMQJ%xkJjd(d<ivOs1N42qmUyEAkn{{Q{Eiz43wZ+iAm
zK2BV}f6DSt(oWGGy#Mo|>+No5ZaskIt<E6A2^chFC5FEJ!&(iAW{}*%udR%lmVZAy
z9jG{rxm$!M-kUO-cv0bE-vo_c!MZdj_iGt6cD=vZu^Db4$?a&qVHw$cI)3Q`T2LuJ
z{-1w|<#g>oMphfK{GsQ`;GE`#bX+dE0=H}@vX&j(7<cb=u-kenKWU*QV;ZWJ<eTE_
z;Ew<N23d!des^4!9Yho5>=Y!Uco9_bpE}^0{b~j<RxD1ikKsD1gWEB6N4kx3<BHSH
zeO|bGQL=8!&00>y^BbOn#1Mo0Ln=vO|1$DD|92KQz62sBy>_Zs&jF<YPh=h8rJ~x&
zic%DGh)6MhFgF5=bH?y8s@qwJW57^zv#wv?%s#UGYDWrUUWN)L7i%Gr;NwTW?*OJo
z)o&S9d}#hZGp_&hqW`nc|F?GH|058@|I#BaL80gq-h+sBCX^1vQEgKml(V?9R!!fz
zb4PBppn3AKB;55(E-b}3G!F3?p*JN+c6JEy{twIlVDHVtxnAGCU)5?=8m&l5R92-D
zi$-NgB~41BWN0lat(6ERvsM!!8k9tc(wsCI5+zd+%@Gj_l?qA7@VqW-{r0|(z5jXs
z*w1n7{p`ErKGyGU@bz(huIs$s=lgu0@AC>ZE(*X8Vjn{rOzb=tdoDrXvU4A1bpf$Q
z1V(p}n7P3cHU99Qf9@xLZ0+oo7gxuTtTj)jJAT~=VF6+p32y_4%nLqA;9V+uQ{^~l
zY5N{`k=QZ#EI<+l;_Lde?#WlMI{4?03$lZoZV>`y&9^Qs$>QG)@BAK9R1@CpL}K~Q
z{s6!aQ<Ap}wQ*HB$OR+evman7E^in4bYE2o3F%29I1ZTT(&z5S=NA|UQ}u571pXkT
z^B;%6@lB<3;pMIVgMW);Rhn1#myp;!Py|+WdT&E0Ayv6<7w%O?J<Bn`w>0N(j9f*<
z9#ztVi~yF=JkC-dQ-YLW8Vf@bg;9ggY0yRZ)+h553+m9T1)uE3m#3rUpXtv5SCe&n
zw|??q35jf-&c#%}$s;rPN5|k+7`eC@SJmY?-{#-XimM2iaTzI?=Btc*`duWHCv~nt
zql2_xRz}7<+hO8^x_5q1Ts)we7eg6UdBzC<rWj}?y>&iK!+Foecdx--CzXq~u*OE8
zeIa>REQ0bcSLFh-va*=v93Sb&BKHyBAaY(p&DER-!D?RVw!G*|q?#k1@lfwOe~g1h
za9X{24L+|i;Wg@+!H54W7TRcqyuEoaEolu<3R{R`JGORy+SzJy0IGh`TS@V)L4v(6
zDG>BJUw%|H7cfbl&yx|Mfg`$0>~Q~8Y#84I%6h^YTrH_ODoq6=w^{!@S7L2i?@iCo
z%h)dmKO>)(OyrXWh&QG`5Afd7_-Uy9Vd017)2s7|uFvt$`S;b!xQbWbc}>i(CA%%z
z`qZ#9_G{wA<3=)&JAov3dgU=I%a#2ljOqp7^<tEfjnCu1-LFtNMjbr)#Yzc@QHAKj
z0FN>v(kt(0R9fd<@wZqTeu*3WzyCl&;(u~IX3iJ3d)Mbtp$*H*8#NnI8w!_tac5@w
z%h3Z_g!YI9XKKl@pdi>j-3g3ggOd3OyM<6)`s0Ew5|*QNUeLZja~1Dp!J<WdUbQgI
zlWakCMg^f|J;@g!pznuTy3fo=7!t?E`Rr@=U4IjWA;}lI_ck;)CZygJb&v{N*r^&A
zHG0H|GLnWb3<+Z7cri*KACL1yF>Y#65pc}Y6p)W}lwl;!w8BHafAXm?Xx>Xhz#HK=
zEkErFNgl~AzX%k<vY9i0tfrZ$fZ@)8GmhOPlqJNz_6<Qc9B!W}C+1p9xmkF8d%jeo
z;mKI2Fm?4f&b9E?W@@pr^cse%S_NuPL20}?cy3u@b7g3RbNN2`Ps>P&NWL23<6dow
z^<Zj%qenFaA-|AgVFiAJ0*1*=A)F;Ci=d#$&+6#WlLKchdgOg0D3%P1TY^LOg!tlR
zJu{@YGfTX3q1x?9<UG*<)5pEIVf1X+wr5L8p{n5*yX=oRcFY-K9?zsI0vL?hwl28E
z4nHFfg0wF4L<GEP(7~0a`IDWO>9lV_5*+kn<R*CMS(^9MH^=G*zkUX>RCLu|z~zJh
zXcBs%goK2y+5;qJI{9!tD0|{+wI$s^XG4|mHjNovSSp`pIjvy+JNT`;M-b_u@bV(Z
ztE(HN)Lp`I5$A7a+*Vc~c!KGajho<qZedVy>|0PD9_2lu=;8&R(^E^`C#oK*3U^a|
zaatylQD@GSQ^`xTIR9cSGz-Q%(fRgl`O@%4qE^08K+W15Z0=Mj-utX+JfC#uyE{$L
z)$ipFkGUqsKhgFqH($uLHxP<q=75@4a}JcoE~n;uitR?nWPX_qi7IN_A`zxxgMQBM
zz|Iabvt`Sdvzz{LZK>sI!e7RqI35n0r@sGQYGTTW9BxpUmBWjtzh?`1Rfbseh=bP0
z>h1bjbQhk<>MD_KU>nd)A~HenTF}7vim)W``hKhQoqa9VFyRO0qa6&K{^2aYS#{QE
z1bdv41X&{x0n)OW3f55-lX^y`W1BHPu*Zar$cAaDft$(^yHZGjxO>YyULhsrB(Pxe
z_Sac}Z6sUVE#x?^tQ@LIIhgpMAk?>w>^>EatoSL(%p8pW`n#XS#>LFE=CS#*=BcU8
zhI|d6gZj`{1kH<aF5rV6JT~Gk8$4~x@0@BO!#Z$)P_S3fNgPq$Y7uqr#l08KpYQVK
z?;e=}!$N6!IbDPvOap2Mk>cFqS7fPS3RCZqAj1yxGaRKzBi#9K7TMln?2V%Ho4>iP
z=G1h&+0{P<5w)fVFsI&d$jopP1DPGIgpkX%fuR7Q+DyT;fAM^3ICcGd=H8|FDE`5w
z=y!g^nTWkSpDWUJHt<mU>jnxgzm{7d1>9=Xh0?tyisdF%dT0gP?qm^1wy;n&@z;&5
zkQ_?YX7C2m*Nu}npSKy-D3y~_E*wU<d>tBsvfUMhO}y0hz=UMa#u)}*L}L>aV~>?7
zJb*~gL)GTdhXbrR5h<FjwiRuTKb_i>-PQBU6}+Zw6qKX*p^0zW<tL+60>l06vW=9t
zG9H7W_d3;!c0F;I$i5NTe^}3|xyFP5s*h86RlhB)AO!qUGBG)BKzn&8r~K26mtu@C
zQ#1Z(dcmX}M{1U3pJwmv^B!XqXE(wcL2z(NlN+;s;rgKY>n&<&D%v@vdz<=jX-OI_
zoH$;BNLe3jK4|^smOdi5P-<^w$kEHQODlX%^xj&Vwv{dIN!@@Oq2CAI_m)zR@7h?l
z8_0j~pP1Lg_8IAU_|(Ehi`?N)EEv-Oc6xWU_$<TGuhSps6G|2rk($-Qhl<dC{P+-i
zaZS)46_IN2lSKAtfeA7)IhaRIzw;e`IlDgW9=uR0^~VrXh~C!RiTTpb_7q~Ip{rhC
zpe2l|R6fz9%$&yGo4O96{5f#+qMl3s?r4lpoxJ(gB==U`_PRg;+fst?kvSS^sCRBF
zaSdJ+(Ll)4ZB>lU6b!^v*vADCU56$A#&A*xXe>gwre!`m`djs5r3KytmuT5@hzgsP
z?w$^|Xc0w+9xE&70F_7H6m`tfz;E}*MC|UEx<pn{(TjOfl|Q7qNL-!W8SQ99@fY~S
zabc52HAg(>>PsGg%cQES+UHcR$%eGvYP(}62@T>}w@)|9M6YYasCTJv4=o(oNFU4G
zApPGR#x2MjdflS26gq=w{_gV@h}UrZNQ{P?KH4Q46Kb0@*8Ta3;RJy#k78Ju6u?+v
z>Z5j6(kG$7LjhYo1!OZ)Q$6q(YW7SpQl11k82Pa{+Uy93Dzfup7|LS0gP^ZX;;l`p
zXnAK9pgFp=z9iDm^qLh!AqD$#26DN7PONILoSJ%$qk*#|%P<6zAL5P9P<^u<QM@v#
zmpeS!nIi{H{Qb@jZrm6zDtEULh<ktMZw=FJv|lDM%7vO<jA7Y+h=Ihiy+i9A(kJFk
zt3bdIm~d`p548sb(^J>-b!L>uP?Qea_QGV3*^-3uvB?_crDJ#$KOyWG9+1F?aHc1f
z$B^dBrnB+^Dlh!5bsP1!fq&NnsUvkFe)H(QPh69&E+-)h?OOuMmejXWx|@B{h(I~g
zbGN=rdNrbIGMa7aGD(zsht?IIA<fuVz~s6z9-;Ra0nZ!O&u5mzrP@p6R)Gq`D;xF&
zaQg=~O?yD-09LFs88e0xHFh}=HifU9Fkv$@IDzl6tQ7t`MD{7v?m0qpYeZLm?Q>(v
zt+hrcbuA~EJ2=#Dj(`1vs7Sj#cXm7TliZB*Z+%W0pX`rU5`%3uHU?8*<6aSYiKtKj
zYl6luk6Ljk*uY=(c~>__Jysla{FZ_LhQSjzEPC{wX}6PKp6MoGGN)N$kBa*(iRPL9
z{bsiN{u1py>m(s#L{3b@q-5+0_>pt2zGoXrz8g0;KZO<D6DY=NMYd-R+ipfvY?CUB
z{aM7GxV(-ZsU0~P<BtwHx(5G&I7nA2pLW=crni%*(w*Yj)Fd)PqYhKM+mn;M*g5w)
zUL90zJzm#!n3lBPU~Q_QLCY84f1;~A-;0=XF9*To!bfk!$2fQ3R5#<S*JY~)INUj{
zhnK*jvrnSFgk}wH6Hhq(L}`0teBz8@5<O)b?7Ni-ls{8_V2bJj*yknF%64;kJx9*q
z_yeajg5T;#-+0QA$#ML-iSPX@EFS!~da@{S;GJHYR>VZ^Z?L3Qmv7Puvo5^t3!s84
z34E}2obhG(>p27fsS@YA-DT(0w3fJi(XUCR{}UWMsa&;ewoLUK^|~{<ZGZA5CPQ`&
zlKl&!$XU==N=}Thcn{fhuShheM(#0k5dKhrHq8qEiD^?k2;48tx0~xIj4$T3?wrCY
z9aMd?i^Ry+PcMD?j}}0%iAK3X*2}iYEB-}Rb}E->t+!wM$F&`=aA#EL|B}tirDa83
z77%L^Z)=Xp1S#@F>!;g`?G4h0?Xqm(#G^hHs47|8>frZ2T#2{DW&Pwyy@O=oR5XRO
zmS`up0jvWHbaU=4o#;j=P&)dW&sGV1sPgD;?@|GM79AT<(vErO9CwEe=~X*1z#d)x
zh5<-5nB(3n;^U%{%`~5b8<D~iPBq-0auJEhK>?(7`l>l)ORBp|_+0vj0MTrGwtUHY
zm)|M~D`wP&awo8Z#JDa;4<DkKs8=AN6ZPiLbA7_A#67@GBPtC+I-t5CME;=hj$dCA
zlyEoe@Cn^%vAe%c)9{Vg$>(E{nkG~P<UFx}FaQW4@R!MZv&W85P72V|Zpp%haOme@
zQI0nH1RE>5)x=5l2&_Lor!YwA%n7GgC>KPVs<&MWu$-H_i3nYl^LNx)GL}a%$sK?6
zkr*y(Mi>3ex{1IEBJu{?`-e_)v>4PxCcH9nXevxOSfri#R!2Ve(Uz8^&N-)$&{x9S
zVVjLv5CK@IVmWiSb<wORCW$lyhM<3G5RK#N9}u*C0p3LWhs1?>3=(-$*j=k~!b`7L
z*F^-$^f9*R@%lk6(4>3c6i7@T+4EJ|ppJNy>^^cWKw|_72R?Qz*L(s!9%;z3sO=^#
zzA7?vqgHWu((d#4$4?XVPYfd(i_xx^x97)*<9_+pXUSJB;9s4FN3*+|&sLUhp*&3)
z_8GxGvH+FS2dSb!XY(ya-nlu)q&WvtbxWk)kF+?gurytFLQPTp7eo(ww?v1a|NCoS
z7-4oo+s!Fmb_++PD$H?l**cUJXCd*WiU5JY%w^Li!%K}8vtm|rsh9;B31c{SMoTzO
zbpXCF=kIt?lEZLxjvJ+p0GJ+(JtMp;O8X~41AvjlKNU$TE-xi)T<#MqLhw>2q4-X3
zRMM$YKRdg!3xMX>JVV32uY%exWZ>8$Q-T}8fD~8Hz2rCKWE&h?x5O7a<UBjKT&ZN1
z=yqi1!~c$WxvOd(^-t=B&>ZbLb7drk|LD9r7a~b;W}45=n>y%T5<OL_RvMej#CeD)
zzy?MZz<$8EWZ?I;$eg`C_iF&wZZIj+zQ#63k*@i`aly4{-l%=Y_<!?C1^{9EhDnKd
z0}A#BxPbJN8&FXA0#Ei`wIUr?x`~GK#+IMeQbwwN5uJLV@JRCj&B|lSBiyzEJX`>j
zBR}h~J0XQ4f9sfL4-VZE`#9(`x{I;bHSqTnoSKnY_X$NTe?3s-t!sf#?**q~V#3Lz
z`3R8`<5{C&-H>Ri|2E1`wI!7cpxVRpYiKF9Vk2aCRo#jlm6)&ny*SbwIEOF8)7nOA
z#~pm(o+-}iGS}S~1PzrYEG%VG71k_Y_56Bklf99K`Z#!aq-^3IV-Vg*+v%j|0J}9C
zkNjs7B&@+WkJjTsCG9-guQZw2=8CKDvYBSG2TdKKzck%EZ~DiJVG5^7#ygtxy+-Po
zYWN`>Xs7a%|GueNL<^}RO0<=c=C=ITXG{~U!Mem}Nnnf^R3h?mZAl+5`xUT&vOziq
zOX;KF9xxMFEEH<-GcknNR@D}Fx=;jKU_J_4o$?o4#;z&K5(CC}=9+P}y`K>M#CNT7
zy(3Z=8(7@Hf)W^#u*~ZB7jB4Iu?vMcDs$-kYTE;~>gPeCIBxZEm%V;$F4?0k6X0tb
zS@tTDpqxsh?$e#sCwsNG`h0>45jha|2*pldPF3{)G$*1}l)A>fsl%wlksqptjVX=+
zf>WaEzzMFLPOrH&g6R;wXKS=2BuqN9_Dl<CSgNe+Y(vO+5o<VVzhQqPb{JezLgFh>
z-@=NxEQJ|d7-0(K+<5yFY;!Ngm%O4jUi-0{T7L&;c_r>7XL>v^5UPSl8BJi8Cg2NC
z9|o#L&k;sWLCxKhBdMfwQDgGtZz0y<V{`BNokLd=3!W*D0Wk^{mkoyln>v0W{KEP~
z3zQ_kzm$L_E)i}R_ou8P^t3v5vBl3$Y({cVEYzt>A4nu`A~hgGQ{$v;Opl-ZnyoE&
zy{W&%%s=LJ%f7~2%53qLlNj}Yk4V$CC6hh;u?xhbY@L#jUX`7I+K`U>%CUzsMYOn!
z6`8vERRh5oBGWA>KZ`dq#=RDjMi98kUpK|xK3ehz|5D1pFZ`u}dj^|CHSz5sv($|y
zB=%z-+~4pN#IUq!7w1{qvV#=2>a=6PUc-12q1tvXUcUktNX<mgu^OgHTmNi&!!a8h
zdGKj>kJ*)Zur7nQ4wuOGb8o#KmF_s1w5T-@si1+4E;~`UQi01zdpQL<t+n^4b7w!!
zCL*|?KFA5WKfPCJg1yz;#_9A_;iI|d%M1{>a<k2dWAp|wU-qj;8l`Q82IU;`onyk*
zvJRm{W|D6KOaE%}OS6G+r#qf37EEb}n0qItrcar{(B<`CFJB})lD#W%EtZ9P!d+~B
zoiw>3f^2hNf@cO|GvV!GGicQS6ei*_tFkR+n>N%GI#NSWmrI_85dINX&nMc(#0UV-
z_EULuNK!}>&_}-(H~2@DE)vq6$zW@s5V4Vf<sAAb#ocWVJ*c;7wY5Evl-B;q$!$D%
z-qo9nfX*_;X7;S8iZ<CiKFQBS46oYKT`(x{T7cnPPO#(k3NB8;HHds{)h^lPY-F^T
zj!uUfK)mK0cyh?HP=^-RK~ju$vj-J_VS;jJ6|t+taBig?zwr%(*qh$T=mrypOCn&o
znpiUzo&kC!rM6MroBZVc?K;ph@IwtNDPc_n^D)CfWw^jBnhxvO0&9^~)V%FfhS0Id
z@^G{iiZ~m{WRn3HfW%%Ou8=B2hfHWY^ogkFy*<1ATaEr*A_%q=Vp0fiZ1vNhC6uR^
zTfwVKNb5t25QiQ4j8?L)oBWobNJA3k)*O^E%ohEPyI_e<^RJ<^!N`vEDAv`PRF#m>
zC^F=eMh2zN!ZZL)>2<Ki!XGRkVjsn}ebL>;+<7=Qp_&EN3UAe?i)UN{&+>|~ouc34
z6@SLc4^lO2B8f9_kzBv%XSIM;7~*qSaeq~7GZi45?ksN0)v%&M4QUuEG8!Z06$HTA
z_20?KoIYPrUW9T<XG`aFz{bz7p^x}=ooIn5@=t<=plSiCP_+X-4k6X{b@Y=zATE9f
zbdpv~4Vd}<IT(dpb5A@Q_@3y%@13e3q1^H|YF$L<d^F$N_>#=AVJe~L2;;646+~>5
z@QtVUYfkcH;qa#DUaMF3lG`!upTCNNJ$o3Tn<|JOS?&Bs+yyT!+%9pb%R<9}ae`1=
z{kMd~={+Q#;%%^-VBz3GD}Zq9xo@QBqJ`HCBty3d8os|?UVhE1VclSd)mppUgx}z#
zVQH~v?G%Zb_mxD!Mfr&spU?&f<s%A9-{B>MKB9yQ`rVb?DSCBIxrv;E&~qquM|{5g
z$5|Mp_B0l?mmqo2P|g<)D?v9aX#h^7f|rmm`h|?LK7!%#Ky|D<t;~DltUt2z`}v9i
zJ_JI+l9H!sz6)9&pxsyimJ|~ifwWZ0!KEc&D)J2QOa~{C_Dpgp-NjgmZhKBC=76YS
zMf$AaXGLEIIvdvcRFeY-Qx3oob*8vA`PH+tsPx}|@V{m~>f%&UP;hfn!LUy6lwLS<
zQ5auZBU2&(GB|0Fr?$jQsqf%XY=(7TtTgi9_xSIwFG*Xtxc*UVtWI{eH}Y=}PHjE?
zD5r~rZxW!~%zv@PKl7C&B)W=vA-{%qzI6Vd|4sg2M^|-@Z$bxS6AUiqjNL~39Pm9=
z5NWGryGy&hG;=hpR8-+Kh&z4TQlojfn=P}`fA&>+`tyQgof+=iYh*WV0*MwQiM-AI
zqq*wafj|Ox10m822@0pDji4!#pJmIa>#l`x=fDV3oG*ndx@>fER%4bR2!w2p%CS3Y
z7!I}2fARW6MMRWWzkL)OE?C)#?<2subg_%R@$~%Z_x?G4!es=$-v%_y(-MP;zkmJn
z8weV=C-cY1tWhUd3Bv)I7yq;)kuxh&VDhIut7abjMoN_jgT<F{t-MZlafI>VS%reH
z2u6m?fP%2!2blJcLrx}W#Eq>J>G}&(J2<j0B8?${amiUwvH+N3BpFHVn%j9AT*!)J
zhB1AOc07@&x8d`adgl2n>ZU~Y76WiH%cvZiKJr$YAz+99)KSz!Uc3uZ#hN9*zO<HV
zV&4y~un79C5vEX)d@}>h1*pf<zSmpG9^h`$YI?JabbgJ~hV7z2!39k@CvaK~r9##-
zv#H6(+*p`FbK}PiSUhF+d##VTfJDa(8RPVf41Kk$+{UpfEguAenAF6Xp(Z`l{?61A
zXQc-L#^&2r9&Hz3p=9x-sidbv;+nvRHxG<AA#(u=4{XzB{#FeWwiDo!XI<Xky54r+
z_|a2F7184a<Qmd;W+RPGsm+x`W#(GH+A;_Kg<i3`t1^t_o1MDU|2_D&GQMZCGX1WQ
z_^l4EFn)M!zM<h}DErEeU0}cxwx&@>Wa*UICuxDZ7bTx<7rN(td}`a~AjNA^CtN?A
zqocrIqms+xHGeO^--P#06X&##Y3ek{LN?aCOk}Us^iQF|4UPLxeP!CdmqFN!T-X5&
zc-y?y@uG5iwcyz`*3D}yji!7CT9R4gqQBYlKyUp|(ro&)^q8M$Bh9A)_gb^kBpn@3
z(C4^5mah&Bv)o#r{i<5RC`yaWPi))Dqnj=#ilJ$Vga3khKp~rVCP~?b%T3G<7S((E
zO>8Z8u|ty9dbmeW_%-)kGGj+oUZf;fBFMYRlW*+ER0K^z{080f;J*z})+97|@iEBl
znBq2>G3<B727jNJuXXsbLt3vf{wQ04Fs%6kLqj1GpxczP$|@u@4gI*zgvq*K76%{c
zlW}OD;rwj?N-k4(Fj8TS7hXFs&-Hs`EtDUf*BIX-v*EYRJJL|l>^M}xg<Q6zIO%iJ
zM~A7e1hE4K)RNWD6=VbnP(Z?a5}*68V`o#%>#tQiv*g3L)uY~AjBu$8!9ZW8xQT-x
z#KT`TW;HE2g7U3${hjh^DkWYO`=>rh4G0YMLK?~z2?b)_qLKl^9gF^)YJ#Apn462g
zraN92@3lcqT9)aMLN246_RO}ov|PSZ8~hV{yKzl`lP2@B4X3^qJmMgu`6kkBl21dX
z%seR--EHQ5N^W+7V|L8pE=TQhh9oxF!gXxx;zgDpNwAW!K4q1ka=w^TOt*9kC69{~
z-P7|x3C{>bX+_N8Q7BnWF!EM#-3UqpD78^u!H=mB4$ymhD8&qgB_m`P<ePZ0G(t*w
zuZTf24j(VYGeF6UR$+i>DUpuJ2qMY~^539B5D8OXZ(Xk72cu+nv?!|7xnlpaZTqWs
z_Zp}XeqfNIqR2+uNc~kI2Ga8kD22m|Yp8WN0Gil!`FoZsOFMGro+JNJaIHRB+ek7;
z*V;yKH{nEJZ7*A4$03^0)A|sK&RX6+-^b{e)kAM9U-49il!S=OuUFhcyVB$B`>yWZ
zm=yW0_k<B^jy*B%^@XQ;G&=+aen3O*1-9Vc@3F)T=X{|`UKkHfpDXVOH{I-Y=#K?e
zCP?koTOWZ=E#PS%sX_tVqNteC)>x#}Ps{;OnYzlr%wTfM$TROn3(ERfL5N$QIX+e0
zY^XN#Z@%vQMNkJ$jni@M;>2h)j@$iy!P3ydZ)Yu?(cLjL&DwJLGJW?@tUz4=(7CTc
zcY#nswc)6!4~kd**5}6K`%1vW)31cgnN1e%@L()2OiV|jm%!`wwwLcLPv0c*p~W!X
ztD}9)iLJ}1IWX<Q9!9Cesc5rcUi5;@k8y8NE)1b?dWxvPX9#kJYvK4*JI{xcUnVy^
zS;+O0^LV3+`(4)HNBY1e0`>E633S-#%_3C8?=x_wqZxiR|NXglIk%1;JvxQsOTa{H
zUrr)La}<h&Ml|3W>#`3-{(AY1s2P*Jn+#DS^VaHz9Z?vC2$W~Jq#r``AsmtUF677P
z{LQem5TyY^fELF=#2SUHQp^F+OK*kde#z^=Hy8r3hH>ZYE1_w5C`BdwF!ejLEqA+%
zItWsCGDfdwOAUl-l?QJ@TYQ_askQ1@Z&5Ynau9NP2pWSLBLWz=@s#oyCQpn=R&Uj{
z!`qoQgu|!n*pjiFd?zQhp6JU(&79JQ+CnS%SJw5DPxIm=G6n(MNgV5?vb?HKdV@U<
z)a75xn?C-f?nsQiPMjXqmiSlV&`Zv841&+h|C_|6wXDBacAf=!M@U4+6j%keq0EYX
z&3B7Ohx+CfXQq!1t@p4Ts1biZ5sG?zavitxex|V(Qvg3Aut{<i6hZIvS3AbF$n0G-
zG_p2B^=^*^0hacYU61zm5eAs3N_sqZXM^D9^jq~7lfBrLRET@do{=(5x^3^5EZr*-
z1=oDC!wD^#i$kYOuqsf=U%3Q7&L{nUP`^Ds&EB$M_U<3Na=Er%i@BoR%lD0;KJ1ll
zJXTH3nSu|6?5CU?=mzBswu?2HxSrD6id#p^xBxc6rUo4%_!}KF7PO<Ii6mg+uGIJG
z<@%=Z$plEU$Godns;aMCDKA}fRC3~q6*SNKFnG}6?y<b+QxS_r0k>Zf={W&!Y|908
zWEPR>Q)Z)WGuS`m$noQ@=`U{{7^dSs&9cdjQjPa^9hxexZXLfoCYK!io*S91_QHHJ
zb8e!uZ<Z*VE+{ZDNbW0oO`0{Dh+!JB)0Ytrxu*LKnWWt6Q<!+s2Jls;dl)0M<7%fj
z81#vqYsV#+S@vZ4j+&v6@tG<3r}g|Y;KIJY6Aky4Ys9O^^;&jpVCW>r(BkQIN!-h!
zfUB71Hn;VMAvbwi>D?!*{y6^U?7YMwC$t<UPd6HJIwr;fszmO>@x5+TnUyS74N%q8
zOr#*C_fnp>R?F9?)IS*Cf7tY(oHbU~<fl`Mx6E$vTaI|blTf@NW!{J=t&}$ahvYXq
zdt-wdI_TTZJ)By`4cW3sAN%!s_w-gHl^710W=s8XjM+Cr;!?y@vi0EZsAwPmWmCVs
z6hcZpXDnQJwCJ~Mv5P`FLPm({^5c#*PII^#c$6(%?P8ZJIfX+7LACt2)2;s;udI8#
zA2F9uN;U%tzc>>79TWKid`!6?X!ogwC2N9j+UFTWh7IANT?Ok0qH|nfp+`S~U&1P9
z8W8Ud6(U8I38D>Z?R59NFwL^q0<Lh3xPRTt%SwfmFj>QP4gS7R4FJVk?Lm_nQPcnE
znQ+%}8|R1Bh-wcI;p(Kl9n#}mM8T6POG7NMMZt6G?5!b6fuO0kR(~{S7snJl+`Pyp
zK)H0xlo5`h$wa7y&Q|HV_WCxEOk%8Sed@(UzMgwVJpEaDtNuIV;XBeMaarRuj60?$
z{U?m9TMUf?7@?r`)myjqBw_Bo2Tiv^F)D9=77R+d@Eu-O6wKkS#dH*?5gN;x$)aL)
zmbMJ7+B@C4JL`1+onXBE^)};{mw<gX6)?Cz*jr2>;w#vbi&5L0zi`=i@QOV-v9ies
z;aK4|nlY*7A3J{hcvx5~kX^LqAPQUX<qAXx{s`?p-`=*F|6r`(L_%@HS_7m{+c{-i
zrEcsEae)KxX=$8T8H;;4(Rm?sf@}Ku`p@W$@}D@Q`vvY~RIqN&6K`&p?Uf;01LW~R
zqoo?!%Dr*BCLE#p_Sci?IIH=hf#wW6GTk)-3aS155-&+nw!6jlz2?}qW-t8&+p8aq
z7~MO0P8_#IZDxZ^=CvmDTmwAsexs(?nynOwEzop^CUOH%G3lN9h|yp5nyzal_4a9g
z<U?VFr!2JMJktA&vN~Ew4Y&=(y#c5~9{^IF%v-AM_@;*$cpczrL+jV8IjXlRk8i49
zy`^c%F_@Jq``joy8nWfq(o24mt3>nFZ7`;s>~}F29e|LZUtYg2(a>q`_+^em2Zky=
zDBWk6T|7E}Wc&K*T<W3wPa18c1+m%esffMDr_Ah=Ucbya@|3cZU()VbCxv{W+;Clp
z;)ZSLOEy@ag{00uK6!j{5h!+|=fLr6Q(F0>dq3Xh?9VSdQq|^JMAOrm{$h-QxLeK^
z!SEaS4;HnoTEph1okX3wLkm^+)ejD%Fq<nIGFN4qV(ATWS5>@tMY`t7(@0%#<y30e
znkVzKrp;K_o>2)*>@)qAjc-EcXTEb_f>nBHY<t5Q-PZ^sOoj;Lm^))bM@f~?P$gSW
zxy_p@>ozyJPBWB5^$I+fN;h0D#%+)OhUev&FU#)wI)7S+D2Mw_sWF0yIO#J?qgnY@
zYHz)plcUDZw?5Ha<-Ygh<y)#>9mn1~0hkomcf&#J++zn_f?+i`w5nx)dtE3h$88_i
zUaJwz;;aO3ZS#wRw%vF4sN>2A^^-zb;_ok2!s}2mwVJH{Tl{?^v&t!>r|3!kVR?9i
z?P2RNVOwT3oy5aIzT_(*GGKL;l;RZA^6`^43mNwTgUfS!EQsbtf{Q0Pr>+fIemL&x
zTk?^ECoZfIxDsZ5PrSf)R+q1UnrdBUFFjhPd<sAtj>28<wJ=-*0EIyoJ~qgy=nVpV
zO{r;)OP|HKd{TNjT9?BiXwI+xw!6OZGKNyq+g>D`+D1KWxWxQ9dMjQbUZ6;13c&Jp
z_7Q7Rk?4rdYz-eIb>V>@dFr@i_44kCg<BC8tWPx9Mxscn^Wgn7cDk^lyH`Z%V=5!|
zD3JdP<U0j7pR364!`rf~_slvu)HdM{f)}_=2nk7D2bcUoq&69}jF^0HSAU(sC0|K8
zJlTm;$RYVjMVd|6DGNS9jGb7jxcul0@j4Jtq!p<FZ#%cIDy?`eV0jVyJ=T{9$*0NH
zb@L$z>z7%#kNd-VSly}?0I5#PjWr^p7A!f!#>jqL2HD96<^i8@N*rvj*BmCfc9Ra~
z!`##^iiyg&bUS9glrqlKjv|3w=RVB-@s23^4gU?tsiQ?WZry@{ZSPc3GcvIV0nT?&
zCG(j!OK<JuwGF)F9cg343nm<hOKsz(%!*Wa0%RLMs%H}n#I$ieQ?lSN_16)T{#=Hf
zE{~Z>{%gk}5efVLxW2A?|3k-)O%pLOHZb9pjYY`~ZZUy>->)s*a<A`W#c|uXq+&|-
z+=;*t$DihnIXFhxsT@5Da`M?U*Lpdx)t2_m+zI#rc-F*aIZV$J(v$WRYvEl*d5bJm
z$glCh2%Yq&10N@ryS2R#z-8_A?FmOdkcM1<Fd?7WhRu|B9z!A3+LD8rRC?*OH&w9u
z51*WY83$R;hb?E7O;KJ@^s5Y*@%b2<GUrBD?ilN9$z8v3Wr#_DCPVoCL+awI`(!D&
z)c}vhUCl!)zb0tX)T~3X39@*y;$IJSN7Y=!49Ov0ShsSS2D!u??89DQ0>U@WfyNr~
z11fj1449MbFtOEoX|KaBa!0F|obPkUW&O!qJHy9+*d89H|Cg<m<>6t?lRNuFy?s7z
zS2MYK3ZVwuqK-5TAgWE%DIEIG9v7vz&NKfnFHo@hg~yw1KPq`%W;npRr8|!ypE9%0
zka@rM%&Ajzy#88#a2b29L5=HEm|a)eUw3!E8=aj{zlpUF3`J15i_1#^d+eni95J?#
zyL}4YE1badey#VCm3E2mj{4uaw^P;XeoAfp`?Jvj4=M&q5nB(|ojCBBvT2U{s0~io
zW{yv`$Ga&cHfzy#aF}=O{9z9opLz>XTkVXCN}gXXq4fN!%SHdNR!0=B(;SaNxa`)A
zuIGsPG)ulGv0jn6GgXd7#ye_Du6ou_pSFoMRjVw{T@x&WpQni#a9LE|fZ>FdF>2%Q
zSH>RN{z5O`Cntb1A!O8)<UE>jM>@8=pwwViTB)dN8NTw7?Ac5+XqL>Ei3_bJev!Ez
zWor{sw9eq*@uFh|>z1c1vYAymrTHW#9c29d?HEe|lHZI4w#e@hRvDOBXc-tcRLy!E
z`{;TX$%w(jl=ttnU1ca#(EhQ3({|Z#vo`eOGJM#S6d5t`aG+{(ZCZmGhpUj1EK%?6
z`jN*1)tskf2}NM%l52SWNg+wtv=Gah_WVW0X;2d;8*=!>R;e#ac?anPevK;IB;Zio
zO&h0+V)-ieY0POoqH$I`bjj=_@~DXlr;d(jm3{T7!!vX<z?zd{GJO(aP{*!^Oe(+Z
z-%16}*I9GP$wWvRcCJE+4h%1)QC5~?iKf0m)ZW{_{7(K!(`-X`S+fJ4U%YINWMAvo
z@pIl51SiP^V3-^Z5)tfAaL>ITFeeT$ULmHUN>B1Ku_6Li7_*r?=als&`mG}l-;(*o
z_h9abXFyCx^;wm{XDOpJ8!h*o7=J%t<YZ9({g~t>oojvyy9Lr3vndhPLzkwgO#DUv
zzb^5MT*KhzZaS?_*(xJ-+(%!_-89V>Vam|wGQBwHIv+1ydy*>r02mktt8{|mi#Du{
z2&aqHa&u#4TVWH1A`axSMM4a?Y12hQHhAT>?|rpKblI6b&adlEtw`l%QeC9XjOGmZ
zY5kGsNpBW=NN<+;xJLGw!LwB@)eB_~Bt@$Hbyu?IiUB72PIG)$^S`?;_S&8M!yr6!
z(!}i=h27kBy!uv^xPAVTw>vk)E-%a!Akw6ARW)9=O=**-#SNBUwL*`*+iOhNfhN0N
z9jiW+YMgh4i_mCPtFim+?Uk6oLNtF7K0<{rMhv5DTH`l(ZrJZ)N@1kpJ|D$efo3^j
z^7+p}oJgU*ONpegvO<6uccGY0@8L|gbnm;`<(0z%=|mujGcc2PCzOYbFPPg<U=_Ij
zme&e;#0b39YHt%b5w760yqUsx*@Z`f2P3~+(k_`FY}Isr*Q%c9mPTiSZD%k#n(>je
zkAPdS{q&Zy8-8@#{yhL|_&@8mhqW+L@0yC#j5S_3WslDK{xJIW{4pnjny1-K>~c+E
zB?zVK*>jO{Md?v<_HXWEV;P+Pw_#YE;a!TG&)MR}G<hcvv-l}hccJs1y?u=u(yQ;(
zDVLi%&!Irr(ogfU-|=m|o@uH5Zrnf>QNY%V6_>zqm?af&dua}vK}dHX>FsEFnYv-v
zRfB^aV+36YUHzPf?KWUBi@To69N@sWe&1C0pO>#{c6ol)bEDQF{g%}Ao+mU-BV0nS
z55849Ml;iROcF#f0DJvrIK(YY3x?=%YNSdAk8(YWwAI<ll`&ZvKD)qdW0KXg0*kvl
zc%}LTn-oyefU3~uE!lh$)3Q>UESt99xkjPbt?Bk-2p~q_>;yo?@hKb_t})^HK$ksb
zgwW0Tkk%eNV{__mt!X!7r&k+4?K`#jz^7<u03$$7Hl+&2<*JSa0i;rwP7<e_3GvcK
zr#nz9j7nD59r!&!(!^`myUlR{7ae`04dOB~3Pv9-S@5hN^Ui3eX8D42xgzPa-1wf5
zuOSR}79#?{|GVR>tWUNU9C-^N6j@DetA11~rlD^Krw6a*5hFl5@QXr=_UUl0P&Y3=
z;R@3zWaQ-^4s_o{<|zwj1d%5HV+5vBOK(<It<3rQ%l$oVhX<q-tk?5M`NUOSIqR$(
z+M3&#RScR})m^z<RzKSP^$iLAj%$X06d>#*R;7(uHD>ha0Dxe*b>c<wqH9e&VV)%O
z10YU+NpV`BEG=$p$F<u$hkZ33cKwuF`VFO#Q$%CI*Z8$JOEozRf27a%DNK2E;v^U_
zWX8G@x8Kat_t;9lt=vD#1kQD$V$~nohbcC33dU8~Bk6uovY?+;Xtb@3)$fPL8|2&>
z{kdsykq}^$Z3=;?cOSVYth|3wgvf^}%flvp3<Ek)<*hq&os2olrE}YM{W81jOck!{
z6Wa<OY%xjwJ2awuy=8N`^&dj<OVY&#fw#KOF5v8mAp;?6=b<25&gFs9^bsY6S>uMI
zqlVlLnP{1Z%#`&GM<WUJyFl9(Utbv%2A0ig%DC5>{8z}@Dr~=OE5AoB+zIDwy{VA;
z2)k3qF7rqkparej=1N^seT0ctn!lCzxu|^j)X2pizJY;i$#xlovR6DBxj2ZyKkkV6
zV7aH8{vbfnlC_&Vuc~172H)V7j}D!a3VvYDDA+zn2A!%XJHPAbnt%j%r}Rw!bJ5-F
zQ*{OjROf8f^d-tFcT}+1p=M{KYUj?Ki_2dS;O7Unv#9ic1zBNjK5}%tq}}ho+c-WG
zQNUq*ez(wRR-tM3IA_c?Dc5ZcxJaH3#I_$H`%Q7d@11MPES~IOpZLKK{-UibB=BBp
zVQ{nh*sOCi<J{}6)iPJ%#^xd?cx3eJ{P0=OTbGkQocA{*!2~`^rA8YMN8xd>x%)k|
z7?)PZS@Lg5jyDl!*|^s{FHLUza-^zi<qXRH8VX{z=&!{a)t5K5#h`N<8L;Z>8}i2c
zgO=khu5|R<79~$P7xI3+h$%{W-Nn~P+Lf<NmyAs)_cgL@tqb&0IaQzL!SO(N(rd}#
z3o9=)270wB9i6198oWGr<YEVv@?K3@_g6-cbQ2aO?MksD_217PGF+sRWTGJ@+f#?m
zGT)0kEJY1_o#(C*<jbuer5?jdjQd)L^hYXX_WZ$v2N%h2vl`khwD`f-JEQXoVzM7%
zeoo9Q93q}YK6Urfb-ov7-TPVD)pF|&_YmGf!!SXrHYY$B7T(;D*E_pzilM&l_lfhE
z)Jt3xr%k}8==I=2O`ZNOf|K^|0r}#>`-yUY>(e&&LH-3Nf`_%%yd_H_C~x>%t8^90
zR_6+0m$^}!?vI*lg_Y;$tkoPdo8z^Zol)B;`Xq`wAgp@XiTuE7oca<WvZ(zrAM;6T
zPNb%S8~0jj>pQi4t*?v9f(8k<Zh@bWk?WYc5fZB&hRXif?yZ!!!<2DV=_lx?8)Vgg
z&)bBg4{+SOr>dG<{WUYQR<UxHO#9GEa!wmw@6s1WDio<M=Vo!s1$r#_Cg?e!yMLDG
z;ed{7(i!y!yK&(5>tZ|B54Y>wj!`$zdicaQ<+ZpaILp|HO_+}8wONCWsT>i(VM6DB
zq%Hy0qwGSDb)J5kSuf66?+;Z}#zY&1#rzkE3-!NFcrW@?zS_SS8Y6Qt_fG@ZmjKTm
zmuL&gDzF9<faf<Xp3saSP`*OWZzk{VUaE<rj@#6H&u<sj_LZOZTmK##7Jtt9K!;w>
zj*t)<<<bL>tLIJ5y9RJxw@ENZ^b|!c(q`1n8{U2%>@fN%@jPWSP$BP$9bJejuJu{7
zXhO9s)p4({Bq+>4V>#G((t`o^>JNEs+U`e<(3A-w&BFeB?C;xWm97*hy=*`H*u%y?
z&@rNS!aYB&J^zql({xSpFpGm8Q<Db^5+r1P9ab*QGz5&z>?X{1a(m@*P&WZO{$1X~
zWy(W}1{b{53-2+-K66U<hNHrEjiTu2^3qEag*z#ck#uS0%|_6LH9>J&?4v7Pe2+yK
z?{IHMX4d}Qt7GcMkHW>G<AV<RzS?)oua%hE;4-7v=1;p}R<jbKuWa|Qiq_b!2fz9G
zN`2X}Vz|bak`(Z`^4i#TN-CBSvyMa<UjiVC(w4IpMX=QNZ?Wp(*L&{Hx1lz+v`b?L
zS$^eTz25W@of%C_;b?81#ymupDS(kcN{~pDHWLBf_r~mOd+F6?Tv>c)mZ;I>DqdT}
zNB+sbpsXWC!~stg0CBJif;&Z$2e~NnTK2>_3EkYPjnvk7l>Qw_9vnW#3J**<J8LVP
zfuvpWjZMa6A2bq*>L+C>^svmOieuS`#QO_o|2?B$Y!cO;vfdsxuamwBp*a!@GSKCu
z@DR+PZjtV1-!z5SxHfkdmCZj@))tAzU1vFV=>1|b{y^`!@ZC4fp%L!rt=>A*F-x!~
z-3sZD2Um)+a>de4#`4DsQ8^er#D0_&r{zDd)3D^{&e!>?w4`CURx?6fdH|l2QCw%n
zI;_I_WLe=SMYJVwwLqR<9kvo79f|SEJ#u}9MxM<`(ik-JieT#=d<hZCvMar_TY^8q
ztIjR1qK=whJch`(z|KT=u5j>+-Yz36YYYbF=ic*0#HQT-zZ0BHa5!z!;kTkhoT&p?
z5o6fWzT;Fp>qX&eibE{AWa$LQgAZY|M2+%6_sl46c6PVn{Q13Jji9;?Q5Rw@IsmVI
zGBR4?{339VXe~=u5EU_-_=hgE@J2s7oNv6Dnivk3AwODEJHB^BnVGw~ws7O%Jk>-?
z!ule?Z|Az&-uiS?`9yj(Ew>Acy|9N`9AL11g3{A&{j~<zR{?>hx>sEsnerm8^yc_t
z=!LWy6eX;h!r_m5l{ymc%G|)4>P5C_jGe>BCo^~MUj9&f{WqdOoE5Az&-3%_0<c?c
zKT7luw6+)U2n`JlX(?|j8qaeCi*mp3F-wHX`MM=ZUblY_>@-cA&c{-=$e!KT<tg%0
zpZt4N##c=0E(E+^PYM69h>qq6_P9Z@BU`pGT#PsAvCY)1TQ{rFM~|o6RKL<|*-za<
z55zr8eOEO>Ei<F~8g)?Xozu1Yjqf8j<?XT;sd*d5l@FPAK4gEt18TfxW>ftIB>#e*
z99i=x_By1RpeUFHE{x=J=LZRaAO}j$)0uWL+k~>c{#l@z+f-rbU-<_0%)}69dJ1Ur
zPHm)_{C49jaN9T7bk|Yrst5jI@S%-_2ztXN<d^QSh_lQP#|+ITxCt2PN<z#rX5cr?
zxhBvc1Q7s<4oB7hgt#ZuOif}LO%S5k5sr3*GC6aPJ-FW~p_K59VA^7|GDsmD$ROkL
zKi;pp@B-=~*DW1!bT!IXc$sEI{GvI8UKY&>6DB~^5dhKaq|$w~m^N{ts-Q+jDpR^-
zZokWkmB(%3>1X*<bPDyZmF}<IVAK#BVbFl30BPW~%%_08@%?*LzJQiKK3QspLNZ5M
zN_XDdvm`OR@dfu2W9ie9F~$|I4Xt}+<BUAI?z|{fOP)LBgOg&&r<bqJQPClm%C7I5
zK^=Q($XX*zQAO`_DOUoy4MV=R@}>dZV%!Hwp3wfX?6=uJa&AY*t^5FPu|gNRD``S6
zu?rHSHw%Dl?J~48kvI(fQuTKa1?A7<qQ<s})M1{Coo|2np#fEe!q(Hj>F-~~OwH4$
zPj^2dbo)}qH4i_W^j3J#a--U3;>}qnjT-b!n|}JFtRXw);Zk;eC;6DaOedmNC^51$
zSl|QStX4J@RyTV+XTGgqo;A0PA5`{rpQQT}oOUJGUUcN1dEEd0^HY~-_@aiU94yzc
zMv?vih<pFG_Y3CDb7oo6#&m2@u9c%&s!VAFtHHzv^SN}|LQ7jWaC4X4KGIr9YtBY0
z8i}{9D}lhZa0&)BpUxiJ@43Qk+g>2d5moc)nUX>og|zJLr#UX+s|G3jL{)K;-Q>51
zartxm@0rD}Nr5e~au0#SLfodfUcoiSQVy)PZrcifFvd>lhj&J=vl&Pv+wCYi^4jS0
znDeD}+Qr`v|1sM?ckY;oPP_~CBMOV6%{l!NnW7M#Qm_u8ZRfN%1-PGH3xYvI;2Qg@
zcdP=nneBCJb)$+hw~ob?-|zDB*PONHg}MpSXZ=QJ0vLS;nnmuXmFqjsGdsi_aV@@(
zeID3rz{Q^@7z%>a_N9GZXzkFxhTJy4FR_Q0d4|Lx-ax=|RJsGlECwL|88Cy&=x2hF
z8-;pbhX;=2))ww{Vr<yn*z_S7O8<pMp1!=-IAP;dp-?e4s$-7KZL(fGmK*76d~oyn
ztT~XjET!<OcDSMAJpb0)gy^dCY{^^Pp7LJzo?!?NEEGM9^;2?m@~N2Xq6E4)L7=xu
z8;yW>zx+zF2+@H^54gQA#d_APH2OALD22@l@eV&%(1STJ2Cqf_t6M+i%D#y=_xbtd
zl$w?x=4k~JnWJ=gyxlJ_P(&AUu+R!gEvaEeG8;5(c0@^SDwmjf0ItymX@w9Q0t8mU
zGjRX>W2tb1ScFjLDGb92tNt8XuxYc96@W2`sLTL~M#RRS(`5FQ$65oEGkE-1<IBa&
zvJ%Y<^cDyu(?2?({G#m<Sj5HE-~rND#LfBn)s2so2dI#qN2g+9ed0M46R10pn7UMJ
z^%nwsE_y-qpx+Y}<Ww&&QVK-YLbf2((=N(N9kFX5e28eMm3X{*DyF&N+A}KlF`@Jv
z*$^oG_>|4A|9BT(+Nfi_Apv+ZbkasjAA56;@^kPb*M~XgjXk?OXuR=f&PD$7pK=!o
zeP!XbDk1qGn2J&+<wk(aD;|Qpifmmp;8C|0bHKZ=-fa6%?a>L*K?qodVw%u_fD`aR
zyM?}9$Z?4MKv$t~8Gkupq-)R6Z{syWI<=i^%3M|u)De-~rqdgx`dvN%Rqr=i32UHG
z5aMwy0|KX!kqPvCf}ELR&UwhzpvwVA7Y595y`01;<|bJShQMZ%E?8P^V*KB!z{@NH
zBTyJ%TVLk?;ySv=NpJ-0MHR?Yq?4kVz+==XJy01KH6AikSZ`1#Zvks?6AK>(%rMRo
zKZP&`7F|r^t=3&wDE;Hkor0L2(lufe=GRXYoA5q6kbXJ3J&idJPB5vQL^B?jq#Fb`
z(bG%F0XUMNLEpOkHB_lWKupTg-?uuyS#4C~mn0omkosideS0VX6Fak`WuiyT@7CFh
zypGoVj&*J_s-VDf)$?11c?Zi+w}_oCEots>e^jBW&;;n4&7Qyd*}K&at;q`&CI&<V
z=iZ6gt=w$qGEOCKLE4oOyY$Vo$6(}iNBtd*-FZ`#NT1aOzPjuhuml{N-I7n~Rh*cN
zm5Uxv>!+sq+4=Sd>@2)cJD8Z#G^N<yOD=Ne;DVq5qdF!m*)!_!AKkKk!SFzSIpV^v
zfbk8R3n}_jjIc2KChtw%NwmpboIqQx4flZGh47JE<SiVE>s+4$-vo<x^C-%otE<O#
z3Fz_jlu>&3y>gcSv2T6BowzxO`?HP<jbzd5%fvcQa{lu3Rn8N+moMfw()lJ#a4avL
z=_$`rY)`iM#;b&j2)=Qe3cO~CXW!wveQbQqJ3sOMx`I3Eb2~o~{!fT|On_vC^+c#a
z#@bn-#J3GPi0+En**MDj#r>bpNsq}o#<o@J3YNL{h(#C4qkkL5jqhAl-rclJ>3b7g
zyX;7N!DgZPqj;tKnX8_QToD3^9NAJOX5)XJkDo<|6Q#Gy+_BHP6XSfOHC{{R=%T%S
z<*h{PW%oaEyvQ&|`}WVm|0NMHyK|LiN?rLcK7DUr*%3G&rLN@kCghA5Z82G*XSsR(
z*@{u!c8nAar~LZwo`nBvOUAwobBQfAp;|8Wgj(NQk`iwUgwDd}6RJfws>yyg2T8n9
zK5+c_n&MGgw)%17XP9@}@(hnfcVMIV0#<ePef@nWI(u&Zr#b|TEbV0_<a>+Uo$LN&
z!UeiW!1EipZcR)-lUC7yEN!gKafBHB1mMbk<v@6+fju$dLznDvI?|Fd*l|k88`4SW
zjs|hv=4Oy2m7OP+<9bL;8+0lrW;9JUn_Upq3IVX$H;HxYh0zkf{$JjW5_e+w`TuAE
z{(HavuW6b2f3hzrXmv5&PDWSd8GcF5ht^nc1}*o)O&3ymM>r*W-3#zWAv)~X#<#uL
zvq#*Ib&+_ZD}V)}YtiXprMr&9w)AERqGt`x=bP%zy?T{K{<U2iUyzJ-tzoxE`iLH!
z?8`uC;On>pm(iL$1_O6}&l!wdEjrk;RT;7ir0D)+CHMHHYI@Y<pNkMW$H?eo((1=N
z>9kkkQ+(b)eqn60L1)G}<Jb~Pphm;LaX6%9g-MST7?Ym784)M?z{oM9XC?-!63VJ#
z-A3kgB&n|2%I^cZbpC!5nIAXUSNA3~qEAV*x?UvI{_{LODts{oW*430NZVyRrcttZ
znx7sxN{8q7#5^$atuQGVpq$_P8p+{k-QH<DM3B7r;HdjWySEDaM~SRH;-Mb*8T^}Y
z1u*s&Pe*OwK<7hQcK%y_2n+Cx!?tWtAjP7Gv(Af$_-DoRY>hX@MfH}D*rI|EUgCh5
z?K67x=*$F1Y6~EC$s0V$jwsQkHoONy5MjS)S)BOqIzMye5quBP>$W2-BW1Cl@HgCH
z<=^>x6i(Nu!fAe9MHWsbEu%*ZwuHX8PiN*?=4M$fDiRAdQaq#OFFAirfOMI#bHV@d
zfydrDe|dxn3Z0mjvpavBA)Wv9P*4!5K)mvd223(B3;3^pC*xrryn2-fpF8b?c;`M$
zP5%VMl(S#{kCbg?dgq($A{H%<Y_J*{BHa-48Y=H<<()sx4)OG6J=Y|j2E^%`u`Opl
z_HpWe7VV!a`G@{0UX{Tf<cy3O9kXd62pg2Qk|=7FE{Gl)F0K)@cSH@GFqd9rXt$=U
zk>jzCjD7~e>p!$hfFhbr>F91-%;Nsb1<jBBKV$0qf3s5k|LGqrQ2E?d;>}qpMa5E7
z9fD<~fX1BHvyf|WNAxJ&^6F?ODG@?>atYD151R|7H!jXl$PhFPS;r){2KPCo@{9ji
z6d~v5Y~MoLfE-k0O#tyQu7B7g-&?DBMFAe{!E6(<Bzh&>|7{&8Go>w>+$jUkwCiZ&
zqTh+eQkuuw9*mEL%8TAOUV9BrV~v7TM@lF%2)G`DVlm)`R9X~xME4j8oEx-~@M<81
zze%Db|D%*a;E+~IwhI<6oB?5j1Qt6AdK&n}@oYi1=1DS-%~u8}94%5<>?ZnN-)xkX
z6jqGom-|Whny>~D&{=HrIWAd!9Kgh!@<HEa<cA3*la|&!xDI#^7@+%c9UQ}Vm&AVs
z^8=5hwsQ=gQiE97heboj&pF5%eH!x#L9WW@D2evK8;`pTe<Wue7Dv)X<I7o&yvsB#
zLM;p1-9pBMX$ox!dLY^iMtmc?ZAKxGPjLcnMHF05U1L>^4iR^UU*#Q4FiW^XT9#qn
zQ*$0>YRIw0tD2>~kN`%e^_@}<E*i2Q_VHE9e-bYT+$G)C<|h@7o7+dEpW^#NgQ(S9
z!-7GpK3j=i2n)-oTRD7zya8Wu;5FrwX{6-Ja<1qvSo0qk2_=zA(vdM1eE)9-%%U>5
z&<Pem-{h0}pFHjot-5r$ab)keQ9eM)kS}gX_mNuqxzPnenl_8Xo_~EL`vkHt{b>yM
zOXTj~f+Z!c{y>rYh#VpE66w-l5)fJvuT8P#t#Gx`Fd{$suN6dFAfqSb#uT2d)=xX8
zRkTUKaFQ0Kpsc+3i4rQ5?P};8v9+Kqu$okFEufOT7I_!uZ`se~NTb?GK{GQlFjHE^
z6pT}c7vp(Ys4r{&jEtY0bZqrmG-4_ySKkUg?{Xg3-kqfnY_>`?3CiLYqRO2&k*J<$
z_Z4b1*zjD`9R#;mg0ezAZxQw7&v3<)$O3tM53XoJENrUyY0!xY9s_<lmgAV1a_Dzl
zfkd%{<^$1o6%!NF4i=TWmMX6A9oF}cz)zuXya+Ft`s4MNtLv*xL|+mcRL}J5axdNw
zx?M3to;}f5K`ViE5VaVDTaVgBZBSjOv+@6&(G^5hgSKyzFnT7$|B>4>$sJWjj5zuG
z65&|_cg^HkeHVkgEBJy?7pZdg%Asa9N2`^TmZJ0wczw}{8o8V|t3}~c>egl|hE&d*
zMhG!dz{@6~_-?&}{-LVuy)Sb!=FKahER)`Xd!aBr^zq)xtcM4Q!v6!gk$UxeS9Z`>
zr(P0Uu7^(F_Sd5yN^eb?`s}5D^-6+Y*Ct(NE8S6J6AABQA7QlOiF#L2)8SNP-VsJ!
zv(Z)7To;y|ai_lGRw9pYyJ*kTDs8&Xx)X4!`8gi-<Nt4Cd({$JyzO2=(jXTfQX;Bi
z%T7o29Ru~YjrI4A7x}q*ko;ZvVIk7;38i5$C|+k5(ygn6JNA}TJZl}#8SL&hm&39^
z+$^n8yhkQaEJ!y*W=pS&mEH+ub#;51Vnq{HL!aqp91w&IYanil!-W|d_~JZ=DZowa
z7ubxxc7cQ4`A0v{jQc$)f|<ED6jI~s#3MhL)8SKKu|v%$25=~RLdyk{hO9xRNlsuY
zalUK)h}k_hQ(26cteqpg*NR|Ll83py`tuBRCQao9{e-T`7|YPi6H2Pj>j_LX#fcM>
zwJnXpM*t5%SY@AhcwhVHO<Q0k0;k>9R#p8hnypMD9`~hTaAUX|thZ!^KR<PlGV#4B
zgP1Zb4!^30_#H&%JwaD0yxerP!_YAc15Lf>13zE9mX}0j;_76vtG~s1^(_@G!vu7{
zf+f@_M3t{3yh=t0>v57w^6{EQKW`z0kRw&L(FI7h$}{|wY@v=;{t+exZ2CnXWR!Xq
zgqq%n=&;JAaqAS&vET$akwd3`Mt5+I=!O9C_@qDUMRlO`ud(-pZrS^^IoFmOXR$*R
zA#o)jAlGsE{Pf9%OU!$!c;#jL;lp|+EwxVH_d(Qi5v*9MHtQl<vW7I5nyT-z>Z{t>
zH{yY~%xd$%gfr*k`mtl}3Ie00o8J2Be;gh99RKP`R23adB%M$-v_e1N=qq?Om9eYl
zBxkJW38?RfTv$%M;0KhpJC%nH)`;dD3_b-Zqc^_=n=aXsCj3zJ!;UDbtHv6gOcMqy
zCGAvpkYSsx3_r>F(A&QH#5DVIQGfK-u>uIkU}Wj{o;z+!F8KJ4n&?N4^WYnG-{uR#
znrqX+(yYbfr?!!uq63qZ_+eZ&wYDq#Xg2%xzUN=$E=zCp`4~sRd_-&IX(?GZ(Hlz5
z&1&3tU8EPAF?g1`*GBD-VPwkvJu+3xqiW?ucly~gIDSVKa9zET)!TlE8)0DY57JVy
z3n7yIuBJSo^fSc1#fd{Oa`D4}jqH2c(0bXerPne2Lcin3IK@q(E@Wzj+YjZMtCQ!3
zMd>g-4Mdx{!hVo2UZea%`=CO~dw%AG59cpHtR61XaU4auVDv*E_g^usx}%oXt-KG<
zI^pbcM!o1-{t!M<TeB$<icag=Mta)1PpDL|rIIZAf9xP9P;Hid(mQ++-w<Iqy*u8z
z_d|-b+X~kQ@A-1nJJb(I#E7Y8>oEJdO)>1jwN8cWF*-H3{gHb~Tpb&qa(VauQNuxz
zRukc2r1F7FEN0EulxStH&UEJ%98*9NehzOS)s${xm@y%fx$n29Z+uEu{k@^lxv2qz
z_r%lINPd^wYE2%v@W=z5+$Pv-A<J>RWMamvgF@by(B1^L;lql2yna{Svb5byG@lI8
zBQi)xD~a!3&T9J>D1ZT>XIj6n<z)DL(#+e~@70Kn?~iO*`uWfh`=hSXT(ZJv$oU5W
zDY$uBxvwhJ@q#5w0^8=06u@?`N#DfO*5}-jD<;XgLXY!J@fKq=m&D75V!_v8LU>1w
zyLZ6#v6iY~rh0yvZ=FLQL=M|Iy-zS+uGhyJKX%&dGw?}|K7$Xad*8-+2>kf)<~L^7
zO37OHucxq7Ngsxv^V649Ys>&^tk!+Q`!2#!L|CpAU*}?qI>*np!;)NU$69eg6Uu!P
zMD`~7_jP_OnLmHuwAf@q0C|{?!|#vI(Q9!q|4@CU(9)zHpyX)IC);;uPSAeJvBt$o
z+NwjeCsTGw3*N&P4W$T9Y_jV3q1WE_d3)-3AYZy6B)HmD%+UVHVw?0SVw1Rxw8gL-
zLuec1Ksrn6gzc;45A<BWY@G*0qT({3ggNN#R1mAKg(nLmYMLkS6~z+k(8hD4@3Hf(
ztBj&r#d6af;yu!Kn7ZX0a_{Kei)vda&?M4-)(LD;^DjHN62LUmo+R{+moV?fOl+iW
z675rS1LWnb4uu~$@Ego+?N3e=H>AF&GJRlYQWF0iSyJl=_F!+vNiQEfGP9g_`T@@w
zN?yk=dGPLq_&2~Kd`ns{Q6~VpNT-aImei&ntfQ^8<AnF9jrD?GLtErY-D;$*2eGYq
ztuz?gSwz1Mfk#DC3hfVlhcr;{5R(}e^ylIgzGI-CcBY<Edzf7GB69sqaQ#&DAGM9b
z4?;{8N}uh!UH@SB-h=oiqfM>2A8_B5OZf^Q-qk+s;5mX4UHdhlxTY9@t0>b5-aGul
z!%F~qVdiU&&k{NVrOyYjs6)CcZ6XqB>FD%w21zUkqa&h$KDO!ToRw@pG1ZSpKs;DW
zbFK&6Fnf817+-yUC1qbR5L`SM3^wXMSEDLX{~%b2`!3}@>^N3^8KQhmp^9qOWEx8e
z*R*i0O)s+Ho)e9eE-9H^>ySpIqTMPSLV+gq(kN+3p+<8xMv-!Yl9Z_7r~JvA=3R`d
zNtJ_?5S22(2tUUyoUf2)4{RTE;G!p9mpv}!6U_f1+6$2dh(W5NUqgsYJ|8%CtUv&y
z@jI%NnL8=$F4im;sU7B6fu|#ctblTvi*lqy8`q1Nf1wo9Du%#M%t~(-?Y+lNGLGT|
zM)Z`t{7#4Ck#rTX5myuPo}?uEdAnQ*?Ul8uimd3Ci&sg==Lm;X3NKtJw|<*0+6}Fe
zOE(;#@b&ZS-Pe?n3VsxgeD?9~7B=%IqKu?nNU*eoiE1c^;a6#gi~sX!B@3kLpC=6K
zfToq73P8fbryRpJ<$^=?qqBV6!OfzW=^Jw_%?V;bR5So9nwP)gPxMbE<8R8iqCr>K
zmZItH?@b2>m7o#jF^GU*w}c5i0^7t7Uy%j~H<LKEszZNrUvR%01`QLy9lg!S3ukIE
z&%Q!P5=}J9(&Ij_5r=&lP{&75!TcLSO$%+L-%+h(I6o8Zm$(_x7QP+vrQQ!pir88!
z2z55L9ag=b2CJAjU~nvcAb_9mjg~Z7$vqZqUF8{bO4rnXDoyp`3xSguG=2l;_}!A{
zJNo`@>fd!{o=Cj8c}p>KCXoC^=PXD-csG(hi-~6xvRS190t$g!WLs=TQBPs-s5d5~
zCIa~KXx}k+PGB!lgImMrP%nhHV)X47_CK5(`3r8FjkF<4ejD;tIFLAFHv?}qEagvN
z3;T9_%OQtRY$}mclQ?6Cj@t&xH({qiwA%zRAku11CVy*IJel}p6jmPXuI@t3n^N4k
z2Y>NH?fAtO&lVgf*4rE3jT=3A@?_yypTq8&#;Ynn-(&dr&M7GWP4E1Fn5v3QPagJy
z41U?^Ikgg7CKa+-QB9>4A?IC1jEjoxA)$Q#$<wE7OXK2|IEfv|8tg%zzTlz@mPzcG
zB^-gNc<+tJX0nMpM6KR|o%4x$K5@~(-nfgzsGV}}*iQfZj%j(D#zCv-*o^ci!#vL(
z#|q854sqweBkkBI)Ls`mtl{ZWf%xUBlD+{4XV+;s?1(;ix0^)JB2kJPa2SK8Vmcp8
zOvRK2dTgT;KAimh^qDjC#bt4g9oO}0hb6+&mK55EKOhwRTf;}B8ZK!iNC?r$#Q}za
zC<R-RF^C~M_&-gCcj0vsGqh+U^1XPBc?oaHuCR2=X{v$&{EQ%p{O?S*q;eS`<g2(~
z`CnnM#|Mc7j8A#aE+PFTJP;L@6SDkn&7HPJPE-Etf_78KUa_L*pZh+HZv*uwOl{~}
z!Ou?M6WY`)`NDsPd=lSZ0PHjSLGw+g-p;ip)B%LHf4Z+GAl&4R1V23IR_6&sTOhH=
zRrhBs?Pu-TX?SA#<m$5!+y<Jzb&%H|ak-3`a?poFpVwr+5zR>$%F1rn%nC+qIibkw
zp6ER-52sb$r>Ru*aChQ(UzX5Y@!cNPuF#SjXuXK)xGcufP-9p*_%$G(P(4^tQStUO
zVC<sflg4UmYa5@)51p#U2oFndA%AvwLp1$p^Fs}Fsh8~W?Rs9boayfLO*jN9GTZuC
z|MA*$)YpYl8MvlzHne+o_Q}B+Iu1SMwB+}QW!rb@H{vwXoA<DYfV8OK;4-2=!Cx(E
zD<>-xkwpyKCaz%>Ie;!cAn+NF{Zuq}c`L8|^*HoPTu^%41g_AcC9jCgtBdu!v?}@~
z1#P6bKDRQy8OAlVT-Dpsk*iRcRc^|GBy^^o=I!~zP*fW}?zY@c(hBNWi>CnE8lRLZ
z#u#v67Dq++tf(KK+aTQGc=2|(zoA6)O{wDIac!(gWK$N*K+`Y0p^)Yr@=n7X^w6bu
zT&MSE>t^}0qx(Q?0h~w+T>{*YFI^>nLs_MA{8f3tzU?JeZ+^*6?JB34E~gbVZ&b&n
z#NEaTd@7l?FeQSy@Lk_QKOuvm@wt1fH)Cr}fv`z4WBixOZPYYB=*irKXD2*Lbj^m^
zf4m6;x&q2uXnnqRvzYE&;?#r9x;%@R3()(rYW+Jh7a*1*Ce;2v;LdO)*oFXQ!n5;N
z$2H4raOM4pVh0_TvpxuOs16#f>YZv}lSRUIae~l=sO#yh0wPC3FA0_6h8#gkK*A%=
z7Z#wrB5`zZU=1W?xWGP^n^9AmnhFN;YIz#rdv4vdaWc=cN94UZxqEE0-NR9-X<7_i
zzV+^?Db1;-M#G4lS2+;CsNS#czUk7*SOskb50y>*3Bi()LK2F>M<{~K2Iw!95Ehl_
zOMf%=K6U1d@YH~!*X@1ol+6OqJ#xafh8G@H+u{ZK7gG*e8wRv1I>EO$zpt&{#!dj0
zjH<v<s$3L!NHIim4&5b;64>M%TU8pph4dZZ_Zi@m8P1c4?zyFT^Sz_jbP13z@1wBb
z*rARWW^9*G^tpo0>t0X>hLvf%pdb$I6Q9Q?Plal%UvSgzU8gg8xSMT>s~h;628UGx
zwjEUCgN3Z-UQrQw2AUQqee_P-;Va7GOOSAhu}dqDy%&jbQ=B-&2=7E2bR3!_0R1PF
zXK1j52cjCo>dd|YZ%f5uG^<>Nyb`MUSF{TURcvvpf}__+O=Ogk$j#Yp*CF_&whQYk
zPMi=T<Q_G(O?(G|KHhg$nX>T}CGTmwynl&)n>snTRq`6<W_7GRfmu<2Xu`xiNMgQJ
zOlg4*GAQ;UJJC$(i%Whzrdt*m#<6uo(>Zkqv5;()fP}B)t_-Xi@a%}V1{gA@{9fQp
zn+d=86dwPN7Jw`Gfur3?qiP|CSgS9+Ly*J(_`)GV+{=6oBEG08i$)M)7llb&0D?CQ
zZz|Q*S7z>!6QVMT)W4Zpk&gqqe-K$LAU~jc)ku*^p8f}WZywe2{{H>4HyfEL87h>>
ztk_f{NoCklri2U)GGwR}8Oqm`1|>roN|c$6GG)w^29XG*gi3`B70v2AuI&B0@B5Fl
z)>-SEbFZ`Rb6el_`+k2-pZD;3U9ac$yq?$dnssrZHhG=oabQho*`wp%*R<M80!;Nl
z|I?#OoFWIj)kW?@^5FQ%1Cf+R;m1O09y!zZrjXn`U_G@T*#2B))piQx8|xugl%}s=
z(V};Hka6f8V;{?sH~S}eZh>bbvNN|g9(}qPu1IbG&w_+^F#(Q`nguK-#TI^&ztc8w
zyN<&x{=VqOc#aIusQw2#vutS(i-p$b3muA=ot7ScI!7$L>ld8LChaN!?sE3>+@Bw<
zMEVI0$LknL>Sxfe2e4i=@BhdQ<$&TxgyVh#eQ@<!Uz^0;_@uxz!hZivLzMs?Mm5mv
zDRX#9jozW|ZW;Th`^w67ccoJlDep-H)>5j&{fg=W2^Ey*2zDwdR({!2g}iCaw~QuR
zlTrk%qB||eyCJ~I6k@BMpOb}$?kfNCk#*v?3m+K$r%YKkTrUk-NjpX?Cpc8o+Dd=R
zEiOokI=n&X9dW>roQXY2%_Z5fjz}dl9%Li%IV3{Q$<!z|!`NXa>lfwCi7Q6!RDnS9
z^H~Hq*h?LmQ;)ffZ%HXC=u0TIp1(}2CEjgkuZ{cbujE&g>5UZwkaa2bZ47;?uk4&V
zpYdhXjKv<`X*+knmqUqL-xRR;>5)W;6BN>+c6DU_MewpZbsp@Hf=aKsfTatUZm=Qr
zvRnR`Q35mSP-{ljGJ`19<)I}_r3CHdnon`m6?Wvd{Nx*Q$GD%rio&${{wg^N^S@b?
zn6+~LqxDs*lD__aeoNSsycQ}fZ7&+5jg|Sy&Qr*FR@-s%(o{7JgK?08iNQxUpUgYb
zScbQT&F;x9)nLJzHLkB@SW;vPaG6(2($%MQHbADutIzfQ86tHjGw0lBeaeIK3BCu~
z?mtK=Z8+rHR`YDuF%K(ZrJJadiZXb2g&}%m(8<f7PxiP?q0VWy`bSPS<aFWjrsbTd
znpG{2otpl46&)>NJ|6Gb+i=oqEVM?hUqjf>&CQKed<Hp>I`)qr78}mZlu0-jA#*Ip
zOIk^p&CcUg^Dz~hqd_rdhW1jd`<U_sxJ3AMYGUfIN)<ISeA&|@bt&2Fn~a?n*7Emo
zhmf8o*S2kIbG|{(6Jsqb_A53XW2E2Nsk`Y$qg@?~nyG*6J3d?4u%+c0qlj%C`)%9s
z+xRI{N40HgqTcQ1rlRQEo?cEmXNS8wI#xG`eKS{5_b(R)H!n>dTySu3?}OF%iQtBa
zwL0dm#Ko}W3`w~|t*IEyrOjsz)F4&9AMi~Gy{f70Dc~I?IpzF~Q#;k*XI?KupBBO(
z1&xRdo+l7E!Rp%>zd;%r8XsG}6)5wc7BcsJfz9$GNPDFML5EMC(i_e*6|+pbj`O@$
z`E}^uXNN_u<l6}qG_OI7Tcn4aWe--k?3=bhj8O)z`$B)H#&E@#!_Vb)@2BOlLvM4y
zmgMe{3wL?nPf^k7(K>$lW@r61-xsD`e{pP5_`!mPQpMDJ%Vl@UJu_ykT6FfVkVoKO
zKb>~wZGz?|zkzy<Q&qg}rDAv?6=qNLsNW(^2eFfs-Ch_U0b>#Shc29+#_hsqrLy<G
z06{$bFID)1hgnQf1&^ugl!8xE^cxO7wQl4)V*8i0lg|)O>)1X%jy`A@o_|;2<<jqL
zRw$gdmbw=YyX)E?DUg#;&bB0ELEZ;ZZXc;js<Tbnryl!Ann=|iB$kbnUYfNm_);}#
zB-nxho@nIfx<2|7SE|QRrF;qzNL-oH&#gW}mvoCxIz1M?OW@;(Eq%0N4T)ef!iXq_
zqt%03N$3dR6l*zpFrC%2diS2Tn@sgCk<(8`3`II3(?%P2o!-Hx>wFe89LD8fqENoq
zQwJ?YeXPIf<RhWH*Pm$Qv&PIllx;x$E!Szg_N}ucA9D!(>VNG!?Nk~&V0PHSAJseY
zC{vy{ukqND9d;$%T(_;cczc)k#w3-|TXq=q@XT+s(;#nBj#<a!Ao+ag$BfRbehn+P
zKm-Mo)0CbP0=nd$X>Z`sV8a;Kn%${DOPN+lNdJf_!~t?E!Y2<#N;a>tp(hR5w_e;2
z#Q{t=jhnyBs%xh?o}$S@$_SzeN1p|?q%Vx91r_R!bij459l2#5iIHHV{9IvI*xDq6
zgyRX5CZTtDLvA4irr+Eeb$1=*=@ye@Vub;UF6n=Q?jx>%DT(SINayCDqEBu`J+$Ke
zK(Bpv29;jVWx#{f>q?)w;Nk@GsG!e$DG-=dEdUCLclXj&5Rnc16<<ivK2>voGx2h;
zN`a}#bL!+yOZ4A})y9O)SKR>B5Kb7&45*zV%>@Gb>4Pf{*P&)4$%1=0d&bP>{GS>c
zxiU$^ag9wna&&f-&5-hiW??xpsz$d9dyPji^S1WS898P#9Nzqv<B}q-iE0@kNycA<
z<0XY4WB{iYPiP=5uQD5;c9Vh(4m|TI9BJ1lsMi$7+d*}!4WY2t1$dmPKtX*s5P)Jy
z56J8=oY5bC3u`DuGDxJkPRwuW9tqA``|d^U(c7$`?=iL}xqA<fge&(02dP0tUuCPQ
zXA;uq067d=Unh-9cp!~sW!Al1N*Xk1*6YDvKkSxRh9Fbs>^o5&<M7(&C|IMog;-dU
z5ci7C<n(D%duRs#KOW0t%6jjpj@Hmfsr!7JX+-n#Ei}jPuE1Dr(>3s?q*$aw0R~ey
zOsXJRmkt^>QR6K#m|1lG2IgrSntz)%iw5(G{5N94&Iz);^K-7L%dHC~a{KN>K6#%N
zykZnPvnFCd!y59P`y=~)bc~g3!g^Tw!|$A53r1*Ll{=XIT-o-xw6(h2K5=D1{7C>2
zy%fnq-@-<e4i)Dpm(o7(Q|YHr^7tUevC)oCK8BJ7>)z2wu9~|rT`C6IK{}`4%@<;z
z8)bITz{BOm6xu3OYkz-O8Zn?#r%qz^6dfiKAoW{oZ^mG}r5~re{AH4UL+P+ealaUo
z?V;IvzER?)<nHeIJ1n0h8fWWx`>T&rYiH=1u)`p<LE8sn_s_eR(RD%_v^WFq1DY>A
z*d%bx1-d?xNZ{<kXA~tO$k(#%;MTBRM)BY-GE9v2f<-VId51|oM&9`Lp0%%((qq5O
zd(3=LxQ~gLUCbRo@QV9h^oahpV4_GY09Y1RU?2k3MqGpSh%$Rv(B2cM^%phu2OYe~
z1oq-nS#cplvXu1V%_Zfhq-3&hSe=aF6j1kSWueR?Tka=zT1cq!I!^PXYSklJJAPbW
zBM}xkm8al=y$(p=(Qz^B9RbZ4`?cJhSpSLx3eE;#AZd6@+cQq6^4UwjNzDP4o_Q(a
zStpQBeK?Zk1+xm~bIsC*X`h1N?kB>!baoB<gXc1gxSU`eMIzSwTmKmvCO<b1B^h=~
z5qpYSQd3ja!{>Y5()8SHxjaI0+9K<jV%kWr9&l)*K21!$Kebb5nM&m;@~q?jSwWgb
z97+Vdp>-O)Zc&}x+SSZPZ<rYxndXK8*A|Xa-(hgyT+O~!{OtTTePZ&u7Z3K{k$1RT
z56{gRS%i*v)AsOmSAj2OBL)r5p(Un<s4o_A9@od4nm&-do=`>=28Gm!va9u*%_qO2
zcO+H4DCAd0X7{e|qFb*2+*<}^7SKtZFra>tDh)+7i|lb|Aq-w^eR=g}!_wUh<f)5P
zdfQ%2<WDkwD?e97`QD?H)66G3CPu^i0x4c8uFX76x;DJ@j{@)yB>&*!bJgbx_u<qZ
z;>hylG=DQrC3z8N`V+nc)s_xQ<EYT6D<Lt`C*kDdE3q5z&)z2*e8Gf?4x%yH+0yD$
zp=`AtQ8@)h(F=*26L%-3bbPaO#%DJ0!>U{yy__KooYpho>IQS<B(=c%01HP0oY18|
zU;f~i*Ua2%9oK2jK@*$UY65^{uDAe@9#Xv?M17dU)U>vfeV?L~eLh{^@uEuHb0{%O
z1=ej}(0PA<f1&KsqcpO_DC~ijYX0~%y#q!m+0H#b>GX?Qb*jZ=(ABrxGXpJhb{OQ1
z&0mr1(CYBU39=)bj#D%DZ|d5&^mI8nfJxZGuE$CDD@bRPU&tmuF*noRRlQg120Io8
zgzKz<(z8=}v}W+?fJJ63c8=4H0l`|9IiU4N>u!*Tq#qcsxtkN)VdP#;Q9w_r-pqz0
z-)V0yIJT606gk)S)WL@5UI8-hy4s6cB)vQ1h2CQ-pR_LY?-u3Mk9)3*G<%|WV7AY?
zebPebmYS&Bl4ipi#~mM}2TOm#5ObMCWS<K^8sc_s;V~>4&v@0<28HEjsfZXasC<Lj
z&rGygU!!X)fKg^;9yr*kKODvJn{ch^sAC-Cie<-R=#mwOzM#+2WeddyERaSv#6j^h
zx(~4`Eh*_%tl{q*;v9fzX&KEa37yV?KpxLrvxFdYaQU}D#Y;P<rB9jC{yquX-N4f6
z+aIl&{>1lM+=P-H8vBg3)Lok@tt_#4yU+N<hMfW^NAL9BxTB(N{tL#+$QZhsbq!@u
z<hNGs=hMX{8wchnDEGsE+agU~PdFgWRqx{Udd83(`oBKRi+UIDbH1f{sDR7lCZLZ+
z81>WHRl4hV2ew9mE-9*?Jm;AegO?Id$yvhfroIbY2T*<-<?3nQ-@w6l&2iTn@hEGe
zwz^xU=G{-(4?eJE9$x!lJ0q?2O{34wU%pfIz5Q+E^7%nsuNoWH4#s{@rYA+aouE|u
zqVvS*)^$U5@N4Svaq9curbdkMk)F86s_P^B>~Ju(9b3~>qi*P3PZ%!tjwnq>P}WAH
zh9Iazh9lucoc&)#L$=6Pt+6BD7k28~<KRl=A8(dtcbKl9Z{L43#9Fu38usfW#FMOn
zw$hM9;{f|shwTROaL1JFsMu#bw!~soJHxvMA{azh^Nd3`LGys?cd?oV&-poZMRrK|
z&*}7TjJaBaZmBAU4&T#1B{@7YRR8)Sxo7;WvgiTYo!lP&xSvKS42Lh<7bqegrouuV
zcIIoIRMeDFMWpO3wpL1OUrVobU`Oo;LWg6R+jK(7ggt%s>^6Dvyn&t9iN`6Y%OEC(
zm)-&6PvQ~pt{Y!_u3g=pPS2YUys;xA_Plb3Z8v^irTeZ@rL?{&)RaQLo$ICBwrQ0E
zsh+f|+O@5!(v@QX+>%&46LQ4a-VV91dKWOHqUQ}jRYEJW<8cZrz<8!*12>xVB1@>H
z+#L+6T7KM$Ek7^UXZPE-fBfBYGIv3)XoGl?<4#yS{<0{ePE-dh2%8<ee?QHOb!X}d
z@x}@ZA0DIOaQ5Sli2=*U#d}k1+}gF^^U#fF9F^{L^->$F=F-Epsp<h%_nSoH9LHAO
z6OHeTt7%%gfM!0Uxiw?lDDc9TX@1;UD2(s&eLEuC>-`I~BNueEbg;L-|7AyR^xxK2
z3m=u{+%Orar&me!;qlA~<gPC%yW%?J0&iyp$fCAMKhNaLt(A;Q>NeA=aE!Vp1?F2E
zszP3Jpc5|#KWI{t&}_Q(d)$|*_LYt8U{=?6-4(aX+NDE}>v#rDatS?W-U1LW48p6O
z2M?-woL<xdI2`EYZ>b{FZS8ZR+b^8mqjH4c9<*I=dLsq<>N_W57yL&YXsMDu`_`_J
z<t0?egp)w0l$4oQiYMeX-BkNr&b*9~n=zsim5E=`RRaB)KyhiQE#_s^0@bZY1=X6R
zSvytki(b_tv1;y^>=&OB^#9h4Z<*n{C9iw@VOh<CU7Nli{tsf+vLIDThYr-KJ4uXM
zNu=dW63aBtV!yd#ZYs;PS?9A~cJkiX(RlwXSi)2$(a)*BRP`M()jqe`PJ;%cBJ@0y
z3L9&=JUry&((it+Gh`bnp?cDgdrf?G*FO<joTNgMFfh8AO>FeM{-|*X#%o|=a}D3K
z{<Q<d)9UkIHLD-=YZJ7@xkEzyS8<+EUi9cYH|Pm{Q`z<EZIhdUe?FM3)>zBE_-KPa
zozlO_AP_3+VvNF~ALQEP0-<9l{+eXCY|w*bA!&xGz~XROqFSBZ-fcUvoy@3tUHs0^
z;ia6uo!qMCUv`_PV?$7Q*Xo^36AjVe25oN~UErD{5xBxx?Zsg6f`LQbSE<Xv{%N6)
zafwB`fKRn8yUFFl@(y<txbkkV@1O%HUZE5Pb`gwduFN5xMGWXB{?_@rV%W!!1zi+s
zPM`TxBtNs$2DY}<4cgVD!Pp`H>C!hWq2AwW{-d$hU)w%Tsajv0wSQ(u+`(koQXe20
zmf;4J+(n2};aLe(oXxdU;SgJ0h29&ezC)*VGnjvzN)C+KXQZ~JjINA?WrvGyHmNg5
z=d@72K6wSxnx7kL!Mpjk^p(%si_gp-zK8}Xar)%Do;z=$^_@%zw`KhrJMa|*M~=a;
zk1A0eU1K61C!-nebK`(}bNge9+bZ9HDd=RkCp%&Na~MzT5F$sn`e@c}(A`6B!I7R{
zw!hk|w`fRR-Ju+5D@b^d;FTYLlnt?^s}#2Ibg1=7L^97kz3X&Ij3=&*`D>X{Hf+gZ
zi(O6j8HXwiWZ`Q3yKq%lxI1{=-*6b+HhZT7c1BT?&`0Fbk5fxB?XbiUXLZb7(;=j@
zOkI8<i)7P4^^ZNrTK~;hiW=RL2lu`;xvJvWC(d=r;&(u#zc|fk-z!7p0)|+R%#VaP
z(O`E_kTwU3K~Y5e<In-#%9COkYz$`Cfy3mbh};-Gh4u0q+v_fu(=RNXa{eY>%9yZX
z%h~Fy?#o<-G<*FkJoG*N0}-XK30?BT^rp=k^bkhj-jU|hhC_-rU%q%oT2mRW8q`Iy
zABEmGvY_Rv;4mO$(N)HK>^3<@HpkDRF;vR9Ri6aY2hOCQm2-QU*{F(|+oZ!NZN^t#
zpjf)Aay4?vAlL7<@L;nmoS`4ohxfSqfA&yac&IW1v<q;#ZSK$z63>eZQ`Mul&ViSd
zA-Jvo{QkrzOg6lwX`n6dDr$IgxS?T93^6|=U~7RvevDrl^n+_xo5N$x&6fcd0AdtV
zz9xfsF6UJk<jFigovynfK!8|*#fP<Ok81(bn1a5)@-0ySgcMlnYx367u-e)_mz}I&
z|E)gj%DR|7EY_O?P-chtJB1^=ORJS4n7tj<=bWq4vqR~}$Z#JG)y=lgMpzwCfIU#&
zauh*QvC=*KUg+(u49v?I340RJy635!?dM_6gSNxq`Cia2`Fg~{q&AoAU>&KIpUzg>
z-9IMTvXqM_oty*UCb7}!8lq)s3)l`?-|k(QRY{vZ$$3WkZH|wAq%zuL;y*ywcRG}i
z!WJyp)iP#4TpwCHJU=B|x@rhgLuai8<%<_=LEK{VZ+Q(HwDZ+jqX?6;t+;q3e20wb
z&d+7srwob6(BHf~I+NcdBir;~+@wifsJ3m*cDlJwOU=5!o=Cm5tK;jGzvpdnG(qRO
zC~8>68dO9%DuT(_)prNa6?RVd+>AydFZ9Uu%4-pFqY~S5`LX=ZS>Lkmoe=|z%3*`a
z08Vt>S~|3AFHfo|k6ySyE6~?)f5ye{NKgecA}T6BJ@@Bg@N1ANd8o?O$z>#^2MW63
zU+qG#z0903Qk7wEL@K(w^k<2K`#q)5_6Bz)%ZldvbCZYK4C6lK{TErMdwFMUwkmIR
z9o@ID)3|-Mcpe^FakPTWruq$PLfuI>s7~H(A1b5<eY#`H#%*3TC(-!W=<#HV&VAC7
zyYCMCCsllRf7=R@IGH_=Ht^*I87+JenOZ{Gq4n0HNds@2W%pm1?IE=Mz{h)zMd;s=
zn5&h&A+iGL!OyzdH+9YRt8o9{(yOub|GBumJTogCp91ZLp^WXw)4h3AJjvqq%qR72
zWBoPK|C-s9|1@1bAeLft$5mcZW-CtlhJ~AGtcdfDBM<Ux^inAnnD~v#%Q{Wy?vVCY
zD4eswy;@xtvE`&J3<X($<S<tVjiizMF&VJ~kLnkCylhu^O%DUNs2y3{!8^kX=gD}7
zElw+>1U-wo6f0Wl+c0nPfWY54v&>_=N*3|knLqQiG&Jn61i(Gg2%3+@esf`J1myld
zwm(7)6V=5AkzB=o_-V9YLijfwjCP}ki%%fimZE|vZGka!=^g=kzW}F7IGA;Xz3$lm
z51FTWuuB<vVw+A6gem#*1j_NyLJj4qT-VYsp6~3eF|1$&hElCFWgn3}1UUXghuJc=
zt;}?_jBV9S^MNqqQ@U%m*72@TTAbe|X~{pBdxh;7!tWhv-6qZ$Rwhnj59?FpkoG|`
zW0Yp*%wIzK4ZN>@IJdTh7vJjZ&mGS>2vjpFX!;>gj&NBoqtYMY%;BPRZnj?ioA0OX
z+c@a@t<BW*hg&sooIm}=^4I5Z4?46ga75XRm`rkD+ZKD>UOy1VwakK1AS4ms;5qj7
zlOPSdV&b!y8cWc6v?^kHojj7(jwCRSM)hB;nosso9HE>^rtZxL{@z((i>$|vvGWY{
zSAD7BD$#%b;YJ013CB%^FN|x#i^GQwemfIBQ)=cXG(?V&ak<etC=|Q<VvdC@SI+P9
zN=l@ibFU1Cl0h_qd90kOXFkJhH)~SL7VYz(X)!{`+74AwY{mGInoBy)7t$iq{g$k`
z;f_k*N5{u%E;@oiifOBKs&okl38TJ6R}l>)9HjQ1^rF9u)rksuOWSWOOKa@M5u#H4
zep~m);!8!@anm0L4K<Pz5@v`ySZvm8ksg<>xT^B7(1-xg`D*qj7~9e|#-5l44f=SY
zm_Zzh6B|%DZZ=o^%j!xE>vn;VFBL#LM32a%FKgzL^Km~O_w`6f=uLgZULi?EN1?k%
zf-&q~kNOm#(@yX9-R`1H>Dl_7!>JijIjGsY=q>OZD&1Ue{tMf8_t;kDHu@O1)4`!f
zXgBrPep06O9j>*ZZcvvIzWfyIkK325O2()EAT=&WPf0>AH9{G1Mgi<z`<7pL^fI8g
z<hul6VE4X232&(;%H#uL2_u~PuR<wSv~s%r<Cx{%gU#Wt+rJB^j<BXsT=|uMe3nrf
zD!S*%?*xfaAiRPw{ado1D5GHPDOZKz^SGcu=)~+(cd&?V>ZOma`#wgL+UTkP66h=N
z5dI8D$OV68J;~8u;i&e$R3yljtD9fv5?;GJ?9U$)S0gcgp^C^93;_UClGYLJ*)_t=
zLSl%+5EN^rr;f`CqP9ckZ_XeQ%a@Cj4?t7&DtldQWEx%m=wB^>3?fKa8=iZ^mR8Er
z6}HP>BR{Zi?jG+ARmgpM_4}e_+q<f%%mKhvEi3zV7LTW`hN<G@v6gsdYGl@RWCK+R
z2#xUX-qxSTh}H-<!Gi}qMMcJkk}*kl9<wgQX_N;olL7f(Jkc_0xr5gV`y*~|1lEmP
zC)S?<Ck6>JtWg*aGW_#jWZ1L5`U5nrVvDSiM!EMUn$Su!Y<H(mO3x@1#-hrjxA<ns
zAf7@O-e0Skx&w>2io`W9-P4*m@Mxl3lM)t9*WUcn=8+}Gb%gmPt=qlUQMb_Bnkp8k
z;Z?RNy&u+{4h{+`rWt3mV{n{CrvLG~d$%793Tk^bw2(@h-S?E|o1<aPXb^i)p=x*I
ziS)zaH{7P~*`Cl)nW_fNHdXz5Q2uU~<!j=5(1q2uS|moxQBSaXOarFhcy?(#qiIQS
z;?=%2@*gCYj4UYs!D%dJq}{`oD<_pYeNSljjWAx)!Prx=+UYx=$uUByKCaT1wjXkl
zxV_$L)BAnBvD5oZL{B2Dllq&N3K|{>OW9jf|J_@;BktaKWGYQVP&}I+^t+K!=(EFM
z?{@RHgS)7XwO<fv<_tR_k^|v~)(p|qd{57d1ShIqcK(ClhYtO{oCYHlJSTK}jULGW
z2XYp_<Bmojo}Ye6IB351R2<bAISC?*+^7alLIRYz@>EA0v$#;`MZ_bAP(?+Zn+DB8
z#(0wmSULOCAoHLDTGDdWShtox0Q}wK-5wZH%~^euFEg4TG0GK;254u8mFInQT{Ke4
zY>3WK-I(T$jtYWdaHp@0IpHoSNzkDLmbQzki?BUyeIXy1x}FUl_a&HCcvb{mZ3U~_
zub&#fYV&;3A)6bkodfDte@Z(!m<2Agf31}R2mD4IETQbtN}qO`%dUMm#?p2s^YcgC
z05ql|u<i9=%J-z9<jw)wtw)?me?!S)OPE1|N{&@J#KyUQi4E$L*&4}vfV+ZI<j1e|
zS#!8Pwe~6B^}y#Cg$p^VtG~t?;*m;0R4m$g{`6&ir85KFL^{cjikXbQlqm%CE>7A%
z<O8BaBGp<B-xR(>?W#2*5l%s8)5$c>k<5r%l4G<AK{+#eR#Pvz@p}JF!t=n|CA>QF
zR7s%FID#Rjz?;$7eHj3Opv=fssd<)60VV=US1%79nn;hn2#UA$o_xd^{+<Wq+_{-r
ziXzGYNyx>?1W$$;l=?6>QS5?&E41G|1?5Vg#KMt(IAYd43kd%5jCK2EGM1Tk$f^tB
zKhKPMFp$xgLi5j_*y_<o`o1}=m+;YpIr*fN$uU4F%7Cljo+a){wSd|sh-biwg{6l8
zw%nFj4v9x3<hx)9WbM3i(a>?}q)o1%R573;re<_decbbtjn4eo%16sf+Rx%bsFBLR
z9~zSS&p7{<D}Duj!<dNFMA|`^a?G+72F!W-G4tf6_o%fqF5segEb$N#s%D*^Z=656
z1BI!x(@wtt5N821@ITb4Z8xi-*nTtcDEt!dcmSdnt*Wu@1RaV5N8&z*EPq1gJ@bg<
z2P==eAs(8I=_B_z5^t`Yt`~l-vx(#-cOR>auKbsP4|vmbIhtT5>*bxUlg3uwtMukW
zlE5nP_+j+KSx%Tn&I6!ZCKPJL@>P9j>;xw3YA-mYoEtH~WMJ)qfN?6l9-j_&TXN~I
zE3e&WIX7{;TPgOIPUr4QKX;bV&uvZHPlrs3{+UBIHXwLxyZM$OxrsidLsw?$I?vJB
zXS{K2&9ISed!A`DPOaa^(QPlBXucz%-^btLH%Ow8xj78&tS5rAIHeSyMR+?}d-bCo
zJUjShKLho@{z5@;kO2}ef(6ri4*f%Gl#X|!po-<|_?}$rsHDObEwtbvs1=4^a$plk
z2at1|>u)suR})PyT$WCDxZ%EMSF8IBnm>ZJ!;G0~v@_ZIG}3iWP-$nB6I;wT#=+w@
z>>YV8it$*>5-xXr39~`Pe#Y$L(1nu4{m8gQ*A5b}+5W|2+Re9}iAEa3(df(HQGLX3
zf7zg9Qocn|ySnX@Hv5dD`0F#SlWIUqI6`T>J#fi^{7BkgxwMFV1gZ*WAYMG~S-W>n
z{q+aeQ2(J@LZ!Bt#&h4<xk`ryC8+q?)L)&pT%Gn&#Z?VP-Qa8w3ktH#vJ5>G?djWv
z!QJ!>S^M7^qj~rKcg`p(-_7Gx8?V^7Q%~{SeXrd1^KB)-*iOd%JA<P8<gD&_qc%6#
zVNrjr4H^{lM(Wk`jV7Dc1-_xRq*EOgl%qMVykk~BZ=hwGy){5@4n1z&mhZ+Fgn!+3
z=->guKYoKAKN>lB4f5X&+Kla_TK_+O>n$V1rE$Xnj}rbhJwl7F{QJ-UqJP9<e75;4
zgu`SB&4(&BIAp<5Z?aT$@@zO?-Hr{q-J!Q9HpUo<NKm{^m9$KrcW%%i?!DNH;;iXA
z9CwrkV=X98%X9j_@=SNP%bkJrfR@)e_J`3MEM!epmw#<tH-oB^2!!l-@lg;*lbyZs
zkLfeL(@&__nKZ=(ZTYQ1k3;Cr>v^5rNyvstER^_pI(6=|J>?Sqmwi|Nd;R`T-;P>G
zV+aVo(p<|iq#AjTl<vUic#5lWg>N{0kgr6{+sKb>Xmp53(d!~j0AizGT>b`zuc%fk
zt*b0plRHMq>g;uqjYS8^g+r#mx&@@P3X6yE-LZ{`L@P#5FgO1K!)Ubq1&SoPL^gKD
z=&C`x{}8uA3(4B8ER3-}jy<%Mb-tm|VfjFrn=0ap%C5&n9n8SiI$ilV3*yxV3l6AK
z!?B#kSh$oJN+j!{NM!3a1h<Hq)=!~9YFf*l@i8R;Hl<u~9^6pb=^c6il%vi@aS{!E
zjOCkKN^YhuZ1`o0?Dl3RzrJ!M+5;ikf6ibz!xtrpHU%b&Cl0*q@&aH?G-*KN&h;Ox
zzz4rIEV(hL-ZQF!Zh*wvx>TSI&n&;}45MVES;bWL@{Xo*`E@6do{Kkm<^^>978LL9
zZj-=W|3JQNGU%6LYa%Xo0n2%Mw_9IB(ZokqREPAQl23nZpBt<B=Am@?97E9_lWXRF
zeM8~`_UbUFK4uij=X6bcuHL`i#OrO$aUN>TSRfZNJH<&U{Nx<4%NEtv{3*qOatmhT
zLk7GiGg`eyY_|{>{deE&c3n4QalN@#gI>SMgTDUjeGhgICtnn4#@M@r$qhMjtk{Qy
zhGy}i)7LZb{Xl5w>><T`$S7?&ogxyP?{uiQ`&zNlLjHeIu1J&Z%8S2VbKj);n~Txn
z=2Cy)PF85~>oxo8bmXsS4d9WCTKW4QEt=}5@tPz28}U`&AXZWv8^9C&*X=Sqp&I`L
zuNUUZHQ5HW%S{Jv;dxLWmVJ(axF!VmOMH<|v6>!N<q3pkb(srC@RqQzr5}}4h^ht`
z?Y_SK6dSzV`wgLp2&g&88(28Wm>9Az<yDFeELQ&S;QW6JT+|1m4R)sNH|cUCRpCD(
zC<L6t`2xpBQoTgBhzEc$dIRR%r~q8zkl^6lWNiy^)&-_Nh9!rk9s-FCkp5KzPlgPk
zpm&p)CLokj6{V(1Dx!YFXRu}hyjV#)F7V^n@l*gZb$P!UfxqmpeBsfT5ziJpgs?h?
z>5d`agoz#aYHZ*%uIU!uapHYNS`rHVZ&{782pDZV6!J|P{l)x<6<5yndZH62n+wXV
z00|5PN~akMaXP8~-^7Wp`5^98B8rr>PFs0FoyhdqNt04ue#N2VcGLikbkQFV*KeS;
zfR9##p~5t~Yp&|x1XMzBXS1?m??laqwDDNIb+C@kKV<-wcv_devLEn&>tJ$(Ms76e
zw+1SGS-tzO44b}<=>UIb&V5Bc)U+(4hBe;x_E(4pf52wHA(>~FJi7M*+H(=9^nc7u
zrgL@S2nG>aC@iXh+#qCoks<t&pLuk*!GCGPm}3Nh)IS-!!}FF-`Y)prj8Lcujul<V
z$7!IWk*LN|PEb=<`nZKxF?I;AqKyePFiIygNq2KvZH}Q+!X1un%lbVW7^3xoRh~F5
z>Us;FNwYD8<mkQ(7o|32oAv~pUV{fzu5xU>;SCdg-b0bUgn1zBOAGivZp{(?Wib~S
zQN)MCc$N7NevDQk8ULv)A^{bH`{|!Au%I9?`fN<6X8QyDu6U>Z65HfYZaB(aC<tLK
z4@tqSEB;q}MD);QvTTn{l#XobeRTElR7xqM)WVYVubCIv7luYg3drO^($57;nS3iE
zWSE)wz^*VYU$Mv5F3IA&fQ6f%bKmFcHCrf7QE*sj?WWAg4e?1bE-7TT%Qz~TXsA1{
ze1WnW%qR_enpQ74@7{KF7>b)n0wN=6@i`R9BgLZI^>XJmLupczk~p;_adpsjJofVB
zlmhH{#n48?QPjxX5hCzooZb|bGejSE{msv_+}hRJi5nr8aK2ER8MERYn&FV(;L9M?
z<cv3x%iTVmm&9(v-@+Q6P`VZ~7UO?4J7xt11S7N3_I1CW_+{#~Og57?L6Y@VS<Yrm
zQn09#2DFah10;}7HT~s(wG#g;HF{z3zoI}Q?VUyMWp0j8xyW6$vx5e&hZKV&Omg;m
z$PO|c@e>0qhD>91BW`vZ2Lv(q8$E%U>SazADlpp(-lGA3RJ^JN7Fx+8NBoPtkM~tn
z(4^BIjmhE~?O-$6Gig)KQq$@9vMM9RvkdOH@dz7aoG#g(jT`8`sBcC6*|wyg%OCgt
zYwKj-?z9QHxa6K652FI8%TbQib;%oRWYZTairzcgnTpsy2G1*@l(0fuB0YPhH;NY#
z^eAaRrL=RX6}y}DRafIxSMC+$EsePtpqXAqQgCRD&x{Xas3c((mYh%~RD8gWNIoHf
zgFHe_P(=2F79zDG?1(|w2+XSANGL2);tZ6$oGZjzMp7Hg*3fo~pllku9R{3Gffp5c
z)RHGbXK$1wNx@O{2%nW$2l^BzAyV?vk@j{MX8`-PX?M#P9S;+?=Z=v!Yd>EK+Kzg&
za$nMVT82%Q4!o_b<%Gi=ypYctElo|4S2y-F;nlX=&KX_vR%%50->LfDY%uV<6#is=
zW<LhJeBnwqj!!hA6EBzw!2JxpkCXc|c`+NFhmRfYag2+lE<ffn6kG4d({eS~O9MWB
zFQe2$QL|uaayneU(vXy!5Sd$?>p#<1KJyS_49iR+BaqU{!ivt@c-W$n!FTaP{f0JE
zX{td`=b=ukAtD`?Fi>NyhzW;F!vrC4yS_O=Npy-#4pFjUSOX0p%0K#Umd%7p`Elua
zZG!iEnPI->1Y6>N_+(bOWhM;Y-hzy08M`TAvlNWXWFQ*b8OBzu6bTaR#Z%_W%%qjl
zr#pX92-N3mHmcOz;I3R>SUU>%Fd88fcoQCo3NJc}!_laVn`zjITLdbKlkuu0foW%L
zWZslz>$uf8U}^oA1J+q_z)}ZqZRlEG_seX4YO9*U767f{kb<9RISSvz*5ST#@fRh<
zZOeF0*hT+qaSp^Y5%e<=hL8jUy^fAqQQBhMK`Dj9Q}@J9&pVBk#|-s^M6+~!Z`MBq
z6SH?9RbMAB!k&(fj*F7VF+{{oKA*1=zZHJlqvK$6x{sUcD_-0?lI0QK5?EPbge*Ps
z^4H8O=MD`>rJQu)Gh)?uBHPV0B>P}!Z9gmH#+714%GtVLNsCv87dSU9Gb7K&KczsB
z@MuEhfHD+dwXCxTN?!uG!lguGOTTJZ>+9%8N(L=JXRi=`5;}Uznne<G!eo(jTi7cR
zmm1|X&4>ZP?G(1yo;fGRWM*NvWSE^<*nZXL@VMbb;Dj>Po$1fNH}%XHD>`C(;)}gI
zcRodq;<VlJ_zAdm9$<fGlTZ`(7>GNQ(;>l_F8!}hk3Iy+xDH=%mg#OvA1nCu9y;~+
zZT}zLDI7ZrAKMX6{a%fvul~>!kyR4T##~1HpuSMMd@Q}lj3+~%mXIKpefDdg^!Oy5
zImqUHoKmI_ImhinA8K5GKi@T#3ChU_hJD4{&DY=}TYw!~%c1o6^WLj6w(^iMG`d#N
zX)+;;8)OS_n;13DkfNpG*Wp_^J<a;d0k?v>jkfu08bH#@pl0JdK2Y!OQ}_hrq0z+-
z3#^$6u@GD-^ih5bq+M>HrTW_a(zUskr>+;j--I9{!vx)a7K3zvCi)+hTQnZ<;H8qd
zakUljUY{6j=(<i8PbL$Gk@N@E6-%0OX+(UdM@Fg}gXj1t+J&T@xR7Qijq+lImQY54
zaX^jkN8bUQanxfM)+n<JWCW4&k5$6M%dmxrlZU-WABM3t6Xs`D!?OoScj~FP4PAw4
zDYQttwqjZ1{+)N_4&B2C{o%hmIJm<~EvuuS|FZ0LM6_J4t*WIJ9Gu}~M%Q~gH|Mf$
zLQ0Ea%!vcH9v>W0O-tWvCLJyt{rEZ&*9O-{%hqx@nY5QIGfvV+&-1YNs2|Kmq(RF9
zgz{Z$;C@e(Cb3dxkpOn=^@oi#0#>%@y<9}H33Ky6Coxc*#)b!j)rKOPryhA<uNKuH
zEz<3;_WO%oEBYD;PC0e8-I!7CZMYk4)W^e|*GkrsD=8|=(z=84k$}_YhXpVU9JDty
zG&E%b$YPjf<}40~mE)-ce0+JXYlbIvHc`ri7w0`<na(A+|9Z?=pdttk^3KVcP4UTr
z1X?<t(jp1+AHbFV`+hV&i)dNY`t|7cpqOBpfJa6-TT0BSjD>XNROIdndu-DO*5f9b
zo70%K!oIOoFf6O3>%LXzkDSbIX)O%}k5Dp*Yya2$Z8JN3CAx8&GETthGwVsTi>@lA
z(6ak6w50~qk!c_rxgs|Id^xk^e4~j6Wk`l}5rWAh)Ob@?p9hlPulN-LWqb?ay3@+<
zNOR?n(PH1mZZm6MZdr_S_^y~XIhkz=ZYa`g)T?S4o{5?vARDi3YsMF!)zSgJvCSQp
zoM9da+Fp&}hCCFbx#pW8YaMt+T>o~1-3Jd!`XWlIl+=8&bP%7YlyJkk-BZkcWI)#s
zHhJIZ)x0h()828gwLL<qHpPM^Ipi0lMG}N(!yclVeZz?$=0WU83YzPQ7UB#}Y@x6q
zBej7Stvjy+in64H<j6ZrlXDITb{#2&*c_1|kXV&}dihn{ywa8qrF~KGfI(eA;~eqL
zLW+9%0f&P9lEj3Bj?!Fmn6k`Zswnk!`6rl05bdnf-*lqhCK(UQopM1fDzaDP`w8bp
zMPDOiZF*4v=6I_cNh***;rtw@3OGWK?(0Z*TIWCDn3|?^+jS-DO)6^(Cw>+aEQa75
zbUsK^303I)?i+q%?GGG5vB2^)Hds`%<9$mxwj*9VF+OYot1De?t8*&Rb_y<By=JR;
zXj=UoyuGX8$dYQ=c>z{%w;QzR-greZ4Z(o&gZkFe(CVdsNOfz|(0BdDQ??Sg=u#A%
zGIwYy;t??f8~8;_UWmS=C9e^Y5|ZeZ@V(nnYG{Zw7H2=UAtbVnZh_u6w(Py!rC*RJ
z1psf3sg0(jqYWxy<a!x1$l~3yc^bM0Av8uEqP{6dh?rM_eAUb+A)sOAZT+qhM>2+w
zJM&a#Z?8m$ywz4Q@Lw$e+3&^l5Z7ij_q~9fv0Ev|2S~-I+7~Mxq?I3HDu+CRgF)+y
zWVc$A<=Q7KUi?vHMu0|{HV=MuK6nYZAOk>%C^AUSA{+^T;6I`hnv0jYy^jt2<<`pK
zg+Zy~ja21Z6t)F)n0lw1wSPmkm?7p*!23j#3bsn`FDv8||G3<-@rKZ)&?x+eKe&pI
zV_B{T8c1HwF)evHwIQ<CSacMknP6wa#CQPHMNwUlKq}AOS@DR0v@5`2Vj;{bqeqAS
zi42Yq4q;lXe+ZI?d7=+?OV#lF3@NklyhR*Z@<f0P>1jmk-c}4s+mE#$dIu7bUq#3z
zT$UtO{I;~P19uHZjJO$N$5C`hS2Hd~yAmDBbzy#VI&)jy?tu7z$a(TbOE;U*lY4qk
z>AB<~&|f-ebA11MQ!K)*{QI9y=KlY;%JctH<)F96@TKq<?AC$l7UFqDdcxW>y)6Cb
za?GX1|K6ZSG(l@FegHxW_SkF?oGK`Y{cv@jal-}*)9Rrt4uChoUM4Cn8tiyW7x9#=
z?#OumYo5*j=L&Webw#f|=1qN`p<A0cy20;S2NbsL&Y9fy)5dLYH(GQsit4wqY10`m
z7e~!|6F%YjD=WnxN{9cxHEqmnzZlE)ag7@*jv1ry<?Yr(jd!mPN@y3H{59}mY1o2%
z^PY)4I_*p;IQI~hh{3*ngHqPJe9`-VsyZ3gPR^<au}fcHyA*xzq`Q=di^CrTcAcAH
z)X+!9s8+2SI&9b~ko6u_#i=U$jNk9nyInh$s_=*2*Q0-Lurr#@Ht7yc?T_z!p?Ewc
z`Ff}I(b8H?fa3J}3H>DAk#yiZPD;|x+hI)fkp6gJQuxZJ<JA9pi#H5rBrihsT*YNX
z<P99dqKSCB;djwCq|f(OMM;7R0$=r2d~cV^3L-4$qBV>Pr|2!RND*tlfB!x&Pi6Jn
z{}3NV#tHI)MxRAZ|9);3QZq=*jO(kGzcWNF?)=*ArDTQ<Rf|rGucI6x;LAC{dro@#
z+a#@}jx8D%GzT))th=|$<P7b*wd3<G-XE>BD)nqTU+6?bupHJ){g7xN06}v|EoXs<
z5x^Hjdd}yruF2X;9sChaKb@+Z&cw<ND_?-D+lS6@M`N}q?`l=~*STj?OiH9*HlbQ%
z4N0wzuh}P9aPB?Pw=GKn91Q$&j`!pS1PFiIo2S@eZSb&bzZvk%fT81SyxgEe0W0cv
z)fE&JWK;^|4G*M8t8*3{x3FEi=iQc`6BJMewH3L$I(<I0sa4h2y4tTbH8uB*#GNL=
zN2H?T=WcD+^we)z>{un0L>evyKi)sLV*2zK2>ir~Mj>Fk;!eG(5qX2X&*#lI+<kfO
zqiV~|TVzNb0=otF_B~qtp!xbBbeNHmk<TCWet1rwK0T`;Sm)r|``>AK&Kmoj*2rn&
z$4*^}!lh5ixa&-~>Wl8tE9%6AnKS^#tFJ1Rnb~_(TkTmu{w%!6G7{UzBG-s8Dk&-P
zZ87?&7=fep9?0YbVot9wdPsWOo9iCV>7{<uV>4J#%&*`R%UL?To_T?oslTnp$a14T
zXdxFCuLf<!u1+!^S;7<7!m9~6lu737cTrYNqW|1!qG6<?2S<#1I2TQEN<oCG=6t^`
zN8q}+`yEcTdvjyjTeCw&lXvQE?$LTt??<uwjjN3}x23E33Fqxkq?-LzTJ`#J6&)oD
z0z080m3QD#hti<~W3fklP~z_mj&yfenckY-YY*H#Qu|ByIdZfK`j0`du$)2HVzhSg
z!_Wq+*%sr<-cTo__Wqsg4_Pub^{L_A$?1HF$HVOQ8&g|Wz|KNzy=Q;5_$bbDVm5Nk
zfkzNGeaGojz+;A|l#9vbk@qJe13Hr;BdWr^FeE%YmW&nl*A+@cBt1pv`QYXQV%{Q%
zbk*F_qWb7m2WfBHKd$av{`bH^Ie@#H9Q-Avr4JX?pfIX}0`2|!st8+*lL~iyVMW>w
zYjhogf4sJ|6N&r8)J7%>V&n%Dvh^|Vt)&&i9uys@i^9om-viID<d<qc#n$%S^1COZ
zC?Cfq4{GlQV8O7ZVUd{_ZHd!`nxmtmz<aQe*j};g*RpF1hy(8F^U)X8in<!b5_S|P
z&N~_z-9Ze~$HiQ~=;h?S_3x_Z&)EIX4DXJX7G-fjD(O-O-CT(hd3tc$9>_44?N06f
znhCHs6KpDbh01dbwYZilXj`1bl$<SlLdL|9Q`=@Tvrv*7o-DCQ{Ab-iiCdoyaxV3L
z5mY#YP#47*7>ge&BBF_m56OhgK9BiXhcdMW8uKrM1+2p_49OJ*H1tPGO3GYdG`A9B
z+<L^2Wd($T<p{}*>WjY=R8#xIs-N3>VYtMrgs|$2j0|xwMY<_9(wH#9k^!*w^x12D
zp1NT5tyF!s;o+(EEsC$k*1yB^Pr#`i%3FWrGD&B?^8&9DK-f0RNsK^?w4_QYXTp$L
z=qQ;ntn){2nip4+t9<wZ+dAN>3v^G&nu{xRg^&p8uRi%LPcrf0!@f^tFfsNELN7|y
z1{W)073O)*C?2T4hFrWd@6K;+w~juXoSe*uiZ=lmK$2t<SWu@_dZ2rWVYA@~(9C~<
zzt1^%VBU*a5bu)D$b+IE;wzSDFVPi}slyIPE<q!W|MmH=0Kc3@szcs3Y@+f6xng1P
z#*G_+yF)&7z1Sz=;X}VUAKYlU7apeU%s*-578yMJ^YvFxt2xC+N;_YPWyaVpAL^cm
z*0$|X^=P^01$r>mw-X$oRZ^%&ydfANRuRD!QdSg7B}5T=W540Y(+Wmp@P&$p1$-`b
zHIuK&921WwH>!VzQyM=k?23|$@amj}s|shBv{aD72^TXN3$9k)Z;0r!1^3zo5CzRS
zYow%7=T-Nn&MIQN!_#Nnu-F`sIUR@0qIY>7RV0qFbjWL@%?3dBs9AQsraw?#hw_j1
zokETh4o8tL@?V5?x56i)8N=eC_4umDqj@&Xz3&kfI-(9om6{Hd!?8o7Vj(N@I;Qme
z%Cx6Dk`{8>@nc#~(+Tg<{!|^38~wQTbRTL(VyZOO&^9MMXOPTXlK#gNkr1Nn{8e6X
zJq{{f+zZx%FGW6-(A^uR<J0+jtz?J_osyLBvwN7$;G~C2$fn4y5uh8v)78_{Bim)^
z|Jc#pSdH@ny7RDq+u$_zSLaM!+6;<{iVA}vjO3vNe3k*MCB8*(2$L3CX?z*7CyY)G
zBh~2WXt6MeX-%(!=H)0hY-s^;uCf4vrfc~bHr#hC<+fGV_fmFC<4o_c2%>U)RG0nJ
zG-aqnO@w$$Q(sxC<46myv{6jnz&cg-eSQ9{%Q&^2-Ceb;-UsPzPU`M4woX=&)>VS0
zqJnrUh*xCL10Il+Hb||S>z@X`NZD0%->P+RGy8|HT9y|s6F5$KH56s|oV#}@=3F{-
z0akGT+xyYg(5)IN&nZ_PS|<F`kxbSC-u2iSzdf{9mJmgfo_cA#3W*9ax%Q6rjug$d
zR!I|Xy7)X#s<9JSw4nL?z_UVc7gyI^tIJ0VYm#$(+Yh*Y&Fp!nc<!v5^qXr}MV*^$
zJ?GmBU73%L)hf0(8b*V7kD#6@Pnz~#_flPYknwrYHX2J#N>$8SXQ12o`{r64(AGIq
zq${COu`&)_OF;-h?Q`eof#NiapK<g3_f_8~{^=7+8|89JYm}V<A1c<4X`-@XRMvbN
zUD7!}dOB~GW;%;;Lvp||_54+7Y5UfDMoWrrcdvrP({Am9{&xfrMtz;<<~L|u<Hn7h
z3jzqt7t($gTUGblLWSP3W*zze{JIP)!lQnN(*pGJ_r}EjCTPIJ@W&+2l3fYUl%_Y_
zxr#Sx`dF2Qc_;6rs{B0$V;V{8&y!l5vHA|Jf3y8OPF!od_1?Rf)A2Pwzj{qdd`nvH
zJLk+9su|J_%~G`ucZT1M3kYtNX#D+RmAbr0U;oXE)yrC!Voz=7(4_IVfzup^uJQD&
z6=unL<)1GQx92`yv+WX~mN?erk?$+cdCF@d02h{MchvNQq)Il(QhW*?<22>LQ%iN#
z%ITBa{1)s1P07i~mqS;Kj1wsI%zU^2NuN#4juzUirOHmD?*7$3QixvE+nsp~QLUw6
zqEp@UvL9`l$JPFMnt*e(<MFYbiV+Y<VPbsOdWS4z<KiE|6p%W$YM;xcU|*q3AS!!$
z#@lTd&+vpGqtoUfp6_%)wn+NQc3zM58^eS3sO<e!(@$b7!)sPhs^PG_Jm%e=S7^3v
zyW~zknKQH&#~={@g!MBl*|<H{!@cqnO*&tQj<!`^FmbY}scDwdQ;vtQw`ce*i328`
z+t*yZ=4(wI7G>6~)$5)ZH9-_{Zp?AB5CfTme)sldY!RHq9IJX^(k^eH%wd<YYESDe
zKazU>)M&9~v}7Sc!EF7DMXQ`Pk3P&(ZjJQ{-6TQ5^-z4~igBm%1+*PFHN_MYg*3P>
zJoH4}4X4CXcA$JApoCq`#VS-|i~E5HHo*OaNWf+COmec!aF^qOGYY^Uryfr$=rA$^
zeO`o1jYrZQ3M<P|>fg;J@-mU>qXfZUI%oi}v=?v|fZY2zYig-a8@QT&Sq`9bApRL#
zYYYv&*k3ZH8e(QGT@cFdQUws~WSiT&`g$2&n6sHUyv6DvchC_5E}0Cc!4;aorJ)7t
zaOa%m>E{olo@bBr8+t1G02=0?VA}%II5*fSMeE>V8J_gGWWuRKhvp%kr|M{4yIZPo
zE1ylglI1d?<am~Q=2BLTSI$UI8Q<&kmguV7Dva>kxo?hg6p8|Fc3Sd?SIz}y(186D
zA6RI?$VDi|e-G!4+ZTbn0qOorscbFBd3gMYBA$e87kI$J`;Kw7B~nj28Zf_32h5w`
z3?`z}Vp_zrNlXk&$h#$R6#Q_q#Q>8Yt>ap0x~9kKcw3y><^8?haYScAT~jzQlD<>w
z+u0Hi6B1T)ES*E2Letq<pa{o|B(^|l@g7TmbFv$F$~W7?0dF=Cx|ZGT=VIY-ao4<K
zRkfE+OrUK50y%4KR>Ue>Jz&AEPJMdyDv&DE=Q5s@=>SDn7Z>WEe(587UmZAnXi4;&
zRtET-+&!!_c(8CJhnA5ltMNK0?%H)7|E_)mQ@ia)&Y<N_1Yo4%U>)<U4O1%WM|uYx
z@Itv2%_EDklpf5W;X;L4=A{<;ofRjoLbC~TvNSgcZ^!exS$3T@6FIEB?@B4~htG(p
zj?NA4x$C<Cm{uLjUljF_t^gjH$sac;35*~13UF<;bJfY+0i4_E$2|e7wrl_B|9QO(
zTq<qaKArOrtxl!5Z7z8^VCZ=^%}NrOqiP0U)@5=Av-f(9EJl|*CFp`Mly@e)nffAP
zK<X6#3FWg>SGAVO#wKcipIJ_i>=+An<=9ZY%~Qe*b`NlEYK4nk?LOlfB^E8!{c8sf
z$zlAFkCUUSK!qyaNe{KkEy{DYjeHMY@LTs_{LK>+bX)Fh;&;vB4&t&S7f#`gMSX*x
zmV>6qm^Zeq-#R^7`s{nAW>Jnd8EQH(jkq_q&jr?e`o*4MCPSCV<Pfm}OZ%Gzz<H;O
zabx>Ha8u0=<5nJ=0_}>PP9de1i3~1F%j9ef=r#57s--mPY#COEM|>`do0hMlzSUo@
z-!POw+_-X=Y1ZpF@uRlKJzm|?<=MnxN>{zgaAB#Y20uYt3Oq0`T{Ar?X`2%*K%xKO
ze$`c3ZC5!w<*+(Crpr1B^Y8{AkZvHkqHV+KzPr>kgWS|OT3yBHsLTmmgkjQ6f<Fk}
zi+2U|-X5kT6}O}7n!lLg{O~$=rd5(8UbxU4{=-LmOj^fw@QYzXX*Ac!%g<LMmyjnG
zmgbIs1}Ben`yO2z%&Cu?u=eHk>;b}r`MPCJ&DQdK=?2xZ=}{fR8#fEcJsUdf3^2qr
zAa_9L9j$t__M!Fdo@a>O=E)%jN4&&{@W|*YvyQ78A57fUtJlGtpZN3P*Xl{OVWi59
zw7*$9W5I%aSiw%KdhOYH>hZtP7Gc_2@_|ii+l$PD!|CBD2=Ar#ks!+8H2Wu!n9DB8
zTWnADzp}DYv-M?{F=su<m#37aPOjM4B=OOsn;=((KtG!nSNeT@MKF<6?rttq84Zdr
zTS*SleEEuOxF9(C8Ck6@d$M&V1_iT{WBhb_MS|U5+N`o!`4E~&T3#o&_>;!o^hut8
zU<$LZWeg?KDdGkYOh7;#k30@<EUF-R%2uiXW%7A}NuporKWijtn}xK@>56bkRpF#r
zq!ErmBdORI(eNqGw8pLUj)0vxAY#}o(jAn3^MIs>J?{0r6f<GP0Ll)~rE)o<hy(b+
zSn<B%JzFU3DR)~?Wj15-7UkQB;{t|T4H$=AUjh00v?Ebib2;}NXmUU7LZ|N^r)j4?
zJl`Yxq)viS6IDgMY%Wp3cKn}-#!>dIs%`SyjJK)ZjGUb#)U%=;CVVg_L=^yOB3^3w
zHTAwnnKH0;*)`m|+SR;{8gt53Si)l;cc<byw)<^r;ek7T6NfV~Gc+tr_4`6zEp2hl
z1EJd5E$_2u`hK{g)wmRGB)dE7A9?g!<4N6ghApK;O^eT8PcHDT02k`6_A(^%es`=&
z-iig8!GN9P>hg!h`eYTh9TInJjP09a9DwmQD|N<6pkCKPLG&1@7sl-RjzRJLq1iAx
zsn1NB|3qfQ6vh=!1wDqD>F4jYv;EpIusWlfN>F57CD<{(%!5U5`*0M;g5R7*hn|qB
z+hkiggK?LjSaYhDE@s3-0#w~Tt^@+#Q8=u*Kg3GmL)ZD^5O95?J+t?OmY+y1{Q9+1
z4WjQ-oJk1O>CbYfM<=aaym-GPo+=?{J9k+x-dKAklvwnm$*!ks)1z24dhp0C|9Rv)
z<gaej9Phh_1uxywW$ycsgN><seU>BP=p0q8>2BiR^m=NT&YCATFK%hYx=L`4YW4-i
zQE-eg=b}dZhx6WZc1q*?-A$4#L5@-T8EsvXrg-b^xPVfiMbAR(;8p!yCLRb2ThHBh
z=rS#D=T}rG11`H-k5N9T5O?#Lo(CxWHhJC5yx!`cvU^IH3S{Ph`WKVs7HfAN_*16o
zMW@UfQ>3}7(9m!J*z6~m!*}IE4wjYMd(Qms7%3I7oX11QOS+GOO*c_>eeU?gu)EEU
zcx^uVjW~clQc~(b8@H$#GN4Hi<NeFnZ_41fME^GB<yTfV@B4XGgr5blU|?L+)Pdqe
zH@VG}Q{&WpHC>y2_U1V%B^p=NQyjkh-KegC+c?v|5Sd@YMzA}S2hLf?PKYb3B!@Zv
z*S>bI$wtF#YRh8-YX%P+R?ey6>*QzSUVp%^oZ^;L=%aje`(s1fuP^#Vq*wop4B0!I
z3S_&~qg%M2gZS5NiJ!av$23IO^7JY8Wdowwlyh_@JTRN^yt&|c>6s-$XIq-7J_pCa
zQ}}D8o^IIl)`<ZRua}W_UidbT04JGFS**j)S^2R_raw4amC6Qw-O|{0x!2>w#NN4k
zMavlU=j2}Ip)(>HcgTJ0rB`(R-o}yboX;83f_MQ^-ZuO+-GEWs;Sshk%_3vVTg?lI
zQpMxrVB5C>+CI;1=x0u_s1=sx(WCyaKS1@3U1|^o?s~e9?P&42=UI@Pm<&0GjOmI4
zw*B*5TcxufEyhI69XHLGh^G;M^gy$Y>!+{rya>L6{VYpK?mjg?ulrB27V7Ty)Zp%+
zEphTVU<JOzqLZ+Lw#1m3@Y-?Jo7BS0R~%A*lBpKHxz3W_?eIBf#(#U{&V&79?!H@*
z(<*L2X6Kuumz;e0^7LwJYip4euGEi#Dwp61Y=+j=Ej+YIq^F=-#j7WpPA*fX>+yX7
zsRq^)YsJ1F`6tL6fMveINNxC?J~OSeFBL9%wcb*pw!h02w;=5a!;b5CL=AM=`PKY4
zK1B}AWl&I)iu9G5Ic3FYU)DBqKEi~DsO_X7sY^7Q$aA;t;`evcBe-J47%oUcc)wOW
zhBce!aOymR0J6(`82w^t2~Tw!G34gA@{5NKNzr3A8H%K1(m4{=whXyJkE_>2kogBj
z9(Zu8f2OUQ;~>yRzgw|I)OLq`(O0_q)H3bjliQd{eFwM?dpW+lhnBn6KI4UT$tt5A
z%>T(@Cie-`Xi*i{l|gGk3ZA$iNz-e(_`>^-oWl^jyl?=}Rdm>s6+gsIEs(mz#OJ*u
zk6RQ48<{B1+BRtxr@C}ne*f?cRAiGkdI|9ZyMtHgM@9Iu>St1#I9Q3WA1&ygl~bT<
z+ZoIZUiJ5p!Nt2e4c4@Mne$iT`&DcOb-#-p){rsEITsxm>f-#>ZsLkz%M>&R+&c0^
zJ6>flxSa#Zk&G3>%;jCH1r(bYnf?Qnny8&wH}`*TW0}s`><qagums|rWhvx|9Jnz{
zHW+|=<62!}sY5M`sLJHkg$ozDdgKMgysYc1X^AAETZ3_9jHWh&;?uaSt+{sO@mPX@
zX~2Z2gNgi(a2|Jasnr*zKK0H!d_%hQb&`4$M+_5<H5Wj2->Ypi|H+)5&Q@W*nKD~3
zp{~D)XQU`zV#)_7S~oc+jWrso>a#k;|ASO+Z{Kt30=bmwz%!ltR;j&<KJz9d`R|kO
z-=SI*7X?_8OE{vDEuyEYBP94_4UA>^f^I|)BTVG+1~j`4M{}I8Im|9_g!l{y3k0Wt
zZ{}>Wse~hk*#Vj!bM0Bk35$Oxn6Azn`g!FOIV$^0%BH2Il{;rXQxPCszJP-A!r!eB
z$^O`qtQgn(?ULX54j<fCLnE?g*{g|(e?3<iuv$!cnoa+({k4_UhZb}E={5P$mDSr#
z2u<dFb-CW-A!lkx#eE@SQIX7vwjUQ%_IoScx74q&RInPSVU5LxlU>m{k3B1S_oe~z
zqNp<E_3DYLjK&eO1!<yaTjR5LXP;O>{#T{PzsH`r_^2BH3o}MO#fxiV=bYNO5zFk}
zsrKq!88+VZ*D<$uyP8?n#2pbbo;xTvy-RBiWtX_eQ^)7ONGqAbgCVJW%}5hSahk4j
zi91qYzUzJW-8bUDjc@;ookL(L6MhB~NE6wM>+ZE2My=#1#HpwVZxs)%%*=jRa#rR6
zeVfOZoE_Iq4Ng~$;)Qe>wr{{m82~c2?Fx!1<IX<sQtpmjfK?Nl*70GdUO<zU9O&F(
z*j-}XmI*nM`RHx2nk-Gwx7K+CJ>bhq!vmx5awrj&oSbsuXWJ!Dl~a<mzvC?9<%pA@
z->rQ+uhE+4_BNcycC_Q-Sl2WT4K$t5(wCil!rq3b!wYG|3qJY0qtzRn5?m(YFf^`D
zEFpO7KOy+=-D%S|i)a~G?NrU0-|6H;Q-ZsD+#{3I^^m%Y!*^ZGp2B19QR!2UXI1ts
zS!n-ibky~+BE{1TJUiRo!NH9}5<5~2ub%l2VNZS6tBzlaVfU6MAYGZXvC%IgZ*D~N
zjU_TT^c|eu9RSj<bq^H2K~fuBE+*WMO?(T@YZc@%#Amr@;KElIKQI8NQ|3qt6EQ!-
zkyE8VxbRrN@4kXQVJI*aGn(~h_-@$na;U<-Wm#-_s%gd&D=IfnxApMw5P-K-K!045
z_Ew}4jHv)Mq(9*D{+Osa3or@zxlFA^{WDlRlp5s8ufqa=zW*CGi#0-S>)ESHLz{M|
zyyH{P(bms1+7iES-#(cr=oG%P0*-HU(qNojdYqg+*xO>!zqEnxw5x5)5UgJou58@U
z2cG6h)Bc<U{Q6rfGG3e8X%te<He2phJ?ot_??;R-)m7DpWe^UQ0U(WJMLI$Q`e^tZ
zw3((o%(?EPRjjXy+1G;LnL*Z;(7+RJ+<A$k)PcOc4jwNdv)k^gbRlE!Rq3zvcAG}P
zwwuL<?rWSrsL%<(G2j37)u5~1sWOcxJtdhxo;#;S$TFSfC69?T>iuC0)$k48XflC-
zS_ul3K~t0(;&@WF!<ij3%suL%j7QN=Acp=dp(}!PYLot%xv|oS@smC7dr``jlkc2-
z+}3=b&U@Bsz*?U!pJA>TO1;i$?URou<VYqjv+GqOf1cQ|Y^R^eeI@JH!*dWEo;rX1
zZr|6kgib2PBK?x$#V!MxPwfBK3R)af+M21f&FrOmOAIHH#7^heL&rhq7gw{m{X5ih
zoii&qc$6zYw)Va!%g!l9p+d{_bUhWPH21c%l+SA3fEf>&N&HVZvH_d^`)9v&zWF1A
zxT`$mF_Fls=(`jKOauJF4hFYNEw_B|I<mK>`hkbkC`R6S{YLJjWbloolC#XE5rxAR
z87w9yk~`NuRcJH3d@(xxlbzR*?`zIHs4E`7+ebK)6;RX*MOX>XUS5<TfEs)m2Hcxd
zWN=A5Ac9W%y|?G23ktNpPx2%hFV2E>-&4(4&S+ROfrPZONa9V~hxBb(#a2yISkf|1
z!TTO=pEL@DNdW?!)Q?#?&@`OSC}^L-KFaVkNZ2YN8C|ARXHi@IwxQJ2#ATr@-Hwf2
zjf<K-G!~qpR8&z(2$3v20=h8j+cND29JQ_A8<oGI{Gzst+72fIyO4#22(X2bQqIu|
zN`Ib?JY#ln!}5VdMoyX8dEr=?-IIj5$U64t=j9r68%g~s%hyzE#DDif&;1X(h!+a+
z(e?UBnS=cp13*+1qTwhgko6%aJZ!JuO6iLIK6j`5fc^Ap@xONJNL&wG*^%*178jq(
zsDkf1-&T)we;Klk4Hmrc@a>(xwWrro8QgU3H7`AA83A+Lpn2b~ZEtPxZB)+T(T6SA
zVvfms+h$4(3;HBDT;yz*Q;3k1KAKmnxFmEZM)-a%WdoQ@>VQWdoXaAoD+xM>aDBwI
zt(oU-`|nusa}RG|Uo5;i-*}X!2PM>LJ?#z~`+eFcMtP{L+6=KDnQ(TbR8uytPO3QD
z-{#syo_Wwn@m1RxeirT&i3wF#v8GyC%@S2^mXG1k*g>O9#LMW3mp(YNU3fcH)BE`A
zcD+(6=WwRp%qLyKYJ!hMi62pV!?lEoOjqtKYJ_M4GcVA(V{$ESS?_%}M4}Uw(q+}1
z9B{<Y;>@5?JN;6XomrZHw4Cu;Jrt=)@4J1cBDq4L{~*9R_-0=$x^{Pa1(y?lw)`4-
z`O}|8N8OXP<L|eurP}}Z-UienHzw>7w#%!wz>Di|f29Bp-ZxEEy0s3?@B8Thc7p>3
zu$2{Y#=84{hrw=apZ^=DH$uII_dQ6XZ8agY)P+>(eei6$&YB%7$ExK#HqolrhuHU+
z+CIvwU+ll)5bZ7_0&o-=XlNJ<;}J`XA=8fbhJ=(e1ws_1vieBLJMTLxqdV$~91Z<O
z+~mX+uGU}<q#aqEK1QZ8=@&Mi#Vf^X^!#Gz3{}QxMqTF~BKbUWOq_FYfm%bvK6=Zj
znoq!kt^%^m0NDn@<UF>U?_ZgM<23>ep0adUs!Hx(V9+c3*HB!IO?WPxzmQ2(zt^)_
zL`lRRr$(udu#*5Uyjfz3L7&{~FW3bM9o*TFqDA7%)%cI$rJo0=6I};f(Jkr)ROzL;
zPa0gr^7aOF?z<H+$pcNbJ#padkP){~(2DY!0bvK0UF%(pF_sWT>1)F|i>W%qD)tIa
z>x$&PaHGxhLBs;G`&-eGV|&z>JxDlYMG{x~mW*?xx|`?aFWuziwo09h_4m@l$U|!$
zNOW_hV2h330uZ%Su6glvjQ*r3^@zz#Tp*9|J8ek{YZok7P)pqD%oLg@=RbUV)Yqfw
z9!H^h=Xf}_I{dU}>kr`~#MpBrM}uJd9&&8LwISm<NJOjTqR@MQ@O~kUG|Ltb?^5rQ
zzV@c#XOg8XlnGCsQONpp;q%wq<YueRevWCMB-vu9j{E*;zXiEj`8`KX6;nE7O2x)W
zdj}m(A4EA>ZT<6*)#lb2Z!SS(l3JTA2-b{D@IPJxsBGnJ7t)?6?zFL1bH7E0R^B8&
z2i`jV_03dFi#YYnYu_g&C0#@4)LcO!Ct$joYZ9UtHk#^izh3{L;j0ER$~x9_ebK@_
z!);E(+Q})E)A01H%|{Pxmj#F`VE2W|B?q>v&HC*Y<@WLAn?g*ErkHY|u{y5uig$U<
zcXf4bpKKg>Vz*%9i!r$N9^RJz^-{@LwekC@Hum|q+W1FXQ^w9+;$y~bUBh7m8T9wc
zNmD;&6n?1p!D=i%SdI%K<DDt<-TLujmP`@PG9x>Z5#0cKoxXkh-f4=Cq2;S8=H@z5
zycqY`A%>qs%(C;Zl&PHt&ASaSOj>&|A6Wz3T1Yon4x$OdLaIMI*-oqadIyCq7;@Ia
z+qiwr9#QP-uwX$M+A7)$v~ou$j`Mxp<O3Y-q<0knw#8tq-sTji%Gfd`a6A|S7st7O
z;dl`#45av;!fO}(DW$=M8&WGy-(qV=HDn>DCRK*_ghtVK^axik?$sxXT7z^p%LwH5
zt0>-dWEUd{oKrOMyYsZqO(z_I;6yUO9yynL)_2ZXNh9+HJ_F(jfzsGJsAu$r3y>(f
zl$sd=+Vji4J$u|pDMZfQEdQ|C)T&|F`PQd{B}tgJ?fF&a?csr`!L$)~w^0x0s5XnM
ztyKe<<7QkV{ThW~z4~mZ&#I%XIzN9z>!%D&km4@p)KZP348j|Lh9K7~mk}nC2SjOz
z5}lB;2Go6g;(;;(4U~g$v=G0A*pYFko`@keJB{Nkr_HI7^(cr}G(QxBuW)2OeZBuD
z2a;FCa^h+_z<}K=_GeP!Iybit?Z^k#G2MG@hFEF8hZcDA<8$UJ+Ga)j_{-qG3s0N$
zac)hHGK}s(la3BvFX`Rmuh#Jxy|`UY-*C!+@0yF2=l@{uO`x%E+jrr+r%8$w(jX+W
zQfL%WGL}qbrjQ1O%%PA-QYv$ZWR5Z<vj&k2$(YPTh|H2X_Ho^w=iTqO*Z$Vp-?zTK
z)_?8qz1OpzRl42x?|1#K>l}{rIFAz<>t4hu2+AY}(`yjZstX6b_do$M4-Srut*9Qu
zjo_wvB{4r;UBnsJzA#lR8F<m^V)|n9!UAk$I4b=39TGE{qkfH82xcr1@h9kWH$HB<
z08cA;x+?JK-*Q&S)etS@p%e?Xy#_)cGH=nr_QF9RDG~vLP_)Ul?bkS0)xr5}*w<?c
zd>!?`Z)9v8O3y!F3L_ByAdk(2b{I$p2s<#-olxNqci!2Kh`S0)!X>C)x3d8Ld<qPU
zGh-GY4$Xt=1WP}G$QS&rL||9h+1Y=yS_pFhz@j~PcGC(M5`JTQFJNJ+c9QQze5=a;
zKH3ZWtO5x?sxvpb12oa$m5=Hp+$?#g2m6bNKLZztjfX3yKiKuKt2k+#*Z>DV#KU<=
zLK^;who7?<9s@;$&<a3A8$oOw0>}o7wp#Q3bHJnsM}be1fqK|wybF`rOhi1XQDL`w
z{(2VMgli>jHYaA{4-e~%e>L?Wb?Z|4=Z(`Sr$lfL7wm#>DW(4@d1trT6ZBupIFA78
z*HGl@ONl>zTJ6T0+t2U){1sTjT77KB1)5W4F$!a6RapdrW%ONb&a1L~TdN22Aj2~t
zj@CoCvc#lo&*R`1WETLeX4+h~4q{&y8%Jumc-}tdeF&HbUJFan{_6uJh)fh6M68@S
zP(EGjG9_(u+`ym}i>{|VdP*9V9Wv*_rVPyl#KM=}<qf2%`HE(J?{t)~W7Rm9JYV}*
zT1<Xn*Wq=sJO3e9O6mcxk(u}o^rRWjNl-BS=Xtb}S0zuN?b{NBWlbaT)MLTnz(`3_
z+8c=Gh?H+o#40`6qtaKs;H~W5M-;L8polfG7Wg1ZVxn|OyD@zADGl+u&jffW#<>8C
z(+cDS?vISe-TN*fQ9z)6f9{v-Qu;gk5X6F@o`61e>8N)qak?P<yHp<`XtL64+A-Cg
z8b95PSz2}=s%jYRw)EO90n!UXw+Ss1xh1F6MLbIb=FdPnOqnn&mC>+LG>ej)YQr`r
zX$WFzP%o82DfxFnte}ON<OSM=CKPv0HFP{=8s1C^_*7r%T&xSN1b-iNa+247gs}c6
zT2ZA!EtkLm`BHw$ZKdlU((L*@V}TH-5y0ube@~%YLInRb4T4xUs-+?`Fw>744^Ui7
z(?H4rl&WFJgq3|5Th8~ttTE>Qq&*c_$Ym7LnK#f<q(VY>qsq97@tH2lH?L4|!e*nW
z@M9zEHJ~lRL-CX>>0l*+8&p?Z_mJiZ&YegFyv;~=jzH>vS4R6^zJ~wPb<X}bb->^-
zlGp!NEgJvjROn1J?Qna2M({mo@=>ld#4eN3AqrQ`&CO5e;m5DQ^_W5d>)RLPxi*O8
zk31M^&CiuDgPBMF<)39skujwqc1l=9&k=d=>H)n5VG1SM13B3nFbq&xN!WZ9ASGZr
zrMin0%Is@26##YO1`!-a8kyHQ+Lg;_j8s`Dt}XyJ`#`S^FIYk<<xD;+C@E)hQ36mu
zpo2WHw(=$|X{H=mh#vV#_&jitC#auER6ybIcb|@m$>3Z2Fc|jl-GV5RKtKWhYY3MT
z)PUG?d=yuKr-&*bn<QVNg1`~OmQR$ZP)djd)07L*%>m}%zsBS?6GduWg5pEyWH-YJ
z_^(<1a1FZJTo)etCguX+wL>*%;5&tKikqaAIH+ZZeYRe7r6sB_BkYKq{6a%@KOP}T
zby8B2f8kP!ipjb2=jp*)zjq5dF5_tO<Q}DP!Sz5h3uO=>(Yg0-X;m$!aJ)>5Q^1W8
zU{}#?iM%iFMM~8qF9%epI+QfSo}Z;qp8JA~nF@H%WYhcW%b#`<2`naCpg>rJl-fA3
z+ql5t^nl*fDQ*+(25}TXfqW<#?q0lAKG2xr5LLt+ZVA_fvX{R}#StJwWpD1&Nhy%e
zDpdU$>QUw#AwrR=(+Ff9N$UOU6w&u`JpgM(0!p>8-fDWvQ>~cekC7(Tl2u!|RAmSC
zE#N*O1=p&61;GPF%MBDMTT!IPSO?$mz0f7Y`~Bs_DPGj?(Y&jmMlfbnK}xr8q_{o@
z-p&JxsX(8$`Rq~wZ*ue35o8_$%~;Xq?(stuSFV6PmZnh40vxPy@;!x%tT_X|;7Eym
zMMcFBGWF@@SnmL=@mp}3OD0JTNmP_M&9$_{g&hKh+#y_b$RiLNX3e=Rp~Neq>$?p<
z=$g$PTpNFk9Qm8h6R#o&4_Lv@?$Sz%3TH{2vgB;Tmy;L<ZAG_&V+py@W4?JghtSqU
zQuRRl-4x2{5<;&+qtZD?NOWf9-C80qg$K;>2&g63E7}WVEH}}OuIS8p)XOeuiTpw8
zhkz<Um5FZp9J7@o#m{B^;|#XC59C2<X;yIDAqyaUH$R6&&!{--3&TO3;x4&TZ7##|
zw3_<djlt9dZof+G?>lDRu$Cef@aV~t7JwGZwvxxOW}}MXD5Wq4OQmuT0#TUlyz_`B
zSLQbCH;v9y&yh0X;3uY>rF4>3JU#=r#9W~E2w@U9Z;>S;qn1I~UnXJt=C6nJkV&B2
ztw-VjBayY1^ppq+Cb~goEFO<sWCcCz#>q&WJYbs|SH&r*)VBwPhLV|-S!*BTOI2i#
ztU65)8=!u0xrO9Q<rSL~33f?%+|2B2C{m}bfYFk|Qbx=5<s3@q+&^@b`rQ0e4W<dY
z4b&S{DN;qNJU<|9*~!Ou$2<uDENBA)f?PP|+js2(0nw~#CqNJ)&iv_9e9}W#({5mE
z3L|m319#VWJU_Qc^3g{SMhV!li@cK4q9~MjH&m9u-#LSCs4D`F%P4LN>qK7vX=$bW
zzg9VWH&0sjJ2nzkuZ8zZDBqOH?xiRYn~$Ahlvs|W%JP^=F%5;1KxFzTT-0%S<tJ>G
z3&X6@u107pa1RH9^3dNXu?tTRAVGXE%;D1lB9lic!dB47cQ8(p78jgD0UkCy&fVlh
zON|262rn?FP*T_-#z3_4{QQVrI_F5_q>%P?6Ai#qTsV}JryB_m0bk(Pr;U$+rwOo!
zoS@1hvXfnn3iY3x5GZM%nPQr?F9ey#z#7CrsdI1yV6dElLquPo9dr^1T|I{FmC=fo
z2-{XmC)l8_oF<uv?{P;giN74$w=E1N6}D(&(;L(#kQ3!af3YfDCs&S15q#6l$WEc<
zzS;?>=iZDDXz3EXj4t!%t}v6!W4t@Ae!GI}<#+@g5;Ur8&-epG6t%ZFqb5;NQPM}Z
zUP1=xH*wTL`Z>#N<DoA^8jm1$gtmd8i(O-EtQ6OuQXyD$uJ!hhfJ<f&6cXT-HVzY-
zH8r}bP$4JNffan(;Q9#q4HTnjf)%n~;1&3?X8c8NFJ$)?L^m|GcnfJ1$hbl3Z-V*Y
zkBg>0V6&}NMH(-Eeh$n#$T4s}7O{^+QH1xcze6@XJw@s_5xb-;+YFXN^Ta1BA<KH(
zRowYiy%=-H2&TvSZm|OtN}VTp_zY*zx=6V9k5uTG|KS2OfhZ#iXI=0ekS9^azB_0Q
zN;7Dff}>M#^pEl-j3v+~NoFI#y`W2`344yT455>c{rciH_@13rl_xW^0{(O?r#!t5
z&_d8$-~eLvUUSlt^xbX?DyE&I1e*6Z1A$v0V*wB6*32%0ye!U@ScJQz%~NIHMtaZB
zWRG>O)(MXJbW|wDpgsmB*0KqcUnb)u|L|x0L9v~%uRDbW&y(lLOtu||&n<efUTAZA
zn_&jfG}@`2=|X`kkqkS@xB$I5RDONi`Q0+MJ%Qp$CI2e6=bvzp;x(w-@_^fh{SnY)
z@V$TJc%oeTl?D;>StzUBI_UrsG~*gA!&~r*sA=M$SaM}QJ06M9MU2(LyP~i^rxyJa
z8N#ge$(T&l>yN6z1|}x1jf?wMXYsg*p?~SS(A_ZtNTHo3mB_@0L*e&|6V3sdulV!=
zct`im%D;S3)+=iT>(Nc_-|p>c{8Q~NQky^U4h$vjXDHiL!n7K10V;^+(<>GO@MV>8
z@}2*FLkkZdJ+eeoM2u<r-`jggniR`t2M9Bw@*<QPl=W672UCBe>kjnrTzXN}a?`O^
zKuAJ3l$Pu>;6`<3M1TcxB#Qf5JoO<h-q5^s=Zz!k2igILc4nMGt?Y>O|L_Ll-GFJu
z*_EU0lu39(gQUNwc^(^RGYZvLW>0JC*8!5@ib&TZPv3D>BL~g;ejqd^F4vY&<nyt^
zgXAVQW?DUY{CJwofge1hqOuzW3*2jW4|!!=uBW!!@$u1Yc0v7_`5^ZB>2gFt>(9L^
z?e(Wde*xpTi;qeRFCoNv?j$>~_vV^g0#&3(NinS^>w8dtI|&NB%6h3H&GX%)2LU6v
zjG9nSJOsszbF$ZfP601vYifl-AyuP{KbU4whK1BB*zI@vCD|J{sC#2U1hy9BnQs%M
zWfUoI$-NXXr;r989Fn%&zZqqfVH7>^F@fjYiIOBBG~WTe|M3&D28pcr*e!DYEt@>)
zw}u1+FgLvcJcme^SA788#Et0W47<+NEkTa9JN0POdK?8_Po&X}o6uYGhtQ}bD_~3m
zgMwI^TCps!gGCZ1sMr$z3V5%~d(h^7$O61`%!)#-oOfDY;4C>&w+2dKQ7H1OEm`ze
zc^oD0Kq<$M{Q`UhbmbQI15W$qDT}svbS7q9fZa<1hY~4HHo`@MwIRwy((n>hAZYmm
zH0%UEkPB1TWgJgEbz>qZPM8$Gb520P03R<);t0G{G|)zt15o`I+w1=%>i<=7hxUaF
z!*7;G@%>uiqwIK8KGJ=?k?U$VDF$Ia2LAPmJ}S{ct_4fn+%BeY@IGa0iE>!E(vx8w
ze<bZn#nq~gQV-egGTc*Jzh0i>=0l~U6Mu}))Z1P-@AR0i?aP_a3+Kb92e0%r{Fojr
z7sZB1lP5Np`r`23vxv<7ul=~JreVp`yi0&2Vf@1&uK{II-U6V)H8nLw-D^(ZuhFDR
z$-)wv%Y4b^n-+)-5OYz#K;4P$UJpY;1cGEQoP`8cAy`2cM>aoE&|ka~1W_`Z7vTu}
zKs$O-LLYzp&yW9JAhIUk-oc@?IKvc4T<}z%r`se__&FF2rCVNH03pFq_j80~+mg|x
zDP%OIjk2zb%K~c92M+Nad8)8P&u_U>DW28+32D>?27{-T#GRe`er$X{MKk)_wtL#*
zPq6&&4zfu)&!NE58;CONclZb!4e^v0=wgK(gt`>zxkEt&;VBy%8(3)s7d&y|1nhc*
z%e({Uo)%=@a3-X717<x!I<y-U7Z=AtQxP0>Qffkd2Z8L?4dtlcVk4>R=scf(m6CE5
z03?3NR*T|$QLS`;cSK6*1C`^dgI-T-d5SP2AO^KLdyEG~I{l@hLWqwq9<siuI}wZo
zE)Rwh{DPJS;lEzqotMpyo7-j6Ef*gXFC=%c&mw+_x6M$q_L!t8WF9A4f1}!vp4(p&
z_aNVXr8<aYa<h@3h8#Q;Hf8}iBH7SM#73K5o&%ZJF_iqXc{k?NDGqUSaBv(t)b!2(
zB^B=0yDsaUKvz9|?3jDU!b8+u!X;h8Y3`SP`c&DxV0fSrtAx5-c2ZJ#QYmzB2Rb^o
zX?GyPpt~{!Mo-@hm+@ZJ{?_={CoI%1D(m@-DQn?^Bg-d_nK}EVnbrw>usFitj>Ghr
z`R(4Y1G}sIFmyJ@_VNe0j~_l5qkx1Uh2DoQqiCd9dt-sWMd`v(O=tJbr2|%crZkKU
z3>?!N5IdhgD_Gr|!f?svhfb=I#{>t&6^j}=Dj(SF>96S(2oMk#*R%fvsnXV5du8(*
z*@xpfXy9Q*Q?}1MrPt8XI&iU)BWi(dGlZHD7HvDa6q;Q>zkfG2N_Sf&d8RMNWx*K)
zwC?93%_}89=xCimB$75Ws8f-|<1ta_!MEUk+Sy&_u9<Re&wl>ssv-;<DZaPKc;-j{
z@bFv9@3NjAap|eImitd^*f@V+xpc7AiRm-F<hiHn)638q?l>MLx?haqzY;L_*UN1{
zkm!;kf`g|}%k2wXgT6#QzE5uYr27*aZo?ItP^Y;m%j?&#tEu^V^+O4l#H!eZ>6cIC
z__Ig8Bm@~OoIpV7$=?{!Gl38y^-2>mD-dh!a=0;r1Ytp%j9y^MX}_aR&-h9u4FlLC
z1AUtq-_P-wKxox5pWfSE|1i^r4XA`&Ms(EZi#{FY(0zm86+qWjBjFUbCwL;mm{_6-
zCG5|%WKs!g+@O-|;`LaQ#dUjkcfi=&O6uq4vMO#V&~k+Z7Pmy%zwk@PCSfCYlh-hx
z%HiS-KcAjmO6re@5<WLZCnXJ_<UNH__aKum8@9@8sC|HH)tEew@*V*B#*od0KL%hP
zUL;f+?Yn|a#UwM$n>4n)R>j$glP^KRQie=6!VcV<EFmEQ{)H{7Q|OM8laaw8{@m1b
zXvdBuRCz%fe@iq(Ff{0Mh>3~Gs#U8HQCk_by9yf8sQd6Qtk$1iXUg}stmEe&x@zdx
z7B(Ptfi18-PMAVi&x%f32z~Irkz7q5h@=+Sp{2u#6Xl3LSK~!X>+9!Nq@2euAb_Zq
zYu(Qr5S7vAu}f$z4b4VaA3z-uPoC)O>0O6e0Yx3Md~ZQSK3}&$81CTeXsl>}%5%Q%
zI;{DSfB*9<P0^L%eC~@!sAD~BMl?@Rcam1>(OLIB5bOpniJ6sEWo5^$x*tA#7#+>k
z(s#xHrA?GK>(<@;OUP&235S?PlhLtWpI&p=;vw`Ac`Hzh`K^d0yR774`g#UIY<lbp
zEekMqfFe73sH&<W_Vn@bnfUc95d=@9;25B%36fQQ9q_i`kPt<89Yp9~f#r?U>ZV^L
zgig?tan*;pxC+b462Trt(!dUmBLv}Km9U$B44LO$IIW?9&fa8%w#vFjUz<I2pwM>5
z+q))F+y1APXQ~<5o(zlI^Y-y{UN_$mrg}9*^3HSD22nh(7=Uf7dI5ZliIFkN;Rf=0
z@I<~Cj4tV>kq#bHTs&kSxN}%vwzhHH_u>6Sw!s6o7R0kCHMzRF5)`@c&bkdJK16FI
zf{!B#oD?J+BEk-{i@w(ZRNWs22A+SAvXvju7H$j>@x42DXlQBCI{X}O0-GLA2%A^)
z)TwhCW*TfG*rvNM&J50L-MV#4mMnSm<_+oA0E~>|gB^cqNcUAqT&zYSY550l7GP;z
zl6DQe2i^nF3^n<3<T9V>O0>b?fR~BU{D$rc!9=B#Cr>IUq^7Oc$K0VgNCOe3^SX7a
z32NMc(9rpacKo+UqseIkHTFf4<e@`^D@9^#gZ!N2WD6}VE$lVu;Hc{8m>SU$Fa=YI
zuUxr806fi?1A~KG*RQVzMFokU3#tT@!)=)&N393HrAq4^z^X$vzQq^oh7?6TuLm{b
z0p^Bb*qo0jOM@}6GQZ%?oja)AD;~B`+z#1YctBM3MRXRBA>gx;|NKJkkv_EyTGT-X
zUOn2ffpzCMIql)!XQ8|-gy|PQ)sVU!f6ePA62cVp3c2)P9~m7CVn)6H3*?115epeO
zBje*XP97ImuW>mM9oC^dy=6A5ltK1snc?AXmq<a$qwizz$XnV{-hWUUJ@&IX`&E}q
zExIlS^ZViPF4+S?WiPQE4-cEi52vQCWxqO$O#t6(m0Oc207w-F$@lj)M@QU}=Zfc!
z$l>-}M&49A;sj<-REAEzC(rEcY^arKvd-($5urjPDXTv`xByMxF{HJ3y}j?VabKp-
z(s_Z$Xpcp1X>MM#tOkb&|CQFA{h+MWj|}i4GVq%ZVbblpOPST_U;jp1Us9(1=yuue
zJIUYUPbUUey;vjOV<0FX@H{E$goX5KHtV+zzs{?he*s_YJIoASE#sCgf~^9gqUyiw
z1Wv_sXp>eym+5Yn4|3l@_=Y~AN5_cUty=)oo~tF?W7E|IW<JkJ{~ag3b1%2tVNlFF
z9}Fvat4NQ1ErTA!N1gy2Q@zQP3uHn^zT@fZva3I`StncmsMu<$?$BAdDikUUt3#ON
zGI&ixn2s+#G1DMZ<=U@%`8%Ird}CS8<b@Ty^e2E}N3}fRS$#GKMXF1gO7ijy$?Zw0
zOFyvh)Z8@hKb!iXk@@y+(;U`7wRx7DvF)dFPX$S@lHT}Q=f(cadtb4+)?7iTyum7R
zk0WHSV4Jz3>d2h5Zo4?$I@oxOGL!yrnzUQJJoBu}IeMsR&yIr!8%BC?2FuG&Fm5NB
z_lW53z1b=#x%YmNH;dD?Ya8kKxwyK(aq!*SJJBKgfX!^AW_Lx%o}4Lz@7M}ax(X9V
ziuSv3b948ZUDRMtPf4jvDqiMr?Hbqz(l6CH>zCYs1m7CXS@cbo$i%B908hN&=#KXH
zY8!{GQfcfQ4n81r>xt-;ylH1-j{Jzh9Wj}0D877)9-f8B7HnVt*F2*l+Sj$4b_Rz{
z+`V_Nw&^NByRV+6Vp!PXbh5Dhi!p_%=4IttjXhY{!>caYw5%*%zH+6c<M^kf&$h`o
z0|q|vZ@XVHB6roJLphs$Ha<YYUA3k=#b?p+HrC~^o@6hG->DQOny|X`t-HJXt(9vX
zot(f3Y=X=_$kNDeh&~8*O5z|ZwATjS-CQc0=!iyVAZxutLx%z9Kqh#0a+rmQiBQzB
z{TrWS<_#1eE)`E2uk7pVL(*NHH!(V@ATJNzm=x7SDWGuyjuK+Z5a!q`DJiLL`l@b;
z1uP)^HdI6X0|H{y5<vMwDdG^K$B~h%pJ@@>lBawBqw$}(3Gj0tY!dCk<O?^BYYq-{
z05O3dz@rEW3BAD~MkteRFA8Bcw8Q-Z5Fs07^$b3}FJNOn*O3b78jbTD3k7>bLBY%9
zbwEG>dLbo+g{#Ju5cZ3s9R)aS@S75}^04gGId|;|O!n~#zj%#A(t4==(zFUA3eBL0
zVQhZiKdf3vQnIb^=n;mloEsU>=4h55$pSKsXyi97=)EW++M#m}Xa;5(M#CYrwzd+g
z9r#yAkGf5T`T9M4_%zfSqO)!A6G>~duFpH82SMZOr471!`7_Lj;Y(aRMnd)BG_)7h
z^vYm;UnfUJMcue@<8$L2NJ-TO&Gx`vnw#}>S6tW*vzMTx`XORWY(jN_YWMao+P$ED
ztE{UoAA}D&`%4Ph?9Jhz*|u$)ul=3g!^KYla>t(P9vmd>QZG%?qWijKhlRy3atg2d
z)ROLvou6L3eN0A4ggkF<X#wADWp*7{+>GnjLz!fk!MP7h!vg~9z~@R&-;=!?4-eeN
zXD{SL&Terw#cZ3et@}06$eF!b2*jV4`MWMbg|(BJ$!c#-S9vIK@cb$|ZYO=xGCO4a
zdJlYdbZ%r{kQHg}W=X#E4CFB{LH|6T569pxk#;#Uqx+~Xtq~kD!_qio?)=Nk_I$-m
zTZ3PKgZ%DUqt>f@&Hbn3{<k_{C08C^eNqh)C)iJyvwGm~dd=5w*bw`N{Z)^e4>;+1
z79ZZdlPWHEj><?^u#|Fo7ORjNLXD$(tp>ds@AAmx)$n$h+O!<JDmmT7xZ3m@pw`!W
z;kD!}5?+0zU*nRlXELQeRD4*GtsuV7N=u0vTU2)EuvTm>xOjE17{}+&pUsdEB3SB_
zf!nJ&xdl{0J9TF_VD%ZT#vXey5h2-(2F{`AzMJo*2<!D*xBfwkblD3Hs6-P9T}-gL
zO5z9Pt`-exMejs~{`%{$edf*agAHI@vE2T?E@_s){76{%XxxH=>XD3Lnu$=hH-xD!
zvJ1RMAb~V^<NZ>8@bR|`p^|;Gh#4XX3Kl0z5JE%9=v}iTntu<w%wM;3fMZ7HWQU4m
zfTAs~9ASri#R{PA%WBx4Z>K3h6o7+;s0MA>uOZ^?t;r`Kpm_4+*PSKh<zag-2VVP(
zl0NcaA$ZKJtOJ-hjz+Fd?<zo)=YM&!c;y4o>ny;X^=^HPnKKBt7Xp&rx7LU9N#ZcU
zFc*4wd3oJVegK`Rizq}xaBoNE(evkHhpu}J4Gp1j020r>dm~NnSTt9B`SKR}I=}aZ
z7Gx;hebLBxlc#8dze|CD<k6tZ%F8b7=NIBVbVzqr$Ers!3RQ;08&5Es_EzwKYsZ5!
zI+xP6Z*r;98*d7E?&PD!^*MS$A8cp0L|3WKV=i<yfL5#HufI&FCY)l~Ww1x@^tEeQ
zH&+#5ph&_Wp);?}-&c$gk<|K8Ik8bdBkAG|vwh<2*tvg@<V88Ws6Wo|sz<%7yc{z1
zPkMf1-PTA7_ke?1-OQ$?rj`c&q_IW8LGW%_^4B9whcD35GlYINWQ?Ec`aXFgVbLmX
z-LIW;_U4p7_Co2S?Q&$42~HEvCkXA%po^lY%Gh?Av^&D^gk~i}cniFcJt892h#LXU
zqDR^VOidnxLBzDxf2@)7z`9rK#(_PB2xJSuYm`;;^YTp5N{%OS10UEa7YncUp>Sje
z&vg7aGLYh;BB<7-$H(^~xBv(Nx)#gAmW$tVo#KeuL!<qxD<2R*Lp@Pza)ryz*E$ja
zY!TF$9>=|$JzHtMp(HV;i~i~7h75(&(@va~(cj$ub<ElakgfEd{9rVt5MYI%<aM&G
zq(=AIDPvk0JfZIY`zJ)ldZ6;1T8l3K{qXeB(NSdSR>xafT0GZmdgh?7qH-04KkEXU
z*vd+k&)#rXpakn^htjRhTGCi1hKKtL>C`5znh<3(BnXXXf{Bg)u<bO)wj+`ZsqY+)
zgi6(MC-#TU7Pa&o^Vg-19_9MsD5?+hJ#j|f7=}Ej{=(x<cu7VU5NKp31$N%Y(C)(G
zVsA~<TjKZ+nL^!X>do;ImqygW`{t*?$Glrl8sdwhl3~XBTCXKqw!d5AWFHR6ww=zW
zyg6u)g(GHw75zJ-qB~K3;p^+&8&sU#-8jI?|I%r8TA&W`X%S1vPOTIv`@${E%qF8R
zUK{xu%X_vk82{oa;u-Xk@-te!m>`uQXnqyS6r`@A^7Mw8_C5`J!R$@8PrN%G@H6e5
z;}0@sFQPYGmcyFyl$rqcTIti*4<i4Qo94CW4Ut8LlAe6uSt1Vr`fbfSK0aUH*u9Gy
zxzY`3%fy}{G}0Oy8KJ;xq$FtZVX3*11Pc5szNYu6*G$kL&4w=c`8@(=7JV>YvYK|Z
zEu-PLyIMTXCt#{HDhAIoSqAU<#b(*|2M$RePv<%mV%5!ac)JODaT|*TFL~u=7ep*b
zLgESdR&?f3U`<0=@<BtA0Uxm|xm~(4x7piILYg#mYR3+Rj1WdCpRw&s%Z|(qN7f;i
z17z3r;|JoRUL;uP|6xeB#$E@9sy8qU=^8nS^L;>K>;`oPMZ1%D9cXp!G}D$}X^qxu
z3#UQ$@obJ!V|rbVo)hPRTm{R6ISTQxS2`-UhEnO)tVus|BIDyoG1J}6^ZsS24@5k7
z|D{jAMU@(}+GGoP_D}4WkBXXGy=s+k%Bjf#=C=N5xv`S<IX1i(VH5Uj6O$pdbI_mT
z9sYnYjr@y@X&~CopsK^Je!gLv{epf9&|9o>1kvV0XrY;)R@Btgq~tlIr*nHY>g#n=
zKpKfXG8lE^LCY-e@?-SSWZR`@z<rYU^b7vV&^y9Vn8}LRPWQ|pCNzwn=he%xN$67#
zRhS=O-LQdY@-0`xAg<>6+*E4S4LoZ_J-%t1N1dHUgO^9Gzbrq3u^#q*!VY8SMrY`Q
z`(I$10z#rmC`4TO1)<8OeQd(ra(woh&iGXzEA>DkU0o?>&a}nH$HN0C9!{5=eADyn
z*)xRDEx*j6T`jwX`_;vvxUa_J{pA`1Rnb}vWTi8ru+km>eWjtRUiBtpKvAmY1?$yP
zEg0*6Hmk|vW{n!zWpiW{zwj?}tq>mJTWk8dgZbV`$M{2UBa2?C9}M^OEQU;3VpDKW
zcefdWRQ!k2MN3cb<Ly0*b7=ea0bR<xYg}9$v7j)CM8z0r4l1q)ap-I=6b142reQM=
zt~>lyb>h%q{#{2~9XlD{rFSH^GD>QE5Emz^X9>C(6DM-1nu~G&Tk9V_S(Wcxqmi?B
z7Nn(#wD2o-yL)(et-fzpXf{^6t0eQ1HGmD^JXsSNXGpWb7D?y4tmY7Y9;LVunr@wY
zsA=`tx*#`1&xgcl2&{C?jVFPDv40R_g(JVoxzG8>MMg%Z^XO%vKw)o*5GKJU;`d^i
zoC|16eJZ;f8Q(W3-7Tv!)7lZ5Bt02o^dUj(g5i|0qS^mtuZF6PvE<zL-Ttr5y7WAQ
z<W=bhRmbv-Jlx%Zc0;}hJrqNZE&c@s1xVjv-S%*D9<}<pzX=nVP9U&?*t6jb4^Xld
zfjYku9y{^h<+l|Jx<-3;tr?0Oi@E^fy1yOqDKDO|sTw0b=2Z?)SLoiy<uDt3GqnB6
zr;%ZaeYFUIk%VAi6lwkgCR8tefN%Hi6cl;UchCsFARY?WLMA|@Bklm#19t@FozYF(
z&C4s-#MifXIm~N7-iX{VkK%nCqsy9=hlh_RrKG6nUG!TS`?%nw(%r;#-oKRYzEalm
zkRzcT)#n_Ok-2n6cXxz>rR6Blr#;D7bDPWR^tUC2_ERn6ZE~b@SXX2(ZnLxp;S>%u
zVo9_d>?^dfNPcyGLQs9^XPPJcETDlG@Q;|dffNc>;;>Ec!wK^m#9iQ++(7mXRftq|
z3cv*1=hyk>7WCPGhL#7hGTaaSs#T-Ef7i{oIwR$092Ov3t~GHB8{5nLlByW5h&wkR
zc!$j-$X*pS3fdWd<>YLf&WrE^me2UNC>jJAwpXtzS<ML~WWYW-Oi75rdgQ<_BnTqA
z55`Vu%f5gA9${UYwTjELa+>bg{Q+TV8fhxl*?~$G{OV{x_gY<aA?L}rWQEi$iFFs%
z2BaVJCJ8N<{ntt%2Gy9I0k;C+@pz)|@*@YXPo+%UZA75gcXFMv(Cq!K8|Sl{N8<E+
zZulQ`U)O8(&GWCfmB0CRA0@emxy!A@swJZGhb52I!&<*iF3KFdI4dRV@?(GZOQBiZ
zI+CiEMG)B;cZz{pNaRd6B-H(c*r8&pt1%<|Vq-2T4_hLbgh2Gi!op+6jxAfU;%Dot
z^D|1s6vD9A=3!ZHcMZn~$;~l^R9CB&4DVizi{B?CbTX3*Ab`KN2)Bi59^RzKerk;N
z5h}`!!&YlbXppf*?<+=olX=`oc>u_A;o&beza^;iMg{FUc#m;Az^3yTFUDtylbo*r
zw$I5?Y)E8ziNeAD+Jx`H`Az?<9jgzIjL*NEE;Kqw65F4bmxt^l__c$P(X$C#z%GS?
ze1-$>MvA*{OLkigM3ry5Cl&Oequ1(_<v*@t>3Ooq9n5O#>aUZ}T<ch#AJfSG=icF4
z-AF%r<SwwF0wpXg+~z!MhT7L&ZY06`99eh7-Z+t+nK_A`<u&8HCnS$yVZb=3$^#`3
z<8Jp^K2V$>9p|fylr1v9;gfzlzuza`?IU~fVlTR8b_yvgD*EcVz$|qZpg_<N&9D;#
zFqQIGGOTh-aaov05Vg5w0{J&l=fI*2ForY`eUkO-3%jw)t*`I)4P!GzKvm>{*00^E
zDs#z3nw`HgA<zZX)2AU<Ka{DA8m6a)ITvj1Zl=*slvo$m5}n1_;u4Kh7~-t(8`~Q}
zf0Dx>pa#1Etx++>jfZy6t<y7`QWq*djBZyNn}e+0LY{^%>$<PD%&)j)^KK31aI6q3
zHYH<3>O$%PuXYw+y=<YHjbzKtE<;ZO$PKt)`&qq)JXZN+FSkM|tg${vSvqa!;NInM
zUm|nY!Z6AxSSxOCgwB;KNfY+p;q4x0$#D~6&6IN!%8X6V=fT2UgQ}w&-Rq#a-Via{
z=Ti|oxw&Q1mOi&<_WDea_T=cSEC}wrU(rK*9px6_Wuz<shYm)sZr;&LmlCe|WPkc3
z?|mx|$OOcmtq`YXGox1yak?EeOY+C!<M{^PmAvLmmwSjQC`>E}N!5mr9z<A;j9@fm
zVqUX0>WQXiKdK=?x$HGP#icKm?iP`#?^!M5#HxFhui*m9d6v6z+}pfYF_eEE9Bwr8
zx;)P6*kX|vng1>uo&@W}m>lI!fHyjp#Tdqj#4tX*J*2{sqd|48LJXWwSrKIPqZ@}k
zI|f~XF%!J60YrHJkgdIOwGuE__fL!ph^%7wT>95vh!qW*4H?bF?6)Z#ZY|_Ve{dco
z;C)wzyWoc}&-`A8?TAGh1kT&@8z?9~`7)#Z^K+g3TWumr&GM1HJl-oD{}Dm8VCzrV
z@~WyTB!lP%!t+CDiGupiswmH?v4?Nxx_31S-X1syhtVnL33BmUcnj!`K!DH&I1>Uy
zh(V$m>@Bt)DDJm!-==3j+<aUAdQ`3HZ{(tcW(c@phgM2EEDxXs*k)2##<RoU!Gr}}
zeM$ZcoLr1=IqI$hB!an3&jsu~oC^n_PAGy11pGu=h1j+pMVsWzkOvPQ4DB93E8)Iw
zjRvwvAE4lvYv%#Es!|!y)Kv3*5hjIB-hN?R&r_5(aqOu=>N;wzmprl~OW?IvtyvRx
z$o}O7QNf2_sq581GnP+=;1)2z9#=8=3)`KYit0GJi1PGMQJ(FK#ZPh=I_7Wld#+&U
z|2J&=a!bSupd#C|k^Eby_oLFgDPGC`>nrZ#fg4{cGAv_fk3jv;OB1IfBEud?(2dbf
znNcxxfz8j)huC7oAW$vaQ8Bo(+Wgw14DNv0yfkYnp7PQZfX_ykcSOCgsOW`Wc3OPA
zy!r9~#X1Z|9ArZYEjrWf1$~xzgatY?eW{?y@Z*h0-$fW61VrfQ=!7IEC)+(C^|lqQ
z`gp{Oii#*XXlMksK10-N`h62}&YOxSw#|Qp>}RT1lQYl`$EMpJ9-CSe_dCBpW1fmt
zH%b7Sd1*o=F6(R9d$wnws%q(xnwr|V7lg~A%Ldt2GI{55oLfzyFpyBR?hX3eJO3zF
zFq+QD@t^P9<KqZ)gj+LmY&#Acid=x*8+#geK)x~Mti9~)Q!AXymSdtjV`F1WUsyaa
zHaX(g={j*t$#3kC%##%5v+kz#m#F34oN}>>eLo6sdVK`Sn11N@*x1Yu$pvjSHO}nw
zCr|pcUPw3mup-;^$zJmO>|QS-v&`ZInVr&~o+s67(|RCCHEk)|TFzXsg}W$!r=fSB
zJ~$GYHEPvDLPMY<h{GOYcO!^^uwoqaTH=6FG!~~|vcWWneCsUO7#Mb=^Wt+qo@XWs
zCBnj^C_>@%fQ{LJvIBetaN9F_daUg176*JmKEKEweHiBk$I|4&1>NgT-%+%=O}|wK
zkuoBb>*&}8v;2+Iq&cz-Rn;oQ#-vpq5k6u=Qw``I`TdFg{IexDGs9v(q8y}Z0~iYd
z!+<XZ4<1tdq;vQTMybusU9WLT&m2>~TpRyrVu?yoQ4t|(ojO$uT@D;o_?JJ`)zvt1
zXnuh-E>Oxy<jDYep^BZ^n5rzgzYGTR+2qJW>_)x3YgM;%gTcY<2vy_8C@$&iWszRF
zxbfuWTSEHLGy8QbZ_rxQZwb2iDr$Ej|2iy%J(P%gS~U@=6+_IuiJwaW!D-!PGs4ql
zIdi@0k?^-%h14Vx2Zg!w$dsSP!Q7;D*IV(#iGk`E?;cwPg%UK2ac<v^E@5y7OAst{
zc6P!s4mD-&!WE;e0k%L@Qs2-J(834|HdB77!+j4C63xqfbNtEwWZWl@21%EnrdFh`
z9=pxnUFGMn*HKpX_BjjbdQ21pS)khv1}Ps~a?4A@>S%?WNpVF3pGb{Fy39|VtJCQ!
zD|9ZNLg)zv()}OgN=lc$fzW@HA>n#temOT~YSS;d_qEO+h%#MDF5B%fWwfMQ@BT}T
z#3J=6$aL3?@SN8##i!vmF11g6`s0Vf{`{bYi|C&ND2PDLEg#jTw0)sxit#3Z4k(Tk
zNrrD}@9stey9M@&_hsv$)9Vck48XG%)muD5I`G5z36E?OGFCkAGP`r5F|3_kWw5Dq
zNfvR&Cgy&f!Q=pO(zq3z{KmWc=+5p;UejNpO!}y1sA@(f*8M~!0k-_{)VkGf?Zd+s
z?P4}N<H-8;T6s~=%|$*tnyA>aZ8S$6sI(8A(&5E}^HO%@0Jt%|gpdm9Y_1?jN5_-F
zyA2`Ew}LH&jfp88l#T3_a@U-kzC1ZpUwGy(z`B0H!5h`TrU$^x#l*&vc6Ym$<42Dc
zpe8|=K1;@sWtfGw7SXKGT1Z&swj0P6ZmpEAIvsGhv9U4A!|GA8bc`RYX-B`yp&WRw
zNuP_0%l-E4kX#5#w!vFu!>s|J-K>9*ho=X%6x2pU5)WPE28@S#_;5?ErnqF37A_KF
z4qTwB2kUp+-90#K^@oMKx9h+|0#XJ2{Fsh91o$(P!(gV$L`}X_R0Akvd&Cqt*a+0*
zWR>SeMn;f=&qEj0um5gC30ll?>JxN35l0yn(pdMW&TBRNv$>ZYx$41~aUCq7k%?m|
z`hH`DW8>q12l{~Vp~gU$+}F_|j6x5>pUqkH`el8_D;E=^;(s=<8;AeoBF|_<$TM+r
z|B27iODX|#euR0#s$gBOp~{StijpZ=R%DRS__&5J@%~pL?FlLr3Z<slJ;(|3tbd$5
zd-f~>&+na`$SasPY&dVk<?QT?$RFt)(c!^yMNw#aq(i#Q{E9eCRze#**C3(<cyE-E
zCx8EbfpO<R*>U~ov8hBMaJ4|@JM^K~<1Vl=u}Es;;v^AQBRxmZ_Y*T`(7v*h=EID#
z5`#OQ5qjxnv6e8NZSaRZQq0WE%U7&G@PG$9j)V%sJmU%&dp844LP;C&NPQp?B|;z#
zRzT7VyN(|}CJ}9g*lGuAKoAj**Bb3Wu$0JcF%5D%I=`^KH*ek~a}v=ch4Lkg$CZsQ
zvW=)Om_JnZZBaRbq9AHWU?zP5Qv#RU(a`~mOr(~PK8*Lw)NXf{%I2T}!6I;PDgo+2
z9|K6;jL09sCP*VLP!|kPJdJ^o_-&NkcJ14jfvOG<Jq_qLPuyzpY^$(#%8<B#vITN3
zJgJN3+mf+_7cX9fpT-42(ZhUZt<yfNek}6y-arESqyD>MUusQin!TTq5&?7h`rL1b
zY_C<ba^lpgjh5OWCN=|v5{Von1cm{PptcbH<jJ4*_e(!Zc0t_j7v^d16%r!y2C?;h
z<0J%KnKC#Mm*;bX=`hJ2|JhevLY`%l`&sQ?i<fTVZa&nvF}5vvWjEhsL`Z;NPW-~$
zf$jGZS2mrjI$9yQFs+=gj6jdRg8rv6xCf0gZt0$g^M2-!!sBgeEp#qMJ!h4_ix5Mk
zuXbg0@A_i^BJ@lBtlAz4xrB#3xHo_MDBnWAFAZ~9>HKXv;e^!`@oYSg=J^4{IV-+G
z3@y5KxC`13pT2y_v<{x~>*_k!BKhaf9|!{VB%S5r2g3kgnvH$Ue&DK9W;9%Lr$GYS
z>{h`LCT^AaY$^A<Fn?)|zfJQEJpW*{-^>=p$hqX>oTqroZtuQ(6a>)lW2h5Cwj<S?
z6iZQ#f5QBQQ9eeEbZcau!*PO}w$%_f0!!@bBaT5s;Q5Z+XbBa~^U{C7taP`Ae?gDq
zDEBK69x269ZzGqLF63|>EfQF{$W?Nm!R(#b$u(ifv<{<StUW4z31M-PQ#wSGA<3h~
z`*C|LFnyL))jicUHK^3u97l+7Zv2tl#9<ViD`MuXZ(@Pcc$QPb+~#lN>#$$}3D?H~
zb}oI#So-0^cX;|Ba|K8c!>NZcfAKp{d{{kBG6H$UTz-S;b+kh3qdy~_evzJjYW4z<
z%feTJcN=E4s-h1BjeS|pd1^;l2ovQh?aY>wan$9B9jj`)T5+p=3FWI6C@BcIUnIxj
z?3*SOyS-mxb4VH8tTzx3J=u7V3xH&n{KI{daC>BqV~=^atFeq<=dn^rmOCn`J1(l-
zMgn*46T__gQi^P4+f|`%7yb2-BHH`hRX2mqMLF>Pf#iJe0K+{3xqB~5U|>xgQ<{$0
zweBy93O#jy7)qY_33QA4a@|qg>$;9K*Us1jWvQTF$<v`u=*2sv<Qlt29xZV-2AAsZ
zH%o1}gtn5!*SSak`7v3E|E+5h824(al!G-WC+L2tMgJgP*Erz<jt7eFMt6YkDS?3H
zx{qW%?~^7Zl>IAw$b)^yc>m@qbf#_Nv8A>U{6D|Uz)+nRc6Mv7DiD2h#B{Hs@Trl&
z4)ZU3VQcY4O?7H!MobXxa1?(@c(Rhx((3sJM16>R4hpliud2cU1xSz`LuG=W_)Ri+
zj2ExmnT-hM?AfQUI&o!E(~O%oJtokw=)Qe+NJ&rLTj}~PS_q#?eQlKb52c*C5dY!R
zFK&qce>W3!_YII`@b82>$8@8+H<`iJ99Klg_s=0ACw(GC6zuPRF0`GxSSN`)rnE1B
zimCNNV*!+mSPS1rj||UVLIL=-Fx4=G-KX|%%txS^8i)7F=}=EkIV7=Wr?-&Lc&ANm
zHJLL6c{=0ybHpYH!!l2sQ9u8}w)<#^8u0|L6w`zYc}_Qs6cF(TiYp`4HB2#vM!yBq
z$LaS^lN%D$Uy0jJ{S{smyUoqbH8iRb^F4ja5ehv6ysion_eF70LrjdmZEfvUqB-Q}
zQrSnyok&FlD2%*&_YRZ{T{Pt5_4hB#KoYyXukX@m55iP%d0!c#w-zs6PV#L}cOjYK
z;rZD4`7zWG@G51be$DzFCp~l@yHQ;mA05@Nt=gbmE&{ICWgd6x>ip}+l05#K^vUl&
z4Wr&P$7eV9YX~IKJ$UKCS@I8*3hHW0>A;BHa&s$h<T`w~71WuNkW9cU%#HNvxb8t^
z5y9u<*Z<tvv^4e30<|xVB4vP{aG~@2fB%IfRIEr}L}36(0Y>#_ssDCzaXA7PAC;1J
z4H!Cz(I<SU_tJiiYH`;HTN(sm6)|GBzIE|4Sszd@VjGZ|J=~HNA7sc)KFy=9(%Czq
zp`lew+4xLA{;Z5>UB=^oPksCw4aXZI&r5hQk)O5eEdHzt6)#}P8?|o!y)vxSXL~m6
z=CN!LazSWk-L!`KA&(Y+VgI#}v-|h&FOfmmLq|(%g+_kz@}w^4S-or5t|4OQViAs7
zO@8cevFJBA9}BHrv*w<+H-LHabBgEao)o|s$B_uwYVqL*il7aF(r2Kd5v#iq%zj9M
z9Yo7MCCmumHA3}!RM%y>CZ<YML<E9p)}9`4k?mGOdYBqlu$;6uxk?v({OF5tMqD+x
zmbypCH~q)vl2!OG+rSlGLE15cI-qreu1+#63lue!XbcSvQ9L<(_%Ldi=pDu&EpcE7
zNWGEKTThRfTv*v(fJljP>(&;Of2gszW-=1YMM66;aYAH3C7raJVh16!L?$nGhzx&6
zydp==>-VG5(x__BVsI-NjMy22bQmvuGe%~(#dZ;qK>+<|t{-mB{{Br~Kj6UwW%Q;P
zU{(U^f?uDXK5q4-o%%x5mTulM1A+GvOz}iU4^oL0^4_T5l3=y9wVkn21!@UQ3NPFr
zkSPeAfxeWCQ-m}EMFJ%0InmLD&nKZ~0F!8I>Hzs&8#~152Od9o@C8@^dN$~;8<&!?
zZ@_Mz{Grza6Uc439J~VNquTd>{e1Bcluc8T=8+*|O=6Mu(I23pP_pg<@z9wlqJ76i
zI(Q_$8yOov*}9Ih*BT9o9Erf2fO|k4);A&|VtMH7%*@wB%`ZKwM>OT-KY@Y=PLNsL
z4ART*-@p6$`GL08_)KQ?p+V{idwPx>@P}W~lnyisV<SH`+<Zp^78yKDeS&)NODAY9
z6ai`gSn#u_=Y0KEs!8_0O>EeDNLvAKsX$lS!9xteg<rl{fDbj%>J7dGX-i8!Gu&5F
zV?q|BeHGO(-(VKh0J8AvCadG6OMR$zlgWBW3{O2jRZ?8cyJrsp2mqC#!OFzk{N?tj
zok5s!Ph#qc!4>cHcmBf#Xo7ujs;}P<<$N(QM^IN+gikc!2HM-tt}G;N3YczNy5tZw
zyW(3`ueOBpbeMf%N)E%xx&k4gN&1#he9iGd`xjm)V=0|DG5-7aRx}x*hlYLArb`QF
z%L0IK?}Bpyltvh+J2ewdp(>9Ta9J{N0`c=TU9?63TLvqnDoKL$GogSFPO+n-2H1sI
z80fE%K?wl!@psP5I#9Hg!w!rB&<%aPi>!>Rcc&?K$$T{M8bDfpx*@<3Sy<wNr0dmB
zok~Y<3tEBD#$m~0@EE09h}(`}5sA74APc9F_9NJjkZ;w|X#^-urmUjs_o~e+4~`62
z_IKdZK)uuwPN6UolK2)^clkqpa16wZQ2+-B@CRaw<5s;nAbajZ-ku%vweSo|Dt>;}
zz_$T&0p7mt>G>8245%Vj{a--9_wuCrhIW;I+vAOZi@M%yTFZK%y`#ghsb#N4>vaHV
zsD-U~^JuR##z0`3Zo2Wt1Q)}#pFkiSG@vn0p##3lPW9BOap(^m*Vuw?6AcZX)-Ir6
zqBCr$tX&wr{o_kSX8YAuZusCwp`n7C-rnr1Vi&*0k*xqaF5N&scntX<el>Rh)e11N
z!2Z<C1PGWtEJyiz83muAO)Ois^bg#j#Am~5xJqW_=B@>)pPcAEw4hH+jT-9?mZ2LE
z<_P8l#H0xRhG{6kf<QyJfVtE=oJaa#+Q=LML=7b`+P{DQg&Dx#+ZN_Cuc<mZ&Y*(C
zA%sgx$<2)%ItHpKrd3_CoJNbm2^#GJ7zY%wxU%E3vc9E>b%QdN0SyCz07a`l*a2y2
zX|r58>PGmB#B1$q@jO96@YhR+%+|eoYjg)c)znx6t^=n?IPBI2=n%}J&j_+nCPQpd
z&z=DnLu<}?o6j(okapGA6M8)L5}jx$Fw)iT13+OJsTh;iqc%j#EFd4)G_tgs%q_&w
zDsC1BM@Pc8>T!bn9EGwNX3bYZTO(TrIA=zO51CzyqMUYy8S+|SU_3#Jfxj^6Z`rbC
z!C7ftrx3zi$#$B2nWZP9`m&?1@8H>Y6w7I}Y$}EAH7JMK92>Jc!hqoEoc5F3m=d5l
zBsK}sFO&+%7gA(M(D9Ca{QS0BwZu9(AJ)`516V5w(^|gEREjjzzYP<w6>i3ST+Fh{
zKN_I;8_a!gO-Ru{fU!F#feIapJDpbE7i`uo>i|>}Ft@6M(C&rwQnGZGnqzJn>Nko)
zrs!0Wn1C%sA3qa`JLk5cXESXMm%8DM#EasVmK%8bR>uKUHD;J?VPzfo@q=a4ri0KB
zc=n95WH09W(yflQ!~APh0SchB!h#SIL^|{_XqkV;4c=J|3n3)5hKA;$lS*8d$4@-s
zmbWf%OSUa2XlZ>z`ztrgrUTEv`mN4_lY^I<Xri8esZ}H_8}UT=U1Fu%UAg?oo%{C#
za(f7k{{Djp@ILK89lh_~1%Sw-9FAJTg%UZvnVA`%O`;!@^vQT}A}WezHR$0#Q%gXJ
zUxXav6Eolf?h?wo+Tqr9dfK;d3|&>41`aZMZ&1s7tz)LqwHe5<uT#NKfYlIVGClHy
zCkJ$vX3p%AajJ>$1|0%s37dVpwt<Su*+l)qj~^8S7a&5hdczKeAnf<6C<UKC&HV!%
zeQ@q!29eFbL_ZW1aFucYFz!R?c&a1D*DoK4aFSkjb@bJRB-}b+c;Avg>FMbpm20WQ
z^>Oif)Q^A&*#DCHttW|Rca{7kp^%1pdq$&2*ExPSj*Y=*OiFyYeS<9aU;cVxton)q
zka_^93_%c7(H_$;(+_@&(+YFl&x}$H^w{K+RmRr;M%2*q=0;QW9b4>)AHb4vwmN0d
zTsjPx+u8X|wrxlevV)husd<e5J>t|{7<ljOMYGzKM1AO=X{OELkXnT{=hzBLNa*W<
zI4-pRaqc&84vs4-7IvGy7TdK;Sx?CRk*%R&2R^l^+fm_HWhD_o@z#9R-a$GKB&@pR
z!B{KuhH5UM1(JTrX0gwJ)S0to5Fe#awo-4tg0$}A=DE*d6;He!YND{SvzLsI!?B@!
zOm`*t(W7nJ256RMw7)Sw=a4M}U%sAYbJo0$2s3EJzb=G3$j)B4emyTY*LnYUsF|aq
zZy#GC5&3XHHtGB80W`?SuybF@-QU7)MhJo!(CtbN1%-0*81=gmjUP#C#-J%NvBsI1
zTeog;m)$q0+L7tjTq|g_BHUNNAs}Z7VgbmrM`z99z}3A846EW67EVQn_V`vJR1DYb
zJ3NY=N_yB&(Cag}!?P&_V%+<;tF70`ce#ozS~_@AD{rn}1wFo@dp`CW3v^@Ty~>Fg
zwUV78g0Ej}b2u^sZ@ITVMNU>0Ws_4DeT1#cdm-o-(Y8PH)?h3$8IRlour*ly2e;`8
zA%1|-^!0|nkt;4FUq!jfNj>fzs2joq`g)@cRX6VBR>A&V0s_8yFPp!7A+@6w`}~`J
zBg#rzQzC-;2lLxUc$Ar!YGkK=ne^!(k|R_vuiu!ra09_c=Ox2sN4{lQ8iNqf_m*92
z7&=cQV`IL0>8NgddQhL`)eY3|jZ8|~!OyQ?^(Zdx2LcWWaq*<)A{<ptZ78<rW-B3t
z-OkBb*bV9<`ldnyMzwe_5EhQ7a}2n19a2gZcRFQw4(rv{)QAcTTcUpJdmQGCG(#=b
z{Mw{vV>2EsY|O<H5Ew|dm*FGoZHIYK_wpD@c)38wtyJMa&;ey?#M1&#hyK3KHCGfB
z?{>|gDKA#<H2HtmQc2Ws0)m1rDDIx%c)WoObG>q9fHatChChCcCI<Q_Ta9=XL+~|G
z5oFl~o9zPsNSI@*W1prHl?n`c-C|*3Q4#|$`eXVTxs(wu4r1ja&Ye5ALnmF{yblCo
zI3U$z1Bs=3r>3TWqIcnR!fQSU_oG!U`;rt2Mi3U?C1ioX2@SFw*<RKcm=u(hl#U;-
z5nMx0kNH>$*SfKaYD2}VUSL^K`piUeuTut80iYvK;9`raLdx1OV`Q>TSxM;y+Ecq|
z5x~HQT5A0bON`kNn1rta>P5`}KtRCJvATp}6H$GFFU4$>A^cN0R1(n8jerzc0T{?S
z2w-p?A3kKrE=5%xgxasz2&2zXfQ3}}byz9X!TE)ReD2)A?2AgoW4zPBN&$-KGdJ>X
zi&EMFiv|9&qX+76j@&RGXdAnF8h?Q$cSb{Dp*NfOwMwXsIW3$AnzD(NRS#c`^oHp8
zW0q%Ugp>`DGq?Kni~68TH<5kTD~M_}rtpFwgwgd}nj_=mta~m#>{`<<i|#Zsh0Ds?
z6ao@>g{pl)K?Pgo(h<tOakx`NQMw{!$(I2t%ZgTjC+k}tx!O%7)u;%T;_><~1oF;v
zuvY(C+n8qJt=)eaxryxDEOnPZ%`;MigdP`y8gL^~^mLktpTk%r;ZcWBtCQj4{F4kP
z<w`$){)0I0-pljdp{GNT!eAD1`67r;Vf7^YwAzRDAo`VV^i^{^ojQz3yCmVsFoQNT
zEFA6X)n;3KP42~h0uhUJ`5ffg=md#^QRE7MoX*;W+mn@$a-e1y+NfP#L3j>%ZVhxy
zzwg0QM;oE9H+T?OVp2*40ru6;xrrtVL{DLdpLl<TNq%XNZ+g~)C}gp)9NNJ)dk{TC
zI9b>LNg9bywo(J$cag+bpOpuB1=a*SpoMR_zbO3AMVGYGiRu+x{j}|`C3JjchALH8
zgYR(MK?nF>Zkpni@^{XrmCD}NR;gv?*4C%v96ct!eYF7JQw~lBhJ(jsj@^ISJb7^U
z?%CoiVo&;BhKD~6?_1&g_N^mg&EHD(dw%~ePSlJO$7D8~Ddnrl$ZUQRt!>x{Fe5kC
z{y^aw<RyDlfJriS>~0^9(odXSJXh@1kc;EoyZ#=OShH>Z>v$%T9dN=h6+9cy*JZY;
zO>HYl;rZLwMNw2za1v0+o3npAtB+3<v+2ExO$lHz8gvvGP|$8k%z+>d(gdVJ4Vun4
zVio;VBN?~tyC6&gfD@D(Ll_%pgpZkU`GCKOLKrG}2q6(`<ltwR7>8-+Xdss1o;kFk
z0mWQVOd&_IBV$2=8Jy|klF>&GBH1AvC8%zqzYxDn{(~n)HRLmjcQAulkZ-^nV#Q=}
z3u*}`!nXK9mLHi=X7xRiiOXA~q(?f%)VD5YeT@nCn-1=ZgN29BIBqB|_34=q-URVE
zsstVW@E>XI|B19Vk@3XKRqDU6Y_iD)vrL4tfjbkIt)<BM0=_`)oFgY6n-5!B=+8Ci
ze%pbpiz6N@MhpEgnH^C?Mgh*hgl;*7PZiw)7}M}2stMa|&Qo$Ovi1XJE;b@B26Zv~
zCm)4lXRs<cD3FmQ!hCpf{GX>z?XeqeP9h5_pcJ3-v@U-v;biyN<@cxU52D5RIKq)G
z?;8`<&9>^sn#PLp@kxky*7_cN_eh=y?(>V*EupMB2x1c^&u(r|gY_YrlnHk|yTovi
zK|YeesV4&k0I{?|ga=bESMi3a0O>|R`5V(n%0<608M+Z!1i(NDoXx;x0-FSgku|FO
zrZDTssR@GyI3kGjF&oNZdL$s^2<}`SI}k%B^p?LxFGS1+sSsUwFeu6G#di|DYal3y
zO!22ASOk1JSklByFrGZHtxmY$kz&u)+T#nytudmG$ALnj4cu}C8Q~Epv$i)96+T6o
zm{K~D3L1|3vtmw5Of*rM%Dd8(v9qp&RTmV+pkd=Fh=mpnkil&q=f0gcF##eZT(J!r
z#O(P&jQX^MCX6>0FJ$HbF(V{Wn(wplK`0=VikOrZuKk_#s)97ABOlO<e~&mM^wFbr
z*6+|sUf3{-My!f@w9-!(&B}NS#nWEJL^6LXE>NDJUhKSK;RL2`Hrv!F;g$ruzsYkx
zmZs$nKiH+n<|EbTa_Qs(%`c2=^0wYyg-A<%{d(5zI_VeJO?&kt%6-25FRPtDeZPi7
z&S0eFI_X|~{$ZO5hAr)eMg9~haQb=kweR;mHlM@tO|RJ2jm&00bW9Ro=9mKd9CfMy
z%7n=qyZ(5|WB&0s%mQzKXC@PLP-k%d0`~jwzM2aV+w)pWTI)f)9D~b}p@ANu;k3>W
z!Ys~v*0l$)Z4m7@VJaJx2RD4PMvMxv5{JRp^@_Y1m8zEA612|C@9yG%ERA{dfz=Du
zVe{*5V*-V1Atq3)JBxS#^DFhwm0{?XF3x;`519;wHYYI$OVu4feKFhU9o#nKKm@R@
zy3?8x(CFvhd^&pk!R`UVtnZ5rMUr*=;v5`~3&v8(G)>~JhH<Iq$_RR!#KHaoa^6iq
zM&QlFo2?+nim=5BE^N!bE1xPV2-b{nZ$CzX0&_T=`~jWWfC#9<pfcM&tmD_FUi$fe
zyY!=~vYs~fuu8ov-Si=(BeU@*vQ097QN)dN6&pr2K@XPMGzj4~ad}|Po2>HiSDujz
z{sFv<%#JO&Pb!K;3cG02wwa}eaC$q}T<gk#ybq=&`%a39ivB_`{sOs`oc@>C@j!Ls
zu$iH2fi9!(F3?PL0UsJ5!<T*sTnU&@fjwb12cB^YThl18RiBtS6H+5tAdVRpPgDG>
zwzVih<GoFP8sRdCG7$#pRR)+MqI`IoYZKNPN7fNTjGi^24`vVWepCX7u*Yz2svfkG
z{wuou%^U?OMEM4&y_t;@G+$A{LWf#D;zBaG63GP^Yu7PT13-*!{3t;l5m06aADQT@
z;l&`NG!smXFK!>Gz!ivH|KW_V0^{8Gz`!7SC>qd1)e_+i86n;gh%SXuWI#@rmeI;(
zb&(OV(v9-eLmn&#i^X(PL@YWN+OYy>%1navA&HA_y<tq^>XbRQo|H21S*7)4@*N>a
z!z2PSA(}imoX~v&p)%@v3qV0pbW^PBnXXHAevK&Q`hE1tOp~EW7;Ome0abxqWDnRg
zbJ`c4?8UA}ii(*#`1oA_m)ix~fv%IF?vv-`)A1^iG)LGlFA}YNU-plolSbb?(udsB
zUA;bANXa35<-eMYQAEUwIV6K#I3L<J7aQGqD@xvqI?mZFdLicEpY7$_Va4vY^TEp*
z<W0qfnu}rb4}iiz3{NWMkJBT&O-o(~ks(azARrO!ULe$Ji2unfQ_Rg+Fo<7S79EF>
zizw^{88}`7Je>$+5f~FfH#F;xL0stu_^}XhI1^uWclvfRj51oj@ze=m4OJDKdPJ>l
zSOVN6Nvu-d>Zn!dn41s5M1VkLg{eVd&|`-%69j3#Li^ngBcSmSQ7PM)i3=buJ$bbH
z_^zE5!DOV-LF@CDvUm0&$HI|~gE_&rc?q%$ToR&6TbxNU)e6ImFz`)VxR?x<UAk7h
z5`iNOeIy_yvhS;+im|fj;P-2%0hW~oRVS=)BU*iNaG}sU2JeOhX&PrS0#*_Izf7V9
z1&;77QCmFewgf(Yys5z#q94$@5y|<1uTT_7bCF-5k&diHNfcrff3Dxhb%jH<7s2YS
z2~&(rgW0PFMDtg%3(g0G1GmnPLNWN#Bgq>G^GoYR;Kcr7zi78-6A0;3x9qSk&2~*@
zUIV9NPdp|V{Avb5alg4X>G;kPY0h)>+|Rj0JXKb4{XLn@BM8=z3j*<vx>tnadn*l)
z0X%qRJW8fJT~9*Vv&`11H`249g~ZTE0U)cotcZ-&11`gLpcaP81={#Uth_K7n81EV
zw3a(a&$s>+p7*HBXEH!qd;m1U?fcQ#e(>fi7DsT>zq4~T!rX6KSfWn0H(&^O4VCuX
zQ(^nbXaZEzfC5@S7BT6hsuKY%UUvUs;j0Z3ut*V>A2(SZCB#{5%JmM3ew>a+wXs^?
zhsTHu^^#PEFM}@(vnUR3W5q>pShxyn?#Mam1-s&a?8Wk9A56=*0p3?Odmq=F75pIs
z)45<Ryx2S7`RlNGraDe18{B~WQbR+-sHfsH6oHDX43W-@<Xk9Ue}d%nU)obQ7=i~;
zzU3Wy3*PdWjDR;S`*J@HdH~L)DQfyva$zDSiZ`RoTJsLhMC8JyadX!eB9_=8Hm|~G
zL!9w=BpDz((lPd&;+5NZSFm7~bh+k8KYxjtzbQ|<j`T2ytrziZu2v!gz&{0uv*xes
zQDmd#x6-+B7-xCsXZd!m8kFHaDeJ+}s&8HQ&}qC}79x#rVB9}=zm!FP@AY!+wN=R6
z8n7dtYOA6vg2-Y+uGkix?wlV(#DGk&L3@8LTNH4})_qsBx>)Z9lY@EY?Hgo0Q2zRK
z|4>9Hn(q5CW9t<#)2a9FzBXUOaR!mo6`%QI4Q?n3n`j<a9gih;rd*QAMWRr=Y9X`^
zzJc?e-|&W7FdL`Poh0O_6{*6<1m%%7*!4;rPc#EDeRKGDZxd=(L$HsX5u+ES7m1R<
z28_*-CKf^3Zi{~#qVsmpcYn*U=%-_F<&+!&Mq#UG(O5&I;XG)QKOxd%YT5Wok@=M+
z3Nzp(C^~Dc!;otgzHvxvc3n=uO^)*FV|*&w@YjI7jsT(~5bEFQz;9&OQh=IFt97U(
zt(~fZaX3wYs@8twjij`!k}DS|j*j^7!gs#%V=8Ac?C?(Z-o6s)w0}QVMSg!DE7h$M
zCgZsOipQeFd*Z^AA^MGRyBTl`gffabP`bb?3?lH=H%8vCTu%SH9}HH4!bA87q#l$S
zGEvb*&b<>LE0{bTkvte`2ln*(xf_b_a9L>3s>rK{Dcp#(8@N{rf?(vfaG?EAJ`Xuy
z*$KQ=9||7FWRHK{(h8#rf;Jf(2NR2Z0<DN22#?tiK7$L?V~a3&mpSB-|N729uMKPY
z!#uk?Fwde4<BD>7LN4QQWZ?N@RLT`#$I#{J7<Yulg`MxirLMmBd<x4KkR0<aTU8jE
zAiLVq?vcRcT;Lfk^d+N5Wf+YdjBVWne!x&ec#$VlLeR>&HjrA}ISEIgRK8pnBNCHy
zem*}ITknDVDjC>ZjP(P%^4EKDh993b=3~;yx@zcaJfqpi-T|KiAydEPRB%e*<4Gh<
zuDIspp%<uS0(eM<$USkPh!lI*o_dI<0nL-34t4;6W5`T0f|tm-Tq6}wBOiz8V5qQ9
z*`f%BGJ#fn3mGjw()9*`#}(L%k~KmKTx*xpb6X!Ek2~$3nbAha7ihR)M5~G(Cpe$m
z`Exz32lPy~{$k7g#G9%IF>$Xor{N!<=OLKWgfC*5f2M`MLq(c1M_FyC%*Q8hDs$-r
z)`KDz@pu$u*6EQJARc(0l_(g?StxdHd4obI0u~#%#v_IhntX*#++k_7BCPNg8G}9Y
z`7MO>J5irAmPN0Fvd3^z9z+a(ncbhOw#DSd<$p}Qyt9{$kZ*TM4sQEHCWn6Vob7hN
zx^vp^sJ+p!Fv`BLuI1nn)HZd-U4<<^pH#t@5iH2!9jY!&X6cAbH}vdauEJD-%{O|{
ztt6cDgUl)rt-!}lZhWwNETS?6Lq`wK9V-n+9QNPXdk>(jv!>gZadcE1F<~Z163mK$
zAc%=*071uysHm6*K`>y#F@YojQ4EL*m?K8SfC&L*Kro?TAPATc5EYbge~mNeJKwEa
zr%s()b;DQ3dS7eCk>~k`y?d`-y?XT(Mgdi7$$Zz4E0CI9svG}N@Z&5mSB3-@X2-I-
z4!V6?88e<oW|SB=FUc5t@Y`B4`mgHZp?CKfC)&)0vFBy|eR5{&wlKf1vSgG_{dmlY
z<!--^)brBo=^sDz*rZ2|G9P60?|g0aE`6S<OfQNyaD>##-PA|WY_j%lCKh*jwIO3v
z=LZ-5)dGC>{E*cz;vU0cyBMFecFFY_2n?%bFExAj2WST8EWrw8mOTnPuf5e&mmSgB
zJu-6}=Qml^=7c?_wC<GUH*NdrL?AEjQ{gM#!Zke1T=p*gBXxuTrQD>vG1UD~nl-YR
zpu+MDbX~-T$*{OWu=@PMZ1Ai*z5urNGmo%@#pOKXBgv&Ii7S=iU^*6HSzp1?^_$pL
zKDgPJb*Ev|IGI;dm7y5)WsJcK<rg#gm#QaTlXy|E*6J_LffHq%G)FUK^->wa$7q6;
ze|%t~{I=RAPv*%4LeG3o-STUggMO%@SgYc(F*LGrM{TRSSGTMsB`R~7kC$-*skEG(
zc0OE|d2bDSup8o->ujsUrUPv~o2~-Uave*{EXR72Z~vYUba{#EUt@<<6_mc;>Hk~I
zuPGy{Il<iS%uzZ}pz{ypNC~cOWkdd@Ie6v0;`Y<rhVdaAE^gj8b_aKiz9C8d81aSZ
z-}L#)+szF&CIpV&*;ixm;47s!4_Q<WwgpW4nN4Z^s<ex=0>0TXwS0Q}xDKxFY54QL
zm!#tAQ{)<cby>$*<jcJb45i{o`J+)y>%?+e7z-9+6eZJPGAtNBUC8rMdnUu&XxfTv
zG9s{J|E}Nr$fO{dyvgz8#?%?C!)o!htbOrmtpP1}#o5w0{Qge+iK_T9qm-bWJ^8@!
zO<<a;bQ+a0lT>-}5lk3VT^3@g_6%13`Vh>QC!LuwOh^lv8uaEKGKONa+I%z0vnn+o
z4zcF@+${C$S=c&)L`GT|k%`h(lmO_WBa)=F-}m$`w$|S1qwAV*fc$Y@>`o2n5>B3D
zaf<}xG-DotN*kxTQV}vdczgI3BbWZgF9zQm+52|uI_vGj;G@*ms(P)@^eH$nRNqId
zw?`A5KJ~jv+skl}W*GGi^JNub8PvwqT^1Jn{kqX1RYqear>6e?90v3mP*nfZ7mklS
zv+><eKBA-NQzFn*qSdUR9d`^&f=YAF6Ry2U4cu3@D>{LWM9ajo@YwNFe(ISdp09vo
zSD#ktzm>75xt&edAIeg;shEF^vn|zJx6J^eUFp7Ik-P}QB|GM=f4TDg$zo~69=s%e
zUL+@?H+77Ry=_(-VhQkf)TdvzR$cl=4S8g($U!W=d2v=n-5-i)|CVVDyQFgx+GW1V
zLm2Ram|9F4c=!@n26%?OivY}sw!@nZ>~p#a$iQ%U$M$-I2mc_|^^YtwCK*Z2oJtt+
zU%kQm{Lbx+s@lD~D^S1fTF3JX78Oh%*5low9raJqZhT=&n`Py}6{#9wF$Eem^|@5X
zOtaz4rL>!EY1`4Xc=~e_gh)*jdiT9)%COM9<uZ1?a;df1Z@OE0UoE|LwLLwm#gC3Y
zB<&m*VKQ*b;uY)jSCci=Kg3N550U9a0(%=-w5`1-eI;$PE*ok#v-O(w@@)iBd~gY%
zDP_5DY5b+8-yuOX_eCnV`#fr~-KzC17RGaA-%9st=3jeH_4`w^a~#Q%w^tbtaLqA1
zYtg1#`C_efj>K7MgKVOBPSPmXbV!E5p_rU*DweQXHJ4g3F5|ESAnKVpU^c+!%C8?f
zz9<rO(Rb$X+4uQ588^Gh8-QGnjJd<wvvbN4tE=1nIrE^_*x9WXP*X7Ppiv%p4BE8y
zESo_oBmSDLbU>>wD^Tm76=)S&vDHqJ>t`7b7NU=fI2lr9S@B|b)t%LY#>OP;bft0-
zKGbvm0VEP;HE(S{PsdR-oSg9#FqkK*zUo6P)im9+vscH=zLqO)v-s|2<61*0QcUQ8
z0)5(OEE#yC(&?DL$CFxWLPgh-Q&NmL3iazW^40Ivj+j0k&FSdrb)nNIb@~3ah|HzE
zN2;7?uR0wCh4p&Wz`tKnf}CZ=wwi_C#M_wGs3SL=g)ENm2hR{+M*5fg6il+E<P7>m
z{a;F0RmHUj&lJyw=9mulpi0sHPZ?@H<Z{`_QJ<S^c=8+MN*eVsA{dGG1kf&wi|y_K
z-3p^`YfhY)1p|T-Ow-I0VM(`A^&P`graso#*PxHJr4R6Dvs3#&yUr+>ey}vjsbQ#3
zgA|qFmU9bRc4t}E)^kTD^VJS8vsmwJ^x*)t{<nR*3B%^oApFn3X$LMOzkk2|IukxH
z&+V_TO0#LUOfZLyV!e|D2^B-4GVOYww!{4DHuDGHqx?k}lPDZ2fQw8|26d;kFE4KT
zM~I)t6U~H&YJP5&>bI6pE$V!3o=5jK?Q6Q~u2`0r%n;{+3&zhDBS`_RcPz7DXg!@H
z@~mLqqB@hdsa*bJNif2`kp}H1wX(XG-`Kf!xAjxf6X{6}@8WwDC`bj%-?dNLluKJi
z6*bu~<ckbGx!xyNupe`P+r03gP8xC=H{n*b!T^Y6^)}_Vb=e8t-k#3xhd!td)2!Q~
z_k5*l$V0#UiOvq@<#w;_^nD(L>@wQux7Q@byRc=biS`6zOFf}NJ`{J0)*V-&wV}M*
znU?!5g6o8hhF(RFsr9P;NuH0e(7P9&nh$r8`lZLVg(%?a0shBJ;9S%r?iUG$ZI^Xj
ztLfa;ykbg(b=7x{{ZFNuC%t4j-~ij+yCBBjOMPL)758m=Ys+-D@K%Ma00+21cM7C<
z*(Q~WnCRue0DptT0KNU{M$$Lu5n!1`&b0GmzxF}C&*DJ;rDi>=hNRKv>t|Hc_tnW9
zFX+8iB`w#!n{l4YwJ_-VW>@;#)2!Xxr{<H=3s|shjr3KHa0ba4ay-OrYrJcOrj4FQ
z6IG*>5sDJja;<#px98;78@30{5td-ur2bP1Om8JlPUsUd)nU{vjy*Ai<jC^o4Ztz)
zAz%>9J6v~?!O*m<uAm`~_Ev}CzcBQW#H+8#%)=zb2#ns?dVhZiu6WGx0^5>nJwnp9
z((B225f@4q{7R3)@2?K+Re~1ZcX|#VO)Am5@Z2q?27y1bh+m~!k@moM+Pb~k>2W77
zHP_Fm|8el3&hm$@_H&QzSaS0{bDFoDvsql>@tBiPbT6%y-Z{n0+&Jug&!lOFgXcQG
zjLFFJ-rxE>NW1uM@z$zwZ<&(VJfW710*bfIzMU$Hn;0+CIdWV!Kb@>IOYwMtr0K24
zfKi)vznrQuKH#4t;J|D?Sh&yRREw<|g(HGBPr@Bk8-KrhyR`k~1}@!omMmP5aunNg
zC}u{H4Po_ibf!lbeP(yv%MRPKXD}MBotIj6O@u|h{mo1a(}9qxbTFf%k>RkA5Xqev
zw@LVT&CuoJoi?_qt76%*JNpl@*ZIC>5=nl)ZEtTGGwG4FruIx93b5hRcY$74#IAqC
z^%hAH$Ce4U$$4~UXRBqD&@c6^mD$~PD>e<}?@|dXS}0Qezdq}Y8goq3r0h<c_tZ{b
z2pQk7v!wE=shK}M73@6z>pss+Yf2}Id+{*fRi&=F#`>4ZV0Kp+^gg6Io(s=V&CCVR
z>tz54gZGEtR~epYC6oqavCZBBjbTYkq!wh+(R4HU=-p3!#eerbdd|VGo|K;7)L@_H
z1(JzdmJ>A@DTs(wW*|vL;q)BBL8I#}%m)#1ueK+~OJ<wDx|>aPukLA+e*nlLzvZzB
zVF7JReGk6YHLJ4r%x9^FgxyqU9^2~A^YUI*tNlS>a~Nx+6TgOm>HU64F(@OjXz%jr
znn~Sv%LKDm+a9pmXtYiM?=<)(4wK+LFO(wTqvvg6aMLMPp~WR;ZN^k=*wXDnD@gyr
z@$S(4O$S-j?F~Cp++s1%RbJfgeld0h`|iLew%KPdAYag=*J&thdFRDL-nzZnZnDuv
z-z(Ti`^ZS<f{*(*(aa?K$>|r#tZ$}PZJo)bx?g`;R9GlnkJo983ey;{v|9C+!LvUJ
z8}P2f@|nx#2J>B}wI#xFg0X?S6WMs~(~xwJQKw2z93e?a^1ksDXczi=XQI(U&fVte
zt0gCFZ8EFdJYaK0U>%hGeEZ*I1pFzgPEj8_N3QBHA8steyFWGefT}r$uRJftT;Y*@
zzCb!P1Ln)UU9T5F1sQ}6$?=L36U7eSlHhII+FZ-R-?Ep#N&##`@5=?DQUf@0n;Mr4
zRU*;u>9h*ub-QhZv&pT_u5lXi;yCuVfg@*@lz4(lCTDv4RBfF2O=&RM;;4_OO2cr=
zv5T`ut)gD{*4Ep`EeS5q$mC`f(f?0w!E>KyT)B-}d6*q+jg{S&xk?9S{Y!u*jjrFO
z^QNtHu0*q2YgYH!Y}nMg^6-R>x{V}6tz7yv*>c}CJPn8C%f$0WV;9pcshH5)ZRHgE
zkuNPfe(pHNy4BjaYVIIEr$!VjU5DI0&Xy!&C}9`9d{4Wj3EQx#+Ii)X>DQ)^1)Tj2
zDf8?DJ9L$<e7u+)wLAxzDx3rU3*-mi#a#vXho_G>z$v=uXbt;P$=0!_25Y@p-L6f9
z|0`s9loKDI$zjtAk4Lg5&l=fH1!M9YSkLcIUFp5$!sZ6d7DH20MU|B4y?}COe22~x
z{i%MEFi?nap@74Wi>AB$qd1U4B@Ex_1z2l)?M@Q!3EJ!`=L%802Phfj!i_y8=Z9ON
zz^ds#tbV5$d?4ZMYaGDgxN$JkN6qw}1IUUNz9mJ<U7M>l88Qu`QMJ~c@8fjp?si%V
zt*bSTP;F23v(Ph@XlkMp#su}&6Qg#GIC3Oq#CI3UFSq-(sRuAohEIPKAHKd(s;eOZ
z2a+}Go?b&-H|&Ml%HjiVLaQZq?7QT}U70CZYj?oz$nZ^ykW>C85*<pv-3HH5Pc6z=
z0~tCQ?*Hl#-3Yf^9JWzEuJVRMwk5L1`AgAPHiiGveADo)CgoNUPJLT(TzfN4mvl`1
zk))!8z+5fy1FACCe&2j+^^=uP9hWDn<N}HQ{x*IeN%D}fCW$5sA(4&{<Kxoz)m0Z6
zCcL1U!KwZXum}#fylBRH%AMgF|MrlfmUao#e1ESw4*0qAN=HB748*k7+@fsPE4|`9
zV)nP$Td8FoEz2<DS6=wQ_#&`k)T&ccE_9EV8D^iC{<y7PYgZtK3{Q&l@1kSoWG%C`
zRB77VVqdy20|Q?7>i}V0g1}R?R`-&f<cXRpr*161H4HoF4%c43=T!4+Il_Ga1a;vR
zccG`pD!{2WQ$J57s_u(Z-BPI1ou&qVJ`el!VlDFr)^8s3c>Bci;Hk2+lG}tC73>&V
zZnrU2<I?z0pNe#qB|0sVEH+*a^|{r=%*OaRRpe3qTN^&XVN|ocP5aMh8-?kg%Wy7+
z<H9A4M2#7APQ6K!xgicM%12w}c3_x&)U}uLh-<PFYNl3{ASe>h_wqf54fcA=U@Xk4
zOK$@!(bCEIqzb5LYHG|K?tsH^wG};<Q^*;wK84tUAqeMwClz!)FW&Bd4}2;j-+Mhy
zS5ucc%S1Z%1n}eoFoyLiSA=%Y6ySX|jnuN+<xO>@l&ZDMRzi7UQ~i;0#bnAr^5rND
zD9y7<>gZ__&2?V?t$tolaKlzxyYU4gJ6D8`AbT^;QmKOmsSak8hE!JI13*DyJgd9c
zGZPl1PMGS#!gfFY*1I7Ul~3I)IAUt34{jS+%U5=+E2->4qU}_1ExoB)+1dNOd)x}z
z^o0;M<m`;)o<aqyF3$;hr<d2QrU@ao0P4Tp*AkhICse17vi)db^}qGGyLD|v^uF4P
z!v{N#FlPSCYIvmb>2M~c+zN)@Z-AEVD8aG);e1zf-tizi;grd~(jC1s_~|*$Wf_Ml
zbn5<@59N?=vVFEh+G*A<N(Z94{=@rp)AVYh>7}r)qV?fBYg5xK>iB%;wN8GXiEStH
zyc?>z5&`L=^fS99_viI=S!&ZcfVwh|d&TqraNqdUvDX8(NHu#b(L0>t1Wx>iZuFI*
z+R2O8lof@wtvf-a8^WQGyKBytWP|*p+H<2h94-3j8bg;G)-)S<pr?Vsa>}9ZYm*Fh
zSE5`@s7&nB+WQKaqJ(4-{FtP2QoAq=ccB&xI9l1xR)uWshm>exh>A1p2Rj~TD+nyF
z{$=aUCbyyDClN0jj`;ZO96Z)w|7H6%ZT0+n{{p1DI<KFy<vka<1hknl^t@^02|K;<
z#2v%W4{qBp;glcgw~Js`zUBI<?IX+Wve-E1|J%kHx^ZTx&&ZvDN=Ii6GLF$;qYx*b
zqk-{GGb*~1vwPPrFHQ+F)JHhexm)Z+bMqj2@xAaqVmaKfNFVTOO8<pTjLX?+q!}Xm
z$k@7ee3WHJ76;%XfA%Tpt=rIopT`{bi>`F(rL5@a@%KQS1`}SdZ<eQioHq*~kkMb9
z$a`k2NNiQ362s2xwI!c%3FlE7nrF#G8rdWwA|#B%O^}YOBSDaTapcp%#PKlQGS;2y
zPDZmt3o5}<x~Cc}Og&ZZ!8b?Tsq^<cjld5%@m!R8`OAACxFw~Rduy;bYU|tLpuQ)(
z>uN{s=-I9Gb~ZoGYNpa$ySN}P<KPSq$*NkHc&?Vvwr2j_;2{jTrt%|v1sE~u)2@(^
ze$4izD%U;Dq~c7Nlb22loxSy^Oy+-g%8bxcHh!-4`!<NMO`*jGfzD%aw_>*RXKZAG
zm_LGwh&}jsYHBsl(UW<y)4)<ywG;auB6DHjZvp!jl>QVcDUr000ws-cN64SyBgJfI
zeF*((^AR<0_N_#qg0WvYf%y}jKJzZkrP-%qn|CF)!1$)&&i;h@@%R&yV)<-o2IFyn
z4b~TS@A_i#+RuA<_JRBbWCRFHUeZ)X22FM(r<)V+(7egd_eqZMPF@6Yjm|o$%zU%>
zh*u%ZJ0cvJAy0wBgNV2z?dfY1%3<i8%BU&b?!{c@>5zW5Ys<IA1UOMWGONyQv1zwT
zr=y}~|6WlASb!%h1Cj|?flqzGRS+>TSIeMEFDD+yrtKf+%^$#SkRzYuB!vDJUKI-Z
zB6moX4rf9b%Z5O{2oF|jeA)OUt@c+cYU33?I_vb>FURxlEMw}cqi2RWGw!N{i$TyN
z?VX@Vs<Zznvv{QOoQo*48uoBVn!9nP5Rsq~BHuRJo=B`bKrdd+CbRpad3gK_&Z!R4
z3DGB3CqwE$yr|?I{=MUmAf7p>Zv7pau7)5^gHA1t16Y=M@E$}~hK=3JeNPfNj2u_w
zZsI1*ayPbVt=|95#OWy`thWK$V*u@C^&5G3s86)%t-itrHgTVO?6BLo@XYYaF-b97
zor_MG51G;{U5cvX8!t_*I$ihu^lFz}LcHx#ySm{U=f7tM|NYIM<vmy@B<6NhZ_91Y
zMf#qP43~}Fm8vl&rf!q%#`!(Qgf{-S2oh}W__~Q3q<O)fGoMAelXJQMgff*q)GO7h
zoQY%%MhibSV^-^#E2pp`U`Sg&%Ou?@{&ZKtTHC{_&0Fu>HZxa-l3#SPxA4&#N(ZAz
z^(`-c;Bis)v6*0!w4a&H1i-#$-~82B&5+0I(AIW9t3xN*<JwO5G3j+^-PG+vuu;s0
ziiFZFeg8yj)ulOFtB~CG36<v8i`)a%Ojoj+kNOo&k{Qj(%YWchSVaxip@V$TVK6hT
zj{VNQ@#}F)uk_|Zg+yHstIPEtW-1?MDj%kyrIin7;4u3>I7rIS;}bnj?CR%iQMr`*
zJ8b|?kQvH%+l<-xc0O)aWW%}YTY*%?1E>FDvp!YhO=jIK-CGsfav+7RcM1B{{2}=p
z{$}GX-2?op+Zj-sk$iB|%{$t+Dk{LvGP?F?Py<7Y6czg(HGNiVj<zMix3Wdt+6Xmt
z?T&XX6}t>qu<9WXQA1&9UeAAAdXRnAy;_W7I>-A6D-V^NMp`F@ZUy?DciitGkBu7l
z{{x+D{cv{F@?B&yHNTqtKQ+m9mwEO_ajEB+Su)}m1)K38#ni84D}<A%unlN!b12I@
zs5VfD@z>mu!SS&t$53J2crQr@y{pT{*AG9PfA3$#wpB-3!1?t*7+kNPzvwtG3bxj!
zmokf20vF&kSN-_-qnWVZ%6ETNh?z$Gly40?wqi~N5KgRK9}VYn?0_-U_}kZ(oqss?
z2WOmphZcNan7)5}{TKf~{|Ns-EWiGLZCU--cqaEoFJ(`59|O!of8%bv8mLyaRt1hQ
zqL#WuF}cQ)^)xB28?Zk>Ng+wP!cFN2+42nz>r?)>Ptbd@8|ZdlH(=Jd_6mlFmZF{o
zhoy8n$?uZWpS6&m_Wb$N6bWD_bQTIhif~wa;Q$u2Rbw8d+X>Mwg${f~l_GSYGJhP?
zlt5wlgzlrlFo&_-Y7G&#q24^_Zzfv!7#VQi<q<hrUW+B*f3V4+%6$HV@LG)(hLdUZ
ztnGrhv`=jp{xsMX<@`E35)UYe2@$Qv!g0W1XxAH9S*ddYL-odgXdKwJ%;=!}^*`$W
z`jHhq_JntH=-Tc0ac)9RG(H(lfa)^9qt8%ZV_b56&IovZ9Tb#it-Cku5yQe2x;X05
zu~T<Cx!Ji%;ySY~dMES9OY5Ww5VU`*WINdV*_`n0Bc&%SSV_LFj>(fBDN-h!S+$e`
z0C-2ImF)YNU%A!S`~1$B__q}k8pC*5^L3O@Pt)eF^HThw!>28=PD$h<6}lbfJykm(
zZ-q^dSWyShFP#VlVz;#}%z-<oH9yd9er%L<7Fn9gM?2y0FQ#-`Doy|aH?^NWDN|f)
zhAyc8o+ua)DN)1}A-tMSVJWoUXU2XMVKBMR&4W`;6!ZRX#155zKTU}XSQhw7M)%0;
zKZ;46g2GXW^#$P}AsOhwD%aoux-TSToN@iPD9X3=Q;wVQ&)r$blj{fmlh@w#`Eyi`
zkhoU{<jW0=mzO(BtbrdK>TxPA3V!}zz0FLFAJ4aCHid`{gP4hyFhEQGy0W~nS)%Tt
z7!Z{$%ocAwY;FBRZ@tU?A~cZ9F6nc@jy38*R@Bomet?tn&j1`AD!m&rw;eotnBEON
z)3STTBsZVf6-RNAQE8>d`M@Dz@^~CizCT4;+V%VLFGJtP{_j!k|91}f{i)+={jm(7
zLm`4XtSlumNWso1+D@D}fu{gWx(gXIe>{~05gPL)d1if=O_o*`JMj^^zW!;3(SULE
z{70!W{?rA_q!hGsP-ts;dA!bj_@5#YbWrUnTf&Sjp(iGdOs%Z;cdE{#FcR>Rut8>W
ze9*iHubDoOwy+8jHRMI95KK&o_S7=9M?)>pJE0(SRPXrWC?Uo_ILN+wd|b*KW;9H=
z!bQ<xhzrqY=-hNXQWRCZ0R~)k(B*Erx<z1Ex`1-1)(WS?{|Ge*XueZWMX;93O){8v
z(ur5iS=?HVu4l-L4jCIOSjF-@a$sf1Kmt7ye?JR#g-!P1?*@kYzmsP-`Q!#|_YXtf
zAn6tPN5)-#|Ar5=@$!;sii=sy-42c2JLP<{9>Kosk&$cj$cA>m^IbP>Vf3#<_fr91
z%8*zDcP^k~zcZ^4cJ0CWrQ(S1c*2&HYG_K;8NmFH?Ks!=mHduLxo|!k#gTBRF_C9o
z7Rg(GaxV&dhmBh*2JhNc1<e(~!_iq8hUPMrXG2whN>;awU)jI)QM=1#E@w0zucWa@
zl{Rja44BHP&YCbz6SN&a2uo?H6`mb=B+kA>fr5!#A$h#%*yg4@Qd|n+Aw9ujQ{e7+
zP7*j-48mMUU?E&(27Em-fa^k|hNLSNRX0W_CkUIw0ZBFuZGXsO<4umzxJc7OHAdy#
zHq~EY*VKaz)>YSjHmf_wy$+j-6qRvwulaoQL8m08jbBQZDGFL1*bkaA!1)&`U+AcR
z7&ixofeOniwBwK&9iL$_%X|b${1JT#6HMG*P;s3n`abjEeOUrWw8ZFzf&}W<{{7GK
zL`3(MAJBg-5^4zH)8yyFR$-^?M%73twPs6YO$XV{#oIJt9(i@J#)$P7p(3zDFR5Bk
zTm?y4M|H%srs5$EW1+88D<JEq!%9f3#EYn$%lOFVKq4j^jqs0H9roZAu0`<=60Fbg
zE=6I6nw_edI#o^-{re1l{GfI*HgBBb9EeyRyD?c_UCpaHQ>W&pnT(QUZ>gNWU|5C*
z96{S#F|I>p%IYVJUy=RiB^U;?Bho{?@75&_s&)3;>${?k1}Q1nWTV(o@tji)eE{ng
zx+?Bmn(_PXWX>rmTRhCuKGWavQCb{It9DSNb5DJ-I4*ZHtQjvu)U(5UL<}K<-<`K%
zaVaF+&bT67u8nGgq|c=yNUBw>UY@ek3kwN}WqIr@OPZ{<hHct!`DLbv<*U4>1eihw
z+p=PrDO=h6M_Yy0Rt7}>{qMExR#O`Bt-gby@xaHeez`19-fVlXsb9%Oee3r);=duD
z#dc?dwnHkBBr-@)*tIR=i>O_FkW#icy)pjPuOG5JSAhhGA&T$pkcZV(r`J_OT74=W
z!y1-pn=}n7sZ6@K=&ENF3C-b%B{rTvI51l`G1`pIVGZ?#eg>+BZu{ps|27$k86_Rw
zkh3d~*PBlbT{V)`ktxlD)egK~TS~M$_eh<1aYD_Y$&K$Z50Ti&5>QR}P4y&|7dRcf
zkk`)KD!nO`eV!12ZbTxrdy692b5-iC<*@x?nD264#9s0JD0LV*T$a3f=SORkVKZ4L
zD`+BjcIC6LemHwSbZz}51lC^yfn@z4>6x`Qc_8;HxMbdsNaDjbU0bE@I*l(^74D%%
z+4_Tu$I%_c9B1}c#>alm-gv>+vAf#8pbmOkH*Ue|^GIZpB3`N~ZW;ow9~Ic~6@2A}
z!plx>Hyi1=*MFxVhH8iF6Pnyc`zD2y*{U)>1FEfiS`!>C3OP=~PppU%euzk1)#=>1
zbNdi<7N-&1Fl4e%btgnl;i8MbYMw4evU>`}@y@zUrI5_4DfXDF4GMyms>YJu|MF4Z
z&wu)2Dsj#i>Dc&Y9%;JQ7G}^gphK(kOR30;?Gmx3$An{#&_yBZTg>k3x9TO0Smu+S
z%Z|c;>4z8r$!r!X=9JP`*rkJG%;L-7So~a5sV}ayb4lIuPAr6pU2T-AoX{N2fV=8v
zkmxnhzmMK)LBfIzgV<n0jqQlf-snT7q<)?K_`!$+c6`l-!mF?71wq1xs5uQflztmQ
zn^zDdt#8)bx61cES0Eivn~W>qNk!}ILm`dgX$oy=9GUFs4c*`JFz0&aL)o?S_+xGA
z5--3h6^Yg_hS017VKUr2@-U%nnC+xBaGD&}<2Uwea~>{k|CnL^Z^4glkA7Yqdm!~I
ze6r5_mi!+0td<-E4!>*s8vJ0SHS)k_#U)9q%X`v#;^h3`bAQUZEzg-2uIwo)(Tdxn
zoO^#=LIH{^8Fav{PVQcat#fg5Q}BY<n>csvT^LpBg$v8HC?DFmySxFJzuvb#QhPFo
z%4}85!dcyjy=mXOvsXr@zRh-taJNW3`m7%UCb75;a&^K$EcKP~yjitufV05i)q(hY
zO`PU#e*NxFheu)*&zZhav)1fuiOG;-?-r?4HXJ^5(6b3cAGFU=@Nd#<y`s&qh7Cg;
z?Q+{D`i(kp(Z=ydEB`62o==?W@3*jR+e?0mO(wN#b<$U_%an%W?7o@@tLi71mB07g
zy=(9;?QOH0y!zy1zzQl%$<h9tqf5)fn~iP8?lkjW%U*F_eiC~(jwQBe@~53u*ny6V
z$>#|Nr^$Vq&1!bX2*Nq!Shyhtru>V{7%(Uf#vEFX_1KRb%uT1Hnm(c6OP#i5wwjU;
z71MUvY6DV4B*{G5)r(f0it^&n<DaXuTpO*3n@82)70Vo{gVD05;uiqHENzZes{Ipa
ztnGLnM%>Z7<{gnJtS8v3a_j!}i-;oI_I=wc^kM!Eaqc@u^`v2A%o2=USMl0HSHFej
z{q5s(7~@6G>#Lr+j}mivW3-1_PV@qu8NDlK{#_BIN!iUga0!h9fA8wATk<`hOke@z
zVUBm-m}%?1CrUjku_a%wpd-XiZ9dy*Yjz~53h?XJRwq6+?xR=QJf(Ln&sW%-XlQF4
zICzxlUlhb<bH>Kj&(fJaxWun_8{M&3c8smv1U=uMPQ18R`M@<&iio_U2YiMfkUGMj
zug=eRP<;9~&{jyuUvGFby{F5(i8=9ZYa8|Iw5`aw`v&8N%{vHJKkQ2Rgy(6kmJk09
z;$2dB1_PoKK!eL)#fZNo@58WD;6hQ*myB?t#NdH;6~NKkOr!lYp8r%T@5`UUil*mn
zkxZzyfbqIm94~<hNiQCQQ=~5~I3EWSgOaP|(Lu(?$T*uS{l4-o#CLoBnDt(~DCdL_
z@PKrw9;l1lq1KNx4&2<;sMhpb&S&X6?%k}7;-;VeP(=lG-gYYzkkE-darskJNB+IU
z0@S8H(poNC_7&`qS3=}pjm&`MI5jz3sulSck=6Dt4(3Rq+@W{T6omji+r4I+G&;uy
zqA{e~MVTm^o}3)V`Rlw~r^+oOH8oYbEeTKV--c8?7(tM->U(a;8=3|N(~=o?1)y9>
zO;&lTwyZ2=Sy_EJSL>D&0OGEoU^1ZdO}ociz}Pk7*{mLCU<drN>u4=(-sv5`2mZK`
zF0UCJp3=}kId4az3|4^b7Tx8<A6)HA>J}zRR9Nz0|CdqCzFVT`a~t@vn3lP_4SEi9
zm;o3N%^HVRUWpvXrcQ7hudRd*shdNPmqOr!mk`I-k;lStjO{y|HQ(pThqF?01yM<l
z;i+rG?yo~GG!q^=_M;FDQoxDx3H#uu$$`rGKl#*M@)KaA#>LNQjapTtZ}IIka6#%%
zHJIOvr6_7hEd;4G3L2C^H)#b;LMu_(C8>Cf&QSslUPA}$*Gmk&ZS3Y+vqG_?8T{$P
zlatxxQp(FeQLH*c0}`c>1jeUOn=vC9i{l6R9_k>6d5&D!Sn46txiX-VXT?~7CnWM&
zj-Xtr6ltM`AV`+{9CalPKhl5jt=}GS_77_3_OCzn(~*>#@oYs481p>r)DAa86ZgD3
zcSa24^kQexGrQo!DQMYPfVO#!*#_1DcP_aZY6UqtIV+v(8nymN#ijX_ecZ=7dkZ%9
zl5DE!)SakbuL6K1eI)zvqHJ~7@fQ<Ysz8$2JlOpl-be&U`uWmLA${GyG_X1#V1RCM
z|EX`b4nn$V(j!e2q*F(<*fgBXEt1Sazic(RB(cKgkjotw!;aInMzU+N>eP9lc2g!}
zE1OqI<_KblC;5!(AQo0*%DwgsBfC+qXqlELdK&3=On2Q>J@?_?6jcG__L&e13U;fB
z&Rwn}=ScvFppi5m_q*6J^3>aX$fDZutvJJ(0DQFh(HZ4ELl%RmrTMt5p~EBT1*Diu
zGFWsuZ7`p8a0{@f;gsvEy>M$OL5fBe*zR>Hzr#!Emz*O1#O1w41mO%md%9C9s%+I!
ztBt~a$$`9C%uS6=NEgJS^5P1=G-^Q7<%+(izlu`0r{o*TJ=kqlvq5hH{olrHlkXw!
z=z2YpnFd#0)to|tN?NY?&I7>$;4qq-C$CzfpICuT_bZtUV3Tud%%Ja4?Uj6lhHt3a
z`qLP1(AMq&^X<u;(5-%^SK?gEZQ|ShhAZyqDHO<{a=tR21uD6h5CG{Wa&$;;1pg24
zfa+)Fr*M=UAks-LI$~f}sEQW`Tf7?ay`tEdMua1`dCI~9n{b3nI?kR4@bJe5w7qbl
z?cRQyG{e^P%Ok+$eS+SVg`{tiM3ni2^<h!fFQ(Tw;NT>e=YL#>)ORUaR>SM_>OI+*
zp+T)!rJ-RVukRL+L<;$57|l&lm|$aJDSIn4sDX_*&6=m*r|;l1d&;J-T9?@{uJX%D
zs5klbKd4BsfaGud7c>r$+xoBKo?|!nrqU1cTJUGYb0|s8#|YQq5)v<>24TbJCaSjp
zs|jyK(tA86T4CLJyBqr70sSipYKI3YE9|%0-p6T(>ne7FP#&o_3FM-o)_dC$R+>h)
z`sejuc%_ujWP5k4llIt9r6OMvmxzEm68*SPB}FTm5;QUr&LN{EZ~N1_aic`bLc6oG
zmD~)aWWDlU&$kNeVpOPmgeO$b@w^{PAF{|S#+$_a0elf78BiE;xDMipXt0p9ZL3k?
zPDnlxP;wY@R2*uutpeu4md6j@1(AedUp_Gdg}itju#GPeeaJNN9GDW@$b*DM#GUYd
zsa04%zW}v<2822-`XxpNxkDLFmAVE0hrpxUO^55>=$9jw-0Oy%j9_uUUy_PLX1`mx
zHidTY4%S(Um5c_R2j%UQ1_>z<eR1)t$k#V+O+7o?y5Isx-7)60dAS<W<z%mCcwJcg
zc+3{T?vk2LLrQP_>b^?wl)Lwz6f32ztzjQ5^R8t@M-<O0<pdLmR`%*0*<f$S`0*RJ
zU~%^y@z{NaeW)w|T#Lyy1^(5|MP)%?y{jtf;igJgsqZZLQpf66{1(`5nqAx1r>>NA
zFSG)`xPg+jhd9X1(rg6HG%+<cWVpUh>#)^@Eq9oxY|gBU7H0_!FRPw614l~JdPwl=
zCL$6keieskD%^AA(5$H5^FK=VoAbA{IBoYd4BJWl2RvdNe^AtMPHUE4|Lqwl;M6cZ
zcj9+P8o}WXuk+vTn6gT|us%?(8!5gld6>wXz6o=v{dy^3)9u_lrWk%@p(o@Db<8T|
z)&Xb`EN#1-3DL4FTyPoeq}X@tQFj&rn%Pt2`e^&6N18$rKj7`l2(=Znnr-*9eQ96X
z%Wg$C8f-#+&WF{i|BZ9aMmlS(BK3_`enSjaNjss%CGjTYfJpmf$RNi_B<k<-qU_`>
zr&|v`TjyCFnKhqUB}N@F<IHIv){-uvg{LmhC3i3%P>a~{8^ZARAF_l4XES8<61m>F
z_0y)n=ly`AMsii#dcG0N9B1t{G6T4Z4^&prsE7q2or`vyr4dRTx$j^JMd8<4v7^$c
zqxa<E2W<O~i%)Jp$J1&#=Ur@zhMf`i{jymcSGbW#zx`H!p*O4JO46Td@Ha#*2MOoh
zcAVAa81Icn*fb^NK(v?})ItJQilcg8tvj{1c*Ym|{;GEU6#gWX_6ioc%$+M{E%X6M
zBboF|a!6AMfMyp}k0ASRRJ@>`yBtOf`W3TBKaa7-Do^Zp8g0~WF~4h#nb^swpt$Ig
z@UNZBQ^zV2c(DQ*cP3*xQXD|Hc3byxdhFeJ9>T2b!Jfr^S;v3|q%a;R&|<NZ;a{*m
z(7EL+SuS1zOBbyoqF4l}RH5@C(zt^`xx#u&CbsO8$u4geRzs<M75$3Fh7Wuie?Mv}
z0+S^T(@IvCr|4=eFX0YFh0ik=`XMKcL@RQ@{7m>stzH6~hzkxXf8Cnd`$Bz7)}a5k
zDKL!!0c5@)FNr*U5M10yt(S_@E1_lleU(>`br3an9gk)8(!;Pg`{KuOmZ}pF5%Q4J
z5Q4);T#g6REG8Y$OmzpJASlzF=nNxB$P$u<aFhTa0|{`Pbc9XLy<WyV0oMf)%8uT3
z)$AA8vD0yRe#JfuIe@M!45Rx!mel^T#aeE!S>H#w3n|uQ^tgt9@F9Yqm)aK$eU}$M
z&~RA={5N>v=`b#_@j<Hv|9Z(t7~45Q`uT=-!%GH(KRTq00G{IGMFJyb1GZqV8^f36
zx>X`kd&?rl$K?-hjFL-T0VFNmT-#3n<0Vt`rjPeq5G>C(h)-bGKc=!rFonJJ%H7nS
zZi_vqGZcS%$)E>@AB74T$BV)u#*PR`IA<C$eQV}bHBao9Jh_!yN&s?&Tn5LfV4WjQ
z0caD+aOA9p9W6A78O{;#kpC?iNO}ESUuQh4oGC|c>ZtJqqF=TwFF{0p93k>}`FYqw
zu`iNs0|DHlppvTBRTd*G_X+19qeIkVfsuggAJRTTHmhw}$)o&)^o>v`Ng03@fivsJ
zH8Um!_CPp786g_ns_-dim}<5m23sUf#p1CwjT0``RC*ty{64(;X>!3W$grm5$I=!?
zmVvK-pW-8&Ty%wxqL`NAQ7yD2tXoJQac^l`dFc}oCFOBBRjoHwL(gRvaCG`tM_s@c
zWn#kj3Hp|l`=txpe%%TV)biCc98+kK8&PN%<EQD-r0uYO8x6gQhvMm`JjpN%@Vg=C
z`Smrc*8kdUG5)3@Pv5a~>@G<wex~4QB2y=lHM>;S0b`G6gpR<Rw^q8e29L#^sn?~P
z&>2SvD~5+%V4>Q0?}0`R`uz$D?P5pXt4-dAO_3BA@APT)o`655dfidpZvZ+?_OI9V
zQJcJO-_Di6l(1zuAcu6>iKd4!mOAq`Do0-MmHG<ly;Vzl*6*X^2QCAz6|6@N8&Bh4
zNb1zRtz+BVV<CoD2E&Q!4XZYy1$x)sy?fu^h?}Uy)sL!Pf+LxYKE5&^vfQy)YkF6j
zia(5F-`%eY>i)-x4$HGJ%UH{X+@iJm{mDzn$oCpvCEa~N#N<mynT%Jgp(J3tsDmUO
zi_+tCc1CX<V*K@YwfU_!-_DIQSeHQQt6Rrv{?u>!;usda5W*~pSx)Yp%3eVip_H=B
zmeVjk+WJV!U^;YW<&)Nm;l{Xx3QY1^dR>m_KD4|yZ$3E|mkr;KcJ}*dvhe-sETf|D
z<5u=lpsmy5puw#(1seOJ4nHo^2YBQKGz=QRQ5aYo$UJ8cYs-ckN2M>S;4Bi=V~F`g
z=P{`CN>2Dp;sF~{NZBygTou_ktS(CM`v`=`-<UKijouIOK3Y+d3No+^Qt4>wqj_rh
z^;}XqV)AFm$lFbe9HjdKUiT|>RqMuWY%(UV!pRH6`JF4iq~rN1B7gF`*el{v65!pV
zdZn+?ESzOX5gP<-d;il}hV!H6il5FhH2o6d9~0?mt}J@yg-jMi-MRnV(RuyV-fyO@
zD0g#b5@prppTIb<N$ta4L5|nqQ7@DUfO}`<1aQ|{D%X*43tVja#1)M$lHX<Bjhk>R
zE>QXnNG9#o{|EzdC@YTHH^g+JAkL|k*5L>2<_Z7)TXSRFmSf%=@KKXk^jBjd*4$I$
z4V6E&zj}TO#W>cN$3+Pm)Pq*5syUVuUMv)U9fjXmWXgtAjgpsPx-I0uHV^)%Eme6e
z=n$~0&}ahg3B1sq28)@U^CZyVWgHQmZG{ls1yq^v{qgC0J9_PTSj08ai0K+R;sthm
zpx8bRhf+~c1KbEZW;ls_W_iW)JTylT8|Gu8WnNjM{1?=S&~rxF#6=wA?yv;tZ;*tS
zGVfY=1X0`q3WJQ_9hq%0>T&0nEOuns*Kx(-aWTI16|qqyKVnG(VPK?6f0w`X7N@UJ
z^se^}q#dScS}!EPuEat52}Icq*Fh793E<mmrOaa|`^HntDE)}yVq);+DdiEJ+fHPM
zd-S@$0^PBtOI6kOy?bAvOo34B>YWT3=SowRbk9_c5SKt+rlI>j%0PD<&kR&iVynGL
z8GYJc%ox~&{zYmZ;}$L~$%@rVxr;#%k>sG^p*pNEP6<-qJ@9fTJn1MUma(o$<EoL>
za+<8<ne$+?;TI@7->Dtms3Xrv3Y5v=>=0G`*;X3A2GzV_r!Q}2RG2#pW+CJw90Eqz
z8csgaoF0<0MxYP7!lC0dKP$Cz6}%)@;FK|5_g`}^u0otL){pAzP*XbPk&dz={jq~f
zOkFc<XJKbxt3Lw&Z^<xXUx<1ldDAQ|3sSjNwyw^%IY}1NE3^Ls9q#Qv9y6`lvaQ|?
zmkJ32xMD-Jxzc0P;z|G*VEQtO^ZkACn8<)uNj7x<b!_hBA8@u>26f^pM-_cs&Jm&Z
zlJ{=sz)iVe@IC512YS=*PeR)-Z~7Wy(x7ad`{^2S-YLdaW(DQs=oYC@*E|zG)p@Y)
zTd;bNYTuHR&#aHx9k|6^do*9c!)d?hJaXWIqMmJ~xkKegW_=kVofu-~8X+o+R%$aV
z_C4_BwOT}4b$p7E&lf7*I%-uIHu?{VL5dclex|{za8)IrdRG}^nXQ|MtIYXOp4du1
z_nXt-quvb&D3y+y=TrY9`~@3&G&yYSr8lXtW$zvJeY<B0a+}cQ9PHF?>0hBJ;;cp-
zbQ#1<?7JYSR|JYIM{a&g&(%hhQf7C}$l<mxs|7^nqZ`>8qYtIFb<D!d$ptgVun$dP
zrq;)cix_Bx;4{i(ipHx_5I`XpKB_b<lEa&+=|)Tb{7S0OHAr`S+_^z;eGz-S;iA5g
zt>S^)y<`*ukOAeBt_nBGr+8UR=pY5|Mfu$hN7$dqkflEDTe`e`8giX~*(`}BDa)+v
zt)^eI$bUcZBxNtMP}id5PIBBOowI3Af4xqqy~tTA)FHYtUgnmMXFzt--kV^KQUY2J
z{P+yUUj~{`&*A2rz*GdNSuTm!X{qn#-EiRv$tq$4vDZ)kk|q*ywCoFWTKTQan@X7a
z*7dt=$;g9W@YRzsNmTipzB@zqT&y#a<Lfi&sI=YoJy7V|Yvv~o3V7g6>P9tUfqGVT
z`I{V*y$HKQKF0mDlF1V*nJhNQqkpYrj{ajM)1U_~Tg`gM?Wf)_Y?YR+u5^FV#@D8W
z9G22VI!_>f+smEABZN{;c-@Rq76rQxbE0li@3m?1ieR*O&DDpPxC~BDbZh4Eg!TCi
zaZAA>X#*;AJ}uIph|UpWs6)x4=*t-8duUYl9@oM4F3WjqudI0ln5)#cqo;~0fi$_@
zj3e;iNwoz}PfY#{p^)(I*l{tNBk6(SFN5D+lOJd>Yw;&X^22if<IF2sd23@+B2EuX
z9=GU5R*KCT&ES3?zamnz^fm$-(x6fBP5>!|hch(92>lOWX3$9`%9O910={dOJrs3>
z_U#^L;zqfxC7@AkmCjBO@)Zz0QEe|_C)VsYBfsy){cE&ve3*JuDTgDIXB8FRaZ3}Q
z+tA1_X-SQF!&(_@XPg=y=P_u*%WL*sS9aWS$M)Fo_HgCN%PN}MsMD{RqM~#K_mlX4
zTHkY%e;Nf$dfP3!1^L{~m&JQ-ejYveHzk#wI3S)U-_Q?tTsArFD|oj}y_L+#sRl02
zq%XqQYgmrT?HoCDw(E097m}{*hNiraxoHs{cWHTG_@S$G6o4#_2x-esjsDKt*qT$^
zYCvLFr&+cR?2!E4#=ByO$s^bPzC?5(8RtfBtKexT4tnXHq?+;2+@qS}2R3pu78Pl$
z2YKRclKAphwQnw;*eP3U#UUu*BU1nLJ3Ib@r%}GvWM~joqjp~=Ok=Imh|?~62vtea
z36-Uf#&Jq=bU3eXJN)y&KBpDH-wl^{ESgUeFYSCE0-mx4EW=-d9Vfr;T<x5Aa(3G<
zj~)$(rGYeg+0A3I8J3Ms9+`SpyM<S%(Z*SLiXGeL6GqpZ^8e)$-@j_U2e4ZA>|U4k
ztB@$@mI!yBylfm+GMmIoCL&N01@N?Zc+D(-q<lTsI!LCZ)jvedr{`h90lmrnk0tB8
zA}ihQQgq9=x5*pdRfj#wfvhX&v3#>~OLF`r>+hU*(KyMp`W9ty#5BccU~@GS!}5qr
zzFDJ69(dR76s!(7PkwxWoHAqJ?s*wzosHl)l$T!|r?|bFLm`XSmc};sg9T%A8}eYX
z_67RFcho>EyaXq3uD64hfg#NrBB+*Twe~N{3GFYi{R!!Ve4>J|Y8km`@(Oov?`9Xz
zWX{x$Z$egdszE@@8Vjob0~_@_xVu2Z7VD0gumtuY|6rsTtMB#J9f~I~Gve>>ow3-n
zw$-Xjb9=k*3vAPC<(3r66*{dPr^KQ46Y15PMRk<Z0Amx;QF9=E0~mSZ#h22vqk69i
zi!15sT1zu5vbB74QL=-OB3gtMKUQetG`@Y|i8a*2jXJ{sk~7aLOxRv?NxyDKo5zUm
zwm(cb(!QPVwHl)yJ&p%E3kCGF(@{vF<fje%-HQ?)opv%Sw$v>7BHg_F*RZ06D5kq!
zj3pe!2Fii52yfJ`7jILO?cX4Wf@N%t|ICaO!jm`@NOqf*O99X7Pm|b8lUw^Q48wd{
zOudBMqi2g`lWtbA>&zcw+;ip9O+<TfX#O7V%Bkd4J2NFbVuI?Dl+Rj6J&$l-v#-qP
z|4a1N?wi$X8w8saSYtkELMzjOfBgp@S%0oKy8CN;r2fjB-?dJ;8Aim7h`fK%;`D<l
zufKfpbAnP8%i&^AH*zErw(9^(l4u7420gvfoO!u-_w~xP%|&TJk%!LQl~Aa}ggHp;
zDEo|Y8>Tk+X7?#G)237m{CGg`SzWu!*~U+IHrnH`>uQ#PQOdb(L-~@VfR1fmECG#5
zgLGA^L5p9iWbfQ>Xb-!Qbr%A&_(JFgdAb+-rTY!FQ8VlCEqnB3BA_<`F+`_H9wLGo
zdX=2Bo1J^|XL+X4t^KdQYmC@j5Z^@V)R;QYS|xts8QE0N{8xji(*?PjhtzsQ69@j;
zGd1kmZL5lXzUxmuZDo4W(0$;=n4b{t|MbMdY>NKli76%y(bo$%X;wD>IDCx2OYA}6
zj6EFZiAs-mN8oiORml0$!{I%f#q|A&b5f=*XhAt|?3qh_U3w+-FWc6!zs{_ud81R_
zEGHys`kyZpmZ*KNDdQ_c9D0|mInlJ_?wA9262q0nyq7QvB>#j--C1f8Mz(#ABOL;~
zbN-9P47Zk4(@F#|AT0```4^KM4K-c!LGzm4la|yNi#dh+VfDiJ?Sn13RxMRcTh)LI
zJcX>gVJ4rAYVhUIKkK;e%V?1o0VpCqqU<cLw~;v_HZuBtVk0voWkgdkQofzr@_<+o
zow?DthiVEbmfE#5b59hn+Su&eS||NWhuUqT1Fx84PiQ)zyJR&;Q4_d@kRNZzs<K(a
zRI#~|E4~d35o>a4X|`D=>j*`D%?B5=g;-y35cC*-q~)#w=g|gffAa5{I@$lkAZmjf
zlK*`Ec0U0A@)Xug#NNG*$u65V4L?BePXM@w^no*gH3>BtRH72w&?7D{E%jb<vR?e2
zWe)ld7j`zAPL{fC{e+dy`&emHE+BVOcDJk3X|DDA6c?kCThH7(dvpzKSKGe3HA&Fc
zKLct8_Sr7^<ulRyVb~U4eW$=qPvok}0nhpAk$qZib29&yf-_2{IHSZw;f%8BUuP6Q
zi%u5eLZq0wsKPLBNW*T=VU0UJ3%XX;cZo)L(ddJ-UB~n?#`;rw6}fCE$+J)b>|{b*
zFPDjiD%T;Y@?H+#*&6!TH>XbQ(xrEjHAa63lG1TtuIR;A$JAc@b}QlnM0jr(XC9ur
z%j&SOEZtco^nG3R_d4-MV*2iRKcHj(&}qjFN<L;1S)0eJ1r46?5f8E+@keL6YklBw
zG6aVvnYA0YqG}+=B<Ew{0{P>}&wBo2tj$;CxF<2w*5KkAYQWZmo*Wsc)&E>&?Ov7T
zOJ9qkC98qnFqxrA?!4q8>Le^IPrrFz3XZh6NFJt~t)V)E=3VpMH1CFv_oR8Z<v$h?
zBVxCiXuq`@OM~DGz38NVJmEX4B)Ki0)#3@Vd*R~Q&xc3ahI|Kfma994=)v$}y(6kT
zQiwO_@z>T!|90R87R!6jg<$dsV(J>FXH(m&d+OxDBqAPz&|WfFE~KDGyA5J>9UCa6
ze4F1AAorI19c{h(^Qw_Zcf`&&>nkjO;zu}<^rS(;AEaFU<EC(omm@c}!$XLcVpN$}
zZ*o+tp-_eyzcg?7{YzxlK^^*OUaWzk(5s!1*Zo-&#xJ+cmQlCbot?eM`i=k4XE8)$
zXKb`0`i1Q;R^P%=+STmfN?xPTQEAHvq*P|&vxdAaWl=S0J4ZV4?L+W9S^sY*e&VR|
z;Hz3Dv_77X3Q$V|<hP#o(&u)rq-sXlxZ@cSrahMW=FGlCL_pT^Tc8J#79w9eNPFe7
z31vBV)*zR$xEQ^!XvdV0$>T&?UA^wKNPSZJfz}YEtL$R%!R;2hRep*4sY1mX)Y9KP
zKxc#49tiro85Edm8nBEEFsU|`kuE(-65UR{UDRzky}oZ2Jfi@?UTxiTN<8+5B4iHe
zpB6`TiHUwKvvaAQ#QHmmr^k(9u32YtajX@8#`VV1y-e)*&V%>3KpVBb{Y|4=T33PI
zvpUN+PkXqO{VkTI+2@XUD|u`=QKIV3pqh5sj=4H~(##N>`mRANw!5YL>RsmVG_tGV
zKl-)UHJQ2F@Vbch4JE3D`VD~i)cn^zaaO&3Vx#<D4TF$e6rybKc<@qONc6kSAnL>(
zzhYtbeNlNq3MMj?=e)))=5)51JvjaC{BEt*sL#o98E|l9L2fx@@o2Yi-;iMLs{Z{n
zJ5q85;bxyjLW3cP*jC`*ZbR1PiQv4M$9_{cUFms!hi*_M4h3yjy+)HI0xA?YP|N~x
zc`*de#Z=K|;UXhk<Lq~{T6M&GrXZ>3EC^s2A1{o>#oz%*3UHL7C-jlXmw3u>nX7oy
zht50+KI~>#4Luftu{llFYG!Mz@tWlTUkP<1E;1A#h$d*5@4*y#sEUgW9l;|v8j1VG
zmH*@-!(RX(iz);iX*org!gqhr15!NefmhMq@YLo%*!_OZok<Zxln{nB@%zdfOMbir
zMO)FQ&O;)=e$uEG(#JBe(-%p{ydL$zf*`r^P1qaNOHr+Cw%XE(9i%DVu_-+=@#T8(
zZ*t<hrQXHNZhiWXti#twF2YD=SHOWAQFZTMa^UQDZDCS6G)BZ8t$*)u%{3p*Q|O8#
z$j`Yx$f(3+dvIAO_};fi4|+)6#IAud8ONK97(yl{gNbJSkeq(y*-2OwQ?FfgDOrC%
zY#>FYrq^%I_Wbx<WTt%6VmlvB8}6cCxu1lQQ%Xly-i|&!MaoQZT4-7ZUyghy336B8
z!@(}F)^iB*z7<vA)ZMSI+)qo39JufgDl*9R)nC*i<uhN8{pZ5g0l`=Cu^v$IN}SRv
zgNyrW3L_J@*Za$Hr^IDJ?L4Eld27!<!2z12lto&T70E01|9F{GGAoksllJ(djBxr(
ztD1UC#~<<T?R&biBJv)F3=Q?qC|W{Y^ta|K8d`Kv@|lR)L)ErDeJE^s{4%|FdXL;L
ztlj{}PfN+<8u=dRI~G1V?$cez<u4|f{33Qh<xzw#v8ka9`d3`Njm-NOYf*8n-`wk{
zyfMSmas+CD7upXiw_BI0F|B8Pbf1}WJEYajN#()nA(qxjyYZ=aw`gFw8;D|2cQ>&n
ztAXFHXT|Q|b7$5S>Fza-4IyZ`A@Purl_m=WKGH%>dnv5pK1wRFJ`<1H&p}aOlug@$
zAxKN=w`A(FaVR*zuYKb2V!8mZG2@c&gh<hzJU{iB8mB3}VrPhK4PNc9;hmyaj6;D*
zXnV$%jb*$Pn?k&|sPrHrfXC1!36$|w5ou%@-Q%&5A+%pMz6P^{4nXYLNN&{lZirA>
z2DDO2N{7jVw%O++!7e8`Q$j4W5dYUnArf3FlOo{jYKgPbcEqBUnlh$WfgSqgAsep5
z{7kGG+zh7`8)&tVc8XJnJIEkM*6z_xbr-W;N3CAII`~yx?YG!(Lx|P<<!(3rs#*%|
zZPEUNjSqQr4uwb^RddZm^;}2^LsXa5FJI(v>Y$H1>^t0VA*b2p+F2?*W9gCd>%ZBv
zF9C=Iq+k(G@-+;CD+72Sj;OQ2TpXgRFAlLFYBeZRXsr6GQ$<f`X>r+y(*Ldb9wjMK
z*p2JjLf@yL?NYof=lxS+vKy}5Vi*Gpf~U6*%i5attQhJ%Xw^04kKDEtksjQqDexM{
z))vb+vDW(ZJZSKNo+?VZY2BO;Pu6KUdHFzFG0H}Rzo*i%dqL+R_#K<v#zdxB)amc7
z)4TPkp4`mmHmb1kv*%9MU-I{g?tagTbKMM=8D(GztzEqDbkjMxZY!Y4n_f?9s4L3p
zx*rL3d#bZ1V3)x|(H%T^aLGLLeus%IYnVwLus3nrP)HgP{BoAw4f^Bk!wckd-vnW(
zl{N|JJByYecE%|EM{qy{w&)ps#=mYlCsyaR^fL9_d`quPE}@e4NC5h%^4Z{3=^fqa
zqc%=c9@IQ{0ouJ|_boIJF)1)$`*2QS{W+=Jt>H+-BvzZB72DaTTKEi@J!qv00iwbT
zHkk6{;F6aV5;<+ENW59Fi`KUiX1Jnw4>Xk^3#0H{&;;WaM(CS59HcJ|!Kxv9%D-ZQ
zML>ssgujNn-b1Z@;b<t>J$C!h-1I4W%AF_!!TQsz(O!G)NtysXOLD5H@fv;2a^Cj)
z>Ath}ZK!}<KI|qvTTDy6q|jImGDR9#=+9zo(`S>x)Rt#5twH?pub&##tQ!%RC@Ow)
zj(URxdP~6`Qd!h4Y@D&>I<4zQ>MTPD!}+^M_@O*kd~u*I=`Zw(G{{Jr5U&-09n=j5
zyML~JgHZ4${d3|1YuQK5@plrk?eEPV!t7#?Gtd2}i3|b|(GRe*x|Yy^^Ne)`RWY$?
z;Q`CAO5M{12kJANazA6Rz<__SK$A_w51QU;g5yVUpOoqIRptl5!~)8%5!6wcR}<86
zz1f3)6&;u~v<Opls!s9lW2zsNtIQ~w0OjssG(d(TE{|ktT-{RO3CFsvekf}11!x(h
zvyjJil~@A=btN}ybY<WVp;!uD$pb<#z@wGuz}bQG#COW;Pq-B-ERFx3F#7TyjK_c$
zyO>N&VM@lc0O_Su^7zD0|2T3qYiUvcLKiUAgi0!wFeW@8QpTdI<H!K{k_A9sj{ddc
zjz)=mMehwB#NlWbE-J}@^Nj8+W}W>=v0WP)TDFy1k(hE<VIy-Au9;LW5|bWkfZ|ET
z_cFP2DfdC1D!L?M!xH#f?tS2djCMnT>qxuKP^ut`T%&{ipYjP@?{qGnYe8kapNgd<
zE2JE!%zZqzTp!gn4~Wob^ddf*vo0H2*a?jj12ATQb6E0X&kgI!PdtP;hm%~67iP-6
zM<E?&jwDT{#iW+XgxJ56u#4_2ZFoB_m8k-*=uEhx2i!)%s72jmN!M}QWc1>zWU-c6
zh7J)t#tK4earHkb!3dX05$Y%5h`zux3wMqTY+LnComxo+7wG$F*N44n|AxKiS^>e7
z{i=dz99r45_dEOQ(1f=;E8GhrE)-iHe-p>mjWRx>m8M=>>NCwHm)GZI-S~^s$ekC-
z+QQ#=-p5#jEh9sf{NuyATBar09pxwWca7L>Ro7w&-Yo;H)o?4(NsQ_pyqSOS`HLxC
z-JfpUE2$WXnoisFkon@@<ivFPx759Q9G#ITCNiu&z+CiHDzzO}{rP(^s;>bdwd((R
zql~y$?A9Uq-o1F1e)Lp4_o{e2;p0s#_XB?L+gg>;i<96L=gK7evOBG3J{Nb)Q#)uz
z3%Am~5i+7MSShn^112}4&=kz4!9%G<eXM=&H4Ua;{D!O(Jz{pz9aC{-4J$=DA9Xs|
z_QXVn#B(hE$?<PE7ykZ?iOeQl4?Xv{bvOSteW9bm$A_AZWQTsPlR>gzJ$w5y6}RXG
zh;4sr0p<(u{A|d(Bhg?3L<Hm3%Vn<YgqJOPd@UzbGX0hXPFxq&<v)M;_y*}JjTH}Z
zm$`0nG0}#@&7BgZ>W{gbebe9Cgg1HkP~JRq8n$~X-)M#M?`-{zzO@z;RO^naT^5T(
zf0Xe?6W^8!nKs|OOSd2Z^=h}2OG9nUM8&#SVjNVOAbTa|wnjHBWpt909?VSYMnjOI
zXy)bDe_^F){r&wf+0e{s_u&i-YIwJ3W<L6CW2xNzvyLeU)vT4a9(@&t?gn7@u2<YD
zp9!etkU2C+NnwAC3_JlC>ZpneH>O+vd-X5le@2u&r)w50K+zhYXJR`=zJQiu2-y7S
zsnSdi7+_N<*lE&w;h2eaP^`^(i_=s|$wi$O9pJkzQ@$%)a1`hBMOu9K!7b)uCGo!n
z;XH!>r4=Bi75w_&Ry+zV_0GlcEUem7G8IaGy^`pc9-jqN05KbVK@7N>bI-nw3d`;m
zHP36Dq77A%Vzj9d1>z9EV8a?wQBycM+|{tlPzytq-fuh5{%}hSY&qs!TwD$>xWVAC
z(FEiPv_{w|26Q$kVYZx2?m8xw*n^WD47;?vFBVO8P8^=niK?t5jo!*ZzbOpu_JDuV
zpJV?=$Z<#$BDMX%PAx|g;a;1GuxWhr2~?&@IARuHNMqweUIifRJt5i=J94Z5Mn9f6
zv?VPq>*|Wh=P2FSY!)RNIgo=s?OX?~6%V_L)9-{4zcdy1l4@v$ft3EjLJ!!lSlUow
z==@9|Mp){lBV<mThRw{Bf_3XZFEq4#IhQMsuF{>E;A=1zUkm6}UK>H)6$Sz-%~$Xe
z7n909lx|N@lF?|4(9xTG-k~+_Fb8dI4!L46<0l=|H@u}SnS7#|k*`?ASiOeV0@y5>
zX(5nX+RMi${;GY5=Ck0b8vd`jN~8`Gj;;SiVSj~Ki9_=!_P<4z&|hEQzHN(OF|8Tu
zDg}J-O|#XNFly5>=@y-&ao9Dkrj_5e`u~pBkjK}(G?&E8c4l9??QFn6y-ObjaztGq
z;2jNYT;HS8C8$Aq$$SKqpd$Ouv<4ouC9R>W%i)Z3<K~^hp3hi!M;cY~AfshEoub?R
zD+|#^+yCWYdC6>eKXI?|ji%BACcpn_9N@f>p&~{Y50O4jCC#o@dkjR|HrDv4UfwV3
z<xSj~xGrUwC?BL@OOrKavyWs#M8XM{m$`v`as#Vn7=-Xut5x$te^XeuI%RQ#$q?Zr
zlI_Oz{rTpHf;hu=?Ro}mi)zN-$t<G4oBH(EcL;>-SWr>^jeW%tr;FY7u8o(Hc96=O
zx0hSAzpi#F|Er~5D>01vYb<9rX{`3MXF7OzH0lf9)LzBV)-OO`s|~;8$*Z=~PRUzA
zElcdIqm<;!4nq?Fm1!>Ht3(O*s7xVNo>c3O7+3Go_c}ZpKyidO-S5t?n9po0c+{K+
zW-V?xRx#)$q%$w_MfsP$_6iEVPuL75hh3$-2Lw%Q(<0bk9)ntW#24`^5n5u!&n3Ze
zfyT1+H_^><LH@dGU66V!tTU-#E*GOJhWUWN(FkA9!d_^9MT*C}&21g`H<aL5aDXB^
zJcZ4Jn{ua{#Hy)b|Nb0MFGvgX0vh|)FT+a>H<eQ}I0ev26eFL5wlI!;NT&B+<BxhF
zgm%CC>qvR=$y2d0q^PUB{3*l+%CxDeWwge9*WMYQ%j2CakJpQq1zG?3B}=S%4mQ)+
zzZ}j%e^OGYU;cSp8~f5ru3%-8v&0J}Fpz&j^-cL0G0H^g6#J15z*OorwJX;72Ecn{
z_H&yQk5K`%z9Y?eL(7`_+!irkT3V3n34#gxXH<7uOl@`n;$0DkPITt3dj=?(f5hF;
z%I&!OT=&hx=S#*xj&({}c3NMADP@Q3@FaMI&4u&o`L^Z$3MSA_Tz~Dq188!<D75^a
zu$%ke&0YW5R8VEClm&yN+qptrMPc3TFxIz1a2UZsnx`+`Q^_Dvt8crrw-uagbq|O(
z{m|&B9=_r#mBs2`H!GSjsi3Ws-o2{r#A(QX$C<EDhQ`)=EtY?MxUcp+eA<>D11S8+
z>TN?bR_@I1Q0TIlBJ9|i23U0kND!*qdTnrv;A6&Zd>otq-E~hBy)Ns2Kk`dIi;Odd
z?(-*W>b3GcvTl8xPWPs}tB!8zq&lttFsmr%^;T{p+-45QTi>(0=KB2Jdakb<D4uP0
zcEEr~-z;t49rt{E@>I#GlK96pCf*bFF{9zq@pqBNMJHFKH_|iz=K%CvP)iE1n5=(J
z5-0Qo>rULNUZU{Fz);Tx$j;}%y>D0L%E#->=gRjhTZb=@8IN=tJv<R{X#M*2q}9tg
zY^0fqcqqenDUCRx-z81%wS3QnhlUCYk0&eN6KdyKZD|7qySC7jm&xt#{E|z?Uq=g#
z&JMcFFLCzGc`=Hvjl%SO>H{iscpGqfehX#O#NLvE#mT$BC>;4s##s%R7GM#b#zs}J
z@5nGoc2w*8LRN}JXzW$E2{DA)wJd`@$4tEA`}cP_GMpx3aor!yb(xtZjP5mYS-yh8
zqlTNe@YcFpx1axvo0ix{J&5dEtlsaOLzq#^%<$cFe{gksNpU%uY)%^SN;z#~o-JxS
zKuKX}^V~)9R>cuRV9K3xr2XwqSID@~u0vK)<A=BCzxV6s&!5G$*fTRH5oT#ut$pG4
z=RBN&J_A^MEu~Uwec$>1Wo^f&I-y?!Erd|l8S~I(<Uvw{p?_-x%4!n}I5e%gJ7l0O
zEO_EpoIXBsHfhJEL9jV;$UXW<Pil{4i_=F+rGXP$`mb5el(TL38V~Fgcu0Nq&yNhv
z`x+fQoTKzyO>QHF{Y&sgGka_2hQmnNW3h~ndAwOkb$5IX<;zBagTN(wCV9h34*Med
zi(`w}I1A=V!72LYyaDg(AHn7Qn<m@zEU+(G>NwWuO<2qubmP_Rgs@3#(fwRQw0$)D
z#BO<u<6=0S)@grz<z0tsdDk7MX_6OlSs&VsY(<c8XE%rVLRPAuW5XAlJs~iM?TJ}&
ztb1oho>+v)n-6U=C1*9@V1B#bdYx1gbJdtP^eihrFA)7>#*FO)d2WH4_0KJ&(64tQ
zX2Pw;Eb~u$^gdn;=wthK3KW^~usy5!UF7l^b05q5{P{Cso>?<}bR3o6KqCmjH>(Q$
z4t-ubC%pvxId3%`G_MZ=htWOxHG1!ThFrS%xg0+PN?V{@zgJ@RpG%frlzHofuOB_e
zTr-4T7~OWF#s1^u_|IV&@y7&Q>)z?V-BGRi^ywsVd&7@Dfsy-ztq@}?lU?7wB=)y|
z^jyP7CDgdZ^nrnR{YH1t!Z)if4--7=kcGXE0~(|P^45VB>y=q9)8mg^P_~^gM;5tW
zCWqgR1gkYp4}KpM2l9@2BV4_mL$}Cd_x|H`)Hwlz#OCSQAUM8y1H>H(baYt17lSg2
z?bbaLIO-G?6*F_@|6D_@JvVf*I{EM34)%hMJ`s@NoSC{5JR+%Q%oiCV2d78((RGo3
zbRJ~FVf%>&Nupw-3w6-<f5=6S?~j;gY`@*xEe=P>;y89t%q!lOej;Zc<MNq))*f#Q
zhwUmB;z%p{Va&&g3*@JCk3KA}KND+h*&Aa)X?Mz_qN3P>haTR77`(!_+$jg*s1hEw
zyb&XDdBpmgcb&5NlO?=+ghahcGm&6TpFVw76_T?!uG4c0s=FkqhgKYoG3eB*dv{~*
z26(l5?nde-w6}c9JTdL4yb2F1g1ACm5rNNkib(e=I(s1k+O}hBhkj9Ti(B4MN)b;&
z<CGGvYdVbVTTA<A@}nc_e>B8OJCj9mTWN65*ww%j_VeuTA5Hi^X#?pgoF&c#q5lVa
zZvxD9|F@5-ZY^{xdx~V=ijqQ-R3enUBukW{L?UZwA(8A$icn-}k)^UliL#WCN)crV
zB_hf0Trb_f-}5^&|Cw{n%$YfJ&iOUZJo7wvS-zjo`~6z3*L7Vl`$6E0{2206OQTD>
zkH|X66Y?=o^2H`KG7{<Vot?15<zVpLQm?j0FsuTm(m)`?y_N8E#-6K8VIp#&jO=vQ
zfy9-(j~03`&itV&7!?%l?L~hBADBa!7!ZVnqy7LPVCX>F%uyL0yk2MX;J^woK*gl+
zc^`~(h}3W$arqQx>x{}r|IwY#7pA+1KWFHD;?Igu3U&DNS2$f@f|=-G;?H6YdJFh7
zdj~yBYMki~;m?1+$^V1TWb{4D>91cd9|(pPcK?GLcpXQ4L7AfFBcM}Mlwq?zIZYw6
zR2M|aZQyZvvF3R*Hu20Hi8NNs{COj{%dbHj0_Tw=UR&pX8x;S4_vIdZ4AkxClr66O
zdzRF(WoxXqncJcn*?mL_lP&SAHy8<Jtnf}0#1-W9X$0NmI69j5ERt+>DVyyj)nka%
z60?DYB*Ph^-W5!*6IRN$oHph7AfX-GaTpdI*#|J!D1i{9T?_5WFQ{~7_YH*ZMFPWY
zunWZ#S=4xek%^it!>n*vbsnEHxTf4393*9>&Drx`{{{*91qiZ0S%|fQQVYKbi5S7<
zBmYa3XAmNg+$@UG33l%&n(m~frH#D8u-Fs>Dq33;#TXhXo#MefJ!{$2`~zFwBy21F
zkkXM$0R*K;yo%2hFh!uCR7v66(kd)<UdJ-bh-lp-pL4D1`(V}+Q3KoWb5F?({ZL=R
zo*|&Vj>lnJk|}I(uMV-VgIJ-e<rgf34-_tg-o?5<*m(jPlb}Zu7zow4D7<Nf_*LtB
z3%0CWDF_4QAf~Ycwr2-)JHbzq>M;k{Uxdid;i(9kHv7Xtfpj}^0YfEY;IjpD)kom(
zuttAH#1}7~5;~a+nrJgdMMzkCZayQnBVuOq-!<S4m7k~l5HN=Q5b2F7CVD-t5@Qi@
z>{D~rPq1<^ANFBe<ry;wl0?cBH)1Xb$mTakqy#UrN3h`%%vkcJTf|~eVu9T-AQh#z
zHCC*AonW`dFTJI7+Hd*P{EL206|5N&+9*;ZrkE)cX5mJ43SYWr208nsNAvNTY#%&b
z6MXv!r_K#-E-vf?lP>l~i;C#ZW~gGbaB09c3{WDh)So39+2JEnJWuBju^Js6MZRLU
zWJ2i*^vL{5-S+yCQOOoaOWvVoO1!BbqS}g#$o7jC=Y95hI8oJ*(yensV`Nbyy_64A
z!Y0GgYD<;%^z`biW6Ywb#z{p5SKFehQs$=SB;L|tnafiL{Uo&4asBfmKM1I&-CW(N
zbz|Q$zQhd*u*%UI>PH2=Q_lyS8<4S$PE~?T2uh_%uo<>c+9K6-HiiZ@fgywMjfvQ`
zLzBqjnCkSSOHOM@Yx7enb%Uc{JM215jE#+Vd?%R|)jFdOCTUql{SLA67#M_-w*Tce
z)6TGTX*C}<$<HeLpYmI3UmlOpcwZQaVQ}P<laZe9-aG}ei2BPJtm`sO7En28M2$-^
z0a*+)Gc%|szkrQFi~QyE92W7r)WOk|k!V-ZTh7NFWHK=Gn`Ty?V14H6Ucvf6LR%4;
zyg`Ag$f8BB$RoM<E#!Z64fiOB%YItqGx8K@m(k6g5QqFG+gONm)00Cr=ll4sAr%rT
zbji#~nw*oVlLm9Z^IilJsrvo-*7vmP5_~mhUWf6@iOC(p+Gd*41z=|%etE@cLP14^
z7Ts9j96&|iiu*Qa(MHn1kVJ?)3efYMLODVZkK;>x4s7)p#z3sX{8Olwwq-`v51%A|
zl~SwE{3klb+)xcAK8pnf1uRr9i@_WGlilhnV|w|^;vgJs`?c3F3u(U<I){KxR1GM%
zKcGs&$QQOrSKXdp#XF`w8iGNzNNwu5&xE>_JB~C>o2lSLA%i#Eg3pRgv}?fh039JR
zeG+R^z7pev@f<I%d4LuxK3G)L&kWfn9K<bx6Z7mwgjPng|M}!Kv0&%)ve@EWgCmm%
zA`S+t5xFN2aA8`uykw#pGR+C}>$}wwL)-BIZ~gND71wlB{2Dkfw_)=oC+I+;#7Dti
zF%O%2qpq9QnT2ts<>B!=qvR4{g6r!rrlvzE`5`Psu>?H#H!%lLw6eW{tN=x{73(uo
z$8t`TVJr0-(P%=(r~_YC6y&yOh|!X=D117NDBcH`F;mPj5yxIZq79gie=iz=ew2%l
zd}-ZI);`MptQUzC@Tg4PiHZ6wBhehQz$szjQlH1?@n#h35Lz2jI->DL%`jQcQgMJN
z3taNZf;Sj$n(6B5PPpOYleL<p)TXrU|EmwZt^&$#FHoU~2Z7!k5~u-7e?Au|+oKQg
z+_aI~0ID|C%fOxp_XVSd&g?Xd+*`r4(B-k*lL8NbR;Vb*u93Py++OO|XEMe~w6Ay_
zjlig4O>TW+E(UN6PA!p^Fh~U}1-775P-&402>w^Y2$9m?N}|JDS1r8+j{#5gl_WL7
zXy6+bzk!(m3c^>|h9J<J@7-DzjZ=r>v~b&JSnKY6a)|IT4857wvpGPuOHQB2iBJhQ
zf@c6=9qYf^kVaMDe2a0lN)bl+oOV7qeJGv*;wiIi=uX5mf#|=WeO|_aLU6mXst82%
zhs`In&_&~$I6s~>kEkPHIlxDS2bz6Qx2EyHjaHXtdmz=aG1Ix8n3xy|2OctmE!<Yb
z_U@{{-h@(q%Pb<rBV0-9Q73`?v8iYBW*~fDe__52$6l@U?{tw+6<nAcg_=IEc^*;C
zH6uA%tJ6;1iPEulmf-2v5iE$z{*z;c6e-8nrF6v2u<Krt)DNq#H4on^Dk>ToFQI+$
zv=8p~;<ZnP#CCp;oiZ%R*SV{0DR`dO`kx$InJ%DiA0jV%+{xfbG7huo<Th+eDadwx
z1gDtBzVm4Qa(4wNrJ|$r0x+bi{)A$gp&>_ZA=^$6yM|j~(;M+1A|b&Cd(Dz<z8+Q?
z-|P4i0^be4rbjS*e|}|^<7?Wyc+<&71<aXr*RXCphn-%Y(Q{Z7_5)l%1d#q^lu^OF
z{L*#;iU1LJb#|uAKLo4^Sm(kc0_XZoUSO@RGs?;2`LGAuH3m)WYxi_1dsrNa3}17*
zShpge%aIpoN{W3{+xp*Y@M>vB67YAm<JEEM{hZtn;Y1*ZM_!Ld^CDtT`<dH(4ukfl
z^`RFy(I{3arFyR;eLw2gR2B!Y=lMYnu(Y&H&pdW>g7&~!EdjC5(duRL+5|8GMT@=G
zttz$JA@gV=VR|$nNNx(0q%w=1R%9}yhLr;sf+Jn7j$482r_n1(7)g-tm76D=C5+xD
zY&$8@$vIOkY<J2~3D|vGYe#brl?l%*nbOf1&G-UjO=m95R#>Bmr*@!Fs%Ca_Kk~w*
zV|1*`nBHGex;$0ZQ8kqM;g#3*{MCK|OHV}Z87ZkndXq~%&ue+PblZ(OwxiW$#1Ads
zUEdvPz>-o2OlmwLVM5q8cMS#L+P=1{V_=aWXvGzz<1WkFn6iCwJUCGj`vJwCfrI%I
z?o*iM*lr&@$3L+jm0-%066dLRfgka2Y38*q8MQ2_Gk5H}!2i??<%~q@rFtXODQ<TL
zkh(QLO7<*~8)(afdN`)>p7w!SRE&{5>s9f^#o(u$n=7I^rxqZp;K>|+-V++SzJ7jw
zge<6(A_{OVa=*2?-f?sNfewH{8cUDWXYIcfc&`z_hrU$_5QVCWMl!x2>Sg=BYq+;^
z?^*eFD1Co}Z_l=jxb~cj)_HYy>u20l3HYinDLRuog1(seu-(}l6Q1|_O6c~G631q;
z5I~@lJ+|D3#jgI0>5}^Hn<`<Zm!}NJ^L6B!4KAk%sF(ZBB>3%4$11{pNoeO5QU9Vl
zNlUR1b1<Zwq^)fLQxF!(`~^XOkJ(8hyW^`713<Y-X}F2yQWmQ$iV2X!28K%T!b$)f
zAdJ(>Xng8`b+iZ<0;4-@(+QgQ$YmDpKqHQrvOEjs8%N@}4SFa7ACeMCTdz<Ud4r=Y
zkL}>`Z)Ja-1+eIZDL>^;6ql%5fKQODj_NtQJE^-rz*nXrWy&*y0oU({zvQV5o#wKD
zZKUIZ%?BnoJ|GQHTE^?cl<B~pwWPem_^SDy<<@jP>pJ8;)9c9%25e`^YF~tFTOr~k
zezZaHTNE?$;3t8&pIlOji=2tVo7AwAI}s6GLA^=6n}bY{{7rcKjXG3wGVwVkRw#*2
zp<FY1nG&`mCYNlouJ{20v1yEv_#CpDEdVK!#*QLPC`cn>Vt%8Y0qNCRo$sU}2E|`N
zu|i{k#A**FW59F6P(;^fU8ZDz0#caK>Q80u<n(iU|5fYobZL>C$sOBb^ez7sf;GRy
z!WLu2qBVIvJjY|s!1Ik=PwTa;#D04Lt3_6;I(VGNYAj;c(WEO=hVSxqY*JUVZVhH)
zp62!#5qgm?pw1E^pCS5O=U2qm!#C6g!*C-+A8iqy+BBE^RA1X>p!h;Su(P^P2-f-+
z8BjZ;vWPSm6G}ZGyxn}@Vh^opY`ck1`3vY4qY=CcUn5CyuQU!FF`R4h1E(35lCgk`
zRdttO2B%EE8{dWhRyf!H0fjejq5WN)KUFbP5{PGg3r8;*Wh@|k_=cF7`%irKl_#Gj
z-ypLm(WwU}<>XN?^7m_%sCL*V9kY5VrwsGnfWJkRLxgrxH3#urz+5cu+Ee_NB(yb!
z2C5=8tm4BJLLtI$C%~o-m8;ojpPzDGeaYmzKKU;wjDkKgOQ)lFqs*GVj~sm3<!y0D
zd)Mv`Gf;$PTP3RT9>yO=Ib%C>i_A1Rhs5;EH2wUhPyY>5r*{e4q-s-y`*KC&E-f#a
z+lRYu9Lz;5Mq_4LeLi6K;0w!5l{NXYlj*}w39EH1ub=1T{L@@qcG?=;yUQ|AaT#BD
zb#*nlc%XZY9zp$a!>aljshi^y68oVTVtdFb?NTL?TDIXVhVf!9OqN`QuXdovC6F-j
z2>#_&c*>&^zFcshsrVDRWZJhJ+qRq=<%&0G1qXz#V!9!Ja${KFqoW|9|3uFq7^Z>B
zj;uRz0Hu4Ri9p@BGX?440Z=pj8&!eG`Z8HRE(isSyPgB+Z5FUj1kd(2YKNHn@!(RC
z32Rmn{tq;HQvg4i&~4l`b%%Y6sM+5?VE(>q`sRi(4cZd17qo3?8bO>SCsNVfsIJ~2
z(6np8A;<}rR_ra-XOmzGbk9Px>@*N2%JoK`?-=6LjJ&+a#s_{xfFD?)D+j7T7oK$R
z8Q)jWB{_a70=-?pHsXmID*&Bs=u%&~H#-^M1O(WH8TXJaM}9HEl=T<7TC@YV1(Ue6
zijLFbWRMPSyH)p8*u{jWPYGYws`UNAe-9V8Yb-+JL1AQE4vv3gd$LqlbOz)eS<h)2
zx>-Bw@sL>8O_eteee6Io%0h+M|1`c%_7!1;?SI=kXOs{ijFr~i?;9U^;i+J5e61iI
z_-Tr<ZZFByy<*^*cjw-{d%aH%soRy+;*AGBjDEtsQ3J3tS>m=m7umibKyvUdp_gce
z%R^+a{<*<5X!K*h_nuSkpO9k6>!^Mzpmzl2w^n3FR~JkWGMfJe5q1L4K3Il$dbH2*
ztqT$I+$!Yo7nmm$LNN$?9W(qbNTL&ct|gv8PT4av_JN^*>gURrV!G2f@+k1`L;VLP
zDJkY%@UO&nd*bc3O~-);i@H^-K@knH1$;iKqf9)<l7S8%jU)EAwQhocRlq~-d}PYk
z3p1O;TO=oOIl95fhgUQn1uZ6lSTO-IfdKw<R#D1D%s#$h8V5ld>ufrXsu&AKLrNa2
zh%UVp%&T|;tyfk!$x6$UA8ChQTg5pv`;h6O!bt63VytGGqQ3vWj@chkRgPF2>$j%M
zRv>}!d-6^{IPc%1vti4+$@FNagxXzHMaC|kY4GU|HuE`X+%TAo$5I@LYnMtQvm4I2
zpkPE-ruynFb~Iv7M|lVp*l6J+MPAq)0SGa04(wy{H|4Zjbxy{ODeLtHz)nBVv8SZ%
z4Rgh)!#&I5@+(-yHVE(?e>@AE1#$1V5oKE7T6T91)7Yf~CgI`|(8#q5-tHSiid;56
zpR*8orTx__&F6TWSB>3(R+S~B81;hs$cNFORVP*6@qFN1C=L`<C-Z+mqc1x33Qde?
zRt;o7mlejpLWwcq_pqF-39FWGBpv#|?p1iK;AsFOBcoveEdY$-tt^XW<CzUuX7Utx
z!YR#7TwGjmR@gAGEeHM1vsZ3t9e4L>9c34Z-@5xf;~JiRecAeiZO#&Sc7zsvG4)}q
zbcqRfLetFUX<BS&H?dn)>g%DH|Kn%PXcn-$_wk%XLCVs<z9D1n(lAQ`pp*67=(tpo
z_9$qkCKI19SCJAIcSA&$s;@r$gHb0Z9$`UH?PXxm(JWR=Z|^B`gFIrGRacJ|fA^8j
z+ZPi0Ikh}aTE<cjO087Dqq5y`+RqX3*nM7yUrwI$GI3C@KTH<YLgV)+D>nc9FrUd&
z2sy(z_>y`_Xz&JpUUwVd&#*%L*f}QTGZ*Rm8?uhpPjllP)<<@pUNQ5)guORfk(&=e
zQwSs|LQ$~E!vy!TaTmQjG?b?*QJGD{0a$wYFm5%?bQ_gBC8c`=Op9-YR|~Y8X_k*i
zT#y#o`lmUL`=%rKzEA(eu2XDBaECf=1g7rOM)tW@RdTNroPCds8LaBs^f|6{+d@TI
z;R%aw3RLHL^Oagcx2KMgKWE_K_K=pcqw$&=9)|<m2AFdZJVY;bZ?VK%ClSL%ejhwc
zOzg?~5)5$<R5daPmA742y=q{d0?t7T4!{c8h@zC*RfLElg@L2CDa%84cbJjM%|(6=
zPEJ(&-lNlo;|WJ54IB-Lx#l=|+&mhU7@LeD3{mXMuhMD@t=qd4Z0nPUV%BSo<v1aw
zp4+6g!`}-L_&r2j>EmBfuvuBw7hdc|!#k`nP~3%c3OwF@7Rf)##*FU0-G>npaHKjk
zVbKmh(RHoJ84zaj@m1Z^NgynIln05Z4-^JHuq&3N!WC1~<Mp+XZ&!f*uETJ9q}?8k
zT8fLNwf-sPlM7P3<E=H(jdJ!<H~L}$5Z}T1PIpxT4i;)vDx&-iQI>9vJw33`=!t*<
z5_ix1H$$l=HD0BPegR8P9Nbk=QY)17pU+*(<CG}prR@fL#81{e9r^Wl^`i|&Z*Jj0
zVcy2x8Vft7Q7}{a;iiS!m+VxbC{O{0JJ9-E!f=Ao0?t<8_yog)8^xw2vLS)k$ieyx
z53}p&T-XCx!FyYtd`+=WpxG_s>n{i6!webOTI1zf-!_0@<kfWSVR;N{-7fg+X*vDG
zOj4NZ=H*b@8lH6=o$r8F<XkFetSr;#3<Nd*lIkgmbtET9z*ek<z4GbckoBNC*It=f
z!4}cO?Cz_o{0!+xbD_`A!9iMoBm8#86M>&x9<o$uwI}(^Z4H}B07NFjwY%eBtZaIa
zbjn$(PQ&#CRQ3yf4*quvWpB-jbjS)GaSAj@waO_U5{rNPPaHlz6Nh;?zsL+(S>g-+
zIm-UB8bzPCFGvw>+j<Ll1ud;Y(k9<ZtIM=xg3xQ2!=rW4E*RQ(h?wyRu5JJ^>RS)u
z14~@HQ&G74t$|*>?b<@Nn8(!(Y2<^S)yaBM8Jka_7IWFY^i^VVOh`)JeCc!g(=8FI
zU?`bGH;nY+0R<<yIF-w@9KqcK950Q$i@h~PnE3BdX+rtMZHe0%5iRz}T*>w`j@^NH
z{eZilW)5DPPQXR+i}9FEWHDupdx-#1RJXF+ZhtvCSo5p?i4>K9ZvAYp$LBYe{`<1y
zidltJXI-mA>yDphGNB+NBU60;$Lh(-{2rbAapP!)7b-4P)UQId-WM!C+d+Aj@CSfx
z6IYxA+GH|sReaskWBSI3hxYtI&*FtPnZzz=&;S_>p%8d>cfA?f13bY2fEtT<WRt^Z
zi%Uu_#p1)>u09a?PawfjcmV7`#!dK0&%GU{38yH^NE7Tg={5O?d|fVm;)QL1YZ>)c
zC~qX)E?%<&s>YAwliZk4XETJ}X6NCj$z9gT3MU177O=XzyGOa+G2Omv@vS!ZFYwn3
zwQ_t>=E?KDWO@oHRC-6^l$rvp09xdF!td{3zNquAs$xx4FbYFaF45&ZQOSb^ktcpr
zcJG|uHEWrS@Bbv6T@XJtwqv<AN~qP7IpIQK6~Ni74mY?!j^Z6~IXj~$|6Gqw%}$?;
z&StSVgUc5L)H~;Y=Jr@15cJPMJHCEHq-y=Rf7v)5%7GAenWlhCXEZ0{RpM_INa}#F
zLE3cMb$}}g7zA5Irf^RK69Tv2fN;>gp82?}6>@6UEq|(;2_M7fUnMTyjphgAiMGtW
zTGQ9-kf{K>NqgTzMV(v{^ChmVuoeMSWRgY7pQscF2LmGL-gh33dXXF>PD?**7G4=z
z7=~ffo|c{^JA`_X&~sN(8)BsIX1xyfreNttiqn5iPn9u>UAdk16M?cM9F5`nx|gLC
zBvnej?lRCQ5buJUp2<4T%p&QO=9%8H(^*3G#J{2)jXyCK3gpKou2IN#R=)m%RpJ54
zefZv1eyd;bY5Sb8|Bs}T5htJ?8n^63M;~G9Rr|qiO_y3}UZgNgXqn7_U9WnHe*cp}
z6kRdO1)6&J-e6S`#0szAauJMIkELgB%`dk%pu|Eh`sHqdlQB4AA82!S+dSMWa^Jh+
zxiE*cIRSnFt`0H}$uqx9rZGLHS>i1M;~#1P$TB;=si61_@FU%gi9oa#(y&(ZX>qmy
zEE@TQ6*tl+$^HT_<`L4BXQ%IOT9>{|$;HlU6lAQ7O_!6imy<f4otXeGIc9bh$tcNz
zkz;7T014kZ)F1UBB{Gw)Y{u@liyHN<<x88t{z=h#6Z32VR`ew3KY}Y3aimo4Vn7I_
z{+hAaLZ^AJ6EyDr=h=HB+o7@Q;lBgi<xO55flnOemQt5~5l~&T!od3GEG<LYjDc>L
z@jjsCYbg9vf`zBbd^7U@Isu(n*7E_^tyQ-h6&Xn`__e8hBNT$R4^V9a<nYdFy#&%_
zpWN>yhtSn&7PYvbAc#FOopvL5DF(NS>x1G4*&QIX3gCZ$vRjPrbhN@n;dY7|Y$AY6
zmLnVl-k**0r3gPunw8Mf3Ou>1#z|>u;*OW{Q6)tyRgl^iK<RtDK6XP(fR)6P!$~r}
zUWCM-6<}bmX3V2%j>$=YAF;Lq*gLP+b)yZ5HO|Ra2^1-N`{oi!8_(T+{$k{)QM#FC
z{!9^cd!F6jL8~94@!~1uz~GU#CRKCL{tch0p9!y<f6!E=_pdJ6a)2fy;AH7)*VU5}
z(OX|998vte;5&z?RfXPxV!{yZ|5iNVS`;=%V2j%+s2*t<8{bfGD}=j^I!Dk|=$I~B
zGxbb$W|fo@*?*?1u+Tdg&TEKJk`<KclT1#2C|`M@@lNbJC|}w>R8oaK`BMn(7`)G^
znko;$)?yN+-pqf!T5Dsgz|`GlPFGZbF}9YqR)n3k?FInciva-EwNg)u7fL-X##<zr
zZtD@49(I`KQaX?pZi-Sp;310zJQk4v$PfhYu@GmKMu-T;SS>={=5XJJrRE_V1RmbD
z7|ErVsKLSr+zus@7{u759sZD;o|a~VRtaRaB8-zl6xuEjd{?#`-+=V+lBF}QsRM0e
z^d7-aqro|T5_;5DSV>_(^a#T^kqoc~WxSWVpju3IO7F#Jn*Hfzi0E&ZLTtd!@f{tU
zW+Vx`t;hYPZt53C4$gF8=l^tJ`M~zK)S?T6WeY53Hw|~8>l`QjR>2$6VKwTKH|UPF
zOT&&+j0x<l9ZD_ucZsF(2MDH@=vEX~qctbsKS$^%CtmHwIDdyp7fr`+*I;C0kjqu_
z^}WUP*t-TIrXkY4#1i<d55W3(mzlvph13fHCm+~<{D9Qewt>(Qpl)p6h`cqBYO+m`
zgyEvQFWyPy4%jBDeL1v^Ut1k@8daDmaTtxnSFR)@Yr7DXqIv;Lu^vy_pPz!f!>gVP
z3Sfh_YfESrALu4Ia@Pw*5`d_XH>2p7quWX-r%sIi#KN|R$f0aHA7NIW<wi;jj!Vfx
z1IxYvnbLM@q99d<_wZoV$y*~M2kye3AUP@%JJ!mo2Ex5H>PupEuG8|>tS2-9cx#$-
zXs6Xn5ZmOk4q?KhX;=sorYgSA^mNN6Ma{(IV5_BfrCK5SIIxv@>y}K-mlCS5kt;y7
zGIli+C4+cHmpxyX$W<VD46^Mu!Bo#tce2XuPE}osp{ftto0JKIG?vDtODyGF$0BSn
zaCWoyh+iuW9)K0O`q-t?iOQTF%ftZ^zLSq%i><#7{3@=7KZsYI>D507bQvMg&sB{2
zXNVCm;|0{!o-O0ynEqy0gxODOV-@7qbJ25vdq^7^$$NkvEzUoR<;i^xkAC6Y(hh;F
zrhBI7xK&1;EMk*KBpV>mgr5*!)PI+ubwHO#aT6slGBPs&8vqyGD67a)cSE>Pu!5kA
z&@Uj=$AG60`18B-5-b{8;B19J^%w*7iOGH+lG`H~U`|Bax`;bS`(==?aMJde=EHJ4
z>Y?3Rq>Tdy3V|ecVsE)829M;;h@XvDuYwUc2Glh=K4Y(u`)^L)@M3Je55-NHJp++Z
zFfZpu*c_)%g|?fR^0>1=J;-?h4{$wZBwKNwa9_OejjtPd%KwxdegDdPciR^MQ0vS=
z;GFB~h=Y_;EWR!s4`|&$_)|I8-w@>$@h*#gLym?c?k9ecMJeH?O?t`UQS-^&9nDy_
z`5tojCk!-vizcMX?jjynx3Va9fKTUW^i21?ndayiHKkh45&=}O`rgp0)I(XI2;4T1
zA8zSgC@K3O!9Diy<1nHbnLZ!_5e9?{w-G-R584Vg80G5-a)UWc&+VKVrX%bBCoO(!
z7R#L}$)cB>5Uz@vnc`&#KGivCxc1fdX>b+7gV(pk1D8}&Q&TN$LamF!z~JF=eB^SA
zpO=T6gKXrk6sYgosH)`GnBkWK#r;$A=NmK1_{6*kehT_eyK(O(E=&A+-_7as4*3U=
zCBKr45MO{8vu~H(5WYUow-uit+9N_XzKZqIb{a!-Wa`xVA4Kk_6?$WtPQa)fU<d-7
zO61t8_AW;ZrAkXnY1t!o>2hG`-~=!S?XGmz{oZRLdKe&6O>^?x=a74{XM<vq<dKQ%
zgIr+Xfz-IFHMq)4Qr&8>lL}?Ml5q1|*)9@BuW6u$k-@>4{3i-dL~gH{W>H!E+q9fI
z%<~bTsRNN~Z3D9s@r$D?>G?tHzWrnDGGf3Wz55JS;$dROBg~fGH-ia?PHYVh43d|v
zz{Tphvi{f@aQXP-r3-c>ynOYFOtSesx|2BMjcIoh_LH8q=hegUhWBFXsMaUKuXS}p
zcfHhBgvkeuH!1}D4ZoTLpi+VglVfs9joWMpy+MoukAPob$x*cI=*B=MT5u(~ya90)
z=K;XCEnt%?pRLu}zN^DPBH%@z6($_;%aQoD`|qW4Bjd=%ST@Xy);?voG(aA-UF5_h
z`}aJlG=Q#f{UfvHwy~f9lUy4|{4Jr_l&e7wc-^e_HN@%&W@sL;BLcH3tiZ(Fyam1s
zz*)_|-oLd23>%H|vq4B!PUQBgzi0x%7xq0pFwIb})x8}~_%WuYAu`yN{FcVTZ(Ss(
zOpIJ&^ot1x4%jA`fBf<8esIBB61K^{OKC6$8i=wUzl<Jky8gS{=<vyq2lTKMg+<Du
z->B$|L&$6Y*N{a+SpVb9NK?DQpErL~i-!CnMh;wwH?RW261%lPstGv(V-DY%=7;EA
zG0^wI6d?M;7MyqxJ3hir-3BR!k~JU@u#4)^dI3SOI8iBrY_W4~IG8^~6+-*zfdK%0
zT6BLN`0~!edjr%QcO{Tp0aR6BO5kr@5uzN>(*R(FUDioh5uoJD=e-p3loqX_03cs$
zIJ9H>-UvT>&|+A?f&<^@Bwlmp{2OikCc<orXryW0jggHjz87Ga69dd>-0mXdWB&66
za)C`i^=g0>go@Q(#FpNB=-lxwFGLlg#04xiMT`vX@7M7IX&*xDuIfM@Sz1}Mm5@sa
z=^o*r!tAG<#t{y!E%4m}j?d><#)^HkI<!^&iA{AzBK<)Xw&^mm5OFa8VUYz74hhl|
z$?wpEcntTt<IxObj9X@sAR^q7xREV@=q{WA#~!1no4e2$hDCTIF&G;lB4cC%Ae2ya
z0F-3Mh73#gJaS^bxY^-ngJ6YF4Rr;1B-3^<AQW)>JvkuKfjP`kV8Z}<n3<Sr6xxG^
zd>X40$w383V2BLQ>iwG$B=qg{cVrq_yGn5O5C*lH@7}t#rI%O#77IIZQbHIs9pa2c
zOPi7VT1|qA4*`&G?~J3Xx~LLL+0T3IbpmX5rh&I>mb|?E3N=*5=0m9}Z%+2H57I#5
zM!^gZ>hkJEiYL5_^sEn88@%-W{Gl=<Z0eVXuljwTcanK&T0<5NhYfpaWr2vARwy$0
zwB&$v9A*mdS-x8Ms+8IU|FxHgI&HS;5yT072YvU<C~1*}r9LY!tbTzyq3oh?n(f5`
zL$NpBE6gSv;vT+m6A~_+J}FUJ8^7X3>c6-E8K2yQ&cP@_*NKWF_vs^OOeDTqphFPx
z7E1s-1--76qV`ld&4Z=^*?q=BxExMxec+qnLaK+FfdkY0xGZJ}BNazOlP#T(cH_Ct
zpUzrQS}S?<56yJRNki`-hG)wxS*J3!SkD`b-dvpo!Gtb53c%mLOQyX}RieHE)xFrR
z++YqT7Z(Q2w+2LrMT0^in^WgHrbv}xE2N%+)v6iUjFvE6KE2_cW=#0W&njHomK?$;
zo+0Df!1g7+)8F)clZ^g;whWZ-bq&}SsN1$&0mja)5KcU3OSgr9fn)jzt>wrEYLv6&
zU*CJJJ<v@w*C8^BC=#?Db?RA{;FI(+{rMyoF|$KCpDo+V{pS<>T7t2}oCaOoD|A59
zycXHG0VN>Iz7I^j8H?!H>j>{Jh&S0DTx>(zBdmV*MIN|+`DoXMyP!LX$t$qBUgdqq
zS>iqC)B#oe%Ei)*E2M(tv!y32VgwhaeqGa2QFH0jth5Vg;LwWl-*r<W*9E!v$DX9k
z9ymu@Og@)J%e@Ol)Bil3wS72-xV_`m1xB1j_}y3j^Si6Z$byZJ8p?2&0z=B`mWsPS
ze9Qt@$5qMuEm#JMbj4X}!(O_~PXS9XjCX;oaMA8J`$k$VSZ8lux0pV#zB|sy3{M^$
z`y`I(0sd*$$y6k6m8Q?3T>VR-A3S%)xn>}GmLW!LGym|vg!=ylp*}L<J42d12}9sn
zBhjRXC$+8=+ZV#4uU=+g*nyKDeL%|AS+*BFIsTDrHaRrt-iXCt^89Pf-v5~g>t1Ig
z*R&=yVyeq$stbs4g$oBXEIQzTxnWS)u0;;&Sb`w8Zvfx^(J#;hksnlFX_6=+X)Oqo
ztrK752Snh=aRjyg15(wB=A>akGyeQnRPkS-$^T8dnZzO#``?#9pmU`Y4a!P!aSgrC
z$O27yjt5TdKzT6u^XIv*urabmA;WVK`p~w&ShhmrBe?@b6fm3#qCS>XQiIX<9&S6}
zKtn}aL;ee!x#G~J#2%59k?BXrS7~?@NBu0>jhtRdo6PVH*2?l@U;J~R@lq|MdhDDL
z7c!&a3ME~e&`#?gWiI%BGy0-~>f7sE5$>VfwJC3csBi^oiZ;8%oqW3&^-TJHArO^`
zI;H`4qtHlFAcgW_guD^vI8u!>J_u=cGhg!7aP<On6sjn%tR?Z`H8e?-RB;9WKMn6X
zTTY?!K_vH?+m0Wz?a$Z9**Cl4KX<k?2jasS&Y~#0;%ZeqQ|5jqEX%F`^E<vq$wYME
z6Lw@MjPmo~M*O9?>dX(g2{?i1KAnxVGv8w-Qm?443R~>(KD!e|stFG%ix<rNz-Uem
z4l=pZzDj!=AY=_k5IF7j=kKWMLE|X-T6DKFLPJ%D+KoE9=ZFmr&Q#!n`C5Ow8Mqov
zgk@wh^|#K3j(J8#o1KX+8+2gfD^ckDMg1XzjG9qdZm|RoZTmL&I)aQ*<;)8dUkOSZ
z7Okc1#@e2r1Y<HHTy0)+cqU5K!t@c9Y(c{jERRV$yC08ZJAKUg7xniXt*owA0+H>O
zC?2o)A<~JhSV=VndUv^U;S`ke7rABbR)s>X_@3UPG2sbl-z6mT@^ONUbY{L&7)QV$
z_-<GLlq-1-pZ>y(f>sjX1)4RTF+){VRYb@P1w7(V5}IvE#!L*b_O7u=vc0Hant4g4
z;Ww*6HU^n4F=G*SxrbY<To7tT7PXjyMtv7&INb|cx2w0eH^{-KK7)feNW9PYObZw)
z3c!6LGTi&%vr8pqmR>rfH1~8kdVp;77k28n2n@dEFeyvbjt&1Vo|g0<Nt*udze2eG
zYq~yh72InEhV+ikPU17s>;la<EWsL4oq}li4!&*(8^8<9KoA78nz}$ZVbad!_bluE
z`Ov##n708+HdM}NPEt{9;^O83fFsBl^E-49gyq6Tjk*(1ZYx~w;X0@P69_di*Q8RI
z^8UfSASy^>h)(-JFnH)`m<A?{mdgcV5>(<z*#~Yd$M<vnnw~lZxA0|f1O)Rku0|BU
zpjUU2zdfwp&@V-sj3Stj{JuUlFfb5OECS+yoe~cQ&=x7XVcbM+eI$GV@%$tF$lg!!
zFr@Wu$5lozY^SdJx5y<H-!NQ&b*A`;v~~&t*$fI{A6+oj1S<0c1TjgWdq)ldOdulI
z^YdV2!Pj?)#H|G;fpQ*40Q^joHWUvBKRN1v6GK{ba$Tk1QQYXwdAuldshI`}yC;~}
z1DFZm3iM>M%Q*+2<tN29@%+O-;In)N+5#;I?N9@VxToG6VKjoSr>_P8FEopIJ27_i
zwn!j@Q2C6DpR5179+wmAx$#h>DJ;3d5{HQ*+F~vU`a5Vh4tV|eDwCyhHmiZCRJCXk
z&V&yOUL^&=?B9$KBhGvY6VwY=6rBzQ@_In9gwI}sC5(XAu46O=+yK4iOORr{pj1@`
zp#!fnqsr!13SK-}29bCJkOC%|_=^t!4h#FnrW*@?GNfkYSbmyye&?_e<2G^0v&@p>
zXPE=%b24&o=DKfuTzy|{M#Fs>@vl-__TAxPbyX|B^J(^Mrt5-dr4o%9^>l~Tw(mc0
zJvh0>;b%%jv-Tm;wF?HvCpl?{v{>bfj>@O`UCjymAEF2Lq_M|Bt8cGUUqLW!K9_qo
z!;0Y5o4=%k+FCHZVfN;|)Y!3|QXn!ljkM0bhNp=v%ww@`8Y|rhqkK`?a^r6#2VNGM
zQ6dwQyRYs;aCM<5ex>SwO=4bWcacP<?b->fR_$<wgWeIiRbuc-i#cg=9gc`joDhAN
zUzxKRdIgZO4Bb>`Y5G%G4)w;0y-yI$usb_rqYO}fQgOo?I3nUF0aQ?!yfby$8NT*4
zp5g|?S6UN9wvp(TpxTOwi;DyQag}+|2&i9=;2#eU!+jlx!*$WJ)4FY(L0m$e{QP{p
zftmycx(^NJ;ybV(6g(`liiFaMc81O)XUG=dA7_K3g-398fF0Xo6CjQ`DTx3p0N(h3
zAD1{qFj#@P6vOF9CI#O{7#uHOVtL#m`owSuaLhw&u}FVpin@k&eB#@1jwMatko@X-
zEfzbuic~v!Lf~(Uz-Srwz!M+g+}-eq2q)r^1iuIJ00Bnh6xvgrq=5;*bivAwfKkfb
z^0y@E`Y~~0Vb2ozk-w^MaBo<qd>7<_LCpV{Jnj$|u}ztgq49O>rCm#7NiA(lKGndg
zo-0%4BP&Y)&%55-hri}(^on>gGr!c=31~>hLByQ^A}(3^ER~)xkvzcYn0pdNKZDZ~
zFyjU^FH>6#We2(zu?N%CbQhnx-Y<*3BBXF`4FqNp%=w9Ue;_mo_a%WI!r|~@1tCaq
zOuBmY>ayQ9z;U1kiu@oon_<T8i6MC}LMCx?^87LDOo8et_3M_sd-u{R4xCB$cZmmn
zVUG!@w6++Wu93#OZNjh@M8A8r*XURvbTP?2TXhNpgAw5MJ6AUVOI*@)`Rdhu;pxqY
zb~??&9L1`v2!VxYn%kFGRaJe(?^u;NpJDTCvhfK$gdxTJ4TPws{@4ObUloix;4gl4
z-O=^n-Y4pxn7zl;QEjmLe)-+EgFr#>vjrX;Zr<Wv4A)tJfzKYy*c#kj%1^5*RQDAX
z1M@3iwl)i;>b#du{U|PMO<7S&{`qNC7c~#Vo*ZF3`Ku)QzZM7~cZah=fODDPYomcO
zlwP-RV>cCu$jko*FCgeUc|qYnKI2Qqp#*+=;h4;DpmdTh?Vbo6w;)Al4_Fa<do9_G
zk+u+%IDYFQjcK(9qM0KbnbUGdfy*@yW5|)!n3KvZj=fEQ3(f}bic=Us)J3Rp!Oz92
zrXTcMh>3~WzaoDcH?6{hhYz<NfBH8`!ZHlgPXLdmRYj`GI>6*+uOj)b-9x<Bzm7HV
zxLzMab*ls|@O9sb>2<dC2QkcdTqNRVq$Mr>3A0q}m&zHs%gnHO)}O}#c79;fwTz`6
z;v57+)yDlM$_I*IK-eU!aXN8=F!1VC4`D~Xu8%7QhX46+$Pe`R1+G{EP<0|0;NHbt
zy*dufC6*Zrr-ge%?ZaU+3c6f-O;yM~=}f&F&j1wjHIaWZjFuWIwkGLfHgML(*bB=@
zXrqr+P;4^no96N{nP~IF1}l*qd2DIHYu<Oniu8hQC==Qqlz<>gBD7=hD6IjyyDvG3
zjr>eu=3$(;#)kPN0NAqi&&)znaOJX;VxF1xS_A2YS@5>(&)bpgrr=X^Kx!Pff1s@#
zOhD;pmn$z8ik~8pTh!N-Sqw8?RSdRVxe*7EVhm0OG)CzUiWr;HV#YuvuW>-T!Ak3t
zDv<VRk!s^Gs__kd>>*0>(>VJhcxKFUWb%dmCYuj5OMJy$GRL$sn{`?pC!+-;JzSz^
z%T2(0gZ_=SeqRsXD-6-B)SChLpM{fsQSx59GYaHwhaZoz+w)^IlsmU;6ijg7qDz}F
z0vk;%Z*IQ=lWDM!4>Vy<0g4u@ruLeg7JPdUFqbu%C)%`NLI~t~-GhB=)^i_2wxleB
z^Ha8b44gbNgaDRRoOXbVJn_!5V6V&l+)T6!zJGoRUDXF5L91Q)%}3ir9gn_7ZDE0$
z<}5qr49g!rLdjKW;f3g^>Q5$2O-tk$mFy*oaFPkO???c$jAQ=EH4Z<;n+PlAWBla#
zkLsxXu1YbWW;mynPcAwisr93k6O%>*p1;s2Y$9VNjU0gLvi6=h04NTP2M6#z`5r#1
z<$0;fcXoK&C8;Od(;W||&l^SHP(j(!IE+;=;)4o(fzq@wsk&NgY6~h6(iJs#(Mk+r
z8`&M{PSQ;Sd&04ahz+GDe`2}JjA2yrdw{R!Sjz7(a~suOU*l(Qz|$aL^EmH>6aFh_
zdI2BP<xn!~>(JRxf+dJN3_j6lwf*jk<sBFZl9Z?2eryA79{7Yb{83wwKSFjrxY&9A
zu#UNr9f0=c)SqhykjNO2dGl8MiZDFiyWsh&%E|zxOj`Jg@vZi8tJ)OU4_cIl26o}%
zT=Ej6EWJZ4o@1=vwucq8=Px)g5xmp1Bo2Y~K#bo4zb%;0&?OF=nojg31}Q+VXUcLI
zI>|E6VK^N+^HR+)C@KqJpNYgYFrkIVrCO(Oz8ss#7cbF(ercq^#$hBvlyynI101g)
zi=lKn^N^8|PC(uoq8!=r<vNlD9(Nv0wA+`H%QEKtG5bxts<5nJ0D8tiPugut9w)JC
zvl%we_?yW-X$2DnhooHD8T^V66pEFa9!RYaEVAu86OVeYd5UbVYjx>=H-Fqj#6Apf
zMA)@$FYM!jzmzLt0kz<>zoR$wz-BxqJ|}F+#NcEPiW?43ab6avcYLOxchPP5by%|v
zmzIhZvXuMXjT+R!M~~sOEV<G#^gKK@I!J4VySp*eZpGTub1WOgpUg)vq7~irv+=iy
zn8`5)&s{aA9xX8}Wy&8UV5&!R`euXr!s`_{J@C6WP0Yfa4fK<X-zV@R=T3HQqb)C3
z1mg&2m-6Kw@WpUkB%YF$`aVOPl5sdH$zcG6CP&`hAJq-U+4o-n23TRSyMWwNP<3=e
z2cm=}E9AzIr;f^I45bzg^QdydkD)RXy(+QbkR0)UVc=qg`g56_&0@I`01OntVSrS=
zd?{GC`1<AOp=fg^CcSXkCq)9<0dCi;FPod25h3n1zFH%KT0IP6EG`>6q^&d2#}zgL
zu`JuV4qILEALxGfCPB8J0$xJOq|aQO0~&uJzEVLBMHui-`Qehy*FVBTMD^QG;6ff`
zUAICA*$myoI9i}PC>)zv@Rf9$3TL||4Gs>{ips=w_vgX3asT7f2cNm4$D+xn)nB1o
z=y>AG_#RgA6NABTcN_^t^41Mu;|AgvT0v3vS0ITHDss?cJM_F#YF`uj5oH4=r|~SS
zP7^Y<!T==^H_lLNk-L-a!YL#vvHaFUq4@5?HtaRoxxoC9%JXaOMNVbM9s&oJ#Fhmx
zOlj02v%}4EM=@|iOv*!FRKH{br`!X)r;DzAFbl`^&&LaBIB&SoFS-IT<~JBAx9rYJ
z97E~19Gg_okT9Ed@iCm%zfOlQwv!Yc2oAK-8*kq#?pj1d4vM@ye;lMY*mp;(Y0*b%
z94H2z3Ez6>_z7uNB$6>u57?b)Qv_OPy+Z?$*J#TlQX_fbEAZ}eUCF@UDMOL8LP1ec
z5v(sFUDJ;NwMoqVEy{%Pn39PUz)z^K7k<qFW<+NX!+s!2R<h+o;QYp6_ag?2sQnIP
z*Tod0Y9bw3cj$=C7fAiSA&wN{cZKr=ys^YTer90k{cg902{`%@<TL!p60mYUfB=M7
zLjq^a>xff~rVi^&v8DL|pgR%>B7-F0C()0G1XEbu9856uH5lD{07)mmD#S2qRB@Mb
zva`|6(Mv)QgyFU#o`nhiefp;7R-QO0SQ@<-skLz{6)^Bvs@{|=2&AeBHxHQ-ze_Pb
zF@gzM#-sh(2U!;ANt8jrfxfA<oq=IWNAdUE**vrY9fe2Fj3rV92PREdR$Fc<z`6zN
ze|aLo6`T+thBN;)W&GcpGXB5ine6R>Q}=u_u0xlQs>Fo<2l?|M>YB!37#d@v{q<hp
z!?MkEw*U`0tIwKN<Kb(Jz&49Ei28Q8axiRWrR_JP1Sd8Q1B)WYr8Pfxj(o-q8OAQr
z6eXq+ng1VHeV*2q2NmVSbQT(0XaAgcu~>eTKfNAKea`b3IoimAW$ON&-FH5qa(T_i
z5y@A%)nmu<{W`+#r^~YIfhvdQ!)0ZXu&V-VwC06~->*3rrAI+F46(+GIK`hJI*g!j
zC3;%Y7C+?7hlm5Xw|$r}Knf*P+Cf+Y>O1rnUx^KeAc~#`n#_F2_yAbAp(g|DgBm<p
z<}b8X7X%voqp0au2uB~$Ws%r7mNB#vIdcNx;}tyDFtbK#ds$OM=rx2)SOeky3~B<X
z9Kcg`C^+D&e7R#*c%)wJR*gXhKc+bcuc3x2n~Ye8Zl}I@Xorwb6!OlDF95hNLNP$9
z_t;tlA8ZEeF-+bX@u-L8aK9}6_3V=K=%*ZXyMQLD^f}lgRFbUolO*7@VC8C0vk!zb
z;FJmP;y~!lhRB7L>j>zb<lBr=nN7V3$&Xg6QO-C=gKAm@rSu6TT(-CoXtC~Yo~FZm
zP$&1YJsxu20BI(Jgktd!Nj^Jrj23PKPNIHDHuzFQ_{L<tL9Caz3|TyWp6d;Or{9C~
z4@4dLvUti!coC{NxMjfUqMmJ$?^8>7g6Y<xb6%bw)eno3ovsSWFTtCFMJCTjeWYj@
zLFA^jqXOVw;y*+lqZp;o0!N~s_9;R{52uif-opac-oACqfheNvB$6){>NvhaI)a7E
zQFLvUi;B@Tfq-%xg+%#$)I64qCbV<eD(L68d&pz3Q-P5EGulI`+%8C?$RY$uJj>3C
zc~WwdyAPwznMjQtSatX5-5tZV{-GVSG|5tXaVwY)_^JNg(HQ6ELFiw;<Op&f$TV7d
zrjZpW_o*%czfO1@u);U={!~G0uFJ5A<+sem2cASR@#^KvRVf2-UK~WN!r9+U9|<S2
zZWVYBGPxHfZc*Zc1H*3s5J9=TXy@UxXnJq|6L4qHZS1*iJU|vePA}IX;JoaoutSt5
zxC3e#RNPC|G@ALnGjjjq_f|JIH}AFIPu*;b4e;)~&ggxbyc`n8@l%R0kx|<qf&pEi
z)ZG#AK+H4+I73HSw-bhdsgC%g0m{7qu?mER%BE_de=HE5)0adHU04BhsMKI12b>IW
z=A3~7&@Yej7rSH$MaMjAjHo4m=>W4<_s7P>kcB{YTm*|}0SERm&GU}ZRC1cXV|k(+
z)lQD^^G<neHVXNFgPuB{QSYn-%nt!3=9W3?2?8-J#DzN0gtIsSnuP-j1SKU;G>%#0
zE?+u22(=Ro8ph$AS;KLRPz6BL<S>fblQt{ka)OKmt=S7i35(<bAG}NgQG8X00pdS)
z5dx3%eUqr$Ch+gWtbMYxvq}986IML+HTGTT9khYQP~Vbk3Oo#X!E5hdOrFIJv|7r-
zpO9yyg`Qx+FxiXu{&@qjbaI{p2-HucVpxP{H7^Tw5h6Ti!adOy=H%os%b_QCLD>vY
zj_Mr*AY5Kq=m5#yY4bx`^cAnP$Cni|)?c_``1cO2Q>YA4ufM=E{(uXHUXD@|E*w52
z`9&g<kjYo^sn0xa=;`{se-{FgQ8$r9v<q7C8(kyLEBBI7TEOy5*E5aQ+*+H%tscQA
zbIU6aEb_zV#kgrVU~{Bxw=JJnBAvbyOYnGRQrn@fM<_@^R+AYXf|$hIcT{uT%$pp-
zn@oO2gBxTzeh|{g;Q_A#=NX%vK3Uxs<Yw`w0Su1q1N;^E=$J*2PMTy{*X>tEoIwRQ
zgd+MH1jlGV($EJW#v&@Z;(ZZfiF0eZYuB!@?NXnai{Ye!MuZk)(!T|Ej>88xxMq}{
zcylD^RP#C_;r_1l^AC)JGQk|j$kGEAFqzb-j{PDH0OlyP1WE1hbYDD0dHg=JFOiAX
z0{?h;?}PfLrJ#|E*FZ~?A<ZJ<98w5&TpI&31v@wy(@b*qLKM*}9Y9O#iH4c@%9H@`
z`w%e^#X6}G@RvJ+R|Bml*$&V`u<dnGx1#lIz-z=$r!7v9UW}jvGcBEe@hcdo7mo2l
z-i&1f&RU50chR{tUIU#<=1#?w&`+TkzX2){@Wr@vU#`sn;CX=13f;G1iZlVp$p+s&
zsh!4Tm^lSc{G<h|!!n;x>F-PIMDruztdnDLiB^ky*98NJ0Km6CHU~)<&oEEp*XO7`
z<`T{in})(S3hx-y&CnEb*Kq%`IDBK|6$Aymyf=(XX@G)ZLT(wm)Fakt7y8@HJQ`n0
zdKs#W&s?n#d=|oQ@gW8ZLd%X52=jqWT>u3y4ZTP2i4TmRxp?TX3p9x6J(&tj&m@5?
z?uE$tj89BT>Kk&q_UZ<l*Z9^P6y^a5?ch2JK7}vW>M>oMt?<!aH;mudECk=;6UVu^
zrlYVzR-x?*e#M$1?oWVz;U}Sz*GHoN>q#7r3KVPemkRjxo0I`fb%JG@*nBi0yC8Jh
zDJyeQv#dsxb}l6y{s=s7!L=>LxS(G;J2@5B^u_oS)R6ZUxhnF-&NiF|cbp~mAmXw&
z<{&9VO*?}$nvnN&IhbL1>}G9y+R=%V%Tn79C{mSy?W2+2V$oy>#MYaA*)zZdd`)ZZ
zCsXj%S>SSVSr+FpSCF7mg69;drgd7uhMDG*NFa}}S<WhxOdPhTW6Ob8uSF5^Ub1k5
z@JF1?tSA0klIS-WP60g5m2c|3V1;{~UleY_GZ&3_STQEBW$<3A*dps#MPHFcDO3dD
zZe`WsX|4a5G?|dAZk!Rjb>!XUjWcL&=3|`5D%phuc^<s3oEB7!*A4he!akQ3HH2+#
z=XHOwplSc&QU}H<|ImHOh>`{BDlf~()o)~ot@so07fso4O#Ym4Q2aK<vlSm5T@EiN
zCQeC?rTS=^$N=2cx7#6!M1!TSLK0aQmm`TL*5<T%^8`O)J_&I$<F1wUF`toHIBj#X
z@s-V>WvQmTNpvVhKs?!Ixd>2C;(RA3C%W7nLcJh5Pr3HXF8-Z)w3+`G#9YrWt+!Y6
z%nV%JI%^)U?xzAsA}ll|M=;l|#;Bi7d60A&EdCcKcPj9*(5RAe7aSE1R&L#CUQpFt
zS6G80%M+!c+6HA0mR?w?aWx6BS9OP@G!RHysI$CE)M>ufa?MuoG7p@)jdSGU;!?16
zT*Wrz-jJ;Sk4!mt-R)v?_BJpkZiCG>xvwEjOnzSEa~Xje-6q0$F?tEQd4nGhK{y8l
z$UbQ+{fiNRhgC5$K3GTn5;SJaeQL0;Q^jYBMshqy@N_UHlPvEtU;^?>Ktq=b4*4J+
z0o`~M2j83X*D5KE<M6~QIJQw5KyfR&)+EFH2Y!{H<;r6OlLKtDx9>we1qs$g1g%VT
z!J~Z9EBBs&8wuA^oyykxi0FezeILM+5a-B%hc}QaRMG`RJ4W6@`;BpX*wr;aHC!=i
z#D6HRP17XwlF@`Iei>&nJ|ttbbrFgWEbrP|R0lYR=CuphMTLE;kc<emB(=t<0k#1S
zfLkowmxJJ*$NT`pPC{>Edrq>VP5+KC{=WGJpX;%g00@NNFJbaK8~83ge3bS4mL!Nj
zvAyB?=5_!Bw8)J>G5ppW+AatY8vuJQkg_t7Ccz-NG#Sq%i+ZHH%>i(NG(I}pT;B2V
zanx5a$6qLyqP7<dLA_+$z}K-$;j?eWi(+DO1HVgOD6J;Es4_^@y)h@;zkzm<Z;f4|
z?BQ(JjtM1MZw@~xi)xv#^(Q61NG=RswD|dQxn;^6OxP>8qVI3}x~6+X*NsgA)-i-m
z;<<|_j-`+V2Qon}prgi360y2m%iJsb83`Dg(#7=62ji%-1+epT1W|{gxpHJF_bJ>g
z($j!Kb*{7#{06ihtUO}&k0nmMD~5YhMSKYnnQBjEkz@XeeVrJn@HPqDXRm6Lo<$y*
z+#XiVnLgwOF*M~p3LN=NH+w{a!7%o|DLyzHZjOZ8IIDe^xRyRL8B&Y!e$oPQIOJl4
z&DCipd2{yKs6k~6^Rko)Kgx4#fz_R-yYJ`eDXFDvkn$^TMmiFi1F2iBEsPfCbsHcf
zvW@acRr+R^fU(QC3v;JK?{9lY$e2OZ!Ci~iqv_(BQZ#MsJfGppFL_+fjFwZ*uY7v#
z+QvxLFN>o$Mi!%!Ecd#JI%$Wi&9vBe)@XCI9lt~u9l8;5=Ee4T;$av|&@YD2aY@-3
zu%7zvM{%;&nz9i#WRx_!@OqaLs9{3xFAxE(oUUkdy-!aEMus7TrtU-Zh(Ykr#PDs`
zMw0~=KX9LZBN^_vr=^1|j6o8|sQoak2gQQL46|L2f&PzaDYM@e9Kln-oB<hd|7Jtc
z9c*{;H!QYpo91WZLbL&14mpO|hIQ7wBH-f8EA5q#5eO?yYsb;Gf<0o^cN~RQjBI^?
zZy{Hpo%x8E=NKdM==1qy;jJH~IGD&L3o6DJm6fzh5=HOwCIK<A!>ESqRw3$AO>X50
zP2U(LdL&(Z!e|+bP1qRIxHpSTXGf)vWRVY3<1@l>-K^=~0ux``H$O#rVAW5dg??MI
zmPKMaC|%HNDZen#r?L|+hfH7Zp>FH;GpLS{nZ{W64N|9Y(wB>7o2<|a;eVt?D|qa|
z+6a5@81qra5EZ0Q2n|%uUe-L*p*pzuH=C#PU}(k*0bzG>cE_0gV;EkyqEydbBzhM)
zNOPu9w#E*f#BBxWzR~_A-U$0xiS~!&IWs7;+>`JN@LAi@ajSy3Ef^>d?SCya7%wLB
zQ2XW&#zp#|0g`?EwBwn6#Vk$u1ex6bibqlHa|)qUq~+&Vtc<EEdEyS(51wjAEOoAL
zBU$O}RREkRKV;3jy(hEfZ~5lMAG7g0P)tl~$9Y&kELH-rfYv)e-;#|EdJm?qIEj#H
zyru;3ZZ5qLyyb)rS}UrEDhhQ;5so72?`c}5NLarM$?9)RD+2z&gr2+G2>?9)J`+Kb
zYDUz%q`4!LF}$nbk9j>1Q)tT;SzT=Y!hZ}k0RZZ0oIT86@!mF|L7+L_@beWGw%qA!
zRZ4Zmi8$C6p9j1a+@%#BDPb3Y(StjJ@zFLD;tu`w=BZ$S(y_JOj1%hu%+HGq?21xF
zdBW$;JK2X#*IQ+U7kOf1<e#_XMQ?TWvJD(eMrpdA$x1_OW0tHyX?+FnmhgGvP8eH>
z)I;j_m--EdxiWHHfaf=(UahU84Vjn>jfSAlp;^gw-`HKaJykxdG1Z`9x9^P<IM=L@
zU2_bzIA&*C&t4O8e@FWpVp}1B;1*#?$7}_#{i&(;Ms2+Q%;07lE;%)M_bdkk6cBB9
zjpK_@;cWM2qmJ6Ej$JKcfa2-;7W5c(uho8%2Q@~G7rsxRG!}KD3aPw>VfpXF{U~4q
zr@D}Kxth9Yj=yS7JO=f3QVbnbziSo#bqWmda{DEmOn#_nTJTelJ4wk1M&yEMZkhVz
zBCCpt`CF4}F@?S8Wsek!iplD55fxMgL$n{&s<oY8%3kDFZtBfLXVGI-w6_@@xbcd~
zBv|DUGDzq~-{Hp{0wQ$cb;r<xM3iXz<l=k*TNGS1P*&wTUrT7Ft$KmBDpI5NQts&W
ze)g4ePn4k_Vsj-og$_ra-lP~wVl)2ap8)L#h?h4;U7GqGnxC1kbtvGP>eS{4_nUMu
zAKShAE%^l67>4zGlHov7yp~|Wk4TxmV>N2o0Jk`dc@?p9ID6=rEFMK__nipAHt+Vo
zkLaApxryx^+t<5ayf9hJH`dReLmFt9<81av@4t=P=DOOnT?ZTmTx;!a7)pS?fSxzJ
z<K`*Hl#cbE|B|IM$uZu}iiS{~_MZYJdyftf!Wrm~c$&IUpb*%0eWn5-X8zzS1R09n
zVJ}x{N%v(QABa|O%-P1BgadC9#CqK#1&GMa6*mndHVSVYT>N29gn#&9O2j?ztll~_
zcMdC~y{N{Gg#oKOXW~f@C<D5JAZKVCm)kIU;nBXF(|OCjsa-U;I03v5)-0^V(ITt2
z>4B2z$$m6cnSe4-aZRBO|AFu)nse|}uHmCkh}E6253fT8rrr;43yD#ZTf{@+aan7c
z*5F+)bOfV|5H;@Oi5{F(_nQ`<<{#BlhMszKQp?lwOrvK4(!AlIC01>l?!VLkm{O$V
zu~l=W4CZzV#=<U)Pdee3%6c7U1wp;QgEf6`pQ8}Du+corO3lt60Wq5SIN+ADVVQHB
zVnpe~Yu|&g@0N-<D}NscXJY=j=y|qa=^?X{)C!OzZTUhfq;S{#@#*zFdnzg`Npm&S
z{}IQCd9q1gc=Xl=4&Cn5`pnHp8G<1f<=Iz{-Ig7xlbkyf5R#PFx{F)RbbrN|1g`c;
znX-Y$yGiq%`SpTZ1+|aOw0>8_>8Lgj3En-f5SO<#srte8%vzJ)U%#r%F~&9b{C>aN
z`8ZHb%%cmiXC5MCU37k^0BOup*DB5?KFUl?CIt4M_nxv8f@%bI?I7qQcFlQfo^TH~
zzBj5vNLTYo`p75JuOj%+gnLPjo0DV&_qd?gXrJ(IpuYeD)E@SNw2Vu?Yj6lK@ALJd
zX+aH{7Ag<cz(U63PR9y7?0{n{b9qC<zt3ym^fK}raAHh*guX2*qntJ~$Bj7{HPK=1
z=BrvVLW90Dnn!`7wn;zQfEK>{IVj9%sZo99V|R)HN6&O6=Jdx_Rz^-Ms|_VwXj}D`
zRXJ6`QDVK>pK@{j_t?s<3kaJ_itOy@K!~4{^*weir?S(3KD>j0jjPTviRUg($RZXb
zqC22A+JyZ0;cdLYM<^R02)6?Y(OfQN@C~_!mYNZ&)S-i6me%oHTy{ON*CJ~VX_nnB
zkIm?BLFH(EFzoFqpb8VH<io?C7>M`nyuG0B+0LZ0s2C31wKu4D(U;S{bKqstKJ#*L
z!Kf-AZ)S}7aFE)jE}#Z5_r=6}fAer2vwGPvQ!sHm9JJ1~cB2_v*SD~&u`nO(9WcKH
z`2-7Do(0BV;MT0Gxx^#wk!`u=?OIwX1#Idke~i=~X(!P=+aoq=EDz817hUA3`DO73
zrEve*B#3Q%^OY7n1FgO(a?LoF-SWr`pjj_``~W*cRU_i<jXiIIIK9pt-@!G#^Gk6<
z{c)t#cnAS7%|yXL5E#ndX?!5Hz8G3YiQfn{TLaK7eWv|DVMxS$u(akMV{Gy?qECg4
zf^^OnyJiT-P1xPw;EtC26-VZv=fBi^BYpvZCDTM}WGa~Vh%{TvwIXD^#!;zI0muLw
zhzP|seSc;B<&@V}^Ma7Jo>kSM;jtOkrav}Nyk1$itT5dKaRI=OS(WN!%!{DaHu0T#
zM-M+g_98*<N_cDf!Q1e-d9Y?!e)3%xO3&bpQ$WH{`9vf{@iO@lKd8d_15gun8CC<s
zl;2qJ9J=w~g?g>!f$@sI!ZS_1!%PWDr|O1abGxj6t$EzW&9e{Pm~${NPickPII^Qq
z$bmhpE}Xh3<$Oh7_IFD3HFP1)P=vyC85rnZpjrUE$jWMBEB4bvcIw#AuN?q|sJgv3
zFJL@GMVMs&pphvWSG0h2t{+8xnmqBeNRc-E1o%<=uHPCljo<Gf4syyzNS_BykamI1
zAl*8L@>}wLYXX9k?XEuFoPLB_l%AK&7nglT<F;Hnkh+2t4s$&sutHDDxRjg*wutFP
z$I5LdMvMHB2Jr_}mF!qgh@I4}-kUjPW_)>QBdobL7Z7faclOe!2xSmCd}yS?uw7tq
z1oNMe@1YCOOU$d>dS3idM^fWV?!7{*#>OLPT|BN+N$<G;ssDU<sBdE1@X2(VX^(u^
zb3*dY+z7K}_OloGg@%elxj<hJQ3nsC1cUf6XBTK){fb%2*P}a`n1}pOFvH~)(wq{J
ztSoc^1ENy}ETpRt!H4fzIR~xTLA!R?hUl-GLKS1SUvg*L)z+V&+H~Xf2OmpNaI|Wk
zqT!QWJ2shz8(t`<(v!P!!72_W*m;bjKw3G}jH;06M?`?_@!fq8^J!{xkCh<GR&BbR
zSHJu-+6eBZy!}tk3NJh0W2kA=+wXw>5xAjV^qOR?p^!2}nF!eSph2NA(X=CmM;KPH
zHHoorau->I^m<7AO-GX3pNG~5?4a=(3aAHmZ?kr6zp0jY*qSBjk&^oioif%6@Wg`R
z7XPQ8QE6?Mc0}PvG|Jh0g078Zz30OE_yUTRU9qwvj>2^Yx2nE~_JPi>wp_ixE7~MX
zLJwoDT}6fOFv|(;xQwz8$R1bh`em-;H=Gr_6VF*WOjzJI&v>SS_MU8}tm2l$dM-Hz
zp=FlJS}*{E#&_mDc77141gQ*DE=ck1XPKon?%DJJMJIM1InQGzV*vu5oj*~I?mD{z
zW9C+>fTbFVK0w^G)%8a>yoShvyl;Jm++em>$$RehO=0aI6P~6K@Sr&^XDws#r%hLl
z*HR{L7;ip!rvCys4L4UhEY!JCEnT=$UjiXIWM&1PdmFY0vc`To1(<{vyHT>dWIspj
z84T4K40TDlvXNtEBBDQ(zUMJ$LqyA%eF^C+Wa^H#olMa2S5>@QJY5GFR3~OL91J9}
z>5OQfcr|jeLI4P*FPME?=kYBE+&sohPAEmVH#|y#ZE)Cgzg;?70o`g#+U#p1*}~#G
z5e2V)>g$J3*s9SUGd$lb@(x5R{-SaX%t3O<?NF=U+*@`F?M=$@2{`O<+4{Ae!oNS(
zOQv6_KHD)9qly@~3;>E&5QUH_g5pz$ueMIy%ynqz1908(e5Dm64**DsdrKk*6v8;F
zwDlf2ECfv`u%u<{j>v&O&M_sUmISOGtrTt8!;1<5yoCLCkeW5)ML^~c!T(wWA403p
zD>2p(7zKz?xby;W=(4+mS%^mlyA-Vf_}~WQ=jRg=kKT`f9qw-8jCOm}(k*(SJWg4_
zZV6&ufHN)h)y)O`XL%+X-PC;>9k&o8&__TNR-aPD4@82yUz<Y)MB6vbN5bTMc%8K>
zNXNW#>!!-BqW^O(Hni;;BBhg7zkF7j?CMeu!iVN0Yh~X$HY+lz+rg^t4<H^qW$_!u
zWBGUX_-JreB%avReOtc?mhb`oOnzHHFe7%N-9v;lwG+ia3_*k4ljViFnVQ>)P5A8n
z82ea_(<)?-Y-4c_6Cmt%4=a`?g1(9zu;AK$S+J$aNTQ>o<FAtyFJFF#CZi(|ZWI!=
z0!A`!LwnvrMIh2QA1*Sx&zO19^!TrV$RB3=V0VL{CFnH;o}8$g3(_5~Q!o6uF(_l^
zEkNuM8Zbp2ykPE*dbGE^V-6^$B!+HxuNEZDOI|#&@f&OE+KS&*_Dgad1+OP-j?1^V
zs~&WrhqChY22`1)Oa`inO%~W&Vx^A9I<;oFz>$09rrvu`nY@(F3g=(5n?<p#_A~~l
zwtw-k_`~aoYxLGWVfp6cTh~bM>6ZVz0<-r^W-{Ig%jFl68fS*qGHNJ=t3*$uzt}EK
zOj^=be}wqF`5K0ZQa|)dz}!=Pdh;7@9e|_>lW>383f|Ii|7r2hspe%={WnTf;+(z$
zB|3utV1LlPptb%#mS6jWi{Dy<bDF(y&UMiELnY9@#9qBhR%&oQIRX6w&Kx1jh{zN8
zNLlC=!_1Q5gcFT3DmOfYLY#Kdj)s71?F2>=EfPU^B<|nXFDc=sYyaW`d`pGS2opk@
zzoPi3mby&?BAmUrZWGVRr50Be$n?J6Wx1&8exf}ii(M!MXa@{kGxYksu!a{C>w@f^
z_S6BOQo<O0qJ~B{J*y@#3F%TRzCwT<@fCm>kQWB-GJs?9tw*6Vj>&Q%0H}^szNjB&
zw59PNb0Kad8APBKC6N>Qzab<|qSB+PC#M?(?9~8PaSy5a#(!k?e=6d=hRG3=5$?uv
z8_>UKGZ2%EBaVSsvC^iyL~0*y=@OQC2PT|VoUtMC)SRF6y(3K`VUq<nG=x{N!gb;l
z;7_$jh52}{WKJOA$MFuzAEt>qOabN)-!(f(;5+P5j)Q~Ip@LnFT>x>8BdUjBwF_bz
zxjv&m_r{~g&4TPD_kUyW&BLkw|8~*Us3MXf2~Cu_sDwx|PthVFLWrcul&ScJ%#o=K
zAw!vyF^Wo)DWW7KDO09oDhlU*uYS)t&w2KB_H~_oJ=fmXIr~|E{J#C9wbti7yoURA
zzwQ^F4xx_b<tBg?i_`bKM^TD@QxG?f*b1fTnfQOXy9ir=0;VIy^G|*0R2G(|o4izh
zaMzMt?d59rY@4^_S2{_|;Vr&iHVh_nmEliTn4Cu}O5(~d%tSMK@L?%{R4hgiJ@vj*
z?)t#ohenwQRtc07WSW733D8H8y`czumA(5kWNzp7em9cjBHwlU+qy{R16=}rEOpgA
z4kJdXH~{Ki>eu_D{h#$9+2M1Gm;XKu=S2F-cP=^-oCd3b7$f9>QQc1RTR;*?N^_(~
zEd{tbtl2bn%tA%uyD_mNWJ**JUBe#$4mE+$HwKO_R%=5Ti&&p8CYc{t>Q3c60c&s9
zNro4(Gwp*23;~|qFmW}E{H}+R0C1Hn?t{!{W_x2~6p@q*?1>5}tR>e<hd(-1yAsCx
zrKo&VOw2T4KAcIcl&Ti5=c@K}n=q4*-dJPj?4x-U_m-SV2p$J|?nUwM>ZbFw;+~zT
zDgFvwb9Cn}5fC3Xd8gEk9vE~ojzL^jHv_^`IJdGm1M1VHJ(L@0{FR1ju(0A}+#UZZ
z*IOO^#{}ps5+_7ROuphT0<l*_5*P4t2e=uL4EK0X{gn3rOswk}1gr-$aWgZmk6}C%
z@vZnVVRhROnjo3%0Q!*A^{N}OIvMIQz-{*6T75Wn%A)n2sW%8G<PsoU>;@dfyX*R|
zor3p?tP5n`4j>9(xd>hLkGR;2AnLMtuv{{RdCRnvSH>vq@8ISOP5msxr9qwHciZCy
zwyDME8J4hHgwW`$fq0C72C-t6foO06a6o9)xR072i_qGH-Dn4OGr^3<L$N~+0b69i
z^R~GdKBgD-0P6m=5#&6gnze$HB!jr4F<#@9ywwa%1j<_#JF-)jVr2#ZJ^>b=)MG&G
zC6rIxSwN$7Tn>j2Hef_MnVHHaUHuOo&o7ia(JP<}9(N0J%R=zukU5e<0-QXVA_}lS
z(vla@8L6g&w=f5?*JYRjlC>gw54m`DhFH08-oikIz+WI;<yamv!5bAOoGZ}9Q5da)
z+Y@~G`2`WWBJx*Qpac}<;nHxWjgYgT=OUSaeGp4i=HT;g134O{X%yH<K~^hHoV2jd
z?$QHw0MR+(jd%DfbeZ=ce-}vj{$v)WoOE~rUQspznK%lM{e|eywesO`&k;JT>ZHPU
zQu9U@iR*F^8&4F{aF9r`97Yq*{T}HcZVb^00slys<J?g)10H7+SOFaR5O9-m!1eh3
z3$P8OUWL*bOe6%z96(ketwJl<xkb0kvBw2sttxo$-$KJU6c+o*jfGYfO%xN;xPWp2
zaRg4oDPiaKMxKro`!M22wNN}}mN}Lh!Hj?_2@w&Az0oQax5pZ`UEEtF1Ycq=&NeQ|
zUv8Yr1YapDIrrXRYz)Sdybn}d!QZOlim0K3<uZ?kLsCk@fiQ+Qh1DRx`Db$;JlYf)
zG=T?&ZS3xSrTH6G=GzR&==C~k71YED(D6Lk0|*sSN4DrBeVnO`C2h(#cSX?>`t_<m
zo!xvSTcEeIQDfnnh4`yalle;Db7H@v_Nh?e+c8W=>e>Huu?IqO8W(jI`B0hozEwV!
zmQFrlNd}9sJ>dH4kcvO?BeI(n9`EpMdbq>AW_jFV*3N3p*sZm`N+vaj2*2{nLkwD2
z(=c;s<o+uJQIYv>Ge?Nn%z`yo1c=Cz^Ub~hHdF9ajSwLGm`dF8W)T$lgXYgOMBw&l
zM$9IAY-=!xq*FbFQQ?92F!P<}rZik#X9(JfcRY-66G!5tb3N9h3)!Ue3l<@K<x%pL
z5md9<Y(zEBXxow*p2&dW`ajyT%%{a`m+_K*rsbp#VrsT0AJ*;F`HzYMgjtYgI!Zc5
zkb<8H-KpQN!qK5{MY9}76pdkG6&c=(3;yRnuF-Lf44A5;y_+!JjdU(7CrrKWg_*L!
zSlm2Wgq7cdt?h}Os`-i#><Gqmf&laH-xTpZgzJPRnvnSHbd5Xi)3U`~^<Kl`I~Mbo
zV04`N);_a=@<A6kfdCid-2=gudQ)W&Pzof&na(im$ZrVWT^z7H&XKTJ!17BP<M*kI
z?%f(DzNoA1w^3t)5(EsJVwp)}gKuTs_IFv99#XelxSO=<AK7TazsG9}T2T&3&qL&X
z&7umxlDFK9YCrOqL?|D;I-~e$A9^Us``~rx|C_X8(&K~II$FC6{;>2v^>I>io#`k#
z%So{XCE>T|V%J#??-A08F3({L9NQ8Y3g-aIiWtNp->xl5NrBNiZEa2X&7}Q-0gxj)
zeZO71Qu*&UgnVPla_YYg^Zjq0!7Aeb#Yd796k&Gm;Jvn-gW;Y=ZlG93WbBYls2`W1
z9O@vIjjpW)KLtSpd&`Y|eB!+kJT}r|(p8AK1R!0y5ab+YPXBK}il(jVfA0SO*PZz9
zT&(DrF)X=HWg`(chIQRJxEr$tmx_=q6X?{HIntAZpi@=LfkGMYu0z!9Ohi>iOn_px
zf_Mi1o^JaDLOB==_s2S08I5x`v~yZ#gnB3D(CJE=ArhfVa#ARYy5zRBq`*dJab^RK
zA*4v!KUqpTLVWg;>Ky<Es<<BhN!^W~tt-*lx~&aOjx8y$%~?i4eEuxZ!`KolqKx04
z0yk$DXrj8+@<;>AK@y36j{9?g)=i%N3X|>#cy+^Joi5&;{~WoZhdQFYOg4}<AN2%c
zU44p4aT;TS^+xRZU|JL8yf51lX_%$pOW<Ss_oIy%+2HY!w^~XVR?mP^4tn79MyPHG
zJP|Y|beO?N#FAdYh$CpsYO1Rfd>I5XBw<3=5wV=V&*p54(i1QiYA8$IAX@v{b!hLO
z%dz4Iat*@P%wT4qP*`bwyy=EmI1=rp>ylCx0<0J+6nxfc!OK!qNf{yuWO^m(7eq5B
zdLwz9E%U*8?Dx<u0MP%K$5TA%-oHOaj{(Udcn>6lJxWTHt#0`3!?bsL43wLodTOqv
zMB~Ovjr_D`l^Xji7)UbY=Z5QUeC%z~w`1uCN!k$rdj76ZXwXKKU$+HTspjtAzeMZ}
z=!UI>NQ+yo!;>h|Rw^#_jf)8~eA2miLkn*fp-uY?)}4tChq2Y`ok<!$J5lul9dCDd
zQ}Dg-@S{!uav+Pds^caqp7*vR2<=^BICX!LNrVoC@?g(Qgp>rf`HH7`Leog{MYi6L
z@>qyxjUE=Hh4zRw!bsUeI`i=*4;fh!lPueUh_<k(h&X~j(u9j^Tz+L(oAme**?!2$
zSEMfWF)c$LKn4$>pGiErq<Km3SGe$Gi0F~dHj2-?2Y|2h0qLN{Wn&$Y2Eq&X?AGF|
zPnYQSURYq_Vp?}daoxJyD@E@sm2$p($~!jv$x}ZsxYQ`m=(kJCd_{-txjdH=K|z<!
ziZ>POj^}yatDwm1M95D3-u~D=zJKM7h=?uU+e~|EotiE=Ui^p&m*oSrkHP;$Lg?>b
zf8G*ABu4(e{_Dqo*WrKJn)omiu?k!{lH19+PXHPD@#DvtGiS20vUZ((<c4w*a`ydp
zA4m-obi)S^9}1l*K;pqE1|Xqq%KCo0YG|a&vy~HD(-uxl_B?h)k_I{RhiJ$q7*n|;
z-?nXE0Df=Wxbeb;3rv{Sg!T&XWsz8Z2X(;c04~hL_L35EI%<~5$jE>?D29#?*qC7b
zaWD)gCIk_|BS|Xei;Igx@=ip6$gM)1*O-C*<&Wv<OMiadudM9-6^S&iyV7(2eucwH
zGjr2tj~_qIw~vK|1>&V_n>LZQAT$pIT)ld-!nwAk=jk2E0|(F?GlU4|4{9`+4XZfp
zMth-|<_;2bG(HE0hc6=6!`+%jEN0hvLyB{cxOmH3^3tVJztCP7em&Ox*c&+%l0=h7
zZep^svMR}IDL(t@X*!Nlwj~~bPmv-{j*suQm8YW%KqmsQ@-6)QHUP#!tMm8YrpUOK
zWA~vQ0r}J{O!li8^W}R`L0)rnbNU2T`Tj{TW(avGvd~Y3Rv{${3Zfe!420x4%;<Sn
z*Ck}wK<>FvjK^$8t&=52dHOl6qN1W>V`Gr(+2^)jOzZ*@f1s^v*RI{?&PJg;QoH#M
z(ujws*s~+a2j>B1Y)nvS8yFlkNK;<BM|s60CzSk4*#P!C?&Roo!UA?8Awr%1l9yL5
zH@#hacz6?#I8eN7Z_k*Rn2_cxpnb5mK;pX(pM~RN+DepCw&bH@i>QbH{8{F@r62_k
z;#d|{sp<$SgJ3$x9{52KjL8Oc2H0m}YN{B&a|VkBBmqq6`-4V95J7@st>=v0Wv%6u
zUZ-^UM~b3ew%#|ibTh2F^cBll6WU5~%0f{}_u-Zkcy0*$7ET#)adDxE6bKl5NxY%?
z63Xlb+LkEf=jQ{G<)@|xYTg<jA4g%7008h{_zV1FKK1BK!=iq)FQG{B(rM}BatL;e
z`n_u(`c(O_IzV}8e+!@o9Go{DM-5(M!AGC}L1pqi<LfI}N$kT2Fx6-{phN#GMefbG
zYL!8$U0lQ4g9luOqNC2dqA?Xv40AMBLf`QYB8Eh{N<>icOlm`2ApD>#yKT!BoDibT
z4|2bnnwo%s06?-zw6g<YAr%N53se<I^$_2`e4|t-T6O7WH)}Y9V{KV36?k16Bn%}v
z?vbG(BcP2$a6wG$sO@z`GJU803SvQXB`p`of&nxqA~HhW3bCo^_;}9En>X|D5H$~?
zxAE##i0JU(;D<X}n=INg4{&mGb7OUlP=^6PpO4}VM)8Pi;Fj;&z5C4vWa{ImKPRC-
z*~eP|T@j!`W_%*xJthuVwVad&)`NgRUq|N+>g_;JY~FJ$9!1Tblz}1Oj-*2=nDxiR
z1U07>CgcA7`yiT0!D`^k0VLOU%F71`3NJ~&YSsAk^z`_6U~n+1B%eqv+pc{*8(Yli
z7c@+<O-1t`t$y|DRegO9b^)ww*dQ3U)vH%CZ9e!2D+5dic+n&*h0}^%ywb-ss8p(>
zW8v<~gD4N6R`|m4DMt<nogF6M@fsZ+?d$6s8R3FO;pV=CKD9S*-XMT__569SmlDW8
zm^w{*ZR%0?LAMvCI_)wjFau3=&c)HZiIkDCc?eM<rsd@7dMz;U0m4eyvcrcDpR%xU
zS`c$O=Q+g7$7h0NMe&^={q7T%n0@N+vUKTESVo9~2y{m)H7zUmioBmlSjP~4LMjM3
zEMaQ8s<H7$m}FSk2Z(fmF!I~KdK7R9?~a8=1OZEBnN748U?-?|Q$RE3e0cI!XlUpl
z?FDV7m2`C4Niyyblz}GSiZ!h3^5hU&0|eh#f?&r@!G=P43Gwr)lBRBouoa%4y7(LR
z7!M=yeF$b}OGQ5Vb?SE+V1toDp0>2K^sIwmh8>6-g_oogy?BvAVh%@?k&>6!AGGa_
zj24x@73<Vl`{?pK!Q@cGish3?2@lscGKzP2`uX!`)bivFlH%eFv-JeB-eXm6#KiC!
zgF;vp$)u1$?W($T;=yOch#5o2fia@Ak@|F4LsQfEfy3>xva;F}P3`To;HAB8XxQua
z6HQJMHYx|#Z{_pUzl-uX9V1u78QY28+Ky+xQTnW2gDtyZ-yPbiJB*{8D&h>1yX|%0
zxN3Q1uDJNH_;2?osbQO@yKP_Af#2C#b?KqrI4&iz-;rxieN#A^<m;<oXO}yeZFNUw
ze~*n`mL4!lH`JHj?JUXK%(~>}4bA-=6j-3$E6zB7*0Qp<-xYJ9rlBEM=k}HMHk-rj
zAD%un$C<3|z4Hhj($m8uo?Q*MeEs_Mg(fd>>wEWnQ(lpa6F3Q8Ox%TKwwO>V@BEa|
z70-3(P36H3)bEAYHt*Sny}aJN15dKSqzyG*znIaH5z=fqHRWt|;z(Mv4crz2UCZkC
z*umI34pPutIm#cdlKdyB6&;P&^oDdyRaFDi($emiYZTdQi?FP}fyaq3vH+z+al`#L
zWHnCkg%KIf%*;$X-$kJLePuyPOiKPR_k0yCDJfgBlr%Nca`duMhr64e9w?fSlr%Ox
zobaq^=@SWPR$A`fk6`eL+@JY*_%+cv!ni!&U3IJe$EvK>AuIv<bN6)K?p_H!2DhH-
z6>@TN7Hy<w3nrFTP>;Ve9){pViEYgNQR^Jl>)UW^CV&3iZ(o$Ur{ejG7u=E#pCMv^
zrC_c);PClS+)5?6hsnvNuwIt2L*0)b7wq1*@0jGLx1T;8l$68=_}i8+v+N=DFbvJW
z#~-f>d=H$st@PXpiyPwVrJkiQmbY);R&S}o--31!cJyph->ASYX=$S&k>+sw=E*~2
z$(F*ey1N%)-Qaj%m-geR<LmQJ=VN91*02ln8!e3YC=}frMj3pi)BbOnt+qvrNN3D`
z)1u@*X|Qip`*YC2+Q0oe;l!4prwB!^XaR^goX=7WjmABD_}~G5t_*5)a55a5jU@2C
zkVy1M>racoT_^#>OJ9$Xo_B72v;fJb`l66t<`ySETz-YAZ(gZEjv^&ST}ZB)cl$*N
z?=pU%Q|wrpn3(t;nl$kJeIdp6-Iq}Yp?;6Qpi(nph>!&*Nl4Au7P?2K&K?&8%FAVo
z_@V~bZi#oM4*c^`uGXLtMV8+F{yPPQ?^rcs53;g3y10lyeRS}|PVJ<<zlZb$TVaP<
zT3Qsl3#aVu;P+_iuaw11qmpc!a^<x9I!V?1us7TQSdSP%#-|a{Mgv=TCO8r&huWQu
z;0uuQkTwm(Qm}f4D3_NQnZVj&I12g3Fq{Znt22Q1$BzmxufudugJLiI^2%Kdf!EC2
z)U~w8R0RJ%n2M-5?o@o(!Gi~@9ebY4kWmE)(o(~g)6qfIxflT~;sB&uEk{Pq>)Zw5
zX6x3ibpa*`GbaMwe~w_l7Df@6uvF`8yHRFcO3dGM?08QlyDU?0P*kYD9-w+KVX7TN
zc!KLlU^n$M2;GsxxCgNY7J7Q`5LPR7pYl+k9J*}TC1pj~68@!r+4vo(X(SJN&!ag|
zq*%Fca<QYc6XENzp+|%cPFfbwHV1D;!gXYR@K!djdwVUL4L783YxY0~9^#=(moD8(
zO!T^R2~Ob8=cmiE>M&1;dA`8h+#L4wwy&@6ABokgR;g-fu}N&=;!2EKoB8$g=XK1?
zTaA11t>%ghK32GB80-`kJ<_jo1%4)Q<@SA*E)EU_1rzvC24(Bc_I9$2b#nb^*Hjdu
z_m+w1{>B;BPwoslAx}Pf5EZ41&<@3*a<Iz~ceJE%7kxMM^!$ymUtOsL=G4c>2jwm_
zuk&x&a>#kl(5jv3r5|VBiMvV+A;k?kl9AVJgE(7OMuwp@w6Ag5d+Az(^XHxOe*OA|
zsXCw$yazJ`p%A*HZs}?DOfm_3@CzPKRa~7F*WTxCy4M}Q3kR=$iR<FUi&1inl!O-_
z2Ynb2X_-N;@=1dF(310*MiptKn5Qo!Wv}|$BuD1h6q?~Q_F=+`qIo^&5Z65pB@P-K
zjgxr?pANMohM^4t)LvYz)M;%PxW~7O*w(M#W~c95<;c=9NZJ5QEb-WLtyO;}G@5PO
zr0k3l`_gm)V6UQsWQ2t^ACLiKJ95AE_s2#@^XAG})v@}<!(Fn6S;CZ2%@M$>8Z4p^
zwx%rUJvQC)VAdJdv38huRV74d!sbQ5S|hS|HJBKk_p}wzG_H8?KrcuC94cetP2Z6a
zWXgZVAv74~wP@b!VQg#+$GUb%!>R?w;Iz)$)PU5(Nl8gmxj%k>el@!ip6#8(wdI|@
zfL>Uej#o>}>S2S#X?elt&4`&&`w%aca|IECuVVe;udhtLf9}0Ub(vk%*hWv63ZJwi
zG5a?HKx&j6xS6{GqPAm7fQrH3z`)Li{)9uer`Uaij5TkzDe?VCX@+@Uv-2_gtr0$1
zS(BR$G4YumB#<x%FZif|0qVlI$f_dS?99<2u|`Pfu0xZ$?2T!pdPqN+W1AB8YunDO
z-kM{l9iv`i@rrTY{uJV3z%N9K&MK)}=ucbep!<{Hog3?CLH@uUJf*`wJ>IaMPn{#$
zU`j`?y0h~)TI<js%v^=x&->!6Id1w7PVy8Ro0y1H*itBVG(l${Z&leNT05Fz&Z3)}
z&j>BbYqFH878g9j%#5sK)Q>?UtuMSDN$wRreG;GJNK`w*65Tr~8z+GbOBMzM)VCyX
zSaDbp#dgqn4ylIuHo9|mJ$sgh2E*zJFe!1Hjy&!KLkjA4>)w;jVk|~RM>i4!)F&C%
zwAz+9oZQf?5<d2+@`U7u(!u=jYu6wRSE$DUp&NFQ8`8lH5dQ}H&YyMWiLs~t(9wL-
z)2^={Yx6{Wan(vkZ_oH0d-poyo!7M4#7_E@91)LY*|Z6pX=FZ&Q>WN8Q?24_G^Nv^
zk0cPof8dDy?PU@7Ly7~*@BX$5#ke2t)f7sv3oYqd*sAGc8)vo7nK4G>gnltv9lugT
zWRUM52dZ|&1&4nWFuwPv=vYR-J#*N5#51UT{e)PaLk}WR9pGeu>V^jf26}pW1_v{F
zZzk^SNASldU0y!;d^pvgk&zMiAj(!wKtRoz<Dq=Z$F{cP&e1u8W;2M3SJNZ)v>cOK
zOHa?T_8xvfy*^CO#B?M6s%mum&k@9WB1T&MHaFbs4~^Mtvlycf14$07a1vdv3!-L(
z?1+EN;cS9JLRezyRSZr{=|mp~u!lVX<3j$|dt|Sq3mg6*!AeOym{nlgwEn46sTpmI
z6dx^GHapIgmFt(Boa_*N>b#??D|@ByK`E(Ow5f{ItMe&TR#vj*8y(c$-dZ&tKzH%t
zPvi*q42zTnkKOfqmwTX2AhB8LEb@fZ?dP;^wq>9<24Qi_d^@D~=3m9x&VzM;ELfaB
zx?}pz(FfgYSN7KzJx_|6>lp$#Llu$n`laLST;QNo{#Z5)fk74WrPbEd!haq6a*yfk
zcPBt%uRh%K`D{IR@C2a*j(iYn-ed}p0@}QLnoV~#9O*BMKtA4g7<uOhMq2v0i14Ap
zro6K9(xF3#MCsC+o;_12mg)TALuHFz6x{MSI6a-W*>V5={l1YA5fMn%gp9NNqcM-j
zIru85JV6IVCves=sBc{?8X2^lvrpy0;b}Na+{6T$!IL44gM}W`w&pWhwrvT@r|q(;
zUWIQKvVXhYQ-5!ALS$rQT%15jpmG|sRw!pd-HhHC!~~G`o06{GJ#ECI{ou8bs!rbK
ztf-PjZ#bqC-3-PS);8q)6R<I$5G{GpaS$j1I6<j>&CSiF<2CCj6h{`4@B4WB1q5hR
z@+Ec}32op0Y^<BAtNWVg<~zGR?bmW*vbVIR+|P;dh|ZM1nt@dTE9%$-<@)QKN<`Q(
zt6c^{RknJ%MybnMAb)q4dl-ml2~t?jM+x}S(m|6**RAv|h5dDN`L!bqZJMes&l?#G
zj*JL%Ro}C)w%+MwnG<oW^lo%IBH+M4WNjE^>QWai076YZ0_}H2=Z>)-{QH37))xND
zUW#1gSDyubi+pyQ$G~ea?n%2PGqbtlS7iWRxVZU)C;Bgi3Q{QUOaDG#qMY|ZadGNY
z9xfuJo5gQ@ZH8%N=AQWW<qH)Y+qe~oAGaIvu&|^#w4(Yf(PQE(p8}=@Yu*fSGms0w
zMC<J8>iYEQ<WJq_h`d;@-c7u9>(+wXIM+~T**(oQE(oNm=Z~QeAAv~bf~>HKHLf!W
zu4C`sxzoSE^}OzGI)r1{Zk2UFOh++jJv9Uvh|!LsEQ^@YEPUz)-{g<5w1`lU)DX#L
zU?V^ok5;W97t}5CzE$i2RweXFUqs%V8WP~=2W4H+{oaW;W!wwj8IFNvTFwsncZn+m
zLw+I$fe*!G!7OSP7dJP~ioK&F??C#UFEN~=dDJXKKuXDXZi22f*u33;7+tsb8y-@f
zx;PYtr1u=MwY3HP1o^8cY_6Z*@_EA=5ixE>WCjht8Wim9zr!qRRBqb5`7tPt=CeS8
z0D~!hw_C^kB%iNh7#+x`tsj96tKB}JTX)z~-?nj0TzQP;0yD#hiJ843ZZUDRnB{{w
z0>HU<L=q(Du6DBM+!!)%Os78=bmCU}^5x4xJW`&@LmG}l*Js3w+qmlQZJa;@DN^Xb
zDiSo4JUCB(z){sGfN3+k2r!!BciW&xQOY7Crkz2Zhde>5g0K?lc|lPl)qchD<%kKi
z43e&2zr7e*_czc|H5f&J23NpxZWc+!F^(k1iMIteE9IMu-GE(omYm)BC~qs^TA02Y
zwt{>2>KPcI?ZVX5)atrXaNkVsyD4=bax5ZcMMY8Sv#Z-jX7s942bP?@WoZ|O)aDlO
zTq4o}1hBenFXw9nFNBma`6^cMI2;bbqXSY>K~>-)vM23YL#YY)o9)yCwSe5H==2!m
zav8ZK2_NetgPO+BjyYg&AW0lG_=F4!wq;XmbTpEEWW}7U9v&X>?$qK1#CTJiWsOn}
zEoru;%Bg}EQ!{eL`S|fR7}mJ5*TAgZJUy|IJUpVBe4cJ@f#-`6MybxbrBSo6#XE(I
z%gT}^&c#>j0Kg<&N@xb)&%GDXH!}h{g<?G@&JcW9WsZ2)`y><__(XE-+&l%%n<stP
zw8bUr5m3e6xN+RZ#s*f~&pzYsT_!-gKy{V(-DClnNC@a|Zok08;xpRH$7coZb2&SP
za7ZW)nX5xd@8xCrr4eNh%F79u(a#$Z7gr9e06yTcdndbIy;8~3-yo=0)6{eipdC6%
z8*6ITiB2GWY<59onY=R_R&Rgi8GC!oSv!^$f$r`P5c7kTwzeKX=#C6?Cmd<<X?68r
zc&rK4VmJXgz*~Uq4oFKA;YbYc!|q3A%UIT|^Q{4{?cWNx!B01bv~UBuiL^D8d=^^0
zRuXu1oVIl#fbm)`)qno^N1d}4wL>WUX&!)F0%Ga)CcAg<);TSY00|@`P|jt&7mU-I
zS9x!yUj7wS_%W%ZsOX{7z_oY%M4gs|0Yl_PhL)n$qt@;d;Q4`$nRg!Zb0s#sFar98
z$NCJXWX$rSLI;1GF0V6!fH>>>4F{8-VAa(#@O<wOWhoBt<mH8(I3jr2zfhJFK~q`T
zMI0h@QWN<B$adCvEe})PKIgl}Xa7q)c%SytLzl^%3r)=QaQoVtdGssrpVKn-y)t1y
z-vMXGPpU$V!B&s=WpHRHD<hQ=;YOYpj0KGA@7Y;&C?O#@4-1ZrR6HypQCZHhH_)a8
zyCxvTgKEmwiU&3)HGw<|@$o6So;`!x0Owt__R#6?a70X-BythDAo~HtiTo4qg#hn2
z?m4ER-7|fQ^BcqW6REmIQJ1rN;MjKhbJ`*ycbaKVqL0_29)Iu8^a3Z}DAP(LE3J7i
zvf>CMW}p4b&Nv^s>T-Gp1?YkcxKw~r=8<mn_w<;L8Z6vLFEJ9wphn)5H!!^EJm%J-
zU=zVji1B|cJ`**_G7IV^cb0WU{85DB2pmL(k1wP=LYt|Gm{@abYir8K?(WA(&#03=
zuU@{yyPHh?lR9U~Bbr?V02ZI*ih!`XjMO$joU}Vpu&#cz9mdNXhb;CcpQ+WePTkco
z^%Kw4cHA&T)Tsb`awQXw9VU2C3OM^oiFw^c(EgDf0+Ec**cYa35$g(;-gfDtqV>RS
zF|ICZA-C@|t=!pJMqf{WKd{pM!vOa{JO4%KQ4HhLnGy(uP-IY0(9xE|RB8*hO{rVF
zM?P&XH7yMjLo#@;B87p4IHvacqh+}){J0d3cVBqcM;u1@C02d?_MsWQ8ko0Da<#n&
zH8uyA?%K1*33`+$MMb*%p<ZP#L$lP8s2h+;;9y0nOFl9xN4$9nP?}Wo7k$JUX((!-
zD)hRpPTck>0+zYozvV8?z5B2adHguCJat>bBZTxos(nUA256FE($dnM%0O#LCm*t9
z&J7v4AchCWqFvF@eLHow8;bJ>BqdcvTu>;VZ6ajN;^Mu0ODv&DJ=Wz33}emk0At8l
zI~I_1+N#Zo^YU84D5uKBs?Oj2&`*0Tg4ldG2;U?z2EptlOO_B2v-~o{N&pd<Du=Xn
zCyW(#0ykq3fa|vH+hcb9LZE=ygAv7+rY2zh<1nE8YXq<^uxh}Eh|GUXuN6*lH(G0n
z_Vde^@dXti55hSwg2H}%Rs<*t_^7K9cr-LLL}z1dxV^m>AyHbBJB$;Ly+o><yZVAU
zAPR^}KPr$B5jh7IV_heSkPjceQt${GqR6kV8>y!?lcJ0G>$=B6cIS|I1D%A&%9vR4
zxI^JS(00Ttx>cfm!WGs8;Q}m;!ll|VRZY!AgcGibcIfx=W3!|tmzK-2A%Na;LtEVa
z3Ow1S=$kh=xVQj=3?sjd|Cz`@Im83_0Ksy*CjevUke~pG!ol!<!1{Pj3_>^(w4M+A
zY+z^@Ht?190!#{1hIP4Tith|jY;mc+!~@<7Ua+dD+q9yBgh1+DuHoCx&Q5fF=IOtB
z@d8UmDDOZmAxRbE&2VQu5^z{sNJWgG0)qGi;MNd|5nsNb)Zqf&1q|i+PmiPo1qC}@
zdE_qRCch=bEkMDrM^#l-fbs*cT|>b^5=tnlsx=7p967Qh-lYoxAOeDtG2x2B2cgG2
z_V(^M0l|w;kG&tiy8;6Z2JtoEzvFq>AO2)!r2&Nf7!lo!k3eF~wr<@xT3FbH^c&zd
zy1G#Ilob;bL&*`w4nF0fW1qT#!3Vf(5m|=!sh%@9xd?MGGNruq0+3v|D1^hvgU@wW
z$@22{0z^Rk4!~$;W}B~HA-90NZ>O#l%!T75Z`4kMIh_V85SYKwUdV8eR^xU3`t=Tl
zdDKObL#t|Qm$`-_hFW)EtYImI1!m~U`f`vy%s5B){NvvT`fV%r&Pk4^_XJlkWg;J+
zs2#tzw&N&8B3O;aUF>zoC1n?ZJy#V~x#D$A0h0yxucy@vDefmgxIr12PgwiN5!A-q
z#$Ne)h0uP3?eA}tW&(qT;raIFw(p?kTWJLN*vz|8QBmW02n5vnn;-Zb#wSyJR6@xp
zzyc+BPL$yA%rD^lV3}NSynuusL`IF92LR^Nr%wwD3&<CS+l%fRiA+@Pzy*RCD&H*e
zNrv`Tf9Vtn`gj*`IZ-}8Ji86+guno3L8-A`O@Jip5=u=O4b-7L*@QV$7+QWA+xwnz
z`7C$9HkdL7Wg0)j%}K^%$_5ELp5eahmYVP6U%Exn1~XB{cRavefp_%)&g<7Nx3jOV
zc!klPp6Opt&%a~GOGGMYRl}a6Dlmt9e6*+MC}3KEaEQsCKYxy&<B)fe@=tpaFaQL_
z?``(Jzs705g%P#TLu6<a#V3p~xnV;v1=N1@P(=3yevGte`VmNm{n1{$w1HQ&KOXwm
z_K^ScUq7ZT3FZH9YZBSfv4rv@;bm*<G(y_<M6;RWE<7JRd^N_W49MYv)X#%jRgg#s
z`3Co^ai`r^<6Q3E15gTdX(PgpHRJO1{Bgns`Dc0IGRmjj2)^DkE~&Bjdo^fFr$&Rh
zXn%}Qq#efp`tg6WCX)umHrKlrA9>BTboBJ@f_pCdU4}Ze!2y0sJBN&0!NTLboSb3g
zE4kdP5uSrm1{-eC^&S-UI+R44FV}kxTprMC-s}><6lQcR=dm%F)kBV|6igfZhd>cc
zO{m_scXg3YW>AR|fu$nYcieLm{wurHwksKNE80G5UsLQzTj|2@?!{7MPH?O+nJ}F2
ze5l?a?IumfsCDLYZ;1$KcCV3pFur8>283=AKLHOO+wXT9R_#3_T|pt<1=Jw5$LQKW
z2+zWthv#!VAO->r;3V<00!LH*90XP}8P%Ybowp((T0>2(c7S;eLF~vRjH9W~5j-NG
z6~(Y5Sg+#ZVqok@*x?B;B0vH~APb5_h?(|rh8+#VU|{t;{gBIKo|d}EF=VqKQw-fe
z`6*e->E(0nUn`|9%{fD21BaKizi^{#9|0p+S*jWuA)pU{!2oxR-tI3013%H>q@=8D
zbiiPXcIsrrW`~|-W+%x>c5lVgJ6E<Mk~BKNWuRhY)B%-<DGgL!_9Csd=u}fx-EBkX
z7WD^k6Rr;8Uxm^$SBNu0Hwtj=x=rEX;ec<wPoKWC`X8-C$>tPGJj{;b$(JZxa~*Ic
z{XW2e;?BKB9#aL;6`+1+$DZG*`+fY=BObX1|3-z&PyCsD+G!Ts{R4Ic_(i&@IE2NZ
zBa!gt9qRh{0Q)-+D4U38Dx#K*J9msS=K*^ibQ^zCY(<8O1Gt=zMEquzAb(kxkB`p<
z11(X6uZYBxz$6TyMu5?OsQ!g-mOJR!TLTA!icK-wdKMPA>}iNW7zkLB#TkG3wVLk9
zdICQ9JQ5TV()hI(+jceXy%E+1aEv1ttJw+CI|t@MeaT~#sz8qEO<&QoID!n#vBm7u
zKFEC`NVPG|X&nOUOJ-vuLxGn@x!1+fQSfLgfsv|^J|hI58S6T6VvmrJ5OU-FuhN&f
zbKVNr&$Hnmh5azGF1^R#>BSxhJx0ie<0@YMl!#0rj+VMcudk0>W9f=D1_#zim9zIC
zkVH>_u6a3{RS_PBb4g9$p>fKP;o^3XSz4e$?arPjL)OtND#cI;@B*(nK72-PK-vLd
zKM&8m-7$K$#HbSKInW{`8^QWBAjL2;^6WVsns15Dd$bE(JimNR?*(U`Pj<$?C<ctg
zy+<4ecR+h3pOn{NzH%C=t~j<VZ8!yjAm<8%kKpU=F!n<5qL?|2<Q{CPO{^#;lR_2>
zwa{Dj$1y@E38)078YnRb4-YBVVm9N4PV2>eTZM$qB1xQUsv~@otu?@sAXM@xr4N@f
z{*B(Jn23lJxfAD6K|mN^XhI~hp&j_H>N5TSR`#i^Uq1U+VP!XAWn&evvJ*EDF+((m
z$kQU(b#-zQf}|04TJvV8c3LwuFbWWg;K+*VWk%e+YYk=&6eR54J#)9y1(gQR1NB^E
z+)A8Lf|LkQ)3HG$IL25BpeaZ5`W$@a5CkIcKp9!&czz-@%>G%r0z5+Wb4jAagfGCs
z7G&rer<~S~c2b;3$a8@oc){A*8UQ0MLiol#D9^mhbaNDymi7QYYNbD-`9||1U`f(m
zzZ@7hi8(-m0g!eHl2x!NQMyAY4+Dq7Dahwv(NX4t72qme#8H5fC?ej^L~Tze<q+}_
zGQ<LLeP}3yzX)#qX|%DC2o_Q@xEjb*yu7?%lZfmW(k`4e!2o8X8e}|3_SUY^K!$Qx
zdvnh7?Ur4uc9~R6p&y!qH5RxBKwk{nq-p?5IEN?_cCEH_9v=>4p!stEjO()#a%RAT
z7}Nnr-5gcu`NRvjw`D%96)K#_Pt{{VXFh>MO*E(o=`qZ&q2P0_w1Jd}(J3xHXCBr)
zo&*0R`JREHE)eGX3f|F5R04LnoOE$=Dw^tm=1A%Z6C8u|Y!Re&$3;-57ZW4xg;aAu
zE5h5iE12sdEr7RRKL`BG8`e`?i!|MJ0JMJ}y?7HjXvJyrHjA-2IXPD}5{DpOdfn9j
z4T4+#MIR!o3WWSlMuyeH7{|h=`#4`C12-x=e-2naizfCE`5YjgjThq@(IbxJ|6H6K
z-r5zptbc;^fL~fI4gn4+Ky2~N^7Y-_azaA&X7bJUrf@40IY@yLB7Z-}<>Y2(-|_Yc
zzym)GB}GLWxud`hl@eeHhH%h&xBmeKY<a}Sh=+%V<kMPOuU#km?|``ui%FjbT5MK{
z!f$W=a0lGNYEld85)Mm7P0_pRIw0S!O3yNo+pG7DwHNK|e1GcHDZjf9Vb~M+^!gai
z-)}PnZ6@Qm$sNs&T|>ojtbJ47F{>FESoI6)Vs0e)9y$3<YSq`<lPOUNx9B_=rJnjX
z@e}_0%Ew5*KZY);IROjmBQRmiDJw2s2yn9Hl1Yqu2B+95zsNfa3%~7~pGQ=oTU7N@
zQDNV%XLFl%iTqmchNhmC;emlV9BM%5aL)RCYkNfBqQ&1I)Tz=RsP4m*9VuQhe=h}$
z#k%wMmfDhE<#FPQOw^|m3dSWqeg16byCQ6UM`Coe<-5r`5IoU;z4On?e7<Qb(zO<A
zUy8;Gs{EZhcdkKw#N7|(``zXFQf9b7_LWE;<02x=M&B$Xw{91P`q|Tt5ST8P(2aet
zueeoj?A>ukIKy7{+Femj%L0+)=@q~mweFe9_w_mF;k_$d+}+%GVc<$0EO2~)cU~i8
zARBh1Nj_$S2P3C6Q;Xx>$wN(B2Of-X^xrFOCcMLCiH}#)^1{x4;{t32asmU6E>_Sy
zn3$NtLqar;93k4g=udGrwTCNk0gh&uzzO?;Q#>*{ihOJ%56=Qj20{v8Civ-*Ha!Oi
z2OXV=6>{M#K0dr?54^LZ4S^590+iN}`yJ8N)>c(T(stg#VH~9s6Jz7(;?wzSKfwyZ
zRe|p<j@Q0@`wkvl0J1{}SjV!XORI^@cqqr#Zq>$%?O<BVK+CV8K{H0ZZ*UMGa0+T}
zDoRQ~t6n^RKGOZz^rFI9XJ==CqZX%5vD&34N)rx(TB$`^I`sVO&1~yYpH9DfH{pG}
zdIlK}*j_Y%=ZM*biXS4JDTH%^rj|(eP({0OZ}Qg<i{{lQaOF^OiDtO>b~O`|{rlXw
zQO$8O48gny7!f_2VD#^&fraP&;OkMUzAtZ%E@6~sfW+*>m&LoIMj$OEMN$FSO)N<D
zqJRMYGZ)tlhr<9CdcYF;MQ^y!$Y&Ch!9X_#Ptq-4&hh(KJCfILD^-E?%SuSRPkb7;
zOqPx>uU)T6UTHvn$)qdKpj0mx@!r;>G6yNWYVPOvR;Xu8{P@9KsjahkMuKq0TB4((
z-qKheuDvzPT^E&h?!1dgV9oGNE6upw6N%l7GS-b;=X7UgW}s^lpjYzY!-pqNp1>5Y
z?iT;LGCVYt$ezLRrB|Z_4|n|2WpvdtiLati+lU=&>;{Cb5w_tsnVeI02h&^)44yt(
zR;M_yC=1l0ls@?KJW;-w+B|l*1Oe>)iZD(c&`0)~Tlrs}pFVExn`%+VN<0LZit@fN
zEE0JoSUG`D8tl&f4I7)aIoa5({$7xGa{faM7+3aFQT-@liwB?{c-4b9{l+vM)xC;k
z!aIE@&{urccLfTEuf_dzq|SyZlj_uY6W9vz+y3ZO9yF448&7%qP``Yl#pzA0kSv|K
zj5Ckovo~V)u<{Cx7kc)vr!_CJ5F<`|#eoA4Q45&;{hPj<6E6DvmwH56{2LXE%$_dw
z4X8hh=EK8n+?pC1EUs(M#f_P30lLI_y^-Dj?OQn_Vqj`@WwFoJE@1Y@D9CR8YAT3;
zh{6AU8eOevUNTbcR=Ce<>gv=iJ^0ecV09oOQ96#UgWX+we!yk@>aAs&jc^<UwZ6I<
zfk6Dn&%j96?kKy3?*cSW%Vr3KY=kw;u^WW+?V$6AVO$zoT9w|53#gkaTJN|eZWTO1
z=ar?GjT+KPG(qt3^#S~XMO&}0s-`C8JQ#6VO;PbVoVNcy^i35a8mp<PK`e)KVdaVy
zt5sg0BG5S!Znp2ALB%!TZwUY-8l&4@H^jVV^SNwl79@K_<XT4Yy<oPw;ZXGa0|Qsh
z$5{gzL~r*7ZSAH+ziht5gWGVq@zGND?K7kt^K%Y9MdG$U8KZnsMMWiH>kg;1frtSQ
zMnP$&H8*o$xt<CW$GewgNV~{52*d`|??XvPV&mgIzP?;Ldj}~N-yWea;5WK-D-#zV
zA8+eb@c#DiMvK^4_J#*fGtjwk4Ta6$9iMZq{gXIS2$&C>bk-CXj=B(c98v-0Gh_Wz
zQ&Wg`1b?l&<2nBMANe4oJW>7Jo3(2ZYgwYUGRN1HEC)|}f3+5xxb@t63Hb`}9e^z)
zL4;xk|0)L@7I%)`TA!>w3WlutXMv7^5s(YhT~y?R-yu^X5KpbjA*8iAZ&A!om6B>t
zIYKrrQR*$As<KZW_07%qaee?9LkXsAeD*jf8oC@w{vRKEcbggCRL^j9aQKDHFn4`;
zgoq-?h7HyoR{%l+-+h~WT2Upf8Ev<Nz!NJX(L7RP$rQFt`n2w^<OSuH<f}>x=IDo$
zk#W8^eK8^|Wgx{*YnlW)(PGC*pkqj~wKK@pwwMlfJuGW@IJe)|y3=JCMR%jTSdX?(
zpQ`X|9c=)Dp-OQ>%-ksI-o1M>N+>IWtm5<`e&hZJ3|$|JZ{EC#&|mFXeXS_{+O@4=
z_ME%Te7t19g~#1}Q^kInjc^KFE<T%we#V5BmBxc`M1X##!eQ$<I5H0~6r&uNE2pk^
z8wDRN14*ZT&H0{G!gV0!|AGmmr_s5Y!2YT|pD2Y_f++EAu^h8p&2M2p4R5ZteS0Tt
zftO+g=TkPu@D2B)`ABIF(TGBvU!*Q`xO80)Y`oS<iGBM}yS!b^d=I>lR_#jM_5`o+
z#}PI=>RS|a&nW`rA1gVt+HgMO%h#{3*F_mNJLy#4p`Fm+xQzdBLRGsoax&y<50zf}
z2-97AeATK|Zz>OB@WquuyK0nB?q2!<e}IF5()6>;PK&iSFZc@C4S+#QZFMZHcQ&<Y
zwFMKbq-d?$&r9Ets<-PmuAa}0iXA-Hs4QSmkn)u4Iu^bn%btx01ud{-Ix9Y&QjZbx
z`rL{dmvfW*TjtAs`t!echlGaW_TQ2B#}a)Tu$lU{na2n)N9y)cM<O#hb$9i9F%Bjh
zC5iR7$3MKu0z3*{E6?W5$q2+jcY_IDJ^T^!tDq&!@u7IP07+06Z*%~i?<iLLX<47h
zGQUCM+^Gy16*;Dun>TF^)6Z?HgWY8ii(`Y6vdbPwNu5kFXgt6)#ns?XznP-39fh3P
zD@eR>q%KMJtW8*nZm%t-^4`I<FcvGC+ewS$Bd8>Fw1H0s>439P^Fppjc$GL=LUH0J
zlglN3!NS11cOqRxuFWbyeU5W~Bzz$>Sll$*sj!{|d!u;xP~>dci?4dTM;28!wE%V~
zo)(eK@frVQ4gZ(8{{ZZ(uM`zzy)*B4V^S>w)uZ{hx~K+TUJ(`;xbB)2I~q=J7wAlF
z3_|-^MH^f@yZ(cV-;-|nA%0jjeB#2FM)-4^fzH;p>|;XBtNrt;mA6I5Z%PMvPBLD&
z$*zwd*Rb+`{{H>3)4-ADSNFa(<wrzDwh(63;zLCAr#;V$1-t6gEo(m+`H2BtCe4)+
z=k{hSP;o_02iE2eO7WZP%F?mjqO<5NAm<D<x_`5Tfio?NsN>GAX+P4LEvJxE>%`iB
z5VUBZH$xY*zJB=v;0MWW8}JiEbO;k+n40EH1uc8Me7g~Pf=>sY^AMnlpCj#v;BI~R
zU=a7cw|a@bG(6Irsm-2g%lYsz{j<IjTAOsT5*_G;KJJ@9bbIB>l_?D{VF8rqKk`Rm
z%KWyb<MYnW69%BRWXK&aj0U@L6@Ag`RX6*7_@-{&vjH?CVbdyOgYyRzn$PV#oaOxo
z^_WfRhG+m#+<8vUsXugK7)aT-1rla}bkP9WQ(#%l?4Xj~yeLwtw#aJU;l(}oEV!@e
zzP@~uyxLibRwtci_ua~j|M$bS``UPx{K`VD=bgeh$1u#_QKUc|v(mH}j4mv~e#DEo
z4nd8kXHFs`B!kOw=<~g<=jIA$mErz{a%<jA*D6}q_~K0d_fJTqTeVuU2|pOg^UaJ4
zlj|!t81!E|y(dz;%AgUoG=G-H_XxS+mh^*EYFxJon^Y9rfH}yV>JN~vf+ggs_Y@j|
zlvK8Geq_AYdf!6DWJQ-Qoxh*LuVbe@9MeYMKGuEeC!hryI&%BeOGxRg4_rJw_{OQm
ziMI3G`(hfOC3HD`21wLGK%&s7(fR(lpQ{j3#y{G|S9ZtfFuVzwWQLd-92`_T9|M<~
z`<jR5nHFGH_-=-xKDZCSr@y@%(L*$k>kY?Vg-jRD2HZ0jiwBC<Ev$95qHneK>7cnK
zOiWm|A4S!s6y6;>ln-uC`H0LuWAUzXTAZ{?cD0q#tH4@@uj{*XP%ny2h<z%&{a9Y)
zB%*1fLd&3nkFHl!JN5jilfofK5=D$Mx%b-k7o-CZTtn}__{}~YLDA4jf;UQj_j^=O
zu)8a;1Kf#0ZDtE$dK33=QoZF#zQK*YKQzZ=-Z<^vAnHXC4ntMxiDdibCA65cWPHg~
z#c$;z?yU6Q_j>gL2p$pN-NEn(cmcQ_>^zZ&=^(Tp1n5fk&-R93O}>q7<%Xge$)%6n
z>OjbWqfq|v;c&NG&UoWb#07|C&zhp`RMEL6v}D!hfSBvox6Yk=z8oPSYJ!!QYmqXt
z9%&QucsK<tIlYQ0gz*@mm-k)08l+6x(bB}Zq1063UmbNHX->~mV>HJDKkmQ0JJn9G
z8gKu5Zf;FIs<CYB9`5d0t+taYJLH{RKZ9o6Yedo4Lj@Rw)ib6jydqkBmiYphP@ZQO
z-r0Obj_H4IX~3qRV5@%i%;g^D8d;C$pC42y)xKVV2>C~|p&yc2bX}kZnp5wuq@-l^
zkXzFg<@!;2m_h^z^<Dzh2SGUgp!KC*x*iilM56g}>bX(DUdIlE>8<f7y2qY;S}^V6
z^-qfel?|SAmoO6}BYWOsKzQcn2v1^i>-XZ#_Peb|W(BP018hW*iNu~glYnZ#2duI>
zT>ldAmA^kw<holAn_~(}Rsn1f6qfN`cm#szg;PhX7rOx!5EyK~ML{aKJ>O35GAdS`
z%^dpu^ZW2sDO9E*->7f!YUub$560|A?`d`*=nTOGPCo~%t3+U3j-8T{2K~d80J;jy
zeP>F5gh5e&L2&TAUTARe`o=A}QAtTv)=fOKs#7Y*`x-BO#OY|QeD9r<;QmcnX?AwD
zR_^mzU`gtn4Yi`MO}Eaw>NV5tG^&TJ-1bL57U9DCZ66~;M)JA;oMG3v!X+dw>^yr2
zA8uaUZg=AA5%ge~=|slH0(O*(bJsx86-dnqAWs04u?MS#(2%xjCj%xF#+RRBWMYy-
zE2cc@6I<}kc;v;=gLu~8AL0Fej{7;yHi(6UbI*MH`o`zHt34Sv$ibWb$qb|1-tmPH
z(+)8&<(?}jhF?D)O3ZmTm^0?wv@>JVcbj&f)Nr4AoA5_^(}k+>CGO0N$rQ@hKSWjx
zO$+ZH0_$#Gn+w&-<fmUt^%gHj7@^%t?*I-?y{lhHLPElrOX>D^4Tl#moJei55vSj<
zVZ&$$n$+q)QmPPl`EP%O6I2%~(g|Ya=rE|wBo3?PbP90z+={Ha;llzzKR>#H_1OOP
zqQ7t(K&wGzRz+F)&fU9mz!zbc@M!C=>5|bUBC>QffF)c4;u>Ma01OtQ2L^%54$wzV
zpH@~;AwV}GB`7RUOiy!5J7;cmTKjT6_uxDL-xFhU%?1)2=0r7D*Go}JX&;3C0B9m~
z0wxauL~EFT3{yCz^SZs=P;TQpps3(D{Ty!7=6H!4>&ymHqzKz6SPN^_k?WX$q-(7I
z*wwW`#3=o@6L=^n|1`K}VU)$7=yKq>?622k*?q3lXp$FVqQ%mhRz@V2E+?_HVcp%r
zrIbNPg_{mN8tU&q{jFp;+;|Xt2EBdriARi$6E%J3CM7N-WSG#((h~#`0q2T72a<Do
zKiQgVYrnyhS4*Q=G(SJTLmGXF?}60#yTHXyy$V5AhxB4q&bK>Dj%Dc~%)LJAqfWYy
zkPDz2kje%FkDxd2#8>1zYXxtf^&R#;drQpXstr8|7kk~rThI`*jZKsBVmoqt1S}~z
zljuRO@hS$g2TNK8$}4)zKj*;<e*3a<pKY|}_BHc}+_SxfP+BO}_!e+`5(S|k{Upe;
z6%-!K;?DbXJ^e6}-0GEt#hJSiy*#-XjKOXG8u0H%z2E4L(7NUT+D-LzaHF9y3JrEM
z0pg^GfU!{!U8<Uzp%cmz)_&mWx7~6KgLPzO(Z6v{LcR`o;dD`$5eoG?o=33EN$C^j
zZrv>ku42ZCthBVR-cTz9rM@{H?sw%0PuyYZYok80F;Q**g2Qf4Mlg9Ya*BGruNues
z(BSxkqv*6}<raH$nwP;5L}L~V{yZvy$jG81BF5qBk=i-oD}!+KURIS%cLPs;z35nf
zzl7Vk+2}A@4$(yq5sx_#1pw=Td4i74LBbkHl6~9K0u=(G&1xtSg|G^wUgBnDEk=@!
z+k;eN1|A9jRu7B`q{blU_u~s8r$h>l>>TWyMxeRqgiugWD2ah+phbBt3+!h0Z&%a=
zK7t;*2s8~fFEwp#z-$Y+(D3Z@AZ)#O^-74#J^Go6eug?**W@sw3=Cx34uTV+9)jM1
z6j4`G6BrT#xB4E{yJgFkftLLi5e*1YHhVdp>=>j-lh#v<6hsr4RJ3+a{wV-^6Bl1;
z#O=3F8lSQDKZ{<AhpXv1hcU$RjrNE32k-&(blreCH7~9#kWL6IK*f|$;%69t;3F_i
zqV4e3pfzavmpl_KY)BdkU%%$Cdy21vC%H5_j657UD+t3P{QQIP7uZRJ;NVZ8o5;q@
z%nWOY=J?D!uuMs1>gQmyLbf5z(q&E<5nya9o#ycW89CHc7<y4^(k3QtreK_dFPpZs
zA#_CzBC8dx>xIa%Cqy)p51K5#Jl8B{g|0COCPtl6lx$GHdVZ+Y4s9hf;5kj45D3`h
zh1yE{RWuEqn76+qp4RMCRgVCJ6f~dJMXzJLDE`|=9K_B<jlKN*s+XD5Rz{!M8;_>b
zxNw1@w5Z5P44hB(3XzhO#Pr4esPq-@D+(8QYqJ*?36LM_HNQx-ppb#HazI8t#P_}Q
zAb)V%UT&0k8HmKu(cXboXqQ8~JKI<v@-u92Lg}_By1v!APtY&2PvmN1!JLhAm@>Q@
zYL2S@2Q)IM<rNimp$cua^Fq61=`Oja4y4?k+#+&OnrSv~Af>1Jr+vA1j|IVn3FRhK
z_klwX!KL7%Qom=n4q+o8J-Z8CFybPlo-*`99yl?|iBw4*OZ0f5=%q(}YYdJlGY`+n
zLQA!fe7@bg9gw$)S1ux{1s|-_Ofn`q@w<48Iwn=6ASOWPm;PlFpp7uz2XJ>l&W(N^
zt58j+HV-BFeiJ1#vlIXdtLYK{V@5zf<T)VjeY!LT{d?{JqG6LaO}FkP)9cZ5xt^0#
zFp*!srnA!%r74tQAY~MX9JBgZ&1StdX!w*N;7S2*A+#H6-M`NAR)c;C97n8ku^Yiw
z9?Dp7%+z*EdNobWkLY*298_6}_QYq7dscSd*3UtPxRITmecie(rf<<c1a2g_r7;@=
z@-4csqyaK*Bm}A1_*NQuZZS#F4Apm~QYeG0Li!`{U$C*L>raC{l`P|)s-FV`-Hx_1
zM3TVO_jR(X-w-9u`$P*Ig&45L0)v7sT)c=1mos`z0JIyTiO~Q0g9i`LibvE!QO_R6
z6$6_u9$9xsS35k}+^{!@)Qd11z-|#UvhnLYdHc2q@h$L3d^$9G3Rj0Ed<K&MXSg2q
z8gN&=usP44E85%J%gK>?FHo7954#<!>h%PV!7eg!Njl(*0b}HXA9@&rswx~HxGIPf
z+R<o<v}U03qFEjeBH}gRgPs+U$dZh!yvl)m^c(qygw&&sb>`h?ol8J8VTn5dQ^JP9
zQq3Z`!8gUHgNOxe5U8dEv=%@JKF#pK0EY@DgOpmqxAKOwMHB>nFFMUyU<*1sA3_%c
z=EV-uRS3)(`ne@iyl;dyk|^`pIOvX;uDVq?8JOpNt-j7xB?uiHQOhMf07Sqr0+%Ox
zgIq=FQI0S*?M8i(d>XREI$gWdWb`O$mwDWa1lbLhNk_T$oy_ajtz&0*1#AQ)Y7}3c
z;QWYH@U>vf(aq8f2okcLp6Ds?9?mO^E;~az8pdXyNZ_Tw*qc5X;SIn@)XM`0#ND?f
z;MuZ6NtRt|CXUGy3D?ZGPc(_X_0ErfkM^}ZZoSJ>-HQutW-7(Dt#uvp^YqLexk#Bf
zy~gn+4f)oexcROx?7Fw8f_>zlqP9Vq53AtHD#yjL=xN0r^x%BJcU(IbZHh{um)E0`
zFSTXB8`Br>f&&B7p34@>zpYpcWqO&*e;!;`!_N56uAuDRF3luyg+7^m!@({0wh+$k
z#BQVq&@4Utn0M>e9>{(c>dVaP+S(4G`C(dvUD!Ye;?ZYcY*+M?IE!lkN<IMD_tkkR
z6HL-fr9Sk@jQ`g=m10IRhfF8|bgEU*;f&V4sr_yU1Ycg`KKsJWFYpa1RqR68{x}_$
z9!If2UxxNZ`mg-8L+Ce01khN2?g_rW&mfhXW{UQsAMkW1gp$1Y(Hru!<HZXrbhRKM
z)elt(Ty^d(f=!f@oW05Sl++fb7%aU-m-Dsy?w4x1-hKLRyb#3laYq4GGp*U+K=2=p
z#e{LNWRTa`zc5o|zI_e_R&s1}xmWmjAscnM(6cOJfc7O0T}uQfsAH?8@2Vw~0s(Lm
zdZjdn2!WEke0Tv>`;1(I{8#+qkMZ#WSIHWlhYS=mCF83T+h#Z^6mgQr|9$<hQd;`F
z_~GH?ugvVC3!hn5nRj1IHwX#dawVAk`Efl(9oFT{@+mPJHsrFeIQ!H@DA;J7(R6F-
zHvQn=`WNI4Sw*fK$WfP1xiYfOlKJq`+4<|;2a?_QIPHwGj9fiFzu@8o8S#To)-_S?
zl0VUKO!-gNqx_$xN2sptLZJl2<5)tKNw5`EPk-ngjNKe1g3`FC8d4k-{-Y?Q9dTLZ
z5OGwQPPGt0lNU0VEbswUuK}h6li?uwhY{-kFQg}<W07hNaZnf!!ngraQFmdbIMjGR
z01mJscl!kDY$$R_9{WLgvh^O~X_O5#Tv&CKGww=Le0<3aR4VsU3RfNaB4X}kpb`%B
zLQ4(Oaa{ev|2m}q5V=6Sh%hSoFXg|Z!}*`8CHg;W@BW`<MgQw@4*f@|)gx`fR{zEY
z_#cuE{gwr1L6ICG_$@rw#Y9FxCa>EH!OQ?LVC6I%EoQ5qbG!nI+oF&Ny1`mud?^7p
z&~kyy*4oO7@9YzlI!K3CE$3PaWgAA@QlL1A2!K=p*3*sPDy>(+KUbxRP_8k?u*zD_
zY{WOfyN|pFB_|@BCLy_WgNgG|<dc!&dgyAv;MAn(+jzG4zkdm8&^IJx)T`=Wp1V>g
zWDYTflCDgx>&8Ek28<E{VFt=_++_-dN``48Y7yz*0#-m4beE`KphkI<rnLDk6I#d^
z)<SKN&n0S<YBXg|>YbZP<A{C8)F`!JJi)2IKkkQ$whe-n>r8~W2!2!)2RiQls(DiH
zApS(AE;P}j@4SSP!kXR(%@S!em3|(dhSXb<W1WDKWC3zTcqVNP4Gj-f^d!~N6dPl(
zHf?}0w{c0DJ)!IaD^C3%+Jl}V)0CO72{zgi#YoAJ3aqw3DVop71R0%xM9jkiOzE=v
z1>DeenouWoWH2rKJmTQnj6WP;vDCw8Z{*=G7;-lwgY6tQ2litr?TxaCUnJ^|=-j*B
zNX9%su|cn8>k>4j6v3|eyu)ygS`3Z7Ze}fnPu7X2Et&Y%r(fLACW*LJT^jGUinca&
zekkLiuaxye9?lHhKEFR$(_3hpsk7r92!hy!6|TcrH-Y8&w&t|`H$3{eLKNwD#b9cP
z`cAxAO&ZzBXtsat{ugJGcY{KwE`n~0RkS;ArGkGz@|3#@SHichfSaZniMj|${|PYZ
zZUGa@*a+QCzrWwM@6lvq4nDiV<X^il@CqMK`}B=ez_g$eJ^yH>v;hZ&CX~s4;#6!s
zB;cJfH7v#hVq@!Q{)3ei3A7<*ucxwv!k2?W7I4gPJA$-8UNtV;4LhFbHBk%r?sRq<
z<&=&xon#G$s}TAtaU<%TSOOUNm?3^j%JO}nO|)*M+^1a+?CJl|ZT*ke|9|5*p=}YG
z6Cjw819vHQ5hL_0QGv0x{@(R){~q#kLirQm1(B&o$88mKcR=i+USp?F7`Ks?bO_Xd
zdDGSgea{gG4y}d}_M~i!L95<JfSG`sEU+54y^eMZpP1+2bHL7qUe8OgEzpXVJmf&3
zq%_cM4(1DFA_oHCQ#p|23ABdF4IQTFt>lBAEFMP!gmENa(GjevUG`@?^jy9F{!KHW
zGYI<cAhE_FLz>)dKp_$`TwT1NBXJwrKgg=McAGu12$v-k3P~08B@&`u20OaEqpi+B
zXFluN5}(5c&=^25g$j9Bt416?w0CiKeL}?$LsPa_R3g&>R?fZWv?d61rx;NY{e;vJ
zMR~#NMVO(Mi4r<Yf-=r5@SNkAwt&>`JV+0y574s<-^6$Ya73UZK+QbPnhwS$LZ%U9
zq`NNv2tlxY`t<31AZUo|kt{o2Y1W6q{^-<%p?FSE)p7xgxDA;nt}AX)H&9_RcLi}a
z{0QmlL?i^T`UC_4Pzu9*SeA|Dd+#ly1SFyI2kwR(mgj!&QU{$`%-*EllBV#7;AXMy
zxd2|6P~%bb@L`mTiB1cYN7VlrAj372F(?81@uJBT&(8yHIa28_;jTUY8(zx%sS~<~
zk;d3_kI|C0D@X2QCCJd3g<sJ1BV56ThS1ID4w9)30;Y5nMr|aAyY#LuRSzS&$)Cr;
z97@F&!TG)T*a<o=TJM%~C&b3OzSY_<Vo>lIsLMT_+lT+KK$)@+vxSZDSY!ryu?{gO
zfSkePK9H#ZFl=$^uAJ<j@n`bf@OQ9FQ<%q*f=y8h*?NasMeO21B4sR@OHz_>nF-Yf
z6X26%*l~jia#~M}1j0f>2`%YYInM8HRNPa<zS^|qcL9CD4|L3$Ku9O51^xK}SZ9#A
zL`Sl|3iS=+w5jtWmwD1dx<&z4gp}vNjEi;R{W=l+Zc9YV>DTYxi=GH&*SNgHjCoxw
z;qn7y2w8L3D-1n_Koq@8i(>Qyg6%DQ#P)u6C_~)`qv?eE^RaO%`;%ar$;6(91+f3T
zaHMPjJVBL^MO)^Fz4z}EAXHTygll)l3<>&ummcZ>0cr=n;uR*=7D&cGlOgrw*yHTt
z!NG*a954&Gf(w{<O6UgI+-u&c;+MgHSB=-()yR0L$L%>kMT~8fIn5kty}iy*7638J
zCJ>^CXyxgqM|A)6?}6Ca&07|YYneQr+a#eS<j;?AZ<(RYG)Nb@HHf|X)3o?UTc@e>
z{6mrM8Zr;axr|_T<vR#h8eoXr!jEt=m)65R=p;EGJYs9CV%u@k57_`1lYfZOkX*B0
zr7((}g#H&jNz(<!X${VxRTz|_Jr$QAjp!fjfMrK&K8HFddXHmbVnDK+z_9MI`>%LH
z+NGqcQ7cls7&4iSA(|K~0SXE!f@6i1zH7B&(Nm^q?(sHo?Uke1xLUp8NO{8BM{fu4
ztl=~V`to<9c6ra%D}JT-M|@L)z|Jd7o=F8)=1f&ANK?ueH_2W0B5DlbTdCRk@RS)e
zo$6IEKYSm^Jky}_WE3Jg7OpVP{(QW6>z9rg>AeI)zj+jwHe}Y}9YF5>vxXQ*>s%Hh
zT496a8*q{rE}Jq!IZT=ccvX%q4DySPq@6ssc!Y}Uvf-rP<a=p%*r878oD6q6pe%d1
z=n%}EX2AgX4KM_0olZ_pFy?92(E|zcn=$qhv)9RG_3%RoSpq3D<Ve>oL{aF$%GzVz
z7|uDzR0Y+M9EsJms741y@ro-}N%WBkoWbB4m+LTayPhTi&|BS60(^1s<zVeJ3KYpu
z6WunTuAy<Zvvd>ywHz)K;6Gqb2P(8_iz7fY(SD7Xhk0`p8RRwNV~=oQ)nv{j3XIR%
zb(V_Q9B(ij5<@&>gK%-RYvWri9vQ3W!Rh($-XY*2tVSXtM6rudA-BFR2QIv9&_EvS
z#ueG3!on-AP#oOyR31g|H!$UoF#~fpHcQ}5^1{?zAVP|z=|jl>BG%I!*@NF-H!pJN
z)n3B5k!{`$&{D8lTcO#G@&nMMguQ1bx-s>Pm|_@q7UQ}<WS&ZIINfh@hIC|3vBD#s
z#VvS*ag`Qhb+xtbm@YTDFCSuha5uA@Kq)!93{Io0{gpO+Ce9k(4j~!V{6~)I{Lt~o
zbH{*0kq<ylpt70Bp?Zk~JRx>Lp7sTQNq<X^BGZ1sWw_~y2Jsg)10nr!Xe4mXc!1&*
znZ7$_JTs4KD56^|y!Wl279JjRkWZfb`91j_sR0vjCbevS-9wng=jKNdzDaVCiA*rC
z&=)v~6Kvu70rmiviMeY5-NefK(PA(bd*Zy>pLeLXhF(5zZ5R<1HTj%H`}xOX@PRd(
znMTQR<fYkIVK9r^a*lt0@b%h53@YNmOi1FHAf5oS5W^5b2eBf86_-#y0*W4d{_$AF
z@KvVGJaSi|^7s+n0`)8pa2Gbg?tZwT!v*(q3A4A!Fv^sypXNmI-`V2Gkw#Qfw*MdO
zy=gqxd)xP`R;^~0Xw-<(AZbDwDx@@^c@V2vbJ0MFCW$0Unq+Lyq<NAkMVd%yE;58d
z#)>lR_u)Lx`@Z+<wIA&Jb?qm6@5_VhTIX8)|G(dNIEK&hIX=g1#TW9;t!GHKq__};
zqK&P6j&R@6*PQqL@v~=l3a@|QXa`&{q-mYn6?w*{Z>LYmC?bF_Mke0C0Roy0$JfcY
zAYxYn)5ym6)XnfKfAG|lN~q#i!ef^aERvwH&~!}8n0uOfxdE$y^x(tGAfCMYV2UE(
zkr=SOoP%Z6;9ucM3BF9EwkLW(Y|O{`%Iu%;4MMsjme5d{IS@ei1oxOL-#68oQpTQ@
z1`&8}1<xxH&6t}|ORnL@ykegJ3G<NC9cc?<IMk%CrUx9jelm$}{`X<$use6>juU<S
z#_pp+X@%NQ@V10G_fMZa)4Etj&jiQ7Eo+bNcld+K^TvfzcC%f5uOaWD^Q()|8*x2#
z8KK8DETAws!Qk}^uf8;bzx<#jpK@%#+);S4o)ZHrm`~TsDHZ=@xAE%<=5ZEqM0!t(
zKCu|rCqADT$rYZb-Y3OM_Gk+V_zIs}ZD4H(ANc`Qr-;qZN>nY0Tc~_h%g8D;uNN*}
z+|Id}WuD{sY{-p|4o`3aFNt;RYJa~W+Bic#!2hFVg@EH2guV^ZHZC*TfnY0xUd<d9
z#3(Db>U=xWvT~h=s*d;~8JL1GiKvMM|Gn;3Tf+;|pRpO^lI~LW3_njoca0Bh@X&aM
zm~S)RfW#I@75qB`_AkNRlACZ3A0Y=xNveuu=D|O5dH893Tv9K7{u@;Kf9d!jfp4}f
z5aTM0QW8x)?`S513qr7V6LCAU*{1vkZ5$~iABL;W;4mni@K#(sMxVd=&dgNK!`-<H
zs2Po1xVCo_g>A6lrgZ46*yDc#_56SBDN=-lzGvJYJ5o8zb$T91TIU?Ewu7Hxxz0*8
z6EPXpO24x(Ex~FgcUJcA;;DU2-AHl$3r(vAb-3B*ZgL+xt-RU&@(F^>*gv0&CaJX4
zSd1+c>q|UOG9wDGSOjt@?EVe>Dak|;|Cz{XoYo<d!uiB$$N)JT3riP2<pKt8p|$1m
za>{+Qe*ENybBnJK>0l`2GbACLbyZKOLs&YBU?+!-*4s7K#p5E7omeiJ8^`3o+eDM3
zSNs(y4cZ_w1z75~WlskIhR?xnPciS<R7G2RT=)p<|0GwLRnxi-Trg&Aql-j}#a|%|
zOnz}ntu^^ti`d?ntka6Q#ifwfbN(zw1mtZ<Qe9T!b_;is&N!ODE2?WAv`#U1zA)|?
z!3O8gKOmrOZP7P|@P5HU5)6(A{hQ9RmDsQOih3-LULyvR#Xft3ajJ~p6lykjL89-W
z5(F%+`Kv)xNOP-x;}-9B^zlJ$W5RZz`GXG2)ek;6ILfn^(afTMF2NsRVt6beX^ER#
z&grRVUrUSx0j6Sp7w2q0Pw#+a!U(12H!GZP^*f$IZ@H-l3R!<9DgHIamxM0lIrH=m
z7f%SG#b?Yqz2i^$Z-lL#>qhIIjf^^TF$;8bP`aLfXR;RQ$0hoAL!LOHsFHjhPrI0%
z@@yHy8nHF_yV%BC_q5mPVfTbk#CpdX#TKNz5-dy{UU*fkd&8fB%&wr%x|&4Pf8z2`
zlOD@d>uYa4KE@(*|BuH<aHgLeo@fz5k3U|9pDK#Vk7Nq*I!>!b=NaQ4J$}LwoFNX{
zY%rx)7u{X-A>G6cn6+O@Asx(P{3ILoXLB}u8p4XP5=tFw`3TbRsVP2&B@f!jH21^+
z$k56&OpVkf>H>>(WfoqX7SqfN;LgM{Ux?GRXyHPs-_ue>cy$)<ZtH>jfofHB!L53k
z*QlM_r(RYHKDXhbObsWlVArNT&YdvNHemGDI05N2y+8>}jjap4`bKKLw<qDbO(3$n
zv+|?zxAXUY0w2)Ot$}?~vDd9CTu;r^EsY3rdSJbllcvvE_X=^C((4Gq2<3%+`$!3w
zW3dcHsL`WG6Zw;A7D!r4qo}})w#j^1a}@ap=N`X*ufM0K<?G=M@R7PFT&9nFQFeK7
zQ)Mr((Gk|;j}FyJfBbi^AkfraK{K-J**Rsn6N5#)UE`-JUUh=*C*@D*-~B)a<tdyp
zzKchvUoRg|FS85?!IhU*@zohemPPhn<SR%czrWxKkF0p3Ne~=UMPuduhKAMd?QCyi
zk`c{rb;Xe=NMbUkq!0b*@M`Mx>RjX6-g`!5d_MT`L9hG+6HYO8nfyjZ4stVk)5%Ph
z&uoa*pc3=eyJq{DyOQ$1XmU=y^nbpM0rNI1XZ638y)U!*p`%BK1Lb6Tk5H<v`4x+J
zl4VQ+Pwqx-3&SRpfTgwt3QDXg=?5||hqz?W4|u`Zkv8|-{^^R;Ig=a<`Lp|4#J>GT
z`3m|D;t}JA{<h7IP{m1b=)7${u4l3|Z{E8&f`H{J*q8GMY-#QNt~FMZO74Yf++qj%
zF2jZl;&RJo(Ml!LL3xB=Kfg7p*5+^l3aHzoI^<m5on-5<Bw*WVSupZ@_)9Ql`yKld
z<+y8d-2L~VjJ1@|Qzj_kvS2{F5+TX>VY3--19WwNfj$-Sdf3*p08}7ZbgNZyTl=NX
zuYP<IW61G72gw3D9bBzN=%#%moo8F2waXxA?4nkxeSGw1VyD9c=EiZqCHECPj1}0W
z=A%bUDbss-kTVyxwR#uLTDekWuRL0GTK{5fuOWabVD%E0XY6PrLy}-mwjFwW-g`cU
zUn?8(lg~PIKk1N;+Pw?mO@0%Z%k8FlVlb*)f@#rLfwTh%zy}!N3<woMPQ0ocC$ObK
z@d#p5TEa!%nB?GRUdnv3aGXaIpJD(e(fPvn4F8=Mo{zYhpPzNCYqhXLRP~BU+$#AM
zuMTrl`q7I3MI!OdTONvl_ZRQgc$lPm5}1?3G*zu-l<T2Eizimpg|22b#pPtcc28)T
zYUc9oefKP8lQwr7Gm*ff)2e74zi{wQBJ1`w7j~1nRb!5|Pq*4XGIN~ogXv)wJxVkT
z^WD{Bdg!zD7>PevJxGp5i#z}%1gq28K*oPg-Q-WLMStzCIZKo^15Im;TWz_lmOlSj
zX@~!_J8E+!8NB5F=7EC-5%e1Vb9YO~p$j)Ny`U6@y!_H>DTb={hh43TV<2QSymEYp
zZ(+OpKh78CPQ`^1>N1KgZ{yn%=Ws%_G_UO(75W_ikQv<LJ*ZUc+9q>AnPl&A15KkX
z;%|Na_ZUyIU|ut`fg{sVQC%Hkze7D{*<2>L-W%KNm~yY<%u02{sd?IbJH9K(Z^MJu
z=E_c~i~?24)yQ4~B1aNa+PQsM-+81CfsS_=3h(*RKAl8{Y&oIBDRQV3b4G|66@4M4
z(#W|eL&FLAGwt@^%ZKOXZ=3;COKcE?L9c&dg~J@@F9%ZYNzOjDgM<QmZ5%{b058^_
zSv%z&7#Tk8vpswq&zd*CImqf7sSR3m7^t4};SHJMKBFH33l)DYZQ9jS^4++Hid~Hi
zqD;<Q-ua+i&yvmso)-pRv5kB%Jx-J~xk`Ckq;>d8Eya%x@3ii{%NxmVof^7Fn&5t~
zlb-RlNs7t>gZcA>=aBEvj$XauDnUo}_YJY*7DR~j3RuU;myaDB9J;V>=~X$pxJK=H
z=S~me;xLP2SWQ(pj$FK_!{5usa}nVR3E~DA<d`+_zB733M5HTlO$jY~`XAp=2<5Jb
z?#e2aDF8;2YvD_J<fLJb1{b=ixQ+Z~rCAOZkh4smuxIzsDKVS&$VB#bb}pB8%_Y^e
zvWMe2-PU;_;hNA*{>kOWuK$+01k+ZMUSiL2;t^&%_Y!!#kPDC<rs0QW+#o0g8K{VI
zR$W;R<CJXGAphJI02`}_Vy~}|ik`fA0py*TzM@rAce*?T2qS-7{X`B$*~H2QuNW|h
z0Z4w+zPi#t9lo0jhQs*-1<fejZ?kny3#=CbHjHiF&r{vVQ+r~to~QQLncDu*4-S@}
zYqzh$!VIC2rz$US_<@NT&Swvcv;#n;^Nq%mI^w>5GY2S|FtC10?OV|<dZ(;IE`rl)
zilP?+?0XqN6$!VB3gyB@i|FlNOdcw5EyuX*w*gQCX{hy9lQ}>#(5F@a+(yRy{>HD1
zfCnz1igSnMYun`;?T}q5EHsG;QYyv>r?3C8!ubQfL)FNoLVN>%!>bdD%8Va=xt3A#
zM?c^2UofZm<;39#>aHP0j%Dq9K|u+mXeyU6wHWy)`Nhyo2?nM%Oqg7?p%B-ISzT84
zxCN&2CR<6Q9Rnr`+Jb1~LYz&8wop61BiBPkp{e(o3L_A97p;a#+Ra3u!X<y6aKZc*
zz8<B|j+{7XJB%ReK}!+VSiuZmplx;6M6j$f>Z%Dvl3pL$otBf6Q^Q2Wu?xKBiP;qV
zZ_@4L%{(Bk;?Wx?E+#4uo56d*`U;!$3Y04Y+9fq*+$;;ex5l6WgMBJuYpaWj_wu;N
z@T}JP%crJ*u}ilP&F`XYge@#y4+_y2e;aWhZJ_ev74;+@!FKKIV6$(Gi>EhN)5cT#
z)=X{nk^^Bx!{)lz>1Owv@0h7;Q%0g+HzUB&5&c;q$l~%aj3f=R?!m}gNeu?p8`NeO
z%L{v^ggNW4ZO^fIK1r`lmn}W?^4hLz;y#Lx-Rrz;LH1uR5>=S)glOzU01<whCmmk3
za5$9#LayE4_xo@95U!?OQ=KZ^OP!!9I$#e+T0s-N-U5~)h-}L3eNOk7`t}@WEnI1L
zDg9urj149}-Lk(T63gGu<=0w5TXLDQL)H==WOQ<Xda3%Ud+F_baHj3HbGFUvA1>1H
z^;>#Zs%!*2_e&o0RQ6|dmqQhWg+MLi)+RT&Q~J1X%3lf5Hw;IDPQHii8&j_cgBiBx
zHSZf2J<hCMWv6Zhgrbd&{Rf^ThU+)$s%A~1pl{jT!I4|o1^3YYw(scCKMeTRNZ(2i
z%`RK@(Cp(&tq&$v#>-^T)}!^f7x&eo8MR*d^JJf8V_fDc+4;4xbGTx=(z=aZ?*(^k
zJTF{*5Us6lXt)`#t@9^$cIt@{07aW{9P<^YIymcFHhmdkHs|x_c|(UOX!O<6D&!#n
z@8*<85{Gz9S3EX!#jGA}{A6r8^+@v@!I;gA*e$=1_pg4ldS+YK_ft6~$;0x`g{;Ju
z2Z3!-Tgluz?P)VnWWWa-CS5s1NvQ_-&LcM6h?u`5tBFd{oK8$UN&EDYt0N3aax(uz
zr*NKq<My@<BqAiZZsvVPkBJQJKOOqK{M@0Hw+pYkG;5Gfpw4;jQ=rFb??Am^?It49
zAgbR)4GT#UY;KhTChWd-H6LN<Hg2P4X(~tI;Q$q*@zRh?S?s}4_lY6P&L6l$e*-as
zanxdh4O>EQi2?~6v3`iidIrs`$~n5KXZ6^%I`PqaU6FYy$vydu-Z<`z1b|jDK*l_>
z$0XYBAUQv<S$zGXFo!QJl&dp6vR_T&!jFVUaUWuTzq;TxAkNynvq~BfrlZolKJ}D>
zAG9k|ZnQ3Q+_cxt#=4x)AZ)}z_^d4A35d`4J9q)p5i$+fSQ236`Q^*S`$PBCcLy{X
z@krK(Fksuyox#NcsDaY}pU_og;_(vA-fsfz-5HxS0qKaoea{x(J2|>y(|nA%m+>qm
zZ6{xRxgg-q9~$=8v<cj>g74CfE*_QkleD;V?X9^0tjp3SWWIa%jw;6Jk=kK12;B}`
zV7!H4&bH7d6idFdwraBpX|aqgRn`a6!eZ)H3iAGkc+7@y0Jn&$vmS-4l{A{bwH%cG
z?J6#gB*IhMYB+T>3Q#DyMiHZ$cV3q&Bf*<#M-Pbl0Nb}iWfs&pT1xj{NXDzy5-o~E
z0rf*pi(YA>+NxV$0^~D)``Iz!-ELckiaDW7isDBu7hMc$(>peK0&tB<1`h7HQihuc
zew!QmAn<yJ>fVOsrv2|L)s8(dk}A4Nnvctd$BLfXt(DS89<11R;KwK>yCt`{Qc_)X
z*(3L^(zt<H!2bP%2=MYwEVt;PZ9Z)I%YUW4{^#hmc=)c9v&rVIZs!%-59r1k*{{Mq
z(Td@J;|_^Ph2@t%lWB?DcSDD`nn>!>p>U0oZ4X-qeUu4Y70p_x<jy%t{lUrS?JC!@
zqZQpn@f=9o`9n(jL039aJc{D~d`Pcp>lRVpEOvre#2N=%P3(9}H>FYocEE4{iZ&(8
zsWR71V@j^y<dsq@(}y6i>wDnmLDx8u9X7Yw{4_a@^rDBG3+02bKL+GqDJQOSn{*vi
z+zrIN_U@<LseH>#Q$%+qI@_3#PYIyMxK$*&eM=2%Qszy&Y_W0a(wP@|tK#)3XD<1f
zQ;1P(qN8#2!NJC!+R6ti_8t8(O2fBe-#%H!(@tcGaV-ZmhH!M0O8nNoH|WQZ*|bu8
zrizCzK@Tg2I~V=q^=uM<pLcGQ^C|<zG(9lh)@SY8lmCelM(V2NE+N@=x_wu>35Cjy
zRf@!L%~z4Z<~A{%-ZKz7rE@1y5+kaYy3MFhc7c|h^=(-+(+r!KoEg0=BWwh993tc?
z|4hj!db%cP(g9_^_}d0Lt5mf8X&wE@u<yzwC0irC)*R&BF6$T0czR}})uwHGv!ggk
z_gg-0wmWj>j+&x}q7Fy0GABln>}_v92Q#m#<=HrGx2=wky|jOya6h&>t_kwFNiYil
zEXo9%DR?(1N*L-ijdpy8>Pr~n=|igKV~m4A+KcaN?bgnDc)|XOnyO8Y-BuKj`o}f)
zn|>OnJ$6^`cj_^wV_ZVF4=7PFmi%Q>$tmONuk<<WUD=f`7Cv7!_bz~(F%E&!!bZxS
zk=WAcQ(LA%+u+``vsrylQhgGNLrrw;6k8k^VVjhDaYFcrxeaT2=_{z4RW)p*R=a#p
zOxBQKs@y|eJraiuBYzDZv3B7oDE<N#iRM(ORbF~+%`d*u1}jXmxb|D>SO3WqNh{`r
zK3?%C>azq1e3cZM!4Eb_LFSg@;^(G86;nOg1);Q*4khgFp_csYS&R3HlV9u}WSVY_
zs=syBTuGnc_iN*dKMrI(BA|!*%{<4aOD2m?{5|GDl_H7cZyMVkR>mC3z4~(E?v*m~
zcx@IrQrrhzg6DUM`pnbscQNmMZfeJ?24mo|O4x-@$*z7&0h6#YoCrE)*_D`*gidHu
zC4CTP3er41j0d6Ls9BtD7Mgwjo3G{8SnHPca4RiVwObK6FWWj}!}alg6-wPp+!d1^
zOn+Jz_`a29tJ~kBFlVESIudv~Evf0Ib1K6eor=5S{zxZW3JEz-6euDX?6Pv~N?V`x
zbw?Z0T*!j>s{4aay4W=iSGiffBD>EGGh4Gu<p)qoK=SMx62Lb#=zS$u>_SMo<kT@4
zn~EnE_{1;$ai(TXLu^%%_RY`Rq`c-c{QDuSGGH%(Putu@Z}U1V_46u=G;7=Nr@3Bn
zY}|%(xt*@qmrZKZZnIw-mkpVH(T{^h)MGZKpzu}UiVy(<@)Go3odC-;)9pgx*Wu3B
zZY=(umHjRr-{)_xm*8^vkxF@}gp4s3%a9lBT2t(i7bMw7_D}Mk`^EqZ`De~XT1+}#
zn~s|e?<85l$1DCL{kj%~X}xZ(e@btr<)d-^J+=K()MK9CJ6z#)SnS2v_y4LVg)hsa
zRSS%0uWkBC`^Z~4M-CNMevcgUKh}-JLC;=U&+*jVQ7(t4av<<~JmPt%&6P2y1-xc1
z^mo<_Pg{1V*_NJfMDK2GUNMOg0X6AwW`S1H?Y3^+pM>o>teld6xaV2dHN-?7?V_HQ
zG0xnTs`J<m2A=cMYHU7Fi8;Q@?v905$N}5={Z(6ESwjIM!>wJV(H`-Rv-;^2=@!x4
zb<?Ec^!5}1#qtu48{E3&En$???Z)|sC7_@O#`yzp<rjXSGk*VI$o9b{o%_wyUK`OX
z#zjN{37+J%*lof_y|grDJj1R9$0A5ph#$RHCjUxr`RLh`6vYo9hcf59(VqH6@AgJN
zLf8MjnUQq-?RHtd#)b$65?~Hlp)p6E$fh=~nYMS}#q0bUB2m!<LK^KxQ2E76-rpbO
zLqV2bas$;TS+Bs)GJGU%|L1xA%{oqAlXk7xtA*q}THgDDZma`F$X-TXGp_24MW!Up
ziCrN?$k&-Fdh0*xpJZ_J(DdF~TIaI^u8(=VJbjX8$G^{CQFSaVIJC!k|LFay(=KA^
zO?5}RS+q2yir#dU8#zkK?uid{X+SJq`(D^a{aa6-p>c2sNoJ9+rHfh8#_paM76^|~
z_+aJR3P?vPK`J{eUgE!R@(bt6Iui!>Wc72+;!ex6OOH_1e0)RKC2Pgj$#;})-MEo0
z19hJcN|s_2IK*|%H>YI|`LgqE2S-P_x`#Fd^^8X(Gw!UV0C<4i^2;QZ0=O|v(lqT>
zV8x4i@y>5x(*)kQ#9AI|<335_x8FPfFP+%!DkF(`nbz9n-9Vy7(*Oe>=h#XDwCzos
z>YVx`NL+T#>v->CNN8G9PUnGkjBDcF?)A+zSuih5ZxrD~?<HWLj-NW1rnYNdq+-<*
zyPC_X^~W5*6kq3Yoj*|TNNLk63kJm`r$2nop6^y|fk%3k9B8d^u)-@)?oX2nqDPTO
zL@LgukP!B$;imO58_rZ9{Mow9ZUbIR&8^?pXqW%^RXzKs$A;^=dT`3zZ-`HJ*>r#Z
z<fbmo-{72^cAeviR=o^}NZ*)$ew_B5!50>i95uA{=<x@wowA`c#>l{Sz4Sh9Voh0{
zB)u%}=sf`602XTZ1_f-x5}E3{;shRb6c|r)`uEx@zko#VSYBITfs{hwV!_BV{AIH?
zYv$r(2ds9Py<k}q8|8li{7Ay0&G>u?V8#t}{5#P5z9w3+VDx%XCH&tp7j=Fw42Z}}
zwk5Rf(KTs(W5Gw9Hj%=6TAoIsYN3;lqaX$Bah}n_?j%T@rsbL`52jxg+r3XH2*P>-
z*7m(n%|!bnRD%GY+jKpHDe-hM=z^5lp~94Q=S|B>GXh6%9R<Pib4UpW=3#F?>Mk{q
zT@pVC*^sa|@PbK2+EW5wq-}cJ2LdusD2v~#?yzyz*N=^jsU;P!&mL5=udl)R0&vb<
zwL7lDu<m_NXLng7PDU#_t{{A`(UH#`#@(YJ)A^3msPfeFJp{d#w%#3o-7nKslQKn0
z@L`Tb{JEa$o5N-rwX=1a8xuf2B6&hm+yz0d5uc!#phv}pry0E=5w?Tv(rCUT-m?tE
z(|ctVy23%L65zajH)(08{OQVZ?fFI-4P)+4gw^3P<h02v$pj*m5#2f{p1+MjF~x-@
zmRk#eGUVYto$9xsg{Su32L~NAj{N)hl`5qZ)5sX6yiSuR8k9V+<ijYf0Apj0R}z<q
zDc6fHhn3nt$MjE|iF(~#PFc_J%=vO(pS&jXSAEQy=w8>B8P~SC`U`WUSbM?<WW*)8
zZ?o!jj}(dMM`jt20W_T0`V4-bc(AwY*w&TNmk+hPW#L6jV(tn$`g$jIE~mkLTIS?z
zoFD0hHNj=lhak$TH0B=tJa*j3^6YD%8iPHvpoI<lJ;)|R;N#yB;1kpm*O2LPmJ*Ya
zr1Byt5{X6LEmz6n;$9k$OGIEMuIrVHd>-v}OCjb!OMMG^iXEne5zPgpyNzzlFVKku
zHCC!U*rT<;iMC(nbB7i~?gxX%!W<@d!%;!QWaHjdMomSd_i`8Ee``1-*I=c?@8i&C
z-84aVV0+jxV>j2B0AAttYVC@UsU*I@P3~@DuF{>d_SD}Sm9eu^T<in5jUBG!DD~+y
z(<JkWTFHTivPl7ABle9CGC<!dEhWG4y6+#I1r}*GTBFDC;nMdCsyhs(AOntS*?%xB
zjb6PL102XC2tY|DxVJC6zMVp3pC}zqXZ`nbCE|Lg9<fn4Zd1M|z>n7CNiXV|CcVj{
zs{GVQDz(2J!C)OUEBpOH?y^%=C@?3aAb_$ip9;|jXzuoqG_iu*CHxPo7*0m!yqU6s
zlo0cbph-Gp@3?KHc@>mAZ)L{1DJG%M$qcii@0zYfs|HT!9|DZ+!^l{WPS$Wq;Gp%l
z^8m_Ayig5Rz3FM|G&nIX$t#aQ#|?Dys*FT+?J5ns8Afj8RGHCHVy>v)5T?q8-^d43
zZt_rAnR?sR^SDKgtL+Faw2C7V2gDJKt6VADqnp<Llm4WqzlN3cx?;A8D-zzQfZ%&=
zvVZ}<XkB$)n@nc=C4W7pubqCJl;wX%>$KY}s9H0Z;hD(9F09ZU9qO&Bf4T9gz8Af_
zEMS%WUL_b1TJ#;&+^ntsGQCNJ594}mA=-7foJEz)n8=xoTxXNa#YjKl`t`)YZ96AS
zeeGM}pb;~#SIqIncRu#eh?$hQ@s?70ZJ&vmeUk<SNO?7{-tOp+E`*KST(=&yvYoi1
z<DSZ34IvUAyYiR&LV%w}dnhbYU}f+|^#-^D*u*GuJiv#|yq)gH)h0j<af9qv!U4$g
z8Q%YQF{LBnVy)W5VMa9L5XE0|OJYb!*`*YJ=D=sdxP2um|GRnnR>u3_rR|7ag_F&p
z3i+1uyx%GZJKDr0NJDXqYPBIquH2%ib<2oP*j7;s1KdS9VkP`{=VL<Z+c$fCCGnY}
z+nrb_wL&@&W-SRSD?@PY3X&xBEPCrCGc7qiJKp6Dl?WBr=AU^Ku|GI(DfB@2k|Gp&
z2-O<~ul{^7P8*!C3bgGDR4JK6O$$xiKl7A7OU>rjxibW4V6`{K8pH$5wsmvgdI)bD
zllSLg8-q{M#IxhCX8FXI-u*TP%oId}uI*-<EH|)n-vQv2n)W`4N6Yzhw>Gm~Y{icY
zZgY<49RG+mY7#FIjf#~_<^mAei1U<@9_YUUvi%~c7aj%4J`}llK>xChLM4<C0_N5B
zom$Pf;($jU6EgeG5)_FJ*F7EYkl2h&uCEmu-4Da+`@WBng8MqX<k2E_@>Dy}52o4W
zK&euRK1HJ1UT@<#c`4`IO;5)<Q)I%hcnum;Zg5R6ZrPi*d3eU(5biSVPRz3{2I^66
zKbpg-7CZD|u>bU`4fJCDqNvxz^4U1ka{o*F4jpWFr%OIrMP}DY5zg{?f~A!W+@6*%
zCFj`MKe26MZ6LGC9?~G*%JJ%P%f|pV3=~3Y*R(wQf*!TUVw{vK_aY_y!1Lwh1`;Yz
zB~@yE8I~+^otp1Gy$m`TMm3ePdONM1J;qfoFK1W{D=FohdmH*FH-(frB*~!sqQ5w+
zeMH?Zu>s?H<bP>vRB~YA>n8^)sJPuT?o_>_&BO|TzNXEGe0A2pwj_1=TD_Zx_9)&R
z8IB#2x*A8Ni{mOnb8s*4Y^B_mzku757(|xWdOTEpuEDs~J-@<pc!d`*-PluhCzmbU
zZC4d$nmo{@q8LjZU+L-qGDpOGbYvU%FqPUl&D=+UcU98FK{nm7)&_VC?BN43x!qMW
zgJykyFPVNJy41qKQio3<YEu3gfsX^{cq#uHtfU!)5XaoT2y$Hiu(EIJhClN_$lD_z
zOPMNs`ujadYThA<?l=2-0haAF*vqz4nx%!=mE|<WBh1{;^(QYflvLZTrYzVm-~8Y<
zXce*W+5NP5nA-rd7K3&U>|toQ83%z;;9kwI3X;tum!L<!_2MoYXFR?lp=iAEo9+Ss
z_A{Ew7Pgt=yrMs7W6ckZ$>mV)>0<bssLh@fV(YQ}M-QH`4zW_Io=jKEHva48s*;C7
zuHZ&sE3R@)DVo|<Wwyf#bso4N5ggF&b7OOOG$`_&WTYMQn{o^bBKTAO<pOxo0J7h}
zT6?x1%Y!d|dWbu)DjFOfqatz`IK4SnKgd+tiXXMJ2n&V4wBo~0k}Z;o(|jQX0DRDV
z@=2x2rGCn875BTM-($$y5O0YTA1Fa274g>Qs%w(;b8T?%W{txuia?l`Lei$>eS^*(
zf<BSuc5b&6*aY~{o>+p7k`fBR#DR&OyIWG{!&r&LipOjo`Ya_T=;J+So{9jxmJ(Bh
z{`I-A8DP#POZkwFc!|OQ5z>In$o`j6ci_OVP&CeUV8bluU3Hg}hn#|FKJcf(AmTuz
zR|q;V_0xfY_;!*xh(9k$HP%uk%9(vrKR9vrf;a&gd-Y8s{}TNHfzKFgr4N{b&C5Bz
zP`-+@7r6fY63S-Z$#pX|uB&h4B(rEL?;P$mY$w#v6J5u4NTH^}W5}=w5a}Q)QgXT=
zkdO@HAcmsk4yhT^d>QFd&_4@%S<2Cw-c&uN!GX1zeJ6?qdo<1^p4s#tu*BNHF(~=m
zqotmGyNSYrFpfF#tROY>^DwM9)y}Yysgs_DGT~`CprytTp^;@A-#E-b!F_+J*&K_R
znHyG2Uax9YG%Nec*HRna@3i8#i)93w`sue<vz5n(MNE1D>gsY~GZE2=uA^tpNd02v
zP*K&?(4Rj$@?d&f)g($?lY&NTeXHN7owP72ryQ7WdwA8gtwef~!A&hu4ML~6b!8o<
z$LddMIQa;5++vwxah(E~MpO?4UsEN!nXBLSjLF{jA6ilvrMcm?T{r#e5;FOiv7*b)
zHRHZ#Pg}5VX4uhwU+0<lKaScFw&z6NWv};BZ=YD-d%f=IPw$WKyZ-uH(8cak`gDIW
z^=tm+QQya})LHbjs^(D9uS&z=$&T>Iem^bVSiLUC(J0q3EqP&dZuI7)+>aaVoSxta
zg=S_N@7SF70!iiW&xWREJY>lWb?Xx4QyBPcdl*^cNA;-Z(^<(buJaSm3#-S5kFb><
zkL>gdu0PJIwj4~ahDb2BQAxhE2x3S^(N8lPC&89`4hWieg_FTWaKB+goq;1bVT!@s
zOD37)&dncMqSAkfa$+Nh?pvz*a^r-NTugcN`uIRS9qSU+WjV>gUEfC{;h+5oxwHLk
z+FM?dr-5gc_fVn_htoD&!n&1cyw*8d(e&Wprr9RnE&DU#Z6~OQbO8_Znm@&9$XuA-
zaF~|X-`;zoL=GfQ2lF(y%=dqAaKyvqnSE_uX!^Dk;+p?7F!?oMCJI|J)k^BAdr4yz
zKXqO08t(;>KUXXLMy4JnHZDsQDszILzeqgww}QE$aP|55Z|?H{MGL9_K9*K<yzoB-
zA2fyAtCNkX-Sa+%G2KG%#nh^$bXRcp4W!*Duf0Ozc9PYhOH(yd=s#4jN~A033V9TN
zX<FA}@qcrrdwaq|j;7Npf%GCUkA_!!1}dzEypW+KwIZMbC6$&6%G>Gj={2lDx2S9w
zBBl@d^kTc7;}5MSg4}sy`Nx~oDJ{oOF+lL($$twT{1>`==TE%5`#)R&g;ryvuY#Zd
z!Uz6;?n5;78KjdUS_TgT$X_EV{pk$D<1Hq}&d>tSv=}?K$+vW{NMZ2Ja-m(VXK}I>
z{UALAE2DDd0_a@Q^)hvy_M3!f|2c#eA6D0A-()pBrM9oFP&a;K<2YgN9y9^;>so!r
zY?1p0Hfa@U4naYPXxsMEv|)W%RD0C#QYc%CXIFk~<*s*$UF%WT;pb_s`K5>=!B1k7
z&sp>hwGu1XO@}~f4n^?#iZAHKp?08R!PRMXpsxSFJyoZ#1XlGLT?(TS$83Tef-#5c
zx4j1y^*MZvxTZW+SY0Q*(n*jvcrP|FEn|s2Ar9E?D)~=iRJXaJ!W6GEJ^Rn<I#Gb;
z((1Xz^znw?(OWcV`QNp_?hD6Ew^?IGtW&n;9|8rTu;j^SD2lls(SzEGh}Kr>J@@L5
z8*f?n0{2R-TU~%vOx=EW577b`S@)0^g)G1jRZ9<oF@}ep*d_7}tLr{!+W|`cM3v6N
z9@Kr%(f?sdRP!0M%bI@B8%{T~J>_+l->jpOg2IA9o#fr435fz}!(6leJN(m*bUDS3
zmhaPSP2|>Qv1Q%M1=iUe|Hc&5CV&e#77F)XsKM>!8I~^+`5%o@kb;t6U)bCx_gEL-
zO0);0nN~r=Eb{C%26~W0-f-F`e^A#)Vc`V^FIoXQ{U`l<*E5?pQBbJUZ?NzpnkMG!
zKe|!Nb`2FmD(bdeV~qDGXM%P*xh4nRV+HrgB3p(1Yeh7^MY2Yz#HNR`kb{xM3~4|p
zH4#WW?E7We@FsPQivInqpM0Ty52f3@q!CY;vMQiKx9)%hCm>X-4{NG0<|1kCspH3y
zG~J;awD!eVx2mkJZB&#L{8hOV)n6^$j^t^6e!K7A4x_@3+MO$P#x``B^TKbqGukA!
zc#MZ_>kdE9`rleH|23N{?o>g+M;{Dc#0gU8knGa_%n$ru1TOjc#6c&vm(xjrE-lxF
zhIbByBaSgYP)>?}Hv~lap^!yq@HYr)zVnNy!KIP7bAz7JQ{O$qx=JWK*pAWHsXDun
z1jl!6sc=^(F^oHEwR8%$h1`{EFdKP5pk$@aJeO)4x&I;MM4Hd2SkNkQ3{)d;2A%Pm
zr=+3m#3atG`JRqHKzB|YpK+$Of`V6_T9N-eX)_=PQj)Wy`mFAAj?2*m5#opRaH3wu
z4zhnRy&_=1x7U_U_YMbJLSeReUNf$;&k(T!R+*eu$r>r-gH_yy#~}J0STKb4G$Uj_
z7Tk|XzJrD$9ZT~hWwy64;Xx8m+%#VP_TpO)rn}QiJ(971mB3vP)GoT!)^Om3Lr5kI
zlo|_zRaz!|eI2CNhIq9ELgVsJQ9t0MK}3))WXBIHZouNyy<no^n9ZVZoT=d7p3=9)
z+F1%AdW^I&UfLrMRLwhF2!rqz8fo}ZJLsNwx9^4Mf#13d%7-`~2#fT<bk=uJe^^Id
zwZF@k7Xj0O(<TnNCTY50=|A)%E~!;<C8MA+giNLjbOSwz3m-oiViNIO&5d5l@GfcW
zNklc4q02Qlp99XWq_DXi=lVi|NGO}8V6{;weCbQ-Cw7(A=#aG#O|{_#`O&4bu81c8
zm*9fYJ~M$n3I9#COw7-B1E@oBIz{-ed?D-O*jc80ffJXKvCagxicSbP5AF=@Udpdl
z9-O%<2%aKOFn{`s-Np`VT*tbNTS|XfxwN-OG0*%9XhKr{7VzYBGF;Wgx&HqCd@G~Y
z`XrrB?=Br6Jhng?Z;A;_Wi~X(oY9*gm<VnNB5eQJ)-9VBP-nvi3R{%wx<@VcLu7z?
zsP`?hqFLrU3bn%+4j6LB?@9^;idgznhFG3ztPf)YtYXc_H4ndhjQ}YQ1)N^9z6*`J
z6ehyH_o+!d!0Pz)tb67obafZ<#XeDI0i~Tg_KC<-l{qcUOWb#=#8HnlN6@Nm-o9sr
zj}X~B$oGH!^T+V*2Q1wLTuSbIWN$q=I$&i{pszx#;!(iWDo6-HE{3Oujz2sQ=d3u)
zg*S$!yoQP*zEp@iGy@cK^sG0ft$jJ6az?j>xHzeeus}1|FNn%Dv|`VSb={mA#Qq3N
z_Wo=o6=z6VRmbWnXns|y+BI1rj_D$rDNE{Y$&?cx;TeKgE>U$wx}xEUeNqFzTuX}G
zLG4c!tnDX1yg|yGpJ&h}va7%?#jG8;v*yy;h?i0GO)EH`*q=e>wSc`a00%Cu1rm$U
z``1BSZ}i1~a#+lMS||3tuZ7^)%&BO<n41U>{FZiJD+QmG)OQpguN=`@NBpzJVL+#b
zg+d|?+9|Ej++v5q*HpG53WZr9tm$_(_ZoE-@)f&~i2lU=-1x=l6i<IXObc8|(e9yv
zwck-9snvr;lx54Lo~0SVmHQDzr!hX^rHqKP{t!&lPlyTa3WroyY%YefbqfExX0-He
zgt&i3CHgAD0*OC4i(pkAuH&a<Tbtuv{UuV`GU+6E&B)tO;#<|~p=hl*Yk`4>)RR|G
z@%Xh}@f`spHE6XDqQwi>wxa&{-)!ytf@hap$Bl8i>gi=e$nn`etj<u5_n!A1$T_R)
zVs2H*+OstOr(T=K9HyI;7TY)a`&AzIsdKV30<<i^Z5QvZv&I>2_5168yx&QF&-w$s
z(lWEI;^DiSYvvn#B7Ey|&G60PB2-1Do^6}#>}VKI9jN^^QGj{ZsfoA27{0EV9;%1&
zby5!&`>Aybnag@ZLoKx~>PbibP?`w4p8H$xWr*U)zjA&kXY5sK(Qe}${+0nmOE+P6
zvs4Lh_L)ckWwBp~gjmhpQOmx6yv5!i*buZ5U*$Dy_-(G1^y_&4C6&pDLT<WuZZs2D
z351g#reRSQuTmdS%A=XB5Qz1E{oFkO8qx`8y<y>k1=8tYwl+;lmDV3%`Dcchpt10r
zhO5M%(42{|9yq57n=v7cKQmj=a_{vBHDtGT4;y2@`KKx+`u1yWdDwmuEopWJfC_~B
zaR+w1^{<%n7vr2a#@Gk-n{lz*(m#y&R9&0M|HZMXzN-O;W&~V<a~g1lbZ}a4AY6+C
zKlCfaF=y<ZsP!#zdw}y2R&~LA?rS58FKg4f(JtZnzDxN84$pIm=|+s5j)#+`2M$go
zwXak~mI$iDl$3_8uG);}e=u{q^*dLHWOXf(FRey0NYVxat16U8>OGTEYcR6S+{*2(
z&0~0yh-0u}5}2j3;|8NTq{;~W-}=wPx~pAyD$v2g)}y{cH@7-Z`+!A1J?HjMOSj)I
zg)1fw7Pc_8X#enjoxUUOvHAHo4flC$A>1MFsAYIgLdrYzMKM@Pl(RUi%$dvqWc2B{
z0umApGk&KgOZ5qXj}8O=8G%K50@)WZm!>WmsA+vFi~wG$Oz|BmG3H^4t~rMn1DO7z
zoPPmjHZrY9@M6dSqVszz`n~<Aad&cyK?y4t8~>U4+;JL~%QU4cafyO*EG#m$uI4ao
ze}F*tF8?QvzJ)Yem_{zUKvN>*e%BK3K|se~+C~=^uR_d>Eo*0)EMOS*#OM<dYPC3>
z_1{0v1YS*?LzrF*i#)S~!S#^Zzx6;!$svw}mq<ZMN`Kpel}S7#=pWJhCq?|`s(KgO
zPb`N6n0qMi*mUBGo7B~t_6>?y2|1SV;fsC07m2i0i15v}=0Rr(vI$j`zxF0gs>{GE
zPKXpmhe0Q~u<_I8-}w?Sdr@-GvDo=MeL6O*c<PfZ9p}>Q#LG^;aQS}HWwJNw>=-p<
zYh9{ZPN;^XvhR%oFCa*9beDz-Bf8farWew?fsdg&_6ItVSK)Dvrg+f5S{nGmV>$KL
z;%7&=wr5@X<z5Eip;X$?aQ>i=6u0nWN>^`;QA;Po#QQr`Cwp_BD2(}Z((P|qZ$T%B
zyaubV7G|b(w_V64bWon>Y>4gweM&Y}^;G#K@NqzAXJ}kmy~mR8MbiUKKU%Ym6Bk8F
zW$j9Eb-#q&pHRHw1TjGFO;-Y*1U<8km1H%@m|uqy6Qam_eMOB>j%B`;^?BPnQWcdl
zae4+vz*olIk7&Zgzr&~=1qGq5eJ{1{hsCSm*vQ86BCv1#_|JLm_XlX(3DXcGFZEH-
z(iAu^Q~51G@{`<}lP5(0nQJRfT*?Vpgl)m!32m0(_bHq@k<80HXdE^RNLAqj7M@~y
z4WZLY1Ya`z2nPP7PPXg}`A~%{m=n&}X}Tol6PyZn9~(t6b@JVoD`BJwK8ma70X7l}
zOFabJ*SmLr;#w~ghc#&pnvPihzvNiTHwYCS_;bHPPJb3%W|ON*uq8vGE+AgfFU9)q
zrFe<g`T2h$>(F~m4&N7wl^}v6*48Dw|DDqPmlsze$^Xp{v9bvJOQG8a?CDBwR#-UD
zJt@yaj3?p6L2cUYUuN!Yq+sFBZAB<xC>vHw5FDZmEwx!r(y^o9u-%NofsLzZf%RU#
z>iyNPc)U4sd`^qCg-_6HSkupL|9{XV)C`4S6I|scb$wG0$EP!IuAW=!c9hPUu<hHT
zJJ2y%vEe3xA6c%&S}QBn?B@^mne;AHx9T=Ywn0I24I@V!pdxtOAbc&NZar~|fpR2{
zMXGn?M%DZ)>b}R09ErlT!v&sBxR1$z-m{dlG<gN$Vu$j^5X<c&EoE*eD%UweqlWuN
zP;`#qkKdK%Eb6KJYl?3D^4`+HDh(xSYHEL^A#@EI_R=Zm)%T~;QX(!%YT6OBQ0K;%
z(ZX_p_>^!fj+li?sg$|D5GPDn>uu%rDAZi1$<>?l8d)6gr(ogDvzb-j5aVjL`5VdH
zz_n3)tIChByB-QX?Y9-fN7{5wy7YyQN-9}aLC;CUbs`cFR+BUB^rVG6wocLu0*J(L
zm)B;CasAZehtmO>S*dmlFj8<NhG)#%OH`WWH2MiqjqR>MpV2q#k9GNJ59Xgn7!q*!
zaEEoigbPOIwRp?y$~%1zNpsNBY_UT*W0;;JLShN7#za2QiJw!nQg&Rf++Tn2*39Hk
zCHp1aw=U{i)1YJ4q9ee!?HmVep_ug9O{!0!!*$=Bo4G%4^%$8eB_pz<zrbpZrAS>i
zEQykwgSW{%I80xJe}gaUVwo!~{ZPAU@cn_I_ZoIFf|mT#<VDrNwN;C!O`BE_#N8``
zkq0Fp{Cj7Z6FvcKlFGzgCm!Q7rSV2|IjDm?y_QQ|U3#pTz;(@O-ne0Xl%87`D`_gb
z?=#VOzHfB>XX}^uwCW~a^>Zx2n=@R|3@yCtIX2N~ClH44dz$v{uk@J`K7H@^g>hUN
zEn`DS1X<jsMyW9VMa`l(0O$E-mW9z{;q5YRL!=ExX`MQ@{ym%!d|(Vm6%jn5?ijEI
zzUeIW_4OWnTRDsuk#pEKYUhhH7abxR%z8gI<Zg<&JXY-!m}D-^-;-;V>Ash#5FGdV
zjm&y%#(l>d^cKovdwR;cE4Hs}*o7FSyN_MzCs=XtNS~%n96PoHof0xyoWHxi^wl>w
zAJAK?_$#MJ8V3a=0AhIs?y&!pD0U)h8L!`~v&HREPeYaRtVh&u=rd3(XKlJP0QWOr
zUDx37^Ic_^VwxeUKyI3;wsSwn32$js@tF$}O+v(5@-M3SY)}x})FgCo-U_GHV5>ZI
zY~{06Jw6c74utGY&a;?v%wT~uLVCgmic!zmcbkWBE~kT?#m!!n6zE(qzAO+bTVCjD
z5H(3Xgf0L$7IBO|`$5D!vsd2INOeG;N_e;Ntu_5!-H*x^kF9UGUP;q;JBZxf+FuZ&
zv+Zv|J!m%YpJ~%xgB?%xZLKLICYZhUNUk^J8wu-y?t4Q5M*o|n{wMbj<a%(aFHu2%
z@{^!@+=NcaS)M|%4QVY1Pget{0p|0uUBYsqlx*YhNJBmZ^;2$9GDwxsQQ!wW(T)|=
z=<vX;Zg0D8__&6YyTvNlg7SI>7b31gnzyUSyiTHA$O1%~3u}v%KQXNaD$;kFYr?!*
z`t<+T%Z;u-VOW5YW~;FsI&Z0u_+LC{l6Lc%STO(~LE5Fw<6+oVAR7GBdaU44Mw^3W
zp#+-6V#xM)Dt{V0NqxwW9~=om%7~amx`D;TZX&cf58W+UaX`w_!L#b67juA(prWF}
zB|{|GBFw6!F>>kPhS?%QXlpks)MY8(3SiWM1`aa4$)qHtNuo*&7hR;eVE>eLoK#Pk
zHCST(j)aJ|cjv+$YL`1?K_I7TJ60%`7?;FB=6;d7EqxQY`Tdep(u*v-9nnas?;~A_
zv;|pmy$I!2!m~sr<^gogr3{`W24CZI5gLN%dg3G`BV$9bAa)@25)&7t3pVlrbh4h>
z(_o@OW~Su{50r&SRTMh8-kN!-bEN&+dgsqrBale)9bIz?Epg$Wc=HhH=oO(hNU#P8
ziQN6fl!5HNP4J!d%1+>9Ch0eTIdW-J*_9iCLNP7*2+MZ=^N@gYJ~@M#%DlP0CRyqp
zZBK_KhPnrh*q{B}#c#y@&T37Rcl8LW{r*>olb}Z5VqZZ;>HlrAFqA`I3Np1c*rUrv
z&D36OAhcP0K(oi%vg?V7jfc(g2^DG`#(Q*Ul0adEd+DT#Tx>xQT-<2|Zz@QscI>R?
z(eH1V8fl{+bi9(zi3dhz#|e`A{G@sE^|?xza&(dFz1eWO+bSon4MddT%v&QfS33ku
zfy#Ja`yHdIQb&n0B=6a4^cY;BknRjttqa{1IGRD4_igrW#B<Gy-m#q1iN5R8HUFgM
zN0B$Q&lL^W%v-}_*lLmW)0qf)u*Y6z437KxL?6xGFvU#?9Tp2g+OxWj!>t&VpeQp}
z0j@E@4x3l)wgX}#k?k?Ri)2AH98txhwz&(Ux(U9UeOkCZ+q@62fGFe_Tf^pSS0t9)
zq=P8jwGBzgwA4CTKa*2W`nx<RdcW*Ki_XKryAu^XEk-!l?^2!h-qC(d&*%O0ukIQf
zV#YmmZT!0@sx-Hq<|&M#gw)L#T?!!=K3{sp55PdEc@GDra+$R0-VV9-52sMhD{>EZ
zJfM2;rJhOfc_r=J^y0ZR7<;$B#DKQ9CMv}((qV$K0MJ2#RUE%Rl{B7ur080%A28Ov
z{PcuCy-TdrrGtiZh@ePfa4lWd6#<V1Rk^K<`g+ple$p#yLXKBO;=|44%gvD$fCJ7j
zQ%gHc1M8mj0{)6f4iapT8ZExHJ_ubpthoUc@VCKE$0?|S<@B@f@RG>a7OBXQ)7N64
zV-Q4?_B!yu1fwUSi35;Y&9=xB2gUXMn`)nRJE!v?6$T!Q7cH`7(ZTW}@Nz$!KSVz1
z8bGHkfdpy!8H7F0QYOLE+m_za&);8KBnKtwshQ8uZdapuh46(E#Nt)>51CHvJj@_k
za}X5sQO(b_dKBtl?`?Equt(dC^(X8z+3nt-sp<~^mc+fdN10tIG(8@cz$K75*W{<d
zk>m1dVjR~8OX~w1wB&UW$jcUzWlA;1x4Q|y9?e%_O$cx(N6ZZkB@P_QqV#YvH2a5Z
zN<d@zb3L^JL?@-MP54hR1Tm|m%!hFiU<F-sPngZUlOcIITP*EJ=gVy;{rV3VVD=3}
z3PACH5l=V~2~#!k|0E)YVhJKkl&dBSW-pDt1sz9X@t_e8AW+4&bvANw;{0M!o6M;M
zZwzNOiCUiyIkSzN`Ip1h>|wq8jp12L`i@#~n~)gUrZU}GTnsqu97@&N?%-jVx=keF
zDz)9gxLxrC^M46unR;L;lBce0PQsuI0~U}ZbAv5Y>uc*!+tRpPf}h`;`xeTG^j#43
zGIfhH&EcCxA(ka{elusoF-R|f4Nppn@2sX9gm#qqJ_*u4qpx2{hes4Z3oW@)tspiR
zoll%uCtfOBcj(SU?J`_so7#*MDS+&)-iw+(hP-TTMZ<PBtu`u2+C=^3dtCEK-^9=o
z@17n3k*c$9HW@rtQERl4=4$_OA5EPU9Qv5;j#Agqpp&@Qw`%VU%TMC#vpSpsP>?uX
z?e-lauh72PJirBf<YmjykG@?xINw{ftqWAEpH@t{awdM*9#zQQB^073%KPKgU8^c^
z3%I8k(vy$@(zJm2^+@pSxVpRRQgIA<0lB|_<V*YJ(kY@=oY((7mA=vI8^?Z@!BfxA
z?w*`_Z7*XbkJ+^J-uP(~tL1L|cEATNZ5P*ns(u}>1?t~j(BTw;+7O4k5xU18o(2u)
zE0c6SW98UK`bkSmd#l`vcsXnGq)D{SNp)${3&(ZrVJzUOwMWxh-98IA?4gV2jDAH3
zh12d4bnqwWT8<u(-x<Gg8p`gbTHln6Vk<bj$qyTI7FOEt5OJG_X{H+8^aq<NY8C2V
zZLee>pgPM&{dpKVm73Od29C*4wXnAd75g~6!O@sJ5z648pm!!<nZZ<fI=*uyp4?Hh
z+>lw6*})ezX6Ch?%ANT`<7=1H^PD+!>JqUss++FXkI5j$+Dx=I&tSx9z!zcl=7p7?
zJ{9EE*>ci7qiZ_z6BeYG+7N_@>L~|h_eL*KW(aDehhODZe(5-Mr0_rd_N<v0>h%rT
z&PSKd6tPtJ^WZmYWXQ|hxE+Z<u_J<g!aJQfZk#mbqPx_7o#AT|g@oD}S3sW_uH+)y
zsn}W8m<UR3xYRHo)^CRuAb7S}kW82$p$31>%gH!H=n|duJ3GOya<8el>GI_9<KIYb
zwR$Ho7j+Zm$GsW8g&taM;-+NM&9Jl}V-Ur+lhc^F_5!0KYQ%3Q>ylrZKjEakIY0Yg
z{?-$3VDg!KFs^)LS*kON&~wxW;d}2m<h4b$kH=_3v(rhT?gu0^$n3GT{=@p9=%wMp
zeWj=HeW#86W`hSqeQmQz$ia06yr_ll$5v5)A04)CDLOT2+b4KGk53auadAr1k`^r|
zVP7O}-ZcEp)Gf_QPg8mQ!UjD}>q6^JfyZXiFH4TOG8_{JKjYS&UG>?^n*HGvAT+-9
z30?!1`5I#Q{n?zx8fq-L>jGh=s+S(M!<Gv-duOMW5g@{#&?12>ymnShKfr&2g#~Lp
zr|RruAF7QIg~d&fee=#zpOVV#<qxR!D%E7sicdA>P<S#kF)k6tk!T5ECj$dI>DGKI
za@B=opP0MWkZ^I$&grR$*Zm?0&9Og6^qN}VJchbOdHx&i(uq~X8dWW*McyNa^U(fB
zX=>`DqkeXO!pBw&>~|@6N{8c|mC!=sag|Zy-6tsS@lc)h&B1=I#&dQ3t2f7l+`Tk{
zFuUqz&XHyF3R=WAGkRmJ^XEtY!F~I{ZmAh^r<u#g^UX81IBP0h`#%#9b%Wder@qg$
zU8|g$hjwtrV1Hz)xOY&)%2XkI9(nr%JbB>yF<=e2_!U3bXPYIF3g$oWyq;>!f`U*7
zl}a74XQf`lEJf@1ch3czq%&f1VnS9r^I{fcQRIrteY7Z_IGPx{$MA=h-Of?LH%qJR
z^5gamlG5U?AkQlAa#8)+d(KTT*o<*q7i^+jHZJa4B+1zfd#xN_!crVz_QtGAPDuK{
zh9czPrQTw9@vGd8VR$*ZX-x}bvBau<4Y0KFH`ffGPJb^IcdyOAEK@g5v$nLFa1I~p
ze!Xn^o)*Rew7R#NJa@lIR(Fq{2jWy`Y1SF-wis=126l8a-uAs)()ze2y>?u8tbS6n
zty%wFpF-J$&_dFL_||hgSHD|$aN#Nf8WD$(*5ja+O_M@fr5EcxoJpm%ZYf7iFC{Oe
zhihV_wR``y^9^R5vDz{H<7KluvBNEK{5Mt4jslv|zM8Yj0O+SmTJ8J_t+jt_f1l;z
z^O_#`QI<EFdG;HuUj6IMa#$D~;KwPU5G*ndoL-%gzGpeE^U=B<Gpt4D+X>Er*>&dF
zF$FnDM{zkFD!>zWcxfpYbHLuU)S-T3gEN%Ieq7gPXxe^7O9{O|`yH>E76!;bSeZr3
z?Wd#YCJu^vfZ0!Cr3u-GGQwPg%WjMrwWPJ}j$!C^8jWu3=7#$Atj0PjL-hxDaj>7%
z`)O!H%aE}S_LqD|Y@KiZxpwPcBbr~DuHU?N_Mp<CuT(|p@@?n1cCSAYMysYAXiQR4
z@h;i*C!0$uP%ZlUfaWmjQ(!Hz%@r8gWyB%agv_qzFQyekx%y2!BPvyB<4e7b7i{<}
z#&eE^X5y)zB;X5I31!JV=)eJ$tA6Jrrcc=w{Z%AS8+KjKBEVsn1~(HeZnd<x1oj<b
zek^}2pvd2wHNtG2R$*<h9y<L8UXl4RetY1nc`i4=d)P~Rm0xEN{7%kU&dr<H{8P>^
ziX#Bc>9Ju?=ix)IUS91Rai#Hnp{cO;oxwePe)2yq%(;D?T?-8enQAUf7Q&zl+=X|>
zIKZVCF>IxmP3b!R3D~8%m-?m^9;y!2y`rW_%4<K|%~9q4-lm%yXn3i1Y^gu^^xB-~
z^?S<wC+@$L)yT&PFy6PTb?p)FJ`bnb9JhIMwuRl1q3Pqgwv7e4VAAYa&S+==mPQ24
z2q6grUCGi2lzvlg^RTmsw<^|k!7Wmxh{vSqAQcI{kycumgr9W2|Fb7>THDBO!Luq3
zRQU}!T-$5yUS}VjGTp<2UFjSP44$~0bJLk&8cyRF2b6s$c_f9X4)5BBt54xO5);U)
zip|2d%a9_!J-1$uZ2INaZfm<mIyA*vaTC?XDG;QyGrOV<jq8$yh(i6I9)DMhLg4`O
zdb)&a(#%zG3sc*EuK2;rmna|WDE2OlbDLaxV}^M}U}oG#3)_r+CRrUOC{E~eO#kYZ
zx*Khwl$UaB#MTG`@i7-|PJPIpwM+XF=l%-KA-tk3IlF#;Z8$s+0SQ~S9}vPmSx+zQ
zXG5gyjOmkZ>x!1rH+ks>3sCT?SDEgj&|Nk=u3q#~1Z-%2e&W$Wy6a|vFN8L?-4&L0
zB{0A0_Ndmkv!|E}s0=JGxT5}h;-erE8vqyO*PcgcBLx{r{HgkCe@Mdx;gRO8udTs<
z>$p0~vm_>ra(12X$-zJSe7(`@LHam&Fx$g-pj4?E&3e?`<8kPkJ}30AUR8g7e11O<
zi;%JAo9Z_d)@HYR4tG+Fs*>_3AZPzE6$|m}Qo}X$hafJQMUv<woQFM~Ht=tDCJ$ux
z@XM*w=lmeORGFhd^bCil<+CZnvb+cJOVUCkO&;&A^wQGO8tka`?RTiPO7`StT34Q9
z;Lk;CetR?xr~WkMGA{q^3EzrpoUvl9PT#)y_&cB6yAgx3=8d=A{kuQitL;8*RCXU|
zcB7*2W{}EiIuw*|1r;EY^9zhT;pUawc@4DERo9LE9<k@0qxJR#$>0{u&wf*5?;vsz
z*NqXdOf>5?Pef(yhpm8rcK_h6M`y#vjc|xi-I$vD+@+V?3i~l3?|i;DXlg3jOxwO}
zeAgkS=OVx~hk_QLw2tOiopV~5s5MMDTlCvuFbqjv1ixTH3ld<7X)xf|(K#{t&YY`O
zPYKaTEZ%(py!ellg|JY;_K00GZghI+7V9?B=o-=2;#Jb?{H-9!)YC<A1`QIa%}%-8
zBtYW|bW9samgw}lx6j7?6+~Lw&L~*t=DngYo!9e$*`8t;anB}i&ZRj;E37z-7(wcg
z(u|~9h>4R%N(Yxo4R2sGB4R%iq_(IPtt@m<cNJWR^s3iMGQE9D6?x$i%Nj_t(a7q(
z^}%>&k=9(kF?sT2wO$hwh7#;O|H)bH_%2jky_^0NMXl~${XKGqZ|(Ug)NDlk7Wz?b
zH0xNxO|HWud)t}RQ}sO#V4csM&<+~6jE2iR=)_HH&B92(q|p?Nf-LUxY{L40{69=d
zy7SBS=V$^eLCf18+*T56Mnut{nO+5ZumL#zc;|$9tEq#I9vaCMiHixfFN15QoTQ=Z
zt4i|VJ3==B9`q{&#M=G7mS&Xc3$h6A%5IK&dXOddFm9p!0re1xUTpRnI%Fr3zl)+n
zSQD)%B=bdhQ7XbIX_(}H_i@#PMn-)*;}p}KdCcDLktF$n_|N4J<XTzFy;TR;EQkPL
zb*WH$?_&YL&};<`)TpNSE4*#9XKVVCgQNV?@!cs|FiGn=cn$W*(Uui+pTLR{Fx26k
z4Zv7uc}I-~hZ253L7^a>ND1i=+p#j@BkMyDy*+_x-oa=~6^x0?^;b#$luvxpbcKg;
zV&rP~oNYsvza8*%{xe%DGkOpEC7u$7)AP$0rBgd_algL>BFy$`<X3jSRJ^prbw-NK
zCqy;Zg{GHS^pr-tumFXoo7c}!6f5r&KxKoh!#H8-c|`NTaUW9=Fr-dG(r#9W2&|uK
zC`;6^$vU`N;(!9D%RCw(3(CO>8jW7XPZ{e011vtGVO73eEwSE3zB<(NMIAwLL9&!`
z^y?=*e!LMwr?x4fCdA~4i$~pi_nB-SOb)Z7PrMvQcJhs0%jmBskB*w$jcmmAaqm;6
zKYKT6GDh)PPmi89emQb?kIx^1C3~lC6|%bW8iq``O`?i8a+4Ixp9jCeHYJz%UOHau
zy6q$3xYFRH`Fjlm@}q-TZ34Ap2d8VsbsgCD$D=9kVfcr4oq(<PuKDl>GHhn>n`yq5
z=Yco{2PO02R3#gC%^kaYWVZJembz_bg*3)iDdq|$v&#TbcMcvAsjN~$dr%9A7{(km
z35X^>|MSSwRq4o;nWn<b`F+G4E}snaYl%1|><rP-f+hhRk_DrWZuiH;a*&b+KrJ0r
zW#{EYV15E>z2B2=Y{=Y<ICsDNIS&N1A?tt*3W6gk23&<pm{nTRfIVVGl#5(|{4I6n
ze!*oHh)9RZEvH0|mn%44GX4ypg}9d`A4_48#_MCMpJ?7~b^Ga6xE-omG9l{wg^T{8
zUJnjrVl3DwoYv$Y5g%5?8kQH$C6x|)ZL`l}bp3rMS006GjAU!7>XHrLF(LZ1d_c8K
z{Z47zP|%!|QFbN{*@9aUmYt0rg#n3K{}Bpas1Y!X&gf=#7aeF)z2sqGVR44hic~dw
z+?JYXXtOLXpBrGL-)2Ay+V8&<7-7>sfy^uj=Im;?1lSOh;_uLty;`w^f0pSd(RZEJ
zGTNkTZN)=C|82vLpEx0<jI{~GNMG`-W~_3mxb~Yb=(O%%*uh^1MMAKBiE{B|Y4HcQ
zG1)A6T>A_CsVLEHOKavkJdDpA$u%^C_|Sy*>Jw33nexJII2}ihhJ6gYeW|7NU7Aa<
zu6Eh+W<>D1s=bqUmxE86D!%2^BnFX=dlpDtGckw{Wb}~?(TP!uM$SAuX1aB|Wk$GC
zhf0HOD%q|5*vGx91JciakxDP(bW453&Qu=_=q2o*5@vCvcd>PRdY$$Um)OJ34kSb7
zdu|ba$F~_F7+No_>5(7pYDy1={Bk2&5MSN0lxnVG$D!78m-d<L1tV$RWN`i=3T5^|
z{@K(&Js-fQ>>t#AODm|oV=i)l8R4{!rNrynGJ=fc5L}=Z(G(F<#}qW3u;~^XB@oK!
zF18HB0X15*Y88bHYzGSFBAEzhn#hJU{{xH{CK>?6-z2HNq^1w9VbZP%m7>i!ZC)s-
zX!uKG$`kee`kVT56AafUG&BBSjksPiObl>?f}HdYQ1cstB6vZP47)5C(xF)xsZ8jy
zQmP~}JdN=H-*KKVUcQtmFldI)VVugrIqTn%u4DzTntIprJa;g3@h?Eh3daB7p<mJX
zFTK&&GO_|VJ)#;Q3}pr&3GWb1I9@{+UB<hsM)#)#VVCW6H<d4ug4rm8c&f#1HqG1;
zQKztUG$za76E&D)d;co?hH+|kVTzIock(0x8s(Z-pnHx#GPw7J8(te|u|Z;ISIS&w
z7wEfg7eVac>Vn12!^V9bf)}RRX0S;B4662k4DA6_;J^Xtt^wVXzH2$gyaN?JBhGk%
z%(<6|I3&282QMEx2KaxU>c*$p|GJswngB0N0^Xu4km2*}T^{+y-i-&^e<!4SCiRw2
zfa1@owZB!9zRQdoXU&g|JxGkhPQ<rYcW{{5Jhb6r$5*&a;Yzw5Y+A9?m`d8MohdXA
z7ccwtcF@<RHK$SYtSg;Yt}WRWm^(FZ&{ybWJ=$4I*k52$I{0vZROzvJI<||RFRlhc
z6w<L*ZDAj}O8><*>#*+Xs9^c#0|5a;DY7PZn62ZFe8QX9njg5w@V}j>OgxFE)fZCJ
zW6L6l0VG`woWDN*^$N0tOO3qGH58I`<)%($-7~JF01a~#q+C(kOA^HPWzS8N_HD_2
zzS_Y)&|;CIR#1Z@>k7VDH5r_wGtRaCkWQT~l<RH)=i&~o1sT4PsdvcGxqCt-m{)ae
z4k;yb`2IH3Jx#_aqGyKB)UyRs=a3p0HXFxt$Ur0{-BrHM1o#lm|Mu-vsOQ{=yti%9
zFVj91rvm)_2P01jYhf*Xyq~uy_`Sqd2%_oMcU8{RY_&7Ch)Tp6#&Zgp3g!DI90yj;
z5Iza{qy!duOa#*SUJ+Pzj_Rfzg2}WzG(gUi-;=D-XP|4MdJ}_@5#@Twoh88<={L5v
zb1!%ojxF7D%8_#ktaXLgAZq;bz7olxO1{CNy3n5Nn$4DukK%==)jGp*LG30(i$m4{
z9XW%!U58MP;WR@z34<~l|9vLY_i*O8(Vhtg<-R+5N$|L5b7XkGf5$>FNR{&3_K(==
zvP<{nUn5dZ(;P3fbTVzrAcIQ5Yfu@nuy{-#TPmjtj>$24<!|}Oxs(9_|EAMj4z`+t
z2a@!p@cp_(-|p}M<^}qp8ny(ZgQIeTL?~)|a!J5@D{uFzmI7~i?KxmjJT|dK8Z}dl
zy<iADFwg&^@Plqp{?<`z2%~=@jgOB-g<u8|_SV{mX~!Ckp8Bae%W^=!%a3Ae&o$mv
zpPA`9ntF9pF`GHXYWazq|6J?VU*%Ur<*VIZf3dc1eP-&%YxCyxv9fO6pxre09^>L4
zPxLixmi@~-XmQNT!?*T#KfIv6^PQR_f%?CmK03C;GUwT?-0$zFQr6|IWcnlqYi9S8
z*5w<>sLlf^Nvtd?3h4HOg5(BAfNY6S1hr)hWIkn0?5jo2eK+A}$aDg3Me}PeP!=dl
zB6YChvTY_XXDqO}qW#?EW6&m|KS<2^_+X=IOzmunA%rk8gM&jsVUgV|MXWKOO1{$K
zO(D`Q1bE2_b)`y0-do8@g?9NGYcK72oGwv%00w{0?+DaZa9vfas?T9x<Eo67@q<d@
zWUwSOm|V#&ajbAWX!6bA$jXl339!C04h#S|wjy3zm@jBm%81UAA8QC#xO75>kHzE)
z@q%NE65jlpwjBh16ex&q<^i$tz4m>)!=N$zK=V~p?;RQN=6W~1_?|Qi;^b_jol4Rv
zsQ}-eV`YVR1c8cw;=O%fdq^0{KUxfU_vm10{o^|W6ow2Q89&El_oNXoA5UsAb%csl
z$c*C~^$iqiEt~4D>pj=gZs~-t*6Y#U>-6}GRE5s;n%Y)+-@KgEq;qGA<+qSj7G>8U
zq22Ba9I@mrQU#ON*9<Z8_dm8_LJO@H%Fh>!b*^{qbTj`xL)X<)sL+;n$Omofi64#G
zyy;BN<VDCem}l6*o%%w&J@(ZCMO#tHV3<zaY|*hh)=WX&z2mHonjvFDx72U_KzJ`F
zM!7c?*=eUIcrbw#raU{foXPUPx#DrNMIf5<baJUtqYJP;Os{$)tEDlh-sAd127lp!
ztVe8ZJdc+DG0>ZedVC)>6f4Fa9bam+@Hx5ueai~8E!R{3KDpYc(P_%99DyOxU*NTV
z2hKf{^Ot|arS?mE*F8^)JddB^@PdnivPUl@T@;9z_xgK6u#sVT1v?uILmgnKX|AJh
zM@T^^o#j1&CS|_rP%~ef|M^KxV{~*BMs!Fe<&BPxo{}6^(A2UXhvpScF3J;gJ`ce{
z0R|u6*mlI>@SP`p>mN~Ws@ZA+EfXV7OBHsCWjzH&qo$f8bS|By(~+MBN+4tPbw910
zoK`FIKF^#o85O-u;lx(AJuOG44N=g0>q$QkKiz!W)3Cr#{r~<KvJBm<MxW`WFybEO
z^BSPrl;rvfO@&&b^8v3rMOUG*;61E&g&W)7Xe-#+4K!0;U191N=*59=e=}R`-pT*4
zkN=PM&NHs6b6fY=Vj&hRXb{DMQbe$!SP+#GtRQNzfd(vy1&IiXGz+4F5<51eXhiIa
zB8XrED+orcC`E-8EEEv|k#+uqoxRV#=gawWKi#|6Z~t}@g}LT@=lhQGjAuOKpX20Y
ztMF-K{XhRew`X=cNM$mXj*6F^sj7<gRSeK~_D+6Z@c~q~?ToclbWK4fw-@V;+P5=&
z@7}$|L%CY*Ubc7+)Utf(GFysuXM-{`dBK#wr`c)C>~|WpN3rGQ)9gQgAj|OY>yhy5
z@=O2Z0{myO{?E1fAM3Gedodl_D)TdhIz#YfFe!+!lt0IXbQyk$tv~uZ<%*Rgw>k^y
zZY<4x^trs;&Xj9JEV-MW&hESmi)Dv#9`U4%{+%Zh0QpOb>7v*bN20H4c)8o=MzoRj
z+Z^3MWt2&OT6TYLC<UDJ8(rDjP;hn_YpAGLw2`)Qy-cbk(c6aU>zRX;pPfBaYblIm
z!@4JpnPD&D=QD<%({}%LI6uE47}|a}V?>IC>(-;yBPs)6IlbnO<UH=o=3tI;jl<zn
z#J1GDWyeX`azIR?^}eW*_dPg45tpCKOj7H8q}#?A+swn`M4ysp*YO(1HLknzJKg4Q
zqrfT&YT(EmxyZPH`imSL^K(8hYG+!3e$$a|bFl=Y^c`UF067PmoaN+FIh|C4o<g^E
z|M(SBYT<4Knq+Rnj_mreekq&+o81zQJIHB48140cuY)ICB$1X4Xf>4Jkc&A?_$<-@
za^kzC()cFXrB-HlPY-%qZ*9Wn2Rt4gF+GVQzpxF;Ch`_;h~*D>{2mqzno`^67LnCX
zB{Z2{#0__^ztR`NbPvk=I?tZh-;=i7ldZmXr)^I`>H&08O@G3^sL8uII)>7T^d8L>
zDgCqf>t6k|kTTE_Ycwv9cTJqu@!3*%jJ^8(_itIQ{gy}%vP_lfHIP64`{#qnQngUd
zbRQ-Z&Ug2!+hgK3mp9er`)4wsmdZu9NkNU>%v-3ajNSF$w($(vOgLG0|HG~%R=o3Z
zt9apajpe?#hz=B1w*TSt?YpQ1?5f{VNAD9>J6{TBEvz?(Otn!7XteXtq3>k9N4?cm
zY(u1MSy6H4yN<ADVI;x7)tPpoVL$<Oz>9E=#5w<1Y1P!^4sll{9K8u&3^*><=RwUy
z-xe3)boSr8vsj{0-DDm4h`W}5{^t+=xAlmSSHrTdqi{js*sE{~!hX=_`2Z561GYbS
zKuP~5+cFujoKtE9Ls&=W*oK8fMoraZ!8Hre4Cd1cIv6b<aG#6{o)at!Lt0=@ny;n|
zFvXy}tG7?{XF{r#xi4@iT!8&309b$Fd__F2-|xini=-w%97$z}#C(b={#VyhPWm<F
zun%W%3Qbtxy`DEKLotmh^?&1^KpP<<emjb1@ZJ}7mu6b?8DL!?u6%?MOD=@c)TbTe
z4Odg9C3$IE6hi{ZFsP_oy|KkdXLvh0I>r{MTx%dn9UXxCxR`A*(qtDn;8`ERw|s$p
zqF<6ZjR0_k#)GlYD-2+%mVy!S@a<NAq4*#&rX)s-`Fdb5V1jAX=)hy?kXb}tL;k7E
zTi;s514iXe{+%5!19-WJGFSOknN4dKz#=qW5K1$sZ3pv~NQX9zHgYB@7XeO@azXJb
z267XOm>VPA8j7SmnbUMK6H+EDyo}+X!_>mc@a|`oYmve*AljPv#UNmMxr#{)<h|C%
z4ZEo9vLd?)nPEcNxLr{UAg~B@j<zf!m)0#^DYPKsWds<jzKB628InPbwOJX!IAK9^
z(AMUS@c^rER_h&|KsF?ys4YVVrXQ6RHc-`tt;<pC+Euz5?^R5jL4#?!5x1U+8@MNJ
zu98@zmQ#3vsx^7q<jiv@Q4nnZNjfXMQ=aCwe<Xo^E{*9Jw2$Pn`G0tW{R>>q^N;QV
zNn7fXIF0D~uH;JT<1W3{h_+VU98D+vd`1_Iy4y|~E%#Ll8MN}5bN|DrDoQVt(@4m{
z3AjLMNTd^{>?xN1H5O@oT1Yf&J8;1RX68ArHDwmn6Oe&08&kdq?ees>$sRtve7bUg
zn#)PpwGyc3VMfNb>c;}9U+{$`$iNQWTZ`dC=Zm~2X?ANxjI`(D<C}X*q8!~>Od#|x
zt<2!wQ5zm&x%BF{@A+jEk|<6FS5GCFjp{7;hN#XiP;#rDbd|`wKwhE^OaS3C8yOiz
z+29@Lpw8B4X0q=nv9Y9eR)f-VdQ#^5A~lfEp@m_r<U%I4a;8~)P=aJi0#Zoj%%=KB
z+WE=z?4RnE(dyb-cig>`eokL2bIP#E5s#*@xk|%%J*b&Zdk|hdq~X?noP@AL&Zk?p
zPE2)42ANjU4JO=qz!NI0Wx_@oZ7P#yH*FV{%~z**h3*++Deuu@SS_d>80!|lOVfGu
z`8#KYj5$7E{L1f@fPX)jW@CLyjYRTx=cHkJi7y4FAvv{*w_(UBTQ=(>u-k};pXii4
z#qMT%Qr=ie{R|{j1|$;o_gGM*_~}XPv_6Tm4wei9F455`qJJg{Ug7OfEv@fMVl!Cu
zUN(|V91($J#c?SF`9J57g}o9r^awQ}3M<km6IG&{I?lsy(IUQfcG4<$3+K%A$q>NW
zW}OG2P%MLfAgr*070vd>{%g~zEMm!;SMm@8qdUvQUeVVuuY+>fYvj}k9I-=8Qt|l4
zMt6R#`9*ZZU8+ItqOS;dM7q`~R!XZCQYNi61}Tp0oD>U3tjCmbcUKdb^vX#m?yx?=
z`^%i7=JjF;>5s=@;Fv<oS7R=W65*wFroijr=<dA|pH55E)u<6pO?=_h$)o@fT+2{f
z&H#s>{|)wez{bD;y{C_s29415$!B&QB;2sOQ6HRr#2x?WT@2H>g(-8toIJ)jxBxGY
zJzKP+*Z6heF96OwoGS`yM7k2u`0Y=}+1R)N?SN83_D#SlVX>^fj?J#W`T7uUCYHtd
z682Hg$G!~4fp(eY8-ul>X9+7JjyavS3;m1IwgcRVb%#C+k$~R$5XZFe`V}50aahDj
z0qOv&G(R85J_h<H@F?2JG6uNa2lTi*R6Hh5YNt7mbHE<;XNXOYOULyxLC9ay^jNxQ
z>LGc+F|r5dzmnomg!Wg_vQ`d&gD`5K!frl=5@ad_^(^uGdPPbLEK&1i^Zj6YZuDP$
z_AnP{wP}|05AS=z5nX&bp?AGM6<U|Y8@vM~iFR3+$%7SXc8lo?LLFuwjWIK4&g9rq
z01|ATnLSx-kiU4<9RKLne}So4-0!|z4rM*KaPKs2@ntN4bo4z~IET4RrQd=J4~+En
z_OSCoazt!i;)=iGaVTz5esD{^s~kXm1bf~IjMgGvcmeo&pC&ckJ-O_ro04GLs`F>>
zIEnk2%gG`+dp^U|(L?vsD<OmbJVX?Vsg+EuJt|jKeDmD%1F-^)YaS1d>D!;yta}&f
z%hj4mU>Y!>d(WQHpgl6~7js*S?w+_CaUN_KTP57|r7(xGKFZ?)%S63IFyIB8mRpC-
z={^!z78fVen)s>VnQoD~hDoJ20QA38duTb+TkAKVbIT!2&0EHKO?ubORQmVVO>{sd
z$hMN(>`)Lxwo^to!?x<Fj*hR{$s<|TO_d1H3cPqNpn=#^7)(ePzQ71_W-f5U;DcaO
zZ>X4ns(l+Xp|_s&|G;%+_u$&TNv9s}QM7dNvEtC#7PJE4<E7*v`17lkPc5~QDZPQ8
z2E}V0DIy&hQy&y7-(7s_*coj6U@K7^fi^x2LXP4X+7aMkw+r^q7$BNR6sw4SMnpva
z!|1H9L3QnSlLLtaYD$SIBoZJ40{*HEmnFfS%p7oGTmprH5ggb4hxMoMoe>Y}qhN`r
zIERdu{F92=?!9~M=N3u5kD09ySP{cFT$^w-lMycu0Yag`hdtbL2)kvwN79&x38<eu
zh1Dmub_7{74h%6WYeXvA6|fmgF$o(?deXeB$brJ{{sOco-bB?ncgF=hC(MOoSVaOu
z^MlKRYB}%MUK7VH-}I8XN|WT1!J{J9%}3-F$kjjzGY*?6X$;A@dsnX_kH}35Z^9;|
zH~j*khe*g$?XU%o=1dj6efxp>9~V$wglHmD{8$HzoD&!3iGio$KSKnj#Nt{8Lv<!R
zJ@Z0&{2#%Sz*5{f93CG6JN!JojA`J*?KG8hhsur`<K3g<l7_b4|4BmrKS-$Rlc9il
z#ZS-Vl6nYws2f7r{=GPxWhH1vPy`<DH&VschyKCU<nPBwQls!+LA!8SaZrc_xV*vd
zT0N*Ff{0m0Jn_5OBtHMoBExwRLxBA&lp;z!)#lg(8L?ec-}xS^)1>JFxU%_XH(|dD
z@p_7>o^Er1`s0zi`h97oJBXlpR6g!lW>l-X!79Hl(;ud$2fckk*eAjiH(L&op}r(f
zkYA-a38L@|ihWb4{-CrHPCo+Yb&6@a@(ZR}D$An4L9^#GeuxlIDqv7CCCbI*3KPiW
zw(wbiiz1CE^7)drbX4vde}(Ex9R-<UAZX~(FgvG^vr#mVsc*#aVUTY?-0Vz8n_NB~
zsTo7QoJ<8tSOfB9ymP+n32i^p|4k|?<uB<ZH(BT1w7p|jm8QF?&vnm8DK_8uqu!_H
z^var4{DvJi`n0#F8g5bbsveRf$HON*FQ;?WqE<t4k(kc}oa&#4h{vYqDCDy~;MTU0
z_trT0;I0PZH85bk%8NW`4wQKWm~ZM`j3Tg}7uQUNKdx&h%0uM8Z$dp8%OnI+YI>nV
z@3=82V^bY|!8C7${UGjX^U%*1GKQVluy6n&+MMsj_GJ%?ehum}nBblmY<FpnwW0B#
zz_w>1TPQ|rQ=$Ls(NQtUROCyq<bFSmS;kRptF}=zXnBQ<O!ARJ+wK^1S$60-o&94T
z*_M>|m0`|TXf-z<*Sk^sJA{)abBThQJG79dwPW7Hj!f|Mq_`S=a~3aJ)X2LkOrSMv
z>J;>tS5pLuE|FG3i(#-xK9Z7%2$<c)!7}xMK;WGe<|z48naT`iJT|iFg9<MQMcv{)
zq0GL6dR?bN5%8iUC2jZO9M+oZOgbndcloKS^9uKPTDOld&zsdvZ-<?EhfF&cm=qEd
zut)pG=L~>!LH;pfyJbNcu0x<YX+E!wWc;!(advI_-W2H!JGiX4i>-;2#>R+H;H@S-
z;C+rP*wnK{e;FYG<9zp#)!Q3Vc$N9^U~=QDR!T$wcZi7{9eiULqZ3KCo!_^9Lh|JD
z?hA0Uf<M%E-QQ-;5^E$63Dtr*DN_GM!i9iQnBIOt{AbHf{qle?#%ArxpS$~{Mr=UW
za=A(H%ZGF}U9aQ2g`5l2*8cJ6tBqRxOyq0ZNp?qa%NJ0A94}`v`J)73_Z~J|rq)n_
zAHQnyTS8AEb!)fk^u5stj=lD_IW;mUoRZFc|AX7N_h(Pyeu+Z6aIDXeHFIQHg5PLy
z9YNA8o#n)!eh=CDp#U%1PmkQZZn)WaY6H`^ulOaAmTu6coM=s$4erBYyhe4G@V<pJ
zk(Sy^HDUgU+E-*#>1vWnZ;yw;+QZoeGDwQRpHa_{hK1@c@DjvGCr{Co;L#lhwEM<W
zxY0wZs02lksnb-L<#n;;&?={6#f+{hGG=v?1;l$SrTXgu1dZn}JXlfw_O48?uX_XF
z3Ztxg&#YzALjNFx(*ed)jKn}Y*8~O=kYoGry=9S~w>L)?{YEk`N#y<$eX)dOzOSe+
zlX;7dgW}418S*2&>zQZgC%tYE*pt{<Hw5P2;HY=>is;@8SiO2RiKfB5X4>}L5<IEl
zVCK@<$#}-&B#yGYgO?7Yp_8cm532fEK~!F)_PzPX@VrymhF2>dMW~5Li;%CQ&Vd05
z_XQ%H4E7>N6zK*KDEg7g>4z}m^a}wN#@oAlP5c_ikY-{*Z_I+(B~4xy$<gayu|hYz
z?suW5kXuq6QH)_PAxo)yg9i|8hE!uSlQ`}8jrtKs#{mL_0z*jF&Xa~CIYHW(h@1%i
zvHh|*7+G_PXw|k##{kmsB8W+apQ5w$P7xJn-F^)+G5Y47ci7SWow@|HmD_CHWmpA<
z){GOyT0SANg9??@=1QBwckjN_!_o<*BzBWM91!6tK${KquXWCx__<XNtSsr11zZUJ
z#*^`PyI26sDDN58N~v52x6nxA)FvJ2WRON+`i}Tn0j7Zgp$S)^#U#R7;s6?bf3l<e
z^t*TOe1UQ}e?s_!I!uz2%C69agpWdVIuK}Uk8zf8<o`fOqFdBCxNer3l_UV;`fZq)
z$&z@}n928E<qeYG8nCt4Roeh-@NSX@aa{Eas&E26d25@JFHGmA<Zc)(&aXiN$VgA0
zxO{_18DK}=vjvEH<rpBV##@#Kk-z0g>5~dJxp@Qf#1xdlzN0nfrO52QAPC%2MzZ%m
zJd;^R?vp<Hqp)5I8)Kn`Gm*Bu1&Ak2yJYijuz44;GwxQHMK_E4#Qf9q(#N-MP~RL{
z^pRvQiIG)phw4L3gS&|rexNMJbTgo1zWc=fKRwQDd^|EW*2LB2k*dpnjhWM>c@naS
zoKJaJ{84rG8NX?CW!-3bkyD*%HLH727tiS)BPH3kO1Z}oOd!LGxPE^Lw##Vyw0<hP
zx~hG3+R~4SYz5(?s<fiT!aQHd@s1qHI|tBBe7j@(x<auVMTLbwT%CQ(ma>YH@9D7O
zAg5Y-CVwiXkL6*>b4puPxo=R9QP=RKTR7jQ*X3O@<7ktzZB{rYsw$q9J@G`u=noxo
z$eW>muld9K_hsq=iBebJrAsb@w+WKuvOzm}ad*)a6x9ZkDNJ~NPoU6tX!cmOt0AX3
z2Ti(uVVqiOwR-tQw_HCClv>lYAFW7d-XuL?)XDZc*=4V;cpYZw-$8fp$`3#JKIV0n
zryK&ryRCT%>q9=X&3|wm-Hc5GA9pAD?}XwXLNaTMF1`Jjo00-;D{F<|0M3=lR3ksN
z*L|;ND@IGJyJD{&(NFMz1<8b0KPziQcEM6>sG5WW)jO9nuB^k?@lq{{+<V`~uC13e
zPtaQ8WZ@AswL>@KK^84dpDlJ7tDZT+IIfe0vPQ?QDZaGRcV9@M#tVwVI=p3kjjG0o
zi(d?LL58A=a`10E4mSGT<V02jT}QHJu)j0rSJJJ(&mKwSXR+}GSvQ?SYH{iY3?Kr<
ztxb3HYv-H4@GprnWlC9HR3r{r+V9zxTg;}a*;GO4C+!<bA7|3#V0R&;iJg3QwtL;Y
zM>{G`o3gE0!_btEf05*KEGzJDX+A=evQe6qxeeVNA-cU$NAb9Uo%Nl9wn|4mDy4dF
z<B&dh^axtp3IxjkIbFvy9IOR=9V6RL*(DSO5Q5c>O+GeC2HF937d3;z3k<J<I&TKM
zh%*^nb>=+QQL+ip1MtD`4JMvxOu_ATPIqA^*t{)ezbX;wU5YEk(v;hqNb>}CLg=$F
zxF_f36G8rJF<zPx{Y)(YjWADmfG%11V*gDdkb;#e*FyXxnlEVTv8qbx>I0%Q_Tcok
z$0J=`4vuuMKX5b4|9S52(UxMASO!CA+w-9dz9Ct%{s*H21m-OW%u^TN+N6d*UWAJg
zh9aH;EFWOPV8;2kKnEt2V#>b>q$xD<wdOKt3yhfBNDhra#tYN@J0qyvO=Yab=&2b8
ziHBE%lzc9FsD*WOi4n9NmBLObhn>#ScP^}Pr8~|7t%x#EJZXVpOdY==z%;?Q40h`s
zmfq|A1}S#~vkM6Pgi*%3cV8qf0<VNJZw_^6o?j64Xo*RhWE@kJyukDLryD<gyRfn7
zCVV)5U!+$gYFfAuPg+gGbTi8!x-N<t%L6>1dT?5n$RVrsZDQC09i3|dncLb_Bh#^s
zO3IC`mjHr-4HISj&t|Rj1iIY#cY4Yu)c#J^Cu%lj-@pn67f!l}=N2uXwlnNINlcAN
z`J=d?!URXz@zpabs21DO@84eoUsNO^u}}vNh;)n~UlN?t5-5NmsfuNek5LYafFPuu
zvz~$dfMDTGYawC|<@?}^iJG#syg*799o@RkTYJ_7ttKAatmW6QhW%z9ilT`y=Nb)5
zBIb`AzFp|c?45~)Z+BR<oH%cgaYbHi<lVjzi(Fm)`b?45a)>a?7Zg$F*EShr{nJUW
zQQMC!@5St$<(W&1ckJ2o3${6|qxHtzv?2l(>v=J#-_#QNgnk-K8&sywTmGB*R0fGQ
zc81oW;j6QZ)=|CksIvv>{D#)aPmh<v7UXuN`KMya*b(^{;}YHhNYwg2&hbiN91b5e
z=|OXyGKX1sB*l&*@-Ltvv%-@?RLh*7Kuo>uQ=7WAE668C=Q^x&`s6!;FK^g7q&mS1
zN_75^!4JL?CIo4PRT=+9hvj0AhmD2zm(?gOshjC&>X0uW?%xWayGcK%UNCHJ()d)v
z!t+n?2c%kQAGFRLAHSANqz0z`ElvH}+?d~mbnS8GewfT;<%C)!Uo@jrN9{J5rm6Rd
z4F$i)PIS2Ow4+2>ft~V!>RW6rCHm-D@ceJ1+1I1x#AOZA`z&eMcWuM{$2O`TyRo#m
z?PoJ%=@AS&wBGYX>)89N^-g(JQ!j1WvEEivK9KMe%sBe61^}YVm50O<z0P|p;D;-K
zN<HeWX3cPu;lqd5xD`AysQB3*uIMDt%7|q|6h3s;T=Dp;t$$pPaOv@vHW2quqo!kO
zq+P!vvVs(9o7v&GsH=<FV;L!L6QKM`BUa}8B)sivg<pUK7b0|(cXEppy(U>(o1Y%N
z?Js<G=(B7u>UNY7M`XH-G$moQqEZ41tOIv2g>nr3QbP2wZ|B55Z_a&U(6{eL21?I_
zYe|elN@Pr~sb~diz4ck16ZAiT!>ym5^Lz63Akr9AHalVAgbJM^2%b8B#fSEZBi#dK
z)Jgn~vX4Q<*2g_QP#q-H$vO~EXGdz%2!GCM+0)GQL&M|$Zn0i#<V%8L8AsoqFMhm9
z*HxuJ3eJi$(07MTYQ)nT-bpO4_JiY{VlbQ!_L{^jb~)cdLX{fq#_gn`%gMN%DqFj@
ze$iFu)PyI)BXog*+MkgWng<yDaO1N3dl&*rS754Eu_z2-^fE69Wm7Tm8SzpWy!pJH
z6SrmtUF8AQedDylOS=J*Fv}1V-tn^P^2x6+;@o96YIW!tI?t+xd`m?2rTokXaM9Hq
zU)7r4Tjqg|tMvuT=pTP~k}du~p!w20Y~7Jq!=}SBh)^`Y@0>&QBPN5kL>GM3KXL|d
z`n*LjL~YhRL|fPg`ehFDrN)fDEm`4Inub^d86b?RO6OO73!I7}$kk^NHXL1%jGz_c
zF{mvYMfkXq(#UJ6Nt=T^^<VdApJlwl?{e6)q$nNM;O9ujFFHaz$+9m%LPvesHQb3P
zKlwD;8Rlg8I;8HL*{%(4H`^*F?H+yO%>e)u?O!L|{bOVAF)d9MuK#=mjh9-qJGawj
z?ML^)aeC|j<*s1BxQDk)X_0&_WripdafTxnA{Xq1Hc+A2>vPUO*Um!Fi#K=AUsPu#
zDznt=yIE(%w#+R_gh01*U1D@gUJ$ICt6pJ5RPL?AcB~5rJxpmbtXHjwU?i%Sh>nNZ
z)U{E$e-MdT`A3cHh8b)2UHk1CMP$$o`#Zf02xZoIyph^T@}$AJ()@;ah=tdr(u|o>
zjH}pP(%-+5F!ETbk(dd^NM4SQM_o1tuk=*wa$lgDqrZAR6;@bW#}(G}$dDST<|G^?
zZ3T;3MG`MM;_MH~^7MsEH3Eyd^|_2#P#ZqtNUFnh`v}k=$o98VYIGKaPs>rAI%P5L
zH1c-fc;}m;gjs9a7VcW&>eAzJW?YQM{+x#UfBDgPkY1BPkd1xGhdNv^RqXBI5xS+B
z`{_x#|61ifiWUVv>^fCb!&0Mfgju&sggln{MhIN}uGr;M&s|x!9(SLBe-VhNfkS)E
z?kw_kcRW?QV++~XnL5?H9I%9Q*JcEcRHP*gakkr_v#i?5G1qq#wIm;sp`Vk+xjf9Z
z5Dh*+os_7T&VG+uyV{gm?Goa<;*)3htkLE#{FPy+2zF4$R_j9aqaHpXP*wT-4(&q8
z4ejkZ#-knN1%i|Hk%*y$3d6xh)Zj2m50@|uiYQRB^Uoy>NqMR<iieT+k`u%M?d1Cl
z;#*RH*|aC@Re2W@IgH_Jz@x7po`2>y!`bRtzCwyo^uL)}yV@|Pck`yfrET<eFK;%g
zt(}TDXue|HVB<k*t}ZV|Pd5*C9QEhpOglYA)1C_dbr&vNaK3cA`eD|9=_LVC&1w%X
zRk7Yu^|Eo0vP{EwpX&zaZ9{we1(j|!7V(s8AI1d|7Q=>1IXJ}4k)o}{&14wagtM3%
z`o(180J)e}+~wYX%sW&n)pn5=N3o4a1FDPErXyPSEsNd|m9$vIf<9|8>3ZYC5s(HH
zqmv6>Pwrk5Rf5lM@B0h~CqpeLytKZ^__i`KM4RQu3d4er0pzr;=&AO3{a1EAOVLs1
zXu20=A*&(3ocN`0V*lEKn4rbc!%rgCUvuNLiQQ#q=?arFyi6`IGp4|jnQ>#Z)9D^R
zdT}Ih1Vf}b#$^tnJP8f*ed?~wU#vWWV1S5`N*tUb-PJ@{uYhnBrhk$oeuOL0G+n!J
zWm%KDcD}Y!_Exy(tA7j9>fo(9uERf8{Q3;@;OYNZ@t9{#J`U^fxa-0br|VbytVWlB
zf(~&Le~sH@<DS(b=geTA#V^K`beUZ)>~qxbBr=NJL$HfyT#CIEMP;l8#i=i(hfVvQ
z+iY9WPSldvF2^{8?1Umxdug-ezf5M0vM}nSYQM{?Z|%<7<bL0jHi`kCH}zMI4;$p3
zr8D%pmrnA{R>PN2Ys&8n&noLcP~+!x!g)9!w2q+ynY9%Vu63^VFl(eGqB^TS7`@}S
z+|;n#1<j=(D@EKxk;jf5ql>3|O?mAN(SUNk^s_;$mNge8GO!3ZfNHg5$zN-!#uZHH
zS=jfLP_s;eG(Vtn&?0`u&sF;i=Dk`lIN@=oh1C@EKU-yJ%)Hfb{~sIoj3ZSKNw#;L
zS!!lGTE{G-WU@qllR@B2T-smG`Quu*<j1j~p&~Fo@qln)Hihn$+8_;vGNp%HGd1Vp
zzyx)}*;3G?`Om1#9Oy`^=sEkYEZ)s9Cv2>m@5<%p#{{u|13izWe<iY#CX}&1jo9qu
z*w`Vj=E{*xT0=YRBwb>l_6$tOFBWU8tkURU^5W-o-UrG-A2?u>waq%uxuFSKmJiZk
zo2FK$vqii-ob5j4PIo1j9$9g@&9{BHYa_6Br8|8CRW3(^G-5aQR&+ZR*>rq^iG9`>
z$LZcU6W>gfRvu&k+iY6KiCR!Kc%0h3xmO!)fepa4*lVgEZ=MM`&0j>$aUo;~9C=As
z$WOfXh0rC~sAW7X1gD(RX$z#|1Q<sKcNZL>MEeQF^D%vPBM4*)G)nN|!&2djSCFB*
zyqo#pyScOey({W37UUBpJQ&_4_3E=&Pyw~Ru*J4%1!{jH1%`uQWK;s;IQH$eus&}J
z3><@3?|~T?>}Iy*h(T80>LVo*g3@cDe)aGY0&K}6Q=e-bt!lh9S98VqPhnfD7o&p^
zy3f`^`AlmZy_#*kYr%n<e*Ce`Jd)eUcX1rG7P&nNM|GIfz|K57(>!^jdSbvI$0LtU
zd!XF?V@=Bzx!F2(&AQJ&`?`T8?4-Pi$$7UkN>joq#I`A0qHulQ@F~O1M=vWoPM$%k
zA{uXm4@+scB25<Unz|`uL~^1Yytmm|0Q5C8(PL5F8?iP7lSpQcC#4XaO(;h9x3)`p
zouXSZ6G-i|mPpR_->Z6bT!rto-&U+x|HP_G40*P6{v^E<5N+P!g0PO8Ma*Ti-lzx0
z-sE=%$xYn*FO+<#(LJv^Tz$tsE1X23ji?x%s<v5<{0p7VR!LJVzfya+UE73E#;<BG
zzbhm(k9YuVooCZ228D!NJD{DK)3XZdPm5yzW+sJU&f7q&YHyRaW!ZPVH3Sqc^Nn;g
zdFi%nV%>6@`S{HRCn5t4C7)W94eFst15(fj6M5S9&~Nwj+|!Ky=~0_*1*^MVA^}O}
z8->|^^YH^zle`4dY~|ybT5$GeD}4G(66-ZgM|l?y))0B^v#pDm$GY_xY-yEtAK9X}
z-<B=qr%~fwRNa~brzDi^Rkfv81g9u#E2#VUCsmYcMkV~F{POe69``FKSjx;7MJ3KV
z?Sm3_jpYzQwwg~oXR9%INhjZxl^g}A>B{!2K48P;?>N!w4<9=nfFJ@8?N3IrsS-0l
z34eIu!f)L$(39Tkyd`@NdEV9cHU6~YtB<_i31|Bk&M)ttHHAIqT4g$^^;ll2s;Q{?
z7L1Q2@b575jcB0(w+&9TS)@D4RRZl{?H9M~JG;TeDdyu(@cmY~y{bs%L+f(43rV5#
z5b7J<GOFfZ05$I|`>{U^B$EUyN5|Jv1ggDFU_bn?BRhKx+i*Hl7_5Mo>(h{jLCA1s
z?Y<3X^!n^O@l={Vy*2%MUU}Q%RFlPEg|#pDU&Wv{-!@~H*C}ZlUizz?@5Xb^H6BDl
z?0tK^qRIV0(U_fs&{UvFzc|H!8~~Rn{{=E8c3?y6n)2d$<vEM-R24Y<1s|Ryav|o)
z9;S9d2C_Bb<07aV-n*pq;vP%cXBoAf=%?AjuO@D)?oWAVRP<FM3tpbw%^M3#N&X(J
zTqhC}E|DI0m;#{o>a!3EZ*34Z1{5_#cv-F@b8v2VCj8l13|Og1D2_cbW^9?TGB7RL
ztMW+eyUfhXfd~f9@$*+b*{SKrHvTdD+>^{P9Ww_R4?1o<Xa)tT&fgtMa#NImcfl?K
zH%FcMEOw71q9A0>;P+i@UFue<<VBPTV&`91<i9T5mq`O-svq;XJ(dmFQ<=M_XEM`N
zM!yl7Ws}qIkAs~n+}FM@-_fJCw{!oJ3A@`12d7SkgQg{X5b>0WJ`S^p4B1Y^6+@I|
z1mUt4jpxKQwm4(*4Ntbiq>h?bwEsp`4~oJTS{Fe}aF-JaN=OcIq>>y`Dx|pE@#?4&
zgaKQfdT!8LihqjR2J;9$*?$@+H^M3vz6~KFZi9m4JZd`<Q-T6)WaQE<<x%Tf#M|WB
zov*eWgHgh(b-0>q*@F>cjP-d&!Uhv%1tF3Mud<daR;=K%RZN%v!u)M=&;Irk9RZ-S
zwh|5nzaBA<{SL%mLnrH5;xP6v#vqpOr~DM6_yypus{fy4tvzQ-O=exn&Goa1u!Wjf
zQ;yZIPxeL>ONspZnpIRRZq?cZ$32=H@Ho>4^J06_IBt1;*nUcegeYdOp}S4$(g5lQ
zK*tDMfVR20t68_Lj(t?Sw*bVvBw+BPhTYyp#dmBr<Zr}0b}I$b+aCs^O<{YO;Kw97
zaLx|GN}Dt-eL75G?a_!k+9MtWz@*~(Wf~F4T^u|PV<zoklcm`376hYuRRp4-I;Yp(
zJ$podK(Q<`-G)#a<?A*@ohMjOO_D$smFOWc<~Tw2zWEs$0==o#-C_CphqI;R7}gU-
z5ap)J&iV8_v{^XP(&s5C*-Q{D+rv^?&hUx{h{Ky!fXwo}YBo@4og=G%L{cZCTXSzx
zt3T^z5kIvUq3@&7oeEmW^J*~5O`kLk3)dW~`uI4Cm~?SgBb?vB#3G{Es5LMdd&o&O
z?V0$BSR*hzpU(D<-xK?LxgvG2a;@9XnwANc^LWJQrUn(u64*`m?%v%z^e&Z<z`BXU
zyP;0F;$Vp{#r*tz&dG@Hw}0o;+t4R_`DK9x=JjslSM3iSQOQ*1CWn;lJxJ5KK0DY8
zu9;_#>tpu9HD<om*k93bzkQ6x%!W+{*|hv_W_dEQf9}b~oD@UvYbe!qbc&&&<ksPZ
zuJ0X^!<09({tZ(%agixY;Oe;08o)Ru)H`98VXd(O**wovjC5vV78SWSR2{aCOu{4s
zqh<IffnZbDo3wurIFKI?Mo21b@>O^p2o;uqFCzL;yDx6Z^8vfgZ)HpmNziLxlgKj@
z=Fp`(f3JgZEl@<Q056!ud~ZssIBVYT74{D8F`}?08X+Y)IOVDX8XYZx!N(lFTY_^F
zB?qty?5fs^-+sd~R<L2|MJoNfslVMEw`sS$7!CeQ<eIPZ7cP{?VdMFeDepCyQI2Pz
zrh^!)AsHQZ?z{5kmE5q{VzX9xLq-L==vv;>)&8ZWa_WfKUXxyC#^|0}aN1|b1x*`|
zmi8A05B|3P+>?VZT>Jde{=njqkqZ(R=<eDvZ*j)yIFDsnvz-iQg;)%Dl$g@=!|hJ1
zPVepGeeIjK>ADLmN~)J<<X!kyxFY#jY-};-bf1~O9}-Ch>Q@vy-5ZEP<nKCzJUu<T
z((l{8^<)Q!F8(tDK6?|x2IbZr%*zi)#!TDM#13S4Yl^L2qw?}{^sF^13^Jplq7oAm
zx4&MKjBXYwMe6g&goM?1+-ygWzD}4qFe*A^&X+G=P>HcsK4r@~kUn+s%9%WC)*3(%
zLCmq8L#dar_9tRvw*&_hmcFSo<ruQGBH<Dd-$k{hb?dE)NWDMJcP@@lGxazbd+O97
zEYc=NyV<jAS^m^qoz!n#eh*nb#Ye?TfA_I0=|6aFvw3}ALNCnK=%DwXJ{cMqgeY5O
zXJ@Z>vo|Zzb}&$1Lz(DJ(%sVh^Cmy2AG{jD!%#2pRW~E?PP=46Th%?29UVtjU07B-
zHZeHG_1yH6k&(a74L4M~S(Id$+V54V<<!NC7r#lFXK~Ymzr1v5a7O4)9;cL~p&<A=
zeu+fBWW_S1;Lz7sR=zt7)q`0K2KjhbHJh#}!{sUa;%|deUhW>YRYb444)ai~I|@N$
ze$|_&>kLppr=n1nyL^>;ErCxjFO8jTY6f{5s1HZ<9es)ddwtpwGBj%fY&CxGIxMW~
z{)q1W`9_wFwr))dat!}L<4^OR+wVL(U=^C`M04C`v+JULbkl8ureTiX(}#nGg&6&j
z)u7W5qz@|Cj7=s+qZuNJ1tnScf>^YJK2nQUB#cE$YBs4y?a=fFXo>vB!R@hT+aCB6
zAK*?zMC9%0dNelH75nCHmq2DFbtRV(o|jp*?bh>n8WHCusIGNJh{cp{2LRX!%l<0f
zu2C0qJ1s3O%hY1pv}t~Rb=Grs@7q@m7k2RA!D;>{e@Q`d<H?gJ#l^*HzTHx+xebx+
z#EtdP+RXpM+7Lw%DafXEcAHSzW~@=vZ<{M@LUFgo+Yb?NxVpL$WSSv#r246rYPe-|
zmTH}+3~f4n`V(ilbCSWA<yWs>#T|D_>IKk?O3TC}ANrtXX64}EFl*K<SgEz95R+JF
z8MSD#LeXf9+bHjtt{Tee_S0KWJiXScBGX!XK&7{fcH0EXX8R8u5QSkG<dLzAxLZne
z!ti;v{LAoe%{N!NsdQ511W8REZq}2hPvhg`y*xb^FdeIX`(1nYnt}RoOcL{kw^vuU
zip~P>6YKFkwCX+_SObHTF)=YGPYNHS-AQfwZ?N;g085uF39*XC6_<ScIPp_>1-8;p
z=}-P9^E$J#v%f$fc0@gf@B{X!Zr{Fw-L-4}*eVs1B>BkG?CgtiaULEX6_u5+m1VRd
zHLyA4A#LZ+uALTa(F3_z=B;MAco1-qN`Eq&%=zWcJ^lRr#*V#<-Vq$?bN4^zr{of_
z3&^@{+qSiRo3_$)adS&y+=<ouHnJW}IotGub5odQc-_=fr%t7(rvnZlQ{`fBZ|~;j
z#w8)#y1SQgcKhh-Tb+BcUG{mOKHspq0_O_!1v85m^y<|s&e5Ip^h4MSAYaB7c#*fZ
zp0RZGrt9%{ev}i|tIvp^xOYlt0-g`j5Se|%K3B1G2y(tuRB-3Lq2dr-1qKEJt9-?y
z=<DmZ)#?{&(E~%X#MQM1W_kS5mrtm%;$TuNthhn;E8mFdFd#yh^>0s~JzGVa$Mso4
zITqUZEwj=pdoZvkPb%>kqLNHU2gv4d=qb#v7U_4I`$e;Y)S0(xT{@@dciCHt`4K_A
z6?R`UgBOXt3;B}iM_Gk>r4JiWW*p+5_^|6woj9?WssW_7L-<+-L!O6o*0@CH&ys+7
z8%$fZzIJCra?Jhtd#X-73F&ftMfMM7wx>m5rK9I$y&`Zo&EGnR{H^EYH`zS%#)DEd
zot@Dz?M9TL{f@`gi+y^HJ#^xaplwa`UYeVicztIk^qXL&%Bbv!!&P=4-&o4Fe0hr#
z;g6-c@_H_;Fi8l0?J)2(XN|LaeHPLq{YJQ`bn4)(nEv~I)2S(qb_jjQ`Vj;(Ht+CN
z<85pnirn}wrg|}F&bWa}tdD1!UuR28%U`nxo)%>5&Yu1I_uJ|>uv^<zBXQTq$eJ#{
zYYHc`99GcCw|eua$(gMvoG|^%k^Pr0UAn$-4gGp7*pKJJ%T7c`JCj$gkC%z7;;c_}
z?bbmNnLjSQYiw?VvC?X#ZDeZtU1aZ)-f>To=*WdQ1#8d1Z+QRyz0@_@y#FZ$a{`h)
zq!e17<8Ytrvjeh@25szPye7u4sdnF<A82Y$iwEL1QV$Mvu-u}d*+6DBVxHD)N)1`u
zvi|d{eS;&Krh_bwKR3r7ERr_cv=e-#;fJ;=b7wy?Z(w!nUqj45b;2qs7u~9JV{`tx
z*L=)u<A8J5=DaDpGp$j;xv~FKRlxc`e;^FzwF7P#ssv<7p%C7W(Fc8%Za=~IFe#~I
zZ-AsY7CNcq?I?c#UiC`jxs9oZ+3)&P|2ccY^L(ogTp>}o09@kn%a@`*fA$3YfR+E6
z4)(t*W;k>RPQ7uuL4aD76qpF4<O(WtcR3IZ3XAkgi@XuiPQ<58{fue(m;W1YPvPHq
X!p)Rn8&4mXYmc>@WN~`T+;#s2i8N+R

diff --git a/docs/dev/articles/getting_started.html b/docs/dev/articles/getting_started.html
index d930a6dd2..fc59b4bf0 100644
--- a/docs/dev/articles/getting_started.html
+++ b/docs/dev/articles/getting_started.html
@@ -29,7 +29,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -261,12 +261,12 @@ <h3 id="training-and-evaluation">Training and Evaluation<a class="anchor" aria-l
 <p>Evaluate the model’s performance on the test data:</p>
 <div class="sourceCode" id="cb14"><pre class="downlit sourceCode r">
 <code class="sourceCode R"><span><span class="va">model</span> <span class="op">|&gt;</span> <span class="fu"><a href="https://rdrr.io/pkg/tensorflow/man/evaluate.html" class="external-link">evaluate</a></span><span class="op">(</span><span class="va">x_test</span>, <span class="va">y_test</span><span class="op">)</span></span></code></pre></div>
-<pre><code><span><span class="co">## 313/313 - 1s - 3ms/step - accuracy: 0.9815 - loss: 0.0803</span></span></code></pre>
+<pre><code><span><span class="co">## 313/313 - 1s - 2ms/step - accuracy: 0.9810 - loss: 0.0842</span></span></code></pre>
 <pre><code><span><span class="co">## $accuracy</span></span>
-<span><span class="co">## [1] 0.9815</span></span>
+<span><span class="co">## [1] 0.981</span></span>
 <span><span class="co">##</span></span>
 <span><span class="co">## $loss</span></span>
-<span><span class="co">## [1] 0.08029716</span></span></code></pre>
+<span><span class="co">## [1] 0.08419506</span></span></code></pre>
 <p>Generate predictions on new data:</p>
 <div class="sourceCode" id="cb17"><pre class="downlit sourceCode r">
 <code class="sourceCode R"><span><span class="va">probs</span> <span class="op">&lt;-</span> <span class="va">model</span> <span class="op">|&gt;</span> <span class="fu"><a href="https://rdrr.io/r/stats/predict.html" class="external-link">predict</a></span><span class="op">(</span><span class="va">x_test</span><span class="op">)</span></span></code></pre></div>
@@ -276,7 +276,7 @@ <h3 id="training-and-evaluation">Training and Evaluation<a class="anchor" aria-l
 <pre><code><span><span class="co">##   [1] 7 2 1 0 4 1 4 9 5 9 0 6 9 0 1 5 9 7 3 4 9 6 6 5 4 0 7 4 0 1 3 1 3 4 7</span></span>
 <span><span class="co">##  [36] 2 7 1 2 1 1 7 4 2 3 5 1 2 4 4 6 3 5 5 6 0 4 1 9 5 7 8 9 3 7 4 6 4 3 0</span></span>
 <span><span class="co">##  [71] 7 0 2 9 1 7 3 2 9 7 7 6 2 7 8 4 7 3 6 1 3 6 9 3 1 4 1 7 6 9</span></span>
-<span><span class="co">##  [ reached getOption("max.print") -- omitted 9900 entries ]</span></span></code></pre>
+<span><span class="co">##  [ reached 'max' / getOption("max.print") -- omitted 9900 entries ]</span></span></code></pre>
 <p>Keras provides a vocabulary for building deep learning models that is
 simple, elegant, and intuitive. Building a question answering system, an
 image classification model, a neural Turing machine, or any other model
@@ -334,7 +334,7 @@ <h2 id="why-this-name-keras">Why this name, Keras?<a class="anchor" aria-label="
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer>
diff --git a/docs/dev/articles/getting_started/unnamed-chunk-12-1.png b/docs/dev/articles/getting_started/unnamed-chunk-12-1.png
index ad4a14d9fd6bf348607fbff4fc169aae5c1473a5..9a387be7004aac15400ca8eb4bdcee9d7ed524f1 100644
GIT binary patch
literal 32138
zcmdqIWmFtd+a=maa19z<LU2ifyA#~qA-D#22<{}o0*w>gA-G#`cWE@ZyEnJUJKxN#
zyXIT#{+M64p{o}~*QrxS9@%?85lRYD=qN-eAP@*$`jfZ{2n2Kc@<xIM&d|$P(g7dH
zj-RxhK_JxEFK?Jhh~Fa!L;;c(7g6&_JA}CVsH#6Bo}M&PbMZlhvvKK^bgr3Hg*#wS
z`X#YMun{8<I{Px=u$M6dHzcu?LQ%00>4@J@^|K@hF#X2mt#NiZ7HAg=%&U7f`QzL6
z^@E?l*u`M{a$1^?v=9&QN>-AAp$z`!^*;L;B#M;1)|DlNM48zAza2M4Az#mVNMbjJ
zn1)`9wiq}Vkp~(R%2;)@M{Rqq99$u^XeXJOX=B)e(Yo2tf<k(`iW@4?x_4cgI^3v9
z)F-ms8})jVU3p3SzR*-4Bb8||M&{(ueYMl5nByaf4QTH{M-XFU=wOp$OHF)BzlJxy
z=Q7h$8A##Lx?+SoyXS)NXh%;Nf(I{x6^z0$RLz7Sur`e@ic~~~AOIXm{+pv7p1wf@
zmMD#~Oug-bN&V8)<Rlt`SYb1cIomF_sPWo@MW#HzEBlHi$;Z2t#sf}CNy*SAsde}u
zvh00ET3U<lQ1r92Gx=2ZA{Xk1o0HXsFQzIgD(!wxg!yR;Yc&>Qqp2Jvm6eW{yA!S6
zH!Qv%^CG9tR-U2BHLlL$GE-rFE93;&B5#rT$*3a4e+a*(0U!BQ<k}>&gDNdU)JrsC
zh&YNTNy|fl-W2D5`0&A?(P6qUf8b)wf)>5my(<`bi!3QV-n3@n$B!RKlx7Z)Hs6QG
zfieDtwyO-Gx|Vi}n;J!a1##?U`!2S$5t+B&J@^rmSW@2LSQWmhvY0{Px%eeL9`N>c
zTv#~Sx3?QtupNvD`1b91AfxUIGja*IagOY|lWMU9b|U$#NG17i#hx4AmzS3}hT531
zY&ms}$f-W*Qiw}Q5#Zw^Y$o5++&sTb-Oz;(8wznWGgeMcPGn?cv7)j2$%<XA4xjrm
z6H?$wYJkJn(=Jq2xtwE3Wo*#+b2B-?$s;N#d<i`#ES&1Q_^F2mv!O<%w`R1$&TnzF
zrV?^mX^Mzs2iN8Zy2T5p82F#(D{u@uKc?`~zX|h@x$q92C2R^~6-wSZk7c0l?a`v3
z*4d0)?2NIp4AoXKMaA<6h{ORuhObZ`mYv5WNLR>9<|=-qgrE`-#Yo<|k4NR)0{zjO
zua2;rblIOSx;mI+T|&Z^IBkgNgCpajj%Y*e*@T(cB$l^0T+%eemnTBc`*xC;6D_%M
zrq(o;{2pxvNewGJE015%Gp2OH_Wo*)EW*_z>FMsa=#TG7@lqxBge7i_Fb7f2gp$F<
z#iFZJd)l+lj@#0hnZL+goZx-oaviIF(_g1X;O?r&@XC#ByS|S52-(FnLPD#Jrc1Q$
zs~Dg!VX>tKtzNt1**%P3-+oY_<tAPG;-KAlPT4zS|H&Wx*8c9Wds?^YYo8duy<q~*
z>efNbx=tmi6aj?v)iNXO2mM%e7Iq6svMW?izvd0-gPoiA$nCG`{HzJ37-3Iz%UZSE
zs;EigkLGMC7SDBy<Y1l>{XT^@1evxV>jO>_L7SwdqvNG|hq$j|gZ7!T?)~3j<1Bho
zInaqYeM~2RmAE!2r1J(+XqC-I;^D4`lSy52=diLI4#d-1jHGTka!baMK}s|$Wa>R3
zU9an-!k^KqBn*YUSrC8kez&cQE^DtZdxHq#RGzkUg;gqiL`_`ee@*^o06a`#Swx|y
zU@ekGB(l*UTA)gRT@)1#9}{Zrszx0kDW*N^zC)YgLB(-kiyq{6mQ6jUb+F;cEyL`I
z_J?t{-rFoB5a^2_ca9ViNH}Th8H4!Tb9H6n>^r8SZ=>7Mq8UMsk_-WhUW4D${olWT
zrQ_eFjF_)piT~)>`pi$}$QLnmRIPpF2F@@ej0<aktbM_$N98AzywoFeH^jkQll26#
zS+Qp&csQ668VVD5^#mT4!q?FEt@96BH(l*H^Ypwb%fS{d8KJKp^}HV7YIcXA&>b~P
zFHC2*A8v3WFZWYRn6YkM<m?43#>#IiRb%7zz&J0kTz<*b8-hB`FNBwcO>|65%k>V1
zfO!!W6<yG+sjf~Hmzw1OgN(FfHuam`^4R5IlcUZWI*+LAxxO@F9Ny+0Wy9fsxdxq=
zE>`^uG}3u{w&3fd9v_@TGnX=>s;R!Dv3&S}qJsr5G2SHvXNMIWC7}VPo0V61rz2qb
zsu&x-MuMv6CM6VczRsj~vDSu|%l`4S8#BtRPA@LHtcv!*q;1I#J2*2olj!~Wnam6t
z5~YCmb?LHi)YjJ4NXo}P>skzA&QjNdS+04aV)(SC&LZMOImqYFpXZORXUlY*^?16I
zP5WY7*_>{({)X-)pL@l=@s^4^?m40&tIwj#sSW?qjF;^E{O8KfX4dGdmc&Hi)CW8t
zJ0*i$OZJ`WY8+-skP|yAIT1(E+}l^bbr>v*B(rp2;E%v3t<PAr4=c)%b4|CoIVJWO
zWc=LD+j8|w0!AGH*wm7GN0o&(vAc|Ae;M88D~*W}F+zW8-JU*?)|kJep;0fFPirb-
zCz+t;@VM(GU`D)ot20dO`ey)#9H+&=k(V$zzCo{N>#e<JcUqcZ*s8N9_FBlt)2oz?
z2YlA41ErGulR)ip+_4DNSh%v-eHXP7z7*fDdc#(@c;63Pe0Bd4-pq>^PpOv0TGy(T
zz$LZ1>`j8Z5&T!`xj-Zh37Yc^RBf_1D*CRL!TeHw`ZdxCW<>@;B?7sZmzRJ^#USN9
z<u`0hWp`Ndfj-_PHak>J6)29S@l2W5yf_TSOu;QO>LDjLH#a%CXpItP?}Oeb0xs*T
z@f<u*2L#Ul`RPUmQJ|O|%?sDW)6?@}d!%ORh*~OEPFkA&<|kk`D-2qj{K``052oKq
zg!Guxxbc}XQep5_ErfRqGKx|_SSoG;INB5JI=l3=caMIU*@?r?>=$O4p-HA};+1Jk
zSQ_Y6R<6~xN#M(Nav6^KU4HkXFUp#kc%6x9^49A9a<8E6d1AcyD1lzJrPe0dD`9?q
zep52a?Z|DdAviZ|_UYuQTEO#mJ&aZ%ZC-G5K3mpv)6hG2dFNpxigk2=F`#gl?4-5c
zs55YFBtTVWEK`UTNwg6apFVGl|6+KyFC)0H-*Cm#M%Bk>DqmifG#(6o=%<dejSUaH
z*lACBJS?kU(b*5^@Bh^B4c!2ln+${8n(g$MLV&;po$!(_0axPX=7y1Rv9o(n@8{k6
z<R>LAmZW*x9(!OcY#6CHOO76So{Hjc2+@7+uQU~x#3xK+!wdDp_Ak{c3V2ujT~OPe
z>|-Sj$7?h@d(-kyS|64i+V5hy2+>%=;6NWfeiVX5|5XPM+8s9=n%Imunv&rdy?Fd>
zHD$E;yF{R=Z*zgt)A$dQ(a+&mh9umMSyFLid9e|3apdfl<6}b4%MG$O`J{aKgC9WL
zxV;Z!B-Vb)6C};3Wd5%~oKICp@iE|~n_Uc0i<BC!M3Pyus6<v@EGeJMuF^;<2N2{C
zhE@$l-gr$_D#-M{zED;uaPE!7EAW6hoShKxPST04FeV(ZhIq*;7OJCy)>a+_xqa)L
z%O+3k+RLDJ7q?;A3i?VdQ%U{~4s$E+zQKR<o=fL^?h-Zg{ofOOzH8eaq2m2@<lYD)
zOd+nh0qN~w>pki9>azUl<oDI}{I7Ia_xFSZwc|sQCWN%yKA!uv(TNkltqdhG4-5>{
zL)@0Os6VHCnU5G5oVN=}@wX_|TgkO2g`IvU!z(g0Ve=Ffj<S_wTDTw=^R9RI!H}AR
za22|1>g8CWjWzqRt1AR`C<F68f;QvU>((+n90c<4JI~AXztmT(U8D-@E>mizb4xvC
z<9}kZgx7*Xbm0)Et;ai1*Yc)W7(qswxU6{}u$pr~0jkR4GB8=6lIUNB1juXfla0kr
zS!5}Yf<R%^sXIRHuGI{YCnv*sbQuhH1e#_Nk2T{{2QKpvd#(EXo$Me~e5Ub(E9f%m
z&aN-p?)%*!yvv>gNUQxw`Ir!C!t(BqA+zX;^w);*T4g=P?V}E!T1iCFkprbaho?V7
zKMzdWl`W+^Oz=)I4=M^(nVzgrX`d<WtKgQDh6|HMIz65Yn+ZX8KhL{aSw$2T<>YYF
zyE0Cf(P2oO!KO93oRv<YD^E~(!iVgmAs9L$UN<7bOCcScdEla;t$$uHVJFf?0Qu;S
zD9Ky}>kSVM)`$l<7<ohWhC<O}(7F@n!va*iPS-jEQ)>fu)=z4Vh&d(?uHJ>RId1(7
zLd47{SH|(dIY7j?JXy^9%=H=6oHK7*2RJW=$k)Wgx<g#QpX*`4N5UHT;^Kir+N!xl
z%DJH@Uf;2H+S}Kzl=d~sahJ>Jjt6J_M=Ho?%Xxfm9sOf_vu-*Sg$vK=@}lsW2gk<z
zPdgFinY#6cjh$JmB|_v@Xx<6M&6h&xk5;PE)3@Z9EG!nfIm7xLLngkwHa_$Uk}YuG
zM=yvhm4ZU~;*4u2s?d4e5_nc*2$2g_vlS%Hl)N^eVd?Ks+2EN#vTZgAUvgU7D#xG6
zK-`$1<nzHseG5=J3`m&T1H`}8isKwi_yz$qj3$OeNEk{ROk-rC1!7}E%ca|UxXBzC
zDl3_Bvs&Whq=E&$LkVJ{d>0A>n%G)w_5}}Z?a#M~If;@Q4&f(VZIvW_Gm}joecd{q
zN31v8_|%+7?suDA@ddiN)C>4RTs*v5yQR9@q4^&r_U5H}*KTs?mUA#9*DI*oEc#8q
zr=14|2dx`T-ta$=fP@uBD+6T#pULbxY)DXZoW}VI0V5RdfwphXqx^XM90hf}<&jk(
zGM&?!%VA}?(zv@koe$SgJ3D*8A9|4)@^t)lo)8qES#W4rt#!X6uy#J3)ct;Kq~2)R
z0|d&zyuFpXauhd|m%4oO<_$2P*B<=TlF<zxic|voHa*uv2-^6)9*-M-Y8v|1dK@*%
zYmiji*6WnbcHN<FJ74XmxO!1{hkN<LR8%?XVHePEcoO&N@~jMe&|sk>Mz<!+p^g4-
z#x4xM0RHZYv9SpQV$t945w_m^gAj_1WQjcSF*+2Bdg#_E?bRu|pYH;ZDZ30yL6Y%r
zKieS=@9V?8QiP_PU|E6(mY>az>s?Nm&DT;i%D9G}TZxT(VGn)1a_PLa%Mwpp3~w&o
znxD@4$e;7Zy-;Nh{O-0i%5)NWwSN4%^>gn8Kf{Rx#0lM=wJE3bx-u+npKbQJA1`ay
z+e7RjO?M0ELh8GCd={f=1f=%6&U}9jnoL(L41KTE7fhii&rpYQ2(t3TY2S7Hs=w}|
zO|4Bgb}J@=uNH!@pI=+wCjCwBWvwSoQDG%rMd4Z=k8^NF{fCxN5`iB{lfgvt7+oHP
zCa<VqxsjbajaiMf>J=)K7?g{VbZEQh39ltHLSiXljfc^?Klb#q5yAFQAM<~K-RA;<
z8ooiI%u);TQwh6p{&ZlGb|axxu?=SkJ@Z!sQQ=4`^>Je#ljSg)?s+qB#hh-5gV4`&
z2Ev%rSz>TOOG0YuzC|)af>iuFc^zJtx8jm8k2X2&2^1p{iTYBk(vuwchW1N!M?si^
zT85v!ua5oV1g__2Fq)AMb}M9m{7N?P+KUo<<C{zUGS~>{gNCj^8bS#knJrc?ahm&*
zgc`Jh7$o=@2yMMt@%A0!WFrhEx+3>Eg&Be}`UoVLM^3Jiohcb)$W#AqN=w!|){`Ej
z{1vc%c!*Bp`aV`AN^`Oa!qlfikrDXl5!dV;#{N;vq*g~D0H&k=gKYb)iFAy|C+Szs
zPj#5)ECwwNYJACv8L9<AE1;8>#=AU%@}!A5Xs4Q_<e%n~wgVU56QCQ-$4x`ev?|&B
z_v>NgWI8ou$E4_p*AV~x+vqnR2L%rEU-=5Fg=F2}K#htg9b?<9u_ex~<_v|rE-mHc
zhJnyzV`D>x^jl`;^Zo46>Anr<6f!4!Z@-8;Rdj(}zw@)td6_yLtJgWlWy>;*7!**M
zb*I^8Q<xJkE|90AJv)EIan(eZG@~j((IYA>R+KUMF${AHr)#=2@9&lrIS;#SKNj<l
z&TpMo4U6%;-{z=W4T8S+hb8GCbLZj3RgJ^NEnQ<3r<swBT&XN|J`gDPi@Y*VXWzwR
zsJq2*8jmV&?>cqpgV1YM8X6k2<K@Q0=89(o43bau`t_GXKo%30uZVfwO^Hj#ol~Jh
zeUB-4Ym}NSnz^)oStKI#d<sAtcjc^P<mm&$w$FPcggiu`4r2?nKh=r>x$G-Pr3Y?r
zdDE-NG?~*`bSmO%#l?S)V4FwvTCzJXue5p#-LAm_$c*D!he!}Y;?Y2EoZw>K>16E1
zLQ&O(?#G%Wy{+W?gW0m7kr7gU563eS(Sa<(8uO7tR)Wq->`IH9xA-r+axn2#@q*vw
zRu{Ga5RIctU8>7^dGYeI!GN;-g0^;QeBV37C)!EOj(Fbjr60Gl+-xNvf`XJ`Dl2V2
zicHPh%L{mapy4X`-mS58^r$;tUa0P1HnPZP=xO#4OKL{mzvtq`PZnxio+PE;<U+3l
z9a;AC<;)o2H@GQo*Jj<p&7xU$Te4%kydME`(Q?Vs-yFeHy5Ja%15eHCsso*hW3z=P
znP1J-tVphAJ$7uGgpf7%AhI%3M%Y`BmbePb#K+=0Z(g=`g&y_nfj|sMDnoUOtXh|z
ze%P0BA?UI{P%nlABLUNSGSHWd*`(X&9OqwaWH<zbWOmU+9M)}g7?6f4Am5zI^r@a4
zE3}y_2U5ARO9-|o7e){?e{f(Rh0nblVoxsM<+5NWww&+|jcD>Hj`YEqWnQOlQM^CD
z1$I@wia$Aj%8h+@>A+>gHPB&%102un)BLUVCmd+n2zSZtn@(+RphnG?1FCKD<EG35
z2`xeUr8ZlxY9NXHBYxnC0kSVkcKqA_y;?*O$#d;1BALh9*8l<-N|$)`2_CQHA$6aW
zP3}-*ab>haF-H>GAt$3usG`u*--pKMpYKfv%e~{x8vUQ4#9`5(nm6X{?zU%|uKUw4
z0YM6Li=qbzC3ejt$1G>*BFyNk-M0JIoUWWisVVY5)HZZAq~8+H{UFs9Q5MFs99yz%
zW?y3wGO@Qm222s5FVwaUa<{mV2RKOP8Jvo_rFPGKL|-q-gIdfAgXcfzi7?AbY=j}N
zKxG<0Yy-Q(G<NaKU`qkNhc{GRnnTw~XUo=BiXv$YV2|2$IU&r=<dPg}5<!9_?a}&G
zCcVLib6wxEPv^=b$aVnSMV8d@{O>^QNAs>8X%Gm^rSPfv#CpXk9rB%7|Bj65hiUV^
zD<nF!9Kn}*eNZt*#;>;5pnnqjJB@xA5#$gQdMx-<xS&!eq)~E3B*ef2c&t8hf7|*c
zsTiV}M&}(Fp@ZtK8@0f%frfuz{od&Iu1(8#JYFw3upLvf5v?D?kP8TTTBpES&z2&G
zNMzA<u#F!j@t~o;k^pt!Od3=i7_+-@?dtBsU#AL9{JlBE<t(GAmUHz|tXRrR*w*Ab
zUlbyhlauoS(z~GVQb^gw2tk5I`?F=JeX+t|w{OYbnzk+jU<Hq!%(CmuGS2gjp=2T9
z_qOsG2WIS-oNt<mFW+6?6Q910le)c<f@&dv4r_Q;7yN&_$Mj@tVtNH*op)y4%pF6T
zSN)%TwUfdC)Cb6>s}bc=+4qJ&j+NeCA1pC#C+Ew<1(=e3G5dZ*Tf*FZL@!ucq|wm*
zxn;`MHhapz-#yX)Tbk|DBx_2SkI~Nx3=*Pw?!6NH7+N28J`RrB!{jC;O10Mrt%l3?
zH>A%uRC4~L@JFfjtDh7U>?eOo7nw#&Zl>`#CrL<=w|H1-`>j{wSYSNGpP&izwU)Q{
zPS4rOtW~I9Z1%-DthTiRTg%ESS<$DHTtLq6>3cX82ccr!o`)l3`yn+C)R8WIH90l)
z9>99%LPbuJhlk}Xr;*SkK>-Tsm?YdvoV#V!)sxn`87Zx0{rlidV-PRVr8_vrJw=|Q
zBKbrR$XT8r7YhE=S~hv@X?i1|`CYDug_MEW`RZZvoQ~k<0<yoiY<J`>+ctMIStgP4
zK$o8r4U~hhMDvWsA86jc$Brcr>w+EMTfZ5xTWNACR!o_c$JpU|B~em(y%CRCJ86Ze
zJq*X3ud}t)%rtbU&Pkx#<g(XSze;2=eq*EMe>+>&q=o8!4-XozBTa)mfQGC=2D*<|
z{^ld+9{i>T-ZjSEp&}l@kg5Z77bTgy@VGLpU1Qz|Ies8dS&N!;&iVbT(o*{C)uYfs
z2#$xntxp-0SH86*ZR#BC&~^v)cR*%ZYA}4JSItil==l6$i~HycLhBIYn59Y_pD(P|
zEaU(~en)pLv)39(-7@{|tsFDxL_dYF?LH-@rrzJ}WWoi=?Nw+zJU({GyM;Mmw|zCf
z2gguTF{h*5|15E^Hm}w3PvJ1VI`u~CU~GS`@_o3jMx7`3e*~g8|CW=KVMLu@?*KH4
zs{sKc-i8)_>n|x%_C8Dy^aFUr4M0K>(1{WtZyK~Rm{<o{^|X`{?#47a>@F84zGL5L
z$w_YCA5dSMpZ}m@YD?s#e?OtEt!)+u<k8RTF?&twjBgf7#Kjv!mrHIHoWEj;CWeS$
zf?n*PeSAh>-O-xkeODOm<<V08G)&lXhNAODGQk3n5_a^niBS)m=2EHD20q9A`1#YL
zpQQHwZ;EBQ+j?g=d7D7Z;rmp5e<bi6unI&4<~p*~qS{v^ksx$vUZ8+6!3VSqZ!+sy
z*ovE>`g{V!R#D?J2vpmk6Tf-NUEWmWBMXCxM{sXuCU;^N+we}Nf1Bk5khZ=zy}QU0
z;#FdJY%{2F<^y?v>s%my1n|4IN=Vt(7(rwJv}tzg>w7<j7X4|jK4dn|=|a&%93Y8W
z<wB)$s2$!DJw7L@YU4+lZS+KJEoad!a<rc!-oOSFm}*wpD4uTW%^*?!#eC@_5D>Ha
z9=DH|@f|KX-?ZNNInqDZv;>k=5P)3J-jEr4a`RSaJ3g_B>$A)O(Ev3ywfsBPS7ZiB
zwe6*AabyUh{1ES`BR=)}2#R1OZNf_sh*{@1ceTgFcI$&1CNb}fEUGgLj4<v{Y;2?l
zH!syOgAI)=kMI4Zf9Ud5t-@mlYVx19GYtTTdTb>4{uUBhu~1X+)mX27^~i4aJTwDY
z6e&@eInV&N7a&P^(nkT@R-0+-a2I@#D0Izmv3am4U;z3wej3*+4k0$4eW*&xpQp|A
zIiIkr=ZuUz?LW^N6S|?@jd;74#BBU=OeM+rhWya(=izfVz;b0ylV2uGA_CFt-1N2)
zyCBc!-H4NR#*{11>c{oGN6))gEFrMMa539O4ZD>1IEH8RPW4N6(bF)Xlhq}-@8Xaf
z*?DJw)mV3QB1+oXid|}|;BP`en-*gkPwEx{&4@rS4*_#I9Iv(ufp9&xQWgk=#uy#x
zP(W{OLJCZBj;nr_w~{_`L=$rzXCrS)<-e;C0x$S?<<1HI{Pe?qVeaQSQZ`ulNyLGj
zD0}TZMUzr!%+}siGeOY_Rd8UaegFl;j?W&yX>t#iAC4(=wt_TWH8c7T_V;_V*h3^^
zi1ar<QAIYL;8$ty(2X@CtTNKRT^Wa?4j$(y652I2F)=ut+`FP@4Nwr`XW9N#gC9pC
zkVrcO`u5fmibE{WZ_?Ho&WcY?zS`n>QDXOXO{~QLa$d1e>YOmw5E`9uU^GY_94Pw)
zY$Zj3tLZ1xK3|Jk-wpisMn`0j^~GaUtqlg}o1c+sALiT{)XFq;@5=kDulr6GhCe23
zzTW~OAE{cT`xv*K>E*y~drXjc)#22aiOn1;Cf;<q$jDU@A%~R~*Gxa#wY&LafmM%9
zbd`uAADP+`O()ton4u;_w9W)6BnSxu=DnEx*feS~=K?7x%i^b0E;V^v%((?zwF(Lf
zO5F(f;-V4;bh=n41mt~-WxDl|93n+-M+}10q=bJU?pgJnweQZbMd<?6h%b{{&WU-F
zlkj!SWVA&kC4W8>6Kw8SWehSfF_qf1J(|uTU?%~*SY#cAEu{Zs6+O7Y;eeK-Gya(}
z*q#gzq4m>VLF3)YmVtrV)Y4~<JW1R5NZXIv8YU(i-@|`8e2kT_u0_QlgZh}?pg_`B
zZ<5XXG(e}9yVM*GzSIIDNZFApy5kjfi-v8EXQgy7!Z>AYWs<gv@#}xUW`yPd2E6Kf
z1F>p;iZ5~ymryr~^EjSHKDsxA5w=~Gww*is!v}aAyqt_Ip<`^r;n4+4cUWNSLAhAG
z<ka8^8U!d@c_Ba;$Kb`vNvr#?d6Pw_;#6GVeIo3NKpFg4@FUBgk=EVKzD=B7s(=2|
zkiSijxZ2l|<&+#zNOhk0>%`l|j5avO9phl7Pla!HUxj|BJF*>b*JdmiweUqD+c?`z
zVx_@oevNlT$o`sJa2g<F)m0gg?<3xSk9pOW*+|vgfryv4XsE%~m|Uo^L-YZ~<zC4h
zhyr}oJtRD5eo_8H$>R^GG|N2+Q!a;!;mL{@HV0u9&#Jw?R7Y{YtTf-@1UasHHS5(I
z{QJi-+b&$RdHSZm4n~4Z*^+G8Vn`Do66LQXo_mJChO`xli*U^!IIkfB4ydl~SfWTE
zsmrh8MaRi_zbgZDsPy1MMewv%0aiGG+KXIfZ_d!a)ods!At3=7=N*!$8NQ`;-ScRM
zATrazo>1_<3(Uz|;CV1&cVR$DOxiy#u5xZ{B-oc9UII$S3D}1lE8{+VkkLt#R8czC
z8k9149K9n85i}2Ay{e=-QdYIyDnyw}mE}3^w45c!eP(UG81o8v%WZ^|!GUt<A^?|s
zU!+>#_i$7{$NhD+mcx45(#9Mgl*YpJI<>xZKHW_F!m97@e`x`3Hg9^i)N)Mt-7hQj
z5V1t*JOc49r_*n4Qf$FDhMI$;Q5Iu*l@F6zEZp970>JBOJR<5CC2g!CA(2rtR_lI5
z>WELC-{miWkQeJwEApT9+~`~cocO1uB^pi?5szGtBr9G({E=oX8RlSuu~0IyeBQ`L
zx%r)sI`~j`r_@}Lc^E6b#spDVlzZ*ebf3j6m=^2{o^Gt&^AK<Q-nj?mPoPz?SES%-
zE?iVX&xcsq)|JN(D63Yy#}xglRAK-@0DvG=ymtRQmX7CUj|ft%;7gcqkP+Y(MS7*l
z7MQW^U})OxI_cbs_QtT~H%qR+3Mz)MV6~ggVYudr!=E%AUSEC5XGF{Lx~cH@a5j)N
zi%5XG>Mn{OFXh4gdlOaE!Fk3N+W-N)zld^640vr%97|MPgDudY(W6thB<edzm9{g%
ze?p>DJfKNmt-u0An5*3O&Cb1&$D)qKW9M5R@r>!Rn1D4P#wbuo2f{3-p+CL)Z%97^
zP~m#RdGq~Y%~*!t<C$TS&x~vZ`1}vCCSNI2s4;h}DsJB7SiE6-DlVQ%y!1X_<=@R}
zCfyX(FvIwY`mc6Vb7KKqLx9480R$rH3z0B>6N)_9PLRh=7-AN53XWqqk2GWO3$#>6
z84O?91Q0Y3QlB$f7-9!}2dA~_Ciuj!cr^ynxSdc^A7<y*7c9-ZIBTZBl3MMdyEFii
zhhKZTSru~Lj_x%{5%Tx@;v9@68a|)~%`kfxZv|xHiMx3xn4yz0egavCD@7!NNWqTX
zCm(xf=oU&lu>zcR5?@CPfo%w`B~wF)=?-TAD0B|d4@ZfKshbTUh>~zbBcyAk|C54z
zAogXCt#K(Qitv#}!TZ5aKALW+w|YIXz634;-fy3+-&<{{r0&(iQ&bttA9yl`42hiB
z<osQz|EK-n1T$B*PFvUJ&}kgIcHrXg`s+x7dN9%jzQ0#eQu0xCarr%CU3Ph&S@NC`
z4I_-Y($2b8Cu1i_Sxfo{tijj9!a{2UEfFk0fAS6G;L5AQ-%<iT7m$e{1sc`rJhh_^
zRe%5f{p*n=H_D$KBF=1yjl1-)bf?jBxF-A}*l-D+SLI!HhCa@16U_S7X@X3@?OD?F
zq4(aztb;P^Z4l3^c+>8^3&2>-Fh<Y~a#(vdT_U+UV2SSX2EMNJ**~3dg{ECVZ)TUs
zcJU<smX^u-^BLQ&)?01oq}2E)JqXM*rrL0O_%sh!ShO{-DGkC=gF~l^#YZyI4;$O*
zhngNRI?(Na<w%ZvyR0XywopWzU#KMXtV$J;aopO2p+885MM|o#rKOc_K^Vy8s;Z`@
zmW~>w%Ej%fYPHU8_S!4KTJvd*Z=rs^T(!^}u=;+TSrq{r(!_H|?r<ziqDUdga*Y1(
zB|=h}iD85f7sVnv$vdfnyATjs{pn`Pq1d#*-Pe(VR#+di;cU0x7-`QPJ$;!(pu{y`
zt9e{=LIKYL;i4D;%{$KN))&9jnbj@``BMwn9ID)33i@IYf?%v-HnjT#qmN@wVVIzR
zW_QEE4SF|h(RIWis7^SO&D<kwf)}_p;eZ3_#rhV0VB4(-c{W9j&qU>IZI;SZx_Lzz
z%zzvOE&dtkGcsxiT05Qy+lOrzVG#wKx*tNT+D{AgQ+#6ejLLN`sv##I(ju#<|6DZI
zo>d@WlVJ;K7;0r5)^`5Q_6Td3w6*H@U<c4!(39$(bnd})UJSI&y(!E<M+Nd=tR|~R
zRIk?`ZI%C!lgAXg+8#e0sdh1*(&u3cT#ny)W=`U*Mw1CF+SeQ2u!tx5qVD7+*=@bH
zhyz4Wj(?v?THlqcKhv%}eqMRN**{Jf_~N*e@V9((WeLBh{8c}YCzJXg562(ruCDUC
z{#KrE__FDGlC#hlL*HM(o8ByY10ELxY$${O&hD6lE}L&xIx~@*hohfH*xrHS&fg5k
zx%G4oZQxR~Fu){**vFkXwJrHVaGGA*U_S#+@6_es)h#KLDBa`T4Jw7sZ~rO5woPi(
z{MDIvqlxj%QU}NdC(wCa)>n<*FjxRDQ*Z(?^@k1=PW<Jvd=$PpK${GSu?c@Wz+o*t
zq6<kQqY741x+A!Jo8L!zc7ovtRS9ad%Zc;naT?D_&~EqRXDhnNqHt$r%kYQZrG44d
zW6^B&x+?d53-$~ZN`eV)@Lr7oCsPJbcaAaDJo6ru+><}O&qc)4I~X|N;`ca*yprnv
zyQBpI6)NZcaaARw!{aQ40XbMZ)3lAs@8SW|3lANg;7hkv`wcz+AU6Bj%uV}9{p@oM
zSD?`E%vO)DPHD|HWB;Za_RpAp6O}#j)cqA=m<<{EBCp24@`sb$(yyy^YBLyyJ&edp
zoNxj2*{B@4q7YJGj0uILF5lX(Ei)TpaAYh$PK_;99|$~Mq!Q@b1TNKpqo19LtMHXu
zMXcj!zX5_VnOBlqqA{GleeA9AumKi*7h;;x&-Zt;3~#1!!16T*n84tlNO0_XA&McX
zd;7hsb+Vs0)in?#l~6F#U1Yx`<1HnB7-8pZ&~J|Tt;>le{lnDele}(d6DtxWu92St
zWS+B(o8$aJsNRR0$!@tNrg=_KzodVp<TD~6C+(8EHTGCp{dyfI66Mv|$6b1m7!uOJ
zz~qWAhP+cjQy5piZ0FJ92-Ra74Bgc`j--QGOjkp|*`Ge5c@5d!<oozPtCpZcbjba0
z7Um2yszwvi3PX#CuZN|9Un9aX=KBA11lMnLBEF&1(ALUJF8b5aRqx-pq+bOicaPA;
z_2+J7nUGa3P{8!>kk>c_`zt^ur*zvmC*N7;Ib?Jir2iY9=*Hj8`kCTp|J`~liHUPH
z{g5Z`CQBj`<;=OWkXMi+w3(;fKq+t;43<gqTy<=-LwpRu75&uvNHWo!mR}fcq08~~
zOJ4?E02L&6NvbO7Q)xw5k!x{%Y=Ux8q;aq}&NJ-`7#WIl=NlRQ+?BYeiX&>8a%}54
zj7?V4h__X`oUp(Yf3)X~!&2cA3j3?(v~YC&`XiV4b@0@7VxARr5yp*}BlTCERrWC|
z5I!Wgt8&r$8$zFqSwR7Z!;8(v;F3Aoc2@@?(xf1-4Yv#bCm4`PGFif=*uOQyB2-{c
z<)r|99Gq=k!8y6wYp2z1WTj^c%2=&UZ-7j{Uhj4OL>XNCRi-RtXM>nyd|*ZRW-Fh9
zCE0vT0rai;27!}lwGgmzqK|1!^vm3)=gDELQq8sk)t-h!m=?iXKZr*%btC8)zEAx2
zWIPSs0Rx*rJ-EHZ{)hb=_P1B($yLD|f^+%KHT|bc&r#fmk1OpUz-kBes&-KY7b6Nu
zZHYpD^WiMtg~)vVHToUc@_nyv_~c@nC*Ow}BT>>+6tVm5aBd_f%yTdPel=80(nl||
z<`4g;nhE)m<>%L8Vd*wMK$sb79H<NfTH|J6LK$KM<e8nf04QLDn}ps)Vvgt8`+++H
z4)m<Zzp?x@3JKb&*Ku5fl=roh@K`DjmKq9?X)#)IFx?Gayn2dh$B(EDU#cetC6M`Q
z#a`6D`6UDn;5v{U>!_1PLjGhdi>xw^BN})nMI7CQu#cx(r!5oYlw*!Zz)9%8w%$LM
zRsX-)jd8E@d-Se!Q+$%vhi8?*E3<rnTx&ZdCP{3`clr&U`#Q%s7wt~N35pBDqKY)>
zkh+x2RmPRj#yG7&gH`#e;v+&HZB8SN;_(X5baJpVf&OVlHGaHx$ZD?uKw+2w<DV!T
z#L<|Hz@6Ql2KGMZ=3FJOl4#2&ESS|InmJ;>{Z;r)aTOC}GDqg&h8?U_yRj24xKa1M
z%+TMlVf!q*%rtSB`9g4z=+<t?QOA|~a2cbTX|8aqEE#vdjUsqlerqQ@qX#{$d21;9
zPdwQF-c#4DwZPxt!sOtTCycWS{|f-WVC-Ov&ir`Jb>eK=BgIrC_Vk>Ip-<aTnQu(*
zuO`u)(rq^`d<eu6<pQ^_*73QwPc=&?&$)St2tv2|HSEamDEZ_#3urMOw$^=;kGa#N
zkkrNe5PT!lx&C*%Yx*EF5?j`Dkhp2>5l<BM&kVsb7`lp)xIsI}Kv;b1j%1W-@<l!~
zG3GmN8&?ePzpO|O{M$Amp{be>tmODrU|M@d>(^esXs2$Z#xiX=<t{hHM-TFF;}>%2
zy{*V}(9c7>dNkw<*Hp<d{q=|bKi5ZRN|&YSvOgvzD@uJ=T}a|YZw<pZ==1wnCi7hP
z(}O?N*B|C3s2~}o4le!(>^WPoMjFxGQ>LyUNfM|1${&bLX1V&`?GhKS^9MKVdT=W9
zN`m=++WwI#q6kAMD*Q<y>H65&@(16HK+Ai56#F@Q(D&9<2p>$xRQAUVNfb)8k-ImS
zNlVA_cTi}7260oZAoRGVqMhF~ot~#^AAzpkAM?w6(aM_L3hzIogLUB?3rro<M0U#5
zMBKW9^_~Ckw*q#fJ2!LX#eDZe)#h{LAd0$!PjJ5baDac|gYHm}Fa4qi32#Ej6pqM$
zL*YTm-$%8g3bPV*>d^n0=gZ>#!j?%oJfH46&Z8m?;(s=Ew}^SS5l1RzIVxc@?|)b-
zb!2N)uaVIs5Jjz}gzNo3msbq0{z|}GFn)v=5skX`i11~BDXC`{XSMUEDJrN_QT-?O
zVwlFcl=CBVKPLozI^qj?nP|0;LfT!9e~@(R+eXQ<|1^h{2rux3x1oCvv}P4obAd#8
z$;yE2H3CfDVWR2g-yVd21FX($jpRiY`3Qx8DDu{0dw|wb83I?f)U0RolaS~et^O@d
z^;n|oM-?XQ(d=A6#2ZbtQV%$z%IWy%h8jSFohU0H5(ij8J;F;M3wQeee(YR;{kj-i
zs`IGY)O^iM9qLv*@KB!Md{cVsbtKKRp0wW0JM(u?001eJvR;{qvywvX&lcQpGFqgX
z?`u|_0M-1s?7?=a+P&c2G{sbtXS@EdEm9yTraJ|46sPU4HLgv+t1S=Q>>0@gC}BYX
zH^c9xtaSv+lnP||?vz4XTTB6T;FSSezJ1|H@lSd9fS=3=y4f*-!T}M{Dx;as<4uPt
zsynC$i)C9RaI7t-qUPrL1D{g`yxqPyqjh76Qc+RCF3E-({}!1?jKMPD1k<CRTHuBn
zdkA)1=yozP0ht|M;dTg3Fi4n<!RSLuS59RxR;V$9nwEqz>{1vTUUU{M|Cm=RNVxS!
z^4u-7YT=h(_lDrz>ZQC_P3au_+7KUtWGu;^QzLh*GdD@^2y$QtTlCNtiY+S#QOw){
z`t3o7Ehk_4XaHBqj5WmG-DqjrlAWXv089h87y*&|&{sQr)S-)(96A`*(t*!Tje0N+
zb&<G9#)L7FIP><wxkm5;s?-MYzAMti2Q@Xz97}JIK_J%!j&w3lSC?N-CTAtY)?k|-
z)qTQ%h|pL5@AkIA9#tQ<=v!%^586q9QlHS_)}VrG&7YGf<{crP5#hT}A?O~j=<)#Z
zK3X??E-Yy6%BvNq7Fa@)1oGXp)~3goB<<^w*iRtK6)aXIA1YN$*87mwzt2{VT#~j_
zY2<M9eE_lj(N=8p^U3$&-^kSL$`Q}J?~owENR;p(VOT+~9WjV!3c;_!;8aQc!i%m5
ze&25R8nvIEqt@Yy;gdfm`w~=~#4D!uxMOP;Y`-35Yn0$gykDcf-`(Bq>go!>YzF+r
zz+P#<xOFZ67-DW^C*g#YgajTQp88G~N+4D=wy)ZYU>E&1T<||72k<2``_KAXNiosU
zXtrEk%9RQ<pMFoc9y~c~8`gXWfO(wn;}kiM|CSRn^aDO+uo{5Z|Nr<Ht0W@^33F=D
zE8iF*>3fex9ccp8gKTnZC&584Yjw){HjzDQ_{U^_;LdYaRbpv`m}zS$Cm7*s;7YtK
zeOoq^m(^K~$6TsVfC`K#Ho-nd!q~bat|3tBLDHAl@{S_#@_yF3mL*v>X{&?NhK&#i
zVE`OpqV%maOho_%(tj0)P~go1#V2(OK-*YAaR|dS&@9*M%kyl-KmK#Kn?<cMTq0#s
zoJTu+Zn7xxT-a$%K*V`YFG1^`Z^*P$rhfO(yi-s%ihj!W-lKRo9GEoGVEiNf+ih<l
zvY>ZxO}Byq<B$3m&o_I;K|ENQoZIcsN8o4ZN7UFu^Am5LyX3h=G8RwO=eP3V-i9oI
zq&RpS@3LzH5y5w3@22aw6xjA*x1<Iopy4iYB2)UN<s3R(pz0=gm+4S;l##+yC%q(X
zD1XwO+(h5b;I|V&4bZR*Uk&a(2aXL?3D9i{g6bUP4cdHoK&xl9M>a^z+}@9+VS}$c
z>lm#JM&*^?21D9W&|WT*B;_oiVM#&+QT(~}K!sbIf8O(rD;;;)U1909Wrx}j{Uzd3
zr%|snMWS@Mcz;%w+URHKuLlcK{F6kQjEm*Y@)+&8v4fY{`ZOs<n%hc4Yf=71H8Jp|
zLrtQ)f}UO~tA6@!E~V+oM=>(vu&bqtaZ{`@ewOY>T^AV5j<(OsBR8=h4?g0z^RwD{
zrGO<B58mi+y>pnxx@>H01h{f(1qB5;IkXo~X&fs}(&ngmf+IC5A{xIqN^Jozrgzwq
zPCEIK<;@QjV#25o*ZR@F*>|s>LKSE!irh9Qxl)St27^!n{hpyu4zc8O?aGtq?PFJ-
z_d<YP5yKVmKkEwN_VXWzKorGhK%o^OAdu4}27yde7yDpA;#|jAzKP)CjLOX1OrT6<
zxy_f%9`ZN@b@#b}xMv?@0C72wG5ICakW3^jNG5UlHzKeQSPl#9$$T#-!j4HNN({OH
z@r!AJ0_HzdV-WcvOHo0=pwe?FTP!T}9ZhdfPw)#0t+o~HI}R5xVJRmoi|Wb>xWz~n
zv2kj^g8{}<W5Yn*hnMoOJz4^o1-t*Vr^IzsYusn$-pfm5ZEKZ3-y95}gb^y@2e6$A
zCr!v_Gc#WREB0#j>j}p;uW3Al$<u&@+mMqtIZU1p#4MbXi4O8Pe^r!Yi$5@UL)Xyq
znle~sGtR6-D9ET80)4u)3L>U$H1k~ln+~uyxZQdgsM&#QWyxyS$3yLlEzb=%s^$4J
zDzrP{XGO`WcD~Mk;`-9CNy*cpwwUGzJF+GQo;~-Y(PGQJ?QXg@s~3aB)VV{A=PU0V
zo$hR=+5}0L>L@M9E*Eo|8B*q79#Ab6;C`Zl3{X!^k}=wb0K8{st2c}uh(Et<!9Qg^
zG_gW%dW5^D1UT6j*!+Xdo@DvJ3i^*tRtvtHNf}3+e|_$;m3}P?B^{DB+nsHOWgb7A
zU5kt*x^GQ|(XtkOG(B@pHaPM$bgLM2I179UMDhT@0}rvad%e=kVute#%$k~r@=+){
z&#{ve1NQeh>yYX)`}U7t5f`^sba#BAhq>B)*96z6xMN8+i`T8K?@HXK;}ZClRuVH>
z{UZeoDrkScK5pVze<1eY{m%nkw4L=PBX>MZ6_7vq<-KL}&Ga~jKAaFd*@w-Yf+oqQ
zXv-~mZgOc~yYX{;Rvgf6bb8&;lUO?NcVI}9LS%5PW~INr=%pzaQr7mz>dVIqwC+mV
zM#G9ENqFc+($>-v9Tp5oPL26#wVgH7o%bM4KV{=OUhL6VOTHe_{%P$w<L7L~Qhlaf
zk7m<@y6cSm0BhU-gjX@MXRIZZ)o^JlM<oAMZVgtG%$uVl1a+??*@AV-l6LL@WBKD(
zfF=b4LotA;@)7{1vtkZN*>uW)ay;UBa{z7?0jz1IP8oDH#B8Cz-)6Kt0sCjGeL{|^
zx8mSCgIL?ew@EOLW(DMJpRLlY2NyDN7N%X+OTVSQB^B(<XOA#j-oZ56SH;l}Q%r9A
zKy!Z^X~)j<UdDen<{4k%VS6H2h9$frN94q*F_P7pq>n9{?*7?ofR&_gpqk{gpkl^a
zfK0Iiyya=|1GeQ4J}RoQx>k}Io&>U_anSn>;P^iTKQntSc)V>gcGlYht%touLV70Q
z<6av}ioevoj?D1Ty+|gKUrWL%l1yxdlL5Nd-ZfTJY|&3!FU5cVr(>7kE-6M|PXx}1
ztc{S{;r#IMu#!7q35wGyG4B^@EaG6sFvX5Dv}w(OGOqt+HE0zx4haoWL9dz3-#-bz
za8#n|3iF7*Co8S1D^&+M-HmhtxzHo`&=&)Nmy-Yx!_T*8n_MfCA!7*9+S^s+w^m0*
zjP1URTHG$vT%#&sUhX4ae$x3@t5O;Y^A6UWCOzS1{LiZ+V6f}n-Q;}L%e_C@y_F^s
z@_S^EeaoG`X_~w0b1`zTVs`OJZoSM^iM%7wbY60Z5oB}pV6WaJP<QcdU4H`x6rd8<
zO%GC9S~pB89f`KvQ_jK%)sjD5frq5rRA!m)x5IKh?Z{;CQegsihVv`@mv`*PEk3vB
zo6oi39h82*^MBg<)O67Xm9J>#5S@?Pg)stT<baYmgUvWlG8g&9x?3wzKUp3MJx2(6
zS<n0>FzU)Zovv|~iiu(VqFRq*G1%ICJNJO<<}I>#)r<I02F@US4jQksQTa11jauWY
zJ35}DFYC`um67hfD#my>{cE{m+a4dLR7kGH>9c~~Lpi<c5@h1-1(-bc)18n4$;Y|x
z%Uvzd2xap0Q{J25%-ca7*@C+AM$GoBxvTmLqzCL5LXt!Oq2STMwF!Jj(|Yx7uUofX
zdIj3o)|&FFF+b-u*k&RSy<~4*vF|EEbsy61f)e1_(I<-&758@-8zG(MPYw*|OW#<E
zd)w*4QITx{s!+5}^2aob#TZX>qk?`_fn4*y`u^Q4|2(ms#U?^;vV^kFc8Y}?yDb`q
zg<+iY;%}>7uCrf{(m49{&GCY_o+&quC6uh`sRAyODc%`Ge%|eF{%KEh%E7l#RPo$(
z<bEmhRqsUVXC9&eHZ*ly6!D6=?;AE1h%a%ajO=s*_efZ%A#@^4x85Gu1y)v8J(;#M
zt4{7^&*0m>05jEvK(pET?GfwdXCH^Yg>cks#-&&B;Ti59Zd$7wCfK%`d-u-$69P;9
z$)!8Ij>Q%9LF2z5$_Hij?VpIr8?DQ0$NFl2GUIvk+Ro^CAG;Xb!KUuqu`P1ySIjj-
zj)9D0!ZNVR&WpzF=DJ|P$QE2awHH`LE`UF)IUjH`NBXPFiDc)@quhU2h%mHh*YA?1
zyMS=08mPRiP%RjpQ8B2PdwxKXPwzpJuW3G3h*j7f&seBg&sz3-6k~Q8wOV42nd;+G
zj4=yYWS+dgl+A6fPY)+NUF*0<E~>wlW&HAMyxiO7VgT+M;Yd|g=lAh!ec0XUwm_W*
zPC=~n1db?sZE8W!`(P~nbn}?p@(Vl0W2v}@iA_MkrL46aM|P82!Ol?%1cQ_>B|-o1
zy!D^kEHB(JEYTR1Cz?=oUs<)Nwu6D1kOTbg6(ky4fO+w5fUo^52F+r0)_W~PM4HAg
z9At8xt?FYPPuT7U2?0F+hJ9D~fAN{ORq>KEhkD7%PHHEkTIbukB?{}pKx=_J@1l1T
zqX=4Rn+TEFTwlxG5$NN}!Gi*Lul=8*YVAp{2eW}OV`X4?4WI3ObFA6a|AlGDx^}U{
zEo#S8H1OQ}y$BX8l#eNPGf?>}1zwWz&j8H;ER_+|g9kArNn;Ko^uAZzD1pjav75j|
zDAoAcX)2*F2WJ(K#SJ@ygt;>=xcNcvKv@$0%-ZDdJ;Ke_F*Pyd6&1%uM>z@6UaFeg
zewb;CB8^>`>9kR{nZuD?24lr;0$NM}NGufomWn>fM~BZzs7oPS%WCWDh!dFclR0`u
z5~dKc&n^k%lS(_s5<{f&QuB%q^?3b#)!@9@Nt^&ctuyJKc?y9Q$(`UPiQRvceOj~o
zsx*R1+`^2NreHw`{E^*N&BTZ#YSfG!Yu`m803N40dmQan;IWv4(T`Gt>B&ihRs0;K
zaHUtEm^R|9Q!{G7U_n3$(4H_~K59Tti@K$rTR)FJWDwO0urWpU1N1IL`Q{A!DIR2L
zAou<K_BDRPJ^(HRZsk83MG_LN-YIQv;JD8n+^Dts+eaMpx{Z(Q!UDEKx8W7=@!uAi
z|98db|JTmHBu9z`Q+S5ko|<>qSpTI3$T0r-V-7yJLR-el2>c?EEuoV!TKdb)@(B;g
zqlTofii!9X1xvImud)pVK3Dr?fdFpZBLma6M*UZ1@VAa@O1YQ9{ALKvGi%?MymDkD
zWn4I@Mef{z09o7%cmOOWBQ=Vb9JuqP!XK75okb4?fEth~jn%Z2abNO!FuJ7iE0ID=
zz$4EbT)i-_#W^B|Zds(K*<4RjMqp-p9>NiX+|JWt;RakzLj7J%TN;C^P7i~1q@Dl9
z3;%cC`Ty?u2Jjih#=;Ew$dYlM&qhbYO?}K;5QyDjXBkt|99uGC|EfP%z`BYnml}v+
z(ILUu0<|YP*A|ele7U=mvb)s^UV1oCfRlYU!`u;T+lQ^*<^cyARzq#io+A^*%$Bn@
zKoxes9E{!i!*?)so->0+E*=+><V<E+h)D6JmArYCB3z<iY}<-juE)Kwd;ObP3IImY
z&&Vv;M7~dE*~cG&%Nmu+$mdPZgit*fF4r{K4;fFq)2pdWw8uv!mK+h)!oz2{jYOAh
z?3$2A^0KZU(jf(5r7JitlHy*%hhx1d_;UQb-QBy?GmFEV{QUk;93Z;9J?ro9ueF&=
zvX;OKE)f(}OCWqB&AP1>n_ts<eA;e-5R~0D_6HLi8@tm|qhtmlcfkg)E)uz-3y35U
za~F&%wU|0B>lkn~k{n#>gh}m1$8mzbx;DcAN(!=ckrQx^`}U5Q_z*U0-EO1P;LcJh
z>va`YK%Q*!{V`CZsq5pPuF&@-ZN$9!kLhXIBES_h*+KN?f8%XEdZ_@x=L<T(@YFMi
z_m2c8JWWpSG;-u7`D&+HGIP%Ff7GQ>uKiy+0H9EIU++Is05%kxTAly37_s0xt!wZ7
z36jhJ-6GsljH3(A_HN3ebkcAx0iGkNP&vEh8d&GdoGHEP@E3tFwuHp^IoIvo;8H0K
z2VuliC+Q-201I{Sx5(uLY~Ke;tIQGa2T~*J)r(L1l7daQA+9m`%QL<gI~8r_Yj_Z5
zQ-ocI)Aw{kGIqR{b++S=jq=tjM1*>rkyxV1kc6$o)ZnE|kG<=SXE6W%(PvPAg+WP&
z&%J9z+2T20-3N(~<%rysKlCuRZzzMiQoSVG)Jo!Yej|WJGM7aEvF)M&;`@R`BQXsb
z^qY)5k^WC)xWGJWni~%UsFD8x{%Y!l47)JG`{Qqr))h!CoLT>ofNC^_%)46+L@QqC
z0V-q```CaWYMKCO(d`fg^!Qy&&e-oC#+zNxm6d$(cCZ#lSed=qE%+aW;KhLEb!(wC
zM!GC}2m^h?3nZbWt_ZF5BGiqN%v3|E5dZ%WjVAkOUZ>3MFl#r*u*o0)0qjaD4&z)3
z%2c`GZ;`#9QT_qs34qkDgUuvt6qL146aUHNpM-%jcO?t)l+#XH-rSMLlVlno3Ep!c
z3YaYn1>L7FIV#0r71vFZ03EKi6^nVd;dJ(J^PkQUF<^QPI+;8u0IfBq!{2|l0=QRv
zHL{nl+DoBJZd6_%{<}Q(Mew=R<3<W2Tw}_J0Sv*+w|hWO`)l^7XhoRB?RahFA7*Y}
zHwfPfV5KKgO3E1oOZhMXmCk(qPX7XWuEchc69GW3>TUQ8QLVh*!=K1cwWEFdAr`b)
z1&UvE0i|bXMS1+|{+rSE?q!=*51!oaA3ITg|4(6W9ah!%{P7;TL+M698bP|dyHn}z
z?vj)e=>`EQX^@aqy6Xr?gLF$b+{Nen{GR*Yz5jSP$Ia&Kwb!0CYi8cB@i$hQbs856
za2^PgqC56)G^w!+)J<jS7-2(y^>u?QB_AqxC~(qQP2g!?R5`DnYDO3V7Tkfj%Leoa
z6n{Nf)er2j8h=hRF{BYzSE{2$!Q+Tn2rFj$BN~@mmbJHL-DpB%ws-NPaud(={mt^=
zSa+)wZSzZegExcN-@k18bX?c5^2!@(@O?dFsku`zt+%~AR-0<6VOC%$;`~RU2v&sa
z18q@8RS=XDv&9c#>y$xW<9)Ll)U>9hD=zz!3EL3Fur*8PjAJ*t5`x53Mr__Uze>vE
z{oK&reXd8Kt(4Kg<#I0&7)V@--n0;~`3r@?L#fDS-s>fIrA4EICzk_^bP|0u%s6%E
zuhkRp4bdQ+pLO=&4{HA(>}CJr|KfT7#R#3g5PU^=F6&`K*T}cjOXVU&e_mNm&m6KH
zPS2Z+Vr^$GZi+Z=iBDbKskedzsovZ#GZulI%b+BwXW)3%|L*3?PMJ;nI816fPY5J<
z=BjgwThOY5emIJ(1$AEsO`ypS+5!oz0Qr5mX^Uy4x3v}|z4kC^NjFcIM*?io1rCsF
z4uhjh+c)jdsU0aw(H%+O&!0N*vBjiRy*F_*HL+h@;HkkdMN`SXz(&B`ZI?a%0FV6j
z%Jsq1$!q(L8k-vBbC`lF%Guq1^TDX_mGc!SAUzm4p>7~U6T_BNDmBh$&Go>Uooc=r
zCR>^*A`9G`q7yNyHml!t%D2gP^MZ)VU{OW3Hx|A^4k`7-V>*8tX_@m9+NwRu&0ovk
zs#t9C-At&rAerk9_C}K6zknALglE*=N$AkfvAl@Yp&jFUDOBNhQjyhYGR?bR>^sRx
z0mXs*NeZGaFZ$m}4Y6h`vR<0T%<xiT^W_XtBiRbzhfsdp(^?zk_^y&fFLE=)>Q*v7
z7j=vN>9RUi;oKHo+_ky`@-tUN32$TClM@77r%OY^$+w?w)SJ|djPgmo+izLTKdv0X
ziNttmJDE=_eX?c@=Z4kOCY{$=s0A?v%|g^q^;wOqB2&kaI5B=@K$S`Xwf;B3MPQh5
zC7VN5463K|b&RIRSL&nE!eia}=$jk6pGN>6kG$n(w!#NhZMsHUF)k@Iq_;-)XSzU5
zYopoo|G&XINZh{@S+H2{0WuWQrLnBe+N*xNRJx|`-xii`O#R1YGQO6(?<w6{CjKs!
zbBL6{hF|l=JD8+CZgz@)yi`x;S{n&Ruj)q<c)IUQePftW<vC!uJhoK{BkE(!^E=Xs
z0bH(Ei%~Tx!BINTW-Hp$?HU*L;ok21PK@-2-Z1=&0*#dhI=4-#oSeaT{H%@k8VhUS
ziktsDFmg?OCp^9XT4kXvAo@;Lbt%_a4|ZQD?9=;X87$YxOWnduGac3fC2V-L)}w{j
zI`B>V?~k_&c?CSQvPD8iEPJC0{T1WqF(WCFyHfTxY`fn?>LILD@<3|(F{zVk{~I9Z
z*Xlb?E{(QKH%L^IWrB5`{|(E3pGtf@zMzNH5R6aGWi*gs0EE);XYnVjGBLASIbP$6
z#isKA9h(tZG*MDqOiN4q_ii+K+wTG{x<JE`Eh^29mhE|7Xj9{?)!q!vXc@fW5AhRx
z+oJHZaxO-FY;cA9pxstm2E*Icn`C_N-PtveXEM#V4ku~QB7{9hSkc{Ptyui-kPrD8
zmC4pGk7~Fv=`>hSLQ0Wr3uKOCjXD{BbfZBd+=2pHA7-@c|1<~9u@)aCSt{-&*4y@;
z*e*`73J3^zU41%66uNcEJ6fugy1P7Vy*d$}HPkJXqNP^2%%egsFf9Eb&RUeKg1I|C
z6KaPIHI#3r1!%3tdZpWb?%B8GqnUSGR$1jvCop2z?$HKhZaLy>&b&tfg4&EXM*$E~
zmwPq~=5LM7$RU)rvKZ(~#`SVD``C~vKkE@GOLTd7P0_1<ILOzwr1HiE0p&{u@*g*=
zp;%K>Q{VLu?{Cf&$W_#E#Rs~(3vElT{{3T<YPh~WPC_hlO;U6s<W?-4#N{d1WE>7s
zoF%`W(vwx1B%PNK;_FZ^kwD&cZ@GIURMq2DS~2S>vn6M%y|TWniVCQ)lry#XF~YTn
z1=fmb+kHFJ_m6FRrNs2-A1!Ru;33TXQrc?ldxs<Zx6rgK#hkn>VfyPCf*2q*)n&%-
z?Yvt;4G$?Oy}Rr=UIT=RrOu$vY1{W?BnnV>7Y*j&=mPJiS(xb{<oW}qkDc^j)<#Nc
z-@ile{wH{R%JWsdY{O@{!2uMpCSuHTGsz_mbEr1G90oZ^9PkhG@SE{q>B6%9C@EwH
zT-4bJHG*l;{F3`n+>Bjr(z3A;XL$Dse6M9P0-ZN}g#0#+PZR1nontuy1S<o>cUFEP
z(FM^f8(1$TfvD{yF3}TPJcM%b;^J_3w84JZkXORt3HoS{Jc0NAk4*VAY6zv=RH&rE
zOsLftX0XL8<<Im0+mYul;4^@O1_1@7GNw7oCG^o%4vqXT(IB`|THo|qx{eobD`|bt
z4d37T`d^Ojy&G+a-8oUR;d2;2<$!TeSK;M5n&{(#Bs;bKs#R(@YMkX=;Yj`Ao547s
zk;^;eb)@~@Gpwnmq*yIUPG;m?C}*4Mwle@>!T3||bJdvbl$F+>86Bzgd0uACQHvZ>
zBR}f3KYDl@BUHCSG^coZT6OplOs8)J5%|e6*<PX2O-|)-RKFp{Y`;f-Pz2XIAaOs!
z<|OnTUPCXhIf57eoqeQ4{wqwL%4AlK{Hh~}bb>R-H|d9e?f*<Rs`}2c#OVpnycaCQ
zah%({x(}`O`Sd6Z=T~&@hYWs%pO>lx|Ia3Jf}^oGn4fkiA;zrP`A~IR0aEZF54{TS
z5bvQ~8!O!ke%s{0;d&8k20pfxLJg2Q2|zMyh+t$^C0*3`pTUM@IL8auV(d@}f~aQm
zOCvf1V=Ez>Ukz_npE@>R`FsX6qMxsj3#ur?J6x{=Lu=)g@V1yR4=WFK!B=Ap!@V8)
z#mGhBX9tOtSE)P%zfXA-k9#Zc#%=dHLgoW=`Qv<8w)lSr*@fckgid&2(7O|K(#@m8
z|CY+0fZsRb$L{<qe#h+~YTyd+G6)C=TyE+WGrWix2w{KFtp0>uXYw(CRdpI1`UFJY
ztdqV6*<4l@v$-HNSyg)$9oL=XFb@~(+MhRIUQv7Y5-dd8IjrZ~V>zMZ?GQ*o=A&!9
zbV<`$z^$Bc;DCva!vopCOfoRu<@`oF0v>zt3yrXlY@suEN!^r%XXR&_QQ!|Rr`NPf
zrkQI<Sa6xe8cBF`PgSR7DBx~rfm7qqDtarF@7?`cltW3|8p|<gkoE)t;q#Qiw>o2a
z_vh<UG5A{d9}<lq&xV{H>nNXns@&7_`95e#2q<b%WEMZ&tvm@EQ<HGp>rPcS(Ulxb
zQt}=hca|t+X`l3?lC@3@@}y_YP6yrF`dCsy=3G8gB44R*`<=g8+uTesWrqot!)9!E
z5ETE^{Nhh{aIBPhO)e!ek0h!P<R=g>7<FcFt&Qgt@;p8ZvxbC8%QYLDUx-2#r?_mM
zs`fsDvlR^O-A}|RT;JVQ8+CSk`SR14O;uA%_<rLHc84FuW=GD{r^@7C2B;1@r<Bjv
zjcC$DYFm>wpOa!j7+nDWt2YLN4V_vpktt8O&-xHG85>JKN6?1Y+T)EKQLu$x33Rs$
zF(30aJ_N%1Y29aeb95%ZXNr?$*wupuq$KoG-!IsE(4-dIWEL}J>88kU&a>@N@f=PX
zRL&$WFs4MCq)%NnOM0|*sCE8oS+sxm8LK>PA(t?cNnkO^i#V>BpB=cF9-M#|@LJB*
zp<cdd%CV`nZH_-#w@95K+LY_Hw);ih!gK<7Tc*j#G|j98{W<|hROrPrQkHII)Fb~o
zC85t@Mg2b$0QoF_>^d-RL?}p#ZTcohC!7C&rs1==5ysa6OaC--$d|>Yq#rJo&&%;&
z6CD-QfJXX`oQx!<7Vqo2p|EsvQ*({a7Snm18d6-}i)q1h{jn#PZ<L<&<Pc~Q+4I$B
z&PQ73NnZ;OnGX>o$^ZVRqlSZSoV*eh6?o!IF1}9m?{O;e6)B|>Y;VaJwVy~Cet7Ts
z@O=`E;0>urU|R^3oUObj-WPTMT+`&k-t4rj`-vt8m`|wR=)yZXm0LwBCJBDQ<TGnf
z%DdIO3eeE9jIwUL=h^k{izUi3eCk}M81=i|E%C{%9+0R}wxW3CwZwC`Y;oPEt>cFk
z!)_?Z#r1`J!4^E)J@JxE!h(B%)uoleT^SnLO;7*M+^Oolikg*>@G6i0vU7Q9`C#fT
zW;S?G%2wfCKZ*8FsEMlCi?^NrJVHVb1v3kEPbHtV_{)T((>P9o(iXz)4V!zhhCek&
zNwe3nvsxr^{?L16W^PQc9!v<N_k=nc+W-C~92m3S=_YZbSPARIrtPu;+T-gxZ|L(c
zGI4AtgzXz`;^~&VGuW%DV*AqE#<fQZ^ZT0CEB5#GY45F#HIjwx5JFUs=EmOf8-+uj
z?dFJ5ny51hmz2`QMefewkG3y6!);xpYS~FCcRq%{+;Srb`~^c9=jNXc&*QyMOz)|w
zJ7>2hzmtEw?nX`h_IN4S-_%ra)ah9m01-lI=IvY|tCEr$j(aB0Se`HQjWneaLG)DD
z;(a5^tW|Mw335sn3be&zKRrF2QPT$k2_p?a5-7|JWp)d)H+M9wmZt_@W{S=eT08SF
z&^YJiX4div37t*r2%r9zVZ`$!3d+fueY<ws^E(#GErSs*u6Dn=re4VQnZ^<7TX#Vz
zhRs#@NbpR#%i)JRr$BzY^u;MCFR33Zg2!-~l3u?M|Lw9l=aq{5Ck6SNZtK)<LATV!
zHg`2|U7hJ)!e3OE?^ML@2~&fCeqcQ(CnpE8@_mXqz1-Yf+>l?sOGlP3g2K6?p>_<x
zX54qzzhg0VoS4T(SxdJ(aM_0Er3GIB_cKzW;n7NcAGmjp7o8J+K0c%SBxp_}!9R`#
zV~T*jJnf!AzwowJ>|gBw?YJL_J7)?l@3zHugIu#P7fU_rIv?RWP-MjszWv*cCDL}4
zIG3(5*Dpwo0RIY{`N)brLIPW|!4*nYe0q8+??~Vz2uiOgzKTV5|2k`njFcG^^0F5+
zP1tM*@3TOVn!Pr-oBjSbdA$vt8pGG2mz&^St!cOgG~&tAzUj_#&obvK1Oe?Rz+Wz&
zoQZFtv}EJUsvqgQfIBlmW7A>aOK_8)oSw21$9Y(<zeuBN%>3W`48`a7NrRdvkrSMU
zySux<HWl2i6tecYyxV0oN=+@uy_g`2vTmy~88B}DB(@V08pJ}~x|y=C0cC$`oIAer
z@xK1ImPT$<`E^&!B(4tgpw&q>ef^*?H}M|s#j&cBu2e~p=HHv=J=33Vu#i`N`Ldc}
zm5V~k=R2}931h=BOah-1Wq%scsq!(TVdylI3i|nL@>mp)@HmTX>s{=K<!I)itUO{c
z(+$F2W4z`@n{H8EypKLAvUr<BLm*y~Bz`p=a@U1yJ`zN?&HJRW*@W$>FyJZY1n^W~
zMIpoYdDT~gANx8vQHl2QH7Jv5kk9Jf7U*b&JgF_D#MIO>oF<a!RF=01*zadq&Xl?q
zf4DJu+m4hSsKllabBPsbTrB8s{pBMN;R&H6_r16o`V&1oG7|Pd0s;{fd}O3&{E;VJ
z8m|qNDQCbfGKwqqkSgZz>{ly)D4l+^v{ZJgwESo&AX<I##Rj5EkNXT&YPy;NVLy3{
zucq226p!#tK5%YAY7Gwc67Kv9p3WB<G{}F~v^a!(?b|%9_fy|**M$-Dy)VKbVu$az
zg-xvLo}Y%cdNTM0(@ChDUoO6w`DZP;K6~Fr!k`PYerP#7ysI~gQzep--PkgbZdCg>
z5U8dAd2jg1&=rXzHekov>P#EBJ8tEFCxq*YLEfcGyz}m;bvgi(F8~^{`GbX@9*cS5
z1HVip{p4XM?^v~d4(>8-a{JjU?ySlU?%_IIDU^^H5gNjX1e!AMDK$WcqGKKZlGNYd
z<jCjpS2Ap3iF~tnVu^EayIshvffy@u%#|PQHJ_ZbBZoo980(@{E&fm)Qt<YB+P#rX
zuKmAQ3iViI*Qev15$HtX&~uTXLi)D_46c86N6=|mGPv=hl&Drc1yIAgcHK{AUZZzU
zaK0ytO0k&RtpB0Oh&O)r#wnGik+;@01^<pIH7~C-=qWI^wyDW$FegRn0BFQ@Ds&Sv
zGKlc-q?R@h%^LJ$qN9`LEqb$x=;e#lweGKPzad6AZj6{&2_d#k7PV9D%;zhm<>VCN
zL4rrOx&1Go6eSpD6x|RnmI|(7kRyJ^N3Q-t{?e{LZCII!v+6f<^v9IliA%5Qtm|#o
zgRVL|gj2I7e(!s&z7kKIzSNShClK^EQX{2d9HBvs@`O7<2J8bI(qrrUo_@?8MApq^
zvXP=Q+xkIacl0e@9buJ_W+m(}o@H+{3-Q0APC!*|6&V?+lb;Y47Uo#37`hx_d@&#I
zdy&p&B92GQ2k<rntep{bxd1vfK9>9N3yA`m(7=3-c~DxeZlF#owdr8F{nAwMYcaaZ
zDUcbl45#jrzx2|93nwSoFw5g+hHM^|bK%dk(SuE51CfwhW8mX#5!O<>U(=+uCSyvG
zoukXsM>mglGrdN2rs4kn{-D!Ax)4giyMuZFg|!}hc+XoEQ?dpZ43E^=FmRnV=sIGq
zjGsMxik7xX;EFpfSAJ9ky_nheCyMy7q-DbJZIMn(Q<Ky2MCWa4mrenNTJhrViU&Dg
zQ)X^&c61FgqSE3Xb=(U|p&r*)Vaf7E!&#LQQ7nPf&1`Vs>mLr$zWBEo6STQq=yL|>
zJZX%ab-ow2>6eV0_V6z8?I*W(8pVnXQK{>^S<D7BbBx1M1=a}`<<CDe%hBpJ$|g!B
zvLu=(Z{T3QSWZW%fTg19tX~L>R>2)|L?B+q?aG4vnL64TzgTMqAPaaX$K)$x6pUPF
zW}q6t>N=Md8biQpX=!Q8nK_k`Tx=8ed06CUOjf0FJ_u4mzoxBL@_m`M&lEWp_x6^Q
z>>=y_Zmt(#BDQ{g=0;7;5hC7*N92W<aVqK3Ias%1o{^PnV@$Xd+k|Au3*{8LU{tT5
z;4D1=!E@_LJ47wTVK$s`H1xvZQOnc-l=%1}s)BFSaJW*_j;2iECWL7`21;PGEEOUg
zE425&$|cQ|e7&6B-x|+b0jo13XL@q7yB%go)78v<mn&T3!XeXa#UMMeStRBKGV=Gq
zI;lidsYzP17#cmAUCCBVV>XbL3*I(5NOG=h_BWvUA&bj&+o^SHE0)sYo-)hvJ+<IB
zWY&>ZLKPbloYpjL`&#H$zU+f|d|O-Dz5k0YRk4oI`h-jJycz(sWa?NuWilIXSs(i2
z*Xn!0C@Cp^f^A;CjD$swj*Q&Ao{mGOjGlQ%etw8mJ{$OQ!EdnHrLOZBA=0d~*LTA9
zX~CB1WpBX*Vp%@MC9E}rZO=s(nbtQ<#;Q>LqC%&O^Q5GtEJR<JV8b$hn}s9Ryugs5
z;qbgvP<USZ-CDyRZ0wORNkOWowzd|F{ZSWAr_VGt2zZ7McULmf(jBKmO!}3LL!-Vv
zK6;fi`km~TtG_})I40)fM?xT9Uc#^Nlrv3AOpKK_1B57rzyXE;ivz2?fX0phPQ~0t
zsC@{aa{^`A(tH9<RIyEK?4d5V0(jUQ>g<y&3YiN(gsx3VeLRBq@lhL(4LBOoilND>
zwNEhz^U!>bW?0(rb5;XlFjgHJW$Tx0oG_W?j&Mq<)3BS|K~CbVQmy>lT>i`{)#QkR
z@3=)J*i6wRI+gr7y70vXz2i#OD{UDO;`-62^cU1K3VAGlRtb;;1OMIiEtl(Pv9)t5
z>DVfEc6|FFu7cp{+tc<p-i`mbn?RzakHL8@MS>xs$^mw7g7k4>Lbcfw&t6;A_21QG
z>T#RaWjY>{_RkT$)lGx|@&~)e*7d%)LhSEqE)JTDnntT{g>CAq+p}wgm=*|0O-<(7
zE%y!U@gUY=R4LJ(j8!?&c3iH%N8-1XY@nt=@+~rXfsG|}&IUnNFRQ;)W@+*`@!s~E
zL}Up1xMnS)vwrg1i=@!5&^@7?u5YOXxEJYry!6{3=UdG&fn)#6IjaV;Nw_(X<_k)#
z4(@+~+a5E_6!2PZ^U<=Sqo!WpnJfeeKCk_#&Zoz_n5q>>sFnTJQ!1;`8%A7AA^|br
z2WHXho8PcWNt}VL8UMrBWe#x9^epCHtuSt4##Yv4+pur`vHzmlwd4yY{5SyxN{R1c
zu%hDmwb^F!Df-%}9?a#13_?;i^zuq-V6eY1V4B$(8V3KO)~%jhu38Nw#-RR399>6%
z>+bJBJK~Q6p?sX8K?EsEqf5agqpTx9c$-vt0x($VgE=$y4Bw}^nc?Am$H*6(-=gwK
z7`+FE3$Z9V4tDr`{7LYPV@A!VjUl4%yvL7{>oCy^Tu7TP9()v1{of#Wt;dTjSLpHG
zNt8|7)};R$MjOhq{~}o`q=ezOuXMc{Bnqvu-HcHY%J{~UT{Gt@wObe3d9bQOK>SOD
zkmdUCcD8+Y+*g;k?>`9?S-McSe<D|X4EPu&Qq4HhNwhS&KA)2Da0h|-y$S@Y8(`|`
zHi$fZ<M+R21MtxrX7|1OGpEi}gRR$cK%<`4Yq~pIou^z<J|ER+KX<yA$oir9X@O;{
zvT5js&@)Qp@fyz18>Ox7i%XKza=O|tcBfQx80Qn!28q7yYNM=LQ-trw$gNixigI=f
zWl37LXKfh}2pQpXkAOZ)xjHlZB=aQZtei28oR(iTlaLw%T#4$}X;5qXry$3x<4#_l
z&TKPX1aJDIW)&qhTT~nRq0nM2<T8eh(K1clEr=bxzRW9@=bI)0<wtc%B<1lSZsW$m
z4r#CpJXkIQ{IAVyWQaKIakE6Bnn6sz{lmQ}LjMs?|IcBPz^8_B9iwFywJY+td>j5Z
zZ>s9-mau!tQ>8|SzWtt1lA@ub^YVT~u%1V+GJOAZ8w)t+OD;db9BuKx<x>nqfEbTg
zvBLTH#}dIoco4Q3vL0fWT>6f99k^r8JB0&Yh{7Q(=%?z__VMuGapRE&IV{YLe5vQ}
z<*YB7JY!9{TUJm}0VR97z<dJqN?=p)u|StI@ZTgNF{PxWQVdVeI^?mInDy#5gfc?@
zfa~<wiD-BRUCpfn2yrt5FRmc^zu7#U*=i14tm{;lc&X6?PruL?)N4U#9}0};yhvow
zS$J0e2D$^Utz5fO{)m~p#l`sGvjv1rp$?rAE}aGiV$Y6WbI@bl&E8kzxIaVY3kTL!
z!Ih2QC(ZfY%yrDB_Q^CTX7#Z?dSi@?Eqdng5o)##xFz4y9@I`?tm);WR8N{Cp|-W#
zn<TszNuROdhQDE}t0jLDw~bcUNVevFScxAt*#LY`4Z3am9jm_BH$pGf|1QU50=S9e
zP#uM6CHcbFljq$G_epKGszO{8QC#2Va2lZZFIOgMikYj`Gg5~yr_}Spiwz79a}c82
zHnbcV<tSTCtiEDf_n6pQ4Wh`tzJ~&a%#zE90&8d6TE@wjNm$6D>{NJUR*?X;IosuT
zy;BFwo6{n%h#oE;LUFymG>bOJ&Nx7&UV1v5n`olQ=4=0=QvGc37p;%q?_f~~3mzne
zWAosSW%~+HRcvf03E<-50-)^Y&!4A8Q*nL8u(w7s&K`QNUx>lDZH=6uAb&R@f=HtG
zwtg?!+CVqDM~Dr;hphe{EDIENXz7Clf2@_dwKADp3<&{5FOX9tBBWmo{53?FWljWP
z#vks3mgHO5L4p?-JyB<{Non9<Aw)HDi7-DMo9MPSf|apgt98ogQ-~qn%&m>sD_NNZ
zyr@yqg;5q_`S7s;FRR@F7kUz`*t}t!xKmKh$cgH5ox}Qh(IOw3SC92RXZ{}in6D@r
zzNwe%JZ_6T%}q1mw*?r_O>wNSusj55&IjmJC)3E`_6~h=Pnu?0LS;CAlUucYco-1>
z8XGinz-Q4nZ_v;4JNf~F0Ch<AD;*EJDdol+*1R5nwYYffW-2sG2#=lLCH*6+zxfv0
zIV6zn^&?kL>=i}kW%2GJiJabY#}`#dA4WVmIlnu7<OCVDIf#e?gmi=1;&-imusG4B
zR<a{PvY64>@nMvWI<hynv3ggE>)A=TtyS+u%bC+gia=cjrW$;hE~G&}psCDWTF$kf
zU`}YD4yl_mOcPg~vWMzxK0SnOf)iqmZHhgm`mu-}BV1Q<mtGPM*z=f4-t9CaN=JIg
z&kPQZ>FBo-5!N>HFr^z;mjGhkxSkif#j-lzkix1$&Sot7<5jkx&y+d4$Kia#*Dh_(
z3{wzDIU5%tiR)7<lnxIMUpi!~Q#|Po#=RwGImdGW&Li+Oz(^d__LDQ+6u&!d9Au&Q
zwp#1&&g53oDzE<5kcKWiw=A5YtmZuY)PJW=6$bz7RWv0(7vbH_^^OfOgfL9L5vJK?
zS3Nq2(pmobk?Fr|=ekIRomhE%x$eySay%FBppn~Q{`&fQ$5BNqAA1Ul3#kBfMyqvX
zWOR5r+a}kW5T9e1;E;m0Y1`JNF|`4G-9Ni=de5j(h(Ke|5~cO;QZ+4a>3Cqca|1(v
z4x^<81~dhJr$h#nM>I^#j;pCTuPwkU29ZFm1gX(QAb(K&k@H~MVQBcmu;ci6{>&k1
zV%PRHL>5NuKKPYSt}tIgI-T52;&;noiq^!WSKndMSzv6n03(L<wDRqasdl-?RSG<P
z2QDQ?AyY{#Di-UT3J_mM)|HgQKg;B@d-(zZ;l?SBe5pzg`nTVfK3WPH;_Bx1bT&Qv
z@0=GBY$jucaDA~0MS|d>hSNtEiHHeVw=|cJQ_>$1r&X;(01VZsR+kvWn~yYdAG)Wi
z7d*l~$~yDk*)O#@FKTR-6r0>g;PjFcan>xk;e7EyYxcZ&GW7Wow>LAbNUUs=A_-w*
z#8q|o6B1~x97+>l0aQ+K%lpcKF51w&Y4T;bxJ&$sGIH0ZZ2a<}@A+m+VNh~fnpM4i
z_gOt2k|%Wu4z{Z_x0~nDo0}WBpTH{@#^Bu0b-*~kbunnH7WDl4X@(hJVf{tnYUtAk
zaTqj_N0Z+lUr9tH3o-QZa&SOrXRjQguV@w1Ykr8Cu@SDukG}3KZ1ImQ1C`tWaYuTN
z7QB(QZ<MEZsdI$Ncq9`U7PfP7n<$__cL^K?L+2wwLP4a=IIu3Xi@R(mr=U^^D8lSc
zwh~PhRww!U`~U2hvHqiq9;11*IJDT@N81CE)V+Ns;w6rglau?6Pbcz`jWSd0f&wY^
zgjN7@iJUe0cq%E^7l-Z})2n_-sD>+YyHj|5?QLYV;<fApHiRVyZxzzsko^c~?KLI(
zw;wSz4Q+shgPW~1u;L{4;D4VI!>aj(LI{|}k}2q@rJJTE?hp6eQWcPZSun>64fvg;
zDV=$GETK51MY~I!<bH^U)m^KcQ`}GY@$ti%<rwo_>FVmLf2-MHjTk?|U6)bo0%$;S
znSR{))S=ha`vlf@VY=hPrj^I7ET<uI3F!G3vCwx<6+=MEM@3b@e;L~@;ry@N^XT<Q
z{%G9WTTx8Z+Ia7fu~CyJ;Q_;TaiW+Oh<lVRjUVAx>btK$jrr&$PfwWIcTNmep8j<u
zzD}gU$l`8a3_JpW%v8w;7*RVBiv15PT(|znA<hiyJNsCJvns^wiC%v)?4lIAUen$n
zS}+&HN*(mNL4t|2FVDNk5%rEgVB}5`A7)-{vFYb%CDzs=M{obDMfLY1S8#VfA964#
z_O$ln6oO(e-8j;L+SR5x`?a+-^NLW)MSO<y3ydwzu~dg_mIcbloF5sI7GK}$5z#Z@
zL@D$0khs<#^BQV6GT~lCi|_7CR88#0BL90Dn~m2$A=54st4z>hRJEhYA@{pue%JB0
zuTi93>s?~dKd+Sl8GKwC<SR+kfc9{{lzJO9>-i8Qh+M&$YiWVFIdg%cv^ldlH`D8X
z18b0F7z?Bh3nIP)MhM<81cJMfVnYsDB_lR=HiGbGCHZ+g%xlMACbbRUjHXh4oR>m<
z=>AX)qUM87MocOX9B8Vk2|wW=yizm@5J+NJHjuK?hd`sUVG}KC^cY$(=Io8@)%&QD
z^Awg0J1Q&lSii0W=XD@t_93zk^~<n(D;O^S8NQn+x+atPBrPSiWcJ<&gud~7)?Ywy
z!};(a5TulrZ+|P(O57WWNqlx<L+ZWUWr``W<X#Sq7oE8&Xpiq9{VPSqqs<-H{fzyB
zEbaoa;_9=9uU<(%F^i^mLWS#g@Jo2L7hflEw0N9O65)_A-|=LbnVFH<at*L+>Hf@!
zb5-UZ8kR!v`E<^S5l#=k%s55@QGU9cVu|jujYCmOQ`b0ypDb}n@@)4t$w8Xgt5J@3
z@_Ir;MWqGWij>?$v--{-=U|VykVc|tVv;{<Nqt6*QoyXf6ZNf-e$QyOOg;JFkQynH
zx0@=4t}O_3cL;?hwMHQ7!J9~Q><<14rB8a15&L}(-jig2y)In{VhS*b5W5%mH+LeH
zWxo@ZBQGMphxJrbiIX%sIeJ7V;Rh@S_72r0JbR2?^IEx#RW))O0E;fw_Yt{(^E8n1
z+ay3p181}G-z>`|VjRdM4yWnIkVa9lleksA7dYK4DADIc2v&?qxjenpsu{3_iT&kR
zc*s34cq9+E;{o|%l95tL^{?ctb_~w0LD%VXgMJ0n;Kk;jT945hF-XnE#$1ko7bYoh
zjXfI(g7wE!U}0g=?T=t%k$|hV{W`l37@9J%E^~U9=_WNLP21EFo3t|hA>tGbw)0~F
z_1ab1;o%>csmsIJv*_zlGGd9+*3rguqE0WrzUY0+nMk9l36Vm8F!U-hz%1{(*cC+%
z#(#HtC)~YRY|=oiNvu}{n<BrZ$?(I6iGVd41!XZi)yMO?l#Wi3naB9KjJWz#TBlc1
zt?g>+Er7946)6B-?D@q-z-OEnk_ceA+uJCxe4_@J6{;iY>Ux^^T0xN(3II7Ro`?EJ
zZvlLr4|wZ^(s79AM@kuS$%7x6EzP+#EOF{G6M9X@WM`qcS{X5oZ2cy$BYzX?<$q2g
zbzbbXn`I>zpf#N%n)+83H6ALD!HDT*TcMeHZ_OP27cidT;NZA2Ppd_B0gBoQjbH~4
zHU7=};iW_dK0cq@kNNGyVYy>N<d0hmcq=H0%(l|_T;=TQJ97Eeo8m@dSAXQOAvALH
z9S&Jg-QPULF`tU*%fTjp$OH*IV`Kv+3Z1@O*d|gwe!E)m+OE1q_=LCj4f_?{3AT-r
zmj~0OQVObN7AdH_4I?gJ)aN8^OvjJ1WJC}lXz@9Y_kWW?@2=&^-=GTPq0QKFcb~<C
z)a!KWIoL!9$8Zkv-b6$#ul7JD83}{@oxcGry#ZQfS(!_BggXW%e_E2j<>6dFtA`XO
z5-%_#%hr;Ztz06kfeSN^`M)5l#$Igv?YR~|O^{+?ASakT0=&W2q2!_~Z+>F2t5Z^M
zeyX>ox1F9a5Wy0hW*^G-PLL=7Uvh)3|3hL@$~)`<-y&EDeb}EYIjz7UfHwv$HbAEt
zRAl5*SLz;#kn<h6_zvOo9j1`&o1UJYXTSY@+f)>pfad}7#<@m00lCP<Xk$(<+|Iz7
zXdB#|!+?p=!^0JI@<7r$#5hRaca~h}^HRhCcaH$}xXvpYI`LTPm=t+c*E;d26!}%&
ziVbx)S1&R_^Q6sx_$;PKSctD9Q%K99F`V!3?*R?+CWtz+m)q8vmdWbpIT7e&!`$?s
zQKAfXH&b^nQl)&)vEX=E&Wntwn1|JIdv$i>+_9c(TXs%Ms<q3>Fe8UWM1aFXiCpDS
zpZQ$A@?5zhia`UEa6pfeHM+Id9htVEBn*(&_g4L%oS=V2<@I+;>57$eZpT;5i4M5U
zo6e-dBfSojD`s}ISuo@gcu5CLkRstoJ)L>6kAHd!{3X+hPY3&l5iW!8b)}|+=rtVG
zJCn`4s*GZTCPgn@7%Mh{*M)VPRbK?DRcqS6oCMz5(Iwhh08Um)iUoJpajjzxC9+Pn
zAqENxLpz75N1G{8gb88aTB+nEP*gR{H<<jEa?}U+;Suwz%+spSt*aJyEdv8XBo@V|
z*x0rO4+GWO9;!$(6xhFHs`n1TUX-~qg7NUnsd-)x6CW;N9wEl~@T;2o2r-NrOMuv=
zpFeF{^b%KogwY(lpjN07A6}j>3dY5&$R@F=(zA7k$cO9U*P|NiRcr_wec##ZyAWoa
z#(4VT>a{d9G6MX0SF4f-b;HN>rtO~IUc40*-*d<C*hw78*Pw+B@F5VfmNB<9(=}j{
z>-l)YPqIX+4SMQeh({F`JjbtHq|D7}AkKs_+)RxN!P^{8i@I04Ayu!$=J4&j`;}R!
zyPN_vk(-1$ey;t*6bNX=s!D#%Mx?;M@mfUg4)z*W)DiKWu#sanhm;)OR}sj@!r^IC
z%rp#4U7`eY3=uw4%1_3ZS4Zz>sM_))mAg&_I-UFv5<EXz)QZC(f2K4JhlN}oE!~aA
zKAG8koBlXi-TCM{R}W+;yU>y>m7G-)?pzKlWzeRCoU%cZPc}aH`2koye3=E1$Grqb
zxcK;>*-j(aboY8Kzuhmcsrk1tm}FvN0vZJ}G$mZE1R3!*^GH^=7o6y_Ipx9*%VGy_
zCuq<=%JQ}ZtxzNH<1#~N@Npw#q(A(m788ZVs`4{3);6it{f>(ND%qrwF(edYSy(q8
zk_y)FY*dZk2{?2C98mpRUxS>n-m*DB*@h#$ojF3TfURd%{e9xLEn?J02r)G^Emq7>
z|5cwv9`N?rs)4k>BTXYJlnGRC7;0+%ZgencY4a%pC<l`F>C<l~Rv3okbD~*=CUN9m
zBE!DC#@g;6Ny$5`R-;fD&%U1#AV!GsCeK!;;ZOJ(N~w#AmNcG&F}?4troQii_^G=9
zXUTBpJAYg4mFl=&y@K%5Q5^2AsrgO><z8h?O^x6E1sJMqP8;>s;i$ATwA{QWwOtXD
zOvDt_a!n3`z(x>HIw&YWsjWrLsj`2tgm4DbKk~ZL2nx1=j$*GV9Emd<Gn58)C1FME
zwnnxCFobO#H^YZj6a(k_p~2b8&Zx49FPxC%hU8RO7295J+0^jmaBIIfIE*t#$)@5(
z(A)~G^y;3tX3}BQ-H9^&Dg!3yR~aLomj+}6ictlCu~#~~fBg2T&Q_j|TSO#icd7^k
ziU1H(veol&Na#792ZZA2SXe<o#&>i*NWh2i`6~ksT?f(M?P@ipxsI#Om4}In-z{=i
zqQ2+7vS74xac^4^{zTpnJqt&xJ8p-}*`ND$ft(edKQ6<0@4y78z4Q<@HeLnn>vwaz
zDU!9(N5`NJRqVGK@SVVm+yB=K;9zxtf!**tSu{yAn#fnBmsE4V5d=Gn0`2l)N6M>U
z-cRMf_bTwzwqKU3mkqTiKZ$9|n7)>KDK)tn1qcj@G--5zsqjtnTgz!Qo>anmbF@fU
zLFrTsok&~Ff<c4rGyt#L3PeZmEwy^d%gZ~+sJ-gEKOL^ntN-uQ+3FQwFYN@PQRl!y
z4!UeSe@YD6KY)&P&&L`x+WTBg>wp9JqtKV0FU!qib?DE(fB#w?KCMT9ejA|A<nw{x
zQL(s6a5e}<!~kuF8px!=IUN@XTQ@&i|3E}1RMkP`AqGYS>=@=h6)3wx-}(Z;W0%%q
zrwnT*$5eE4Fg9z|JqFWKq-MJR>!ny2>}>-4t?kPb{O(WpznOn=(5P--RwLLew<fC7
zn25$&b(b1fw!YS_+bp)2yZDu#D-RDzWT9gr@Rzv%{pJy74zI>|M3%|;p7`3%rb&g_
zC_Phu;F_oTg4kofr-k8m$!n$Zknl*X<>n4a0h3^pe9>F;MAp4@Xvy$H(IAn5$lX!w
zdv;5U49&O#*l9<Sifk9}pcW%)3dEX|>sgAbb`#}mGX=}Ss?oN^e{rE`%Ev0IJm{;~
z+!Xhx&<9KV!J!eWkgUkPMaEy^U3+tAR|Y*BBD2{O(q8E7M|#c1c8w<^POS$lrd^9n
z+SyIVNqpswvc;XV4udfV_)=t^D?O$QKP-+}Vh9ecf|k;kn+@E?+m?>9bgwp-+j)0S
ztCBGCG6ddCwV6E@P~YSr-8G*Cm1pPmd-#|h)QC1ceppEGwcp>m$U;MGJeo@m4WSko
z2`C<moqqSYcKAMx0?qEwuo{Ws?qG(6g!{GdsbPF0x$M|*cw=`i5y@r*g$(6ua^3~+
z#e0{^bmq~6lx>H{3?$9&71EU#pI9=RSbvNM5x1&MH1BM{7_8_Rk?d^7-nt&GC5z=O
z@zu_%Y<mT|^i5!IC#b9lBnidx{q2bIFgv>F>ws&tw6v3uuoi8ZFy7V8MHfE0>x_2z
zF~@_&@^t-@)%Sw<$-mn8N|B`!!*=e+Mbsfk!EKvIeR$u1T>4q{z1SzRP(P2)^bJPP
zbeXY5`p2~7v>ey1w0LOEm)(EoF`?azZ?jr%)z%6vxqsI}8Y`GfI#7m$L&bgRVVs?j
z0|ZqNy}rrF&U&0`&ft7__c^76bEigtF4czX@k0AkMGVp<1*cT+&FTU3l~M(X+x{}$
zzqTYbkOxGkK56rNvRC9G`q%Ya$0*2d2B*cL95ZW)Q+I}glWN3B<K6h%aRp{L$VU^R
z+-*Ca?{%Nw&{I*V2Hi;!|2A7p)>%ovxI~MKuU^bJG(?fkE{UJXXuI0K>HM-*XrLyD
z)tg~Kbl}|VUHuuUcgJ6FuIp@H<o4XYwdD4i%1fCU0>NiN@S#QVE|+iz6PUH>&AIG$
zel5``p(wWjeY@WU_ptFpW&`>fHa&3!Qu?Xl(Um6y+O80N5)qtn%wX|_b1TK#BZg0S
zlS74s=v_FM`xaiyHS8o(cpP14-qZSMqsDSTZ4!a$x$m5LB8+)G(f-BJ<)O0d<Jr23
zr!)0(=7lcd_Dodo<wg>L>+dxoX+^>X_TAOj2g^|&29ED{#OB%>{vJe;4*Kp&TCUIJ
zJ3hQtQeNhLI&0@keCK|t8MsbwbTwxGmjAm)_xbt>H^W_zkJkRyn-24pUgzacD#DA|
zjK+3<%YZEX*HMb`$xz=)PXb537m7M8gzZIT`%C4o$X%m2Q5}`ki{<2NT@U3d4cYtw
zvKm;Zi-NVto1}UTHw6d^H&L@?KShfK=w{-6+zI+JZ2xKTh^$W;MG~17=?d7cg+Pq=
z8<zH)atBia7O=>mt+jWEF)Va}%aZZFdoS=HjkJyi?HBCFD-I*+)*@c_G%0_1?vq#S
z{JkjP#rN_rvL2NQ1ohE<{Z%#E1zr<o#XmJ0Lofa&z!&e9^5og<^Ik1vF7f2Mtk6X^
zoI022inY{QtkreKfdK#M?#bWp%hw-UY4IThxeLv<1_fF~i#FHkxfp!(t_wXaZ@pO0
zL7`X=oNIZt4fr8!KS5jt$Op#%-<RJCGtlzRg|tI$4g{2fna|=Xzh(Zo+D=ae=smA8
z!C9;yL<N(tle;?~ZoCk2=DapmwGBVF1<WC=Ty{;i^JGX1MPMARjzfYk=8xhQ#z-(6
z5Ydlus$d}Y%71>i^jiowE4+K>u6n|F_s4Z}-086;miq2#yH*>K{YAXDcLKT4nszpU
z2=MBcQ6jYyTK?`DCN*wE^E3tCFHWS*Y%=VS4Cn2g{M`~9$;+Pgnz*C$ejRks{vp02
zC>m*DaxVUpt~(0TX;4t@tQ(0d8$`CuZuxNocl|X*uF%d+gO54>+Y@YRA>XbXUeACI
zS<=hHkJe<OJ%6AoLG_zUM7pPef6wx^+||0=Mq}njp7;!f8%B-~*JJu_UkP?$6SP||
z)Wd04p#~kF_fum~_P{>@`YO61L^Dzcf1Er&?QYH!y)&Ek1C~!M4i?ktJl!I4<F@(2
z;Z(x=R%=M`rz9E9qBUlmNh*2J%*b}R#l^BKWkb=yJedF%va!S|-%iW7=KIYHN>2s}
zE*SmeH4j;}{}S4jRSE;CWtNC=_!vn7_Ir!SQxV~Jgavs)&pTo_Lz30w-`;+d*wscq
z_TV9_IIMJc5z=9Zx>dCJrjVZqQ96|{K>YJ~GwbvR)DA+8+S{%{ce`ME^%WFY2yj%L
zS?$vXHOf8|=JJTl{yLo{2$BxC<79yuGvqk)iKhePVAi=Ao8^4%P+#Gu)!h<9CgLSF
zn?r&d)(Y}JY(s86i8_5_$|MycC!_3ZqX8b4!)_s^FU%`Rx|aSiL~=XwrIvl%6R}i9
z!?Jn$ooe*`>CnRV$|fdg)`YqMjtFA^qs4E%(M>v@8%|o7I3EV`&V9E!PJnd*)>Jz7
z;AKz9#rfWJ@gJ1-dxn~9{f0hLfqWgv&8@s(1@NC1v)rD~*^_Pf6jt0w{pursjJqrH
zmGn&m#bU!NB2$`=NWjwqll_8pck=(&TYADO!geE@$vS`n1qeh=N?Ec>+&JX_0Ww{)
A7ytkO

literal 32193
zcmc$`byQW~yDq+HBsYkFfP_d$s~`=6go>1;fEz)&J2ojD0u~_ME#2LMbV|8N>4r_$
zo%p%uJHF?4&hL(K{<z~ZWGh?tT64`c=leeI^E~egR#K29#G}H4Ac*k!GpSb)gm!%W
zjf)Q6x&Ok175sI>=Gkj|2*M}5{ze-@dY(fNJ@i~kLe(XH1L^EeHady3v;1;mB9<t{
z?;~^rP12e{5)=O!<|-~6O9D;0eg~1S15*tljl7#o&pLif`TiLCGYU!N^cXo(1S>;a
zMSXA2S;t$emy+XtG53UB8<NgkXL>g_CQ6Cpbbma0v<y=wLu2s!h)^y^qWQrjyIPWE
zaTy|8|LygiSuMt`(y2|R2-8?*zhieX*BPHwR%J3naf;rO@BPbP>R-oZx+X?f%7>(~
zhP1Tk>#6BzhNL<OCDb8tr5t%z5)_PozDrQ(;CZ>qSu$W&P8%T$!*q=sT|aQDeREh?
z)%;c|6`f4_{^0kq^~id3p@f1&il_yFr2Z@^y*osK8(vXgSvfjvX*$$$z?mMbpaojN
zB%d9S38pT)dJd*aJJpH`2_ahiZVbu9^&bAxwicubud1qg{)IX6L^;@s-VcWEAHFg?
zJp4{y->5&)@^PHKon3lZ9KW@Kf&#qCMo&Vb*&m;x>pWYEA0~<WUQ!k{F)<<Pc8KZS
z*4DPMv5}q43RO96YDLhBLiA!{V(jdBJrTWe{5ak>Srh~k6aw_LDlA?P#X3BUGJNeS
z{Ct?VyQCyy5CdxEA!FLtRSHS{n(Zywi|)N~i{9_8-VKHl4G5wPVDRgl$|Sw@@$_|u
zQ6#Gd{i}N`U~KS090<%<=|p;1zE=6^Vze#+c58#j*(qvOWl7dFiM!!=ekenZ`HSi5
zPzF=JGFe93k#;cmT?i(LLey`J@hFioVP)4@l1dU_`)>5I+1lcgt9R)$u_X_$qNf(U
zB2vvMdO`CO>#nrV^jjVQ8Y0L>D1qsUEhi_3`mx3FT3#V3Dd`89ca5GGFRoZOA9-E6
zufCtK;4+o3S^HsD-Zk~yzo6y@F>$3P#|s^rSErv<UYnF_MHxo2ekavH`uY2h?zgZx
zjLK_jej7DYo6n`5z-NEs_^w#<-0Oz?Fjbh$v=!!E>C0MXE-o%cMhxiT!-w<p^XXwR
zF)??99KJoaj73^SL_~1!r;#F`;Hq))^E+Rh9<V6HXK#kn8{Q{HJ%9e3f*bx)PA)Vk
zC@3VPtEWc?%>MpTFTd6As<+KP+1c4ED=QDh*#-&yl$b8h_xru3#Ro2TV;%cnz9}~w
z-7G8BZx2cGx;!yj5PuxES${O`wh+c&Qc{w8!IjFoHBoBZ8bHXD*4*ipZl$0?Gw_sO
zT3Y(R)lZ^2B@PVrRA)Sw2<cKk4ORdJ1qCq`Z?=Rl_M)@P$zpWYdtojPOo&#qakccR
zY}a<FVRuw{sj=zpns)CI4PB4&bHOQJ9O8oWMlUgB`NI2d4g6YEyr4ZZ<jqQTd(x`(
zW&|~4bh$n;hp0$t-r3z%QGv%GdG29CO`GT>PiBKV77h?zzw2W9`uDN0j=)$YOmHEn
z-*=3?9)Hm(N$f&5q1*NN#n#$87P0XPMh=diq`L#Wcc`e=jN4Vehfm)Qh2_R~Ed1>1
zGMg+j4f-0l_%cz%HF=CSZA`q_;-{OqmF0MmUd){`t^8gy3@C}-<((D;?IN0&nP`b^
zYo!YVorU@fJg+()JiSjZ%&k*%K2Z}j*c;S=flkatG3PqJU(wlhih6%05ZmxbsB6vk
z_RO1SI$Sv(3%b^2*^VDjZ{ah-Z~G7T-Yh~g(wI-Bc%Y{CQk!C*h~cPr@87-i^`#6s
zKVFEqI_?n9RWCdx^rcN{vOn2dG$kfjn@ArF7TNbLUHa{Hb&-+8M5}li$!^&*Mk}|9
z1%-VG2&Lwia@>3&Lc3EH+<k+Vdlvx@wL}P}3Czsh{aW=PCyjZte9s!SEsS_Ze9C8y
zt-;+gT5|ih&&5mzA9EP?<%^Rc3wYss!y8XNb;}3(t@u8LAOTJ#se50u2WiAT>j;_F
zUXw!=21)Bz5rJq=(oK%nYXtbEJsMULScEEZpuCP&Tb+Y7b!)u=M%87oLO8W6$Zp*t
zq!kgkD-vI&)3%w4nOHG(Kk$xD!3JmC7$NOIXEKG<?#|^KgdA<vo3>5n>q&k>76uh&
zp+0M^4+@Zm?9E0eb=Bcx(1%Sy(7A`yvk?<-J?SV>uT3ohX;w*vrNv75hNpC-cT1>}
zO+PJo8%e3ZJ91Iaf1Sq3vph-jH0w$?F#L{mJ`se{(*pq)!4Y<~LYX;%JnD-(iBJ1|
z+;6CIyr-e-%8;`T)?);3SpB2Lt%t%e$i9Dif!`h3X>x3>*~;gO)2*_C!01@9{XpBX
za;2de(XrrEE6+T<S+;7d7u+o!Ph=x>Kl)}oQP*PK?)ajJ<FSnj9svg<M49JS83?^S
za?yb&`5MOksDKpM&3Kz^?`C1CGn(<H%3#F*qNKGtLb!&I3+8v0G!^OREEtBimslZd
zQd3h0(_|~Hkf%o5YMPpqWMrzREhl?3nu*Ko)g_)Y4IWb!mgb|mqu(C9vvTs4laP4d
zy=eBks`#qXd5*Fk)0ft_fflQN!h(p9u>Lh|zINr-@}itCFO_9w;&$D>LizUK@F4Gk
zt?*vK%`6NWqOMga%hygQXjp=VXx<KihtF46n42G}#c8AUe%MD{dt>+9iN5~z$JLj5
zpNgMmB|1;vHxp29eXqb`Ch%$DNoz2{0!Vxc#*53G8PsH|?sr77MP~;Hb6NKG_mkZB
zerDI1Q*i$B;?y3@(<po>Jid9j$sT1%%ZCX~iyfbFH=f&V*B?bq1TLQ#n&R6lB!!3L
zCJ8Jx!=f^1oNUJLXWfi<x@%a|s^BgL4b0R%nrx&Ef|7?rg%YwygM~y_9DLN5Gh!5{
zy%eb;E`=C}cbayDaR|uDwG{F=tJeasf~uXB5C}w`dOqpRoAvyzj0p;XUr=|Ol4TFt
zLT<6cEvW|6+)`WDsml$XuO_otQB)zf;^V(A$)iD1Rl=DuHe6q(7?ijus^X<ilZI~)
zoxP+=w#0|L?-&Nwbr<2Ma*0TetV-2OjI3S1tV<_s`Sy!l1t#7YHbZ+m&L1Z(=u&lV
zj+|Hy%tvs4TkvZOm%||@VW-+P8Oa_TNO_v4OQM?ec7Tyj!0m7&RXWt-@l=)V!s)>}
zjsni2gU_%;HFDNYiObo#u=I_u4Kz^N6F2zj)G;r};c#g4VaQ#G5-o9a8VmA%lGTZ=
z6i?i^sIeDL!SO5GX%ovA4TY_ob<+VcKutq-ur`t-cY1{7J=%CN=+g4%JqAvctDf?o
z0^O_gcQ!hq>rOarEUM$jZNWD?KHLWPw6(R>Fq<g%h-XEalbws}5;d%7`FJW@IZdm~
z#CDxxN2GWvsP~27Mw=0nlkm9ZuPY1q1n=jJyBm#{`lpt3UxoKk-Z=(!n55Kxtq10Y
zPO#hRf2#B&I_?KCgi)7DG$)VV(Bd?4M9+%ER%@84$#+@z%5++FIbJKz;L?0xq!S~9
zXJ?OW2`j}_mdHF$P2EF!%?Gn2x@}bx^II%w|LBi-v}}|#WL2r1|Es8|C@#T8#Bpt8
z3t6jOZsrKqRF~M%y;rvh5b58y8?QXa3f_RNkSJiQf6|>KtsKv5Iwaoufdql<PLt)D
zK2Q1)MvVcYesHhk#$)~anHG=r%a28@EG5K>$!^x>yN4fzo81-Vv5lqffAzJ~6a*Is
zq4&cfT5sOB+8(ZrCI@`toMJ-ncy7#B&X7VR0TbjL;DvY~W{5Vq#lFFpE}`i67G<56
zmxBZ9cMSK}8HR@shiH)Vt?#m^Rvg(QXmwA12rnKDnw57)apu*Tgi>%PR*K*8@!dYE
z(a6_Q^VBLpp-|1%Tz+$DZ_oO!gzXG|e7GIS?;?zckH6_u%RAxXn#x+BRc^LDUEdqa
z=XQCr%r3I)wlmw@9!i1lGf-(g!;<7)f<Ov5tPFf+l7r|41y6Fa5_92<2jp-^UkxOy
z%V=-qjOW?%+ELQxpy9HAeElb7>$=^5GU7Rpx0N^J%4g^Obv5!Shco#gor;N=ogi$)
zo#NNQM@GvO)-Xq*!MCJ@*n=i}W>wt33Ouskn1^Q$6J2_aHNu~*45^Knb=UpriFtH(
zxETttj_4EJnxf^t1nn}DPork_4=z<WafoQX3EF3lt>57@8+p_(q7p`K@mPDKb>Zz6
zI^-QTlW#h6_#TJWyHWA|4uCj?cLXS@WYD2Hl>2m){H@24pFZi8YY97VxtHRT9|5?}
zE?RHvfGW~!rOfncZQwH>t1rHSzWHo1&;PEr@2)I$-CxqU5+QsV*c`KQcDS7&>|9Xx
zCTsY4XziZJjC~8H%cbG0z_A0ei-Fi-bF#ce(E|WdX8G7>UaQ^7(#88LgF9V$O{R18
zhBILvMb6vPw>(ozxZCoqAPB_Z%g6ZZcGWsu8mqNCaz2U2b^D2KRx*A++f1Ri%l>lr
zJ<dq;BFDeCEq0vt`2`_--{N)EMCDbIN#n$D(DqYD+g)84#hUO&3W~cP@64U-FAFF1
z1=4vQ&jr$P&5X~r2CC@B;rj+Q(a9QDSn*j-@|wCm*Ta$ss<WmSxq66+)7pY{wp0(F
zGhArAJY6q%QwwmM%m+$|;WAvMo_&Zu<Shwi;n>uAamQ?RwD-bnX#X^&V4a$~2qb_x
zh<_KV!Wz7f3koyM%hS)}SjsgO%9zF9D0H~S#?FYJWc5MNSfTS0uY6kEYuERwtgGv5
zFuzcZ!X-0-fz>BugeOZu%<`9q5qE>EltFf|l7K?=T>$tVmmbt$@tVlnczcRIcm`0<
zSt&O*$XN}E*cTw{MRUDj!)^J<@HBv!4n8NQ>%Q9#_VZ0r(#)4JujbScbnZLNAHw1!
zvfQ$5|9loKjX{58*!M2K{i8KA)=8UZ-fL_}MDyFKue!%)tHW3d(V2RGvV$Xn5xyq8
zH?V!PhcTqln!w8XLat|#JT_s9Kt)-~RaiOSxC@c(ia;(>Hn`^<bY68io#-;}&Qg7X
z6%}roP(kjgCTHt2&I|q85gjbvaAN3VgO`q3ZW;2aGEoEt6tA#c(epCM5)D#PCCk(E
zX<<zF2yum<G8I?p_|tM1QNdi_{?RSRM#dtEOszLuFGe3*P9B#E|GjQ8?D~!cQy~+c
zy2o@5LYe|zV;`cY%FQ*5l0atMi}976Z7#M!Rm%1oWn^R^T_j-~M77;!z5Ts!N)1({
z&J-BxI{rAhY5tOVL=;=hL!ijY-n#STH^CcL(#rm25@~rCzla6zOz9Bx5bO|iNhxKK
z!~{b5Sab#@o!<iYs!%Rv(l99!5qKeE5xL(@DSC)YBORV)_!=6xnVP~14ZwJob;M)`
zVWjA^)@h{^q6id^l3xZZSQpK_gitCgyuk&HCR{yV3)HGG{dCp<JxmE0ri7w0RY;Sr
zT=i`enf)XCVnm43)c4jUPVeU`5HxhuNH+0y-20OE+_Eu3?9`-s8Z)8RyZ4v#=pM1y
zaq_vWtJP##*c>5!Dw@UVWuvaA@+BL+1Z>)UzZVpregOe}@q(oWvl%G<!>&m7?S?Ze
zTie~I#G<NmAG_HL>-ON4_e=&!rAXlJT-$-1hE6*n^AmX*W8YsdboJBm=}Y~Lb*S`{
z;BHPXl$B7_QP3|J`rdBL4cD<>31Ti5g?Xo@vMQwAh_JcYC8?~SjDqQ>tVqV?H@!s-
zeGMG2O3W&oyc^=A^40gVt4Ea?ex@mUGJ_w+?&k>th7<^jniq(&97;xm2B05}my$}f
zN6ork^<B1aONm!Uid{>M_^IZ2ne69cxpk*YJ!<=x*T;+B&A{sg)?_%amHBBckLN?V
zC0kPDwQYplyim>a?V)J&-x_ww8Nhr~aOr4PSj4pJyRM`JZq@Dg?I9DzJWh>U=`dH*
z7kdN8y7Sg}+aDSC>6mhP?*ui(+%_L8;A})sYMldB=H!uv{r2NE<G%QqZ{IW<ND`L;
z$kfu&5qaC>&BkOD*-#AP@%jT-Je0!w4=SNJ#5EH=UtNqX%Jl~FcvkS;&t0R}j}JtI
z<zqI(MY0&jmb;l$(R#_vcxmtU4?bWI5sD7Qet+yzI6*^AOG|<5d7{ifMM#2)!N407
z*LZOl!G)BdIy_km<?Fx7d&wx$uLv)K0wU1!i5kw`2ajcAFK#vOXnf^<UFTvybY~!$
zAxOV#%G&*OPF5&)W7Mx-^ys%x1hLQ4tsa4s-3~fAIr3<)LUxS;UW@S}RJ*tu+xJm3
zkEIynOBGFb7P<aJ(OQqQ!&_VGCqE-V*wj0(oN0YZ+)C#q;&k!i=Z!V(+#j!eSQHaE
zI5<GAFlOpr9sP7477`i?C=V)X>Zv-{{kTnE30+GUSHoFBo7tw??S}RJlK?{M$4(pL
z`?TBlquIMR1vd{3szr@|(@xD?B1mQll&sV>82jzv{L8xmL#i10zo5G;Arka!tRakT
z_d<TPq^i_9)%f^+7%Ql-o)N1y*I>*WTyqc#f3$Xh$XA!^1ch<9^(p-6uzsjI0MEf+
znKKG&b|><^Xf6mpyR{yy``U4dz*L(J^emYbG(qlcopS*x0jDfy7CTi&9e;dt0P<co
z4NP<I!f4XAuID?foZ%qvzEf)Sn`6Wzmg_<zo$4j?@MO6;ds6&*)GHMg@s#X4;+DY?
z5p+imSdTEevs-*oHTMMss;p-k%8~e^EsUo%j#AqPh~?fmXOI8^1EB$T%`nhQ{=C2#
z-95TN7tz$C;jK5<;;Qn}!r$@1Uc(ZRn|+pOr=B$l6{#Mt40d1uCQjhmK??63NfiPH
zz<O9!*2>2{z6IAf+9>VsdpIyC!7M?Bb8)x~_yRS(!2Tq0>vFB@DEJUdb2ZtyT8F!C
zr-j0}KPw=x;dCwU!Gi~***^iUsTDgEaM&P-xw<(1$<j4rju2kkSbVv7zf!>EjX>%V
z4(*@@HSORSCs-vqv}AZ!i5SNHF2=oxtBa$I&DRQJeJ%42f%{{3@#_!^uMC5N%h8pf
zVvgZoM$fB05C3{aX<f-{KN-I2(_l?rd%xs4s)c{%BFE?TYdB(+>7sb1#c{IwyC<^7
zanepBp_F5N&lU16L@6cjZTvDTmjk(MZ9A(_6Cl2f-nastYEoKSS{fR6w=f1(I=z80
zPC)Q@2Y*3>w90<Z?lIpYyHh0WRW5ATO=Ghw29q9da2!n@X^!UhqY;*Lp@sYgcoh5p
z<aE!RiSQF|c|i>`G+KkUiwEmt39F4<jaP6e=y>7J;pP+$A(g68(tVXo5m7UjnH3}1
z0yy<mS6-L9j+bfqBq%WNKcEwpJV3+tkbe}{tB|#C<lAm}hB`SMdA{-qla^cLsAXTn
zE@P9IVY5u<Gkk0~Ike96qBNr$fPcBCIL;FXjKx}L-qz!taU)aI(GVGv&16DxnBwMT
z&KuvI>Po6eW=uAk2J^wxb{4mztCQs<;`Sf&BU-pvWkNzS+(;2lJ<5Po)`$+qZEYTc
z*8RE-I^Dd{8st_SpRm{Ed37T#<U1$fYjv<TIwU9>K~GyWQE^MV;?`>gp?Idk^c$<`
zV+S`Lf<M!@sd6dg5)_K2I*F-&<}5%^GlTvbtk`ktGt1Ey@@z{*QDpkQZO4gRSU?1b
zlZcbVmq}u8oKZI46lx<Lc>?n4VV@S9-9P_vlwo~ktW`=b+3AlVUQS#eQh&91gGf^U
z4Aa=m)m(CN<3E8y)TmsM+i^)rNjq(--!7=2rs$-?y1c?#xv_lmb*{KHl;cC&_$7t3
zln9Avi-Ev%M@QLJ744pwm&k~?rd3F?E}^DcWsAJ~jq&H0m}&qQK}5dC!ZPzcFH14W
zNhl#WIQUlLhy03~x)xPemxttvKYrl|I54sVwtk9?yhTEyJm=g!XY=G|b=%piBwQkL
z`+^tZF(z#7Z>zeqw~cvuhku_>BP3=IH%DkDiBmP(iV{asYJT0C!KJ0U{?c`|7Ixam
zmZD|wTcaXCOc)p&AH|-MIjo`I_VPSMji6y*fAu^}@2-EH_DJ;MrOfvLzaWTHP~PiH
zO5@LO>0=KnV_(G!Y}9bh3{mu<JyDmK+pRl~O!7RK3rD&z;7MlNQKBwu5C>~(qj{T7
zXRXBII&&VZpz;zsU&k_`!V=q-weE@Z5o=w%Y#6T@)E8Yigdjgzn6ky3o!EHUF2<Z5
zXLa0kq2{mZM*#;(B!O>|SGgmuPLyhY+3KYz6gqc=-vK1$m(8bl{YG$or@6(FwpX)~
zrXL;Z@nGk>?q-Oq=UV`RFUF$|&i(A@mr5BstBEQ6lE3{IL9{j^iXpvHLY%czfTwf3
zXz0N%K>IK3Wx}cWUI=ODom}vg?;FD~8F)o1t*<y|!regi0#$9<RN&&hn%@T}!~`c?
z?oXPzIYCJ}i4LI}x0~K$Fv^B=l=g#qSv*M;M<(q3z`#Yn$eDS;ZOBCOGhq#bL}qhy
zCXe`5+)RmK_jJ3p=y3jU`1V+Te;2~=2Z7a;*Z55@5HZxuHP)0EJIZgL%#I)EV%x?u
zrXb&qzM;clQ1yB{aF(H*CS!9cvLiCU8!t4G+CRf(NP;%B+1Jl_*KvQ^Atbdd<BNju
z!+R#0MLDS5vy`WSQ6TahAY?yKK0u&W0Q(2-1xNjUxDw|Wvxr5DSH+sPjB+$L8iRfp
zvTv*28wPy?n`7>HS0StP=S+0m6WkO$y@9iK6JF2r9+d?+f(Vyy`SSgn8i!96%lqdP
z{_S)WWC{Dfs|>1ccZ3l=2*Z8Y9!c#5!nCPMmir@%kKu-yb(I_s9u!+nRTv(cPL{3y
z{2l5M%g1R~J#>1M#-&U8I*i>BcdVNFYmFu+FE1H4)l=M)8@|)J7&t^d?0gBo&}!{1
zB?ieuM;Sw)CX&xo1~k;9B6Y518CCP^D?rK+;tk?4(}Y&&s6F=n{M?hc0^EdRl9)$F
z14jSB+HFY{eVZWrV@k19z3K@T63I?oN)-Q}#l?%mDQiFigolOY8}LW7da4+l&^PY%
zH};kMu#dIXV;~N|e9q6W#v8arI5;@ivELJ&UqP?1zqz^*ZS2I}sAyxelXVqafULoC
zNwkKo((HaT$$e|4>j&GT!xn{_=)BHtx4R!D_b2SkP2NhA`9Q^&doHF?*T9D*okUK4
z>(~VvP%`-VlUIyGU_z;xisIIg3Gcqx)!HJHT)e{m@riYX$zZCFhbOXqa|UB5XH241
znw)}J8W+h$fPlP-#m`~zQ+_|$revJQJznK62a@cnqvP-!xGNCSBoNiZp5vlf*;Yxm
z#up`$ySuyH(cFc{g32RSSD!8xa3Pl(2j?Sd=zX~on`CxDOEM1hmJ&O?G^b_e$1mrJ
zjqYCmS_}R@AEy!1EWbda8xKv{ZYSr^CXQWJed-ms;a7F`dvVn@uxFFeZ|%YLa99#u
z=D%bmz2|x}*ROIjF_tZ8CvCLmUg2!^FcyS4YxQL=C#Y!JGh#^f#+>VB6yeJyulK1s
zN9hP*C*}D!Fv$aQFlSHP-Oo?98$=z~RD10Ew~To28}}tSRik+`8Auhaf&*aY>qgg;
zk&)2|Ib;KQOIcajvw?5d#erHZ%bp8ezIKal$@0yi5RD(KMjqT`lY>+I-5bMud!<NP
zXn=IjMzd38E=4)Ahv~!_fIaukL0jUs^5dPM#mjy%arfgL?FK&=ks52UG;Uqe+rRi6
z%D5{QEj_UPV7brgaS;(6K+Yf!Y<*xdko@uE$A|rUsmGSZ2MLXLr=KZD_y;Q7AhvA!
zjrid+^y~ZUQnOiPug}Xb!j)dL>xI_0QCD)V>I*c2Q4wbFjLN6D4B|QBE7HpD8^s--
zW@Bg{eq2%W37y3qFzug)K-wG?`~`6t@2ME2MQLgG3|urdZ!y9pmA{sSSBRTe!wNNy
zHYtXR>%gjh>FW2<ZiEg6s)6I@)_xMifU-}hb>8-zeGl%bTd|`;U*tOl3W$=`Y{*io
z&H(HX6<Vrc@rn~bcGJcP)_xWZ?nu+Vy6?dxYrxm{iFj9eaMV);%plVNPW?{<AD?q-
zak4)+SdM~g4)so))=5Y-h=w5|!2hK+jMkQ#gxeHZb0AxLs8i=+>tbLaCDqpJ#8{_z
zgpB^Shnqr`c9~2}Oq?`v7G7Nn$FBFZvDnDsoPV59O^?pw_kob`Qp2|>a}ViZ_Autv
zt5;SPMjVhgy$Qc%GIt`y?H_`69GN1L+1T0QD`;;)Ja7FGru)j^WN39y(*pWPWq=T^
zF(r^b*QGfJ#UEgGOQtp>ME0Hq65YuiCRQG~tn>dVM##HK^$@RPkwlX3=>`9Uo3yQ(
z=bUKuQpBc01ijOk`PJ}U#2be%E>~P((a+dlUbLEdN-%Q7^FCl<$)XaSbFzG=c!}m$
z^(ydGt4%Ep3oHg~>F^cTL|p*5fm9bBp2(x;Q#NQ+T7U3Mn<$Wxi_2;#{W;*@UY9Nb
z=9c-(D#%RCm0DD_2+xYiC7tds1#uq+i%J=(3GfQ~3_gNPvOQnoXgUIeXw9NOCMLo?
z{d2l(&z#M}(=iwd+^`WhC)Lu3aT!!aLudxGva&kb+IkX2rz{>v)0|sP*9i&=y3V!u
zV_{)|HSkPPk@&`qm<R7<0OGW>w>R8=8a4Ik^gJVwTc^*-&Byd~REE&g@|Je}R~&D^
zawc_kAK{@xG*#~9hTXZyiMDu=%!x)E=(oqItFRi6@B>NQ`?8-HPk{=YDC&m(L|<Ru
z$*F>uuBqI74CE6u2!xi`uU~_cJ&bt?P%a4amT59!=@g)9&nYTW${QsD`d~U=l;Qo5
zp;ZUZhT2F9a-Hf!xK*MCnW-eRlwWU%84cVj^Lfr|uL9R=>6f2rb2Fu1{38<%_J4Q*
z<ZiAN+*f2~VbLflRnOxTMJaN{XH;I`NGWd&`EI!tHqtF@QwHodqI>3=>0MYyWtv_o
zdL9SEUS$@QD5Vw=v<~p*OO+C|vMOtaFKAY?v=uP=tC?ie&$5!D9>>WSj$wF%bT?r!
z^aGF=7kk~@*uHu|S^*Sa@zmy6p>7h7Ud!2b<4m%IZ}=Ty^e0wUR&aPCyf8~HKTTel
zjKS~louMVktekywH-(*DJseuOX@JOcRSk^mkzy`iH6Fld0v~;Gb&}&ZGDj`0)9+s>
zhL%J;fAj5*O-;=s;<MN9b1QO`O1<*D-M^Zs2jd-Ih2@)-Ccm^B<b9KbuT2NWiS64w
zX9IZ;ukun74E-1x83DDY`@yKjjT<*)(v%CwSXfy2Eyj76n3&kv<?v2HcziA#*Sqxj
zmxIuqRvsoV9xLY7MbBNcEwtL3P;ALYybNIgcH9#(t3%0r_4QpB197o=fxbDLXVDs@
zwY@tNfo+TPjMTy^fgwb)<#BwJ1c$H8e{@Oajh1k@b+@Rey)GPc8^(W`Czm$(l}ud*
zO%+V7yI7K>|G|m--s3zeJ;Cq=fr1wRHEZRKc1?vnN$;fe(9zM6k->%x3=G=ZkyUl~
z9z8038b~ypDG&S=U?thw+JeOd*&J?6d`^io)Bb4QaMt6G;gl6{cbM9bE-3Vl7IMy|
z8kVG3IfwsB=N@*Tfb54`6D8*UjDAYJO7~X2eE!S@ulp7iMNUFOa_d$kz}6NMB}=_=
z*1xN4n|-iA#sIk-DCmIPhy$BMLPBEk>Gb7*QX6}SGbO6A%;a*r6E|T4-&gg{(Xf%w
zQEsQ{oz{*J?01&VH~@i>T;b%HzdA{w&KuSGc}c7M+o&8#r|O^P;vZnu(%Kpt60&mP
zRwdd9G4$yXpKAS>@ht5M<57i=ukq!z4FFGoCb&e15ZMXWy#7Q>3kSIz_)(LC{7P4H
z*ulgk``R<YR;~Bq>9fdJaAj2Wqu&h)(CijcG3u)<Zpu{3=j1|RseQ-ZAB7Sy_gzby
zRSiF-H%i<~H8~I3)^|NFNo=c+13nd;1){@Sf!2+&69iZI-G4<=xa<Spra4XS>~zP$
zlf+dc#=ao*_xA?`d@bU@tX%u;l|YM8JS6$sS0<?0kb*-BH)YZXWRs7dK0P>@u2tan
zaQhXcr(I!@BZKcS?P#^E4kd4|^;eoOIpdH>W}otaSfE~@bJE=8mg&5GKJoEP>;g5M
z+Ys*}GAX|v+~2`(t}x?~L@EjnY^L&bz^#7Tez~t&r@*X|K1Bau`4lF-89|-n<39M-
zfru7uG@%!p!H<k|C&vNUJOCMd<ieOgF`##Lv|R{i`~tE!s`YTO@s69pu;ROp@cbe3
ziFb_;Vs1ZC?vrTUCDQa$x<j@Tt<i9{ReMJ3g&y1CJypK3$)Cp`Ct2&XIXzQz0hqSJ
zw)W>=*v^94c<hBM(Y&)X=){CFsjS0bk}vR&lPf1I=#lf^=d6sye1W3md2y<?$tO~H
zp~txEdLkqLThvYCET2U3nKBw33ZSu_Q~u@FCJ7tXdatWXzLQ_S8{VKsyYJ@N5>gQf
z$UX)r3-35t5jbwVmN$WVQ_(LnT5Z2f<)$mNoVwJB2H7AVDd7_46(33CrsRTO#qo~f
z)K+&S8Jm*QQ9yb%U`Xs#Sgp%#XnwgKpxUh+4de*kW_VI?h9)mo^>%$@#hDLGveH{Z
z*ASGMOCz@QZLTwKmA4inz!;q<QEE?rb8S}8K}}OBkLXm=7NoBe3V6G9^EgJ6Gxg^m
zh3c%ohwubkDEjI7>%$RS$cB)TV}1zaap=9oHut>Ky5}@O>WbyePm0iemAiiE*x2=r
zRmiF9F~t1#Fh^BzTthEMln}zVS!wT^9%V@C3+qo<t)R3IN?jS7j^<M8@WO#8MI`BH
z;U^7334Jq07e7kAlitd{XgLjoz3Ru8yLJ!Z-8yv2B+mKxPtk2U5tmQAW->6fdHA_Q
zC?Q<$+mN<F%BYzrZG`TOwZwq+G?cm$X1!oWMHqP~l{X3m=)`5ib>J_fB*123YXZmb
zMbyJ5*qdy1M7e>LN-Rk+rR5Yjx#HAPi)etgN+ZL_H1^jV-++L_0&(vg2I@dZIn8#)
zKQx5i4;^e3PMBQrQWGt-gR0i%jWMeg29sw<iK<LSUNY_?z-kalsQKlj3zN(i;#pyD
zH8SEce)vB1_r-PxA3BL78eZ#qX3!Xz>D!P<Lhb&iF!R4hkiW$C0C^<!BHNdD^(<)g
zE|q-Zw3?199;nJmy?h(AE3H}siD$}8zl^P`l|9811Y9E!YR?-3KCrPF5NBa{*KUiZ
z%A^^L0YBL5OSr5T2%MW!m0qW-S;JY1bV3d*$GZ!{P6Xeu_!B)2CxV8vd<1(*!}I`4
zMbm-YWBMJQbYpuVPRjKrsd;iBE<-wvSHy~GEG^I5roF{(U6(n3KJ#BFa74fy1K}ty
zHQ`J9Vs~_oxTp5OQFt`Mk?y5C;1Q}8<yP#Czdee15_Gn8=KYas<sN_y9W%*SK$!zr
zEcV+sm-FLY;1^n6UM_^!BS0_?aZ+%)3soE1?AJMUC_=NF<B`@Q>C83$VDs%cuNZ*>
zKSK0y8~`W}F~C?Sv*v<ufltn%RAygdGFf(ib#kFS;mk*H%e$}au%MQ0M%Vsc0$V3V
zxe$M%uQYCsPou<4p>E^PFD%A}hB3WMfa<CT72iTf1f3|N3|xqumewTrU%*a`SRqYf
z3k($XD@})v(7kJ&b#<6w4CzFd*caCfERCQ(FR*PbfM0eySmX0Ncae5uE^L!(?8~t`
zb0Uohv>q6kBH6w^sbAweR4VIZi*Y3c4GBq#^1`h^5GxP=Jb-LwT)C%lLSe56!3<vu
zS0J2((Fndu8~m=@=-K&^A;c+9-KMtkJ^is(+59`(QET8yJf=ngTh|$=Q!6c}miAAW
z6%$2_dMKr9)%L&HPk`L;M7e5>U-FsZIUzy<258)|5<@v?$3^7+{rhf5TWTfm_~c+<
zd!m6oA3Bh%XL$r->YqQ>1Bx*Z4FM+}&8@pNnAQQJE5;VyX;(2z_$${jY4rsKg`CX>
zdcTi7=c0>Mz`>=Gr68XmbdcaYeNBBwvTKlSs>|Pn6CbTxK|9O`<b*#aw}>QhfA-8I
zM=u4Mjujw*A_;UlU@7~$p%$i3yvgDeM@hD^8l^hrA%V+qaw6>T$e{BhU?;dhMj1$!
z09ikbI?uq<Y2LvEpkD8@@P@l8ve7TDQ{J3)87I8%7_4MK^|GsfAFp>ia@(%g_}M4Y
z&#hI~X;mpJE35d8M`F^XbYdE@&KcNBLP5cAI+PCbt8^%NF~_w{0VsLrzf=ZTpO$*@
z@4EOE;aX1}mJb2T=c`!Vr~L53BXD(}mY5WtKab%vSAY5PCDYlzL+#~1f1XkETTjQw
z$Jg1+wfOk>;8So}fPK{DjaF*Z%f`VWV!za5>Gg4qzFm5wcm}P>V;SiZNGAC#|NI4i
zl;P3#44^Yz#i@Jy`l2#vp1pk81OD+CUh5=ZBlPGWPbJ73$XaJ^F0P8H&AGX`(bY4{
zg~P3TBK`bUYv#9g+3x@@hH7?uAnd2X8o1ej)s$!cQf3I;f&5bwMy0C<2xwr{0pQum
z@m#e$UC)#Dw>%qDm4$!{Nl8husubqr4AY}DHHN;4dwD7-gb$j;^IOB8o0CbCNc#Bt
z=I!pXNh~g&UWyPwJxw@1PbL4DMZxqR#z-lD&65D$#Sg!;U|-odq)$rA<oIISTkw47
zt>d64xDTkp%s9VqB(Xm^Ipcny*W&ki@KF09B>_<&29QIun%{nWoSB(vHC{BASJ?34
z*)txhFr!jI5fS&T-_>B0i_N|`Dz9Eavjw$VB@-4@VLuk_BFi!dO;|(N>ZQqJlPbJf
z@$%<A<WrwC{J0%pTZ=5YlC+~)cB);p{9*_kv4-3n-6bOZ-SGEKY#gmTB@|J5loBw5
zLT_NKwFPd|-1u*Kx#xtZpE2yc+S=O<4Giwxy9XfX*O!LzF)<Th6@neW4X+ar6Y~V5
z;?JKyLBs<D7L&35m(Nx@vnG$@<h0UQUz{=>HDXjs={BSbUa-LCYtrbr&AVmN1{ZdB
zM7ZC@?P&D;oMLA1gQZH9!V5JH@hAxb5O7Us=fwR5*{}73)5iFcB;e$Ql*rUFjM$T+
z>MnQqR~k>2s8=NbECpme2rrI6jhL}mPESt<`wob_(lRnK($a=TMj(DU&bI{vfpg1w
z#sl~+Del}61$&4|E~<#b{9z2O>mQQaVlUKe+?)Z6D1Uc`rzBv8iowr}GUo@SIvfrM
z`yKqrlM1|-5wYTUT@$G1sTKTK#a~A0dZCjox^LXqeU`1$`{ExQ`&D2A6w`YZIiqH;
zqauron1GX3ijJII9YFXb{BGNLQd!`q(<2X+K4e5N;dKMa*OolA{VMv}ZOOL3?Lwd1
z$L9~0G1>*4)bC3qL|?Ye;WyehDh5*Lmt32N6@PYfRI;|J7KjW_(*%}&-xgZFJ5ct~
zF0}m}04#UF{rbcx8zqv+Yx>I0uH@l`GEJoy1>|y5Sn?UGi%9uzYOr+^8zq1f@~Bpa
z3&dxq+6*}s3=(_hFEw_zO(wsL;BZ$<;*Q?D3?KA5oG>mfE(XOpKYyj;`mgn0`6^}F
zAu#FQOHulKF4BAZy&Paz#wXM$u!{xl{sax0d<YH}blv*{GQii3$h+UZUk!~F1o19x
z-OgsA1bLnc#9U=E>c@4(Kj2bQQnN@4=8)KnS`cb;@=Wk@H>c@AMY8Ms;j$TrZey%)
zrt=dcA5||%XMqH_Q%x!jIt-FB_X|P^7MfdK{ggx(tOxqz__wF58<7>3asA72BCdPD
z(nq!#wE>_30ByzWT;xHpT<ct(dL1*fHQ*J?$g?stPx!c|*Jto|Xq&6F*&6<L{glc_
zL-VZCTU~Vnx#o54IY4Q2k#d=51yb}1yWo@IL|PYt2v}~!h}S4#bOQxcw<$DRzK@%7
z`g~CXIY7}rEJ*P_RYY`0pBjuSV=vFD3d0R*8q2JQQK?^t@07_o>j7XQDtIGo>S!I&
z0cx%i=WRHXFf+#59@FW(Yci{I;dD8Y0x%_fAcuavuL5AnwTG2f6hR}3aQK`3<;Z`h
zTtKs7axh8TX)ps9i`zfPfnPLU$k7aV(||Xz4?rCtl!3{zqKmXP>N7z1rm*&x#vQf3
z#1`$ag}?x60X$*>d)d(N{sjANo7J-?WFEzwC!Ih3IlKHN5t88cLxYup6kw#H<~31L
zS65&7;V2{;obgT20}m1rwRNCnP@Mty9eAw7KxzkZ8TbI5wx=(q-L~&f=1;ZVnIsnB
z<~8l&H;4bddaWhvft!*-P8+xoNVDMUh0_fv88yFf9Y%ADXsVGv;sx-aY54ufI?xah
zn;H47EHE{<^bcY&R~_w2wYtcYKTis4c6$b|8dF^_5;)^I$3C^Nq0|Tj)V_^;XD;T@
z-qXYjAJS%%z3>b=cGs7}$dw1|K24UX7MZj%nRR91Vi5;Bu1}q%dJej0s;O3r5!#bO
zJeVY9z^99&$3=K@xS#6L->Yd#K8uVc{Ju&NsUqt;4T7pj)Md^K>l{qmTEfOVn2<(<
zu)CG>qOh9??ZkfPTb?SURpRoZC16j)*_G5@@U9;q3=6&Cg@w!hivoSi`}dWvJ+J#G
ze=OC5TkL3Ll~p5SGfXGn@^oHT_Z3$KdJ$I&I$R8_<Iw{Je@h8RpBkUUWxhr_bjbP$
zp4?-?EQyOnqlP8=#rCsih;)A6RD+7;5f*^EcxqVnOV5NL7a_-66W%e>w;BAtW+o;Z
z_a&t4+eybBI;|@IoqZuGQHesqGCIf|tlGU9uxVpo_$2mqXK*LrNQ%7qVphuy-mA>M
zw=&_|=)($1aC{QN|9F#O1>E=E9tcuAg8#EZ?AKNjiq+BvT@VK;1@O7FRP$PP@cHFz
zh7~Y6q};CSFPJ3jlN`MjfSi<=cve7YTKu2BlEt9EUp@gyWMcCWIP-7ZWPU~s<WnHX
zFWLwIBNlLxsm_70=x31CN#X(QsbCQT7ecR*JxJ1Jwg3Wxlmy@g2P<op&77qF8R+Qv
zoN!knDIp=@+c#VYXoa9(W9SC}>E#v^q|p4rLTO0}+ix7u*Rnx&yQX-;0#Q#30q0e(
z(U->+W}|D-x}Fax2|(??4$1`ovk!NKAG5Qw^YS_X_RetfkAOmO0UHA^uWGQ7qfi2A
zufAS!5d#Cm3*I~>J;u9t&u3zmU71c{He@$InKx@A05z>TP!V98z5{^?SR$@fc|SkQ
zv_Z`|G2rrDa!rJNPD;`pidD}uq|E#i@m>eG>pK=mFeRpqT3cFx%fj#qahm|U7RhBD
z*L4e|$}TvN&y+&jNm>03UjzBPYwDk(Ex1K1Up=Xx#k%wb8R(D+lTpcN8RE?!NGfhh
zeKyeXZ@K*gt*F-IxBpLehi45(&Sc%&1v4A-SX%jO<(GNgKb95gi{2(Ti;6<ZZ?pDF
zKDkAndmN^gs`LKYrw6AX4}-vv3Gy(A@qhC$0Ic7ESCE?DN~h?pgsQ5lgM&ki$JZ5X
zNn8eq-f^EBzRHNWZua@GamDa`U_jo*#RafV9<+K&CM(QEKibg2U!W~PzZkP?fbd^A
zGk+i_p~;8;CAei>z&-!}<n=*0@_h)Xh5x;7_k&5m{#cSjpaR<H1ye1Bg`Qmm8nM_w
zNnOGrr~H+4Ml;ILG&ZVIevwTn!{`RX-XX)2(ZS?p*M>jE^tgT>L8;!%L(sqf;(U^I
z@H;*33)LrN(i3rj77WG<O3juc`D*ed)6^d?Ps&;%=I+R+mRDp%TF}Tu^3jPqQ$&XN
zLPlk$j~*rHhEf{nzPv>ttsIQUs;bGv20>y;+)8*L5dQCx0k4begd1QU!Eih+^H8L&
zXRzd*{yUPmIW*X!|2JRNHhTGIbV>}Kp8)r*2Y|JFGf8IGJ!C1lw$(aG9*3_jk5cij
z<Hm}YyV%UIqWGPgH-}4&`$0L1;msNXc40(VnC0)P-9JCS06Vgf;~FruBq7#v_7q2*
zj@Z>@$36gBNNX4PJwdz5LqMAXMhZBZfCWzgWG#@^)YYR{LqH?|h8j>Kfj5sq1s`!o
zSbZtySHAYxAxk!NJqH*4eF?%HVbocJCU+Sb0h>k)-4S-uX>bS53@=ZRc!4{nN~hp+
z0=sLjg7UrpDs#-)eAlT4Hfzvq<O1+0Xqggl*`bHVEh<2E|GE(XlPnJcZE!UYGVjpP
z5I~>|_1OE#k_8+-Mm^sE*Bd!q{W)umO!~)vZAb&Z8wM@_xPII9L4yvJpq+u%-!A~y
zwrjV=m}=ew;%gL`mlefM+SU)q$;m;-6A9$BIT;xnD`Yx!FG1l5)Oju<w9u^UrIF&O
z1Q1M&fwf^{;wA(P|D~m+X{o6YpyIK?xj4Zt?C%F^$*04fo}Ll2@-S-tC{GdFdB5RR
z1IkQ*znSs}dF65SUaL<l4Do_qKYHA~$M_II>sK0*PXAO^&{C#U3eme^;WB7yN(OD*
zf}j9BTn2!drLfXH)pI7yzM6xsJlP0Z`Df37X4t$kn1(j{NtV_XIM=}~CZnJ@>ST-%
zIol`&oFpC|UM!FqK;Z8dDF=p-+=h3AfYJWX9dUp?rt?4zw)o@oQC)YKmDw)*2#XNE
zboq(j2)u$ofQfmTnIRV)Ws%OBySIG=3_-+nV!taar^d%MuY_~jC`ce&2Ec_Ke%U6I
zevZaw9p?x0ZKiiCO=azZqj7xy+79+-)-aZoUItC}E4@71dmTL2h^Wir9H>}b762P}
z_9o`~{(I<;I?uZnBEVYS{*^xk5&*a%j=-U(nXg_s1zNAP_F8~rS&q2J$o{9y-_ST}
zmcLUEl$46u-5WrGJ^!QMSATW%JBWnwd?Q^w@4!W*gl;e}Fi@L=d{uniz<`Hx!>P8;
z((Jm=&RTtZ%JiQc0rmhWu>Q&pr2gev`GD4&2Aza|dnI50L#da^q<yNS{u|3dOXP<S
z1W;#ZXHaP9m&bb|+!?v7Aw!jGz~>HpmkX8+W5BJVku@!xVSg4uS7i+f>&gi#5H}(~
zuKoJeVRyd0tgH-J-av;PSTSIMgi$`wzZNfydb|4i8rBPHp;>_Q6NDUVdXG{~C)sf!
z`jJuXTcBKU`c&h}9)0zcAJ}^?kDk8U3O9}NJXsqTEIji<i?I7IEq0oV0ahI3p&@}*
zFNlf8w9&j;RlzP8+nRO*dsLZITSiK1#K{>PxbFs_ilMYY);GPu;71WYOt8y;W}rX^
z(Au#LdLJCT%N8gR@!!2ZWdDmOT6z;NW_^i)n;ggMaaTt8G!G!*%2QkNtN{GpI~dai
zwd;d-KQOVdB4<*1)cAoFxf|4GaQ9tlu+QL!C&*kp?n1!pmyGui9Iw-l8<l$OpPDny
zYuC7VBKHF`6_el(V+uNL8@@#?>Ry5vX7rbmYiW^8>OcBHZT*3oUm<H4l)Bd~3!qK_
z`O>J=6|8Ko=+Mknuq{1znsFBS$tfs+9W!>>RkHK^WPiJAA$-k)h5+FQW0R|0jr*KD
zhJJn)PCK-~1r;q$eMb5P1vK2ag?RxiIT191ps8~YNO*vKJOxZrVdEtNRKv1iRBDK4
zSs<TZ*VP|xKLSO##@r7O5zqINu7JzH#LjMS)BT^5ghp8CkfNMwk%)sADql#TK5}w?
z%h3=$t9w~6yU<_<#L*TZ1lm$B&rcBbVjx7L$|UO%2k72AJ2r)5+H*fZlm`S}3~UIv
zPNq%-C{X}uMlF20;tZ0402y$)YW*1D%5V^43U<<war2W^R2DDwCFpg{Qb>sX{P`iU
zwho=C3OMN^-dR{|0L~&UEe%YF5D2Vho|JWK0HRu%0_y?Yrtr>5bL-!uT!FO;u8j+5
zf~U5ibpXIf8bP~sc;OYe<L6#8UKi`S*1(VJjfMe0S(%0I!N_2q)p=58)uc_Vetae+
zdTx<`ylwAG6`ic5nhD^-^z`%uCk2?b%<}^A<)>=B6JX{+JHqFVq0a*)+=YGD$-czP
z9f*ofwT9MFh7&_+E+tJZ09k>`LIU;`^vTMK0Q{B->~wirnh6RT85irS18yRTVD}jt
z8Hu3=o`SrV?Ii5z@Sy=!?Cl^`*rpdvI%5qE=^%u$(xeh>6e*CpECBkX)l(P|IYnp=
zQo7d+eu3G**nPk*fjv?=Lli#vf5K5%?$>%kRw82UQzG&@sDj`*`mZgeTjQ`QVB;lj
zw<D}f0yu%3%`H&SC*0hT@LKALNoNj2OLn#Sxcq1LFY`Gn=!goqe}%N~7q#}RIE*Ps
zs7S1Z3BP@Nts16Frn1tCd+HR90f34$QUMwELr@SaD=VO}P0c{(fGwbQwDJ!DQNU4e
z3>x@G&A=PxtD<=<v%a1^-yTqF9=3zNhwT;q{XyBk%0c-qq8>#3W8&tNSsRgQdUqgg
zR?aM?KmTG?-47OqTvSetJdmXz9{n~DRAok-xM7lOB(8g?86#O&&j8ew2F7<W*BS$R
zpQx!Rw#@VHvxD_Yz19E_QGv1y_Cr0$CQRLO&xN0bzI0p}m@Dp}#mB?56G{N4`#R7H
zzUrTyZkP)y@LwPu|M~N$_+3YhP60Wm*7rBHhiMUFZ1u26-rIIe=$;?0J%+J>WV-|n
z9_{Vz(34@*dSPP&$XtM*eaIeg&3pJ$z>hD_wlhrc1I`|f=nuZh5>ELbRV_~)Dl+)_
z`5xBl{C{`>z~=xx#|@?r)b!8*aAt$;<ODDmYE6|tEr1ckgU@3r$K*^AqO|>`{bF&`
zY_uEUs+9}nV{mrY5T^8&Xpj=m!5JZMV5UP_xZZ~Z?3X^25qw8^vCn`hri(lT<SY>p
zQGtzPZd~*A5!v#p12|7~;|F&(_Y8HNw@X}S>&+>8xGz+1R?NR!H0q6ATwcDWq5vTg
z9Tk1L(@G3_nUswxfuG=c$3h@6T|gswDqFqr&+j9Eh{7SJ#q>s@99AdrG_q<yR~k^w
zu%ZmXO^E*bRm#c^S9cn8WwVFq+t}EEyU_dX!S(2CL4rw7+k5C11<y1C=4<m`-&dz9
ztwHzxY_o6P<dsm(4sHRDnheD0=g0nX4o}6K{pIKb7pea?_Ok_!f$cGHHL^RLJL)S7
zVo|yspRH8?{yAtn+;DOOMrFSnH@ig4%e6A5yhwnalfwL89wP8Uyd0=951nDx6{&{&
ze*?DCH><0wfgOnm507o)k*FxrsMM4<7GTh;!cc;z%48k3dxs%<l&p?{#UHL~j3d3l
z>mq}Uje|kzn$iH(M%RVgPdFO%_I1UJqv5ls!FWX++N0LrulP%HwdFDEdyR(I4?qv-
zs>S2DGG&+_P(j4@IvGKM#xY?JR3u=J0se^CLRm>*)BxgO*A#n5s=<{V2E^dEATV*|
zIl#+#ra_L6|6W~_=@lI5pC9LZIaU{o`=4vu1^yZL;YP2khiUtdjDCMZ2_|PfDbmV6
zI8pFe%zTo)?5}2phZ)zKIe#B!C8*)%p>*>d3ZS*yw{Krd=*lIq=+?P$;QR#PTrpl?
zcOgRj*Awuxg>28<V8-h*^S{!)2LNzkd#{fbUdwqHIP+SC;2wPi<BH7$a9Vxuo%{7y
z9jXfK+kUY#3qlllki{BjogR=6Rf&ATw)=nZ8lJOBr21i@>T?G{E8w{ip&ZxGdB6oq
zK`E3xXku{f&0%7LJ@IGq!&K6XGyxBc08fWF{sC&q;j%hQHjVdee1N{HBu2{(#}@71
zXNtg*p!fSaeHTAK&m~b~eoQ9V2ulJyF9YD!X%6|}zttpDNU~HMuRJ@b3K95SWGUff
zQK>cq=*fN9_5I82t{fM)xq*;C&<dX2;qL<nPuyS)0iR20VN*uCX6N6((gbqBKO=)6
zAS6s{GvR^Zpv<*1qC2MoA%CweAh7<Ad7u9&PxSxy``@Tsh!dA7IJFEar$EYQdHIdr
z5;Usll$nrpmxNLAA<lO@ZZPl-+ysPcpxJzon5d{IGqVvOU4hKP!Xi6a1t@Re=<bQ-
z%PT6{0d1F+S6pqIH5W4Asyxi@u7JAe=FOXE^gv$|7S^5|2g@8Z+5w3HB*>keohj3#
z60-GQUYWYJq?i%`TX+3F){L|bg#D+T_kfpoWYZ}tBjW)9qbNQEP%-fDy91A`ES-ny
zo*iIE?(y=XGOd%x&g=@khi`wjfA6p{Zw%Hns7A=z*F3Uc#1i7+eXf*;l38D*4R%f~
z<>PB3ubEB22>APBwF8Er)G?(9o%S5<NiSx7^IIDM8fex*h!ESiY|!E{C>Qz@gpazp
z8%11p-oAK&2fd7Y+`H*C%xhV$B?s@brI8<gh0ovWw9%Emy}$K)Dcln4lJLl!mkJO{
zXGin`0^?Sd-P7L{vh0Nt_9|8t_3|WATClOO1Zjw_X;Hv=n(<Oa8G;HQJXJ(hvldzK
z3yiG2<omT{hmYic?{;9v0((A9HX`aVP`nrf1*<`s&J~^a=cPp`IcGYKQEBy`MUL)Q
zWQ+SJY_9kXD$WaY7$XPf^Ai>guHVJ$^(ke7b*ca2YF%mx7+c{JJ{|&lV7MxL{V&P`
zOvK;S4onW)`I5Ij{)_A|)BhlL|Nbgy6Eih61r8_xqyZmr&3!*K=mZ7+WfyxPa6U7E
zMuR2oAm%SXCxT3fwxiR!L1>``;a_nWRc1Q;!t(=o0!c&!jyD!QdFL_L7OgW+R`{dk
zzJW3OlYmbTD?Y#^(d0$>U*4D_0hk_27X17VH-62|5<`i!KcS}cqkF|<V}C6X1xjiI
z`-2Z8zhpM;$IcLs{_9N)?<{jtsHnq}kt>FKQxdpLGAfdfepHf3wgh#goh(#NZs7w3
z=Y|<RBAHe3=pHmc^PB~4{a3~|OIq`xZ5nGRHQ{|f+?;rv_Lp!jSM-<Q9xBT+C;n)a
z_-gjCq#q2AgAEvzcbwL^%pBWZ^%%r|zt*;uI8}@XLaD;CR5YMjqDwqG#qL0u<a_Ca
z>+xGXzuWx^%ZPd1;V(ID_QlMX)&XIC)(5V%s<5IK_-x9b*Gmx3u1@2d#5Oqji*dKU
z^2cQ$OCF&?=_?ggTH<(LS#OJBq*m2nLrJE>rC|j@sou|uLkZ<L#`Wd(3-2?>N~y>{
z5TzP^C#Ibn99$U8{OevJvr}_GeS1Ji$TMY3g9*6NlBf7D82pKjP4t@`kw%-Cdi%qw
z_<FbHzWS!Dtvx3oS^eR2tRBGdt~V!6Hf83I`&l!;sR@(p6kc;liN3&STgpfCo~i7Z
zDh6V{^T@6wF;rnC>YW*rcl+g`nso=b4!F`0emxEBPmuWy=p1chgwbzHdh|Xjy#ncD
zusw;v8uGKM?T>-0Fgb`VrUm4u0P4j&GA5P8CZ^rk-fjfev50_0xSdT9V}&5ZN;Dgf
z@EZ{A7?l(U-~{Np+vCMgGhQWtX9>kwH|&~dS}m_k-ee8wt?Jb2Ke)aC)ifPW6!4(|
zO@A6dIbUKi@c5NhxoEQq&S$UcPxW8Yz*7HLRqq{-)gS(kA7n&E5sE}1aU0pPWkfa^
z$=-X9?5v6?J0vTTk?g&bkiGZL9%b+Eb-zEq-*@<<KYDQP`<(L{*Y#Yl$CE4QyULbU
zVI!&#Jzma0-hce)4d4@m7Hy6c-sRyLbT-sRUD#dvRRmjtTlE>+_-DHXmFeHO5)1dc
z#XLykOc`XIPWEubNuWp=PXab$cGE|FB6j3or;(9xq*jQbiT1vv5ln4V$ENKN=^InZ
z5W^2Sj2d6ntN7lIy>|Uu0q{Lq7+O$wY~|p<A*F(r!^kv)=#@c~B~~uZ^ogjdf>k)l
z_(ws%IGbr38sXkYTlV54(8Dimx^53zl~QqT*IyE!;qM~td`L}HZtlNy<Rw%n)E$<-
zoc1F$eR;|ZzR0Jg7oQ4LL8)qxhC579TVH<x)A6Gyw5my9k_uiB-=~p%ek|~(oq>^s
z)M&2e7RW9JgS?D}6fO&Jj9i_*h8zmFO+^5sV)d(*b1f^V){7lR#=WriY}4Ur(CK+@
z#%l91EcnwdMaBK@vi+Z54ag0y1C0mRZ<j$h0IyaE(kDP1?8S|GSLD8S`N|_xeN(nC
zUg%Nqk6)AE3`#`kkBTh&vl0_r=T(aQ{4<^kBFgKEKW~{{m2lO&ZZ-*MIjer%^SDK*
zZJ=57;6^DGyRNk*3rj8O&UYm{a@#L%IKWU1e7@0J4+%uzQ(btWm(g3ZKV}X67L1io
z(%jF!8dJb!WpUh<<fZZGQs%7Qox?IF^5WlTxlBxdupTnd(alb}Pw$x+(8z4ADJW8B
zg}F|&8Ak*uP7wkBA(V;VK^8wj|MbaL9CILEK!!6PGIOwzHQ`hKPy+9WNYiUKno(gD
zU@O4pF4F<TLIpWDGR-+Rj9R}<6gJ0*7x26N8d?;we-%g~K{IQWmLlxs1|n$m{sJl<
zTnIDTH&N#a`kt3>O>DTPKhapb&wE)ibKgx8W~Z+^+b(PcY&(yorGG-@hqfIjytj(U
za1EH-#;D3xcC<*zclZ16;+yVE5@_ejuoyKsKH)t#9hXiNK7u=e$6JmC8z^73bJ`XZ
z{DePdr6MZGpJ)7(SBx_nc3Z>B_-`OoYP<5{7I@{=;v!OxOWkvuvv_G$xCFw8|HSu5
z=K}}BC5`70cBbBHkXGL^m3!Ne@*&@=>i2*CTQZWr_e|e8!Q~%w$3a@YZIH2r`jg~5
z`rdQVF_np{!n|U1%YFTQxiP4B${Qsv*ze%&+tp%-IT{d~y6x2cgnMN@kW3Q=!W=4F
z5y!b7AokMq{akD7CBPI^^z3M|;wwrbJHo;Ba!@s$Y#-n^3XE#-=t0u7?%iX<%l^Eg
z$FBV3<U}cnH+<;@aWXUtwHJpO^+Ftv^JhLCIw%6q0=$^*J2Oh_>6bY?9hLp}sBOoe
z7={axU582VA#u6)HopTcC#MrYARvYZN~{YPbW=lF3Q5|<X2BsLkQUYIY~KC*w>)0{
z?ayQ$AvehTcL5!Wz0EGgd);{MC5i6|<43DEpqjNxm+=F*@5c{G!~()~0E~zH6AZ{c
zs{`E@x{&3MaHe@N(}0oDo2|?Q*wQ&+B(Nag+~P$U8UJ`pg(qp`hRmPK*9+}TDt2XQ
z#bG0=D);XN($L<B|FBlszGJ>vwvp>89V|iD;k1`tdn?3n;yiPSjZ{arm`tit7>i75
zF#U(;=0(TG4FrjprwnnlPx|x=X?1++(>fH+Gq|OvK<1gzyF5`S`Bfd%X-k1<VX?!)
zugJ5LgZiB=!7kkB=x8YH04Ij9lP8}P`?;l50%+B@)&pq0Yn(cFl|)_nsiQ>?D?qZ(
z&^fi=$%^+`=9=tid!|k17^8ox%b5$A)L$X}*`?M=1InKp_M_GEHVmCrwN?ltdBdZp
zb$0iLHVSg%a^6T3XykU9aUq~$E<K*k^p8hrgJKTCLu@nuX)yTgbn+#Of@vp)I!m2d
zI}ViOgG8zDMyCBPYIm0A=8Tqz185-#X#8wZdE;SD1kLBYma<O2u_&Ah-)}QB1^`c2
zdNI~L%jZp)5W4!9<Wj`^CSzf__0X(Cg)kN<xDPyxUVNgE>Lq={b2B`hsj6ICi9QY%
z(xpAm+BGFZO6fG^3RuOn8GkNA0$yKVA41E@n!g}nVP(bK%mI0F>BYrc)_5IbUI(~H
z%im^~M#BGQ<e(;La2=pRngx1b5VU*z__2q_A?!JIfPcbrnG6z*{BjVS<(KboPU>J8
zs>j~_7EE>(JBr@sjG&<XPLNeWB4w2!<sZNrn2yYC{`(>Pc(AE`Laf4i7T4A`r<meU
zqFAv*3Jc-E1Sy~V(<Ew26Qy+`<n-HyM*+;3xF!id+z7UVwnn;`mOJYcoB!>E4`H%m
z7*bka(21$6C)6kWzVhD}ei{;d-u3zCRim&Wgx2OHENViCfDjfzLW0^Sd9qlqRb?qA
z!4$mz-^R<MwX7=<!>qG)4qmmf)&<=iH-6ZUvEjcLEd~|eTHzG^wm)W{Ay)^*2Nr?P
zb*-)2(wAPg)u-{#gt>bF9UlC>M-!vh%2xOU1MxE;CYqrQvLZn3KApFSrj?csMC~LE
z%_Bq-*%}QJ`ch=3)qYH~Wb(c=eoTYoy|h6+$%bhKoTJ^-L)Ettru{9$Z;Kv+Xy;a7
zmGUz1*D#A=A4s?<n2}+N``{rtz3kx*U%OpFAviFVc?OOp{2<O|zyDLhr=;PN{z}n{
zQ)j~D*z`QCS65>bhpzv<uVc0R8E%NRw5|ueu5KSE^^}X&93SiZojo;PVqh+H-3}B}
znP{7zI4b6G(}Myh=b8R;tQ-TS0)n`C5!zouN%J_DSP~<rW%W!MNeAKJ+H(K87koZy
zo)Y{brE$)FW`hq_7mZqJ6YX<QX@Z9voyZbrqrZu-lY-%-jAvTN#FyWyn@u87_jQQk
z7`TUzqI~P0&eeY=K82zvPwy^_uf?eH(<$CErLIMo6_w|bO-@5;%`Xi<u*4r?dF8mf
z!KRxH-wZ=6O_|$%>Ul=5dUY_&awW_0(6np}m8*RF8n!-i_ql%$<-(Ixb$_7A_EZo|
zYyaNElibn+TNJNu9VTLqkI%CqsIj-QPxK~Y<d4|ve+8ni>qgH0(gpSCHb8C20e+WS
z!>h#KbT7WDctd7VPH%v1PoR@BhbgZtiyx-8n^zD#P+PLUZBJg~G}lLKk>O)2^lRh}
z)_$4cSB#<`^^r<wukJcq(sAHl_~i5Blh5Q}9%=UuZ&GKitrInH;#>EIWHWj{befg(
zt)=}aLRlATO-$w8yOs;C9`Yx;3MeTRcdl$5^Y2qcsYa?__nJ(!vzX+Z4)Bdo9aTr&
zjGt*%1eFV4VAbBkUgdjlG&ecK>5uAwg^}*ZMknW$`8S@Dcz-3Y3H35IbI{1;-Al*r
zrf-zF_N1TR0%N8)SdZ*tGBI|ZKQZfx=o~q!16_7`(w8YJPgStp68vqewe(@Tf?Bm(
zmz7jPC=!F1K_iM@Y2dC3+2mB-t7|MBMdL}Pls`#fNtTV8pvk{?t%Kw3PIsA{lB4~V
zQZnC(${Sopd1OPci$-JW&+y5M%bQ`pMD5zOI_;~FzSej`v8L{K9zkJtRYFfw2ER7;
zeYj`fyXP8=R^cy3nC5E2tdv60e=@nY`Z(>P@cVLaZe?vB*U*}eQRj_H6=R)xaU_*L
zrv20T`1mYs(59v9=bj|fCN=fM&hsCG6OHj)2b8s<{F@g2Q@wRu2vhla+w;zCnAT}t
znz!!n77P%Jkd5#5rXmsPL`nH~znKtLd%Ko9KYhYS7qkx2Q&4|=>n18=BqT{{c}ai&
zLaTFO+QFCuHutKBs>>LubgjNq94>zH0~Ls9TsqMQ*1H8J#I@e8@k#Ax5TxV0ej(C!
z;`kS~)K0VEx(7Wat|PB?q8*0vQ?GcIOG3BtRMj27RxK#t<c?6@=ZV|gI?}prG#*E*
zX0?Cq*am?WI?{dYm8-L4xiS9s>hi*Z2}R0dtu9V))L-J%D>#c=xlFj5f}n!4bj|M(
zlwz)6Z)zs`A*q13yTr7bJ^F8Z@T1<!jr!|rEmo{Ysme^h)0>ZqN9mZ!@P|{L4BXt=
zb_}BUP-ETXH|&pV=K)GKxq(iJEwZ#3a$YM3CPphchg2~YOWqST?cr<eCaB7<T$+xa
zUKSP>4yqTHNzv0I2~}z<fY7REe(Bg-)uEwQpbY7(pg#?cUITMsy9@;n5Nb?E+F#X+
zP&%Fo#a6#4GVSs0&zG{gXxk8yADn7y!SY1A#96uDPYle=K;mix@vRJA<T$vvYc5sz
zTWaR)pES?i34axh<m8*qU@6Cj6GQ9Jw$NqWfzaY%wmP~ejNm81#%))xCd$N=T~RTz
z(K(d<<Dbyh8wL_kB|`-2cs$FtB<zSwFF?B0RLT}&h>3+YNhNKo9`Ciw04B>%$^5gs
z&fPVhhYfi+drkcNSxi2jtOXmDVcn+sg0`-%<O-m6K}V0(a^;Vpx|}XJQknL5r#<do
zm&pEkxSVcgXTq8A)0>w<<oJpfuj$mpt&K5}(VRVd&*t)ADRb0N?87s;<htr-JPb?*
zsVO0Q+xv4yxn;az3DcDdjt=az*Y|&%2zfpXA_?$HlSzq=#!|?5Iq#pIstd$61sOLF
zTf>%EN5ym!?~T;K9^@<I&O+V<5d~X!2Yb!z=GJ%95kc2d*GgjKaI!7*5nPGwzbo~~
zNJ)u%J~YvL*(5Pi#|oD71x;a|)$3Qo_gkeEF5_<-FF${E=%^>&6x^X!?Bq_RFj(Q{
z2zv~m9P4AHx&NBig^x#)OMQpk&Kst3QlNjx@lO!;Dl0e67F!XF`nc9RzA^amWf#Zq
z3O(UWSL-<?nv2sGin)3F4cxisyvZcjesP?YB;Id)8P3l(zWIwZ)0y|gD%m$*(LQc5
zt43mDYziBg{^RqOrQ@);WIsE8M>Wl3(oCWIcS57+8JC!yc>;|n@7KsKhO?02ziOVH
zd*TKV-<4){!NOQ>^zy_051LJI_3fKIDD3)C<&!~4%m54<v+Z-K{u0%%F7}kAz39GA
zNSp*P{I-*E1CEby1JJ0&4<x48tqxI8*#sfT6BQqxRmYd-SS40cnRgGng?4_QD%Tri
zUPD5tg{z$w+Xa6%e6e{-Er1|pdS~m&)lZ#FSQDuOitQYR2xs!BZmS)<vU~0bq)XOa
z8R9czpRoIC_M7rY4~5oUymw^`(>k&B=uN^J`H!p=b9dtP>fPqQk9t-U|Md6wN7Y_M
zuDDf)os5k{;fx5<V<Pp|)puOKRO>-68>?|(K3QP|r8mcps|5*8`eFaQE$B!2{SC8l
zaODY!_zerc^{x1^3{8_|-o_8}WkOl~F1(9%Ds$<OXX#?;d<2bhC76GDBjol^uJF9n
zzTWxoK$*w^Rx=aP#yoU>dE`1NFP|pr11aJj6im}lNg>dIiNsJv*nc4#yB~aTb6`A&
z@7sM|8$QZn>bZc^ChVi&2Q@|?Qrte&m&G<<iCI`IgZO^KO%Q7FL~HxC!FPCaz6V4t
z&@2vG-8w{Oq~o)tpI5n9LSzZZpC_xH@;be?s;=^Tu<s_v_7hVK?6JOydu&>$@5jT#
zKO-z*2}lUM5gK~^aQ1xcej`S2O_ThB%_}Whxs%Q~Q$AN~m*cf`rXCl{4qiOM((#R_
zwg2dX`RJIhCpo@p>o&&!muco`D&n-ANT>JK)yoLT*|C#SK3-|5f{94^6an3hXZl)N
z71qOS5|4pSY76vuF!W{MM~jwh)SvDYd(MPJZc9$++xqIOypNn<;vZezNUW#S(bDqR
zUG4?Z#Ov31nx8(g%lWMk`mKh4Wpr+nuoS<-m8a)mFW&p;{-Ndoi>UkN1hjPEr2#pM
zs&6Cp{`&pJKYO&Ys?81^um64?peiBP<@ym;Sh{GmZe38jZ(J`<LXLxMAF~|!*)jRO
zWeP>}HO$O6IZdF<etd-0l_Kgl?aVHwdYj+tPxO><ipSjLrTI)!prfE1*Hu^N24)i=
z3QbCR@bLhj1i(0`nf$8Re|S778?!Q6{sWmDk&%(`v=gCtcXM;|@Zg7x=)V4b3X~Ry
z{ge;|yRN4?0MtNzKNqc|=@23(+tyNUP*54O;jnw~!R@93AA{xS9#sdT6>3#Lc5KNb
zd-ExkxC2|ZaryZ4Cw|}29t>Zhiy!0m+<Ij-P`YJ~lj)R{;RSs6WsOeCfA%-fbl{Qz
z?XOyP(zI*9nmZIDO-&86k5!*eZ1}8x{W8QC{xUUH^QRL3Nk=F|rNj!%d`a;QpfD4Y
zq`%@bvp1}F|NgF`nRXc|=f=@edHEV*7%u9uw<jukLeI}Vpl-!$UPzw|D0{xXZ!nGU
z?D|ooKqWeb<S1WW?4sM8s-tNB=0NLZ`axw<;bXL%iIMSVYpeZOiR@Q(=vb%_$by1L
z2)SQXAS)FoX&1`;<F)G^6BYGuqkI}LJ|gNv;B4bX8>D~PWxR-gbjnrF@ns->NGxvU
zo9g~R_Fa5}{aTK*{?8|fL1(4Q*;zBhP^#74V!Q}(nq-d=SBjDAcx0Tcwr;J`O6s*M
z(wEWQ-VS-8k&0iEK-LdhjP~2_BU3KsI@q@ez7huqrzj@(rTcg#p^{AJ5Th8A$oEvz
z^Iz8Z#I{5bq*7q=Qk_bF7H`69q)Sd0B`Z)_d1OV%PqE2MiHV7cG|+ZLnd3hx->Rb-
z%Gv|f`Ewx}uU$nRmF0R6qykCCW7%f{zF+AKM-YTF)O#+0<p;XAD2a<dN%l~DKp+yJ
ze6iGz(5<rF1;H1~u{~y(AehJdf$x?uJDO@LUrqT4($S@z8Unf?6G8nZEbQ*pNS#m@
zozPIVJsj(=#Y`?$X4*FhAwNQi`lS8^!FY+?(=6=V!Q;Ls?4!h4b>rolcO$SPGGyLn
zjM1y#e@5|wtuTnByuLtAU(GQC5XUe5pguu8d{`AlSGT@y6WvBeMWwT`w}s5Wsw16I
z2;+<)oZs(zN0Ca@w>83on!TZ+JG!qL{AIq#f2Fu;823gY&Nba4_Kj5BUV4RZf0lJ&
z=gXzp)mV+f2Pk&3CsE-fVf0iCZL_sh7W|=oF~`@ox;a`}s_rCW1Y`Oi>elNWC$SOE
zbQcA^Uub?sS9PQ3igzW)(WoUzG8`747ojvQ4_;U0FoxKHVs2+Hr^QGmcQBS(091%!
zElou>D~jaZJKPGO+&YlXRA!P?aryi16t?xAk{b@xh*<SL81O*r$m;wXbP6YR`#8I6
z#M2ksi&&3RK60<LA)+jFxNoy`ar)T;|JsKQ(vUdt($3nfUGi{Hk7|ta_-BK=)$NKP
zx0KLl1jxj4s${IDs#@N-iHLyNP~=FU8F#-ny+7sYHmf>qfgYQgY0&|3sm9{Xn6MiL
z+Czi3S9AOF=w<CLPPcb<Ao0rXqJ;EPqg82@$2OE64@LL5T{ei!yS)YNenaqX`|X`J
z2ugKY=f8wQ*td2LqM&Udec8ZEsajDk>7S$Nlj*NNaVms=`r-x#YO(pmb1zT2b5tql
zWT4heM`_!nwaa2n4CXQr2eO&$7K|k}jUG>lS308H;UcW<v{U<{i5W~h3IyB?Jta};
z9*G+LB|DAlESl$aE|;}<;{slu9e<^iPQ4$O&JBTo=RlZ?QAgCUXi7YKM1_OB!;OtR
z3(@!+&}pXDN2b*82RvZd^ab9PGcEaD0MnP9oX1@rPQpwm>qjk2A1<B=9JENr3ed8%
zvv0neHuyt@AjxQLR3F6a%U#n6$g)K3Dsu&kFfuZ-va*6H4PcU{#?WhMfyKG9=W+;|
ztex=1PBEl};>2hckC^yX*vk7C*d@N-jjg{F)S+&7gA*DlsK4Au*K=_`E$V5%K2nG@
zATAV#o0u17wW(_HYzNx4<RQW!jqC9AxAko1t;Ikr{LO08lT}}c(@FmP$tvLYEjeX>
z{#Lc$*|sq3tLt@>3(}wXm&NbkArOyI2a8p7+wO?*yQhb-6MI*BxHr;%JS3=v@H-G-
z0FX#RNy!iN0}e!=RclMUzPCw`-?#7{R8T}_HJIT^{-}au1C5n5lY6y#7i}*eaaa=}
zHe<zW#S1ay$%2Aw9)<EQQJ+aOlZaYZe#_tLD3XiysJ=`W*-LVR5P5;T$P)CqMnP6$
zOTCHeIp@w#y?<RH7pc>&#qr$3i3nb}#kI9YK8^bME(_kLYD{R>b+XJ%%eeGAYr*Pn
z<~H*jYA3wpVVO-Vg%?)CWxuhku+5$9KHl#jz3pKz@t!W7>7S6k;XSc@jf>;80&pDe
z9vryW`L$ESU5)gcj3m2W4|rs-_NQyBZe#8uk^T9_4mK=t)h;Xl;HU)H2>H0VxrKz#
zQx5FIVEQ$;o*CF|a-8U8nDzmcrdjb(r2Eu2hd=%jX~@zrk!;SkpqI&+=ld#(5D|n7
z`-$J7%bbU?<bq+?-s@3SS~&`veF9#=-DP2H1P7DsdWg&NUq4&Y@k@LbDT&wI{?@F<
z-)3!Y9pxV5`x&myS)nhc)eys{M{w~%%EhJRmA=oO&jA4Rmx3C`N$GY%54!}>xZf*9
zlhS8^OImYK(s(YXaV&91S)pxLjRwn8wbcoRd7OnL>GAZ%@+e8(_J6_4lsy(02otl<
z%=@L{f&>#)R0obW=i?&1s4#|J?L`d4bN5&jsYujmv|N3KDYnwS+6u2`qnqRWN=D4*
zjaNkCNo5GKvO=ovJ^fAh)bgP+NlXx4kj0JT?afKjxKlh*#>O}EN{NK%Y{;W%p&nt!
z^NDwObU@uE_tgbNN>h9tTy3v^T^(TfL2zgyHFaE3IF-o?R96s68?5++>SQ4X-1<|d
zLI*dWul=-rgaI6Ia*8+pSV`cHC5w5@2&km~wdO(k`bx_*R%@M${sG^VN5;Zm)GLn(
zBLTazZvrov*-=-ri>7S@CB0Z`*`{pp39me?J^j+1(HzJcC(5ojL`b=Mx~hDW&IA$>
zpp6K9_YTf*Sh7FR^!&FV8`cu_{=Gx>RvE0X)3sji&!10#lW^<U!FeUD5R`}dJKH8b
z#~4V>xz%~f8jH6a^q+bK_M=6nN9=owa2Ci^y`P`D`T&T2nBTeqxuzrN`f`RutK4&~
zd>W*mYxJ^r4Bj(P(P>ER5;;N+kM_e`1-&ktqhYFFZHn^ZO>h;Gq*U(7YDhQ{;9vQO
zsWl|bmzU5Pes+o}Fm#b}ho*r8<4W<y!8B2+B9f1T7;Ju7@;emqH*B|Y*@xiAZ+$po
z2DCQgQFUgeJNV`aVTB;ko-8z$0ni_ea9U-@!o1Jd{^<D!PbC}Ik%uwBBF4|Jl{=_W
z5DbsfSoT1V<OX&+<qHSD;+I(cdHWyn>odplyzLa@v$J{kB$~8&-_A>DF}^ijb6iin
z&SLgaguwV-L*ZTG6afMIshWS?zJFUNG!epQ^;1h-D4jC-OjTGa85m4|^Mx56KIZq`
z9eke(+@}WRHgFq&RzYPcM{{3y8C=vN^S=csMXxGS-}d>lMJ#gst1{NOkp3!COWz?D
z_wCu4<`m&Jv4!OWLcc@9&@4jENQ^5{^cW+TD`S$K`R`HOxzk{PM|x{7BOPhD)4@9@
zBoDHft-^9gq#2Jp^D~oZh)OTny|#$CIpzGUI*r`)`>YBl58lRQm0sCs<?!Xs&s+)f
zOL@O{40^n{Ze|1`PfYs?at$UV4^IxsbZ0>m|NNX%;uowAVPRpQG6VJhN8-73Gp*%Z
zz=o69!76T^#{zckG0_Lui7gU8GfNBV|4;`E%c|n^_)%Qft?+&@X(gQdw?2(Ykgk;b
zS(Hg)U8z}D%dXzW*ikPTHkpiC7ehQNc}F4^Zknwi;-9POg>#I{jqfg^iOJE<iF~Yf
z5Mk$D*O)G&6-_sMw1mM)hjTRK1qZ4;le)?jilzKjebMX8?9RC9R!!sSjgU04C;j-L
zsxZJ7pB(K;*t5DgF3<D_;{4;@kWqUkZa_`s7vlIqzMq^e=+$j3<ndG;2eZ+S9y*EA
ztj-BXPg6O$b`ZUz%~kRr8UzZqs}TZW@?)!Ji7lOA8ibli27I;P%vm`%XiHJ}yBX7P
zR&6VEQLaDJDp%it6R!T?L1lW(y-+^tIR$r34C24W7B`t#ACbzo*SXzV9mYkNP&VvL
zP0cs0ZX~8L5h@guWv0$-L&~iBO^b<3zq2%|-Mc)Tm=Qd(7Bw=$nA?xA#M(m$Z=bsh
z_v{`r&hAWKq^>LDHGr^QPbe8@{$A5c^!)zT8KnJwd>8*s27vFQx%Q`cc+2jrOA(}I
z%(HiQmUV-I&(Xtw3EwAJ?%W2Cu0ZsBge%`jr~MkDb(1&lRrvd})8tp9m+zBgQYYK3
zzP%wpcoS#UmGqN$4gdfTWigS<O!J}z?!YFu5=ze;F%?@b&op98<cm*5B{+lPdXV^u
zq8c|P;cfX3Bp2NSW+CY%P<9+09YJb14x=zW;^1(RM7Vme(vN%YK9%F_hKu0sUJNYu
zeS(EZUw{S^)9Vv|6^y>9$k1!B(PYN!J&R8spH|`mKKUQ1PBJwS)D77VB|j3%(JB0B
z2YV8)4@7UCot<^i521xCp0#yANV#+AU=au#q#ZoaTDI0s!$pcmYwQMdc&kn~i;|;E
zo9?d*M^SG)jPkSUhn{JSg{#(2J$g;LpDW?T#Zk9kJKu-sXfF57=e2sG0BVENU~03*
zvv#Vxd&{lf2izA!GLkR-Xitx(qpU-o)^f#>IQ_@vjesRb?qRQoIf~p9wL2KnK{hPH
z9J@N(7LC#1bobr#uKu5&t|BkO=6XcDzI=-)yM@;Z)@-*c1layKH%nvOvb+hR=pmZe
zIYovK--hh8sknbizqG9BM5r8|3P>dn4-dx)8Nh4q@gt5E8j+r=(pYaA^@#Lrn7SHU
zm3VZTGnnF~AJjz$PPfT`o00sQ{uW2XG}rSlrtkE(tFpDCkA**E@}UG>d7T;>{wdXf
z0)wi!&i2ex;j|?B8P7DV;z!1iK=Ivu%ACE|ZLlmH2Ge~%1o@5<1iLusbTz8m?*>^R
z`^(k>iDmAqtB;p|o%zT+dWf3#u&BBd;x6;J^n43Qa(XASCBj>I8VxGlK76||2H8dq
zuVN>f!)Yn*WL{V?Q@xDdt6JQgx#$@q+&(|~<o5uN?j2A)6eMxQ?vG2#*M21*={T<P
z`>^u;GP?QgkGOmC>0v45&|@NNdY&us0RiG)F%fZ+Psi&!v(r8@x^i24KL+qym!UOu
zI$dq4iww8_@Y}_EG#RuXrFeR8^L;3cw_Y<3Ia2hE?3gF4Rc_Z`)CZ}(_eVzA{j&$(
zK6}LSlsA+SMIc0s1R3IaS)-=zbnK9$I8v0jf`aQgjHT6|Q>$#7?$<K2vFX)$SN2*^
zbD7Dj599}DJ0uv@P!a$#(rWOoj_A%g#=fHNMsk&^8L2bttFBJ7f~CG;=li*_P9tAW
zsa(#(qXO;uI;F6;4d;)0DnyWmp}p_d;RvIrl9_kpV)J_2#Tz^P^m-pP;#2R8Ek_dY
zox1Xc{iY2P!X@z0l~%U`IxF4SenZdd0oSonj+PA7ZcN*+htws_Y&XMW=mR=&kQh{>
zs9>@x5$-Qek_6^bAx_f=YUEr9|2?NV8Xs%N^1)`@rHitttsYtYuQELNxPuqJgNN7l
zKt^syZZ3m2Nr>7#xRxoWnnSy*QJ-|}QsO-7+gpyu*A~C4@lt(wThb!0jf(m9dSOXd
z(21Y9ffy;l4<54d7jEln)((lp3S3Afi4M>|AC_L*vo<FPr8%;(JL!5u<7Eqh98D7M
zo;Pp=sazAke2$S>V7VSOx{~5eIs0WJZou3|m7VbdCpkrCElfOWIl#M$cKY^A!s`%{
zUJ}G$nB2*uAaC#PGttrRciE*yL+>m~CG#1ciK?z_l7J{0t=xkEK~MnMl{Ya+EW3U0
z$ylXAdyXRSGRPWH{ygsv#O#nedhy;buu9v*Vdp1XfW+VKBHWnKA0HInyXSkWX{c~>
zJXelplw4!CE|JWQ<$0FA7kzf>PA#;1;<Y6*p0@ExxD8BnM_qzjF@ByJ8bc5>ZOWOz
zsQ7Ut^9hAgtwE60J=E^9U92v7i0R_{V^sqVWZAZFp<FLzDsAIiu7GzrO+Vw?FNB_Q
z`tmn+OM1nU-YP$r-1Q8}!j8M{FQ#F3w2w>5>GIewB>f>Tp04Yrko#&y-ru{=SbLAJ
z>LG}B96DGB`$72HD~qi9U5VNhzrTx}S|m9yw3Z`r0X6Kn=&*O&oVvy5aR?<GJ4Q?b
z)rZ*Fy|aO~os<1b{%50=Mr%P`o%5ED9+-0XNei3ysA;fhI$e(sDC{&g+zE;LLK=71
z*H<ItEYC6{j;;~emLBdJcHew`l`K_*G&51%9a|0phETh|iUCuAD!h99j2BvlkNQ0k
zEZU87v&}xQL{p9;Y6oZ-5wTHF%dEGg_>75ZeqO8u`2%(@nQf1nG}hy53=gxDf|>n;
zdOW-4$N92x`3joZpV$kh7`{%Wmy7+Z{ISaA$ZX0r<JmaFS0ynmN<6tcc<N%b3jTO<
zO*{9i-F(jWhM)ggWQ(TpfkHO-z!R{`DvRDbDr68_4M(zk#(xJsKsf~kum!F_;z2Mj
z?Sh_1v-+(X!*(f}s)8>$eDc??Ux(BH`GVnhRD7FmPe`!*pXHZ5%g;>8{LMIK0mM_g
zQP)GC<B7t%7#dH&uG@_GwcBK&%(Caa#!;biahV9)ip!Xp<4Z!SYvNKdm{-*$m9NG2
zb$#Eun67y%TuoQk0+zfdn^#%rLgz+)L|DsQ8FZL%)^U$G-`)9_39&}=&MAri9<Py~
z6gg7~@jHPqpDpASpZqML+F(d94MA-@L2X!$w>gR~W>7-lUf{Fe^{3xWK1wUI@FOEh
z$>V!<Ck;kvHX7|_nR!DwzVyE?{iBZP`NgFD6*E(nJSJ7WP0loJw8*aULU{TC2EwcN
z``7ZPUGNRYyx0UOufFE5tLMJRc0=MG3(+mdzn`_13wt!Fh$d4XhNiz(m5M3fr&FPK
zOWBzm27Td3Y-}+rN6E&rP9U>Y>6itliUx}oYeD@skmrvAunjva37T&wQGZrf(cscT
zl?-B%bJ>=@_QaBbBMiub?(_fZWaH{DzUp5Lf$}qEDo*0$dmG+&m@c!Eh|6AjiP!!L
zbD*T<F+#5$VK*AZhN2BXYQ*Ghw`~ixGz#?c`&Z!1kzT-6@b&c-6}zTty^M{GAtz`e
zNkIL;;2ofKt|c=1T~Hd-L8co(&kzuT-;}T5{^TAnJG(uw0&sW;WNV`RAJa_`=o5L~
zJQx44T-5RIg}uT|G_g`TYo^F$l-c(WrWq={H;!_qapWm+JYH^NUJX*C=nE7#yU)xc
z$Y;6kq3~E(J8g#hW~8a9qWGo7mEEx0qFlD#9Th&Km4vr%E*>8$y%@=;r#Q4ogM<DU
zM2pe-^hvvdg+nU?ZD57V>%IRkF`+|6xq>`OUAPeJ?o*J#1eSg9`T17c5@^R|nvQ;B
zhXH5AtENHlAc-jqJCFRkR)Z0vS+Q<`+G9@J;($(k_k%Q#Nele&H<_1SGI=O(rz-oa
zUdFHcWis|L%BkOCb*fx{TrrsmTgijQ1B0Ku!p~kQx282?cnvTg8HmgTA1ss>`0unq
zmIsOiBV%J|ya%I%Z$7y<WK@{<tTZPz)d1l9>voRwcS0a?{p+eAh#NId-nxdoBlleI
zjejY8{IJH8VbQ^VP;L|z(<wb_lde*KI{z-K)E*1*TUx|zX?cZ>qv}Sr@jY-5pO?zC
zc*^I!B#HkbUI(XDxMY8jsVTOSBei3fnvs@)cGJ$ZZ~f=^hU3epeWd1}4wvV9zJ~-Y
zUGC%;WyNa-lMwIDm)p~;Q!n)0Hm|)o1E?c;BVKVG=mXA>$eUIJx3@%VX6pTTkEh=}
zRRnz~q==yp?1Qd$!>Fsfn=h#oY=LfPn-=T6f`)0W67QfczE*8DnB(d!I>=Npvneck
zvd*>S>5$3(bL!V`@1@d&WK*w*#>1JIdxB+Fm*M-DtDq06ikUR)UHL{MFp;Cjfqb|{
zG=t61p}cv%kAr9=Ffpax$459_S`_By#2c|5^~+^4k%rErh*~3Ab)xQ2VKr1XB}rZF
z-4JxIE4|6v4F}3d>DETp@6Ho%5!idoN&5~o&=C{fR&m0*q_QoBfq4Jj8(vwvu*g<P
z>A_*sttv6u&UeyU2JM|O+{0Gbn^<~TvX;K;%=iS?yP;4GQ&90(eIB&x5Y;7XaeR~P
zOC_M*;#jagBD`EnjF#;C7io)=-G3EWq<vBCW*b-%EqgWG{u#<T?CrY<DqJF5hBq;t
za&PWjmHMvLbz69XlRhY2%)B&b;ATXc@Lx+62GL|W(pTY>kch^DYn<lZq|JwSbZ5!K
zg#q_F%Bx!3dC6Y??0CG8fSg3)ID&dD?;j%Of#Kocwi(54`uj2wrax__)4cw8YxhOE
zx8Rq7m+al1GHpeXSU6oTP6(-_Fc5|}GY<aE;%!AHyk`|AvW<;r?aLozTEhu1<6~A6
znlxb*>g3ZtWRk6V9~b9(N(rz{8X;^-GP1C_Ajn5joy)zt^;rx>0fu<~ZJWy01F_y8
zv#&p|D;Kq~y0zYGj9xTK?-tsn$Xq~|``PEBGimbS7qg{r3j-jxubPYudY@x=PnoeW
ziXKiy9S<*ur2G2qnVu<yo432GAMC~s!;+-(kEPuTn-<|XnRLI)$!<wD(OyguxBmDt
z9-afk74!&uN0a6?0H?~B6pwxyExyNDbtI!Cs@oRnUT-PCsILr>Nt&N&_@UJRwhT?>
zoL@g$DmEN4cLsD7v~Yf_6Gv`4$vqneX)?=GJP_jt4cQAL8lR_^1C(bO_cC4Kpst*q
z<w(-aS$`4<S88%_gyqbChad>$Td#uA4my4m+Ai8^m<FK&54&XSKQP13(ddh4ysgbn
z(d+oZOrdu5hV2&IIqJ;A6uM54w0Ii1RB=Vj`9looneJO`!Q*kKa8Z-pfgoa@Ue?LW
z2-3q;zvi)SX{nIuK41L#CTk@Mn`g`N$5Oslp%KWhrXrf2LR2d7ZF_EHY}#;iD2TXM
zHkh9Cz!@>1TY4dgAAbt6C(uOalxH{$!bwd)+ik%w6am|EgUifAF!b4F^kKEx5xLmz
zKY0O^ZGv>ze%Q)E)c}7Y3lO11TE&iIogs+1fFpyCWB}8`k9#tbEZmSaH^y1f3$8|(
zD(Gvq$+SX*0$wEn(IhDNZ!N7o;)|m1Oc9+n;Y4#g5N$n0^o0|&pXhT0!M%-P0DKh0
z<nFpjmb-$l7?{iuLn|rIC+T~OSAaza#D8qV&Cl-w-bNv6t>X<VX?~-@_UI4t>(}A3
zDf1P8A<HLlO2wJ#E9P~h56|7$pkyxgyYP)Og#{j9^L@CcK22ul@87#SJCG>*6WB!W
z{d7z^dSPK7P_%<i%UpY?wO@zvnSR$cux$`{k_C2Nu!JcqzmJb!LVqZ0Glxjh2zd1Y
zCG>^;4I&lMaRFeO_Yo)*=@@=x2cfo*>BjQ^i&1NDDVqNK_Yd-&(Ozs2CwdCTjTb%3
z2T>yPXid}q_i+wj4+C-!5zFXxpkA6s-yQosHwdf)gF3W8yb1D+jg2Cov{Dw=f~Q_w
zTt%2-!FaR0;lP}s`V|&iF4$FojDj?Z_VHPD`LF5_h1hq*aLZey))f5&!On_}gF^!`
z4HxHU@Fn*D{)l@s6i-S;N%<HrLC~E|N`*PLy_IyW*xTo=UcN?Bry25H3ne0ATL`Ui
z*KMejoWU#&@f2QsG@1kHy8+$i9&J$!(iwk&f7D=(txFl{>pgZd0_Q2@(}^C=1VZ*8
zGzmQiq-#muwKBI!Nu%T9a1b!NfOQG|?jicZVLt@&G2k)R+=_T&rQ+FG&h;@dv7O_0
z7<=9Fy8W?-QHLbFYF&6TEB#+E8mvmei!=XbPrJ_dK^YiAK?F5kY_0^u$7^?a79NHg
zoCJ(!Pw(C1y252OkX=1(*6$258qfXpwj!JS$M&)%bsuM8JA#00H^5dPtVgKZePe7e
zo>Yf%46uh|8zWE;gDSSj>%a~yk8f{~($H*nz(;yClYw;Eu};JV)aA=;|1-IX2*ANV
z=%@ulJv*XV;@xU0SaN|Z^Af}l^2Yw(D?<tw^x&}91Et%Y$U}gTv+D`L!U6$9kWhdI
zMIbG@0df*Sb6))61DvJzhd*TxW-FwJ!%v=8AtodOO+xt!M%4#y>ssYo#i7a@b8Am%
zQ@s=bK7&AYD976!Cm=i#9Swktz;Flr6W#=z%~t`gTu?ItmVXA3B}flsPJmcCx~&A>
zELzIHwB)`%;s|pN0<u<CR)B~HeHo<RZVnful<VJ+&<hNvIf{#ki2>v};szHE0^sBR
z*7P(~eCKB;d=@<jLIa-)E(;cBW-F*%y2cN{Mf)i#N)^tivKn9tmgt2q20^-DbBBk`
z!_172K;Egcme%qlMeuufc$RfinW_qKBrw2Gm;fNj<7wA<3n)zAG=_j9GAV?qQUQo8
z?KmkF)aXT7YH9<x<wPOc&EI(U*YNJzusy)+;8)hjb)9K=*>ydy%x*F<C51h7PD%7^
z04%Lgz7~N0*{U?L^AEI9jF7y|z>u$9EKSw|rcJp0-pU3#iC>q!c7K(-z{3DjI?Se$
zQ5(oL)k4RcLSV|Y&w(%uYls+El9TfUpXZBDqHq&11~Ci(gAc?Zyc6`1-!UD=ADq;@
z(=AF&M6|!ZCid&qOuE^xVaUbg7Zf~!F7d#P7H?34A3cs-srQ#73;I`F(wRbnf;<FC
rf~;8nXY-itA^d+v*8kaQ&Tk8(n)yY)d+7pavLmumijv<X3<CZaLvLfi

diff --git a/docs/dev/articles/getting_started/unnamed-chunk-9-1.png b/docs/dev/articles/getting_started/unnamed-chunk-9-1.png
index d31eac9bcf48f4e6b7267ad09353d096b0c8bcdf..df21a321026e7b4ead2ee93be92580e3b0df555c 100644
GIT binary patch
literal 24497
zcmdqJc{tW>zb~wLlqQieLrM|~nN?({Pzo8kl%dQr&s2&cWC$gcA!RJ{tO!ZSoGwGi
zJkN3&-p}t|`+eX2{;~GE_CD5H$6ot*?&Eo$y1B03aGu}qXZoH`E-TAZZ)ey}MMXue
zs35CKMYWEPifSX>*3I~v)Vi?}{Lf~+i}JEmYs7zE7bOH!Q5~XEls$XR@zrRD)Aeg}
zJ<^j4{ATOVZ8+w;mEpwWd81smcTc!N8ID?~wb)g=es3C49o9EAz7b~e;ze^!s`_p9
z6MYw6>TSCouJt7Mb?c*zXZLH;?xycwUL8!5u76=ud;B=7Q%TgF)tj|<#5oEj2W)Ep
zK9vjN+$JllCu*S4#c_x|ePfyL<A+}++1c3y3Lh&d?BrwLyZ2xDD+ArG4Y?WtERuKo
zzCB-J*}SL2@aKnH*NdF(+-PFGHq((bQgzPS+uOU*cs?EvijLOplF6k}VSMsG{?vcl
zcmHSnR9ou;mF#{N7XNqej<%iL?72=&P0fK+US59m_;J~@XV;xPb;|1RwBa=kjn?|d
zPgo>&^6hx&?XBNk`ba(NikjM^Q+H|)9X`Bi+qP|w0|MmFpMT^hr!f1w*H~ZD>h9f7
z903v!2BJM!y4)$esiYzw!|KpCPBZ;;l7j4P{Dll#X=rG^M@zHbsf#LKoa!BM5s{^4
zDNDLvKV*A7$AVev@1K1?$ur4OU-x@yY!GsqHZW_?-J_L~UF?0X>CjZkyw=^hVYOsZ
z(Nb-lg@whBA=|Axcgm)bihlR^zfFt`<dE0<JKYy9>0)nA{aoSt&NQuO5fQ%UeGeS$
zc-w4JpP<6}JxY?XzP|p4kzl%c%}b#jxQ(AhE;(xY(=#(NMn*?p3Yl(JQBe^Q742S{
z8~IvYEn{MGY;}1?CFz$&b!R6_cXzi@<6CN1H@7QVT0T4)sfR>G54NV>$jBedHYc--
zh|o((N!_)v@yy6LWzkvqBq(U_emc6?xVVjZd3lR7b<$S771R-<!JeKQ&dJE^p{L)<
z$HzBlD;5$O%Ff3}6UL`o(Ou^4bU5z-f8p?TJ3FDEprB_VArC)$Zt1P&E4fx!nrqU0
zjFXe|bz<VT<l+Yp9`M`rU)p)_<i@jS&!!vl>Ezx^(as^BLZZx<`uaU-h0{J&W5&kD
zKlAN)s6LU&jC>`_C!?JD55`h>6!*?gbOqq`un7sNsjcPAjgjfDzP<GM#~T^k=bL;I
ziNvBDE#-GLLHX9NOyk>g#+H@=S7POE^?Z4<b;k}*Utiy<^uE465y#0xQj5Pge5tIA
zXr`%p#t<ILt9^lrVb-+vm1sKU*~^!S`6<7D-+ce>ox+sO!P6G6hJ__1CF9ddR<#mS
zQ-vurB~CMMXMdTdDPm*!tmNk9Rb^-0D%-s6o}k{Bt?xg4xU9B@W!f-UAFDQ&fUnIo
zZsf{XsHeACUFq%ViMtqfLbIt34=!FgQmmOW@gXlQH&-OWwN^N!bTa+7VfEztL**3}
z$J|dyuPtXJW;@Rf8&0vdbl8vmI-FNfP(2^dRrppp^4P(hn=_VDQ&U6P+1U*!*%=wZ
zd-v^2X)9-Fa?RegYgdCetGkh%?%~6SRc&o;i`H096`Carc+pV};lk3*8Fp@dkF}O=
zE;0R;Lse1nNzE$n-o1M{Q`AkZ91RT(Uoh+B#3v+Fv#cmZOLfH9D_xnfoc%ixO-`0B
zBJokv^S|!&2s^)vuhvOucalb0#jh+=C5u)gL2R1J`7yGe#lrOTIagPSm%`>I>C~Q|
z?|u0q%Ovg2^!)kr!5?onM7d5YV%;9(U0du8(60(Q94_rq>?A(pvAnXvo4AGQ^XJd>
zf=0XN$2&}IDSxVhm0D6#Q!B8rtcIJo_a8V=isiz>!m@Si)-wspk-K;AzB%vMUmNMW
zZ6DXy(5l+niq#cLU~upTY~h_uBD<NzZ27g+lbZD!-zsgTrarH(9@^H+c2Ax}{pTys
z(W44e8Jf9w-&8G46jM^oI`k|36>$Lz%{zAP3|U?5r3s#!buMa3y&>-FUpTH#BAqZN
zXDd06q~yc}NUvW1R$Hr{JcDP$IM&+O*k}+r(wfFo%PnNtsly=YoH;v`5Pf0OrcFVd
zRjf`<E-n}2<Kwv~$#`-7H4#VJ$mG^{3H3&C2T$GpS}W;3RZcgzN=fc--W^a#PdZR&
z!cj=2fP$Ak=DswPAUE=NY3Tx1G1rLU?b{cy=D8?yQ&WK`2udxwITn%i&L(9elig*i
zXU?4Q%g5?y81E=B9Ep^0YA`B^6n7BTyfgMI>$Y3Jva+(OnOU;gi|0HV-~4UH+w;=r
z==exx>8Yf&FznRf6$h=M=G|q!4nI#H`uq29I+@jZkS%tivZA82t*y<V=+O?kZ`qNk
zSSnI~zx<iUPOHl1J96a678Qx9ji}3}g}Nv@#QW%6-NVFWHoI+uxOihpc7W8<Q6?s)
zYXNC7F+H_oOch3g3=%7|ja)gpg^qi8)RUi$^^WCq2|GoHsV9|(3+Ojdhnn1U6ha+8
zoK9BS^tk<(o^ugP#A!?QQE_f=#q6%h$-CvkPc<`*_H5s=!_T6?evCo+mB_91RQ04s
z1qB85&Tk)jd9kAnoVxw}jI8WNtxTg^tFI>-vP?DFgoQ0TlXFjKD=KdDXBNM0P5E?x
z122oNe#JJ*X4#?5)!I202i;0E(nwq*?YUMPYzAu8M#*(`D)uh;AiR+y<@V$1Xf#R@
z!d0PF-@nTm8NI6-!^&KE)HNpl2fuEfsk|`WQD5Y;ATCWuOUouH$viYPq;T<~jJ5Si
ze^%*To}XXm(pX3HQrQmv`244Fzr)P*bkLYgQnR(yg;N5B49{-YysWa|J9aF8Lefu;
zLt7=GQM2eSJJsUS(#so1BqeW58L1BKVZO>^nyg#Gf-;q#XeBZGLuRm};I#9?xZ6Mx
z)?;r{T5+j7tJ_1|!HeW1+dsmnRZr;!&NYi%va+>(7#0@R93U?wEMabT=vR(q+Ty4A
z_={oVF8H$Ms&D%)d^UOdajG*ZJ-s2Q`z8vuh)w^~^LF=3=I``>e<9*D%}K0f?bH)Z
zYSJ`&(GKfNJUq%zqLgPz@80)kob}ZC8qYct<&H~Pl14ucaUMCMeQixxUS58g(zR9*
z!hJQ_q*5!?wb+jt{nfGi5uHH}<Ic@-adGVn?ze8;iv0fkXo~x-r<{sDhR3P-sB&HZ
zo?Kd(2nY|SA8pI1OnlB5dR$X)_Ui)L;DH?DT^FUgD*SCo)Sup6iTy#m=9Apq+;Ri5
zoQkD$Bdy(IW9FG8d;9FsNQz0Mv`6Q#u!$$9;?F#rBlhF%Pq>s`zE+8r8f=K)aqir?
zAK3LM;G5mu-F>=$4-Gx}`c+}{S5{I5<-*KmUAMXCN>UNkVSL&ha{6VHiFer8*sf>X
z_R|y1J1Ota^!ciaq@k^LFqhI~e}tu$-9rwa>1k^;#6xUsZz?F*o~XrJ@O~fRPJ-tO
z>_+T$9hpU~WwYyI<pX+F7P@j=e;5cv3Ae<u$#|iAYg!femOk3Ch)=s@(UoHvU1c5>
z6-6&*`>tHy*3K@)ihTd^g^L#h!^0o{{CT}CmQz2tztW#shD3_05<^+bGBJ~MnU}={
zC7Z;4)N^bWNKyAC?x;O)-_+Dp`odbfuztnPHu3r#b268BSB7=pg^aY@Xgm6yg-&0;
ze!c8^C{GF<T-JWw8U3A`J<WDM4-4yBh~iXCKmFH!q=lNVcriw^m+D?;k&8aoy@=}~
zzs`*tHK{5R=E<~x8$uR3xX?!l2?=ecKR0}$Oq8VMZnU+x7gkRr`Mo%GhoNJ2vW%t|
zU43ytdhH3e&C6E`4h|0TCGIPe0Upa61Ox=Ammc42PE<3hkJ*%M(c$ApneVP-_4v}9
zq+vMn^x&z@STS5VU6iej=IFKMU!Lq!a%VwhKe2wphL4U@dcocA>W%OPTXwQaG4)pX
z$>0eYR$ll<Ps#%3AakAevgpXC<tv&EN=&~Pdi<=84wF{4nX#=4Iotf>ojMi+L&I+T
z`qyvYN&#P9HxB(s);wi9+_a5Z+&;w3;=>!`Uzyw$0jwSaE+=fPt?$l{Y8n_Ayauc~
zZQ1z{MJX+f4;$0m+UX_v(#jFEDbI3WI-&`3jnsw<Y~U-LT)&H1?3z9|3-Hz=g@U!f
z!WFG_xEi4JosenE&Ecl^iI$!Kd}X)}tI6MI+e9<fQ?%ZwC8=jK-1h7l7}z2$EiEf2
z_wt4VZZ7w3_k*f2Jd5+*yO{Yc+C_p5QGtJ?=}OOzwAPKTR=M^ppC&7yt?Df=PK8T2
zokY9Sv<qOBCUCc!Lw4g@SYY6NKxNX$_Bh4h$$EL#=Fg86sIk$n=;%Dn>2%&PM$W0k
z?&Gx>Xw~&N50HsQrW?-eUhzuQX7d&rmdyM}FO#lfH$btgulP2sI{$E8njL!V?=L4O
z=hf_n`lC;oA3Ms!laZfN@i@V6PByKz_xp=eBDRAB3h<pMS?m0hxTpWvv16qVx9$6I
z^Yi*kkz%xf$6L2;yJAPTfVUqQ6!ak4WA&iR{ODO{=X|9K15^A_6adO@K<?Vri<*pA
zFJHdbQQ**<Q72V)f5VpV<1V<;#Z}70RMS<JTMcoFgN+FRiGPn5kph~Ut}Rbj^C?`q
z^rhI1La(F#L%}4f`<=|{>gw0p+SG|#U$@iHgx!$Yu{e@mNJ{f2rO6u_9=>ql!exBN
zg9qykZr^6-;Ym^sJ2N#oc?J~icW>{8NaZb9%h%M@-ehL(fAr`PcGM>{((hVrpI%Lv
z3liileIYPKD{DIdL%R9Twl=myhp4d1yu7`)08071Rf;IBsi8*qKFY^eSrc(OFf!5)
zEa!Jmxvy}Mu|{nDAx_SX056~$_~)**^*Li>Zj1KZ`+)zym;cV-8$T?bon1OWN9WC}
zo!zv+KW_iH`@F0VgPx;R?Z26L{%4TC|6lg|?|BjbFzMF%Sot@tBO@;K25<ZrPwS!P
zEfke_thr}eb``V8$!7oV=}`rRR`J}yr>onw#Ij!VGZ5b(1%=HB6^S{(i^V{~ob%tn
z-2XkV{r|YnB3{O8FVG@-)P$>Q^EqAar<PXHQmmF{jmwv`QVx9l`0=)10K*YKPDNEW
zng<NoiHXMmh}fm2OIj#CK0c-RVbdbky`5}P!a?j?wru%QUf!&zi%Oar?5^OtlTWn|
zusn56>Q!W<(>H}gfkK7n*yg2OT_x%i^wXmf5)vj^G|qSV3l(U^#S65g<CA=2#hvfo
zEseTHL1QKWot81Vqk|b-*$9;*B~yilN%?26b4R|NjDmtfN+vTSWA$(L4F1Be$3504
zU8eCjqD}_oIsQ_(E$Xo<0m5{rQ5Z1R|J5rd=m=59s(uN622o(<zg#;_X9w$do;b-^
z=KKBj$W;ma;$h+!tE;Nco;gF6n3R-a!NbqL`{2QYPzQwR4+^nk+06i_zrjisHXeC#
zC+gPS!1MGZgFV`j!&4!F$IBb#(iLvqVk6XmD_3q=F#@WH*bRqJ%b)k#s=`=xQY>;8
z{)Kv$!0qq!jDB7W9A0#q-(Oyp@{=p;v;2B&Ctv#YOP6-Gx3?cpWIS-dOW6F^Pe<9N
zaNM5Wi4!MK$+O>`I5J$Ng&*`e%of*dVmfwE(L@55@Ep9Yth%~7S!~0WT@Q0|9C-a6
zGwi^}mj}&_iC$63{xr#+^ulKNtKD#wI(}ZrrX^=UY|elOAN5jnggkSMqis7Y4SNK*
zqV$gx?bB^1brPb?9tUg*GzYXlmTFmt-wefXp1QDLftxjZf+u;(dW38f>74#iCefXs
z#(C?h-uDa+KK}OY8>N$vheu|lC1u3@4CkxaSzAH#U)MgqN}SBM8#$1kk&(i(Ux);n
zc9fSlW#ZN|Dgd2C>sK@X7DeuVi@iwfpAT8!MY8A5SA`1D?|he-m_Kvu#0fbpfbHA2
zKYQ`w2|#OXety2c+$Gsi$3tbN|3l5ppy*ba*Bh6GajqOb-C`!-nB=lO%^MUS0Zl4U
zJ#?&jaZ0}C#EW=nCV)yWnhKp}oV%op7uO$3Ar)!OJG4|*CUjOB4JF|RYN8Fd4-{UT
zoSgiGeq;r7^psV)xY%1p_V>dgO<OVMmK1Gokjjno^EB-TPX}HG4JI&&utodU6m{(`
z88(wlhK}wsH2jZf?c&k}d3k3O8riTNik#;jL_|a++ROPfJ?X0s)1M!0%M!ZA9#lVA
z%5(ZOo$mqOlle+X%`w3D5Z;7iVq@Rb8yS!xq#c0x@xf)mA|jx%x_ZZ+J$trp-~PbY
zcSq`t0xw7xuf!c{lkEW@>p>?Rot#ki_d#Sidi3ZS$dI*>;@+8=g7iB#M+9_1LZg$F
zm3<8`Mj%uaqi4^bKl<~>00gq^_KJwOcq|?y+6AqIq$H%P`|0T?_wV2T8wEkcdG;u%
zym<?2bCDb!$r}4S5PDi%9Oq6J33^@kzlUF*wmhMI2dxGb=8T%!^Qw~s@H$l|p>IvW
zRZIZ~Q&B<w@`RRo>(AG~g_@+IwN*;EkZEX-yX|=UL12Y1zdw7{C#vzZrGru{C@WVC
zHz$3>tt$k@{aiKo_xE?~d#-WGLgnJcGQ5g)R1i--fj;`|;gF-3%rC+gu^;_;|L@E|
zZ=LkoHA$}mQE_qo(VrjbPg^iK&knAG2oP*|Sn-^*vskuemp`6EH@@NaoRHzS{fmo>
z^d0Kz>Ru@6+S%@R@AAt9u$Wt;%##COzH}ZKh>~>C`}&N{!9BD<X+0IlC5KDW7l0eI
zWka&jWRp&?>obez3Q9_99qM=Qen_m&27(VZze6CWxg$z1IoCala_rgmJ~#KLK^~!P
z%=eCr@Qk(RU9~EF`%1z|b-1G-XSlAirba2dh?-GYfhvys>RSa`TH3^lyEFYt#*hnd
zuN6pBMl)0`EK<)?+(ue>jmv##gMx#DhgVgvURBC=SXr_wUn=va31e;q7BvZpmU7EK
zPr2=rkzrG`mh#ehN_`t681hao<yY^nTTK6{QeSA!vD7u|DrO!ZALpw68d<T6o}cRd
z`}dB+?-P!*$Uwt_aB9%#R2Zh~evs8;S%8U|S?#%lon1MWg#aF$T3>{q@dL0|0=s(V
z<m8;!x3#t|1^u0Z)Ov`Y|3{4QIM)lS*$biij@*QR>P{I$p(Hfm`j|)MZqRMbeq}1o
z3(70kMM>^qWwmUkx3GZ<?nBGN@3v%{MD}_5^h-nh#hO&eQIpYY%Uf2K=B^&kabIzM
z$}HZ}z}7;PUXa%_5Fna&l_gp57hcPy495vD_-)459LSNj?36$gJ{Ee){!d?TuP6A>
zYsFx$0}Gblzdh&KbMT<C8Y6i0f!1at!Nt*xI-n`N^zY!ad+6xySg}@QnzrtO7N62n
z&LFxoE&ua27h2S9+G^XOhE3!g%ba&BpvujQf1w8|C@P+Hcb7sl+p~ZF!w(<$V?8dX
z70th_deTdtb(X}-3JeYP^6}ZeanmNYW5+Jc?^jD|?kQU6EQ<}0G6|NhC@np=45FE9
zJG56vM`t%08dl!7mTlX&m*N5YLZuokHV4)xbV<DuC|=lbpRk^OkB|F5e!SIo?3Y@0
z?Y(sy_+1z8nn4Zx=%!Ckvh(bY$L-KvR#j!ERI)&ow3CLY_J#PwKY#vEvWINNaFeMO
zF2pAb8V8w0Knr^8>svO|m|$8s(bxA#MS5jFC@$BCJ#HE(XK}n>;`>iuiuunE>mXjr
zC@L!YEntCInYE>FqLG|^R+W#xKE*&H#Pc5G`H{2|f9(AZ2Q^+UuFZ_XX3vTeCj3r!
zeh?`%>1_$-koW(MKjqHMe7IFcywBjiF1LjnZApHmy2cwnCHWoG#RZ3$FBD7^pVHJA
zGKmGiQdahFN;kZiQ6WfZ1H!&OJ_h8>oSd+KRv4k6nWYyuG|MRl#i73|Ko#dAKlbra
zHH(GfAT$hpkDHHACA(I9qL72q#(iTfm{W24wrv8My5lDLP-1&ZBJ(sDC)}n!n~}_~
zjdjL)xuBwGH?LmgD|$FOYL+o90TDKxtdXL1nEaNMR*7PCj9h_-pZs!7^TW;iqk&(6
z-JY@-caC-b{Q{M1w)pB;0hH$Mn$uk&<~FD<i*v2IhSf*iir*?gh}gH2?>PXTK~7h`
zU3E`(8frYaQ1}fM<J!S_>KJSE0)q)NLczX#`B|I(hI@sD5(F9a=|kmXE07cv%&e%p
zKmp{-;dYa{uoA8srgB{HHa_*wYQmO#F){4EWaUej1YLHWi>?3a|2{vT0XtmPtnw!R
z(a@{T-#&%xJEHtEuDyuz-uB|xsoqMpxroqEu7Zw+mACnyCsp4-((s`%)6KW7>X}!G
zk}$Ssj#NHUkXQ~<V!*1IPevUu%;DLT!ws4!%sntPWZdG|G_S2qQ6iS$xwX|5+Ulb#
zd?yk<eE89m|2F*8gkH_7P~M8yO(l6Y3faT0s`mEZJk+fh(gMYH^3jqUyEnCs!Cr$*
zb|A;H%2o5$tv91#4|59&l1Thhog5J^+{|ynd;0p~6JIqjk1co=H2ll=rSxApsx3iX
z|5KoF|5FB`3sY|^LPGZI6xgTu%gxd$1{tj7W@_<@IgBU&^pA{;)XT$@%&6Q#6(BMF
zSm|3wLG*=ayRZ{CuFWM~Q?@AS<LYZp-j<MgY$Lb6V-<+~XLRQM)$h5lGWs)#vKL4S
z3#)(hJL|dmjd9~!uAG5^0YVMC9YDQkS}M-Ye(s}=t!?J$Jh^&P$J@-V3l!Ukh5?S&
zfcEA~8X6jHv10h|B$n4(9K|<(nmlQYHiPb`zRUrAWPS19{s=J!H6_ZzL|BIu)_JB`
zTf-lF7Z;b~q{8dje$nnrRf*Ce5NfJxd8uJ31W3+@n=vj#1+j|TU$SRAi>_a|x-^n*
zz6+`n6sUB=KT^8!N!8dAf_dikG3V1&Oy9kkRde<0*4t1vasnhKFffo19ea9v)oLQY
zhYQHesYz->WMVwV!=q%e0NLQ&x%Fc9qmEs|u=%jz&ri)7Vz;aNee_scUF`AYd%AiA
z9N-Cbfa(_fh`h72v)=-CZ1U+lwM@safA}<5sB3F0SiC%av2E#y@~hm5z{G+`Nf-6(
zRqRGjSVXxMF1V5%2abP&P-8HX_{O$z7NSzhc<dWf+dAQ;#l@!Uo0sO<it`}Xd8C+W
zF<v~Agst@`%d}MwIuuv6?dZ=v5PB6ZTqwm#zM`r5<n`-Az$@3MJl1+E1CsLV)vA7e
z)E9uf%|LT|VYvC?)YmQd>IWFs-5hF&4;QhD8J+Q4TWT$tT%M_Gb>uGqIP<Xyr`lhV
zdwr-$2(S(p{t4g)kLx}f0Lu9{u8~LU&fLEJc9c-5dp=gOlvM>ExdL3FPX=U3$rtOq
z`OfL_;92MFdi7$;xVU0cb0S0Vujkizo&GK^UN3y`jtMT-c{#an^C|spa;C4OJxJD8
zRswBnG3R|v;=5R@vRUN*?<|1F@|6Z7K_7}ckcQTfZcs?H`A%x;tlTjH{fZ_^*ZX$g
zI^mWWLYHfLcO^aNSkv?Nb{$!H`Pjn3_P1lTFmAW(V)~Hi2cG9Uav|v9K$OdPajaKC
z(NqPadYYqE8Z`NCOAks!k3<*V<VTARF<(m6w{PjWC#d4yzSXkRpY^=gZRmxT1*FUp
zd_=|K@<kseQR|ImUep;?V+?Q`9U-?(b$hdFb+ka&+RwTyj8MU)2w`gU3!lCUf950+
zkN6gdX#Ute?~7$TK0nxW;L#8CJbs69os5x=nw$JLfpXNd*u=cTw!?68guV5n*v$o|
z0>HU`6B^S)fMGodW(RTz7fxdSr@rGv(Sn(9DqeyfKxaDLfpG%cq33slsi~_K7Zwr%
zYne@*<Q*$VjceC_qXWyx%0ASTKZUPAlQ_xqNosjg=F68aM@hWbdOc~RJ^>k;+KHE{
z+E};!>@Q4AOpHUZR9IV?@Q8<JuC=1XlO;M}INF?NbH^C2xGK>>iXmrRgn-?EuC{-`
zY25Ao%KRZJRcN38__oJ*d4;sH3k!9A<yyry<=dJ1E|A;WG+#<MrO(3V=5f~!{Cx8n
z++$u|&GawBS!pOcTu|Y@{VB$?RV;9r2wj>Vb4l>e8gc4yHx^_t{uaXBu(rA&-8Mfy
z|N6U=F{k2Rt|(Ub#dRPqajG9Tvq?xWfla3y?%ccqcm)4oPhU}0wI1-je(Kbzz~|2m
z8~+}+DbA^H(aN##J8>ibwPwcc$=TV5MMaXZu_j!HMzNcCNu}jLs`IyN(u$Y1>sLH|
zKNXdf!~^AYgNMf&gRWRaNXRuFT`@*Ljp>Vbc-@xeQ|qMMmR`f4g#34uoBOP~`T>Z^
zwp|@d{qB~r3~>L5=s;mnQGKTIRS{$Wcw2v@2_K-P<vwwC)9)V}9R8GRbQFCTokn@d
z;v<=J#%y8uSJEf{a9yGN!&YDIad)P!n^uY#rv09AN`cxZr>Cb^{4$cY0~MF$<n*H3
zv+W6F#&dd(djg-)lm09Jm0%F3;)O@>2p$y{7Ut8_Y_q@`4+kf9o0#Y-VFe*e^WGC^
z1ScEhGQ@{DU86A^2@9<(6T_8vgZGJUJ?qBG%DTEMP-!YN@ZXyC`QJiW_}}m%g52C_
z?lBaDON^3pRMgtQAZi9FpxEUA)wCm~Sz2EzDqadnnY<!i+??cAlb<Lmzeh%VmzS5d
zCU}k=+m0VjnR8cg^oGXf18ZbA9i7GzH$N81eB~v!DMA}eh<49!q1?5%@8{-tUopw9
zhI+6ATTp9EA8`g~8Sk2<20wqszZe}IHCYKsJ>fr&KW}qU9b4$~kcSgy8gCBi|4)B^
z49a6xgOuc5ky|RmcCgGK#AP+uo0%Q=^!;8s$)2%{QkVAJ9<pSAk1@xKKBy!{t*Mbp
zWV6mfQEO}Ko0fGhZxw|gY&!kQHaDGdN`q{DD)+0|PQH^7DJoNr-xNZB;}>~h#;;qq
z?p<nXc7DpIPiJ64or$ggdDEcj@nmM;cc3S4z~*0ZDR-irygr{laLY09yqR5M<yE=#
z{SXA%;ZID?&7DVTs@(kH!-v0{-d`1FqWwo{tOlEewxroy0fXL5kiq0B#5Vy`<-dXq
z)L|uTNrWf}7K3b2#ygAa*Xhw5^n3SQ(Z}ui=H=!d5)=CslmF`75mSM}ug5|kK2uLh
z$-2Y&e(I)UpxR~{mYwjuGKPuZMrUVdmLR*yaeO@OMW}+Zl5OkDe3Nh2zUOFte?%LX
z-27d$D{7=EI0fHn(w}Tum-N(t$5rC7Sr4)qXL2k$465%!1!R+Ak<H%2z>t_*EJ^&P
zo?~Fm(T;+0-0WdI$(pYz|A>ecyM%JP;#ee|e}l2kN$yO478}cf;x^$@lAOE~MocPY
zU7*rVM&X?Zy}$%&`tiGT1V8#Ome4c5=b@o7nVFfz-e(VdtE)?!U~Bib`L9##Mn_zz
z*kNnQU%M7!b`YT*Y!vY{>?~M=F|S`!y?F70u-+j{jTn8ArD5?3+f-Us_VI4_xujo^
z7)#;bK7$AM@Zm<Jt72nfO652>ohPR<KSrU^{0@|7`HXHGU|F(q0Evq7bi?W&5DJ%A
z7y@_Glc*8g+tObb9guOSj<6x;B$;A12Xr@o!ct4t&e;|1I=yXl%t>9vU^_AbXx;z4
zAoy%i*M&FNdF~ioWVeAkr%#{8Z*APXIR@W@MR>Kj_G=)=N7ujBh}0EidtKl>?xn%|
zwWg*VaPO%_qe&KgKbqThQGSDiM&PS$*@P#L+ZgREA}0z!R%a6w+z+tfnUHYg)vH%`
zAprg9>q8Qa)?@W=g4$S;A;M<RTZyDVad9zp<8o*P&tAQH3bwD=VGmG`49rdxeZ<8+
zx3wKaw~$5V#K+ecpx^=U;xh6>xmLYiO7m5pKd*c9<_*GvNM+DL4<!u!urO~3w!KL9
z6b9++fYOi*P6=#Bpl--{Q+;8&$4IwyvP@{fT`#Mt8LI0dV;3%J!{xQ}-~#~iwsg7^
zI-dZJiKo!aMhM{=B&YH}^Uy8LV0Tmf9vkx`#HTumnWyzD&pEe&3030t5P`VFH8Sd*
zFNCt2|5EKP-@p!~Nl=Ndcz-p4-~5+g#u5a9O2jSrt$Ji1K6>;S@XnwOVFxA&$K9;1
zlj}foYKFZH6sRv%hw>h#qFxdLLuqsSe}uei*RI7a<~?ORt!x*8NJ9f$WL`B5jfj=M
zU4>3poyW3TImN}r%iZBW7?6K`xaGO=41<8)!)w>Cze`+$`C|;jByqm6y!<?BO>i5<
z=&9)1s+E-$1g)^+!vjPl3KSfvrl0K^Iy$eEr0y_YxaFuhnQrK~?wOvxKCN^Zykbo?
zZD`2pi!>udj3}tRqr;G*x{riR;d6KZnr<MF7oiQOy8%8bK6&z_WmJ#aL`oa}(oh+V
zbY^}UB9z=>V)>mO0S8IKE%aqO?%B=`UTG7i-t5^FbACsgFpJZlJ@4b<zT93rfH-H6
zdQz51A)wOtU8&@T#zw<{)we3q7FP=1zWaN634F-BB^r6G%{%EYPn1-%d(#aRGCYRx
ztq&N-m18Q%K%`wEZ^-4iI61LNNwFY@B-G~h`Q{ltJ$8hP{J(v>0NzEQQckSfr=cFp
zQ<WbLs`et++Ep0&R~kChf#wbOkayh4Ah;EMP$=IGdL`G0x`swC78V?${eYw0gM&mG
zcH_p4ukkF9y5o0|0Bb3S`e{&YaOaLb6o;=>Rrj%lH}7J448Tdxo78Ob5y?j)d&R?3
zJvz1f6Gz%+>p!&n_s1=^9oNbIVX)%R<))E6{!za&?)UHCl;KUAH($U7dRI9+7m-ZR
z^3KbX9C-Nn!r+N<70tD1H$dJ#Zg}SoPs?9C8^8GxK&kXvh#|BCA%rrHj;*yc+o%!`
z|6MWYPe37HZhS+7YOot_=i*F%P4MX0m}yg^kzfOIA_fC2B_PqxYi{Z@CXeMw2oIC8
z_K>LJQ1II9rB`OJ#?C=tjK?Yto<pMTFxGtV3K+2h3If+ins!dDkz`i4-Qwge?n8%O
z%iX~0Fh-|OpHqfpli3Bz8j+k~Kc<x-Wdlu|8{$PaA%q%Mih^^sE&0;%32SPkwO+(d
zom&-8nKgObT`oS#_-uLEwQY>h_Uae(_4T_Ufu(ms=P)GKqGBD1cGa_{V5b?7!yrj}
z9*J^XkGd?>$%mptybsH@t-E#!FFm{;89Ag&h8ngXMnz(!#+55p1NOW0(V?T<myCnS
ze5At1v^?rx=cne$mJx+@7ZHzy2r<0*q7ZT>sN;Lv#@?=%wJ;Yz8UnnOXFf6{gp+LP
z=$Na=SaDn{b1MLXT+WRGdukALeoc*}=F*Oi4kO8{2n!QY3?v#4v9qt|)ym}e_~9^w
z(3BY?{7!Y9uoPrrz|BtOx&8ih|3E;RK<`7h4KcQ!eQ2;xi@TCE_=4?-%tD^cz_IzU
zfq^3hlK3FJkzhS>ha@x_#2~QlkgyOGZ2DtQLlZzLotS5y#;*%&C14c~b{0`0qajp2
zi;D7xzZVxD&xW*VWdN)9=bM+?x8&6wpc1nkyg6;`Q{d#}R5kym*(6US{sr2QLdNYH
zgB&FFIu@ejVgpkW8i$feB73N?5^7R|qoPtB1>es;_J99es3lD|0A=b%vj_{Kk+b6+
zjPQhuXF9PSr<yf%yP-6f;tnpgsEJuK3g~T=drJStgz2q_m{=*y(BaBh2F561;oaa1
z8m>1fC{znY__ASxzHRY0g(`kJPN->W8du%@fPxKerXnW)lP={H0@QFZ_7ZEESA%Xi
zF?!23w8O!!k`lca03N?sA8)f>2zU|G|M`wkg2mfkEJ)Ep4d+ethyapV1+>%$cU8~Q
z=lv%n*WG-kaI$nqI+=R&231wnahH-0A3iAeUkFhz;fI1yeliq^VVz>vgDaE;dDy=&
z)byLbQ6m;5;QYr0ZP1r2=zTcaQom9(Z8M>N3=*&T1vJ1|8}=YhB9}fnKQFAs2SInf
z)-9_xTH4Z%G3#s)ClNUlcNizKXg`p3Cp3NlVE|ob6z~m*FO}Z~9hNmP;6(6Qt=BEy
z=^Z%mH?HU!7<(xeu^(?aHfKD0wmVrfL+;}%ofm|%{XQY;WPS?pwc||9=@hZjJ?EXA
zM9~nRJ$rV!mnhz75*zS-&t1Cov}y@PeF0w~AuN&yfS{)k1Jaf<N~c$09I1cCCWG#s
zEYfAwUAoB?pT?ClLC&fBLkMv@nZ=rV1mK;v<c^(@g-wi%kSwyaqgR4HD$YRjX5-_l
zGl)c4(-O*t!a@Lk^{o2q0)@lj>PhE`J3`EGanZ4j;$aaXpzk#rIS7Cd966zxXC2?B
zk%6`&04oJ}{ixJ49aeO3_5!Nc0-HV03P1i<T|MEV62#dS;flDR7oDzaeX<qRnTgIK
z!op*ca-~PA@OJAFWCpjhvPLtUK7k7@MMz~-Xms6g1Q#@P){BXYGZL9?#H;))JQlmX
ztp<OvEzd-+eL-qMNjVOo*;w5Y_r*%-)%)S$;hWuLKOo1SX;i-+dDe1phDoS9J@E9>
z?JAxg^o08oOrCIiIFwE3K=9AFqytdd>F(dZf4$K0W7XXzBS9kb?M1`tkBNdectYv=
zhYd0C0L)<?%qRnp?h{6GR#a;{d@=OUi4(gZ#tY^bLBsf&afb%o@h+~9NZkRxK-Dm6
zNj@lMH~bKR-EUH-fBt}ZzaK(l$&6%}sHmuj%lz?^Cr`qHd4OVb4n?W8l@1$II8UO+
z{OxKm+SF(Gi9{0&w!8B^{5~DYaUF7Ti68d|0B?b{V=d4G*S!rgL^WuF62xyee62a>
zECAmv+y>?cUXL<qC8?+UfQPD0opE+*X2vr*nw6b{qb9XLzjBvO?%mIS2w^ft#(Nh`
z(N|R`QGo|r(^#<^?%1yxNLj5$m~*173b-#jUW$@9=v>s9oR8u};O<0m^3Yl}a$$hV
zY1b`~ELj|k4H&*D8@Cx7{pg7k721|XkKmW;!=5{kGXujX1|XAAgTL>R*2fTor{@+L
ztqx_GEXyt_Kr!yxuyr1iMo@)pAxd&^aFh)fqyBt!qu7T$f3De~uK%U6!a1gO#H#=M
zK_)R<0tJ+Vv|3s>Z?%ChdJB2G>mP2e!_0vX;4zrsyL?{?-ab6I&v1aBA_H(pQnG_f
zKRpY7WCDQ7x8X>9AD`^Uee~!j<j<$t?nHMZSbck;$h7r@>)Of!VSDl$H*n4lHOe=;
zh9WzM$rbm%eRP~$TzW_e?mWR`#3@vFw~5We!vpxDb%EdDk+*ks>YGi=cABKLR++Tc
z*!1+hFtGod1&{s=QB5u3;%i4o$JYo<PEJkTS5ELOC=dsUxr`_}<WuaZ--Kb1Vs^83
zLHOlKv!DEJl&?rmJ_2B969#q1B*8tbva+(W4S-VTA+~q-_isk`4hakEMaCICuL3OR
z?o`kDh|bKUdZXQY_dWo60^Ptrm}nq^eWxwj39In#>ar8Q@%r*1XE%(2lwr<f*oEFf
zoX}9tQwtP|4E&#=hyAapdFg2r>)J<<e*Q5sN6_pa6@HAUd4-hDt`ZMvB+zwfIM}uC
zT3NjZg--bVC@d|_(v9YRG7{MTBA#WMG$;DY%g~&R65jh?T~PdAd4T_(Ix6upE|-B8
zJ%#W1{uqU9S#r?L11&|wZ7EaF4H0pK$ulbC{el4r-WP8<ORcS@Mhid`h>5}xHz-86
z@WWZ1G}|V2t1v1CVqUDLr)M~h(4|@xr3pRw{WM^Ct21kDZLNeMXEQx1?gF-X_r!#5
zk{kLdM*LW^er=Hy!N2f{jJ%N&-OyZK&nzhdjVy5mjTwN>B_(qV<~z8upm9T-`mcm7
zyN--17D0z)<K)Dj{r&t5?*?i@DGDutTaeXoLgiqU4q|Z`aCcZhm@<qb0AuBqm3Ki$
zs1tl&(I|IP1&!BSYC_0))@U=`3%~C3w=?FrS@4T{iC^sd^XCxqx6RGX>N&LZ^pBrD
z-3fJo@##~BJ<yE)LO=Y0l^T&b_x)5If9_Xy<kcRhD9;fB9F#lFxDOw8Lk)QM?%ju6
zKR}gNVs_uWVq@eUBAs?HQrR2f>ZD)vr*1#>lauk=Q|5d8b`5v5Tx|FBX#-kF=*MB;
zV{q_5Gxd%gG7-Y&W)kcsmk6DW2pGxBYhMf7H)p{6PmFGNgGpA_{8LULNN4Haz}vd}
z`V7@0VFi_v$pxWuvVNPOn(0+7%|mq4hRW9bUDWQ+e;2VMf^x)n*1Vb<b9x?V_6XTJ
zC2&R7!IqQ=vxBKuRfNA(dVUoKw5M(TC!BW?-*hnEB?mW~{RP>=gL!jfdCHP)KiSku
z2x_de?%p*_mdF|!_PI@jJ%9cNQm38tp}oq81tH0tcwJe|5eRGtq7~{iJ`8D~X_zZ)
zS>RWo0?<(tt#$e@V-yhk=cksVqex6lj6pS0U<i13IMHI15(@wh{nzi`zW}sexpCu$
zJO>y1MTuwKS^q=LEWMLRXB4?O1d}mj=?@&!(*1b&eq1TgBq!CmLd2qCZ-;$Xy>^Wg
zQ0bBha>_*$th%o2o<aZOrHGI8_FkRP&zx0*sM23|@{tYc7`o!k)#bkgI>KmzZdVx_
z``yyc;|P5s23ZDDx!#4cW&CM`!cov*0+TSFw!o}H%EX)9LRx;~lSG6IbH#I!(H-sW
zn0eTZ?LaDW{tS0YO&r+!%XDA00D||w7>@0!GP3P{{%@OQt*IPB=mmiq<aD4JZGyWf
z1L*?P`Hs3c03_%H4_p?M{cR9g5R-ci8FHXDQbtP)6O5iUm6njuc3=XM0AmC~MKQu)
z#yQMZfk58Ma7VruQC$bLi|vq=Ak?4H)I5j?<lV7fLQIm*`$6S}M*@aPu{-!gvCpBW
zol{la3k~`yW>5E`UWBD&{J%r21}^}-dJ(P_$aBM05P4K8D)V2NpP^;)TmSh=>2L1z
zSabXSQs|<UG*S;wffe*7e}6{W@3~nUC^aUZ^yh|~F~)luHoe(DGf;)o-*~uk65``;
z;Z;!GgLv{A^s)T$9*(Ch9i7-B;`XCNRB{Rf7>O)p9k?X{`_b#r%63UQ&)$OwP?a9U
zNeq@DBvqNHmPE`QXDq!uX}I0h)%B?b4-ZcnN;<E$G~7NUpoQ~L=EtDC70uWn3MOFi
z^_jSIbqCivDx3cAfkkS1Q$6Kq%jxEVjwY#!QVKa2;6f)_@Q8?9Z#E<Y!B=IUB9PM(
zs9O+~0_Oc5UenPrOK&H%4a(Hq+;i}o_adaLkU_ZT#yhm<273GZxyYSf)Qp@|4Vd<d
zLoiECKNoYff{+U{);gly<~6xbpKgs=DYhF4&D1O3UR714Xy<{192dF3eI;LS1p~9y
z{>H$fLf;S&=!kJURo>Q?*SV&yROFcSkeV1Xflqwx_-PDcCsBsUIeZ8o?t<kKZ1_i6
zf6Gq#dzk2um}|OvAm9x&EWGISA&E|8=bpkRphxO4@moQDz8?BS_(?;b#A&6Mf}erc
zfn7O1e%e7r#i~r43lRMo6GAZpVxpqAK!q0}yJ904U14+x{x0HtxEd)KLQE}LnU64g
zLFmCo^`$*h&~TBK=5%?<k$_opOrqzQ?934r5g`osa6|{2&5=aVLm1+AE2OE#I*(OP
z_mw5?Q0)?TNm%!$L%U5;h8aO*vGI~kyt~y-)02+1yV*6+`^kCYYkbMkr!85~kZG$A
zice2ZO??J8`VptN&zExgHckdW<H}5=2B-ENF%e{Noo~AXWF??^8<}5(&oXjp^pP_3
zhAyu@rRT@wrO|;T?sn90AH#gxp$8Bsw42;Pn@yHor~fEoD({fAG%Fe#BC_YMUk|UB
ze%DHJsrmC|${d54cyYvV<N5QOFh#nXmiDlVPhzvlRHw^CH^zpts#x`rL3Us4-lCIl
zyA8o{3<Sww6&jA90(65YltZzy?Ggs=heoOUu?y-KGKYj6a~D%!n7YijA3LUf=hMA)
z^ir-OPzK-rEEXWrf|IJaogJ{e16TBEwr>599}R)62cLM}+IrfJjF*6$)aaGOCtrFg
zsF+;?T?kVF-QXZXoydBX!sGKH=J4zaaMM6Mt_vLww@1rYgw1~5L*D2OZoGnm#Uhez
z*7m@g#rdm22mbmXbsB{9^HYKpM9eBt2gVB$5A56LiEQcD&c=lbM$79=q%5nEDufp%
z@u&8`PRObfZMABe7&jhmmg`OvT>raS?s*z6_**um!=Z~9(agFM#L0c+h`dQ#dc3Co
zOXvh<5vXqI=D*0~i~p=JLQNCa)cq%<*MMG6DCEKB@a4F(>!Cg{*a<TDfwvsd9*A`4
z3|tuvNu5^=;)Jgep3LWvCBVaTajL7fJ;`sOc=4uinsB21#FerlRF@Q!wfHd7IZQ}u
z+OZ&N*fs{X8)SYJ6Im!s&&~Dm@YQS9kS{tjuV4TiCl??%VL##81I5d^>-~HOBKE|<
z<F{;78vEs?k+gKf5GmcOoDe0pd4*j7*h?o{cH|pckx}mnUi3XRLGX?(W%CtLh~nuE
z*@T*X^yt^okL%80te+r;<<_Wz&pKEb7_@rt#IFGUkwc_d<J~X4uRd~s;S@gAdaZH8
znJZTU%gv8)alMq>LL|E@uidyooCv_3t<O)tznWXp@aDHC`;M5Yj(4tH%Rl@F$2ue~
zDP6o6++i`~ezoJK<9j0VQobdd??%3X73GLZlwg6sazgNp<{l7~^nl|XW2gz~<mV5_
zA)t&AAv3eHf~FECl$4bJmfS-y%f~Qel`ib+_3K7fH?Dqv;qdXca`gfI%fnsmqMX`u
zrrhG<ebPL6lvV|1ByBM4_h8f>RvRQUUviE?aWW1Nn7?=q8ju=axZ^}%O;L#d<x5l3
z)2eF9L;mwFWjBof<@?lUj)w3?c^Qq;leWwGF~mDw)xU6IJ0d#j@*GRP7|8U{L9z=o
z-|A+SwYAD0bua|l*1O@JJd4v8uD<aOmH4`R%mgMbsU16#(`6kzZPN1asut~o5P?Di
zko=plyL|eZLf%tjkTtM_g^@AoXT|HYo<9&92`10X%n-_0M&;&Zp>F0qd){0>Att6}
zPZ{fbU~%=AWxQLy($Gxci2S(8+oYtV(JPVoZ#9}98w!~>+;n^)Xz+D6Iw8;M9%v=&
zx?E?JnEevpLSn48h!)V_OiWu^7a$08fRA*XT!G|~K{M+&h&F};p@h9q$>MS;+s{R%
zDu7v>tKivdAOMobid~zqHxwqQQEex!qthCi`>-Wmwr3mFv$rJ(HC<jhR$OuS<Na<s
zy8qA2Yhu6h?eF1lmJ*{<AUIJY+QyaR;^>G8-F3(>7g|QZ^23IYaa>A=n~_xUv&6&0
z6=^HEwS6e7U!VyXj$oRVLsGJ+!Ua3FCF0bbT?dX|ziHdaCnm;#(0;tx3OZB!f<kjg
zZImRB_8NBMXV^tLy^^?+O*;<UBu<{NYrdRs><ocPeWCt`Q2rf8H&<7a;~U&4ySv(j
zM(>xX3#a%Ue%*|%RZ2v{p-BZBA{+P^{vGGx!`)cP?~;>cFJImb?2^1#vPPszI|dUC
zh+w{$t&!iB441j#c*j2*J`LLVQGKu-QoZy_<jVZt3Xhc$UB}hG1Evz;9S~y6MK4kL
z=;|cf4l@Aa;lghM=wNE_3>pB+xnb~JDb=A<r>@!)iuKe_{eW)7u`#!q^wAvfMg0b|
zC4ZBI?Cc>)vJnrW#gP~UQu9+ByO0=K2sCn>boQD*94&}th7|P~vsfW0!hDDXgpYb4
zBbn7OfdYFJ39E9q602UhlvzH-Z*{_6US3-5^N1P@uS{GUFTcZRT2JV5CQV1%v}uPg
zSXi9EB*zsJ$)eHz4vbqwF0VJsA+=m?cUZ9$;WLM>D`g#>j=j`L8il#7(Bvm`N>-Gt
zq9*Y6aFk1A`3lrUO&vNa#QRh71(wbt9x8)21*9B+VBzU^htESMet;R6wm<fMa2pFD
zw=E(*o?<ZqUCW2YosmOvFH8w9jB1oyL*P_<RiZiq5v>wC=33XA5{#?*o-;z93m11t
zZJQgv%}@UUC@1OHZsxL!k6<_z!rn6KDwc$)0B|lajAnEW#qlEqGhZ^{!m+rF3KTAM
zx(GA2;=v2Rb8XwP!-2qd#6&KbTGBCH5{VF4yYmL3GDHdyY?(yYWn*VAO^$tmuRvpX
zL3<M+H8wpxz5Qti5~n_IVf_tcn3d0b>%C~6nVAiVIJg7_G7j0ck_LF@DXVXv|9UlZ
za(=iyFC+7I4Ffm2hLtYltau36#^_`|GZG8MP3|tcsovwh|M6`Rf?(j)CXUHy&%I*O
zk)Ksg85<ovt`sGqxdfY=wj|~2O@0Nq$Dzd7{@gDOl$v0uaE8`h(CXqapnDMrwdiW)
zuem~t5;QSE3_uKFhA!I7#lhurf1zMdoa??i8tLVGATFx!uY2zP?~Vup2Ejkr$AAKP
zGv||Gj)4)R54nblc9D%t;StFpaZJo?J32d;;+P!RSPsRu^VmTL__}+5R9!OqHFVt`
z5%EsPy5?qDx24(pB_$qye342QE+jtZi=>7D`Zqo!RBDx@^$vzVeq@_#L;ed12|2{W
zvjbvs{)B~_O-arbKX?~JbO13HBMgNjTJ$v3bWnp4&eZ;}*X!>N-WQ+lNg5~hGdWF_
zhy9{U6W&|D0V>45o|9wf2of<6$K44PKUgJE|CLkN%L?T{T~kw2-&`;zA5{EbKQ8Kj
zHodi$R^&qCpD3{`y{aipcZkhu-yu4g^?gC~!y%iG?mod0ueXEY!b@%=Zhn8y3%o|!
zC$k$5By4E0{}7w0Y}FK#ovkNswC+~Wk5FGFpG@ob;Ub%2w{5$LP_w`pVbLz>CC})P
zj*M1=1-6OSORIP5r0Sf~znpeau4Q>lN5!#+k4lc6MD<j*kc$3^<2wGmQ|oVX_H6p3
z@MXunz<>WAsJf*Y+Bc$z>4$4quU^$n-s0sl*Q}wYrRChwivVZ~!Wrxz*u%N7{&#to
zwKgW|Ex#~>eTfu;>|jzUHQ(XG<-NTa=UCp!mMmhn@jvof{}Ud^f6!Ny%Q~-n>~3yu
zevFTAVI*&Dc|clHilnao7L~1Tz%ilmJ(R<HE#9ld>!2Mve<<J1&W@XdLk7fYY|gDF
zQd|Rv6BQM+(x>^9$t~^&D#4*sQf;k32NVNoe%7f6y5mTEle#KVm_-d9c0HzBJ_*wl
zzcn{3g?v#O260FoTP&Eq>~KHGfAdTlYr|Z5fp|NlMw$||!ZkQ9!On)`sN|7SZd#t3
zsfVBcwJPg=D7kzZWVLT#VBuRk(k%~ntYaf>{n}cxI!AgDCL^~I*RzoTlnmnbX4eOp
zgw4Vk7#V5zl?gRYGJ5-Mg|)6<>fUJza*bc*AZ@*#OEV>6Aj>aeCi@9L`5d?#!(q<`
zzExFSF*i?P+9gk~?&bavpLzoA7IS+$t92IIZD_jBU4BgWU~j67<gSBzj_+dJ6VlS6
zZfIbD14}NRt%D?+hC2MnN^MXcpL|#6L>SIRVr*c7WxK7><Ok~>zKy&gGE&}UpT{hB
z@9lZ?ninu0T<@%3q?lbIfFx}{?Hg&_+Aqk65RmpaT~}`6xb|fS+cvIq$_W{*QXEfK
zJssIN9ontD(JwbJ;L@1jg3dMg{LtOa_mDH3Lbi!v={{4qJN?Wq<~$Y2(npSL2KU>$
zq9{!qI*;kj9r(I3_LrBV6e1(r@lK}!W~Emm_4A_b2`slDGVW%9WF|FXS?4yqNBPT+
zTfHN{cyXu7Ov3DDqvQYl^=-8)3&DaQ`ncp{4b^9!VbR#)RGJZ%wLifsf)HYgb$6Hc
zVB)9=O_{yusj2LZLY%f04<Mnyx8YVxef=9Se*E9FVlMM$abb6vYn}eXE6o4=meOM{
zqS{A$;D8B^XUeWaZMlMj?wZrGvxT#=vlHXv&tr3lIZemq7Zgy8%0KyeJ8pUd`T#qS
z+Sore^+a=1)4TL^RhW3cKI(fTWWN6`?fTPFyBIk*5k|qW9ZBOQYwjyUDjp3Gs%9Ib
zSFd5Dz<0KRieq=$(LfmBAmi6CHBbzWc3`NF>Pmp)`GaBA@vxn*)Jaj2_4Gs%w-mCG
zu9!f|a7MlCvlZhg+JETSl!gl*aYhhy58Jttef#$HS?7fW2gf2pc2q#1KFz6kJ}b=_
zU*`NLP~O?}{k@NmHTVd-fq*#y<X6fG1Skhp0cbgIEy9#-BZ+an#CPw0KpK?Mx{BEk
zh<JjwgO}su;*Ls5k!oscOiWA~pc2ht48ltgc?dl{A{~Z<sfhZ)%*+hWI0W)t0zu*b
z<W)XBhye`=oN>k@AaDhcu?d5FO`V;o#M8~m8G$UY1cYY?(e#6>3MTQ69X)y(m=a4K
z_tJzF^!6<m&ec-IaL4b{RYmt4=RS^)Ta0xSXuv>iW_D<Q)3rLEBjr(s&LM}5^bW!O
z*F>mT3*LJRl68VG9moSwB39>Mt2aX&TlQB#G|hd|lhv`q9kDU<j(oKsh~jT2fPL~Y
zHX`9(5^6|9j4;s9j9Xp#wML8G*R+8~;x*=Rc|=9qCajc{l$<e^fw3F`w-N2G?MBf^
z;^NG!!11C5l2W9Y8tmcurFKxlv91!`lI1?WEEJ4;^BWk|L-H^M#jAK_Wv*z#t!~~0
zr}up5vcYj<n&1F4fAf#u$Zwhuy{f93+y#KB4sq>M+8Q2PGV98qd<zEhT*r^#?b@ey
zt&BT_a{e<@ZfcsOOPLrRP6wGAZCwy`oV;m@2;e%8fT)er1G=iJdTz7z@@K~eP_J81
zJ~>uTF%4=U+IXEbz@NxTI_vuA6e%bf_%%AM8rFATs}7F++qUG9tU@k!D|n8v*#bU~
zo-CYHlmx2w+6`Bw-I$<~)3HEI2*R9xIL%8ci*6Mb7H(N;PIIT7E{1}V0<He>8q>Zs
z{*i#GAOmIz#~=2!QBf?Wi<8~iHtjgvO&zmeKUa_Az?tI-jR$QDhet*>>}sWjNSr;f
zip0;G*Zp%9(bBQdh`r4{EON+8J#;1XQ_+UOz2%~5M~Up4<l6EKDMy@X7xwTl#x5sw
zC2@Emgp41^mlW34H#WxN`UZis^IiX@Tie*oJx_Bzk2w~W=(Tu0H$@!W)e238;pgd@
zH87M-LdrqY)6=$;v7Dsj<nO)nh;KPE;|z)vQ0EcMSeclct6>Ep??ZGGFhCwCLciyT
zhz!!BIc~}_GM@GmohcO+a)*x`Ar3tJR#P*#sDri(AK~5W*JtMD=GLWc**j4f^Cz+u
z_N-g6FsjL%*lN|0=uBo8<S>xOo5;69(4qwsWW9E;$Ya^iC>fi?(^N)AM>7c`KJw&9
z)CYI;Qf+E#YN<-Z*4ujBv1$tajV&zZos<dM(1M1Yg6PF(6Jz6#$E#N(HMF!|HrLm8
z+eITprfzKfo_60pn;H@CvRlL;&U||>ra|{lMc?r#j+u(2o7&!!bh_tom5HfoM&!aH
zI?|-b2_o4rIyx(c08H%#C8fFiE+~8>n2;t_5;;GX%D;?h1T}=?)cRCOLE!^rYGd(n
z$Zd1EU3KpB_&Eu@uysZKcmwKIRv*V~qMV1h$#$ul85Q=+I2)qS->#se1&;-bbZmr&
zp|_lN{>mSGLpIqme~2p}0M48#Fd?Zj<wik^{WzB42n5Kvd^d30>>0ALnORd<$+98V
z(2W^Kd3o<pK3xl!zw%IKDplxr(*ItmoJS0&5P8%UQWp{=sifNOCuQ9~aE)Vjlr5~T
zuXUGu=&~t`f9pKP=fwE%;t-ix%(<@P>3BYHe%_uu*12BK!BQZHoI+@_1I5JTK|vR?
z579P`(_riGv$C3dvrz)C6BMt=-%czNXBJJ1)J8zhQd3ult*-VGnl!>XEQ>MHh7jDd
zZ3^dqnMzGmx1c&A(&E%1b*J{#>(<t2t5PYS#r#jxc$X}!(U8qs1qSM(Tfxm;$Hh;b
zYP6@Iip~KdH4jKcw=_4$W0n1sN=QvrMvEuqAXcGESHh$5HGoO9p*I{cDHywo8f%pH
zJR2OfscvVN<<BCS2rf8;EiybbH1${VXHj-gNJPZQT;u-e(3dYw3TP4<t7^D6$QH(Y
z#oVHKPBcPg`ISt&yvIbss%fYF_8ms~=lCsvXD-#Hw6yftKS$iO5YxO%Qvn{Mtzs5<
zafo^-A?i`EbR7c=B&R-F$(iq`mY~D;+9jtsSnQxX(FAFEMa%m%62(|r)PH-gtWbOV
zUD3mt4pPz|R-kO(SO~A~8x9s%S=^p<B&vl{asmfse7DZT$)IS%8->)DCKfWB=S+$y
z;{_<+S#BiV!K9UTg~u8pxntR6SE&K4tD=(Ke50wXGZoP^QWTrPe)DnA^BZt$c9<;F
z;=7mH>fGOMVR1YZ=X8$t`v|#fPukI`-<~wTEG7UBV71x><YZ^hSLN;1btswql72k8
zZksSE$QH6aLg?mN7?uVjrEQ30gw8NA6=`W>WAnpiGkbVzPD$@8js4frCb%wz-=w@k
za5!K<_IR}_o=8JOgM`1Sk<o`D>n>fylxUp%J8x=+652Z?Gvgnx&l2ji)Iwezx3sh_
znn^RdZpe!lL#K9Op!lVvOE#<+_nO|~TA3nvUALUf;S3#|W?G@1^7(+Ut0Z8pRP>#N
zSHz(Z4jp!3mF9rm4Jz)_!OuFo!(%cst`g@nynC11KZQ^}p(B9%e}KsL+in@hyQTTI
z;xU;Q)S<Ht<5-_v?)6(4y#4kFT6Rjnyr>)a>kl1kkj7)tvvmyJjFbb=V1&kgHcfzy
zt@~y5u1iEeM=5@XV<YBo1SJIbgT$lL2e%4zE%%?6(IV3oPi~@S5Ih<uy2R7dTuA_w
zx_=ZyV+2{VTi0|x!X+g0)a#=Dk!K`TB<HSuDyhNNQbj34m2z`GB)S)J(;b%$XH+nz
zYbMjw*5=fqA8Ob}`Xdfv;Wi-Nwath*2x&G!;<yUHipw~E@Kf*STl#4VDk>%YrRBJJ
zT=_;<8W%-poCfOFxjz5Ro3n@pG{Bvev{?IoM=BQTlpBTI1wr6N>%JlMQS8Rh;we8%
zBaD1rS@FM#VKoLR*Ib;OmxyHZN*|v`J;0w?<#Ek)SxDt``70jMrMBg;Hd9?Dif|g%
z6fpHM7!i>7W*FqX%ibV@;Wn3~+YG{CSIwLAGnKY@Pa5f2Xy}2^9YmNfal(C}z+Do@
zWXv7o=B~#O{-_lN^an>DoM>)sMS%DJ>g8-)nhK*ZZbk?}!xe&-xgabu3M-Q`o%WrY
z3XP0zH&H4yr<<*LXQFKw)nFJzps><DaLDGI87|s%B}FD$p)MWGj}4V<Tbrm&+tSwa
z;y+LX{Q(=}-uLW%&U4Q5OT#^5ESq_P?&KC$?{HJOYR(B-mqv2iB6eStLv1o%8w{zU
zPXrKXwNgZpp#((<bv@nPu%OKyWA$HF3#F-X@$qn&N~u@ZO_o-SDFzTBMHEHD7recP
z<bqa9Y^Tqzd>%V;JWPe;?fz!kFfiS_NvTxENq<#91KH*ytr)g~-YYNJu}6co;uths
zh@r5`z5#=gU$^hdYu7wh&uO5ZD5F?ynL?tha7cR=k2bsW=W+XcAOHEog5u&qjDN`R
z(lF|gbwz>Wh#h&atA|;X>h*dfI4qT1jC?$1@LJ3v+pCgM`vrWyd0kfGWB7IsOcPn<
z_0q)&jxsi{DyQ6`l1)sdW@J=@e9W0m?AIW{O5V`5mnZ$xAs+~k6?LJ8V4bX#Fbq)`
zeK5q<&eBV49RG2q{WOn94g~BX2j0T>a*I3#1A$D40Sl_(FdoGz=xZdE7F;nNC`r;S
z2u<D^{7I;sPUlnYwh;nIg=(iwp@YNC+9V{s`ys)-_ic;Eq&R26V++vKUQBzOx^~Y%
z<-%wq_P&3Va(mqj0nz#a_|%cg>gae-a<enfHy3e-i@n3;R_skSKm+ciHu#Oyn42M=
z!5%@OP<SRiT|rS4z}}61zn^Pbl_nBB+mcnU#g%TLcVh2P-PDv0QBNcfx@7=g*?o0u
zf_;(=Cd1sv1H~<yhmrNgR}@W!$L7iab=oVMJV8_sV*l%L#t2!2!)bvGKL4WRAcTrW
zt?tJS8#<(5HnMI5!vnJA*4o}^v^%gmOU2@^n8A}C0=WvE)oOi|ERVX3AL5_j1i~Gc
zDyXoTU^_3aOL6u8nO1+_QvRE9w(D+AXCMI^up<IVL+uJX55tB?M>PFz*{ummk!k+O
c`f3QQ<aS-CPPsuF!Qc-{J(_t$cPO{&H>?fm1^@s6

literal 19900
zcmdtK2UL@5w<nGr4=7?oK!bo3r9_G#-GU&}6qGI~y$6wAf(n8PNCy!Ck)|M3Ae2B9
zP)aD$O9&C^y@w97U(fyK|NYmRxp!u*J8QnV=d5!)hmhng&)(0k?f2zvHN``JvHV3t
zLvu*!ro09X%`P+z&0Z$j{qU3Q=4k@_vENu(QJ!W8`TKo&dISy4SsEqzt9RYrO_4pF
zjxTTR%&%~>I&$v1R($U4?xXCC$wztFPuL!c;S}F>Y}n<($E+96KR8N6qEFLls0Y5d
zal_UoiNlfac-Z432VdMcdy4)zZRGC#d44JE{C?+bySWxOk`;Yu%S685rm*3ogYVX=
zwhTr@*urBLW8-6WL{;1N(zNHx@pp65&|r=j(nQL;?Yez*VBZxbKPDP3#{ba|Ob=0M
z>^m~G(IpN8*Kz}P{o=kJa5!q17<0h$l-v==0q23reG)#~UVmOXapJ@`RHdDc*zW(}
zQ~%@d{Xg;>$jRKjckdooUS3|Q^E<Wl%cl=Nx_sIgNBVC5<z=IJM1H>T+WKV9@X~mj
z#_ii@0~pv&22|Zq=RCl&zmo6CXRS<~EUlNXUq`R5x>=D+3yB^b9UTiTifnbSPVk*W
zqjM4z4}N@UW<vWSI5;ZLq^AD$8L4w}az-92GiRpQcd>`e{UucHxe^Wk)NigWemBnB
z>v{P~EeHPEZ8x{#hIlFM&MbW%Ej~WauO}+`{DwzH>g%KUd}Zj*o?|rJCG0qKGf#)#
zI8Wys2gfx%oV~!OPoIQcrgU_RY-8`<z58auqObUouw8EfO~}dHAvfAY1-gBUbvLNf
z!|-L_Q2y-6`dkXHPoIwy$GGUDjveD{mfiTgNyQPOK7QmSd-qJcC@CvP{QhkmQqxmp
zr$0_15SFG%+QP!Z=Y)jr96fsUoUpKlpdp@jczD=xb?#vukr<&ayCx4m8k?7GxCL*1
zH26blXM1b-wbV>#w1ECAxTaq$n(NaOu^0HXKF>{@x*`AJ!w1u^Pj(4E_)3F(o#6e_
zn_gacQM=x|_I<_dQBhGEy1M7Jvke&I8o5WC;*sN#a9>C{e&R&kC(U$x>3y!q5%Ew*
zM@Q2t|3lwUKN_ilCyjdxZPH)fti{01Ev*qUuBO%8N=Z+*;o{=*U(C1exDfKgwOMAx
zl21FkJ==(2?=h2^nOS%v@TAqtmoL+(LQZ41$=eC;Bh0Gu1x0ow>o_qx;XVS{zH6j0
zZnWi*m{qHM+$Ec6^DjLrZ0USx!lq+wBZk*bLPA2e5koV@iO2VCo-pM2uslQ7|3aEB
z-I6$cSU^IeCl?*~{ex<v?ES}=PaE#AMn*+h$!*Vv99l8s#(OMJq{7*{7wn`e$Gm-u
zD=xll*Oh(WkZrWF!C%|E2|hpRl%c`mU%L(*J{>Af@zTV-?y!z}_wF=V^5MgW`R0u*
zx&@YD>J*J4d$OL(bgx;l949j~bN=H|HJtcpsfbzqab_{wHxJ9<^Xt`p)-^E!qj|UE
zC<^HAdH2j8ZX;YAdT^ozy}f#5S)4sNgH-eTd*IfZdeS{@ZJY7&al=ISwco4rgQYdS
zS^DCHndXLAZSQ6{u6OR=vXD|)AflZo$H?wGwVWZff#`~lBy8i`ORYIZa#o$6W91bU
ziCZfaG%*Iv)Y#}MCL|=3|FFvceV<MA+qV~DV`J@!ib1Caiar+<+)ePDkto@iDdXbi
zR)8`)ckbNHTen`(vx=#hHzx>?itHjLCal{Att=NwXZK3UVmfr1@$&MX!%^Dwqdc;*
z+eI!X18;0hWac=Lyj|z}I;c}$nkS0lFFt&xZea*lxi^Seq(CiR(kd)0Ob<SpN+n@f
zi{eBqUXz4!3`$z<%CKAOv2SNzyr8xF`TlC}0XDDqaNo#w<qwb>o#Zxh!_B>99|xXW
z^{we;x%*`<<yrS(p*xc{mvekB^{qgq_Z7)K`t3&)Gp{6?f3aHW_sk(u6?PRp8dSgy
zlJm_1NiOkHuGC3_bMA<EWYKMUbaXI#bwhAw_KX0lv|Fxy-1k5xR%)5YvNg_(T<RPF
zAIU-;sQ;|{WoWD=S&$f1kzz@yC@1A-YUh*+cy29srK-k>jqJC)#@mcBQE+>H1jAO{
z)YRmnuVrW`sQCJ{(9|GhXB*eG;V{~os+@29{V&`wk9tDnWLHj^00!<5DZedUy-dJ!
zsYNk0B}IOFt=Xp@D#`5Ivwgz$eV4~uk^`!%tKYD0s$O%(vPxwi_eYAFrSp+<*jjek
zjH5o`UZ$2k`jjWfrAr-yV(#wlrLI{vZ@I<d8mD!``;XBhH_m-=q(p#$jxJ(p$w^6F
zonV&;hZfc>w__lN>E|zVofUhRcsri*<F!c6k=CT^2dX5YYUNgEYin;-iI=WlH#YuI
zV+Ys5TxS%H&7Qc_B)_=0Xa=nT6;fxMSL?k(=Fv=7i-LQCGzin7+A#acfQ3yl7e8%U
z_pzh^CnEfXDeG**yDwfjJnm`U^cTi?yp^j;vI4*DvAH_$IMIGShZMLu``NlJP1U~n
z_ceDo+e$w9b38oC%?UEjlf|VBQwzhyr!1`u^78U-eQ8BSow;iF@5hT*F^27so1?<Q
z*nL*3*rZD>h*6Zy5pls;KDyyGyD8e7BSI!W=&RU%O_OjGs;-?~Rx_da0^H;bl{hig
z@Q4Vr-_<V{1_IdXd%jqdmmXejRYsGkx+aN~+duYg90`qz;)>u=@5~x>k~ACq5iCf2
z6&RT6xiSO)@#<Ay^rdsKxbq!{>t~JnPeDhAixAZmFQp7u2l=f(P4?C;M^a4{gWS}@
z;}LODM#HPruX`A3(^TV9lp?rGmto6sC4FC?9U2}k+1htYRW)j8$P{PCn(_86i-gly
z=xIr(v5jUZ%FlX*w-iE9)A_2`u05;%@uS}*?fa%bcXaU8qpYlau##f7Hda{sZ(HF=
zW;1;y_C%Z=X{3mspC75=>E%-vP47N=DnGjg`|jp^h);Xg;DqeOKS#9u)w=K&+-IlB
zPSqbl%$&q%fr!VWO?4k`1j3fnn!A<bvt>gT51Q%DyS}1OH=5|vpRBFqI^BEkq@V%s
z_wU~q$|ze+gWA3A0eyW2Yb44xf>JI?+m!2upk(@@CF0d$)FniJtlis1{6Vv}LD`v!
z!D{6g3MfbOw`3(-y={mUQIy?YHHjI9HWJOR%b_kk@3#r%5beD|t;12Ud@g18^$IN6
zVY_gIjpU}edD`X>k6c(#5IrGCCX|Ddb2zUC6;XroRI#?UHibo#Z_$jBfPU~!+I?Yz
zbj#YthD%UTt<+`uLP<%<LZ;DHJrqI+%6V<t!sMONi6e&Wwy>WGKdjuV=?u;If^v*t
z6fEb`tccjyPaAKU?|rYXu6=*i*A&{5>~iwT=0lSN%I)v@LVL-@kFKBO*Ex$E^W}K8
zwrNARfql9JtT&Bfr^6X37nOdGNddBC1ICgae@z98tuQq+tAkpQuF(FJovjA9Q~+JE
z`e9~I^J>zE4;;+GrY{Cm3;+K0E6&b^(0~-yS7y=P>q|mj$r&isHq*LruJD8e!P$z9
z?DqC{w~l52gl1do%fjXjr*a6Ybk49U7hrc#hU@fB%C+e(>xj6MLnGgliHV8o0f!e=
z1EDru*?_o?!6u)YnyO)7zz@qK?Af#3=lS`U#+l``)e@x7M{ui#Ls=-hxD+AB9D-fq
zv?3K{6<D@h`#m`LeZb)+9A39=(QM7*rLj9bWv)4=rCb`)6<v2v!+MW|v$|I5G|qAU
z{Q2<kaE|+>PGPXvl?@E~R!FWsxV+L*Jwc-i{&cnY@WGl8BGe7iRR9f^_xfB<G&(V-
zXtC?mU*F{<iTR9A!fDYow>a=Lyf5c=zdu(=_h0V^`TIR-a2I%akEFLTPd)sq40k9h
zR>Y#8>TYP5&<Aby>M6!ec5ZI&?(4;>wa>LRG&FJpXxLz{F5bl3k%}B^&bYQOTK){p
zv)%h&^9ucE-Tpe3sa<E6?w2^;klk9+)O2P*Nslco!VW>4AFMe<3J3@we!P<`&BDgQ
z6PRYDmwVvp&k={q=gyy3x^Usb?uwkJdk)l7r%8J%*myB=M1@nYUc3l>_wJoXPZt1%
zNi(_IYy<l=p^+Y&^OnDHBhS>&%8yB9oP^6Mz))o0Z=8Y(7q#i+f4_01>Z;qf$lToA
zC+P8s!KuTC4lqqU?A12%=92NG(%+=a&(CY<>0M}PY1uW-cIYs?XsDpzpOTuoFJ<;l
zvI3*Q9PPlq5b5pJ!Cm7qT%6M1^BGYR-s=S^DJiRk&01M{N=iyEb+Y<I4z=lsRf<3b
z&Tly9v?MDYGnhLC?_BeISrg9K|DcC9Bq}oUX29VyLqCF9WmlX><6Un(dIW6eq+Aev
zFfCERO;AEYcbz&t8B_7&^%-S{Z+w1C_}@9HC}jn=Kr>Ff@s}Rt`aqj4pWze`P=(!b
z^YR5z4t~~w_>%&9wO^m^@t!b&%^u5gh=sF~&T!GJ&LU68ZR?QNHMegzA+qZqfr7MW
z76C48{iMkw7eX6+aA02!(skzss>-d2&plRO%{^1cA55B?n|mU|`SOP^{O|rlEawFD
zi<UmBLX#@6YU94bdCdeaDo;P2l0dgScc79^cF}aa)^lP2?Z@ObQnxYft}e6reC}DZ
zv^znDB3x<YCQXaIqoczM*uB)*4PJav<NJ$E#$)NC7eQwBQ-B<!;T&tSCr(GWG7+vK
zW~m?33?FB)p}KW$y1zV5oaSpJz|z{~$u7(WJ}pf_5;)<`o@b6~?=G`Hdi2PDS><^Q
z**<1*w0T3CSJV(VtH%BNXQm!r+2!o)oHp^O@n5x>+2kHLLJO1*@I|J6&A8L6s(~W`
zRqU`A+ls|?+lrm%<BP~4Z0o5d$_nh;w-4c@K<Af59cj|6xNS*As>c}3iZU`Thy<A3
zx_2+C2%vSA&(8K$IYF_a*&^7pCb#>5?um_=EV-0<t>woWHG1>29s+Dr-<@Y-R`dEy
zcw!<N*!p2=C=G2FUb?l+6@zRu4T*9_{3n%j1NMko`^5xQR#hR0C*PhdnWx7(#C843
zLBE(7$H$M&hU=rgq4xi@yi`BOBH@?<>lxPcQAS4gEPY&bd;48xDVNXu`b8>G*DA{*
zfVpGE?DU=}(tV9mU^WzgEE%r>P=05<i(m@GYGHG+SsAI3?@UI=%gFZK+a20QQNU`}
zoK>u>tQJ7IP=5F@$pHGw>B9l<<a`X%RAP-wLk^#jtkqPP;Q|I83b?4Q<nd@qM#ilv
zmokH5hpSL2ri&wu3jjQopty}{X5n&r4`t!{!z(MVofi-|QAQS|cr6A`^%g3psl{Ia
zfnlDAy*A&?NMl|RVQY!I6?96d@~#{(MyU6f1Dgm?#3d$O511(O8Hu$JWo!X_R0u58
zTHL<RL3cJ-y#M0q$AFOvVaeL4rMEfsiEjW}C<O4<Xq@*Z!Qx9z2QW^%+5U2WlaHT2
zx5IgSc>esk`|_+|V$JpsR?HZ+r`*$(Ol^!4@3JER{$*?58Akhzw+J!6eD&(D_1RBl
z!TMaAziPuc5!vKZR#r!rxW|&ElA0RT$_Pkcz8UdktgzWDzvm^}7A+h8BGS?ZBh86u
zglofL9AscvvD_*5SPq9nBr39bpA*n4nERNOFg0a+o|iZD^Jo65QmCzbr*RGUovjse
z(^I}Ak11OIN+NWc14E7Cl=du>LOm`Xo|^P6{1%lY?C|SKPOn~;Ru51?kb1h3jXbpd
z53zE?d9`NWmr7h`T%%-hvH5J~A&n^61<-zk+}5~y+Jyf^p3U2l$80|p-D2l@ZF1`1
zn4BlwSfx2Qd=D^0gSgV0n@}c_12|QGKLcuNc^lJ_XQGIF5zx2Q_GXc0X-??jhK1e}
zpZ*Xe|K)4T-3N2Na^h8Zp@u-v(7?S$yY|PeaXDMGf4s90XtY^5Nu605Yq<^(du#<O
zGv6dNd-LwyNYcV>S(i$_Ia+jt?9S$6@<MgB{L7G#6?eHrZ))t&(9lgs$9#DOg<DWP
zTvAf{C((xdF@pEkuaa|fFUHFGR6I<y>HHi9bXy5l9han}Ua9M>ARUvS=%$!+Yl;$9
z_gcJ+rxv%`JKpKu!U$l2R$bZgD(^1eX-bgEBxM8TMNnppSrT>%+Ls~jxstiyq<ANW
z71RHRoT~!|a_;FI(=VBANRS~V3&||ha{7B$%vI9alU6{_LIj#=COE62^4LL1kEM^%
zp`pL>y@7NT0{kUYK`^7MGZ9}e$^#5uy5o|7CAm%h@a^#;S-y>B)ebhSO14;@=vV-?
z!K2+A;CSgUa_aoL`PFu72{N9hEgx>kPFK~osmpnjm<L^<w&`QMVJ$hjZ=G)si5K~E
z-TxZ$?S7k9x3fjraqkm^3x)H78|LqgUu=>p?Xp0*{o0j-VdLfJ7sYE8yUrS(zi`1(
z-@Z^URXfKJ=Z%`E$?igL`pyY|G9yO&J7;8OmQQ$sAmQKTV<1%eWruBNHQ*r|Flon;
z24>ta2u29E4yobhSb*;FCIbgZfb6JU!#q)==C%(tU=|+B_B%ZMXwJGIFYjEjs;Ozp
zjPNoL?gfw#y6ze&e2abaW~}aK#R@F#RkNLg2M@}vVotn1^Ar~4($DJ*<IyoOF+@;`
z%O^HS7{YR~D%@UYqKpKcXa?!Cu1K4=yNEwfO7cQ+aWUH_?ynmA>b7(m2dPS;9}v#*
zo-gV4XOeBCJn#q0L8cpv_FNc#dgp$!_nPGXWY%k#c(xO-$z%61+r?RFd9ZHZ2$}c;
z7a!~FsqilA+VWg%j0C+-#^wdcvw-sBCNfQGUY8Ctp$(TzEh%MiRkt)|J28CsF1sZ2
z&;bvdxJOvsp+)b})zuYtUoa(0s-<g++4t$FbFS|I$#y1gEKgc=E5d5-bMCrtPn?{X
z80sr|j0#beANqEHP4r&#SdtxFz$p`P51pOPNhh4}deQM8wCFPo(sNI2dkd7_iducp
z$AFY$X!LpIJJ1VfVj}2VRkDcMMJ8Q=zI+a(lEJ)}o3lc!+Bk_mem*{BD2JJ%zHz70
z&1;<9HTDJX@E|3Tqc36l4HCQ-l&a>(OCAmS+n2O7+g`uoR&wN@xOc_bW_uhF4sz1U
zRL->EuJ`Id!0>Y>LkIosU%m|Wpn-Yy286-KHBXmoKtl_4rFd)Oysut;N~Bx=CQEH?
z)gTwi?W{F(2@5Y2HD_sMBH!JbagUEoRmjN5a3YpDjsMo%xpna|8}>R@y4JeT8W=H1
z(c}V<IV_=s+F-Rer~qnR4|(r}i}kijOxM@ZdE+wGEmS3xZCI}7Irr^AA!wZfpd~9S
zE4Kl2ec#dEK2`)}saNJQjhD@y^#X0H7o_D+uzsx=85#97G{UK061P;O9*+b)B)vSw
z<3E%~@!8q(-{>#*#C4{tvk6?daLcCU!*el^as}k%D7o=z%yK9M<&voY4SEsoC;;4L
zj1ZDwZ_dVG9XT+JX7ieR`e}ae1&|V1FInIH`WgeoR`yIx4oKO~lvU;i6P#X_9>`i>
zshH!26%AV;hWc(xW#Hy*O`h0w=U(Mjd2?jyA)=+I5Q>|RK`R%wXgW<;nS5Cn!82L0
z#r~#d#z}G*RM3@;ipKipW+vcP1zQ2TD~Z=(FAUvBce3`Fx-_G(oHVFB(LkcsvIvM^
z4v_N@26btn-*Yy0os&Cz6E?RnXno9`bR&5LYOjr<PwYJC82y<ploGf354Pmej&#|(
zif-;_!EeaV&o}MIyR$D(y=@zh1kJUMF9-j7kwgzl@dbdHj@z5=i8WDxZK?to#PP*m
znNYWJ*t!Q7L$)fP(=D`Xs=F`A0Y8?Em3W+1gE{iZY@{I;)okZ$Z2qX{O^!QYGAlOh
zYEpE3{PeQhZuiE?&Mb@OgkIRs`nLCRxc>FmU%RPYz8adEg?kPhKJ_{&KVKtG(m6xj
zN9JS2>#dRLFhx{LQ=%NEI1!{f@14g(lN<O7()!YPviD-6IO+`MddWzK<&L2@Xl#F*
zG>+4~m-2dS0q3WHUJL~~45AHU(!fRk?Qr>-j6vIBlr%4hD@dbbzWCq?;$6&?&B`jN
zs6_ht?akB6f;SI=5{l@3AS;AS^_T0zdrsvF#b17OO;|)^80tC_7V=Fkt!PC1`(oah
z^7*q$(>n=H@Jyl$ZMtsh>c)0$9PG@)0J;PMJXFV%7|DL!|E7mW={auhS8%y^Hb><U
zB&z$xJlJcYjt9G=+e`2v042I<Z=X$%lGgE98e<0#6$VvMr5G7$!TtNL)S_`_VwP+(
zZx2iJ4_q(}i|D_rFnoGSc@}u3x769@%FZeD`v8&ux3!6U%eQIF_wq^8fOZYp2A}@!
z%)ZYjV$p=3Uc}?25nsUfBFD!x(Bi=S3d)$qy&dO=dY>sCJWNr1z8}rb{%v>^if7n@
z77G;gzn(PnpCvr}(XN11J^AT6Tz_J1sZCwAL&k0HJ+T?^fM87so473-3MH7V5WHaN
zW%-W(ZJO0L)ZS}xpx33Oocr|gBkgL%{==tvNjMC{D)v2oi=Xu&FPM?w#!Z+RB`bzx
z$825k2C;!z#9Za;3gd|r-@_Tx7!5C`Vki}@72X?9W=v{B*;_$Ud@{4=&;gJsDl02@
zSEOWye3248ZIByaW#qjawKUZua@M=1rbgMy>J!*V2U2`QFT;!O(7VE;qfdrbC<e3K
zaQMdQ$0WV5Iv8>w#V9&x?l`CfL*U$k*7v(hR=?01b`LJ&LdRxN6_Xy?W~cGi@3URy
zAa|-dd^3S}j?8i)>VdBrr?i%a2FDpmr%+aL`_;a#c-Q{(AS%OJ+q@|JM?xq;5DMrg
zM(Zq6F8ts2<s31bdqs%$;Ab8Xhz_1ZY)<6*WMwUvh7=Tt0#;iX{P^=luC1Rdq8dXF
zT>2CO+RlcRsAUi>+U?G*b)$b|0UBet62BV{EQ`+}dV|dRSaNHDm1a(_xw-i*kR~^!
zIYhZZ>o^JOK6LAq3<0Kz0X|R_=DGcVckZDEonv8u|78}n3@xzjX-yXE>FH^O9f>BG
zHhAB_zC^c0tfydm6&o#jwIL#v69feU9DOX$EN)b0d5#IuCx*7DUPM}-f%&tmH-JEz
znwpMv4#10?D28V~zyYe}wy5}Y)V93aH%!nB*|?6@$GC@<(Bk<=pGcf$C`n8)Ecd8;
zPRI02hTiQH_zI4o29soSK79%U>!x$($uFk=X>LF+GdnvPgy9DadkiyRiBXT(S6%v7
zZD^mrp`HK|$~Y$`riFXmHhvuRh{FNbflGrP0sE7NmKL{>&sO0(S#Leq1&3fGhuvwy
zYUoPU6}Nq%9o8o@I}knXC>t9ZGyoUf98jTU*T*<@^NgQ^1=F#}yX#V(E&VQtC?MW`
zPo7D@(CDZ#P+?!0b7zMb4cXINXZ!Pk5g=qrhd=rO!vR`du)6+YV2A-`Utj7xS#51*
z#!!6C?RhRxS=|yxGe~FnovY-#g1G<SaPXfD1n7JdHf^p;m+pa{#)mjAz?(I<f>_1X
zVDX&%)v|jc58CoR&m-bC;5veuAlqGa(bOrk8aTBXV0(L_iLvoek40h>Wt)mSdg4TC
z85)!V5F$X$pPIi6-X9+rLNQSFw?J2b-SHOqM!*c~!RJiH;|&h8$?^lXf-N$!7Jekt
zAD|0w{$o5s@}Tcg&6(DGGiA7@j?^h#q*s}Je|5s}O!5nahSHz9cx6BRsk877*wLAQ
zfyTgL*9WtTrvTQtz{dy53m>xI0~$u$1;qCRb|>LH5vh&K^lRPRuc4;)+O|8Fy}PQo
z2itN@1+b?m<H5BoP)1fRD+HZ-n2`vMWmZZ`wSyrjnZqrsGbN)aJNVpQ;NPmv-rzkH
zS+;Np2nn^c%laQalR=^Y(N2E(q7%XscHkuXZ{SGNy@=By*#y+o65vED2v_5SPQ$}7
zMu$t*4y6hp`oP+uC*GhJ#Sto%<qH7N+JL}?lMARy0`-8EYL=%{Fy5AKb2y-C(#IEU
z_vsqB?Yx1C20vhe<puDA5Gny=2+hTlOiby(nTeBw_}HGgjArEXk5>iS7_$EYQDR{v
z4l~gSC<si)WPKj6<vKs>at59hSQyu@2gXFqzF$VaVsVa>^SaAS-#B%|z!hW{4lt!e
zKI7o?*T&U-CZ^}4Pg*aM#`Yl|{3*nPw`ofY*3#F{B#pvnIB)3S-SXsi&KH1PFCZgB
zo)keFotDEItmu?EPmXkL#Y!w@%))U6i=d0|A*F&{OSL~!E;Y;MKE&ug|DDO1l!Mq!
z%ac4HctwnjS(fZ<Z<wtu41XwDYXbbPAZFWrB+|1Hpe2`xNSoc-3r5bq=&0!E`p3;O
zJiupgAbGrioCF{~5TC;k@H^G>#SB&>h@N&H(ueG@tbPw_MzdHZV`E-e=HM4z4Qy7|
z)KEH`jPop!-N(LcRxqKgH^G~#OD;Obtui;>kcr?|Ae^0`4F4f6N=hyL*bTcXm&YRw
zkEBMT__Ryyx*+$Uxp;<=G3gMC=*<maT*!SZ%VPTc=SDQ+Y0+rh6~8X9UFk>D!Ck=>
zPQpUBk0Wd^=Kzy|N+(fy)ZRJt{7`Z0u@##J@bBEy&-)cv>6i;%zsU9xDI082>=y2D
z=ca$_$yWnl|M|TsdDPKSZU#zFr`;E~dE+W9tc&TQ&-ys~R#1VeWDQ~=x^h~^qungF
zCFVG|46gbv%M<r-W+0uGf@J-pk4BXi%<2X(tA9`XcEkQ*Ppl^rhoR8@y*GhnP(9dY
z3`(7pv_9($6m1UGML>6V#tp+AdaUoVzGQ_nL-g~=$jDkCkLtQ*E?N9KxmQ5R=K&Qv
z^v#=7k9`SBADb{HHBpsEkfN@V%oY@;AhdcCXWdahz0X8FB`wXPj~k6vU&(e>ZL2o_
zQX-?=>iO&;+&E!Ozw{po1MwU|WyDP{!l5m!jV2NypJTQ%L-t=hma$0cel&988o1RD
zbkb8&h7RnUVP>{kRnYdisiLw}+z5Fr`bmQ_mqxqT{#Z+}L8tmk+6IMm`A?RBk7+v5
z{wd$4OAu)gL8m2J?Y3?wKqg`CFLnl*rBHk5Q)_lz9&g1ls9JXK<_6t?EC?={>eR$^
z8HKRP8G;Bq?Ccq~6MqkYeciOK0TRhjbbJ2dze^k1D^KrNcuCy8zWx+KCLyHu&Q6xj
zPnnsE=9f7Z2pc7kvKnIWS@l(1u`OAMqip^d_BKnBCIp`uIKg4w@aEy_9z5V*l58e`
zj5Q{0Gq}l7R&8lYpidy;nC#}<bqP=EXYH;eLyz%PlB~_tB5Qr|1Ge1=8C)2u`;&LG
z?JI6E(*<M2X||tirhDg37%AT$X_#8s25VbHR_N#{urOOo9$Sp}m3)EUx{owjC?xw0
zHxLu<W1GP7ujH#m<f)zls|Zm@M&bHl%ds!h*@*|Rjw4)wR9g7EcT+`n*Y9a+a)MqR
z4qLzXPAn89#6{}BqPkzCbn|9tcXv0bso)LeXLY{NO=A-iGoXL@K0DYd)(g`vWr##-
zw+8BbUo<d0q=w5;gFyM5F-o^z0?k5lRHe=a<O4sDBg#ZUG7lko_T<t!kcLb_OLLs*
zlRy$uu;s86?LbT?Ys0?<{P3}U-ZfCtJ(n(wL(8iN!kb!Ltb<)CqquM0idzqji{taz
zAW=-mekS=_WQk8g3Wg+{-Zs8)KH$e?!i5Jtr_TE9Y*4mZn<qLl%|J0Ibrw~Pka#>&
zTf^ovmEQ-MlyENPp_e=fp@^1}4@%n_#Erh&<CLM>B2hYIBdKd>h)jeGL3mi$Kuq2N
z4G`R<sof?E-6kPQs;a?ozKYbT9!}^7p&$)lee{(eJ5v4qyAou?xWvV^vCEl8CDV&Y
zGKjEc8?>0Y8xn*6rb#@=n1N<&Hr|>l>^yN7=M2==*+?Wg11b@LiS#U@wdvTU1UcJ=
zMb-=*G5_~~duyNVqeG4v++V9wG94O!7GgpSziL^sK}W08$CRh)2do_u@Y8-#*W4Tf
z`gg22CZT5K0=P-YEgc3`vEBO@vm68w&t3PY7s7vv{`2d!9|7H6vgy1D;$*6NqO3pW
zZU-bTKrNMTZEG7QyHcV+70g*L0srN*5y22_%fV%wRZ=#rN<-|L8FHPV>BU-CZTcV1
zcHejVe@PLLl^GnZ2Mxvy6lGygs%>3UQ#2g&T1yD_91z7Qumq9RAxH}lfjAl0NZoRK
zjgq+lk(nqs-L<7LwfBg4hG?ZXLA^f5#dQPNehLV}tw6e81_o|mH)dwVw6wK}U?UB;
zf6{c^Tz%A2@;Dv3iYA~z?G4Nr+>`Awewyn^tBMx7|EihT|J0yu=L3dcS|GRmc9MkD
zs+$#{5n<RNNwUol6)l^skfP;$CI=hzUQb%w!IHUKw{HDKJ@U&J_Uf2KNDFBJqkrvz
zI>aIz9sIu%PX13l_Wx-F`d{)}bbV!L{3<{wuLb+ZI%DRW?}7VRfZ8q+*v;aH8*_6G
z%$KYq6$h^mnzo3l(oxSI=><;^*0Iu!8#lhLTmm-=gmcfYD|?2gEqZ9LgUSlHP!qku
zBIDsyK6PM%6aSzd8tLwe5C{iH2VD%Fdv+c+peV>49B^WT3J)`He7nzy&GIrW`pwak
zCr{3C-S>V+P8#_2OhyGsVy1u~xNBTWR@M;IRPcEu&u%4u8+azuK&-6|Ehs2>vLs{O
z&ijlG#g6|K2Axy->{h<1UphG4oayR`(UFnI3Dh2to*})%m>K1n_K*emeHv6CmN0V(
ztUP!pC;pd^4|*PqaaQp8^T56s|C!_qA?T{niW&T(+Id02lY!rV2ZSXtJ_H2-;IuNO
z!h5T=3*6m<f^J~KYi|~|o7B-_@@)OI0h%M%2ik0xH$)tiV0UCwe6HXJ>4LToH|Urz
zJG@u-4q}t*S5Llb0#w+hf`5(wBwY^SsX9>lX^4SG0R{2s5EBv6lysgL8f8opf^#cG
zx~{BOv9?{wZ3ZLo@?VOxA4j23xfbVI+<=MKL8vPQOdM1p(xk-fdiPq&Tk->2uZ~+G
zJx)SxUvc|pk-&fEpJN06OyczwR_Q_xd_(_5N5=)&9d{UBWSe(8vI%kw%KvYc&r?Z<
zHF26fsjRSijHs|0-2Zn9!ItnTF@;>_y11X_4K&JQCE&{0C3dy9-vsk+(&x(UCXm|C
z2?{plCVZL$dG3#HeBk+yOyI2GI6;Z?zR%8Xx+|0I{vXW|{9h^}B=NsH2`C-l3+G8l
z$3Z>-_;|HFF7WU0nVFf{!g7}p5s-T7`}bE6PI@>XIYHs@SJ6g4*eR?yJ~U|EyJt3%
ztMVx$qZLAfS<AB&xv`q>Kd9s)uQf+<A|O=Ke(#`u`xvy5*kWc$h*|Q}?rt5#+Mj~J
z;=<x$DlqnMsG~FkZ(|f;7C->mX4V!*1HT&|q&hrN2%cN`|K<N}eDS8VyieyAG(a%^
zY)~THT}8u^1%Uug#n;CJzjFYC=iuN-ah~i9haffVS5dHhpst^P!XlU(m?zzw<*wpZ
z-l|D+FtrdUZ?v->gXH)Ue74<@8o{sr4uqiFArfJIA`{8G5MG1RuWIPC<qQG5Na!m_
zMh_Ar#K`lSDl07@sk<ZIF)(n)v7hV`t~2jokm4nzO2E0j36)~okJkm@$yHZdJ2$yX
zhRYd(U+&b`fLsgY`{1jF<1iIrV9nOSUWw$Eu|4X%D5>3F8(hN=O>4bm<WzVw!6PGc
zAL?D`T_OO*ajFOs%cnL=F9vaQ=m9<|0v0+AXDB#XKL3N&-(&_n!y@dKN8`8?%j*qu
zgiA)p#bpAbZ(Fpb76U3xn<kfaInW~ka5#?waG-Ox$&-uS6t}@cWLd}_XOk)*pIkZ(
z^9F0a>jbb_WF)yc@7%d_bw|b-_bAhk?X-lWJR+ME78aH+BU#D1x;oiuK5(y!Y<oB~
z(^SH8FLv$ldr%zwA+`mB7Q?``!vRAM3=E*A2>{dc;Skrxl0#Bp-T+nD5fLAcgjK?$
zqRe!}Arcxbc>lsIVW)y5;lOn_<jm6>ZqR9dJ;ASy#FaWj>=rEb)q=!7-u~g!WB2qy
zKLRQk22zj5lJabQ?!&5-+}zvX)Q)P(aqN^84h@FLMI()<9p)(>Eu_nB)!ilEmT^&S
z^P1VSPK!Qf=sA4}D!tVH48~Vc-gNHQw*?SaSUVrPjw}OCCI0&6Ti(Bu1t}t4<cpwv
zC9b<ej^mu9WN&7Xy3Fz)f|P6c5hXQn!p$${AW}mhXiRj@4>x&=5K?A>ST5bd&IcN~
zk^AFOG9I-6vA6njc>3K(&p<lEQ6E+RZ>DoAM0vCI6W+Yxg`}D6?C6(f*$5bLfRsiw
z82{tldC8_07DSjz7zU9??j!5TliaW-#it(sW6Yo*Ujq8DGEhLXfy(Fplr*)sqIGEH
zkBEm>{qk5UEV!f0%;%9L0kkqCW9>{Tv}#j8P-Z@Vdq_F7ugRi|bq!`)Ki2IA8%cl2
z`6u8&f+QAizLj`*ho@+oBFeowx0%V%?VH^OY_isfK!Z4C<RbCM2GC|ZU^vF02-B0F
zs&)5n3JD9l47>Z30!7Say!FNkrKt_l31qj6aTr!rk;j7>EjDgTpTP??^j^Lf(hG3>
zCT<fsbuQ)TVgU*`Q~Y`bwdvlc+}Fp`F;XU*YEat)`8lwb1ast`ujf4;L9V;fQ?DcX
zZK7+0oejAVl9`3YWHeg8s3AARwclNFWxZh!gY>WE={>-n{JZ{wE5rvozdaF5-KB|+
z$)e`#Tiwg+&wY*<au<SZoT~<MmbUjSg+Q5G58+dArS1%}?OdEnhPeRnzXpq>K`;Vt
z{cFBkUcO@WH=X7(Ac1uKnTgBrL4QX{gcq+HXlVStM}g&y!o3U%YQLXmSmvU-4HClJ
z2_o<1W5<rkF!!!htc|b|rsUY!2M+9HRpxObSrU}T7(01kVZpHkqVd|^J{!|G6wV56
zuYC!R#Zv`m7{4(uMPq*4*3?YaM=>$2`0i*8bO*bwAE&X01Omh#`603hnHwiN#Srl%
z#NHq~s#i!~%&;jy8`V5;x;~b*m|U61T0YsxhU*Y@fpsfl-Vh30oGb};#!9IVrXm*h
zxKfKlAkQKwU(?%ps{0OZ6P8~B{j?`s)pVl?^159rL>HD!uL2J%pJ3Wy8!$Tyu06MU
z!e=C+RuBDY(n@~E2uQZQI{~{zzi|^*bJ<KO(as&H3zCKg!~cc!>M!5431*ps>50p}
z-ct$;lN68m87&1$bmggm)WN3*P@&KyK}c71ngpG=X)<uRq8sRi{Tk?87Tq=Q?{rVr
zS_oNzRN^iz7Pcc}J~$y35C_Gm1UH);Ed`->Bg?%T-NDfj;Ns_y@<A*?2DYp&%gTWH
z{8K=DJ=Ttv7R~g}+?#0##=v5wT<v9orvieaGZ^GHbMZ)Yh6rZ))mYHL)Hg07wsc9k
zb_O)G1#nm=iqO2g4(?awx@!#e^fE})QMsKGVBP)H65#ys-RgRU)~Tfan~=e?cLzG<
zHDPqne5MLh8*)H5OA7?lJ0L~Ji9xK#4tTpBtoB$dK&?BBk+Wi7(9?@fbaZrW$Oy>g
zA7E75FvzaHN3_lx3JTVs|MmiNEkKebE>M<Gx$#{!%kfAo&=aKPLhxk-z@bvUb*mK%
z83HSanG4xUJ^Ec8Dox#(B~TnFunlw~&NGyU6@(&wB2w<~=qPJZo9JMTFQD91QUySR
z4d9L`5C`46wbIhoI5l`RSQx)K?8_i~@lwK-=O875YxY<7ya2?2(gzprfpay(D(CAy
zrImS)4~=H%-VUxL#+0zhVZDC(C@za7*wN=6Vil|PJ9rE{6GTfHXIgRp0IeLX_CW|-
z2_c~roZQ~U?s&CyJgA^p3Ql)zZ4Kd8MLV~zxVewPTFZa%ZMXaHmps=-0xUq~$%ix}
zVc8V~a|Ch@G$t#+t&-97sjiMfHn)5Ts8)G=lcshVdNxj?MX;b_cy`Q4K}sePR@r)c
zzyvLt0r4Ba0dI&CN2es*hY+H$ZTCfYn6>i4#(e`}0fCud-p?);1GX_+og46Hiop~@
zxUvn+Oh1xm1fjwNkBRdk+dROJdBDeAT0O*(-SAGLZLq}db_2g4>*A#xh-D%n{r3DX
zAnd50&2cA}E>HDnl5!vVk_cCZ)UIB=`gPmmwzYMJ%dfAys)X)pX<0*7V0k>}Ku1>|
z1&^e<dI3ZoT6ql8-K@`RFYm-ISAB>s7GNsYek2dt@pj=^rYV^@nVHA_HfccZfzc#Y
z!1<BU(MY`boy21WFlmte^YmfG+yI+ipizjp0*&tMp8fQp(Ch1w`639pgolQng<A=4
zMwC#{`?x%*Q(*v}iSV91Vrj>C)R3tVq^^;CT3vm8IB0*!92`vj)j^IP30da5OzS~t
zb<soan<N#`r4o#yNv{n>2%FX(f!U}S$kIKMj9*CH*;GW3C=!%|0n6%HIpvL&*%+Wi
zb&ZWt@O{cK){+AK;V2W6#p;Xjg$DU|))t1g%U?Q1nYr`_dsd4WulOqbyTiSg=JXvH
zZ-PjCXpC)e-;%?~zc&T^e+j++Z+z`Z)i%-OThIUi?M|9V-|hoIgoIXb-U%@=h%exJ
zL}70S<lha!{JK-Pr6m#%^<H<zwQpeK<Ii7Q{}^J}r2m%lFFU{dzi~bOkE$66t2(lt
zZBW9Ih+Y5msq_E?%4uy4@W3zxr2MC=APx;7QdBTQaKztJ$X6_305&`_@*GTv@1k@8
zLN<ex<t_?CBreg6ll}&nk!osc<x@6bHZSM1eGCDy!Q@BK!Btt976y6`X_z!5&eMy-
zu>JI=XBL)!d&%bHPw)2wXYmhi>FDTujd5Cj<PzZTOGmvpotaY+4mr&IDYG!A4YVd|
zY;5ewQpo-Qco7Y~tF^Utck2pFSkMZ(`OQ63MEPvaKUpG%GbVtm9fCNwp!IF`T4ZX0
z`<EjBS!*WN-wFhVSdjczw@Q&A3s4V&=APATmRBR{0r>nipk<|i&h0q(<51#8DZ~vC
zQ5n`6n}R<gAxOsnLIM3`S=k5(8&~nK6EK4?-=z`4<#g(1H9SE;7kMrKCm!M}hG2Oj
z*9sVi;YwI^G&d|%?CBFMjF2=?0|nU>+&tN3A?&7`T=4#2MhFe8<qv();m2?9mqqjj
zNGsocD`q!wkJQkhGT#0vB2m%r2b^b@v9Yl!bZf)H;cGHpH;@?xKj58iUlYRQ2a%Bl
z-~ETeVZPI&og3n)y#-ds-ZLiUoyGq;hJlF#7=PFUX#)3rr=fp}4Irrin4hpnfN2K|
z^8i{JwU+>y)kv6&BTK@I*y_QnbI`0jPE}A6R@$o)LDvXwDc6VKggRVNagV?Ci7#91
zD-rGg-%|$Am%vwmL2WE)^wOnEU}M#+vmMzq21P#KYi;2H!{C8yf_OoL;<Ka8i44?>
zzbyXyOv6{@e|rc)@xN+A3%}I8=QafBqF~Hmq7&LZqDR0?9NC@;hT&j;s;{pfn#>_Y
z!$hSrL^Dszdg;LKBolnv@)T_jO+LifAXg5#X&8Y^1r0!O62`UvMAVUdKOT?wKykf?
zsZ92>5K%z9eJPiz(9KP+aoDSJ1NQn!SMpUa!JIl37y*zeroit<rv_-)1Lyv_3Y1mY
zb;dv!XCJSV^?7QSHaLR7sQEw+pr>AZ5-RVuuNIs)NXESQZX5>C8F>f-7<YfBB>=f;
z+_?j;IXnZP4s6BU(t1~9pe+|XVEAngt|Kyk0uPPw53-Ls1)~rk81&e=tEdD?Z%lnz
zfRRltD0Qb5GaQKv8Q%#EltEd5vXTn(8m%~t5$Fn`usb5Y5imm}@)!wm<AJIGF0c$`
z*M^QEKdRZd$OmsmL=Yr-AT?8bU9ZUYJX}-Aj`JbC>a$K>Wv0n-MxdeLyY}p_eX?uM
zId}&$B?t@@3Em*#TEH5&%*;}eK?>OK0kW$hc|-6$HNf0}OwPiDOVsVRlq>ahbzy+v
z_IjQYC!zITAVGC*4t4_E+;LP57}F}t5eW&|FxhmfiI{OOQ<{Y$v(lHGWl*94<L0H~
zB{|vIo!~^Iz|{NO-kP6D*PS;!{~nm=Ewal*oJ5Bm*wP?-n4nnG0|K2LSoUul%|*Q5
zW#cy&i0tBoq=I442Sf?Gazo}892d3SAbtQtDvkEIN$RM><pRJURTa>N#Jty?8V4NQ
zwlKG~w9>X#YvjcGVq)jO!HUt+(i&;m23T%w-T(Xd?*ib=?%wajMOY1&CE)X)8@oMf
z<lB4tvcr{ruSHd|B%mP3dpS+|fh`AZ_)py3Te$}qQ%|8y%)oXK<cDe@?M&`}H`xTT
zi)nZ6L;qz&=0L_7%FiU!Tmuij{Er7eh<NaDqAf7E<-E~X>YQoM7RZDK2Ju|${{5`f
z^mI?}k>NlX&zwZrK+JfvR8GXga9IY9>(R(Q98xMoG@(5!*>!Y_GenvL8DzEJK<p-<
zUMUd4JbLDgzzW5Oz;9GxASzDj0rR02QmW4!b94b(nwp!J3a|+o`;qb?`|J+B+z{}Y
zXplURoCF{~5T6n78!O{kgv=9x=&2f#p45o(Ju<b5zh@$|c$9HEO4|?=tJT$2QZGnP
z$nMi>w?h8@Sepc#K9l59>32{z-W}2o)<8Iaf-;<f5EuC5UdYMYb=w2Iwy(iE4&TB}
zL+-)Cx`LB~qu{FVek&xNjNCWe^0Q0GjrbkQG%!4T^2Hnk%fkmYou+%2i(Fu#ciAm=
z5w<%VC{XDRq;1eW1W9iWqT?FV$ey$1f+W^`-B1$b*RGKAu_rmt^!4QIbbfnFx!rqB
z2oLt(5fSn1(o&o)C_$xmFt1_TJqzoC91YDy6;vS0?f^cBu7rc0T*@!99{vHgF?%9p
zxhIOIKst>9$@&fJ`-6yC%{*}88s7Vel#~?7PS@SN1jMuI+z`YqHJq)A2kjntP)c8P
zqO`jPboZGeJlr9BBFr027fm7hIVkvRKpxKni;F_C(Z!DjQ$SuhdhD2q{hav04w$Gf
z);M(yirPx2rsKxQA6mTy>70`C2U9O$o+^X{8LILA!fdOzkjKJVD0!$~L~b1N9FIQ|
z26&EKGBR65-YC%ZkO7t+y!#gj0+SrDf1YvBgujN2A<|Dmnql)7fs>Q-5M^wF0SEvJ
zC*d;n=3zV>TZurM)UigeLF1)dJ*2rol3gLYCaT0s-h;<*a3S-=VVsJkS#0%?Nr1rC
zP*MP@RRE1f6Q=dlfd%Z5NSTe5;KO&oNEH3#D$H#`G?0iq+{7NeK=_gYz`jH#@E^|)
z2>XY$0ps$y@GOl3);HyjFftB2ih~DtR7t`NOh3J%{I;u{Hl(aJ2N-<j#u5ayv5Rq(
zVtfvN<R}iNC9YLN_MhJ838Xx+Bnto*vOB*srr;*S(_%yswG9!+0@$#;;MpusNVY4Z
z%biFT!QR;G%uG|)uK*cTNs!h>@@|okuqx%}6BLY|A(xf$^C43youSY$ktfv<eB4c-
zqr<|yC+RxF4_62IEh5rn&!4ZeV?)FQVsk9zCu}4=AWxlz$B%?VGHR?7cnH~!X_62T
z6EjL>;)PTixKZRHyN@tUKLm{tMCYVgxi5NI$irNiWjqRqVlaUMbAg_u*prK_Z*|1>
zE7{t9R)I&YP-iI^x9h1OO@l;QDnKc+izDSDvqw<FT_!;J(q;q>UO;FT<f!i8UT1E6
zg$#ncTgM(yR^dLk!sHpmSA&qTJm_LPfVje8%c-IwF&$^h#Zr@+!(>*jL{QR047tdK
zHrl$y4oSzEn25936{!#-D@`_-{2RvjY{b!@k!+v)ylf6B7zh0kC&~M=!#_`rYVYG`
z-7_@nvr}fYu3PR=h*;sc*J)PB=)fAJS?@>d^PkksH+?BRUwv!=%0U;9H4g>5k<~rz
zVg)k}5T&b!jf4|MRTm->H{~HZvMEfx@+v4O5_tdt%5zExG8kc?ht&cM9)f=02-3hT
zh7SU&Q4l&G?=8$6f@cfmV3+w|ln9bR!??uYN7#E;ATjuFnuJOR%&koB*vxL<(^IOV
zIZU9_-*?=$n(n5a@=L1|8g9EL&gdNM?RoKQAFc7=A>JWW>v_Jtwn5Kc&}z9H^ZOY3
z#<nnEi?#VqHQpleKx?ClY{l;b5e?!-Q?s)*>+9=2D9tP_#R_xoOpQ;+mpx=Ame&Ka
zMXa|5cHFTwQKTP2OY7?_Wo2b~c>L)t>`qxTsk{4yLh}4e&JFiM()8uamruC|t(`&t
zzGMzhJo%uN`JGCoGRS3$o}xE4Gvg8y7grQ?8_5wwqfM-=#v22y8)w}ZbA*0Qy9}<b
zUIy{V1dH8n6wi1qeEP5n27@-Yuo#@1yO@xW(C-X#%dubuPS7qWTbSNrPcN^a`1tsb
zTA7c0e0J`SzM;$yi07!oH+|@uyphutI66AY`-<b$hZiunU0hZcNFtFe3-uns7bIQ6
z?u@LjOLlg48r#_Lfk}!SDP%Gm8XBshTdAn2DZmHV<5+Lp@nUWM{{6<ggam~$^7hZx
z_+}R;r{g?4Ja+_Rk~Kcbr)s=(EIHlQ)&@_}nV(*)sR$<9i%Z6<_4V~tcXZqnjcc?N
zi<<`lHG6iaXH<MoT+?n0OO2`pS#0<DmrSAIZoB>;`~#jOB=dw|Q>fQ6yF&0d^))f_
zR%#BaQf?FOT7Z!Z?Gv`8IFiCyuKC>>N74V|KT;{XZt!VdbXj4Ug2%DZDBVz#&%XBH
G*?$A0OrCB4

diff --git a/docs/dev/articles/index.html b/docs/dev/articles/index.html
index bc266543f..7c44e5355 100644
--- a/docs/dev/articles/index.html
+++ b/docs/dev/articles/index.html
@@ -8,7 +8,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -116,7 +116,7 @@ <h3>All vignettes</h3>
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/articles/intro_to_keras_for_engineers.html b/docs/dev/articles/intro_to_keras_for_engineers.html
index 4f83cd204..e331ae6f5 100644
--- a/docs/dev/articles/intro_to_keras_for_engineers.html
+++ b/docs/dev/articles/intro_to_keras_for_engineers.html
@@ -31,7 +31,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -229,21 +229,23 @@ <h2 id="a-first-example-a-mnist-convnet">A first example: A MNIST convnet<a clas
 <span>  callbacks <span class="op">=</span> <span class="va">callbacks</span></span>
 <span><span class="op">)</span></span></code></pre></div>
 <pre><code><span><span class="co">## Epoch 1/10</span></span>
-<span><span class="co">## 399/399 - 8s - 19ms/step - acc: 0.7477 - loss: 0.7456 - val_acc: 0.9654 - val_loss: 0.1189</span></span>
+<span><span class="co">## 399/399 - 7s - 17ms/step - acc: 0.7464 - loss: 0.7476 - val_acc: 0.9627 - val_loss: 0.1272</span></span>
 <span><span class="co">## Epoch 2/10</span></span>
-<span><span class="co">## 399/399 - 3s - 7ms/step - acc: 0.9377 - loss: 0.2072 - val_acc: 0.9764 - val_loss: 0.0786</span></span>
+<span><span class="co">## 399/399 - 2s - 5ms/step - acc: 0.9376 - loss: 0.2072 - val_acc: 0.9771 - val_loss: 0.0776</span></span>
 <span><span class="co">## Epoch 3/10</span></span>
-<span><span class="co">## 399/399 - 3s - 7ms/step - acc: 0.9567 - loss: 0.1474 - val_acc: 0.9820 - val_loss: 0.0623</span></span>
+<span><span class="co">## 399/399 - 2s - 5ms/step - acc: 0.9571 - loss: 0.1475 - val_acc: 0.9816 - val_loss: 0.0633</span></span>
 <span><span class="co">## Epoch 4/10</span></span>
-<span><span class="co">## 399/399 - 3s - 7ms/step - acc: 0.9648 - loss: 0.1182 - val_acc: 0.9860 - val_loss: 0.0490</span></span>
+<span><span class="co">## 399/399 - 2s - 5ms/step - acc: 0.9652 - loss: 0.1175 - val_acc: 0.9863 - val_loss: 0.0488</span></span>
 <span><span class="co">## Epoch 5/10</span></span>
-<span><span class="co">## 399/399 - 3s - 7ms/step - acc: 0.9707 - loss: 0.1007 - val_acc: 0.9872 - val_loss: 0.0475</span></span>
+<span><span class="co">## 399/399 - 2s - 5ms/step - acc: 0.9713 - loss: 0.1004 - val_acc: 0.9870 - val_loss: 0.0469</span></span>
 <span><span class="co">## Epoch 6/10</span></span>
-<span><span class="co">## 399/399 - 3s - 7ms/step - acc: 0.9753 - loss: 0.0868 - val_acc: 0.9886 - val_loss: 0.0403</span></span>
+<span><span class="co">## 399/399 - 2s - 5ms/step - acc: 0.9754 - loss: 0.0854 - val_acc: 0.9879 - val_loss: 0.0407</span></span>
 <span><span class="co">## Epoch 7/10</span></span>
-<span><span class="co">## 399/399 - 3s - 7ms/step - acc: 0.9761 - loss: 0.0790 - val_acc: 0.9890 - val_loss: 0.0408</span></span>
+<span><span class="co">## 399/399 - 2s - 5ms/step - acc: 0.9767 - loss: 0.0788 - val_acc: 0.9899 - val_loss: 0.0378</span></span>
 <span><span class="co">## Epoch 8/10</span></span>
-<span><span class="co">## 399/399 - 3s - 7ms/step - acc: 0.9794 - loss: 0.0685 - val_acc: 0.9878 - val_loss: 0.0443</span></span></code></pre>
+<span><span class="co">## 399/399 - 2s - 5ms/step - acc: 0.9795 - loss: 0.0688 - val_acc: 0.9867 - val_loss: 0.0438</span></span>
+<span><span class="co">## Epoch 9/10</span></span>
+<span><span class="co">## 399/399 - 2s - 5ms/step - acc: 0.9799 - loss: 0.0666 - val_acc: 0.9893 - val_loss: 0.0396</span></span></code></pre>
 <div class="sourceCode" id="cb13"><pre class="downlit sourceCode r">
 <code class="sourceCode R"><span><span class="va">score</span> <span class="op">&lt;-</span> <span class="va">model</span> <span class="op">|&gt;</span> <span class="fu"><a href="https://rdrr.io/pkg/tensorflow/man/evaluate.html" class="external-link">evaluate</a></span><span class="op">(</span><span class="va">x_test</span>, <span class="va">y_test</span>, verbose <span class="op">=</span> <span class="fl">0</span><span class="op">)</span></span></code></pre></div>
 <p>During training, we were saving a model at the end of each epoch. You
@@ -372,7 +374,7 @@ <h2 id="writing-cross-framework-custom-components">Writing cross-framework custo
 <span>  epochs <span class="op">=</span> <span class="fl">1</span>, <span class="co"># For speed</span></span>
 <span>  validation_split <span class="op">=</span> <span class="fl">0.15</span></span>
 <span><span class="op">)</span></span></code></pre></div>
-<pre><code><span><span class="co">## 399/399 - 7s - 17ms/step - acc: 0.7350 - loss: 0.7737 - val_acc: 0.9276 - val_loss: 0.2389</span></span></code></pre>
+<pre><code><span><span class="co">## 399/399 - 6s - 15ms/step - acc: 0.7355 - loss: 0.7721 - val_acc: 0.9287 - val_loss: 0.2384</span></span></code></pre>
 </div>
 <div class="section level2">
 <h2 id="training-models-on-arbitrary-data-sources">Training models on arbitrary data sources<a class="anchor" aria-label="anchor" href="#training-models-on-arbitrary-data-sources"></a>
@@ -440,7 +442,7 @@ <h2 id="training-models-on-arbitrary-data-sources">Training models on arbitrary
 <span><span class="op">)</span></span>
 <span></span>
 <span><span class="va">model</span> <span class="op">|&gt;</span> <span class="fu"><a href="https://generics.r-lib.org/reference/fit.html" class="external-link">fit</a></span><span class="op">(</span><span class="va">train_dataset</span>, epochs <span class="op">=</span> <span class="fl">1</span>, validation_data <span class="op">=</span> <span class="va">test_dataset</span><span class="op">)</span></span></code></pre></div>
-<pre><code><span><span class="co">## 469/469 - 8s - 17ms/step - acc: 0.7492 - loss: 0.7477 - val_acc: 0.9125 - val_loss: 0.2952</span></span></code></pre>
+<pre><code><span><span class="co">## 469/469 - 7s - 15ms/step - acc: 0.7502 - loss: 0.7455 - val_acc: 0.9107 - val_loss: 0.2982</span></span></code></pre>
 </div>
 <div class="section level2">
 <h2 id="further-reading">Further reading<a class="anchor" aria-label="anchor" href="#further-reading"></a>
@@ -500,7 +502,7 @@ <h2 id="how-to-distribute-training">How to distribute training<a class="anchor"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer>
diff --git a/docs/dev/articles/intro_to_keras_for_researchers.html b/docs/dev/articles/intro_to_keras_for_researchers.html
index 98d75eaa8..c0742d7ce 100644
--- a/docs/dev/articles/intro_to_keras_for_researchers.html
+++ b/docs/dev/articles/intro_to_keras_for_researchers.html
@@ -31,7 +31,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -1359,7 +1359,7 @@ <h2 id="end-to-end-experiment-example-2-hypernetworks-">End-to-end experiment ex
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer>
diff --git a/docs/dev/articles/making_new_layers_and_models_via_subclassing.html b/docs/dev/articles/making_new_layers_and_models_via_subclassing.html
index 0e0c3816a..52cce1c2e 100644
--- a/docs/dev/articles/making_new_layers_and_models_via_subclassing.html
+++ b/docs/dev/articles/making_new_layers_and_models_via_subclassing.html
@@ -31,7 +31,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -469,14 +469,14 @@ <h2 id="the-add_loss-method">The <code>add_loss()</code> method<a class="anchor"
 <span><span class="co"># losses get added to it</span></span>
 <span><span class="va">model</span> <span class="op">|&gt;</span> <span class="fu"><a href="https://generics.r-lib.org/reference/compile.html" class="external-link">compile</a></span><span class="op">(</span>optimizer <span class="op">=</span> <span class="st">"adam"</span>, loss <span class="op">=</span> <span class="st">"mse"</span><span class="op">)</span></span>
 <span><span class="va">model</span> <span class="op">|&gt;</span> <span class="fu"><a href="https://generics.r-lib.org/reference/fit.html" class="external-link">fit</a></span><span class="op">(</span><span class="fu"><a href="../reference/random_normal.html">random_normal</a></span><span class="op">(</span><span class="fu"><a href="https://rdrr.io/r/base/c.html" class="external-link">c</a></span><span class="op">(</span><span class="fl">2</span>, <span class="fl">3</span><span class="op">)</span><span class="op">)</span>, <span class="fu"><a href="../reference/random_normal.html">random_normal</a></span><span class="op">(</span><span class="fu"><a href="https://rdrr.io/r/base/c.html" class="external-link">c</a></span><span class="op">(</span><span class="fl">2</span>, <span class="fl">3</span><span class="op">)</span><span class="op">)</span>, epochs <span class="op">=</span> <span class="fl">1</span><span class="op">)</span></span></code></pre></div>
-<pre><code><span><span class="co">## 1/1 - 0s - 161ms/step - loss: 1.9081</span></span></code></pre>
+<pre><code><span><span class="co">## 1/1 - 0s - 131ms/step - loss: 1.9081</span></span></code></pre>
 <div class="sourceCode" id="cb38"><pre class="downlit sourceCode r">
 <code class="sourceCode R"><span><span class="co"># It's also possible not to pass any loss in `compile`,</span></span>
 <span><span class="co"># since the model already has a loss to minimize, via the `add_loss`</span></span>
 <span><span class="co"># call during the forward pass!</span></span>
 <span><span class="va">model</span> <span class="op">|&gt;</span> <span class="fu"><a href="https://generics.r-lib.org/reference/compile.html" class="external-link">compile</a></span><span class="op">(</span>optimizer <span class="op">=</span> <span class="st">"adam"</span><span class="op">)</span></span>
 <span><span class="va">model</span> <span class="op">|&gt;</span> <span class="fu"><a href="https://generics.r-lib.org/reference/fit.html" class="external-link">fit</a></span><span class="op">(</span><span class="fu"><a href="../reference/random_normal.html">random_normal</a></span><span class="op">(</span><span class="fu"><a href="https://rdrr.io/r/base/c.html" class="external-link">c</a></span><span class="op">(</span><span class="fl">2</span>, <span class="fl">3</span><span class="op">)</span><span class="op">)</span>, <span class="fu"><a href="../reference/random_normal.html">random_normal</a></span><span class="op">(</span><span class="fu"><a href="https://rdrr.io/r/base/c.html" class="external-link">c</a></span><span class="op">(</span><span class="fl">2</span>, <span class="fl">3</span><span class="op">)</span><span class="op">)</span>, epochs <span class="op">=</span> <span class="fl">1</span><span class="op">)</span></span></code></pre></div>
-<pre><code><span><span class="co">## 1/1 - 0s - 139ms/step - loss: 1.6613</span></span></code></pre>
+<pre><code><span><span class="co">## 1/1 - 0s - 106ms/step - loss: 1.6613</span></span></code></pre>
 </div>
 <div class="section level2">
 <h2 id="you-can-optionally-enable-serialization-on-your-layers">You can optionally enable serialization on your layers<a class="anchor" aria-label="anchor" href="#you-can-optionally-enable-serialization-on-your-layers"></a>
@@ -802,7 +802,7 @@ <h2 id="putting-it-all-together-an-end-to-end-example">Putting it all together:
 <span></span>
 <span><span class="va">vae</span> <span class="op">|&gt;</span> <span class="fu"><a href="https://generics.r-lib.org/reference/fit.html" class="external-link">fit</a></span><span class="op">(</span><span class="va">x_train</span>, <span class="va">x_train</span>, epochs <span class="op">=</span> <span class="fl">2</span>, batch_size <span class="op">=</span> <span class="fl">64</span><span class="op">)</span></span></code></pre></div>
 <pre><code><span><span class="co">## Epoch 1/2</span></span>
-<span><span class="co">## 938/938 - 4s - 5ms/step - loss: 0.0748</span></span>
+<span><span class="co">## 938/938 - 4s - 4ms/step - loss: 0.0748</span></span>
 <span><span class="co">## Epoch 2/2</span></span>
 <span><span class="co">## 938/938 - 1s - 1ms/step - loss: 0.0676</span></span></code></pre>
 </div>
@@ -817,7 +817,7 @@ <h2 id="putting-it-all-together-an-end-to-end-example">Putting it all together:
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer>
diff --git a/docs/dev/articles/sequential_model.html b/docs/dev/articles/sequential_model.html
index 7a4153833..2b3e07528 100644
--- a/docs/dev/articles/sequential_model.html
+++ b/docs/dev/articles/sequential_model.html
@@ -31,7 +31,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -511,7 +511,7 @@ <h2 id="transfer-learning-with-a-sequential-model">Transfer learning with a Sequ
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer>
diff --git a/docs/dev/articles/serialization_and_saving.html b/docs/dev/articles/serialization_and_saving.html
index cafe18ca8..97e20cbd3 100644
--- a/docs/dev/articles/serialization_and_saving.html
+++ b/docs/dev/articles/serialization_and_saving.html
@@ -31,7 +31,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -946,7 +946,7 @@ <h4 id="how-custom-objects-are-serialized">How custom objects are serialized<a c
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer>
diff --git a/docs/dev/articles/training_with_built_in_methods.html b/docs/dev/articles/training_with_built_in_methods.html
index 2fa841a75..3a7739bdc 100644
--- a/docs/dev/articles/training_with_built_in_methods.html
+++ b/docs/dev/articles/training_with_built_in_methods.html
@@ -31,7 +31,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -211,7 +211,7 @@ <h2 id="api-overview-a-first-end-to-end-example">API overview: a first end-to-en
 <span>  validation_data <span class="op">=</span> <span class="fu"><a href="https://rdrr.io/r/base/list.html" class="external-link">list</a></span><span class="op">(</span><span class="va">x_val</span>, <span class="va">y_val</span><span class="op">)</span></span>
 <span><span class="op">)</span></span></code></pre></div>
 <pre><code><span><span class="co">## Epoch 1/2</span></span>
-<span><span class="co">## 782/782 - 3s - 4ms/step - loss: 0.3410 - sparse_categorical_accuracy: 0.9034 - val_loss: 0.1855 - val_sparse_categorical_accuracy: 0.9460</span></span>
+<span><span class="co">## 782/782 - 3s - 3ms/step - loss: 0.3410 - sparse_categorical_accuracy: 0.9034 - val_loss: 0.1855 - val_sparse_categorical_accuracy: 0.9460</span></span>
 <span><span class="co">## Epoch 2/2</span></span>
 <span><span class="co">## 782/782 - 1s - 1ms/step - loss: 0.1590 - sparse_categorical_accuracy: 0.9538 - val_loss: 0.1323 - val_sparse_categorical_accuracy: 0.9621</span></span></code></pre>
 <p>The returned <code>history</code> object holds a record of the loss
@@ -239,7 +239,7 @@ <h2 id="api-overview-a-first-end-to-end-example">API overview: a first end-to-en
 <code class="sourceCode R"><span><span class="co"># Generate predictions (probabilities -- the output of the last layer)</span></span>
 <span><span class="co"># on new data using `predict`</span></span>
 <span><span class="va">predictions</span> <span class="op">&lt;-</span> <span class="va">model</span> <span class="op">|&gt;</span> <span class="fu"><a href="https://rdrr.io/r/stats/predict.html" class="external-link">predict</a></span><span class="op">(</span><span class="va">x_test</span><span class="op">[</span><span class="fl">1</span><span class="op">:</span><span class="fl">2</span>,<span class="op">]</span><span class="op">)</span></span></code></pre></div>
-<pre><code><span><span class="co">## 1/1 - 0s - 191ms/step</span></span></code></pre>
+<pre><code><span><span class="co">## 1/1 - 0s - 172ms/step</span></span></code></pre>
 <div class="sourceCode" id="cb16"><pre class="downlit sourceCode r">
 <code class="sourceCode R"><span><span class="fu"><a href="https://rdrr.io/r/base/dim.html" class="external-link">dim</a></span><span class="op">(</span><span class="va">predictions</span><span class="op">)</span></span></code></pre></div>
 <pre><code><span><span class="co">## [1]  2 10</span></span></code></pre>
@@ -462,7 +462,7 @@ <h3 id="custom-metrics">Custom metrics<a class="anchor" aria-label="anchor" href
 <span><span class="co">## Epoch 2/3</span></span>
 <span><span class="co">## 782/782 - 1s - 1ms/step - categorical_true_positives: 284268.0000 - loss: 0.1658</span></span>
 <span><span class="co">## Epoch 3/3</span></span>
-<span><span class="co">## 782/782 - 1s - 1ms/step - categorical_true_positives: 284173.0000 - loss: 0.1207</span></span></code></pre>
+<span><span class="co">## 782/782 - 1s - 973us/step - categorical_true_positives: 284173.0000 - loss: 0.1207</span></span></code></pre>
 </div>
 <div class="section level3">
 <h3 id="handling-losses-and-metrics-that-dont-fit-the-standard-signature">Handling losses and metrics that don’t fit the standard
@@ -546,7 +546,7 @@ <h3 id="handling-losses-and-metrics-that-dont-fit-the-standard-signature">Handli
 <span>  targets <span class="op">=</span> <span class="fu"><a href="../reference/random_normal.html">random_normal</a></span><span class="op">(</span><span class="fu"><a href="https://rdrr.io/r/base/c.html" class="external-link">c</a></span><span class="op">(</span><span class="fl">3</span>, <span class="fl">10</span><span class="op">)</span><span class="op">)</span></span>
 <span><span class="op">)</span></span>
 <span><span class="va">model</span> <span class="op">|&gt;</span> <span class="fu"><a href="https://generics.r-lib.org/reference/fit.html" class="external-link">fit</a></span><span class="op">(</span><span class="va">data</span>, epochs <span class="op">=</span> <span class="fl">1</span><span class="op">)</span></span></code></pre></div>
-<pre><code><span><span class="co">## 1/1 - 0s - 460ms/step - loss: 0.9638</span></span></code></pre>
+<pre><code><span><span class="co">## 1/1 - 0s - 439ms/step - loss: 0.9638</span></span></code></pre>
 <p>For more information about training multi-input models, see the
 section <strong>Passing data to multi-input, multi-output
 models</strong>.</p>
@@ -620,7 +620,7 @@ <h2 id="training-evaluation-using-tf-dataset-objects">Training &amp; evaluation
 <span><span class="co"># we don't pass a `batch_size` argument.</span></span>
 <span><span class="va">model</span> <span class="op">|&gt;</span> <span class="fu"><a href="https://generics.r-lib.org/reference/fit.html" class="external-link">fit</a></span><span class="op">(</span><span class="va">train_dataset</span>, epochs <span class="op">=</span> <span class="fl">3</span><span class="op">)</span></span></code></pre></div>
 <pre><code><span><span class="co">## Epoch 1/3</span></span>
-<span><span class="co">## 782/782 - 2s - 3ms/step - loss: 0.3365 - sparse_categorical_accuracy: 0.9041</span></span>
+<span><span class="co">## 782/782 - 2s - 2ms/step - loss: 0.3365 - sparse_categorical_accuracy: 0.9041</span></span>
 <span><span class="co">## Epoch 2/3</span></span>
 <span><span class="co">## 782/782 - 1s - 1ms/step - loss: 0.1606 - sparse_categorical_accuracy: 0.9521</span></span>
 <span><span class="co">## Epoch 3/3</span></span>
@@ -628,7 +628,7 @@ <h2 id="training-evaluation-using-tf-dataset-objects">Training &amp; evaluation
 <div class="sourceCode" id="cb36"><pre class="downlit sourceCode r">
 <code class="sourceCode R"><span><span class="co"># You can also evaluate or predict on a dataset.</span></span>
 <span><span class="va">result</span> <span class="op">&lt;-</span> <span class="va">model</span> <span class="op">|&gt;</span> <span class="fu"><a href="https://rdrr.io/pkg/tensorflow/man/evaluate.html" class="external-link">evaluate</a></span><span class="op">(</span><span class="va">test_dataset</span><span class="op">)</span></span></code></pre></div>
-<pre><code><span><span class="co">## 157/157 - 1s - 4ms/step - loss: 0.1213 - sparse_categorical_accuracy: 0.9596</span></span></code></pre>
+<pre><code><span><span class="co">## 157/157 - 1s - 3ms/step - loss: 0.1213 - sparse_categorical_accuracy: 0.9596</span></span></code></pre>
 <div class="sourceCode" id="cb38"><pre class="downlit sourceCode r">
 <code class="sourceCode R"><span><span class="va">result</span></span></code></pre></div>
 <pre><code><span><span class="co">## $loss</span></span>
@@ -654,11 +654,11 @@ <h2 id="training-evaluation-using-tf-dataset-objects">Training &amp; evaluation
 <span><span class="co"># Only use the 100 batches per epoch (that's 64 * 100 samples)</span></span>
 <span><span class="va">model</span> <span class="op">|&gt;</span> <span class="fu"><a href="https://generics.r-lib.org/reference/fit.html" class="external-link">fit</a></span><span class="op">(</span><span class="va">train_dataset</span>, epochs <span class="op">=</span> <span class="fl">3</span>, steps_per_epoch <span class="op">=</span> <span class="fl">100</span><span class="op">)</span></span></code></pre></div>
 <pre><code><span><span class="co">## Epoch 1/3</span></span>
-<span><span class="co">## 100/100 - 1s - 11ms/step - loss: 0.8017 - sparse_categorical_accuracy: 0.7806</span></span>
+<span><span class="co">## 100/100 - 1s - 8ms/step - loss: 0.8017 - sparse_categorical_accuracy: 0.7806</span></span>
 <span><span class="co">## Epoch 2/3</span></span>
 <span><span class="co">## 100/100 - 0s - 1ms/step - loss: 0.3661 - sparse_categorical_accuracy: 0.9006</span></span>
 <span><span class="co">## Epoch 3/3</span></span>
-<span><span class="co">## 100/100 - 0s - 1ms/step - loss: 0.3009 - sparse_categorical_accuracy: 0.9106</span></span></code></pre>
+<span><span class="co">## 100/100 - 0s - 960us/step - loss: 0.3009 - sparse_categorical_accuracy: 0.9106</span></span></code></pre>
 <p>You can also pass a <code>Dataset</code> instance as the
 <code>validation_data</code> argument in <code><a href="https://generics.r-lib.org/reference/fit.html" class="external-link">fit()</a></code>:</p>
 <div class="sourceCode" id="cb42"><pre class="downlit sourceCode r">
@@ -675,7 +675,7 @@ <h2 id="training-evaluation-using-tf-dataset-objects">Training &amp; evaluation
 <span><span class="va">val_dataset</span> <span class="op">&lt;-</span> <span class="va">val_dataset</span> <span class="op">|&gt;</span> <span class="fu"><a href="https://rdrr.io/pkg/tfdatasets/man/dataset_batch.html" class="external-link">dataset_batch</a></span><span class="op">(</span><span class="fl">64</span><span class="op">)</span></span>
 <span></span>
 <span><span class="va">model</span> <span class="op">|&gt;</span> <span class="fu"><a href="https://generics.r-lib.org/reference/fit.html" class="external-link">fit</a></span><span class="op">(</span><span class="va">train_dataset</span>, epochs <span class="op">=</span> <span class="fl">1</span>, validation_data <span class="op">=</span> <span class="va">val_dataset</span><span class="op">)</span></span></code></pre></div>
-<pre><code><span><span class="co">## 782/782 - 3s - 3ms/step - loss: 0.3428 - sparse_categorical_accuracy: 0.9022 - val_loss: 0.2337 - val_sparse_categorical_accuracy: 0.9291</span></span></code></pre>
+<pre><code><span><span class="co">## 782/782 - 2s - 3ms/step - loss: 0.3428 - sparse_categorical_accuracy: 0.9022 - val_loss: 0.2337 - val_sparse_categorical_accuracy: 0.9291</span></span></code></pre>
 <p>At the end of each epoch, the model will iterate over the validation
 dataset and compute the validation loss and validation metrics.</p>
 <p>If you want to run validation only on a specific number of batches
@@ -916,7 +916,7 @@ <h3 id="sample-weights">Sample weights<a class="anchor" aria-label="anchor" href
 <span></span>
 <span><span class="va">model</span> <span class="op">&lt;-</span> <span class="fu">get_compiled_model</span><span class="op">(</span><span class="op">)</span></span>
 <span><span class="va">model</span> <span class="op">|&gt;</span> <span class="fu"><a href="https://generics.r-lib.org/reference/fit.html" class="external-link">fit</a></span><span class="op">(</span><span class="va">train_dataset</span>, epochs <span class="op">=</span> <span class="fl">1</span><span class="op">)</span></span></code></pre></div>
-<pre><code><span><span class="co">## 782/782 - 2s - 3ms/step - loss: 0.3654 - sparse_categorical_accuracy: 0.9057</span></span></code></pre>
+<pre><code><span><span class="co">## 782/782 - 2s - 2ms/step - loss: 0.3654 - sparse_categorical_accuracy: 0.9057</span></span></code></pre>
 </div>
 </div>
 <div class="section level2">
@@ -1072,7 +1072,7 @@ <h2 id="passing-data-to-multi-input-multi-output-models">Passing data to multi-i
 <span>    batch_size<span class="op">=</span><span class="fl">32</span>,</span>
 <span>    epochs<span class="op">=</span><span class="fl">1</span></span>
 <span><span class="op">)</span></span></code></pre></div>
-<pre><code><span><span class="co">## 4/4 - 2s - 559ms/step - class_output_categorical_accuracy: 0.2300 - class_output_loss: 1.8162 - loss: 3.8764 - score_output_loss: 1.6820 - score_output_mean_absolute_error: 0.8826 - score_output_mean_absolute_percentage_error: 1826.5676</span></span></code></pre>
+<pre><code><span><span class="co">## 4/4 - 2s - 540ms/step - class_output_categorical_accuracy: 0.2300 - class_output_loss: 1.8162 - loss: 3.8764 - score_output_loss: 1.6820 - score_output_mean_absolute_error: 0.8826 - score_output_mean_absolute_percentage_error: 1826.5676</span></span></code></pre>
 <div class="sourceCode" id="cb61"><pre class="downlit sourceCode r">
 <code class="sourceCode R"><span><span class="co"># Alternatively, fit on named lists (names matching)</span></span>
 <span><span class="va">model</span> <span class="op">&lt;-</span> <span class="fu"><a href="../reference/keras_model.html">keras_model</a></span><span class="op">(</span></span>
@@ -1093,7 +1093,7 @@ <h2 id="passing-data-to-multi-input-multi-output-models">Passing data to multi-i
 <span>  batch_size <span class="op">=</span> <span class="fl">32</span>,</span>
 <span>  epochs <span class="op">=</span> <span class="fl">1</span></span>
 <span><span class="op">)</span></span></code></pre></div>
-<pre><code><span><span class="co">## 4/4 - 2s - 418ms/step - class_output_loss: 4.1489 - loss: 4.5443 - score_output_loss: -4.1125e-01</span></span></code></pre>
+<pre><code><span><span class="co">## 4/4 - 2s - 392ms/step - class_output_loss: 4.1489 - loss: 4.5443 - score_output_loss: -4.1125e-01</span></span></code></pre>
 <p>Here’s the <code>Dataset</code> use case: similarly as what we did
 for R arrays, the <code>Dataset</code> should return a tuple of named
 lists (dicts).</p>
@@ -1107,7 +1107,7 @@ <h2 id="passing-data-to-multi-input-multi-output-models">Passing data to multi-i
 <span>  <span class="fu"><a href="https://rdrr.io/pkg/tfdatasets/man/dataset_batch.html" class="external-link">dataset_batch</a></span><span class="op">(</span><span class="fl">64</span><span class="op">)</span></span>
 <span></span>
 <span><span class="va">model</span> <span class="op">|&gt;</span> <span class="fu"><a href="https://generics.r-lib.org/reference/fit.html" class="external-link">fit</a></span><span class="op">(</span><span class="va">train_dataset</span>, epochs <span class="op">=</span> <span class="fl">1</span><span class="op">)</span></span></code></pre></div>
-<pre><code><span><span class="co">## 2/2 - 2s - 757ms/step - class_output_loss: 3.8547 - loss: 3.5725 - score_output_loss: -1.5209e-01</span></span></code></pre>
+<pre><code><span><span class="co">## 2/2 - 1s - 706ms/step - class_output_loss: 3.8547 - loss: 3.5725 - score_output_loss: -1.5209e-01</span></span></code></pre>
 </div>
 <div class="section level2">
 <h2 id="using-callbacks">Using callbacks<a class="anchor" aria-label="anchor" href="#using-callbacks"></a>
@@ -1156,11 +1156,11 @@ <h2 id="using-callbacks">Using callbacks<a class="anchor" aria-label="anchor" hr
 <pre><code><span><span class="co">## Epoch 1/20</span></span>
 <span><span class="co">## 625/625 - 2s - 3ms/step - loss: 0.3695 - sparse_categorical_accuracy: 0.8961 - val_loss: 0.1873 - val_sparse_categorical_accuracy: 0.9469</span></span>
 <span><span class="co">## Epoch 2/20</span></span>
-<span><span class="co">## 625/625 - 1s - 2ms/step - loss: 0.1752 - sparse_categorical_accuracy: 0.9488 - val_loss: 0.1405 - val_sparse_categorical_accuracy: 0.9576</span></span>
+<span><span class="co">## 625/625 - 1s - 1ms/step - loss: 0.1752 - sparse_categorical_accuracy: 0.9488 - val_loss: 0.1405 - val_sparse_categorical_accuracy: 0.9576</span></span>
 <span><span class="co">## Epoch 3/20</span></span>
 <span><span class="co">## 625/625 - 1s - 1ms/step - loss: 0.1277 - sparse_categorical_accuracy: 0.9626 - val_loss: 0.1217 - val_sparse_categorical_accuracy: 0.9644</span></span>
 <span><span class="co">## Epoch 4/20</span></span>
-<span><span class="co">## 625/625 - 1s - 2ms/step - loss: 0.1009 - sparse_categorical_accuracy: 0.9702 - val_loss: 0.1138 - val_sparse_categorical_accuracy: 0.9667</span></span>
+<span><span class="co">## 625/625 - 1s - 1ms/step - loss: 0.1009 - sparse_categorical_accuracy: 0.9702 - val_loss: 0.1138 - val_sparse_categorical_accuracy: 0.9667</span></span>
 <span><span class="co">## Epoch 5/20</span></span>
 <span><span class="co">## 625/625 - 1s - 1ms/step - loss: 0.0823 - sparse_categorical_accuracy: 0.9758 - val_loss: 0.1086 - val_sparse_categorical_accuracy: 0.9679</span></span>
 <span><span class="co">## Epoch 6/20</span></span>
@@ -1256,12 +1256,12 @@ <h2 id="checkpointing-models">Checkpointing models<a class="anchor" aria-label="
 <span><span class="op">)</span></span></code></pre></div>
 <pre><code><span><span class="co">## Epoch 1/2</span></span>
 <span><span class="co">##</span></span>
-<span><span class="co">## Epoch 1: val_loss improved from inf to 0.19488, saving model to mymodel_1.keras</span></span>
+<span><span class="co">## Epoch 1: val_loss improved from None to 0.19488, saving model to mymodel_1.keras</span></span>
 <span><span class="co">## 625/625 - 2s - 3ms/step - loss: 0.3793 - sparse_categorical_accuracy: 0.8936 - val_loss: 0.1949 - val_sparse_categorical_accuracy: 0.9447</span></span>
 <span><span class="co">## Epoch 2/2</span></span>
 <span><span class="co">##</span></span>
 <span><span class="co">## Epoch 2: val_loss improved from 0.19488 to 0.14489, saving model to mymodel_2.keras</span></span>
-<span><span class="co">## 625/625 - 1s - 2ms/step - loss: 0.1777 - sparse_categorical_accuracy: 0.9475 - val_loss: 0.1449 - val_sparse_categorical_accuracy: 0.9590</span></span></code></pre>
+<span><span class="co">## 625/625 - 1s - 1ms/step - loss: 0.1777 - sparse_categorical_accuracy: 0.9475 - val_loss: 0.1449 - val_sparse_categorical_accuracy: 0.9590</span></span></code></pre>
 <p>The <code>ModelCheckpoint</code> callback can be used to implement
 fault-tolerance: the ability to restart training from the last saved
 state of the model in case training gets randomly interrupted. Here’s a
@@ -1394,7 +1394,7 @@ <h3 id="using-the-tensorboard-callback">Using the TensorBoard callback<a class="
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer>
diff --git a/docs/dev/articles/training_with_built_in_methods/unnamed-chunk-26-1.png b/docs/dev/articles/training_with_built_in_methods/unnamed-chunk-26-1.png
index 7fea86b75827cd61acc186745d8a2ae0acfde73b..00956151abe8775f548fa50b0c12311e2402e3e0 100644
GIT binary patch
literal 206413
zcmeFZWmuJ6+ck=}frW*rNLeT#AzcCnAyOhJBBdZ5inNq1jYx@<5)vXIh#;v_f~2$x
z3ex2Q>DpuRy!X4`{b&FDzI`0~<2jB8Z(yzKy5>A%jB$>+JkKf1(QMqik&23nM*fWS
zc`B++hE!DR1lO;_Pxh=1mBqJp`U-N=RLkUl&kGU)si^i*$xEM9cZ?YR;Skg~*(9=*
z$60WOTIL+P*EZ=jo?bgntd%M5C`~nKd@AkQ=rU3!=lkPIsY!ab!RE5NI;z`(k`FxJ
zsk-VxJokkw4t>*dZa1ZkjEo(2g^b^d<nI%Sp67OW_+#zZcUSkDJ3OTi;12%%V)Na7
zG{}M6+P_~~{7P;A{u|#l>w;+h@Be(f_J7GQ6EAgapt2a-B02N(sgUJmr-W5hRNr1p
zOtl{Da9>^=S6)p;m3>Q&+3A|%(_=64zvBKH9D9N_VhqatdLq{q3(2bqoe&KdzOMTv
zQv4nH{q2$FE#kxHI@qbGiiP&`@Te3p;4N;LH>a?V?WD@qq^lIiCI0=o{W|QW#lQc7
z7jv3a{*M)^xch&8N%jAmML1PDu1rOB#k}#Yc}qogwG#dEcomPqua6JHc@I`r={k?l
zO%AoF&`0wdD}Q;)Yrt=`h6K*zNTcd!GF2g5YGoQJm$rA8`F``ESJgFZN~VhxvpXwp
z^t&QJ_1wAiFus4NZgYhqpl+Xc?DiFok@eln-d)aUqnN)iUWH5W)hDXPntaQ)tegpA
zS3AJNlO}E6VgJptK0zfqh%!G`B(?N6K`PipiCp#exNU>{k<v(qjxfHZ*}iDGM_ZVL
z3}mmF`AbZB99DW3=i|RukfS+OC!BKYwbnbot&%hQMxMU68!BsvSK!yBMM_+F_GT4T
zs}-ZmpAVFUiP{98=$n=7yw@roS3liOE<qI(S;d{l$({4_ZV)BXzMZ|dI{Y0yKb~Sl
z?g3e;|NgpS{*oO@_V3reP2`>b`|Hi~|K~61sz?Iz1qKGv`Z5Uny;KU9RaI5xyp+Dr
zs5YvgZ7Nl-$Za@(qDD3MX3tY`2dgjP!aQDHUP4t%J9&hKg$G;GSZMYNcnmhZlIy7q
zK6vEFkq{m|W^90Wet!O?w`Lcd5;S}9+jmQ<>gqc;?dIkTZE25_4{UYni#&)oz*U%R
z`f7fp%%iqY6=y#`AAh!<{N6MnQXlU~>Gqc@`}km!a8*ZcZf-`mpSWz`VI}?QXM&VZ
zTcs?r3qC6bb6kiyd2gL+nr=bj<+o<)N#|KNZrrHf_Qo*7Wp;dqQqB_l{Q2|zubgT{
zF4b?%n!Mr_Ld)*1-K1Y1BNIN!|7;^2hqU)Du9&Nz9%fkmc#mL`<<xm^n>n(TD*Y{~
z!soV#*$wW*E6?uL$~02)VV1bHIN>_|L%2qKEMMf;ecBk^f?Etei@zSRbV)7Gs(imS
z^qSYWCc<aR3O6BO-ppyZ@Kr_1x9OFJoV0YF{YV!U)H_gFjNSgbbvaELmv+uE-CG0k
zdqi$rQH(k+8e5?C){IUmTu6zYMcm#ZMLU<zFkipa%ew9)b-Yh4cA#)>@YQ6kd1K=C
zNO1?<<$TL;*`^cbh*sON4hh2rZ=OPrQt#afItDiVh%npH@1j$!2L6-&WP2ZMOuBww
z4V%GlT;t<Od!($2r-<?Nqb<JCZgcECGuf8kWMgH0pOm3t`Rw66(0%uM{Mq2H;?-Lf
z>7x+<iqVo1u~>^_y2bfv|9e{{Mc&(w>_zM-KkMv^c4r<doVPl@`1dQNbhqBMEPLtC
zg;FF)LiOC7{*HCJ$BN1^*{SD`S01Gw+eD)B+mr9#zxTJlHLsjmn4K^fY)slt={$YX
zPi!zzq}zu-MkQL3#2!tVGFB$ac{Dc%SKyQMd2vW^=F=8&&YI&BUql8IV`FYj4l?*e
zqHE><`MAk?d2zZtM#h`3-MQ0s+V`2D`C(sV9=GJuOn24<gMi7lk@mL=^vi><()2s>
zora>vuq7Guqj_oW9f-cHA%}qmR?EowSbxd(16!_<<r7Z4GnP@#T%UgRQ<hD0iYBev
zOGVENgYvS+`%ml7{P`7d=FoY+v4;36S-tC0TG<ww=#l%oT^NKc_7u%lUq4@VL?w!G
z(=N^mlh=#Wol?pO9P-%1nYEIt8C_+sB@{w9JxOJFNW-wJ_V|Qiie~2FZf%9}#d>U8
z&S3JZuBQfmo=o<wSMO!m4K{i%EG+m;ucd0`K19W)VcDJ~FMa>fqw<SWmooHa+Q&w4
z>tZ7d*4<_C5#k6&Y3V~|jfnw{3zTwN85x<<B~H!Un}%KYX_YJ3rt?PMA04^;`tsYe
z>M56at;?Al4k3lSemG;5JhiNHSyf-0dZ0gDh^22xZA#V2%bFN!ObWDL`uX9W@SaZm
zK!1YA6A9@O8-ddNHp^x-Z>*4Sd5%HS#j!cfaG)XHQ-P;IAOLm7)_>LNbwM))XD0hw
z()8NjDvOU~>9-_mW*T)(eY>uL)XS1i%S=pUqjY)m9KM`oqH5I1vOPfL_v6z$oSdB4
z%ju^1i$8rbsJfM9`t_x>losmmZ@jYT=JIsGqepUW$5V5<ys|pSqFtvn+h+yMn-#^E
z!;GWkrKJ%NyF686WexxG_u~E(vMj#|#V+tCeGs;2J;rO%n(i}IQ&pu9;n?L#KeE_n
z938K1jE+-UBQ@6_`?dFGv!=1-v85*p+@JdeuO%Y~7*dL`B)h~Dlq1Cqmjo~EdZ`i}
zCGxlKwAv`;yG?C;L>~S~`p0`a{+em5h;r`5t5>ha;89Cc>tp5oOFXt8Ji8#)_a!WF
z&#@~f%5sgy7q@TsB!xc1exyCfK)5tI4}03{bZyRohl+K|J;}deHP!R)0%nbDvVM&8
zeGxXwzkmPUDPD-^Cw)V%v(PoOBkyc52d`~^-AwfIY&4oDm%R(g;b-Qlo(WO6I7Z~F
zEi}zqQkkYZ-Q71c2yt>;NX}ASz#4h7Dul#ExTA*|=eNioi45Y)wrG1ZGhMVCX}?5z
zE*h%$tJL=#cy%PDaQm`^Vc$%*zduQX7mt`P$K)n4-s*ornHx$=kO*cC9K+^NQME>m
z6fH0D+7Gvi{Jg!IZ*LJEu;dYo+tVp!hp=|TP#yih{q^;E4r4xLWFsDKrr)t!*YRqz
zbLMQ8SrbPGT1C$2`=KXg)LR&z-n2sTiE&?Eq8}km>qx|nQ<XE$V+Ff2jcb+ZYpbh$
zNxXL8Cs_`glnnbK9sJ~2Tm|r^Z|Y6Q;_ZIewUqvF=#2b3aW6XQ-fAif#g@%f#k*c&
zqbP_X(z?$Ikk+=2X0LkJ%O*YNnO~0-n!dZXn<lKhu&KjS;NA9XUqu%FeCiI6V{Yr8
zOTO@`3vEV4bgbf_zYmkc_l(xbR~KLJU7VX#pl9K`b|yu)K+G^-EkT+0*57^;#6yO2
zyc+tI<3zRa*WM?d04=4Pc=EShNLK%B#`IrnmfIIa=?%XwTX1XgtUww%ocvH2t16y;
z1+W-ag|I#g#R_bd_<N4g^>2NAhZaq~%XgQFYU{aygzL7hSX5rC?>Rve>o)IsTt7$6
zc;iI7-maNGspT-f5WG-DYSC8Prdm)_STq{=h0?5bJ>D{|Mj9d1{W9FD(u@Y+BA7rV
zE=4?v!C1jGchh%*JxI+>nHod&O6psp_`CabE{;0HN2Tr4xrNd~L3t}j5$!_j_EV9Z
zJw&=y7EP^`IKnjs&7MKW{m%#LSfuzsgcse>G9;SLaC?FCv>%Gwe#7HByzi0O6x3#*
zGQoz?j2pi`&iwtVV({_)1_~N!JY#5hI0L!;VHD)j-pEQ^F}j{wMtMC!I=36quGI%u
z%le6Qd6?j7lBNlYL9%D7+t}Q^*@aE>oSPg9H7nffXb)7z@NXxq2smf}Fi9DC*C~rc
z&9FgokH#gONf77C)5)`^kNh1qY85a$J4=!)h!TFiP*hp8Cn#%ZHyvFmV4{9wqACT!
z70=|1R$bb;yfpoHLxL*ms^i~!IsRnpS1SRhCWL^qnjCDR;K@txZ=lsjM3nl~X0y+S
zHow-|t>;3KaT))7gdm)q;*!8rqsX80V+s3o_|TfYUZoX^i8+iiA{4xI9DWG4zq`&t
z&18Fy@ka3~(koi8K4OZI^Jk_k&5tGM3}GER`rF@ix-p~`&Y4w;A|-S)WLjudj`JE0
z52Y2Y?2+LD9@U#{?#M!vgfoEv`sktlC~PwK*3on!AkMZa>H@sXluIoI2=&=cwVS8W
zL?MX93IUn-@Ea)t8`icQBhef55v8_^U>;nOBbH*QNnKl8>rd$}FQ1`wAC!_RYNZgk
zX`WV)t?rz}67mHdx~oY;&v&9moToj@lvdEJ(c|Ll%QEGDOtI!IsTrMa^D+~?)v*9*
z)|KoDC11k$)l!?#i$$it+X%mx1X8X*=@uPB5mX5`P*zrEbYGybcb|T^S>EYs1*u&v
z+BQm+xuD^rcvO?`Io1ok!bP5|w;n$RT>q~1deLGRppc^I*vCyg1mne3iwC2u$^dT)
z<zzcjx}F}pdP*@ugg1m+=W?>$Dyr<=|3S}Bj&gGPFkH{)B~6cXUukJlpx13>W<*dc
zi>3ixsUnLB98Q{6KD2{Uv^-yAjeRJ4V_4O!C)NM_-bzq*;*LzG$$#pDr?7Si2Y%%|
zyMlrNEmPe)2qpRYQmK-HN}{CYqLc4v9oDOHnSeK`g<o5FO&j9$%XjVES@P%Cr*8Cr
zFBEtw(2NjUZS2zlVPS_@se5-;Q<tgAF$LQy1N3%1KE1`$q0?2^G#}8p{@wM?pxk(I
zR%i51PXm7mAww4cvU&h!-;KpbHBwiea(rT<8$e86>>W0y9#6OjO}=u{djjEFf<0tf
zo@*%*67kIkDz-+38vRw+OoAf4Kr83QRc#kaDWySy=XILWD4-8%^0I`l)rkKoUY*(D
z{P_v1^+0@R8OmY$(up8+vQjjnovW$0_#mz0BAln%jLYA)=rZ7TIP0SfbyoX(fS8P~
zv-?AIijk|k(OCUJ8ejS(i2c#G$(wA}>K=1ncK#BLsLQAd<?-Kt`e6we982I46+cw@
zLrg8`J-*6ih;GZxdj;4GV;9LK-Ipe!)r5u}M%wLS5BKVw)u1lw3DIg`b{U`VT$W@;
z$@NY3r{_QHHl5eGlEv?S*ei%lKV-1-ATX|{J&YuUwo^|quYQRKO(8e1%=Uz6w_7`R
z?!5L^5@hGhN+lSK@#gkIdAGd5#lyqnC$%*5RMPcU<pI418|n1Z3Z}eJ>9QJ~_3YaW
ztpR2&ZCw!@^+2;E9*5slakKXUx(~#gRdNsMj{!BtI8TobMMtBAtWZftcXXL@pwvo%
zM@laQ5&}rI!sdp&Vb4S3sef_$(H<$cg>@~RXqr!tJKe05T>wOjv#1>XnxN8{qRDTX
zlAeAn3fBml)xGgyZRxJquaOli5E~#rV*Fk>iN1KN$|ba`7_E0!g#7sGd++9{89Zb~
zw!T%Dbii9v!}JKOwskE>lq2?(v$(sEZ0h+wHS$dk37e!6#m&a%K|p<(A5-3kNxPxu
zQvTZGwERY2JOU2N$y!-iHOyjX{SNvc=d&yj4O6QBa~zd;kC<)$T%Z5(@iW@EJN9xG
z=iv<A>mB)gwXrGj;wQy|K}8*BdqTT`@fG$0qARE;p1T?=^#~Y5B=!AZa-w=tgRD)m
zMjCTMr2XaNz;E&`iVp88lGGAjE~FUYhhF58J>P$QdUV9HVyW#3t3nsq=OfWBe5Uz$
z;*1tuXG7Z@1wQQ8Wk!eZ*76K9MYFx(IRh`l&%WixUZ<XNIs$@QMi_|i-Z$>q0341Z
zZ(AB_uA^%tO-Etj+h2A9q0(t><{DvV)vDaUsPzi*zf$+mepYJWJF6cn+sh(8nq$yD
zOCbBL#XHo@XGHrFWBt6q9S-}(AC)aa)SIdVj)HPlwgM*$iOjNw#l*zy5wf^c>FCca
z5p(K6!{S(-rg3C&*8UZ_pyIp4^!H5F(^IvQ^TU}${;SMwO1_}rv`_jlGgg$`^gy7B
zr00iw)CoGV0(Lh9{|q-FI#4+15IGLE0}Y1v_Bu|F{)ol1vG<{~iknKzW05`Fm*+d#
z`yv4><<m40&+Yfa-3LuHL_Ud3UB|qgUFzUme3#@=F)X~d&cU#?n|74|jD}4~+fh+H
zu{lrLCEJey>h0SRk17mlSc{}-H-J=c37y~Zp}Jn~3*9WLK)7)f!d|f^`m^IbtnGPO
z3C}Fj2Af|q0KRxjy3R=hr|Xj}b-JnWa)@;Tn4;YB-H(vJ_L0FyHT*Jb#Ey<4Mrcx7
z5>*e*{T&FrnP$skLQ<?g-)SO<0ytoGkG4)tesYqY(%JHUy2Cj^!tzKMC`BoO@&eQ=
z!G7*Q09*EeE#3g|#5vBV1^lQ)us>P?>sspPG@9Gj1#+N7--(uUHa~5z<7dI^9moA7
z=lThzjK$-zj~#fHMD!t0Vl=lKKZ`{M;sT%44Ai|i4L0u-E4ze77=L_oJV{Ka-moC2
zxc9GmYH!}W$(MW&3nav$ks2Y;jN+VwoAlP{ZXsKE2@S)uZxt2MfUbB8xXk>?qM)|F
z^i2@o5F;|6Og-BMsm7o(h&=;AaLBjs_wN#HRU*4HZ4z{ug&rRF6>@^@y<jwS*!;bp
z$0d-+<CGFw>o&tsR^MpsFoU4s^vE}1WK)#wLRYDGcd&-8d{{ZOk_e$%0xj@=U;RFp
zSmlje?0Px8ywE45`Cn2%9$`l4S<&`iUMAloWeVWPi%A$LHjC0AK>g~DajJaaVQaK&
zndPNL1=p5O*UmS6K*j7Pwa|9BmDkjz`1UFWSGOksYS#;AjTS&p*t<)cKJd?$-su|7
z{$>^bEMdzQs&lstr+GA}s47dy6Hg7dnU<q+8U>C>p>X(=v5L!gg`SUmuV_zr()EJL
zi{Ed2qA_fL%(PuKLw~Td4LTjQTE}IZ=^qZrrdt*_xSBi<fDc_g{G;Ut0F&3}EQ=tc
zX!Un$Ye`J-34@^Pms=!@Il8&Ip%}hzwI(7N(Oe9*X<WXVB1l|-+pWv!je|^2+e^=h
z%@p5RtyZQiIs2Q&Y#id3HN>KL#^qY~C4Nih@YpT7&m^G4p??;(uEFHMf<?d(bmnRZ
zm)6DMryoClRLWikLg(MBp?xx6AR!1mG0r#P{o2l{-yf)1IH$57hJO_9bfHZvTDny^
z1J$x<smL6%&I)9)|MT;M>2zgp-1HN+>PLscyTkWErbHz#{N2~LTii$uEBgA5TwK~c
z4TMIC^KOmR{^zLE0NFM(i3vOtvV7mRlA;D5?yi+L015X|Qx-K`oWNR1e|kt0=%)i>
zpATYSI^E7__otz;IFv^(YAO_B=NX>KG^&57&T$h7f+=vYr0Y#OnnKX$9I(?M$1#+|
zc;A*+8fhiwRtjSF!+S{4&gj0FrprWtJObM4&j%Sm>R8wLsrU}}pT&3T)AWksG6<j{
zS`={4N<B9`>C_M>&+E1@lQqw&o1fDt8X|3~d`UAyzw~g$ikzIy(*xw1wJr*Di3kbw
zdhcybJ*T-*wO(vlA}|2O@4_&-u@Zz2G&(xSK+SrHAtzEc;@h2xQVH#?ba~!=nP(+B
zc}?n`d$2881(XUdU-uKs*(*X-+?#-uyiJp3^+T-8hd!`A$M%V!;qsef&lEuuzLre`
zYwJ(6heSe|Fzv=}tUx57)rh}qLIJZ%wXPRj7+mZGQ+_IDXNH1r6nhD}XxA71S|4a;
zM=YaCf4t#6pbs^}v>PBo$7O!xU1_RX#8}a?<d7iQ+1sxb2(v~K@H$ToXLXaZj;Ler
zPTF!<UW1xv<(YN0nOufjGf1|VqJc;JEhr@TipPCcA!*vZ|HY%O%A;vMUUb4%2?LEu
zY9u2IXS%#adY>AcPgUJz_h+lrLi)@#Z`CJizgAJPF|4TGCjd^n1`?u`#X(+#i255u
zfw)CbootK7>&wCRKgxU=5`0;leg~L#e|ry367S1m^M#ieidGN>B+)SP6L9=Cz^*F|
ztq5fnf2k!$Mkr<S-9yc<y^W(?vs)#g%`gM8DmjJM;4Y3468j}^Wxh1?fUF_tYLxvK
z$QaLrB%Q{8$F+QI`rrUAm587w@Wu2vnAqIZNVt8cr4D$BTG=jc?L89CQ$D!aAcHrr
z@6|?%3qRb#w3mny4Kue_t?udx)$;?@JmO0<6<+JE68-5}%rUzAWBrxsrO*!ypd^N|
z(xAh|{+{MoMRg_I9txctsLofvT=1%|Wm2=h1LKaU#~lrll^zC*V>ci|=G1+kuAL4X
z((koYtu{0WzUI-*7qLXyDB1sr@rm6Q6^TjpN~v;mbbi8i5>dSyX26!nAfr_R4@7rR
z&QFnk>_?tM<--w0)P`#=-WZu;;__`GL28kGK)yoi6{K_kJNGkrW&QQBfx0Rh)U}^K
zd)7pXV?2STlK~wk?#DY3EgL3+^;J;_f++ryb8I8?TF?ve8o6et&5R<)42K)lRD6N<
z#SL9hSm;NH(t`c8YgDCw9e5U#nK#e2Qk@te<Gw4aFTA?w8>)MYeXo$kQvo-qKc%95
z5rOu;C|3Lfljxa3Bl#$LBA|@f^^+L7h+@s3T{vhZT{7N-1^Y_(>D`(K4<4`^04pjx
z4^;salqDrZJ+s?#^kT{x=b=)YS_xr8&Oj}prD4*X=*?TxuO?YZg%Ld|%28kOnE*S8
z--WJOvLMCTFVK-m(#H9=(H*_$L*yh-RNPTR^9S^p-nh4;q#IkVt7&Z3rdB~E^vH7f
z<KrPh^bP<z%|1c^W}x&wDMRZF9;XUZ9NPs1@p1`JC(cLDb+U<FDnUL0xksrxy|ol*
zN~0^6is~9Fq1!)H=|9>L!Cj#JUceKaXb6FXN)EptO11vD;fOS8s5LtHNoC-^GJlq6
zdn>};7iK8dkZj6IJZWQ*%ho^dP?sS!^bQ|k*zIt30_qY2Aw$R%5fIw<A51&ZM-5yS
zCtFxMGNHc^X1Al(+hYbD)Bx3VD4K*fp!Q>13m|cA>ry&v5R2<O%|S`?m*$2vh2Jg@
z1B!YOrI$m!V!`3~#Q3;A)|TGi9XzlLLWCd0F<#RJf}N!gYzGwkwr)(7kLz-l$I{|_
zIlXbjb}p^#FCr^A>dH4e!u3{BGj3BkPLU>)5#XB`lnCn@(Y~GE9!o$>3sTBMbI-FI
z+Jzak?<3}0Y@jrBWF&RaPvihmXLdActFfnUaYNOU$uz1a9Nz~aX1OgKm6DLc5@$1x
z?xKIa0L->^C(!`BcGQOdXWHSqIDz~ow7&9dK1vcJHH*uQ{N<&2hEf9A_)N`~+t8`I
zApr8!M}P-#S;kKC(1C?RMq&oHc4%;fY-)fJfi*-OqKq}9Ucg8L>eRt5Kv4fe*Lhit
zx(vpDe`c*@qPqR&1!)Fx04BRNbnGs8_DMi~&H%iQ%LwB)E^E<qXF>^37A?ZW%6jPa
zyAa1tMa3{aRx)6GBPt5*lDlYOJO|Z7*h>P~$Qn5m>{yH088-pm*wmRZXCkUi)^=j7
z-w#w>NfG{Gt2PXoiE#7M2vM6D2+;M=R7Eh$i6t$!a(@`PI&_Wzp-?^N-Y{b?%c!Mz
zEM+N;@b_)Vdy@u7(75lmUzK>T6yJ6Pizh)C5E{B0q6>K%BKi0+iI+i_Vu4iRTsHp?
z$*^z!D|ITT+iyp0-s03ZwXzOO(AXfVU^EnDK;$h#?mVZfrIU>)(U6ANmZshnvHC`W
zl2Xv(f*fl#^jzPT*c>>(3}BY?`-y^3&J<9NbdlTQ6F(vf3wJmkT$-(QF9S~32PUaU
zyA~PE>5g$2>%gN0_eD8#lQ$(|7hjaxlROz&X+i|8KeL8PqZ$X6VY<dZdqmlXs3k;R
zCM2CxukbiBnvA0TfxA{73==u!wu&w2>G3N)J<j|xcMasKZltS%xKZ8nlk6&PY&ED}
zZ_K!L-&yInKwvJ7nktCDg>Waz>sc-B5mK1DY=qpii-dswcL*9Lh)GC9LM4zP90nMd
zy_}3?2@VKxv?Qx|Wgh6^eIf!C)jY^-3536ErMi(3IZ^oy4Xmxvq=;kz2`3i&8?PIC
zoWz!CXI^FS(ekyswYTlozHzd_QM2iTJHRNB`nF?b?PR<AYNJ`M=Tl@or~ALXyY7M>
zrHUJgGq`DE(+zay4S<#+ZN8f7oaCFLX&6_&1rha=sI|e4x`4b9P$M#K_EctdlO1}>
zXL#1LWf?u6DEX{W)8^>NH=50ZLT4xX$|qa&^pbNSvs%Mj5~9hCl5bDy61^{zl$%S&
zI}ODiBAGx+d9QA>GTU||%1O8EH@$EH)AK}OD)r0j@;a&y@WiUSjM<Ry(bSw6fuz^(
zP}d!ig_)OQj~Ae3nTB_KorPpr^_TxPDIOa)ZQ|cMkC2i+dGe&^G`Y{eLoGbJI4_+e
z1=+%hVC?ieq{)Ha(PCux*|CO@9I^(Owb6;&1Mpd6dCKubk12D&YW0Asj^GLWQGu1<
zZw3{Q_iLu9>5djmcaT};Spnmvg<0}PM0*MA0f6j66|FF_kiXtGqku+6I6(=RIvwgY
z&!kupYKByB|Ii9+Z_NfPU_vd{kjC7;%!Y7xGAh%jbYT(5I5dKEYD9)l3kz0{5#?=V
zHr*F@6Y4|R-+`~0x(WFaw+54VZ~m^>QB9ADGnwI#8pu~gm=Ohl#aJL)z`=W=4GK9y
z)nrx>e9Kqw=x1Z-y+KUGoMiDED!l&|Cb1IK|DdUzq`*Uc^%SyfFTqwlqs&3I^hKvB
zA;MZ#Hxb?mOv$?5DJko-Cx9|nI?!3<zA%~Q&h@oxxc4sx6{5-#lZ`WwRU7;uvgjO#
z)=QXa31;+_$NeEhltVym@I&8JI=;Lxp*c4BJ5Z&JJ8yW$@`ThfEy_IyWq}|DB3_Ar
zrp6+JD9B#J$Q3q|KKULq-&nkNDYMWU#6}LKjKQF2aq4KjX&F!9j6UPdiv1LH%><@*
zsJ}Ugm{^5SZeEAc?>;Rm#!wkD&<M(*RX(Q3*J?e1iXaO**p=qKXrA%N4pgbUTDX%D
z_`&zVrrjY<+(ZcRX5t5-ox!SRwi|;eSR?=CK>xb+mgh7)_s$($nzx&)mt!eK9<X4T
zs_}QQ>F6sLDk`=}ubxN%l4cVO4#`v@h|;yuy}R6BrEOb;ys9vAKoSB%5QW8kA){5$
zYk~}7(3V(8-6ho@ba1!QGXYN`sv<nMLD}{K(n@NS=|UiKlXgb*WyHi0Qglfdy?giW
zQ`8R^lqL#!s(Wg3pk)n(qbtIG!KO`{x-snYhY^CGj9Xe!9SvG^9DH#pIS`O1HM972
zpl&qsk|>F>t_mDF^^;gpe~dYJvEp>-;vO)w$YIoM0P&XycTYvE^@$IMA!2>Owa-to
z#`LU}I*PIpk$a9i-q40sg1vmjLJ-Lq=t<lhn1awb{IJu=vS{ObZ#%GqNzB#*uv?Ch
zK;VH4T!%F96<bS91U(THH_#rw=E6~d!u+*RE<U%1jnq&JOZZVDvz3Jm%6zgW@Ni#i
zqJa`JL>0q@ECX8x$;=h?-H#wj-f~7-0$j+<gxCQv*zzSp(9yK~^Mk8NNN;BnxA!6Q
z1z6zHO`w@HzSa^(o7jc6>bL!%Yzc-hgpue-ktwsO-ZVre7#5JZD?ovS@gpYtON2!2
z6}DO#_h9E(Bl+Ez7JO5=(0eQ;+s-&Vb3m=qc3+&du9cWxncl&y;RD-|G%iW__lk!G
zGTa_y!qmF4PPD#;-tg#Xy%<I^*ako2i9u4cmtV7aSE0usdh?nOqbMJdLyHziu6N&F
zy~Tr!nJ@{Ug(U0iu`KkOM=P7N5TQ@ZGA@%%8id;TFyv+~`fSAfjhG!{@&hsWeM=_n
zA8Ai$<_@Libs^+c=z+t2cylFws<p?zR%FJR?Q)FU(Ok-mzN~-#n@oRTtzaJ$Ku>%1
z4i^dSw=$y&ugtOTjzI~LHy|qwrfAjc)B22PiYwSW1N^*%cnLnNVyDu2Pj&*x@mz-~
zm4&x%%DE1DY_C?DZNmNKKcBV|{&BtRKUTx%7V)p{EzkJ{WBWTvd8dlE$E-KrM*=uo
z_0pvr`plIcbNfwj>XxMvS`~{jMd%<6ro@|g)(~s%gZw|XqLDL$p;kxyYuG}WmcYy+
zNB|`;%Ms%qAsbm-<TFQz18`4q$k@B0MN9QGT?xQieeARcm{qyVZk?S#IA_F`pso9W
zp1biIZy>Qr#i{|&Nc=p~($KH>gC4DrPblR4JrRWD@9-??J-Z!5{x&g6KEJe5U7;w%
zW+2FX%~NxYXpL874JyM}@HGkm(zWADWCBREcv?8<iklNxWNIo4t<-|4_8I8;mE!-g
zp|aVmUAxvAWk3$m=9KUc7^^JC7gXnr$efEL--=rl_i31l$u2isIC~^++5YD*6ZKcb
z?_NQh-Y+v)f4K9&`ROySY7vVTDZ1S#%D&PER67P%p7mP>nH6F<{G<%MH$4o0$qUnH
zPJfJd7b+^}m^!xL6xo8-n&YT>6ufDu!(IzRs3)|5Yg)_sxQUCf0Fqn-5iKWT$@u9l
z<lS#q58U0v%{I2PwK4^zId-)rO4<P_DJi2bPq%-1wACLhTO3HBiv2X4dFzNqaa%tU
zUAQN51+P+B?7(cpZ}dkVE{P58;Rqrd`rlvwUvTpN?{E+QpKJcFvTZ+D0RjW4#$4|m
z#&l8SkdD_0@`S9n5|k(Ql8HtUBzqXM5u>yuwgH_ikfp`TDr!Q-yHYVGw1)RJgamPV
zX;p1)3$dJQ<-iV<=(L*Z1icgZy`bT0i<AE+z)5odZ?WW)q_UVaYfhmT6%~blj@Etd
zYjmYJlTa)io^oKT&C9dP3nJqc2dl&dZ@=y!j=Z_qXZS<d0=UK+ao<<Mn=#m*TxRLO
zsCQiXQS~q(nU@3U!3UfY)}>dA4W(oX&E;xAHX>jTKP$fYl9cS?=2y?3?}NZf2hH9a
zGuSdf6ljP%D44`ATaTXH51p6Mx@5a`#r{XqAY4`AhF2JDK-Y->Qs})KBFt*(T5Bqq
z#1M}i=?X+)Bk<u8<n*1pb`hJ;N&_S~5Q1&DL(#m>-_al95Dhev3()I#vAFy>Nv6dw
zl_U28)_W022BRm8P7e}X0@cU|vT!MU63j%OArzQ$l-n7Cp9dlA54W-~6I%`3JqJTL
zHFtn*`QgpwkTS7E>w>)1|2@}+aJ^?ZzK~qd3%*NQ0cP(4<eJb({E5mBg)tU1Rux7V
z6ATY*2OHV(+Fisst`0=*qi5YVKijDfJE^h@k$ep+gDzpxL4<;2IKqgn1v6S%JS5>!
zST?y5=(FLJmU56`=64;p%we!~y*IStmmU4#{L0ZEJ9yR%+7NNol|y(SPB%Z|t}b+O
zYE09U!kg}d9^}_i;4C-RQ4k9n+7rp0?i(iIO#H=63_4PXj#xmap2ypZLkuybAilCP
z=!tt*)+Aa=Y8Th#on*2OoxL314*GuwrufdUI}PQ6Xy`)N^=3VIlP?fc12H&SdWHT|
zoXxT7M6MZl!+`_mC9QIPJselIJ_sL}CwWQ6H><q(>cvoh0HH(BTl`9}qUs$3nKa3&
zv1M~<Z>3-$>xtYF{uyCJi+KaXV*UUYJHv*U{ici|Rlg;+Px_t=Sc4L@jX@FG3Kg^W
z!C!ZFGxKo~>*7_^hsfM>N&MJ>9iSX%H}5&NZ-t>v?S=oM3-dCY80;}?`;h}X0NBqg
zbS^J;9(EHxn?OvaK!Qj7;2e>L0;jOh4v#|E9#CLeXIGR`Pux9M1}3PXj^F_>GrS*o
zSpwxoh$4PoOwVoBunT5TdBFO2C26lwjXmL8+}~Xgg)vRs(08}_k?=osAsv8h3JX!N
zpF+Cqi4-H%7*{QVze)#BF|N#?1b}j6i&+12pJgeybcC%xtlfRc4I;w0a-`)s#6z?D
z>k-HWWS#0^j1T#1tbAB70}Ll&G1so?f`(zxCt~QhNfZ6SLJP=WeYn$ss5P?Y@pJ9I
z9wm?+gega_+7Mf^KAcZ0TVstJJ)$f&5GVZz#}#J2Ottu*-XePFl>~SXEfL%@q=www
z2tKv0;(8a9DD}&Szvj&QW|$tHTo?^+hZ3mT^=t;Fq)Kti+m=ns7uHj;-2u-n8-pcA
zWF+fr7oM`(kvxAl84D5C1ectU&fWN9j(>jASe+IC&l*=Y&BQ?WM|05<G~iGNvBv?#
zO9lKT{N4?7Cc}a$BwaTlpUp$*mQGlf?v%_LFRYmy2JQ9VvF@eMPu?Skkci(g%b>41
zeD?}-0jA_?vZO$w4__hw(3<3n<SM`Z{u~$!07`)^NE|Z6{BnjvhpMW!9^*21!Fskc
z8T<C8Q(a8ow`=Dnp>w<!)j7BGUi@q9%@)FDPb=uzT71`plPxjv`Kg`T8gCzHUA3v<
z)nB0^shi(FrHf}~R-Jc^w&{(Cgx_~sx$REDSapHE|JKc$r7^I!9{(*L@J2{T=u(lJ
z1V-V9vm+xS7;|!SU%=&lMoEdz*~QYp(C`5Y&p~l<t>FxeEz6J>&}vU9PnY}9^WQge
z`<7|Ewa{&mBir?#J16#STfYY1Oia|()vtg!_IDOZF-y9zHH&QBx^)jd{V%Nj+x{0b
zhYlZZ(C;YsXGyyrI>@+V`iOvRS6A21?(Tzik|uTgP;2wb=C(2)_lS#&+dH)1MK1*c
z^RT1H3zCa0_%*|pEn8aO+a2ZB&bi&+ZvqQbqgTP*yZ_*bz&|xLHK&gDZ;VY(-<F!1
z3Vo2Lp`k&<*V@{8dwZdEw`@(}j+MLIWk^fD^HV64^sKC%>lL*$HGg6Jd`erJ@l9qX
z&gI;O8cS{&v~Vv<xQW=hEsP#ucnPULj%-`6<k8xCX#t|fKO-Ra_ujsJt8U@w=(tZ(
zQkUcX`uo@8UcUU;+#F(4mwwpM+1UWn%3G_1RCXSoa-8<S-09xscasaVHkY)thy$_N
zvPlpo$x@ImnM!sJj*o3^n$IQp`Dr17J6q<+{@2ptn;p!c`wuHect|ZT1U7dhsU5NS
zmQAA)b^N568OPbPXa7Ji`JU&n70S?qmoE?NWj}vTg=))M6tI5HDPJ4`*~D@5=-PG+
zxVh;TQ5bL7*aUiUf1Lh=LjH4PM5ohgTcH8s9p+#YypKu@r8Zis961{h6~%o%+qgC=
z{mq*lCJ3UPLp>5?*Ihav%e0Y=|0(U}CJD*O7Y%;=+uy%`e+~_K|NMDHL|pt+YwMos
zaN$Q@UK=;l(7+pbXl7>S=jXs9$N2fX{i|wf9=?9fr8#_N-?l?WFv*_c=f^Na&rIm7
zg2J!CK@LIJH6Mk~pFf`tlqi;AH}}_c&6+ihp0D4>;a;F(Ef1Dl)YjGxtd1xG$H!)Q
z1Yuyp?RY%u#5`<Ls(2|w>h_g=Co6d>GX2Vje`e<AC#HLT|Nc2T>bILmx45Z^uD7?h
zuD<>bLa9C5(i;G!3mfGKT?k#jj739J(-Si{fgB$^?hDG}$B%DA1l^QnZf^dmrbgM!
z9)0l7c#lHTz(E%li|;wAF_n+j+!eTCXZPCrNk~qP$cxW%_df`<<RGy2&XRPwKSuqA
zEu8!-0(KZ)1j(n)makt=-?(w?u;SBwH!YywZDVBIj3Vf!rK`(~+j#iq&6~@V&=dOv
zbP(7#aGK_hJzA28h{&0yW(d~X>F8K5MBXp+_wm`Bl9G}RLG!rF%r<s*b~Y|9nUg11
z=^GeW|14gGvHZge<$!IX-vWb!OM#tk{I1wuQdWkVzZ*KQho4^=nhOpq6^pq+7K*TJ
zVm*BLFdn898%d#b%K<CJ<mBw}^z;<LAPTeSDlthuz8x%*E<Y{OP(xY^UBwR^JQxrX
z^6<%%9VO-E-B`@JuU`$Aocrtc<JC8_FKF}+4tC+BP8llopQ#b8-@UytDJh$il9F-&
zWPo)l#R}u%)<CmAFtTUQo-6%zFa8<9Z$;df@=KOhGI*uzoO10N7?Aho{rvg!FW8hW
zy|ZHa-P5C4ZicNXFQ-PX@V|D2`lP>jkrm5+o$<KiHaa@Ge@0MA4n)g$ol{l4gSm-<
z63egoeZ0J?&y&~OIk)?`)A$<dt<2YkF}2u&b;GH#`!h3+DSAb^CU>_}<=pJqn5tXA
z(a?)A+H)jZ-RRhu%9EOvK%>2;Dnv@6^!U#Yt0N>`pB3>OJh)m<PY;%<u*@xjf`Zyb
zZcOd%?F-qcgr7crI-QKlbpK^7=34iSjNZO^a|)iEcr))9Hs}Hulf^dp?!Iv0LdvC#
zf3U(Q5Fr;+wLM{tJfWendwhJHJb|a@2B-{oP!b{21*mH1>&H}lIZ#<y33STZ)!qGT
zY|I}E{jsEE&EVkRxr-NxKAdje!o6ng+LQ9~o~pGd_5qQRToh{sT6%i=)^FdsA$@rG
z`j(<y9z#Xw9~wHXsYws3&#`02ilNPJ-?i%@(l~SQ)!fHDw9s%r!tuI|p57CHWUcGm
z-#gjaLa5KEAJjNZvlC;1V*&!#8p%q+=j!E}|5gT&;zmy;-62-iUzq<EqbXmxc5RKF
zon6J}&nM;NsL`Uxc2|tIp#$|^3mmx2DCMSYvAluQ9ForeyeOQVn>>mFU}7n5KU|F6
zC~2^n`EwP;bz&MO`*a-KNSZ-$DP%!GvF`ms|H2jX<=Dm%C@wTG*Pm}uFlo2wEM!6^
zO^&R4ap!qz>J}Rt8^B*CoFFdCwdwowcR=lXk-J{9z5VES_P-D4sAd3?e+~?6prxf{
zV`oRP*wUI|K>hv45Bf>V&O)6yB?K6{r0nwo6sf4FsPB<z8*4r~Y|;!;BDWnd+?M$$
zyXe~I9haNUQ%y`wuc9kwWy4&2Z0WJt7c{@(A3vCO?b?Mi8iGMVL07I^xs9_eUFgAa
zCA)9e)t%$i%HE9Mladt^a{yhpF56OzG(n8SQME4kT0r#ghV=oaR5pm<)TsCU0|Or_
zD>uSbbJESNuvM(8s>*tP%JSJ5@U%&N%vyMuJYY5LnxS;Fad0FDj8Sc)rM(ATI|ik{
zuA$*>axw@0X1$!8+?sXkn$I}c+6rJYdhEu}6ItdhTM%q@b#=D`0=AP{1q+lwtKyX6
z2UG5_dDi{Mj-}V^GBCP!?I%DF95vj~3u@fm5|fip>FKdx#9xN#l{E%Xmp`{FsP2T^
zf6TI700~3S!NEap4u8L8_wL=wo31<v2oLuL9QgG4^9H!tPeBrY9uxBqG6W+WW-!r@
zxc;{7+xPbze8LTlwiHftvN{28z|O8!i=0;737md^isE=2)9SSl06${#T?d!wMJ@7b
zj3DXfk8*JQSz2@skZf6t(wE~h%Xe#fl<WLOd|eB<o?Y6D06%GB!iG80eYeHAjKTV^
zUq9)&FP3tAvZLJ8*|u%lZJcR5g+GrQ#%u^1Sln7`_hAWZ`@m?=1{xX}ef<MRnEpGC
z@6??WfrSiX%-MN7l>TDqmIJ;2dI2n}*Z7y;d-RBUCmr3aWnp0<O|L_azzepGnE!l0
zOkB0F@WFV2(k%-<DQjV2;hJ}8?3%H0cdgWN39O4(Oib3((9qyu!f6GCO$dB4SiiBf
z;Iwo1ZUu{2#3)X{Jt`<5D)AaaLqiiYGqUBE3U0kEDXPcu3q7<n3<`WskhPAByN`>#
z0Buaq$oK%*Lg}g6?YX!_&<Wb$G5wPDFhj2bAVJ0d1ou(8aV>LPLPE=#X%x7hKYvmg
z7#JYtq2I0otGfH(!CE1!?;e09%`<z<?jh+8@bjxZ*FpZ{XlEK-xLvSM9B>+*8^e>e
zoSd9kO)6B;B^M=hCtjOgMWC-UJ<PPSmoM){if%y9L~}VMFTVk2Z5{w#9G8$#)z#$=
z=<x7ZzX1A*8^41%6%Z7B0gPm6Y5DZ&)8Y>wR_W^MW)%thhM|EDcNFB+SO}KpJ{cds
zN$NG0whRwbSLo`T`0CYZRn_gFTMxs+=)vRv0ghWI<<5-ahtrim!Qg?dKA}clH8OgL
z#gea<m6ct$u(<eO>T_i!A9uPgOl@4M+85xqzWem)o`7sP(b=+oeEY_*Q^u1~Wd$EF
zdXW-0>6MYeyY{v<O42KhWmIx@X8-bOIa%3<b2F6Kk~ySDA}Vx)9|bys&p4)VeN~hF
z_4hI(p$#`!l$4aLuRxZ)iE_UB4a>jB4iZ2_44W4*EXJp#d`T{EYz(?J{@b&n+r!H%
z#&=?Jvd3kMqLR{VHwfIJx*}{|hJR;W-Q!Hx!^FGnyNCB~rV;}K1BZ}M$c2RTp&L;D
z&T43cUU(N3#Z1S{T=RMC?_XO&#MRNpL}y<oB*;Zdx$DLe+`mb?`^_7^N?_bL8gX%P
z4ne`-z~xR4fBysk+lqlbZZp5A;gr_WA4s!=kWk(~hIkMe5{2F8(sW^NuIL3#O?Gt+
z#NnQDLt|r`@gsPLIn427B*l}hE?>SJgd<rO)YT82cXxCYiH$<fJHJy#6@BwLj-M5q
zsVOV_wWC(|O+*9-%jD+doHH{si|{uGx=_D%?T|48P1xhd+se9w0s>xmd3(njy_)Ov
z9{dK}gMV@-rKH4NyLN4l|IuT|&LwB#AjJn0b7(i`GcUBjZyH3&5wd74)>MHz{{CQV
zp2KynTRhCr%k(8Wi!95^$|gs|{^d$_hj8pPx?;Jgsj+eJn^H2;Ngz7EpujxtIwtx8
zu277w1WDZZd0Uzii{)(3L6g}~;lm_;ak%sJL|<*fKq9j+7|%-p+(cA3FCT{2I(Z=+
zJUqLdIB6tCc{YGl#0v-rh_roeY>eB&C~_h=B!qA9)h(`~$k^$zPBkMVqm=i-K|zVI
z;N2L{MOurwE#y=<B`3ef@y8QbM2|+UFUx=a+&ByD@;f)+_q-!;;#n0HmEXJ-m6hko
zUBJOnnyniCB|=m!u?X|bZ3_M&At9Hhm!RU@9~4E9bw9msb1Vi3G#*Y58FE%EQwj~I
z5lJFid%C@~_1NH9W#yM_gGISiXqy-FUcP*(e(KaeN5Zjhuw4KAJ3acXIp4{~3}Tf!
za@v@|PDWAD=hCzfD8=&ki{&3a+_7eCnXP^)rg2_X_2eRKJ?hZKL|8(R<2~KopW0+k
zznIH6ZnA_2On+ef&z~!esB)!Z_CVb}2JW*xy92WM`S?uC4<%b^_s$GGyM3D~qp*r9
zF+Dwf&YpnuXB_T<iYrX#3M<ZrOl9}Mf$$<NjVD7AU<scx654$6t4|S)Cm;~9^&H4N
zR9;s0vAzBM2jd69LIVQ>m>8#DYFWS;lN&d0?x?E4`W&<RE&{IjrZPS*?h4M*M7)W9
z|Ngx`9<!pN;^*(vXU}egSJV$j&U^!1E-*4NS!4MFZUPafXJ_|DNF3ti+>Ee!fZI4S
z3vi0y*o8WBB^aaIwT{P`-G>_DHyv4gf@ZWYs{ryue``jpx<y6B`X%oLU>fb58z(vD
zUePi#`alHeLbKR;^wTXPqr*6CdlyYjJ|g%RLq<l%Ldr*rpwQ56Xn`NlDWwNp{`POj
zDEk5M%Fm&ev<TN+_hm^!ks*CpGk-njG(E~>poaq(rkYu1XYo9m^TENvB!T45oH=1=
z7+)u;sj2xiEKI(77aAMNhJm|*#3QOp{8^I&Ums;MA3b{X#fW09(Z47U%7RC}>wHLv
zQjXI^Z~D7;X#q8`tb1m62I+SIxDk*8xuJqXLs!=i{p~6S;WUzY30ZA<4m=SN5hBK)
zsw*q+dV6m=b?TJJH<X)4@$sxnXlxgAySntjQo9k32kdlnZ?2VE{7c(6TwhvRN?6QG
z#juZi(ykuh;MfR&T#irOxYPq0hAwuSu<*bRR5Q%(sga3i<mI=s>j8_yoplA~nga%l
zK$-`+MqSZql_fnl$*AVI&S%#Z;@sDnYQOZHX`Z5m^G7>d{8^;*Aq!l}v*&se7%1w0
zl#A;@Lc$TU!YF>tt^R;u`Z&6NfS(;fMgG($nC8{oJD3beJxG}?fO=!mp0xosmtVHg
zLObvO;0KLsMb5l#HBYu-1vSEfn?3kM3_o1!=2Q<{U9gRb!8hD`<><F&p@4t@AJ>><
z>o=1)fORR(l8cM$>9c1WM-EbMkS*}I%UOAtpzWiOW}c5|X=$}wGM?P`B=L%s!7p&9
zOW79mRK<uZG`|uQc~xy~pSZX~pgwy6o?K@0^N+W-w{PT<Z&B9-J=?Kk$ILvc&!6$}
zk9fdymo9l=6_E*DAnZ3B9E=w-s><f4$4F;K+{J?Vi{OJVB;6J~z$Sqtx*^R76yPjv
zDWrXop_azR`<a>iL=uAhKuBzSeEb<@Wr)ba$8ZGqkK=Kw<+W<dI6j`SjwZRgr>Au5
z)2$3bo0pcB&L}95R*ph&4~r5Dm#oVy)E%IIowyG0b~q58>@vyD&VK0czjX&KEqS!>
z-?gQsrLRE0UMP)0$-0E&g!g8nIzSOjoZ<9~Ga?@!acXxJviYpUpUFSJPPewUdeA*h
ztwG7WYG}xM^VB_0Se^0HeOz3=6HjRvgz5^bnw%D9c%jN;NFaIUx0<ObI|<!_h`Mv1
zzkDgd;hjD6uT+|wcdEuq!2ij4!P)MljLh2RG`-qZDeEq2ptiG3hrc;w@J5}c60z;y
zW{_Tlq{o~?z*tAIw*ApomYA$8^#jZQEr$H_A~lJ5B0DSV{*x!?eQmewIYtFKxv+@m
zzGi6HbjJAfRj5Xev*WC2dOv^vzK>o2n%;$+cf_s2uDcO=T$qog!J&Fpf94jhvCg7$
zJi^R8cp^b-m_4vy$6~?k$$o+OLF~KNmTBz0*yVsH93B}7K%@bJ0dO%us)OB98gd7D
z9b%@Tj8Mn6wzi1BN7>klJ35$Qcqe+*<A8uW2pZhxF9Z#2$aGLF%0i1=X5U=;d<NK)
z&<C*36G}=nMN4x>bqifQL6pdI0A1if<Yvn+I6WuAby+Xx#(pxg=|mWn;MTW+)~2Lz
z28D!lfn<m{PjSND`yteM{PzYRIT?ks>U#N(l)JnXFM;k3+{rEzZ;)i@lkrBYsi4bT
zfpCh}E1z-@8lJtqy^nWOZo1vjMMg=Nr<n^VAnEoam&@F|VvgYhSs2A^&*(>g{PYP?
zQ98TUZ>|i%)qng4xPw{#Os<%cUFsomaRwyz$s0G)r6pG_v#tpjG!H51X5fd;VmOSp
zrO#9qQsJp|{le3z52(4_0#O7vcJfX>OI*>H4pe=_JVTIed9<@g&ma<TB<rRWA0MB(
zwROh1wsuZ~85*?oXT$JQihkREaicX{UFuh_zBo6F(pXtj<9ArfdCEd|5d?lm@HC(!
zpP*omKOBxbUZvc*vr4ZS{Wp`w8k8|6k_WgjIVDB$cMjfdLg69r{$VlrCT?e_8(CR#
zH^F!GYyr6RlBGpQe%KsNI*0^^=D973#V00KB;DVoM$e<6p;5*gE#>|aVxj3c3d*%0
z9#CvWMMV?CCNgdE-_OVQ<l-Qx=3|P>MV{*|xi-cZe+b@5%I^X11N(=<RMs~N&dUzI
z_`(%ozpJk4r4oazjzy1g;FBjM+($lVPY$;QQuby9{|>a(%CmocP7BIX#MuWo+n^mK
zVf#GoEQh>-FF1sSLlMd`@E?A<rH4V!Q#E!y>sDx~u>}yT6`C$uo<Dc)M1kAlGo(%o
zpxY-$#ap*-MN~b#$$IF}{rV`>W&p3l7o6Y9SX=X=za6Q>5%g!ro%n`$f8;x<asH^U
zulJeTd-ih~{?s!{`2PhbHkG45)h`5G+}!cyi_gnxkgrocfOf^AZtB7RqD0=x!-o#l
zUGs;*c~%7Xeegy#5+RX6*y@;HuEQ8JP`_fJJCw~>m(%|9-k-o8W`vL!BGoUDPO1Q&
z&(S6K7~=5j(Hl&7qSJ=gt{wBUwzZA>yvf_}Vh@N$BR*;(5f==QEOMH7Xx3RMq33{c
zXd)_o#zIR=O9F@reR)n^-q+v0K0b*J4d-QNkt+?qF-<!IxO1#BG&ool2ANpjtKNYs
z@B$tl0#IYqqE(I%`LHMN7axl%UACjmsZ)3Ba(BAAx+W(kdc5p|%HneiZLG$m=*^oa
zdyieYEuBGsHT?a1u^|bN%7cPhRBWuQtdO}C5?_O!ovrqZjEub0@>VEGvPjJib&JJr
zq+Reo63}gKYHDf(#Y8x=?0yVeWqbQ&GyAErPM)Q;snkz<C03oow3${R^vRPa($fH=
zji}z5wMRl7mtkN#Y@3415;oGD(I~WP)hhi?HONVmBSiWWxUh6M><v+=iqB5IyW<XL
z<=&yjy1M6v*$`$e_}XH=WtrZCg3NF9{+zmc30R6rUp?5Lyu7?!kAA^+78ZXf(MwBM
z>iszFbMrgkW@$}`Icf{gdL3Vux{Qtv6BhINTb<mS4=yBt_249`F&n^XmX!VQrQz?n
zfkiAkcH9NCxB0UmH9S0=4t>0A2_CSY$c?OtcOZ7Mvyf@_adZ3Og8;^gKGoL;G(7^x
zE3T^Y#`Iy)S-a3B?B4WSIeU8{gNLh4jnjXmUtJAOv8Eji2eWUNq}Ca@Lp7saQ7ZA7
zQfmP0ABToMDlR^O^EQuFCB?*^h0w`rZ{EC_l>IMX6cLH!^7uqEuGswO=;*pBm=lkM
zTnT~@4V?ta>;sf(Xjcc>*w%wdoa(8nto&27yr{BpFX<pKW(AKPws#Zc4A@aOfs`A$
zq^g7q(cn3aT_ojC-MS@a-W&>XTcJ86%i7K^KKKrlA(_6KNZ5@T%XZu)lF^wn>;Fzp
z#`jXosHPYSqsVMV14?QOJPb`D%O(-BGeWMlt?d^?sZ-_s{r%*g^$kC-yA8eP*p!T%
zoOitUe*{Rv=LWsoJ)Bmb!1&T+UNP)C!$Gv?ofX%#gP{yI@Er*a3GoV+&*h!K&}Ye5
zK|$fp*%u#)riIMC4fC)~_SZLt3YbYio$*%fh3hNUFl!gx1hbUe@u|@t-mq%obA&Ps
z<j8jDvGPetu}DfvG8RxHX^DJ#XH5PAnws3JjkEA7lo=!@Cf<Ja>Zp4s5E%&6?bKB6
zST~e?s3diK(W4MHIx-DW(-lpeamymrayMjDRJJ1Tx0x)17WlrA(&)j1LozZP#>p&l
z?Dfv@um!}Y{eZzp0^^|Cs;Yb8;aa{fP|{8*DsBc<CNYaR=BwPbd-p-m<L45hqV%Jq
zqdwiJ-9&7Hs`(r;4u%FqEH$882dCTj0%r|uEH%ROevN|mdt3PwjtmQdrramk**8Y&
z%;NT(xM6Q}MQByszzr~lY>8cfK|@sIgvmy9X$&HsW}pi7rgY0p;-s+0-wyzACe8p6
zC_*xRE6VHizsjpDo7aw+lLssUVh*V<?{8G!Wtf+nds0cs_kt``@a&)2IGir~;ZE79
z$p<MH?tN-7Lf||OxVkJ{-roLf*K=PluP4b{ad0z+$>7|%ZG9OQn^KCdnwfb)CLW*G
z1IWNZ=ofDQUi;ifao`(%hLX?za6*vD>cS$DD*S6s``y=T7`09)?#Rl@N;j(BduJ*u
zO91URt!xhM{6<&FhO#BNrn*2$$Pg3aE+z~Q0du;ci0qMc5$P7mE<VNpZA%6k2b$hG
zB_$<ndT~w77L1b(87Jl@&F_raUA(w!NMR@NcscBwexP_n5lc=^CS2g|y?gXWQoxzW
zSW_;h+w6>Dq!``y?c4uQDBEWpPBB(iRjnk9X9q4yW|-$QHKO55*$SPm44+t0hw)?5
zpuK}bIYhb#Q%SF0DJ=XjGr??W9p)w_SOLu=U&Bz`u0<poI?Mi#dClnqM~`}EtLS$O
z6V({QN;>vwSXAqG9Kp0S7EvOkW4=w@*T2c=i!+?VDVzN6ERTLj=DDIE-fL-OB<roT
zQ7gZ)1Xrqq!V+*3R6nM4Doj67BUSq(R8{v*Xy+Jm9GI;7{P`+ms1-nq@s<~(M)<K(
z^*E-9b+_TtUGF=B24!>kIh$AVXR~AjQ&Uge`4w1mOgi#Kp}N;)8mnO55Mi&B+&wbl
z3+i<CcP+Z*&tJdp0F9plZ-BwO7<MGW<jZ)m#dId$irl3D817ewKYMoIz=3}-fPH`j
z1gy9U5g%mf6T1KMfzx)`*E_lEd7G$l`#ahTp^fYvvI1A5nisUUwJpIPUbVEO!6<~_
zDlniwfBzD6df)GJRaId7rk%cbkQ2>r_yo1Rlim$m7j~-{-!1vH(sI`Q*OWmPvs{3-
zljeaJ&t<%iLT^s$f31QFOh}RPo2jX(EStKWyBNJ<KohbSu@g;ROsiL~z63%R)J<{h
zZ_7O5uGcOV1&(97>Dhy|<M+|wJ~(Ck!uY~;{^MvYp1A(aHE|J<W9wEuP_?V?xJSM9
zex>b5JKath$~(lObu-Tv3;LePzklPbu7fcBGtEVXNLSex`y>Z51j0v5qQU<`ANFUd
zAZ1}6-`y?f0s#SGcO+X1gh3YOlr`<u)zw!a(Ag@2KAl%nGqt>P=Z;UHC779yCNowy
z!+F|Db{P^Y{ruidkcwMz#()s~qybJ0-eEPrG105^Vi5`<BK((n@N-^_2=3t2Z2G!l
z)?XM>`P{K$_NN9C#<1~BlaXscP|$sZHkqO9Q1RiHxGv?sq)prMKgd(f3`S3u(Ixq9
zVY~+G5?b#<Ha4n>_}sZXH~Fe@_gZ1;<g%?8@0q{m=tZ=Od=3i>qevM;wY+S5c*}*%
zuQz?2R^QFAIE-F|PjWi9<Yr=$SbUt0>ic^;?`2Iic+iIkS9+wYWd^g__j%CXtQ_GM
z^L^Zk(~6cQ`Atnt56V8kvS1z&v`<ttY*Ks~164jFzm6R1gE%?Cla&T-A47ue-&g>T
zR)&`>-`*~a_jsT9h6tnY!3I{(I9vnb_X!>(oGZy)%6t=Aln;xP?sEwdkv%{*f(<{^
z`_5v~>7xTMwM_YJN%VS@w)G?C8*Ny%lP<ez?ul6v#tA#)k-qbEp(CP-UBAd^<I27N
z5)KYRrhW_w=&k46s*>PADchWD$4vTd;o_gLCKEJ-%EBTlRZEB#%|oQvM~KozMX;0-
zwK}Bw)ZQDPvne!8OiW9C?d#nxG}TUuP(#kubcbWW^yJ#B)2frVZ{pvgL)86-8ebb4
z)*OC4xI^e0E_?o60uV)(Ko<~#Y1=649rlAfe>xQyEUy1ypr^0jwV-iCMC1~uUSm@e
zF(;I{Wj~2&DiXFR7c{Kgc~J^upXxn_48DH51PQYgWx4L%b-iXEUTt<x&Js*>Ftc)c
zv54#B+?wPtc%G9n-d}$ZzQPSe`!OFlm5L9pTMHfBV7L=n6zHzGMZ*4or%yd_VDSS|
zIHu@>VyzbUNsIrXb*)#|J1&R%=s}x?77;h%zSVjR2MH)reZ9TV6!#*$*?4$DpV2_4
zDO>xhDQR!uQg5I@Qt~o1s^@HHfA3-Te*_bgO}6T!BnKyF94phTY*UT{4(MXuQLI^n
z1+lD;KeS0WeuLWgJR$uKsy`F8Qkx2*w=f^S6f5)4f>#@(O?+_My$n4F=7MgVR$mxq
zcxbIBMQg5m&`v1J%U@NLHZU<c<GlUUo~CH;PPCsZF8iPF?;0KD{_B#e^==Ckza2xC
zl{T!3wQ^AU%XaU)y&R<qo5iVl<YlL9=F_5=9=Fzo1P5nK@pDZruL}(wZ*|%CAH^yC
z<bRoTwln)5Id@ZaZ%@_zfqDdmSI69abNlY=>RdffgM(Se%de_oR2n(eJO>~}KB(+D
z92S={jcB;_3bnfx4<KG63|}ohHR6!ai2V5R6d=Y1qVREYan&bZIEoKN-3%)Hlj4{)
zf=_o{u-3kJsOvj9T9h^X8nW3xd}MMk;*Y^9kj{eA#fu?DpDQZ<yt`h0P9oFVHAKg-
zYLG*@PKOajO_ljnH>CS&_=<K68E&%j-q_{!WV`SUjAUzWEj2ba?wL=PdIx-6!X^&x
ze`oAWlM?1QydD<=ZGO~|1`a2k^`D<7J?*X;8DXQJ@t)3x$N`V3NtpMuy`=MEs(gqU
zBI7EQ5yGKWjUb3%#vflIv{wvT?f=Ezn}=iFc5TBNHC#;^WNM;8gEG%iks->M6e*G^
zgb<mUBV&<FDIr6M$V`KwQWTksG8Q3Zp1ytQ>V2N){qy_d{oZZ+?sL1g>%OnzJb&l!
zcO2_jYhU}?cOb*7#BY=dEPFZ7v*mBEL3-UMVFbQZXQuwsa-ru92SkFOoHtZAXew||
zm068Jr;F(QojZ4K7gjuAiq)V7FlC<iT%=&oz>wb9i6=R4{`?mx`F1%CpGA1Qk8H~q
z%-$Q{Xz)nL+bmku>bTlh>$T3~!+ujEcoL!d4VujjntB>ZWp{<2vkxUE+)YPM_d_Hn
zaMiuH{m(|NFU*<oByZljgu6qOD;P9LtXTrqmS;1Y`saE|;5NR)tmmux`eXXJi+<$A
z#%JCFu6iTK5^r<GJK^!;6?ccs;*5=rpA6oNo6Q}k=kMYsx@xiSc@g^7HS2HwEx%6a
zp_=uwZEI~^kGvzlthsmg%x_-Hc%&NQ)SO>ju(Nw6@WsRzSkudrl7+~Q#)Uj0sN;__
zj*<2-(Sa~5Of#H&`A^)q)Wyihro-n*_X(q7uh7oM%+Y*Eo^o#7h|!ce8;Wh-@6x?J
zCSN0|A$ea1=Wjg<mXTwXYMx5++iZNfp6J%sImYfmYuMN!2aorG;hM}_ets`%iVbf?
zK~B-r8SMi7k#e-Gt6Z!uO*PfEp+SP8FKXaD_V-9*5IT||I<4EZX)a2aD`{zhojC2P
zH*DM}dnN?y6;rvhsQ;a+ha&{>8~NVdnyk*tXio>YF3wkO$#ouEg=FakzM}Sle?~_~
zn^0%!*Kg<HDZ<zJ_U+sH)bQeF;v{&sDzmZTLmx@zqp8|E(r<)}of#7_Cnwf!TfctK
z&o@HR|Foj>um58`0mRcQ?#gHcLW<joC<&`o<wB<l+_hYMcrAexqK&G%Zk2ynm=`c0
zLbQn{2%fVyuR+mQbW8P%`8MK6jA(8^9rhjRGt@FK(HRV+Wq7!k6Wwvq!sd*E+$c%=
zTjAk{j-8QHUGeCya7AVeEi)%)Gk!mx%a<>Icvp#`;9JJW=Ncd3wRDs0{<0JyZcNnd
z27<Ay@jquV2fu;&Ip0U$?vSEgS~CUm#`deBywT?OxH-D^VaL(gC(n-57aUHw{l8^Y
zhvNd}Ztc%}0Ti$Nk{%ijh7+olM`D6!1TWNCHA7eyQ+r-$t-S)DOlthsWtAWVqm$r5
zLtu9s`ZsJc+=)<HIDv1@Wc2>)(LCk4Ays9*C-VA9LMOLvJREY-=E8*wNWbQ#0+pNr
zV!CBQZNfI0?TVad|8fBW+zht%9tl)gQTg@23pA7W?%OwuI{GLUtcJREi)8bkyS~T)
z*5YBvi3NfZ`uxOb+6Nb$o#3#n#j4Z)VZH^BIGVyQQ6OSd6@c^s)Uzm9;Rg;h{aTQ3
zUI1_S*4eo&72$s|EwY<8Zys@EW5SchWagsB2?@#I4iyv@J^)X^vkw#({oLOG^0%ah
z9mtgX??>6E{qM3-_3`rZjyofkpLmqK;b2bR1uq9s=jgfq8XCHukdW{}_8D3h^XAQS
zd-#wO&>u=BFUMd0rNC98uCm3Q`P<CTO$}Ua?-#2pMF~wu3o5dWrw+F!aX()*WHAp9
z`T}40HKKz&L1*ORAW-zL=xIG>^ocxRaCh4b)<$Rb<ySOQCa%Rq3Ip!*bVl@Y^Y-?Z
zA5%&`>-O)XUi@$Jd>G?oO7)b2c-hWnWPDoZJ9ofORr51nU?nqay79-o!QBF{AEz{a
zsH`N06p=-^xVUH)lZFvQE9{TD-HF&jJrE!^{=+*im;eIVcWfAJZ(`7dN?uq6;15ls
zJ>X3se=yb-#%#>`X9>Zd<z99_=wMN6Zf>5Fa{0znKvtSR4Q{Q%^xnyz1`8AAFXMsF
zPgPxU6?wZtWKwi3TMgg(CqEL+P5EbOI3Mzszl7k7z4NkW3du-m76RZduxl~mCTpDk
z;K6!`GKmD7JEKH=TF4ik`pMkRzG>selh6bJANt}47bN^T*>=jtETu)B-rkErAQvv0
zME#KCG;-_d?&uT3_58vA1%I$IRW^S_WgzG+dF|S@@|3#ZyLaVt#ZEaFUMR%gFXOdj
zSdF<+9Ah{%E`t?{vdaMM@0mfsc_;rtI|;x><~@ly=%6gYxmG92A!^Nq8Rwmds?A_D
zye=z~1NCgjZZsZMlg$+_bVuR@N-gxmGEY=}M|rXS+S>vrJz_uEy}N)}4D<tIq_e}r
z*W#NNR3VjWs#tB%fJWr<QBb{{#R1jgkGt7=O+1IZBA?`b9c+8gR+X&u#$4t{!h-sj
zmP|_sHkirwFO%=cbPn$K6EnBI{{CFad2in;p2tLe9j0qvY|%?w;V)9VL?(bMZjsEy
zaQwl_wU@E9mL469b3C?Bit!e4MxZRZ5K5qoD8sLzH->IUh>*!L78dO!%UUhnchL_X
zTm(4<Z`ylc{0GP@_`ZC6zg-i!xXZXpoi&ag^}u@mHH1oP_9@7y$krPz+$(SnGAPe$
zhQR$inkISv*NJJ)ndGirOOd+hcN%OX?Xr#5634DtsBRoIT`pAr;K2jIVD{ME4wXlC
z2?`qiY(Dz(7x+%#!<>A~dcwlO8}P+yR#wFpgMHV#*wLf2e*2onzK<L|3Pgx`?--c-
z?x;rRPuvrpQ&YPB`NVOY8mAwx{)n$=^<~3<odJ#yQ281|w9HjXTs$N>f=LU=?w<N~
z*8aH}#ZH~u8nzwGNpbwCdL)O*SjWm$K}AjNQb52;;8={VuCCOlL2$qX#CE-F5nG^&
zwHk@TKvOf=_8<@zG!CI0?y67!_VXt(WQm)oskss<TB_Bpf*f@<wSuFMxjS|@^`6K6
zH1xMTv#T)IOuTK;yhHnEGT>}rI#xZjx_=F?zKoZC!x{G#tXskGU=CIW?^-=#0YhW6
z^HX%zS-%J`-=q}tiZfOZnOSko5YmFG<TC*(1oOYyU2k*ue?K3nX|d2ByZH$Mhx-6%
zFKeK7(9&ZMLKKOw_@lki?t1#@Y%Wh^HwW5IEV|+D{*C<zZesMD<#%1iH6>l##zeuv
zmj#z;bL}-MexEoFn1ZJH@lrny?O0}4l=WCRZyN3G?1;rF;+1eT#eW+fZKxqWOerLF
zsonI<{4-au`;whax1ck~m*+rW^3n8soA!ejuMRP!Glf;2EV<_d{IKe$63!@tMVpcl
zLarDd9=VpDo?bJG0-O=YQBmP*uNlw&DPUz~-B3IQwD&Gr7tN9<4Gd^2AB-Ag-FBv3
z1qjslp_Hx29iSl2&^6@~M;tI*)8iQpMcy$=H9Fn0ZWG!i1D2fsm6Zz2Mhj7~9)(d-
zf%aX5I};Wk^ctR=@73S$pkR<Uu&jS>!A-4e*RNL$W)|eKIeyk-1@VMRAQW@Opo<O@
zBUnF7&qOUP!UU~X5v|3yTUb?AY&>iEL{;-J+M{Sd&*Vcc3Yz?~&fTWQa3;_-z5ss7
zj#;keA23uv^I$W(%Do-XTW<ou=sD5a+M02`H$o=MECgde4?rQ@JBAQiaPX8Sy8B>w
zy!`S-m$kccArb9Si%E+IZToh54CG_|@OYGT02}54ACG%r<aLusfMAsw>m50=+MqlV
z{p+g0*+g!Bvs+yvhXf;p6c*DRiJbM98#<10kcVAsAssLv$Fm%1(^)I42Wg-xeR%s$
z^HIl4rNo6^0L8!d^pw>mbp$DJEy@GJ+ysoTjjI#1d%*IOdC1vSL;-5o`6wsOIAq1V
zOG}b*QPk{HC;`DqD{4HXr1Y3sDlTx|id77&_8&QNS0-pXH1V#Wfo^?E(s@+CT(8FY
zQ`7<Gy81e(sG59Qs1vlw`ff_2;%#VnoNRj+r1nwJx<$T7S`9N#U_Gr%O+t?s$f*AI
zLq?}gT|vsMSSY8knpgkIhsRS4z|3BVc4`1B25vB0U0r>6bTs8o^TGuSX8IH_&)T7#
zA`{WPv<NlE*N%?28BerUBc}jcevLssdW@atF^x11eH70bbs(|5Y;0T$)>ZS*+q(c$
zvXq?!4bUtyduu~O15ryss(b061ZN5Ka>{CUcCa4@eGf^j(Ywe>KzdiMUZvO+>a4F1
zNj}eZzXZ={5pdKEK*9Od-@nhnLG%$Xpc-ml8%X&SKQz2yaQTNJCwM0RGWMtQT!F3V
z&Kobh3=Xpg<coAS=z;^1dZri=gE+7codzOjr7i0ZIQ)){!O6)<#Ou+P1xS`!d*Wkb
zKQ?7+nkZShn(#w#v(Cl2Fc%?fI6Si`r@y~{8R$BGxoFIsy@fT!kFIT5<65-3q@P(Z
ztj3Jplns`(H@oNtKfOgWqVL+{y27|PYm*JGf3VrkgRU1TTe6H{uHjwAb34cd$;-<x
zt=kL=IvZOE>zLN^ySEOYBl!}1N3`Tu#zFvn9&R?FeISHv*)^mQkmXUW=z+BgBG87F
zx@eC@0<G8uVnrP@7{M5Zd|G<JiRvA6dF4=mCZm^B)v>@<P)jY)*vyRIWeO!vFGk(x
zb=7CuY&I}tw6W=D-Q38BmUz(iYm_=P3b!Q9riCxVck1<0%rN{K0_S$u$(OTB7s_3n
zx?~GvFZHV_0uy5VJF&#kC5N90x8_@+yPMpgq@}L@u&bKu;Y*wsnAdr}75DEF#@c#^
zK<xVV{reV;!>X!&t4?Ivbp^k+_D1ugCe!B8>f;q<d-lEE@}jwSI2`=8jzUKKc-<sg
z#4_I8+qNw~AGC)ZgIL?t+FDBjfw;Expr8E;-OHGtI3-8I`T6<h9QQU>+GVKked9*V
z@-f`q%MlU!UIS44V4}{*&qfHAu(PxO8tyfqh6#E#r~$5`Mnm1Yx)@I<u_&){Hh`P$
zIvFa)dK96QgRgL_M?WtsTY~l<jcOnv-0d{%hI97C`}d1b0Lh`R5Av7a(X!c`ixKf^
zki!7qBgHO#U`xoM<(!-(P#&pbP6draOvlbe<Kh)$F=lV7jeDhW`Ux1)5XGX=spc9y
zz-A<>o-;bH$pgejSA9AVLc4(0JQnLxX6EZ?))DyzV;)%HFL693V{iu6DQj^vJ;iDj
z1<~c!2Z8{?#)>*ib8|1^O{%?J<cHh{&@M@CQErSH(+a{N5-#JbapA!F_8_w<uc+t+
zD~32r=ypFglmM9LhBlXV3MH%$x=^*KISBok*1}quy;t`EFm#n=)|j^f&K7rsn!w)Z
zGEU}QyEYe`x?-_tK!~Xsn*;CK10~+UmcAvDrGN6*w^`5~TjFPp5+-uTI@_~nPmCwj
z)HF?&M|Pw!vhM#Jasazi$g<(xa8ve{(r~Fm@oU}|;H)=U_#F)eG&LXX=xzj(j0up9
zeF=F#P?H**iRNi)K-ZEcNV!((p6UGoK0fri=b_Yw$%QPMER^ICKTUA4m_gA$rSs>Q
zE(90hVrGQpl`B`CV+Mmb_+Zateh$5Y1w?5=c7+9Z1=Y{dbPF$}x0MQ)g0h}}{HVHh
zVL?cT&U}WVIj<O=xK6pCK173dp=D#9EFK;zvR~j|%k1CZ1ogKkX=$=(C?JM<U<ga>
zcg{&jE{PS*`G7O%?74F?m~aC(hQ^lw7^LJ0Jh#)0d$nSTQ9SDby3eq2q7-g#uKt-b
zN;6J?-2b_wUe+UG3uvW2z{w&;ssNgSzi_Cf2KkFRc>B>_n%IlT6^h>^Vm`n%6c36>
z4*N$?<EsHko7v90Y11_;r88%?E8785*f24=U$ZgK97Em_(~Xiqe}5HJp^<M8wn8e4
zz}hnDgKpfyhi6cG3IZ3O5socVU`o1lPTJTxap)TSesR-1JzY2r6NyM`v_e}hU%Iq~
zn9*j?ZShReOnQZ3P0_Jb{C!>SMo+9sMLD_aJ0F3I3_4zrsuAYZ1OfM6Gy^$)i^e2o
zUH0aWfsJ-L6~MoYErw_|xnfLmKapT&j8J2<;ZdanSQBa325+mYwTxYHm?_(1n~DAi
zxFT%Xuxt0uObPWoictF{p>$%%k!{_oSk-P<xdR!n>LE2XCh*b<wF6@ka1Ll_YX_+u
zL%!tMjHvtCF$d#Fo(6yq=d|E-G-+fpIdi6+ch@dotIx>1wSAP?Yf3M3g@A94BxL=k
z!vj6gb9pwk8P9)paZqruVi!A_rAl$dk%@^135WZ5Pat29VGF2~Rn2V3$(n1x*N6>$
zg%e(JqTz(&vxTodebR_oY;SKL%N{O(qTlbk#A-IS>h#Wkah%zEZ{NoWv0-ZJu)2D@
zrSJ$^gaK_Kr~}i`Y$;k@B8+`}wP?X4aEgO=jEh(c*1l@)jlr3~{}$U?4GqebPi%2J
z4zmG<b<ze3e9%R8g=|;tQ@Q7NfI*{t-BHB*;>8PpYy!RGd-hy(1lR0eS_)!U43244
z<oz2A4=5_epaEJjuI~{DV#}SK6%XZ>lulxj@WXF@-Rp5SGbPseWZASYj&DGXwqRm6
zzixWmq-#Bh2QdW&GW*#odn7qIIpaWbR-FXKB`|`6Lmh0!6O!JR=q%%d!Xg09vlzw|
zk2vGB{&D9D<eBD4=Hnb*abU?Y2~YFgXt8L2cM5#sWxhBX^plj-RR5!&6?JvP9H%BF
zA4=k=C`v}*<=I@Le+H9Fdy4_kYMT~s?2xF^%()X$WAKseOE@~A#x-)jR^Rsz?c{;Y
zGr`)LQUl%h0Z#jBbc~g0D2%`E=(96H8fW0$bg{n|%jRQSY^F}nDA&4m(m(q#d3Cta
z0xi|O>aCWyxc^KMs;H~)la^*c1u|let9bJCDJD||z1K*3ATqAU*^LPxWCCtr3*&q`
zfx!d>#b3pc{ANco{pg4OQ`j@HPr6{16eAYYg$7<Q2j?Tm$j?z~K|){!&CB3o@Mz|?
z8+3lo<VSmAc#H<GP=aFGZ_{CgS7=nwWCs%9y(mt;f^1nfJ0vvpA4^M1Fn(WDR|iV%
zKX!}_ps!J7Hue&ZgpcFkw5`IxAY_hiU|r-m4lDq-#Vt0L8`;kij6Oi$!2M~DCjWM$
z)2H=ILq*R+!rzQ#yA_%+zz7(jwKWrGGqqX}gb6A{;GwZN9Ab#LCorr}G|2pO%s$1h
zu$`jXZi)Z`I~qOg@4te0B;YAv)J+ya0l!G6AE?a~rvbTN!*m}SubZApV*N5q8FH`y
zoUmxfaYwfgP2#Vhjn?4Ihx)?)5)C>BK?UIJK-2yT@P`DhcwC;RrzbH0k$Cn#{RCq5
zN&;!=L^Ui~bvGo0q}(w4b62B>QQ=2TGABlzcv5@7?XqnG0g3JYg9oYxk~D0Rl$0dR
zt*L;qn#-WhXuMUbS^-#(hEsrB4|FMBmIWAyfW+C_Wy*kWiq7=?q$It@L;=&#urTeX
zURoQux#xpqxp$0aJCxJ^BDR>LD;2S7nvHVuedEq;+g^c8`610r3lht-fI{xXK34?-
z0o-CXBclQOxDct}$FRakGYa)zn65vG*c|$k2kIA7K@nTfJY0iNT14bt{2BrV9Z%4l
zTZ?X*7lw*&uI!)tomUTqKA<2@3|*euSkwW&%>yuRFpB>Iv|j4YV(1TYNNdrgEJQTt
zMf2fk)K*l`kifgDYGmLsc9RM=oNHGAQlaze4qn0rP>v3-{D}oC;?T$S?)`g{AYD1n
z)faI)54>l#x+`k4X^LM~l>jN+6RM+q<Xp#xmmwqrfSe=MxxM;A&Q|#D?LkO@e+eVH
zV6(71fT9Y5<>D6*a20t6+6B}Q80SUT0^8NdD-7??j09_ZVy9l-b4=L*wkoDgfuj~}
zCl}3Rh&+UfSj_}W?h=kjLW7m;%%f`*K_hX!7b%4KvD=DQmC=I6umuL$@`|#+#<TkN
zat>srmIzjDTm<IU6(F@$>1>M@05e0z5mV$COql_iTJ8fq>o2m=P6YUG8u<`KtB7^W
zJR^tb_N2VL-5Ak5#;He-`xl2wlG#h*34Ac-Gj!#|_)JG*etv=FeY@`;E_!%*tt>f>
zbJq|9YqdpKeK9!3%j;_RPOGetr2&)M-H3=EiZshU{6tAtdkAw_rIzKB5RwQTcxLe>
zSW!tS((vhu0B&(Gn;{nyB>zRJ_FksS?SEpCEE`nnh6JT>dqkiTh=oSu*+DHt{RRL0
z6PYVmCaC@l!O7t8KPN-5^HHJ{{Wwp6#*fa<15O-w`7VAkj6UPRbT5ICflAL(F;m38
zh%KflKtfAdj$LcF1i~Fo%EkrNQP5KS1iqiP`6t^!MvsUXfNLGlp7R~f$otZWdeLAY
zDcSICH@2Hu^cIf|K`FB4GzpX<jyaEaY^eHlbai8qHk3Y_LI(LD3o$X$`b_zyQ-7H7
zRJn8n7Tv)GOpfXPY-!=@wcHM(St(0OJRcP?LJ)&i1Q+pRos8kzwqRbGZA`VQG_VB!
zNh6R)S7c1gdGuUdPL57XuW(C7z6cls`y<ApG3N7l<#SVUbs-QA0mm2iIUvKI*~`Hb
zlyc$s_dTf1{cSd#X845TP?Q5*6_zxNk-F#SpGp3i_0C^cO>M7<sp*0N)ljAATc_Vf
z$4v3nU{XM>MYNfZ@%G)juYdH+sECe@RnSE<CNd1aqO3ZlyBX<e(a%gBArzCQh?bB0
z#k@AsC1?P*c+}j+RA&BXj$o!HQ^6OUO$uF&`Qd4f((0x7{S^ZPom!?Cd0_e>Bq$h#
ziKGItyttJc<17HV#=kL%7nF~t{<g7-@1G%>rArXEu3Wp;genBn#9t*6k?;~@17Z*-
zK;gNg41OO1_kE;MkF61sF-x`6Wnw@p`S&b0P#w{KC|sk(0cIq-q@(|x+M}y!NE2uQ
zo|4E9-8?5zn$+!k59-(jP5}Xp$7Zv}1Cz(_X*d*?ndz8UAdL+bw_l8o<3PuCM0ot<
zLYy;xDd>h}AddhCpC>sNX~D^FFRyNBMN)bl<=CEM$3kusA*vaDi6c)=Ts*?Uh`@gh
z=_|PD_xkVFox<b^NKY$_^Dz>GMc!F(9l8Pck?8<0L7^)nC-<thmIYZ`Xt4?AenC&V
zapT6)ZTv2}a6Qt)UmGAvNPR35fu1zv7A_&ly(i^bTDE5Ae9WCN`2+}2*O3Tam|5H(
z!18c?*gmj?GBPsgBno{hF1&(O@$fl3M;hTqc&ve5FouEJld<SPG{T8%k|{Tx>`w!6
z1DC)$mXMHG3e>lRX9p4bMMOov;k}p)mkfd`BVYlr2O0s_YFK3CyjTm%EOeHG0(y38
z(%Fw?D?}8C!iIU$Yw8jYcnc?@-XpzQwr0)oo|Vs0Ng8Wz2*R{^5msxmNyR32a{wzC
z240HX&@jLZ9LUmpk|mke(YOeNgDarK-V_{)eEirGv9n@pNY!RM{JS8$f7l&;*E3aB
zUHx{QjwcYIH4^9Dkm>C`bjYV)1YK7VF)>7bRtz8oc<d~PYVHgq7;=cMXoGC1szGh}
zvaoOt>6n0%DN@#{@o0Ua*^K+H@H4wvDO8!sBq=G078>K4HEZ;><mh+<bNVv%@(K%O
zo6a;X>dNz@vDfvWr2;OXq~H-Ro+2^9fj7<6)zvM1M!5sPn-4skoR@L^HdXs@pRe=;
zUQ>N~nJ^Ep1>j`>^E(Or`26s%^^`@Rb_C@OV<3wFJASCDx`x96rxx$&cdH?m{0|Ox
zrh0*0yQEPqBJUy2a5K;*S&Wl{utJnF^iH3@d82Sj+GjqJHB{cz%*}xKXvD*Z#9;y)
zkbHoJ=g1QVAS!ewq7Oso8cr7)u}4My3P9>E)Tzm?BCu>(1zy-N*dWwRbQ=9VANe=6
z{NS$YflYc6DFM^elY1go8?aZ;qC&zp{R%D-P3&W!KxQ<K-@aQ1HR?KWMZrD;zD@EZ
zSi9i=Tt(+PCN_2tM!q4p`W$)W2_vJ;hrdP%!BJ)w;g9tFK*TtYX$vGB;67hRqD~1C
zNwGi}L&YE*OOEs^s3$<WDHy|gmES<=5k|Tw2|(84k$?k|4S#zT!6H#7^AZ*VRoF;n
zr9m=7b~6{)DDl*2el9GG1M{wC=OoJaLr?)l|8+=(C*%Uo1fZDOmy~uDU3!27O51$2
zmxCtp3(DswahPSmaF*5~y1~d@XJJZ(rc$sn@G~DEh;>9;`$cJK;)`5)RSq1JluUty
zUx2BkUjRQSCBnCTU0O<m(-<Y4L6_9iz=71W0Afa{OJCRS$Drc}JFM?m-L(ufAP%L2
zUrmJzF@Z7@c^vf#@O=RM&`dmvD4M57BZNRe<bfZN!ePefJ_-P}e%ZQRVfz@4q?~g9
z^b-S%%=nm=@K-<yzXG=cZqXWEj8Yf%B6U(QDNiOvCoou!{qhp|3|eMt{c^z?hYxSy
z;W=pBh*pybqq{6Pc~?M;EH6Dt0F<@$a62zDHf#hHkgmb+KWS`y1Dt8vXoOuNa1=ZX
z0s9H9gNt_f!pS8@|D76ISYFPArU4es3*4q*-Wo7tK?bL>bulq9h>k9VM%z^^`5S4^
zABh3V7zS_M3TDHhVzvv=Ad|t&Bo;O%GnatGhPl4}<+30=-Y|a!PUzoOH!{mBS7tjQ
z566bl|3Q=c;4PkQFXf;FjWln7ZD8p!f&`BFFD87A<>KBocbg`aB;c~Kw%^;sV=ulg
z!aJpdK>UfqhRKXwLPE3%0q-vW-7dB!ksA?$CWwG}U!mMoKwt$yeA~9s%p(DKQ2#vv
zP7h=pVv}>h*lNa3$~gONCbG%vG|L8v6--v>X3!Y_E`Y`Evo#c&{&E3U#n;u>6N`sV
zEQ`iGFGQy4HTCY@Y*|%enA2xKV)X?&Cg@uvehk%|ZjNK@XdXHQse@>s>LVH2aC=bx
zGQPk!m>$VZ(`CRiHNZiG*24>2(lQJZVD^h^^JeHlvEi-}Ll#+pJGSfzbd6U_x#WDh
z_2AQ{s;Vk5kRU;@9bJN}XeNRp0NTw$V1}3rkV7W(5jqQmBI1r0K?7f;Y$1x{!$(mE
zylQOJ+getMDc+aOy)<nGs>C`l5D|psz;njbpl?{%`|$9wRm%Z~Gav^L-8?k2!y++q
zHy%SF$~qXvoS8j)_H(oc@D$t2gw2eNNh1Z>{0p35C(O*Y9WK9bNgOb!y3g#wX>k>!
zj{=LYQgl{kUxR$dS$q%ve<ME<3L1Pf$VUXh-q*2TysUssX~s_sqZa8!%~SbQ?n==#
z6&hWn$-JZ_%ZDWl$JWx;#OWzF;A#l+syFJ2%G+93Mo(b-V|X=MdJF+A8T1CnUopLe
zo1<H_|GqOOljkj15Xb1lx)$4MKbq8Y>TJ)QLn*ph^=tb2Z*6VYu&EcM3K#TfTUhXb
zG{Z9}f<;zzb>J^rGxMT26hV+poCv@<xetgRhl9-0Kwxy!rOE^-{O&GB%knHKnL!|8
zV8CD-6l1<S4#rp!^-Cw`yz2c40Ntc}EqPuhW{G^xoptk1CCWG=lzj_#om^(T1Vk`s
z`E<VE{0NrnLM*CPW=hBrM~8a!4y;)=@5<Cu`}S{|cZ?>V_iUGb$uw`BvCLJEh?N@u
z+_{uJPgZ)*QtgBUu^m;(hd(8URUNez`)&Fn$|8+JLq<00aPMBLJv)xBw0e<sF2?qO
z(!vc}Qsn2)F75FdNmJk09(rYnYqNWJnXRPjuif=wR~*B3kM!j@CB8Rj?@^h{;3oDR
zx7{nr-WhuGraA8e+Ds6m1=o0TKt*2EL9(ZvoOSXlo}!K_NOBp7m(ZvXsQfMIc>_{`
zY{5?JvH@ROK%*_}o;?8TQk*Li8YENvK|&XnhpI<1Y_2rJ|LFZb!Iqo-M?2p3C|b9}
zW-_7of&<)71nQ)P7`%K`AVsWP4!fN-fLp{R^I@0Wr?P0|7kXDczZYQGxG4kXjpW9H
zIeU}XxHu2;tQD7>dUMEy31{N#eX`2x%EF%zUN@2GyoAQ>WA+UWMlk+yeAGTQXgReC
z+s`3p9xm*@1El&c>%)y!TvCs0tRPkH5Od>NTh1NbZ}uBAZ+F{!Y8`hloTW8WT747r
zJEd=*gGylmY;2xZ^dmtj263WaW!a*H=yg9g2I?^3FvPc0u7E=ho9kG8<L-f<WadXU
zHHTITO41)S!LMcP(^GQKfZK-eBM)n%vB550P*cM04SoOkY6lef85jmm0EHzn%dmS3
zXS?hcNix(L)>W{8QG_^&{)gQ=1W)ojh{Co%*uW6H^#Np}gCD~hVFLt6e#aWcYHWq*
zc#4Fk%)Zf|dzzW^hYN?{|8w_|X>fFz3q^JV<fQ2P;PZp;8QpW(gfs~ki>KoSVt!j}
z#Agzn%z=l8xFot4j=*VJ@JWzdZZjt*%L6=_ryjK<uG<_NY<KHFYIn6Uprr^d6)d1s
zhdWl`%7=d(at(gu!C)MB1-i?|DQGlOK=f0L+FeQ35l(^oCF`G@+?NB{@~L4P&|f}V
zm+bh6877ItmA{<^pA33zOT2+A*m<Z@KskJ3&EmDYB{m(p7y8K5y3<T=tnT2HvvQOg
z88JHrLK&Cg!<mU+*QFl!Dgj&MYm8DY3$0eua?!daPVO`Zr@l<UQh-b?M|c*wZHg9Y
zBtN%Lm_O5o2;7cFLjeaefv}&i+;GTsR9)B*f6Kt&E}9gTIG#A0fpx7gDT@Dnaeaj{
zxBLkdA0s>$H<2qbzE__vT?T`+&bg1WRv%@VTeJTO&cG^Z&&Dor%Q;Ib+kaouGxttE
z=zsrv{=EP3?(0au_wUQsjrPRf*MENT|FWA?IKso=_Ra-cyU54o_d};Zh|GH(2a=I@
z0M9H138i%#o~CFY5FT0ylkO@)(wrc`MUb8>L^9~*RsiUSm9%32{K4zYmrl&nfAi;G
zAN=S4zg-|J+Orwl%CfOzxyV!k*C)*Z7XVulY(FH~iCR1q@o|KEo||qJ&@rb#eB%3;
zpx~_qbLOo*3j^2hua}n;A>!IO?gq9fp)pH3(H0qF<3<L7V{;H{0(uY`?#Hr-oIOG2
z9pR2b<QfEb%s!1j3i@33->xzjX=>@FRoL`0F)+L~wm6x$?Z16J<F^ze{@06XDM|AG
zeaTC2nRZC|^WW=cP_X;^a-031U;NL_`MVkz82+<z{=eQG9{sv_90IGzh72MgxrYtF
zxy!70O71QY$UYW7!mN8vHPoh5)HZI(^n<gb$^+Ccg*4VN$bMH1=R9j~BqNLQ4z2t+
zTXUT3Ix8a8Vm0z^B2X56g&UhU(t@NtgD9Z1(R_;EGdPpp>1%%iCd!d|titST;!rM(
zX`hKSd-8WUuVr8`KGn9WEgL^)yq)Z;QM?W09fFb$l1_RpiiTzf2KJBjj3L)4;C>XU
z)-#5)+<=RY;;B=}5eM>$VZqnw{O78NjTTu=O2Q5h#c2GHGXfwl6QY)zE5Pe37&(kf
z9{{hiIkn<oeEPt2Y`ec{*4%$?py2z|xn|L_ar|tNB1N`I1!TJHuuvR$HVL;N4ggC3
z^(#OwNDJBVSUz$~QeAgEhA+b^PLFxm6HWK(`Z|mG)9=4ducO}++0aLlY_;r!?!SSS
zuVR$i2oJtB>un&Yg)m@RgAC~Xk1U!y0VRn{$gHz?rf<Qg!^rfSV6gUjB+Ex~@Cb==
zw|CBDkY3JPO}<dSSOBo`o_g&Ynj{WV!NosexG!$EMM*dEU^2@V)x)6`%tRL#v&WfT
zvAW%C{h#kHgnroyd>K()8}M2n0ux#x2>;?(Zuuy+SSM!ukk@<6BtHaR4VW`J^l>$w
zBlPoZQ!#YExpVqWo+m=EKpL=>h*75`4w_!r7>gR=@1&OtOkbhOWt4`+uzdG_Islks
zqwjPUCH>I1L|B|W>OTTg4n^~WK{<d@sz|?*eLr*(DGy{|P(aUZH~LdHDOJE<A3`mE
zvl-rbI_+mLAun8{*8b<i0OPFYJ0-t(>-3UyYay4SVh($qgH^0h;Xhafi;?Jq^jOPr
z!pl+>(kn*3w&ZBUxN`b?NL#ctE%G6Cdy>Xyti&>_*i6ferKL$ljM;4ZdZc;0@Vk)Q
zT%Pc4oL0$nWpwK7?~ihFXbsyrU~*NK9X2(hkExV_zcz>pH<urhunKHvhG!wlb?YNI
zeg(`a(ewpJ{$#|yR<J4e15;eQ?a$SJmbb1L0P6)+=o-<-ICF!A=-IvA@a93Pmhkk?
zd^iHVx($Fq@(<XAxz1vE{)jRW=~dW3AF<8E$~pm&lSAu(%D;!b7I+%qro{E<pjZ<b
z_@v%Rm4k5>oz(YN!Pio+T@+YO;&sL8>t1JX*ILxM{GiBsAf!i^Oy7--bT`tTY@1=0
zWHD`Yb3eI0z%8xc5@s<$^yohA1=j_-iz!vj_}T_gSWGJR`15v|<c<#E_k?2bVh9K9
zNT!bup8ohd9oGHRyEqwX2wtzlvO--koJK;g9m%lI`1^A}N|)jWYZtG?TnCies~G-$
zcGiD4vFRDznV+#M95zf}tOoibavZiARaRsnfWgd!8Gn}uz&s*zX&2+36}oS}S3~b5
zUBOAeXR8vzCjFUd`c_M;tA{-80yZau#>OlSlpD&2_G}GMY|X|KHr|T0%}i#CXTD^m
z6-@#WD*)M(jFcAXuuOj$E+DVr=V8-OJ}5O<W5A11gt@dhf;}@WC^U&-=Fw;UzH5=h
zjg9#?@F}ZB*QaFvyNG8UrVl+}Wj+`9<#wv&M`?QmUjc@e>I^jC=0cu~`!nLP$cs8d
z8pltwpM5@KgvTII0`^nAuQ`2Mvj4%Qs-SOps9p<q1m0j)({DOLs{-Xxd}YlgWQ=cx
zzWd3oamRAywK_R{@lGy~xmiaBuswWsZzz){-^>0#B^86iCzPdtc=nlJE8X$9M}PWP
z{`@PLpO)<3*MAbi|Nmvn7n`V7l76+7o7>$0*5Kq{Cj!e)ifYL)+Lgg(1PxOJL)WXZ
zGJQZzEjCNaT?O?xYY8}NkbhQ7_PG=?DkUBs1w?R8$uxENC3D_VGNB{sMFdVvfbZT7
zM&x;TN;F3W9w+ZgF>;B3B8OaF$+X?Ad?XH!#TZ``!6rnE;v4*8--uznV^C}0Av5l9
zr#=Oo<m<tWUrDCz<eollExvYYxY#KW)>Z<uC}@W-C)YTt&aB{LCW~)T;k^x87)Q`|
zTqpa5pUKOXLDkqB`sY><kmT<MK7@MH^nTo%X#6pm6|aT~T<F=C^Hni;YvqKN1`!+R
zn_<GoSVjht*c)8n;#izkeVh%@jcuJPHt9!5Vt5+F#m5_9%O|7Lh$h{E$1tm=5QL!b
zJ#C$ep{Db_h0M)>@VxLdl_svV3Ee<R6oI{CbEn>vBg}%=m{8c01<Zn&xPxwstAs1O
zZktNQdpyY>0(9Z}r5m53`Qj0Qn}w+{c0k+SQ0JBb+Fju-1-Hj$1dH0Jzg{k7A~dr*
zB{BuRqzBm^(`At}8a-G<NH%o4yvV=<4?N?5E%L{61g{9#U@FTB9(TofSUBKM1Xj@#
zhxBI=o_MnqoF<#^8hBgT$a9iUMgb=HmM4&BNHN;p&E)S3(7AQo=omPl=;?va!wMwv
zBNAKIVrN*9lHp+nOw<j~@D@#&KIx5J=y}7L(FDvtK^*4x4q+JGmLgl<;fA3+X1%G=
z>SknRr6&WpK&>wm`F3SF=@XLis2sFq`gF^s$X5rYcm&e5j~J9An;x%bh+j;b`jM|E
zA%fr~o6XW%z#zSE%R75~_NVaM2gSG>a2k+*hn=YJgf?d0ynux5*lhMOW7-ZAUfzX}
z`V0@8fSA&9G<Q6abxW+;d7K@L>9PYH^f;xP$+)i=UWiPM@PS=ECuK?-9hl!gYxy^u
z`LThY(MvAl<fj6H*tv%bPF2%B0k>ugrQg&01h=ltUIEuM3z6FGwCP)a?p`<>%Am!b
zci_SNTI|`T>ikLD70m5JQDXRPA2OUt%xLlmowz)GGTwK<psOooetE-%JKM5X#5{$o
z-8Ku@AfEsAf)V0)+Z$Vv<?N$v^Zo-8xMuhpd%oPeb|@yB=Ec#V?bF>4K6xQFEi0Di
zp@w};f4<+@4|^C3;VYpdzxTF%xx!{L6=%ju8R413;KuG(4(#$Y+Re+!g3HV{`xjip
zSHKuk_aDHK$hMaOxt+E~g<<P*$_>avknqjvC!%qX2gkJ_nGlg7FWC|yDEJsd6%ZV(
zjpSy8=3!E$m{F+Jeo3vP7qD%;$g8ID5B}&-g(pttvZgCvgKFW8)h(VkTA8#TqT3Om
zh0*e&RSnCgf?A_+4suk2r)|uPt=EG&;1Ek;MRs}}EdT`dHA@utKN4Lfw0T*;L)0!l
z4qwvVhifOcP*e*WC8IsET0*P07<kP}G=AKlmWpjdM~)S2fkdBjh3|-FUbJ!1%wLIJ
z_a)X9qa@_pR}A$4b~rw|hgf8BH#<^>h|Mt!^oXg__AU3L0$6B*l}399Jy)2;dw_ml
zgsshj$*!A_5H!>^LK!%1TrF6wb30}YJ)!$QvXBUn&5L&Nv<WtVrK0f3-;Z#$2%y**
z9^}<c-w3ObiX}1^Bs=H<RUy$-+|cuIAJ)gFOomf7#%W}n+Ql2+9N2;1>XI7dK?nf<
zT@H(p{ruO9yfA5oxS>+QWF8raTfxw#kaEcEQw$6pK_N(CgZ)r-6{AqSXY~#!eiPi_
zy?gMMf^$>M`f!qvu8uwy=1mx7^5E$mZm%1Je~n20JN{<4?Rrrm3_nDD#97c6?~nm}
z4t#BBM-*ITKE4{a&Gxvf7JBP~d2sD4zI|vn3+&&IgPXaL06a?Yn#igZflEl9Ltz|d
z*xr^-&1{s8gAlOa>*CL{(ZZcB%fK^xClZar2z<zU54TQff#!l{*(P!>g@2<vwI%4%
z?!p<B-o0}im6t3yr`AWOO?5vA;JfW6({V!%Vf+Hd;E&+VgXw4K;lpRym`<ncb2fJ9
zVHiO8$LhUcbpf~!Ysoi@P7K_DRTP?Gh_&q=tfMpf-flBMfJ1F`Q>;f-3yYA~WqddT
zLB;@^f@(iX^Qh5bg<V#N)O*#{t6oxkw-lqA!t5K|LoD+|QKz$EHF{&cN#qUX_7ozI
zTn5B-gok&-<dFSNoRl}|h+2cj)$rs{{#wR0=TDFo>SVj?<mS$C@;?JJjP1S}9;#;C
zDmYX6ddyR<5WqhPe(8W=MLUhM=`>z+R2@TcYB&Te;~+*Vv{rKJB-mZCICZ0u)|=*P
z;e_N*bU+0dxnV;knirq&_|<SN+MSrBvWUaMXfi~7l#0?XI7C5L%c&ZGTpzc=jz{+#
zO-o&P=!~yQa)yG8c6Dv4iF%PBN>|Nz$HlvczkWh5@oAL+h8=RIdX%QGPdVQC*pm}`
zs>;rLRT>8B-Wlo%k2VR$KZCC3XoPD^y^@5{dn}5ex$!3OG*x;Jr?j+_udazXSLr0{
z%bG%2H5MJbPbT>|m$<uvT}MjvCV9INqz@$=&lNM{!0D;XfxO;iNE~aHzmyJOGu;a)
zI@Q2TKa^dC5yt!jQkTQWTQ;+pp%ea{XUYi;|8T1=Sd&H{?j$d@0lc4)&b9!>C@iqt
zxf5gY(-#?CnTWXwo=&68sh)<Ssj@XQ7-9ncI_r8G(8>(dwbQ=*NgWh0EgdUaxQ{O6
zg}4zRDi!1Oz`Nm}br;rdn5}7loJa@SJs4vTciZQ)p=fz0Hi=`6Oq$7>&Ftq?Ws*yP
zodu$VKKjRLiRR$hnv8*CM&_dN%SLu({pA7-9Id5&s9j6}+i1+NAI-2r;6S$q4&TO(
zmqRr;OLH145SR<1D=%Q4=^Uv!UT?w_L(#c=Oj&UP(tGgQc6`F*u5MUcH}WbJbe`v7
z?@l2}G(HwIRu|mzBEVQ{5MJMiO@7$7#EbIC#UfwbSjT$MAfzo&H-bNNOfVxS^rDc=
zsm}b#_$o6&!L=^KCp}pOw>H3j*RvU0dkykji~DhaT$W59e^^PJDc+fbb!HE-WE;%5
zAx{XfVHZ#C@A85qDTyDEx5-c5U2^C%zm+a9onG?V4$S)<CHB4($1Him2@Xo)*(YPC
zr`niZ*zyHIi&B&F!A86K>5TAbf+ijQ+QrZpk}w*pEW)w3l#azBUc`5+j@U;+DzeiD
ztbqzBmPl3Qig@9ju6H@i#09QHKCtWF2~;86IBeEpk#(>DxYc_mg9u>ycsRW~r}}o<
z2B+gBEJVaI%{kJFFz7C7(-s*fZu%+aC}a6%<zWA3J!Gv5U;DJfodTL`oqwx3p$$_i
z##1k~8wZEga#X#>n;={7o9v2=kphaK{0`yIXZL{_w%w?cx)^j+r>1LGS3NoT%5J2}
z55Y^SYtj24i(jZBI3rL(Xjg~tI=L4!4v)IZ26n8c9&b%Pw55kD!YBPJB+U$v9cWvA
zUjtNVmgaPiP+)FtX+Gx$a><(;!$mV*y=$j>*NPs48j(O-X;{eMvyM6la=u>eRN8Vt
zZ&VF1QFKwBlI2aP>LgUYFTKugEsvygv$<=uWbkeBvo{+azq0(Gyi+YOgOTRj><rTU
zR!Ce7e?sr8nm_OKF7hkqXdEsQw$AH5d7Wbw+40%rS(qa`DAdXytz;cT$PN&a*HKYX
ziNE{{zB7-yj#1AW8Ee|M)H~4WzLh+cnBDDx@|?aE*xAYO?DjMOGoeu$unb;ad!F_8
zXlz}HxT2F=o!ox!LfiOEuNybWES{{p3E<PqepV{c|Bf{rIc`ZZvYga@)Bb0#mD(F2
zEa5>3%!P~Z3oz@;KbVeCuvp^P^SN<ACv^TiEjo}qC#WpSt%e@PY6=2XR&-oF^nn^C
zRGD#Pd;~FMxYx&(9n<u0L@bd|(zR?#<PT6RR2KfSf=(tV8x|4XNqHCMNJN`+_FSG@
zIqZZUG-c|k{d^oEaxyfVk8!oYXHwPaS7=p$+Y$K86~TcQ1|^~0+RT{9eOuFrnFIj6
zX7^3$Hn=&J*iB9#Wfg?c`a{R0W=5&uJ@~xS?ESFdE`s105$+(56kmf(R-Y~pR+FCn
zL?Vwr_Q;GX1oFCuXTm<x?L?Ge0+X#wq#gB#!kAF8(w^UB^WL244A>4MEbGmukacl@
zbMo%UF`HXm<Ki3iVY+SZrue^hy)=p?X#a6F-@PW{JNZ^C@76%kE%ft>fLS*O8QpUS
z3#l0Km0XuS*N%T>BqSHbqM~>@ek*Sv!E$3_KYRgiSUPte>xC`YZp^Mu+iN1IZ^|1?
zOEL_#uQ_==s;gv0Xg4Ot*x@u`GRFUfD+Np?N>TZ$G?{60jIx&9+?glRxpQ7eh2CUe
z;9~<ydFRGV19`^((56q$UtgM!!z6rA=tut8drOz)Ak%q*i6%vn83?r(IKKtBQg-sM
zM2fMwc^uzmQzF(u5aC+qQEw5?mv}(P+8CR#)ovczJ;K$M1F_juVSz@6_Ary3f*5b{
z1m+p7JA@(ivek8^15sRFVZagH+LQs(-e(~<_bT2+Ruu(Wd3^z%6J)im<2+DlG8Rs!
zN^P0T9b=y}_|)a7&JvS*s7}d2vSQMXLy5O*$Cic6Jf27oc6NO}_o5Rcs-BEh_{mbo
z%jeBCLE8mmW(`6t%TPAFv29Ws6GxFKKlMhMqdZGvy0)QU@tmz`e2l&V9#C}G)c1!z
z(`hA8H;fx2$!#2O6Xx$G`1<jT-%50%4p+NQRI`p@ijxY6GLd}TG{N2AAvlAYDBP8!
z#O!m*@pcQr97Jn4qHfa+6#2fx^d_52a|X6bNL@N}y|F6{e<mr{48cp5_WrYc%ucZE
zE374pRdc<q0Up8)=Q*3l&oMTk*NZee$gTm&lbO>%Gugd^aw9Lkv%+kBC)KUJ^^KE4
z1eDvo0|~Vm`D5uD7X=UqTZjXJn7r(b-Q0%NrTOi+qd1^TMA`s|-k@M*G6o2s2^*j`
zYIMT}vLE#fbjpE<2}`@tf-6Q@AG0vVLWR!H-#G}Q2zR*$%!IT`^O2GZPZyT=!9N0z
zh0hx^KnA(he%w0j%X%pf(KUc4D2v`VkZM=s-r0wl-Mm3%NfT&Br6QZ^cdQJuh?y?k
zk;V!Rc0H@N5NzlR`*Ft?b&!!HSCW^)?M;RImPq>dGB2kU5{gVVZNYGLy3&i@{F2|n
z%G2M?K^wQfXfoP*lA|}@(o-qr(559)<7bN|x!N1av3VR69Qk^^3hg043lC9_3wk#q
zCnRfEmQEAzP@wAqmVb`$@Tu{3&k01J*y;)Rv<CD(AGCBGJBG80G3}_?NM9?<7}R@o
z0?Q}Y*YBuVHg(KsQ;<b68R<#@_%a!j8e2e{y$Cm=!n>?+$9ljaC*R$Z6bDV#&+Z+{
zU0_4YwDSkAsfBzX#v0rGX+Zow?`ylsFCfavSRQ%ob*8<%7Csx|7FuQb=}6=)a|aF=
z@haeK<IAswXqUU{k>MZn6r&F);80^*>pE^#G!e|%-iR7J1Sq{Uzu%8*9ye}WUb0(+
z@OASkoD-*hCY8KjTw8S7@v8nNI<l*Hu@38_(x!MS9{hf@Iea`e`~}Y9@`kS|*xVV2
z>voym*jTg|R{`NJJMX(2fCm9v2<r}kLvPd0B(>{62m&5U$rt0MftS7=^&|F}qJxO?
zljCNu?Ra$S`G1XyO+m=OxlUMgQn$Rn2Mk2XuKHCNn%4&d6BMMZbh$}g9-fUR9w}k2
zaxz8f?lT!aWKg?9U^#S<oQfu)hI1FWKyFuf*MMJ;#mB_Xnll%o)HO8&Y^fHsGwI5|
z0IBL}k5eyas%@Z1<T^g`RM=?hX?Ay(S)iV34Jx0zAXgDfh^1H;u?DJ%kp+ysnbbl+
zOe+yNjS^##22OfFSF2BnPkFy~+GfMk((z9YcepEQVOyc$@u&-0l`pUp0c~a!S-Flk
zucbJzpm9K51GN3kVdvC}c-e%`+(3~jNV!-`@boj`b)6i4Y>?L<))Of?TiF!H2Ftzz
zI#ZG39ZjeN+tG6Z6<fjqlSQW34pzYG5Rxt|&BkC2CdhuJHrox{@x>!6!Wm9&Ta#im
zXEVU!BGU%YpzKV-N38)2%HibX^L9YPiY*4673z>u8r%c`tW2a0S^9g&XQ49BY<|>r
zfAKHFnhMuN?J0)oZH{F@`})GGkEI}@w-70@h!<g2iTxfU!#Y>$;{>})VKY&3;b}lw
zGqnmj6RuGlHkd*WHOv+|!yYxL0V@W3FrfrkL$$q*wH!tx6>EamARGONHBMckR7^V+
z@h0VsniBcXw!t1`v74yOwqoqzmfbsiO?^}hlnaPq4iNI$s~bnF()qk&0mdMOO}hA}
z$-{b?=GigC!RJlsHfXMdg5`elz1iP$<Hkd8v8^9>j|@_tE#8x(ZlH%H-|Vk`(4FXQ
zNW6DeikwC?B-8h5BX=(~;Iz%bW{Gdax{<+HxFP9?**P5;UL#XX@gAXK2dup|o`goQ
z%QM=7%_6`L7aXKgjio&t&2{Y|k@irV2o(S3Q)tc_psB1~{pJxNqzKGrKVb%O13}F&
zXJboX{<vY-&^RI8l-ChP)pZx-@gnTPCM1M$P(x0@nb3=d&nC3ogRDNIbXX4Et20z}
zP*1+8U24ifLrcqu-3E`l7zcOGffqkfgAx*4ah%<+Yy1K_=pS*ho(6Llv95rIxMDrE
zw21>XsNcT@i5oqfL8#1#HjfMuQUsB_Sf6s%1W>t)@!!4!cYaGvjg-;a7gkaQoVaj0
z>Yu_?hqmPr!f~K%LM)?|m|aY&%p!z<SWgX(oe3gn7n3)Mp!Wq_Ab+9_@YUiu%sRf$
zrqH$Wq<p2BwRX^WJ72B~`C%(xV0(*pbSHaoH_{Xdd<(+NQw<tirzS+&?i!R|E9{Sg
zTn5qn3JLE(G=DE*hs|V#B`%~gjowKcq<~|Ufg|jxw>X^Hn*bV>o<BZXMyMS+pxy)p
zp_pEawe#5Ov-%;)l{m$fvk#(HndWC=R>5G3a#-O;C!3<9$IpMe3d3RNgpSfj$1HON
z&81xkDz4HtiV6$)@iQlrvK>qh7m^{k9J`fFRWEU4pH99)Rl=GcJcYmvZY_aU<hzWs
zO%70QQ8H!7p;kpKIEU;DIDI6TT!&w+)d#4D%}o78tmI{cY*GSFh$TGd-eyhVzgR8v
zMozN>CNnA4fqV5wOb2EBhz@CjPmD8F(bV{$c&UR6XBi@{F5!k|99}P=uR$L`rs7?n
z(t*T9h1mn-s`~N48L^liu^dzwpecmtqmMM%!L`c};~J1eS}z8}R4g1X+EkRR4+gS_
z5A;k;_Bc#UM;W*A5uCHDAy?x=yQ?_^9nS*SsUg>5j3+fexW2}N?B7ukFU54NH*Us$
z%ym)c!5h6Gk&%|$W)j5{k11Q;k^Y$paHfLzX^<*<k;@bT+i4<8dujurTTQ}_J_mN@
z3JhY&cwV4;pxtTw4=s}O-ydAphv10#01B|E39&}x9gl^*g*8zkLF#LaTCs$CP_K&O
zP*r}sD<lwdG7wZcGm+MJ=>BiGlzS?FsR(Bp@XR`=PxTw+);OH1%SCyrYdLlthX_l4
zH-gpaP**&}QxgpnP0abPGf2qBY{G_)&cp3Vu&%vmJE}c}=n>z)CQT4Q?eUmpU21H@
z;FM9?NbMhl#>6XW$B`%Z_w3$Tb-jcqdwWl#)E{Kbyxk7z$0GuY9mm**FRvY6*VS;L
zJJZM%)zUyp4|)W*wr<(-dN^%LYOq^e%)=2X3b*&JE0F%aPTA`-!&VN?+4I)z2|o8>
z+p48ARxZsvKWFx`CHF5a*!orT#opyA(s2(Idz9Dh-y1l4{S^&GvGB&VlPQg7QuxI~
z%oWb(mj>C~3L5dR4GVKVJs4ERE^00F)Cto)mDSxqzAL+4;q&!}ouL#N|Gs`0-K7=Z
zC8eB4Q{S7j^rA)e<5{-LgyW;s5G@&*D+}h$J3D4{6O*LVuQz=yJw28Jsbg={u|rgT
zq7yB>-<zEqfQ_v~J+LF|x6O=DP1po!qO<37OHXxn!lsoQr=yLUrAH!o{n|JgzIS6H
zVEVOX_Q87++yHm{P+A41J3}H2n2HwraZ5F|TNF$p{H@P_DMfy>AD_>g;{pv;AiGOS
zJ5`nqpa&z50PrP63cL1suRJgehAq)Zu(siO`N8xkFnt2>?}78@GhTI|aes>hxw=1R
zHa`*5{e~Etf};9Wr#NR&?iP4M-+?gBYyL6`PBh@UyU|*tm)l-_%tw?X_Y}nqDyg6B
zEnaJgbZ&9{PC-FUl*8?J%#-0+@yrQ`%{8EXF1h5^eekF1$EV<Em;nDxT8Q2d?_M=<
zF{F8)L0^r|c|)Yxi{tr)gbsuGajqKTnzw*q9YyoNbjbJ%TCag<EF#&|`WaWO>$&MD
z#6(Jv4TXLViD*U(jxHz&XqO)=$G`MXH?wgNQOh%|41#t$XGKOuiAI(p_ZNkwZRG&o
zYZC%N+65JVU~2OhFTR5_y-w37+X=_$HP=}uRBlb#jPw~gfP5N>je^1u;>!5?&oBPD
zscriD`jI%;%Hi{+jdSSY{*c3pZvd;e03P#$VZ{Lr4gX3$jHUG<2rI%r2<_6BZ7n~c
zM=3uwIT3|{)?2>5Cv3tO-bB4B56j@n0rVs`Lcf^jyXrSvsK;q%or^`(+Jn4%J^C=#
z_C&9x*9v`*Joq4ckGj^I+_FG$IDFp`huvlz4R>%vyurD>C_o$1#Kf-PgWz$dZRZ{3
zuEK(X*9iX`ar^`zI<Jf@x%v|4x*>Mq;cBdDKb+lJ_vNluBXMel7shRTqqjg#w__09
zZu;}T&Tg2#5ez!bs6_gJS-izrABmfZ8KDjE((*@mQ~`Xk2`AL5ojZ4CV(tdBOC|W2
zR<s|jp&~hggKMD^?-<^o9HV-%n(=oa&!B)TCJJD)8jxmgy}TS45*)y7fw1Rm(1Z-3
z!jxiAyjaw_<;G>b8AI?6yNy$*1n*&69U|qLchj-O5t}XmsuOVR@`t>Nf|8Oy^hZdB
zil%FzFtighT6Yiy6;7Xy&d$a%q}P26yt(_KUU-B2iW7n;OsuS1m4gNRFtL6FfZ=*L
zmBGw{6H@LN>0FTs2_!g>y#_RJ3b(Bfdu~O>7t~x*eQ1b=B)USkxLGYqaV33+;gJyq
zMMV#M2l;d7Qb~LtWPpuG3w8=s?`?k{vKgJMa@}^I7lD`%Vjg&%=C2mr*$55dTadcM
zbunQE31`eL-2b;gl3OwU<LQCy{_K)pDs%76$wL_U@j3J9w$W+u5Z`9YCKy^JY%*SP
zn3e)Y%3#Pa{WbBg48Q#R{EuuuT?WtD8U6jkzyEt*gp(UwvH$t{DNVcdf4*uiM*cU_
z%%8t|GD`v7)BpK8!FUVl_Wyi6U(AJ)=zqT4IHtGM|NKL7Ut&1@&zIf0|6Cx$f9@bH
zivO%Zx(okjHi@O2CO|l@d-qz-W9!~OD~=Ns$}`41Nk@_+G~zVgT$Lm+(jywcx*|@!
z?Cftsut;KBz53s*#x5JifS>c6kJ&g#OS@^rAAbMmH*_|0b1Nhs&+YKdV!z|>4DEgH
zW6vHxjH3V%8Ud-EJ{W44@<6XAL+PGGtu4p&$HX9wnU5UIUikASI2~d5`^y&!70)ez
zuKAPdlZ^v}3tIuv*q-#kUocb$;duCmg|A?}EpMtI{mFlS{bwoB-S|JVl>Toz5H6=@
zi!)@r1PEq`d~CypJ|xj4fTg~C=FBKZ#kCnAsX&NUoCdcxKvaJo-8$vmnQjq1_z^yI
z5uEy=xyH3K_xCL}$Jx&{QCV(>#Nvhx<q#5Kf$nO^h1pt^#G;9F7`&2`Liqs8`_OaD
zh}B|Hkw7-M7oCCImXlcHJV=|)?ek$+VO!8O#M5NQa8@!gaMrrXymY_7nG6L5qVjKx
zB)=hfivEf%^X7f-#ckxX%m*i-%nL(_wc|k9FF$8K$h>y#PHe(EIQieA>BI$XCm6>>
z0qU|w$49}7Rgi1T7I_+9Uc@T!Bt8L9-5bCvEdVI}0Z}L-+xiSsL@0+uqI+wNO8NkL
z4N)MEsXg4kh|&B(Kd`<xpbxg9F~o!_2Fa>5oPD@8<NPVpiBb%p4`3BAUblYz`lc@3
zUmx1Qud^ms4Np&h0I&x@-iPAGjT`bHY(ybTDvwd+r1Tq9%Xg|F-_I#>i#q;}fv}7Q
z{F_>kN8bt!-HvJMruO#Ds1R<WKOhePhzq(hXpI?UbVCoS_t!69z#w<uOF8sKf+?uH
zRU_u>*My6&rNVFG&tF3jlHSRBV$_zNh+@E2g@CP^k?5n8gL%oVl^B|x?V*1L38po&
z?q$nNKE?QjT$rUb+E=oZ%94>bA6CJ)i*jG&w&CHYKm}41T?)viYojcHL+G(_bK}(T
zzg&Rv@kpg0zCN@Fpf2T`YVoD+B(EjHB3PSyC<TR22;H*^dkYY*BEhI+TDtVB?=zhj
z0qt8vF%8NC+|?IPQ1Q%}#}LLnKsO4Ptd?$}>oEG`Q51@V#U>>yLmj-CR;}6ubJ92V
zUC2^ZNAxWic3sMtXpsH9m|+x>I@_S)<_lVbJPKVHa4DiD(}-1*rPLhEk^ofvitbDs
z|EYd6U_I27CAwf%Q_lrv2W|!rCTsc7q$3**F;ct_p2)H-;VrSS?8?ZLWB6=NpER=(
z?yYYzC2+1p7l}1MxybBNF#M!qwUX{t?p6vBQbq$%->Mc%mR8T<PU(++c#Z&K{edV`
znbg<idepQ!aR-3m0KiV}kW_oXM;A<g5A>`RYf?a?sG>q0xcMnC;E|M@#(2hqHTeb#
zBBx;Uor38Ota9%M){vbm&v$hV+I8~Mw@7b%%4?KAtq=+cz}qW;z52i!3*Hx>^XC4Z
z%LUMRRAJGAS6KeLa49`l`0j)TpRh(zaj_hI7>e1xg5`%+Z{>IBJB2a!gxVZb5e|Tu
zwtM$ozmyGiYFG>@;5ir$*nDek-3Z7q5S7?l^y|7K*38_}jY{Ygkl&FuVdD?dm>7yZ
zHV9sjAHI@~jYANh;VFo?p`tKITMNkkHrT~COXw<4Fo#!)shp!?1-D=`z=K<;PG7YK
z3z=^P5yTI$?S2VKNplN(K(bW-VOM?9O_FhEDu=Z42-=in^DdO$ydY6R?#+%38}@?L
z!wFol(%c3u8zy%4?Zh8`ASq>6DXNHt)e5xro&Do4kmsd<L_KET?uU0vE-c!H4F)XA
zS$J@sTlTW_SC!kn76ctb8H;lBi>V~$vi!iZJ)%>je)eV(N|#L1?3R^%0~nQS>sCct
z=nxxPRlOF@>%_yEW3dzJ?s>ya3#l4!<5iStL2JjC&+zT*&ZK-E5J&}_4i*epS2t&g
zy1jJES&_liE4;viorzg(E<^^mE&CXT&E~1ms<Cpb#;$=!d#^6&?mn2@2Us)?Dn2V&
zS>tf^o!#9vj}g_Pogbo*1ZJTl(^(@3_l0=FL!bx_WK=`Ba^D10(%nvTfMakoH^hB!
z3pRBeV&vf=j@(%Ar#etHrPmz><KU-h@vB$+azU})%FkahC&|+sy`-OY&#?OA00bT}
zC(5B+8s~;M3`lh##PFQgJi~vcW><E47iOVUQ2k%40~D>D*SAs8Uj=e6=en=0QbGJW
zmp%b1l8rs{`%^|n@n}-Z!9#w<_gp6s6hA?n@JlYsY1DCG`N5UU!`<D=K2WC{9U5Nq
zDeNi8sf0j7BWN%gOFA=K{Hz~1Wk)4SfjvDptwphXO!7Do(h?P^I1ICC%OS4Q7X>3@
z<J*dgip0B{@N^4{i`gvtL0^mm6oEC<*%^<S4YtjBtaF;*+W3FQxcp<N44PY*0YinN
zbcsUJ{uh}Y?6TZZKxhz2qWUUT5tLr_W^N-k&zDJvl-S!peU28h<QH=g3h!lS-|b<g
z$*|8~qV%t3aL6vR64yyUD>iG)^ZNDcVvt>|bc6;~yf$Ng*GegzCUH<&5Yp=p<ug>g
zSTAAI<{1ag=J&l$Ae2-hYW$O~{wXoA{d-OF!|u*s2+!HsH&16+9G6(;`f$sJopTcR
zk9%EOcfBABVo5(NQ0(S^jB`nGj#>Bl_QCMoJ8+`=1KEEItry(F=e}rSXopU<g;<J3
zP9ng)2DZws#_Vw&Keu`fukrx|PUxh)1#)3q4WH?>c#bH4Cy>DVASiKD?QLQfdlyh-
zy8!ss1=4PC*1wjU^x(CW&w4X(y1qEu3CR6!M}NQ<491cVS>Rfg3ez?KUq;_vs@oPQ
z4#lkzoE2Q0oUa6?FywffHhJE7-+>V{m`_21lFxP0i4uy>K!#7-l3+eUVTZ4CgDOy;
z(;hnjOK>Ce=<dE%!Xa|@dFsUea65(!bsV&X+0_UkThTs#ysr)tPQf^NMIw1tKt^Q#
zH96a=6u1%m_!N8l!R83ai5#F5274zhGYr@X3#NR~^UYrO<s7Tly@xH}wR%z#{!881
zdJK2$6r>L4Z0%G^6q8%FEEa`8q2CnUVh{$B<JU5rk^Af&aV>-6s-K!77?+*h7WaK&
z-Q^2ikBXnckN81|Q$Qc96+|VWz+CHA58yv6lghrGt)<1bAi@l@$u4u?-g4$f{>k~G
zT#2%3LT>s$2#*44-W~#t@9hZaXOD=_<~w^0^}x*;9{eDa>EQl}2d1`~q9+&E#<`e-
zn~$&DuqQCyd9c>V8ltN4olr-XdmFej+;S4&VGHIzOTX7i?&01@ff(Roa2vS1Z$R2@
zfymn(I*Y)9*aV|O<bPSO0lpZ5Gtb1qu>)ihKLoG4Z;@aq!3L-mc}{oD3Wmbx(A;jt
zt{Q2}|6P9I+8W(y&8bre#EWAWzV3qfS6}|rL|KO;UE|NXw35VUVA?OL#o(r{?TY>T
z%Tz)5-U<3`YJ36~UJ3Z<U2zYe@Hu*uRQt_-QFv%PUT-0HG7b8rwa?5$va>H<&@eTP
zqr?Ydz&$got&m<&s<241Eetob(AI|G>U`YI$2ao7ZEx{iy^rpu3BuB<_9d(WuOD>b
zlI3vy?H|NcCC)oKDq_{NfMR$DOvtxTp*#hObss>$*PtA=;ucrz$cL`Ag1r2$&-Z+M
zeDLVrptaZnxb_xs#W(1jZbm0~)uBTZNn=3C>Lh1Ac={<xPY;}F!60GFdnCF;;`X2|
z1p;<cKu_pg$#GcTfLeMgW5<#uOVHGI2X$!=2GzI_hi}o1z@1CT(*q86=k;s(SLT^&
zX{MF53j2WamVgv}3JtUrfXO3Zz>q^tmSKQeit)z{@#g;9wbdpkdR*H>!h?dik#X%f
z+t_1^?CTafH6fEjgM)XV7p?#jK7a4fQ1$lqA3wOU&H^z_Z~&>7Qvbt<2y$sW*v0SG
z#JyvOa^7giwwopF9C~HJf(Nh*+vvt20e%A~)g6FU1O8273~Bp<YVqUa!|RoI?UA!&
zmB7$9Mtp%4W@k@<$IP^54G*5RKS;*XHeYaN)k=z`@r#VHlh1b_rY5v(9I!MEfzi{C
zB8v}$<ePD#I27aDZ`^p2r)u%l!^R)kLJ690w=7+NOeQOhGBSwFwGv;DjbmX|)+3>&
zhFwTyMD4pn3k@3@8u~iolimVKHw2xHvWWLK4<X6LB53MgIhb+0TSrHjQRS|0;H8(I
z!)!-iueX|ED2;{z)3PtzC_dyh4{AQxJ4fA>lI-s872Bu;`jH0uZx)P&p5kr<k2ZpW
zYH!bY&<+TC>$h)yJtCknMNt(L<Bi*+E!o|w<Rj5{1OE0^kInE5gAH9rTAAg{g%(`I
z-^T(D)pJIdwF5UHtCI{1J1`7X@g*I+sR!D-(I+~mG>LLe6AtBVS%@R*E-o%v=8(32
z^t`^y7@=IlGN+g5BOi2k;(4LvY?D<lh<ot9506BBLj+{_K4C#s;;a;`&P5}H&KZki
z`0yRSgv5`6x2*v_!6)R%=X?$+atlJ7`Ae56a=`~ez?jXGolrnPjqMJnP{~PL<fH8O
zNDpz|#Q=kUg!<{-sNny@)|<y;o$v4CpEGl2tEoo2R4P=6LP)}yX%Uj_I};&$mYeJ|
z(^Rr%OOZ_3vS;6!$`V2)?yUD@joT8kZ{O?bd_KRwet(?DIiF9(eZSwY*K@h9>v=s9
zwxiinNO1zHPyW=E0fXWCOR((}zD)GA!uXW7K?{bLa+g7lB5T^*Tl6qPX^oQQq0K~3
z)%M^>bZeppHMp7^YM2x&><l5CV;+Hj+tu-bCqKbsC*KO96+L*$6?-k_j};@0Yns)C
zzwMNSP~5rqK-mXuY<fH2iYv+t+(dgsW@n`T5NuM#-lWI6!oZ|i9`|En(jR%v=r)J~
z@OOP>{m=IZ=(V<Jt+v<Cj&?9PqWg}4=hX^k!z_Jl%=~v$Egr=kTA4;NuC5|1@iYv`
z6qYDet#gp}+uY+)sDxzy<G3WQVsqLcK6_OEoNeYDLFPjSLO^R6W1YHXP&p4{J%t8l
z=&}dO$^0t1ok04@nVHr!d!+jLI%)v%h!t~Q;|Qna>xche)*qZjT6A&UQDo=6Y2V5@
z)nAO(b>5ug`sU}2cNjq9o4f3Me0<_Gm-~dC9hQ|ff;RX7_9|1v#;zE(R4t(&)~^r4
zxX7Kz$jB~RgTiY+V+)P|G!tPcSQG(I{rlT*6~TO_cQdXXy`)%+^L&LG!-X%FL??h;
zpnmZb6kt?MLt&{U@0eZ9InDs1fM~9St>XLOYPK1jIBptI89%IB7mBJ}0kt!BfPZLl
z<p@DwsYesj2QM>lK>IXPG(s>Lqyh3QN!l@z<Iu&2cM(Vx&@{rqjrtd$Yzo7z>4IPj
zLQ!#abop3dZZE5<qzsA|AymZIu6tm=)om&OKe|d=6~Qr-f;xQ~48KI+dQ<j=kW>M!
zM*^zGcbNIYCv#L>1i6<DVvbgrEP<B^jfoMW4}|tc0axINtoI9FFp+Io6GZd{{1^_P
zd<cE?NEpBNE}BV2yzB(fKi_1qwte;W*Zq))sAXSKNM&oKYAI7iY5Dfa&4H8d!Oy2=
zXCu*c*J98lEuV><RS%?`5}}*D;1q+>XgCOaZsCd^{@ZW=lgf@j%%(($V;71O@35zg
z3FmEfP^E_<Vt=fWuBTa$j*7b;aV!DI3O=^21N|7i(VSGzsRqegMyEMKV0>B+FM#9n
zBVv;$&T)i5)~1Mtekz@ti%Sst-w>Px&Xn&~VN9I}Nb~**1V3@`F8DCuzDv^Xh<(L!
zKAPEv<Kqx4gyJZijE}t;DAxO=XCJsu)aU)kh>mcHR=9YPs4ilB-luA-K<L7c(XOmT
zgG1<r@JkP3#32%UHR<g9`<rhRFb}cL^HCUDVgxU-X&K<4h)D4!ReoqFw!n__2CDp;
z3~&FXEj4fj)p8l{7Q()|P?)@<IjMjjfEq;PW^*5`6tKf|q^m?5k?J7)((j>}L>V1I
z7{~v@3u+}Uwj~A6IkD*As)IL<cFenObUgjg!vvr1R_75mDBMS&5uRNg*gxL=`=5Vl
zd#aY;2e)x0R08p!6E9lI!G51^3jDSsKYhB&ubUnJK2290V>t>~sOTSlSoO(&ri}oJ
zpg!UumJRWa*gG8g^T?s`5Rkuv`5=yt$kvkLddObvuji)V(bN$DhyH$hB$m%9d3u(%
zUDc6-){hq%GFvADznpr!foQCEcbkVGQ@lbQ15)~&GVZyL%ca9o8&zRd%8=i+iez~R
zl%qs6rEetp2!u-|EM6@atyoz|HGsy(Ds5~Xg*Vy&L9FaR^jaOGsF=-`{^3n;fR6Cd
z9v835(N>NE{d1icJB_wd4FcD6!+)6d+ll_eaA_UEUoah^i;LG7im@F<Xe#-Dd*QS4
zfj4x}^Fgu5i%K{Isr?mZR(SXA`w#SOJjCkbD{p^#F(I5NL@Zh3Lz50DfDcfeVx+JT
zVqU1vGTZpw<I~>}40YQ3w)4Gp+XT*Exe|}a<cC<^M#{Fz%cCSTfQi8oVRlzZ0RlY_
z6jsVnufT`)1Bl4i83O<wfvJa2`SsI|gZ~t}J^#sv?WAd>s2Wsp_LWjreBofiR_0vE
ziN1snFCr$e&n*ApDK|AbY6xKT^$_@bu!a)FS0ug~>OJ3FtX`ZQ+CH;yrS;s#JN}#s
zUBVeWX2D#N7jNFWMSZw9_TVLitDzqm2DG4oCi#8Y;wNO<TMJi`193Q8VAtsHdZGUU
z_;xK=4oJK|hhZ0GTl5+ompBu`7a*0n(0?RQ))^7Ec~U?#<^I9|ppqzi%VO{4;*YXd
ziPUz7m?VA76<Oa3&8e2WNA_{G{5Y0%LikDajPCFt)WADQpD9B{Wd#WH*6=7URQW8H
z6ZEDa`bAB9T)+Ml_IwHd;h|Bp#^Fyo2cF>CCL?10>a^U5iH_6$%P}IRw?3Y<DdJH*
zqk(=}6IFIq@ot8E4_Z+Ta10R;Av~6qfkfR3wH;vumzI$dw#zjiE&cq5WemRlQ>tZh
z8gNS5op-AZ$9m0UM5lRA@;(dlbcJ$V1EHZHIq-!5aKzbVg95u4^ER-|(d~-L%9_Yo
zQMP56XPdDL>Od=MWyeJDJYL_nuuC9811cxy;5%m{x<A<?U5FbrVUFh4tRReL)w;zT
z{q2RUnhOdf*y-o$#$sia+CK|W+b3X4d&98w_0N9|EI^X-YDdK`q}Yu)<H5kq`t%PS
zO04A^m9ub1`6X^z4=6%E?WL~6TNFwS?W{77f{)B!Aq7fw3_GY?SKtW@DY3uE^bMl0
zwcRR)0xb%?gg7}ltzbb|%=2pCf|SayVCtLGo8S3zBZ8Sb6eJ`AkdlZwx&%+N6x_sO
z5s$|A*w}e`@KRBelVe2<@`;%Ac&&7Vu%Vt8+87B|QOu)98M3QOPMQ!YcddB1yQcs(
zH##dX-=R^F(^cqzwP9|$BU8Y#e`K20`+6K$;CX)1q_Paa)HG`HWYou8W9rEAE$GZP
zw5)=&zv`WVRn+PU@8x0}Xh?u1A~(Br=gzBz4ZDHZsN#4^<SGML97e&Wjg=k}d8#_L
zta3wF4=AlHvmSPT-vlYgdmAilP|}1%G$&1l#Trw4;}i}#$JKjo7+-hp7->n3G;gE(
z${9W6{Edu!Edr|EcJ>_lv>2uvmx!9n+ui&fY8OU7=>!#axm1o0_;%Qy?p*QN;!whF
z6IF}#eTplrlUAe3GmT*#V*+}I(SnJ+kHyt^3e<5}Hc+$x2kq*ZBo)Gn%VH;;P`)~w
zvQcy-qs8Lt&H$c4Kjta0^3xsDD4bK_?z**(<5L=8hT;YFts&VYr*0_L!j41TF`)ZV
zNRxsVp9%_|IF_A~zs$`a%^on2bah@z5AU5)^K)EE-|EJnFA(I$UnR97e9?{nrqD_`
zlpYHR(VDm`8>W*05z*}&hOX$=@Szo#U=N|&XZUM^!o5@unGGtx$0FqXkL&kAU2*G=
zrBYjzI{oVtr`taOmu-=#yC-BVU|P>}38KM_MfW|jo*xo#TyMNGUiAuJy1g?I*v{C`
z4co>UXiYv|X?s`;q>eBzt1h=2vnWS6%kO`IJagy9e{!<3-$T`C)hvN5^~*B8z7`ws
zcuCt1Q>2^6vTeios!?8f{XIG9&og~puG!jujt&WVjFw&fuco+Ha)<cOeYHM4EFXec
z#5U*_6F*A82HY!k6lioZ+Fj;AQ=so*V`F3529RpI)dfx7o32LmswZ=0%En`#JPc2F
zRc(1&yAIt)^K#lYP4_$J^Z)%4(-#Fg@`(BGQGLf0{7XV$M4=6@Op+Ze6wKL54gbs7
zdAL5^Z_Vlda}j4=!nzlMY`(R2PxH<gbjR_!x5av%VBJN>l3@%v9f(Bag#Q>fdn2@U
z?J#>OVR2NN(dWG5RCyP4myW-Lc|OCwvreW3n&da%v_n+V(K(KqD-UQ;m;2PCL1>we
z4g<A1)>#dt!v5tB4X_d!5FuS|mY}+84{Gy*ZbXawRjoz#Pn>cI3LQzxb9z>4qdfN0
z+m-#VyZeGr?ZE!V+{wo=5hBW~Mstl>AWr_<f8jc%AY$EW7u3zZQqJ9e<W6v~HXgyz
zHxg)y^!LblcUp1Xj4?8~h7MJ{erWkwuJe#umjo1RPjqKcJGg${e$x2oCe<&EJ8o~_
zO6(W_{&2j0X1JyP2mK+>pp)`$EAxXwtr&Xo8Dw;y<2BpS5O1Sh=~y|Mzu&GYxUfs2
za&-Psj|($PoIf75(suj%&S%tKp>$pEysXd{%VmViW0n_)5sXVYB(+V!28*s`xmgX2
z3T)f5#Wb&h`q;9>Wbr(EmJuS*%QtLoof885(n#d4caN9#)uE;ut`<fu|E%3xwaDmE
z+_OVwPF<azJ;g;{69sCGwMzVT69ow?JMv1mYFD$WM2;pD{`bF*Rk2V(X>F4Ea_RS(
z2Lv{Q=FqdfG1nn8mbZIz!tUtsQK4gQ00>DG1<DbdBC#AT8q;uJheky3RjjU*cp_S$
z+~3B*LHeswKv9lMa}LqB=6s;8=R;L`<_zZH_K_@*um|{83&{uV4nbUag}O=E7IU}?
z*cq#OmHYPKRdkl(9>=jv=e9FW5jc`QNMSe2$;qKY2)TcMXB(!m0MaWtI%b1EZ0<4r
zIrl_w*+oUgP&@=b-}8tid`FI4*oz^LWZ(qhLi$>8-{(h}L4XY@x_o^PCtN`hqlU)P
z4FjM*uGPdCj^sg&g=UJ|4G3NW*c>%v0GyUcFj%B<<3k;OX2tDK1;qA#$N+mr(^o@$
z0HPBbCAvI5Ix2)Aqq~UuN<f(^vCy728)Md{3^H6J{7a2KFhcP}MB*Vf6ytdcqJB<X
zb&}Jo3t1zv;4Phb-11%Jep@Q-0e$zO12dn38Odo*(JQ{Ex%WTBGY26PhGe;;FbJ`S
zO7ANml_xQsyX|QZNy|V3n_ovN@iBhnKFklsGrw>N<sY%dx^OD{eIYC=N&r~EU0BBd
zGI9R%A0Qnvk4IubROiRwAC)d%Y&Fo*sIZ{w8vX3{NNSEBc`sPr&ez^;{BmyY`aEDT
z@koqYvaT6|eBC~XjjDB+)Fefn#O%0H_?Ec7Yn9%eaHS^<ZfF&t<_ANI%m2JCd|Eh%
zG(ccg;wokjh0vZxQl|t2A|Bilm4e9A5xAiBXft?`@L&@6V}S!aS?Ji2yp!rg!kJ-H
zb1_9D`2dh2LYL=}<CVuxqj(M2Bdt}wvS8iizGd^~3qH(reQk7a7{omug6QjyXAG{h
z9+k)y+Rwga3(tYoku4xfLUHk42EV3xKe*JWpE-lH$9uGV?g_dRf8+q{?5GC-Lu8+L
zdIh3HvO!UH%LjPEE5J&<4^42Q8?yN1Yq6&Dm???J$OiSUs65xuT|~dLXNVjoCU_oS
zBmo4_b~cEB{E^P0<}W^=a39+WeV;zJfLXJj(Bs|Cg~JV38y;i%Jd}hHApfM}19uZ4
zhIkB>pY!i*jF>;dhJiXft#x0%pTMpvf@6j3a`+(Z6TSA&e~1l7NvLNqh@O8<OAAd+
z!!ifQJjgHrGiP9Wo;L4g^npLYIap9nMA_ZF7zG)3RR1tDGqb}EkYGa)M-gGq1vZy8
zP;{Vsv$`iXxyy>*FY_@^lPbRa72rto1NLp%XT#NQE;!sjeDdM8`u=mBc6%I3AT(I;
z?U>@RJG%SGs3a)q$)Tj!I;b87-{gaxIDqO-;MzRKG|r>QB)Sp=W7AhH6L`$Ea8cmL
zzqr*J929g9C0CT!cYS&AQpe<6+|Zm&D$9FO`Hven=sE)bsQRi!Rf9h0)NC)ZOtM~H
zin%TzUQ~e@Lz0F{4B|)&yfPxCw=)|8c~<fpx4H43m01z?Lrt-NMK}c&hWJYPjwv+*
zP9OWOVoh>-`)4T=`}8;=3?c*vgFazjwfwv^y1Vc0iQhurk9u5=AKA>Oc)!lFYyZ86
z57Q{{<cu0+ncVy+;kftrOMoXD91*+b$L?-gzrNY_r7n+IQhfYVyvtPj4`z7ka_#V~
zm8XQ1TlYSAn&h*(lIVg&VK(is1*-wHjD42#Qnb>e#!R0WOc)(CEY<+%cmXS#T8CW)
z^4I;J7Jw%eJ#&humzOdFT~5d6JrPois>o@Y(2-Y716I}mqVsw|&bIw{mn@MK=4?`r
z4rAINiVZB{<%mcF%E|X$x%y*GNxmyxgEQ9=*oF4S*|<6Xp|;@4`|Wq%rD9@8*I<S$
zf>Y}sNpH3-JFQ_4;_s27g2G!HCF^(_ZtmEx@|d8A8}c#wIC|F>y@#mttm$PZ_{fO6
zf4SPgu?tjR3xpE8+Xq15IvXp+1*Z!f!9{GrXJK#;Nn>Bq?feIT3+Lyqai|)5O-G#W
zaLUJ9$iw^41P|C?n=R3!Wdu0MoMm5>gemNH1$~ne>O#<gH7t<&yM!G8&3bAMO2eqE
z+hX!dSUVKiUhY-03sWp@#VD9s1B3v0m}}k6={;Z(H0J;Wx!I-PyVC^%=;H;fNAi<h
zWggnl6g0dCcEd=MmFdOxw*r8Wg6vSt?)T<)>|sORZPn@nD~X)zZ#p`pSsu|$@HmMC
zu)zsDt;7hZi93|u4c!IQNH{H*@{IM5$m5v4_l&9s<?(uN#Nfw|m!~;m5Q9XmEihOT
zmGuD;um!GkmuCtm4#j$^<&2ePSv7_!KYR?W+V<cGQ1n^Jd&@Xx!FXA_jMU1I)_#6i
z;FGThzyno)E5esLEiGY0`J1`EJCFC)<K-)8caP*ppuK?+*bldD)G<=r0v^i7-M(uS
z`C%BXa|*zWs18kJiH0WE?q2V1eu(4?meEz5$*}=76O&bY@$$Ho;+VC%8x3i(^S!X4
zrEB)DV+KygUvLf{M*40^!>D2%-UrvYI^hK#GikwlJa&?3IYfJ0I({m!R;e6qOp;V%
zAr=b-{WLs29(%Mk-y*&orO!#5*omnrbt)?8Q%`Smu6S!Tj#6pf+gR`udFvgRI?Rv{
z3{Ridf8?8WA_MJRR6jKHwt1fat$TvQ8yHR*(b57dT6J!?+)^uz9jj07NA}bKeds3;
zLO;1DI#O+C&p0GPsfZG{M<of&weg?pV2t%s$s?^Mq0m;8{}}m5i<^gI+cq<o=jnjr
zCGz$ku);`uvDv$K?@*mJysrBsBV4bGzYNO?r9=i`UAw{!`YlgYU%qcdbhRsdm{4=1
zZ@_Q%Bp+W@;^pZn*^8IPD@De6{Bcjy4_-1Vc2<~QT#N<=u6_<_y}qXysFite4B;^?
zObz}7Bw^E;U0%Q-qS(a({Q7K>uzcC5YQ4|Vc#nHAQtnxgAEgPwx31wg8@xARaCQ1u
z!l$&CQCbU}I%Tlds66=?uMl>;qgk=8ak0M`yHfKY=AK5BPeq`y%wUg>+B(EadoLA^
z_4WW~O@wKOVII~!t^GA*d$#vSIpEM*_J3ZBgj0O9%z<RkUgmPymSvfU+|JjH;fR>h
zW)9ruWrtHZy4Cw-w^U9|{E3Zk$E9cbVz79MmYnVBj>Pz_>a16DiAYrpAf?wvFR@No
z6VqY<g893=CX$mLiGDG>s|{z)BdwC~Z#`sV_<ZY3{0Kf))U^D#Nx0DDK_yGR2U1f#
zjzM5fyKTCVFLv40v^j)`9Tgtb0tVS!?7rjah{IT?RF3UDvdxc?EhQmJ$xGv>5fIa5
z^=PYyxZo5Na-T9{A3n4xtWbO)XZIeW9GHUW8XUEEyzTpxVDc^~Nefnptc|b)%|1{J
zhEP<07IPOOxf|S$9zg_e>T%B~E`(*>o(A~^FI;qq7BitD9<aeZvmQ6T-Jk)m3-3mG
z3~0)&u?aaa6UrE)u!TemF<eU-upfT@#e&yBfX%an%KYo4!U_rsc1hUW*8}|2{_OG%
z__gE?)ofI1hJVoItn^a3D}6D*jMD&gCIA_3L9A&(AFT--N}O3R(;FZSFVq)Xn;{6(
z6SwR9Fnp7Rc6=KrXR_`fxK~EKe#zAZkgQklrDGf-&iJbmsovdusI>}~=#Lj6P~q#E
z+#zgm?MwZyJ3z@}roy^)9#G%O%WNBY00}B}+Z`?*k?*wI?I7(?Qs&!nr=v^xC_gXG
zn-`MC_DK2wX^;Lmc$3(E%?tzGoHPJf6pj{f>RJ#q-=mpLLVfSd3I-Q@H9skGYp58Z
z8_JJP>87J%PKQcp>ej7?mYszT+dGCqxY-pUOr^rY;i4_1x64pQ4C76_?ZO6_SmS4s
zZB@G?!etzK&NR3KLo~}<LRBDMUxA6=TML)Ag4H>Ume)X%DZgaPWr0}GKi8qQ>aS<y
zCm5FPuWqF15Ln^wWaL-UYo8E444`>z@dL`QX+R;V=pd63?oyxM_>Ns349d0Tb-}(B
zms5`CkMQv8-zZaP1Bom7z7g&(mICT%+@jl$IsTEPv2yf4*>5=~_{;{&#Y~{V+?dJ)
zjPUdEt8f0+SVcE*oLSGcgx=@vrDeB@`?+IBqE0eW+T2^mQeKY&yc3I9|KpGE4Vp9+
z+Vm7&z^f`qd0J<^cZ8Z1_}uN0gJp+R-~6L})Hhi*?p76dBJ?&bfS}SBzj@3k&A%5e
zgDl}`hPhKyiG>T@weL3Yxy62a*MpWdR&A+;@}i6RvK5AJW|(=^s*rO^L`Y7B>6G4{
zACmXMl<=wBi;qJ?mW;}MuI<@o{vVldunrljhMKXGULg#Oye<3TdAlKNEd0R(%Q<oE
zq1He*U<I5`qU&mSbgRgD`@X~4f_Ug|<Dbev6P+Gz@Dfo!^B#j}mJVf4TpZfoI;c9}
z@<6niV@v$y%i+dF*(JPKt@`-H#3;jSdDA`d;v|_X`Aloc{n@&TW6$m#yU#Pb<xh$!
zC?JkxFo9q<yCYGHd;Wd^$fyl#C!}Ci(*_KOdH62AIi=NpQsB?LzKsdtyXX)4-Ry0I
z&pXyYpY!^hM%EV@23k7LevDbYKfOsNUiTvVw!8I?NNj}?k}cvNZzw5BqRtq#b<TA2
zyLYe8{ZuDSO0D@&({MEb?5kqU{?gb7+PTQSL?0n=j2$)WgjsCo2JYV-`;-BIxOsT!
zkA`*odgACX-X}`Kz(f^LJcFGzv-K~ejp2K82);bHiA4*$`DQmo7ah_MReg9TexxTP
zd^CUnyAwSwJuV&`1zQSAqVwYK!b))4)~(_8y3kd=M@Uwl-g=z>%+Zin=sOWxo|{Nn
z*Y44)J=0nzgOR;q1j<HjVOKQ+ItIL$+iSM9Antdu0Q+hiIKE!5E*i2|p3f{8*lo^U
zYnA9ripl8aEbom&Pw-fF>rHmQ)H)9}31-7Ihfhbf?pkWqO8GRHP$<e@QrP#unKdS|
zXKb#c{eQv$_de6H!{lMHn}Xhr3!8nU_e$~JlM?0cQ2!N@=3h^<^RB*q@x>L13q@be
zH7<07V!7Q2aQmUQIksHkHf8?OcB>issl<v+yHEU+>w@B!>us|7p=K5mzPLs;KsXG)
zn_TUJ-cc7NrSd9fL~DE80egxrcctlyL3Z-s@xs|NFb1+;SOgC>iJn;Bz#bVf+iTIK
zSpziBOWU2;KY9=|yW1nXpnBS9Zx4;YJB*omPVG3NL;R{six)&LIMx0GlK@BRpe4L-
zUk-M}6QX4`Si3j4130j9@*%K-P^|KK_-ciqSh2b%jb5v56;=E_RAhPvOxz!UvV5S<
zYYWm)nNFe&RDx7a=$;`4){SQU{|%%ACD0ZXl^SpHaaM&fl!k(e{QmVI!;S)0aJ(8l
zriM%j@d&)arGA0*ipKV)=4g-%Jt6~t-UB28&n+reoF_M^08E|mwV2sZ0JG;&ing6f
zH!V(|Y{iDqPj}tBYB_NOL#6Y#I~aKy;&}`KMQ)i!QNTx<fMp*rBe!Sb2(tiL*t~kK
zmyJx~23MO6>Cuax7x~R5P6>gogU705^D0EArJV@tw}zo}k?;_51+e8%h8ev#NAisd
zuRVUeV4L7N;&RS&8JINsa6Bo3_FZjqTcBm2-J1X#fW)J@y^=MGTDu+t<#D>>i2CBD
z|CWJpGaRX+FM4&{XH_4qze&~wG^++hj#W0@_yK$p60JQ8c+mIIw7)R%)P<0I9<fB|
z9u1fx=(M#T-0F@eLkI%)KNhpQpg?zefLGb#@}(6`+S`Da{$Irn9(K=s>^seT+dc_)
zq^tkt1D-}z-^`CT8NU6#9n*9LJNye{3d^2NO-_bxlAYVfS^`;5lHjiCOoJjdr#3KS
zC})L{2f<f?H&54<n6<aar%j@mNfBZ#;ftx~T<B)0ojEhRya$rIA9o)6yC<7UeIuV`
zhJU@VQR!jyA@9-(3LR$VA8}Mp1wCgtb^PR`Kj>E){$#Jujw)bzRjexhH7sw|-N;Y6
zcN}?@YRMzQ*3gesmB#I^WZm2%#rw@^wV74Cm|5MW%{lQnf#$1+`1m;OlF-N8Vm5gF
zE`n8mAZLFqv{Bu*d9%sFA_=T8l6ziHFY0v1wH6Os>t&yn`J~69FoLa^oK{i-&ejj1
zjM%adX8S$ako(b;;)T`Qw{Go9k(MXaV1+dDRnve6;sK#0*1oFRsLzK=LKx({>va#j
zfk}8^^hd`+20*#ZkTKp_q>g2KLc%4mE=r){-b_&A=|23hW&@+U@gtzqEQLM#LgbG}
z_Is(gjCCfvT`tF|S3|~aC>GGpNOLpE2RxvlqQX3)zqaQL&R0Bcup*%}xG~uYgb%?l
zG@)A6Pe1oZU;ihNT+f5}jYo4!IQVT(A8d|Opx5U1uRd7cMyBcIDMIKnVwZXhn<ejJ
zgyKAmP2-VqzFGH2p7-DHrO)-QLi#0YJb6DZ$li^eHCUPa?3%NIMkgS2d3-RNg;i3y
z8GFxHef;4M41V4UzEQ+FDqqv;#+KY=+~FA13Xwui)6<0ItE2$|5!}2wQ^^DvmN(}$
zFgR$jU7y-@iM-3<sl9Bv8}KLyv0uS4>w~~0H`T6eXCb<;?MFZKlj~FhTGs86{0Xl^
z-2y&?rH6MxLystdA?Vi4+e*jpbi0(HG*WwV=7)xUWWv@9`CX9snHN3jf%b3`zO?57
zNWqNow3AGkQ#0~S>gcERUht#h{H4dl+=EO;m%7{8S8a>LGPn!<{{ODXGI!<ZO?{A;
zd?!wvU@eid4j|Zi1ABnCG#Bqwl=nRQ%e@_a1>q=V74U9^gj0dl3^;svH>i{8Wqm_s
zXhWiU_zaha)S_q8_dRYk{}dP5`c|YnodvfXT;XrBR<*@Eh*UG^wVzabxWmW))3_w^
z!rQWg89QK(bA~u^o8k2n1ydoX)s6-2QOy266}ZsHw+d5ZhxKah+%%tF`Wf(wRiO$Q
z(zdPA1T1{nUR|4i@`-c{c=vVABF>0^G}=t^I*1U7phi2|ks*Ps1k*Kn(nOG!hDKWm
znl5GtvT7Fz-uA&JdqrY%OI|zHml*3zK{FUtekm?4j!#;;V?>E$8`(KI%Bv*4@LpYZ
zZC))px}v`zZNF~|H@Cg|>TU&l-YG<2se}f+Qw0LakXOy8m}$B^_#e$nlLZ4Lf2t?i
zgu!jsmYslE16B_}(Z;s-JMhZ8!k(1I&G+4#?VAQc7RAO=*I(N6ZBsjTq*s1G8CZ1f
z;~#eWj5IK~UW>>Dg>_X%*i&lMpn-Y%pp%uZFA4TUA>`ubw_?zm(F`;0z}D+uCTMsB
z>WSAuEk)ma`)wV1n1S$+)%7<BnMSYFhZ0<(>L?Cr{)tr}<3YoESpyhj01;=-la98+
zAHx`3jKQ+I+7SAcu$?94YtB{$0{sWeQhhvX72Ie%T5m$P73j+0=Ryo+Gg}U2JL4sL
zASg4iy7j;TV1&Rpw-eS$)5}1|`;idL6}BE2*byH1=v7BHcLbOKny>~ld|<LkO3%pc
zsFlhC(X%AQY+h|ePGDw6M<5pz42iq&x>hm)cZ4BB9b`Q{15hapRYqISr)}pxRWXZv
z{(SrH$=US{Cv5^VF+y_~!LZq2?CG34jf#L~=qkTj^#`!?;(k9UiguFlnK%o)x=A5s
zM^diIQ0Eun7jTFacFl3YMRKyVt-N3@cV%lNGwZLx!<Ecy5^D?+aML>-qi4a<`gW`@
z;1o0OTTF+57!gnC^OhS&5wK^twCzYV0lu(n(!JuN&MujzT}p$yr7*iu#bU8cvSSj?
z{HJXBoc((_j%8C<)eifS$F4_2MYAh-8fSe6_uOC)<nRA45<0846q6$h3{>#HDMii5
zPprqDV{s;Wm$=n=6d%X8cwpV8y}I3369v{Gr%iS%w5`Sp{mY}<pWE&0T|)T>ZTmOO
z`M->FOd}l@sn<n4=dLve$4bSmBQ6xr-B1Byr4VRVD1bUwTo-Z}@Yn-^6b+C=0zCY!
zi5|v)(hr(-lyKu5K4?&a^jF_|W<<B^=q2k;w;lf7y}t!^P8eMRd{b~el149u;5*BI
zDMl~YmqTRc6;~O<#_PfbfLjdiu4LDm>)jk)U)sx|yR70qAMnnjq0<Z*D^b-mT+I<p
zdlB@4*eR=s#`h40Cf*;BwtUA+WA4Y%I1ji$%E%8pv3mo)io_70d>;z>TTSOUjA=*u
zM8ODQ{)UE7P?y9nKBaVD2Y00-L7k=zuxyD(_>QW{#mk}cvBTRY=O(m(ZPxcgLiU5U
zAURhXbPg~CQ=Qs5Pk=X}M$K`5<c3E2y;#xG(Rh04`AE+s;>f6Xr|!9P=gP;IdanJq
zZ~6k4X437E`9-1|y7jv%a?2XB5E^jITU;rJLS_g(v-8M@w*yVqKlfVm45|{a7QiiM
z1jyfJu2k^$+rr#^i>enLZznkAmd1*Ea4P}A9?2i)!tqC)B<j6%-Ho8j0Ty%*un5~v
zqSZ@fH~Gt9s(70bDJ@XmxEyyY9S`9Qzvd^qf1;rj9m<J8LMOcv#9LCHbI(YFJ>?V&
zDufC|Y-1N@NfmT&8dBt^G(0wcx4ZY*dQ;IX^h8jm5pgzE_}A<%PF<r9M)KORaX%Tq
zY<tLA-@nhU0DaKO9v9on(dyRycCHl<Su?aOK4+fvbqE!8u&VjhDaL&<Afze2`y{4<
zLc_zaG>_l?rptbEObq{baSe5SyT;UwZcY?zwRkHX`OIaniS3@aL4H(NFBwj~$S#7f
z|BD<3mTs*P9CPE(@T>4QHS*hTQOUxOFlW`cadz3~hGrY)u{zlkl2W{l+Rq;IsP4;`
zwG$s-INQ(h{md@Vb-U%f4Pbu7>Hl2~VhicU`yeR2hs*Seps<z05B@lTuJ5?bhf7d-
zr+&qE#ee7eH@nNY+Z>Y;IP$+;4vKd*0arJi>jDcPVT8U0NgrfWNGRLCYDq3o;cENt
z%U5dp#_O$vKV<b^w6yTnTHhNexEfrx7?hoIa>xIyY}hk&p8G%lQsvmXH9<G*dy``a
zU5;xVx4LJW>}GZRTZND_fp_kdxtbmu&be%IHZbPH-%-DQ{qtRceWHK6`8>N{V`dj!
z<Z6EZ<E%bAP26d?942?;1+(*w-M9+^+<S+lS@RbWs1II*=`L*W1$&S%RSX=ewsCU3
zh8A=WnW82HMe-`p9xVrO4i*<)$--{d1lg2f{{3fsk%S+`@_^`62g4X{4kpOkGD{~F
zPhlI6F4s4-en63X3K|suw{h|oKNf$7r3;Wd{nC0lP~srM1OMPPh*TVaj@Dq;=@xP^
z^u)o;YQ89K*0}<cb9^}SKCjXXERm{DStjb_Oc}nKv*CGECI&pD6SA9;uhV`oXWX1S
zS%A^PN-|}ITKxP7tQ9SpB=7GbgMKn-B5(bIPDiB;k)ma=gA&53*vRW3%2KzQ4s^`v
zzG8%r?b>X-V+@<Lloq8IYvfi$dcWUZ<eMpIfmr0$^;d?a^jM*(+l8sJaj&D^b1%88
zfpzYJkDEU2#2{CvfDUgk<_?f|Yu%4c{V2t}cBk#$j`lb;#?y?Giup{vS5h_@r6AvG
z_KbLXBKvTU_3+mY9(Fx5ot2{x?sTli7JBv0mavwEN1yH=jBz!pmKe2JS7r{d@S%8e
zS$;;e@~Dv_qgIPmuc=a-cd>k0X2Nsgs$^5_oU|N6{Ahx3C5vNHE#*dOczbXK&#=)E
z&9DjAGhJnN5_#6=HY!XvtGjGSv3dFGmdZVinc5PAl*xWQb@bQVBqTHiD79ukMVkp|
zX$dWpFw>!+SvBrGTS8@NFVNV=c{6=OsHHbUujb<kX+i%Ksu8ZmksKA;$AI+{aX9O5
zVPX}YG_cp}%iF?{?0Nx4nu)PI*9y9j!5S1Pf{cLaX~3^_9oti2YDRxMcU#r4_v9zo
z#E_5lvtRfI(EJ6X>@I(|Snu++WNeab^M;D7i)OJljV<-(%-ladZA?nB&gs$I2F$@1
z$eB!{O6*>QVXt-u2=AfNS#2$??GSKzu6bWFrO>62OLBxTyLtpq*^Vrz9V$8G;|*8v
zy&&QdBlB#x(eaej3_tpLz0=1X`(Dk#rq;;1LbByCF2&kP2}sx9NmTMGY$m_W?!jZt
z&8Y}W!cQMYhr)|YW-s<-O|BYyPE?(M3--G9L6~=GeB4>3#g&yXchU#2-W|AgSAtL4
zOE#ah&2j>8e+@|0NVPFU*;DAHchF5Tz}|#Bt!{iLnIQBJ;U9m*8Lt0;zi0}agS-^4
zf_FeVJDwmiR8X;OdkO?+#7V9FOx0rRt1sPX74Q?{o=;XUKnASYZ@WHXv}6$)CqKyX
zHR>Ir#jUpk?D2*0AmKF$XQ5+#thl(<Slh_ZKNE4869Z}1YbW7+>oj@3J3y|M_?iuE
zD;;FAnRbZ=!NhBA&(-Fx$$9GfwMIw#3+s|e9DSnq^H`DlquP8=2my6HxZ8JbD+xMK
zo+$fM!Jta#qX&gDVh)R|ph=hVK?R?{lAK;}lPVA7fWYK#QXNqHB%X79q9?6leIBC=
z9*P1{$1Ou0RdRKP+r`y=Z5Y{*s#PK|)532~CjU7vl~cy2+8d@0tuz`N)+;3oPPqPA
z8a{mBD4$%<_^S5ETnT!<x0pr_kl{XRRC;3Y>x*2y44##Xe7EDTG1wXB_DG@_A%D|u
zm2le+nm&m~A5F&&>&VU3B}m+!dZi8BT(7*`g6h&Ic{xAdJe5SU*>XT<UbV-yQ*|Z2
z4h6uP6*R3&$cS(R7%(k@0kFx{8oW_j@_h4hh~yUpC5;6gOcGfpv>LCT1fr{Jliy~}
zUoUBM6>=B15%{@B2s5kyf}9>h!4F*SM^Ce2hu;8+B7by&&VOSx)_wligNTsP+k-fA
z4Mkr0467-5p^hxPi@p;h&t9gKj$_bJ!>yzJ3q>r`7cap_hA=;ep-=O_vGL?#nBn@w
z8+HT6UW@&3$`mRr<%%2k8u!qRY(h-S{1wx<M_-@J9j53``>==me-eNgBmklZ^9#SG
zymtMB;^19lX5DA`=#&jUw*gbG6k?V3y5YyBFa2g7d5k@9=qlc-zg=3`@9A9Smul~g
z0;z!i=nUOE#3zhZTq}4`)p&RzqMmIGrr{0{M}8;+o`L^^>0w0dB4&BXd9nv{*01NI
z=Mb2)-qobMW`y%9)VZ1zpKQN1L5#WBj7wn__FRbKY17i7ygiFB`CT*c1eE(Fe4aK%
zMtJ*6R?RKY{%PC%w#2%{41bx^;+VpPi+!eI&2icYM3&`-GniYf#&l}T#QnD6X@(HL
zM@-6lRazQEH|m|;6#m>?#>h+Nc3YpmJjrqtzxyBd7Gw2D-wr{QPNNVy*&g0(-ww&W
z6)V%3z4%$_cV9WfwNcC6xUF#%QE4~Jb4t0@gVC1p0PW}2Zoi=a(*m>(`-XwBR@A?3
ziNkPn|0lsXVIwikyatJmwPi&!jm5ak7}L}n!3GP<f*3g;@76_+`Bu%`$Q{8hQ%iR$
z1T!=G5^U1e&Of6sp|4<k_~21KO_APl-#iDa#GQ0#0>Og@bih^RBrW_1sJEUf=I}`h
zJ*~D=p-ucL5CgKb{DAEblX^bO`(26Bnn+Fr{6u?f4m{~;wrGQu)oQIrSKwLDsV2zf
zB52W0AV`?P17FCe0a@B(3CAq${oQR<lSs$+RzZx9(<1kG8&{t_u5<3MXao?#0noQD
zHR&m@f(gX${j1|JsovH)1mZyP1})Wr%i!GEg!;U!Pxybe|0Ag$Y*=-i_Ibd*+yu2b
z|CHcVk%7hB=Z~C`gzQhCP&40fZZFDQfzgs;uvrIsNedsZxGfX*5l3JRu@&pK4*zu}
z?@&i`Hn7q<-8+x4?SZNx%pT_D)}#B~1+lM1`{RINe*=}>gT?>K#Y3ln-UAEky23_U
zghM-*5+41Xt`y44AI$eI`L%McGwV4A2llxb;`1h>jyw-@;A8Y8eRZVcJJJif_4>aP
zG(zSHFU$rnMO!BV*qN-CFRS<d_nw@1tyw<!E3{3VVTlsv$dSR&I+4XP#l~RMAgM&p
zHYnx|1Z(#|R0=CrmGD&`*y0U#v6VG@*cU1u50P~^zvSdQLwT;&S%`GhEW)J8ND?@n
za)<TLj2b3oOaOk>KyqV6eW=3OlvPxYA($E%>_@n?fnMBQfxR$db|nbF*O9ez#!B)7
z@E~dXS8hEpUX7u-VjPl;qJgLrfVS8M<CW|y{u{)qj#FV!9X|pF_PF;<PHckV3=EfP
zVa+>X8gc)*Z25g6uG0*Zv1`t)c2m$cRAa=cx%++Kd0Tab0#@<UHz(^(ikZKDbv;hs
zQ`zu<-O-1IhMkWBAsTn^=pvic6aC@pJxH4>4~!!`_;JT;PfRudS?dL^xfbAhXSmp9
z;l6oH<>K$SnS4>&Gr#+TYN^4DpSr!Yy+RU(fyg;IpzJ?nu+JHm%^VMLI}IwM-oyEu
z(f+Equ2Vie<1cIj-1x=!_|z-aX2|v@yl2iyhbcpXd4IySvVC?2<*ic52vm-B?ZPW$
zy$ftO%taRtb#XqcE1Cl8BEA*swfp%mjHRjiS)p=vOD}+BN_aF$`d?q4vRYPTq29U6
zT!G2k4p?JUOzBA;n69k}OweCO1g}(;-=n#*iaC<h<>Z_Ttv)?AxLgf_`#nH}aM~}J
z=U`ZW4)^c@vzUx^75NSszG+5|3s3Tng;40>>Gk%nMa3V{uA3I`5%lhra0k#0_nL;~
zW^c18uAU#(E$-{h2NeD**%Na{!o(!ypGi&d`pZwy16xMYMx+V&t;&php;BD+1Cm<>
zLOk1hhWZ3t8142UT*lNd*01_dN(E7cJ$mfygCn}xyzH>|q>U(ErkF?W)YZ{yw9N`R
zh5M|+D_JZK6Ma-;BJ)uext9!{#Wos-UJT{c6}-U_GOh5MKpb*k4)~!BJ%a-Z&Kkrd
zt!ux`pxh6i*nRPVXu@A!nf72oIb*e6_d!Y1)PDHh=<W=A;5n$sjpGgP4J|(S(g>_5
zd=(AI3o^z|n9YI0Z*EA%CrWgjlFP~5?uD#=AZ&uihp&Vt7pyo-*rcUddFqZ79zGy&
zJh6|<SdOZz;*X^w)l3|2XGPPRTQV^%;V70x&8|PgbO%r7C_4D&fZC8Ys>Em&RB3D1
zGm(Q`jN@2@a1}_@{3?Z&1oxn}`?Lc1qV#CEi>`7dQ*Q40CO-+A5q!unhc@^NhKc-q
zz{Hkkwz&J61ZijVmILc&ifOCKpjU4uZXT_ufj`B6$W)i}_TB&;>9os<_Xqq0GoI3W
zK#`I5<yi{3skg@aT@YyiM*b*kwiI2UFj=|g!gh5eJD}zyw7_B;XZ5++hw<8Mt68m7
z@x#kB#-$C*Uv>aUbZ9q&HuhEfZ^p*#!|eJmhWe0oT6l0+H1paW6<+ILRu9=3D|H%q
zIe2IqUVHuxf9&%VEK021C1ykh<eC1cl#gT-52pllv<~1>jxZZIEgo_}!QNyo2Z4A1
zSt#eXTcs9afk94&{uU7zW$d9as5Zl4H*CK`V7=L>rbLG<GpVd}E;o8me#y{)OUX!4
zxrQ~VmYsUkjelBvB_psjT$yhr=#_U@r*pG1=ay^rb#*;d#T1UZ!i-$c>r<{okG&M{
zwB*5i;f60A4WF?MyyyG*e_w(k9>ogAb?&Xop!_Vg+iK`Ol%UIHEs^W3y#N&-qq{@1
z0PmsIh2wz9=!OEuWcqK|3uLS5tvDuGF|S-FA~rvr=@T>QlgPfheCRMXDq#FsKk|yz
zQjC<7dHbWyexqol`P;)BXj5;s8C4_?(Wmg2#ik(%Q(#SF?Z7%aee_XJ5BI00*krPw
zz1--J?)#AKH#WHMcsTZz+$*1NV@1Xu6EvSnZo5u|A#H*-vP`P`@+_<#DXa&JFeth2
z!~>-HQ%<fws9$sKVrWxQZXI_v|MEK)QWC8xtNtw?UZ;=Hn5CUg*Tl~+4VgHlH89H?
zO~(M{)j#3Y`-13{qeZQANaF9jS+rsn4bmPcg#u{TO0O)S?C7i)dA~mYgSr#3gJTdv
z^Nt}nKGC!y<|AP6nW)0vqRwkCH)P?dRbhKjPpFdc9oqVk5Mu5{o4jb>!M0u8OeShl
z+O!nS%wqbi^M2`d#pA#OKS<@G9ZxgRrBco;74%>~#O@4wBH^pFI3=LWD<j9-rU+a_
z>8>tvU%!sw&M9H$t(V8QZ8k;?O4<h4NF*(-dhUIq57KD~(YC3m`&Nyvyl0Hn%ULnj
z+yeQOM+6l>5ntFBs^pkafICd<#Oha}vNUb~HVT@*YKUNrS5}edVy6#|R|~6q#tcbU
zd=OY{tGkk#p6^F+hk^Ht9}ZP+knFs>nR&#p_&BVBOVc(aCOvdiG}_d!6^RC_1|jsF
zuzumS%e3V$11M51cn|E#Tw67T!gwzg+sOXrdnB$cFPr;WidH3RuJ!0>q2u5w>(D4Q
zNClGUOs5D^D}7=U;E7GwdG;M}^6UbicmkQNS(v^oIcU?;yND)Q;Y|0<^oavl(^hvC
zuk<)Nh@6q+=j^L*SH#zUYnav>9+I0^sJ6W*GyOg-HgiEozU@-tcsr8Pl?w~E10xE{
z3fFK8^*bHR%>~-Vokp6Zs5<dZ(ag-hEJ-<fI}7<tkhh%tEojB1#gZL#xs2+QzbH$g
zMG+0UzfdQbSxFA?NT7-xGt$@^@c%7Z!3qAavAZ!>%AK7hiKB>jWd5C7Z{GB14;2}T
zf3%<!?Rgp(BrqQ6b{aKvU3>K80;=HHwJO-(T0NkGr!V1(y78piu=Ua&)hRdrXtgC7
zD!P~~Wa=$+g5jfb&};>|`|afSpOf!n;n8IEi)|&#@%*sEg@=8Y;n1XT<_b;Ggg0fR
zw4nGmGJ1p>?9|`_nt6U>AGEh_+z6~=3P2CAMbD^4N0L5nEh^w>hiFi-ytuuB#uae?
zsfCn5EW-z{DQRRC?)<fSE10X;X9?kGokUS16YTh6_%tTYn$y_Y$X=cy29C9SwI~3l
z^n5^7)CNh9=@=#A!|+18ccH|NQNhHwxWHcn`Z<(Q0sO!-yUjdWvrqzqa8tSL0;~vs
zQB12jT1{={9*q{WGjy5Q&3v_Pf%ZYFVQQQay}`Jdx-f@%st89QUg}rRw=VvPego84
z5i=fZrOP-_sqWf2FDU!oq;;$ydDhLx6xq3BH;(9-1+Ev(D`(rWAXmJC?^BC6t5*JX
zvoaEQEgjRnnV9OfEa)(gw#_c@=CUi?-zx&%P3JPnE)KSt_zO8cipJ<6A2>h-m=A<}
z8SRjS5wuxEqbcOV)uRBbVyE*o&m~U1p$PzhAhp;X5JijAhtOJ(^5bD<1=?nE!p5%y
zrK(#%+r-pmor)64+MLbqK%XhL*BY(WS*hDJ%+|h)icGW(e=7{489ttK0p|fAmV~ov
zZJM#jGiu^G*`T-DVyH+h*@<?An0tAVoqe^b=D;F<Bo6AjMSJMnvRkzD`EbvjvhmOP
zpdM5NG8WC0+Kp9@-8GG0cDVv|d{S0!LVZjt^AuW`c43oHUTe86Tq}vAC|qmNP&?X{
zz|g-BwPeQUvPtybJcId=Ke#ETQNLpzXo!K5m+{aiY-{YLI>8|tzyhCH7ZqI=MXkwq
z3O}KqNqcYc1!Z-{r9^7>*%EMkx1TQt@>^GIr-=)#UW+<74D_f#=CmBZyYQV1G;o|u
z!&R?W(_Go*VHZy(dzOHPOER>;spIee0ONkK>iM;Nxvl#K@S}LcSpXGEi!wADXwm6I
z{s8Dal5Cuwf3xF?fe$!){#X{qB={g+aBXEi`#3R%!$+LFc6&dH`}6b5Qo~Q<Cc9nD
z%Bb+HI0_|K^sG15IaOTgY{DswW^`~bf0B1wHa~kN8<99*(G3Dcr|Qe7AQfnN1|>9$
z(kvK7Se)tx&KB-af=|xRW-p`feCW`D4lVh`0*Fl)_y@+Nnc3o4uDvEG{!|DAzEUq#
z0FyRLwS=iTJhm&H!rC5Z%;Kdt;mne)K`6QeKIAx$Y!uLkXaNJeotEQfqlgcvp(Tm$
zkbKe>KUk%vhhoNIAkCxSJ(z91hfg8&3ZAI$EWvbFuxn2N6ZJV6lE41_+e6m!bXHL4
zxb;%(>U`{z(sC18P|?Ud<u=*ZE^HnV$So(!wg&59Y`Y`t7cQ-g6UP-&B=0G?1~9@3
zWn>&m3d?pP@-nv-Xg~DhEI6bPO2J*z?`te-dNj57RWU!atc=P8so5-m2D~`Yuh=V<
zncmZ>0X0!4989W5G5GW-t^Wekg^NQ0RcUjE5BR>pBi3l->v|XIJ0nmlDJm+aIPbMS
zP;JDzHM^0iT;?&`s1t;Z1)`%8m?LSt{*V?R_e$N-SzNAoxlx)I2+BpcMn0PM9CQz6
zYHPRGGix<tsgkqZA)>H(bl`z+#gul88$Uy8rTf`Z3uBL_1!}lz)-;4L<<%_ln8@tp
zDi-wyPGfN>Ofb1|9nM#)X<0gFcxlgs*xppx8boFZ;kHTEfdiK4UCT=rqT@$S$^Dz1
zYNh4^{Y4n~JYJ3v7`gJgp#e~DH3a?bh?lK-z7-pNt#j@TVl*ZB+Ud_2_DTE9Lu*I-
za9J)A#(}Dw7gx7#(UWco|DqgDXm<DOG(U!9F@?h2KzD0V55*Gpy{(l}D|wqAq#Xr{
z4~R<%Sj-rkT>yxzsB2HBtK%;E=h#Dt8+I_6@8vz9)_ucwDKz^#tPGr#r3F}KT~r^>
z*Q$7{ef0kV%)4<x&*Jrc)sH8!k~=086_w;X`WnHs3e+Wi$UGMgB5E4_P`jPiws0I1
z*A_x5`!`CDJj-RN%PkM7Om)lee?Csh$EK+Np4;*zv-RzvS8V&GwZFbp-fZtZ6`fX-
zk{mz6s}PDLC}zr+AkbX>Fx;?(pQ$`!aA#jcj3eJxrumE0nn}O}EYd42QqwdTb%_#R
z52sv&2)ll+BfJp<2hC%Ln6oO%se%e^3Zc9-hIq3&5C{<;Fy-^uEP7)dyD3ImvIwJe
zVeo59uu;cq0$Nb@YO#<@R_6rJ^*$J8DC$~3MOTZZDiOK%EZI9cT#{)O%K6p(;Kq|W
zHFlr4_AB9frFB|c9Nq%Fj>=Ys=syg#YSu4b4??szeg#esN(WjGE~c-}*S8dU!i+z0
zcQ3sH7|_HprCu?MS&0y$9YU@*pTw(93}j<nPY7;)@-Mybk<}hJuv%hXIlbC{TVpP!
z{n<)5uQlrigEa`H#D$`RlgeW{1<E8OV6Ds)Y#Uak)qxb{6KtA+Xg!>MyvAAZy#D>W
zz4u}2$u~OQ4Aeg66e4S_u0@RI%4f~Q>i7+&Ed~>M`{^@fg^U3dZ19k~$?IB*;Oxd1
zaPEqV;`vbC`Kj1dgK*X>kN}68PEB2??tOoy*D;j07X7Y9I5VQjEv+Gb<d;MwCEh_1
zc$|EEct^A-5s2nN3oc#Ma(JtrpthvKI-Hb+^m5M=qV=#QC72ODErf>LbaN}l|1X6z
zwSNrFhPK)1M+XmxQyn}+d^?&|dS<?K*d9^XkM~Kfmr*8MkIDd>90I8$ROg0~Ugy^D
z%&bKD@mu=P4G6c3=VbAj?U<6KO-$K)*;rB{G`ExX%_bNb8(JPUPRiglJ9Nio%DIv?
zpW0X&UcB4KUT3BFexw@MsnGYCrQr*^e3(~^>Jqp6WMXl!6wZymqmjp#Pa)#B)Bg9e
zhh~d?_z;#2-_QG<Sh~mG;R2OAtmm#~gWXF~_C19)nxUAi>O0bm>tl4egPtA`mJ4fD
z1eRQywNd!%qk?PE6KaT*7vJS#2!g7q`?NrwUnN>Ik$H34toUsUL`RcG&s=c#a#hS|
z3HdxmEJepkidk&TSn+n)<S5^C%tOsEghi&JuxnixLCbeDUj!S<Dzs5i<%r@x-lia9
z@J}FqCr6tlnFdbEMwzpy=Pm$!)VREeHsT=mD=?t3_zY-Ze=YO?F*EL6I-%Lzh%mWK
zH3YJ$g;RVAm!I3?<x_RW4A#u}M^d5H`#bEk%&3#|rMe1FBXa75_=E5mUTgVp&d>gN
z-7JaPHFonjL*(N}M7Y;&(xLGqBhAzSKa`nNs%w;;#+P!V_DFUN;v>O;N{*}<>BG$L
zH;bnW^O+s0ZGU9YD*lh%3Cw~^aUQ>V3BtO&B!{rH_lC0k1Y3E*nXY&4rf<OFNy2pS
z3<c@Q+piC_&b*#sVrQjUmrY>3f?8Hq3Ti3^AW5OLM3*n0AmazwhRRiEq0mWvW(%M!
z%+RHc4->NI>+E$%wu3r$?a9uU8U)slmoM7goU}#nSqlptb)!|R0}Ro2YgU-;c#l3Y
z9spFoip-{+$M)G~ehamOacg06F;^rwnXX*suh3wILiK-$in5!9)95%AEW_(1i+W+|
za?u4{^yDgSzTNvW2~!Ekcw~4sFzud+ED_#RS8SQnAR*}E=Z%J4F(1rZF6%MZX-9dy
zjwT?RxvZZfie%R-jN#Hsp~szCz>d4l5MBk^OO_-9b{(F&7x0_P(`ud$RrmAidIaT6
zKZ$Ij?ah7kVxo@gF+;OK!f7S%JxaS~XPly&a`B-0HJRy|jBo}*L|LiFG1fI;Rhncb
zImmd3K1`Q|<|5if4*6KP&RTH0LR)C|l*}z+9Z(zSRE{Qx6_(lY-lX!5>RPAr0nf3~
zWgvLU^Ff()xyUwsBT{*ZVZXq};lXaF*COAPW%%aZ2|+e+YMpEuTuTjD=#o>nD2G*l
zf-PazhE}-BbU5y~IE@Zy23MqtfpM;34b>SlC)KP@u4LH%7lTO#1|zZphaWdb^ab0U
zjTXNQg~1$2i0v@!v?O%<O78)F=^Mp&>A=Pwt4k{QZOg?I`^UMA5OR<=UKKTg%cuW^
z$C|96<`SB=;qYXUi0*@={jJ(9T!w^@G#)x*-Bb1md&sk~S$0r!fVThX;#c0><W#QE
zhPLH3`&%;ZbDYU7e~dAMgTzlKux{hU3icigR8Y%NXajuCY5zZiyR6(|e!0OSFSt)X
z$aICPe2(T-0NEYOixMZ}%i=o(_)_;jl|rLb>Mn+Y^rMjLHv3n$ofWD&Q6R`^%+8Yx
z{*RaGm=N$#8gz(Xdig!VNZ(PPc_pL3xX|oo70B9D8c(v*(gGES44fLoEoUQbpFVC^
z!wCqxGCDqm(m-I@s|W{vz^1RE4>2kA*dI#8L$vla8$&%;03P#CIqjLny7c<gLk$~l
zKTb_9yo#}*ZSJxPl6mNxmJ0GkaqB{P!7Y0?E?ZddYe*2NSZlv9lzgA@sew27!I?+I
zfmq_SH;PIkc%2$}s_3}3`wWcrj#_B`oc5-o^s;T@X(MjhP5fHA90rk2P^~W7y2+~8
z!40M@mi6$I?4XHXx6!yaeL;B01LA*NkgeT=-%oif8|PSaK`a};jTg)yJ`eNUS{@^e
zDU12LjXJ2vFxv0!CjLiDMEnS-zP{hS-gq7~zR{2=P5x|i&#sdoS6(UbE|V7v=}t_0
zp_+;q<L@U5KcSmo30>pF?t6&t#@qVyw)w}~D74M*Jx%<wVW-iNo|)Cy%2CFlwS<M{
z+M|*AXuPZp=oTOCMz3Dn^B|G*pB0N;<CBVir3}U2WvoCZt+i8Ve{3?I!y2lWKl^q=
z%k0{<K-A7tyQ2%i(nUg06}xg&LW^6B$}bu!)m=o1%iiKMn0azaZP;-sCS!>qbTlEc
zcYFcq&bW6x>XlNPM(;RsC-z8*(gx9lMq6;6p{Bty5`$ldD5zER0-ofGG&yJ{`Mp&q
zEv82>(%TIk#apa13pWE<Mfg$;Eu*I$)6gFs8(*f5DF84-B>_V)tL3jpOS74hFc~^b
z)Elp4EB5b28%h8_tv5Xb+&?8Z*amWqUSys?YK>`uF6jxX86A1nJHFfFd}#?A&+O?8
zsKOrW%cKShm$DxC75mbvSXz&~t{%1hZh}qLcAnukYhE80jDwuO0Z>6TZskrwSny`U
z`w8?1eRZ%K8H;!{fxK~a&LqO=#T?ij%%{nMVT*7n1A<p|cmHV<jOvQuA6*}KU_pCC
zHNBOq&``w@@$*g?QJ}*r{zqg$5$p(kGHUNHl<y1jwg#_f8v(2Um^Qxs>*s~c&$t*>
zz?XJmTEPz@qhn;(lIyrtA9FO62ZEq7s>D65|A3q!<}HCDe;a*h;>RGQ-$jZdBW5R<
z0Wo1NaVw38`try02E>^LTL{i0w1jpJKJ1|474-%OVFUD_X+t}$hxFG<G=YnsR{K2V
zAvhmo9IiPmBoqksqYMzdhjYLoxq<b)E~2{D6ljtI;!h3jz9aPgq5E+1TYT{+;mP+!
z4gGQOf{;_Ujh-N_3+-**DAjkDAcaOJ`fHzHfw0^C(_;s)kX0lySzEM+WS{xoLCmY-
z5$J-5_Dw?=oT(<k6pC7Z=(=lgmeoGa`ovHjNvpNzkj1wJ%gpFtw&5?u8k!6-#kF^M
zEmCs^Italdlwyh44y4WDP%o<vEyD*sZpoFl1!xMGo|&#Xi<lQoe@xi%PSg=Mm-F$A
z=*JU%9pPh5e0nwDA`|>3U({^cmfURt*w0A^x)Kr51dHa<F3IZ{qj~NHBoYKqpK1U&
zk&LdWY=p{cJi|RG7##8J%U?|5B3C2Q?S#@X5dC`QU#5)Fc5$Lx=TKEYoDn0**c?Ht
zR966YEmd3Wwk_x~fj*2Dq@Ex$#JeIN+s%MmY$L^nCxmabGE*M)Sn3ydzjV-g1ZHaA
zuK=({8~AVxvF*SC#tl1ceTWRBWJe+(QnZ-jT)rwC1E9o3%Tm{l1qO&d00Q&V3Nqos
zLn%e1n?T`FO&hvVvkq-y_y3<3;Pr=$H^?I=P;B)A>JC7!SA&tp*y&SgVM675v7VNW
zV@~5qL5f8uaD`2A@q*Cx9~qRur^-UAuLK>wmw+xZh9>k~bKwKvk6;WH2oZ^N0vTQ`
zUxlzX%1_sO10{PPh)`ZMjA%@}BtHbG8hsK!Eruk%4D+mC&X*XWw4y!AcZlYNQ%SO7
znelW{P&Vk|Io9ejYuywmbiE;$y&rUZJH)oE34US_kl|D2*Wv&1@2@u<(Odk4r?4G2
z&$tJpmv?^Ha<CHjaW80b<q|stC=(EvcLXyE4}cWQAuL!dzrglK>T~B1$hj~N=Cd0D
zF3xe**)&`3Qz*g6*ys*~w8y<C{EbO0;j}mIpvg$P3r=k%wsfs6diNanU@0wjko0Az
zgpG~?(%uBfv1%~-hRGf?9PRbielAn5As20#E)9<gsOV3)qRowfZoOpKff`3@CCGP;
zeFg;*$j&Gc1~Fd|;9!c{)f5$h1M*W1KwK6{X0pA%W9CQsR`lO}Zwy3;w@8cp(ww?~
zqg8My0~4=)lj=wEJ5^}aIIQJf5kuNK%|eX44_U{VGjMPebwS814HAPjz;(dJ^{d>d
zm8+v~P_;`-+c-ExuKn{ju^sC7BEif6Lo@;Q_7;AYpKxA`xa$mPgEtxaAJNM_OzH=`
z!l!ROeZ~M!gm%W1lrb9$24XJfa12dQQDh+oPnzM7G=2D6RhQT=m<aCMenjUdk`DoU
zK7+9VfjuLwl^$h}Nri>q`2?I~vGdSoWRTQ1=ZVe5q1{xs+T=V2zdTO*hrW5n7kn0)
zW7Y6^%1;qIjhU*SNk|5nlkWR<b=t;W+Ykc%AtpbEaQy^LXI9mAeT|fVnJhP1oIw`g
zNtsT34hTIC-AER2((rkx8Upb(<x%Zr+f@LTyAD@BNDc5gMj+>Oi|D<%!^oEiL<3oX
zZ>4qtNjWH|=Aa!gg&zPtc={1=l01WqkeXXT3~7H(X(Js@0MWIDUALGM3vhRgVJfqy
z(K`yzrgma2;b%pMEDnQMZ*fZl?tE(N#Pyj#RTKYW2@q)|PDaeEMX><D1`f<N1mRy)
z0?Ic<*+lsq5#%he^v8AMUT3PBG#&#l<|5p<WQ4T1xRC`fmJk(d063cBbsyD}lAv9}
zuyl2pcf**ICDCk%y&Uunkmodu9(z(KgY(>zd#<P)>I%~!xU#%43|EVMy)A;!Uu>b*
zIq$XvKq(GLIlUY7DFHx7b}U6t0RNXqw<hp<9BG7vB$^8&T1iN@!qkfd&?F=G4nt?v
zhHX#%%F*=Rmv*(IW$-|Kw!sx60En@6@?*(Jh~8_6izKZf)cG~r^=(x5o5w(8gS`-C
zORMPl64cBd(7y&V0u!NolO1hfcRpdDbbd8&+m41<IXUcs=M>aR^F^KT2|whf^&$Xg
z!K0M9NnPPKUV$@*0-OU!5;{Pw%3>=$%Ak0n-$crpjLtrXcY9qg8!!JX;>i=x0TJa{
zGw7pgpcLUTvF+*7(zlIZx}c=LOw}*i;OB2Wxm3b<?_l)!1l$03P{D;Kdlt~NVCi?n
z?hx?D5i;Ld23RW#EPD9rD#BT1u=kg^a0bj%cM@8UEg1)J0k<OtpDhtVM*-hTX%dQ0
z>LigOGo{yNXX=0-({I0e$=^turXZQ53Sv<uctXr$8A%=b=#6wPJ-lSvPp`{tL!L@m
zy4H%r-%C%G?g-8b-4P7dCJBhI69&3N_T@>2cvuA=!0<x9v%)BS9OKfS{H~w+BYIWz
z5y89%L%)5NPP~Ern)H&EFdy^5$e06kBpQ~n^Yp9TY|0SPb)ayAqvruSX@J0m6g?DA
zf^WP70Hq4=%M@xQs(|R-MQ{ZW_#KH)SB;??&h^`m_yUeSgU4q|HORYJYd}<KPF)OU
z6?!a#%d{4h8#J)#&x^wO*wdJ#d`5=_^3fC~U*Z~1dWt_i@i!rHN8~JaMkkCuld##j
zqi@UY)-y*A*=3$|Sk&v8N#Q8a4k|1=k)4+_hO*C1rt}2x!nr1&l`sc4v$0XztDrDM
zFby*45lH0|eRGHg1!JyqaZDiWSK*|-69$|&N}Gr|3%qE9IIeE`G1|q1PRv82_z0ab
z`ocI;Ca|6+&OsjIW-0HBv}`LsRFyWp1^Yxbq+nSWM)6JI#7;uAYMit<{M5Ig3H~_A
zOtBvAX9IwgoxrCP>zzcYTt&d%4Zf6JqXRwTDTlR;6b*wQ-}W|A1P2-Yi;<!itPNr!
z53H>CD1_4bZ<Cd!`7FGatKhJx=0wyDoe~SVh`-0vch|-S7Tvl@IP+PM26?)O^G^<`
zc+l}G;WtLs>0emvK9eTrtYlU^b9-8oaX(tQ3q@$y%;NAZF1x3rPgHNb+c0lHOJt}1
zAA4`&j^+BjjX%wzBoQHH42d)tA|*qHWK1%ap^_={P=*FFgfd4mB~*yaQ-ctSh{{mP
z5M@Y&%)j%}-k<N@@An^gfA8^mj$`lLmgjly`?{}dt#h60Tx-4BooAT4kL&FXPIBh0
znKJ5KaRQp|RQ$NuC+W7<zbIRhpz-qZwbvTjb;(d~O9M0_L0XIRgkK<^cJZk<H4VV>
zO;yLXTpW?H#%T(-iS9#jjZU5VyYYy9%3r=;*ldnG9;1;y;eLR{o1Gg47V{y_-4}gT
zKkEK78)BEtW0R_m2fm>ojJwNOAgjFk!`}4Fyr`_*kRCqF{sG8%uEfubWBwkq!Qo?E
z!am&Vy|mnrE|4a{(mU5eEX&mJcUxBDT8V=<NnwLUyh7WE|IzPR+)7>0E@kdq0L-j!
zOBP8-qaA<hF~Sm&!QPgMdz`e29X{ak6OoG8cMg%UY}oHY5D?p6{Q&$XSY)s9R?7X?
zS@7u8W1-vxyd-T@mbG>TC?h_Z@vAYgXiHm#y+=~8CK~_~86ZToaRRRnptK@~_U38|
zWTWc~v>3-Fsxp<@>XwM^AJwGu(qcq5T1=_+U|AkEUNtt)81)#Mld%)o){bz$7@}F#
zVa`v1U;GKhdU>R>_9VPAFA)=jyAdmf#C}Rzg+_}Dq>yGlZb&WL8OfI03}!+AJev6G
zoeOroO}K15Mcs+Wa3S(K7VU)u#tTq{mWguBxsD>Qm#ku=1E4><BmAC?=3#yVkE~>Z
zpBTSl#)5zl&T5SC8j4eB;buF@h-00mgE4o@*iOFX5?=9B>=?(jyPoZPjCdZ^0A)T-
z!gLob2joFu%*YLd3;7ihWvFzzvOB@=upswd&5#V)2ZX`r(&d~QaFk3Tb>Sa#hWQXj
zdd4zcu98T^bJg6CtJ_B+I~ty)3B|~Pcm)8#r@SH~NQmk`Pn+{x6STsp;_Cv)$xLZ#
zR{hTB^Ps^AiRpVUa3J+h%@er`>|4gv(k1F={M*H>vBi*X`mk5&g2dB%Dxj!UXI0H>
zIMNqbdW}{j43p-yQjlQ&#KzIlodEN!02FdAQM);O?UhbsYlW<ZMa=TQNDN|opmYll
zMN5n*RLSWD^$7JnVcg1Gmp+0cBMpJ5#|mJE^!V0N_9kK?GNWfZ=OKf#)qId_5LjaV
zDSrM}YCl>K?ubbH^8hlF!y_Yg`V)6@k{~%xorUs!m)*=gm5!swn){ebcrAoieQ>N%
zX?65}l1>ffQ~X`4giix-&%dym`GGP!$n5l5bs41YGApqttJuyTB!SWF0`Lo|owUrB
zb5cX#uSvgM6q;q?R`8?Bw=no0fGW<FFjJZzfu7FJ4?MVocnYc6k&500P)j?V5hLme
zYJ9vKtXVj1aJUsYQWFSd?hlzz)Ybgj#Z<eT?<f6deO0l(O0%lRUSfSQ{8il+xtE9e
zf$DwM2==o9LEbrwe9wh~(UpE9C2b55^*xRZs%8!9)ZnF*8@CGM?$enrqi6^Hj=g_F
zR&znAHuE%^4qT6)8$7dqeIqJN^@P?swGh_=^NpW?J;hHb2Zz=(X{3>`xA)r^#4XJx
z78s1a%}&yXzF--pM{QPBq%fG0AkJ-m{4t(zIc{_R$+I*R@4OfC4Crg5<GzDJNju%;
zOwD7nit;l4cdW_z%}Wm2?MWLVV<`WAHJkryrXlseM${%y{`;4d|LYgwO`5KxJKBaM
z6KP!AW(q|+5Xu74fl&erpe(aY+k?UGE;|g1bFpn703Pc*t-@aN!S9<JmIFX#3sWfL
z?_hZ~ay=MPR)BD^vW<~KX^sJ++eai|Z9<qz{^({bsBJ>L<OgCfa?3DKc?c9wPZs}|
zwQOac;osW||Mf0D^3U1J2bbXR{Lf!A{A)1t|M`~*8q(YIFdh#L!1O+v<GV!HLs=38
zmE7^@PSl&6#R&Hu;Zzjw?{C1Zv<V%GaKa$?5^)MeKE;p$_jU)tU{9G1LR+Q1kt;Cu
z^)e(E{9wT}PR)~Csd&!~*>j0<i`UblQo@IOwG3lvc3%BIR|tdkubI#*{x#8Ao$}wm
zr2KaguwDMU2><_E1aE~!Q<dPU3!MxQkaTTrt$n5>g2&vk9}t5@?{hk5R6SEE?71E!
z#6#^(i*w)ul!&2tbd*>XYbJQnv@PXOc~rx9KbR8-|0i1t)ZOCAMvREJL2e*^WxYtc
zjCB<LKkWc56ml#Q^~kPR_P|o%RdkjKwu$o>85Pga2;O`%CT#daUUIx$1R^IB0LN^w
zbvo@Li@!d)Xxpgv6FXYj3?KgicK-f2GlW#<yBDwU!`VNd$hl~1<&S-ob$ZvVp0N1t
zsU{)xbwz<%XtOg1u5a2D(I?@D2-(#*`4Jp_G!)_JkX_V218a{<78pvx{no_ZO|^mv
zO(V%0g2VC#!MRL2|Mivqi(mQY3y&MhTCgN-AaDMlYk`~dR@MKX)iNeD4-pTeFp|+a
z9@^7DkaTijWCQZm$a}m^8p$=r<|?_@;tB#X5$2M#5Q9@ZDMaXSCBH-w)oYVqLOJ=C
zm!B2axUrI5^DSK*`HQDjN67=SDptk=I#)tRp&Xpc#S^Ip8LU0xiQYS#5r6v=`Xo`C
zyn)hkc<~LPyx$f-OO!OhsP&{p>>|ca{@#0^%;GP^EZ$zrS3j;`zFN;Ujf-xgneoqc
zXye;7+kNos-?IQXMLuDCkAQ6_$d737CVLP8`lT#Ft~BYzH-zn0QftMv<q719ML{oG
z{F8hYrJ4WF$8jxQiVkRDSFrHo_pfp6BhS{mSYh!}-2Yt4eeh@5V0m>c+KFZxB7{H#
zPVzC{-xqg@=zap-yw4qPF)}0@4-KXH?SUZix*zYXf#;L^kF797m>WVTK`fb)e}ROB
z2!30%1<3Dve@`YpRg^z$aieONfRlUy+RZTC8d2z~zgKF}i(2U|Uf$lM99buX)A=#V
zi+8TRn;eFx_lOASZaxj}r`d}zub}0sfq?BZQMy47>7^e;d^yuJi@Q%<b@4q5*)E-S
z$09*y8DA^YKDcDXTE>(kB9q?A8qIsSb6@Q`Jj@uJYah79h1TE}@21QSZn<*-yF1eS
z^eg<A$>rCUdwKnSYg=1Xz3Du~xS_gxy22r2vS?kO!ly|m<4IQ2?Jl*vnAFUfWOB;d
zItc8Vu7*CwVH&w^B|jyvB+U>vjm1BRtzt6YfDSC{HL^9P(~hcc`KE*^IDcN+nj*sJ
zNiQe&>$Ljn%UhZEQ7yjNeRoe!PXbz>DPqEswW*28WwbgMmylpIHT&~E@^ASoe}BPq
zo#AlR#lNVmdi81znYUww?no6h>;CiG+NXE^{Y9UXtQNmSK*v7ha}U=(4AlGcN_%+9
zD4hH;ZE&}UBSlVG>cYet5fKqz+AHLzzn(sSzLb`h*7aAvM&iERyLY2$+zL=`yAPU?
zueocB1mUHOjEuwzzA<(wS8w#p+O%&C`Kr*TrF`W6zx53ct~q}Exa>6<o@iyd<<H*8
z(A8jsE+>MwGyOZsmn^;%mm#oyj^k(6zq7U!p|c5eg@0WI`GvG2NL?~uDD9ffD<cf_
zgVt<bLe}?xy)-PU<Xiu<1poTOzQ0#taVh@!ug&K*ME}<(kssWoE#!;+bAkWIzZF**
z{I&l7`VO)v%Y^=Hlz)E1;-8zA{Ix6o*LQvM4$S$_Gx^syIX?aG?fm!Y{J*x@cztQT
zvCS2=&{HVQutXFop%qL_OyS7#ZQQu=WlfFD>C?QJ=R&t-O9=Xtc43UMUW4iB)4oWt
z`n|)v2TqK29{3rkY!Rf-gdD=<$B!Sk6{AJFph4lIXf95HvW|}*p>bH(IE_)wV4M}K
z4N>8RcC!3!E|BWN#=euN5|hEaHgz;a#C*36`T6;L$1~Zk2L;`XjPylYvRTBd;x5B%
zXcN6=v%sGt%eY%#U%xp+p9xio8`(B*K4oFC+SJq(y7%R1&4W*^!kksRxoM}kc^knD
zuf8W1MTZRSCFHcD^ZV#LnaB~R`eTIzc-Yw3MnJC<!rm1zIh;yHnz4ey%A<Fr{E-sj
z*tLruIZ92b)LXYU-9xfc?+HW`^hJe5xm6i~fIEp`-p>sXO|V;H3Vb|fYevO&LqS4j
zGU)tt`qs1p*z1l}NolDRY^m%6iCUJF___hy&yARv00@H&o_KCX>oP1i|7EH=Ar}{y
z8V5xT3hGH3M1K#z2}AV7_Ak}dr^jf!1f&Q@e*L<dn!4@w?b~KE=r{WXXclzYUbJ(%
zjs}!LeQ`TaEq(RsmE6DznS(|~?Dh^0H)HRf@dSe1N>P%N(T)uSNr+aBTX}d^pwT#}
z?v<EmxA#ueABz{Rm97o>&R)_(3OU6kU}QC$vl0?0m@x!V*Q!UDs)GT5Od(qi=t$XG
z%t;zY8Jx1Xt0GkW_U#5_8jtU3K<4}^Nf%;Ji0|5;NYC({V`XEz4o@1?$7gvvE=~km
zte_dRuHCwS|7HxORy=a#1#+)WZG~S|Q-&nuqJD5P-qTE1$DO-gxKQ!#oldTywl)JN
z7gt`4Y>mXDgoMxLHhaXxn8<sTlw|DfcL4=o!58%O^;zycfNc_?`o}Zy=mEBhVB{xB
zw}l2)y$IiaoEB+Ld5BL?>?TN7AA7qBO6>c`mfotpgXu?~AtKUgo&fOo3kYaBdy$5Q
z1`<eWWCp{nW8>mp%cJ8NZ+u|*?b~eV8b%G#c<j4C9sdAQH?-4vP#5wd+3s>)US44)
zyM}0yW$2>y0Tnz7rewU|&!D3w_>QaCa2x^x)JTt31JgUQg#20lJG(_i6`)dq97zuK
zy87Mo<ZvE)un~vT2g}d7ZQCl0eA9)xXkz%I{_^!&7Tq{cMSj1Y<bG5q5eUjk!jkf#
zs`{E_C1o6*@yp@n*ym@<kda%4<mGd$^Y|`a4UKf23>FNIHQIRi@ZnZzv<Ymn(8$m1
z%<JedKr_`9cz%@ts#wWO7-|+*#SLK9^!nwSC3N^vIsrxyHg%!ezW;4%hz?CoOF9e)
zD(ZqTI#yY@yD^prGYGq-p%75d-5ML<i{pdlnRM6&xMPxa_oemkXT*Mae@<p%0L?M4
zBco=5QfqfV<=wzKxQh)?5BT8OaAKK|6Be8h|FcB3^k`>zTik9b8cpl%c&Sqj<b^DK
z1+tanyV%Q4prB{zh7GSLE<V7rIV@kPlcLOs(k=!h?w$eTs^foUugPh@#LhpDaSzBg
zuEMuqP{n?R9WRAeA5m4s5YS7|`<h~pC@LzF3EN*je?Hm&2H*ek;2$ia%qA%*i2<xM
zK*4X?O{_RQ`rGohV{h;Mp8WcdjR4EU#Dp#an0J|mG+bQs{Bq`RQAiPpBmwiDl^E+X
z*ip0r)v~EoeMkNvl%>TkKY~FtC%~Zi()#rCz!F1WAfGG^wtTI&z_Io6axv3LDxLq?
zB~Mz<y-B1id*0c(g>+3#tauZMsf{O5JywXWT*%#<_MazLODUwkg#Ritz+z&Y>N;s@
zX?%DZ-G&VtP}d@7H-o$n1#KMop8F}XqO9vV>i(<%2m@Y0O{JcCbhGTWC73e?+r1L!
zXtyKZADop^h2S>`jqaW!r;ZC@OsZMQASx>Q8)|?+A?BulxVX)j4Q>k~{T_cEVd3oT
zOtpITGbA+kde&0N`LvmRt%eWd@y*T_fto8AN_K(aup2kLF<h)0<svZSzC}e6gF{1K
zdU~!xy$}^s9joz;m=4&{WfBq+Io3rg&NW2WY1LW28A1^eM?Ujf%oMqC>lVKrTKXGY
zZw$A=>@RWmUwp98*C0B8E!hZR(6QXGo9XHNRUunKu*sQZ?>51lLjAG?sybS1{k@Lj
ze{PgV4|`xe8{0$IGhb$BFW#2+V1ZP;64&_i*Eg*Xt*y^KePSOT9v&GP`Fv<+(*v@c
zu=nf;l(x0!qYqS?dg37wIr$iL_pCsL|57}9*=w5jFdlR*k|G;24mf;0BBCJ6QwbT7
zI;QtJ{`j<cYHG@t_8J~rz>)>(Z8&)JrVufr>_2GHjX-R>1hNr1E31cQ?Evr<)zzzT
z&JN;qpEP^1%IAg+ZsrZzb>bIga^gfo#<+XYUj7m)8=s8E;2;_p7<?WY3W7zc35Lc)
zN~reTV}$u5(cic7^RI@K9%h^MJft)Xc%4HR2vcML6`aMmuoI%AqqhnRtEEO0e*=Lu
z#dSWHf}@}fWeO54S4&Dtl0+Upd<Zjf37NAmIF$P;WfvDyTc@?Xy&H0?YjAf~?NE_1
zK}J`F-#~+Hj<))yCVD)YN{A%`unXBR5aHdKm%n_ck$T&S8AkF$Y5mY0Fd^g>UdVuV
zfDQpu-nT$W6S_YNW8{-SIirN|K)}9cu`1Y9I!sqY7(<$YcA@Pkv5PWa)o`An@i3Zi
zuc}&!U3uQgNgi5{Q>LcLXK%xCF<?esHG~v_0@^s20y2M{_T>nkj$k}l{zt6hmmfc}
zTn}GDn8hJ3&Wt|3=bfFa$mz!5xqg<+dILj4NJf{DC4m*#xp%KT927cQ5=io*s_MSs
z_2b%DqR+j(K6nPt8XMPRekuC)3Sm}^pxdd#tG8|2R#;T@tfpo)#%>)dvhO`<V?&E0
zYj*lHCp;#$bm93gOEDTOVBEuUI}EL}#spy$E^gy=aB#p|6Y;s{&!6`$E88z9D2R6l
z1Ag31P37X^T87%%LsnLTIJg0O?sPfM{<dpeUVuXg?mj?1aHEpZ#Wdq5(m-x@&^!k9
z6+C)$5E?4=486S{+S`*4`B1w~|Ki8pU=W)hoY)trK3PH@T)la7J@(=%z!&^l1x^Ka
zF@a(@Dr5!+=CnOl{3?W*)<-^n`{s}Sha}2&9qn07B34MmxG>1B?`Kd@5a8izF)=ai
zOoL5u?}c6q(q!{uK;8E*n`h&(=%--bbTLtFP-qsJYX$6AJ4nB@vy2RL(R6Y%xpGwX
z#F@0`T5Ldxc4B`0vu8`t=2`aOK?=4$cXc+-A`A-oA{jczfIlH$nqUluGAE_Zqqu1~
z>ajH(Hw<HYp8q{^icR_=GerDVXgzWr$ZmUiO@10#3+V37J33y)B3oHm-2h(g=`q7p
zF)}gH6{jb=($viCD%h)JI5KPz1!rf`;m#)^Ag#W@>g)?LBs)0MkQGx;2nIfBX-R{}
z7k%s27dUD3m_Ded$AZR60od0ZnAWQH<9<Ksr;6r=Te#!>Fs1ei5#d>|0U?lWjC3U2
zlsA_3S#2%#>C>mx)3wQ`xV!ITm%iwS0f`+Lb*b3RZv$3M2Hl9;_F<h#-_Z$wB6EA|
zu944?Wfz|7;Fm~<(X$<eB4~<CjY5AZ_g_>HsXWKskJ}Gom5~QkDJ(2}+R?$bj$P_0
zJi&RGD9#zHAF#9d(x_gTaMI>sO+McY{uuh6;}a%WLX*paA-XwgMa<w40AK>q@ouBc
zUDyZh&_4lY1QoKqo3U%T+dE*IVG-mpR943U*S-ZSM}$I*hM(UQK;(U@=jBakw5|*a
zm1EAC_{HBlp^v*nKCyLaKF%X6+^HNcyZKXNd_;uLqenZhqZ9kj;d}6ckT(lsr(ni>
zFhI=q;6B{ihXUIzXjD#27f?{8N1zrg1DXnI0p%%Mm%$sMq(qzBLHI;YPj9x9lT*vt
z$w{b`Pr;JIyWKR*ZO6<j5WOn7(wv-2(PTK(R99E`B%oJP;9DAMYB~mnyJOE6-uL$P
znW7vWzF{e8sM<OKjkabfUpr_X_~1im?@7KjIyN>ipK<vTH19{(3du*9^_kJ2pH+=H
zit41p<)gg32&)S}KJ5oSSp`{9s7I2wG{!@+-??)qX&mFAH7-1>2R1?t4BZeDD{COS
z!BQG*e=#5M#PjT5vTD_}L)Hiq%*?pU7v?X#c=6)wEIaJdPKW`2{pd6v`Vjz^H`w`P
z3v8lS(RNpS2hn=^2Tvg=3!K0G8rB#$$rASi%KV+EM<Z?`#}G8NlTrANUPOM7^$DRR
z#I}b9Mf7k{H>^6*F&TP9CIBv>oM^Z>yp*{xi5sFHwfv75U_(H8##cT*zNgT?`5#!A
z-2h<t?%Wzf@qPQcP;V=P2da+oUp1AMzyN;1!Bx=5(b3b7*4jLsHnX%0ga~g5R#Ef9
z<Ls?e0PxQdAldc5;gC0dm!`1{TjJuaCl2=Znrv!_j*^m-voR-(90JE@8Hp$K^pc<m
zB(dAs*!5fr3^g6_g|Z;UJ`WGm;<VJvb_`%h8CYQ2t5>gnMu-q<4Zj<JgumZA47cIJ
zSZ-0<A2^!ubH=&S`}VN_>@;1n<PZ`94eR(6iiwu54Mt?!Ds;1qvK&^Dx>j2M@uT6c
z0~^`dg^Iu2;B{+3e;5<&Z6bVaX1@k?<*Qh+b!1rCcZGET5Cs3=E)&}Z)?xl)HKqxQ
zK-3%6mk#LYZ`54^lZkFLOf?dK*+JN!U$DbjsxA3ok;NVR>2Mua@y|&d`Or?W&r^Mc
zz$x?>h63>{e#vbWSV1PVQ%)kvIK`h_+0>-@^3|*GS;HGqQQP()<D%Q0lbwC@x&<t1
zKvg>Cw$rh*XTJUM4C2f0`-kBIsajiG15EHH1~UKb>^z1=WONRARaR9|LqDXs=M3h3
z^pmOap&neEoZQ`o_PqgZ?}?CZ-|2X1kLC$<?q|Ya6!|`6c|tw1rf<Z?Zin2{aOdvb
z8{$$Ce;-1#iGF4=+fNJtl-o89KupEmjlGunF1`+bHVYwBkbH24YwVy>OzS`MkjVPs
zWO%CS!!SG({Dk^WF|qgG4VtkZIWXSrLD32sIyn5R8289`rvW26>`Tcroj_N-a<fhp
zb9{6BMCJkA2n|i@_vrijl^UHneH5yTp>KK)^T!Y*Om+E@NhS3&3=)8G3pbMlTFQ{V
zklxkb&&P@m=kbH$_D{U#=H?iybMbqv(U@{8#Jp+Kw>j5G<Q3kLgL;A1&UJT7Fcp<~
z_VH=*1RBeaU=-4X!4`IQqd_~18JBslK#&|ez2+|xFQGRVjON&MXlRk8(V=YbV`OCX
z9D)>vN_BPht+SYo958<e6^VH|KgOKUr6pjDrMSl))?|C@aX~S6j5KNN#tp3HIB>w&
z`U%vWD)sNNA~loln8s5xuZ$_{)JKmV4Rx@_R7d;kXs@A=^W?#U2cRwl+74jmWg1Fg
z+sesQO^<cpZgLK=qD@copg3tqzB#oJ-o)3zulz}%a(w~*`jD<}bk`(1BV$Vcn1{yo
z>({TMD|TW+!okMh*bc%^9%~*LfI-pom=LZ73C+E!8i7YHyZ$g6^M(z{t`E`k>hP#|
z>6;Q4+qmFW2+L^#zXWzT$+&l0Ym{fO<D_Y<M>Q1J!;B1`cB@aNo3^Rl5xnR33r%E1
z{rvr9pdQU*Pe$H^90LS=83lzT1c0I01a}$Gh9HRxh>na7&|hPZ(L!))c3>cw+HZds
zL=ooD0DiToKXVl#`g6v{k58|~po;O0vH3tX#A;$Fz%#g6>sSDXE);BT*qjVE1(|jB
z*)lYh?+F$MYhM8)7qKeC2Rs=Z-0_^Cj?AgZ$VgoXVo?~C(O9prU=F|r>)DMmEI9U-
zv>(Xz5g-Bv=<HrG%0+1Z$?YH!m&=~ajp_qtL{LH{D!*w!IzSAv452ndo^5m5dQc|h
z`BsXoWtU<`Bi6>V)$SfQ-Q`ATz~2b@D$c7NJm}qTb3mA+Cz}@p>9L0E*M9@b`p69e
z>Bg(K;o6y*nJBTR0=2+36D}tb92`tVOIro21Kuq?st{uHKy10^t*s|CE<`B%;C!A!
zaeq?L9*+l5FQd>`w`Xd%1Yq(xmrbNkBxcpMOl-m|%{GG?q!<`}h#NISdwLn|-;)4d
zlB_Xszz+pse(x3l=UpLKZ+V9i@23!e!ed@;bC;nTLGcQgw6xLVr+Xbq|8!CNDGgbp
zZoc$BI$7Y}r*Ip~Fhr-)0}Js8GXdnSgRWmE<O3E%Y535*v5=6EALayftesaSQ;X%*
z%J}Z5KE&N?)I3|H`pwQOvqu#{n3=Rrt^pb`UvP%4ckj{PyLV5w@H__!kyzuZM<*uU
zeHTd;h@i&NsYV7c&0&r5t(X`Su$(v<=dxc<6$6gkS1pf?jeVec2gTm>uoXhTinmfF
z(Uh+3h2MXUEiN|zrxt*dBb!?eOZ6L3K6ROmt!)U#;x+bID=Sl%&yK2q?6Y9c^$q8}
zQ8^6{u@O$LNktH!R|PbA>dcwcG!GzLBxg<n+l_VLIl-VG1U7MEucO1&5dDzFtXp|u
z^wxs2ZT0oEgn#I2`1*t_KB&gC9cstzL;QBY^`cTgkUypm+#!n3v|F7qs-co@?c(0h
z_(sCi*4S(e_95)dMlHGNQ@B)Yfg}{}xo^rsfQ&SK(9t4x3PCv7nwc$KufPjYBNljX
z;i;+$TkiPN!A?<G`8gQ^_b$CX`gv`wvZF>9B#GlJv6<*}xr0m{`E^EM$BwlKxegya
z8r<ftqN0K^%Iw(WK4d7pB#Ic69fx5<fFI8?6|QugD`zGyb^kTyTHtOC40jHiaRP~n
zkt-=yIFK{6;B;GDFIrk!l8+ydA2U}ydzKF<fTtZ1<o8}ld3+S87m_!;`K*Sg1%Zxv
z7&DJ5K+XD%8}CiVF+r(xcdWiq*~MGqaEj(BpzcRGIhq}a*;b+d{!@G}3QFurq`Lb|
z{R_+tquaRA_v+Q<HS-8J(M;|3JCT9=;8W3?DXgs+{*pntIB+Pm3{IG{$TboS2<9Po
zgI|$XSEmC!yoBPt^`N@M-A1`b$HsVORxMxtw6an@w;amOLJST*JBPDN%|mzmp)QBS
zX(+9Ds+*Z$%36IMU`R)j6N0o#w57WDKKk7I%ph=|HM|R8H3G42W%bq;7XARn87{M;
zckwajJf8KSjfD&l4|ltuQK@dKL$Z)ZUdphq-@-lsY%s;x7z<b(juP4ZE)5)u<kj?n
z?>56;A3?nYaL~_bq-%rjWW|?dyvO9<0bfAtfjzzm?fQo+m5Dn`-WId^vN!tKcOlS!
z0Us^`cScYRWd2VPnUTn|W*)~1GT2@HhcvS*n*rI0`q{`W1EA)dt-G44Y9-*sIJ-U}
z>k%L#;5R`50VEN;sH?dGvUV%Dd$l`KHwQa02kav1>XOC@Brp2Ls~U3>RnS?b?+~TW
zq6TckkOPG6CUq{`r%o`|g)G7{>1)gZ26yxWnm6oN@)><@E1|B=Wq)E)89<Oju@|bN
zDyeJSBa~ODsWq4cz&`k`NgT0`#r~12eD(UZYR){+A>H!C=%mIFtUxmCOeh(IOAI+!
z08U9sRvQ}|DC73mMhOeIRe$)vEF&Xx_QRt@qhKW*dFLOWg20tlAeKY>je}SU?imu=
z!$2g_L@fCX?@vrUBHT;Z)af2CdT9^{qQC{6ehm2CW$psV5~z;^PI#HfetqN!P$0VS
ztgUT2;yo}cwU?KdReSzg3L!)r<x8hKR^wo;L^v(0tIG^>fxEr9u#mKng-|%K%^jgB
zSk&A%l~<TFwX{&>V*DB2P@`lZ@XD2CsM{I0zjPLn3(uA!G<!86^Kq&k3Pb90!{hEC
z3OHNJjq=VNlV>_im^c`VE)cAIDBpsGMl+>lHPWcjy@UbV{X#u>@K}kQ>@#L&Cqae#
z1qNOLS%*o~A^4i~x!Xo*nkm%0N20wkrm(iWAXH%cb{g>ARlwF7m>~0eYN|#kU>@Cc
zMgYPeWIWJH4Nm&_26_2xjD^Cq>!)B`2=n~S#7{^7qormFkIN3f@<NP&E*)CA<p2mf
zMMWdaEUx0A`uX*%ZjEAH;f-Qm23hXuF%)|6x9M_<if+`<&`?xSktMR6Ohd`mZC_Di
zzz#nL&Z4~D?rTraNgT#P)UtW^(QLqpqe1GvQ5r}idY5^C*}Mwfa>4?`D?dXml@HiS
zm8wQeSVI)G5~eD%%^f^<u66rH&|}n}BIeQmaV5CPV<mngh=bhB`%JS9Z=f~!aVa!m
z7yc>bIzmAzf)&F+?M8c)YNU$t^7aDi1p<FJib9I-eyj7?Ha<S8PoF-a#O86^G-2?O
z_jnHF3()>0B+UZn*Oa}?`}*~3Kz*t(xGO^Lo5h0GXpEg+>lT!pds}u*FK8YTDPv!e
z2S+Bbk1Acl7WD=&>{kH8edMO0c3f1{H58$KMwt=a@+J7_YnW$hCN!<;1Z3GNN;p#N
zT#aAP5N+R-8SzrYu`yG}lLz-7Y8}p9jdte#@N1TPTmY_aa_<KdL82_h{Gy9fyGSKx
z)anH$6vX*q?T*>6YDO*XI*9Ho!Sp!=I3PG-M$f0jyAMJ_*5%uLq=r3gEJp;uWEj2~
z_86`Z!&j6v*jB7q@uAdhHxg!#5C`U2gLrrYZsW=j^)HicQP`tQ_i3o8TyN!~T9OK>
z0h+ubB+SUvu_)cPf<?C{V1t1i`w4N25v0|m-X|Y;)0(>)gFVD~kKXn}$<F<Z$+0n6
zRKz_(p-oL?ABd|eY|bokXKMm&aFEHKuxf{^d*I=LI;9-{eyrq4jQ)LDS9hUz9K{yP
z(aP%@tmxTJ5KsO>wZrc+Vh4pp6B83jO>|<QJ2s8bAz4}Z1C|ubzTAnf`uYOGQn#ht
zNHJvvfQV7KDAH3_AB#zi9mCLjkPMCcctVcEp@LxPvSr3aCr=}>gYRXS8QzCyLGnc?
zgy(7diY`r5YuE0mN;7i$`f5{+d@D=^D6Pnk8h?Zt{i+E){yNI@zCYe)#rJdud`F2W
z6<AG{ag9UgmI#i_DvZH9a7szdM!Z)ASU&Z9dCg}19fE>8F<z!IwT0gagqb63&l}%W
zu!<yI)}uNJ*m2&@?is8Fb;}$$hcx{HAtB@PhvJc#D)W1Kx(gm4N$gvJ#oMZrfjO7@
zSdXUZhkr~}+k!;f5>P@T6B9bIZv_Bo5CM<Vlln)y!4D#k`0$-u=&9L&3hT8bhfLzs
zd{Q)j%G{i%=<F=|?<*WRa=nf97q0b@&BuLck3dIAGpwV0HQMz&>E~TPjp@&-AP&}p
zqYbs%7RkEz*WTc!^RyyYFJE4|eED*QX=wsIkV^^`3rE66dnm~YM2F*0+cEo3fVWSP
zyTOL>>K#Xn@C8bygJ81vxH6d=8yK*`&+^zFee-K_FF;qsuB96MXj`{p35CQwDw_n~
zy?dAQWEK1zmt-&ki>fEa;H|i;a}k&iVw1V&%|SZN(JGADL|DM}aqkh7f+~q*Glhyv
zuX+d7Ws>Z8_VT5o<8>UEx?l~YddlVnKeV+e*E2CQ^R*rS?pj%0ty0elPcLB$kh6kv
zh}?ZUD0gELh=aCG$e+h<_mJoWRWDavQ(G&L^P>&%w_M-r`udGHQrt5Sv;|Hf>Pj*N
z`Amaoj3Gl!MP)-rM+b)LOcs&hI8<xaTuj066D=@k>ygFfYMt~!HqyWRJ#3xQ1WG;c
zn9X7%<cvK*f2(sdzP`S-<t|9~Hx?nk$BWNZ2^K+-NaK8Yd3mS@x}j}tb3tWWsE1|C
z{giPxGGpR07|XVK0OCV{k!L%>g*Tc`jC_%;aq}WPNy9Yy8$<?uMAR?bX&Rx-gKV#|
zeI|0`QGGjjcy{mBvX_Fh4C+SSyI@e{=g7#LIuR0_pW>e~2XB${gh|l_&}HZEGp{lK
z^LditEnM}W+o$Kx4>{jXOXCaj@_IQpm-f1I3WhK$?)dNzvvjG*tyFkumlHJ8<!0)7
zb;3eJm#<p2s&O9i_90A<>tRN&zh$Q35H@1}MGU=R7)n(UIs@1m>hWZL+H#3EV7>rY
zQDbYuR8e`m9^H=P)q4?@t~YLHfytL2-5*xhf`%@7&TGILafpc>cf39~<1MgN>TfN;
zBn|9e(l|DS7Mm?te(r9@t9|cPjT%D00jVQGVgO(<o1wY9m82qvW_?SmG@2SXG~crZ
z?8*|Lx;%mwQI>OeDl@Z4HZi9Y^SI<4iYZq??^6-z#m2R7pI+*ZA0(KUC=a#tGm-xj
z%eNFlPH=<G1%cAO%^RRL*|@PHk_2aW`o0Af_((45_P&v+oMv-?(9Oe1qlZA(!J6_F
ztE?Egc!mLwY3i&3KQ8)^jLfC8b1*LiSrKA1$r^*t7=QdWv9(=?d;|eaNLxJ0&-YD9
z>D?aE9OhoiP62z(7xN&^pwx9e+JULhe0YI5@N819>O<XZV4(s4CYVkbz|0}`UJ0gH
z11L#XU%#TQt@uqQGl(qu%@rV5iMu)wb$M$7;5*T&AOlR(5`|gb_F`Q~*KcfdQ)B}A
zISk|C#3F0;ei*4W<zdb-WDg1Q3}d?hN`N>D6vcA1RaIA~zB#(9s;Vj{<m+t+IRN<O
z2Y{*qG3H}6scgKhZ!F2c7zLk>?U|OL0_6mPHF>XkI0lXKi&KZ@K6Z3)yKvXl)m0#c
z%iSF#<+cm<ZClZXAnE9s7*1~P<zQmzti$5s)?!n4pv+#;^d`Etv@hg5dc^0exQKcD
z(h6&O6&9xYw!kkd=C59}<^={q$}c!OBH?lE+A0%M)4Tu?fC*&H=`%=D4Urp9{PVBK
z?En>?XuPPSqqC?fLdb!9TluTbvhwZuWo2dk%)a$OKo|hhjOSdoA~PY5BlEehFEu}_
zIc(f9MKk?#SJzVV+=`3KrSh>ekjD^iZ+Un^%bty=dJ?1v<lX`(>Q2u$5=dp*u;B`V
z*O&E){@_GqQs=Rr8$3eeNhSl_6Gun62P?9r3>m0H*`g54#;<3KP<g@$C!)p;-L-v?
zhdiD;kFX6BTX;YU?U1OCRS-H*+X3aIM(ZCPqnSEzXqYuuq@X}L$N=6GTQF%?lEmG>
zW|x5M(=)Qbb-=E1hn-rKcK|KE%!jcKGYE%Ln~RjmCygUjPYbGU9hgDVNS~#8YF~OF
zeWf`t_PMYD@>u3N-e>5iPF6pSiH=Ty0sIXL?@M?0GUQ<5s&fkpqK@2Mb`aWI@b?7u
z`%5byi-Yww!3i$*O*;<o`5S(+<z0=(HRFp1dawHWDht)>h^c{)Ib~_Ny$uszSCH+#
zr=U5>s%TQTsRvI2do$FUm|nLr86Rp!>K(O_=hjI*d5Q7Be674l38P8Jp~Z;s+jC@1
z0XRY(;6p#7+EG?pTl;2ZqBh=U)^28Q&M3=`4TBdCu?6Ps$dG>lh^nLu8vb>YRTV)S
zL}X-ubB$ZBp*nv)-S`tceQqZ6x?TpP`bADYr2xJs0-oitYQE%OUDQbK4uzLKHb<}}
zstH~2l?L;5FT;10RX62Z>ywli$?f*6omA%EUk>Vxqy+cYE;BbX`#jcg;IOlEe*Twl
zhOm~TBbc7;OG~ztFKi}Mh(8t3f%)tlkP-JxVDCUry*UD|C=w!D8-&lX6wUHEB;#Jp
zV{N?V7J0EWaFzDQ=_v?6lttu|C1He$Dr`JGrOBjT$)hktR+2E{LEX!|kuw=ciTf~<
zo16OriT}&kgCr5;e&WVj3fR&2b>hWT>((8Q?a0l^IRv=cGI4@Te>TOO)@QiqN2g!2
zU?skp%ZMh-{$}6jNQL!0E@gwvEh$K~)|Kx9eyk>ka$xi3%{(*lk}gLZrR<eFoH|bx
z6=6hU*4b+pt-lJ@05A!bl=yWVa@!F0e3H9v^EFJPocg@gtJLXBf4?6JHkFMAF&w!H
zP7oZTy4}s`{l71l+v@y@?FGnSKeP7md$VaAV*<+9FI@-#a)wBf;MSF!)9o(ER1gsv
z;3f?YRUDJqG9E6m81NWK+}zLD&>npfxh}l}LmF{-V>hlCenk$E(Rj{QR&~#d$Uvbi
zbOi3W(sbRXUCUA9`d)NT8Mq!<9A!Jep1g72@<qSnYY#rQ#@*UJ+1qn8S@}Sin7@H%
zcoEdzNjX7_+TEA19bMc5Db6uFECvS$xn1b$kpZ=_$xO;dj*5FALeO1nn}oy(;i2Xs
zb`x`R9~`1n=#<1Vbob#4uPtI0BuS!MezB-)0>^|-)y6UzT<C`)hiy39zeK;%#ISXu
zs%?8)#uQ@u-)I4$@+xy3fG!#*Z9(1Lb|@zN^226q_~&RX#x}ou-+^QZNrhyH0W*C}
zxyM}$#>OfZ(y^fGhzs-;pu`x1xF%0n*1dkc76+d6Gn8*fqQ47D1_fjfZ@x%ao9=q4
zFXT}fly@TXe`zUlV4jyG;G9&_`Fk?&%1V+$vKZdNyuBAE6W5kBZX)4>MhIjeTfiO8
z6#vj0tSjyuP*P6^@%#{DW^Enbr)#ak63YVS|8;AisYnRC>&fp!=g;pXgJ=y8y?p?<
zL4%g#5eA<v`z_gc=GaX5z_*g8;OMyrVi4!#HJ_)PyhC$0lqMFHlvGqzDdYr&UKAtH
z)#$tJHDq}>1r!kSL>jmbAwvy!q%v%EzPoBI0QgYEdlK5gEDY>8>hZ?x_>Cmx-0eX!
zHl9x1q{oN0A$OAkE(Y-l02fEQPFNs=J<|AKbJu8TX_W<~X8>;xekx-{TpKzx{He@x
zq{qa@Mu?>LaENA6^q3>*%a@q-LtUfZP<xCuAhbcX#%HH0|Nc+Y9Y4SM;5-qG8Y)TD
zYE)?deh3MwPkrwrl~Y2?#N8pcl4pAQM<noRIk^k#w=g>-xe{s(#y$w??&C%Spkr=!
zbY#)Uem%D5mT@feK9}L^z-4DyG}G<7UaTfUK}3ZS)~xjONl^xD6lPXdF8E>`^>=ly
z#R0*=3MM8H?159LWVl%fxdVbFHrR7<r|;qQ-JA9uu;@o#{-sA$?&~1s&1$OB14%DK
zB6V=aBCWN;)r=s}<E-djtT5PBh{*dS<`{)qFI~EnXlELy5pZnXxpO4qJ!!Eu7?^_=
zx;NtEg^R=FkSaTcs1KmHHM@C8ZqB%E_>~?gUiqOO=Nqa@N^ZTKNZn9TQQg=)iA=M&
z$Mha#K#f~9Dqg;%0%xzo5X-UObC&tZ!n`B_8m+?+Kf$!UZ!d)t4EMNx8atDgUCOVp
zP#pWZJe11%?Y*_=`?;*_MH5^j$?WBSL`$<Pq}~9Lu=}maCx7XcnB4b+CL=0O6_pAi
z>V=Xd$mQFOf1EEs2_U<T84WcxjqUQ;T?yRjS+9Mry>Bqh3X@z{p70~i2{K1I`0JwB
zDUjvR$*dxxV)r=x<0QUB5NoAHFQoV<^AW*-EEy5kz*KE(t7>RihwMkF*b`W&;y0Ec
zIrt{zTYX<W;Da}%K@id!>+|~6tCI*D^|FLg6Sxc9FisX$;ecOAi107RBG0+$L{RTJ
zrp#>sNaik2H!hc%P*|!;l(oobgjt0l$&Uzuz))C`tcz7>ASs7$`?XQ+{P{;oE4m>5
zs07ay!0kmkHyt`43&j4HHUy7hJ<t!))mD+<Gt`q-grR6QEgz3pOAA_%g#7yEW(EX{
zp=zjD1L-Xv`=OK{LLY$>#_H<c(6nsS+5-n?M)gN5yAlmyg&?DrvtM!w*YF6X#<;L-
zX0Erlw^D(vr~la644M+rpe&-29&zxEp-d*;eUf+2Ra8{S3J)2n={bLIR<dX6gC!x9
zD-uV3gtxEL%;sZty>lc=KfDa=c-ipsfgeAFPoNRCGMc6+?2$P~a!K;7;L<7p#ZH|*
zeJ^bek188B4<w~(U(vz<=B4`Bym%9ZShi*wbdOr2xcyMG$F<(C6C)$uIIODA;X91G
zi#p}(tM!MaW-m9zujV0{6Qt@DJ*$t#O9x?ic1@LW2VzsAS7>$in*DiLMD8z}>Q26C
z)K(qOl}NR-0+j;;LQ87_<ISoYd9#wPf`$C#Ib%%*!{-<x{le|i=CNPwdsi1{(FI4K
z03xA+P_(i1D9ox#WfI{%O%aPiCB$Go27ld+teJbPmXMe`Xb8d>A!<0L--%)LWgcFL
z41CdN?TAJ@0433sK<O=qfYs=$*C{mpP*hcI8Ho1&0$lscLGc6uXT?}kPa!G0uB$!F
z)31Ueok^wyLVWIaw*Bt8FM6F|wG^LRU$dDgE*l9Tb9o2$x*Dl^GZ$*1)eyP!pXki9
zJl-mT3)@yC`yK9RJrh%^ofYNt=g$<l^Hyi*bwbCXorN7h`C`7=3{j;Ey<ACwB(61D
zIRI#)CeG+_P7W;c!-6SD_jz3Qk~v|xAKoHUpYq@z1pnX^STt`#c6IN$<~Mb-l`_MU
zf|GL>Wrj;KZSH{CQZYILUDlO?0+*Yw=6K!bq$Y~N_SSg5BN8~`q39bR%ga7hv9e}G
zD=@nl$^#8KH%ub)7F9;I0T<o>d=mj_>l<Ua#l?R|Y^zO2n?5yC{vOTQbVu4g8D@-#
zD36XQ0&oXfv8p;GL%JC{<TW!SMne8kg&EALR1wV#-HK!m=FSG1#iHx}cc-K29S*>D
zJ(sNxdKEPirV-Kpn>01}YNDSkdu42}2JGmb+p@LvIDc7LS%b{VwwincpFAcF$tUN>
zg^J>wX&2iqmN-RF0dmU%mzu?*A0iWMK9A>58#A-JW^-sw1_;BcXA6^zE;fq6<OH<F
z*{6^lX`dQ}^7&e!T^Hwmo?E1yykHP(+<BN}mFd%IXov}BKj8VWpa@v{lo|KD@g!iV
ziu9c~ZEY+F0p3akV>5#oFQ?03nz{85**VhVz9`IH$sYV$O+y2c4TQgCs{lC#Yj?vc
zdD!44{6xt~@P4<FLe{PDhP3b$d<#p%9B(mT9|AZ&JzHdvNc3fYoq(fH5=PgRQ=Ws;
z@`FiNz7va><{F#=7nqdPsb52tw9`m&MOmYMS1@Bs>S7t5oVE@&?>eHTw<;l+A6U7P
z?&d^~JuO=PERml7ZCWFG>Cz<#CFnBL{zScb0qZDm-aJLSA0h<-SI1rb#>FNVpeW6t
zD90=|l(dxveX6AoE%h~!!{3~w)4O9sLph7Kvc6FpZbdMIAYkKMYk2MZ5N463=xyg;
zFT%h?Qps!kZsf?F<JwFCk6hT3`5W8zDX{DKW}Wo6h+m)y)OEfR0-pv7FgrR4-j`GM
z%cA=gSdZzO{=fru`C}s^Pq73$&(yPNV~|r$MsreS(l6);6{Q0nfs^=9;v$S$g<5I#
z5KIDS$`cNz4}|pNqat9fSs=I3h~0`ba}|y(_Y6j>Zh{)jvv+U0Va%nz!0(mKsF}e4
zyfU+eP~JUz7|>i&OP`D(DXZrxplE&nu#||`k)F-cS_7+Hfu3>l1JDeG;3VthmZPp0
zC9jc3amzAvW>1z8eH}=HlO`sbY4z}Zh@-*=F-*s;vAV|wqcU=AAps-#^t||Q$fgED
zG_1($g{mK<MtLQwW6cmag^t69km0;EADIcR09Snqjmh`Dwmo+)0M6|*-WX;z33N+Z
z_vA>XG9q)j4Z=KHBCIAcNc%h#5?}-Qr=R}eB9s%6so(@%pA!U?Dcr*~TxcTFnmJ-m
zSve1|V03)UnCmH=M&vad5*G<(w^6g}Y!aF|3s^YEwns?_*glk95F<lU(=xGX@IOdL
z><|`S2ZA=M^qz`MGxS2ce(2DXZHN%}+yLHr?`>alaP!X$#P-ls%MBn14gg;|z*Yf5
zS>9!E<MJgW=Kv5UHv@;sg7(@tV_hfzgqWI<Bn$SHb!c$#5gcot+Vh7XutDh~oPUtC
zCi}%MQBmK50x@K4V74xSCaIkuP0;;oe-I@?BVXxdQlItmAh?dzTefT=ee^<ypfw%A
zV8zdn8$889IY0f&f}4kj1Ik_q@SzEmhhrK+xi6zuPgXHR#h0&KS>#7WQ6(ya@R7*B
zK(KPa#DQst5N#RBKT6CaOCh)6Q%rx$1156X2%<*k4&R6fMvUSRDk%}B5XkFeq&1oL
zY$0a%8QS122ZAMChD%4?!9HYLyyXRZd~Rap8bK^7qDlpLFk`p<ARLgfZ5c2CgHnM6
zjJ!)7@=dfPeTO6RB=ksrPR?_ryv$rCZki$>gUhUd=FHjyDNqP~3A#aZGIk^^wA*c9
z1V2E-`pKo1pX7!Qf$N62S*Yzaev=s~527dqRhoc~j-;B4sIXgqk{*+K1x^DxXg|7|
zot-_&8r~@Y&D*}9pR@wV_CFOLOKfdz^Jpp;EP+9i#28Bb@%()uT@4&ow}y@vgeu3@
ztrQTJvfy9XHg9e$_qb3drQ?ccjx~!iizP}s+*0TPvF~ud3@Auw<dGun0ENBoA&&rg
zZy@9?iKi-jb3#tfXN89^)+BZC{K=pw$`M*;aySUlmdYr<^3yTa{DtWOVG;`BG!;7Z
z(ZaN^fxrfZ9z;8hwrxevoQ-JLtO+muQihrxVD4%>L^?)BAC#4$XgLwVZBC#LTr-39
zFh?rr)k5UXAPKTVOp~T90K>BidG*k#trVoH4<~j67}(@`EzEx?_mIRcpR6Sv4=aHT
zXnL7duAp;*myhT$CE@IQ7uCwjN+tGW$SPJKW^y(EnXjC32qAsT*POjA1-8aDv#*i(
zBPvO5UfwPwkIvcpJt`=uMs6SdL@gT)t5m3EjsF<gF03)Wv7~3iu7wNbrAU`*3ow#o
zFI>~huSpT3<Kv8mFO5+kSW}J%CXGO_<sBL}f6MhhibrL<sU6M@xypYi`CT89z!x!{
zUiBfk-4wrr+SCSx=Vd4;q<;Q<bY(b}Z#9#!kziz}Fq9Rc*&jZ9P@I6AqU>xZ)Ce@|
zwn#T$c;)8ib{R>_qjm$|VPYAfjHP}(@cp|0LVU#ulHowE%?CYuRug(LDJhAefpya+
z{^>i($;t9FBzeM~1<_EA8$>dh@8f@>Si-L!Bky?HNIE5_nzY|~4#~w;SYc9hu>n<K
z-0ginfQiPA!<b~Pn)3tZa6Ts^EDUuqH@3D-Kodr4EC=mEXAB@i(eCL>1`%9Yt+oN%
za)Y)&L{A5}D@JhkzoaK|1ee3I))aYoKoNowkKSiHK^3b^WE^`C`Kh5mxE2N{H-^=l
zP)#-E#bll1^uxvy?tN{D7rxK|X<MMxJkuR1H%?bKH^wQKNFTgX%pzg&7^i-)&|m$h
zOi_(Y(a4^gzfpGqBE@p$6#5BBh`B}7ZtaqvU{>y>c8Z9L+W`SvTAY}W(9!$_u!24S
zSJX(Wt%m{hK~W+KF?xIb%Z*X12?HN7I0Kc}z<8g&kk?y?67%M8I(A&Rj;O~6Y{Y61
zeH+mLRtTjus`(8x@SOBPMZovZ5tGWzRK0jXg-p9}JBUSqSN;pP(G7!>^>Idq_02nh
z1g&w$%ihftW9H17?rcyA(Duy3djlHe+mI)FH`S4lf#;=IzukqlISN_<8c_1`({bjD
zNuTmv=n-KuH7AquF7kN%lOrU8I*GWElubgYeY#)WaTIhG=sRK$+&8<RAW;1LXK$nG
z=Brz&`a7Q-xFhYMmo|(u7Fu$g;*F9VBEiCOF+sP=k<n526D|w$9%a?~k?5C80V5S^
z4U}*N>F<+J+4r|T-UgZKubiXp-9&JH5won4#$kOtdL6*TAH+G7Z$Extdo`=5mG@1J
znguO0G?*ht%&5B5b8a8>(!4XSuL$#nuXB%^!9|WBJB?(NQS}ZQLNC6R@RJM)gtF8d
z?kSi}!lYmQFIVh?9k~hU?JzjCMA^oetyI6!hVncrTgrl=_0b{Ge3t3Cb5IM4d(D@&
z72|-^OxU*EXC#t2v~*hsA3auQSTr^|I?~_BRaT`JNtBv!HIY``Acv60F+mv{rf>5>
zK{SV(EDcFX;mNmZT?!IBZye=6hyzyH+uI|X?J?~LQ>*Rx3d-ka)zya|xCVp~;=k8H
zQOOateL?YWCXJzVMCnC7JC}eyA_^bqPwI?1L2f4HB==-pMc;@dv3i8Uq$CPEIsq@m
zudcJdMNo}nz-{{EJj4{pHFJ{slx!&CexiHZ_g~R@OpuAB9J_pkZaMngv4J9|w?-`>
zbH>JP_fk?45w^2TUI;_@tTAx)Czd?d-72l5MACD9>P+J_pdOq_!kYp3&lj)*@-s-&
zu>g5Dir?wtrCN%NFxlhZ%u5lpCF8)Ml18W56Jg75bcr6fUoR>mvL5-)MRLeALLSXR
zu&lc=Pj$~QD+>#0>bg-zJEap3%OB5^)Ci;OZ}AbPX=Bj%Ud-y2#lgvV`BmI?WGnPM
zfmXASd64oqFkE1y5Rn~58qCbrF)HlVt@QDwA5q<P66h&u9HbN&+-6q;8M>X|F*xT+
zF}irr3V{{{CX;1yzt_AJ5-sU1rKmxW0UD7TKwxqKQS-Nm_n;_$Vk#`h{{4p5TTqTf
z^nA#8>Q!prCWrFw9ogFZ*fJd4+;ZKXNFhQJbO5!|zqV-eM}>x7Ml->aP@U+-mEdO5
zes%o;+&58OhjE^|=-omEB;~FY4tW`=rG!a`4pV^WYOg7#455?(wT`5)Abp$}_MeO|
zAT=3L(b(gcQHVw~rFQXH0n$z-l*aK=UrD7is2V#vyMXF)xL_QlE0Fc^4UgT6!EA4E
z-VYilPob7QRMIdo`_ZGP01YQmOO`Kv;J_B#I(=Co$&m&Qvn;2;E-hvzf*%wDIIn@K
zL1AHjF0cIejkvfVi1A=$Yli#Bit$I1#SE?1tS7o4gwZDWJ-v{hyl6+e24vQG=(a08
zNUaO@as}v+oVaoq27zzzL6~QF7<T6;KNXA;UV8hGS|rXCL3{sD`f+R&-E5N;V7FG;
z+Wz1)3|}kdm2(H><Y>yBFn+lV3^bv`37|2WFFy(n9;o^!RC;{FTvSWR=0$_L!}h6f
zK^GE!=fmSOAzj(=`!D$7$!rxD*R_imJ4Omhz`773$u1slC<(-;-rPekK%qr9G>6NU
zXJYOwcA`O!n{j9W1)d}^9+0~FuZoE9V_><6q1-5J^zip|1=b3U<Mid^<)5L(J@;d*
zbI04JCK6T^mXw$xvi(ypxhA1!K{y^e<I`C<e_8yaQQ~a9eC0|`q@jOul`ofh3V4}~
z00!kI>+0+4q3>Pc34yUmnI&;Zu!2{R+yr7~+~IpvO|6=&KLS^fTg$-cytO<6k=8Yw
zV#umy7Cw}WJ-UB??q~Owl^O@-)ek;%(^0>Do^maB&$XN*w(kYsKK!Mk6LYITQ)?{d
zqw?;I+f9R~V<OXePD%9mY1OmPN?fKEc<j9^qjSr=cvOYN%$nO3{WIfIGj1n;<h|FM
zSu^u6Z{kdr^)AX!tXB@4^lB=qN8)~Scmhqpo>Q=;JV<5i{)i~R3@5Yc{;};#b23I?
zAykkuHV5F2vB23e1KH5@;JEN6%+E_aeR}~Yo`;W5(P8G=mdLzh>A$rAo@aAt-9m<&
zr!yt??W+YZvF+2RPde{2xCM>N4G0~IqPUxQ5a0KQHZ!M$<5|>TTG>|e;Ls0(VS0VA
z8`G(o$sfWHUq7z-ag2~Ub&#>{fR=9*GeOofTPA`KRy=t!0Zpv+cT*kMw(T&2TmcMS
zzX=Iy4G<MN`GieHPFAn0uCA^_pl1$QW)sw#-yn5@mngdX0R6Wp6mukAC(V$t)P!sW
zBcJ>*`MB-A^l9jxQLbc+GV~qLs%!wDUyTebOeoW-jB)eRr)v>dreJ~`#=sqs@R-hR
z^(LLaTMKOCOEH^Q6NFO9DnazMP)1hUY~=GZduY{OPxGaiA+9DXNXCH*vP2zKQo0FZ
zw+8bG7HWVr_rB%Xwyh3mip-aGzM$NG<ugM3a6;7p-yNx!x&UFqCNOfOrK$bQg!O|u
z3y=mFYcA(718~uc&hk5_PCy?S@z9`1(_w}yM2eGeZ7ojQSMv~J30Cm|!qZI$JWIhE
z8WQ=Lo8cF*?W3p#RK|*8xZOb{MP9VD9NXPFDlHlVu!XoM6bP#pjg#-90&%on@(08B
z4)h|cUG?^@8+tm0(?hU)11!@^aIoa9fYvGKX`u<MZRg;y0q7C&MDX2v_mm+>A*?4X
z4EO&1WpA*Sr)%-1Jq7XNcF*t60y^=4<je8%Bk~|f(C@<QboKV~;EaWWs6?L$d(a3H
zp0lorOD!_eT@)02F$v``Q82+G-#|<X!aM}<P!U-Wxx<I+aO!yW?rr(Lj~h#i<Nz1x
zqJb9wN}zdO)YYj#eaMZvo%;9#wj&>&VQs?3MV;20!V-7&^__8yeRrhZ4c?xIXxzU#
zUua<dHPsid0@=oOo5#P_JA<eQ15{Rm>EJ;VliAtXyI=_H2j1;~yw6uoK|vYU@ktu#
zV6j^Zkb*dQU2t<#LdEsh89Rvm$f(lS@5T^7jp8(^0DbR$yL`opO&$*)6iu4LTA!{f
zD1aoj7TOhb&*srg*W~$zk}uT5ABW)e1;}!u@Z#W*0ae5g?2t2)U@0ee6&BkM;RDgN
z!@UTgnx~62cPJ_f1^y=x2~I6~wmjMav|m`Kyq1^-NpIOB=kR(6Ypn3Tz58MbB60tq
zpuFP2%lM97NS7lJa4VlZo3;m~7;qn83*<@d5zeX@Un-&x-myL7Cy?}XOJujU@<(F*
zu?0!bgzawSOW}rU2sVU!`}V~p2M7ZG=E2INfv#C&&|<eR@8_709tylt1Du1_OE_Uo
zrEU)R)?kS0l|jPs;Ozh>+|Z)|HSvJ(bno8pqw&GBTm)5I;e^ZwW3m6|=x8YT^Ph-B
zKt0=`l?sAnB}K&>ko!qE_TNz5{XlP*PFi><)G$h<u8_O|H4;C^#@+^<GD0ALpu@K1
z?aDEvscS)YnPctg>FBlsvUCj%?NH|T4-5>$hb4>6#j<ei+^b_%r+XB<YmG<9X32X+
z89QGEb?T0~#~<)Ay5li%f>;k+uM#MAN7FV`t=ppm*A;(7Wc$yKilYD`Wzq|gn>j2|
z%r`t9n^E!6>u5@jU701pm4D^R747@lH7891VG3g(X8AD;yx8p)Fu-z6BxQ)r<cQMs
z2Vcnn`<QuVsP*w-cK3dcpOcd)1u6D}5E|v<8a5q;9OR4wa1WFfz;}t=is$L>`C7l$
z$2`-@q3__$nZ8b7FWAmVIL41>c3?}P#zp`__l;`2FqZ6VXlrB6#4qFNf!m;GX4Y+L
zAr_RfpHV>j2B>?~U5OO$M??l%*ZimHfh80^TTq#Ip)%(&46?Vyt&O4pKhkk8d6(2l
zBQo^?h<{JP6PQdgOxNV$dbDx1L|`(C{?+(Jb@#cGvyn&0?^8~`LlS52vK;y027B3q
zqKiMHc#l@0SP0{E7{e@(W1cj*kL=XJZwJ*E|Nir(_W&_yPZD!%2D-~PgK;BRzJe6J
zKX0a}-$i|W3T80q{Fp>FsuFmhMKg2=iiDFY<nC(OE~A$oUh7H`KpUBun3y36J_>&s
zi0JvW9@yYb683l5_BkrvM{Is;7z9NQXqlump})T4gAe)`K%A{LKRc%827yUsXJ@C`
zG^}V1SZzMr(x1LHj)3dgXZGTYYL^c^c(wRWse@38nS-UmJU3<~6_qfYS4C~@aIhe?
z;1w%z?jeoiMxQtl$p<J0$tN9Q!Zj<kFZ}hze5D8qi9In#D-?T_u}e4)Mx#fKA=C?Z
zA(_%dF_%SWc^NQG0Jc|6Dap~JvyE{;`d55J$|@OjD#VKc^{UuMjyLfyhI0!b4(d=c
zG-$AW@oy;<K6hevp@(5O*NJO<i5WE3eJsQ$tPSQdTwFX#Z6IhQH*|CFfgV=q+x@@Z
zdN8{!+w>Zl7!!&X8%@XX%Y5sX|LGNcaup(*2twSTOQ3S`Vmm^-P2CC_k{(R_x&hMw
zD{7^591V>VUWi9IF8Xr{p9L|<sAg9LBEJjY+66P<yd@?4HdHOxj>Lx2Gcq1qv-i*M
zQz$bi5{-g)Q^xrtt!B_c#;Va1oe}IIY&}tApF%=Sd-!e6On$Pt4*$Km3LuX?0j9^#
z4_)-+F}8#s6qqJT6CmrPgWz66SqaXLcpW5*d$2RG@|w^0;(E!;`?(6f7t4uaS*-t_
zvu_Tvh+6T11Kfc3rlT4XfK)|KF9si=LR=Y862fbF{$tPc@DY55RHRWRAXhwkTS*`@
zuXOzHCd&a4n#x2*M0lyi#ST~c&!Zr3`Rm13_t!`K{R8q}Ryxv>^xrSht$gUW`R|w2
zUuW!}KmFjHfNr7xemz?9*FO5kk54Z9>s0;YXHQrD_2B>UGw-$ky+Cwf{qG85PyFA%
z2hR(Aqoz#c*mWydfTjC$8h2X2g6FvWFnT?twtqaFq+Kd?xya!s-kLPrz58Ivc2XS&
z23Vz-1vkofM(`W!n~kte=9K_JgmB|ATpO;j+&sDCY;)R8)B}Y7aiL-G5UMs==)N@s
zkEYHGLlNc3crpd=YI?UEU-I!plUTwIwAHN*07Vz!$f)is5&4&${Kqj%5ajyr`KJ6o
zZh#hd=hG{n!Qatu*&@uziBLZh!BY77Aobe-;=RzstQ<fNTL_s|!AmSl4o-ng+Ch3@
z>Hg^J+Xmug2Y$*YgjggJnnSd1Gc3tR98h@G@@f|}{XA$c6!2^s(UuB2&?qM;cn=ZC
zL-zS6(jA7(7G2#!MG4qqnzl}~N`1*^>FJeEuWSd92R-_5=D^OK>L9~4F7G{LSNr;P
z1VB!prn0REv`s0X@p!biKNTf$3zD^)9@@6%r2PK<JN#gfcb5$OF~b#7&=2xoUXB}t
z1gLYJ=lC3V3|{L&miVpyafD4p=f51n6|9Q^%q?;szYi#-jOY=fekEjofXj?=paSo{
zx_*5P0ryDAr|V`XAup~~T2@wBQ<JV(yPX!jbjkWbVMQV_1Qtw${bHpfaSK*Y*}&j7
z1kHXBZh+|}0ZZWB(k`^U6b9x8$`J%Z*#$s_#wv&|w?Xr-(}#|_g5Y~^ATimWF*!aS
zft;!maKMnZ12Y&?eiGM^j|Rv?2Z2C*G%h*9v1Nn;sc8J&loWqI(t*wl!<TprxEP`U
zXqCX)tZi#THu+w>k4ac$BrmQzG(4OejZVndgI$O~Hco|Tv?I>Ax)u_$-8K$5gaoW$
zyN1*)DSUIxvL25_e5n%zlWf(cF*xj7)qX=z1E9wd{87){SsJ<T!o)+O^PF4&h0u!z
zje^fu62*6b=P~k}GQTnoOa<doS@r3?{dUwxBK!$O6OyuVVvms0x^Rq@&k0U`kkosI
zBBRAQ>9uGI3l%8JdI+Z=d8s`)2b#XJp`oFte5!%J3I@;*$+Xse;Pz$f@9ej9WZkV2
zsRlMb2U)_^>()hKAqiAQzLVvy!};@cC{9>SOB)F$#bD(+b}KEq0m!D!@oYvRud4$D
zTbo}}TwF`I5;f_myHXhJ2<@+Ce@;LJ^m1m%J_Cbplz=!2l@iRVePAPqSQ0|hI-uPh
z$iZ!QyoDVx`dX~>B^XXYN$0_x;;YbRhAgEySkTeuG{WK2l75hzbE2Rx6t^yM@%JMX
z@3N?(0|+nTu_!Q6b&!Sdh>L&d0_czCKtA*h(VJ`oO&pG3;lsV@iZ_6O<~S0g0MfM>
zc!Qw&{Z>+MmV#{SF4vKhh~4=w=`H1J0HJKTXXhEDvw2tffa2^y_{2L#yLN3}|58h=
z&Wp;*O&_b)DD-Lpq=GF_#4bQzKQ$7O?zFu5a*nl%;87gIsBL##58?5i`Kjs;MMe}(
z<2S<ub>$ke>A+C<6!$L)*CBFNK=yC*>u70(a#($rgs;7xm?YjP=DnkzHQX5KiSWf(
z8&#&H>Cf7HNeKxh5KCs@$9JI*kp=XctDET$x<%lWfZ!I@?YyVTidpYBYCLErvS~&U
z&!XHqIztv`79`cve$hK-+MbN#c?Q7CZkQ2DT&4Q}PKzLf7`veVj}2ME7r~&uk~CfN
z_p9jz$--l(dJGeR3_TF888x-=9|sTBjI#Jd%<5dnxM|agrk3nB?<LC|<H4gRgl5tP
zXecWBJ)5dUm??kwaO_911&PWJm_APz`2CUNH_g{lNJO6*;7DRnQdK=hYHgAH+&fcZ
zVU-S@C*%%yT6#~LnWcfjxQ|rv2?HHu7qzk6>r1Kg^`VHQUb9B9*-{t9#*AnR06<{s
zRD_Ie2I#)XN^>`4IkDX75TJ9qGFm;xV3ZaVCH8*)te*D%Ct}uPsQW7@EJZ)2D^g(X
znz6A1NBC&6Ac)}&@q#(!F_d2hE#q;*sUg#q^Sunb=97I8wJ06$B;B?gt?Nj;h?LaS
zD94C^kL4aSj@1hwTO%Bu9)2AjK6>Hn5=Rh?$56C%#X%LS9c|?6YCMoov6#3$4k}Dv
zI|Mxebasb;zisbU*HFF<&LaqlnWEz2jrUba!gU<QeJ2b)!S@7xeT<A0r<8I#xT9Mh
zo0Tm~o|Ixzcd&eXUDJr?8g?n&tfMK}Wx6^M<IY0x6hc+%8n^pd6L$8qMjzLiSqV9_
z4zeLR-{+t?yG#r`n6Lun;^~#O(|eutQK`|NcMW43jhc?-kOT62hugt2RLCeah!J&D
z>*8J#f`CN>)v~9@B#>D=&Hk&QHomJgGB!Fo%|&&a*9_mbZK-}Cxg$r$*Mp`vZfcSJ
zmj6zJlCb7CLP0XntO?1|sp<ps8ci)rNT(AH6z{AYSoj&@r3fC-x_$w+W{+g@Dt1AL
ztRm6M@t}kAMhA2vGAHB)jDc>c8bUkLstSr6erPxIT%Qnu9gb5kM0I>ejs-Tuacw7x
zljd?_+u*z~__hR2ORuyO(ke43=j<VT51dpw`u{QY9#B!9YZs_FMobhmu^@t?v7ktk
z-a&(+QUpXmKtO~6>C!uxke~vhQl%?Mkt$t=ra?d$igaNpiu5X7`rWVpy=&d9b^dk!
zoH&{JzW05~-uv0l)i7A+(a>Q6Jx)H`#v?#nDhQq@Z~8F%J@8)+45oxZGNT)^f!_J-
zjc}xuP0aA`)R376mt_D2BoBB-;qU3o-@jj0Regd1LPh+N$|kjP|7FA)1po<;8eBAk
zVAhy!p$p}aI#2-3W{opWWVrC~CUiFnREc{RWmZOx<0=T>kOe|pT(?gI06D1v>X)Cd
z5j7gtQ2q(Tbw|E8Vi55L1I=ZkwgSeQ)ibgV$daO-`|fk27bFi*emBCrBVPT{6^22a
zBLA15@^d^_ok|xle|Zgb3<>I>xO`4&nouY;e{b9QBljPX)sNN?89!g1#LmW+*1^7O
z_inz?Ge4OC$5#Y5b?bE*FlG`eqvjq528u+=vRhWV@)SZ<D*tsi!htd3`M+n5oSr1j
z=ko_B;44N}u$s8unzA--u$iN)e=ZiD5FbFc)+b@w<wqVs#j!&p@S1|Y=YocR?HLQO
zfXCodl6ujM7zdH1*)M21rEd8FEd22wSHm>{mXV+b=s`5f2{fcUUu!0wOeR44eQoZy
z$dDJI35qb{VNGn%D2Q!#n3pDn2_8cYD5jHOq6n%-{MvA*Mw?B6q}w-2pF)^5yK?A%
z_w0^=gTzi-d!Gepl>|J#-z^S-B=;0_64{FQ5vbyR&=Ay7^bk7od<8Mm?wVkK)c3V*
z1*F37>(~6C>?gnfdHOU{kVf_(J{=gqlH#q}h+9Zq3K(u6V~+Uk9H2FUs+=UyI$(?O
zX3(X2u%YFJ%<arWxh<+!fzelw)DtVPew1I~0)$zC(7~x@=Le(u{F@mYJxB@$BX93M
z7b4xry<@*Y0fWJrvp4zx027sMPwg_Kei%;p&wBaAnysB(5E+1kv>k1fHi$`^CZ!+<
zJpbwx!VlyZ|8xTpiy-A#SloPTv3m{vOng8@1$6bVkX!;zg&^73BKjws8*Vwpp(cuu
z&ICS6BlbH*J&5=?6DM&C_ClXr?*KEw7)-mgTgE5^g@=a&H~aDD&06FMV&oJDW||)V
z)y}~o1W;k`El+qfI&acH13R9r6Kou*-_i<s^g}<tAP8F(0W~r%;t)I~1Bc)Se{dew
zrPma&LPQxuPKw#BhtDlzli=+HpW;TTxqzC!%NF(EIRGM43fELOFw&@iw0{!pHb}-w
zI-Nct12-?MX^*=RQr@m^@uuk4EvoF~B?grJRQJM0)oC461B2)`NN;NJ6(~?_$oCOy
zj0XB;0PYPzsNKiglV84s$WkudZ3oAe0RlW?LoiFaGkizgdeUB_Q=s7TJJEIzs;lYK
z;I#1yl6aPMSq!YI2%jsuK)|t!WS&ou+8Y~_EJX$top=AK0>0z~H@6aD>G0a^FpDMb
z^6mMLf<upb0@bMi2Kj^Eb&`0A>~C%jLBHRSlKin_1R^D+9-5UCq{u)ubQu`mqa<Y&
zmBR6fON+!s=%AF#Ych&NX0|vupcxQ9hSudNQNWR*Gpto6z+CXMCA6pZQMjjWqbF<X
z>xHSMh~{?CcaZ9iC~*<hOzb#5A_11RN^9n$qL~FIS_`?;N!%tPb0?n*k%t2Ikz#(8
zjcKBO8hd54Q@Ara{|iSjYdYNZF&wQEd26T_N7`Y124V6)7au=%vbQIX^D@lKtYfYz
z#9}Qk@16g>)<I=Y)P3@WoyDFpFhs*o%(I9+y-JuHRG-0!+=>bcegpzUl{knh(GJAD
zICf?ugEShmhZUHyV_`5Lf6*J8IDHDn#>Owkc5gXnR>^WkL2!lPtZ&)*PUnR&jnbME
z7cQV=IYlzWMNv=askJ~`lkN;|X2_%@t*kX{dEJGD;Rf(n6}pkNz;??Ya+VM9#hHja
zRa@44mxqkSYj&`mo}8GA-n!-DF;F{8E+=qJ{$Hl1D?flVSx%FY<jtX$;XuZ1a-Y~U
z<qz)waAAsWZd9UU23K$rdg#}>8y6S!;b}kv5PJd-IyxpMkr02=(>6_vS?yu!A#n+4
z$ckL&k_d^25KVCO_~KdYO2V*H_nW4Xg5+4Y+Nr_AZ~`EXPOei$!xOQEoEZ5b4~uX=
z;5BCE=R<&-e_ALuRDeneT1PnQ+P%lTK4eiw;T}Ns2_=OyLl`FV#X(&oZySZL`Co%h
z-WWf8wpo!c;*iw*+*~l07(IYjqf(57hF}lKy+i~N`9k=<Ua&#Lf&18LVu!BV0WTH;
zt-plyGbl3B{Iv>-?KP-5n&F6Czp^i!q;Hb;VUgdgq+Vo4SuWpPmBp-d2=2RCQZc&K
zp#r>}T3p;m)^iA7sN7IbYB-EzK%KNQBsmab>zxcjF`@WfaiD~0+`NxvR#KE8okIdi
zh^VTmxq`7_K8&UXVMra>L@=_U)#8}}P8qQ8x1WWT6%>M?&-nlYdT!Y*A26akfa6Bp
zl$Me*Mkv(W+yEtI0OcnE42~V!1I0Aax~~n(Zf09}Dh6XD@Zq6^dw&0)7U1E-rx@S?
z4@de<dg-J_I|XkUW$H-~fS?24x3s8}N8hoihe;hchh9bpO}IW3c@qLrcBqobO9FKM
z9@dr}Wd()tYlp~35KJ99n;_gYyrRSn`xPV*8fs9c_Ek)~)?cOY!RL|`Mv5@roWw~W
z)G;bqoP^)NM}zP~BY&iH@tm5P!syCN;>k~1VjQi>Ai^r6LArt?(p~JVf@Xmi8may-
zUku4Pz{8^8GX(&;2a5_7AOn!u2eh<p9%nA#kDw)wnB<a#_90kB#%x~xAAB}Tj2gq+
z^-DvbEi@qgN(2GqwCSm-P=nTwWO9;&I_8XB|6@sAwquc7KT@UYZ^|ePaT98SV_+kG
z@HHQw{{=N}XS9tryIekl=hhVlS2R~gNP~#Bx@Qxeuh5%V!>d3KNS6qf?ljK-v12)C
z96C{bbv|Fq;fDMj6ODtAf&hZhI$?C1qK73BMEnT#RwewV0K*s-j)%LV8##_PDHPP{
z%;I7gIG$%7IvI)9ePujoDZ<gxUB-dM<>{F6jE$etB2ro})70y)j^S^gSy(SXQt295
ziNCWO{!1U=dNqJ$Q#tOvNv9JXrorL{A&L1Rl@f@BETkV8G^J1wa{@?lfsY{cZ~xGc
zNq7m?$8#cE5&-~|=7fL+N|mSRUYrv8KEL}^oX0h&aTz(dkr$$HhHeOP?o-tSlM(V9
zxkLfK?m_qNsWB7-54w+(Y=Cqm^BmO4W5)o%5Cx1YHi1(hU>6A`t}euLgCrBaluS%k
zG{Ym3w3h=<FR^E*1dvGu#B1miGWF>F_o%4a$EV(>p&7x5{FoKo#6l6S{XeX4iIoCr
z6pUALXNaDtc%WFQmv}3Zat)j1MjRXFTGO;tRu(6@;HcDG%m(SPN7fv94XDA6gQyJ0
z=dk^(Oa%=DYY>vJ1>jI0>~qQ0@2mawj}SW#()h|l&@I+~4phdGCRGv?gf|g5A(u!T
zAD%!Z^8|1Z0JHY=h|c0)c3R-B5ut&<zrW(sfw*C`WU~;_)gbi`;=jwAeqTUvLC~o|
z4PcDIlzI2h*iAsY<Avu=oxXuPN~Yv<Fwf_s9gVHMr|>1iG3!K|n5f)_YJ`-<7@OhA
z!LU)!QJr^R-SFH7AWL}yQSyyYZ`)T|S+N>|SHY6E8><#*esn}Txdu0}wr(3ZsmXV&
zuCCG$e2i!ME9DFM03;Cs!Cqlrc0!VhS>Z_}LLWc|VH4>2`f^WKtJ6+$bJvYrB<m+4
z&KmwrmW0bM!3#_K)i1TYVl59i5tzfVo;ohl`FxZc(@r-ONskz)z+!@)TXK=2YI>7M
zr6Ju&Z>UDuN1R@rpWeYY5Wu>R`kfDn{sl?P|AAL=MKl39N;jbcU^q|+XEh!`Y9#oU
zE}&W^Jb1!#0cuKEQ!>r|)?~Aa5?zgG*ih8@_oj=HN*r%7LdrD<W3jkGABcvKpr0T6
z{THC$XkPbqb$3^UnP%@I^aTJd|9<d**n1YrfLqV?oP=&1;X5bA1Hg^l06ObXVx|hY
zLY7w9hxJ80{7FVuq09O%U$)^{>%XWag%)goRS+#tp+uxEka;g$tr~P&1Z_%{A`YIU
zJ4QW2Lxy4M;zG2W{c!y8hkt(E`x2s2=0WY&xZ_6T6Ch>|f!^g{_fZYxsZq%LO0Yi~
zmt^|SL4yz&#+f;+qfWP0^RCjtzl}UWZOD;&<{DHuPeM{Ti?l~W>>0$|3xgS!=8&Wa
zq*63(C6;We8~E^h0-dO6Fz$2AZgaK#*1P}x@G~f<zk7Ecf}nx?4*g)@ayp*2(-=w0
z0SePuWS>mbyGyQ(d|+_VtQ;q@Z`=cIoH+cV$CYNp+5iQp;I}^qyo#%WK1*i2B?iyu
z(Gse1A4tE~e8QVee!xq%4hH~g7Yca(9F*VU&QsC7GRTM_7<#C|<KU+9E2ag&n>7M{
zz7Vu^f9^NJYa2j%55uO}n)o?mGn-*%W}}qN3YJty7CD}7mN85no%rLAcOwOmnDk_$
zYQ94HvSmDqg%xH5>tGDi)gnNq?-JtWGfHWHZgZ~h=x$G+*)OQi8H^Yqf91S>&I0BR
zM8`HeRXjb`?g{T!4xb#wIpKvh*9U7hO0%d9$<rY18G*cPHqjQ=pcLoYT<Iw1rt-qz
z<T&OX3#T?qL<+?Tr~qcFXA_>g{J;FtXmNi~)(a|-qZu8~(GNPA*%Qi~e9z9;hCWbI
zv00MYWuh7*mQatS7jL}q5%&K6W&1Mb0F8lkx{a5Xln}L>0MyiJRo~$AM*#p#s*z3v
zX5{h8qsmC0gW1*W!FVK*(H~B$*v_Zzpn9!S>rhqzz-N+4IJUGZ{evyTUU{duElryE
zXhDfzgFmeQ`r7Y5h~ox;vxZ=5Vc(SvS4`=15xk5P_GnCu^47@^Eu?Va<b(q_PDC(c
z9`r=*i!F8WO;(7#tvx4P5I=r|dH9S^KcvFMwHe!`$gOOfY=e>y%I17x<&BR#4n$Gt
zvO3xa*O3?ZjnNzlU>(Law&!|#I_x|>JYE7gaT|8PlS@GLXI90sg%b3?+I!$BIDh9a
z!iz|Knf8jPL()`GThu2V!TbcTCoq<43a@akQXs&8Cf#cq>ncxy>q%+=S`98#!nqV!
z-0PsyiQ*v>(S?mbTkDb<1tCrj&guZNRoCA1Y0u%qR<?%&irkkn0oFd#pvYkP`x(Ru
z!Nz`<Ea<=500@{ixknWev#_2L*A~m(${F$6;OOXdj2Nfl`f*G6;*6zXay)l=3fiId
znd&3dK^!IKjv5^tI6<8bEq&L@RFBy>b#864%PC=*TaS9({)l3Eu2$yFCMjNZMP#s6
z7^-leNr19Me(nyk4SpSJ-&dF_$)$)`eo|e>VP_6?h2%{b{>Fny>ut7p9*TM;E=}<Q
z3c2VP21iCxT%(b%gsFW1IE0rVq)>%qEs{clsD0nPd-vicT938iP1PHrG7W3;L<qbN
zctm(f664_}u4pYe8lQ$mcPK_7e3-GTfL7pXFhyA)KL`6VHGuRNU4CvS5uwJu4t`7u
zd0yM-w=lbV=H$uBBq@5nMRNEsQkZZHd6HBI*U4RqME~JNiI#G0KKck^p0;udIdCTI
z&@-ze!i~YkQ8?-`EDLpT@jAa^M@&ku7ZDnL7lzlt$4OCOO+J%9Fd#3Ql_O4K8<1^(
znghNN6f;80?$^F~sw%uf?-XezCm|-50uG@&-VF;7w6NT|C$)nM2`;|CM$J2Lb`@aU
zb;KKP8DC(G-4o6mNZoE$<M3&?p>fIr#yPLhwXkN4qn71qAo*RhG#ph<WO;Q`FBe`J
z=G5=^I=45I6=P$FpC)sCP_3G<lSLH@ocrOxj*g-}tTW3%HZ&~Eq{&tfP!1$MANn_<
zADqU)6ow$lpLV*QAm;U7)33h@s`UJdvVX9@|F%#ZNh1#)|0jVrVbb>%0_vd5p{-=-
zoY`|&p<<OE&!SY~phe-J)@vpVo~MAoNy9wN?L`;eeDg$@LDjd{+&GAAQgFvI2{((Q
zdfQycqElu*i(G#C2E~eT0VuxPq?Doz{ohKKu}x-+D%B&N^J?`TI3q7mz%n)DRxzU!
z#V@f|ix{+G3@v-N$-KgF@qXD~Gl?(h@!RMgAT<g(%aPI9N@M|`cPJS|X1zL4XN2QE
z2BL%c(dhE(Y8ndq2<bp<a=C6(4r#o9;3o2)+bYkE*Kdz1Z#X?mBrV$=wdEua6cmwt
zYDS8X9JS(R3e5>F%AifvaqvL8KO-Q}9Pb8Gr$2?QyFc~2foe)cg%nX=D7K>Eie8`%
z*jOlz&&6u&GkHbCh`GxcIs%T(vz)<`zKPySe!L5qP}*$v%gJzW*Zzm60*36Ng01&V
zAFc30qgM<3CsjgRQZfx$>J`Q>a+jgm))kCH>BE&6LiUm1rU>XxVSTUZj4h^6|HLei
zw&P92el?Uz*>3`x>-0P9Q0SNo!7KX7v2@Qb&&G~pFthAs+Wls$jLF8=TW*Pxp=#pK
zCTbA@8qxu_n>c2scMIpC>duI)5p2ZS8^EkpN6){y@H8BSnG$d^)^6oetekM&SRP(5
zQFHkX<Fw7U+PC5*L5KbX+y&90y)zPo!FP(dVk-Z|8U1NQBGeME+-)vS^&MnGAA>>K
zW(@rKMEpf!GHMdimiUkJUUP^`z{JuO6o!N706P??!EX|J4yo2hdk;?DaSZ0!p{x&f
zmBpPV(0xz5aMXhgV(@bJz8$-J1LN(9z&XfQwb%7^h=)}(?k5ZnUc2>SAnyaJ8D79x
zfy$290TBd%h_$q$=sB9pS`bByCZD-aqkn?|-vB(gn~Npe_uycKLv>caL`{c1{F9~#
zw+k206~&9wO*cCC?7&|F;_lZGl!Oc6#9B4B6#o*u#Rs&i!i5%|(~a2p1cMHKYCRl(
zZaLA{q@<)?LQEY=&o?RKtS6OTQGKMnDZtDC8GQ+i!lSRyPoYXb*wJukwpCrFbIqk2
zhA|`S2JMY;%a1^Y6F{VAlT<N;p9cjs-*6xN2hkSRCvOYUg2N?wE<Oc7JlkHPyrZ7l
zHQ*fp2KC3`mN#u|7J$n|1kO<=6UqTO(rSQ+9jU+YA|y!5z*7UffTc(y;S|X3fl@~p
zq7X|h^xRL;>K?4K%Lakry4Y#Tlj$5Azj6pzf+C*TyoteqBH7~?K`$gCV_w6dDYYVg
z32^YNH!hw<C`NE2fNK28K6wzrCInWt4i^ZB`5JO<Hp_dR#=Cq#$bHe$6@5m4OwYS$
zP#uJ=paAa|X(Rseq)#VER+PxGV}GJg<WAKE2WyPWuKDP=F`$J0N`E5VwioHh`R{sE
z5>69yW5Cz3u~hLnL&FuyIoG|15A*Dno_|7^tqW6e)jog!{Sx06eP~AbB0?;w=<Ca}
zen698a2#{lVF#xHJ9{othyddnP3%FCC8GU{BJJ08_<hawwGYX~h*(AHOJ=34GzYvG
z%k5wp!&iR^=8idkzo38sDWrg@Qv`Z^M0^$c-U_WSgeBOhO6)(h#4ToGmVqlr!vUOc
z4^6a><;2@;M#$Eq6NAEL1PgNa-dWb+S_G3N*!ER@v*x6u!8X9s<OMV>SCF({B1JbB
z*_)Cfg5ITd{1m8H8T@G<pk_cxssNbkL3g=VD?SfzlHa8tXju+eo4#8GcZ~qyXNX$3
z>X$+`Kwch$AT#On366y(H|NdNv5Os9Wbp>6ZLsk&Ra7nu_!>dn@z)9HJ*M5%fZmaS
zQU)K&U{}K1Wl2jpPu;ZG)p<t8sP$qGd+!_D1c(y4G6JMrCu!h`YH_*KyF{_Zh|Fw~
z2Mm;#mgH8l*kwTusedLP(OH?u?j`L9SP!Dmh48f}l}g?pX}IG&7b7%QL**B#VQHa@
z*hRU!`^}EpqCI{L+tX){)`jR{7(@g5*ptV`(DbWe4!l({>gm%LGRp&dON)yDPS!Ut
zM3o01_>E7!wL<jj!t|1U7Xt4BNZEtk%5Tsh2UGU7>6Kz+R1HkND)|q(Drz2TP3l4w
zv#!Z6NJCk@*`nsHm#sd%em9a{3nxoexR7X}zq@wq_Ux`g6me+bGA>*I_IkBGPPT6(
z<`XG70642*7{;yIZ0hRg{_lq{MUe^`W^A$rjZ{9w3O?G_z<3s9WM?<01%!kX4BkoJ
zwdsCim?QIVKCG+j=Ry*5MJ?LIH)p{Tvgyf!5BBk}j2Ilm!(i?+w6t!D<QIF~uG9(7
zmQ=(Z&G4mn+mGA{137QmlujQCL)Dr>o_mgHUI@I2sODG=GL(&Ua>c$6ZO|3_a^MqD
z+@hyWRS(3}=;Ydm%Y)?<7H1=dHo{AK(A5rk0TkxfU6k7`eSKtF$7~Kp;CI*+(GF$E
z7+82_p1K-+<HDs|rOh(x<Q*-H<syVc&2fuWcmW&~YGEE^{2Mu1gFIP)U4q}s%=3A6
zHb{hcGyk_%ei8_s;Ip45FojKIMQ#+06xH)FVzz(IiTT+jO<?|8-}&aUw=*hup~v13
zA25g8)PcpE4@|}v9RuhoZ~9KgekfoC-~_pB$rs)aDBWM~yAQDM^!DxBLBc=B^i~@1
zx%hAq)Z9#fUb5IhbX>#j;ux1fa3vKj10&0GvK1*oFduY43I-}!^XPT;Qlz6!48bU4
z&^sklj57(X^n8aNr45b@>ETWQqas}oL~HX-*Kl0?(zm%$*kn&ZFw3Z|ZnbnF2Lw!&
zD7F5EY-3pGH^OVqio!V)m){%@ICAh=BvTfcS<U+Xkiuu?$w0-s<F%FQd2o4nb94JI
z%FCNKjPV{D1JRLxrX2rQZ*HASx%NLxm)(Bb&5|ZSkDd8l4Irx~*)C99SJ#?YUth05
zCbDK{W}4{3Fb>a74JhdwMj?)@uQ~dN2`ua@U^B09Ep*rDuda{U4D0ytU!PmxOEXH@
zRvzanSl`KJl+xv2WP4a$kaFMGS4U4DqoG2yGg(F{^Gq9sA4N~I&NJnO2U+^V^gM+`
z&bI2H<jc?IaIHoM#c#j=r9LllInwRj!N|FWO<bihv=v&2O>Pv}2&?}4&p#uREwBO$
z?xV*?b7CH^PZP)EW4`!Y82S@)_D!;pTy41u5bbDb95is6!0$xelz=>#qcb%J;#CW<
z^^W8Tm3K>jWKEN{PZ;qC#m38U;wEwilZiHQ2lQRuS*}M3TO@(|O9#%72x2Rhs5hQ)
zpco+6VEPpw48J%`19=0zpvQ)xyS_~vipw+tihdSDrP@~FIyWDCQy27=Bg&KdlARke
z!_pf6^#T?bqCYn9TnN>0jfmvbO_S~hbx7_frzDt`^{+=ZUoS5iT)2)Dr>xH=B&_aa
z*i`{Z8AqcYT89$7Kw5jUM(7E5eVX#+%OC4#&B@#MA26x)Woq9O@<`yz`Cn=qbO{3%
z{O+f1l3Yp9k5Y(;s|P5PeH<d{_%<^_&cJ!VB8e|`m$4Vj_dg~A87zFmXrYWj1Fb-;
zA?Y=WQX<em&GQF{Q<%VQgUIBznkRa$Gz{2h(CA<r^h({@Qdj7!_Cf%HO4q8PmJv}k
z0T|}C3T9nS(8_ijr7=phaYVefr>E!6x2K#*yr)6or9#3F+R@%#P^{oTeO0)&39kMU
z3&RQ+XC7^XZO-7}AVnYpsRySm+{9JqeCM`s5W=Y_aH4n`ufOe+-CR})UX_rLNG1I(
zszHr%FVA2zDCshte|jTC(170K1%MpdkpJ6YjG-*kC5AEpN*IeF-LrSNHfP&)Z6Y$X
z5L<+oHi5ZJEbsZmjj@<%VX}j81=Epj0E8GB{AT$O*~j1*6S<6fQ{BD~)nD!ctAs<!
z&;}9SNE`ew3LWoQsV{16HPjax(EF+4Hn&LaLPv#(ZHY9RI|!?Qp$;^+t4D7B5xV$%
z0zYT0t-AH=J|47z(i+eRm3qLA?NJg>?X6HLu*UiNI_i)2^L+TUJcr>d;i5*2Q9cjv
z{dCVXHlDGzfsgN-_JvjeKo+(K+iN$aZ8|bzE<PTH&D?c#quDhD&$E}_-F)2!mf}@-
zZ@1A_R4o*d%iN9V8bUEQ#5;{U6RBkIBT$WX<)`ry0|}j6lSei42yy1H>vW@Fwx#7x
zS3hGd@AtAn9u!r%XO7tabpXLz6*3J5I;`z&J_hnxey#-2X||XhL0TeQB9hUtuCT_q
z3$fKg|E<t|m#5xf^{>IZ{2I_z6Z%)317WCx>Ix#bG*Vxpno5HcqUGY#{6!a!yK3TS
z_%oG>A}Sw}>jex9rT(;);h6{bD&9#tFYZ~HKx8;@?wT?eTdv8!4(txv&UspLdhO=f
zvnR^`9_~ipkw(Lut~%=I=I&t}BZlE}gh%rvzRAg(7H56UUDs({Eao_%g1)IS?cFS?
zubxk=`D3~euR1qXw=lDFmP`uerO!%`0`2>seQzb^nHWPC%iiG$E{tIgaDSS&eLyD+
z6W$g5sc2S~q<@y>#=r)PCkm?}!q+p1<`8mD&{K_pGF*gQpx*@sROh;Hz85B>OYKNE
z=)Wila=HsQZOF-J+6R1N&C!088l`>i(AAQRoWqVfPNUGeCnu$Wv}nDb%3fKhRyJj!
zYxLLeA}!0$X$KrIa#qUb>$W)YDx%=c-#KJZ%{5aBb`l6PJ8H4xM-}_3_^DO^Hr$p!
z?Fq?fE93jFO?GyeH^u*3)2Hye+HTpk*CyIYf9K5c0W?WPhGFh8{*I21jmH8cyU7wB
zVCIfre^tX?D|z}8SO20YR)uS}2DCP674~untsaTSP=_!6Bj5S$`OlQ?=ni9)Pm}}w
zDg8;t{x98awAb^zi2!KECq)r=5$%h%nq{mVxP-dRqR_+tYiWTdh9~vUen8jElUZIW
zdkIj3TP_1Bu9<iL!^kG0R2T<+=+L;%&GVfg>P6fFkw-$?Tq9VY)P49s3M^4-#;xJk
zuMg}a<^{pSJ*0uLd3ypSr6*DbhmWW7w%rkIBShVxe`+3B*ceew6+O<5aD-ObGABV4
ziO3PFWlBB4grZV>B(#*h7&zE3fI2F-JZ>4P%>(H6#e#q8RN(X}o|@xJ;++^IRrb&Z
zVctu3_3`f5HFe|vYU!}{zsZ`S`gLU{hCzN|-DNUEl~qwl)~lx(R6rpWS+d1U;ay*B
zNo7!hwk_x6`Oam!)3MukuX3la<>ozjK=oM8X>v%EzLEHO`X-uqZu3GC&~3_5SFk8+
z_2PYb$C4jgJgEE4lLZ6Bot}_7%`avT$RD#H*eTy8s!@w1p0P%?cRT+MTa)2(l>K&k
zbFTtO>^nkwwhR<eY>FN!ng#m)e>YsgtkLb-S(aC?|2jaZ&+aThf26~3`|Au~Hgk|U
z9NvO?yiPiMT!<O-e(YVN4<i(F#8tfzmMnU8P8Fe#l*^;_A_yx9AOK%<+T69XJClFl
zbb5?cGSaGvsZ<+L_11gvlp4Mi3t4>YzrDt1Ux*&G*Y#WeTbtd!7kJs7-hFq{8Gq+L
zcrPgVv$Mu{q57-lzJ`1Tqm;0=a_+zv%v!1<Mk)UukrO@;qK7+4*vZmc>$s!i^>*5K
z`9%t}W2;q1*;shznbI%yy;U`vX4`kj-lF;CL%mrZvEJ)Q96!nqYWA&Lid7BJ11xy?
zE9VQ=s;fF}7-7IUSMgr?wW_}RL_^e*b?bA=Nza28WXo^3Gtfy5u*pphxQ=5G@bwn8
zwK6?gCK`b(#thrG{vEYsTzjafx!I`Ji1+j=(uh3$Z`>cZ;FNtLrzN3703l>C1>Q+7
zAK|XllT@5Uk2@<W$0gPm^n3y%_F7r>w&Z?a(b&P;D7_SYUL3?P$0^ov2s_>lkDv<X
zpu1Mpax(d?>SSt4wsbpY^vkP?SiXj?F@%*^-r&dvo57om2y&^L(06ho#xsZWyUpKu
zrJbA9=>V*9ADpv$HbDX9y$Q(SCaMo-9UYygq3VaG^Ti2ggz5azjXpF(m8)H=_IY-L
zr;li0)a2-vD;LgRSHMkeS|@v|Z6VuA;l(=9kL#Y8lqSN3CI^D>Ap7;)o?75b$1P>k
zn}+;jDTlBPcpMDlP}gvP3&<7UM3r;Xo=!SV*e5<AWnw$NYz+asXFbSn9y09M=14|A
zi4&)U(?p8yFn9qd!=3`fvzRhXPbUVMA=l(vuyytZ&fXvBblk>^9Y&8KNB@q%ny?CL
zusW#XbQ?;M6F^Htp~QU%oUSGh6=Ecvd)Ycd!RX9%a|0#^NVF!>bIjgED$y+N+{6ro
zF@P#NSiZn=K=;M$M~Oao@*fzrad&-+)4mkNee8#xpk(I|Jeh8Vdnk4R+2k=VEWr%3
z+h$8y!4VLeu1h`6^+^^mED$nlzOdwN*mWWa_9*AF9$6}{cJ;<3puD)YRCxdXeOQv!
z#JiDEU(%GUdV_<7pq}!CYy{1+iYb3&qN2b_W;B|tE;g0L0?ENY+jMI^gi}4BY*#xm
zd#3>Uilmprd-@<E#C7ybg_$rAosHXEyF{jS>(|S^_wOv7jz~n;atsJ}hx_5u(ozaw
zd^g7re9&FtW-tMBk~PRB8?djFs8;Hf2Oa{t$3M}oTRUKx`XNHRB;LPfF1@)nXNUQ7
z_y#DwSQK42*s-F&Ww#&ZXRn}nt50GRyZd~~2BVvB2Iv>Svf87h=ZWH-E<8&FG*A$0
z-@K|_$&yN`?PoCu@2M}@2)l>)wTwy@^QM3oe#Ff|C)Rc7psmAN-^tH_W)&d48j=Zk
z!5R;S?QJZ?6T7y_<_5S%4&ihVql8WeeSQ7ev9zqQjdI8Fur(BU7q#=R{js?$LI&62
zWc4x{`Y9UJ^iY5HE&ZOo`DPdgF=cfXl_&GC&9qk=vKLwALESv1^z0};j=7D3QVH|-
zuU3h&S~!xNm@MgTEY0zsEez9i|J2sj)^C(^9oKLmF<f8ZhBvJEz<$|SHf-X(gyZTT
z>=^>_i~kY|UTZWX3u%_{Q+O!PBc>sJ-JuZ&My-}Kuw93cxcED-@-If{qgMMdWYCA6
zhzLNi1WMD!v)48si*81|;3pOaJLemAEQy*>^Cjx}%>MJ-@_7grAAV$$5;?iKr5{-#
zgLT9k1}!`a!&6h49p_}c*4|)OiY^{~8$PwC57YnQ_{Zp_0sU@SJWW_&)zfz=@KCmk
zth;w0+1a;BZ){Y-k-2TFhbrq7s=lW6a!`*6qd<I5V_4!N&*D_wK$H2ijuf?uDGG57
z0W-nWjs!siCS_H-N=DB2?lp3V0)SK(DXeJUM?@bI>#~bek}`hgo15Pyf4y=!VcY-y
z;cfokgX@3fZCBWT;iunUf4+Eu`@qHDer8ktS+c)W<od+T{lEYA)BePRe)GppYq<UT
z<bQwUoh_IBp514eBN)L`!|`RoLu%H=I5`%JUe==pO=X6YqLhYlP#dkW7B@Ne8iXUc
z%DaF0MHSLkr4CY=O|D~)-bhAauRxC`Qj1dKI$EeG_N>2e<6<BWK4b^yVtH5~H^=;D
ztw%+*61cz767-$JZ@j#)*yoQPDDaRP+yg}y#C$UW0$xSgqKPU@+WJjCwU>F`H?ZU`
zglCqbdU`$o%wGUC9ixvjaMieY1D6<!6<cfCA@M2DKwot@3#43xZe^{OKy?SCz^{Pe
zPtAD}tqC4{cE07%cL~wVUnAHIC#yQjWF9kCe1x^cp)71KYPakwk6yIZfHn_|-4m+_
z1Xl6Nc7m2oP8J`B@MzwdN3?$&$wbmoq1+tumK7`(F6k<0n})b~!2$B$$c7_WOgL)#
z3=K0%e5zh7imNutc7FsCPF~&yhtl+!1x%(Zv(aElI_h8~l2$xLRYc-+6mJZckn{Sn
z6%A+#0L29?!Q^MU#cRJ^n@d~LkUH0sO<xRk?8JZ;VJQnbcv9}j<tHC4coE1Z+Q8XB
z>vH{1>yBJ~j7Tf~4N+b;=`xC4s<{c)=X&0Tr?3s|)}|l0Gndy6!4G}uTCV<5*1t}T
zbu-^!yt07HHdO0nUAmyDyZZ|k3`<D=JTz`0haN3vk3{FAhn{*_wTE^qYO(Lj<wExY
zF;ZZM$;@5YyF~YYsoO2O5~MA4=<wkfu2nWclYkbV=Ml-+qtK9MrCTWiZq7j=BuE=c
z4`!9Os8eV-9WgOiSbMF+#TNK{`=FHbSHraXBlw>`x7Nt!qFZG;EYUhxLVhZ=;7N3z
zU+m`ed2N|?1RvYr-Exl$OqrFIWzNQ%<fNpo9SzU`w?Uarr<kK=En8pIy`uHE52)@N
zB72w_U~%Il<>(EoM&Jx;1jr^^ei6l1jgPp_Xd8qA+C*gKf}01>${OZ8lc{}#o5R1H
z7U*?tD*6(LExWZNY;5BpwjB1=uhV1RXO;!}`86ZbovfLDL^bLB%dS>x;uYPxSQ-6W
zZCIiy3|UsdoJck3+-bW-oOhBN&Setu0D_K_JKZGx%%rCFu^cB}9Q;DFlSF^t1sMkn
zY0Rk7xphZ_^5E_bIbWlL*ZC*0v6<Myu1huogZF=lb6IqgcTTw^eZqL_>oK95m@0_g
z8Tt62?+YU704|wEmb&ojQQzDQG`d3?8q%_n#0b6hoBYhSg)pT~@txS3f$5$WS8THx
z3%(zbV`AHwvHj)jHDn)+mk&zK9g$O5MXe`3y#11zU}4w_CD^5~hj>)uRctn`dsU=f
zU|gna>T*ml$}zgWh`mQli`?Z*L`MtI021&80=*eEX<|M!9jIORG&HM`IIxKeAG-uV
z+b=6?BThJtNcrt~_Bvfn&!iXaH;{#T35E!dbM$f{SkFTAH^6KOyH&>-hK9)c(`kE7
zDx?5oTY=m}sc^EvbRT9MNQ2Oj1qB+A$pJh4_UU?vB$Nu(v9vUK`_DJAg)M+qcMhCt
zJ7hE0;m$%^m@7tz8GCpt_Tzv5?j#eRh3rp`CjC^`xjyOKmT95|EKFM83D02+_2cBi
z&SPDwW^xP6Rw-H^eIv`&M4wxKR%~Tj<Fmi7QIqTRht>|GsN?~CDfAwP0WFT9Hw(&q
z<kYBZU33wZke#vo5mXe!JPp5g9CvN8^FjSofxEt^SX1zvBov8A7)>}$j=<0?`6<Zm
zUter3o{TNa)S?95_ut1E4?&N3<(PQByGg85Mr8%c>K5eSTxlT}SSyvTFVb20Z~s$G
zU*sqtqJPjlH)TFTYnuwhvL(Y!1Nm6bNhUmVZ5KpN#U5MQ+nlS;6cCaxz$)D<4zhFW
z<=^zrKV@68duXBh>?0@Jhf{U2@pdOIezOrhZi3YYJ*iu+XWr}_%};yrVMi3qWSI_$
z5$gHrOD+{xWhj6Vu{0oF#4Jq)2Gz)0BA4;f{?zV`^rcC9GG+@epigfD4l=4))r!9k
z(T2z8KAdHQMzC(i4Y7uOxBanY?;B@xU+kQ8%A0u)ecwdRmwA92?T8uDT_!_~C%Dok
zSe(MLy9&5)FsK_V<I2NP8tEq}Y?<xmeGsQ-*HL4LV%)vN+L{!=$LWcNEM!g-qWF0H
z{it2CXY|CGfgQG>TD>W<OP6WfbU4yz;eCFVFXMbDvVjSW#ciyW6KSl`E(OG3Bbk=d
zfz<_k6l1J5@*NoAksTfQ1z2_n9x<GePF2?~V8B-+Lq~W4^w8o)dS+qoE<X2+)7m*N
zTNVU3IcrXd?#(et<6;y&-933dTAUYPPBa<clt{NX1TU*H0%b!x`ooiYH4bOZTTcT>
zSAaBJ6@_dYDkZc&nvgaZ+&5^fYx&+~9bMwKVD1S@S?A5&7jWyfkWABPkFqoI@MNdd
zbMOOez{qlahNtmR<M*yHYK-oTBj!~w84$r^(GRM10b#Yc=dDL5U%y3e;DM>C4(<DG
zb+uYE3#g&6ou0d}HmnmI#++WYf8B01DC$RWswKcB22P$NrtM&R%_X>tx!w!i6E7}b
zk1cqiFtskz;;X=x%9Zy*CxJ<Ln9}`d=ES(zW0C7i<jUOomwNtPa6ZqEr$Oflu!;?G
zpSg<|1Mw>%Qxu!?f8UMJFos&9*>kC9_?ZZZ&3X{|(9IDD2J--^xa<UOXYVVlsds8b
zn`D7DV+B_#57ICC;f%y?v!y)stlI7=SELwH=PfSgQklPg!QCfdd1KgqH}vML7_6+O
z=d&IbO#Z}yk!i#LA<tj9MQ^id#pL8x2?s5#QgNwz&S;q|^RqYb{ol11kP?x&C-qd_
zgvsD_ZSAN+yHZpc`yziiDFcMO*1lUFHb%LTi8;t0B@EKWKD7PY*<H>N_bgp{V@6}w
zSw>_$QYba!z~sY2rO=w=vfaUZCWGU1r=h~3JQ?S-x!*Aprd^%d&8-FTF72&1Tw8Qr
z-}>tkJume^mzmgyNZX@=&TK{s;k;Iz=V@o={0di{?xz_(TC>8mZQz&uFkZb>{G<`k
zI0{Nit&qxCM*gZB1XjaDVqdaG!K?I6VZCe#%_3p$@MIF=g_JshL+0BLUrkI(GDop>
zb$@F*bbt$(-FXu|JjtJ@!%=>?0^oD9(HrdRwd$d20Ortv^Ci2>=bvcP-w9~L<^Ekm
zvqFw}u_3Dr()jUtH!#uqHgiCiuQKJM!Tq<NVcU#jL#9apX15Irrpdhg0%mqbAc<2)
z2?f|)SQ_)k1(`k(gPI-~21$iwnL0$`9b_P5vbU6;Vdy*gzCCEV@wh>s>NislE61LO
zdOMU0LiBvqAbs~qVOo_7*D_}rFXH~7N56$V>1&QBX_t8vp{#q3HvSI8A*^tKqa=qS
z=1Q`wX8Sy2I2+#ArIuC6q7u&}eCo#mUf5w!<r3r=eJNo0;21-+s7VUjJ}F4|oz0`r
zcRW$aq1ez1>Jbnv<j)T%bK7UUuR!mb1-?JjO2`kEKk)s(jGhK|cs<vXlnw$hYvZ4d
zk+kFD=!G5-a~MS7J&JqplqOdYA~_lBG^SBzxm1GF3NVS{{z-{C`=l1bj#6Di_@X?L
z&CIQzhR_RG_e42d@(Qq+rIp_7XWhqPT+5L8#}osTcKB{4rUi=aYzA~^q`hRtTo7qm
z>tiIfa2jjd#o~g3;_+GI0KJ>AB0&g<-D82(W#Vl&51LJsJ`V@MP(A~F<gIHM5NlC4
zCAkUg>Y=O*fnj}eN-ZexTh~Nzx@&PHqAf$W<Ex|RyY$<(j(0wG3qak5*TqM#kAN0u
z6z|pE-Ca{|xKn(4y0uA><ixeB9qi6!&QpxPj_{`WHmwoDXe$GwDoRNR;RRd1$RiY3
zNJC!LnMAgb0JE?yy4pSwt&NkZBhub|w}L5beE0xlvl@V|TS4EWOj03q0Bz6<a?J3E
zyBWUGv{4p?cDaiEAx11LEGa1V#u2ctp%PNg(g2vz24zJX7TD<0(P0b~a`!K5;HI_w
z7EBfe;C+>J4T0E;))z%2)Dn~#*xzsAt`Nrk3PX(w1Yg45TtOaeb_AW$!0bMS!Ev0c
zRv;L{h`bA<E!8R~%l6R7N?+gvjUrhW9_|5E7LCFr3Bda{hK;^C{2?3X?q`L2brp+_
z7*tnQ4t`>>sx*H#*B!QCqGVEZ&;YQ81y~~T5?kgm<ZuV;Y}u`Xzjt|V<1EWWZX**^
zI8_=b$CKhZ&|6zcY93GeX^j@sH*6tXmB7@-YL(J{N3Qh0aQ;|h&-Fj5<`^BfHn2IX
ziGC>eJHVHTnWAc++rEY{Tz1r_Y=&nB=s(rQgV3VW_(~p9GDBT%1Ytwo3zU*0#7F`u
zARlHImfD~A^h;78)wDpUAhjitc$af<cf!yLm27B6G!QGlhY!`zk&(e}{O07ozrZ(2
z$P9Pi`4ONG?ZwYagG{lYFfOz_D{jN@J`u8^4bvrCM5B{bP*cEo%hugpe0}3U>nGng
zLw+S?;aGQb^0VT$O<u7r^lnePv*WWS6<=Mt9>nUZge@c31%Nh60){dQ-*iG6V3sKl
z>MI0uo4~98`WlXeYU!)tV57X84$r|(x9(AsKeKOF!t(a?!bp=G!luL6>7Y(X#gZ?v
zyS~rxTwam#^YbH@;rJ(OIByXD<-*0Maj>Mjf@;1&M7r?)3o`k&v5@60?)#|J96-{_
zERC<HS06$l<>5|OK24fS3+D%2&DB6W6wudFy^CNCre~-d*WxeyN=`c~+&5{<CUO^v
zOwJ)noKn_>jgq4!s7G{6rW+o^u5rHGzA?h7whTN8B6$+3EHULm317rv325ArMMvzs
zfVCR2oYO*ZF-{m*rV%zBmd+6-EEs?ap7bl}T)^rWu|6jINCWV0U7VTEfJWhZ47Z8o
zn*5Z_YEpc(GOV}iVy3`3mzMJ<%d0z3bzt>n4u?u?I#?Idc`ATa1@IOR1X(gi6%4ix
z5ZFS@V^LO%r!C0buloH7Tpx4HY>=rF%Bev#3XvNe&t?`}Lxcm}C$g*OQRGweAFmyQ
zC>@lKQ-t)1!mr~H-4L$fBV?45LPRp#xL9mmozG&>k*?j6)~BCl6lQ?9bD30c_;(K5
zm1%oS`n8$HI$k!|Qs#`Jjjp}ROfHeBZGV?)Mw4&koWQbhRWUPN^%dJExt_NN9evga
zM}KwcEK03m$7IVc>0CZ<6x0YOZOA+cpdl~eJ|J&lb=SYZ>@rd1N%Sn;K$%5u0_GRL
zPoA#?kKX!9&Iw9XA<Qx7W5zZ|&KU3_G=cnui-Y5C8bf_PX~%__uVdQ4PMk=Ks2F8{
z@W+l)4nbZdCN)5*$|R2h(QSFgZ`9qZkNz0(!9-&AE$S|Nk6-?ftr7tfct*eEP0KH@
zwLtEt;9IiW6d$dsxYpZEJDGXj@Xut+&#VMxJq-Onnu6T?HB?B^`k*q~2JIr^_Sl<>
z?N_<)G<P3YP}rvTCP}53ly?NeRtlLA&PmAev?{Kwl6b&_D}yye94tvpEZSlE(fa(n
z&!G-`=Q}Q(&J79Byolx08|#%lC*}^cQ2<|!3!abJM6ZIY*C<Hl0zf^qSdcC-NjCNP
z8r~3LMq~P)L#!+p+2^4S+=uo>AL#jn?VvO)Nd<z9yqJM%N$~OTD5C>ua8l2~4}>Lz
zgF`IM5=k1K%Wp=V_U_rEw4|jfc=^U*xtfKRvXcrSs)YKN%*G7MHPYRS+kkjYQc*ML
zu|B^S;U0q{EuY@T)NR2`%3bkNoXjNAPhZl()9l!}(>~giV7CYptx)nkE(i|`O9BKd
zraTUMVsv&JXF;CZ$M>OiwpWn?Z|;(f5Z)^1CCd<dMlJpyf}|rAnq#aJrSLssZAohI
zpqD5_$0JN*XLOS%h*inVCT@2YLdmC<jHMH0i{#k%@-oikr(NwVI&83g+cvlX>R-b^
zL3EM<oXzo29FQhw7H|OOvCWol=b7fYB3M@7W&mih17z;lq6kxkD#ii9xRr87_T8xT
zEqh>gVw5_bX`iOFrJJnfz<q3N&&hkSaimLibNBy6{=Y?9|Ne{mmvk#rPJ;E5O@cs>
zDfO=lFs(?s63pS+y=;r_?&@yKkTV(z051ge%8pk0Oyy}I-AcqHJM|mZZ1A7H{G(9E
zc=0~$hXT#63c(##mjtpSWM8^J1VOJE$ppYRW5eNi>A8a6e>iKJI?k*6D+k#$jzKhP
zsjNXqnt%#c?Y2V5b{!*1g|yHk&nhbrm7#D-n)Cb;>ps<lf+!Qp8wus56YOD`@3&xv
z!d~14z+>3(TO3mE@hwHJhwLyA_{cKhkIPy?rl@W@65bR%|NmW!te-!#=0b!<r<v^f
zKnvkCTG{$+`C$$Cz)!s!VE+x>wY_(ljVi!}A=Yh-^AWpr7yl$Oh7PJkgW6zme*;7S
zs~)8%0|IOh4+AH(E?YLK-t=By2nR$ns#Y|+HFuW<lyO3yf3|92dht1tqo1Oi(+e_M
z&Spd$h9#!#+Bct3D(J{&q)p{vxHt8V&$_0%%=BY<P6W*rLI_-owC=86Sm`)|>9D}6
zrX(7RF9|Yl)}SDMq`R66*QAM@V27(P@V1o5Lh|L6%*GYhMAPh06llnj6!CBR(3NWd
zT$G22j&5E|A0q`0(#}x(^*((^JFHo)Dq!Q$xU$d9#K2?b-%E3ZJ6q|&xyl;lpC4ez
zJ?JcW=dxDjQy{tmF>&hEOLcXu{yT3o*m^(21_uP;X`xAo{1;r@vr1<DT5N8iC3S33
z7LDMI6EBd0+<$WY`|*Nqr2fH2*>%9FV_3uL1JAcxo^bj)j<5NquIm;@d1VJ8VRLev
zS1Mo{v7`(P7=lM~)el2bnd~X<m~n&nsDM=gk_RQWxQy>BQfCaxO3-mnx=3R+*yG#(
zto_GbhQe7lbGjH)>*gSZR0=5jw>E;Nnsr;kBv?fIQ)J2LF?H7rputI^SpYxm3o{j)
zL)nB8_X^YPIpDfdB~+HK=(eZjW|>FlEJGOIZEOf;p#`jO@u+Ai;H%M(4)6PV)8GYD
zsK|A}ycf;BksA8>FWmg7=NnSk#BhgX0n%tADF${x<na;wkyKZhA&VQrKunS}rp6Fr
z1oe`S_NX_qejUn7H)ILyD6?MT{xEYF0TAkgh{@h(lD6@D<YkR6AQoF?5GzfTIyIB)
z%T)c)1!rr(K-S@$<*%uxq4ddE)KZuc2x=@a-B0)U_HO&tj{E7o-kY8+u0Ysx*PmEe
zpH*ZD28sz6u<8se;#7r+LseuJp73JkDdNCm%=XvNnW@3}voF*|?mmiu@-}0TOfXpW
zmUPZ^rO(9-tmQT1_VW0orV-Ia;hCt2xafKnxqTF-_iGC;t%si1Sl^;zs&O5K4f|wc
z$8ZzWWmEHA;i30ROxkMddbl6NVkKROiU8Baix;)MrjNK!@UKJB*ZCvDp~OEsd2avy
z{evBBJAxdBW1J%!#(RnjGa9a=F=@sr%OVv6wq1lBj)6l8#!Dr`z452MjYz*CHsc^$
z<g!W^BQw(fM7V(7rww7~7F7u$OKBKb)JKP2C|_=B)+Q-M+?L=r!hY=ni}NN#(jgr<
zB{C8PlhAwQqM|kw?@7te!E!I!gfN=YB<(){2-~=QC666nLbG5LWnDY!T??vgTH_!Z
z2LWkc;t!+OsE5F$iHc_)Ka4(hwj2*+VWP9ePCLz>pDWkW&_s0a7gU`U`wktNa4E&?
zo+pMQtZhsID3T(oB@a!Jc4n}6fzp*L{8rI56KEN%h2NUg*Y=xvn6da0k%BG_3l@7`
zeW3zT;|=$J_|Ud?3uP%uZYbT`X`uSM1V;P{yY2j2gbYN4x<d2MH(PtGV<gOFJ}QK5
zSZ%1U6`KNdWIhJ>Md+0y=u8RN$7PkS&C+B#uR?Uc|M1~SyI%NcIl|GSsO~pMSgMox
z7KdVJR~E;A8w9MDH2eht+5uvaVX?ePlxGAG0#$P!kb2z=75$M!%d;WCG_&7JjTZo3
zr2vvy!5EySXCcyd8s5xDhZFm1N}qAq>q!XSW<6?C7cJh>USSA;-IjHK5~M+?D41!K
zo!6qrEq-*{Z3nGV@?H#?u%(Kou-!<o{^kwJ!T?8~75e3tJbWK>gk)=*@zSMUdvmSF
z4<39(&&u7wd3mW72GXHMe@3R|R7shXKhi!TkxQQ%y;|(x_2@o+@px-*>=C3fFC|OS
z%=+4ng;aQHoR*U6o>_sKV=g7v#p^e~G@x~gqr(p*^b<oBfRS@G1v!BnL7hMebEMze
zwV@E_zSKO@nVZ}}ne%O1Wd+e;)T+XTHv2KQoo4`*H$#Ia&as``xec7?q~j-lqdz}u
z+q3iMb1(8O_g=BO^Z#i9It4s2M@(IvDpn3B#(V_|WS_HgI2lxin)T?YLIQ!N+b}rY
zGDsdNPHmCK6yR^M4OuL*h4AY0jOsN_7z13-F}^~lnP&WqZn&%CVooAFegh}moLJ=$
zO5v+}^f6aF5*YBtbkzBhwHv@&QgE8baQk!zb5_>SAw~cF9)5y1#qVQ-a4Pa|FXetK
zd<&p}T=j^H2tBR#C-Cxi-~}bU?I>1-yf-|;=dg-41w-*8aL(66&*;a@lS)uLNQ=uT
zU)z45)4#@bu)7ZYUJ7kGFIzVG13*W}U>s=+$WY+7Z+FoRDL}K^f}4j%ElBF1Lfc;8
zxIfSd=!}Hn3@qB?5DdibIZ9HJK8a}fc*Rh(xcwoy-pEn70FIVm+f6D)Sa{j$@km<X
znL~Oi!ag=PH<S4Y^0<@eY(RZrH9A?|c@udxU9`_b9n-@}sBB+Cr9fy6QbcQ@CELcp
za0WXe$h;`T4a0|W-FuwqR(=m2+!fElQZzRVA<YvAHQKvF&v_K$@astQxb^91UGz4W
z8@-jdN>G{-8nef}qKNq55R`Hr7d3@c0H_nJOJ?NTF#AvHS#X-z3waQUPz$AG2c)sw
z-j8d;cr1zy$p((3M-e6}c@_C6*GC`|Zw1&k`E2o{-0oKZ7N}nXB`vMKLRgC}4l&U(
zeJCf=K$VA$Q>+|_!M=UT4D9yqA}@5F2_HD5tDtq8TBmK-J=0~8HHz7o+(XB{qyCB%
zvyK1ih*nrBD;o<UQRHSP{56RgdR0s*3_fZ@PBp{)ymhGL9RJ<nsz6EN&W>*jDSz!5
z+4mjVUu!^l>I4oQ)vIwDRbB&BG=j6%ip)HDXMhAkNPg|{i(4}6A}Tu|(7D<<1_7#l
z#=F*ul^Bp*0CET-P?Xb`M*KP;+#5q#HE%8n-=rU5z)?6C9Rk`9Hv)<Pu}lU@mjRUn
z8Bn4O1%cjj(cm>;0S8O^yIB)5LF0(EP6+7w=IGrC)O_rs-)XpSpm@6uw+hnY!8}4X
zq&w1(1)Z%b{vM$dG4nw>O*D%d0M4Nn3&9tJ;Li-9fh37Qxn-To4sFbm{Ws`9A{^B1
zcc60!+by8jjO;Ky|63>8r<;LdC-;|#b?mT8S={$tq?Pb7qtSu>{`Oo^X?mo-(D@B$
z%t6PkV+`BQql#NVT9qiHVI;{MgB4-i+VOblpSr3*07hB0{!4g3RNMJCv!3^;AA!Uc
ztWr!yg1rgB$x2R*DsUL(e*FY46t+8KC}@K+PaJW<;4XKrE>SV)VDMm4z&-L2U2m$Y
z0_I!})WWZgE@HWObCFJrWkkWYf+Hi{6_+&*)4R}Sf*^?Yy2v$${{Bl=fe-aKnl&`A
zR}T+n5frrL&Q4f+S{J_RB5<1LeAA`Qq*EtO4AgGqxWV&<)QO#L=iVzI0g_;rjFe1}
zGEBdd{PPO5xB}MkBGeHHYot>uaHTgG&BeHKNgLc@h&BLy#?=B>xc94)$Ae2W7AQGI
zIzSE3$E4VS6$`;i$g@I%a`E)A_mQ0*jEw1nd4JOJGe2z$`f1-5TnOP4+8Y3_+xjYn
zFdcGu5an%y?J`c8M3@b%MJCsY+1Y3*`j?(|8NYp;w=WAIsF{1fS)Vst1qmAfz?Zyr
zkn6f(WT^HkjkMXHut$+bd5qm&c>q+N7{fb`E~8JR2Q4xLkE9JsJN7ts93&MG@dqGv
zr(Gv}G*}QhVp?Xf^7{>qV!xUT1^??7J6KGL@_6IMB}hfcZoN_kNW_lY+dAeFRQ^{7
zIj_BZz>gWGuaL=wjqBl~ZUAm00>cEDy(iHq8(s<mV7hJ>>Dfpj0`g|5<3ZV5NNUrf
zm-(cndmKtJtD1oPE3Pkk@}P>;%s-Ta>@ZK#rDxA+!Bp-zC<q4wR7JcUY4BjA986Y@
zGNCr2Q!Kh^3TF;^@HLISQEj#*5Ri|$;d<nI*i5~a*cH2m3atW*l>!E3FftxJDbZhJ
z0H4&rj&4HGX2v~xHWF-v?Tc6A*ed`=7y%>;TUAn2v|r(wnI{0LxPAv|s9P4G;nj^h
zBv)}Z0D|}L^I~48c?*`T7Nt`_@;3Iy@d0INnw6IQ_IB$AZ@eLkmX{HPYyH;?Ku;K9
zl%<M*<sAU<mI6q^%+SRe7TQ84NgKa|mAJuo;>7zKd3T=DD3~=gU0dzj6nQyeTaT})
zw3KbrCzDKp&0_4_Ps^Kd%s(`)_sAb@n#mf>9(lw2K+>z>^ODy~zxVrxt-7F+jSedZ
zYN>4W0}v~7j(B_?jXn1gc`MZf(OBrr7s`dMg%y>Q2r;HG0v#*PyNM_+Ub$7%)_$n*
zk7O5)+&Wu(;HAppw6zF;JuGDMWHbi?B_Yx;P70;~b;+0=Zg-uBK1+hf0F;6e-bmOu
zOI7G}8?s>l_~uF%Y`oIr%|(o`?3?a(s9+L(Q=(sn1YA(~G`}ul)0#S(bf|-gkuh(f
zn4)<}fU!ah<Y*J1khpbutXb|!S|veZ)D2Z+fk!#Wn8%}3f=?7Rfbsv5yA;}DyGH2u
z4g%wG_eDvj2l`~mW~Hb1H>276J|VsGCx&eUr#-QFVS{%__zR$bLPy@A`;7?*!dEFd
zessp;t2R0?z{yxYByRWKtQ1gQ$D&N;h76(qtMHU^c;cQuHH3n+$t(s*pA~^hsOOT`
zWBZ=-Z029A<my+}qxp?|2c1gKv!#xgM|Y`8I*u7rGd1W>G(4Ix9uYpiuS(>4#dGT?
z%a4Dc9}}FH9epsyOeXZNQn*0N&S3OQw@L))X%%V}vNtT*ak^-jU3ay<aj6x6<JY|2
zwi{#!y9(g$&CFa>^T{Ha@QpIUTMAn(8qX|_UKE7SV+#ewY}PP>Gxo0}Yt3n%=xOS+
zEy4ZR1!zgz;vlkuOcsSEB-rSX6Y_ilqESJnmA35mG>coB5%gV9czv~t06i_aUI3=1
zo~#2K6)6YOeEiIr+R^AI#*Aoj24`K>vLv9nQiJzMvLn%iACh$9zJJ`9g*5Mk4oJW<
zh9uh-l5^W_5=BvN4bJMo62le)Diw=PIaS}LZaTwbz-qjZD|w|8v<YP&pMJM#d8p<R
z!ztxRW2*kug#)?n&qxRe$4c{Xif2Vnd34~{{JVS|+?jT#WNNuI*ZD|{p6>s{d(~H|
z^R4hpxJ>vxdejCbUMn=LL}!BlN3c(a#R+XR7FUT&7R2KdH<D5Pi0`_D-sW)fsrM+Y
zj<kX0A*vjxC-?vU`wOCU5b(i-64-kkLi<jB@kM8v3`(o|n-iRB%6%CR#zP&=NuJEF
zT~NfhL|ZK&*G0;BQj=8TG6ABcLTG}yx2LG=20jS5W0eIlWhTZ2Ew5Q0OU*v$^YwMd
z$gB|Pn(7(XFaI78GOA)+hrwmDV~1Wc>>#R5zm;#UtDs8p!;U`mfT_d>BExgA-F4rQ
zBUz-?=3nc@S)GTU5}~ZGfx&V>x_f2WnX?1N))*Bf1utqE3$OrzjG|MFOJPwkXZb3i
zJ4hr@kbcrm)q1|b?le%a(S%3MDqH|hVS$2w1<|Io>naRngn)j)L91?DVEIq{E9Oz8
zlCo;PW^pa1sH%g^IpT&znYfgtA_b5x%mECQi<Hz2R4T&goM}qGY*PdFKz69(S@+o$
zq3Q-6qJr`_H;`R=tf>s3A{q0dSJdr4v0Lify}{S%+7XnbF&IT;`8g<KPW>b9Xfi%u
zyq}1&N@uFrz{W?SH%bDWlY)vEQx8q>k|xtAm>5z-b>+~dNk#+ED?k%7^rjCt>=6t9
zf3}axfP_O-OxBaU!u2`>q82-ikUb?lQ8Ex<75Wt+)c%+n>uq;W!wz!1%P${X_%TF<
zk#TwQ#qqoM@7p(UqX2fMN!r$!9%5gIOtA~F77C`9ARNee2Kj8b7QVupS0<AOptKK<
z!?yjC5L>$jVQa}8ShX#%vNBRlV$_oqmpkx#@d0_b@#}EE5MY!e|0vkwTjE02Cb-HS
z#5?E;3UhUo0SlxK$639aP2{zMAN$3Ln<9}fL1!`U%Vte99FDNiO2x?_OcGFBEKN=X
zv0&maTIVZVw-atM&yIg>YO{Zh(@$jbn3wPW&Pb3L@}dBaYXad$lg(3)KKjQYoX4cQ
z8FFZ#Av6Yi+c3V4Rl=_jSEzV5X!0}gn&5{EkWr6FS5QZZ0?}0nq6|%yw8BI=yh4~L
zl@W3p7uHc9tB{{aXw9sGy<qH!kPQ6w-Bv@X{YQ>W3j_z2m*9iR%ONpkp#VtRta9l)
zEf}w)kjt<*Yr?NgFj-UUt)F9HMyLSDw_TC1^d(#ZN6=-c!`Y6ISh&ICPz--WGkTat
zq@yv;y=t$nB`OtaKk4s|hQLa&OVwVU&D;rafOjjCoqW$*R`512fPG@iOa#(+eEH^v
zZgXf|HF8utP@h&}YT)wq5usSo89xWidj~O^_)h~$nas-|Ix_sC7a8l!FWtw8%XzQM
zS0-3loI~A9AWp*CBc3RLF!+c}O72{;BZ)I;3w6V6-aq(#1$0fw)2DxoDo8D&LmtEb
zs|om)++y?>24oyA-)U0kN;9Y)G!CbArwyphQ?9~EL?c0jm0*Q0VOP}M+5l95u8+SS
zt2m2U@MA<I0Bl<w0lXDHiF$yT5wt?^q#V}1+Y*=*!*bY0S$$|62k>ll(+PV;0QSTM
zrdrqy!-fcT1yiL<`e@eEklYkhR60<hf_vab2*c4FMGHl2Il8+HtXtD?ZOBcB{zL&-
z%+nLtMZMVqwQ*aiyE6zu9fa&_aO=cl%RnV+fwG<OU_@Cr^#_gcv1(`#4_RevCBA%l
z8ht?ZbTA*8^g!ENTD|0&MeXA4#kA(*ORQ9Llw#IBY<0@uB|1>Ak_MqVC8YN><IG*n
zscu^RhYx2n1Q#ah9pRP#TWjp=`R^ym4<VCvElQmLq|O3J&ZXQCT=n<!6TpCiVVV-E
zvWCSA@tp>EUXKvq50Y!6TTPn^9x;hfnD{pT@HO&~wn@(i|Hc)x$$5Ic@J(ig)VOic
zI6Q`%ugI1Cpsyi;qOf;$e#9P;_uo;cIRW(oLl${psd{0yfB*{$znFy`sCYALYzs+>
zos@}_fTI4JyYF1k!Ah?vWqR?NsH~WVg-7IMMM}FDs8Jov9{1@2hYWy|x6V|;b4qR5
zXi`)F5ccByX|icsSN>-4ssxbo49|Tpl{#?>Ue~3J)gt*0e#jYXFu?nK1h_?a_4H*G
zEzsg@1ZdY?MG8$<*9@%lMh!UMxjOgn-$9(Kskql)(JP67OO#rXS_PQTCFj)csYBVU
zT|2O6&ux$m)nvJCOww?OZPS$F9GX#Q9X<=+A!LzHfqEcuEGZLb87s)#nzA=r?gs8a
z!(~`8kF{rvKMPY&nm~)t;&O5>PNQHr+QnWk24v-6fxV>qBBa8PAlaew+Tff70N;>k
z)1EM?ZK$6+jq<`}D8f{#!CDj}GvW05aivZGcJ|Bcq?3xw;7k@U-Jo%@^|lFoTE5y<
zbBoO5C4iD+_m*V`s(omDt7{BX<Jrt;8w|$Wt&JX-G620AfDS`DEbAzxV&^ZvoFzCd
zDubf@X#~)s?p~np9*Y|uiTIznCO(A3#E3@ZKsP|?b{&o>D4!Omj2EAgtuC5+&XoVH
z;KdLn$+F@(lkoWP?PMllK0%6s#<MO}CLikgA=xp8MOD$)$w%9nu11R)b_45S_e=up
zP=%rAo?cV0(&VW#a%4}T)se^8(E6tCp9w$7nNRRg3P4qvY$jgWi`bj*7$ZVkw7j3(
z@h_6l4U?pXNu^F-Ll*oH0fhA1#dRQt&H^`?zP|l+6sKh2n1-+IpnaxVg72j`Irw{+
ziA<H$g;=rp@<VACPB{cut4S-DU{3nkiAO893%v?pCYEfuFtvvEo|V8%cHNisO=`==
zFcKsLi0pIIYEWiC)wf!%tO1t7{fq8JNU)AY8StSadwR(!+An}-+4u8~x59}eqJr=O
zDPa~7nny5PNS0gSSg%V*xl{sR&;ibny1X5@(Kz1<xX8|F-9ht+eW{>eth0s#0fnSq
z{(aTwmk?-%$Yy(CV0Zryd+!}p<+*(elc+H<c11-gDmIEBAWFwX6qRD3Nml_8kS@KM
znAkwDP(+%7f^?-<F`|GV5V~{)w=@Ci9lp6X=H7G8?~d=!Z;bDb!yk#jX7BfX-sf3m
zuDRyYDQ~)Sk5uttK2&b3-p(Ij1~7?`77$4sa}t{*e72D|A7$;#<ddy@0fx~-c81Al
zgpu!^`IH9a={;kZ-qR^*4$)ukaHWDwy*kR|WW3{1`z7gvGY@mWK7!At;13yGB#ONg
zkNIEN1H3kabKWRvAJ@yD54DJ_-n7QO7<4;>$O^tbfH8;#*&EVP5_Q(#+&QT-E7px0
zseRfEJ<P;mNyzS#NUzNUxR!zzJc7t;Q=cJmrnrafQTDs@g+j<A+bdQjq2^PCxUu78
zZ6ERjaDY@1-_}6puLcwN(rXiuNn`GIICLvC5itUEK!8az0?_<%0?=u^+ozh$)5yM&
z_Bp2XQ3MLQqA9{aV)bhG%pwRmrl3%=FBjP)n)H8@XZ*zqh>l7M)M0{Bg*Nu8c)}XY
zV21HGgJXg!Vd6!jqOz=V^1r}n?TbGLsKE!*Bw$=wcB<P^XC&65P*h7lC##{Aa^Vq@
z#x8`Hdv$l9#sJzr;bR#3chP(ds7o!j*u}tT>Jdj^MWKF^H!f#4O_^rguV~$bD~Q3|
zHtik^8$3ZGHYnz1Hu3l>9lE);0$x86gEU?ZvP0Ha75Q-r_^auyX6G~QIDxgAw!OXq
z@a-AL9L!%22CP;CZg*XG+&LBqQ5Lf6WTiZ$?~Q?Pb9wR&e560$GxtLglu~49?#x9K
zkEpH(L?nuHi22$-a$EybSVusa!S9ra?-}3v^HJb`_h~$2JcRE6#U<mE8fLocz1fS+
zKO)Abk2r*s&P_o5o`g;)sc5-0$FjXoS^}~>;3sL`6)a<m7B0MiLe)lax)~#JxWGw}
zVn(QJUu%zohS{I{R9(=A!7OO#`DqA+3Z5W@@5<c4=FJ}253C(P=Y-tCt#~sBxLOj(
zK;qm9`hc}d*rKk`Nz^qc8!4F*!z2%y@i2^hTdN>FP<rZ1@+3aTY(jdsZdM8vi=ege
zw%3MqtXo}&Zo#`Z>(wP7fXEnegYthJN>M2=85B9z6&uN;br^>lR%~Yp1IVi=EH%lZ
zbG8n>jd%9D#IV&vs7N7kBliGaC65fY5&#r`vUSJyWXEx2m3D?X<>nAeduu&xn2oAr
zI2OAL5~^zO$>#AYDHl%Sq#OY)ZdB8YVTLj(nUrB{zbw1*=RlF1kqoB*V?idcwJ;<8
zS!;=CY=`vthbx&4SHBxyV7&LkL11{Qko(@|B6}Ullf%;Jwz6k`Ln56p^Y&E9@A57g
z{g^`>TdER3ygdb&6475Mih>ijE97u++$-mVZ$}WQnd1Ks5Z>E=dH)Ed<~L9l18%PY
zj)EEvT;U+m7;{!g7V;tb!;=poW)%_Jz&fBz8x-xHwYQzC#QTA~R$-Jw3gl%_y8VSd
z{O!G+Xg=aWDD)Je1TR1Z;NubC#y7GK=S;>yj{g_pCPMB06iyPVPLCm*6$2qiIwyS@
z?aQOQIf+o(@ItasAia#c>)4@Ds%#>J1kLW$3=j$l%*i3y_53|l=K;Z74Kh|Fo@H*=
z{cd{3_mlN)lRDtc$Z5n1JKSsNoIx2(yTi3s*!X-$PBvLR08yBWT6l@gp{x+NI5>`)
zL<>e}k^oL#H|V*?jvcFnbEnFOpMb6wp5X~BfEtqY#zz0g2>&wJ%bsGVqoyG3)2ykH
zWh3CjCUO`}q;Ur0F{lhcZ%O5tvNBm`u#)3ZOpAqaxas5L5L=mS!#EaD1<pLNO&c5E
z9!2LA#hW2gPN&dqT7ekb4qXfsQKH)yS|r4cM+&$WXaMm0Sj6VuJ%Q<dz3^h4S60YS
z(VYsgW_Yv|RT1YJh$0yT^mv^_5u_T>ak_*In#6*Z^NeEP$Sz<7u}S(QBGKEDs}1K>
zL&CG9cK|Dk3%6%41jBL{<_UmoPE;SNP62j^g>_4_JQfJ_wF2uTiaH8;9@xD~^l|{`
z5Q28uF7$o^1*Z*)Kj|2?d-p}>3o2Xy$Umsni)aIDgyVaRQ+@%fN=n@vc@cD<Xpupn
zzm|<H8pDFsFh&!hm^y4v^9u{ZeNYh%sB#m1Es(Gm%|MI*itK?K_}#w!VLVhHL!^z0
zeTiU6I61cAQLsjaO+A3yID<<<7fRZ`M0G-EB=yPB`$MWYwyM+<3n@0k`93q@79JwO
zAtYvn(l*L+2jOUJH^!7J2NNbpJDpEG?r((esW`P%tHVi*(a(DUiId)w1}=~iV}IU{
z+EWbtMJHYs1B}+7=+nCuHo#V@2eJ^6i(BL&Tq6P^LG*Bd-Pn1AkHFjR%-^$3H*=qJ
z45WJ~cSrRu1rUNcnmYj@U{gh5Ry7Y$ejcco(MLl+*&>s%j>&aOL&z16VD)I<8xi7d
z2JaLX%~`)&OBG_t6)MN285<HrQ0amHISI=LE=fF^?Z6C%fJ)(WHwPbG!0MHL%B{!{
z>hLsaPEQx;TU77Db*x>zIvoC<2}1!&f&xd7K`u82aCA+b-}!m4mUKfvB@$``7!cLC
zkyFSlPy~V-$!b?)2f}NtLhU&d?k2n*!c_K?U<{<Vj9Cx{g$+=J2jYtcLBuo&qiI2b
z$p%3|bQ6fzEwyP+_(Is;wO9g1*<Dxa%9-6@p^Tt?7)8cTln<*Byrkkp6{c>+X+gE#
z`!sBlKh-+5t#4^6s(@DoEvBYXdz(RkLETg@BrpR6F`QE+lv-_H*JAS(Lkt4$2M{+k
zQ6b_X<bzsvdV?x7-5^!xeKuHRsxEJqF=EMgTc0LwgYaUNQ^sQv3)RaiySII}ve2m)
zHG*cmW?thvTMr>X6Er~$jAD-zr=jy4&?KV8M;!@_FL`Rz?9I~r<)&@v-&eI98&4#=
z%l67F?c`}V(QJ7z_v6(~D7}jZ48<kLBbJC1YFW0w>2|yO&kCDj<eE2&FhlfR)1-t#
zdkBF!^+don(PSJRJZGe%#5tzy7*dJon5pfN6M{o^hOx(f5hNBopcvbf;{^o;ycn~9
z`0;C}t1F6DHR+c6bX))|WCIe50agW5aumestf4YV4#v;Rm)mEpt2eRj0XVaqUF;#u
zf(e`rZCQdCE^#FR0kQ~~iovrr1r2}>c`C%@2pb?j0+fo8VXuE{jC`hY$Z>g2i+{KP
z67@IRBf`UjTVk7AS05#;zg<R6InNkrB5?qyM}(X-iJ9EJcMlYrCy-yJ7!&#7rDJK8
zh}XAcmY+@kdxm=MvAcxj!LO4X7tvS{N3q-8L&+V}{LTTRRl^Xv>KHXXZAJo4YL)Tr
zD$6J7bWyneE_e@z2kwNHYNZsZuYMXQ<a*v-A}oMULz${}__bx0m*jxeH`<)^#HV`=
z5U8Wb9;jFk{1!wGP67o(6#j&K05&H1fF3<d)YX;R?*(@Y29~$K`x5MdhqVz6?$FnG
ziqeQFeoZhHl71+<#39GUxG?~KHnbtpQr0^v{rf9d!7-q`0+L3MAM0o=8^5eE9UQai
z8Hp^5FTeE+8{A{Pi2mTEBKgfCW!&XbBh@Mbi5h)NbN~3=5>#Y?@+)W}X`L0&`fUKE
zmrN;8K=-{1ILOA32Y$Z~Mc-tOeAwZ3c!mLoft*nUyc>`>)VJuP8SCPT!21E<e_mj#
zlVKK~w(?_+nV1Zp<iXPm60A6#A`xc_BnDCF<4|$?8sh1XgHrb)G?3@{ITdB{Y<B1L
zBX_nIpx!vGD;2P$Z6)K+8U!d7W{#RNE9#P#Cx`g@e(K-)h$Vbk4VY)I_FOplr!hz{
zg4N5Gq`|z2xv3M%lIQuy_t%~MjF|=Q2ld$39_yDmJw+r%_@{1Y2a%>6DikRS-W$n=
z5^<jAwdvkXE(XS!Fr6+vGhC&7e5F2OYD-4F)xocMz(u#Cu}Cd;QYv5J?C0(cKP+C{
zgQ)HNF76;S5_Q`Y#Zuye^DuDmhAT<v3nUkKp>4NaaWBi|X=BXv5{Nn;D7@WA;d>7Z
zzp6n_n4|>x-gF3XGp?<ks0Ez?L0Z|xs>Mq;e5kmM(c)_trabyJd>LzB1dhQj{w(`m
zlBqHdk^xnNNMla=o5v5G%?s#>Stg5qW5JxSzjGa1@JU7j<FRimP=o<Oj8V`Fh;0)4
zbH(;w^)Mo)64b_Ir8#6v=F3#B@}V-|3vl{mMDW$sS4A}npTYi-9{==-jjh<H&|Y!#
zHFpNEN~d4e_eYl7+QgE9MKKJTMabO^{cQJLs|Fk3b(5E^;sU5jYP9aWcPw+!cvv5a
zSwdaNCdUq%u&@+uN<jyzs^jlj-?2R06DVeVF3kia*=-RSSJ!?<8t%_uK};eg4;157
z0l(!4uDrdt6Vs*-2sooKuLfA{Viokk;1H)zZ2j~Ei1lvm!(l+AyZp5+22^e=38mI0
z<TA=t0BT}8LS4#^Z|A+Ar2Z-Y)hmjyc`?D{`1ZjNG`ymbY<uG{oV~(tzo|N~JbZ>a
z!m+fTDFn78KP~I`qtn}((i`xuygPOjtavgI4~uGwGDUr)52PyKl?D&qz&%N(Jko2#
zHr)9X6R?-ugXo<mY-Z6uqUP5jMWriE#kq6<A-X=&k6*Krap~c+GdT_{EH)?6#Y(c5
zxC*|;92_+~h{Nu?SO<^Q7OPNj4))1J^xj2P2p+mG4edr3URBg_>PY5NP<w{DmLj~n
z-Gfo`VTjnrRBpKL<mNsIER4e6kjVYWGEw0eQi*0^$&AK!a7GOvsrg|Gh^UC#V^QRg
z^;_z+#mv>M(LB8in8m1yiE9tG^4oP)IJ%tx(MDDN=UoP&k46_#E%ltobl_j8EF<f;
z#OcwV))FYhr7z7dTGvP#Bp?^K^RnKantz5p6%p4;c50s-*9B4-{X5CVP|?O`JFg=t
z5L{Y{3UxU2#3rB1Eh-0vJlOjz)<6`^1G4@_4$$=<#*$8DZ(zABb@+!{W-6wpF|eOe
zD1IKxTD_P=QQQUm`g2foo)!3>g~k2QrtVbFeX1Fr%<Tr+)&u1AU;59zT(M**w+SJF
z8IsIdfgf&n(DmWW&}2G~^^aa|LZ1nM?Y91NYcC-DjZFF(Pc{d@v9$q4B|ltAhcX<8
zl4Z%7KF0Tdt_Vr`?bifUY9vtE-^P0L(e4ji!SeXE2I@6hV5!`_?)gEy3`3m<17P&U
zjh^tH`a^rC6_@!_yxT*RJ8W)V0|wPwSo!c_xZWa`m#ov@fB$C_y4t377q0z&{=N7i
zD&2bMz?PRz3t0}#R-!yXJyfl44tyuSi5CWfq!17Jt2^FuX2mBjoXq~0`OWkHJaM~%
z=YLFT&p8nFKPIw&{XQfKcJSZ6UjBRu!|wj$OAt-a|M4T1|Gpt}MgF^n%w_mLZ=1CP
z4&Slp)38b~4QF(h;-xPuLCM?QQI5JOm=y0KD)BSNK5(FLmYdMQf98;!+&IoYcDmu0
z>SfmteaCX(mpdT-_iug35}~qh$t{F><Jmvr1()j(14ef)K=7#XY54MH{OGne{Q-;B
z?jMKo$^Y^7a5Mh?KYqsI_?vS5fBx*lBJ!>O{&jgh(|G**5A}XT?cl4!mdn99$U)H@
zI^ZD0d_~=TQfx#Fs9Z%NdvHVJ(Q+&&o*`R5Oe`1&p(VMLW1|_q^$iP)549tqq+|b}
z7VvT|sIcz5tC(+84+1$68)3c%6D>W*vlm|G_?AXYGuM;lrRa`5d!{xUAXl$G`5nsy
zCcZsJ+)Jzd_4J7oC(=xSoV{*lzT3kCZ{3aO-}ql{<jki3ngJ}w|9K1lUuXF~%sqGL
zs89>&n0?m2@Rii|t6R?A{^d^i<ukt&6mLGX*6-Y*mBHTXo>7iksSST|Mm5xOrUn~m
zcW6CdP^%uBV(MA_=*OQ=9{!nk>-DWme*EL))!Sn}v!CpWTG(Dl{`AT&o}s1G*O3_}
zSs#>;d-r%K{4MkA_GKCe`OnvZ?|=Q*L-*5<|9afEZ6+`M&)0wd@V|@m|FUzc6S|mo
zKBfvT!~lo}LksFQE#<C#07tJQz|GW|YD_U%10X%_Z&!GRJ>W*{4}bdfsg4<5HxpP+
z(w<F;*+2jMa|Gsm|2jl(XQ-u)sQ0MS3i{@RZbIhg2N{5+Vi<8~$Cd%nUv)PU$d|u;
z0;A=+Sz>OQR@o$|lbm3XB>E@Q);;BT8nbIzSi~OZ3>NcMAiCt0jY7`#9MM}*Sy{p$
zUiINXGRQ&%6#9i#A-X00WVpS#<x29C)laX@sIO$c1H0`zp$Zh3d1Z}}TYQg`!=B_V
zHCWV%s=8kojRE?NLc0%_KkUDP|G(TqX{efe0UFhG8OHd*Z|{02iv4int{TWir>YL(
z8y3|l;6U;yOGm`G^8;EAI1RAOZ}<n6%jKScUJoO&z?{vDW0N3@e0&bP8<PNH?#|Mm
zvVMIQM1kYwgUX<4gnP`*s@3L6SS(z~a(RI+P=a8qS3jxMdI0*Ni)DHF-NE^_*alcN
zq1gFnoc5P_Y78@j#B6$(1xMA7s^uORWWMwwa!{Zq8^Fsez0ivC@@uS*NW&Ty7MVn3
zqiYu}TJ#)59<Nbg*Dm~8DZ3)1=^?Fle6Pg(tS!-iql7S)$!p9`n~Q~IdJc0{3_x08
z>+iAKk5^hfT*VA<mum(tx53PZLKK?<D?(XVzB&bT)iq-lmUXSBu~EuSy!OB;rk#4Z
zg|r7w=DE*!jUZ<@Li8+y87$}-s|J&(Lty#{*Ct{WDnKpq8llLcXyGH~Be@Mgx?POn
zt0G)muKg6ZZHyj{&C00Nd5sKZF*nG{pP0t)T8iQ#3E&B9^e3R#R*6hx`MyBsW>7nN
z$GbK19^BEx+p+|1W7<acN@WZnz-%SWHW?6nymE4KmMSh}VObo8VIOh;<X4Ze?Ynpy
z^NxiVtfA#G^k<r1t0P4IKWvdPx~?0+k`{7C1Kin-DaU3qx$IB>n18H#m1pDuy&FI+
zpX~v%f#u3a_0f8(A&BOlkVnYySy+S>=AY>&k{bbHs@63_S00Y!lPwt0*>V|xYd67|
z&#9b}zqOH-Aeaww^p}C$^Y<`_qri%R5a*wax%_O0cu!khT^(3TTIE!3AEhA64K<Ow
zZx@xuov|EH3Ig{X11IucO$`nWwWBt4D>yiq_dYVU<pNOQxemxy%~2m^x;C^$c>?~>
z_)C>O?}qf3NO~BTuanRUyY=f;tSt_pXDl&Xz;ZwwEo&lxT(myA83$WmJwT?$>cAT2
z4jK309a8UZ`IUo(h4*t@a#_q*&Tku5_g~C|)n-SApGG(mbcS9lDrO*|s~H+kWh3w3
zm57R#Jg9Lcxi5hJJxB3|y2DtN7vWQG7v;MfP#F7ljt)5M5+lss)7y+a!P|*>dtTVy
z0WLF|#v$8pGoQ`Z=~^-e9}IX0(>mubs&M{TpSkxh$C`Z?x*#}!K>8%P8~$^wMV0W0
ziv#}mu@>vi!lRJ_2cHcazGg2Q-ub{2sc#GAV*;ie<_RahvjcO%UQ`iVdT|sIbFq(q
ztSQ#n57Pa-#j*W`xd}d~6c3GdqP-{f>QP+b2gdwC+w+!}rzh8u_TO?Lyqgc|Snna6
z7CrLqSjRdcp%u%3=r9i^;he*VKB<NE=A=scXFwTP3dY&}%;VuoBd&?i&eOxltXRRq
zazqfqtk8qKlhnujw69KW^pwS3>wq<84c2G%@CKH}IjA$e2GnpY9xxLxHk?{1&cn;U
zjPX#OLk!}SnV!RfFk=)t9BFV-WFbB?raV8@%o)JKLua&visO)}R2i_iCSX^&h)Uz<
z?87;IKgg7$j-yt^jX$X2&yHM7SNrX2Txx#*eEasu0p=oo`)>1hOv~`?yUPoiC+^#K
zEUQ=kYwrEuKm7mE;+WPAGoRwiM?A8!ve?eyfGbbJdcQ~gWi2b~8xT`Ap->*(H8B6Q
zi<RcL#Zw!cKv0}slAAaO#bs;s5*&t4Ii^d!-Q<;X(6LIoXs=PnV})Oi{d|KC6$?x8
z@o(=T3~JadPzv54<%~cHh|vHY?KOxGd_Zq0T(n+iC9dSA?DWSK)&KzCpp+8pcy75W
z>Tx#^&G>3Hq9jFtT8-2y2Kw(jJ5jqWo`m@@K+8;x7&B!&Q1fkf#=JhdGi0qpvz-P+
zeFRR10n05A?+9Mg5p1i>4uR8G>DQhDPAxbo&!J`t%7&gND!!^ia=wKqskVcn;#NF~
zJ|zQGLI2lV7k!K4E?-fwx3`}jj*`6vSyy4}%~MryiuWqq=NB#T1JjhVY&J+rPBy@A
zgwDSD;{hkxQ=Hd^Z=o{VP;%Hz+^@fVLC{nH1pXYw6sJzD#-Otb#90^6UC*lg6Z3Cw
zYtCmP2M&Whu|pvHOOj2z8=ONMrdT_H9Vr6TPUa-qf7DSm;Z}wL6hnVqIAoL5ganxH
zaG>DZr`{l{s|6Z%gzL~*C0YgvmGPmlxnLV5P(Lu!QucFs7?<o+eR;Tkp1U=gQC>rI
z@xEgj%u~VHYaHvGoJbCO4M1Hl#7PAJ5W>-nXFYclV@Phmbne=<>pJXuAs&qZoU@t@
zQOdXl1qBh{M8NqyN8sZKs+R$3Kx|!*`Hd(lFZY4ZIfQ2zfuh@U)E)lWjpzspxMLfm
zqNmp~I|rK7`c3;TUdJYS13Wgy&bz6piQsi=nXOIK=YnjQJ&3V8xpVQ}o3M;`0s_{f
z<mX9`9*&9v@8|ey;%$9t&fd>$tQO!C>WMod0V!DRgK_$KG8hb|Qx(D7xr>G7t>EV8
zmagras7hw9bn@iMrJPb?7`c1;Q43o5Zb15_5FNI>?5E4H0-*gCpzHP=9$p1%xTosK
zIOFbELA<yqg)$VYOcd&NLU_8qBdE^Ff&sT~)hdeTP*W((><HZy6>M`1W%Q0UW^<4y
zdeiSdmVF<^fbg%DxYu^{E_kEzEe~o_1frENO%-ez0P@b$y_O!H*EpSbff=BCz7=aj
zM>QO)IW;E<m4gU?`6gp1M!rFDq#bet*RlKi=G-6KpqeTVzcuF31X_x(*`~c0R)VIj
ztl5f=O4<><PODboS$m^UN{u%LbZ^+8<#DJ<-vCbZ2DLek6*A*z_|XdZ?gbYyOQ74l
z&&UnjTzZ_>uhV04?KRW){AtY&InX_1o?E~|p!w4)1G3QucpXE=12JBhV-pHG(>MC9
zP2jj#0Yl%yetCmQvu(%|l%^IhMOlJ^j{dm}#MdTMXe1ZyHAqK`%6pJ18aUxVZ{l*K
zDe~N~_reL(r!%YVpcSkxg)-UzLKavksXJ*DH?tf#WaL-<dITZ;fm;)`=`UEpLWu4X
zV!3<;6SM81#VvtCW{7<=qT{wl*EnPRNnn_DZbW~M6*-$zpsia22odcd{rGTAe^_jE
z^%vPnZ0MBz%5$UXdAG1$hf%z5-=8<3#Eery5!ZU72vY!)RU+*o_nHn=ot}%ij@4g_
zjDC0`i_PL>jlSnXk#ebfHl=7Vuc|FELl=_u+wB3b(Is;%er~|W!%L)0jC(+Q)!cGP
z=ydKk%EhC#E)}hNjoSan1sJT%DSuzz*I~26Q5mT0w}XSgZ#%O)y=k?Ej`gA9Nz4di
z14rb#rTb8jH5y<J!Sa+y&7r<{%hF@2krBE-8V#h}kgHk?RrE{IBL?2$JHQ0FJ09$V
zS%|_AV2M6uhCMrwn&`GW;l*TkC!`VooGgQkTtZfNj3c-RB73$YkAN0rwoc;IcUaPb
z{CZJl62$U#Vjmod@Rf34i*acO?2(Jn!?Es(PgfVa&9RD7m<%uz4A75ejiJHD7ZY%l
z3Q&_Q=6H7j@s|{O+)OOhXH$^L6ks<SV8$@(azlB!;l5@`ft;_emh{JZA;hPhPDxdb
zjUF$oiznuM@zB*SS(L;*%dun@8Oug&d{68Mp#gr{I#7PxYyk$A@Pc}-5D!Gm1z5Y_
z@h^iCZP&TafkHTNg=_lZ?RPSB|8M~s%J`KanNF>n)@Tn^!wn4dcT9rQUBYpJ8E@_K
zoClR|A94CMwPr+_1Eu=wkaH-+<6Scbj*G4o6bZ<;zRpFf#A=KZR{g|as9n&9!awn%
z8Cou|_4sg+@Mq7S33Jz8hc|qU$e11V&`Y<1J({cw-=GJg?P-MfvFe?$we*0*P)uJl
zGQkk^U?Z3J=a{Y&!;fud!S90^;BD}Vn(ZKs2sAhT2@w$0a2d>Q9H<JLP0>2oOK0Qd
zU-PLG%MqIJXVQbo-(Gn$y_RY+@>t4t$OL-AAREUSgJIN$!t5=O<+gR(;!@Pc*M0U0
z#*9%y15(Uzt!v3R7#i0#`M9QEtX~aL5Um_MUm?)i96^&ZK!Rt~2g0#6dDQr;KgmYJ
zvI)WMUw~0=up^@n9$x{d1nszM#Y`yaX$?TZ^DyFLK53HG<x3Tml$4}rd#o2VKL^NI
zZG|esN)Y*OuGe=J!XL1r+s3Gm)aoJ-^Y}@mC!y)F4M*VyF}-xtVW&<%JM+uC5Xt?R
z;C>~=ED<|dtS2s6SbPU9h}Dm)=fmFBJbZshShnP#<L|g%xscP~r%nh12KIojxC-Mp
zz35Wmaf)@q!sb1xuuz*jF$?Lf^&o3t6E6$ze5#f{pKJ$S>>pc)|DMB!7ebSVBf=5h
znfHJPw)BDDvH{};*a2@;zzkxzjRNg*T2=k&w)GJZe*~Agy|H!H{$lG3X`@7toW${7
z>e1r7X2rfd)JK*yi^}aA2-ifwT30|}L=05Mtq1@UAph_lrSxbQaHKaPbFe)88TX=(
zdC7Y}mxXKstZ$USh3%1PaSok@8j4{u(gInX;p4B?zgS6@nA=-yD>wtJltqJpAOrr~
zN(?|2g0w2gD~F2p&<y4CvtZ7%U-dOk1?bou#q7jwMs9G#WYOd-EG&@F@S!bo6CDbq
zI&VNKQ0dD?Z6N~U5Nblg%%#8nRJV!%3l^3zOX^M3_Mql0*Q@9$e50nJ>t|nzl(y|A
zUjoyo<qe4O@$n%tS?ADXEl;U5&sZorvj*M>x(k6T`$W#x0!uqPBs(Vp6Dix9cOwHa
z>;PB@^6*Wk`Zn!3bD+2ZG^}8IOF22jMRW=WJoTry!ds8t(8Ap^Ltykm7HZ^8*4Vf$
zLqdO3A7jf+EWKxo5c{-)0&oMx`$8B3m4r>=HirC~NrrQUG}#<0?=e?MQ7zK{fGh{>
z;sMmmBfsQ=-hbS+3HrtK$csm?OrEGidF*(djkb%7zN{WVg1g8*of=!daqq4lR&cUY
zTv0r0G?C9WyONv|#DI5f-HI%6eVvZ2k}K#l$1`{03vG;R$g&iR%1&1Jw9U=-&$Xd|
zEogfc3tB)9mTPXj(3sQHhc31Wtn4Y-76{Lit(}s!K)CJt@n`)bPG#`?m9y5b(F`^=
z*N+=09D6aA>uctK-4E#=x@V8JApmyFScFn{I)lJK%D<RH#@BTysMQO7xB^BV^f)hH
zN&=Dfn5i=$hKG}`J=(1|QH7$U_~kF*s9__V@=h^BGgfT|0Ls`)%oqfi>HVY^X!_p(
zl#6WVlUmGTW+ik5=lsbRvcqr46`nE!4d^D-%tNfP$Q5}Wb5I5RXlz=@>2eVj?9II`
z*d>QeWXs%O>f0c2sGL29E<L49*JqEwa~mMfG06)!%CG-I%4mE>&|^X&7?zc140a~g
zy85KNpOy=bYB<6aMrb=Y@Lmc6nb_VKgc|naIu9)b5jmip@Z^65WAjL`bpc!0#PmOI
zxS`%152p~#aL3%@Af#EbLE8CJac=8GnL``M`^t@Z!k|ZlW^{&o9*alVb_iakxG^7u
zll5qG?N=G)ly!Gi4B8j3W9wVf164UsP?KcLee`F#HLU+yv>xdaB3d-8g;)1WBh-N_
zmL#|dC;`BLj&Fzsi=TS7eZA~C{&Mr;qhO#(na}*TnDYo*k@s~+5Wm|Vm?hjm6rNsb
z?8``NwBQqkf5v3oSB{{`MZjb}sRd-BZ~8SG7TaV&{i;;~y7rox-{T*UX*r*R;EtiC
z{a|`}TI|cSe1hG;QOyUuRJ*ggEA;0s3*Pb+ENcmm8HKP72kc+x9h`;6ft+HfRAx2t
z>bJtK8=VjbRXFW6(ngi5zt4TWCi{4qQ9ADI*ABPGh<z$Rg_`lvh9{CanF#2xI2TuQ
z3zj6HXSDY*!7pj<z*>Ck+vyP&pRk8x4fNQ|jPY;ZocGQrOyJ@8ZI!TFV_dxj-MSP^
zfZ<FyA)2xVKV42)Svh(HX~$s@9dlrwSVI$$_w=DnGQ|vRx=>+X`2DXvQ3vaB3XV1o
zp}+Vh7=?M$SJX*xvn{1c*%P;a2%0~urE6g$*@F4ee%Npbokpi>Tb+C2H%K{J1-gML
z6qMv26toGnou^hAV?G|!1|%WI(*s7@@$5Ws))*2|=;t793$d=p_uB3ocvIrFq;$dr
zZ(FmvSO)mlsand2z|&KnvfQ;b9JroI4}$u2^Dd0{(6@{{g>p|AwQywRJauTI%qi|b
z26sf051~CGZ@+p29OdV*OJG*X7mNBWGlGPpFiDu$mtujdY8A5MGmul6j4|^MuwIiE
z$2}zE3=6Ok4C8mF%e#N<m-n5K-X!Dd*rr{<y%iJo^gmklW6obtnqFW>Q6U6p3vi$9
z_rbve)~#_N?NUt7bA|PsxE?1dC=A>vzW_#P8z-3zKo_>McFbIQ`B5m59n%JJ&A~6U
z6;XOh`Z_2)?}jr_PtZetprl=1UY_{(MJ^|7*Aks%{AkJ8ZnIDjeN;1yHOnM=;>M9e
z2UZ-gmy?6ov+cDbMqK&bVo`A&M~y<cxx_!;H#mn3!Wt*>{NCFfC#8LKHRsAio@9jr
zw5K3Retsd37{WlB^q^PlI(>Nkgi-VI925s^#(2%lDY|a%X-AG;tfLaRDZb<Z0eh@{
zfy#*e2p%8;yw_;FvXQjD@b>-t(_gt1JBO-{){^jFAsX2?QuU{LcEMl2O=mDSt(=@{
zc@2SjFq709U|lBO@mSXugVhUHxPQIEEAD}Oi?9U5=uc{qtLyaX;YcWH&mn5!-P<S+
z{dw)r58zbPGAuQR3@2+_!=^v&WGpdEg!^cLw;P8ulpUCWNYi1sHem^sLm+rR*D8yC
zXI^b2eVliQ{l@kCdreM}cQr~t?MfvM`vUjA4p{h<Y%3yjiqxN!mH8&kh8Q1<v|Pmu
zhL>}ELo&9a74$TsMGljEFLItR`>GPv!DYzpBH-kJhp3z80uF#SL==S*fw|ri-A_|S
z5Ot`U7Q`gMIi!prEja%r4tWZKeD2DG38cWAD4#0s2VkBY0R)3BC|bDKc`2PQg1p;o
zY3pXlM*_j$XYNMn76Q;hL?~Zx>=~7JACBO9<B&Ji9J4V|r+CsqPgQj-LOMm)GGr2K
z#%RG=mMi^p2jIlJO9Al`La)28*>cG*&oUs;+K!VRJ1mZH`XrjBkQWsio1#1P7~~-K
zFj`RgcufW`WyWQ=a}~AdX%hR9nWD|TtT_V?m`M-$QH0a4!6#c22D$F^lyt-L`$F71
zl!Nc5$y4E^{0<577@#eKC(j?mLIF6R9MjMkQ@|a3M&<xsgj~nJqKD#G=H~;BO<ODH
zBn5eSZvr^b=s&=`4X$>#ar(#Q*AUA5EL=y9Z2)Qt5{OE9n>^t82EYw}vNS*jshP=`
z2l;J;m65tbA^om9&E2r;20&qq<Vtyv`}Nh&&rvn>9f6zApwP{}cd)g*0tfPxEI@>k
zlqrZjgr*^!nCb*Jkl8Vwf+HIh2Po=klN-g&u)4=Ixe$0Xe09>qI9aksel1}pz-CP5
zN=X<U&IMmWhDEoJg@yH5dp><{Fb0rL+6WET_5?=yXKR?8YJNpk2$mC^-=5Jq3O)de
zJbh<}vZda;2cmq4oK-(JeEIT)(jF=Bp(zc4NStwTs?nm~;i!x5OW2tfpBZJ<&{>6Q
zfkI<MoQUZwoP;(^Z#QcYMN{0lX{1#*P-;l&fqz&7*x&{_5pt`Y5i2&#fMjaYLntPa
z_axXx-&DMk1iLd`NaKvhjDli<byy$lZfiaJg)k$KzS~OlXKnr*)9`!j)95i>=ply)
zgwM@At8n+$a1?<VeSWy@>lqAx?X8m8ZF!@eAYan<vqmhe?nre>mY4Fy9?~c^0npgk
zI$p4!qSOi?cPTv-{(&9s3W!FGvcuEpfhM#I!B^X}{N<i-AXIk(O>#4u@bwasI{*Ws
z?cR{`eZ~V^;$d{Iw<FqBqjIKwe+viP?M>(<6hersY80|-W9=<2{k!Y<YkJX<=mi8r
z#C(Gt4klG)9f2{mBbxDq@W4(&cJN>+fO>;JHo)lUlQ87tIKA4YKHyD14r$LBD)XG1
zX`Q=;q@;gMSb15QH`nafYXqd;^sweiMw5zfDjwo}88>H&&Z%Nu-c%-q?CCuUNV7f=
zh6x=nb9k`^XN)(s75MHZSf&AtA{N8^iF@`5@a5Dh+>U-jrAuoZ)gUHt6Px%+&Ydi*
zw$P={zsCz@X9ofs;Q3Up4wvvXx361e4c(CfbU!iiGdxDj@_aknK=D34pmA;hp!7K`
zwRLxNlz)R6veyB;hmZgi{Rv)du*H`FG-Ju190r2Ny|rz9-uNM>Jbrp+m&?CF*TU!M
z0K>o26|6$TAdjsZj(P+f3Y#Q?#yR(3$F>`QVei=}A$;Jq+ti1q1upca9ud~_n?UHx
zO1Yel>%be6I@pi>2-Y<|-}(?uF)x*;5hpyU{Tan9h=&be+X(~QS&hWc2UPb0@T0k~
zS!1gMxzZ8oF~d@4^jx*0zm_+(7B-$J0HYo`h|Mz~Q>2nmUR->Wf={~7abO5*p*3z3
zhr!+0U0VT1Km>vfjrlhT#X-Uy_~qx=AIhbM6pJ_8;_0+O8^sHulI$$HqYL2jBJdax
zD;qza;l4Jgg^l{9C0CO0^}%Vrgp92T8ErSqtHBO_L9wySKxV&Nzl+eh*Vlb#UJUvr
zzDAzgUhXk#`wjsw#C>|nm2smCNA*~Q_};h*6o{q4VaVIc^rpgv$Jyy0m%W_e4B8*x
zQ@#lW1}VHgcI3;VpVz$x^;OtCVg#lOy~TY+7s?fZz?V?iKk!7*ExNdDmw<pb?4Do(
zXcWF?I6$0CFXWtMWubZg-ThTaNOYf!`hUhnd*dwx%<t#IE?Cza1sHvK_rM0fezMd^
zUJf6yQy}xx-HcCn6^_O4bZWv*U4y_?XDOFV&Z#NcUMS(;_*o!hGjO|BIn<$}&Cm|J
z7A$iwV2$TLDKdA_;E@@3fBF}T-yR$XF_@fo<g#(2$^BTf@s_a{>k-qH9LYl##Q$P6
zHPPs?h8ni8Mlj{b=t$7$5@+U!h~Q$XTj;(J>1lK!Ee+*BJ7mJ}Y^4seXhLAlr*!<*
zNoJAoFTaLt#N@9%d~bKjnvOU3yRWBYfDrizH|9d${bzso0af!tW?P7eo@zHYrcg2v
z<b*yI@m`8`rs$X|OLRSUgjwNpYe9LaJtj8cCSuS+{F|>iN`i28u%cI!kqFrpmxZ`E
zC6%xbQp8#^LhYufT3+d_0u{y`BcDaA!+fBG%>>x#M;B51;FgBqbAC!|{sS8TmW9Zi
zc~gujB!E<80k-+-kQwY3%8gf#jHiU5lYl@vn;DN^ozM*SnpvD+@*Q42I{Gc1UJa0G
z0Ob@Z0zxo!{j3xkfS<#jEE#bGuG_v-cH$_4>FwBnx_y6tB1jLXXyBHmYs9aQN}*Yp
zLHHE?NOo>Wmi58NfoI5bGoa%_!V84e`bn<T*U@2z#K?8j*+e1oA8g+XP$&XV2SbEX
zge|w}Lu*R${+=y_)L>@{jMd#KM)P+#-ia;mPs;DE!vw!c2b^~rFa2{Y5hjup9X3eX
zU;Na;#4JQ4^mh=RL`w`T*dXw`VY)gXhg8~lj!ecYSk}7jHC@`8#W}Kl*kN9RfCvcs
zxepG+pJN`#S_s4=C^rJzeE<-}9)o1&PPi5thkm`g4>(b%ek#*CS(|<iSe2;s4=h&K
z(LiKE1v)%zAGn5tYuB!o1L8vMPE@@?P;(QUphEJPX!-Dlw2%mQhdp`@2@^xsbRF4d
zQAvp>wsawU-#Pg3=U^~Vpw9-3;Hj!(27<?dPxp>T#!*y7`f4fC0_cAi>rJUTRf{2*
zD-jYz|E+`QxdP6lvo+6y$PZLbLW`d#055{6+9|~&NExHs*pLM$p6I`Rox(PrsNmJe
ztGw|Vn)RH6pl|^49)*CJS0QDdc|D0>L7Ff+(?ldp8^D3dEbqfTQDF%`rWzXsPeo+{
zGn{wfNRxln*GGy#6efyvm~}Xq@Cwu#-eZ~yCdTfjY>yy)!tGJP)9(w)hc?(+;}ARb
zVrNYsYl5gz1OhL?DfEyWA-}-RSkZWN9iIOwa%Kt!Y7=xfk-3>BjhWZW2518Z+m?y*
zOeuv*iNi;n`kXmm3ZjGdV*HI@0YX4dO15#?{#m1B3I3cFBWMxAA6x`+;i(-0%!g^P
z+sK-clfmoXAhb0=JH>P8$I`*bPdvhMIYt<jXNp3Ue6UGW!>;`ap3jGs^F$n$6RA{Z
z|8RTAs;llQ)4a|9=Y+s~toQ9l|7B_Zm%=&^6*D))|1~sSkbQ7NA6{O<9E!szTA4X7
z#kZS<J!YpV1}=7{I!t!!jD@K3LY5Owc@w3UXc1y1lgeUsndx|s+2L;sP^-~pVt|Y9
zz=Q2K-<6G6+8TPNYOM!YSbhRt&BQd`3;use7n;RfTxpq($+`A%+vHNO(z7BZ*Iy{#
z*;@QO_{FP$l{e+K{Ce~b`;sk2%17_mY*=zqPW&<ZuV)|0E$3kWyurZhto;jjk4wp|
z3B8j#Q##oRy$y~{mgznALrxwoVPjJx6FOn#31fvLSuIT1X5LgNZ~mX{Lcs@4{R@qE
z{&ix%JqBg#-@mNBKT15#fBX8d@GkKZ|LyDLPuu>LG|v3p<uynCl|MEA`klA`l}<kY
zv;XTZ@W6{9v_+TX=O5(c<b?QlWOn{9B>1P;;1W{9w`*6?$I&2BGymuiG^rzyV!Ui&
z$)va@u&f7MpdM_^zU8H7;m?zua_7)t*tdLHQ5r<~8l)1TcEmU#iRg6S3|rU7kB<-!
zfBu{EyCU<Z70vY<xIWr{iWy+?gEBUKdvYJVQ)nz+ebRQ738V~+j*gD2*R7(&NXa5U
zzXly`$ec959}0k`?KUnhE>=N39W*_w0IjM{)R*0@nIs39Kl`UM2vkb&6~m7yz)0L|
z85W;1kei#KDjitd2q8>FBH3HEHu0blXarR<nzBTLDM{KIxDx(UftqEYWTTqvY)8Tn
zxWy^}4u;Q9E2htZ@eu=6gTUfzP`86riALWN8yc%Z+8mf;bP9Aqn$)udp$+0d(%$H<
zeBud$3h3jz9Ih?-u<!Nwk4|r(YeKS<u>i$L$-g*2P!Mh!EEyi?10ly`PH{oz#Um;z
zS~9wUi?(ux?)<jgTn?(3jp+|MEPL+y`WlM-A_|NcV`bB`djTmW2j3b-W+B2Vk0f<j
z<2as7a;jxfkVQ*QG)+|}Z4%7lLS7t!7Ee<10JuPN5)lDjVidE&avcL3qzWbMYSeCN
zY!WV#W{N3}kp71>5a7UK0Ov7)dNQ5sY~F3M%sFdYJBfui&NQDC`Xjy$_@CmNJrDuR
z--j^p(^r`-mXZ*=re^gIAI@Mx*J+eoy#ZQLaNULIgW8EfYPt0B)gqv_N6_U}4Xlpz
zQ%JIflqaLzm#p3OkOC?dCv+>JD7-+Uf1S=JTvsiri&D$x_}8xH->UxM0?;6dD@cS^
zKy~lvjT<*e6=k%)H4*9XDdfL2kc<WyL%bqe@)Aq>sseedbOeqxMFwJ=Cy~mN+!cgh
z<RPD14S2zB=(7?vSCf(ja1d#<0?T0wRzbg_SLV)5=8nGHxM+SyTP^Meiq#Mh36&9R
z2eAVQG%gr}L?Q=K0YCZY5>}h68NiQXd(C()JkV{0S4WXtvdn%w2l`-aBn47|iZ(}K
z@nUk5;wxo6D-4@}a?xIu3A$w<X+>S5$eIp@NW0W_*;~1o02QXu#hA!)Eqw+cKs7YO
zNkRwQAkA72@D?z1En1O5`W^_S+55P}cVcWa0?!kG#Z%od1v4!OntXH?fXaV5^M|)2
z#23_7GeGxf;y9I{jWJ#|LSiU}WFP!K2|Kwj4Rk(|dZ0ssC#Y!Y0DazNoKmN;uIbu7
z-xlS&Z^wLdE8Ctkm-*Lu*6>M4G*qYIv5?w|h@Xp#OAkgm*&KDnSgbl^6FQaatQ4S=
zKynpG?Ts|kag?B`Zyh;^kCE_I?Gzx>GjV^2CWLSAZOw5~0fEW-sP+M85Wy+{CMCyd
zfI<*-{Cozaue$0tgC-o>m3*6PVWv8CypR(<L{VGQ4+GCNq19N2eL)4??c26>;)zZo
zmfi~yISg;u8sCSHq@_GRvGBuQ?2{-g9*s*7pIrE(hL;aHh?mX(_z?$R)OWgI9${{v
z!Vl{iFd~9}W{6B=#YrEyOS1n^J_?Fy1g|-W0i`oLqay<AA!38kZ@el_0G(?RFui3Y
z8G|ioUYvm0iVP#PXJkRO#D4_&&h`~6d=wer>p)=aSLHu`SWK9&R`PkJhNOUi0HhF?
zAP0nE>oHs=DGEVGoCau;<^`#E&Q1><Cm92iMkty>4(%5wi-L<R6N}rqGpYc`5bi<t
z-cf)4&OSMO^O~0>i?J}~{$rTf@?<7h%zA6py9bvc5(N3(I6z#-kxta1;_Qdv->7m>
zw&)DR0J?Cd>9@dY59*G^%?^%@m8-shs1}E;yOj1o$JG_G(<iV$1B<imdUjY^S_X{V
z!;np!^n9b<kE17JRC!3bjw#Ihsc4lh82d=$P`;@oNnNl--Z#r^+42j>TxCgN05?g?
zEU*}YAn0UBx>&fP4m^_IWPn+x7y<l3;Zl+NuvgKb<w>kF4P#V1{sF9;^_Sou*l(<V
z*?VhXRnvC(4p=wQ>A#-i_DIx#be?eY5aZB*6b`LmDb!zKtE@WNja;eOU~?Y#w+N<5
z$bDu+gC-^h7Srr$Bwd*~dAM2)h#U!6uAeym@ztNy)4>#QO1G#%H&7%>-#D7521iHz
z!BjyBH4?lS6V%!8)4ON2Asw?v^yh5+ib!zLQjo4e_H1S9VRH;>J5Xvcv~5Nj*flsy
z;}v_GGOHN9TRog1W=>i+un+1qYntu^yfr8t96$<6D$5`>z-tM+0UfY!gsaF?9GO7P
zy`Db|;;;Dw`F6k{BjHuF{gm$S`31hg=4l9II%)Hpl;Gr4peU@fn)|XObN68DjKCl_
z=xE@nk?0CNwQA^=`k{vv*qv`)ULNiz9L!S_SWfDRhY)tm)<ZY%ci)>g71Lhx@m+du
zb+szVw}Oq&94hkk#1<N)0zkSN8_TZk^-ob#*F8O}5Q~hwH}L6#THc++#6+A3noFk?
zD0~R|gqPgq2Qy&}Nm7}_`e}d>P~qanMs4W-#Xu3BT5WM&d1Pc{8r?osyW~3b>tURk
z7HOlR0h!U$6N}3ug=U(6ftt@#npfRu-Fo+4$4Qa{#6Wq3?9o8HF?r9|$Hxc65COmk
z#{rnU;O0~@_kH&_^kP`~PyfxBfhr3HA_A6mPZM!csI$iT!*7dK0jo|-PR0P><wesO
zDeNtSL;@a8$(=%FZT@wDl8(Z8r(p3You}Z3bf)1M;y_7~)b8B@sbUbfHbSn16opZa
z;=@D6lB+_Mg5>LHf{|9Pb82b*f}xvO|1Qu?G$3n{6t`*ZT0@7czk0q2hugj1osMeu
zzLEfb&8?nq&c)nn8pc2kgfI$$#aFIeQ3Dx}y~}p|>qqfQi^H(qr1S_E{}@&&8ger@
z+!~-wu|oO@4mlPO`2)BU1sRo#n9}n*tscFQ_W8(&8a}-m^T<d>7^BSUO&w68bOxAR
z4W_GF-071T>~b6HNpoYdy3y#Cp<&$#db!<g(l+sMYq6wIed6H(e>;6A1}lzenUxzR
zm`2OM5D7AEBtUqpw;NdNMWERmJ2V0bsA?Rx&sLHMdFje(HW<LqLYC&fLfbn9_$3-~
z44XoN3B0EsA88Of12YC!aXb6qCCR>PNd+^v&Aog!aAhjs*nO-DgK14cCy*$lg-h%q
zfg_4Au<Cf!xp|YI&+O~Vh*%2s%dI>mB$Cz45wm^Io;|xmU%q_F-;-`qwIkjgrWzmp
z#s+F=3V`6h^u=N$GjN=RSntKlAS&R;4oz*8Cn-<}p$OH;qrP30<^rfj$=u-iX?{5H
zXD&HFbPT$4%f}@ki`KQl3m&M@G%ra)9hK1++l7QGp5-<<CSn>SKN8ktRgB(HZF=$I
zg+C~;7$$Hk>7wNRvWQ_Yb=N2yIda5*3Jd$40?JIgfdM&W_yn(p9T%znZov<RvC<I+
zsO_*<nX`*oy?kT_&Um>Eq^E++fe4mbWn;eu-rNbvxIq*xH4(5IN<>1es|3Vse$2Lb
zmo#(z`j7ecC@vV|;{73Xt%Xf{A^kWEbO|Upid-Gg;47g4WtjlHmHIW1JMTsmAlfXi
z2Pk-7{><?3aLVh#!XVR~h$GrQemt8TLOMLx#$H*Yn3>Mj1E8+`gk)FY%JX`p>{x0K
z2Ax^Pt@`p#=*^a(hQLqBM3*BTAU~qPq+kiVVpT4<hrJhjP=Cin%hNzEzl^#-)KC+>
z@k%2mxK>cvlcD5QH|$R8B!e*V+7N_P9J><$=UQ#0k9K~OtPqq>TF-BSuPas8IV^+%
zq*1K2MzVwM(WLRIsccC(5}=MhS6nvEmkaZ%mAW6kAU}&9lUg&3m5Q4VMi822;~ETO
z?zt~D9j!s65W5eP+3xbZKwPCwL!q*J9i5ysudd*-II65%W}^=tM_5{O{PQF>P*c*o
zhfvE4>|QIN^Zuj3Gw5Vj32pf3zN=?ElS{Z@yt8$Edqfj>+UN2!aCH3GdmC%oBZ5h~
zHou+pARVp|KVY+i+gn9uW^JKe=)<J}+&2a0F?$4~25Ry=JV<SJ`PcyPQ)-~3ZP)Bf
zTI=BNFE!qf6{3U@!~n`grKhUk{zW0wfQ+W}5|{XZ9*rIaASooRt%cNsr1fbmAcjdq
zre$uADvJP{2c!WNWHl5~B*)R*ZKNv{azi-AzUPNUi{9qvbN3oWx;aQ<dhbUe3kbL=
zOnx1HH?6-Lg<le`RcI(D!9dDw5j;MDwA6!^KA18Z7eLIOUbZ+(ahz?^(CvNyJl`T0
z3YWQHNs-EC(#0Wkk&;|!EFw&#A|b*YCCG7VP@AEU8bL^8dWk30&P+4<fa33wPIftS
z<{^qoGC@56fWzY7+uVF}j+9$1Z9`%cPQD$rqY+l0ixhi^*GSs_5E%q1sBGAtEGdcQ
zfS2lqb`0f0)HV)aD&L+G&A}DWy#zdLB)-N4Hb)PZF4d&!FnV)@ISOI(hCvjMMuC2!
z&6b;Q^t>dfsn?X!tH8exVmS3~%vGToA*SoffImee+fq)gM=~ADpdh+NPY+|}V*$(4
zY&iQK(z^#M&>8iYq2YVGZN48ns9EAOb`X{4$2Bv5gFB%3=F_eM`8l6~lYJr*W80#E
zfr~fnZ)cKUCRpc#d<zBKP51lf2*JWhbQ5T`CUhz&=7!Q8rQ9TVsoqEvIff%XZ|6bu
zM!wCWDPxO=hes|1<4E-M60$A0%O2#iDtL(TX;cp6n6#13gHgYK*qc2t4f;aAkX)+l
z?0AF#25Mu|AZk<kHlLidYokv*`gQrpovQv+I1-vZ3p%hVO1(F|z11;5g(j<_)@4?7
zdka~6tr?n64zOT9hlg^gJW`@64BUK>rUO||rrfYYUo|!|#kYhgpj;aRzMEDqEYBg5
z@`MrZnkiZVsp50VnD5g@p_{_wQwYqLE_{#O8+9*Cq~V2IHY%`PC$=6u4A-4ad(r57
zuRAD3P#i*2j=Ripccf^UKSHvt*w<+m%@5-5Z!>|kGCiE++}vzRYKaS^6DE`?$OpTd
zsyPdikXW^{GMxu3#|7Jv3K_#f+j8lEXsBr(oCI{CQ+sV26qbI$(<eRR;NE>iw?_S`
zUg`SVVjSdn1a}V-JTTsvG!U(Fv4O!1izQPe840V+oUZ-*IXx2C&F8x13$arwfh(pP
zU2;c7OBav6{^pAMF?4-s?uK1E01tyLhltF{#U*MtmyGLaf@xw(bJ|cot$_aHmgC2d
zL&RPKS|nM{BQX?jLapV(hnK&RYuB0~Wi51tTX=~%Sj+l~_Fem8;Off;^JOrc--s<4
z16NHWMo@miyyXRp7B4O-ER?6{4x&!qqC2Ch!lZA>jCNme6!lOa=tF1C?dZZNjE(xc
z`H_~i)_%C<lF2mALYi5j4JqJd>0MBzQk%h&NHobqDReiC;Qpo^{QU1?+~JXWWo}B$
z7{r&#OxB%yei}N&XC@t=wDv<vwaQJ*6bYNjjH9zN9(Mg*#y;G28ifH#<h~>_2lUW#
zTIavz;MLQ3RQHYEfwGZaW6+0u;oOZ|x9oPgYlO+OafwW$N9Hs#M$6N((0v(1Jy_eE
z^1}Q{HZC`}o`Ld&b#Ti=merTPTflD7gSMR1dKw$TZrT9{aMeQ+N5>2n&GKS`3-tFz
zQi*_&pN5pgXhS{#3Jy%q%GDsLOoU+jrf5K)rfj+%L<#mcQO2Da7lwRFe7GeG`fxWH
z2<rTytX2vE)V29Tv$YeZ!+LCikPa`UffoF*-{L76-hevN*`kn&=KI*DJGjXmKCHE5
zy}-qx{;xn$;u``b9TLt;j><%+sHp4)gX~g$Ci}{jXU0T`7%7l@*v-vtH!W9v9X4cd
zk~;vqj=!h6m7yl-k8`ID(wMmuI2}vJ-U6)~q=9eMqextrHcp`g$?jeAG#)(sYmLP}
zjVACNT?d<zUNDA4MC%Rev>O6aXr5K9J@n7cB)7m-TdD~{#7%Vy=|W0_1sXf9+F{sF
zd^ehj;O7+~%va^|xxL*$6N%TyeSll8Jb_hFA3;iaJbx1A<9Mug#NlOC&mm9R#G3$?
z<%U{a7<08|wAtV9D_qsIBG5-RuXzROcgB$aS=J+QWnfi1x5L}<%)7IQsohF2=)-Ve
zc=(--2S}_IB}mfP`&$2(Pd@E{p~-z;v29sCB?tyG^URm#z-W)4G<p(V=WGZxRJBXH
zDcQRQCpcUUP|Xh2!wt1;kU&SX`hzrA3=Ft7^)}HRIeCf)&gRSNc0Xbez(bKjqj2+a
z<(xX)PaV=Y%!ctbnuCyCC7d7dh80IE;q*vV`Cdi#tQ6sOIB=EzsB%7q@u86ZYOf3U
zVF*IilTT8G^7y;u9lStu(lzy-9Aga_tpcb)1AYUGHR4V`LX?*M*ark8c|eK6I@HgA
z{6%5EMbil+*!wJ+E{H^GQ%{Si=R=w;8U;H*X-JlJ`&CMDkx>>ro5g-K_otI&Z)|A+
zNpQ3a#v!R9aZ-G;26!^ddVC<Y_-N<mPm%JRPTP8|#}+MFOY<T}K$tM4;2u*VO3l%c
zV1jJ3*0kWH2n>g5vedqP>NxHWVxz(5F5PN@6&)NHu-23%-A#0|MGj&oN{VKmbw;$c
zKUnj`U8XELx*<r8)@djbRG0lDea(%}+;KbDRPCyY2!(2(22SITfHx8Jd)2PJy&WTe
z3KSXOpsJvy@uNNR5fb*eG&}7C#E4{CY1sWo{(Rh5EsEG2t${kvh+kn=4d@=`l`=*g
zktlek#-<%md!%^xV6ve?CA!beGAw{0EKgn3VdOfQ?=d^X6_65S9c;XJ$lQG|7z2tT
zY*BUKi^c}t!x$Wn2S>$a+ue)U`>uHj$((cjBmiu9zs`CHNh9HBV$t?sfj&W-ZDpTg
zd)giuFa}s|>)&CPOFpED8Wye0*3UY23TljTtBq+5u1z|km6B%s7-?D}xELIsp3{B?
zXLxP$ZDN9cUg&85@J}oW{+s&k)M3e=;0~j9H%=ziUXLx0Ak+m5U{V5_H--uoCAlF(
zpr#>{5=0cWfZdat1oDn}C{`v1NCoS{A0}z`b*`^SlarTkoY^!;13CfNY5=zYlVT?<
z0f}bO5v~z`a3@#pdl&E=BH-2t4kOiSg#0u)*$q4DYU4BjdUX#j3nw6y^S%Hjc|!Dg
z=2PqH1uLW;BE_b$mHX1kYHO~(A!*lDc?iK-CH5+#zrX+MC={iy_8@}@`Kk1Y*SWVh
z%r1d_w(s6M!s&pwM>U3FaOlS@WYL|7Hcr7{Bl#5~H~=3-Gk9uRKX{?sAC82=h%Xc*
zKQkOM!Y-)DfEfYuG#rf6W7<039KchZq7mrERj{<?wRzy(P`@=@6GVjXrl+UVQ3AGh
z5W7iPp}M4~#V$<$56rqm>_AiHzwC(I$aDJH!?TqJdRd42*ODF963|L{s?e|-+W$34
zxm+5Ioxrlv0p{N~$_2Ol^vp^NU}+dXQ1tfi!G(R--hFg0+R)BjAm7*4&6N)IbxmY=
z`*g#;0X#Rq0X*@z+XmHKAP7%pB%uHfEJU=K!WH8fJ(A@H(<g-i^ao;`Z!XDBr~$~R
zn&nuv>a5sQY_d&z?mq4r#n)*s;zZlJhdL1@Doi<GxFSx1v4sB|rW)`9_-hzMWNDbJ
zhY_Uw>l7OJFmK`#SDzKyRNoX7H0T7Zc}_73Y#ty<FhJt<^ID?QV1u3_Net=%mbCZm
ztA#;5Ik~wb0Ae?}4|u?cLM*@fqa(OJagf92Q*nabg8b;ApvK3Mk-W%yWBBo(uUS0M
zv!`r2e*=v-Mro;02Of1}e(#_s5@Zvc1*jierlqAjc!!DG?#Moy8C+p3r>ql!q0Rg5
zO<<@-(#(QCwc&=0_g9#sSaMZ_8`*{y&Ut#K002OA;#rA>0xImjwZG+AW?4%INHx><
zqA%itu8zoJ7FGTq`Jk1(vl{J0pjE*LE6nT#0gNVa&@8>)Z7LB%Qi8e>4mUx?*W8`W
zduN}4<l*+a(<l^#f?uQIKO!Y94XOFD6b4?ADa9iu*ydgoJnbk;)w<ZJEy?}AMQq>_
zlCfE@v8~N`ug#|;9LP2<xfRHhuNvf@ufaC?W0b_Qyp)n%Vq))2>%Xkmr-~y)(7W!_
zV2RQGj0`aj8Q1s2X%kqP_{v$gk#02reTw+z6m-r@CW~&3HN*|~hDkY9R!1b*wn-fe
z6FULG`x>ViT<tHh3md;|>-v0%ZB1O|(9lrH9T{l4sy;mMn<4{X@1BwS#@rCf7t*PN
zWI^n<yIxe^pz$bHGhmN$+}w`y3=<)&W7)rIm@P36D8L3i5v-z!V&&~wX-nL@xotDJ
z@Xeure~81@APIYT^XHQJ4_CQ#WW=+TG{5PTD2+B7MhzorVuHBX5`LX$`!~iT2nDbD
zY(@*}ipcf$>GXGs^3F(><izJ4x@%C^iR_jMnZ{|OwiYWVt6k~g{v_b;-E;T9`iB8y
z{}uG}lem;+VeiiPR5*nOBveQqD)^KMfIZ`JvseF<KuwpSD?pDQBm@ogzeJr<heJ6u
ze`e*_!LoC%ug8mYy}XY*V>ST|MTVxt6I?z`d!Tih-eQH-J(zr1WZsT*{bR=|ZL6{q
zaer$t{J!1`$AJA<<dL}yi>}tda0>^Z-Y2kJl{jTnmNcvIufP84#_SNPJVE<e6Pa;+
zA9qx14-i~rdP;N$fSlA4bi{$0C<1vN{QC8oGYtS$Zw#CW-?J0AlaJ*r1Z6)l7=EJ^
z_cxSORaJG{naNh}yoXfh?WK!Ofw_SS97ijy<lrGOudQ5>LHQF5<K15S-t7ksN1k}7
z+p#O9P`=gN-qGK0nIX`Av6Zx`<2G?}hIToZjDyT}uhblIryWpO{?Tn7M-a%Me)$MR
zh`%58e%am~H1n=x+2J}9W!(rs!W+@m6*-79FNFSgH1>XeVy1exS#H0EZUpGyC4(C|
zL`{7^jv`J-PhLDx0)JE;rzvu64acp7svJZOJlOZ~{<@+n7Ygi;`R@97*S<L(C5p%%
z8`;-0w|6(Ks4pDf)g$5MV{;f0Npf5=hPohV(8hwI5TK4K;o5cU&X4K9`QGZ<St#@$
zkI7K*yWM=}&kx;w=k|-A9{F5<SPk|D8Cfjh<!{bP)35OL>kB}fNG`k^#5e*15VBD^
zDD3t&&@b3EaX&=h5ee&4;hD@XwCC}FkC8<l*&d!-A82Vk;&f?d(k@o8Z}&P+poNus
zi1kephkY9NNvunFz){)!C1j^}sI526jt1iA8;x+Y!hX%0UQ7wzr(?W>XD1j1R?XOm
z95<gJsJ5?@yvyHqq1F4D!@DLUzz-&Mdf1E!I+@{!pfAi$PXx}MKhH!+WIb9g=_B~R
z+H)}^tnb{Vkk6k)8<55XiHLwB8&&Tfeh9v5`&1oI-P7lYAGF;L7rc2x6}W~Rr(_%r
ze;(`oxyYxlt*P_+_F#ndq9FY*V6wk`X}9;b%FjmLUPhT*lO9bgW@MQu=CELae5PQm
zSu)}s61~2dK=~p>vo0oNyG|r^v30UBi}FQ2^L4+=7lobKV^<}=ZAlM8`O*{{9u>J;
zC|{ZlDKKE*;9E#A{{$W?X@TvRGwnrYtpXF5H1i~Rxc`flCKmbAJARk?F2THA7*zZ*
zD_Ak*$7RcE_!4vOMo)vCOMN)rBSG3p&KI)+<@Gv#5c%nsKG?bMSTVE#H3vMfTzYuu
zE=1r7U=5<i7qe#Y>G^h}3AdWguT@aiiy-V9y<L%m+R1lYmw(ZX(uw#CvHT+lR7gD^
z4LpyL|52cGb!0vN=`+sjj=UY0C{?KI3<(ERpk4xEpmkO$84@w8&F1yM;Z$TN{J0J3
z8;x|jc47GtLdq1#i(eVQKqsU#*>xQUS4bOXB!G}in8?RT)LB)8zD1AUtPV^4r_P9X
zA~FRpfSe>OEqwHv%fS(P1F1L@x|ueI0RO8YKt`N1pePbTV=Q4?FgeGK0n_C6&I+hb
zytfJ3afack`7rIGR`YPsKU@F+Zj`_Hfi5AM+yH6=dtXUS2hHM(dm-lz{=aZ3OGgn6
zL|NXiqPG^===cWHWc?qFiOO(MuOD$#0>CQ{i}HHSwlAd_=M#H_{qKof&5{h<A&G|R
zSS;FxggKD)mC}tY)q8;f$iFH+dHDIKVbKOue}E37Xpd|q72xW%#`cYNI_^1#QBV=Z
zl#=`r)CCFhQcO3-=6+%}tc>J}JC*@|aek_6T{;*E`KIcP2#@PS%%vrp>;yj)o%WyF
z(s%75LM^QH+&Nd@DJ0;k2t%@ak{eo%;3!3Jn5T{W?;R@PF`IMnCqLv%0W~%d3NLl;
z;_~Y$;@3x%w@<e&#MRA>hM{1#(65wNB50y~3M~qcYk+-kMCJZtgsCkpAHccN6b547
zPM8fE?4*{cCrtxn!2c;57Hjz;^?P$o#~lo#f~d98={aR}%xnx|fB^rt!rhaV@4W3{
znyhAIwocB)r(qT;d{UD;Y7O<Kl2P?NQxIyi;|=@_t(&4Z($*oU0hC*XaQhrXsI*CU
zeV-K)+xPtCT`<O2cf;gqpiOoe@sa}@-9X3<&N|$?eSe41C?MBqyN$@U@JaPL0Zh+{
zs-qpk`KaAw6yZpeQ`s=1Ds>taM7CNb55diWd0bX?xi2v(z&uM}?BE>YosDA$M}=f2
z*SB+z(RAo#bw+m04cc7n?7>E@-%uru`4ahoeVY91AVz#YA?!uMd>QwAvfgC<A6_pm
zCozy`@7LWg$OGGqgB(d50-$~-42ak?Ky+RE9Ok*1HRK@yh=m(}&>$e$|BG_4ypPP}
zg@!zj3C8+sWf6__NIK3WPXoMrH}C@EHlHF_JkPs-=*XDYm(tK2<p0^pPMmP?IHYgY
z8^ufYT2V%{9Y0NlbMT?65W9C68ul4j`ZuEz`oN;FlU)$T29?ndRyyKuIyc_(@=AGd
zcHd4Z(D9uKGxE!x+$as09ON0x3>k58O{Bc#c<WxYpqv6ka4`XVp&**{bixINe9W8o
zMsB0CDpcE0bWK`)PjySj3E&ItH-AbXSUZ-1Cc^VB>pmk1)-HAY($y7>qSA)ia<Sy6
z_13}JeT&%H5{=o1Hg2D?mK_$vCeN<wyT*(%qtGU_F@aE?rFC>aD7QS%EGsN5R6Qm7
z?;!MtKOn9U7(-+fB9)nEf+HVzo;7>>jKS$r_h>f|X|AnSs4ABU2;dx|!4KNIk%qn6
zZT;9oF5%>$>tPhV>QJCct_edP3b+h83Yap;w2r2=Wi@ZuKFcoUxTkPxbYz6&8g(kw
zVR=6F_99LM#7u#3P!EVa!+@PO%KVTGpUVV;&%N_T&Y*xtpO)DNfR5_5(ejRAvL1S5
zVla5@F<LL*hn%%R4o6Yagfmd@<Xl0vWo*K7!1XGzc8ZoXOPf&2t34%te8ySv@R;xU
zX-b}ybno2Q84C{Xx8oa?P!{szI*#6hzrH5duL6Fe4xDdU-HHu+UVbfHgVhHg>2axD
zXlN*vwyOrj+hIA=HA4Y!qoUn~OiDA&8Yny!Xc3L8qV|gb(&+Slaq;@E)3GQbr;Tw(
z5R`$nrrI(XbxP*8Xs7@vUu2bw(_eED<Tc%hmX?-n9GORfM8vPY0`sFme1m$^Sd<Uc
z!wxYE^XhRIPRq{ZUN)R{KZ8RWX;3V-DP3f20ag7{35G@jRpmc3GegDfAiFBAEASzJ
zvbU6sf-XWs+SaU^EAcCQ>6gKk)UTYwMZxc`h^CEPM^7OgUuv9+_&pud{w!>pLF8Cg
zsERFl3PX32z6AYyVrun@3LVm>W0X}*hUcMQ<f>iO&h4{U9jXLJk8*tM?++X8zkmP!
z#nMZDYow&4;{0c(r<+#GKl=4gWG{oCj;b_zp{KRcN6}%#sT$Z;tf4>P`alUE3kMMH
zQ#Co)tT7Pt4B#8-tjOQ4e|PbsMfJcdz$%M|pKHMMa-!=I8x4k*oHC5CVMa5YQLuNZ
zCYt#pKT?kkjU1wyxHr3aII3H`JUmB%O8TH5FrGj?a*!~k@hDPsVfivQWV*rD5aCJ<
z<``3)<cYNfTSTLB={zB(BYbTZaXr=cOwkM!`8dui)UoZSsu{d<`M#1{2|qbxBJc-5
z2+9f2dYha%*!XD3(g>K8w2aWON9oDV2rW#bL8(3lb6I;^^Yl^cs{op7p3#iLtbxSa
z{ZVMIc;77|oQsQ$1_tmz*ytwUz9{5cJUD0oW2;duPsK*Z5*@_mtp$)1#wUASh>ucV
zOtC=B2;vzJB-3mk08s`eLaoZ6Z$TmbDE&R&4p5OLkq2nb(Agkh&HoR3?*Wu$mvxE0
ze&(DN6jaQjC`lwJD54kuB?}@b2q=hxfC>`KBGEvSEIA`d$(R5o2LS~M5(NY#XPC9Y
z{_pL&b*E~kYO3bm>94xF+fTytp68r>_FjAKwRrb?^Kd5}MK0%qAsj7T?^=&LcNl0?
zmWE!zbmTB7FswSwNGnP)rB~vNLmdicQ$YtGM00#fUB6ZTe*G2$aFsfI9pV1ztDmyC
z+X?ys-e?6OFFU3hqBs%Dj)BOeb<09~^qKQZ4Nf#l9&7m<H*)E3L1vV^ymKm)Z3ELO
z6~Yyd+CWx_0v?9Fd<d13AOQJlvmVK0kz;(fh5u&!e*rdz6v`DMCjPdNK49D`7pC-+
zDp^dqQ1r|xMnGA({^XVDGrjm2x)?rQlyMF_GS0GQIEH5%68odx=KU{!+A6jHS=0ga
zrJ#8<wz)B`*TJe9?nv_7b{v4vvbo=9u0!_$xBFK}?v*%4p(O*F3sb(jlQewz$mE)!
z?AKD@xt7|hv}l*)gu&+Hu0-sq>TaYlO}8cxQwX&@S+NEtk5v4ZgEiAkOjjUm4V|4)
zNYg9LCupz-cU8h>K5e7l--@pK{=qyM^4LA76_p?h0mxMvG)?kr!3duu_6Dhz?mqgP
z32YVW5+O61<nDsjLuJjuyZ@k%XN6GfiwLGEkaH1q3W;rUYALs*kvBfhGj7p-mOGz7
z4Ya!5+pgEIaq@@mSult<Qg2~TNK(7+^v7`he#Ko()hYm1>K}uIys!;!+X+|Rm}>(s
z(*XtMQCoqBfFCZF7Vt{sfIXo)Si!;)PL&ky=M-8fWUgCQZ@TdVpiMlw@YLwS+DRk%
zLbBCZU+)J#EFBjrG@V68p`l{cBK%gbQY&_U6o45w4E-FUR^M5v6DMSV5$^?^beFcA
z1NxVs_x1s2P<ASSFBgGA$$H3Vd(klR=dc7dgy^pq#we2qaJVS1_cxXul}R9nkoz|Q
z@Kwyw2YA$m-xa8+YO+P|JFd~I&((16M2xQ@Sy?1&T;b`!c^+H(yNz2M%T8=i0J!(%
z-y6d5?Pv(wI{eTBl&)a8yqJn3;ueD={RMqE)D@=*)?RK;FA-Rld#lU8huhDA7<^C!
z(pv^nq&yla<O&4+5gC*()SpGGlKRECDKAy<QR#w#=)f$^Db)JA5hSd9(H(C=a4E$P
zAzwHUBmK9WOT)PZW_%hBdlFe*T{eY)Dv1pM53jwW-MfT{3XHu6Na_rF5#>0!<j_ZJ
zjZPSNe&o-?YdLNPE@q|9sv6*E5(qa2O1tcT;U=j_M;%O<YhP!L8i=I|NS4X(7lUsc
zswro1_D;HYwS@$ET~CPe5Oe_RDF7X+K#mDSF|b&p!PbZ(+8H;u49E(Xc`V1lC4ot5
zIdxmfH4jbfp&NVxx5+Xq_J?ZBi=8lJQ$Q{lT+&WmSE?ItFLEWO6%I^tLv%yrYrLx}
zEM~`|O$F?LY(aW7G%<A>xkdsk-$#>GAqz^@eWgDI^0p6sxg?^(0l!A*klaXOMTZAR
zg*~U?BoFbd_oNMql0CA=)tr^prfnQyV)7Tu!1ySM{q7AeA;9-HiTSGB8?KAPrHCA4
z12{k=z6@ZSaOtfQu08+_TXE-*PbS)o@c>t;Q;!ua!X7Ms(}NZju3<@sl_lFaL2$#`
ztAd?Sio81lg(c8g9qeyYz}eI~pjzG(pd?Gh+unicRGVN=!D#yD9|w18ec+NI^L0Em
z6{rD-6D`QyhJoMr!Ag%DcCBq}PF8B6Nqj@34M_yyM{$j~E{LcH_BGMlY3uy@b%~Az
zp;^S0#FIR>KZF!F+@h8bNHDQSSMOHQ-~Rr<&L4|4?M7aOa|vhLf(_Bt-%nO?zsI+K
zvqkZTR^a}U<<qWQ_#fqgNeslI#};NjLFM2;7YLb?t<zT56EWnwE+>IpCT4JO&{e0T
zva&J;Ktc|6eg##5=<F>WM;B0XeU2@Ni<0Zzd!x`C$hwX#`SFHfeJUD3sfW27*6z#a
z1{+S698-yAlasS2-6LY);Bzzx*dOTyr3-ISBFve6t2ndgVKB=+E8Ipc5&hi@<jpN?
zZ13gPqU_+@uZ7>3cYOpks+-_Ac4Kx-1eWIv*pHD{(zB3$q&MOe$5e{LB=KiIxD>I;
zdPEF*YiDJ}`2#X8hR(PIv~Q<xR)QeIF2tSKku-2Xiu6Z*duiCOn=P)2g^wAznd2Pf
zWty#ZmJrK$^Q)|WM7i1pa^UOlHF!|Tw=cS`$qkpI;n99|?b@}vbLY*=?jEB###q~W
zbdLf>uIZ%$eY>CprBUZG(1MF!FT8E(?Z6_So^TGu(ZaVNyuGG;1VplD?dCoG{clJ<
z*BN3XO#Lr_86=EhjM=5pfGelkK)<08bTeiEe_QtMEj6kkUTxkRl=d#|V_epIcJ10#
zsh1>^*(H<df(1)H97i)6Ex6?06xe{PdNyse`{d&hb4)c$3;=AC&B5`A$i`gyg|}mX
zDBN`rO{x_tL14|$8N=YrF2u2@7O14Q>r4{8CP1B1-=P@k(Q<>>W=b$1m;a8o;_rTl
zjOfei?SUXf+rC`Jz~Q%@W2XUpgtvm@@l{UZ`}@Jq!x@9yZ)S=!!&$NrtN1)(jI4AI
z4{!-!A0LP68|ROk;*rt>-oJhLx6#b-IE|DWf-};%d!o-q0iID#xfM-(htMiP<5tqr
zpx#-PDRxlt+Zu~7fa>n{L(ZVYNFbFryp`<H4^A;TkRs$6>#`uHm&aC!30DNEH_t;r
zmJ~tXjAV!OU{d%bn)HQ$bpnvKy|fQPtGyLgctlt*bsr_+PZ?A(JWgsL`XE@CM}9hu
z{&@*cuvP#PY$CzYPzT&lUsmtt<UEa^>`Wm6ZLj3bs}wQVQ3ODc7dCm1wXMQs0<1x*
z^F`pf43)6&UXm%uAih`Z-u46Udl(I3BX~JJ)>@j5q?jmOAwNn=WD(hJ2u6DF9|b;4
zY#hDFv(f4gb4Fq_YHHL_qFq0J7M~iB#F#%`A3Iu3y*Kcst>BU%BW*5x!Fk;VY$UL*
z@E<;+ZY|OS2iG6zE{^+nF~-n=zkj)s@zf<!8d)MC>@*aCTud^LVPaEcG6qN91or*>
zpg){MAS**v%zZ#F)9as^^MXObfv4}KH(HBTlZ-+_+9972iIXZYYEq6)QQUVsPwBWz
z(x{1#ARCT5TYw)yr~se<iF4_w<4VD=4ENe_aq<yiKZg!|)~kdqmmvpBb;!rcMqV_B
zt3n12+d7P6H|Yup2q&R-rI7=0mfsnPV0vnDdI_wkj8`K}e_KJZDP1bcw68vA5eN;b
z!;@-eT@UGwLd!^uW{{9U5kDJ>E~1n^eN+pjZY_ZJ5{Lmz(vU%sEW;&FNfsCqt!Nlf
z79{U~?AetJWLlcQBM%~FpWKAF*VqLGm2tc7zCEKG@&;s28<0w7$Jl=@U`P|)2w9kX
z$}NLQV`vOUfr|ihL3x+n_Hb;W%V2n!LdIz?(g5%S4`cWZ`R8-=L*!6D9e~;kT}pVG
z-$s@K#_NvD;8{z-b|=?g?9EhEn~}(SiQ(`(LtPp_jp{CP>!zioYcqrrPauvo4M45}
zvO}m!K#HzZdBV19mjvB(pxu<{fz?Eol4aNfeNbzlMsYFLFckL{9JxPZS0Wys#;6Bn
zS*TpXz=MNghT(g!1{PfHly;H_BzR<@2-4ee5`lh`*(=Bi))-MBzgDE&*b;Kk=z+id
zff94#_G%?!r<T@tc<#lOK~p_sDvXV44EzLglsiC@0OnkZThs{zO@y~FqOikTol<bb
zL^3cnMXeuO+KCs24nm(qBe?9+09qyEpPkLxo{HS_2u}QI81~5zAz>hYMH7?8QOnDa
zFR`IB2z%!iqqBez)(D#=5aS)_39O>^z*Q;)*0N&xa=OMSSBG>h2zigznfFRlK7pd~
z1{NS3Kz0Z^C{(*cAWjIw5wgS{K+0^4{yTVSM5LjXHf>DGB_XHE<GEkQX#99FF2~=Y
zRj2X1NZUi=bbBEaOC=ihDTru-1prdyf>i>;DV%B$?3*NnY-07q>y)0rkupxGA?r!x
zo-eaO-wuI)Mdr-gvlKCUa~E<;uc+UF7)s3wc;QCi*7=3(1=9p_c%dDs1FH0MpgX02
znZ%eXZ*aru4k7$!xVPjKO>KDsKO$NSwavkSqns3r<lNi2*nkCcZh@!}e<2_tl}#qI
zY^t%rCVO=QI~hMg<T4@v<(@cRkw~5?2;;rU6PAvezYU9K7+i(E6^*07m-QJ07k9Q-
zB`bo7OBr_J^+u|eu*vr2h^cNugQFN*kj5JcP^XYgev;I%ylA`CK%`;LWP|0?;4f#O
zGvAjnF$IHRM+FtpW8+}gOZ`ctS%pab<q)jkiz$PTqX4i1D!Qix_8<LAuTSrJozUuj
zgQCjN2fP@7>Akz!=x+PDC$Dn`92ydE6>U3Cn}R}blhxDp2J2obVipP$F?_N^NS$~w
zEOsUSB``1YCVg4aheQrpl;;T_(eYwfN)sLTWi~i(dYB5SShzHC@-c80wb<d@;SR-@
zi=hi|viczt(M6LD<fcfK5Uh8@ne@+R>2g7RsL4J-G-5>iM7E9PYk0J5NeM@yF<({f
zD39^uhYu4kYF*Su0#nc-;W9oxjuIs=!>X3#Q;DH;IR~5~6#A}fH;Gz&8av(3DSQvu
z)*eFbFt2dlunNDNCNos*CoS&Qats*Vq>&{2V~^U>p#cd?A7)0m<0G(9i8>py9~cQZ
z0nuTHkF2Av2+xH?921j`my48utHdT+;%WAIK5fs};00&S#8hiMDhCkA2Jw+2)iFd^
z(gf09h9s_PiJdMmmrH<Ou)7_%zrxjBDz_HAvix49OOEg%`8>h2{?ODLW0<i19R#ek
zF9G?X3T%&;*`V)gWMpMwD7WDaxH#w1EviCuu?N9)cL8I?-S@gDu&(dB){9uAwDA2f
zu_cbRF*qEuLx_~UxHOwC)qGxyd?pSQD~U!Nwz#1Ip>#;RJ^=@r+iCGyJ_(R2u3EvO
zwj7Yk6jc0q<i95NSTInGgbTCB_*A`6KuW-~@SSz&Ux$!Gr0jiS^~!V+fz-p}1eJ4n
z6nH#U_Q?)Zhyc>yb`pAi#(q>6Uqj?C`R_EJ2~;~_$HsDiI{2zHf!)Z)A1iH`#<;;H
zoD@5S?A%Fl_&d(7FHV|DM9l~>tNZi>^@`zPvCops%k5z#5C{|49Lb^tivRF;z9Qe$
z+#G~*S?DW00!}^b>i<YIg7~#C_E7aSZZAt5LK&21fF^&ec-97yuwy7F^W=nq8i84q
z#I})f(Bl2zul&qG+A@r8;lI=J2fd5|nWKv#8-=^&apeRM2JU!6kkQ;@3x4)51zjAt
zi$BM0Q&S!EXo(Z2Ar}co>IaGDK02>BTP1o8P8vi@|DIBfP~#b9uwg(`LeLZ^Nj^ko
z%f2J^%J>;^CaO@WEqZkW8OqgRo?`XrO90>m;=~zuejqmFXZ-8Q_S~G*D#TMZ1Olsr
zZdjs_?S4Cr88<6|=q4Y=UwoFGYIPl~&YGH6pwINLe`;rFO-=aE4hO-nimjH9Y*OL8
z{0kAP9J{BZgY#*DfymT}xvH4^f57bjzt)ES_dn_hv<m<G%RBtJ4tps4(;bLpCInJF
z9g9d><(=L|G)`}nZ$l6U9GbWsN+fq)pd@JOAnzwe1KU(@^7Um<$T}90kia|Etz2Bv
zxK0gkl4&aaQ_FekzmD0|zoGvgV*VSh|F3_=_@5Bg_+$CI%H*kh<4ci0%{T7r&9~zy
z{Jin(+${+{zK3R%&w4U%4sS-&veiakMW1hf!7XR|B3Fz@Dl=Hxw%*oiWEtz%L(<bf
zxfZ?Oa>+M%hUljI%Z~h>C8jhoKJKa(Zt(l|ZGkHD+e2od)t33m9*@r|+$|WZHk^9k
zV?O61;fls^TTtDE8I=u;<6?*8%76*@p~g<!J>BQKn~0?av{JiwmV+a`2JP59=_6pY
z3X#gaJG7{D%TdMt=7;j^_0vla<`hP$rK`B$C-O$i33^I3Crw_sP3jm>=t62@-dKT@
zLLHZbKq&s^x;FBk(nnFG^u~`ih2d!uy*L=#8<wJ?IQ4{+|6TZe%FpfZzxw};O3r_z
zpeFTCr(WxSOiBMwFeUXbZZQ%pM+iZP2!tUB88@5_i-*RrS{<POiR^e3BH{?xy73c0
zacp%X69e=2{=+yZdpxLQQtA+;ldeiKA`xJM#c4NE(GM;HF?-DbByqCBqSDqto1rfh
zKJE;khlBFTQ-P%Lo7926uY;X168HwumehXy1|bjt&$iv3#yDu`d?KI@3>UkzR}%xO
zwavh)>OwQSJfQ>X$jd*IZ(_xR6qVKV)P>(56RHmGo}Gm^Iu>M~v}Q~gA@M74DYph;
zj6Ga?K0;sF;5MmkR=Ab-$ghi-aCmki;O|#LM|w4i+ACjglUg8j`~0furt(RgQk<-K
z09-5+6;OHbz@^Zjh4}P`UHSBs{k%j_4nrgH>et)!3wT#g29Sg9ry=|dj7~4EXE9`*
z$!EaN6Lx+JBcd(k_)+s6JwQTIJ_1uFofWp33p{U~fN~T^9^k??82h;B=Wz}{12>+o
z<D}%F+onk`DW5zNStM5?>HMp|71LqvsBtNL{c2|?rz03S@bjSaTgWmMo%J$t6D&t@
zapjL8@Pl=L#xMOboW}5p&Vz2DOo*JH2i-M=&Qtqb?}WyK!!x89j7uTthb_k}(3L_S
z3{k1039-<WOiT_MZktRs<8}lD1QY?tn<YddK!H-873Z7r+GKTqt!D=Pa4%XM<v!Xp
zYQcN+?~*)saaj!&m&>nvPMkbRQnU+iK#!d_9_$e@6mV;Q@4xn(uKJ3Ne?REu&rhx1
z!M}{-9O@C<HG4x3<+tKp+A?wZ9?GX5ed+k5bKV`k=7f5C8vx2%g@XVKfnfEOu@X;n
zb?;Y)g>)Gn3y@iOeHA_b$&FLbe;nZbNwZ8S_K^GQv``of{cwXE`EP*1PYSxWl2fEl
z#FSx6g+C+=w{<c9vyH)sTLlb>3-4S^o5U8OZdrT~zY2e`>+i>enV1hhW`^2Q)7eCK
zVsSk-j}=`wJo<+&QlEgoKQests~3=|-o9(sm3KP0=HKQ+rXKq!a$p>V<KFXo@x=^|
zDdV3)t-o(8bpZQLSLXETC7PO=eHB8epqSt^GF_4CfWqs4U!nrs0w89))VNv^0`5R}
zbskvLZ)hE<KP;k5Us@qJwZf8Nr*2&Yzqkk>%h6z*sq0~Jal9BnR0=u(T_-N!M@K0s
zVEAbMJqB^8?~meTwZ&!0GNv&uMRa#}lklc-g{-pj8Z;%Z`Qvvn7|OH${syuE0GFfG
z)Eq!eJ>)g6eVTk2=y_KMx6#5W|7%}|<-CZ}hSqI}T<Fp;dLIDKZeil~wSq2`hJC}s
z!=LS(eCy+X-#X3aP?;JQZ#-8Y=iuaCJg{=aTAU9kweGt<xrTY(Q%AmujLSfK(oxxW
z>rX8m$mMcai$r56-+$HdT+cy2dUWyRW8Vor=h}c#T)lfIuy6x9J@-GMD5rC`Tz+z?
zQ>KoW*e47fDAnth>yFGrqI+qf&hSrl#11f^`J|_QyZhvCKPdJ$s$>q4+Zz%tx7Qm{
z14f{ADuiR0IaHCNM={MO;*v|-KMXf_8vG(xh&K4w++hh056|US#Cr$ysJUEWdf|1<
zWp3XMkr@>i={*=OCVQi%qy=}h!ca34mEWS>6wv^Cil+;t=_Vg`e2{Yb2J$qE%wb$o
zkU`Y<Y}?3idjTkmtdR`Q(GwiB;X-5Q7Wk#t&mB#L9?cf2>;L{$3>qo?m^;Fs4;ps=
zCnfw(+WX(jmkW27lgkDRRGGUl)>fKu6`c<?djw=m8?oa=Z4?Gylgt#u(Z|PbUx%Ko
z&OO9pX!=G0h0Z{Dj*Scn4IPeI4T*<*90OaZC=Eh+R~_hcG`eIY_e^8iPr$d>E8U3i
zJbi%9uuldVe#C(uv2kNR8XLcW_s6$kMw7__PCAozOn~J0G@sx*7`#N1C*MhQb;*bh
z5=#LU#<-#Ez-3{>*QMS1_3AMh_e!elz}Jg4G$!GH?mQ}kV9sF2&Gfb89J#-@@GxD8
zDlQZw4sQAaTS|n!475ko(f>1mD@J3|X8cvfhRG)^&4$J>aN8_2`ZN&A5_c15bmu5}
z11A8EuLl*w8h2@95(|CPlBsXn38ion7f&0h>TF6{fxlt6y#vNp4~A*~EDAbqwiI&;
zxBAP3kehqm^OL*37}H}jc<+8(B(%#5{T<vi`IG-HZW_J8gEso_rp*>(ce0jaUkNe`
zeGDeBS5z#98Ye9eaY5)9=VKg49efEp6%P7o$A4VU;ZW<pho%h~`52&Ql7#?U2O^gP
z7>}i`9`xH_qE?uYf#7;jEH1~GUafIF5+gEOyz(P~s~Rs5m$wjEjduC1AKO0ObZvZV
zTQyS7EV65ZBb_ynx35FhWQ|f>T+ELa7TqJ{_)4DBM4)5exl;r8c{C_1WJ@q9nT`KH
zJkF&n2aomTTP3V;sb&CfulKpX$T~DM^c%X^<XDd%cIuOc!+}-u^76*Q$$<VyewzHB
z)zMugr?biZ@T-swSN>8AWIaJsgY^69_jSR0O&n6ouo@5|kAgGkpZJ%w2-XaM|6xbs
z+N3~Et~^NB1VGMU!EfmTRQm~8_N2r;EWI{6?>-0NkKe&P2e3AGoC88baX2PV2dOEe
zC2OQ&n75Z6E1~^OB^fy=4V;2FH*U~%7y;5x?Wpk>UI}5nOQf_+qW`~e4`T~&!H=P1
zLW=BJ;8e0UHZ~dy?{hjqMs6CArGxW|_$Ri<EXzQMum}2b6!bMnLAU2v6vEMmECixK
zNjn2cNx%)<B#S<XyQI~urs=CLlH+5$7QRzdF)BLJV#T@OBj*CYxmoB!ovp7aLb@L>
zsEho%g<jpQYEWU?<TQ}P4X+>nxxN^*&&a;?op=q!<l!<5`-reUb5a+uZAKsRXU?S6
z#xUX0(T_+z@c>dP<(i3i(-%wBz^WmBctNZHtVvsYhsc$!g_xEt>i~ZoD}X&mD&kAS
zX!->_(Z@?|lfLic?N3iSIXMYTj7~l$hdpA8=9m!mZX!zsh(DE%#@*g=Pc>CluY4H(
z$l6I3MPV#RrL2c635>-2Nr-QJz;a?>O=mi`Z!g>$&IXBS<FX$<WjQSCIqor04ilKu
zLAUwjqw;7yY%{5&04w*|cNBgVrQqdWI)Z`gSEOxj*G4Xv%H9IFYDqzz-f$7FaOKXE
z0WxF=I2qQubM{2_rCq`KnzRNZv0c*qt{?WZ7i3q3Hv`?}cDl2nT279LD%ZbkdFJh1
zv!t;s7>{TGQcxDUlYVh=aZ<N$!(;Dl)s3^Gh*mb>3i8Q3#1kDfd;_8pxh4T;=4(rW
ztoadqq=@`0SI4Z$vbO87$S51~X}S6e#Ku8bhq9zYwOhxyUu3LABm+IAb*P#ol(i!E
z2B|`$;^5)Q{8rf$W!n~2c$<amE;`U#Na>axd3g$9x`zdDcwb}?$<YH|&jMxnOefFA
z@n)MKbQ>n^3k2IKZKOZx7A7H~C$HT4HjA|~h`{wAy4q1bUL=4GYE!Oed0>%1!_!%z
zzz|UoklD;2yi?EBCqV$@YeHaR-%fkDFi~8Hu<)2jr-#UNSnq9|M>Z8TBh>@X9Tm0n
zLduc+6c$ouc}JlIAuP%ONw-E!J{>hOxcQu;>pEeKw-#0hf;`f6EMnQn{P*aUksCEn
zz8Wl8<4xUe9NvM8giM{$_x-V<8C2j+cz!074GXW6|5Rbjq>(K4p4K^NEdB~N_iFkU
zl>k48*3^kdYp;IhcR2wfiHPp8yoi5$C~`kRk6iN{d1PvOpf`K=2P8O@(l?vJd`=zH
zfkg;T-(Y1wsq+ppQ5Jel@PI?uaO*XGJyCgN^*Kj_-IZ<E)Ns~Z_nTZNBS`Y_#u}4s
zJgg4IU<k)Su8wEW_J_yu+p_pk7$lgZ=-OC63LVf%D+)4UglVK7SejfI6&=jGX7R$H
zfZ{7$3qy>l1U=XmgVCw*9(-=4>E#g-YFw0dBcqQugd3VlFNS(667g{|jLXlhhA5gH
zZaNx5r;+g@3_xz@;IISDC<|VFOaw*foI9!eJ0ACkndF)~QDodvyz~OoqYF$&m)j$N
zzsXGZv$3&hGwZM%Z?@DrU<eqKd7#>;L{e#hB3BDKUOEG4a`L8P<JsQEkl53%H1H8j
z?Yxfw7A0FnDHmtwK(N3jjEHiMj#dk*oABGz)HlMDg{F35kwU9t(P*Zxh!F0?GY0@6
z8SRB`eOoX2TpOH&Ge|K;N=-5a;UQUqPZn|G8s<lbaaj~3+=RSC3z}>QP?skIlr9CR
z_~CsLhH^?uFMA3KD0EcRok0(|EZq;R`6fYc0ql}B6CEP35)G|#53`soW5NYIrGmgn
zavO&qhf4<>2I!<5fWJg9u4$bg1LQMXF`|lH$vmKflNdjq3mUdt8vGVY$WFD)II%UJ
z84LKR=#~a}9(%{n&hW7_`TAn=2<YN)DK*1lO1hm)>oPT6&N&;@fOlU9eYCv6i@!i#
zxM)7sMuzCnBFk1j%STS>9ycm3N26%?20c7-6#xzY3Hiv92^{l4J2EL|?^&gYbUINJ
zeWMlM(1S;vb-<wFa?qqtwQU<(x|D%1H^cHOn|!<C4RLFd$(A^_QhX29rJ(sLmlALh
zw|j5jy^Y)h_uUNBePd#{(I;k}v>s1M*?*(s2Q%mn_UJrhU|7q`xD$dxVX0mrt5--T
z7j#7_gr8YT_S;l~)w?xS)0K8<`(!p07LTYH-h+%IxZm{h<;%b6xd*CX0Tzu6AzWJo
zHr&k3fH+nMdEQVi8++;K;*tfoC+YSd*daP*lAVWUv&yyiNxW|Z?>4Jutg1$HOeQ9h
zYJ6KmuzRgB9=4}#964(S(gY`iaTqcO-_)aHlOr#?XZuD&kzj#=on%LQ3#P`Qp2?UB
zi0z2b#b4EteGj>&#qW7ND=%>}L_D2}TRWXjl<JZrdQAQvO!8;)AVj$I<@mdGJo`eq
z8(O{0%{=CZ1{A^fwu~<?4_Csmb%1Hvw-ggY;txije}B(a=;l=kSO!s8B@4Fpc@q<4
zD=f}xj6#(%Vl>WW>jf-_OT4-Bi9)dm(#C?Nl<UWt^|jC3+iZ~mw$B>yQ*LI`c)Vu&
zq0*q$kfwb%2wxT^<RN69kx33ESzBCGmTmLU7p0uLKV!o3U!SP8zg+;!F<+*{MU2J`
z`b{O(o??hSrh4yKDI!f<d?$+UFM?_{LHGBlrc9c>*p85sEhUGzn8tUOLxP5s<s53p
z``?l|YZl(4dmngJ0|mn~TO4=DZuFGWKH$fPNX-JB5%1;Rg!#p>#Lr!zSiV>%f6RKz
zN#{>UG<ouaN?M?XZG}m;*A+t&sXlN@6#zA94Q7tBZ`}Zqf!yxZk}-zTnRN{o*yLcX
z0nJpU^&6HHrQD`hyYS1@(F()diBdBZJwED0p)ViLm3`dI0V$idd7X^dCxqB%#=<45
zZPQ^9rrT#-GZBV!O>X35l~nI}v-%0W0sv3X4a{ZD-bk8E0*OPxw6D!4w+%5h)|6jt
zo2}V86U-qVi~d;e(nK$-36s`=w`qM5w~(o809fS0`cRdW*_Rjpf4^x$5VQ<(xw0;v
zd(rYm?4@d3#naZF&~mV_K4Ck@Mt<pU@OR7Ky>n;IVDFy_K{%k1d6i9xlu$&)P~hN|
zO|1h%gr;y*BG*-^cWWI!l$WU11?pQ_LSWjMk2gGXDe?ZQ+o`q^mOAKZub=2zKk<yq
zwMu1bsU*n`=aN%Cf4yz^sb5d5h#DJ~nuf@mxxHnlPH2Q6%r?UfBoP|Sv!7U5kwr6>
zWYYTD*AQ#(jp2^j&_@2HYbe?E{pi=u9@?JDk*IGl{Z){ph3m5mvrFdfriLAgyjmic
z06$ABD6Arb^UQtOrDZT(&!b!|-tga7=Q5zzy^$sWNJ3M`@d0^_1UNDg3pCJBo2-E_
zYg&h``2^gV(jgB_)D6NJzAwH_Zzz_YVNHROw&zM9G`Gt3619r&-4ExHl>lQvEnbbg
z$qs5MSvakA<epSs@|NNt@JqGiYG)BO6t;So2_eGIsi+xmugOGEi;YB5d~@>aZ$u*U
zwN>MnDokh=BN){|U&CY+G$9AP{+Yjh4&A4X%60zT9e&VkGys|xK?1;5+tJyX{L#~H
zTov?;?4O@xWrYx!ftXgLZ%I8_pVBsb#vf)qd3z&oW1=)HHBxP9G}wi7Q<0(D#*GvY
z0We$PdK1HVtp-$@4T_I%(_EN#v#pF{5jK2O-;dr36gh7`Or3;A#C|(+stY%$lH6D2
zXh5PYN(1(Qdh?CrjZi1plRXX%>P)7d<Wq&yt2mNflMW=fpghtX!=95lYQ5K+!SA#z
zedR!&aqDcO&#6?Nin%Yco7HLV-I;b3w=cc)3{XewK=!bsyJ4v~%TkoP`AJbp7EdXM
zEmW8UQX@n$nO;Y0Tn9Fp4f2O@T@a<n0q8S-I^A)Yk;@!Hdqc1yK5WG1F>pKyAm8<}
zFO}W!?lVxoxEWZakM)Q34<irf>0m~>PPVkFeUx`y#osVtgeJA0f)0l`as&wvel84B
z<uSp>(5@{VCcRn6GP(2H6u;)P-u$r?h(sR0+BuP9QOK|^?VJqys|fK1A7veCN+XpU
z-!;3yo;r%E6x{pR%v97Z_BOkdwzs+HgMVlPlSiKVQ8;+80${iXoU$yGHjC?l+c#36
z{!F$0Me{t}$RGM27aCORz0KH{)@5fK)f$nHea4dylVR~vRXlhl3Jg|UiAdfIvUdc1
zZ5<DhW2b7@m73GA03|mhAS%J7<ZtDpi46?pVHQMQ2lDm96mAl^1FIuK7U7a!Ga!N}
zvUk?aO67E4urtXyG6RNCEL2HW--MQdmiY!z<1wE%-{_jI2uzyz0qA4HY?Qcb8a&9<
zO%=ZAYs$wOwuwAJ9mWaf<mvWN^ncW$bTs8qhdP_`0bIIAkiAiXO!pS0q-0Tuo<v(;
zoOd8J(_LHz47JGVQ5aJEXE1?6_RU~(uLQUC3Gk<HZ}iS0HQqOQDWX$DC<zzF0Bp69
zuDO@0)<8Ni*L~j}2-e89kK#sH55~R?nWa(vHwbT}_297?04n$eHW~|dH<<-aJ_sZ`
zFq&Vv!JB}RLNl{wp_(V1X3f1<_<itkC%FJZPYRT%WJF@OH&R!3j367#8KCKP^b>6o
z+BWqFUXqh9ly*BrX_wH;N2a>JA3y-P&0}S@!svDV#r=jj6og*rG*>83?ko3E?|Jy;
zTk(7p|H^{8N3UUGK-vSi(d6b*RqfluOr4lfc%SQi=5~Jwm>>hAF&gL^J;^u0^}>@s
z29ssm<x>1?FuY69X_m#|09%dpV5!l-@M%}ZhoGkbc%vJi7SX>Gb*=``B`&zui#1PT
zI0!*_s_T<aB@{Z5&WhNf;Ey~HLPjk6s*Gek`^gW}0Oik7LC>rH*~-Yn%PsAwD4Z%*
z5bGU?CObEbVnc;B#xmGp=e!h}@)sfoekp#f!4%Cp#dF5;WqVC=Ngvdm>~F;>6H)^d
zCxg0iV7G>sGgIBGj$}O=RKi)2k$uCZQ5GvHv(Qz6Byu2`aKRnXr^{yYH$A{yL>9vG
zY6727mX(>LTO6sPxo+|dCQQXnw~u0BnZ)E(XH3xK+aK3YvfTJ~!2lQ7w(;lMh=vA;
zwONZT2S1v6Bd$f#ZX{bmDUq>Wu|8f!{3|>3XhlHFUqv8NsTZNEyHoyrm^3FvG2CY%
zQw(ZuJ!Z8WAO%YvV$xAOSd2T)09M%n^sHK~qJM~dJ3G=R#>JXj5NvSGlNYDmCWrmp
zz^9bk%%>KobPCb^T86NsPWuKCi7=r_ESD;8^nZPE1JPlj#0$CK-a*~dc-WG`FQ+V)
z?fapB*yDQk4zBABpUGCx57d8(3$t1D4~mJAT^!|9m?0(;Px8lGJ<+E&v69e@i2K=z
zoH&5w7pDQGO}fIE`1Q}Tz;@1nz)NZAA{ReoF45G7#xtoCB8Y*E$&?dFj6GPM@zvfV
zn3Prr{jdnM7+NDLY{?k-7kH8^KtZW-je>7aT1qDw_ML%ZQ}~(ajjHe~Brn0WgUjh?
zUgz+XA|pXE(4}Aa7nvqlk!&l0>;jjrwnDomK=o8c<Z;~&MQm#}#OZO(^}nJHyt=<g
z@Yl{%A1oZG%Of$(m<3C^g(>HHs4-b1c@~6>g-X=O+y2QLi7%jdp{E?<(ob@V1IM($
znEI0pj!Zxlb-)F3jG6oJxtu3q-KoFI#+8#A{ZDO}g_BDg&hvl$qs%?o<NEvOJ2NJm
zY5)4$Wq%D8r~Z8G;FK%j)Sos?`_DI0==%H5e-`I|=bdx+M&fRC+gp&f;9}ZVuE}8J
z{lLwtfq4$4a+d3#*`B#Hzl~uW?0JiLPW@G5W&vDj-XO0ZMt_vymPipC-@%i=di82_
zCk{}{_RcVZrZvJ^Bbu$N;B)3<PnZM5T~wyN3FsT&WYwYz)oZ73z~|h9>D$~XrSKLJ
zDc5YsO&`rk|AcCYn!9(jI^1hUwe~})fVnv*JG&04k<JHe2BTFQNdsb|2fj#|#o;0)
ziT@wldE@JE53sJx=o?==n#z0V$g`^-?-iX{x>j25%6Y}#sc)Hd)qS}#`=dWa7JcLD
zubyx1v(fj=yYs8BZ2uNqER@^1*z&i_ch8re-$i>&JY$<V+pC3=Kjy~{R42MD5uL;K
zV2P;JtYBF_`ahF@`hWh{r8|O}%m3g0@;~4HzjEo``MjQgklsl9+=UA>;9by=gAukP
zsJiDzi}`QR)YjHOgU#y6lM4~oz6LKv_H*V~z(z41ShWADNf&`yzfv13D{C}55~A!Y
z+5CN24w$Y0ugqzWT7v#GTy-3umOZ<N0_G9!>w6cN%7!uft`0KsLpgR)-k#w;UE8($
zq%M!)g5(53ExLUvoBw7YTe~q*ypRzrAGQi}2J;f}D4xRRYj8~877t_Q@D)ZIAD-SQ
z%6k)LE$EwxoM<D(Kz@Gy6{1h(u+5}7q+=&fPG|IX)~p9o&C!6Th=aKG;LuQX8S~n;
z^$;T7j*b$Wll79kQM4EmOnSEJo`Y|QJ_JjWBImOC&+wJ{jy-#9E~achg$AGg?Q!zN
z8-OddXd|3x0|Q2SP5@gC_oEBY2Q!C3$xO(MTUlF+nwjmFkd(Zjp)p-tTpW6AYrxr+
zed&Dn?lcxG6qFjuOhLu!Z7rQog2+WmOKNqqqOs^dF*Qctk61El7#ACRKP+q~zJu+>
zmwCy-{tqydc0Pj->;CY}8W~WH+#VwTik<u{2!#)2H;Bf)tFB(h{ZgW(!lL)rQh?FZ
z=FhkJu!D)s|IF>t!RY;3GG;zbj#r?>0wYPm>U7Hj<<8n#hPZeZUfqxa2ID_eR<4CW
zAU%bV5iirDLzwmWgGz)XdJ%B%2iuEBMn<|qexvwi1&RX4y6wK_T3TA*vnvQynI{F_
zBAeLR=iwn=$Hd&Tvs;LvlX-CDmJ2@p>b1>@6P%Womj3?!07cNV`y9YFvjqcxX!-28
zZfK~~hNv#Zs(3uQskPM_J7-4({EfGu5v4L`%f<&wd{2Je;l4!F`V|DY@vD!f{jRGM
zeJ{vX16~+^fK=GKa_N|=s;Zd|jccWMoWkIkKDGuEM$&c>k7o%hYhI$=(S004%a$*P
zmc_hVw{HDWojp_vJ6*sPpI{AgUS0h|at(T**vzJzH*bDj2Gii_i`VVH1Idw0&21Cg
z_U$7>?T_FAEM2y2V?rl}+7_KXdlm)LOq9{{<!zUX#$}e5uf_*aP!_>ZzZ~4l{+x+i
z>eOv`{rU%dpS&}8MT=2`Fg5e7c^KD#YL>*8m#kUCZkcL1p&D|CzW_BtAu7}3CV*?^
z0C&`g80l`DkKWX0fsLElOje}V+~}-^!jR4ib!_#AdYA}4!5T9Si11AI^<EgBdt2Q3
zdlb#U2atdd-gQ4HXd7hBCqMldpGvv~D~9jxrKgKr%XJLGqoWJGk4&ZB&z=pdKnIYF
zt_%BHV(oBr)Ycx^HuT`(!wdM?{Kl;-WMpM$4wROci=#vF8e#%*XdT@-acs`eDimCY
zTU2ErnnQ*3y{sK?JPp4m8SkW_(U=7Y1NR*q7B};`y-K(^G&;mXvzAVA7?@pz=1X-u
zysLp=dlrAx#A0NEB_WMT*$Bt`IDJJ2J!5bfnNA!(eqrwiC1vGR>X%P=VcXOrzh<Nf
z#|vJoP7&R`lr0n@6Z6#4%%qh?V0!Firw>DbCBGA9n6W;*jS8PV$6%yi4@|tfhHLGv
zD#u`yZwUqx9eG73kWN6TWx<t-xVd}O(i*Cw;K&5TXFvM-Im#1*T^g{tn1PB#KKaLk
zcGPA}T<S9*@3DE<75iNSqe^Ix?8HDW6m(60y?#(^-Zo$iF=><|qG@KBU6#W{=aQ3R
zR{W$BQ!IGkz=2VOEOrObOGD0IF|5yA2#VsF#$V6S!boW;-0r`n0F%g^7NfTwJbrxt
z0#i}eTw|^J`ntM%IQ4AkTE<%d)fogFD~<Z}%jM_S?$-^Rt^Rda%y0$Iv%2kDx6VTR
zmQ}`pcU1Q@b4CJMhENzZd7ExFM1Iu@Lx0+_MQ>?rT~ifz(dNmM_u$ePU=FJ&3x~57
z(S#qS=8IVP-;FHU%%I&3DfWY=rh>X-pn8C8p=Nqa?IeyY7lxMsBF%dF@};jmq)qPP
z9mDj76PONa#MEO8GY<_Ihc4cf@dA`m057lNED%M4&qVvyMqBGOtAN04gLe?TD+}+^
zq{Nrm`lNGjuaQ%^-5F+PX7+)fZYw!CI2HnZeUIK&oXjHR2uG%mDQ`hoUI9pEHi!^^
zdU`(MyXPnKKL}odLoVR!6%i3p=JiMQ%@t;x_4wzjc9?*(WRHUvc8={!-@&`SAoCnn
zQ&YP<hT7}dix++hX%E^lQG5w1?=p{nV@#L!`?qvU5s1Xtf;vRBAUyc=HlCruu1!Cx
zs-gi--P@hyb8Rebe5glzXlP_;h>e3I;<b&K)d}cBeF9U%u?#im*`1p{a4)zoVAyoY
zW6=&Jb#-;f5D6V;joJIY@6I8>*yOdnVQ9JNx9wURGm0=q|J1`HJkA0DjdDx)a96!J
z%F#phM#sM33Z12`t&MM884|iUCN2)l(M^tJj~_g!MVCdV`Q-R1ydfxOTj3gs6};#I
zzyPK?9{?Qa!*T;|5zrZ$j;m1KEs=6(eFK9-@QggP+Iw|Xd5Ey>ic<hgYgP7Wz1zTr
znOd7WcNh`=N2~-<3fluQ-WgGodV60roDipi@y^L$JcrHS6-I>1o9A2xfAt;)zZJTX
zE4(K;0Ia*xI2!c-FkjBd&tHTUQBa<!%b4#WY_8lq<YI)W$zK)mJKKKRK<J?szv9;5
zA2vjLTfC@`d3g)LSNMd@r4PMkh$4L`EtR*Bkdj&iNNK_1#SS0%7vg4qW+@gsO4eWT
z`?2rnOCtirt>jQU+=p#?@i1`AdYo>`6y~sH92XV+r+IpGW@aXUTmQ0dB^Zp=UaDxQ
z#GLnisG!wCK~}Ip7ySW`wh%8F!v!CKT}<OQ>Gn7COiL362jniqi|Rf<+k~DxH?{%#
z`1_E)EJCMHR7`9JWT@6V?o^brpmzUkR3rJbkIZ2cvqdh#2EB;OsV`nEN7(j{H~YnA
zF5`NUn;8U$s5oIEA$P-BEcAq{jkC8>n!F>kxTM5HJb)GRyb3*}cV?D-Sd>u<BqPO7
zk%0?o9cGHGfi%@2kNMNzpNXuB6Q_~V4P*c^7G~wVc0U4MLS7ggfEnP5bQ@zh{yi3?
zD9!x#dyV$?_Mh$TcHnfsE<>M-6(CW`ag(`hTPpD-(4dQoO9R#z{`g?r1vtaZ<YVV%
zvxbPvLtH7_BMP#!KcEXT|CJaX)_OTq05)&gvLoW=&6|%A3d*{VA3OFAA_(!$>vk|_
z+Oyg$UN8LTYLraCJaD1x@$*0C<je;p>?3FooRBj-bN1}X>@a|caDz=N9)=}qEw0Mv
zIa1>VK*S7S+&UD4><J+N`Wk%izJ7Ha;71Upzy#vgBQS={%r9Nq4sv^F3HNf*6OpJ0
z4I|5M2$68hUbn`(OC+H?`p>Lcvre?(Oxi%>hVsk@OU|7h58OuU&6bggk=LcHtPyc(
zRAxd<Orht5q+|vHL^O%F2cEPJ^6+2*2=xz=8;?p>78ZSw&Ax^yI_$H<L#QL*l67hD
zezfB6!op;`-2zK=Q}&y5&wx1563c(Yu3|=R&f>)`d$2hM5^i;E1RZ;mpdC68Y~0+-
zV8X~?_y+`B0EOSf4|$H&lC;Np5JF+VvLvF(7%3);(2WH+(ZxDR{<nDe_>SVfg9ZyL
zHo$Cb1R}c`40N1}XeKkP{6*iOXLSZzDrHe7_<UQaLKY}TGIOTQM6WspoSWP`o}Qj%
zC0jEJ(5f*Jifw6Y(*Tf3*UAM5qs5tn>i9W7KNDlie*n}z$u~qHfVKhN5b{={l+?{X
z)u5VgVF#K(Oc^5LLUhGedL9MMK!2!PlV3pK%lh7$<ePi&DehqB>c*!$c>0`U<HmoG
zPn~G%>+35@x3r9C-Us~$qt0qpQ&ZkYTYi@LA72iVlR`1q(0aLL>bkn~agYL-M<D=d
zfV(tcnm($P6{n`ckLcmU1~BeSu^3x}*n=eVA51#m$2IJO-K6}b9v_<jcK<F!zNM>I
z&xV2zNdFA<z78TKkAE(1;Oy-Dh%#^BsJD8W7GliXCn{>Op`js+*j8hGMFGd}z3Paz
zu^@!qx)8cIZ{7MKJ9{49+Qa1J<O=V@w{G!+bomI)9kbEFrMLt+d3f$4^M&r<L(op1
zjI*(C-no;u`+ekGFx-3U<z<hP^a|j_uaY2#qkarcEv-7dy#&)fc6w4))`c*MIisXB
znRO#XBANzv3H<tinHm0a{MNS3A~<qj0qhwB=hf)Bg@u5Cz%8Wwqhn)TQ2O|kk-;V`
zyl(5(t(^PzJ@)iu_3-p80@34r(OKSSPoLgF2D*u#p9$s4cKOg1uq3g@(eg#n=CM#h
z9VtgW&i0%I3&b$il4dsa&r69}NYY^Hw-WFDacnFPo<5+Fr`QWucP%a(k4h-*GmzOk
zd7s3(qDOHe;OGWngjry<^L3R=_e0d_9W*+`0uGNCK+rtea>8OLMq=yz$B$=cJ9F&Z
z!TM6-5sqp-E&<qXPG`)RIrAs(V^BK%6mnEnz>KdRR7m=SSikd^E?Hqke1Ycb8yi1-
z_DuT1>gYmnPSB&8KWEXRObpOGb##0HocLpL@iGjKKLX(>Q`SN2;>B7VmN|3g9K=gQ
z`f@>6*9IYo8d=-b(o&RLX3d`cv8HA{!WDgXjD4KHaABH|kPs@2cNrNov7l?5SPR>Y
z9kGZz6(ei0PJ>8h>+0&x>*!eH6Z&Fb=o=cAw9c`vc#ijswua%=68S4U@T=~HP)hsl
zq!lQ?5^fLd29Ep?l?=Tr0Rd(#<Kx6c0ouLT34W+<1$RPw@1{N`=8=EH8D<*xs=kqt
z7`4BlnG|1qk!jW`ul;wisl4`^-23+J46@dlXo?{ou7q$Q@w8qcmZ3Yb{TPb2IrHb=
z$AOpmxU0rEBm}_!`-+M+=&yzM;pF`3>N*4^8KdWW*zR<6_zg+t`a>wRpcXd--q$81
zlz8gxijgdkim*Zw`o6T3JTc3*JK*=SAd|+E9KsDKAt7;ITYI*mq9WRMM<S8}!^3xh
zz&afShWGe+OV+NPKYjZ2Gs?=ocsfuT4tRKY#C4}mH07;j5jJB&BliOm)AL$dvtZhs
zkesZma*cWE(rHMhN5S`^vJ{gtkij|~p}aRCDajW>XXdg^$BdEL=hxKJhj({(FYC$q
z{P`U+)o2(BMg-+Ska8BRAt@*O1Bj7UoQG#Q)B~SRcu~q%Hha_OQ>(dTj~<-`IZ51o
z4v<~hw0rj=5Mk*sV?9xZ|7>klZfMKKUdc<=!)az0=iST6c_%qp2#Ekr)du!cxc~B{
zzC&(E9LdPDQ9W5ik++L*BX(<|1}2>2x}d$iJ({?D$jY{=y@jUBvdSIq6^ReH7Tkx3
z$vbR5yy`)ed)R!Cr<#e|V7io)6gIk(N@wt~)p~Ao=gvi-2u-oujUx`#th)gLTQCrH
z4-0xZLQ>47ya>nJ0Pz<|(?&tTHB+_t2P~fhw7T9E6-n=G_R+(Mto8U3j_vyT_3J`Y
z3#0%U2!ip#^Y(dTt>{4pI%~uwBo<(EH|V|N<Ku&$gD7q}2%8`AY={cG5sBOaG^+W?
z<DjoH9bfnK*)t#x^S@P8_)>5~!kn9z7jJenNfDm}UNKVIEr?z_jk9N^36CC~EbAVe
zM>Ky8JzM8#??!MWcmCeUjT}qMDFmzlb_Z~c=A-WUfCI>hBugDULPP*pTqg6qt^w+G
z1*u%Ru`V0o%oPUZT-S!VU)nOGbn9u%S?$z<dPl}9&BkVaeP`#MP!%xtkjy);e+%W;
zfOaUHJ-fC?KGZSAtgNgoe1A_m{DRp9Ers#Ia?zx5MgB9_ViCg?xd;z<n5$oSdF8vS
zo<F~#%@sK%4}Ka}>LTJ)miwDmuh@DZ_rQ(S_RS%m-#`Euwh~*H@45+P9amhpoNyUn
zFIhNu?$1$M$isd3_U+rBCTXx;wNT>w=6G-0wCVm3=w8gay1MTT3rk8g;J=wKCMI_5
zgjzZ31}5YI`P{fpPZ<efDP&B>8Z9Bl<Gn8e&)2hVThoNc1*WZ_<3fs#Ll)(;bR!oZ
zpB9qi&+rQrl^ez@ZX8Qj#ovD9REz3@GRplO=aP7@rS<ps+d}v!cz=(_f!oe#k=h0K
zaP#nJAQ#UC400^n&*zWlV~MRAU=|xWcjC6=?;-Dz{BF5&9633QeD$=o#x2-kib<Wi
z>Yylw_aZyZPQE6c9@Tv6V@Afq<Csm1iH&_pzZ8NqKgT4XJdz13Qt7WI=)j-y8pl9_
znD+z{Hzm#6JxzAS!X^ev;B1{~)GhLz*?Ak0h$Cr%8QO5JD8PA;%CSdHV4Rf)h46BT
zYq)ki2A;#vhZ9jU<?`7xXE*`WEI-$Q@T7%JWSg3rnti?U+c!>}vWwVf@_}cUluKJ6
z{^bHoI#Yh2JPb^cT-+cpF<JM;<X16jM1h>M@D(|3fE;IRa|R!v|HayZgb*CaeE<HC
zkqHh=7I2UIJK&NP+>IB?iCjVx2SF@gqVn30mNm)$;sT_1jjLs4W^!ZCKuAgg9Z!><
zujS?10BJT4o;@tBqr*-UyjH{M)82gzmACm0P2mRD)a^@#t9CVxXK13<^JoFx=7iA0
za^JesYQA*%oXY=T`R66Oo~XHoGqdW4cbo7h&WcnAnOuJ_h-q03>jeh|Wf`?sI3-$6
zq@6b%N)PJpkOs(*3y^DO{XHbtJTNo#%|RVJUkkisTNF*%F+%+OoKWpw<<vo<O})%O
zF?mo?Ef2q`QxdokCchh#Mord<Smc5mEZZm`7p7qMJ>4?Bq1RXsqXZs}?XW-fH$s#a
z%Q|UlYMP5P$6xY2(QsR*0kX~|=e$4(<3y^c38P&xrw|acbFq>yaaBmStX9bFMbV4)
zsemocVQ>#_GFC!Jidt6t0~_jgP6n4rFt0K{-=3!iWV|+BHRFWD#JLukV7hR8?FJEL
zN2!U9jEjqhN3N@zTTF2>F5AxWBV%b&?JBJTCf#r*I=axQ#sIPN0-X-C;ojq%sg$x;
zos6pr{KC1w5c)rvRaNIS56oi(BH;VhFFL0tZVju%0^Td%PB#j~>6BU{>9xm4&U^p<
zZfpwXp_X|*Qm;%5RfO=rNm~XAmCV~Js>xyONX>EqeY+x8u-PpOn&y$W0xPwI6!rBv
zd4Q8h_oWXH59j_W@_UI}=OoS=u6qOLWPBRIyaAktlujiQB<*nyP56p_%CV(}w$v=K
zx>r~DIan&_O2$0fr%#szLf6&TVn$h*;wiZ;chuJ*DEkVn`x>RDrg`Kzq`s6z%s?TA
z2-%)Niyb!xm-fWGeY@II#3eUZ!fE9(Gm-2eedBClgSr_SD5Ti+qF3qz=yp`;z~|uZ
ztVv%j;Sx5AqqD*@3#RO1H@k5aKQ~+V=FJ=SwW2jCCLcpb0g62c?hdX87~#>#yLYeC
z*Y8Jkg09?2S3g%G^&Q(x0wBs3m)V2EjCWWI`6hzns(En&J|W0(;wE3|8RR-^2q%`-
zT*p%p(qyiXFm^IoLE5R+-9Xq>gG|y*`>gkMViUCRUJJ249P}IHxtMH2%);6g!yHn3
zug}^wYq+67<}tFWZzM^54OIGa0RW2yWtD&7Nykph8E`3n-vf{_JS9Q!VxvROr!?Mn
z{blRcMV8$hL@+feX=rsk01UC{GTDx5CTeJCl$xf)MD5c>Ev<Qh=MXtyv3)SJtW2R;
z0GglA;X|`Z;wCVo@NZ4H4}5brmjX}b&)E@w|Jhfm@3=g>?X8C{`^l1(<kzIMJ5h@@
zty`XwD0DS1w1YnZU=)*^&cg^qRpndv89CR^s>7esK#qF%>C-1G(+TT&=Wb;8|M^p!
zx;tb4sPt3o!{$$ttWR$8!;hb7us*=`xFB?<-Y;rX*@Mcs33X4(W%e7{Qpb-k9Xt3M
z$y;##EhKo$s+9OTx4ih`ij5T}aW;AsGHWMTTOVZa>8;*Aj0k?D{6%9bAeQj%@iR}k
z$j~86NoohH3)4|$yi1X-C2_Jhw6p^KJGW*iV$-`qT-<4B#nD_N;KhMRKy^IJ_oR=?
zG_2Ed7c^$SW&-JYBewx0?*cs#Q1p@Nkr$WlbTkKDUBLuBfN%ph>-0ttOcqz0+YFm~
zRwVkLg|uXYnSA_(6=Lp9!1%VfRAQ5U9q^2d^4pK=H5ZKG?JX<|7A|~=p`|&2XP>sK
z0O_se2y=Y@*#AcMtU--!l0ur|M~`wN>wTW{Vt~}bv+u5$=jXW-NXtu9unjdc7)<aa
zY{F(`@7aFI%rgcBy(_>4NUGajK<DH=VhR_EZFLZ#KK!nTw?TKTl)n=u3i?f{nmMaD
zJYM#%a2Z|(s7BqM`OXs2&-=C^jfV-(JG{IL>gxA^Q2Qasy^fAPq^`aZti;za%R_-b
zZBWpu71Yz$(AFz5L22WMOy`)C)T3w5mI4*GMIJ|)Ic!^~9$d6ucvnF^?jnN9Kxmyl
z{r<K3{y6y70MN^J!F^+gae7L(f&s5!QBdbTM)3ZC1P-U@$c-C(z(9{Qch$b14m|oF
z071cWA<K~ISVMOGU2g6I_(Nm@^Vf%z@6b44eNm)UobXzFZ-m5)!Qg(mAU{7Jb|x#=
z@LidM%J^4H%TJ^~gek)s;2zA>8ShZ$U~1$%dTTo*XXkI*uz`U@aXJGf>;<s;kT~88
z34uiGMyh7$(odTJ#&<za&kid-Cvffh^^1XQmdQnFX=!0pY6)y@=c26wbaKPy%`=dv
z9YU~V)B@#2ZS@i2o2Oy=nI9)BBU20KMYXa<xlo9pdbC*%{+t~jxW1_vygyVOhcAy|
zoon0L!kqSZQ-6-Q+mfC;jv&#;YrTi;?L1<VPQbI5FQ0%Yb=`|c%?p=RCMuO%K?C-v
zHn7G2phX#<O+-+BA)$}xk6}4<0$%AM!lqPWcyT6!vR|MB1irwIcH_2fvyp^Q=Tb#Q
zr4WM?@oAkv+t3fZ51*(2J*Z6vpM4wIz?&|QRA|oJxzF|>V(IKJHOAFJ<~;jv{b9T-
zznq6k&jtrs*YT?KYrJq45epW^X~<_12!({_K0rW-EP|{*ZNJ&z>9h%m@SlbT8fX+j
zudP*TksUq8NBr!9O4atz{aNpeik2!UC@|7gZ=%3`5EZot`5fb*%#~OGTtHk_F$yv3
ziFx%on5;QqYiHMmPmbvyaj)GBa8#EgyLynCdf=KH6*RXX4LSq=<`X6x?*Qt66~MgD
zsJg&d{8?W=9lpyAO-=tW&=Il5EQ&hR3w?8<!rH(zor7M_F-b|MsMxn}sWDgk>z5Tw
zy*6-hEk$bd?d`So!WN@*YTFzWJ32b3uh@B${7{vmCFSHCfAu(7?|D<yH0H4L%@bSO
zPXG)+8WxRcoX(&o0STN6kE8yf(h?fp(@>Gc7bocd{KsOnOT3|>!Kz{c0shR}tE`~j
z@2fUP&vGtGv!QDOAxLXyZQHhux(axGIswwp|D3BqX-udC3k%D<1Tp{3B^l;&OEV-s
zpz96xBWKyoU_ZZ2fa>q#4*GzzgiffHt*sbJ1k5QD(2qsf2vZv)VO-}AQK>V0fU@AO
zpB)N<>+j!J=v8CMvH)3NPWSk65ZzYXyMKRX&~At+x?~+fVS>^RQ;-}`!qjxc<TbwN
zR{w8i<VTy*?U^3EN7iVI-%e<VZ;lASRFr-HFo38(Ku7)+_9tu|4ndHhAGjwffG;^U
z0Vh{uzbwIKTP^_sY>pEpcL8yb&c`4+chVp(Vi$iXC|HaX($P{15Bw0PS0Pg6i~Eg3
z84}o)8l~EeL2a(&`+@GueV{94-Cy$aVMTcWd3wZ+H8>lW%EFn^!necQ4|RI9Tb(S4
z>Hs6khwW#)hO{`coS*2UKmBr1C&*6{^HFswYE&fxzeW<80KqIm)eyo6$qEb%q&SLY
zC=8L^RxKzm4@AoV8$X~A2md@%;O7+aj6>C9Y^MNY@0E(hXZe<B7-TktL}N{m(u&hU
zE;{%XAAkuo<{+YBob4f+>rl#lg1nEm9&#M<g9jO?kRL$s9ccF53rx}`zd|K>qCDp_
z(k8}lV?=!&JnZhq$L|Ho0VKl)goDL+ahVWMo@SQ1mG%LkYch-Q&5p`xyMzLmA9nI)
z-`~AspzC%N!Ii)f03a#12Npqu^c|p8hFL@uQ2qr_%_u}SA>YVGRLJGYlesAAeVz8B
zduoH+2RXp&vIzjQ=)C`vo0}W=eDR<7-9acZt*oq03)unLsKX2%T^PXTwhY}m7iZYm
z)I{Tj{Dw`o4c&}`Q2jtI_XrV)dXL!NPoY0)g}^_rqX+)1wUARej1iO4?2q5SGlNAX
zuZ*B&2T(lBX)D`<ryKpB-hQ}T^w=|5mZKKMM*hE3iq50ApDkql8Npx4FMx@Oi4#}J
zDo^oh$0OKV8eaQNJ|tQKNG|@o8@sGtQq5)49Ds{FaJ@jexd%zrrkc0Nce?wm_ecf~
zX6NN;0$ti((2U_9SpnZ1070%uvbj+O)>Vlc__w4K;9@S4xPG0xnG0*FE;+Ljg0xPJ
zH;s%gqQUzSt16|R4n!c}*&&cd#9Tx=1ySm~IJGGhD#kDN_Bl&zl@(Av2icV~L#X;5
z9hZ=3_<`f^@;$*|lOLX`M(t7sd&Wcoj?WzDBg+wF>qot?gPr}29vsaE-nmG(_xb3a
z$dJHN)*V7HIOye#R98hg(~x)a(u`?9ZR5KuUtgH6t%~+#E<&S^lVMDJyr!_(;D(mi
zQYW+l|2X%c`Id_wlDyP8MN}eMs9<&M7+>~fe?n?)r2nRMXX_?SETxycGm_&=TdFW;
z5?c(a%8q6WR56kKNvEs#9?;Hb$#01UtH2dat5|!a3qMmWxooeBKZeWBta@B}y%VN5
zA=Lv1<-@l%zzQ_n3_z8<{jcsnKM!l6ua%48$1lkab<F@&KmTfL;{mErH#+_q?*R#5
zuAmmc$a{^0&=-<L?9swIlz>A^Z5p~4>vehc0Y|(K(Vbr>I1IE(*ZTb7!-usni%2cH
z=5ZvP^bwUzFgX8UpLvF~60V$Ge-?%8VQ04<Nbbf#d|te4&5jX-01rRimu_y{2({rx
zyv#dBkSr=EPE1z|x@qyJcG@*nQ9t3(<(23W_1OPIK;aiZo$gd9|Ee_{wQTUqc?Mh|
z)<isaCC>?iL+p$CwzRVE-XwBaQ0C3<4!;#V$1w%EN<kP6>?J=$@TwCrAXftRZ)R53
z<D2QDHL@nghgPjxm9j}-E2<Ev)Sq{1goxigO#!n(PKbl^P!8(h9Z{DylwJwgdF!C3
zcRB>manlCe$|1jP-nC9)VPWTA0+h;1+EY<cG0%Pu)4Fx}yqAl`($WC|u`tK_s0ZQ(
zFt8B+a*?|_i9=X)pcnmB1@jBlIbXgU-@z(*G}~D)DX1_EsPMhcLx(aYFgp3uY~rg)
zp0|ivadYTSG#0{#i$<e>4in}fD`5B>ElD0EYE8)HpZBCAAG?-R!4x6mZasYLJAge+
zr)E6*nxLWavD*r6K;u(tMy9IZVu<xT?MoaP9-i0X{rvd{^y)oB(3C&FUwVn28yttU
z(6M<pW7e#B9WF0legTi&y)SX$dSR76OsIo_VPJ&+e0vRSZknLhkurh=)-yUT&av~_
z<;!0HuNrhvQVuft^5&al7`-K66R*r0G(TfOtjX6ydZlpUWYx16c%>4!5b<MT8Yi|+
zeO}mx%0xhX7UM(O5gsoPD1xOE=g+G-_FYC5-C(&ePi+vwJzC&lEj?R1G_6iX4Xv#s
zxm>M|l5^PRSs&iy7w8#tH%eq&UJ|D7<-5zcP_sHZK7R7#$@y(7A6;-or;Zz!Nw)^F
zfUHWBGcHc@H1{a^b+M{s3JOc(-GnEZho7#$ryitLs~fak9#aVK1)Bgm-5UvOeeI}*
zKw5$%&7?xqb37rt1`%j0Jf%1tuxv5*ps`NxOhf*y)h-13hIDfmuDodb`E%yf{)qBn
z4+DS3Bxp`$ysnFXoSbpC1#WU}zmsU4I|`bGRPIQ|WqYrNY^yX;?B=VanZ3kJ@phi8
zKltBErK3mC9C3O@Xn6K*W<*=qDXGH%`wI#IZj3g^p0~i{7DA8pN<Zs8SXM1yHn{+Z
zTrn1b*6`3ajjLDllhZ@9tM73`626*y4Fq@1EPGH;w{A)Nc?%)b<1%Et&x&h@d-+!H
zxQ*VIU}D;eoe8+raud}6<{3&3u?PM%f)Z7@21(3!b%f<S`n(^rOIq#VD4i@CItIfN
zwt~aR|75hcyy|2`7F@t%+-uRic##u0WA_-xu3a<MGrr{JUc_1)aT&me3ZMM}gE^~@
zH<tGs`}0?IprDZp^mcd80dgg80ZdBZ+5*+nthOsO;zsv`c=A0E)Kp6{dh$KZT&S5Z
z!~StA>Rc%)t?-Bl6Ngg5O@Y0zK!haGY>2xD5LT#18M-`gQikRlb7tR$9R8i}pe)ok
zHfD?hZ?%O}X6?ADs;Y-cVsdg=O4J1X%eQa$8G;9>H9FXJkIS)ey8DqQyM;7?`SM=<
z;mXqvkvW}Ly%n-7ttY?AtE^$(+wK@)GqOEHqIdz0iI+H9pknJ?UcC5tJ4%bGv2zE4
zqf<%h<42ETEZPRp2hyK_{mb#irR(<JdMdy5m6WAQO)LSH5+Vl<ctT8-xy=<j_fJO+
z{5^*(K=9J=5;hO*X7Tm&Q!25ZZ(gucF4v!Fq%jrEJ9Yt%S5n^z)x-V4|8~a>XT`U&
z{YW@wq~Da3lza(&aXSI?VO}!Nqdg%3sgaouO0UmPyJj(SB~L(L;C0~{t^?d$To))y
zZbICqkB?Kj?!8nBz^Nm?CDC>F6Nc1e+U|iNRsfV(L1r9%h?EvgULO<Jj-c&QSytvO
zzwVWkhYZqGFQ>@5!%-&IQziWuOYc$8sIelBj++xwH&pNj#2Ioe_z3&JLtbxC_bz8;
z<wzQ>NV5la>I&#Rdq599M%e?77nqum!i5iBLjmD2vhrfkZ!}J>TPWJYMpQ;HR8UhJ
z`wB6rIb{#LU|tu?;FBXr&q$v@#q|;$DsHrzbCE^3q94qKmd}#?O{f_>Ojceb1nzb1
zM=hF<Zz}MnPvzVNku*#OUu(h^Xs${J5r}<#_pa>~zZ@5unIGghaNwLwH3eqLP9!~o
zf*LK>C(Y-g8_ku~3l43AJ6Q!fm2fR&6<mSb%xTYTFb=Xc(D!E#io=^)LqNySb-)2s
z*6;AH+s?=59It|IUQw2to11lYd12ua3{xs3*5C?K$_W*2Ll5G~?)%ZvT&UJJtBqqf
zFIl&4_wXPJXCJiK?v0?U=(2ZVB$Ag-xkvz{&J%j*YT`<;Vb6kULbltio4_1>?Ms}{
zY#fSvZ;j^&Elr`s5;1LO1Wf0bE?T!g*kBkC9kCKR`>|B-k^X}f`M}+MEf_^$Ei49S
z;*V67Ga8@x)xq-c7+kKBYPu3Jg3?AJ!eE|e5fbml16$Es2F?&}j3j6fO_5s;Lr{cV
zx?;rxq?^8gEMRga1m4p|{>E<L047oG_j=K;sn$dJ^t_nlcK}R3Ms-w__2*nYAUL!w
zHe|2pJW}BnD9z<LDtmX8HjczcBxZ6Q@aW=3KQY?{Eyf{96_=r?AZ(xynVCm>qh-|H
z0ot7hD1C2)iJ3XN<1Er7^g-*ZYOv7g=KAK4{fsRP$<lxVUoOhh84)}oB}G(<#vH<O
zUSIeq7r5ue7yR$6p%pcN5})TEUagvHnvtlw^+m{m3;}PnSPjOdC>Vhp<!3Mq?V&?x
z9yjwL&-W0qTt~fc9L(CLrbi$dzN@S}%Xt;ej#+RADP?y+`w&H!2<A~&B8jIlA^qlj
zR-nvJ2VAk3J0j*UTC~a|upjn&L}b86nU3Zvk(!ZpJ^-uE*4CEaLFvtvm;OvX{&hBb
z=vAD1w}owsC$b;w?OMy6MYNvFto9EIdW=fPsv5k`Czq;I5^IHVKOTixMng_$>pjqw
z8>8<tf#P@scTq!*jJh3`Km%Qedn1nQ^7E+m7Ewv@HVtKzD9lvMOi*kGD@J)`U&rdJ
z;}6}Mpc~o+`j*d1b3I;Ze278((EQtj7*>wEK9R4c795P)<Q6zy=snLwVUF!P6AXe+
zz;g;?wNF*o!v)8I+_tj)B3v(HV-139*9@a>EpQP&;d!`lYoX~&B9{pW?Q{Vdtb6g|
z#eH0Ip~g}b{&`70_<mX(%>OPm|8MQRc{tWx`!0SrXw;xcNXk%#s7x73lX=KIlL(n*
zxRtplDG?PnA~F{;rHo0VGA488k|9%yO6Gm8p5OPk_xt^O|Fie89mo3~@2leetj}84
zx~}s)uk%uc2Kb`Cv{o&5sKt>DEvwh2cnFz~ycM76+*$EK?(Th>htMbcdAl=bVDML4
z%jnRgkvcwVgFF{#T-5#vjPhvq9vK@W*hy9N`ls>hd5b^3t*l%Fq(Rdt0m>1SZ+3Bh
zNC@ym6;Gcb!@D^D=b&}{y-=iaNu=i*Njr#=8`!CnJNGi~U)!K3%}m*6QoNb|_4%D7
zyF|75$$x2%19xyBCM}T0Oh_2yvC0xR*REgxn&8D!xyd7lf`1@V=V9k^57l-!3BeKv
zk&4}NzJ==%|0u=vC<-cEP6p*WK+s5Sf}&#W);&QT@WFpH8B9;XisgUx2^OfRPQlLk
z%AJ7nIvJz^|FnHindGR*tx3a>&x3=jjyzjhFAbNH{UZ{f;4wK-v$-6ZM)(rElFFVv
zi~LQrUD!MQHGo-|O=r%XIrC!nc?mZ3ke`@c^Cb|8^Jx!*>5{_6Eo9s>2#45i?uh+?
zC2r;&N0v0{WZ&_G@0=V91(fhc^eD4UJi;g`t%rw6l7gY;51;r|-r8LjPcI@YjACGQ
zs}$t&z$9G1O#mx`o+E#8|D;Za79E6ChlX50ut;EPN~ZP)$)u|iR0OIu9<thRE`*+g
ztOl522fyBRW)6;eBZxW6^cr<l6*}8g%vlbcp^%9{u1ArD(z8qHl<#0*_y^oQ>~Q2x
zEPf;|=I4eEdyE-q#{UCAvdXB{!=VY-$xJW&8k5=X7WlGb+ClzeG(^1$FtjW8c;kBh
z{H^^{72;@Ow5|}x-MBga8l%sn(7-@C=WC~_L|}z`OnLNie*OUnLPDFEgB>CEe%s0W
z3l~lba`v)!q|^ay3LN8rsHkQ3A#w3-Wv7c-{jXTuf~DuA%;pb|&MM4fvPxVh1|dsG
zi}>3Av0&rd^mq%W{P8XYzP~4a-Nucrf2KuAY`WgDf9V`M9<8}qCp(JEzm~gDXw8|n
zA8Te{K54-Iykz7u<@b?hY-?;^CetXOMvwfS&dtriZQ<j=2%(Y7KuM}*49Q%|uIsRS
zS_{ObtP{W@j9%?{vkwUize45X>Hi#Qu+qrayAKjA_Tat%^_1Y&i8}A|*RPc&<Ivs2
z#iy<dFZAy`XJus$%K>HT))TzejV2h*)A;P#NWdRWfHX5_*-Q)Em}KwX-wvqKrZM%p
zkD-G@O2{a$zXLfgqV}z-I<u4m81+2Ui^kAFpIhmP9VHM;LN!%aqQ?BF4x!qBYVJw!
zE}TGHT46#02k3K&qDb_W1F9KvIQtMTX|RcVc{c};+=Mr`tns)B2&3<llYL#iz%{WO
z`+Q!!{P=M@GTT9RgJGmZnWvI8@Jt+}ssXrg9%|`}o}T8A6K;n_vwGPZ{J?)Oov*=j
zt?3uWP7rlJ6o|s1pAE3%N)*eQa~<vNRZCG?$Bkg5*!t)M@>N5wu&!s``tC9nEnr$!
zgZZDC1mfy~yQpVxXv@IZ*gm`GIHZrNxdt9@OM&1Kg0ATOA&xd!(1@>x2|r|KGFTc0
z;g=zob4hV=&?ydzJ!2Xw5ZAEfF!5y}vWrE#hUCE1!*8~*{!*w8O0mE=CKf=d`bJB?
z$BjXI%cVRCWUawu#a$n5kEJL52I><o{gsu4#T=Y)Tqks@NEd3nv--|)x}S#Gu#Vl=
znp}ul<Jb$AeFEQA_Ku11AeRN^w6o*8fw+~3fEanF+V|Z-$j(B(HS^NKO1Uh2Ll^~Q
zajKt$=U7PjFy&T+(!9g_R><|9a7fwiMglhkiXS{>d>FrIIKC4P9u9H*+Vq>+4|i&#
zK|2SA@S}yE$XAABkl&oE5T7|K@7OoD*eYkT!YAIg0+qDw0+JfQwOJme$WPa-Ti4f}
zlNk~7!GkKb9UT+FMG<sn*LGc+o7!A$raSZ;8+Xb{VBI4%^DiNeyg9o&^fpgx2FF|(
zSfU48aR8J?&o$2$T|lDm-|sP5Cu|+wIDEQ$lvheB)~z*U{-8FhR=Fb*x88Mi?En`#
zIp8d44cC8}a~%i)t?65zWA8cd{h}P%p(R)OUs?eEB*1HB?~Du#h`P9}bG?W)CRz6T
zeyV--YB@0P%LAXhFYZT8c`ly4Vv9C2cSO|zOy47YRiF;OfB(L@Y#K%OBWp38K#M(f
z7-XoNkrVk}2WaWtO~_<nhoiXjh*{)in<hq;_>c0PZJNkBf%x&7wWh|e{}kq^Y$qM9
zy5}`CYqZv({{B39KK*MN^YD1suajkP`+_&>;2&uGadA4k{i@hU+V&2IA>wc6d#kyw
zPVM}YbJ(jrt5*R(&nTF?c^{GV62PrbfSE2%_OC@x^bN!w@}O`|5CJe8idYR@eAzKD
zDi@moW-yGM$qni#fq>jnMB~|dK#bw{k?bZ<*DVF?TGBDdgi`aERt%JdyC&v0BYr`B
z+Yq_0;x;}sRN)7Je^JbOk(10NLyra_<t7{k3IKfg4BODukNdQuH{%^Hi|-IbBDq@&
z;9OB%U4P`vy*qb!myE}5D)F^xA{=XiOPUeQCu!zBdq2V=pnmMHnW?G$Rn#Z?6B%Q8
z&soPUc7KG$#aor`JzrWx|Nh4$(dsJXnFnw`R5dLOc+O1nOo;2ouflAy<v`PW9=ru8
zQfGHqw*pF$JQqEiVA+c5L9D>{Z-^<U!Ar^K?emFmNI2}tdC}y!hS=p(kO2`2sdWw~
zW<T2eDG-khX1#%}_vHr<9#r#XE?X0(cnhP@CNECvE?%>ujO+g76TfKVwO@cgwJU;g
zdZ2V~q$o~?0up`~6nm5xsNq1btih^4s<9Qn;3Tec356@SPmKBMKhiRGaEQ2jmni+a
zo?h4l;aN~<s2qmSHBRdxsj)X?eiV9#L|#CnxuUML%#VAPp7AfyUI+dI4o25x1qClt
z%dQMUMEG-kX1CRj#I>+%@7`-}cZrZ>I5?2?LUT+hnVeMCnfgZT(s>6Fw?WmHu4@)H
zZ43N(Kkr9{IAhyp$EIJtjEQ%YBH@g21`6cKu#dogucB?KtE=k+%v+u~c5Xw@UrYL_
zS`?YnyUpZA{8|Aq61<zB*obzV5J+NIDaVT+^_}Hjl7`|I6*2#p<ItliJ^xaY!&Sle
zY3kiy_JTS)7!p0Xe`qOmD!W1z_sktJ`)#PN-WV-G9!UBv#y6^-pdN<;LdSDMkzLC=
zNU7sGk@fSG9PU7Puy~2h7%8&6+<)obg-)b2E6?dRh#gGfBNKR)MN)w+uL54xAFoD=
zX;?wTqzh}*rb^-#fFgyzr6rnATmH)v6${_Z1#TI?F$a$5QYT-YhX?B=yg)KRgh*Q<
zwI$MD&7E`&@0nOw)`H<b?ERY_`G=yi@(E~$>~C>gV%PqfEX3>VpESxVDoUf~N}_(U
zKeJModOVnF63EiYC@-xK=Kn*vulFD1E3FAfytjd$x%^A`gsAP%^76MweQXs&9oJ>X
zkD?eOU07!f|E$VK^W<xkqAR6?n6Ds6^g*@HukdnJxgWD>*+m9?NPSk#!#8QOvUWq9
zIJ==llTvv|QJKzo@zg)Tk(*91N_%A_8yrXGWyZ-req|n=AZyTJit{62HVqh;Fpjt_
z=)O~B9ii|pe)<$mA8O|<O@&J5v()3I0>OHH2J{_Xy)q~_OKV7~dX!ctb|krvw=DKl
z<h@(BG;HGYr_{|KDArbrjX<aQ8ffQ14Bu>>y>n+fjEdKx(u|C=Mvu)yhGVzy@Qr}H
zf78ilJo3Z|%91TzLZYuiEWTuG^u*rS$p&@!RdZQ2=^TMf*1pC0DfRck&yaJU1&8zT
z@wrVyE&wBwrL5MgwF~_6o;1FHF`=&4{KVGaVH$0&gj`~h(O3AqD+qlFHts6aAv2{o
z=+z3p8rs^ffh9zAx&)nl3X1FSsm*MiynYIq!H&Yh!g(%V<cZ!ryMdp|&>2(`Hu+8*
zDJ^TLgvm#)yzdp_rNQ=>a=Aj_&9njW!!mpN$B&mlHZ=aWeod<Bt`D+T+}5!dJ01$d
z%c19juOJ_EQPb0l`XrVtVfU2r#6)}1<^6?8#aC_a=KK}+a>FOqRRS9Y7V>=l-=!gk
z{r~AG$2ez@{U)ZS2a#k|O><F4FH>3)sFs%8I3ibYlrzMXT|op?pK6b7<>NoC06I7;
z|LvPM8nxUFh|GCllhY(TK>_ifTlSgDeTmta`Hg~|>E^^|ca(T1ke*wxi)3W3hX6@`
z1S8y_5QBHe2ImK3BzPP?Y|_k|&nRMx$#rxOWqu<~NnV_|0^JRN%d_Y!v%Bnuw6cM%
zz}klm96sj~#{%dF_}p>juD}f(wY2yRn)=k<CzqT$OMtFj0``NZR%5N&^<N9)1&&sX
zD*o^v2k3a2{jx4!?Z^8EI!Y*KfF{JH_#@|L!)MSDs7q$-WYm#v12Ze;l~<T|mhISV
z)<*oJci`3s0&@7gW`FX5cA(20KivEtEj~PVi$8B6P1N9|Z-{n0PggZeCDJ!IPhnk)
zFQIsK`!)rD6iugzxD@k1eIxf+Ureh8LG=SBc*F$Dj#|0ujsATbS9TZElUnlfn@`&D
z#EY-hmE{lX$xs81{v|HV0RWyI^{b9pWYQh`W9f~uBi>1N%7^*5?%H_Az9sq?goZ|?
z4vP%d0#%!8*`(WB|Hw0IaQ`f^J8PNX9N5QL3zpqTX&QU+@(iQC);FbI3q*2I-o<Eu
zMtgUq<bX9e9B2db>{|HI;JF;&D%2i&yJ<8*R3o78Az&^vLdR+M8*z&qF2}dxfh=v5
z`QF~r8}^hl@bir0ItUbPLV1Ix#hUJ=3ZKQRSFc{)blO&=kZ+yKhVPE<uXF+@+|r7#
zGAb~NWVC`uIP5#ih~c1pW9gB=!5~;T1A?2Z5^xVXg?d3=G$o%ufBpq<+-7W=ppcj+
zrVrA`d+U2GYITsF!Lh;2))%|-D?=EQEvty_W1%&;QRoo`f&311>*27xT&*gay(#&Q
z602$RYJfXIEhV+@-=DG-zuJdH<V_5yRZ#B)WOzR@1Qsz)!eu#XA~_^j2hgDs?`IZV
zfU8^7xn~E^V!Q6%^~v$;HOj-%<5k?yyr2LpwaC_92%;el*Zy}8UGor1{C<{=S^$mM
z1-91*k`)pjcnCo=)NYgN0O~?&cfre(E~zFN>eUaEPhAVX+=u}yq!k9O8RT+9TJHil
z2!(N+?l+7tgid-J?uH61!@BG;urM#64>26Y*!nPRu@u;>gC_q6>@kp*iV+RrGcPDr
z>PPJVZm8f<G<lFl4kgci!aEk+Y@wk|Wug$#@?&S`D7P5^gng&Ww&R(tvep0Xed2j=
z&dzQf(CePPuOR%?(fcHDDl{gc`L?1?sEnYMyPav+Wv!2)FQ96_+=;f!mX1AzCk$=e
zS0ycNc-0Cy3GTb<=k`ba7jFQ53@z_;b#<j#@BUEnvpUelOdY8W#*?J`J|8G1zzVkb
zYlv9BgP_D!#R(Y^E-K{><0!HqG9afs09O>V*7e;tb0@d6vL*wtLIW}#E=lrW+u7J2
zvuc2AP1K7)bv|IG4>b&$6Es|7CK-RP+2?~Dx>M`{^ro=zlEBKnGXg>Gi^N*T)$0V;
znd>aR{JVD_|L$*c_faUV_-%7DqrAMla?=j`ljiTl<d&H^oOId17Y0!69r{XOar6>B
zZVa0@QvgqNmHV;*fi&;lj|1%j=D_5s#|8*G6`&zomsSP>3dZ)#qszoaL_}aR&Vs~L
z9#O~K$tjG~ck+Q4TK@(x5%)maxKssx?#Vq7Rk_!IBYg%lA3;;_i*4qzC4EIhyVjP5
z#Xz_k1R(Yka>cA)fE+bP7n((_8?FC_*$nPBx^{^0p+qPP5MI+V-S0TWrl#FAyDK;*
zmFT;DJXTTA2BYEKJa-tBm|peC&UFuk6oZ{53!%H0A-XQ;i%F~aA*Zev049mULF+^%
z#9l;126XKuARrx&oh^~*dm<f>xUnXw_%I5nol)GcAUy!MV@Q9|=<#m|*q_0;g-<qu
z;VGVnE&dC%b879JaoLcip+SlVEP{*7i^S27)=0^TW<hIc7Gm)gct%>>Y0Y3qoU$V2
zXd%I3Q-{4?|4YuXrr|nHgi^;*+#HKV0Lq6pu~zK8Fr^=ruJutnB$zak-`IJC6}qlf
z->K__0aA|jK2H!!Blc|oysn~x*mjWxkPN{qXY$cafzY!8dbG%QnN9zktfsy>KyTmA
zn-jde;_72$LX%-yr<>POV||tuN%x9-)HHlt^E}3l5W{E%e#pHMpYsa~2iX~hpvg!F
z)-dr1%Y%s=bQ8Z{WfgC(Cd?XXG3>Bq{;eni3EI}KduW}&`Rj4?dE?RZ>FH@SPOQ)7
zd@j+5UVCWiNIE}Lip$p=O-^<XhQVEBsuHvJ3HR$2<K>Hx8a1%DPQ4SO_zzGQrxVi?
zo=ZWr^y$;5OTgThQvelx8XjH?T!u)0UbIc!lW^e#hra)hZ>|6?ge->h7{2=tApi64
z*WETFg5eq4IjL%nPDoD@0ZfI<Kjl;Y(-V#E>^E=Ujyz$oL849?9vhF?K_(I2w54<j
zS2lW}Hvm^Jn?s&@<HB8}2K@|$PpUdPV%<{pKpC`7FlDTh&f!}J%qoe2XDweQuXQ*%
z`lwAJ>Un9XMxgiw6|)T~_vHJVsJ_TqMO@~gCVz;6;4JUkG6bj!5T0BY)mJo!#6UiK
z1p?8A>BG8hx6vgEc}zdo5Acm^-ha|*{;}Xc>?KM?1lwm?$_S|0I=db+HE82_9_(cl
zSzB?^;x}I=^*EtU(X70`peYA~+dJObsUR2`#(Pd_i*e@N|8*QMuuOza6Hs#Bg--zD
z@;s-kL;GASu9*Z{Hwc`XrrHgKzlQ=0_51k?7gm4&b!Y)coH$Tmx0%U|I6;DieDXo-
z5m`1u<@irn3A4aMEf`E<=8eaIzV}P+cZA4%+yv`aAzpa9$};cq<JT}}i!2<LDx5*1
zA29?-JDvw75LO^-s)l_-9@Rk%8w7khBi_UO$ob3+1lgD4794$yv3$q^anUT&hR%_0
z`6s(1Bn*Bwo6^7jgqwfTs+_@qNxFkiM({mktA^tlU)M(fNQVBD9@H&76Qc{pp#l`7
z3Dd1ynRh?J=$ThR`o?zX6_bzV&ZA38ZUQrLU$N>2j!M}9KelhLuUB`Cf#)JQAWw*L
zFn%x_%?Is4wD|b!R+{SxZQqPo{%?o^Wu8&!k+zk3e;J()0zHn%S|_hwc(fl!?y%R-
zHMmO8g&x-1j_tehK{n}^>Ha-G<JRNo;o*@yIh2`7cL`wmb*X#v2>`hBqK+IsEE>x?
z2{DVgpn9Cd?F+qq9%j+5>RNHL`l=3G12P#BQl>7L@3Dc9EXX6^>`><8ivkxUoC&Yh
z<j<e9CBcZQ^B6lGiKONeK7if8gK<*$MnD(}v^-rmCpi(JxUAEiz4K#xyE!^{l}B5y
zn}KBXR9IVERi)YP30I4$vd$|O_b<O-?(C<EMzeN;#vzSynlpKAMVBntw8BvYvOu0K
z>e#`!1b67otgOM<i8Vz>eE!_h_1C*K`A(Ji>5-Ak9+j)GxzPpIzv=t++e=ai1VcO8
z5BS%*10#rxR|pX<zr%+y?n7MZFlx9pRi1sSmYcB+g}nNJb9pkWNDDlpi>4X@M1AV-
z=a#faNRmf8*Tjl?x(rlC_^2b7Ne=rwqGAzxp)%!z5u%X4Tml{LgWb;J{d-JPn%I)&
zmxo{%=KiE8_*D#(CW=@zAEVEXjUb*VbV$d0xg$uxaH%wV6_i_Oe(SFRO!x+FLb6Bx
z+&|@zGCrD)^~<`(8|)^{{DPzNYpKv03LH&>NZUnrn}-7zV-dxSqxk#uw7!6&=~lZT
zuesmZ{l)j8)Pq=54zdrqO^qo6z$jotLhJv6E9%wLx>v6*PsR4V(r30jb4DYb>mxj*
z8aDnFgUIfM>=_O2f_A+i()0Ku{`i^jFG{T6uz@LZ5!I8d{zb~z?RCM`%}V&BmQ*3#
z9ZAKByfW-P&<<<9%aC^M!QV27e-VhT>l+2NRQ3X7!YBcv_j5qUuED?X<oKQl;zAVH
zS$zLdOCj{bc8;vocy&`w(04F9SIA;12(j$#0A_!E!BqYSPuOB>xt~0NKOn-`|Bsuj
zgqc`3A!*B)OgpmT&{X&WH_F8#0U5>cz0%Tbq*d?8sWH(IpEf*ucn+?+tgGGV&O6(X
zEN7QSo$OZv5?vLormFfK1J{3iGwb>`3K2T<Gv`HbA`(Jme_3L-s3`)+pPRVCHHYDn
zV(4zjsBngV-J?*5;=GC<@V*h(&D1%L%Setr`uW*#MzHaW&ku>&D(-i{CEOoxm9)#^
zE~?i@eX>-jm|o7c!<@IDD}WEv-QOX!dETU=@wM0|c5|>JaY_K+pgBxhQUMa3Kr8hp
z2t$e|kltN9d$;d^a@=;1sec(h@nA<q7ukamzmY3ifJ4oq%I&sx_3*fQ@7~D$w5~1_
z^yX;=Jqy#)(lT&~2?$s*l%6^>Q3td~!lmzu(E$ItKi@V0zPCFn!3%{Sapx%r*F&A;
zd9E6&RK3#QQ_@H%KD`bKWDz0yt9c0934YjM`wV^K1keYQaRp$<<ToPOVA~%GccDqi
zH7s!Hajx}4Gf>ZMIiRX<=eozge&x;7-q!1krs%7nZZZ$w%fpg|6d*<SnHi(kvV>n2
zK=Ujru2ROhsvHS;Ys9NO1&nzK=vHMkl4y4sR^e#qFk5SL&@hC4lxxt}D85(kY`cvv
zLVM9!*gAbdr*(LZt`5q7I!$IvF#3pZh61*0TVf9T%^T#IS|Sf9qsl?iz}sB<4<Luh
zjrXBw2?Q7wd<qGJF_iKrkvG#VVWT{+XX~6_nC=a)|3@7mfUxTru<!|gD>+QkwLj>S
z<a*Q9!GCgoU`Kv`YfL;HHvM~w&cRfa-|k8j@Z`KGXkFiznBgdzJ3FI2tSJxRp+L!_
zT}xXB0j`FJwWJ~XZ}VWF$*3P>x(_nnpxmd#Xa?Qw76c8LCN4(y?s~}7-*c2hSAw}M
zaXykzQRaP7(R|m&SDGmQzESUF7n(9W*~~CP=+nr^I@q+d&Z5f%0(|51jD^TvZPR>4
zGT~edW0LnJ;|NVQFl#MzXeTLa+Bkmg_g}v%fCZVKJ9mxP1VJCkKShu3XMpiT0#Z^&
zU)-vl;qd`}t-q@fT^M9(vaanUqHeFBHa)g~_6Z6$KxZ5|L!@G1<*C`L>B@Of&MPm~
zU};|?qvB4R!kv^RUyMuAG|#-CxVsH>znmClc5dzq)^~D%Pz&!JZuwhgVh3w(5GAHC
zNI!I#|14?8`4y+X5r^t9`9W%1+WaFmaA4^l#Y9DQ@^;Gw73QV@(S9Vg3rx^zh%FP0
ztOOj;@OXaKf4tk%MAvHZ#gF~VFR!cQ<P6!x&l}9fS=kq^EtH{cm@=qUk~-tYHz(<o
z)=D!sjIg?YKO-$IY^v#P!t))E8@d$K92twcwlH^n)Y;sAua~3zQ^mkNx03VwO%m(p
zeo5srPiOVINlqWG_q{jad&>REV19LaC-h}ed`fQ_wW+rtTJnUEcc@<+71B~q%=)Ar
zO>uH^n#8zK1D^6pZ1Y>+DsK|uI09h}qQVBLqV;L1sg*b@mXH(lp$`{{Ra4fANKnfM
zyGuheI5KtjdbM3ZgisArq|?!Dj>&q+|Ag-P_STqMwR>|T4|A)Jy!s5^c=J}9L`?56
z+tAlwuf0RYr}%jOZ{SUt<2SuIxwxniqWmfEn^P1vJ>)rZH&etHgD8Y;8@U5!Px$$j
z<L}8&FYV*p!olHu;8mF*#wI9TXh*lFTH!e;Yn7mG=YwJj@|MkpWfxsCRw(nPxsR$?
zd3yaL3Dv9>_~_Ug)UN&Ti$9n-w1^o6dzRYn%j*c8nwZ_VX;Yiv(L`t1W2{=?2dPC3
zc5vNdl2sCjv>NpO1)rY#^QRJH3{;)8yQ`_QC3wPS$Lgkf9!_Cgv#uOX9c>M;U<z4$
zUsEG8n<Pg~ZoBUfGn*2?E`c2W5aQ~wJSsu(-IKse_9)SN95;RoOx6nA|DLU+N#D73
z6Q@9UhN}<CFJbD4P94W8%rdORukVM!c{m^gWgnlim%P8caI)M-#k~R~3@rdvU`h6)
z4L0AhYD=9P`gFs|hlWI1?ORQkS7?x5_f<b3N;-=o=G^Pt0jgaI?m{(Z>g8&V%IAx}
za+BK-$4hTgw=&GU=2+DDISt5Gg}t4Khpqf>c>mBe;<sCwjKsYXL8^9V+6=;TKFaRo
zwxpyarY-um!H2d5A670C9YASM_R*aq9RX1{Wo9cauQ_4LQ`yf(6<2)BWKVQvHK_fJ
zm7-?8RAgVmtQ$eV$70g=U}HQ3)8q|{92Iz#RiE@bnPLtDH{__B0gX`TwLB*Z5QpdN
z@B_h~o9I=jVz>n*z&;>Ax4?W{4-Jh(c(Dc5T(xQx&LP{o_B-$Z2DtF#8&bbYv<X(!
z)Tp4*hFe@*5B<m7NTf+0CHePwBUMQA`6cb92s4eY)r^v?#KdD5s3i8|_iyrhhv8uX
z%(L=PuTRR;=O_s6!)){#k%Nmx<RQhaRUWze2uA$d#rikxkkG(Bv~B$jmOA?HC%J$&
zYP(4{)AZ8FQPR1dn#$chfQxSuw(BglA}u!Ez>5Ope_iPDR1k@I^Z{Tey`PC?Sa?dk
zu1>c#pg8NBF#OL4#2nz3y^ea9Mx$vdTqyBdC?&Ivw($fQ%}Me1m`6uP@5N?{hL7@5
z>vs{iZ||NOtZVX>s=~Xb9}x%Pb-=sVDowwr9ltM-{c&;IX>T>OwC(FrcXxks7;|Gv
zYk+?6Aw6q^&&c(V5b#W~iVhB0Rbgqc7zzS`Rq@C$85GK%=e_X~E+_7Jjqm|#C`UdZ
zkHDJ-B!|R?o}L~8S`lL=w8kWENf9@z^78pH3>dJqb2xPH2yEIOhdrDYB}3?0Z*Z>B
z+1niR^BJbJxS~RS+_rzlW-L?V97H5{A)amTpM{1h5@iBy?jpo!!sywFSQ<hb7#sa^
z2(s?a$?gTG5^xc1AKd=)UpO%Es$%Sp(CIQ8*H4$KGeuf5v=wm8<dkIPkElX>SB~W0
zuq**JTWprMph>X?Qr%R1bmW>`kVOcCrtR*R1pgN!eIcj9#gJcESPS~O^rt>p&n-+$
zJg90Lq2EA9dI&mQj0n4l;yM!MP1QJ{waPt@1F@%u_vEspAr$5|X%HuGVP@WoO%{d4
zQ^G|tIJ>yGXuT;jQU9E9BxSWPptpR;sslVG@mwnCZE#QM)9S)iV~G~HRV&i)k*gp=
z7DN&!TH;6*Xd51CL~ryyC}X0L-NzRAIXUGVf2sn>YStusSr&33%MTl$fs57Z^n|hg
z*%>f5HNL9$W(wFMhyIw)p=24r91(#gn-MH;3%h%w@LUO$o9+s2xi?BfA%-u5y}n+1
z8yY$6MYa5skT_IB+4crFJOr#fSQIFU)D}Ozs;POX%-w~_UB>}8i$)I&4b`BE;=u_6
z^2vj%8b6gA*CIJ@fJ9L&(3Dwby0q{Eqrfkd#%58wW<_xI<Zu#{fR|ShB-)wDsufs`
zXdJV<RCLzt>*lD}ELwJn^~5p?J%$KqT@U)!49%JGwRCienQv~wY$4+jaR3<tMoXAm
zCN^gPkU_BH{kwPX#?GQ?g?Rxb+Ii}aKDhvDM0j{JC_9JqJF)a5h!1;iy1)u5z!4!Y
z1x9l@xF_garGbs$V<&83Xn~jGz{G^}?>W?V`vT6EGiyiwmlgmHkBWHeyIuFQAZ)!a
zD~obFoiyDf_{}Qe3F>X0K?hj?^&t}@qvhr%AcUgR-l$Xqe!;KI(nA#$8*hgbLVoOM
z|G+>76ss&d$b?Bf=&J6%^g<zL2SHmV_NcO~4o1mDK6r4hK^!7KVS&Y0qm__M)Wu8d
z1B7MCBIM;m<WK>QXza;^0fgGUQK(WLjE#-a7qT5Va|y!PXmsg_87(1c9f9|`76hiH
z>wqJ2Cc`T6VaY^zLOsBZJ2xt}VER)vAc3jV1=yR9ryrw9uiLu^r(M;_8=a;^D>zwh
zC=ofo#DsCViTAv}Nk0Cq-O&ulI1&bU-dVV5zB(cB7(@(7qfNfs5#cehqz>bJY$K<H
zhkSkD+wNY3W}AR2H-Vej5SeJ4HS{@!TfB9C%r@rW*{g#xJ2AcCVM0P0fEJ^X5(J@6
zDb+NZ7DOf*aO&r{(T9nr@0!U&f+{+b=Xd=0KOJyhV1+we!-6QPYm}dLSzq)9Tz%p4
zxT=#WN(nrm9R$>?4h{|uX5v_8q5=+Z1aM+9;sQv~KCceO{Yl7v&gH$w3v5A=6kOd1
zye7aATP^A75_Y358J|W)mG~g%^fMzn6D!?|{$p@g^}?i$W?a~!N83=(jG!$`2eh|f
za22|=({l{c`<GreAIU53@3#yXhp;ZIqz93RKIb!T)=Y$r3>;_HtmxaflgJ+;G|__6
z0dlg9A(d#ct}xSVYe2irhC&xt*S~KLT~4&YS;BY`KM~0J^wBV_EbCtL9_HAsXL`M;
zyX-Ln{mJ07M(`xv#V)VM7HBt0{yafe648m6*_@UC5G%B28b=}-H{?DcuJtulxR1Nf
z;y0>XFxIA$zQkdtHYggp$4_7UO!qJ@T5{qn9B#iA0(cfLaWss~dS@(!;Z$u|UqB+-
zI4uF`GWs>2l_d2fnx>DeJ7cM-@f$~EQvX7o4*RhT^fnZP9gnXPh2a#oQ7*usG;(&p
zvn7`FK!DPMP($ChY3*7~Iy$=8QD%RC<)=m4ih$4toF(3aZqL9hvt+fwXRH9<Oqm=;
zN@9nehH7}@T;JXn1Gi2syLnI#nGnO_*r4yU1O65-aTD~2WvT}U8Rc7EuX3w<|6aUw
z2z5**$%A4Ctgw5e{1?lNQ1y(!gcStmq4u}hG1U90K<s-S0E4Z5!T~FzZtv4o6PI|7
zz)!?F@>*KKNe*xnxN6?<TRGcWowI6<fu3zc$2-Az)n_*$q$usg(1DJAZ>`gq!?l!x
zQ3+NJlB7{%{Ntn7SznL?w0+uX>Vuu@92l3v15Vd=vI||)wxgRv9D2$OaUrq285$YM
z#Jnj-DvT0RemQrSreR#(dZrya9P&b83xyd&H)~~_bcJv|2%CLgXaGKs8$%U3e9i7M
z&nBB^moG0bfNxg|l->b%py<hdIuZP>X}&38(Q>m2^r|;Tb|@d2<zS3L;`4h|n=I(O
z+TiEOIVnJ(SR)4zzU@$x$1a2-CXL`2YeJwU=X=3xVRvMEJq{&twktCWV>|SI%=gah
zDGqjaN7qhru8~RH(e;aLoj?YcU}W-*`Sc1TX3a-VSxvhwzynRrF{t$pRtjuwTelM$
zHultK?*mtKwqWYVuU{<-P7kbrx5v)8P^r3*rRzy6*ryAP^H4M@=0j5*D7;Bh?!tQ4
zQpYfMLN|2hk&a#OQMWuq>c)hwY}yMRi_(Z(iR`7@aF#-Zj{wFgtjOx&?yduVA_cLa
zpnakZi13ZRsmVz#oYYJ-@yInn@?Go4p~A0PEhK~LSA}l%>W3)$nb0r>kGD`(SvgsV
zDAbF#<|IetM{41v>c&1=jp&ci@eo8FA2V}wvK3esJ)CNBn**YU_~tsDpX-9o6#J-;
zGE^5y@rK{}^HT6?y=6G`)Uql(`P8dEZ3HLXix-PA3|t#)o(Y61HlAL0H=i+I8VK`b
z&s_eRU$#nGg<8rG#8qd>UvJz(0nFHpR_uLtf6%rY12vkcXI!;YmexpvcUGIH<u^ze
zQ_&OXAnaL^k5B8>yelPI)$;w%k7sXYAX&ZZTL$2{Zo%hQi$0U9UsU=jrFP{bc6T_$
z9qw4ip<iBZ;380<;Bao`Q^W;HOsqilU3TVxSJlcC#QE)_ho6>|WFSViAU7KL_RYzC
zBeYL?B%}N_4_5wR%$3bswj{&CRu>Pxq52`fwoF7umMmkV6M9u)8=zEX+P3XJ63Goc
z0T>yR^wa;&#_YKibOsXXWAOFsnqU%0jEuD!L54w{HfE`^-2d!nxTY#uVn8_B<S!uG
zs($%0$;J@rYcq%;`qo2-_?}`9Bc_R2yjfFMpN3S~uDARdSzhugX8$q$tiu$JH%M|l
zk@EfJe_oCG8idp0qoW@{Hf%Vu&>4572rXFKyK<{VNLU0lap&+UtG!l`-KC?hZAhNl
z;z`rVgC}*wkX}<nJ|yj86Cr@R<&XMgugg5?t>md2ed_JVEt(p_UUnh<g$vWk9zo~}
zuH!J1mXyqZ560JTMc7Ap-VGh0>-Lhp;}ILa)(a<2WcE>nq*6Nw7Hv?L2S4Tj3g<^^
zRe;Q!;Jg(F(L>l>nBPbIdFNjD;XXqwdXPMu(0busCLRQ_z4PsN^l`*K<4_6DhltNc
zPx_kV_y_EBT_m&vLT=+@W6h98sNhT@e4c)mi8`s;_jJY4twz@`2RhuQWz)2JeWpTR
zPN(LpQgQ!lOSb?e5t~LQX45+9qbfQNQXSH0<nDcWiNAs*<t@-FF)=Zvfol+3Jp|*s
zW_DS5^?&12LeBzJHTnhcgQ3D^PEJ<Vrga(c7q!-90;TFx&{D;=ok2X&frwGp(lXWV
zV_wqBe9Y2(k+@LPl{yw&7WR9h*O~F$=U+QnSu>#yCNX{;96qv6>PDuHLHK2JS_Kz?
zb<2+@7D6(SQRD&8TqFLFUgSaVuV>RekH5ZCff8b5!hh*$)!C2938+O)OUHw)Viy0r
zI~u#AIjxk(YBe2Q1^~IJ;g86?n!4Ix*qi0y*Q#aOp+ZRv-p$8GQO?{Lf1`zmPkQg(
zcm2#yUsbiYmozsx1Y^L2FsssWdyvW6gxyun5Ej$pOKXO2lS=pVj8)N%Hau3*#0)+I
zO(p3FlOx}XgvmA>!QjDz50Afm{rV*Kry5^cbtLVXZPOm=!Ymbc##61EBp1yOI8lQ`
zRI6OdCnG5(m5JJW)Pq~k<E*ZQMY44qSS)^N>CP-)gIJm`aEQn9V#0@(eL|M3jS9_b
z_Bnj^(=qWaqdAo)&7sKv;TuaWpg)<9`g$<vr_sN!9MGdaG{}6~eV&L)EoO_8(ZZo0
zk`G<ZynEL;X3J9n6{<?L5DFnFkQI-GO&>s4F#<7+xS{-TM&>}N-j*hONOx)e5Or|S
ztvgs0>tRvjEZUgR+6?DV;)zO9<A9kosw~v!OrhT$>>5?}iC>eXmWOMjplL?-rAij2
z3P(d10*$(0N$|u_ur;KAY4~?1(7b-8h-g=d8ATvJjFoh|{|)0tBJx$j-WW>!8g6>`
z_cUiWr`%pr_U^4!yI50OtAkllAI9cI#Yn>-f=W!wj6wTa`9J`w=YGm^xa~n|^6S0y
zgiTnNu9lj5bt~KO6R3XljvTVj*f`EJpld3{^6;dVI2Cp};rS|evBz707pna|!;xOJ
zE5r{(uNK2f;;OL`?3RQN@fl|*q|nDsS>{0CeAg#gRtfLk4$VBVN1G3c1zx>MV%PAu
zqWn=+9)&;q;(Gj-d>>Dn|Bdr%aT1T3r|Y^rz7k3HFZ`z3gqr~5w)v{>3l$tuMT%jE
z`x2>rcI{AW(#h2Y$f>kC)f^&ec<%nJz7=+NL|~c{CF!E%yt0MNyJT{=j>>TQ8|8aZ
zs;0K9%kGN<6s2u9lldSsYV@aDOnvi0Lo@oL%|<E*A{!pe0A#&O^)H0#3m3mKpYAuD
z!_qLU{*0LJ+hH~2%KG3*n!3*6T8DQ5KY-sl31sRvtup)6s)cxY4|r9Fv|2hnAH@xc
zhs~~@_aDYUKPf0mGv<nPkGA2?XhscVHaPjJr{^?MO8x0rft>ts(r4U)@I*~eH0kL7
z`ZXPv#P&bmR5af4TIXupJLiy8`XriijA^RKXgTE=or}9W4V79Y%7VS>hksl_0nM@x
zy$0oF7&qk#Mwdm}BddKb8w5Vdt}vc<H_-Y+9l6*)m>OsZFd&VniGzLN*OQO^Nj{~7
zlvI>G$9;lggZAPD=R^YpuZXBjZQDRkPs0EzIdEE>VCKY1oUA;j>JC}q=4-?@cY%kG
z5lSa*_#G9J1ePRt4Uf=2HUOM)5=oIV?)!#G1a4CG#7R+=5z(rjKx%gpc4w(X(29QU
zT{v8^qn;NpD&pr3;3!7`O(L~D&#qmGCPQZ(v3yZbW^~#3d3lxM$x5^=K*vQRfiQIR
zN55k<p!OP=s-{7z*7x~yQpvrPloV)L3;}!bLFRS~gjq0>q?W$CLX;gfD82`PBMO7H
zwuPz`8YBCh(a})>vaK<w&jg0P0~p_RD={%q>!u5$Ox+Bi<VJX6h4dXt1KJ@BPDK(m
z3e0D;k)EfNoW^qT2nn5nZM*;&2Vs=c8M-y|x!h*Hzaj_caj_}k&{s1_-+<+^;;94~
zL2USJ_+QIT_rUYTbQtwG6ix;x6?TzPWhK?=BRV;#@l_D@R1;)OQ8V3$LL+HqIFE`Y
z7|lkM?vJNCfhEE!f`+?V6JT&?)CF8Rgs}%@B-B>WgPL`w-hYvcP(Eg6y^xrMp@auy
zco}{68mNnIEZz(p8y`;t)C0WN0Seu?`8NLr{lGWc+aRvXhnBz)1iP%9oNZ4u5Rw7x
z;%Gwn0MQVXIxT0az{v?bB9a_mg2)Uvq+GK;q9)6|Yu7PQHrybLh{;bSNRNlI?Ao&K
zM^{775Wu~6uf8%p+>Y~%N(}M%hQ;TT=7E$>;y>7Drf*RFyP2Y$aJc1Rv=4B7%sJsE
z=={PyT{>K!;mBGl0rn8|d(TleKn=IB#8oJ^`XNx7#9(w}K`I&VVAc{lIFIDETgrl(
ztY74bG*Cct%Zt)<kzr$$N@bV?H;FmIv4`d3rKqV$8z#}pRs$>)je9~&Tht9#u3k-W
z@_sZNL?SS;3c}KZh7o|uXe8JNncFRtss`?5sCKJyQw;<%pFYqL4WgPHCp5|vf03pI
zr)<&zg1PeACFeh1Ca)8k+eE`9_zXoOJ%uZ18G7MPl5K%3Em>~JQq9XM@N~Nt1LzhW
zl)ifHfFf==Afx{(LPke+|6UGd7YpcKa)dA<qs`S{J&S(h#*JhJYoK=edyG&dDq*!m
z+o<@^&<$dQOfb=uf#T&ye~3Uqw`{JUqU=7WXhAh32?MbInXEh1TZtq1(EG^Bl`AW8
zN(ON9Xehx|5ZG#EAUnGXICAMg>_mUR&V3oWJA(j6&HQH4U_V_AP&O65DW$z~V0L$t
z1OvjLD%^~@?bgm=BU(94^=b9Jy=Lf2<jF0c%|si0i)ClowQJYNu*S)o{_`h^U=}c<
z-pvPDh}vAB528r~f{RsX6f9b6P(s7Fi%Zyd);%OlSF`PQWy;d)qf0aDl0aqA_@zAk
zv;@f}b}WI2xzQ_!Dc*cYp$BC4+LJaoumJW#53@>1Z3%ln14n;?$ALcreHcQ|CSGf~
zbP8|HigYI;hcZUR!-EZ$XSAbr5Uf2px5Ow6y_A9g^&7IMpO%(Z16kRP%fl7sw&cec
zw#8+X61YY4+eekj@855s`I8VB_)Rs)CsMJwN*8u>aw1QIu5vnACI>Votwg2_oF9R}
z8|D#$b*@kNQYfQ&+RA*!tCvx9tH8Zm0yr<N0W-sg!tvpopc|kt$;JygjY2~OC5W06
z>5i|euV=ok5Sj?V$e~e#=ZmxU4a@-m$?neQ_xJU2Lj1%znSPdxWm#487(x!f4B?Rf
z6G<Cl^+VB>gc!pv;*pj%2HsgwUtgd8Xcirha*txB{FtJ_u7WkBL2?BeJL*;PV<22x
zP!_w)b8-UOfWNd5njpxJ#1z@Skr94mtX9h6#TO>-!A~&tD;3v*(@B3NkiFqRofsHw
zkq5VjenHmniuN@nb#-PO{vT6QP4lmaRgCAK?@2+2l_gs2bV#ph8upl6?jgqxZuO%J
zPhLu)p8}vT6F$o}9E%^*(?v%4!CJ~BjN?Xjg)9Y35QDSxgs|A=sCgeGL;5HKvbH4L
zBO|*T5kymF2**b5nbHN6UHynxu{9*Ymo`=@g+nQECfd_>{pL;P)vK>*P3wkhDWfRa
z4J<4ilmzMq28fevo6~PQ<;U;=jEMv%Vka~;H5GxzyKYC=TL}WqYBt%%z@Ryp`W?><
z0xb%jdrA<)LwX%lR8(})+dJC!Xdv+(Y(G<gqFH$&o&0Wq<tK4pbK!LN4G;6-aFkxR
zNy96ShQcwWf8XlWt4Z93smq&nn#R#R=N=d1#el-_<B`*mQRY*Er5O+i6(IEJQ4WTE
z7aEunRh51XX7UfrVu~~`S%r|(4*<2gZ{bht2gu-xOV0&|1u^!nL!OvEh)S~Q$;ZG=
zxCMwWt@t8bGI5JrHjLu!M`i>}VqIM!Nq~a}4JKpBGw241y{<iQJ33llCs#>DB?4}2
z*KrLarbZ9<_~MNr9G+;fcdH3U*GfAQiiS}v7hb!_=pC|%I3`N4bR_*&*oC1zy`XO^
zL7a!29({J9urzGlK!3lYt7|?+!zg-q6v7zsdO|{i^|_RL2tCzEeL_y7M<%njs?SOc
z@xl|7b?9I{Onni_^+ta4oE6GJC1M@FL;@y?@6sAm6BBM>;r!+W8*o{-q0dposzKsU
zvIp{-a&w(y!mHP@AN4bVNr+f922+jf>54833xP_O2BIWNl8lKd@trFK+p2<Pg8)Ve
zpmud+9u+$qG2&n>dWCrbf;B8K7<dwl5u-k9EN{TQ;WjRVlo?P}u5dhA$9Gmxm%T%!
zT7Zk|qzFrTF?OdiEY1Y@I30N=oYPvC@u%!adv6&77#=CdZKztRDl0cbtM&r{2hbEP
zLzT!-usA`A?bR^$bE-o19DG_qPEH9$PhH1vyxqSqtgbE+@H|;N&6o%fFz8Md^kUz-
zHDx9~58G(ce`(sEle00$R2tN^8GY2y;9xa5CHVJ!gM)iu&tSb!{LUn>*9^vovV4EJ
zB%~gXd)G5`Hpjp=A<Hm|sLI42*0~Zj*4?`gOR`XwT1U!3ehdzky?emQ_L|Tz5;NDS
zK{c{hpiOYOa3LB=1np^EDt}818v6K1P(f?~yU_?GFEBJg{kv7T@dxm+#AAEWhAWAj
zE+*%!3L$K;VEx#W;W06M0JN9^?1#v#$DWGBg;HU&NWv*3y-`<;%~4hGki6vEB;mUN
z0UBOsh^)t@7qXZ@A{gR6NHiwl11EI(0Ym23B3cA@1b~Sk*&uSU*TQ?v_2m<+U6lD?
z;}j0fwJ|8vH{cma;<mWDVs5qSzgxDu#7zP2bApIjTIja@fo+3h{CG6F@aqr;N!@<&
z;zhDw^$FCuI20y!-pLsgKgp3}U@+I)0a8n>Sb?S!<<}Mt(NRC5o19}Bj6}#saNQuT
z6FFNlIXOvQOn6k3UUQL=K@p#N=Ue<~|6S+ugRw`^IBJMr8#ArH6N^a}y97ZQW`NrY
zx^27f){G#$MiT(5&J1}Uxek+k^{<RHo?cGTtz8wKzbx@eY*&>*gR;FPMfvP0E%7IE
z3JNut|4u)MvabpP6jkx^#K8|LW1b)|`puSR^&@xMHtq1o`yWvEpN(jk&+J;OuN|K6
zpR25LV0rCJs*Vis_9zH~lP9(Xf^F-E_+dcVME=)f5dSPt;xx<`^?0e^NSqA_$|5sx
z@-e9{Wk$DP>#0=hqUaA*27WYZ-qptN{3J(Qfh3`0t>ch#PsdgJb``_RZQwxmArn1x
z2Zv)Xn44QDG&}e33cxfv6ln`<tsC+q+?z)8j48ode?QyTq03_D*(2+L0`&)Mk~t8I
zaqYq|QGK$3^RggX0IET-&<}cgd)MGhRmX2O(}gD86xMhaKVMfOh@m>riFBAseYTQP
zZFRDR&6N(5H?4l7^xj^ltz5M?!N(%N=FED;<wLQs5<Fs9?&DdYUb&57=FgX#p@RfN
zNA718U_+rWRe_zvdA1Z?!d54YIh*WC;}4mgXg<_Kg<~6FlO+HH7$j9zqDMguU8QHy
znpw7tl9>htvZAW$jfVN>3**l6>M<wkQfb7Ce0!Zg2436Ap7@P&D0EA{oGiSgG_AJg
zFQW_`A_|SYj<=f^=Uh-yrd)AgxQ&U>Ssi;VLxKn^IHe@!{Op%rq-y0V3XMal9S<gU
z43%3N!iM}56!?6Io#H`jBFB+AipF9}{Dcy6cgM}4)bXBYou7UT2jXeCq3y<BqG&WA
z7~h|{GldlCbCR=|8MunEu9BuR_mD+HN8Xru)4ecKU1sKu4fjQc+h9?Q=eL_sUQE>B
z=CTppf9Q}tq>Ro1HX7~3s)<D2mt{y|NU(QgyiEln<}E*krm=zSqMpooIBGCK%Ax`t
zpt<N0NT*3;_U*ekb;Ps%a?aiJlm{tpLlzW4S7vG64Ixm98jd@sa~CJ&m|Sus+%OZb
z8swe%2SC&ITJP_l{y`G?k(W~^YJT3Q9X#aAG7SAg%g*Ic&iov79&fLk`tie@=0Hh*
zNyuWbPGYhCP#D4KDPjZX2`eec+>i*c)Xq&MEFl$Q5D%{OLie4h!VKYR;G)yEiUBY4
zG3=*Vu&Ep7QJYjFpgC@4;E&MjgRG%ORDmVii6p9!4B~_x>bZClMYzDxU6f;K7f~Ig
zVMwC<RIk^lpkL~ev_`(u7gWhHl_u<h26@K_rStdjf;Tndy63T;a-d&hmLIP*eIuBm
ze$)U+a?B6(;(*IDB<0iQpF>Ab14W(P8^sXNslre!ro3@L>gD3*_E=Mga;cikP9k^R
zh`T+KqEI|0U;A2S0O2!S*(1DoXH(bcpsUJ1E=su+dj^U@VQ7a^X2g=zDEA|?L@*QS
z#FUa(%kAgSpEpEk%kEs*yo{pZhV-xs8+HK4K)wd;wTk}w8={bl5+33XMkicmZQ6)F
zN3&tryuU01_oxi;G?a5WMTZ-ai$bxxg|zyk)XI;?K+hjnhy-E|^N5ZEoTHCWiIJcn
zMNhf30pjnP^vBDeZ>m2$bpu@njL$G2DF#D>+VMASKO`=rvQaMS0<X|0oImw!;diGn
z1VG7cI9E^1MO$0SJl$-8QDncSQ2zcRq&xVzI{DZiGrdp;CAWcw0$Rb$CU0p9Z^6Qi
zc(&_A#YCZGmf#|8xzj)fzyqV_1K#_Dx*h*V*9fGFBRjr-NmoKzdg|m!%CXSnTH=%!
zcA${?@fJ8aw`|^g=8KI6(F4Ca^Q0HrAvXCHls`u>TIklxuS_U5#BZ!>VQOQbB;HTS
z3J}A_CDJe1EB0$BXN-Tky<{y8!>{t|C7;FW`-3qGWo4_9>~S3fH8rm*n;l-av_zU6
zUqQ*dg8M}Uz_BCys!YioF3K@A1;f?$WME15NebndEP^NXJUbSAzPu%$iXZu(UmEeZ
zX&e6YTNKK+T>zc`+t<Ehc)<VmWpD}1*nj)Vqx_9R|MA^3%l^Ld|M=dE<=j>){^L7X
zPP%_r{y)CG{?`BdfRz7>R&a*{JB6Zi2z9S3H0j3Iur)J_kr->WAwb{22p63bqSydY
zC_Ia(Ys4dIG+FP>6bi2{s*kDx2zQ(Lu&)b9ubJ^R3Z-QW_M3)0K;Z}+vcEsh?h<mb
zPN|jnDR;~P9gss!X{Hd9M`Q{eh%5LyMwekVO8&=xd?iY8{qNrTtBqG9WT=7Y=9ZG`
zXf^-$_vx^qA}idO)P>a4W<3cpctkGvU6E(f4U5-OZkJl~?_d3&|DAqcyg8prp6frp
q{tr0A5<!)61UUI`UoFkqORIHdGoRPenCr=(DaoCZ%{g)A^8Wx#+N2i%

literal 118726
zcmeFa2UL}3w>3)Q7fm!`ELf=a0t!JuM4CnvK><+^=@CSv2N3~jCPqQ3iik+F(VHl}
zCXtSU(mRn&QQDT?*>|pOzV_cS?!D*!?T&Fj&Kc)?dSvhSeV+BKHP@VTJ?GX*RmJ5?
zxtDTqa4c6krf`abW0?sD$6|@E7vpc%FA6$}|FziYxS|3Fi~j3TUUUEl$3_k%g+m&S
zp#yCW4>ZQ7<%dkoyf^CqO?P1ZbA>}|-g@4MY5Zkd{n;mvE`F(Qd~(gxca2}(->9_J
z>$_D4jpsd&KJnyU^Hsvm^<RB){N8r&%lG#d$d6o4UMBtWrSpYlIg_sXz1a_(8l;@h
z{+YI-w=42gtOg$9?>{Hk`q#w`@csRd_}`Xq)|f2%=byJe+R5|JKW+TtNYg+6#PQ9}
za(c19evTbf-}m?b%)xQ+ORpDS{PT~iPTpJo&p+M1_Qdj^f4cU!fB(vp(Jv%9PS)_b
zj=dW$+YzBLfscJuY+e(ka^k`Fk6&cqV|DXRt4eKacj^r*Z^t9t{#~p}X7yl8M)H6a
zhsJSyvtU!M(`aV_Yu?7h2CrkDQ0&X2HL(^Sq;vN9k3Z+)rQ$LrEb2AzcX5e2fqDP(
ztk)8Q|F0hJf5VFW=PUmwh1!2{VGnNDZI~~EBh>s<(Vsss?}(tM1v_wJ{LV*McIj|%
zTpOS>@u=AG&!3m<U0jl=n~^lI=Szz9gTZCLf5Gu+<bTwe>C>1%<nPqZxiZ+9J6h>&
z-IiS>oHMX$k4~!RFTecqK-$4N_tPKCL|vvnJrFZh%}Qv?vg_T?%pHAlJV-c&;U|3*
z2g-P1ZhH7*bVPBDnmnJs+|0$-5%L9k^-+#*LL~k7Sv9Ybx`IP?vFU}0xJ|oEUqiCM
z^z?KJzUplDWwYJ7>AQO>!+4klEVmP32PBI9_;@OUM3M*m)i2=1{pg8SxKF+M@b>Xx
z%Z#=w?{FkbtE&8uc&{z`<Hj;Z-ekM?Fr%a-)0ScM?#Yqv^q?z)%}1)jWf`Sg{C#96
zTgAA=FMhXd?QTzLX0F&!0c+us?mAk^xYZl>hs2jjWq!-c@uSfFFCU#svufGLwUX1r
zSV;Q5h++8f<Q0E@*}Z=9AD;#6^<y@@G!@)@3Xh~JJKYtq%}qo*@2Y*DWWlt`8Rx1i
zgG_hUY;1#}dtPB$kg)#4zM5#YxHE}5+mc*6TqjLJEehr*D)#4_)kJM5@n0|c<QU6v
zu5;bCp#pKNN5rVb<K2p<@Y-t2dG_tDi_?rxwcthO+R*!<CBvqEHc>0tz`X44HnU!C
z=_{6JMh9}3`LQdv9DTNQ*|L`l2|{NQjxFci|9x89m6q<YNkum|H}g+_{vJEt->|NQ
znLj%eKgK-G>9J+@sZ~q5w!0ysty+5Q?O~yL<|O{wLGkXu%cu)~uDtX~k59?U7faIm
z{wIxEQM2l22?bRgCk0tCTpS!4N_^5+d?q{G<yB?JYgf(o^!Aq6xX-q%mvQ__wO+30
zfw56vZS1?q<3X!t_nFu3T)cvxcZ;H*k51a$_$7<C&(FQeoa*#8o*DaSJ2zaiuEq3^
zU%wfC?;#Ks_7tzDo$P?WRSej!Qk}j+Ku(4+m7ANJ+4In%lXiHSNbdK2^@$<l&C9vP
zJuQ;5wnwQ&@O$uGIa|-zTZV8j4R_j(PY#%PSXEI`QHDdr!EsW*?6+NNx~DmX7cJq4
zk(=|z$ywTmILaNXKAti^(Gp~?7a`}`*%hFsnmX~=gP+^D#Lw3%qb*&7QHf}A{Pg>h
z)rn!sk*j>8FYBaUT&X5AtZ*5D*f_E2@Xcp@IUR1ZCgY2|Jv~u4#H?DL9P#dKGL2Aa
zF^E5txL)kSiEptdSdxIct3^$2vxCNXVcX&jruUrud>5ul<cm{E_(L4|=}}{1&CM6{
zZ(5RkQB!tF(9erFG(@QK#XV}zb5?wM)VExGA}MD`IrRzx+c?B3eOFDRw@Ci}_?(B&
zw)iAD(xMbBj3*bh*L7k`bl0p|^NcU#js6dG*rG!(DK}5Nd#b2-?%cWNiJuoODb_X(
z)9gIZ6oYl28&7ukUBtQn(QCWVpZ@s9wrpF_y~Zm2ALywz<~@A)vtsG*LWeage>eG)
zgQLzHHSL;t6vF6T?CEG_T~?L*f?aL0VZjOG8}A}tIu3VE&33T(gACoAcx?+;rc?;$
zC}2yUsD>SQ_vp~|Ml5dUud7u(A0PH8ekfs4UHKB>c@)Pi-O;_fJaG8as}<3F4n4u9
z7hHIa6%;0i+Qb%q+E%lLC&z)wuVv_JUp!snvthL&udRZ(S@qWRrVAnaKmE3d@$(Y?
zW6padF23E!?>vyQx5IYp#}#_?3fIo3_k1h!1E+yboz&-rm1i&PbCZhuM+)5Ljr?UN
zJZHxH6+`w}b*j0~oW8~Ilj?a;Xi*g=E$KG#GNMRTdT<Tps~T15qAL6i<-}s!M4dD*
zyGp6(6BgEu&mW&)m_Jks*nR{rVl+FL;Xhg>FVBzc{b0YXk;`;<P|D?x)n^ObToZKC
z1m$K2UOFyJ_da!Cw%d-qQ#8%pVxO%jv@q=Neqz?%?CqmNb7Rr+Z!#NBI7~S*^Cl_&
zh0^zkh={mN=5&~Z=3vuohq8K$3vaD9{#YHEa&^4Vku^Wb7^x8PuMjDa8O%h^y_d<f
z%$*ylpm;0vlO9@6uh3$hQzTl#yV{|{)nT_m?*8F9TB4I}b*Z#O;fa0W*DHQQ)a@XE
zrdfM;+wsoluQ<IaVnU_tjNUwy5Vh-hJvCkBZeX;tE#Gxk!NPs&y~jK3=6X>h1()$U
zE%mc!*BiQyYx=Ag-OMRB_2&ao<D<pHO){0q`}-S`MQSoH-~4vNDkmqW5I_9Zg@J-s
zoYHGP7EIGygbpPqa7qt6pDJL@7Mu!8#)=l<WCf(pd&tk7OJP<8i|q(^ooGsB>~&4#
zK~~=pcKKBE#0iHoN<MoHN+{cS@L%0&HH$-TJU2CL){>kz(VS}4l({g|&-n1v-)~2_
z<K}$)Vdw7MK2>tFySJZs@LhwRol2|A=$l(TODJOclD4yRp{3>|uij!#j+3&xP>o}r
zB;li`{=B}z%PNPpFe~QQ2H?0l(G>^gSzg8Ok{vFqgAA^k6%SLET)dq7-c@~vuXh=W
znO2@?Nk{M;8SQ!FZN-CR*=#F6_4|s>+lM)Kr4JUaR`a>JoM&xII_lPX*{heEoz(Wf
z|25TW<x1M7;t5Yr^`}Q8t!5V{3j}fo({*pzJqww{j_#~EOS|`{ySWi%4iOvO<DU~<
zI00deL*DB|DjnCo_m^MT&)72cP~7ZPPg(~a`N3Wj<zl(yf`zLmBIIPPoWl;-KSAD$
zG>{Hd4!n9o#o_ItB}Z5b^TxF?>Z102wNr?zr;i?OM5);lsLUlbGs%pwO}cz;?6^qY
zz2VC5Ke`>fGo*fd<w{DPsLGL6Rbx}{ZY*0@oN886oo+J*^nKn^q+*Ne=~#_#*MI-0
zmO;;bjTQa5<h+RSW7ptES5E=mpZv;KKwhpsNk1fAHc<J<p+kq7CoVgfloneg=iS#V
zKuDNaqM)`N&rARQfOMKggI~n_fT^hS#DHUOm_z48i;cfsm8>L%UZYub*b@|7^$hE_
zQ&lYq+AmYB4&VG%spZN<lPM)tuiOR{rTxQ&t84tq#uf#6dJ<f)(yh3^C*$j8Z)pSf
z`EehMRG<Q5XWMp1hA$oc*b+8YW2vdU`E5(rpG|Q?|Mv8*I1k$<wA0Kro6}Li_9oAx
zBAC=AKRYSh*0%jvlr5uJ)Vej(5&x`$cP2U@A|fK)s;l%)%K5UpMlIa?EZ6sjJ6qqG
zRF#|VYP=9)<uxd8)1JFt(ps<Bb?fwF<@9tM;#2;?2kgzI2L471X})&D{$P#2Le-TW
zs=uriG0-!LIN`8T$*9PSOM6^5>yk<F#2e=NJIk<SmD8cRAtn6IyR+^4&%EzXI-9bY
zp)b47{W1e}AuyfU!CII+n_;crJQ1Z5a^LN`$y-#mQ~uMP-Xi7acF=d8d|CQiO~7?6
z*5?EA*SFRuXrDHU0D`$!(cIJ?_W02C#T#0J_UNAJOY27!Ov%H`M~1&VAT|10jL~T9
zpO`xC@$H7Y6WJCW`L3v|n>H^@zVc-F%gYVQY&q&vnyH#CeWy!Do8L<Nem^&T@6Nc-
zF>|-}*MIruN1pl)Z4x`?Uu4yk^;FCD*gQp&h_1Cevtym!Plt*pwjIA;=quU2ehJ^@
zlj7E``_1d(xY!v8OS$LW%uCZMKlg>v2(uRF@jkAVE9H@jpSsP@6io!<sL&x{9I&WQ
zD4VgC;7ZJO`;NWp8b0N(Y5%vye)D<%XDE6PuJjOab38Mjefi_v6xYg7DcxMH>Nd59
zk~ZNf^AC*ub~zc>ooz_cuSrkTez~1fe(uv$8`Etavb+J|VDtoc<WwGtE9H9petk)_
zmGoq*-P3cgZrY+CMh%{>)_)->I5g9A5kTk24-+r=gNKaRU4e#LcQDoIfVf@%5{w5|
zqL=me<CDHHPy<lbXj3p>ot~_lAsnR|mR&Rp@M2Z$1USb^wNeS$=SvWaF?Bb=ut3h(
zIq&NDsyBw==>U|CS6ZxU($&y37fPQLxKz67*5oiEp;%GCRl`yj$LI+avR2y?u9kcx
zal$=k`h9SSt#n(`t&B_f9A2BVG0runU=_R0D5#WwMiT+lA*X+D+4%e-VkgB41;(4I
z0<S#s&e^O>-D{auA(A4DOoP0uN8H>N)-oGnPw${?xkoqs_5S>=K+IK*`3(ln+a4Ck
zL^^-lc@)1`*67xF&ZB#)kDurIz2}cL(}~$*6~h{<Q7b`7d!fUa?u}4uIRlXPp`*b4
z%&27&*7zMj%+wnjx6@I>6luTJC+F_{^tF`ZDv4d6af2G+hQ>b@W|;z5$t!w}?{B1_
zcqz1H+ZzLjdM*>r+}M7#KhekG>sy`#9lc#yVGiwO`-iSA;`DT%?>Bsh-YLegz)jqu
zez)p@kIx(j(;7NcYgC0sdz&r}cNT4EVP^Ly?eDsM_z8~At%mZnBJXu2IHc9-4nwU{
zSK90wj9Mmea1xj&!+h^-QH;uKIc_U|qVeL?r}x^)>|J5!T-IE#yuT1$pnn>vE+TU<
zy~PN@XxpIgP=pK_G0v_WkFM@Y{`dVj-NQQiHOE2UwArDY63>-elx3~a+L>LiS|LC8
z{n|Y`Yg()_^qYtEbFTOz5eHsvv%EKU_SN+ja{ITdh3i?iU|j{!26Y_I>p86^tJx>I
zlGW1haGiyeP<U_GY3dn^COQyLRGV!XzLIPV^Zt(;gcA$VLGt4X^_z3@?-h-7R~Q$1
zt{ko~T-e($YUnm?Qd}ZxEgYy^v|O_FJCvF|>9T-CLjXtn`x9Pv0=;_?URY=7vNB#P
z+3&)ehei0Jb$An-%YC)G{A5N-sfDb!VDnOShEC_H->HuI+yH(qJ_l9xZHu@FiWs%j
zC+e;vJQ1IqpZ$x&xV%mN^o0a8;)RO5XEf2ctMVLGoMWa0WaX|twjE{V7U0lP0|SFI
z^NN1F8+i6v?yTbmz_9$5UuOTz#9&CeTC{4|lhe_vkq_DJ=c9zZCiDc)%ujtT+ivEX
z)QiXHAinR5o~-ObYvi{h!ey^<1fqc|VQ>#?p6PBFWy^#1d?z6x1v`zXB?2zr%PmYX
z_WJJGxziJ;Qry%zSi<6r3~O#wds3u=&vf(Vr0l*pG#m@+Q7IJM93eu?zjnUjpSYgG
z%W?FLB;(#`)mz;aK|$uNG3rk{tB;H9N$Ege%ZHV<#R3TTXGvN%em6fe?kzo(`BcMM
zP!%v*dA=#m>yJPF*pJrO7#kYNt#$I`cO?SuF0UU-2u06s0PHz4Z{R$b_CU(+LU4WK
z^K*wo@dCDhjMSC{_pqO<F(y~+^P$`GFI`aMALw<~d+=U5b|QFhWMMw)v$x^GNRaLC
zU#%+yc7OShh1l^UGB1CoFMic*CBD;7cA|0p{-1t4V;6M3=XJ2{KuSq*Z-jezz_eiV
z>tL~`hdnq&Z95ASW(x2twm7b)<7vAS*BLl&ikH5BZ(O*E2P|jiWV<tijxY5q<$*#K
z;#CBS1z1yxp+Nud3fx%=&g>MNZqOaZPQfu|f7OGS(@o$9ADPbpcGAAvqU4uI^@gt2
z&%S(%&!JTmkZjdxZ<W8A{JeO$+ssGDt~-iGIB${REF2_%M7Fdr$QQyV>}S9!7N)4A
z!;p&y6TR^I0Re*q{hWQ#5re33p7c9zBLNYG-|Ptej<P=(h$EDuRm0*~T3sUfQ=9$v
zcY5vzT<*jR+(Q1^eFdqTveIt7%<l;qJ7P;1=uExK0(Jzo$e|{EyJ7EEB2H3W&m`#u
zzf=@)L#O@@9T}xhUc1+#)bq2`mVJNyJK(H~t(ms;RA&;-uKen&uS88NL#L2^8ORi#
z2qjs$xlgYs0#qm1YiHS3F1^VA_Vpd=tL=W;uphZsL?lo-QqXxdhXk9Mdg?D12ToR^
zRB1=p8k~KVmUf(R#}AJ^1b<Sre8VkazJ?YBwLP<O0(r6c#yW$uzvq0pGc4VzrPybK
zn4mMZ;&~lV$my)L#b2C1yjV4<?DwDhE>u5LqO1)5_n`gQ*~Id8fV=18I_Z|$yK%&<
zobM{`V?W|SlYr}sIBxHu#^_+MJ+KPpVWSpA*6QTk5#=K5uh{{7=@=&`$H8<q6>YNb
z$KS4{TCI@zbnw}!*rlzXsq3!$yu2LOED5J^D4U>_zcn$~Qc`+n3z30E6WS?8M`S?F
zaOH}zKk%3;eV}%2>;V0`X2+rxGT#O1U(v3gU$UHg4KTiN2_GAf9vmtKGkB~#3{~n{
z=*$$upD|g$a%B_O_|5Vg^tGLA-tbZs;C!Xb=xdZ%6BC=SK0i78iw;ncyB11a8?RH>
z9Z{?1)be&xA~-m@Lg^Dud}a(|&Qfmc>h9K@*cY=HQN~3KTSkYQG{5`Yup{Ebc(6Gw
zFmDh2=A*>{KL;GY^b=nAFtPh5pAf6TY|j&-ZNbk;;jTY9IJRoWgR=kU&vS{>|Ig|E
z{~`zbk8k<!P#FI6mH*%1)BoS?B|DBw6D%5%Y>{`*u;rn0-Io_sTG1m1=Haw$ybd5h
zoiIC0Pwt^gcGQ<!v#EvSPge9!o=0BNBc)pyJs~zpNKPCF11ASFIs|d9uL%mv%NzFK
zBc%4C-9XOJJciWrkLM(GdE$SN%=Moj{J#ytYl+RE9yg1K*wj#$UEaQm>qDKE;hvH4
z4#!rI1%Y`I*Am;0tkX{|nrs00Dh7M0-J}bgc+Fg8NdIM7-WP)5GQ8LHVyUA4*uT$p
zX3P-;TxVpEfftq3wY$%LN0<Z%xe)ML00f`fWoEl<HaJW0gmD9dU)GfZr!ayF=FrlB
zjyND(FgK#yxKg5wa4x{`aJf)&A+Wm<`XF^6C-f#>;E)Q@yYPUCwi|d^CHF4HxP*<b
zCfgiDYIF!XXc;(d1N7z<a2->B9V#WZbm>wf$Smr~26-1z+6FQienifFXil>Tz1kP6
z4y4BG-q~u$86d>81EC>V42)-l4ko0XD1x#61`&|4`uqalAZu5b<^Ej0b#ig3<@885
zkrwNrJ)8*Hx2LBnLZHKCRD~E=oiuYU;tLoc(fwAd$%uj{zcZdY@)O6UZU3htuH52Q
zqas=KMzTZCmZ!Wr#3p|OvDS>V27`(?cNC1Lrd8p#0HKQg^@$Zu4z*jaFJA6t$eL*g
zF<7@|jRLeLAzMX_d&KeKY^D{>E?UCrLuvq{5sbV7$o%SbNk|WJ^J6unH{F_T#A#-b
zf}3KMRMNK%{lnBSgXm&U(+Jmn;IxiQEGWv)sJDWjClnaeIEkh97Bwy|+~O~@id3oC
zdb8$KGd=q{O<QzeN6;E8OQh#Yd6a2(N%?qYLLM^)fuoztLMA9iAEFbNCiDkNp>7m2
z2SmB=2C)nGUio1`FMI_4I1w%*WgLM#QdAvzT<D@=Q-IUyz|uSm;Ven6GKs_5Tx-`B
z144BMsL9;|cb8c@jOAJ*9Yw0v((vOJuWl|M1_@b=Ht6;hU#2<uVm<Iwwr?aF6x=-9
z9UK1c1#ojaV_|9KG<B3w?CU+$mJ?+D`k7L|6T9BZ?$j;UxCc|9S1;m77X~fsAAydM
za8SalkG_nXoP68>>RK{rD!MaYRr7$_y;JfXn)CP<85-|H7p3pJapQ(<e`1F9@B#u<
z*YZbw9mK!T6*u$rk2TRH0FOLsa?|Hymz~?P*}!GE=tFCkoeVJFwh7sxOvV#;)<UH4
zc0bRvFU~8qUanE0z<wa_?iO1Tw4?X?EH4?4l`FOM?KM-F9f;9L1A@_lcb&A<D!<U(
zk|n9>YjN{3S}Gpu?I^5iTZ%rte`So2*XBAN1j<XOlQwm1s7<<G0-$xnqQ&2O9_EzV
zilchFZoSifjw~e3YYB;ZS$j#$nyQId(CC^6Z(f*~(avp?H+}{hraTVgYD1UdBXA6C
z0_U<CdS*wh13uoiUfZ-tVBwR8SG(Lv-*2JW$)VfY(_Zl5iT4w<p;Fq$Xf}zJ41N6-
z4>Utzge%wT9jd!0)4-{%MVEFNB5-bNL)@{-bC)8eJeKe~{&;z?+1`5+Us!aCO9<6M
z<H%l{M}l{pPjh8mh_MvXO7g)rmi+q7>T+>6M5Jx4x*|!S#|MN~EZMjYnDcfTh?1hl
z6l2M}By=1A_IxC;6<_;e(G&DLBzY+!e2u`A`aShKAV>-w3v7NT(o4IzHA2g_J5cLP
zR*naUt_L;k;6qPawrmU1yZh1}XT|upYhOr(tk~BK?0(Ci`%NH1Tcbicc)v$kU=MH9
zZ2N(WZy)JdQ}Xx&$s_KEAr;N;hmT(@U;q8aVHF+pD=YJyCoAKUkxie`^9H_u<x%qC
z-P6y{ixzzg`-mYSiTPRK`~tdfPs;4{+RU8ns}myS<WiH!NmpM!>GZ}DUs70XA(8i7
zPE_P~+S(3UIJ8n}{B4qc&ie0vJUBNR!Sbn+XUR|i5GQ~DFUFTKK)jdjcPqcQD{@Bc
zIob(Z&>BVHCyajo<*O7B2M=VOuTa~rlVKfZJ1{@lA%L}ieKpamlXtt8dhaK+GX6Ng
zR>KQe>b7kV4I0hqbrtZsJ$T*R@p@fR$4`Hz4zAO`tWE?X0c4^X7v^Ub!4agmM!1ZW
zInI2HKCX9~pYz};pzfXTUYJyL#>FQIj<f?u@JC0uU5)M800N7I;1B~5?H60!#^VjU
z&-)*IGyL}O3a;6O4W^-YM$^!t>EXOeMFb_DM!6wI$s6Ka@z}>2M!l|$F9i*0`Yo0(
zRl1<rLP+H94YgA}@lbphHH+s%nMgLfY~Mh>mu-g*LJ9%p6grobMQh`=M6Q5z{{T%p
z7l$qd6m?XXF=Xmt&;x-pqxYBK*p~12AQ^JC>i%=ylSC1`@>n4<YKrD0W;fzE4sjfq
zb)iuY;xtj}ZgUI)MH1l92vQ>sL(L&^%f|0b`w4)hfPXb=xjHeh0iWVexetVcF$$l5
zi%kI~5x>+?aFbTsl6E2XcL8GE`|$e`0Uu)KYWlBp9@v=#D&SOAKv2*+M85Cb+?)y%
z2mLhp6nf4_XetQ?dQr1qd7wYxB6vzmNGsI3?*H@JqRxhb1$%5;-iwU0A5l`-y5QJ{
zcUvX^7pSh*ChAI(9t^3^Q19})Bg3#76!z7{SL7C>Sg)r3Yykc8JLKDoP*DcZmXoep
z2o2**lvDv!a5k|(8^ty#pnru_?;m!hdDj7lAzmWEV8)O(X&*h>?X(;`1hx6t&K;Zw
zK0E@q$+uZvUY=;pMt9ac!%Ha16Ap+S3*UC11NB0c8cR3!L0#A)A790F#zzY4oB1&+
z<cV&EHErI}mlxhVfw3Y5OL&BYPb%IyL)k*eO=pDrMb19oID`lIfO3^;UMEN*GJ%j#
z1T8s1NE-<54VQ7cG&kf>;0+gu@I<pk^2PRCCkhpG&l^z7z3~PjM}5|-cU6VNMS(K}
zkD*qGP0};!03Im;XCIiNMgmVdIO^j5<N|)wJ4fEDA0BV$HA4!RLuRUM?1dgs*=V31
z`l?HCgx9ShHac8qV3yPy4Z*-VUZhFQ8?ST5&bM!i_k1n=!%&!+b(X_KS9_Tqv+uT3
z(9t6XVk?tE;@sNi-mh@?M8Ya0mx@E1y_OylEs-*cs9kpH#n3p61{za}((97+rLBG*
zPsV9WopnT+>jZDH)7A=A<JL2d7!A~Pl_LtxjHnRnxPd)Uh{BA`&fp+4o=jJQ-`PB>
z@6;V=HcJYBdo6Q`{M4a(3mMP_N9svEO#e^=(aEPbKa)X9{T->roV;hB<)I^&fY4Km
zbRBLRe!#x(8Oe>cAb*wTc5BY_-YX);O9NW^3!P-A(VkdmW}c+x*|TQ{^om@UgJ`*P
z0Aj)M(Am4kh0FRZZYxKg0&FTcg>^jM-Q8`H#)Jg9XB=4K-hP`SdWSp4A?rm`JSE5-
zh9YqqI_2)Vx$Z##7{M#a(u1j|V|%ioa~`*Ea{%3u28!IMw;7UJwV%+=ou%gTWQOSb
zZPAL|b>Vo*`&Z}38$iV2WFD8UA|K4hP;todjE7XE&A~ovCuY3^+73DZcH`xO9wY~q
zwo4UzV&R{=XC4-J5!ELaXfkp>0hroIwZR<Kw>b^4Ip}4Y;9a{1dP$Y<u}9Sl8>W!e
z0!O1u+rh;tkLO)xyaUZs-hKcTb6L}Q4qt8wutnv$1$L#m@diC%%4K=o?JVi8_F&C3
zL;7KRnQ>R%JUf*XlK4J*mr}O8H(EgbI47W<hr`ZJ+ll6EPQeSf-m<XS1Nejr;u7mt
zrJ!JrlOLF8b*-;YVCdfRgGaGqJ}4>HSK9#G_YbDlta<~FP}SySs^AKvn%h8k`=ZRo
z73QcDaXLigK43zP+@;uTn37KFUHurT&04`EmS0m#pDQ_k#{A5D13(86r#}3$*@gL7
z6>o_8xrJ^dAJb7Q_7{j4XSunJl<i=UIzK(yYg*_2bR_rY0%(6lbQhg%4jpgeSYWQX
ziEGSgzba&0K=|0epgoFg2ik88*V1dHFdYV)B}^@9<I=hf??4b5bc7)RXW=OIAK|jj
z#l*>zIu%>&2Stat!%TpP)vV+>IoBD&+eUD0Y?xI+jidh~A-Y?3YuBzl3Pq1W(l$s$
z)4U*3;B4GRL-wP>8zW6<`^S3`bw)IKdLq*MJFw@WiGj%0WUy+<f3vTp%QDfci+6ia
zyj>%<y%A!#x7G2<ybYhhV36PIob%mbPKu>@Vt1TP6^FtM(z6^?DsqKJ9^U$?&C=Q;
z&-#)XYP?yC^3}WNqb5)h?`Nh1cx)x$@ebw*@7y6kaF|!*--VKs2wB~srk^rIXOZ{n
zH`<zGgDsMBpl@R|<Im7u4Hxs+6av`uP_u{6JBs&zf`<>8@dr+^hQmv_)wXDcn9iF#
z){SEakfoDyN6M!_#PKa#C+q|5wHQ^Nnmi(JGynW$nX>-h9Dun<2zd(d_AbMQFt6kj
z<O2c?0&Y|1^jyd5j^r<1Avg8o9LzP9R~IV7q)X7!cA{_cMlL9Vlg0>65Mt{xaoiZr
z9kcCG`GoAAz$_Bk?Czw0n6wO`q)emJxAhSs?5+hYt*HCL99bdAO$P^^mkg&ZN(j6{
z{%2D!s>jMHIw6;7dAYIxINmgIn}*q+xuaNVEe>Y$Y|a&{%K9cE@|}8uMIqwtt)jA>
z`TizPCsu0rC~rpVrQ<{%_b(8NCXljW7E~x50tR$6k<olo)u}rTbXPDuK%EE>QnZyk
z3iVUGn3Vn~%7SW3I>^SZ%ZXlpA{_ed9t=4T0t74Pfh@dMzIjxbFWBJ%*BuIl8iaz{
zP?)w7Lg96;P#=ZD`4?&Wl38PD%%iPZP=WXQwIO#_h%+6^UVX9iWqHnVNsd?z!8+s+
zAAjF0lu1E*cQP<#4Me-nP8{iReSc$}*PvFozi2B7ZAWJA_oUc|$p0d;`P6t{-EmZK
zJF^mM5z-K~Yd*5jOzp5@#kp}R6Z~pRJql2EN~B=!(k)}?@}b!y<wB?FsHY4fbE8-W
z55H3nac>u^+`omv-}5FUs1NIJj(%g?kOOu#7Wi#luzXE#f~Um?6hq;W2Wfm)8aGKm
z>hy_VP`(~Ho8&E=z0<1Wjew0qlSvRvLJ8CYr*;VKIPu_QT-ILMyG&GQ6@4_ZMd{0d
zD!sGHXO87#wS#8Vje`A<aAXE`g856d?vwN#i<A_t&%Cc|UDl*S{;}Bn^guH-KCxsc
zkuWOs)F|=et8t5h9uZBgPiWG7Uwm?uelm<+v0p7)+jTi_gwS$$?<d#~g{~yQIUVh4
zX?8>x86r5p0SGYtMFuAvWq(k|+}TmUg8~}se9y^r`>1f(NNMirU0DwM8t$~_bji9;
z-gXXhIbWU^Z_%UAl@zQ|ZvPT{y3GvGF(u+LPGIE2Xgsx9RS9IrK-&T2<GtZkmLm&^
zn^tsufS4jOE+GLkV|n|&gZASq9Lg7Oe2Rx$<CF&%MA!Lt<jcI$epps|&bP<?3G1H^
zQ0ykFS#?D|E?~}%s0K8=M+~$~K|r*O^$Y7~7v53i*Y12kDkH!h5Bbr~Sbz>MLVrQh
zlG!ZbLP8fdS7omZH2SdO3y?*IYfq~kf!9`9uK=c=VMqvMFH8NH2VzJ>d(ic%p5bAn
z>;9;cggRi%1)SjfQ)iQp3ukR5Jgbtd=*!JTQb)SLW4&BdXS0?|u1i{YtTQ)3O?pu@
zODB%p#lv%S;)vZcYXRuvqvS#MtiQK6tw1eg$V)Xwfn5i-MQAi!IB%?)5Q-6^np)S)
zffp6k6Z&wuHkW*ILwQF#>F|uN1??4EszM$ow-*QnUjpTWe6it31dbmggOn%F36`Si
z^pg1_6}TY<$F~st+P#Tk2l8O#LUAL-x64uh2hTcxx+ir|e!5KU2|RgHx*|cJp&rg-
zdgCO#wA?7F!sDdO2wrC+y^op-J$#!ch5cG{#=6=~D)KB&_2gxZ<X_mVCs{X~U3LRP
z<XL)b@~>RrH9R#-K%w<IlK%_1G!7mOkvB@obaxCY>S2911J>r!UK^9y_R0KpVD@&U
z7{X)e{i_*4Ou{C;^>M#47mEXJTV*AXjCd6JQ+dDvSkH219I3Yx@>F>fKY4ew^lXcw
z8a{eyc?a}cd_eB<Lt}crrXN-*aZZ^}za>mu!aCYq6>me%W1zTJn(H|Q!&sfdb~aGM
zS(8Sc1-hM<&UwUWA;K@rae>g@Aaj}+Mj~-Qld-)A)XY;;gm03g#Yd%@D)<A>FU^mp
zaqnO+^?#&YqO`9g2Oz3L2}ww7=bY$Ae&&@b*~voqjlHZ|P+^Tny2^;#ID-1mZV>vD
za*l!lX9Syu--4aVsDzrXNY+Ws*EqJ^1OchpeTQ`pt#n0r6auU0S?Ypv5?frLyBclb
zv))fR@{VnK+01u@)qP**Ts{vZ#@@m9(AFHw&pUXjN58${#IOBP2Mi>$KA{1=VA(O{
znwv1V?Tx^u6-yosu&XwTr&n&JW=!U^48YZ!vfRKy#JDfu^1a4n)XQP;IG1bs5l2OE
zNb82{#;+oIQ>DNSHI+TcQQ5E`EW*pa1^|xD$#(Ppz0CGI=V2TlMw&2{5N>FA3xd$Q
zb@`ujqg7Sr0xMUNi5*$Z>Z&ZdD5Kf=&UFQz8iHvb$(;Yf;Hh}Ez6jhXCwOwV5qYvU
z>w;u9RyiZ{=_xaIX=t&|9WLH8l#u$4te52OYEEHe%E@YkI7z4}R>5xaDDqlG`7AIo
zQG8;Tx=K^s{pLO>%6n1vZ<&C*Tsr<*f?x=mE^-z3%6AbsC!Dd9?G3%TLVzC~oP}yb
z!fr51N>%5g1panrLO&9V{$ks|)=0@6lfu$j4j=8955t{;dL>rB&u-GQ&$7O&1L?7N
z;_kwUSY(Xi3Cp1|VJv(Cd*ORn8+U&cIM~!YCvd=*N3xPu9^kVg+Xe^JD;Eumc~Zvd
zZeECgA#U^7zT$U^0(@$+;~GyCJ*|?Sm8`Z-1#n#F6xRQp@#u7KWmrgECX!vp@_P_v
zS5sD~A5~oAI?#t6Tv=jY!#>u}O-F<I)q@m30e7O+61*aS%xDqZONEBmQKgpfT?)Hn
z=L5A2c8$A{QXYdQGsWuTAC;b+?p3d09D=EfwWh@#A>P{Q@k2<<+{5;ZI9X=y`(H>n
zk=KEk9CqC`o_sOh(y_3Q?bL7ur)rB<XPZH^pe-kqOHMJ@f;V*xWu&0;kpDJ-2fErL
zEAmX*yB*#rWLj6payF=(+UB}K5AdjX0-4BXuxTz@2F;_1=7_uESC5WbAlmLU9KXN2
ztSmRAss{56p|5qDK<EY7cQECC5$SaIZ;2lwT9Pz%BVgE-D`|wO5Y5+GvPjWL#7afp
zm2iIiM0WH-*6*RJz@#txGN$F9)yV!}X(x8!wI_K;qD};_VhK1;@?S8~6ycnce*Kff
z;;$8socekVSB-h+#;pb?zx(c6@xK-AsFt~~BKXI3vL)rmZ{9DBH8!~rW5Tbt+06XO
z*E>=&?N_O*ixwH*+Pu?2L3k09>F>5`(M!{baEGda;P<0W;o~v`lP~HWlC^((?U3)B
zNB9`2vk+;)eD1w`OP51_!`qI9iHtV)q?nk-Ue_H;rx+3P8XpvsE?jldhj7wo`C7zn
z*qi*OjyOevd;rre#ODsD_`B}zm5ZMpb3vPCq4%n8TR-u6(hftfABZO?gF)IGF0^OQ
z8^@%$U+bYL1P{-^y<+OpWw4L@n>rmWZ(lzU-2#xnRWLtjV`0w%dS*NE!QPnWUZ9Lu
z#HCfcuW)e{Z<JPuRKn05|E2|bh~6hJUTV*}Gn!=J+LgPdD{pc;Gt4j7VZzBqis>HX
zEwTDsTkTO}9Z9oPbdb%X!3%@0CtOtK6-zx9opioZzW9-jGwXRsvSQm{p7Rgs-xS>x
z4iWfMt`><$Ik_QGH;DNYcW7s4C$(J0fByD!pdZ_thp`6Yk4TDX&ae@n{?Bl}PfNn>
zu9HpO{58F>xVtINI{7Y6gu2OuF*T7<z~?VBa*SP8@t)kPYgPt3<TUDH(c8c24wH5`
zV-C#G+&1B|##nPC&Elk$$xL%k_R&k~dV0J8JJi~Hr@+8;;#?(ZMo_m`Y}YiW+lUUw
zr<c|ZulHtGILiJ7nIMZ!7Il{&XY@2AZEjq8wOn#wLHZ^BMPsg$9X-J?9V0pP9$K^B
z>Eun~2cZ*M(wc2=fsXcov8b)Ox;hQ|Ft)7a;kJ8V;97akUfP00vFJ=3u%$;Tr6n)5
z*TwsXI{XToOBX=>WThN4ub?|uEcRbl-Lh#f_=3P;7nKg<+Of2%9@e($pE%5qLs_j8
z6cI@tJ0RfC)0*$<OcP8whU>Vv2E|m311olfSSa7zr%8-Ej6gg1;d3DOdf+!D$m50h
zD-z_W1)3ta)wN@+k6FFp-h>Egwm@ex1#g>>R58h2zTGywYPj!Il;W*A`|3HQZa-o_
zBNzqKZ)}WU=(9g?#WiMV*dl9Sqk6-O`LEO-ui@f40=AKKE>ZK^n5k|N_YDy8k|)%Q
z3Z{xgE^!+pKM{9K2Euq5!&9eLIkY>O5>*@R#<LX5rC;^_{vO@E;T1i${>bgie-3r9
z?#i{OUvJmGa-==WM=Pi+um7lVsk7u*=w$oDnTNNYTXR3miwJ!ljZcbhcGkk6;;^0v
zCm%^c5`H2AoCmGoMtP(|m>{x&o@UrqULsi1#voE{N&xnrS(&|i6A?x|X$6+uBQX_u
z0{smD6}iQyVx&szA-piiss;v=;5V0KyQA^90&*@9`rniP2S#3E-PUZa^z!l|9y3xc
z9Ah2QR26BYEEm*a3P56Gx|MDupX35at9tWXm*mH?ENiqm=K_shr*X~HOhWU!7+BH)
zAe0cmN*l-4xJMW(-7N%h5@=^wH`sdn&CoY>u6+2Gr85^>vs=H2dt1O8>x-F=8am+T
zWeR?J|3|_s9GV$M&@qN#hf66RMaz7vV}8hC_=bqfriSl%#IeM3z0b219HQKxA{@E7
zrI@XyPNvBz>>SXY$2!!WbKz2ZV7J{Hg|m`r@76b{SZPjvk@0Xr{wgw;c^xyn>(v)A
z<P_5Arx7%GtSl${;fykA)!(a&4kCW+DhkYo<hD`ulUV~UxBD3}Bin+0?P{0*mfz*m
z@0tb;jXfK&OKD!Ha#Cd7Be=~TgOalADFhVDntixc$q3U#b#mL|l4RhyBk$*s+kBCT
zn`pT6P*!GNT)Yu-3PtfwA)%Ju6q7Xg0+A3ldGNlNt-`=q%*5%o<@(r(co_qnNXCxN
z8dy>Th>oA-bqBd*3}#1aT(gbA#5SrMKS{~v+rcT-b^Dj0`jc(W7{53fXXrQg$wNM&
z#Nn<y={k<qs7~7RMkde<$QnoEeSy8w_I+zuqw=iN-P~47U!S`={os|S`y5Z85{~?(
z<;y_lZ0yRl+tM4Bb2EI@@Ajniv93JU<HtRD-MK;)y%fF^LH4J0(Tcxu9Ky@TPPF%h
z|5_cb7LiaK1b{+jtCLUBP9#+K?>Ze-dsWek=*Q}z6;vy~S-t&yZhk_7<zxhiwSe*c
z1icbFk6RE6Wtj8h@4qM~i_LR1V#yPcjL9)51ImX6Y=-3W;S>+;ku(g4$k^s?jk?_m
z#!`Emt+_O7+Tbum#m4zwfQ}+;ly-b3_b<fHF|q(Z>!zn9+b*9Lw@7DZRAmjF0nro$
zer(I$@B$1x#pSUI7WxbJhD&$fJAKJ(TL-PHkAFZ2javdwOO0pXY|@+!Kb<DIKZHC%
zNJufVs;jF78*0PJa1b&5+O!j}&^fS|W~0Ck#_yAAY7BlgR$cktWAD7K^KV})QNTcw
zIT!TPgPRb;YL(J?H}C&9Ydjg4OqJ|>{H}zM=Jja8F{7DVZih5bHFfA3daxpzZ1kD;
z1H~0T-2GQ**M0!c9D8r8E1c<UsWovX8$)$MgLB^lyM6!<THHVBGFFAog4#(5KS)V_
z6vi+7%$Q;P)iPfWgBPgiTiB{ZYRmXgLu6#+aIBhZT)72L3vX8q+C`0z<2*LG>P*pr
ztkp%-Za`G1=*ypF78IOebUQB(!)%kfhDPe_xWD@xL#L?jyRm^LF~$}(u|c;GHFz{j
z7{4qNE<JAOJmmE<gIXkpP6RSeno{x2=M_xkt#}y};|WSimd&ZL#c~)Lxt@uZr*n10
zl!D90<Az@7*-diOpFG)&v=V0tw!kv6qojQZJwGvM3=#L~<Mop==qL!e`1Haj{mGo^
zj*3puvMO^~?GCx&TcmB1n@+n#gUjwB=gaXwmWgOM&7F>z(Wt-C_E!?0=*k_9-^5Nb
zJgc0THzOvk<D{6)<?Is>S97mk#fld#$A(^8;Gkh9Yk{d5o&&OKuUymP+*}zy`Gu*!
z&cmF@7YBXI%Y*K&4*$Cspqx--pdZmcBpr6wR&@s%8klALg7RLpC!_V$pe#g!Km#?1
zUd6W7dE@n=a_97PbaX!23rpJ#43bhY*R?}Fb~qbC(97y*ixOX+btsn!BhAJ8mQ1Ib
zo|4%71eE?b|6z~`4lE~Zy0)z92)cm8@k(ACmbM%Re{ddyZnB^jrs|U2G~*k&e>ow|
z1EXb(K`{$TQphmVT$toupLuiv+c9V)fOP4Nvb6&g%1WEu;iIkK3#mahX|?J!7MML7
z8Fpo<3Gwm^?E_{*ty#9OMN0}Ha_|CP-@@}00qijVSbTB*ilI1hF-rXXiNB%v5-GS`
zS`Ucw9SvHMye2o1;$H+NmVwyx#xU0|G5~OKait)8v%S$|i*=uGaQ6k<R0QTkY(fg*
z?$b-uKAoLn#bi!&XP9RW85aF*`5mbf$@#rM*?{{^7@2}{^>EY~HD%{kDds?#Q?J(H
zQjnI;lHw!n%cYrZ8g}hwT^3YTW^1xzedCj|1&zNL!^88@L(}7+UtG(YbZ4!D@5e+B
z8rFu{3XM{IE`mLKzIgP=4{Y_;&3n_~h>3c;X=oofX5|#fN;1>GuW-8!aqDUGEUDz@
z!TiqVX3$A07L~MaeE>S{%A_oHVI!bJF66p_eN+N0pV-h3D{ZM`6%5#CKQ;pq)~|2y
zAjZbE*c@4Y1U6IC++MtyGWwU7vojzlwi4A@+|M(^+eO8s3}>X`Jc2^PfZQ-$I2Xxl
z8qmm3WLYzle}{DR!pj7uCU6J}Y)f43&<^=&rO5vL@kATV7AE50?u!>MTzgn3Yq29U
zGiMKzI#i9kqTIk5FqMy=aCNze6H$f%#k<tXK|FSd%k2U2k+_Bxls^`kFq7m<{ua#z
z-6jz5p)*a=J+dw68np-Wz}Fl}V_BQ#@F%0pCd`&ak&?kUYZ_{Mw0M@>&cpebI+b9!
ze*PQM<Z|;fXv&gtR{F|7siPrmtYYm!@$kZLNoK|-B<!9BM;~sR3q;*xpNqoK#QX47
zXzx#Ia3H8s78x*w&A`do-z?FvApeXVjexE@aG8VS;wm(sx{S#VWWpPqc9K#|SHwt^
z@TGR8&CIYj2kuU`2jx!6>*kNPtBCf7ZC!5sX6r5cn#OvSyBY;o%xh>@e2iUT5qMLv
z9&G>XnR$u(E>j*CDno-uU1%i#?YTwbnv*TNV0^*R*aO8Ydv=_Ve@MsN1zuoVQKylv
zv$>A8Y>^+Cku82=bG>e@-f<M3vd(sAW}B-a)Z<Q?$RvfTSoHYu<BVm7Q-2IKkZ43O
z8ExNM65<hXTksosj-9u5z;ssSOOtu&HXZ!%;k(r{u^<LyF=$KP0Sg;zzJtYwzp9+R
zbCYBS%a-&#H0eXKr!gvT3MToe$AE2F%oZcHC1B4C%o-|fT>Ev&2_q3K6;p{XSVxD2
z1d^}T>3v7odAIh9XVpXD#JH2QK;6KAAB1AND--l5$s<nQ+jQXtyAs)^Yz^&V{ejCD
zWgu0jU~&%j4n<lEHc^W|7?<AOl!75P9rrH_gzMiDpF@oB{MFfsL0j;iM=@1K%qorh
zBJ>YyXlN87z&jCaw{Y0xx2uM(hal@sQfxR@<=EY4XPCPp<Yt?yjw&gY$(3jUFw^nL
zEBpy`km>yXsy-K{b4SHgW15~m7^2~<Dpz3<*X~yi`J-VpY4!Nk#pejBPE+|=ZI`j9
z0(&m+^t~^2LZm?|6egBH)(ZMdU-#)C_c8}ZRGRT0Dha*TS@=jXzjm^5LX0`XSLJLt
zts}Owet8}7OD;WP5lVH6wBPix^$issi*75Q^Zw(DizlQ_c9hA>nK;AEP&>0a1_%0$
zdB$y7gtD9z3|=q!8R#NaF`IAj@vOb{g{?|=^&lW=kLOJ$t#D>ee1a~tRguqLqdp}Z
zd!uyZ$dRYv*Y!yAPt%b^iKxm9Fx6+Szq?gwM1VCcAXOxdUZ@Mjs?xl@&tmyDrMs=j
zUx8OR%hF7mH$9Sb_*UueR+=Y__pw1IuRVF^!lVbjm1IW=Q;9@s>9)e>aW1i7dr?tQ
z8M)>Ln$zN&ZTp;~2ewX^RLVnAFV74(xGxccQ<+x&NZ@(Yt_>w-meOm2?%&@|%(*Y4
zsk@1H->*`wHQ@K-2l`>tDQ}srSR#!{+k@vM(!*W^C{*fJ=DhFD4Nz#;ZJ&)$X_tOE
zQ|T<37V2<UJX9iJYrP7n0;N+0SV5(`fE&A{fc?ztpvLBj$#HE8D89RbbsOe=?lt((
z+aY)F-X*1##8QIN2}~hjVL#FVXoe=?%Ce@@Xz7ip;(?Jq<4hb1)p%Pp<XyGy3WX94
zdaGfMd_A3?homPAtD1c6lvx`-cB)6RU;{KOwKeP3nbz4D2VZyL!PxQcv(Fzr%6Y6B
zx?hw?Be0&y!_9QZ0KePRdqIb^(xdu`k3Gp~QUcGJF|L=;E~!C!-p}w6BSDirLgQ+v
zpAE}NyvzcO0kyi{!tv&VYWr*`8Wq$QnYozAXHvISC#g`<*h?lzgQqbuJMDK**F)%i
z04nh{2y9iXPw3DfIVREk8X$@?fs7gW+yu>Un-<z)R+yNW&<zzdMP~#9<_*pr7CqnR
zMSbr<rcJnE?IIY|0T8b={q0Kh8SQ1{A(Uou|4=FY19r*TaHPF1V6<H;zy8Fp_NN1*
zvaQY8lQw{_C4?!bN)_eq*2&T}BtV72$V3H&hnDc&8$Ubpp5vKjPQ*iwM&>xPLmsN#
zI+5H@Kg?lB+n<^=vbK`t49lnprR*u-Jln2!d(&Ymic<5>UuL{oAun@(xArR93!<nX
z4)PNFaxXBj2n{tqQ9rcf*;fX2;ulE8AyXv{hYVvC3TY4tW<@W$kfT>|1GSFS(~3fq
z;v(3DBr8V^__l4^hTr+SiIdrn{Agr~1eZ|5dwv8txd?98oSZz&{g5<7nl6OgP2}q3
zJ8;<(%EkeLVaCNix9C5M7)Ot%CA;o{`;7}ixHIu5xGII8R-C5nIAzD)k`uxR42sjO
zjHLRL3WdWZ0s<ryAP;XPuEnu0Hu{P?U%GW0H+GOcIs~IBLo8zYgb%OOi{H0K%3mf4
zg6>j)@i@>=1UD6gyDv=dhwZCR<|ZV^5(wWsL3)>z3}&R^WMhvq!7q2<LRB~)+Wm-5
znuQ{<9Z6S!zJxr`b@5u5KHrF%az=zHCIH*bdU6lL?;%PvD+JQn_ksWt(^U<0RSk57
zHRKE+GL8l|*Q4o8z4&f({DL8-KjGmS2~6(lw<-iiT91nbjCR9r;h+l|oz`%lb?bWg
z8$1y`M*)B6>J_>(gDep2n+U*_NYIpbg4^$!UsK@9dC;hkB~E+<1J{{cpl(LM22%u&
z{_M&X3mC0T5bY%%lG=mffJtjXDW*&5B)dWOK!c3*CVK9VU#DOK5g=;{St4oK&SDmw
zhOwYHts{iTYhSBQE-vD{E)O({_S%B4I2ONC>F%?)hd2hR==uvs6j-tF0=j(#vT<U;
zwbC;9BRhFIrizCe@G{<*Sz7n~W^Nvfm@95r@M7rk6H*I6^%8C<I9wYyFYN}9^dry5
zEU!F8#Kb_;SDrD$q8=Y5#-)g|Ao&Z-Bdo!UX922$2Ni;?aFfOvmOn{XWGt*Zq<~>8
zcd7TkuBMxP>I_cn=y3B$S|>z%mIk_PoBoln&M#MTrHXvRv9{;dBhjMWujI!7NJ}v0
zX_sw^NjjS8J>d2Aj|Kbkv>$JQTR=}A)*MZ__EC5sM~>&szXm?bs!9Q_F2+Qj`n8MP
zJHY>(4u1v1ibn7?($IM<zogOgN1}Hpw?D!CJF(LBDsG^7Z?B+Hsf+A;VLGNO9VVpc
zavyMbZtB-Ocm7I^bHu_}w5wkSIh;X{e5p`7_2AXm!x>(vNonN_Vx9V>R^hX+MX9uh
zM1@;Cvpy?1y#;)n9Q*nw^=xV}@j@NFQ?&ZzLOZBXn}IqtIGV-}hijn!)H8Dm9T^Wy
zLi+=$*hDapWeq}clgn40G?MnnEMT&|K%!f3i+xFU*$Q+DBS&cT#P^h~tt|;m&An3?
zY-GSqC`2EQ)Teb77^VNgJ}Vutm0UZY6qoQuHQa-Y*D-GmM5A$_9smy)Sh$^h<cV+K
zc(~7rIj?)VQsR_DLde{0_%RKYm(p!;we?Ufc$Mz1Iq>b$^GpTCc!Qw;Ddj&uF$)C|
zq~gR0E@X%tE6eyW4af1D+mUu>R^2^2-&(%@?eYlA(sk^CyQgEl@~o`4o5Zj7d@pOq
zp0w+_uD2!aTpP#1SJeq`(Nq27NBjczW#f%x3|!yYS~gUV4*v%ag)PR*+5c$y4R+w*
zKhFx$NB`qz*T3JWb?D#5bC%Lp|KsP>zu!Xl%fDUowgzm}KmH}hvwt7Y`RYGk`M+dg
z@2!CsFm4iNiy~AC-4Ea69Q+R2UH9=}?2Y+z*pH=bI5<iJaC1Q7+f5wLmtbR-w;~l8
z15w@J{JevwTA&?Jm;L!KHZmV_;bZ^tbFKK_XSn_=K|<%_w}l}L{rEs9!};L2bQ!!f
z&~dK%N7!rII$u->3!ON~QL6*=u#68gsN&y{Cn-TM>&4r19Ji|h9(VyN+IpvmzPWah
zqZGfD&2`{!*S7yXm2(_4t|1l{#_C9t704;LZs%t93mjuw7|!ZIhOv)elpn|P=+uO8
z9bk{TetyCKp11qY5%|wRKlZ=mp#LXn|5vjP63H6eqNt0T6t=}%kmc9aPZs<&DMblG
zWEtlt#>LKS1v?Lz(`61cQIL?NkJ;G3l9zAXSD<mD35*2eMSw0({sN912M>P&RwQGB
zx`;8_?%f!9(iAaf7tZJI@i8%;P>_p3-1WI6z?>ruAG2XE1JcAIvA@vSP4YE?pE2lV
z+zSd4DkV&SuI=8vdxdDb`&~le@F~K;VxJ3a{}(+e&HTns7xf(wV8E@DJSxh5<{#C@
z@38pgTpZCz)lKi`;CTBTP1|3S4F%-@$y?8KUJ^I3RKgW<pkILfD{<98iXSkQ9_FVm
zaH9jia`N-9fXLal_}ld-P3I#nS1Y|kPw|gmIi;bo27OWN%Y_yoYcy-|FBfpZ;A8gC
zy7LCiN`{SF1%1c<{H2k?cC|z@VVokaa%Ks~wVO%*T(=j<-1{Lt7=uP~8YFy|a;dxg
z0|+OBS@v-LHA`1a%lL!C?#b)WetDq?)YVA42=;PS%^=OlJAVY`rI97(Jm-eJ3Wosj
z@bi`Cx*QX_IflF>xs#CEd~Xq0iTOc)Yh?dO%5C<)y6p=_>j)%3gIQiYG*-JSO%3a7
z=>-;MDIab-5*OD^x4cC|HT!JZLxUDT^{yta54Q$x6q#pl@Ifoczande^&Vs_kUF&^
z^rs)%XAmDpFJ{w?1do=I&<ZiQE3p6<Gzk8+w}(o51NcZZp=&PL|Grq6kk|R!RnIQV
z&9dv!HF3HU6kV*-$A0KJo==`U$sNP8__FWUE&UOk9I=SC^K^q0n%qziHY&yCg~S39
z=^m7iPY4NeLolVVcV`1Ncku6p*3O4QEk$$TEu+A`)`?^Kj{goBeuq|D;#GP!0s3E@
zqd+?-F>>c#rQ+vsxvfU?Q>m~lXbOxVT0XJw271Iz3Oa`P(&{H++})-M8d2ux=a2mu
zh^8KUWZ~+Lm8P*a|CL{G9Are%#mI3qc%-4!O}7Smd3%qp2U&riF2#poNwes_i5D{#
zM^I+6gOdB1lCtyVE$n}@ybdzfQ%I0$*;)(uC_(o7N$eC6@rRqd+l9Q03=*37{(8TU
zX&C4^>c<;A>Po}__{1d6&0lg{D<<0n$N6*MY^6HvaIs45p{?w1xGkU7G)$g$e*{1b
z=M!8H@z=k7yzi@e{x&umKX_Ox<V%ZbV5-I-im=ij@V!qVOZE5THcq;UHR09YcZF0|
z@J;m%8K0j-+k?Fl&uArH+^eV*Aa<DV=w~c49dLZX8K!IXGiT0tz(9DA@%i6f#)Yy+
z(82gN06(#h@>O=bgNse!A~&Sz>v#*hU4P=(&$DB`^c0M0r29)4I6&N1gk;da=`Spl
z4E6~TqRX%+N(o`a{w!_*LhHA$TVCNvInI7d4qughS%bLuWe1F!`YYIvbNb#^rBfd^
zQfSdw6fanq6foz4RgwrXGAiLVz=YmX_EK&PBT|XTVV_=8urzSC#g$-L9=&^D9Zjot
z2(AnxFl!sCR{Zl6`<}A@=PADGgQUnEbk|at?kJ(5WqUrHYw@M%)HYNNf^gri@=(K{
z9_`h5q=@S-zXN?SKX|AM5d%X5y=*MoR|ghkZ+JgcX%_}hM@-2@Jj5p=eMoLT)36pi
zYwhKCFq`wk%dEn;p8fc<{&1=mgTY$LcQldA{4}e{1?);cOJh<~#X)&IcoSI@QfCju
z{f{4&f4>O(KONG>|6jsGc2maDr-l4SS}XCpl&1J)uPWd$K1KM@;2<%HmaUnhbaaSO
zCDxD_O}b>~@!=iINIt>TLJ3sDvJW6Dh$JJL4x-sRV2j37v(t2g9TA^ZGNUJU<5D9c
zup3{P9DjV6uJ`0Pc7zh;&aHI1XK#@f4DzZIlDrQMh_iD5vdvoLRA0<E4ahCPpCCbs
z|I|k{cWxqRU4K2XmYy{_?23*Hu88%e>wk&*G*7m{+*7Jeha3o$wZsVOxlMPIVw4Hr
zZ|6gcq+3W=UE&Mii-`{erKPUE5;~$MvP2<>O=G+s`(kWxcDdkWhY3+Z>Sq;$?KMIx
zDCNt>WgA&}tep@9Mn*=6ll+!jd?)$%aV=P><2=d?1Egzm-C3$lUAxQvc)isiw(IX+
zfMBrz&$r?#HTZAta~2SB$;7u66~3%Tw}9IfzOynQk7Wz_sqU%J{d&6lVK#aJFn1n#
zSjx8m@FbW0Kf9pqsi?}&kNr{=A7%Cs|9wAVAMS_EbuBpCk(AZ*@RzEH&wqc$HGtJF
zuY;NpKu-8z+%80zCp13_7pKLJXO48xZ%3KwC~U<(Ho3JpxW7b3Dmy^qV(`(iZwUkK
zI9hE=#&n=Wr;pw@yTO9{ki)PNgW<pUbb!TUPZoi6^1&<zji4=C#oqF_pUT{s0+ip_
zDj*a(Te<=iaBCUgsl-zEMH~YHROH3gk-B~9>QYF1r$3y$sGzK@%$^3mjBkHrO9Np<
z_0x19abx0ZDWD#myTghS4PY(ixR}lUltW<q{E#b;BAqYgTk3lfzibR~<_6~$ly?RU
zl2_&+!rX!w^MePQYY^uc=@9>lwo4^|IRHxS0bwi1_jfdZ$cFobCfywYNQeie>**+*
ziMyql0-DMuVjH8jD+Q)Y75g?_9bwM_<Ic*dR{{&)(N(W>@i2ufJ=rMca%t9s=5^Ct
zL+$IeY5I;hZ`mVZO+ea&L+Ng{H3!zP-|t#ifth(%!5174*V3U;d2QdIzYe1Zal_f<
zF{UeYv95h`y>w{lngWs@NCYMi8{MQq4xjy)fbHLGd5Z21U|&q&Z8*1?WF6N^8%Vtb
z8tDot8g!uZj!L{?mP}d{+t3rk1(V<&DLTADcFjtXG}x3r;?xBL>^^kw2(I~BO|>>9
z0>_O#EIWmP-v@&1t0x}~-(_7umPv6O?xg8{5}3#_Pd73g+fC~}kf{9_Td~odF6E@j
z7xP?e;9MGO6I@-b*#!IQ9ONG#HJ8rSwivN00-bK>M;>@@8oeR|h$CW(L5?g?_);Wq
z&{DAF1<0FI%?<f$;jxjLG~_AawL3wwt~X8HVRVKM)D+w00{pz5Vhivk&!F>|F=QN)
z<f$=X*Xb7<ooyiWdBN0H>!_dK;ST+jt<_sLy~qTYN|r!d%!kA7sJ9Y=QR46d7^M_-
z7ZuvaYjXjtO8wKNKkralMRYL@?clZ-K9E3O$aIAT^DPBMbk8#$=LV<fX4>&7*4^E7
zvCOTn*$G2D@#hR(>#WFP<{(LT&oVPJ+l^*$U3FM1CecaG-Pdt-Xaa1l)COQYv`Em(
zFu|XkP7ad&w>-NrU!>|b+tN|L{uT0F0Mn&cy3Vf&&F6V+&xYA42xt*CF7vpQ{7qv6
z@X6?e{Vri)Ra^+HJ2)GGiwS(^vAkK$a)e<uWiWE0H-gbVE!I~`ce~&b*)`Q6_W4ow
z%u+JqE-0~x`L|>=QI%fMTd<{1eo6cKkq&a{r7>;MM5Haul)GDePDCs3+`E6jek!k(
z<xoKF%R^7ZWD(p0T7wce_rq}J$z2(0)$SdJh+0CtHf;-fXhmRB)yYZJH@Yy7eYq=i
zv{GF?Ocvcj4G^Xaq!?*vkA2vRB7Biz3<=r|dCaG8sPgBL6n-dCG1M}IBUy}I`p)ni
zl!-f<@=c9dcBZ%_Hlf)97K-qUVe=96yd`Ld&0Fiv1GbZrcGZi0SKGy}y9(_3r46M%
z6+dKIU!;wk5Fifwp(iQ^O%=rDUBH2%f!125-P8!+;`NdY7~~UHzX+zgyGe{B%WYp~
zc}0aW#<%Hun9!jdoD59=t=v$h6|`FpQX~T;>l^b>y8M)pMWP3~>Hs_~{Xbes>+LsA
z956%9qRV2=IIAcrDY*>UMZ`w~^dcT_9$b)?w!gE;CSjU+^C+tFk1XQLAKm%@2=Ci4
z2Qnhb<Je~Qg@%}!AE`GaW)#wBQi^BhcDihvcB!wO42m17WYIyBhUwEAe|(>N%(G`(
z@rD7zb7;1V(343_1SFzDoXHVVQYa*4Cf7G8k^G{WhbRC1{d)f{w0g+pn{KlzRrR}m
z<1t>yd3=|(_)k`fL&tkfja#e|`AwU|4=cuP{AHub5e0?YPmXCnf43%O_t(F0t@~-+
z$)D8CW)}U<rFG=s?)aFEnNO;@6b`LCq`dQ#g4e;F-~8=|rDbyqpEel^XNfrsg=Ga!
z_B-`B<vLubl$z}NPJXiM^|#j=OQ?0awjOhNiz?=Cp1X((HV)a)KsUy1IT|Y9J+hFv
zGH|w`A}t=xT{h-ybB8L*%d3$yO|M$R%Bz`hR<vRScp_#gm*9gThuL?&kQA?VZ)E`;
zq?uH#)#^`WpkdT`dBL|^6yo?YbqjiH{3v|cQ*gb?l2e8oXM<}L8IdE!pRzrN=o1v2
z`?Fwuz8jNPStF=%47lu<Dh8H15=S^5?Tp}=ga;UZi$NPMC?-}jk_+vzFtB8QRCKfj
z=2Nmey@WI{PFxpzS`o!4yK2-rt9xyYD+-rnThP6G&j7;C-nX)syaSWDIi^hIT1>Uj
zr_x=`6~+T|yw{5wO!wfr9Oe7FH$17Q$HmXrl060Uj0s*x@i*>@2YL|W97a6eB-9uw
z#GZ=WJ?YW;&K(tD%&FuqhTlH`t<*v3vhOVX#^k~UK`AM1n4~mNqCpdxTqSewIOZ{>
z%>NJe-a8!YzW*QB)m2?BErqPo9+ZR-8lp%fWQ4Mk8M4QzX%uHN5(&x7xMYuLpZ3Vg
zN@Wwu-sku6I_tha_w_rzfBcT$AK&Bn9pBp@_o$rX{eHcl<MDhv9}j&^%@B~&WfX$<
zAH^0*DClSD`ZQ(<G*6N{qM;|RS>uXoDK&rH-MVKs%E}KQm2swK4(Hwf5?hdY+H*i_
z(~!i9Jz!W8%#BGbCv;9PC)cui&dp5B9F76Bv)LFDIpoLv<g3{%*Vn-dsX<di#;<0v
z(R{>ApLF6!?(I~5=Gi$vIyxGW6$$ce&`jY}7l*==!;RPI_!>f&-ux+RJ9z+#DL__=
z`wMWC)t_i3X$qCjsZrIxS4G}Lj#o~ZN#q_1_V<qh#j)XqPF@L|RWRFB=jT^Eb4Crz
zGZ*6bDWR?$aXjs{cH6NkGL9rae(qD2GdF~maQBY1Ro@_uz5AyRzmBJ0SeSVZ9rWL&
zP8?<2<_^ALv@HdQBtiD?ja#x)JGGxAaZ0&ez!yK*TIf1sSL!xuD-s0ig7l6vkALLu
z+r0TGYKP(6%07PHaLRlRGmoU1MhX|E8#x8^d1vhDz8eij9o7p9UmbWnTn*jQ0PWQK
z8KoQ((`t=idghK`n=gKc>0AoID7%<E&G!;HP0uw({*m1Qd|JXi=5ZQnF`e^Zq19hz
z$8_HVQ+>nSLZ!%K2|6!x+syr+AzJR<v4eK4@DiJxf`U-%Q2!tcJ3IR{0-?f$bEr$Q
za<n%nbw)KdiGz}Q=dle+q)>2#X`{j54fJ5VU5{pEFk3Pfx1)O`7^m4IV%x9F$Hco}
zq;3398|K5x-Gxyre?3k<zaFPI4SG#L0Y!7)-r0A;n<e$}-x4-m_mq{D%>pwnS8|um
zVSQ;L_R18nw(ld@kf;bAOtV?C{Q0R4WQ;S=#R{Z3L;W>n+4^DbQe<6wK8+XeE{CaE
z<o0e)Qcpb=xaYjEgb|_vLa<LK&%$HH6+B!xywug{zpyB#BU>-dH$S;pd0n<?*8ZR#
zX@@ptrV{*RlME_}_tumP7ao%>)bwd=S(@jxg^P<%iq-e6@m7Nh#NY_q^E}T_SoL=T
zwV~n~9h9CgKb!iheXw+<1oZZmd2e-;2WlDPMqP-@C}SIJi_y`{tr=HpK3LO%{slaF
z!u^lOcwi|W<E?m%_x$5ACXza?-1<Tb`T|mDvYM?Y-nHqw)){W!gS`L&i_?E-weock
z7I|UwNx-jlac8%r&^o)iOrR{WAnU8wjwe-M8YS|c@DAoq(|f-R5k+pvD()58+*zve
z$|s<*6V@q!B6;aQg5bc?_xLPj|NbmIze378)O$qswNxL<I+26`x^wKZk&))-uTko$
zdW8FK9@zQZ2O6$pu~#-vAW4OcpMVlcZG!5aQ;UW+S{GoM*BGm`3+kkd%GeYh9e6*Q
zdjG*e=W#RJ@1Lrx|D^0EdQ~cb+xe&(2$r6jjLrb}WbJ-Bm*^V_I#>Cx!|wX~u!C%f
z?c<<Mfu&2AI$;ZVJr560fbcA;p`aNQe*<avcB>}D0W*)885tf<w;$3eY~COTu4OH%
zcD-CWMDUbwqY>Q=o$l4O%9L&0y44stbEAOXtxl_RF{i7M%k}${cSY7dJ+T~e%KAYx
zunN_}-KMX2YW_Y;zdxN2ooj&c>9P;~%{WMR$S#<e^LS#X08U9gXj}_Jf_<sf27pQy
zVpiE$-YOj-vW4NcWZz<nDl%$>86_UYF^x0X8Vb87da92jde<60)b6sSLk=H=NZlz_
zU9^LQ(L6TL+%U@acK;}l$$oTv*JeHCvV<&J)G-uQ)zf?0X~jdCO=TSWj9>b-Oh=ww
zoENvg<uoteBtQpfe)Z=?Mx<Q<7=a2}olrOpAhsghO9D+AlFgsD;N<i`rlJAKa}14m
zJf005>d2V-OVkm>Unyn=c_5xx2Lu^PmqyK~rRXGebQq+}$UWpw0@Wwv$qa7B%xwFc
zsgfYR=ZH9cpjik?Y~Qg%7L8avN@C-@7NwnMpWY(wFUH^<Ca9-A?M!|2=#dk0#_`$%
zzr%^&t*N1+H6_RqA4raP^?t`@hMcf_9^?}#j7SNN=@mz^)|C!Zhw>*9T965d1}`rv
zAb`ctz#tAEo}ALUXQ$R>m^CT`;r7_A689x-5KYJ~BtvL-?Ux{W$hqicMo}0NL7tL|
z3Y?z|qUuD%vvwmT_2y`3Xb`a<*UT--5XsWU0^xXW3Qmzuo5+gehyj0V%i^14fG0)v
z4Pt1I#cbsyoJRkex#f23*bQ=RAbR;6T^gcbASAx7Ow0N%mjH)Tl4s5DTeZnf0B8Of
z{^M7Gci&q4_a)G#?`#$R{gb(~5sC;VTd?_grgnJ|^ph3(2T_hx^@SaFe9_rm05bXB
zN(v;RUB+8+zfM<T(YjDq$^*>vjkvyUt{<RrF{4U@)bZ@@HXiKsjcL$v({8BIr{+Q7
zzBAXhp5ymVD82c{qUFaPu-?1`Ij>7YQvd!GH=5sz(fLl)b2{woDjY~}^nwjc%?q8b
z!p2X(f3~ut>a~k%-EkXFjC*jJRlAHWI)^J)u53<m2HPM+*gT}aGa0Hk*p$2avgM+f
z+iW%#2>KgDU-#r0iy%OZp`B<c|9uozF9@dlt|+7A-Tal6-~7tT1lGqTKZf3IxnnC#
zje^lpf6;WrvPCr-h{${QRy35XSWnshh{`GYl<3GGHB-1&TwY#>hQBWDvSOHovv5l}
zW0->+yZP6fEb;yIChvZ|NhXdKdB*m=9DQ&=P*8i9pr9b(p}siW^}R7I7kpJA#5U69
zCJf<lD}h>j?aQkLvYod|amzDoy3c6Q5U}%bteCCuJ_UziHUb+1Jy7_3xZ>?!_BOx2
zqcmGr;alp|(5wCAIbEHnPG(%I+wqn{E|XV@ycX1W9+K5(_majOn^^RHqN*u6m{67v
zJGxj{Btsit14!FxG=M*yt({Q*=i+T`(Jl61yMIB3=vuZrk&vov7ONk$O6>z4Ng^pi
zd@SG{QxwKuEq+%*?tgulwRaGpoAWS|IIG94t9SsJm?Bdtbhni}k{5t7-or1)Ks&Mr
zK|&SprHU#HJ=HD&fwQP6#2km8F>@b|0~M5rcWUN4q(1QXUzh;tr3|13c`8t<$gphF
zAu2JbGE}hI8wk%71a$<%I`C~XP<=wElGwvF;3H}eUjIRs91c2-8n5TS@M$y{8Q%nu
z&1N&z9!QYvLUa)o6%`{z#kq5VAt4FC@O#15tik*h(Nc<NtCbWNFSWP%eYd6Aak)LF
zEE&o72j$xej>=;i*;gY8uIfGh`<^iAy~8swO2E*T^1q6}w2vh)&%e(l)2*|ph0i=W
ze5Jdp@%D|@Uo(S${WAXB(~yz3{^L*ZKbJ6W?LYsFnUO*M`DdJrQ2oz8JM@a2%76Y`
zb!QY=|M<f%OpFx%k3W00V8eg%v>1QqzVN>v5LxlRR}iP+zh@8^;s4A-a{r6kQEUgO
zMjlK2i6-ud0q7JDU!<ieOg!ADtFwN$%3qi6;={Q2fD;ls(46Iw$9~B_I5-Y){_Q+=
z@;Ca<6N7b#=J*YMZB0Z&=3l>2bxK~o29=}suF{MXjaZP3k_+73sb+=WIidvJH(gZ8
zLalVPM*Z0EU#@+Vyx2|x5CrmTiy_K|;ytdQefesE561LG9gT8^!%w}MIYiJJ;7Z3n
zkqPpqYrv<IF_g+v-jJ0^{^dV@{u|<OIsWgA!K}A2K-<mF&riXU+;V&W)yJ@S|6_70
zyctt^mcf_>M{4^48(eQo*(MiIg%k9)^&r#HW0?inen|^9ITY11wMbhOX8ScS*mSsP
z^woQx6>dLyDGOiGH?Nal`qhGo>Cba0cIJVZvpN(iN6R^3m+);B?}fc!m`9`CFv0#r
zQ?^wUpz8SAT9q(S8yOV64bXFbirPCKBRCW6Jwv$At}anSFd`RFOd`;SezeWCsHI8a
z5gik=TS7wH%gYPRWI%#S{8N1NIIIu$VT;Ne!Y#u)?08&zF6^BC0jH@tjN>A(UDs9k
z(9bUtTNiz(@vhBmIjU@J{i0xYSRTOlX_PAG<m9|R*3_sbUl7$sDVG3WB@`Rah<F^<
z?!H>^yR#k3lsR<!0LK&EZmb&#QRZFDTe8o3tm3H$$f%N_vI|(}SptTTPu~hrhu!Cs
zAIpaybm%$WeWTUGz?8|IqiBHXB42RZxuVgpZ}UanuWqI+^b{_$@o>NN7NyLJQ$bpQ
z`of;xFambRJ<y9Oi+c0Tq60wIs6YKeLN*R=-z<$WR@az)Jx0WPD?<+$dUp=x16kg(
zZQFY_4*2_;Xfo>yWTNU~hZT!3roKj7w{M@GrF7rc!q@act6Z$6jT&44Y&Xk;CbdOF
zPH2G4i8qrBC@uG$-%&;*Z6d$FwShWLp(9^E{Nw)m7<(-S0B?}%en^B|6{&uX)fEn0
zx%9cTdsM0j=Gm9p0a<^w5Yt=HIQ?zCai;K!VWB_P`CiaxyW<xcYMMXQ6)FS!G_$&M
z*yhy7a*>RYk<n_W*X+|(IZ6|k(hd7DZVz|;8}bKNMFcOR?0hk6hNf+~{kli29=oh(
z-rsHURnl+Sv>MHx4DOu(TJCn3YaPUz(a|Whj6rRizqPk>)v+fG?Wzvytpow>rXht4
zNLv&}y&Y|&*B+RB?=|bgc#$%j8Kd!;*;$H~u&eS(_m%LE_NDQ4f||xYxP5`7;pNMw
zkXK&xxv=Abd~G3xB5N2M8|%!+!J3738mP-r($w48Wwv!g1zQ@0bOUBr10$cXyV{2>
zl6tUNW|MdD@hRYoDC1_@PX4%5Sh;FD`L^R<MQ<)yH9kF$MW=rIxt%*jM9xFRFdLqH
z$rzbcq>R80F9CJ6DTW+J%_8YVL8A+vD(+9qkS@4m|FuX6LKAAZmt!C?5lUWWMR8G4
z75YtE^uT*?4u<R&ufqcnH*?=?r8M&a@uzEYxcoe=7Fs@W89oI^+9O~R5SEsZP1kAE
z_$A}^e^G7^J1=`Ac`#Uns(?3JImN;Fh2U;xK9r@N8fc=ca2l8Ax5X;*1nyQXv6_OM
zdmW^}xE?NiJcM4WuMHvyn6{4Qf^!px{uVaF=Z>cA3Bd869e9fca2deDyMD}rH!vHr
z;>1xjBIJpMB$BLJw>FX%(<$krDz?-K$ZthqaAk^aM9gLKZ-RD%zu<~5xAD-`2;df>
zl<&x`q(z)VTg+@bvf$nQi<eEkTztO#*wJ7BXaS#Xdb&SlVS*`i-*R#GY_X2#UO`pC
z(VidO+_;hH%b&O=-c~oSE*qO*IpF$3z>;~B^8La!5VrnnaaYa&AaD0zlHZ}7@7Rjz
zkJS0TqKb-oTxRWE%Ma=5=sY0^$*pI}cnscks6HUCJRaFNRF43gTleg)GLSU{iP_l(
z^i3P3-132*CIV_qzzY)CBM=1xLS8RHZJ~>sfJs#^y2fHDc*rRdlZk9`3X>;uWKsRv
zsg1AoMH;Y*_VMR!0`ynI2jW}Yy_3>BV%>@b+u;lH$59K}t#~*dzxgR~`!vsmve1Q|
ze|A_(ANa8H#%)InXY@3iwwH%Xwe`2(!{~9NwoSQEa@`2D5Uv2?N)Gh+1S1YtA5Kwa
zf;Ejpf(6|<W)dbKND%y)oGh&;g+}EY=Kw!t_L~xM3SX}0=xBFt+jatvWYbIMEFHM)
zswSuu54QmN-??SWTRZK@V?1K4C6eDizsv)-U|GHt!e;WDl~q{N8v`QQRe0v)$trjc
zJcI21bEx42b*sfG9YxptQJ*T&_0tU(ac<gv0h4mi3v&8%E_R22@!ALAO{kR|#cexw
z^y9V2yJP%&1si9M638t%(~v&;?)vxYr>nx2tN!~EJ;Np9d^1Ka(P46lSa6BNKCCR;
zU&-ZyPHVM4%BI1wIS_!dHK~`_FhB}^ZpiC|iti+;uV9Pz43EDLw}d7p)E|KZ#vnJu
zpY-ERI2&~bcvCdQIeM|k^<wW3hUQ}$mw|bYA>SCIe|Lf|NiDcQs`!4$T?v>5VBs$g
z_#Ox(nlXkLbIYKx8<`ndyburXp=qTD-mLqQ!5C$Az?o14?7pHW?Lrxnl#s)Zi_{6C
zUyA2%MzsPGvN3qza4}I`Vu`O)j$^@v!^Z-Xwhap%zI~t}9}BQ0<>h0!L(nYwh|8sl
zo4uY-Qw6uL78<(;-DVwXhH*blhmJ31CvQG1ghyocU)JKhT<wdy_W)!A@Er?Y^Eq_x
zb*MAbtvltSpFN9{_F9`<w=Q{Tm%;Rxi=R1<Y~4)R@=?Izs$^Hr@w+D)w<BlAB7vww
zaY2{lNPrN%*fCMToQLR0i1N4XU%{u7uMP<gpblf#u*DsRX;m(u+A$b;pDQjd=6CV#
z2?Az-E2@oc#<Pf6<dpz4s6(XT9KqRf5U`;qTT)hbO+|6PTRnnfO-Hdm5iG^TQXX*E
zZb*6{IVKP=1*9P&YZ+xH|C&n{LrnnOB(QA69v|i=&v83<W3oq9YkQ!sQG<~`7v;{y
zL)IOhm>~4n>Y}8-kYZW<VB2Qxdq`<=z|`dstBBKqO3Z0?#s@2XcpgXO8U)??;mPw6
z_v)rH1tkW&M^3a}eY=vNs;E^GNCW^P+kl&5i+x1*k~#3C2xJ8a{wz+q8rV=IQeb)-
z2ZfJZ<cCr4%X=UR!cXCyAT0wjdOXgYSUBwijYi(>#o_HjLUq@4k#@#l6+yBHl*=C(
ziXv`*1pDHWV;Pnu`6fJ;kt%8C$1$XXvFUjfN}Jy9H+#5{c*}4X4w37DDdnX089?$R
zBLRHwC?q6OeA&`KJDIh^8J3Yn4F;woFN*8OZy(U5nPdhPHGB^43;}z>>G1gE8|(7<
zo_WmU<B#25rZbVGou`KJ%L_BbomCzcxA*A!C#_*F^O9X`ohM|SXWwCG?VNtG<wsV4
z%{<URQ#=jiF-EN17Y`O2)IFkmCvG0OHG@p`*|jWmhPq;qqhC4EukhvFu~|O9XkH2V
z#uM{KWslz;w)bQg#{T%L`vy6LzqM<(G0?D%>L=uZIeZzU6z=Y4F)<sFNrgeA(+LmH
zg5Tf=!0rnZ+QC*MPa<Eb;>nZyk;sjn`{Lva((}QTXzbhfcK<#6JLlRQELZ+P3)p$~
z&1KyS*)O4enb~}?3;WLp(9;<{FCefkkZ57jIF>GdU)i*O1yPNV0XlgvJNuB9`A~gQ
zs1+zv;hlH%IKzgppe=xFZq_ddvEUEL>mItDOPT=$zRU!anvhuv6AW+N_5$P&xOzTm
zctMIiCmj=$gaP<J^~Qd{f_?k>Jk$-BU0}YRSuzB5l`Eo{FO>~FgnqbIeIv&B?SS3V
z=5k2o8wK8erqnqII{QG6;B`?+$qQ}SXCPMy;<On|YdqYm@y$frup~eriOkHC>(<L5
z$ytnlMRHc4^JHUyxyfxA9Z-Db@I)DCA8%TviRJr*aShJWm$-aXjE(^!kle4)-y)&)
zAZ1~@5kj$l(l+{wj7!ow1-^A==4>7PN36?HD)7;@HG7-0QhIX;OLDcO;nix7ik!fd
zsBK!Y(a}@05^4XN(s1qx>*X_ttp)Qg-EVfy2ye0QZJDQV<VcLNiA1uCq#iVK5tyM^
zxwz!(cH-E`r5jA!h+XTsC>le|dIq;2uyp(UA>Xm1Jd{J`+U2XmSrk1X3ddrNzLpKe
z8oNJ!SgF5#n09&-f2N}a?mF(b#{5112M@k)7+ZSxMWuK*IVp(VW^{d0{j}N6A~{O3
z{t;G)!nO3WgIC56=S$&@t%>ug9^WpzVm{;E<XWeCo*grh;pd7s1bizTkyw2uCmjkU
zkKbOl{vN5$)614Ge2t`G_MGo}AM$2{9Z@XFEd6_#fDygs5m4cV7mx1*BSA;u%r5ZR
z*IRii0WC<#okSa#MUTMod}9{|!Ny0Gl|kSU5N*~km38@!BeU-7iLkxllug>mlHC;b
zxzC=rPQ+c0SQTcIw+egVJsF9(vrBr{P44^1YDh=@#67S}J2%~`tBhUNkRSfWQDtxd
z3+CqNZG7M|AeA*D)F_tU*tGvxXi95Q{hc+UQ(|gq8JF&-tlaYBg1*pph?MOr71O26
z+1A)#n}X*am7Nq|7G3?uMfH12eCJOz;^~($5?NuLmoo0(oaEOmA!6-pEMRd%+(OT)
zZ1P+E)9)uIpR8T^X#Vj@-GK7^obF$@%XyDr$bmys%~GjBg)49FHOxGr7vvpa^)}e+
ziNHF~yXH@t_qS*VXrC-1r;i}j$hb}**=IP7nV^5{BPBz=W7?e+>`x~rCuiGZ6+&ak
zWIjcVl*s0-xvIDuAE6hWj$a5nWOb~!w-=9C_`$1Rj!?x2&jiDwoN@Lotw693f_^mH
zfWbMDuO5ZcY#)k5z}aV=oN`H@i^r@Na#f-5fa<cBcs;Ly7<vIbzgjRLVv$dG?b#zw
zW}rwC!~)|v3W+J21uQ+l&X{;U_j;o&D85gi^eBK)vfP7ZkZwXLZi=+UD38<9Y~ifJ
zcOko~-EWVqFXvh0r)&6=LP^=bA-vtvKZEMSCVCd;2~LMA`W};I7tjm>#VU$ROKXsX
zlOTrp=LH$`?I3JM5)-m72Xq|1E+XunOln(}WjG0fxyaNKKVB^i0psnJx#ml!46=0d
zlgcFE9~$}`S_e(nC?L$KVhP}U|6zw85xoUyK65qClUTnfYE`*g15xaD3$zh{v0F5)
zA7hp{<LalrC6Ha@P|d#jb+&=jwiqvP1+wodYN$Wvmr`8+*)M4T6&ykZpI^$h%Q;<u
z_=P1>aE^n<2v;*^Cb7gvnw+X`(-%U=u8aOGI$|;5(_-P%GU(O9Ey^ZRKdB{m^~4)g
z`7fP=MJ86Uw`}ZJcG*JFx*JWBMxwM<5HCebzwFc7!&%F}?`<tm9}M7|bF>YtOR9!V
zY_F9I7~!+YTLc8ECvkfeDyQ&d_@MA}<SfM6ZcY15#}nT~S=|ONN^CSh)5v{(2#!S}
zl9i$un+f3j>4QHML%W3tfN6ac_zR?ctrQMF?D!FWHvY0xYT-gG?ZzltxT)50Sf{Pr
zvl5AE@Tcc>kRTl-_yf*w)zpq30jaN<pVn9M0@{@9q=U>I!Y=V1r6B)3rG)$Fo>DNJ
zS^h|nmd^kDdFdaGs4q;wducJ%CiE!NF~V`pq$Gc}n;yMF2(`o+NdKUDQ|+iAsoFo~
z*{3|WO;JyMlU{J^)}I7WsGGiVTk;zJC)Y{TYz}-W@G6u{v0fQ?vhdtE#5XnujFY7c
z?oNK4{7?N>eoy#{8|IhFAjW%$XmQB&3}-+@SSz%$?O0%nzK|Bj;;E)9$-7XA_2xhv
zFwN{OPCVa?;737T6e9aaY!4ck4hC?|*_Q+buM=69P-el_VV5_sBg<LVBI|R}z-wCe
z($Cpmbpq{owd_lLZ=ub<-&^T5DsCufY2mYCkf%QSld1SrbDnx>+nu99FVapZ2YKHj
zK_OW{pY~p|(wwbjm$)&ueMZrGkUlJihV&|n2Cj@EfySgTWvk=Q_pEiGHP*~wY??+6
z*la^%V<UHpej`5Y2t5x#LoPJ#T%Lf9zz*({2Dk*7+N9`QzelYWQ5L1GYxvlXZQHJ(
zo=X5+JMRqhwxIHQdhvi2nUG=heR6ul<Q)jga)j|=$bo5oEGn&z!g+8}(=o-%<Jr7Z
z*zD8A+{*I2Prhsd$D&bE{Fq<9FFyjEZH_VxR7_%ZSTjdTI+P2ez{jAfDW7LQRFBF*
z6&Op64b-#}fP5`(iCM64u`#Ma)W@|bfX2|!g>r{JdK3@d6^jEMzzyk7dmYh>G>~45
z#3XTXEi$M@cW^qz4%kp7M)U{hQm6ky8%l&4rnJD$Qo()^M!~XR-1|4!kb)5xp$CiN
zHc4D}F7*Jy7Hp;9JCgD*xvrameoS0Kou6Mo8SXeW@cCu)&@jKuu3fvhQ&O_OEjwS5
zgtmy*4;oyyvc=X-x}nQY;54yHXO8RsJz|?^gdTt3V1vQs%X>iTORqp_u9b;12k5AD
zE>ecHnl1{NGCTM0;ASrx7{A~Aa!Zzf^A@i5JB87=jz4oz++36A;b=qmdtIK&|DJm|
z%RV)raN7%&y+*r+Mf5)~cjeqf)6z|lBa~6yBiX26I_dUmYC#}Ro8<YT2{MI-sU+1I
zQ~bm8m*?Oeh09)QCsExut2<BhU#j|+qSSl1cAFe&9!e?Z!e_gEn%jGDk%{g0u<k0h
z{YM%u*!U!665X&VYC%zPaW#UjQGfo$DI|~lg7EwIBXGZS)D+QO9K=ldkD=IS&w3y^
zWf$!W`dR3F-abH!=o+KJhdqhuf*R;0DDmT=(IP_#Y#?II_q(F`B&N#?O)4=^hc}0c
zkl80Cq^YyW2V;O8eo;SY&0AUkpgu!?=^#-6U-xNrAOu?>a3&TH)X?5AC@t6pJdaMG
zVFZmLN7ZHTt$3Ll#Qg-cDG9(ylY=DCN4g^P=}ZLf)%bWoU1r^Mm8i({k6Ev<<&Wlk
z@5C<Wq4o^O+ic>WJ^lt2%Y|JcE&q!1Pm1cVFSYv}=gU8p3l(s0?in~so&Qell4bem
zQ6XB6PUF@eHsAcNw+wRRv~uQzwQ%>ciOlkC<K*omErA4D0*@6IKDN6G{0ht8E6)cN
zDW?WU7q#7E&l0E{klJ7v4!-kXs%;CxCKg%3Ut1xu?%urlb+;8s_vn?v>0*GEv9Fh<
zt(cww{Cb`O(J{>>%$$G8NkmR+G+WKDr-D&chXiaM);)emGxu6NZM`hvv?ke|R7SUE
zOZ2V^b>LlkT_1{SOWE(BKLG-jSvv8sF*sf+@<uPcWg$?wO;iEr3Sv(1#bn@!3*>EL
zW<U~X{NubuxJ@Zv7GZ}9N!Wv8;+7SSbLfQn<K<x5^aOd<SsVzPVZX#YmVgB3z7bgn
zsC|NSi6TZgfv&Xg6qzky4dTo6!y<q&#vn-ROaL9Z0wx*(E1uilS?=2Nn8VdrsB!h+
zo`$FsC{}2uS=;~Ix3?Rub8-NOMTe;=qrFpnT~V*xGi<vvErX@Sh3k2vCa-R2YN+X%
za7i%n8iqUohigGO##iX;5g<|rNn5w<S6gS<VP`|yx`T|iPU&s{fC8t9{g#6~2eO6g
zqbDoAbXeLjp26DNdfVjv3fToM^qw?R_LfLeQ)qgRZ9uAU${+9Msf2!#@ysXG3=a9i
zc|tCGr}}m2A@nv7IF&#!A^a=o3jf&7BGrGSqz&w;q>&bG*=~<|WHRtWKG{t@=++NI
zHP+jZW0Ch``IM*{_4QvWgt`TD))aH>f2K#hz~F7w9I|`NOwtYSlsV8FCh{b}i&8@e
z(co%yU2gZJWfzM6rcCqD27Oat3tyIZ-{$A6?C$953UQnA-i}p`n=si;285(8saBbc
zYDEV`(`UW&fCVvBu<-5o+KOuB%uH3^g!#dJaZ96@7gkM8FTA16I!*zmp#A@NCYG!H
zekNQ~{Y_nbE<dNem!n7q)yeH2%4AtaUWs5ek*^**7;MU#k+BewwU~Gr-oov6k7GV)
zt%=_-z^);9_{<$&ob@3$Kti%|GbQwHAgO%H-W@~bBWTp};cynkMYsOi8ge?5+W9$m
z8f7I<d<fUd4ySvr>aC<XzRN|O+lMRD0Ak73Q-$->&r#gPfKN;u4oq5eGYA4fh_c=5
zOVn=6rG0*3co&o*0uT}}LovJW!2qsY-Ck#LB)4+QViUo6o)RwMocXF-c*WapO8B}#
zS+PeFyr{ASc8(F!8<;vD0aHkP4i~dcz!G&LtfF`$tdo-eMBf`yDPXq#FT!DIX}M5y
z@tXqb355xOOOFQgu6{JZo>dHO5Gci{AGaN_NAWK<TgpkO!v1f0n7By9h>eZS7-FL-
zK<XIUJr0pGSo>Ovb#N-mohr$6r^G0!N4<K$vqh;5AVwtzIA0Qlh_Y>T%hfHqT8ymZ
z%&R57?%wa`q1iCd)J0<OQ^OW#*8%%}tCFuI8=XY*+QJgNj=G}5uB~XGt7w42n<Xpx
zVI@U=Y4pjdC&yy!SB-SdZ@#IeukAgw;>^4->z*ksUin45Aum{_LIu{XD6?sP8}_17
zUx+EuVfg~1*{S2o=+`B5Vq;>OBt<me1MwlWCR}2P&x#=D0EibtkAmmn0;X+haJ3ji
z9GS{jZVbIRmk5GSw3g*dRWR$SFno9p`4v-4kO{x5^2?q_x@YY-XVI!}M+%9EH0WbZ
zy0oo=(sEbCLOYOiXIi-VruKKf=vCpik^E`~gdWf`rC)X8sMlT#58G#nIUUIJfXtzH
zIaBupn8l;Pk?zv>XgFpQ8hF%PH}FH9pqM`K0RIMc%oV(Kh`4<+nKNm)d4)sX{Rnlw
z2t^8HaSd~W4`Is&r{I9=E%|4Wz;SJPG_}`jV4<L}Bv)N~()n`|tybHidtsjcbw6Jh
zb3T@-5VHX<vw_q5hkus}GReCDU6MN3=}8Lx{N(hMY|{-rGK1;*XR`p<V28DHhdu$H
zl)V4<QR)Gy6gEBG#fVaBn2hjmF!TtPI)7$TFqKOl%*6;CkSoymBCNBQi$l<$^Qr%R
zgoF<ugQG~_cOs*opsb64Inq(!?5eK}F{W!2s~E5{*b$tkp}YT)1pnUP*ZBw&WR+rB
zJ{_sp6v*?@=OO`#+}plZs+znum{@h_>snUUSd<uXXoJWM1q!;_WRNo~V2t7I#B*O1
zL}W;Xr(z@*OCm>+7BcY+eLtV$cLOJ!#WmQ^?^5o*CLRf%-3lQGe>x0&i25TfEe(X|
znArxDcZ>SG)H+Pto(dqC4>Yiz{#sP8{Jop>?H}I<rtE`D$P_b|m_eg2R;fY<_n;WQ
zPJ1h}7MJ!S>BICEJ!<7<pn9U#&R6d@-@IwH&aGP5=&>iN_J3yp%j)35=qf|tE~3Ns
zKki4ju=cm#5okvl)=GQ&ef3DUfEcUffoH&jlQXyBNY)J#DbuW8Qi)(r3YouTbc_8&
zDC}xrD~8p!1aAdQzw(?s-6g*Sx}4osw(=OJ5QLiydg0Mgi$IM1icTxlHf(-uRC2YT
zMvd@H5Y8J_cslR)%(8H7?K-*AQ8Z=lm|RG7-_|=+?q30F-QkGg#AN0)o7+Kkm76g{
zH=}-D)>YT}b$)ogC!xY4hzPXOf!%!K?|*fDA)bIhgpgia3i8ZYA$?8}J6Aum9<A|_
zyjD>z8IZUmt=vua?Kv!CO6a8E{&lw>a0S(KNFuok>K|^`X%b@Vlg&;QWbV^Mln*iE
z%Lyfolzd13@<k!0iSu9j`t3_^Q{B?FES_P~VIxRV@1COX^P5S}gj7IZmm}e{-@-+|
zEPKn^g4p}Lt#9*lC`Dh!Psq;)9SeP3+}5SHjUsOle?(dzxnSl)@GehBi_7W(Tzd1j
z!j>0)CyimhRd<_3Tln+R*_WM^Y!=K7FvzPW^F#DZToO(y0849tdxlW%U1D>Z{Ym4q
zwC6bux$Cpm*4Fd=#0Cu@{YFWbY~;=7=!oLcM=s*_S=|#wL6S!OT0tKI)}f^nxwaQ&
zZ130{+Lr13zE&@?&4WqnqR+Z747?ETd+Z(6sGmN8H(R-1-}ZzS1^{pEP6XF=Bjmq~
z+NVo)vj39ua)y`X#a!T@`XDY~u%>`D<SIb_F9-Jl*)N!%Hcin#xB<9CPPIq(_qM{e
zYi)Qu2Nl!!HBm;L1$sB4iwGO?`DJS|z!8Uh3%q=9YIP84aPN+y$MEI+T2lWYj{$f|
z7mRPee*wy)f;Bd7^HY~grxORu>M-glwb`;BMV1!gl8^dPqg5+%TGx@nbxyHkyytcH
zPalDHU%*@c&H^+0K8|FuH2qH=(B7-zJv+dm|3SPfXCZkcW^g#ndGY{CQ5%eOe^eBz
z&Z3Ssdsqk#Z!=Bi0itsTHp<|%c6$)C*4c?wJd)qf({rzYb}>Z{!yo}mEsgc`oqx=+
zXG{3vUIfzHz}nY@2i}nZ??0g|ozYu>E4C<O5PPC$Q5Mf;Y%Hp~k8@B>_}sY^5V(W7
zDX@`?nIZcfh0X0jn3{x2@>>Ar5WD)tUAGf3RUgj7M-W?Ejdj$`T5~at02Q-Az_tLy
zfqt^%hXVTLU<$fDvSCwNOf!~6>6pH%^744zI5v66lxNnW`Mn9+E$$Rld2Bygx7rPT
z9<b`ZgJ!Z9fl)+4vgyUPsivHhr40uCMQbYXcs%^|c$ixLQ7nYQL=_m>%O`0dBG-8J
zLI}H*R*$B<M%Hu-=HvgqFZN5&N?6=sq=0LeTGapNVR@gQ<7#lVo0pFwA6d>QRY?*E
z&`!ONuA-hz@UuWA*CAwqPf$xSa~0UX;y9r(kLaTN?*;RA%<217mX#D{*x03U2@1#3
z6EbeI@y~Gn{0IEX+yXUOmb%mi^w~@0#VZP%k`OQmrzL3wBAov2eKyXVWDG>Q?G#-A
zE$&V0my$hz>66>i#J}Xisb%UemO9-UO2p|2&C6dxoH3@|6+KAM(d{kv(=S0dM>iQ3
zmIS7;A`nc;(nFx&LVY`)Oyfay{+nz%IiY5_yq=Y{+tw61u&=_~kMX$txK3CVnAg;g
zi=rE9fWzDAiH5kdZ--|gc76;4Ai<Y!W}zNXjaaD?dxopoeE5#!vsH7@#NM7VL_Vw=
z-(bNf*sQ<OzcGR;xE|qe$i2h~h6{%Wo2sJaSz&72JUVGy8q{HGf8rP_C0XCAMy9u|
zL(l$;3sBtpySfVBS9RfB<grCHZe@Thp>5D^+eTIuoMIyh_Xlg(Iuq3%bPvgNnM|3)
zO=)>lG6?|wN3!~lVdV%mkpcmu$ICJ5w=5bR)cf2bc_2HFck{n*D?tGQ0@X{DJ1pDu
z9LYS`DRnBI_Y>Bsk|ESCg^nMoMIw1E*=bgJc5+P$O&G;)`xvgBYmk&-2}@z}x07;m
z!wv(Lfz;R6+>HI7VdUZw&4q`els<%w2QBLJDOjTXQ7~OOV5I`(Dq9%c(#HT{kP2;4
zZRW{~c4M{Rfv}C{qQf+wwye`7`Gl_@Va%lZ>22)ZgTBN}$kYUx!NWz^<H0)Rc37@<
zxKPZ&OfbPo5t{;lFl1O5vj5Tzh&`}I$Hwjf#RSlBHDVwkeT)sZYT=1Ki#qxEaP)^{
z^!`0z_MguZq6%Kj7-lepeV%F7_!N77z4(#@Is)i4iott4g+6bBg=aW_oG$(TG5T(A
z0h9$2u&xk8wKHkXOoGzIrD@MpLQi1rlN{~B{P)N7e7Q*5_w%Bxd7yuoe0%4`Jw&ML
z;jZJ4bUSD9&`cC@0Ob0Je%Q?MOuC``a0c!V(%BaILunZuj9C!?;MNi$6dWR9v(BKK
z*XG$<nm4{6?1O0lC&a2<0!UR%=c73*DMCAW69xd;bG?DxgkN(xW($7J#RGo#tn^9T
zy-Fcq&|OEHnW8U)-nVA?ZVTV^FHCwbM^FRSa|9B;+bMuSA{*QPi$P*jLO%c=?(Ua>
zvNQYd8?fwBey0w;dao#YND<8OpHN(Y(ZIDqt)^WOjC70*dFhjw2y6rpD=6WHuq30W
zr+3|stWJ(T2Frk|ze6F!z$tMZHc*90fwu*aS9dC5KVgD+xCwDyIEA)&20CE$Sz3EG
zs%*`mJa9pyNOpBwjH6qSG`>Wi6BG#!pq68Y$Sp{p<?JThFr;ijG-bw|V%Oqc%>yBD
zNS^~JnyG3L@!6@LFQqUR54KU>z2(g=^C(&Om)dx&VE;B^AJdw82n!F*hK-GoHa6Eh
zj~ZT~5|Zc<y^h%O-7k@X1EW*v2(y|Ca~H4$-KNm|(BEI*Ne5Nf-kCOHpJ6j5Nbn(o
zH(zl@2Diy5z;L$8V>GhO)gcm($`*e93R4Y;Roe~Nh+&IDqO?+i!8;Qm;!zS!YS~vW
zwbSPc>6be?I@F)ocNDE5{uq0HG$L-U3%8YV%G|0uvx5zJm+Lb|jFIIH76ztoJsHhf
z)Bd6cJ7Mvhu41fcMJJO*-}>V`A3f}(<uk#b7@5jph490m(PA?!)&zwo0ebmr@w9E^
zTbY-0=AN|&$f)t<=Bycy7K;ME=H(U^UEAJXZ28ls<y~K}z>0pWbw$FU>(%5Q8<Xff
zR{h`?53RF%2m`R}&|s?L$c>^e^Ov;cUL(zqEAH*Ie7i}$v6Xk5{E0W&m&&9Yf-gO3
z=GL8q=$S!QoMQGlJI_|&_9Z1H%A}bCgxClCM+ImNVN?R_(;yd_7{pWa?0kXE5(eUx
zN1#0Pp%s|#$K%ouOU5d2HW(l=q6fjRpf7d}3wr&7c|&TTv5^^g^8i+Ul99_F(w*SO
zLD0&82M@mWZP6z^3-Wb8Q9uSrmVyzoR^jaMU79L8?s*lMZ%GKl#%N5Zx}gN11MakK
z65o6e=G(No`{?=3#-0fyD%AsRg}P)B1XK#BbL?rw%EmNsoCJ)?7rO*1?*DKWY$PH4
z0t0j>xuO=!VdSAA(>e&jN1`y?EhMChi`jsc!KZ*Ldohz~fZ`~=(T3vv_uM7@L&^H{
zwyj%P91yAInxK<W)z5<s!U5SuQwOlEUL$KTX$n9Lrp6w73`Mc7HYX4M$qIWxUNr!m
zpIt^2wCx%Vzgo!E68g}yRT!xCjnNm9nQAhWOGHBLtFfbX2mH_gx)?4fqOvL~T;NQH
zFbz0EW^lNsP?L{9OKgf@O4y{t$!B&1ll5n?N;v@y*N2s2QGA$+C!Nqo^7BvH*YQ74
zPpB!T4YA;ARG-zx)vN<Ip!YZ7%aULULH8S7w-C4|OU7ctw}pEOV=gwq!M_?9TsGiq
z9PBNc2fHkHrL#YtHRlT&&XNww4ZdHPBw*yC|D<`lxO`U5xy&6q>=%(uLo3Xr8w*}U
zz<PpEAY3zdVTTP__0E8`>^-(ikS1wvmw3Ks?aH^m=n!UU;#y#*nAEu^zluW%*SVez
z*eJqUV~VhSPu1&Bxwg-f3YOO$I<Vrc0|NEo<TIo8BX;YR<Awsv+3pbO1t;@1j26to
zhB`_PV6={DDwq}G`T`)05udq*^N-%qXefN|2|ylVZ~xLU?W5oU-UD?$RA-Ih<vka&
z|Ar%mA{#~rF=NurT(w}A-GdI$R3bI+aE*swl?N-?WXcZ`L9}X+jN*9Od#3Tu&+bWf
zh*ZClszmerDYu3;#Vd+CFb5b%A(PoMQ}|w0w`nXU+PpgmnFSoyjK(g@<Wq`@BB*s^
z-1|(bBChv_IYD*rSHS2x)7Y}B`3#o<fp-d;70Kk0K|pXCti8;3k3+Y8`v%&*OSi`H
z=4(QRcCmT~ou-^Q*t)BEf3V2q&xeP*%2#g56~rW)0u+=?vNKL-{(F**fusZ`R^Ax$
z$hEnQ_*6c(TDNvBCIU)?ZQ@O(W~mM$h15d7rnG(Sx^+TU-=uxm1wW+N$$!M{tA)I9
zVde5nn6Sztm@?0_DlJ*L@fiUk!L*a4x?+ei3XfUmm(~@&7k##z>TAu*f-=<-KnJhw
zoXJrIoz(=rD@fK421ai5f<87pJrAbZF=aYt-73*?>(7@EsfEvOGv=5)!V6kxxW9#v
zKM@q_JlAZ27)iIBbISvC9+<aXL5LvJvstk$m#N{6haLV#xJX9!i9=1lAmjC5;a#Pj
zFh?P9AJSd|W~E;(e9eJ!rhfFmNVFa#>1Fx4aY~zk5LPB_*9lmQ>w(FJD+n*7Mj99a
z1z70fE2e>#oSOPD$qH5lLRBa@NoYzh#%{NiG^B=r7B_bLk8&Z5YrKN^wG$w@tN`dH
z!}bguRz@-EBFs||wDckrf~0YyDZE-+>hCw+yxB6ymGg3wwX?zIEY^3~sUL!^WaA3a
zRNtgJ+L#(#oj7p}P4z=G)djzX51-$e{2DE@ygQsl|08tn#uhTKLx?3i8|A?Q$Dtix
z3)~_Q6Cc>corW-r5^AzK@N^1GlJEeM@{QbUqIQ@8aYxc^6CrrA#*HZ@%IhNlp_j+C
zqRgy8Wi<-M`xWrLrqRxlJ;yLf*QN~F_mB)aiZZ^k4INYxspwEVidlWTMHpxV!$poG
zR}Y)7@gN`MC}(jH>IMo_RYLLwkqLgH{CF)*#Iy5wLtD%z;@}gKoGz-~e)7RaA(Ll_
zzP;!zt7oT1GH`Ea!%r9)#Xub+o@{)hK+i=GuLC6MxwN3045Cq91OdJ3{5cOGIUVgQ
zF_Sr(REW#&gyMmrFH%BiqXbO{e5Ho7P)7(8h}2|bQp|NaoxD8x_+)DUP>Y!TAm?ah
z?;RXJ!lsowbEY1X39^Ptsze%<hk!<eJwcXz@jXcq3nE|=Q06@t?yF6Kn|7de0%0Y4
zOkfnF*aVykBp#djOw42n3xu#V2z-RmDItD>{79IBz>uTxa)ilC=xt`Mj?7B{gx!r?
z)m`=|3RiHw10Bi{WDhgU?chmz_)GCsSqX`(OXel6vaQ<m#|?L06ZY+T$KGw~Q+-|g
zckT1Vded>&ZYrf+AJi9#e#}cN*(E4cqx_ZE#MbIaDT~0-_p5pCN*`S9#b&V9;8UU0
z^rhXGS9K>!gjaSuewJz(%B!Cqv`@*XEVZX-@4*mwBj~i=Xgp~{ZWDP9z-JheP>^q>
zd*U&YMOhJr;=LNr*P*Up{cq6Gl~<`%Ib|x}-#@zB(be2sTT<?2rf(5_I#;uyeocMD
zlzIb^D<vvh#Dt1Cg}_yz?H%#7Qpd6=k@i9YeTV`;iwNsM7>ON4%d7`?QmBvUL61gt
zt#dw!je=<wl|Hu4M)d>r13W5;W6F|;EUnugpGvNxpPAZZI{nR0M6q(?;g@nc6j>I_
zl@@bh8bZm1TN0x~-e{Igi;L7<t0=O0c!YnV<_YB6>YM!uobYpEK3UJl_3%=c`y+8y
z#M0BLFr`9@KCA2@vxe;Hjy+jI?0({pKWFROp960eVyE$Ph}aqLwKW!Rgnor!vMFkB
zmA=)*uW~(~JwEQEZ(h!<a7p`|F@D1&{H>g(7T#*R>1PLM6GStoRipX|`1x`GQ;lfm
z)V&&(eyb{?KBQV)_At@i=SQh{G@I~DXpR}9^=lt8hVY1Llb3GtV!ZTRIdi@8L6vid
zF7Gy1a6xJ`y09t>eD?-zg)1AFR863xG$tw8<10Q${6m5^@ydP0V7a+BBR4mf_|M3%
zRxmu^QLH+{SwZ}zCuvfHqtgKA+93qC2YwYDp&3HrDl@q;Y-N|M>kfO=*E4Os)LOw@
zm#_15|6{=^eT7w3*VHL#H#1mPst$CYKBv2GPs!p0HFOB==vP1Yj-dc~4=(P=T50K2
za|nqMu1S~<=quLWF@D_U88;mKwj<;XC-DIqD{#ZeZUamtmc4HZ@r+4L-Ua{C3mKJ>
z8A42_)U(!|o$KE=Tq)|WTR+OA;4vn$24DUHzC3JU)z1nm)m4b~Cm!mFxM}A^T(y1i
z#Ap4v=Im)4;ig_2BJV_ulDBy?-ZR0KnIH{B$<WTp71SGIQ@d9B&m)mIU<3Ujdi!x*
zMmROKV2x2fpfcj*_2z0t2USSoNR9rS&#&)$Pbl@NSwsmLAiCj^7#f{Mxci1jng6IX
zOAUBpQlo9?$-J5^^7u?lOypS3b_y`Awhp$0hzsAUm-PrZ*~2A!<XP4*-PSW#$Vf5u
ze$Bc4e7TLW`N|`b>RIQ?`?oE7E9x&rzB6@9gxv3paKZX_YcislnIfu)z}v}a<!I)c
zYbSe-bpihJBC@R8ykibLdfS0-N%?tK_Z@K!QubJI#7|vI#JsK`dDmiGSRuxRoziRl
zoUZzx-_KyKHC;o#TB**hLY>^t2yZB7*hDi&e#0x4ZDvy9OEi5T=u!OWIFox>;zQ$|
z-7BDEOMau`QC~x1?e9Ho#@e9DffJ33#*-{v_bA8L&XD{O&Fr<T<M1C$ldS!`53HFN
zFtG%UZJ7FkKPd{!Sa(X)1FCgIcktgoTmR`gvEUyo(M$iSnSO*wVEpUn$RGbOx>);9
zqYL-{T(@@r^S^KTf8@d{g{E-51VO(VGhvqjW+?6IVbH`rL#*N;n7C6J!q7@N^`NHr
zt(G&vU(F9-y;YUi&wZH3Kv*3r+DaLK8Itk`_E?^11Z-8&kPM|W15#vXqdt@)<||5F
znD(<kc8SjxoBK*v;VMNTntsv545LIRP5HO_662^S75{&}`4Peok3!bv9(u|3VjJ)=
zywpMA0(tymY=abbd}PtQ%B^0*#B{p~Si7=El`(t?J@TRA?zn9q)IkDclau+oQ-$JH
zcB{Tp$EUG70DKOneLZJV6rpKPWMvVQ$j4MtAAXAetcwxeWBXgJhVkhG|LNQD-!p&z
zzfbmmLnhPzpA&<LJn@@S;zEPV4HAPlrq2dkc~B;F(xt<cY&|jkK0}*<$1n%_0#l`X
zj6g88l|+>9O8#=B9Dv|_95XQf3#hQ8zR-HHg-HD17D*xIBBBbDU~)wIw7FW}Uyupt
z6MzobQ_jnPEl-I4=fAi`JRzAJVGfY{2_tcRtVAvnmr+nqfY0$xr-?!Bzx5e!-XW(0
zb=|-0iE!O;JOZx(oP=+1>tpxB&4KYRO77(EgzU|lNFu0J^!%7qLYg}D4~8K`Q6TxN
z|M=kCM3<VV-h=%RIN{n*Ur!XHpMu-q*v8PSIig>D_YbBc{vvfm&(uAixc5GW1?i&3
z36!sMKN#;Az~zW>5ThixZ-3h<AV&5Yt_l-VC*!k*WQM5h4%}@EON%46jUd5|_SNk{
zOCOTu3c{usP>F$0*wP!$gnI<8kc%JdA}06k<_W(p2;A4x4#Dyg!hete`@D*wV|}FY
z0r^KU3Zx2AT0Q0V?B8#BhFqSI<o|ffCN!pcm@#@y)L=B@fDus?_Rt+Y8JbZMS36F?
zX5v&x-x-Kj9(jFlqtNbsGoqO6{r+w$f0FNJ9Ky(4i)@;%#C94=aaS{h7*&@2`a2Wl
zc$I|Oqxkf7{pd)20dWaA%^>l9z~Cq<GgFv)mQkv)_>#+iQ1o2uF5H@(anI~+*o902
zco<s4?%urQSSe6RxfBsW6n4SgoDb*z*PCr*yqVLl50Zg$`om@q&W~~C6X;cln)j0}
zjElZ=?|=TghEI99@Y@wbRXofIH6(n;uZ%ldqx?U|*MdyXunoqYie7rLxd;|wiLu7@
zr@vm1lktMeF}df+)xL`LJ})eSiWrAtd^Poz@y+5GIkPIAQ6zdljimN*!q)maFXJzF
zGBaL4>_0A(C6L%^z_6s;s81CV5=zCAK@SWozAj`Gq&rtLe&xXL_hhlPwPkqN#-;+s
z9ed08H%dPlx2~5_ZL%mNQ~S(%wn#AEv4nBX53XdS`@T&$RWV=D1>FU&ng&ecqi=jt
zk`EvnX^5nHvk@^6*s&Hd4*LFI2mR@l-v^zYwt`2p<@d3xyU#dw438oul}7H@y=%f4
zznsB1w0-Bb`j|GcV)|%N1|QVgOYnhMf^X^DYS+w%fm`Z1T{LV_kR1^~@v&BVG1HL(
z(vK1!5@nCGV?lgc8=wckPf96+O*jooQ-~I8&^?`xe;aYH?itQm{7rW=NOZ0KSILD>
zzL50x#D0;y2C<qZR;?O+3@?R<1nsM@7ZJyWK(Q4i4YfwmzT^Y{{)u=%oI(EmbMn7c
z{r|PSOT^Rfm>3DT%}z4>a<N9`i(Os`2xxPV96_|niEsdpr!PVeKOwGU_dLC;*Res7
zvf^J{fFeJlufCYa6pQ?Ra?u|%aj-Xh;o5#a-fBlROPe5z+=Kq&jNl8<DQ{1&IdhkI
z)nl8Rd!)OPiqP{OI~MxjSrMsv2dp8r6Ay8oxgJ>o3>FDPHU2G5Ys6pPk2rkdS}5J;
zO<H{FYCrE|dC-Emda5o6K6~zFn3%E=FRJ#(HWgk(cb%PP2aj2zN=BX}tlIsUYsfNh
z-T;+1FHq^;6Q_9~GV^D9{*GS_1eOZWCb9jAW7i`hdA%{84(ySJv}is3v)@ma+e_b9
zH0CuePA12VpC<!rK4f~L2f+0whFmma1_=5#gOIjK?qo8S%^429tOY-APquVRvH}_{
zJ5+%FoILd;<gMTk-dix9kwT>5+riNF1_u8g@H5A?*$c@T2p=C17xWM~K--8>2Lmoq
z=cO2N&n4R7klwZwzyee-c$tuDj6O<3!?j5;=j<{CxAeos)Q>6zG=UY>YRCiv>X&&3
zBbKMQ)>mje=#Yqhw@9D5brjQ)%_Mf0sF|Ca6U}%}tg-CJl?*o4g>D8FBOMxM=k4Bm
zl0C89K!c>m<Jgl|mJ#3*q%bO0QzMTyhvAa>6mYCrdH(!)R)lOUDpE-zAxV{&7guE+
zb}AHF02giZT`g$_vtJUJwl&FO)smq>4~yRq$YxyRTaB=^7S`6+%isfHcV;^}wbdXC
zCE2%N&+QvlI_`-NW^W-T^$m!YOd%U{&X)Mh8XMl{iC*jaU7zPJQWOne_#5;$PS4u!
z3-6Fw;C{zMtq=dWuLdhP4B35RNB`M>ADCyJNM@b3;RjdZX6oGh&`?|Z4x2Gr5yr-z
z*B37Nfb7AsN@7&$J9v!I%uAN(!3^y#^et$oc7&-sXm6;idw(T1Yy$)o)eplK4I5<<
z58fdcDfM>9>^eUj6)5i(U|k0k9fX5REm%FX`gm$bpz2Y(<3&DyFo`p+{TU2G+wrD%
zu?Z9)#xi*yi2MnRHLba&u)LT^4UEqA#l`@`&;k28e6~B4R%c{nSYbpgiziBcBMsvo
zhNm_b^rcZco+2)4aa*mU0tBb3OLj;tyS+nX8IwXNDvXFkg89NS_pH51#9lJBz`{uA
z$mE^RqIbR#Xzy^0e*!UVB{tjgTrvbo=FjbUN~kqYGcz-5zv4OuNZ_J}%^KU@=ehmH
zpJS1GOsTFLQ-oP9t_EJ7xL1q%fIn_G3-YiJ@qED8wp<-s7iMh*5@Zs9`H?R!{QJw2
zBaaS!|NHH?h?~s2uZ|vE4|Xz@L<gXDK~^mYHXNkH_Ivs?lK55VPQGC!XJD`GPSsAx
z&nJmfI$nb-%|Vr%HfES<W76>+6ND)8_`O=Ngc!l;V_uHc?z{J97+%K&O#sfpva4^D
zH2)5Wa^En(*=LJT!el6K5g@l&N|UQYc0|!4U>Y~+NKHFR#3PNwYJdk6(PFVd;H@^C
zR`1pj@26~tm6L1#bE1$gK|Gu>Wbno)<paUniB~PLnxiVZ_MIabel`?q^e|^T;jlv`
z4TGKcBs?Ux6R^HKM8kr$1S-UAD2sSZ$YTVW>E!2$0#>a57Ufbc@qC`}WU+=H$Iv@a
zeYl8605;R!f&cYw6=m``k?~1I4XO@TZsJ%$fb~NJltmB2KGA?nKs=nkWlK6F5w%tL
zgl~l5HRZ8g(-o^41rZW^Y?>-DJpTsYw3D2uxQ}E!;D}Z(q6RIQI-5|3a7g@4iNypP
z?DB{k@NSL0B!=i`qeN8NfbhK@Y%;fcE8<N37)Z3}HWsNK#94V_%ASmC#RkhbgywK#
zvp6r(4-i)k;v3-7cP`u;avB8B!8sF#HZB*Ix4ntE3Pdy3Q(#T<4dc*#4yVDp@**xF
zSoI*aD+#A$u91(0SccI#x+qo<dnz3%5P9$h-FxV`Ii62XQjC#gKngRoJ?}C9AXkRr
zpM#Y=(!7y~I$3V%whq}~-xu*S;3`zy#Wj*Lhq{F}u|8QOYIiQU?L$*GWRk+Aq`E4@
z32Ejo))&OcP$j(`A+0G+eCfksF@D<6edWoH9a8L=l#|MthW#dXa`Hj5q=A6e4uuyV
z63w8pGbS@cM@*(7bkb1XuSK@FM?NI*Zerph0F;|J-IL+Xz9B=f&aFTNW1IpEk1MW(
zO3n2u5<Bl~%+(M0He)A!WX-o3hO3BLq9-_K?;vBdkyyu{etbxP7?HmtrXRQ#71?vE
z+<x37gZ*``1l+AVX_21pxN@rc&K^~_^j&4<Yu^`9OMHSnOiHLR%JB{Z84Y@f=>kZj
z4wzR-1?(uN`Ro`zYQe%(q_J%GE3m{KdlUbUB=2UZ9{7(9y~a>l8JyAGny0ZmSjJWf
z(wxobhFE;)S~yiJqwO@WYI{ZM3)_-ivCl};!fhWe>8gy$LdnLrNRa19c!n&Cktml`
zJO*R27I_1i*l=w19@a>LfzyCpsEFm6<^~LB(q}kDEH1okHiMv%{$mg+uQrwqfgbzF
zN=l^V>hXGYkfk2{e5{LroPcV2<@^Pd*L}CvohfpfCE2gvg%h^(hK^k4-`Jm&%qq9d
z!y2}PEr!^aUEA<<%MnxXl~1~n8pMgplR+!7OE`y}aTkgX+E#zN?<fX1=5j8ZAcZQD
z@2&qB++ExHxsG<tyPnuVom0lvsakzb9`4q%13bjaXetw2W`ntzBi1BW7E76u3-~jM
z(<W-oR4H|Xc<jg8;iywmv2rEU?W((Ss?EaXcdO3M#<lbmvha#D-8T=_bGtU&QOvH7
zhCqMgOUs<I<P`)4@GKWihbj6CzaP25<o*Bx<{G3iY!8MuB2hCY@bC%m-BJ*^;wC$i
zv#_o5K6gfR8~)&P!~nMI*swt0MU0c|khSyQ!w+7T;rjyUL4yVeMUvsTL8_^zhb>+J
zZzTpx#73CYrc7Xwl+Je4Fn430I3;{80H;V~z^->1%%+0;cY)H$BQs$`E<3<h3(KTG
z*PS!rgjCFx8RC!$63+Q18#jD9D^eaHg<7vFxQ-!dn;};X&TXQ;oFmu6e*2TVh=>9i
z_#6W1B+b_x6`2cISJd#?jvb<iZ}j<}2*H;Rr`X3me|~pQO8$YE(~s5?@2e@csdUVu
z9Uv7h=<A3?;LTEA$Q|zb&1;9zoH${%A$`dDk#=||#X72Y{ehItQX!!xrBvbQ7Gr6*
z;VA2<&f{+BF2%^@Jj7{mXAW|>G<G`$eILOPeHQd3K4s~tvszk=D&oSm>Jwi6V;Hx(
z<bw}f4Z_qrEA0F1BokU`Fb6lLx03jafwY3l>;(uvixOX?g^O40Y!c|VL22ADLzE7<
zk;Gbh$D(}2b~7LP;vS|0UV?MWwHgISYD(&h@*xu1FmdB?I{jry9V!}P2_yVy4nyTb
z<2cOeG<tnj!FrH|_1-SDbVqbB;PU5V*DdMKzEF5v?h<Sde`5cNm>zPZw?aIam3-oQ
zE0S=)CKaBfJ@QwS6or)D_+oQ^MO?!2Pld2b`u68A9dU>vHxrg~9QE5{m*L_Q-hZKV
zpDp$=dN`DSXV&<wiavr$hs>4--LjoBr5>}!Q>Y-A>_|5s(1IfMg>MeB=H1n&+*%#l
zeXKx7=rpHbPsbN&VyEdE9buHvKA<EXG-*cW7mv-`-yx3{p&eQAUPwsDQ?8LH(prF#
z(AqnOivCc4VjAX^h~&@hrAR496b&4y`@GCXX25lShq^P&Ziehh%^WXTy5sSEt2|QQ
ztrTgg`r*#8^WmSixjbriJ3DPJB)4sj-VbkEV_fa1%sFBgl}hZ*tPq=or!7R01f1Y^
z!|&51?1eX;L~rrLstg-1+Ru#V2QV<LaPk<uFqjDxqe`X7r)zsSBg)3t&l+cR|HlO}
zypIdQOYk1Uh=M#L7bT!*r!Hx%C3c~+4V1a_pC_7etg~l_OGgUY?qnnFM`6~fK_j7t
zG-zy6ZZ4d5pDv{mu!ukJ`&v|s)u^aQ*TG4=O2H5@aCz3eU<oI~c~!FQ*iGyK2*JOi
z%V`G1$dAapYu~Yldu2$>Z-9B}O4Y<E*Du$X9QYuKXJVCvo;^y>N(Wj5)FCg=p5+ZE
z=1R6@sJm*$1Bb<uyN2%THDeg04^fpXGlkQ2DIZ9o$14-SqdWca;emP2Q4Fs?;89e5
z=Q(eP^@pD=HkF2)f~cN(PvG(>&3DMyoaZN(hk_)eVsTL96nWU{qBKXMI!_|gcrwn*
zb(Bzs-*t1ZT|r_@I9a`wpoXNzdf}Ug@BWS0ZGj~vgQ)!!B1Kl%<E7z&I$$A!+n9gY
z$P)A^t9p&gC%{}KB_nwt$$8PkpR;G7BW?CrTh239CRq6nCxi4MKKopI<$u<M&mRN{
zCte+!8Hw6?!nn2@2>L^g7D|(QY)C*PiLeiGUq$c|9yjlLU+FfU=Hx$tvu(+V4s3`%
zAxpXS=_?WwYCZ?veW?L@Ze*2$xEMAOAFiKsHpHrm3KDFsxCLS;>4w$8O=liD_=wm&
zL1s$Ow-F4=rU_L%y2=H3bWs`n9I&K_!ww+a82^Vf_o5Qx<8e{G2@_+)m8t<fXJRv%
zeKCu6+$U$52>6lzoei514o*q|jN`aTd11imUn;7y6C1JyALj8Xq*2>=vj9vy;^@!8
zr>%3`?Vx?<1D-&Ra_t95M8;DOc=!+|VVn^0M?MZ`Ji|9IA)RRTv(uSLfu%8lP}uyX
za3Ao5itI|zqK#VI2#LbN9`9--F;1-C(rNHUJbn&0tt5L6T&d4J+2`U_Ot8Ec4_a+q
zo-0T(=P=7G%AUivf&awFMUZ^1P_sa-zTu*_#!nLy!lnU92j3Qit}$OntizmJ<3}8(
zOh9NJ^JT}~+0Cnr6y*Hjkp1`PtNDs&Wu*J`d1lm2P!L|3_z8AxOa*50_lAZ-+h&a_
zei}{xt4R2C1)r<w86=8ALk?b1#&)eOdzNCfay&8Qx<Vx}w59O)0c}N}R^f+3NhtMV
ziR+$IRw2%YpoKUJIj1ZR)Zz8*XHXLl%*<iR5NbOEyR5D;&CjsPJ4dDuAN286Ga?!^
zGtpCICyopQ(cn!Pc?#RZK+90Vs3}~&)xKP)Az(JGo>((Bln9Aztj?x%KNL&xlW&Pv
zvF()g@37_#k#S4^XdTrk`A{X{@2Rsn)uYS&ILgPIpT~E}f-&{jicak1q9*u667bqW
zF$6Z$DD1!TF@{moTO+2Fd@96rg7?G&p76Qvmj%b^=!B@KT=>oDq%;SoS@o&D0eo(|
zpS^rrfi5CVa^qF329b~>E#OiMo>$E{mNAI&W<W!`kV-C11s!O0D(uAHBi;4+{KS5P
z7nqRGtG*frUUQ?PZG3+oMQZvSK`(gdle8-{6lDJQh)JOmRh$apk`kahcr<~CtB-T>
z^W~81K2(VhZFwFgzg5<u1^(btUPhm5qjLdbO3B*S+-^Amw<afDsc)WJ@c2VYG#kgU
zW*n{;of<r{j}!^IP2+anJ3TQJZdV2><2S%ayEIRaX`pSb7eG69!V^<rlF8iEQ-?PW
zN3A?xJXLK{>LbG<EVWxi^GS#7o}0j<2$^DM&S@(1J<rDvJUX|w>VL5->O5-=$r2j?
zx#@GLq^^Dk|8=D=AMdx@tU5&@TSJ9R)iBfEWKKcWB;`zhDl;g_*{BiP&lgUsg6Xm&
zXs^av)cU@7NG)S;{lqT@Tj@s6S6@O)_7h5MgQHGp{Z9w)-;7L{@068?&PN|``y#LS
zPb1Ei5)<zTO(=bCOd{kUuO61!t)O{mbU5~|If1P(zK9vvha@+MB=<K6dmx&Le;lV{
zjj4Dt%K+@^_=zzzn+Asmv8ccaa7VE@TrS#r`tWchxP}5KE7?HaZ4w}a_ja4`nG1-E
znL7od&*A4|J7T)Z+MMdJ^<t=?nhtdaA!;aWKZnSkK3hapW6n9t6p=}IElAVi>@)IV
zW)^XzzSgIco1Zx20P58@-2A&#U_4{v#E+dco46hpS6!g|c<szD-0q9}!jI!9<UarG
z@K?kYLG{>vPq4z1crNx|I4CW;^%V9>r`aHZbS8z$BP9vjqSTsxumec!&i|m)H$}Eo
zNaCtSEAjXP@tVX)Cak7)vv((AkuM?ou+sw2-c6vr3l_C80{phkoU%L68_zm%U+WuC
zwD0zVSj3KYE1O6C7d~vYtvb5oJvjbPF#)(j+pC>>(D%?cPgFM|%eWPV(GgaN#VFC!
zFa<gi?Z?ro5jle6<#WDo);PX9bE9?Cn#K`QX~xEP%kJ^Q5&J>_iDJV&oBBZrZ=&BV
zY$0A#?fNuqqO=l3)32rghJ24x3sD8FTsq+Tzu0>ZsH)HHTa+|SG-`~Z(!?mDU<Cvb
z5bOv7QUsJ5DI!$?X-d%~Dk4pa2vUqxsi8{|j7mok=}m+UB0Yfg_U77{bI$*a_wKmw
zj`4rvje8H{jFF?lX7BI!ed}9mt~uw*TGwwV?(8J>Ey%+^!4J6oDK^Jw&|0tJ`R(<3
zR9NQXFT|hew5?jXdiBDD%ffmZT=vdWcjYK9tf9j7`I7|=sCg9_NE$nWT|V;PLb>uv
zNt(-ld^l}&<TKA5?cxz@b#R*E_32%tIrR)G_zc`0ATYVlay{P7u!VrkT#?JHv;?|r
zY|B$Hes(R?NA^1TmwY~A?dk-mC|XUuBg(eNb{OD>NgI|8<lc2VE8!Co?s2+h985~5
zs${nS&?4|mb`Te6Nkbd%eULgX-g_()h5V3hhGWDf--x5Rfm`3o<0nnMEEIa;&qSaf
zCFCN!Blc^!S>3^C%B|hQc!y3pUb?$SXW*msYGJl=!GH^;PwoS7=vHqo9R%j-PIv=&
zu>r!C6ySfvP5YC|>baGHgKK>iqPQJ!BdkRZT9)mFj_2@FFCGPGUF>o12sx_}#tpF8
zU;eP|u~$&>geJRy4@{`TVyg45^909!5t)B{)_|;pSNN=h6st#S&WwipQ@*p99z*Mu
z>dA&lnT7MHfY2p=tzeEVKDA6}TjR$4L41GKzDaKM^xsD;+Y9{*7hz$ga@qrCP}xcX
zsy_5_qdPxfdP==L3^_+_TdacGnwpy2y)s+cF&ua2wAU?gp4|zNM{f2bm3^07W|}Pl
z4M-D4iObnWVTMxS?+_)t%j_KV2Mbex<E1&_nsB@DI%UB(D=LZ?-ecA8^^L4S6)wbH
z2pCq7rzC<Cb4`<RT5`Gblc~QIhN+6EyYO9$Sp^CK^Ogo4%K8js?{V*n&f3+Dl^9uk
z>TY=IJb3EIn3(Bvl-PYvo$X6Ow&ov4_PF_YjXh-y_DgT!Y0!vh>n3%eqqwdxC(G>Q
z#i3?6v5p81hTB~bMNAO%)I};q1sU_>A+>o7bb)*N-bbMMm3)cH*dXU-hZg29)maj3
zfY6=Gl^OhDVyV>R;qNy}i1s~&Uatu9l)iYcLuNEnHu)VlM_+I6V}Mdj?2!)5)85!P
zJ&dJY(PYP$#E_j_`GgliOQBg^MP&`LK<PRs%Cis5iB|o97<*0&ND$kEIQHpezmb%t
zt<)xYz%>I~?%LTyZ}X33O3?j@LPUw*W+@E?Yt5BeRjx0(P)<F(Z}r+sjUUl;(-E}D
z6_*gffk2-^0fM@5FQjh%Q5+HQYT^9LmEt@5B<VR&O4>9}Z~&Anjqe*aZwAzUzxt_}
z-SFhcnetC3Gk{|hVWMUSikrJAcnVOA+sYQJzh!)xdeGJyDG-%FAI2VZ3FhMtHFHEY
zTkbGRa2Ki#X&p~(xyW44vN`dHAVTTbW~{RwD(w!FPBbw!WdK+&oBp<?FDYM=hMj&A
z4>c45t~%MPo*!gcPL(XS(?6=n!T2|{KRVq)xOW9)`E$&z&#LNGvcTTp#op*woL{h7
z$`bS&)&9i)bl0ayJ4utcQ3D-!i?!T{>3;5WfSD_^z2HwEjXSexo~G_HxV{(c!O$nV
zei<a@q=C@Y<Cjnu;es5UcgY0Jo^lM^Sv+gv(iPj6S2}(WVw))|!NFtzojl#_f`ur=
zca7TSS_QVDs(eMbh<o=1C&`7W$|6d>Ulr|%LW<Hg&!sG%a&YrB@{E!OT@@8y;i}L$
z9Na=c=nP!w3gW!}vIV&3HxQ;8zEN7^B&7~5D6Fag3%WB1M4^k(Z7nNq4pfb;4swh8
zf>lULW4HB={nMAIW(r0dLJl!xaub-Hht8b^*x3S?+jB)u@cs3~>`flYsgl#{eu^zD
z8UL6J`)a|w`LU$v0d5hqvxF8_38Z8}t*nRMLjt5-+IAikNf_hQM}yr+-a;F}h@z6s
zUdL~9TAe&T_wk2&^A>ymzEn5HaelBXVu*lWqkL(Bx;TiaayfGd+-yKLPOd#FKt<zk
zL*6E&8ZzkTMu>LATT^IWAaDq-mVTn1NQVoM2I-^2Xk0{BZUr2=C`5NKx>NwpI`(O{
zx#2pb8(7e@6#{=c+sEC|qkLpl>@ly&%-Kp<cSvchq?I{OE2D*#k;Tfy5c%SfnHGS(
zWT<cA8J-d&#Sf%hoF$+O8}?6f+)6oogL!5vVbgfL)Xj50y6WIQUWMSZKH^3A_ATdz
z$}NG77A855#cjaSQCS5IKP&A;2Gn^TpqTF?QNL|7ohSMIt$y1Jzgi`+H`HZ*N)$B`
z5d^4FLGzn~EeqKMkfplY-MHgQomnuQfrppc?++u$A8c+_y!=eR@7U^uE^?#t0E*N9
zX&Ot!KyTo6UvlYrXYl0xh6R47fX|Rx7?rAi_eC^<I~i-8fU<Q>k@ELAmvnP$NMPxQ
z(NWpOct8&qwQ5)a;j;vp0n6ARXmFC|>v}>6v(xf>n8djuS{4AVdFy!&<rXJSl_hq@
zR!JQu*nkEI%^4@4k!Yj#@Ev2s#wC{8R}B0`&$MsLMO^bi{8?2UAA~!leRrrir*VEz
zv$$e}$MQQT9?KCtmVbkjze~eaLo@1B{1ywy4Q2516ta-A?QckH4!!bL=9{}SNUTq-
zwNnQ7$2JxLK%C1CKfso$3A7jGv(gh7ZANKim<2YW8>!L;cN-O2)}E)xly?*xrPF2B
zY$z_x-elY6G~sS3Is4a3<IG)Mvl~&)KLOGe**cCvi1sf1C|m?L0%RC0F-{b8+Kt_A
z^WwX0YiDO-{OT<axZF0#4rR|V_t77`NJHeC24L9hW1E9f+y)Z1%IJ{2-Y^07lUj;?
zw^c${8BjUxzf9Gt_sQ?56aj+pcs9Yfm*!CWxYQ9mJTkt}kTWFur1YB0dtEWt>&w)3
zDxEpAZ=Kv)4=%-IE&uKJwUQrRJX3RxRWo<>aSu0vpkw{}mk8+P=q+{UTFxGM_Qr<8
zJ=$XZQ*-TJ9%U8f6@%sd<%4-`N%grp<$RS66+)GR&J{1MKlr_ExOnV!QvKZniTC86
zS*w~b!m~W%YHki!-4YSesO|RQxyj8fEE1&gQ{joxVQWq$yCUhhiR8gMe{%t17fAxa
z&J`mC4KZ*WQlGmDz&=TB-dWM;;D?7PUtITr(6fP3KD1Es8&pe5fF_>nX;vO-hdYZq
zwf}~?)QSjq1nLo&HjCza$<sDm6dYgyd0^;XA)qFG=+rH-ucDpgwzIOvIqh${7(NNT
zDm%29c~>4tI$foHP1g|6{vXjYlF@tzoaOBK8D+KdFvj$(ctlMLa+6^k9o8K7F^rji
z*#T<84qIWgKx9LctIA~!wO&qM91s%Gm~r<I;VhJ1Ue>UkWuF8ufVr*R(6oC}j-TWu
zDz{a}=a>5~?;N@d&_D3b9D#f`P81SdeN<TvJ5sYZ+P}YH4})8k#2i#_3^?Q@DP<vV
zyG#A<tTPGkcAS_Ifzugf<_$n`t1Ya7@)qGz*|8IM_Nbubts!dgfuI=KG{$#JZ=SQh
z2pJenfTS{oNB|z=3-saXTCz^)y^X~!f<x}((_xZq+(HO~&-2+h<P**m3cGu)yPaIs
zu~(#N?*{H&-|L67yty1m;A;fxsC~5S2~bpV?mM}g_(pVrQ{v{Z%g&q{a)dcUXFE;K
zicjLnkCys~5J1XTjAeJ-*X#h^nK155O*$X~+Xkko_k=ou6|J0DHRh+L%!A_QfS52(
z!NoSR4uiNY&~^^aQcO@fvTOPJ-R`=x9qWk^;x|JoG4i20jyDji9o<AvP<f6^nHvwl
zs0!Gy$wJXfdVo~wA&H1=<*lzwOuUM_STD?J{7&2KhyC9ycsYY5GF{kk@%Yrn`n{4G
z6B{^ot<(=+aeKIGR#0Ra%Zi6GR8}E4oclD?CWY;mLo1;_{i#QIw<)0vKoE`-_DduV
z>dgYg968%uL%m9B+E6A(xVieZzl=f6vP3v|g%%$d9rjZpsGIIKAO-D!Sg5#H%7*y!
z>$R^a1p`MkUg}d%_O&GS0!(ztKs5lm$6d&4PkR-XDVt{iTpZ+}-BO*nfXbE<u~Rwo
zv<J~|h|q3ByhVd_UUA?m>rnw0&r`#8C!NMnY!p%s1`H*}su@^qkqb(@0GF~71j&cC
z%|XsD<+Rg)+f(I4WKg7aAMC`DORRMFSZ6`IztzAjO>}YhTkD&xWjPyeoV)}0tUnIn
z2`ULtGP#3!wu9ObSe}D=uz>OaLLfo@!^9w=RyY74eGoB+@%OL52KX@yHqMTyK<MzP
z9{5@3BsECE@|4VIu_~jM3!#?DK$uP-E!X>Xg@`LaTX;*4>Jip~DjZ9cyPCo(it4CM
zV>JKvQ(yfaj0wLliW#m9unF8g1D1~g>k=gU0AQWO+^1^7_8<%sO4)s_)Q>M2@=a#H
zm9X`G<WtFw7akr*)X~SSK}}a?w-ao|VI*7YI%j)Rn=ZDk5NX&%1Q^&-=V?I_c|6#v
zlrI^F)4hxGcVfRZ(Cv~a1t__L0A6I1buppo2@j--BFDue@`8}Jd;%awS;a_L5<%1;
z_2^;E1dQKCP^i$y@1$<Dt82r>aF559j87Bd>JuEnvP$rnh}X&zBvKZtdk;#-X=SiW
zhXL>iLG88{?h?+9s6LZ1J6unFJu)C20TGQ2h0=Yz`r!roW@+cK{R5MMxn#3%4fQ|F
zwHKY^2E^OD+CX=&FC%CLfRR?y7=cze4EgI!zh;T-LN4lx!yDgF8#95dY;AzfRpaD1
zT;%)q7D|6{@gmxw>+$5SCC44^VZT4m45gdGku3`_rlG5kyl=uJmMZ4Bdwt1r#z2|6
z%{9-lvK!G^lF4yKcpD}gLp=}t0}G4E?O`RB>!or;ReW^V6elkiLM}@bLMNa|*XE@|
zF%0U4G(sstCL$NL%STk@>4&f93P~<w@s^UmtBp6vvONlMq2TaRuC+_Kt|*oV9}t>H
zZMTPbDU;X{9Z*DjK!qud$M6`JOcm68)GZ0V{NlT+G~?-0y+kZQKs6cONw3ZlTRFL%
z<<R^Cnl#F?&~}2ud@W+3Z7?E|@a}N98Ie@(0bsq2rc9!x;#sT^i<NJywKRU<I{#Lg
zAHkhEFM%n^1&K^X@}UGlyy1OU@`xkMGV&e;ktbs5L9#j4Q)kk+AE~i%f5Wnu`3+Wr
zmL0E1T&D-Dijr16b*ALzQQZc7&*XkznrwgZ4LYk_#CRbXEg8w%D{2Qdo2AOfXTG~i
z8X=<BfqX6oyoeK*hb!$TG_u%{BkEB4hFdM<q)$XcG_4|wqB(txk)jAQ{DD0eS(Zp1
zkRy8-GvC!l6@rhZVd%J)Ws?;9`W)uNroY3NO(25nKfewveu&S1v1uvYssH%ue9nCN
z&wswoxwy!m|LFY{^OODk%~n4%chToR9$LxVEPwxF@s0odQmn=Qxy$>AZmX#gy7sDY
zW9g}5mkEji5MjUpO*Us9AopEdGl>-56on8#wEKY_7bGsJ-3zfQ1#xCX><u(aTw#E4
znFs~Mwhoww?Xne)370F4dIyhq6-vt*YO(d*X>yI_9JF)W>c3>Md4Vp1oAx)c#%Q<t
z34g&-ob&g~z|i(ea@--Vx~C1yL-TpV#+3bM`_fYUe;6Io--B!6fxE#!A6zcoetxC9
zq?BG3wh&YgR;8P^@O7~v)CF;$X3|kk3ILhYQ(v)t(^Pi5Pr{QV19ld+q~Co<S@OD@
zSlD+vJU#Dw8-L5e{y3@tqRAQ~>e!(95c76KUZjx+$bc^E?<e-HrL6|9c0ml~swGMs
z!7P9i*VB!Ts%2XpQsD;NO+=*}_6?im@f7QgX~4#^miz#3wGA*1@&fOT5*a->N>Ctg
zKtDkra(P$E&<XXT)^2D;+U2zK&TF%|l2A)h6;F+F$40Q4nP48kpNJo#Yu?MFZG==Y
zYXj69aP-^?ek~i(jEMnFcN?+RP@74R7@B_3U2W}08}B9X@dhqoStecaRqv1WOU8T!
zPCt6kdEQMX<HnaKAFU)7!jG%jbb+<)G2trm;86cv61n<r6Q}KP1hP7Ha~@LybM@5k
zXjtL0iJA=|;|hb8A+)c`C&iPX)hPj*ks5mxE>_j|!}rwfF*GzRA$A#Y7-Vqi$IUip
zjmH6QEdfVQcYu)yf;5pweDSu8Y^eO(mNh@!=L^h(^%1L8_=?f+jOJlL;_~OHN2c;T
zH9iY?zy*l~p$gIP230v@kY-A#Q6Dc=ghYx5*Q<pfvOqockgF&2EFPEyCHNl8i@IPC
z6pMLn`K7k@?KjuEX}@05PAF!q?z)-Ly|-j}1$AC-?N)mDiOu9i`r+E%-t-#+r)BR*
zv|KKEStHUIWzyrkd&_1Lj3viO8()jJWU2Xrh$%Ov&tV@VH>m!?pKvyLGsxSD##2bu
zIViJGFx&${m|^2Svo*YOu6WmlD4N+YA2eoso|>fSClmlo@*a~Us{Y@=;{zch1+|%-
z=TdkQQIs-qP}s@E+a6pl!|_4dJd<WL23t*mrl&HXuhQ6C8^@%p(u)BLK9?{DvlmVj
zLgo@ktG|gqKUQ+T>3VjwccU0Q0fhsX#F!dL^lm=jr5s-A?PmJxbhM<W!-?WwSIo^m
zd5E2GIOE0(L8L3oI$@3J1>8CN?MStP_%4D`t*0O@Cdd44t(CKHpw8(4=IOcroqc$z
zIk1u+sWn9Tp0F4VO<<<<fShVCE+|LhIf@ND#oH<Xg9U2@x_7MbBLzdan%+STB7(kx
zl<HYZOsEFdLoG%X0Fgv#Z&O0`^!+c(sj%4IW(%qZlbMTtr%BHAp@YgF&A5KjRoNi8
zg2jNhTgmMc+p<5?5BKc-@SWd}4iB<rGj4oBl)nBY3gEj?B}Z!6p^y5ZYXcsID>#;7
zdkp_v<?rub;4)Y5qMFqTFfOF_xEe3jJhPZuR{q3YIby}=mS$V`lM6Wj_Ab3P@@Eew
zDigS{I`erT`&*(jSO^;x2AM37oQ#ej6;o7LtrYwCh-$qbP}!a%NnY0se%*193S~!{
z^{>BG=b(oZbs3@Qq<lsAnI56FcrtH)E?={qb!4wih}l<9sNDZ%ujAxjeqOJd0BRDc
zy{y@idh*QfVdhAT6iC6*!~rK*o1UB)^`D8zhq8;Gz%bfOL6!>uH1+^3c8S!Mf(IhK
zF>l*BqQuiSrQ#2M^-TEy)#WYYF!FFGvTV+pxZ_QGrIxjNntmE=6-VWxVmt+9Ve6dk
z(N&U@KNFzZHH3x^J5jr-9uI=&p=BFrS$nDWLTRsavx^RfTG+Pdt7_FUJiU`-PDg!l
zeb#q$yW#ZrJx@hqn%y0~j6Ak6YHhKIw+m9tLtl^vfiHk$Z<T$PK0J32O(c>GmF>tK
z40ZWad$2so!$RQa=hsC6{psJ`mEj5=F0%h!dT|LEZoUd)e572YGE?aO1ddOHwjGhD
zadcD?#i`&UjY5cR6CZ#CHu!nlP(y*b%V}LSTdpz0;D+(b*_hY1TKXY2b$7sEQq|B=
zQ(H%$+M_b^cXXtc2#99AzXCr<3eW_rK|}#jK+$Dwl^`AOe)j2YZq4e8q9!{urHkv&
z@Ki&^R^s3%=yZ$Q%3M)u!|2zt9ujiN#UhGlc@2-=p&5}3w8%VQoVJy~Yecj=0Q7eg
zgig?`^=>H{AA69a!T}d0;~*4L)Ik9PRZx+X=K@M4PXL#l?K^ujs6L>^9l6mwhJ6mR
zhCs1QXtynj1<`@D>JB0{Bj|Qr2|nUjM{PuuR|jc*DP1B}2dxd2B?vUf2}Z!(QvT%K
z>YD@7GCr^9f?Kb3`EWn#yJQ#+P;E>_8MA&NRFFu-Q=H4r<JS%YgNf3zD;Fs0oU^Am
z2<=4Ivz9CGa7jHyuZfANBlSCr!kk8`kxr=r&yN8@lp8|a0mvUTy&07#2nnce0C8F$
zr_U48n?m@EY$P`tY62vK62b1z?1f{}4>wWP&A>oT{&NY(G2BG@m@C12=uAlgth@9u
z8p1&`e6|rs0Xbu+E7(M-rRER!8UIK%b)h+K`v9vcg7_qwE)7^!%dgA>A%$K?F3W-<
z7HB{2V^3D#keI}A)!Tv|6{RI<Q0#ruhbK;O@*VP2&Gv~_DxQh8n_gxNbesB;w14vY
zNkIiYQCDPW<gvWOAjgnEvk(n)qrPvaH1g*62QHBf6R7F%9ggMgj#O*24s}=0W=j&&
z;`q4V6z7K9-9ar(LQB};smaJ?laW8PROC`ttUdds&2t1LU)SPEOr&vC>J!CVTsGE&
zGTDCyz|Ah!@jw8PDFBo@fl(!nIAT^@xdG_7!eZ3Z@+I~VTiGR?bt~*OBOCXHNw0<N
zZPZbE9sA6OM+Utiwe|;4K*y!YgM1ffb~jy@_}g_hCkbemp=+adSft_SUpqmJI}K&!
zy^P$Z$#`EL8RGv&pnoDmvx#G8s8^5VA-dGcG@N_xO&cgn2E<wH<s_i06URPYZR#e0
zosscmvY)z}_@nV^`9t^}$`~n=5)V)&suc-}!@2q%JiwD=Zf;T=hu=|xQ|+%hMSGyh
z3doe&yWKeV3WfT%J#V*8bxSP-jyaH_O9?Vb|FHuoHLjL`*A=i+H&3EZ-!B$kc@gl`
zGp@9O&mrFghlKs*!mtv#3bduuX6SbMMzrk3+gnpg)wWvA*df*PD>#KN$la*id<|`<
z^2uu-fOBrQW%VVA9Bc1BY>MDg@tz1@em~?C`DxAfuV(lqnV!h2(6Vb3d**1+$8#^K
z-npdyX4M-(c^{q<*nq{&y6Yl&(Nj-7{1l-eWjb)jc=!45$7bdXa%+063QwuK9N6=0
zI%w1&cStufcvv=}lIM=I%KB+}n=Fx&bNsgJc71qY<l|?%(eJkDiH(Wb^COq9j{KFc
zYbD=f>e~1L$e#SQJk`0^(SKemk1*@~?Uo?lHDFXFIC}wO+{3u<a+-=J!6@+WM-_IC
zn-Srw5Gy!am$^J9y?JM|dX%>kN7xu*g~^#AF58Bm0s-i!H%0s~chQ5i4%plBu{JAV
z)_;BkxrNbAyMep+h+Q5xU2_a=f0epX?7@m`ce(YadtMAeE247WzpsF7*nB`5Og%BT
zW3vyXc+}f(x|<^$n|*y6*I3AyKxTv9UH(dQuu@CDS+AwY80|$9dg`XOHYR99xQA(G
ze{o-Kt{n}%*Y@S~R*WYG0Qvo-kopLjc1;?WN+V*F%6%bRAzy|b9`l=y=m%-$@ewE7
z_0J4YD<1GafL7EpL)tC_8Ihs8y83#3lZ5KEGb(m+H;T!0_ND8N*qq;&Cg;mzqu+7-
zXq7jwA;(ewVXLWlkKaojR^Kdf5Y*{a=ev-r+#@9OxUDAASw%Z>g>t@I;PJZ$KLqkb
zhAi!B@AbVMmzXaR94MhRytUph<JJRnxl-PxUcxIM84y}ZA@U;jw-)vRxf4(qfw)N+
zJ<d%b1v9K79A^*^_me)wgp2WF`5t%b5nz-QXLqo*0>MMnBA`M6{x4FykyR}#A~2b*
zyi46@dkKsrdf?Oy(C%YMQ92Mjy{K~tmLSdB6aa`tKR_TjxmBBRN52WPK?|2C@E(Yi
zLd4~Un#YTTK$Hg}(Y}lFA;qz|QgUV|DGf|&^A+cqBBNltcMjAsaCN<itAU1#W1=fD
z3{OCVy_ujeqP*Xf9xo3P-a*3Y2hgacSA;848Rl0ufx<r54}*)tmU$i1WF}v*zej3k
zijf2bL;xn28;>&5Vrtpcs#=16S@YDLn2>Zq$7hCN`kodjP{#G!WPV(j3R_^q8#40f
z!(g$G2s>ZN&Q~4nAr&Vnq0;&wN@_qXTcMmd1Nz<D!Au;)3lf6}Gy%5SQC6q!iIjV3
zH3|4%$v4$SR4}Pe`28yGCjQ(u<UOS2Jm8-N4D<wvWZGN+8dJkKxO6A$h|Pr&n_7du
z@?{|jTRQv9itAN<jUSC~IJ!%cZI8H>&V@0)?h^Z++I08srbi~z^0itVe!TjjwOdvh
z@vbtG-dG{GG4W&~pLWWMnnOz6za72LUb^3~Xdz#GPxTfyc#r?NteH}%q;UagL_#GH
z>WnAw2GxC#DXBKh<7}VGar$Y<L|cDgbazd-AyqQ%iGSw9WL$Mn2yZe1^Ak$Y3%PIm
z5aAmO9?A)+NXc*Ct%;2vMYq5g`9SWGslbYKk?AV1A9UJR3qG{SHukrReTNK=b)&fz
z($mP+jEq5(*H>=82V2R<=0M=!#b=-$`yzi3?aC%Psq(q3ma40uGS^HdR-Q?Mia}x^
zS+KQM!&e=ttS>mXQ4)0&nll+BJUQo*Sdc;27dmvOH=^kU@z#_~lXa7Uue=ZCz~%?w
z6e5>11lp^A@=COUUN?lIo+_U(?)JySa!l9frw^%ezZN_W{-MOAU6J$hjYNRwuj$6_
zdVMO3NIy5Nh<57N>`!qzPCBn5aWKWkWU938N$>2-0LPHlQ+bK;zZ(Vd@XcL%X?l|V
z?S4b)4M*jr3x#e4xdp3F25)23Cb+9FbK;fx^f0^bQfx=lqfbdI-2Qp=@7+#K!qE?s
z*jeu<)>ydT(05In--^fxWJkl0ED=2lh&k4fNlPD@0G^OE&Skd0^@<9q@&=}VT`9KR
z8C1@Id|fm+ZcdCf;_#I%Jj0=Th5q;1E=fucTM*5ug<|AHi5{9VmJ(w%D#<AER2^nP
z!+M3L9GR4Nm4W0Hy3=&wnd3w_I##a`hBtT2<eY;=@>B!DwoqL=NcQfhc4tM*&Vf~+
z2S#^e=#j&Y5SSU)lMd!mQ5L9Fk3qrrzpo2z&@ejYU0qCR5>;W^`6av$Um+>t0IArV
zZd%^cp_~-l<5Mfdj3=@n%pk3{aANg=?%*}ZG4A336;N@4%ve5WfP{9qj{WbSekihw
znXb<ulT~20i;%p~n8;z&as`+iQthn&0(Zz7q#CjFB{jcL)d!c&GxJ0Szk>IwAU=>r
zm4wP*Dl*0*Z1i=SX357($EZfcuZ=wT0Z>jrvVv+t7)U0CW9jm!#(!!O2!T*XplgI;
z9o|`-_Qb7=;2yEQyOjD`#DUKaT%W|hoDtd__%hcP*Y}jvH{MH7iA$h@wG98?_|d+_
zGk(oV$Ao#tZKN0puhVZh`>v>8V&f%Hi94a%_jc;GRnfew%7i0NWZalVeJ*>F^bg2v
zR4wdCp1g34BpJ{dU2t+otQDld&Ywxe4gYV?h<<|hd#a>@_SnleS%`e=`bE=+&#6ZO
zAG3Vt5O($>6hE6-WR4v^)9s&snsr}wiJeps)KC&=iFXZEV*QCplM-8o*4VlE;t&Mc
zLUcMpYMpP6Pz$STgxY<)56yfs8B|uFWsxFVhGrNW>Qqe~PfJ+2M9*bG(BVY1RT6L%
zL_M@%rwSR#mRKTS1A7$Z=$h56)`upJh%E^x#>icmGGr=KU^{Z$?AcfyiKvUfF9@^^
zLjm^aZ3;G10o+?uEU2eykAG#>L}<!NBg`^OK<ZbMNx(%U+@*6CPup*?hp6y1QfrqI
zeWHpYI}NkViQSms*Sq>`?K#~5?$*VcU#|Q1ufLj}nh+0it8;#JA@<MgptCc#emP+M
zp);~Y4@rP;QXH>lw7MqM@b{BOPe<K&A#h8>QBcRWGxt-kN`>!muWq5W`J=YZhs>fr
z^=fB(d1rfn@yni)msaxHyv>Jd<w|e;Xhj;f^j6HJRcd##ncp4)DHa-8)%=;JkVB)h
zp1<8CAB(`O_s1-IV_%Cj$IrH*b?JF2Zz^h2(G!nKF%GF*Q(Ktd598K0O4rSssRG|-
zM@!FS7%sPSwOckbd$TxnIkR&({P(PKh-!3?M`ntw%PXcg)ZR<1G{XWJ@8LIzMa*ZN
z?q{cG7%4t)+JT60DrXR5DxO@KSYRjS80p1~C!Bz(#cBu8QJ{O$uTOAJ!U>Pu|NcQJ
z?2R)Q%5>*_Z2yWNnkAZ?(OMJKGn~5f2kt@(szX#)9@(jrE6SLTb>6=ZOmW7vqf@h-
zz{Qud#Ti`HCHu~*_Fs~W*@2YgOuJLw694+$M(9rCOf_`&m3Q{Vj;5Q1=$KzXt#oz8
zl$ewO`*u$Kn1XE#wxh3Aol22<9Ut4u$P$?hb@t^>Tp{}_j@7-}LcX`V+O_`6Gf=M8
zSFNM=Eq`uO1@{2hCj0xCu06|!st@`+mOKFAd+EY6OwOj-o=h)Fd#biSsVOwl_8rLw
z_7Xo{W+ye>e*83~EBaKvWY4;~c`bP;YN?k#{yH;#sZCw+r^%2Y4Wken>PpC;yN%%5
zd1f3<SPqPtIo>oDxZ68cV*xmlT8<<*&>qOhBD4YqkL-RtQdv$!L%OW9%%QsGU!f6i
zJ}{qK2qdFgC%1kph$jk!%{b4Gh7}<^DqE<i>L&JDzCjc#ewydqwq20}8LvULXODqN
zTejr+XBqLWw_~&KZPGN&bU|BjRM{fAq&~2uURO2CYjH`1yGrXvY$QweDQ+lKynDKE
zhCgnOHufoOY~{FBY?bMMPJDVeYiX|NSSS(4%tYqtTiQQ$w#U{ZV##ORD38tC`dy4L
zrT%Ko33E&NTek=)KdA`c;ymg4z;%dwsY)9cZds;C3uVYNS63oxkaO-`>|-fd7H=fk
zdq4IcwR!pTpX~EJ;!;Nd9Wm31lu&>Ki)j-A+(-35)t=_%X}GvR_0{cK&`CE@jT}|*
zi70hRQtB`ICGy(7(?9^0Tjk<T5Tdp?E;WanU4(8@CXxY00lWXho|UhF-sP4`2KO8>
zKnYBFYTEi^Pp<UU?Hb6&T^XUBIo??dyX}Zp+qV%i#@t)|Jb;eY+qWr?f);rUeSdm;
z{&lKjC-Y-Bo>H%!c04HfM*}?qlcFni63)B*dkvy)#p5xvz9XDyPgkOL#S#uP?*Oa2
z@_7|<d5L1WWpzEZwuF@-FFfLHyeZ%P_jO_JM&Apok8MRxSgRf&7;5^}T+|O*9Bb*m
z_~*uM^EK_2Y+Rx|D&MGhL~yRXgh$m?E?R6&I+ZochPr`gjOE#=qAu%z_RQngRgzC%
zj1XEx#2ko_nFuwmz4b&M%%APrqw1`0%ztw_9xTPR(wvsZ&_>%zj5lKF)5J~6X8HGo
zZhqT#FDnDfpT6fRm^Rw&s%mO#^V98&UzAJgLk5KX2U8Pdq)JZU7wbWE<bOX2?1;qd
z-dOekakwp|^`)?}K&gDR$p-BMdQFpBBVb;qa}veYM-wTj&N6O`%0t#NY!2P|Q565{
zs5?XNfJ7vPIF><{n?B+{vQx#sQyqiUv3gLww#_Y8az^=~Vm#pJ9|j&MMCBxFVs9i|
z-G9b(A+)kSd}aOoxVza>zf&QAUmUe<^6n>8#|_YB4b(2Mc%LHszc0QU+a9j^CDZb}
zi7R?H{+*DQz0C5toxiQ#Z)M|>k+KN0i$BL&7pd9Ihd-8V1fZyoYe)cCu4r#exbl(?
zxLdsf3yplDJcu)k<VyXwAhl1J$y9`$%|+rek1}F!ndD$m&2|rtIOHDDm>D)QT|4q(
zkWL8&^tU6#HW~(nXkGu)biV2>X0g_GW<ME<vncHZ<=zb?aS;*ni0;i~$<s~D#vO_5
zd?HdSZ@jfvY9B})?qgJaHQ7+I7%Bb8|3~u&+*4*nX8<Pq*lK6DRh_B95nMJjI{XyW
zT(wtEEg7Y4`%xXgwe01e$~HkIlVWqVptx@JNV;PPSKyM~{cK4lf;zPb8lAZ>yoX1U
z@+Jbmo#W-!;;31{3>k?*gj|Upba=YAGicV^e3Kn8(63ZNqy-Y@8UiT-B-xxp1*cP1
zmPuS^y2^>_n~Q+U1m2Mioa@>BnrMc^TRSZ=eC(dd^c=&1u)Sq=wKmwNlpr#DW$4>c
z35*Ig>9YDqn#vW<=*r=q`|I9$@MsNhim6Sg_KPpaB#sA6v@bR9P{JMHj#+P0cw`<!
zn-1H#YV9YDEo?zO)|63HZERC3zO;S&c)>M!3Q?BubcxVj06G<+6>6-Q9NV_TRO-ha
z%~q$V;t0yn=MUsfkDlR@KTbjI<^&O6o1}XTkt@E7{cFW4q@pU3(ziauo^4dls|a-r
z2|^@JPO<sTF=W?N_GYZ;Dz-h>1a+=j>0JFB0@e$Kn*Q4VaX)L8$o7LaHswN(MI973
zR)k6fGygYxFe_@%+x*d0<QYq%#BMkY$ml*aOOYuZaGreN)!PzswpH3=?J1lR#`UjO
zy23%f2*AER@(2>JqA|IpG*iZhhulfry&AVikU0@5ztXlaJc(iKQ>NC0{PeJYkK-QM
zanL6+%FC<NPO1sy$g+ilXu`VN75ZmWvez{Nf|n|0GiNn`7$1%)8UFXZHVAAcTy9`Q
z@L-b>N(oh+V}{wjw!rCkHHE|kR#4CIfOakk8IuZxO2WVui_qSW4AXD@Jb6VmInh86
z_oA{-18bzfjTao$j3sa<f=eY3VyPw@sp%@OSj~hb#>h*&2#%PY0|~2xuJ9-2co6w7
zNKqcqnvrODYASoWv(L?-?~|K9vl=g{_s9JwRcv3ojiCzsRbAIA)V+Q$;w~+^KNfvg
z!NyVd7)>mCqNKx9LE6ld*R14QH#Z*|r$zsO4_<n;8K{%GeVE1L`c>RY#1)G#<3=mF
zuK7LzTd&ZD;6$}%Mii#5GNbYbN(tkyqv1o5KHEl_GI?glUDv3t+|)LA5la3Y3wF3-
z+&2h3J~77s4SDNUCnDdfTz0h|r8gh@{>0hnXS{aR7P!!o;iZCx4`dqdGg9c(7}s-w
z;#!+{I?oIsdKKzwzdmZCPXa=ZReeWoMc}u0kRu9k*Y0p55NYww>sO%wLP=VeBy;C3
zAIB5}BgcK3aQ}#nF*ha{)Yn#SYwJ)G>h%3kke&(87CuJpftwQmR!u$f?-?M)XU&@n
z2c~K{Kh+)Gnn$ZHX1C%XU3aWP)Ff&2PTBzs#Myh;s{W0T&Tqj1GERh|+4deqI;|)?
z7bRG~Br^Jee&}kO9ji)TW>GpP*V#T)B5ZTl?5jp0J`Q#gg<P<)4CIkv`Z~C@VUH2#
zQQ2bVWsx*6d$OfEXE?P(J<%O@FGfm?p#-P>7oJE&eFR_|^Vd8IMSj_X&i0uGfSkio
z%AOk_Rch@=`>8=`f|K{*fXosNRyF*p|1T(#L<aP;fXC^+(H{ck6Gr2+n=f_E{Z%oL
zj&HIto|(b)20@)ek=dgeB3x+7@dYFS>7qb&7Ht53L&Eb2u%UvhgIYjP?IFKFAzX!@
z^h<*7Hmk5ZIfIHi@|8fFwB%|S?9I8!$ttZcv&kgRuMB1M%d~;BxEaCXGc2_sn4l@A
zhx{slvY3|IH2u|P(sXO}SSptaiK|Pyn9`<Szc>bM%K_FUB~my|xLL$a&eV|{W?9>;
zB|0{*aDhCuoyoor0Cd*GbxI>cA?X1XJ{9*J=Pn7ZsNojWaTb2%zW%3fL*+0Wk^mh*
zo<UaT-XYUXmk{u|3phWYlZrhKqCyT&H#5&k`^X<=U)_w}cJ!-J$g_5k>%Ll6E1Khq
z?Sm2IY=RcaJ*#FkR@LmKFnxo+EFY@VIGc&H=r>ZX-kN&z)5|y9tCK<ETN}BNY!wzJ
zh>LKwvkLjLCL>3BwACvECfTCk@0GQAfyn{mx%`?Ki^xr_+7GD^`P1tl0N^yKxLO;E
z;JZvg(?}-t1T&hrA&u~&su!n`6&-JVrt8Vj<JJog&G}na=zNcR#PViXAoo@B+KCRa
zbt6rNB+6%csstcH$j=I1g%`Q9K#I=s&^+TDNk-DGl!6ne>k4hkT8dnH_?J86w1o_a
z0mi@`l$x&qwCcg;Qoyj(&mISc#73Uf@<(5w`o066OoVnl3?|anA17J@;ixq7koxDd
z(9*e6)6{vUxh-nfo>rdDCRDbdRalx+y-?ak^vd+g$e&CRxLc6!vECm;^OmL(nD|>)
z9bd0nOqRNLDf%!ZZFXsv)=$;`Ko!?HfmFJbR!nI`-IzXESeA6_!SLqJ+}N6cMYt<q
zdHoXAxTLwl^uu#X<pPzi+5P~zXML#h+5w?;&=FtrN9%YLm%2mNd&kv13F@x(>7LY0
zQVfHlMm?(%fm-=|7;;iWn|QR@RU3u?H(1vdxw_JXloDj?4{XEmOoY#&peoIz=FbRD
zMO3RZ>O&cHq{=ee0Gj(UlXPb=bcuwrqn0Xa<B7n%=a1Yp#|$4o+h#2S(WcrUKsUfT
zt6Jn>=n8eGSbN8i8%MTeya^j4M&`0+8mz)r=7eO^7m%Ua^!KCzi0{>Pb>0U~ZUn$0
z){K0BNJ61S>_mUTsrA9L-AC<^QgiaBS~eyH05+9BhnFxmzx-$skn!Wo&gSGvIcfuD
zWSe*fLyXwjkU(=?OjC8KGc+7!hdilCp9=`5@dWm~M$=p#fYY=Y`WwCp0M}DtxuNyP
z*K7M_I3t?L<><nWPB6p8rUmts!@+30RbE|}Ih#U#9W)Od4%7)0arzF6Htn4dk@J~G
zPoD_3)Q~K-BJp&lb-5}KCZW(#Hltvre*XJ?H~=@Z;lj3as&&XYfcVS|XXN6t6uD%(
zaj3jh9m<s}?~mAw71#S?duSh=Dz1~+teZ6F1B9g7zzBSN)zBGax{CcKBHt3Ipyrg^
z5O!PO4BMW^&KGjejO{rl<ul5^<0iI`;ROFp_W+*ADkGm``A<&K9<uFyn|@fJP{KDN
zZS;|Z*1?fKB!X9sT%HclQW^FilGxd7uuVmrbSn+D;b|mW`TfQLl5_wWs{*NOl5C>s
zoRdv6`qVjyaH~2V-`(v-UbBEVODD3BP!b{Qq~-;zF;iUPB831;J!S^+VTe6TG;WxV
zc!EmJry$z?8~(Qj(`#5Q0p}ZO50N7plv-tNtI$y)eg+B7b-2YTotFmwr62Ag;zmLV
zrt=vDW@p+h{@9a0D@c-)kLYEBWR~JcI>ciz%@|b+=oLkZ&)=eS4PHV;$nH1pdG<+J
zE<{Y2`9?5`257sa0@PsS%17?W!g)D=py@_Ii1M1*In1=`2*{r`yn+dt^XLK;p(m`5
zZZ~zvsfFZ-h{9A#P`gK3YFdfYs`w|e#XRoABP<r}IPHbD??qI=kzJq<cpw)@dta0f
zY&XTZ<%}x9X=z>`o+OeAh>(Zw$at+e_}bo)bedk9!@e~R5H=IE1EhM-GKV}GvHTCK
zlgU;P)rXyeq2Fdy8I2hXcJKt`rtE$u4hK;drV@32y$o2EKDW-iGrT}7VcAxI)SC0F
zr7PXhY2Hh5l|WixV7$~6`iO=qA-cD|XH(A8MJ!>!=8vg%Q{9-#%rb;frKH$gD5yY*
zVj}_o$$&lPnn7;4L2iGsLtt$*cAqe3;)>X)HiHjMkM^zAM>=iMb(pRzQSp(>7tml|
z%<R|;9{#Yh5!E$o*T}9tG=V$?Tzdm2>1tR_`VnofR1BmusYa8ygp}PAP}9j$zx@2U
zs2mM6z;|QjUwo;8ozVpm*ms@{7tg2r$=%Mmw$O;sGxfL%G>_#ZA%Tc*O3av6LLV4`
z<V^cuU2opo*159}?b3N=e8|?2uWvu>JL1PHyFBZ7WPGs17ix7HKNX;~<ROnE4=L2z
zK|G|Jr&jr}Xu5;CsgI+J(KP5V!Q%?$W;$@NKIII!WE;98a@T^h$IZDwI*}pI9L#_N
z?$`;vd3;rJ_LUx=afMIZZ*OH~p7*I)u$A-SIRI)6Iw-h+azk__Ya%a|D)<b|z+EI*
zg;ALTkola8W;uIgFtttZCB@B0nKfpcWL=;pcj}xO_<1@<qe>LCr$ut7EA|_zU-({l
z^i*NlpyBtb6=fMxrS?6#k$%HI5m%<=d_L~CiEMi+Qmg<EpY4Uz&$d868@SW`+oQ)f
z&QF|5d~&VLPrEJE;}SZjVCJ(2VJubw(FyqT{ZommMLP)HDozY)FNtj-)egbS?#*Ps
zY5{=xDF}#hsza`>z=uW1)Wc**{URDsj?_Kv6VG2330#zh?ok^rwq+dc3OA!r2i~rb
zfX3}-q}*J|NR?bM$eSp>uNoCHc{(!4i-IVneRIot@H4Vm=#+3L`7<Wa*%bl7Ki3aQ
zUL2yp(hmHC7t+OajTuS>Nnuf>s;<q7GQRYb(?}%lLrV5G9I!Irq%6KE6I>fhF#3xS
z3pP>*lOSB-*rORz>(hUS1d{o0Fdt$((Dy94nooH-4Ur<lHnTteTyk!p8ozlGE<`jU
z@Q%m+cTb~KJ-;=7;3eW7!#(g64K|`p^6$EYdovg{-&DNbuhqb&;mUD27mPSclMJ5e
z5|Tt#zL@ECVa{B@v?kJwPq$n5Xjy%VO|9GColREMQkf<8Q}u5IDXaIG_5)?1v{%3I
z?qzfD6?0LWk=8%iKxzFpr1ewid~92#&bumLWmc46BPJ=O<_8}N$)M&DX^Id!tw>-k
z0D&UeV(f#@SO)k2AE7L!{HhAMm5F~&t(s;5g6&!&r!bYXB((xlBOI%fjDI|yr1OmV
z-U;$MF=l~Bfx>DtWkL2d;tP%pZWLn3q$9Ivg%kd67(<1^XNBRdNu15{5b^zlE1=!z
zPLPEyY|t90>-(9Y^`Wa_>M~|I6~d~CgOEl<#RB+%l5rJLoIavkM{$?bfw?tVigVjH
zlYHBQt{KWB31cI`qbrOgGmq(Q1Ews13o$$HvP;yB-D=&QR!H(Sa?u3cLiOxCeM;)V
z!(qhouvZ&BWl9x9K?WF6y8zZjIILip`wM*n20ptxkp(|ggm4fim89+1d)&k&23WKQ
z0I2F;qvf}O6pGTpcz^^d7f|gmx+g^>69e~|smZ~?VA8*!E<6EhfF44)7u{$`zEb&J
zDEX?NgEqOM-k4d#NX;}>>cL~UD~TW+aPa9GO9sk*7Ro*{N&!os3&Z?eA)k|^g(chD
z4nP&fCM>-ER62w5$PK?j6Y}UIXpI=!RNm1iq2v`?S`_wbw?$F<5|J$SugRye7hkqC
z?|nR<$$*e;BO!>9U4&{9OKeIseaR)zHSeHc-2uy@^(3w$b7V3iM1*ulBC)$E^Ab0Q
zd72XL-IbPjaoR&x*~X7{Q!RUNN=*D8w~S43J&L3Cz-$C{#yfM%d-Z{y%Zf9)eg%S1
z8{o`ndgL58KpC<xGGrj<{Ut90(V2M7o*sDHkR#@sg84*?Vj`HvIqrOc|J>Dn?Q^DD
zy5;$*9%<oGJY9jVp*W;ge>A9uuOok|Q;n<0AWTm^q2dh+^0sD6JcH8?&SVmk9H%Bj
z$=!#hMG`g7S6QxFJx7JNXr|thQ_<*c+EUt=$^;vv%cdoiM<)Uu-*}mvsK7)^m%{M9
zz1Q+B%@08*UA%NI=0fLrpNRSHAf6}J7AZszjiD)1@9$M=+Gxtja_BH|y7{;dk0IG2
zRpDL5EpF$TUE6d>m<9n{Zmx;4Aj}}Fd6*6v?x<&PM<^vj!lnBg`xoAqQiM&5_pvDE
z7dsrQ8FH6yuj$5uu{>(`zI`XAEsOU^#_#O{L|KkNTpWh`SJc}P(te?Pqy8JaQZpHF
z<LF+9?4l}|T8PJElv;^eo|nmhVnBH}H`<K5>c75z<X*xZZ)286l0Ks`Lu_6zVup}+
zesSpKPpeegtmS_*JNi-|+NHH`sP3jRI<kxK3{>oC-wHn#!3%5pOLo&iA*5_`ZOmX5
z^K)6EIY7q4vSp{mdLk&9#44~XJDza@<kxAL3e$#O>%#EMz1CP)mN$PDzg(f|R3W@N
zWyK0#sG$@O;i9?&CT95C5nNvdVIXpBu5oSJhiM|qhJatQ632ybh?Byx*9i0Kw&lbJ
zqWt4U3Nh-*fs^&g36`~PW*%~1v5ff!*Z)!5{Vxl7?>0xqbg=1Ws34*iRS<SdIWHgL
zkx}k9QT*Ce_YgrBgdnj#5wY<Fh&kQjNp+?Kt~E$+SkjMWSTumW9FyOfdn-=TNMc#E
z<OE0k5|$OdpwQQLYpwaI`ORM9Fe`2%m0EI~VmBDIZ#-EJIc=(j=t4?xqww!-EG%O;
zwZ0lV&CpZ-EU%&eJlYK;r#EY1I%&a;N2XL47cXAm7*_=M7c7pacW+jo{(;4Bqf1!H
z(z$P0V&XbCj2xEw^7;_{wtsy6M}o#&ivJ0*T|-S2+0rw)0RLFN|9&~z^9B8(ieKa7
zD<oL*Tleh~zmhui^^HxN)*Q0^f|d1`7u)Ztc9ip2zVL}yBJFW%(~WyievdhJipy_x
znbnuSZ}#CUc;)(P!;>3-ZusWr565pla{caP>ULZG>Vqx@9j{Kd>EvCTNs&0Mx^Osb
z^jY^aC&^eziv>}LY+0^<{Tqoc|Ni>%hyOZbgU0gw{k7?f-$_CF`Rn@EBPp#c5oO4!
zwn#}yHJd|%Eep6n{=htGe?%_5(>m3PjjaWMHbe7K5ZSXx6VbI5+ESyy*xBaisoPmD
zN73;249)l7^ZLC$M+&!6+vS)<(f}-S?kG3pX)*m$-(sjwl>NGISl(-SKOA66Q$N3n
z3WFp6s?hxI+kcg0@@)Dq2jIFQNq0Yg9s2S>7rq)TzT{WY&|sgL7^<|Tj~c8GFFkH(
zk>{iH_Jd1EX6UIktDgr(0<l*@b!e7pSx3yb$1yS6@tJEH2%`e6XRc;~C>&2gEW-EW
zl^oTO3qD3XZnXHdx;`k^%#KC%UO>QlWG1Dbf0dQB)XY!%NNGndoncYf0b3eRT(m|x
zE3UK0sYF*8<<(>2OeGeP<vCkdYW}%@8SwU~^ndxW{pSb&2YdQIi}Rnw`5YqfpZ_<}
z@PB^i+{iWCPrpO)A?k8gboCP6w!D4&w$gD3FuND|eyC<MXRe@CV={)j$mmz#?OQxK
z4=6`w!EJZ4e8SDmEq1o06^6PO!HP_uCL#LC%wSBmA2Hic;apKOZ!~1+6%6}L4HF0+
zL$P~qhv^qA&wC8OA{z#RnW<JsrnAKTO;1^epE>kR-D3Rg_9I_4@W{P~ZJ3j2#{2KI
zw6wS<LC_!l5G^3G9u~?|AcT-iP;(`?e}G#Hi$N`WJ}U*O3O?%O1Hm_R3_-&7oxxWu
zmn(YF`uiTX;fy{A+Kz9ceI~kV^7SVC_i_b6!-e(KcQ35RrBof*ZGpdkujQknqoWf#
zz>7bym{bE1C%2J%Pcn~_-GjK@VEIYhJEAJUh^6SA!W_2z<QzJquaG`(*|u%lmN8n5
z*z+Dt0seJs1t5z0lYh0d#p!Zv`|?XH2RbUY#3XmG%?B;;Yq|`tSKngZ6_*u=0a$3V
z9Jk8zFoTiC25Wk7m&Ho>Dr>qwkySu&;Yv=?AFy0rc!FtlTtEZJ=PLoq>GyJ~XU?pp
z(#yxkC#`M(sGJ#$Iq9FUv(7S@egW@;YfzDHfKsBVPb>k`E9P|%A3p5PB`z{xbI^_+
zYcvP*)_ouBKe$QS!oor>RN}zm%Lfd04+aMZQ^~JkN^gc|EAzq&PDMSq77pzvHJ^c;
z%7bY%-WgZ%zC|8@gtZT<G4nARVG%ad(*0JAX4GxsV(#AS)^ulez#c_T^C0){nD`e~
z3nI*9bN*a&TJ2~HrIW#)pTAWj`%5lZ`P?;4%;MdZj%Q`&o7g>=o104trfOwe;w+m&
zK7Vf$>a}oOg?IWK(GGNf@#4k4E7&JSVUPo#K#O8cJz9Knq_JATMXM7L@Ok~?fHP?*
zJRM(+NjQqhcyV$>=@1LeVLm1Q#qX*(#okXW1fsVa_aZ&0P>>1B0SeBIwuntoHofA>
zJ)c+64gkanw0TS&=O7?mgUeS|AMO^H$JqcSktn{fUKg)Z1#ixMwzQYMC7CJ4^8-of
z?9&(#cHN;ZaD9Dbb>P4O>PF$uS<KBG(&Y!|D5D%2w|CvF8DritjV-sSKHaQQknBZc
zFfw9E_LnTzQ_Ike;)IP{%(x8{!yY6@WZ!qu`v%n4*czIPTjRPb8M49R=2=s{e5k}Z
zIx(>^U|&>WAdrdfMC2YPnZ;9-@EtSlnGFka^vm{AT#Ct!)bUSG8eV*4gI1+wWo4<Y
z^l_f|{0wZ~^3QM9pK)*ogH8`e`QF36g=b%3pe{D+4!9bf>zX399nSD^WGjM9x&NWf
zrn7~ZipcVQ0ll?Mjo<rWIKLbR`;hS1$Eq!$ke%t8qRI5dl%FYFb2bH`mgTSYv7^Z!
zC68corAgu9;mIiB+p}j+rR^I0GfVVch^tQ^goUSM%6&f4@sU`D_x&mOq@=tHmT(a4
z%Khh&j=nEO$0FvQumUqxom@`X49rSZ#V$cwp8f{AG&qD5gW!cGUU`QW=r_zmE%Xa>
zyFAx@Qd$%a!gRtqF${@J4OomcW{5P6M#IcAt^Y4)TE%k}*tZ#Gjml;4WO?2giM)?q
zeCyV&(VRnzXFB0N(>COvZ)pcPi-J>oMpVat%^Y?^&9Y_7^k5mz>UnLD)~AAp-u!jz
za^L|BkRHmB5k@~lb5vSMsfw?tDZ`ute32(zp0pC90JI8#k0~7Zki{YDdl0U7B=t1Q
z#Kx`(6?SMGr1j9gy^7ltkIWG1qfiqn?0^9T_4o1J^fUf`eg-*+-jD8gU7)k`_t$ly
zK<1)+{_~+P|CiJ0|NJ0D$IoBh|5j`K|9No^>0~UXd=)DwpaE2W3OrLEtVHnQyN?%v
z^Z_9CURW46xAWwA&VyI?jSsbPlaCcmU%*UmZUWRYG?_*|EGK0aE5_m&a$$hqErWZ$
zotqR8tQDr{pAj~VR?miCm#moXoPj*$DnJ^3nkBh()y8;_`-^)scI*D;#pN`Lo}Hb2
zicSJFMDJWlk!!d~+@G8T=;V->81jqAtWy<hX?hdlulgj(Sj;ku`5Kr>)_bk?MPr^#
z1Bh&F3uC$qtcW=Zny&%UNn_|qHS-3<Wu?u$tXQt6vBPnKBq2(Qirxcii;vXgxW(Vg
zM7~d}nNbes_`qTiVtEVReql25IsR`_PMtc%0K|Qlq@y&XbP^*>HWYbrdl3&4Zg6^U
z9v!AZxP%?AUAM08$i(9B5Z<;}uSOKmX9H=$F>;hi^b@zU=x<8z6VgYS>V*k*MPzhL
z=u=ISu2hsl2s={(1IEJ|a|(?}1b7!;iKh<GKXbrlPcBU&p-y3BWTa}Wrj{0`m&C|P
zYGRYb3|;!#!31E}%+?SA;mdU=kECJX*0SZxw;*04;p{-An~dy>EVW5K#l#t#OwW*E
zF#}SvyJWLTNf^2IK7RbTA=T(~Jf<8m4FaECv+J;d-efI_^l&XokRG1WCgcL7=Ucme
zeGyu<i~9=W8=(P6R$0<@I`WS{_TsrsJz@HdVeua0FM4scPfblJp>Rj`>;bJ#0d|BQ
zZuMc<35tQ{xt`ncufrYn0pFoXFxTnKCVKhjG~B}UVSrVv+$gaDBZ6Gv09#1bu28=^
za7x(k)qfT-Bbj_DJ@pbT;`1!NXk;7`&GPs|x)I(awQy0Lv%KFqIydAZMSMy>YZ{+M
z9}4$WirUp|a=*W?X>kW<hAz)o$5wJ%K8tA|WSgZ_n|V43G~ctC7TPx%Or;N~PZaLK
zJY=R)hR={KPGVq6J(CPIGk=OC=a-!NG>6lZoOa<J!AnAi&1jn+Xw!pshfnv~+f090
zoV=VwJBE|2%rMl!MYJoJX0MTLDz8TO9^s2`Ho4usyMuEGEQS`Cw2-+3B_#eP8>O5j
z=m5fq;)!zc{^ki>{Jy<|94v?AX&Mz4=>7nZlj}{Xy!)GRk>krK1LXj@RhAyWcCw@q
zKeC&0kRQ=x+=djxQ%F*?G;Z{}x{@*lm^QnY)VzKAm#<Zy<lu?O{+h9RH69>2fRP(m
z?6V5o(hLY*%;~Mw4!V$e9ZU(yPE&orCN7*LVB~0v4%cclw97tf_?MKF+#!x1jSF*`
zuyaxG`uWnaV{k-eKvv_1MFYgX<?Y+k=B?X=&jVP+7ysYKLJ+NDDx69WU_X1H#cYrp
zfJaS6NqXRD+<W-2f1ncP0k2N&dA1f^my<2^kWew;jeVSEUTf4l!7SPx-*CoZp5$9J
zMuVbct+~Js@U&Lbt5OHra1QHq7Q9`ew4pXL(TYYZm<Asd!gldtOjuL)X`dA4Zes~s
zOx{^|l>NOrbuhnS*OecKVfQvfV@}g*`;c@{*RBc)`7s+DWL3taPoQkX?z;pY!?4FC
z$M6%T1}gOV!3Dm9x<Ets8q!%Z|4FNf1;v!yDaZkmy*|cchDW=f8TFq>FXoOccc81-
z)@ASD;6M+Z;Y73V=7+f#1O{xDVjw4aVG7q;W5sCB0Ep>)-Ao<rZmqMYxaRWFdSz;3
zKeJ^iYa@0fKJa7YWu#XiTOHE3$;!&|$72VEkxPeZ@~A6KCLjWHRB$TNzYP$i(N7{7
z{((6*5vn~`yo^f4CF}^K<h)fquZ30pXF4nUo1+b_nUD2)>g~lG#``3SI~C8N=FB}S
zxy1acHd<L*lU~=qV#fMxD$0*4v^uCWO+p(S1{EEQWQZr4eTSqMm-ssJ_aS8ig%|7z
zvXoc7N~5imJ=MGXIy*W#ApM{8qB%L3z$F5L=z8L$q9|orWYd|ZARwEh59x~~=5tA-
zv|&K_V+F^=uaa>=(D2m`ctu)*{K<6iBVq<!H3E1eG~0s#G?le37?NE7I)&SKh;;R-
zRAjZ^;?P5KORph%_0vU)mbm5gXu%;DSe8UH6^Op1<Pw;Dx?l6c1oQS3&HVe}GRdm;
zQC~H#UX5s(@afec8pp%Fe|*3&o5xR|iqg4io2szGT*$<e4x2B}O^@k=`*-K61khRP
z0%HU^JIW4E1gC~{Q+2d@Gv7WRfKd%UXAY$c#oj;hA<&K0ha+tth9><W^^KX9FWfG(
zmYfHycx(q-*yz%NU624wJJ*phD_jNq#%M?;#RvbREvsFYeJzDu&guzz;)se%9-54s
zw{7e8O#;I<I=&6X`svpTAm)Gn&WlZmx2y6KSwxf0k=1h#_LqoRgJSz60?B~ivnZb@
zGdwTy46POq|L5EN*vROpiYGY7I~fP5%<h^bawkn7<mYU|%|<7+irs$-&aGTXa(elS
zs6*!mFN=oc7s<GI1-sHI)5eo17(!b~_F7Ah-6dLkFjo>}yX#)zJ@Rw*VOQa9zhKas
zFM{*y$7lonVn+h;jg}Qrp@WPl!9sKOy{s9Y?`EhZU{TEaj0>>@H!1{+-m}WZ?_e7`
zk0vf((GKzRqMJL2fJL(xt*opR4!i@|Wc&8*bk&+n*kK#f?1mkjE}&n+Wq6GeTCpmg
z(PAF6@Mt}qHI4=rE1qDYM!4`B{3PvSB1uwGL>L8}ZjJ+;e~j>k#{b!~XLoFii_X@i
zN+>CJL^e{G^z^dz_OJ<wtySk8^3A`{#;+5ez4+dL_~~=PaRpZAevEWWx4~}7i7Y<`
zDK{d8srn_)7Lsd5b5_27?E#Kh{dT`+2oIBp^E)_^VtC<$T&#|I%~!;MFwmZkyX{og
zI2?3h>kNbuQnrzUFNq!^rDl5Z!B1)?oFPZ7WkqCI4lw^ty(q#{oRd7>-M2Gptip}V
z4DkB5i*1Wdztc0|D}t=yEMrYn+-76~h8xDH<4VMF)!1~S%Fty;aYR61h_`iFy&9n+
z=}HBYpmz3nv|V?1w>0n({@g)y%RNzbs=fPJm=ao=J57TSH2O}=_KX>57*3qZuIbb%
zsd4)8d|Pa6luHpat+h%eC?8nEg&FWn9$HBNp8^gOnmmRvkfBzyVC2MGC6s;;K`<yy
zpMQ&zk3Yn>_qR^BGviGl4`>Vo`NM`l{=_fO(j!yC&SfJ%d)!oS&?|Z8k4vhmzD~^K
z&-HT$=f@@h5Ruk2bU1kor^&}ES1h_uO(ZiVonIMWJ;6A9CiMq#z)Ococ#v~}pImQ9
zDViW`s%a>ddgTccg!CkQ4!yS6MBaaU)XZ^cRA*XM;4<UDhJIZ<T~JWKclxOXGF-YS
zWz+flhB1n{)_4~Kp}zw>3|7zj1kL?80sa|d^g#U>behP;6qY*SNak_*a-a|ibKHd}
zZT_t0jEsy-y25EGj>UZbdik2g4%Lr*@Untqbab#Ux+;{+(jM7zu>ajN@vccwPWcn6
zfRBoT;mgmt2j7dVUo9t7?i#wM%ATizqM4g|Y&@9#A{jA_rh|82Gr3br2vF*AZtkp#
zHMuus==hwp$;aVmsNCobaz>9Pah<M`$@oBd%VK>|zWMXkw*^7sn6a^O=>eQ0H^i&@
z?GLnpuslHej`XJ6kAcw|dz5vg5V2+J$;F|wv-6DTJdP)-ex*nolV+L>X61-XlPEi{
z`Ut9nZ-rL-BnaFb*taLo(U2cuodpQ3`HK4R>AN&AXE)|@Kh5h!#igEA*He+l+-)^0
zKX3bF+-0wb3DA>EF8{-+VRJj7C)A0|MheCphzcSv1AN50$i-DEuEE7yr7nwneegzD
zJ#}f?2w9fLD>bq=j?fgjE_+~<v!GFWA?;Svh(k`34F2cGx`bw<LfoEVBTdMr{1~t{
zI=Oam4&5@`pktwKuJ)KK$In@Y^AB!C0`|Bvma2o&<!_d+tNA!B_s1oFajwso*XZYF
zo-xs2=+cQCC24I#mbDR0@pU5$3z`H1+TLB<Cp0gQDcQ?kn927dT^6cEF3QXpG#U8Z
zC@2ewpVHd6XWuWSFiem9@#o4GTf@_3D!}CqxU<sE&dxXM@B8_^0?CNq$-MC|ND!DF
z=~fB|sd~86X&wv>GNb{9N+9<TF%%|zTi_+Onr=0`Tp=c4)fT^=>(?xT9ki?i`!w=4
zeu;ZII8mgc?0xo!N_<!;5;(HLrKC>Qlt#`C1AILrDhb8kWVFk4H2=i-RHcg~`P|a&
z3o_#|{Qfvc^Qgv5w~&g(6B+EF%zcCT%oV9zze}V&8qYv}BBfc#=mb!)Qns#9w;SgG
zOZ^JTq)G)%WW(hW-ZV!2&W6lOy7J9sPa%vH4uvDIpEYG+MDs;%QGd|`(SjIodH~O(
zIf(@a5yDS2N|>hafQotyTfO6y?zA!y=utW=hTAN}r@Wb?Ss)lobD*mM1l%sn)9%{l
zvK!H!?lDwqcBNA|?KFm=0Oyx(a+(}(5AtU?@)DFg<QYO<cjUN27Gna)-GlP~_&)WL
z(TByIZnW|9PB$kXB1$!9;icz+BB{Q&iSz_acQZ(De@d{?h2;{7I^JpswK`*5M<GU6
zknI(lb~n*}Z??a4bw%r7o8-By&B!d4#QgwpZIKJGum-gbI=G>k>5$!)P~p;{**2|q
z)0%2o<iPS}R~ro$lMTV4)Ud;Z1G9KPM{K1<8_sE)ORTGh^@ajve8zE^xNZoL=S2Nw
zE-ciAf486<Ioy;vzys3g+K)}R6RE7zVx$9x+;AELi9C{|{E^F2RTWeh6uo6}4|qE)
zVoZmc23oR&sE19yf=F4@YIg44?MbFIUBPHk5avJ=29>~pz4(l#oG-zwTp+@y&P0pg
zB<;WeyQEi8Xn5b7Mytt{iXc=nu{PQWQSKc3cas6nvA#Aa){ULaNt_C#_$izQD04NA
zyOy22Sj_}}#xe|!Sz21sjzIIrizZLz0_+x!9i<_r6L}NQJ^~_wHYEG)6k0d#xCV+~
zq|pKNB+gcJ2FzS{$QEAVr5VVlgnJJ>Brt)7bl?z6xq>QvrmqRNHO-U`5uD9K-L(Ow
zzFEuDWLaSs_?9v!(a><p^s7%U4$##$KG^#3X<e9qKtK^?L`Zb+QmCFh^9gn$MO3Qd
z)tNdo^SQ{oPU7j+PK#BBkn}%*G7D0T(1#*Z2YSPLIQ<04enc&BrN2P{Ll%Fe-UtBQ
zB7AUtTy4pKW*aihH3<<xE~yl51Fd@j&4MASHW8tT%R2kIZC3toq~iYpb|Le)scnig
z_#_+)82rU_z$OC+5-cJQCn_c7h1mbnSbPFDXrdCrqZ=x*LL|>Rk<+AGBQhU~iZ%3s
zp=UgXT@a3*oh6{ZhI|*7Sqw9BgWTa2z=5~ZO(1jt{QKkzc{h_ik9oUQ*xDDt?CR{Q
zo!vD=G-3jnAZuveCYAc<uS=zY$b>(7v<Y1_cB%r%-W>x#)>0^=9Z6*MNh)z0FbTM$
zx>4wLUP$2Qn8x<co`vEye3bP>lh0_TBgG<WicR-_3rsNonI*VnG1*`I(|=PJ_ut4c
z|M`uJC^B=+|D7rHpT+s_SRCSCp_;fFge!$8!bHJhF}g=|X<V%8?e;J&=zv@?t^jQ$
z6AMU3h9Pae005Ka{XS;F7W@`N?unk=Xo1>6^e#D$8EzLxVT8~;RI788h2=Z5ce&>X
z)1N%{Hv+zRtVDf3gA0p){2!C8{#zR$3uW&@Zp)~K-1%xgh%32K;9IxNC$?PF+O&GF
z#n+aHSKJjBeY}(1W7&nmttP>%zY^PQ^2zAPTAP-=gIfl#yDcAd`*OFE_qU~a%Nt%R
z=Sz(=9Ie)Ft<|p8ewI5lS~L3T;KV!YiiB)k>y=|;S<KJsz2v_*^c#CG766Ou@V~)7
z9!eL2NoH|L37>reWSh;}kAHy!(bG3Z+D)cFocQVwBA?XjqeSZ)oz-V@p~GF9l|(L~
zn1EAmTVE;{th~mM{{%OL8G>^h4nT#}->`n+|JpIeR!H(OZ>)^<2DBn=eVOZ4BZlno
z4rGi^PdebfV4}e5t{yWV0U4%<M)l&BcZcEZapcjP_wV0Nup2eaNz+2mjVQ8a21z0v
z@p}wo$u`E502cmckmt0J?gi$$FWaV2vQ9u&2EJ5lpl@8@LQQ{#?p<Qluiy}Irx&Ba
z;zXb)xnMHJGMm}4#@ulCis2`v9Pt1dekuxgX!aSg-sz`s{hdI<&G>DNf-6-c@byde
z>lKnj6Z!vQ@6F@6%=f=>bLKR2PMvC6v?w$Rp)`?-LW`2EWJ#f%lw@rv6_HLeHIXG-
zlC1?<!cd47XJ#zXqGZWdXDk^+gY3oqd|sOG_uTh?_v7*V=l8u&kMlU^Bz!)f>v~`B
z_v`iCUiK5YM%IO`K(n_Yr6~$?zG}Lr8-9mgLY;8|v*i%E<}HbKQn9-pjmrIkuq`;D
z#Xtk-B}pPXO0V4G=JB1k7ESlL1T>J@ghV+&{sgi>XQ8|E$G@-lgruDm6R}5{`WPJ}
z*k@2A1R$wzA<`>naI%W76L4TTrgciS0K@QwLq_&R_>R2=a$!1tMwh`FteJuV)|?*D
z#DR%|=K>#xlVX}%TZ>U9Hxl*W;63IS9)yoiNG^mi#_^bV;(EWphuIj_8_NN`!uWO+
z#KQ1601$KOEZ}cJn?`)So3qJLvWb(q7mR?+ijfN1VxHa&>7=3Y7Pk0_XNbIQu)2H_
zP%18-$z@Q196}J(!<Z!^Cr3Y`8&b$VX`cg!uom3+LYNi4O(|9r$_!5B;k+z4aZPB$
ztsfE3Q_6{++!$Iavsvc=t~GTKT7Xs>!7eU-;m-JHgrF*ueVX`oZ%xFc<5Lr3_|X_F
zl7Z5-n2v1g(|!G~V7OteiO&8;L6ct~2$IX|gs~omoeYJU?xPh{jg@23o;yQ!s~_{J
zwhW<E<iO6kZLg6X!3mpllO~<KjcQ)7w!m~1FLC0ZpMA-Rt9#-+G3v)gSN`!V7hN}#
zn&oO3?qEH(bkSj8oL_MrP!Df7gv0_-24FGEl4;jy_t-Bp7+J`^B(pRBF3P&qXgZn@
zdz1RM>vI*q?rAkgkOWx(_w79Vh-lSjx7c5>gJ}tuUc^3#KnPtH4XYTKkQD;{m!X(N
z?cYf{f%hH7Q=}fhFQnlG>6rcDN?W6hqASnNNj0FdC`5Lf;>$E}4n$bVFY-s}y<|W)
z80Pf|wo~LxoT6Ar`3C`BG;xgC>O;7lWJcw4tXCR(&5;Z;;3O|ggffJcUlf)`gN}gy
z*fV8l_Q>(!&!?<jz;s-{>^=lW7vs4OeXf=?m%d8ArodCe(^Ej258>?R*kNph&74_2
zdTYlZ-1X<Gjn7s_Tlw(Glch!)kod^o|I@YA(sc!kr%eOVxs4+K33V*!@heudR!@X3
z$)`0Nf+b3=pszc56?5QzxN+o=h-2B{ImaPP_|g0D-7U~R<TN-~p0k!7mX!&}1fJ5_
zbi$+43{O&TT8P~+@6&E5Mn_sk3r9N@)#)ypPD0k^-k#(eU{(u&w>RRMI8Q%37IzaY
z$NuPIR2Xi`4GER^(yf{w-&04&Tr%mZ0XnoQRvx4A2PRtIP>u$r=i1duZ_|t$`@frG
zV}Xl)9t4}H{5*)Tu+#jg-^YwZI)+boawgk5%*Xb7KP_=Lo`j39;jJMLVLc#DV;-#G
zg;*y|2x#=<3|ll**V(aS{Rp2Nju4U?tKr4JzBRhsc!uz(v~T63VyM85zy)uo2(LGr
zKn^Y#Guqis<B>eRuyc5ZkJdeaQ7(Xvim>s_z_Q;*Vp>S}^8QflU4R-!>08mtItb}F
zr|cXS`zC-9eA1}cJ_tOUQ?iyLa+nZ#6K>*`lxW|<o3mk7LmzVZZ*#s`K;k+#H<t+$
z+}jA}@J0_}RlR6Grd9Y#+eb@^#_>C3cwq$fgYr=op7c5Rq=Ge!ju&e2YO4*y(339V
zH(L+TjlH3kjE+7eM;qD|pP~Vi;N&&Nl>I@MQ;UGxun$rS-x*t1e@#d6AcCCa<32JN
z0d6`;BvuD<DktCJOAKLb<f(({7qV2^r+K_aQ-22n#zqW=3g8~bv~p+w&+(>E^oXa9
zWN;BcO(oAw5igK;LR;T;08;T1_^$@M<UHNA2XM9Ivg*JPSU&=Z{?nF4)8#6+N7#Er
z`EFmHcfWns^{hgC@?F23R0pZm6s)HxcUYam5jg)3f}~qt!x0Zq5<J<WnKHfiYI(%e
zu&17h$BVR!v}<~B@=?5C^!5SB;zA^rN<ss$EGq_^^eX1_6AsYr6oaotq|zM6Xrd`}
zs}bkU8Cd~{dTOs}3*IwP6KK<tz9=wQ8P8OKx9PH{hR@M9)Z;#37*wpemQE!3x>Kw^
zCKn6?C=I`ZEyu{dX;uD^!B6f?9nyVgh7*VJj%RA4F!L&CAG-G7^49#$4$v&ya;KJj
z07|b%<}4oi7RLzA)tdpxRHq}wuS_jy%Od9VGPZdJKE7O^7aC$8z>cBjhivM#>pT!G
zI~_AO|E^p8Ef$?DvkH_dt%4|&bJXNm**AH(U1`=G_E>Ha?HpEHI&&jXbOY#WvR>w)
z$cTM^53Ymw`WwuE<#guIBr9rH?%|!tReEC>oouQt2+j%b9!**E*3dd~cis^lmO*k>
z)j*6Rnb2qaAOYR8Y#Qd>X+cU`>SSVyn{7a>ye(@hCQHP7z#SIz{@6=g6sw=iwjWhd
z*;<%5#~?w<dd03-Ce5b}D|G@JU(&5n^5n^rqqAq=VLEYDLmj`}uJRb*;@QcEK>O$f
zrnVpctoLfqa!L97V{bN=%$5B3`q5ybD|V(OcKNYj*C#sU<UQGonk3}C*JbQc*>|BD
zGD*|E$x5(dCk`SZVg(ztNqYv5P0l*Caoz`lil@cJyvCRKFimIYIiF*09aVX6jwatC
z7YDN6XwOGGG&#$0)(hGMCzAr(`>0QZsg_XF$)y*(N1;*h^$+u>10B<88<aF{_IeLk
zQW_Hr#3jAnZ`UnGGo(&Vj75`AaRc_HppX#b-_-#57$s{)5?8f;1Pkr&ep%b=n=UaV
zW^m;7SWL#s>@2(wN;X-nq3YR{a2UIg7{U~GV&aWVXJwHhyqVlS_JO6BmSA*~5LkeN
zMOq>p2P`Kml9Pk(Dmrw8FtH#Vq9c>5xurMx8UuZycrbta0JT&RM#F^~sZ(qSedNr$
z%XaDSH#RtLyxjcKoTbTmg$JE}=iEUgq)f^N#U#PO#VJq0xrx@sU_&zRGnIa`F<x0x
zK6=Qo#4m6(uzP!p2dAPfJ%md9OlrBrUQcL7O%4S^KzT~c0XLaOS#5~7gwzN+Ly?{=
zN(2}naj|r`<#a&eOIMGQN&y~}s_(>s#ylh_hGAB1H%gE~>_L>S2&sj@hj%9W1Bhql
zwbD_4(*~->0D~`i>5TnE7&wFifIe6%6DaOs&VZm#=zKl8lmy9}qw>SIx=`)?fl%(e
zcig49rKJ!!w-6vhp$KHqhTp!WJse2iGUG9Yc8xsH%Z;c$mTlR|6P-rQZUlx4Nxx`<
zX`3W?cy*acILfF`SYddZT5=wg#aS<V=mOs7B;<et!Ix@b$-y_duTx8l>NqBY5aP}4
zpeV)?MLT5y;dcWF(K{o*Dm2gvRlMcV8XKJf+37w)FN2O<;aDu$WDfI4d4K(%dSD_h
zg8EPh?!s6@Z{mY3!@&(rG=4{n?09Z(6y^K8x!~a5O1aGr-_bt$>?mZtRL$5`x<CB-
z2cF84vQZek0RC=c?zThp%|>ZuNnL;8zE4!vNESduFatC1zT=PyC=Y}h0;@d(*~FYA
zat%_R=MWu$<+0#RX}cdeQK(#MOJB(Yh*-ChUyU!m@iaz#P0B}ZsaHo?O;jAt_WrZ0
z)*_H3#*Nnv-l6!|P#lYf$j13Pey0x{;w#%!`B|u;I}-p7w=6`$Dk!jH^m%{v?(KoL
z@J%0CPeH5>999bPF}E-xvH&7ABA44d^ldz=C{JNLJQKx?M(&}VD*%RUkvJNhmZj8n
zol*AB_FY2L(^`Pq%dF9@9svMnB&MxJO%PY?JoZ}$`m&sYyhjc=CUqpg-aQK^n8EN!
z7oHzQ<K5Eg{Ue-S3dA(|r57buL3!owZwExpEQ5L$W*EDp)!L_1C*St`b&I<ShT;lC
zEbH>1gjcS5$`QT^Wh*10m5gPo=XCTDhY)5vxT)O19uHb0qV?pA8z?`?a8m)F+_tdo
z*42c<q&cG3lS+^H$#m2)$J)|t6V)Jk_m)A|a;r9<d*hkm!>Ho-$Gt-Ib)w;Y=AvX2
zoA%|N8ZxZ8G&TN}ckk84xsr1wZ65XK8|QdU7FS!rvMx032AFss&<SymJI%sX^>GqX
z|I^l22FG|^1Hd5p>V-R4YTMt$S|ICx4dA*?;9Sns1~4rDG6v9fgxNpx(6uyyQPImA
zV+n}Nr})1EK+1o1)PQqyW`3y0$<6#cUG*DSMr0C;_x>P$(RGbz_pUKymHUu{K)r5h
zKZ5#;(bXLDJ*kwlbK_r*c$d8QVif)9;%^|f)<hy(E?1zBm%s<PmPk*i;}Du9mUq6X
z)JrVaKn_RTNI&{uzic-g!3?PI4Yk4s(hWSpHqM{y7wXlJE^cvTLb}ldY{JR-IC9l8
z&R{CUQF`Hh6qH{pSXp&x_(T3eTpN4-Mt;*%R9A7uqbLli{+DXuBmq#yGV1AAGu`Y$
z;JTsAkmMLtb~<^d+mIA34pET8NTJDTgv2C2TlkK63D)AVzs*%z#J1hKOcezc2lsM_
z?8T&1aBw$X_7Jj1N?&RHfUi+VqlusI^UOrW<>nT&TtDkaqZ$<CK}EyC69{dZo0KE}
zjil~V&&6(UHWn&tC~1^l#0Xv4hE*6@5NdW4r&&%D=-2=h6!-Ky=LvOI81=DNirVoo
zp(MRq(ZmkobZl==Pa}#5y<C5YsS3DDx4UrO`1><$*!Je&td`3$-t9C}*om5Z=iW{e
zBmiIj6hvV4<L>(Ktdjl18XWAh7Tef_I?lOaw-*0^^2Ei*9s2;EFnmzA?Ot*hiekm=
zdw_Yk#rS2T2u-0L>U`)j=y5pY#_u+lj_gPAa{LqbroW-85JIu>#FBFv{rv~c0$yhn
z@?)>pBWK}!J}{P+_A+!z9FLTN&3-dJ!F4F0x-r8;B|vj@Fs>Cf?w6ZNIX?o&*pmK1
zIt`#InL#KB<+i2|YV%^}yPdPlEL}En7%%IAx9;A<FJepG3`|P`b^gX$fpe5?G`#En
zoYX0O-rq4%$3=c~;lW)d;d4@#BBp59!Mk_<T-ZaY>(x$FiFjlkaqvbA8QMSoA$a^w
zq1LmxnFdIx2s$+mo!B-sf4nyg9<Ps(XJoEhEPmwK`ZL*&dRm?$m}(k|2Zg>nY!0Eb
zn3N51#cUQbrJW{bbaM<2!BXoBwkLZ-Z4$~>(Ymy?ISBS*$f1$ivKIr-Q3*~;opltj
zO+rQBxpTk^HbWIXDLcf=ZWxrLB7lsaIw0+GD0SMuG4ljR-09^1Yg;k)-$@N)nz%KQ
zzBm85OM<bD<z95$1XZr~eorAT;ikA&4Q%K(U60ti@e*c?==wgvtw`HaL%GBIK_F|(
z?37+Gwv@Sw86yl~+zFjaxkt)(#HpyWb(NhE7W!jIXQgD0DIbGbGMn{w?6$t0P2}rp
zr@nN#vsD>Rdu5|T+2}YPdp`7Byj$owGEqkPRl%4uo8v!9x7`r%xnu<apYu1Zz3$@;
z>|u>)t>UBERt)qmEe6PW>sddZ$?%@s%1Xq98UNmklI0oz4W{cR^d0}Tk;^@1xV0=c
zw%k49L14DQ{6zdSvuNb1C8wQv@HYaHrAM4iv%F4M2I_juJgUa@jXQ5HJNR}MV@bY9
z9zh*_p0R~G1L2`R8rF`osCfKQ6_NFnVHI}iUEDcs`-)Mr|BPqOC_aiRcFXkV{n>&3
zl1??5@%jpxrlzFwSaNPfvI05{NxZ1efV+&urdT-@>8*tay9*AwR!VPnINur4$|ESq
zBWG+0u*S<l9=8C=?tinf*I2F0bFpgE_40vvlBRz9-<4$}&Lsah8gpwbB0;$bps@+X
zv%bd%yh~ATxw@bB57Rd9qf^k#8)N|viEqjtyCz$+&-T3WMeT}1BbQpvd8-@cCUl@~
z;Uqaf0F0}oq@cI*CYk|QaTe`Ntb?ziFQHbHl~`_F(r;zMc$)oXV~$HvK1Ps8PzT`3
z7?~Z{RVn!zi&t}OsO)s?D(utEc>oHMG{+7?K_EAfDzDdn=`DM$osonR5UW?roZCcC
z&PhXqzzEDz(?dEbr#o@5#}{qX@%fA!IU4u4_Y*z@;=~C$Vg<Whxffc#GueO&{^_YI
z`G*$Z!h6V2T4lU$ACpHzV%Bvy<uZFTYLOYeTiUw9@#Qyd4>CVQnmxDIMRk@r<9g*n
z+yj=&wNEQ_uhvz+Kvm~g5$Pn9VoH*_;oZ;k=SZ49z*!S^$D7$@To2FkZP63@V>gm#
zlfp77qg9=fQIsx52suQToUEWg1DMJV9jkwZ#@$(F9U|AMm+XWzr+d%(v0WNz<m4se
z>SD5$GLQcaEmWxhD_g;iged6>1V3@q*&RHXzMdr!zP`gZ0UFE$Y3<Jqkzg%Ay#TyC
zhj>SnFFY2KYM*^y|Fw3V>=ejh;-5JHznGBjB$>NAN<Vc<gi}dZP*44#B)%>Oo^KiA
zcF(f@*nI>KZ_8U2_I(r%a@mF)v<xguBZhCmL9nR7?Bf0))VpCx4VjDE)E`~MX_&U4
zpO^Nv_n4Ouam_9cp2YmHOPBf>cg@izt`^d%wRK3EQtG=aRT;Wk^tz4g<`iwes#$f#
zZK#Q(vF@#wudZBFn+!YxtNzW#HYJVW*-h6=&CD&Q%vFkesrgMQMCDWo>Xe;;RM*J<
z^&nhqiD`>?dE>4wNs$PXuDL4%fr<gpoTZ%BAshAWh72ot=##%@Z@8KdFKlqscptcW
z;|x=ho}*{yvsAjf7$4YJP4UIRcKN?fk8n&^dx@D}=a;7cw&*-E@U^h-o)0B25Qqlz
z#|?F9OQh9z<Sj|5G>tQvaXM2PdO<1+Py*R<xZr53_oys2N7#}LJe~pw40HVmIoOuG
zWxE{rf44eyj`a#J>)S`$urFTiq{*;Tt+%(AQ{pGo0khato0RtW2VGkY6Sdx}&3=LM
zXS=tzabno?fn&ItmLoU|E|$Oh;@{4o3uiOeKJ^^YUX=p`?z$IB4!zACk8Xn*(4gLz
zQg<daF^P)92G8F-k`0YN^8p#?Rc5qptnXY0ihfb^h-8e76%wmpvGqh(l!0~s@X;fM
z+nKeLW4dOAw~Zv?NSvTaZL6W=qv6v2cC0*T`mC{ltYTa<x43gS;igg?QFJl4B&Wxj
zHX7|pN{uU9AuD_1W#v=(jhoWuT<+X|xrargCCWve0kmCpa@AP?2I5A73pcN<`1iHK
zZvAQ}+^7Ug%-tRUX*sha3!0k4SG&z+_B$y4T_coNOE|R07mty(n^x#xv7L11h)dXT
z`JDeQs!pv4ZP9BLnvbKXHqWa&*J-z0v^kTWsgbtIf^HZ<k;nPg=_?gRePUH18r9-k
zROII-Dc-qKP+Ykqjv5$%oD;J_feAti{s<prL1_;y#KV35k4FB|kEaop%i*8Oz{4$Z
zIg?b&F*b!4u?^%FEIXn6D?*yiL;IArWhhk6*c!l*O<H=Kw$n1pJ3m&$oekcRU59%3
zR<@F*PPJRbcJ25u{ll1UYtk6De=DLigS#w129^4(#&6fpI|MJxGKU;_eFGok<?@>s
z+8@A^hJ)J-biviXpFja89F?Z65bsQKlEm)=KIBY%C=+#0sc_tQFV{%+<U79YP_5WD
zioO4$gB4hr>!<_cvkwW_gBh_0k@DYBuOxWB1@~ooaD=T>MzPb~YtNADXp}Yv#8GGH
zWpW&j23{48QE3EMC%4!Nf}@C6ui-4MG@Epis?>Ufam#s?SiCA*faBxp4-)3T_hM!2
z1J}~hZf4K}*VWp#2T(2F`?(J4)DB?Q*TyCr$5i4RbePS_X(3&)@F4rNYJwEd84ySG
zpiS4dgtth$BR@3z?jcn0-7uzIOnN|N61y};amnO`iy}>HSGgCuoWfVeg*rzA-=S~f
zh;P@vR4HK<CQZ9Ob|bIgppz?<cC%_F0Bz)&?TtY|<yFzZj*6rijF2vGN~yE_r{`iX
zZeh~Pg)Qvu9K;=!d$ADB?Uz20pKbf^7w;PcnS@SXfyabKgb8(c2QJUne<^z6ZFb<^
z-MB_<CMt$kIn~a@CcpJ+@mS_Z-vg#C+h4O}y4nfS#5P{CE^T!c0|&5u=(ehJ)Tuq|
zN2aY|Gnr7wwm5g#G=u^0g&MVMrD#tOeKksGA8Bt-cY4*O#of72=)>+U!SI6b&o=J(
zIYl2Qr)NeB#i*c$SkEQ7LBX}0ECmOU2@%kYaa2EwqF$)J9Y8299dL7e$FZeoQtlQs
zf8U9qQEhr<;zo{P4-4%;#kM0ui1tm=xgORhv8m!F4uuk*=hw=8w#yd*Fq|m;g*U2y
z1{q#VNkea){>;9ZC=c=Tw(d$NKJB}3kyHG8ziwLCbiLE`&7dbhDV>3jWn&+<Wo%mv
zc-#dE?b7PYD$5>?9~O^yf=4U=okM{}UyqG=$4eZ;vohPf#TM(2N>A%RE~cxUIDm20
z4<Tz!DfOSBk&zY$?=DezW*X;|zdG<7YDMwP@<0Z!eYgkeC{$wJC}`8r6BDxWn9~!6
zjQZJ*+mM2au{95{gk*HGPuJpUGM4KEqp4<BA6^MSF8FpD+6^thLHpne$iS}m5;)A`
zSWQhpyYo<8S(lr3GP`y|it>S9z%5KHe4Wktb>wE5eNO(C3=tPIDi+>U7bu+z5H~D}
z&v71M**A=WO^k?s-oX<fxTG-nQ$nyf;!*JgLQHw^OX$(j2~HL0NtTunMWXZw<e^cn
zHqzG-?mEI+Hlqb3^ZtnR_-n!`ZhbqU=%5qF0MW8=HIqTkjO0|RJPfFEwwNU4@&#I1
z=h|1z{zkHNN$=I*%BrCY-9g>kv)@A3>Vln2^zNaOU}-$6+QOI#ejEstnSRS@2Hj9#
z&pxpgt)cPOo=*i8y8{usJX&&~Le$Um##8NrBxq5`RdCWNJ{W9_E0zRd@f`Vur#VWo
z7@o}#i7{MAOm4H>o}h7PEV0^mSP5B%4_vTZ2{9xUzX1sjUTbIKlA3mRQKvd>t#oh~
zgdM_;f8eH!C1RHq(Kd1r1QeN`)19Cdi@_Efwd>(X2)8<^${B!h3P%V{pV`qo4ot6!
zB-?NOh(L#~9SvDzw)e-Sksaiq_#<N!x?<^OV3`xZ<<ZdngXV{?K_>(5pgNYy?V!;x
zGr%v@0O;j;BZJU|fsL?X&7tN3tyCfAgv|tuU=@WM&kq)MY0)lGk`8=~K*B^#0}J7U
zJyn}jAQpuX?FcIcAvfvKSHMT@=qQTsCpFD2F=K=u$#wU}P04aX*KFYbymmiU2RXl4
zj+nhFM<A6d(|y<hq+^&&&D_lo&_*kooJ6HoglAw+V{WAjpV+7&!}dR#>I4W_XAVc-
zOXd7VEXsNO81%YuT%-5cfOklh8<x?H_ba9eWsaIa!(KQ$W82#0FrEnmc({dk1JN!3
z<lYFDBoN(?umpYsCwFz(G#WYwYsyYuy#17XFA%by`XTJT28fNLA_wu_tWhy$;9Vcb
zYY8$ECc49k)P#)vExE6W13|%+{Pz>kV+#-xrg<q3|JDUt)N16p-+YOQycCKHA;ORx
zOiEP)!EaRr>`I*N0S=K_Q75p{3jrlXXFLckL~}JURTpG(weYb~gjf?J0~<IKC+vBM
zmlZkjjT4dC<F)L$nj+P89k!fe9@yEa6Ps|PorjsJ$O<v>v@6hm2S{u_n+N<pDH3)R
zt#ahCV7w(Eh7GYLFCqCi5^WVYjJFK&!8rMAq@57MALza831f}p0!tNB=%!-V>Q@n#
z#+vU;@V^!>L6JL(V9X-KdR{K?Mk&2?WdJ|MV9PS^h_<fyR2cxHsnPZxepM?$5iJS*
zTQ2Hxf<^nV8vQARMO8J29}+^4K{T>W=wAF`bSuKO5ohQ6lRL0vAI8&fMMk_)-A$n`
z367W=Uxdz}NE^sm@_cTwSUg(MYHjhN<ZAJT@<IbtgoY~%w1{B}0q<$k;vYcafH`na
zFRP>&dP7dVvt5<>_*W03*$XqWKwBw)<P|SsXdUfKjkcO2yL8{#X)PkqR)GFduk!^|
zO*3#1Sd{ir+T+8c!LSAwT-Q-}6qCBJ?lj{EGFF2x-8Lka?d1l@s~gIOMX5PT8XZ88
zbM?Kt0O1LkUEBSXn<I-sQBpyp1&>8(pQ@G?E55?26F3}2)w&==GDy<2fiL)q^++X1
zX4lGU>{Zqs6akdFIeK^2J>U#GqcTZ+LHC`P{6HT>6w`s$AI;T#jYmO-7o-3Wx}IYz
zP+P@b^<dn542iTU?+rj+Yg{WlS<>8mC?y-nsswyX5xgRm5OJO4N!o^5?ic$W#shqD
zL@nE^<;@$J7|0fsvb>NaKHH)iOFX802xqrb3h%n;^?03Co6m>b^v343^jLI+N$p5<
zWMpKg;;c-p0UCZwy{s?-b`j@Ah9*5pUW+3^eqQg_y;t>7DCLKP;0i1QP+$$!GJ(DD
zjy&du3w{Xa6saO@(IgOnI_X}}Mb~sm;3nrvWKgQGP7LM5EIQ$sP%A9~tmi)0oREo*
zi;-aR4gfk=;L`3pz#R(Eh#Kv2I%pv<{kXf2TTO4HPE{qZ6KZzp=3>g)usg=TU)&5v
zt`Jzq?zinWNr{Ogi=`SFs7y^3sZv*OpguFOB92f8L3}i{EM*Gr#SX_Ipcp&sy!a5Z
zgkI>zk?>u39FOAwX5{R$7rC5V1uAYS)9G*{O15|_#iIjbr9`y?v$Ji~*n<-mAw5CK
z=y~iDp(#_7igw!2+(M-|v{NZe7B6j0#gWBcSE_}N5!B`h4e7offdUJ4t>Cw@q~GR9
zD^sRNITdx@;8P=d-0>3r9Ge4FBdmO4O<nR<b_0(=L#3@O`vLT0;b7u^QB@5!jIgwZ
zQqRUWp2#!KK4GI?2<>0|;c@jyn&;RW;2!J#fr~*w`*<Kce5Jz3hdbZy#m=+rC^ijR
z1$a==Po-3O7p|j1I?d^$BwnL|dW3BkO?+&8DUl&b$)W9`DX$sG$KH!7M9jV2NtfpX
zbdNg`n^h!H^O0%>+!vGIF;R98wX@8lM>XQZ0|t-1HBL7lW>o;?7K4~>+veJ*-xFSV
z4tWKg7z7cR&kK!6AWux5yg;ViNJP$UJG`kud)hQS_})wyrc;Y{0ruAuVhe!~e8sbj
z%0+w}Ng|hEz|u<>N$#%&<1`0+t1mFBS@L73xdalFS}W04KcJ*6_6Byq3Rr1JD~#<7
z`{;(>gbo}bXM-v!4m-vP&cX{e!%c6olemBeDgt6*CRcqqjgn0Ip<ojhI8%T;jRfq%
zFd_supH713crefnZlst7tjdp^_C*>|u2K_MDQ<+RtTydRjlfC*tUk7z(aU_g8t276
zNP!rpK@^xJ<8wSsG#bEDs0L0govG4|q?=%wTpRy=Vkh}G5<Iiz;Ql_u4g{i=UGJV}
z+gnq&VjiVmusG}n3^!Cj7hzu&#=T+BC}h+!Z1VhJZ5(XE;1ULR2tovW*4Z==l{V~g
zyFBUARB2nG`wu3i_HpsNrd~--dF8PuxVhmlV;l+NJNDssQkcn0$48S!H9m<D_gsDK
zeRG^wKc)&zQBdbrl$v3-2(WTNw;k$H5A^v_pN*}-DyVP;$NA|_;2@ir9Q|@(^1^}b
z?O#lln9qYOsE6Pjzu&YKn#gHv4}*HnE9Na`io0v2PTvvS&!YJ7J?3#AxUe`hVUl|3
zO<p&`>{2awKmrljza(BQvroduSgPQlyR?NW2$c*MR2boRU+L!pfMcz$Gtm0xO%Xa5
zg$(V`GcHRa{-q%=*|bo~FEII<q*>Lnr&s~kkjD{L4?Ww0^Af2co!CG{L_#ZpUBJYB
z<Do*BRwva5BdZ62VQEj_xm1}9nvIa|`|h6dr28c|Bb04tMLvE8Ojt=;d4Grc^cz~|
zo|m(4EgZG@%ep7~J+;245^$ZVECP?I5TZ?uDEj%(bc-BesP&Tf=NthgoyRGBeajHV
zQ#i~YZwOR!K*Dz6MRrFmYMDd$xCM(vW*+by|Am#GexV!nt(?nsJ~R;WT;zLPIwP^Y
z<`t4Dq+<IB8N6NiYf2=Zts)fvzM-)Iz6kM!c9is=ShUq4$MLndsGRZgiR(J|d79jv
zOFYQLOMo#>nco0L!9sbKoE4Zch!DL?HUuU*vdp>NfK^Vl0BoWAle8n0v1ZsaUCrbu
z`O%U5vAg<x`$up^vMjRS<@<&v&6#ke9}r3dH${=;2OdHYl*l?IM&a;^i5d-n50LQD
zy0)P`p}kJy>aTcq{8@hOYBxC0fU1*<u!P;)OpgZ^*matUcIEp9C(U7q9R%O2KTi3~
zFm;Y(<(?u0q=}fuj_2suM-#%Y+nv~RPkVpJ$W)Ol->`B%m7Kmcdkwztk3WXdf>d+y
zcdW7lwWPNHZV&P+`>vKsGfJo{YDxjpGwS(^*BM9)ck<H@E|@G_sy}{B`GWuaqsafi
zFG62t10I1Z_Ibb_ui?z9VFa4tchr7Lca`BerHk!m)Kft0c3t(kHC;H`4bKQdWM8*p
z@nie%B1~}IQc92+-xSvAEG&-DUD3$Y{qH|l#(TFvzVLr##7GS91cfCM#=Jq&?+5PA
z6EUq~cZE||&ZG(l8TB`r6<FyHV>s#>rMG3=2efbJ&>|hUsx4Ca3;Mj6PDTEjC9<f!
zGaG~;z~n}@LL`pQl=h?Q9t7HOYy9^R<;VD5miE}uqs~a+GKrYz1Xh(Co%+Ye!2c#I
z3uwe2U-=*BH~Hg({a@X&|7oV1|6BwivS$1;{6Di2&PJgCe0WrrA6vd0lU=7%BB5KX
z2)QS-Y!YvxjOK=vaAXuBO2J1FF#=99Ogd;w0csI3q6>X5iDm-nI3M~BT80elvll&1
zLSV^Ya{A!IQ_}egl?B(a7lX<aIpuql?h6*bK&D#Myz$@5!4rknA)Yju7z}p2Ssd+<
zD`b}m1VK}PTKnR^5*S^VYA8D%1(tXa^pg*4&t}Kh5WjejV)W-as6~%Nqbv$5e+uAg
zhg>ILeSf4-KwR08=oc4)PC*?hHuConD7K}VV?W8QDBSi(Ut0w1O8z#N!>QyA<CYo)
ziu<UanCNspbH}_ti=5J@pGD#JC>V*&sEif<kMD1O^JX)2)e1L9Ss??9@12+e_FgIV
z+F0W6xJD_@hPl}$PZze<b*bBfbYypwd;uRu8@N8=#-c{YK>yG8{?ILI#H5dPri;c4
zP8ASzwvQSL{?G!<12TT|LY~6#^LIG6rpdPlPGVvflR_>C*|W(~pI$`BpF#^=)X|A6
z#x7-U`fLQM-a2+;yHQy6&!>)U+^GC;@q%kVEl$**E;>U@-}=u{Dif#t<+reNMPEOk
zWWT83s~J9<(|vvNEE~T!`)>6j`$mx|U-=geC7)XTk88o_7jIC0`0R<{>#3T#E036(
z9Wy=VW~NzYZo9PoUhBPgX`j8kf1+B|@C5^xCQO*1qO^>kYvVQHa~(9azox473(l-Q
zQ$(ZD;_$dViYOSI+_r98c{}p98UqUvd3J?4&K!f4HAL$P%_>!IK1W2{#m;4OV<7F^
zzH{fLN>yIrb~OWG#e7!|G<!9ooo=ms4SCZwXytZA;*_(mW3$RZRJER6i?}vdXk8Za
zTECEuG>{U*>n6Sv5+gw&m0A+)6q40=&Yl|98h=0kA|1}d)vr}fUfvgw#DTA+a8jo_
zdewXj)iRSD!*cNxBJDFiy`2>pk>l4Ybfie6+jR~liK_#eRrG7;{aRgQ=i-~MTrH>Y
zN@T?rLTJaQOvZECd>*J?A;d~y&*w8F4D0WJ99|1=X4CkaS-?D8rf-{!d~jTmiYvT3
zxGj9jt?oiryQhYA$G`3GHswc@kfA|I*sSY#V(Qhvsc{%nrG_`!tzCY2OCncJJ|BMz
zJE@;LefqS0EQ+C_IHb@L0ZKi5{``5SD(wnj-0qdpXP>_N({<W0bT2@BO%bGF4hw_}
z(tdL2S@<m6m{Vu}xR<6k+zGbhmhJrfIo_=^TqC@>M)8OLxVe)zF*?{#^Pph961yF;
zRZR{`1bm0(gs((Sc~*Y9w<$-Se!90RGdTlc-N!5AYwOvkJGgaZkRGKk{x-+(yJ?@{
zwW0y{sd%jA_iB8)*vCzCd-{M)T)T4R$``_gkw=0BDY?{r=Y4%iVyMg)etW}>@!$SC
z12_=Hwyoo3@``T}N$llTt|J@oR!>bCE@<kn2k`<T?HZr%eKj;OGzBQArvVb+xGD2<
zxO}b-=)vv8TlLQKyRF7A{8w!%AyKZSa#>&D$!OlAX0^o;-1dC%*;(Tsu^~Zt6APpQ
z5U{6v=x$8c()!Ot7D$@(7Z#$~W!vr`Cx%~mD$hW;^Xs%`ii&c;d~m?**|Qlz>abk6
zH-M-&BUq=r`$*SHJ_yeYXTDLxkG%n)Cm1rT_u{LnKSKQ+w{O#?P0N7~nEQQkWt0jY
zLPoh$uA9>x2;A+V+Z++MDv%cQ;dH2R_Qtzt0$de{_WI2Nd2C(Pq!svsX}^dh8czf_
zELFW;q(Yk0&lt@jyp~ACJWlv!?jV0R#Y?!HPw>0(r>~Td573!1zE;WS{;ooHH~$HL
z)_ILzm+`-eEcje_s6YKpL}`3o{O2G4xFEPA|Ib{ZGr3dnu@^`+g39#+37D|<0O>J+
z;f$7LtPrxouEoVI0uFPF2E2XXA-f@`ECS6j7b@MG@eg^lB+6$k!2k64Cs4rrYVndK
zOqS<<N}LdqSd(8}n8Zo%E@Lo{gNSxe)y0p9WUgO|2maJ>-1am+X%DmUHi{T~3#8VU
z-e)A}#CBZZw<dmVzaIZnF@X#C)V(S|w}XJvL>c%tbLPxuwfX!m`wQj_N6!_`(uXIA
z9AgAv5rm7g&z!*r{u2LqN-S;`_R)2U!Zs@AM~;!ha4mn}r&+huaMGXfIab<nlWjkP
zLyWrtUP23oZEDImOp#DOH7q0(Z@-xfZz(Lc*=MgK+goJ8jPZB6s;d$&Q?Z7LFMuF9
zEU%G;AOhQqnYT%e8iO9sL2(toc5JHm;6_^A&I7VGb?+nv$PAt@H3!$T>Vj=Xm~X%&
zBCrFT!7PBNsNlpo29ENkyluCPh;G7gQplfWEYe$uTece+O{W@pE!>~H1NQ0|TZ~sN
zqy<Yrtf5{;7MTP+ltj4%deJOBfUZ+)(!wI(6lY*%3=UfEJ^F#RcgRCBN;|6Jt7qe5
zC(d)-HOV*8qR;no@|Q^ya4q+*|I;Z~l51L8FCAA#J~@!m(uuXEKji%$h;De*TF4p=
zmh8%mD<A8H;IT`-uo`${0c?#Cixm{~)}-&ed(^2B3vMgKOkby}kj)D1jfTb0=iC;?
z26RI?up@7jd|3X_lS9ex*HTZ*Km@+y&emN)<c!_A`m$>H`a2HSp;W)Tz3YM=1`|IX
z>ae|quUl-aZ}(Wb_}boz?<!6UZ}XPvF=ldY9{PNpL|*=Oe75Ew!i<2>uiw7CqQwPP
z30IJ3@#RAZ#u6-8q*ut}5*@G2pf6T2@3(8D87W1!iw!S5mOs+K>5P99SG!%#1GEue
z*=>n5N4qBtxW<<-oR`id5=;6U4fJ-O@WLfLH7s41b>&>1S$blUhp@KS9TgU4&r@8<
z`AjY@-G?}3W*=;E*GGM1O@#z>R@!^_?#*gy1C{VNM*VJipZp6*94<kNa|AS&E~vg&
zg2~Usqn^MllBT|O`z0kauYVYeU!>1C+kX8txAZ;RL;4o*=|1&Vs0y0nIcB1?wA4uA
zFi41m`T1Z$ILBj`&kZ2%-LH}Az;oQb%%#7^mOy^8TNmbqhPmDYBF*t0_D^>2cI)ip
zbJCvfSaDjaeb4=C3GM5JN2W$7Bp&KfjX;Z$TwojY%H<gg16E5Z-uc(*&{o<4Pa@8g
zDRUg^L^A;~p!on-e{Qls<}X3l&1Qc}P6Oey8M#a4-DtxU+g^gnZvqA70|Taz4UFx8
zxr8raIN~f)A0}+!ASDTJR~&LEoi+Idi-@YzQa$N=f|4bKJ>!Dc`0mpOsw83lJ1hun
zaUOR+qKfZ^`%t4L7vonCkdLN8hKd>?oegmpRB(|2(<!Ti@`|!Jop*m-M6sn1^A1`|
zcpw=gz!dTL*2zGA+jasY4-S}&oNi}GAo<AAW6P%#a}OBXd>mpcn)Lk8vno*<M<pZa
zEc#sIdMo?JAI;cg3p|=5eZG#>SmgL*ncG+cUC0@Q%GJ!{F$$D>4G+*@aMWS!`zSEV
z3h-KF$2BpE5_5ai+6Lj)QSeuYAg@Fn3g|;OMqTLY>2ZQH;|y<>TNT86Y$Y|WcJXP)
zoN0n$9a#|kxISDtZ7yy@pYDA-u5A0!``2Xt_j_*vO=<f3OE?FdbKZek(syU7rA6C4
z98G#^RnRF7!jQDPZNBJSJm&BLrT?5k(M(G&#~_wxASi2*cqI6_x9Be0CR{q3_hEcR
zo4Gk)Mc>`Cea0Uv`UPwB=2RB*(cy2i)22<MSOE|E4mD<=TXm`g?}E~jTus{`K>FE;
zS<@-5yKNV8#1SW<PBFDU%u)lDDNBZ*Ls3Z4_Cp{|!pY+&?~S{?#vkL<JU+%2|NR*E
zi=wE+gLOF*>vM<4W8A!)I@m^X;);(!Q-DHnv&$we_VJ~yL@jwO<C;(&;!qCWQ?j@w
z?242WRma@<f&*5;?S_V!O7RBgC)}-|$D0Ra^INLT!yah3^^ITn9O1${|Jy^?s*BEW
z-K%!IgZN@@$MsD(WI~@jsu7}3V|!s0HT8H5nm!~TKF_>@VceWI?5fXYAN|f#R<PVd
znX+W(j|<QPTz)_J$5pp*<356`Zv1~-br|Iqw+vFq29B|%KMVD&%*I;zXoXcy4zLb{
zNrbeQ>V!dVWXxE=N)64%6a`6IopsW0n}l0g;nHDH=~nF9&eaO`nV%ld-Q$mE1!!hU
z7L6o&@oYHe9D*HLMbt<Y?QK#Iqte49ZsA03|9-}_X^=m2LQ<iihRNwMsAih*VnxMQ
z&fdFEj^Dc!;oj{|Pb~T0x8QP~;=0~*d!d7O%t^=5&}6lt(w59KjA)yo)oqm39MVBQ
ztlkZI<|e72#2(Ap$Rjv9A~xK?$vhQ)yc=Ri?{8oowBW?V!@yxg)@4tIkA(iV&|=Cn
zsKeQ@IvY!hJo=e^y{%<<z_#s3-y@D)+?MXSY17y+bD$@r2>s`<Fonn&p6!&x>a@;T
zanSJ0$o&WW^($@6C;1wt3w6%JC{}F?(%(Ynz~gi?o-<<!P)h@a&`!zEXAEh`s8S*u
zG+foV);Ek9E#$moBv%(zd0`(F)KD-PTK=kvXFrh7?lF}`<2&fmn#2j463ra-_yMo9
zi?s-g-ti#H7r~N@Ql<z8XK2q$*wdLX7!pA5W7|$1WiW|U26W?F0yvGUQY}BN2BOJ&
zgdWNT+C9Hkcy#2R@PQICQF$1qZ~1}6-#U9?5qR9dgaQ^#49lb5!C+OxOaw;iSR(BX
z>31e=Lx=`Vrx-`4I2J-i=EIhElXIpV{&o|2xh9YVfvCp{IrAD!zV1kUtt&P(L5n4{
zuJ<!5R&@fV(kT`sUE#}KJ9}^R6xL09N}IW%)-Ne$i9#`CwO6Qdw=nl4G|U|~j9EzA
z++DhlvUw=tsg=Q?8biac_+`#!;0`QgePS?k26|XV<+Bo^0R;iS-Pb{WQ#=#)(!mI^
zki^Jb;B#QSPa{ArL0m=;`*E;~tQUFOEMB_QpU4R{SX7;UfO)~TT(uR+e#n19hx+Nw
zCnr7m_ctFa)b>9y(}*F9cH$$D8VPwA3|%SThZuL}gz7!76zA9pGTBrb_`QjMt)M*P
z?D!^}bzL62;8}}LOD|xv(1Bb@aKfYs2Drcw9tdJ8-s=@8Sf*^J|Mak@8e(%%3b{)6
zjo)p9MCYZ8H{8FLC!Ay}=m~M}0=Mo9K&t4_$=*oFpG?(`VKaAue2MeZzQWF}GIQ@}
z1nV(C2AEiJxo7qG0(FW&5mKR$ypap^Ukc+J$SsH-m|Z|qC_oY+dONF?F)$FbYyzN3
zp?U^5Wcw#HL37&VKIl~#zF}W%9L~6|lOuJnS$_ank+q9%5;k9@FkIB{JvNQM^HPP!
zZg?C;+G=6nKq<YPBP=LEK@jp{IXNFZZj423gw69DPFo@Qo1NK?>XW#kf?9(`3qU17
zICB4;*D~hqPz($aOEL5%3~WH{pQ#dqQmHT>lRuAx^R~v7+2t=6Nn!W#(V<r2Rv8c?
z7NKUD%XkxHf^0(RmhFr?xDhEHstndf#q3AZE(~enH1kr(iNZTM>*Rxc;UTPS$(%Ja
z=yl)3@p@5Vi%DyYfrvYYU!&ST)~xn7eAZJQP0uz?^{1PxdaubN_7>RrbEmNTyRs(&
zRrh=(i{kHB6y2|EPg}_A^XX4wnQ*KB`0M_40*dqLuj{_}&p(R%=OSP={ExZ_tOwS4
z(V_Sosxv@@BJFqRL`MS}+;a&f=Nb&H-MvfXO4kUIuJ)nK$(^#$Ht%Ss7|4*vzysHb
z58z;?t}-Mb2g5Sx)He6~i^vrZZ1&T!jqD-rcnlAZ$#YiqbWM)G3=qnqY1;FHdM;7y
z&d>-wAjKxH;rV8|6M#)ss)m@2aajBd5yN8O27Bo>k`b3g_zIJ&vgS+^S=YzGQkbk0
z+3}4?yLdPz)&19>ho^X|{PE}1-^O9>Pk;Z<KjK3Ce|iy4t#w66h1UA@KUFyiE5*nB
zJ^I*-C&uz$!9p3;@od3D^jELC4mV)I4PqurbjPz(Vlp{)FPm33FLa-5IHOsAa%7Z`
z=OWwFXAPWRd|Tr1(9Ct-sk3BU)=3D)5|k6IM@KLbbUr)YDN@(Iwm^qej}yQJTtFdB
zNK91(^u7u9k9~0D1nvt^l{q;!^12vJXh@YO^M-&Qx{lxwn+>zDi^2D0W2=BT4W?uA
zy5LZT>Q_it2FFBN5zx^*mJaxNWnz_<KTK@?I(PDr?w-%rtt^R1RH!)J^Ml7%7ZiHL
zdLrt-xUv_Ng5gvs^DJ5(Y;ynkud>Z*ixw1ffEQFIHO^y<N+*rs-bIhduVG(4c*TH9
z9GiLy0d#V-txY=+G{%ezvY-1<uL*j0WD<UbfWZUkp%wunVQ0g-Ul_c^PX(!FAz`57
zFgq=TG9@|(rd--{I{S%h?17u`vZwCkayD3+gFj7WA}sa6$5BEw0kH8wLbf)>c+<(l
zgcX(@P3$m)7(W0m#Q-@Di>qDGZeCfD&7gSZOh9@TfUFC70&%OMM!ODIijL;zb)!m4
z#e2k(QJ5#5i}JBFo>VH?zJ6UA?&+CB-XnE#nrqsenaO9IwlL?jgFCTB{Bspsjh2(A
z%pvVf+jv?xEG(=L+Pmz!d=z{I5w<ZA-5=2XBz=2wFoCju=6?*r$`{oWZeQoYfhZU<
z5(ri6jkVxRTtvHKA+rl>T$GEeU0imu{}%0YYmrlc8ImZlsOB5Q+#aLvrtOFxtG@q5
z#WvS@uD`;8E)4qt`qG5ZC>u8D40)r1xT-q1dR@ap+iM!Z%0UkH>xb5>FV=LQoOq~N
zWyPPlR5x;|svpCJDaIz6*)R&~MH^sJW@Zt5KaCJ%J}$Lvj%Dl!nHiMJsTnH$fNfX*
zcICV5y0Hr5ZKiFrH=7_<$Za`_@jJ*}DD8X=Oov)_LA?u6>@4({^mx&(!FWz0O{v9D
zD6M2HBQq=P*LW@{d+7A=SdSP`@TXdeNfTPW>|1ap?LzgluO2#yh)p3$*7p+4lr^Rl
zi3a#1yUL*>6jz*s%x*r>9_33n!XT6qmKE%kke>VIviUB}*l71R5o{$Ky)1X`oZ}6w
ztdY?ZR%64}oI3T%s=9)zZ03?RAxJ8aN<vyNQd^N`AkD0?*0c+Mu8?${CpOW2vPi@Q
z<sdy=%Z=)PNt}cWxri5Xy>KD>*|Z$*_=&-Byh=CI+;FHMRsLR3rE1daS9snz;2}65
z2f@fR(Q%LgY@KWQd<^xp2y_CbXrOtxquU{l4(2&(!p?q8aIk6L2K@tunj-YUxB6TF
zXEG7C8|y)d32{C3WK<#cZ$f{)G-vGbzOIqY=wwB7wAPy_m-)f6b7ynogr3Oyvu8J8
zBams<Z}v4e95tN!r*?d($v)f&NYfeyf(msK?b3uhI`hE}xVw;2VC%J~J)J`B3}6Ej
zi-lqI2%G27-zge{|N0F<K=d~-f%eP!->zfc73Ig5AV7~oz#1%LEt;1J1Cx#H>_3Fd
ziv8tscdy+!_%>%CtWzQ99kG8{wjb#B)+5cIuFKRwGf39~5$xSz{b#XRxs!8^kkd|n
z8*Tkc+THFz_uj<jju(+}czu1%DkA$gJk?4|9A80-LDh!U99e?CGAx@0r_|h*l_HbB
zOK?2(MHj%{z*|q%6tG*wV#<#J1FOl{bM!Hvyt!quTNX8<X7GdIO`CQ_Sh3Hc9b(rb
zcnaxRoK$x0qykzL=6Nx%fUa(i4W=!g!H^6w+8ih$4zA!35zy;y_Ko-VAPmeB7NX_&
zitqz?X}URh9x(b*m;(V9LCp60X(}VC=(eb}=inG?c*W0S160NM!)w^SG7xeiM5b=j
z{CK3O?5fNX<UBI!?N=K4%@#UnFqi>5kyykD3z#G=nLo^+2H|BP1Lyn1Jh^iOgK2uZ
z)1I7S4b7uU<|u>7Xp!P<FM8~lhRHpj-$tVb0GTB7Uc##a;>#|V;|6XYdac2HeF`<W
z&x45<2H7x7W@xTWHMG}_l4FC~XZB&^I1gic0bD&sZEF$X>_g!pEPY{BdH8Ip-;Uu*
zA_yu@<;$^FKyOP2cA@9`$eAjMJ!=8EG45!wynJomJFpqPG{GsjR7tx=mX<il%jP6q
zs^9D!qQ^a^k#ji#2(M0qGj267#3o?WC#fkfstT>Xpx3@WSJpN;Dk{m^RYujA>t|};
zmE@2*ldaAnm+vK555-h#J#=b!bq-O$6aU46&(^5|3igna3U($PowiY*_^YL094>8Q
zs;_HtcxdP%w2P4>$Vywqruo1Kwg}LOiJiV^Dji1!s4G5(_RgL)O9o*-u~lLSn|WK+
zOQA;GM-PTswp2wSN#tlT^3^iStF7xgA?Vz?=Inmw+?I=4*s9CT(}8mOK#iry@fk-R
z)E-ZVmx!6eTiCEuWwPTfY@~V^<8iOt-J$95dxNpTV$JFFOPZJcdgZ$*x_~OL==N6M
z<^Y_ZcOFB!tNfG0zf4|b(st`Lwn+|KkdJZzxv1l@2a}J?yFXx<ngR1mO?~Y!-N6T&
zu$A2tZ)l;Uqgz43H4F>Sd&^}eLlPKIRpFb0*HMqEn9kM;5IY43484dm#o&a0?GczX
z7x{mL=IUmKTRu3wxUv3wa8;;eJG*flO!<I2sx(#q?z-2bM{|z@FS_{L6Y7)2t6pRG
z36Zd<9<vIsGYPCu=oxHTSyJLOdMKo>Vrj{!ffLTw&AnAO?#TotO7;b}n<LdDKK55E
zy5<!{$yKMP|Fv&{YDkpaT-6t5&H7_GB@yM<XNAT9ms-(sg2hFvnpWSHLyqn1T_YV^
zSN2>^E~&=vx=`nNzp|xX29*m<I-S-}Z@)5*7!Iwg_`&0^>%@=bV}o(3_`yo)KyB!K
z-s!0)fB){b_f9nc&<g%`bt){F=zxB)hhRzAEU}(c!>mq+i(UOVRpZrd4enJ##BZ0?
z$Ou8YK}<ai>>Cfi&*C;i?}@zn57_i&IIkkK*V$r_5k2L;H`JIQHL(f9?e>Sg(J{RV
z1J(93BbZgQ=U2;!nJL(t!nwU=X;@9O64haxY}}dK>83jphhbce>A(DXxGG29YY)zB
zADYi=%N&p)YBA}cTS?A?zSN0LjGc&ow>|;Pn(?o-V1bjLyQL(jt+D~`H_~ElhfE1;
zSajw-QN@<ydKd#i*0g1~eH?SkHK@ch8EA#asXpO@Oit1o`((K_L5Z5~22$HK+odZ_
zB~O-I0&kA{VJ!15D@y0-%&%>SJsc8ML(C05+C7Z3Jq(8o)VDYrSEgpxZ{B09GIX(e
z(RA0y{T90)J5BfTjm(@hapH7lWL-P<?v)SrX=$*5LO3sduRc6BTq;e?q+h{E;a8Qq
zEZ4v*F<hi#{lKb;B4Sa1sf&P43qkHkfz{+0M(7D?xy6LtXk@`ra=fFP8pIDM<!+XX
z5%B}2yoSDLAtV^ovo*Uso^03!>{;=&R1rV3bSV_wbt1s60sP`un9f>KRZPJQ@YO*K
zd!0+Zg*jU|1+2lSa6Xb?w4rt&tU9S}zymkJh56Wws3lU5(K%$*9!IZP#VA#D#T`|2
z`!z7KF=eK=fmZ<2TvsD5njU4q;;2toe{!-3`IoGp;r`{(;+mrk%yZX#gj?e)Rwixk
z6`|no5`O-OT<@}4G3HYq8!1Wu2>d+&@AH6DG~M?)ve7dcvXIrvj)%}K?Zu!xv56F-
zW4P4cjU2=PR1vb`+W;9dF?sT;dc)33yu7R<+bfp7Vwx1bE$!R47dB4u0SQH<SIAgE
z;G8TPcHyQ_1;9Xj7%dHkhYYkc%Lc&+rKsC(e2+sr?8I75P{noMd{_e5EI({&Ac7)0
znHRtLaFwWQWJUAJwxGH=nH;%o&M&@s=#Xe2Z2P)_QF1D(#_p~OwO40_YF2!gj8g-v
z)V1ewMUPy%&xN`;qiWYk6BH4QZYSf<g#mQ7u&3pbpaj8q$ED0`LsfR{1Gd)y=<#<r
zOwOIWF6NQlw$<yzO(fGkgtmocHjR1DP^;`bgbK*KYmnBfQU_Z7y5Z87R<}ixfe*#8
zYerE(8hCgb>{0)hL!HLE%X$9Q_&}vz8%KC@#<$Yi$F)-cYAf>nC2aekST55nI31Ko
z`ieJ@E64^Q<j}SEhDTafx2gwf(b9%w2e92~?~_F^5yv98ZVRT(skLQfxlQ;7&(5}d
zW7MQ+<(5fl3%)_KMaQW8{JSu@=R~IDnSDinU{?NatLOa;sBJS(N|P{*rj4B;!zOpC
zxT9yeorLkptj#M`r*8D_@DnHO-##$d#rH)47@q-+Wopkby-^BL>0nP(PTC|SCw7>X
zYM@6hEp&yo|A5UaIN$Q??{T!$nc~X6vFetUrVDd)G*9Ll6gIC+T9s?yT-h9!*`c>5
zWJ^V$hmKq24+@$#O}aJH+SivnjA+&u4N@+NFmjr_)YfQ{uGQ{*BkUdau93;O`Lplx
zG#}qNj%4Qr*LkL1Q?N0!sYAHMbhO}BVb%Y=Go*!1e)g5<%$WtKyJuiq)AM)?eHu%L
z(W+w3_Y2Ek>WN6lBFwjqIPw9U5w_p%Knuxmw1X$2%3!3t+gFBcXg|r1ziEfXyg=I{
zBw{1wU~g?lMZt;t$ayQoX)_aAmE<EINk_1CMjiO5%8T)vm!Eq}*Tr&@Zy~r0DNui3
zuEp8G=;b&+4&r30f}_`3Xr2X_qRTjO8lGnzwQLlewr;6<in)wAz11)Z1L+dhYJRO&
zs6JV~w;e~I76bVGImHcg0)^~zc<i(x;SP>rm_;$p+GOz;+fYNXpl1O~hbr|t!4h{t
z#BhEddO`RMizJbE7}v`5KQh3aFU=MWZq%7nke4gG_`n<~EEm&{B^9Rt>C1@UMG606
zRA!8AJ_iU8mp3)f2g4@XWrfX+Z^-m=%GV=kbRVg|!WN#X^bR%9o~~}-GMP$;b}z`$
zL5ZdztPI!4v$}gWIvGtZRaZ40#xe-OGH~6rbg_5hm!{G6GOD=-9?dGEk5CP4a5jF8
z6MYDz);=7OP!n!5X<Liu^(3h9XCmSrzg=aKtQ$CQxphI6X86U6MTBJf#i(NC$ynaR
zC_z`vo!g)=0P+7~&y5GeZ>A2CC-6@-p7+kp1}({4a0y*i#iAeA#p-8kYV$~E=Vl=}
z#&K}$6jOI%0HT%C{rfd&2?<ba?YMV26a41H0n~+?af5E<RiUq#1Nf)==!hozk~^r3
zlw-!Qx9bfLoCKfXhc4-BzMYrABsj1#0|AD`oUF(kXbFEY)2OGQ5`JJdhN(gf*Ag0h
zLMp=OD@144!%67E@NRX2R^HiltC;=-Pw|;zh(Xs7j^R8A3JJKezI4iIAS2n00#+Tm
zANjTKa3RyH4wq(wu(|mp3!UN|gw7t#*=o=dS@_OLdgSg;{n_{Z9XD-iYd?r0A-5Rf
zhO*LoxiUFA2F@W8xdyJT^KuO~N~d34ne=5#Oq86uYeI9gzH5Tzb^(py24LW9tm<yy
zIut6@;ug*!*n*vW8KZzkxm-ir`SS~}fz7N~M^Tb_F}cVV3A2_jPf^B-ra8WtLAVU=
zNDZvU0S(r6GoC`T-FZi9hFfugmyZs3kL7HxO7b626MubA?I(O~8i`W#R-2K-?}00a
zk8B=-An)>y`O~J6;Kew*)WDFB<^STqc1%V|t%Cw5s$a}B=&kY#loWA=k@}Rw)wlLL
z!1CUC)XXd?8D;*GwwEAybMBUwR?|5P=e@p17c@7?^Yz$h;@XM^#1D_$#?Y6}esu_g
zbprEI<p<A7nyT7-f#J3C^GEQSyW+2HHQ0rsc1NoM=uCPTIu(A8mw)(XxLu;J1l2^T
z`qQdqTeG8W@nG(XSXGS;w8W`-Wkhh@LnkqL_sLm0>Wc?>q4i6!bLL+A_Y<%ObykR_
zMvKZc*GQ+zzWM_eT;6~1v1EyP4iUegUzs#0fGFdbd_?PTn)Zu5+y3j8gf;x!?f}|Y
ze&CM0=6rD~!6et<#C8otbav(Kvefh`NV84#-jii-N-4hto<Kk6^`=2M+ESFqu#1FN
zxz~)&_qz2@6M#8vUut``L-Qvz!P=a1#5vo^V(plDa$S@Sc*0qXh}kix7)v!-{T;~r
zMw}?y)MApqGOZaz#HVq~aY3Pm=HSe}l8FbV)s02(!hUrR&yA==9h4`XzQ5Gf_V<3s
z0vu=7Cci;Ny2jzX0S)uqeVK#Kr-MC8ca!AU>b7nqkon=si_bclT;`tbtn%_JM+D+R
zw726;L%;e#bh<m<JW<Itz%dk9uC~oNBnr)nORmAm;A+Fjitk*vJ$=93ytO1EI7Hc0
zL(w*Qt^hJ6+6%W!Bl-P&F8p9F{NSWVxlgK5fX`qd>Iv3T7@`0&LNNCepFmwMxW=ys
zsL4wpP4*I)=hCRf4czxWzwl0eOjkn0{RRptJMo;(sh^iP8<+&K|L@$F@4hDOT)Ol9
zuy+)&zaDut&=_X6$LHKR4s!v=Ig6AkrYjXdqWKGTCD&k2<d$HcZv5h19~n`)z^Pd4
z#y(xd0tv7|*q&b(HEh_4Ef7QZx~nu1pksHpHG0b;`H#KTu3jUB`*YSg;A{S=s?hNW
znF2VzTKczP8UUd(48PG1Yt`>_dI$U=tGw*n<kMy7zr_$CzqKQLv(-Fijb-5UlonUO
zp8?>Fg*t&8^BVjrbE&xG$-%u@Wlt`aIy=+?Mq?V=HT2CE9EGur^Bb9vLv7&Cy~5uR
z?j7IZ?RCD;*Q8}E`K&Yc*d+t2EzIuM9*sG4a!R{0?<2OWgyxl2L3JzE_K4Ysco=r^
z0g4D#_S*E+WJJK(4J#iy839Cyse9-U{3u*FcEod^Iz719uKTY#F*Iq=jm%?JkD<GS
z$JrZTsiYiH<WY1u9x&Bn)8JU$O`1Q-8t)3~j|6*U(dB|JytcK<jFMhF7%Lbm>g<g#
z(qZiaT8mLTB$I`CHoRgDQ-5wtOd1#_f6hfg*#u#D&Q6+zks}H_Zf@zg7*g`!<AYR3
z?||&3-#~|uTb~VOP-@+(Lr-=DIqqH`mlu3Lr*bGNE(Q84bTej-Io?3;TuP+FO&p%0
z^O0)XsdlCc2^ovT>%ik<Z@8x0{TAnxPI0o)Uo6pIgjJ>bnxyY?N_zY54Js*x4~2RV
z6kDR+N4zBpj~C|N04sEgkzNr~HQ#aU*b<AJ-unKD%iBiF<I+M{y9d?n*LCg{%{BOC
z;U8y(&eKpip?}ZXHh$Lo*UR8hYsvM$CzI=0;vm($vNEWnJ|T}b`p~AfJ^$GAX<ss2
znB4Q^W%38tWw8uwhA2doW!?P4kC%(ik|gV*eBkjp)M!g-K4aj<6se_$kT0>_(sDir
z#G~=SCp9m|@esOo+S165$~dP(%0pR~6|Zr|1WMkVI&Uh70Oswv?ofT!Bxh1esp((r
zjq>sf(sbxgEA2MP3DH*p37L6&)Qe$r<r(|ESH0YWK{438i<X)`!Ul^r&_Z(d%$bfk
zTfL-nlCXQYx<)oPuT<ISY&-y>VwGz`X-R~<tgY_WrzXnH`eEZF298a9uZ#+IkDkky
z@fTHMZup$i;$z;O4JZ6eT86&-{mY$(T87D{iW7})FTS=y%%bqz=pA`^)1(~h4V!#d
z*7{~|pYEq(ebrRUUuC9+p@YKYD_QoN{=8&ocllyPDfzzX$qqr0X_h969w)uO%J#df
zyylkj-CJu$h8D^G8l@WE`i<>?r@MREiOkN2ighBPE2A?%OPrm1uLlv^1ML8C);q8c
zo0ONoVGVWQz=_V|OyR%;R;lD3?6(z|nV1yNZ%7?qkbFa*{B4}-32X0rqHHM!Y`zcb
zrEW%IabG}%q38amG1TeDq65r+>o{0urhF7*AmWQGZ&+*o)BxgGESv$U9b?TmQv^lT
zMQNSs7=TpfZWR$TT9)vu)Qf><I7+VFm${{(eOm75Mmw!HWBTXMF+C(&qfY&_wi7hi
z*KfzhkJMbrL-O5?$1Kk5tma0fqP9)&+g}a^w|#oy6SIu2j!c98{gLm5|Loo$`9?eK
zYQ;nSnvjTc;Xg-%BE({{uasX^U%X2A&dVh#ZDG1;J{O$Yuf42Ec)5SxsqzG{41H4J
zXMKR9#y>s=>2gtDzh|_O;<nSR!@4)DYbeqJdL$$Z151}c?&>)Twm?Sh%Gw&d$OWCs
z0nhmDUgi|fTIOcjkAgDctl9sXo{IL5pD)YIs<KbT>&$}yrr`Ir0&Di$_Vucx(2pg$
z451NM9^LrhD>4CiT2rz;Axos7Dyl!MP1IggR5WYl4jZwoD`@sMooLj2;Z6Z>`q^?+
z(6>G-i#xy^KaiDiA3cG(yA8jc_j+(MT7^oF4|t*5v1HG)oXZp`Sm!Jt(_WTf1Rb9^
z#Nv6}5+R8=h+4PETCd&NCDD$?(T*W8j<?P`zWDG^B~&az?wFFXo074I(hEf;<6mCP
zGEuwzWy@xMXNYc^PP&P!!S{|YX5KQ`eIJGVT<>EX8nvraJNvW6PHHxv8!o$uPP#9T
zQ#MLJ{8cu}u*u^@Kq57#HwVU$R$+7Bc^?zs1%eKZ0~;`trB*oVw|eKFxs9VISu+O&
z0EcPZ6#G;Au=vbjH8rR=lSe-w_KWc@?IvdXzC|8V!pFBDq>=}kz)0l>TwI!>U1F)d
z>^D%C%s{f!jD<$Ms8bT9(Cyf03@THtG}^+82UNXsu!eOFNjf`kCS0+vU}o!vP_wVl
z<nO61W7V$ab+PN++m7<xO(<;+K|-sX;`QSrwsWCF$Pub=uZYG$yUgbMNT9q$IO-#g
zeoQ!O74Fm{Tla8f#KV<$AFg~7Xt^;dD56JJxkNwWq5j>6`h{1QCLTFBwKT>tB>zo}
zQ}gBby3#03R@T@%*1Q3a{&dJ3mAQFQnaw;AsUxLdT#0#nSw$gb^Swnnjd*S2N>9cL
z!HtpP(x7ftNr@Tvz0{G{8T+(ZVC^!w_ozD7<Is}y@8;n!F8zS`p7Tb<G20T3VPqSV
zz;9+pf>0cPxIcL$lZOP0LvqS;DluL>82ud#Px?TUBzAKiQZx5kod~$l?Z)|9wH;pA
zFXqlmwij!;@^v(vYAT9gbxu&P4o$gbhXym^nLXXO=Jn84rrR?8$FJiZW?6HBKvM-k
zRkdhsHI7plsX^0j*z|CzPH+K;hM1#w9rJGi6REiKfgdtsTG)FqiIF{$>lxR@)Iqn}
zYEJ4<+EAd{8a0DT_1ZpRJ<Bbrkx{OZ33I%qclgoD$q^2|I>KIYMq1fMMmat>0(-^n
z3ES6nVyczX#1z#7^KxbG8<na|%QTSjSE=JP2x_r~Xo&Wxk5HJ|%>Q)+s(u~!d7ET1
zC$F>XjhuhVqZ@>tELCsj)sb*plxn(Ss?}V?@0j^_NT5WM>jj9A<TeI6^w`#IZJ5@q
zVWSbgePa8In8{0!0EI$$yU@s0#zLqaUYXJo1Fla^Gs)_=jzstyI6-a46U6v_GYUAY
zDk-~YLBoq~4g>jazBf`o?Y#FKb^!~X&`^f@?mCJ{kRyJKEV3>R5NCbOB9NU%=eY?4
zrHwW>EIv!!e2eXYZ5$*HqNx~d=qx0<WkAx$3=4)saT-uD8td5*%?gSLlo|8lK2tKI
z!Dd9hoRT3Jz-Mh!?=zvE4@`uRp%w*`0*63W+<AgFEzIrn-!T7kXH(A@LT(<V9Z#m%
zGy&CJrXvD-*y_Bw9h;K=K&|E2M_`QM-I}`>UVerB=)dmY!n#nU+_Odtl|?T;4vlym
zs(T<fqCIkalRuGZuvewnabflT9rC#{8?x<uYh<jeH!frH5kD6Lu&}>ppd^5`bmLpB
zIo2Ws#F$gb0c0+V`hWx$?~*jbzUA^)5jk*-(OO!yBxDK|ms72;YXFMm;hj-jrHh%P
z)Javk03Qv>Yyhj4sj*pW5$|^o35SLQD40u79T{hx#FHCL4^QVX{+}+;SF@~9Ff&n{
zN?}(W3*SyTKp}g#RJ4AbE1`wd-^IhouKGza9i$jAwqOW4W&7R7a0-{ed~4K>fVmud
z%bW-K9k|M}HuvdLXgbN1qBziel@Go{-TO@nrnab&pUK3IbmQh2bNQOw-}YnAlm;H9
zOEo<YdzG}q!0B#Oc%xpl&3c!|gja_T>TVb+)O<TOz4~735k!C5)}uE|e=@j6xvs+n
zw`|ut*+@GSZgznxxRtl-09(p*9%Z;|Z6|)SdDBXIP1cunz+?4v&<n)i!p>|I*GGz+
zIym4^Z|Hth*%2rOY}_6Lx0HsoH_=6shr~~1%O!!Yr}W)J=W%}bf#5|t%t}tEs6@@!
z40O!^IBk&2UGy5V=<{zL-9wPI?~?z=2a0vgo3n$QCa%xc(ZqUq@hn&7g`07B8z&_9
zT{4iXUZnHl;IgC?R72+ZFR;o*vx+S1P6un9cvjsJ9FZInu|MHRe((N-v7iVeyGh9n
zaJnhQ3}DHh-)$}0xxTVwWp36t0i7ZuQ(jOD5ag6>i~V9%e5gaPR(8oz`z19As3s<?
zxq($z?J-uLDrXa!@NCbYp8e&?O2i!WF=Q(PjRa?iec<E~8WHKa0a@S!YFIiAOegj9
zo*>ap>*|qx&YmNo!Im4mw&w8m$`4pU9Mb@B<3ItXHj=T^7OX&+MMF}>^4p&#a1{L@
zX0TjxYoKWlLs_oapQMO{Kr@!U*|X}qORXNPMT(v1m*PhZ1*QQz)Z%5!0?C@7Bx%C)
zdX52wOb2Fu^Aco;+t+dCJ2K%W?$A@DvnMoY61h`gWy&1;3aR)sjH)?Dl7~YAIqob9
z%Q6G4yqad=ZprHHn5KuFX|8E6QrrR+((Xu+VPivm4_VyZ#69`q_+i#%g<vH10ab7T
z7h1@t7_SkuNGJMf(uD^|(Whc5s|5uT?TRb6Or|S(xbkEK#~102OuTJraOs<`!4yj3
zs2a*cYsh)2OODR%OQ2;8l+Zn^_?6H`+-bRJk!gMmJdw)i2V#}o4ew5_N>{slix(Mz
zx7$_n(cKej7KBp>+vcz`8m6$5ORybYf;bO$*8i)$tBq>%Ji{-QSp{jWMJrL59~=Tl
zr)>?++M=Kahz5j4uu#wq0s;X&Vxd}{<yb!`--T!kD1s;;#xNpa1*L0Om=tXN2!<HJ
zRpO}y#HmM_c(A*!cOCV|{_M~8XYVgaa*{Vsp67n<=en=!x;flzSrKBXM-MM*KNn%+
z(|E<Inrhk?7Pc)kf0FciK~jR7MLQ4S%$2)Pj9=_G%Gm0{4&$qJzuG_Lj!%7o_!=qV
z*-9~<>Dfgxcgb*EwkbIP$!l!sR_Cv5B$XAf-dCxndTa+4)ke(x;gxbgY?}?p6vH*u
zI-Rod{ktU(cS@C?<xxo#6IO9)4Azq#zUR6}hc>=<I1w#d>jPJtAHmRI8{8zX44$6Z
z+)s6QWzoI~PWkLZ7&g-FX-v^#8R7+%S+05g+Ch&8d#C%%P~=+?;&rvnfIwr;<?isE
z=%4Re2&q!&ha2i6bE&ms1>)-CH{+qGhqmIhD;vAD_R6fcZ4?HLZ60@^JF2^8B`~5u
zatYKpPdR;evdqu#S`vb*0!+UC_co$jwwivG4{NChk_;_CHRXdhzi3QYIXgz$rtjZ?
z_>$5bXsOqH+L`XZJ)vVL#q9Q5Lw2Id-EiTj3glEpfA%;35&*<;bMknX<DOq=CyyPz
z;kqE+J|VH_vcFkz#LItsUSh!=ZBoN0n&h?k2^8vLr7F68;hw1;;@;l_Qq92}i`EdJ
zikoi}w>wMg;62hsDh10)?91w8V)P=BfPfJjl!iW`4?F2Ay&up~1?i&N>so7x$V~yj
zkldqhJ_OHA(P|O&i*Vy&PAID6uwqS(gQUmeEZ08V_Sc#dHT~|hqKLFp<$^F}&{qOn
zAZ?5q=yGU@!#1rwjD@D^y${LO!gm}1Eo9=9U_E!3mYSFj(O{)eQF*YzncO(!AYzA4
z<Z(Uqj&EbLisGKh<_;)wCSQBRj;Q>-01HiK9<$>?@eSP;XY<oUV~hNiNJnH2Z;ZT!
zVE!ALi-YHPgP>4Y(kVQ|Vn_WglsL0kHtCO6odSoDWn9?K$S5><n93f9EE=;eApVFe
z@gr9JO+4fEsf%07UsD$kcX&F;^<`_?lzXHLj!74UNEdjz94kQvxBrDZK3H9QOxF$B
z5vrztW^e%#O@rAhoJ5F&4;pBfNyRqRt$xlaYpz<_777=ZcZYv1%&_N^q$J)0DE$|@
z9zYrhC2a=O!kJ_*?495=iA|WxbC)cx97(;~<MmB@doSKNi!2x<`%`#41XjZqAJND%
zSWR4H&0F5LhWoatV<e!9dVl%Zv!X3&S7VCi%XR$?G>wL;_Cy6w4istt)7vass^bAa
zq;I|K*M|~73)rKdU0Hg&&qB8wu@DW|vKGaSt=Pe%u~argMB@u74^<m9d);M>SZ@km
zv4XM^hUazbT!^wDSbg5tM6EX|PZVF>Ml|NM-TdtvKgf!FtGh`iBsq-yy_m-+4v7PT
zf>A?{h6M*I)cF;64Q!%11(vE?v1N(HTU3u*aK7keGjS|y-+eDQ(~6vZ8x~P?hrw}3
zE_p1$q$4+rj;hQaNfASc-t;v#o9U?c(OOg_mEy=sg0K9irz6oY$i9SZNta8z`MSfA
zDV-sE7G>8Fe?y%%ms%EZerl;EhYoAaxk;z^$@qecemq(b#gaen@`vl`UhfZHC(=}Z
zi>g`*L-s7*9}9NQP7;Y&;uec*Jnj=i#L$ZF_^g+6znl(V?9Etb%+UIbacACDH6ptI
zi0CdxZgp{YMxjQsD`L5ah({Qt4D)Turg-}!L3Pg`lUj~SEowh~o6d2Icz8AhkOGGd
zHH)E_GYcRCIk}X`iz*$`SjitFUJUSkDFV9-8`F>g1RVK5=W^%uQi~vaxy9wgw9;CS
zTc&$gD;t|vPIzQpTwRPbv9k5X+NDp--(^Hf90rg@oxB&$=H|7l=A;8`F>T$6Lx4wW
zyeExrPu8TI(COx;79Rb+WW~(SbNYN6W-6>jtyNmfs<Sz#2e0?L+nR#A&nsO&?l4Yu
z05K}Mf^FMtSGkTEa=m{`Qa$a=-fpZ}`S$VNhdMUGX-5{1UH|yKkPlfCCedx&<lwpX
zQ|4AR=z!{z@76r7(=il1=NO1O54?z)1ptxtyEBBU`38rVk{tx2Z|tmPvCoyt&U+5H
zS%e@voP>qAn}E4rriQEj7SQXQrD#1TuS$>VVgUnGjm`ri7J&Yej%7!_Dgw(cyImT1
zu<K9zg`gYiX;z6)b*N)^hpqSVJWYCDlHl$P%tFwnk)_KvElF&oauXbp`OD<8I!vU{
zC;HX2ss|#UxhD8~mN;GmHz%PTfbC7XA7uu;Bs>ggi0{FBpU>$qo^4YK+DT`|n#ZQS
zT}Ec>c<u%~w00h-LDr1v9J)IrE{;7{3fZy{Md;Z)5+BX5+rWkn*vz#tJRx33-Y>5n
zmp1HXk!`|&sQqFSdem;h%8w&HN*v>T+<i_ebp4K^7qKGWv1vkvwLd$A$8*9T^6H8|
z4x)rdYIkPjSiwQr5Z}yuEa@=r;5|ytRBS822u$V(c#o$AM##V<L9kn<x8W$+;%o3r
zTrzY~WB;#97DJS0;un8!nLkUBUmZ_R>Qsfy2MQ~R{gA8Omnj(BEj8s6Bm4x}ltG2H
z-4o_sfuJCdUfV8;20K6hKjTH-!02W5seD$-@z!n8|LEtZ^PUIzIr>uJeklfc^zwG$
z3>m%7fAs$k#bErmhF6@80T+<FP(0K!Z;TLU?EzIR+_a5KL4f#4X$kOlQ6Y&Rn3U0n
zsAh7p6^sc8woXfHbULkvkmmM#5V>0B*_0|SdrX_*N1oP_KMZ4P3#GPXL*pqSZ7$&?
z{^f4NY6ED|v-xpQ1YZ6(Q+0Gc2>;Q{3JM)~PLLGC`k{5GdFWi!r<`^d)zo~p%q#mA
zN6I>CaRqCfxB$ir*@2m^iu|Zoo|&E(S6qk>8?gK34uqoD{}=OjpCNr~O_%(c`;Z0q
aKmGOc)27gO=FD2h#})Fx75Zhr+w~7y>bmd%

diff --git a/docs/dev/articles/transfer_learning.html b/docs/dev/articles/transfer_learning.html
index 3ec8ffb13..732969531 100644
--- a/docs/dev/articles/transfer_learning.html
+++ b/docs/dev/articles/transfer_learning.html
@@ -31,7 +31,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -128,8 +128,8 @@ <h2 id="introduction">Introduction<a class="anchor" aria-label="anchor" href="#i
 <p>Then, we’ll demonstrate the typical workflow by taking a model
 pretrained on the ImageNet dataset, and retraining it on the Kaggle
 “cats vs dogs” classification dataset.</p>
-<p>This is adapted from <a href="https://www.manning.com/books/deep-learning-with-python" class="external-link">Deep
-Learning with Python</a> and the 2016 blog post <a href="https://blog.keras.io/building-powerful-image-classification-models-using-very-little-data.html" class="external-link">“building
+<p>This is adapted from <a href="https://www.manning.com/books/deep-learning-with-r-third-edition" class="external-link">Deep
+Learning with R</a> and the 2016 blog post <a href="https://blog.keras.io/building-powerful-image-classification-models-using-very-little-data.html" class="external-link">“building
 powerful image classification models using very little data”</a>.</p>
 </div>
 <div class="section level2">
@@ -229,7 +229,7 @@ <h2 id="freezing-layers-understanding-the-trainable-attribute">Freezing layers:
 <span><span class="co"># Train the model</span></span>
 <span><span class="va">model</span> <span class="op">|&gt;</span> <span class="fu"><a href="https://generics.r-lib.org/reference/compile.html" class="external-link">compile</a></span><span class="op">(</span>optimizer <span class="op">=</span> <span class="st">"adam"</span>, loss <span class="op">=</span> <span class="st">"mse"</span><span class="op">)</span></span>
 <span><span class="va">model</span> <span class="op">|&gt;</span> <span class="fu"><a href="https://generics.r-lib.org/reference/fit.html" class="external-link">fit</a></span><span class="op">(</span><span class="fu"><a href="../reference/random_normal.html">random_normal</a></span><span class="op">(</span><span class="fu"><a href="https://rdrr.io/r/base/c.html" class="external-link">c</a></span><span class="op">(</span><span class="fl">2</span>, <span class="fl">3</span><span class="op">)</span><span class="op">)</span>, <span class="fu"><a href="../reference/random_normal.html">random_normal</a></span><span class="op">(</span><span class="fu"><a href="https://rdrr.io/r/base/c.html" class="external-link">c</a></span><span class="op">(</span><span class="fl">2</span>, <span class="fl">3</span><span class="op">)</span><span class="op">)</span>, epochs <span class="op">=</span> <span class="fl">1</span><span class="op">)</span></span></code></pre></div>
-<pre><code><span><span class="co">## 1/1 - 0s - 487ms/step - loss: 2.8362</span></span></code></pre>
+<pre><code><span><span class="co">## 1/1 - 0s - 436ms/step - loss: 2.8362</span></span></code></pre>
 <div class="sourceCode" id="cb22"><pre class="downlit sourceCode r">
 <code class="sourceCode R"><span><span class="co"># Check that the weights of layer1 have not changed during training</span></span>
 <span><span class="va">final_layer1_weights_values</span> <span class="op">&lt;-</span> <span class="fu"><a href="../reference/get_weights.html">get_weights</a></span><span class="op">(</span><span class="va">layer1</span><span class="op">)</span></span>
@@ -639,7 +639,7 @@ <h2 id="train-the-top-layer">Train the top layer<a class="anchor" aria-label="an
 <span></span>
 <span><span class="va">epochs</span> <span class="op">&lt;-</span> <span class="fl">1</span></span>
 <span><span class="va">model</span> <span class="op">|&gt;</span> <span class="fu"><a href="https://generics.r-lib.org/reference/fit.html" class="external-link">fit</a></span><span class="op">(</span><span class="va">train_ds</span>, epochs <span class="op">=</span> <span class="va">epochs</span>, validation_data <span class="op">=</span> <span class="va">validation_ds</span><span class="op">)</span></span></code></pre></div>
-<pre><code><span><span class="co">## 146/146 - 53s - 360ms/step - binary_accuracy: 0.9183 - loss: 0.1887 - val_binary_accuracy: 0.9669 - val_loss: 0.0926</span></span></code></pre>
+<pre><code><span><span class="co">## 146/146 - 57s - 389ms/step - binary_accuracy: 0.9183 - loss: 0.1887 - val_binary_accuracy: 0.9669 - val_loss: 0.0926</span></span></code></pre>
 </div>
 <div class="section level2">
 <h2 id="do-a-round-of-fine-tuning-of-the-entire-model">Do a round of fine-tuning of the entire model<a class="anchor" aria-label="anchor" href="#do-a-round-of-fine-tuning-of-the-entire-model"></a>
@@ -690,17 +690,17 @@ <h2 id="do-a-round-of-fine-tuning-of-the-entire-model">Do a round of fine-tuning
 <span></span>
 <span><span class="va">epochs</span> <span class="op">&lt;-</span> <span class="fl">1</span></span>
 <span><span class="va">model</span> <span class="op">|&gt;</span> <span class="fu"><a href="https://generics.r-lib.org/reference/fit.html" class="external-link">fit</a></span><span class="op">(</span><span class="va">train_ds</span>, epochs <span class="op">=</span> <span class="va">epochs</span>, validation_data <span class="op">=</span> <span class="va">validation_ds</span><span class="op">)</span></span></code></pre></div>
-<pre><code><span><span class="co">## 146/146 - 97s - 664ms/step - binary_accuracy: 0.8660 - loss: 0.3213 - val_binary_accuracy: 0.9652 - val_loss: 0.1022</span></span></code></pre>
+<pre><code><span><span class="co">## 146/146 - 86s - 592ms/step - binary_accuracy: 0.8660 - loss: 0.3212 - val_binary_accuracy: 0.9652 - val_loss: 0.1022</span></span></code></pre>
 <p>After 10 epochs, fine-tuning gains us a nice improvement here. Let’s
 evaluate the model on the test dataset:</p>
 <div class="sourceCode" id="cb48"><pre class="downlit sourceCode r">
 <code class="sourceCode R"><span><span class="va">model</span> <span class="op">|&gt;</span> <span class="fu"><a href="https://rdrr.io/pkg/tensorflow/man/evaluate.html" class="external-link">evaluate</a></span><span class="op">(</span><span class="va">test_ds</span><span class="op">)</span></span></code></pre></div>
-<pre><code><span><span class="co">## 37/37 - 2s - 46ms/step - binary_accuracy: 0.9540 - loss: 0.1103</span></span></code></pre>
+<pre><code><span><span class="co">## 37/37 - 2s - 43ms/step - binary_accuracy: 0.9540 - loss: 0.1102</span></span></code></pre>
 <pre><code><span><span class="co">## $binary_accuracy</span></span>
 <span><span class="co">## [1] 0.9539983</span></span>
 <span><span class="co">##</span></span>
 <span><span class="co">## $loss</span></span>
-<span><span class="co">## [1] 0.1102593</span></span></code></pre>
+<span><span class="co">## [1] 0.1102476</span></span></code></pre>
 </div>
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
     </nav></aside>
@@ -713,7 +713,7 @@ <h2 id="do-a-round-of-fine-tuning-of-the-entire-model">Do a round of fine-tuning
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer>
diff --git a/docs/dev/articles/transfer_learning/unnamed-chunk-13-1.png b/docs/dev/articles/transfer_learning/unnamed-chunk-13-1.png
index 2679c6251cf955e18919829ead7cdad24d6be2d1..c72625c5f53df9c985b9ecb4e37d42b54a46c867 100644
GIT binary patch
delta 370688
zcmV)lK%c+r<`#<f7LY@Kyrf5UCVWnXH+Am5ee<0=Ni!O0Mx%^CLJ~<x#*tub)@#hJ
zy=F1q&tR|@jBUI)5X_Pc4q$(_e>Um!nqZ7ESrVW;iW4+>=FZ&Q`OOv1{?RjH1}j=Z
zV14)b<KBMz?K)LmPt~c@=bWk+i3rxPhBd5V4bNP#A9MrOu!c2%tYHn`Up(^^JsFiN
z`-5)98lG;f@&5l$tnvQ1@Kh*z8XUgkTlbtu|IPdD2fuUk{fqhu7ko0%x4-55>b37X
z^zWiRTRnKsT@RcXm{ZrPd-%Tlk9O6=-+e5kOMCCU^MR8g0RY6ucinvBjW^zS<Bd1p
zJ(mE02~X_4?@(KRJ>IsRCGCd-&m7(X?EJSs@mGKSFZ*JEe?v{)K6Kv$$NG<#_U~ot
zdx<-HP(4v;KLmKj^8P(UKe3*FtO@>up6~SJp>BQiIsSA0_@E&o8=idNj@xz}YUM=P
z7hluxuK(z<`91gUU(HT`I&UAo=g#|2hW6}&4~KZ~2d|%heDR;n+sA4_ytwzy+wVEr
zRYYd#)=zxwv-kFFe0<kU-yGt`I}Q)**;W5r*p+|&m2JhKF<$dWuKN93!hHW<zI3)U
zGBZ&wZ~wI~x0cVod%95fz^ymlynESDGPmodo9;Ruow{L;ef($F2j%gJsyBZ9hxe-N
ztFN7OykfC`SS;4AdGCXRJKuFpGiZ!fgLN<a>%-=WN_$L7o)$bacqbyV3m<>^7)ST{
zA315`v#8OlU;4ERDy5NTsl4-ld}-;lgvZNN_C3U%J*e9DiAh`I{qzT)dfv_1MC*s|
z`A1}m9({txKk)hflq!02^siq!<=1A`H~r#8Z@Z;``^DEZ#Kgu;Bc(ui=lr+NFI&6%
zr8ivIXl^=psxtn<_uf7DF7fG!dv3aE_sT<*=sh3$@y&Gy!F|D>K05Jc{?q??L!r9v
zoT*av(%-tNeM+V}{IQ?8aQpV{=ZyI%&VSoo`Po$`-&><y{@C^Y^e_HP-z<FW`a<<T
z{oUVxyrRDDzy8CX`)>Qj*S~h_{sSL9>+WYxQxe_uhAqzIYd&}1M_yDaJm)>9IWfmR
zaCLCrTkgoRo8Pd}yW-vZj{eE>XaDP`cJI39a3?3*{^Tp_V?XzgoqW$9T;`wm);rY`
zleS+!1*|=JJR^7~BHQ_wU%uoeul|Wi|M?$(Ing?cnz#JZ`>qbQzx9qJyZ;@R1egBd
zp3^=5_?UXvH{W;G<((ftb5I@V*e4`yjd!B&7w>aNK5!<{fAF>UJ$%nU0#jt>9&Lhu
z{qKL^^ZhAR^!;zV^XGr%uXd~CreE9WUi6N;zwnv{Z~Wxno>V9Q<|iA$^Zw-U-giHL
z%ir?)FC08_*Lz=3sJ#5|zj~T@bM*bsb-Y*nf9XR)^WM_QBcFctg!fp)ho5?R;9dPk
z_qJO%{_ofQ<|p=k=S-_1{Pu5bcdIY>(EjYfv#b7O`+I8)58v{YTMmb0PyYFJg^{2C
z@b6sYl$tZ^){O?m?Z5u@g|qIy`k&%|)5Yx4r(QW8Y}hpIZhhm;(b-Ll4}9fM{>z1v
zSH1JD{?}eN<rbRL(>1qo)gRq<``dT=&wBTMP3aeYp&q>Kli?GSmeLOeo*BGT^qt?l
z;>91@{ngjc_|N~yv3t&<=B0o5rftFXA6u~$e)45O@n`-|a=O$LVCtdEJ9fQ)<E%l|
zqko6AA3UCsyx;zoe@pbk_533+l|9-7f9bp5`U9WuPiZi=c=3y#i~soQEt4<$z!77#
zCIE543!gVGCSLU73+ce#gGUY>Q?0MP^A}(9>;GgfyR2!W(+(&$^$VZeyXQkUI1kB^
z<<ZHhks`;(s?@vaMbE3<^41G~C%3=hrozshQ)hgm=$3c9_JhZt{ktE2?KyDld#Vm6
zzNf~*hAUop#dQ0h-~7_w+WoxW|BcJhK<P(6a^I2Ne|i1<NB`_wqi5dzpV#@Pi{a?Y
ze)m_;J8^h${cru|mBCppqvYg$x9wX=+DA`xXJ7WNzy0#vN007$_l~=N{`kW;^<t1x
zBB0|)N|KD8khIPZkG~cG{$ulQ9{kha{LJi+ZC!ZaWM;bucON~AnzVA^rR8sb<gM@j
zz?<Lq)l|oEcDjzo&(wD>=i2(LLG|V6iA(!o!ZV2XeZfWlhUkaw`A1+1*FV|>?+<?9
z^Zlt*^ybiCy#7}|6<+&)TYvSDU)m~$?@_zeA+ViRhsu@8$VkK4{@*|TwQqd=bASE*
zcfavP&4*|emC218#tM%~^TS}~XaD}*z2Exy@BLVC+xvd$fBv8T>G$JX|KPPBIC|Od
zeE3(k3Sf=1tG;g1`Q93-ee*kBeEDm>edS+$=O2H0+r(%?j&4|gUu2c3sj|sZb>`i3
zzdJy|&fWOU0|ukJzHwXkAwp>R){p=BXHQ)6hQIyt2d<m@o4@|o4mQ?JH8>yNFyqB7
zTdhNTwQV79w<EvV{BA2Am$Yw$|Bi0@{^36~?{=w`!Cmiv-D}_T^|spg$&c>#&!T4b
zrSJXJ|GK>O_3y}kE!+I+_}D`W{NrQlbKf}g@^AN*&mL6ziAh`I9pL-K`=8%>*?Z0;
z+P&|Qdj8>=3bRL>;Oi!S;Pd?{wdmt0CZ|}&qu;#wh_RLcKtJ&3Z+hGRe%l*A^1$qm
z{OAQ2{K)mwcmBoO|Lkx7;8j;W_w}D$eW)b0zVgmr`fne9x$}`D_KyU6;$uHCb>1)k
z3vGGMCF}e`y?#mu>wDk(y7%p_UUU7r17H07m+tLsJG<(Ye^X=l<+r@*_ik-o_mXvo
zKl70ff3bJX4Lc5e^f!L%oxlFJkGC#+!OmOHy!(X*PY2Iw^>e@dj+-~U@sEBEKk?Rg
z-~LcS0YCnK`QQG9U-;E`{qfuW<ePfy&JFjy?`Lm#<s07fuK)4-Ux4TS*h_!xhAl__
z{&#=p58m+JuXr!G;R@#oNxSg8@2Z<03j9arU0(f%w>@~^z=1t~_n#-+E8g}eZ~p1C
zs5$?e?|jRrRyJOHak~G&=uf@kIZuqKo!geqyu9dtIoF>xsGePVLekcFhwmHjAA0$|
zcb`G@%dc2?_?~|xrqbIUWrF|6kKgbEpYQzbZ-4ufl|h5UckW}Cz5FHDUcN;fy8X@*
zjThf=PPoF(y>h0#Fn;x=*?oJ(^WXN7-@9SLo4M+R8<ubU{MYBEf98)r^lO*m(L49C
z%U*te!}+D~o`3j%zfM=Y>e}`E48gqvw;u?u{ORkq29LP(WA(!4UO;=l{>87|o9+0;
z_x!I{jV|A{k6rfi>%4ovc)(Qh!*|?z>+O5QWk2)EpT6#lsxPlZXV++s-~72-y9IA>
zWZ%Ah`wsGpUi;=>yu|Lk>DI;RSHA20uN%F8_X}s<-T8{=Z<GLFkALQapIE%|H{Sc(
zKRJH*rkfY1Z+PA&9|17fcI{QQxo>~tuFm?G|NaNxeEl^qc$Rzc#v30n7yk15-}g&9
zH$Cgxt?~Ug?F!HT<qy2))mu(|y~io-O;<O>ljZEGz%zpPhwAJey#0XxoL5|b&9%>e
zJA<0-=Wk!%+x3;34!Bpo;eEgR6Ep4;V(QJ;UG>5<iMwXY70<ump+WVlmsR)^k+#M=
zeE)d=u^TQwljtoMKJVdr{t=lf?YQPqCiuA<F8@K#ccir=5Z17UHLT$ojc4A@*fp$S
z4Qp7#Sy)r_Ygoe?*6<9+4~5fj4gYa}k@pr>@Azz0db@Y+x_$BJlI6u_t=sMP2l>co
zT?iF+Qe!zWTdo(gOy@aEQc{@$P{NT(qRPdxHK6l>l<27Vx+^aE);+sqsp5b&jCTe)
z85mMzIVw=TBLYVNGnO($;RV<juP>ZvN#@K<)R&hQOHT2kbEaBH4{n%jSk)?jNRvhs
ziR!HfRyvAv&A{Oi2Lc(!I`Zx8RTuEdF{L9A08AnR8H0oj83Ra_lE@h{0t*1lB9I}5
znt(9`ARvy`G9cu#sN!**9KUz(q#tbCu=QKtzITwS`e=R7?sfVq*N(MBWE@v$Wr3{c
zGvD_IgLI{xNY5)e20^FFc!8IHrmEQ}7hTOn%3SxAKl6Jt)7zc~Hj^~EfA4o1^``6i
zT2mYk27@*-GB!-IZW#AgT1yIZE+inInwr&0b$gv4@U%5)o*HYo;7aFU4KfxKibDFX
z=j&K=#sFEn-A>{t@LVEfj5*RHBIyVzgeA*4?{|Ba$sni)qL3%K&aec3u1-v?-W;1<
zKQ=Z|EEXC24`ZHYS=JB3K^R6+oJHxrz55Ow*pG}6fFWzFHDs+d00th4N)Q7;0JNul
z>ed(`CByH=EfA1#L<B?vB%7NzzVVHXM&nUhvZm<Q@T8E3y=s5{j*tC=vbj6vEcN0&
z_bc?gC`}*OLsIfmMJ}Iz=$1+zaWIHrrRxR@3%x-nFO~|92T>d`?%Fiv4$RaX4Cfv=
zc2brK<)RlxX|HFyD^aCbW-iNZ7I@gG7W_bP5clspj*>S=ipJQW<Q+U@4;|{wpG+#g
zE6iZ~Mz(3(xnSoxM~@z#ThPJChTTUFr6B|96&uA9tDQxQ&wBNL3+CViUs;N>)LI}w
zO9%vvdB!M5FlANBka5n9<`&R0t1<vU0?D|}Qftu7RmmXdEJ@>g<K7hH-NU`Tx8J<n
z>8BP)+Y>WWGxy$g%W79y%Nrv#Lk0jG3C1RAV&ag6X<8d8)=IosV7bjECT5QwJj~2Q
zg4Tj^%RxBkJUwiGT5ICGCt}7^xwP2UD%-eWow3>C^3w6SgNKgo9mJin=_WBaxwvw4
z;qW=<Y+C89sMXB#WSqre91>b&;7F`h8=W=?$1fBrQM=pib^%zHWt?+YI6yEnGuuo1
zi;MHV=OJU06BCs}ZEoSP23{&OrFFBuHYsfwTQ@y9yMFV3Iko15>$*=*KRhKMBFAwm
z)k?8kj>72BgNIJco!Ilh?p!J1N@cV*RufsP0W=_3YXOFeo&hjG1_0JtL}ZL{&Jj<C
zR4p-%oFjk*w5TD=vPW&pnxbFBlY`97$#34yI$a1r^bNIFt(2b6DitSOjF!7eZESpI
zQ<Ev^^}<enyXS&K5ZWYlWl`zK6TU^+>GaEf;DIOy<>$TNM{oPHzu<8eW;%_viY;S2
zOu~v^3LHt8_me>{R^!zv00f3gsY<1MVsR<%n<Qc7S|L-pUl~8C;w&wG^Yi=4e&AWZ
zoG-7dO`kNKKCzQ?Q6Bg=yy`_KyUUAv_pKX|PGnAh91Z8%VQxWbt8!f^7RnVD>i&io
zU&%Pj7h3my`YT*BH}Du))K(w_j+?8T8-4NSO<Oih+_&%EQ0rc$K6miwc%!C(90#Ik
z`JQ{1D>WY-Z?zp~xfT+$Eb9%@TD_XvEH`xF1v}=B?U7CuSXOjd5_cLk!6EU2^5Wuv
zr&+;&4<4I|rvhAvnT_jwuh8v8ex;B^s0k|7kwUq!usqMXuT97EO5AlCid(}@o?LQ0
z2S_1G#+cRy88|{_dEV*u0>3N;swB&@ba8RPah&n-anB0`_W*EtwVP#%ELR$gR^@V|
z*B_`Hi+)L_JPFmH9j~9*w0*~=8_wC{21O))K>YV4-aiT+N=13fB|iktkTLE^VM%MF
z3W1-rV#kqSiL9Yh=`aLvO3{;rVOtnKrRW)B#uze|F@}gkbwA8zB{G24LHU@?SX1<C
zcv3KN5*<7SWv8D;mgMHqNpdsv1b5009$XB3FKoxD?Yovq%QuWPrkZ0q&gXmmg?^NO
z5%2esewZtzF)$@xUcUX}$;NE86by{X6O!aIwB$3_cL7b7D-<k^GRe7H6dN~;9baf^
zCRV!bN~vs|!DMCPj$03w3&o|?u3%La3(qU}l1$V}1B<Q2)lBQu@=SxL_?|C*v)jw`
zO6wPX{&^)|ef;3up_PmUfzY`tSixj}bw^;K$QdvKnVX80CNm{#9fndkRvXl&>G-+U
zci*vZZf<^TvNl$AIPneZH(JmdLjomzQK<PzY=acE&9qU%^^}nw3szfdFo;W~;;~}~
z%Yoky2Q!r#BRkR<&(cC1_6d^eSi=}&?9+A*F9C-*kE1N)4%Yn&a>}(zv&8j(f_i-{
zHEMogMR*>eu%J}jV<|X}0}6z2ip8?hIRGVj-W$Z@O}AXAH%G=nlN4U1T&q=^LdZDL
zoJlU2i&980%Tn&hMA?;gpRCWdA9eFVHyIt7yy(J9H*Y(i2@c;+UHpB*P$Lsq3sy?r
zXw)`tSfBRVDvNWYBqvL#hekYq0RRnMQkIa27(4Z4hlV^NGD~EwHU<%aF){Sr44piW
zV|c77`ZYW$XiJeJ7A?#hyNr2IDkB$zBwajmyjrQ$8s#igoMg39l(||zI(F&Bmw)Li
z->4Q#VHmN{<o(RGrd0Mhql)kRr=R+%)n32odug0X3l+cAR{c(rBu2Y`E`ZfwnQ){b
z+Y5WzviXz!CV920r}L~HRP#Z9x><5ux62s`l7ZW6g^U-{R0A2#cu5xGB&^oO?DXWF
zckd>}9qS#rW5HwT%1Q*Ohhz#aYu3t+OP(R_JD8<LvZT`~c!5>9L{=^qk|>srH%Jo3
zh%6S06{AI-YL)h-WQ-MmfN|jqttiYe9`ronIHDxD*KMbY(t2a;=&@x2A%rUgkD^|s
zGFtMyrR8qH4OW&9Fc+*z3#EWM?aiBJEj>+&UK_LAUMUqTORFn+ZXAXPO69}_WLgy~
z)hP$Vv_D7_rLBbwExDeIldiFTrBr2%4NYXNh8w7uwXm|ZeE86R39VR`+HjDS%YKr^
zQPdl0jIQ6Xt`K-P-}LR&WQPtOPtz2P+iQiUXPT9XD=xq4ydB%QBP?043Yha)&40pp
zXz>`D_D}6vGL%U}UyKl?QfX>(eCc?7ad9!~C5&@KhDg?07)F3j#~z5tS_^dQ_0W)K
zfShvxFvf7k;EbPtBgQ=T5!!2tehp6w12iY53L}#nPTYK(>1DaA=6fsX6f2c-u~dlS
zC@AX@tOOpPtc>#rZo2WV2M(N^nHux+xXxgv?jG*+!Sh_F^!#UE?K|A{<=nzzrB;8*
zb=O^a*|XmGyKh6z2wBnh@+@`>4lyl}p=PPJB5+CPoCJD*j>F8%%*f<OG{4a8wPQx&
z$l(<fLIxu1<ukS7N_$00TPQl^MxotW8lP;EfrXQ;e|+F*v*gdG);QQ`7JVNrWrbo{
z=aFm=db{?LBjN)Kd4Ev#D$JymC7wWI@}es&u`EkG*9iyt+`{sP+2-gdrg@)$rFmwo
zMTfXA5NLUSX%4{~8L9R{v#_`d$QU95#)G(CtF5#LwF-~3%oC*~N()|L{f2cXmzD`x
zx8H-O%a+wTpI<t*(c9`dEQ|YL7AchhLnaU<5ivB5P0-|t&Qd}muu-p{=keJ1SkUx{
zOp+vo<hqq|+2u@TS<+fvZ7rUtH%5b?<a%Xg5r_nTW4&&t-|H()#E=u$k#<Z$MG!7`
zeJpH$_RbyKcM8`-%On$kU`PW310WzEMmW@NNPqx{L#;}Lh!y}4IGs}4!~dEH5D@T`
zgj0nr5&$3o8~(?kavpx*RCxwOL$mq`b>{H%5*z@qXlu3Fx{VvUak!eqYVKG(=v$?Y
zMzTnMB#@y#y=xZ%fPf(f7&>S)5z)}tx0W+xgrLC!4BhvPuUYhK_&%dpt0W_p2crJ6
zOU;qy{k>jlnGaeekT3`wxqfV7esQI{uqraPbgVNzwlR*g@}TwhpT2hEx!cliYuCL8
z-}$+lw_LpQr+(zeRZ55V@7X%q{P7!p;_7FAT@@wiM6(esFKb59VbXE5N+obEV5!}X
zl$f2Ja#&_f1PpVNS?0~mj?bT1I(PHv!a}>^2gOp*jZ#IFra41_-1wfmvf8#pg@O;#
zkwyQ+@wiYcWH68niEMkNUnrIjA70fqnI?=?mKzogavvoMQ>#{^u+KPZt*2)v=jRuH
zWf3i@TCJoYK61R(%iLG})C-Rsc<{jS1(Y7@+;`kOO-HLGQ0`JE<6r^-Mb9Bko^VXg
zd=RUhb~%^?<*Y8p7P$=-Fi|eM<#IVsvpjp;&#QQH$n)&3UEk_<TfSQ`DnT?%T4QaP
zWVJ@4P;Ji7XPMHye!TgdjlS<MEiL4KDqdWSwKn6ElXE9dAVyj%YteNAU_4H8#+`a|
z%o2uSj|uR7U+WYZ<yqG64;)$aoFdblv>zwXJ+=_I{?5xUIsej&{Avl#uv;16DgBJx
zA`l>vVE_QcPH+4~Bm_h<0HBSLOlYMD2pQ*`5daR=?`fF?L}&pR0s}y_8~~7ifB+E@
z5C9Mm=t(cMj4{vi#>U3BB>Db5`{tC+(-eRMvf*-Kt+nRVT17-ch@qu_YIULSE>3XT
zn-2iP<;PkJ@EH4E*A)F4o)i{)-MdcA)ypH3mt1z_-~p@@anOmgv=o#a-?1v2JG$Ui
z3xS+moF5GOEY}^0&Xuz>yt|@*dkv@8-ByHarpI6R`d{r^BMO1Z;9YNfC%W7fqSJ}i
z&(0iPT9!f(fyz~}Q1m4C181q<at)4;1nq7VDP1m<nzee=TlKxE=172~FW5Pqgi$hx
zT%(JnG8yB!tQ@$-f=><!DW}Ir4<1=?Bp;tFDBEARt}LaqwA5`*j&O&6m%#6~dMA$c
ziq#?$+@>)vxS29X7upk}BT97wpFD0|39jq*I{nGXjjhhgy7i;&#r~&1`PKCs$HFu%
zac}+j#`fw-W*s!#=F$UEbqO*cZ8<WJ!RXXj;M>W@=G0&(!YHG1xg=$hr4eJ4q<Iz*
z0Do*Io(c#@l<WTd^5L<6=BVIgb;=nE#{trVxb1^=d|4<Lu+56)YPnqQg}stfa0A{S
z^afGKD+o97r1c%Gt<@SV<Gj?UUjD2r7fzn|*MI#|J*>^l%vd6;2`m*{mysisgxu4q
zOgfQ{^z7!1+b_AKIW=wp_;+KEma(CdHUtK&r998mn6xq~v&I5{(9pppxbQsRS{sLv
z%Cdg1pJquRDEP&|_5DJjfQV_9I?_Qz!^m>vTrl85Iu<zR5|IHD1jxt|f*~RV4vasE
zA67&44FIm|SF3d*Oj9LYL{eZ*EhgWSHu_zc4krWz;==$eAToZ+rE|*314xgx7Og4z
zH9RR82FK#=gZ%-29f5$$C`s~^){l>lROEw)mSm~cw+0teo2G&5<uc1`YjmOzxE{o=
zJ+v~277uj?S6_2&uipzAl`KilJ#SlSq$vc;`(0$vZmk-tRUA^08>3VTq?p4^&G)A#
z>bcIB<J=EQR>iG;R`Sd9%PYbYNt~$Ix}H<ynC3ZSskYpI8pVs%qpfxn$Llt%ziZEd
zdU>Q!&Z<psa$<COZlPJL=4rUP(g!ZcYGyfEM?z7hsRSoW^kASYa$`EvQ^DnzZ@cHN
zdqzhmPR!32iiMSx<yx)8n2}PUb&v0F%R(Vc(?f%K4m?Sr)9I_+<hj+>2jq-PAe$<k
z<f|JeDvs}e9A7wE9UoCV<6i6v?6y~@r)QRyTPsWLEPYBJzxT)xocjf@-I*7HDrGNA
z2Z84~LEWU8R(fTA1prX6axI9`9xyPZRhkgbi&b&-#Dm%ZfhZQMCd{H}P^^?<m1HJe
zYM(4M%3IFe0+vbXXU5t*V~o2FqVg$|S!9!bKk4;<$H%5FxcHKFn>K(Y#_Vu~8#-hF
zjW#)DNgDS1omSdwM}t1I3Uo@AtR+jt_|W|3)>xG*uvRM#)((i6;J}!4q!f~fkRj(B
zBtyw0cM-Xe5{2&s1?d)?pv0x)I-c}gE;s-qq9?Pvbg1ZE*KJNtoO9m!`wtyl?Q}G0
z1cLy7L>w#tJIgF50z?o(AR~r=2DLu*1QLPlp@ZfDfV0O|<H8UbV_K`r3k&m0iwkKI
z16gB?Aw<RyIU)ySr`8c90wQFF7&49k3<v=bIRF6}0Ly@YELp>lRu&e7kXfEPjswWJ
z;EW4o>~zLIV@x741crzV2@otW0w4yNN?!PXqURf{?!4npNVSR*WM~ix83J%1h6pGC
zS*GGz)ti{ABIf8wDP$bSS&}+Zx~?OI;D~t=SxZ97Zf{VjRE6WVJKZ4gqCrm?on=|Q
z-WVGn@An6T{vZ&t5EPOm<wB%MS}7HM*CngFofhYuNofpOvbi>laj@1A%ov;HN=TW1
zWf>q7Qp&)SQXV;cWPEf?WqG4oEe3_7$Bxw-l|eX&qNr3ZJFX)fCyF8qgn~y=<okY>
zWsc)Gu2Zd4=jZ0o*mAK5maHXU=obTIOvPEXUSpgSK(SP6cRSr~chjcL;UG-2bY*3g
z3(k=xQ8dzM#HngDMgrGeUR-R{>#cTwYqhm<*>hh0?;o~j!eZEqr6{HSysP43;JF-6
z&d*am62*`Vx`?*h$|}`DsTMf(GzwEoY=CrVIrKMOsDknS{GIpTw(FIb{fO0ib!BPu
zd0VYk*5V+J?%T8H@X_OF!I20yFPHtmb;xMWn2TCP0}O;gBLr~8B==TVd*woZfiXAN
ziBc-eO*04_p*PLWKJdW(>(@=R!_aeGLSkrU#>WpHJ-TD(mc^Bhv7uMind&#kecMXo
zB+E49xoI{Vt#;RHgMzD^040$!Dsw&d)o<Lcb6u=3hC=5QnzY_1)vMKw>ze!SIl_!!
zsWjGKXsb9$O^yaQ0Z1k}2MgAJlGR!(q&TOQkWDsfe$7vG>=oUK>5??*B^O@xFaP|F
zG}V*MF;_nIama@S5%hZfBu#_B8-z+_iQ{^%@9AC-7%^_*Bypu%uUC*+3%WVhIJt0G
z>6A;z^Ff{y7Y@tWAnH1<ljVJ74M}<Q<o;@*x^8wV?&Ym^Hzz9lrJ_@R)UnnXs?_9z
zOs8setbX~$m+!pfLg6|b5Km>J7_AbWW`jXH8T8`zvQ0yq#saj*NRR=m5m0Mm!H@xC
z$GUa^03ZNKL_t(zlrfBHa0xkMhKx1VSOW%-p+ipEOC87MoC_gPoQi5P#z8m)JVDk+
zF5LoiTyp$krRtQ5#m0z#a03f~z<?~^Fx&k!5b%sc!yeX^KXkA#&KaXhwKO_02D>rO
zv=pb}u|!~K_`B_t=V<s;C~GYO05U}4QWB9d)>uoH7-tOGnOmzK1tQZ~97o^&&bPkv
z?VH=Jl{|?VBc-fi0yyW)LGCc%gq(3{Q4@kiNhkmX0`Z}Li!mgBBCC;<Km%aVYMrNt
z9@y=9UOyZZibdhLTsm9|E;%AHE*O&xC8Fe9GRA=sAtw|TupjqIwP4G|+muN^{@G9K
zZscpsl|s)mjB#KR3_}3)fH5iU%xux~<MD|?H|k006oR1BYF8?iMy<+`Nyb`B3%S;{
z>PV7iNousQVU(PIzwMkT>b2VKZnt;gg%{m?(<jHqCakq8j*O<|mDZ+Bn~gCZ7kLs*
zPmCSd|3FYIBx#P0vtiTb{re9a0*9kY<9>gze#6EfC>%a?Xrwt-ES0$L96NR_j-sQk
zS1(r=gak|7ZpVV&vGamKdpS$I)pn~=tEE{AoO!<A?RLw5<x&(yd7iIdzkcy#q|&VD
zd4pD(C2750Ke@P29~p_#glY1FAkCFuEW~M2uGf~AmzS3BDFlU;)~eE)b3Rh7EiWwY
zd!V0aIPbjeqxIUL+sy_}r`L+&5a6j%^v00QohVK@$cFE)A8kJO;!E?Wd++_b=d(_f
z+8jW*{!%A@6s{;0_*kXHTHQ-8x#+y{4H`TZC1YDIckkK77`FS}I8A_vS+Hc7BW}Lq
z&VHI#>$OzH7HDj;*>0~Eip4mJeSmG-F1qoyZ+TYQNJ-&hXdETfAh*WYj7#pMQ5@%4
z(J382aPaC&FX%=C6e3(&+Ihjb@u1xtZM3_K^#Z4VT!v{Rgw3L?RIH>K={)nw9%C#?
z;xJAT9F=BB!+dCuz@|A9!gGDrX?Kz^3knsZQk^Ry7-fbtN0E$rrqhcwf$PYU^pGLr
z28=-hVq9BDENV-LKmt?9iX)XMSIXGsRvWWC`qr)I?bw#(Nj&IE=V{stYOJwVdmiVg
zfrONQ(i*VVYORvA&!k`+a+BBVBWaS?>XngZW&gqZ338XamI>)|L}rZg%brmx&r>M`
zYHKLV6)QR(nK%tQM2<xt2qgt7$>Vl6?#9TREnCjpamj_{YS|)@1xu9W-PXyuw7a6B
zfyolIid*GSF1bZzt+q&Dtu>Z_7?Neog0-lBjkRb2^PG$&1mJ=r0%O3CN(CPP5CKS|
zW&nghMx{0nhdzAHn8{HiO25<#UB$doW4t;(EeZucC=wtM4u4{t@h}S-?reIbfbG!w
zHd-fxUfl0-B3C+tUSAdp0LGXcaTq5CA|fCJMBt2ZYXJ#?5ej06#(+W8ISh3r=bUqY
zj`%pTZqhh9@Zi3?@7}ezxJYCGnE`8=AnqdPz#L-S0w)y6xCJ6aV3KiREkhsz0we&7
zU<u4H%)!Yr#@d}OW5^gVhKM93;{p&#fMx(dmRSp+5fvkeoGEgEG2{e*g@Uj0Y@}Md
zXZPLpk&%^NjGPH{i3?)LfDwU3INh;-))mfp!%zFm9;&oA5E3h%Nc*i;9(B9jYNa|_
zAD2?DtgMzx)z;#PJU7*9%}2Si+7p^wgHq9^iRpf?AH{K`(y8fnmhtiFsYCk@_!YNN
z8S_eosg1Kl(4RZO02WqOON}}g637O=w=h3nA0I0f3m0B;;e-436$|A`HQ+jbE)_Sd
zuB=W?P4;^|t4*;qn&r0L?$+xK;W`^OZfdpLVHj$y1LhWMqn_u5AbYx3L^sbP#$5za
z2+Avqtzw~4E>*KkF($(}1nx@Di;}F@kLH(JqvK;-x_8~Zt5hj#W306*%Unl}j5e3M
z;laa)iiPdc^ZJ9nMf8iM#~K)aj~4>zl{Z{__dUClo|&p`{QVDn5FMT_?Rn0j`@VC_
z-5>n;zs5Os@@~0QMB;+MMA7-RpZ)K=-TuxuzXAz4Uh}e_`!83&Wct3ndymc^*|=f$
z*s&visXWqXe*7Okz30JwwMIiTavg^Ua&BRfBLm8f0YY~3t@n4jl*Dm=B?x+YQ1bj{
zx!mcrgCH0TTFudVGKjJ0C~ISF=9W6vP|xJ$X8ri#1C_FunXp=NP1N(8#E?+dH(YT?
zbEZbBfd!hH8NcPW{Q!{WnFVwu%d?EEB!EsklDw={9*4O#P%cSgDQH-pZ@EzD_Nd=a
z!3u$_T96XJV<-i;)^O>6Dx(vXn^Rs@1Vp1@Zf<^Ls>uatn+2YqS&CJ9Vr5R{x#Y||
zT}R9sqkDs8DFx#$qH!F@b=~2Dhm2_(=PCnWd-k5FRU5PGrva?%%5Ja6BzIiTb$uXa
zj3vY3C{3bx{krwFTK&Y*(JYEt!IX=hJ3FOv%3{o-oN3G>lV_lRKyKc6&ed1HaBO-4
znB-uRB<!y&E+0P-t*&@RRit3t5+ly6R*6bev`7|!0FeL~DNAJS(6lzj5?DeqgpAWL
zqKjmS$RHRT`htjv0S^7(1VdSN$O)4|2*FZgfN|z0CQjQ6%r8}%lcoB&TdpdwfD8#Z
zA`lYUQ~sq#hPBpz7-K27lxKNlDK|z9O9>%}$PPX7c*-S$r$YinU@V~lU|=l)kmE3-
z;o@S5kPAkS8H_!+f8WhF-gxxL;V_CE#|0n&1nv;`kaOY^fEhqU4k&;R0}|W-2Sfs5
zXpxA42DAu<fni9-XhdkWR)izC6pj=?mW&dTGXZQE8V3S@G7LGKq6G~{6vz=7L#4Fz
zc-Ze<aNhPm`_qq(6)HB&Q7}DR_XsURKx7CU3BXiJzAtS-l$D95M#rK!@k?dLb&@1E
zguvu#tG#Y|#^J7m^5PvAIb8a_9|XZva|{5uV1*z^li2tDn{K{o<JiVhwZ6L2%EEYd
zW+slpa;dz3+FH7B=gvEK?b@($bGJXxmUiqoe|e=fzp!8_8ylZGd2-%!9h+ya;GLC4
z9XjK++S1Yzsr>lSBZXpdWMpLDgZrzMvHl=xcYB3Gku!Pk-S=(UxCsoPMY7ahX-nyB
zn%z7Y^jpjAso9x97#ktBo_l^6MK(8p;7F%FT52?Z8oT%Go|u|UbJCVMGB|eZ#MX1S
z4WcN^vu>xmw7fbyHNDzuRq9RG_3qfUt6r&)wLu{;=7~S?;Sn%7K6c5D4Hs;l{JX#Z
z;;+By^{Tzx&w8~Y8*Mhv-!MDdDEE)8ZmKt=WMwYaH^#2qy6L{#?_FA$*G8LAU3KMo
zP;iZZ(KmhmGm(ysNophG>$aWurr-I0R{C)!q{|Fv#!As@#jeK#M?^}STobY0U@=V%
zGo~68weH86O_y3poYX|wmBponb}!Yw%km_y6zg$fg<uj{rC3gokTP!a=13`vqZo3=
zB(tE>#2}816qCr*>&?SQ=1Y#BQJiWkgag2TxnM?HM|xW4E3JX!XvPFu)LCb|S#Gy7
zmFq#btupI5K_zfK^gKz-5*HkakjdOw1fCn>00@i#l*$Tzae8c`8~01TA8^+fE;w2^
ze2}IlQ{{p$#M2Z@Ge-9Z?M9=ab?(X{0OOo{o@b0@0Oe|FJ{$ls$B}Uy9Y21&-!C(N
zbQrn<T@DVj(ijUCNr5CM%D@3P5P_p5TCBPy;mE%3XW9r$X{vQ(S&kf8W2CWt=Y?Yv
zlb#SdPdlsgD|1H&tBa7di^vKb0<Zv-XQ{Qu=)@YWl?EUL<eXb$BnX6I8YJUnfh<@s
z+5!SGZY&DH?9c&<j3HVg01ylrSOQ{yr*@%g0006~paIZzICEUz2xFwMD$V+H{jj}K
zsE?M$#$7)E-~b#DwbI5I$2sL;!?SFC0syV`@HK*9D2IPt*W;X9YfnX+>D1!S5Mijq
zhxtB;)))@VT1#Y&(Sk`Dx_5wxkhFrwG6ndK?OXRfxHpPo#<?LN<czz(J>sr^MM1<Z
z<Ae;55o5>^kXz)0TyV}28CV9Efkx5*7RgG6WUV#2wX!0)Kp>;Flp;-2Yc)rU!X=ad
zf|vv*41qRA2nW$J0tUbsNWqhAaLqMWK6qf?$&<_JAa;e5YR#nsj4@yUXc<EWh(-!~
z;f_s}q6oVyoqh;h4x%(nb&^?sZP>`zq{_|WiIal6&BjQU<Tlrpa^=LaW61f$x|zk5
zmHqn<D4hdAp-^b8uC`ZN6OCpd{XC2*H<Oc-fg?FtFvjDKQMMEmtEK9RV<!rpABWNC
z$Vk}j`Hs7BYI=Eoe&N`$ky<T{RBN?YF4nS4Cy6q^ms?$9VYSn9T$c-fCk&(2c83eu
z>-VE5W=tq+!#GvOgmI#c-Mn>cr_=9rx|5Srz5W0YlQf&0oI0@o0GD33J1~ZnHY>}k
zmZ&*0s`3l~@+=ozx{m94{>13G6i%<#wPb_9FITIrPG^2`5jekK`-`3u)zE35xcRGp
zW%8t6o4w+y7a5<GN1I81uf>&0SC5}Kd?ZELtW~w9JTXfP-Cg$|*fY0$pdB6?<i;&e
zt=q`Lyk4z6>xySz^_*u*t-t!sFMsMYpD-Sx#m*o`DV52k6hgRh7<2A)dU3CxYGcAw
zbHh2xoFGm!CM>vKq|gd)ab*xIh;w5IxNy4tf#bR1Anb>o=4hpV(^|2}Qe~oCA;WTQ
zyx)%~H)FMmAmed=(8&y`O2sdh#l*x2lfw1=!6392j2Rjx6a;d5dV<I%Q6|t$6Vq=e
zNu-LNUl6kHITZ)%97+OxM{r|{o@2EVjA=_2Ayzh3q%Cm?=bt;eZe68Pa{Ap>&SaeE
zPIo|n3|TP<1ZS6jUHX!-vDv4DuSZ8OY=8C}A02B>1a5&dXQ+nCm2#XU$BrK>lnPo~
z#)RN*n&z%kOyX2&Q?1krLD3pgxz@^zjZSRae7?h7mFQZj=5QBNny8NX7!<hggY>}{
znnkOFtQQK!Dy8zJmtFd-OD-!HN?EVHaOB|f(FaXy(Kl&-wcr$7AvhYNbe8BmGfG))
zRGuT`f*~T2QOJOXi6;n#ELdxeGR88-fQBx54X3>M1V$UJ$Y?SetOC*q21o-KM8*i2
z1!jQ}DU|~f2Lx*wX@`M8V|5a?mb#se%g}KpazTJ_`gn#j|2~A_<*B3^k}E5%qetcz
z=9hD&_~~<h4v5H-fnf|3h{ys%B5TQ<cGn{S0a!q7jRB%zIud{-qL;t)W#w}DQJ576
z9^9QoA%e{_%^jbZuuKwi1U}q7fN*+K0t3X;&M5-07OVlQffNW0U>I0H%ee%Dahx;6
zASj06z;!*xk&Y178nA%GjA5?p8$(H$`o8A}0Rj|%gz*broLS)5PvUyD+^ja3!+O15
z+4ohHJHD@m{YE(a-GoGpp&!U%v9N#deUfEynh#Dc#c^71jzwWCy?|j-D_2IU86$S&
z$T3&Q!1Foh-NC@GRd(I~KrskL#>V#S+viIDikH1qnY`cW-0_`TFTMD(lH>2Xdso=)
zSVoS2%T<zizV8WHsWtW=JX|hSvLtI%Ytv)nNv~V*gDgoe+Od;}7EaD9lUEA06N?Le
zA()t*-231@WsKuGgCw<#S1J|ZdC?%8o|%clusJ#!Mxim5IPXSLqwZPG@4V-pnqMvj
zr6@|e{jfPUs@lCIN_TAAzB1@Jjx#Yn>IXr8wOknt`mLqKAn@}z-Lz@rvE#>Um1?Ke
z>a;r(qoYep3(R$*EYG!a0`HvhvBQUtK2@s0_q_F+&w(sIck6jrEI=NGi}P8O6?vgl
zsTWBFNzVvxwL35w`ohhDx|ES3#}RYgllPqb^hi*A{<AK<_L7Uw-LZAk&Wqo)^_=^E
z9=Q82|KU@|lO&Ep&V-Rpl0%`$b6x0#rqxM2$4f)ym&=eC239j&8!bEBa_M%`bdU{r
zAgwZi^iXi?xLF#9$|^LIGwbfzy?^_T9s3XM3ykkdFQ`-w?_ccqT3gm{Fk$X6G3fS8
zo@EgZ`uRZFqeo9D?d6&QNs}9knhR!sY%aLFJbzLO7C4+CW`oRSyjT!T-{aQ0j9j!O
z*KwriN2w#Y){2oOYppVX#F-#67PP}*kPMdFAyB^c+zs=Kop2=r@S~WDh2i!d`;e*l
z{e~2ByfN)^f7`b0Ns=5tb}Y*i&Y%=HN!abQ9X}{Yw?u>>yr5hquo{UYan8_xB3hd#
z5rZu3t%QU2@R(7@*s}C1+^=$v=OB5(trarYxUuNV#2U9!uWy|mDY59}-dm#XN*;%0
zj$Xioa7e0T5ScVLxiV=22%IqnAO&N@hFL0vK-MtEhu#cQRvW_@7s!kuFqU(Mh+qI2
zV~9p&N-K8y=ttlLj3Xj4X076Xj9DT?lw5F^Bcn7AxygWBrBd#CzFT&r=y&Fq<3X=6
z+L+ql*GE9|q3^^P%@{X83>bg~B8~z8hD#;@L}{92d79^O94AT2m^*aeTWbm6)CmUw
zoO6c4T8pO~d<0}Taw4$AkRe$x#tb*jipQ~&i*MVq!5B?|#3knrFlo7e3&0uU!xMBE
zG6G->8N*ZW05C+skb{vx0u6y;U=e`<O3r|6ceQPdC4;46CC?NCl$=X0D#g<JbsJ`<
zW&+0vq-<2m<0JKEwS4Zz_01xaWT&SlXJ=;@mlmYwY~6ltP%Pf^ty?Lxg7YZLm~@bk
z6Cwcv<P148rJ@uf6HHlu%4en~!~URIYuGI37Pzqm?j1jPq*x5%C}xNwW214JO4k{m
zo?KXItuD7b*Xj3rl5qxf`>nTb*}SP-sf|xgW?Am~-nr+U*ADx+QOoU>+DQGD+ioA5
znOa<J$I583wUK6;C-**Zw<SAwa;{vj?m4jk@bTlLQ#0KpS!}O=CMs8CnIpYop*GU+
ziiPQQ>!;VvmdX|2Jc^S%Q{_rEj+0`cSO|)q>yM3&JHla%&(5r?SIgx}S(6ctSXx?a
z)a$<E)a&&^u~04*j~+Q3hl5hlA0KTZS<WEr4+gz{o@EnbW5<siD+U249Bno?ZeD-n
z*ipyxR@<$xKVZmzFS+Q3r$jZhTgMOF^SMz`?6<?2omYZ4;a8{qky&C;ECl^-|NaLK
zD{wk}#n55a_^#tQE|<*Vi~(4JFwJ%!KlbfA@45S~`vUGX>(#OG(G8R9?zrpDj>cAB
z*DH-qH#&T9rP-WbT5c5!Rh1Ed(3vqhcZ6+@7Nw_W*G)Ztc<4|N1m!{@P2<21jLw<h
zX|6e!Q5=`bCA4TY>-57Yj<;?;=jf3WX{H5n-*dx$Y_$>G8${Og>+>sdqKO3_p*I-h
zS`%0RBq4dN7L+Qk$)eGE&88&K<reEfsU}3t;Wepho+xl8jFt>I17j6u!Whjs1LT^(
z0E-pHiY(fHLP0ozaXn!<W+uyZo@Q2Srj@Ezf?^<z(N|pYlTVLDpXTv_`CB;`?RJ|i
zgi#DkG)6|Xu_sQRbUfD&N`+#@5=$~wu2i{X-Oef^sXQgC9hU`xlcr%B$CixFGZ(QM
z1SR2DgF;#QnB+Q2;%+E`H>)GfYNOexjn*ru)86WTLNaKh3LVUujwuWCq#wpT6-PRa
ztjYuf1dBibmf@*f#P6}4$pQe8rH9Ue;fy1KwFWE=6I%GGO;AIhxf;5Aj4=-#E(TzY
zF$7lITq|vqHkndss+BUiw#Fjaz!hXvlEm%qDjLJ3$GAWyEOG!6fCz-ZkO1+lgAICv
zcyVce>B!;3bI0ejR*WMNX{`~_o=%r!j9EL>{KH+nr`$4}a{>aO;fWXz9dHJgUiPxb
zu;}@Y^EMf4l?7x1gbPdpazG~kFK=%WYioL)=dJa1f4keW_dd<N=l1tjUqzNgiYBd&
zv?U>p6dP28z>^FDC;{Su2LYT+oB)X(z!?~SNCq+xCo$kSfe}YmED4H}P!bK9j94rd
zi*>8Hb=%W7yZ_s-T^a0KW!eK$0Sf%aXEz5Q@cwJB^}f$S_WUz&^BEg91KZ?{H*Qr%
zfssKXA_D*nkRu7koi_8?Jj=4lXm@XK=PO_S%1^!d=AhsE;upVg{rdH<{rR7j5N_;$
z?e>!B=9S%QHQOEcUA>;4K0ZEpI-f5uX0vC<$Bl0O;;;PjhYvnFKbhKAGei|9Hh2W&
z0gyIhY2bp=7(VyLi|1$0RD@gI?Cy>0Tm9}<-nn(<%B{Vt<4*VX{*}Dn%ko@BkyOe!
zdp?^+X)+iL*NeionnQ4{VIlOp-Nkf&>Lf>{s<sUb*G>KG>^Sb{*YDjrpI-DQqoa$9
z`FbtWq$TedR=SK*5oc-bj1Dv}i)**<POIYj-CIRl<%8bNm0hlc2d`pPw@sX7E`%tK
zopX2Z-2;TSZI`n}F5^T+-7Ggv8%q_e>ty-#@?ta@IqwgSkH7lWug)%~Rar)VQu^RF
zLG5C>Od{FV6-PKeIC|~1*H_Eca=8l54|{`?gTq%|eEH(y^ziVYHD<iMeQ|!FwTV^y
z=I4LqGm(a>m|i~p7dyk@;qlox@8{VSj07+Y?M$XwH^vXY`*z`A-MT=?1xXPyE>b`R
zE{{YSDWz1cP2K8?y8gzuzy0xlM-Okk`0~9sU;Fc4`uq>xdH3;ov8+v7lu?{7mg~CK
z2#f(Rj**ZdxWORb8+VIpk>rY)s;d?pSqH0i9i>_AED#eyWU*RRrE_3BmaW;@-f_AK
zK*soCCj!nuMpdJWisq~4@O)~id%n<l0+Gzl&Xxqi1>`~qhCb+CFCF!NyRNa^ORsGw
zQ4!g#h>sIUCGrpwCc0V5(FwJ2XONOo2;LYiMBF;RZftEFa_A3}y{kR#-NE5$oW^b2
za^RD3pJPy*7<lmXb3gwZ<MIAya$i6CxS$V~-zUJbstJ*ch>Ivr66Cz9+g^7dWMqst
zCNLqm;6kvnShh_Sf)+e~aDjn*RaHnq>z0h^r&%skhov1ENe8}cZPhf(QUr{d=p|Vm
ziBwXAR)SUN(3GZ{`(~}%wQ1J2t!-12w~RahIkzFa1tQ{XGj{bMgb;#r&ijB25EwEL
zLTnr<-g{3@x6KADfDkxih#&;VjZXw<qw+))f(zsUAOs&Mc%z+vw%&Tv8r!y<$a{wn
z&|8jPat2_7)r;v_xoW+$Oi1K{GZ4?ywg?P=*m)U3DC_p%;Na22htHlq_1+1|iD*ME
zgU#HLu^)QyeEKANuIU?Y2EcQ0@#a0rhp&JA>%Cs@N8BvG_wpSN9C^Zc5<nnRObQOb
zz!L;Su8kuuh$+v1lqWzT0}upnO)EKWdPl^V3tn;&3F%wijpN<%)^GgUuYTnVuV1^e
z6TFEdb#4DjEcun4$?W{hG*z$PIXyeRc5S~u>^mYEMc@1W58itB!|%NJ;GO#qFXqL`
z@!7-sA1%%<E)S1ixqCMP80JYin~^n$psfj<1tFE-9qF!rjj}<?<46p9qde<6=O^Rs
zUayNt!(n&-*0s*=zyv*8UdE})vM7!PkwuPqH`i^wH5%FAC;R)ey6*1n7Udd|qAW4N
zEs90jO|7>|DIru8CwZO^2K_XPf@>qicDJ`<6-yz9gW-5I+8PdaCga<;ZXKOGo6V<B
zpFDp0^wG|L?zXkMuFGz>+e<TDmgH<VO{b@)oiypBNxfce4f>5~YSZ=xJ>%T<tNZW1
z^Znhu9nSf~j~{lj?C{_~x9xJZ3?VR3PaYjGKqkdvxy(D=)w)0iWmy<!v}wEju18{$
zpUf`mP%reFBtAZTbT&VI@s)e)dijOV{px2T4OXvztM!Lg>z#b?=nwvIi<s<0e)d=&
zJ%Wo%4DCdzlj+6N>C!<+Wg?l61V$VafeFKi^Gxtw#PUR>5obcoTl0-SdF%V{e*f0?
z_}}@ZuYK?P?>xJhgHqjIR@SZa5JE`eh#|JRWt<O&T?QtL)%y<)>yiYBPL>^=uC0r*
zIJRwn>nKEV%n)3#jd4}$JDnWC1Tb||Z%_J*#kww>wUlSQrp4p)`f`bj(ppCdnB|?i
zZQ3R<<boq|?t`;UD*{Rs&iQ_pXAruQdJ!a?VuVPrSTZG0aUq#x$jLh?g|#*?bOe<#
z8mO`%5M9~p_WQmzrrYgwvn*0^Sys1i-oF2T!DH*ehftR7SHJe}ZEs)uOz!JPpT<<*
zd-TozVC%}&YbU4YP17nVvn=U$I;STm#%dv@^Uenc2to)Uak-cgS?~^#rkAtRlk;A0
z7<?d)f_TQ_ao*2(LdM!k+p=lP$~GE8@YZ<S25W+AgEPUIx?Fi(I$ar4I@?%htZ~+V
zS|5T70ccaoJ}+{;bI!Sq#5?DRHgazPK}hBNW)uS6`{1mz##sXt7$Q7x;owH-H-<*<
zy>re7OB6Q#b?-e<AOh#SwJt!g-Z-bd(@xj5ZmcsQIOkeZmsK&Zi$%3w1gj+jDI)?R
zB8MB6&L`D3koDowqlX`V^wIg*8G#pnf`4ktbl!&_GUFrLbmT(_4u}vLJP!{6;`4_A
zY=*rd{Q9r|`i~ZR0gM9)L=KE2@aIN-1Y{hXcg`x#oYhLQSPEk;m=+Nu&RE3Sx{ReH
z?>TD8v74w=$=81Rb60l8X`Eb~pOR&!tw+PY*5>l;bbB<oI6ed)(l|PJ@_5jH?aXGE
z%VI5r($-wLe)HMc?CHfL9e{-<lI+>T#|oe*tEUeiO-7?&?5NkbM60S25eLqZGva=G
zX9vj-w}y_)$#lAxM76WRX^ds*O;PB}W?h3H4u_ptE~QwnR#*0}T%4Urfym@|d#r@)
z_Pf@*doR9Fn>LL^U6q0S{{EGJs_qlRs;X2J4JKV{?P9S|N*o>?-no6-I=h_BvLs*5
z7XYxewWWmk!8`8|(Yh!m<MGS4?nF_vx|n5o-pzZ~xCYAo{VVr>@Lqi7N~`PTd<p=P
z^Rg*4U1vO7?fZ|d)e`D4O0C*eyTmAJZ`B&L)u`H|_B!_7u}VT~wg|Onj7RN?M^zLR
zG)PH_y~jVl7yt7rFV3rc&UM|_^<DQ&k0Y$FaH>wmbLp+SQsm`sb?<Kc_7Z&E&vi>v
z=;&W~cOsQm%t#u*?dV{WyQrX`(3&Ce|6&5s4#NgVT-7I{3M~YSMhy78LeIu((Xrrk
zs=BT)oR`Wse>b)lQHeD*-+9mKDqfp1-K?ElZ;B!Jx4H?6r{Up72{1a%*;U`8Sh{h6
zDGudJmCw1Y2$hU9E5bwmf3WP~<L~W1bb#)RIs;siq7^VJ(%j#>^Q*vt)y?k0EJCdJ
z`gilhtn00UnemX}=kkt%;>?!AoiaOX!NT=-S~Uqy`0&?%4_}x+{#1A`l{&Z8KUjPU
z>IBiYT2xQoe?}8MJnLHWIHkxb4fb$%Ce=Y=4mogq_M0~<)44}9hOV}&Es8ygSU?rz
zc)(eQ=btX|q@16f#v<?HqQNQb`&qo>3Xk~a7DBusSXU8C95|Uwn41^JDB{@B=)+v#
zi8ilw2r!;QLBEv|dT*}wt_4}hN>sIXr$FpC`m_|^Ek}H<88vL?z^Yb>Zk5OUR!h=K
z#WsHN*1DudQNy<#*$9ubuxScxU2x-eK#E?UHgxHAL%V#jGmDm@hPvmSR+$!Jn^F0%
ztyRJjuPHtx#~hQm(sQ!rhI>NZO~T=m#OlZqMMesV_LxEcJbWZ$kZGVXb^69JO?^#5
zOWiB&V8E&tDa8nzE{btT;$i{$qDbR$X{-cURYG`d66{@N?g@vSqdqy8syGUOIP$8k
zGpcPYn6wpr78HMG?PF`!Y}(l&;^H#&XUjgUukXo*4o{(K|75`$ObXQ`GIk(M?gGEg
z)wpa}{+H4lba<&uq2k0w{k0m{b)8C)NVpt%CriOi<rH^^g!Nvr+#?bbgie{Z1l@oO
z!#{-(aM|7fEmdq}yEoiX8UzBoZHAX>#V~r`JA_ge+r6WLCf3UeJ_VFh6bw-DIp6>M
z2Kjy6t>cNjlG5b<FG}TN8|unBPos>?<VzF=wd>*QClNH9Ly=J?TH|1YvFo{sO0{h`
zX|=mfU5>c$i~*&Qd29^&U^iN)nRq20lycf3BcShwd%CS3-wDnW<ogPo&7b+S1Vn}o
zgSq2uF+%3GUKpViho-q~!42jDi+oRSZ#&eR=1o3Q?y`4Y@^Dcmn$JH%kvk`$#{Fd?
zh$l`VH9WLmtm}jL7pbN)_92Mz@2ypz3KXmyxxY3lYz=MtK&(XzuwKgK_cxZhJaNvQ
z(M+z0#WrVEC!MAz?Dc>-z8s1zZueFWIoJ0Qg$P6qpv_>z#6JOw`*W4IgoI)plULyL
zeO>LA4c@lPV#Vi&rxnWV*4Bomb@E6IwAx6ls=%Ve+uQr0<8X@>)kxoIvt7jLpJ~)v
zm&i_!;`3&D$|(5sAKB}DJ1;|=@{t3s@~sh+QtnS+U}~%@;O!Rq#t(NDn}|e9G2bEQ
z5Q>&0$Hlmiz7%@p)3{cm$V>`x!{2qlPd2q~n&GBb<v8Hm$0NfHMJgKQv9hlDls?7`
zzWgn1=&i&|qrhP`(aJPzTl|qO%|KCA#D1MlPKBTw!N%sytK><#PScOVBcJeyMu<0F
z-0nSuP-hOzh8l0BIES>`m?n*~2IaGpV=Xy(xy3*5>Ub>D9-ul3o=j*6o2pkm^ex;S
z?N%J7=Di03S2)FeCD%bAA8sR^3N=&ejxu;eS<2cxb3Z{+Q-<Pmkp*5D1!Q6AHx$+@
z2q}uMNr_j9TX>6)RDCX`lJ&yY+8Tv>hfAX9QUu<YRTs}4w^Jg^g!<+F8A>ah_6iKO
zmNlI*iM|Y$+@k9m?t+YAd7W8gO-Hej1sb+tY%x|(>KfN#SH3BD8{m=NZ2S~lBx;)Z
z5J@l?_Hds)`WEfK!@6Y;MX8@r@=|jM#jQ|>Q%C<>B`xq}lqGQm@r2LRxG0o|U`2;2
z7Y7C`=&V1yqvZkt^~Gq*Q{G8+^=}WOsd$)0aKE4`1JP2Te?c*$ECf<&J~9?&Qt5!*
zFMGn{g3sOu+fAjo%00ED3p}_ML`k`PnDu<l!<lwI==pKY{=bQreAB`BV2*ed6d4cI
z+MQMC^&QH-GlQl^W3NANItG@g?eDx_2`2+IqLc^4&A_)~3=ZIp1`DawQE}iqBt5_?
z`iadqjorD1oq4waKY#F>YE^E5zb91YCnxZu2Zk=y0y8V^0$8jIl~UaH_Vkdy*8?+y
zMd?W=$(_I^H{Xy*Kk9*#um_ysWfgykkhK_xa)<+bV!vW>EGUTI+1(s(>2tO_FbJ6M
zAbtG(hC2)MYMam)dNs3eE^tKSU_YyFJVG=KX@jqU!ylb}0Q~&Yq8d}Gm}d<0PvmVh
zKj5n#*S-e&S<AuaNX-1^rjIx^Do870CKrY!ju*GNA)V`8>cg7l+4ivTq>8@*@{8rZ
zNKwgu5DS4w+n{dv2LyTG?x`zaF?$y`*I;*oArJ_q9Gqde9Sm|JKKSwvJx`T|S>0Uo
znTPcKS<{H{U`7pN3r5|~v*Fp*t`_;lY)=)GnO}H#6HYayhC0@a=xED$-z35IX0Z3Z
zC^7s?Wo}BLCY<8%ZerlfuzHZ!D5t`!$g6shlXju)$pW>?cW&sW6NCwP@y4}i+xDt1
z_hX@G??Lx6_5GTQnQ}N%ws+xEJAV0k<R*+TYF)0FsY*V!ea)pT#*}9;`a?JCiyD36
z3Rjkj{eT`fg$qbDIgU0<Cx4IkgHG5LdiPD|&v8Zx%$l*Y^~xVX7RTIQ<a-?y{nU8{
z1<k`$^@u&%Xmahg8p{lzkl#4x>%rM=mIDrhRBz<b^d#HGKwH)8blQo^=&Xf)VKsx7
zUgufG>XnQAy%E%4|E07<__O@H183P}hJCqMUnYT%+k)D!RzPcH*fbDXv^t8V;C4@7
znlG<EZ#tD>^fNGwEZ$ZY{wv~WHF0WVrtR|<9PzBv`Xs(d1ycmn%^2%^9WoMRWm6SQ
zk<?+pS<jf{_*RzL1_?S#|1n}ufsAccAz_nQr#>yR^gW*VfW{OoHCL6r+}F>4r32JR
z7_!xkJ;r5P>TL@CO-PRn!EJTXa=%5u+fnz&obU7b=1Pu%qa=7$II^t2e4`l-2Ym%8
zf#O%-viHTgcqjm;G>Kt|?l21vm8v(~s|W`_!5M@ns&EGr#^l7v9obpPaK8=;-rXUI
zdUvBnp%f*s8lM)O6{SG5<vLn}$eCaP+))A1s`ju{!Dk<HTHg3NY?WD{I#S4TBt+x5
zzK1sjSuGt-zx9a#aOX$0oR?gslP+tDTr}L^;I%2B9q{Zu8~$Nb3eJ&Lep2H7QT2Gx
zanT5eRQ|mi)_^1xWHlch+Mq}x$>w!g`-T@oLbh;3H**_aqZM{hB^_YO{Afnng57vm
zvz^&Rpa3^>)Md?VY9Tu_<YsJA%{((y1J+Fd|K_#$=E8{tI0v@3J*;N#mmi3{%-#x6
zaLuU$%-wMwR{yAe>XOLt=@f-9>+U`Z3qyA|xg{6Ado8^{x}QijMQLSKk~ONcQSWtP
z){IrmRO?Tjd1?4|qUtgJHKw=()&$b_j|?=fOZBt36BaR4$-tm!;yk<UudsxHYH@01
zOI8i7u*UbyL>#S~GasCVwswWC-_|l?=8O9RlsqPPj7e^K6-#F|_^i=VY=eeb)n*QE
zs71{k{|aHo8F4u6BW+oqZ_F-c?(P)4eSTLoIOxA>mGw|~wS|B|z)VJhm8pfZE5nE}
zApT>RSdTtV7DcMmSI_zok5FHr&RwpeD8rMsb_$%c&CcJyKLVCy47eCzxFy3PXXG=$
zNM{Tz*Oc1aytKVN>o-5yl`~NrW=@6dmC9hJ*tQG97AW6y-JNY}50*^pUcST;TD}Tg
zTxDXe0!5mX&j&db73Do=#CzHbj1o9YU>q_^IWg}{k|bg?80b|+IvGZ9#YL$%YD-d*
z=qqTycTM+(SM+!K(~kZW(SbcdSvUeMD?i7i7TZ*+<_L8Yc`_<Y{&?z+pyD1Laa01*
zWK3m^97ob0l2Y~~Owb*g<kTqbylu5z%pXk39FwD~%f~`>TQzG%)rVe)Nv~wct+v=~
zf`^Iz#EUUihqo>FukU?c-JAk;Ih;0m6ikAT0vv*;21kczSz6&Eu!Ci;x&X5YFLaax
zUTRutS6@?l{$94$?+M4T*K&ttQGkk@<+lDQjorl&9A2uO+5338>*g-y+r>a6%r~yU
zF2c3rNgDKUW%sR&olss}zgewYXs45LAQt^V^xkI%?t7m7?kWy#cdMqqeHY){V3|#A
zDiMXX!CrBF!OZ2=<J(*h9e~Cnlqme6-k)fRJn%h1UV1W7>JaK@+B#ELoNF;b7&#Ys
z(Aup(FDY_ik3@&~cggP5&E9dHY1u1jnWk=Gb)swn4TR}|v);bccnAi2g2@NMaA`pq
z|GxXn4C0n_Py~ZN1gya*K*MoCkuvax{E0R%|HYhkmT(3M4NPvTAK?49hNh_!^iVKs
z`m02yiB&;nsbVigd5TAaI(jq&H~6miazm&2GhVjrMkphAhC=n6aUzX~bY0F7LzFnN
zT_LOKvMl#Td3kQWD(kQ(L%X}Ap1EAFWnQ0dmTT1xskU#dosKHV!o`?tc|hdkgoqDW
z8xQPzJnpYJSTM$Q0M>bML)rX}*48Jb22)d0uqOjnF6r%oNJ}14oP$kWZmudds<F_Z
zn)xL^(a(W?X(3DUN%n810QAfODJo1#sg6ZNmW;3ZOuQO_4AFa@n99C}8=~*K)K*#K
z#Trii@XW7hOoDJhR$jdE?UI_`LsKxr+EC|h(=iiX)ia#Hop`Wmr(v!>Htu<g!34_z
zdPvu%q~xS~>Bu3EpE$ir4~tVGZM_>``R^+d{F|k_L$-(i`Yqu7MNz0Jqp6by>&Z|!
z#31ylH^s!5srG1VfB(CMENhwrw|T1bm$v2Yqi&?SP<8sWrTfwkgA%8M`M;NI(hRzO
z)?&DrW(NRSlBPAF0Lsa>F2GlPo<bR>`c1ws^zbe%f!!}p{9j8Ld2w`r`g{4kZTlD<
zEK!W<id?dcr4C^TYSx!Mdei{OsCpOR=l}v|p}+)i5HGg$i@1k2TvuMn_Wp!%;11U9
z&L)?`H^r`WBRySi(ANrtr5m3G>VcP)l#-jkSH*i-Ua;{$d@j@(d*EC7nu;4D;^Wb=
zG>9PeT#ZJWk53G$i-&W%l6}^|#^OvuLcHy$b$Et@4532Qe%IFLc61y=%*>}!eA}}%
z(O{Pd8(Cp9$$ZoK9|$Q*7##@<v_C6bKocfNJ&>D=p-9LbRwR6y{$SORYR{Eirv4oW
z<OrY`P76)e=MiPZzVL}ple!&u2clMK^7HHHEj1Ow012aJgC|i{?fRWCUdH{;=YhAU
zD|?O~`stNV|3p*7^8^iVNZbrHL|zF<rOj4dV-lC^YMNYi2c8v&cES*;>$6vnoGp}}
zLhf*D|C49BB#juC7mY)|UHb-Uzjf{dsQV>YC~qE)xSGFzZovFsD64qK-6Q6$n$P&E
z8vLvqQq+Bg)Vboj;?A1BASsF^h#O48gC!EIAeA*)N_E16r*<s+J#a*>U)*S+$#=-T
zx+7v&i#b3opf6NjW4#Yi^m0)u;GTA{453-*1M&b9$~Ky~LIRQ_Pv~bA*96iea>@TF
zo`RUNpHr^MsLXU7brI)vPlH_&Fbj3<wwuJv`U}YyX<+@CTm#cTZDTz1_(PY1=0O1}
z&&Au^ifSjYkf>5hXIcJngIhNZ>h&NNMcL(Whsiiju|u)mGk)>tG%_!SYLIrCwelMU
z=(O_H_HyLS_0^ewqjj!UHNc!Yd+sma2Wt>xI+&RMJr%M3sC*Bi`+Ixa*a89NZ8DQJ
zrHA1I_E)}LqtMd>&GY`Yd6o9C9zvi3(GJ9St6%PsslOcZKg%%h%9CLT{&kyW7;2(R
z2II#Kthx{dHNdICKYXJ<Ly=o+dlO=q<lk$4GMNwahW?w_T@RZBl&}9_ne8Ge=*?=^
z^R8;UySu6XAnh9FmTZlvv-y&n^GUoi1!lVr(Q?~ZLB29VI8K55^?#h?xGp$qb$Sf7
zV_U|v|2kA|>AeTOYiy=P;g6X4gl(TP%b^j<FzgD06)>u6OnbE(=^b>kf4RSQ>Lke#
zqyTTLS*+p5zi|PW%dhQyCDJ=s?k3c?uElySEbYl?y;*D%s;Lp&D(mTi$z%O%6<jfs
zl0)(K_QmM?-L~~a4_BvG1GAu$D|=5d>W$s$rkKxwYSe)J3f#R|J{3Cl_Ks48@b`dn
zd==Dr$Zv`BP|mSy)~tAG43)=bcdlPCfN)tpqXJHF-T<IvE!qv;Mu}a*el5@*r2e(n
zUBHEnEfOxL)0uaaAc_%$@<B<iM6PHClF#;yYmH8D^(6avauh2SILTTZmDNMVWoX~Z
zzn_viybxf4jKm)UdviBiSA^@$p?Ot2|KzJ&b?O9?b&Itg?=-#N6`q#255FT916mIa
zwTw;V%>o!kmJAs*{$2P_1_dDndoBv{$?-79z|>ULb9Mcmfx`uK9?Zuz_RM&i2Q0a*
z)jS$h*)Oq=h#rK}7_$Sai5Bjntql)`B&Oau_v?<0YfdCdP)1!2F%v}1YOC7H4R?^l
z>b&Y+TvWalRw;9&zzH<hJSr~^Xe8g)X#HW?ce($_v$j+*Yyp3S<@G9($(Vs`Q{0Z3
z=&HvzX@Z7WLvgK!Ny|SmY#L`n(AdykQax~g7^#YpNuH_c6QH)qT6CgJ^t}mBM!9rm
z^Dj4V#N7clyqsDieaLN8SlpLQ;R^%NTy5$~6QU2@8bXkBvHsF!8|E@bwIst$?t-92
zzjbUm9(eu*v#3(i<H)9WEr9P-+Z~q2{eAc2j$;Y7n%|2&L!b0EKK5XMOaM$<==;&*
z$gi>fKK3%<4KT=3fRHg`Q*i*r9H^kE8Y)0kt}sLXZq3B%sS*P*vBIp;S6oJ4Ww6)j
z(43M)((6jP#E7nES=W^zZ<}Mb@OC%8^=v3!@Na=kYI0JpjKkqaVwWi8v!#QXm%D$S
zBJ`nrtd!^$4lW4pJ8gF)mdXB4(Ei!bw3#v@<@VD#Acrp-v{<7}Jc<HrhI1yjBz-VH
zhCV<wb2N1Cms1TYWtI)>9?z6RN2X{SKg&6^Ll5i3-qcDPOmuY>+ID~`V+X=5G<BaK
z&Q@aS?0|L5b>5>)Izw1<TC#$y_hZp&==AQ&9#k+d$hO10-E*-mhDKHUz*=Y1t?y}x
zV(2B%QH-mdcMEC?3k>A%%#16vvBq>nIyM&7-&*L}shCIGCyDBdRj*xL?_g#|jOy+B
zyb$TRmftooU1P_B$-fQrwS*V$-|0Du_3{q%4Y}BuEU*n;wOT$$Ut=(58c&K1?IO<d
znrl8QdIWUoz$#umcu)dTVIK*DZg2W`2@3;FUM|x79~O1|`~usMNGsETuS4rYL#^&%
zS?i`YQYslNai12MK=L7yL}v27x4D5FN2c=mA0pu7B@s7+EjQ7`Z-y&!5YmP*Kj|=l
zvRQUcO)(DL-6fbK9bgHVeOJkEgX&mFGvkM3(&=?S%?*XW#t&zA7CuE(;|Kr&*d991
zd89HCP1It;527VdS<$|2#X|Q)NMG-c76x(Lf~r%m+Bn}M(RR}B=HtyiJ-)~sTE1Go
zMgwQ<dKG<VveyZLfmH_Mys5M6Lk^~tdOEvz9aXghBSSlov$be<?_U+g{I37fR~J;O
zMvt0)BhHw)L$KxmQ>x!Dd{=$~+x{s-U8OipbGZX(WQqFpeE93f;=|Jmyf5L}5(GX6
z-e@TS2*ch=w$Ai|J<L&;T!mI&Qr(`5jOLp=VN8M4=#IFP&cCdIFV(>V&=X2WO_ma+
zr;?_#Q=#h)t2cK_Ql|dOkoZ&Q#~ef_aVM$shkSpgTT~SFs@A=aI(xhV(0)4J)+!nd
z;`nGE4T<?(_QSa-sIqc%KHFB*#+2(5S4YZB$D&1DO4E~r7tDDLcWR~g`TbQ)5$=Z7
z$)HL`N>1OBPyTAGgLheehERSb4wA_d8C)9Q-m8CAxf9=r3TEzaH*?4J9H?`p9WimU
z?4pF8MN^R*(z*`*S{Ko70Lma>T-vR8!7K@XTdIBZJx4Aw7g#ivJN)4W^1uB1?rH6k
z#>0uiaW&KE7MKg0cj}So1S3}h2p#g~vDUl}gZ~eiJTG^`(J{sea(|TU^P)V%dljw$
z_#)#z8#zfVXU%Wfylc&qTVqs=X%$%*@6)@=8`S>nte~HFnO?dN0?&NLmaleBN)>~@
zk>WrwHCe(}*B{&dxC08uSF*EcG%aK~dFc9!dCAipcIk54hL>}dD@iM&Vq*F4o?DyX
zby;IoCd1`Z&nh_6Ly_NSTjXEH!lCo30h;re$;{=8XT0k9E8-+<8j9r;&CvBwQJyb9
z?k{$%<k|CsUXJap1AN-Uj1%4P?d|Om!qcKV-RyTD;s$x2wSDFKU>iJ^2)t&4>Ta7L
ze!i!=Oxh5YnZZ*T%n@Rj?ft>|Tf92|5@IKeI7d~6iFGuwe%*9yd!FxQJ8x8Wx;Y=x
zXkDr5EGZum`d0|1sIbYmv=Y9c1C?G_4BMQasV<*J2mhx4D9gDiR$9C%l3^Nmhw?wg
z*7?`Z;RS_81`#LAVeP`Ynjm#AWN*O1a4em2kMF_8<!}F1TdE51+y6rL5Xx6GVJACJ
zG=x}*)S#xj`?g!>AqiautroN~a2sCW)pxbj?){g*oGM2X85S1SSs1Z<y?K2d>HV{|
zb(5|wBh&EdatIENL?Y)sj`t_Ti1iA_U6r1-_Qm87Xurn_&_#*-GlYH{1+1nb3kc&=
zOT9OL$o|nV$~5{dZTol>iZ<s2T37una&qE_1Ch7WZVuWx`j^j^ZnoNq6?$QHvm^^#
za9$2BpIzy_$PnBPP}kI0&0h29C7qI&@LT>>=>Jdh*jsQ9PVmwx@go-qW#NwSW#R%c
zPh}W6eV0QP-bRLBReJ8EdzRULe_+$ScdT#%<wZNi9DzQTR`u4GD`o(4Z?eH_G4Jkl
zJW&G8iGE=0XJzmat)sYCc=tQ0<-YvSVJM${Ty}2ldQJNSi*3GpsKulJqf&$Rpl+c-
z1o_-Q2uqSx0p7!CRbw-ymA7Fhm8%nUoKK4knQgLPWNE4u1wtV~1I$BRi|OKqCc;)r
zIVWGUU-}pnfBLiaHtf+B&>)I?ZP602jXW6;<fshUp*C|iZfo<>Un)XVV&(Q;@ph_x
zWLx@>nKzMVk`_WBk$EB_o>2UoE$S9=3;gyFm-dccl^R3ohhe2gTj3GYT@;LLR6VpJ
zuj0V{6e_iA!>V+Kx1A)(-5;x+g@2lOP7()_gdAcLR%AV>qS9-DM8znY0f^vz&xr(`
zO~L&_6VJgyy_atun^so!vyeKH$}OPl%pT2jd2y099!$(!qr8B^rLM56r3->|>+wK{
z{lMa65iO~G=^91kX}IuVY5wul!&V^Swa9_tqg&;O>vVjW`AN`k?L#U!@oVe3LMgz#
z<`k%?T}yN-w`u@+d3jBBm&w*gVcxyhx(s8y863)?(%rB&gLl{kgUT`l!k4`|b!dNn
zX0WN|JvhL|$_1CY-X&ODQ$Oy8w0&Q-n#S@ySW!5ICk6VcJ3!h-DLgPn@*iK3a=}T4
zvE6~#%)As*m6;Z5mLcET$<S6|vqy$L7w329{yc&{7zhSz&?5hB&X8HPQyG<%aFmJ<
zCk%4a-x8&zqw71@>-!;Xz^Fl#B)k}(n3yONOgReuNifoM6mp_o-I%oUtW7=Ys;Rm5
z*<D+CAF#98?1J4Ze5b^k8T~dWC{Y64=2=t)-RtmN9C9c*>RR%`jX?#jh-v|Z)rEQe
z9r>lmh62EOerHEC#M|m|b_lNIX(WT{+eMkFcRZF2=%r^je@+GUW>P06h<b^Pk@7Jo
z$<iqdvEr;`y^74p=v^*PcLnvnFKxq_?Sfw;ZS_RSv78a7(}S5Dbxbs4>+TK6!0_-g
zla^N^@PkZF<)=zY&xXglSTZPOy=;O8de6@vEdy5=tI|B_UBmp#!$<Q4wyQxw-jRC&
z5x#+um(oj%oLB!ASiQ<|Rf+o~TWH0J^JV0g@Oa=>)=zP3vs0SHifTaZYkbxxMgnij
z%Rg1VHNpHsSg#+(4>TAi;Ll3NKOG;_^<HW9UB2tv?YlO*`7>KA+vkrP_LoU@tZZdU
z>K+7wyJ>>jewTK?+s4`ToJh43km8+^I$^hlSSLb_1sr%*RT;|eehc1A?&R|5crxFu
zaR27_gxEDz-?8^)>+<Df-;F}w&D;%i`4{U+?*TICJL=1zqgT(v71%9Qe^De)mk`bt
z+USIwv7K9Zg;FQgXAoC^Qzw;nXOJt;+4>$(OIpkR3Qs}i?fH4q%{e|dpp}8Eq5$5p
zT(>y(CDpPo(6Rezum6$J+d)Q0Mf@8yD~_zd7sDCqMDoR}Wn~{d**$mcE%csI!(vWj
zCSTIQmp3bfHx)xHdn}Jj&KBA<9bxsdU44!9gNcZl>5K?k!x47thrD2GTfgxz?@J)D
zsnNvMg+1av`##3Xz0`;+TOvFoa!p_8AF1Q9nrGBG^ba{<YWt|R0oEBAbPHy|9xjpY
zX@$Vsr`1~3vbnbe6up%?Am`Cy%zA1!BOI3_iTbVXOQGa3O2rZ04@v_GiF+WdOg)Ko
zOd-zL^RpY`9qMca){K&ONgiEX0iNn`r+F=ONg6a3mWbk*2W^jMO2YzzB>fH4(2Xi<
zYP*qx5jRfO=ZcF*pTE*@%Lq>?jg2Pky38w(@)-EG<n>*j*P`Ep--bc#^$m@ty5N|r
z6O3KM#8|KvwJ{^U&9h8!A4MB_2FEMN1_>Ow(aIOu+S<+|B^NNaxXv?xyu7DRl@X7M
z0}{1}s^x-rL+cJQWITk`weYIcFAy5p`h1J@n*Ub@*w{48)ecYH>3%z};Vi8YkY!+8
zCaE~Co~pynwitPWAlP*K;{Kef3U|J05Wo(&S@>nl9BkVE#}?RyOQZsJ4uV8cMwGE{
zk(LqbG|KP7!+!zEGj5&rseXzar7+m#H|E~p(9m<@;yzhm7x}qni&uv!H9n{8v%AaR
z&{n0sj*f^E3{=3#HgvO6`3$PO>D1|dv~!Hgt2IZ7iIsA&*Nu~cE#1v(XHh8B@!N3X
zqVJVo3i~nVWm9Tv-E*|<{Ya=_!dZ60$j~tD*xA@40Jvy0S$Fe9;y2FRWJ>k<y?gez
zt}agUvUGCnKXeD}+_xFgB~^j(>lCqbaFD8q%mza`0=G3wa0Cp{FuP52vQz0A8Xo?)
z?n%HqM*y%fUH7{hFX%O4UdNzzg07bdhHWRXL|Xvw*Yzi`+qFp3xZ<}q^~c<;lMAB~
z#j|~&pzqQ`M-5rndvbB3*J%AC;Qjbpj9JR-c7<=<dw*B1k4L<3F<~N4Kd4N<LOw}C
zy~(@P^PiHbWKJ^a2Dn09tWb-z<ts@vw~rQ)>#w$z%|`PwQKTB_HN!`0-`GM9JdCfA
zH)l5xB+y-llXIL%b>_b}G;dfvlX_4%X2K8XYQQ8s^^0^t4tM0BuHPkwN41};WGb4L
z)w>&UbvB{c8kQqYsOac2Pi!y7ao32O)4uD)o8VS2@?!H!64aRdKN?+L0>(Mdo`Xjv
zp4}mpZY3KwSk!F7l00}#t-=wlBle2RE#hU1fosn#u8O|`IiuOnlhYQf^mD9b6S{r^
zTSMFUIen|`ql++`qtcsMAk5{_O&w7M#ToO#hR@Nu-bJHvUq$4NC<LEwBC_$+-NQ!K
z65mySkI8U@SyyP-+D0U)(NO-m;D;k8;)d^j*Gv9~b=gYX7r_)$Q?~61v<bqiHM5D<
zsH^qZbg2cQQ(-RIf?jX@umW=gMWq4pE73Gg?DS=V460wEBGsHuP+>IoJ<S8}S00Bj
z*9BydO5g_QDX_5u9PHwx4g(V4Z=r+Pw+6DMf>B%t)OU3J$cNAP>zf;q!30%mE^_f`
zMe;YD4o)9DgdNAoJ~D8qxi9<>)>dbyad|?*t&WlBSMl5Mcqmt(3VRxn%eM)16!iID
zpl7Iz;_ki})x3Q(7We*o_U=mY(HGE^!s*QdLP8X%%d`FUbNTGd?oy5i@C%2%2~nI!
z9iW!aly6p$C(6%#{T<zZ{6MrJs-vuVmkjUkiS?W<z-yttVhaxfs}9!D_FsexLZ;f#
zZd&?P_~8;85hFvoX6rSbu4Z5s`o47PP3n4z{q*q~5|7oe_y>QmgFasUZiomDwG9mP
zavprj$WZo8?oGI_vr9uw${=>8ILMA4q@tU1`e#jF^OowTR9y*7vSTtFIKJo$!XIz*
z;%(VW{j~Zm7>9y(esy*fG6g)jpUD}1+WVrW_<x_?fIV2J*B13O5om#uMdjo?j66{w
z3UR-uHqX~7vtJ(;d!sduMwU;9gC*xNOY@6;lJLpi{dKX1Is6#vczWRVjC9gZ{Lu#f
ztiAULVOzcU;|PZdE*F`osTpr-Nyi7u8(1uL_+sY=nnj6}(<K_dn06pQ$7$i&f>uqc
zg9^AZjE=p6m6iY`RKqJ9lXy)$M2F{n_~|v(3*C=SA`ch0EKF%owE@!;`zAxR>*H=u
zpDMVJ6Z%d{nvaLm*@z0%2egUGM`yCQgR@QI6WiP^_UpZ&TEpDgpFz5yljHL%&L<yk
zZUW2-nh7_k8wykiXOCk5wKlh2McRC8x9rZ}h@G2jEC3wF4w0G6rz3kSd?bpQ5!wg5
zT$aoWARglYaaEtODEbqUTWbAL(fwq&rY8y1*6zXmg`__Ec}2{o99mybIU~av_2vk7
zOc{=HtQm&%KY_`7AF_n_S!%r?En0z}{IQ|v52wbQnYnYT&%7lcFCUdB#3bj|=q=W*
z5^g4Mfa~QOx|@gGR0xGmYpaAi##T5GkEh3Xz2dUpuCydb0XEVC>p3*{P!jityWiv!
zUPv;ojfE~-n#s+Pt8z2S<$a>`XAl~D|M-^rr&>#Vh3sG0)D$ltd#yY4+aC>vo4EYg
zFr0xq7Sa%IV`>`3#UeET(qgJj%;X^5pS9G|0Wd`gb03b*{p6Ro0(gdBeJCuKjCN1I
ztKv8sn<K;kHivz-?{ptEtM^>&nx1I#65$h;M2VphGn6p*BnDP(UmTJqCX6eiADqmq
zCRCUU<2LpgCW}_vRZqyK$rueieCCd5_08xJ=zERbwU51}1m?;jhSYO+t+3Djl;k%B
zR6c{^?a5&@q`Mq~<mAbB<g^_1cZVO62oI}LTlxr9G}S*A_8fEE<*hhw`%kG+^cLmx
z!%!|lln$S2l>IYt>Na}rwxjd4>5;3|V|=@}(!#Praed5eazFnEwVh}%`-PL%_M5=_
zg;f1_n@OHUTB686SEm#&`)_7${x*<w05#SwA>;J-{;}x}+y*YgZeqP!BtqSLG*s}j
z_gR{iUU>(#xA*Rzr*3ckPWA4wuG@$nJ?DzcF%I{aST_z04h``SJVS*M=mZpMyI>7&
z{5iT8&eMzt3p4rTH4F-pQnq8i(jjt}!)T(ZGWH<e<oA&*rf@QUssv8{!K?2f@DSbh
zTE3VS&*B)ZTJnbT7hpH`R;Dz){ap4$%Ur-h8A_*geY!y-27|$Pb^Nv$S_O<sxEw8u
zO&e`dxV4CUmpc|Y{G;!^7DXoHmqZKxm7Cl4C!=TRJwC@*yHxuY=~qjE_E7}e`H3$0
z@E?eQOR7foqNmm4N*2#_%|QgvRo}2<H{a*`H04LIv}J=CN{*{50FR~#+IbpsNlqb<
zV2PL3Yfl(-vk~uaZkFO!FLI6@_@K;QVgr+{Vl^6GD6(ysyFZz2;h2uDJ(GMPnZk(k
z=q#+%VIacx@^dE*R*s^gVs;&2nGfA-EoD9L3pA!ceh<u4x6QsFQhWf2tmAc`5k@)W
z@O0yd&0wj;AOc_65Xt}jKid+H>j6TP1QE=S{p=}=uS;!pF!q(dIBs9QMrM|LdwvD<
zAkFLk&IeStqYZCN&VI-R?00s0FE74d@@Uf0t<cPW976sm-RUm7JxhZ47l}4b50=O;
zulsjnsUD7sMLv0>Da8WJ?|%ce@z$^w4YC*g4Zhg$@%!|$^ZND3dSX}ZsbqzdbPI;|
zd4I^yc$=_>E30L5_c!FBA)lFAIMa3c7>UaDG}Q_x2BEL<y=hWe(ft%gaq21{n!B;!
z9I1^oHly0)k(f`}RThlqUANa{yWl3DM`<)vKT~JTx*he5;to0iFwg0iU^rIQEbo|O
z!+G^q3|-ngoFbRhFj&(2(f&W?Ws)ejt1EAyTv>@jKatwnFPY?#3qD!bLOUYhb?4(O
z?l0ic1bb2zy1e6j)trjqxTrru<P}YI`%UswsRw294Y~yV>U@*l-1q8bM8T-#9c`Wa
z*^kW>(p&TLmda-UT{h-j^wpqj<e74PtC@2C(fOvS$yw$O;PSbd^ec!Q3?ZZinm0Us
z$#@<PLGxF*BF((T`O}12#o|ea3VG9vWz>ZqKVP@@>S+Z<#W9Xb(U076*}4pu{zF%N
z(Dcc}l{%#2d!bh2fqmTuHB62771M;e@0hW;J`V*W#8DTZ9ylutdZvyFC8VFoK=3{M
zxR!eTN<AA(Jp3J6-sLI11*wbLZX7SsVb7v;^gA0I<2x)jLh+^G9bqi&RsS2%kR1ba
ziR7XjeS3z8IATzG?x4x?IZIuDsfFLyiu@Xg?o1b59E{9mv@q*7%R?R~8frco63C!t
zISDy*?+0K>NLI`yyYofmT+tv;w91jW;*>%mxbc;;Z;MOcRZwK@*-Xfx?yiX5JwC)c
z*ccp9d$NP6jzV{<_=ktD!6qhnQX~h6CDrMRl_$##jO+>X4heBqDrq-ES1rJda0$DY
zrye){KGG~Wj-4G2|E}$UzrW%ln=ugfZ=}QkY_Fo$x7N?Cq1kDf$C*=p5Y9;&UCOJ0
zfdMs_apn@>me9iT(qrMP?|NbaW5io_G9fs0qPoypUCm*!-P!ny_6IIw7B?+?Wz9&c
ziARr?ZwBb*&|#7W9DkqQA}Y)Z#O!*38ztefyjRP=f-kn)i+kT!eNJ3SKCA<wLXjo!
zfNOhi!2Keo_4s_UIFmzJx9q3M>;|fh&Q^1}S=M`JoA6n%Xd^`AA6JNy62{u9n!@l$
zY_r*IIhH%%X|~!5dC3v@x+wN<nD@<xS>kT>rp8T@g?nIOxa)-*gq7;U5^`zvEu#SO
zw<`4o71e~E-kMr|1$OV%Z6yf7t|J&|oc=T=0NwK;y1;^iE(YmJ!*<SI)pT<%Vdm#q
zGqi}3`{&**MR$e1vXx$bJv+1SGzE5dSJ4}0Imyk}^Xn4@_k?0k7~Xy-0M=GyE<LvH
zv?-DkZsisO<Z7U?d4Hvb+EKl18OvBME>N@)+t%<Te0mJ&YN28M^_xJB`7q!!j0BBX
zh{TcE3nr6N{#NlL*`!=URjNNa`biNYEh1>oWwvnuc~KxU7wq%#-|>zPQ&|k3uXOPg
zf(G=9g*i!;8YB%fWc9<QtUSe*aSWkC)awo6qZqi89UKomY~8TwZF$X1{*UQDr-9UW
zY<LP8|G*90bSzADcy5t{Xn?5bs3?*bc|_-YwDwwDlCEc~MDuvsfpGiBma2!d!+yBE
z`RF;yJO_PBbq?dl4BIWHM&+9xUGwhM64p!eKg#{2Sx!X5tsfmr2?wK|>tw(9RMt_I
ze!SxEEC98z=;9^fp`JqVxi!)+u?Pq5bVtE7a-WcRmRl<&@k0c`cwhzbkefB7sSb7!
z_f7tL0x7j9Ph!u$D1_6Rk0H-j2+T`LLlQfpvIcP(W#$b%&JDmKslHN3urQ^lc-K8a
zS-Ol_;$bHxOs>IG74N8{DgKQDi*t#>s@Gr$;ot16WT4d^sugZ>Cl86_+aMbITa+BI
z^8VzkSO+lEFt|1y(4q#3+N&5BeYh*1Bkv<B$*8aEhazRrB6k|L6uzw^Lhhjbg3nyc
zO98ihIo|h1aQfly(5czV?qMI9NDq+OIX@&-cR>><(Qg+Y1-cBnxw4yTGV#f-nnxX1
zRvMn{z%T5=@c5ZC@~94bl!7YJFR?loiCyjve?6&BT&RFsbUL{|K^YydS)!=wy_Nuf
znUa!B{+TA>bOl8=^4}WHkFZaD)l{tS+j!JHo^xwmeQUxJBrmkHT?#iE-}BKP_=cOo
zLYp^R?9<$(G9NbfG+pmbJBr5BaDZZu?z}f)kn^e&Ltc-kW04BX?dDL+eCdKVQ7@x?
zR|`{L25-Q_OLABMyB80K=cONA2^bM0=E9I1j&)03G~5a0#?7<BDXO|5>szzaZ(+<D
zruiDCq75R6;QQ$bjmLl1*8X1BNry%_ItF+mW<54SdXdOU7AqT@2;5Rol+^7X$_g)a
zxBvcS|GV9;unmq%;vWikbivQFP3>@Uyz+$n&I2aQ!76NUW(qDeTROx5Ej!oxS+iV1
zG7XzVa%^UnQYN@@V407`Kf!r0jx{&WSmqy1o$KW8Z0~I8a&oX|B$PJ^qRr=eny~|N
zWZFc8&?$S$t(BM#KWa5+W$_UVlhK2toFKM#R#tt-?>HxV!5{mwDCU!@MXL*~R%grr
zgW2+51+~nwpIA#jQtiahzu?k^8~N=+gxE<O`zusY+8^w#$(@wM?Ip>n+0#y}wYhel
zO6464kNEOJApb@h!<mYxDI6y)dB7B3?ERnCQrsR^P@(OoL6Dp(inRuQJ?*Jse`dS&
z>u1!jd%0C#2Iv!3JeWo>a(V+HYa;KN<=ggOzhYd7R~H)UH+2RXf$j)tYzbpbRDG>~
zF(t(c)YZM?Amt_z9An86sIIM%_VU{P6-xFBr8!j(jrQhakm|GWZU98^M>Dm8I6Hta
zRR-<wJ_pu)Ip|5Qyl>FUt}>Bs8IfV`RIJiPh&Rojonp936*004x)tMW2@WA4u9EK+
zXpngUWleyh&Lk|m_M7#o@LHtF8NJZT^w`<%%9@8{EhTnre5<C>5Oy?i_U)BIthMfw
zxgtG_+>GEC5E)Tx+pd9@qHnEy!z7mJZ;*KjD?skQ)AlTnKJEKCKE8|L8dT>;$%%F}
zqv;Wz#}yyE+9Lz5X3eUUdXA=1t?7g1gX?+g=?H}O=tVohs~$l!wjX}&LztGDvKCAI
zjf%Z+uXSJOY_aVNKD&(#mxuqrr+Wp3xVp|K?&9o1rSzc5X>ShZS9iP+)4|8{`0A@`
z=oeomk}K$luO9ggGN=K{<Y&y=DCc;0@Yy4at8Wv37tXcXy+<tFM;tPyew5fP*rICQ
zUt`52p{ah0BzK`M?-EH!poDj}fYwROElOh259vj61Um>E)j6~b0v*l;Qm5u&?&GXA
zi7u?lBEx-LDezk6frSmP&MyxcrM(4w`QG8$slGX_#iutjXJ7g{XLyp~BCK~&o)za(
zbK9-OdQIl>Ev_axlA>{0NUxLYC540e3Xy}M?bIx)EcvA7WRA4j@wjQ4U!>zJA7}se
z9R9uwo%7EFmi`zBU<1qFJL$OEsJOKkc}HyOS>!0eCe8BA&GHLcMH>{~IZGDv>&O>R
z#sqQpo^Ws{B?T>PYk!)}%QLKf%4U?q+Z(*TwYo?Dl1Il>saWo%&Wd685e>6h=)}Ri
zzf7snXIL}izo$Kfo~eVLlG*IW=9vS{=1mwD)_1ujI-xDjG1c`5FEHZr=Cbjw{!=-~
zIp+-yDE#VR`OPKoL=@bAU@qEw>Vd^9a_&M|16#@3mHVL6fdV9K(6hiDl0VQtK_286
zn?|zGSHWClMDh(97AyuR2S+-0E^bGjP9vIfii_FmC1WFRE(n)@h^+$C2LR5MvMiZZ
zXKAEDl{wlZ<U%3cOKCi3$BRV#{OEs5Ja6m1z@wEzb+e&Wtr6yWYaw_<%Z-zB<(zA)
z9s+-KKCtJT`;|Mc#Wt*|U6wuVyR@9L!d2%~B#P=}Z8u7g*abaG9+YYi%f+mmZaorH
zj@+e2>g8!S#f=`Bg*CgLuU-&55Qrh2uCOm+Sv126+m-VkXe+BZw9fnbqd*o)AqyWZ
zghz<{q-^&#PpK~|s7;0ly}Mahn#mjFfU=%!2?&h%7D1SE8PR9+9x{)%r}qj!KDX#@
zYH8`R$lCz03#?{y#B(8th|;A@JB9i$;3Vt!!{rsPR<(i*BpO%Bw2QzgL~P@Sr6MOs
zx&HUu`$DBmRr}$Vth7?o^;&5PhiM_2C^zErBX~X%A@uOKNrJE*qtQ&i!OZi?B2{}0
z?)(@D87V`apIHhkU$J{0w)fkET^L^Qd^pHrBR~a0LrGpj%R*yFvY>on&d-Ce=XD|k
zP8g-QzM$$bX|VTL*f7gL<{6dpfNA9IXYa0EFsI}o(%00eFspAPiHRd9yr9{@G+XIK
z$o9wE17Ww*bn0mhM~*P4u>lNQ-zjt<EQDklY7(x-nyDm$fa_OVAOijUMP`D7ZtNr6
zrD|uFiA5XELVEHZIvySj`GM=bf)l*GwqNBO{DQ+c9zQ-(sz-dB-y!mnhha(4qxZ5;
z;*!w$_9@ewUJMh;r^`1hR-BQe(9j4t0H>P)|36}_^6_cO`gzZ%v`6U0u~e%g7cneb
zX&({l&m<5wSw5J#OBp(a*9sBCKppIZH@DV}diOiMeqHTqEr;T+yVI^(8@6tgQ~S+%
zbxgZpnYi>&cv24JP1@v}=9!t9RE6S6vj^Ok7PcvQe_Pn@)+b~d{`v)|I(ry*{40+m
z^CMlMX{N+k>!^M~O9rPbuEVUXd>za5xu5fPsOn}tY3|6cf~7)?)6Mk!ST$0z(nNK-
z?H^~in_bu=PhbxrY+5fg>4Qh|4mFawLB5<(5+p>pFDX3w{W0pHBslt>ABs%vc7L9}
zn?+8gst^_?>X)GZISEK&(J{JXOu<i0j4LNS6s5-YLku+B)D)#|%c)5(_(Tcn=9#cl
zgm4*=aB*>iMmVM3B$3f;?f6HjL{Wwax?t~*z>hF{I)s%!U(XoG89skmdH62j4zT~N
z*C*jid^$P!+JF*TD#%?aJq)73?sW*K4t!W9(1E|$x(8?Giv!}T(U3AKl`2KY34Fov
z6mHJ!1sfNbn$K!`f!N?NKu32|mlAuX;=lId`MOsd-q&dJ-u+5m9k!$4;~~8fm%$g^
zE+xpw?S(wcx{TwX0(0!q_C%@4bZla#_i~rH**nxfj-=5A&hs@BE7J52y-E1EZwaM0
zyFNwwujhb&0ybM;-IUnrh<(`oAY2}n<|ul8bwF%!EzQd3D_7w%H}O;dTRWD+h)Djz
z*x1+x>i9+b#Q=hi)v=)7HYE5&2B)Z#6I&k3Y#Y3{hY^KdE-ki;+VzG<`ekHWFJer5
z<ymNDWMy!%V;g8g$IoFt{{B9}J*^P~EK{RLnZU(%AXK8alb9NVQHCU8gPmPN#90Ct
z_1K@|G#E7)lcdZV=E7=g6GyaWR)6N1mN^Wq_k1fplBqwu?LIYiUc6{^AkrN~Pzr7D
zY57GcrKQ9zGK;4154NWNkNX&XkJv_>DH0G3Cl|5J2hj%~=WQ)97?vZd;Vywh0ocT|
zS<J#L-s)!G2y<!Gx2tzh_N)8Er?R~(C^&WGDAV0Vym{bE!JFTeI93^3I;zSu>t_qx
zo(t3aka}~id=m!TKwm_6il1>=b*j_@GP|beK&8Ldz2}_Q0>t^^W*Io*ymYxSMTFf=
zvb&jG#t<(g)T+2nXjX`!CLvjYs>gBxN9H%j=Dnw;JmR9ksLPk1<t*G^fuqL~jc6lR
zTY<3a6)#I~C(Fe5k*h<@HxoDKeP_fph6|sdhG`EaBE2{n>L~rE?;y4vDj2#tgbNCy
zEA2k}D>bEL?k!u}bO&YI5iF!vb~QLS{>Sp6d)xh43opb$KIsNR+Kg0=rg;yD>-QZ_
zKcQz%`#OA{pA+*DQ8Paw+W_NGWX&8|vJEmriFw-<O>P_ox_d@Q89E59Z7T%b`-hW*
zL*{-uoma%qP0{oNR-cWe4NciDJ?NN8#h6R)E={{v#Aw!&t|jz71BE)bs9m@G{Y>1|
z0rBvUR4T1TXs-Wq?X$BOjE4Y<iag9_-AFhbKJOt~dAVO~XKS%TNkB9|f*V$q#m1`f
zg42{kd-TvtJ!hAjd&WiRdOP44y{)iq_9&e=%|}+xj-@hy?{OEPWdb<~eyGe7x}ZQA
zBg0I03+xzlsrmYBf@lU2JFt8>BR56kN&0t(^8rWcxL_4nUO!?NSZ~kIHya<<&Lgb}
zVXbMNr<v*t43`Pr4{o3JGOAi1(wn;UFWa=Mg_V>lpFg99yf@R;RaHsgj%)h%m?9B0
zx+UE5!nRCgLO5RswmVs8_BMsRFQoBkI#!Jv|6+)|pME-o@l{ccUvw0?+93%G1tU?V
zuH6JeKb^s&Tjs8SvCbg<suvpQE=aOwd9tsGoPeBNT7~a0^9vcUHP3xVQu@`HT#8{i
zK8TCH(2(50KW^^yaqS!|TD*jwXCf0li`(LST2j~+KI$V1(i!F>X`bDIFi{Oq6DXnf
zSDc{f)tLvNSd#1IQh9J*JB8-CNmag?P(EK!c3XO?z{&yiox0&s*d=DK-rZSFSm;#C
z$BUC4;!Ee%h)_!_OH%46h+OdD&Pm^OZA17f*0Q10Him{aTn2vB_4Kwr-(K_7(zn#k
z6Y_oq_>K=mNj}(becw!KqEK_gvvA%)y!oCQx>NG!IM|r+<8OLnYIWGH;b7`#exe;*
zOmVkBD?kqwt|O>q6zZ(&ztw74^n`74eZ4+c@>INJ1N5x;1)it{lj0Zi^BfO-nwPDD
zB4NSS{ovz%>FSY=7{5)=WdECDHCtPFr~>2^2WBd`ZN#J~Nt>MoqalouNDMVsY{2Fg
zm?~+9%b%wk>c`QTDN^U^>r}dpMvr8y_3ZB119fpL{aF$N@r<POxzAtRwlP&Nk<uI`
zrEo}nd>?i|6q)gNlB?WyN`K9m->s)$g@=ycajDD_iCa;#k0MbNY09Lr-81s>?m3|u
zU!Qt4Ush;X{SZ6(R#DyIWasH0golZSY2HEE=+n%kN9v$mZCk&ATRjKOia-nS-(8bU
z9>7&bMy5+3cB~ZGOr1142?9=n1e)JBD!nQgo{ry6H|!4CU7JQ7Q`|YmY}2`=Tf#;S
zYvG!%?4Qlbuh)Ejd7z8Z7s8pC*9OYf^M_R>1nBp}cYmkKVj^!M^JiP$ZJV9^i{e9l
z&i6w_^-$fG^;WuBZ|}X_5>WPyRFapM1Kt<vhB1>c7?=Il7B*{Ob0^ZhCi%a@4t;Rn
zht`RimRB|}8C|ZFH1{wkfOzSY^2PjA<n92~%>1{`|7>>LTx4ZsXIh9XGU$VHVV&13
z*7EAlnuoLN<~_0A7n3wLo0&>o%}qV9Pl7M!F@sB%2pRU08${{FCG*YUl+wW_Alsif
z_V$MtQ6JGP=Cy4dei+;!QDlP7HKh;-B$eLE7L|{+boVH}*U!kz3`5$A%wBb$*hU=P
z8{BJ|eJjZ7S!d%aW93ra49OlzywG7FaB%ePZ@Di#Zv@I%keA~Ew6wH#p=U=_nI&I!
z*b0B-T@ml93)yZpLY}N{^NzIv?P8G^-OE>N0?M;nB1YBFOm}z4U%Kmth6rzDds|ma
z53hyr7vT-3rONc9#u7R1@Tl>ZrtxKd#xK|&l-mBZWX+^P&v7WSd$n{O89GXf@?m}R
zgZY!gG%-<FRY4U$K1rrJml&!<rdy=k1q)fX1Wmu${MW^xJ^7|vSPu&WXj1Ll3RGEK
zvL#TRx0E2r51=r@`z&zJLp^;;m(C<uy)^x^PjwG(XTMNS-F)Y&NHI)@LdAn$y3n?`
ze+o8jl4d&CZ1Kx@S2y>PYoCO4WSq+3dG7jI@b8ZOp3u#g1x?@TWIDx*k|%D6(!*14
z;D+-uT&we-`DVTEBKFNgU}n^%DtPbzXgce+Ci_5*Qvw1~A}x)8A|=vD2+}FtNQ~~T
zhmeqN5C$lzq{QfvGJ3p-45`uOgwdlpJJ-3+Kd@i+#P_-5bI%v4V35cqvtU%({hSr3
z>u-`dx*I^sffU0nupi7Xbe;SddDqSAhA3NJqTk%~(c8tWQFGW+D*X_ZW!=`>_~5A`
zAte)3Ta}7$$Cj+(iWj0d{2wPCk81`l!@6)yg}<eVb?#~mwZ8hEqSxks4`|(LOi?Y2
zkXm5bCE$ThwSFi%e5-cGjLzZp*&72x+agZNH(tX^T#u)?-}W1Beo+z(S|7CErghV$
z5u@}nkU8%;U1iuv=8K}i79!X_uafD8tS8{ntLF@vE;}{scmJ1lX`e{u?zJ$%U}urY
zJ8A%SJaYsE*O4~WF6mc!Bh$=hoFoc0&$@wOCVoX70k8GUoec}&RWb=mx@cm2PS>JF
zLVg^aJSsfj)cl9H{i2giHO7KP<gk-HJR<H$0(p6yII>-5SV96l$K~o<6Cnv;dn~aO
zoHN_~EpghZ{o3iix#YHE<v#v?@nmXAwJd@2A^fR81V5gxkaB&{1P*p{q-tiiYeUBf
z8UPAxOBOHBF5BbReo^<DBi8}x{4)G<Qe4ooz5RM~Z7Yc`KV6Sn&UfRTQBq#sLbGWq
z?)|~a?LKg-a~}-+0Y&#7!BC;sHEl?sH>)g32o$^lL*Ct8A>hq|J%=ft5sxWrAzN-%
z6TS(u?Akn88pZ-V5VwnImAU4^l`D8#*YZJ1;S~4v_2Am!aZZig%^Eg=>5IvW*jl+o
zi1)NYAYiWfE_9Fi=GO7?F%pS1?Y$rGy<^IfN>&DRR&GvaPNs8`^19EY=>#cwHYDoQ
zo)GDb7gdBH+S~n){%ziJRkJ=V2DZknr6nYMz9v4tjK9BOy*~_H*>!g#s??1gH?BY7
z?eV|=Bp#<R>n8BpTm#*GJ#8b@;dgmew}?TRx(d<%+lXje#>T?1{2d_26K35=LS%h&
zI1Q}Uv}LRI#hZx0mbdn2U8FQcxE~MY|6$eCd#7v)qcPhL*kb#5wdvN>(<oJpLcM_Q
z+28xg@K8vJr6e$MfQl3~CErKCh^wq_f6Nj@sU8Yaq|ZG8H``YDvhmp;BC!|Q0}dD!
zbNlMZ%RYwwE1TL!{k{V9HpIA!I2`?d)&a$35A0r<9#@6Z=}*U#B}tuCJ+FOcd@^QP
zj<+5JB_^coe)NKMbq=4uCmZ|<F*&tzKOFgJPfxEb@`iaTXd+g8ad~!s@(P0t#~j@K
z>*u>eqHkj$lV^@GCz}O1p)-3U!a+VknUQsgmwET6_rV9B^H^4-StgIYg(s8dhXII{
z98{6`z2(qkoaS$P8G+LYiMMYge6Z2y>_+wK{09cSd<~8>rOaAT2sr;0LXrr_f@H^?
zI5nZW4R-Oz!8Dh9R3?$bu1+5IQ1XKf8C`%p2_ytMyNCTo1N{>c1p5-Kj?gOW83h||
zAs^VeTfrqci1+LS@z(Gf;atQ7jM^52v^l$E_$rb5Bygoe-fsD{WMMeNTTEV7PO1;(
z?>u%6wCuI%a4^p5{m{yN7F`+#@Pc3}+WFmj4wJYx1;zFdI79sBNvV_IOV$#K4&6()
z2%mPM%AgI1F|yacrmJ)MUp@6;o>j=*yel1J(KeLl`pvhQwTqJ5>HFZYh_QAVDdu99
zqFKh2+l@hqx+fL&LyF$4YCjM^Q|g`k3Tw=FVz3^Bi`EDJG0~=!f&^|g;1d)%wlB>J
zJs+J*5e`Hno!b_)3$a8zT={yJ-!rf=Gcec8kto&rysPm}^QcRM_;?MM>yY;tL*XPy
zFk)T#X*Ig{8ft7EcDK1Jl=R`#`Xx%dD(j(M(AMLUpnE9Bs?+<i#UBgP#a(xQGmW#5
z!qLsYTd1XD$0a4*Wn?a3aOP-N(bes1d!0D@k$ftZNNpjpIY93srHB+ugB=0xj3YYg
zO`3DvY6DnhgPS@bkl&Woep~otRl%VS?iRreg14h3;kO4doC1Yv2p`(Sy?-UQS#mc+
zz)qfcpupF3jY7IelS3h9Z87n-GG3OHs{IP2y;M^3zk}Or(=y<hi9z=4-@jMg{O9N7
ze^a)_doJs&FCi<JCarP(m;K8WJz|3D#r7R^@9`N8<r);3hbPZ@v15c3gof#AzCwI2
zCaH5wKwwbkI-BcGmY#DWQho420ctE{hu;L1lOXZf!b?e>HG7q;!&g4Jd!5HumLp&P
zTe*!sNu*I3ArcT}Jev+PGO4OlQNp1Hhtkzj<O<;kP%}jyidK%Bm@HMuns~;~NZNJh
zj?~Ro@ayrXixKi-iwONv)`oHT=F9~BjB1;2aiS77!E4O<tJ?BxgpFiMdf?P?b-we_
z9S@5xpS?}{8)%!SW%ydo8L4saC3LY&5}XWE8RVoSXBmC|UHzLukzk<tB3-M_Raw^b
zf9k)%Z*F5({t5tY&A&1@ahY!5=o_-5k<rnIY^-GOhO0pO%2}i3v(CMDds&oOuYefL
zilwzVQO@szNx|-`*m1hnh{GZv95adxa=Zd%O*{ROld}hlAzCZGg=0cjQ1t;2U+rsA
ztxvVtA|_SqK}MVLmony7*1aL$ksb&d{P(qiUBd6LyKY^L+u~x5Dw5`~%RsEPXgA-_
zG7(LzGP6uGqd!(gGMy#U=#3Ry0rPp}!NCHI5u}^l%u(V|`x4l%@tpm2b%ql60D^Yh
zL@lg)FUm0F?_$>1hsJN4c>)h85)-Q+5RaX)$(l2wHgX;9DOa`q^Ezoz)eh{>nmxGM
zsjl&Nt@yc)wn?pG<OeB`BBGff`U$bp9rAXNcXX^&{lm@gD(l6Sm9F1F32ku4FPGkX
z<nm;aberc2I$*&SI>}!b+o!5@e^GL?wwBd=so8MaTG<@@ri8~EtYc!B4~-c-ayy?m
zW3s+oFMI)=?fyMHJgh&GbD&-zh73Wyw71~x=;-L|>^Xuog!za2Z=BkDI)C4=4F8^W
zchwLs!=iwrn~c%#IsXiRJmiWQr5o9W>eCxfyfbF!5BmGw{!BM~VDP|bm4&XXa86KW
zA@sL9f0w)_Z@iCVU4m(*W0{`tOaBy3_SSMySdvMpE+U(f-^-)+MM@jXGlU6TnK8#O
zM}4no=1sD;w!cHvf7kN733wT8aNHd@QQVRVMRF{qP2(dy9GDNF^NGUHiptK+xMosY
z^Nn_?c9}3(6P7h^Vg3}4%11u3Poiz%qls3xiY|o90|s^*$`&Dh>z>?*SB^nMK4xZW
z4BNR)d(-WUo)K?b!9MI@W52|EZjq$Mfh_+pHIW=;rM$*c%fnqf<24ZNX5qp~o({=*
zCP|dxK#?8%?g{Y2kX41fNr@}*Rb7fwwHvo#a<sE2<n?Y4r<T#;WbQY*m?lP@<?c;^
z&68hDJ27t~BZ-sQsW(=kzDNQe{W0Zok&n8D+z*oNxjGOF-sEs8z4y+!Ly%hHtJ7`@
zCB?DQUo(oKI`CMTK)507Z}l;Fqosx5ffSjPygJQ;CPLsXj-ArsNN~`Y8Xhb^nMTMS
zt2N<{x;`QXtAKqKPx}}WaF&AjlOZta9Z<8WbCL28fufl&WC4u*zn|-><K@%ff~S*-
z>epMW(g?naJH3qWliX2%=1fBK+%CwA^f#0^Ih{aADQ@H~&Q3IfQIeBQK&gV1>V!LT
z)+4aF4Ul4{s9q(&e;Xxiq030y&;U$-u~y2zQ?Fx!I~4B(7n8Ev(-6fgD(n5Vd)D&k
zx&Fcf4w6+#0(g5M;sK33_Sh5nI8NCVue6O3%)RqeHC0|oEpwI;Vpg`75Ps(B8IEPv
zcZF;n=k@qttZ$E4Wy6~8|E1(nP+Xsx?QibV0H1k=Gu!z#T7DZ=)bNi@bi3I%8S0E`
z6BzrtQdG^<8uk8YzJ#J9Fqg=&abqz=!T1)NwR-M1o4^C){bm*N+xYq2#zyTlnN@(4
zhN5Fl4c)V{h??cwtGSh{Q<Nuf#X?a))nbm?@g(*)E-vh)7@y2EQ3<Yu1Y!_HWzbuo
zImO-TkJokCB5-(pAAR|sOKXdOc&nJ}e2K+^*=SX8!EacZ9(7esjcb}taczTzE}o%H
z*_})V#H{(14C`-)tl_cj&RM6oKNP)K4^M6WxIxHEz+oD_ucoRno@mo5LCfwx&H_{p
z#)2zL;TNmBqw#}Sw{%1HNkzQMZ2>j#D0XSPENc+%Q#IlXdhzV=@GxEY2#MXyTIxO1
zSh<7Etz54nhLZ9aHsBpHQ`Xl%Utsl2=fH)!fETG7STW4nqMY<#sq1=P-0BCM;Tmg@
zC^x#b{9=vpx(^#`c}Nz_X6}NcuM@z-Q7!38?*r$?j{4!;AO|0p^47gpDhFUg*x-so
z_K>=Zaf^|Q3K1vKe{4oX+U(g2ku(_z?YfnseQcD?_C-AV(6c95B7(O;B5Ro+7+ld7
zHaHhFUV}LFq#vVjC=>qs*!MYdwR~0nn>+D{HsKCL>0hlt@A0oB*JW(H`)P5BOz;ga
z)i=(s_!Gy9i6f@N1nT|$jo*P*J=sF(=4>w_u54y=@!tWvBPL0m`s&m^@Cz)I?O#^R
zuqhAj8)=!s@YBPu0*PCWzYg4Puqmkd28T=kl{SRu!pY|Y$-r-&uLF#y@C#p(ud})2
zIt-<9z5V!vNa!Dz{OXR()=LUpJc5J_I{c%UmkO<lOk!>$M_<^fvGargpV;t0eRcic
z0?%0}W+D`MEnX+XoG6c{^!sfLibCSgAEw^e6;sspZA8yiUHs2M36kULIfRSr9BLnH
zJ7E*#nUwxzFYFjL!1oE8Ib>8%#gzA;4@d=9f=n^X%S4!w&W|-XL-Nn?a~fe6)&CN^
z(O1=bg*`kwku94f^j`oQ%?FwUus|3)oh{9?ccCDtFzhkzm+Va6gV8UT#JAbcVy4q;
zyi5fWdu;hfN`>Z@FRLvY#wuOFvm+Vac%In<mtpgSf3V}3aia8Z!BQ*Hm=Okw?0r!Y
z+mu|e_je^`416P%+$q#%<g@+Hz|1gtbYJic`{y39a8P2U^8(N#ZbX3Vg$M;zcOqCm
z(?9$E)+hP}>`9)J+LxqEUJW}XPYCx$R69+-eUtI#VvjiWBhsAaT;I)+ekOSP8EwGP
zWpSIkhLRq$+E+Ies+zTOZzzd~LYt-Fo%Akl?AxdbUquy`1S)FWAdaQ)0{0|kOtvBJ
zDS7&@QDB@XTTh^cC*4*TwP1pl#idH;<U%Kq9zZ$m7b`oxZU?$nF8<o*i}kQLhE9ll
zPU(VAkf*im)ef%(HT|LKy+{J~d4MZWFE(TEDV8)OwNl$wxW&kyqdc1NTI+Ugzl$tp
zZ7pJ^mB53i83OaGqaYZM+o<q<G4q{Y#&g`zR0d&r0-Ubx8}uBnpV1ZfHvK&|NY;5H
z<Dgh9_Zf5EvIpt;x9Co%syV~98q>CNGvQ*YtS7^0-45K#G!%{=ULfO9z_CsQC|}#7
z2V+V2HFgzc-P?7Qwz8@4U*LX4%xxPoNRm8zy0iLkdi0EnZ)*3rJKNi}&2@X*zY+dY
zDu0J<fr84tutp2Fxlf4;?vllmBd1y(8@yg{T;t}wh?O&|>qT)nda`)19nG+g4(+(|
z(e*usb!#!0Q3a1pLnN-$!ua=MJzrJxp39vZc%;>2k`8n->k=Sz0T1jw=_t97G5*^I
zpv0Tq4o7vuGrL*Io3SlHG?r6&66U|r@j?-3!eAdZmpdhk{B&}6-*r}ii&ZD71y+i&
z>ItoDJxE+7dgXGJ{iDh+liqm=_xQ4M5A!F_E9~c5%tHIGTOIU;&=0A=N9u|EMhnuP
z{6wE3+yi@F)?3i4ETQ?dId)~)IlO<2wD7Ruyj3Lj`t7_N>}o&KPF-H6YSH628XO9s
z;8~?L`SywJuZk%@c@sGmp;OLqc55l6tCAlRvu`AhS}D%{&OF(3DlfXm3ik2PXYA!%
zCs!JC%8Mb3F!kJ*oI4V91aNbn4zE>sjhMX9;;|2P1@k9wIJ3z6jFX?io*(PT<NIfm
zHhh8u&MxxT0(zWVT+`Ah;^$WfRd#?W>#iWw-CnhGnS_C}l%}1_*vT-deeE6z?MKaO
z8S4R+R{cwMWrD?hMAoCXL-Z}~*2c<n0rhiPrUB_%JbRI^J8KFjL#7QJ-nUtz>M-&w
zP|ySk4cHgfE@tFPH1ae`7-5Ez#d7xSXmK+ncEGsQcx-q*P7V_PXBPqr2A<O9SQg!D
z<=I^So*U%ymP;2eqi3v&06yZkf5FGZ&-@>fak-hPBqX>4*ogL^G4dkYl2Ucj>0mBU
zl=^zD>s5~{!A=wj*a034g4(EmS5;OkgNPNK)VunbwmHn4Aze3^y*5Vh^F*=}2Y7vY
zbok9aMl$?MUbS%H%mGH?VubW@>e`UbuF0cKZg~#l9f5GpOQaLuEZlQ_)Exa#&vmc^
zU*4%x!qDov^ILotG>Hz|7{!QFuy}hifC!MG$0LDV+%#l3ls0tcX@_r98rdZBaKcRu
zp~CO3MFhK$CTMfl&y+|f9VG+-UcWRu0nYxwx3NlS(AGjc(AE=pf~XU``K{B&rF#Nf
zWD@t_7dMChw2=sFw-D)P7^pKxvp#)#H(wIIH9EQ%OFC>^Z`li=`EV-szVNm_PyVQy
zDP+3!TBHen8ZKQ#wp}wrS8vVC%rc2%3R3rc%cAUDS_spqe@*>h<@$WTel$&x<q^-;
z8EOg|d<&TYhK|>=vMyt`Vh*O=#EOz<JpQ5(#WLm;D&=tK(Z6$|DTxw{l%4435{P9D
zi%dIa_ox|e^5k2L+WF0_aXs@skt!J<bK01Rk=7IChoAw|sYtcczMDI)=~g{(M*XqB
zf_9r$!!HH?&+f)CTuMU8OJaiSel(+cR{SxlnuB`4ioZy*{|dE}#`bZ}i%Fz?mznq_
zt)D|OX-vE4zL@ntEzDmR^WSHqmznK1KdV+Q{w#X-T%H#OHzMU^ZU%HB1g6F>efU1o
zrwXRTje4cm&S9-AWXo(2J%B1|z?!CKKC%qStg=CV^>F(ye%BexyC{UL<)wCsR2A4a
z6T5~b@Nf`8e~1j7V$jJduVaGI3o)q<VMA@HMANO$o6@=QKxIMR)q-i%{eF;&{Yuif
zYW%8!FRxN%p7-02Xqn?aXR!xAX!XcA^_^?jPk}kSFOT{6x3TwRmnCCrK<J^;>>iw&
zDua{#V{2|Y7^CmQ?e)cgwr0<(eL|c)Bo*M5e<Qt2v+&co^QG(IIYsZ$-3w##7~bct
zu1|h+LK1ixFmTS&Gkan)4FaoDCZ4aE-P5amj%w==<(8S4UEqQUxL>)2_rCL(Lk&5J
zNJuc!?;ub&^I6@q;nZmiO&QFml_%PBDWA)!ggZ<<p-<?W)`U8PF2JjDr7|JtWkAdt
zJ@GABA!DcT!#;FQ%h=O5r7@vN{#!M=tF<n%^f^<R9`O*4#76Q%C4qLb=r8E&r`&^~
z4G^SWX#kD9-Afglb|zy^_LH9l2pU6Uqua2YlBu?x@$MM`Rg9oj@EcVT-5=tg>)|>q
zOIO-RTKz<MuP5pMiYSYS98LQo2>`p|n$&8f*;hQ4cM1jC&$7;39Fji`2$5yaR-;~f
zv(?MIhbs3E!cRNfpa&skFC*ED1H(<l4PC1js0hoTvV8-7{%rGpGTkBGfDCgSI~{Yp
zi2q1}_g5SS(<)qfS7?$#0K#k|n^MUsd6|l^2vWEAsqZg(#kGQ@v-OjNI1oP9HaR`L
z_CjzuC?uj7liJ;!kV`M*wTdetrLBa#w(QxapyUSB9@=THpgIr0C;oGcpqIfaHbp21
z1hL;u_>pWh>lKboB=I-5WgC&;yP-TI5Voq)-4^J_ookwnrTl~HQZn;3o>YyVyY<Y|
zj_-?U5#4VI-v7PpkF~xB&aCgV1c9w{B8q1I!iKA9`*r4L)Bl}It^WGMk`RE%S##ng
z+zVt`>g`x!j`+7fK+PF9qUlLumvV+)4nFK)4R!Uty10YJ`%PhY2u}VU`Fcy8zLbEw
zzwm(GJ7HHExS8<YFzy%0yg<3Y(@E7eU2{RGOPA2&UG@w)iTrTj{BZ*YW?%Nz=hlx>
zqZU6Ze1<yKnav-hwQ`!Pm{UuOTI02waG7}?YfH(;4Qc(tl}E$(M~6t+JXORwB||?u
z*1s~qh<%|F-BzfTTN$-K!{2OQ=Ulg`uN*q_St$;Va=+Ug9d$HU5)M0KLNl`*yc&x5
zm*si$tKkk3l&%RlW*r6jVGb;HZa0W9i>-@tQi-L17PjBZ-658{SSNO-Dzns9ev7C@
zZ1Vu6t`l=T=O8*kE6rbrVRvq=z4uh-k5D5LUlpUUHR-*l)QR=zB2Ig4um1M$HMHrk
z2uTPs_4tYVw2AudMfK~=z{TB1u%sIpW#3D2t?WjvSq&hgwr<L@O8W8RW<61|c9IQP
zpXs!Fk+FW9+VSe<jq3_@KS5&sEa{rsfy*|Nmw~Y2eB6K8RK;G(Rnjh(8INz2>#YE9
zCUr8Lpqdf_Lhd2WHfzx?qo2yT28gD%1+AfMRLr~tPDSU%eDD;1D6U4XM(f|s%C<Mw
zDi#bsc7d9h=~dv0=ze*w=lGhnwpccQMYdf+mRYOVW81;}(bnoM1&ftLr&7~VoY9GY
zUoa@a`_(*C$Rc^%OxBD)QWvx=uP`Hqgxobm=U%8LCJ5`QyDU>lxMMpUro+siJ7fO*
zUW|^+Ix#}ER(DlG;v^5s4J_2q3_E!8|GX;UfFv51s;fSxmyMg9BbqK6kb9PF%WJ36
z8tB-8WdPPRH&0eOsVXaQb&z1eEV~!K=GSEe+ZN28!2FFHG{$ip=n10GwRVWV13ztX
z6hn3rbiH!8J{vTWaNw+h<841m)5UV*wd^veWJc4!wO9Y%>i(&l=tm@^HQ4pIAfNFa
zFmrf#@%MDqtIO1r4&>f=iFGH@OP7a_OD-<%+$dEPD;kxpFtWn(emGX@4{@Y#+1WT1
zu?-O=xa>}XQy&A~aipds9{R`38~shc6@QC>emAkuH5lZqG6;PT<3DZog%-@i0efmf
zenZ4J*0r?FdSx3c?tz6$Syq*=F9`u(+jR+({o&FoUA4W-S2*-R$E&g-O$ITy6!%AR
zmU7^pi(li5%_{bxY5f5Z3#09@qO@ll(9v@7fS&wU?B`iZ)vFYDs}!i#>y(u%5d3OR
z?!0j_z_(S3O6E_=(>huQ<J@|bJy&PzBg6dEb=SEnsH@f0Z=T-klDjX}<}?76Qv)iM
z`886S{d(wcTF(_k@aAH%dEP01?DFOmvAn2k*8-|(nj?Qnm6sMH&oWtA7&cB&pQ|w>
z+4e@Se~7esPM|`NwL6kHPr&jQyaaBnFp}7fVU%7ZujJ(U%InKOUQ{NL{nNB(KEU5{
zdz?wUo=t%<zSARp+)c<HE&-_6^F5lW9urS;pdnn`647dc|K|vKw#pS9BxLn6uX&V4
zmKTO<{hLn8P&el~7pQW54Vy1V@6^Q7m^zP5%uj_KWC16}_eom;#mTRTZpBJ^FaD<N
zJBFSu49VS&l|T>2PH<{#5n2%XM;rBO&HNF^?Qi}R8WsA;6cRr$>;l5~w(|56RWRsy
zwNpKRLcWT`4``>35l8@1*v?$(6fVN|i62LZ%chV1F%H!+l)wGW;=Q!>jc0vN#8w)x
z19Y84Ed)HWYeoA}^C128m4d4#1>Wi}Y7)>UKW58-*PN@(1d<UdNUhMII@|h6h0Z%e
zZ~o24&WkEarw-WX1Fcm%(H-($N`C)IZ!k2;t14(dBu5H=vE_P73e~Ua%!=t35mbDV
zTXzQb_QR7Afp@by#0)1z%S6c&Y|BvL;<I{_MX5Q!)Fj67`fFL@W4zK~M|J2sq<4)R
zW)9e!0q%S6e#za)rJlcH{i7x&72LACQj($)Q}Oy_IBmcUkkc4^CYW8m?{aqIJI7i)
zA-gdx|9}yE&Ul3~P77=}Qh5*NjBJDZz$J#Qx_VF}^iPhwGr2{6Xn|VJi@^}M$+F(g
zwNSzcksb|S1-%{ovL4lq4Xn*soygE(Cvh*eCk2AB<uyDBQqF#fm&Q9<(AGN4$bjg~
z&lVzz+ZZ6UjYY2A)vrN|lR)KIB;l!^>6}6z-ovQOU!#4#LovFE)%$(X)Xz7gGI5Ev
z!CCrP&hL(D@k5>DPGq{Sg*U}!ZKJm9^7*?qcjI#dzn)r1g`W(V%<<LmrOZkQ#b{8b
zLx#U)heh}fv{1ab*YEOIk0qQA1tCf-wb?mI+yL?uS`Y~%?WaimMclXfL@)*rUacn;
zW#E!LUP~O6>CT5ZOQ!9i08i1NjoG}~vOv#eK{L+JKSpj+f1cc*_uusiU*}~DAsFo#
z20LA@>QB0ejs}9KBzAV|e351!z$2V0T=I%j-L3VnZu_T`ua@RM*$6Q_LWduIRNknr
z9sqdV77-F%;)P=VH3qu0TVajsip$sVK+mNhQ{{>D9g+5+qaW`6dO4a*d^!HPzEmn1
zJSOfn-78_2+r12L?lWy=>@6hw;PiD*H?-R+u|2c2KGf9BA6G+HN21Wd`L%e8lI74a
zqn61^M9~pUf$egTh~k&RKh2VN{YSCmc>sAT!E~(i1RKR{{;Ni_;*7yQ(QtG4d8J{;
ztmTZ-bn7<<nF7!1ew%OC{<Wp5&*{rSRS$4BqURjSbVZ?J&(I^U!-4j|{~!ydAh6be
zQepysMtuFjIeF@rKy04v4^v9-<$hs9&rP)cw~y7witpgT)`82FBNdr@Y*4STQ=q5|
zLm}|*w(kC~C#yBM2iUotTM0=a;z%A?L}QD^2Smsj9=Y2+@s*Gw@6OA+(Ch1&Gaq5b
z)MA;?Ch;P&_!#$Mal@?SK_)?K&%ci5{4)xO0;#d+v}!?myD4pBWGRG9o_ab}EDFcn
zlTz*BC<8m5BzpAOpcnnvRK)r)V14mZCxQQ}!}{WhUya-y43nuKHDA(wfA`o1ukQ&q
z%R5o1O2Vn2Au4jvZGGzRY)KifugZPvCe;<PQQ-+(FBWT5wg(Dm#6bCz%D$)$bnaVX
zJD!=~Nc_xo-VS*k_3u`NTOkz^7?uH{+QMg_O1t;kPo5LY3gfXYswjQ|FmUN-L0mbr
z<P3N=)Ct)$Y|+LZ782bdN1nZBm%v5gebD$}$DkYb&d6Z42-~O-H#i+wxn;f21MbuA
zc8mpk@#?P*F6(YJjC)QdF_km7<9TU1c@@pk=?@gDrI#VCZ7$|>Uo&?#d2O&-!@ncR
zn~}Iv;>}RdBMv+ta$aDXuV1;~V+`)Y=>SIQzoGw?6KEtKL@KQ?qC*5sn1_*}us`HG
z>(IPRT=}`V1to80d?|Nn9V>Mz-pc&EOy%VN7+lR%@$@yl_nyuaoteD?h0KKLB(rC!
zI^i+|3u0llp)VE*-^cPwD1S=-M1+=M`y{VyE3qy@HmLbiB)JvnbnU=d9SEKJ#4*%P
zO<<(@J#@MCbufx-uE2fHa7U7|ZMVA9EqU~0EaWNSbMgwz0{$b+dA5yKU6P@>Vl$sB
z$ds4Mj)T06E#YsqIO7N@jA3IHC(>J~=l!?|N-_^&W?wt^6w7cP6(*;7#Z^`KVK8<i
z^YHK1;G|>lv6~fOT1gc}fQ#>$A#AfNBLpM3J^U^0-DBzO3C47yr(@;_nLpR(S)7Ka
zS-o;4z+u3Hk4U>h4TK}NyXB-<1de&0D{#kdwj(DX&Y#t}K-5EAOe^z9dvflv{Yp*3
zK`fZ^WVx{I4hdYZ5ZRvo+nd3L7L5HZX>RvsK>DhFFd!XYX*uAXSd7gDvWRD|U(}N~
z;eD0R`J1}GKj{TJocGxnrD|Tj!|a$}9n_)nC^9KAgrh@)R9Q>EKAjut<+0&y;Dk$)
zWoZnnG+6Z}!-s@l)CgD$5Ph#MsNQ#7?e?XoXnxbs=Z_nbI<Yy~ziBhG=MN2+07Cp_
z$cTvR!NAlMVe{>;3tN$i!klMvBXKal=urv#@kyy3DL1hexxy#GNEG$}#L*pedo>tO
z!P7iA&l`b8;<J!n5%G8;+eNbfGYS@CBy<_eufDbt$+b7l<PyR%>`;(_*LxLee(b?!
z{Lpw~tMFT7^ANUU*zecS=6AnV20bdT9)s=yqMAG>-Q8FuBFGQ|+%5>IAB20w56wFk
zN~4<CY7?JX8>_CVAQ;kP6&C&=rLj_s<jzAJ=t<X;43ctge1uwm`r!1M4dcdHo<O1~
z;i;K%48K?L0^Bpncf$>Pk@1`JBmAdJDVY0P{R=CDTi^_K(6i$FroLn~eA8eI2(Z4j
zTV*C@7@a=q*-!E}4X9ir27C8YfggX=qh^<KO>QjGjXHr#MN7P>&TNsDVysL)>qIkr
zOLw00hX*2hLvHDDj+?k5<-fPO;^cGU?PTdb;+kxV^nYfaYkj>B?8$qKRaUC~@>Ak1
zzx&&Ji@{Qrn{I3C#DUWsbogB}z#6(gyWu}S-!s<qVtmq3BC(|SWvpsSplK)O%oCYR
zm{slm=4fy3<m7so4pu1`KS;9{&$5|Tn@|0mm~z@|(FMPFNKZ;@SUk1i!>kdCSsInH
zO;)~+J?T!vnOc@4cfcoD$W2k1Wx$O`^^d}7bp33yu)UEgF^w^pM)xD2DKt@z?N=jo
ztuWeMx)1F_+#hj(K2-57J>nH2=HNPnwh=;}q3MOFc&P}QJjZx-`F{$L+4CAwYCn6s
z--4lhAe{KrjzmW9C&xyu@GS*3K`s9Nj~Tdx&j`KUuTYU99zpq6<T!9M8znM^JdUU3
z3Vk*8v*BG*B^Y;XzheWCU?C&YA1md(lY5-A48y|>ycY_<+!m#bg}&%uJAgF0FMHW%
zk}}Me9U))DD9r6)-=i`-M12ob{#&WiUc>O%0#62i3*OiD>d_b)&MN(Z6ueUVh?LxL
zTL~~?Ixc4aQ6!biUV559Y?O$d>x>b%4$nxQ=NCIUYdtu%5iJI$@+_3q#8^@>f1j_o
zwK<FJLJlC3va((0OK0`2eT!I2_)Kf7GoZ_gBbabiONW?Of{3JIQ{m7}b|qjxrwwpJ
z8d)S}=$;g^{=};0f1pilWXa;4o(xw_Bh7xgul?ykEJ2=bJ(H4(@3-^FL7qK6`*p_F
z%nwQ%61h^7ssWGCWfx1sU>X(FnPZ`6SADO~*5Ap(wi!{rV~%uO(E7Yj2Y)qa+;Ao!
zbnt7S+4Dxu!L#t;={ARlk!F~L|6IEkwm~)Sc46;xY9^)B_3dx4J3U!@zfD<(jzYT5
zEZ)ePm%y>%kBfD;uy%3Yr1da};qyNpNln#@^`2dEk$`-nM2Y}2E8d^25J~2=bI0aT
z4B8>)>q!s)IF{%6^xsqyzw5#?lEMFd)yZAHw|43}kCA4T@XUBcdEpn)I@&f&Cm6O-
zh;FfL>ji!ww*`C8rd9V!-bi2W-0yh0xKq(*_`-$z!M>`iU{FHBE)GGN&?;GhfuS7c
z(zv-1kj;1Kk-CN%-wioh7S~T+HvQ@K>Tqg#%kop*Tbz+1Q(*$nw)^!LC>tfe#-MF^
zIi^`HAS9%+^uNd6gi(D;BO8Ny9m}~|Qx)I3kn~L~BiFXn6HWbWM&T=}aT}s3*#w90
zV#KV>K?`?YGGPbDjq+xmt&#LZ#{;}fKaYGmfU!D-uk=ygMnVRf#AqU1rsrLGhw6DH
z-FE{=(-%fvqY%r9!eG+|x1Hl{W4YUt`%{hk)61Ra55c|95jw1LPrtSBE>=4$f8lSO
ze5WB?oq`~bR+WV5&?)o2WnZ!{Pd~$Ui54$)!nchb<){Y@7U!%xE<;&VnP$#p-zsrb
z0h4U1R^3+?UCT7%Ij$iqn0JBpb;7N4gc-tw$!psD4cQ!ccC`q?jEC&FLc{~Si4d{y
z6)7KVcim#HWguUH`B9co(#GJZ$-AQvC_l<bRRJe%1gq5{LdsH_x!N_$M0MFEy%0fm
znL-IzRf`h3u}TB7I21}YppR86C}#jntv!9Ku7!x(vnMz?LU!s}yS|a3zs}X~xlOmX
zR&`dv8<8>12x%U!S8yC^{0|?Q8A6E5*ht<IlyG~`iJnwV`7Q42R9zD_k0Of=6yDDw
z@_X|mu342xX2=c3+l67$ifLq@U(pf=wEwj95P7?u@o_bZgglMh&2wy0Y{dg8&nA|)
zpzJ3POo-fEqU*gF>Rl2V0{(&kpEuS<4(7Q}s7y9m7rS{$>dak32o@A9=s85T>G3u;
z)kp%{ON5Lsd%PD2_KF)R39)C}=4M}_ux?<KdO}C17NS-_{Uz{VcFyXl4s3h|#FKU~
z9OX79^C^00{uNgqchrHuXA8i?Lr*dAUz9lWP>@@5vWo4Oa&ymrACPr)iSHYvZP=c?
zGNLo&djJ|qY2-4n3LJ6BpmU~IihervPS6tSb?Zc3%2XcJ*6LBv+LoJt_TO1ZqePy%
zMTg(CS?KZU-j+rlvMb!YTK<8;vcEz$xS@?dp;cLu!9MAdnF<t7_w)zQ`FuUnx-;m{
zZ!KmO`1`ua=bl39EG2>zj?Jz6V!;*B-ws+{lXs8|NKBnHZ^6DE!s}>3bEy=H?}+Hg
zr;J>2Qv1pzOzY<m!OMlEaQ_vC<%sR(5|2)?glNw&)YB(~YT!BqiZ}7&ru&?AZ^YTZ
z8<`ak`bL#BR;;Lx^}~S(&}k77#-scA7uJZv3=*-{8zs$IheJG2MkwvYxi@!7tZJ@7
zSN}Gx@5X^`a7t;P>F^);<W=|;BZr7QtlQ>pU`^@MK~@9n(*xUicb1h)#C~AJoca~A
zC~L+6Vqyr7H$;gAyMqP0mV{lP+g}gAmF4SHno_={S%bWY2M|nr6$y$uU@B5v{76?i
zp2W|kBhvB2xFCMAK_N1V%B=D0nHr*3QPMrUTq3$;9et&!#CpLL$0+{9Uja`84;Ctt
z@&uX5Hi%_v^9o^|b&WU^XC8DC9O--?`TJ1AkK=EVclpDkuAbe8DJtT09ec^I6yx|-
z(7)G<%zfv3fxsPv3nYMzE_kOS2(wI7wx_q;GCq55$9f$^sX}&9J#rEjqaOh63h+i>
zRoI%lt85%q^1l2r6?UG7!3c(Pt)K)ODg}FQ>h5d4qdpYAQi`+pGv4kv<w|irsdF~l
z4fRZ(h0R-vZKcj{X4ullsek89?B>K0AcXD77<N7YdACd|*jslGazrlisrBH$F5BF4
zXBuS}QFwZi%oM89VI<aom&Oi^gUj$Xqtcm~#}w?i7s)BP=qqGeJ)j#${eXyEK-VJH
z8%&mIFG`=vlNCPGIutSvPAXL;`JNuELdFoh@SvNCXrW2FtJK@URi69Ei$Xb)mn@iK
zw+{{AL@?6_jr>wzCJ49GA)H_H3kf(DwXVxf56wDSo9_1$^Mwbxqo9F^e(Qh60e$9w
zo|3?deBUv7wKM(V#0hE}ms!jhV*p{*wxof>_1HmP^~CDxYq*q(%wZR^HjFY7_imx#
zc7ehh2G^k*YNfPjHFU27X^LXo<FD&E@c_bP8TJin`#k?XZZ?zt<=Gd*!$eeV+XJ!V
zZRiHayoO@AV0tgHr$47iq^0xC(w`H<veYQwDa^QuDI0E&`7I8aLM(@gh&+4mune}-
z3W?N{R<>C1`!63syG3;RX64-|f!@=B!EWMTUex3d{_Aului9wNQkf6h?Kf@uNeNKJ
zYU+M-|1N%Zv%1=MDV2`P=)2{Gb^=5+vnm&UTb4E1HgwIL7}8T<$e&?YWm`L$suXBI
zDF1_&R0FT;^Xt%=c|T9Hyc{A6_qNF0mJYwL?gclyuR)Y0FU6;5$9Y$kt=#fX24l)f
zpDR;h$kD~%XJvaq0T;V&P|L}ZULeBV@T{iWa@$bik49u(5v|2gW5Qn>FoAjk7RxCU
zL_e&yc$(l7@rGi+slzPaD(yUT2wO){Mg3X|xrf(ycKd}L{jw42_T7niZ>W}OAt4ol
zSw_o--Kg?mV>@9x6+6~JVLr!Iz_CN6J>~ZrBIBYj#w#Jdr%plF(7?+{An-wBRTZe@
z>S!w5pe{;yp*I~|G}U0)b+~!AQLpS-0fL8W)6$zApZ}UJ7ryk`M7GPoY?{xSOhfp_
zltzcq*yxM>emL*!=GyHE*ZaacR<bTNYAdPsn;mghzaNrFCQLyKBr4P5YUU<e%|Yt1
z%3kaIsb$WCzY9kh>!-WG=&Pje=8$ECDYJIj>MtR!b*@KmN4x8Uiw|=A=S<eA#SQ0!
z50hWP=^W_1dj93A@!tAy#(Kd+(;%-H6@*fiWN7SP2D3g?RN26Cw`WToQs%9sLT(Ft
zF(sVYYU|V)?&BRB&G9{q7H{YuVRDBCsGm15qtLRlPc<=-jkaQdYSt+dN?|AljkvuU
zy*XVy?jM~8OUCQF*SP2V#4d=rJ5N~S@*hpp1dC|7cE$w@;`%Sr3_VjKw`Y0N$tszw
zC$uqQdKkciCXMc!kH~X-i!+i>IY5uWS0tvAMA!LR&Z$*2^?#c~Fl|h8D-?)7@u5zP
z&!AS1<AM!NgjfLquyoQGu0(2%56Q95Gew*V3P@iUPpqfMp!y^PBsfgVp!PEQ>S|A5
zw)sP&LS*sk+&^Ar6Fw(a6riW)H26~IVBxhAh@jVzXHgY;_H=JBDX)2N`qBkD+Sbii
zSWifbWbji;{>3$38s&cCoq3S<tc=SS8nD1M$e!2LPg4UVjY;p+a{Ck$+p64N^eilg
zqQTo1#O!}Nd7AXqcCb@Pl72`u%O7Du0v<MWxEOQq)s;rYUO^yoHu&LchE3q@J0&#G
z&Z{A+HG{cr9fcg=OrJV}G88>2B{JUH;Ost3Pxt<Q?UDI>w5x2;8@xJn>ssg^czf1H
z&QpfI>i|N~U1-?66TgYM_|vu6B)T{4q2gMv;Fo@$VF3{b-HZHitI#7t(7ML`O<Q&K
z-$M0R1!g`J*u#uZawA&UXm9&Cg-+_no)e`orT_J+^hyyM?V^jPh#0(DK05Ggb~;%?
zg0C`?+Qv{^jhXa`*pOq#a}Vq;gsLgC<Z{W&ZGZ>Md4*lm9%0>{nV7dV5cw>Jn4XDY
zqWBa4h^?_%@($fn_r*QGh3l}(&CSDOh>*GoJx$FC2HCsSuKDiq)cono!Kdo~-#vDs
zeJCSsNn;p(bnInXt4EO3y~J2hF@0BoQbreau8)2Z7?-9fkqz@{6=(6Np2WhCe2<AU
z!VV8@8CSQ)jqg`qSew_8;IR;TFP<s2*1LYlc38Wdzhb>b-(Qf*om^H5oy_v(fv$<-
z>8yhHF%kEHtjkFmy9fKyCHFTw+f^WM8dBJsqa)V$Y6bz5t>3X}{AE|;`ggX6zH*pm
zvcm%+R;T6t@*Y+-VeOgE1^@N|>o*D3U3V^xfAeB{{n0J^6GIzrQV(@lYR^>!G&W*f
zkCFr@wONonI<%8MGg9NlvhFWa>pU1BBTo5_q5a<)gC(-7E`7YRm2zKO^0f0K%qcQn
zg?Og>h0EG_w@GwKL7g`Gc^|bc4wYgYK(a(}15Qc882PXIBsSRutIJ$~RzW>O<)EQe
zcHN_3;`#tMlMX>G=C{*~WJ3AxTrfVn1Z74uSq82ud^(iFwO>au!duW8Y(;xEJrq-g
z?w02L2Wi7LMK`xg&EsNh8a;SZ@9iVDwP~?XPoG>R8&g!SQ)iim3iP15lytzLFf6$9
z@_f0g#&L#!&@EkrN(N}h99}^fRtA$)=-j^hTM1(Z&wN?cg3<oQZH5nUtUjGxb_<M6
zA8~t?w<olJL`ZCOV>fBg{ierAPJEkh(3wqDpTX&9&O;WN6CC15r=pkDQX0xM<OG7!
zo*k6)LT94FP|}WU4gy+rA!n}s;po>WcXwvuuLN|F%qa>!B!H~K-28L(bs;TxXNmXG
zM9Ht@1qOsv-zmmg#6DNA$c6<d{p1i>Pl!*%--+UK1y7G8n8l1_NPhompWQg}#jTqT
z5%PpUGf_xeC5z9rN-IstCRuT}(1(6x&r4>a$-R+palu=0s6A8jK?1XERI@zcC>4p#
z2)zWG(ua@9-b%pIWR%&;D_BUK<WAp5jJC-}<d;GDFK(~BM=MY%ENNK1ikzhg?A98*
z<C0EA92be)?TzlFzNl&F{vB8}ZMou)2#`l)ompW`=(}qnfk?}yW*)i+zY4ENOTugT
zwJRb$p`j-obSpuq_a(jeyJ}V~Q`=0Wwo=I6pe>iVGeD=Lm`iO;B1_#28>E2(JN5Nr
zh3KB^yK6MY66EJuO9c)_huRgjjbLd&6>i;GGf^otKFpMVPEPJ_WNdjU@fjrS`XA#9
z=%$|#)2K4UMl~a{C~`;!`ska@UJ_sWiH)DkhBdlC6rN?RUCdwV^HvP}@MPL{r+rYi
zW&@cAEP$h}MGBVZ-0?C0_C_}(QnR?8roa~6g*7*n#>#GGH4b<E$VCWY+;%%VJHhkc
zd7YDMLXBJM=%R?eFGe&X>&47z#!P$&Qgz>{EkjVHj44%v)4PW@u)ck*cWK(%Vh#>Z
z)t<Q5RPIAa%8W3Py1B2TaQd+86n@K(ez^gFh-Douyo*0^gN-pQj;PtxFuQQ}ct5T8
ziAu)o?NKbL8gq#t+_(^XIij|FL>ON?yjd(f3kUWte$6sjU#OZ5hWKBV^j<YQ+hEJS
zT9`iF0j`?i7z&o<l7p0Ic0qd`Yg=XWA9`*I<!%v`VJ3(ggSgcEK%aC?7TFMlb1Q(%
z9l1;)ZLSIZO13sV-#ls~#6&@hs!ROxqH8dQ)N(HM9lpjUvukJ>H8&;xGsO7IjD|7y
zx<$if2AYLe@IO`s1QV>u%^Q2F!o)cuu=Eps^(X4~tI)SfD&LCEB5Cu*;GA>hAy{T|
zZg*>{3@%*)8_oj`J4#*#;to08tVUpIq2BldbV)T6*g<2lwP)f$==>|5qsba$gny4f
zs|aQOM^KXCv|kbXv*^)(iVY7wxuJJ2I_Rv!4o|0NhGv>P<<oiCAG~VKdCvB8PsnwS
z2_elLGj%&bS4AY3IIXoz3mifKe0g=gI{(jkjU*J6MC=s#xI-N?X5t~@1wb1jR0ST4
z-CIE7Kja(a&D1{+r@%?xAE5`ci^#LFQP@&4){IwYDhSOl&LHS#s+=bX@L|L<Dl9^y
z$waOrElaIVQXZkP+QPc+;|Y7Ud*LfSLPS|dPWjv!WN#Vbo|AAs`KW*eq0<iQ85+V2
zlIMk}^@*Pluu)Pee<8++!mcsH_2{|Erhntt(-zJ?Lw6bl`mgwdnL~zU)!4Sl7HBQ9
zK1RqSr=^~!raD$t@uacHdci1&2K#N=#Xmp)-r0(;qeOtj7a<@x+?2h0U_(kZ8gmf3
z1I0wBQ0-}Xwjf|8uHAtIVC?4MV|9Y4#8P4gIrG|smC%3b-p@W01N0)GN4R(YF+jhe
zn&yfy@T|44HYsuWpTr6LY#dFNn!>p0;#P$U98LV2AEq+SHO(rE1*W!xKQq<xkqneK
ze_0I3es^i3+orYHss`Te!AGO2cUu=*tDj!F3CYK?e3_FKQlj#L2byW4iioq1Hb;0w
z=%qTZSJ!l(TN>+$0jnIu65gr2T#qxc$ZaIUJH>}a*$ySoY$@r}Y{feK?~iOe$66kH
zBMfphOH23jHwTZrBpCDSd%V1OB}Cw{Br9E)3fbh@HrTbMJ5k-6e>Z+Y1>8|?i5%L4
zKCW?tYj@K_Gg@0za}G(V6&udrXNHTuYV%$nP^|}x$aaZjfIhK$7HJ0W`|Q#2l9G?C
zbn<fja;LsTFqbH+WpCoHf4a%Vs1RFuu%CGB2X$0*-l6(=IUmxT!ED*FuyoIMQPF!u
zI;fCaeYpGU^e9V3OV2?sd*^bI^&UGY;Y>rQYO|0^saj}2*y+v%5_|*P!s{rmS}&JC
zZ#HgJD%aTn&||)4(SM5rxM7z%z5hO6{c(*iAhxvL0)1YxYm~NpQ$Q9Yr1s@xe9Zk=
zESD>2NwkQ_b|Mp(N%PfAlPHM;?-gzNx-FZJ3yW+7uR};gs}U|fZ802`fomX<!%$KE
zO3J2b4hmYL{vH`mmI{iA(Gy`JrzwLmtVZ+6&!v9_va*`DT4_X*oBU`#*SkVynLFqP
zOycD~%Bzk@svNRJjYv*0oFGlT5dFMbMEI92^VOp2nf8`}4R@zH;XyqM1`DZiS~GdD
zWCL0u+aNj3AkKn<Jj*U&4UH;w^U06t&%fRC>NM*)Ttkmq{GUyYpIm;BVA!ps<@j?7
zrJ@%G^s+DS0quPi3*>cApnhEQe6fLBmxa^JgN!kQu~5%Ngp(F_V<JHM7wbRzz_A0%
zKe>nn2_E{lu!oV67Xk=U21-*qiJhN6W^W5if~1R*d+u`NJUShP)T3%&V@;{=6i*h%
zOOJG^gem%B*Y%|o@hVehVZk*IMR#~5Nn?bvmWC<t9rS{-4u%;3>_FBWADi-1s>g}7
z&7Gbj!CoxTwH_VnH;ne7S@!7<!i@vmx02P$xE^jqj+?~7C3a*&4p}4~ckLyLEznPv
zCD30d$%w>_aM5*X)xJVVtuV#NBdBrkJS(zUQb<hy;;~6HbH#9xzf~)sFD3d9%Y!ev
zzw^A~=C0kZgKRzoBF#6z`&6<!yAQj02*G{0Uh=2%n^11)A}uM)i4s{XyQCuxdi&|c
zlq|+e5-_1FxsA0B^D$~m+H#WFq;5;H2x3<L{$VM2u5H<|QOMslq{hI$=!}Iy9mfk1
z1zJ@TqKO<~5W!2;S95RSk_AF?y$~K^ne1#%u9OcG(y?cuq9R5H3ASM>uHkz3;SH*t
z=%n<cyO=JD21_ule+_n=ylRh4W23S<abJvN(_ZReY!a%HR&N3H2#vd#+s9||Y7LY+
z6^4j1azq~#h(7hH2#I#rtC&yUao5Wp)P0cr@Cm&Nv~{sb>}M_K&a^o`y-0rm2p>0E
zUVSP`VFqa3W1Ae`ca!P~Icpy?b}*MFP5!hMGU#A}=|sf&(vnc%q>lV{|Mq)y)WziE
z{gw4ie8~l>B=qi*X)ECDMDD&hjaBC6YVKZUvb3T1WQQ&_tw|`+P?yfUKKkr(z#D0P
zJ=!MMQ_uh^JL+1wE>`m1zoe6^MhMZDJ97{@{{nE>sxLo`N^*a46R&e#+`uYNYpYW$
zy_o*dXH3)mvHbEa+CpDhWGwv)bcg=%_guUvQSgl$yj{6W35-~;qdS<yRzo_@u9%1h
zYDnXWbaav`-R265NgDw#E)#1*HoHJRjV{OU*+P&2ddd)TF?k>u5K-Ig5Bf-ea$n1L
z0p=FjvJquRZwiyij;sGk9t|o~FB0OkEd(EP(Fdm`{QNHib8tQOS_jT~LY8twqW6^{
zMZ{*%y_ogg2Tz$0htB()l?!7vN~M?<&!PR{G#>B2%#xn2+lY=w9O7g7IknT|ABi`<
zP2bu8zaaJQU9;y5K6@5bc=N3Rq70!koj|vG{C5v?c($+{$Kdb#!NtO#e_E|D;jVQJ
zi=ibw+GmDs0H!B5EG&$JuYj6=G=qLVj(X9{(V4RREgt^o2M;OJqS;hqaJ<vW@D_=w
zdA#_CvMY!^?9Z~5^h2me=XwLLLu(q|SIzIw)#mhfh{}k_2Ca&J>t}dsP-M2V=nCA<
zp?UCxBYho38O>Fw6p`Wvs=%MDUk%k;P5fWgvLkUL*c`vo`Td}oYr)r7r=%B-uGJFk
zwtA#UrJcWSrWh(*2<cx5y-f+cyO2)uf#APSFOR}uV||j&o+mT%GUMR{Ut=VfqqbWS
z75myB$QOb{t6a&voiUdmv~x47R=t2VO|1Ud*5@=y4Mn}kk4`NCF%j0d7T{vvz}YBc
zbq06c%-I%6#+0jakr#bV2t&q#>P9X$_KJi)(}T(xVRmp4cv66tFLB>|XDP&R_PXyO
z+YzaZNVgYEB$LZ(SSROKQzhaDn?|BA><<Cj?)OK<UDIreH!b?tO(KJOy{-Ui+TPU@
z1akGgI-T!Oc6x8mDiOye2M(@+V)}SBQHlb|@)MeBIFt{l^av2|ZQ1Kysmytx-T#pp
zyXyr;WdAo`D_$&9FBjx0baT>sND5q}+y_1}Qe-l1i2NCC^D5BWJHk8U>Zbqklb?H~
z5Bs<O2O&Y;zCB}%1Yfwk>$<J!_Sxlef01X%%*e~-JPmrbZAOI~G^_gRYH011pwku@
zJcmqyRtgD1XoW>km~jn6LlEYeVldi79W$4gW%Kape9|ut_a}<u#pT83WXmTcrGszF
zww^ENZcxluOHMo;ZDcMFk<TwKjM7EVhUl%$o3=X%ooF+!{pGSmNt6B{_&6N(f1K5;
zx=Soo)kV}z9Z8$KmlLJ+HxLp#W1Z0<1T<7N!H0S@$gkhLwp=YVqHn|T@mWG$bzTBT
zgnqvVz+JoQ=T0C+K~@^jUM<sTI{1}e{f(QquJ7*-KmOp|hYz2=`PLVo9zD9X`_f?4
zvjEHUS=V%B)n31L>+EXr=F4}Nf3r(F>d($kukRmx|NW23I^KQzvndTR`j_t>e0=|t
z;4!#i-`JEqC%Ad@uyC}xIDYTF_ipSs-A6kbeDw6#B&rt*BX%7qTVT8VAO5TVv{0w-
z{b>`C`WgJ@ckb`ry1w+VwX=V5|IwF!@+TjE`kUK>(b4f-sJJ-0oc7(ke`<TX+u|%=
zF-Pf`gCa<15cfudd)E#rwC5LB9teP{wgCl`QC6?`;^K6=HE<br9eZDn#>2zSshrJ1
zk~#!JWCT)rz1)R>S^*1dWe{mCCe*^4TSLOQxR~u-+r2tH-W+bc`%nJy?VZspll+gK
zT*TBG2iDl-YMB|+)CprYf4i(Cf;DDVmBZn1RWA_IhRgf?;iCtSlbTdj7>=E>6n$7N
z=9B)wx^%U?bfSS3{D@lgAOc;N_765m+u7B8ys<r+6kd4Y7d>lLuc%V;i`CE-07@7c
z2^FE16g(rWQ!LD^wAQ3GqE<?wiV~R<DFRNA!l3A9Hd`&ri>vvae^=kw*x7mb$-9R>
z8k0mBAt0?JDDAQ+2_>xi#~}nwC`u_oqYYwSBO#!3MXx{V=ZF*)v5<-qQDH!dApt7u
za^aYKrI55z5u~juW{w;Y!B|a}IEBu$Q7X$!Ut66yBmjtrXsn@0KXK!IL#TlfQ7M6H
zxmsM#x~eQ(wzkC%e|mkROyg^i#E=MtY0XRpQJ|C+z{C*<t=2@k^PVEB#DUo}x8BDX
z5&>JSQVJ3yvlgn$rYh@v(2f-C>}+og$4}%Gi9|rwH{ySZ=RE=xAVCr#U_d~_Vv}e9
zgJ{;)E&)P76a@AuR--}o{K=!u$>iRhd)u3PxgKofn|ss!f8%FI-+t$vDup`6^W)?B
z)!B>}H8sypAAk74_m7`GS<J7>YSsA86}{>94l3RH*c%QR@$BNFY5j6pGJ_$T8b3ci
z>t<yU==6h!BXqix?ZL!>!3d5l$=XPfQes40r+3$|W@F6yr*qC)>zqSUtIDtHX0oxx
zfW6*eHk*a$e^;wjp63Y2DQP0B(KoAlb=j87t}1snr(0WF{o!y?HD{Nzyw^`DwOzL{
zooZvd;Ill7AvSHRv;ly!tdvrOWUNv@h(<B9F@^}OwbojN1c|NED7ad!+PZuHy${Q>
z_9+Re)(0d-+9;((n3+9ulYo^LtrvCWQ(U&~{l`z*f66beRw<?DM@N(C#&~OT|H%W-
zu|F6+dHQ^7cQ@-58{1pP>FYNSzWE2=y!Yz8)3dY7`TXK)cJ0QETQA-DtV)B^y1Tl(
zy>nnXxj4Gq-q^i5tyZ)4?D**X;w(nKb#3?Vot>LoyPJc9LADogd*<!4C7&;<x?X(#
z?#}g_fBO&aKVG)+-48z5-r4IgH%f0!#!sGHqSC+n&F`LE)#S4A#%`Vu*mw7e{?}i9
z{U=`i!YJ?OPDSrE>P#U>$}E2Mwb%LxN$I1->gjB8TKkK-nbqy%lljxDSw;Gz=O^>3
zT`emROrhHyjlOW}*4>@0Nv3+1GJ^^^)7q$Ye-=)RQj%C$thT_c$R8eT{OAAmfB*A8
z_tX1({ln{<dPtXjh&Q*7=cg$Kg{Ca=$@3Yrp?us#_55<yR$ZpYA(BSdR&kIGp~L6T
z7WW^Vy!yG<V@gHQudC<-?`%)*y|i`v&R~!yr%c@>pM*X;BiIi+(jVquc<b}ZIMmsV
ze_Jo_>|9F;XIG1}%QMQY?G>wfwOp0nw*aDv2vJh{!2*oL5C}k%AyP^i1hsBd2tbUn
z^Ich2A%rZ?t#g<2MPBspyz*KEs#~8pF@e%5grJluiUE*`0aIj1yzU?)r{SQOPDjJx
zxZfWR`h$Kku+C+9-s|;ygF$~dRN9Ine^>N$=7=<bLI6!l5orJmoI)GBMmQLwvdpb3
zhC(b{zBL_Br=w9nHyROGkvJ=WfhA}|MKA9S2SBu1trm;fs$7Q98I3N>ilQirqHVi1
z&I}2V1Q0->vI@0AMW7H6)_yA@LR3xLMP>mM0R^;v=pccVQb;LCB2t`Sxvc7{e~U3?
zdA7B+nK_qYP>9H)0j<IxTI)5E0E-|20SW;q0S(NE07yv`11ix7NGL|};of9xV^AF2
z{oeP#$1%HmZ##2?4Mku3{8!$4{OBuR{nGExPG={}c0PB?oGp*LWp(!C`1Hwf4BRwj
z1QejQsupD_BF?#|M@Mbj7QNo)f6nf($jkYBG8$ja7p6<y+3e(MX}d%qZPi+B1Sqmd
zWQLSkgG4C?03f0@fp*Qz3n8SGw6a?3EX!K&n~<&+We^%oHl93wbo<Wj=O<@t*bp;J
z#uMOBuIAPfAWcX8#bU0NCT&hHE?XbSsAtE|`?<TSs*EXevd#&jEwZ+4f2+D-0By8m
z46P?vue>FtP-{&f#%hHKBE}dWT8$csl22V5_V*97ymxkS+4@w~ZD7F2A$ddwkciX?
zSVaucWmyw^1Xwj~(a_)fyTAC~{I~zb-rm;f<?K6u`W?*l?|=JSoMdx%|7yAP9M3LJ
z`{T*#^2}xK;e!YJhlh)PfARFu<2PS@qlw|ODhyrMoSvPYo^>Z@l~VclKR!mV)U?}{
z%GK)j!HwPi*2(4K@%_tyi(;~I@%(Aic81(&=(e}A%+cTcoB#V(&-~&qfA94#z4FKJ
zeO!j#%Vv0S{s|}98Re&oCFsfVdF7d}E|-VX(e;DF8ynjN4Ss9(fAPTfY2dWZ0|m|V
zJJ)swh5q3EXIBZ17wxhOJ{|+95E4WIh1TkE&mft@-7U2yc}q|v2V29kt4`*t7DxzF
zh?;<u3Lyeu8v+`Vl(bR?@b<y<7ypmHvGA)0pT7Ufts5&ZR{?MA?VT;#WlNxpLLCxa
z%$sGM7FDhDzU|qgf3uZ9oQ#U5sj8~+Ni`kRbwgC<xx2br_6wT=&1UUnxcSisA02Fu
zOrABXrBU`r7Ox~lKymM2M--l2&Nnx=x^9t>7F83=iniR|&HcsM`T3KRoyk^~Wtx<*
zm*|mMp;D+cX#z5&)-4?oVG+i)vJpdyz_ISyEX$0w&z?VDe^u4~;WaU~YJEx&qX#cY
znn=f(LZaY30uzeDv<C8Ot&1X$0XY~YkYx%b0+2CQ8+5J*$_VBr8<Id>Ho_o+MjH))
zpo9|&fQTjqDh1|s>Vyals(!CfT9eYcq1O8ly5)Q+!U9ZMH_&Q>5u|R*vRc(m)rF2U
zS!b2e%ILbSf7X*#e35g<_1OQPtHdgX)O7(suNR2_*#fNs6hIOn01gpyyZ(T#<Ns~b
z%r7sE>-BPDkrT4_tztkBrI3DDrPqiNkpP7OQ7K>$At6$d1T~{#Agwe?k`!5*L9uah
z_8hRwv+V5bd~d&2WHYcgwzBVi_np7}OTT>Y@ES2xf0tJeKKgj&7xRm&6ac$O+JaRk
z%Z=$3u7?OA1g*{beK=>wqfxzDUOPOTot_$ip>>n~plV7V(!5<Ng3dQUh@`!5dL&tv
z2~zOvy=S9hj0o(UgBMBFM93iot#t^&W`ombr;i>#xpla+xxLec7!og5<^JA&+qN6S
zB6Kw>e_T~pd;15W>#R1X=a*#i&c~S2;o;$@_wRd+Sy7Z#wY9Z*emO&hhO7YFEFbp<
zZ8$ogopg~QiLr$wZJlOLAOfgZm;u(?&utf2plRB~lJy49PfmJ+5vWBzuoT!j*LJP2
zClqFds5r4C4(wf4XiLk*vZ}hHqvZ$hf4H?be{#Lv+2xr}{%3#gUzuO7vMd{JPk-w-
z{^^}NcZZYd{YQ`9c;mJ0>E_wV5s`ZH?H_w~biBRynMF9z*tv&i7gZOuVqBCyx5`q;
zdmDxAEm!T?#r?LXi+K}-s5~LWVPUjMgUk+wy-)5xJDH_xJJZYB9xWKN5rLeX9_?>u
zf8#-qK!mxe>doP_=*xtLdz0xaKlx|eBE9>Ye{j~ex3;(DPapWY6C?YR-rd99x|(nA
z?cV?J(UXfg6EQM_j|rQ)6M#X_X_BqU)c7iOzHPaqR%t~7*SDrbdf{W2qA+kuHn%bH
za#=ZL6cTE5O6?Bv*~R0RU%NYQ%;4VZe~*uz4L7zAN82BN>yK}2?^Zl7JBg9ouFGr|
zeFUJUY@+vF<1<USNu%Mg?Lym7+0<I;)6+}s^mI^+#s(>_mhC4GpG`NWO$S$1lh7&@
zq94&PR%@O0@^`-bCx_Rs-@beE{NyAh%m?F(^I1xGRn@)i$;~%E-?rR%*_&(@f38@~
zS4|zeP-#+5I}*xFhDaztL{XTN2%rEak(fBe7*oo!d_0|;E|%w4SNqql9bUf?nq_2(
zO{J3|Wk6|85;;YQk^+KgU`FZp21SwedcC%)qXC6vlsJRd8m*NzPOD7k!>m6MvPcBV
zur{JJg9uVcf@<A5*MvY4v`_*he^vmLe6(6+d6pNsvA8O`1|)?LIi}=8N)nPz;2F9k
zRTrzyN1!Y#T;JuxoC!Qfg$Rt=kWw#-_&=Lh#~3+@wN7g*j3BhOZmmbD+89&-z@*S+
z);hNyZgXO1tpaSD${Op88Vs^BAp0&c5DKG~A4a6tiUEN|5d*}O5};xQe*mNuqe3zY
z2sL4dguK7E`_fCdZ@;wj?)U%rwb$Qx_uVI3ySr@&Dq`Q_U;UXsd!tML-{1eWCWXV%
zuxdj*zTgmpagk^qL)IT;!!aigT>}7lR;;C+N|Dx7mdmZ__~i7cSuLC;=8zGyBFRi{
zYz%%`u3}_oa#ZM?jWH_gfATzAmvv^b)@rR&N-@R|g3%5DLI_1sfXhvmSB-DGz>H;8
zw?T}{&MwY(wnkk@SyrUBZo4qMDnq@@Y^JmUt!Gy!AqlFh*Iv1IdUl#fm$4Is{$Rj9
zkS5P*Rn?Q}b_g-gv($DBNfQcbg#^e36(X$vRhikijI<Y2qU^_yf1d1YO;=60tWL|;
z#{@~hGaE$=ltQO%CXp3JO-U&geFMr_Ys$)N2PSh(Q|%t^^?UiPgBwlPyz#l$zVn@L
zUtONZ6pk;>=S};wKlk-dKKb~Q2cP(^yK(L2Y_$SomR0##m4?Nt{pd-l%plKP=on0@
zqrbknYoq=2=~HrUf82MMi<K880HeH!fl1+L(3ccXFISHAvJ7~mKiVIcZCEUp+Bk&t
zV_&#6KbbXMd$`^6(_!$f>B8Z3`k(zL|KWqD&(E)pmnvqX(P%PO-9}ZlO-lRIO$$`D
zO;!w(k5yBP(RrSsu+~VLVu~7Rqu1-oGS5tqz@V0OLLIHmf4ZhAv_83*ExQht)kte5
z7rkKWq7p!kd;P&-w(;?IKY91z{f(WGSNi!!kIc3=DLU>p#P?M1Vl^iq;8aY;SM#eF
z;&9L#PV(7ow(^_@Js$ST)v*OMN;yZl&8k(YqM9w5{-ih9B(1y=n-@DLr!)9LrsYR)
zw9+?j+^(ADfBla>Xxb))c>VDDs;S60uYwm{G7cx3FTM44dwOMh<HQm`CD6_lS}Ua$
z!a6nng0iS+{osU3X-+`|VoG&gD{F`2@vLbsFK4&!-q|@gc>g<3N6uwd90Dd0idq|O
zASI3x;}WsG001BWNkl<Zg}IBdD0=;V-&$*oQAL(~e<z9a%ot6~k;rOojmeeCmC2OJ
zLuf&1axMieoV*VRNF+k*OuHh{xyhxIm{epyX5k2uI7JSLL}@funb39u!pMSBbxjwz
zF59Z?nmVS4M(0_UIcMvpWL_^wDI`IJ7fS{Ju$CGcBkF}{g<00H{uc}!!~}whl-5Xs
zTCE$=f9vMNdIH%sb;yf6*Mg~Pn}j+hanANPMq2-nd8{_F-e_Q;_1F~=7=T0(C?ry_
zu2IOD(;>7?S1*=JlbQEE_+Z&^*;b^LZyH}VtFvc6ardR?2iJf9yWjhJzx2!h^I!ij
z|K|VlH+FUp9<3JDd|`S6K*L=Gi73$;E3t9Ce_l#L1e(M;dvS3RLfy1w?p&TJlulBU
zx+Zc0L~FE4$pVm|B~nT?b<K><IYp3V*$dt@b5Rsp>!xYCt}BYdFjJllhl8rF8Xuw$
zi&cFwTR7)ZS_1gC?-ZeLL&WUlqTHJfV@e5yW6JWponLKl>=(IG2m247A6si@vsuv}
ze=4o&wsppxot+oG{<Z7ZtB*d;9EkWLcXbnEj0BQMtg|SLfTUC#d;|dyG<I1xuBz)4
zo2Fx-n35n0r`%Z%tq8}aRl+V;5<}*UXv3sq7b0rcbNQgRx7+Xcvd_Ko+OPlCum8-?
z{JBp){kW`GI<xtpe=%Q7m3{Q+Nz=5mf7xvBVBbr6c6$2itFJ}>neojNtE$P2N!+fQ
zMna_A<;$i?VMV%UjPqS6J7M4=w{571!+1DQKxNA<vxCeK=!By}KY9H8D{tNNP|YsS
z-hY2SGJ`=s_jT}HJ0AB3%KY@#z8af4`KCU4^6c_hdXwv0FWtVg`49flua}k2fArz`
z*#)Q+Qn|Ca@w{r1)*%EA$r@*hj0lF#_7%3v<s#%wh+XINUICzvo*xsybbD)ObIh|P
zY7@G+woq8JzAi*zE^>E$^TuELslRmp*~ceGm!U>8w4c9rd-dqqo3~!u7^a`TX0Q{{
z?xVBw@6FD-ft}4tn-#11>~L?qfB*8n_i?$r1kNXetn+EHDD&}9{QUa<wr3Gi56_?7
z8tp<E`s2ZTxokQ5+NXaBYSm|Qv^KBaeQh?seERg!<@2NM-R;%#%KJDTj4QGrVKzSv
zdNAsZwq8H<%XV>jwF<G7XthGEP%Bm{YRs)p0tpC7Ddmy`g_=TANLCvWf9a~q5h@N>
zvny-+dxv+w_w9GCX65E&NFb;pDr8RDpmUl?m{}AKC&R&{Z!D2w(z)W5CNi3sQUXed
zk&No~qSkqqgEBG1q%4uul%*~SM*s#iAPfS88Z(_zuwu!FBvlE5l9VJ;Vhc)1SkN<Z
zD}hN+8Eqv<-89v_YFBk^e>(!~Wp28$<!sS5OA%yAfULlU2wEG&^+M8G6(S-ENrZfi
zL8wzMLJUa4*aax9aRQ1&p%6<%M%2h0IC+sqQf7$8!^y^Id@-9_EzziAJb3%Zzn102
z5Btl?+GrkF*5PVI&<vCk3MfI1L?F>X%Cmfbf4Zvby3T*>OJDl-e;@tP{$y`386-$q
zkv)6(;iupF<}3aEKl|nvzx1WA{*QnCudC4fo&WQ1fB(TJPeTWjiF0mb2N`A7VUX4s
zBO=C_Jntc5zu!MOJLP2S<z;V_4|_QP#h83oZEs8n(0f1Z6`|`C62_Ei4FJ5>Kl=od
za;FJZjL~~<t&K5Sf0L~{If=QhmdvT`x{J%Jt*vPxO0Rt0%e#7+m@ls8H?Cb@m9xdN
z3sFWr*ERLUVyUeIlzVsYuy5y=WzfZZJ{Q+p)pgmlli?^cMpAHDesp}&8xEbzfMSTr
z8Ut9gO+_!V#R4Lvl(A~91`Uz~l~U^2=@~#Wpq(osHw*|Oe_CnV@2AkFAl3YGy0!b(
zSKex>^6}BhY}F3?#meJsakZ+dP0qgYjlX|z?b>8Gy1cj;jK=f#FNdQMAl`fB<@3|4
zrj1$VdV~J6=SRD{k{89YtbjkW!oe~1j2TXb=a<#8yt3A|q02`Tqf$M)%8B|#(F?WB
zoFZb+LT**Yf2>d^=jE_Jh+VV1y!x|W`-N}(!(X|5cx}FzXSup{us=V$Xggodf-8DT
zQzTj~7XRwc{`v3x&L5nepL~3LzDlY(J?g8^k1So^*=W;M>q`Z)T$G+yoM@VvdFwUm
zB?lC?gieXW)HKzcO%;REnlT+;oE3RLAqM8A?G#91f3=Su2|6HB76qH8Cg&KS?&9)%
zY4hRde)1RoyPI!6c=El!^~?W{U*cYxl<z&aTJ^Q=+YD`W^ZLew+TWU;e(v_x#iD7w
z+v=I^vDK=WkId@$sxaBXjlFk1dE$Xa!*LzjyjRp$M+bYGnMO>II3!6yQa|tWkBnZ4
zlDL~)f1WF%gM)*%>*iNWYi*JDQ-o?YySSV!Xw|zhU=6!B538oSm|sL+yZ%rsO<JKM
zt+g@c2f-yIrL-bL1{vGBUC(c0h+x#lWaH%MsA<~If8niv`fI;3FXwsT0<+SnH5oR>
z<eAHg903qG4hEyqa7!tzRlzaTbvqvPV-!&`f9&`3Y-F+l>D)LMlXxGMphWOKFk@s!
z2K56?$GTgxh9A8Mwku^#0?;vdoov59;Mg_M1A?-~BqgL$67NG@HC@+v?^8<s{$M)Y
z7z_rNmzQnZDy1~7r@uf%h@@Uz%@racN@5|xb-PA@l_ISbvuLd|%<+ZK7Eu5ab4n<h
ze`D<0RvB$<78v@&@z(a<v(w9lDJHmi{no3mys<8H|AM0vp)e=G^+*6Y#?%6U2%$ud
zUfWCqDhUc`d$TuR9KZ3opSZX>V^Fi%Jhq)A-*Mfu?$vvr`@>)Pr~mrbf4-P*N>lyc
z|JVO^nO1F)Ycui2Om?<n;}5PMESu%le`tH7Kd5H2i?j3U@{EjLU0o#h&1$Y84qVYL
z%jswwk>@@%O_N!ND4|(dZ8&z!$z+2?)c}A-O44T5a0=Sl$)vwpEpuy}(F`#w+*&vu
zASWzbX08_9@!6-l*Y<7S*g|h^>^6(JU)HCOk9(u><?`9-{Ayl~ZtU;d%&lgZf4$r-
zE@dznIoE%FT9yrX@5%I<rYWm78}++Rirf^IVqK@0+8DFkW+ov@idZ*3v4B!4MU#Qk
zstlY`LZp7t(^@wOZQHs+cil2{5t7M`u^|;%G8s+>Tjs{rbeKE#Wm{Q7<fKv6_0`U9
ze)jzA+V14mja#$D>T|EX4c0zCe?J|JcCX#IQCBsCx_RpsD%qTFc1>fHKG@xV_VhTm
zA<I59OfeCWb@v|}H+32gdjQZcdY7{*LZ?w~AMU4mas6NywLU&MN5ashASoZ_y`G))
zvs5jo6a5G8{K?T(l@wc(HOo%h{LZ~EeERV8>a0AQuT*ANZN1UYc3-)RfB)bgAKyP;
z=6%CBo(#Y7+u!`Dy9d5Y2$n-!)D>tQ0WKGF67E|Qn3E=h(o>3*QVBznPM}FuUSutI
zlgae>^gIbDrF7diZ5L7^P$r|U>AIL8q0O_7SqUhX{mGPD<jZABX=5<GCOy9ld!wnQ
z2%K`pLfNz9v-+(5@~wT=fAHJq&kv>pGsrTeqm$!n2m9Do!+vIy`t-r$Jk!UsCF&k&
zQo=>v)0z~E%k!!!n=XWys;1@7?u+?fd=N3jZoZf~XRWm%CKPDf(C=mA@dT0UvQy;R
z7=t7;$fD-U#pT%LN&^uPf!0bSy+GU~HrgPOaP-kLvvbb3jU*;Se@U1-cX~BnEUxx$
zT%YXjzx(ZnS#Aft!XzrlDO8HJL1&G1*7k>eLY28;mgQROWqq!-W`^L|W;sKi54U@h
zZRHBE7Dz?U9Fq^Bi?Qu|3T$ZYV@hk45+I5&LrTb?L>Vd&gUD*RQqJ~=qh5bfwb6+N
z?b;9$OI0>yS=Dvje>TmUGdvs)N5j!-KI`gAn2FFDtJMqDA0iR~;@Y+X@PqFaL?H!Q
zl=mJ0){;dbl*A+mEP@OSAP~AvtOFK_9Ahf0rW0d(!`^6fYdCR(jcd2=T)Tcl{V-{q
zj-UjQ0U0PH5m7=`Kw?ZJgs4y;#&&bsFC5J-&;H`S_H*C-f95yyzPmV|jmL+R@rDQ_
ziJQASc5CPMotJ;<Z~d)l=6Y8Bqu>9>iy?mb=f3`M<vXHBj~>`e{oa%JU%h$b(X%J#
zT1(qbihk}~)pS0#XV0JInH%<dy)5T0tX3;YytlJ;xheq|c<m*V;b;KDWVEras_Hg0
zxk0Va8EuS-f8G~aW{qmw76`R*i{&yedgZ(}K-y$oT>|QdkDiV;M$F~Ov#X(V(nL+|
ztU4~+HV8zZqqDPX`#aO|=<?#++B`;?E#^&|x{w47A|a47Hgg#<AtIp&fU%~~#`~sg
z>zIU@U6vbTASFd0xc(j&=NvQpwgH>@)+^AV3(goef4sfBRcO1rHRQNTv0BZ~2U*rR
zzBqZ-`BW$AV$4QEwZyh=Km6d+!9cz8>h1H(#f=*;ee+x2`s45a$(wJ!*|p*N!Sz4<
z!#~{J-uc3tZ$5eYV0ktF%9p-;|I>%d`D}J^vA4VbS(S#!a&$J6WHf2tblMp2W8HO=
zt)kb%e~0&<+`hR}F3PH{R;#F}C@a|7+g~l~t<A0Zv*#~uZuW-lhaY?#BD#L2b#Zl3
zxAo$qM>Aii!EoX+HuIJw{p=e*@zFp0olhP-qmAk5@#W_~|8`ZC@BIF6e}`p%FmR+$
zfY8(-Rc+`3$CMC36AcQ>!8;8Akq`*5>&nS^e-wQOF)pv>T5CwiBCD!NAgj6^<VDwb
zPB8(+#9bGZ(WBk~=F7V`ZvYE4{>eul{rWe4sj%km?zLPgtm@QxhzZ(mx-mUHJ^S!m
ze|r0+Yp1sN{Z)Ha)|g1xJ^bkO+V=RozEZ`gVw?^OpXB`N>e`*_$Cp=kZd_ko9TjGf
ze^O(#$unXRt8(=(p%D5^PD-(^7g=VAR1Azjr>7^vsFlt0yy%Yyqf{8zaYUzg_ICeZ
zwOp0Uo!-EbMg*3`LRsdlwNZd2t)FGJQeE&V#&se^Qi?1fk`)<2bN|!(*Is`4=Dk;b
z?{|OqY(58~!e~iJ68CyNXN}eh5quYmf5JNFilV2KI;vMOM$>nkAThG-=ab#R-XRzl
z5K>I3ZoT(l6$^wUF($#3Vpx00*V9pf#ISCo6Dp*X6mr+L{_=7-o(@LSRTDtB6oZOB
z#86jVT{p|+D)_X{l4MyHW2~DMf@qCDaJc?Jn)t%b_k!fA6agZD2qZ}YAVj3Ke@=--
zM2TTwU?$`>HUSfBZBQhI7-Cc*#ipr3XE!#!__d!LjJJB@&8!$&`@`>Tlu#uyNoA~t
z;3kl7L9vI$Zaz88hP%D-mh1Hex${k)tIf?xjB#sgD<#?4-g)_zSNpx8FetJCsMRVC
z>|gn7|K7re@%7Dr_pkl?FMZ+Of1;b;ef_q|;NI(ZHx9OMyz#Qm>`S+Aj)ub@|LWJ?
z`th&6{Ko5t*AGrlPX<L^7^4A%Q{q^)wd?1BqcbXnPAfI)4+xN%1t5grz2}saQtN>R
zyqIn)(h6Yx#AK8)!{L~jomE+3Q9(ke19v_>d~`Hl)}U=~IGr9`>x33{f2T~IK%>%a
z;Iqql)h6Zoi`qXqK7aV^Xueo<T}vQFLE*GQ@NEYGq+mV92%%e3`kc#+wTK9uqVG5*
zqP0j2nZbKcq>alMgax!U2x?FawzoDmH!`A9QZ*V3s%7QK^zx!!HQHvLd0tkwSNIe&
z=Zscu+wSe{-Mn>sdu!`Yf8TlM?YG|gkN<=JNC|8VMkmjYR#%H^xiX|4-G4BdjDGc3
ze`UFv*LBGpi_cCah=58I0UMQtPWiyAYMD8^HR<;h3&Z!`f4s3Vy!*;asP*Hcvz`50
zj~-us{L#hia=E)R-QAx$1qRj7S(14C{3(O9At|j*rYgT`Q&SY`f2F;xf9Dte^8flj
z{q11K^{V~KSAKkFb8njWuk9Uh6WuZxj^(N$2K31hI^qmrKq_M~%(+(1sJ@}0Gkv2r
ziY%A3Rb*Sa>nZ9x%PFMj6Z4{K0B9}EFvf<&ptLqNFf5kkXs|&_lMnCw+Hd^o-~C1p
zbW!?2D~H8K9t|wJf4%9}oom-#dHL2C-@3QIv+?kQ`(K?N?vUtc7mF?>@gR#Jy}{t|
z>{OIlRO+XG>Z`-TEM}{NYlrtAJj{#1haY~roL7+0YHdKeW(D#i+L#g%mc$&rZ>=-q
z@pyml@b2Av<MAXVQQE%p+UvvdWVV>sO*0$}opA~j0Xl27f7Sp9h{js;gQXT>7G_{Z
z0A^leabS&cQHoO<4u|u_+<X7|x4t;t*<H21>3rR`Rav7*=9D#{6@z$AYx{?Yq~&s+
zLeyF*rIC!y`ffC){)qacko8tEt$b_&k|MM&D+bnOq!bJI7=B<XPr@m#ODq5>MM?p%
z0H+jZvsv49fBoTbG#(jel`+m0nd>2{)v9cors-M%$cv&V3g(o8Clo`d5C~pu3j=^s
zN-Kp(|J)Wak^-#>U>!&j09Y^CASwh>q?A&c6s})angCJ=DaKW~T9l<Ix_R%Fw}0X%
zUw`|@?!5Zy{`Ff*sUM0!Q1?Ffs*q-KqH{2PRt2kqe<>rACm=#<>gDeC7@^tQ+t}Zq
zK6v=)@yYSqUwT_W4~JVfZoGnEDG2~10#-mqe6yHeo?M<DA8bq?e*Ey^`yWrp@YTF8
zYDigHRn>BFbaXPAZjDD1K|H@WU(7BXU}mgVN-Gt-XOO{Q5MosH0zg8D#u%kEB7su#
z#fp^5e=}z_X-!6}qUd?w5-C}W(2Pz3K7_98VhY-55Kh8n+cr%|Nm}oTlrc7h@bvj{
z*@ikKTl7c}M7j{0*4J&Q8sGZZ_yjiVf_IryikvoDAtok8TA@+EDYb!J(T~CB&P0(I
zqb98s8X~Qg_r43UNs$GVwr$&jy!ZqdWelmEf9>)9{zl<|AZo!uKCD({;=l|HSo%=;
zL|Gn~`+32sx%JY4Hu&BL?*U<v_p&Tc9Ga>+e*UbNWxj2a_k{5Bt9L*8^pol4=BJ-N
z*xB3L+1*=})hn<4%x6*>o;`c=i@)@Dv{Au}(aJ#V7nYP9?CqQ%A5A9Hz|QFGyYD_&
zf0VE_y>Wg~&gZM!*ETk{29Dt3{A9K2R$djO>GO*#P!7;EAt^{85@JHZsj+X~xb{aM
ze{%5oFH_dL^U`a7^)LK_>HNvb$<)WuYIQz4zg$$hmn%dq-Uw?1qqGpSBt!_IBVZ$h
zf(j9m&y|X8OOgQ9%EwMH2{x_oLM)84e@6M3=8Y#~BXgTL0uUfK%}ayP<Hz^^$?yHH
z?<%vnI>?5Xm-7b?pOq=qSMx7@`E8MSdU~|FsA64Ho^`8KjEhI}W!~!z`<W<cVhp~^
zhoib}%epC7=Ld(oN6+Ueh3Urh+0%<_hljoeK-IMoK;_nb{pbIUo!uV|kG}lwf5YEV
zT9sv`ls3k&tYHVqca4N@JQx~s4<9_WuJ`sA{|u_$#n~J~>f21C0wkLk{m}?OhbRO@
zj7m^35ku%ar<5eJB&9TJjmosZgUQCrFTZY$dGFo#j*m}_RtiLkpNxBZJKMvdC1OoV
zYhztuOkQMJU6+2j(n?T+DvHt8fA+?7*AAvp<wj@7N<tJLQfPfhDfke&6k3j6AjT9?
z1SBafFEaH>h`}l?Oi{RrsY_ur8kk<DtrY=4B?&Cq`F!!@$+N0yyZ|WG9}ahRcM(wu
zt^tc0HEIot)-4;-YV8tSV<OgkH%<bAoFK)ul_VlyT7y72k|L$7paKk*f0SK^<WW*g
zRS>W98?U@^<L)c{;fPRcGDaDov^Cw<>R%MR(V}c()Fw|tjb~$B2!cvw)<~OzFC{jb
z99YN-{q~Q2@!Q}2{X2JWzW@G*&JFUs>$*tD*mNdSNRVPgM<CMFHO4eS^~JrzS9Z;^
zsvjSnJU;$t-tr=N%G~z;f8pNsTgqmcbtk81o%dU#ktRgMrfIaMf~@ZvNr3<$#XQer
zN=liwZL=&Z%W{2R2q7r_V&a)nLPQ`EKy5TA9YZL~s%+}%c&v?yEOk|8Mc(=(NsO_p
zvQmNu3AD|!telrzhs~`qM_*TMNPavT0Y={?n-{Atu!u7{rGSbHf7_QNZQB@~rIfT$
zPE^yjF)|TZ>&jJ4h7GBdQcS`~P4Gb&wOW>CQS<>xQt~m9QdL>63d(Gl4)cU^adfu5
zvCk-|ZQZnx=xnx(q%fwU=vT`{mOG=}jXN(5hgpi@JKuiCT8mm0S$6Z<^|RB{+c$1K
zdiZd5e(~{xj~Mw|f8YAfzx8ka+j-u9^yu+;JpRmT10oA0R*0PGaD998>~f|u{rLIC
zYRTG`A<*~Vf5ys^$}X;!m#3FSVXUQ(KRnj4y|q6eWx;iedS#sxAq1_7s@g|pB!^wP
zdvG&LgZb9<)PyW%S>&tv$}j4hA-(_MhsTS{GvQKTYrfdqe;8N^3PEdZLmX>UB?eST
zI#Whc07%5p%dBOO3N900EU#uvt0)Q3kj`vwwc^f)ByAEW&~=`b4v8g7hV(n{f49Aw
z-30xm?X8^MqAZUxIH{|qzWOU~e4#AMci#D<3b<;0rt)&RsPe0?|Mb`Ihu<g_E2|va
zvZ>Kz^{m|9f0~j?SC@^+>S5m=Jz1JVBVJC9N^PjDyTq{=^n!n(yZg}`Mf%R|ds;m^
zIX?aFpT1jn_4fAmV9=k<E>nm`D@=wtfsnKPuYBcekKX_E-EaO*5VV?vv2dAF#)l-r
zh)8QvW8$?-762rH01%PVM(0J_b&8aL6obL%-ul9mf5)F*l}puk45_Z$z+$o<#wbil
z6e&YWMNT3yVJ93BopGku?*r=SMY`xLI9IfFlK_>;wTT^~kPD!db}=NN)UDqQBa%W8
zk+qNpwARMC$#f&{_gQ3hdBvoh%ZRilBP^tK>$|#Xlqs|}SyuG>L*I5}B&8%s2~Y{D
z7s_3wf0bsATI)5)Qz?~V0tF(%iq{xRL|m`;uP^577gI_Q6IoKIt#*RAY`s^sRSbsX
zDUptmUo75hVB{Za6Ql%I+k&(v!rW0WSG`O!@)?OQXXvZ#Nzd1Fg?w<Zdwl$SI^CE|
zCcUCxRn4=f#~`Scu|{=uO(9wqDBDo`VShB-f0&f)U;o1AVr;g@y}$ey|J=*B_X;Nl
zFt@$!twUB>Aj>Li17}%3aunpMS|LjCZRU)`AR!F<d1h3KVco5>)`k?Tx(*>IrJS|v
z_<XO|L&VlcM8zVjvaaje`=Fg|>-GGHs=D$if>v2CBZDah6zshBDMWy}?Et`N<FvAh
zf2Nz0w(AzFG9(tn5Tm?^C2{6-ZViEO2nx|y(uP>LZR?^}D5bmBH+8GER)B;8l31co
zv2MFO>$4zf<3kjH^=d9lGMO!_s%xXtxpDbqYtu{6y?o_;G^UIU#<{EzA@<Q(6TBy7
zc6RoUPESMPzwj4-{`B;i2)4F2MN*Nue{!{m(f51#e6cvZcH`jiTGPdkKY0MCKl<d;
z&#EqP;-={+g<Jb$BxocmMOD?#W^>`JU1i+yWO_YUgHU_!+B-KUpL=;TrLqlPY5VN{
zX&qG+)U2!{r#8iwW1D>0bgLEkhKc>=aQf1he_|sW*P&f-U3cM6|M*XaO>_Rcf4^7G
zW>*PT2+`T+oqx0p)XR6ai@Q7dt5bL2!tFfWDdc*OcWr%R819VZ&AsgPE%W;I-YbWL
z>l^vjbdpE|(!kkqZVKYqbq&x()dHxqvZ)hSZ3h4<wl{#MZTC~X;a}L;m{d($*AV%^
zS^d~4e*5~b(Bivw^#@N*o}8B#e{FYve%zl7r@CJro!sgVI%-nZWko#L8yV9|9+&0v
z==rj2_~TDbC!03~@aU{mw%2)I1vy_f^G0U#4$0t;sz=YvG4fzA8uW+Ea{Kn}FaOxv
zhx<Fi9Z1`>i_^15XXj5+3JA7o*cHQTH}9hDU)4<_1*EKT&gL%&S%5E!f4j&!_q=Yg
z#+X8g%<C(!0g-vRN(gsexp(LF*IL9ju}{e-2C~K#F7IoblX9q3mgj+k=T1;WV?qQn
z$`&dcs;p;SE`lMXB!pyGR7e!TbV97Mq@BtJ==zDYXd6%oX+{mGqfi$lqL5ZLb49O+
z!oi2c+{M6(lru@BtgFU%e?S<7lZY}#87qiuctMD9J%lr)G@`@^DgCn@S@=N?2a$*r
zp(0dT>wkVp0uf_O>)%oe0BB6s2b+8K|MT`9PnTucb>H6mnCqVZZ2HW~sw|V`yQ&*#
z12!6l00BWXVkm+#Lk$$c&|jj7#tMlM(vV_+1PM|QB%-^CMw>ESf4(ZKvNE&2{JeSg
zn&bA?;8k}}6Ez}K>N9=uM)$pa&sk@!-%@pg4s?IAqiiom1fdvNBqjDK{p#ejws{4>
zxwQzqy}kL;3wNukbe&twX8-%2Kj3aT$Tn8X=GL7%$4AGzTcfw$e)HSk`F5|@AB``+
z|NbMbGei}wfZ%*Kf2}fWt*UGH&RcI^-re$AA570cfwi!;(R=Nc=l0J|7Hxt=+uK+2
zVx)}qt9e-s!8@fw-PA}}mZbtLvTWfzAhxTOCWs;kz=@SoDT>j?=9wZDVl1;vYXpF`
z=c(yh0EG%&lZvufF6yoAT}vi}<YTJah7r8?S#Djutc<bNe?;ei&=B=|W#=PsOu?bj
z!H1%N7$b@(r5V^{h6s&9L#PNvDIp5MF$Zn3yeR6rPBBuEOJV>_F~rXG2E8ss5kRFv
ziXkLp41!_?K}<=+K$75^ZeABVQRTLBsXjY7sbgqdXHA(H8FlD_QUH;?uO~Z$;ds=#
z_Ja>TFiQRYe{X-UY3jq{qs3zJ{`>D=xqSKg=k6lm!$*(Clg<78qxo#*I={E~`1TV|
zU%PtqD++WD?f?KF07*naRGSTiy#imju{9X>5mu1eE0;DO?w?!4s;?%K%kO>mh*l>)
zL<YIEJ^cFJTU(RdrRIZsk7kG6q})AQ%vdF>^msD9f4G<{rBrS~+4B<*LZz;*J@f3c
z@tfRV&OhHjm|l5$6lSaDqxYW5X<EWzWJ`MC!z!fO^HY~^UfY^n-Y~v9ch0#cITy2B
z<qFr>T#KTJv!Zu=Hh(bpzUj8LVQm8^7S=Hh1P|sju-P)D-l#xA)rIZJXnb<=$K}Ru
z<MWU>e>4u(*JgaJGOU}Evx8Yay7&GYU)vtrmTzosl!Lse3Wh{3jJ3JhAFhI&E;^N$
zc~x!Z!;_;$<3iEz<zzp2_oF<^%BtvGtBIm(@+|j0ot-a4`oFdfy@(v29E?T-QfxEj
zT{4R1vxOvI<cd^|6hnwCKxmMxQswPC&uw45e|~iTqo^P+3d~I8gos+J%w!Tdgs^rK
zh7e+m%uJ|Kl&<w;BLZlpopa5_*=Tp`_6v9Kefm*7p8}*HL^cz$i4z)SwW@jp(bO)x
z>0*YUilQvCQCUo~e4Gs@y6TIu9Y^PUN+OKT#isR5;M3W%SzJJ9azn<j)uzl#<`g+-
ze<Wp$uxCk#LP{A$IzuDUww^$AW^`WIgdoetb<X<)0?24_v@y(jw(B5<L>dgOc`_i0
z5s_m`6m?;Lk&FdqCeqAnb0@)<jhkQY<v}2&lu`sh5n*MiD5)yJsIKK!=+f@x(dMo;
zrF=Q%g*m1anDnpQ(Q9inL->b(@VmeJf4#q>RXjgEIXONIo&BHx?4J({ZHg+ii}9eh
zTwEk6PESsU{eIIlC&y<mzw+eMPZh!}BvA`9=*&bPYUeIy^U3bjRcJ@O(dp^M+2P^v
z>gCupCDE<j-J{dTXv<r-o*Hdz2Z3g}+}OJ0<};yK_NscmNFfkHyIf&PMP^&ye=>lv
znK33r7W`7EQ3Q%GB~B^kSuP^R=q$5s*TuwDe_%4Zs8>o;uUGno3&3sD5K&AKP_#zM
zGYy~s6C%gh1!waDAuU()BFjSXF$LzNwPt{}Yn08cH3R}FDs4EW&ikUYMV>)i2i`z9
z5kgTE2}6uN&n*X0L@}nwtc;B@f2EkT%>+OrS(~Ml7?L0bfzylC)}*_-yHOO?#@6KE
z{(Wo9RoiBHx&C5dbJw;-WxVg6dh*F--Tt@#^uPOG|EK@)`#=1_#^&Z^G8v6V=jRu_
z{$R0MJ$dKpjg2ik=(E6$>rV)?LPZEEeP!5Ng|IvBPsZg(pMAc$*&FxAf9G{K+#D*C
zUB7zu=Rd#aLTgzRFhN8yKYH_xx^3CV&Ee(GA1@@;RYARhE_yv-;KaxbNeqKZ9D*O-
zxQ+@2U9gLb?(x}Xu}QOe;g(lpr*#g4&1N+-Sq9Mgh`tNwhv#@>mm@rN<r;G1om(w^
zh+d&E2SaEyMZiTDTEykbe-}2>EQ1K(MI-ct22qk7eUkh=B@1A$j4m_g=ILH}CB`1c
zgdAC75>1$r>PNRJDRX)jnmNj7Ys=1;oYVGXGVWKwH?!{Ix4-ks-~5|@*&knDt>WRd
zVT$8%VXO)qo7s{R7g>gg-i77LjfOpgXtfQ#F08f6a{RSTdhQq5e`MTm8W)2<JlsD~
zpfoBYw#+3Eki^y-oflaSfPl2Ud*#m4&z>EAQHN0VE1M}K$gIswmJ$n+j}bMICh&kH
z2*Q#ACuH$nVV*04K04<mFN(WQzjXM@qo4iYFI(TP+O}!iBw%c=O>S(iwT?c-km|Y1
zNcDOH)f;VGxjER}f0nA3RRv{v+q6JPnp{LD=u)a(yQo*b@eo3rg2`b#7`TAM2uJ`)
z8D)xGA~SdZP@-ZINFa(k??-C1HQCJb`~VLGKr7|EchOzGdhPCuFRB#hCntpKyeX!{
zgn|OXX?>x}ky!x?AP~xz44{<OiA73@IYEjnEC8C&Ac(C0e+0w{2uzmsszK4a{N(QC
z8&Bk;322oPb4maSfB}Ji75@ML^ndeT{Fkr2`eNcXH*$D<xIDjT=8M&0v6!C_%hq^U
z=2>3i=GGu`dwzb>b={y}e)!?pd|F?>`V^rCi57IAh?YVMv&CYxyEFT=QFi=jf7;d^
z=4GSkCm(*if4#MP<N8gfH}4;>h`8-qGFjvpVz;OlTYZDV1rc~x80|E87LSRoNhuLh
zmja-QA&3Zb%5#IjoY)69$$CmdUC#xmHykxCh{+apLns@ALFgKfvRE!5$wqGwV(hq!
zZPzs`)9=@9mlEW~z@=5^z0J&g)dCtHoK}=sjmU{Ze{PMJlo1$C0wTy=ic8-Vm136!
zz+^_nm>XzTT@zweSV*ev7P(bX0unM)@NBFy#&Q6WbvlYcB8f^&Y|51n5fD;r>t)w*
zf3V%Ukbn^plqdw1VPUKy)*6?qS=rzlKl{mRzxj>*#|OQ#RKoSLzIpSh<Kwfp-ua|I
z?A6Uue=GIrrysxk;`2K@8{1oUJc3U?`O2gl6281S-a9|hMTX47jh-EDe!jP_GyUdU
z@0$cy3Ps8+d~@~2wHNPRLtdRtrwK-X{;z*F@3bZ}YvM4OElta&^^rNbepR*JwJBJm
ze&=hinEQJVPN$(nwJ|B|Hn#q0tv92P?cHF$f1IXu-33WiS=Ou72GCD`_P!6RZ#;YJ
z)w@^wm9kn(D%iUa5D>|jwhbq<=JC-|1=?oW5PeNs8;KY+h^E*^cg@(>wRt;)`zkFV
z4cf5FS-*`+L<=SnL<l4bLZ4~T6^|a&VU!h{Wp329tqrT7j$M^Z@GbPJH$MI7iQ6~#
ze;+3fyqG)C*mbKaFE!;U0!qsA$~oV4-kP!tfFKgR)l_JsQ~H(2s;_X>%jLz%IRnKI
z1elCX$Uta|9E^#+2Esfu>+3=9Jt4mE(yO1}`|#}G!Mbx1MBo=DlgLSF#loNnm9Pse
zERupS#E1?}W>N|uQ7ao<vSeR=`88M1f8PH7U+teB%@!B+a--<=tHF>^dGDh0zKi{8
zoMoA=#{I2p{q0?<`j}_h5P$#?DoTJMLI@It1dOOWNR5(($Z63M1*5qtN?_+x%aV#B
z&wHAa4n8phF*6D%9lLHdTU15QwcTtsT`U%D<AP)8+28qlfBfRBUwiXMf6bDNe>NN<
z2x=e#wQe;45rQx6l>iDDSu2gW9`tLYy)cT92#XM*LgaOCxU8$6v{prB2jh*&#_si7
zw&*v3JMRo==CpRNA_&T_R`&FZFWgPhYeP*_yVgB^y!X*ZpIEJfa~tFFVln4rZasNN
zsC6+ZrCxmD#o=)H&2N75;lrbse_#48u6MEPHY_5oby1dpc<J)xPd@qd;P9}jY*Ci?
zKKq;yc6P7y2fY_Jt}giA#rerrZxB-?f$4Mx0?cBpDF=PGSb6Uo=X9R0n#SfCDHTJ^
zP<!u0ASEUu?_F6GnYFpm0I0QzF@_}Awuv!Gms0BWD`QPekvRZ>QYyw6e?#Dunz|u`
zJkL{t`Fv$kOd$d?YwY)XAx7`LF~)nZl<M_*het=f(by1+urVfaB1Oh%?|q?65@F}t
zrb%dwH3)(TDFth-F{*VzYok<JpGLG&0xE<cAnU&{5;3#$2?b70&bCW47;YS$eZE|^
zoe#pv=|rTMB{2&#A&8{$e|V%dz5DLFcb>j;>xm~2z-oQ}{)0U4XJxUuy?O2W)xn_m
z=1+h8Cx84$_wIc%-rRiaop%pTF2>`HuWBwJLDtVp-Gk8OMbScgv6y#VGtTv7C<SKX
zILp-hlt2FXzN^owTy}OgV_P$WB&5h$o)PJkAf;%n1t2F0K@-4ce^pHP_Ls}}jeqbD
z_l{1UymGV32g9@D{NMc<Eti3`LFj87L}*yJiC6lgO;JAZRPdwEAAIrn(Q{AUFeY$}
zv111Ii8~)woj<$i=JQ~aEto2eOFzp&n}j)Gj<}r_qr|(BceMGe_MdUL3v|b-WDz5A
zjDW0wzRXlw=|_37e_vLb-=EGt4y!v~d(Dbmn_Mb$^I&f|xN~cwLiu1%!Lgvu411B_
z^}O{4OusTf)H)wS1X7VbdS;L^E48A`P;~K^fW!aam*8UAKKx>DG#O9l=cm*2t;?5<
zHM-EOY1Hpe&x6s3c}WCo$7I<bUA}hX(v_Qw**Q`MLc;8Qe^g4Z9c5bUwOv(ft(1ZU
zi6b*}5+tuuior`_t*GfdB5iK(-hS%Yv;D_s`;QJzk2iO>w{qkBiUXyTIK-x1X>0Xh
zW3qdF_vTaCXgeEi*djxv7|>=aMw2)-U0kf3bL=5w-IFYs5-f1?OI;R0XAW&umIlg9
zlhssJrKHfce=fw}y_dj>6Ei#KR;!gT?|uI8<oImW#NMd)-GA^Wum9F>^{YPQc`Kqs
zP^&fOMnFOczwAa=fCNa00zmR5ZxPmJQ(g-=wC*4QOr$^w3IHN?U2I$LDBHb$^U~$3
zz!IVe5HT`vN{J<{$1h-lUkNosVos^++;TBHJw4mse?L4qIeB>h^DN79Ym^wHuquk-
zWWwYx?OcBJ@PV@E1Jre=ly=Tly<Y25;<P@83^z9R_V(B%MBLn*IM*tr>Q!A7*_tov
z2%o%lsqPlZHKynndDbi%tFzoH*R2xAvb5d>!yFl6LTNn_<#`!GwAxymEtks_BS4YX
z6$i>78Y%#akp<Lh)wIqPMbB7^zpOeVTGyVfA;xQ$9}EE%e^bs`mT9dAgMrO+jmdW{
ziztA)ZbB3X+$kZYjYeaPQAR24e5X*9QYmr0YS36j4Z;GVl<K;UMN&$nP<~;NmUY?$
zXw|fPhexBTtQ$w#Qe+@vL1nCWp>>V`tRd&x&CSX8|N8&<2mko*-@AA3!GlMyzy8`=
zZ@uy0;ojEPfA0DD#n->_jTpmlJi2-F`s2OFd0E=5x_#$v>*D$O^sAZ-dERR}$`Hn7
zwR`2p!SrObsZ62gr_)iNo5kGrvgzVtFwCF2ylYIVV>?UX&i2L=8^!)Kc_)tDGf&?+
zI{ZQtvBWa(t?JHssZ9R#Xq%7+cc1#n$M>$@xb?+se|D|s)muLcj~~U<8dX4sCX*~v
zu;fu`R!tp*@^Lw<7YR5u-Mb$iDC72y=hWpJo4sYT%8N3@6at-`cba0OvQud(`VBiM
zA{$6Utv92cSc}9%4zHoPtF0yg?{!4fN)QnhCz2$E%{a2+7)Y|AKA!+3LE660Z@v1`
z@i^^{wx%cb;eYAH__ez?{_B5wv)F$5-h2Pc|M1`M9e?Pw30(^ynN?SIcg~KFr;COG
z(AX@qfL!KwwOU2*dPUanm#4>5{B_YQkx*G1VzhZS+!!8S9NSD){Sp*A-<+SXAmM1Z
zJ#PY{r07$UCV8^u&8P1k9PD*|nPsJ<brWh>CtK7nAb$XBF?wxyS$7P2@3r%gL{d^j
zDB?rV);w|NSxo7Jw}1NC-rn|=>;1uKHeVK|NGS#9o2KdYddg<~$wU`b*6Z7{$0!Jj
z0f4k%WPrtTHC@cRuJ$nq8CwprOvUJQZdsca*qUClN>K@#1to|!N-L|9bjnbaPDEme
z0@yT7k$-1r(~I-vyxbgq?+^d@<=^;LZ?vJcQAID2W)T!nTE!5B69KOMGqT>)vp`}*
z1d?B3CPf6`%YlOkDMca@VF1xuA)+#Ja(<RCF269`puB2VHGm)}PO_Gi1tI_-3bXsw
zRf0;153cLlhY!EFckkoJj~^rAVlf+!dfs;_xPM~UFG~|+`1sQgA3xaZ4SL@D)3bTK
z!Vp517v6j3gzM0OG0C~@?d|v8dN+i0<I1(^?ChhD_La7P>geS7*{AO~#$r7FtDnBx
zUv6(z{iD+p0#S-2rK&8rT`+qVue4d#PT4Hi2tb5{Ye|a*Q58k6Z9778(W_i~Rl$jw
zC4Yg~+{BP_tGBl{Q%X(UcCMQ*>owqQ4K4-%rIgYbLU18<&MlTR6wV6+z*^-2aXsGw
zfQV+Zna#4cZHK+QC<?8$Xr*;@-ie?90ssprtvlyK;;QPA0%LSaG0!VhR!}1atyN0N
zTFau^rdd}f1G3fz2{Hf?TGfrKvV;0;vVVQ~czOYfQc4nn^G+dXtzFj|Q}p|TfB#qi
z=YRfB{?QM9^uyb?Z-4l~z2lSf>8ze?Y$i_M{?510&(G)c=|>-ZaAk8+X8F^1pMS8w
zf9dM2*2US<e^sNQZrU#n4xWE*+k^TKKY9Pw?VXF0{X2JV^{x&V`zJSFed_yfy?+Pe
zic?mf&XQ%uI3~G0(7WGy=H$F-JJ{>m<L2BzQpQF}%T?PpGM&|y^n5Xmdh_VwJS%n{
zpBxlo@FJ+kpQqI_DiErTvRm7|I)tUNpVo^dYL!xQ?fXr$<>il_xx6@8+}s*Bi|$JQ
z@^Zc$KwquIMP`LnmNBD*s1;RMntueD(7JZirL=0>BF_kkAsK*#m_XGS1=%wQVjDX~
zPJvZOii1VDij%+p`t!qqZck2Uov)L*`pk`kUUt=HM8mD;zH#HmQ`LL-meTfe%F6z#
zULBsCXq}4+q7Vq`wzkGxbai2j&1gItteREc|H{&Be~X2LS`cFl+dEs!dVe9}XXncV
zJ}+$YlAB(Z6-8l)2w1%L!ZN7JC!cxt(~sXj{_G<klaenR1lQ;6brbb^$}6Rqlb{k(
zN|W=UYun{w(YEbiveUKE1z$Ivwb`vZ&*awL|M;D=^Tp0)s|Plq3n2vO*GFo^b;G-8
zqnvjx2$Hha)=hUYKR-RWIDa}lYU{dQ&6Lt>b-Z(3=Q=<g8Ol75m~{dMl_e;8R!NF3
zC{vIKr|3eAF(LvH)pdPvba=Ly^*1;E;CKJvJHPk)l8}L{DF*#fZY#|$`M~SuKoAl5
zf4ts^6w;U7Bk(1Z_?Mp|f>I=+)-Y=Pk`JUJ%EfHn*2^lhM$t4xvwx0Zt*xN&WnnJ>
zq!fR(>&95_T7PtO@ad<YzVps|zH4tiaWl)ZvMgP*?B)8_t?PNFFJ@=I{oU_Axc}(R
zGf)1zzxeZ4Uj0Ulm{}WRVi8i57(xgV{O0y<RaMunU3>Is@A%~K?CflNYv=yGPhWWP
zIq&^&G<ffW4_5*2K7Vt&L*3M~rk)p8r2w0w!PVVyGdtaX^e_dNAQGaBaaMP|UXPR!
z#5~Krk4hU1u~+p|j6`}>H^crQGdjl9`T$7Vn;S(@a10>Qb=|72*C2T3T-S8~uy!?t
zE&+0itTky(S{Z@}!0|uo<@j>STf07fwrwXVSZgDK>pTLoK!4}L=H{?-T?j#Gg+#%-
z$Prk&&P!xvlgUb95n*Gjh`25oYl$eOKtzdC4ADhJs552HF6KMiTl2bEAN>$9rkELp
z!t1>c5o>M!;1B*TDSGnBC!4zclRx=S|K*?l*<^DAK$gqJd++{yGMVf>eDvZAFZ^GB
z@oz_?(d|3WJ%9J?i{Jn2ANTshC$2y7RSkw-zt^vN)6=Cf@!7k(%Vsv*9xYbW!Kj!F
z)u*3*SX9}=`?Ic_KY4q1_r_*b*&#c(SYqAuy{njNNiz*)X)YFXn-#X`pPZi&5hx0P
zfAF2({Z~Kyo3r_2T4YwJ$W%|^crctUrmbMn`mk(wCx2CK=<Ntcjt|>qzc(1%xXCA{
zpY8d@vD-g5s*0*PJsS!R%7UUFYo!^kmjlB9NSOld4QfPi(csiAAuHR+D@qq_2cmq8
z4Ra$B0Z=0`C4$5$!)P7n8Yh{_M%ip)m*V!$-NR4!h023fV}Jh9l}kI1&(&}L&Ogx^
zrN~iYzkjGA(dqe_C?u^`?ScT2h*7f4#t;egq+cn8!Fdz~5b@Vl4xUncy#Jsm%C=kO
zRoSnq>Fi>0ac(kY2q|Qnqpg0nnQV=sKo~-ZzJoj)3^uObxcSBB_bzH*8c-7RJR>5J
z*!eE`qyQ0AVgyu%P%|J=X7$WD-#7!L5F|S0E`M}wa0x()jmfPSo;&*d;pxTHur!c}
zBmpvNGF1{pl^4ZCC=V=c=YXY!x?a?0XQyZ9XKlS$ts3W&HEXFP#Tb1Mq$HqKW;@7H
z7@0f9X(PEcxi!5at6*?)xEH(&ArP?lO`B-%<lKSzjc<SNx#wSTofM|$9a3dR8(U~A
z_J0Ox1rit)C=vlCl>oKUZr!;}pb3A$PXncp6p%_nEhy%cNJA1}LPDcSiwHxCB6+XU
zR<~`-A+jbE6rmWqHT^iHL_|pPD}Nt)RnL!(4nFwsy&wJPM=?f8@xu>4EQ*{%&{`#N
zBB5)W&CSVAfAZ6BfAhBv4j&hJfrz(m-GAy_S5{?A>_Q{JglKHGoL&IH(b3Ux*#F|e
z=Vej$tNtsme66YH#a6yrF0WqNIa_q)&er?)4=@Xb#ms^vwLR|T#FVKkyW`{2^DgRT
z-Q;;6%F;U*z0dPPly>Yu*3D#G7DaHrs0;vv5K@w=-&aJ;DMg=W`PxmtsvDDKF@MI@
zYNfSaUxvgOfq)c7P7LBYm-mXqQ8+}@Ym5{k)^(ldd7kHqSrJ7adu18Im$g<bsFX&6
z#cH{hAeoti7opZQWm$@V)|QkJ0VG=Ml_@3XoG}JiyzgwLm9@@$5tRT?YZkef&)Wc!
zTB{W)6)^#bcV1hi5fg`(Uw-{(KY#nl-~FTC8IOmq3;*n&{j=M*pT2hW#&A&m`Jex*
zL9bk#pKT6D7S-?l&hNkV^S7_xx*f&5^x{j8_x6;0WernJv)bE#RAv0lZ@per=4^2+
zsmt@MsU@}k;TH!_+<DR#Tn$Q-v$DMR@X_3dt>LCDR!}>KnKSzQVkTOT#(({$^G31J
z>^hDhiy(jcum56cm19iHr2!~Tx`Ng1o!tj1-Cxcik`gclkpb34%<JVMEG{m3GPn`5
ztaDdHZ;VHePEVbvd5VQ+NUR-G<WW^gVhdn(vRJbBWNGeFwXt>o@KlRhrmhiCl29Ni
zSV2q*3sISL!zypiLuHCVMSqpaVxK<S()+4j&YQ`l+iwJ%ExKG6lc;&Nh=2XVy^kNe
z*q+UfcdlIBJ3C#?7aQYoR%Of8Dup`5WV9xuR`bRO84ZVIw1`Jh9Ar4G;Mb+m3qVQ|
zQaZbs?dHXJvi0E6gBZCe3JIdf7_!Z>TN^z@1OZZv&UvlOou}_U{(tPV`SE@)&kSW+
zTR`;Q39NN{b}>c)g_=l&P$^{)3#FnYGFkwVFe0wKg)s#|(M7Ry{VFu|$zt9!WQ|fv
zDV>ATRev}fZ<PJMwi!YK=Gn!?#o5^xUp$`8m$SvZUNzAt0o2y6ZR|0|5Tn*w7~A&Z
z{N!R#X8nFvC>T`P#(%iquZnV55qmQ$UEtO`5bk1Hv_Vi`x_oQ*@(t&KRx#NYwTh!b
zzdzL0`nENM8Wges3j#88VooV(#exKgfS^b*LrNebM5v9{$;Rjj1rSpbBGg)IZAC;n
zJ?xLZ{>tl<(bgA-M<B9B#fe0%wE+K8h0ZejE1#@Bzc_yHy?=M!dFLHY;JVPOvd8;-
zx2|7x^%8)qvbcHu+H8K-wes9E&l+Q{T)BcuHa9m956`Qr@*%{;A`)VN=o9dGINaXe
z?)7?$#iA&()oOLPf0$=EGt6dF=F}^DWGd6|zy7t?-gxi*s><>VN5jeS{-f(JJ=3rr
zR+GtiaQ}-l7k^d9>Y{11)@|3dE@)$7NHM0osI;Q4^NZ=UcWEajF7kZP?|V^-P-2SG
zs?2JoImsHy)V3`%TWi<5`#jHG^y?Nyjil9@W$(0xx^Aq^*40#1RW(fm05;3ou4|U_
z%;<i<KX2*~U6z%C;9}I)c3qcc83@D>MI<jQ04SvmWq&?N3_isuAZykQB64DF)cUrH
znSJn}OiD>mGsCL!Ixn@LLQW}It-S9FqqHQ_(5uQ<UVZ8Q-X}!#;~)R{#`PzvswWA2
z>%G;R<MG#CdHLSQAG^jG)YYJj0qRv#4z?aY`l79u7~NMk8&dq*vo|mAsIpiU#%_&t
z4AzJ}Uw_o3@>Ut8@YSbx%idTtpLboqzm=V}^0~z0=$J1|cRbFs{wTyGkaDZLX0<)+
z`ECVDA1{~3-ZiajM4Gl~m-E%C&7XQX1G71qJgjF;V%zEZah;Fo_Kkk(RQBNFZ0Xur
zn`ikGs=6|!dRh<7a3)dp>~ytsEP+(R(E*>tMt=(^j{+SFw%&BW+?-U(sK81q9VHXA
zk_=;xG#(9e&2By|sJV8%pDn8rxt*B(OZIaLdbouW?(gpd8}FDvuROS=>$=&)qqn1^
zIwXyGQJTE0R&~2<YXJ-{<(cWcTVony3ahq`k#{Cz6ux!s^8VptYs^>ny#8MoB38W~
z8GpU1>%IN`%a?YC!|}<<ah~@?#J9mH+xRp;U-q+&Uj{ov2#U;jyuH19t(nb~)>T=q
z%OONS8<MmPodYCoEYW(biGV1hGlf{@f(b}$X1oMMVPWQ&gqaj(mJowT=R0eSCL5!8
zfpWM}_Q#}6N-3l+xYhB|-lv~@djJ0YrhoNGgp&{%5wUq+D-8%lq?HCBP6BH<O6$)r
z=E(lqrH%1;%L`pB+QJgq5^ce-OcGaZH(zyNt9&?OvMh#FCWxJP&bzALw|U+z=Ey85
za)`nri0kSoW(G(iQG^f`DI(HqL~u$WcrOBkiU^q3xho>AM3aa_(RZGC_Nk}u5`P=^
zvi6Z{4Qs41CrK%-bzlSwzY1inA3l0?c5#8Io3`s!z0J+x;p2xO(Ydb3px^ITz23#_
zbiP>KfB0xT+MmtNd&A!CC!XhR?R@~s08?@ig*7Rob>ML2#`Wp+Y-?k?tLy3PsLZPv
zVp&v!!C*X|T)%$nZ+`mb=Lhe9aeublynM5bDR@UYO~wPGX*j5|%vOWVY&8ArAHO|4
zJf;31vIGVs;J~BNaJgJ`T|MXz>(z2H8Ch$HFoi%Yb=y=`sjW3eCBUX>K^Y*%)+A;C
z03ZNKL_t(QBp-qxprRQ4S|v#w%d+fJAVh1dLNR8LqEn=m(#Bd+s%_d{uYa$VYT72U
zY;Nw@OgZn%Ec0D(UES~Zz2^{}TKmaDj10^mZAj4}Dl|GJ@F{W<AWV!gk=DeVVu~XC
zWyLuut1t=&??^*LX74f$t!pOz0SCvN^30~hNZ5tug%|Hu!#pMV`s=^>{qO&9I2gb9
z(hCnhf4H~z`D?FzeKtM+)_=FZlOn(M#+%>#_P3w7^W;Y#f3$OH?wi$Q*mw0;HW~=A
zG<r}LPu#qgbn>ZrbZ|Z_tjY}70;|E)b<s6vr_DFM^&3YYzISObFn)Ta$%|s+i?fRm
zk}5f1U_wF)p0zeEz;IGtT-1#C&Vzg7GFy5tWB`LMY~qxZRG_aE{eO>_vyO$4M$`1P
zZsJDv{P&*!-~Q^|S(C;l-q6*C>y*tCP|UJ6dM+yu<Ph4bOoRt1rctR`4aagK{Ki@5
zP1OaSr!>s*lGft@r9hvdV6Z4yyABsGzIMH<n_iK9{?SD{*}XRk_hof-+Aa3xM(ft`
zYUPnK(b@-2NtO^Ih<{s!MU0b5wQcBJr?ldbgi)zXDM*gd3j!*Q{eI7YRONVMSe~99
z^!n8=Y}dbzrE`TSNxW;^#Y_zbgUMvGSN0DNj%vTC=c~KVyw+Wo#jFb<Lh)tl6;V0Y
zPFHh5(3$<k<}bC*t<TWPm>khL#|ezFi9MlcOC)RyD}-I_8h_u$=vg9j1P}y9Aq0b%
zRF-+yupprdk(rDsd%CESplcc?X`A+7fB)gb2g}(aDuX5?O^BKk2Ok3@W>!jRYnWLP
zA)r-yI2^L1)#CiKhhJQ|dL{3Tl{SgM_Qw5<t=asd>*{6Ot~y^9LsrNnLZXez^IT<=
z02rmHs#A{9JAW0t5(NnBvW>NXE<()62}BS<lfn=gKw|Xq7ui-s0$>7BT4z-mS*DA5
zxw-utzxBQAPdxqb;ApYxP_GXO0GO=FopWv5uDv~ftH(N@FMZ_1Se89u-rw6Bje9=S
z2+8^Q<Wo<bPv?ybg0fgOfB1L*=*>6Zs9pQsyB}A*tA8ORn~O-`=oB+)U{p~9Ct*of
zcD5dVvDbx26o&mly_)SG9`Ed4dwg_ubb3~=>L;#Tl{`-&6?tJ0Y;Hv2-DjVz+i<*$
zRF1bUU3sv7u$awuwl7Ea(~DVQbe?JCFzA&cK~W+~&aDRJh@Hzwky4D{QbdB-wgJG$
zL|RX$Q-21rx$UA8V6BlvGNZ&bT9VPmTGh23i%JPK_#)2|r^w<{(jr=6X0>a(&Uc)Y
zHCl@@BqmeV2m^owZ8A=wX}SbdJBKM+W306-(sm9M=6L~Hr#7-gW;RO2=y(nFLx?Q7
zcYSMF!!kI<JTM9rp`z%$K^~Ras&z~X7_!QK^nb}mpM3I3=l$vV`K8^fr1Zc1vp>D_
z)H79oaCmZl>*h6Elt298517(V-g@&p-~H~p@4ma7ohRS0H^F^n(oIVI_B$U>&#Qia
zFdA2;?JiY$f6#mQ_$YJ&(Y<{4n?L{H?N02!`O6;=gD>UxRKFmnAdPb~6H-lTA<eoD
zwSUs0Gpn4$$wog#AAAxBTIV}e<%BLSny5LFBhxRe-5qY7cB@0*gk)5b?Gx8sERIjk
zo*WhC@@8>xoNF!<lCUV$TCWdGh?D@mBv5pV%B5!#eQ<d6%6IO*5f4vREo+nad&l0N
zHq8d}ZCe(kMr|mSE}>`=41IBY|LDN#vwuPVBW&uJ$e1Q#V6$3v3am%U3|ops*g*=e
zF=!lzEGuk5NrKAE=4jH-%%i=%PLW8gvZ_r^D|>0OBhmG8qctBKpL@qe#oGLu5D0*h
zxC<$`RQnL8bJnE$z42tcb#^k7IBV+8n(S-8`FgvGHqY1AhV>QU^y1t*XEOsJU4Pg7
z!V=1eAfQkg6w;OefEh#-Dn_N0(Z(9*TWVH5dLLa(o>LM?h^R#|5hw+u#aJJsVc$70
zS*EkXdC%k-Iry+#H2Zr;`(GU9HlsXarF{|v2!X*l5Qs5~tZfsZND*RTvy@_96(+MW
z;0K?4)Ej+$GMSXQ?i$R?0a(+VyMNBRF2syDVhkJ-2P9FMQAR~hETFXsF(QB=wN}ex
z3WQ(IUrtI$i)au7pjZ+SdLL5?fFMk3wxtL$$6V{-VDMmnpB26O8^7`V>#v($KjEp1
zoGUQKuDwic+pXXIYbiK{zlG~&I-Rv`x0p}crWOvX#l`OKsH+z#hS8`h%YWk8XYc&@
zCl{v|r$tfx^FRF;*REW$nf|LE{QKYh%|En8yS7nUL=uZIC{5&D%(Z>;$*0av&y~X4
zw{9725BCrH{XUVr|G@{gD0g<YI@gKF$;nAI7+t@9EzD1DU%xUcsVefuB_HGY#Wb+I
z`uf-3{`lVX;=ErCNBv&%K7YmtfJ*7aVy*2}WodPmnb!G`LRsX>s5RZwb)ju)Nno@k
zi8iyOF~$TWn#gzUaBFL^SQJGOLMSrZELX^^hz9*CrYHg+tO;mM7fyi?w4xAvkr__O
zXswiTJ}PCScX^iAb<=e<qn4Bg!-0sb>wt^G2oaEk=rn=VqO@|I6Mq2_j8PJD0zs{8
zmL(D$90^!bF?mvoh*Iz($$M|hB0+rW$*1d8EyCV)&piFqXf%pRR?EfZ-JQ=q`)o2A
zbWJ;z^!jVB-T&;fMcw}Jhd+4vrI#K(czET~B~8jT%~v%Utfp^${neY-cmMGB{&43~
z^~1kZ$BQpcFQ$Y3=zqOW4kW3)$DhwvD^jD$<qc^=JX+<pyuJIxqU|u7?N84OYq$$W
zXF!x<Y`a!x`NipKXJh01>|ASObJXOj-r%Jdzs7ryVpP8C3ZRV4t;x>n;P66qb&3-7
zWw2X~wHzNkemGQm6&e#+|ALQU*P=mUQm9!JqDrb#3Q=jZ@qf<8hlx$mGm+I+wbi%f
z-s!`Im7gva`$SS?&oQas23CLe;X^gp*q3l0<;3dHMvGu=|M={DXLn<Bqx|&#0ijws
zZ?zF2ME}&?C*J<~2Sjjk-dL?!)A8(L3K#?AxoIL3sj95{MKzsH?>v2DelcAvFBFHu
zf_E|gx_0!;5`O`=cDF{G)$!5UVzD?E-keNG5~Y*^c-KYtXN&XFj7XCvBWdj<=~l~`
zYa6SyFmg&F0;B~sBatAYXsxxffY^D53Yb$&z(^qkw_1@hU01u*`RJ9Rwa*+$KmoIm
z1_F!>PDw;0QdwSfp)*ARAg*05>gmbResHbP8Tin~ZhtLd#~396Bvz!>G;xx|NB|OA
zwQ6R)elN4c$<3FGx@nWiqE$);m{l&pD)`z5CWtI4Nr<Tn&c)7BVX{Koij{C85+qoA
zY62%ABvk7dln5&j1)@^=(tMPnib(}AUYuYOjgkOBl-7vSHFXSp_2x6rJpW=f*$!xJ
zQCVw)>wh|m)@CA<*sl*L>%#HBO{K7&O_#Hac~duC*AS$Bza(IrnE<IMvitWx)jBPf
z^Rx4_o!#AytxcWl!=uA2%akIKlv`6biv&>_V~k}Wo7u(1^wn3s{)4~x%OW!;$LEXr
za%*dAb7MT4or_>r<fqe%&b!&+VQ=e3;`Ge3&wm_0#B0|sUE3Ty`uww>e{k=jqs{Ag
z?rvXNE@s<Tum9VB^Y0*rUT#;>DWc9NAEF2Z*KshYi|Bkxu_(&A?Y^uMVhJhcMP8C}
zUAr+Vl{WL$A~(bdNmAAC1?SeFI0OjZrR2+^4<K!`3NcoLAs{gm03#$q&WzH8YX*ZN
zG=JIvK<9&ivL@@gZa$xCqat(FD<UWFeb;sZkmord7-JO@6K7V6BKN+7D9jL}1V};x
z1Vp&bJt<5YAjW8{(uTS&8ltkuZEjJ+fA)|6(aFj2D=)tkV;GdxPyXg_UVZh|!@WHe
zdH(M6A^7RV`C;2MtNO-`>v_?8@ZiDm{(t`VWPI=5y`7z%|JYXa|8Z5leDC20|J8r;
z51)DMkD$n3m-Oyi{{ppc9<Q|Nom`wwcDC+4JX$Ol<V?n@YIv*6XQ%t2_RPFY>G8ZX
z{oZW8M5OVkbgqfP6_%<@#j<Exmr|N+4qR}beD=;xf^(<@Gqi<*HmEqhYA#M{0Dl2e
zZ4zAZWu{#m0i;D2oUmf?ZEIKzArTo+T8S2wA&O5-Va=VJw%wwF)YGEExj&itEM_{W
z0OQ`7HgCj0^Z2rmRT%P(_oRK#&_#4TtIAQ&cf4vEt>MPi?Z=NE+}a*SRF6;2N#oAA
zA6+;*{p`lAF`#ZbI5?V;ura-UHGhGu=e5tca8Y`O#Wh#1Y@Hr|F*bV8&mzFA?V3RG
z*9E&lq!fu0D$2^TbS|{*>frICX1-X=&N(_`)M7FH(Zl~Rs&>Ei&F@l%2|y_od^bHk
z?V44uDhb31lh#0#*&;Fvf;LJ-2#F9&sTRwbb1eX9We~`@0BsjSh~flNh<~C0wN{W2
z5D5%&WNjoyZF9v?_WEl-zFy-bo9^O#ae8{uwjQZnt(L8KF~-CkLqq`tBtarjs0aiJ
ziHMV+qUGv5&kQGDX4&JT<EQRED<)T3^)|LuIh?Qj`D{Mg+Ch?(k`G;oE(G6oT~!UU
zEYF5m_9|;Fd565t>#(rI7?<=b0TzE}rbrn<0Bz@ux8>;eGj}iTUS$DIYBU=5dgb)2
zi7|c&i9<$Eh>)Tn;@_rzO#6;!7c=L&6rIwTXSVAa;aC(|Uf98)w_MKq{oadPFCLwo
z0Kn$f<{NLmNvb#K_k~$VAyEP#V+EBWLu5vyjg5RT8IL~y;@)C*?tE;!W>tS1B(iyS
zKAjz(oMc5wl;67j#KF<oZ+-9Y{QR$RJhb~qUwr!6{ScF~*_&^@@xsfmXsw>S{p8;L
z2ag}^h0qm76N1e$n-!t!G-8oyV6VwACu=LKlOl>CB~DG#C`AMb#TUkOzA21qx|U-y
z+GM$2HeFm-m!#+&6LY4O%`|^7#U!K%5n@0^P~_HVYQtJkkI}oVDiMU4QAC7Oii`Pl
zb8~Y2`puue`|<SRjLEF?u88P@U&jcQR`h@K_9iiwX4jqHIp<E_@aC8!Gb3`WtgLz9
zP-KZkih~AHrmRN0VL%HT?uHxi$_9F&je*{H;l1IN)mpgGGC;SamLY$s)tX3(s)?*(
zu?AMwkW*w#8Pl7-@lMB!mz0HGsA!<T6F_Vt@Vytf@y@;Xod5a#36+Ex+HRhWCv%Q0
zr1l(u5P$(VLg480G)+k<f~GD3loUn?cD@bXN<np1q^Z7iar^GwyOQX_#^$|`Z~gi=
zzVXqWTjhLO%;#B>?A(98)9rSbhf61u@u=V1TwT5W(Z^5Tc<hx|UXedXoBrpFJlA<g
z|GR(xpPv7<|Aq}F{iXMB?w%eWpG<s+5e2>T?yX)wJF`=)$@_41a}aE~wsPT2O^Abu
zRR(x$t8Si6r_;7wSZD^o_INyL%K7qgo_9oJOO({@c9&8eTh)J#gY;Gyb>j9O|BWYZ
z|KaP!gq3rVB|zjDSpsvc0IU=g7)S>6ZHNSj=w%>Hpgo6RBeNh3AXCyx#nw9}i3!f;
zXNpC-Y_v+o$1@vA+47QE0oS|=)qYQ7FJDtc(pFVHE5fX-N#fDbVT$q62HoA=S?P7X
zi?%%5-d>$m6)Jz2&&xbXN6Wg~>DXCa7n2CAJ0c|xvAMBzd3SejJUJqAKGr%H)3TlU
zD3sLdA483QilL<L<jcoL$8|GNT4iYtDxFVf6nJfQ)ZJbpP$wtHdpnaGSNj6VTZf>e
zpviQ4a(I|Zt&oFlNQk)b5@e7@FOHmx&w`c;JTMlFA_9LRf>c@vMMP5Q5WOJOT1gU;
z2m`p#u(vWzPYUZfTE|J+NwW_Aj3r>Lwbq`ToSvPX7PDF1mX&o?Q?=H{5Pb-ML;#CD
zMoYOkXamAxo{=#Kq-brHW!G*zE~FMZ@fOqGa(B4t!2l&n8DMg*g_@GA)3mi|(_%JH
zr0Jz8NhyB>DWkCoganL{7g%QY-V;Rt%^VSlP(?x%!3A!ecabqNA|Q|uQYR*IJekc`
zE<AGi$<K6pOKlraahj!_PG@{_(zflw(j)>e+{T&X!a5~>V(Df&p4L_Eoh1TbpP0_H
zo&p3RO`4je<)yNmG;M=IT)cRxs>>g~{>Jk1YHNR~C>lVCBpFGXOsa%fl2k%SEyBm|
zzO%gC-&kJ(=E-C(2<Ju7>vh*QHusN?pM3hcG|iI1>eVMccXxOH+VyKgBMxu9_tp<z
zQ)%8;S+H>Z`ZZ*ywPJ7ASBDhZ!~F+k)gnL$k)&|JB}TPHArl?E1?IMGj8sA?1~fWp
zbz^^RMGW0eni)~oC6O>nHX&-I7w(bF;GOSxyYtx;AtD6H!g_B^T3hcUODQFxAW=3&
z-3G07C(V?kx+*()7C8WWLh!!&%=1sItu24{b6@<?kAGT}KJ6+*0t6uNJ_y1XqX@t;
z3ZbN=`Fth`K%xt6fMAq>7@~95w;}kZZ3usyNg?{c5%}S55e34a-wEDtZCyx=S{V%=
z9PG7KIq2j{;Og?|=wQF!?~TvS2ykVz^4cq}MdI0XGCw;Tbh{WF2e-ERsaXl45F1z5
zhr{;Yf9HQ)84m8>|7)Y%*4Fyn-J8+j`STkG`?JmMbtUcixDgTV&mNpyzVgw@p__j+
zWgAX77Lt@yLV?v$2pJfa0t7RjS1Y5zh3!F};dnOnA!;r1Tvt9=2pQMMFn;vQPyX4R
zDP7*Ug67WK4}>G|5D-If1hED5kqk%+p~j#w3V|F!0r|jzA+qocGD0h5%@IX3hHO!^
zbz{=p1}6}+S!H}MgATU#IJQ2xGxmRVRL&Kd(et?-kIR5bx6|e5*!oVA_A>L?r!Gxu
z+&wt!_4{cTli{$bXQ{$|KUuA{Yw=+0geLF3@`%!|bkgNsc9cZreZJs%C~8B~&vL1P
zlD{z44Gwy1TI<%OF?vZd)k!g#ix@MLt*$PY^Y&oxtZ6Kf$kWU=kr{&XJNJL?P0vmT
zS;nqv$^xYlLI|m)Hi86`0D#EBIR}I!Wtw$>ux%Ufd?K_~#+Xc{!b6uxEco8Qh#`9L
z=q)z&Sf1J1dXC+0ceJ#U=ebg9!Cw(V0C?w{x}A<^M>fPo*ViusA|WZMj8ZL<=fE6^
z2mzh908tC+T3;2jYEncO6&inKl1z86J^swcAAW$y{oax%4gdgL6w_X>*RkDTosS^`
z<h`!e2AWz3ArL(WU|8HXhzKb%1R`SQ5EzkM<I8zDofkxm2vQ0J&_<z@b=eG-)<5&b
zmo~Q_F<LoB=Ymmk^W4Vq;bGG>3qlS(RB1mXgN6`3nO=EvbYfc@y_bI!g#hPjfT*N6
zIel>B(aTHA!z`0;y!pm(X=z?oE6XeA*3X}gPm5V|@#1A6lxNaOj>wX<0LjQ4;dFfb
z`~TO!{pMF+I(KgE<mBYhM<1EZ%9FD(BaSCilXW-GU)b8(GTr5=7Z<j#FbLA=-M!<0
zy7efut|sFk+~YT%9}a&8fBoH86D>x)j&B<8Pof}p;G7R3_WF57p<`50%#kug3eG*e
zoEc$~B)CdS>UWc>D3~LlDC>GO94Vpes<z%)$7!!qRuut`di`#agy6jM3|>m*T;shb
zB{@b8alw$5gx<M4%LqUU>h;sKoBryreA~McfP4M)+rRO(-}`^PKZ+3}2?mfr3?7jr
zkrWCcR&~vc-8^foD;sNDCzTKY6d^cgT?+zz3~02owGsjt5Xo8J?;8S_B{Ip(&i?MO
z=pOFvC&rvxTa!}0_2!$;eD;}xy<MYqO#G4e!TT3J_j#G9AHMd=XP$j-KAD!&`MJ%F
zPAC7g62rWlA0B`0Eq8HcB|QDYqlbU;r(b>XGvEE2ANG6QxwB1O0gDHxb;o33B!uOQ
z%Uf?$<NJ9!IhhLQ?}^ZAEs2!k?XAs&<I;v8DUK&olgU6_*IprF3|U964s$sVK)F!j
zsG4kkWeEqSyj7XhL2XGw>Odkj7+oYp3>cZ2f$Qi+CR%@Q(^REt_w>P}i>YUb0VBrX
z-AwAAwsunBjlRCRIePnW{y?O45$dU5X3Sj7CdH%CP+c)nk{fn$)X$CJ>70E4lu-+p
z_vD4u!SwXO?YpN>KXIj>w$l!~=nhW`Y2>ieIlr<x85iERuHf}<ziM2oft2wamOH(s
zDYVAbVRnD9)1NiviJP6@UY<`&Px=?-4GtW6KCTAK!#-t2F%B^}4?aXhD9Ymg-hB)h
zU39G@l7TUzK*s6p?D*(_q9?X7IFt~*lSG100%7FZwGbIYj4_}wQb=PoLR3<mO(sp<
zXl?Se<C#66l+q}@!1f9hk$tcg>a07QO~xevZ=8SI9xW{gtr?)Tt!<m&U7n|maXc9#
zL1474Yuh%27(!qIB_%O?iY`VH#K6IG2+?^j5QGeE)wU>`y0NV#AP^{o-n_7N<?$!m
z=zIN~5t_E~-aG5c`8-d1Y2LvYA^0rs#2AAMKQln$pTUO#0Ah?GdH{$3HaH}X?1S^R
zuC#xaj0}uQ3(xW7bfS{Zi(mZZr=R=0*2cvU7^1aCnqGI1ce<yir$S1tl=psdm0?~e
zIDd{By|>=BmWY^LmS<HpC&Cc@rHk9$ZvMm9UrST7zOjCEd@@>IX{u&@WqtSlA)>f;
z?M7e`8IYtDf|(eE55cwNWHLEEJ^=Qs%gcZ3YpWM8UH#toeqfC0bh`7R$c8BcN@aAC
ziw1<w7{ki?<_ll>+R1~x{oT9$L2q?!rPCW6AMV?7nq^5dpBXJvC03S3w|5Wo;izfa
zBuzq$Nh<rjd^{efS+8;47~^Bqi2?v?t=5{r)pc1njnJxT9WW5d);b@UnL6E+NVI=V
z%dW|XJ%Fer4f><nI_o?~N|Fp?EQ-1)YtmXEFhpaN06^rXsxLo!<;!1xaWX!1ZvNb}
z&+pv7ySce`?fSW!w~i4X-l&2L5LuC6!4QJC!L~D%B|<17MhYCGj3f~W84-k5wrN@t
z27-_>gy>_4%t)w&NR3X@ggF2}x7&X~fU2qzV^X8v`N`W&(<1YHKKuGtzxw|BAH4V8
zd*`>#b@R^syZ55?N2e$E5B5f@D_{DQQUeLN_{dUJ@kej{&HEp{w!GH;H^2WUiSAKC
zid0>o(!$nrH5m4F=Bwh;6BobvrGNR(PkvNR>x-Ap|Jy(ROZV=*nk!qfU|)Z?7LaAP
zlXNgoc-T#Mj!!$OPNna4*_C!Mlbf}(HD^hF>;Ku=H^=|OfA`WK{JWRyCaw+oMN><J
zE`nFGjqEY9cTpk%o7DIE`FPqk(zY@r49Hes1g190-iLa;ngtHd^Q4G}Q(u}g5h!ev
zu(=rY1*WX#;N`SSJClM}QmubbD5<H*GGT*Flp6N>yOX0B;`(Ui&fz3WIZ4t<*(@)Q
zj*pK^QXkK1qdI4kVw7aX_#_)Fv4dV$n_T7!f&A5t^|R9lcTW!1HkQ=Xp>NCid}U)@
z%b!a^_i3!A=yir{ecBzi<YFvOr>EeY&^iEEXLH@@=Voq-NaPrVRCRw<9UUD`$0t(6
z=v%-L1vuu&L2C(un#Q%!%LppQ7_5{+Ng1_D6XSi`wALeOlO;*Ufh~{_5A8S*QKAgN
z1`^hlS%0NFDh<)n>INATK`lUd-h~*r+v%;YZNB^7dx}s9jVPql06v6>kr9{#Fb9SO
zdzP8m2Ok+F34uhUgi3$FXdnTEJ_Ht|RJwld(*EHIf!A8*S&Zz<avr^N&Ut5pYn_!z
zk}93Wa?TvRk3<p{ZCH+oyzt5O7?4qbR1#SrV-y4kL7*TZScuxXAoL4g`0|s_y)YcE
zINy3801ynV^(M(Xon8cJt!3cFj6&PCi^0zr`4c_JD8^Qai0ps8bCMK-L{Pr=8!M}K
zcJ8lT+^(V@EG?-dowe=8<%>VLd2_U}Ui)A&?YRX61S5igNS-60m*g`=y)3h~t*U8Y
ze|9!jCQY+^eSPiu=boLO97E)+yVTfdRHA@G@Q!h`ej!_06-k<=X>QEX{=xpv&Zyga
z?8?@BG6v%1wWWWp?Q?lTes+>_(UGl=PNJAjDrN))O{m)7kqDv{XgdjJT3`rOQIV1u
z86g@a+NN%ua{!)WQ&$VSVUs3E6oFf7t8$Lfx7Az=kFg~Yq$ZPUA%(NvRnC^JQUvJo
zfq3zko*tj=PUfdh(Azg}Ub=i~_twqx!!&_Luw#gpIZA)+iC7X6vye&&qor?Zi9o=F
z6kR~1D3nu%g|RjYsXz*ur=*2cXr%0P(uam@TFob?qhar-Z@=~QQ%@WnKR7r$9-Z5u
zBoQjpCO;WZoaei{`;q8)eD>~7Kj@8C-~0H^)Y@6oe)fwmhM(gI@N)(dd1RElb!laH
z@9x`g-rj%PE%y%VS*eewwPQ~Law9Obnrt_fO6#-n*-zek<JS9ct#mqrTsLhQYmN~M
zA0i3@(lM?nfi2EYtMkL9wXQk8wybAV?oPHVj0cqqgbl375f$d<_wEhGOXjerCt*lQ
zMwpUJn7Sytgxb(FOxCxTFKw@vWoUd9QJq`sx!8Xy5=N1f;1r%Js5My0@6PMjN`&5M
zZ6)vLX5`@-iR&!71xKX1BgZ<OBRQlth}-uM>$bkIzI1+NH0);cVxDG$#<hD79@HM@
zo?RmQ7naAf$?EC~`x-(RpB)W`>1d^M{>r(ssg=Y;`Q?81SAXj_fAcqgd!;v&NLMdz
z+S-3@Y;K&puzBUuITh=LEaMju%qyeiJn6ju&dr_M2N=wi?MJrHUs+qbVA5XQxU;j#
ztS($|^I~d412Q&MadLdfu1&Q<=Ax`-Md<)ko(Czs5RL!=LI_n=A(4+xNH)f3t(8hR
zQnxe8@*a`WCQVG1n9S(KVJ6pFCps}ENlbrt^ZfQ-E~48203ZNKL_t*2>U!QE4Tein
z=R`6D4q<VC7Tr#Ndu!Y1)cd&b^H}^qDP>F|g=R);LkJu^2hYxi7+`T%MyZ5WD;wt?
zx$#7Ixa0%GfZ$P($+JO{bU4Z^>*jeU&$`RYYrWn89@=KPvMh;2lIDR$>llSbkV1bb
zA=KiHnKU6OV`OWs=ZBVoi$7m$qm4<D%mrwp_{y(*>&35pef`{aj1<FydIn9?G)?2Z
zH^w|1y#JY+$XeSpO;Hrn=_JNa_zxlH8t-h3L2K=-Bc!^njY+aj=Y=nQ5hSgwtmIi<
z*VXdM(((BC%8f^Kntk)X_zll7ME8FXHyQ~UkRwA-B$h^_qm#3`Zl8Ph>2H7Q*Q#o+
zwc6O&c<jbwufF=qU^rmJ$#gEIRz#AR2vLZDl;_<`m#;LfZJTy)Z*OI^{DbfRT?}!w
zG%U;N=;U-V89(vZqgxxRy;S9fltk~mb-_olZL75K-Xd~pWNI*xkQ!m@%C>*47SdS@
z-etz*S&}4XI-gnR0Jv_d7^5+U+1sY}-sgF4w8~S{$r59vAYcwEQ89)XS!+#1ZClnY
z{?GsVce1Whl9SAQ>B}#zZmz%j`YT)8o1<Q~eQp^;71}Ds=wl3wF8Hdo0T>9awGcyH
zmr_XYy&xGG+86_p&_<F(5YB(OxNsPkT5BbRBoZXWaCsSs@+^DlD_{P>%YXOyQ%_&L
z_J{+1^ob`=PS4iXH%FtDM;^U#c=X_fFMeTZW%-}|SO4`pZ@>4|uYEo3^pZSBMD>sQ
z2><^IAN|Q>e0Wm6_3k}sI)P)>=@xZ7I4M1-<+Zh@wt2S`y;mwE3Cw@yXN{|f@!E}R
zAmzL&2ZLdrccfA=1SAsPo^R~$JpJs`RsGwSuRR~!KN+q6tIN;*v+eWGq<E<jx)or7
zg5XB4@P+xiZKt8ME_x30=o1Qtq9#trcO`a(T1f`gesyqj+Sb)G*M|ONEXV$w&^fzY
zP|s*D#SyuhB4QjaXX$?+O6kh8F`SH_r{u~!irSGt%i+DMcy&5)CUKR|QEy*3PeKLj
z*;cE=yq}ADcKq12t*ckIP@_zAqD*X>Y>>6S>h=4+wRxUytaN9^bocBK(#9llV0w>T
zy7;T#{^y_n_1}K{vF9(YUpRR%BQ56B;^^?mq)Cv{x=+n0<{y6#A;jj!S|`soH#WDo
zE)Kd&p`m_v?cC<&BpW_Bnd}`rXstyFjjL)``RL}wWO{lmFbZTu%(A@GAEjAWNF_)K
zA(7M~?fwHrpkrn!m6DoC=iPp{H%hafNji&(bSafmT1vI(aUp4?Go9v1-j_*=D)AAV
zTX?220BLRVyt{wAyt=-=K_r23F-f%e{_s&KfN=o?CQ?!wA(S8`Pz28;<n_lN-@LdL
z30!;#i$_EyHKEj6BS>b!3{vPc$%T+Yi1<)434rLLM}bmFAOs*ml29Pf!Z$D?pq3%V
zs;(C*%OpvZQYd5-LKE=W&;8QVpM9~nv}OY|-oc{j8smS@z-K>mp#lbg1w6Sd%h_x;
zpU<b$@h6VdNe+JTlx{a`>yo3FQc5F-OZ`9igYP{5`Dg3NxTn?j=BBq6Q?t9f-|3CM
z@}*yPUI0r(6iDnDh#3HwNs7Pv^Y8iqOUr8?|MYzuV5!&P2ujJu+JN$dkMBJG?B}z*
zqm{N10SSL$0lV{3spI404?np1<de^Cu5T<aFB_$f9z00X<o^Ag(P$-2vyVUicsLjm
z5&?Ab+*bA3>7>)|CEB!AnI);WmA9Oj#M!#7>KOg%>Uu=L$V-D!U_cU;YmlMe>*i@%
zmxYu@5-!>rS(-Fe4TOj|pU;I5Dlx8Y>#70L&f9;{uxAtuM%ib-@XXS3Z+9Pkbf5do
zbGtituU)+cHhlA+{L}yKU;i(?(dxk;{>P+~6gD^%A#&pu1z|5Gkra}okF2$pLI7h3
z0T7VcS{H1aX~_UpRmBh#DGeNBbgu1m`=@7<rM305c@aZErO&3*n;(7XeEjua|K@Nw
zy!n5@%?F1Mo_*$-^XJch^wCEQ{K~5@pPrrU@9!TR>^bK?_~4^QAARK0N)6sqYx@*T
zuz}DD0BKZLn6j+P#_!xekj$lx0D#s4^L#c-Gtq?R{P_)~uq@^#qw%C%TV7^&RyOse
zbL-DrjNgCrcQ$TxP9lko<i1XhCQ{kwH?Du~-#Y~dI)@0M6>b3NIXMZoV}_B@qHQ@a
zh=e#W3KlxV93dbB$k28UKCDO7??MzvMren=Lf&bnb>FOQuFZ~5+i67wnW$DMIj3p6
ziOL25sudP#XI7zQ2pAo}!NJkR$IqR+yi!f3OWp3t>0u-_OuPGg6UQ(un+xYh<AZ;_
z3tKB*g?^rnk4lqe5c&4q-CnQz==IC@Ztn;bwX+@B7cl{qKmGhSKK}8KqXPwC0ZZGS
z9vo_A=+oPn|M76HefPaLg`l;yWs?{s&B4K;AVZWSNq=Q^eRgujz>F|2XH{GE@?K<1
z5WKBfsVvPrGe+-SfXKmyh(HWJMr(g<Crzc4F@_KvfbibOh)V0>@@m@YmUSb9l#+gC
zyoa<Hl>z`J+gj^`$vTomWxZ;Auzs<B5Rr3kGMQ<euCH$%A3o4p>oi$ta*CpGE-ql;
zQYvdLAuRH&F~;a5F?v3os3gDg==I)csj)uQf=GBD0HTjhDI$~$?@tD~u<w6e^m{o%
z#mK><Vj-hIkeUJzhK276A|enX5&X<Lc5zNuN;x4IVg#yfSif-Tktd!o-Ci3gI1k7S
zfmr@*QJm3IYn>zs05EgYH1qlV?CfkoX`Ic^y#Iuw0D12b8G(t=dn*LS=)3*yz5RRN
z`mJxC+gz)P*_DeIh(jk)<!paG8V$8jAw=JLXRWu^xtbp~eFHEbA0Pe6pZzBr<?C<!
z<idsXoh%&=y8T{;g7*#&@9Z9aaQoh0egE(3#(D2}QC^DdV|2b<-&ozbcek78Qc7d=
z>hkiS-_5(-*|fCQ@9gXx9vlt_OUK8jbzRq0&CEaqf{ZaC1d2@T1+jnD2<$@e^I~R<
zVGh%?aX0TEGlb}@tE;+`We^!S0<hAOL-5vl=d88HXm6c!&RZv?R3rdE5PI8;miiG|
z)bShN`r7KosCH$Rr#EigK*T&xSJzjcd;Y1H|KiUgxf_pO+*rybx2c3qY6PH<$fN@N
zY^8%YNdhsZ%J|koWFdcOp|Au1A%r9f0LGXYS?WYc2}B`A=l!H8vO)j${hhLHPbbr}
z@$`wOpZTLd{^O=;UV7=Jr=EW5<6F0G-MTdx4LjZL{=x1GFMO`9OXpnMwrgu^{K>St
zpEufucW&3?X<J#gF_~0wa#q%5$!rBuRaup#C(}vWaw5~%!V7=Teqnui^>Ft@fQ;6j
z9-TsTxfafrMxkU3b#-ZVCG{$q)VYPO&y>o~lJ?wZuO7d@uNv%ihUkf%FqA<sI`$GW
zh={-lAV6x=#HdhEfK*TlJ_FZbe_o1FNS6e62n`8?m;niauA3^7)BPi`bqDYqrPr<d
z4EbfQ>!xxtR*8Rnqb<H~r+YF2MnH%{*Ond}o>aD2y|8jPJ6XE0sXJMB(6LcW=JWob
zD<pbU7q%XIuv5;)v6wZ!_7~Qc&#x^<A8+41@Ij@@Bw9um-Z$X=?yXx9*k}YP8y{Wd
zbLTed`5Zq<Q}Jm95~0FrHT&S!M>jwG@bLJ7l!`evO}&4#)W5W~-Rb5QoJ`4cRo9b<
zZJH-0(E@}bMM{zq-nrmfTbIE#t|{5o$X+D^QVCK>Apj#&XrryTp^oY=_n93dv3h7M
zv9MM$y30`@C}N2$6RjE86IE>;d>KL;qh$`>TkmZMPAYln^3`)&mpnjGl`i-Q41|o{
zM;kaWg8F|4T2E$n-nP!xF05^AUB7Z&qG+p{L%<j~2J|2T2u56dUPzEiQDq<#(g}kE
z5RAkqme*F3USFgMV~E^Dv?ze26w(lAh?4!HLBKJ@7$XSuQk)gdS<&`a)-K+-(Oq6f
zmce_5urQo*&V>*j>S`jOG9*mXxTbM!?W>}0>$ZO`tI65;@Z>mrQt3tlMCYAAu&z}~
zR!xx@vweBH@@E^H8%O(lS(?20#*bFl)~m=z$HyE=>0WekHZRkpm*ojSM2<)hAv8^~
zzrWw<_3nIp`!BxxgD*V)7)qUY5XV!k)t&o$wdJh0)Ov^jLdf7eNeKvnW1eM3%4=7z
z-2Q*?W(@AVci!$8d3JjG^>6*w_-y*jGtb=o@Z*!y@o+H6^Sqi*7OnZJsTsj~XLPEi
zB!o_$1H`&+>$>(anmntmTj;D1h1O;^o+1G7;wws`%$T~GGsh$`N?=vYfzY<KkWxz#
zg7a--ZL6iKtJ&p8w*JMx{BJ+L{nPEMS9X8y-^;VSEUMmMaO>`!*Is?uHSNmM^2XNs
zk6!)TepkD|Cq-CnX;R1#0$`Ag-be3To>TN-w5dhoT#JIda}rS~8NCgJI#J93gf0YN
z(mGMvjAyf=Y2W?+-)^11kSFQgyE~ux%;%0CoFpbI%ku5F-`?EZTpBF}@9*5c?L&XK
zdhN>I-T?r(;EzvFu3UZOY&`w6QbX{um<NP*qT58NelI65Z5_HvRuxvO1PJq_!`_x<
z(*gDQ=U(V_b2~52t#1{s8PBa5b=Q`AAKtn@9IZq~6zTxOX&%dEh;rsSbea?*$EV(?
zd*eJaiZG&9;ObxlDe~+DdiFM807QRcMrHwyNFWk15vj;bQJEMdLIjA6h9W~igp$LN
zYgp)n(t)D22lGdjBf}JYy1M#cyqjj-nNf$>&QZ28ILU}40*9Kk$XA!TgwS7IpSwn;
zn2ypFY&!EEdU>=n6>{{afBsrZIF-T3EIQjZl`G2Nbs5O}RyDPB^pp4B`uczPu$|s{
z^?QH(V0IWJdx(QxCsm>vpI+SBl)tcMEMaGmfOI;+@nmvx<_En&<06)@yn5A$rM+E4
zD7{k%Fq=&QA+oQk5|IOhvYh)6oNc`g0x1fWCRq#t<3ljG5JJcW3XNKh9-u!ON+YYP
z3Mknl#27+E6bz9F7kUzAP)a=oZUNvSKn6I3fGC-{Z5JIkFvhH}uRr$q<B`MO{{7i(
zwm8)PXJw7W**eD9G|i$mGFn=B?1?9|)=k~ymxoOOAAf0SPa8rXnj*$138f?<WDK}S
zGmD5JEY5MHlE78b`g7YCU-+f3tzOtBnKezF=Q0G3h(vPnCKpAd7$Y-S>t?e#$52<*
zY&NT^YBCwmiz)k0ZsAaf4uFZ!xz@TC5z^G8X?o+SC+B7PlXu@PtI8zl{lkO9lcS}j
z(MvDAbbq*ipp<cqEsA+lmmFP0%V67jetL4q(f`tyzT_C9O1}H*n=0$?A03ZpQ=O)1
z)_L;j=b!n^3-|XA|KczI8Udsvh#mldQIL>I3c_K(G<}6zlkNNWKw^Nxh)23Rq`N^H
zBt{6MM5LuVZ&DfrMoCK7L`oVY1etVqN_R`Y`@YBV+p&LOao*Q;e$v~`?SIZAJuBbK
z`G(eO@8iRx<(4{2L^@1hY;4!AaY0ut0tH45@gs!j83gD;>44&)wu;yi$7gvcm}#PU
zG^U7-y~=%JPS8z3j_>w|6PV7AwknqIeH>(tj1g2w@RLfuET;CGLv&N^r2kpDL+5=w
zDW{10JK*eS0m&p2_!M^PV*o$=-WhPV`caxkrp0sr<l^7r%}_rrfYEggO~F+I2Q0XF
zBz1|Cz9*4a0BBzx5HN=us3r5laO8dxg1`mm%gx$pFX+B7tzm++nN{2bO|B46yRFCG
zwMUiJ9cIHZX3M6=#uC6<_rU2XdT(Cw+-)BgSUf8?C}5ir;$d+|q#~Z9DUFzfpv!6?
zeIZQqKQguUD{6l&4%>KsN65Ry+l*kya_o9VzR?c@#+L>^C=DZ3ZR`vMr#xU6(835>
zuIL!#?$ybi!|l_gvPfscCF$GyJJN{@B&Ija3TZ27lpH>)&>aN^$rGts6rszKN5m8n
z66q!DK$H^Db@+)DOXx6d!plk6K%<oOP+n41=l4zGjUzH_or+=mZ+;_4&5i;*M!7G6
z!6{VE)jy3|QVl3rA%tv>QCGs-Ci3pV$>#X62I2d(+|}+cQE1k$qN88OCG_heq*o<0
z#x525EkYVaUA_`=T!K%GeczYQo~IzWpUQh&0WL{bngI{*`Qf|vgY?{{<uypfP3(x2
z_tUV{wJPbk7Uy0`(B<Xv($dX$dsd{vXTXuf(}7jrQ#dPE)0+6JVsh#D@6(R~ZeC0n
zKi){?X}$KKZ;H?-{FM)C_*^b*m#kp@P7g|%!&Y9*>s!iLA03yJ3C|tc-q@WBkj{Oh
zX>+c+G?-OlzfjR|S@7u@h))TeM210xX3Zkh2l~k{#<RDJC)33B2jz*Y<1&6Hlmoaw
zkz-q*{`~bVrNpjt#oWlq$jpqT08@AZ*2>H2IczXBc=HkzaI$;{krHThBrb>@>)#<P
zoSZyFnM{O9MZh^^mV!i|R+An()6I%Lf*|tfD*L~tiz#{9E;vU5j(&~pi9skp6%{Vu
zOeKxfnn6W8*~S^Z(s@&F^bO8;Jpf6XJQ`Zsp;&g1jjH@V96_dmL52MK`XABD#IB_&
zZ#_MM>i_ybZ|1}go!8XJXK{R*6|TNt`a{b7aikP839jO3%&3%tyNx(cC9$ZVOi>fZ
z51CYMWqsOo5!$7>l1x-m_Z&>FnxTHh>U$hNV~&ESY`&RyI%T4Ci`j4dG6A@or|0EC
zSJ0GRj%Me=0BT}I1%_))%IZ!%ezeu=ktmi<u*dSN>tk4V3l0_O^%fOz<fPDJ^l$98
zeWXsRFg@7H5!8iPIxfsr(Xkg#>pKYPoNy0KK-%gM+R2!H2~>r6G#L%kjuW8ZG>-QH
z_tBL+K{t34)>z|N8c!GPfN|K`c!NWt-PefZpTCf2Uq;7UZqFJxr@99IT?{E?c^&cR
z&=DDdAq8|%Xyh0;$L-S;U3l~X^~=wUG?K#&YaJGT{8yz?zWNYg24($^Df^mU9DN-m
z|7UaUjPdQahrdIO+#N1_Sd;||hQ{u1B&hJ~xEAK-#85Bq)BBtyz+X$|^1#-v`P*UJ
zw>q<H-9Im$Z%nwdGB+&79`G~n2|rI$?pL-^aPtyhk!)1MNI6wdMF_kYqh0Q3;3xAv
zj&)o9WMI2N{z>Y^VAk8|{pp~K)gURN5vS+rZ#=W3m7OFj9?x=wI5Cst1`c3sXe0ub
zke|;(;>WFA&@k1}egZ_8a&cX%s(gO(yy$^+Sm5w*HadQ{XJ4P)hsJAZ^S={_k{3jz
zaB#}c&7M1O(*8G<RAMLlTqC47KdL_XD^c)E2JxFcP8k#OE5>P&+`x;i#AVeHDYvJi
z;i4+qz<^z(8*FHyC#;6AWMyBYPk-v{=5R{7Kb0e{e}ZdYFckpKdyM-dSw|#Y5y7g@
z-iv3xSqV5?j5tg=y-=13G{0&Q=gXV-3mkX*5oB%SP8nPI_8W_;pP#sayq5-YSV~kJ
zs`Nzxk6lY^yT3mYf%s!D&BM|R(E(r4e*z6g^UABdXG4zrrGNRzOZG}bkdUw7oqSHw
z?`%&uN-4J}M?4@L%oULg<ylQ=KDrP5)`-9a!8GXkpUHy1IxZEc-)JYWGYe2!wk&;X
z@LI!a0R`PZxI^sth9-AC#swcDJKLJrB<l}RmtfQoy(!)3%HKAP8(hT!|LPX`KtF>Q
z0%=g4;5w(bFDFS>SOq`-{@ug0O`MUx&O6DFq``Qp2{^fr<&i=Km@*8>n|8e>p&&{+
z55W*^<b?sYaXKV*yB?BR@mVe3VruOun1q!c9#D%sHeoUZ>hO!6JD{$X3o*a9nQo5N
zS9dD7@3GQ)(}fD1$)W|Q#4h}nhK61g!|NQA$>jCaK8-S96_6Dk6No!5s36FHh1m`8
zH~ZWGd&F8&3)I}e&3>ob;ca~J)Fn9M!`#zpYiH~1E1UBIu6;7g<wMwURtegc<jB?a
zwM@t5sfVzDs^407#O0+v1G)lj0Y6MtG1f^sGV&LfHN&_nOz3Fm<9t1eOomMp>TrQY
zq2XkR>uIs=9QT(=#SlIpL_MTi%p4xk;0RpAUgEdeu5lK}X+rhkb(J&ROYQEgy-gi8
zzJB5ZwZo?aKHJO7*t7wQyB;gR)7Cptis$LW=4~IueMLqD=@oe}FU}O`So<=1Rf=B*
zeSxaIGmvLvFM?1#>y-sFsjw?5(0_+he6G4o8Z}C1VplSxcp;IKKo_v-44jNS{Q>}B
z_ulsdUBO%PKu2D(?A+YtW_KG=DzDZhUsKc8ASu}cg2=|>MWZFd6$zgcJ7z4V7&eIq
z6FaN%56gPPk-Hyq_u=S89Q#bo?$U2sJP%XV&Lk;cy(s1jEn|z(W8De6mj13C)$gyY
zip(K=<++v@bXm93(&87W@8z9XeE_sny)f!k&|zwP_-D{7yZ&{4UObWQIZi|{)iZyB
zG(M0#i0*6D2XF!}Z%sUkQl5P~E+5%SC5$d}WIH8J3@TDkY@d`ZIeuCC3X6#-iCK+B
zktxcEfm;;MN(V3G52b^+NJ-Yv$D7}ml`yPwkGXL~&cKmV&mvw3*Uzqtj{x9ijh4tV
zF)={Wu;bz1zj0%y^WrT01@II<vrVtkRj4yf8PB4K^ZN!0xI3Jaijc+*j^1h;Np$*O
zqgZ|F!p*t4xsPU4Qld)XL8u+d1JpEeDEP*d#2uuWtN_M%&xD1aY(&Qp++YBC5!sWP
zni|ykq)iQ`Lh<8#pZV<Ofq@h*MFo`T=}Y=l6^aMq<3t+i3?gfv({m-%*9|CuuFf<s
zIx>b(Su_7QrF!Y`YJ#8fz+)eRWJb>a+Jq<oD}<&yE^&M?N<>34hbmEmYgbzg$}?Y8
zM;~6n!7_2`AudIkC+KK8h3aH=ToCsm4aGx+X48ILn3n#OI&}$L4!uc+vN0EbhJtHm
zlYP2g{hQmyeD+=2{$UYSTktB<WDZ$}_p!Z^&u<G7O3)={v+$!~nwv<+Xvq8eXr<l%
z?(fBSJNxZdATWDsns4p|Ycp<&HsiLu%ZtchIr={+eAv=vJIDE9ZwB~x_9usnH2Rcc
z<>sW_`4Kq-eLV;qjX5L^v$`wJ4%HKqQ7H4(x_7ng+%%*E7mX``*zBl}jQff5)0&J&
z%LCp_u2K8d%?FPGYzi5le28*S2y%+@xWm*x5m{uECzmE1IviP92|GJo)7-zG>d%E4
z8O3%{)6!INzevY=P(V#G<$Ge(RfQ~D9b`L(WF6EE0|u>h*ptxhL+nR)^F7DKJcdQ&
zJ|o9X)QgKrdevmV$0O{cRv8;Y)bOU2w;kg&gHu1j{$dw9c=(;xC>sqN2W%6>Hs@rg
zqA+HSX<DlIkr->uPc*>osht?X&$rd)JXyl{wE=eDo7>=wphmi9uNMMUs_e3zV=HR9
z3S|)yz&LuwoCIN@t(q5tMGq<u!krbIZkheG`V2$qd6N)TQk^_+u>V<7?8;r?F{<>5
z^;>q^WnbbjqI#|*4gIbsH_>UMyKld`!$4zew9gM(q{gh~?C2N>D-QunytZl?7#pBc
zWJ$Fk3Hj=!!=XDhbULQG@Iy_tSn=d>8nJK(MEa{%n4Z+_1_baEGDSwSNSGR}m$3D-
zG8X`k5hIU}-p)_QfV;O}uX^>dlQ4%qKibRi)$YfocJS&fUpP`YzpP}n9+zhJ27G`c
z`I(UrGOKto2>n!Mk7iRKB`}mEw)D@W-{uhh0Y0u|Aq8hx78u&~t4)vb|H)c3{IdXn
z;9}BK(0`VLaO*koxy5nC?OoK4H?(){94TX8A0<je8y>}4XwLerG!5*25QK4N@#fB`
zmMXK&snhPh6Yh}7q{<ymYT(tmDBj&XHzGEsKp732>Zd~b8cL^pRtz@C0J&mHYwfZZ
zEUOA)4P$-pUe`P7Yvefcf?7qhH1~kooey@&WuwA5ns2?u4OC=#Lj_)iL*&4`YAi~n
z(WAKf-ApL@Ka1msqpn`pk3nHKcfWw|t?t`@FXT+$4jVh{x3K@QH<+qx$xQuGc1UFv
zwK)BQg&D3I_}3RD+41E57x|LLiSJnh-3yJrjUP_+dTJ_#^=i5-99dZ5+IC+6mZjdm
z;a?bi`xT(cdqvUjO5X9ctbPOs@g<QI5{yj`4bDSxdF7*it}^(#AP{>MA1ex8Rs0}q
zAAGKCR+856Iz9?@<7Fz>wo{Um{|bl8ei>j&BFz6d{~iXfBa(b48-;7=uS{iig8s6{
zpi4k``y*dYwxp=&Y_oAr&h+6VAh$6p_MVlA6EhFvI~*;esKSw2l%fzGlGC2!{50L@
z9OSc~)%oafesge1`>4IEfcx+kxLIP&YYPIdRZ4TZY!OcOBV0vZ-0#894@}D{)L6I1
zi(^^F)Yu{G57*Aaou~R5m230O1+L5f%n=G@F*kFks1m26&KugNQa}hG<C2RD*|2T$
za1%bdy6bqA_Nj89p6pdT2Z3Gf?HV?%fudBdMr8?Mk+%AAI|&0HRt72M7rfj8+3;Nt
zj&MOGY#T9*RO>EZC2cjdx6>^@s*<@#<#CBXOxnV480(XiN}I!NaTKdpQk7B>c|un{
z(1uyFpbO;D(J-l*pH3+-@GF+|8Xw>Edr|T32h}f5j?OFX-Zt)*8?4TchV0DH%Q>Ea
z$A`DeNlMA_c&#%El(gOG+#KPuMHS@@(;95;ane599|{_hVVpBCc7?bk2%1>9wm!c0
zi%&~S?d=QL*VRit3aoufoh>z2t2=Vk4_kdKpheUfyT2|>FYE~1-R@V*oL@{BlX6oX
z96jtrRy&%u-A^a3oSl@9K?p5;Clzb$s%M($bg+I#Ja%$XX4)okC^<MT>=&n3o0%;c
z2JRz7gs0QdPCi2ORSWYAxP;%f<a7kN^$eD3Sbbe?b>}YH>nq&CrgcmgCE-^>$L;Q~
zlHK?}#Q>1;-`B0LUw(y#JmclK2w8vxI{RexIWBO2<yA5GtTe7~%x3?72alJkpA3~H
z<j0;RywzVCPw_mH_1kCKyj=WrKir82+nIEU>#P>J9>Nd3JMWp|mQ}5JK%~)Tbjlwo
zuK$?dZW+r27Oot>h+NvPc3x@re(G7EHR<r8iJw~loEefZ&9ie$#8v5{QxWyf=n7@-
zjCa1($kTWH)hbeaJGBjALAeg@k58M2L+@X;e-o%_zq%N)X*<{RbjUTVN>-^*(RpSS
zL%}xi1MAQ0R!i)o{17IkZFU|xhzH{f!EqIAmEg8w%lr+c9<SEJdPmfOJ5Is&(&*Rw
zn%@;Ifam`3M&^jjk=^DRd~pEpvfF2^V=+Az@4&sI{R+0i+wK<NadPiIf**VE*f%|<
zQ-Vw!biC;KBbJ_^KHR+e(b_tJ2mtOr7Z8q@S8QYBpuP8CzK6U*pDi4I3~=yuvRJeA
zj}(1VuKWU5i=XE&6v{$HuWy9eRvSuL91m3lk|^m0ubcAq&GZe#=%UhehMV8TJp*A{
z92PZV75<T=BxdSZ4dut$sjsPD4+@VW-u>+EBuZtVj|sxV!<#NU^(ljYX`=`6#T59p
zKQi5rW77XUW{_YjcBRs?^-txHd?i{6%0-o><8FUMyrkEGIe?I58MfN$x?<a8rSiN0
zh__qiSF)c#bz|LfO-+I>5}c#~873rTm-7ha!w%P_0K1gbDzP)0ls8Sr;dKSv+lg)e
z#6?BpQBhkw{{Tr#<?k!8;{atsOu+JBo49)qhX?v!!=~cQQfZ{@g?d>PQVWuvUcUa$
zmj3P=XT^9hyMbS=-bZ~p_miod_&cb~=W?6en~!GVyUHrTZqdKl-Wj2;5$THTO=QoN
z!n$}zI58?>K3e(O`FYXD?J1h^Vp-+Bz?Q=OFiY!^i1M}99ZOyh_LExM1Ze%;bvb;L
zDvcew^t3T0`4tr&ymRHZ;W~XtfTLXuKlW9n61OCy&t)xwAPtL!Y(o9J>BWE}kDO;#
z7azDLDyGWy6X<x&43MJxHzC3$9Jmf~27O{oaLp=#Pa8Gh45j~Ay6%zV>gM<8ZZ*xF
z)XdBM(m{{QciuABH|ebp8(Vt&sB&o@1ed@tjpIi_<5$I(ITFY9Ul2WE4+e!gU&Weh
zkvJ=}i^E9yg=lDum|7+Dzh>Y}YDBEEK}QXbo-J6YPmJ<ml2li}`1nr0c+{wm4D&g9
zo;xaI3)(<x9m)hVgy|`Ix-{jb{p!!_M0FaHDO!+W6|Sv5hTR$E($G-j?mnE-KAqag
zoRfp*X)odhkUNwkzyo8*r?4<Y=l!1*p8A2aHa7Ltbb-T9UKrp?Yye8063!PE9U3he
zxOIQ8=iu$1W8dy}Hko+5EUp$mv5O2~{C57GIg>;#gM=kE+R*F^bv3Qo;cH#oQf<H7
zW4taYWQ4|g?kn1|#i?XMdP>)Sk^70xVh_OemD7F!#6kr8^uc>Smp$cI^Y!?8^|;x8
zw~N=#mG_7)TmLFr58xonf2z=8ZSS(7jqKN{?00&_3qwl|3Y*>U+6e;t9Pzdxx*+*i
zbZdH|Mbm^`AvzP3Gy|h%qMlo)^I=RlTiYeuTQ6y3X1VJ((KLh#f|G5y&BUb-jI1gv
zsz!aj7Rh2l*AGNr*=kpc1RAIM=Gtxlv^pR9moL!f58MqcTmyI!(?13B?tf)ACms-9
z(Wm~mF|&MA=Z9#h_X!mku?&*3R~$bVO?u8&7{{<#H94OsZ{sScVKZOXn_N**H62>>
zlOW@SYy%baD-~N-!EyGQPIOq=F4w*?<CqXXP2LLLha|I#Gidqp7q8&8{ODJKd2xrF
zDYdsd&ugYr<Xr%1qg<i+3F2Lb?Q95?1oM$(^le%?xC>P#Ek$mob!HdgQy_`X9S)J_
z?E}jxp!PGnhIt(l|B$E++wO0y;T{s$`-wB4Y>3BuEDdsIl$ug`^ebbpQEB{?Esmw0
z+E-4~O=edE?R*R}UbSElh)BG-i;R?{>h-(#eooSHDFD|H1jp)m@S9JP#2xWMDqskD
z6kXA3=+WWPM<;3PICM<;VD!zOqNKUUPm4BY2?YcK-`CS@Rekl1AcCs*b=z8QS|)~V
z9!;It!KIoAibG-HA0xR^Gi4$|O;y2abE28wDq^6YWSfsAq)Om#Vb^yLJv}{|-ad!(
z4KfeI<^Uf{`}arIl^ag(f{`I^9DC6oNv*`I>jd!}$swlg|6ZcnJf4mpH=izeNZ+Qx
zbOjdAg{Q^G@d%UXq|E?5GQY!H0aVK@7;X@;4kz~LrFTu>;~=Z9zzyLWCIk~{^jm`1
zkkEfbV~Q5=61%}FIpsKFXs~dfAwo#JC|qJ2cupVul8pX&GX^;h57_J>qS^`NVQ%vB
zeVp7q4tnxD>Uum~UPN|x-mW~R6LWIf@R&ud+d9~sv&#7H|I!My_HtcYZuRCd>5x3S
zZkV>xS#0*U@qO%_L<TqnwTKLR7(jPPQ?Y_a&+DoJlF4MrEw$5%kA0(llTE>N=t6jO
zJTXB#zM*J&k~)8&#YSROttq9Q^?*l!=-xs_cm-~Rh?K-@wu7T15Al9?clY#fCT0sh
zdgW(*=OUj958eWoZ?VI=mX|f%{q}4fIPHX`InrNSqcH)Ok+f|mg8~|(b)}?E%jPoP
z)(NNf)*XTHOQDvB9x3(|T4%|YfUEcZF8>;P3fXMGN%{OkBzC$Ko9%lhz{~<)#~-=f
zI@8L~bp06l8#NDgvx1)+7f-B+M39k?O$qi~UyBgI$uP^=mw%9J<8w(5#|f#>gV6|d
zhaSRR2R;Lk_sI%O(peSC<;YKcIxN{<vR$GZI%;GyFEK5#5W>v#yfj7pe;o*DGN7JU
zt?^|Yx!x@mx7Ckhzzy&?+<D}MKf;sSdc7)VQ33nx(rwFTq7RZ=Kkp?l4Wl8z9VA!(
zci%0z*G4lQxj<<gIOMY$;ksK@Uh$+bdiT^!X2~ZoajpujX1)vpW+g<gGA+XMLqzc_
z7d<E7{{%Jm;>%A??<5x5m(DBG7X&8ed?caiX*OuD(2~=8ZO|B~Sz73kvx9?!Ow<&F
zXs6M?Hd3^>^2%yNx8u6Bh|{0hqP+qNg;FoEeRJ=!p)s<bLHwaAQHn+3F5P5kN>pF$
zCyxcliJ0O*j2BNiv?v!FNwB!1T3^O%aVZJ!!})%i<)?h_fMBa%uC->?gJk*!#EP?#
zIR`hKte(Z}+91v%@`Oa%s=#3Ug`rzxk(F`e{Qz)3vF*G#yEt3LtO_H!J_ejS&Tgs&
zUUE1lfrw(lm1wjk#^!sl6K81KAFdYQb=%I8Pg_HcEUE~T-;^U#cUM_59WHC@!;jK(
zGh))i<9fJb$CnH%3-dm0%{Mz!)sJm8;*^O$XsV|+R^I_Us5&sy7B&TKyK|B1r_UhD
zc%{&DrVF&3u`l_=**a>}vOszLMIgIkwao7rB>mUyukOBSMB5mmOe(?lmhtW?QJ!oO
z@8ami<t(x{3n%Ms(7AWY&()uXfV%~*ziBNgHk#GW-)KF&z@Jl7Gap?}5azcAM64+B
zLu+emsQWH9ayF<l&uVTc%bBBigK2qr&PcSTt)u-gB1j*e(_>lr5fJANKW0SRr-JFi
zzu;(!VXKLeemxf3u0UdRXzR;ClJe`gW8@Je^gr3i3MJXC=(e;p%sa01WGHiFfMxg|
zDz~byC%(V%^q-4!9%Ov~GA%9L_|OZ*K$x35YrRSA-xz!6<Ka>7x2IREuDr<V6mX6@
zMKx`%w}xb_290z9Uiofb*-%BZ_g2sPj4%DZBowjkd7fYdY*enS_DY$xdf!~!T<P6C
zT(7LcgEmnZ=d6zjkrW4>aye@mlBCq*@ZE7(koT<p^0GvSNQIgI_wyMCf@PZPr;qOE
z1R}zn?g8~YZD^w*;e^{So9X9qiQKl_L1?nz8n~OeKb#*R<)aVhAtwggDXKB0h5eO;
zgnW=jzr;Kzn|?nQqrjy2rLtbTm<6jVA5%7Fk{<NmOKNl|N~Q84?v82dM|%7#OR|5N
z9Ir=En`{nKLOg?if6vi(u7=k?q3Iiv>D}j2JnD732L6Em{@?Wgvm$qQPyLG$W1%hG
z4<u}V(A*#ZR`KiOiM*TJ#T9ksytlgq%XfIVpW`Rm4-Xmsd{S<=M!4?Yr~XN=KBlIU
zT3&QnZ5ny?#i`W%@2}|MskvQDgob`vA8Z@K_Kb<x8@;bg=p`GxMRL!ZqcqHq0o`0I
zMPE^e1V(O|U5EcpUW^_zXnUCLK5SdqP*EFDRf)6%ir_+7Z-z7Z(s-5F4fk1GKE{>6
zdsRs=Lh6J^YS&ycwwtGsNjOp~q7tY1#U?I&8SqedgX)>&B7E7bc<T&>Y#e$?8|vcG
zeJaVI`%wx;DZUz^-0vqj?$bt&Lmz@Q!Ms>;I9*0{b5sP%6<lc4g>YS^U=ZXT@jB)<
z2AMmcq0%jO$#fpj$BZ)WhDRq~4H!l<xW>K7PQ)pO{0kxdJrp0GgK#WVQwq^Jq7s_d
zhXvl?RkI94>g>+``tzrxwwGx5``k%Od;43nn|t-VKs$H$b7X4e{(j<qe4mft<3*`i
zz~SZ0r;Urnh|<yp)&QQG)Z$<gw9mYs{96|Qg6X5ujZ@U}s)fi^vu~K`c%p#N^u<>|
zOH|}GW$woacuagzdV&Py9>F)MP<6cXFpKIjjpgRcG}5WzTq)zfVPQ+J3Xv16Z<Usu
zUM0*lUdJ(;k)VDpcY2Bj75?{dB@ZaBCOed@n4JNwD{y81%CC4FBSUzef)SH-S`}EY
zACoO_a1y1Vj8$_0y(0l>gQ3;fmsoZ0f(}K3b}u3}zteD44HDE5UR#kwTBVsayIBXh
z(#LIJU(-_2dP+Yvr&r4aJ~*4-Eo__nB)KjhU!u}7Yg<}idRD-w#GY3M&(ntY6WpH$
zb9_l+kM=?=-OdR4Q)FJ;BODq~0IbM?x<RlZKckXRfxe7}Pw3i8V_)l3et*c`AhU^+
ztEA4A^pCk)zTBVr^dQYpuGJusbA%9a*3{B+AEre}x#6$)?H}g(-)uHIFRjIOxNO`^
zO-;3jwU7A(xOkkL{pq|(lI?kXygGjBI@Vs*%rt2I<^lVU;=jv>GPc?jJV4-`AaVW6
zx>y@c+K-1kT*FK4z<%$2bl>{HH7dM0OmjOdUP!GybN2%;jhB+5zgJL3$uKm}#~?<t
zn#?P@iw;|I{Qxw8&m3`qaZQcMK)wkFK~<xkaTF`4u@nAeWG80?zl+BuPlYOFyP~B*
zBY06&_FrkfChO3%dvkvt1*)*4`^6%la|Ldxzi@~{L3}S?h01wg#0#hd>sSlDthjLS
zWQhQa@_PBW{DqrIHIw=4@uzK0y^HG`Lz1x+y9&mH7I0C-f2o*U9Q<gcKcwBlA=TKD
z{G#7iRqcf<#LA9>FB5<2YWW|h`OQ>BL?=N$kKWaT_tQtgJ#YN$GvG9z<~G({0G=co
zgJYS>-GAd^auM>i!N5h1-FdR4Wd0l^?~s#T!G>!S$KmGiqPSPA&@*$>*Wd%zsQP;T
zONaGw2Ca$u3J%n1@iLgK632B|S_e(b&jpdnhvaaK=dNWq32Sa{W~E>NaX0a?jGRl3
z;JBie9BOJOr;m~WFPf(JFEz6z`KVqfZ_uZIb4DWvG~c}WrMhzy0zqdRE16xGn_XI#
zcWC+x*Kx?^vzCt%85Xx8bVYxN<L$y^{>lCfw{H*1?(th}1CFLa71aFtLoYf+z6(dS
z38cl#^KKk3?5C7#q0%;2fjc-wW`wO!R;sH&^ep_^=Wxu&hpQL^-D}+{?l0==nU8!X
z<*9T7n21KukdhAP_@3TqRRy{{-L3eyFJ-`5JB3B0t}{u^P0X@sFIqdVI9JU6r(f7Q
zI9u#4lVQNmGX&>T;(zkL!;bpkWD#WNYXJb>u67+`DrA<0RW9(APN35$5$XEW-s)X1
zgsqIN+`duE)o&y&B_&06VW7$NaCH25bPAlFu8&9a+WpVEg;HL!hOND`v8mLl^vy*X
ziO+iIz@e~Y7quk1!Z_WSoioXF`K=0$3N=n4n$kX7`X;y=1N-FdKy{R%K5o*zhqZFO
z&NRCfk}XL9Z)oL_^@*+A*Gg=DWnt%%U}N_6q#|FN+vecoRo_5riu2RoL;vFrfcE$5
z^pN)>Dl>9z@nZ4&f_X-cL*snm*iNL(Bg)x-)+h6CXr<}mLU+CixEgtCwq~+tImdH%
zpU^d+?83m_!=W#RuSd@jXMU#^-OkPt4>X`W%-+nVEJREh<x*4<CUy6&XSpI{3h2VC
zzI{`Y6QoKwmsDo*K3cr7dAb0AdMx&nbhBGA^oz62(vlKuGvCcr$~>PnMC0=O{5(Zs
zzs*Rl81Q(njQK*v1fru=#s<>eg8tj##m7YZBlXQw=_ipw<?Pz@Y4yWd^>T;5gQK^)
zKU6~~eV7O3fy%S37EGu;y4`Ym6=&IX-u0cFccsu$Q3;#n(89??iSs`UWN0j3x}qBz
z8(Mj~lN%Zpz>zlL3gL(R1T|Ss@`*>gDtBiyCV)1fw`h&ru7-^lgWd9V%P02HbxQxC
zzUtCco!46VuMb|FSMe$@Wg|~!9@m^dC^pvtwrpa^7T4BO7aQl^e>JtV2Ul0vMkGG0
zU)J1+2Kz$Z0a|OTJK5iGUCL64;vz!ySP(H|w|#ej_wkY-w}iJvgF#}QYlZq@9q%a$
zAQ#-h(YjSd@J)RjDY%g6Sz}Xo*Dtn8dU<15Ni?Qqui7pbqSgD55)ut9o{r!vc1MbB
zj>7dHF*e^6dsMtJnDfZK`c#<EBcIhR#>spgnCHGVD!N@IWMpvF($f=mbanOE6G<>L
zJ$<CV)Ea5q!^V^!Hv~~bJ|{J}?Z*2+7Ag0awPE&JYTu)BDyo1^ty>hom{l^7cwKyH
zgKI?8MlChvsjMPqV5HYuZeOh&x?eGFs8_<~X*LgYuaFjEPG53IqX2eOOAzkF5i`93
z|DG6kF!}k6+x#xe_h+iH`Twy$Exgf7l@ag>&ynbSp1@BhJIR{sZ_g`=@lM|;N3!O9
zJyD!iL#baos^hGy{f<_JJm9L%yi848qTM8o8G;=S`B|&x?Jb<jG5ST>!2u)~?T{!U
ztX!DQ>lEkNvge&>i~zQVmlJdpGfTs|Mp3aB&8rU<TE;2cQ(ceCoYigjtIMYPBxu|1
zh3%=E%!Ew%|8z<#|1iKmRv1;r5K%*Mvvm~IaYQ|2$r*6&^JQqB5zia02NQY>qq}Ak
zImUSw)f(FMurXqQ$_!E!(*^$uh<tmPdHPHHw5nV!WIV+AmK$(2?{GhEFEdCiG(@Mo
z!m|64mWms3UD<hY$hpmJU^O6y0-6jBVHI&kW+wMt1VL=YhNh;67ZQn$!jzY{s8q0w
z+)SKsQk_o!*+VZ>spc9nf&TmK%)i-amf5wu&s%Z|Q3}t4kB7XtSUK)Y3GcV5GONn6
z-EGBeG=4TDngdQh_AwTRM3zs)Xs?v5U;&qZ%uP`_H@~#nGpvkYAFv}MM^>7W68cx}
z9vw(|Oir4ab`o0pBXc%*ajQ+A=9ST3L`E*i_&msDi}Ss4m7K!;QvkMi)!(%J&LSlj
z7niJ=)`EPc^{V4yfb?nR>A_sc_!6&_hbvDa@OnCrpasA#9r-4_C->7(Y`8f;%=Ig0
z0sm0G*P`klqUirtKZ_54hS-FIx-jY7NumXxYoiOuvcI+%D^w8ORv&@=VM~ge1anW_
zPaNj|2iDZ)$Le(h4P+{-3Ve=_QrE=z{Aq-A9!K#tJ1#vrkz=;2Hiu;r89Mk~G~-X!
zZ6#0yT{;htcW{AUT57EKb9iiQ;8!Mt)+_qR$bK~o9>j2-DQ;+!o~8s8o<{U3>Ug<j
zw~9X4jA^6AhQoA6O<*1DAhJIl7b+@`SHj=GbU(Qr<ZADA!zflwP~E2CEbX&o34IHL
z*0j|hxN5o32dYL{qM;B)@0K~k28weub&9pghzaZTEFC;*eF~hfaq8G4WbF0q4XT`0
zAu*!aW*O;n`C~6z#$gXRNq-b@1T2P8F2dnpLrYFe^?i6(_bk#VLczGOo2^RZOAy43
zKJ$dkQc>rxFCP`Y#8P2bembd9T0)gByt$StJ>Ae7jU6W&e<f%?Uas>`dW<(3{j-fR
zNcuN#)C|7yYr=U=5+vU9<)0FmThua~ZgVo7pE_ZBh9*wg5-UYf5WUd+VULGJ_3*W3
z8X{Mc9y6tznb27zWvuMUXXCMZ{a|_{er|U+AiHE^P<(NDA<qo;pM{A5Tf0BAD%4yO
zKy~1Ganrv`PS?N1Ii>g{1g>|+Hc(y{6vOH52RxGx1~Ho?)U6ZKwg$ENKH1s?0`7sc
zyMhj$HXottC8Z0pIM}mui>q)fsC3XjkRxU=^yKitat$*TYWHOOZ7WA6(9PzUj+2ue
zqvngiJB;`ml8ibwscHc#PI*KP{bncx!*UcWWUzw6NwD8YHJDCZ=|hskK2_2^Y0yn9
zAR>Ji=k|*;@2NZNdItU}Z24q^CjfdI#1(oRc+De4o-OA8xb{Fm#4~FQEJ}!)n42$%
zEgFdjuD=^qKSzCL0bSjGW$X=|r>vFRKGdokwEM@rDyEivYW&g1%j@y<wveB?(!+#(
zKfCxjq-H#o5e!GSv@(Oznfu&q>tzCmM@M}PTcWzKN?tH0g8UhG#<oAa;Iqc3QKuQ3
z5@V=*()PAy?(Xq*`u5Aahg?!D_t@C1tVGoOA6;n@g&&shF1tn=PKZh4?yI`{*#l1}
zDS7u-sL|XwU0Vry@?M8Mm(}EcMPigNSx4uI8L~F6Q;K&Ip){d(ub!?c3d~LDPdSa`
zbv&-GEw?9{c8a@S#i&L;9Bo*bO%>#immq!Dg~acC)6k2d4$8q2f+S^(Dj+g6da~CL
z1?|WLL(QDOp#mH%xD<5HwV{3WVp6)Typ$s%R-7K$v$Ld^qv!1Pp={RH-)Md~1jWl^
zd>2erAYwD@6MGGZ!$QN^0D3uGE#t~6rBILuzXxXMPTaIUf0jqzgU+`z9sBw}aS0Bm
z<BJ!A9*2LUcVScrDQPaGn{`EIaSX3)rP#9((nZE>V~oxIHlyp;^5~jH7)7$wI>o2$
zE}}^3^9X0H5w1X{?aMX&inFU8+Gpdz*vZJ7x#C|pR+{Zfm52To0L;4GLgvk%S?6pq
zUpwC`1x7oO4KIOQ+#c8$gTD7<ilOQ}F%h(iLtEnmx}2rMeTU(^Fe6g~eY2NJ`nRQ4
zInb6WJtKHC^uOeB@nX-f9u*oT9!BMhwvKTeF`+oqP9wZW<dM_wjSSjRfj0XDSnX@y
z1;{Qw6*}oy_ra$((`EfpcZ~8Y4I)-K$S0j0Fhc<!R`in0Y_tFNw5=$eZM{H>Yf?iW
zDr9(UIs@m~q?r6Dq@Rr$gPl383D(|4R<t|I_A9eUmJOoBg}qM3+ecNH$?sa2T3^yr
zL|-%#EEWrthFQhYhYZA-xoDDSX|rw6Rj{f`qr?{{vh0W0M1+LDftil0%RID;(h)9K
z{Nx_w+2Vmo>)1EnufNYO{tir}Ry4ueyWZ{ojZW|U7twsvyixeLI~Dt2mf8C%k|G-I
zyS@?Ck+ZXt6BQMeq+k6NipdB9TrT&fM4UT>hT)7PIEI$YrQXMjetYA;T(?)>S+-LV
z^tC$)AI!*WZEpNN2P9?w#!0n3J+3c)-2S)P<?n5L*~ArtBqHyXBg^Cmt}JJ6+aIr%
z(Kd*pUwZ_XAuKe=l(Sz$s_L<rHnUY_*Xv(nl0^0RRpxFrGivN|BotQE;*+PG5%ZYb
zbyxBn(;gliIDPw;8+1O@cwVsg{O+1&*;~^+e|tFR_zLSc9WXQ=t-k))WuX^~=lFpP
zszu`|-XI$Euj^qfUuNp#HyIO4IR35FdiIAlV9_)q^Rx-Il~Ba<Bc<{uMgN4;VZ?dH
z;aPs>cdGYodn{YBe}UEWQBuiaPK)?+yn&St2!hK5zHk1uyW4D6IJMBY4EtifxFRO$
zz16+87G&@|30Ph3%7yTs^wes1JpWt^GtpISZtskx@**I3JCfb%c5_aPBS=O2vHc<C
zxUKVtK;!pW@(Wyjik`m0_q-el3HqfH9d53vA;sB<Z-WtBGSWWZ8@*j_yWLBx+eNt(
zV7jDf!-0>zp;eKB395WY68nT2UO&zSi;Ig>Yw)vQ!vQD}hB6D<M7HjyVO-oeDDH1I
zD>$#LTy%fPGye6m9?;_ZcvG02W`7Wqf<Vj{)yZ$zoC5J*wC0nt?fy+fo1-MSoc1%X
zak0t^Ng`I|&t^B`VY~}YYDy@@?zHy1s9}1d_srYLI^0o~nqXD1y{YIQ;~~Gl6WUSU
z%N>mXFL2-)DXU1*KW*D(r2n!`U;w#SA0h}YhxcxqNokg3A+)5PvfpXCwbnH|>Vx(E
zkM`~{oE4+7E$%7u-Ox49%DlVV*@@-9J6)O?ch#g)k_))KC@*?>MD;OMKl&%Nm7-lr
zk(5Z|q-{MiiG|zni;TD1C3-RH_9$0PHbXpK80ru@dA^VEGOWB)EkIBv)JamNr*8zI
zaRH9p0f<-HWBb>M68-IRHlj4~wQD-p`7(QCRS=TYa2$|zFyq$*|7<)pQGU7NejHJH
zF)IaLwNcmMXEL~<V75(pHdB;tj46)2YY1^ENDH;6q9~3n_+4KPX;5^i8dEZ|`Ai21
zkA`B6p(l(_ivJN6m==Fe$P%&!Gq&oo18_5cl-KIi4c$Y)@!h?|^63-^O!RC%d^iJd
zellKH#ybt%VRlxGP&WBKn(As=>VwFT5lpo3w*bp1s!K%;B|K>kI#f%0VfC{R_B`7t
zJo6?(Igu7`MaFw~d}9bvPz64lN+Go=(K0Be?}ggOmnYy~%}h%h8v8f3wz>xZ9;oEw
z^R^~b(_5Cv%CNeEJS!fajm9btD4%0ycT}&!--Y++TDr-5#MU&?0c-49(&q5)`e?e~
zCn-GtMHt->lCHu($QIc~gQ*-Dx%3DQwnuER@%PzoaLFN?gq4(3D?H2LQ!$M6!whtr
z_0Z}39~gML!Y~~pgJJ$vIV<3>`l+wln5$uEd&{Qt`r+6J9W}Tkda6>^2(j5Y>K96=
zXR46E5Qa&MN6pM};L&+C)pd)+9d>JE<dE0?za5|1-CKE-jx5s&)mc!&W8StXHLs=2
zXj5NaVcp)v#$NGWl7>RPN(iow`#ggwT#Y>ziX$@|iw=^$o6|aW1e`a0)o}QSSvV;?
z5(KvHa_f}*Y8PZ~2TNrF@2(a~-)fn*pG_i>dwqRddadLv>rD${uIs(m=Kg^dQc`<~
zIT3%-bJe<i94sdi3pd7oLUSjxW^(;DQBZVG*rCkr2GCQqav@5JGAA+cRMn`}Il&aa
z(uUgq8`LIdd29><HmBZC4sm9YUR>OqvvLP;*mnkxLg*bZ=7w*>N{4eg1NR?XaqaeO
zj1*<bB8lH}4UiRLgcxox73Pve{bplc1%V5BztRdQsD^yqfKLWE3n#qX^AKLESxqK1
zA0LfWktI6mEsg(OiQPtc3j3MzS;v8lewE86+m(YXeUYUG_)v!VY!?$|6v?~WieVXw
zRwE8JoQ><7^SNA9T^HO8hf@39|J$*(@b$H@>sUGbQ3~_FD;)c+B=l0L#y!;w^Xo9N
zS*pH~Q`q;b=|lg8yY}%U)P}$4QJ7;Daomq0)pH&v)mJjIWKNT^XCmL0pWiJwry7aV
zybujbsZ>t|)V!~SAyjhf5o<Siq!=kiX|zYwmx=vDeT&6-GbIb|;j*O@l%<7q;Gz@w
z81(nmq#$NmiIKry-!G|(-;mIKXt^7IOL+y7T$#N%n}IKC`V9JsDHFE`tCS|Zfz3{6
zi!{x8E1ks`icuz{X;fTMJ0n4X-&&~CIU%+4{OZpO`vdH%IkoV9_Fr@tEa`M%Xp}&7
zE53Mof^y`Y1nA^+|84U-0-B(FJqv`ZZG#Ss6OroawBEMPG9;~urTr3z2z!T%H<?LD
zMs<<tJSC*oK!U5DW&9a=Z~PD1nIu<`(~CY=6YI_19LL%g%Q$tEEQ1&+vmNK|gz&^q
zfz>{qJPu!#K%SFncbh$A7G|lFrUUHnNc>cZJ{~)Luxjyn@9GrPbzMl?dD?fdXyR{u
zHMa3WVaLnwZM8Hd){tChTz3Y4f5uSV*v6%@_eyJBUHcGQ+rI(Q&WmE6&eIeTl$GJJ
zhqhhE&VDK7`nSUF=;3u<i;t_bGhpc+$>$X39)5I0?S0piIY*8U7ofNeq63=BV<f5X
za%T?lyNi|&uOi!r_1c9g63(EhscOl+;tr5XRG47OWa@Wm)M}bomyE9GyQaQE1SQ|U
zIZ!tbyrI3{?mWA64mw;<b$^@Nv3Q8@eTC1t-*Lakss3jo%b1|?$4mt4NVY^LstI9n
z9}?qUCom-{B`GTDdyBGr0NLgl!1YlpPshJ<<3?wZdEHJ_>35$FS9n>Bo2Qi`9o&vZ
zY++%c_3>iD_JwHGStvt|lFU>0`MG-C@p7vJe4<ZVje0y9ubdTw7;2QQkn_>m2?Z)h
z(mW4HB#o%WN0W6CuK91gA~|;^yTYOH%B#^q6;%lY9A7`~>`lA{{Ck{M&(~b)uWSRo
zoQCa9Or=@{J3U-HG7|eqWU(GEx&!RqIN0Bw-K0ws*^=SoKkmrf)22dI#J_PwS=(sk
zAl({UC0+)x{cP!;$~b#XGnLYLvw!SGAfbxz^|LyiCIDRa7V#KzixU{1(^fy-tlyW<
zblx$BAo>@Ixe1ffe?*|E1Ia|;gKyY~K=<EMUbvNlUGXiFY5r0X5%qO!CGgHliQ4{{
z*V)N2G#tjd(WqOpSI`|2J#(-co}RvHCKX=?3z*4@^)N-{xM>oO;N3ku&YizKT3#8%
zy?$)RzVr82l@gy7W2@jVORBzKcNU^JrycMO^eLV@iT8f2GD{4V1(cGse$vo5626zx
z0x9+>=2}~s!8WT{=Ct25l2%Jyek2mfVeT38k<=_EiF4i0UhOt2Z4jRStEiJ2VSu!?
zLX5$TZK@hTdZ5xcboXiYurE^@vlj&=Cl1+IW#5xKNSWV6N?yG7_`&r@6dyb6@twyk
zCDmnQq!AxkLL$yz9e^tZwu)^6ei@FoJ$Y_Oc4I-N#{P?#TZkSbo)3nKvdGDzYY-~o
zsBGgxN>WO}TA!{KoR1DZ`sHS4Wl@)+h(pTpjTE}1f&Z*n4@qGTuL?Z1Jle2c5qZGT
zHwSrFU1vZ3c;)gzgBaNZTeO`iv2i#gm$7qMVf`t3z0T%33}o)fyn4r)KHl5n)_tMu
z0obf;-I;az-ycaFgXl;}f`MI58N3}HGZT2SX@}3zMBd|v_DsQf&+a+oz<>IC@rQ?^
z<-b%Ny{u`HAmN#r%zY_d4m4oMd&r;oeuCq+yWy+pPshHcS~)F#m-fY%1VR6LiZ?gC
zm)mWkJ%}yE05w+z-)Ak}E^Tdo(}drLJ^cc$#^`CEpARA8*`6oh<0lXhh<%=@WUY`C
z>7zn9`l-A&n4$y7{Q^e!NmIxAplV6l7+#BqH*z2h!lW1<_}iBu?(Z<95=mjjF_Qip
z_s$O+E1<QYv+2Y{|HwbAFS8I^4%I8`8UZd(SKHFpKmgUbQlZy+FEXs*($Az)WY4%Y
zi}U<kS)w)MqV++VAPj+s7bw*vz$(-&)$F{SMsU5Va_xQJYF4M%(An@X^|tfAz4P{u
z`}Qzeach?G^>X0x#UA@W@8V*cTZs%+ws_!eWc3OY$%O0IdiQBJB{VQ8)J~nfOwFz^
zUKpMXR8+x*F@pKFsx0DeI<PT%*x=H4C`ZKkKU-%jyS46Lc*wG=i$PK9A<c%7+|E&3
z;Yy^m-#-->LggAyuk+JFCB%=WdY|Ji?Bfpa#wA;+^V7F7M7Fa_Mo%d+w=)RX$zol)
zeNFqn$y8-Zf8`~vS2@Qx-<<@1kK|^gYY@}|lPQw<1xqpP1bxr?t?F_#T%n@Ei%lr9
zz5UU$e^20$*x}k65nc7#@|8uvduEj`SA3W;)2h1?DBB&2x-brcjv>{DX@l`9|CuZ<
zT{CaXU@Z}N+crQ`sw-d^gLn{)*?_kw7$bh-zh@XWICS*kI?~bL0yrdgIwMFY;QIt1
zMYU^xCRNnE@xHrwgJ#I|2(wEE$R@=d{0&1z?<p(VNA2Qc9WMh--jW$+xwJ*pO61Sd
z|Cuq`9~^opS^oK8VwC+QQ~cn0LP8yfNG_>ueNJJ!`u0eDY2p5enW+0wv6B1g|3|dO
znx32=m!7h%80F*WzkoRO2}7sF=>bVKyXu8~55I<{|9(e_I$!qQoZIBCPgZi6!WE6u
zWZ5v0AqI;2A0jyZW3J~{tBlDHtR3|j7CU*+2Jhmif?)-a`V+SBH2$|cHC<4@Ni<RQ
zDn>=K>GfGM%lHB~n=XZJ$=vsG%#CoIb!<HqEZ#9CF>Y|N%1?UOc9i=!E<i7r)aXm1
zvZ4{$e*O$Ky4!G(kb*;LDtoO5@6LaU2KuO~lj*H3EHqM#E?dFr5~5J8v_8<gQ3ls}
z+x<AtX*R`dX8aIHC_c-JC`No7ZJkPyo+g@^1sz0AeW?PRSLXe=d$@uA=ABOAasga3
z)QH?Ng}0ow&3{;IJ5DM|7kG0ZR$0;Jy?uq!xa;8N^01xobljewMm~E;<@^InpU@p5
zW=(rMtA1(|TUwNwrF^cm%jO`M<?oM;de3|_*9ukOm8LQ%9Wg0a!<HqhQ(S&pUG`td
zGFD5sAme)VioN*94{wxNNh?SE{HXPs%TJbFn}016g@8ZLaK=diLTP9Ze9_BZ4`*!O
z#3l9gysj7iGmd(yEft2L3WWWh9{i`jlugrxhYHldqdxGeQ4?`DO<QMdsmQs=WuJzB
z$FJwK-a-*SXCpk^Kb+c5PL=($wySe*NT1e$riY^;&Y3*eiiwPS(A3vWlYnD?88nc?
z+0%MsP-fG5jzlBigvvS6ahACr6p&SIPt&vW+PLj6q$4A}*>$(k&%H52l_eigX*QIg
z^Kgy7|0bkQ5SAqqw6VA=&>yL+W!~BXd|%1^0?Wue_V()lzuYGE4_m&vLhWBd#t7N7
zEN5vtj2>REmm$Hcg9Pa4y<~4dECfI4P7r(;vh0i)c^Uvmb{$M#9(Q-bCa=Att|4e^
z5Qp67L2d-iByex<`f=UgsyG?8ado(mNP_^@o3{<pULJQ`A9XF$<YVr6U0-+HAe^;(
z&q-yjQ7rM%@?+lP0Ii9n#M7hyb#CYBMfHPCXU=4a^YVSP90S|yeekxvbgn~04wv~C
zheFTnOyRH%Gb6&nhoS3Zh^NfS)pF2A_4GB0WT<ObM)d@d_p{=(3|r{D4F7r@EzgcV
zDMpHs43Y3nFN;TGK+g<FAB;!Qofbyar4`>p`9I;aJ~ve;t*H`eKYyF9XuJM}6l`Qz
z@Yz?DNy<Xwzux6Z4@c`Cu6q{;&AR9ALx!oCu%#-YgqFg#@l=Cy{)N`NJ{|q92HYc@
z<VYn>d2Nd~aN&o40_N2-FQt>uZ%?2c{=lKo@UF|`F#$G+>7-jUAy7sT3pnQ*8uysh
z3gn5oFB&|(C|9>J;3<eGV2f`^t;lwda}D)c=lK++#x7t~N~mCiBd2g8xpFSmM+S@S
z731yG@3!OzIjZfeIl!mY+{hzc>E!kGTI$s*8Af6mh?B`p*-eGN2q8gbDai~BAjA|!
zI1vI_%b}+qKe1+;VNh%UsIgm-ZFIbK5VkIvuhZ3aSJg`lRiY9q0^?XrljLl?ULmfl
z6M&w?4<R1HmgvAR#8&S}GM_Y|7vUa=jb2ozuE2f13}qm!*`bnS!Y?nR%6zQ`ei7Px
z&bX$ipt5zt4-UpdpJZc>Sh1~_9NE?b{p`(^Q&R<N$O7mW@NQs*oQbS(xjk_I5&Kxe
zhRi1F<kwqZ&Bo)RRI_0oVf`vP?QNR$!+7j9O`Ca{n~k@~Rrg%xoMz)Ssy^jSL+MAW
zGnBdPB`LS5%_Qe}Xz1Kh;QdKgpuRza-=QV#-RwXLYbunJRTIT{^C3kVz3MmMsDf3N
zPXw}r?3?ceX#j=fKAK(HBaNy34oDn?g_%BM(^m%`_nC`Y;AiHpSLs6F?X5sx6F+4m
zidfNKTK>)8)-_Qu9+gY#qLkB+`_3vgSGw^YRq}V3cX$6Hcs*i7()%H0W-G1qrS(Ez
zdFR8}zH0mOA+^lqq;6Q3t#hLc#Ra7&IX2V9t|&Ed0A$*v6b0RM|C0H4-PzBcE#>RB
zxZCb|jyQ4&CP4u&o2kV9t#@BZ{~C>n9k_L0dpG_W+xfeGc0YKa*XrpK@XyCT#brK<
zBDM~1fpQ@>Hg>p~E+Hv&d}?lLVnfH<dwUdyWXls>hJ;#>7<<Iwpw#mJ(KBfh71r}9
zIlRCq1|U&jjIhzt#lYK%JXEwBr^U`vN1!@J2PszLvfuXP8HaXDL;C|N>h5%Ge%^gV
z`g|VMOI0V5+aB})+liI%cFGa26Q_xvZ|I8_$h%d^xA1q^f4XgaI?5OEV47xp$kYlv
z9iqKgG%B)#*=>4QK{EayP3IZN=J&qwP<vC<-Xr#|*rR5sS~WuLt@f<-*rN>{#NL{!
z5~`%A)f(|d?Nz%didt2B{h$AfU*6_bo}BZX``q_+U7xoupBY@T>7Su2A1}NIPABNv
zEU27<j!uqWge;qN<3H^;GAz3KIegRnCg19TI&(F4Rg)_qZr^4N>RL$qO^Dm0CL|Rp
z?M<P))x*qFHmna<co1*pRY8KM$^QYJ7x7Ex<sJU4pl?DNmbAe=ipi4o6~U>$Nu2R%
zNg>9iSxh@RNh%!MpMrnCqS1^)0Gk3*`WkHE>7|lM_)KWj<9GA>29HervzT>A-9--)
zsFxx~7+k+<`AOy%Jb|fhr#C#8u4Cla2g97hGA>071S8wIomW@SANw4gqc_5pOD8Y?
zxZgT5eO3cEJu{(4+o##W-7uU(QYi-^l{&1VV<fUdGu0u~%ObE&n9S@yfPAwUVGFCu
zGKqLITt@p^O6RNfCK9Wc`ha-46g$kkSre3P@=_|9inMGPl{8FmAzGU3wHv5q$)La$
z=<vAAYsHZ6(LLAt)}}hcq#mAT#2bja=WAj|WTNXc20=OIo`x49?&9`)6vPQNBYbcP
zimR%m*-Y{y5iNuA^R7Q>0BYjiy`14m)&@Hoe*HVS5Mi3MAV?Mpg~_@E!XxXr20!f>
z`uT({91FAyWuhKWtyj0ld6&+5$m$s-FtO+p$UaDWD{CQfw~VBHTU+Xl=5uDx_*JM*
z={HGe{LfKYJ;+nYE~{(nm!}qwBO<R=PC1%}<|1H($l@PIqj5*CfY%a_U+0}}47A@K
z-`>u67)|Bh{eDmB@>1sL=#lEtAAd;}N@qABSgd)G3MSqV#JYMG2Wmbb!Dn*L8q<r7
zLxUXMMn;fK!2;<L63mt!fCY`icoPX@z8ytX(c+O*WCa<-PLzB&Dl{E<R{Vw^Ddl)v
zG|#!%_UwW4Mj$*72wxwpi~gQ=vzq7JAd~Jq7_|02Y;*caQLcMQ_(sZdwDTvazEqjn
z!sf`G><8yddv<T$3O-UDs~J&@H~B!EaMDeDs<UDY+p99Gba!(jLwEgI_<q&JZ7{ot
zB34ooGa{^TpaI4&W3#OWD1%5`e+89Xn7jMgxc5Wh{El!>K>cI3OO^Q?LNko&i~Z$=
z8q;Z{nasqsZ-EuE{0BnuH#HL)X^P?ZO0$rGv*|E{7M&T4Vr|&o;d&&spYaNV!TN&~
z#*1s0&_DrZouGz>B&RZ4ij7La66NSWX$^fH2QBWWC!fD0#I#JGEf`4HU;ozWC%^x>
ziO932W>E<SF!dc?ttuNgM}ns_KjCs6N+jO+<GXoqTSu2qzvscKbHC4;Zu%^)eA6?$
zr_nhTeRi;t7v!$HMgmod-uN@Oy?ymh6wjjxF)@)j&$<)TFhQ@!mXY<Q>u&81>5KuL
z5G)7YT5JCoR?k8*#b{b3=QTr%)D$`ZHOfcxpiJ2kP-ROgwm-Wvuz$ON$@0<{cHi6x
z4GX1`;6SJb1_y=g_Wb?pAp2u>IlY{T<Vo(6-+w~Cv|_-lul2im6!Vcbou@;7T$?P7
zY}9RU)gJTO%(Es)8c4?{m+><m{?^1Uax^-NC@--i*=HW1a=2KjIfe;7PozsIA9*bb
z4bcNE%aLmlPp#dadeP(mN_K*$S`38P#^b`Cq)4M_G1^)_;5DHJ4MD&)NO?&-Y`;O(
z-=W@tuy}%~fN?S6cs$iWoqXi_o(E>O;zaGc+22Gu0VZQm@FUz$!SQd>LU8z1AP%cR
zAlX8vFG~9=07HuS=n;(m5!_6}A&8G}4}=#IYs8tr0zD3UR@QR6a7!`ODu#3gZ{J9{
z$uj--24<re_%tMhm$*NvqIf4&qcZ5QYis2r_~fgPFGjlTm~|35FIG5(@&Yxe#fCXm
zB4qV5O_&sTpgAlPW-`Sj#&_y%^$B(hVF7OC3Kgh+>AZr0pwK(qFZH+}v8f0HK<iFV
zx!g1hju`^a(?<zsdS>zr!9zGI+p))&kb!{r*kn3sioUh_+8KAR^sQqICZav&Z+%~e
zID0`}fi~Dj#(R<E<p1kGZto$7(Sm}KC(3PS;Dye|x7Nr=7WoOSxmIm|J~C1g9Urep
z_ZTA^@5B%*!!wGIO{V$i0XkPr^RO(K*U}GUrw~}%D(@ukYSs!#I$IBZAxES_mJ}Do
z>vFu04zFpwzv4S9OYm!YlW;>4O}*1-sn%>EBycuJ{(e{uf0nUkjp-b1@y-{Va#!&$
z6=kZCk-wt1zhUw9x)^t4L+Iwun}PO~QhQON#MOKD*8_c&7ZV_C_X-F`uQD2ZTX|q0
zei~V{q;{@*YtG#IxkH2bFk1dO4(@CWWk*ESjOCh2r{gM&yxk@LnBjv8sGYDc`YfAO
z#h2C8Sv?p`QE01f8d&J_CyupA&C<m~$IAB!v6`N~<qCg7yPywgo~#I=Y1BQ_2`EOP
z-6-PIm`l!>==G%{;8G;4Yvn7`zw6SO9cclmySko$jQ7Kz#97iaBFM%+)ZaBKyBCn8
ze~$Ys#qEKWmdgEwwwbxvDbLTJ0#7GrWoGK0T>hS8C^zpvGR*_~fESbC%qHrvPb~%r
z3!8Z>J@5uhDw#T!h(0LjJhzw8vEZnY6%^o5im?^Z#-j=a!g8{+IcgR<!w;h>tYrvv
z_5b<%9REd$NC1Z@-`1n^dup9c4G)+v4>pF<s;WYp)*_a0a}gLjPKqy&RE{VA{rgvU
zGglYAc|3D^suZ{XOI-_%)JV+~9Wr%B-jDcXXP#&7K_j|WD8&N{A|O{jJv($RmhfHb
z2pOI8Yx+_Na4KD{=*okO`Q+q2VKeB4f6c3FVPPR&-1*Z#^y)Wwu<c83@acd4+#v_b
z{xAA1RbP+!k1j>8=>jcnu!)+v&0@8dmcSN7Xy{2q{jc^iHr+QgU8F|z%y;H3vVsnt
z%V|yT;pD}a!#93?Hx&CF+~s$(bUU%kd2^7iM9d3FCbHs>b8+(#(|Kb^XmVbB4T=(f
zYYO>J_eSW~v)LMKyo?TnCJ#$EqmdMA?&*SjMo%SAp7Jg_h2>>X1tp;i!8e4Y9;b;O
z=a@zu#MnunimtrM^xDcNc>HxaQLL1&&_t|WER92<bdI*b!-_<&c__VNx2mz0ck|Cf
z7<?5#!thBF6TV%ej!~@zG4*46y^K)?>ea|VRTTx9-;}N&J{c-z@S+Nf3zR>}=HOO_
zmtrcG!v9vycwtMDg6~$YHPyD+eaII3R<y{D4rUWvkrYOGO_&80nxsU^m<7?#epGwI
z%weObAg9pHG*V^=*N2p5eX0GgGAf31DC{HPL0RSiioT8EqD&aqcQ_(?TIpWNu@zW`
z?Ti>>HTJ?BEj{~GRCYUstXF?vF(LFt;5N)RTOa+9u{ZK^K0vv+@F|-@#BWf#u=MP-
zX!bmBwk5Cge9wM=-|vxZfkPrM{E_Opr|wyWNug`5z7OpuEwugz7dPbO9G4dcZ4T5+
z39+~$4afBc8l|L)T({C$<xfsdqDCmMAfiF+B@n^HpzYGnjB#`sI3{PIb>#D_(T613
zvRU}XzZS~O{rAZBthgY2_GU~Kq*mN)@b@$s+Hah7{)W2hc!)3l^1Zmyx=qd0uF+|4
z$JV#EB*B$9ZdY?UFQgJ~f<}rr7JypJGH~!yqiLq|+2tOHjk}(@e57`CHC1hyHBA)a
z<Fq<3<&DTRb1#PGczNa{yW+l&w_iWO4+c6z*1C60-?0|#_;p@~Hcf1uUNAcNggi)v
z#2Vv7xfPx!L5$%jbM4L_i%MsI8XmDLH2QDuJ>O(7<2U1fZFT1-k;WTWb>JyopSavQ
z96kG920mP*;M2H*6f}JY7tGUg$d>GrfVx6ArRFp}KbcmuD@k4d*s5^!yXxr`T=V#`
z)3|)TLfvt*b4wVaU^>ujCY;5#k<qt>HUPJDwq4BD=FQ3Fa`e^J`Pj!Mm`aX*74jXj
z?K(kz5bXrOuIALPE?Ad=>zkgQUkxM5I|_S?N?}aSYlqG9OnXi9)O%`C#|Cv*XV<5w
z@1dXXFGrphadw7z((ByK%naMd7>nm;W+se5jXs&s7b}Nsx1TQUoDg;1VC$||I)Qf6
zqYo%y;RnzPbDD2*EL}gsmGj?cC3ef8N%1gzf?BY*#?XJs*DdG3Y-KU1u=k+8=<ZEA
zZvy^J`(0UfwzRZ1bT-|B>}ple)a}=HNVL9y|6b!N{dVWpcar69=ERKOB4T##7rLub
z?Xo{BA8TWy7O^vrod5K(;#&-l)a%elV02GlEs<w^urHfRVwk<Swm@`=Q}FARMlL1M
z#ZLz-s;mC^H3py^i*0Y|P6?y>q!dO(S5pqf4z(Y71kNau;@cZ<OS*kyY0MjlF%^IK
zerV^$L~{5mBO6Xg!<B@AeiR`Zs5o=*PQ-|dMB^b1%=H;NgZp>H$7y^#0!boKbQURw
zR_yhFZ`j(g(dXGT(?N<PnArf6CuW${yd^}q2KMBDZ4W&ETO6vyrs4W@tzVCB#f<25
zo7Vyq<)UH3U*jvp9=>hFWxwF(`aHA4obZ*2#%oksBP{f!TtixB;y?wtIz}#W1+O@4
zAT{LYY?Pkz|KDUD^F&AX=!m>{XMgWczo85jGV?GT<a4uz@BR7yH2YoAiAMgL3f-I?
zEI{fkxvb)aL5|^mMFu+~^(SQY-^iwA@(o{$DHuMe67BoW)Bj#QNhk;!R<!Z2Yb)rH
z;X`&GS?ZzhY$bWhF~9zs4MZzE$8~GuSQ*68aK(NlMfR0Vdm8({6lb4XnuB@lKMB9m
zY(ZQcbnm1`xW@R3KlCr}@ZVS<Oc0=WkN_~HI}26t8LfHLm<yw_OuhKA1JNaEgwEd}
zR^%`?O6>6Hfa4Ep{7yuzgc!T`{ddXTQg|Y;9O(%-Eu@NXBiFq&K2ozeKJuk-7HQK7
z?b;}l4=D}42PH%^=6xyBbeQRm0qyIfw$>z2feTE&HxwGmY<elKpLM2Dik=HU1N=u>
z2|8{EB%&5dZ@b7V!Q9%zDQsllqvGbseGTWp=Pixq!dS$vp<Co-qzLt1U7g<-yE5aN
zsEwqYME^AyM^XZetZ&U^=cj|~dWYb`Y+hbkT3XWid0q5w=hLRh6^8}R40>q-&6e=S
zgyCgs7KN}7x0?GDt`+78TdU?^U>%X1uaq&4uu)>;w}ly!==?4XIp{hX@u^jTg@zv<
zl=W)<3~OC9q4y?Nhf2nMrO8ESk<WaIG0s>2x>o77)@)xb`>05n`Rw1-mKpufVrv-D
z+D9d_N;fyVi$4bK7In*<syAnA_ausM54yeO>+FH!shlLNK}wF1YUtGlfGOx$Gn)P&
zuhtfDa@E4ym)&uh1GGzazUoc02y}LSAbIk4Bzk3$S~b_k>Sy7ieaq>bcPGtDU43Q6
z$nE;6l!DG<o6aIv7Cb7>ClQy|)GV2Iv2M|$-c3sE1HB^l5h44f$3M2vC}QtJkM}ZL
z&*Jd%h1SB0#}$WMSSSmI09?4TCWh@Bj2N!}^D;n~GT(cq?rOX4hKz^9UR7bNFtqKf
z7(MHd8$CZ5cAyNF0GB6=oo6p*KoR<~;&0`ns*#&55Lx6<nB{$ZJXBP7^VbjHsVG}q
zx*0ybZMO2%akTnyq=^~h%v<^4UpdK%CL=BP+V$0Kr$Srj_Dce;3zTmR182mhJ}BAj
z!s0ofbej2lpZe|!{$_@lm&($?N4c=(UX+P}I)oL4Z}<jf5#AULUbziXnqu;94;Ud;
zHq#w?Sr)P$mSD*9@3Lgy?{U2+G=Q$qr)d<3NJ1R&L?xYtXtLp+q!R=Gx}LfQH3rg9
zHa=_&Zk8B>zQ_aapVw9WY*cuxt=qbgKJhfScPh{`OPf&&l8lQNAuF!rcdN>fLCXD8
zjHdiFUGv`D(^ZJKLqe30oKGKZ{+pzUsTpPZHBIQ=qUKMv%LU8r;}Q@!PM-u^50~)k
zcL~+WZ|lGDs%R-S)z!bh3P#{2gAS(oosbl3{FHk+;Pc;-MJeRk41ABDf(v@ea)6dS
zll*tq*~#5@+5GTI57d1!69W_+RXU|d9~O?asJibJkZ;l%`1Jhy_dlQfz0u>r7+L-I
zcYD_o_I&fxMOloX{5`@la#C>~2x4MX;4@!0&_Bi2i0I1@V;JnkV1}Id-)BAVc^={F
z6>#`*41oQn#8pV@>qw_J$VwrEQF?iLMJ$<R*#mmclwK8$!!P(UMf<fY55seNX!&bq
zb76tMJX#mYb+6ef^hhLiNYNl{e4w16tIXDa#dw286QzNXs*~2KM&d!D-ulK)A@0!N
z_XMyNgxOuYFz*~JoJ$0z7r%-+H27O2vsowu)Hk%We7=x8h?lr-x4+2MxZSGEz<+Gk
zaJRx_aw}=wYYN}4HKK*hJl+W}-Lddxh1A+o!$|bSWLiVkUr<o!yT8A`TkBV|>$^X*
zGvr?KY_F%k&!|Kpa{C{%CZY3QHx^^$Uiyk|ShVcEWeo`)L`Mk1{T;75^P{|)^-kjn
zFwncIQFjkb8up~o;hX8esAd^WGX#tUj<{M#L-6Mod&FXQSXtFcBw3PQPwMc7s66sy
zCRozw^pmBIY8rEBXpsBX^Kqs#Vm)EUQoKkpMiX|`?_M&$0QN2WD{l{$uW!a*qkTK;
zuyadGnzAwzchw@+h{2XFauj<^>0A$(R4IqtXttX^Ywno#YEtgI-1e@E27@ByRj&@P
z_E*{%mrKyf3IMXt#S7IU_fLw9b{qhA_w%#*CeZt}*?Bd$Q`c!H$}l17Y4pfNxcy}1
z811YpD9F+v(~z-twIbiCpw{WXGrTL&elalbXE*M-dKovFEbYrY_(^n78W>I0QRu?+
zCaB0#Z!ryhQsS?Mnq4jHox)<%`axLVNu~YG-&Vg)VC|&nrRtZ*7`f^DpL1yzjuM>C
zD=w?&+gENJa+#%NnqYtW+E2hOaG`dirPHn9ed*V6dV66P0?VjQ$F-vtzl}0&_v9Iw
z4ko!n3A6!O4)cMjDz^|IjLEe64W9wOKEsLb0KN$7Ls17?A3+@S1q*E)q7*`pNAud`
zCrCY_awjukc1_dEl79b+Noy1b;o?uk3m|d#U|(Ts6i3(k*CfTxC4Ja$U07f6`t_AY
z%<(IeoX6k6k7Hyc8iUZa)cf#%MrvP_Qd^%38IV*GA$6+bgbX-=JUtTAnKUQQ8q$G|
z&pL2tk`N&YhUfZ>Ba_~k_S(-*k-!gj<hys3GDTa4N<_A}HNMv#LcOvR3NsZ5k!*tw
z4TkV|;(rdXQe@1^QKv=qwP@5o6X}Sfi!erNvHJxfdFaKM6s1Yi`Q6i4{GhfCa~YIy
z|Mxv#>c-2S-b#1Cosrpb+DjJ4Ll-ML_;&a$9BV@4D!{9SPS+}6{!SCb<yD?;I9_jR
z<vp46h>f*xO@m2i?Z0~(Eymb^Jzrn&yg6EGw<%Dfj%a5mq>Yz;f3fJ>G@}r7@f3`n
zm9)(*DJjgIWC=YzaogB{x-UA;i}=s17fNdsY1XeQE`<a1ltk?sO=4)x$y63v+PBzo
zrXsK(2IkNZ25K_n(>SCiOPN-qs3nFk!4vs(_-09?-WZ<?uK%%zR=iw_UJT+$&`hW)
z%H0*zvPqZHTI+x$(>3lQL1#(gJnVjW8g_K@XNa(^@WhWL)4XZDVc+}pa_37S*o1$E
z?HG`T0}8HtH#axi-6=i!mbXjGx2?n~%L`8;FD5o+80@bPDvYWvG1F$uDDrO&;C&az
z>L;5&R8+Kx<{hlHkN9&Y)OXzHSoqeYX&&VI?lwlIP;;t;?Kw3lNGZrY_MT)L0h8@E
zeI30c?YZ%Wx%x_N{u^tHg;>boWD&#Gq79mWkO$|Vvy6$S=DBNH`pHP^8FpQvO>rxh
zewQyM`2_`Hk&PDO!Ad?Undpiarxr7W4PR}^kaVU8(aw$gAJZaGcX^pWX=C_;w>hi(
zNKuebOTan;J$r@Yj9SY9u4k67R;b+sSTc|2EXMq>h>StT4R>&Jiin8F`z>Fi>;Z*a
zLEzB6f&2(;>DI`(f%M?R-+Oc2z`^nI*>qC}zboUJUFBkaQPGKVld&im(C&k&>o}Wz
z)OmZ;8jv3qd^CN}F6Ml0su~sHD*1az{5$yaXkCF_jo-6}<0@nZH!=JLJ2nVc5s|R$
zY^r4-{8UCl^}Jo})`Z8IU<My3OEiDMP_*R%qv+cnjq=ImCy1jd(DI})hkU=$@N<f|
z=^X(!Qq-<A^0$Fo;^m8j4UEj@2$fE&iC3P@+yW@*zFr&bJKbMYI$g~iOnG^o;&<)o
z?P>sdY#L=+@r_B3_hW~6i`|d(?R8=PC^kV*$oU}&vV0`{)wJwqHa<}GDcK-hq<O8T
zE-Vk~Qbkz6$ZjAu^B-whqO>eg4W6c$y0DB7vRh%zki?N@jOn8m>m4EmG~+81FQim#
zJgF=h4f`;Y#!J(jBUw}y%}4+N?PQ?2ItlMrSS86EkMw}?Rv}(N<&QY^H}CndMaCJ@
zK2SkDD4KvNU#o&Y(HbaKPb>Z5h4mS$)Wh!yZ23t7C0=h&xL$l~Uzhz>dNQDOab>cH
zVBKO%{KqI>3iLHPp#H0=C)<Ws2AtK%Q{jM|u+!Wj>C=zJc_GW90D;UB2KX!TdjIfm
z<hBML!r1yoi)z$>woktY+g?+oo=1h0qe5m+b0NcXLYxAk%NT$-;>xG;B?p-9F7)52
zsNJ!Vg}Yim9TnEXU<bfNE^`nJDaN=R&w#~;gF|v^%i87EBW_;bdN2V~*MyC=V7fs2
zjt8lRzmT)^kZ)A57)45AcS1_FhB1xZ&ok|nh&4WamvUS^)f;tl3jLHdUo1k~D;D2h
zJ!xDIPO{zuAmE~#g_7ARMi2M0<Z(1kuVb|}NkmX&cg-e*GxfcUiWLth%|<}q^ZFe2
zet8)i2`^F-qE4WR8Lz@26|%`}4#?D*xU~`p)jyKJ*9WH-1_nu74S#{z<Ait@x9)F$
ztvEt{7RnI4L^@<aV$BC62n-AirXnyw9}_r<IaYv@^R7xxU*&_SoB$4j-Uml1)AIYb
zVYmNEv=?FQ+QYvl-AiuGTc&?3Eb02B-}lTvxxC)p^$}V(d!!P*_dN*_Qf`j9H;4|P
zhgaO>fS4MSjlCvf9NivSPhqK}7@m`uu>lQ46ycnDj4wa9+WlhUdI{1%;c5}AQsMN7
zWC$?vW=&s>=zj@1<F^&sY{?LqcG>hFxwM%INS8EvBZu^CM2i+p9>I62F1v;px6wth
zvvgm;KiJbZ<oMf2``>|*o7s|);bE_$BE^n{*&kcMClXI$z$&JsIIqYPORf4PWK%w7
zr?b5^aP#l+&b{J@z{tqUvvEiEN(@FU1)zKR>Gr@KIOz%+q5jfNts1`ntACAkXD1=%
zbUbn&>>GQ@ax8b8k9Bov26<h)yghsAJ(RY1y=&QV2JZwtwfiSkYkkm4Ed$5hy^V~u
zFYTe526b(JyE!`;sJdYfMWzP&8(uB~XOfS{DF+xy`D-jh?>bi;!t}E?&lF!I0c-T+
z>$0!S&(>g<d&?cz&kgeOO{2`{>Ps76&p_|W-Fsg9ZGaDqt+}>4iAm>TJLPn#H7YV9
z5KNj>`y5q1zo-8<jWPlk-eU2ZiHumL>Q!#QTk~?R25l-mPxn_AA0|%;>3*nsj2KPY
z%$;;wd3)Ck(X->`(8Z6Zc{F5^fRv9I!sTnDHeaIuJ0z_pG~FZ1TwunS-4e<#R5p#k
zH)P=w5P`M*iWIlf6surNH}N1qdo<qnppC_=SfJC&4AZc2rm!Mc=LS-`c&TvyOfO~1
zMITf|>z1M!NfGLe(D#$crd=drkE?m_vbug97d7HDv*wX=?26sf{j|pkD116tnC!uU
zDjBdc!3ku_a%`Iqn0^zafR&|5_5AoM@)qP^%M9k!cC2M@Pa#N+`Kz#;x&+s6IlKc3
zGZYssesEeE$W_tAjeO^L_q{!6^W@u2J}12aA%pfObJ6eS8_k<uFY+wrSW2_VYEhGn
zq@!&aucd}_&t4`fCrp<E_Z574q&LEAC1iYLMHswWi!glTNd}J!qiQC$h*gghG+Xbc
z%cMCaQiz#8MDpUx(=2{;eaoC=!osinQq1z2hvoy8{U33{%72HF*#a-<G2y#6#cRzJ
zwgOONcrLa_vX2{5_$5BbS#r=+Gga!<ZNLC=JvQ6)o!s7^KTlesfFzlYkN#(W&dmk|
zGeW-<4O)r|p;}Z&cH%kf?b7?X6Fl>!xZ_K|GF`vqv+}0Qmwk9-J;GFh{B=9cwDydR
z$ppTx8CCpau#?W20^N1<_TP8>dIC0G^KH4A(?c~pb)4PR)uAbR)MFYcP)$8Oi2m!+
z@VT>HEEN`WrBykx1yB<ezK}#dDp9o*wrE{kT3Ql!|MRQ=?&ipo+}zySX_mwY9AZS`
z>7PF_x9OZN18(CyWnE#UuGx@Z{~r9OE#NZ>+4LBXi42PQU@MdwE<3kRWq^Jzb@H0S
zAY6dK9Ul^%)9bkT>#w8NvA-l+dfycMS&Z-Ju|kjgm2R6G$q9gJSaWyUbe_Yo*`p$R
zRVgL5<Kmc&B+dG$(K%+S<b<1_4H7ZFd!udRUSW|Ue}Yd>!|r9Ud0}l%E%u1es|#K6
zge>%MrO0{Yo0z7Fxuqw$`GL(7Ub}w2J|!vu_ruJ=90`@vg`msJ$Zicl^0y@!fBD09
z3r>oZc_pQc9Z;M0Y-iz|oZ?UEvKZ=lMt4pIo3(9WlaA%!`m~0i0a`n9XL?tsy`{b3
z=!M$t3&w!**{efzH6t^`^}B|1E`Kjv?%;`}TWG|GdEd*JuTXiCxi>MeZ0SW650%i9
z)fhqBW!2v&$#;j+=1(m`L#dm0rdfv~sV)4LzR1Y6H-84cg#7LC6km|+r|U6-T?lp_
z_v^{GDir{Kbvtduo@+Ga^fC*-%|bYv+}YIR*7ITME!W&NGe#7*hF@&#GUz{#F)l;i
zHDId8+w7s!Dy5h80uP6TR0q`9vp=o$;e}FY3gdVPoLQD}x8gywhAfV;Ial<Ikcfld
zAzd?T&IOnwG0Ko7J9^xA3gu<Qlp*P;JcQ7Rbou&K>96O085?6L<L6ncFdA<pH=2z!
zU>+vcskAcqdD?sqq>`4#DMp5<jmiO}hpWrCH49#dNr1A7_26nfvaf)Z_5JLUMP0jE
z?PErwqFRynd`QI(tjAlO2h9%g54ioKW)LFXP~hqP4{MSYwr&4bEh_;5xn2|Li(z7b
z2Q7^HpkeGZS)WZ)+RzYGsC3zS8E>pZqQ^$PmWJy}B7YQNCzVN;s!e{5))Uqilgz%u
z@b_Jx?8Cc4W(mTBrao_{g{cXw@ioIu^8)C+kg49W<G!60#sb27R5p(Hxik}A<?F#k
ztpLrqh>m@wT8-c*V9W`89p0|+>~<v~tzW-tD&ZLA)E~sgc-FDcFgDKD4-K&$F?iF%
zwBBswTla~?R2UT*0^lq6^O__YJJ(3*3DtX`rafp$KpV*Uz?4Xq^h#=g@*ClXSJXds
zTRC3P#|nXzOXTxHo^N2Pkxe5(|07=tAo0!B-|i!F8s~VaV)B2QB;Pnl|D79DFz_Hg
z_V7TMa0<rRJrz4MwT(8Bb=Tu{*#oG~HPrgr`<b2UlelG{W5$hbgTaOb|5xv3?ZU$N
zLR>%GU$DZ^_YASxkT<+oT#AKE?i_<fIJ^ac8EG+!lYVIu*RzK`y*=K)nXktO`eHXN
zDb;ZdEPRPl8mT%?=@!$a^$4>@qBrX=Quf0vsqR$=Fnx(P&qwn;jOJSN=xL8<X1+sH
zNmn?IkXL2uQ56(s5fw9(;80y9)L?TjP$Fh&{umU-Q4_FTvB8`HpIA*O;51!nyJ=EW
zj(HluM)h_)Gbm_A^@{~2+n`qyXtxywy)OS^Dpi29S9XAv(7i5=Fi+3($mR3Bdfo=A
zig)7>*ZcRI8f0x$pS7uiV;t-DsP!*9m)<t>8UAya;guw91(rHBtHmF))S2eos?DcA
z!Qox~9xT!%>gN1jM8eA5dnI6U!)m-wZS8acr9sQV3)iNzmUWpGe-{kE>#U-p@K_|<
zE^1@7z|a29E)8*wT)dF;M^maqo2b=^ktLf}5fdV=_h=lMQHx>T4r&QFbrWQRJH7or
z*-mU~WBOjb{)<ZL3GrH5L^RV*-h)ATOOAJ*2w{dNBd-sn?zP|k(}g*?#njqLP^_l%
z;PaPGd_790OY_G()bmOP5}K_7<{U4UZyNxChrvhSE#KnOT8FH6VBycd=J~Zmp$f5I
z8=}s#9*ydp*GpW80VW=}7*<L>Z_*TR!{9MS%$Z`8O#Nj--`9%&;$rz7#Q52KEDo;w
zBfb)Hc^Qy8i_WZnV=>?wqT?f>U~={#WhyoI@O&*rIbDk9Y&`~Q_^qtnNM>|QbfFbO
zcn_Jae$4t>8gVEFcnF#Fy97A}1<8Pkg`XtT25XgM6zacBu<CGer%t>N+26CRfb*|H
zr!@Tq-V44F<zc+~x~>`2Hz`7))3&@rD)HZH>g$rRG!97cJ2?8y4!JYRm8Ng@`tlTC
zBDV-(r5fWj3P$1pM6YRa%=Z@fv!%N5a-~gs=_sYUNc&_nr#JpQt-^kf7D(0E`p%U1
z60C}ai};*ZYz*FZlcP1oGAf{~+MbCH$|?8Y#e@~<6a*5C&$q0+^8)4HXT1n$F6n(v
zr?=Mpi_wTf>6=21DU0(2J#4+kn2&=Cmk<DF8a5OoVCb^~1Pv(0wA`!4StwNlDXk$7
z(^gX9p6NQCyg^bb{ykdls!S9SyHuOXX*!D*r+;P}26b~dl*83&8XqSb<6yCp7B(;F
zN#H14y?@ev1k(6h3pV?`ZgeADNH_J$qeQ}zRAWHYneTL%*D={x-GGFfkp!U``)f&C
zlaM<eu^a>#OtJ8Cj-VYMDNBX?I#?0S49FK3<0QXw2y%8-X6X%>EG%}6f7yk{5_ER(
zw<_pS@=7iZ6lHMt?%m@HEt)Kd5QG<5>*sT`vwXhTDpFl+uW1k@Qze_tRi@5tZ1lDK
zwf>u)2eOQE707q99-$vpnd#XHli&&u%|^o9HaAIu3jH^M;-5s1<6~){7ryOW!s6+=
z-D&+*PE{^qwY8Ymj?Rvj#VyKJ*$Q(K@r#wxaeoP0WlN!%clkr`F&q2HpZBi9mSP$v
zd>#v<!PYf`9PUhC>6x#3dIcs+oZ7Y6+AJ@01aHlOIm!HNp{JptEZ7TwNm$IP7K{u$
z1cAjJYEf>_?Bk0SnGscYWLkxud!BTJTg;e94zG`nCy_d$D51T5(XBmT=L9p`a()Q&
zz0b;8?Ok#{S$PtAy1@uyR-=wLiWuaf?ZB3oFhBQOH8vd+Q&9_0?MR&P1K%~?)Qc2<
zQ`BrLt<UJGuTQ+jFw9EHLXm{bkqK=I$n-gPOmI>st%cXlYzNbbu8WSWB{_xzK4vVk
zjM8~1VFq*nQ^swt8VGXz&(;I`c`4#(J}Wm+pXA-3_6#Ebtqy#>Ii6h5uu)@>S>I&p
z$ql(g=fd0~^l>y=TH;_Wpi9(Mk0wnDv7W_1Fvwh94?%c)KOK{Z)j+>}!@>lN;d=7Q
zvmi8$sf>;F#R(5fELC*O5eIA-TH367^Mm-<UTFz(9s=6J&G7rfO!!tWTB<czAN<EV
zHqC|0;}BV(YIrTpm}`!>tJgS`rJ<Q+L0@VmDj>X-T&9yP@MUHT`2`%#HcYV6+)*@&
z+M6WnU2%|<e$H6v)GjHf^vnYYh*;Qp-1b+a^}!P((#NLK+)R~;zXNAy+_XFy&tjOW
z`nO2mfx&YzAzs_=UC4*Jev9Gl=Av$%jM~n50ti=YmO)X>%s_7-Z4G~HA&Gj2WK-1N
zGiJ@i={ZC-YM9U$lO<&eQUxXvo?@6O_#iCfEnJkVTp&>jC*(*Z=Lr067JW<_l<S|n
ze_m_Jq^n=ONV~vLE7-i&tbsPK0`*XDA3H&52`fvbU9(W84<8M2sD(<yoxID`{`fqt
zq0=!gFs-txHFtUQ=g*$uL!%+Xk@N~FaI(&yZLI~lgbEW+BlvQ8DMC0bno5Jrc%4N_
z@G&9TuIH`jkWsMP58MM;+H93GJ9tS^c4qluC3w+2+NCUVG^=EmA_=LJ(jbxg*|SgI
zL)f_4W<pdX4zATnR=yPfq;f`*6GH;qi)s9m{>!Xyd08BU6Iw3}QC!?TPLTk%?is9i
z1~J=+Su!)=et-zzNgKa*tj_)HoE*+2`!p0s-9kZOvCa!efT>vNbm7L%z}pPUP=nR@
zsgUt;nhwXjmcPn*HO9Hi8T=TBbS&nLh4P!d6M`TrL`!9b)4RcOz_zh}lZ9wD*^o&i
z2#7f()k|jfU5*~DF()~__fnkG9?=oFHa&11;&i3Qj1I|9`zC4r#iDL%!|k0}l^MEV
zD9z*@wtfYGn!>$7Xp)7PLSGJ5*gU2ssrPzu@p?!yz1-3+T7m5tHGo+BEbZ30nD6n|
zxAL}vvpr3{IO-xPZ0Rx}f6H*rcy2Ml_XqX2@uGxbjZL4b$!fLR^@1k9FvXSi>Y0+5
zy`%#9eKPRRj34vDKHFTGo8F5Fiohw}aIs0g5CpO_Vh%S|m**Cj7r^4ZWUp~9Qt5KO
zKxv*;sr&nRb?37jg<vT6^9d+R^gf2#h)ro#T95_Tpqt${+YK}N8hdajj}M913!hPI
zGkI^WG*;}Lw%^P*2#JMwICq)or7LC7Rmq_n$2{=)EByZ`DrcAu6v8X$=tWYK6_p3*
zfeK~ljB3?bDY9RbN{KlAx%{8tB+D#4R9OH#^}(3e%$hjOaISg(HO5EAs8}^*@1)(2
zxk9J=73<3h2(2{AlpA?J+ccd83=hwHlNI3oQ2fSZpt(;5v`%O?Q;d4_!)83Hw28UZ
zMU|ANc;xOssHUO(X_Xg0*f%aL!X~K&92D+2(n4c33}y78w8bep%;`iVR^O2<?od=n
z)z6G-41JcdI;4CoP=KVDso{rkUkDjh-E`vGD72o{hCRj1go=b#+UdcEZYzX+^LLG^
zPcE=jb7Oz=Y}({>go2SNW$CBz2&ZYf-3V;~I(Q=-^}yrDcE2R=iF+a)pE?@=F&C~&
zE<+ltyN+XzZ5VZjs-6wUofm&fHeG+l!6Td}ppVa3BlWByJ&W!ulQTqbk4!#a3KvMM
z@{7s53+n8UhAn+Nj;NGQ(+y~`3to&z`(n#&c2&I`$N0I7*E~FGEEu2QXVJ|t4KhhE
zws{m<8@%BIgV?fJhOO$YYAk_);?0&?)(>|1>12e&J|7<9U;(zeW!#g}MsZe}Ri;EV
zcq9lP<4tQpZrlIr*i*AU1!9X`gee~EN#6N+kG%Wvy@FiJvTu4H_a;6q^Br#??i0b%
z9z(Pg&qeJ^lET9U3W*7%l#Xb4MnQ$q+pNM(VdXIbbQW?;nY)*14H*QKLJpvk+MJ6i
zi8SQoN2rLR(U&bpW8G#9DFXi!69nnPHL9le-dW$5Q!ya1&MkQDsn`h%+gRB81})4j
z6lE}&m;ftq7+cb!ugw7cDBVYOQP&>2%Y?8c<=>wxV0V2r{MGtDNwQ_&RG3dc1;|s)
zcQ5%~I$$Q2o&y!d^`4@-RYp1{vKd24`?e2VEk3{ob?@CtdHIFd8kYMZr}N+Wh%GFn
z=I7m%JNyn9IikV!z!vj=1R`e~<)nRUL%Y`~EM*cD9mwupn87-cZug_JJ+OMpTP8>4
z<{w!5ehJ?izf@A1m-kx~LV2tAG)P2U427vR2l*Xs0>BDZR?g12HR5_tzT+HsX{ma#
z3A%62nm}1;>G<!&0JK8%DEVDV;O64gLY@jr-BucU-QAjDSu5|BpefgXaOUjpyKGmG
zry9IIK9r+&>s7}pCeC*I`vm;;*N08Ft94%zULU*&aH~$wYZYOYIU3xJS<QJ)oN$P_
z-0`yqI+mE-ncb@|7wZ`#9pMRqSUVzin<5v}m%0_cbx)w#dv$()un`%SOzFJjKmC2P
zgi(X>M+l6gXkL;1&)R!}*7;RnFKxWY7dSk44b>+@n+nSi<mTrMW|8*Q`^P`>l9e0=
zm5t`K=s#N%l`iFkN3?Oq`S0>9aCb9$-<klV+0+sgk?l%hdw6B*fu6o@8kO1c6<9oY
zE{y=SF1Kkx0vNhXQI_>Ze3OE#cwahuM$p=ugx{6~g3m&TCKT}MEgxZ~Vbbz^|1|af
zBm$vmXf9%Cmhft`ft2nABW1rI1z3{fSOyeaZH1W)*gZebn+YLi0@z3SQ%DDpM1U*3
zmHK3H-yp+?vt7johniF-dh5i^eyiE;sn^49kA>aJg+n4ox{Teq>!og`)}nShX$x;^
z#a5e&gizVp!WSu0a--0rmVUBrS`U;~f@rqB+eN!cnEc+p>yXz&{}cUoJK!|IBbaVS
z7Ed;t$=hTgSeYvGsVq|nHA3lH0}!DAowuOM-Kp#+%xGyNS6AgoPGFObq%4c<8}MfP
zS0W9r*%aY67hv0W2&IKwy>9cttTpSiQ4^KzTy$`r*{zJOnqm}J4q_pKU&%C3u@BPx
zi%Lza$o6PxY*U@eqCg*HULLcz2|SOl|IeQ&U&F>}VSY4=<%gwzPkGsZ5Tb5IZobn;
zWZxhZU}@BtJ)N(sARr0RgDP(etCuF4;I1{xY%viUuUxEKOu4i6cK4~fKPpJAToQWg
zb9g%;12u&1aOp9Jcj37vFBbQ|okF<gnZRB1z#f}twkur78)GU=(JcuY05pseZ#?he
zJ5!jv#9hm>3T{C&X#ogV`S^1gJ7JXYl;U}Fh>9Fe)pikf{-f$t;)&hz0ytN=K56P4
zA_QYzgh<hgc1;tz{GQ1q0c_MMU?%Eif^WW#rX+W>M#wHMT}v2kK6Y;l#+H)cFH4fr
ze)y4r>O^-%qptIw_8lL(1n{_ES!4|fn~UrYh}c+_!Q6ZVq-RSRMUqM_gk;=y*1JmQ
zKOu%6J&cHFB#$H1lO|LbWAQeZI>5^6i$=D;QiwC34p6Q#uQv%8Oez}wRxeX|vq7x(
z!#5202WpPnZr4DPP+g__hT73=PWcbL?z-sknM3jzp|c~<yj`^4$!rkl7vksjxn-HO
z!cgrruOt#U9j9E)61f{RSSe2LJu&g3Z(wMPwaJ210w>gVyfg)zUIKIi#syYaIk$gr
z{fkz&6Luq>{axgIlJ6@@L0td%YACEOlxUXN(r@M+HM?rSA%=4~UnL^Kr;bxU{`<0b
z@1msG<H}bNM#jvQ;YZ)x0-RQcyRlVi`Hvn-vr%3E(coLwB%j;3&gzXv(vBqV6W@IV
zV-z*y%c3~mhpbTBW%B0<lHaAL_Ng21@A@{GS7nP#MWH&InsQw8uuOex;09;Hkk|6&
zR%Hz=8u>Yu&+bsi26}6_kx{1UtZ8BwIR2eR6j`~SPVe_2R4>oWcET*k%D;Z&)ssde
zwY>ohO*{mKG>eUbdgn1He<M{aR79GSY3Bs!8jsLYxlbc1C<Q-%ecH<%=RiP-mc28s
z-!Ph6k2804JLAIEe?|sPCYzD<d@1BbaPg^ynhw;U>6st~ai3S83R3VovyTWhm$Ni0
zPLNR4KqfWfrdi?)`{Hz=|Jk)ZV9pxo{R-6iQ+W<ts$gWoSFe~+LMxFTQK-O4ac=C!
zKPQ(`&fad5_pV+KLBSN8H&NqHn7Dz@eIEVo%LS2wJcUa0j|le?u)h<g<{`;KMaqIB
z;eiWfTIpk;&eqwxyZiqxiewGI{T<F`0>@gYu^%doD_Z&mHqn=5-6j?|f-Wn7SA$pW
zcVuA@j)30~j^eAL0T=Xjj66k3o}u!H5Q@Bb|DERmVQctd8W)r*<pFo@rb6vbr6*jp
z)qmws(-`sONyw8XV<;3I8XAgQ(*vDtcpr%#x7gAZQlzwleDbG$e$BQ1Y<Lbr{4E}L
zAP~mGv5gY)Y|j3pl%V0ib-<4!IGPR;+zMiD+KY;cBq=sGHl8O178B@Hi{+N+SmAMT
zuC|!f&+NEcyCM5KZw&_|Db`>{Z}=Wg=!Pt0ZScf&un(*C1Ho$7Cb#rG9^Wjgr{L`x
zzxx)>f45AyhkXu{P`&vD;$y~-K;mu7(dMI~C?sv*VSk_f#i~IC5CD^x?<25Rj`BF~
z=?~g0f#&Rh=2-A!V`J0(@t$D!{HbqD_*y880&l_*s9Z$M&qn>2o8J$;ByLZz=$zO-
zPBx*}&tvg=Z(x-)>Qwk%%X6r3_v8#?&~*2})RUwPu_LOlynLu4rKV&pe9`-r*t;>N
zQSIsUj?RFnfeau7R^vX{cZ|6^XYYgQ-R;k}mnAH{?TpXR_BW_j5$c!XMNF03=lASn
z^VOO=TLGFP;40~XK_zOX?s~h9wer&<$IT;97OK!1J}HtVGUbMz`{A7=L*93LGDIyZ
za#IZ)rAvOEO6%op3)vdb72zc<AihqBS+#rz3g8Yqf!hMVKaD!3M`&mFRRX1Uld6Tb
zXm_kk>dbwmMPtl<`c2>Md05QmX=hT*+M~|PJyx}w-~X>I*+b2zyyUcB_A+?5lg{kT
z-WA<B_cW!S>-i%NwTP!39k3^<UH{RkcU?Nt>sN?o!TEL4Owt@CSqvwFy}UkHiM}(f
z_4B1n1{_yz*LH`B#aio*J}$SVfJ)qjmR7$<9u$=Le0s0BH0Vn85O;h~=E%%XgO54r
zdY_AgqJ*D)6glL4^^3p$FO7Bp%B-qq%wd2|-=rM+Tt<rK!5xo(!L9qsh3ADT*YB`@
zKeE2E)BKIsYrmLs8934swjbHi4o>!uqtjhLE~{9UM&+zgO6GM!LQzp9m;(FQln9QY
zh+y3jG5v^Oi!vVNBHWiPw|Fw_&Z|l2@A5pywN{fOIiNVl<o%p}0e3~>$y|~B^>_|P
zYxvgc+(xEl-DK8k;hk6I&<{iw-a|sw9Zfi6P3CV1njX{XwO1x@He0e98R7b|@i~Tp
zaNt35k=4l#-r=yd7A~j<Ettof9mJT(vEHw2@8R<r&fOMPDC107MgeuU<)Y!Ee;wh#
z*gN@O`?Go%a~auT{eC{zVI^la>D2Pnpsg{!kSQ-Ag1Oz3c7{xlO1uB-O!T25sblSg
zZw=LS!S1vJCjHa*MvnBs6{s5Xu*}e^4ln{E%fOjnXa%n=qK-9hK~8nJLDH_|@37TD
zoB2nofh+a8#L2@gSIj2zrUOP{i&ljro9N-(gX4E3ngaB9h}6{yKFsb+MzR}>8JB7r
z<4DSulr(=jKB1s`%}u8x#QjwP)21!tiX4#Cd;6Z0BTzsgvoHm+U@c(vu>T!>J)q0j
zM+&+lRv|BHC|o9eD3+p*k5>+>_{dad@(PTS8@6~d5GGcUM;LjIKTB0pJh<QIK{@yD
z4;u&{@vINXpwmUX<(LU#6jfkKEs+sXoKfEh(eX0kP##cyj+FXJ3&XGcuZ6cq+Mlgj
zVn8T@1}Xn0qI!%X%T}B*^YHgS;FpVPfn9U+;#0YH`QM@%Df?vHwDs`X{-Z1Zu>G4|
zrL|t`71N(fbafgwjfk9SSmTn-S|Nt04wcz)`uT7A`#_hXJDN<Jrus&&{e_{g^)ZY4
z|7I`oRBq4AY}%)$DkdhDuc<rbZA4)|cMIf8BKKxHua?5pj=Dyub=R;{LXWF=dtbbq
z+Ex*&@NJ#gYs@+7eR|)yuql6N@w%O}BPvKSuXK>(Pur6mhCw_n;72c|lk~r0(mY1-
zhVZrNCk9&JvV!j$A~r^%H!|h@I#DRJkVCNa@V&t%J0Kmz4}2kzxcyTgu?H4Td+Z~d
zvn4m@O`A!oVJqAHTe7uspx5MOM#f*|p`hqB+(V=FZ#Vr>uc<BMPpBS^+xq#I*e}Au
zE|-?hU%b3(y*-jX-P`>`#sU211!LiY&#T;3LnAje?B?3nGR;|E$UbK&Ih#Bp7Mwtz
zDEodY?TA!tYt141j>k@xCjx5K&ZbJ*GION9;|UU+hV7Z<*n-~x$0N4|)C4R>DUrZU
zJ1`>TEmtsH&~Vr`waZ0VrDdIsoG|_i<3kV6x4Q|Fm9#`X9@$7h#U1@yw_N^g#O3G+
zg2d@HX*4lX@#K#<!A2(6Z-IcfE(~Wn{G0K#!EqE{pT%dHEE|`z9~iNP?=qwEdQIHi
z8QFL_;>NwCa2Vk8WQ5G>Pbe=3JMRJy2|rDov?htErtIii-m@K-0N`EDe{!(iXQB$|
zT!i%0v<aN_<dX7(at@8wUw^=ao-ulKEA-K$@itM>F^olYJrhcYGlROEt58gtT63-O
zDDd*i!cLsDIe)dt?N_@67J^4I&>x&_Jz-NNuQcW#!3jv6&drTSUzy$N1MBXz{O=#$
z;CB7Q!DSHT6f(9GT7%U#;nU|ygd#s9M1;Dzgk;+5L$|u7KaUNw>=a5NEVWfYTs8E8
zcRH~y3FEWkO>67yDa5O&uv$HmGWzx~9vUi@O~WKdrbH1D@8JkOB=;#J(PBq~@h>7}
z8g(2OK=SLrrP;~_bXV=3D*1JsC(E7uof<q^|FrZJ<3nL&lu;v+pboJja|v<9%xrc3
z{ac-)E4X`7I=hY2?|#-eY_Y%|zF+8!?7TOUo{FN?bgiL=5G=E%k1LPR#s#?{iF({x
zIiXlvam;6AgYfW|IVTK-vU}#?;h}29tH}g_%e+vAppRcylX;!s70othhumGwz+r_*
z56ko7g9(I9h~DG4{q{PhtF_GT-CYBde-md`V;=WAPBwwi>pb_Eb<5kmu+F31?yW<f
zCowxV%Q1T|MsBD5t|gv176d$aj%Z!n<e|}2aCkTdN<dW=RX<M~ZZs4vE^fK}d8)uB
z%Fb1ftk!9uosM|+4l9QK8JFc~^`NVea7Ve^33*>}5mD{RuAjwblvB3=%i$Wqg8^sW
z%V)nf?U`yQK;;*Ma~tvDnezzWG0s~)M$h_eQdbqs{ts9DB2%)NoI5^o-i0k-uPQIt
zU(g<y+uf5Ed24u*p@Y<Z6!qKLy`mS)|GSvJD2du>U9>9+I$jm2tMm2Q<0!Ir7SJBs
z#&#SQaT-+~DG!w+eid$o&IeQ+0hho(%3X(#9SX&+j1$kiEF^5*9A$s}h}y-nBFXq&
z=w>{L3yOS4e}f&-lER{;`PUN49oJ1A2mRfK=?z!At*1iH_r&d1v%crZs2je(*CzQ7
z#{aWOD*5h`=Y1-p3@VO42HXiyqZjc3ZhZ}Iye(m;M!Z!i@L_=<_$!{>5I~WWUY2Wg
z{ks-iOzNc*{{y=$?bZopn5x-p)2|yxhdh%u_vGKr{yT6vUloy<z7rBxU($A!?ZVB)
zM4ZJ1UM&@aZT<Jh6m^AohY8EdV~4F)O~~*Pvv9Nn!n7)`=J=${_wOYA46@naFd|nH
zMG6=~%#aB0IHz=lV17dXQXnt2P~g9(t~UB4bfBvuPwD35(8=9PAlV8B<4+1H*2+3l
ze`O4R^q!B0il4w=@%yEt=?ni+FxzYH$?jt2OFtU~kH)aa(ZWwz6$8ZU<8@_rwd=B-
z*Kh1^aR%KWu?34Y6fB?4g|QMCYA{#1MDC1N*L8-kXjV^rZRo$j0VfE>JaYFnxBmc(
zZ8!w_&|2ZQwO?|jyCz{kp`<{Y?(o>QldP)Ej&8cqgB^UDL<c?P*t%&Yk?rnC2?E|k
zB8^v@LOdAZc`lM9@XKRK2c$mEnu)a%PX)!7rB65Y^_6tkuv6=zufXnklwE*kDj1RR
zNlOSyhg^j|cgY8a!}gY!R6E7^CcZF@xW~G+Sn-8y)l4Ma)6_DNWurPO|FP5ozLi7+
z^p)yKlEmXox|r|rP~cO?>WSE{7d}he_UvKe)oTk|QqnSqpAcg*GtGQV2Nn(wHQE<#
zo>*_4T%-^aPshE|BdMRJ)AW*ioW1^SHbGwzp$5{7hbI79njAi{)v7}~|0?sJX*rLH
zA;wQ|&nc^=D@l0xWhowT6F|&mjgBdd;h5SwnG;nrw)@2;wKW)~JImL@FFUW+fuF|8
z=aVVzrN9AJO*A+n<kG2ndbZ=t${f4^!k<`NLcpx%G{NGwkI|^KgQbLMV$b{x(sksN
z%P85K3;>yV(}G#TSwx1kk8K@E@@O7F_2$xt;n5$b>Q28%XY33mir_LY@HHW+JajYh
z(hZ5P+Sno_pOK|zab+VmwZA0^!;ARzs|;#$X++J1N6fs_4~a;<{2pSiH|zGVzj5A#
ztMWF`b=+>x{~uLn9n@AAt$nmWp;&R3wm_k{I}|HY+_j}had$W^QXGnF(BhIpa4n@c
zgyO}56EwkHzPvNvy)*Z}%+Ab7=A2~hwb!%O^NVD=NuzE5Mf-l*qV4#?ExT#4N(Mm9
z$;{PCQLr_(7;vh+v7zd#8LLqqSEA5A^-A7OPoUU>!iH+vZzI0iAd;a<l@%}R1&l@-
zp3&AQ=LJ>1*!xte3Ab^WIR2&6I=T#7yDDi$7*RtZ^#*Fgl5-`{*p8WfWI`l9$T#3{
zgmS!3J2k!J)qztMkzMS2@G`<19~jui7S5>&2?JSHfW4ouE$?i`G1>4l@fUp1Gs>q;
zNvjaP_VQ}$Y{iRc57`RfDl(y~6169qP2OJJIl9UslJVZ!!;bV3rNb|hHwMqO#s7Ge
zlfx|jVW=kl`g=IfwtrQ+*6&2k_iZ(8ETs&|V{*=(5b~!3VzOKoZJux|VW4@?by@1{
z*H>ApVTS@`++*AYu~=C#PnbtPdr4aFS089#A_MU+6T8mh>SU;vQq(FPG2JiHrJw9i
zU*zd1tOsqWze~}o_A*(0`ho$Cz38nbhx>a0rJ6=)Psf?>_;#~WY~k_CB6KeW4E*}~
zxVLFo@+@vj)RY^?qv-`|hChHlRd9RedkK(j^ORm_$&@imas6NWc_-0$<{br%mz!ao
z#k3-@>OK$8PFU2PNq63K31&X|Ut~2EF?cz|?zT=Xy(MMs$4c$xx*M;3@ImohTXcB@
zBh}2sR3Z+$Bsxc<GCa#$_PfsSU&W9-K0Q^frzyKpBu6kBU6xSaStCIAnMCCriM$?l
zk^b8+vg~WAs9u(WIt&h^6h^2oSs@pHm5^do?@z9$?^<WFbqQYB{h^%KQAwtZl^ma&
zg9`1L_b3WEYCu}+T4zfxxyL9_S7(V1{W<74Bq~Qo7>VZQ5m4kk?<A7-abaJvOVCQY
zAPS?uwfx&%sD~UiTPmRc#QdxiBOhOh9j&csG*?8v-h@;X9Lo>lQ3}M7U^v^penvh#
znL3lGm)jLLV-@_fuCuGYbF%Xv{$kzpOtBo=oUF+(Z>vU*#y2&=nf#l%RZI`e!`+<q
z)m?1icQci5(5$Pc4!eD~9v>6_%}d0k2=mT5`F1~sA%oXj?dE{nnkg_N>6macv{C#}
za2!g+?T6OW$3IpcA?t?r`Ufrs$`317V1c5<cfk=e-rytM2SPIgCHG%cY~deDD%Q<#
z``C+f4XyE=LAamrY#f&DcRe?98=+Z6(+O+#k_0~r$o9`~J~`EMrRmrsOHvz6oJ~q|
zPf>vM4+?Z_Ukh|2UiOoM=I1Q^W80Pn)a%_Oxz$XDMU8)4Uu-EeEOi6IS!oF9!?Ank
z)J5Lr=MXQyUZeG|8n&nJyWe2y<6{i}^Ap$Bd7^y=JzQi-XkSL$<6LIY&!N8%z6ar$
z4I4C5Mw0lsKJ^qHfB8YhsCf#2U(lo4fhUN)P10?`MQIXi%!A#KuyJ75@Bt}v=aAt$
zhw|vvWenMVfIgJTT;2^2-=Lt!NH>}vBkOk;eceOQvlTKLy1gaRp(0UPLnC&>HtpuQ
zZP_N1d&TUFHb<iVvS$f}>+u0*q<$9y`S)Qf_W^76A3$XU>O&~%V}=r|4i1wTw&o99
z+#4Oo@|gy{1f3`BX}H+FP@a<|)~m|Bq^CgE;K>q`CC#7|2c1^WOEz-s7aaYeLe@j}
z#&!8*94yUHUFL>~xai5GTf5z=c8KnV?(YaPsCa<_CdOL0$B@m7!)dV}ghD@4Ya80_
zHhRYzPa#`>?d$zrem2fKD7XDpRH(t|Y;3KExpW3Q!tb^NmK-T|8+7ead-pCqnGJW7
zzrVFtTSv3R=V}*h5@WL1F_**j2>hBT!M2<^>9Yf(t_!5ii#QR+>HC{qoQ&~*=*$2U
z10G6=J@O{zY&g-fc}gnJ&O&bXElE$+$g!&KMTs8oQ`~UblA8Snn%=BB16tj@JO*EJ
zdUyk$*fw?mq@LgJU}E3bWw|i*GGIBZ_ui)pw(Awjo~+8UJ~X6V{^`wo`?c(ICB5fg
zwa^M0rJKEbj4V@M>EYrkCh7F2CXWJPrz)aMJ&{j~R0|XzD^FMB0#wnN#ywR(>tyJh
zFnx#mJIBmBwLm=K6UvO{mX__ob}qS=rlzz6PnO@W=t$+pOLsj~Oig?!)YS(U5@bO5
zL8zcfQ8MknKN08(5}5U1{^1Fg_=H{+P^{XbK#hquOhtO~oi4Ea_EV>YW(SbZ9ZL_=
z;I<6B+paM!>)k+AA1a$u7(34HXA69{_8R9bf;Z02oVv_>+UbvAm~QWDX5@KmMg&bw
z9MKCcG;_akaW<mAD*DZ|@qzdy3hIbWw942fNh}}!J)9WrQ#fbEIul{&UkuTH0iuc$
z`98S<!i$R%9a$sr;^LB#Clufy=T(X=S~XK}AEkoKzN@IOW{)q86C;Y_gEJF^-4}=T
zZcP1{Abri3zzRO3*K|2NV9ZM>8~IqzRXTFwCC)?lt#LcZV$>D|UoaZ9=QYaU-)m^!
zn25q>w&680Z$z8gTJp2^5r_2#b(QzME7^4sHW{-Wk;r)Ajbt7<RRpr1$o*=`<VklY
z+n;1&v7_r4pBxRO$|Jzme`nTA@qSzhM|lADR!1$$n+UN&p9qbsGd4h5-CJ1gV|?5>
zxY|6sdhvbIBxSoon7dL)-zc_MVF>jtqEi+GLG<wn<K|K*b`5NM+XOt9rx6)%VzimX
z=M!vwFR^lq*cY)F0i5R%Uc%7)xmiuy5h6l5rV8a{If*fpaZUw+j-%~}puD^P&uulS
zT&U?gjqsxRIF@Mn;m0-LK}06toL*UMi}Xu?<-O<E^VM9O-wobz{697e`+rFkVhr1S
z1_={NuBi*sJVQ0`ZGTB+VEb&qVlFBJ_1-3KElBl4ODw2Cr9pv>-;FdRnF{qC52TFH
zBGGo}nZ;sqlw*s@@JVQ2dp=XeCeIiSXHObe&7slcc#~U4m?D~$Fz(<E|CGv&ul5o(
zbVCn{Pcfl-5<{9@7SX(Eb!l|+_eaY0k*q4?wDEL;F_myTrzyNP7X;1Ss&S#sLjOUR
zVr!k|-!@q;qtUjUh3KaCR|Z)l%aR{{Azm_}VauWs?XVZ~SEJid{PKf8TMNf~&0a7T
zySuAWrbPNq@&oeqeHH)nUplH(b}&}v>fxp*XeDGXivsX7nwHQV35Y7LBSFer9B<i^
zV=~?>qf-G42N&&~_=}W^RDI{ro9&sz;PnWi;2n63+4!~q^-pyk{0tK8OCOhIe|WrP
zj^P#J7YbG;@`Vts`dG1W%Ikx=@mO<`-zPFLOw})>@>C@qlj<1XP4oa@&(>zcY|=sc
z_5h`5A|P`N-6yw}*Y`^8_y1Bf&DzdX$YzFrSBrw36|^@yRu&hf#HC;VH*+`S6%-JV
zTsk5-zRo_W6qp!Z!Q>LKnOd2WQV3!AB1!O$r07N;_AKLW8J!LbRT0F^po5_sVlDb2
z2s3<P===oO-|x8pM;p$lfT1e^?I(Yx5_w{;hvq*NQ(FdZ7A0%ho{Cb8hn&Fm2Bl9%
zb=|A0=Te3?cdp8(88gii`J{88Q^)1nDep+rAlNw2D0T0zPZvn;chX`x<7tkw{c7}%
zF2ZmG4@}uesN{yJ4ORL2>HJ<`alDm<|7~|T=&y#vc_4p@DRaDuZ-S8m@=-W33T_Fh
z)JNa`Dt@MXqBAZ@o>ogp%wa>&$@Gulhdi{^E04H(7WP$Vk0YSpBW&QgXYOyhR`icj
zh3YhLF}vKt3k262s7KE8r?Ti`#jkPyZC^k5@eei5MorB-{RKaFkSq}X=2_^__|AYp
z&C^bjjo;zr8*I6-$We;1Ou%NUaI9XL+bMW&BT=;~=g*VUxS{}HKPfR7F=9Gr!G2mW
z=Q1R^U?ia|^Yh6;FyYJaFOXR*{_==-1)-t@cm*74Rq>aD{;RomaRSgt2t<(&lQ6GP
z`Ekw?UU-02_wjS(H(8FCC8YI}(=yXM7y*ggoH0XQ$!W$GJ2nP{PXOBG+2PhPHTWEl
zOLbo0dQ#c!O^$ITi1i+}@G(z>cDcrrGD54QDOKA#=A#&nd7C7@n0N|m)1b8QW5!+Q
zdP8MxW1MbNXEfR<q$0;IYzbw9xV!%C^t<jL5X%TxB@5Ln%D6mBLdL(XdQy@t%fx~~
z6xyQg+O&9BIjT8z1{@u^0e~g%ghK<wrYpCMY+lvX`>(P2dww^;e9F>Hk&phuOMyRq
z?f9ki(OJo=XJ#E#NhT!hKq52!CNJxh70`k--QK<C#6xm2V#AjLDNoDjzNNk|MLE)K
zrZGMq;gJ2D+LQD+JT|lphET92$^=MhoT5P7NG}O<)j_MjC|BqJHa2B&_wimlt?N?9
zwQvC&mOhiWwkn4{73P+DxOzD|s<WcFg+kxK4EnFVROZLiL~JBknfQo_``F$jRj@eE
zxMi1Z+%r91&{$(wNcW1v^zoVwzgFn^izIm#()~N9y1`hFI`#NV`SM}nP~P0pe%ajP
z2;6x=P)yVSF!m|SR%OT|3)XRLKsJYl=Mmx-G#YO7Bg=Qzc+o+KUQ}~wuRox3LP1Hq
zGcS&nMPc_~cgIPP+D?)C#~H5nn^~m{6Hn&tQ^l9U>@h4O_3`oXHDhDE1eB-P#a>>*
zlG5TVs~Xw~(DxfmT4uP=ry!x6cfj8YwjR}HRG!=h9L{=)I}4h=llm-T${B{3y(vu&
z)z%746VK53=3dm;oF^Iy4f7RF2<t8GWKVXil6vuQcYPk;2yKl=^d0v21s(M=0-VW$
zpl6Q{z=b10!&IzA`>eC#e}#^|S(B8;meP`^_LDADDK+8RtLDPgm*R(}k%7LN@2kr|
z2El-2zGbV=XFd}@O>|-@BWEn~A2EMF5ukn7I>K7};LOg*ollG=tExwwTB?3p!abtJ
zfuZ+$sAaH`gIM6pZ<F$(9mc0iD!&x|_-o+Bj7rFJi1horl1zwi(3=pIoGKS@{WyDB
zLP0E9H_HNF5VIOZQ=fF&twxc8)o4_>Pn>iBq|$Oo`@HItu+7O53sPNh>;N=h^Oviy
zsds3@-#V>$6@PewW?AvII@XJKS9_|(bB^>3@r0WFJe|j{wX<8vMzkT+!Mn)24bHg`
zo5O__3Ez9h*|6AGHnCZFp+pBE_h-FS`0~cv{)(1gdbJ`}JA~5FRH)gh#r8+)8w_}W
zsbeSUA_5M@VTJ&26m8X0zcIn+df!8Suzedn)$r<}cDb0%9wY!|vDSI_%(gC*o3i4K
zhQ=$hoe`OUER6f36vp?-6n;>DbuDbUH<d!TX$;ZcU4jegB~yw_cvl}qB^$Ugy`WhE
zopNH9+tlQeb2J>xxwhgtWv3>^K8SVT)tbdOmhuR7e);d7C{Q{bbzsyvQvT3VY6$I%
z4zr?B)l6Wg995;mS1bOkr8aLz{yc@(Soq4fk#@>u7-7mh(i6DZTT8D=HOBixlgjb~
zxZe92aK*5k!W_)~T+xGPRH$_gq70qcO&vRZ^&8Ym9B=joOdV@#8mwgQq0;BT**Pj9
zxt-0V1$U?OKoTYICm>TOl4Aa2e?D+(x-ql~GHsp+^M)Eh%($z10-6_X;<gkm$mZSr
z4AK>;hK7$dRNh5!X>h#YIhE(&z{-%<yI$3gL`;y2V_T;0CTE{m!C~`jJ<w$daJM)7
z$gBD{Zb#tH_Iw+g;)(LPj~Omo=<~qi0C;ovW3aVzYi=!}$y|Pm*ZFXNqq4Bf<>@&8
zD3Ol~eCk7U?cUQH4Y|X+u+S&))q_t4rRiEu4aK5=<iLpKdT8d1&A+(pKTb`_AyrL-
zUJ@1KiKVD>DpYsqjVH^e)_3veDpbVVlIRM`QpD?0XoZBbZdwEwN@QE_F^hp@UP3H-
zw3MiDDyFP~(YoIO{aLHwe4kR?hp|7TPJZEY?qGv5^A4j|*kRDU43|J^AR<|Btq67X
zTJ6&7ryVd8NR;qS!#7uTHMvMAI7oks_VeKqDz^_nw-wVh2m08g&o;Yn^B>x~@7I)N
zPB*vWGv{}yg~ZDK7?I*yB<cY#dMK`Uhwhw`HSBihcT%y_*+E!8NKTJ~mR!OeKU`1F
zJ@6!t;U&E=;u%ydq79Tt*($m#K+N6JW5J0f%c>WYUq(Sq&hm7Ijd!G8lk@hUNS-Qa
z+7@@1kt(qVUgd_UwBuWRwbpk+DeNyQUQb0PA)x4fXiRZ|O#PWmxGfX}EhiE0@Vv*A
zCRK`BD5U+|E(2E{)iM^~=<>imVoc)F1t$rYg@c?j+5O=!<5j1eeP6U;_*MH^f*yae
zPM<s703W=A{V9COO|--JGS4bm*WJVGX7!<V=wN1KE-&bA*XSke`^G)Q)OFcOFu5$R
z4R+Z0gEW<;z!#=37njEUJR1B2($`*=ZEkw&_V#dj+St<k=o04!9}bA*0}rh)wjWQr
z@5d6Q@AahK|An~T4T%89lp+3bSeRYjgInHK@^L*X{K9h`iaUKjw;i0fZ47<l&pNxk
zZ=8`Mt2C(E&BerE==h%M`LX7F737&Kz}zxxlg*6+8r&D7u!6Lfj|-Qm50JUvW9+^j
z=5+18S=hPSY!q?D>bkkf$za$;e~T5^?5dGrvx}|{bRPH97Eibd=a|^hp0~|z&Ch=Z
zPl3kmjI+gcU%abM-8q-ge$kU=Qzt4WCh8mFD4e6yj{>5McX{p`NLB1hjzC7mQYihq
zS3poZ200auAD_&KRcpMV3(8;%bM=04^Dn6G6m<y6^S`T8njw3=pM6YpKYDol>}>UL
zDZWeu9(w>9{@bK7Zb0i78d?duzN5`7<2bxbO+YPJ$Fb`D9|@Fqay)eSctVRKi~DI#
z-R^O+|0i`LeB3`q%HK)kCDB;cI$}ItCVXL=MkiH`$9eVR)T;ACs3;syXE^g5YLpm`
zVN>FXEjrwq%x}e2kSf4J^%9rdz+H}n&IT(Nuuj9IEVyi#-n8FyrQ^DwRE(Zxt($!=
zuSX^h|IWK#=Vs*h6ZcmXGxpn+cb}$zzwh?H{=F9wFZY%=u%r1h%vDO<p3pMzkg`_h
z;hL=p+|(O>(qI%|ix*aNem-L%q1^!OJM#)p;Ai++HtXkT73e*~Q^w30N#sJ`{Xp|x
z8SoKzu9Ze?oJLq*L6Qymhs)iY)a5Ao3}dj{c;78W>1$}e1L0_XAFk)><P>9t45`2)
zSmIc9Fjn8ZB<wNL<(J2LZbd@R8m^^8+!J;TCAJeI{uZwLVw$yhLHAYh<Q~_^$w%pI
zI%3<*BjZ*hW6ND|O*F4PxVjp&_|%b^90=!SuSZqysLD-@9A|FFY}HgN8lk75Ix-*-
z0<j`2oFUK8Vwcj&?L_x&CkUmXoks~LY+fDgKqXv@=TIgKTp9+OgkN5N%k@KO6K|63
zzExh^yH==l)=$vWid4z67nJyl+Y|c6scb$kp%mP8RykM75YaAi3G>@{rut7Akh$}8
z>^%QICyWCp$2W<$En9JJc+M>;$4PYxl~B+m#@6{t;c+A;|862wfT8+aX>qccl_27W
z+^;m!Tu-Ew&SlqE>~+|ps|I{3EyPZo7DS-T85Yk52JFSN+~sTYB$VS&#MsP|UH(SM
zkW;H(+;1p!nTc%GoG$Y8wjBm+q;Y1ru5|K*N{&#Hu*nZkI>aRW2=H`tc72qM*+xLu
z>;<7QC6M>;`l9~!2ckgnl@&^mr1HE^@gq|ai$=)%oeK*MVX7aF@L3ncGFaPHD^;>U
zzJKx!9v6h}tJwZ8F;%f*>8F@#%%^c$d^rh@jei0r!9rQs)L7HmL-xRuDZN)9J@lWK
z@Z#dU4LL+{A{6sI?n`ZAvNnSC@AxM84g(u&3k%uYBfCyY9EWGw8Y-%tsieK%HA?%o
zXALSw!otj-a>pdk`#Fq!^t5pn%O|Oz0YOs|W%#9}d=zi?yB`uCP}M>S&ix!Aj&G1U
zD1|_Ze5pV`f#1;3M?S#y58e=52w})m?JH4ZqQgiLd5GAX&i3K?CqavO26{3?jg6Bh
zs;XKBmbt&`*<@FzL3>T;IGdA!Qr+gd&hUYo88Ct7rOd@$8Mmcmb6tqbr4{l9rSH_7
zJs`mMzb}ESUZSToxf>Eh!^fL$KPwclJ`i&?)D`4}PETN)WpE%+_UG74X@m7AP9A84
z+E2Aja8kEz@EZ7`{3wAc+A{%fxaEwn`Z7+{@#zqG08^YPRu8o@3x_`0hL#dK+i0L;
z1@X>=_y{@9YOnt>^mB;awsN|?{F_3;pUJ}h-^ijauy8qZF<%kmDJtSFE?L~)J|4`C
zg97hF9wxx0*qk8abO3R&%<AIYte>9-+2S|HGkKo1o#n#Lftgpb@nZGe5{nrj+q_%-
z;2S%m+J+S6yq25w=5Ei!^$S7H*m>uR!<E3hRUfJu1cGiQXd}Ku3s;6yS3tEj4i{y+
zrUmw`*J(Bd@sBut=!CeFrOk3pdhuuRV#1W#i7uo)q#Q-|fK`n=_|&c_ZYYyh+P8Yr
z;O9DG`_f$Epi=@gJYt2IaJ(lBHA1-pL=*)LujXuK%V!*h#7!ohyhC}<zL;dBQs{(h
z6N{@k8JiS~wXJdC>vgo%!^)d!&o6Fn42sDUl^{&GMWQS!rI4&9C3ZqAwCdS5_LIgL
z@8PB8cGr9ma0sBVB<&&hdlDimH1UZ~s%XDfG9TMZ7M11UGXG-LJ8)P0%NlSrE76yh
z${(}-q5|$;IFig~u^BOgnAZ&2>j^`dMTCjCF4_C;&E>V69E?l(`uV-U<iX%KPDnO+
zMi1$CC;!R@d%oX+h0IMEeM<7h7pi@R@}$)1Uk)?`lKs(`@d{zuB`=r{?mJzw&lfIc
zXIpE|!N5k<c=yqrrp4osVfW1d$_Vf!XbWYDTpa&mp{wP7vHKLbLmBK?J|Z4(AGtC<
zH;W9O=nNSRS&7pSDgVbB^e_N}YexI+qGq$zIev=;xREV4cb~Iex^$nT3hBO*bm_)l
zXMqTLk54)(uN91{XcNu8eO}vc1&$lrYhX%buN@B%(7dTqHxph1B62z655b}|x^4_o
z89h$Y<EQn@4oY8>li$gOX`L`-j@iCrWvKQQX`21UHV{gB*dXv%g~t7jq$JGp6TT`t
zjo~)7(Ev414SIM=$_6hk_LfaH)b6=z;S})M);J(2h)b;7XDbGGb7A2r`sJOSba0T@
z)gY^3=Bg>YndZD@*?Iic#>yS#<N02}V~OP*3pR?n(cx>ho-zl{r9Z$@2VbPPu(7;&
zbCNYX5+rPjJm3GYWNA$PK`>{mv=?%Hbar5Md3LpWWM#=~vBN($9MUb5L&VtW6RoOE
z(eKTKwj+r_{Dgq=`Y!(g@_53^cs;nieY3WPs?L3G%0X3lZ_+GpwtAO~!FLG8m99YK
zSdNGqyc;!(68J@lH2CB><6}WV#*-6t1^_G*(nzAPbmCP>f?4<xDrQ5#aZW}bs>~E_
zpn>W_;@ket%Pm!a{W0!k^@8bIs*Ii}>r0!@zXA<ko@nVjSLI`&etX97Rcn<1yYGut
zk>B7u#k)YmM#4#PGy9I<h7JL4+h1=e^E~$_{Kt)6a{h?p!+!yR5xY2T4MycoXdMG=
zV6&-PU4zParITiv)2lrPCvxX%BB$I`z2m62@dpH&3MlM&gti%tlqHw<HesUjhB3{n
zXO0v0`IOrahk=fcoUt7p?SVm}*(*{D*9gt7>s#Q+6?xF%$GbF7o4|D!S6Ub6PW~#w
zme$lzSh%T#fFzuFt)SZg+#J-$atlOqKFlsia)vP}$s!ArKkLzT>QsJ&bHXX0t*QLD
z82wbqktf#<-;!Uvqf+T7(Zc`QzXiE2XHTCchtZ<;Uv9P7>uTP9+ue`i)QQTy6FZ}y
z-*A|YZl_vx8tA@0NtjOYDO(oF79XfC!6N^{6{is^$vGl#`f797ij$!DItcKALnR>d
z9DS2-eNMOc{l&S*4zA96J_VSN&G$~bsuC~va=EImsC^4a<OCCB!*11T8gT_<)(;P3
z$AkB#K<{PDxi5V$6D~D{=f<Tj5A0$}q?Tl6Bk+F2I3^v#jPRp~@UDD~{j7&y&uKY2
z*T}(hXJA>wLgM2mf@6?|Heey`+5RnE$bB@*PeA6r#`Ur9q9oST6plPxlKCg`DcrfH
zM*4V%yg1%{T6e~Uh$3Yk*0U2R<9E$zd7a49eRYCLu2K2s@hseUI4fJdKiAVbUdV?p
z2Z3CYSPmxswr1~jGb%8CfTVxUl6%^pn(K+`_lc$s^!wT%488UOJZ|c$9Q*E^E}bq`
zd%^}7TT|0*A6IeX#^|YB^B#Ae=N#_rY<e}Y_n$gt<xGb95Ta)8m{uoP)B-yFlQL|g
zXA2_(Ak=fu*Xf!6+0TY)k-V51nx-;Ta~QKd-P}}gODg?L)sN~<Rm>AmczHFy`x!EW
z=1Pm|I#apnH26RO6n_$?w%hfP%yR<CJX-la9vv}hdGJ%y7GE+f`8MtN<e<t%mO=OD
zB56#^2El(-yUrrUv+7%#1CjHZz-v_Y9ow*YXxV-?xb~3b8DjC{i&1QfQ!5!P=(shp
zra8q+*xC2y+fe>ei*rTW&E^HlAT|60#uGf6_Z7#oVjD5Qc=|?XZ&>#^e)rwt<Kkmi
zUF%698>5u8-;o;1<9rSF!PPr#&`*>hWN&KxaboUanhc>VbN4XOvo`m5fU+`_xgJ9`
zgApcPbU&1J&y~lra3-K6;SIh_NmY8DXc6d`03)Z;;HPBJAzbU}5T-L-L<|_4`vnWl
zz8K}K1Q8&sQew3Y6;+B}Z_tkw6@`D{ccHICjhIkbGmuDn{KiFl%f^9?^<p*2?UkgT
z+`B(UQC*C~=i-hy3T|@Lp0ao(vVE}84zZ?NF%Aws6|N5ylFa2lP|=_;7lYVM1r4t_
z&V8@*@Wshq`bq4vVaVP67n8m_`=*9+gD3kD7rnr|o2xL&dn39GUh9HtXEyr9CBX2h
zs`0hW>B9sbe(|m#_8~+2iAtN%Kd&Ug|48-ZsY6^mYMK}0wznv6q6x`yi>neIaP<Qh
zX=wvz_vwvD+)G2S6rW~>1Q7x+K{ZOX)46)KpURQCu<(0orB9Xb`J{DxKra<0S~!z6
z7eMf~M6aamN2)lynlqM#CnlHp3|j1)N}?L=IV%8`SIRv7p~r{r+5p{qXZCVO>`1?s
zlu7ly*w*R?F;GGz8$pl6hWh8Hovq&7QNL}(A+<}(oibw->hHM8pJQ+|QiMmDj!D~*
zz$Y#)(n7!@G$8d?y<XDMy-cZcc@+L_W<aTcbxmWQ`>U5YXAS(Az86vq9Ys(nS#oT0
zgZB;#KRyq$L!j7ly!qpZrLI2F5`wagXm0$Pc(V0{G-WC~U&L||U6Bddw>2;6jMty^
zw1#ozn5pc{-1@0(2E)JOs#MV_1(kn9hsQ{=VYL#6lX76|zW#nv202@w>$(FkfB<Os
z6`Y66hS}yQ=W&LzbZY)0;&9h5_m8&ckvntmHGlK$mhmhSTxx8Q-pY+{{j6B9cIJo9
zyBg`NF*M#0e8U<qF}3f4+3|fr6m!YLXqZWQ1P7K7AJ#bSKap?nl)LA_&&#?~DbRk_
zH6>@<Ab(I}1iT=;D;b*!7|g$EU8m5FEOC!$f`)lB;F07qGe=;)CfCyG%&56Sr$>*;
z7`Z|h0tDt{PylATspExVv}$tIkQ;meEU`1ild_Q_fQE*HYadq<O6!O+xBMC^(Y#<t
z&<Rm}Vp`&V6Z@#u_kp--<%$j8wZdvrte4%LfxTpq_w%2Pp@ZQxPT)9lrGA!l9WRt)
zSeYl}3GSO3Rx*@GAAC~BGh392=4Xd4^YbZ525^loj=@lw(SD7ttvbl~4I20&Dm4@b
znyK@F%fD4n=I-JUxR!ay!Ex<6i<h|}Dxvl^Ic$b?>{6X;2`Agy*_>}5Y*R{KzU(hD
zM9FeIW_hQEmq>7i!43Lrxv=CzOtIN)v$7nxNQ%-%4-7g#Ur0(A3?e4Z{r2JKC`5AH
z{PSCuVGl9OxVW*<?-3>}v$+Cj7E))GVVBa|<f<7xH|S*m%+2fKdK_4rLZx#-9%oi?
zX>0Z9)ndl$OoC$^p2rA|5tABWss@cUDTf3)SqwkfeJPw(*_U5*Ust~Rg#_(-iGxr>
zU9iUiSfH|_kC}IogdWJV%gqyBo?dNbT>r`X!$vOV2P^8FGpjfps^=>-r@WHC2yDM#
z3;W>;hhPG|AJsjm_=Sa6*@xtkU$Swj)&EGJmIJvwOp28`AnzZ_k+(M&X(kTF&czcq
zA&wQ^$X4&l_aO80#r6xBz8xGY^@rN3)GkPFt7nb7^B`LzqsbW^?O&HQ1)2#O+RtO_
z@WpYR34}R(YQ1HZ`7s+bkPgM+h5-|pJkrOL5FmY}=_j1d(ZWw*{P;VZ=D@GNR#|Ue
z3}d2bu<QQTME872zUsAM6Vga_;`)02?Ipt@mbs-x0)3J==gDUX%hURSw#r4_8d!lH
zY0Q#mLKP?^%1GHQ551Tqf-31J8Li$2@$nn|YW*46Ffzs3KNO+iF<&Fl$P^-y$p5IL
zeFgBe$OdKq6vezN@ixyEQ2HfDKG&Dt=B?qMYUJRweXSVSOxFe^{TiGWh>_F6mc<Lr
zdaLwm^kXYnsyiOlXMUC;eMJP?VB?<W7g%;v>qIl%e(#fkXYN?pdby78+Pbx!__(j!
z<YMW!Hmduu)j$4h_^0m*q7gbdt?u~ukOyFpX?48K(;Y~JjzhWM<Sn$>goJ#s{mUZ(
za)F2~N6!>2a*ouB5^SLzbOhfV7m`?uhMcWN;FJb{W^NN#B-jo2E9pboK7M2rtoAnW
zL<($ITB`l2s1VctPNOF$XSN;ffkrIw@%_L(CW+_!OaUSEU(cgxRd%TH{5U<JrL@35
z9mDp@18qS=tm4}85n6mBUu!*W&~Wx%ncB>`14v-0&U0?<tuI^}=j^q|5kex?c4J7h
zRlej^dIiSNUM1pb{r9B)o&sR!D7f&6t+daH7Qw#BWwyFLtH85JLVD*GTk|<A6fFK{
z<M~V=vi-8c&>+LyBLZuZSR<7~=@Vc@$xEXoAQx8oG;jZnXN^?5kLjPvWNEM2Il}}7
zu~q=7NNI8O$1(5Z%z7Xp<V1H47Cc}`w=;9dX6`;TqfE&7aF8MWhG}Phx{>bV3t3s4
zC_*!Cax1X>|LB%-#fjwv6m(4G1qHr@V+9HNGlyRORW6=p)0GylzsN+s`xTt|ob*-Z
z>W43x<O~HmJLB)ig=zmRy|xBQAkv+_8C7-d|BNN)$Iaw2sgtO~?c<C1sLskv-e_B;
z8C7M_J)*M!Yi780Pxw^xW3aJC|7}Hi9&R59Cc)9ZJSBaW<<OrE5|`N-`1<X-=5XY~
z$s9S$Mi5t!W>?9qgf5K6uN0jv@#O3LsClca@^ii7kiGu>-U9JPuRU*>i~FbZkut|a
zp!V(eZ|K1|VZrS_%)e14PeM|2UC6(vY4ZEej4g-&U_*5*;J$v|VJ6LKo?m9C_Da&V
z0p4z7)NzZ8LD^+wDly|EXls`(IkLZhUxex?wy;|+-`iVgQ|avRsGKq`8@>E^=6%Ev
zss*u^yx$$)^O3;mK1tJ>@x+G&UK|itR2c;z_XZWU<3!GXc<p)pW|6&d`;L<Q<&&Ql
z_1p6SubYCbk$Lyoru6O0(1nwm30<|wBA<B?euB(EVSJKbkjmQHH#J<`4*KJouC05O
zuEX>6tw>_5ZkrDuTwIzMWcB5%>VCq^K@QFz&TM8x<X@E;#A1*tCDpf^q&X7CKZ~i}
z@vRJtM9eH~!%=QplqQE4lQ$j~68T}*KRq9Tra0P2q|~hOeiT{}f3da4XuJ~U#7Fv@
z*=IEuO9xFQy7wT7?b^1zy~f7#?uY&EJuh*_`+t!x?dKO<rz4zfB!|-{!@-e=6^N^A
zyHIzNxRHgWiz~>bv1~}PfYsf`*tUz>8jI<-*Dp4b`qzqKlU%1TO8-EIQub^#VrvBe
z7yfYq0HudNm$djw1Mc!Ntu`oM)#X$x()k8Eem)>?Tw7BO3%s&#UXR$NCmu|({vKE7
zrtoU2=G(N8JsrW5Qq?b3@~KU`vTs<T8jZAeA~UzJ+tn3{qo`xPqW>Y*g$jHjk1`c-
zN~z~yiZbElL?_c1BCVVzpCxmoHPSBxE?DR-D-*nn3Y?dPJ0C-@y^Z$?+ElUyl4xQ=
zvuI|u|KJi&e9`c6`aEvyFTuG@agx!wevRMW^id*brRntF`U272zww|ch{aE5A2o>m
zWjk~B$mxY4LVVC4wx2F@I^|lvA7b&7nu{9@SKL1;>`hqS3wYdA?T+zphB`O`wLaEb
zj3m(<oM!DaZsDZ8bMJDW#20;_AY$xlu6(U6+S(u(Rj5Nf?OiOSJL9prbKfb~=nRL>
zU+?Owv1RLGi?i?Z2V1<3Sn8kW7+rE`%=L^uCYkWJi&DZ;;U3XDez0a66LlPY_T(S<
zhs1Z>zNJ}bvrnCs2Qz6*THCiaK=<NdZ7s?+zSe)GWj)3Yg*X7)3pFDbeXrFlZ-&*l
zLr&7KgUfaEnnW`{I2L?jaw?JzBM#KQ@!pB@9`UGoEfE&doYkCv9S1k}?wiv3i`C@1
zd^}y1ce4bSgulzMgyjxjnzuO4MJy)1VWctgy?u!2S0_wBi=QZ^8Jn*FF7%3z!sAG*
z9Bll}SrE&CJ$cJG+-*nEI2V>Cn}@{fdwU+U1^j0Qv*~l<`vaP}S0~QC^b&O~o&$No
z?cDZPzaj6qIH|FsY>ZzCWhHMZTBjD!>}%=k>wgo^R3dqsnU|fPWf9dQAGOOA9Vyk}
z@Nq)(sQee!qFqI8<?ikw;JKK?dCBNJMlmL7jV_e3Q2()QrKZEvVS<EKHnnoXX+V4v
zX2=rvB8K5_4t7HnaTspsTUiC{^1ZgF#wVLhq-Ypc^zu*(g*_^P+3V{H=ZNLzhF3QF
zoUtWzc2TcM+&g3LMehdjnA4J1I(1+uf|2N_KBrk1;~7(@23w>lAoUfRMa?Rn|CvKO
z;%_uHGw+|$un`|;Q>jL;<G$NlG8y;fGRun?*O1>6YDC_Pntl_G_WO4TIXJCBBCsxC
zX=E(T@d8zWIcVPrZAXgN*PAshA9T$hs)0)0mbIOl`j|u+UF6gRHBME~<D|)bw$Zmy
zr@?;BKV2%XFW1}-a9a*k=ST%|m+yIK5f0azneJ}Jb4@dQW}#2_%+4AZ8qUF+N5`R>
zj9vTx`ZQad_7kw5$Ulkdy|3M9**a3tArVl!ehn3~6{LtTsrIl4n;~2>DQidhVLk5L
ze{-nYqII$PtTm(NO-&MOzzQCzq)FA+Cs(vC<@4=%U>h9;G{80BmeIob_4Nwd3BSpd
z+63EJHS>XMFNj#wzCEt3_cGX;L;E8cveTD+u#PNpOfBEdW`tg`C7S6I^m4Ynj3RJ<
zzv?h+_ICpIuzzdwB()ypL9o1an-EnsW|$rSI4(5ML!Bu~-Ilb3JI0`U<@zY{aqG{}
zMwr|{U<@CCRS{zgJG|5*&%)@hxHUns`j*>xS9a;Wt(f)rzHd9PFA*q&<Dtn4RTinJ
z9LGyKqENg{Bc94#^G-LCx>&aLp17?VQ6Ztq*j{(8&C7vfy1v4trks~6zFRi8tm5+q
zOIKL{Cr!=nESXn1cgckMQ2aCP^MtApp788>M?nBBMt=+Bi7S>uO;uRHPh7O6F7`Z|
zNlQO%hhIQYTN74<CcrPIN$~5HF14Z1oWts0v4L=Z1A^f_J<E-I$J^#n;+v2XWrI1w
zi{AUM*%0H2RRhj(PdvFd7#I>6eN<nVzVNl3NfP#bMP%mqKS1()7U3?&mlXB>&e!el
z-#7z{O?Gp6n2u5p(`+!!SW8%!<K7H7;Jij;`3h6!DERTp5P3TZ?^?CJJo_qr6786O
z9_M<KbJ>|cK`-GmTItr4-fED?Q&a7Mc#&SCRM$0&osl)6y39k{VdU#i)pK{VHBqb+
zSh>9GwJ$DATjNu;_Qjf>b}TM;|9nKL3U$yC2)?EF^&8?5bz|zUC&U%L`yN%1#EK@%
z^!9lQ)hDmex-2}jpSq1Ed{ML1{ys(qofcqIeey~%rwfTN{^~!X>@mNuJ1Q&NOfa!o
z7P^9!IURXwxI*q0fOPTlZ!>*c_1%Ipkdn&4T6@BqP8L@|nwrYFw*p!<Z^?f%1sely
zl@oC`<-_9Xw!_WK0RzD}L_#$2jdI>6(-kx4c_VT`xp{dmjiyB=ky@(A!MfY#@xwt{
zn{pQ8SA_~{{%^k<@IK)Smn-j`oOEcOBdM^3FDj;J=8F1EX@qqKUTpdMoNRKi=om6e
z`MgjSnZ8#__b0k<(}`}p=)A16k2eAq^%O|rb7PoPqk2eKWm$PTM)aI{sjeig6-04L
z^NH`V5}!hgb0SI_&tn&K4W*A|qumA$l><55WBkYOmL?r`zI`3Zk^2=s`X`P!OY*C#
z9WN77?&WxH7F00S<cIDo8XxOZZ-p=l?QaQL8Xwceu=(F+keBH3l_DSP(*+7q)~e3B
ztq`izj@~EMc_On8yPu!fR-s#!RI}C^`q@iWKmYSvNI+&L8;Z}*F=8R3B&+>O&L<R?
zs;Gih37f<!J=oL^AGiHtyyk6Vi_gg}Ptfi%8}fE|sJLB|>kUtB$nmzO*-zgKE#Ebc
z?fj52PuMY321|2)|9rBrSc0b;2q<Bkxk2R-C-ZM!x7a;$D+jOmOryCe*q>VkhXh>g
z)V+k?_faB|-s1Z+ud^E#Yry%U&t@qk^31oF65ilDf-K1QaB<bwk2d#`iPbeU872KR
zx%Oj;%$htG6M8R~Ok1uy1M{^At5=Y${8UW*KXiA~GSmJt7dgm!6rpA%d&0?I1cT}#
zVJALG-U4$R?P2wpM%Lm`hOs~c;sK*{b6w8tHy=4a)a;$nGhy=j1ngHWH*TO<sZPyy
zwDGnu2f?zV&6CPM3|jGV3O%ivc85jwdj13@si3_RHgjQ_?)!9}5SUPwns3nOy2|wf
zJx3qMoU#`E2I~XwJwyS(NOKUFJ3jVa?MogZ%NNCqX|L20B5ibqiMGMh_jaAtH#HSO
zg_)I4x|(6e=F&5-Ppa+JH5#ibE1z5srIft~i7(jy9Jj4(3$$7(2}d+7BA{Z1@r4Ol
zB2So@Vln8l1Ps>8HwDK%qYH&Sw_}Q!bmJ$qR@cfy){FC3wV6+Wki|O#Ci{jPHqE?_
zOGNQc*nB=^yX|$>ikJ5c=l+i4S~>Ll*IzdFQTpTr?9P(zP!c274Gv_JaQ$-4cAk>8
z>DuLJ!^Qh%yP-y^VAVAJXudD1`i9I{mYbXZ)D;zHB(-aK4L29<OiXF6)0HfUBjnPO
z*kl!I%Z*68tUFQ21eI~L)F<BajAxt`0XEp%WPbyZbm%(z5e>m=dQs?dIqI67C(*u3
z?zf}A)r*s4uHT^qofICsq!Oq{v)NKJ>x%T)H5pGI>FEIo>F>H7t2Jb!>G#n>P~x7S
zNzp^X7inT@nnqRCc0kyM8J$V~<e<Y<xa^}UCx)ok&R_*lqL@KKX{0s|m83m!j8vLH
zB@E`~#*UkcNpP!4dkXH;KhV}9Z8HD7Nbv($>+QP?bJTFSw36@RJqmv#x%)4^+ZjZ6
z_Bmg#Qh3WC0Lgdg%_{(5PwsM~x5h1mdgfDhEgDOCTy@ETtN0xB1g)@MIW>DINw}Xp
z^!26u`2yJMA6!;F6wpwGW^y}aWrG}rO{SfC&YTJja!^O+4+SP7CoHG_<-Nm`>+6$4
z&)E`PSGt0ITneZ5pMCrlmm{FTC2MZ9&%*=e%6hjg7OpDf)3(2i2msIJfIvApruM=n
zP#d&g(AH7Rk5QAmUSm#3*z#e8QdC-|ILbr;4qRCJt^Rsgw$(2WOz0D1e_O7|%jw{j
z+31UGbs9k6`y4JW<~k@)VeJ|gVhOs*suX`|N+6u2KvOCET<SSC+2?g%$j5wx$3Q>L
zqz9Rb$E37CEKR5H?p9qQUnlU$9HLJ04|f_i-L)^PX3jZH8TFQX3X;Vg==F!hm~5?Z
z08#yxiZ}`ym)J6MY^1AgJ3dUlsc%Xi>IrMi6kt*ENcc1Tm#n&)lf)AmD^CV7q58|J
zjcd`EWaC&xFK=m}uZ1sL)^^^L)>y1>OwsD`E%a&XR;rEBj5j$m=p4g}c^d7@<{$Mj
ztuzgI87)_#Wrm?y@Jwvp<x;E<vSrsAV9Ysz$^GjpnJ{hjT?}e!^9ouV9X;L{&_o%3
z6IG(de0|ow;xBrhl3(|^v|7(S#66AvR=((iFgheCvT+2}FV>_R&Ia7erNh7dT-X&d
zd6uC);dQ!L%77qPIQ_;o4(_<)&<7*DZ|^O5RN1qs`Dj=q49WUXPZH6#QQHRa#8;!*
z6euVu6X_-L-cOCy69_vcp?~p$&5AygB$4`k<VnpJZUK(ORVjGwZ7!1Cmq3pn|9j3m
z4Jv~vgdzBB8R}35LC~>gtDrn6@8z0fI*Iqn>!gGl124M_s`T~4tOt<Yj}OJh>Cv)a
zQ*npDY!5+DyRjCbzJ`_Q_BKG-<YH=JF4fqi<c~1v8*R$!cfvID(ujPo_bE6SL{&eE
zZeIowW>;frcHZ|v3fQt)LE9zAPg#^-!6jLI0ztL{zGB$Vx{eef>@?S->Jq1B8JN05
zl6(?>xY*N396w-D?)f<!>sfr_4ihKb`}aKR3oh<16&_UyZ;9?`rWgQ9Yd@CICED-k
z@pRwR<n5Vv+uKY^1kTI_Abp;B*XwUD>CMc$x_QkWhU<dhqtSBu-sMm8246ns>DT&X
zp}-Xkd^1v{)5#Oa63i8dj$$veHyE%-d-Cg#Qj5h*DB08*4!Mv##;5UcCV@Yv)b?>h
zWgxzad5<!lRF0zT7-N9CQLlVH8ok2rk7Sw8BIO$k^TUdkdLG+sF~SL~Uj$e^B%x85
zi&Aq7(#xa#x8Ljl>*YfRF%Kt)kdfoGEtmOLx66by-AjodhZ4Sa@f9T-oH#^`93@J9
zl8&kJ!TC0dkiGNmY{Z2Gk!tGU;`Yp5TBDqvgWywMJnwiVZ#_W!L%0&FCt+@LYwK>A
z)L~Aulhj-%W#W@3D0)!dc*ui*H+$f{OHG}Wz+q77H-DJy-?Y9GqexV*pbfX9aEzpD
z^7WEXv&$m(FMa*^a)uQ1d_s)aIvRo-eIX`!foSIGy6U(3Ghs^YNAcAO=y)-b<DSy~
zCz~h3{LJj-BSAnOyb|r3jd4!T!GYPfx&Hav9TH+{61SwSItrGvEEG9nvt`1m77QMT
zjycb_k&~dPc>(tp`o0Pn?AcZjJ}Ig85{fq?^>e1aaklmK9lyUo>&DF{>{|#bwNODx
z;K~_a0&d&1<?rci<80&KN|zdM8<y20t4H-c>QwUi1n?fL`|a;&>?kVgN2S{6r+ZJ>
z3uf1Z7|-w}+j-y?sMOhuvP5ZfDe0cnib)Asl075pEj}WdcTYD^G0qD!)}J{LJ*6h8
zvz8YT5IFs-JgJGw#qm%jDffivecS*43U31S!@_zl3H>ipy~?x-9T0n^AcI;_KLSJt
z!gttEOLz?MOjNe@T%EUPgYJ1zK5FHaU#_+M*?%ioiJt2Jd)@zYH3ml1gS-+CHt`M}
zAbjN^!<NffBy*Le8T6pidR_Tu<+x)1@z76l-+9*mh&J*&BFVJd@A8E8HpZmm*pBD3
zG_YUe3+NZ|XfFAjj#-K?2>onqckMtdkkqzc!(5Pm?ZBGd=8vttVUggf?$*0fgybX)
z6w);C6#A-5ne!6-^td!1m^?TI;8iJsjec`**Xzk7=v9MZ=VJ=;A;mZ3=tBBear$)W
zFq;Q(Fc|!tVT(YjyAkBtx_JGImfHo8JZ(mj3QQp;c!Ib4$6=j%Ir@x#hw{fKkNzw7
zmJh!TDXO&R1Fl!$t`E({=0f-)xn1r8HzM8F{-!zji?%tpU8f)%#_B))zXx2~uS>^Q
z+G$2N|2#uJY;-^V&b}ZDzGzEiGrP?SqnCp0;>aTc=PoV&=ECYcoBkc2$Xo&I11oOV
z>Zm#9=EY~450xUp+xdW(p#e%@^a(`xyLPgc=6Ib$X=^(lW>Y?f_2zdX+Dr-0m1guA
zrHfJV03qpzgqkd7jr*92D%S<Y*J%gAH4wJ8_lqHqTkRGXG6+CR45WP|IcTnVKcy*I
zw`}&0Z|>j8f#t2&-00aqpxvSo#0RWNB<vNZ=yg_{L@zN4c&sdaeAFKMduqBkCyRim
z?QFq(yF7fYU69zOw1;IK$5U0Pdlb6pzouwfnqAv+CG)uPZl(_ixu3x|<!N)e98I9?
zKKBGoh4Lf@J+$q+NC}Zfb9cjMP8Yi$T@Ckm9&d!=M}W0$$R(hUcwCWjD-dc&t}p9C
zG;jO%(<8l6miKp3hL-Mq<dm+i1%Z>at`*}>0Y6(6TalM}iye1x8D0Ik?uXy&)UMa6
zMIPosk3^OYw`oAgO0X-i_P`i`tjWK>>q!fEyWhEQ0laJ4Z+txJ^X>Rlc|N*AGonD~
z%GF&VA@f++4fq|wKqXBO$;EDw{?m)Qq2=|Al?%o^-hAdwo|{L8fZzFSt@}eVA^z6m
zM1Y%l!u|1xDdcf(<=ks441D{|)qkuR>CkaRubd9$Cmh$%aqU`5gqGipKYFoT4n!@9
zG`3{9N#A57s_4L*o4^rE<D3&<;6UVYL-TH^=7Srs47BnB!{CsMV6gdW1o8?}ytb1c
zG9Z77?JDE9XqpFW*<@P*=A@cC&J}C24Db88<FI%x9}u4P>ob5pAt!;Q-;{oQS@#y_
z#PH?gYS5{~%9NDj;__uzrLp9tNY~E}uS4MRI6v?*L#B1YVJGCo4m@aQdDaBh^Ua+<
z0|L`U8Z-ZPpT2c{)HFl}T)4uLYr(-g13U{i&X4<1dW_e{%l?l0c~YL&zahZJcFz@|
z)G-VRn~-#y$h(`&?|Ntmk%8Pj6UmzZxjsgS<d*o_RUFFYAFj}E;B~)4<Oe?X4&L*~
z1PQn1%s8?oz)rg@ArQZo|K4j3-UF&q120$0WX^WO%%_Tht4UOytHbqJSLQM<qw{c<
z=y4O)kH6x+0v{Lt@9wJE={+#6T*N+al`+^1v5n2ag{Qe*M<}yboSNp$U#H*tKkIMG
zze$F#u5@iDoAaQ~J_;WN(JnVByR9M@uah&(oqb^!X<@!Al<CqZIWl*^!XrG`V@$}@
zJ9yK$cv<EuZ?WsH$?y|vq)zu74r*B0HmkX(K)kA|(r;MJg7<RQvL#@S9Nb;&J)<9Z
zb(SBH|1iUNfB$ynYy>zertA&Q$wnX~A$`*at2>mg9rx2#*DzMjCk2=H*RFT|4{u%X
zV9Vi@LGYZ%14|iSWG9dxJd)OYcd;nLe<85mguL6zZ@U}pfsO#j@VptiIWODm=1$ko
z*1k7tBb_+NWn+iH?{<M)YipY!b5c1LA+FP3$2@S3rWSn;VUxB&N2TwrQ<i*PsfH3&
z5H*<Ne#ZhAT#r+<7T2-OU8{32w;l-qfxG|~O9r~U8YXAj0hdnSL#~aZr5E~^Zc1B*
zGl`W=(MxN0L-(T;BF~`#a1zzMyV)mja|AqPXA2>7Gn60TdU@S_#fH2}Qj4^tZ{8{)
z$22!KsoTw(e`88f&lz%H&UpP#`Ed=@eVd*ivRmtZ>S8+=fz#Bi^szLe=JaPrOEqe1
zyreO0Zh}e!Xcr+!%aFAu&5nJ;t{VRh<lW%o?c7Ll^JMqh9B_7t=-lpp=+QLW`$8cs
z0s_q!fjZXAfvZd9uC;UE0^EH$*$do26TRh-*Ezl`emzxP@Cq@b=HC&)?Ub3mbpK4x
zC^OzGTb~~_GIs0!^c@*+F{8)-q<d{+?|J~e7y-DAMAeWC9ZfjWi`IBVoR?V|9>CD7
zJ0LQ2pR}sp`Z#v8bHkb<@#T%>V-?SB|6It4F>q4R9lZQ<M2&puq+RBrxchu!%k4Js
z0bUaiiSPAH(X+GO(p++%)s{-S7260Jx+idYl^ZO<nDhhc>THSMnr|4}wX$zIo7W#G
zgNm6_MGk1h-Di<9r~Qu){iy$b#szTNY~ej4(j7ED2Aq{X9@}+41doS6H6QLP{vTKG
z8PvodukkJ+KtxJH??UK;7<z{Q0g*1fOAS(_R}m6=r~)F=rArZM(xrE)(ggo>5Kwv(
z1%aD$&z$?>&g{&--JQw)@-5Hve2#M)&k9!lrQH5Y%6<e3Rydk6vGkt6^Y1X%=ifTd
zRr=$tj}!Kkn8ASj$>2)I)ycPzNui)E$vxLkTifwxR2%fmmm8fvyL+eB9f#wTn136J
z7~3DQ*;&hr@S_6Ezmj|Zy5GGx_2@j_zxV2`bx085oJaG)UD1gb2h^-Unk1f>*|FaC
z^wz@TrW3P&eU#&YCmii`)8DSW4L!dkRJ>+hd3AUb)wC!kaW)}%yI&o2K5+1I_abyR
z;^M%qLH=lg%^*;!Me*;jQ^&vlV@xh(a*y-|f%m<R-baEL2c0+nF0Q|jTRgpda7A&m
zBzv>^wDI;!V!&e1cJ9-g_<Qr2iW`X@nBD;S`;V=TViVtB5T8Ba0YISE(6yEuJjdgP
zvO3TqRi?8cgnV<{<Hd9~TgU?vkJh}8L9<PAx6BW{0#<xaQ=;%&jP(Ql)-%k{U=y9a
zz=P?Ehs#^me+B~neG%}d3x0RKPRLt*J@|@k`Bc91y7AVVd@1;J-*>Fi`nb6KVEY!I
z$vBgD95T_#-eUc1P*UM`>Za4`V&UMm^+j;!d6s6D)AY4e@VQT(TT{px;l01gem^L0
z=P*|t`@9}4x%h|CjDPrYRVNcuP8qrt&Fp?z*R~bhan)wsTJX~&=V$H$T;koIe@+vu
zH=Bd*W-Y5moxiUXPZx>gUY^Eu%!GDRhW__P6o(&(9sc}l_09T<yXDfa^Ga6H+fmk8
zUH%LoL~yV%ak<cRGZnBW6}WUhc=36ih+Z6O<MX#t=*8v0fu;YdQgFZ=<~!Sw+rih7
zy|2A8crC-V(q`yF{coXr!F(-><P-NV43EF(N-izDoC|0_o2rky4)b35Jv(egVRcI>
z`JpjxT)t@~^lCrg*r+;mEsCwn_^|co_<-os>yvHyTcW4!#nzqpxBhPW*8uPGyqTuO
zpWAes9WDL8Gb2wSPCI|AF#E0?7sHbF|5t+UCFK79D^EF|oY4O%LhZr$z?15mt@eVC
zm96(lB3|>piWnX}6B8|D%%CFVfP=6E4D85DN&;w1l8q)0S2%Zo7z`d(Vn=KPRAf2+
z&07lIa-wJPadK28$!$?<#77qV%)xj|h!BPGxs+;+L&#wZjD=XDS@&wE!tpjW>kM4<
zYHa;zgqf8wc3^Z_n%@-x6f=o2f!|b6sMbHBY2h@c{mREgKmeO57TqcGQLU73K-`-o
zZaCBAC*dK84Xns@A|iBYYO{_p*M`;ndaF-jn0!AURA6s&m}^fQ@7T#af|*JQ1B*F|
zBI1>}H1cTSP!z3cCf{jL<yRMRZbQ7#F>fd={EN*bJuLumsd->IP{8lyZPm~P0nAWS
zDVzC+|1H&ZRJhn=e4iS<X%DS#TW&d=R6;DsC?3ie@t4&Q!kJ+0AZ5D{)FSFkitB|I
zYQl*sgoZzk*zA)Y#%a$)H?L-yNmsl$Ho2&vY=-^WlfE>3UIr)eTgJlha+ik2Qn*?A
zkFoS<b|cl%bVnPyOL-))ZJFL5dWUp~WejBcK!#5<%jF9FI#)E2X_rr2xRw*#U~Q61
zojD~pSL(=mRdOav2ND+6>@9pWRagIgKcKS0uvX95{1Fud6G2ME$jV7?mLBZH)aEL1
zo}ZnPq9)-7<TL$Z*frf8<1;2o9QvXaG?(GNu;*L$rDWrgj~Zn((~Bh?!zV~C@3e>~
zX5Vl*jzL8WGk=bFcmB6eks?P4#wlyJ2##qPDa{I|oc@i>EN7ykst*u4&0|VL*Dtt9
zu&UBLp~dLw86gmaq5uLc&IRH^@qhp{S`8{z=<83P5r=|{h!U#nzJee*b;YTvU@+Mj
zQgj7f`-NEwj9TOo-fkC;wcAqoKfM4BoTw3pQ;$SFL)#3X6iz4W>sZnvvdBycz=q;K
z6$s<|k|qsN%4FhHGznTL2+)NS!B{dTHVzktSKx5CiVPe9KSNwr7C!*uV<%vTrZ6&f
zNu#I#zgKdT$cEP*G5`Z858{lS6S+nrES9Mxy@G{I<POj^igN+EL=&U4_}FZA_%yP0
z?}R~tZkUK@vZ*;fxuRIc+zE@H0Ly^wz$oRrkjR~kc1{8e0>(j%;_~iTJCzw=5c^GR
z7mj754-1>Dz1U-=;>fb!&hgoki+9=%Q`h6Ge^v$|&GcE8ezW#EgCAr7e?i2x&f(0g
zJoPBQf~qKCNFkgG;Q3*jJg2V0!Tl&JGmm-tfBVsSpigil$0ri367dzl;vh7>&%)L@
zOI|W#M9!l9q0q4pf+L}{_-XDeJlFLP-&ghHQSQDnaY9i&)@%33p~%Q*y#g+!rCI3!
z4vXq?z^*dnaw6lNsjGn_`*JSB0V+=8NVG8t`I8wjlq&l>*+WelC8RQ8yl@}%D#F_6
z4HeO-SX@RJHZF^SG$Vt`!p@0M(_G-E1R(~=rR7T=36Uc8$Htb$g=dvBVL>$@5LG@w
z7P*H?pJAnr;Q#c@t0^D?93&R7-8t>r+-3)f86YAash>1wZW2Tr^JkfLuIzsNa8KIX
zq0#eH`>B!9T6ebHx|p&U60Y9;)POLuFUzG@<QX<P!&1)01`-WHLbUom09}!oWnn4~
z;43020n8aJJJYTFDNY&Um-Z~b?-y=pdk;FRRiq|Pp`^TokJQ+|2rl@jR6trlF+d6(
z1G9t5g!2Ouq<lvcti*tF0jxu0uMm%wBXPbAnKy412|y$QW>*nmaUn$-B69xU-u`IT
zcIXi?fW$@av51OhRjf;6j=ykr+vo{Z3*M6rm1~SY|NFC|0xnVn0$_-IM)vOH#h4Gy
z6{S?M(6LEi7D?MR^~Lr>ZZa;16R?L5#zht)v=~GHJz67`BGZ^>Um`0iT!_mY&~Da9
z8W^qMz$9urhwxN54y8nR_|mAKA)J_-6Iuf)tq4lOMve|eCSb}cO3Q^1fQot^mxFSx
z?P3-nEiZ@2Q1%|;e9xfr<+Cu-CcJE0DE@z^4)0T_?rG0<r8YcPnM0|6^66lzE4ZsJ
zZtRFx=6rV(rkHRrG0^kd!$QcgkZ7T+X_QAPkdPrKgRTIOWXPdWloT0g645_x(+t$C
ztpj6P7?k$&zIHXW$PYQGu{FlY>;)fV<+Gd<#(gR?E!Mux8H<C%GOjFh0p>Cp8zhZE
zt$JS)tEiN}R?f>)nUgiLYbhIS?`Uzs6>D-b5DOiSc1cJk=!S|fsSS1YX=wFrf1-E!
z-x&(=Ji>?D><kCVkUOFLg$<(FFZP*=#Ybz$85(ids?R9$yx6c+!{_PuAWqSr7i%~k
zz=4b`Gg5k9`%oH!SwsCXM9GKwnmjP=6C*G?m$L6TtPljT0q#Cg{X`KKmqUOA$&R1q
zbV;5RZgvoHaH>gocWr$oWO^p{MA^=N6q)Aze3Xm^35yQpV`#M{BO`mhS^ho+qe41G
z)z2=yLx8Aq$MNm+vHOp=_{6?#R(x?vNhzT1aWDD+fkJ3jXG9CcK1oTFa!n^Dc>dj5
z6F}QPAT1q8ic#_u&G!ltpFZo7GVd)u*%)11Lqymj<gax$9<frO<`;Ektk?jkr0Zwx
z{(D}g&eUo}?kb7?TPuSt{TsKKwWEQ=Z<5A;Z0Ol0N1rWvxDC0noYi>|Fw^U?zFF{I
z?H{(bNB21pkS9uRFaG<NuJitzTSJq_3IPCD(uS4ScYHUqsH(Ur2&M#O-0$NnMOT>R
z#TFkHuGe@0giu-rZnwbS9Fcsf^eyUOz=#^Qy*R0Fpl#+QJaDESd)E;2CewgG)Mp%K
zU}U^}#TSmMAdMA=#zM&q#eH;)2#~5NlVtk2_%8;3N`esZid~tJItETyJOmCEPRf}X
zZ@WLU>~w!U!$2~22(LW^kbMmHSsXb`MkDoFhWhl_NwO|2EdsZz-~0IMVp&ugfEw{w
zTRpWfGh<A1gb-i=6lzwhYQaS;JFz?8chsiL2ItU+MWQ2z_SAgtN+$hA`EPJtY@#tI
zud5g3;@HUSHAAIN^g~w=jks6-U&tO6WipC#KGAq7Tt9=zs9-89UDN=I6MOH4k&!1r
zV{x%+6ZR?*fitcKw0BBvTzJrZ18x<=gbE4E6|sC-FwAStFLh)v)!9)#?dLkS7Ir9O
z4}e20nW3WM{z%1BwN%qty0kY54^-oevWoTi!{Z(s_6~tn5;Ogs5?3O=4_Tp#gpy1>
zXiSKadSC+({?}8>L?!BUW4YYecJuVsj;3D~sCm-pJXMKGTi^G+eEv%+P8o$I>#7MG
zhYUzkU{Xq--H$-^I)c>z6)^hUeXS<0Hwd;|oO(7cPlEblOTc%4@8xR#qVEq^<k^d@
zmHmW!WT_SZy}Ty)A!!vfzq?UAoY14$SdRVm)8k@q*IoDF>voksxrt0h2;*0$g`4Xa
zCo@wsXKr7gaWZj~xj?)o53K_B2S=0wCJ%8a7^bRX6=$DN&Vgd_u5@A2e0<$Cv@e^J
zB1QA5A`aqHXu?v8%wY>oE>q->fPzGO)z_UZiBi0#%d>lrZMI?!ew3#&Yz&B~jCd6*
zr^+za>m~?~)cF>LmrybUZR=0P`cRM=MqYbre}HFB=)X#h%BVM~DSlM+vqlw+{(LIX
zgn1~4_c&w=oDjoh{PFgMnG^4(l|x^d>rlmZ7MSCv+{88=y0YSAJS?S*p$-O@mT-h`
zFPd_?XT0avJr>Kz%)FPk1B-4*khaaJF3VWY*_Q035)KlxOL&4(jY-F4X2#y<vKUi~
z?$w(H2&j0z5xwUMLiEi9qmYIMdfj|v!@&<LkV$|x`OIYW+eI!ABL=;dv%#Pu&oJW$
zCQ?S*`7#PS<^+w4xv0J&_$%nMEJ{GdygDhy%d)V|pAm;<FSX-~E^}&LdPHi&;7<%T
zvPYrS26YnVgAON#3s?RO*hVZ!@hpBFVUAF{98gmv%yDjVhX@hEDwCX^OmV3c2)|xv
z32hJ2y3YP6xA%ZL0aJS%g;Rrr#6`l>v<Vz4)L4`a#SzJRgRNX@NNOWB0Xe@p6J=@q
zwk6-t?u$1XUy9ZZQSlQZAsi(nqV<W5VN%-}45Ni#ChL{~6wB$!JctO(VBea$cU-Ja
zMK9`uPXnkRmB`>>rv2~F3TmbNjW3Z}geB_p#ZRYgOZWxG(EE&#EBi~^r^+YtZ45;h
z)^1YT8B+WJ%xg3+r=m-}1@yY2^aSogXOjNbB;5o$DXf_=$}$#bY=N>%Ffs}{)E?XJ
z__%w1e(vA)ydqM)&(0=&S*R;>`nelr$a#bV%P?wD??IXQ_dOZ&9m<#)&)BTe?RgI2
zo1c5?n#~0m4oaxK80$g3_++sM;QZmwYcEC1i|jZipxAHlh?^6H&4R&&c_qU#ORS7}
z($8ilWk|#f7?Z@rl<eSxLxLQ*Oa_!55t!$<HH)ZlIEO7+LB)MS&%LjluOGhbeSNXV
zXjxXoCnNk@`YH#R%!uO(hx6IK$Sh1}IdI`AEX>E?p1I|_WR;Rpif%uUmT>#CTa%3_
zzVRU)ZmoSUc)gp^`>d{eQjf`euIT66y#OSa5#h}7JGBA0o7kA)>o);1iN>fLd)9@#
zVKvQE9f_g92Z}7v-U882KPDt_x|QiqZVoeGQ|HHLq50$Uk=mH40<$p67bAgLxkc-Y
z!4_;7(@+EX4Yud^TZTW~vnaLy@3~v1g6H{b()UXY6*g!NV}%hw46Lkt=-t3zpCD{F
zoBz#XLhk%UuDjFNYYQ`L?`JtqkPOEbhG(o@G<`S(K2qJNU;6C{YawH4RV;cKX>71)
zqZdDvp)cw0uXQ|Z<2BBWHkn;pH|K&i^+C{>jT99ef6h*Hce8D-bANKEb`<sbtFc+%
z5W!u;+arHMO|b{5XU68;>~IF?R3Ja~FC@-}{r-Ir^Q8zIUI}JE?gxbpI`pBR!1-E3
zt1ZNsHqOa=Jkf6)rNza$(P(_i1WGG3J29e9jpD$EF_AJ1Hc?c(=vDevnG`17myDsK
z1Tso*)w*-@%(s(&^TH##0U8`%X2f#Pp<D&DHsEo!FxBG+Pr%U6K47|VBgf30r5R;$
z90w2%!a+crLGBJNJa%C9QU^64DPuwMgcyV;W_!Q8(t;60ih7LJ-tX7ys$c+7MHJXO
z;LAK(;rK$Apu8gG58-#TbjN5h9f(>z4c4wnevz6-+u8~9uojV>$*<$Z9M5o4X-udT
zC~b~CLOg}$wraQZbaco?%$r0Oj!K|XF$DC98t@X3k?e@t4e`AN_Zpa^#Z<MN<x$cP
z`5dzMCdcm0n(RNZ1;2Hezd|IN>tig^A4Qy>Wl^(5zolj~wV8fX{bC2BxR`jmUBv<w
zo@q#Qt7TAmobkcqkkr2=6EyVYWc_5_Yua>{yP`tvj<C-MQl(VQfjYVAR*7kb*<gBQ
zBp9WR(tR(<M{%p|kmWXyYWR6MpF|A6nI6KXI%B6#<#Z#S4@DGt>2C(vakKZ>+VCzo
z_ueG!XLwmW?oMdG{rMZSSsvq1AFL=Al5>|Ma61~kF<AlHW&w)D!^=Q+TcXU;`GzyK
zqFU(0_|jp2ukpt;%7)b25W~tkEuA-f-GvUByx>Etg!TDECu1WcvvlrC3A6NGz6IRE
z;MR9Fm9<LdEK2vU0GYPG>_))66u~$;GPl3u5sFKUF`6?|dIVEf$oCSeZ$zyOvQuZy
z?cp<hlcd6~;WfVbme=l+*wM3{Fb;MES0%AuLZKQT51e$^B4dil>}zuDa5g)TWll5e
zpi0dNtLycpNFu;Rv+Luf0FdU{tR9~J{3IV067bFfqa%_uO)^WtL%qx4yHEOdrUzQc
z$MS6iq(n`a9U~5{g%(u^b7+<pa&l&{KwlSxnSaSN(c<)^8ENp*D+5+_EWUFVK<&%U
z!v<E5Cf;)fr<WR;7e*(a6h^1a)%q#Y9ZN+L4Wln8U{GDm0k^g^LuN2Sq!2dlT%#aZ
zg-OJPx(()RFk=J(*c~L<RA~lOInIdmE??7sqBeFIrnl)j8GULggA9{odPZC=dr&_H
zB(?CJe3e|`Qg1+gAef0k^~sE1?hr*RRib;`pc|JRLuT2}u-8lw4laMTY9jlD4ZZsb
zS|`VK179a??qwe_f2)oShY3d}=e_^>jEjDHKK))t&J*8p=|@;(->4Q>MI@&Y0fcim
zGSeQ18MTahVTwr^OmR1e-F$hHkY+T?capaJR+=8A9hu>r79{f0rpvvl&1ZYOj<m2R
z8PqQ`9#KPHOFoaV8?2_6_~3OLa)gymODppc(l}8ipLI!hq5$<hqcUJf#U(S&x-tCG
z^%6$eA=7MIwd1;(*ctOno@hjZi8-!wNO_Aei$H8f#uyD5B<NWy8S5wtmB8hEj~WmW
zWv^74WDt)W!f(NcUcw?=0I>I-6@^g6NW@4Sx;~}Jymq~C;bSEp(WEtA>%*l4S5-0K
zQUSpoi2Kq$=kpN9LHhcJa(<h~no=FkYo;^!GgOB%NO2M)i3-;IMZ}4SbzD&(8;l;p
zfK!~^fw><IeKHGiM0vTPA$)qmRJD@q^<Syx6Ech%C}1(QCuB&1NjIt5+!6Mubn+>r
z?m=0KiF=u(MAfk_W%2Q<f{)h1P}CduPt%Vd)*N3pM<1_DM3X>YFx`1?rSeHBxH~%C
zZBatunfwO#A^O$W0^J6)W)kP}PYg`W*zi}Dn+tbr7Yd5h4;;fnVUvRC*9Y2?q2IOe
zXHUGa2CK+nVHOC>V={W=%L}k0g~jWI-@JE<7p7_v9QoC%<G^GuRxdNKcr_h|*+lB<
zivMQhvC}RMDXi(+LZ=5xlUKEVq`ar_86SOE$S3ig1YypU_w)H9Kz*VGlR{^Z8(W7z
zDLz)#>MBNdCBagT&o;ucR7mf078Kd*ShK^ulV*MYbv@=N@l(ljGP0`m!ogrE4Bex8
zMs^9;p&l;i0cjGfh<Q*k%;1}N!ufjO$iHXSa$pJa;%a6I#+2+(Za7}gYaPLeEeFAc
zrTzC95@w~cI2bqp;bcsS!{f(~Z-|M@NT1B0V*c*%Pg{<$@5^#)^Z@o~A04vr3??pj
z?(G?{m~@Z9G7}dqu=hVD5_XfycXN~=V-`t8n~bx}lg#O5;)FGsKNn4X=7%%GO!A$m
zr97IbWy3~)shtxKho^D~=!qF}`6kuNom#`fzD`D4*x$irVqzb6qXctFg$v8pZf8~(
zdvbJUAl=;V%7FUMS2Xs%NwCLXeVb$ARmr1ck-e%-WPWb?!x<nG9E&}edAvmI%QP7*
zc9>1ysCZhMgYuX5^0QZNVy0{`wtUJ}Tin0O{#!FP7R_z0L^>@Mf8*hJJFjCD&K8aS
zu62hDg)DfAnTrd@7Q~IWR8;8f9zMfm1it)GnVOjSk8QY;f<L_CSGz7tmfVtkr^d-t
z!g=3(Sd!C={b^Hp;pT|Y$Vpcip;<aPKI3chUiE#g=!1NQjd_#Tc^W8v?&Yh*+IOg3
zK#!(>zV|h;wcXzWokX~V*uQC?QoU}(yg*^0!pM{x#&(NA<yZbhh}GK+9pj5-*>8t$
zMn9mnk{vV4X8m#<^KiQ|#Z3iv%YPem0zyL-BJZ22Ye`Y`%71#RHsPBelIw|<n`1+s
zA$)tt?72Y3gL>(XP5+HGIK)3=cPUdV%$E}T@6K1g7OatB&`z}y!#h48n#YDrrAc==
ze{>QD#*`;~jXtJTwqqyeB20wCv+`h>!$InqMg>U~4E7(9qvHkk>2>$56v-;O^Oe73
z+&#&>klrRofFxoFiuk<5%8FuBQbfHq!sAF8XbVdLMb4nN{2Y9*IT-E>ms7Fxh^UE1
zrO6aCSE)@GoE}&fugxG)2p@29G>QvE%B6;{|0t#@#JoSakd|>|n%Jc4=M~VgqMvdI
z+aB=$c-rNeC`VRJSp3<Pxw^o}A!pO`yNzyyL(%Tms6+Q7#MHQ0@`Qhg55f0;>uk*|
z%ZpPg3iAtW!5inbnrj@t7~d|Q4cMr>Qlr0j-=+NyIG@=*yO2rcfVE6&0!L%7wQH>c
z$2hzzh>PWezu30J4NJdX#}9lp3LlP8!9?FtehI6f8!JucCVv#plgb*CBw5-<cE`PW
z#s^6LC*j=XU?w5KUi$w1bbqZ$2r5Z~^PbKWDf9efs)l-&bCC&e%ZQP9kX!0+rV;Mi
z#)>5FJoqQ@(XGJ+PkX9^*^Bwi194G|3I^q%X6$HeY!bgT<^rRFNZ^@Okkvw^Ff15y
z83Ou5ouBw=-J02%qe|=(owokcthjEb8RKzYTlP{g@q?sHdCWt&2)y!f+-E+As4G!P
zNl{7J-hhr%*2PH<NzQa|R<U|j!$sc5sLa<)l#q7G6lKdYKFtD1c$X(iZx-;y1fah}
zc8|U^BW8Nwdm$f$0w@KG0;*Q7?p(H2(>}I-Wgn7xD}M1&KDRy1TZ$!)M!c=F^P4Gg
z5_|{Cwj;0*iAzz%E&)#-Si)jkz%o=&E^>LGr#p<n6nQ^FC4aj*C*)qFQg$juZCe$c
z%ZPrx37<tc42Hn4c<NH<JaELA>WqB$EmqL!MI`yjrTv6<JA|pyK8B<~(-}+HHQC?b
z6B0Oq+%PaBmRqnbnIA}wjrZhVrK|kl7`XW?tzp9E`Ja<2@Rh9!sGoIAo-XQs0#_K@
zBzdcvNkhJ^4Kl7vb!qAO&xHa#jXRG{5fO7T?Cj={*f&ST64XyD6*Mt_)wd141y}cG
zy_r^fF;d1$y&Td`T^O`-b?;Ai)61{Hl7XMrQ(XZMsnm<mrMAYGr`Jmi-KrVCKgvIy
zZ)n<=fWkF5kl?z;jKRMz&pd5y9d-rAYBOI%B2z)1AKZO`Rc($5s<H?DfC;qUhhK#l
zZ`_3$k;2V)-!!L91yB6Km_0RLraTpSuQZtOQq|lQV29gUYY6KV2(HyM&NcwZa`DWz
z1aPzWM}}C9u(6^&)o+Q%j*PC^K~+?+iuN?S#usOJTr9byjG3}f(2o$p-KF+6)h7M8
z#_>^GJp52oSlraWUhdrl4wuxBrYb8+DFR=6na9|iyIoE(xZ#D#;_=iIhf19455sZu
zzFvujO@90MfVK=CO229SxStwvcd^JB+H%Z}?UH>O@89Fwr@K`2exmzXKU%Z1cXbv1
zZvI}Vsu?|HskKpBBh^CE#7ZKdBC~O96kJQA^?qf|PF=6GpwRRXXR{&4u$q}Scf`rR
z;fCfJ=NdV<G>9<b0XWZihz}Qut({#;Yc$bv@I4pXRCAK`LJI9F;xCA~%3Tp1<T^!h
zk|R5r@3-YZw#WB@Z(bipcgBD8l@P|KwU$-H#7JWyNxkewizkrYp2Y?W6CX-sy84UK
ztJ1ySN?=}LV0vc`2h^6y02k=*gp&l}5d2^0sGGBVI>tKAf&Ddoaru~!Q#8QK^hp+}
z*+m|1{!SFCsq&DSdSPjiU+8E~p1YIF=JPih%;~!hTS|{m(tw1U|MqU%v)jMDJS6_1
z(m@3L?kM18<9=whWwG-6S-_+0oHOv?^^<o`fPzet0x}~b`SPaGpV(A+er8#2MRkj5
zgN{z-FE25e7z2Qv*+4XZpg~ua3jKJM`!GlT)YLx8{Pf7b%&9YFYI<|v&ezSG(_$dP
z3}Q?2ay8ug=fQ^5)<uunZ{2rqX>AtOueICSd{T~v%4tTXL<VeOR6=@G6jMingc*!g
z?b;6L_KV8ON*q9-tX7!5J+hZhBhF3Z>$K&VSB<Iu2&sr_t+tHzEpb)4Ym6tECLSMd
z!Qpg5C5x6zbZMuQPeU$#|N6oEa91kL9j>klibC<Zu2LY$O7cm9^vYS31hHw3=tuzd
z3K1YK9@%3U+q)<)m|s2BIV1?U5oFoeHfjVdE4m*LN25lSEu+0YcXU`I$LsSZ>H|HI
zm<PRLY@lWegUnJH)?dEKKS|ZYO+hi0NzjjpJ$m#{$4{?wXP#uJycyAZ<P_d(>rxtK
zb3lO%)4xkaxJv4(ge+@$@kir_cDjn&*?Y=J_YaIcJIhJD+fJj9b_>y`Vy1sy=4n<D
zY%-mO4+e9dARp<$gjZX3>pm?fTHx3z3o+1&Mv}a(3O2e{<65A9a@>a&MDN=#68%UD
zZV<Vl+B-rQb#y}o2#P&D$<pJLqm!QZlg(sn*i$gP|MLmZLH|UOJh~z78mu1ly0(Un
zD(v^s&#&)R*roTH3EU^vlIHis9|K6sc#p?UqyRF)lk`*a!AaZn--<8$1qIAtA`H&M
zuVty`U}dMg(3gPxd6v#DF+G+NK&nlD@`8Sw=Hhzl#V`+?q1;5ik6D#)N0B#nfNf7Y
zoja((x;~#@uRrYFUfLTdjjE|%6}96l*<|x?qqI<c5y0l$5Fp0CGQ9Gj&VqYR=gZR8
zf%oV3i%9pW-95#C(61kblu>2`5}10(BQ%%t$(}etq?X3i7{i{<&~E`nG8NYcA^KDg
z7IR`hU6j)_<sAoQouEpR1~>NipQDgSe`h5T9iewje=cl<eX1g8q%Ezq{Y*P<uU}F2
zKDsBb$U^Zs@C_c<VA97z(9HsL_s)*#|EG_;TZjGY=)jzk?4K{4tq3&(pqR%R9OkwI
zA4%PzNzlg!uyFR#&?<*2AJzfqDI$*6Y!y=HnF8HtF280OODoE0rC-<~WC9l!lKk%*
z5|$bV4>O}$MXIFQ=1@#~`iP6~KV4ZWZ|vGb_LFmale|K?njjTCL+%VyTpq2MYgzfh
z_)AA#=6T&mud=mQGDY<UVH#~e3{JmPxR_q749u{9<LS2czNBfnR}R7-=)A>j^-7Fj
z@Cc#I4&oy4+{^8Tf!|7yr#^dlkpFLP;^ZS|=wW%gr1vdevLVm*j#5i}xlPHRIy)lk
zVRYx;vC6CJr`PW>Hy2=ph4FHF%;)X+Wv$I`p8wf1#4MHl&~VLGlfazDCQQ~RihA~o
zsY|sO(NRP)e2Um(oH;aO6@C97E<Us*a_^|nM_h?*KH=toN~BbQg<|_ux^eu~nZQ3&
z4yQ`CsxRQSch5o+7hh*%ln69DbJoH?1l_ql6p|J-ZyK6^r@APZVY^H4`2(}-pM#j+
zyEVb(`%PVfY{jHN#$7KANxi2TBd-qXRPG)HiI#LPcumOCPosVH3-a%`XGiu`9}k#>
z{|f=j*0bGv9#Y%y&)gd&3t~s}gUXNjek`1BH`z<GeAK97d9x5p)Kl+7ij@<a>1nMv
zUp5&OjB+V~AR{1lXP%(0BlS0zBu1I+aDF0HluN}=x9DXkeq(^S8>SbjCMBYw3Qq5F
z#Fj$K=PfG`HIn38GFm3nYpn^?Z%C2+o}K(lvRse!BQ;C(_3w6vsTZfe@Co+u36aX1
z(EsAb^w>73JyFt*@W<~-LGfMrIpfv$G`#^2y_^hopO+hbUsqkE8P#i(Wq{{6kwmt&
zO<xZac9VUQj*{HRjIGYJnj^5vY4COxN!KwH1AEv*k&)l(_Jso!qhO>1Od+xNHxBec
z_iM0A>g^5FC2#o|2^#|_&<32@YbPc7=iKAzg@LSOI}>S0a2h&TvhF|79T<bjye*6l
z1%3rlLB#+l(Rzs5uXaMaU${(Js{W`MGIv%j0m+naMMaoIv*!OMSyjUx=<BHSKKRqI
z;G(rO4L=<2rYI;L)ifV{P?vPQ(Q?fF+8g5|UR_g4(SG5am0p4e?nWrq_<Jb^mb0#Y
zC*V`o61(Sh?i6MNEsZ;dj_ejT7>gQ6u2I#t#>joQWk}D<aqas*;eey<8pLBnu%vO$
zl^>%%il5YN5!aw`eS;f+e~i36mFrd2X<W9+1|O_+fAL_SBnAkvoJZywA%v81MhPyr
zhz~5xmi7CI&m$_Uo(4Z4)RRvynRs5VPKVa{H8;i?PB_%V0{O=NRjl=Vx$1$7%K=9T
z64u&UA(&6f)&+r~S!GBHaB1dN;-@dElrR7Z{y`egmfM1wqXZ(!OAz1mNumA$L_qm3
zVm^vJTXoAAFBwb9(YZJD_KX^=BT{;d`DED711DG9>~{Qq8Ku4Fv*(F@_ce4I=A5PE
zf+g5^{$2*K7Xn*4TfZh*1t$UWVOZFb=s1!rJl8x+<Lz4+v6UKB8z)WuR?>12rU?fL
zIQc+f^NE;EMskO=&xU|aj}Hiu_{M(W2|+04`j)A(&J4-9JSe3jV1uO8$w?%fzdR_h
zc>s?R3m1OpL_)KrO#d=lu20@HIK1GgT&sZK(dPBy%^mh5?B-{$L4|Jl<sTl)4c)H~
zLvVxlFkdwoLzztvvm>VsRtPuOFb6h5V>Z_CQP^ETN(N~E`R(DuRc<x#?(55;U)^&b
z>jdw@!r#i!!9@ZBT-(coGYI67bZ>{6X*vY0zN0UD<tYUvbv}i>It~bK75(8Au{B+H
zIB*w69j`0(gGwPhF|ppNm4xg5*fX~(qdQ+eU_>6O8v4Cj5j9v~)pg5x$BLO>=#(*H
zJh40h9rQ=E(+0?ULu{r@rL&d{QIUuk+?MUO--wDMPkmqAj*CpQEH$Sx|DIJn_k)W?
z?d=j?oB9>Q-NZ=7k<J*;zheTtR?sJfebTa=Jz^~IqL~oid+QPVm{ngAJ8-|h6EUFl
z=Kwkv9SGhMAs?cF>Pzv7VJv&^lxJ34uq*GMdM9BWou4~fk%LkoI5B05rs=)*wo>v~
z9KuAr1P-ymB8i9)J+QFk*yKAIU6Y)}a|rj5LJn}aDURzMZv;($`F1_)#B<?QnPJZ#
z72A$KSwTvKtV4^vxf->!AyGq&VF_K(PcHX-B@yQ}{Z;Vjv@L`%vtU>lm3lYrNWPLh
z<jc&H69pH73bv2)iOnD{fTKP+e<%iZ?yV5OBi37gp>J<Z_ZFiFqQ%s!u}6_mv)k>i
zEkEdD#Z7Xy-sh;UYI8@qO{Sb}ZF!ob%VXwbHp~mOcm7S4_WEBb1pVD?{AbMZuZjFf
z@nr<%%a2ocUbe{-0yaST#$t$j&Entda_t}3XGf8>|Ds<7?YciCS85C>ZTZ~Z!dhIM
ztJ(vfYIvVXY*BR1T0D-qKJ-R=bwl8Vdr8{U?>aa(*7p>Mu#qBmyKZq+k=YjL(Yx72
z-{re0^7iiO8sTTwOIK3^N;D~R=K9j2D;K%N5CM2v0@fUFGMztc9r<;<Y@3}hEI43o
zCoAxe$V{n+)gu;?8WZ<b#VDOGI6AiM+`Y76#`EqgcZDO;d=5;#Y*|Gb0kF90D`RD4
zRb{suR5r}~SnO7L?CQddIfyp`P#>XXbBB$iC?VsH*8-~?pH}cb_b<e1#gJeTLKBr&
z3^+?tXe%=e)!b4p8<Zg(nUbugcStec_WR>fPV$}1@I`xtiOtGVITU-iN=9~c-4C)d
z<SHOW8G+=n!zec`2Y+)%V`)Kh>a56K4(w7Gi=J>9jjtJ9cJ?{4yho(Dh!93wsLX`~
zRSq0fh^t}Ea4IOS(JwDqqY+W72x~1z7RM|gIl7kZV_1CHH~h`xPB$xs<cKM%i-S=?
zQ`88>#fXg~bTl;72r`Jh9^-CTN<)Z}_9a2bl`*aNw*mn{tH<*!r}X;go;V7ZH3GVi
z#76m~;TBFHvU({0uie!}h^J@+Uw^Hv50vfirKb$P9)B(B&)8UE3{_Q~&wq)8l{)eV
z#}5x0zI=B4{mJoF_QO4M<$L_*@kbOGPnZ|y=P|(!f|kjm3&MBUG+nXi-@1pl`F@nT
zKJl36c(L_DmD^p1uB{JCN$;2G{`7ks<SE|UrO4~>a`8ZuU5%uP27AMdE0@GgjeUCv
z7*tUf@ou^E$b`A?5q~*5_MU17|0=&s#AS0ixD(Q=ZhRUkNpm!t%9}IKlx7rRL^(mB
z7Q2FOMvG*hHVQY};=dmv#;MKhEw5r&ImIzqb9uo4!=RTL9Xu{@==Qgb;w#as;8R^$
zq@k<TR#a5LW=Iu+@RG=$cUHzkj}0z?dnc6Dncjggm@C6E@*7qqajWhdy|LLXTulpR
z-a7TFF5P^Y_|nk(<?lSRgW-cF)375xy@X;vMfD}Qv6zcT-dCIopRaeQAsx!8&S~(a
z-mcLP`~@vpE~1?E5Td1+SUvXA??;8KEc_WMA9lSJBHqfWfLJi-tf=PXvR1R~C*+fa
zgm+M1ARrW1dHXJ#y5X*;lqKy)!!HOOl`2zL9BS{u4#QfQ5~R=4^+#{mev6)RvBKw2
zalA|Llpesh5pS12R7O31lfB)f)jayXHeIEP7OQpT-MHLCmJC9%&PD*L#boCy(OQos
z0n&OKEMiYh7!!aLd4)t^7l0+WjpHlEW|fgf5DDcT)7EOb&+G%pBG#z>q!5<E`k8{|
zfFcHD!ud>+PFl!6D$DA`4_YV>HSLWsESq<Q;jnGc|Li0Fu^TmC+rOzT4m`PGez@`V
zw+BawyhPC}MTWn=^556*TyG82%kL0WWh?F^C=dr3V+0PvY{JH#Ifr5b6z;Xj{@dYw
z`fvE%XNJhhli{FNAz1N}TsPB*f!m?p3!Z_4`zk2~GT~u_QEdx$06WSy41%9b7{RrK
zVK!UKGH`w4=vh=`N*oQBNK+qQMoA^Q#mK1F_$W?-wmvt=8QqIuw;O1wvE@INtDfS^
zex{X$=|m>OaSVXI#wCNK0K8?wD4?Lh7Z6As6PEklSh;a*>AKX+BEb4g#ZU?kvdtK1
z_;fded-2oF8}99sPs=Up5<WrAe=7<w<z@APBV2w4e8I?8b5~eWq&iI`aW`QM5fYlp
zsTT%`*8_pha8l#jGmsynhBi@MCXB~pQJt6v`G1oeUfk?|jUUMtwd20%?^S6U;o-f>
zJD#}YafeRa*V}=q0~GGfadj4vyj!Rpunw3eZ7&sYuoEImo_(1|zaJf~Y^fTjp79>6
z%1;EXv#b0_9408DLs#qYN&uw4=_sIFZj(T<6J2i9sR1o6WK0%l4Ht;w#9GvMATjJ2
z1<VYs5{HQ>Z;H4NCjWJl!+H$gE^NJ?aC*+l$Pql&Fu{xIA5!0cNlJ<P&>x1N265^|
ze}X{RQGG1sKlmevaxDe{e*Nf64q7ua8B+$TNJ6~(OU)z(&>w-te3gX~!(P}l)%YDe
zE`4f4;6!-GGva_9w39b>Dlh`+!Vu~y$xcGv6Z_Fv(p0g2Bu2*qD7g6>{T+QJQI9td
z3RiT0S>eWlQ*u0r&R_NtAKt7|x7}``SAOBw(h1$dppHMkb`AGK&;2@s+btD4Z{+pQ
zSl8zXUOQ)X{5u&*;Etkxe#$ysQ$n5MqDx4^{3m(!jt8NT&{{suPkEBU6~i@^`^59<
z_oQ}9xO6S8-1MT;0z0KQ12LdZos<6vv?z&3y>AFNt$47ZP{niUS`?|<K0#mp?m8-@
z@W@hTOVE{i$b*AgOxf-u%h_@UpW=K1&=T6Ov-ah&ws|hm-y724eXCv!?@zD2wgu(!
z8Bnv8CP4&I40jB8lT^RAV7}S<&URU)v)bfNupTUL|G)#&Pk+39>OREVZGAWKJ*@XN
z*J1Xz^d$l}*gy2i!<F{MS@-paH?>Uv@PuOY_$t#okNeF^4CQsIP%T4-s3+L@Q<>^b
z>J<H7*giBH-OiFhTU3zX>?C`57@u^iI*MyN3F!HO@RwAf+=o?;W5OVJ>w)ZwfOGep
z9bqb2Fcew<H4u@JjR>H8e(J5@g;bpQE<=^+hvI|DNxvls1F`qCODaR^{bick+Cl@W
z-*bv`{RwF8yg2&nYgc@A)pmNUFxuuk>+KhUJUjjA+uG)18^%7nBd<jeUJP8Y1!g&U
zChHG4A5o_8O8`w6n=WKHM4XoxE$nF|NkU0M3Ex?QMnn^Qsto$|wT<NT%U?@q^3lsA
z87!nZ!h$CS3qfOnHBUDLV;+6Z^9}_*7M{f{-ky64$9_twQN!|6#d}Q~9o(O*#F&@^
zuBx0mWK(%%7Xz-wHs6J`-vzZUt|v}>7&>&E3<=uQmC(iDeWY0%mEQ&-o_F_;;@@G8
z+vz8$?*^WB*;T3>VA}r;hSoVN*U_I{UcT6^iNPT*`aTd08M*R1_fLsUm%>{r7mFRW
z(uiU5Y&6dP5Lvi?Y_{g*Kb_!C8rjpT5ND^Y&5<*q%8;~}+RQ5_(M_$t1m7JVF14qn
z7WfwamL<jLTIfFI0XIAU*v_+XwI#h8EhT^XZadSC)?rKozYo3(^ckNHeW|Kaq2pkk
zq7Y(z^$3t+!S`5>(xNBsDGdSXbyP@>0<QXlZU%KrO;^0Rv?K6%{pRLojR*5z?$NT|
zR+WJ3S)1xB`LM+KERlP3@;7!UlKhy7uA8T*E-c3IHL<dKNdVXet6>;b39*N;6MUk1
zt~(G?rC$6?#m{rkc2!ATl{Bz9t4$gXa1=rBQX^DtE2xM+ID}caYS66JQoM<8ShhQ`
zx?7OeqYs)Sh=8bK`=PCV#7Ie_ce(*(%R2az1o}s=?DylV9kp-Ly(EhbxjIp?i*GlM
zZ!o*n+a}d#zHm~s!+=S(_4zV1B{n_Cddmmz<t|srrBI1n>I^-frGH8T@*4cO%pA!9
z(lp9%`S188alZMkFT-sZ<7bap8DzL4Ouc}36<Zos7X2#>R-uOa1tmbULM<X7ovlo7
zH^>Oa*qNDT2%(nU-_3KBVhrEtCH^Zl$E*b=El)gDQ}RiXnb>c%uAZhriqgI0!#$rQ
zf#E;PW!fH0Z??zM>P0ZwO;r^>-bp2)tK}ld9Tr<ZSXu2folKd<`1hIPzl^aQr26RT
z-yD9G2dWvdlevcC8QLcNZU0s&;t$(22xGFbtU&Q|qDYqV)~5HRj`T2+|GcvMFp~a!
z=)=VC1OfA0j-ExmPLAS|nvG=7MWn8jT4g!8*C~OjhEIphUcpnEKLp=9nD(TT_wkgb
z3suZzk*Zg=C&g%Z{F<Tm;UTF>N3!v$)Qay882jXv@(Zx^B4*6DgRGw>ajwfqM&P1t
z9oBnX9|<c<A-_~n+hAcQyWgs(Ia2m`Z{{Y~)~qk-cC_V$my(I8zR?SSI#nsIYF5Qd
z=7G0AH9>-S3lP2OtWD3{T7^(p+Ne0s{qQK==k9L>v_SkAr~L<@Q5vuQQpPU1zdx#S
zw=2C^;sae@<o)-vY}<#wFM_=m`Yf`Nhkg8xvN;k}wN5<B7?F`(AkfSSaEcUK<TJN&
zK1q#K=nIYgIA#pqYK9BBx|9-^LT%ZAE%z>v<|)@~#=z3jXynCq*_(-YM^~o1&0)@T
z`WF^9BwuVX?g-0~rS@eyYr1pn>7qvF`(8-wMz@NRo4}~T{lTiX7w<}L(06YRlL|1h
z%AF6uNadZG+3ld8OH-4``M<@%ZZ~TJ6DE4*V!)17X3GOSo|lE#_VHh=AphO?pW#T^
z$37&}wne?XWvN#AvW>5aU&KTUJ=gnQ$6E5U{nbVWrrjjAcJG+q-8}WqYH`bo5uNvh
zR))rFuL2A^9181*qugFTnV%{>mH5roBHxzaX{k<nC>h<x3*){hKj=#neLtgZbQfHA
zHizotj5g11f;&3L;}LlW(x&%a_yYn0rp36mPXpzsp8OtcY~1`d@crv6_e}t<oROv+
z7r3*5ajb*L<m%k1UfZ~ue~1$Ie|H%s;7=CX0B$ZiPf@WqVNTL@U%<wkP=B0u^iliK
zC<^Yvb(zL=@Vt?d8@1|XT<jSOmL#JopYEFuE4`!Hk3*37Z-#%sm19Sg)a%Tn^-1Z~
zhB+>9J$mzgN`+n@P%Ixye{Bioi0r|Xl8;~*q*R79N{T>Ar0dt?QSzQ#I0MUYBGqGZ
z$Z3ZdBrX}a`)nzF?y}Ah2jiV|fBrS-Ya`7^8Z8JOeA7lmU{gxL7Cz)0{_WN*GD3QD
zHJD7<0Y%gcQFG|&F;v)lOY8bXO`?hYr~~e_Q@ljjMezFFF9O^oeFnMQn=Mfb(HbPi
z<jhp_=UfDioyCXZF!Zk5Z71+qp+HFQkd8!-rASYpUd8eeq?pMG+RUj5IK%@YGEyA-
z8RPz>X-s!0&wj^T$==?UvtsBoPKAnEc>nyz?2!_H`AlG?R;24%KWzTr<zs;TnWKYK
z7vB)D!$uGp2^}W|(*PR?Em}`v!n+s{)YOhn1N`pJoo)j81b<nGhmED#3B#{^G#0mp
z`y{COV76`Sg$}T?^m{dJ2llBhdoRKEf-zZ^jg~Tk%gQu^k5^;sTTC#&IOlQmPC$E_
z?)>p<R@isjLe`Xeqe|A_i5HFYqk$(89{1Bq4tj*vl3fFAt_V=5Il@xstZhtCfQ>Ug
zYle-ZPf%GM$}!p;RClB|OS6WhT_8by+DCcXpR<4Lae}8_tcC_Le*3PrtI1DTP6B(4
zsBdX?1?TC!pHR4RGo#ad-1vN(Qt?6-+T&qbYizLr^Q|uucc!{$<aBR`xOT~=g%S)v
z2D@lK#Kij-&YnEl+%pvyaK{|D+Ym#0M&WXX4ENQEM3KELmrPwtkHN{<IkRdl%a}Oj
zE(QRJRTfG>sRl}EshV1Nj1+$5Xo#(~u+XA4i8%Q6-6{f71hLoCS*`h}edmc31i|ox
zPtLosE_2K!e@ne$Ijh2t*>#4h-NNFs4t0n?L{|SZ=>b<G%E?`VFl`zBlIL$~R-wZ@
zX6G<({C4ba0fN>%_j@qL;Bk3S{0tAy9h-yT<Kl1N&j77mGBfQt&iAcwrlgPLUT#F^
z^N!FYUE~6SJ5qz+0_w-ADHH=2W{F^?FFa5jC|W3N*7)B=Z3w1ynYQ4Tjo`{c(ffES
zm<s2?YUiO*1g^#01Z;=}Vj0KdCn&}nl71AueRuh0ne{L$Zz<eaFCOXe<ue2O^7b0%
zXuD^X>a4cs>~&cQ6z+2eGR?LoNChb8=<acAbo5{BW*)q7@Pug|@DPKHk6A*n6}G3%
zcb@JwdNkeqU9iO)OWxldYt!5zmN;LXZ_N?ea<p5a33-h<4GlVy9R<-34+?#+?|heh
zok#G&DUvd$f8MmBW2{S~B;UNErJHBG-ibM=<L9RLc6qTP8R1C9ansY~WhU4;9kE2x
zYkqwflF~qqC&;tqY2D`P#mXf=tL4O-?uwvVC9yCDb^dqE&fE4xW{oNNl;MzOThYdS
zqWXilV5wV7SDCuh+>-W^;G(gy^$STj2+ko0x7`{Ehw+gDYXvS<&K8&62kimSuWpfW
zaH&nWQJ2WOs4jp#(t&Jz|D{as#I_!DoFSd}-bli)p77BO&S5|<I|nlD6o`#cl`4R4
z9ngNRzR}d4wHV!(dz}UmBlk)4$65PHH=VXTbi(-VJ_sJRurPm$ZvizP>sM!&x@G@n
z`?wbUQ~M+%9Kj;DD5%w)TklE^vRpCG2&yZVB^!+temF;)MF=V@eNQ9~;R*-{4_wT8
z-6-F=9$SD}sfq;F2nMC0JKvtN-BruYdnk`hE-PyTnh89Kfo7^iPwkJKYzynWu-zXo
zDlr~`2;{5V>rGDGz4QJ@bCpLzuj&;4c}SnfhbXvp)^kp)RX^2PLPoUpEV(IOE>tVt
zoKe2RoaNqLcK$sL8>tTcTYr0c`)4Ke`1IZbnNiJ)nr!BAGn?$b@4<45v$L~>@hW5@
z%S)X)Bk&T3l~RanAr)gfF~p1Q@^#Hyp)HIg_X}!zDx$aZ$3mxd*<S-Kua`HDb9cS}
z?jG^XX9D9%Q*QOn_q6XNKo*O4Q~!1Fn!f$polPGnfxgZw;?hawZGL`sq*S2g1*rq@
zhO*N1b~GAY!PecU!V3R<VJ*?G4XBU0Wbwe2zbs5Q(uEO}kbW!z8Oz<5P&&jR!I*sL
z$Lgn@p~ve{PUaW$*K#epp|A3J6`P388P|SD#SQ9hwDwapYyscS@VgDZ-*n?D*hghq
z)kENq>yjliB6e8B7d-<0890+Ppve~bd--jt6%3i-%AYxiaI_7Omno(}QQO7hI?kql
z`d(bRn{C}!HPx&0<?2e}5{xnhFi*5{lQq_CJ(YcXNz65yGaDT9bX&zhs7+u`(_?AL
zi-8w@yeDmz&QKu~-$;k<$q33L$turmv5=}>M=@v(Mdq<Q1*=i1#pVM6cb{_0+^I>Y
z!f9E?;fDST4AFfHC`(F`_!$yuTjx}bjV^E#))<Ggh0b}<*DILq#=K?2Y^VoRzrC|2
z_NeODv;28^((X4kEip`9VMxg#y*R#xCblTUnuWL51RbMa?uA^~8w5YD`fugsw^OVD
z>NVRThj_=o-5+erm$PlGpQXI#3sphb12&HMsg~)Mjh}9YiZ>It>+ht<REZ6Yy0=%Z
z|Ce`yxx9B9bNd(b7juJot$16H2g+Qo;5ozpB6Jly?7AQt>Ho4x8apD8>LyLo^wlg7
zJYm{7=(#+cosV+?VmnMgR+0WbZQZhlpz*N`4W0OY74=7uR38~cA2pVG{=5Qa-MpE3
zx%{wIb>~i+Q?b@)-nt`tj3(-9L}TE+rtl+TZbw&n8rUmH01qad>~qu9xMTNfU74nq
zboF_vb<6DsS+4#dxjUMi2cSxHG?wiE4VIeGNHb$(U#}CRBUUEEdoEXa_xEOb7E2kn
zjE2`-N9(eJ^XI;cX)ArTV2+L;-pnAy?z9U>5K=!GXR1qc?DTvWBrv5K{N2K1uW4}y
z?+xXw?R;d$GKv9dyBHoGZA6GkSTn@!MK4*foQk5dEBgTNyIkoqGx{+yGcpi*K_M+J
z?ippDxJtFh_nTs}vcsg6lL+{0IP9CtPd{{%_*5ikW>pCJSftOH^R7bRLQxuxFk7T&
zxBe3p`;|S+g_|b{igsc8ZumTJ+cdpnbYF}*6$hpF#SredW&p(T!5Nu(6B35sC_X~@
z=|A;X;RMjM>0WGb5o(`tDq^sfvb-(y(K8B?00fF(d9N~EYJ+6fP6=e=EDxAR%qiK9
z)W8is0D+I*936fZ%lGS4R8sl#RX6lzQ1K70{HZ+Vcy`yMDW(=dE*JQrd&wGD{IDP4
zIs4iV<LvNId;dq#Ihy_5?#bgoQT7)nH*IrGiP>!b4^3wo*7X1N@i7nv5+ejeBql##
zfRwbfbax}&CEX0k(KYE%T4120ba#nJx0EnoAl>)=5AMfY*B)?v&-tA5KIgn&sWt##
z{Y|<bteyFcTQ^CMMG}X>e|gIFzv}nGm56$>{&GFUJ??kjqaQGk#M8&~>5M!5$A4}T
z4}*_iK)ttV>@pEKyZUF94Ur9kFQb7V;;+VY4Nk<Ol9GkAQf|s_-ZXXA1R0VBKpGf}
z#vk0lZf5p$<_0Iblu$qqg6Lb{EZ4Ukv5)4qiyCnp-WRJYvkMH(HXiJkZ;~(<ld4@i
zi7$@f@UL=UW1`ST0Z!k^ES9VQ2b2lU<uwyvogq>+Az7z^xOXq8Czm$IN&x>AFf$Ae
z^x<ZE+PR>!O=)=k1i&oynZFVg?h4b1T16<?K#vGWH*mKAbfkl)*Zcbc`F7FCaVo5w
zO3d_rf)GHQMhZY-g%9$tORKnY;<V8AhsUd!6hY}TnxSc>n=;pdzA+9QED)(HCq$*E
zZ>jwE6n5EOfHg;{py1plced~g92~Et>P&ZKHS!p4PreLEz(Iw@K&6`@JFMs<mhfu<
z<Nc4Ny{hP!YGn>0U@a%ig3(Mas?Ua-A7v^Qsw`%b!G`2A%>fc^TCn!GUXCA4Vl219
z{GWsHhkuJANXr~(lwZc>eV~i_LtN%BUZA$?&_>55mD(c}SZVdOVfK6DY-($jslJwd
ztdkP7q~4Y|w5Od5tl^GifXU#H2}cE(wWtX(O1WJy8hmH?h;>0c{;oiy;=$GZ_N_kP
zO?Ko9KD!>kalzfb&fv}Ppnibw&A(C4zuh@&AvJn2iAPvj|2d%KB>URg{WKUL->Ukn
z0YwdVo^~{S?`c-p(=x=(Yxcz2p4+GID@^19lZ;!w&1U5f^&<&j-(g5+g%HrZMH8>T
zXga^Zcm;DIWQ)}r&q38JFIxn6R^sONQ6adk)yb82RrJMxG2;7d%BD9sI{19yQv5Mo
z-R$A4tdmHdH*PhPggBl=rDQDC>eOYArWfmd+Z7y+o8s>p?GVdtE1+$O?`whkhx=*h
zDX;n_94`Fs6;ZiMI4qs)FCY{_I|Sw}`n{FqS6iI*eD5L1OlHf{wyHvHy583mNkuU0
zW{A|4J4yQJH(%v5aHT%auUI;kf&X)N4jK5_+sjf0!~@y+@l0lHztT;bFHl+%Yn;y}
z?eeD(WI!bI`I`#IPlcu4v_Q$I?DYq6jsN!ERj@&o6cI8-*0A-hCn^L#=|ba%#>a^O
z_1h{~Vu;VujSSAIF1AryE)M$oU1PJY+Us(ff%+cT3F^BdFF9td%y$kBW={9+_RmrN
z@4Hfx!P7s@2IM3$C*DKELmVnN8<Ei0VsUWpxNpZZ8{m&M9U(-f!e@8q8RS7X|Mt1#
z$-N7Um5)5NUcy4d0K(q@$i>RgEwH1LrdOmB_#^kToiS{Y<TUx$&r(R|H=S2s4u?``
z@PkTd2X1dSPxuCxa~+P9eYd~75B%Lr0jf#doedT-5z8S4PVJrsiD*kgZCIZHo>LP=
zS!kSczd<#?L2=2TI*kiqyQ*wqat4tfIdTyt_J&t}-mt$d`zWkx*6)XP^}hc<$H&WT
z{we~Lb7xwU-Is92C%aroZ>%tT$T5DKnSAQU!{4MTCW^B|EOEc>g!lrx&N(%l9IYf1
zA^vxeZM{`Zvt#9B;=o<<M=5)cPKq=%6ZfUQ7bS!)No_@a;uVO#oLgaGEM0)-OT{ra
z07kN|;zCz25_q~cs3q{N=Y3@xt9dF>W^7=2xSTh(cP^Dg-k&=3<MH+A<omU~70u*<
zrQNkp9&bz}&=cT~kq65WHGss@`<<NtnhsX!$%2H2p$k*=s4&2_pc72WAd=U?i)zqI
z0(0gYeBNcUjR1Xp63L|){o@x{EQZ~|0OlJUCT8)gx>c<V59dZG5%K0R)5u9NW}!9r
zSEJhIZzC%U)$ZKat2J6a9FdLQhK7;jLbjlSz*U<mpKq*?)Piie<(625^dNVus{>rY
zc7-FR@cX`ui%wz)AYb4nNPNzS>X0e-->V-Es9n4{TvN#*EUuERkK4&#SK_vwZ^O<F
zuNByvha!eeW1!^zz9igGr^{KP#P2rBA2dJRM^Rk3ushczkdfBx7FL|r(k7MN-jWLy
z>rlBw(KU7G+sarn`ADWn!xsZzk3}SBJ!j-m-Z<S8E!e$&5Tay8`E;~GuC{h=>ZrIF
zxJ7&PgK&YNq}U&mv9MXa34r*oOkm@4jLDK?^`-&vA?V={X=u)-g0pt93a^Sm)M^k0
zKGzm>(w}izHeI;FFtHdwJri(#V4Eo?dAA=|gLAxny89|4Wk{@DiEJ|&F2x-SA_r3L
zu=S{9rU(HUga!sI=5xt3x4=@D)fKbhUA?bZHqvQu{&dp~#LFz$XBD?^!QT0|Nx26P
zrdIzH@wyennG~&B`T3?s_a!N}Cm1C6JlYClkajZHYtXXST#)2jbF_OZag`?FcM$~t
z>75w4r~5JhEvi85#y}bekYpzI9d$PbV0~W=bE19O+Azc2OluWv0RoiE%P;mN;6p-l
z8UCy|rwhim8~YAlT;5%%!oG0;!-3quR(B`ATGFtB&Z~(-`8nim^M-y6iaX?TGfABD
zyb<y4x6^YLieEk+Vi}Od^{f3kec5!IBQe|<jLpz)I0ebmKP3L2%PK2RK7dZQMX>=c
zE)I)xls{)E$dJxbIDLKTqFBp>&x3OoQ=~cJKeYE&@*F<hC{6f5UH=mu<G6KylULDj
z1W+F7t}#k2se0$!_z58Bx^;7Ze}+?nIK384v=S*U)=`^AINIDESdunNyz~{$%c(>$
z)$N9)+hmnOFB{`F2M4l-Ll$+#>G1J=`Y>1=)#=}=gAV(pW6Mj>z@k~SERAvow>gjs
zCOX;6?R6eTojg8(18~Ip?9JNFt}jwBIn5_UKw$<KYcq_zNp29Qg%9Cmz6*UfV?5`d
z)%#IH@7eHy#TkFDo!U#DxuX(XNPgm_)$1Nw*CXK@zRd5JM~fuA+yuI@G&gOtllc<2
zcSTe)pT>WR=?3T@D{?Q)e9Zj!BR&Gl5Z0M!fM@rLky}b8QMUnr%NyN%e%h9sF}vdQ
zl6bCZPH+;wJ&!TRyS_Gi*v42w>33e=oL5fszUXpBW<8gBjz{5h`IV4fBJiRevnG5~
zr;cfdO+IBKPF=;D#PygNa6Wjd2UC*>qS~8QTOEBpbx*y8f*2Uzd}ja$2zZ3Iz@A1Z
zFfxw}^Xd)djx&h%Oh*X4pB<3EX~Y5~&oY8itX?;Tyto;QuF4~oLM>0=f-%ifi2|~u
zIji|e1dsb+*+8VaYR;xxhsb{C)dI=G`u6zv`SsP!TP_=Fg51R!p^5-SI1XmZAu=ba
zGU|;xgZ)VQ&Y2O8K`%R#45%;|ruKv>BNTfW2v)Mihiu_~Cw4r8izx-LKK59#N~}3l
zlniAvY+Zfg5c!gcg$1q!=Kh{3<1*ud-IYCl^DW5u@ZxJ?6Q;14sk3PNw4t}}@|;-I
zG|$M2nJi|&v(t?H_qVnCi?~Xlp#It24uNFKUk~6<ngtgw-w!cxc$`8OZ0YOe#h-<`
zXEn!oAz$2-FgM4TAsKjuqex$HoEQI1^hdB0N|yj&k8&tptUW8?@kAxSJ$U!4J+t10
z{mMU@5_Dm+eNM@p>Izy~W%<MgbT}b*_~=5T^M`4nt3M}s;(g&{9@nUr0X04#N~;x$
zUz6RdJleZ7jCi}i3<*HPdbtl)bcdnY`#+78&9=cD9^wThnCktxBV9!W=P2*2$fRe!
z!cD4IY~a+NAn}^>P*sB0#)-;M-@)52YH;)RPwtD;r45MJh=}{m6F=|F^wZ~p!;>n+
z9VK7n3v~=55)NDRI-QXE`ueuBIY>G@;WvA~%UGnLV~{>zn#wecv1h3kOv(sk`4)yH
zg>U+1ZP^5NcW~oF8+(Zk`Zqj&KCZ#W83m3>t_}iwZsDc*IIMmkR~*kcqN@N<{Gf!b
z1RG*Pb8lG$oV4lX9mwA62NTDX`E#&SL5Z83&g-eQhr^`KyM5om)8^J9jW@<smmkXB
zKAl@-&oG<4nTF9nZc(jWAJGS6<b(B7bntF}oK`IND|aOS7#e-l+nUXVY?{S{^-}YB
ze|HKvBJZaK*AQ8TbQr**Ar;gB(IuS4*QVUq`T6#uwB48A^Mitg--pw@oMT2=9fxde
zzO|eqSk7CU%EzkA^NS&(-cag3fy7PG$%>nUg~7p17%43pofM6F2Mv>~SX|T^v}l$H
zK0m*HIE$lyxW^EtCqpXV=S#GTcitQpJRGY>d~JANsw6Y6uRE)IJ1h#$YJ$!flN47}
z6A<7)mxyG(TEEQs!AW{fS7A?y$qIy((C)j$hJ(|;MjL89!wi~?$yh0R+W>1xK6&ZA
z{VXaNd*E-UX}ia(f;&5j+iN#hSF;b;T}Dw~>8eTQn!H=QX_eY>t=G$~VgQMbvrmEC
z4@Z7@`iI+Zd=6n(*Q0Eqm7puUJ4rwZ;hbR8WsMoRWW@IGhIxae4aHlCpcPx2QV#Oh
zr*>a!r5tKfUECIpX2!zPYeGP@gLGHp@WLRh)CqU#$FbP^)hRMY*Y<@cs^66MN&^ch
zTvN~55`alhe~AH#5#u&Yt;W}c>C?{-ByKQRxl-utZ>&rqvE`%WS;gjvukV|wuxA5t
z3**C^UrGLhWj=>QZE5@?;Cvay^XlmSjC^?E`OV%br^O8ie;F0+ASpv8lk)ZtMd~a=
z7~o_;+a9@MV!9!F?yX+bxpJD*n?~NV6pmdgLOAX9*OZYGH8F<4&!u#~0KbY;Na@|!
zuS4IjUtBacJn1oQ&B=BY3h>wg4^(dKf4JJ(q>{KgXytONHq4N?x;v$sEXXjz<Ss}F
zon?VnN2J<LKHqO^1hD6w6-`)ltC0|_P-8Y&wf*&|=;TV9T2BeiKlJ;x;FE2A*#FS(
zGo4rZop;+c@7p_C-_jZD<4Kn8G?rg*%X14hsC{osydv0m@V&fW?vjQ}v;;G!Sx37`
zM#!T}jiXX}D?>FLpZ@{7j^#e_|DgW*C8I5bM4psGTACGzYfckEX<jkaQdtcP%XT8~
z#RUW(yRr8bF9{vMOq~QzWDp?Rmwn^A*IvP%LMqeuCqfjYUrzTPwojAZcZMWv(YO0u
z<g7jTRJ?N<wOhNLIC(I@$XA%SD=N7AV|L#O3-k=U8mFphj+eMzDll7H4tZ!tll`o0
zogS|<E+qR57B)|v+jB}TzlYHz9Y8Qxk_gl3zLfiiAY5N2X#g2%x$^Y2XdMIT6@L*w
z19f!wD^_p0H}5(==ewx<cR=I$Mw;~#Ru55lXfosRgL9d-+sg07KboZ*7>sS_)$rtz
zt69*^KB{%H<1)HIjY<3RJ;^km>c^Qp1M6Axv~pQefhVLwFU|B%d_rL77{B)3wTBy)
zf)KQg%oW?yGbuIWl|lEm*47t;-<7bibbXuH`m?2FevZgQfb_Me9jF(XZOEw>dm<1T
zy>y>ydMw)QWMbf>wI5QB@%|fK_Rt;VQ{If@&`os${gB#PU|{3OKz<8!f5t_96Z6Zu
z!oArhd#b)714mV+t)&GPX4+@#$@Jsg@$kNDzwwm>X7Bq_0$~YK(8<*0m(@{nRP3oJ
zT|csV$sHIlUWq7asznlpnrl0bYHoh}I&asrbp$H@qT2=32}Hg#+Hi?#O*I@Lc$qYq
zQ`uX~(FPSp`#PAceqZ|`+OzyxsyS3f86c&kxbYxRxn1re0~mnOQl;?AL2R2ln(yl*
zZsobmLOd>S&Vxi&+e5s?>2^<{VID@Vp`Ab4?~lsXt`FW8-0jidx2sD85TBl$CKX7)
zeAs1aX(9JsY(xF;{r(xSJYxZC324Gq>R&HcIc2_5xJcl=oYmD;CC!oro}Ay`Pr~jo
zgz~AR$Nd^iMFD#pPeLcMa5gl!(gDm@vK`*2ekWxh;BZ`Romg~vAQS5$#5%m?<T9J(
z<>S|ox4!Hk+lbVpjww#Zua0t$_<&OC8=^hx!d$oM%7C%qp)nC7++P|&zPkOkbQRez
z$slBaMSc@TuxMGNMES?yUa~HRFhu0-j_M`tsmI|HcG&zssI7`Ax-Ga2TmDB@$mMhf
zW?Hz(VEzzY0DJTj`MO_L5E3AgcQjeH%k}>5Xy#FVA;-ti=l!llAlg@Gcllb+TUQ+b
zkD=yEPL)>tPHtpdD8rtfC*)*k`2E@a$GflAHi{2SgE3^r@#7H)7o-BOBAf{TWRUa{
zY-t^M3oF>myO@Z+T$RR7RoUX?IMXm~+MgA(>kO2*eeR>cY?)23o)I^r%+;1@X@aEm
zJ)(V`QOV@tB^nseO!+1NT{LKqTl7&5d?@vtyL5j^4O*h<uLzHceD<6n&EN}J1{s(U
z_=e12H_%owMzGaZgD3-K6aT#$wYh19yg_Kq!Vo0Uz1x_aYCvt33U820{j4XGIoJ98
zV}o8ag#uJ6uge6vKYQvWFf>{`5lP)Mbx`im+TUXAmR(VScnRASe9R#c{~;>uAJX9D
zx)Q>OPoeu&0dLM@{(P8}GPVLigb#v${LLUc*XlG^LsWdY;}{|;)N)Q>v7-P6zT#en
z8C=l2p4_3zP(m_JaSP5Jl}?eF>{I|!$Y2DZSSM;5iy;zj^tCFfJ!{5%O&S7${>QU0
z(vnGW$j-s4G^TkGNob!m<moCv@|=0n74*hOCe(i+Ge;49U-!^^9M368Pv-U0V7vaM
zu)*JDh8KR^YD10Zxk7G9m)SKO50-MUvGmx208dBc<)NtL`zk{x>w=4Do`^nLA%0U9
zsG(I<_{KQsX|!eVe2;RzCB7~l^kOY__AdU!Bc4aXU!Y;KMyx5im>2K_<o~Sd^lw)W
zkDLypx|R4<-}d;D*Ywny%DhA!WXr}`#JT5WohfryRP7*`@^*-^bURDgxw8kxB3r?F
zU@x>z!JF(BFAarX$dIMzP(*%E;sLM#E!23fe)<l_qK)H88}yXW&jqjdNP0EkjB}Nf
z?fE&^o3{PQ!l;4FZu8K^&{>x`G4alz$KE9IyQ__&L3M%F#mER@!B=L-Kt?4vt*iz%
zn=}&|o7X@$0x`i=lXK|Pv<mZTHF~e?Us6NuF`6nQC0=~$fPA*YN|O8+2&A@x00@cz
zG^$3g3YApG7Rw!l#tesIleM?jg>lGSC>&&gUB46+Jk}w>QVeo$GvFP`_2&yH+<fSX
zV3KHNoPzCtfAa6saeGs<^3XV13J=yk6jp7p$B_6F3=LvZ;_>cq3FhALo&;}mCcs5C
zvfMHjupUmIG{_jS;EW3x?YHI48n-Q5^`)2SuqJL&PimV#Dc4eh$7@BBV0}#B=XpZO
zpaVxt%`_rCNlHM0RG#KPUF%Sb8gmJkA>R!Sd=96kC^mkM0l`xq&CdzhNI{Gc2!+3L
zluVaow?9Cu5gua&`@`heaGEn)gu%fjl;31x_zuSu!BxJ8?SWi6Gswrk`j2UNNpt<j
zTaHtziLSaf!nB`Eo9blsvlS>>U5kcd;hyg??)u|RIi@wwm=r?>zv^aAOcbb;>6BMZ
zQzo}<MV>ujFhnLaJOS{s$Z#t?lhvGi4RN%RVJiNdFf3#a`}e(KMEr{ItG3T7)=;Yq
zX?>*OT<eJ+g(*Yfdicn_xRe}Y)J&JKeCA=&Jf-Zwe}BR9C5VwSC@WJBbs8iRB8aao
z_}ID8vkG?L&Go1=DzB)hb{g|mNP(ZN>gqm`nGi<mY>BO>t$?J_39~&Af`B#ll6o=l
zyrN|+fsQVWy2Klu2g>Y;$ehT?AfXfY3Ko*+1Wl~8dtRboyC_{lEC@qCg$x;~p%#3;
z(PjDwck_Evw|yQg;5tZ4Ay?YM1)aw`z)}<cZ`i3Mvj~;epcjgelO}*VhA}zEE0OU(
zoA6J^AycGaB8@<#LY>bT(|q!`4aEbi9uksN_F(hsRe^>&ocY&hsyw_;*Rfdo4&1-F
z9ecWu`s5q2BjMTt8$gXvL_wU?JWa(&56vgiILuPXhchz%;)rj%zq#M(^^r<biuGYm
zy<-eh!ot}}51tk<ghDWbHnTW#Qimc>?VnFNS6%5Pg8v<-snbaWw-_LU*Ae4^Sfm#`
z@-PmDuo!X~NnFS$Y6S%i9x3$A&r7oDZ*1~K8(w_rjs~nm9E{Aidc_K)WHu}@&OiSf
zYvGPX2+V^2lk78hv524e^A$%WI~1gtGP`Kt;;bPOsAm}>ZHq-AgJmuA04bag>a_dW
zZSc=k-9{PD9W$jQCb9TZ;I!{%RMGbKml3Q^Mlx4TzIBBdCpR1s0AMAcCLCC89qlIr
z812dzO__@>9QtgY7y$H8D{)*MSHmnMRav?DtA^Fpcs2{(uA=$Xb9S3k*1Xk;Nr^T3
zRGTN6_?o)(Vn(9v^KEy3JZ%h#M=D5omMRdT;Zw!Px9K{cBA<SY--Gz8#dO=3-#+~<
zSnc$}{5>+=|93{u2OtkDBq)+q+zZ14-VBe%{M^}G$T`Xym!-_?VQcu}@81yA`N{D2
z*Dl8!X@eQQYVyH#*6W4TiP6*BmWD(ep>@`lt=T$0{@$(>e5ldCK8TjfkbmoyX8te1
z8f85OF&~AnLPGXintR=Bxd0y~KH0?JReK#-BYB{T;z&2Ovo_fHXo6h_+eMkVCF%D?
zFjSg`7?($efsh9T3isQM=rYQ&NHeJo{zq^zG0==XvYLi>P^f@Mp83LBLccPnHslS#
z)|#Cy6K(b^VV|ApkAq%yclACMXc6H87pI3bvkYJ4xILdlQMu3o<C6wp7i)Y^40&pY
z2}kE;J39DmItdm!bo={q!hWUy+kIz;ArV{Ac;56<+Mb_4j>@2wTGc0@jB$FFi<HdW
zScp&wAC1~#wOro;_a|YAMe4N8-*4InFbL}wFxh9&3*;$kREAlmrU47Ed;fuND|734
zMpF*^hFu2EowKXwgfF?p!rN!GT^slE7M*B$p*#OI*<G0ZhX{&rg++M{o-zvU_%X#P
zdxK}Q_DF>koZ(f$g;ae6#K!9b56>IAJ|2cKGVO~Y(nuwMAJ^>o^4qbcSHE#0AYdyI
zj$y62Tiy1+0RQu?gT3uP0ZV+0_y?x{g}!-%jRp7kR}NCJW&+Cs5i&t4v2cWi8tU-k
zWT8Juq!kwIrCeaIOUApKP!QHuE^a^ovDQl9pQWyBuwlZ|Qp%9UI9W!ZoXrOqLS>2@
zz7J4#&wu7u8ZivaUAXeIl8Z1SZ2p+5iVgTo)c2z^+U>)b-BXbu%J$|jKer|5Tm=N8
zMrscN2MLw1SOAQq4ln4#lf#ON9Bf~nPmpR(`(nc&CBNHSeQRoJDm}iLRI_Y48a$JA
zsQ+@}=NF93_Hjbq4@c|QEi7Mt@tsJI@;cwT^JsdSHeAdayScf!=T~o7Ju<S~ZVXPC
zn5@n<UWOg57LkY#AOEuF-CLUZw=W%5spQF5z_OZdv%sgSt83&m*JNZUty+%P+a1u(
z{A&|-GSu9GGAi*}!$Z!vHoYK@HbPOmx4JZWWAIJ4sw>XENV0!oUhP~2GYzgHuPeLf
zSx^74yUEdXT<>6Lz_*)4X}38di93V+=AzUHqlKcPv~^0CO}SU(dFlpZ19dB|Jgf;Q
z8hlc3HtuVm*I%Ic;#cb%z0s&7`upo3y$;w_&tXP>=UdhxGPLZD1nT7PRFSd#8zGSD
z8x6q1*(cWUPRZf~%9Np8AsctFXr1$91`boqgt617*=w<9VXRe1#UKT_gj;-vf_mTM
zsId$^SB9Go`G!jmta+|lto0~ee^1h$H~j3Uyr&;N6YpftF81YCI*JRxK?Vw%e-N;H
zPvk7*SF?EUD<!C^((!r<gNhs)`W-%!x!TNhHQcM-ai8Z_Bal<bL6kz5IrR;Kg`BBf
zEpNP7h5b9jXHXzu862$XxErbDY8TC!$ZnY!cU_v9U1Tcs2z{xE&ePJ9Rv1%C;Sul{
zvHOAwbRy#3GOR!c>WtVQ<_XRM!(g)_z41y=wRl`B(5k<a%Ti+jYQ6*H6&^Z3UBY)b
z;|LiGl<?9T$n2q+yX}h6q!zl3i~$kw?plj8VyRjqW))JY4Qr=)J@p)za3%uriFr!E
z+JxPKB&$yJE<omAg}Mj9N`wgEpMjOgv0hTJ24IM10C7>RG(v5q<U;ryMlMyT0E%a7
zoiH=0nXIAO)z`PvpG%tLT3V&D(IsRKjWv{3Fh%E`GRgs@DS(lwk~mNJ8B*Olmg<QR
z+dpb;c-+Tbn^#5s&aP(f^eif+mmu$a;6`b)i>#`##|*AYn&m$i4Xji~(m{&)Ha5tQ
zZ2)<d9yw!;c)4(XB^)al&{hFj1}1>qe3LLBg*6uM9{+@BPzWq(t}M#}l8INw>s|-x
zCEoE<4Q}jMrz0et#B#<d<7AcaNiK^$8M&7Ab6gD5Y2PXE1A@-S=S)>5ZpRy&gKsYu
z7BH(HG*%LHNlT6SlVv>3*HgvSPEO<I_R&5cyn>efn@n6PHmG3^b<2(6!znWMH}&t$
znAb?e=4;muv*S)t;N^h1w{vA%4YU8eG%5OB>#~yYAX|&BLbNVTOBFO252cGM``Jb$
z9Psn}TnqZlr2Gpz`x2jvCPlf5Asa*MB42YuRck(lXuwq;Wq6}+?2!G6^7@ec>0TE_
zc=z#n!R>z9=nCxMXm7hVZ^A7yd1r9Yn!pA58ZWzw>{rRjO3Q*cE#b(ahn<Vz->vMI
z!wgYyS|Oad-vo=)%S-4sQ?w+=r}bfKGg^XHBItC+*sbGY+3zZH^)5-`?(bixPSGrt
zTOVJ)X?$}t4cFQH>y^%Kb>icd2eAetY(SY<1OD@REk#&FJsBy}F94KTYT|pBAiqf3
z(+%^ZOeZJ9bj6DVQPPiqvU)rs;s@^Mg(bb4JJY9eBe3DI>uW=Cms59i;j^@Z6`Fsi
z*2drOHL0YD-3D?M<q<1XEgkBnYYi*?+x?HHdFZ~(SV!|JDToe*v?Ma81KpAc9<7c4
z*r0{8d<I*&r4KsZh^Eia&2770L}7esCvfpsg-a-)FNGJJHHHcijj=(5GJR8TFhXTL
zu54HG<=C}pavzUOLSJ#c-s;mqR;8z<eOSZ4y@UA*)($F9Np^RfA4nvu(K~&BRHu3}
zE#Yldz$K&V4?*lfl>o->Wi$&w4lHYs2`pryf+c9hfXKeVrN?Zf9b^><zE;E*ep6yp
zuHPO)2zZ*-j~M^V9g6P?ObZz%ri=v}GGNuteujHcw(}Y}Bosf_*}bnu>9^iYrJcrM
zAhP}Y9?d3hP2ST!tMXD(5Ea6Bg$O0~aU;o4Lu#p_ilJCZpLs2aB_ASEODPoR)9dK}
zX}6B2X-V`da+@QIiiUc+Sdg(4^3UHG#c1QTd{AGDIjX*A!-EK76`foNit2GmjVT!A
z$T;JA-VX8ri>73SsTEV#0~Hjtr27>JViXvaE#*o&sN1!i{sxi&{Atv3<GZPgS>rvW
zoJKq^Rrv?8i+0{n4VL8un@4r{>hf#AZi@xS>esm1{b1h>03i0gcKs1zyv@_PQS|*^
zJg&|=8c=CO(Uh~mDHYZ3>9zIu#TD^pZx4^A6VtFTi}jW~k`7vgB`yJ>hHsG9uS?g4
z_e9aXf7_UW$k;+MNh--b(q3C$I{)3BI`ZAcMknO=qVA#FY`7+ty>dmY1C=!F;Pzs*
zJ+M;s6=ZWbR5*-G8(Q##3f&sf*s@wbEhAc5>JVSD;wqSx80__jN6x(T+>$4=U?2GT
zC@TEKTBs^2Jn*u!ZKWHz`S$OkX^>IE5tOcPo#RVC&)?dgr8RFY{@xIk(ai&%&<7S(
zuma~<#lFeCUP8&7e`d`$5nHH&)w~b!Z(QZiKqZrzxUD07tLeo|?+-iAKixL5#jcwF
z>)P89_aAtak=bwQ=xA&4MqM4aM=MNt+XxTs1|Nla7UZm*yqISy!sEHI)3u#wuDZs-
z(Kd+u`4$gOCT$bP@nq4pNiaSmXu&349M+%1RfZs=)MAPu<)On3a}@djGvx~8|0{kA
zed_M^>*2556#+M0l3<Ria5-;*mJDL{1~3wc-ewpb$^JTq*I+}MbsE<ad|U6DeYpxl
z7bS@YT()0&q~%03G!*;uX5m|c7~gSLqqS8Cz)x;U$9GE_c2@7H9$MF~^ICn6Ow3Fk
z=pS$C{ppjEl8olY%Cu!*ScoCcB(q|#R0E1<G{=nb1Mhalz7a;h+ogSF?9|ZUvg8P{
zB24+Ebk6h11|QG*e7%tw0(BbXWxnQ^KC{!q0azraei9pFRR|@Bi8KV`2=mysGJw0<
z)!14Wtkr#SG!;sqp*%Omin}-YF&BGLdJ%9yB)=THj6!bx%*Z~h3u^TecstHn7#0rw
zPnmNxB2daWS@Zwb0qlG=X{0Dvz+DHr!wPvMyZWPpRQe?Oxt*Ap;iMWhT{_$!2rQ6M
za46wR0ZLL%4N%BtEiA->n6}qm$Qp*SK{WelO3F3jftff}P^}M3jj$k}5cEV|(Vn&W
z*yjo@LRj!L&39UCAhnAE7>xXG^vVH%L+;s?IMM2BTUx9DC1Z0*kv5ky*IA_2#zDP<
zUzdG(opSU^OBj`0Yqv`^KV<6(BV@F-Td2U0x~W>h?{sxd=b({-e-4(9wl88yTv-AQ
z7cIn;BV15bE{I<cQh;rp;ZkGHcvRNlre^97v6b1h>-mQ9&m1kTmzdh8mLn-ijb1DO
zW3iU^wrsTjd@5+M!F25s+T+LPDaJB-UOMev)Ih}8mnr^mC6y5y4EzsfXJ;45FMgr)
zmGsDW8#8%2#h|Yjti_bq-dfdzN+M)={@!-%Wm7K%PNukM)(4webOY_|GUVmu=jKKi
zs{V=V*gQ8aa5B`S=sh_*YZkQPbGuufd>*da`JZXEW9B%sW}!gjtXyHJ6c(parN&3S
zE-Cg{#X>Zazw+Esr=n7CJe8vEQlmUc;sI_Iw3(o<qbZR0c-$+mPH*PZUL|-uxE6%U
z==gW1AM)@7+Zd>4DFNfa283R^H=7t5Z@ryZti?Rcao|@F8mAOc=3Z7_!fdQLEF&@m
zC_%PLucQ0P__qCRSVLCdaP-weTgGe%kddGy!^|;zM_wZ2G6M?+d<nySO1c*WQVV`W
zqA_HdI~2w?S#Xi+y&neY^LQ<BYqkb^3}<j}H5P1j`NN$zQ98Xh<?IvS5hzOc#{CZ;
zEeWP(q@W5XHX?sQQBsb+y$e?ow|C+DRn8W$OW?FrT}LT}Su#>uy*=?Wne)yca7~{v
z`~35r>wLbq%rI|UJB;Jmx~ISk!T6)L=5gwFy_@89uWzr$#!nsshrnvFJaY75rC@0l
zp3;b}R0Hi=C;<99^s@rH4q`enf`au)k2FB|rK38Tbb_=}U=(Wnf@|&J#216ZqHnu>
z&y^`@)BhtGD;?Z-%kT#aK<dIMb7T88?&@on)huWzAr8-$6ei58A)^x3wJg*$@9Yz-
zPh1JMn*QCo;=@Pt$=$xqlaYtjDh@NB{9D*SlOQLf<dkMiiEj5~q_<c!a3Zh{?Ssmu
zPh9`RxOqnM+gP;w^59}$d}h%!vwl^XC5)3bE@zm$a_P#=x^p^gTi0$uk5N%`tkpz7
zaju9!!SK^BFdJ+cI8TF%F@+Cr*4D4A8O%lYxb6K-hS}|LFNs;mwGHI_>g1iGiJPz7
zUeA8*m@uIYsN2`nbZ!7KOMJ8zh!L<<N>8`{#F7|-;hALW%xSf(W6pY%HnEycfvBr`
zm!-oCMcrqn)A56^oRB=aDpJr-4CkEB2Kt5={VDTcV&DETOtRW#P;s@Nl~5Ng%_J3w
z2W-#iGtX#z2=s1%^XNX|q(p6e_>ABr=qJQ{_(esVjfX9>2?J?C-F?&_&+-B<TQxdP
zPCK1nUx!Q|IMVxUA6zdud$hMO*(7|I9JwU=P*50;t7}|Y$k#({#nm-cwskAQDfAsB
z5dK9CHr4;r7`s;qrL`0ma-H!WwsGC+7L3k26aaM<1V5&&UUvRdp41)6hGLDHJwIA3
zHRa5ZKiwo*gSiQn13p)=ImN{{MuK}RqDJqs7|5gnlmHEotmf6KP1^)ZMh5L-cDpCQ
zVx;a2%Nrw*3}ois;2eJq-VDKab^$S3jqkzCgRSa^yFdVHD@n!Gzke80*kCvIe2z<$
z#5v(XhC7j;8wE<4pbkzZ1c0BuE*6B)ZS9DABVhX_?R1ek-Tk_}M2q+B+0}(2{oTRU
z75e^ujZU5A@r>kwp-&-VA`&buOxxA{TRdurXN99uD>kr45CZ3G4d8qEIKB254u<vA
z1Rv@bwEAL5v4f9SZ`U3!I<Iy*O{zI7rSRdE<L3c&S6ldoPT}g^w|#s1ofjxy*EM7X
zRw0;_m*F24A9)D{h?QyXLW>(bKn?XhD@XiU4hum0wyrn}cut%!3arQJUQcRV<hOgs
zJkUsRT*td&zbTiOnWgat6$&3H!Q6VVFI9u-4w8@S`V=-Y+u+R7HX2YMup~kO;PI`B
z0SJ|Zhnd1%Rv1!Wvz44}JDyLoU`BaUOKzQyX@hnds-Vx#J|v&oQnc6t6}Ua&j1eHb
z(giz_!W^`%#QQCDL9MSYa9A=Ve!1nr!#&vDyCLwQLkg6dnaq!G0X%_0nJpNVI2ykh
zef$tB&Wk5G$N@|x3#5_(z_qN=;hx0OWAr1@WgwXyu$2NW)-%#nGNuPer2*h&n1M3N
z8%g0uf3z&Ta_9r#rXgy|xxY`_Jqp5GRk(G{>@-}*11v;y$w3KLyn~fKBaD!GO|B~<
zVq)Xp3vz!=1+DTQR+`kbS;x~P$?KjNX;Q4+KKKfvPz%nK;dDuZRJ*&?D2%Sqwu0d=
zNg8Czr+t4-)br3G-PckrzsX4I9W3!Okxn8+v|t=lq(N3L?-#Dqii)m3Qv<u~`HVQK
zEi-~6Fd0@};{FFm^hER2)m2dv-FzW65zFcIFiR4>pop6D-gCmNZu_c)=fs21e^Ynq
zW66d_a{*NE;B#|vrQS0J5;s@aMO&D=g}vqL%rlLzEr#m0x^uEtDw^5nH-$en(0X!#
zPT$>pdpCY?ufJP9+LzM3>3DZxgHK@7&{iLXWVC|eKxwc_)@L<Od^=AXn%1s|HwWjC
z;pp~Y-OClX`++$oz8<Kr)?_VLM00gcM#tHIPA*5VS7jW+EG4eimDg)!2;QXEc*1P5
z6V0^Q6`z>@b6H!e9a6Yo)g#n|)(-HtIx9=cJP+%HZVPf`=X^U@UfI>vP*ME#p6r>`
zSmiqG1?!%iNb7=g?Kf8SbVG7;q-u}L9M5V0%yuAsTj%>4)1kOd03cgu&b!As&Qgea
ztaQ4i8dbd{9&lCDd4IT3@ObUVdp*MS@MkXMa4uw{((FF2^C}K@A`h!KsgCDKB{Pcq
zvBokgKB!3@L5gKS9YHQHU?iAhL>=;o!BWC7F*PywI9>CY@wfncIA6OWGR^wmHh;nF
z(zu_F%mv`H<bPCJe{ibmAEqU4eQ8`QV%P2l+!9B=cz&s<5I?gq^r==~eN`lrf++ua
zEiO1<J<_QanDvY-0F9{zTdUK;2Y0Ee&i{SwUhDXN?`(tNdWSrCo=Dc#&pNhqoe@X#
zc0Ovn@7f9)8GF}vpHUnId<EOuU~y|d-F&>fbrWlDI||~(c)h^BZlSW05xjvwW>N+w
zHNkY5<M`DyfBIy$Xuz{RmBf!Yh5z0rVnpiss)?E!PW{o%VQPeQQ8*^O#(qiq_#~3G
z*Z{7EA9gv>EE>+F_QQAQIEEY&Yp!w-44;Vkb~!P~6dJ7vjG%CQ2g}MQG1IxOBJB>O
zOR6m018)qcjqc{Rt~9=6!PG6Nb7s#**S34!^z7Tx9ABPCul<%;qh}?&%x<d}43n1M
zhsZv4n}0w*1kZVUw?FOXA0MD*I|R$TfB2OlfI)3yWS!w@)Oo;oE?KX=P)my_eY#hW
z7W2o#`e3ikVZXZfFm_p&r6Zr#LOvyJ)439Xq7wzDU=NQ}t(&cr-uPFBPSWtQvdz)a
z)9av)r9%u9#J)9j(He^mQmt3{)>lox@)KumCGtII#sC7&OQRayYcVHmHTJ^o?#`#3
zxu=%4m<eQk5Xuk<jw_`2i1=mB?e><$USR2{okTq7@R?XXY+|Ha$UVuLj|i;0bZAr%
zYq`@t*Z9iIw27bjy=br&rIx{FuzG}zMRvN$lF%9(SyA`xd=tOCnVvmGY9T-|G-aOp
zF4D};Mj$8T5}70&dMQz^d_OS<yD#bV5zdLo^Bu5<Kz<98V+p&hcT<yFN=8BoiMNz3
zn7XKg1P(M|G~{&xKS-jKHK8)nR$h(RTE$>%jc1&GIM(3G`lk~{<W|<})M3ArUOcn<
z%^cUJ!H8e?tNcvWkHa1b=%X%aaIe-^!cx$j??6X~^Rf>wRkM4MW(A@rsOI$JP%L(K
zl`|F9s38-IbVPceQO?A9Yv~E2q$PPzV>MSzO{6UB#dr_;?r$_xW+v=7L*g#5^M0xO
z@^M$<aV<Z;;9X)K5Yfj`2%`8XjS~tk;(qFB3&VyB5a6hM4MZ>ZaMe&Hh1{&OpxYk<
zJ8xS%uTDF!Fg6O;8zgH=j5I}xyA%k-_d}d2j+o-&O?ntA2@?<7f@XnM?e8fg{|jX@
z_rf?KDDg-dnVN?EDt6|l4KpXK;$v5f6_%1g2JBXJn>Q`lXl_$+5Zu1M(Aai!uEx4>
z#W{we_Kq{Mb`vos8nRwiBR<*?;u1O%k=ikmF;j8URLw;rZ8n`0d;=#U9BE-?S_2_>
z+Y^@bV1v)N+%he!4UJOkC)075t>ip(2kibron9hF+M3|E0D3FyzdDTlv*yo%<i)oO
zSY355LavsBQq1U&Ug0wyh7*tjSt)SU9859)U43;<F1ezOZ)}KSPMiY~y@A0lX~N%l
zB*@PnjyWS3fS^9%d))7rj3pp{Ev#6@7I2tv2ikNEyN+M(dAJ;^mIL%gFUw+=3x=}-
z59P^9C1hba?YF!?+uOz<%$GtE0Dj+|jRmK}RhWCv19J2gn^SA8*YV$(0~%+9etdRz
zzny3wcQtm?8}kV9aX{jCb7jO`-<3P;Z!Vbdp?y@nMfPq(_@1{P|MKiWCzWDX{&!w9
z<vCptatEM3T8ZOn3A}WZ5bl&w<Ag>?In3qQ%w75(t?cEk9c7ogUQr|_?({nTyvS>q
zH3&pM+}GB2&p@>l?v~>+oZ~P9pZvfLf8aVbxVdI&*y~R?HU$Hrc)g3Z4mM-(<%8#R
z(FgyAt0UvJ+kY8m_vgEhvzvA7UEM(;OVJ_i9YW2)Mnp=y^DXDr;vr{K;nxQG*zYi6
zmI-Huk(1SHca>ZMBx#uRMaT=S%Gb=es^+W?2+8<(9x0pir<ZO|`_LhmurEo+n*}CB
z%opr6D~(EB^1uW*DI`=Wa48ufWSHkBCM}~Nt)NsH^p=O^<AJ>#nRb{n5f?yG?+}H>
z0@R4riYEkX>nM?1z_IGnnLNJPF(lx~fg5;uWaTs?V)~yHm>1wIUFJ0iynP_mRB|_S
z)c`Q!4l-aGZXBRhg6ArXMNU>fuD}0$CZN69h;y;=oTo4Oro(pvTY=eR2MZCOiVaTL
z7-T5)B;=E-b3ueMxBuw4IfXsi&1GclQmJP#auvP<wXUc^b#--PNj>B3s{btrlQBGH
zB>lv2F#b}?M-}4_c~hpVc^Ce8L*I5YlCc)}XzGL%$ZYr+0V04&`W1{RkugDwR<GAr
zW;}S<*dq&>o}03oh61phLTB-G(-r?xb=)ZD$|P%@(Z)1r7E670-N&dGv)=7?(=hk5
zozlM&*sEIw^K0whJi@2YMHt72jfc@cs~^E|2E4FhgbAAW4~I%u6$mL%GQ<2I1tU_F
zB^@rQ$h2fPfPiUjg%cGsP1Gus#HObGHib2;RyFe5Yg1xt$Z6=x8h+B%M95U2Z%Z9~
zT1EN*q=;X|4e&mM$C;0KDmBxi@0(vB#vPC$B?DHsSc)OL<YCq;OX2+Zq$FIMj}04*
zGqPWll<5l9#Uoc+sig(hEuu0r|BD18el6PnGNMXH7`;L}PF9V0*f;pLGqixEm2!pp
zizCI9N5nbs&WEXlJNQYrzGl2l1kzxUX6PmUZ$No=JbNPQ?J9l==z%sjz942(EymCl
zN%m%uR_NaI^Zlch-N<G>0?ELL(NSfz7$6XhUTyXa3Bk0taTLTO5IKoFx-tX?E4LIN
z-Wh!2^`BN0LgPQkSupLS?dZ((s(;^<PZJ-x2?ymOHK|6d>v$4a*FWI);J}iXZVm)f
zZC8dS+MS-;YU=bhy5QN6hSkYs*zDa*q|1swjMg$p<Mu;bmj3H5+AJDIsH2xV-gEk>
zii*>`UE|c>)pY=Mc_SMqZfy@uEk1paC{VNf0RNji`kQpF3F?k{GfVdBotH$OCMhzK
z(%?w%h&A9!S^V0Y(eL=}yO-xn7f1qIyMIWV-W(q8BaH9y((osTNXLIi895Ouq=pv(
zAb#{(DsWcJ=1Nv1UrJ$dDu~@i3ohqGe_`<wdbFJpWB4L1!t_iAVQDnC)Z)4ykbxl#
ziC?u;w_kQgSEJfIG4*zW>sm)JCQ+oNkwbW8XBzVwm`fvi1}Gq@mLm)OIs=Ocr+1*G
zm7~4sGOD?_map1M;y5KbyQbQ6d7e|Z?d#k;r03Nvke%O`KqV|mkkn>JICf|?dUWc8
zNI_G^E>vcNS|A5=Zz&PSXt3;7ywo@zo}r(Nsi^OE1pn9KKag}@swXD7rleQ9H9&KH
zq5lo@IdmNak!8J88QI5Er+b+sS0fIifA_8D;H~6qDE=2)L$miquX9#@5f2hl<oSVB
zjW9^NIUmM|6TjT?<VRTgmED4u$7x1Euao%GWvio!I!j3drWdtfvQ){$V92Mq#4RK)
z8g}cV)^*J%(HHyS4~3^#Kf-kCBHn*t)Qd5SLdRN;i3oxuS!!8m2nJS+e8kvR&u=aS
ziR5FXL8QubedWyl%}tAIc6>n}_2mT-Ly+I=yD@)eVU=h0oleZdns7J>fec}C-)}Q>
z6^uJ%KIZNVHCpO67kc@EeyQc0v(dB+2FAmL8zv2`-d3XpmrT{hDWs7)#NJ?A8BGx5
zj}Fu=wc=haTcwmk!axggaDn;}$Jq_UtqK>=7ku^sgaxIPooaN+Z?lXa92`WQ-{1;2
z%pUnHrKJ^(^(jDYLf>wl55PTLd7mFPF70_qgG#wg=T^kUo+rrB-=BrwUc%G_mQ34G
z6<X&m+UWhaO=_Hz`Fi0MGCDT8)NkmLsN!rasnvH)*YYI%a}7ATxOj<CivirM+Dg4A
zOw6m!>3SPpqKI|eNZg;}u7O@&d^CEa{j$b$9lkcgL1%xbJSXX&TeaC)#ksV<Ds<FP
z;5J2*jqdHeq%zV*(0pcP!-4VVjOSh|L}UZ;O8S5@oo}0Ofj}p+qOTlJ(Rm~mLgiY2
zDk^Yw7tUSz*tE%9Di3Y-RR+`xL}^@hirIQjF}?QWPVczwjF&2w9IDVkzp{V(MUB|w
zRd?^^^-u3$x%!+^e8zaWqV0c+&91!J7p8+_@Ypm80qqniY9dGx5QYUIO~=_7pEFbZ
zxca!Rd3Nmfu+`rAXm;;tlR1UON-7-+ZeyNqJTw<l=`MMfELf&PDG&i*Fw!Jo^!T-a
z5O<e@kpozmb^NpSj2N{PczE&F%uG!4@(K#fOhiNOVo-UfiCNkzi`lvV^)SdeB|`g)
zDPlskVk6bn)iEgwCP<rz>0{{#qm;f<hkJIQ7OMu%Rs+$PS-6%;@LDwHU1c4xk{Y^c
z#r$<bm`WPF6Rhw|Iv$Hudr{s)c}&qdkUZ>OPlG+fI%M}f`Jam$UO8MfO)`c^>c8gM
z>DPJwfwWe?qZK8>*JWq`IGGweD}G}i89o<Mq|NT{T`$1sAL=CT<U3E3I?j=>Nc-%@
zCjip^TuBe3FJzKT_9jkJP-N2*(J6A%T?!fsW>eCn$fA~RwKD!&SErs2J0FGXMWm@u
z$xvv-#?N5BzpE>uX`&7WuhXo5!S;Iv_sj9tup{kbZKtgJ@Q;QD9(13;Pi^LzcMHh?
zN{zCZBlsb2iKt|au*)IRXUqzpL6J<1(V5Z$e3M<mp_2c)!&z*wSW7ho@2d|*(6Yv+
z6h*_<eEG{A-Xb|FxVp0Cp3ahn>8L{=%!s{dsik`R`Qd8aO)a(oNf4-hh$|PQ^FF{P
z?5kF&7KYWAL`PGZw0qj~_rzLyfMTYLBA3kf9v0AUK(+YOSzJOyiO^-OMz69CGhC<+
z2E~U08d+_pp`MgFhyK;E_TmLZrD>a+*B`hIOwDGNUcVIO*ak}+*F6nZYosN_4PML3
zU2Qh1s;H0-<>0j2+4x{EcRoBE<5^wZH1HQJ3ZSK@&&x9{?qiGy;(2bh*pl9tjgY2E
z#KrABS>M{qaQ_5kfE}h4vL+@bHn!So5mi@D59hTe<>x*lNZOsQ<%l^xLZ(USs+QKR
zRJ%4hDR%pwp9C>OqrgwKUFJq-&Ob5o64=!(ST3*`H<}8vv61e*W38;H5j3rX5<0$J
zbk1({G-5A$tsv8fLiK+Ta(v<Y_^?t&Re-S3obwjgtVXF?z+C2L<SHxJGO-~}sspUM
zi?tm1PyVz2m0Z}+;?@YCZ%B}Yg*?=w^tpoet_#1V4G#tGF4eg<WvQQSFT6@{Xj)Q8
zZK<xCnsF8&S4<K&Ok#PrQx3w6{w-D9>r8ZUk^vObN<*2+SpS6uZZ>ea2;`Je<;<w9
z2HbAC-5m-tNeWrWjg-BGSS|^am9gxq7g$@oCH)B8D1L@zi4H*WGjC8>fg+I}U&dIY
zc_N>#QrKvtt7~c`FvUT4QMCTtT*hQNP#r5igFXYu0xNobRb`$DdUm?%7W<|)X^26o
zlDY`<%`aS^TVAV&TwXpLlRRpCh2YLLz9!@SpcEP!|83<^gH?%^LKc+xpFfqn9;5VD
z1gO~J^XTTab<VHr{)H7PGtTIA@_r_9oF`=AKq%%JYdNV#Wp>(sd945QsQY)PM(anu
z|ADb74y?OfUpnf5^n%#_cA9}oW#nsY^RG8$nS~^oVZO;gHb5j(ESy_PDaEpIC`^eF
z|1)<X<{n>&wx3BqV-BO}x5u3-14PBd__rAb>J2z13-Z~Stg5C&tdwLm<1tE>28K?v
z?U9vnalF8|bNr97z%NyM?un*`CY@3y_S2_iFp#D7*VdzeX3L#gRsq%a<!X6#Iw~7O
zv=1hG-k&6`1|O|@*0kSb^V)tl_^moUSWe*NAIiwIQH>V<As%(;)4t-oA3zmULy;iA
zNp%u=J*>%3_i{yqQgL6>ctn3QLHQK6{$=Bv{Yyeze_ylo?ep^nRLiil0A13%IgCp`
zOi{4NeW7tY0d<Mdq3-(oBR#7f_H)bL5y%y}PxCeYRI<{ffC%NOVVQsbv3%HAw?|Pz
zL&Rovmb4aJcC<8ZYfL?Jt(MV`Xs&jGY_|*A_@ic#u*a1qpQUpzgVk0~URISg0qE7R
zy-<M~c->w=N=;XnkUhslkd9cBYHkiT=b}v&X~<cPDQun<9lo0LHch4@j^@P0VkL+h
z>e6(0ORX-+stE3xkFGvS+>vK29=5JsJ;3ZUud)JTn6#qnmt1DN3LDszT1`wR%hF~X
zpJ9NYQ}#HyT3*m6q@T!c`Z6-8<Oj<}$9+Wev7$!f@XCQh4Ddx{wcVzrXg<uO;I({_
zqj(N7U7wK`#9C514TW%Kuxr(tn67nr|7m+CgG4siXd)C82$CaZ6^iRA$X}Vi=(lHx
z@BGJJ!_{$4<r3<<)*7(6{kBYA=9^%^7}0N!X4ZJ0iKOV>+{m;yh6V<HW?y8O`V7)b
zw6e*|D2DAzJ|%VcA72=wJ8of{^q6GkQm5SfIXqd*-<az1x3QKIcIb!|9@~7;JOJST
zCOSe~Mw*O*X<J%B75S-@_TRxR*wQlU9S&h!aU&KO>z7hE{4_XyKKuZ?*3H>C|6K6F
ztH!U0VJVOg0-~_F{-azIay;=C2j5B>N0nxMnvDeS+%H5rl3E(h=m1L_CxGI){~u3h
z8P??chyBr`WFXQ=jg$~UKtReNozg9ebV!f7X+cUFNhKu)(p}P0KR}udkaoc6j%WYp
zIG$&(_hLJ)d%Ldd{GR9MT;D2DCl6I0|Gx6-ErKFQEbDf|RQ4vW^O*7efaR|7{=Bin
z>+Is=i4OIXf~fzlSIqI{5vfgsOv96(!F+vik5aGi<RT?J#XphO#s+IhZ__wiCQf3y
zL0s;$u}6yaM*RDzc68uPsLkDRMb@~yxa_M>&F3A7{7nGW80X7Y|BbNVvjBONtI{PT
z73zPkrU>8RAf}4u;eu#0)!y10*yALD=yW55f7KwJ$HxRKD=UqxB$&vFddz3fJL02X
z9Cgp)iEucIGU)54-v6kX!PKg)-TJqmTcQA;u6Yweu*%*W5cK2kAA0ZibC#czQ`%39
z?Fkm?82}q<LLEj%jISg*C0NKKTm4#1s$_x=teQ=lch{U_YRV$R<ahjWN?(WJtAJDV
zofq-9GCKm@gSZA4ksrizlri&C)H-}qZ%%993@)Gi2eR9fnf`9iU`kddf4YgT5=t9B
z^)q*E%=|Mkk+?wQ^G(-5I=5m5ed)x->b&?si32D)bupw+;t;4oUrntdTF>{*EG(QF
z${}lR5^!Bv`%JFzj1t!eK}L^~ht@B#dg0POq1m*TEv&)WW7kZTH|=#h{QnK0!+*pX
zn^_R1o%&(jC65>aFcO}MZy^)sWB!1b7kyZ#$8{m9@<Z}N>YoPAnIB~GEG^9Yfo*If
zylif4<zs2~9E&D@OElIMk{Y2w$675wmXtZ3AeObam)we;-Z(zruShSC8LnL`c;gsb
zvv4hutr@A`Y;2woxwcRr(|3woEXg|Ur(1Gq<bBOG2@70|f6cK8R%qa9>2O(uyxH}i
zZ5@k{Q<b^82&lJ6>q%D98c(MMw3dB0E~U=*D-E8k*e}m4{8+(D9HTEBYli&ATKp19
zyv+TV(-nuvnt4dMHGh739zywu4N^1BkPI#wD!kaRygcf>KF$OGvJAi=@BcEo>b?HH
zEAsp+R^yCw(@E)8!r)O>@0w;gsp7(ZV=+(92IjCSw;<C-=fxZBAn6L&2_wou6VzTs
zymC<z5#l5!-5&q<<NRgSpa^2aVkVuojgp;sgK?Z_`IMP`vzJGkPbEH1O_f(c{Z}}C
zHeL8vPR_S)S6_++gT{S@k<`me6U(5P3?zwsMzX;bb{F%o40@CPYI&EM_ZPHoHsr6o
zVfKT!&qv5qi&8d~YCC*@e{NfQmdvUa8H#Ta)X*0)Uqh5b*R%aE4puvFlYwi~&Qnv_
zd)badZs3Uf{>n12)>|W~SB1<TQviGLT}XSNvKjDu#`u8kO!@Vf0Cg37Dvocg5b(!<
zqr<QXHxsfF$~h^eh{X50vDTc|VET8s0P0|$^d-cle?V(CndljydWZ9O^qR-t<DDEe
zwLB}k9AR_fo4ptw(t%yg{&^X;Eqq)O9^KD!iA{7*9fjP#rx%`;t|$~f=P6K5{>+P*
z%_<)*FSZ}f=W%+QvcJ&aVhXBh0I|PVpWQ0#hlEZ_W-Y9-1@gbSd+XX8=mED*TmLXS
zn4MH~_G!jK{)=obKnMFTqNyBRqV-tZ`6bhXgoJoZcD9}(PBu)U|3A^u`OC{S?u4vj
z0itO{@|3$o##k{gxp7kUjFA)*x%u**ch;#xy>V4u?n`sQDi)c5r+MndHv0B)?EQ8H
zablaacc1-lECYYsT)848vEns1cjG^#w-`Oq7H84CXFu;FK=JTi=PXyF#%VTn4mD83
zYiVx2R|(wVu;M_%?c_N6zfeB>4vtq=p}Y}$cCY$9fh2GOX7^zm>OMH}ADR5GDEx!R
zxy|Ruj>bTr%L>tfT?9o*nd>r9l5g_Fgbq?aH8mCPo)<V{;g7IW>^yXTNLBOzzpuq4
zJ?^x!3|-~|EFi~fj8qWeT|=H7d^5%<!5MUm!@rh)o>FM{m&W|gQ-#rXJ_|ZxHOuIQ
zg)@I<-k&Z@c0R5$IjZ3;Wzb*yWB)y$ZYLr_U6;Ag#9VfxEXt7rpV>90%J|xAKPeGu
zn4c`p-}Zb=`r0+3jGoLj*UW5DB|EJdq4W9(-(lJV;$MqTT{`Iywo-qWURq35z{}D%
z{wK<7aBNk6YQ0gh$Q+jzGjAsfI`H>@$F_a6>nX8Ku$-|3`S^}Lf|cagHgwc_%qXQ?
z?rk$q62GfM-e#a<59%(=rW%D@<a_`<xC{GKNuq3_{U%k-dBjOfEve8YQ*;>jBN*jx
zNR8eBJPCuRc6PV7|1-7}A0sMb$ScVi&kc0Q+Gr0;#0^oXhR}r>in8}-%sQZ7p~RBQ
zhaE=yGKn}uwz)_+g)?%_Q>j%yiwb+`^Oj0nc`0}6Bs>=TXP28l&B6M)rZVWHfFkG~
z8$I_2?$4(-@_+p9f;3y}RFX89r%;9?2|s}Ww~iy>pmyIz;bz6A+T;woSE7a=ic)m=
zB&`SIaHzmA5;k8Y_1=&SP)HOmBx+A%F!9mxzlJ#3w$W`4UZ~6W>gg80BUF0Hagi&-
zdm`mv-Idp-(jA;)p#G)=P=AL1`Mv;++6T=kPl=_81)ieV8J_TZnvB#GO#5woTG=eH
zH$rnPJsfRUZxnc`W~r;dyob5NvK$;!o!ic*(9sjb55>I-AoB}E*Z!`|dunGU-zY-B
zI0X0!y9L?-`O~`}2zylrT{;=<-h^nY4utEgY?#rJ5$eC`JRe)U4}krgxjX1#StvFZ
zpqZVWRj_hMk{-^OzqUe{KT-r!YkdP!biqR-?ynMIgkzJQu5IVFmbZKand&ZMT3YR{
zJQ7w^xGyA&K+t|qlS{i2*=;W>flT+8yA2{7Lt>CN(3e4!ft<zDx2#e9xxHilTX>X?
z?+Qg@e?jNpO9DaeS4X={g_A0{3V*ytM|7iF;@eUoF1PC$(Y}sWf8l6_hcQ53;_K$p
z-hM^y&lje|Y&yy?90+G%3brB^JR7UzpYZsl*n+wXSGu7gOmzF4MK;JzTku0|`hZYN
z&i=Pq%&C95d%F=oS280@K&B1SqG}mKndtMFE^2V*)S=o4L9@}q6Jsrz^GDw%CCK+a
zmu3(}{Jz>)B0{|L5-mA&cL@xa6^0u;>1=B10PeMUQ;X?E&i&el-v1iGcC!23DaK*=
zhg~*n-P^}Z5j=Vd>!y4B!?p9CT&WeqrI2lAV{}&k2jfVxiOIX6WbW2$BHFDIjj8ai
zw42qt?)9r@8V_OB;wUeyZE0d?659dh-HP1tgdFDhpuQpu5C-S1SjjN=^;ak{fu+8F
z64c9lSxR!m6mC3%k@YpwM^_*}c54>TqU%ZjJbj*YJTc~`FNTno|NgIx)v_uE&K+`#
zL`siiXLA%NlbCJ_KRf~fLw#ID^I15Swp=epIs5PYf0otmc7G+~w4xqOoj<Cj*6lwW
z4c|NrRjWS|<fj2hNq(9Mrm8#0rt1pQOnsujjHj3LYqa+F{>{J&8Q8{DHujl%#Q%~@
zMS7XV3c>Xal>$%4G_|5b3x3nBg<?Egjni`4B{U|N#Ejp)%7sQ{S@ddfQ03!`zQ)h*
zRjygx0`AJOd)M+Qj?B%<&CqHkHRVc-w<n|2a&aXG8u(ydsbNi>e{S(_wCrhv#?9z?
z$N4P==rlrnosX~z^N@L(PxZ`1G{%`oiL94QE(=8)rK|U`8N#-(DA)Ow*te&%n68ea
zD-a#n{nOb#T7(F%tGJ0z>^@XE4blYfswj}UDSmVMyo=2t+u1fp!#mUVvgL4i<KJ!G
z<v)tM4}eNJ!`~g9q*vOyIC7$AF_Waj8<VExDUvT<yxsh<K%*GBxYv+^3z@+_NS{y3
zeD8Ke5PINly1n^FF{fb#*Xz7mQ~Z0C1~JPQ^_JI&<*QK&*$eEAbtNHPNJ|uqcd?nw
zeSI5G7MBE(mfA|CnK9-B&EW|be8GM6Y;N;t9q3!_xkl-LA8TxS7KAm-JmVmtH`lnk
zbaV4ri*xC2$PYaKniuf=7gCKw%x2y<TAYj~c6tQi_bal8YPc~giT|1K8>`4|c3(!X
zce>1v-~_>P516uRC)Z6FaowolI3~D|{PSFesK6qo?;3O4B~*QzZ=0MaH{)ng<&@3z
z0>A|y=UZ!h+$0{~awY<BXh_@n7#yU2EUhBq$|(0e#ij8mEUf1(z@yM5P^%2K!+r)^
zBYqmR^+vXbM!khVSw2M+Qz!?Ef}6%(=E8J+xuYv1xVup0$*^CJn0_j$Ca%&c(Pf0*
zXL+11cVlnKdH3dBPf09}qS@@sFMk;Wh{v@yH}812>UcP`ThJ;dGH`$lhHV;WNykpF
zZ#Ud07%{4Hw&suc<l98H1!#tC+8@Mm$h|vp>Y$7B=GgT2YNysEzd}yc)*>ecXKd0M
zuflRfRq}$)zE+iMK1p~zIY9=+ct?MfejL_Fw(HFC#T4FRHbQB{8^JTvC#mJh`=+%o
zz{^9#RL~Io4ayX!q+!le^eIeGM37bl-Ih3r+Kc!8*YZC6@+N(v1ots8v|hYbI?IUc
zytk@0+J79oHC4W|uKKjn01XKMTkyVFoc`w8=1LgWZ9A?tTqDtTFGnK$btL84*kZpJ
z_FLE0dUyxzX9HI!Q|BoE^PMRzA0H(y;p0nOxuwo?IYzv}sh!h}X^qUnYa*q*psUjI
zJ<krF%?;}b@QO>Lmm`ouM^vVOZ}sj=pf;}5-p#>J!2MZgTVV|P%1+3ZL@pkjPi>fH
z(@|aT460tnaQtfFpAxkJQH4k4VTm1ick_S|!jb)R#~)qZ4`Qkik=|XU{Dj}Si-JD*
zUzl9YNwOdb)!)tVU-!Dv?kN`j!0mYZws)SV%=|ONI)UBw>k`nIp+4fs+VB4%zH(We
zsrtyYmoOjCdgL#r&>j}v1}}5qCvC4l8<dQy$owHHAEKUH(0-CyEeE2^T43|V1UgkT
z&Kwzxi440h?Hll#&rUluTL>D0>Uc;5@KuSSlTekU_4e~+X0od|kNzKJ4YmW^TpHKN
zZY*MB`7%uQ2DqI*xf(49Gpjy``_C>AM5D_CRWGz!m%~-YRpqtmlr^eUB?%JFDd}O>
zg0#QLpr=ov>guQ*_HMl;u>Mu<%lz(Y>(`Z*;&$UB!zHUz)AOXoL-@5B*~wy5XAZ8I
z1aUBkSI3IFG$hXRxzZpD<$s^Lu<06`z0YNWSzJ&3H-VLQBCzRy=3x1@k3w@Np7N|N
zkOCY(IoQxqYSO})gyj<*C%)>~@UGrLg??U~umtTwZB-6c>fccN)~Uti#lVw;Lp$3)
z6{H2JX}DkU<#ID~f`E1@K7*Lc0QoWO0nTAUO&xa-`zu5l=-XvlF)si4Udrj_qRy@L
z2GN$x3P8R>LlwUKjp>DJ+dD)x@t13sFni1yJyuD=*J&_8@VxPW!?CWrcjZxzyNj;v
zf!^K?jZ!X||3<}g|9G``HWGnKN=j~BW=*6P=sN=7Nn4lGqiy_*Y^#gQcdd=6KvTX~
zWVEzz5P!zT{)@26!de<~R=+Lov>l(Batm2_0+KoH?d>)EvuLm7jD9hVDYD40jrFT5
zY=gH59ogV~84qiC78>2t<4>i~iZnND;73oFz5jra5RsfHTXfN%sjr1)7MgXBdRh-V
z*fw@r2r^)paH=iGxE$Jp+%PCV8G6s7443ok+tU^&rvlFzmqt^2t}o7Ss$vwApxtKx
z3XaL~YBwO6go)emf;MZT8#NgMMRLrLX2pmzPnv9VEpBlbnS#ag2n7X2PI)bVb8~a{
zICsIskFwuTeZyH=ee)&tAG0he`Y#@1o**h+D^qpHrn<YaG30FZ4`JP!)n+_7&mW8l
zakmyI9Xw3Z2S?G>g?1^Bz-TaS*9XAH&=SGL#nkcj`7(#8(kvy80!f`dv%*tV4r_CL
zlvI(IaT?jd(ankf@-?c~c02_;2;PZ*V0yaZyoAIRU2Og90g`4%H8c<Shc-4ax@G-Q
zdTG)zdT7~w^)Z?}D<TSFYB<*vx(e`j-Yh+K0q!|Mb~GQc)tr5v{<Zzy69WX@{)M+L
zmX<%lQ!V|=AWNg^64O$R^;~2QtR33bG2&d6@eP;RIn@f1QoqOc;{^Xm0hMN!!EaW&
zt8RZ+?PD;F$2S)ja2@LF>+6{5zdc$qx2>3f*#$pa+a@z1y<(Ae%6TGut@x)rMO4Yk
zykCPsufSXrZz;jQ=7)w%r2q~kwx1L;?oS{(Q=*CLl)`iOM{$<^bwl2V+#=WH1%!&R
zeLpw0SwkUUG9C~yerQwNMbqs=(APw|0(N0s5g2-xoGuDBA_HT_6DWG{EqtP4-Sb6r
z6LyhCXwWrpxp<AhglT*V{M)!nO}sQ<a6Yb>Q-wV!3KuoAj)_eJS__mxB*Am^IA9*V
z-5mls&iXHt{xvc@rL9OAkhI^)Hlh)=!2G~Kq3+l~Y8(m*XT@1Cezsk&)YH+}P=@w1
zr#`0N_zL+2N9Q<ZE{JoAlTUm7a3n|}QHQ)TCnZV5<AM}Yn*@GWW_dD}vSr-3m^1p}
z)2C0WrKf(_+vyY_T7{P6V<-Rg%F!9?D@cNt>Bt1#(1A?$>Za@?<PZjGyv?mrChFWO
zNzbNMyK}t^xJ$OS(qSMbnX7{VE{$*nfsM;P{i<1Geh0(KBxvB_-@bhmj={*=0`*0g
z#yDYk`6EbV&&>Q~Gej30t3lAUv{wfjq+`a7i@Ys%Q2-7!)-wC;@r=t@cH=0#eVpx*
zV0fQkPfGq4kr+En_WBl2&R_+0V_Tjg-^;&!1qB=kV{oj8Meddcl4*}w82;rc9a$81
zEvHr-zyByB{oaeZv&e;ps<4Y*Uk_iZGfeG!XBK1E!+?bF*2Oq&YKJ`8h=1mIhaUs&
z&&cNi9!vXT&x=_1S0)2!0f~(6(lj=hS-_rbx1IlH`(N$UimJV0<<y`4bKAXrbaHa!
zpos}93BFmlG68|*M}AP-2~~Wu=zNZHH8-yqAET-5t8spxYct)Vr>YEz^Pm!6m8v9f
zJMZ(my#`J`rOz5t{`lA}((YdE+uqR<SVs;VjVvxKHAM36Q_HT#KfT&}J3x!ysiV;D
z-|39|ac=i(0S*CW?=vKG?@E30&yB@{vo(vvhx_=qbTy61+7K#R+AVnIW9guyu&q29
zhO-$0J^g5&V<-ZhnVk^Ll7G;>X3+COW2)Qgi%scyhdl#Jwih{`Jx69gPLghB;wv{m
z{pkXI&>DK#cPqz-C@cgPaB(2tio9Brz3ZX1<xbwYIXapYn}*k-SiC7veo&BJiW^Mx
z&9rD71l&7gj2r*Ypnv<%w!VFdxOiVcqK|(!5sO;9ZVeIp=Ytj6DB?y@)%cRDRKj0r
zX7^gKw9%bqD`eIKywBx=kw5g>QnBjI4;FS*ccj<~1<YYOa{U*p5i4#DD=t}x5$_+9
zOue6G2BUpPCk3_xEHC}<j{aJA2!_dNQ;DOtsdSzyy_N;ZQ4_`Eaadf<ku!Yd`uo84
z&;>UnSd}M)eZgP?@wfyMY79<<?qL59#{>4z=LzwXu-Oei;Vh=zgoJ(OP#?gg9I8?y
zx>tAXoy(>AY$&MJO9G{l@;Q$B)@69x=#Bk+@kb2@Q{&X4oIliXE=c69MQ3V(50ms$
zHeuB?u<D@DH(fH4`VTeykHAPqZx<SL*x}vuJpFk%9(z%ENlisjij-9A$Vj)XvG@J9
zMd51+*JE#sG|q>r-|JPqBMSf?c?<nV1dl2FqZMEQmq$#&do6@G*XL(rV`IS0P_0Fy
z)IV<%WJmp|L-lsU@uBZHU3-ScjN-)pnOD-kUe=>-)b{QLy+0^Lrm0^(vjl6}>y*l*
zt+nXP@sZl7!TO3l%c$g$e@^@b(&XfyN)-uPfu=f)bBc}Gq2CV>{6KR@Cf8+Cb)1;o
z2*QHmg)($j)Hq+eOtaCnD!tvJu>rL`ru}u=L5BvdM>6^yxlb9!;9y!s@JUpE&urQN
zUxEGc9#Zz@OUxP3TKm9kGCkh<u}$k#AojP~^0Ui9GRGf7J3Q~+5pc9Vbn1!P^PHV3
zRgkk%lF@hj=rJ=h4}4p0?P#VyA(~A#W`~s|w;i6XCCFZFJ2aq{hv#)NgHhW-ufWu#
z-y=a8Ic|UE#2-J_5R&nxk-2<-RbM47F}K(%EvrOmD30Lg&fjq7Eu<zf2g}~9Qr!QO
zz4zLvIM1b&en>Q9+I`-ta-MT@t8drQx9v*F*p$q<DnJ7a0M(NFFz`olt+v}H)3%6)
zxX&Z8xWv}dzR$<oGhgfvvH~vjL57{zXQdf&xRi!AwSwwtyUcs$%(5|i9B@>oI+sF}
zW=YS8jd8!e;>kOWr1zo>idwxqONf+(AGA`|HL47Ru>Rh{0_0(1(4}F}xzFA7>Rrgv
z(RtMNu;rx#VAE4;EA)yL6Q<*5`;?X4x=euJASyQTk)lN~zFf*m4y^*=xQSYbV7S}W
z-@YsiKupQ&Ybb8wh`-aWRVX-%&U;J+auHH8t35!|-)e=vs1lTQ`Ys+~T@Gok4x^2t
zj!w0Hm5;QeaRqA}9+pQ(1i$m8QBWpUfJB0nc?wm4IjH6jFPo~(<Eh~trSbT@4(V1C
ze#Q@xarm)w`(|LawItBTciEsS!ly1TdHc{k{@7AJ1f@2N_Q#&i=O4vGOi}7dyDX0b
z*S0!eFg=LX>7Jh-)#uidLU;ywkb_iJ5PE5x#$^OwK&o*JXOg88Hu`=nM+T$>42bJ?
z3;i@e!yIm#WzaA-Rf9~zU4+m4_K;`urJxR8d&Xm+5ad^AIMxhRL!8cvCkJ&dO91YE
zSE)WI><w$eJ~SAfE46iYc2+!P-?@C=VDJ2-8as}kb}=`%Jo9UZghD3ilIuN+{y2%n
z#1R{()tS6=y#2O(y<5B0gVb4jrH}a369g0%7;j+%N&EYi9)c~Ff;16SZBh(Hh%@(Q
zs8#86h&>{^{+ej*GN2U8#4$GAAJoZ&>g#_Bj_FfXA&7$2`MCYiXmN4UpW!YkNkIAH
z)i<{U81u)hwwoi~XEH1v__{XkZD6rZ*;@k0M6Br3+Jh6KrCn%`tAHbyNyk13m|Q{%
z_59b`r#3fshG|_lUR*{R+i>j_?l`&EH~P4F!E;HfYef`#vo7<O8To0-uPGis@Hr2<
z!J1<C94)0@NZnU;T#~W?M~8QJ7s*;O`=@@!4o)^%H8mRbtc!a^6#IMWMQ942cuS_%
znv@>z+!BBPHG}eczH+mjnoL3<QW_KBedF$P;+=wO_b?`SN@~SMprQ>e>}p5mXbxQT
z=eA*Ogi=Zr`^<t{Q+-a+Bh^efjV3nwX+^gyJrtd{jmzF48QRKy*i+1xU@@PF3u3j^
zQD+Y+l*?BiAL%j+wHEpmU7USFzYg;0s-G#&g@nP|U2uN@wa<69x&&}p|Llv9Y2<3s
zm^z5(+uAC@x3bbqxj9@0&dWP*VwP_<XYY15L<q6yfVQf#vN;914U=POOx;4_%c>N!
zsu<~U9H@xSH}CQ%g?5cC<@r1YPe^)4dRF&mJFlkiS3B>KKFdKRi!ObttRPRx9QnQ2
zAxJ7B2eF5``<b&f2AFKh<xWb!#Zh3T{6dPr#srDq%d;xWKX&?8Hzvq6oE3BYDS^f3
z3J+^nBljRB7~nc9whwZEHNEVq$~ZQ)*q|xvKp;B2Us%-lldwfZnM=cZVGy<lY%a-x
z)>Oh_gfI|gb81uH&^rUI6<lr;l2p(;Y%(`Lk<S%wUnZHYgH)Hf+Fc(Qx9E^)&3vT)
zs-l(#Oc~tnfeT8+2^Hh}Jb@(iTVPmeGbvpP>K=E^a!X(o@cUx@<~Chb64UNe?;Nh{
zPjqd0-!HCB0{gaNf=JJ6#H9laa!ni`zQfhYwV`-$P8*|WOkG(Tkfx`iyY-dYt}Y1|
zfxxDZl%sz9BbK}WGjhE^Bk*&lJB+A&151w;Qx#aB9<4vOsHP{6)X-PlAHBUkPaozK
zv9z>I7+CuO55}E@Ntya%>Zx8o0e#Vpe_fC+Wcm6@bd3pzb>;_b`fnUC#xO-(2B6TV
z*&fM5;rv20q-^17>=XYE4!-fwv}>aFyv>8ls_k>bAjX+I-t3~my;k)qZH>lhsc8MR
zdAMjly>r@Bw0-^l9_C5W@BU0^*4S{jn^u;mq-n)0nqOrB?(O+AowF=>K|+}KGgKvO
zftz`1XMJk^WZTLq18X!@P<-dO=u&S6mK@FYD58cxu^cr>-`Y~iWk^;|gJ&)p_?zop
z`=3{GKeVBeTblj=U=FZ}UaW3Ye~i^=T{1W0#8IqDvezP(?JXTQGmSw6luA7lmpd`T
z<s!D_m%IPavgaE?Rkh;#*?E@nelP!de;C#;pK_`2KbxOw1CZ@s)l-N~vP2!VxJvq3
zhOW3K4I0w*<wi>0{bCMXVTJe%+z!VuqOG2p79q?W_slpbBOsE(arI&sveyZ~-O<S(
zLQFv(nRV|PFaFBxNciQ0mxnH~QG?H)f!H}<$V)qb^2G$3(*t$oZl6hKb}Cu6)++`-
zkF7Uujv_~@z+Z3DpQs$}{r+<aS)G&2puqb>98!Gt18?|8Nj<hlZuhIJCm(o=%&?~d
zeo(vD2ow3we>GIWs=P2G!E~Lu`dRdIjfd&~*uMO62V8;FhBgY-D6I6~e!9unuKB5L
zO|JX0tTe0GSTH7;Rw(#l+X@iQtB7zdGGVJ1>3cj0XqF#pmPFUf^YT(hLL4k9;VHjx
zuuNON2$Q!5tCzJUn4k^YdfhR9uuS+5{Hp}Q8o+unU$l|?=CtJSboNN#^j+3Oj9ozQ
z2pN<Tg5S;MMca7C>2u}cx(xWMWOV!M#)j)cVAYnTaZgo{;kD>2mVlE(?g3j{!{ePY
zCoo`4uPymhX~J8ekV=8?(du9Z@ly|{zNmJSDhGk81eOc4*W+DXFzRq^o>ZzR)*`2D
z=j_^<uh1crX6@hpL?dAS!uLsRGpXYF8M}e+vL@$$`!hQKN1r%^=TFF+SdU*&gwN&b
ztS_~QZ^{A3@6H8F%yH@FG$P<S3O6N|T8{u)ek{FteBt<V3H%bZm{pY)Y?ajeykQaj
z!MMRp(6C&;6=^Zzlv!+J%vV%5gSMXkT|3v61XbDPpjD9A@@uy+pB=l#itD1}i9T4v
z!NiUi!Ai{50Mcca2f2UpIn|_UmdO7dO!PHlL9`0^sjfZZcTGiRWZO9T<2Y}f2w>WQ
zAa)Fn2y?AaG1;1(TD1%Cc8?@u7i9=5S54-EEj9hOU%a(NJ9deV5#iN)qE+TTJ$AKc
z@SBy}(NX<bjv(Q!MOsmLpA&XkPH}~e(6ssRgl~JhvnPNW1Worew@)7yJ##Kj9`62$
zi2KLH)GVnCWx41Z+Oey96<n@A0bp%$2$!Vy;>q@DHvJjH<yoQ{Ra?nDJz)j@N?7Ht
zT344()c)wS3b_3wdl6**NN8M8UB=1h=+0{0c?j~m_=QZ+1D>s88y$lL7HQt{c|Y@I
z-~fNZ!d%mn^Zx$>>;2$KT#=2`#Q5|)|5|>2`2r@uRb3~dIL`xQTKOn7h0<({E6{vT
zpKFND7R#M*->+)P`hVQbpP*n@dLkZg9-;%zcCFo&`t}<c(y;Tmw`rZ0J;_weqX#|8
zd+3~X!_e3?iY^dCWj9*#yYw<S27Y{fsj|K4eb-bj=`UUk?B`|8a&Uy<xb;=MnilUo
zoQ_Yz1~Bgc!4H53$8%xIpCvzW*u$-Yw?7edeH^KpMVF)15UHSMlaa#z#4bcNq;mFq
zu9cRrid=6uU2kJ<ks(P`QI%Cf>BB|!GX(7UL7nHPRe?6{9%mcp%Nb_2m1cH3wo~On
zQk}W#3&@cAP<e3~9=QXVrq_-=>Ty0yp;Ux8i(g#~`T_tytWZ`1xY;EY+B0wzPTX_q
z+?B|SH*Q^eR4FZmK8<ShX78HF`ZwmGz}g;qha)Bc_^P$d<8p<}`BB9~<*m9E!v3>y
zt}2eP6|x9Anm-(xcA9qUVG39H&h))x1(yJq7ZMIpAS@QR^?0Y){Jikj553q}AH$HJ
zxBEVXq^tl5*8|49(orXx*<juBRXb6vz(LZbQDuRpoNNCSxBpccN`9CDQfB4hgxKKw
zMu6C~)z~0$22?t%tPkO8-8e)c{H(o;>Lir-A|$1vBr0SlS{J^6%bUb6k2~-8_XV4^
zXHQorCNBHn#%bpi37f-;B~w@<pfK$s5*fEq3=GE8DvWR?i)vt_%MQ(0&-6zz5;b8D
z<Fqk9{EFZGCg|I`sh#W&Y1HyE(b0_9h%_Bj{gJ0kp18)llx$6IxJw3hZczBzo>;*z
z-=OMKr9y}<wBaqnt=XvY$U>*0Rq{BNHa&S`v>v3p4t@Tbm!t3W!~vP&hxq4yd9>aK
zwD-J(P5e#@>Kt8<f=%h67DY!h^z_9x-J*g=tm^j92aYeD0^H<K<lf#fUiNh#8zxM~
z%RkIv=m~7Oh0e~E(zj0C>bUhk<UFJj6cl_Azi(qqsH&(%0jrzgLoEBD5nl64o%jFw
zW4E|v0Hc;CbGzO*bj+{e?c<Xt9V7*4n3NdSp4a3|T30uO1(n!;4hsvUU2<>!=ec0x
z9Fg{YrN&|5Yvu@v0L=HN5_NtonaG<js*dbJ;rN?_MwH4uZ^cJ@TMCoWf2T^YvcJ^_
zM7?ma%kJ{chNX=EU0XvMzB>%0gU^C<NT+w1#vOKX`G;|TiN=RDU}f-K{B8`jvr}AW
zLFKyhzg|fC#SKa2{L=fybfGS6(f!f~w-Kg?!Yz!=`v*eAMmYG6MJt%w64~pk^v8{@
zdzKe{mRM`>?Zf4!#RX<(fue;y@2V+mVq3w9Y&BetH{4@?qNm5Db=t)T3v<YAj10@E
z3fOpZe`I-QkXAUJ_j2@DPt`olMn^A9zOxMbn$*Jv{&L=4Z^W>q_tn_|O_iI-hRU*{
zFEu=dwCN?tyQXDE(Qa!uV~1c!#E6FgL?ReAsvt1zpl>)Zo|iE2?!#J}(pKA<!o{~v
zBir4*oq5qV=XHq1Cl?F<o;{nW3$<*Lw{HwK4CU|6jzGA0!p=JEEzCA#i9nc+Tx11z
zLlQxVTq*O)sME)rqum{O+kq0=^q9$H`w%hkXKj3zfNz6U^COSRx*u|Yd4t))A+2wV
zs3*c6Qx^OlL*fLqW;VaHW`h!OVRq7N8gx(wi1hewEX$vd)j9#p6E3F8v-&E^FI$&{
zE76WK;*Sa5@eC{mN~H`5?y6RV*}bFV1@XNL7Iu37xfA4BYgQS;HeNSVjNGS=(MLJZ
zDo&tJu%v5|6MsBRflsYccnfcw7}<9oJ~Tg3Lf4)*kq)H#_;KRD7L$>!D5Er+caF~Q
z%?-(zLV@#cgh50tH@-9-Sw(%`OW<sMq5!L;qlbrETq<f2Gj3w$49Q+CVf?~qK77Y8
z+&&9jPDv%mNWM#`*MEB(WAhHp%C#eOl%e}4wa6T{ByyCv#uu)`*x|l^Zqgk<9m}9&
zv{$u3sc~glOjbl~n~9;SZT&I$YKYM_;OyifKqSTAd)NKecp6bRPS$R5e|1%iv?p?v
z+N&@U#8P!X{;EaC)Pt7w_><dvKIt~Rx160ev9GOke5cFd=2OA~rbYkGC|qRrKH4W@
zx$Bu4ugsMWI^JtszFVc}d}4}V3A~_ucJ|3%%hDq7V&YH=bH`Ndu*ZmY;Ef9D$I@(o
zzu0$tpKO_8sarsHUT6U-WW)JyU-RlY3A3?L6$SomY;KzSR<QCS_FPFQzKUzgI!I*L
zlTRdmXr}Ck%*6zSj#~<8f_OBH3=9jGQ)NGX@+iGE`k=T0u0+n&kB<XkOokQy?Yp@W
zi&=}Si8hdw(5eoOHH{<_`MekZZej;$P$SU!7A_aXvn|LO9svnaR_%`B(nv`fd$HO@
zKvdoJY2x~!m8o!~sXLz>LvS?0n!q<G0DD=xJ22absk<NUyj;E+y+0oC(kBLOF;K#N
zb+<(?+n2A~J1-LMF~iBUUgjLN!mqzw%-&;>9<IPv<JYWWhc*wDq<U-yz8t{gk55Rr
z-|M{enfAMM?7S<7m!vkd#I(qCN<11|zCX*;60`(I(&(9lDNN#K6%!WnKTrEcJbFC*
zrm5-Ts1;+!8_lCaMO#6Y6s^N9yaU_(TBA{9_O4vlR)9K_iLDkweL-670<!*&c|@a?
zm$19&K@`ha@Y`3jaYPM<F9&d~gOqV$DMzte((e4?*K-Q0WI9CTMph6V0~|WoyTEs0
zRJ!Pt@QMnl_BQz0-}SzEMdz>DR5&CA{PbVww13#ykg3O!#+{Dv>p|BFDAw_Cf@%1K
zqlF1|By1f&+eUun)ueRGddJEhJi*8MoK5TlPN;S-&-<iJgyz;n&j3L5>3lypL*W}f
z6c)|;YDJD;Ux77%RjVt6?dcn<BV`<3&_faT1C*G0q9pW7svH#N$jJU0Ct6dd8lO#H
z0j554B~++a(%(u1^Ynt3dYR0UlQ#2z)-2axi}g7#Yk^-J=onYhhYH*4RNyC3u;N4f
z*}9;{Qm(=xBojN{IY6yVvl(rl)qgK#WoJ!Ny?8#SieYD4hp5W))$N1Fl%r`eYqUaG
zSv!pjt4vVrFJsQHyo?dU1Q9{NQ!ormt+AhB4_(Qrdev$}#gSQKw_!ZvP``lKvCi~4
zBWfROug(ytn#4bkM>%cq7wRc66qzEH(!9lTWTab`HNy=6tg|k})@TH+zxUzTT7Tx5
zy?p@|=r1TlblrNJgVZrLgy(*MKaL`jYE86GfvS`sc@1g}T5TAg)Xy(UWtepZ_zC=}
zXss1#?MUHs_G;;?u{*yCb1`*j&m27#Jr3tm@$aF016y%88Xa9%Enhz@5XU}mKHlDi
z|0(Jyr-%dM;srz5#?v;!+l}cPp^9l6x6|%N{RQIEA5r^k5d*F=S9d-gS}?^|?bs=a
z0=+;4^en0gXPlDy=W8}?;Z<KB&zs%P{K`0NO?GpUszKN&m=sH9Kb%CBm(ymKvzdHY
z*g$D%Q&5rZL4hg$jW==I$CitsE;k~DXyA2Bb_Ot~)bx>~xIv@bjWOMIk@oVzP(tuW
zf2i<#qI{@Y;K6U6QhhUVr`YVdFc24|?D?798XKJKJu54b)|AbuEpY2|-FZ?z)ZM+l
z^CoXw4)wfV)AK-687!y&CZy~Ck_BWg*ORdo?TC7fW75t8wf`}7x>{kwpW9SC&C`#8
z4F%$tOz|drXN^IphAx$Y_f^34bzH4$`+0}v$o$Z7i^Y83d{Z8M(Qn04MeWdXulb$L
z>TZ!Qi_0t(h>WEQtaoL=(r;^2|IPJE$WX{Mwx3KJJhY^$j&PV2<razuHdh9TNJ7Fq
zy1|E2eN`ZwI>=z`h+jcwQ3g%~es4PMATA3IW4ibtE*GHwftz+VU0g&h>RF;Hsl9ac
z4<pYK^8`>-2v3j|Ye<s(QwSvvxKur@{%?vN0u5<{{3uMpwa#?SN!8_wcBywS!qPD{
z%>AsX6+*%IA#!{oOY|(vFXx!|oV!ZMk9F~?A(9%V2m_&Ee}y7evfp}HOVeHlC~?hi
z+B(9}3qSzzsj;8Ck)53FCx}yrkDHq9ncORUtnn)KPN5<zp9kw+=ut|*D(WCBX_U}s
zTuN_mZwUyfV%D*!@XIrN+t?Y^XK&2Kb<3bSCTQ;}Mm1*}V?mnZm9nyi5Cg8EvNBz+
zxc>h_4nIh;kP(hux1}zly&uc^c?JgVISJ^f16;}SjfStDGm({-F+A{TLB^*zG@Cut
zsak&6y&vmM6(QN})UF<9XS487=iA!W7PjfG+!AoK{c}693^+*;^X>T@FNohZT&z#6
z-69j^v-omsi4(a)(^PckAXX)_JD6QF4VoE~(iBl*QXDd*W@A~|hI?}-&^4k1NNY?1
z#rBi@YH<(5@F{LlKd#R%)=z3%%tnjLU9##coy~px@aUb|Gcx~xh-USWcbcve7HJx;
zO(NKa6f7R`s83`$#r0*|{n5?eyW4A!o5xG9vABV?tSofxS|`u&(G&{{u%jl+jU~vF
zp})jGP_NXVa7){q_Djv2NghZ@unYjOyhj-Q3nvxcD6<~jWqs3FJWF;~?ao<35R7%}
zYk;}9YUx%U^Zn^^K&$&$B<O~ZZV7kkBNTa6gA(k<L?Y=nF#Xmjl1mqM@Eu3b$Kz*n
zJOmDG;({N4%Db$wXJdkY`ILOlAf@{0OVCs+N13g4%uj_a4So!wOx1PapaJlFD0XO~
zj}C{G1f#TR4QD$K_p#8m?X^afZ^|E8ENu728Jxz`yY%Ae3ry!Z67v#%ureF8VOo&G
z>{>3Zk=4RWo4jIXUM0jWe!on_5^!-i5SD`^I%m1-uey&h$n_yKM`yO*hOzw5VSRHa
zOFhk?kV;oA`y2a6)%;-vUTmp;rDHL(KFT*UC1w3F``e`1fFJ!V;OeASd^rFLewK4(
zMVa4I=fhUmzYd@IVkVTn;hxyoGIL#A>cyUTs+DjuVHb@{8Eq)~h>(p=*v(1MhK==~
zPJd`Qgff8Rb$s^(9RbteHU9*PkTt4dlL`5U;{b;zoP7lZjXN3y(wK4TbRSmKHt4GI
z64T9z%&j<eMfOz~Esty-_o;3&+oVhqtHt5(+PzbuV4S~}@Z>SFCIKZH2+`rb2%XEO
zT{LC$bsux&)G0GEAR${fE7Pb#(qr~WIJuLt!J0!Hs6%upUG#rH-q+POwze*#&kD>~
zpW<-*#Q&o3Lv-{NP@ew599OYIr*NTei0YH1(d27;z1R@E@7BE9ywRYL&|{RztkD+U
z_Zih9J1W{3(8w6vN{%f;*D^|p<uG~P$@~89wzjrb+7;_moPTd!H@}E2l~{|3QQKl_
z)z(LAmYC}o>0+fryJ{osu#V3FJ4P*-81(NHoGtg7_)`Sv{@Xk%TgGJZ8;;maQ=_RU
zqYsb%agCDUotrWjoG>Olei*(rE@LW~ju2Iw(o=Su*)--JZ_(FB%%*yM=h^JhG)VRG
zDG@j4KR+Lc>2sLFsP@($E9%Pzocx#`vy=vC%w9^*Iop(ZdZpPEnHExs;zx*S#6ni2
zq4ubzZ*{<q_vrg)^<zj`>9)+u4^w^a^^Y;0(amM$-no>ua}3`c5i$9L$Jt$<AbvUC
z)jWa0-EQadne7306-MefHg61}Ka8q4t#h|P|1Hz+K#c{t5iV8B_VxmdhILYVSX&k5
z>gX@l&Vm+4!Lhu#t@U+$(AfC6k5X8TQTwIDTOj{`cfPy7^F%B%for3d(B81>9SWt=
zRG2(SJWgxMjPo)<=7s+2jDgl$s7joZLQ~vNUDZuiNJ<hr<QXpx41!&uEx2ER`xQ2c
z-I*tpp8En++|O$5WZAAUB0Hee7T4^P=m(4a>U(qSaWg!wLE+P#w-?k<j<Y%Vdu|s9
zx%B?rbXzeC2?$<nsAzf+C5)u&8L6y}o-CVokK<&w5sd_`rdq?yzB$F`afU1R61p-U
zUM=6a-aAhIQ~&Mvvi(0c9_P=~s!#f&RKL0Yn!o(&QOQ5dh=Cr}4j&JlZXB27J-5-<
z-<X`1y;(1{Z?rHo4?LI{0Is$G8HRWuXot1~?&<cFi6ZE{^Lno@pe1uWwCeMQ=ZjW|
z@_M;^l|BM#Cy_l$qo`v-tDY}u++ZRQJ9Pck-cxt}>T>8(O6ilr6CCP#!f>kBtdw=8
z@aLA7E4GHTmj&@DzX)6r)0XOJF5j(+T8)QZ-TM~vqa;oqGvZ!<;x^yIti^Chm?Bad
zS4_=-J|NwA9QQeZ(tTI>N2P9vTwT?zcROx!WH*P1nk?UbH^~4cHEVQ4?KseVxe{IA
z@v!ULvcKE?c6ZqI#P-=mp9eZHzfkWpWxeLY-u{>y^`c_qtIuq{;Kw|2R#s#<6{P3|
z`aK7wkO;`@C$IPtRjK?b@k6Z?K%)JiZ&=U}A9hGV&7%rDBQ1DXME*=0tS}d+$AX$j
zzbFy+^75L-cEjb2In@8IajG|&@w%N!7UxPHvaX(6TtuJyE9(4-Q|!f33~;PnU8}I5
zJ)2a~<I8*=Lm@3q>vn7?oPa-FjSYr0+RFaJ07_8-b(uWD=a2a%mWHVMU6ZrHo@k#L
zkKH$h%K`sSO0(jIJ#RaCW+qm!8k-PTa+X@(T8@j4nE8;X>4@QL`y^~EoL{K&$h$aB
zFs<C`B^;NG-5!yqZ$GOz!L-YA<`3`WEDUZ#dO4aB8C9V99rYV3au#JJ)TM6oy(crg
zkY?GIu|87v4%XB^&@IyrLNqZJRp~ay`udU@08W>k!Ec0tcur2G*yi4O$D|&5<1fSf
zsNtgs)-U7ZLf+nlvxc<r*4kPhaJS@>az7;IYw`CWpUyJ#$;?z6ub`j_hx#Na{2b|1
z<)W#<_i!W3ay{^r(Um1mM(}?7{_a?O`JVpE&^0Dt<H$Xe^Xlm4{j+9^G25{2VA%N%
za9A{Pwdy;k5v~`&O-}?%i+c7unzeD!!Q5}lED4%g51bkT`W2)pxb%Hbl1%aSF83oZ
z;NAPJIOl?dxH?wvhpU^X`FtOt8;xWM6kxoHW`YtJozFZ`O)u_@`r=e2swlPvET}=j
z!5mtm+6QB#Y`<&%nDFu*i&m9Ai`R+*(sUOco;vkytkI?fQ^8nY6Gjh?A;vWg$X+mT
z&zWA|Nny;)vket_3+1R)O9v=eZhK^JdX!$zq9a({r!(8$;67J<AllrKs4L0ErVj#<
zq|efJ-YWq-!Jm-cRjHo3FZ0aCxx`<WOHYMa(9TQFuETL5u_|Sg#r6i`iu$+!VK3yj
zwJIMGPVz>}66r6+x5^+ZUg$UN*Y?4BYBj4D^HT->2`MU368I#RvESeR{T(<(-tdty
zw((VkQ)d!rac1D%%7F8s^woo~6n+>8+ADCV3u05t?;_i^_Q|Jmg8sDTP$?+T4&x@H
z34^gkSV7Nlpx<Deu*@vcDOEP0cKF##CW@HJv^cyvShtB%?+A%JE-n#pJ>U55wMs-3
z*-sq-@A!~913tbt4@DsCh47lj#_R2g9&WA2qqH%#F{1wxVF}@oPzpAD3ey_MR6{mD
z0rs*@wvz4BEn=Mn4d0HI1AnjAW(X(im2qQ*C_q?o2w&skD5xYCeNF}5(hlPJQGF-g
zW~2^{&S%;(=j+xYK2b@MaKzIonTjT#QXQHPNZWcrIl6TWRcZ4v!B*sqhj(2%RU7eR
zkgbLW(|+b-w4_^EzL)hgSbL*7n{aeJxv-QHEEqQfL{M1z){555VLi1~LQ*1QSv~#K
z0dT#)xVXp>Up5)plmeurK5CxHD3@wSb$|2l!cNLO=l;wP{>x0O4+y$)|IhK$h@-%4
z2l4x@+DFY#cgc<2n17WWiqDF#-}=^-=;c2w6|Honm0RNfEa!H3baM3X+4w_8bftl7
z$7QjKBscdKaf=egxl=o#KrQvwaWmv-dz*)^mCbre;-|(F9n~a2?C;gpNu2nyvjc;y
zWK)akl%BVtS8mbr{g}@g=A~U({<1w{1|8xt_S~ebSa54N%ffayVEZ(3?Y3a!2Akvi
z1SzXAMi>+zs}X6H-LS2(Q9O;H&D)9d&Z@Do+y_1=nrHNwxNJnth?!Y#Tx3sJ4_;1Y
z$@rx#uw{Q+83>?Pb;OT#e;ZM)33!He;mQ+FpuK)OVEGioWmnq;39^A#BRWwGE50S!
zbpqc|f~0if{VG*`^NUgRypFQb;{k%#cT9yIq}RRq{{*@>>#JK!_kvsqe!{z*|6}_A
z#-WS|20KY|U$vtMn-W4_C<oZyH<^n&J<xiJ7Y$CRVFlzks)_%AS2eB<h8D}Fu8vkG
zmTyNb&#$o)^_;Rwi`-mkY)@R_e>-(lI0#W@3>iHEDJL0-Z%g_~>BAnf+-xd4^Z%&E
zgyP?pAX<%^+ZYo3YKO1_*z3YBnF}tA4{eNo=H^KErK)2DKhM*rU}An{dxCE-KiXs&
z<PwL2r2t_Zu(oR9>*sM#V)W}R>XUm#;4D-*H-Dns{B1BU9~nA0x$f5W`=w0ZmwwaL
zk;{S<=q6C(hU4QxLOGv-h@r%VUs6E26*gbCdBHX6V<Ad}%-<c!74mV3b<YTMR(~wF
zJLrqOZ$>RHc4yJ!T(U;JXJx?^i6E!lPLqJCHvu0h<(5`fvsMXW2e}%cY+=jFYIO_u
zvhOhS1*YCP(TMt~A=e5sujBTdc_j67hG=%;=ul`Z`COz@BsuWxY3Be7*#!|@Kiy9P
zxOCSX$wMnOsG_%cACn|1;7P~fdhOKFr(*-8^%K$Y5-bxsZY#TfHfZKy(q0bY_O>0_
z4j{+uO5amQ^h?SethZraNSpW|gcpnljPnFF(?$Q3%Mg6|4*B)a81s^y7PkmG#{7a}
zJ5g9L$|GTf%5R+~h2$N6U2evyRlB0gOG(U22~2x+u{a{k;@PR<Cpxagl;1`yT%|ru
znR(32`&BfwAG1*{@DhiAGxhWDu~tHXU|>g-;;+qo*L;Q6_iPtu_@W=?0~Q3y&5iSF
zi`eN2=r2^X-!u2$0y{3;&Vd~K!*{xg^^0x!b36&jh3T>V-8~s8c(oLAKU4B^7{A)@
zmdS<wkJ(k4DER6AQ{DL)(Q#7xffAN`cOOw{S$G+>OB_+ZK=XfIh_6Jca*}k2f56|>
z<FCS#&11)>Za))vT55Jp_!yEugaw#S>oY4q=Bdnw1eOo7?30x@3=dDYG^|Oy>llui
z&hc-2p|qwxe);*AQFY~-Jz6rc&<Rf2=bkFm<GWkP<ihMb#iJmYBm``r%}qTuJ_ZFm
zLoWP2e(~^euvy9Q@mE~9uUmmpx(~45c63XT>(lam2^$sbzp#Gsij&ken`ySl_r5B(
z?Nalks4>3dc26rMCU~AT)*wdYO_jMLg$pXzDWUu8#}<#(+0L62=;IITR!@})h_uF%
zANGQzUnzVE=U3P5wb5x<zFfovU~bQyT={=pB^GlN_0n5oUYsOt&YgQ_CHVnELgAgW
z8fN+-==v2^zpJCp%EY;-!DNhmf3RUCp<DES96!f9e@{caj_1^H^zu|fqsFd-xUoD6
zD$f0*gB9tCN5jWg4{_!4enqwCymZG!+1<vu22Cl@#TSRH*pBBd`R`3jHM&XzSJ;k<
zPsf+)Q6QXFEZy*a2)MPp-2~E?nG5Y_>+shHbIeEX2JUx&v>#68%*I>HwoxckD9#7T
zp~>J0Ih9rK#!vVid<HsnthBK0m-_mO+o7zW{7a|YeqMLBu&0_z^^Twgo>*L*kfbsC
z_+S3GgzLm~P?ha=TW0B-i-N|x`3zFcu}1ylh({q{Kii|Am;0wKx4;g%vj=_MaMyIl
z*!DTX2A7`b{o{`hN&1Xk!$C}gl<Z&dhX=nkDzieMjZBix?K^`EXmU7OSyd4J1M*c#
z{>Rya2G<Rx;j(~}bw%o-9HhNPs-j}H4xRFre{meqLF~sdI8bDP&_(&<(}0KQZKHHm
zL70kkRq)tNv7uZMu<n)-H&C!@$BMTTf-RVlpQ+3FSd0!PB~TulZ{hpRj%nf2&;=dL
zyg&>&ynZ}H6fzm|o~eEhqGP%i6~3abS`0M;3DV(G;0a0)+Y<+qRw~&0KLJx^E8{AS
z1=pjzG^V`z6MKe*u1UvvK?jWD@ik7`jb2Dk9i~)tJ>&*BIQSwQA6uki&0LqGnWX+3
z6e7}}1f4KB#|e)uNNTGrFD^az_UVz+;4awP+Z%e%6!{&(4&JbWM1GG_z<ICF3T7oq
zdC!9%Plz`*AJ_BBaqp?-a1B`xD7dXeR4pMvi{|(;Ov=Z*cq(Vl{;LpWh3a2y`Q|=m
zpFFbQ(6<d3i<$6kF)^g2J>DDg_3_W#LbvfP#E)ri`na+%@5cEQ_v8$*#^p<BOvxK~
zrB3Xh1WeRjZTBONZBzzpoJN`llbQyPwPY`M_r#au^Xtl<ytnpmiHIeV*IS2E+2%^V
z*KD+*im7yLBeHgIw5<{e5@iS}Vb#okN*MY!4fYmzid?J>&atPr-4@uYnC{rA@D_{O
zu!h&XL<;vWwRn}ITVi64=PwV#4EJ;Lsy>r+oaW?}^61xm)r`kpHfa}ITU|Sk4SwKr
zTq7|z_1xz~z{Nn-67bW)a4bIsC&Usl+p|iiIOoq2qI9s1BM`zZy5Q)nIiAK6bat?&
zB?J5)n$G)?%J=`{$38|JRF3RD^0tz_GfGC3>`jurw>w)LDug&FvvA1fSP92Ik{lt&
z%&|{)_V@n$@cj>r!*yM+*Yo*!c%rMD{bLMr6T9}_w<rAa@LK5X?!p`NxZI004Z2d@
z4#u$lf5kruINZOHzR%R(^;6g8{a8qxUHo&|@Tisu`+0jNMgr0wPu4wPUZ@;qTF3Yj
zR+4E9^TAeDY^7vV5j%*}yi@ak{|IS965^6Zib?I=gn{Ml;6gx6A!>JPmp|<i90OSt
z<f9GM&i_rIy^1BZS6-Ho1`>|)ffFIX?`m=?p?J`zZZuZC&sg>raht32c<vb1twg(0
zHHmGwkF`=!QsE?*rf-=XMVmk9mUoSzEjB1rtjuIe_DwUG>wA!1_P|sw&hMREy|x3Q
zqp^#H`dNd{9-vNM@K2DIw$&aVbGQ0AdL3_@>AXTseR(n!bzE2tHF$h=cJ-HK_>W($
z`3PZO)q?>xTkQ^tssa?Yrg`%A>A%J=Ab6ZKlR!RdNXXxvHX_y){Cug|h&@~R&*>}c
zGCRlH#wLc7)pQ0*Uzp#=H`MHiI_^xo66^iyI2vapprdhpK$0%qJb}@SMxG(#i8dpN
zIJg6y>i}sCARh`0P?fo{vUJ=_tEFUo8#a|c8c3>zJ=|66XcIh)&c%n4(~!kOaJQT+
zG2FxHR?S}cPJef4*3i98WjnE(_0wWw{_Y&of^mBVh}u9K_c`9B@o`2{DJ=}^J#}uP
z#}v?r+fW`WR1JTD9HpF2WlPjNXlC^hdE@kYLh`dSOv~4!Be<Zw9E+~=GShAquhh03
z$=a+>3AK<zN#!3V#XRVH7Y6b<ZjvvmMOtcm`^j;>R@z21YQXgcGb$<)dwjE=p(jj4
zob+3<&6te)6vF=$$HE!y;R)(ZVpGWdHXxWtX%gn18y2b1JW{e^@R;noumPf0{2{_W
zpDImQp)&b*z^G)3k&)q0QA^}94(D8P-4)T$Xt}u_aJsc#uddV6Z7<Dgi!A)44sN?8
zIZA|kgp)W`!`g$t+=K4#hu(Bq?*4>wFK<-;U4Wn(#C2cFn<~=h$a%uatu-1)fW)5e
z^+6}Bb`Ao|;-WYszOlOEA<Ihd)CXK`cAcTRz%bVSbT&RB@^bnf>@by2a>nOpbn+&e
z)c}P3?8`s%H1!p)qVPsEKK=Av={{vKdi%PoAAj<snBn7L7fHB2`ajnV^uRK|a!a5b
zw5VOU!qLG~0Y29rOuXFPr7Z=Xz$zlP3ssdZ-W_U3jW`8gjYe8o`)JL*D7GR7W;q*R
zz2-VE7AEETV@X0gyMj~de=7X@S8oIv4f(CpbD<&oJUM>-#ik-@3c{ZUe#e-bWWljh
zi!$jO9F^IU`@Z{`>WRtCUo0<px3c3lxkp^2I(KM~ZeR47wos8ucGjK)okg8|5I*(!
zwuz&(nmEokHpvyw`QB`Bl&&#7-1x{JEbzHB+j``-R??m7%uLO!&EG+{*bm$9uCh9G
zHHWx);UxUqU8BgDPVm@%*`pq=3;9=+*l^pau8Sk3%E_gb=@gj<_gpe$&NimH!U6&o
z@<FPTnKwHpw`TcWl3rg@08Y|PjJ!DJj6C{hHGh(IXC^kVdXBT#g1su+U)%g9g<>D0
zPV5M?AFGUufQx;q_pBjQTtLoJ=%Y*;&Vbpucp7b4Rf%28&-bLnD|u`P8l<PHeIkOp
zeq~_+rN_L4F;0~zJ~k-~>dFz(N#&7Ow%)MwAL*&>FSt|HHtIKE8#Ok?b;8uV&SJ!%
z9!baNgoy46KmxDaLi|wc8`alP&?3p9A2LrheFa*yR-n-<Y-;R+DpH=W6zf%T_axo^
zQ&CZ!!%yexCpdf>2rpK0CXI#-Sh=fb@bd`xXlbP(zu(ca*Z0Bv!q9hHK3oZB{g3yM
z-}nZ3#`-_(oYEOU75?Csndu}@7iBcTaNgmMS0Ic=sGd!aMcpEQAwSSmmN6Kj%uY}4
z3K_8?Ptf;a6}}OVv08XOFx{zP!u}tRul-9sUcx7SUu=vz2FuuhbE`-x8wD{A0n-~r
zUMjpa*!cTJ{K>_VbW!;u!4CRU4KMWvxcxZ2@hyzn8ZaOZ`A}TmNk1>`4zAJW?JU2c
z)n-$<RZ1UkDzlURM3x3@YkON%<Yv9a<98=v8d|>Wh57l1BF5O0x~h_NMD0r{YI~H+
zLbFh@d|LgYJYCPBpNKr!9=aZJFa4R#l(zA~WX+{vr+EohO{pM%Y<RZW3;L8dnA>zV
z5ClZTY5*TNwzfUAjW;}+o8@?uZ_ib~c)nRz=?hO1<J8iGy^%`ziprjJeO-WPrZ40y
zTnxK7wcp@Cj^gM}ZU=j>p-+m{PM3)t$FWz$Hm={wzZaQq?*9oucUayu(No9w=S;e=
z-d{BT*&37+SNSKRF0TQDS(ohnewNk#3hbD33Id<$mLgh{`40ar_Di!;_^B6oQ}qDH
zz}5QY<-je@1S!bKGuq-;?F+mWoAt35i)?nSW|Kla<Ja6nm?-vBEV7)D{!8cY<mP>E
z7RmefVpm_Y%~cI4v(ay&4?bZgipN7*fPgVpT7>1?TD7W@if`r}Mrm?&9N|dhh9pLI
zA9%3#-zRui11En>XKb6ywb_Gpv&@s!roZV3-^q46;>Fon;q#u~T3VF9>W@#EqT=iQ
zzu|WiYE#doj0<&0*|_{>N-`kog_7dG9?CUuE~pZaB%8D5lmjHJlElGd(hmk(j2f?B
z9G?D1n5Lw3q<bY|sKv5>vgn}9np#-_R=`c+otK!+<;A~3U=t&MHu$nl?IhbG>(-oa
zr*Ru3dPkAiJ~^XA*YnTre4OjrkAD)E8xkA10E6HVfMD+$c0Qs<NmpBiHqM=NVG|$;
zO`&U8C;bTf>L#aOS2Up^EH6s@=PDtdtV4Sp`HJl<yn{84&#&OOix#IVdzBm}0N^ZP
zT8Q+G!lHU!P6RHu0q4K8-W`6WB3w+j?j@f5&f4+sA67vIuW^t`$kITU)I0e8JiDf@
z30d}VMFbYyeos%8VT>#F<&I<0^rMIRurj-Gd}zu7`S?$?V=Mfdy+LMwv+lAnQfuH5
zo5291yooot>q`inkN6&DorL!XUM2ApF79rn9^*-2PNWLb7=AtX&@2J4q3GSdJF_@W
z5?V-b?^fVF3c2fw^#9m~J<KqrG&PO-H*A-Aa@uHS7rA^mWgG6JX2={3p=-6bPEs8=
z)*&(b_QsRrV=&%gPbYjcDa?xVNesEG$&fIqGs?f}h1jFXrcJa6o6TK7QR1N|;)!Tl
zhJ0F&q)(U5v#R?OCDm+%#wJyR$6sW}_IJ(w=nW>EEpkg<G}=;-6zYQyO!@Rp<@wp!
z+22oEC~Sj0L5cZwZR}cs2fq9>iF$G1^?rR3Cxky<$$dta<{lO@@F1ETyAc%ZS}Ef0
z9+=yQRq)i$EyKqK?)(GJw+_zP+vR3bF#*1{OQJ)=vZ;T%FLNSjQO3!9E%G$1Gu-z1
z!K4?GDWux$sa}=cKEKzn;S5j%&&DQK!X4HmtlRwEWsi~|W^SvW+uWm|weUi&>)>4l
z>Y1z2+lcad<uNi=(l6uBf68oj+c$jj!<H;~$mqX64mI5o{hSI|p;Gucv_FbjKmPd#
ze<aR#Q}QnBwj4SIL&dF~^Q_uwq{j?6W4l@%yPUGB!FK_~M%4D_>$}s_eGks7CS5F6
zG^IQiy2mf7=`NSXuDraDeZtWa=JL6QLP}KFZ(ANG2kW~pT3f=8vB6!(+rm%XZV5do
zY1{D#dP}cM+A{;>l?CZ2VIACw_<-RXp_kwZ_9UDTQKKY86r;isok(C`?MCNewJZ<5
z3TRMNAMTl%a*9`Wb?J@nc%Nk-APV;Bvp>MBReEk{toeU9nbcg8h@RwBjd_>@jS6J)
z;Sk&bzTl+QF$&hzGFkyPV+?twRADbT>r(YPgAvI^Pyn(s-!-Pd;y}7!P;M};hq<5E
zvz_m!dbLY65Bzy|=5}lBY32CyUU@S;g;h(>7;b+Y*!c``4L_Z~YQ9QUzsd$p^`uo(
zml>~BXXj@8o>C#&aA8I<zBVjwMe6d2nb*EanrK^0>SWOECn*2$=-9|jv!W|$0f?Ok
z_Nf}(K(S=irXrM7M8MCmASq+TnmZUf`n0t1a^>IP4r=zVJDn4&Q*8GKPb*YoX1e#1
zpB!OrXmTeE)kXiE#G|2MMJ%NMrPf#%{~Cqq8b`=jd!p*fFU5)X)e~4!W(U}_m<Qqt
z%cO6-kLa+E%UP-V2J4dsp)!7~+gx~dqoAfTpv`I`UnVq7ekz|aMoRpfWI!5>HRKE4
zFD8nbID3PKoLAAEV{`iAVjxkTK&K%2=wq%&DiPn+)_7MWdFf==H@(avE5Sjyl;0T<
z?N143d!4`l)<n;?gVEJzM__EKzT7BkZ<!0<8M?fa58jV||C%e1*1_D`uHu<m3o-?&
zjU)^eVCnb#V*I(sTA}6@G3p6Y88lV*jMDcJ^|;3l`fgR(8Zs|As-bWmY0MKCrQQR9
znw{^!5)3P2U=uL&`Fp&>kDqtnoNbNwjQiBJwY_$EFM$mX3ZBxQ7H+&-@BxZk&n=^l
z{m@l8<-)4FyK{83-2ZwQothzLC=CIi#U-Az9$$y+wt^!Z;5<we(Z{xWA{6dQmD>@(
z@mGa<VEOOXK|_TV@!}GjM3O=}hy+WSc@a0_7mEU3W3%COgVW>OhWuID22J1`7A^i`
zQ0Adl3ArVQLrJN7O5n|HO8tk}VMl@3FPpcF%zqv}4<8FLmD$ESzkaCuhzbxrxnzg}
zGZwowmy=h2-_4((QY?HUB7%Yxy=P8OQYO7S{Fi;sFoWx1Uz*3>-^Q_!e+(4mq<iYJ
z^OC~&lM*vI=w>(UZ|fBqK3eu!3t&mZ#OZ}_K5QKhiVh5|h5j6fO%iPL1`Zjb_VDwF
z`p>=|#xgbSyJ)ATk*TiMMqt0bumgztp)D%1da`BF4E)95v8f)pJPai?fo7?g;))kw
zruOuWf~)?g>k!fI@(hkWk0#Ogk9@QKxN9QV<{-Z$S=%o*Hs`Naqz*BzNds>nP-T+Z
z0<z%Gk{c;)v-+QG>7CBl+OF)}x8Es^f`q=aNUpY<Oy9&GYgC6_0h<D3)wRCW4i^%I
zFZSPD?i>^b-;N!Qe3{7vt65W=nrFBe`UYYDes=B6MB8+n1K4kfrD8z(_X`P?@1NMe
zv2+T;JSNU{cdOR#fa&<l!b|U~FlpAVL!6acF1|hTc3Ot;gIMk-(iNMl)BF9KlUv?b
zt7?%aa}txoC+^KefSs%B-&VE}Yk347-`O~puC7ff?>kr?+fnwJ{JZ@d)YH(+XVT8=
z&$}u_UDyAu`_9R(EMQ0TG8W~jc`L0LX|)d2v1r={hXzJXNenez5k%89(sn(iK4Z>G
z4t~{08Qh@^6L0&6XQr&aaa!qrpzy`Zzb-5_&wqh~Z%nQpIPRQbJ~E_$tPEO|D_Kq1
zJPL9tZA+^nkkcshEp$%F`5LMxyz$3UFi<eoVuLVC&+FRi0r@$VUvo<AsfBUf78Xst
zukMtU{BX7z%XR$<Q@FPWk)XQ6Hk-iNQ~a>)4$;HAPq+kbgqimFgwmiCedNo2%>LYn
zyY#udKbGtfkmXX-d8(J*OZU7M*=%rlFr6{!ce%FcYLUzRPd!=Lu1j8_I=9X%g`0`}
zz8_tt{<66qt2TQnUAjDi2}5B%45L^UO}n+(>u9m*^_DPm%B`PC!^+}}+IL1Ua$E_?
znwuqtEFYQ~jj?RDJB#^1jN!rpvE09BX0l9VJekV@GN+3jyoRZ%r|W$sA_t+WD|o1m
z+bGZ`+hy9HvjOK~s_oEXS<`^q|D*UIy~(p!n3or#rq3(#XQEu9sR@J0kx3_nsJ_=P
zP!2`*zJifLpH}`s<u#+xl#a!NEw{6Dreei9LV`A0TLSk7%x6QcmbP+=sWY`c$WX_q
zTe<^|0w9@dMM=rZxP+g~bA&q;#mC^-VcUdB!&nh26%D8YC-rxBAW4JI&nM?kPAfiL
z9jl!+$1eWcv^$HfZd@Lu7jyme4YXV2(Fc(Yjv3l+?QE^aPjh+&+qkrI=wo{2qeW8`
z*|xU0=uF`>S_w}z9MMBXCK|sc^WTVm`vsUI?nylH;4oqZE>IF&+a6bkc2|MMrrOi;
zt^=&x<|Ag!UdlOR;&AuNWgtZDx)taDl&UCaer-sc226}_5}-)(0WDrH6M57CQeF)v
z{B|Us5wONcLj~I8ItscU9?t`3f4a{##Ckx?VsY0}{fgtZZmi(XR1s%G+YA1Hz`t~#
zR~JpYick0VUSIigBsMmA->9MKC*&Sfx*s#iJ7zo08r-AL{#a9}R9#?je_A&%*k%g2
z3Hg=eIQ=I->Vz9IWBt7VCS@=IUc(|K;Ss0n<r22-_K4DAhsUNSgC~BM_+zzy-3(n4
zZ{!T}*UN<WQ%*e4^%vm(Js1Yq4N>N?_wGCta2k;ScI*F~_cs2yI^!zT6-}a0ZSf5t
zV7y=z{ja$S3;x?c8^0Z@`*?ggzR$d>vaFmO0V5Y0yXjJJb)IcB<*ou3Dqj4XI)!hr
zOpH#a8*med3&g;C$g};5<98Tkov!&UsVAqHQ{J|j43%ojA<nZ#IcNi4AL6PSo9w=k
zRI?AM&JXm|yqJGp?7O*HMn;t5)Xg)zRTUgr1P%4kLi>AezHt}Km8@VHR!p|!AA`wD
zV-=*k<9U0_j`W}+6ujQE)x!Fdza(#!Ov|t$f`b^Ent~=)LM;MTNaU9hBw^mEd_wp7
z6qu0~GcDC&gj4zOjsJjmNvC5e3;9w8kIhU?r&1en-cKHoLrjKbMl(%tX#C5O95i_L
zw~0NRk37zA?mEB;6-J!@1-ahJQ2|k5r(4sYe@=hOP>`>CIwfPdsZ!|<yZ1GL9?3Ux
zz9USW!Z&msa&_Ep#=Sj`!}wsyKDWSpi{C%ltc&pVLuLoU1OW$_uXsoB#rNi}fv&#|
z$>Ywnp>Ik-tRimarU7$&`kks|<(&JjD6$2UK+Mh#7rF5YyHBmf-|S%*8@b9q6a-}`
zsgu6xJMyyanTh@SPIops$|T{^t(pREsIq$HA-=m>>Y2=Y^9gBjlz(*k!nOS;@7|g4
zFg9Z7mW+#sQ-D8y?&&w`myN~z$pYt^-jO~2v%hae^%%=dKj|pTR{A#6d=ndz``?~?
z8Y|_x4>#+iCqbnurO7LGJjy~msl@;E{9m7klB#N<)aAPO)eh1B;F#ETxYf7oTBzc=
zF_SB+$GiL>JW$q7Uw7Hhx3(=@2pbgZ@X5?oy8}NY%?LQ7eDi}GaxG}V`Au9fH0E$s
z>+3sRp>LA*6>jk~l=|eAfrB~nzq%$<b^m*X7QI}WznCSwR=Y-emt7w>9GHF$WP5B$
zT<sC~EyPykysbWeXZs`OyOUQD)O2=Y1a_<BIZ}n(a!cGSO%1LK_V7$uvMq0*$q?^{
zV7^UN%K&7kE)&727P-{nG?kNv^(3uRIk83I3mb2vCJrqD(?4>}qjwz-O_#ewE~eyV
zLs*$&lD7OlBqrf3MgEx5UbQNcW~cJ=fNxo;k*Nc8%r2+Ldh;u@z9d@PyNzy*%|{!9
zDzLUK;{^doJWdh9&49iGDxs7MkcEZ$1qCZWXmuukk>XhE0_po@j{=D${$~Tw$VTgg
zXu2v}*?0bz!CgnG-shKcBvKE4lvOu3>t3zTUxjYZ$iQrD;$y;{;z%L>0S&=^qK{->
zzIe~Zu!qT)OYG6GngoZMwuJ@SCNVv8a~JdHscf1e{G>j;79XyM5Ie~D$NOJMlk?64
zR;OQ7O=I0XQJ*6TRlCvY=!uhv(2)HpduuXDEqA)m!cs7aV~|b`{+zZRL-PGEl-uoW
zY-0ko=+#C3AnP$l&)cOst7p^)g`uS|ZHrqI8`$Fm_17bow$1itGc7LH)!vWg@tf$d
zLF^H%%OrC5NW01J(Z9r5&b@Ei#hw8B8S&#{(7N>iZ^;=&)qJHy<gqx{ea^w~;zk&P
zKNu52Wc;cmiAz65N@2cd_~UM&Udg{C=sgWQ{5#&`9Z8t2Y2O1~8UwgW(>WQ9>*nsm
zb)Xe-8R4I>hP<o8Y@E&yzGF{2!h>g;TIE$Dk5&e!*WF+P!gRRJKgalTpkjoGZ@cIR
z&J&}VAe_-nm<(=c{2NhyxD!5fg7rG73gfah<)BYPN3&e6p&imL$21yUk0E*4d?A?U
zf5}h+xgd*=@(&KJBq!!7nrKYP&`nRT@18sUFDJ;I47Qo!2}oL^2(-8w^t$Hz2hNGz
z9IhKUE1}S8Z)tgs26BB3R}$Ahhz?g*vO1Xz$*1irzb<-<L(x33FooK)3p{xGBsEhS
zru^_u8o01c=rOj)8Y}4dHwrdIjMVDmTnYQabo{|?Ol#GH`+*gIv7^QLzE7^ZV<-K<
zr_M9jYYwNtTI*^*UiR0N+H(f|WQ)H}Jl>hDPN%dYCcI3+Y)|Y$aBx$-Sh(wzkL;M*
z`QM(RpX<(w?{}Cq;1xwyLJrc?xd{UW?IBXE<re9Gw0lg$l<l4ka&z0s$~v3lIvYz=
z==zy?$eq5%L9~np-a+hpIkI_9(jM_Yz*#WInYi0bUE5n&-KJtGAH#<%kIGQbhZ8$u
zFN(p2?&#_!V8+?r_jkO0|HjGc%7NSLU4sc;f4cSE<^Ih+$JQJJOza|m_|f@CDNHz=
zUWh^MY*5MWatMd3u0=}O%AwLlpZw0_eRZ<5RpmuXxgbYoegOGpW=mW!6G=A<z*Oeq
z`Pd}bw%z+1V=fsWYgI-8vto1-dApUqT-Ib4HMToIa@zEMyu$oK6GBQ?!hsxEpPC_R
zF3Ee%KDFA+r{HDibz9C^9X#gjH;eDA*+cq@{|iwzocO6QC1cYCJ**Lkf1{`E`_qE?
z??C96l?Ex~$4Nn=Vz&Z!6rrOl@@$wGv0Z6*XC_STbloWGe5cJ$A?g%N19m-vzBCkF
zRq+9Mbc<_uB5d0~wNbw3cjZ!3Y1$E)HuSy!BQBu1x!Hez?Iqs?c;4T`Z{Tk6gxw<z
zI{A;$T8H8QsiFRa*YP=>8>^q=3*(<D^~}|WEKkXak;c~zCLmvZ<bTi3KAiYz*=8*s
z-lt^qjol|dG%yq$ZrgmkLaqF9|GAhi<x^fhU`KhsmztbM2tw+{9-P3uY?4wJQ^ZgB
z@Y$fSe7qs|^bwge+oQMN!;SD6e8k8JakJ3=7VgXk()#Jq!Cza|oe>t;QyY5jm;N|&
z1FZ|4?-mcWkPbCJ?m1&yqfVR=Y|Dm**v<<SQ8{vt{{iR-Avc?WcMCXmEZZ_rNoh#}
z7_plv_4N}n4<IjjpjfKdZtRX1oYE~<SY&m7x_+6Z{-p1@GKWz!Szkvz_7*E<phbXu
z8PYD*BzBv?ey@S_r4nhT23*MPdb!CB^0jb{!seD;$BV`Vd6KVX#k&DAvIcWasN^MA
zp?+=K*E(dp5$6YkbKy1k)^I?zkgE_tf>^FQ>pk*EkU*QUn<;f-65`-1{HG-?PgP)k
zEy*%+_|FRtZN^X{GNtIgyUMLp@Hr+p;Wo=!`VC#ZU-V*D_R0ClKH*I1am{Ra@XkPS
z5AGjmV_BvqiL*`~G_<XP)QaUvdS2Tt6Q%EOo6SI+y#N49S!_6$Zf*UeP{6W!V=+Vt
zwzL#cK+oh%T6KqV6XR|$yzvb##BVin^LRcg)>n<4-x#^Wfj{IfECByR*CtM0s16q8
z$pw{V{Jd44b2ooS=AW@5Wg2Fpe&Uakcjq?^OV?Bv&XDpp^tiF!fHL@9D1GH-<t?u8
z`cLXLUCvh3WAIiX`4pEI0a(yc#wt!JwY7n=$Xz?oRhD>lf{p65y@T=k-7?;!Xon(C
zoNj$z_#enr3qQ6Cznx5}ZIOAB%OfS3bbfx`AV$lx;$EYY^o=fpB_Vbqqkiy6d^rlP
zi#LDFQcRv4G5PZjOeM55nD1*r>wWAoBCB%A9kFf@73NP1Y`5?N&#d1a-iBw3Cw_|D
zCSI*Feez|ydDQ6Op~kv-6&g9c#ufPLzGRp2ly|OPi%jbZ!VY*+I91Vjq`p(C&sdHB
z$TxMBWwTyl{hBWWX6<fsT^mLYBD*&46(@n`kshhz^#sU*iW@3AlxjqwzePQb-MBg5
z&B|LJ{t5dXi`@abF6}N0)y^;GH!CVe37#EVlrg^EscxIa`KaQa3nMPoi@!au-#+-3
zz!&E2;U04zuRPP{8={VLLy_@-IgQ5-zXCaU2}nDEY+lmkyc+?_#2sol311nk-Hfs4
z_rEgheATN!nv1?Z?LuQl?Qv)69oKpdn05Y8c2W@0C;`aciX>l~n?!gvZi`uH65a=W
zrDIG#9LeAoJkyEE*%EX5s(zoLcK>5jvYSQlZ#a=7(6jK{3M9;NP>_E`K)b%K#j8k?
zcd2JQKpJ0Nfwn8y{o!QpgMkemW1p%2x~JE9thy<eg^+R&-zXvIL(~gc)K15Zzy)EZ
zi$$*U8E}?}d<enO(!!krsLd_h!B0P^cxBh+@oaCVh6GGWiNCpH1Q|%@w_<5Me3BU`
zBp*6!6(vv1-!~ygQWGNPN&I<8ORK0*jY2fvgp<>`5xHf5I^N0?)!4rPkgoGYnhR&i
z6~x0qgQ-cENrn15bGI05Oq(Taky{?A8)yrF;KD96U~c?W6p=M-4c$=zqj^nkxx7m+
zIz&^_X?fzShUsBH#H2C8rC)QD$I7j&u&t8GFYk*?;sOgZh0`Sw#k)S&`-aSw?5asc
z<Xw$3Tk5Jebskg)J}eL@7PGtI-5&M#@A`u(XZVmhuMj82i`JajTAf+pq@lW@{TK4U
zTTL48-+g_KjYDrkEH-gpE{fU4Y?bZaorEZ_VDZYovWK)C{?&6`?S-qIhJN@lzT=Sl
z?Dm|6ch7V9{gY`SE-xdmTeqL>PvVbkIFnqB?6lkxcq?WD*>&Fq@#+iiuAS_P_||sS
z-)=d8oMu|Q{QoQg^5N!3_Zj>WFc?CG6=9_vRFUFo-O|buzy!p0`$!_PhyRSHDlw>C
z4p>e#y)*d{5ab#h<`0~2Px@J`_4%C~OuPD?znIKKbc3~W*np0^R!~cqUDo9EwEpLf
z2-{M=t5Y^GW2`V#WNJ(q6aMLIPkmz7^#x?$0NnK#OAYQhBmh6G`j-gQDYeo3iW@aZ
z2=K(wbSdVy>BT#l*;2IdOmg!s55)gHlK!w!M`ET|JN%)0la^Y#9rw+<AdBZ7so+_-
zfsaVoS#G5^6mptV&zVHQyM@}87!TDt@QT$4jVpLM2${yH^^$9Z{Mn63digc}k>1Dj
z7=KLb66xs!31-jPr|1ko?}13M`dfdh@H6ku%V)wjmcNBf$&7-8=tQfD!L!PC+Zs)6
z%q^qe;We{mtIBlCc}f{<vt2f_p5>{HY3&b}B-8E*#82LW)0UF*+S3_(Je6Naz72tt
z7^Ef7Z+!}zPM1oc2A?~8+)M1b#U;YIiCmZGw6|mYkN*9OKNg1qV5sgW^by{w5;ciG
z?g5XGeFx-mZ_csp!EnG-qqeCSu5$mx(Pk2LNKX-h6}yM6BGB{mFp}AmK5EL1ChM1!
zEZQ-*hz<JQ@SBx)FG)(8)W{6ZXcPANTY~Mw1r782KgKUfRyH^O_4#(%Di92&Sc|DU
zo{;FSExK^;Y2g6#1n#(`qF-mJtY2W?;Cq}U*EM~J+a9&ntD#vgZA#&Kea3J%&nh9(
z689&hdr?w_YO7UtS4e*?ak>}-=&hK$K^>Ohwf1~zsolNJ?)<!1a*er^8>;D4b@%Cz
zx?cQ1N3e?6P0EUGgn#uW9YIsC4@{7LSh1E2asx@uO8{W}k-+&?X|sJtAC9^IOo9z*
zH`2rJV6qxo%4Wl&Et4NY;4RHX8!U;9rJzA?rUlEU#x17rJ~EETny0rDVt9_QkD##f
z2eVD69mQf+r{cv&yc3prhKZUn<&$U-&~nxm9%{_R!O6>D>6FfY7X#&~tWZ|XHxQb}
z1>JlA2#`11Y#X%QWlhvD$NhJr*Vla%a)0I;CD~GyaMtYXJ!j~Mt(ei<Z~X+JGhMlF
z!kgCb_{U$=A~HxAJzrCi1T^_$Cqlxn@OjObt5)-aTq?6ObMHiVwLA?dW@4>sz}(x?
zWcYD`YN+3pS3Z4NFmVpJ>LAuM2|hr=`qAM0={*V;n4^>M@>SBCW1((^?B09sx3+2L
zZVi?)(WuPOf4@Vok*-yL7^Zwn*2e0|+?J!Usi*Bvxil|J*Ie{qLM#ZtaT?~HyQLbr
z`-J-GWGmlh$L&hR1;-9?kYw0l4SKllu<o?W+aXytDSO+0eoi(_<Kp7j4zG-l-p>#M
z9>A(oZjh;Wzy0_|1Frq7F$hB`%1<EqT=PX`qRfc7jsDw#+^neth)-gNv}<8ta$crB
z@68w4rC%lW!YHh1K7P70SC-`8fIsY;3gGxIpb>)%PD_e2>9Dy;_8K~nlWr|6kO_8h
zY9|ulZzz0(<y{?_2YTQ%wAhRmEw)o&YV{X$%XsK(#9RN!c0Y+{T3X`d%)iOp2<k~u
zB74}td|AwW5mbHToq_OW8}Gp)Xz^mmd1_6h6F^jtXIw7YI#^u1q?*5YYURZjxa`30
zv!&v5M*MC$D`Nxal~*(5+vGn)>tT?M_SbYD=~=on%2npOD!+0hSbHl{CS3q8Ug%GS
zKMg=<nee*OJ(4WV@xe4XlZp7-O$HwRFDZzu&-Sjd6<alU%%Uw-IF?fMeD<l{$)2~{
zom*X+eh6^vh0M>t+76!28zYPrDCdu6$Ly+$mX@sRmm8F$1SHegx%F-cQk&I2lFDFA
zhUIL<jt?lfBZAY|4?Sy+vA{94k;+{VV<j(5Y?GvOhWYU{g4{RXgo3ac`evL@f83I$
zB)9_3dlS(96FAtdMMyyouxt}fFJB77a_*{*(N-4r8~3*Qq?OQQrW;K7VBeXz<x*-u
zAU0Eto_U1Of*M&;@&qmZ(iMbD?4o0>*xu=S27wN&USmt&VBml|>dD7+l6zh#Mon@D
zSif(UfaZPo{|LMWKgR9X5O>S&Vys`B<{XDmXkI>!?{ur^3|-d<dceb@ufy!50{*;%
z|Dd&;^*)2I%Ro0Ks7Q~bN83zzP05DHHOtM(7cT$#td?Zj2O7&{4B9tZQuCzquPdb$
z2?v#1VRGct&PIU`PLKuAiME3anhH!YmE)uf7-PD3G*Ng?{_U68!-<>0-s~igY=b8*
zdI;rrEM&{>EbCj|Yw`>V3NA?xo$&HBYCpbs-x7X?JtJOeLGQxpVE+t}5qs0c@@zk6
zRKqUMQB=*b{2Xz+&k2&O99gKj*<7sa=Ju|0tEGUic@|*WYRfs-(I<Ze))f^L6~iKG
zJ8u|L6%Tfgk1zL+bsZ^PwTn&pjtta`vAHfYo_v^;HMqwk01;-*sV0Ltz?*IUj4Qe3
zM@L~<$)k^>>N*0?w@!9b2q`C3mF`+mzMa2vcl@v{DUJJY41yMlLiWz-AeUQL=W@i(
zecW?N55THJWiD{$rYWJrTX8hgsw0F^@cqObKA@s<V|Ri%+CUYM<)pv4pYjN#3l77e
zrXSu&m1Zs|rxq6pL&DyPveqx^9q}Ds5^QgVQ*<61$UvYx>K}Bue5eZ$JivPH7h?_H
zd*b9QCY0=;833pFT3FtwP6iQM=GGp#7M<t-yk_Bnz=Zz(WRvCb!L)D229~^uhMD+u
zzkA|P(y)tXbNW)TCD4XZ*c{~Ru&%8ZeSC!p3%bV0NXo(vV`Z`S6IYPt-h=eAik2$J
zG-4WTu9UGc^+6dkYwQP|^b757<GS_x%V=|{8kXTn5NhtXcfQ;68t7Dg$ZEca4KV@`
zhvDIKhl9G`j-{A&l)PD8XA}uY`zp^Fv0=qx3PQ*$Ri1RIe1AhdFVu2OI04n=Q9a1Z
z^u`q$6Uwd0jpa!rm#$))Y?=e#p1q@A@{|4tux){9X1YxGgIRJ^<Au()Wy*zL@($9y
z%}chh>#Cc|)RYj4fBMcI{Ni%}ADrbTf=~BDesi1#GU353?ZM|VoX*AktUg$U$9M}O
zay-8|?OZ2+Q}(a2=XW;l?$+z<NlW}Ddhx#c5KQhUj#9zT&zU-mh3^;tgzbBu5bu$9
zG*$f>a3^+%3Zs(=lw@ENtqr4V&0LA}CgGP~)~k?DPK?!H*N!&9P8R@6jkW}>yj_#>
zHu+@Ab$dznlEJSpY;H74p@jq`F%39NHmtLn<=XL0<Ll321Eg6?-6}F^Vtk|v<Tpu6
zheW5)bNr=;h)Dp6KdSpq+Z`6At+u#sNkB4aE#$vNn)lx81C?2zb}9RKNA=h5MEyhu
zsQX&Hzk!XO2&mnR0lv0k6>|0KQl=x64N;9TwGO%JT*dmf52lqI^+YoG-<mYx;7+#T
z+o^v%Y|T7}tLW~P8GMQvx0i&|-+EMDZlvS(#hRs8d-86^ZqVfOgqC2}v9`Ufh*lL=
zB_GUtmj3<^y*NozggQ6(TxTR<H^a4O@yll=r}eefQ6nyuW8k@ME1)*a#OPe#KGeoi
zs1hM<m$hpjvWeLVdBe%U!KoH5n^(B)k(yga7vtlu<SU$A+A$nD71+G<TS=t@@-gYN
z<bs5uxbjf`MX}g>CK}T5_nf?c-|CHx74-ACg13*0_!;-1)3(yjmArQTl<Bo#uH<QU
z>2^bNwRokR9RMtYD7y`G+P)ByPxrj-xXY|&)h_CIj*0=K6zUpVXO&vj933;5BBQPn
z)jC;qCpR4LDvka)z;E~H7M{0w5{CEtK~Y8|n1Q0FPm|AHMs54y`h4l=W}7JfYNg%T
z@r^CCzN=*B^rNztkU+|#HUEQ`mwE-E<b<x~YhM8xJbv9xO41Wy6&7*%JRJ3%pY%q&
zhOGYJ89!Z>EITV>!;GIu^ngXm0JXZFTZb;TG*bg6T-tJ7tG7g5iziNt0s_<E{^=-1
zmM{Q&^>Mdw`ESHdREWsWjdy8!QVIGr;vmTV#K&gpLYtF<2h!_{05@<L<B0t=oe7!O
zoT@<8rXCI4d3-S_h+w~0%y4PxbGcu5u~e>h;mc0hHfz;+J2PE9ixKVwwUF3xVejZ*
z|1V@dh+~<Zk%Gl~fS<M{(`Jg%Vr`%C)v4lUhqo03;-z3K{fOix9|0OejrkJ%&UJxx
zImucQiKxq{-<PO@LVg{WcWifq(9ohLPaxdgs{OjQ9T#Tl?x<GRje-61m4n~MJh73Q
zRH1)YJk{<)9An@F(MKkTgyM^i4z|QZFn7~&OWl1tpKK55$<O(dZ6ajvJKvVp%BzSE
zPvGR9vvD7Pct-k#A)P;43lG9WXcu0+y&nCB{R4}4@e|~JNzK$n|5l!t6CLSejZDDh
zePEd<8^Pe~4c^U$*6?HQhZ7YQVE`LOo+&|;y=0nul=KF<=%eT*WV5H0>13ITU(+W&
z*N)?X(p>!}a$Z7}m#9;NKhAPO#>Vunjuk=qYuP6q<6rAp`v!HISuQ3ZPr*LeyWCCG
z({d9fg}LRP|EDX>UNo84Nx#MsRRPd|&e^ihRfo|}vW%B60u*(eD<_koLdLvtKDgsO
z65g83AC~T(r7K82o)H4Ah*X*r!J7TEd|I`wgu&z4X>|4R(NUQ8;fsz1{Q@M?+1GCw
zH#$*I$oDlc-QM<KbMl&(kMNacU^psq*0vCT+INJtZ$!RMAe(5?Hr_(i0rJDp2CCan
zU~%h3q6xJe_r;x`1XWjWpiuvji^rK0Hd9>28*mDbt5armsldO^y@QBObk6^l%F_4m
zf$4D6qVwt6pg*c@H^fn&MdSQ*vnJ8_biF4r@co#$c#-oLQFPlemvY#-hj;KpXczEh
z?v@{xkA{5QJw1K__X}i-^5Ppf-kO@o6T8DppV&-Aw#pdLmmPVAeo=`yTH_-wuA*Y?
z&X-tv7XQ{Ka%E2rt$789Z=l%dmpV|W@MACC<j`4~&kYmogJJW}+bURkG@jnLO?h`Z
zB1kd(eDSP&OfAaCyJjS1|M5wi-Pz35Zu!;Jg_*62k{w;YuEAqK!4)-`bRW!#eJ?k#
zE7Lz9o9z5BAy8OX7$o>0CR<;Rmd)-S`he>Qf9%<{&|Y|x)oD*z4g<xQ!jjalB}}WG
zr3ObW4-?-6qBr@o(GzNyeG*)%p{jYKrw$)ryohDaoSTS;BJUTpR9fOJ^W@Y2SM8c~
zo_+^byS9&2cL4NAUZ;MR1_=d&%uF2{a(qpLFHKOBW<!JKJ=Fz${ADb!d>H&a%Y%=v
z-dr!yd?wHCn|p6UT@EU1P4Y=JbFkvz{S<q12WQYzCXSB|%bg{yf*`U1JL3*)-m&)4
zYi+#s@plehL04q(9i8d{stH#n7c@wcnym;4@HM6&&jkF%1xhoeZrzi_gb<vdFvqwr
zO`X0Mr-z$nHBU|dAg<w!be!l06qHKNR~8gp9b8PD)4@ib8C9mGQ-f6AyM>s(RpjF#
zK0)$uBBK^pgk`8>QgR9`CWVB|Ii`F%FqeZ#pF3<2+V@i-pZ@N%k+(9OWIWCR^gJt?
zTGYza_J9o?UONeX@6<mty_YR9iQ?)%oZ0#_l)--k&CzWu#%mo@mUdZ(!=*A2x5rep
z)CZ0_%HqIjOgAh&h3*>@GZ?h97S&>PsDsm)>AYL&eP3-cf?TKOLp<sh>Hf*eyvX~J
zg$HR%ChiK##7YrwAkn@z+ntb3G57uo%3qI!0yrwWOjgZtg;^PnQhgEAn=UP1FQQG7
zAyPwUhRk}p%eyEmxl}1#-NgBbeaV0Kso%z+^q*E%@UMu)o5&MlX^{?^!Vq&OghP%a
z1f~f}$FT|@d(CN~r`5rDu(JAYu|;%9>~SX35Y&Y~?u+%$a#vR&A=i|XAMhS{KfDBB
z()%p-IJA1GZlErPiH5PN9>Mr^Tz=v>az?!*{-pVOn7VHu+GOgkCe+~r-Gd>j<yQ<2
z(*fX~{}pzOMcdfExBYlt`Sh7&4qYN2Vs~|Qb?f}^Zt<Dh<9I1HdMZTn^=C0swiXtQ
z<i1!<Xu2-~Ozv5jUdsp>f|EhCZ8HJw8=7#21dv)OW9^VmU-<$I9*HNTtEc)~Q#7xi
zOdWGkd}Hi=CZ(kKIapTK|Fa_$aYH8<nY_H*(=L`RE9M#q3Zz7xRmD@1AFw)k%m#US
zZvTxKe}3<46$<rq@`GD$^L*RIdW|Yw<ni9^;KYf~xSejZV-J4o?~{xFwpM_(ojJ3L
z7YCCUi7tW_71r1>KKru!6F~+Y?5Pj?L`tLiL<Pz-rA4`yw4Qh8Y~VS#^1(YV-Ae}^
z)--eXQH7dKH1Vu&dZI?{O4)q>U~oCx7X6>|(G=H(qiUhVs?cX;U-l`%W&6hMGvn^O
zzCrv%;v~03#dw{U(DLH$x<25p<gS>c{;0QBE=-o4E>BxX8-7zXB-L2?*SuYR_0>l4
zn0#@2omq{o09^Ize7|VKVE$|kvbaEE`~Aj`R9L#c`#7yBPpQmDB+DKY$^xb2!C2n^
z8K<=?kO^}{Zl9@Yn36j{Yekc<&ecv{d)aj1#0iOKV{IqD{ZA&d?SSBX5x>KYO0`Y&
zC1`#NB!`(O61XwWub4HVQudlvjR=^bTfO-ySm$A5W1Fh6nNy31@(Hm=T~h@m9iRR!
zo%i}^KnR~-3nFYQ0P1h!3Tz<9{f@Z&py|nsdyS`;2Tiw#i;J<jdn?m*hwJTkF&0YV
zby-e`D*otuZ%B;+(ubao^8+x2=qh0t#$ZDDee7#t-ej1QScrE=y$u6|d;-EPoOr#D
z^KCs=eiYt_OI|5tdqBrX@+{|Ufrod(0weGA8+3cVluCg?on8?*83Q%$U?X~Q3g1H1
zy(E!Q*_sUVKXm$7PG31E!aD!>1`G!18J*w?UPCk$UcvxP8Z5N{<T`UV+!ZvMz8gwO
zr+M&xMU=&Y=Sf1$&c^$7kuq?ctY8$$q+axtQA^l91{O~)&HEA_`S+$1fn9JQltn`9
zY3y-UddXOzs=Oh*y!Sa7%;`R}U=pL|ebq>E9wt&B&B_-(7pK>VDozpJ#P`SDmJ;MP
zVjHj`TrL1hmZI*qa*tntj4OB}KIZx@NNH()G7%k1k-_e&K1cEB=&U<S-ZG7?6V7Bp
z+|l$+s@*)@eJ;&eFuI!3NtD9CRF)fr_v^&uU7_0K1twVG$6tplR??M-;xrwyndd<a
ze5a!m<-3E-h+Svt>EFNE83e(8D;B3pzD0LU7LZpNxIq{XY5z3pBjbnP;!-8(h4Pl=
zO|b+j6^?m4e_8kPZSBR-hP;^xwPDPEDKWHOh{ZTIr>8pa?xw+(0&C!4>L+l1)X+$$
zcswEs3(p&!nVdA`*H*O-{Q_7Eo0=GLk1!mGl1Ligp`H2Yp2rsKZPWi<e6@Urk^4oh
zBmiIbeK0P6oB-8%n#k9uP1K)@Uzpy&q<SB5%FTG8P-H({bPQh?h`Sz*7Ypfq-fZ@&
z^3rdY8>2V68OaUz+O~BYOY`<XHK&}`TyETplOhlI-{kmc&Og1lz7)_&_n}ZVG;De3
zhIhd1$l!sV;2m{N2YqU?ujAodrh>XxOTg8stHgXOIx^B7t*WgmSuulm#8RXP6&B=k
zauD$?=%&V<i09cybHSj&ubaSixyNFFVqn0wd<8Gu_@T8@!dRj2TLs+;$J2vx{JYyy
zVRJH`-phkOI!|XfwCAoa)vk;(QfDU`IlyT&HUJwTcwd-=BH`j4YJKYXI12@wt<Pn9
z|8A+Z7B5r-o_QDvFKLAlt!5t&m1epT3WAoh<vpN_Oupziul7S6x;Oc+>RX@mM0)##
zNV>>*HlqATYO8#GK_x4z*PP1MZ1mtoOZ6=8b)jnH%?S3ir{KQ8S#>siZ@?083O+u$
zj9QwM)D-uLrR6eLwz0Zh`2vuK>zhqUmnQT$%r+ZLNa~D@VbNuWc&m&5JPEXs5>8@^
z_mOoggo(*(H-5k1D}~by&&qkvbp}?V@*2SmZf}ID(b?QWq-mC6-WnN?)=lmC<J-5F
zGm2G6KYeDqD_$}J({vI{oD@dGAaLf)JJ^RK&*`K>i0wVl&|9(o7~t1}ipH-*Sdmbi
zbx9$4N$Ng{qZW2nEaT~HuYFl2lO{9Yr}55uv|9jEekhUJ3Cc|9xQ02>fs(>tZ@%9G
z)j0Da4#~zeP=STCAH6Ak)VW6QE;K(Kq5>D0@KL_^;Cv!2OUTE9tYal$lIXGMaGkP5
zMfam;6gQ+DUs$g-0B*CmU$(*Gn&1EJmBQ_lVLCaydN4tV%v)R14<#quP{tU|g{Q|X
zjhyi?vBenJt)hKAH5Rk{is?nGM^)ehfj1^ov9BcpLXu3f2q&kfl%>7lSL@}4Sb{(5
zn$ExBa66axChzrgjQ(O3Sj5*_#ZO2b;yMn6dG!>WkiLd|Kq0~m?y3pxZsF(8t{b4m
zx+Qfhu<_JYh^F|V(a#W4ph`CXE+Ykad_y&=vQw%(<(Uyuac}lzhy5Wf@ZgB-UyKm6
zluZwuOw(poHGJ<Ni6egDNmf}viSQoQ-brP#BzWKqR+SM$_DjQbgykVNw17Z$b_RFC
zo~(}#`T&nYh*Ra2G2_Y^kyA{Xj0^Dg1?jQ<`g#kfgUqc`FW;&V(^a=)2ShhN5lzq6
zfJV>!44i9(Xi4?fpC7FChB~u^MS4a)an|i0^P|Xkx9eByG21lak1;>!ZnH;?mE=gc
z=H6j%Y*s0D0fRbW?U8|m-DN_?cqZ)mh6O4c2(}H~PZ?89fllqa*o>Ja6n`Tv`VCjH
z!ZtVIx;n28iJe#V9j;^V7OeHK)Aq_7)k)!Y^Yg&b+QQ$H)c~2o!tB^rIGdz|$0QJJ
zsGd~m^*7Dfj(zMHSmnMfJpWhj-Lc=%X7SdE7v9~`loF;R+}hN3whvr@SeU4Amq1{-
zM)h9?*X2Todz&0J^VPQ61y@k8cZ4sovK=1ec{Ws8V>gX|k!$H53u9+zS9FyN3Zlmr
zT%28X6P+|KPh-1~`cwUtl@?jAr}`JN$0w10^~u@63D)&K?Bx0sLt5u?M{oy&OTon&
z6*2OBx4+5gui`TqZp4Ik6iIiNBv4_{#i&{GP-V^?RSFMBL_isf-mpU?zadQY?h}zx
zly`C;GHU+CczJm~3H9ARJ#3t8)b8-erMzQHue+~($K49>Q5VMJXMDujISM-i&W`f4
zcUJTSih};CzI%qrX-xA^kix(bQc}M(ES{P4hAi(E>!$2ed=n<s+jl<r2IQtn36}OE
z9O5*gO;a#Fo#&A*hkvr|IzvJF`6{x$bNi`|Q+G>n^$>0GXozxko}jrdR0@B#(1?ox
z4;`W6NfSR;tM1PqdM#6=q<SkgrNlST?}Kx{d#GAr;4mFkjEosSTrTy7Ny)%>d$N$s
zb222=#{?8Ewws&W{J9dqUhNXc|05-niRXTH2QI9=G;YCU0N#yBg0w71Y1T9{Rfthx
z73La9NH6J2+!2qeOBlK>lj!~RsClH~)<iUujA>e08j6fEAL~o<WlCcdmJ!UF!U84&
zG(75b>3jM^)-L3HG1OonY6K3Aeo`PMPVK_TqsPhv|9Bm?iUf|ar!e1ki030Mew{56
zO)Zw$R5%GQ;R$R8cR21W`ANDoWHaywPHJeVu03k|bUWKC*x-0=7v$%<+LYfThr~&J
zL~wtF@C-kBTdDs_f45F7u&cA9@fD+t0W%LmsHaX?vbPhQx5G3|SIvTCgT1%BJiSWB
zwI{@+Fq&V)fDDTQE7$Dd)AKDE&96|6q{`*Np%8uQqiBoD%_d)L*RKl9LTqE1=dI2E
zG~fCrONJ1UjhT;1Gc%K6KMo^{&Y3is!sVk_a0XBZkhq5b<}5uZ{Xp1jw$D$2frN)!
z{d=7owQ=C=2?Sp}y%ix}FyiW}cQZgnS#?C%N&D>~fF9NRgS4`;x;<Ryod=Q%6*+ye
zMuvux7(}hiV@~?SS4d~nsHvfd0UZ())8<HmP(1K)+0@*84~q8w+6-Sh#o#W-w=nNE
zGj`R#m{$Y}6yX#lXezJqkvmomX%HWME-{IC1XYhhW3ViTp{U5M8rwO%nquVswp`c0
z6(BF7uCC5@02vuYygbDm`AAmyc^PE;>wiUg83Kn>U7VSrSp>C<e6HCfJG4KQ<N4}k
zHh2y7L{pCYxVUzB8?wfh<*UsDLG|oXX|3u~I+{@Q-{Y@1u@vKYM99?z5b))Auvmi2
z!)9JIk#E!5=jeRve7zpD?6!pO<E8*RHA|be%kZl;uB&dY^C@E2)#diA+|?rS0&fQr
zv^&l>Ds}sX*e2YnR|d!5J{f5%@Hbwi;MwxIdM;y97`3;*>%+!(iankuV+TvD%pRSc
zoet>Gr2CSTJe1^$Eu*QaR1B}{tP@~g>*eqv{|(nMfDNgWOP=n`lzB<zuRzzJfj)oq
z!zV&>v&S8WJB~ETz8eZ1fu8exnv&h(an+mc$M4^OT%rB?{qnIR#{>B9ZL7AP?(bSM
z*7M3rb^#d8Sq91;=yP+^^!n6BH#&#ToEikibD|ec)uS=hWbxK6hkl8>bt8P@_!_~_
z+%(v0KD6nWj%UTrT#WyxFpwW@?ZX!1U+`Gg(&D817@wyvSls<ZjCQk%;U<z0cT>9?
znlG^mb2QBiv?+gh`w;so>;vi*qqg`Ff!2_*HtR<cEK8JGfsE&-4Z%&;Jnxp1;7Lgp
z%_j0l!-bXfx+7HeK9(6UyuaA)4KA(*<u}oWK3KT>>`u%dCNfM8aHpW@DK@;10`f`U
z%}KI54yOJq{54FTSouml>F03PW6Y3=59eaD8F}0z!JNRfBK7DySZFsaS}x@!?*GFC
z#&IN!dK2bwUVA-Z!8j&8iPaUPzerrBiA!C@5SVSAc`@p?=aemCkYyrB?m}lwiqINV
zhYpnt)FT1fH0mH+7|{lbZGk|NwBpG0YVG7p`1h_qx=BXLP0k!QliN6qvIF>e<n<mv
z{;I+Wtjx2sO2m#c6~Yz<y!6c!qup3iPFD`j!QZMAJB*%0GNI_rrh7)^>}K)qr;(Fo
z-8RoV#2EonWmX0=LFS|=Ri)4QI#M2026mzA!0JPZInGpFoT7U^?W^op)TNp<PIP?G
zm|ZN1R?J_YN7uD60(UTCd11B4@c*OfyrbFr|34n1L@A+4s9L2)6|vQ<O>MPn?<#6<
zSM65pS+giwvnXn}RP3m#DEWx05u+N!p1=G3<Ck;tZ*oq~y*KaI>-l^>o*WD#3|wp~
zA2IH=Sw6i)oOPa|rV9Vlk89fGJLp3k-?K;2@ZQH|M`+rbOaxe6k@9a+-#}F3DQD#q
zvnOMnH=r&$!pzjuN9wlh57uz9yjl<)fL}t-&NRn(Z)HVI3v;BP`pe2%B=1e8&R(yy
zO|_F6m-6E6`aI*&hAl^=WiA2*z2^Dg(c80BN9O`s$oxt;<&<2q<FGhO&j!+mDS7Z@
zZvLI}6l*esKO_!Dv_0B3hGLHN%HB&AaIVO&sIS$VykU?nmehYMgT6jp=xH7VxK>xI
zH`w-GJeyVuiF)x#u~>BYw2H-iHKe)OeQiKROTTEF8LuEhaYjiQ9v(hKd6xYVoB`xs
zM0WA#)KR?2j-jW@c(VE>d=lD@@zdA@k4*l#lKk)KvE9Al5y$JLm%&$Kz-j5l<*1cg
zZ0Urn;&0+yL}ybjKDOsVB<v1Qu=)pdKV2QuUG1c;wH%@*3T?0#eZZ-k+-*;nG9Gol
zYCbQ$<njP3|JS0gVg;=a`TCAql){f+9Ija_S!3T<d1j?dw4RL6(}ozP$RFEtMQ?qg
zu+R3_f9n}u$#_dX?sG>g5q<Rb#9YZv$Wx_>n%?o_)ZG|?3d}DxpB-GD9S4~x0PI=X
zu4Ih_kUl}+geJ2c6B8+2B!A9>yNrDs5&p4wv3aq3b&?v`^7jFYWDw?Zk6f-0R7iu2
zeI0ja;4*4sW28r<uUmA!lE=jz0wGxm6K73pu<MJ$SIWhfXBV01W<U8#2_+%5S8+j|
z;)J}&6km+iw}l!VMruM`blm~4U8|{8N@It{YoXiaRCrRnLQ(@x&!(aVDQc2WA(v@K
zVFJmZH999yZB=$htq4Gk4E5`-8vG`Uk&Y!kp@J0Ru}~YKymM_!*oZy3+|Zc4v<Pa-
z8K;6Q`T&JRBBA^_xDH+IX)OE!?<=Hg#PJQ2*MNUnBGTmXmz#_%fF9Q!@+FvhrioL`
zG{a+-0Vur+zmp{?2|P!I(M1+cdoRm(wsmZ<+zY`KH<j)463UD88#N)atBB&tqwlUI
zIew*5W#qTFhr?cvlBq;f*)t$WbRikwADTnRQWC35qv-KA_cD$JnL%f3?OAs%&dj_D
zj2#K`hWJNrqy7PsPUFjb21=KO1|0Gs(M0f=A48cGQ?@H=XU(E}@2*Dwzt#y6LL^hH
z%TpT4!TEV|^{aJ~1baA_c^k$q@&q&1_#eHo#c~lHiD6^w5eZBu^S**$_3{t7!s9Jt
z7lF;?{e1~FLQFCQmO+;a%ZyQ?8qD@rp(6cGQ5jLkJqhdy84!0EpU!xCrQE<(e_CC%
zT~|6OOluLk|C;*|Z_L<1ifN8-{n=`{OSW*Jv{LCzVC44vLC=su(ShLDT^W_Qqbh<N
z(l^(|X7;eNGP@y_&*LG73g{H@UF}(ZdI*X`)Fhjo-UU@PH75}zc-e{!e>7Bh=x!w=
z>RrR_IKXgL%J^0JaEXHA(|VN7T6OgSW?sTPL{ZN0b%{hVgKWj856muQhDexdl9iyP
zS?uqOgDJ|?j(7S;8sQb<N!@ksh|_0RL|yyFyUm_4_+|Qs3JQx>EJIWI;wsQ8@*Ci9
zHX0K6gpHkz?3Q^@!(k8mlUp~RNfq8Hy#t~RS%9)6WI}D?94U!r(p_DKYlB=OmEZDV
zLX>3X^|;Z77rRGeAW6(XskTG`thDp-sv}r1GtV|e?oSQ<Zwhz+bPD>{!TkEy@cZ|(
ztVMeEPQXge<5dxFT<2*|Fz)mehc~af+waUfa?#@y*z;7#l{k3fEar!HdvkFCoL2x>
zn|<eQrIGt%)3vH9i%fC<wzn^SFG#yKY;R#?s{`E}pB{2?3OloEsp!=HR(bA2=sv+%
zb42bNx!t&qbJr6+{CXE@!k_3b^$i{~6vsml-LhFcLU*l`4oYOG#}I{o!AMnp$CA-}
z)kJkLyc-@pI5;T1*st!IZPbiLwk!h`7b1#R`;)*f2+?8WPIc&Pj)C^a-3krFqjM%w
z>f!QE<NFFI9;)4}fk)&}uKh^5=;U&crOC!Pw#vZG{~_ua2llZBgXQl7eBt4kxkr*P
z6@IjTj+fVST;xUT)C(xW%*CZO2y@Rs7x{6t#rB9x4K1sZ<j7EGVtkDpK#J*DV5J1|
z^}{Fmvz|W0$32N$bZQV@V*HRT&E%vDp9V`<-lT8f+#f)6O1bAAzd(z0ZOAX9WeM_0
z+0DsRTrkIy)e(I$6Bg1>7D9Z>RFI#6HjLNwt6Y&Jq+4tnzY3!vyc%p5)`7b|iQX`k
z4r7cZA(956H-k{oejfmQ2-4r2_(orjq~d~C=EJz+biT`1pn2oy?vm4&Q*xMaHwNQw
zBDpByI2{=gq-?4InA=DpzkmHdkuY+&Dh*8GH8}$^>yO)yx_tXMZHBiDY`hh3R*=(G
z8wdyOfbF_qMVJava8a$$G)EOlzU($x;-=%LK~e@Q4?bLXzXDne6meFiNR5gruQb|{
z+Y2T!0+le;9i~BLs!_3w`>l$_&vnhEW~;?Se)I+@`ZVjtL!)73`JBp#I8k*31Cqb2
zCr~-b(9&oPWE3?$EwDYZy6ss`hmjRfzgKApTJY#MTd8nPan88{0<un!w`cmnHqe)j
z48XMX4nc4q2(|-w6Zqo`L^l0oo)+tpt9B9`cQSXUiIr3s8EG5r=c@yA{?EJa#jsHU
z6b~C4`>of7XuM7lNyzQ0`&+SMc`D=t**v%^HZZ%nmq2otz}^2FxvC?y*~#{#e#(-Q
zKkPNj9xO8HcXCm@R;$$6B%PELX2=vVEatX3!Y=|)=J3RPe=8=$LQb9SUn{02%I9k#
zFfwc$j7ZS$nk!eZ6Q&q+M=04zCBtBq1G_KwcTR}kix|dJZUQGul=TN~iumchuRSOB
z7Tl!{*OTL`o~(t=VZ3Z}2S!HiF1M64^Z)BO+e`736OP$U;Qj?3ZwVvIgD1NIftk0i
zank@?Q$aQ*crJ*B45_;(KVI01#<97>e<BXI4^}Qk9M>8wM&h^SMJSgKGz~7v*BgPI
z0kA`CYUpaR@AzbMbbrR&doBL_^lKi+#c38t&sopX$3HZkZ6ES3dM10BVn_1qA|A~L
z0bd%P_OVu27JY5KBWpD$NReYJH1oFG&&3ezV!)vS!u$dvLOupZ`29C;Vg0GYKLzoP
z>XyoWrzXNU?!%MD#h$L7juVlqeFFou0X_cJ^^M{I_)Q(v9eQp~p?BZla(E;L5yvb=
zc6lw&hL4OHnAzALQaW|0%=X^uJ8CHf7a66Ip;x`e9n$~ScS3EQt+D2nqwmHu1%K=p
z!`MWZitN-hLg*kKJycxpM8zsvf1fwgY_H8W{{352Qmk;iwzRpqdAP=;{6vMxvLS^Q
zO)YTNkfZW0VlrFlPfwRK(?h&y7o;vGnB2j~>kYFPFwI&o>tHkMz-xlYF;0LPZ&zjC
zI~odL?*6#1O^_OE&TcQl0VViX=zPH_m8E#t$VXm>8lB;9ceq9TQu!+*7pZwcLd$F$
zoQpSyh<*r+r1itALp4~Nc>deyVhj_w7PxZFU0GODa?$nyOl8Bpno?9}S~?r10d>mu
z|5pU`819fhwYbM)MaxOZ(o^VsA*KI>1+Qy-t6Qs^D}O13UV#dstaVM%;T8#mo16d5
zU(GLYfv=RYHRHs&=V<U_ZmJafqw>EkWX5FGI2P6ZrP3IdeJt?!)d2n7inEglOj`=<
z>CwXzm?#QH>ZI56)T+X*#d7*nQRZ*Zd`1L_Z!i?1e!@I&Y?FRF1qd@B;TjUA=PNa?
zG_X<ibQOCE*wgi1&w{q?PU<X49r4Pz>}&>TMmpLu8REQGVK0;rUC9IGy%jcZADI{N
z;&ftQXtt4XI8U26uw=iY8ypfQcxwnw0^^cF{g?mZMF=@m8||lVL#|_FBVulT&jy%1
zh|Ci3b#dD|xhxA@T3O-Ib0<@_%oV&3`F*@1!6h~D=!4&Q+zZ5buB3`@q^K+7W@i)T
zM}q)0J3TKsv~tV3@W3xa{2yES6N(plUHr8<Qy-m$QvK#IVRQQ~_hbi(G@Kr|Fp_Y^
zd3)E@=}%oyE=KIE4p|xncunv}o>T%hZ&ni%!>5fT*%|qWX!XhW{@;)4Ys*xJxvS~F
z&Kpn9TPiLy)}>&U>n9$Rd`Ov?aH;J(-`pA)$ilV<Q>`N6F5C>Hg1ptCJyKD<*~Tb{
z2j@$KYv!{Bohoix);Ak#X4Bo-H^l1Qk~<}7VUFsums@F3kwg?o!i$LA8v+6s{_6^E
z96dp=g_XEyo>3}=9D$+q2AM6i+xBl_4*;&{_7-MsMY0v;7s7Yu^2~#0Q)(BZj*6>p
z#!mfQ*Vm?PN+Z#X1Gf<>0Y2R)1yr8{|5FE*{8Gm{tZ)-GR0UQx>_@Y%xp;MnKkEH_
zHKEEWLmI+ics)}Y1`dP(=pIB+ooPToNE-8}u{c~9he-Bfwgz>QjMp8`m7+noLdfy&
z(sOBGXG%mV<cd$}WH+2pk4#wRNNcd|I9!<%*FQ;Fu)!Cli#1hKt#x`L3`djvhvgE%
z^&C!Cg<s=WXkn6{#%*Po(>vs<rE4(}?Z`dKy@;FFZ4pbL0w5u7EGQ;QPl*_Kj3&IU
z{S}paic9t>%n_*qCCFF&l+_e{`CAq*aO1V(v$~NSQXSzOwLFqL%<=b0K^I+T!*Y>b
z>UH6XT@p@p$#-S?R)RG(I%)?JS6c5-&Vs?(dHTv*Aaql0_sRFHK`2BNT`xKEF$wk~
z(1ukFu;w}kf|*Z06`r)c=8gm`*F{oPM6{*y4Eo#;1zaR1mPaaWj3bs@i(+_0vQd{R
zD)|<+c@$!C=@Pd;zyGswvQsahdP-t<dC#8Q_%&!rW2FTH?yRcHJReL%B3fcwVcdFO
z?viHxBIw$yh~7&gMJ3sTl&ZvmGR(Y%c`onrct<oqAA}ak3!MIhw$g<%_lMg_w>B}E
zT!#-Oae@18S&s6%P*SdF_h@)eJ0=LlFe<}5f@bGZqt`)30ZsbGF`{?l7_P%*{5OW1
zS9VVOCvy;+H2QBm<6y_MuF@;$Vr$AXP0C3+Xc^zo5ZyneVX-&Npn&(Bmo_z3Rn2CR
z5+K<2or$^fqu$2#C(Q)*T}3FxG2YwgJ}NTu3KMWKW85!adR^^?=R#F^6|@yj<e?I9
zzjmE-jOPP;w!^>tE(WRDx)Zt7%H&|$8}d5thuN8siLn@IlyOe-oUWMsdq0}5hc`&m
zi>y2gdOcg^6zpJXVy4`j7>kj3`~Y6l1h2?B9Xz($A9l5Rk%-6#MUjdu;#A1Net?gn
z+=IQkVX=inn|8IL-R{j#W<~T!isjwd#IzJ@OQXoWIX)0$;w6{ZF%JYCOLJUJI7apM
zv<T#NJ|JTUR-0vuCzyzx+XrWTXGbY19G9CFGc&CV)*=&Y_<=_=54N-=|Ioi_D=jYV
z-d!skd=bB@T9*LqRVy`S|0+TLdrlZPFi`O8*`}Y6l4=BjUV0R=yk7+DX>73U9NUS%
zF|$?CuhM>X?VL{$kANc%7!oYY<WI>^@k8Y>X6*vU=YQ)6;e;YOMU;L0SGXJpLf?7C
z`3~I4@%h~dk2`Em`?w5d8x7quMQ~t9dRsnx_Z#e~iOAQpb~~WLhn4HyXflX_UO2!!
z{HkuH4t<1KhsU;P_$@_d8{hjkkvz;DM^DT34&hNAW#yTIC_*5OcG$`Ixr~tfY8otf
z8UVtn#^t>EK@|#tI_R?85Vns)wj!GXm^HMtUePzBIHR9LGkmIh#}mi1Y@nhwuc?~!
zLFnPu53<ZgyPCbcc)17iMMUr=F`(=AI4pu&vXZhIZKM6l&Vu&c5Tkm=`;fdneVq;U
z2nLO4*rQ`dga&^Tmnu{xX%-iSCp18TPJ-e&5eOa0=;M5RS+%_19!RxQ|2HE}%?;dD
zc8osxVa?(s-1-E8a=w|Z$O(CY7U=kUdW8Y!uaz5q8KJ2VNM$jz&Idy{<k2^1roZc+
zT2<UX0*5@%cC|8M5~p2Gp=WLm`aM#>Qj}u*m7*J8{ZSy)CWMeI0!ku=zz3vnD)<yx
z32kd>IS9Y{r_>z<V$np&<4WWsg90MTTSE_Kx|XNjZ_Up=H)A?UxeYSCcGJKC!Q9Mj
zuNzLIscvM9%4f{>)nRdZ@<Lb!U&Kw$h5D_a5oe5ohjS72SgcX3HzR)+J&Kz<2}P1{
z6(OC#FqxJxxdwljPXOSPGG#N{R*$&%AVsdL(GYd@54b<m(^&J;-&IjA4ejOS?d9cV
zH7lI=Sj7UVIM&?bmdJ&w0T<q;Wl&VhN!`}Pb1*I>&c=wIhu^Juw-soa%M}Nv3FQf$
zXsg1yw+YL;U%$R=RTRsiqj1Ah<Ht8fHUggP_6|W^Iv$oz0;u~pz52gqDa3uB4~T(;
zIi<O^ZhY6_C9lL{Z%2Y(yZv^YB6d{PApa+-4Q0dyJ`g7JY>Wn5F76^l%hpebcXM(y
zq_c9kUNx{ftH7v5SsK~?0|NF}@hb6t66M_f8EQT?#-VPGG`O=WOC8&VuA{lrHOaCB
z)cn#5XT2vtXlrMiz7evjb>ZaXuluLw;{SCSd=PnTZr=H%t~DEmf|ITkGsgrM+$<S$
zZ0_lZc@c4XyxY@Q@?Yd~aZ9MbnCIjcWw2eiKVj-_^yx`URxrqy499O?e$BGGIIS%`
z|JT>^kR^rHxZeAGZ(8K2?_{s<_$bQ9O`;^;#mvk&)p%h>y0Km<?0mxgiTcWXTObCE
zZf=Z<a;qU!5LWdmM5>!ws{`Ssh8g6Kp1a;|>^UvC9RBGweGnN9r7vp1b-c6DE=9v*
zsdPpK_WE~s)$_Q-$e08p@%4{P6+(~KkN@iC4F3JwEhsp`bMN!4f(uXB5KFvruGeg!
zmCK-GQ+J+&7?~}QSB3b1c0|^<dOZ45lnC3CH`jTPH#W9hI(p&aJ<DtNzQI$#<zcaL
znos>|j9Vy9$Wx%wA8q}ZU;7{O)p%BJ)_yPv=S54@)`QEn2i;mMUEO)G%u!fa1}&Y4
z6pX5?8p-Q6#vC@6yx#zK3fbMuGoV2{<=^}9iac7NM<N&?LZD#Gu^*Bj_XH!tO}!j%
z*%bV>bN%kU0GH<Grrxr50VJ{<q|upzUGLYJ1s+1uY3;-4*3|`RQCeCVpm`}5@}B>?
z0C@${XcSJ3X!G){H+2C~E^Wa<W@cuUnNZk-*q1m_Z`IG=vmLY?Ny!mm%yqjMjV}<m
z@l!9TA`p^*Q2qulYEpr!eSDk!n5z=@6rwVMpOM-6gqVkC^{)4BO(VLWiUE(yt_B;U
zLYFsl^1N-B$z3vUsYiwRc|Y%|o9v-emn@hp1Zk#VyBJE4&|^IJ_b<N)g}VM49$%XU
z*0CZX_?BxboHCLa6P<c$5FCZT1g0RENa_G<og4#cl$NvS?dcvLcB#m`JdlSWLe2Kb
zMQqF&0ry<aZB<l|{TV54Q@kQJJ^y!>wcGePzs$dG%A3v6rnzTrc0$;0{*dSz!H(Ys
zC3lpR6ibnsMZf73&uc^RajdkpTybx&2lmR;YsBIyCz%3!ntCFG#0Y_MYz+aCnqL8?
zOh)wvlFhIWt=s&5!d*duZr#oF>7PhMuvJ!*Ra>{=MjG@of#=<pb^y!!&G`m+f=@Dq
zopL2Y1v=$KGrzCNmspY0iWOOiJPg?Ry~Ld5`_(x#VtxmAg+9CbjaRywb}$X9@9GQ>
z4nJG_v+oYH<v-WGDlR?$v=x?ke+y{7k8Sty+KvkCHV{BVEz8QiQ~iuB^$SkQ2>RvH
z`X>FPuN5hdA^-#%ZC1|BM^jc=@BjPo@Eg2(dShgS3Sy`edw@FwKf=e`i=B;Q!c`m<
z+Vb?DLU<6NmXg_w;Bhq^ip&^Hnx;4(3_5Fm_zebogfxmnIMvNQq+ePL1GZZ|&D(?9
z+QCMj^ALizEZgO|$mKaI<#uMv$@51YoQy{{iBw?M^A=WqEYJsxa=S}xb~bnw2#u*g
z$;))Lz@UMyXxsZd9UU*4Y9W&5ilz5n1@dE^4JO9!TX3hbFoC$=+0}QI1zIsGNu66U
z$uU&S3g4)sNuAhK*OB)CdbkT(b~OJ8N2vs-?I!k2E$@Eb>Kv13_}O0_me*_WzSG}Z
z6waeb!ll;#=Vr2oK+e4z8($vcGcCA0S{cqkjoM{0yCw%lD*8}fUWgb&ozYKb%-WUB
zeF}5ZRsLRigXM&lVbmw@44AeNu}`?-NWK*?b*_o8BcdrwwYY%1Pe;^`SD6_+;|ytb
z`<1{zaK^h&-jJjzRx$|G=?lpcBp>DcJvnnBLOJPP-^BV;DE-ZcthBS$)4DEtZa==o
z*z0CC8hsvS=>G4SeOxTvSIRiUBoZg?_;304NUE<M>TcPH(W0X=Pq1Ve!)h9DJoiHM
ziQoV{9-d`l<;V-rt59(vO;o5<V6n+)4HRcc(yv7KVt-hZCb^R?3s=68W2-h|k!$Rx
zNiGPeCg+IQEQFG1P-$e_>xkan_j%b$5JPUCu8k0@!wopPud+B&6Q`%MDu{DMuPhjw
zg(zx<GYl1f3ysw^ae%zX;^O-%T023&)!ckQK68$L6c8vTE34Fwg1b@W6;G)oRS}TB
z9ivE$F%xEf%H3V{@8-W`k=~z?6JNm~DqC(`x{8a+-ob%@Xa-pzlYVMi$v*`rB(hfP
zvtR1pJ0(oU&cEkzB89y6&HXy}80kKHS-a^zM&vl+=W{Azv(%@l^=&)PMh6npvPUN;
zCyU?e0Dr2sT7>TAUsIdMI9r;ns_JSx`^pHQu7T>W$3jl+=K8U2yBF6R<GsHt$4gnU
zvy11rk2PjPHRCKA`^P&cBTFs1hF+(F9L#eN$M2Dom9Wg}|F|JgVaRb3X?eFtuDa2R
zCDUEpuV3SgSDRY~Tb(@@KkgunrltCbEx=<KaK6D;nlq4sZT|-vPk!$^+ugWt|E#ir
z^>O5QdXN*(B5+YWOZfM#^Ntq6B{pk%I(E=0!xZ{AB&^jFA8WK)e2BZRo|>lb^UwAg
zf$BBhjjtySch)^j|CTs64hJQSvxWgUUClFF5m_#1S#u5KF8)71DUim^u_Aak`;%Ut
zxGWhTa~g0wFfnm(wVZaz5AO?KM%8&VE`E;B^OYro--uAmGfvg0Sm7)YeL0qwD@$ey
z5I)w_RCx&7?A`R_k4*pctm>s#r(>389ZbcuIO-eRv;JoEO<QN)HZN<&YpBoC568sf
zcKteZ$D~mFC#o=INP5V)z+br=ulpXqx@yz<!_fIighup-W-s|ak=>&Zhm)-zi=^b}
zg2&@`Y=y3(cPf2t1@}KxXhJinc~OI7VSrn)BqSO}p$jqj!Y=xf*>@R2FKVdrrDHwf
zH6u&*>S#!2?Kr&gZXCS01?65ZNkXWnp7N~6w#`JQL5f--%KRkIq9*RK7=H~<ygX=s
zxF@qE;DtrWc3B9wN}>~?(Ns6M;9F03u2(?bH(;UpqNDMg$Ujj21pB%psnc{I=6m<i
zzy9V249IKvBE;9wQxQI|0y^fZagQEexSR2srH1h7QTj0HqOAxJUdE^U!U&BvxLUEh
z7(4<uU=P_p(A*!9DdP{B4d(h*b&#LVmsXlqbN;*udT9T4edZug_%VKY>AAy81*L~(
zrX*vPQn+ighhAztlePM{1(I>V+h(><P8}nrsgpWNg2u8DI%MD5LeKWo?Oz898J3c)
zOv^JB)5L*_mYH1{<Cqt;=m_HRuC-5ZxtXAsZiQs&16khejNNx)^Fp5{TKa?W70pkD
z<?p=+@#6{YmABT_FqM7Z*-M8JF~H*}Vg?5^4a^b-Mn0~Tucs&~ddaW=F?6x=Y3#J<
z`4RWPqLI@T_we9|EmV$!HlD~r64Z9p{o!DA^B8qt!|^GNOZ}rU((ohZ(5BYKEMWfO
z7E2X^JkDtGNZnLQ&*@ZHMugu`IoSKZo}<-jVAl<33MZnSJMX#ZBhGh%r4Sk|gMPSz
z?S;Iv!Cidme|@1uS~uXw+1&KdQE<a^pVwXvT}wF!d}nk$0@$p(F`k1T4AoyytGpL4
za8x7j3S8c7R17=bpWATmZqA><^WoKn6S91<M{zJ=^@Uat$~n`ME%w$aSndCHaL#Oe
z2ck+Sxp!93BobRzp8YfsQw}X6Lf>XzM^uR;#n_)Z14uBIxvv;Q6$j(ia`6*ob1{y*
zI$Lb4sqw-VN;g=01x#Q5K2uu9M#&T$!{JV_@>vD*7;#okdt^~!4eNoP9IO5sQC%$q
z0sbl@?tTDUiiX~_mT>-WD^J1Cl9+XNn7wXy8tB8&b_+0LR19&d8*ORWYQ+pJy^ytr
zmTA52-o!GIpz<O85Jkwi8*wRw9D)l6HSg>f8nQ%1d!7F^ImP%IAM!fiYu&V&)fQlJ
z5{UGdVMDvNJTCZN>5`r;_H}=!Kd)4QjQLJwsklnT)O%Z@n^~GVEF?&ABqCr+dgxQg
zJIS@KylDlp9B8;fsmJuBtAE}LMKU8`JKlE1)aSo{Dc0M2=ftx#%spbxNQhTwSDhPa
zGFZIu<2ZWbA~7v+_8GR~rW|{&dAkqe%=<0fMw{s%sZCEgtO2`H5E@Q4|7^=k*5bof
ziBUlwnZV=v1LsdM+VTntAPYS7Mc+nPF@l}`ke=Tq|A+blXWZ9i+_BJDJS-L{SAjR`
zniZv;9n|?dD9YWhc^MoP`S1hY4eittP&G`l)RuMzdQw<&6{YAmrl)xaVwnBfEYMr7
zxBzt{5>v`VI(M^Aksx1?8~P)1h*zXJ=58X_DO><W+uwR0i^+#O1cKt5!d!1ymgC1r
zV!)&&oq14D>0br^hGJwQ4oLhe!yH4tNS_dsTu(%Y=Hx)4BUWOh^<oi+I1&i{XW4t}
zo~9;D;KFU6Ua%wlERW-S?=qGTH+XQ+db?}(b@}j&z1!m#uhq2^ZWT<<H^`|&FG{b@
zuNn+!SBmqxJ+5Z2w&_|(@z*v_U1i3ht~x8i++1;i6kKeopPPHE01-4H@-oymN6(FN
zc{gc~ocbN^H8)#b7V@U@M7jQUsNE44`JJ#N2)p-U?*(|*B*5dqE2|=0C5fRwoDQDj
zf+(DHqyZg3g$;GR2dKi@`}fsNLC+Z>{HrY2xBakAS=7B|t#^WFwzYNqPcWan)@ktn
zY0}Vl)5XCanFauni~I2DD$lV;EKYxs=+XYc-?45*53mXJe~SFSYLY`=5WG}A?Otil
zg-ed*vYR`;IeFtX#^yiS>SceMha^)Kc~7Q{o@^zq#F?9sYs2*F@g(s!<Bzdn8UY&K
zn7gE$FEy~(qr<?V?1g{gZ^DLaHmRf8_MZd<6j@j6&otNot>&%`PdhT*H@TzVluJxF
zb}g8hN_rMLu*sjC_$elMG|4z&Zt*{nSqv@~xz78ww)yYT9ayf*D-;bv0`zH+#=1}R
zV(;HH&8}>#cp_dADoy`E<in*2mjy9E50~R)L}6!n+$;q4SGBf*Ko{4FV_cw(&8*I<
zx!xVrD;O5&WP!8@>J&bV`?%m;>^~sEV8lcLK^O6E%nNyKixFRzopwso%OZr!;f=xE
zoa<H}mbcZWfx(X2Hv*o*fc=Ik&&W7`I*DJKKxH;J_eZIyFLzv@nif$*NX?ES?owr>
zlW;4aKC4eTG{}Ex*+pKI$-9UCF!71Ex!3Sor(zk<+YalO&==4s=EbnO6lD50e7FrU
zI(X5(7;Gb*>ZG$VQ7Fw7=K#;~WKC?^I_ne;oKZnijlfv!LhKkZ;R<y9LNTPz8gtXi
z@0up~=?h$cuXf{_)}$D5_xZ@~O__|c@(Lm6bf~|+Ebm1UApM<>>lH~xKF!IT_wWI{
zv?m0(SXr8p2zXN|Tr1LO?);n-CPuppR>ohg5H2qbF4xm+dqbLJy54UE<)cEksf=kr
z?L+rb9|v~Ma$H`6lMFL^@Vsf!M+&PD`%Wk$6xqZt&;8+pA^LBPPBh9l;^Nfm=*sr#
zf8rGN#58uLljDJf!osfP7Gj%V5Y~nv>n`AoxWZnYEK{c8EXVV(7g3wUKjSB)y}K!V
zlZtjd-BB*DpS?br*^=Ix;KRVTwu#Jkwojk3IHxu}FT4oy4<(fEqK&UxK37-Yc%44!
zI#$kK9C)#?U7I&)zBRK<NjFa%OJ85}UNBe!4K0)&e-j(p)i~)5{T!7i*%IKdhJh0m
zSilYIX$b*dH7TZ3@1lt<wyeMvf8W5stG<iI%dK9zW%wDGl^IN)owOC1%xkIIl+AH@
znS?f1zFb?IpIHSPdU1_y7k1}!X;G(}C-~B<KMO%kGJS6Ddkv6`emT_`t&G9+N700<
zufhLp>JOa`XUb~|3=~h)P8_*wfMh>7h1cU6&d9z+e9=}XX8QvIbrNxjS2`tfoVO@l
z&KUrs<9Iw?pfZt;qvS!J3Dndk5w=++P~=(Y+P9deUbgt8wd<prEPj+31y8LpX!pV<
zIQ-|`9%w@tz`bHrN9ZMbuYgrSgM;xw$eyFZ3D=INxh`+s%*Zc)c>@7bFTSJQxT>>_
zu@cJ?gYX{zc>n7!3MWwElLEoZ?}G5lGOTVn4};Nz0{f8!r0I8|*D8?oU!foe*stTi
z?IVYxB7ZIZH~4L#O0{TyJK59CkhiUeaeD=7bOUe8N4mP?dWPWQsmqQ{hGQ0bS9oHm
zj&^0b;PI5zBVOX@?^(dqB`LtmVf195LG4}9_)ExC)V)kx;}BVA@X>?V%d6p(#s1yt
zQZ{Ii-dLLD3e{pxv-!Ug7)ih*XzAt3Hm!&_7<WTH0{0}oW}ko;+{Fj3Dbs^?brgeI
zG4H5E*gCsu5{onVAw}Y(Y_8o9Q*e;Mu}}+HK3H{pEs#0@1FE#sQj)rZeOZf})R$Tt
zalzo#vmMs+uQT=9nZs}q?9oP5{vXj4Cn!5)#F<I#_jR{AUe9Z(n1WYE7HVStI@;#>
zmSWn~WMXdG?)9{Axoryz#a;V7Q}Nq7-%rTaB7~~ws)<_>Rc%6J&HG-f-Es!xb`ghj
z$s%@Fzv+4h00H#qpT7|&e}a14lPA!=3;zxkR(Hpk8>@9WB2WKb4*KPTm(v2ljPBne
zaI{-t@W(_n`}A+boBaOl5Bw}eaA_sE#eYN6KHk;&QjNB&9ztHrPf0q1(_M;Gj2M}c
zMzT#6Qk^zHE$l>!#l-v|sB27_YI{t&R+BD~lCHrPu<d&qIL$a6q39`6a=wD+5MgT3
z1Xu0ArRCDe?ZuG2<E1&r&|c~drSpBav)=o)o$h$-_G?G^zRyV>45%N4qBkr#<9$cP
z?92w|EH}JlJ$0?^Y@UW?r4X6%d0dq~Jy$qCTO;d4POR5etkUsC=T4WeQh@a%UsKD;
z+yoe)MY%}5#uy8mTkzIYGl9eCwW~b%b?#@rf@IiuJX=<NZZhNWdS+nAe2j?gDog|E
zo*Ul&LB1FvxA`S)f$gr15wI@%5;znA+u0I5jwbDt7M|NiozG2=nJI=47<lf!a#f4_
z6a6#hONc*VbC2!$s2K7URI*V5wURyK$q%5`ku$4y^S-9D8W2&G5S}>aczKx>9IyzZ
z|FHQr+yI4)v@^RUPG@6xtCD?FLJtGi7pOt8iN+Z|B=nRT$Bh^K*wo74T**!ydt~Xm
z!{guG-3{J-XgQ`SGq;UohsC*aL*<_xg}oq&4>@4}x_paL^;>kHrH7zG?d{cXz|>hU
zA$hs(CS-jaq=Cq+JHG1X;whT5oDICrR--qnx%VXPh8W#Iw2x(9s~2dAO#(>@M+)>4
zt}VR;@NrRVeq~nJgv$3QSR=~PZ$#8pT~&8{%=)1*$93^2jSJsx!S?!=r1sD<%q*3P
zrpW5+vr<SW?cd-9wH2?SP+4Js6u5oV>S~<wo#fb&975=Mf1qwW5@?38eQBGK2dUS$
zCvoq(oiei%WQ_Q-+df=(gOqFF3r#wUMemP23#V@J1gKhE;8m%N8*}pD#KOG;vV^y)
zHsfOZ7ZNv$vP2?u1gg~3-k2_Y)5FR~9Pdn345=zH_HveQZ51|b`HlJkGB3#bxlrSC
zR@ON=N7o=oC-?cbXV~(Z83jg2JYATj{fNNk>RVYkQFqCKaY$w+hz)Hbgga?buJ6C2
z&+yLpxtf}!><qDR`{|G7J~0fu(0>bD6%S1+o>u0x)YP}VQO&iKyIQX>I9bs2#e9<~
zE}r;zxOCY=bkHr0I`;rryRcYO$K&&4k;3Np^XfWpOzN%^aao#O4Oe>;RYGIq$G9w?
zZ^|}0CVwh@P7(V9|AJbzO|IO4OE=(Rx$mhEx~B8Pvx--I$>~Idi)c{KLW1antx;BM
zOdy;R?E%+7tOofL*SxXlysGC%wI6K>V7XMs=cC_t=gdm;jlisQqn(h*q{t-Uz~mH&
z+M^!Xz*YFTHWe;><WTK$zw0*Y=-=>|F4HL~jYFZ=roP(4ex$6tQ=rH<vp<{=CCB$!
z-iQ2SohfVnV^0%Wv-QnCLi^!l-mL95eVf%JHkR#fIP++3<>l|x9nE$Ashh_Vsj&zO
z`ojfR%qQn`UciO6w~odhPWf2MZ#}5VYxE*DiV(io(;G!JSGo*tLj*1;e!fDluFhQ4
zYA#;>{$F_^U58+sPl;_`*o;qmWh<<@OL{h%+PsVOKDRRoCnp6$xH-&GE!hy>l#m1!
ztF8T*^bXeJKOhRfQzk}{+WW0QoK)GS)1$zV20zvTxHf~i_oHwFOtw1ugCMmJRSG!Z
zB=f&FVUJw2D=MySWRcMl&>p4Y#MkBSUWfGx?P!U-|87zl$ER43POk2!@zSGUv(qeU
zzzS<{m-CaJf5EYIl9%cI=Yzh+jqEcGFOb#aRtL!fYNS6~Yq=d?{d{PlM=6f4@oMd;
z$@wM&q`t2CpVmU}+qHmp74IpL?8AhWH+V*VOmgpdi@qVid*e2rC0c+`wOo^%bvFRc
z=UHF%T`1G>XGY9|tDGkm)-QUETSd(GdR%^*<%xIu^b5bRvc=-Fo}9sDU(+!kp5u9Y
z!<6*Px9O43dR$!xPo{n1SAVBh<VyS0i#nYFGps7Npf3@eq$*nPgrr5qAk@xLXFI9k
zi?7ryjz7NpW>t}F6zKAhoxP61z>HMqc;A+m&{?y2;t2NbW-DnUyxXx3aASRUHipb}
z?|UMneo}Uf>JG)z>wl8V`5)si%8FlDSWtzX`yumKzZ%lsFsEmd&9;=ShTY??so7)%
z?xY}C&-6F0u{av73T|eJ0uM?Z!wKD&{lm%FE<yyJtbHNRh4`#hihR&!_F#Z=qD%Ho
z%b=`J*w*>n%;0IlDMHwFyZaDjtvx;9A_oxY(?btV!~7#}pY|L{aeHBAuP&~3l)N-f
zx=JsQaq&YA;pVJ6V<!fe7gr06z4btCdA(zi=vsDH+lPb;bne8M^yCrNrhp<ClRAz$
z6IekwGYZzftP-f1t8Vm0k$1hCA3G8*Jw7`UmNlqssSI}N+v-}{+>2Toop{z_y?9=G
zb#51NIyb!_gXBVGGv{1<J-P~MnTz)rmZezp>-~>#)BqB92PJ66_|m)(jX-yG+qk-=
zoOD7{LoHl0s6ptuQsY0VF1RyDD|L9$HUDx~&*sR6S-fS$6njsXV(;U2N{bp(W{8t2
zgI#u4uK{N~mwkYm$7_r={V%$TgA&eUHM_zFy6w5SnSo%C^TNl0+rBEi`X`4sKSCl_
zE#+uTCd{M?XGNj{Lyc?!Ak@Dtj#w#it=i!mO7Qk8NJ4%GAH6b}yQ&6-6lbcgqfvfg
z!c<<4ZM!V5{If5&a@5pr{@4zSO^RH^=XW2ChiWO1q}nu!-Vl;?vriQOjYC{9ysFWW
zi=}MT655*I+Pq9npqh@4#Q(e5``*BL@knnPlugU+<BI%e10UY00V)R^+@#%##jd?%
zEZ4tXKf84j(Qvu^`Y+|uT{p=4t+0Z7H=cu7w(n{={`vxCTq*t(>DOpDtA$7djTnrI
zV(zsS^xc0|rkhLwpXn<_QA3ooF0xono${|gmk26Y_Dh@gsIJ()&waflce812!)7)V
z_{H(o$1b#_M#$j5RCdtnx*mGWYIv0<1dS-{T&gJCz#j3x<!Ax6<)g0ccY^-#udhv|
zuAbcAF$WraZmLB?X)^L-ZA%!r4sIRnL@N_0zNprkD=U!Da6jXFGtR~MGP+etuD0ad
z@23@MdW@f`sh*9wsmG1gqO;yvmvsSNPa*d5>u^te@?cN6IU2}w7P1m`PHt|-^eB`G
z@|A6rExe74H{m#$)j!xE-x@0LSF9gaC$D^$e@!__o6~dnXN>mZJb~hLHRAC2a)4RH
zZr`RafAZ?D-Njhx<$wsmn^4lziOW8`jMc14YVm86S329tniTgRZ6XBc<+(wrZnz{7
zSGj#d4*$k;1_9H(AF%nVnGz{jfar6H2fc+0^`kZKi_&~qPuY>(XDcI)m=Hod*S$6-
znu3{U{y{-)SVCi?4@j2xdN{t8?E4zk(YsaOly9O$Qze=vOaQmgL}K!wV-!xRQrzOW
zoY3AX=k&fb>hfgq;+)uWvIjbZ!*^Hn$`*UijyHjWg@uWQ?clxx+wFDS;GaT5Ul-QP
zw)?!<&g+gbTkMAd*q;c@KK3A3DTFY1FtHrGIOe;VxKcEStD6)-Yw@R@OiYXoMl{LT
zee>?>3m31$z4cxDQ+E*bD~Ks2WkiM&ryCQgl&hXW)LKD4Fc}XtH8)ka|HdiHsj)RS
zp9?e$3y8u0bqVuOEa(4-9Wp5CyZpOA<TFoO9MM7<!pY1R3TKE{M~#=WzQ58gyk&<p
z=|)2z^aiNX_=&P8hGbgs*z2otS1$~|4*(AvLr-#7J)EAbSBPs&ZSSv|{J~ioUn_G>
zDtP%iiBkeRKV87@-I;DjF}idt24lD(;{cwh%2fCGc79J_%-j9=@-prN34Ir%Vc)ZS
zhSnM*qxGNHmm`{JxcDUqM`v@-nT%Dq?T&r_HzpAs3V4W)Qj6Ygf!N`xN9ts;txPFX
zU;CSd5*sw}5k_&#!ix=Cw4b!0YK&<NcOTC0Zyi)X>eJ_sdU_+(&*4s{rr<vVP-m`d
zPT!-bs}yhY<h$WJiGtDXR)qB?>Jgfp0upV=!cOQBQ?c^e@om=jwNolK*)6+;nk6~C
zNpV3@P#U$?ekiu7H@S`YI;p9?ih!=v-LT?z6gwU1_Z`2UmVGWTq^YK+{BnG=j|3{j
z7wl;0jL}=v()1YJ_()e_`30f^Kuz*dzBER)h-#8GSWNzsJ!7obG`L!<se|#xeZ)OC
ztL^F~nCP^9iN$Gt@Az|mz6WyHaN#pu=BBH!NomEeAJ4={aY8T{Qi#~FZ-e=b!<m*u
zNnNUsmvXCbYsatg`2T1RO-dvs9skk^durhM&_ee@_-$#>Lg^)uBkCL&vD6C1NiiRx
zOV9I^!U9_^kCzRUP6sEi(ts;zrQ?6#+je7X4dfxW6V{TSDRu=%Ayi?WAB48dXQgRm
z<hm^CrIM4wvo9t#I~UW8DA44!(#rR=B|9>2n#Z@BIv@kvK4L1fgKhA22e^r~$M~_%
z2f)GZw7^=+V4VjZP)>Mhj9eB)V_9V^V|`gYeHMYe-!ljIXm4BJM1%-_FXReb_+?Jp
z_a815YfXy{AFG`o4U4p1+qpPeBq3d5?ArFeZt7C<6z#uSq8J|Kqj4)mT146tY`m%W
z={x;H@JAgGabccsKhJqLHvGF0EQjuJXU*#0ngjxGJf}(o9GK5+j2LH2yQA1uko2cE
zM(E9OLf;uBkg!@&^ByUA=FKlS;K2C_eO43~|EjK0hcerk>B~2^PZGS%E+$Ns&yOtX
zY{E}sl`d}zJ`;MbWAYiB+9fU0cRrYSwL_@#W<LFQ)T_`ZUond3r`h-t1uKKrq!vSv
zFlttrr)CcUkS6xL<}~{;gQI0G#<C~#)^|xy{@3LkhhSGKP3!@hwSr05U5g=(ajR{r
zw)l+^PrPXD7K}k$+Dnc`)x<28#wg{Hv019M$lo@vA-J0P^`Ujp`SPsQ3Y<`GMhh}B
z4T-)bPxAL}vY;6-Y)$F6Gh&SQU+`{}@SF|YjtBza_jx5Se&`&5djAZ%$KELj_fHjC
zR(oX1(-M514{lJrU`xn!QmowQt{TT!Nj6N8KUK{$emIKb53=^m>Hop?Vbzbt>1Rpe
z!-C=jj{}mkizvm?A=o3{dg!u@)w}FwjP-yhY|2Q@pgO4GRAQa)ePV!E&&fmBRodmw
z;ne|f+wac{#`&LA!fs{?O$@sQ6PvTIl7`us8kdbE0rw?#bvZw7MaUa*D7Sl~X;v6a
z-Wg3QsKU5}glMJTH0Z1oB&UBw%7lB~9$Y))-R`w%e6{m=mPcHi6d{)Msin%}Hr@uD
z0usOmkGbcmB?0++TvCjtzClq@fmgjz##+G2Xy#w@rYpl3w#(2j7g@p@s=sy*O+$>y
z(Cw;Kt#-xEYHAvo?}g&Y`XM7_(WejD6e_27_UnBV=eD~d+pTRVmH(c)^9ZlC=o=TH
z7Q$9hCE|AI$NsjraFT(tH&jitMfxHZK8@MTrd^2AL^MT&5ckwm1=MrG5DYNCrENfl
z)>QI-P0gDJ-S>-$r;qHO;xz58vx~A?^}k58oJhovh10PsMm@IymycarSO?Z;K5Kf#
zmEU}?Hi>B}TD&=o1D9cwvVR{~Bo}2mjc;|;#u(?n{`j~SR$X=NMd%;w4Y+&3$;mjz
zg>C4TJ~r}b%~tXqBUNY==BJC%kHC1qnAP1q{acKO(b(rRvA6UuFA`>kez9=ARKJ#-
zsUF1SEsxZl%9YNv>5bYwt@zW!ahX=gWo;c|w=$0E^w^}wB8B2`jH<(=bFlb)mKQX#
zc*dJuVc>&e{-4>|#h*V>8f2IksYMCf2h$GS7h`?rb_o;7V=d*R{vG?wk2y{Oua!DG
zA?_c9^m{tw!!;UmxfrXv%xHW?lxlt!w3qkD)pf$+O=y#ab1LMm;rrjDkZB27LT%NX
zDuLCf>|Sm#-*&9I%XX1<y;(u!<}KQ|7d)S9I~g)2qN;O6_&-_*1*CrRvW^_5n$Gz;
zTtsg~uAljpOdxxXrO+QEYwDm6$lfyj00Q<S;P=&$CgbDUGVhWgn>f`3<I#SQb=MzP
zg%hZ(sR=j>+Mv#{m)0W_dI&FrHoR)F-JVx6T<yw)uv#wtIy5Y$DYjg7@-#0qG&rVG
zZM_FwekaQ6CpacXQ*mA5K(<v(W==e5_&ow{^da;{?&<iIw(=F0pUwwx*$qp#n~=Vv
zH@ZjvDw{9~-mcRnhrce%ZdR(Rp?*889Bd26IJ<t0O<*5}{&r|*WPg_ldDojC7nk(L
zfA$Ex--<LSNWP7CTQ&N4H_X=Nv@oNhlj7hiNC7@q4QEF3I7_Tlpq`q5-)!*KMAcRp
zJ%;leqOwT+GymwSUnme1)FAliz6yHDy3IRr`dSE76+8fueTvrF#~$|2Ycd)2K<JAF
z*;Duh+`bEL!(DP#@r~8*np;Oq`bM<Q7M>M2w!cD9vF4pk$l+6D8*_=4%U|XGODvPp
zRpSY@Xf%NuG&a@~Wjp(K;G{z(P63Lo^>?YU-q-1CtI%(*Jq2QPdp~zyeMw*R!sJ5Z
zv6f_Lv)-aAQ6DQFR!3c$G<F4VyqD`6m8bQ=eW&XcKR6RSY)?YD>h)TF801T13u|g_
z6TDucxvM#)R9Qc=Q0k<*OAQpwqP^8aIHc9IAQD!Vvss$y-wPt5%CM?4J@cF|iTus?
zB+{iZ_7qa+7$6;L=|n<ugR8)){r1Zimnz0*!zP<t6;yJu6m6yYwRY;ncV954eg#z*
z-sX()P?~}D(V)^QA!mA2dQuX}RG8Dq?6%BE5d5W(g-QFw>Wqv2baJ8F2kZKhi}N|=
zQs8na5!g#yWDB00V51>VRZ(MbDTrw^HaLkf>(e|C2GZHy-!=u(mMvroO-B0@kegMv
zY8cBPwE9IF+aIe(Wb>SqzjZdq4iy#_c0Q+pi+y7$YAM_X#nGC&Tf;xEYqa>$lQ1d&
zP3Udd+77xO2X3mtOXBs{GO&BEAWKlLY)H0nqRK%{<qkp(=ER>m#p=}U<hn%{M9Nx}
z5S2dxxZ=)0WDMxJZAzkdRwf%3YS;ns3{7?N4EoDY&ct+Y-YO|dDH7K*V(kAfaKW)@
zn7b=V$#b49>)(#aJP);J>rz@mk5$lnoQ7OG_Ya8_CWA03i8GAt;7GUbQ9Q#PvzBK8
zPe!*658GbYIypPkIY5_J3aY%Y2i~hiL<|nV;qLGlbGsN(z`!c!u9-EWOls#i51?Rp
z>}7Gay?99>*LcKn+GIuUYWARl)xpxKyGbqJmlQ*cOfbJ4t!HLFxn#V8=%LoQWp(>{
zI%`&@L=1$SoYKE1I}5>-;aGHd;oY^E>hN>TN7tjLH{AbXZSBxi%4f8nH+A4wq71Oc
zNMaFSLp}OjY8t}kzkcSuxGkoM7oz5NM6;Qwe5vuCpX;ux8K?OoWc%h#`y?GTbZ6=$
z&A`o2|4N|lZqzLm`ivz1e?#<7V&aD(th70a6GI?fjOkYHb@2#=P1i20#CTWPO$nxt
zA3R|!)!&Drx@>ZA4{30uP`Zn#R!gALgFo31hb|7zOAgX9Y46jQB{O<`b`XXQ9$++v
zHoMo1XU7$E2C5e`U|cSyM5L@i7z~6HqOJDmiINpM56_&3UHEJ<E^ajzRN!Z6W`CX^
z(q8(9(UqAwz6H(H==~3_F0=T&&3sIjZedVXcpQ}~?Xf7rvL`jne)D<-Cm@Qq=bCsq
z$tp@qFV&!qPonAZHzjRuRYN6_-x`0#BZ{^sW%Mjc53v*H%g}XJZ$_-ADE)AMuFZVF
z8O>Atk1NyTx&!h;SiiVkD%o?`RI<U0V^RJBBy!M86Ehr;<FW=9Fb#@wc>e2qRL)`;
zEW#E4STpFg2}P{!;nq?Z5SXWyJLd8?c&p1ip328oG4kL%E3&=gbN*s?sMYs{8Fd>t
z+ILuhnz73}PC7X6NVuMUB;tEn8N^wQf%Kj-59k#}%cN3%b+k-fhF3)=jjf8|<x3|e
zZ<<S$CS50(-qw%iF8@pHBJ>vCtkpG~P|h8zR4RD3T9Mr8dRp)rsDjs4nF;V&RS?xs
zbKWhrqnU5ovD*=fl+nIyrqpSqj-ujOt)nmVS#KDVa+QnfbH&xvK$NdNn3f2%2@VbY
zKoep;vn*31rGf1L3rk#EK*a?Db5(%w80WH)RKe!kCS<c()J<I+D5wV3d!)5SpPHIh
z9wa^-zVFoVt>F*Q%gVDZ|I<Ut^pT~_4gJ7ZernfG`6Mq~TLfPBTUK9C-KjV3$MbAG
zQJ$TC?5qMEFE{C$r^%nhvtH{{3?$xuIzg2jwT!49UH-N<=BuKlZ;~H#N`%pK+<Nc#
zGoLj2I>A73lJmv-qxA+$2bipWA_H{H;S&_5EGaaeoi_OfkiD@CQ?Xx)?MO+2=uUqk
z9AX!QQNP)$vQK+Ch6L`IRZ??Wcw$4h9+!e}?|68AM_tXF&=d@np3Nhjz(fqSF@=1Q
zjeR@p8)osd##HSftsfGx!?_mA{e=2Xp!`Lvc!YfI>4d!GOZ|=*sSrmbDMO$qmh0PE
zk@&e8&!r|1T~dT>N-kqgC>N5|UzX8#B20vUQj(s@q!r<T4o^~o@gF$q=gqEl*fVY`
z^yWC5>X#o4P9l~<#wO864=upc$*@f0`$e2o>#g74WM9<<Vf{|*`|e=d*c17YZf?ni
zh34o}54h_AX?^hnbcxcLd{s0mVSMo8<b@)(>;NdccP}|JmP9m8D~1y%_XSCz=P-5C
z(&VSOny&0wvpqcpmkN1nT{EWh%b!SuUd1M_SQhabJHLJ-qbBJS=AI&k=ttyh^bjrk
zUxbkL=$4iBS3IexA=m@G=<Lt~%a-FWcw!ejpRc^jFxRr7`3)8ZmPCP2q2V_Am5&vA
z?)g9`xc_b-41Bgm-(R}e%CoDpRglb%I-dms0Y$PeW8p2nk&%S*nh9slcfLtfq*_cT
zi&u?y=O=dOE#XJ5NX-Lk_p_2~85uBbJ}D<;US8g^%(FAl*+1x5#`>r-xD8wXe>9!<
zBh`=h$FJEnyY{S*y{}EibzQp2=Hgmq@4YhbwYRKf?|H54vNEr|xk4y2yb+Qj^}XL8
zKL5jOoO7Pzv8}$|p~kH9$LHf2OoA&G)c5M<&-u=%`?^P`xjG8GoXsE8AXWj@BDuq0
zy0We=9>zw{P*0?h?YAPg{9R5zUEkq#5fzme4aQRz4ECwD7(vEVa9J}_nrzUP5}Sj+
zSIySO43gQj{n+&AWbIvoY5lI*@Cft~)4Lf_sBQ@i>G{-i{Y-8&9>-PY^2+V&Le6BF
zUJh=`s1CHZigIkt#ROt73x!iI!R0X@T+(zJ9WfV6yze%N0GR5cX@?<kO0d8mXHZ6Y
zi`9L^>WP{jm^glOJ)HSU3jfAKf9XPgzZktn36{ro?#li-&K{X6FvO=dn0+-zhky1f
z9}j`jUoQeC{&Q{7jpiSoVo?S7%f7*&U_TP*@%AtjrKF<+EjnvKF1DAvzefy1>o<`#
z^|f<xxZq0srmuJ~3-OQ~UncY<pF!Q*MKis0)s0t+@Ych0(mJHfQgjxBv9Vv+%3d9_
zE_t&Yk<$t+7xz-$R@^NhflPTHZ8zA!XMOfR-&fOTgbfca%5m!NyQOFgjL?%T?Zrl*
z{r{qb4w`tH*l$wEuj*Q+yw^?xY0IxWT;6&0Irj=BkI+|@R)!&U=acaKLD;arJ59->
zQ%Z<?^1!KTp@JuxI6>mRog1@(K0}q4?CpQYNum&JmGf-3hZu-wKfECNo16S2wZFTq
zj-8!#e6JftMP~~;FZXQzZSS4bnFLX)ucA>Mk1zfXsBaiVbFxkK1=#Ku7$Ad*#|vtz
zFdh<U?}sVcZVkShIXSc-jx=Dp^bC>)jz8C8nG=1A3|eNGRc8p%(OGonJH9iy91<KX
zYDGTuscBI5pBo-K08Yo}++9Uzj%v6bmF-gozvF|ihp)GK{{Gm`SFt$LUf!gWx<yTI
zFcVR$=}X(WTQ2R0Yv*#7@A};yjf8jVICu4i&<!&0ytDlHA>?weRZO|ByZbjTu<wnr
z`I*l{nKPxn<osETWSl@HrM#M*Ky8iY-|?+KcOU;IsYsIo{-saotgGhv2#O!>_HIWe
z2KjSzoQ1pzjRD>Cd15r3*-6UrqB|;kvK8uW*K7(vm*x0Klb0*F)a2j`AB`!?Snjg8
z@VRmKdICm~V%;*&y~4?s0<3di&h;Wcuv)aS**K60@6ir>4D%a{EL;pVb_G%qVO$l@
z4>v>F?*^lY`UueYzkw|!u)u<QNSu}1x*4Y<aFp|g{LxjA(N)t_U*G2dRQK}3oq%qJ
zbbS7&@ff5}ha&Ognd@#`WHPX?{#%lqrZI}QU&jYWn?Y(?BlmANZ_6-p2WqlBj%U;t
zp$LOj+fW0eI|Wk|G)S2>`))1pVCCv6v|uyz_P(TJe7Y|$<hjOr3*Hsr*1ruNJ;)R_
z^y7zzF@Ax#Jo=qW<V=LhAOQ_<h{I<Jw0X<Kqc_4kH_kAaLc9w7n-V19c-GTwmuxNn
zBd$H_K~?`0R?d3f^RJtX(1gM37wdly935l-(E9*ArH;iC@uw;Lpp=v0eOi;JF4yIY
z@ry+o%nY)`=yM+sp~|gi#UrVycKwcQTg54=3r3wO!5Lur161P~2wb%gQcXhx$&rjA
zyyMgg7b$u1vQgOl=^IBMKUEn%(cT8kAhLxzTBbk_q9VrkGm}WP=V}vaTd3<KK=3%y
z*GVf%DsYn%eXavYX@`pQbf5k~{Td1Gb(JZXJQ)$M2swBS%TUKlN;68Ra|h<w4)>W^
z(UiLC<hKXcD3ic-U6tR$*jGXLi@-eH=?a7G;j7PE%k~&0WORJG=HpymomQl|*%QK~
zIv^!c5QeLQllzz-%w>&O;mTePr=}v~CY3_KS92tijMUR@6ps49Sjy;v5mP>7v}APi
zmV4}7m(<He7yzgtaa<WQbHLuvM8n~G=KZ$&bwC|z*49F0>(j_yCnM8+{t1_L=}95G
z#j}8*j&ajZm_ym-^#&{DTf2`NcGQ-gy&pb!<lUU^PY0QQ^!;~!*BWA|wdfZX<gMJa
zwss17PIIck6v<L8)ndmV{%dFp4`lz@g!k|S{dn~(XF2sqN-&J$$!l`x2;m!ALHp$R
z8=-bil+ioG*r%CccS3JG@s1H*i}?3eO^EM}ihWqxkzu0N;@0h%xvM^5DN7z#sokuq
z3*6Cvv$X>?%>&G~g&QXUjo%G{c*D8dz7`CgMbmkF<r5^6VypQ<v~JqLwy<#a01wTJ
zYYXn3hp_>4HX^Tr8=5MmpK6UGF^LkD(Eu>2+WFD`Wl)A1Hu(GG53B05oZqLXI!{3N
zWJ%hgQW`Wp7(o$NDd3@fEh!mmzfowAX0NoW@sS1{>I8n@vZ**y<8c@zO>KG9n%RT;
z{Dfq%<%;=1JGdmLa`)59y&JmvV%p&0@&l@ZlE3{d@nibrNo#yeq}H4;*Fw%ZU(F-b
z1RRFV8w@$<0UKYwa-U?iO=-rNsr?k0rY7h3$n#>p8yTHy7U~YW4pT|p2@L|j5YDS2
z0qna`${-L7m6AAC&oS~Sj}}4#iReMtZeiWTMBR{b8ISpm4{pX5dfX|%(~_Nh9+N+9
zer7Sl(#xl2*Y?{U(01!b8Gdxj8;R1aC5nd|7(9TpeULxc2~bZ73U1UV&f%$Kht7@F
z@X0%B(4Y7dSv$-KR8PG~>iT_ZGbS%TinBkMQ`m;V$Ju!4TY$VQZ5;59ma2=af?Jn6
zDhA~;R<_)-?_grcd7szjYeQms-g~&fQv7Z0weq1!;hB$xnyLGX+_Kfvr>L{qv!~V{
zdfs33)%SjU-(z8ARWk5atT-jA=ko8pl*+q+?jNQLpN#D4ONO*IrAcT*4vxE01Sq1s
z`(Q}`oPvW!OQpbLK2`n{nzXctcr=G&GZxRm838@$Z7ux~W7Qa}jm>F`m|1*@wAs3!
z|3s%0%^o`^WzWoakfT~P_6=qE#i@=<g3*3}w(8we;&HhK-#16+SN`wwZG${zKVZhh
zialSkp9*K@OComU0&eqHnu^(;YcG1N)WzY!w!bbW?Vr?<f8O#pe$}}tf3Vr}VYACL
zf4-Wz`rX>(rODxS`1fD)-S=Y(ym~FVR+>B?Ds{oOw|=pfpb3l;A*1dnovv4vtqdbw
zjZx*?7Ecu<ciC+hrSL2ko`XdE#=kNh`gP^F-JzuVkn>F{|B*mVA@>5g66U=gz43OW
z??dQr`RT|*CVv5sqx-Rx4ECDitmP)yPuKCLjFmUj!3Q}Xgvl$^-WwQeiVW%OJP#Jv
zUUbr^WhBHgL-V#PAX-?|H{m|8Mo#agZH~{(D+|H8+O6>qlmPl;-K&Vcxo5N(0)umt
zAgKMq8?$-Q(?d&Fk`jW~@mT{VSu1~5LLOa|erPC?VI3Nzua^dZ)aGPq)gZ&aaEurK
zex0Co08cA9_*oi+Z{}5KQ}-{AIL76Z!g)h~sHt&B4RSmikxd{$5d+C6o%3R_C15*Y
zq#CY7dNd>+ZfFOy{HGl^Z-S2Mr6wd~Yr)BZn5&)1OB-F!{E@=!KQp2@;#J%1C`(kM
zBJaZEIio;;1TV{F{JF;E7+Xf*Krd%5OjGmy>Q~)sL;R?Z6znOU&BI*aSO8=KX(Pg+
zCEXC11R;7s8YMqyo%N-mXq0C7E8y7c7oT{mmjiThsUhCW!Sw+lHiz_Kdmi4IW;k<H
z^jMHqTOLJKaPM~Gmh*s^IXyZVK*-t*!?8Gynem~u4c|;FRxtChoK0Y+kZl&;eM`cs
z(e3Bc9h$<ql@07%oaWg^h?bt3hbkVPr@w|~yNVl&QRO)YV)uO1_n3f7QkS>a_^$dt
z537mo`hGW@g`NL#wz-Mg`d8i>;<#=iM(*=T<w-G<y?RnB(vU#^Fdv(KM3^2*0Evu_
z%~>4F<bf+D;$IX1kQF3uyrFYH+^6q*?r2t|81>}8D)FL%3ZEG}tuX)~%jRMC-)Yx-
zWErNOsOXYoY}Zdt;sZ+?aZ-1Eo@cllQJz@g57#kC^7A=nAK7;tK0UiQ2or4l&8rPQ
z4eXA;UmV(BHNAs&md*eA-yi63Vc`q6&b#gZEVid7-@dT;_-c)QWcKezRL%Z@+DG?^
z?WpUsh8KA2<o>G6i_Ng#A|b2K^gm=6^ac8NSb3rV!lmyjI{mFQKRabB70-UmYvf%}
zq}lR$uUz}F+S#Xas60PWT4uzE-d<CF|7+F<SC>!;pLs7IuD1xo9N(Wa;P~w)?02tN
zh$W)pjY+HjFaef04ajhVA<vL+xVd~XLXOUbK2>G`(&dFNTq5hUmdIA$DC!TJWGUv3
zJ6)Eba%aCdJ+^rJCgV#Svlsie3DDR>b8DY0DZgsIdima%clE;hKb1W?pMu1SotdW2
z%IwxvHw7J^!iwUXoC3+*A=>W+ml4*kdES()y=??~**r<nO`X<>d%){^IVxcxAmY8b
zfTdEt@-GZ@6x`fP)6dH2s<{T!Y^lM!1rZU#s9Ib!U8}Ve974*lefUyC*vpg#-k%I1
z9glyhT*En@-fXvYIM3?Zz!qD2e4DdfetSRJ*Opl=C-m#YV3b#4gdrEfOE8?j&GO7<
z3@}VtVE;kx_U?Y4Z1FTuFXLrw$XgQ=ll$wd7beAuCqI=WDfrH&R(%3Y`Jta)_m*UR
zKpAt~vrePL^ONKV9tAkmHnmclKTuTs@rR6q7D(Z{KZP{?#2nJ0xPBI>zYwY{yNjhC
zl;S|wrufUm@<$FTTvu;x{V|rkE+)NcjZKIj0qkM~G-ST@c=`01wq0B3U(0t7bp=mO
z!R1>`pK<}6n1zR|Peq5F>N&kr)3|-NQ7u2TK(riebQjC@KU!o7Is>obP`}Uno?}0d
zc^1@#|BHIm?v9*!UWzrlyg9vcxZDrfx_!C1G!$IeJ)nJzB{j^R_xdSWBSIJ_MN<<l
zQ|a8cnGWWkNmKv9GA`%(V81{Zd3Qr=0H=&6&XKDIyp({Xg4t}<x%c7Jt#%AWD14cH
z{9d$mhJaj)faA_dR{2dLwe9!K=z_!!oqoI4r*zv`vquI2js5JCU1lfE!rV`tJ}+JA
zPNrU-B1u2rzZG~dcO8DdX(La^zZJ1J4ZC&v)7C2p(+s@a@A$WUP;0(3bQbU!rgA#^
zS$pWe0Dt96uy$j;B}uVj*xk3U2o!-Kp-TR8_ldhE5#eK1o0k#=0{|MY!Ko>KOFU0+
zItZ~;b6JN!eJQWK5{Sq&D2qx`5G383g1rzJMsSw+TxEjK5}Ngg^*wio9+F;t*VNY^
zk~{9z<#frYb4yVE&@QuQ#DK_wCf4$k->0?l+hmcDuWaG(_BgAblt0J<qQH-0F7F{J
zd=(pQ+i^u&FQpok_ck<lDzYXpS^|f+f{9Q^Kl&dGk+&P0^3~@Cuk{X<!wfZBzdf&c
z`fnU`;^&jKF#q015H00v^23CNkEYV=-Ltv)vM}3&&!IBiy}j;htK-jX*!if$H{S-&
z?(J2EN19pv@yKBc%fIN@*+miT-;ZRHUa+PSk4<TIk>?<k_tli&V>g@$Qf#v~)By`*
zuz-Qg`YBXp;=BLBsBwS-PREXMvq>zgRCJzCE4%dD$iZh^wuCVB6n;4hz1v@&-f(w#
zIAny>#WK691JWm2h_k)(0&bOB(V*7k#VpA>vOYdD;eEWxwWe+Yg`9jwo4XMZ(!KU)
zNKE<9o|tmz?+XNQz;w32H<dN7t!geK!dD`$NSzb{0t1udw9SO$zLF{DD2?g+UgCn5
zO9X15QO@L=M|E2LA;psShn%2GHyQP+9K;iLpX@^G`2Oe;hJC1^`ao<vr-z(vaNXCY
z41=dzkl~&@4sK{c0W<vjxU;iX75MBxbMi|*>+++6UHWiCMF1p<@qr^8d?0snJ*12J
zJ+F)UNys(T048Owb^lx^rYt=&vF?Bi*!}AH_h%EH(`vICdj7kwls~l9sXNhgX5FvZ
zt!|K2ZS9d7;J{6F1my2r4NBB&!6k5j%DsUyj$mx!mjTogl9q<kejj|V8OOoJcQVBM
z;6krun!}sA09tT!chtpUGsKc3EM9({s><uIQzwA*3uk@|EtJHj+HMQ%pNNM3b-Ar$
zDild{EOvji6?Od(sw1va>{w#C6=OFu^C+{9V6ckq;R{P|=9#=!*<#Br%DbfeWiix+
z$@A9BN>5ZC4!L$(9e@j{+ei|Vy|0{8{jOD~@9rc{<?i<F*6n`z_FszxY1nwdowx6`
z=jR*i8puRf3-;%bo@~3|z(mVZXa`0xWlN!(9?BoeIR%8+RfBr2XLiIXo!Rfd@^f2o
znW$%JjCYzfyb{mLXOkHH?Ju1<;9Zc{qH~Tu^Fvt}V30k&qK?c%#2nZ)Zuv^L27mPd
z0ars#Je{QXT7Z|Jwl43@FXEaBO)7j$8Q#XGV8V!HO%<pzl1*W6ux%r<ZlD5lm0shG
zcywjHeHaXwR6z&dcYb*njxx)wi)H}$SvtOcQmsRz9qfGkTc6A7cHNA~*~-|&e_!zj
zD(X5_vRv(~AaA~w+;E_mpWG6u4zpZgbFCKmrXb0MnGnS4A=(2+0y>wM8wDJ>1nb#$
z&Pam^#^XjBT{y*FP+294em8VZnK=T2kbdxrK1pOwjPDHC(OTKAu3E`3dAPtwS<3D{
z`>gdVW19O$p=!}^&4DCfhW?sH_1$+Ag!PcRIy9u472C`&#0Y|brgClL<t4dzQa8d;
zfq`r`l;D87cV-Q7*6@Oe+~nG$;ARVQc)Y3)Y1&D^^?$2OAAgOgD2DuW>s(hj0CD(g
z3UJ|BeP3$C^RNR<B<zzf<oq7zNI!Sf5paIA=jF%O_%wv29~q4xPx;dGSd9}^5&jKn
zx<a11^XdLQRed}X@57D)b9B_}yEDR@-+Mf(<d&|Jd&@)DN&}-GZ-VqqbRvNc1h!=O
zEy=krQ_|cuB8PxXgo8Hm=Tr)q+d(suzw1B6$}EC01>1g#RBMYV%&iI2)<02|K4xft
z^P`QPlUSO)8SW2ZH?7^`Cmn$(+&bbT=Ez<P>Gb0J(cinD^EUFAmgbqwP+m4Qeqyd%
zNi*}6_Qn_DGs9WLepLV;QsN1Ysf4j%dIB`i7dM-%-@@T2FE_ATU`jl|@4L@Fo&^Dn
z<@qog@VXA=#D`Lbb|*&IuTu$J4FL_&SrOhMlrK4WUdDQ$P0Vlhq_<G0gKb{-w*Ucj
z^SceVN6J0_j`o6;hw={|ZESun*g%QtUmb=!y>-N6zSpaZ!+R`5Jzs=w{&JJ|#di=R
zmyaz!Zu}tSC*5?9TU~I>*+gCUb}yfL{`+~rggURz2Bam-<*2be?N-7(BU;W3x)(fe
z&8wEnqW?f)uj1i+*fT*?w-gpM$4p4WKq5$&?@LO2*5gC%!dELFFWUJ;#2%3}O1D+^
zUvvPHYFjS3i*3#<f7@D=DK(%kZ|XF*v0%FAa(x2_iWuOQzp}Jk79e94o(%r)9%-)4
zT+e$!&5_~I&sP!gVD%jfo1YyVWuLFXe4Zk9Mkkt<EAj_xYmLXKssec)uSUG1TMJ<#
zc&HCqnX7!0Sk#EYw5gODEgZJ7_jCb_{F?6h((*bRyatetn=sV$v)7)5P<8%x`d!zo
zdRSj3+|gawG{IhmYm~Q#wcw%v>`xW&iLZD$S1Oi2dEq}U@;eusbZ+3PV}o(4x=^W#
z6mRWYb?#vfK3~)hk*Annj-Jh~+cF;-MAduuh(!&@?3(rs@P>;U28Z+`O^H>3WCSg6
zt>#Ch5W%`Nv-@c9p0(aXQE-xs;_0a|C)S!(w)<pyT6^Uh%%f^(X3nU-E+b#<J+Y>_
zdD<an)%WMm;X!AhEI76cNj_ZQtEIXHq^Lt1m<;pFp!!Zm-}+`xC7vD1VX9I{IOvsg
zFo$!7*#+&s6kWC~hC1?gy_7tYed~398yZ~!!kXiQd^|OwR2iEF(mPtP&6|s)>K7n8
zhz+ELBvMr>p*&bPjgy9x2x3=QST)R{u;7urOn21VY5e=~n|lD??5y}?^?v%76Dq&<
zmv<O%QQDnbSDqfRTDRN*?Lo~Ctai<3Gk;^3iC!$8ID|FxdwLp6Gzhlp-ZHe0{0=>}
zAhzSoiqgVysBK3>oV0C73Aov4Go`11X*48MBRL#*|NiR@dT(!n5L3YBOq;o$h5eMT
z*138^P8I>OA*_KJK~yVRKimjdjD&2_xiu>1qplOIdIi6!&+Dq|)9<dm^mT{y2Jn~N
zub>bli{?#BgK$(V7@^iS8rt^@Y<pC&W_2^Qv4Zc@DXd>N^;OK&_L@i@nAxWOX8U-~
zHg3lDc$8yi&mKt^Q~?s?waw)rB`rI*V<pn9Zar*c%=Qu9tH)z3muU&3i3d3(<_*Zi
zQQ9K{T(wa)@MI1$T8Iq@ZBzP4_qCS3I3a6Nd^`;jr&DOf@(bv5@r_Tg@A~xd(}&(0
zZ>dbO1Fw`0{xrL^1P6DdRffe~7j}6H`6&1wNkO8A(eGMZz1zBwk}Spc$ge<;Uu91}
z{v+z#4E*so?t!8O;!OY=0&_oFP@?LhLE+fd#l()fWbDw%3cM*xS=FT#tg!#CZ_DT5
zW^aW2`zFZd3o2qeHK&(Ayq?Q+hMK)rXG^t)iR(2@CY?2qRu`!NgL~w8$29be_t5iB
zZE_ZroXL|k1d$L*eWBK_<9dwXtR&vw%X@_u6s!$5KF?(?2jK)NbB-_zWz|`zj0{v8
zE57^?517+7bahcRX*z#Z--v`|olE3u{ph1-JAC)*E?dId*u<ne!a%OB9Ow7Lm#Va&
zeZqCY&{d<Ny!e=KkyDY}dYKb&=_R+Dqvg@|#3kE)v1RI_r}0=}ske*?3dkPV-ma9p
zNn&l^gFhI^G8@lxkc~>wD7QkncdC@j5WQ}QM!IQlWIK0zl%9g%lUgMDV=tp_DcbjF
zm~AZI4E*l-#xU6(aD7W*&(NW<vwStJz{^OPP9o5}R7yJB(kLS%qtnzFpjm7uK6&?Y
z&g4vF3LM+eFpqDD+{YOMuUK-w(zpMK%$25CvkZ!}_M`Y3@U*Fh3RHsPC}EuZlFQS&
zmm@gqCzI>cnOUV91yA;tVs)NNGtd;C`Mu<q?f#Sc?U!%bx<FgXt5+KmYzG5JMVi1K
z0FCsz5)}v&OiRnwhUMhIDws-O^^(aTHstVJT9Gv_U}lajf5IGyiCI_4>wZ)5r^QR5
z?w7i{ge&$+b=mw-84DD$@3?%lP^Its%jTN_C$r6;V|Vzr&hN&-4o9JO;sVdKA}Dz)
z#=ay=M%S}x;mzMJ1W+ktG8YhOY=jsQAj7*zVFF8vg+`?sIsvY(Qg{@B%u_@mlw?Ye
z4^9}}PuZYA0wkm%x6y<$vUG}+iGT^H>Kp{Us#5>}G?3E>1y7SRB^XMl5#|s#YF#?p
zQ_ZK5EIEAiBslr5h%ONI>(|Dk?M<%LKR?EAe+pmgS5W(x2mpR4#zTrIgx@zfcLiTZ
zD4zYU|E_#{Ar{gDf<_+ttsOH_J^31nYj56XY=;X7)HJAe8jwocAbxz{Cm+d@j-Pvo
zNJ>hYIC{ImjryONfI`ZBg)i*l$A^{BlJH-@&UXMMFs;2Ir{5V9gEh=ag!Dm0Le>_U
zHB>@Xjr2qFp|ujeek-3;x$Kduq!bhsnd7i|uSyX@NMJQKXUw^Dg$dwza&r@GxdMUX
zqFW?BrP18p^=)?FROwi#MueP&VWPsdaxachOI--f%!Ss~`o`nR%1Vz+JF~*~I*Zyb
z7;Pzi3EwT{^&R+Iy0RKS)reyP(34AUrpS(FQ>KjKm?vebvJT`~cKnHGh6cTA!uiN_
zC!c2sY+3<AHc<=zWcKO?1=Kf`F_0vP5X)AmDf0a0uBz3Fj{dm~xIV>>$NO?LUgI%}
z5YRuEKoZXu;{b3mp{FUDp14Yx7M^5^O^O<byj&e$WaHB*Ff%uQDy1w0&_1>FoIATu
zsysX4)6v{6pQ{5H2)ggoqBz@9Ut6L6Y-;h)&OHoYHYO5{pRQaBB*FT0`L65@k~+~#
zW2YrYPK0ik>)I;n(-4q#GI?M-+6Z<1`J2J2;QDSa*D$j|fll@D@o`;U-Hiby6TaR{
zkQngAov`3-MsRZmB`uz7Sy(0`$XAgj*ZgV{5Q&vW&`=ViMdNa~k~oP7u^^&Wfc&Vi
z{{y9#E96B-DwL9tOkOVsvD1*USZyO;ZY0V8H6kw?AL;&E-x!vCPiLqy81-*q;PLgt
z+1+15O{-f0$;q1{ri=H0SUtF;6n#0fPp&_i0iZ*fPNsBU@yt`pkxPzM8@q$Fj6;&y
z<2O8e`MM*6rPoA&g`i?PfLru)gbo9MnFQe1H3xyH67fUULzQGo%!L#i?m5z@GnY0J
zQW)7oiNKLn1jz(;IWxTY{IR6WuxT9rOD-s+)9~P*+)nVL;9FzrVPwa<z)Jd0#V0mn
zGAn2Ak0yFs$EAF|?x<FJE?+1US9SLEfF5TKSb_QR1w}GdJ1LW5DnLNGYFf1Z^Ej4v
z``3P1Pqpg=iw$;OkZ>_FF(GCl<ARv}N}V}oNoLB_2TwZiNx?2u7v@V8RytAnShdR~
zTss`&lN8Xgh+Z-4ZfoP_+Btu9eY&x8Zi)m?tgbL4ANXb4JHaK3Z`7B?owB}Nk*pb8
zsW4M!&8Xy0Gcq#bR#uppn4}S33XY$s<Z3VLpZ<Rzl<h`_wt&5|;mR_PB3|u2h5q=m
zxN1a^r%|3R8-TL5R*m7iib04q5%2*Do*8lmJTOwwC#G5_fDn0EKCExfTf}|x5@&XE
zb93S{dD@b6n!ml7#dwcG(|3E=ALIX3i*r}Kk&U&%pu?``y7ll!*+YT=xnRX(D(8z>
zcLF-82R7wkZW{n3-?&pp%d9JG=de;hFhk=_CntMiHVdluT4hLsiRNpL0j&nF9dudD
zmFziZ?EI^i5AE&vUg~hvmoThwwQgbGzrXizI2p>I<7A-R**%J=kv;W;jia@rqqV%F
zaZ`x|b=PClVOH8Kkewn+6Fwl}=21m|sdx`o98%BH9KkM~ssY`*D@QjDYx;N!Au5T2
z{Dj;9bSPgp|2;o6>wQm=4q!4TYsD~E`l6GfmH7)5pR-=B7CcJ0a?k$lZ&D@~O}y1W
z^gl%hI>@OcM2|q;d7-N)M@=(!EKQx0#-5PFtR-8)pSw+_0^J|UK|1&RfrfxvdMmMo
ziQyM^gMw5YM7cjjM5$kg2K3nSfy)l%qLXz_IZB(^2AGf4Y3%BfeK?Exj&3G`?MA0d
zGSXvxWYk_P`C}&JSUjhz{2AmVV*1V5c0lh1D;ZxBEH}R+Wu|u~aE+kjka=aulu^uZ
zu}wX%OYmub?#hIc^+RTx{nn=*Iv29;BC$wfq>lFlOm9V+ql84&nvfnuG+Z)ldMqgz
zjEbXv5q{b2A@Sb-SIG<Hs(ayTdrS4lckxopKTJ}Uy`~&ros>8I{$%;+tDaT(?}#@_
zfg3jiPp%dB$m|NmD?n*M?pIRi|2Ddryl-*k>C%Q|>QTc9zQx+rhS03x?Bqdf)=n4S
z`>@YzYk69{w?9vg<AV_hXS$WF1ur=x4nh<_!LgzdxfSrn!o&`FZu1#e3{Nt8o1LBd
zPzeKwkCe#^Jld>NoPAnO6Fgr3m>^*saP-0f(ZG>#-Uub{DrTe>Ae_X*uTwP4_$svm
zEa5lmVil&dnly)dF*jjOUxD^xekOzf!JPjVS!mM(ttckwl^3*=8x^`3=BVyT3ZdYY
z&r3G@usiRy%{*lC_mI>g6Q$Sq(mQ<Po!=j6s=D>Zt|vI0akjmga9L7>C}*#g*ksGR
zQw153Ev1)X{kjpUMV`N@?=_Q?nl4&Wiq@&Pnl#DNP^`ixV@`zB@f``tbqzvr(_<ru
z0HVh%=kvhOjuE<mh^3>m)*PEEM>U)~Nnwr^&KN`|X`dr!PJ7S(hMX$5e2{-g{gw)n
zT*`;EqR0O%1u7To>Yl!SR6L_OeO*Q0AZ_RO#vhnwz^Xf%U!F2<H(by%g*<2ogmEl<
zM}FXWahSR&MFQg0IGW2pv0bK2Ia*%2LI%2<Z8$gw=?FrXBO+KMs7RykP`B4-TOVIY
z{#$=r@SLXeo$@3SfcM~&a90t>V~AY~Q@}E2Zlo#RU>@rh9Ct)k2ACV_l*p_`KF<tm
z)3->bag}*iQ)XK9*;IwGj)siR`V<Ikm{cEgR3^)fjCeM89XyFBgupY}v)nzjWOSx6
z3LZ|bHJWk*Zlsxj0uC5TTtp3#kmi{8o_UT-4+{U37D<DS$tqb_Za;WLBY$xHesSLX
zkWrr$VEiEWBxakDwWUkFB~tO5NIz6*&xn=yxWtr|Lwtpynx2Rv6HOT_#3vK;_-sS=
zPyg-fho3rU8N!NxPf}Z2@lrNZ7Z}Uxo+`bUV*c4O_1eK4yUi4aI=}J^Sr1YE_4V<J
z;t>rgGXbx;`RMCMDzCgyA5}o}r*w{6diOE{ewFMq;LEh<g`mlalR`O*h*?*%K0+X?
zx(hqzUq!Ad&rBRh%I8dO=jIuMAUYdXb39-tAu#EK|9q44eCWt@F?rF|)>a{Euu-##
z=2F2!s1+a9F;Q@$U^>VmSMk%r<Gq;uh|Z3Vyu{C|lz`yghf)0hohBD%wZ{BK$2;sK
zT-Yff!Ad?RC3Rn3(<a))`&)_$3x7K_qOtb%ZQVL^Z9t!f+ay1K{v2=A<ti(`=kAHv
z;0!v`EmXk!TxXUM4RhI5S6P2y?m=%oo=QC_qtIRbneRV?3OX=M-Vi0HLLl}U>TIQ8
z2qa<IK7*)>kFvVtr}OwTko57zcnLENAIu9EaFzg+tu#hQ0-_-~up{P(w6b{>2*02k
zhYf8hfi~r1LK$(IwkMs;P*Y3GhT~vyi?PedIU_`GU$Z|ROri9>U*Ed+7?Nc}Y44oQ
ze|DvjIgXmQ9epGh+-4;35?{MGUb28rPb<e<r(G*gN}h!oY1(N9?bD2y*9CJ3kaBG^
z%WLeg`)WPwIA8<KgCf8g26_<K@+fPIo?+=`J$czslwcHII-#Y=R4x8c_V&ra90yTr
z#Z-`5Z0A*37f7*t$dv#B{lZ)y2zXhbz1BgKj#5X9w?Ti^+%0p1iCBp%@LOX)Yg@d9
zeof|0?-)Z|qN9DD&dZjsHvBC`Fc28)$n&Lsza>inioiBF^G~+acVJj2Xfdp`L_#O=
zs>}wL{UAUjM1l-7%u%Pn&)ecC?{Syk7U-1s4$4f1o(cHeO!7XNU|(|dQ1GJ~VVMJp
zFBWJwOMj0%f<rt4%t)LpjWd0xbWp(QTHlH_|LD4soGptz3;X9zL4tgKE7n~`Z7-dl
z(ji0Qa3yu}(Pm5OesbVR-+q57k^V$TYR;`r@*{uuSW<x2{^i;LK5Y-!n-FRej|F7F
zF~=_U_UGW&VY_i2y>+R33~~3>$gfeGA@{i+D)`mZxg1*QhB;pOBP{wnOr=sxdk6`^
zJe3nQ`)_~N>+x$ucv5|p{{FUu0b}6Twp%8EB=l;a7jb*6DJa;`s26c23Nm)KQL1W$
zti>~|`{1Zk<2t%CHHPOj4G%YMyuI~v1@pMaTC`}Gd(Y3^TpB7hSfCu>mS|MhLYiV5
z)%)e7LwhMzL$s)Z3>vIKRlpEkYMcon09(?u^Kwi|Fx)#4jPA}5vSdvGTly0MoF+_k
zE2JJ;(o5HsKCZgWShY)x7l5^3gH2pm8UhFL=~M)RG6g3+@>D|QgVZ;N?xd~LZ=Ly8
zJ&hW96o*{e1ggY)gf;%G7*yc`6^mMInwe&$kN?ac{R?YfE~7iIl8)SriebFeXKOoT
z+Ukl&W;$lK)XT`q=(4MP=;}BvI9O_PqvT{&3)0teY+ZD%t6njt<8!)s{S;3uVbnY7
zqV`9uylHn3XBHL{duuN(Qa7dr3@W0DXpfbT(Tn5#+=!VVfFq<!QCgNO0va3sH0hiR
zK3?{#06`j$5Qz@8Hdc1wVRSrKft<owY|@ZY$Tf$J{;vC7LtBQEfV7MJ+x{+sI7}rn
zCL@`%<+F!uefzU$DhWP4QOO8#gtWP<At8+I*hQu&_yxxkt;W3NndRlSf@D7P>Xn@~
zR(Btuc#$@5rB<~bYNa38Dj;{?9$?G*D35urZ8enC_$O}}Grd-cy%$F`Fmeg5(0Qw>
zitb=gy=`sN?^E@NHDn7al;pLX1zcU!2Bi*|LbZWN#=+A)DPF$g`1<piU{8x!T*yIT
zE6&We5G{?Rb4iYZ1$`t{^uPJJ?m63U`n0!lqoZ*kIY$vS_v?G<M&M^u_sPjt$QSbX
zS{1Zd@E?DCW&S!5pYF-}vTXbtt<(ofy<J`5SIZ_;aYP$vmgn&{^<le@8!fE9h5b40
z-n@>W6F5KJm|S_`+bgqyi~rk$d)DOF<4>hBpB5YhGvb*%Iq$}ncG)6|>$(WL_`d1+
zQ2F4?*|(Qa3Vug>br=J8c7KG~$?DbYPREwt0hJmsFYpRSQ(~hnm_bMT<Y2;BDmJ%H
z&(`o|Mm+HSNmGV>!UyUB`Rc!FXEk1%W<m+=qxIRE1d!zPdN7=PrUsMeP|?hj5ZhvK
zYHJe5ur6y+d%ltL?S+)g$`=!p@+FzwqpM&%KspvRwe3}N{ysig<1r$?1~Z|+H>JR#
zV7w7qYq-61MMhBQkFb<KD3H#a=O?(r*RB+5S~3NN>%T7XocrlgJ2M{${}S+N0GWRf
z6sPSyYVBnE>JzpX+<@P?tzo!4(U&KG(Vz%<^2Fxd9+E2J!QuE}_M@(cuSb)ThDKsY
z{990a_3Ze4(amkf!T#flm{82U1|7}x1Ak!o`76a-LK<Wg0B&mcp|t}uacE~@(yF&I
z^TDz^;DMqSI%Y*y#?gKppq|bNy(a$_T<?q7<xGik4U46r0W_SDYG+nqInq~^Cl)Y+
z-s{xx9U`79di7?{E#<HxgnDAU4JD^{JgT{_U+zpZG#j}wWOjh*N(Iw`xoFpz4K)W$
z>F7ptK=0}Wp~92&(1K(~d`|;)wN3%MyT^2uw<3AD2v8cN#u7nN574N2z+I^>U^AM=
zf}yh|!9t1ARrks?t9;A9>H_+s(ZpKr{sd6MX7`d_rIh=Yf9VuHHa*IekvNk>0p<X)
zmb)HodMQRlw3M`~Kat$7nI?(Rmott?4To>_oytTKb|UF?7Zt~dR?Vjll@)pPEa+q)
z4cXJPLQcofk6vLCF;{mZmA$RgDM}&tjou4Ym>T<^9)ABC#rb=&j!)$#-S#%Hsoujw
zl-h*)k#d{sN{c_es|<0MEMBujtxr;)tbg>hS1bfTo3@TiF(_9HQm#Zm(e%KPrrI)6
zHW`aE(Oe&6qmkB5jL##iG!E0tEJILYncEx`iu}5x!#{p(Zf<`6KHfb}eq*g}YYA?=
z<-Yf`Ksl^y>r9BSBnOMbWplyTfS}$aFgXQMGFMyuiCgr`@oy=3zz>Q)r4KPL)FIQE
zD15&Fn%Hb31>58W<=dxL=FPZuH@{5l8AAj-0$>-(<b|U3O@a6YvwG1v3j6AjE&I>>
zqJ75+4+6h*Cq`xfjsA@y;2TF>sv-^IO3PZbkkBz1!YWFQK{mW-!qDp7L`BMzi>uYs
z?kXo0RJ;lOj@hmMjAFCRVKD{1dqLl&{_aC@N;USjXGLg|ma=)QURdmiW=&Qt%Z>EY
z*uyr@q`xMBVcM(2(78uHlDLNTmfJnv)wj<eT`9kH5Tdmn>y#Ua@!Np(Ol|04sS)MY
zYEa_XEVwgOx;j_BTtvLA3<2}=yRu?~Jgw1c-QuIGL`$J=MidiMST@DBokI8J<G|EW
z^Pv54X721^fo5N1(+WQF&6c0<|DerA21`5$_)k)n`NvyECJD!h1rvK@oXc#{hVZr|
z>gdN24u`83j2nvhsm<~uBsVZyj~HsHt!{&0afNY=)D&8hy{u}dft4LjXUXw!f3w4D
zknT<j?kR|56gV~y0_=a%DZeo#i5l5^TSwaqu4>dI%+k=q)3bh5zW-LRCLv{iESSv*
z6+D)E@OH<$y&YVuw#K@$bKchYN}|><I;W(OfVhAF1SdQc?&)q=)+Bsse$p=5Yx{|L
zT;dzKjNzn@^H+>ymoN~$P~*c{0^|blk`i+=5D`LA;y`ps;(aKZpXx(WUXfHhf3$E}
zgDL@}oHXTK*=(;r7J24#kbI?PD;XA-0=9c#=DAjs4U7KL1@&?>M3D2=z#qL_%HN@n
zkuQBt$UT#hIV_OC69?&d>)6L+7z5zAN3F7b<0E0I;{V9U0oc@c1fjfrDAy&IoBFWB
zb#@3Pippv2=Y&I}j2ZIa@FX2Bau8*3439VvuKkGz%0+1d<AOZa@8=+c9UicfJHIh3
zw{vzIeMmziLa|^`oJ!V)k(<Gv3fgtzC`b}JjBB=zP3%|Ct{4&p63h6>D@ld2N^{jV
z7>7GB@bVB6-!6|(%;z%K<o)o;7Nd#8NR+yAAfg7pUD8lpaDhY4Vq%i#1Evs&hWh$=
zc*BMtP07iR29rjXFI|ld&CgadbDNG>+uD}XJS_Y`Qiadm^NmvurT%?*_YMNXU+mXk
ze;bwJo^5fPEmAT&w{u%$=%BNln-8Gyka%eBI{q@bHi<o~NWOS>LZVe?$Q8Eaj*_M2
z`Km{vrr|3%P3wV|&^RPanX?v=NPIcKgX!GVz~ZU*oDv<LHU|f@!>O@zuU00L6fg1B
z$?^Um9jrn%eW1D<y$XHU8BC5Wvtn%A%(h+{$66;yp7<?|@j1&1+>|ihIJrzR8qf56
zTJCL$5Ww!?O&FDxyjC>vrOMu@KcD;*REgDIQdJTQ2q+dEfi%j!k=avac~k%heDj0i
zrFv23J5}-Ms=+q#_)@Z5nqeL1_wQa7at?D%_O{>PdklorC9%vS{mrTcYdR5{0#F#j
zL{g(yDxTXq!W}{>Tk)%+F$>t+KKNe&8#QTye6#}g6X23FmiY+Ns5XZZ-WqvucXszz
z#RQQvZD6Si)O!j-H(rs`00<-t2-Jx2q~=O4S=gKvuix*`O3QJb33tbM5Hi{MNvqR;
zo5P<Io7zeP@|!{LL;~&sAXSG|sb<c6pi+YG5&g_IHqS``WDV_%;*vAnQ|1S5`wdq?
z;5ndWV5(CrtP3H96GX9?iorl+kTw&S>k10GP7D51^34F#PG<(a+~o&0^i*3MSSe;o
z98TysM9EO=>CPL=By(4Y-8=xe8}so9BcxUj2-T-=#t_pHk|(xv4`S0mE!?yO+^#gD
zxgPdrG@{zHE{VD@8_nME9qR;mK3$5&W6RyPPfy?8+dx0IL5Yc(<_{$>bK6o@FCAZD
zx>{xq_aXL=n|W9TjT&lAQeqgYR(`c5C0%6(+gXb9W|Gl}2n+iu$n2cw)X&%BXXOMQ
z|C+FUGB25Y5fKrAASvuOiDH8ZcV7dy816TJGHvX5p0>!@GTfi<ea?gyDrZ^8tuJK5
zT&O#e-uqkrhq?;JFO0#fiSI(rNHU6ao}IHow0JY?Nkr)j=1K6Ax2-!8F=nvRKHxyW
zTh=hoFXW>ZYrBwwSL$=Gr`V743|s8vcxeu=Z@XsFd%1BbZpX3Sn;$`?oaVbhMfR~;
zcq$)8Ufs4n5C<w1=hF{rl{x$LD{YCDb06tYe>IH?Se18Wc<+s$6cfa;W@a?NCTgV-
z{7oc(rsq1u*f{o%pn918805G(?k7`QmtiOsy~%2A{(NB=yN|`~@8jHz6W=v&G7}Nv
za;A`uRGdg#{G&##Zo`<QC?xU1u5U{6S@W30v2dIEBxyo?k1w+5I&o0`X{_E1JwqT`
z3$@+R@hpRr7-TQzr8dg6ufagOrkKe|%jGF@h%HzRi39Z0E`Kz!2O<!$3QOSF=aC+#
zJ#}speMdg#dyX&wCn+dfhJOXq$elB$M)gePIS{hW%H$FquJcfiM}xQo+x}P;_;NaT
zsRjrVIAep``T`&{sW-{+5Yg!^{Kkih#+Lfe$cI^lW~ZM76(jS5Om>6r7q-qI&q@*)
zg>&o6w_RYV$z&O-LMP}UmIY=G(-T@;gwncFNSs+w5DU6epcZ;`G#$_b=ZFruJ`MX4
zaT_0BD3(Z~vwUdChiAnQMpErlBmjW}>9A*{<L0t;QDCeoHxZODnM>2o2}O3-&~DN$
zqMOo%E^QC6wN+DcIPwD}BwEc^zKI<wb@}ikSZT52k8x+68y!M^zhl-v@!1noLAM9c
zdNuw;sH+o$*3xX<qq7RI`L*iY&x&Qp`<(*0gEmxA<GsR2+hlKVZ?F&R>rnt2o;m&=
z`4Bo?>)hB7%J9J8u4zzp4pkMOREEFaX@|=-zso|@hA2B|xW?Z7wUf}tXMb}#1>ac0
z8`GEo!^TwF6f{O3wbBbxQd0hw%sttwS|oK@_HHssep)_m_g{g@FZ0f)rH!rifb8>E
z$v9_$+(_HV68{mfy7ueG(&m<qB9bQdhrMFKy(taaZ|&dG-b4W3$)HL-2!H$ji-gDa
zo2zkvjqpNH!8C9lsRkGn@aBS^ki^4l1j*p&msz7JV+m0K{~$&mpLLNvl|pksNo^?;
zScow#QciN?ehN>#|NFL&jcvvucj{HG>9gq#>dA&{e3}hYyS=5Ug4G>FMwy{sxYwFB
zbDj;eWFBmq50&mSqheSHsPG@qC;029%B-4x8vIl^Eio0bc=*XIDthyGS)<%i%4Kuo
zNM=Mtesb-*pAc@n!d7l}#4kX!KCvn#0Vj(`hXEk~nj0A1502Jj=%pN~0QhV2*S`6q
z+yBf!5l<0oT#at{i6rv(Od8I2XDU&slgUju(L}HFKPWsyWok%PtrEEz8q>jy(nC}B
z8{+6o4{wtI0Lp$wbQ(djwAm54#_eckY#co1YE$8$b;^FTYwAPMtmIB1KB@3@PSP$J
zjC+Nsa<F`5CI@ZmATY;+B#ciGS^zF6@PJf#y_1iLo3l+;bGo-&o{43JvjtZq=RL^G
zM?1ei^c7`8DUt6c(R$832<PfI(@JGFB<IkY{zO4rvqXyKq#qdo0`!32p(2hyPAly<
zYNkr9l5cQj{KOUYg6{9jIeQxK<*EF)hGtg@4_#@r?9%aT{n+BlM3XXVdr+|Kw33&#
zaB*?5zh5=M?YBAI-jc<qTZHFwY0Bw~QvnRt%}~0kMN5J_{7n3l_!$MHPix{|xR<+G
z)^#>4*fkpezOQdnKd`4+K!!Ld21Zb_Uh)lmso06#ojFhki#XooPWxo8_nCagC?v%P
ze_wUOAu<J$+SD%F!WwBkYZxM9jX(4a5p`o9Z$5#+1zcWSbTvwvm>z4C>x+rG2|2UK
zqhdZu>8>ou*v9ALDIBL8-770A_wM={!M#?Vin+T;YT(Xl<v9K5dnLGArjlcLl+ADj
zJ#;x>-<AGn14We$<6d=ZWG$S7nxtRK=#@1iLl+DRlS$Cd7=B3%)04h$_QnoM98J&F
z9nW2gaaeQ)a?V~D+r}VyxXEHE?!H1%D$!2xurz66io5WpmCG4(+XNu74UiveZb<*z
z%R1*Eh>|ll4XOpMU^YyqrSsw<e*TS8AO;S2PwkI<<KdE`%9zAF77r(H&Jv(yDbq;J
zF+Y#O-bDpy&ZE#8f1h4*9O>3Zd8Yxx91<qFM5|lB4o2!LaS}f##sSS`5)x3u7I`(u
zGR6!V06-`{@}y1(FdZ*z10XKxZGMK<fAY1X+F3_~i+gTK%yY6wmX?Vdjs^-tVBa{I
z-`Vx2$z~UjT-b%~_g?mm`M!1h7PlF4AmN&F{m^n&!LiU5aNHoRv!{_6Pwr@w4B+VJ
zpyVE7B#~}=`4YgXlhG0Q;Rni{f$i{0_PNy}77r-A7y>YhK4ZazJ)b!R-EVkri^NS4
zfaY?-QC>g#ZVl^=I&?>kEFs(R{266DMDZDUy{!SCrEmZaaxQWp-54Ko6KPt3JfYVG
z07MK8+*Hlwq#IfkbVOPem`Gftlo7LyW3hkXFn`LOdH&fvt49b+RLBq3=A#vV<~QeM
z-_?QE_-FXW{|-&sny9BHwBNYWG)QWuRG0k9p4eV}^V<&}!|-i@P2vpRy!q8D=LpS(
z)q9ZFfWG+QFIj40-_T$qjt{x=k^=L83L~=-i!1=S1ET^1dq+P+yd_exb<@&iudrPL
z^7Pr{boSe~boVw;6E)W=0srMxfnXf7VTk3#L07^dv$r;w>H&EZ9mn3XIRoJx;m4GW
zaYP=BY=D}&E+kt^eZAm`3?jQynlu5^HL+1C{qWfn{}z~CbvaW%y^XYxD~u;G9{7&f
zFo~Bz9{HU0A04e@LZPi~dRj_)MOkWst>1V#P=K9yqSFn<6>OVs$o(A6d<=5awzG)`
ze5n_?V22qba%6DeSM2{bm4uF6*|@~>m!I|WI7C@BjZvIQUp9!~Y1nD4_?N_6PPj#J
zB*(_oMxOhBUV`}2RL*FS8X~Ss6AahPQ!v(f82OEmtp}g^Np2*M2jv`NWQ@)2kp4%5
zs0D>!0z%IR26U?d>!^_gn!Kv7{e<ywY_Jbb$57C5k-nU>a)qspa<G7s?J*AUKB+qh
zz|$v%)CP!G&F6e`SxrNis?OC8tPukY;`-dpzmpLVlzIFGrnQ~RmpI5Lt5fWb|F)P5
z+VWL?+w)!2S5)YtXqMcEMVk9<tdo5MsuQaxl%aFNTN-o$TPF@gS~!38UcSE|ZhWU`
zwh?}GAq*}1ZAakTqJGoskm5qauk|FOku0Y4w<go)z;+a2&Ri0#l(?R7dD@x|PW+5R
zNJuE({8(5}^|bf6P{BWs$JEX-%@7K86oEt&r}jW3uz*#3gqW5_a+nkj25_OsfVnYf
zQ#1r%lfOKdWm=8iaI+sSxoDAoROj)0XF_-u|3SjbaR5l+;l+gnV99-J<@}Xf!$!L+
zQ=5n~GXaee=V*{;FI&pfnXbJ(9uM{D8pJhX_+!*xd=E&`fw{T)1BU8MYrVjaU0rL(
zSJ~OwRg1O3?w0;t-#<3GzD4~8K|gL4Blg@R@G~UV44>X0=d^RrRB;p`6r7cqY{A>=
z@G81FjoHoR;7b7E8ijbpd-*l9v5<)Z{a4jH7DWpFF3!%+c(}6S%5(3h++8ZJxCNNh
ziWlnGy3q5#v65$|^xGjEe5h4U@i_xeI(n9VoziTUi7;6%*(%nmM#VcdI3How;De5v
z3Qt&ckMyqZ5oKuz<0Ta$$o`+zY3z(6@JoZzlcKl)l~2ki&5eda_>;AbZGm=m)xgVl
zlm*<0;snO+zBr^UIxVdpjC%rwaTovQnoa5%3sFCnEViF{;#ugAGd=FP%H=B?e`RL!
zrd@RDzXo5s#LsW{P+4iMQ={?y3zMxv4+DZUx5EN{{2Fhe9DG^tKT`rv#KW68$OwoC
z6pn*~^zCvx8;U{=nM_AbUB^KLc#G?YVF5A<C?%q1;!7a`nAEle!c0hphG_sa2$#Ym
z^1>ayRw=4nZ$4p?mq?rVPukLKDM58dpDd5pTs>RYiT((LI)==!luh0GRNVpno3?(P
zB;}oy{q=3FI=Jh#0y_C^*_>kj^PpsrfbJGNw#8cljxG0l#d;Ej72f`~%DzGuO#1Hl
zG|zWWnRzBx9hVZ#ndgkSg)Hob-exfxpM}f$HaLcRtA5(E=x%H7{zVh;a8y37)>yNw
z1>j%mhnJ;L&SP-Q)eIQ{8-z5y<QbzESJ?^FzkpU@<)lc|W#J}dzlv6=<Lbst4&R~A
zi9e<Xf>L6mH*dds$+4y^IXZjL(Y&#eV5gRqkr9mee5>(=Bwcl)1*4|KPTg*0WF6@5
zp4Z7vN!oz`FUG*Yr;SORl4qN}Au{<5jg3EV?{rW;Pnj4j$$lhL9T4M3RsDODL2sc(
zH~-qGH$zTb%AQ9KZfk1c++o@FnX5)^q8MsTryT(KN2~%c2?wry#RrH6Q9n>HLM*SX
ztzBHSxW{}oDl(L>%`^LAY&vLGS5Sk0eo6ZL{$&}gwJR3r@M3d7HbDJr@-C?==2l_D
z7Su;8jCV((Dwv;`8x~e}1Zq({zP`1$<Sg6I7BQ&oE_?o+oXbG_x${=7Xtw8$Ac)v)
zF_?TyCTJ`N`k)?++%F9LKc?O~s>%0{{~t(9Kp2Pu3X%#6h;(;@q|zM{($WmLK{}*E
zT4_n?22r{jVK7QM1`HTA_S^gO`~J@FyR$!bcJ6cU>~&q&>%Oj6Jf6x7{8#!AkaBix
zuvgwk*>@7TvowMf77S7ERK7PhG!(KE1bdQ_vrcC;Qd53ga3sPVv2a4yH}nY;GgvzU
zmI5q)qyc+33wMWP(AP<k6490uJFzG69^QX_sJArd>vpZ(1MXQ?l=EOIxw7Ez9K*XF
zJT&p;pVDMFU+Q>15>Dji&s@Cuy1h>O6I33@!yCf#Ac_&!j!03EJ@b)Nf~L;P(8j;I
zZMd~B)G67zb(yiE#E6c%*E5WLrBMcWKqBS4NJCrI&jeh9voI?39=#*x%9yK_GjX2j
zXw*hF)4?ID@>^L8O$0#{rGcOv>HSox4~rM^)+9khD$EdvW2Pm(`Ng9z{0fMq8e8#X
zn1tH}!rPH=oc2S%#m@x_CFI$WF;T0=dd61*DOPDvCZXwKxdSP_8pr6h`i}M^fM&=5
zJyMOfp*TX-iMM&vSS=#Ioye2m1U;9RftcgBXsd|uVC%j29%~BSpRjA^f97-uYnuPc
zTW;Re{HE|It*C|M9Fau*aJZtPs^yxykB`4U^P}|C%L$oh9WE#Em$PJx^RmMJGa}?n
zD%mNf;S7s7vVW%HbZ|j(YQt2cz_lgveR)Fonywh$A`g>Vh8LzD-L=f=fORtDc2i?Y
zPkPzEm7RVB2Nk2mOYlH_nbc1$H9it`B}VQ+p1^~#+Oa(e6f~b-h=oumGfoAs>w=_S
z%1&(^dA`kDe`%#p4QksAj)X|VIPK3}Ea%dC9|$l1h2SRE%!N}|a&z+o-gL4vnREfF
z`<f2yhbj5r!!ZT=!cx<z%HH*dhgOkP>B5wW9|e-rOkY;kxioA`*YkrZAQbtG^2YU!
zY~N~jI_e%jWKkIz_?@I})W8<WCWjTq3HRzV%7W(E`sNDO+b|V~|H;Zu0x{lqe;aF^
zU4uPX>3Z6e$h(B~>dW>47pIOw2V9Ql9d-P(OeDV}_4Q`2uNN2h43krM2yiUHb%P%D
z{Dwv*m3<o`@71{FVs&&zhdzm9%;@~6Ppf^E9NA#gXGub1mGC@FU(a)MXvQN5e~Jfp
zKbDu^i_LTCLH=LwCi<B3Npl*yos-Aky?o0gn_VxAM<Zv;$rRe^4Ez>X|6XNP+VDxH
zZYiB2W6@R;bw!1qSlSY38mD@w%-Fd5hu7mNysyJTiit@UwD_G6q?{U_N@_j`{=CWZ
zl_GM=9*px;-j7F)IHcQEh$GsvS;(07^=l4}QuUGsT`L)uAs#|byA;dy(&n@`CDgL(
zf9WY>{)GJ^O6?bs+XmuXoMOQ&Gn>{tR^1_lWOF*cr~d{Y_Q!t_XcBskXm_m`iZ1^s
z@<dI+$hf7o>V4^p#<)P9XY)E4mXxb(xRKd!O=hv&pW?boj%j%+oIMDnX2AWIjkkR9
z4<b9y*s$mJ4v%@1{;qcZ74n_o)>3hPXC6E7l12s>tV2^4o>2zCaLKh8&sF8<U%!S5
zG_{N#A_pK}xMAK+t<rOTW*|rc<-2ID9u`on6t}R$_XU`PB6*yl+f0)8_b9`s`4P?s
zDH2cjzQN=+dq3uekT3=xF=PlMGoB|asnX#|MY3^S%jdrAex3RLA<j2`YZBk>PkC<Z
zEHdCtM}hLahpcf}?bJr?w<%f_*3wcxC5#s8P_-d9`(+jTwPkx)U9#jsI|qk9ZU3d7
zKCJU)jOu?Kc*4`cWTtPdPy2ey_o&@{uAdsOC*gy=pCu08mCiDapuIs<<mZ_;YWPnI
zi=WbN@nyu7t1O%(Y}2??B$QPNDR8DHq@;u|)O+^OK?cGA$D^HX?EgM|V|D0d-pNis
zY4=1rw2+f{UIXh@Wn^S*?t4bQ#I}&LeP4S@%trUETx?>-7R=Ze1YV4T)Z=!_8ygv>
zFyPW`4DgwRC%q73dokBu{A~0|vad^21*1zPh(gL_)W&ns*sQu-`yCZ`6t2gH!K1`9
z)Io+?|EwdBIr~9H98sy1nL16Kqm(ppDk(&>UCU%dO`7J-iMs9(8}ArjOd<H_mf|Jl
zSWT}FYrG%Bf+sEW5vD6m$iS|sNC_{5#J;i3V3kwzng8{YiC-r7(3<Uq$9)0sGxsdm
zk{30_N0Dfl`8#0wL)X9)RcDwa4#-2T!z$Xi1_TD~S=_Uvt_@XCCWz)JpTS}L6rG)4
z*9Ua~TUFq2P~b{$ecFGklJHeNXuw*+)`p?fYLtPSoA@Wo7q9(SLi>arIE?p+qfaYX
z!Zb&R792l=#$XVcNBJt^{_8t6!|`-L4}bjEw<WWV9Nd(?Hncd(;OFitBaC>`<~F3M
z*z^J*z*U>C_pf8g)|A_~HhiGIr8|7PSWJA5)bMxsfZblbvN&JrKc#^v*;4(#H05XX
zaW9MBN3>QEeh_B2{h*_(qoe%xWe>Ggsf^vLXItdx|JJR>q&94~{$|w6ucrRde56VD
z1F5G_Z+ovdoHRGM&ZX=#JEX*wJc?<)JPJ>MC+crG>@+`cKk9OJpz1MXVKV>O7ln@}
z`wyEws1<Y1WoFLzevLSXhJ+=TJ7g6oKN9Bo#t0%ILD7j{+>r(TWLrwrGj7zhPAJV*
z+t+0L>|XWCp-A{SInJcSPsSOnv)ZE(!F2-Jr3MXD;2TFZtx)nm((1_$4UYWw1x}Lz
zge5Aqc7`|=Bo!MyQe2zSJ^LxF7$WtWo91R$g5<yPdzLvTUqFQxcV-8#9G!5Ta4ZSb
zYxM^ZR4T|h@s(dG%$&<k#_^I;!ixVH(&Yn>q!BqNxG=qp9Md*dQ-+GiAZ^WOD07EJ
zhQ-FzY^8bMiXKHu5Yp~#2iT0V^a0cb*JgGa7FJ4*axHHs_<f?e4OGaJaYEE)s=3A7
zSwoAKdd8j#l4Lo*bZXGg#1BfA5&xb29-@eGMW&}%?gcW7#s6;s1&GXr?QORnyiO}A
zo}z2QOnbHM7`*2tC9f<xj5DpdDg1257Fk+^@DpC{HsW$gje<QVd6L@+s5}>qh%_W%
zvj*`Q6Y#0wh>+{dp7oq;;GOg9nKeLtcL+NKd=n^m1x(t$GftVy<}I7<W7}XUxtZ+v
zEzA9Co%*L^2`FfK)MbCbo-yKnmhQoWXp4nh)uVCwlgTYfJn8mRWDf44uC#|%x8eew
zu_je7Xk5O!fIMBKZXX7UbN6?$R_6^_+U=q>TLg;_+ZWHDO%8?kye$6L!5<D9G%T0M
zEd7x*&1DE1$l<BuVb0LW)b9<xAC-c9D2t66^+kq;vo*9*g$gT)Kl!4bC(+ST)08JX
zn@IY7u2y=f*ctC{{q(6p@mozg9GxGbg}FHcLH19}aX|HDuU<X}?5R{u7(K*l$eFF}
zMTgx*?_h%B(&a5OZyDX#afguB2mhPVvwr9>%yC|4fxMm9FP=&!I1NLf?Z3FgCqa<|
zh7Gd=<qV#$)MqLP`}i;@_#^q6_V)Lb;to15AHOFZF)XOn*jvGo3M<i$MpHkX6929|
z7_Qm#l_Ex}(E}KGeUPO#OwG*jSbkdM)<RFxOIHUMJoC7grr=Z3+%M)TzsukNXgKjb
z^&hW>7d%9=Qt1|}By^;*>#z+w|DEs@bVhLocTC8oSc&i3hHETR;Hm$8Jo<{1>p^cR
z=Y6LZz9lKak-dMij-oI*b&g#Ui+6Ih>+6+4<8`9g*$rUVYm0cw!Ia3@h|_<Avm{~b
z^#8i#(L+XOcST9;Gs(%B%q>gw`o9nE&`8+Q<oN$Cm0kKj1G@BV{r@<-tqVv~OF<s&
zKchu7kJA0mAGC<~{9m^u6IbMVIkT4J5%}Lv1K0=N?-5ng(<IEY)xge?5!_ww;)wnz
zMJc%<zI{K<>S&^~8$T_XIH0D6uM-m|s<B|wM&!Or^>78fVRpQe9_|v*osMYjX>5U2
zh@fj=M?z1ibi)^}UJE-e1>BfqEmRfVzH9zAw$~63?EF!jc12BW^p1PRrLrDf{>otm
zteX|P?hGuvySB|F9JbI~^le*v=`(ro_B<M+rt=f;t{YkUuI+(DX2m$#78G9vhx4)*
zl2F2L9Q)0Yksy?eqcw)s!4Glcym)d|=e+%{qisCT7gK+*Vhh)Ba$Z7B#ybKeq#P#J
zmy)n;lRFgW(pf<~N9V1c1U8M^50EWg<Yuj)PG$`Y{WgeoPi@ghi3RkX=LIF@7t`m9
z%k<5gvShKo+f?I?yPRETsZc+5Xu4lIfTV{9bZR(aj!#6LeNlu*3t*<*9<^%d=nA}b
za2mSey#bqrpX9bFwj*b@pX6OfAb_)PO|?MFUvyy}bY5T1+OOY5Y#&(V1@_ruX6Vr!
zE`EQ1AN?T=|Fa*sqvkNm=(ACk1wR>sX}ri-Ms+@Q@kewd)1}V_<Y?UeAu|><7IMAH
z?Y=ANS`Vp1X74A{anku7BCB>Dpy>67P?GzC2ryMn(s+V6<yOG<-@xNTNx#E&Y^5*>
zK^QaYZ+iQ7UMK3?I6#Ok#r%1j%7sxXe-zm5TYcx-zpW_hvVH3<nCIQ#R1ITo0#LqP
z7rW!_e*IRi<#Fikqjt&CIEiIs#EM6$zJz~ouH|tUSKxf#Q?s9zWMa2v%f#YYlBZp_
z%gS5Rn6uygl?N?*nX)lAUDulk`kai*(>t;ypN+$k#g}UhfF$&&Dgbj;J-<(exqZ;v
zAn7|ChTx||$6)q{?xr^5#p-{aT~aMK&(ZQOp)*|qmOST8Py+$@i(Eb@(EXf}+#L75
z<Z54Cw*_zXhOWp(rtNw9-LzQ;A{Dbn-X<n<QIBy;@kNj4w(XxHZimT$<E^HSb56nK
z26&*;rwxF(`hw8x_?oQqdY2f@h1qmgUyo=4Oa$)!J$+n-KFLxwy@_#IxY^0`nMNE4
z&9A(bxSE_wG{Z#08zfJ1%lx*9>2rbIVTmX7p$5=52aInJRMbWi=x{{q-h?3fB$2uL
zx-24KJ8$K#lWYkauM@oq+Y(=HLL+UL$28in0F5!gI}e2nytp}4KH%j=po);_OJOkt
zSy{n~AB^j6Wd(E76tVC{^6dD4*zEdFu-&}3!heHa9P;ckZ>4z(x>tVGx$!}Rt?DjV
z(Es2eBH(KJI%fQ88gORCj!`u0&20;&+>s1RkbGt&w)TPki+1PQRh!?+t?dSJZUAt*
zJ-)kXc0$}#RoS_r*M;cnIEn*c&-i08=lxZ~X}K3e@w)U2zI&2JqWiHtoi`>ff%D5&
zYlML}H0b4(J5*K{>$A%qGyj|5TYJm~TR+V7IxNs)d~KL(1@_b!#q6?tpgzuecWZQk
zs%qb>J<yW89dkPI??CTsccE<Q0bu8)<Vov1X3X2z`_dP18}Ikdb?~bxh>CoR@I#h&
zZnzQVoy8zHH|u)FBoT<4o0yTCyjy_mt{u2ElN|M>N_6k{LXNI@&*oNRLT=ujry)dB
z&QZo*3!fTK3(cB8%mr-V@-`rmg3Ik2UY%>Kz!|PL>PiA(+r^dFBLVOtUvJ-N{d4z*
z)6ZBRtj6TtS>Nts$@Qz#0Q%v}UN+DLw;6f{eer8L?@Fj;KQ2CCeM9m0o>AaQo9Qt#
zkmuZNL8f8I=`V72E7`X3r`o^K1*d4&nrKWI?+iZZsB5ChFNyhGBI$W$B6-yuGT(9B
zcV|TpwS&>A!Oy1<0Qw8G>!1s>g_59%%l6zm3C^(%y-PK_o;CAZSd_fo5$xExD7jmT
zm$VXEf*wyC1y2?qa=*~OY}EsfA2wm&b#Zq`E*<t$fe%oDNC4CAfarprRZyXuSe>Ua
z-mosxDM|kicWd)qTTOSXl_g}frbjuRRqK|D_dprq$K5~{LX6+2ZTs$gy!J6JRTe59
z9x%7+(lB^R=mO`0h5HApAW2=v{+=dE-i1kybs~GmMdf^Y1u;|f9lL1M60$Sh1Oqqo
zoi!S|V5j=}BPS2oT&Z^M%Gupp{CE3*8JhRM+r)p&D@Wd$i(H?btXu`t-y%<dccy1I
z)ptjV{JW10j7+Fj&u2YpUcWwqcdS?60H!x-*u{DdTt=s(kEc-{(;um@V)R1+=nGrJ
zg*xIn^=j@J2Zv<K@!Y#G`k+7KH`7Ll%k=rXk$IH!T?69oq-*gbAlb3I-G!P*w5*i?
z$Q$|(!Ma|#*EEFaF<#t(ML^W-dW6fjyG!B(g*mM5+RmgS!<#tjV0T8O>XF!;OnkQ}
z8~TfWxPppnT1j<1g*Pp?yM4cZA9FpFcYQN#daL2$w*kYf3iksWwr0?OXZT~D@Pfg?
zEG*>!fMTe+j^p8H`{T_BEQlJ`N0l3jx#u<MYql70vFaVa93lW9fR;I|0=fWVo5tv_
z$E@RMi#-=vX_gWk%Wm9qI}y!&Q!SRQp;0~7I-Lg;W4ZYc2w*O@EBtyD#;ZpX<J;(W
zHRC<)_kJx%HeLveAjm$-N#?xmB@6c%hpk*i^>Bb7E;omm>vgk$6SA)B#)yE?dGRf7
zGi1fE^HomzEigxqzHJiU9h5w7Drs{W+RFk?+AzcAF81h>mQ$MX&g+N3^$KvkZFYGo
zxUB)bNWVtrLO_>y^ZtLnJ9KT+Z|-qn)`w&6s!T>%H!8U>on~zTyX<+`>F#pd2DEyj
zCq*>^_6~S|U#dx-I7zxs=i2%oZhB)z%K_vA`uhNET;?Y4;$R<fy-W|C<hmP(F>CYt
zGu`R<z2kVg8uKH)6FEeG7lw(gz8OsKXbVMV`u=T*A96--<e@k9+Rx_UKA#W`dj6~O
z+v4+KaW+TjwmiQ(BygM_xY=9n2lb!XsNx(0GBNMGYJ7nMtMMndM_FX+?c3L<{zJ$>
z;OMXOvd2bcjK*!Wq7jJA<*s2{{Km-ia=7aZbGP(#b<hm;DCRKC%p7a`yQiJ#7MRQS
zelE&$yQ^*d#<$DkhGbL}YkUE!yB4hVA9tLpx2_4B-K21}EC${#H!XFYj@U|m@{g|a
zU0-Xky6IX!nI%O0G&^6{!;I9C{R1NM&Sv#;{rgus&O3)Y&-F~NWaImx&MnE#K4W!V
z<`%cXUAG}kL+OFAcU2RW)Q6a)60wih-=D4s-hD<~otxcVltZZuZ`^j>Gy-FnPUZwV
zH|O@Q5!ah}=*AA#6)5ZSaklNhrp}Y9di@o2ZkDqjD&p|L$~7nEga)RL2A;<x=h(hQ
zbt2CNoz9SxtViw0F;PjIXE#gLe$C0FZSVj+L=^Vg_&r46t)6%|qWPqFBRyc4t7Dze
zMSo`a9~84CXzG0%C3$h0cio5fUb!6GNEg5Pq4&sEM4{2FYioi^5;+)yLc?HicMrt;
z?O8dncO8%35Z=CKQ3U?L3-V42H*fCFoe(Bn>%x+#r<ikgT?th~G#q{8oofrHKM(!0
z&nmIdz9NMoMRp>lvrz8+j}dK~qfHNTuWzfn0?FOhuB-iJE^55nHyY2H5x%|E0jTo_
z?#G{bJ?FORp|cMgSH}4Da#2TH$SSWb&$iWiAbg7sG&M#K%u0A~2f%r~{<rsjwtg{D
z9-H&O?Q_fF&;NfXEiUwJWgPvx+_`N-U1spqQG_mvA+e?}nVTs|+tLPvt)G!nAi$#t
z#`~?Hg!iqtJ3+Iikd#{X7fVni81(D<DF`z0=RN~2GYFT7fiXYOsPHN&2owz7j|T`Q
zsR@I^IY2x?U|tNC9-x=-Vb|?NT1J=f*0qt&@+=&6x$8!4ZnGLL3jx(INY}BX<0O?+
zX?>q{4wd23K1haB`lWnck8U%Jvca$V_RZ)%gBSspr%Ww5EnaNWAo!tLd`mFHPiXdm
ztMXBjrKBU?O^jb@Yt>Q5b~YCh1Dp=u7Q6uwLSnfoLO~5+B6ghBorIq`FxeXdGFo_L
z^CGQ_)98@kN}C90N6@6mLmW{4Az74K1TKhi;zE?bPC~XIjS=2Squ=by17g!wvA1%9
ze{`yFv8!~ltQ)~<@%gelH6bylf|xqdPcujBuaDLd_m4;>^ZZa}s4fhkl4<B<&XWNl
zPplEpDP*pD69EA+F?|NV!AW2seuGo+Ez<EF1Bi(T7u$(j{W727Ggje?LBJzwO0zSP
zFFVR-U!4gLPYMDmd_nt0MYat5iyJrLXAp$=4X-(+1w;-vP6iapu_~nmmM-KVc9;(2
z;_Uk)pw2h$Y}^%SrJVy9(U>$hZtAlND87dw-s(55(BV+r?nvHa!d<#+R-?5fhd><j
z@sj5H3?CNMi{{jF?6+b`HwMMt92W#3-ABu+-v1Jodrg*lSLl~$@>$6%2Ml{@Q)I}c
z&DWHM60~zOAUI%h2p;8-S0iyG-meFQ0$j5R^h`|XoyWfza2WulQnwrOd?`HX(8v^!
z)kChJ6m3ut-h>><$D{zuxRHuTCfx6F+_*tX54ou!0kKL1FPZQtf7yX(-hd*Rm_b1m
zDNnu#^5~4kXm#-#Qy+gh&+sXFR0{@0@?mr!<XWIm8s}=oOr<8ebO=Zv<QygdqLkxP
zxCe4~^_)Gd7X;cLMl!mZ%qN=UVpz9o+G5(8V=({RY$HvDp2$~nz!n!7$mMV?D0BxR
z@rdqghJx@C@#OHr$b&)p;QJ?D@vq@yp1>I~!|n@3-3f>wpKm(|hpopbm4Fh&s69Fz
zWoPuKJKQ2F<551_bKK~lj0|}!VyDpxpKX={CsPxb4KV!5pj(P>8OtrQJ!JWgJ=LJi
zK@so_sOA1d@|l5YFD>ytxu!lOBY~Q_QJzIgW+4zS3T*jMy3s0!XLNLgqO4V4pZb-I
zUOGb=7T=4zZjwHO9Sf7&L)Vhn78!7Kf-nzp{)CeV0b@Sf+kBAzSDdd9ys#k9OI%*?
za{jpjP~t2l_rOY)d7YVAnHi^s!CZFm^8+;jzD1T{x%x^Zn;ibYXA%a({TCfn7UcIC
z<?PuEiA6cTmwhZxu{D=sz{9crQmQl?X`Hs{sEgxOuda19vnTcvh1g~1H-o(j!wJPv
z0$nP06GFoNDYl9YG=|6HvN7O>&MYa*5m6E;0q^)CxohVYgsFlip@{686T~)lg<#(<
zpG9TQMzTK$GxQ8JR~qnTyn0n6PH5*$z)5WY3BEr8$=Ci6I{^DI2s3DpT(b0)lY5LS
z#YEr%0+&+KkW)aIa47LUD+sjUOG#muPZ1~)UyeGeu_3U#Sf7}?R;i#2xXtM2I0dfi
zwyT%^W{ZMDnN&1|gevbpdYU|={EzU}|5r?)9<Aba7G9GZrdgCJO2dV4oMjx_(mDK^
zp$|!+fZ#2qxr4H%-hla{Dd6k3hrdtbe_YTm_!8p&=E)mPW^wa;kZ)<O2W{6G!IlnS
z-<MC<NtFc|lr!-V85tSM067Sp1i~T5VI&8CmvZD_>rwbqRU~cQ=}`dDa%E5L8l7o+
zr$^49#@)w8E*+ep{kZ#8!B=J}Jj&k+zsOmndZ-JuRpjbr5|!jYL{d%>5SE}VQ)hqa
zu=42<IMf59ZItV8jwpA@y;_WM31=WtVIi97V*yRti9Q)utQ)fgy14XPK8kR?=+K1~
zDj%-CaE;Y@RN(GdXAHu@DKfCVz4>AfoQxMI#Hu+yBwvX0f3M@*A4j(_$?_(YVJz&j
z3Gt-szyH;>-OMXjouV>j?at@&_A{>3mw&jRAO^WYEs#7VJAs^Qo7aWp*=4}fg269e
zbf_Y~`HC?!Lb|<xvNkl+=L)`HHYR+TXuen4a#ABqBP1laZZMZ0$_^~J<nR4NOB7ge
zCafWeEhh#%pf`1qg@yGq+;XyRrd7+(!Y=-0)X)!Aoj0je7MA3dI$*}jpN;&>bq)zn
z_WMX>C^sLs%2x6uDS--<Ij8(_%?YIKqA8A<KVfQIw&wZ(uBLP@5X7pzgU@a3x$@^o
z3L(4;iEvl%te*dk7HmzOT-?6M-(-^0MTm%j7l&D8qezY%e+J^5zkA%BBbugdW%+S}
z5X2yj$B^jM5t8^s%?JmFy@+4+JzF1>*4g*3YOD_X^RgaesgWuyOyszU0*AVS%1CG1
zZ>Q=>)&S#Y-Mz+sEeB5WOz+=49QjO%Jy^O4AsM}4ms?}maQ`#7Ag9eX*Q3lNh*JJ^
zG!r<fwMC+<QvlXx4pL(JNvyJzOBd3b%^c+39WgLL>Dx4#0_&Z3dPS88Zi0_+nkSF)
z5^fDZ<fRhtGrn;zpM6~%2JZU>zX33?{{{}g&PcNLF-SmpDmy+y{u9q!e61&WBb7{*
z3E<BRLAZ8J$N=c$cP3BEDy2D|unR-T862N#IWKUKD<?h%D-hDZNFK$$9_Y=(q4EP1
zI7KhJi9#ZOhOF9#-2L_wpB35lxQU5rOI81hp9qO!X8x|^$dM<<1_>~UwcYncd<P0s
zUB1i__#%ENBD&nrSCfS!uzO!9ahXfOLe=X~+p6;^bY)?0mF|-OaXON=KUFvfBU=r~
z1N?yYoL|H7sWvX`0zUhw!+eDT_)ph&%XUKqa_@CFr`FedcZ-UJ_Lp)PuYqfgyDV({
zOe2PARMb9AqrbeMa#nv{*>byR7&tX17c%3Cz=~>2e8rQ(nO>iYbC|DM^2P-G=`AX&
zfI~58MS5R8W$iTT3^V9?L;o|VnA*EqKY^T%tLq=EdHxiX*E_)Fl<K$`(t@(pW7o1v
z(&rqyk~4O7c`sfPIcb}QA|cN&|A-gI{R*qyQLoEH(Q(fvjoMh8#$`+acxvQO4zWx+
z*O}WR75UTq{6l38jqA_i92;>aieow2-j>$B(v%q;ZXUykk5Bens)v6`aHsXsWJR)Y
zkoGdVf)kY(Ulmc(P}tzPc~;&h*LwNLAR%QcGee6d6s)wUZ{;}3!7US8#UVsM`aH<4
z5)uUpB`FJK_{Nju^^*S8G;khtxF%z{$Q6J{bqP2uAHVB~FuOf!vZ>qs&fJgtM=|ee
zrzQ1~)yVgIgfEN==HUVR1>^p5A%Pp`^L4s@I|Gab=l*<G*G;*=QHzEcrorXYyx7f2
zeAnHc!;c2kvNw>=6PLRX&zeltg*n}zzx?%W1w)Uy!B|RgDHL#80dLD^FW^|+c~nUE
zgm1|Jt(eC+jCJL}(-3*pL*4MxN-wFHO7}IR6u8@t7AxhxCgW|<?1$d7vl7=028E!|
zy}g9=(6f$r>~Nj__@j<b_C>imzw!?nTY@4}v_rECf6Qh6H0b@@P(1tWGE!9)T9*H}
zwVmhST<v4viD2L%a6%@j_Vu<mWrr&$jHZC1*>BCVX~pGSBs!_<s2irY+E`<xpAWuA
zfERe!AXu`o9wF&-RYdPEc`YNPUQ%ZYfbzdKb>+8uG2$UQWH_nC2(1olpz+JiUQX`r
z9N@gm9~!Ae(7Syz9R?P#!YtD_xn3o?QFqx$h8N07rq$lS!QUFxisa$P6*>}-n|uBF
zl16|fPe!)?xlW~$gE*H>XwQK`LOq>YaU1HHP%~!r3c3Bds=bTvpk~{-1lStd<Wo!Y
z;y8kPx}xCsT-r|!O&(YB$!6iD2Hdg+dZFSryu){<`N+5W4Dul~AKC_$3*Ej#aKAGU
z<J2Ah{+0)PbnPwts@@=Dr533YXr1R$_=tstsVK4n+wYds23ss0)PHUEs>&p%&|X{b
zwN_&p-FGGyh_tp8WhisMbJKGfPyRNq!>OsUdzHGP*I<s{OFq;%atxnhqc)4eyh>ck
zcG6KtT|R5;yVJJm-gLe@!DzkClSAU|v6oJuB$h+qL0Vd!u7l|qm#Go(?=RDhiskc~
zvh9KyOTlT?KVDklJmv-wi@c7|;D3#p_nH=qL91Sw8Z)<2k+jbhNF%{)N^i2%2+%m>
z+%BSSM+1yEO~56v?2Wl+I`G05u(<!d)8{DX<hqRC%T&~cr1J?0%r_T%!h<R&rKB2B
zAK?O8o`+<#e^rOR#^~iBfzy6qs3Mu=3xy1L<!UTO&^hh`u|oEk{0wn-N9817{`~WC
z7y5{5`oc2*i~A8-*JTOSN`U14f#2DWJWiQbiBJ3G^ARDY;HoMjS-!0Cd@^NAmdjNW
zoyWhN54Re=H0*CQJg{uA#ngl?iz9Zot`&etOGld|F}1g8b1Kr2?s#5qy6K)ahLl@R
z9oTdD*cAnbPYiY5(g?eFgH$Y4k^>E#*~)~4!bIfMr{6lrW}JIIpkO4xX)H7b<EpSm
z)kK9-%N{d+OlGlv0eY5-Te{`0Hx~K5O5sbGp1o5&DLGQzOomQ6%!4hbR~_N{Gb$F?
z3+v8x3Aovmya76ExSs@UIGDkg452%Tths($1yt1)cQDM223Bd526581;+K29P=%Z)
z`3hWB->uRI-068hnRV~}vAc+2rpU;i1@3H}pU?-asrGqbj)~1K!^|%6p(g}>4=A*5
z$;<-Ip^`qn{qT+Xd0Ro3rT20azJOIb_YT?C^h>5sk>!Fe6qa!3uIp1(R&_Cd_jA9@
z&9Pa7%v^~81ty2e#NJsqfjlUAGD~KPE64B;T$!K~JM^qxC#L5iy^(=1^yIYnlR!-&
z^^<T`tGClE^unKKa{b*SYACI=+eWdP3EJ3=3)F3M#6!SI`eqc{VZ6>v58OS?8D$Dt
z`$?LcF}pI5I>%Q0HWK%^1A$o`Bojjqk(rHAAWz~0&Qvjo^c?RocLSXu4(TWPwl>>)
zWffYQY&=72e<GZX+Zw&xQ}LOR|KTu8u$Ft5ht8<>-^D897T=3XYCU^W+TtM}s?<w8
z#V78uGXZnyqO*$qJaz-ratznr7D!$f#0MN$u|9*|&gz-XafdH*a+2x^HVQN<VEc4U
zo0<T`QM#G$0z};w>CBFLtvjp0%75tWUC+<hu*NMJh>5BzSwWl=#Cjc_Y%OjD^pE5)
z`Z=Aqzbg32)Elb?OG*fmk!E<$eN8PF0*cn2us5n?@QahqCI;X!+63e+Zbcv>W^#&r
zok&V5!ghC!@~&)K1BdUZ(uU>Pb1eg1U3M=Ii_PmTJ9qT8T1@dNh30;Hs!hw}B3ot3
zeJ$J=?@^9$Hrx@dd{%tM_rAlV;V&lDUz*GgZ4oCkRChox>^{W9>k#k8C47&TQe(;;
zKD?xjsXRx%QU+j^@ZNnSbA<O90kOhf8SP5D8>e!cLX#I4s=p3*+Zd|I04G+Q$W&Tu
zOf|)W6md@udojwqoPHnwl<)`cY=xF(tfz^sZ&G2naP&mZ)J_gpm9dF|@#3R~rFID;
z;~np|96x~z_)GbXL7#-CUlFDb%`Wpi;JB@#1{;COSfB=r*RoUTH{ADX!^&H<SoN#Z
zi!&W9)Y*Rcu=y+a9*DZ#rNK}~=lS7#(Bp8?3}pwpg&bkj(YLy@cnQo^z)gBadF5#9
zyc1N<urlY^T1@>Nu}c3=!|zs)Zk^xHz;#FFh6CTa-0d-wUR|;~H0?cK4GIj2=NhH7
z@DL`IARxpWTSBUgx6NE9NAJJszw4h_EE9zaBk-DS&yLmFG9)j08TzePZd*F0nn>nF
z0<|myR?Zr}D@~$99vZ9}%dL`(YM2uUU-TvWmE%00nmYQ}!OS9(yARSs?NBe1ljaGE
z)NQ{^!hbtO7q8HDtJ>wiSE(m?niwB=$ZiI20j%a5_UGg1BxQ)IjXT<ZVhfnA29sTY
zJ3&kh6I+a>RsF)#vvwq&>*>s3|1;n{>*blE#M$qdl}o`cBv--j<+gNXX^%%gmCm-u
zawNQN->9l3)StF<s3HO^53yk)5&{PY2j!&yG{>=-H(m~S0BS~-^%*Jmm<{qv1OF;O
zZS_-^IUhya!L@}P{qlvQ0qdDnz<*ykJY??9lW>2fI&E(PO{4z!Wff~C%h>yjJ=D($
zTKGA83}e1MvVUB<Uv{Ar85bpGPvDE`n5H%O(RGan4@@*7Xd7x<`z>?0hhZyP-jgG@
zh{JVhKfl<B2b3H-nwu}jL`%Xu?y7;iy=rtf{cYf&dMHjinBRgi6a>Ph$RNKQI4NKZ
z9X)^0`n>(sadp?Vp~2Po?ZNilK|DsO{UL&)$bXeO!yq##eS0BUL*nXszdF9(s?w7Z
zOiM~Dq-pYMs5bD5a0S@u*`Nn5TGBf)mL!1^T+{72H$Jz215|!c<zmK8tAOR{6PAX?
z*<H297oakyRE0g%WBs?8=L&V}iDY8QALy(WH+)l{4`ObOFn@BmKfqW{y6=G!Wm0r=
z?}6&x#6*I4!4DII<SYv8|2jNqZIMw*Vo^?p#VO$%Mrqmce++X>5maulHS0vKS}xMr
z?B*^MqUU~msRnLl$O3)tW`X%C|Lw54aR9X(lP*{q*N=LAWW=L*;q@3=>9sf|GA*Jb
za_?uPZekQ;&#I+sciWBw&K9-wH;%n5y%N3GsV<#vyUV=g+f}h+rf@^LbiH{!gT0Li
zZ}ID2-<_)-`N~U7a=$m4KG~f7x7eb}er2vS$MCs_+P^5FGTege=>>pvX2rK=vinH-
zsr(pr`<ephvOmyVbY-EFq|w_?P7Z<g5gS8t^idmt5F?DT`0yJ@`3~`F8Cx-khx!as
zR7NCo;MI2M1(nEqn25q^|MKI`+@Wu9aLD&t34*bGu1^J>Lm`=)SJ*C~^ZofID5WJ<
zVgw67z=jqUYOMN!y;f(Qf?bbB{+H=#-i;#RMRKln3u!OdwjUyh90q2Kyg!`j7^s`@
z%SJZ1kL=(ktG#r9zK?~-_a)PSo+}eJJi^!7qb5QSNAmNOM9Ej;PvAQjza5Dw^GZz7
zQ4uWP-L+odE4#W2LG@7w4;-+x?hI?DdkVjUgj9dg)lGB-V&VOo%+U>*U}nPYVTg)+
zw&A`kLqUuEt!`ljw|zck+AN=nR3)FE+dH=cUEDN0VIe`xnFb&eng>04$P@}%HigYr
zaj0hf{h$9zG<+Dc@>k!*`;#UE8=LFZ=_WejLeaASe1lBv-;gCLcX@~nG)aK}+efU^
zCvfw44Y+sc#WblJyX+;ZG2KIrtKa6!+n~$%37lzH-_qh<gV;N*Y=WDvP&bMGv6N7a
zZ8TISA$qP_@gkK)*#oZUc@|u+WgXi{jnfVqSf3hOb4`yVU>MhOIPB0<;Wxd<#Q8ar
zD5$=%kA*vWZqYDH|NQ6%5pYC*w|<%{&KS+41{B_jEN?aXN`G~!f*){^;CExC&PK5~
z&&>6?L(<zWJOYSZv+LUJB(q{_1IXuc;(O1YH(cUsQxY=;++|@3mubQyha7Fq)sp^J
zj8E(Y5R%nT9qbhiQ-2N0N78=QB9MpsZ8W$9BKNzl_w&YY?qxL!?}cyfFS07nIeTjW
z*+OpEg3A*h5)iinBFFt`DvX#R$!Hkfg}gAbRnNQn+YjXP>YF<`&0P`ZUA5$wngZwt
z4kP}2W69R`7WM*Jww2CF-Z}lXr%cvPPsE}9h<21jvasvF<6!Z&#g-vn_Xz!vH=z?A
z`HP{~A%_pXKB*!!kE9K;g)k{Stb7ApetOmIyKPBgtpuSCQwGU1Qha{#9Hg>ac45m^
z(doIj<=uI8$Ig}0*37fG2;MYK4U5iBuI)(k`**w=wXw{4gjn$|6?Y7dBQX{F6glQg
zB|{?NO2QC30LF<bN|gS}l7N~cY~!y|??fIACL>H*+jVE_`CnBT8+7?Ou_OS`B)t%0
zbZvF*iYYHd@TIqxGmz}%`$pW+GTL&UTGwStBo`ve)mCjeDQviMg~g7HxWClwFCNEZ
zE?+{~m$!{Fr!MYU#%`SarJgQi@{f2cXg@{7*g9kp>XoTN!sJ+4t2^(xj6kWFp&p07
zwVRTA-Z+oY(|OL+N4<5J=_(6sVmnq;URuR+7teltBu_5$rjJGW(@Fp;N7y|)(%9s1
zx#{+FnCp?e=iO2mG*0|@=G{1H`eO${`5zDED|^Ol3a<P=Q|y0o=65R%6IY%fio#P)
z=4K4#P)(<zYUNa&6xBPAP_?(kaT+d-)RxJX7c5xirvN{-l48J>i<8#^*sPQ@Ad`gu
zGzs*sTdTwgXG-1qS~Q0W@0mluI%~sB*CMUaCj41yU`B$HAYnfyko6XIK5_TFr#ytb
z(CayfVHc7U$qa^MU%tAWN|le*(qw6+^sf+<;U@}I-*KeKyZpth@}ng0`VB9Oy+}rE
zqry`xat(j2HnYi)=1l=Gkb!~+98C64>89E2+VN$kht(3*|LgeoXFM?Uk@YgI5(J;|
zubf%Or(Wth@A29>^NP&aKHtp4#g?~0Kh$bjJauwPXow001}Q`r4H4&ySXa1xdQ`F8
zAYf1GQNSa}xWrc3KtNv5Fi8_cP`vY8q%$zUNv!gj#7WQT*gZH9qJ1Cq#vQpt&CJVp
z$4ySz-Fg9H!D{LNPEO}rqx+bCY;S(BY((zuYD6}w8`B>)bh0&PE=1DwIu?ui?tu68
zDa4*sJBpA;jq*2R@nyyK<<ozPH*zCGs7Ul&Oz#d?uFND+$S$9qfsQe&e}_l%pf`Pg
zSr*z{5J$_<F(6x5Hb_V3J}(X~F|N|Tx7~JD=h$<@Z(HDKn^5#(Ol3SaQe+&KqmI~_
z+^`+X&KR{(!B&7il$87HtH?BvPs4FM>Y!&DaCmNZJ8Ra|(R(i<&a}nLNhGJ~)Zgum
zu`xA;RNL&w29ZL?T<^!;bZ?P?N!E~-T#fy!imI6WY~XJ7W*k611+EuW$FWvMjzPq+
zTMR54H!0<<3<(nx?eo3~wlStr4yr-zWP?-ZXCdWfQ!faVd!5p3UyM_yypVcs88Fb)
z=qjORoZ&uDS4c?yC`Nww-&fn&pE~08lEQD;(nRY_&_C<;QJe;<si7_Iyi{Ak4rx-6
z_2PHoKvbwEWun#-d?`&Se^>0W=Xf<F0`9k;YN`AF12!(*0o}TWA*wr1#!|1utnIxm
zEXF(0D7rS(J=L~2)~mQHv8wDgTCq>>i9D13u9l@0saL^T_Pi-7KL=4Iyl7ap5gL!r
z*xA%|YP5f1H#84xxpksTY+;M$7zX_qe04McFb<!p>j>LKa${2oRWFjYVwVtPR8$cL
zxue58L?A6ZZsoF+4E2(}Oo#dEB6}5;7_jQWx!?3|eHESid&8`r3j%aL+y3Gs`46!o
zz@0xhG4gRb0T%Z5j5kL(;HsPO*-gXB<$1sjiiF@$T~ztqmtWBn)a$FY9QQl5Lw{KS
z0y#a@LgBy2A(5rY&wKd87w5xt70L@|pS^if^z~bsehpV*f$mr8d)$4RSFL5`t!%{f
z1+~Sq&dGCGg4MHF{nvEqe7lEQRF9GyQn!b$^2^He`4kAZA%(A0Wy|jOjm>7V;E{oX
zjCJr;l_(tQFFND8Yvkxd%F90q<Y*{(0^D()_q$)>d-BuA3kdlH{}LmsVY6Y@#^Ogm
zR*3lQn72*)?sP}k)-7Na9CF2>*OUG9TA|xx*`0`BXk`u4?ZsfzGyk1H#2)qU1zuZH
zdf}Pfe~cpI<kNBx^&AnM_nsaSswAJpJoj?`D8?t`Q;XFhy!@t5nyX^yhE>0j01Ph+
zQgxRDxrhp;<lfJk;uR+^$N)GxFz5%5xqlzwzH<Mlg$bO#sC}~H2Z8dv*O?ED;YMRI
zk{Bf5y}FufD|tfGuPmn{LKfP`CIhZ{{O?yo{8-p4B3`9j+wS?QY@_^?9oWisAGBoU
z=9pK*ti@leNAUur#Xuw%UQ9gzjO1FGr1HM+E0s%D&;*ygygK?@V<nQKUNXJq=<+@t
z=J7~u#CCRXbac6m>>CJ&$L_uEgq%Xydf6cTZ8xS9ipa$t3(72sq#ktL|ABq_9_lZi
z%sqACbD%8nG=Xl*^4^37n;bIqwPs`TuvI>}H~;GL?lSU7LEmP^u(t%BiCvBjHkH)T
z#3X2!Qoc(4&_W<jVrb9uA*s+Sn9{Vx|40>cq!CXi(ddQ5Hin#|Z@k4(563fE6jPIG
z3+vQv8HZsnMpg8<4wt%q){kFNx4_-ao65^|Ed_O}pT_Jbf04Z1knH4@L$p(o7o}*Y
zG*Ol*01P(wu>=zNgnHnyl=N=>v^tDHuzJteT0My`6#N#C2<JW@ZVf15N<gu8@t?wG
z=#N^uUo5j_A3lbZ=gcj_*Uo~MPc%zR9p9A+eSc$DE=!nnbHqzXW(BZCCN9MIJSM<p
z!F^v?FL0lN2W(LEP#*Fy2^vTH#H-GtOK09aO+U=fUBD#AahV#ZHtBL>o||6sIX<_0
zujuRzB_yNslY!+tg4sEmOOwmIwy~0Rn^2DE5StvblflDDq#x#N?7*rn{P~t*`EOwK
z%{F}l;h##=l~%9Pl!gngoLUU#`&**Ll+=fXz6Md{y|W*7>l;=7;9GB%?u$4&$Bt)9
z0LZ?+zPZA#X4eCtb3><epCL%3kBtJBQ!%~!mbFOWRS<Ih;+c@^$?5sdmnmi22|6{w
z2L_gErQib}<3mZGGpx9To;Upv1@E|yodM5dIe%f;%#IA3B^X2`{4KKKF*&oUR;<l5
zryiArZfw;lufjxeky;F-LC8y*iS0cc+V6S%EHvlos4Y7X==D6^FnZ9jRXM7W)9BGg
z{u2#WdG_(?lt+Do*6RiA(Y2ZB(~~1fCpg&~13S!=I$@FW+|~O`eXUj7AZ|R#JhhN=
z5}u#LS}AY(RPIHrI>HI^_0t0-qGm@@9}1drDa=oNslO<ymS%=FiYUJEZQ8F_@mp4}
zM*U0H_?TY}0EZ6M?RTef*bP?fwt9}l#jg(1Zx#EKbBklWosUoIPAjp2qJM*mDKdWs
z$klTL>HD0$0+3f!oeop64w5wSW2R!531-#`l{Y~GeSbMd7n2A7LaQ51K<`BbbJUYr
zEVe|Btwp8{c_XF~UET;ZN)kO8pICzfZm=k%ZXpL~E&s7CSr<1T?R1j1|2QLtit4Ip
zyDrz8#&adniMDwSzD5f@5x+(ln<4kyCk*fXM2$hYnC3Ef?Kl{>qGs{9FR|$ge++ab
z;AS>mP+;zU@N;L?T$qEQMHFs(P0QRqS*9>EIJDH++nAK#o=s;ZJ6r-8lV4W07e;03
zzX}9Lgpq%Mtq_(DiS&Re<n39Wn_Ug1zewIKx$7P$z{%PQa$TKRV^M|+dzQ6i)^RX1
zG0|motUNYJouZI$=PFKKxih;f2_U?<yX8WkCR(8mj7=r5Jxqu8abGGl<k<34nY71z
zZK^)zHHhG5Hznn$5_eXh5X)f{*m9|1J>WGKV;ZdqG|TQ%rZyTB^Lbs1XR%by1xX{8
z8upD=0@rp+B#1uSXKq$r0H|@yRlJ7vmFRblEtrX9bX&*UeJ4(tM&f6hv#$xrrHafa
zhD9s+MkyG0qCDZfY?JP;iFEFflTba~&N6wXd`RM~eOqcH0r#OAjJU2)avAUk*47@P
zvVB&M<8rlm9<#|0WO^l88@$TgT5hBjuifY-M4h!&q*6sL>Q5Oyb#SzCOltg;mC^2$
z8WmC{c0CbXBH^>XwU;_Mr?AmV@qXkR9|uWYM@-HQ=|%jJE|e-jFoUM_@36-!J>6mA
ztRk9&3FXJ~#wJcqRkNEM+@*{kf!HUBp}_`>5U`44=c887bteZQ+6-}3JZlyyJf`<Q
zmu)sXtm5o5d4_Wq*|8~uHCGFs47<vwv(MGsZFE%{U`|DLZw{8*Yd$)89j}r*B<q%+
zAO9gkz?!&}ldzf&`8KB#Z~AjYy>rEHI#G<F1GvzO0}<%f+e`y1&!n%tK$!gYN2gRC
zk$o1us%-R(Mu5+DoVEW(&aA47naEAwd1{@rDRoiI!U&Y=v%7N~gbAt*8o-^riVmrc
z_MHweNi9M?w+qKB_~IyRXue9Fqh+ZhC5dz{&#oMs=Nfa)&M}cx+Z4HMNs2BfpG|&0
zly+&1ABz=^sTW|xvk*`Lij>#O7wwk8q;k|&^>ZTI45xEjx?mg}KE`NJ-HTYbY-|+v
zO1ZLqCVrLsAKxa`r#7?}QXph?M*O)qKbdMYa^RWIW!8>BYNrap9p&CfhMlf!Nt0gg
zPRRA%n2x}Ceb2znrPHd-nXkLnNrgW^a&g>!?0QuPuqG-?N>YGg<E8|DRb+Oxc|-5B
zT7(GbI(_EpcfOb@U^9PYBEVS5cpdF2QRxp^(l??%J6#PhNbCFBT7_lc<!&?lrCs9S
zrFxwka6-7^kDM7E_dm$UfnNAbXuVPRQD&HUAwiwos6>o?ax;LG<)q((lH$2tAyH5i
zqQ&P<{ItO`0_be@yF0_4@n*@yu4V#IlB?OWI3X;bM}kUYm1C75ZbG!q&dz=aW<@y?
zZUq~rOtoU$6-ngjaCL>TeoFwHBItAY+1g^mew!D<>6#v+l3F~y%klcPX}ix!KdYQC
z!pvtaOv8Wup_(kyWPGkqc}R|-FS}X=dyMx=;LYzLpvCOwb~~A`4bvSFwy`i>lCf};
zQ2c}GLu}1lbR?NYn>IegpdHl>HNWauK`+l@LBi)oE;)G8_5=ium`KLjzBle|%cP7m
zo?qBJO>FFmr5^4+6|7<vP@N_r*VA*s#%9!;Hki){Rd2|ik(<9a$M$2x3O%bOeGq8(
zu2V7qb8_xrq!_;>N=-r!+}3PvnjG@A@K&tlu*y|xYW<9g>n3MHUT|cr(xkd+)faoI
za*@<}4I5h3R)j@XdEw<#(iHuOiZ^7ps^_LKAU9B>w1vzi=<&bxI>fS&WHrRDgIP%r
zgugPgP>L+`4JK7g^Vr)YV<*wRubbPZzc7%;@`Sz_F_24_V=;baRBx4lpl!V7%rP4C
z*x8J<m>InN{wF5y_N@Ogm0whYyq1t&KO_XHqqR9lEBx~&Bz^^O9six!fo7#Q4ZQvy
z*3q%pNB*5lroC3c=xIcSs5yf)Zd7+_bbu&-hwh>SYlc(v?<hPF!To5E=2t7;6HXv>
z)tK(Hi}S*>P6e3hJNZy>LwPO7CjCnOCL)|qEi0=33*o)s%ypd&76)%b<^2AQn(_@t
zLtP!u%i}fm606vSHg8043{}jhXI4^_mFHud*oap(`v;jwrl~`(xj7dsC-Mc8(5JHr
zY)E@c{`*7E@tQx&jRps6*I$|{+AM(E%u}leRbNMpb((5NUX4`7Rh<Ps`;6=qP#CwF
zE_>BczF+pXa<BeVWzc{=jPZjTl#%+&NXycz{`r{Eaw9?ADpUXM#js>&j}M>D&lZck
zhbkSSqB)s|i4W86Z(bI!M-MMJI&{oCzwFt}QSF|cyea$2Jmeup0J8CdM?lJeG9tJd
zs}-d#*dPg1$EZ@D2H^|B1%DT?e^m&^|7{$CxDYXD%WJ%TF<Q^_g|td@Ya=mQo&EWG
zK37WG%#P;#qMhc95w&_h(!%9y6Hk#=P|Vd^herWIz};;~e6~~p|BpfISe8mL!AvEP
zRkO$4Z&`YugAz$(hETOBaM%DWMl``id~Yw^l0QmU8Gi>KbznM8ps0H<xYzB(7VEtA
zz~4{2`D0G0Ga^sLj9Oe@yBH+;xm%rQtz2LKo?0iAxMcj0L|uTA#>+;^y;3vkTZC98
z6(ukHO?Cg0l%%ZrT@HNbV0QHijLnK~<GMrIH+fi!D0xwS%UAxUlwp8-9CJT)mfH#=
zo_u{n=gAL^z$qwjabU3ZmoGiiHP*t@({uZ@79>D1WcK13*7x&I1X*=O1*N=J+NO`#
ze?(l#t3i!dMUhw*HND2gmj;O*h=qfD`PWe}L-^jPp4k<$+W-9c1lm8nFgu(c%1r)W
zi=u(4kwe3r;SY!m0PgNzwStgWHq45sYF8d~px|5_BSA{E1#g%^iUc@_js}5o_vYto
zq!H)^$GC4*CT*SdRwuVtsnWpTLoaze<;CSRwJ*d6_rN<tH&^YC`GSP9#<A%_tB$+h
zsqA`sx?^T7pU#)P#~My^r+gxc^@!Nyf)a|$V59v7o-&jGQ)2P^N*OHH;KS}|zSDN(
zf$?F=99W{}MGY+jXzF51rLPuMsap=S$FF+G98z?!Q}d@kj()LiTN7}LU`w@YzB|ft
z>A10c;hDK*Q2rbo%&b7cg?iS4N+dfplDL!tn%k05N}Bm^>a-Q?A4t(!X^r9hacq1f
zli{`a2g^U&-7o5;L(z)2dbyXE^+Fmel-2Sz$D@5q_jEtgnb7O=Ktz%!d+nuEpd-6|
zHugpp$SQglBElk?W&vxeYu_V&Bwutqh*1F#S8hwGRGB<T?g9^1cd$0M_0B#w6k^%z
zG=+meum3jxe`q@IM=JmSjT=Xfh+|Y%_9ioX?-6>VV{h5nd)`)sjxvr!R%XT_bjVI|
zWE}~|K9uaeH{bj7!}m`(<Gx?7=k>g<#}xpIhOQX8UV1bRW$Fq3tRU*=g&<Jrk<X=T
z@eW*(*iCgaI;kzR1cSJLkmG*mfzZQ;Hf_5(P~z0zdhX`d*6;OsG>z8A=!^UKQ2Yt7
zDti>Ta9IuMxN7I@rRoinw6%*^UGS#WhDoRF*HlM|s-6A)-BDCsZHkA-Q1=a&Nx<~e
zODY98RG&Fe(ZH%h;4U-<fMtdu6p2)@s9op3#{_o2;Nb*D)Gm46O~nI4X&(%YN@jKh
zNalhEKoA5joUeSUu<6(K)T3H;0-D50_ob!C?MgKnw-<~l#+5?1({#RlCse8DGD_4-
zBVbZzCM0yEuoNstl0K69Q~Al&3T2q}Vl>!_{oFr?t6TVJ$(kPp#Pk>G%Ohx<_#!2_
zGnx*@NRucEu9^#imM)LX0JjT96|j8a?R>yfPmXUU<%Ut`4Fy^pNqdJ8J97~`iWd3)
zfoW&^{XL8_Wy`CIy;MTmm&*mfQTJX1)=uW;Z2t5Piq=I$jlI8rDkLKBnzSHtdCAOH
zpv_vjsotrV(#0ES1I0pHpP1xG5CS^HRmVT|+stxtqqeQ>i&O9D6JYn>hw`!mSqaFp
zkSD)W)KVPz)r+pP*DDV_AQY4OG;|m!^Kjb*e*K7v71%bJkNOjFv>t?d`0-C-o?(`b
zq@J9_L=6u|y&SZP-U%uOA(Zj|rvryed(VPXHYqiAKLF<2{ru99&V`+r=;jnmQ!w~m
zA1!|}5frtENh!GOE4UOK1qf6iF1OfrICQv{HK?^&4kIFtw>ra*^lan}V2~6sF)iG4
zL=uIiC&%qU_5>&qfu)HHtw!~L9;Cm>aCDX}+)S{WOZWW?0#5A?=dW%Q?t1)QUkVZ&
zLPQIIyQEa;6XBicgA}v+S#!_IO6&RAX><qxH!hbL-0w4Py8GaZe0yajYvig%ld`q>
zDvM&{&SAC*K|NaX>VD@{?aju`alsrz(FvaXbK<c0sJpXX8cxrF#_7v5>jzd_Qlvj=
z<u@naku~fM!b#Q@y5*#Kq@%Z^-aXTJU*4DrtZXT)g)}%Q+nBp4`v0{x`n45o$#b_(
zPlwjaTJZu~H7_dOD4?;UzN<si_O>pjkaSLQAcAsx7I6Fg_0=~4XAd)S9cW$O$fUZ5
zeBy75Rnnae^K<%n<)HO{eN8;Y2JyLHe%k`6%N9`I<ymyb(zA*QyzFzP8cLSQ?6NPP
z0ekI<CUK7@R8VHfL#C)dv3y$l%uytKkAL7wQm2?-%bDWciQe$VB;3{@p!?`}i>+k~
zR_H?Tp{Rv&rqzXz!`|XvVj<dF>cM5}56jnF;BzauGEAgmL8t41LSyJT+|ZU)8yj=p
zAx=>q`u?Z8X+i_wf-arTsE6JO5#~(|uqW7~K!bQ+Xk#&Ytl9{)n?$Vbzi7WW8AJOt
zS%QC*UHDb0$jwy#pTBACt@`gZn`CR_Uqcx!-am5x{=GS@&iUaz&Dg@!=T66NgJIut
zEkBe?iOCsw<v7@&1f5+rJhKOrQB2X|Rk<4?LPg@b?@0VMzema$>f^Q^qR+_z6;Z*J
zI$@I1aYJseCb>K=-S45f^fNmrH8$;*W3($CJ06-bmBbJYf#CqV`yT@h*z-6qT1+wC
zcG-;TP$9<H%#p!OS(V)&v3nUVVJwas#c{6-tFd)?OB;)1sWjB~J6Fustr_Sjcm)@6
z(G@A>Y{bshFmP4XQ*wxHzd8o2sFY3%jk$ztP_>}$Xhbd6d4^#E%@e~?vZfw~7dssx
zfBSmXi&FO_Sa15~&)b0`USMDIUHHkhqK2NrR06?AdauUx!%Q>VuzmF>c{Z{V+4>v(
zcY^OoNlSld3-$EeIBh@4wUaHsD@ID>HKGEAw1pp>KhBs+dc`WZ5e_W<CHcUUSeeOP
zw<O`&Ta1FWad5GzJeq#=UDlr97!HqK{4lJ1{V(IA%rV%c@D*kSQcc%zxO+4LrO>3r
z{+8tLrc56uU5H<vJqK^LVeiV_ThQonuO?5k;8vT`F|Ws$0D*za9Rl#0cD-)adEPu6
zRWst6GkJWUg)Dt33_uJTwY>#5`Cm+v&-yc~E~p-WJM+<$4`N!yk5mT_G%|7!JUn%}
zjvvgR%z`UuQ}-M<XGa9;izY$IZbrf7KX7z-sM#-Iq{mQ|U0PjPU&$F@L^s3-$|HA;
zMqfwY>@CiR{W&w8zFHtHFmLvi)52L2LQu|MMOTl{6Z=L0&J)gG)933$AEa5MHur2=
zLu{;No9g@y3a%S&)|;X??4CM@g3i7hZ0Gqz$eUFsd%x*bdUg1?#XC}|;t?KbC{^a`
z3hb@+9#u2<!@3&q^yY#(-c1j&G2Wtk`NOm|Q&=dFH18kY>0&N2;EUhcbf2JAu~8AN
zomOVq<A@VL>$oawcrOrzVn_p*^<mufQPrCNsq0yXNi)2i&~NqRCt>FTPwp2`J-+Fu
z@iAhi{P1@#20{BMDDSmjVL<W22GzEMrq%Pae10}fn^T|D%2C2I)vl0Q9kI2Xf=;Tk
zR_ySLUfIuht9<<vvmRy&vtd}5)u|q6HvBlF3+V3Q=%)=|RJh(MMpMPUFk)~>Vx|uK
z=d#h|;wnre4mwqJMiHRJj9T4SsLqIZ=O=qWPn(1}*QlX<{7C57&QV6Uhf{`-SQ2WZ
z{xIuq(Cw8sEHYyZjRuZ7u~b3%GcS6!d&pO|WrH8-n%p}6INk0uv2>7iKP!0pk!C8m
ze}2<(h0Ff&>dhyL_7JF_Qr^~7b717n?k{i8k2%N~!jWMOt<0*%{d*p?vu+2=nF_-{
zT~rGD^vgy~^I!YI-^KNFRgM?gJ`S9GZ`X`$OgBf1_bl}GERUZpFZ+gdavZW3){2*|
z)EFL6O01wLiI|o>q{y8=s?=fkkvh8gKEMV+AS3AdoiYk#y|yvUzxE}4)zRC)MMz$`
zf_Kh43;MpjD(W_yW7p)(TeLKJ<7V6T1@0W9n74;jpP!k#pThDhW&0L1juvzav!~IJ
zs&G_%6q>4kCI=5oq|IWzOf}}NY2zyOFEzfK<=pp<#;X3gMF~WY9-7LQc#BZ#7r@D=
zc(Qedb)6oSUilL<W<pnn*!k0zZ-PFth}c43-^|;Q-Y*@3!E*g@<m>{jP|hV^UhgLv
zDkRkL(evX$zY!`tO3CW*aK0{qFE5K$bU)XO^~a$8a|s)|zwBa?c5ey$dOqy>Z)w&1
zWgeiB>b0}UWuYcRMzFB9jKdD@1LC4s_oujc@+{%#E7qIfZUW-4-L%C904&%YjObN9
zPE%nA+l3!}R-so7yEtCooE@tO!CQwdEIf_aA8XwS^wMEU8A1g)Z-C~ZNSVJ0S<Vca
zBcFte`G~ma2;<G*N5?bD)98p3r?-mHYx~YSF=RZ>$c>h!xyPR<q$U>tA({IME3-}n
zG|Vz4R7Qy*_|t7-ss|UF!@%B)EZdx&;8#DqT<;Y7S7)&@LFoxG=F>VG^<Qc_f)=N{
zerU)32YJ65JP+(#{;lE0%5t1p=yN+6h!Z^b@){f(BK-DV9LjF)LjWF|^t^Ow-c5_`
z4vBH9({5shnp~9x#`OZg&I7rxopT=6G9m1&clgjjUVMKDe~~_v3EvrgfB%K|fQ;cN
z*JDK$l}TXt+xWnJt+3{Z;ds02^@jQL<AYd%C-(nQP$!DxTBwjhLoN+Mo{N_#`IC^X
z>7&)+UfZyxyF?D2Uyi2V+TKiREB)R%;&;}}&noWH8kY|}{rSDd6-ec!egbNXa-JBU
z_9M$G2A}G!%LCwA`3m28Ie*=Ib7uEx+Qq$T#ywI;E}US;X7?R=^@ioRfG!Qaplur`
zJz^XwUwX@ZkRe=6p?Hv_7LDbQlFxpi^f}OXXKb6S#phN?lU{|HPu>mK58|_$$MmDs
zY}F0b24vqm@e2(AuKp&W27J!&s7A<3d!&>e{D-RS&!CU-eha98<`GWzP@zSahN-ni
z%b{CNV_{aN+-ix>xJpat@<tc*l_r}*R$P2M1!eA%=l<`1)m9-3VPqzI8Gy}j`JKNt
z@eiG`(;0QVKGbTZX`%jE`FzAvr1R3TTwbvF7MrWJvDy~^S1I7AYIP^A2PT%$gEU%K
zQ^CAvOXI53&|*Gij63>6JP=FOP3nVcG^M3E4V6=4xTT_jwKz@vm)Wk$%CFZaw{?sK
zt4}>uzP}*&2Kzj<uCH%+5*|hH&zj8`Ltly%TW`0soDSMpe&|ZfaYrRx$jaKbCG6r|
zYFZN|wd*o)A15sZ{a7i5tY2Dk#(bW^DxHj#%aDVZbht&9y3JCM5rx#Y&wO+CNq^)j
zW8X6Y`8nK}hO05`gBIj@n*VdHzz(LJJkKZ2<~y@<Z9z%2KYmf0oz)lk%vN2<FT_t^
z=s%Ps?He+vr*@|$$|k)x)9f_Wr1RtYIHlj9_16H{dTM*;g1X#-kz9B>k20(1sU(bo
zo!;vys<M82>#(ZwSV_Z31B$AWgV<F~Gd|>;@(%e5LmafwlbEoFQgaLGg&}2D#6ou|
z<4$1EMfYi6?19RsJMfW)GKi?N1?)+@dfa#Vs@r#C-k}?(3lsxfJUtucw*LDr0*<Ap
zA724bcHH|9RaJf*FiGx(l=11~TW1<gj|tO?^oyP-?eTkVf%eW9FhX`jtIJCyq0`~{
z(Z-MqC@qqQ?95)B_9|<px^I>7th90Jrq6Z)fTQEza`M5-%II}+G?uZr2Fx&nF95ml
zL$~5I2DAG;pXYZZczQjJPQQ^ps|NG1K&2#}K5Io*)&KZ#?O<gXWVjz`|7iO}RQPHk
zjx=uZ2`1Et4lIgsYQ><d%ur0Us=pYEku+)vH#fwCb*>jF(s+6+6-?)-rKim=klyQM
zqc54UX83!pyCGoCRDP%)A&&PbDLC)$#mk2*XJ*;uu72q_EESQ=nKSkDZ_5P|X_W2_
zq%=_)>rr(G4gB$SLn(f}`PM_ea}wTO&j;#=yVF#*SNmoii5&dMk$%!0^UrC8NaGlL
zM4GX<z%jVP-(Bc3vx^$1r|lIBdULk>gSD+A#M#H|a6i92L3DR}DeHRwdV2h=a`bLj
z6qPR)bWIj0hi}%QkMIqhhk#(`g~)XlZ~nP{@mx-R{ss16Udh*f{RjUc5~8IG#pkqC
zdZ?KfXT$#T5kbI)_`eCFe#{VZ&`d}LA#md|=67!$V!Shtc&c_@-Da>qKKWd<J)^Lw
zjH4_#DAslYK0MA@t@YH5TAczfYoJT8Q!Sh~jt#A<4$vBRHll+AMqv+f@3i1%tY$Vm
zH`=#4#<AH856AkEaM{hB&F6DB*PbFByMGUwd}_9Hwpo-v@Aqj_u_%82sxW1pzgaDW
zmC8}G2m&HDQcPCnWrB{rmmUp8dkGQHtf=j7Up?Ldz#T?&@S-8`o>Wf(24{jZe12o^
zCS5euhk07g9qI_Q|Dtq2tdR1)Cw)qoN4gVt62qcHu+wGqmro8$wvubKo;NY=i$RRA
z_H)myQNiIBHhzYDr6=39UOqknUDRSKpK+AuoKAra*<g*2ZW^K|Z&6vdY39L_>_>&<
zQ7s&gzM;V|Wpb=k5En-r!mlEQaY>e-A~}2q4-Uf74nVYvKuY!qrw!%)pJ1EA<UfT?
zbAsVN@N*x=KTFGw;5&VM${$!remZzqP5LzDQ{N1AAgx^H*}bZ&Bc`SEls+EXZX?F$
zlDnEdpQC%1)xRJ)xVr^y2+7zXSU&Cl;P{!Zu2l*#s*DLGKZzVvCz>VtKgE3{Hi_`D
z-bM9okIe!?&SRD{J5;D6cxk2HTuT_Q0aFkv??BIeC$4gX?;U%n$`k&V3)RA?n(E4g
zL?Qp|;ID}os^`>?x0q?ESJEnSCkWw%k3!8XOnhHfHgf1ulMz>Z$<DQyWLak4ZIL2>
z9e|J6`I9$V{zl>Qvps@CHe&m7#7P_K*TqM((0%}9-OD-=RpVptBIPe8Wtr{eQ{2Na
zIa_?6EFR_$--suKLQ8}Os;Rd&pM%oa*<9;M&E{NI=}>TMYe$E0>FngB&gfIUwOpnT
zxxF)wS8{{yQTNw}E{$hpXn1?q3KloRghQzUncpb{XLz96S!-G1akpa3H@I9aGbJP>
z2or%|JjrJ~Nlaa#(gJC-zw((aus*g%MVCZ?tZP0y+uGXNdw7219Z7Op(X5bAr?z&s
zQUDxJxl*wzUCx==&1~P%)<mT=v1o(eo+r8ivk5H3KD{mcOWwJ9e*x;O4HgcCEF&ax
zB+hXkd~BQj&klw!a~nm40*x}AZNf;k0C~?yxIl4Uu&sqUH0<AGodj6d-V_pdhmI_~
zv$^hg`lR!sStK%c08-&gNw2wr!l6%`#AU<&oz{^aZ~iX@bW47wML>FD+tpYGPW)B*
zECZNdZ+ynwG);?H|6{*_9eS<%uhq*vC~A9isxN7(SLwV0wROD{dGMipxZojB_1ofV
z1vm!E_g<B>utOM}Q7#R0-nZW7_PIznLlJc_BGpx{=Z)O1jA}GX2BmO9`v>OKJ&R}7
z0!HHM>T6zxaUTg*yWBVO##;{YM4?XbiUAd{u_=z2kjI>@n-f`SYT0b><;$L6Sy4$Z
zYqJtoOMmq0hpJJgcybAhUI8G(%ifJ^!djL$O6_>p%vrUESFFxiS>`JF|63^TWj*iK
zri|X_6x4nz6#3Vrm*UjO!re*x?58CaS6V%->w|yHB&Eh2P9crkc5Qw^f~=WRGXJLI
z8*RsV+&=V7X}Ibo{#;#n5MXG|>NnDFX(I0N9}8g-EKY()#u@Up5LpChyGA|ac4f|1
z!PE(B`*F$Mv}Qd~-l$`L?v0waY|o(zv7Bq%^Nx!8VX+b=I*o4$TDiU&>g(%s>V3=T
zl*_5bM4&T@3}ET?<Mj0Qem>EE4=sk{bjqtZwx0~aF*-e5*xz2-LZM*4WGAQaYhM~n
zIUQExP;jx|<3HEpjD>+>m)u9PoCghL=R-&Gt`VMDjoa4r)pmD0#t`lTG4X#3NAALb
zsL#Qz)PDTXLZ|MALb&XwfoaY~quWS`51Gt8SeuD+9U2YZGzeo3jMx6SX}s`NGT-?v
zRE)TQ(JvzX2>z*wc7<^@!;q<BpB?lUlZ$?~5)ekT6LJ{oy#+FtF?^+TDtrSp6%<%2
z_{S#Q-@Yn8NI9Nr%NXPsuqOJ^r)s)ug>9lZ7wG-4PyM?2RoN5%AtekU7rZx{1*!`}
zS_i8muZneO;!d{b3-WXH67-8s*U?qlEJ|twktn!ru2RJQ$XliFWJfsJ9WD6Yc@iyd
za^Sv*sU{&b0g%wj?(@WaGk@s(3<c)sNJ|fz=9SW%8Jw7j#)1^J>i>n|-5z0b?xBuk
z9)`7*h-%5Y5`G6Q^UKS@y8N+hup%$7_&dtFCcAhuI6U*nydYksI>@gZj&gA=yMO<F
zGXfLrw~-fweYJ2o>A`@ptIV99ou!v%z>~(A|2rE51_x1AO<!<?-{(K7Ud*WN%-Th6
zh=7!AzEYIwMiOtqWy8&7fpk+7F&Kg4o+bT+V4@55{3OuZ{~MXfzwa^0x&`>oX9{NF
za64vT=OEnTuIs$~)pmA<UD)xI(1vIliT&bx6I9AcER2y?w`%VC&{YW}T<m<RbkU)w
zF$DPI$`_<3cA{3!e4>u3?D9U?sR<A-DiN&EkcnfIK)aJ%9#*in&HCP%`IqH(5+Vyu
z?nKDKVP6Ja5;85s!Norr@-^gOx#04Zvh|ew$)Lm&#l7q&H|3x~>E@L8M%6!jb5cFe
zbXISTQ=7%oYW_S`E*xw(GyYB@_;`rV0HCF26BVFe!pj-1J#A%AELAn@VH5F>dylDG
z`1eiymC{W>!CdRFl!`Z+_T5Wh!Emq;@vllsd`eVtZh-4^3p%W<UQ+zGXIvQxQ0xmX
zI%);TXWRt3qOwKkW!T8c0{@~ydQ1ZBJ70q-0}3$`1SfI*HQs@BwcN|MnFGyr|D9}F
zOiC)tN>hvImqnZm+I2)Pov+UkP{fF0+8Y)gkO@0-WoY~-LXwtdPob@8_-Vn&B{uXM
z%vx!)xh0u^R}Op@8(}(RuxIod$UrHnTUli%#2mnX!VC`2T|EOLoF2X@)~cdb5XCe$
zHjz0MkGO%ZN*TR-_p@?_w3vHAv^{d9;1eP?prfrV;^^2$4Z=~5<aBO57OJhm(?_%;
zP9lzl?b;&UWdvT=_t!twi(crO>vE^i@?U_&`B|P#r+<-XKIw|GOgVI$L4?D7n35ei
zbb-<m+*5i)B{@;e@1x^mDW}1RPYB}Vp3i1?LD;jorRXCEclENQ#8A<)$?cu$dDi0y
zm9u}d{+_`hv-#BGeT;bo`u#!%@(TUW(ly-&I3!dSWj-Q*>$sp?e*Y9zc_&qx*z5N*
zgLV`QZCb88aPuPjNbujg7r)T~!APnX&lCW6K_>$nOZccgA;SuFsB2&fT``yQQ#J?p
z)|s6AQt*>YcEy{%Sm?@0JJ8d^$YRoE|09pM$J-)7@XXs=DrO$PE;;>BO(!#JoNs>n
zq$Oi?HM?+@A?Rd#=;{><TTHGocn3P1PT6>(e1w$VuDx0IfsKThBU=KtOdB$d^Z@mk
zmKtg42?WX#rN6d*m>m-JIYtybaNmu;+Rbnva$-*?;)9LQ579dy$ogRHy6NqL7rf;w
zmZ*EqtMLrl0Yn6#BUy?^kv)AuIp$Cn#lYQe`<85P!0sj-*{zf0hYMecYr?<SC4@XQ
zFkw|F4OEca$G0C(dxBv|weC`&wDd%z1K65IcmCkJHgN(YHN1qoY1%9dBFAE>c4wH{
zsGac%!|T(zKWwaT_W$@Iv&w#bzF4SRiN2a04(I%Tzd%U-*=3gUbsaD3Y{=?p!S!M3
z>z%6l<QIc~6WISG_`eAOouRvb$?YyMwbD!`F&vP;4`}}7Y8nGPTiz>91*fN0KJAjs
zlqfM%CQl*41v)YkJ=MGNz_lxI-c9?2MIln3RL}eoUK2PhX61$emy1o)L+}1MZq86)
zJ$?2oKjScJq0ZQj01q#(t$-)WPERMTK|65gdAz6Du<%a)*%<BHExbf+O7!N);?1(%
z6)QM0iDC75fnvMUOV{Yr;wI%>wVhn;zmcD+Dy#J}xaxxLCQ$k3#i$_@4SC&r0>4B*
zxkK8)w~l|eKKEE>O1GVHmgPB0uZWW}h3j1$gN+s3PY-1Leyyx*3qcqsucAzXdieh2
zTCKzwt1+ndW0a`$s9!Xc)QtH~Hci8x#X!;)-lsf)0&E_h=p)?M7G~KGtkX@0Nxg3s
zBcI*O&gwUfgO~4wAA^U_bAr5QEy>xSdISP|&Z{l_%whKDb+4A+)C>-~v~lZjq*7Dh
ze7ujPq|T>{j_mUqY?-WH+4IQfi2!H!qt}|jLF(Bfb5-9fyyETEl#hg5+;rB1qJJj!
ziZt}J0DAUrbFXUF4gR#4CMgJ$CadAcas<;#gBCHHudrpF_pFX|W7Dl4UF{JkmBzz^
zLPNqr2Qt}Kv*?!a2ZUk}r!PUcFlBl9keh2p=QzJLg4k~$wBEZ0uZn>BVm|4oTZMS@
zR%{wd4co|EO&0ggq?DD=De&kvOnHm5rw)7rdIo3pm8SRMEJm|avs^4?Jmn;`^g`t<
z?HK!k`cJRawV;2}G5kxv8ND{9ksc8dsc4?Ed-q&d%ILX^^SX}DZ^?X+VoV8cl927o
zka2JP5??*FEMA5r{10+Vh%oJz0CW){(JxE$ov#$}3igREp}}~fSV0=cR~q*U_RUfd
zXfDyoBQcRT+i2@lC%~-d)Z_Z)RswA(N}m}qvpY@8c_N-s@RN0!yorTO=@3Fg<6`Mi
zx7xk5zD?hbKT_>W+*UqZxF)8ua0b&Bq(P_}+vf1q-DodCwV&Wox9q}GVjuU|^G7>6
zMaJuk<wL!SCM-k5)%4wsEfJ-Yan1{X(l-1e(3TFfkuZ{>EY}mtsLjt#???FVQ+@tw
zpiEXuwS;tMx>R}XTstSD`xde>lN9bLS;h%TuoUc1SP=XAALT)zq+DzBC~&%3Otmc!
zYC8&&-S0YZ-DZX#7W@?2g0Ku51_bej8V%$JuGqGSjpd^(+wi^7CyI))4c~xlC_BV6
z&d9pnC5I<hN+!zJ&u@5}goZBeY-1;K<(Im4vp&OpWc_0A8>?c&w!M~PEqSjv$u2ja
zL3+luFY2|`au@hBL9<$rtT>G8r)R2Gr@E(mIz7=5yn)-=iQ2hHAm`vRrho($>V~wm
zdOABg?Bqw0`pkaK?x{dHWdq$D_l685utnyH2nmzpo6&-+k<sBumz$63j1V?%{Nt`g
zF8T1&)d<&NS*O1;T3QcoEh5NcjSXY{|MCA}UiZ1l0#53B-(IatU)vI}aHohw{vn53
z<ST{bK-Z&uu_pa(RDQV-rgdvw9)4=S+)&pVyXL^_b`y{eI5H<V08nmDU&&im>6y)Z
z2r<3K2qyE-QtW`^;p<Z>aSYM<P5}r|%fI;hs<?&0HRKVan5yl>((J4-t3JU(9RCZf
ztbk~eJEx1Y5!i>Z`<=h?K^)`U7L$!^W_$oMn<yRPwpL{&<By0;DV&_41&uaxVNwEB
zDQ8IYtlN@Yg>69wIF*MS<`bw`m=>$fuTIks$?R$I(?XVFi+<nxI))pE!>LH)^N!5x
zgNw(XOV)qs`qk26pHMPqQ$FARDNm>9i|IE$y+iVbE!X_;YU+&0&o*YkVJ)Mzv!yc|
zRVN-BBj#<LI3`T>94sY)gKZBMx%yTMOmpLxvU!!m4kPA)2u@EuL)wk!T-nm$%qA=Y
zpQhZLQ*c988}V%=5X9c9E>s0{Yhzm%6mL?0{zmFPL8Ci9(PfrNiiKTZQ-URI=B;P&
z+fNJzg@sXEL#A{CB(B2+;S(MPkdsw{_=8_=Xu?9is*<S^gjH>8-nOr3y_6_>plo8O
zMyxoSnFwJ0U%X?mM~=jj&0WoUV>H6pjTfI*vHb6ofGo-K(eZhd31wMUaMF9h+B_qS
zd$|F)v`K6+cCJx`>XjhjHkPH~O!6t!?Vmpn*#81I!sc%mg_3BqyKg~t(xh_GI2t2g
zns1p)iJUF=3?FqHNI)l#3~yIUy#R4sQ}E;lH>sZ^+98d(H}f-uiP%6tFcS+A2R?<~
zw4?cwu({zYEBO~~sdNbCh~-Vrw=9H@uq5oo(oZ@Lc1IS~G_FS2)C^<ul|#apz7fNc
z>>`)%UoYL9N6dpbw9d<Mk$3G~<1^Bx_vXIphG4;DuJf11Mk!_G&bp2Vyd%H2m(G^5
z16;A)M#a8e^z?71OOZ~_P!2yV;A~Nb{EK)vINW}7-JH_eG?Ricc>-+914nJYNfzm=
zon&rVzwm>m9SA3UD_sp{wAwhytB6yr`U$e$0>RB~`{P!;Jn$yQ6$vT6!=YO3wVKsZ
z?B(OSyPeIdbh>cu-6U}<f>;IWFi(8JWU<j&8+p?AM7d9v6vB!r34j9Z|CLObff>}r
zCg6hgve6Vn{}##~<HrN}uPpq0t-RK@{n!jRGkdFeefs2m4j~(#iIpo3C(WXC{&OAe
zsRqZl7X6f6$(g`2I0cqIqB%!(o{F&E{43`vbD594y68^Jp2}{gVv#vpY=Xo6vAwOg
zod!v{A6JN7h{(K#q5)%I*C+aBrSmjXB<gGc?0HuJuCK2mVja1@s@#*+=3i1PMCS%<
zNfP*Ca|4X^oaW|QL+XR&S?^kXU>XvNx<Z>7wqIa<Q#XubQBMk@vI^Tnj}PbMPf}Aw
zln$n-$~{6^p?ek*w^`b$5bCv^2GBD1M~ug^E2d@`#`|rFfH8k!(>v+~3hFy|lrrtf
zhPC@GT{9byT4^vf%MF<ox+J4;1xk_~26jTR;Q)ekbdC}PD_j0OXPSgchd;}Jy$h1L
zvssOD{3f&UQ|H#O=Y}9yubcV6oV1ZKki!KzZ9d=6Xv1!a(4tRn`y?$LZeVvuu=7jX
z$@R3<N96l+;7?S9^7(FlpVn^Ua|223W=r$Unxp<E!vZ7p>tc+32~17-RkSLHt_&}i
zN>0gRRpj=SP3yZd%Rl4=%3!RDS{*89kiLLG8e{lB>L*WHM((}%h$dm5GVMGzmFSP!
zby97YbWy0itW{2U?1(lFgY);l=JHhkR%xuYq6Rh%^z_U{gaq`ZRLB;mxP;$aPpeZs
z7{h}7(+S3Po+sajtg67DzHeHfOUB7N+{MFdYx}Y$4CTdf`u%Y8YG%XLu!D8a&?c4=
zkxKVTH|M|m>RaYj<KQ@AfZF&@XSIa0NrF_1Yq8(8hl??Wd-8E-iqpjeyiq~moc!aZ
zUg`iN+3&S5{gF@ObS;OtygpfKrXa!=jB!ES&Klp!jd8_W?`bjFdu*kJ36@Nb{OC%)
zqJM*OCW0yKkd>CXMlvoM<B5L8(HZ<l#BFR|laTI#!Ycx6|6vj;1noEbVx7WqOl4aW
zbpX*L0vds5TqJTk?XF$%O}MXhECr%!vHSoP%H;<;xWZ|tqU85j5kMb3?!27sleQDq
zjvpUqYTChjxTH9mYYDYPF71!ng-%D*K3|H9*!kY<bKP`v=l-OgNv(|yZbdlvW?kee
z9X8StPa2G>Zo6JTQN9^-ttn8ZqNY*9O-zhFG8m;Jz{ICnHF;x)gx@W#_S}CWWeIGR
zD_;bFDRrxMzmx$te>DL(tt!8R@}2gN$TIg9pZN~KERif{)wTzl&p`y|V&8D|_SLBC
z_FQ`pLsl+<Icpr*Gm1+nq;=G!aNN!HgV=U8aUA7o&k1%IJE*YNO$&kqerdfbOeIt6
z8T<RUVg57|)Dt8A{MB6feoahD1Q47hsR}^e>Z~l^XUyAf>bx@ReakxE9wE&h_*@q>
zA$3G9ftLJp)ciT>T9h|IrLZf)Q*o)=%>WA58y*_c);<6ISmc4M|0CUWm>Sf@X%UOX
zVe>OdlO3p~c|I5zn{swFv<O=;llbbvk^Iij)h1hcfymWU&S4Php&Sitpw=oXLJSMk
zpo<O@^DjN9i_q6r%Dx0nUrv-D7soR`N|&p9^3p!u-d+}-Vr;hx(`kPY5!_Y48Xr1P
zbQrfBs@kjcpH0$0Kxs0&FNyrVF!D#NeG#?x6-r)i=ZSuxu${~X>SUo&YU>qlM1s<&
z`&E>-yH3L=>4a)(+J68d!@N6p#JhaauaIwo<=t`hqtR;YMa!zm7^GU=BcVqB4FQJU
zGm1`^p|^k4w8;s{co;ryNfeWafcdKO_pDn&|EL%G(a=l6NfMgmyk33ZJZ!?5p=%J8
zf1k(gcB^(jD^Z5V`C?^#^k6uq_c?ZnIS-y0W0IPbybYw;Xn?7I`{Ds?UyL;^T+D&x
zk*M#>V#&wf{kN{#^E_!EwlqfjOl}wiyUgTmw?}NW;|mX?Wo{>YGpC0TFj@6zzj<)a
zmzPFnwBK=!p<LX<m#59akD0I-YhgtN-Ji<FU7=H+6{QSPUi=WsGw!EVi&Altyqcm3
zyjG@qZCt!g@CiWPjweV6;VIjmvv5wp8ax4QoHD*0QmYTjpSch3gOU;zYjh%PIZ|c8
z`sbXWpmOxZVOsH97!j<Er1j=o#~-%~70-e}uA-aY`}uquk(&X-PtW+4e~GjeQcF_T
zNtR|bw@8j8stDfmz(#rr*6Mw49=!`fFAO~3+zgtvPXP;!ZYC$@-$~mgb<wV_uS{AZ
z<~v3{JyO!ima!YgVl6gG^s5i_llf@EY#HH8SE^j6)6;V<4eU;Y5KIV7=uYBESumDV
zuS;G_6yg&|VylH}Y%&xqMOs>0Z{9hVm$#A_9Yq+E(#7#E@Iel7Og*VGo}@EYEzJ6l
z>g8BP0oPydr2-nlUq53+)zFf{2{GU3sbF8}#>ANvX#OF_gYq76+>h}yjInD8mt>`z
z3qI*iV%5q<>JzcUjo$A2`+L0nT{uw5Xo{UqN;-BMYm(!#KE7!0`rPz2BI}@$;<#^E
zIp_~)DF2s4yE2?Fd%4<my_8|x58~aweg%6xf$jK>FAXv9s=3K&+m$$PpXZS`<7Tcd
zuFCt9i}!`C_q+Us*IO9hPNO6j`QmzdG@S;)rYFT)a6y9KZeQ@>2d_v^9+;cI@AO|&
ztP;c&w+QodP!ERna`KO!VvioX=YTIlaXo(RF8Yn5aG$n4Oh!T2`qha@=8vus?KjHD
zz-ey?I4ny_gq$3wjE~EAULMsFi>ml8-Vcu1)5(C*^z@nKM}g}yyN+O|qdwjOE<yn~
zxP$|5`{?=0A%M-<8Xg4wgx2E1tBWy<VNjB=#D>)FDU&DB$Dk*n%(YA|Z#%o8T;^}r
zE0+FOnY_;&!)qAte5@&w2~aw<?**F262$uhZQzK7sfP0L>v(Zb>$XUSkpOl~VM!go
znI0Jd9G()S;cStv66;8(`7<o12TJeC|B>sBQ%qr{hex;;6EQmy4zH5SOga%B>K+{I
zZjbl?v1k~dWdRqp>w$8`(FJ0ZN~e|%YAL9U_mcJR*}B!<tUuV=9Gexo0_MQ{z|DE(
zo-}Zv2805^HFGnjYQNT|D)RCOos4}gmhZX-wmdwp9ehlkz(wizej~c(_p(M#*+Oq0
zR}7@rSSSujQ~AZ&x?9&V&r9k<j10c-KQ&@VKRr#d&R4nwV_KDdbAIB4|AbQxEaB~W
z4~z4R(Z9<opIX2Z9B3qL019vGup1C5#}ywek(ZPEIWaG(FSr4fDK6Zd5**-fTV{eI
zEKV>|o_h0U`@`dpg<Xxj%Wl7?m`#5a2~|$N|Fx~J6!~wJSg@5>Zp*MiR$98%)je!@
zKR-(F!lisAC&kpK$~8z)eo99M=96oRw^D<uYNbZpkKC=4>LS&n2iRJp50o4t!-oGZ
z<HEbcpfXscX9y8t;mBi<9c8z*qtJN*Tqlj7rEkZnYdA&aa%txG8nyY%UsqHxkg~*=
zuaXgROiqPhK{ohcsi>SzHWBp@uDdUzoaN#tr}1yCds~*a;h=jo>cSL>i#O_!{b}*r
zLN@epVK1Jx;a~=k|55o*C@<<?MnT?8lkv#&PjT!DY&w(+nT2kMJtcxdQWcHzffjnz
z&oY#VkpFq5pXGXN74iLz+ki3Vy+hC#W+T=3;dJN`d@CD?;J(A>8lftgHy==C=$8Zz
z38rCxJa^iUFfw3!bAF9pDLj7#8cX{RdAlpd-*pai9u9DEc7>sk$lq>TdBiJFM_)_}
z<2OPALSMXbw|58;S+O`ehy?a5SCnNO&3Mp2&&hr&);!H#7i+$dNvYzDe3pn5HRNl+
zo80R$a{AnX+wj4603c}&2)yZ!?|S+AzHmf>Hp>lkC6^u8=$@hcUZ?$rokW#Sb3dN-
zQbqOx=r(zBYlrA%aCy-6T!xv5&E&dI^wGw!mvt*$@7`Am1ATpZrx!3<lZ<#(;OTOn
zq35Gj$(fX+qm;I4Fr>O*@7haG9HUpkmkcw4<~vV;VPI^oZ1ES$1SY3o5s|5u%GZBN
zH?pz|-$wo$8A5qEsqdz8`Q`HX<p}M}1zg|(o2nG5pD#4j2>D}rW3UlwXVaMFeSGE*
zkSKUMS$y#Nu^;Ca)%pDOYo8me^bIwKpVC*@E-}3j;?7G#6()5UtGNaL{@)bVYhU-U
zC$$s6_1VtldhK-%Xo>Tx4pACI-;Q<LirmG?K`h^)(S}vie2U7-5f{trp7RmtQEy;#
zuHmh*?Sy33Zy9wqUv^|gb|E%`N(mAejc_W>HPTz{{&yIEK{;#LPECEOdrz2yV<mES
zWVC>JNAVVTZ=qq})-^I65ES%d_L*(%O$%L#DGc^@^mjz4d<?u4E`{E~S(I9HdJ=sf
z+pIpi?q3q5VwE*^muE)g24P#VV*nSoJ+e9AA|EL(;?FYjY2@5jE?nbhy#{}2X12{L
zA3W?8$8A#!{?bCxspMZ9kCp^r0%vT}Z1qQZk(K-nCCw6>u3*TMGM3SX#1s4{k)whd
zW=yrLpecr}zxzGa=}ju%a5D7tof`ZO_o-$T{CZH-=)g6KU<<nUywuIe0MLJ3k<AYy
z(oNOyIo!5YJi+0x$J=Mv&B4RL!%a`%u+}c}3~hG-Qoy)UzPdfZ_ElSW7EwTI;dPm?
zY<pdJQOeI23C&bQl?#r6lGMaR!la><X{*n)`0QuetIYZ!ui(a4##-3sHu`&E9gS^`
z;fq%efMVPj2U(&;`43It5uV&~<r@h^g)4WjuOf*rxk;P*dCpl+pT}y-5m=kG8!t3P
zwYaOZFMd0B#!{C*flf0mfafn?c7DDp=F0>$5g#6CoXnTJGiNhT>zYdj5zzPxy^84E
z`!FgoY8+SBlfoq3h?Ca(k><x<d#7~js&y%Err5OHD6_q-GyWx@k(V^UbVA8Io58cz
z@dRkTd$6&B?}*sS>#Kf9wEOJY6<Z;1SC~zF6NG3l->4|XR4-F=YTb~(6czvI@e8_p
zUAG`B-@@Q<d^}7XqQ8iY^C_G1ecGK?U6~Hk|M0-(9-T@$y)~Oo?s!ls!ZZ!(mu_tN
z!IHsAwm4lI)}s!TLOGnqENM0UWf*z<Gzh(HGArv_kYrTdCjIn_#u$a9_lY0!Dz1G?
zDhbU8CP|aT+;hTws_$6uF0@fCe*3E;lVLRbT&n#f!Cx)$AYaTOFY0vlPnXE$_li*;
zDIO*;6|!V~n`mXYAmY-5wFwjY`qQ5jTIKNVtO8o`zchh>(Nfs|y7g9~kC>+2Q0o^P
z8%y?IzKDjL?Z4on?r|f~AA_eP_+!Do+Ydu7InLON`oWAx%~1#E|GFQ=!nB|Nn)!Me
z9kscuelmfpsloe0u~z&fr=rr*5oi0(tVC~%9u3@Tzdl~RMO+9gp_BGOQNH|94o=1a
zE5e9d@^L^o#=^CzwY{BNRvQCj6VVgqS|Lj7(Up>y?)>Z-<~-lAmOV9Ybs}ujxwp_2
zPw?mM^%m<D)y-jnB~{|8+AdahgL*ffZJA13#COnfrt{{%R*Ah=;7pHdn7>?Bu6a7d
z=Y+`AgoA69-=XUKwUvHc{v<m6MF`Hd+y?A?GZ1j{>*FgP9v-qHDR{1nvjvnr_W(h&
zVeBet%=a3^(AE7;AXBCO#Ory2OXu@!YHjIPj;VF`M5(Kx(MZl>WSEZ@Csx+KU#1vo
z#BuL_`ObVZSa=n^CukSFRGKyTOQM^=9zjm{H8Bs3V^Dh$q*Y<5l_VAJ;#ijEJYi}E
z++VqeR}9<V{n7g@-QB4{OMLSrDrc@aq;|yP&bcC2H?f2GS!$unLJ3mSR-MhLJN^Wb
zWGWBmo}0psCiD(HEfEt>VTY=f$<c~~60AC$bC@_S=~#~1h;y+g(LIY;5*d4t^O${9
ztFv>|U6Qw~>{UAp$D^X<)4kJC)%O#?G#!HTb!bV-NolZ3ddX<EWv=z)P{vdLbhG&D
z&%K1_Od)FM(WJU;QN0S1+qNBh@gbIZnZW4@py{YH%0Kn<(_U+^Nmt-_=h91g3$FBF
zD_OaE+X#{_y@Zzt0dmU1sr>$W3N^m|Ez1v&bQ$hqJw`nI*0t4Hv<EGm@7y5-V*dj@
zZP-jIHI1}?h+Cfey9tbH6$r;SBvX^qhK{l9swGPA3A!1tBo-8-%hVavF~s5}gO*$B
z?d@oW1}%6vUDDgocrvrh)n9yfaN9Ov3(H#<uih%SJlDy<{&)*So_xIWx!#S)63IpF
z&w4T?Ug=`nGwIMn{KZ|zc^`8DQ7EBNb7fOBxz>+LT>hAxZhDm9<AUqI^PN#kTPOHv
zzq+YRLs^qhed0S~?){<!b=U}b>`4;67EUxxCmX%qlFaGQ@Bb7b`fYGC(mnR4hF)@)
zC`AxlE8WwJpZlrrJu%jGEg@72lB6_n)<et&)P7%z{_Ud0xw+z-EBOGgm_mVK0sd6l
z0a<Z6(9A8m>mbSuaV|-u7ngk6O`@7KajW#7Q9|wjALFn0;@gp*wb++lf|bwsP;$yo
z+JBfsVo9;1e3Z5fKE>}zA0_4*F%o#lBP_`XZMW&J+H&8MSWYyV)kZCubvWwZ<%}m_
ziccwxYK5VygjRl4S?2>;5m{N=^KBR3n+oRFyq3@&O>!m;5#Zx~y$jq-Jc+zKZAah1
z%4CSYS-ci4$p2tv11bZ0={mP<&4QSFilgm3ABdf|XW7XGZk~cHkG#+W$kJ2K_Ag(b
zr94RSf5r^cw+Xh)mat5Q62<-Y6N(=ZX+QjK*0KM*wkc|PYzEkJaY5jMH2H^q56bbT
ziVK&!uDp9^_U_$DNeakacr=YNY-){Iid7iCIemLm+ah$+2b|2C7<&1_Fl(OpX8V+4
zq&TJ;jWk|STRQwo$)#J&R|%#8Y5P+@8+%GJN){m`y66RyzfZ^AcCQxb_o!wK!n&cT
z_r6`f+MWeB86GeBmI;%L8Pnv-^G+~sUKGXV@3=WM(w^#xVSA~Kx)N;2$<1o4!!GxF
z!9{}K#dZoThGQ|#V%Hc$m@6=#p+l`{>Up2&HP?!QSMW3n3q}T@-<f>!L(f-$?eQ$T
zA?;$ZN#o>hdTk?$<P=a1rKf=2#Z32YcMl#Oo*e)yXyY~Sw385F8->>TodsO5&)0Uz
z*=?%eTShrv&%vbd+?<J^rcol_)#uxa=IWWZt)UEJpej|oU47qL%V_WgGVO~1DFqyE
zWhJ;MhhXcneudk#m7K18%`V0$##WANk?H$_?xJPjhBe;0%H#4x^YFqgVos1W8(u1+
zIN<U3>N|HUAGy2uJ-7Lk;-ud=jXHbrBI)c51j$t4uOd7{+D0KnB+t3oMxT3!J9;<@
zc)0YwWtnOVshzbDd@Yy%z9KfjKITpR*SUP#_jWZ-YH=a#e8qwAFsg1@jI%~@qNX8N
zX}xBL)cR4ARK)0LD+mJAAV@B<`^28Ka0MJFd;8BBw5~n?pDZ&?2Zi51gL$}Mx9H-u
z`5oEn{NLR!N*RCoIEgOF6iduT=sNW3=}K&9V)0}}H_4MEk!11ozWtgxPfx2Whg!)I
z^Y-wiv9~~5ecR<i2PteaFywqX(~9Ni$6{?$@Q7#r2b`DR_<3HLir$rb#jfHDAqKb?
z#1a<BV6IPX?u)7<cQf71y+fcv@`s3TRP}EA`SwOr^daW4Hs$toz-q>Qd|1>+u}n>d
z@RTwfI|Ktu^2c+~AV}`BxHn`P$C}<5ITNpS&V9C(v=9a3db~)G5aXJtbsvn4KGgzm
ziAYDzjm_uJVXR+-8J5_8iE*}!ZUN%SOMKcpv1c-GG9EpkIByJ2*yGZDL6-~LBI4Ee
zC1dCE<7j|((VhLIS7)erJj9{LfX1eHnVO6?6dO5EK&ivHuyRDWb;bvGkeoew)FcUU
z#r=M|i!u*iizmCkb73)JDn(f0+g)$E;$4Mv!#BE<G8L(oi&fcDUC0Rjdj@0)+vVrv
z_+sVMM#A{h!B2bDW7DeU3tVlj?9AF~s>S?jsbLhqi#WX13?DD<eVeBYct7e-p?dwA
zT(4qnZ4Hc?a2UI)cQ(l?D$aQ~MYQLOVN}0HZ+T3wM=lB4$wi!l!?l85E&gkl28V3;
zPMeXXdvW01+x$x!O=Jn?0Fw;oeAJ%L8<wch`3~Z+w3Fa)YYjxbR`Hx0+GXWl3F6+p
z`<yHWh6aB8MST61Lz5Skum0Z%AoE6c<>25zucEY+oBeALp(4Vp8h=rzQp}eucJms>
zp>e31PQ@N%Q|@jF&or*A@ucc<lNeKX8l1$D69gF{?pWXeOAI|^n7*pAk}p|dcs^pM
z{Y6hR*8u&BmM2K$lG)%Z&CSp%HZqfh5U6yW$shuU^XI_V0bvX%Y%(i`BmMj~jw}7(
zoGF4fZYXvYp^}qWV@CR_fn>hBJ$tj8gGWr#I#94?ByuU!tmE*tUHIOM`%;ZkDcjt-
zPB+)YJAmESv9&<0;#}jhr-~917r(mVMcPj}R&vG4kZNklj-f~W7mr(yb>uRrw1s8V
zetuM^0v)aF=}Ad_-XZ%xvPU{E_g{aon!!8LyzX0jE;`5ndRRU&QbWpp?VY1NI4WM-
zilQqvS}L`FblJ<e1x>T?y>-+}3)_H`bd$bB`U4sqFCgGcQ&opAS<N@QB%4ar(;0ot
zZoa~Ati>(v66Fi4Zh|EJo0YfoM33&EQ1kO`5#^J%RR<4z@75*9_>_sW1@f+#9xLZQ
zvN||;9dRr~9(QrVS|nbi=HHd;2RDihXkoE%;=50ny{BaUgzOI@fck)MS)9k3<9?ot
z1aLQvtxZ_cDbCll(pb-b>GxU01a>$i!~8Zy{_mStXz4BVVaxU<&N3qWr3C&ApTmoN
znnuT+jO5oFbNkV#!1d;N2&tjm>eoA@x~ao3F4bSq)HWy)=^Dh>vDbShedr$d<7B>c
z4h!aPj;dtMd!99*S1@9D7ZG6@=J)BE2v|NNa-AgIn2JE%Twn3s+OdAW^epe~;M1m~
zXT}z<_1%ponkJeWgSFU`4YfV<3^3y&==i|P?!DURqh{bd{c7W82^elLbedU>H!7;A
zvh~?rP%mI-->I3G^>;hUlLodw1k2^;%lNHdy=JYMpF3~gx%^|+ahx(U)4J0J01n1<
z*lGXzVZ1SdIVwR^{tHINq*6>&6h<@I_<f<v>K4H*ElQ<Fjed=HYP>!ipQp@Z)!-GP
zmp^`Jnn*EirqOcbB2>{KQqggKZz4Wr?`Pqxi+C4L`bJEHtwKNN<W@?R8nl9@qY^`U
z;bldrfeK8t6<AYzc4q&G(aNfK6@W#)qe6&Dh!`P^*yG;vJZ32LL*cY&?mDmKJwdQ@
zt0$_gqN>W5E`71BE$(HB$oCpLc`mZIKaOhfRvd_7-oK~N?M@toDbi~EefK^-U~qx6
zDVHp-LGR)RH=6|<UOsA&J{!U6Xivn}l`lOZU2EN1-w?91VRv&bf|r?C147>9Dv>E+
zFgLrelp{|isPbikDCi-H_J2ff)_pooH%*eLmBaUX6DC;?JNCy{YOT*E(O&TT%rmO_
zvn^o<NyAEAV9uPMnr*ft%GJRkmZ7pzO9OSZyzJ6BUm&SplAbZrg~j64sQ^sa*VZ~s
zo)31mYtJgY^meRGCcdQtTs*`6G0;J-gmg=RFj2-k_ukongK1}du-nnEG<A@BEaxBT
zytV5a?cfmYFv&WYp2iw|`loNSJQ6A~GJdhsKED;X-FXuWQx@~XCSM47h`)Y?70846
zgAT#pZejU$ZNBUS_3|M*lq<{hF%Z9nDYh}$SBWZ*0#Jdib9+E&HAJkzcpXeEgsxYV
zgBHhauj&k#UgFdT0ggA+)G~ug<Pg@GvLmHXR#Q{comrcLk$)LnQ`6IW4cHM{-hO|N
z_&az%iV8gvvJGY$lI7}|fqiU+NxAaZP-=EP{uf9!J(|D81FVV%hoh*Vh=>SFUILjs
zy1_2`(Y3$^ApqE{>$Sa}8Zb`IrYIj>OQ4UWAmR_rD#@ee#m>%5+AoxOr6uYpoKMnP
zW%QMg74st4Y1qknDGWxjLr1!zMw@|yh2`jbmO8q0P-xM}cg=}4E}hg;el~LczW48-
zDzpN0Ldcqp-(mBXRq~29rUoDm1-EqvVk9e@nD43w_`NaoAlAV~RNPhH&5u9e@P`}B
zowB`zK`H9qL71aV$k^YofMMYlzJcp~j{*PfUv<6cG<q7sZy#$uEJRvm8i?Vi@dL<y
zxeBijE$%OPIJ#uik)&ioFaC3{LDp!GmI<8aZQH+>p>841YqGoQ>zRmd`6X@Hw@{s4
zLPBT)Y{dI`ko|XGEIccuHT~c6F?PAfvq=hPgB;v!uc{PXb&XL<un?GP^17%HvXI1M
zjh&6vQ#ZZwrauSB2jZK<@?6ZWK_b{&<s6nx`s@~5bmWms`qA}|o7K05xe5nwMcS<!
zl~^gPHrQ{m|CC};ON|je&o&`)OrV!cccQ8T9C%^$#Y?McB&6h<Vl=809bF65>;!j7
zhm_k+542fiqOX_p?7~7NA2a0eLn(yyP}ILPM;R#Um$={5%=$0ny<?P>apfeY_ARMr
z&T4IaO+FAnE*ONSg5{Cxxf6(M-5>SKS;5~{$HR+@_z0RS1lUC%n^3{PlepA49g+qR
zq_<xeav9A?5o++O!Lt_Kn?^yfp{9|#imOd$*5h<%qtbS3fgS=e48u3dM1s5^Q?jjU
zTG$sFmbide?hAUAm(73v<7U+TKbp?_k;?b~<A=mCvNIwaPD02*cE;%#$Bam}?7jCo
znIU_x!#lI&P-ez4ieu#nA>>eY#<698@6Qk4|KR*`o%_08ujljeNammLkS3o#14qh&
zs2H&XYhTAMjEJE8U82S3peyN!^FVDIL*>}RvDe&}cp}P7aBPZ8frSE<xnbvX{q=;Q
z#TUIp2wDg|A1I^i+WkTeq?oV!l~5~_2gHu_DsnN)Hm|+(21ky~lYr^9wK!zyW}9Nb
zR?ysRwUFa=8vB4W&$5YTH|Z$>i)s3*-f!RTxBT;$ml&Z{f-v{p^!_Y8WA8dPl`VT*
zZg5;dnt%KF^@G<bG4!Z$$?c;vWCqk3AjNyrpe6b27_;n6CYpg>Y!Fz5W2vZNic2Zt
zm()lJ(9l6$8h2$wos(2FjhOqk2*SMh#HJ<|eJ9L^YDRd+>1rWpcV;vqhauu3v!tl?
zx)XSa%vHM=kb|nL$4%9tpr5~f0bFLnxr^M*#c9jXZHW@p>9-(-oq%?rBF|xFhE~Xh
z%dF)sxdm)LahjWISBA=?xg^$tJDZv$8Y~=4959on(mLTQL--qXkNvLZ1INob&`awd
zs(Dp3piNx(hyL|jOQ}jJV-%ys8k8z<FlYj%bbacYs(8F^E%rJ1;1~ct`x&Ir24rWb
zO>8`?vO)!)t|3S_n0U>Mj8ygiN<PF*80LSF&!xL74`h=rr&6!?)M;MiOiFj2eo*>1
zma248@i5$UPt(AFnglvr`H9nnBMZ0e41IbTA;)>lz#f@yLVx-b{xQn>J28DoX4XOc
z?SQ~-1~CNRZE{!a0ij7}VwClOVE(DgELX=<G#_R5f(LHaGDeZ|*11}HQ@8lh)ERw3
zSr#0`aQ@E5&Wqq{r;_FLG7FkTQMNaQ!)E%b510B#iK)B$<v1nfSd?Y-_RHMW`+1I>
z^pV@5ZGVeS!L>sTUCM6_l&6_ki^#r1P%Ju_er{&R)46+<x8_zGTb_;e`PPa0WtFmF
zGf|q9h!jzMMj3JP#1wk>M{mIgA6(T_@g<=r>SpJejT01%WLp-9UJ$JcOhRl9P%<1I
z+11@W%o;N@3r$e-u$`NQYN;61O^0t{rtND)q@#O$2-^`Gr=Ges=T~|tO@Ojd>E>?4
zC!0b7AteJ9Oa_vrAtFtb4?*<dU|{~i8ww^FfC%oU+<O^M6-DR29Pcz&shrmJIq33=
zNlOWJy_&0pLS>LYCBZoIzy<w_AJ!VQaA2e@a>v6alKkWCWm9V3MQ<MEH1!N6r)6o&
zv%kK)L<{S4$@oap7l`fnk^Acih2V0sGxnKe^}&%zKFum?rmi;`NO_UAj47zM3&dD@
zwwatqYKlMSZQ*3HcW+ZuKY{N<1o+4(RhZ#mhA4U(Q3R`PTGYLpD!oNKk>Z-0HE!V_
zNunA$bs)NGp}-9sFH7!6pw5iXcR>u5zKf4XUvpozUq7#ZA>X0tY}4^MbkiZj&Y*g4
znP<!b^DUOz6Zp#4PmD<ZWkRW^xDh>4)hHIu2`4r*Kw98)$cil)%vFZDR+cT<c#76U
zKACe|QY5k3eG4cZPEQj<h?&rQZT{#UWh5o|y?&&LCidfJAEk{++p6$)#Q8HYNfuCz
z7bHUfMD`6!lvt6KMuTqgNro`SU7nNH`1x~(mfCU2=m9mMvV=4g$>-Wl$x(+eQuieH
zxedRN%d?~F<<u*1-t|5ZppK;AmhoMsDUD~i3g^MP&(#Z006&En$>=&49X&P-)K>4K
zF6wHU82%lUTwl0)c64^NppJi!{_q!xSr2H(%R+%zuDN;9<Yn?U1~5WBc}+VauIg_g
zyB0F<F(H}li+V#%y@Z`NF?{Phmp@}b^@z!9XJ@AuJk|=ZXdyZi^ihFLQ$bgk2cwGT
zD+ivPo8jy;)V<(%cug&)!h{IaAJtrSo=!KpbMWvwW{>MkvXwnF54!$7``fYP@=t2S
z#WQ$8AqSUih@l{`;0eT@;2#AS(2z%CFql$-B40&R+hkw%m{a?^P5Bd#06DDH<<UaW
z`F0Djo!HrR+D-(fwJQ69FB2*6-RNVh=(RDGcv7&jSjq$4?@y`adK@Adqx_1IGCg?;
zw=S%8pqu01oB>rEtgdFGELR&R0hLSQ1nxQZHa1QZzlp>w;?Cvqt!^@P89=`qeE|8l
z7>@!)W-}XTF7lAEQ8E_ACzIV5C^xmH8EYa)kwSEmU@jg%1MwU^U!>jVB#k)2-8S3u
zv|+fmcM)gfAC!VmN6^Tx)rBM688h^D)s0e`<)(Hf8Gdp|!uApQyyZh;KyE5m$QE@;
zi51($Rv$6^aOwkfD;VHR!h&cmkr^>DD1GD*(|a+2UFO+l;g+5G`I-3)({_;a6a$>Z
zmZtmLpIlR|RVVj@_rm$iJvt`^*ju+dC~n7ewnj{!=2dNse^DUm<w5e_-$5;2z2-jM
z{oBuN*~uIGZ9<#MfHAX|9tpoGjHRA>M1O;1^j*gQ!)=m~WsoS37TE|heUUPU8|p6B
z)(RB&3g3_o=YKCcH%@-?yJ-$$*7<4l(MsA#lX#u1Nb7`!ywHA%FGKcD9cLv0{{8EJ
zZO(zPO_74~HSCC7FiOTs`~FCfB*r}O<DUwZGA!Ky7(z?&cyTdYi<jI}CD2e&j9;WI
zK98{yOvP?HVOSRJ(m0u8dr!5%5oJ*5VsL{;BZm^QA0)23My``o(ZrA3@95VgCwDAM
zqk1aUW1NX>=(42?2yg2Dq@1Mxj>Ye`)8EpqZ+De0{~J>ytxRZkf1;h?*CO$hCw%_-
zN}lY6<MnBmLStLyD}iw@XP8e?GA3~qP0gyObnz7NUdMQlS)=0$;F*N}g`qI9j2q!}
ziuClE9&1CBfG_O4wOXjo;*^a{lv{2RZDtTWd&c9}_A#vdT@f{B=}G6Wj=+gL=%_q^
zRvMn2sw_{7j6Liw@5T2>5=WhvhQ{Xi@L0sU<i;{O0`>Xh=Mblj@9D8kWz?@zL~8Gv
zm|JX)(%tA^yi8c?r#n7*g94DvLdLfbqCM&gC2Wo4SzlCEHe=TE>`h?k*|v-GuJd+J
z)cG94OH*m9Yut28ON&*+sXYh`adLOKtwdJ5=gsrQ&TrA;fNtS@r{rR<>rQf&{bVnQ
zv{$<1I*$Jc$Ke^qdpvQ6;r2_j?B?3ihMwl>I*X%P3)7;y4_}&(P(4?cjUqDg0~4_-
z#6O|LG<~k_Qw?59+608Mqfe9kEK8gp|7~v=;Y264M-$vBTJHL+d)g{|W$$5_$H(Sz
z86xtpq|#;g6M=6#XWJ1uPLTNWEV;PvS+>Ty=CGR!jB1#bBM##!X;0L)nU(t!nS<RY
znQfgHW0T;|9nEZ|D2W8}<pGj3R+hC18qkf>7sXakR-hY-#bCBuuXfKLhP6zm2kuxR
z*Ay#$+`7@%tvH+XK2IuXj6#04PA#-=k4b&Dc~<GCZ|%)512s*Fiq$-gEXdNkZ$eB%
zrx_{9JxsL>1R=Ig+68IKpxqv7`}(U_78B1^jf}8;x92|{p_aR6St`7A=BeKL8BDr!
z@G(#!h{7s%b}GV1B!nJ2Wl*}WxUPVClMv{BDAla<=&?qFhld=f-W}869xQe>qos5{
z1~@zZj;UK-owiC;rC>zp816-uK)Ut8P~P`h2FRQ;`l{)i^RK_BaI3>gVSC5wtY>mQ
zjRsWj%b5|Vf$o<M^jTz*Po$O8Xo|P}%(rV7IFf0jQW1>ZeP9@sW0_PXCByH`l0OqB
z;;Vb>%Ffx~{K}bi>k?3H#9_~qRD~D1{rKSBz2$shB!uqtKM!yZ&!MxJ07LR%<ZCd}
zk5dJh7@qjk+)sLH<@w$;ubzBVt3<>kF)-LF(}Htbg~^z~p02DN*z4uY+(}_FKC(sY
zM%LU5VKO_EcGk;E`%TjNH|G>(1nZrh`F4PT7F<;plOO3_bigV5v)O;ex>h1@(-U)$
zbf5|cL!a)W0~anculG}pGDw5s8{6Aao`DCEuergnxNhnPZ=7x9)XaKEAD=GcaI0sl
zaX;#7ILNaMq}=aOVs4E_oPID?><C*rd-VRGJN66eT~it(i5QOU`-2G#hkVz^O2f6c
zuSZ2|gu}1TQK5g=eF%4t4~n49mkEX1uG=Qd^Ms3w3!2m5y+3_xCD%T4Q?F?;QaZkX
zaHF3q_;7KN2bYJpG^z6UXjUQ2%{}R5?L3*ULtPd^V%Tiwnx|$P97mho&;Pu6sJK7C
zhs#5;oBPK7<Vfw;6oB+vT~1m(KKbR_KuBq7u65-V8m^z)rX`OCK*R@IGYy*WtE=ID
z&INL7SuAy(XD;`N3TFd+?d|Q5nCR#m82=;~?nRE#oYe5}@N}&ume_e-SqAexggEiZ
zV+p_~w<bC!UQrE5`Hpm#m=wQWd?tA7JBrRK?Dy$f*X-~<ol@Jyk7p%a`(rC@fKJ$y
z8r0w^Le|_|TP+0pPNb`DV@aTbn78`G+U)}u<c+U%O6w;S4V^g#)eFSYRi=iXCil6O
zm250#`||P<s$)U(b){*~PGNx?G;v?LIUYD%I-nGN@Y@P@X;s3+x3{$&KfZ6H@|tUh
z>$?|ila$NLUEmzDL7o<KAC*;J7-t`)l(kGbLvIV5EzXccEAy7q+akW3#C^kua2Gw*
zgp$vxl`hg3wCy-eYzJ&M?*W;5lCkplbYrs8!_`h`9x~!x-rhz2CF~eJ4%vP#zsU*r
z{C-4|_hjOEWc2u1gPwxEao?!~oha6S=V(+h;CPH8mD}>KUcl6wvEm$?m(H@A0saMV
zbVtjDHco0z)(Bw-+1ytjXHAH?x5UUEplFY)AmpGk`|vBjUOtdAO~LXnF9A@Gy{d&p
zSAfYt&^i9@!cV;Qzx@7U;7!D(Q;96_{_AGO0O28t_7}J<6bYq|%wI~fSK?TCq_Y(D
zNZ5=zCcPW-wi^s1B}1^pP~Cz}`X|G{5M`g@z>ClE0y`MbhqvN*Elf=Ena1@{Rhezh
z7Jk5ac%NM`fA?#=iObK{N+E@cC=N4Yz-tq1Z|}HdDN8_`nR=iXxOEH!6Wdsjfd)T2
zx(u4L*4UlO+NfjJX+0j;1G)c?uarq+L5QKYR-kp=%SM8<f8WyVObxLwaDLCsR~47I
zC-bklw#rofo?u+FsJb&jOB?m&m5+>{``uT3=gp$uy*c2&pWeT;tkZ&j3mH+>@71`x
zqUqbIVCMXcxo=h&h%XA;{hpppM@Tg?$+kC{_HW4q0%cP6o-!H@<?sKpo$mM1J*TbX
zH<^;m<S_R9)%r<PRZOi?t}A$H$E*(f2(8O%eCAB}M8J=TNa{2+V7*a90(PwDYA>_D
zYyV=9PVsXQ-6~w(AHNbzXvBElp&Ppaosa!}F+0XYS0v}Via!%kJv^T4lJUL%x)pB5
zip*fMEjM&75KHK>#+w%M56S}+nhRg6A?>Ib#!rRzdWu)QXY=jRJ^xK_j*#c+Or&K1
z!z6Ose~Zy{zyvD<i{iRbQ&-0{Xrii>vFuxN{!8m2R#IB}<lj;%;60HXx&Nh!4e;J&
zpSJdSw6wRjCO{j6KUBL!L_~<^f8(Ojdwld`v#u)wh;7Bj-V}*}RXYD|A%x_{<BJp@
zbEP^O+6k@z0K|uS$auD>Kw1XYwN>ez+7g9epYU)C4n8tBl;?b*PLizJrnw6<g9Z}V
zBa(@hP2)5KhpV^j2VLtrFCL7BZ!1PgkD|4*>3fEu=<peKgofsOKSzH2{DKmXiu*GX
zWsB+H;*;~snP|u0;8VgGkPq|ZRtj0hkIeP-3y)A-zBHQFnXofq0SL{aK<W!|BPKcd
zRz;F~cm^tfF*985eIZapxSpGv=eH63;b(e6AqX!@ax3WG-@h1pqJC4&Dm5JgO6}%+
z%tKX?1xRi_Vp>L$;3C-!&<bVVG)*e^@s1CEvyDwz?~LGZ3I|)}CkRMk8^jtyhGgf}
zstCmrv|pDusW3tf<9nKN@{4@&Eo84a(Si0xsMqnA9-fU2d@?wCiwWT04Y7!~{#jB8
zJVj)^_k*74Fk(`;0Dg(@C~ey*VG8TeF|B=U{){pM&d;T)?1t{z+XISQVKC<NXtE@m
z)puE0F&W(gkl3f13~%Y76}*fz692J-5ZpJw`qG^OX3pRKQ{qgn94iYHJ^MFkN$-X7
z_g|UoI^wuKymwB#?!Qh*yZJSLso!{Ym8qf}u~`txv8ki4j_WVJ=rD9}3}j6QUTm&{
z_FyiM1}Vc^>nPCJO*85q172Q%d*x6BEzHMm4J&0Tl&~KC0qzy^N&qRKLf=8x<C8>?
zC!>WmPu-9p@9Gfqd)rC8V!}MU--lA+FinYvXQXM^u=D{EX?w(YFl%fMze+|!7d_Qv
zGv<5#E#3{9H<j5)e(hqvRm8Ypb%0nGeYSY8v6`9N!doYIi|KO;54leh=?@b_#Ba$&
zi~7z<WxNnT#_LR&&Uw*jJ~)m9&9;ea3PF^kvJAimLJTo*C-S5u9(yuKIf}t9af+2i
z!71DCzR~37t+04_K2*nXA^`MMvv0uP2-4%E&>D97b<?=cy{2y3eW}z+-p?t*<M&s(
zT4wvbAFL%pn?jE$O)lC|=ks+}vj@Ewxv0JCjuS#CKz8#icz(}0<CdsZo6qLiEs&zm
z1;Qa??4PykCgI}Of9|F!DkxX)O&sPjdh7|^V*_R*3l`_%)c%RzUO!Fg;71`^uIhT+
z+64>e_`r_3T64uN@3qyiF4i!oaC1{jL!5vuJjr0gQrbA9k&!zL=;(Q0eV=v7cvmUf
zj{X(;;O2424Md0k=JT!v$JOd;{ihkwhIgZ_zH1Si(x`^o!uskOUlcN9he!mVe^{&z
z0?6}s1s|*ey_xB$shXy#+3IQn>bm3k;7o&_%vz$&IQJplu{uZ@;aq_Cn3Gvc;Z}+`
z*_mg5Sy?><@C#d~H-R3}jis@?e8)m>SNEz3hWHpmql&EYR;hn%VLyNVdLbXSZmm$8
zh%68gj0ewE;u}!RJmO-!Y5!FMn=ci_su+8!YzUf-<Im*@y0s<3n8D>O?*(HBQ<=e#
z?1$tG14c>*53>i!LkGem_0IY$CMQd&bu_QPM$~m-v5(dR>1b(ftclU|z<mP%5iN9t
zH;6tUQ|6_ICM$%WX&(y5{G9#8h8OvDJ-}3gPf;YJV9OPfCWDYf=W0uniJ?Y$(ozX#
z-24rc=i1gH*)JV<8S;Djp_2#!a8w{E1oH3SmqPkKAG@B)aN)C?Zsz;1vC>w|8*b3V
zT8f=1{kv2WP|uV!a(~<zdQfcW;(hea)r=hcUt<0HFIV^;r$ArNHA0+Ql?iDABbgkS
zhg6W)i(bMV!^l7?&Lc+3B!=?MLI9AT(oX@Mh1rkv=HlYn`FMN~U-Z1fB?gYG04SZ7
zyMCj;g<gNv^i&*H+Y`bSv%Pr&@8(n26|LnY%+BsYY`lh=2#?5kvK&%Ok%m$1;fFMl
z5PArRTICxIryP^;T&EHIQk;-E1oAz66ceu;dB?*wUm!lE8%f29GA9H6-I6{;-$;ew
z0ISoQbj0>4(jt_DAuZj;j)yA!=139}8l#+LFciCD`7F0WEK(9_ZZ0P^+u?p!nDen{
zQvyc|<<~Dm;+Yu+BDPs<{_%^Z3pvd~rb)rG351lBlg3wVMR~o;2h_Fh1<mz3d=0&?
zrC>9E{@7%^YZEO%RjTALsvf?OzE$BoJJxv%k&Z|sjZa8d`O5}2fhlJ#w~Cq;9!ccq
zoqFU(+Aez((5|i~*4xn<+=GR<VzU_a1H{eEU$_n*V|+dJ0$WO7v04WGPJhiEx|A$k
z9D3DSFVbWH^O|d%uCu&mb-d~eyk6+)4A}cS^p~p2P3~(Gis?i|w04V#&4D`Gl&1}*
z+unya8Xz1}2tu?8--AnU{|4&ob+vcpe4K!0ylYaySox_K@i?_4!YkbOS{|5#b88=a
zE?jsW;QNQ8%}eR|r;(lQ?b!PI(k61JEOBH;FR;G$!57Q=&W5&NAO8(#2Y3on*LJHB
zT??B>qA2svADf$->)RYK+X3ys!3y1|$c;D&L<M?!*3E6-hd<Z3Oa1B5m;Ya?!(W)1
z&d$vh-dlMCWaL-B`N?7ZHpbKW)ElRa>mtmJCxHR;`)sx4*3;JdnsG^V?4RNKBp5%k
zzP<fH%nJ((3oENG|4>iKJHS~fmh*Q(O$nlpb}&uO1s?A1_YA(-u`TVfC44AlF#xhH
zHm2n!{9<axqhAVeY(%Uw2*jIVaD0Bb_f8;@*GEiqL^opPu)FCIgOPn%OcvG4)ftCk
zorG)bR^R!5P0w-k>D0f6t*-VTJnY)rY8VCZ{t0>e3U!4ES^LzZ5x)Ux?8pHsOomga
z;tWezoeynxZ^v3zO`K)}!?8DA7auI;!uAL6%LmKVY6>deTXhFZs8}(RrvH?cXUY@#
zcJ$<MZf=hHE_64h=BA3v-*wWmS6=65Knj8fhNYJ!70P7D;IZXM5>STnOCV;OZieCn
z^atW&44xa^%fcV*d>MDi44`!g*zqQVQ1Z6u+yKGjp~#|J>5v!q87iK3(_<+_8kv(w
z!V0OH!q;4!z8r0~tdV2!UZ?3*#V^6qwN|>?ogE=?{#Fg2%y#(}Pq&5kruL?{*#m7i
zBpm-%yA;L7ZI12ODG)k*wQ@)0$)uni2bUYnXS!7_R?IxRJ7>7r@wZK7u5YXtuGgbZ
zbT@aN&?#}|3M!C^qG}j>RX!Kokn|UbeNs-(*Ig4O5eeZHF#?0w0-KMI1JqO}|AAsc
zrtA^Qyv)#u1bWE(CVB&*?9I1t${Q?m_`iiYIQs_%y46F#ts&y9;*Cc!JftDy(?3+1
zlsSx8mS)cpL$BCwH*LMC9%8~Y2&+A8a&lE$UNn0<Xr`IjfpR}`vKuaq<l%{NF>oH#
z?3EEMOEWRle3n#NE0ji>#`Xy8qu5MqqD&YpA1@cft<OsR*_st&psHvxh>J-p2nSE~
z58>EOdj};*pNZUtyk1O(0cAPah<&GOe;6i)VGYP~d>{L|6>J;cE_CPKvc%})@QpLv
zwmgSAT}fg(3i;=mRoLIR>WZOz=n~&u32F|ZjRzol(~UOHKsoFmYvQ8y#qzTf`IA*y
zJ?I4C*Do7H$iGkhDO^fE8yAJ#T_{CZR=znn-vKK;OySEeX8Aejcc;Y3%H6M=S<FiE
zCuh^0$B*CuB^7G27J{`T!^@ps9Y+-N_ne>Wmdp}RgwBAYvZedFTW*B&p03c}bP?Bd
zJRyuBo{qWYY&5U&QkgbX-Qr`SKfkSKz_t(1v4m_oQQiA|)>jt|RtSnE^hif$a-x2<
zgle7!tXNx{ut!L(QZ3-`3GNMA$bWYIZ=zj21wB~bD9llo-J1h%lApQnJ~X>(vGgHC
zym$w-1ElIL;Lj~4$WoxBgC51(EVfD7poHgy^O+4JN@vWy(zaS;I%#ygNsg4!N-1h*
zWAu8FLh0WzP|+*nzSPiUkJ3q|*<!bGF`W4kT*br}#2x#i*djiwa^|3Gzb_ZHzl0gV
zM2`_zi?6<1UG-r69Wy%ev=`cg2A5B<7?e&Xa3M^ctIj9Er*O*dA5Q39dtUc<jj2sd
z?oeOvEJSQPSTNrjRq53ev`mn25x}v<A{L#@7B=j#mRuHjW_izKDBZJSy7e{0aWQOg
zn!d)6YmD?73*bqaI*X1pNYwBGg9bmd-PU=uPB&pl`YKDU)FnC)LktL2*|70L(PKvC
zZ;ah-o`mAMH6Ud7pnyvo-4IV7W~k}23yX6q8-Lf7(nHNiLm>0`0P|G{+xltOSV9+f
zo;l9z^re}T=tkJ%h1|1%xtmsH?Chr-3b6CGb;rmyO*#dJdfKj8+dh_75qwx`tm)r?
zpf{ZcsUz6C8yp4G(ij-Pv9C}jQW3`!B=B7i@=&Y1O}&E0jvVjBe0caB$VouEg5Dk=
zP?oqHve2g9DmT|-;n9Z!Pm7L5rT0I-&VKp|+^$lNf3~ynrN4R?^+rGXjqts*?)8pw
z#x%yv5T^%orW{RE5-UX~itzYuoo7L*ruH%Q_Hm10CIyB!V%bC(NNgu^pDL5la?E45
z$NakdA2A6Gk|(7P0qAA~v?+Zk51Q|u8>~zwNhbf30L}fROa@2MR+fG3*VRpA(GTLd
zDM(HV=g4#DaoKo=Z`={437Jyy6Wj8S*e+c79PS$aQO4TGpFyteYa}7;WZSXx=s|SP
zMLJ)<V%uR4Fm;+*b&<GR_P{gz(9-+k@EtDHYQwPB>skx(Or$M_l@#sD7oBCgIpc1r
zFKra)af$)ZQ@*I{ONz4k1+%U4nf?}!>to{e)wY>2k<eM}n82f!F>7Za<dCMR%bzLX
zs7u`0*0>ipto5*C_Y0Gr0_^JzNuBJ~8JZ+-03L5ymz(bvP}VfJUFmFUxblz??5Mzn
zTH3VoRD21g;m=HXI^K+-FQx6grNUE=6c-dCn<6a@oI*)Kci-PvO67br`0kEQt4<Qb
zy!MnoCiVqq!1>p=CmtwN2{^G00+IIcT*C>x($?}c+ngnTL|JjpjE!C_qOMqi5-X-C
zK-}<$5_i!lnt^A_gT_|D8~Ig=fxo}B)00FdfNkQx5zZ^!$AyOvI18JrtMkX7gZBU-
zS8?dU5lUD3yq$P8GNGpw4g~8RXkYR-B-_g#y>i-i(_F4OA@tFs4yI8{0LyaUeQu;7
ze&;ff5Vknh03XV9<odVxC3djUuNGZcI7c897jo=uZ{K?Tl>)fl148f7tF2ML42m_l
zlWo#;t(E-cVwuGBbk%M~dDf34h6Ekn^g(5m_`p+08s3=hF8emYnp~yZjU=B@Rll6`
znV=AOvqh2c^Vk0_3tlqPc_q1O$o#gateN_c3d)fg0qsNU&7w?b{)Ec!yl3B^Q%pEJ
zMUNXe7wBreO_H2-x&>2PQ1hHy4DViS6)j}@AgtGj^e?5%taN{xrYj&7v9z<YR5gmi
zKb)7YR!HS%P%RL8e>jE5A72R|lpV;(Xfw!2HOhEnq-L8xb-FWAqRHBW`86ISnrddp
zvvpV>r$isn!$@%hJiH7L5(qU3eJPX%<ZmLAiYGy;4qT?NOZ!(EM&9Pj;OQMm>OOSN
z4p}%;L3zwcaav2xba`)|?d*7b#GrJ2F^?z<bY8>38DtXY)@9OcW&dJ&U-!%^Oup9-
zY`(8!LiG|3))WNZG1f-e6-}IB%)RAg8T)uSZdEuIZ@$aIqvmmok+IR;sZgjM5r1%S
z!~I9kRfI{_c-R+GX81RLV~Lu=v>~>#rs5IX+2(8mN&`5!qR|)s`uT_6DU{BpR{1N!
zbeV<Htd-mk5&oMpb4I+gmHFC?HmdI$bgh2~>15O>{47a1o<m;>!_*!`nrmxj43ONT
zw)ZQj1j1(Swrg>|(Y?Lj^)i@a+}QjAKe@<cqV-=!ew;J29{*xzzkL9O%?36!x>1|W
z32!H<G4b?q&NUy6IS+v;XHXpIlzyhP^6bBUKGA1gXP#G^*GH(UX=nTz9@|vkcKL&~
zn4Xm3w!{zr@HIR}>7q;t!Ok6Vcy&SuGkV&wi<6xxi5A$YT@ap){`$Z*TE$9(S+hVg
z+w=y=*AN<STQ#lv=|dy!7FW<><Qvw9K*VYi_&6SgpxXnRuMYLDQGOlmzS~2pM6I%H
zo>+~kkTQnI*$^Sh1BJ{Of0K@Xmx0_7>tlfW2cz?hsY9ev7b%NQYGiE2%y<{CO>8hT
z4A(;;K!Dr-cxy&-B*D%?8ozCC!UA_Hh|>t7*U-@J3;J77xqu9t``7JnQH!bVJf0oy
ztoF@BY=${eL|>sU_BG=_wtmXDL-$yF0W^}<mtDetr$@1Gn_920&W%~=EL#5^+d7V_
z<=w5lv{Nw9==8|eOxDO@J=i1Q!)iYazH{5;b&&f_0c=*IwboeA<^is)lF#wh(Fg8F
zeaw8JYp?7qrcDZv6?v^SO8Ja>RTEa%M^;?{+Xe1-5HTpW1-ZbTy?*Wm!u;KB;*CMw
zewjHQ{iO6@7OqM4xUZv1m$6-!g<Ux@YDiL#Ay@^vWg%>-b*<}YtY6xc6~H)@<KxDh
ze13mX#T)_g`|RlGXy9>uZ@3<5bPrkI@Yqt`=d<u%h0%od33eQS)GFcP+@eMrP*uF3
z>PtBfN3}NBGb+^7*4#91%(O_n{mI^%KWd<N%Z;PRmmsaf)KMtA{Fp+SK8g|3NDup0
zeT=yk+%mE4SLC|sW6Jt0!wz%Mqu#8h#S8$M1<JEDsVXX%4KzfupKIPXr53C=ys!3p
z5##>~tQ^Az&=D5e13k@i{Fl@w-qXExnY|nf0$wYr8b{~RS6+Dztx|C@WaZG2M%}D&
zs<$#f=jJ}6-=(psurY~DAT$%ywCq;bA#o;#aq;oWU=k1%sd-Z)y+>~CiIe*~7gqhu
zDUukv7n7aNd}5>$3?rPq4;~D6UCJ+9RXv=!K1E%sqFzQQpvj5Ro`C_)l`C2KxyH}w
zQN0&m#tBY#W^dxvTrtcLHrtpqc$#4bs3N9~^X`^A#T;UdFZv_`mk;bECN}PJCUvC7
zM}ZG?_3oWzey4YSI{_Y2Z*C^1jfeIA7o+lJz{9z-y`|xkci>X%LsSA8DLrrLQ#oY$
z_$+i5s_81I^s*aY-St${^$E26JBUpdC1-5wcE4;ySbwGGIUD!qYJAASd8W^I?ic_4
zY;37BU$vf4U`?Q87?)IWxCfSrqSLXMHp&*jl<zZ?|Immcql|=zH$0t{lzIoH7Jv;-
zT-l6U4Ya>8GqbfnLla7o=dTi;n`EqL&|{lMm`RRl`O&Q_!o;*JjncWz82Uy76vjPy
zs^Nuxf&HuUEN6_!@O$vt876gy(2i|rxkfAfOOY~t^P~Cd;`F4sS-Lp*n7ub`)9!?z
z=2WoPbM-k-oMLkmtj`gC_9OiA_wGe&APZ|wqO+`0m!y(3F}iM3PCOTQG}-miiK6W0
z#H73YR=-^WqptIjg){WP)p#t*D(P5CJ|PEQjx7DL#iR-B6b!&BpT)YKP9J%Rb$$N)
zebXNs2%@)j`m2_<zI@mE-H-v*yrfPB6yFL(rp3psABDidDK$+Vvn=%GPYnjzs_&Rp
zX_mUTib#kMLbe?juFr`q6E^PlD}P_-(mou~s#NJ(Xl&xqDEA{|tEQv)`Sn(C=)zw)
z?y%0#&9h`z<CYdFF8T9y{Mp^wN^YwNl=sV*;kF0(is>rp1)^6&o0vMBxU;q5ZRkGV
zaoQsv(3aZsm#{mK{uK}(VGtQDB7(t}t0R&+?ZH)_w(Sg@%AU}~ETcUO0~GE&z+@W$
z^)+d$U|?m0eH=;u1hu~0nqlMW`(V>=m6hccyNr;#L$b$OSW=PWsl=$lL0QSg3!X;*
zTOhRrVI5ux3Cs3?r62Dl{EbvuYy4G#Xlg$66i>V)0P-n(iBi|cj~~0b7B)^Agv~5X
zF)m7ii4@$yPTm{etC;H$AUhRi8>l3h3Ng3sk9B3r|FLuzC4$Wwn-+=A)?`!$!WtS0
zXb+(zU5*9A6|}sxh1b{Eikj(>@ub_qfYUzxCQ^q?RhD)vGfRA%r$6b?Pp&wHO2;n4
zu+ZEY^CO9gJ?f;P{7nNbX*rz9$70%m3e!YKl1CZKSdJJ-5<Y3^$XoUOpp&0b-T|m}
zOm*tDN7_)`a2y-f7ZL<@DF|5Dd8C7FXxMP)lYoMIBS8*t(-v!Z=zEJF`;w`Fc%XtS
z!s(=>43ti*ddk*7bllEbBh$~-vrAahE&<#?p$Pisq**eQ2;6|K{#<M)C*ztIU3eL6
z%`HtthYivo2;!yi?cKpX{Ze`bFEIWSod7}>br(owHTTHUu9ubAQx^I5r`$C&u?SsG
zSFdT3c%z&V7%SN+xLF^N(iT0bza%x~Np#aAA>q?3rLP&>?~AP{bM?JmrC6}AkSl=Z
z8@^+_SCJuPB=>gatJx6y59ff-c{i8RoX9^=jj7T`F%(6=U_CDsq<e!6zahh}m#aE%
zg_4`y&O_U{z(Hw7&Z-$Q#Aju0vVH5bmHfSp6A(yc=2)XgTr^kAdN_2KM*}>zeYCbl
z+k)|p%+`&JPvXynePBZr0D<G$A5@?$>v(sk8@|H<jHUBv=F?yB2}#U}?;9;ZRcJgV
zCwP+*7?A6F#TmR5jLmR%QSJtonHQ)NrGH&(t8t<Ox>Z$K)m82p<p9lL7MM-9jBB4y
z`M0C1ichnNSKC$%Qtu7X3Jd71wP3|*tFS#FE?siPdR;%U=|SA4U{RecXuKiJ_ch1>
zM7j1P2{v56Qt|b^-hvXBUXp&Gv?o;}K{qEZda+a+%%h2P&dIg6w}2jlQ2({f`-sOa
zva;kj-cVy)FG}8O`c7z8K}M!I#m1m(o*7LhDO+1`9<Is!!4w7p;cFl)*1zPO(<*tZ
z*&XA4FIaFM4P-jkqk){|X4#$-s&31lT>bg;3#S&OD#I-o`?JTpFJC%)8!{7EE)^IC
z&<MRPzrfJVGCC#g@ntQPRm9HJcDwOt*Y;l<F*Y}X>5gcL-|3&@9egs~LyM5}U)Q6v
zp^pcHEdc{&$FYR-FzeHgP;drl9bv$uip(;N`<vq|fz42x|FBWKJ|CYj^Yg_4om($q
z8;5ZFX1Nv6dwl$@@rR{>8G9n5AcG{p9f;ahzFkdalx^g!bTEr*U_~A&ef(FuUn|}<
zq_7ZNa!|A|QLrH0LhPRNv_hB(2nGhKV{HzGQy=^8eYt5yv*{EwBqP>0kY6oaJ#%_V
zEBP~>KFa&KvA;>q#O(#3zVKmG={l@orbrvNh%p#qXM}6u`Bm*LWIopWY~qruYpb5k
zusH*zi2Ga$rx&P$QNE`+PAuUTIs9_PYufzGG(l!G&D^2ucAk#Jt(j?t)8Aj!i^Gor
z-M8gxQHv>N(5}9tqoZkhc>EXpQ}M{^EkC*VKxCP^9Y8n{Gt_XZxlaR6#9|3ot>K-7
zgO|`Ca8j7N%^`AjF-adRr%$6XkmnHA`@UKS#;zeqg37a@GNFOclf7qSMJP+%QHC=s
z|1vZxZpqT!tftCV>4gra?IB@Qoob-{9G8NE9G)+YhLF-MA+=Jl!8^7x`lOf2_4&_9
zY`#3<eHl<Ww=Vjw@og5P0V#+C9#1N3q6LDOI)-~Xx=Q{u8nZ#LAie*yM(;t%95p7|
zM$<?rQ2*Xl-ry0B8(?79xlu(zsjT^%w;oI|q5dYzZJsp20h;K)KJ1%k$B7CHuRHiI
zL`IkW<|6mO12azDs`lo9&;kRTY@l5gDg{OkQ*+d>g8lOm=SqbqscF=)BS?uiCc=P4
zOGHHH+%`Xw5fVgv<6~(2Xeo(DOr!LHyL~*WqGx(L|F|<-Jz6$md;I)^v}Ye1?@hs<
zzsyf6tUA_<LjNSB@`-`0wJVWN1S+y@l634jA!N2Rl?k?RF<jVgvBDGCSY|3-map#?
zTq{D>j;sF^Y}U*a(+SM^9+=vi-IHN%Xc)#PCk!hv1Wp;h#-g7W#8e;l+}{qUMl{T!
z&fZP6$W6^E_sFuscUk@-Z}Upd-u+`x4Xyvrj`S6CDI<+Z-u`%*bP|m0&$ky!ZNaWa
zek>Us=^)R5Hu!t&*i!(rJcDdtPRq5wEu)wDjht7RD8e(<yoGO+M+Zz9Ja}+oBB8Tj
zt*TGx3_tJgxd96SztZLeRCBCJr(@J8r8eU3*fc?ADeC&$Mpa4B^84#;)HR*Idj^y}
zwYpS**mbRfc*h^TvQnWD5g6D~J6tN9GLLfcZ~#CAp~rWBE)a<qhq+2K(n`ZX%}3T4
zLe^{TZk@ys{NK|Y|MvM$Q!rfY6qfKjNi3g+j%zj*O!&&oAa^s7fq@hlNTVDYWbHR-
z;=a}kv*xbm=CF!}2dw}Ax|7F2O2JmSeETfe@DucOi|71dA6qnW`#3IJKJ2i^QLp4`
zC-5rsLEOdf%;6KIzt|;n4x%y9tz|a2eJ)>$m->xFVN0Dw*!AVr$A8In;lNK1T$)PG
z$%*S7*)#@5BQvM5CZj*DP-Q12r~O^eMf2t-<-Z>Ft?*qc7Pki>TUa{(nJoleHchj(
zjr3TSj08L~HXTBEIf}r7H}<p7xx+pNws&h70n+&zUm>lo(`3(xHTi}7J86|B*-e;&
zU@ak76Gg#n*9oD0;cSW+e$u`m!v9D)X!-4Hd31h>!ZjdzVJ}ig#pLIEWb>E(wiXvH
z*sh%_fCC7C>^WE0(Xjo7sdlv-Z}K|q)zA)Jw`?qr@O#iML;4PZBJmqVRTj|yUKAD4
zM)uj#!0Hk2*eLVXII~%tZo~~4e-eMv7ybN;x_wQgh6c0Evbju66LX@sT5CqB!0@2c
ze}>{}_?t|OugbhpO>k@TSg-*lc@%{2HXQ7nA;A1dyflqpi1B3uLy>aK`V%i&F#C;7
z5mNCUkeC{?IcSXe%Sza8Ph5P_fd>BelXxxIXZo-t7-Lzyc_MvjYR~!lV|Lr*z0AEB
zoA|dlr*aJj(zOddFIxo89)!TjI5Los%+$Fs^~G~L)5@T>CmGUwq5F~DY)J<9khpva
z56#c{`{%Dfyz}T3Me8VDS%8N4_vph?^&htKEZ~1_e29eSra`l>(XD<E=_3ZV*9Kl6
zt+9LY@h~C0ZSSXp5r@t1V=LNx`pQ1b_s3c_)^B*<>yc6WWtg%FeLDOx{Uk2Yo5fF`
zoo*6)a<>q*jR%^SZ>!__Y5k>$=AqZ~zP=4_?sLigE&Z*a%D-z&^7xu@iZ7cC2jbyf
zGRLv;kwbTGM4qJAP&Qe^$PM4Q-<@qhZb#)=c?)S(H1^&J4m5Z{D4@IjtWC9+)6KoF
ztq8qI6HW9yl9idy*B{BB2@AOVJ1JIDU1O5++WG;C^h;v!6SyNAsFiixq%v}Ez0AMw
zvj75@2E9QJ8cRONqU(=7d1#@AwkN!KdI~lgEU2H*$V57aul~K0?u`S9-Epj;uzrz>
zeSV8by73ux9d2CLVVq*6*EjvQzH~t8^X~62f3$O6#s3l){_|oVHxnqzEfcg39QBAJ
z%(M_{34ka&Rbd+&iHorZ-1IN2<`n(@y-E|~q@9-b!25C71Kc{J!t5vNhO<t7ZCVt@
z4yz8=XXmIf-AU^z*6<hRFLrD3elBPVKV;>-X?zAE-=19@m%ZigjuC!%PjeFA6}a<-
zFEvcKcSr`ncgelqlpFN1E6}nJ|EL&5tV3D0UjgZrJI`Wx=(ry%h>oP9_A~Eu$wRQg
zm$?t!YUZ;T24%e}#cJO<ExcW-Qri2m*#*?ZzET&;2OZA`a^GxQ>&!IZ&m3akDZKUQ
z4qX88FwAIc`&WQJ#P9f*aaUt&6E=KPlM$DwPjFkND+=4357bMvTUp%asOva;Z-qMl
z^_ti*_$RN=WqhHvl~C4ro{c(62BK>{rvU^v{+Ual-orU*NQsMCw3l_R3fhitQ9AjR
zOHe#_K#Bb+l9rdS3feiiUXj06)l%(oeGjl3_20ME>co$c3+9(Pl}=J#Pbi&_F8n(q
z0!5ebwKrD(eo+uh+Jg@q3cM1VQ1prMydR;iAg0cRLYkA9DU)L2(Rta+>#ftkl)J&!
zORE?TX;J=hzG%VO3E62ksn@E>Y)KFLS7MwKqNtezDW6?Tlx(>z{4NqqC+Cjomh5Tu
zz+p#%$gFs01@9>YsRL$-XYUIHcvXAymy>)enH`h*==;kOU^%ewE<zw56vh{>8k3+)
z51Z8d_>|@AqA)M+sl2g?F~_r;julz)!ZGnN%paTipCXpc!4>0BU~4o*zEt#y@LC`<
z$vfHH3!nep+&N#WRWtU7=6kg=!3y4PdWf3O)KS2B?n4XHc7NZ^V}JRbCWBfwWQu_g
zz#{R1RK~PAD`2ciLEv~43@20Qz{$6iZaw>n?^WvJ4Fq$q(YBLopk)4wsP}_pRhJy3
zW`vfDxh`K;$usWDLmICn1jnXpZ8dwEKR#GVx+`yOTX>kmWQs^v!K{hIJq0HyGe-wb
z$q%RB^vpJRO?y|t`;UrX?4t@;6TSI;O~FUELhKBlXG6Fd(uV;f!%FyFoBsx$I`KEX
z5xbMj^W2$~f=(1z;Q~XV%aQRfK>9r5aQS^`MV8PQTto17W)!_)X~lCm5+}n;%0pi$
zWHyi<xl4GqOV-2u<XJBL1%7IEYfD(%(y($2t#!x!EflVu(K~g|KsiROUA2~AZszV_
zO0CGQS6%HQS&)ZXHfb-Eyt7pA=HG4(hQ>4Vmm#-pwrv{ZsiWQ?-Wl?n=qk!=+U%Ml
zMNN47K6iaE`H^@l4*P~X0-*BhYh*jfjWD5qOlekBQMUM7+(%Y7?kj3k#C`|F%uk>O
zLe5QXOTWRqMd6o8)GKcuUfKjN41Kxyvv4&+SA4m3{}|{=m%g`1*lTM<onX1c&gT}|
zE_zl@2-nstyF>kYC3vIKn6ok9dQT{$U}0T*XmR#El+{)M7i#(d#2;fIhen(HM6Iv?
z88>L3{FkY8y3{6m?B7oFabm|D8z3^6F`MU5_8XmPperqfrBl@xzyt&Y+)2g_b*d7>
zcuXx#_fd*KY7&9_dFbSX5qTg#@-x!_Sh%mgWvVLjcP{qDkY{>V9%(0fc3c<~SGN$t
zn0S|(sqjf3LOc!pbLkY%H}ue`1N)nz*){VXjb<Zsq89I6qyai0b6F38gy}Tz|F`eW
z61&Iw)79HYGF&Ht{l=c|;b+rbHgR((Z$I_7RT`#2jBdZXe=l5ry_Q1CEnE+Gc{o|7
zwh)!P-Ck~d|2wrjeq_cw*m;EUba$^E9mQ|a#=(c6BOn+&3BuFes17ZaM2cEo04=pJ
zhrkvkX?fp&-9!1t5h~0R!jWu|fBcB$lWt_h>UuH%M;7dydi=OP?0ioG1s6);gdj^S
zTGXa0hTe&!hE&$yx2>7<z-g$OJkJKAx04&fqQnnF^^;CC$020p6||5SuC__hJ+HK4
z<FS^bAq)BWmp_^G_{F>ApGw$%Yg{{21nTuTLtC+?)d3h4%L^J#>w<^h*r*9NAhuCu
z2<swPg<*j+#=k7RslcRI5M|7Y-IfRBhCim*93SV)J!SUa_*+`K4)^LQY^$w-O$Be7
zn=sqU!i@BDjVz3w-N-!3f#sAtXE-{~rfJG^56_m-n+q^!496r$(yNhzBx`FEKjqVC
z!vZRK6A7W*Q3o##w0GP*F}0ZO!w}aN&#vpMSd4riqsVPXgtw*w3PHB_VF5$}VI7cx
z>~4t8t55q)P+$WmJS?anvU-ioMZumQs`saEBwF>^!pmQQ$@WB%hp(cv1U)Ug6n}A4
zm`Kp`h>`uT$R{t+g2m3c1OOG~Do@sUJ=U5xrFlDL!N!gz$KRh+0j^s;M3Drp@)!4`
z@FJ3r)wPxP^WLM{Q|Q^&y3=0O1mdQpqWn1{v-=1=(G8*Pn<zjDoxO&6-g5AMG|We?
zy;#oO2C6Jt?qIH&?e!L{WOH-u(2m91^sFk73p6m-bDSg5^uu}FX@Nk{l5g-mrv+o+
zUVZQMn0XyHwzjT9Q;j31iysn|crzVPhNi4-64X4tb||^QS%nc(t>KZG;CZLvnm+#&
zaeaV=Et5GaW%%T896t;8k&&#!Iy*JESYJ$4$v)@Vqo4Di{r2D{y12}jfDhQvNVRKW
zIzc$mhKgnMKmoU2RDwftJgaansGPmi(9&WyU6BP=hAEpEE)9MbNSvRapKRjj-@)FG
z=}wa%N0A<#`mV93j&uN7X4Xb+$AX68H(E$hrOUypOdhsg;NpJBKWVc)brgPn{2Dw^
zzUTG#cmEnIog@dQFLox^@&bknc{|6<+&-dq7+`O26(1qVm82F;?<C2|q;Hh0%Dcw0
zMh+ZCd_d*$)Krc?E^mBRPHj}S^K^9%p%H_^$q<+PjG880(2qlpdlQ(J#>8T*iyjU!
zVgbZ^vT?*d-K6!}x6z2>LaR&DzyJb;)t&aoY!#TDwa(qv53CEr`FgT2AK}k&t8`pl
zfq@6dn-9B!b}-YPmRE~kR)$N?{;;yr?r?|rRta?N&RH$|{JB2gS=`6?H*!kcb3GOn
z+sgNv?j$ty{8$8aU=@Bj^_rFVt5b??OPf^T@^ryqKHTaYy?_nau#qropEENT5<T*g
zVJ=!Rl}KzBeX}PP^Z_beDQR@NhPujqb8Z~c9_rb3`TR-Mh$JPhcQNS$-%WhCd|9?r
z%N;^~kC&%o=rG%aM~j1_V=5QcSa1bZRB~L2kKHg15i+EL#J`(y)BRl#T<mBf`l=qm
zOR;oO?nh0nWo9OCZ_+Y1KUj`n8FJBP2Jw(1mLZ?p`R0R&l_YtQ>XGkm>)rmicuyTn
z!lq+l7^&gK@Q#+C4&kj&EvHFLDqEytTZ7B8vNf1`tA{CYN#qNNd=l~-cUng2;7T$5
z_&bgw>B_iB=eJ}}HvS4_JSkm&%sQSoE1Ow~%&{MU@&*F3zvNE_0(e=b3=+Q5+0%H9
zUiv&n8}U2Lt1d0|-JhQdk@+ansx6J#{hFJXG6Hnsf43(mvGbNU9(*&1Q@|}^b4gq6
zCWH$&K+iPU!}An>R;eCX8G4#<lR9d=>IzXtq-Zcl(ZRfMGE3k3;=w}c&xEBIbSdhu
zuOAxzGCZ0w#s`?WLz)%^qn_+nN`+qbjE`Ua_VWAF8@)E##mXm!gM|?1f860jqw~@K
z+Z@H`-n0e+^MvzM;Qn>JP5rPV?>@R??COHGD@~dS_vOcQ`kE|*4}Z3%OfD=!vYi|U
zkB2dmz5y*h@hbM^Qjl4biCq_>$f|68^O3l#&(BoL{FxD={0{v{d$$qx737F8pj@$t
z@9xjktZ;w*y};rxUX&*@(}t7pUCrR<hlz=s!p`Jq)H{9Hh*;11Ehf99+i(z>;LWFv
z4m8a!g3Va(cIllZ^|AIlcBCN91YvgxyRZVTAvWFv0yU&?Wc726=?K5Ke(D>C53u|n
zM*3bogrKF7!E*B#9@6a?AmU-s7UX+9<ddO^^AZ=V*M6prIvodklNAqZ>HwOE$$c;|
z=HEnl(#>v5-OSxo81<{0KENBUED!vSj_%=F6ZdSS<CK%iH`F@MP2qan)){em*7OD_
zi|=1>zJZZh8L1Yp_59%cgO(3;08AYcw<}`3aXa}}gq`sI)T5KL`TS-WuuEMscYcYv
z{AsIaezrUG_Q~sU$wxS^NW!^<BmS4~=wnp;PxSBuS3W^`(dIfU?Edl0<Yw-HTA6&y
zZgkh_+*;=$$K3Fn>%*_yEe^R6m(wNfdp&c%Qpm`p9^dg5O-IILo^Wp*4WrHEo%zD|
ze>3TKmzA7fBzK+m7#D|ogu7$b-6h#+`=6+#QMpMuupD!1j`x~Ma2;U<i%)wu0vAxh
z+gl$zZ<U5Drtf8rhW)K5>H0ZhAFm?4_R7Y+=0_p~1TRZxz4^o6J^2$(4A;oqKz%-o
z&KGvAUbHaMZE%aB=d?b{EIH<W)~>J5Y`*8vo(bImP*iJj2j|Bax7vl2&J>0u!Gjt~
z&@)F85RVNF38<!NK0p6?nd}zbECouQhz-2zeL{x@q`BL39Q~<EA3uHi)X*^J5g3ZK
zEM5aC$52yKM}7D(T}!w(eJj!$u6d7Hk4($o%?1Bt<P{8n(mQGl2&hH9;*5R-(NK2A
z(&3}KHT0=4tt`K&9mrS_zf?OXxECzVO00@(IUCe)_XHsyzD(P*cigTYb3--zW$>tY
z8Ns;7pQ-ID#<bPyCw9zyPZ*brj$c9oR=)UjaJFGn15NDWul{#rf+NY4AeyH?zP?3g
z4XliJMpsrRjwi?UjWkVeO4s=gsz<-<Qu;~gm{^#1e`u$%j5abAvxBLT7ukT4AWxGb
zCDJ%=z@nUEyg|GeCQ|yM85CRDNk-XayaSUhO;H0q+&$??Tn#}BN205LTPGFuSPU^j
zV~l~y0^nbv0035~-|zx^e_1ttwvR`199mlECjlq4cL~=&CIDZ<aceadfgMBz@xq&p
ztO9drmuKERMX5=!D1n$|?OuA@7`7^go1~yD1Km95Ivp6VZY0XXy#B`zY$y@^;Oy)5
zrqWf(w1@df#?Hl`xA#jF0Hh<s@^7o<fAmpPg0Q%x8{(wpf&&v2i7^8SJaLPI&y)p^
z{9^8EL)f$Wf&$($<hwP?700us!kqdbnVOV;*eONz?VKAXcJ6Q=IJApDrxs-%bz~*o
zyPDcr#f20jgRM|3WA}m|cyWa#^*G9nWteS>ktIedi`fwhH{vsVhwfRnhY^gJiYvqb
zBEaimiZ#Dt?fd@9%r*Z<(|P|>{l9<w5F+9j<%nz{!lCTY$zIhFvSoyhz4yyr9oc)m
z$;i&i-W>Y~A!LtZ%iernpC7*d05`|&b)K*5d0mg|e$P|N#t()iD5sx-n^p+U>l+9G
zF7}Rm-uv3Lua}MUHz&0g5*bx?t#4`AFjHyZ#NdB;1m;IuHJ_cZ6O+5!yW8NVS2)u(
z@UOu9h3`d5fpD$coap6*5ZG2)JFeNe!LF@|L#lP}m#$hqhjbN{`EQKZ)bvQ599^GE
zyBD}^1!vv3TYU`a{?X!zvbD<QRmS`=d#;>d>uj+^hVFR)!T<4GO+mA{vC(UxEuuDm
z%vSvg8-&30kw=^#6v`^?y#8x#4ZH^d+k=CuHnSLvd@Q?1Ca03g=DWn={qFz{b1Xs2
zZ-#yR()ny{dUGz10{4*icJc1Kgu}%Y|24K2**&4nPbw5(2@{0Kvg9EHgd&LQ(v2it
zk6JDbeGZ@5aDIDyIt*rZ_IomkYceQVz8l-5!#=ZvG~?*fPkQ-Qno-5GZ{vZ+$WMnN
ztp~@qSF}KOL&~?DF|o?YnasY##D}M9%;w!@NE&6*C+6(K(L>;0+F6iq(c}7U!-O3h
z#VFx=o-xTS+aAPORQ!3fq-UY|6xsj2FXzXc7q?DoJ_`nPGY#<H9MS3)-nmOLS9g)U
zfb=!f?Y7FMMcDi6cM1EQz6WUJ++24T{9oHOU!IJ7J3aAijFLR*N^r8%=vuUgnS#^`
z7EfFW_<LRRB__9yA4N7kAXi^`-3EOTu+Rq2M-D7JxYKS`BzPAeOSc@$dm4tFK)?wW
z(tFlFKFUC&!%U8l0wA<rx3Ian9dnUL6MM*lI6|H;<RfenYu?8KL6$5R(~0IUZo*<n
zn#NP!q}}IpN4HOy!ynO^)<MTyy^!dP`$&#nS$0`j(4tF2T*@s*B`?Cbd*Zqai(T%Y
zpZc>)apLOcH6t;TUuHLNXdx>(En7=y!4Pf&S&-okVTQ5d-3f!sMxooQSJN4dOweI|
zy<x=uqCHT23lKy9gN3V`L2y|fqsxqkrUq}NRZ|R6JsQhH2KAmOz2?Y_a^D>kUjwRG
zq=Z|E_-hPd(ZI+hE%q5de}>QD-vX)YL0p#C?sjpx?Oa(C>4-wgm%MHdyHc%)0EKpv
z5A5oQXvg;PnyPhwr?2QHhZbRoU$2eko!^jGCaUj6U4b`V^Zg^zw^*>@IxJnnDLhOI
z{!_NYqh5Mkz*kzzwyI}|%7_W!EY66w5{LF~b|pU+2#5h-fU>ErKHBq^C1-gg0`hl#
zLVj7-D#U$W{e@S?{)O+)XuMAf7cq9ihvUH)1c&m#;Doo;YOni`X7@J?d$zpJ5_t~g
zEvJ}iU=^iTTnikE0FA;g-m#N+3D`Wh*0kat_}m2xfr`=Ue_uEGm(43r^Y`Fj`z~&-
zOujKvtN1yeY>CQbe`C#;mEQXp)x5)@pJs^9JKyYTbTN;>h=e53f5dT%--U#hef|pz
zM{X^@V|QBW+6T|kOuq^6sx{x@u{UDpxPbl!xDm>p4+!y>FzDX%=gXAfclN%BV`AFV
zoWH!8_gRa}I-mM7Tzgup_OfC^g%7dA5jGlho^p*utt=f+`Z^e!??xtNNlS`IPkz3G
zcx?S$MS>1WiNjB3P?zueP}J?N|3{W4#LjMKQ23E}^x;FN3~qzThbARNY^BucMc-&F
z09`dq(6P0PsnGAZWd-twfjfUJ>G4H28*j93C*;0j4xj->sVIt26N_c?Uhl1QCZAQ*
zt>^Lq;fss61<$AnWz0auE;iRVj4Y_1IKX)6s3*(&uvA3TY4>p0*Hu5Qbk3Hi4}9%U
z>o-%DR6#$mR%9R4pN|S)bO5jW1QG|pufBCQyOo8Urv@coz&<?FtgfCO{rgge6u(c6
zlueuwD}5fO#rH_U1Y~>D)H0q_4Xv#me<?1mW%5~xIX<o#v4on&2M7?Yz+$Yj(~kp`
zII5NV7wgQFuKV6ddadOp^^t34md-Ou*&lQTpYltf=pN`p->&@E7f2eS0^g(nV5U2t
zE;HFqueO6}SK>>{yBfir&Xt%;yK-B9#b^e2gn#@fN&+QlC;o5~y8F_3-XPXCK#nME
zz2w;MX<Ym7$ao$+v|I=5^$L<ehS^x7FCjplSP`L&%f}vssS2=+kT8*1;G0ILn-}iV
z+9Ji<5Zrx!H4;KnsSO~Y7Yop#NcG6R?6+wOkC5Py=Jyjs%wr``14<i1<!k<`viR8C
zPqJoB6fB;sX6QT11cSdPyoU35<(J=v`T|krLX);7r$5~?+*ZCl8=BsH#wswemii8@
zi^8Jg$DS+86Vu>9gBGxIVA@<Us1+HF*B06avuEK5DEB5HBOtKu2RiX&+67eJ@6<&u
zkT6gx^95C7It?np;%#S~nv==$rxWx~ij}SH?>6>Ww=!!t3BSeB&?H9RqkX39v%JtZ
zFVF7Yu0_px;D)lUw=LwvGL3YN(=VZNVPh$GnAyyF>qeCx*!YRg-hPE6y9(QXUHU1e
zYZQ;`>Pvzv_<Mg*!0rC6`#yF}Uh^y=p)+pOcTr{7XL~<19D&C3S4x0d^or3^i6yD3
z^%%AG!<LjcHlAHux+M0srMt=d&2P**V4NEa@^t<El~Yy2);R-bpY1-@bZx4l7Q^cp
z{40fi=ZZ`yP^)1^YHBTGw>u2&E;CN0{$y?s@?uw!_rH+vMGeAUV;mO7wtr2Js|qk{
zYpdT2%RWiOTvt$e=O<(VTQzeT_w&RQNt6z;YwAhE)&wx%s47>-Fj2E$_SFL9of3Dh
zGaR4I3@YieCVu;m0%)y%o@L(m9{HNx|LT8|=uYXY&eQo;(M;8bf!W}^JRhM^&E1Y*
zC+#xBo4wmm0N803v7M`{sjG7~b$OvQN=+f%pSPyZAnF{gbaQ)Ea9hi9o}J($r^#B-
zpGfubQ3GHwNfZP2W0K+d<F;}uN&H5cSvE@9oO101v8(tgQ2cc^2eEg`t0W<9q5>R~
zB3ks3@^d!yxV-p@3B9f2*sF{2FWHYgmeUVoN@SC_o}(Rtznt=NlByieEWb{8{QX1N
z2QLVjE^#|0TeJ)fGQTuYmA-k=%nP-(*x7R+nE~kO&Y_<CO)J%#+$HOG{r37(x{}M;
zy3Zl03cUZFj<!oh)AIt@?!!Ee=fatu+xxV@iRtWi=1o_DUCA75vZf+ofgvo%4a!D~
z{}xIB`6orD@S3Kwd&Tlc#?zbj`dW)M+Fc3Pd9Vqeh2&@po}N<WgbW6Kfbqz+R91Rl
z`2%rwO*bm?6|=vZcWg3Nz!lv!Fe@zOwksJD-zFf%LHqWfv&Hh)_@Y_M$DscjOdC(7
zPV1t!KIhNSk%xlw(wlu#&aIpl&(+w9{Mi)xs3*EFGJV&kmMPMwYaPFrZ8Zv^BiA)M
zwsM>kV#wg4?RdRxtbz~OJW!_We1vhjK+${R$FIpo<idJWqGVYhAOGvkMUzF7U_X!<
z3My+b`m4$qKRqMX7fcaldR+}ABPG{Y`}v-DM~kQL5vxrYyBs&dKL2S`={F}R8@>o1
z(USfrrMXz~F9eu6a6D)EGK83tSx~s}bB3_<G*>^kw6J}AD&K8s_iKJ?cKN$d2H<Yd
zRI42ETmxK=r%8Wvh<MG;v~Kj=^9do#>nG6_#5DfiWSCUwOJ-sgkdjK=OHB9r3CxsO
zXb0oXiDKuCS#Pf86Qv#FZT(W6wwSB?!LYD&@AxD!iQF>da{_gCytkK^ljM}k+ylA#
zroDxP;@aD!+hgfy%d2+a5VWp|0Jl5sUh0H%#FBV1vQNU)4)LQ|nR61U!kgSM6PgB%
z3nYQ+u~^m1-<_vfqAmf<lD8d)x2H_E`&qZ$r)7pluisyS_rTcg`Rhx-Z7G{tT+Lqy
zF=y#ECa+N_yAe+szRhNu7a!|WEvSyI2MQ0u6k`G@$`~Fj8-!@V4ZlVLim8WR?&jGN
zA!NwrW*k+Izcbp5>(v%#3}wt{Kl#=S$z-S1DXPkkr(-*Azf3Ax&~qZqbc*5RHE)0Y
zgV`LDjKA=7=&8o}LLg@z4cn3VyX~Eb^_g(pX3q`m<};M4xfROI`xsLc8NZ_Q?sxit
zR&g9VcX#r-nJ9xgn?!#OKrXe(&Jqz55U<ajk{jW)7FhGsbHSf_q7L*)@jrX!^rXM?
ztV^gMqe>~3)e+*m-#JjIzV-<>=ZPzwP6L&cau}UVT!DJV#hCQ92KQ=zd*G`3=!SE1
zlv3(rQSUkdSGVWl-rMW_*$I7S@$5xoDT~N}6}cDkM;a={$j|ox`SPOA{a-~eUpbex
zsyKq;!@r{M99S{S7mp)kOXHraxt)5ZrESfn!7@&dCr;!GrkIJv`wpA3MeY7;_1S;?
zl*m<(6-VGWBqJCA^ZqfRc}HP_c0vBl%R{B|8N*ddzn-?&VKiY5kbhXKQu;{DL5ok+
z%l(z}L+}$E)(i~+HEaq|MI{yX>+Yil!d&%(;zBCd?_ftgWO!VISblTlo~><$n3JnT
z`C0#-?=pYuX<&1y#H`kGNB1FubEsY?gFS3Loh@~DNtIT%Naok3^MQP^Q0>`?#NK;z
zN}B{FUQA%IHs8z3?W#!m&ic08R4cc-(Oz>KGDrXaGCeK;4Q+m-l7V+R_d~zEN_;`E
zJS5p{zxa2t^De!|n8eykh~Me4Apk@C1r~n&51g^F)a|39qVSdB6R}=enwP{3S@zhv
ze!Vl8J2@FJy1u`)?Eb#nZ`fyll3Ng~iT*ITX8TOlKfY3)M^xaR?#9AaZUXmEQ+bdX
zOdzC<va1?M(CpWJ4X3J9<O(8|f<P5$n7UUwo_l1FPJCYbY|I^dYN@K2{_M$e1D64O
zG5mG800D+wLzP3!5nK0fq9OcCy24(xQ$7ca+!hLzYl3vz>-)fjM@QR60AIk*D43?(
z>plArU!#Dt|NcnR;C*{z4#UJJWaQ^O+0+BrR$&bw!yN0DsYDpyqz;#j&}R-#&MUls
z^G)-^JsBAm{9o1jJ+LX7Hduh{8mq%sDuEa@?CuKj2t~iF1E$`|rE!d#lW^G$H|4#Q
zT`)1(Q;hPOlPF#7xKe(xv7_}mVnj?k;X#3Xfsog2A4nau-LsxOPur7zL(ube_SCEO
z^zsz=>tqJ9RXWbAI@?{t9>%Z{8|y2Dhm*HVg#QXpM5?Qay4Zl?H6!<ZTpIX8uh+TQ
z=7lV(Z+d<Q$|UKrl=4P0Pw=sSM9~U9p*Df)(`R{02z=Tg>(7XcN5}b-G|wb_AmCUC
zesJtgv7L;F7L2j#k((RTgTyqk9N-sz2Z}}?>3x;ZSqh*Yk(m_<yf0NN4K^t=T)kYb
z*ETnGoo@f;^3rrW>PxoRWhpFq^b@8wtQgv(^h7n(h-jLT8}<U|b(*^*qSLbtzQ@Qp
z>(d@{pdbuQo1dDIp`nDS;)O!83-7;qzRar>Ya*z0@afw~kWmJ=`46m@yr7I3fF@~w
zjYcTu9fdL(ett1(2)v)_^S^T!)>%{(^cERLYjo8e_r|xWw#v!rSZl)OuhU+0x)W-9
zW2|6a>~uogcdJ<HL!XOIlK9BLrA3l9`F7+NAJQ+neG+aHB9pe}shovcR>(;{lrW1p
z+4C9%6}Z&JLk7{^a&43-`i%1j0+i^77VV+07lMd}!2O>T)msUyhsCwu$4n4%(sG(<
zTB~gbxmQSq+p)C6jozjG{`7m9s4QP}m}nfF=&j@s{R^kN>b?h~X%<%YhpXcv>KQj3
zyC=KG*H5FE>8J>C_*4$TEJ3rg%{(I7c5`!{CsV^i{LgN;ang25me13GZ`nR)6`WX5
z@3WTSGIvV4b2ibq$JA)S(WfV*5}M7}dT^1(-+VJi+j5}SdUmcNL7$d&I~ezL49o%S
zLm*#gomLJ)ned}e-zmyLeaQ3`nH6KL9X#e&RhaG(&HXNnMZ@==L7FbHdT(A{4-W8W
zaYuy9P~tr??~8j+vcCdGBCeNk(!)+-WBu}R4t9;&Z;G@UgV8FbWk!w8#<$tehr!3s
z?L=Pl<=1zT`)6?z(<tpSfsnb0)GuCN8Rn*T=0Q&Phyv5rX=vJVq|w!ZMe9j0PU^CL
z__mYcW<w85Ag5>#>l^4v_#T)d`#aQj)w74`bu=Gd8~clzG6RH9+Y$fZ_IdB^iTq8D
z+F^VTST|(pHG*M5u`tS*;B`}Cv_npqsLgb~lLZuQp?l=!1~)E}n7jKQRCTAnqwdit
zJ4BYIH+cVR(*v@;3s_G4y`YtVxKQ!BX>pe7Q6&R<Y%)y*t1JS;uDFW`lhG}}(&3t*
zsbA3W-Sj9LfIHS$a*Qgjf3hrsWjqDhaW123-;HNN{7%&%gm^?EM23brXyah(s7gRs
zEeU}YldYVCP+AO%(LUxt^sK*lZR{r&+FL^FK!sx6^Ulme*!w2^__UL<OY?Zp|HSiU
z-XGf+6T0g1)ZElKKOtdIoW4<Q=J=n0S5*mN*Z<TVx47GVV6(chfKhBD!)CQ`{pZZ(
z>E0V>&y_P0yDh!iGF$7v(^FGT4u@-VhUJr6Z=Q5pir$0&Hn&2{8E0TiCd46+TlfBe
zy#v+PD14rpA9K`%1Zob-TV{x-+0{^8#mKog!%D523(L;ZuA8%D?M`doGP)Z47AWN=
zlJ{AG1f4h%vV5am8;4i<uvHH#Wj5zEKh|?cJCxv~?Xr2py3^lF{V~nW;JmJM$c1ld
z`eIzLtp#h6b2yd4fj8o{XTWS1gO5km$KJ=WnKQ4=s{0=l5eB}dnFPomJekp~5%KR=
zM-qnWs?ipXTMMw1)WH8MecJJ{YD%?=x9dCLg5Q>lWC?;cK~=5H1KsBEz-@Y@?d;Ut
z+@{a5$jugTwLgC!)6ApI=X<q0EFfdr^aA{<oJ_WY^D=DHcA(GMVyJ9Rh7<k@=;$db
z+QbeL)1;GP&T@Gv7@BcdEOu~vNwq-Y@0Q))mQ}TJ8?%}WYuh(#w4WkA_PxZL0*oGK
zO6dx4Vog9duf$f-^ssAct(M*J&~S3)O3Zi-o=o&YbMcnAd=W}P%n3`HIXks6>$U#h
z{&NPkwv<mfA>SN2>XwiWRH|oC5;1C%_gbBmQm#ueuXomXEDX7b^af5D8@v~H>gyA}
zac0%I?CWEoffnfssUffGOQz~QfWx)xg3M8+*O)(QXY~A6|3D}-1utV=h3Dm;zR+S*
z&X(KeRy)Jl)V;9gxeMxoEL);SNym+}<-QxJR;P`FV|pgvKYhpBqoe-`O#Fruw1Z1}
zhXv9%`yk%!nGmzs=>KN|l&G}nsl8=YE1RaQ;VgUBw%F&+8be1Krzic>Y{PNvAJf(P
zH}jGV8}@GQy1BV}x(P=*waF&pDAFa5r?)I(gRIE&A|so=d5HhniDdrSkkRmzcyZ#Q
z=JIQXXh~GF@5u{U@Mr^Uk0(l#Vmt|GfOYDDfy)!3X7B_*r)@o3wevls2To7t_jYz8
znnpgawo9-!NCZo%#1=gjmG2|zl^-_isE!Q_lIiVz)Tu~i!At;Ar3jozv68^R-sh3d
z;h8emBS%x(f+6(Pu~^g0zJ2%Y5;y-6Qm6?$#U58$3-`=_PZI%GmwIat(!T*VJa?P{
zIHQQci=mpnQ(Tk^-_7{K;)tc@L*W9y@F9t1+k%kwig!p<7*f7(PJ0=V-EDca@|Qv4
z7jfqNbnfvELUAr5Eh*)bCG_gOfec!aJ&zy)F3%n8L@r7lAS4?>S->_06}3Qw!5*8U
z9iU`%3PfFWW+3GW^;cDe7I0a;1Ca}R=jdBoop$-PoH>BT`+8!icE0KC4>g64z3<(C
zIQ<HN?&dGMgQ^7OLN8l?mqyTIG4<!Jjm@Hu63R1t#tH3z`dQMLMt<Vi8GI6+js~Mr
z<uD30&A03*UQ+}r1LqNI7TI*amyWss1XS)vNmo`^_eM88%ecFO1_;+olNp_%go5fp
zehe^H!4-WHmZ0REF%x!E)o&xr_!`=pI?7r?MuG>J<mz$4ggOX!&WpIB)^|}SqmpcP
z=7}b;f_b5F29!207)>C!J!_l23xxrb_e2)`m06!Yb7W2NG!^fd7Gy&8$J_;`XXw73
zi1tXT<YBoGFH?SJ0}($@CWwZj;dYVxFC0Bgo!-jJg_lJX&Gv`?sD42o>G1%jj3H3s
zCZJ$3LlSL;a}h`rUQq^2Op9^G$LrF>-HQ>!X-`g6%KI+;?lWpSIGihUKVMDClDwwL
zi4AByTx*=Q<qUFP{3|AV(nbA7p6O*!c)6j+&I??p^!2jzH83N6A>v{~)VYXjJxbu2
zEU(qwIj=9Ph|3rn+P5sv5D^x!b!)K6*aYvr_bSG$ADDvh;>KBLtoR<YX#0x&^QWb5
zUq{)6#&Z*_v{n22%szQAk&N$A^okeOM6rwu4h!bRHXH=zh)9Sq2?)r;2J2d@Dr#<w
zuD9%3|8+850`Dl^la+l|dKYBOT2{KpyDfdR-lsu3e|3J-{WSw+RPPb^v{UWn-x=}a
z6^+R<Lx&OemG}jz4|YQ;{8=D@U=cg0nrK;i)_1x04Ad@QCpIydLWL~%(`yw$8{aL8
zmg|f?U)qaK&>*2=Ryi_r0|xH~#<Ff|$gGjpFMj~QNnb%`gM&pVDM!#?&Cdx-mf@y@
z3(m~*-*<8Ir%z<>ncILilj)gR?yXcjdt#VX{#XW0-0bYD#smwUe1f}O15AeAXE6#@
zYlB{UU@*L|hqH4ot@TPLHFv@1u|ZDtm;aoe{U?!%zT4>K!8WNpGum>Dl_+5JIyftg
z3)MaWnooE2Vx0)^nfgMZeH!sk9k4DI;#G%Lpa}UpQ~K*{=jmuxx+qe<SDca<u3$85
z5BX1#Y>ti_X7RiCIVd`1sgv!8rBewCJZwr6w8k5GCxi7^34{6POPTCRvfS$nhR)?s
z7(T^{9>UF1^X`cfYRn)Igg;@C^s#l(*gJrKx1z>(%^zCIeF<fyE-a92W?7l#!22<N
z5;#0RJoR=HEcLxU&4)f*5lZaDl-6jC1;;sC4+!ot=HjN)iGtyC!G}<2)NWR5<wRpm
zMW;bR-hDLr9R(~231vN~RKVt^KLo@0BPtxC%oZBn5Xvq^JXVMH05&vx9&t>3EHEL<
z7xPL#-DaunQ``c|Hm7t8Y42M5w7EPmx}>bA)(F^YU+(#0m_?}jK}wS0q~@MSPMx)D
z!{b6~)=BJ)(9o_S08Q7lnl`pF6|VgnkjBj!T&t;*)Ok#KCxWJ_@IjKKodZktii`|R
ze6(bTf8fC7X?gC@$>3j;su;AwH6S@0JWBbaR>JzZ6MIiI+C`NEDUv%8;MB))hpa1_
zbRkwqY4EO{!9Q~=syr%gnaUeiir19-!OHlP!btlG+Q0?&@mI>voD)}&1#}SBEvu;R
z+#fC+qB6OnKXFDEA9M6D-Id9r>i1$Ii%k`nB<(Hpe!TD$DgzVOe%SaGdSHiv{S=HD
zkO`;>^27i9XmJp-``Q0XSy2$*^s@#ovWY}r!~$2N9``&7&Fb0bOa>2Pu2ZCft3j8!
zbR%7f+gRQq#Fnjzl{YKP3}Dii_56ls3rk&OApWWO!7USNZybek&De-c!O?2{5lc_s
zdY_@pn^|P?i5ppJtMR@C5P0<CQ8eM58@wGRV;GUHU%9t=r=PKo$(MMpcT{qP-K>pE
zg@gT^JwC4bUcvlB&qMgN`uxf@FKzE!+S(4D4~fm4dbpmP%CMVP2?W-bHytO?wj3qR
z-wu0$+#KKAy_-#@prv%4cES9?m}O3js;R8&jjW6hI_|sY{6>wwz{|a3?Ii77ze_#6
zvci&)THAS1kL~Z#O<Uq&@9sG)uN@y+@TumH^;P#f4Tw@R6Th>XsV1jBIS9DpxPWx5
zJlUO<DDHV<A@@dU&SO~gwd{YrqE6hba(j&<e<rI^5$?AcwOXDhcCDAqw7%DRljXoV
z<6j(-=%FT!$nN&cBCwg13EW%^&&L-gAXemD&qvurK&_{qHjc?-<)VP;ev;OMg1(!t
zOZtxvM{2gtjbmm4l=OAz{?YKDvI{R(jZHnShid1R#8GK|9Y4TL>?;bUmYXU^s}}<2
zz>N`baBOroPQCSo<HgGf+r7snj5^W~tw|TMY&mv(JYJy_pr78(2yFhR-FS+qMM-&H
zV0ZM=zYnTLP*N$7A;Jbf!Mg&l0}E66N_p`Gup4>WhMv2{AoBy);(UsUERzyIvojOJ
z_bNl6Xf5*;%5OiNs;s&%sZ7%+)bRCnjDY3HS_0=jjn+vgDOg@OB`s@+sQ$VblMgZC
zAct#0YXXqjd{b`t(ruf}f&n)0g^@LgkF^6AD09}>dJ7t<YCV*<21vC6#^0JD1x$jF
zmGP_}sSr@V5h9321e`EcuZk{>t&vrpuPz>;m}0^Vkxd-~=TML*2{J^8{&@V40=g)T
z{y}B7UJtw#-O%vd_oHU|^z9tbI{`2@_^UQJ0InG^33NffIlbb3OqtHclylnNL*sWT
zt7Mt2l0lKMGh2UX3?^?|Mb9v#^OD!S-G{V{WB3nw_4=}-xi|!v`*7#f%(9RrvLG6?
z(1LYKBM)~t9`lgXi^D`3JOOTEmM5&nVKkJ`_74UpUm9${*?t$3`le!+Ki=rl(o!wZ
zdPskJ!Erf5Yox-xySul&v!mZ~)eY7kqcyuIquY@9dm6vwovgGbMItIggV3+EYt?2R
znGP7HnE#yIqO&<@`e0y)b<(48weqOnPSBcJG-ghyt~Y=k5e2rdVJ-^6iAJ;gO@+en
zW3dKXp=Hy-0mqM`$&-a^4+QdRMsom^E(U@ZQE5aI48`}m26Sl7`6pjtak8)oDM1XZ
z3#!|fvGo+|QLexAmo@*F>$gy^ydf63oZQHXPQnD-xaGEd)hAT>uP4qAI*F4B*xI?y
zb0cR(eZ8$*Oszb2DipHBJq~z@GuwFhJe%k2Kej*A!CmL3+~X*LBY4=^{Q$)k`gvC&
zy%k|2#ST`J85en#5S9Vei9M7}?r6s0gf7c_t$1A&J3fZmMA=eEg;*!isXhHP-O1(|
zD~GJ!y5W?YjgS;W^Tk0zz5}MMUIBam67_9%fBtr7-g|4Z<v0$wQJQ?ggF?x*^5?Fo
zvYG(DEYSTqZ3LZ$OhHfY0;C0RGGYSucEQ7ahZ~!}OsT?6fNKb`H$38GW_bMd;lmJ+
z^1z*a_$m!eJe}WP#fxEv;rW546^=YIxLQG#Z)>PxdryN^(0dSzsjDU;B1~Y)@{ZWS
z#lzvl!@Xt~4*nx|z{^}f)MK$@qCNZc$k@v+U@h40`eNwKoc1hW_(Vn!8sPLFTN*uL
zlLFLKC(r2bPZW?QWzuOLXI#!uP&_H%ll<wi<m}wiTeV^^Q=@2oea4+aQJK@^ge!ym
zyk0PbMe<3WDE>OQl)l>MZ~0f+kvhg1B|bU*3ON50e%(-*rjE)$rmE)4n&onIL9c4k
zM7}ael%bovu8V;2%E9{uT}TtO5_i~J<rrK_jc^w<fdBG3l!8foFX^H7LB~WD{M8Sw
zJ3qumU2aYkgWyDYk<iMmtKGS;q(h_qk77kDTfi@g9!YN{+1lf4C?p!{kN1QCE4qxm
zyVo4uUw@|8dJM8lk4EF(ymZ}bzosr2&Xja%;kxkhx(j5*xyvfIi*>WZ^495yac?a=
zZx)}5I)%ZIhBQvRIej+DEFlnxJZevZMV2*InYd2_wBju=kicGJUy(*jt~4)x9P=!X
z91rY%z*D!>09&d0JRe%9vVd3^B{1P-XNT%T*qIpcsQW*#AXQyvA^4O63y6PVL2bH|
zE>b}-K<PVHA?CR1rt)X!(6<7gAYLBMs?DKVX=~2V*6ST3@s<kLr9o4bPR(we7cYOy
zbIQjf$TI!>@bHMaP4A(Fo)NO(nL1qrz9$P82#yaUIpa0&e5d@tlvM`*X+P5Zo<_f*
zgpGKfyW+c<j(P#0@k~DveHj12`?{ZJ_~jwG0{ASR5^rwjW9n4bbo8yYmA>(7;O)P_
zQ*9C96q^1OxmM|ad|hGOiY$>dRCN<mG2eP%oq;2^<7Z<9>iBt~dK6lTNhX$1jR=1v
zjSQNSni2+TbAv|K7kbxpl?%(ZBx{wqpSaHmbIakx&6Tkr3qv*<Cx*i}Ru@e^{r*ZB
zNMN-m>|)_y(%90M%cj&7!le9$xjQ~~VHD^0ros>f(Qw6$a*eAOect+^NK_CRQ4(4D
z3;&zNUnPI#wLsr`xTJa;!~<Qi^J(2w<}E{V>~)@8w!1TFY#vRVlRA#h^l$BbN<kvz
ztEoM02ED!NhsV6e_dMRMcx0N>QC)v|1q55@goMH6+JlE2O$B9F0w#~q5UL-Qytr4I
z+~%5U<fm!Jku@Ftw`WYZ1)zTG#whgqDspmOw9<F6)NS5-X@j#@2$#NL4xC9}W%sE>
zSOdUq)a{w{bvtmjr=_heb-l7VZ0K#ZLHb6m=wh+?GKsOVy0z+snB)i8(d!w2fsrx&
zrH5c$Sy@4unlkrCbYePOOH+tt#a^_e=qdl9<NyfN6BZFh!{erMZS=79rx_Cyu-E#s
z?VyF+3nYwm?u+tVblIaUE(;4XI~FL%oFysI#{&nw!K68Kps@|0d)YJ;ApO0#p!Fj1
z_F_3}Tk}TS_j=xXwo$YHJz$2<#)_w45}Vr=8F?HWyx6nX{CL{)a6512l!-y&wrjwy
z`8E&NT3z|mW9{g?(XE{O8(mwe#HmZ>gHqD_>RVVdIkcg2r-|Fn+A}eI@69Rx*%l8|
zhiqxxT?*PH182q!d#`!1By*j_Rd{~c)|0W6ib@+4(eHES1U|3eSYWiKbb&WSz|Gqc
ztI^>hFbES$`3i2AjBHLna;Rv&mWfcyOWAT)CitgACy+;jFqT1LFJFstzMam2qJfPs
z!&gV$S()M<M<=CQ$;^3>_#iWMMHmU3Qm{Q(aC0uu#&qDucu}O~P(}KxxYc`a){Pv~
z140#(WGn(Jhoe0K@k?*l$ngYC%pZGE2J*YmeD0r>y_De&54?Ns<MjO?djS=6)cU0K
z)ha9tSC)u)1Q$_I7)c(o4fdLqRq8IgUlC8VBuz?D<JV+kX8a)E7`1%0Nio*oc&a33
zLP!5DlQ5|ZDBE(Zl9+-*H%${oSg#)o9=<!%tMAb!%M5V9zfE>x_%2W9D%+O<a7BEw
zc&emVwc~s6+0pclgd5Y7C?1so9u-s+X!)I#^5n0k=5A9M$>kzsQ}&A|?*1Fx1zYwb
zR%G_;jt>iSkSFoCB$sA}utE`a(E51gr_3Q?@*DA=3DB?y6<D!-{<PWF8;hpa)~aJ>
zZIf&x!%N_3nP$u$BwA2-#U6){QtZx-#P0rW5qtcY&&HOR&#N;yu=6z=fiVGbd>teq
zU~5|UbIj_L<!DzBxK>mTy{CwY1}S3@C=w0Fk6t+3PDt(CZ7dtp#rKQy6;30X?3O{`
z;RNM62;j_tUDd{ra38IfP6D(8XnI%7X0s*%5dV^sdk<ri!{$&}Vy-8a@^P!nTL{a%
z8ka&J``Ts~FWhtiCv8%SlCIU;Yd%lf)P=3e9#e@-Mr$XmUtjbuPUj)?>sVbvrfE;A
zMqNC-kfKt0TxpH!KWMseDmRR*Us$x%BjpQdqU3_93_1Tc(oc`O&J)xp-tz%|rAYww
z`o#K3WxRK62z1Ll(_oSOJ+UZjIHm;7)vg<jeEMu+B1u?eGVcV;qW;TnJXoW@U1qvo
zlqd(7bDNycbNJA3`Tx%MXQT~v89@Riji2=4Z*aso@<w6oVsOgojnwtg{PhAqUBjSY
z-(oX`w5Zd8tlnfiD@%TI!%tES3iuYhGItj{W$9d3SC{NJ-{b_iIJ|OLY}9hhKV+ax
z%fH8h2dRLXa+TD(?p*X~Sfu7R1Gm`eX<cn?#KD5Zefy>Lci~o&)BDGZV-~}<B2|-8
z?w9*jQLS~CaX8;|T6O7@^-g*wTm_Z@U#LrwG#oPh*%Uo07`N&)44el7QBw6D*Ww$c
zJWRe9tK~i#0-}}m{uO$9dXM*ZE_6CmxM&k4J1p&bLc8jIHdkh1cWv$ZOdL$@bG2zM
zH^_13VZAQcFrlNWz7GPKRN}ZGsP2~1>XXp1VBbt`q3!#~2x}K7K}HsB0xp*)99=Lt
z2PWn8YuK>=-NWT{r*k{N_gLHa-{icJ{(91Nj!j^ij}}c79PEhWbeYZ4=}j-LGf!Kr
zu39huhZq{yhBc&3N{f1K56<nX6TcEXXc^PRi=`3FYx6TbqFhF{o+gyH{-Xu<uczAm
zjrSKP0*Md43<Kkh$==j6Yll<oKE$$bAIe2!|Io!}%@KIyhX!hrz43qOyn=TJkx}AJ
zU8!#2=W&VNVa9}YT3FvLB%@@Z#I(T_yQ}_ot$VH_XkY&e4|6&qOZ}Nf5~8)L`VRAo
zuCT^cZSv<c2t*rcWh1n}D!+F}Fq@c!x(<qufP5v+RQyAx<X?RqJQQ5BY{68Y>oW0v
ztZzMA>!Z2(;y>V@0MF%P<+S_qaoozfy1L#+Q+9vP{vEGis2<>;fp22!2w3s?!nnW%
zgD{99`Za5tEE$9n<QXZmv!*Uzzf~3vO(%hRXl?Jl30GA7ScO*cSCr*t_2pAJxk<Bg
zE;C>t^F1-T*|2{;iV&qGCnwhuo@_Z9FTc5-|2hqlT*3YewsBrxM@=`Gci_NUSZw<N
zUEsMu04ErWrrmza#ho`TAqtlG2!aJwWjrn_Q#5s6kkdN7gX&+e)}IwmgOZJmag68K
zS{BdA8ish}^d)<cBFXD8h-%pFyk<z4XjWsmsJt-$7M+1(Y2G-^HyTFPRoyhQCwD^|
zvW`WxfxKU1Ga?{xePHM1f17Vh`B?Jef5itoAp{eNO~}Mm3NyaC#lFp#!;jXzcmHoq
z%~FXJ!))^j_qmn;Q96%lC@Jace@xyCZ+eIbDW?ZNv(=c&#lTsb-8}1l)s+5740FiN
zIh{^!u#LwylnLn&K@4!IKX(!Wp(X~C==Q7o+CZwRm1gZT;)<4^9=F(+xx#tQ8x;z3
z?w8|FRl$hWtYP{3{R_M1{hy~dJVuv~CIeuiwBXV~<otEB9nUjjI@*JnDjb-LZE2`*
z5S`}tI$zu+&=8mWYourNLRqkj_<N8l+<Zbyb2U9KWex3SqmdXmRXsO7zejPw{c-HK
zFd*W$uFD@(J%P=4AubeIx0vV(k}uS&wR1Djepnw}X<K9%qcD%*duxv#x!h}si3wGG
z%suJ(^Rz0fwP4%W<*;rrlCn^iQ<S2E{K7m;Nw1SoM9YztoXYHWV_^7tdv!9xQ1tXj
zdpKRJhwG%@`#IxrNOKMs)0{_J!p1XRO+ZcK(-c|y`=?+HK^DI7$kqD#kF~h2F5_H_
z+uomWr#7?9q*^KWVMRJN@KCIQWAd#qE-$#y@uSlvKTbN|NR5O%D=m5HZExhb5uk7s
zM57_7+d~`xM-8g>mzCyMZPo&~)0;K$xP-YFpsEdYbj1`6xk+T;UHLXJnXnk(j*6nD
z=8WWR<os0z)9pHws7xXWW-RdXR6Sj^rm<?(;9E^uQQ?CAjOsnZnq$l7zH0|g=D<0K
z+ub_#`nFRFyCWb9+c~|w!vfau0zOh%q3w4O3`dtie|`@&jd#GPk%E|-3gjki`7d64
z7!`E+7GFNy>j0*qjQ8IuUmnl}gD(9&76I0^kjhxP+{!_HTob2Bj*~&k>UBG!zA2TT
zk9PGfNu0NgB6G^1_0%4{scBnZN!e!EvBQM34K1PzI>wMmPgUIb$35b|c%8vZlm)+J
zeS_=Zrzz_Ox7*$EOf}MAnl^(_JvXK`<-}=KisF&M?59{GPt3#DSiqxve&DcVJK>2?
zy3mulN_pj140SYJAMlC8z0oXk3uvbolY2@mr6Yb$MC=Zj`1b}|1hl#+7Fp1jigqF-
z@s?8g>9iNZ?0UMmYwRMlM<MA)x<ABYhsvLPIQj75!-U`5-Sm^?Y9~Fc_IX3bKfc#R
z`a5Xa+tgF(zjc15k_)^&B8_QCWZ`1bAS0uJV`2WxW&~ZV2#9q~n&xqby^bCm2dA=)
z5Ixhjiw(Ubt71OQE7cDo6}I2DOso+&jma~<xA^J=B;P2&z2i5vy7Iw`UzkhY>MC}>
z;`&tmv`!##poZ&Y?gp}F6IxuGhN^WQyI(ce?G#1LmNMgLr39po_9BUk)%wg5sf1Ed
zl-{2SJ-P#^x?T);iiuR#R(mgQfI!B@1GeFpw`WN$-Wy+xK77zoHPB}IYA=r3Qo>RB
zxBsDg@CA(;`ttW-Gm=gNHuL)hR9k%DpauS9;7}~#Eq6N}G7PDVvQZ6(+2aY1k(hSK
z1j)Q2#$%o#j{y!9m>OX-xpf}r3AFr1z8BM+BIGB#kENSiDl7i0mQ9*#JxC-DzXO0n
z1&cTmHuv2r@WL)N|Dvi`dwzDlx$$R%<)PtNgKwe`8TJL-UD2|p0`nOq6aIOvL@<?7
zZeF~#r6Pt_{?q8mK~b6h*ov;nzaZWLWt)hG{cj^5zXKM*izD>~S)LcY)1H=%f7e)Y
zuSd>$r(=FTYQ0<=HWBF#BdTAD{_l&8&&{BY#tE3g-5=LTdRq0DL}lGn`+icQ*4GR@
zDOdRvee(%&Pr=CqRDnwdZS$){HO{d{J&8TvqYY#8Jk-_6Zu@R^f5?|HrNEff1^;a5
zdo}X>xNBgeD~dDpx_kRr#At?boW1~iH7<R%gT1kxbitl4#J$j&aW%lDAEt+YoNtgx
z(6Hc>+~4Qidso!p>3VuH#F^#InhQf7*pMnlb2V5dM7F=#(pjn0H@tM4KX2aiJ-wW@
z=7mlx&8ipoE!qMuWB%Jz4&8K7*hzTsPt+*|1#sYdzB7M?-73GSM)i1If_9rkeU;sE
zqf0XnbJM!ISKotv=_@VONlX2CDSCFnf90qbJbPDfA??+0CMcAfG}_SL)-Qdoh~;Wu
zG`sTn7!fEOLV_XS8DQ{Zrv*hn@41PX6D?9k#YdWAJH#A_5Q=gE3aaT&M5^}OLFt`d
z0Nhi)7f;L-;#5by%xhxl)J;HoZ2k9n_HlBuiT)EZCq?Ib?ZID3b;xO}=Aws<S_5d4
z31oA+#G)v-8q|nc{8f1UUCbnfGA#yAh9wVVr;Nf#gm`8?mvDW+R%;^sVjKlCiN8_#
zO$3*lzOYK>{hBU0PiQ@(rNk#njJL|n4-hq*{EE{uOX*YTeh{)A%_6wQ%96xbvC-s)
z2MzebP337q;(s3Pjn7D@F%hzklVeG<Pz@U{7D%Ad(mY=8{*uJ#Q(~1~ah}J?Xcu_5
zJ#WJ3tV2jYBCNt{^K>5I59Z4K@Ik+~iXjG4C#c0$Sb7W}ETW^Lkt=tawbrsq0~}az
zg>DNb<jJZ*khWyS<mw5zn2%_L<KzpQ*}?p{KC{0+BTpa8s8$>rtTu`h;hJB|5-cIY
z#8M1FaGaJhH~wa9))SFqTPlP?SiC*?{e*NBsxqgbWMkUZ0&uga7TVc7qT`7tfg-;T
zKGf%4Y7?z}@|rbEfVDa!39Z8oJm_5iJdf2EiY#mY)9_PRnHZs1QTh~<+^06e<4GF0
zIEufVZt^)_i0HrgcmxbZ0;+dGt5I<KTjlY+Q;iPZ0nBoR(^sKP<mPlDTx;_9Z}e)D
zwz8(?ampuj7@i_GF6R+T*TotUyrr^9JPn)aezos5??W%RKV;<nisu;+hQfg0u%5Ot
zFe#bUdOjfmTxH$V^4PUhIQt4!R@b)rEN>PB@{W;_pOPoi(gsHx(56mk*RWwy%Q6P@
z58vpT4q7^MxiC^&q7su5;gyvPs+byzuDTShp{s+UcveLlJqa8ur9gc}#azwM?(WI{
za*qib*`Kri(YfzMHjrr*f=^}fPGI{ugg#9{`s!><sNju($2q5sfr0122A7}~Z&l9o
z>-MGT&svGs-5fE;mpo~tTb$}3Q5;-sludXn=6UqHPy>u48-Dmu=eXK`vi@)G#8!-9
z&0}KVey%`yRigPu`W-*-UzgMT-iIJj-xr*M8(!^8-%Qd1z{U3JfZOihvLa=xou=Zn
zLZm7&Buw+Ae6DXR_#nNxSyzyjl>E~J^PaSC6De;x`O{6>EU567kDXvWpnSUMgeF+X
z^XC841F^a~Q|>n#1Jc(80B?M{dV!{9H|%cUE8@kuiP?l2^U?=Uvi@LR`k*%QlOE8&
z>9Yxh^EaLJZ^2C8tpvUT#2+(#m6xmtHuM0w%E=wwoLO8KDu{%;cOS%@VQYb{xePCt
z7mEzjKUpC7AM|y9_X}oMS0I#32vTm+$uvgYnO-|<k))sA(%oaJli7O=B|HisW$%d3
zoghoTRi%GMtZ45K?}nw&92Uy{GjwS8ci;dFk{Zg`A`tlaj$x!edM_Eq)l8a_f6mDn
zeV}445oAn`eF7&@&5utl>fke$HF2`yfn)sYCNSMFI@L7t;FbKpqMoe$4@IIfw)lCc
zBq>@_cJ_Y9#b~AnvM#92Oe58bpYq(p-FebZ)dSmnk|1}xzCU(q=R>S&$*l;Xtn~n}
zqENyJI6F-&o}j59&ScVqVAuN>lBA5vgn$5~cRl&=?|ygmTUg#{z7TUcjevamKFm}(
zYk2Iw$U{0lz9WMfW96PVnRL9*n6zx@11dJx3#mK1gYUE^rC*+XqW~dk1FT6C$<UWj
z4J4jFTxYDBq|nNsyp>rrxg`0$W;viuq#AH2%bO2+fXo<T7bN9iOj_n}2qR~lLxfZw
z47~BszKkSy4Gkph8R23f)WFZ#6SYsIV_CU!d@V~7PW|zqeX@nI-ZNc4P)a%DkTp&p
z9}||zD#-et;a-))MXtey-ZzC`KYb@OE2f#sdC8keE4CCV9Tx)Niiph;UZCzdqBog}
zTJ<S;2B&I}z33yt$&j*{dZ8L+3EahXiy#$A-gsZ*Ib{d2JFCgyC39|a;{fht7Pjoz
zIxp4}{CVc*`u(-^j`L>Mo4d*}6%S1o(s3TU)IrerQB}b%XM1}A{a@T1X{!oRS@snV
zxfrGDF6S=oDi91W9aKq49|Lps4i;{=*>_IoZ`+w3i#aZ>(i&Z+G2QsIG(_uMF3hQ^
zfv9xhl%t6BJCC<_?q4<=0v`eAUwOm`%hH0XrvrlRav7u5yE0)P`%YbJ2XbngYn;HY
zT%Y^4L64XaM)iphgg@wquCSU6w5#sLMSJtifXWvMiJ6IH7`3Ul72xs1$6`Qa>{Y%_
zgv@CDANO5`tDC(mtVDV1(Z%r1MjvMu+Q8-D%5<#g;le>)!aGeJ)VsA}g{Y&4Nv9{|
zOp@OA2OZ1i<<bn~x0`pt_QJ{b`Io5eD0#h^OZ}OkK0<A6YSi)Ejk|}3$JU~HqWQg#
z$7eC1BA$<oDBTaf4=iA3_im(t6R+0W#<<TvrNuwY{QZxHGfs&Z?BB2~{wD(7w%2=@
zn%$QEtYJGtzg-9Bt#;QRo<#DLgLAu_y=Rhtr<I>`)(5>?<x6jLcXqnQj7O0^d-Xsr
zyI`#F)Ys0Af->bJ6L57i*>aUo%;d98t^R4p66v!{&o7EA1AIMSV2tlmryuu^TE5Is
zcdK>l_kHt7qWFm$W80z8^_tu5(eQAlL~V7`+2%kwjmI!orilpb1cbrTnu>efSW|a?
zFCvLQ^rx>Eq4$y6?+3u4<zt3zD*3(-tw|0YVitzVM3A88Syjp`CjT&&80KzfQBG9`
z8SV#76mTw6VC{DIa4*S2@Cmlw0ivsiVF&AiuZoBRSnO8Qg1&$ZyGc46TZg~oGG49r
z7y&6GVHZ$O?~#Qs{k>$Vv3h;)zfhGlPOO2x&P&IYKO5bv<2hWWM<bezdTt<hi91Ah
z%T4;qS<*`MK!HCKSla)h+(+|{I1_2X+xh$CUf?>AuppOt{tKR+7jH1>v(~L>sa94q
z(r=Hc{Hx4@PL!pgc4UzxGK(gbv9}RY{4*{z|AItg$Amc{XF>I&Di~&-zlpqEGio)v
zy}aG$jN2C_oS(26tIJzWn7zI>6chD05FP4i)wO+DN83Q|hx54olQMMOoFRQ3C*6Fu
zZaNLDA)cC`{aEZFeDOpf6EtI1Hl>uxiY#IFwzT-kRhCq=`9;&b1G!WNLYfF?n~VSp
z8Jm}C+#fh#96Sl<WK^8-eKqiD823t(%ix;uLv-1r^AkwpYoo_9|Bz`pA`7Z{(E@Ve
z(SB61#)EuoX67b_D8^W(sGdFG-~PS^XAS^~3HQr(x&QVbmPO_rxwY8JG$cvCpN2Xg
zrU8X;6NRkLJgG~(a}*IkjGJbj|7kD@hZH`iv@|V}iF~6*zxi|3Pfmv`uXL#HHy=UT
zDpDB)>!p(V%c4Jw@I($l`#kIf2kIeyK~DefK<>0x)y{jmITUh=#`SBtcpb;Ru>j)w
zZf<r<GW9|6)f>t4Tmq7q+v8I}G&Pd7bMNZJ?j9dDUGkaY%>ut~orf7LU&FHay7^^j
zj03u*LKj=KQC%ebmGQWbZ8ZIRf)1%-RsJ0-DD2KAZAvWK;m_RE6JwQP9xpF1#kVKm
zD@lt1iR>pfH4WHb|F733%n&vH8yE__q{wn#>YHuenJsM$$y;gcf>AewFI^YA@x8op
zwsOxNt*1_t25>vjKtVC;&*$rfN;|?|8HOW#waBU*|D4(HwPTy_N{I7kdahOOHCTX~
zp*wDK^&Xce{Rhfm6*t4n$3N~;iq0s&1KU3(L+A>y&X79$7uZmgs$&jRm!*k^EHtI5
zKbc-X9*V?D6oc^0v`J2Ft%-=7`JXi_U3AA@fv{yWN6XR&h!3dzs+7ZxaL%m1)4uE8
z-#1S{qI5l~DC6?<fdW%QLuve|)5BZK=SXGq2n8IIr2D@=%Wvj=>VzJ>0hL{?hli*0
z^Yb(LFab|&><iC1)qcQcP2<sHMlKFm=tjx3baJMYQ(2(}*oIzXs4-R~A+AR*?>n2@
zyc{^XE`6MFpRw^M9@^nG`F?DeTjBN_)9#)PX?w|p$0wn0cyTLKTu)00sJV$a*rARP
zqFnYns@w#ug1SyDCSW43P^t>-chTV>su0pSr*^t!%%`xnVBmnop2a#*h_3<6`QYwA
zDd6IJ2Fne?NLH4V(f+mOru7tFaMvf`cL<yj0mU4JsGJ+Gjud3->5IjlCM9VLA2fk%
zagyc0o#xV+J`r6AYm4)$jSNS<sK?SD()qJ{&|s*EbEm)gg6J2lh^`7zG=kQ2Fan9v
z9LjGY1@k`(qkqyV2O`InSvk6T!RK<&KPX+rJR*|Xg3)L5AiRq9Yo=%dolu&3aS@@;
z?`KY5QRc|7;$oiG{fyyP9A`@jr@Qm@25&GhI3?$QJ9lSVPl{VVl`3%)6M;o?M%twJ
z%+rscAu~$S<77O^Fy(6j1=B8<&A<&L{FCOam^8q^7Xq`kEg9;G+prg139!l|roy8d
z^3QlD@Q}5eU~3CC?T8m2d$(($0UY+7oYdBs>3l*@ARx%4bX1&GhE>|C$H&qP>0SXM
zKY}o*uCr;ln)N5LxM=x8{j^voym%VKGJ{|l0o%|^ihu&XYURTst7Kt%diw9#$C|T)
zseq@}a6Wfl4#S_h=q)q~6(I$0C!p1L<b(V`HEy^&x(f&@Q#>RJN6!52^B78Av{oNm
z(YNPTvL8O4v&7eD@q5?*NgN|ffYh^jCQt_UA3vaK;QGYO>rJFAPxB6&qIJ8$D~-M~
z#sfj6>-a-rrcqX%ID{GUTZO+xl<V*sxPJWc%+2?5roOMQs&iz!@%G$n{(N?RjwuS<
zPqWGseFMJI*v@4>97o0nnU_miZ|S^P<0YC@?hAgu{8{VdE&+sq%>7=W?NggW#~;bj
zsh%O(aRCsnfiKZlb2m5r{kk2BLFsJy>y27m&zyG=@#zge)#2;)TYR#(*CRYX0o(aO
z4%xkwiYj4W@58x1rzksXYkO(UiirR|Nk?Bdi%ZP$7A?PFu6ImviS!4q0TxqF;JGIb
zxA(UZxa}QT28~pc(d+AL@?ndkc|9i-<Y1pu*`B=7>vAhift*uwWDito6OVk-$)xqV
zyeqFbFUBXEXntDDbkd*Gdtd=%H4YNk8R(r{*ic(&QnQwp*_`>g81(!R({_*_2_F1}
z`~j9S)3$u%gdF6YaTZy?O$PWzyAT0mR-LttpV8TORISK0Ryw@OYch*($OzjB@x87y
zPBSm3Cq;xO%E02@{tp`&Sq{yA@iJZU(P6nX#$=S(PoXV4ClpU`{J{Uwbl&f5{%;#c
zm1wm_l^C_EB&GI<Rij3!z4zXG?}`$o#Eznb*8CWuMyOpYRz*><SFNI`U1~h{_lM^n
zxQ`>b-{ZQk>pWklMa!RoJ|NFt6R(&LErHplo6-E44b5FQmOBqQ`lXh8Hk#MzQeTCs
zFo0{7Gvi4G_abb%>D!eP^f>|(#$0mk^i*oW)M`)D1Oq=kpq=iVJ^`8Z_LdYc+sA+K
z=TMq!RE!@L566Mx)oyE0zrb+O3gwf$;1FiZrV@28d_f2Q4yC{Aemg^aIG#j7V^OFt
z3d}I?^j2t$Tl6abbL_1WgPMbembTQ)o;C>0OF@@yUlHl!1xm=<UsV0|jaw)b)V9<5
zDQ2z_ZCGCtn@@{?3cVB`&Jm(@c`tRh<bB1?_OcYT;ioVs6nez>N(i=@=UuF)$H>t9
zQ)>7kw_gj6YF`)ETUV6W=K5Z)+VMT$v&c0Jn;KCZ%71WVXH4?fR4VZIkC9xqNzEt9
zogW%IkMDMVX%qcz!6T_n<fxnfJp2=;C1_*T?dD2iDdH?ig01!Mari#`Z(?T~6uj6H
zD8VVgU7+O@>ZF*UxAOB)aq@?<t)#skM1FtMt4e7CUS0XLv#qV;bDaZ5y(33U1t09#
zxh!2Z;16}61h@4gJm*M0e>;x$dYv0b8=D&as(XO(R`S;e3hB}QU22sr6|L369YT4q
zj64It_zY9Ba*8KTg?dU27SX=yPf3Hm<}O+doYYGoVJCKU<Ztj<41Bj$*q=e}pt8wM
zdHb0ToZ)s$i>&W9>PU00xv`qmzTHZo!P5`Im7({KIvb$ryEiwsiTV#j^iG$65u$_-
z)KWdd$PQNshak=mZ@z!jmSw>nUHF><s=7MpV>{y%IvJq&%{`{~VV<P_>GtFqMwjUC
zSloV-XR9D2ek`Y+hvDMZ%Xh>8W3@>OEf)&k_&o!_L8TViysHF6x;)FTcY`C>GvSwA
z4iS)IdZZqxM9gx6#P^j{Z^iG-lK>_~3v+Wh8Cc#3Q;A{P+^^Eu^q7b!4l$83Y-2)J
zJ2BfIHL)J>aAb|=hthj#4Qo<g2!8y{@my{Dr@=w*j;1G9cgf1u&;3)7RIEmixU=BZ
zo{#PmlD(=<T^c=6sCgpt_?1L$Ss|~5%9{3z-^NF;tlKUJ(XBsy>MZWd{awp2odmAP
z73QVCPWqD_<u&`y2grK{{#Nb59vt1D7`F+z!J|bs`K?+ngnGU-Hm1{R-B)d|+vEzs
zw1uCYZfvC8-PG27shYxSX=!C@W$Ft5;3S@V2a|SGRHmhQ|CCPKw45h{d+OXl*a^w|
zB-Ji|ZXsqY2GHWP|8;5o=TmbUhrRiEGzyrknG*^PIQJa`4PW6fP*h$04Bv2_E&O}q
z?&&FgbK~8}>76n%@7I1s&WB+#>&QZ?uN&2-N<jAZpLYmq_Evh?FAw^)=R4q`?JWFN
zo-F9(AVw9+F{W1#?$xKOPGMgPpjMh^Dd_fM)XQGI;_#-u*H%AiN*fGm8=ld>CpZ^P
zYdO7k=j_ucM-0T;AmeMBot}P^lvZTtWYAM%)VD*@6xskjo7@}qUdc_asQ7a**r&Jn
z;RCZL-;-7ezh;|F`(Hk#|Bw$aiC1H+#w4k!7zcXxg_3h=R5!o(oLX-#-#NAMHa;=>
zclTB<)ysjprO~?Ie{7cAjFMFA4_Iw=^tkjgDsn2RBzZ6o8IX4eL`E2GkJW|wi?Q!-
zsT%iX7dLo{B#kk<qjXS(@AcesHfOFCG9lGfRX;pMDy2V#m*D~5{!A5o?#pYP*j_2N
zc>ni*Vp6;flM#d6#rfa3iSJ}MLX|J`45HaO;H&P|@Qc&-o#$UT^|SIrHZEiR@l)?O
zqo^&&3RuH`Zc)iykeU|VZa*2~m7ROM*3&g!yV?(49(&whvlRISi*-LzbbrohT+3c^
ze0^O!#oIq&>g8~}Y{7X9pD<m&Wo}|69lB>RGiqTh=E+;ER;5xnJZhB;@G>|3>(OX5
z_0QivERUI$28KUPzK;j5TDXv<fo652<ut2B1`e&{lmd{ccJW%w6y02EuB<!hdCCKs
z5AVbh6#Z-yCjERq2Tqyg^vzE}Jl@<v(a1yJe$*7;KCbN>@TZG`*O4niyzJtUyT06S
zdlWeRyyS(Ra(qw^(u@(}P{AwzGE=Y^*0YNyMZBX8+ovCT<BxW#%h9T<Gk*Wt&O<yG
zYX7$9)6suPnlIzC?Z$C;9A8A2x`wy2I`ciDdoKVu=>Be8Ip>Eefy5uB9CGfrJ<xl8
zRwBTtFIsOJ{}Ve^4QzC_tLyI1EU5OT>b7%?_4$)hE}GAXjKSWx&(qAmlkO9Yjb9u4
z5@}RbY_|zkyEk)^T`n%P^76V9v2c)lUViPJ{R*1B68-KVWPW#dJyc7HCfy$&9=!dd
zyft|5eU58ukpFAhu+3TS;RYeM*)Kgkz5*iBXL}ZG`)IjvBe-keMs0uO@haTP)bx4q
zPnfv=W1Vgb>2IL;FD7#`wUHO&+L5Q@4Q}|nP8cQ^-3VlDT&!mCKR@_eG;;je)9lT1
zi<5=nRY{4=Fn65(pfK*^P7CMgQs}>3fyjfz0*k8<xTv`L@!cK{<*pPq>9C)@wW#qK
zk+PkZy;(1BZ@9M%>w~eXYMuCuve9IfI80?)4F2ew86b!GBCn(<uc!%XU+H!bUXX)*
zTqf&zh(U>|yXt0lMf@skIK)e2l&3!0oDW|Le6*QI@jB#7XGilQTNNMWMZY%dQcHLS
zO|a$gyFRexg@=p?n|dGW@Z0^1n3$e-+Q;A4c_ObDrQtD=wQUV;t@WAFyfEtJpFjWj
zh`aOn`ntZ`Y5Se27!S;@FarEuauI*dJ&rlJ@k-+Tp}Bf<bvb)eo>@?K`*_Yx7N|PM
zoA|rQQONq2kdp7SMC_{?cLsq4i4?+!(UgECH}iz-WJow@d=5W7)MFPcxmP+M*<4-U
zo|&060!ROwQXPTRM$Oxc^<)#tI$PYq6SY<DwB%bC@MM%ec1=oom$1k_{0f&tRehux
zaJ@o!G-W`i*uB=GW1I|nSBO$)#zV}C(skw2K>8X81a16(ZnFp%Z2p%IG`w}RLTS{b
z8W7!JO%f9`)Ai_J4o_B7^BGv5p*MV9YrC-=F8cxxuGyKnl({=VNM>!WKhurQh6sr>
z)RvAq4u(~_OYC_*<Ez>A^n8G`SDgYs?j@q!%=9Xy+q~=@%xG$>QL)GTNV8Y<T4?$A
zt$q~hNI@cWp6MOw@A&#?<3t;nqqsCk%Yrq@VieSU_7{Vk42o{6EMYA-r7-pOJX)-9
z*+{;>r#1~)dQ+q7L_f^<@Bj0Z-8~`8s|RQ3JsnPr8M}56H8P!FTsYvm=;2s>jaN!@
zlviR0tK_rjQgSpv2ioX*g&NY8yFnVD5=6QR1^3^!kj-{gQcwRM#lj}tlQVlkFMCP%
zTaqn^3ZDi+t~SPTvj-C>hBbY$92IN$E6kfGISi~b3+@v1<n7p`%t8pv`T3W5%118<
zFpiB>&8iBm`WHpvQtwPvt2J>$c%U}C`|1^G)lP(N`uB10PF~TBhplhlKe`W%J1u~+
zXr)~O=U^&c_FRz6A|8vyA}A=HtlnHN-85~E=3VU-+$^>C**QCJUdB3To}a>{*}%ib
zQ^wR2{-c)Hi)S6d2i@lxJlZ}{<V~!193v(cm%?n26{YqXwpu7iLZN>kQtOU2T?rAG
z?YF?Q>Quyx!z?g|P>&yUW41T3vijis?cKnJ9`8)h!iag3kuibQV?kiJt%4_&R>#de
zE3>_>2I#HH39!LeVvVl>{fkxT77qhphv@?HvhI$aN;VduS>r4iv*W}NgwUL4RkJHY
zRf(M!JDrj1Wdgfat&`T4b0>TAeTklyDW}1AGL-id`TRXa+@&qo0>y?!0Cy77_RIHN
zxp?c=&=-zZb`bV}=|G_$GQ4#fDgE{wy;bF?apjyRloK-{;<@)2;Pg4Dn8+D}`f2Er
zSdo|L=GZG}32N~?REUfCh0<n>E)J~kbb-e(r&xuo0bFzk(y>rHAa4Vg2<II&hv`6{
zVDG1Wn62@qo2rSdYwg4~H**p-?t8Dd{Aww6)|?Pk1lc30qi+Q~55S)>yk!6yffIlM
zjb&%>Q?OI0r#(CvE@9KVR=hW4p~>SagXXV3Mh-SMvdV<5M0;<$Psv?Gh=;ige+tk=
z@w%(VCAwEj8Vr4w=lD|E)%Kl(v|NePK#!f4C_@jQR(}M9s<Lw{@Bs{#3>6-wWtr0W
zmwI>Qk$FTgVsb??KC*sL)tF<>Nb1Y!`V=|%$@ZLdJ$&~iaduta<M<ZO;+HIXh4QFk
z8zQ^V(5)NC-S#lP);gKa_C~qW3+CsXuYt@+eXhhRyN8K}yZy(<L?SK<)E{}oHeS6H
zAOZ3#e=)zjDr78I&F(t>Ci<g8gJF;}(NJ_Ndr?X0ZL>tRY^1rYh4?Y?2lBJKd97zw
z&b!#_oin&KLFKVP$??FTe2Bwvg)LoHVuaLl!@hY@LsKK=l6=;-&#ZlFbp3S@r@We$
zfjb<D?aBl83V9-drp~6^2#b62YOEnt+0>-Plj^ypuE<wXxA3t?)sy$6uf)0w3ma$c
z)?B{+SMkF`n0MRF^L0ilQ7D7oZ*&pj#}FYYXt5RUJ)kM*MLs~UJ|{mUWBywXPGCMI
zWypLA<-nUxnXs=VHFaG5T)zGAAxD64j;*41+l5;Zr7eiVKy-V^=FjyxX6ceg?&;=C
zdpI!7rQ&g(sHh;8dBT<49p#VDp(MdM0$t!CNBxPG544=yE!$@V(thygv8M??Wk!2o
z+(<YLPHMplTtW}w<{aH%z14-(ANnQtrxMzQB4AH=j}iKfL2i-{p#HEwL>6~cKEHUY
zV~6!4Kz4ry{Wh?yyyLGyqUG+DTJio~P-OGKQQe-$Z^5BQIvjf6h^es&u9*~!cX4M{
zVQP0eV6q*-JN!k(a=*)V=_GfCaR1GQE4_pJ5FR=%z+#FgRD?NQP`zs4;8`G=C{Q!!
z<=i8Ujz-(@Bs;~m&`(%QE<pZ?Q;4U?v~6Q5ui0TDg4k<X!Xy9T$>naACyfO-9p0%L
z8j>kw_ioj|PtE}?yruVyD%;i1;K;K$JY(F-{P}{!V#F0N8<YE2BzI8+iHBbuz@6hC
zB7_&f-p|OEc7y1T)`x6050#)s;zLrcoT3JSX_<W=mxA%pXNo==z&|P(xJY!WoIi;z
zLx!^pgMb~aUXH~}_d>*m-m&{c<<*la7_P7C%l+m$Ue5xQeg2<cui%)MWFteen4Zm2
ze9HDEM%i>JC;^PQ)v<4|8bkQeP*RrM`ON4y(%7D)E+333|Na6zAkw)tt2FVvp?n8I
znHsDM4A(y{6=>jD<sw;lwr~X8Oh8L`<iwV+!21CvB4qmry-3%8LT~LHFe?_IuKI7m
zQoCABB7?7r40E1UTXcl{BRnGk6DjK&7-kY$s_)>-RY9N^peap%8F;?MyL1@c#KD+x
zFyJ=!G&T$rXT7xyd3SNO5+G$6`&(udOQ>J3TG%}y->LAb5Z>%%>xBId3^$iHj~k^D
z;vO>5V0<thUhwb<&LRI9Z%O1%Q{-yBKu2DFzB;ZDLj1FT)hXQBSyZaA)To6!Tgv^c
zfGzYx2Vv{+)2~J60r2$F0|`eN#bOY3X|@kJ9dUWZ2dT+~Vt*_)ESCF-wdw>(TlCB?
zpVOn^pn5G2^LFhKQCBZSnXwP3E3je!^CO0c+7Bf7V4kT*5DU`(Db&^OCF5cJvPBx*
z6&CFOZ>z{WaUy0&+Mem{J7TY@L|95OsXS7L(n0jYH)sEN_~$XP5QdkZ8t9SdWO;<?
ze=t<DD}pi^cQh8@gdBc9W5yG@HRNcxIc-jYmv&6*LeXlVy@n!aeOTjU%r#a84_F&~
zQND~gM@tS;3|>$}npZE^2nVaclGBR1?cEbjy46o01RGVb8)mtV4-S2E(aXZc1mHO0
zzDpRC?ehSQf>Ab&{<3#StdR9$+v}B~Of7-iA3-#m04FL*M6Ggx`TpVhxT*Q%2@vJ@
z2>0$*57a~zwxVn`n5y<hdSNZo*xRBdVhpTL=ZG<&Ce@{WA1YKZQYX_iJ<QdJKmCWU
zr@0Gz82Ej=AyA);rl@zyuuoYYl9la8V^UO3je-wW7**hrsw&bX)UosCF`zz~fC|p+
zkRpqsD#D_@AztZ+3al5QhZhR*$+w}j$sG^h%=7CQ`zC$wE8;&WvHByjZzPyUf}-Xq
zZTN-m_sBo9()+(X_M*<Bpxobh-7zWVo?$DOxC<SI;jzK7i0A$|8+ndZCUdBwbNktU
z<#4QULYOO!Ax@2vv8=5AeD*?$K>^I801izd;=nbJKjCmFxw;pVD&&N@7wwWsfpKcX
z02{6=#Ly52o)lG7!f)j^b@YPo(zL;3hK@XL4m@g2+Ae0r=<8Z^@aOyRn|}hQ)|Xi~
zmz(>q_sN6V7TbgDTmKuzD#(pGk)EH!J)AI`dE~-6EdciNnS1~0yW8p9o|aQ;>5XQd
z_9!nV_oO$O%5Bv<AB57md<`884P(`Y&9uukR;k-6hgiToN?`X8`J`$goNFHST6>@w
zqh;mG0k8B_e})%7OZ&Cc>Y-f<bYxc@DiOlNtAfwNe7`SEM?Mb|ot){zE+n<Xk22H<
zyN&gQe9S$Sw5Fqs{);j7=|ULK2~lJ%B%xbSM?G*3vwEdUw;EYv?2yxxZ`Y>RtE*1Y
z7f<~ara3X=5fl*-A;r{rNC?iC3*SID2ObS$(Ct#@Ycu2FOIraI6|~V(mFS~P@x0nD
zpWW_E<BNs0dTn@feY*lQM+07}o1QNX_kE6*v~LLHo$cr{|9>%o(g%W+TKb@aBDnqZ
za5!*`Wmit_(Ic;_6LXA%RL-}J#GR$EtJ7Wcmr$IE>3iPPC0OOq>?O|odILQt@#4-H
z0p}7Ro|AuXiU5AH%Akt5p0x`t%^i9$zvkxYxl?eH2yf;yfY(eN60~skiqjGzuU`;}
z4CS5prqgx{I>y@|sL7CnQ#kL)l7yqmHVv?2TOg_Zex6W*UN=~oz~SEBM8qmswXq@T
zOv+(0^NW}hVWVM`&6ALZf+d7f4`Gu=ut{i4yL(N?K@sN2+l+88JXZ~F@A>bNu-JEv
zGn!@QMpwf@-Rpj#5dnN)N7BQ-ACl`0I3m)bpjN}o+lSSejZ$bW>ex{#swDF6zuSA?
zK(HbROkJjn^>Yt6)Y%Z73Gnk0F7KEr5eq7QnXvQ@CN#W-Y#(1R>yE!N>k-@={O_KL
z9=#a?^FndR%NvYs9U`VGG@5mN;tD2pp6GLF`43Ka$m;5uceJGQj@7dL^J4OUo=-7`
z*}WMvK$oX8fL`dSkQ0&U@Ll^V!UlF~iCW>iXv@NuXkZdCxCjV(TBXm(@zT)WDXpDl
zWKC^&Afl%9aw5pBJ!u>)(o&0U55JzbY2T;#sYqYq7EBOt3^GFpr5U@<p0er=IZ_Ma
z^5DTB`Il({Qo<m?x9q8OJ$4-?{Pdvnhs?JGRh6i@Vb06Qr+$_0-fsFEDq<*V<Qt(o
z2-tq+n4u-Jv!I}0WT?2<aG@7d(Zfwj8l+^RR6Dl)s5Ptl-c(b8%nZ<{12R(nj<Ui3
zJg5;<7<g>Jf9N447vd{u$xrWS3wtFV2gf9FB$wbKb=RW(?Zq55;-mbvWV+(n_jgL4
zR+fOKZD79}T6NRO6uxwm=>P&)#RNzKcokU>NzNo7z8lJ0>sBMEL72bVUe4jn9LLKm
zFBJc*tgIw5#990q3sDlBS(3Z_6GH|6+jAxqWX7vV!GIDv{nvT(PY?0GZ<2C2^4G^D
z?Vkwh7V&uxl%oYgR|zVrNpZhi#Waq{elv}{Kh~uw<LYV?Khuz@r=BEUIu?il*%CR3
z*(Oti7)fvoR-@t8hLg?hce8!tK9dt%+Vcbw>}K`@8oa>KM&qV)^D0qgJpI<!7se_1
z&u7~GCjH(ynutjaCqyOodHBIYKBv$OZRAj<cj3=t#~m#nZ(6fFLL>WG|I~SK4}W9g
zRsQaFGSpISF!JPN*=+Z8%Vl14#7_KGw7lQ-?=5}k<Du9_dRukSy`CsNPB2!5L0K0y
zNx<*%0ufznT|J!UVeYq}_2)?4j^_F}$(j$d($1MA^4aMvW6nph%Uy8%hL0pVVEp*c
z`B6m9PAPvv9}uZj-?U0pRnN;UMFNw}buc~0Oj+e3liM+?N?x8-la7*TzC2&TERZR9
z!dtV*Zs7t56xGWo#uz{SZr$$)KOz8%mkPcQ729Y3P^mmAY`H3v%8`GMw*V-#XH<da
z5xdLPFV+2<!JC6lIJs`a?73<gPm=hO#>Xt%?{0yGU!KIkTmy;q`ANBT2(iJ7W}A7p
zQ`(M@dX?YQDLosyKtj=JN0qOUohtN?_hmv5G&&l2X9yjnPzj0}Aw;bXX25gcAl9ZY
zO?TUrL25Mj#qriRgtO|-ljjK8+tGIz5geI?gbfb>Lcjxep2(q2rx}92VAZ8Go18Av
zAyaOYVNTiIj-t7YfZuos3*p0@Z*MBF1G#ev8kOV!>FwO%*h!sGkucQV6e)2tlo()M
zOFR1-l3HxZ|Cr-3hdnnfdGsH!PkdCk66OUfi0U!d-(r(P9bqVz0G4rvQLX{^U{#gO
z=Ayy`WFOuM_4XlUF+<Z4NnP!=%7tXTx>_6C-Et{pcFudj60QV!tTvD@_qT#odsf&h
zVenw3Cxbzt;VHBU{Oz|(Z?h5|B!u*vsh%+zWAB@nA1SC3W(ptm(-qlvPN-#w(z`S7
z7C&>hWUeWKmFh9YnWF=rDsltG_ABYvO4&g$%_%tO9b`poGOZV#8e}*0*%wbj0ozA(
zr4EU|)1wJinkG?cpIG;BIAlD$N0LIJinz7DOmiDij#B82U5*YHR1y>FiPrVB$PNet
zs@KKXk;<Bema8t6h9Du9{C!!uu*2m=&jh|BFJZ~He>(-N5vOMvJi)_-5GH%EJ3Yk<
z+i(yxNi1|F-DrunFz+EkWH6hZ2zDpqD687T)7qt}RfCLF1a~Xtvk)PC$t4LIeE-3a
za`qp=QrmCcV4-)P&z7UuckZo|g2_k>Evsv4WWR)W4o(80pmcifn}n=TZuY-(H)r1Y
zGUq>cgRL9s-*8(!wuF*-c~1XZ3E$r`GlzR<r&y6jk2oJ)#W3`oc(PkCY7R`ju8O*?
zh%=wtgse`?bi<@s(JE=*fa1cYfCB$td^bI;DpL$cs_=dLemuyj-gH8Av_Ljc>+T?{
z+-nwuEvk`7|4y`Bbedtyf4OyDAF3)ckGz2Z4oS($`mu%4=T|@L1@8Iua_*WoTfny~
z<V(F0JcO+ArNc9YX669ZxCWi9ny0{fhmkv_KnCMx=^7`fZ1S1?k=JiQ{M<r5d4yE{
zF1j__F~^@w<_?18lsw8JP2it9^Dw9N!heE_C><3w)AtKropaXaPJJiC3@{GdFVT05
zPll!s0$LiePr9RM`CzI85@uO}_3)<P=dMj<{p1{F?pN2dHm`%kN9g?>?j3FrFuBs3
zXEI#|FP|zI8R0IBJ5F%*<Y%|u0>{A1<M8$ce}<z%CUkP1-n-?_i#~zs&g&LR`_=Cb
zEg@U$%nW$X&S^NUFe|Cv%dJ%MH3<m`fY@Ci2lqlcO<x}K$i3cdm1~P!g^vKwx%n-3
zEl12lG;m7{PXT@N`o%IGf9$~zlMNcfSz3A<?hqvhB>+Vzgm9UF6pajx%t#^x5279^
z^0E{O;cD!_M0I$^7-SzScpcBN`KaWZ89A#xDc&qDLYGb*>W!}c;1M9qmc;GsocXlJ
zL}@qV>~M0Cu&&~7Ik^tMba?*!1Ir_Pf0<+2y$kmRud>PQkW|iJSm_@2c(i{MH3OgP
z6I+(7BsZk2%=6ks{pi_p6lI}c0Qk`(*#Zp}qbU;eXd3q7%me%b&j|gB7A012;QjN4
z3FBFZqk6$)*B^SIhLe(U;PPPfv?%hJ$S$6ARA^mb>-Bw^2Bn8jm_0M?mgOlLqmv2Q
zB(^7NxOgp+5_jOD2+~ytLW`3|Ft}fHP5UtSw6mxvS_AswgRQ~#)?iuop%6pBfhOJL
zO)GCE5tu0#=EG)66@@u5%bhIlKnW+WveVge5r=ay+N2P}x}s~@shQn(!ds<%S%S&0
zMn*IdzMV^Dd_esZMBNz8?iu_Rb=y$?_yQeN_PB>jE3Dm$jYxn#QD&4z5mnEg-~iTT
zx&w8ji`50kegecCv*x!KiRIZJGb&~Kb|XIaO7Eh{){sJ>X@ii2pQ4_D;a4ZcO&sHq
z<6DzB-86)6K|Jd1!RpnOp1D$=&zJD?LwEEk_boL(c)CFX<cUVuH|Z<v2wIkDKu==G
zjTES}Y^&ZJvRlke*WDN4=!T`+5|M`QH&Y8Ezo698L#UAc9P4)?Q%lQ(lc@_)wFK$7
zycdf-Zji8M3yb!!_2Xj$XyE>JyH?ZD^}l_4jh`?qq2q*Le(5fvxYOR3O+<|cw5<_(
zbg^(xl+nqW2Vbm9$yR`{By$S+mDqmQ<Bt}oit?Qc;5ne$#zv#}jve2I+T`)%iu{|2
zZ51Ib%=?8e$?_)8{c25waX(>C6CEv`oQw3=s+Rf8`R|{dkzWyWw^aZQp*%6=4iggi
z7x<sOId1{&`?1d!zeKL=8rQr&7*tK!TY|$6BjmQIUd4=VnCcReD&k_QPZp&RLa#C$
z++W*2FQ7Ml+ZAW4GnHvHqG$d>(y35x`mgiNF_qNTpUj=53yGVaPD$Cw+IC3`Yc~Am
z5joY~V?-U_-5%gN-<&g_TixmnaV9pdFm+fzI8dw1G;rL*m^wWeA`-m#nV8$(Uj(n?
zBi%gPo6oeM<#%&;PQ3i}>+g8%kzv|-sZm9~=BQKsJ8BQu?=0oWKbPm~FApborOoHU
zlUCw*>KhhX{dR{AwpM&#CfEZ(Tof@8z|SwA+eNAr@eNaOv)%k$o$5SsLN!G01V`@w
zUJD2da!RTtz&r)XICAu`dVyfFP55X*!3dy)LH4f}#k@vbj?7uZ5Z$m(NE*sFHi3q)
z&5j&-HXc_nF59nr8n+RY-dGOG0Y%YnwgC!OZbx}2*63DO{KrY1&J#kYDyS>l?jvgq
zIRX?!c)g2G65{B+b+{<3PC@LRbXE<yqbOA3Q1X%Y@ci%ZfG4^-Np>i>-q63RK<MM5
zotd>SCV0iJ;HJi)#^7`=t4|HgMXL>yTR}aLW1^_?y65WB7Tkz;@+zNF(J;5zpWS-^
z0Wxym`m(EYc6#10@HU9!E32R_CvwiBSmtHA(J+TK^icxcW2Sg{vB{8tFBW+eP@xSx
zr+fP(uv{tnLi(N^fMwt%O6sn8duP&M3lu{54y){Ii}zwh^KMr+{hQv9B4^~IN9YE#
zNBaj^*7ITiq?b65i}2$=9x|r{Qw&k7Tk~J4o3jFZnFwCV2qjE;WOuMO*fZmxP9Y^L
zd)YS%?||jz%t9S8rWl5hed3qkp@;QVu21(pO*g=|{P?Z;cHk|)1I*gH%ybR14;Ik=
zNsWm3Aa@<(_tb-?0SDcNUro%VtsNHd-_7`9Y}P3|{N@8ej)9iq_ib0AB-e6P-0eZ^
zh7UD6NY_YL?^8ca9~vO$-~+A3lWCa0Q10yk(Rc~nYwg+F@bcbC*SMGshdSg16g<+Z
z8y|wOjykUnBH@4JgTiAHvuZky_BLKk0m$n7@hwh&`51_AJi@n9<d^Awn<{QrPa-)k
zSsAlpR4GBNC)})<D#h_)NRrd!?I-cKg3a|!Sr0SCODDFJY!TF)C^t+Igd4$0%G+r~
z<b<i&{&{)sABgGv`*T>vx(zr5q@|@@$5mJh@{<>D;4A+s^<$`iMhk06J}8<P#F#fB
ziWArKqU=cz*ZIsfk}7_(N7mkShb=2O5cBNvb2vE03o#|NEbLwQxM1;FuLhgfWROLw
zqxpMzbGMM01cGEYxUaapBC-`*-%L;b^KT!qA5K)0Hg<Ikf>L1Ta&CjX^4uNiAGJCs
zFi(ra4UE4rzV(nuirl<9tJ|6}wq9ty`9(}!P3hnm0}((K%6z^rxi2VBH)LDmZcxOG
zy}8^PON`;V^XTpB7;N}%(KE6k&{Zxlpn#uw-)KXpJ;FBH-GbABkKZj1ndj0uBcrgQ
zvLc#Q5`AKgJY*?(TwH5-I6!sOp^amS*(!#QZOqP2zRIc^FRGugYkpZ2^4z=^(+JW}
zIp>wI@=owe#gWL6^@1%+Zlddxl9EKV@}EbJ{|IE_(ax8dlZ9`UWLZDY)jBPZv!`=R
zybVY~><#Rb#2d?ZDi&P5y}eVkMn9N29?cvKzEs)*4E62xc$Xq3ePe<fWjwn!BN-ft
zs(?NLofoYZot&J+T)jqUl-ya>fy0hj1>br!Q^G`W5h5egT&}iK+@Nm3@kM1<9`&#j
z&a+0RY;ci~FN(x;SN;!JaQ%lF;DGy+O{-9@Ks3N?pLZrN=;-nAhf)8WWFj@$*F|Yc
zc_=s1ZFjrnKjuU=dC+@30DN3R0p}R*`{rhtX(}s8QJY+(qX)3ui=Np?<!En)DvXDC
z+S_0h)KzVXPz`feIztLIqn~eAn(Z!bZMhb0zseHu^0fL2He5f5sCGs5>YC4yfj1Sl
zEj>R!Q7EibG!8U`dMCyA28frSN?p|j{mF<A7bU5psKMk>UD7<TQ>{2oxE;cFls*IM
z_@$yUsOF05{e8F0p-|GOj@q2^gAf%^hA<dNIB7jwzHpZS4kosxrOv<|qq3!Yp)T#A
zA^VjcQpB%NSl7y)Vm5TLG#Y=^YTbSmGX+pD?6v-AU!U$@`<!(?uKjFgx?hEz4v#)~
zid6++o=*o_(j;&>r#ZkK^D9fAJ{^tuZq~%d%W;d5)KNE{T`5~n>&-Z1t5^5O4>En^
zqtnHjJvOXa2&pbh7?u1jBxUVGA;)DB`)7XlB})uhzuZ@(w^sq55I%xTr@;Cg3JNL+
z-))^F5JK@Y`)|s^y^Y_$whp@LPi&i>y75JuOYHylypQ(0PgSr7-}3@*5FfF#%Ci9B
zqbjRbtr-IJSh49BmRZAg?~u25o~&v~y)yCkl-~BJ6I_q}5$ITTW<&vX0Mi*J2|g6Q
zUl^bGabXwd>j7+Zb=rB+57ZS{E7#wLGR{b$r>QRp6<#&Z5SPaj+E^nAFN+@V{@xy%
zC7I~^2#`yLS-J-1AK=g`gI!t3&hv{u4@%s=eIA!mqeZlxK}L@XA+s2-vI*?K=tkl$
z8m-(|1xRxD1+s==$wiR#Je;q^)XA@7l#UemzIkoP;ju4RDf@QpP+>Md{oO8C{fBz-
zIzvI>`+xrySAW1@&`+&J1C7=by*tl0V#v=r+iuo=p2J^#q^pzv%FZojQY@zM7YNop
z)z*4`KBL`nw&=}`_G*%Y<$OyBeUp}A?mGSs15_Lw&GQQjlCq^4BF0&L4Av3Xn5m_C
znWlfsm975-tS%?5ud}%Bd};VcP9<R7k<Hb$olZw8oLqG9*F%Q&B<j0LtS&0^9*GNb
zX?nI6&jt6T9C#p4udC={x&iuqq2iq{a8@8UH?tTNKOg=V0*Yx^c$V;V&B8Dii<62G
zQW+4z)$#R!o6J2i-)6R4o}0hNl5p0oH)L@4ny*!%^gz0Fu|jVtv>5xQZ8h>9y3Vqq
z^J85F(IaF%&kWzwwN~+e4d1D1vT{$PXvBw7>A?}WX)ZM_XA$S!v!&NNWxy<FcfalA
z$okSxa9&5<*OjIYBTBCrp)VEI4>rdA(0qoVOizvvKabFeSZ!OUG~LG%zLq<97{Whh
z+acS8hz7uk6d)*+JJ8ZTnsJkt{JAH9F$2U@!YcYiih=-smIEP*t;l?WU~rt_nKps%
zR%Hnj?JII<Dn|=ixV@+~=cB*dF1i>_YO2e(A@brjdkVA&DZLLV88E{)6WSJ-VV2HE
z^*c3f*R#yij3bOkUy+qriGD(IJ4$MFfz(B&`e7O@B;0nS!blfWLSn>wl`hWVqGE|(
zHRj|vy)_yI*3dPh(>GChFA#S7kNn_6w_F2~X!6iWJ+SSd)~K<qt(VY}B;~*Eu-YS+
z+r$_Ba^&RX;8m^GvkhkNc9_%^u#}fVH+NUB4r0ius@pHeV^3r2_*msamQl8OJibH9
ze*tmW-$CujD&CVn*FA|-;@Vo)UxXh8gc+JPNmFusCj1EO5hT5dTD(mSUIpsE9vvxX
z2#)kIM~J?Wm|{=ll393=R2^Y&*W-z+p5a5BVD3nL+pnOw;l2oa|4P31jr>E7GkXDr
zFjEId%PEwu`kMplylw<%G==@_Kb6eq)^A@|8WufDHX~L<97YX)6uKuAOFp$QZJRyR
zVkVU7@Pkvg^#BxV&EU^U-U<h({Zm%w2juYn&kt(_Te_EzieA|BQO~RE+dm?=*cc9C
zVw^MxwPIMed(Lp5993RH8ZDH$(8V-Bn$5REOYHWfykE)OIW1+P*KG6)sQbUMSrfSe
z0V2Uq9_yb6zVIC~-Ey&~a2l{z{7)PkIBV8fUw<^YkekJCZDMWhY+(U!Jw41IA0=J}
zU_SddfBu2><v|aMnaufi=9;`ka8AQ&)34rMS?SW@XN>Ch@`mqDifU%}+Jfl=OOoCd
zzJy{;`1mc`;)NQ`!gokl-%z6T%2dhGkvvK1g+@t-F-Z%9TO6A*YMefAqs|t?+{4lM
z^tHHu<hER(6KyuR>}2?8<i93(lMA12Y_AEmQk01XDllF6H%pO2{nA#$Pb&E7s_{~o
zYJ=2HI0Kb7E}VHs!m3sG$NBY9nJ$=GW4l6aLaB25a`D7jJ3r#^OzsK|UyM928CDqR
zXuEQ>au_}@oYNxF;$lGQ#iIQ0b9A=0YC`b|3BI~NMqDE1-Qd9q6|B|wpul*Fl!ds8
z65@ayLvRu?klDHWul`#8{=f}7H!rP9EHNDG=5{Au2&n)msS;Bc13%u(bED(vbt?+2
z4F;Obsa(W1&I}Xp^cSNobAm9va{nFRXJ7#7L^2DU0igs~Y?#i=TsG^Og&eeuJ<68y
zJ<+MPvyeQTqR3!t)|3=%hCXew3corUm3uJ|8G1Avym-$4fDfsN)afvZJYFF`I+s8|
z$+`tsz`rK0$_#uoO5$74UE=r5aMDBIE`*K~vOE*uedx1i`)WatSc$nSpYiuZy?O)f
zLlIqnFzKY7twD+7B`KHNdBD;;)a;wGxQ6OiC2%HpO5n|ai<nBXsol-P(JNZ1K@&l8
zNPES2HzqCDbS?{!J<a-?R0<n6^lhYpzSKg9d&N`L0pXu%-l(VC8AYW`m0@qwEN5;z
ztkc@cfIaM)n06?fY=6+oFda`7!v2sd%`ZyAWj#E_AYWw)ql+f*+WLT^bTlMwqT7Pa
z;F*2tQ345kpwK)VXR5Qa^Xfr4_C2lBvqM%)7<kFxBmDF@%BH2YX~C_3Q(%|n5sUe{
zb?DARo4t)q>;I}Euh#}0gioWMIVIoDKI)p?z%mOM2WdTtxI8G!9M5?gc^V8)s=z@>
zk+23@;qeLr_P{xDKFlMJoCpy`Q-M8z$1AD5Qrfz(Suj^6;!!K!mf=lsOuH2_X|O12
zH09X%1A9lgHS@F^!trDCJl#f%=&lw+;*4KDCZK5uwkZ@sx@cx8Ii*9QnJKC7X8Cl1
z8pon&n~mFw>c7Hote+CFr>7+jM-jnCia)lE&h5g&bTqWGgPd3EnI)<tPmem`5Blac
zxMvQs!-$;GUKJkV(y|{Ov3=7{F;`>fDmT1Mmk8mo&GsRio^1KiNJBScCoUDDC{%~{
zRe3@cC5+)KYAe@dDKQ{J`FCdvk%jVh>u?KEJ>HdZnt#mF`%K9MgF2~vUnPNm<4@A{
zaiP|z^WS2yAiFuQLixl9?qk*nyp#a3chg`_c=O<3OF)h}Wt49G-hcaTXA2c>pf`d)
zQR7-btIG>Lg6r>(LoLUXYt4jyVw01do|`!_JFocOLi}fOj-47~M2k3>33<P+{QepC
zX@Ok0JecOqnuU5=NPq6EaS1LvND7lxZCQv48J^uuiqp<tYP|sh$@@&;Z!k_rG$i|m
z4{1hIg&UvlJo%K0I0Q%|zW)deo!65!`Om}t3}qh0aEvlIzZ?^_x|?BDl%yIJPp8o8
z_Pcpuf$b7^cD=oS{d4r@aP%z2>X_W!AO~%AbGXN<jAD-z|6qa<JX8@|wv^VK_5HYW
z*>~;VW$OLi+InAaBhk(lj#*JcDG%n<Bv4AG@YN03?(P&qaz6;x)eFso?^E4RA<NCp
z_m8bD={6h9))@++QY2M;i}20iQs9MZ$PSAjFYRix`QBe8?cO0i{a{DB{7jyjTPFU^
zK^31Y4tL(L5WY{|d18ID3J0Z1PQ#gkr<A6}lg&nvWdvcUU2+K=JALbgg;UbG_f782
z_36lH!SgR7&u5L5&sS90IHHqImd7HbUJxSBq?fLpNMsdllH5PMeIS(kj3s)}M*8Jj
zyhxnOdj@%kZFFLko-LmrD^i&VCAcTcPD;R7NYZ4{+Wv7VIXFZC-8YioGTqrkBqgGX
zUz@)oXuFAN>^<1wF%ZgT8)Z|F3Q~-dT6f%wO1yO^+%_B|Yr&w)^!C69rjfyQ=f&Em
z%ti!gsHHT|b>SON(csvTUD&U~jIoT|B4>$|f`|_I{sg)8_0_I6JZRHcZm#pJa{X5y
z&gXUT{^g*H-&tjx^17qwF&|Mh=61ItO|h$cyY%$JbTciSJ4QW}p3Aq4rzralg~KGA
zOkHwy-c*}fQ2l4rV?kfq#@St~no)50@v(2?w<tXg3di~Iec7sQucd+Lkx@Jtrv^qR
zCD&ND(!T8?XN&OJI_zu6V13jX5^{5m$(6_)2W3FCAeo!^UES_#t9H&?MueHcgAh;8
zjJRK8tLA_K-TP+YXO&P=@O@jw?b0B6$<0{QQ1+Ic6`gY0W~G1+7|3>`#(_SyjOh(o
z)kv5u5m6FmRiJof@LLrL(`F@7O?qc}*Kw^?nBStb^F+`q5REd;=DcJL^0p$Bis*p`
zJIH}6#-sMX#Hy#lHDWgVd7c$Rq1w=XQFcxr1pMf8pQ(xC`Q;i!T2eG|a4SRgrD6SG
z`PwgZu+@I=il5*6XIZ;OcZr1X-GVQ>zk+WsYbW1-QnqZF{|wrd4s2*=Vo8LIU#V`d
z*fNvlCW1s~b}Ojfk%sgQsC&id9tV=b7I4jV1_9xJ!g%nluLP44lz#^wZ-dlb{A0vE
ze4*W#zT{|c&p7zaS0B&M6>b3gPSy0iv^z!wU;(QvBZZdbCr6wMUccv3n>5g+YdhOJ
zQ)0)c2HIl-Lvw6Z1T4O!at$6F9Q2S=k#1(p%=GbQ@nQs>lhZ4z>p4dgEEv%ZdW-|?
zpycFaqlrs*clVy8K>iXxf{j2%JVi1b5L!{I@YK6DEmdl|)|NyADj&g>&6-YADW*We
zGpDZz|BoXT$8S<6qE}0+fupx%7^q#^!+Zk5#D{wtavP<0Vpi}Mgi3)<zboAErhpZb
zm&aP@|5DWfe<2`w2nWnd9^2d7yfy9`yq-C9is?y7Vwgb7DYKnEJ6%J#q3&c#zD$~)
zmzU>B%WkPF=_;@Ek{K_8gR<x0f&x?-#K8gtE2GhvIta2ZrSqcV$*=2GRa-FUQyHg;
znfb?mKp@8`3hH7F45~ZZs{Qbgs^jLmtRSMw;3u13#s6ywcB>*U*H+ef8I<RraO&Y6
z><@da<Z`D8^}rj(TEs_EBLAsMbpEAmLu=xecO$~MNtdQg7!-AeScq4@!YIEE>Dg*W
z{<%5>!cD1Y##^Ox7rZK6)rVZlCN$HP+}k@l_uC~@UW*(LS;He+T?5kA6(BX%oqMTl
zNjObIo0_Vp5U?T$2R3Td15w*_#Ttc)(Jbd#KB*jd3|9o1P5y#f27ZLC6Ma5>s8A_~
z($O%X!&cx>SSJ}h`nX+e5PZTORBWKq=Ne5-tYKh@PSOUMIAg}1TJvLt4J`??jTh$T
zo7E}`4j&wa*hW<2%KinX6wJ#=|8<Kzgh_`#XOoqk=`7Grrq7s&@Iz@QS4J$lV#S*T
z7OuD9g9Vp^Hz&q&*MeU60LRv?WTc@6>Efy&vRp&x6^mWQpoY@B)H=wpimNsVM=Gcc
zZu3XkthEQ<PrVz>j--XY7&@QvI2_i7OXiM%6V-2pmN42b-r9`t_THzuUM{#=&H9~k
zlO?xTjM*KzNBW(avf(V4ZC?AQL%99U^71nL&#%~Eg2|2ymzuNS{N-Zs(rE|9i|wk=
z>tl9m*Yw82h{8Z12)Vv}q%zl>nbnjWNR>k!eGU`xMy{Z$%}_Wg{^FzjyQY;NXfLq5
zEvjaW{9K<5d{)}-7Lh*iMw>e&s!ZzANhUf_YdQJz-(H!Q>W7G^96h9AG~NE~CuIf4
zbK^MZxwG$Rm`?r@1Zhl|xL3us@F!sr!fUl&2F=+IGcp*c_65-ls~L8}>LkU1Ua7mM
z;Xo5Q+TVR-Si(wcw%qUu0AL=fptb-T`X`{r^HMp@xpmi?i4a*wCF~lLnwV|8jqILh
z06`r30Dp^QS&v(-tipXKP25S08LkYX=t)id-ZHt!gU}(;n|g@*h$I0{Swfn2<U#G%
zzkhsgCakZ9ZZTEg3tQmFDM@U2h%zQ657kLYNx6O481EaV9PYfDV}l3oD|{$*%(%g&
zEOrJR453FmLD65O*y7%)Ms{R#AV3_$jvKPSchQLx#YTo@6ODw-Kuy6AcnZ-iMabx3
zP^%--)JgIaALV15sV-fpX@j{ZmHh3L=+CZ9NCo#;A^ebssZ&x#z4C!}<o^GNUf042
z=9~!=L9Mz8LT`k)k$t`i#J>D_C%ToE?wwU1ti1n>Pi8r=0N^AHde?GLlI$Z=2U-Mk
z#YzS5&$mQur3VHz^6kUr!e)2QlC4-+SlYitoIG*VppI2G1s54iaH%OmEw_$muOjme
zakE-A|1M`DS#E<r6+$5PzO><M!x7M89d&<vlGy$8x@s*gts1LHHCOZqcXSxe#ney+
zWJ*o@CwrNl67&S2x68-9_Sc~Si<SO$Z~yg4=HyAe^XRy!XA|74te?j`e3KbNfy4je
zx&vMza%jvZ*9};mWF{B5KNWUmR2fShc6}LIHes8$p}gqYw3*m^1|q)waboKdYivC+
zq`~tbaR^2?z-a$_CV*gaIi55XE$CVC^xk^Uzu@~R1#po!*Q?f%&!0bfpmDga2=c0W
zf{G4Sd8oBM!23V_E11I>vx3BI*tMO3A0w$ZyUpT=X^<2gAVHh2NKVp4NfkyCpGDph
zAtLVH$6ry85lJyBjQmA<&HuFv&CB#VMCBzGQR}-ss>4YHJ}+R2G42glB(nFxPhq=}
zs;=brv#5-#xB~6_`;!ySeZOi)wRL2P+mA<R%SkleX~H!%Vm{C^ItKQDT+bR&pE@rm
z8S;nfexwh-%EC*ulp={gA%^M>_|zGOyCKXtPfv{r(QH8ksFaR~W3$a@0BE|LX}x?)
zp^N~>AWt0iSst~U;6>aA3cq8CiK2<VEHg8^-yBj!-mn3Pf*rOS?<*^K@R>!K9BAAq
z{JSg+E+ZoayX*?ze^ii%#A~(e2I(5Jt`twsS<H0=_-5(sx2vcB>bi<kIWMd>Ll1;|
zGKb&b+nH<H8XJXvEQaj2`N|I_P}?D~X7-9z{<9_}_Z(S>K^)jg(|mG8J7i|9Yvm)5
zkoh-Bd)9N9S_vH@6{1xfpHNtg>;u&1BTR5lWY9toGZFI?QOGW3MLoB*nIR8`Kkbdt
z`&axi)Eq7!o~Cn}^H;dJZqK^i!vmrVXMHq!#7d@qozDhluEl8LsP3sr$3s`oPOYi$
zz3Vk7_+8T#r5A|F99L&70i2ekC_y4Uv3pd~(d5ec49ZE=O6`%~O^@n4-}!(bUGQu@
zcUeaZMrPcBdRzjv8NUd$q-PqY4bo4l!`&jm&^dpTN)h~o6k&<@J5jLb;dS}jSuW=r
z_kBs^0oL7ENM3qZCiK`~XZEpoTRp+zi_a(iZ`L;_DR1O;@yfk7C!N>EbYmVtrDBB!
zqAlgyoFGLUSP`N&URaf!Y$d2Q4(D?18(j0CeOs%V?+g53B74Onpmk{!_&@~^OV(^l
zE%T-7i6zy(r%r!HuWYK~ONaZU3L>UgGze{P!iEM#9A8vQcr9EMeya;=^*=bAtn>OW
z{uFPyEUa#}TUSv|>6e77UvFBxI4wB}haXtqAbl_JO`TUO<ZO0Y<_m!Tl`ET;3MpLr
z;bJR;r+%<-Vdrd0V%F8~W^W;=as*s3vN1Q;y7<CU0v*K6@n1z)Y;nP6$+*-0<#`2P
z-T61^S48?M^?W<`$U;qBCV!0F8PbM0W{#(@$-%c~Pb&D#)$+K;0)(yj=&ux(LQ{zO
z^X0M}LoFr#7bV9xd}a*?%qqM44vES+&lI7J;!^->nN>D|nv9kM9wxQ8urP)*FdX?p
zgex0f$7Xp`uNfNBAV5*|;&)=OE6V_WtAf8<5VEtqjm=6?wcn&s1jfCuZw6Q67MUoY
zCoAbQ;KBQYAu<nL|E*6g#%J7zrJ%1a>r;5!FP+pS0xdI@;W+~D`~Rud*Q>FpD5t5a
z4JpS%$rUlGLrfgSspFj9sV~6x(Zu#&qku(Cw4B<sNKDoZvDZ)^l<X&kJZ7b~M+$DY
zn=}4bbI|~-bR}D`vfdqN=!7Wv0UjZd?9Yl2*3|z>8p`$MBM?Tw`jVyT0)1a~<oFN1
zYoQasQqREi7F}^?2v$|DekZkvzIR2_w^6{&Hv%57sG`s4^Ape2PUg3s{R<@Omep8x
zruqpwuJ941eI@CB{H=TOq8izPUc?bjV*!3u3lKaQ?i$V>gQ_QzOu;zWq`Ed#<kjP6
zRm7CjoI@kUb6D6?v{Y9!QQsxYum`L}QPlE(;3L0yG5rdt#;_jtEvDt{I_4S+YrcBv
z;Eq}8P6mXXIQld^=KA7o?dj#BWTm10LxQN@XCqqYbEK_+qmG!NDa#|)h)X#Dxvk$Z
zfQe}#EHYxSXEqIEL!#J5oRe}aRN}s2(y2U?J@_tUW!5a^&k(tVG!@`1mZr?8JDi0p
z7NL)NImUKs9wQ%p5lkkFIWcOrrvdN30VRFDodDrM`s33ite8mvT_W^j>A=zXkDV4%
z<cGfE!Lq(%(E>CoaV#Lf#?VwE6yuhLYsV@qqFQv*QilhTI&3d%8^u4;nO9;7mHaeB
zBt(iijbM{spGoO;@_)?+OS!9syTCz>$fhQNrY8Qr7)|b<nXhfAOI<5pfT;?zb!?G?
zbsn)~w}|9P?cJ&-fix5EQ+LzIg)_`jTfazw^r%>XVW8pYgE(%e9(Q?HLo7Rp8vNdl
zv{o1lB6&<j>_PgHL~iX#sM=3hjDu7i%H$o{9n>`E3tfi{+0U(&EiEk@E%2tjh&{%$
znF()vkK-lPm$RP8d8+W0rq7Y?uctOEtEfMnEh8JX9;h>}-v3nJHFNck>TS=;;>5ya
zQYQ1L9NbUFhagqPSZ3ivB@ca!09q-7gUdxJ&vP_NeOcZc9+U{YhJrXZmpx!Z!xamr
z1Bp@si3bjUm%C!lBO-%>xZz)Z1xeRI_Ff+jE@tq&p`HOeotvCUKvHpwN)gy<pZxJm
zB~=q{Kx69ZFykk>)ftCz7qNeduXij>4b|N-yIW)b(Qs8)A051Vs$v~_wO(Si38bGA
z4hGrXp8fc$<o-si#kP_S8%9qKzuHrs8?^}E>SOTCycwx!rx5;U4ZrTcxjvDwnwAZ@
z+#AV_znk(hz0{V@x;e~KCTEY0bu+ZR9b5n93!o#tIy*Y}%&7Wm;d*gVU-4GSf@Ds*
zP*63ul&rM5iJ36Y!`sUuazn3k0iDFN98oqAdbu{({LR(Zw|2p+;KAMdr`pzG=h%vh
z46eM*2c(S)TX#6TUc>+D3gx0^*7mOA?!My1b-4&LYdXcO{I2XgPKt@Sg+%+GI7H-)
zl&+5XIEUYTQ=o;5?q2hte)qZa%XU1BN!u!7_XKcflW|g`I7_hU>IBlq&LcyIdWP?T
zBamPQ_uHJ<y~{ulCrlV!92IoNd_MXJJF541%hzk$V@5?^23PabqAfxVwwx};efsP1
z^x{nJ5|-;^X!U<Ioo77N@Bhb<ag;bDj!{|JJ7lknjEL;L*CBh$N2tU(L{>Q{LdGFG
zd*#^27dfJwaO`!mH~-J?#{a$>9v<houIu%Fy`N8}@XKDY=(~{SfD0?r+b1p#E^g}x
z2SbRs7P&jZNHvcmUZNW&=PS`=16g4YOOaqw4<lc3n_Gd3A-bCFkNf1^Nn@xT?~riK
zMCgCl<na<AO$vvGTKQ-}Fcx}9)MP+N$o{*JgFt@)|FWwoJ$<|*qb4Wc-LteN>JYhu
z>u2%~_{gocYvd4di#K54YgI_2TI7uuDdE@Y`A+}*c_r8h7aFz2Bfcgxyp;H1G-x@e
zwBVmaiMHhRoo#Ic<on&871iNr9d)75Z<6^V824aX7@Fj>RhjO(uuyHQvb=Iz_}Ow7
zaGBSL^$3y_6$W2C^}LYY!QF+wr#SbgV^<Ha7CRyj>Zi{1;0aJe7PtR4Lh_ZTFOJ`*
z7#gNTv~7D!-soPP`zQN$e6(%nHI)v9!$L^tk?-GjZEDBZ+u=I&Wl7Z!dRsN}(z5I#
zm4@8dBQZ^5Rsu_?UPwhP?4?98eIM)=z+Er>k)g6EP1i9#S9{`w233+S=)-@jaEHe6
zk5_17fvDa9g*(cUC|(>>ZCoZO<gTRlxO|3HC&+QIX(FT1jy3DHY2!o*VsyolBfdh{
zmKkemW$MYeGClmYK=gr{YG(M%V-im6^uI+YcM25Q4H17!@KF4qVWhz_8LX587}Tey
zgYfpxf1H`fjnFi=KaW8vZo<85C*|+vq3yo^h!A?sleF<0;ZJN0=7@#SZ<6%-Tj`l*
zGzBI@yJ1K8YsI+muBLS9q93p&)D8(ofn(TfU+n6nBg&7MJhl`^6&k%Y1=Oe3DWA{b
zgsR#($a_^&=&1VowBr}7aE<_@nd!w2x&_nht;`gAIs5l?#I4q|Znh2JyWZQHz55iZ
zMKPc2_C^x?w-A%MS9$x54nqp2r;tFX8*@p|%|ku+@ixs6^fFyu3X`NJA1!{Liw-W|
z(5jWoeWKhR92gl`II#wv5h9|ZD<>#FN|=Fx;S3rVq}Wf@SDtHD3M54-PEY?iUVhkq
z=?)yunzd+p^3><Jen@cgnLoQC8k&jOUCx*T0&nBfZ#zZ#dlDw6{vPLFO<$dk4~eNn
zUd$Y|AGIZA<$-eq0qGA|@FU_W&8q%iuU*TmZc(*SIdHX;$NhD3?X^11>o|U&cwvpZ
z?wU=JK|xy|gjSkA0WlJ>5)O>TJ3ZQT)SPY4%_06lK@26iisslJ?>sT#LwXYKua)Iy
z<pYpwDb^!6+%j&}+arBudPbIo7S||-%Dedr%8WA`@@AzPR+!z8dPL8`;m;_|_5|^5
z?tJy1xX#C-pIet)bPPr1^7)4ouS<n|>JzvN+tLylnBqB$0DbZ=!$T7j4~RzP(;raC
zchWd7s@HxtUY<&_8x_@-c6m?06MTJl&`fBpKRh_%-`v<miuqKQicmz+>S1$@Fg2ee
zLtTqc{L?fU@mu3zs^pv+rHxn;hm|h56zVW{MOh-kzcYb@QNg#Z5o_5AA;I1(26flz
zPZLQvi4p_=&z<tP|3l4JT)l)eX@N`WmxXHXY^XjdJ7>Ra{^*`~Lgz@`YYrMs*p0jW
zWk32kzSIlKjbXe<I?0p7(`#f5==G@!Xe)l+r_{GJG0Qk);+FhC6MFA2JS1l;neJLN
zap1|ypQu6k3sZIeTjx<Z)u`KlfsHEg#(FF~#FWVdWJSD4+}fyW|Ic%G%>f9j+c0@6
z-mUg?T_lmCG)PJ8r-!79ylgHFr9X^wCB^+dfy^?VKDeJWj5y-XQ1h-Ap>4IHD!y4K
zq2=}B;AF$SHg&RjFj1>qI#-{~vGpgB!3imn`!l=qw@S1^{zzm}d963=>(;5~s7)Ob
z;?_<maJ5L+aVhlIE9BW)@ehw!YKm`oK@m_|$AU;myH45#i!2WoXge}u%y^}n)^GWK
zq?YIHP)$XKahX9Hgp;9f-dpCGyc^=dU|KfT?LpC(v}SB@)%3q)*{wR#;qSGX9GhR9
z3U1CVPxzabyK&#iL-HGU>e{NkjxW<IT9+S62O3AT;wZm-G#DJnes&$tQPMfdXl%;d
z-@=t+=~GuqUtqbDGy`Sl_hIK{`dZP}$CYK~ko9B!uy@6?rOj{cqoxM@*`hRoj8v%T
zW^9bny6~dqT|R#(xv6PxN6`m`PJ{#J*YW%6{jq&Cfp_IZu!nE{o^6LJ<gViQ9{$e7
zj05UZf`j4;%IyE4Zzp?nbU=R$_yrZ1s6klh_>9X#8l`hT*k7{Q^Uk$Zr@xz@(*Xa<
z>!$hx4t`Mw<Mjl>d+)sZ-&^5%2Q9X?8K~N=Es+iE*xQPUHT{wgxYrN-GxLLH&;uS}
zj_yO26OYrh$|pQzKq<>Tm1uo|*<Ul@{!t$`yl5D_&xF)ED+yN!-^;t%`Efdo7y1D)
zA)FJvmE`*;{NJ3n;@O~ihnbMv&=4_d1qAtBp1>bgO25x$;8`#$97`m8uUJ&M)V@02
z%D>#aI?lb?YP>4ET1;u*<I57YTMLp0PThf@-Worex_0orTlLL|*e}5+yASM-fo^6>
zK0Zhj@uD~X=(KHOqH8Jl1~{ON8}L}^?mY>@fk^((R-SNu#@F&w=2dnqNDR0_y;G!@
z6(VY*sC1|PW3F}zF%S!dm7^yU84#%Z2y#v;RaeXB-WB)Q*;}vnLZ?DSfqz|8*pqLQ
zP}GiWwmt@}b?Y6x-nxX38w323i^w%@Zfm;;^Bv`u6mlt_rKNa;$F9`MBI;^;b|4B-
z&s_~z#;n4xk<u9?j$wO+bP7qaBYXs{nXe{}by`>`2Tl&-VXHT|*1Zq_g%k$mJ-MgS
z>{~$_yINbi`Zw~b#}0T&)jA0Rtdo4dJ@wcSp!*`+N4QnE`_3N@2T1xfW9WN+#XZEb
z_$50nlA^RB?CQ910_uKxINe;AW#n_XIzL#bbj`kgjZs_;KWXY(UuR~$;WeW9XJelk
zQhfJQ=6KBKba~g}o?GIt{1fSU`8BchE^?vqvRDcAcz-PxWI;9+se3Gx0is5IfO|=$
z&+qoK_VK??ZhJjX0C59G2NPI+RJyFjQxV7iT%Ib1BwVOGAVNlXWyw#m<0|ZGr|VPw
zw{F<wAGlb9lhv?6iMw{0=uklrZMS#`ZA#z7JlKb1M~+gO>CoWizuvjKf-5Aswcg51
zYG<{6x#fF-=?L$?d~>z-d#dWMg{A$(3<xdQgA{5#$TcVd`p(XRjSf#Q3i#!7@l)-G
z((%v`hPbN4b3Wi=>O?W{?ps++ytykizO<~{fGsIcQSKOQ-4>gCG+F=`0I!qK>%s^L
zbK;5Uy_k2}7G>iWar|jkI`@%ASg&|h@`_45eP|a?VytkEMR|EMjP?y~noBGr0Fn(I
zj+l5JS_61vWXZ^8mdS%2JId!-TvHpbf#~@xhqwQNGf)3ubpNGm4Y$vcWSfo(IS(={
zmo%WD*U*(_Mb2!ILuE;L$-9mh7a370-JdddZDe^g9MIP~Arp0D4*LAAZP&Ckv$bHP
z4v+~Ru_4U(YfUaOuWpN~-^c<e^mBUNvt<d)T_DQTbfwu5D@W(V!FVf;`4v^XIAtu7
zJN^e@;zZ?C>~g4#wLxKLU~X<iEZX1M`PsMQW$i{pR#tvfykc^Xi!aeLZ*-{a@x*I0
zCTOKAz1HT#Xo7FPub`Lfa?g97w<*af=z8#B+G3cZiE{Y2=iN|U<DLEkra(-WS^4*U
z00>jeclT!q<fKiewc5h@`85R2#~NR6ttzT0Yv#X>@UNQcf!$+QdNPr1{Lxj;In}KA
z@7iPA{B4z+)B$$f@J?3W^vU1{NO{lmo_o=NqpMH$?GZB~_{)BPN#{Q~yXXyRB))xW
z>^gnE>Ib|C)_i)t<$HzQ$dA0}Dx?A(H2<DAPBedOD;=B;ll{6|@q6vo@bZK}&FTK?
zKR;jMMJOmC>^J~dviJR@K3EXD?Z0Oa@F0<sGyJ-5wPhf(&hZRCx9UL4%=_>>qVhBw
z4W;@&uzMLL$<zOal#ZXwcQmIY{+&3@U>Qx1gH+z1^W*pYI_(D=9cIf9d~Hol0Z5lr
zjz1@qvOM>w{Zf<pz7$SrR>i{7_SQ>}#vHsCnsJ5nGYx&LI1?h@SC>iCDAToM+RoPY
z9_nUBy+5~-0$a#g+gO-K`aXuk9Zm6V#v|wUQkb0$uv2f=(s34MKBSm~LftepZ){M>
z(YWflcZCoi^KLLV$2D&?93}h%I^$BhzoAB5jWke_ulc!gR<YtS6uzzX>5>>0^<NPV
zCSj`S{^}H{t0%g-_4r6#IY-2~y1C&fvIm)ow}^kY_qAfvg|;mNUH6+-mWnPNsxK~U
zZocFlwSZoXf4Z4#YBm*5azEznBZ$Nx6q8=fPa{wSai&v+9$*DZN!|hJ<5uqIo88|g
zy)|{p(myb}`hxk^ZS~<jE+m^L>eq}w(O$TW?2L5HA3WY-wH}Rf`f3jxoIV6j^n@;o
zh5OzQluAji1zDv397nO|r)S71Dl01~Dak31@IEl`gvFa^6zb<b5fU5^9HZ9?2@jut
zc!_g=w|cc`2pkXB!vTono6ZPg8yNiRvLZ2fLSO9!>2m9shki*Qd<z7TU~3jUWZS}5
z#{A5<?(T>)sb#MOB0f)gkGMVPdV92Z?6FU1{CjfxFj@(UAj9d#)iTn@cioB=ceEi_
zqk*tlzcE_H`t(6Q!y=tgwN^SrMY4F2{^mNXfii>c1cpMuRgK!iGQ8Z&lFhkJMps9@
z^x*MY+Ovv5eI1tg>IA33<N2}qr7bkyk&4`GWOe{CSWY2_-ks0g@P(=AWb0U>CL;%p
zA`MGnPwlfsr6QESql*%2UIW{RIO9oF-jZBVT^Z7!V}X%A5kfJ{PgmcYEg|~-Ack5S
zb_<ctJVOe&aTW_hWV!1qc$_QB$^s?tKRvtx;$=HsO(RppyYb<=Un}5mpmbJva^Dw6
zXDX*%bmAaN4_iKE7hW{xaZBI8kgCP--ruloIT~m{^vc9pu<c?vek-KNM2%6<^lH88
zg|n+GwX++t52AFr?~Ju5pJcxY4p=A05g0;L9)PYC-e{)nA>mvkm0el+)TPBrrS=bu
zCu)SlIsT`x#PMSKJ>sZ8$LFP_LOB0t-hfhQ_w}{3L@Gi^pEmx!jq-Ho<?9>aCnKTB
zb6m&`HHd5UWEZDD-(T*_VHWN4|MOVAq)ai#36UGDm;fpyx)PohtdCnCbRTS7t<wQl
zerAlh&xz*1k~wg>4fYBfTI&?=AX!{>((m5ccC8)jnkX5Onzi_Qe4%oD0g7QtJe$9N
z=Pz1B^rn*dL0&?m<a34h&mf&F3ZpeO{1vrSd?oM=xZcoEHdnVFhlO(f11+W9*g&7*
z@O!4Z0`aQZ`bn`*tE-!u8qZ*nzX5xTI;^zy6CWwwZ2y;v@_%0=!B*K4eK%rjd&{mZ
zqSw;er51UB91BB;yQME#g7c3r6$Du;OUpb1J`3`=`EM2LBO@cBz^M$T^VMqIjFq5T
zDU=lRLxH(>XU$KsYBGaOjLPxP^P=7;N!e)?wmcDE?`RoxPC*4ki1v^UK$^@xcZS1n
z+;Bu?DzA$^?+-V_(mm=zOH@0gGl;--#U07OG?BWHk$#ch{f}kRHx+vxb%-KaC%L0O
z6rA|ONu3kl24lFt^}~x$?lnJ+&0(aR2@Lx5A=tI%0t9QhVgXDWlff8H$8;;@T@See
z6}fXPHRrI^4Oov@=U?D1sVq^BSyvZPZ8qp->0vwGEMxX#0K4QK`@NT<5fe}9=4fMn
zJ0Qd^9n#_@!{EsN<`{Eu<F|EEy<~#Xx9MCu75}o~8c$zGZtsvXtK|UZN!g+D>DI4;
zBJm!)H+2O#B(sNC!e9B5#GnF=U|cq&-$!{|bOl!0uKBj^K^y@DQX;#csHn(2uL))7
z_P;W!+r)WS92Gb=uVWN))%??TXfqmFZld{a%IfyCi+nC-q=D*>L8RI*vjfmF2nsA6
z(_|QkA~C(>kU}m7WZf(1c2Gq8J2%!Mou2UDM3FC^3>4~+byUxfXM+X&3-MNPFq5?`
zyb;Uj&Td%_z)e3iQTVAMaG2~e2fV^-vRIxXPeAF;O>Kdi$P{9)qhdb$S(F9xXr<1z
z;=_}mW=>xBvmrg}p>jQ=me0BH_|(uJebn5s9D4M5NZuO3G;xIbHkz_LwKf=?s2j?h
zx%_JLC^Fq2POmNn;a(vfBY|8?q)_XBq|CO@s=yEE@S5f_Gu$WVlww7a#01`b{dre8
zv0(4c#ZH*MMNDch7H1`A35O>nrKk3Prc()HKCkHbPvH)`sMrz67bT5(8g;N<ltPRm
zCy943@L=adP<W2Nt1n2;&%1kY3f7tP>T#@vu-lmI<_eoRE{0%*T;ZX?6Jba<epixE
z5CAmD7?^9;&IZKP85x&-&B!zj-&w$JPh1pD-d=fOIl!aC;)vm=<Inu;D)j$lg152F
zm9L=Svr?BV-Xcw^p|Me1%`o(cSybuSx}-iEVOkg=OJ>8$#)YWUd^{C=F*jDIJAW`T
z(slXF)TMU%=X#l#!ae*|P~h1D7Qjxn<VVkV{6IO_E;stlmhCP4y%+(mw%=dmU+%wf
z9-G0{PygfD3L>gBs|YP8-B3ac4$jeCt<w1_9-TXcMO>`c8&s8a)mx!so~G<KeUTpB
zGbjlR`UavG#(AC>4`|<eQTJ<u*1h)PqC~9qKV$*ULow`d8Xh<Q>)AGFn@i^|iFd$R
zA>=TOUl`R|Q6Y&nsGC$9Vm843>P`lHbX(~K!)tmHepghWv2HvF1yO_Q^G(#0wO<}o
zsYLJpTU<PTwK00<?%jtocl4=4@+2}{r<-J#=D@*g&xE(2P*Za?z5n6Tlzj*2LTXD<
zX%1SG<P=5}tOdm)l6$@aKgXRJHO5}u$Wr2Jt}GJHeSl}9TCCOC$G40HeLvbCXfxdI
zxZ1YAaKHL@Px)wY$cl~rwaFLsWyI_B3GT#J<J+35P=Blg4MG9~1NW~n0uN?57mLy$
zmbz-5Zz07Al7sYNaFH-Wh_YBCH7D50)^C*zwk1s_MI9x)nvVixZzG^FT%*Xi@X~92
z!s3h9bze#>pbyRB8soo#aPA;l)B~(s?Ar&Focb_zt9&s5P7*Ck^c{j@>7U&hw@`nH
zQAsrt$2T!To?T||Yu|HxK}#i%Gje*~q<%y@hpgP)On>yD9A&(nO2TNO+lRC>nw}mg
zg&L)O!OQvPN*Dm)CW7|5?k^!sCS^>D#>M;j^pl(O@B0+avKquTp|PZ$3flC+4gy8K
z?I+8|q4PaC4?)ZlGm{wzbo*HRo7oJE996>3l#0C5;`1Cd9<x(|xcfW?E*<^uZx!w~
zJcz(Wmfj#|xE|=$DGGk=WIVFhoY^bou1^DzUS@Xh1gti?B7ECIgKfPE>D6Pe7p?DY
z@eJ}yAOf`@>c5_WyOhsm!25hBL@vJVku_ZJ_fXg5R&rU%Y<+#ny-UR$7v?HF?A|>@
zN=PosH86)qr)4(CPE;gR{>BSoa@y}A8)Z*pzGWJV%gD2EY*63QLF6ERSlcpYR9uIv
z-8L<p222Ov=tRn5BW9T}!`6M84mbT$J!~UV_`A|QpPuZm>NJa@4rQ;qSz;h5glwo2
zI(KxX^P2Mw&Ugtea(d`8ecGUG%6CXxdiooyRAATs3giaQhjNRyZ5R(C#&pdQ{wx2w
zo&H8Xl)ba<T?u~{+A`Y+oKE?(d-xD!IT+i3({-Vl$3&>}X6ttH1kdQQtEFUDt?;#t
zasB{Yrt4c43Ist`8e>V!y#}wS$5=*Is&<4D@q>ecVb&S0jO6MGWfh}MwIwAb<>lp?
zMlpd&NlCTSo1m2)R){=!a^KZ5-5eD9t=YG`LZC5}w+CUI*;G7ey7A?}@*I`30U#xW
zID3*|e`z(ouAvqo&&q2Q7SdYf6$h;;%Uc4mEY@4JTIV>lp%<ayVZ+e$pD@9jWN!*&
z@hZRWq!@hp>0UU7Szh1SF{+wEuL~L!Xd8x9*|sa*MOY>IIM*E|N`x4=up+fw<VR{Y
zQJ;Tp2gz?;g3BMNM>#q0l)GD^fN+&KB2lsG3+G*ggg-k-Z!ihYV-B_zX;6rFc)h7W
zY^`kzT-$m728?AZH-vLE9aWWFPFknE-A<xp*|ED$uHhdXer^P1GDU!l`nHj+8(hg)
zTL}&5t60IJFJ>5!3M(UKkN-XE${3iq>yPto<bS2ms1kj6y0=%bxt|4SX`f+E#J$2`
z=d*v$N5YLXjd;jxn80q$Nyf$l-F=7_VLnVg!6_<ee{rnPF#p9)bN!&jV`~x9rCQ4Z
zm*R{ihw8~#E(9qF>F4Z#4N~WL2rbLR%$HbZQ)-BK2+m>{;yMg-R(gt&WyX-YRt!JT
z73dz0Na?yyZ|IcDTMCSc>MoS^S4am76iHeC&f4vonD%f%)D%MqF;C6R?_>HckxEK^
zA{KCoFvQyjOeCD-^%eA85JwI18)PJRCGrlL?+^&#Z(|Em>-39Em3?2UYy5AL#n;z2
z+UAFzg<JX}MNySVcbGFx{gi3u6R)jzj}6}x1^}Px(M9qKfI@lMjO8721iik1nrgX?
zVa<%JZs|812YF0e@L_GhU4f<j!#+@);=?=g<AM1UYdSS7FoQa5cldw}y@YQBj(+P6
zQ%N)h9`)!kN4%j^KA2<)#IC$lBT*lI66o*m6A~iJa#v95sngc}Dh$O3ss41d3-724
zoSzm0_BLJv0~|q*Gqb;yjgy$L*iV)H-inCP9pHIVrn8*Rgkpua)=sREJA=kpP|$8R
zD9?wa_-6$#N|#_laYOuIWF);X?(|5hzO|<mLY|@E3V5~0i*jq4qw=xp3JeBTC60_R
zDqlet;mvJ+2*0!BOgb^aQ9Dluece#sP!~@tRs$F@x|7B_Tc}`O$Vc}c8_17q*gI7&
zo{aTa3N;rR-uM#^5>M3FUH;EEKWAur$jGl&_`=KUs*QvgPzqFp0iMm^*7W!QMCy3^
znD_0y8Os%F*-j>jt!jbpo;P&(v&;NvUl(Lqa&wJeyOt)kdWDkTuuPn>`H`}%GGk#~
zOanBSPrV~0lTA3D{`Yl~7{ZjP$dJzAdxt|D%i?bSaH@~sZ<#qXq*tgTQzd9J<p|R<
za^aDRxTvkIH7L18(|LL#RhZ7YQ6T$vLBnh*NcPus(^El;m^9g3eM&WG7;(Kh1Z$*4
zdw6J#cC+nSp26&ljipX^fSM*3$rFXEc|hg3<8tnbXDLXM^}xD%a+oth7Nz6dV)TP0
z^03_g#nnZqd4p){DTcZ4{Rh;jz8oKdd4dIeF2NtImKYwIi`m%NSs@Bmd+1bJzXg_l
z9pryo(j!M{_>Il*b*_QCe>JmaYS~cNqIcn}va(_Vb^ApZ704otj0(VQ*A<9=0p^CH
zo+IhihgT@L(z$|;E+65a3v!EXdWRROv<166M?U!BV#T<zv0<5MwAnVX7cMunHy~Au
z51tv3DnL|+I2U#N7qK#Xl$DuD|8jmh?8S?T$Q&)j$K6)zJp_VX6|)~bl~X{(Ix1&$
zsW*w4mX6NQfL6?=6{R=35z;0d4%}Y}cSKp+OlK=9ZBoz|S`4W3{4t)m+2RlTbNcTh
z$vpqV@NoAsmY_|i?(i@VIIusz>^Q}B?1g5}o0*c{$IZBv6G#D>w|0K>=o@5|5+9x_
zfp*azQa!FB;Tu23F_f6?Iit@TAAY7m4MB1UX$)z1wv&2Uvy@h#dg~Dlup^tO6jt=j
z?D|V>SpIJ)CFv1KOtmzV<FKX&y60El+yB`4W1HVW6P@zb<oo9298)1s%6o|)Z@HpE
zn}stz(Zuz^EH_$fAuvkS{%3=&!7G)9SQd@1#=06trRmz8n&}Ov<IWpKspdm)PL5=j
z+>{gyT^?`StMRWwqq7#%z>F&6Xo<7sNL!rx%=|GU=|@CTO+$TX-Y}xNIiPk&DrcnP
z$p2%*M_a8_)3UF+8=uT7$DtZGw=_jEs5aWJrpRZ2y*B|e?6k^F7h^(I9ZZ<iwEH)>
zNJ1++>#;K9%R*JH0V{;xU<7s+X$V{y;s=+Ts3TGSOEF<_t<uJTCZ|e<T+CAZ%iyi&
z_BsRN-9Kkr&XOCJ7lZasbL_wG{CSYzN2uT!CUaxFA8-D?^w3Q3ReBKijnO+EBaw1(
zMjZv}ZduQV6IyW+xV1Wrr2*Jy6d&CyE3-nISu!G}(Cag>N8)Zc1*Ig8{9L7PQdTz9
z`l<P-J6xu{>gia(l22kcR>Jk0aK`9^_@OrK6m^V{X$GUE2t`k>L5>3lc~<Z%h$EN0
zHj;jL#>fhLeId@q5N<m4dC}1Y-5dXaKE^qn940|fTtXGdi14|{T+UvdrKITDc<2iA
zriZYi!!3DqG(LPvkN51cf^^4F$G+41coTCCTUe8+axw<A+-uvbbrU+slS%ttA*idt
zSWLG~v~R(oMkP0lT4sUs)ivh!Ig<XvSGies(I=O23dXv36GB?RImD!DLNmVrZ#qNK
zu#2R?P|cW!hEZKd$7oyv!|rD7sB-u&C}Ew@%M!i$<`Gto(?#+Fx!PRoPh2hvC1sT>
zoMZb3zYf&17LM`fXXWHr_wi}xMq9&Y*A9ws#)H$sCD|kH;k7JrQ+i3WYdlwTade>1
z^R9B#W7vzO8}`vxxsAxj0ftHbW|>Q`kjR*K;2>fn<nQ@;PMq-BN(Z>wvZiOR)5GB*
zBjKuO5HWk{A$f7Ut+cg+**!hQ@CKYBv+T|<0sH8x-^4y6r@H5LAIF<0-7GUn3n>5d
zld;EJ^pTroG3Qo8PV&z;`G$Tg*m@9&MEhTOp7<wx&Pq6>8<oWF>cE7oPo=2HEktS{
zc7koxPKlKU{E~x8z21q53I0U>TlnfV5WFD6pvLHjgFo!LMw!ypr*>R#f4Yvx<7WYd
zeEj#F^SMPS9VZcF^*T0Jp{ng%H`8@1<m~*}ikmL}EF<Ig*5*jgL;J9urP_HvN7qu>
zjJ9a89Q`MDP6!C+#F`ZH8XQ?s1bLJxBWmA%4nDVv4uVrdekUcF=Cz%HcaM-!OK!O9
zV7G-7`xCpcgNzUA`A-U{sIXx)-+@$2JO>e6B^_`hd>CYXqZszb+GeDL??c`(7x&{^
zx)QxJtj_7uWHAtWR$EVBdb0misrP814(ea2QlQFxd-B9@RCHKQLkKBNjD(CJwErYP
z>Q`S^tdCQ?#4Q{1`=or|vYI!g!j-ENxS))6Trd{*49R|Ob*Su{P6ha#ZUAQ?P!xE{
z{+(N6wc=)Mzihgc{9c)^wQW~Z?cq+`o1vkN-qzyudpET*=Pt2szn+gGr}kP!>q1{P
z7_gv4M)v}YV<=yW<LYa4N8K$!squf=duAm_`Wh=ikvkkWJ*0%{xGBZCKIz}*=Zt?k
z3USTX`C%8aR%G}*diRowSO+wths^V;2fxLDR5PQM#kc&rc3?lYkM8d9%ATx2iDi3h
z-JkV+*zFj8wYy-&yO{3{RZOe@;V!0ktO8NMKli!OV3NZ5cz<q<@Fvpt;^f{@L>qr%
zp_K0X?=5xz`7|=974qw{3J`3VHQB-D(CQBuNgqrC``Z5z8+X!23De~wf0FpU0bzGU
zv$4T?tsU@rSeN8eLceh)=0LtJqVhhEKm67Dz}zu4_uRKfOpxwDLB8M+nMysMFS*1m
z3Phx-XGvCzNU6{mEjp9&7TSwb!)_>xNKL^7YbIlep?W=ZxAyP(pYp4%j$`b^Tgr_M
zt1N-*6!(CJPpk8BhVB#ew-y$9^>Cfr?+Jt+0f%w~gW|vVi*4e0Ua(h$OdxjWouONy
zUB_P2D1etUXQZH@A;CvtPjJfywo}HqxGu|<$veo=hQ?4WcfKLL7Y1W<i~q?r*3{I1
zkJp6-C)+HOA4`hx28BKE_V$Kufsh&L98dPX79cW(`7}w>l9VL0y(}Z$c#;xSM9iqg
zAj>|Egm9$w_Vy0A1Q79UV7PX#V-IiMe*O^u;;5t;dp2vp_?k@dATw61CcI?wmS^GL
zEf79p4{~SFf9?JDYW~{yi?l|bO_WuN+BNxKG_mcSuv?pEOmB!OH#T+v->)<jtofUO
z7mnI)k0gi0Ki~U%;^G5WNG>pApECXMMW=kC+0S!DTjg|T9>0ynVz=?pX*@$h<?jc6
z$zRQDH%2BU%|7S*@`K;j^md^XbaSCMDctB!`-@0_e~H8btI{18Np@N}XSCUkms7f2
z@vRd`HxP8A9o+IE@dk(sivGDF^3E*~fS)boDl~G_dlpf5uUztJ8d<=9Zykmy>QB_M
zye3(_$nR)*F+V>0CS{yIv)*R&D9AG0Fx?n@UiC%^#`6yrlnEw3hLY=~;X8o83Cl@S
zRZ*t{KKEU@kv-x+RLgW-!>e1{_gskITTWSSI$tvo`0PLu$e>*+lg6=8*BjFY`0FqX
zC06>_d{>C7K$#o7A??eC=t8>ia*Ho_o{GnI$-5W*#O9~4D{1vx#=TPs<?zSJR|%8w
z!2$*0o^_<eR#SYFci2zMFy757vXHqk@1XJICeanadyKsj6mR83b?;EW1RFAzgswQV
zL4`LEQj*1aR9<mCPlEAax-kZjceVNBv0^%DZ<RVjzlbc$-`rMeKH~QYnfT-xi}`c>
z^YaYXGYc10xkOKkxZI!5ph_jXfAScEx6_wYp%%IxGx~)b&~$n&(pEBa)1vC=kGJ33
zuxg`2)g8z)p;%aB@~6>$r}bCncBm(qmXssq)48^2Kth)czIC$2=KK$U$rVXxRdwv_
z9$oFeQOL;(&ei;s_ovAe?k-$_5GrEhG_QdW9>*g?3YFMzKlV$Bf>IB%s*@vjzP{wY
zZ?uR`ZT|Mmk_EkuZ@v$Ehy$67W*F9I@N#MORg}v07^rJH13VWR>;IbmW}J930i-56
zLkeY+Nk6M5*i~f}mjWKrNuGs*SXOq+A09W23Rl)ke|~rb)k-Cx7`T$U1#{ge!b5Od
z!OYCRe_d0feOYT@370MOXN1Ewpil2Y2V1*m9Enou@kvkpUXrZ1BH3>%Hqf?mCs1?W
z<fP%&ps<K%U+Q#v+HmLd)w#+gU4z}JO>(u#`o7C2mpc@sJ0w61J#Wmn9lU(3B3o{<
z+-sHNHgK)IL`ATiU0hc6MsDUOee$l0v!nV3wA-1U{b^$GpzO1$Ln_}FSEmQ=)?15I
zz84u#r{EWw&CL@;TQJtL`F<0Z`*4+B{r$jCm@`@$gLSj4oi@=G6&4<Ldw|%%;XsZP
zNG<WKx0+bv0U)uSnVv;SPVEgfC>|Z#B~+k*D>UKY0Bkil%LRd>elUXdoRL*l{u~1~
z79uXTT_kCG!X)#GJ+0X$wn8G?@I7Zokr$WwS94TT`7c7kB8)w6zua-q>~{}usH<CV
zrPJWp+hnu9&~+UiRXP0!GFs3%`FJ2AzzH!s{RfD?JXxtIMV3)~PX9I@L9c}@ZV4oc
za@9?HgVShB@b(WhcGL;W@{x>aU4zOi%C85>({_PtUM@<Y)$P-yWv8*fr$OJ|Kv=-e
zLhq|rugbD-#cQ%#!`YA+Z<!ll8V3SzlEA10U3T55Th=gVV0|!gOfcVF=g-$Nb%89v
z)e`DQGR%Klz4dtf#zU26Y1qv@D@Q-E)_?3;&%>)DBjjbxwK7~w>+Ni8>I@Y8kZ!D+
zUGMqm{vI!gg*&6lc;8FB3{(}XOfrT@<UDDbt2fCQk53F~kshomy6iq9UPL*5_J!)q
z=*4R92|t*Tyc^2wB$>~3Wof#PkCD~_6gGnyF~J$5ZgKZp;^@UOas*5SZ?}}som_s5
zjjf)OJ~Z$(b#G@AiN4&DL;aB!ygc{`^$ym79Lq^$*UEe5jg)yDshR+FuYWMZl0-t0
z-ftBlSASqKB*!$&!a@-hcAoHEk?o$5{jIOx9>fL^J6~QSQKvur;jw|*F}tM>T!=xA
zSH4P%<78Jl30k$bKdpY0N>X+X6`%}?ALoRdF@=jXu$G(W4Gj%ASb52NHb%l)$1BT{
z23qCWMr$lBO*KD~KT+d<m!LhT=xA5Y`ZB~y!5uAWh3k)#X-K0K=SDbcNCY6rIG{$K
zL^!jfc<CRd{|TF)CKUGcwWK@;@T~_YV(rh-nh*&LR6U+Pg%dJ2IH{L%cRTOqYtO<!
zzc9hH*N+RdT2%6q!<6{&MLgB`Muqai+Umx~x_kvD)*ElDa`&+^aauATT;J1^cx=I4
z1#WcnP!n)MVja6|SBD)L%)M*V9FRx)I(aL#(^5L!T@K<o`T|9+rgAGlMZKn-1cr>0
z-x;DRAcq?5j4V3X-kG17ndu>_JZAI%Vbt*?hrYSJzEhFU5mDaBI?$7l^uxqBQ=04<
zb74ZYI24iEEV|dK&nBS`p(iK7U(<?z!|C88%<m&WuaoAbS>b3xK6omQ2=<Y^%^ErU
zJ<tKk6aNon%R~?rxnuz^dEFP%wY{s0m!m@t6p+cH>l!?PI@hpdU0{o$QCCyPS^neI
z1aLZ<_ov0yP1|ZjyqF2f@u@WhU9~Y2#~jN%xw@MntsnFDoAACNa83wOctHrP0*)`y
zb<^3SE5WuMml(E=&P3Thf!-LbKO!<R5?sCngF-~LTkZ4)&>&9f9>kc?{n05;;68{a
zG%z&WQ~W(ML+PqLQ9Dwf!INB90VVTEKS~Bwd!QT!>-FO3G7dx_&0dsXvNI-o%^qND
zS=9718zxt+eX*&^psU-*$0sO+K|k)Ir|_Z56(&yMo~B(*3}+%WX9kXN@@dwa(x#$P
zgw&O`wgb2v7!}53NOqi91EacI>0H0o$~YVm{P4Q74F9Xq{OFa_19;E;zs*JQ_aBX$
zo2%QZn;984EAmEb4H|1(Yc%9WlWdkW8>c#=FHYOETVnooIV|KBaXn7sXX2rM@<s_G
z`<%CMN7nmCuo1el(t!89I1T87^xGL+Lf*)!5eBfJ=&*3ld5DG7z1{>fT~i_Mg!z6V
zydfsqwsSI-oVT5OcS|hD+EijIXvT<<#TmQuFI^p4Jk!+EQ<%gohQC~#2pN}G3`<nK
zriQbo+>9dU+WUB8x$TxZNl47se_cyMS?U-oq)gZOp6_MAuT_=}0ms2?tDJWfB(gV8
zK=J2l;X97SI*HUzMvRRgb!?xk%a?HR{Cas5^GyO9=Z@g^CUqos!8WoFJ3Av|S-Kdx
zd0k0kGAd$jt+3zkm-nt$>wktwAHyinK_r&Z)mmJt2>&>x9ZbKsjN+!SFBvtmk~+H?
zhSg4?z!~);%a9XBurezz+sJRQYh-Uo0qlS^(=DQsR<`jvJ$j=#lht`af~1>P(Y+$`
zHf8!KubHWxRpR;dzaa5E<ts^iQ&1NSoUG^G<elMkUh-CMNccH^e*_y)GNAbUzQB!N
za>jN8Whh$oAhnvSB}*adp(Eq%;%w#`V>4^Ae!pav-h_7?kip#t8)Mws)@}7!K*zsj
zI{axDI!EQNm9_n}O(7{xiy9-&b$6;#@R3HPUHf^oe;;QOOL<xIZz3x}e(P1nq&pSW
z*&C`iqxdk<sxoN$kHW+5kA-TP=EepRq{7GISqXf1*(_ZXjI{(X8tb~)pZ1q?qkk`_
z?L|aIlwLG9)(3(g(2Mm+c#cpG04Z#H^9Jd$*4=vpZBR1lc`bHkRg3#Am;Bw7lG=(f
zFF_G>t=3S{(yO?Oi>heCkgFx>_v!t3`tnC^y1eg}X=9ev<561E!jI-$Z66X~nkE|R
zW*HOm3}n|1Nz(lizt&SeTW`z2Jcd5Isa>+P?$z=^YF$TS=C7^2eF;9C2<&n(dve8#
z<KaRpftoGxlItW`qnnuD*-oln{}$(o7dym}@JH6d;(me_Lc5SKs5}*JM%Ezt@Mv$%
zvgEIKaKeOAtPVOo7`$q=ehZUle`fH~fogeYB2G%DT1ec<D0`cDm+Q!&G;yl3<8+o7
z?d};qzgv_v*X7BU5Mz8D!0VZY`zqWRxhTA*9@a92BsfUrh>D8RYWIlGZMjt-LDZ4D
zOoI$%&(1^-p`t?6zBye>g0F7ui6JFlHm8P<X$FJ?^1}({z@{AS*EqIq8@>FobjZ_A
zn)R)WF<h>S4|eOujMYl3KZL@QL1v=M0kl}xH|t5HC~CbXBs>FhN1Ljbg1)pOt)FO1
z^LohViwajZ$lu7vw}zh#GR?nx;%>x{c@iZFPL=SHS6f+jZOPRw;~o2BZ}JmtlmnM^
zxw0#Qe?3MW8L{fGfV(NgNulYCqPOV!E&q7}M*71;-_tLANt!hl?PcXZ(P@<bt$rp&
zlBk`sozj!-^REG}pQzMsf)bTiW`>5p*9BxOUB(y(^ov8MTK>`Z`84iBH!rWfKT}#j
z$+E5d7Q&ql@8d?y*P9K3BCJl<ytx=#>S}j#fMdUv!Gp&3vyg8XYiCrxffgL@L+F+i
z%RwK`kL~o{FjqSM`Z;nVS2KKPptkLXg2II6n733(Z4@AGp`tQ#+XkcP)Ia3<L59FE
zST~j}mV`Zhb@=Is2-Y}X5ZjK5hn!x(Bin5nZr90g5eNi>l4quqPRiDddUuMlLMo&F
z%@gq;S|;+S1ON9gaB(n|GlH9+CykTOV0|m=j}5XEjjOj=3%9fR)f7SOB6uBHy88{3
zn&{lb0w_hm4JF`x-?}v2Q%^(sNQ`pz)k2I+a(zO)Eu)Q8@Xt!G3H_Qm0a<fV<}?rn
zpXw^?*Y)m(>OQoyWv1DtnlVNZ%HfAOzy4AHnmRW;!%iX>(A^mW3xIUV-H4}wkvka6
zSR2{3EG_!U;U`DhU7m^=+De!xPr=Gszc|POxA2jrAmMO9MArQ+bl&>%!(Kg3;MVxH
zfG$XD;7c+tr`ivqS%IM832to)i-ItbI^AckznFb<^{?yA)qkF$Y5Y5RD8AdJZY8-c
z=#hh9)l4lT9&XO~U@gtWRyrI>YVGG5OpC)EIVcUqS1C_QpX)VU3cFYaddk-@68e8Y
zugN}py!-dSEzCwqp#m5f&A+*KuCrnvjQcH_R@tBqfi_z(b%0Z5_o1}=Bhwz=am0J~
z@|lJ#KMu?XS()6h&rmO|-O2Ws6I<r;F=<%-@!ld-L81R!s;hDs<?u6TqJLLKHOV&{
z(Y6De6^WwUJ;~DT*|v+Nd&+0C+8wrlPGX(eNfbLDbpUVOg1Wc;_v-d`nH)NMVC$f6
z`qWEmdS<3Mns7Sp9c#;8H}!i9Q|sOWrl`Lars#9gNddnS^sNd}S^F>kzZHPU2`<Pe
zP0W4?Q8Q1x=A_Z(=t@(<!Tq^zpOb`wS(sC20B-WS&n#r3t_i=-n+7~v6kMwku;s=)
zsY$y3zFGju6kI-*{c$jiS#TqY<o<^Q2s;5UC+0e{)@+v#ytvYt&(>JJ<KGRwq)HxZ
z$9VXl>sAjLIG=r?HJW>j$#x$NsTJ1CK|U0<TaM*^L|BhZ$$J+(*_vjoEkE=3cjW%6
z-BmY+sC+5@?rO5?5)NSARk$PAoG<U!@ai%M++S%eOBDWXDe4Z7J1D3iR6JSKns~*i
zIJ(E<teVH2BgVA#e3Vr|__Z~-<VSVpnMvEXV#&(pUX+fg2CFR+$&Vdn?!d80Tq({<
z_QM1Vp&|;uMoG4J&1Hi0(lM`nffWU?HrLxlpaSE)(^q;~z~2*m3-)KRQFG1LjetOO
ziY~JzX90q`DcFBw(<R$0{OkyfG(sp|)yOD@?|R7RrfZH3v!b<@<LFdLOoUh3Lbyi*
zKRKWez&U#JEFN^lh-MABc0sLj!CWRDEYK|u^-k|NS~5Y;F%BXlto(HxY2*3qcYP}9
z2BTO#S%B80v>;2fhrVt5$6Y;T!WYXCX*{NOGS+Z(MYV-~TX`^7$wbprFUb$fu9L)I
zeN+3kfTeup<2lSoj;j$E%S$Cgt7o^xMOjC14lf9cn{^dseTMn=P0_U#Wm`$sxgi^S
zV#+PSduFw%sg0ElG?pd5*+xo49Z*utL%l*`K*#>#^j_h`%9I^oUG${ViH0*QON#Ym
zn&msRqTGjCKWr~H=B})yfq!cNEIIgRCf7U-R842=jt(v1y|ejBY#=AMi|}}2>TOgE
z>Y%t{8Ch|=;u=pUM|_O$wf%$}Zcj8{QjlOQ{Afuzpl4?4nrVCdiDZxIDlP6B(sdY3
z08GxRA;r~@VM!v(O0pR;8Uy>m7`|dY9R>N*G8cD?Oo@~5N3QP+4Gp{8KmI0~$9S+o
z<06=q^lEOE49Y>U@#FhW#0&&ADgKEREz*7eW!l@T(ij~YB5FN{s#dfQgIl5X^=$O9
zZGr3qGc(u7p~~PH`gT&d$^eUT?TrFB;CccKsxJKCt~z|fCQOa`83|12PW9@@!@r8Z
zBBP_CBhC)ZRW>ihf2w)8K04k!vY)+NpzD%lk&ZWL<s<Qs8b1kGzD5^QOgkbRP-!9D
z*UzXe+FBz2rXZnCk&UJ3V3835dD-M(q$4Uz{|aiB>X_z#N86vL@XI7|#;hEe_L3Q@
zaiVD}u{)a`aAA*Ht)eTQC>7$PYmbV78RX%Ug1TLHS6a@x;^x+=RF12myJMGo2VJ<U
zf4-L^c>Lb;s_=lQ^W2A(NVh#zTcj3-by^BXA8K}HL0;CMJ<fa6Z5`z)Q1=f<PWR*c
z&lz%*QGxZt<~Z}4<~ds!MGIg+vx2K8nG^1(jDKOblH?_`lVr&%w5A?UQW<1G&dg&@
zL;hKd6oWJVV1E(XaeVJ;>fWdj58V)}^6>$1ywP!4-d1vSEuu3<wV#E9u8W&1T$NbB
z<0-`&Y^nHS_g7P&y6)mvu7h!*MtszYf?|wxP_)dS$jfgsy$2iSD#yTFW8?GLR(}H1
z3hJF1&(Z%XP5md{mfUH^X@UL$0mhj_mk#3jVl!JI5JcD_ix;|9n46r)ufuTvcx_-z
z5GMKdy=%KgeayJjP5c|^mt?BAu7S08ftBQE%h;`)hrTEjDyUgRK>C<Owd>WvrIrz+
ztDRDK#DBypO?Lc?J;1aBrW6%c_1jwEmx6bQyrgI9e>yl5Pi})}Ls_G=+x5+}NU(Pp
zeKiGK5~ELh?kNH_zwOvIw^V`~5}n@iztUgcb7k+-)w{EiBi074i$6^2T{xJsS5Z=$
zot_@EVi*gOKfs5!*nSbuHDpe59rqBt54!d+ASNesXo)Tw_!l?t@$-43wV>!O>Vf~#
z`chPNyaPQ27XU7&_9p#)YnJLU6@1?7o`11!-XU54VPu`hZDYf_XI{d&UEm<9{cD--
zjiQC`_gRexL;t>*g6*BG<o|6A-(&we>4mO&(t6alQaE!=s5ScNnk8ymHk8=vin!P;
zQMmx`f^QNUfl7OL@Qkj2`yCF+&+!HtKFM@+ou+@XGGhYW%8mV^qoY~x`(R*v|7nm`
zplP%(orIz7$EQPuaXl7-<f`v<7jR}RlO|gs9c67LPj3(P9{YtHqKrO&c`E3r|A4m$
z@g>bM<NmJCs)vCK3)fSqx*sL(r~37Koo5{J;>96YdO#KOG67NFb&KLUKHR#OtjT)j
z{8v`VF!M*nU$L#4iwWc;hzA8EB*y&M>`<BUx8v98<KHo=>Y};sXwr|ZbC8JMm$aOe
zCY`KL)g_a}P>4HTV<pNG9&;vh3Alcg0qx^)NVF{c5Qj7JDk9VjN>ll66l5zryG7wF
z$dzWy0XSnmnx<b5Xo>_S?}{IX;TzR5S_uh(99*grupp$s!>OtNol`kiyCP>Tbjn9K
zu9rU*br)vaRwf1(yK%jCjl@=WUB*{S>IbJAxps6cPq2n<p$b`Ix%nyS;~P2G?!0Xt
zrqI5lX#)Y>SW`nML6VB-p(U<+qgl?*HwtEerZk(|W$fObnSs;>l=t*$8(cnDW<Gc}
ztZr$Rv^3q-wRr9w5v?qDPx-G3>IvV(+M*u-$mIT7`aYWwv9*7`|8IK>yqcV3WIea(
z2;V|C{&k9_A74I=@s?FkesQ=6hUb^t6MW#IH|e3sYEWD_fJpkq@Hs(AHC;NKaY_xK
zG~s_=HyzObMfb5s%14*$%^xD0R0zQL(AU!xlQsl@n`KTtg~eBU6@oT@*3=6>vm(FD
zTAAE_Oo1_XHNdi3zuYUY<e(yP7x?AiW&PSyG<vV`YQIb6{BZPY6O_&@Pn4lFtNU5Y
zruA#I2H%mp>#+N)7HOpMBDIR^i@!+%E3_n>?~b~PJa?Dbsv15b4N87{hDKlQXwTZM
z;0s$2g#633{D$8??$^$18=l|xT~H^%q~!n2f0BhtLZhExr^BS|!fPw4s<JXz^5`Zs
zrZ6VK!4VNmd~aQ3Ev(-g-_2(~qI^+mQah%~k^D7X<8`=CzVeIWbT2ogR;G_PK%1OQ
zOTz!QcBzKZnEusFbFZW}=8!H$bx?|7%Vd<V=W@o%F5RC=rTvN@-T-cAh}kw>hsY=H
z)<pPSje+zQVE^Js0f}*vXWgG^?6@rJXqRSBNt5CvLUsJ;;rHXc>YMFf1iJ4nEThZK
zV{6$qz`m0*VuF%6jqH^wCL7?<NpdQc9!)Bn-3E^5wG|Y8zfFsa_JunW-Ww_#j3e}_
z`{CqPqD*OlTA%kt#YP^6-ld`3u=2m}_upEOHCn9$SShcf=YL)Y4#kqW?`SHkmQR4U
ztm<+3Ckyok=GVJe++(IH?R?yjqbe7h8B_N6p70u>$Vr&H3$+bpR{*xSdSn;4qM2|_
zpzHq1;)^fo3=8-+sx3RZI+REj>PGVL&w7VX@&CnwORhnuaUma>l1~0bg`KRM7BkA`
z=zzzrat`tfgJ))2c4FQQ3PxEjO5(8KXMef7%7!4J_B7m7?ir=^_2v7{{i+ZF*I^RH
zNyJN44Tx#RBM6>_0}%RBqwI%)IpZT2VNw6db8+#Sw4?N>QF=UFZb=0iPLl!2?FI|5
z>Qos?Osf8?&Zc(=e|O}OR+)wP_BB#x&$MvgxCcUnhowah#ya9SNLRg?jV?yRlCjGu
zUCPAtbaI-(PuUumaf72i6GsKFc+gNmBd2Sn!>H-%^HHZ3pj{oDq1RUD`p?LCQE#KO
z`6K61JVu^`*9D<b88Xz(F(p6npGB9{k(L03gPG_^c~Zx_3o;T6j1sJIvD<?etK-e%
zOY<Y;%YTIi<>jmrTn#l}w}am%e1=Q+o_lZobMURU*pJK|kUGL%@1hXDt%<f9Ep)IU
zA?4D=W|$-c6?B|ol9+T?>bWOI90j<=)(2R5uA(17AM^xdnT=N!#blXZs<Uh6DKcvb
z#i*tQ5(lFU4yJ>i1@YGE;F4eMo-86xcDr&%2g?wJ{4n1qPX%2&3pA#|a(<wr1T99G
z@q{lWFSG?|QEkB89O3x@*L|#Io0ZDgDZbiY`U3{I!w|*KwEc)<m+#|x$6O?YS7(ss
z^wVA2m&-WZP*uC}R6TmVu&t+4!YJ0}Hp<Da+#w6+`ApE|Wky4uh7nH_BMnNdRfK#2
zv>09ajvkK2sME*n4K`T`kaEZp*qijF#mOP>QUeVnNn;j8C<;^|F{%)VhG{(`S#F?=
z3D9x1_+Z6B##Y`Q<n^9eTU#@V$In$*KbbQBp-XHFzFeCqv7E7$tM&Eq$-Q?^p!A3?
znYIaP3>GN09xZE1e?Z8s(D=avou%&H*NO5!>mJ+r9I9+=#IW(a{DWBFoi)P51GaPA
zvRl>E16NI0rEQCRj>z-B?Jo!z^iEg-7a)@1C7m-G5~}mfj47CPY;}`Ao}<tk?(Xh`
z?!c{G68@e~j$R(8sGM*8?byRA?Da0ZPf-rRUYz%-cn*35g=PQXVy*DG1o>s)lK{&{
z06pbA<4KySyaMrm!Y>DiDqFr4!Ogi#QDt8|>aA4H=jnWbKq7VvtTbL=*9L&K`V9qB
zQ_h949mkT|osrx~aO7fbZ@*_cJ3L+{L+rR}ufD{`y)m!+YlC&)Xu~?SLlZ^~M?oz3
zfA^%uWxt~+vJEiUCeSbp{w#ia`XsldabkbA)k0=`f7QeIXQXlIP*#<&+q2jhM(-^v
zG;UVmFN6~xxdL`dM(=F`o13x#j(C)klJe36F8(xaw6Dj4>(|G=W?`Z0b~<DI<6zxn
z>9FlAtICdEUEFY1_{n<@i5ziNl5oN)a^U!QEzIPnGMkM@@EOyiV*l@FJPegGnnuWv
zv)f6er_E*P`f4)-g~0#Obk=W8{{I&rB_T?_DUnuET3T{S3WAI-snOkagLEmK1C$n+
z2ndLjNY@aMAt0j}JsRon{rTa$zhKvPz3O?+c^tM07vyx#ld*Ph%%5=whjn;w$VRBE
z3LY2mRU7Xi+n2@UPf~X=0qo+@Y*MenXET_4`B<l7e-Ka;6@Tz?iL^6=>XQ9{q<ULU
zXL>Y}vydv+eHGZ}N?uNq67Kmf(ZomPaDs)TZmOB$N_hB&!x5T-nRICU0DV1tN@-`;
zhy9GUNg{M@Ye~FQlzE(9Smjv<6!tvCfMFYu!an}J$~%?%JBa{QE~YvF+j<2Rk$Rsj
z@bQN^<Ws$3mvp~$ww~HUY3B%bB2~5Yb1ymbByP=B<9}HO(TKZnym&}_?d%&NYpPVR
zY$=*LS5M)3Mn{#qJ2EnIOyTkhtYxk9Ui$f=v)wmB*=Pi%3J-4eWPU!r!u{Ug?=u_l
zZKd73FQ9pHLhozdOC76_N@r&zQPPoJmoBwuUk;^i*3&jgWyumQ`=l~;j&Hv_^FV6b
zBq<^0l1o=v7;f%55+Ie7PS|b+MrAkuqMt_p+6FHP>VPx9?6VBAry5m<Z8=BZchp!$
z$X|KT{V>6~Sg^z0ysTpq9aFIz0tCQpM+kBRp1WS#9unE`Vc8`>K4iDdbY<MLc5=rr
zNR_z4&b4T7ckHg1r9p{$OI`c<On>M3aR~*-OorY~#0uCfsWZ8$&yfh_h?bX|RV_6T
z+xC&A`qm9^Uij8=(r~sgDQazN_C(qSA;$Yd*Vo;By1{H?qJ@e1kz;o-i!xwj98MTZ
zg2@rp;8e8CCnF^ZS5~x*kSm6egfAATvSIPLbz$qQw+GF>uruOzzr3qudVu!%Z17<r
zp3TCGJT~f7w223|v>|?5wN#gnI7<yOxj#KQxpBsaF0TH?>#DuY^U}}5;+iJOISAuz
zze5h4Obzr!xkP_|MY8Sn2C(Ym>go{^WUw}zd~)yIs}cG|UX79&pO}b_%Tv?VF~1gC
zKD1zz``$t}!e4fGbac;0T)9iE;Nk}dvi+#Q{t^zlr|FBrLazHpgU?tn?V*<!nV|4=
z;T7dpm!tLs>5sQ`G$y}UdzOb>#k9-1cbxb&OyLP;4SV6vj~csxbt_lO;Yyw?_Nf0E
z3PM(<40Yb=$X{Mfo+Kv+Q?U34w}Uc`HFzWks~0z;gS({qjLaXfbKNa1`yRMEG(HY4
z`4l(I-#>q^*7u&pxCPsk%n!U3lS!_abgY>)+O(SYk#Lrnw7l9O1`g<flk}pi<(rZU
zyWEBEo{n-%GJwrR`_1|JmvJX2UR#b50qvDaBRX8j*{J&-Cb{JJc*o`FXYa5qeMQN*
zb=?oeIaAO(dB1PcxOyJ_zx?f>Z5q@?y5J!zb2}(Ap$!_V6Vv0vA!-_`NQ07T4cEn1
zxLBFK9n#y+>v&{L6}76KS6>=WZbAO5b=Cv*DoqG@na79s^t>w9m*8V_)419@R5psS
z*edDi$I9f<`f_T77`8p0W011oDRb^oR-NK1<4RX>IOIo7Bf^sdVjC`}&V+D=DnaD+
zJu_gFWh4;<gz%~sF_27{H_*sQtfnXG5v6r!QBQ-6g#c-!f$GKFfu`sY{#bk_`t>uN
z(m6m`U$cZDvQp1@^<mbSU0Oxo^}h-#`m5uAa)0vJ)gFqmXZ&hB!Wa`<i{}0rFP&uV
zaU@`T@cu>Q2_D!Rne!X^CO6xJw=0m%dVqfI?JFs(NjZ4h-Tw%}8()y1q~LMB=66^P
z(Bijinxwl3nje9S)ZyG01!tYNhPO-8mJ8>=^3dpm?AisW`?p78`3;kv!6xQ7-{DOk
zfvL?Ihw3@Td#Wahd`0DThf{Kt0nRzxc;p3GjCxrTl4zPlO>y}A_~1GgcfEhP`X<Du
zD*R(w+?69+j8~?fB7{GIyGASV733`^M~uuwSEg?Bnal0*VWwf*;MlvfVK0zu3(EqO
zg<0{jsor!VQt$_(`bO9L`IqbgZx`F$zvfZwEo3IZo*%fl*G@aif@z~7xrJ$s%yTdA
zqn|fjsruh*wr5SYnLqx)bnO2-uwjp?rf+ckoFO0J2l%EVKBgh!ig`ZYU3V2?8{s=T
zD_Xcd-di~dyIlS-KT^+=P?|Bk=G6g=P@wu7HaRpV{`T^wr^cxDKF@t%<6D#Diw_f^
z<`8XkY~sYuS5dr+vx5=gXcIH^+U?ULOh&5Jwqnm%9Mtgw+!9CXZvSQ9f>5{ZTMK?<
zRMW!j0z1THd#GMdV^hklkts<;wGi=4PDUfGbl#=c6HIs&J`0xnVwjPVPz)p|``+o2
zi6p5*L#&bx`pjCsULT(elI{JS;TRKgacg|=66<14r_L;7-vCw@OR$^-?-ujCqqv_|
z^Tr7T&1qMpBAEBQ1t;vg(d7b1T<?@Uu-y%A%}2wQSL0xkU?K5^|K;IYW7mbP%qPyZ
zwgYw&WIxz->%@k)vX7c`2j)A1o2M3xIz2WIFyrX;th9Q;LntW<md1m$fp6FIK&~`}
z=B1=$;+IESJY*0ER@QrWBK^3ZW~Q2HFkUVW20{jo5S)M8`~ttAJ8%DOc2~60Hwn3(
z<(KpEO5wHwjY<2YK@~XLZcy4;f+?0;HL`0K3}gL^WWB8vEu#R!gc;0cF=00@{?6WD
zv_dL5l0x#c<uiFHCtKcWuNEUj(Y=Mx+e^%Ou_()znJugRm6?pYN+`Oc=*g;@Z6npk
z-?H5?Z9eC-U<XIob)TITPx^>JdD2d?K+TZ}>6pIbaIzb*vee)CmCqVnMPrYXPwo^e
z1}Mp0ScSr}jI#hDAKP|XmXNX)iIrl%rvcM05-4tRB@i-i7|5Dn%|u9(#6Y0JLP^D?
zQoKk>5P;>nizfw7s#Cn5a!y%uPY+6x<~iDuM?sHJZAb1w6RiF8gv!^f>9`#`5SxA~
zyzC4|UDBp^aJ)YhF2jjeo~Rr{)qY5OqoNn3`8brLrT{L*E!v1ED}+c4i<F)#Nd&tI
zoI#iWwM$8M@3^(&`tYob`_>Q*$G+9Pi|8A^!leH2nwn7`8Ghv8gqHJ^lRYXgP8DV1
z>;%FDTJ%}AXROu9OSn$)mZg;01am0l3GIiN@B=Tc->MqyFST=Xprpw>H!m6`F=5*j
zMIkp!0B|#EQp-IjBKG+qx2f}-+}{15-c_(zSZ><bmN3RhK)EameElcC@P6dOsKN1r
zQp%Jlf{0#DwXZD6N6u0pCxuuwmQWz=Cw}u<hRA1_*>sF|U7Cy-NWo@@?#tVgF(yhi
zbTxcejQ(hy7T<k+<cAJh*|n^K#wfyICeD)psP>iH88euQ#XPSYNo2INeDFRNYJq=@
zmPDPQ7*)Rf{KzcGjFn3X@ss+kp-j?`J*%oGgwns|=J>Ha4My*-qzK)g7&~XC_32|S
zDe-M;yvnJ`sj}BVI}Cz6I;BaFtGjA>W#)J(EGda5Rc|(UjM1<4C6%P|u1%LzdJkY`
zM?fPp_36x_>Q7*&1_9R>Malu{fF1~;F74ee(jN?BOxSME8~5AY+jOUg+Ya7G%bN}x
zJW>fUC0n;woi{jesVUqF-O_liBabk(9Cc}2c;SA*;aW~l7&5>p9v250`_}ipvrE1N
z7P&dr<DddRZ<zw$yqAPANws|~$$&loh`lat`{}GRMUkAm^s|<4p`WI9gtG5;fdjNr
zY=eFElQMyXVjePfXv2(IhP@n%WoKZbh`_LD{Da9&AGyNMCqWwV-x0Bkj99GAQONDg
z3ikcN6-I|(yv9hBtb3E8l8Y?z<it-_)9G2!7(cGMDhONPFMqXqWVeQv1cWd_o9N9h
z3;y^}u!~6a8<+9(>IREWSMz|A#%vw!zOf>l=O;loSF(3q-Y8PW#N`37<2)K2G_`&x
z&!vt}8?>ukJ8vtq^dl-u86wTnR=b@L*XBO}EyR+vO>Y+#6z)C?{>ysI|4R#cUSD}A
zXI(0#Aj>3%`dB}C^aeDR)5&a}p<CxMIN<CO^oZxycX+&UmXUs6x^X(UGXBb#4Ij4!
zz1x@LKXON*NE;n|4QgbeD`lp-d3~b>mEPho+D|_MARv~oCA_=jdNldO+J{T4-Du>2
zbD@as4pHDdAub%Bj4p?mB(Znka+gbE*vps*jq@QqF7GED$^dA{kd&b${}Bi=&?Cgy
zNz^=FcC{=bo+3mb9QYh5dQ97?B9M+e(=90?YW9C$@;L8JjPlpMd%vu=lYWLahF3eu
zzrl#~4FnnnI=f^z2uA4*5k=lH_}WX^mi4_!8#Lu{OoZFk96&TYQQ-ulBpL^ch;_9t
zstDW4#zm6kuK>Jtex`U<;o%c(AWYUve~!(z)Ys&9pRB2XJv=?oJ=7aJ)EL<wYQv0A
zPaa`r4g;&7V?-=WiIsbdZB!(O`2G|tdks)hMrV8`!%$(BNluiTTwqyYviM40%kM)$
zSitT|<n>W{2?cQEj{{?YiKR@re5LV8`lhGA?8OQ&uW&0yYnJK5zRglpL=x@?n8RNd
z*#5BnK>3yv7wO*>%wd}(y&bpVqZc0ckPj6_B>b2bO$>Jyvl$=>_v9)C^ThiIPq+Yl
zgn_2bP=mJ&Pferjk>v>1x&7u^)pEU8^j*VH9*=frLR5+;GeN3es$PA*c{!KaLYVBk
zMoOT#Y06n>aK^=i_O9WyU7_jM@*0DKr;gM!q7+(vp5FxyEdb>B0OC_jw{D)j#vtWe
zZ#=dG3}-zhz9@u?u+$y)R}qglo|0sc^X&-<UK=Nk6QRB9vs&^N2p;9H)>lUm2sA?E
z1ynwq(rL<(O4$R&r~whQ`>MCkexw{%(*>k)Eg{mIjcFa(aSN@QP;jLr?^gHeJJuj-
zYmSnv9_5&TkR}s|)U#)uLAx5Xsv!En$Q;dYb_arLp)H<sZI*rFtX11w*yCxu{zFRG
zf^i%FoH9uyHAR<Jq_3(9bPR6#&IQ&9H=!m1>WpfVvJF><|8qd2#_Yu&1;{1>D&bvL
zlQ~rb*@-hX^BPq{>PWK>FDmBu{Ds^;hn}8S%=^JFC!T+TGBsD%3k*Ajwm;2(x&7Tm
zoT=-3NlD43OOxk8edCFb45)Z8f1GlY$ajXBKRG(L`r;Sxb&Sx9O1*4vp>r%J;7E3%
zKwEomx9K-2foM6?`+oiqOJ04jcu}`Bp~?V(Jdrtb17>!R$dgOu`aut7RhcOb5*$M1
z1vE^Qk-MM~K33>z`FEqKp4i`wdvq`Pfn=XBUh$yqZ<ntw#-!{Tww}y}zM(P-Ng4JK
zh1<dRswGSFt^Ory!8n=SzVQRyl2rXZWZJ9Ys?Xa&divYW?M9mS)v+hQ^xm_ejIXAV
zVTO>?OJ@wj-M3U#JQgEI4{F<WpMV9{MjrbpbmpCKMbH?4MAm#tO=16F>zD;K(dO>=
z#DYLK{7l#`0i#X96JM^zEHmLQb<hp)%?cXf?bfaWWdkQYF#cW!EITEYmmmb@>H<;j
zLmWX-IehUagM%QKt}}3L4X33u{bF!(b0a6ta5#Y(926ylC^o%)YOgg)Nbq6Jnu<I1
zEGuet$CRFvU5fs%Ohu_JdAsDQ0NMRQgDoU@=nRw|4+zuIQl{ip3f7<UUJSM=bIV~x
z-FlM^h;u3IMG8owz~IQ@&yA4ZYGM(^GXwaZAW)Z1K1vaIcXzj)zslrM@+<>ZE9?1U
z?kA*G!kb-0s8ZD*_vtG2!`(0MNQnm;jxQB%mKLso+s&nJV#CjvKOG$eqyL!k32mPj
z)-(e9!$sHEmNykMQnzbDVQwHBZP&hf-q=)=>7@o!_y~>1rzC$p4d;h&2XiCAteEO*
zS^8&`cx+Svr&VQj6`3uS2q7y;R7xC)BEoz1Z^T*G-PJU2qz{)h+qU_QByf$3A0y<K
ziND%wbvMybR^`*kIwHe^9}y86t*;GQkL5YFn0>d(AKh8q`AhUd4Jlxw))lx~twhvK
zc~z`Y@~smyuAA}O7d8#Yaz;elHgI>KWz<)l7~TPKIz{rkr$N^@knj+aYVJ+MUzZV<
z|2GpDGYZwF^*oYdqh=;mB6}R*qHJYi1tIHsB+6pYWWaB`YF5($k3NCKvBlws!{aCi
zzB8#&>Xx1dh_`=f^gSOsEvW6NZ^G|_s8B1_=(BaU*OiqL($^d<eE%KTxixM1w*~O<
z@%h%#F^=;~>;+q7o93RdM>S@NWZCm-KZdfaPgLj>`I{gyCXzyYb$QvE#+hr@b+2??
zk{p|7K^r5;Z|ofxeQBL(hkj!|4Gs^nSgbCsCrTQt3Rigq>T)GV4M!`H$tQ^znp6t*
z9#8U!Vje-w$&F*M7Haf?dni5Na)YjJwm+KbXT8C)UT^fHt7fg3H~T4AB^!)9V&eK3
zaOMp&v~TmRZ#H8)LpRq9nT#g`Kw%bYl5H13qQ{lQi+V%_5=t^jYO0HHIE%+VSu)6x
zOUho|)~Q163dB*SNq}5|*5h=6mjh_`_M5F4um@$k)*^6ss&t0`C6Fvnp0^nyk}aAy
zx~^QMN>OBMoW-Tbe~Ot~2)uGRO9w9cB^QET^Y6;Z%0F;q&(s^1E23}lM!ix?tDMrH
z8Qq40;744@)uqlfv9oVMf2!ZR|0KT7C4ZftUzi`}Z^dj=N!zS@(n+iQ&t!kob(3D9
z1C~OFAeC!p3f|cRN5!+>Jx(T08`s+OlT<aF?$o`T;6kP4Y!Kl+6V=yn{)`YwO+ep?
z9JH<RNfnp5Mm@rs>k6joLQG-9^)cGb`CN5+%ptDr-e}I2#Q|At4HbejQc(@6tbvUS
zq*bQkRn&w+@lx7qqr$n+FxTwFBb#^gjYx}(c>4-wifzD){XdS2HY1O&(0vQyGJO)R
zMq$*SSv+0lPe#@C65tWTO_IcGh479bDX(esAFWZsA84gy4#47**lIzO#l7XD77H3v
zlSICf(v{dL{zf+5jXX)NA2Dx_%Sqh!_<HFhZ`m}r%Qaz_>P^W02<#veg9!H_$hliv
zFB<|oW`G@Oa4`qs+9)+PE7vi>^cD=8cMLE}&;>nYm-HQInq`OIc>%8Ta@<}1;$O+w
z9eN2SM-`GKC!xt0cJFg^vZWd(C3)CTi1qj6d@zNbuGyXYE9X3{2|Vhvq2;s#zl*1J
zowV)|etzU`b`wv0KM8puP0ffW4KJmJqJlNK0E0I_2Oz0bDJL&P_mW7_Vo?dwuVshk
z8@DC%9j|r{Y|7H}C`$&yzp|wJZ9PLaGqEeqNxkR)oN$+&q)U)i6YgT3z1SuK+rOW?
zj}C-KHorZR8^aEp^?U9YQc^~|frNwA)m<i<U-^ijw9nDA2LUpczvC)7?-62TZJ&yE
z0Sw(!ldv=~x=%de#1z!ZrW)atl=uOJtDp}iA`V<m*I^2Om<I6JbU#3#tC7Z|y4}qc
z^`t)<(S}0qm{!?)AHVszVwbLE2STdn!33_P<#*rb9?!$cb8Gn^O&9+<a|)?-t8*WN
zI^tTkWmtPh69#u`-@uFgYclB|?tcdOwTmSS)b$}LF3v@P!+(Q)>*X1aw2rnlliP3h
z`}_M>4*o6og0K`pvo=L}xu9dp@Wwo(Vu(zBnWIz9%QEoF4&v(!YZuhgzxl7s4lZH9
zG!OUzA>nelzjYBL$HRewCI4=&`_(73Cbj+{>QbP~NJ)$KqZ3w>-CbwE@Uwo|;KU1a
ztjXZ)pPHF(XB!8ny_)K!=m+hZrl$TKEn~hxj##Jiw;7d<pR2YmPVy;OpWT0DL;*hY
zoegWv+dZGXT<s>-c~O!(vr&6bCKY{j9<rwX#p#VS_Wu3*cdjiu7ibVCCdNm}hID~b
z8^^Pw60l)#243LUJeSk}q_5Dh&F}glMu`XsXH9UfmVf^JJ1g=Fc79+=U-n8=udlT-
zl|?5;Sz~Y&neEQP`Y_;amr1k7QZ{g<3N~1Pb$v^k%@$7L>8ETb@^T^Hw+qHh%-_+n
zk&Hf*W_<2zJpawF?AdT<;Qqe?f6PNcXZmj`U}K_=d)whV3gH3-Fv^|&Q;@LRTBOtC
z(G!b|;72NlXfMbU-@MY58Efw^ywUIVznqA=o^aly5glMhad6<+zbW>mS>flLVKn;U
z4CH7A3191>8=pTmxc>E@bK%|l5R@@QnGNfNQX)}%p7E5aAw(h<FYbjuHo3}IRCfMR
zw%7}_(Mk<%o`|l&Q-EHNl8?Q*?G!{PJ!v3kq-Pd^C?a}}+EVn;b{5(H6|=ovY<rqn
ztvQ$~_c5uS%Xpm4<x7V}RMmPHTgxorvT=#n26<D@)!!%TdYjk9;+fT07@n~}D^jbU
z3f#V=v6Ns$!MJ2<FDtft(<;<PbTE<|#gIM{OokGRp@oD4b@YvPNk|hyexh^&;va+`
z7MHg)YvlrNe7?2bp0IY@Z0dwD4p*k!#Xf;2;Lm?Mh@npVRZvJldU7y+)fswQ_U>2r
z(w>LhCwuN_e6;I;cH-0>XLtcti;pFeGR@nj>un<~M_)7)#jI)&D8#{-qPqyCd~2tc
zx`@eJ<bU!5kj*~ic}JtCdpUph(tJ#_Rl;cd^pgD*B%$Pw31{>%yF(dCIJ@aJ7+3JV
zZ#91zP7-G{?aTZe6%3(7WZOlPpr?GzO|V+#nXm#O`|im3ljM`G^}IAPNX-$1;rZbB
z;NVXqE3*n-jyuIwRSf+kHoR+F#hA;g`)_Sv*$6Sf*>%4VUm9D}$R2=gU)1st<f=^w
z;`KhfllpZ6;ayfbVZy_r{-0`|SkhPU=fojBsj{yP;Nr{*!A+O@2S*mo>9!Yrjfs>}
z>?DG5zofYdO*8K*lVA~dSr~@<C#yt_vs~OiKYY0RUiD%q(;3Vkmoyv=`h5T9<M}J8
zq@@5*)upYrZ=L!(H|H*qZ)%!v2WvDLJz8?`_xE>p<~R76uP0{QdWDp<47>U_;~<4$
zHmgFe=9yN}!2$z<K+nIg;0KeWw?75e6oPQ0trV=+yOs(8*GQaT`>s7%v<&MgNGzGY
z7=cB6;c;>fVO9l>CyNvUZ?1~uuSU0P+kyZ3<0nAcyRTqK=xvvkbIF|ck4)whlg?U9
z$kC3=@$t<(Qj@+V0Z+MR4J0|#8<dpquk~A+b%ZpIYsYq9pJ<DPt@YW~eLEU$Pk*yL
zRJJRg{KMoUw5B+{VX}4M@7&)%8)v~u4RT`->xn5?%ahSv<Kye-Xl?#<Wy!pPO~Bey
zrcm~0vXZ733*z%b1HVs%`mlq&RuHvVe`bq2Udwi;lc+5yNPZ!g(2{ZO2Y1Ioq*YS|
zomP}`hAi7}oYGDT<$=Pc*1$}iz7?s2Wv&BueI8`A75<9AJ6qkk*;HR^o6yUB2QyG{
zMNv7Qwtk4JMS7ne?;A}?KiU2ZFe^llp#|qdl!lC648&6wP>`=pRuH?l1v!6bXxJ>B
zN{(2rl3Pn7oiY*disDGfh|+qag^}vCqJ3+lS10aD2<hYGee~Yke59TGz9Rb0W)25F
zYa1y_w0z)lQ4=+%;mgjT)Dt13$)wGl-6SC-E<5EV{ty=}Glre?3oeiY2p@p6l{a>C
zTW!r~r7S)yK0K{KanTi)=FePdA&??MrP}yJ#b{QBexOa2qmSxk3pp)a;*71=IzQ(#
zv8;$Uh+-)>E~JEdqd;|%W-qNWTYYNET8Ip3)8x0jEl_FB3SoE|lQ`jX$M7h&U&hTg
zIp&Q0--j0AcLFdCez+|_@-z7GwD%W(yFZki$^cfCkv4EQoYJfz_v)Xj!qwb?C8)dv
zV<#G&C2KiV`o4LFJrcOj{>SQ`qS^}O(<2MM^l#&xGc`z|>T?%S#`{=o{|m3I=cUo{
z4>?u@?(mumk6eg{XS>+RJ~WD3?g<>mW9+6Jz{5LVd_u%c#F3f?)SoJO#OOs)mM$(_
zV$KYH{)F*<$f(z&rjqHGreIg&-Jd(z{nX^u!bWDCMMZQz#+Pi+x{08$Xr6P$5kMGh
z4#7rPMpv`8-)@bJDoAm?R~ke*k`~{e7HYwmCG(I|lSW<5Y{!@<+0uB@+gsngP(s<7
z9A4nW(sj3{v3d@8%n|+}O7YzhJaM+g+0f0aJR^-T90FZz;~nxUq+hN1<ZMkx`Ih?w
z*3P(0{fZtGkq}4qe85d2t4Iv=N6S7tgq_OR`Npo6(Z<(!(1^Y~Zn@GcbJco#0`2ha
z-xg{Y49S!`26=@wc`jK4zd(De8SkR98lB6xpp|tGDd6g2uf3pP7+W%U)E?>Q|76gl
zr^i~iG_Url9(MZyee=8bv5kYSRy@=kf9HecjaYfg5TCZ{$+Aeh2r58l9x{<$;pVtu
zI9teBdVs&k@N{dc8vohbY~5nUba~40nShB+f6PDBwq<DGWnXf(%zP7ZV1S>;((<E^
zw!0r^901O<1{r_+G>h%sf?d`3{_i^iJvJxG?$aRRSFoxjGs|Vun2vhyll@g5DZ_js
z`ua}arPgGL>_gB6(BgG&Qav>iCI5^CxBJjCcz0K+v(s#39mZ()aB1hX@xd9ZoSEr@
z!|5H%u<JE`N%!?~miuteiMc;}cZz?sy!trfGzEYabkEW8!^L1HgQW6@>uVJ@l+;|y
zRbEMW9MX*j?v35~nfi|Uko{fM)6boOso%mnfCOqm^N#YedbEs9Iu~!kL@-VT0?Jep
z-=ws^p$_dZ4m@Q45u-Q^1hb<>%FVzF$IJn_bhOod76p=~AQ44e`$-pwC5k{ip&#}F
zr0NUyVv;;N<ebIDptNQ(pK=mH%}rJkh5g&sM+o$zY>}ir42iyfl|M#kY7dlFCK&hR
zQvUtz8_pr6+NDd9#BZxt$&;|mgVK6Nu#d{7Z>qhIT{{n$)hM>d>Xl697a$+X1(PIE
z;nVz;rr}PlC7sc@D@c_(kQ+2bTd#`%JW+UY1S>MjGF1sb=OBf$WJ&D!Q|E8RELM9%
z<f{+MYLPZWAAgA^(jMv6xVt@-o!73YyWM}lBH2oSqHhRuFB!oOuZhP;uXa==)d`G^
zdbLoh%W6MZ*?g<XoAL9usPQbJQJ{UI+u&dKeeQX0l_}&e^QqCTzauJM4BQUIfGtA|
zxl$7=;as@O4~qIXB*NqyU1N6$hjDal#EJ%A-e@8|zTkYPCK3CFA;v{w(9N%AD?zgG
z%OorMu=_LL{VQvN$;+i>GeqcVM=;KNx)a^paT2N)2knW=g~X?*W<Gd}kZ1UJN7d&o
zB_ubV(VL0ujy>V8>-n|Xkafe3Qvd|2O(~ok9u>^k>hj0?(#5wO#RP2ha#)70t5{w&
zTJ~>>({-4iwQQD07X^1VG>jSJ<i2?~S-_2Rhi28Cv0@SU_fp^fM@<Tm%KP!&ZSh0e
z#9})!<&=xQ7^iVj16y%<o$n`(-1wwAUOV`UGi^N>&I~nU?3~0HJIwNW0Lb;&A!Pe8
zkjy_!-6uNEG&p0NPMSMoY`b1~Ipg#FthE!KJjnZ=vT1y^fQU^LogH0SFH)7z$ELPa
zJ<jRTgn0j)S}c~b?W|k)mqTU=XGT$9?_h9|x652pq0aZufv08kuYYH})|M{&?OPB@
zU8CL=VM}k5!VZrY=}j@S8w1c-!}~&o1qJ3UUYqgUg<sK>K@i~>m~h$Zju7&${*<hy
zizuy9jkiIHGD*W4#<ni_cyR;0+ZvmF|4hhCPDb_xp^X_seS=(_35ThU;M8Qk|DKF<
zVvzjYyz2J2iSrP9zuOdzSY@?81ttt<uXSJx`|@$d+IS2bniMJ-4p7^=%(+$i&c!+{
z7X{z6<V!BNs|apyHa-}eik1%EziE0{ItKlH0%B(C9(?z$#@A7}9l0)<d|9N-To|<e
zoQn-#ajl-vgKAvouEL6w%=yM1O<bO`%baV{QovN$rexScusgDKncsQ7a!jU2YF-_4
zxjk)(|4ZSjyH@V{7+75|SP{Bf$9=Eb`0gfM&^(89XyS!7356W5gv-q-{Of^t>>o{o
ztwq{}Z3hz$f=HfdA+x3YT?w9SS^DO=CzrerlIC3JnkRef^M1$o1ej*5*{Af}giM09
z8jyZDDGr5PnP1z9Yc$4Tr-?<#Igjf>%@0m9(A0U&Xz#EcAeZ6|&+uDX?LF2{4qGbp
zd2{}(0)Fe91W19MwFi=WjAjLx1-G%(#Z~krE9HCnL6u7SXtn>OG+<*wq-{7dhq?b!
zw^hVZe7dmR(%%xE4@R5#7@o#3cqd$c=w8HX?EErJ>SG#y$;7tiB)U6SDWdp}hb#eF
zUX{gL!G8bw48Vym0*zDjIzKEMwQZ7-kv5K=`Pvb-($G50?|yYPm!PJF788qujwnNm
zOETW}P=^0z(3YMnhKRD4<v!^Xe!yAsJ;MY$z+g4e;Ve-Nb9hVm%9pf&$&}7F;_J}-
z_;u@vO9f)b?3>p$vZzz0Kqzvn6aj-}KB|#~Yl``!0b9HXm&D}>nrYywq4~Ye*s0KN
z(|i_5uEWl_x&c&~L!DYBfj5lpy$jfs;9yKQ!mPrr!g~3i5e;EN@x{KXWzZ!(?r+f9
z9tE&}y>M9#JaMoZb}{C{A|$!t^(=xy6vECV&pr|kx>+<vwM6};Xi%cS>R6JqKg~Zm
za13q(=2;eE1LMVt9yUTB(Y*~HAR9o-i1Na138=}TKT0dr5~B8h{k(o7t3(t}sg(1Z
zVRbx%TAk|OM+@jFikv;kNzJB4(&ZNe(M#U{FyZ#8BO}FnwAwIrJQ%_=L53fs?@Gr)
z6w_X(A&%LV;$ICiyc7B8q#YlJ5J~e!A5?PzXp{doTT#-oVd$A+fsl<DN)_#{KzbX!
zuYI`*L+?YcN>HX#2=5m@HW9^DkhrRmN@>$>*P>9}vW!bXDQ6G>dh@|GSmIwkur{3|
zLOwIA1?9CP6zz-Sf{&^wi4k#F+rHjmSZ8}%rTn;uMafk+OHH=*a;W^hRXdTL6%8RE
zW(4D_xci^^jylQGY~wbDUb@V+VYk%+`NovZBP-+8uyLa)8rKQnCdxGP-O>-_%6j55
zg)Rh~64~G83p9xku2vKV?;)Epffp_GXjlJ`SthQbK(%MoL0HL9NHo_w60zdi>Y(qH
zx%pg^^*a}I_sUXiimIycw=dpE8M6WpRAZ#?cemm?eBS`LT&!h)hmTy;eZ`z7mFx;r
z842~mQ6F)yT@rpG(O6vD<tdGh#k|Y3qj%P8{M66$qMQ)EA71|Mv`_ZSwAUNj+uKVf
zC#$O5{(ej4>S6Fkp<4^)%Xq*>3|gtSFP0_^j`=#F>#r659TY9db{v6SPIG{IeFQ0F
z6=9>*F!Qx)xZHIZR2dG3EpHAJg*yI8-b8jDoZVm+f<ejfcJlV0!sg}8@v%>%$;+)-
z78&PVN@dh}tir#g_SUo0v$OFxZ^~^^)&nXFpSDUrCpQ?n%PBm=w4Cg1FKzt!cpffi
zL*$LBe9Q48qc<hj?E(3O{5jBr`sH45)`PBRe|_5BWTlzswHM2B*I8c|s`OeG^wNAP
zh=Xv_$aYJdgx*oKe}P|`5|a^=>kYzUJ~?VP&C%fap{AKMDmBJPel1vqu(*;=(=KW+
z`oUP}>glK?p(jHObGI<zf6wjg?@;PLY1_S&UyY(Nd-`ZO#|tH08U;YmRt#w*5D|vS
zV|H77$)gzh=dPif6-V|PhOU)7Zp}VneqLT7q6;&D%NNNf_U+F?t+Z0bU;5i8_BP~C
z9c=B$J%X*w$&dI^Bo3llB|8bikIvh8$n5oAIlF4ld>K}MtwjjkA8^_hYG>3$i$HAS
zjX5;X;;&eeMiaDOQv>fi&*wT%b#9u0n}(wH(-AQtxue4dV;X!>f`poU*Oa!u^T?qz
zU9Sb*@)W(I$nsU@B2$^WD(0<K@;PH?e$T(37<ZZ!EXd4Q74c67+O(-QxyiuPu=#Ex
z3!D>_cS>okx0BKZ6kU)oVQS*XP_AwpTUD&tq<Bmqmwq&}55V=KvM+~)>@1N%Y<XQR
zO5`!&BaYM&Ep%QT=aXX8gAA7^5GxV_4K7N^mnbq5Iu?;<&?xc}2*<mJmsIo#A7TO!
zf%Y5{eGekzc+1EK!g;J21zL^G%QL?55vp<V{&3O5Q==2Ni;v@^7IxBSi+H4`4uv9S
z7>+vxHL+K?YhVu-c3llzZCYOcMWF;99ds;@%Mpu^tHT~r(Y)*NctxZXA*UwgE)(W6
z=|xhxJS3X%S3y3jq@=j8hn=4U!eEaFyGP;7De=<_yP*cf9?dv4O2i-uqwoiXN$d7l
zc4SPy7HE~<6*bbn%XW+=NRiBnkK`jU8BVK@Gy5aa2T(KBr^kDXR}ay0zMqJMjJS+&
z4W~S+nXdjmD=pRLb$fFi)=)6&R~@+f_<1mz*`kBIl(($hm2x*k!g8d7_iQX|qfaO7
zFr8vP;AU-GXQ9O(?jN{lp8P!3n&+-$?KXq#iw0wN6pAHe>vhou=Bxwc{~|kYrf=Kt
z5?^kg0DM04U0pZdxA!pZ%|?3$C&$Z6W$wVXX|&|+%)-rF1-t?z{@&=|I?fV1FF~$r
z{z>_Ri<9%MiA>Fke)yZX;Wa^y>&V+B)+-e5N|yEdPd~7hUbOIaRN=z%25Wf<mcFm{
z%OB3_ggKcrs#xEF@ZOznoCD5<ZfdIyYxjWNy|#-I*3hMfqU%Y+j&p^Z>(1L{O`*}p
zzSB*rA6`yoJ=di>k;VOG(rfeEHJN<M_$$SHZx7RXrl^Rld+;Nr^vh?%*V=yC@7|XW
zA4V>ZSA)nCt5O6HZ@c$4YAKW3tCQr}%PP@Q3>{2Yt7_K+)2u6=NV}@Y{(ST;%J2^G
z^dxjEmf<ZgBm<JvFeOYb9eEUes=Ncq-#rbHm`4uZl~SP8n!gA`k#>C)=B#`H5ymjO
zdpmK)!m|C&1hLHDcS%32epGp`9IN5`Eb|w|=;LZ(D*W%o$99=s);|%P_<h0~zj`SD
z+qDhb7|I3nWHj_5)e!pU1r1IMv>N?@dt0`_95Z_~f(C(Tq8F4~YcZ;wR{0}HH@W7b
zYb4*{FS#yg19dK!r7zjvcJFxovK%$XP+w)z9cP8qH0QEXrY3?!Q>g9Pla~}1^FgXU
zy-7lwbZ32wv#L33^C-HWfZvXGhOOz`>=rSbNg3@tNnXp5>GA21(Tqy(9bo_lq^rq9
zG?RF&Q<Kve2v??cP4r>rS`|ARL%o{xV`EME=_@D34fFTm52p$lBk|x*)CL&I3D}sL
z2AYHg7$Fd~F18JlUI>ITwfq(R{rDFj`b0>r7yL2w95cf4RZ882BvEQnE)V%DS8Gb&
z&d65tNfXdi+FK7Wd7~1B?%V^IIMNV<#)OoQ@L`We(;BGRKLvC{C~c1W)Tn9kpySY<
zym1-6HZf|{d)zMT4nK`()ei@rEG1-}mK8bYNBs2~!XGh8{}GWSvFlNlE5jZ?Xa4|w
z0VCa5j=t_ZX};mKp;9CfV~=;xaOC;|RydAGSHH6t&G6Pvi{+Z^^<n_-U=dFxDy$wZ
zw0buej?py)YBBzDYuZ+S##mqjn+fJlLtrHru6nzAhmQ(E2Y9jHEZomdPlw`btK-!k
zTRR7~V(iEBn{~j;sTOP+^t7LzpPys)mN1ww^J$VlSS2r}7#5gBb!NeyoaYJ=l?dv3
zpLk*6&ed_hf=z*Q9^gvS#gNX$X!HFuaEqb{+x8D5$dy0-D;7G}ad=|5PJh3!<J9MS
z-u?T6M)20pDm~{uvuEAeomb#m2Uk33U|YXyc?Cw}E=EDus(;vaGx+WJEfMa-{}H7$
z&9v?R9X6i>*7$+_ox4I||7ch*|5PXhZm+lA{sV5gi_U=&FbcP+!zvH_5g)zDv@^!Z
z;#xb-;kP@tB)qrS+unsi-l@s8g>xYg8{T?$Gk3e*dAZ81a1gH(Qv5lq8PksA*FoJb
z0Q;E=x8v_`*B7oRkV2i;@Y=cvVpR(H+vtUVr_G&bL;RM&dM<Xj_!L8-!Vu%vVBC3g
z@qqOl*zey8y6D@ML^eP%bf5%!)HmlIEMX{Qi8mFdcE<YGZMNyb#rAX&cs17H#UjLN
zl)!tk!`$PUZ*G?b=i-T^*M6a4<)qI5MaEDCLT57#>f&#<dVMPm={)<Z>8!>DZ6{U@
z81grD-f#4tjOpYNqOCc+jHA<R??HK-k{W@g#n0QZVdi7&QSZW$yj`lVR{jw049uie
zs4Tb|^h@hLwba_4`GxQuE&uHAz0>X2TgGAJ{ApXJn=4IBS+=LXF#djCEFt6T)YnA~
zYAEyRd*)+=yl~OBKfK9LdC4n_sWZQ@%5XLN?_QuVj|>G{)mn7&U(VIct~4|>xi^mj
zDbYGQ+IU*VW1+?gQ4I~meREact=cX3_Bt;gqx|hV_C?C;YZ8hL#M?FODvs8AgDW1U
z8sgz<N||lqL=q}XH;^i2h!?yPHy^&bL5v@<bA&?@^vFb*o}#n|0`xQ0Tz0%!!w%Zd
za2+R;&7CL7?pJ8swY(`OFF~n&YgHoXf02mo_&l5<ihq`mVidFgcApgTqg$BV>ce0G
z-S_*>miYPSd)8q&Tqkc@%^Ullq^ePQy?CrV9A^b-C@yvQ!^UCE;)qLoTCsy59vqNE
z{buBV2(QrQG{jK-5zft^rslBLrCs?jnd2FYfGGVNfA|aJh4%Yk{=!o&A7`Edsi8J6
z9&7L#+fX%V^r>2GMK^OOXT>3%-iZjyin7qYDn7kevA#EYa=1L@q{sOhWqq$#fKzqA
z<bm=l2*($f95ubBK1DkHI=3D6*OJG5Mm?UVMmZ^J-qiLiq;l-`-4eZW<)@+3M({TN
zKDRF-bX=ldb6-OjY?67)zYAc3flG5s+B>cb#2j?VzQ$ODwvOcR+Tqgi*Nw5Ti$u-X
z*x**;z6(V0n%ANJ`Khe?=+B?8a=i}==gMUjEZYwfEkianI}7z=b1IGsx)ZZ4fNSgr
zDeWIE^;Y5^@4@P{Q#StJOR#BqIXKp_UnZ1*&-Q8Ke_X%Iv%Onjyz_jEzVqL6=XSgV
zk(eBm;hZ6Gm2P=GHP*EIqp*3t$-NbB8L-suzW_>IH;F=R8IzGGK~IauxYGxb{(B+d
z(mvkQ0fI#4LNAf6m&9*w7nQXy#%?xm&;R}petI%U4BuWuwYlbhbbh}#f4gOPyV{ow
zKNYgL>N*2{1E){$<$bSJ{`bPDw!FqET%^8da!&C2%m;^80H^IZx!bEF;@Iq)aqv9H
zBUHjIw71GA1TRH{Xww4R)p3Oan8zO)9E^7e+Ai>?lWo<J^E{bbIL~S*Yj{o)seH3n
z6zVNsq?xdAbunGu&>|_Xv2`g?=8$xLJ_ab;>H$g8F>f5Ie_~og?~apGyD5>euV#3m
zikHd34cYlviJ<qfuPe>t^qD=czz{E?o+fS*Y4`W6;cvYyA{2N|G$I}TOA><HwF_7O
z+tRUUzv!9FzTdh3^KOxOEv`NIa{WAz?I$_WY%T;$T_uZX*<MI-^{N{i59{j9Ap_Rz
zzX3vw*3?|FKeBcq_c-Wanbuo#zS>NswU5)kBODp@Hk6t4G=gKN$A4!br1(8wiK}dh
z-)Etwy8ERrtKGTjo}yb*wYZPSoktwXqr9gDzK<Lpzm+=Pc*qpC)vL4f(89Fw9Oda9
z@>NDg{&^4he^YN_x^9*Z7x(6gt6gy}>d}CVs;&(yQ*>zx{HRy<%<aE2%Mm9Vt@}!P
zU!&=p(%yI;%ubHAc^tzPl)e_@bBva&vx_YY6B=~8Tr@TGMP<PLQ()V8+?S)rMy=xX
zB=H0-rJgs)%K+sN9at6!JJ2kW{T>WLB+t9jbrep1##+7z7dX5rkX>QWPWtdll;$JA
zJ1kn3p(h$Mxc?{RtCJUsomf~pO2ULoqjWKa*~``M?4VdzPFfb@lGmkC_(fmqoeMeI
zP=s4t&<<~~?+XuubFP}AqN7VN{*Ro^_v3-KKZQA|#TrEJK%o6Kg>eHWB~^;}AJD?K
zDGWnDtI247m$BLruz1>7$#sYsKm|+z8?7WVvJxnDIAgqInyw<=%nTnET2q<pO}<<8
z>{*N7N=(Khjy}rdMu-)8U=!p6ANmzaTEhU(dN@Y0*TewEVYSXs;!VgC&TChP##avd
z{eeZ~y5DvvRn{HBt*pzzv28YFjMr2x=B%}O{mW$wl#wMdJj2x9EVJm}jUND){5yD6
zRbEltXE?^Y=szVTG`Q-In`k(kgN$#^mByxK+bguG1;8r7V0}61>FG7yucvEk4X`4Q
zb-8{T0{ipuqR`V*q(AsXrH_q`4M#D^Jd^JTzPzTl2w3VW2sZvtKYsIg)u+SA41^Qr
zFI=tr$bgDHhZh!Au<Lgh&;Uj3XF(qNzwO?(8M`-apLAj5A#I0^?##^0S{2}x_;d)o
zQQX{MhK1TGWA;ujC3DX=Ha=V5j8oiR^}W`)y{b?Ea7h18`y?7sc}E3j?$54&i;dbJ
zm9Sm~PWr2H@e$2UIl<^oz_xdDT<F{>ly(KAvEUY3f-g@mjRE$GQVv3|Fu$O%5QMiy
zL6P{}$;^x~`Mh<@PV;v2f*B_h6;U|CX8LTTGVo6!D-~nb;qkveG7nxuAzYp24})>f
z58L?ouw2-?!!z@;s@QxR_{skM1Pb-8ij9M_*3cQpkGl)is9ZQdJ$q=glc&B^7`oYV
zGV@6(2i$ks0m+3BkJ~+B3E|!>b7T4`H?=lkX{c({z4P*>y0+uAcPv!fF=*+hnPbP_
zhi;g*<j;jdLaYlIpEo>GrgWNY#k~FUFJ8A;>-ExFJH>{%zQp4p{mPbw2!BLiRs3_x
z94K6taSuv;qc=M0<KFIi;nrwT@M+?3f0Iq=xiD#jA+Si<40*=WlOt}IVg`#A+USb3
z8Ls4mzTnFC9T%VD`8M;}m!akp!nY|$(Ef7tg~9f(ud~k&*V5vJEOYmEarCg08>GMG
z?acP|hu*!PITPQ0(pf0Aes*p}n|k|%1jFHanVCu(Ud~T=w^A|^d%KpN>V|CZ`p3Uq
zx(Dub;y}Dmn}~t+V`Xt>>&G0=ng4rsduU4Mfntwh4r;1WN1jnVjdK{{8<bWPY1juT
zu-RePN0#TH`C-=`I2xMbPh5k><u&1jS5^IbyP7n@AzelSDY5#&Q0Rq%D)van{&H!;
zR$IkU=`mPskp5w67MsbxaO#x#i&LS=AN}H}4M;?izcNCm<&SW26Vd78KYH;W7a_Ji
zE{REi6NTzqN2ri+RuJjNTV3UPOT<tyKwh}A!eEpTqW^H@5GhTLa4^$MsH+~s3uoM{
zuu(NsigO5&#LDGr^~Id?JSLDyixZE+O9i#vI&w9wgrM85YfZ`1<c5{m66uOqfLk75
zrqeB;Bf_sD!DG8l!&lHINK%~qrXWT|J*d)H9D6a7!1+%e6R7KhR*yjtiX>L0-5WKi
znL!kxVWDR`rNT`MpULw(j44K6=qERZF2sK3O~8LX=l4L_=kpI{E}yx{%f4B5sLjs7
zrY>1J?N6^3*?gUafa`nKWY>?M=Fh=_ep@qb{K~=ubg;XO--k_9Fz&*MTIG+MM9d;;
zw3#*@naO1=%8t~2XQmss-cTa=6<kdP4cgUucy0?J`3$L~aa3<wF){YT12G_tpzGjz
z?xXveKbB<I)cqzUuAYRdjA!6cu^ZZ%P+CFZD81fUb`TGi)Mbk6j12EF@Bv&Dy8?$Q
zD!9SD!*$N3c~1SAv4B13`{|<;G4bk&Q3Ml~#}hY){k82uCqq-$*W6(j4MiPShnkaB
zL((TIh!HwF<5(#*6k1mAE9Q2SANVe2YUGTro_ruF4^q(IhYJ_ilcYVuMzLYB6(T)V
zmBzfQetOfvt*za{Y;VN?I%r#i$T9C7tQ_M7_)vFAe($JA{ZiNZ)iT3cpZ6FD(z&-t
zpjWFORa^y^4YqZ%d|BnBN0+`E7<{yTNl$TBP&@3D`@wT<Gx|cK^Ha>JxLX=L>$kn)
ziWy0$WNC3tPAUwK)=T0Yhi*|?&np9z$va;47sJ}s-yO%^kW`ugU)V)`pE<U^u$C0Z
z)-5m@ieG1m_I>sw`dD70_#=)eB~4u<s*abq2L`KxFm7FTy!Uz|lP%Eqrx+2=8zK$j
z0>Zm98Xfaa>>G@BL??Vqp3o(p_+4UxC0jchn(o{&{O)AHt;D5Ox7`ePZ@Yc0?EXh1
zsfRF48)d+*_A2-r0DR)HFTOy{NfsQH0Qt6Gb|}T(VeUg*w8D524<pzmrQu^qmAv<)
z3H!CJ2_L`UG}fmTH!@8k8EY7m18_88lC7X9WLKx-26uDWx!`L)=MQ_Dt6eerC7Jgp
zkNdJzpF;tS5`+=2C-!5kR-=(tm7bm!@4bhKcBa!*BBNWtJw48y9$27Po#k>B>F%k8
zg?qDz5*)5YqWpH?pRjd2UoS&qF$3piH>n<DY|(Sr^9h@OL{zaKAK1M`_{(S;CbBJV
z$jKLIt892{j3m8B$oRA+A&AC}?%L#ykVB%+;uJrS4M9oB_^L>%IC2${c8|0~-mYJF
zYp(ei=x~<<!n}!%C0`<#lp{$<v~r#^;3>Wu7iDMTIPQK$>xKyC;CKXk1pOKlmxD*S
zFJgQoG@*_c9v7k;!$8SM`v`86#ADv!LuvEZt=YUe7kv^UldE;#n+|I>-E>s&P|<!J
z+}^!g<y$?Lex8H%HPeNl&&BRBRKxL0ALJke9(VvOW58|RB*@<@1bfqd3|?7Tc5p>3
z^SN4EM^i&aShG5e8>EW$_Fw<>XLD0jIe2bju4ilye|k?>*GtJtgYnOKm*x?BiF}PL
z&?nm9Y9KpYQ0{oK>v6K&<?<cG#8H|!-=j9dy6DD)#j)QUbl#LubY3m3?H$G{TphuI
zqR{o_UaSo{dqsQuGmZ{#te?iY3hPA*el;jh-jGGy%x24<3Ek{d2w8R<A8K~~{NOsx
zAOY3a!E*eLeMJY0Q<?S!o9df~hJtaHVF%qKZS$W`&$2ErgDEP+bCee<nDyd-HLwl!
z!NR{j`*x1-HS0I8U^A3rtF$u1T#5iwN%*70qN{(ke?vXMAvb2W6KqNL&xMh_cA$gd
zO=V1sk4LWUb-8qCiBOL3b?r@66bAf>uBXtIx6vB>Oh~N@CF`F3GFOAzRv-6AveW)&
zYC!7qkbsil2l0yV3DO-<rK)InXzN}x=hiH_(Eb#e4Lv-JwY1%aiZU@-Zvfq|4DA+N
ztec~mbHXW+PkoI|4MC&CT^j)swpYdmb}(*rl|NB`0xf04(?lPvBq-zQruB3<8mLFs
z6_Pe@dlN$a?94eG-l>qdqMpSuSi{M%8ZTI8?Ki-U%u(WA1RuV!P10uUyP3=>uD*?=
zqm!hNwL}>*x7c6DBE0@89KdpF10D(41$xfi9sT5$h$YJoi$&p+>=WC%PH$n}>fQ5$
ztz$p)iPFRSFE^U|ytJet`V9qZn6<?0T1oS(+3C(ki$<-$x)qiDpOm>X^jVZkAZdz&
zEZ|@{df~RZ^CTOvz5Q1N*tLj7GL!SJR!WQMrsO;ar&d`sMMUw;0DHtnS8uy)Gp_S!
zJ-A`U7HprtX1(%&e+QXNFVeVV?&~EJqhsyWq^qmWqD3CV@$}ez>;};gG3nR(!C|73
zVJ0~fT(oZoW<SOJ>~+kvPbd0MpVm#HF_VVUFu?xB!!f)nnfr>#cjHC+@?uG4AxYhp
zsFqlEy(xwgnFP7_KpuYHGC0XnHqSIZ9LNm$GpPS`<in7-Q#QwY8e#~Vfq^*h(7{r!
zDvezXw3>);Iv5OQB|U!n8DjH^Cs~LjB8Q5-Ea$Z{Oj+?2bbyy}xQ{_)BZ*yESr`??
z5VFL|V#yfcFya_brOrk0BTjYG$n%OQGEaapRnc+e;5pSVpyr*MtTTP$x$trD<?y-7
zI<qOks9&R*ptlx&Rh6$fZg&?n0O1xF_d)#$Wy--2V*x8qk8g+}c~rrlEmTy1@NxA%
z_}SHk+1uPCbrvj0MflMr3bc5jWG;8aBrSq>`+3)PgIXJ!+fG+$qN=1V+X94-3+(vr
z)=FnbcYXmn7yNcFaSJlvJUm5nS(wdo#_Y+|^*Hc<AsW8UA^kg!d%>s&CK$Kw<lEHT
z3eJGl3HMQ>@K$t2^b>FZyYu8UuZd<bZ}Mb&dJofaz1Y<rB=by4Do2}}x2&G;DR=Jy
zYV#L`FN|i}MFKoRx4<{??JA89M@s3{PV$3vKz%eSMqO{~l8zKn=VGK-YuV~|63-2~
zZ5P|@9>a39%6T8-p0SD6Cj`M?YuM=u`q^g-Sr(4LKZqVbOWtyjmi<4P&O4gz_kH7`
zR#htqLVao!P3%xx%}`R*imk20DnjiR6}3m~y=%`<D~h6usFhL`MbXwSt=8Ut`Tp_y
zKj*wp&Uw!B-1mLm*Y$E^_jPt2ob&FSr+!UMTHCF^26YzJaTuZw6;baeW~V!V1D#!(
ztreZG^Ruj`qrk&Wa3ehpvmJUN*Hv2~Y@Rv0@Z?j39;g}c`o$qd*gxFPnvBaf>(7ex
zAlG7S4E@Q2jy&Yiilmo|gU9c7wCyn)*S5uO<RMA?70V-m(u;atlk$1pznvY&NKxsb
zqYh0jKC>-0>QBi9aop&fCt&Wmf>Q#aTVEj+MAQDnRKSM2)h3cfA&Q1^^xF+A1~Vqb
zMEnYrVTl^Q-C12bm7ysCt)hm-3!=eB*~K>rcPa-PE`N?4dMv~lA7+g-Jrc{LpH|%~
z{_qg7nKxz?<bSwxk?6v#n946eo6!X6%?MFad*{3H<nlYGX9)Joqc(E=r?yCZoh&iA
zEnU;-LpJMsNB}J(FVp5~(Bo4q@yGV%D(VMo1c3YV6TU{gdanh0^PJo(t`}xeSC;PV
zemXx{yVy{>xU+ah?y=Lywta9_DiC!@S&vVhDb*Z~eO2HmAfTklSQSTn4j+9HN<$;4
z$ZWSQwvn8zsAq&kk{J8P*T+l4B1@~EtLX^<_;ii4V?idFx-l`J$@in!o$M%3nsEj=
zzctn82ciVFccpnu(Z-D^_0ip%L{ii$s{zu7zjjGih(1c@jV%Lq?V<;gHTRJBq=P@=
z44zEfe`{cvlfyzJXzNhiZVa#fix<=)@~`rP*BF{>P&m2?C1nEJlR5y6)KxxVgHa~*
zIciI{YDx@HCP17`(ZuHZ(2cVsep0MVj~RWg<&AJ*O5<`X7)-jtZ|z$lh3V{izJVe&
z)`?7jJBw==we(32S}@`<iaw(d>mi*9uid3={Nx*q@cU}%B9Z%ckF&~S-kqKKF#Az}
zI#H%f-4Kj1Gb}Q_qS<Qk`kgz>su!_Ix)p~DUGfE_CWq}aY+!3mXIwHapic@nz{{ul
z;(Ru3x=5#q0Lj*$F1*=i)w(AnQ<=Xx8y_a>Ms+qNd8V^vv2a?yiEjqP)o9H*pf&%u
zk-I<f?|jHNaC?1Wkn#n4O4j)9@i%VK5?#465_qe#yM8oS+D$&;-}7!vL9K}{tPEJ<
zr@AaAuz;OYU=rb?>H}6kfv-l->TDI<aS+gSJK=vL%jlX-nc_X@hemdL0{DpEyA&5a
zklJ^zgq0AA(La-KU$NZuAt8;UC_m*84^8})Jo@d>&MNa|o1@d0yS`7As^<vz7H?1@
zHF>D_<_s1+@IQVNU@Kc!SY@whIZ(acCdZj)j9>$Q+$rCma#O>e4o}_5G1Z4T=jd|=
zOp6=;`1PWQd=4}Mp*%WYJ;zvT)cvTodOLMcm9-FiM=GPiXD({|I)zPmmDPuFRFx6X
zBrjL_rLf45bL5z`v$x!$s3u2Hd3y%Z7iYf3S6x%W?*INtoStk>fk%=SPcZW7EhF@=
zP9CgcEkn+DK;yknAx(W8?S=?rJ_q)U;-`>1WuT~w<>lr3*o$9yn*gc;Nl^!-cfrTA
z`<2WRj|xcsd~4v@K<jdX&=B8v+p8S2lo-y*_Pni3M-1-8(taD0qOdSA-ALb?WB9{u
zzBj^@h%S9SBY@?t*q;|dl+3~I{Z@-DPF3us-dau<`vka&|KAFnCK{yk?KV+Cc!2NO
z#&E_b?8WZm%QWoi3U;5qFPV5@W0(@_Lqe-?Yj;K4KlwR0vts!>PS*q(>t{DOeyN?}
z;6xH2DIrsMO!Y_ML_rqhm`J}qQ8)2+9Bg=A2KK5VzphoEff|ic){&2ZQB~yoU=4?)
zlG5HQxPF)FjY-0ROp&NJWFmM_&F^gbzd?`t>^cCdsptpz8W=7qYL=*~LHd<{Gs^zm
zG7FYh(#PRm;^2#qt<E8J3UYtK)uF-wu-3v7!trdYw19u@C`OVwsm`eO-@LIJ7bP_<
zsrCi{=gjr6zyxUIWDZ<o1h^8ged9E4_`Iypw|}^w>plqxWq#@Lt0UxXsNw?%bWc&I
zl_#o~YpHId<Bxl1?x!_vm`%g9Ba*An57}yfzRsyl%STC=Kl&-7=M75?hgv%4br+D<
zZvVP#$eireufxJlC8g}F#Xzw{yn~c7s{j2<#YJ<FRxYw7<}4CQ^V>-V>#qUtx9*&;
z&)a`Nf)PP9aYMlkHM4|yk2Z9Cx!9eTi&A;@n+FH}O)fZlr2@~kgc#%TCb2pgW*YF(
zY0?UX&KI%Xq=ZlpvS>ap+x^Czlk@b9B<g3d9jpP@dwsJc;0EC1*K9xOf;Ao%#kEm5
zjF`5RHh`K~Bag?JAk?r-L|0N6W2*Pc*mx`5&*Oi`s79Bz&8fhQDBH{xi1X(|Esi`B
z1`*=3kA3dWw|3~ul4Z-Xy&0jgPv$(DL*r^3Mh@oNJmiWtUKXDNPeRno<_8R$Q=9TN
za~aF+Z#wN3e2B4c1esZnkvmcCmJ6K$BVNuk&c@mZEg&zdsU9oV;dHapvP09|_zN8<
zG09?#Fh4iH`SSGp5h&sDmK~(BvxX^GmR^(k$=ysYEgsO*Zl9Pu|E`e7U+))}aS7UN
zcRJNVvpQwpn3_8F=Fb&lP&(HO1Y>7BPWOH>KknRLz&LFV*W_;NbB?TBgxs^UwBRd&
zM1Zp)#r}4;XF{>B9SPb`?OqWLE3R#hsICtk=i0rBnt3uD=8%28nIE1an5&VIgCVA8
zs6whyh7Da8rrYZjA^k0XQ#XKbvkNDtmY$w^7_w!<n$mp3NbR3!QSe$p+hWKjC+9)K
zs8cfqHw_}@egHxH(?2gsRq~zaE(I}J3h=Mdh}cc9<YTYOuH~EXth|V|^hQ<WM_%>8
zq0<i?dtFdCks+xyU<`m3F62!61pPdsa`*N`Ui7O+?CQ-9#!yB{=Hlxw`-k||1p(4@
zGBFG)I9cZKQz<50rMUF-9Zdj0-7MQ`NfOBcc(Tl$U}ne?OJM>lN~A5hPov?E%e#Cf
z2x@>JcGxPED}9jN%l&s^Jh=+3qH_SG1Efc%(E<SI;1|Ih!DXsyBEN7us{9e)rRV&L
zYx}F;IJKw!>LsrKJN{KC_Ups*?ciyDcg4Z(Dq#U`l)pBuV%@YTplXvFVDhV@o06%&
zZ93^*;Ip}}RBnc32NV_-oOlyS@^;EhFOdpZU9IUs`<VCvY7?Dd4qel~3lE)<%6IpD
zQ{lPwqHzLxHMQ0mxflzw<78X>fIGPCDYqMwjJ$!5ukYTMN%_xGBCNvB*Qi|P{8fvt
zsNa}TYDksOGcX2~J|Q|N&Ckq_QP<a)rtb$nfX5Fk_;-#e25o0`(mSu+ftAw_-XEp-
zKco0FDc-mRcS1&k$-2;eRNaFj6#4#K-~LTttl%})SCEpZ@t|$|f7fk3N)3QbhRKwD
z&*D#t)ip2&v0B$rmW=3byFa5jL$H;6bmE*}d`$KB)$e=DX4*`7PMzY2ER=1L4oj~>
z%*UA-GI%M!w&uf&JZb-z42oBzq_|G7p}`w5E!(HRDSFc4#h$b0T>~}iAIs?vZ{=gx
zY}DM%wQg8{MZN5nBm4J-y4T_30+k9V7=VIiCMdyq{&?Ph|Mg7%u+(Vn#>tjSbT8Ps
zPF{bE@H7u&I1waSBVsRI!0^m!Dt<*eLSe6u(w!R&ciZ8o?*L4uWoS)U47jiW7Dc@S
zN3S1$rIY7;y}JJ3_N}}a8Yw2ij4<Y$p>%|>s8cUk%dpsu{BpX?KC~`ip~ok%gFp}&
zwz_A>W7~1CCnNdik6n|8m=s%y#H!3t^^7|@1~W;NV(J4%0RW?0b)s=fSMPR#mzX+h
z3MDu?VS=dD=wm*97MT|o&#{l6KelwfRO21IEc!i=)Rler^80<Yiy=3)`&29dK>$Z&
zok$q`i{HFcGgW9KZ7yxye$V0b#V=py5P}lLR*F93HV`*KBn4ef3jd;YOob*E0i;Z%
z@U0{+0F+UU6`jpb<d3{18}2lvPtiD*u7oY~P|J+pUJs_x$J+}hV|fMz|9G8FTk05%
zEHj12t-@a~k)l$r4z3?(H|-q39jy{2Z!tS3HgNU7ehnuxL4g-L)Ku?-)O7y-b1BW5
z0ayG;i4j))8IU-l#~m2{8k&mW4h)Sxk7^PdR93t=QGhw?lrd#%Yh~KEJOhtJrn9O@
zgTbErZ|^;_rkI`aSZ%j4;7R>vzPO;4RY(0C87VGcw^+rbwHW5$pebz}5yKHbp!-Y{
zj;yPRg=DS>3Mneu8Ds|0XBFu?1m6>NR<5s$#gMp99tln(W2-tx<_S-yH+QFg?QI<`
zZyl9wc7T_3(=0+){)8Bl_Z>uD+qJ<mJGvW&^%%l*{p{4R)`iEzQ?&+@xWJO7D{+c%
zet5mdi6=A<rw#Rk*)+4qvVx*5&?fLL`L+4gersjx@}O<WNAKSDU;!?8QK|_A8-rnE
zX6HRx2xFG1I<T)S^`u^V99EW?VTE8Z(-w?MHobL1C8QYsvU!TfsA-k2(Sqy5sX+7Q
z$#pj|reqpy_zM@(+`eKN05<wg2DU;_40?9@C&=?)Gi|)iaJvf|cu82L+DSgaxJ*Zx
zp+>dTSss{imbW>{RbMX-kc*;3co36gzJ5)`RGa;z7kXv%aF~EWQ^Pwc9p{2O%21$2
zv+6AR@kOb#>Ym_N+bJPxiiPl@@MP&2EZN6NAP(Dj@8(H&lkw|}Y?zQ9j8-kj0P>KO
zl$1$-nzkVzg(rP#+<v#KZ7h28JG@^jZ;sz!Egsy8nbWfqUu_;cU%KndH<MK_PrzKM
zO5WcYCw~aTrQsNcAvCv-3~e7~!R17qWM6_J-eSTL`c`NVTE$60P3>}E&a<$dWTz+|
zhxKmy_IT?*!`06EI{e`RXGsn+cdP<Ku$!|a#eC!_@Bxfdh+w>ECFQx`bLGVuLU?0h
z4wr#drdDSwdLwJMPmp!!4L>ykrTY5*LF_}UyUPg$tW6B)IXxe0p=^Y4#=gC&X{<zR
zMzj45oJ#B;x0FmaC3js5l6<*bu6BNc4d|iRr{|xqLYvZk%*U{>(qMfzuHE55FYK3N
z)at%}!j$C1Rk)cka7<*u+YlD@iZ~}Ek~I1TBn>UgkDZV%7fas#>g%`)6kvI=eSb^a
zyS_#wLdtaor=h`X^+p7sbP8cqHZtG>+Urjer*!!*dpu^*H9`PBBa&2!O%gV+SUM`n
z>(j)9V;w8;<DYNY;;Lm4t5u~Q>wQ0mGSO&1TWMZNest_#Fwgt@bjc+{v#7GOB+rB}
zxuj1D5*FyU_u(ltfX+_kfNeDj??=DI8dG69$vihTD?LdGpbjtX7IEvu8v78+wVRTa
zj4UOxYNuQodX@7sh{y<>D7MBWe#5D=K;ocJDjk^-1p!iY^^Je$dUpM)8Ikm3r$O|X
zHj*MdK^tja?iu}vsv=Eq@?fDEMzqET1`4oN4B!$ThCx@lH|u06l&yOzDtil+>2W1s
zTGgDZMkAk-eCVVSLBK(ZL*V}4AZ6;VZB-;^aIav^LwOIg=plnWI(EOo5OsWXR8_x5
z<nAL<yaT^Fdvi**qe5Kp<@~;4G*5CSz?qlPU3d5X{Q3W70{PgCyPc$dQv0NuPXEoF
zu3$wZ`J;$=igs=~q!mv1GFD+-f_X2)boiN$d*Q{?hU?GQA~EknCqyN;B3o1J`^-$!
zA#o)N2p()as9B`B#yiC4UCdFma=y>IxBtRo4@e00Yg2ZD$=d^`m8XMT-4v}~N-)lD
z`YEQSiHDeC!U%P`C<OA?fHMtD-AX8a6}D&)#hohFziiBKR|I|6SVh^kSkuW6=2T@`
zYAwb&3LUVVtL1JK*B4-%AAV}4!)A;?)zTHPV#6p0=jm;@o}hniJ4{#yjvjFhtD<fW
zr3*GJrODcjq>kYS4uA)eZkHfpciyy?E;aWpnZAG1s$5;;t=dynL-1(haK3hhF?~p%
z(a<53opBd@p2{yKL06eD^;wd1r`fKq?b>$Qb}}oF<qqqc&`E#kBjbq5w1cy~cbkFb
z*y}P9k{r;$Du!1kd+&QWGH>7Ejvnx~9Bp{iSs#j9x0iK6hx*VsXO48&-}$N!2MxyP
z7D9kwA`hnX^9#@S)_N2ki}#m29S#gSo&BA)7<ei}>8;GjepZ=qJ&1^X$;S^!yRq31
zuURz^)m4b!!~7LXrKR`pU)W%eE$Z6^uppcX<;;p93*YwZMQB5Ae<-oq)Ncw_bu%Xy
zoquy4OjixwKKBgSj7D~y{~VE=a%3-k%__udI<C`+5~r&+T4Q;hzkK%hTdZQR_tE_J
z+4pNbr^o$*AFA%gzt+(>c>0z{kdHVEV~TkXibxxOlmBV638hB@!zWybUSB$%VF{W8
z{5KeTIqILeG10IjkX?Lnn)*aAi%CqoG<1|t6Y9uTG*%DOf##V;U;mrRrAWO=ss<CA
zY9Pk9Zff)3r$(#H40{J4gQJWYpW|XeHNfhzF?Gi<9#aIj<<%?3%YIw=Cc{UcqJ)_N
zidrHgpA|VNwW-bBQ86*cdqM)(^1Cc-6V{I=g4{4p9PHY+j7!v|YuPk(Z(ai^Ds;K0
z4zFp#g)~!BWp((I8Q7lo!-mgL8LaZH*Os2`KH)Wf3L6SAHj($QcX5a9Ylt{!r!{gk
z0#WK;XVi3IkqR*%_>ZiRuH`MwAmS|7jZzJG+($jCYw!1M+JKFl=8<G<Wu>#EEk)1g
zG|!7^l*atMD;B*VN#$<!5G{!Q=Gt1bNw<&Fppu~$S?=(NS^qwJ9VHx<Y{?nJKwZYd
zuF04jO2#-Y4_VLMomE(3wD<yxAR~Gd44|*gk@Jw>1vf^fqzB4cx)tnTK33**@2&3Z
zYohHsK2xO1)~SdWV)@wpg`19`*WxYI+^2s~FIMTd2U)~L%M~iVQ*f0o;8mq1KdXGD
zM~rpUp)e=?N5$mXbvP7nD#YqVV%aV8#*u=0ti`?Gzlz~61E(=S(5)#EL*{ZcI&PNW
z5{BXlq<(dMo8hNV+-@(|YdTsH_g<Vmaduui2A|yLRaH?_EW(aDrDuadj7=_X89s;&
zZ<D!_Y>YwHd~m=eV#-kcCht)DY$$-fg*z<<P)f}*05ZL!6lVFG(PDi{3IrN%nwe?$
zK5~)uy=DPI+!VeZy7ZO8X+TvQ%Nk4tHQ2Uyb@4Ko@4i6l3u@y*<w#t|GaIyg;jJP{
zdSQJA_@`7DHY!9{|L(l2V*l}s*VC~&*PK`EY_xDYkmIBo`r#etUD#jXNAuqq=ak*o
z^Y(}D&T|U7+ze)vEL_YGCODX`imK9GqHw<#E=D&czQj_jqlwOxbmB#$jG6P-*M~uA
zfLHmSabLriIMa+*!J$;TP^1CvkY}LmW#WjGr$G&N@o4OEJd8(8iJ~Q3>@i<_KlJQ5
zLKJc(G4A3L^NTe9l+hZdgAhHxp#NIKMd#+`1fH}|jp7M<P|y>=BQ!BI>piti`0MvM
zJDLok8BLoIT&d3u_y9OT8!rAatfg9X^_uWcoSH)xlv<HM8w%8{FI@#F?mBpL8B0W8
zvS5uX-als4(%fKFJ8^dZr$oy&XbI9z{PQohkQuPq1Sln2XU8_lRDgXHK}OHgx|iKP
zO2KUXpE|x)(V{3@07ra>FH>rUypJ)BMDZC8PDwgs3j@n%ShjA=j}<O0st0-NBo_<f
zzjVAVmW?eW`LA(uQh*cOQ5sz1B!9an*s%Hen?V+y5y`p!6*zt)SksVeV<=xqU3EN{
z@lGT2Rc*8BeZ!k*9N8Q?#oYbI5W*Mga~;seHWd)xRXbhCQ)HG_?c?zx#M*$+DyQ)z
z(DCr^-{%&apEE%84%UyvGCrqm2(pUe!<b1=Ym*hNF~pF_m?37;3^=V?gvx9fyH~rQ
zL`!ywZ0s4={t>-WN9Y@9^b_u9fflH~P{?c1+{%L1QFvSulp#4?sR>^O5k4e(@5E~=
z$<OY1A`7Vvhu;>WPVwTQ<h|lFvUi^if=cD+B1)MxN1;*xC9U*X!q`_(c|YvuibBM@
zm#R222%6HG`O)ETb4?8}Y#SS7!||ds+Ww;?)=+4-r$0a2+x^R7myb4rs)bHnfDyWS
z*bBPBVyf?GRan4tuaabm%>VCGw8yuFd2-d#*V~A{r6#X&-65MN|7JE8TC*k*3oo}H
z&5<NHUQP!Y6iXOGH%uj2jiG2)+I}9E>{RDOzOo6~G4j8ELm9unx3puM0?xO#)=@7A
zodGY~*Hd*7dRMvo;g4=kw@ku^Gc0qwTP@$HxK_#l?$Jm((fkg-V_}eVtI)spFt@~#
z#91$KHJ!J_@89_;vnH<QDVMu|uKnH;Vy+2OsUOH<@f6?iyT;os=x}xarsVeT$C8_b
zXXVZ2{BQaz@Js8pPTs-Tz45n1@XGzil;ntC<9_usCKYcc@&js_<illS)YEas42*np
zjVdVi*3U@fNiwRWezR~S6AHZcw(Z_Ie(~a)e9tky`Tl*1xcNA^+9lwyAqU~+mckll
z4uMGtgt0iLYxIGLUf+5;C?e_rmcl-JRcZo+N~f5ILNsHAu3}4V1r!-mtllxJH!^G4
zP1mltI%DU}<Fevt>)!}$<;M?I_DA;eaIRhBU~#)HvuY`cl^!IyQ3WQ;W+dmt3^#p^
zp>nu;o7cJTHtNxm2tQ4^nf-qca8M43$xR>M2Eda7h^+vBj&*6Pm4sxv2@>hl%*!Ct
zDtzVqNPpXeFpXU`VOVdXdr}zb^8Hd~5tfDAjc(5Q`pvkzAh^W<kAydE+seFvd#k*y
ze+?0;(*1fkKjlU*W@}ci9()+CEPrrV+w@<T<utetv9c;I&B!8jM;%&`jk3Rn|MJJ(
zrZ^5NSMOlxXC)l>UM24?jAEK&H)2dprF5vdz06Ard{|Aj8}z5sn%jfp6skM(($4DC
zIhZ<O-j#>YM!?vCoy}`~N)reKsca(Bfg@$ex27cmoaxMIz1dAkQba(xmr%GEqfBjm
zex_2X-=hRH16iqD$_r9k(zI^j1Ls>!4vSc500Ua25gQ>BoTM>|f>Y;ut+D5Ob}9{q
zsSK;m?>^?_-0i$b)##HLsl{aN_Cu^-jB!HozL%Y(l^93XH!G6u8+hvdxSOTG|JX#!
zH5qb3zvpYDm%It%L5&v9DYjYT<pfK3)Zdoq>htRuy!~^7f)Jr>AUoG~Py(Pv-)(gs
z2miUPxR%=yq8hc3J)G;xbM|jCt)lw6>&ItU5JULZqZu_NRV9VRMsmpI=_2;W!1RHC
z0~nLi&rc=(Kr51fH2zW#&<n9;xo&-%g$gE`qwh@P{oEI8?kSr#FKXeL|0!vebFJBb
z^9=O}nV_)@ifYy#ythtm*8WS=c*vK)hE`B6xQWr9pazlCZy$-CN-~VhKfW*;6As5R
z*V?P2@wHL9Yw;_h4P74T+^KR^p^;}s?Mi!n;Ll6<nVFM{{KMtzb4L$676MmJ_UDi?
z;&EXyS`2*NjS6}C!j*%S6)to5muiKt$e@hs0FQTjKb5!7zO~MYjg$8UaM1D(t??ci
zu~=(2y3IOGTbb9!8&1+Te&+-4Hczns2If9hf|Vo8EE1q~wZor6{xoC%O;NBth1knZ
zwWH75=es325Hnw{pnA7r037E!#@;ZTs}QJGGSv@VS+%64el1G`W|&klRXX4}=rHW}
zKVGQ`<CSR^$FfQHi=`??%9R7JARU`~)ukV^;S6c4?GZ!VLaFqXb+rXVFg8=ROWKX@
z`^xLzJgGO)z^3ullPkw-H?8u3(PLC-CT(L8qdcHKbOlG6ZkBi;&8eO8a_Yo)Rwe$8
zCt`9_eaoa)y&*Cs2eO(=fdB~Sf5K)bZRO8J9Na8AtlJHLkQ3qT^{dr2%xX>FZBST%
zuiNG>Qq)Ud1ApQ;`K2gu7i;kP?!i)7wW4qA`p<VV=BH93cqZ4iI-f`~;)?6oNAE@+
zirhS*MC|r8H0324^WQ2GFVKLuMAD`Vph*3_NsdU5DdcjtX7J9B&NL_Gly8c`PsCmE
zM*P&=rHg8+`uW36;PaTOxDU{BPz|Wm<k#6cluxjvqnf4hS>(0^76Rk4%P98}J}HM;
zw$ie3(jjb&^u<b&UsdDdrvxDO{EV@}1_gO>XeLLilJVmiMFIbjTM6QEn%OyB+vbKh
zOJCcF<ER(?h_1T%C>)d#%o|P@DM0huGT#Ti>>Jzj6mP1Zvzt2kb_<u_MO~I2o5BK{
zK8-HU`nHP)3EaT;C5zs}%vGfC0Q115hmU}{6&g}Y3L@nh6T@Hq&Editz?8H!HX9CQ
zWo0JBQ-+s^`P>!5M9_7?kqJo&bD~HBk0$+l#;I`nB!6dKWmP4g^1w}eThB4);_;d1
z<-roO>f5eLm4ccdZ~aV=<$mR|hR|`H>!LdL<?aHMI~t2qp0hEJ+-;uJecKuErZedG
z&!7F#_HuKGbYSCSTcGjYt59DmNJJ{@?!Jv=zU0Zk4DByr8mU~X_L!BiEVXu9yNEw6
zGc#;)p|)mhpMNDxKU$pWZ0~pr#3VqYUfhZ39fX|>Mh^z49etljE2VL5x;Nhw;uqvO
zXr}F;s-}cpZds7Q?78j{Je(ZcTRZ0i)>ykVe-FD}e(!kdpE(k*H~;PJzt62{Kj_uo
z(G$Zo$7dCrTkmmoDC&Mj0Muh|@D~=6IBR>8CfG4n21O&8jZmX3I{mHFtVhrFE~+m^
z)J{i&wtFrU8ZHMov6mxbbvPdH{h?WbC56}g#JAR1zdot0ZCfNGt)O-hgfYe#qDiSu
zZi|p-S%ucTlUS+N{`Dk3DB#Ypwbrtpw&sthYBk*a!y5kfLf7z#Vv{)~O)~r=ND{d^
zEknKQu;vU1Y~`=%D}Ooo?Xh6$gP$*muVR+38ht3+;j(fL=nc)%ure?D#KwqGy#{kM
z4<M#vjjEd<(@Oz`8|L#zH$!DP($&zop_YHwRnrcI9zT?rTfNh~&@l2=^-yQ!NJ6<?
zzvA1ME}!_|5D)Tf_xA3@&X}hGBJSeMA+aRxvtRspK@o_9rI@!c)9)s$Rqc6pi_Z0`
zV)Y47G>iy`w(UU6Ngb2H?FG7RztruKnK56n#-QSB2(k1Wbd;ZM!mHFqxpD{;VG=1Z
zu7~R)YQExNv9BvY$w4EBAsKmsVVvp9FD069qCu$<R2gwYS3)OexeAEv_xBd2{S;#Y
z>UjoL%gcZn<f}kFJ%qq(J7d>>3(;UxLAY-u+Ucbp6I&!H!)`juF{S8N$!2~tLLCbL
zN5q7FL^`r2LZ9N<9X-^fcBPoY0-K>EMKeSMW#g65*5(84hHSaEirBVpSzzV_!VaNX
z>s^QFzmE52-b7G6X|-NCROwNS)wPjf(wZA35D4bA6qmh-dfldy@U5s`rGVgIrsr}w
zm}!HV!*d_hyguQV|F2H|*u>%_?1Gy(L=a@Kkvv)yyzfTw9joQI%iq60=xBT|@MwMc
z?@)YaP~Wewk?2g#{EBql_b%5<B7EZ<>ZmoG;_RWC+KsKvn>UJ7MU*?6ef*@`o2>2L
zNJ-(u!kNuGJH`dG6^i?Tfz}dhBNTIY)}M>t@<jOLjQ`@8fxSbmG%62|b%WHor-Jhc
zWSvr3$9u6?)4&+Sr#F_`0x{o{KCRwai`QPe|8vq~!fpIc+Faje1PBt9ZIn(1nusLO
zeIZ%_e9bX-GyH1Sn=Rt$UelGzeQ(5YbiSPyv!X$#yN-M`#+G*}-Bb9kjDE2N!;-=Y
z(7ZOJotsu)X2!2Rd(0pEziMtJX5hYSjoj6qf8;xwC7zcX^4NpjqL$xVyqAACk;O5J
z5oPI`94%nImZ^}fb8t<~1knZmjyGY7AI|!jF`SwDXWea9AOk-%(t1uRaTv81&Z<F$
zwquQa4Fm$Sb5LK1PdFY6GA`Vfa)eUD0nf;RYTUM3%cFJs6#rYQ*6gU^5EGA+c~+In
z-(rX@j>WRBdEPrR@(PDh2_UvwN=_Q-sF-*gKn=l*mzYp<v-7=q)3Jq~{eGv`(E=IS
zN&<_1_@EBm*aDOD&t=_YuKoF2-nbC)8!B{N0aw~A%tD2VdbyLu17O?)ro>)bLooo<
zVNdbeP`p_jHkPgU!-rXTzUDJW=xC&91Q!6;+HZ?whYV28V;6H@lZ^Rvq$#86j!sK!
zAeatqW||qolR@U8jbY+QaD%`Ty_iHX>DtM_yGo4iEyDJ-24zwhTSTm8Qq22QyEIKu
zf41&b8#idgXKN=f<iKdb<0pkjtGI1?yq|Gq`vsOs5Gv%v&+3L*x5yBMgL8g@Ghkxy
z*zD(WbBQpC=nPnt%2WKyIC45`POC7h&KpMl$aGZwXwiQ~xlODr+9sLhCayg9;aeTo
z-0+962&ufPVaqFw>>{@_yc9ck7GzjpQ^>~LPa2|e#%^N(kZB0aTgg(^5BxH&lyzsx
z5}V->L~v}q%3lU#Pu2M^Eueh!{5{Mu@N0(<9^#|IqWMwp(G6<pzA5Vw&5w+&J5Y&J
z7Gq}qy9UZpA454gIr&7&9a`dHsl06mhayXbNyQcw*&e@{{ih^3L*s^ox%=L9TwX{X
z9CQ+@ut&XJQ$KwOEx*gG#3VyMJeaOeSHgN*qVyY=3g08yUH&tUjbRQiOTW4T8X6oY
z48>Etf0$nb$M4=?W(=KY#Vj~uM2L<uPot2P`o-}6ARDm7H0Sx>g@VUYT&69Hl)FAa
zJE5xK@nX<snkUU-!^vxXr1djv&hl$1S4Y3wg9MNg7RB3{RSD$Z`00hJho5d?DgED-
z-21yf+*`jaJxe8;_Ph(&&f@LaKdF3l`GqvK`AFTu@kS@=XOlm4pf`<*O`t`)*e}I|
z=@q3q?~qUl41De%Xg_Zz6#j&j%?4>uA(An&yK6u5xbB3%$qT7m@WpYSFgZioja_GF
zfzmj#HWNtDb7EctbK~V^&q?p&OYGlMo2liDhuG0CcySRYiMx>Kp$B`V+nk2%b91Lp
zKva&SR#3ST0AI6h%V%|fI^dglXH89=c(#SQ>T$P70Gq~>Mmivf-V>*ZrUJy6?pbes
z_@$aEXAR_{0;Ip;VB!DLE`~rj{(B^!xFnn!$+RT(00BTynUT31Tj<R7un>bg2Dk~N
z+ZXNT;a=4@BJSb>jCG&ma`46rNqNcnlI^Kkb%O|+7H>vb5s(l}2QPJa)kom2&Q*th
zJS?&4zQ*p19|cTYw4v$dY(m1HiD6`}A7j4yBrqsc&?NMv$->Z)yPOOc>~{g=;9)$v
z(DICuQHjq8Wp+&<2P*)Jid4#?1Z~%C4qL}F@ke}9l1kJ7)HMsB>KuP~@npPLIH-ch
zCc?P!=$M#nxOal!nmj-%u@t6LigGfpO6LxKGYoi7hppXIdgPCAc96z}gae)GjX&B+
zkH&oUv#XY6s03#We`Bu@^qUV=Up!RwyctzozX`q_3geq9`-MGKRnxQA5vDj}2+;l#
z2p9<5F<UrOk>UH(B{}lh`T@+^jdS-!;!W?Xf}Fd?5fMbMQN1!`ZY<kGMX{i>bzhC>
zNNQISe+y;=<l_jbM1{l<l`z-&nEIri&f0_J#93z7lizNi=6_zH5h)@C@oEB-!AlNu
zwVPJPWae9{*n-DmdHx-iwhC)eUS_X1J3MlQm6Q7V`Y3&NQq@e+<pA@i)sZI37MB17
zHTu(_d)x2^Z?yf)J|~5-GuwARDe2RvKNL3CpWo>{=bH_mcS5Z1kr&*|@=-i05g(xg
z-%S}~;uNkn>#)aWOpNlEuh;FCCPBH}62vjaoJhqwtVF2d)zlcjM7TN^<^d+Yn>Xo3
zlIOQ~=O6HM{BV{P+g`Zjw|6=`T7M^JzHlT?kSzY432dNF2i1B`_dP={*M0fLc6WC>
za<&$OkLj_y6kOq1K~eYVh+`^C)oelVj{L&Hy{|j*GMrl0zL1iH`pwLogyzwqOdU}%
zJ4XY1Y@HX;agVRU@{5F#%Dt}VN8h1ul<q0U4oy#%c*GUp@y$z1lz+ME%!#Y7bFQxs
z2oR=J!ZtQ@@}4}wtELT?^KDy6JqLatzj3Q?A_S0`V;2`x(rPcx);E@-4eS=*9a9UG
zQ0t)hOaBuI-ur%jxOo1Z&9m#Jqvs$5Bn+CsR>`?;``NkSO&#9CkG0`OC<c4V#<ru~
z&876l<l{SUC6t}>9qj5yUq^pj>Hm7h{;7u+pV1l#Gq<bv+p3cVIQV!@1zJBce4b-f
zUB~ujwRpr@!YQH3fkD@fkyY0m9plAjg3men^khBn+Cm+!d{xVi&>Xct)0+9<tqLy!
zTRQvpI$rkLU&rIS*dg}12dy`aR2vL>U-)2-`A8gq<eB>v)A})vKf)aanHVZPWckr#
z_@?lI_UJzoCN87wtD%vK<q-XT7_mxLBod~*eV;NVVKI%NlE^7N22~+-H@ySw(xF6y
z0p&Rq$_(Kwoo-^yFJ(%8OEO><Q`y75G*4ms@p;Dfs~Lh~oybllkdtl$;?!}kOh8Vh
zEyhZ?NBoIP-_NsWu8ZM5EGf9+r^o4i691raIXoFb$~)mg0-#$_k0P`ch(<`isF9rt
z=W?p;m%{s=3-sqWGgo26kMc0>s71LOtzzyvPc$9SOl&`dh7fKAF{lz-yH8l5b1$Aj
zuTe?NK}@JiflLLUx5lHg*$QEX^ut{RpcxbmhB(vy`?U+(qih8vt?l{<G!m+EiR?Gj
z<1L?*f)Ya}%*vT5fswrLC}*$7ck}zm<yl(mJu99b`o@NQ({`@3Y*y*o^qnKmfpiz9
zNPJvJt0UUI7pAlcb`0_2#-5TmNV6#0eV(6R56KFYAw>?m<NUr?`(I@HfWS+6&Oc6R
zKgXuk{5OgFZ41v|dIofgxg#@5jApz+hAx_Bo%E8z61TpoQEEj>wTQdj^=yeu$<ruM
zR_nxGF7ypD(?GojX}>)`k1XKy_w4rFoc+!DKKDf%xRgGM;8ELwwWl3+0m6w2beRm?
zv{bT|&Mk5=H)u}1f{$CXf)-zH(=Qep)^1uf;O2k+IOU~)uP=aM;+q|Xq>4_P$r-Y_
zGUfo44C>ns{vm@s6hHk{be?0<0bvxTMVPXFA-M)QdV21i(;se&lPIt7VSqp3wzd?Y
z|D!cOmhx~`GGm$5_75tZ=455~Lnks(5&}=r?uc~79=mA_4sC4h#>9x!)*H3X)BP|S
z>66cDjN<T{v2eu865qsF@2b3tRrjU4OP*Ikh_+_>3@t)Rjrq$!EHvur-*sY$L7q5L
zTJP>}=j@Or3xFIUN&c}#`Ljt$x=+unmrd-x@1T2>u$nr3|GWVoE6nrtG_0cdOB)5K
z-;t4|$KOXJ)*mN4pv}BK;B$os12{R$850Mxq%GIDUwB+Y2m}()iseUt|17ey)pg2b
z$NH4zNWO){4S4el=xl;YTgkKfTp!W}Sg3TJ{&)_*10;((b$h>5Oz`g$9Sa)HD)4=~
zpqe2{#OD93&@WIMPrCA$U}NJ}dN>o{)KGzM)i$=+S4f153Dmz){QkS6UTmZU;xlgo
zuWO|qg<)zmpyeMo&425kjY%fOETf`)NX#!oX4^OTg?1J^RRR$#>fUyDgZ|m7gpM%W
zUL2_KGk3!S<+{A#cW>C?^qv;z;$VF3S7N<yVhcX!I*O;`i#8zNw!H18sox7!R{Ee~
zW7At<6H<X@+!+!NDvefngziu|TB-7`uLl2{v{(6-)3uNU6u3KyZ*uMU^WktrN+d!$
z;^Obbc}GoyUCtTF^X2NvqYaDN4)TP>hg^``14fNv4YN@fcPInk229v2Y~~>cK?)c6
z87tS@_tj2})2HN_S76NyVG<8J5(La5{g4U6gNarwFzHAScKoK$*zCryt>qu3(ej=>
zmJbOv-XraVC%hlV&8spFYMU32?|Yv8;Z$Web<g|t_g^2yslL4Nua8;vXq??O>}_+;
z{+}6?;(;h{Cvbvw>hIA}iuYvmoq#a1CzjwBYhbSKKxNTZZvq<tm*0()1EW=x?-l72
zBv;QVrqyYR$^DJ&+zqXOv*q5j!K9N&v#Ayeh8|zrv`{$q%gM>s%R9K+TD<n;d%L^U
zp9SYtl@CU4%Rro;pX~8B&vWuS8<iJ_KZ+=b%HZRXmEN%{Bz^l**^g(z^%?awSXZ{k
zWgih%B@88ySTZ*im5b?rQ`Vv_u}?pHNUjaK_>#Hg5G&u!;Bh6R=1Ufi`k80Q`|2FJ
z=`n9>5HwUqR*Hh9f0K(dmJ?t)uZa2AzgO_&ia&VAl)3BJaHYywe5oAFC1ZhHIe#+*
z@HX$(_&Qy@SiV0LrGVX$j<Ml1#x{2~lSN7gfpEc3vNROAC&~uOjuzW=`m1U~48H%1
zwJ*i_X#a=_36dC@xFk00$o$x{Ed1fY;kE_g=Vp0gD3IM{%<!rtUvkyYl#kJv=E-*S
zx|JISml)dw@?lt7IY%I>#WwY|P^%@7l4L$h7K9$_6ejOqjV=HHeyxXvYnV0v>NtA_
z7mP4sr!^jpux1euKu=p`nOMUS&K>olcE9tl0NWh~R|{DC^Vx2r%=r!Yos5D$x9ES{
z7iDQgSyEg-IPUC79o`HRu{vKa^k|TErX!+y`}F#U)${Dux-MScdx3A;+A3Q|lneKL
zTFhgTd&_wwmKU8YvUsu2{Z$mRMB-53=!E$sm^m^5aYtAPh+c;4tfeDQx}B5z-Kafo
zEbm%*i;F~zA9jhy5io6pn(o6_tZ_b$Ii66M7NY&r?LmD%yOG^osn+Ht!%xxLmx<;U
zwZBA~8Gbq_i8CtuW^4zIxeNeaQJu{&W53M0HFV!?4w|NymB%{Veu`{0e_Ed2XlUaW
z|5?#FL8Oi<@nsJ<DpHfmkVi-DK)JKC5wo^-hkL?`^InqbL5dxo&9yw=&8gM$=a-f-
zN@C=e&iIJ>1a}p}cUJA<FA2}7MraM1N6y7QYt1dXweYy$zei$*Jh#k#(-3{hUbybZ
zW4x-t!66fVAz!gGA_30753Wh79{idm;m_y0!f#Gzp4Sw9RqlD$X2YAk@?+*<$lne5
zg+gSS6^yEUU=~J)j++Iqrc-lW&;KmneSzmaKgvofx@F~Td_;MJ^k1>wD=_#KD4ip>
z-)~US+*DVRRhh(Waeg*Y8jztBC}Bg?|EQ8!+N$GLnZbY!C1<jvi}(C}*mLyX(Qfp%
z>QS27#c)>0<%2nkwiwzJ+=ty=nFN<s6TbKTu|8WleSh`k005Ti8dsh{19fKuJugSR
zcJqxB+yu0|<QcTyIO}@3AX<2niccVB-JL4a6k&ZO6^!06vGz9DVwrgc%&~I0XSU6%
zdcrR1)fbChcGmHCX9~DL5_+ELm0%BS3Z_?gydfPLW?1%)M;CdVw#1`6UXvI_3A)gZ
zdj9*b1#YHD)Alef|M=`&^*~^J6_j`HB}euwxZoWQW;e}M8f7mfckOj0lj#6RyI(0N
zM0Ig~MvkM?LnxF2O&m%mq=)L&jp=auY|I7mq6M(OaB@{3wuB{!H8c{Zo|<1Rz3bd(
z-%576C#l{-WPFCpXMjsp=#FCMPQ2>!gri4CRrbz(rDORk((9VX9Jpg4#y40P*T8Q@
zL<HhsF}m}F9NJE=kQeXt<;EPXZfIySP=W&>{J0oPWI2PS+etTD%7+}L4co2|{C3a-
zTWltKi7{iwhjen7S6)`M29pTjqrJgF#W6+?I^aBB?0#q!rdeJ$l$3V85}flfL)&pr
zD$zV|3$fWj@?EeMfy8}KtZb34*X=)l^UTpyX)=fzz&Md11n36|H&+7(!LgJTz>Ph{
zAyCiRj+<RJv)1MMkMp#pO3815HcQLP*u8f(6(j?TKn+fSCG-jH2(#Mrv+w6c3*EBk
zNtfr6zrTHb`FvZ;!h<&dmYrAu)Yj^HAEYA>vqA9VtM(r#Ygfg$DN5Y$L8PXxc`*3N
z?B1{Z5&UD=n{aQcBV8ShxH|D#+HoU0hmFPaal0!e->(M$l<mZE&TLNEh9Ca37z;6*
zw-jzuQagMpUq|w{#xy?EvaUayOgm1(p6mu4{wb0OZl5AK86e-c7R=9QJwD$fUGoe%
zX!A38M0{$>URBdJOV3-lSj2lDbFmkMrNoMyRY(*xM+yR|^aSFdLeAsX&9a!wty(O>
z5;0@8eYQN9A7)NkpS6;obG?JX+R5&d-RIt-RMj3!a8i8NhrHc=(((t#Ahhu&P@n{y
zK@$e1w}#PERmXQgD9=f=<b~TM_FnML--d3A0D4-mDEW)}=E}@2#04jT4^2o(Z+dZ8
z6c8UHN6F+6ND$F(B=QLun?b}P)O9vmRp^*iV`LNGyJDa|GQn>Md{`6)^Z*Vpf{0O<
zBx`3w1VlDlbPdZ>6s2TBA##kZf*NR`y4mXLUn19cp+T?n$K%9hOX(#=D<1pb<t8Un
z@4xcw3J9`HAYK7QG$Md#LumE@_{zJe5&7d@I#w1Id%f8B4<eDtNMA9kCmcY|E6@aZ
zosptrga(cFHCI{$){FHEqq^6v!18hHn_5q(jMT*dd{vevA?Be#s9j(f(+YZy>;+Ch
z&cv>o>+7)eL+;)vF1(9on3ajs#NWQJ0l|3bt%yNhL3jpx!Aln5t;DYa5oHO6vH7B!
z46G%D**i%wRQDiEP%yf~BfFIi@Aum8rVlmA1Q@0SmnLI{cKDBNohZ}r6O(%dQm+X*
zHcK^e3eXgA7g=t)I*6Atxax%rq;kY--$2Jgi67w%2yhO-IJ_#I4i7UKrz`N{%kE3b
z2Dd`Ym{8ZUiufTkw6K%2rgD(n#clcxq@7;~4>a`+wGl!)5ncwr+MK+6WA}_q1z>5U
zqvPtt);}|#+cc^MEe)XTarVHVYS|$-J!ToE?x(r>SJ@`)wrgj_I_*~xnEk1)^WDkP
z_?40e?^?TVwj4<bRO!06BCV77Re7~M<A?S)8&seF)j)oV{`Yrda+~A8Y3`eOhjmnO
zu7~_2!RGm5LkYE+waM|1=)TLpYUls<ZxS%yvFqz|@w+Qz@}J&b4bz8n8+ESMYITL)
z`J~z3c#882`HEyDU{p~poWNS1Q1FLm@7!w5Vrk-;<E7K2;Zl>0gtC*681&>A^~&Xw
zgwpRR@8!UbaM~NA&GWUCd;3dAj=AF|3b^!BVKm^NO(RuW{HxOO1*MO?KRB{n<5Z!I
z{qTmTUf(&b=53c+*U510+@zZi*7(Ul>1+Ajw8>c{$EFRv{NU8z=EIPQJD01Xo@TS=
zk(1nmurW?!_8Z!&&o9F(TDRxBCYgt#?LQk1m*9u}rSl9bR(xl<F`cBxJca&thd;tQ
zE1h{6B=F_1gHMhlF>tr=svMy!BkhLN*q9Gud~<U)k11T^%d@1Oz{`ts&$9vf?xS)^
z5vT3J`px)~&&2+0Y0hU?+Hp%UBH=yKCoqdz=|vdvb#lft#1Mqiy9>gDt)xE_zp8Y@
zmH@w+_OHtQ*3HT&KimDlj(CbOL#QwRV3nW`wS}ZPFU!b}_pW~8|MVk1HfKi+`^fEj
z$?TKi<A={CztyHov6lNPc6@T7ZIW1AVd6`Qn8|l5eV&H>klrG9e3Zq1Y<<nXfP+C@
z$0R*2RzMr{?R_`AY~T9(!m|3Ul$;&eI*mKF0k0h4>oYO8B-p~3<kwEQqk0vuF=|^<
z>G0N+@I<T1FFKzUGpnAq2XnYhf5To<kSWUH-<!7b8yiX>?+%X*MlwE$&E}KQ>=gK8
zBASjIUm{{_ms!7gPSoH=`#jdJuvkg<=Naq}$<PFF3@MKwkps#3jaqHP_iB#~6Xa-U
z#2BYm5$h)5F$%C|xYHv~E8^8iMZ4A4ifo#!81gws-gs|Nk&XPj!wL};7RN|t_yPWK
za3BXf^}?xsT&a8Ay*(YO{MvjLEW_02t27q*6UV`7Y>#Jl^Hsf!HF2v<WvS^s9daid
zR?yQ=uf<0mHl_)&j5#%ftDhy7a+*yJiA9W>ls7j|uA$d!1I^oP+B_EvTlW9`nCvT+
z32i=ocmuEaHe|Q8d2;syb{#*DAHQ3hQ7Wa=!~MooXC52KC=GFm*g(O|e>b^PJjez%
zRu-od0ksRvrOlN0qXd|scSSnZROz|*Z+7LUpMUfn|I@qouj=Q4Z@}MI!GA-5uN1uH
zB{+6_K?E6lH$tu<V<aP(Ywu?XFTL=sj$-EOO(KcyQICJAYk-pl)}z0a?P%pZPqIh4
zs3<+~Yc9KdJ%&^0QM<QxdHdB}^Y)De(U<`}FW-<KgDYumk1vmG8r}`uo|3%#KVD|=
z8T^In*ZIXlQZB*Pw$49*U{cl)yyAvE!k$fpM?bvtyXBeooc`?7DLDhXR)fiwH0!6k
z>q9A;`$rq|^vv5vSle_77=5i>@yJ?}<K+hS?>6+y%yZXrD)#vIv*rR<)~viPC$`W3
z%>3huo9*dP>uNFpEr$H7X#F#g^=Q~<xRpE;x4i1jLK1<E)@6_oH)7QF8L2`e88S4#
zSXunXF5m)JFOR$mh*78|)`gn_Ebn|58JXat<^yB06(366#tN64)i0~0y=l<$<rtzf
zTebi&sAra5X4IkUPd&s=g|CD@puh5=Kd1hRb*;)nDduy{j46>2A$!6<(VoDcF#^eV
zSnI_ljV=Ub*1SXnQYF!Ln6cjPPBmYf7ZR=H4pZl#(`%_J-VfveFFlYS>FM7--+5(A
zDsLhV3l-eY=Es_zvg9|z9wOni=`?UrOr0JK(<A_==}#A+wsu=87iO)j0@cWt6;?K*
z5!KbdM#coy&8pt<qH*mmC``31ke0=s3ZKqUkzLbKrXMoqc_pls10KW5pkVQa0X&#~
zM}r8jh761Gb&xyMN263}m;w5Ac*j-vRlWd6QI<#8+d-+(fK7M=1A{^$dV>)#uH)R#
z45({_4ZPd7zF9=gr-YV8A<|zdmL{`aWh((T{OfBqV`=iCWqDN>1aJEDlp@FRi^Ma)
z^R>#e_hnl2()0aFK(3A%zfXz_+^+yqLO&p<t=t%wK#U@L{N@JNV)XSl6t(xz;1hdY
zy*6x`-P&+hxZ8-M<n!mjXX^`HlfQr8H1Lrf;7fLzvw<1mGHP$C(Ye(<OIP4-75!XP
z@Bnd*_si6knOvBZC|_)T1U(+yV)}t6mPWyh=;B5-vO&VnY;l*UsHzkizQxQH^FAaQ
za>idA)lJu%bp)UM9je5hPj_F2^KzEI*iFD%1msn?9{fbT@LC8@3<ajqD$mT%RaaW5
zEG1z>*8X%c^Qr_tQ&r0)te&*a)Jkr9V&pBD>sgmRc0Vf-q<W=!ui0k0CVa!jW}4Z<
zBcnOOSoEB-2R&(ULu=jQHmd~ZeCPJU>A$3qz4`5s%Vmd0Z#OP3#GMVSILPfC&|D|1
zH3za}Z+YS=>kY4e3tdHGoOc+)$^4%Lgi6+AHNX259q(QCsGW`8S$HnN`FG+Z$g1dU
z`*N^n`@xaR`MtU|LDvcv7JM~LzDc0l^rJ2jc#~~+Wvkd4GwBAURcn0sXM6kJi$JC-
zLO{cFwIlvxul1~J#N2Ramzku6)1|Z}?0x37hHh`S%i$J0p5U~1e&(KG0!p+NwV3^E
zRV^aQ*y`_Ojb!AnMp^pS4nCy7Z$r+PkAALip_*EY#=o^6QX=`fcXkg$j++CvwO0>$
z|65f3+SAeH=jgV%epFZBUhE$|I8w6k)Onht6uXucuy8z3u-nXLbn&F&#V)!HtGa(6
z$`n6i-iMN1_t?;K;P6H;g-H)gZD<5xQ0~2A$dz=>$hc%e!xml91tp$Ot!F1g?f7rs
z9iru0U2X2~`C8<RrD$4gqjqi!d3|R~C7+Z8`FYCwPNwHbs2pAT>%C?^Bczqfy?R(l
zHB0=|<z<Vi_2n}%WuO}A_RAalVYDAUD_B8*dd_#hyzINf>3n;qCa3IB?X0+2+2X*%
z?%rFalip}Eb3d!QNaKC2Mdkt;@u8+K?>Clx^FQ%E%)K)w)7&%7!{csHiLsAQ7K~yM
z1Wq>bj!@&EF@C@wU^``*j|IG!t9Q1`1PQr?Vx9a$H;XRzN-L8>{!Tu|{$BuO2%Gni
zaR=9LJ{`u#KYfRm3(aJeY1-BnrBX1R)XL~p->>=+6u3$`%}qJJ_~O*nWQ(GjI;*FB
zFM5!sy;>GiVH8Zm=uO#ZTUqTuW7C#HKb}54`q5wi@cZBY(Ytp)JU^c^sY$zMbb(gN
z+BTD>ET+?`+P@KRzl?_coxi>J_{o3SqoebSi-)t>$sc_ADs%fc-+TYjlZ0Ar&Ey|_
z;g#vGdHIFC=dxuVt&i46>mSZ~K89X60k99fQYyyaoC}`Ia&qPBwZ+9<?GRC{^(h^G
z@WIn}@90l{;@4k!@wJB!4i~G7$#mAvrXl#T@BKKWF#0eGn1piSQ+G^{4$ptxI}fiP
zY`^xgmv7#@(e63sIE*XrhfD3JC{c$XtCcT`qL|&hJ-z<$@ssnD<8rq1*-w7Dp0?ll
z$&Xf)_HN#I>635z<?2dR9NxS8`0m|qx$0t%j7H(S)TlLxo;2;)FLtgJuYT;t>mR@R
z`p0iKZQT!x<O6Ub4%3~zcCvq0mJJv)jL8}c46Ut-qJH@B$=AN}jX(OMKYIV}2W44m
z?W%gRd*y|qD5gzqTwzL8H$`1fH8ccAU;JmEoNnEZ!{L{|bnxH&*MIn<pZe0C|9AiG
zfAZ?bo33xt@Y#R)U;L}{WAk7Am%sTt|IV+Cak0}rhql;{)<^53^$&k)Js(3KVz^WT
zCjb^9X646lb$|cC{Re(E6t)3CGGrXYM8CN6t-J64<l4c@w{E?FP@bL51x(v+HTBf{
z!H2<DvvKT&5~sjY0!;5e4G$0Rmet*@?P_<s+27lqPOBJ_w#XDil%}2t7I)r#NW#XL
zqAIRkfAQmQexj}?N9TXX!x&$E<+a-%`?w@&ti!-B-MG0^R;MSYi`8<uT#SA-P%4x)
zihY>Rrp2|r>07VAGTSyNpmphN>{o-(0AQT0rjxy<nvEf`Pp8LccYgWq-MjDKeec72
z_wJ8_2gIG7ePdl!HO3aMEUT()oGt3IXliAxL_EJ(0Qx`t-@ku;_s+k5^62jW@~{5g
zFaG}RfBp6Ee*W`&w_lk3%m4g;_Obcj{@4HT^Phj^{r9`y{KD*8-}%MKY52Fl_?g+2
z=dx)Zt&i46>mSN`K88LKv4|^*q3@MaO6klE*xufH_0^Yu{_{H|0nr$zh!Ut|tN<(@
zKmEzw_kY%u`=5XM)NkzU+<yPwy~ScRozBX7no{y%48st^fWSi^0})6divBzuoz8#p
z?h&#t3*6nF&bB61U5$RO0QdH89vvT9W3FGldF$2<rOcBjN8kL;H<!!hzxCPQ*xx@$
zKKkXN5W>JO+_?F|&6_?ZpN7%*UAHW(aYow0G?gi|l-hr3GUSaP3$0yQ)+fjPMIT>!
z?RK})U;XO;^S!_R!F%t%dwhCsZCw@>qIRw+i&<{8E9$B!in6SgGp=yca$=l0yI7o^
z9Zs6Y`|jO4KU<zZF%Vp-UAgn&!#l$mKl^J3$0y_c53wrc@bTe4{wJTk^|AlAy}Q|M
ztB3+PK68I&=6>0Zd!00`ijtC4LPZ53K%(Lakl;NKYZfdJ55fz;Qy_M%5N#J!ZTbOK
za0s^RkJ#7u&fJ+fhegVQ9Uoa#?)kmKNdLu1=bW=_5N7io-|mhlgFOyCW9;9!ps$<1
zR#hmah(K#o6a@e{+a4YsCQ0(};gNGzq5)8ZS^<AVBq3dvi9o}j?oE@Vyms~4^~0+|
z==Hh{P0Tu-t+KBqwazs*)XqlVVsN2t0MR$Kq~L2ht6Fh90~MqcWxqe2VVdd~=;@Q=
z(dooGJN#|<;Mt?JR~VxV(oVmmqF78uUdX0(Qc7*iPH!hk5@XVCk%F*ULg;Nb%O;aa
zQ`3KFd=^sr{g2t{a6TNa<}15ueykePwkFM0qV-nyO3~{pWt1`sl-5cqr4y4RNuK9v
znguCb8>Wl84Q<`jb-lQM|A*=PnO2zVcB@={@zw3ezl^{A=3$zpg!O!0_qVtA_YXj(
zgYDAE41E<`B&51-F67A=W3P^j3H|xpjrV^}2oZdHK2V^P>h?N_xWB)j72Ugc?m6o)
zhD>)tK<B&xP0%8_Ko&HfoK7D-y_2QoTem*=@crAP@$;jjBL!+u4*JT}^}4RtZDWJ?
zl7vla+ceI%05QgZBBBq~l;mWEp-~EhcRtue>DJo_+Y<F;<gu>m4h7rHrc0AZtrdR;
zU*vh?K?npz0_)bU>Sj?bSIcER9-mnoDjQZ!BxyQD8KV|DgH-E0*Ll+I6h@gWEp)1d
zge=X8gfZ$;d8yOs=tNoPXV0fA>j&H0)-RtOKUl4&i^X_4I^OE{2wN9UKL7Hgz1I#t
z`}FI8ds$}lwCwisx8Hm7^7f70D_MVa0@lSu=$%Wwlhf+O%?!pE`^WyaE8{==b@+JQ
zR3QW+#2B+IlTs?JV~iN1wGKenbv>QUoNXxxAwWt%3<4rzD*=Si3Jj5o5uC%K=<i;+
zT9##1RrA>lk<v6(3MihBk_YdDAR>&2A!0-#iYU&Hz?|O|rHB|~@Bn}^Mu~suqeP)W
zDv&V*2?8-lK?*bqLLm{QKoaL0orokDAq+r55`<7vDgZ4cC?lmJ5<-CyL?9KBFsXqT
zqaOwUvOM+PH%&uAq*?A-zgn%Ftx&YqE}dP+-~p5Z+}({Ie{%Et@9vF9Cd~#qrFU+b
zz1Oeb{dv^QFTZ;WPoIay(%xje0gs<#jn#(-2baq?{{NFP#$Fv46MDuNV=t|X8M9=J
zG4|5p&@;x^hU3sP#@L4A&@;x^hU3sP#@L4A&@;x^hU3sP#@L4A&@;x^hWisl7?$k<
S-}`F-0000<MNUMnLSTZC(&^~{

delta 368645
zcmYhiWmsEX(DxghA_WSCAT3ggJG8jF6bc23TcEgG(Tx-j?(SBKyK8ZG2v(#4f)sZ+
z+|PTS_ndslb>-S0*2>KPH#2Ky?eI(D{lCO9l0ZVCtSftqdk}l}O0lzwN*oWFWq~T$
z588Lsgi;*ci~m3o@SrJ^c5~st2d$S54`O`X$Tm&=mly5x+$rWGBwmi0kx&wOsuw)N
z(-x;nDsYF><j$YV$I;wmBPa|xT@qyg;BxhKy;R?I|7*SV!kWgMoK#iVCmR#O^zSDD
z4?qd?|Nio=F5Ts=etxtm9YvCx_mm9WBqJ(2A7%9;KDAALPihM>{(i7n&GRX{m&tY$
zt&8qh1!E#I@97>cYRHt>!Tazf`*wD8bQCZ41m|%#`f;~5`{985=jeE*;QGzDyGTBZ
zWF74AYv=8pIKH0m%?_T(H5CEy&~JY)Q@hl*5rExsw|bm)amiTd+Pb>Xe&iY{v-Qc|
zbEX<HfF6h>5=Cy`#`$UH`MX-DykqlqrIxO)SJAX@sQs1q#_#Ss{NwKff^J7EH<`!Y
za2BhU)gzUMSrrYYMBa{*$iVIm#+{7cv)a+-u0PG04nrQ@CRhy;+Oh%Zb8g!kdddya
z)00OE-Se}vv*#t#nG%O@;Mt-F1M+cgr-_v|JY<wNgR-EUnZ34?Nc&%!WtE<DWyB;T
z*;2EQ4|j%l%L^UHKF5MCyOZ%GEiEm#wW4>mW8cn#$?X<9fhT?U)r-fgPD+qn+ett7
zr~k36iuO6>;~z@kEuX_inc;&~y^hb*-TKqb1n?MPc)yw4ab9S6TWBcYak{N-Ke2s#
zVfb|6)Vy-&R@t-}o_!sb%@_X&>wJJ=V`CfM41bM!D|{QK;=Nft@V52Yl$Xcr&+9%$
z`<~}Fu6ZK<GlAf78}~>Jp3ko2g!*zk?;v(GrS%#Tu#x?AlP#Dz{_nW9<fe~a#PR<6
zgdMmUadvj5eEb(OVd{~llzH)8v)Hp}I;&0G;CoxjchQJl%Gp!V(9-|48}lu@_aE$X
zP3xz}haIPz=%?#wq?|dc_CL7x*Z)F9y+>FZh3^0LcHXsidhO4a-W_+YHymEI!G%?T
zQQfjRyXKuF?&xQC|L3}9c-CD)(DSCt*B>t%ua4J6g@qrAf!kv9>8EX#M^gx7t0>*>
z)aPlxtnzU!Z-UA3>S&en>3ZPm5Z35^yvBj^tgy=EhKDHkwk-(d-QxNA`TtmO14DF|
zZ+zlmWn$IqxE=V10XWb-0K&%}6YKY5>!{-R11CqTt#}lClZcD8)1=1x`-vxSaIpKD
z2mHf=Uai;F5xHH{pQF}e-)Bp_C1<Db{pF!Da~?=}zhN$NmilK)d-lIS`JeF&Z>N(x
zo_F#1m$%b{ZehlB|JOjRz#Mnt<*ZU$)pyPlU*ajzk{tm53!oy>1)OhXW?_3DIcH-B
z_=Q3_8&}<6y6d-^1CG9b&3*sIXWtjt!YtFiY5(OL&wg0SZa*+%XHHbmZL}+2K+IWP
zD9O;=Ms!|B+&%A&WoY49ZDYLa_uGX}&zkuFI#{kpyraC4YlQjajkX?lq&q*RPIw20
zjx*h2Bz^}bgs*tWA}4$=Ywdl{(&_2wj-DN`&~aJ+)zIfc+dezRZ1MN{MTNVM{VH4Q
z0oQ`-aogEAyU(5?<#iXvTd#v^V-?RaYXj$*PrL+E#i9>;si~>YfDMsvqxD7vUg<vj
zcI(+Y{r%`5?+1jTXhhmo`>{`}`*8=NqV@6!SZ%e1HPCBr>~XvPJ-KuH4^G)AlK02!
z$-Bp})Ai*4t1;d4|D0^!vR9;2?)HtCh6V(#C<b+1J-Bz?t<o!JSk&v76MvFi`MwpI
ztmia@r@9MUyKI2jclsQ!wLi;XVD0`dZ?^nI&B5xi@0Fs#X<WK}#|>~t{LudXKea6w
z0uS~N^Y#y$?2jhTH0~)Hdfx0LcOJP`re|l1A`S_icRXH#q*aAD9qsJwEG;dc#eVF$
z(CT~h94S9kdW;?4k01YM?Pu!woVMkf_1>~!=P~@8vw8*&Z%Sc;`B^Ye3&6vI;a$BU
zIVmZU-$Ezwcww%3R@Ql2R;adjQ3?2D!7n@?!d`zDc&6^UJ#(K_{*GE_&*OJQ*i~()
zZ(Pz-hm!CO=qWb(3a&LJ`mDFR%fr^2zv&AL3n7diS<f&(-IO=3r4{XIeARc|N&Nah
zQDSp*6ZVW#x?Rg&n1_AqVM8(D_Pv4kt|UNt+e6av@2>OduJa#ur^7Ry+<TM{yRvwc
zBKM^#K2{vyZI8!x>1>V7<j(t@|CsH(N$>QS)A%ZUxd?e)Vy>~{qWsy+4-GIo*t1{%
zW6i1&*0|y{@+{(cm|g2(MeC73{G(Url~<+r4s++jp*^M0_esZ$tBxz6{i}h;&KviY
zlPloq>OaRNel@tMuk<;*gxNp+hrQ3;rSKO^Q7LBO%IoUC5B~*CVq)UAp1ahb>lL_&
zJMjniRo4Yv-j#p<g`KFaw)-7+(RhR<m+SnNcIJMn!vD{HkK(LPfhnG+NRRoH#}mG)
zVk&^vu+K%A;c9~|OqxXB<*<IiZnfp&89CEZ^_J~8MXUPd0g+2i8Uey(-(K>ys;a7g
zeP&<lbsh#~MVykhi)$0Sc5_4iBLmjZXtQ)$_>2a0l>TLb3UE7P{`y&49nX3fKD)^j
zbbpEa!>ay2O;HNC?EY#rcUtv61I8{qg>ibVE8E@HJAC?6jBk5-dR!Ou;Vnmn$p*Jw
zD0mctZU@86mFGgoEr^Qr^mN6z?L&LOt68t(a@jg0(L-k17aW&`(Z=_ka&z?6Ba794
zso-{@vqQ7&;auhE+=zqk?523=|E9jwsk62}(nELJQcpQaPk{LUD<}7YF=qUa|0n%N
zlKk;6{htzHv5)Zos|sG8<X3dj`L@MgCGj5JwElE?+kV_}T#=#oLC9jhWz`OdiH6Px
zG4v?^#HIf>d_gs$o>fa*NW%=mG9IF392Q@)_<SR#P#1o!@zvn69OEdyFdPk#;a(#x
zG0iEZ>aF(cssCaA!f5$(Farf1`PRiwP?G8VsFaVbYoRGxat8ODfQJjrw3mq0aGxhn
zPW@P<W{r{WD-O$BN2(a(Uo;D&2F?W86wK<V$bRxtROPWT)DV!Rs@ygXsXr3v8@?Q&
zq;wY;86RI30>*bDB|2$RP!<NR7Yh)Xori7qK6l;XI<vBfK)1>nb|*7ZNvY+UASRp`
zN4>u-M@BIfjf8uA{)GhUIu--Y({&afhyBY_<Wg-r4}n&*Z@2d(eq*$eZ+#EOW!f!u
zE7XVv0=sh}p>;3ff06gMU(8$i^OL-aQTa6XGjI1&APdfrKO;0J<J=8kCk9Rprob^I
zS{k-~YB~0Uv|Cq|HwT?<XiTBn)6_JUFBpGN_wpaRdj-o+lzH_`#Os-9A?v<x(X@{m
zvwK6Hjv3%rO-X*nGg9LVe%ZqjzR9A1C*t;RF}8OQjAX&J4F1#)#?3*ZoRGu;Ax#P$
zec{!?W+ctJL8YSF3dICMQRV0sI!~E9r!9E03h2_qeF-P9JQm9?laFp{J6V<NqP5G~
zue*BD1e6nsqA92tJxXa7h+7KdBq?}kz?QJJQz!e$p};tMe*E7ATMhDG?7Ef4dK?~O
zo29i+NBjw8EGx}h5UTGPxYKv&&M0iueHI8Q`g$=`a<|7~#jJs?#Ekrb(P~J0;qQXY
zWeGgCm2o$yE{VSP_)vWD-{eHzn!_o=f$123mjIcnswr*G>$s-}dbqKFWu=U^6%Kw5
zH7asHMLD)&RSyxPWIz8xIJsD_QY;eam0Z6@5?Q31K?0A{cpQ%}$I&P8HQ_a{gI4!n
zuD7M#Q<r_Cv!g)Z)mr@&EA&rfY#JRdC<)azj%Uhou!pBS61Dyno_A8Zjfu(S_SBY=
z3ateWk=qhBMrQHmpJsnffpIUIZ$gP>%(fb)utwV{I$dpD;Sa$1;tC|NiTmR!EqVkq
zf<8Gq+j%_dV$HKUmOlzw%LkQ|2*hCttGCLZK{syhFa*MuSM3_HyJ}drGGIN(0ja4=
z2Fr%p+D>mUNYgV?O0Mf~cDAs-57{d~%zX;KEPre;^K^1Uj5}M6y?`|srnr4q7L!Uc
z)A*T~qZie)yO;~XKEuX%>%qg0WjrFmTo_5iC4<Bu#?cSLc`b}$<c9--a)u6qw=&*|
zyjA<q#|Jn?fzT&&0$4_}+PMCY^SKrx44>E#F5M_o3$tq+Y65I!>c1S`>mcart4~tj
ze9{!FByn37CT9tlq%CC)l#q^cz#UdDHgprlkXmByyr(yO%6z(9_s9rJRe)&p1#`Zo
zGycu$kwiOsH<Uv{vNgvDLJgKyRcG1>N}L0lD&v17TKj5l+nIS+ht~33--=SOv}Qq8
zoH`l`Kem?$`j|wVsU9^~dJHg=v|2Mqzggj;DVJl<z=6Yz?Q%XB4JJ}v?}X#Jy2HvU
zzh&%<l!tU&zph9*<cH!gy~7t88IQ=YcHJ9)1Gh{6zPk~atJbdI^SC>7B2x~kVF?BD
z>M`ExmCfqaWkkAV>?p{M*8Jj?v%avhtI?ZTi?|$}coU=fT8ns*B}_0fZ6k)yz{AX}
zYi88Foh0}ZTi@s<YsHqkV65&*JmzchLV?ZXhn@QO<NAeyQC~J2<`|l$Y#bfu+*opu
zRk`{TrU$!TagCfQSHhdk7^~!JSQNhknn|o9WR#Td6e<oFpoNbu+L|$cf{@F@otmc*
zc;OT%jkI5~e*DgPW1qR`_$O<mG1Ix?5KYccrOGOsFk>nFG(%AYWjuz{PTp9KQU5wC
zz0PDL&L|X02i~_s<}2SF>Z)RcW^0~9Im<s-_JYaZex{;<V1hz<a;iUzn0@#Nyndj|
zdhL=SB@557xtJM@S>YFucMFc%(7c(PdOhLV<Cc-uVVlC^4U>cX<S*LvE?qD@k|jOy
zW#0?&P-QrG8A`lkYjdgL)MfQRDL?x`yM%+=JnCjEmbrt5l_Zn8z4Ef}(iT#jq#GV~
zVb`k0WNpo%o!SIH$~P0$THVD0G+9ml@a-7Hh3QxtRn^zTGA5E`F*;!RZWwl~9hzCS
zUz{C#dJM5X?(g1LUqC&wJ~y+bRMxh<pTy%sk)d+d=TtLqF@^sAj5Wd#sldz1`*r`(
z>*e?ad*rA2{-4D39kvKJ6a{0s&vIX{Kk$Y{CI@h(P6t(GN+^bSke8cq0HKP7Ms`OQ
zOU`$HjEdACrZQw(+WP~>*;J)bAL`~kEfDSaR2%qk$AqAjo)%qk<z}X7mg+Z^29%1f
z)pc!b6d|9%4B<@!YE>-QSol3y0ZzG2AcfO5-q3+`x1~y6*?#d@BIQE8GL3YJK|a+C
zdA5{DH97hbHM^k#4#c4dK$^_VyZ<-Pj!S^cV`y!}Jqti0q{vdyzZWttiG2Y@diUO*
z$Lcx0eDXS^iE!4JQx5)&8~Gg>MAlM(LruWxc02MC*F(xQco1Cpfd(03vfD~3;E^SB
zwWxlc{$2BPFpc6bu-5c9T<1!488&qJ7N<@1dbQFI!`hk-adQe(z?lhRcX22A=<f7u
zhbFFqF}V6O{9*KCi@F<sez<k}*_i#>?gh$%{FXXdNmHTQ)G@7mUZO1u7%ljPq7_$r
zRWXjP&H?XGsm7Z`bJxl=yxW<vI5U|3xLnNnfdRAWipm$X0{*S{djs-X<q=DJLl0M?
z3TTZ`=glm$kU|6kASielcWFESE0Hyw;?qh!hhn{{eDRiG7Nb+xFN$k<BE=C>+7}Qh
z^oN|h#KA8b1wM?V5D^YGy<99TZYYMZtX2}kiz649rMP7~r<;>qq!^|#c6WZs*HUN_
zg}G}v0=9O$Oz<sms?_kbWM-`yGCzr3u$wq<eR!evS$kCrm~gh`lwt_aX!29~4`4In
zluHatH3kTkq-4Xpt%9;hT}BO3h$Nx*>t!&qQGeI%{_+U~`=^n%RgbtPTdO>otg3wh
zXNxP0c-{Hv`fuZx%r}CbOrO3?x>dg3&$82`t5|U#I}#>wQUr4uVHwRkKl)54y*<D7
zUxR<X{!9np`fy4lC=`|soy>i?NK8yDN#+<QAf+LN7=e*Ig)YUU`oTFf03NOf`XE0B
zsQiODPF5Agl#ZyG*~kAu7rlg}=stGH7Teuh_Nr!uEy)vMftIMi&v2tk;bHb6;Pu3x
z_(aFI-MpQS$M0l`&5Qb1getgGN?`+nJ`NNxfJy&pI=Pv+QJ=rFO|5j!@apGReV^Zo
zzB|27X#DR?{Sn*RHh09F?_+aKO1XM!<##er1(JTBkB$y*%hlDb9M(f*Hv3Loi6lH}
z5IfxUM(yovg(oOwEB|2Sb0<S|mRbfjK6|n8V{xqsWVXinq)lvO!}mGAPURXf+V9@J
z0~GsYlg=J<_SI9A85Jq~b>RN2IMpnQ6fi7|oBPw{hNZu|3`HK<euW-Ik>U|e%=i|z
zu5Y_ZNlPy!+S_W72EjDQ{eeveW#x?+^;prhqXdPt3O)ntEw&?@`()IJORs+tOEn5R
zie=8Q%rlAzGe^<Uyu3j_YQoUCP_WeqKwEMqWyVNNkW#WRarWKDd(>J$N<jXVM9rZ>
zvyC%dEsLeCO`Rt3ex&LI^OABi2c}d&wL#;|@ddVYUku^b*qS;+Df|jo{d0Ejz29DJ
zZCZ;j$>>r+sGqR#!5}0g&YoO1Nn}(p3M&m7DISnt*IwQ7+eb|)KR>_X_Zj0hKnw_k
zwD^zrW#ZXCl;SgPDgo!x-kE`aXRla592U=~kEBCbi(^x4-|u@pq1|)|%OKMjIV#D*
zw-cBkR2=@UN%8#9(BS?K&J5~zsA%X$yB+4<GkNQMXV4r?c^Rf?sfQ2gY465DyxM0u
zNYrD+db(=2=c~gC6EBWVg9jJ)0X=hX8!rm;9dX1v`*yp#inAFzL<dH+@4@jg-_>-)
z`u#F55#t+C?~S&pzvKpI{mJdyX$<o8X^|GTpE%=)<C6-*)sf-Iny|wX{My<t#3Rrs
zKjRAS-i*|xlfs}meL>meWZ3HkYy|^e1?)VYF;cAUW!Hm?$Oh_DACiLi0E+t`^@D|b
z9P~KqSj)<*haB(w>tL=S9%DVz=5JF}?y7l{^;RFS?x6W<R-SH7mX^}!79}Nf3TS=w
z%JrV94^h0Gn|-m8vdXOaE}wTQ%;1&+I;`Txm2O-wmkIIth#8%J6gvM6Slq1}S)(PI
z@7J-3Nv{IY4V$&<vB52q0Q28pxKH$`CThKo?=%}>PPT?5FY>-kTR?hf%RZ3#);HHB
zm<8=$Y-Pf2-1XgEUHREJ6BD%<g+=)JJFEP_j2hs;h8?6W$2e}gxpYV{2acDLAEHH}
zaFAi5#uqRe>Z(RA;$!4h&Jk=pSUUQB$JV-Ws<>kHBFm_et83;KAZ-?>5~CRdfkHDc
zT~KJKsp8N<h4rQquhEdGw=%eqyK--R^yrWX$UwneAbea55G|FM6e=nR71eW|kQFlO
zm{DC_?Hqsq&p+@iXn2Om52fahiwTtT_FPX|N+LA35b$)%eM-sLaI_Z@@-2?xMBiW>
zZ?|K*{UpTv>%0ZHVjEo;c^99o@3^^3@C%i1w@+K=R3XVal_p)Z`9R%AMY)0n7u_eR
zEJ_NozeKvG|Fu1M4v|IKRX?-!1Rs0)x^D5uOCT`%DLFcK%$8ap(*kiYdNTKc5tV?*
zgW%f#T{A+v>>;V#(Yw#uJ-_6$S?BDpx%{<Ygwh9DAzD{p>i3L1_<i8uJQG8ry!F60
zvx%49$t<tbl%u2H3Rm06i-_!z6m^;@YAaZ~*26L={t&bK!)z~z|9XmlFbYWL`P-5C
z`gYsN%fzLCK_E-<C2s>NtidbuaeMnx)7)ode(F-;DFS#n)@_UqveelepFcS~jFkSy
zGP|-pjig!+yx^iT<KqhH-knW>rw#Rwt3&8}81kK(Y0csa2_Ojt9>fG7z<7};iD6;2
zuzm|NrrJGJlE6)rgjRs=2t$K3PqW`GeiLujVN3Qht~9-A7`O{uquS@|s3MdlTTmMY
z7j|)sD@N{jB<r=eH96cJ8dikVr>2rhniBa_uQvds+IcZCf=r3>^lGtbtIZERBW%{n
zk>sF9%4*r(#n1BA$hiHVv*4CWKPRQX)gk5N1}ZxI;j7@Jdznu&a@+2k_Pzt&>Tlj!
zaeNn=@~!ketJEwMM|K<yg{{EM&zX&YltbJPibHKdBeBtSU8|*qJI4!JY|t&A599cR
zZTSvRL%KaTb((LOdUDuvkPxF1!RbP#*(%l-D9#k}J6Cs}DteA4wE^EaNZj!*FkFy@
z*z#$IzAOuf2&lz|M3oP#mD}@F6wrMVh7Fk|rOxmEI9__$*TCN(5~>;HioxD`*<#QM
z|G=ajcd+c^YgxdEI&!CPI%#KXksr=i;Aai!6ES}7W~s2Q5ENFO(C8qBSA6A?UfDN)
z-`G~yz(x_<Cr>~Qt45Bp%;HXB>WyUk{#kg*O4hN>3^wnDM^TzAd+(%aD-chZZ+$j0
zrt?9iH}XQnL0pEjcS!410L`cOTH1N&fjv|UE!@I8hcNd?lq(Sq0fFkOx^0Zd+glx=
zT+>8JQ9Gwv<&B-<I^N9rqEQ%%irTKrt#rlN3d0t%Q~&!IdF;e(au_{Rl69tzi1r&w
z#_#=^i^Pkz_DU)fzbfkIzrvQVFQSi7;T6Yqf{GHK3lB{9x|}IS$`CwXf^-aZFhE4V
z^v?3?FmQNEX+DL*#WgbY_zHONL#axDeU6Ac@z0FErQ`No>f5!f5++t%*}!y#>6VaF
zD5srnXpmy+C{6tNoqVOJ$Vb<y!erZ}YFr%7Z^7Ivlhi0BiQ|ll(^c!ob<+ldWap`j
zVuhxIU@j>r-k7~z9pfo7R5b`0f-a@xV5Fp@h9xgG7~}jhF4r^;7OE!~)>97%87_D1
zBsSIAjKoHbp<x(Fo!clUzkJ<p3mY2>b9OUyxL%x!RF;jEn8Xz8>jqUlnI;h;(O^=?
z5J-weDwNJ#iqvo?FM>ekD&_q_bkjPGALo;jlQq94&CsHA=~#aXg;G07y&P1YcqfM{
zHORwhN`VB%N2YQDAy0#Xk&pmVt~Ba$c6My)TA@Z3>=?&xQj)>NEx|`p$g5at4k=SP
zKO;W^5a(d{_Up<vgLIQ8;X7&#=>#gtNV!3LKNL<WF{Hs7nhK4~uu>+IsH6mme#eaW
zrSA)Eb=5GVQs`qjX$uZ6(GwLkwA@-}E`;dG&aujCSjra2AzNDyk4C^G<#&nOtZZ6F
z5}Hrb3XBf6>Ck;QoeXEX!g<rdZ3XWBZm=Rp`KKfjiD8xzP-={tXH>#z-wEMXK14c+
z_vc*KnacRs9C@NRUua&5I-738dlu$!8k8RavZ~uHc1dPQX4(xJ<?6|Q;i~CQ_<qz7
zOmK^T2=uk1X88gbjtbx>C>+`q!w!!YtGOwDSSD3e+B<af*U*N`kVW#bl0_0sG%RRp
zl3jR~tu9-X8#>NS-2Yp>4C4S_nlLI@iXIa?Zt#B`_+9vZST+wFCr&cuqN`oFpKO(-
zp_0$>p$}bdU|zy~V6BJznEepA7QAXd3|}BC{5b#B{YYe@qz3>~^i4@(yrdBIz{1sf
z)xEt_gIFTA{7qV``g#f^oU4kB)MgYXNKW2fbIkn>&cql_GwM2X#**tcIxccL*sI&9
zRM$IMZ7X3?V%|Rxoh8%^-*o;uFSz_eS=&sylm949lZ;-wh-$*McFCf?RmPI2ZSckC
zSFc+JZ9*RI7CL}3r`PIW$y_HaR~`F={ki>1*0Dgm3?rQ?b*nA5#t1%n*C;YRBsn^M
z43z#DRfX4bj>-t?beq|gwDgk27YAW`akG17UX?csz@tvu&!N95tl4*b+gE4==bjI=
zpM)n!e@Mr^PdZ$z$oCXB>T8OM?@MxPHxU1ZQ}R~};&=mm&sAfip1|->*3)7!4=k}>
z(oDL-&X*sKj|eIAbsafSUm4Lov{P1hQ}lmbqE(6$p$-pJD*U~x<+EOR$JSv}S;hJ?
zwAel0LW7oR3_Z||YxnozhxlN+NHm6!SSnl@>KwIYKRR3;5*_ZMF1l@~<+KdBEsup*
zFtN5Jdp!$~?#%W>%u(Ry5E#Rl;2)|sWpka+w<XhN@j|P_fE-R5n+XsV2Neq6%Hbm)
zKK|p@bk#9pm{LFs42P-~>X2oE17$fU5{Qbos-<!)o9qM+`Dp~+BuazCg5~rSXr`Af
zAcOc@P-?Hush{-zgX+WsQ;o;>^O@@%`Kvk;@hJc&6IOmIXq>+`Mi+;V+FXxJ{fE)#
zS6E{HV&1#!COBb(3<VN7=#v%+!!wX*p?yAOQHpDEDwcG`2*vLAvQqtMkhol_gbc-g
zJ`S4gA3dPUErZg(nO=K0LCR12#~mdtv1J_WSS}P}U<^8OZd5T0aQyV>D0&omw955r
z2|ZA_<7H4hSE0@nsgw9F9$K%N>|AIy7fU-9r_OBK!_d5O@0?e%VAa^Vy0+GyymB(u
z`IbDLqo2sJ;B|R<quaq^X}w$1V7!H=$-&?s*|si~X$n?rSX5>R9_0cm@c>Qa-9z`h
zElhYq)c0Uf&Y6WFvB`RDZ0xtH%e346&OT5RV^jh=1b_Bo>#&$K7<px$zxs35$}$Pr
zJPP5~y6V(SB(0hSf1b7JYeU0%uqn(Z6>4dp=WY?qBNwr;44+bH7S5*I98g4+<DJ?Y
zo<A*@e#bY6@<<8B@f*4NhK4q$f3d!KMpsAJvWR-|#&q0Y%9XEwQ+GM%@7B}B1G=gi
zPBFL(s{+gSVT$Y%`A30xqAdxod9&M~9=ET8xmD6ba7f}SYghKpgSgc#7!jXz0yMLA
z!goVNlVLjz1fl}TONp!ZSFIlRTOv!1`caRuI1UKk-B}uxFHF;X6obT%xq0>C1@MLI
z`<+d6Zj-OyBtZ(ouo!J{>NP?ElNj>yPyO#3rx~KoSgI6?qbPiqc1T?Fk+29VDh#}*
zD!xzJMSWaeO>CK9Og1KaVO+z&jF!mrF40_h$r$=xo}!;IK%;<k+Cj8V(r7y4#h*ns
zM4g~~iifW>N$D^)HP|eTyNrtT!ljsm5g8WueveAB|AR#&YF`sq3|KuF!1qTjia~pq
z$iR+eJFjCqdb+FoJj8YOO`<=^VJ;`xWhwc%4Bp|($VYyqrjXwa&R35u@S#K7ib1p)
zP|C-*X;wmcm_wpPlwx2jAC=^i-*J^=Hy-Tf4W6lE`43wRp9Y{MNeU?UzKk`SM3ay=
zh~0K<C3lCI&Nsu4)|{7sHK%%;nCJ80@Tr}(qsU<e$&{G+i?o3Bo;*FLz#z_HDt}HH
za%2;^U`~|&oaz1?q*3STV0=I9C`>4lB-0HWhzdDI;x`D>-z7Fph~oAiKSV$_h+$QY
zy$<_t0@>EwiRGWW3KD!Of|tqQr`)L!vKtHxn~?cNO(xRwP2XsMaA}(C{C4^ZscxQs
zlS3ItN=ghfw2O4BAM^_O9Z?^CJ$ht`6L?=?oF7hcVtLRazqLD<8+f%2tPc-k#iE;%
zL5%~NA~R5GLmg2vK2vR>hvNIo_Xly}_+w(mfN%n+Cwn<C=zGD|?k!MiNEetUE&+-y
zMvD}V1S0*0;|EBvfaSM;QDK86l|j3AcCqH~=9W>S*9<91wU2)fri!IEQ3d0G3BW%F
z$mbV($n2Xefw69yL);&I;qRt>%MxXu+t)g!QS$?Z2?zxFj#_<vO$@w6r@WbAf!9+a
zwp^9C4r0=>j*xYvc)d4!PTYHaeLh>=E4vcDS_0*N4**toEClCp5^&e)txTyZTmqSN
z!Zj&5X@3h+*tPDpv?hsxhwj^~cF`2B*xe9$5m{O~<XdT;@KPaP+iv;Lu6+%^q2Bf`
zO0)G%aIV@g#O*V=D+N93(wBD7p-?oqk>JO1;=)%Qe?!LPpW2!nbiI72c8UXGalaO#
zlQebnfh~yM^3@t(SgNL}Lk<aNsb%HoTGm4Z%SKaT-(weUon!a}?xL$);-zI1piaXK
zoL3+4gwoE>)ot(Wfxn%3zh~x9zP|C=9dA8)ssvoUG9dKFNs36XNO0(o5fAoHqTgG6
z;0v9|FSM1<!n3`uMhti#CLz(8dIHrxu!9v~<wc|VTT7m&ef!Su8<X8S`MKbwq~F5N
z-I$(Oq#Yd;>ej8J5J7!~&8AHc;SUV_cC~s(F3VYPT=eq47Q&=`V67m|>gltxH~FQ-
zXHK=g``57wOG7SNtV;s>{F-_4ME!GZ8%dkLz@?t(7FOTxD~>79{dgvDa#*#q%uHzk
zB$wsbZ8ME+#}snhojBU0fPnhKBwXJ{*#}=N8B9|RLj_PsF3}2jtWe(~k*HbZP(xa5
z7V9rOS*9pW&yOn=-jtf#Bu6q2j}S`E*dy5w*Dlwvuc9k7T9KJeGFxg8&iEI@X|~xm
z2wL<XAhw$8KbW@!L0~m71m_SdtQxQ-3AMD9tL0OOXIN`7vo6;()DljW<~(!dcQ{xo
zr733VXBkfI5t{ZuM;pw9O?WJg7;t_RMpAN`lz(-(@O57KP1wizH~Q)YMl1O5#-U7c
zG@quZNI}R_$VTd58c1+a{d{m2ejJ@?MlgXst)vv0_|_zeloXD`E(Se#m<y0pP9c)O
zOtQj-a`R3(E*pHaE1k|@nHxBCst2JA#`wKMrjF%5G`#KR=(i}x-3dNriN@K{A?ekV
zz@*`lj26mr3)|SDMxvPv{*1!G5eyov0Z%zNO-T)+<rm}hzZLfW7iI6gwH>x)Or)KJ
zD+a<`p*f<0_q_+r_=Qq@1_6}lv`y$QLs9!CsR&TeEbw_y#$xI;Lnp1CM_LYyER23Q
zdlGWvs8z-m6qs_{+YV{OvlOaSvZgo@67nG-4{Oq`?DzJ-sc8Rw_zPzuC0w|$L@z!*
zDY7i`WMUjbVJpHZH@|eMnn8I*ZK<Z5cIX2&NAr}GeQH*9lWlIB3o!AXIJpAhnZ-Lt
zfdXPnIrgx8)mm{kCfL#JW8+n-v#w=tW-KJgO(<*JtiEPZe$;-w`Ea$`XnQwq2jt)!
zE;)BEds<t1EPej?QKY`9X?rngh+FsH_-!~=`H7CLt7~HT{p=8-FFq6sr4Vvy(ji6Q
z<+(M&?~u$-yYDEMfodNN3$Z}clJ>Qv>1EDVXO7**`Q<OQ+<_eP5*dw%0FH2_3Z(8h
zO`p113!b;a|91OkP1EAJM&s62R(d>I$0?;$lgw9@-Cb=VZdRW8b3bR(8xB{T*L^M&
zf}khfD%Ka<tDAceC+~wAVHn#o|IO3GPUrnPWmB_7<wdOlz-p8~_7@HI&76Gaaiu@X
zQ&K_muZcC}VE@cb=zJC9YjSderT|MtvnTI$!~A~w)4xud*VdV{GX0Giop&ft6L<4>
zj&ucMJm$-CQ&ztldW99Gf~K2kOxn8@(yWSH=dcs@2U6v~>dI%#N-ZUI?>n_7GcA1c
zc=a{FGXqSW0!V4T7=H0et>1Nxg3-*8NOC;I6g`}0Dircg{MDXkyT^T>+K>}>_!qMu
zRUkeRC1!}cY^itqhboreDPn<42q^E!p{C}x)ub&+sv}F~=F&@nDB5qK*ill0rWLrz
z6CWHws4^hw*c&?z;g3_1ckdT2nhvqZS>CgV$bHFtZ3e8r-b4#8v@E|3bjzRm%+zQG
z>f(b(ZCTiY#hP_=urS0%YfLF*i<L9?uSRfKgB!%s1Ajwjm;J<<#Nu9?8TlW+m{5C#
zt9ck#WomfAT^N+hqrt?*A{+IKbGqr@W;=v1J_?^nMUY$(aYm0p&Ky&RVk&}xhXdNt
zwt~(xaRNG}5=i*SPUDgDuTXm&$Z5As{cymO!yshN3S6x2(MhCmC^Z=k2p@+OjBHY@
zlFI}FAv1{~<=8rYTdE+|Q8W(cK}=lrUbYxgvPMS2Dloc`_^5tLSeCOQ>Oydz;>%S)
zISFKskyPN}KPoUcK6tO-ZX_pES>+LjAXvd0Xj;dZvKaFxg@i)q6*%N2uU_I4#4&A@
zDj{{id(K|I^<)V*>2Ic^+J?fq-fT9Ai}&Gf{lJMKfD97EkZ=Fcb9nC7gVBTiCDcIZ
zLvq2U(5avG9c;~FROJZ?G@rgeh-3qM=n_ByDX4Tj{?j^`*i#(H=v`pB?syxA!}_lP
zrGS%zqa*8^zJ@Bb!lpVuPIMd)E@Tj-s%U(9w50k%YEl3ATynz+ztH$++$h7COg~+O
zUL^|)mK`toJMFl1J&obSpl1GUZ<CSH;oI9ky|P$|DE`NH<&lw>p4NF&=LP8@oI9RQ
zPSdrt^e8%lE7L)Y9^&=t`YF2q-tfKz@~2{Bt6<KoUU#$S%hS`RD=Tv4Nix#b`CngJ
zX&1#6Om;2k=-V3Q&(zwaV>9FAN4kYzFCnh-8NMkI_5X0xv3S!THvJ9m0e2O6P%aka
z=buC1fjH%BhlWbug+x2-xwN8`mwxv2vF*kGAS)9ggyJVOf|lH1CKx5iDZmy4%pqKH
zNhh2~=B&zQv~{$R#@So`B~FA{?beYTU0*$Yh0C{p^xFAQIm^Zz{W0Dkf<(H-(dy52
z7?D}6rbhUwR3)4HvDfvW|4se7LkGJU>sFz9GePV#&-%+aW<$?04faHZ=X2)MA@EoU
zs|&QK;to|EmVm2aQ6KO@`5Og*_NE1u#ChYNO@ArOl<|v63DIrnnI<vK&3y1l@Zmay
z4#uu`C?nOq;){5KV^a#n^>#6ZK~bm3q)FCW557{(Z_6}LEkjxZm5WMC)K_v?=;&lp
zh#C+t(Q`R-E+#psC*o&j2&L%4xcuGHc6|lq^JpCO3<p~v8Z5t@8D6*mtljl5C`L*o
zIErvce>8DeiVx!_J9Jk}vFb7<#?2kuInZrMxwfDnn+_sR5N7B*-XQw>zhx74WKpzN
zJ-8d*Q9{OEOkDEU23wWrF%HOcxm%61-zbpw|JZO#``hrE)=`H+zkJr4og*z=uT&yB
zQaPjZ3^@WOoo7{;_S;qou*?(qn+JDZK-&voCW`oouFXz_R+ccFZ0LBS$kv*7kX&Qn
zb<;OhA0Ogk{d>-GuF+c;GfIbqNl`s(FONfk#3L#OJ2+ITMGT#1<$FNRkRPWEi4J)`
zcydy@a-)tnTa39JYdX9R`3r~<6}JGyLj(CS3gYex4X44NeTmr*aHtWZ_%D~x+LXHq
zzDdtDQ``;!{W(F9W?i4*1V<P1&c)~%+Gh%JcDB_ek)Q$hStvAI1;$_$G#ak1m_hJ#
zbSr|u4}_D$rGql%Ao$9UIxqWXuQ6d{hu&C|g8$s&RS+iK1W}`&Ne_qoAi?#&b5PG#
z$SW|F$WaCj2+{>uaA1{yazQ*OC>DIlIh;|>!Mh&f;FmVAY8J~E#!z$=IVJ0Fu!{MF
zAEgL^3OKbNdmTd1@tYq+Gz>B4RJ&|MjJWVLTVTywaauK5wENNPO8AO1xPm^gOZ0wG
zGS18~4UhQi7GF8<Kpsa@)Bs0mBJ;aBD1o4($vE+}vmP++;m+Y;op|@wm-b&<Y;`pX
z2HtYq(}AvU%cMu=Of!7EmA&&kJZ@*r{mI9;5dzl!$R5|{s4qLHQ0{oWd*W0Y_g*_+
zC;ctCil_eiMGbAc>BMiTy0#|u-L}mIV$CG|_4JfX?}N#13G0ytba^Xqr8-lOW1cvl
zFgIb4u`}Sp1C=PpbzF<glUe1=I0iQS({$N93K<ky9;*01>ePA3CmiuV^<0Qg)cguN
zQC&Fuxif4uX<uIPJu7`+FX{~puAKQ(<7cL3+02m=G^(P4jT;tTBlI<BhaV1&ek1C8
z+N_3Kv1POB@zV6B;>-=QA^A3lSy4LyZG~N99|?GEFg??=wENN}xT4BAhs{hq#w_gS
zXcHyEkm4i&Vd!+@nylYqJ>en9)^Uyk&pexaceQ@o_;jWo*QYv|IytT2^baMAxszmH
zlirbq9v8b&FIbA3BfNCkVres2u{FLZ(56e+66;VN1C2K+v(Qm11^HroGzsImGW(My
z90EkYIj^%;94J$dqBPjM(t+Ep7WJ-}m`on-360QDoS%ssD`)93B*ZhL)4EpRGO?w6
zY~HwBp3Wm0t>u#(MNb|3z7bJ`aDdpzd`ye8)mA&tpM9lQ8W2kjQOZRM;}$Q#Wr*md
zBK=h_nRDN!4@=S%_~VJXA;XFG`0sl(KLlVBS&xy3;w`d0$^ssP<l{2Z(nn5za5H#l
zzH$?^9l7wfOb*5Wv6WbcDnso^J5EX_1v#ZPi3w`IaF~2+z-?vuCCM(RRpaO%1?|;g
zTK3X%l(Zq83@p*}p=~erRRl3*EAxH@+)?ZA?sp~32-8>`n#{if>@H~_WOFEzTsHu&
z{=@cKf-0YMOVS?wwT$H$m}IbyZZg<8I5Y?8IV-jdVrf(jRQjSSck$8CK>CR?I3<6C
z#$nqzn<o6FULcKmZs0e*SzyyuI~P>6KFL`C+bY>QBi>QnOuFaQn?S9p&~dsz1hPn=
zUPeMt9?z#5p|LZ*sCZhl(YGzfkpL&E!)-9ia|24!7}`DaQUrOLf}a3vG8ltnQ63WV
z?e`4d(C-OR_w@O2P?+L4tI@OrxgU3b2=6Y3DQXv(E4cUz2%Kj$D6UB5pO#}}gzt}}
zFz%=1o?l*19a)$YXO4#Q+?--#61PMo+17sH&NA~*`^dC&66;y3vsUk84R9O9jh@eL
z9~>ME@7K8eI%8YeLF?9EZ%OBR#}{1#mB)U;PyLERY&?crAFBXA)hH;}`<oTl0bW{t
z@Q{<8pY@hAe0GXZeB-z3D86jJ%oa~#!v_)J=C-wEEo=UUx#KI^Eu7s}_jRA6@i#jP
zw%$%$BTg_^x5Ks7mv_x<^T0BqKK53B{pM(Kbaq%qkgbm93%O;H$LYqqD{`*0h_fBf
zi>}z-0hLU$%(OkaAqB5@jHdxOwFQLN^|8T(*E6~e%J!+OiOgLdMD!k-3CweR%E+Ss
z+PkV(tEU_lwV)?UsjL@P&3cT?(><@s%9tZPSL=UywjW{mHWOtCIRZj2F$%lhP}<Uv
zj#C}rp||^HnGr%D#mb{-SpzR*)5i`48?#abE&0b~T3=sc;M|xcxBcZ;#XX!aeTR4d
zP=a}T=<s*5tUJ}=ng2l+PE7C0^+VD7!4u?P%AH0Qoca1Nr{`wjx95wEIN~YaqO;QO
zE*~+DI<Kgsb;mqiZ2)f<*mjcF(dxSD;(nQD;nbeYKy(Ff`b?R1c~>PdDH3L)Wx11L
z%3lkrHy>Kf_9v?$S@1+3(NB<xi@M>|4a*9?WL1jf;rBciM2PcVhNuac$M_H-wq+5j
z<3e?>uT28+3aM1TBSI13m>25tGM5{?tv~<<(NcSq|4n3S0LJsCJjV)XC;y@&3`*&w
z#x_4_&r<wxrmKs?;X|>-8vijL?ub74+8>>mgowm+^17o+$X2k&Vbmq5OlS<Mru166
zeiJQgR!84sC*1%XAu_i0PSM?&ZTavcDP}%6#R3XtvLz-aVI6R5kE2w_sitKfPGm~{
z1zQp;pivoO07kxLnHO57^jA}tpN`7>@-&cd(wJaZB#GCPmS4er|L>b?fwnCVdHs|p
zn#M9rOBvmkMFQKo$Yi)jXfMREGW1=dOwk}J3TmJjDEQA_d#L%+xGBhQa5B})h;J*@
z2#PM*i_3a{PF<2a(FShus+%`wBSGQNu}02Us=zA51`2EQ<i2PS34E1#uDzE6Tw-Y*
zTU-5gocB1`VqOGN_@PoEQ*+RY-K&t0pQE1)@9&3g4Zu~#m^C<}qzV!wq|UD0<oi=N
zCC5aC=De0!d->_9KxAv7j(-EVInI+Bg@U9ft`636BN=h12&rt0Ud5nrg&nw%^07d^
z^P-Sv0laUIG|f!GJ91J9C?J$=xd6w+okLC7)x%V5*wD~PS!K&@jUs!w2J_veKBu7I
zUHroR57b&$r@YN&SP@;Gfpwm?hGS>DxA)2HkXvi(yww;h8=dy)Fu8%xzkj)N&le!a
zz<E1OK~tE6xKy~k(_LsDFAv(kFYFIzm_hOnfb3e=231zxfa%N9(!=<fF*T2YQT5N|
z*WyM?4RyctUYC{Q!kn$^Jv+G>rlEnVmgCUUq>;=>-1q!Cnh8b0#-d?xP2x-|4zHuj
zxybtc-NBw^o!0f_tn@fdoP3r#;Xl^~zfMO#3sTrDIgZQlqzIeeQAGymn0Z?T=t=}O
z02Iv1gacQJoDEwHh2#8JdSY@jb+CQZX-C%90}PKifoZNFrkn2m+vUOC(CORR&CQDi
z!*!kY(4NU8HS!NaCBnOI`npAqeFVfRtOrZZs+BN=o0rU<XJLbj)nkBu>%x2z9IjI6
zu|`3ztE)d!vr?U*$lm<t8R01-$ip^}xpi@Szp21og8oRfj0{ezdoSF1u+i;xOwO=d
zJdE!ot_IG=#+yWbz1UOBI#RRA5<O$X8rA7}wJQ7a&1#4BRde!D^WR~`x4s*>TV;WE
z8O4E<ZUM@Tlmn#55D+w<wSe=dCN^FAj`*x&;8S)ZZ&>|@_VKg%+m_Dt)g><=`|q0b
zUCUfecwS#W(X47zf@U}_lf;HN7iJD>D8^e<2QEisS~rln5f$m6SVo`UCaMgDb~P=7
zvV=U5DVIsL7A->4>(1-8_OA0`IkCG#qez(QbxxPP_hrAxwcnNDlg{xSi?q@QSB$7%
zffq$#)4v>vLtjwWXq6*|`8nPKW#b2jNBpbkRFM0UP(SV%)z$r)qTek(ZX&g_f`<D-
z{&L^U&6$z^AUjc`N>aLXW=CNNFX$W<E;PugOS#c+#-pl*3H|sHW{~9MB4N8qT2Md~
z&?tjV;684z0|SRppyzLISQ&a<oiABt^y@D;Zp;k%K|W`IzyCY06}>_L@k4)#oeV|q
zn{u)%Iu!jhR`F>_2uM^jXa_x3T^`mx4-SIum5d}Dx`T`~n)jClP+7lkAfw#0RKy0%
zzaiz8tJNzBC=7xYpcuL~*uga3Bljzv$FoFBF)>OocQd(IFnwY7Zk<}~uOo+jXs{jB
zm3ixz0eiKEOyJn?ru7ydkihKbhx>@3W8#ovQa3~q9k<gLx4K~yi(zJPO5q1=&C?^-
zX?W0Jb)%hv@B{FO+NP9s?c4XDhXTsm?UI<gtg`I<C{^A2I|sjX+}E_}Xq~!AL*bUX
zsKKJUZ8+5GpRUgLW-zXfMLq5pz36G_^H`g48k}YOeK)$#hT;!^<|6xMQqV;AZ`d&{
z<$v!?UEV!VI@R7EF}n*nY`@r&2l*>;GV*~?RPgbeKSniC6$yRG0tMlS;kSLr+-^F2
zLjYm>J|s<BaO}b!bww}O`B>I@I+G1Ni=lnv6Gf?Y$XAo0%7agExOq|eUEt>V!S!-|
zZ@0xps40;TycJjn&b*ERL?+&Phx?O>Y8FNIYUfq=>A~jd(n%S68$J^Bv><I_h)im7
z9(6n*M&X}Ci}jXG$|>Kg{71eC&+(f*)5i5<>S>Ev!8X_OMar}sc-aV_hBoo<YO0z+
z&f}dHYqx#~`ThAnhc%w|z=xV;Dz$|7bWLW>N?xTGngHYbAK{y{6K8)MV!o+5=oc(L
z9@+geVcfRhOM?yw=Koo9ak`r?&yS}+`+abd9}-O|x(HjZ*W~>)hq$=4EGi4nI4s;d
zIIok7r}>mN=4sAn@d=hvA$mX#uP_w8UxTf$iY{(ttUvAg#+prJiOlIY>QqW#;>_k_
zsE|^{n*hx^JJR;;jIrVxiT$LuEo(XO*pzMhvj{pK#dhj2yli1U-HeX&{nbykxA*42
z&X-v9GTw{7K4ljMBHquh&^1Xa5yxhW&dTC#Y`&7@)z~+6JNe$=gs+w=_xGJY!GM%e
z567kuVH;(&5P@Z1Dt^Sc_o2%A{n1s2m$mz;{Sz?F?h9|V7j4$^Q)~Ct{E&HuFCn)X
zO6}GH$-t8pMyHFIYo!#iW_m>x&LSwVszqYvXwy2b8oHT6-DIS@wLeZG9c6@^Qe^8^
z9Ju2&lQ6C`T<Z(_bK>FL++$`|{tk8YG;eRY0vD%wBQ)b~!v+@S_bSNQH)e<QM^=F^
z>I<M4O#O-_N$G{Uo*W&i)2Xt8aXBu|8*y7G`cGR-sjn|R1O4rxdl%sbNzzOeKW{I`
zU{`Bi7Q)x~KY!j0vp=@Bwl0Vy-rfEw{&3;`*4u|;rP=9Xd1%iA=B6;rqUb5@=;>tg
zi&#zI*xSL6>IeR-uOs}iXvg@LZv3Lc{Kr>-Y&XU!L#R1R)^;G2-t<>_efe_5PuntT
zRQt$A{VKX9Ha@}(%b)cNi8J+_iM8df{bc6l7t~Goehq~OXEY0f@C=wjn{$)}^A$1v
z5A|_-S?Y~;ym0~Rj1OY|)IDd!FYSVStu4;i1mWIQQpO0wt2IJ28i#RR1Trcj(s>yW
zJan1aofjYsaIWC~q>&s^q&J^#R;9xg!$m_y<J4rf=dI_1qD}c$sOf16gEAvKli)k`
zHGPp1+~*pJTcpNTR^-QqA7*U#**|<L5dhrfwn*Mn`LJC1;E+qw#TO&C-NlCt*v-m{
z^>=%~e{GIxm(xZRa?A(p;d5?6rGEj|KeZH2v+|QDyuuU`??+X9E{E4>BjY2^zSNAm
zjD6>sT-M9VJc<e-8JOq_;fcx;@#H&reRi|+a|JWE?!(uLb#JGSfcJ?QhEZ-Lt(VhN
zE;2*e?a2ei%n2V@@Cc7(s-;4@&)<0)+O)n3lRVAOjD6?KmHviSb?cV{8AKcK_>y#6
zv&yp}o3F|_QAm`2aXZ|d7*Wg3h@YBqpqvoXE<WMuepN_9K2sQLvCkAED%7UNi(;a+
zP#>Q6r_Q`wpDkG-%v^TjXr;q8Q-2Ju9al-L2wGfPQkC$2I&2KH_>wd;w4JEyp2whL
zI9e-oyyIi7hduY_Pomd4K;Hp470!%VaiV~RJmxel4Zv`AV&}3?3w6tB(QntDh9&@C
zcfh`0C$&gJq5V5N$Xq2FHJVhizpv<*+tsC>gLG#V_Lun`tM(R0#JmH~&$QcK3`(x?
zc%=*U2hnTbac>OPn2G1AUo3xWDc{?|^mKCk5O3&tQGeY2@JFT3-3%DKanKVLtmu3)
zxY!-)y*p-1tf?RzjCDl#^c9$71gh{8;t-MeAvNPj^#>(5EBdJkd7U0~-i!ebhtKdC
z0DD6_w{-y{+J{0AhD=VP6zn%!dgSg-LWEysbn@M6oI7~j+)5EI)knjmQHOOrFI{5k
zhb&Uu8X{lq1QH`h0@l+)x=FZ0V1C(+Vv-%F+`Hb@vmJR5<;eFh=#=xHAV~Zmf*<&D
zFXkuWq?lZc!SAb`x3vCZN>UT?Z&S-BxN9m@)Ss<6^v`Iyj8nhDP=6klFlzNXSpQ*T
zZ?(yp-?`1R%%D$|UiZCfjqi2jKo<&%*~muWdn~2EDyWaz0-9JxiY%#qRJ^#*d86d+
zY3cp3a~rEU0W}L0`MSShO_yqb)cxamxMO7s7Txo>rDWYOdyE_UE{i7bW3T|Zz-qld
z)ox2axZIUFWitc2Thd|8%PI1VEMOHGkFO8PiPwdMlp)q&Yi)D)8dfyWkZ11Q$|*y0
zn(~f@xBq6N9GF`p#TvKxB>tRias2qq{d=7Ip;x-O?j6!n<IQTrUeOtxbjbMBVDk*l
zxLl`jYG-{W!7+8B6e>2p%Ke)b9Rtm?S|@MUCNI9l)PJ-DO4J*dvQFcq=GsZy9%KQv
z%0LgzW3s(*1TV<*F&mkxh81-)Ui7-a{5M$gF~2z-2zV~YC4&8+R!kc=|9hDr>@Xt{
zv_DR(T96X5!5xOBM1eVVZ6w?KCx^FcoRbXMuaRx*s6P3K&T=K``BEKkSL>YBlFDq+
zT)@aiU>sMZdaE|VN_d%+k}~_7hh1Ry1p5!Qc{loj0V&^OBYT$yp=K>U$PlvZRDvte
zTzbU1{ty%4+xZmHd0**$SKf$;{QuB&mSIi4Zyz6>N{9l2bSX$nmq@38bV~l{j!`!$
z-6b(lDT#rCGzb$JATor>kQy;?!su?E{a-xytG(H=9XobiXMNAlSM=^D2+d$d=?q*x
z`+HDibmZ}gK7#phC!kMq+4Ms`E!GWNxsMOM%r6pXFTh67nZ90uv4HTGS`2AvV?tvg
zp`jw!!}BfrSlUTh>C*ASvFcYIt+Uh@*8&cfrl^N43Z`^q7Eq}O<RVaOn0X)rZTJyf
z<7E~he~<I~XZg+HjsGOApk{e!4&sOFeKoiTrfrE2+@U>9PO8q@EfNpCQKgl6n9>ek
z<?t&tKVcdEm=%OU8N6Y-WhC(E=En_ebpNy39)i_Q(ZW|sx|!X~J1{&_oav5;u-g+e
z_5Ry``<S#@mqIK)cmN}B#TF-#82hv0<^^ES!MC@g(<a#aaD$v^03CLP1I|S{YSStr
z>%wH^epNlc8!;4p9)zkW7BSfEy*!HUKm18T6sxK8jvCgrTs-iK@_~&|7ym>eD0416
z{tviy)2rJ=AuJ?Q)9o@oWgjQ?a&~iBzoeSLCSR?L0t#)r0B#2J`bLv=A7eUk4y`EF
zPwm@mitc$lkH_E<GrzwUxp{l$3O#d`D6z2@p6KBdLh&Ke#=Fr%g0`(fyTQ1Jbg~#k
zZoGpHHlBx=G@mDByW7Z4@4~Xjs>WE|nKGp^K@+_$romARis9hp=Hb%{*m>l>%R4A(
z><kTM@Q2X?ZQCEzhAdwRi&Sm(M;`8M46Qmxvz>;Q>QDa;79!@7)UZBviYp<A%Rg9M
z4ft#!qbc-jeUw^pu{WesVynoT0#PU6`aM155Q(XQ&bQxP<45j0{E<KnX|wS^9_)m^
zV512aI+&ZMUFdd!^0E`Vp~0SVyHPXZk3ken;w|5SoY<IU)*LLko*%808T}*w=j7rX
z&-b)Ai8YsBkTbMb%f_n1Wr%&DbssaQe%}Kxf&JJLDG=eygpfcG!CBzq7u2_5cRnsh
zHVOsWE(cj`IWgtkG6GssS~YkWqly6YKqu`x@3<tWxse9NkMEbRc8L1fN!QK_8n}PC
zED;3QiK9RFQ()Sr<V_lkE`O2&w;nzeAP!)omE~|FXFqilJ=>NaMQL<Zm##br<&F_^
zjyrtHKybr4Xz^TtynQ?r-tI2K1)^6KCEmLkvYrX*qJ`P_hdbboM<<b@%J(T{R~$rp
z!^8eHipW49<v-(=AO@Kg3dfvVJDDP`R5;+1PfxpdEmFIAWOIG}d{lxw`j<%LUP^9&
zs6tn7*xA3FVBxVmyPVx!{RTvnbjI?*P131--{tD$C^Od44%(m*p-FAo^_EfoUQ(<?
zKBc_v@#Q6IIwHdSU}br2jfueN{<yHKdf{mr8}~2$dDG+nMAy=d$m64Lw~m4&SxEtG
z1SZy95U~%tK24-8TktU9bi(Xgh#DxV8K0y%hAOyAxzv0Jp7E9{wQDAy({V6lS8Cd<
zS--kC#-T3<2M3WDr`-iAk=}vCu(M?Xt{O`-AF0EkNCh_O;J~Xbfz%K6LIuXSA#p=w
zeSN!hcu2Hx)t|-1L#S)2|0uh16!4H&?8RDo1zSbTaSp9gXSXo3v1Q@B^<8{o3F9%$
zt#Y5Yn`B9UV>`p_P0X>8JsKoeLjJf_a{Qz@+UmzkMB&oKVl&B|<G$HRG>paGecF25
z_qOu34{&ZWD%a$UTRU-hTTwJN$LIWzSL;ts@(Vv&Esjsi)!A_ppdvL@GGK&fER6nF
z)%A0?upX-&=Ts)&!;ezJenAZBVmdzYVGD!~2|vchz&N5AR9KP<A&>RUBTc<L0qxZx
z$2;lZteQbQ>S6hB#-M{dvc)WE`iF{^w?o_ec4&Lli1&vn@7GI8#N^%X{FRPxU8WtU
zbw4CTA{=suwEt0(P*AbPL;?<FbPgQZBg`1%*qKbR%!JIU{ej1#<Z*%Te(hEx8SQ#*
zZ}Ns3J#ntSZulFM=)F*O2cIBiftt<L+uYBb2`!DeJpPC_@2AY>Pc~0Aj5P*fzuzb4
zHM;kVowKH7!O2WuJx#OuIGIe4BW2->vL7G3tthjTc~Tusg5L)&)&TTUUdLndGm`q`
z&po$$gHc+-NP(2feC;=&bWVbUn6)?;JJG*H9He^$0>rz$lQ9Hzk2##u#o+jTN?KXk
za9q&5(r&M^@2Wg8B`m@a0p?`F3^2)H%Yicdnq%1|yTBfS0)>b(bpr8%SPG08Lx}(N
z8>t8ysTc?`Hfj_AWkJ=Qw0p?>Z4ZPgYj(OnCvnbSWc>VQO%{9*mlQG6Y#+6Ic^)ao
zk0r1qj~(C!ZqV)5>((a9QF>bFuO13Zs)<`<!|fG6BY=^WQ02oX@0@Ko8Ruy@NY!%7
zawMAjucpGX*%dpNybF&#R_1?52AeeL&%64o!B`e;%OHSF`Gi!dhY89Qi73RmSXS8m
zEW{ZLFsml#Id{DU_I4DBLqBzR>bs>QTJ_ZHmjtJHW*emuvrWQ@FZDfHS3FDtg!7lp
z@7`?;HC?0=!+s0vC0eUcj1pg%-A_<65DyLR=r3o-f>{ae%E|*&t=nv;(+_U8*78YI
zJ{-2D1pp>y4FSK|1%wN-m)|#S+D!_EUu)7RrE<CGCBY;p!IKf&1v7<dL#};a!`CS_
zQTg{1J{60QIK`#ZFT~KuSHA8-3o+_4sv;%TjmSBxXz@j9JdF}l#&E;&lT^%fcszBp
zZ9j};_YN7u$2qS9Nx`XC9gAUKOw5A|`TSY$yaBkqf9~wCX%<#LP7r0f+g<aQI!y@a
z9pN7m{;5Y$rn$t!*SFcdq{*8oXIWwQzI)+uQWP)mi?tD91mYw#bf&>n1g9<B%x3fG
zgbX1Q!0_evCrkfTomPCa`Sp6G(Zc0LMeibA1>0SvGqki*&h;dF^zF^0d(oT8w)jZ7
z@LOQ*$P&l>SVl9ENkSk$nbvW%K*fNAm?u>|SB%`D`}Ge!w@54x38Fes1@5Z7+USpX
zSV=O>HQ_ZwP<-?5&nr*w7@>a%I=|g^u)0H8#zkFj><9&dDCUqsz3%koGd{Y}0GE1r
z=v;^Q9<eYnZ$Z9}=v{^c4p`zcm$6WxLK-k_AObaa5#rielzD*E+7g(>O6ukFh{y&Y
za>zllPCAqv2PirY!o$oj?WjvslFPiA@JB7a97Q?h2El%-8Rhe#iRtNZ@y(^L{%zu=
zF~39RiuV|S{h5=KXd%Umrtm}b4AZ;q?r{$UogH|4eG(yHFjYH|`8Tc2Sex9d4`^~L
z5`5qy?X$Qdp7VrGe3G=6l*hHl!D?J6BiAm|sI{?VcFKJ$g@h0nWrC~Om@^Y8!p&Yy
z$+jf0F_}<^#j4VaEyrai(I(^{E!;jA4#PqMQDDvT7$qZ>(XLfvFCyiE0ju%sc72vz
zZI%p``RA@2RJ6Sv!~`krrjmpNz~qDHw7sE-Fsy*Wa5Moe7dM=UfWS%<VF_~33&Ubt
zhr-)7OUyzafzx38bS7*1ZH<M)O<y^zDMaH4ne|x@wSx7J_a+C^4iANl6KIpmLX{M-
zW4;xV^WD*c7ZuCN9kv`*hM#^i;rm<w9`tcd*Hp^56c#TQtH*Vab)5%%&yneBh%gEI
zYk2z)-&_g%WG5W#W{{cH8`wJJ(IrfzqOyl3qja|eVNB-<85>%7zE5{QNwpL5cXM+S
ztAJEg$a=<gARhW|yY{IPBZg{h1Rvw$n3HS;KJ4jze=^!wTklTv5|DhWSDUu(YJwuB
zPUe%3S172Q%>10`p9*l#J^E<(EtLzG-WojL)bM&zTEB?bU5M*0`Q6TZ%d#E))81eh
zNiDSXt7qGCi#bF5iWa$%d8yi{qyf45Oh~$zuWjS+^^Nm)GxI_-?zx#7+)mU^zjjZH
z#$0l+;K{M6t-oz7hhvug0{9a%VK_n4Y%_kLUdDF~Q&F#P`~e)hMuEW_jiq;oKWF3C
zpIFHMJ0|Ok{*aHv%iY?1N~H{2h9aY*10GWF%KDg5!7RIG_4qN%k3geZKK1(ZQf_Xw
zE^RA%dZD3)h6Wjvm6M-rd>!jZ!D7n&Uu-*iBrFkb9ZmKfC=S)~?|t}#eYlCM$ow@w
zeSW=k`SkF=W5Bofth)5l-0rfVKj;k3IETbUe!dXw-)9lItp=X;w}*z|_O?TJ`jkBr
zm+OpQ6f=lT5La2(iieLf$2lpe#Z}W~$x@{UEHA5KAJMrOZK?j#vp$-#FczW9WaXlS
zbMYpwp}Kkn7+S$O-7&BT*kfXP9dZNA^QN8$IX%k|;8-77Z!v0tGMx`XF!?DFJLnO<
z&?sl}|G+_PVjRxMWc={3_a)|a^K3p$g<d_SYEfa6BAxm8Ua^{u0K9sclG%-coZ0V<
zPTi-BoO^F#(D6h98R$R1<aDN-f91L8J~1BL|2LRK6HBNWxf@UORDhvNLMrQAByzn$
zx%_pb6fho&z74lE_qL|EJ15yN9cj|k#Aa7l|92nh#z|cFq26b%-NbT5^<TOMB0X^a
z9gQZB+uLKLqo{VokA(2{1jJYj%)74t*Kcp+mtbV$3_foLF$!=O{PoX~cuy|lw2-Jm
zoSYtrU?v3REIp*{3q6W_J#rY*Zv2Ce%gI<91L%E1$-8AxAmlyecZm|A{gd*VYaqhs
zduY`^(G%@4Cx{4~HwSwh8EB1`iO_9;Neupy{@vj+k%gKnbM`#Q8!J^><FVN!0V7j!
z;Am{et=pIjj%taiih5hO1=saRr8)OA{S?7YuqkeL8<4LscVmANUGWI?JTvmqJp(v_
z0Fr^_uJgZ6o_tsTy$3b|D~R~A{Rg~P89*sHEZ@C+<!<=#l(SuQPsiJ*=3|@xjvTwt
zWrznwY2!ly{<8Kri4`{+`_A;d`#U?G_0rn(mJRdW)QICW1)0Fkj=-a%(a83X4LYI*
z=+p2-pAs!?LO+DgeOrX^7ZlohqfZP#guiW;E<dbyPw%i9skkk!IG<yWwpFJPZ=Y>U
zQ;1%#mNzQ1K`{DPwJ=>Ue+@Xj&}5H1ni6d0`WD{T9S{}_0PF;Pr={aHGrLrrEj<1x
z@qxwBFX&fjx!Ne!YPBP%Z&VrwHQC^@O|wAxBgob^JI@m?cGsfTzKtfIIR!vJVi4xh
zw@E`fh|ia|z9|1wjyxK9IF?MZOLKB6G9hA^X;_n|-7L`Ey|MYGv9!WAoosM@NNCkj
zfwc>fOVho5!r#MMyxra2h->lsv3U^iHysHPp6V@fNu`1fv_qbSv|7nydjtETm0ylb
z$%ILlPAK1+RAA#0L4zy5<bjzs{RjA(i8xZrw5rrh+B?T(-9v)uu8dPR(ya7)16{yk
zfAOjc=KXi+0@{BW*a2|aG=(D<SFwQE&eoeD^^+itIV{<zweRg@2BZDi-t2}vyFvqD
z06Dz?T{h(dT{c}a9#vvN9zTRkjLKS^|C=>MkUwF=vmA4qFDRdS;NsZ217G>ynD0hi
zP6#6ktvcM{A9z$nsRbsuu=rLPJ-_X|(YvBuw9jTis2o0>_sNynxv$MY;rU`l4nwXI
zWZRIxmTW^FaF>G4=*cL_^F8%kxfTj1@FWhUk91$F=O27|@YR7Rwo*v@KDMkea4zo@
z0j9Q*!)s`xRD=Nv0s@;o;XUIdu$yq3FYA!lM4-!n&S<^aSh5&t%oQHOWYIL^?KG^u
z(|4O_e-#=M3b3B&c$ai2UxY9I*61h=AXys=uWpORqqcesl*11?m)O5+FSOgT*80!U
zFc25%BPC7aL8%f)2u}!Y|K|wN%!4#d!_uLLLeD|F6aYxk$aXq-v8lnMt^P-iqj8E%
z=9B^=prNl#irBXiOFd+^_E=_@?pu4qR!7%MF3x=89NOH#m0oO_Kdi(R1L0*4nt4N8
z)+T3BnK3y>_5v2}O9Vd1^nQBI-i*BqW5JcRP8dm~gEK4McGAg+QwTKDAjoLSRAe0B
ziS@_^V7<mC<IuX~j+0HuoP{?ig@7DEB+K5LOUPQ~F9+FSdBmm+7VqhGP~~!pRiK;C
zyzDU031_1`!z!_!jFLXT!2W%|O<CU@eXQB#Ba^3+%&yEuOCBRL#nSxsO#b#Pbi&XO
zR(bkG%Dd2+$zmQHvs|W26><%KIv^@=Ff=p?AUL374;5!}d2>MXby~s*!(fx4-0{*U
zerT!nPg>lgT)Y{RgOf3ulYT*$`;()C{1czNWgMHeE>J~>!6h}OCiXvtZO8{#{65!n
zRnH-MPtljn`03={kAfsa!?M=&6nSairy_31{eK&Snnq7{d&4?5x|?vsj0xF)yK<+1
zXVIbJ_(}0=k%^hZ<C~z{tu4I(b_%in&R+j(c!1(OZIHf6)hxm<LHMVz+uHTY4Gfc%
z)nwoA1I@^{LNKDcH+E!FR=LKw?{8Blp)X!HE92i9b@+Eo<m*`mUirPZynOYce8RQ4
z?_kq%5ycjgYDdps_eyiZ0K}1ri3c`zyris)o>_KCm(%{*=8=_^pI=;T#__GL;u{`w
zZA2pn-R?{rn$)V2wqzjna=p`#P`;jEYkj^vJU{c_U#K<p%F&rX!K7>ZTRpxAOp5eL
zU+7YXX;szcde?E&<1rq62&$q`GP`=w*{WcB$hAxNUIJx_ewW2uO|Y}@Um&Y=_Pl2a
z?Ag_|M)@niftA@38|jcP;p|Yh?ec>-O@&E8{4YfF8xNgonEAXKh@Kg_O|H%J^?c{W
ze&}sS|FQkeHsyxzO-BEF2*fi_eAF=*{Mp7FW=tPB_LepmQlT`zxG__jA?-7#n<u%F
z+&ZT7US;ixD=t_O;fR>60elGk>ty`4?yGk#P#-M(lXHuf8_5%pvlfDqz-6pTo$Hes
z4(+%!=w$-qxanxviWwkx;IcJpptS*>7sRCEp9XqQ%6qx@zJp#A@|3+6Q75KT0Q-u%
zCammZxJ_o_8||MK-At_1o&EN?r;_CGScW~BDNf5dm!HIVN{aZq9I$xilKhHKebXoe
z5)32SBPXHyPRGDl&@|P{|8yLYdE~LaN&lP#W?BaCbL<jsKnOjtMSYcOM2*b9jH#EO
zDM!!~=sc!YT2(OGC=`EVE&Az@I$4Y_`<@f)^VGKzDao-+lunR+;h5b<6atJ*QDMS8
zRTW)^n8Xp%ke9Je0fk*-mf3dd{u~JAdz3~TPXBuEu0i?XDx?S*A{vyO=Ff=tPuoyT
zzAhiqX@bdv8=xC96Yb~+INQR<D%1pxhd-?{TeRqdez&#lJls9rfdxsj{irn#)#nk9
zRf&`Dc)Esfr;;ka9P$;shVI<Ur4eA#=ImQKV2S=t_fJL~1bng<szJSpU(N+~=a-_6
zR$e@iAzNk*2h-IWHUE41Ko|bW9S4CZ`h86vTKQ-XQOstQtc>peuSCaB&9A74v4KQD
zcvBym{`_5tSm|)2Tr&JRdi|kQMcoQ|G_o$$Kg92_GW07uMGmiw&VUBp#J__r=Qhh^
za-m>&yKorqTYzHVRX`+|B3Eqa!0vNT&*dJ33~7`VPe4`!mAL;{n6aqUMh&Nr+zcj1
z7n(}gADQOEN6#mXsu(7?cRdfD2Bmjpk5qLaBe0k3y^*Kt=52#Awi87^rp5)6v>B9@
z<VDPJuRiK&)MjNVCg$<;Uso}f8Ky~USH{N|RVYL&0BpN@Z>|sXDp4jSE;ZYZ?|Y@(
zr=$@&*o$jbZCU2+-b=*kW{V@S%9>QEwX|>K+0CW258B4246jj&VIaJdR)oH4Rqly6
z^|3a~{fu^Sz?W%F=7Q{d8<Va3K_6^GvFr1Av>cQrDQP_b7PqvxX!{hq+ZPcY7WRkz
zn8h0)9fV*iksV9Xm&?lzN)`PFKLqi|_51m9TK|Rl%Y*CJ7e|vxDuq7++!Vg|pC#T1
zii}L4zn(5D-dE4cGF+9k6B)JmkU+tiql<}KV^YuIzo$!}qBTMeGE)C$`Dpl~bAIqj
zfN4%{HQcB5Zu;$LDf_pj^Tpdu>01<VS$YdB&H$GkH?ZBS8{Ws7`6pC4Mwa0`vi6Oz
zRxXX5c@|_jm)Ikk_*2%ziXAZq;sIG)<1+*y@$%~Eeu`cyGOhSm!;C$89sNgYlfQ4j
zjm5VwN<`aHpU;AZmehT-l<x=Tt<wXF=fX0>G1eF&zimPtf-+75>^CdX*jbLJ?|6ac
zcyC2ZLJcPEE<5ODPtLTR_)2ev#qQClS7?8c_=2{d39|bhZM2>$yFk9_Hc@8isXbd?
zz%49()@%0OV-k1CDtD2TK*mT*pYT3%j9PUI@)6jORhDEXBqg;eGc@!A>>_~9%w<Lr
z9l@TyYE**Fbg(Tjbt^?lLN>fa%Yl1Y>bdd^-P2DmEV^v8hu)sDIam)8I(0t%O7NBz
zUo<KNAY>Rh6PCR>u#i}XzBcqjsJzPO-~rq~oH{vy4d{!G&(ovjDiyvLpHSlY6G`3F
z_f^O_;$Q4~lShw*fc|ucj}UdzrpLETw|f-lXL4m}FSez3NxGnnZQUu;IRR0=?8@z*
z;eApN+#L>D`F@2*TiY}mvrxt+U>7~_T^Z%aX0aF2QxgyRGYpPpSEwMr=C8g!)Y`6a
zVjAm3>U;}x^TS-grU(0E=;L4GS-6l6bac<Z<KDoS?7zjY>g#Zol}>6wD4U*uy{;L1
zeXio~lY?Y5e~Vh1ynStYfVx*-S|w62L0vYA;H$=Z7CtqkBx-gjpzU!(x6fuh+6z_V
zEzA8zvN(OLC4H4=bDMSC?6jinot`@Dn1D{Jnm8Uonw6hQs4^cy&NSp-JMP*iJmYEL
zcX@GVW>XNU2n#cLCiH6qQ`^3QpWQ}=M}A{s$<R>a=?pvBnl+CGHj3ujGd}#oFVyLP
z!C=cmm#*1#()f)qu0(3ig<fz2E~snvcd#PU9d^YiM@Yog+NiVmN|_=5J&w7hC8zXY
z+oGC{An}cEpDr8cYrAH1h47>QM41-4gK_v2;PoWS(WPbK_U>-pX*-g>ICN)pP?IyE
z33s5t|6x2F$#NIS%*<+$?Cb2sV`uiA#mfz9(}u3LJC?o%V`?@eY5n0@6^~rByZd|a
zEF2LZDVEVhhpYHg<<47J5^mi%;2e+A{d;z%R`*|*jVzX3DH8w0rJL%@KBVzo=-!Eo
z+CDb#zlKOFKXqSLFTDEQFk57|F0!?B^$mBix^;Et4}|;4+?<Pqe&RyLksAS>pOg?7
z0aeF8vcJ`0spzKy-3deP#w8_FaT-0g<4HC7SM6UP#`hTGWAjlp=evjR+gfpt-s^bZ
zn@N$|t%=*_TcmR2_TO=l#V)bZ35<n1lw|cr0KE-^K+fEDZ7%Uzt?2!eMaEk58h7RQ
zGESDtz!NS$dSM>Uxq<Ye^0!pq-dH2!8R2>MVe8em%jUPo{U<E_cr+{a^!5TcV2Q}P
z>G*mWVwFzG7KnL7fYtLOMN!jF>dKy$r1Rn+C*70q06uz$+!#)AC&+7OCek_~Oo_!K
zi|qj`5vDLtL2s0&V3pTq8;G+{X!Uu|MolHa=5bxj(zCc<?HR4dQp1F?6lq1suy>{(
z!bZ5MV)PqGiaLD0`fcwO=4xA->6^6M@^U}WBrMi!`R6T?A|V{WSGQraI9DT~uc=l}
zpfluX+OZvA(he_HfyC>0Qn9+|i3bZ<>$Zi!C;4HVyM1jZ=fw2PgV=wXh<sx)Cw8Fu
zVFCxU*a1wIXV;6dOw86(9}B&MVn)t2A-gHyoi(l9Z5r0KDeV%5906(svxm>1ssfKO
zLIw=x%Y%iH%P=zP0rG6ZK}2`(Vga-x6U_Fka1$YiEu%r;ugbu}<lZ~zJce)o+C4px
zg=4cD9{ysRZpz2AGbgp&Gwsfg04g;l;@ysg)2!j@iB{VRtzKHkhJ!6kx)(W6-p9=`
z#V{;?j;FA@LMUQGG+)5iwvHOjk9PYGcU(>M`6E=5TNZ+BS=4hvL;g(c+AQ^SDSM-+
zeoby29Q?Q`*x#`OiTO3wZ*QEto6I*#%gOb0_Sy&t7ahvdf@%1%fwpgeb<0lw?M>HK
z%Z_NDKkCE69+i(h;=INv3s#Ky<n&hm{MniJjmFMp$ZHNW8jr5KN<gS*8d3wxAQNu}
zZ$AEmC`Vi{Mz11<&PV;2yivG{n=8hwtgJK{sCj8kBVXPq{$W(UO++G*z{znlKD><u
zFY;)&*a#Oecq3(;0BPL#`Owfd<Bn4J(ZBT~`x~V4HIi%0w)Qmiw8^F~AR2srwt);b
zB~*1A5AQ?I&(A+H#ACW=#t!oG{?0*ht!Zh}{pb3;>mF^JF6X18DU)BHj6u*Gh4ogx
zjiG%C_7jZAjiU^6dpw%BoL*d;p8Dt;gfZR6k(YqOAdd&oFIS8p1$};%H@P?0#sV(K
z-@5XU9lPELX?-LH<KzZ)pTo!Q((Wd+QuF?!>YB#8Cnx>Kx6u*Z5z7ikWb#F=F3mH#
zNXb;L`<klc)xRA?!;bz<{*RSxsQa*FeG9U5YyV@q#l8NOMp}~@bwr;LIt^>MyB&R0
z<FEp~1lS7&tQtb!OwRU7#TL!g>S%}?4Y0=L|06lniYb)g5GCFrrz6Jv-sPPt`ToIS
z-`yqTFl<p+$E{qbq5o9*mc91W$9$a2%`5ox59bdu^Ma|-Wy5+k;zmza)7_n>UJ=qd
zjTGzUePiP0eomlx@G7OoT_JYuhn#Qk4f^yea0|@t!`z}TT5taXSN8ZVsW0d+aZfjY
zfUwSxRIbG!q`)YL0QuTm4pM?SI02>x$7LC&qRB*<t?4jD^&nYpsnFJ%*_V*~F<A;v
z0M9EfOcOMhXXdL-ukHV9383%ULc+n97tfH>Wbv@)hSyl&cxi5aR2hmFO(k&v|Cx_5
znU{hsn-po&im4Fc-1pPKOAqHIp{Q>6l7z&(p+w_Uv6*~b$40FrN|K=R^i!xm1m>Mo
zqRUW5i;osfER|8y#ob8AiILe})y#kP(%v)0YT>*u+-%vXC=s%xpb-mlwoVl|9cL;R
zCo$E(4z-Zo-VP94o~4WeoePA*tgr&Kfo!ZsqDo<_FamCl2dqS}Mrqb!T2|S3EVX#n
zKsJx2Lzyo|&U!xWR5$^o3c5G)!l6)5gS5U~s!)OU$%<G|jV?4!?fYjxvwO6vYfvXb
z9ydgLZs6nZvlueVT!V}Au?}Cwo4uS*&nahhR+P&{k^P+nkv(h=mH{x?t$p`1_<Git
zsWg_P2<1~MW!*Ky%q*l(R&5X1?h0~yNP`BSD&E(W4uN|k-RHrSBkEIvQ^~+hU`1D8
zu-t~`Wu;@g1-|;kc@&&_yFo*~J9oQp&kk+kM;nD{3fiY37x+bj^T;l%dqO>5dn2M+
z8aD$1(etxW?JW|3AX|m`5E|+g6r-w|FFTwr(!BNPckubv+0kFBW?B#juXbcm0AD3B
zr=$y%No3W+ao-RR>h@Xu3UTgB!<g)l%cOlc!&^KsGPbtV4d;y}?e;h1Z?11_twzeQ
zYUKKwQB80@4FJ6~4*2d1;KqyWqTQ7$Xf!G-E*7vQEjIwBIm%lcva_@Mh}Eum9IJf2
zF5)Xyx`E!AM7nv;b2;`e*?T=$I)h7Js5g6;OnlwfFf|RfH8*v!o)JuKulHT(++W+s
zV`EGkT17$T(+@RI*)6=kpj49^NAe^o)|`b9h7&PFPJV}fgDZo}cyjd6z^UTn>X+h$
zIon6QJ^(THxMs@@T)%qk;oVjafk?!KE`f(FBQWp`{w($^{1?4SnRxmj4S#_lX(sH#
z_7PqWF0NPc1coJL87AuJKQTAAGnV#<{!95=Mj33>Ugk4Sjzpmxan~8UPJirj>+LD~
z?clD6vRc8S43icm<DvR&iliaohv4^A9Q4xwF*k7$HIqMkwkR<*6S;z)u)I_9K=2!~
z^v9s#S9R)^Sbw$j+Z9dC&q48yRk<H4)^Y`0RTG(#Szsf1NV#14U-+r0=B@|m?~6S|
zs1+6GoEi5^Hgy!3QLpNi$!rJ_aO-<>dV2$$u~W!yEsU<i#hsJq<j82<o${jR3ZDUn
zf#m-bXK6dvx<B82tU*oS?SK-?&wbBfh`L0y%ymi8QobLlQ)_#dr=^YXQs+>`<D>bn
zjr_mL<mLM{HtS1N7HMBOS>4@%NJBC(y(jv%GX!xCKZ%K0EgN@~J84E{SRV0$-K@|t
zsM4`n$69-}^v^0{nl6Mh53OTR*<=8(2@gk;j0r4YCKqZu6Rh80Zi=+hdF5PY?;}@x
zM`nheS3G+a5y41&s=^w}Nw59JSoE>%ix11`WQ0k}Db(DItU1U3;!#$f7*?=h^cZ{v
zMtU!jLxlqtOY|ZLChPIZNv}S}n1B*SF2hK?toV|goSr!#6aIENLzD1TdLlskW!YHE
z#nv1(mpi<yt+nl47v*S6PyIpjNCi!hr;`AJ#W~P=395RPj*rG?kyb9w_%22Ih51QY
zlQrCn^;St*AmWrc+0UHY-jWohiLrJiio2>RBLiJbQq{!V*51zA6Zr+VjUME4^F8Sf
z*`B*SJG~)W^l&;tI48wO;eb1$u2G-RJG(oPH^)WY$3u+huxc9-!kpvh<h+$<{~Y0!
zN3$F2le6pdyS?%yJJ3>(&pvOpJhshUtV3K)aXh(X5qScs;e+onE#hGc4fUDj<s#Me
z&HelE$gklCYdoN1(!_L&nZ(&hpGK#25=wAMU+AtaZ{@k8(k%0W0>EkyagEoDKAY5(
zwvXIjTjSBr^_*!^3_l{r=Rl2P;~t+mc3TykMsg*eBZ6jJEX|I>Z78ue&HqTac&>ip
znk&j7aCh<2ae@)~eqfygcON?GUxFwa`!Iu(ds@x3n?76#sd2~8G-P}I6;ppZE1(-^
z5V9Qj{2Y~rM^FLbXXGUOZ>(^Vx}h4C;U?k(t*sKAY9%iCj-Kd4h;3O%4MdNBt|YpD
ziFrK7(kTOl8+u|ap2~prYAdVNrZ2-W4jm8TP;b-5#u)88$8v&$E9ZqX@+{O@_ct)Z
z?MP$t*bxv>=cI!QS~+}g?=QaQMuy-Q)Yh9LpG^I-1<I8!<Ch4Wf}DEEkbmsVynbXj
z8iD0*|4J`i^wF@58?9fbs~CnHyt@vVee#|uhQon@i#Jb;(`Tm0pH!T-b2%lOoZO-3
z&2#cm=0G@9YH`l##cy4tP^$bQ$$>e%@>SgkmAYyjdG<QkNIKZmR5f{dFhp)RaZ7Gn
z@s&4*0*IqY55y{PNWf{uedP}xnSYQ#cwD>PZW#)j*B*`&#_9@-IbgkVKISL>4wdG6
z5lbDLs3JvtZy@a7mkx$po@ac-YKEK~ZQkY`i;X_6`qDV7-Jb%6a!Wt-xQu3k-PwqY
zb2Xex`;2@6xu}2uv^$0nefjTf>~^ldh~`A80QjGDd1JFf=>m@UkMcW+m{w*tJw#+7
z>a+Wi*XJp|D)$^lMR<Z3^aT3y9aCO{*pG6pYLd;ehK!B}pzLfyFyE@oir%>E*!jAn
zNt?c#^@|bi0=Pmdgv{xg<}xc6ti^&aRW~#4HT~mv1I#9><cN+wDy{%j1YYO{v&23F
zWNU-n<Q6wkR=Z`bC|O*}2)ztT96t|Cg%ZAZZl{13HX&rNog^m#JBnZ$>p#@*&amGT
zu|Go$zC*dG^l+c*f~P~QDdXF2*?oN<2zrJro<iGdee2JE%Le}q#D4AYo!D~C(>MO$
ztRdw>HR=3VJQwtwmRy67AFM^Y#`zqG;h+%{5)c93@fv$mAkZGGMn6>Nj(Cx$i2i!$
z_xHWQ`0<?x8}LWnViCT&ckhqQR0xhgQC-eQMBN@=cGU~XIvVSmvR1n<lBptMC7kuN
zNYd%(T1XOd6G3IKD<?s5qhTJ<Q?>FD0#XZ(sPY;eM&xHI<Eywwy%voXGIOoK(s@t+
z;?;&iIB!^CbxJNu7O{rF>!(6@O;m8U9CAuZ0f==MI`!h^56EqW%dKEiw(YC1Q#3xm
zk5ur5HtM&@Rv_-IobS)xY%KpLicnlwRQwk-A3xhgaqZgYFY}fpZgfG?bF%YYCm%O6
z`!;*Ux}Q=BV?Omoe0+QoA|DV@xTQD#(%)nK>H<IOFv#sA%VP6KKxtBpM=VV@M55OP
zbr(a*c-^j%g~^B$HlgOm$XhaH8|&RsN${sr9#Q8w`>&DbcoU?LJm!v3ff&5^h{uI%
zNex>Z|5PiX41PHpaUS1y4c(?f&EcedmBOzCQXEZ8OXcU4!kE>i0Uz|j{E~^0<oKb*
zlPSN*$lk7Y1FEu*WNuJ~08u8657K7Z!OofTW47?wQ}o62hG{G8&sWpVqN;Ii8Nt=R
z{lxF8^hERwP6`;{UnkI>(A>qtgGQ2sw1+XD86y4A{eJt7l|~hYe3%su1IN9c6Eb#%
z(}ly&uR@i3`&wH83DztFLxXqd|DLfJCrc~cd_Fvc1vK0&uM9%Rp*;88u3N{7MEH$K
zcX@wRE!}=>Cq;zyp;#GQ3a2$%<0#&Ma$7ed?aXtExV*oRK34$|R!IGBl@8V5Kdo`O
z?@uRDp=HNB$b3r3>h_uTZp?cgN~UC@QIaZ_Z22O3=x--RU~Tf#yO>yu;t7nF?VJq;
zmPjdxSv6Y>EBZd^rPLt|@^iA`pyUMo@{@BJ$q<Y?O-~DFkjK(>N!=&ZN#IG%m}~{t
zWk}DNA*_D2sa1&+(sqTWI2gYRm8#=)G38C+p#=$nNC^o^wM?5_sz|tU?r}Y7n}+{<
zB7taFcr#T3WWHh<gABWo*|EGz6=S9atF>u;L7FLM)U<B6HbdZk!EPp5KPDX(diIQ<
ze<nUI%f8@M*Lk11S|T+Uk1T-bxw>rRcG@*}R5KXYPw8cJq$Wycz89q*SikSc6BF^^
z!GsYd%vw`boWaFX7E4haTy=+p)<*3vc-_`dsZ)msc!DwJ&?0_7NSYNp;*ZD{(K_0C
zjDtteBvUzx3Ngf@Ej-KG5V>tA*<Hn_Q+wswSXx4=1OiS66&o4DQCVx#@3P{gpl@KB
zffwW>e;n!WgB}dfQ!B`5#@^#38{!a3=QLx<q6OIu#C(u^&Jx5Eylmqv@HT&~jD#nX
zd?dXMc*T%wLH<3vN~g(gAp`;J3*NqYua-5yh_Q|xMB0;Cgxx*}Y)zZ_v?DS*1X89U
zAUNB-ChN~}saxjKZ-s2@>OyuF<xl%h{?61MQh@wnWuXZw#E8M>+rRz5`L~kdlY)tR
zY`FrT30VCHML{$+i`}>Xyc87K^=n*AABKhjQsm#$ewe`{R^P;QN_PLTux^_67@Yhl
zx)Wr)Ft03Mk&ju)jw_z19p-mKy2F@eT-h`o!7kS;mBTiaj0zx99J6>GPr!nOefV{O
zTh{50&4Qi2hvk&bTz9A9XswIskaq9qu2H6n^kK%35Hl&%X@!ENo!!oHc>3p%o~H=_
zQ>MnaS%U!O<Mo0DbInSBI|lsjl4)0O3#;mXk`Li?m3eushpkipjx#c*+uF=pi?Qb?
zCsIyc!NJGdp=eTsz{}e$y~k~>t%099mUh0p0CANt9A-eEEEDKI$_k(6{JskoHFLyB
z$%#;^G)h18FKP5`@-)fNaB^lErX$({ppAII-cM|s5Ol^9F)NUZ7nHOK;VO>v1QzEP
zrwf!6WLWQLx)?P!H-X)T;{!>v)r=TMxhn92^NrD*j;4+lRZbc2LOIB&U+>rU_ULU!
zrQ9e^Re+PKh0f5{K&<0iEwrRwvrEKW#OI}i(hHge67DLMe7TCXHGg9nLO2kkomiHm
z0_!9niF6onCa0BQ+U?3#S%`3bg3ZXnX!-diTz`;zDoIBPFKJWbpmc!!3pb-;NDqx+
zjbXGR+5_8pY$xx7O=aVHe|LQ*=;9a9e;v<F4Vx9p><cW6I~=hT7x7vj{3P&&D#bW8
zw&4e;9K>Xt`rEX(E@8E^^C=rGnW2RxR)JWR@MW$ThoOPFsr$mhT-Pz_r?L#6#=M{Y
zy!0mhroPPp5AGP*Ab_<8PA4k!hypHvd5i)V;l<MVcK^KB)o<486VzSt|0J!)T}?)b
zpDTfTeZLfF6-ZML9UM$Ki-$f47sRQ#+=k_r$a0$(ckP*uTz@>^HXf+VJ6wT{&N`dj
zf#(%Wn>s;ca7Bm3pF>Ey@;#cK6}1RF+OP8Svgzy%ig$lj?5Ueh^yT4rxZoWa@Hnv&
zCTjL_eI0J~u=n|!6uq)Q8$UNFx7GZX_VY!b>&=S>s9uyGZV=+hlBAh5COC%QK+VZM
z^<7%Dv1ZI&KfhW*SGtvV**zG7ke$w$&3p-T?~C3)xDP)MlB=6xS=>38nrfk05&3#F
zB&5$LUNY{M)YH?Zfv?PPwbwU{qq7?j`Hqf`(1Y{FNsqD&(l#Far}gF02k8>;E`~Pc
zy$&+s$&q!sGx%K4^7{AK*4#x>hwreJ=dfeu#P4$ZIlL6R_$E(8G!GRtPw*_SB{+$I
zBu_5H4oEtUqaclQ5ForviDCuO#&iw2fj}36@(Dvaz8QhS+?)y7#`^r{c$%VUX=bBv
zE)Tvj%pS3QJQ)@q5LqLbB=ia&6#gtNxaWK~C`>2CN*P}l<4)$rd1o6`=OmwKGc&4=
z*mXqQ?m*7Hj0?|KPt-?<v>tt*RilBMx_tYO%7?&Z@0Ul2UHHCb9#?1qpg3P@0N>V-
z7|osPj`{8LZ0TyMsD&p=rWA?~$3h~WwszpZKb)oXbAzNxNp_XOPHk0b>2|-&=e%qX
z6oi;XNR{(wNACZa9>j3hsO8p2N(I*B4nRV1jKYmah1k{&FC`OWS+~at`*?YK1s1H>
z!0qH#i_LIF$_~o{=hMPPISvZt!R9E#%!!|8+$?Xy1N@GZZzgU9Qii^~op`cq%MpWZ
zwu#)=z@V*n5A&SE+K57xdh%Go`~;qPYTzA$y*!C<FaGOMLn%CjABI09ZB!$%6zMJ3
zE0Jyyil>z@XN8iY0j2=zk)_+cJb1VWfCi#ny5_f|_g0`SxuvBnP2zaT_@(ddKX|M3
z&QgDW%M?qxCNfeZHs4sg3KW;$yXd2l6N!WF_TOCK>~Almo9uhS{ld8#R`&)mROU_j
z2&FlorDY2DWv&{9cUW6KlhIkFRkdnXoSkR6tKnXP-)y`2R;!GOj|d1;VF0A0r6w{B
zc`&Tc{UVPGitxg~1a!3ykEQBbcthW;<n?l2M+XYE%Co5WhjD}i138)UlEOiup13N#
znQ(dP_<W(xO0#wa$lub%kemq_th_Z##3fmq&{IJOpEhXZ;c;FbMn=DkN{146<t0^G
zaPlwg=gW_cww6{_i;X&mfYCN}1J1MmJfh#kZ0J96MY-)jW^=I8qw<@*!JeA87Jf!N
z`sC`SE|wlo%W1NV_}q0e@(rm>z)Dc+q~iZ(xZ#h&;R@Vt&(u=`8Lly3^XdGbyeQyJ
zt#x0nasTM5&-v*CtuBnIl$oE-YvhsN>VBA%NQQN<8@(~tyTfdj-MoOva01)B?Z`NS
z{5-!XmzSqo$Cj38Wtlq;Z(k;)lkA~$Pt(QX*bix<;~K6Q;{IMu^zKuDh`-U6owmQt
zqM^xx%osJsRZ1%zrp>*$A9u>YlxOp(EECrU$&2rkRyxh&jRq5}<W)q!kf2c(fI{?Y
zEukSnK|;u|&|8NnPl<-<rHjcdyNKg9fz;4;kusW#(s_9@e$BGO8Rf&ey!$|y%y+l1
z*N|ObpKXiW!v#ng+ta7VO7+n9vpY8@z}01P?8v>G(}ZMHp+oBAh7bLi@)iESWz8Nr
zkT(QC0G0QO<?9!{AJYX<^P_3j$1Fwr854Hk)A8Zc(;(`X4Q<T);$Nrp1AnWCiS`B^
z9}L@{>q+lK?^WUloVSnQH$TjeZ!6T71#_#O*dm4xHTn$Hv{@zKuC22h8yiEAao2`S
zTBwl7kWR>42C4a=Cclr*d+~Mwo7a#-0$GARAki{Du!)D$!joBHXgdG_g?5S=)zAvi
z+i>!Zlvvv6njqB6K58`ZcBONvYiSf!*?uY<v2MV33L+46Wf|O$E6IB?Iu{?l)5dht
z47>&Hp(E734(LT->6nuTC}gp*oFvOSd~HIb#_#j}4H#u#)VxwtORwU8^O5jg!pEss
z06{oog1<vq4ObgyIdfte=R<E96^3e^Cz){y*|i|rY}y!17PG|n*ktZdITa3-c!m5{
z{|pQweD|y>)Guf?l7cYhmBn6;?29+bU{`yY7k|EK5R%^t$B!E3XB$6sib>oLvnd#I
zzzb2y$Ss<m^=j%ez1>q`oh0HgDgV5F4Cr2B4?P9z?$DAa<U&KX`{qM;l^1L+ZP_z5
z*<On>=~hSMI-)NH?az&Kt1`l!E?28Dt*`4ACH{A=L#iJ%R9-KxPlrWa9t@ivEw5na
zJ9Y}D%L4ICyk$IZdsAFS|NDrwinf-j4#vyMP{g*2hd882?sEyvaC}zZ30`(rahD(S
zHZA%#^7<Re_Er`e5UCgk&RZ)K|IDAVHQDxr1O&8nwf6nq{tDoT;D4$c%-PU=<Fbai
zYs>$Z6B+FydppK>0PG1AUma9blu5XMb^ZU2w-e7yyR>Q9-8b>$8RO>;EnRI$i%mAx
zM~_(9e7)^O;SvGY7$CZ5(K+4M$D5_{l;-ZLjGd(w8}*yT6MC}WU^;a=lGSwldN+D=
zb7g-rNG8Ok;G>1VtjM8jUxr}cx4FaCo4+S0g)#gXBn$mi<mu^UZ9MG!B$29e)X%%}
zbZzurfU9SOr_w|O4tIUCycM||`CofcVTZKFb?oPz(G?18!1K0_kmCzf{{>RyCd2-w
z0Bp3g7|}Upy(r&1gt^dclXyG3e6=mY=U#}y8Cf~0$}#HY@Ker2=)Oz4y-~isyj{+N
z#CM5;m94u}ngFRiQ&fo3md)1-_G^CNR^(<0!22}1-I_up?j+mawk^$GnBQtv-ITPx
zNgyX6Sj)>w<N&s3ZV={Q_e{A&g~Je6CUvrwo))yRqe+H$qtDy=WdT2FsI)kAP_KV)
zHSKmCxL%6H-~}q3Jv0AFRRH_zu6<$;kCtv=*4hkxd#i^%JxLRbz572iu$$YiriHLs
zDn=0(?Jq!>1+%TE^SD{1+!!<%Z9LFh0N3Ggdo5KC(8?e>_mx)Ta{gE>85@g!F?|+s
z{<i=b&yU&foW~)4oUd=|;(dwHK2WGHTw5ryHwx?XVHLk|q*1wl1ErB$toFw{Z?R4;
z{L=_^7cWkgJpV0neE_v)7j@3?>1gX!BQQ!Hn`z2~W5Bmpcr)RO39NgAk!<MPt~tt&
z^DaP>HIg{(<BOM(OO@(A?!OumX%-5NKSj+J81S1+!eE^A<ZyOHG%wBv9eIP7c<Fo2
zG%s}qUmEmG>tN2-8kgLI4?YloR~xBr%UB%lDLcem4?-a0c}R>ED(t}dv=O*D&=o<9
z^`b_C5QnOzpq@+3q~Tie)W->TO?Cm`e7w54pheucI>y89&8FAMx5MT~6&d9UXz#Z9
zJa_Q^l!P+mzYFsxwODLcG%cY)O+juhZT4&O_d<cXFiucQHzhQ`NMNib)rFovQljC0
z3D>YM-s|UV`Q=FWD=`8;wIDW<!kT&;4h|S?6i9<0kv2{uS12ZU;;8%CHeNQ1-vrF7
zdzI5BZ17?7KSx$mgf@ba+r4Z*G5-#6Q8N6kidrX!ORT>E1%lZ`#~%@yIA=w?0wsgf
zjmrDdj*0Jv$m@+x{*>{iP>qTmI8s;4AdlQD*QJXQidlIg&8cGwD9y8H@mJ4VZuH)4
zW>~E$pQcsZPR`u;<9238_s)PJbL&vH-LHO!iOo05zawhyEcS=q+TZLedrl+grs!2b
z-$!A1=@Q;VH)6d=ss8w4gST^ezzu2HJYPBRihe0<$JOp<A3q4a5tKBzKMBhdEb1t+
zgiDJ14*q-)UkqVtK@`Bu-Xwn@zn_(?%CV{^3V%VK8wex$%%KMYVF5*%;l4w)av_rs
zCe71RjkEs=i-cC^!;<zT$;jvOt%|xomWhFLJWWuzx^&`zy8}#a)xs2bM?Ft3Q?ukx
zpnD?ZS!%|Q`A9gp6Fzmm3Ts);@lQ@zvk5c2oBX((_ZhYq)h-Y$%9iL6<j7_7Y4!O#
zXPqaeFYYRY@L&_O$eaLPVzM{s6v9=s$?p<!jsMki9h8ZwJ`kX__L^KS(=<pI8|4KV
zb9ho$ep;}do3*NV<tihX=*&sXPDn(({5mJfW-k?pi@@~RzZ3@GKOPJP$vtBB)xdDl
z;|HzJeO)S_)Yq3Ws_%f+Y&Z#HG!JsA43;4k>*tl%@jmZ#7`1`c&z~n1ne}c~b?yIa
z@5Xz>_DdI!rys+*B{PM+e1qo3<N32f+C>~bo%;0Kd;VTk3$fak)WK-AJv`=5!nn4g
zyng>}?c|_g<86_AkSd#!jg4otk34~%?zma%%g^`(i)`7fmU%qJQjbKu!;gOs?V}IJ
znxn6ru-Hm`Q46HheF>%GpyPd#S><@#{rO~VP0eMrPm?B8*w)6*a&~FynYuq}*NI^_
z>Xz*A{0dodbA<P;R)&n1Q7x`Fg@gnKO6YsGS=|9$#E)*8DK<@K86eCG3zbe*1*dy|
z72xwUdBL%<lvMG8ZN&vBzNlY0Uph;ih+dhvxiYDU1c)NvrfY1QU8VsyiBL#d7TfMK
zjl3TvE%l(khH$cNjX=rbkfI{~ywUxA|K^#rYP?ksx}iDROQyX2;T_(}j5fh>SKL<6
z1O9_EKV|LJ#X*d@U6ek{lM_RR$|C#oKQP#eu1NpQ@?@TjJ5gSD%hJWpd~@GPN~m%t
zglF~yI2rN<PJ8g4%KTQ4^^KqdBkwy&*RWeW8UT8VGAk-9;?X5B8!Bt;`SMu_CBdIS
z^AcQN|MHcvjz=X1No6gF*&n!=i%%NNFzZ0Or^BnA6%h0jJpL0SEPgiRBw9^s@VH~W
zz2#qA&pFnNS;PsdQ@g0F%n9pez06(Q%L0MdbU#mw$uUCWa9L>>SuG=j!>-~z`nXW%
zl+k4&#vEh;9~XH+uo|av>VINU!Wp##XRz5WTF1?wu_1TvH|rF_9hD$v2>;CEa7r~L
zm1T=_6>2<pklm2gyxd$QYSp-~c$$w<L&E?3_6A9IyUl(zXul;88hz4#6$<?Eat0or
z|6~wB{{aW>;v0Ph(zk2Cc5r*c*UsaU2<&wr-*3$}JLNvYM2*^5uBK@#>gV$ZJA<bg
zVp2@PoT{YHiFJ{Oui2gnpM)e-4#p3MZ_0?0ehd4;_l`e^T9M1I<pFaxCtItP)lA~C
z3x$#ZXLY_XvF2S{2UvSan?Zxo*c<?LHQ%dx(sqhAN60~gjm%7TUzpxU$r7#*h3ct^
z@7sto*r?ytIf~;~(sT8Jk*dEYBG4Ri+SR2g7el<I*Mh~<fO07iplbe1ftdHNEX$@d
zPUa9)ip7RCBfm&%$IHav)ycVfL+)n2A%!#D9AiJ820z4JcRtH_(>nTUA_C}zaUHD@
z@(Y8N5ci1?F+~2H$)chyC}z(ALXaObrzkeL_!qk#Qm7~VcxuBzp+ea2C7~AHcFJKS
zrrb3>Ihiw;Mk#?L{?M(dQQ(KA<#<{>Q^uQ~tP{Kixcyl?aMZ7O`?J-Z!Jh}#H8?cL
z1tKw7OC6^Egxn61v+a}jF`V|qYw5?rUg*e{AXdLjS#djB{)M2Xiu%unbQy(rl)z!<
zt0a8R`QIlLF1^gsii+l1Ei(yj(My*Q;A-MV9r)uGho5rcA}$*`&=JjfI~k(D`BrPh
z<t5HqUfM2lg=OiAkHf-qY_|6;mNr7Q%z&?`$;QBAc5jfT*-g6t7~&h+v*_llRJ65A
zlQcMH-h{WiSH4<CEZUz>_g_POBfUL6Wwl%I{_nLTW9#edeSLkc(YJ^F`=xmsL3R(G
z7~?~?LzKMdZ_l<Q8n(Wx)wDcF=2Kwe>ltG$wo;H$1dIz6wqrAZ`&;Ebxm#W+uY26+
z?~hxZ9gWa_r7Wj#`=h8o#LsU{^*;YW*3FRe`O4|n$@xEb_1;q!HQQi~B5^nOe5s4-
zXKg&%3(?WoL6}3p#sPj^Vt1w$<skZJ_}Zr@A~JHprvoSl;1iaK@%Y8<|A4(|YaO2D
z7edUK|D)+VAK850Hm+uhS~Y8{UA0=o9<?_$YVR$zx4P}UckP<rnzd_F6+4QGqJml_
zh^<C=@_X^*AIOU&cdqL=&*SqsI=yriVdVUnnP97VKw8aRQO;*a*x=veO{Va{ThwIj
z9plW!MGnHZ1&}DoB48|3{{AGRRhzZCh-ZlE&0E9Yk4dFKRei5mi6?_#OJC;1(<y}m
zQORCVZ1PPIiMi(%^?I`P2SIr3x;PE7U^EshR))2^7w`$=1E`qg&eUfMwrveMD&jik
z{_3Cl2tSo6gH<e2r#^;EA~_EZZ<#Vs=5>+g3vLn(v{w)&PG+NHPIZ#n`Br#%w14S{
z4ng($1OS)ol$dRUhmGZ(;MSoa@9BxYJ>M-_DRC*aL8FnoN#OE9>S{_O4aJz|c>Ohh
zXXKy)Fl{QeWR3pZ_{Z@Lr>ig%O|WATD>FmaAP)A#I}XDsY~Ji_<5m~F)M|vO_7okm
z2tz%OITOWIQAC*8VIbjrlbQBRN*T>YESu2`LV~eGk}NT%eTaSE<&@Z0{`oX8acocf
zq;l@M-db-|WV5KiE(3Y=c{5FfXT&=v3S8`GT>Lmm>b|iN&*gaVoY(>>Ez4A(7+K4w
z+s>@@bZ@eSJXSW8^sUKg5Z3y`i0r>C+YC0q7*v8cpZxvBj0|QmSWsXmxtFzsMk&PQ
z`d{c+%!Jcf*08kT2tB4kfM_sVT}W{Z$f4~&pF0z4v~cz<xGFigEz^$M1nx&4c2BN(
zHc$V01_t{YxxKzjBi(E>%I>Bn8u`FJWcpOByL=YX_`aAl)~v8jB%zsmeoWMrEtehO
zxD0;Li3<I<JZdEd{oNN*v+61d>3}cdVRhB&>CnG$^UBm5omY&fUu=XqWi42G<Ql8N
z6F0M8i5AXS-u+6FFj&JB`7w=;yYX2Y>Jf^OvqcYWuBYY46Bv0qq6gJ~exT)P<KqCS
z=v;7lxqyUlX4U02`AimiJ(c$hF$44$3R7~9PGK>JFN~f_NJ%oWV~_yLN*iMS{E8{v
z&oM)Ebf`tM<IAOYhcdyZ_I_?|pL-1UCJ1P?&n_pHzn%(P-ptEB+^iC)n2ToBZ}U;k
zFUex$@#kFw7uVCp8r)lRJ9sSAQp9mt)6<9Ci7J~eA)m!Q!v65}ggk`7{{tRVQ#rJE
zcgqMQ_4P#sS6rufgdvRqbbxGTgN<jH=M}nXIw_OO<Y?q)xbnhG9oeW?tMtT1-))pU
zmHixp<^*OBmIF?=`>_!`wtQ`GP*C1E=NYln-GUG&f(LW3!Ek99$bS%UfZrT3kUeku
zH*TH=s;1=S$<M|vTlg<R3`kM_@Vb-=?t&yI569g_;31<ph{Kb<<FiPm><Lp-2e^>J
zS2i&H#3l}JH$NT?N}|e?M}%X83SVHxIU9r21oB^$DXz)nSlGm!zyzFMpOVGP^Wo@W
zq(CkDi68E|oSH1Ue7%M55oVo@{l^nGF?DNqh{IV1ZRO314RTN%lYNc>he0hYuLm=x
zTj4^_jypSZxN4W)RjnQI1NN<q+0P%*-yfmwiBHDH#~-sW6DyC_IO!{8jaM$X4BU@r
zndQ6!&{qf(D>H(ep9Rwxz)&8))vueOfcxoHu8_@-d!<^+6*dod>0Zn4R&vWPS5Y6D
zdX^Vp;)nCEIa>55tcw#jO){8w&fJ4Op!chSdraOEYc?Yep_+%N%O;D}XC%-(e#=`}
zC-9}eb=u+d-+0)ufrlp7aM*K=>TM}o^T$iTdxwyo9yfZoB6&SnVrX}B$6b3<zljEN
zu}L=MHcHZ|V&mp!gt;$7@EIkIfaOuCbe9jHYarFo;3sMP-m*dVt6!jZu%(7;r^l+e
zq_h|m6A&0ExusM&_B8|E-)yO{Dd)wXwv=#?)yuJ*UL7$js~J%nQfzX`Trc0>BdUP=
zwz0_^qg%{&#e2)&T_0}4I_E8s5J#7nxXNrdTo31IB&9{uPQg{3BE{2Z*HQ_aq2Fr1
z9WXMaI<-1A%HA#rgnCtV7UbRh`}dBpYtBM#9$M(ppO8$%Atq*QyeQUxnbXwR^uDO%
z;N?rd-Z}xlxwU^6Y0&2&ti%(`iA5RUV(M`8tsR#rE|f~xUxvR$Vnv!ZFrK!hHoSI_
z@b{*z@7Ya^N=RCFiUUpj<8577J(ZBgocn9u>Z(RB_V2mb&_Hqx%83GsNIV#3uzCU?
z{{C#TOSPh<R-dQOXE|<m{Bt88SuisQPY=^b7hnH<J(y09jA$mq=IdZ39DD~H@tk~`
zcwwuFbZuWs7#~&p>GC=@Ydlc6(5>zbivkIVjCR+kkgP7@3RIi>eL6p@AzvTg=D!pn
zqwsUYGtWr7=eZsOR~rNYM;}giF6ljb;qOdtpq*<UHuhmr=rAf>djraC^_ttjEBl~!
zxQnP`>ovO=SIBdxKJ~9YQ%j`)j@gACC7Xs~Ie30qwQ|Bpe}p^#(#R|-Rcy|_ebqf7
zSctps<ma@;k%EjspO*|67o*vsWfLO7kkPbQxT?$*dEbBx4fO|Al{4KM7<bp}pNix-
znwU=}mg&sg%CT4S@k+mJyV^1z^Vv0bsBh$*5!cqx&@CG-!>Lr%V{ronaRrre(~9D1
zirB4_$5kJ_aTM8QNbiM*4+eb~=TamsiU19!jDCCP*6S0734Gxu2AV9DBF1&oL3}aw
z*j4P|YUG1-7@9P%Hu6H1EI1184Jz_!w9l-T`y?Do&?&c?Gze0|dSf3ih3m=h+>r4K
z`;US_H<^$X_PaxWu@Inc_@s<bTTqaab{Mb9=H&81fc`uQG(WS@y<xr+RI=Merl9QQ
z$!$PcCF7{DQpQ$g!BUP@@@73b#K}VL(Rl+ZRLtgGHU~>^-H?elB4P`qn9Xj|pjRhS
zkowCQGvDNrx%gOT%n$6n2za<Cy*nA^x)m$=aDDv6s%b$xRU60%^5#n@6xP<TFzlWv
zpPHIf<}|y=K2Y=&>Adumn0CD&k}O=iJqjpAHS~r$*fn&SlblxBClLDY9--5KKPlbr
zC|McUv*~q*npSrEj_zO^;r-6#)g}Nvt<!%Wt6#h763}h&ksSjM48SyBsxiE_m-Hv6
z4FUd`AMEl9AiMHwC+10~Y%e+vI2QOw+Tf^x^tN+r%|-83HROaA)TlwEo^g0}cXwa@
z+Zw-_z~j)x6WyPAYEma6IeaE>AsSS#`PnQ*6!!)0)gv9!*l8o=@%VV>`N~;n@WW+^
z-_6`eQt=Ljp8v(^^lBVwYDuWz0VyV+d8-#3__K)znAIJF$ViAE@C77F1$5YAlai9s
zJw^539tqrq44fW%grZlkJS=(xKMEZcO`hGgr}cT?9u@mz{%Sk_fH5K5?Qb><&t1f-
z&cI2eAM96}yESj3Lj~p@iN4)^Tdie1hO?xPlOEh6Nv=0-GMtW4dz!G|u3s^gkSVz|
zu(AOn{5!i9JQaLK`aR|C`jQf~;^Fws5}t&ge&ScgD}Ij=u6l$Yk<WpYazQmT<!1-&
zjb=U0qnvyfvWtx~SQAc<KlqM?7&lxlniCg>i_3*apa6Lz=PwLy;;?n{yUxk8G)MB&
zwC){a8W@`@337~&8Le|^^c7(eoVtW0saimL%Kp;tk5$o<ET6W1b${x3qXISknWLYX
z?gN#*x=G<8NNd%`4)qTdy*=J|V?jANUAemS?=r1~ENz~>V5;2<1{PyTZ8~sgW?;x7
z3$odP3qi#n+!Bq270=&kI<bO1JvaBCRMg$3F=w998Ivc*PCcuI5z_Cpx=PIEl+Xa(
zKMHyYgfm}Qherxg36U@9nxvV3S1U4k+Kbsdk;4twSBC|$B(ZXnsJ(P+D=;6~ZM`?3
zNmjaGFUjBc6{II|9O*rVv8X^FM;a2&z+jW0x?rOuJJ&YWhHG5~&=_4>b{(<sMtw$+
z*Js57S7uKe0^et}-kfwNO6<&nfycm*k*d+Ik+HuihWyd)$CeoWim|tNx9jnk;;XtK
zc7d<9=3WV@xHs$?`|XrM;LS077Jj9ldwg14b`1@RleXK%n+)RKpxjqSkx_Uuvd1la
z3EXN=d9q);ky)HJh1gYG{aa6XnN0A6TZ+&=;%P9UTyi>yg=~Y(@7kyBee?nVU1F)?
zjeUxZ8@pT}$Fj*S*5${Fi_ONOqdN4IMhaJF3_22up1Zk~{pVj}@<IVoFCa8(-)CXy
zALKXvYXZ-`LlQlEepdQmR*K%9LH|AMRH+C9z;sVWE6!NK1!$nRZ?VZC$bYR9E^run
z+VyGQ=%7)1!6vxiDL7(SGYM#O&cr4tcaTeW$B8XLZ;8p$3R(|FTH(GUrFg9VnnRUZ
z-XvmZE1g0ARZ+>$Pqp2(FMijW`Jtq*4|)NY>y2|_Yn+|qPeMr2uV64ih|O9%{OW9+
zD7C(^juiFVj`p1Na=)6@v*(qWo94<txi9sy=Cqx!8z#P;RllRYwp#_@I?StT+b;wx
zfsN}Ilm3N?8{=^`nqy~Nt*{b!F+A@>s|!wo%QM{GD>!|r8UF-#BEc{bDN~V0UGPuY
z?uU*2(z~p+!<q^+)Q3EIXV!l=i>Q|m!(u^Xz7QW505!d6m5Srg#y`nhxqco9&FD0v
zAJYYk%YH1ZH7kLocLO&0)h-f+0y~9<<>iiViEG{pTlBSiQG$2F>-OxM+$5XO#b(Us
zHy>a^iJLVTnyb`9)WdCBXQ9b66>Js_;n?zbIX{;=_&Mjty)WJK$2hHEJ5H_J-rZ}r
zuz-QH80K3x<NeOCGqINXtG$`ff?m{B>(yEa>c+>ob7eba7ia{HBSKJpyQRSUzWj&3
zPBORav%|yV)*ioJOxU^=q}7T1w79w?NMn)8hTdE(yR0y?8_xfy%fqqKJ-bYV5P6an
z8Ke`xo@}gd-#mv<s~lDva9uvr3fsm2T*K<Y31w%|@!yBo=1VZPjmNBmdUbxth78tC
z_4HIWuj@GuVE6E_N9VM|K9lDUEKq?R%2?*t{|b&%)~axjDXh&Bo&17R^09-38jLF{
z!v82B43#FKrJ$;|)1KFuRDIHski+*(8Ro?gD(cks-WV*-(+am19oc4A7pK@P61T5r
z-%U7!fy|iBqv~klM*bY^&yo*MyIZ&|Vg@Vct%x{aba%$8C-8P?ujPF?*uZ63$F4T%
zY}CiT{=E-%v}Wz0|0CuV^=$%5a(7XB9Z_N`yC1JD40j;&-P&o@OJSmPRnkVtj@re#
zAPZ{0g|?JV1WwR-iO>sOPZ-_i^1B8`mH=GnJGX|b!LIdW6I{hVMLzhdpm0#H3uK|q
z!DJV3cx$jSB6&qUIW5)HN5#}d{LaGAg@V_y1;Y`@_EcP7&;FIbP<cgyJ)Xb&x9z>W
zcV-bDZ$fLwQFjM*5dp7fuJ~^8MI=uZ8x==8e7OzQ(*ajnp!8b_YD)1LOnFXfKB0?I
zb&g&0H#?JKJAM&~^!c1(z=pi~ruyhLBDe-<5aOI{VVunMK^3W-7SHO}S|Gk>bt0aY
z!()92!gV{nYI`pYEiY6ylW49;NiX4~X<ePkudq?lL0dhm6omL#pfAJl=XZnFd>|Ny
zoN=~;Q}S&8FKUQW{Z&f4Cdi|Wi_Hx?V!fCQJKSHew){Q6W!L4#VF0EHfUYY2Hx9(&
zx>|<(Z*6`gc=+Sri`z76B2Ms|uODvF!^2Us4`)pnV|t(G#@~r{B<YbXUfJuz?Q7+j
z3^Fwe&Phm2a{VidMNQ?trPK2mBuxlG+*>(#!MBxk6}^JL-ItkeDT+gqYyI=^m5QV|
z%=$U{#>c}PxbW={Jm0dv^gQ_@4U}keRoE<Rhc5iea;m!z|Jfy0vU<Hvx7(`il&sv^
zS`GO*9Cw8w{J5nUMl3Ts%vgQcp4hMq|DOJBqb%ORBu_8fle6f8IA4);3Y9_uaT}rA
z^X{1#->W_{SWe<-oo>mC2E}HqxSxR;6}`Vq)5x=PZj`y1(LuZ2Ky4g8!1Y8<u+%c2
z@r6H5_iJd3{BI0-qd1Lgv-bCvp`Uj@3;+LGU$Q4~%+dA&Dr$Ae$Pl_WI`Czcx^+YW
zG|O!H>^v(>U{u7!_W$RICJofch5GjsHnR8d6)XA+<+>5qpY%+!QZmL)!n)npzYu@x
zpwbpMMx_!pm^?M}l`sF}VJIzT6g)$xlU4cI6-p~euR0Qe_l+cEwtA{HoddfjDln{8
zfRtj$x9ih%_mi7iyqa%iBB@fmL-u1@Z?Soq?2GU^T+`2e7MoHChRb2|mi`xGYtsEX
zTdjhc*m@n^?TU(9s-RB|9PGP&+2)TAcydNu=THJ2NHRcLg_5PTxX}ml=k+QjRp714
zTHr&@3j%q{?ZO<wcMI<nvy(C7>Dgdm+!PLMf!QZnx@wkgp4P4GC(CTh<6-A(4^aaT
zm`K{{c_#*gBe;Lae3|T1w{W?X1*zy^XA-u@Swd<eiyK{EsXw>%?pqFgA!v^d^F3J?
z9|BaRf&tX3W6yGo9$i#ukHzi_RA1wYYwH4S5;svaZ%-qnKzL@dT=ZFtE^+uPSUv_s
zGRXq!Zd@@D?!7Wa7IU~dJ$eoI#*OXq5}^>*sei9&*`c&Fox5_jv=iD4X(yWNOENAY
z%_OEwR>{Ip5{FmTmPEU<*~rZC*d@6Xkr5`;fFR3I{%&nmORLd;;c<HD)fKbO?dH$y
z$_0sW9NJwawHV@G%osT79e)uiK60;FfjJXd$9t?a{<q&E<gg8j8VV%NFYF*>FC@M-
z7C3rxix%99w6m%vp%db#WO<H!jpf`BKx^;RAfo+PJ{tO~*ye}HN`5M+{{C;DQEeGy
zABgvvZ?|~M%*51tI+ubO+O5Dc0;p#xaPUc_uvA79#zF=*-NLN&PK~n#p}xh9wQzB(
z_7pF8W&kd{OWVwCmETQnEL}<*2Ma$wJP~@KT(N5v>JoH$uvc-rIsBq|s=RWb*31o(
z{oTA4tR?EydckSsF6y#;D-gD4e0eDV++mo{%cWu4bpbMg*$$X?_7mmjyMhTf9rKU!
zp#E}gNsSnl=B&oR+3`@r$j$mwkL_GqAw@@tEPA(k0ChAha5q+XHb07aHk9999vzw)
zr<Ekl_j?R_y{GrOMQ@%2QgXOq%g`q4lu#Lp+l}FIR0}1wF1UAzi-#$0U0ne+r4K*d
ziiMWSn#{%aMkwm!tc}GKZJw#Fn$9m2SLt$0Uf0Tw(Y7wXR=1&5R99<w$vw`YM0h!8
ziJ*+qG_|XwRC-_hxyUdc^FW>sF0OL1?s+}QXX5tkMVG;Bpm5IEkD$IKnIP{L#Fid0
zaFWtVN^+TRD|@=P2y7VNw|)iklB+8=JUvkeX`R*0IMRaZRY|Gj@?Xn0uVim_`a+ne
z9A`Ky^o)&#)x!2if%=B&#s14(OcJ3vcN}`@-rad6bom512L7ZFLDml0V5CUwKo&i>
z*Y+Ix`gsCB{S`hTZl)d|Xiz(N$XQUgv8c49_x>j7PMLY|CEPE&*$-G!%c!qk>o)4-
z<aDg~kJ*z7+8)u`y9#cqB+lB}7{6ZZ4;dx*`i`WzVxLg`#;qqzOn69BSKFRC)&}>6
z@`>Z>XR5`T>#}pu7C$y%b)D&*FK!<*Sv1Ogs-_lAS~_jnA1u!X)&>0lA4<P1qDM;L
z<nraYFLn4iq5Y}69Du=cfh&T(hZE+zPOIB}=Fl&pm!AF}4}Z3-1%tawsI->y`mSUH
zeK(FN#+T_$VMdR38(5*!&cClZ_Aa=%YU&_b4Sv#ies4n-{|%&?{;}g>gUa!I_S0Bz
zbE%)VXY{eJrM7(w@y6DMnKcx)xFVs?cD`omdvWwR6%k=I&jTMNc6$jF-B+mlozxW+
zaRqQ6Ku@89+07g);;rxabCb50mV!SnR+h75n#YWn0fD}I?qU5eMVC#fpoV4*gp*hf
zw73x-4R-MT-`{Pp6oPl-bv<*~KN>U>vSF<k#o0vnP{wdzxBaD4$&93ZW3T?X+9kf1
z;7fQ;#|qT_g)LCato1FbqAm*^H67gku~g*xrdX00(RaI1BXcp{=OU0At!|gpyDwoI
z{uiq^n*z)KDlR-3)Q5w?SY8!9FMs3&8p9FZ2K5db-Ku8HW3)L-_?Y5q4E>@J@~$t0
z)Huj4;+bbjJZ*rr_%Z5jcEIl)=6}P49>0V8-me&`jR0LX3SOUxX^}+%ss8l?t2bD*
zfv4-z!7T<6XiR`EYbl%G`)_Wtu+^@mc+|SpoKM}Sf0;yi_{8}qon+xmK3Rwwo8fUB
z@K42X{{F_}{*@Ejt@Pjb;8q#45)jjY7HQsD`AfpZ4q*z&d^E2n!AUzZ3$`HeVP}2~
zbA5kj1UyLuwJjQIq`IPt(^o^UM}d=e0rX$9`Mt8ziKT<{%Rb2<NSo7S`8>%>^1QSt
zIVNo`<YDY%bAIYXpZ6ciBa>*Vn=I9>5WpjHev9tPV%g%CQR&+&FgJI|d?9gBbnmZ;
zFDQnQPwcP`q?VcCh`-k+l!{$-Q`cC;`{Tt8Fh#c;&sSw*n!9olN**)h*sPjWFVbe&
z2m&Xt96iZP8u5KCPnvjA%VV~spj@{w;di)!DDe{A>t6EaS2D_H^SSh~`gbmHKLPar
zn##1<DO+)O5<v7en|L%@_WX0QLZ8uC1~G+GZP)VeMuq)`yX21OFd5%&s&(<l&S&TX
zx}t@cawA<8IP923vplWtl`4-RWr#}BwUOH{<EsvNMHF9F<`_2C8*Kb!p+{t_%uYiW
zdx*8`Q{clz`wQT-=K;A&`m?%$uGOM03w+X;!?xV9d_bGI?8x@Tx(F-5*HTnxztQ0b
z(cF@Rm0>X+bxiZ(;il6Jza|&?vJEA`iEa~@rkTQvd|n;}qnk;`Qy{ce_dHI0bn$%e
z-QOHjqlSYE>J=#io7s0WbShkf&o`%T<>4VxGviM|iq^2F<bucKqh$F8j``8^d}LGm
zHou5MTo~3cu`jd84Uy~vQ7QWNVZq~J_yO%Ha2CN>6_vF)jX?N<|GupMC@lk!wD$2m
zK|&hYA-R*gr9Jn5?RV(Xf_Z<ew=Mp#7zl2=eFlI>#IXaBe7ng$2W`mZ++;m+W=<0}
z9J5YkH`KTZGI`-;dp}WBd?=Yx2B!48Q{=G%4@nyfIP~jTCBz^&LIbAxgIJ<7vB5;o
zhrlm$cm3L1Qm)3ny`KNxMgWkm4&Qb%%l10?)@Z%NRXG)oXVlX(lpn-?9FaLUKIk(N
zrJLB|G89R{yA_f7h+wBE3nPI9@2ZwO)=6H5rR$n|k#5yDYrC!bDvN2E`DS<a|JPRK
zi-ELMza71H?>S-0+~A&MLoVAk-@!sU-%9~|r&ZED`P0TPNGX+u_<>!jQ6X}d2J#HA
zy*`4Dr?G;h$**9z@77pYj(Q_-ou52!fsGKaaQ@yYJSn@nna-=P2=I^{qksPK%h(<4
z``H6*C`?x97W3~6e`ZHFTwCAManVEiC+4ig3pQ)E+uQyfoS~Ac8XjE|1hH`IMi2A&
znOSi}4mtlAco=$0{REh7!K`YyJli}7l=xhy{JCr_X(@C(ru(57h4^~Xl9<mr<KTJn
zM}GcqFJHc>J}>5DU`u7w`SLq^<vlvb?4kPMP9ydt-D?}m%&yHpS#m>Rkd%ojYO(Y-
z!*~*^c{PgO8;VOSSp`CK#V)$%8cv|{EY}rrr1O>snt94Yvp~#b-X$sCY>e8JHGM6s
zlWiCHd0fP2%&uxGmqYU`#6o43v&mj6zhTm{sj0`XzFd{}Z<SMxIM!yQu~LKX&tg%>
zmCU-HBkSHZ(Pa~-BCT(ViSc|HaUHIsw{btZLavxn35GqwZsF5;R=R{ay4#|889oSO
zNDr)2f4nWv5&%ow!TJbpX`ldd;BG~Qh6}a!;h|m3YC8f{aSXdft@d9<U|cjaVV5al
zR(3`2`yLPognfHH=}i>ZOi3~qhHBKg^d8xw9lMc|*R*KZH%{Xs;HF9CFf_VAlici7
zqv0`YiqDeZ>{KWhUi~V*GI}eSETZVjJ}c}O>Gm!ldr=_}x<9H&jh>o83wLjqhHe&D
z=A~XLqXr&u^RYa;a<reGab|u~{XvnEY=+vkFnmYz{S&3WnryLV6C!YLW$nIWH<Rb6
zcyi9V2x9`?QhO&-j5+riXFm0m(`VqI`K!jW;L`f;CDktC4Q;HU1(#l1gV|9pp@L2_
z)|fw#LH4)rEcG#ynnU}WSgd?pw(Xxu>pAo^B#oAL7=*(K^F34+AsBkZLVsba!X83n
z1?>4z^&JO1#64WEp(`HtcD9WNbJnB>`U3VQ0{X8J_ALR|r?Y$Io<?LXEUJEWD^PzU
zNG-1cMe$?2gB0itQ|T}%L^VyW-v`dZ#schyQxsE}SE~I&W{_#sHCHv_XsLhs{+C+S
zH|3;pGjhc1pwB7!(3c+(!b6AEW>KX@IVw~(-*K^3MHyk67{p`p9aP+09l_2rLYapU
zS~YDF@u_~5a`l~7I3b!kJ>x3UtLzJ=VlHGDANe*L^fB-lhLZ_0pp3=g9~PWa<O7DC
zD^_BOu=sngin#TYU&vd>WOvVNz0K%^c{VIL8g`W6cqKK~Cobn7cC=Z>MI^r2>${Xg
zpS*i)WGrDNiuV&)?wuQ>Ak6tkkTCK;Udf7xUh5(u5)v$0m8~8(IydRvIn$(Mzizx^
zb}4aagl+7WZ*fx>nF)nuU+5Q00d1gUT7B|0TRMyN`b%Ma;$i}5ei^f7-_zZ~s!M24
zQxA!BAY?GZMRHte(eJNFt1?zP>w0o;D_nyL#~Ic$TyY<ItH|+RWjcc~@}uaJw+UOC
z<+KjNJ%wgdpA@@HXw;e>pqQIu2=a36*5kg1CAmcUL-5|jc1!JPujGdtO+p|6M(Z8^
zUm<*~o92A2=azl7)ZLBn&rhD@KAlNkp;(Rg&sLIdV6I8CQA@G?A-6Z%X!h@30$qWE
z)|qA!v;9+b*oxK+Wga&+-g|#BHQC9(<_xXn?wv_W0)`G2q84~Xog|xKaEOK{B)sJL
ztCe7&ZQ%Vy?XpyXbf3gQoHT%IJmu)rw`XY+D!b=_tm0goyn(hwj)U!!E_nNwn-<%|
zig#8ZHIl1-@yCBIkZDZP&$LLZqyF{Xk_P=kGL`gtZK_)-ynWv)FRS|$YOvs!*~~tf
z$sS=7d^T<DV_cJ`JR7s+(tpfk*>$~)b~;02UgKX|&(~*90>QkKfUVTW=so9-m;3{F
zn01sRivf6A<}LGY2d@r<d$FGhS#D6DMULvmD4Z@mM&5J<CTW}e2-8Tz>O1S9${?)d
zz=5j`r4Mh&dr)j*rk9DY@y|{3`uIFc3g8~6KAY}yq{DoLGscF9GE#=2YA<BBj3tE5
zuK1!Pzhy944aWoszS;#&xCX_^gfpWmq*u1<%AK{Q$-$eE&%s=v3U&owIIW%8C+yu~
zKU}ZeJUo|>z+XCaNZ&;vUDo+VNU=x5KoA;))uAMh6``+Xh9jtg3*-8<uU6pr$W5z*
zyZ_ScVJ^a)#!soD@_mA?Dw<~C3?-Ap<QQV|w&CGl9SY;XIV%mk*qN<$h8R08<%|Tc
zGJn^Zp;oAb_V6(R=)%r&#DDJl7{5;Y;dN96(VKZHbXejs4)a!8Lp}vBh-}t$#Tln$
zL{CC3PAs+2b}75=We1#onoe~jM_;R<Sj@T02PVGT*NI`$N&4Ttul-Uui?Xbvvm2{V
zC*y7e13YD*kJkBWBGbVspCAk3hYB3Ydu<}|MG31~8}H>J<jGhgP`x7mi3E*?C&(qR
zq+3`hM%c2(+OWL&9O?Ob->s%EjyqoOfZ^?W0vk1*Q_5by4%ruywqYa5%`sNHWHr=y
zTS+h0G#Cr)7C|m#Z=TqehyO@mMLA&lG!!lZ0X)FTn_m}EYCLCB4-c9K{`5q+ztM5c
zA!}<R|Mhle68*8HUoj9-;p|G+6QNDD+Phdb>TK9^H1ozX_ryhK_z}t$M5Fz%-wvF2
z-aD4$Ihc^!tT@Bi$%A*3TP@~IA!8x2gZ_)n#N<^fieL2G#2QfpUS~?ej?z{$MuoKt
zK;ICC3E)Y^vGhfj!Fb!hZ>iFxX})`Y=luL&*=KR31K$IM=vrN<ax0v6jo6|K7h_~C
zTW8?T&LkMZTw~`&jS)8)CU3oY@D--te;~_(?1x_E;MrLB&RPurpt%;+2yw88@!E+9
zfW8+qOVy=W(IA)?!$fd9(=-GTe<!4Xgw5Xi0&OwVylv6MiUrrcB@8*Z$9sFf!7O0z
z^>}9x3l8ji>xgYe-VF6*x63HHKfsi?`PHzW#pJA*(xY(GfcPISJ-^*&atoLLb@SyS
z4YMo9{xwJjC}$<K%3x6Lj+<Sf!5tFC)0k`S$lkH_?-!?cp<Q~szuZsD>kB#nx8;Ix
zxkPXZKLaDyuMyW-W)kT99-82))p9M!igIM$Y3A~(+9Y>Di!o-Sa{ztF{E#3Z0y!I6
zB`xY4qXUIXNSb+wsn7_7ZWeloL3%nn2bSG2E1R3*Rlv=`R_O!tz^8U4Yeks&D6S8T
zHNvNam8~3yEcsjbtN-2_-cJEE9*%E%WT!)eE>oKBlP+q(79Y=)`U5t+E#2az;&cAI
z7d+tT%g6f`=q48;&+;xe7Okp=s;M|QKi5?H#Px?5J0c-ng&@kdm1jRyL$AG9$X+`A
z3eBGH*k@Va-Iz-p=kko_@p=kAq_+e1dUfc_!|(0q5pAX(SVAOQG4gOAX>-K3s~tba
z78hzVq$|YQi^8jKIsk7}@ymQb$5ae&U<lz0{xWq=O6_e8EAWBSrBZaw09vOMkeQ?*
zod~@+KbbF4KP%@!D&tO2m^HOiMol;$vczZG-T9Wp2~&Pc!6!5kE72~113NwgQs_^(
z)dh_=`weSXa?AN0NPw1&LL3q-fBI)(w_sYriqB&8y75&vhg(W>v1oj<kTr3@=&3pn
zn;_H`LiUl20xOCox%{2mykE-m-TEt7yWjKVy%9ax$E{!)+9wW_O-4!OuF;gZ*75c2
ziq;Kc%V3Sj$=B*QRGBzVDPo+DqqvEOv1TPnw7sBsq{FH)dO&`qT0zFef+NzG2LDlB
zW{{#Hs&~OZZsM3p;&30n)9xlwE8)tBlmff;`6vi)60-=gD34`)iW4L(7gRhpq3HCJ
z4!O}3JH(tjI%OGCM)mRhOib+h#n6{)Dunae29sZfpppt{#2;T#gZ#U{Gm3p=JyTUU
zmct{6yX|;20tC0N)>mbN<jl#2v;*S9w|@VR!89NtI0F2s0YXL1vX$Oaa*)5<6fK05
z?3#+feL-zmUj-HIr-X2!<5`qP)&e$-=N5-;kLQ}3$?!3UTTG>uAoOd%IK`B<PJO#t
z0YzEB(N3LMT*bzp^QG~VC~&%8vsSrv*d_6}gIXCNxk`69{uml1=)3+zMTlXUZ&Qz!
zd(X&Cvb4Rf>(9?UYFgEi<b@y;)gd}>JHE&54W?VdYdbEqijj<31OgU)o@h_Xa%5FR
zXGblvx-~cd;)j2*jBK8|d6)03X~^mM-nK>_qQ^(2Rt`^op+i2qs7aJBUNlH(-{EYD
zn7jrcpke${JGB{Z)g7p7?8l%2ds{W_ZDyA(C1Z7<Q=>KC*jYtbUb_GXZ}EaQ5B)<r
z-bqp-y0-5QmS@ApoFBdoEn!gH<bjl2NedGaO_Ev0W~=SWVx_ltPRNeO<?Hlo!Dcb?
z6i$;1ycmvB(7Zp!j3F&4yY?QQ{Oe?MG!BT1!{G0JJo|99fu?WjJ%@#$!u}l(Z+qV_
zPZt`#6*28x?!OryIB9N+t(noj-yXPg!qj~Ui~N!g=jxf47RfRm3Gd*Y(U2&sx0Qw+
z9$P0X^@N39nFS#3rC;n7_i)uDdH3@blTJb#I&-}GojX(YHbqEdR2WA>Uyn!9mII(B
z3-K=qRUReE;6KUlK(tuqJz)K*t?LfIcrs7uH$mcIPe8c4SFEdBmg?(Sk^&pfNIKEf
zil|4=H)rZGDFOEt@7+d=?!fDBVX~=%!>-MZ4J))ngXX*bgX9sKtoDyAWuJ*3KttXz
zaW(bde-lgfnC`3=fr03%Lz@iGIsuOv`zyri%H;M$M3INt%t#MPhC!30uG@r943nd9
zMS*Px^RE4SEGn`gnD=3@8u;lkF>MVt-;Ke$?3#I)Lm;Zswy$pL>W&hdLXi)(b-Zit
zbZl>gyI*#g?WsMRcNeREUxw@zGNN}1aGxk{9e+M7NSvduM;WAocp792jL{0iMpDuA
zDkrY(M{Or0;Dl<|*`Bc6N*78cSHY?Dq(gtRg(q+LD-}M`ZsLN*XcJ#+ZGy-Dyw(U|
z7+*?_8rP?a70%|y%-JLU@Fs6o=CMa%E>+JCKUR+$)9;v5D(S1nOi9{~jm?uBzh`is
zG(9|Brb>B~^Bpaeh^0phoHG50SbKa27CcU`>H~wiYoA@TM2&A9Z0-4X87QQx3`Y&4
z$9?yUJPM&9sQM_>p|SrrvtQDb$&$OtqU)Ywl`2@r??<IpK2Fd%VFI_LZ^wUPqPW#h
zTk5+pM-HiDNfItQ&>$9YVGLXluwx3UXQ5d{{Dmzo@K>DswJiYr*&@Xw9OaVqlj4?1
ziFFFH`t_szoQeu`yqV9&w{1z%-1$l4nNlf2aN+qMr>^279X?emTnsn@R8iKri0tL-
zm}o1!Ht8}MxxMx1yPWWNxbztH>V<ggkBQ@{2gV|TSgVwlI}W$=ECbu7qh2$_e0bF6
zii{dRKG*`&Ut;<k8o9J=<>`+J+j-8r3oy({Nh>^IIlD%NWYQ~T^y|-)ehYA3WI(sH
zv{kGZ&?_74dG~AIKObMOsKLmr#;-C`G62DhfxVwp@~V@5{q2r5>*BNb*SQPvUEzd9
z#;m?{nnrK%4AO5hMRG}c)<P!#Tf8^GyMUN;cyH)r32<#KlEzq1TMlwp9GhPbBfyU=
z+e|&FM%ijH@D*;>OmOhOV?72gtyR{ErE5M=#fbNN329ocv(})dJfoN{^!;jK(}1t<
ziGI`KK**K6MXp+=Ian(6YBhjKTE<Ad|Ew6*dwxfhrlWm-pfm6-o0!ed@1gdo@7c5%
z{wP`!K=n;}^u36emX^MqN(<ZAbbXRMi-_4e2*p5;Vz@||I-dl;koH0Nv<&{p?93xk
zX<xb|IJ%g8Lky=EK*r>it7eDZU7|yDj1Y19DNNE}9pQGwKg$6|j*TkxBoKtOAwpK1
ze3`Foc#^#}SNdM(qd~QZSN8jTzr|dWvPMRLkym?Er^IBdVWF;i;*Tc<Ufzuh1(yr0
z`VJ)K#NONh0!h#?&r?DUy@G*DX&{5w<reIp&Xd|HzhTf4MlfgGAuz>>ZhHIH2X+wS
zljq%ndS|f<jy`+G?H>2{K%y>Zf`(!Tk7KG1^L0x=5b%Rp;Ixawz)~E(3Is0~CIb=8
zVQ`JC^LsR@p=eha`eyvV_u=$@d?{HxP9HL%mG2$14A-)mvU<aJazYs->}R)J5G3%(
z@9pljgd&MO6QgWA1C%@8Yl|<iM-oL8Ik@y*rGpjAUK3=KeA~|&9ba?>@9Kb5>6>``
za6zV={LN%pg>!gQB*tY>sK>I$0_dq=%Qk2_Q51V31)FVh3~#3Jf5=245eBWz^eDF>
zFh7{hq@vGWN?%=#3Tl%-q9#jxtIqL-C>x&_q`*&0``n<c!NJ@Uvv)zOBgdpF$jP`D
zm6X>yzH#l{j(8tv@e_ha-TUfqJZ+U)u9EbMdz@Og810kHDZgcxtn*uQ;0XV1{e%XF
z;*&cT#1>tAv0333fa>!p@L`SquQqi|oul{_W)J#D(7^3rY7i^<Q4p&2{&>etfU3G-
zEzH=12rCV@XVF9P{GNc>A)&jmr*Y+{S%aC}>*6J)${+N&+8F-?RxmcZz4?A4W0m^p
z<*_v_SD+VaIe5VmIGm0GX4f$=Tg2^AkC!MdA>6=EBXHwS4C&FT#(c7m3?1nVY^FE>
zdA(F8fLPbOyOe#HZsF_cHn1U6nUJVKDyzf3S|{gp_lIuAP`gX~RU%egkh+5fsKj4a
zGo-~rO{i3l`Lii9naI5w$IP(#>Le2EK7Et063D$yL8=h*Bs~XedRnhB54-sx&lxsE
ze}G2|VBJpb+)}?twSs}hySMjwZ@VU$bI-n>win*J^T(jJ)=%37)!)%>O`csL1`fY!
z;y<5R+VXt%SW2>Z*TL_^8^~zV#XW;G3%%B==+js{gDn`2XIUCj1~5y9oNu{H9W+f|
zoa~7QMXSn$o&(%Zj!(DifPV{3FZ0A7kw$%Wkz_M7q!OBFIlgnhNW1R@?y5>}Y>Nw*
zPO~Rl`_NHDO}#g_z;Vdvl^uI`!0r(Z1~0v|XTmTN*{t*7RK<XT3exv^GGOh))pAaq
z9_`)K%-w68-m4)3)6Sl^YiU)qHRA>;(2%^VP2;AuDIjl^?Wz-1D%#K$n60I}F+RMF
z7<6#y1zv`oRPBvd5NwT?o7B@gEw$F9jC@*nQh`r*G4IYm?Ar6>_SCoMFC%A$gWn1v
z|Jec*<6~pG%Iv5M=6k>EWXwea`AuKB;rOaFpx?P+cTkrWBDBNAZJ_>$I1{UvzDjKs
zj&2+H4eTh{y<v*1D>{QdeT0-+SZ*U><7wyEl>?C|Zt=ab$IA3MFBvmEu1L3H;A)>w
z>XqpI`fz?25Yp9m3HRwN`CZVN^o3^1SY+g+6bAb}es^JP3CjfWx&6l03JKi%G7tIK
zDWC7!YEjfU@BxKrv0O|NL{@3ylA>h0!ypr*K;X%7<jXch?i70cwkxfw_o}<0pamDJ
z>HB`D(e}vtUkb3kxuUs(U6Ge%h+FT~I$sbf<B{M`go`(O*M{>Dc_3*?5Eian8*IlR
zr;3d*B$B3~(>%VuH2B#^Sd6(kIW`eT{_D(jyP^GsYGIOS-AE$2R49vhOi398LdW3+
z<X3kHVwxK`P&yVG8(MapPYQQxPI3RgOwg!6Ri&?PXU1krPUH3c%}B(VT&4)j*<LY4
zerv4A!1ndnJ9}L?g2=6p5SvqpwkB#ex_-X;4V$1ctFPZI=~kuefzuMF6^+(5_>--<
zS~a#}YRRApOVkz!Ptcy2fs73hQh<Wlj&|fyP_%Z%@FE}UZy{XuUUENdS#(7UzHZI=
zDU-(SRlT?!<>Eq4clc3_fpFi*d7N8qc2oSDHn!YfhGQVPW^QII$noH&aXPGrjASBf
z75~eva=GBQANI~XsINa$9ybJdNMHRydW7vHk*1cqpY1HULs&KALTG>{bI8ik-h39j
z8l98Anw!?52<Z`G-oXpG8n6aDpc^46N68|QuTUSH>d&a@c8R;P+XKAkS~C<<rLjZ8
z5XGfj#CWzWI8J<lwu5+t-b@kLJQaeT@Zi%=`tr22cDa+{tfiVYD~H~qR-(%+S!%$)
zgHVi(!{(QW$eCzTBp^S~Y2MT{@b9!ZE-gC~ZFzaV<;5_nmPvv?Jk5!!@nYNFK>VU$
zD22&E2fx}M+}(wmwehOEZq~xKwKkR3<@c)#n?xruv_9cnhKT3cdiw3tyaEKnYJed4
zSE;b)=Ir{vEqjY>c!t4jb$y|BouI2P0Z2)Y9E2qmbf;+yKm?{Qm9w_Jk9sG|!6fAy
zWtgsFZcwu&KjLPq(2LS!mC@pB`uCG-%%$D3-LTb^PdQ57!o6nGWzjD$mEWT8q$;gS
zq&~OiVNc+G!I?H%-%7IR?m#CDaecIL5;Yzu`0Sg;%!L56Hpa;%aKBZE{G+G$@-3Z&
zEn?hKSV9H}gS&PtQ;uDy5Y5+Z>c)CvhT1=oT5)~Xt`%&~Ntf0S3M-;f@-r0TqXn@4
z6d!*NPgt;hvo&e^cF)lSb8Ws%ekH1%6@HDj@qbOnjsHm>O6zT@{#hY`xHlmFFQ~nh
z=?C^aG-zvC-YKO6mbnPe;9AEf$1RhyC&uQ$VZ8!^U0-L@W31ct)NX30s5^|sec%r2
z@ls-kz2EC%qIS9QG=&(Kb7Lv=<08Z@KEe8R-wzjgHhN|z%=amYw3JD<SoUT2Gu$8w
zeL*|=&)5n@=dF(Ij>T_<|9#W>5Y+N8OuK5JbsuMSD~45>J!eX@!e+E_Z!f#`=v5gw
zz!V9XqLk0z!Kg8DR8asTgoV|CR!M)N5s4MvTuc}BN)aD~yTwYZVCScoD-2TJZ0>;O
z^K030(h4fCi=T$8IZC~IQ@zn4KkvQ5<C_@qjq!PPd-a*c@X$e8Y*bjTqA;-oZ7p}Q
z0U5px<){gRii63sx5AFYs;aC8WZ`m(lo0?M7H({^xQb0WG5%B1uSHFHNXUw4cSCo_
z_SO%{BQ<Oxryrv0`|Rh#VWELw8L6U@k`e}HSzqW?2nL%j%cPZU0GW^EV#7u~qQk}x
zJY^c-`lX<FTGj&#LiMegC`!!JHM9%CpP?8ZIVB6hqABFd%DEMwGdq(6SIm<3MSvP7
zcI=eNn{ZiAWr5n-zrM)6Gkfq_1MCTz5H=f8e7t)c4=2_s)NNivF*n9MnbBpY(qVt^
zk9)_<4*VC1k%BHS|5npg&UP%;6=2A}?ApGIkynNvI}5em8l{H9k*se>f9#RamV>~%
zJ!aa-Vo>(U))<%YU&-#PV?8Tg3m``izR3mxd!=%-Kg}$9bBdr-R(PCCqm;*KLq?ls
zBifB#_wtakSxx}gP3m*%A=)%g@4u`upKF>~{E{VOpxaf5L9O1fsE4QvdOCV|P;iI&
zhb_}<X34>Wm}+dAh0tLD>iExI_j!C>x%}wd2D^4v%ih<@df67kw|-hqfH{LQvC@*W
z$4jSm&7NOd@s#+pH=*};K8rKENoYOOPDhUiPd)ggA9*yZ@Xq69?N^`L&Om?e+}kZR
zQXAhjGKZ4geOtTh3KP&U3=DL^R9w7@^GqlqohzL(*W>f8^W4PggRft*ri*?i_I;}X
zvkbkzSzTfF)k4S2e1Ki20f&iEMOu|pb6?!N3667QZxfyNpEfZ$v-w(dNl$L&j~w}W
zN$nHZWXd6Fs?0nJ-({|QS#&Z*3lZVhi;Zgz&v9pe`ui+BF)rel{)fAN8QFW+OX_&U
z?*b+2@bPsw6dki`;eW%&wF64LG%B%~guw-umjvT&Yj2ZXox+K#u>cFLSZ!?>b)mwT
z;KZD~{c^{wk<XST$c_=49GjJOmtiObmu*+dX)45N$b=y3_2I?Vd~n4aw_2v*0^)AG
z^xtu47k@kBIkNjjGTncC*58gBOOH)|QVA?OLa1}Kb2<thvl{B6PeiBjs}7;VYMWF3
zW<@hQXg!_99I~Nqq-*kEd_wQ9`v)Ia*_~5FKyj9Ek@l#5^R#WF7QG_xTOL#|<XMG?
zxKqQ@q3XbZ-ERwjl(iWK8(U;eOvi<V{YSE%Mwq1q>IqA*spt`txiFG=7K*L|^gI1L
z(y>k4^8d`erQsEYmD^Zl5l?UdN+lsIblfa2Xzj+NK|EvPb=h1*Tin?)s=`$4g)MB0
zyevvD$)6_Hw|3W!W1PW#8~qv-GJhR5_2woc8(%J0`$8-AGNZi)9qJXFihP))14%JS
zxvvsEBzZ?qG5JH93Y6^QmThN(AJlYL(vED*6kBdRC3~(yk5$70%l;PVtIYdugV_*&
zbB?uavRi+~jGi7Sk8Uq+&y{-Q=~mRrSiZ|39dY$N{3#fVB%Ue0x6__$Ua#(KsWlFR
zcfCsJl=0i{+GTp_m{xmtCVM4TQe-$CDnxtP^?f?@g%nA-fzVMevcO-uw6IpF*pRmc
z7Mw=RfdxhmzB*>7C6oiUi0K}we6AEV%`B0Y)&RS~-NwntFnK?&qAJsGe(e-uhV)HP
zrrpZx;f-*gA*HdvtD!!XMSktm+Sz5=hWWO<Dld!NYsaT}-mmF8MsDs;PI8blyN-JA
zPY<jqtJLy$wJe_Ce^C4ZDX#v|XrS_CeDXE&43f30Bx~@<P5}>q>eU=p|3@r@91A*!
zW5sHIS#6ELl4q&f9V~la(3bZ;Y1&t~H~8Ye-~36{-qf^DlNJglzwG-~J=x`>WWr*|
z_*|g-IV;OSmK5VS_;)3Hd1e)Oe@#E)c-<;{JD<iZb91x!AUs{(G=R>dNhN6&Of=Np
zrf-Z~xc=>hFuC0WI%T?w8mgl<dQiR3^~^)=M=%T4=fN%oW0nO!vEp4ttvt2W*b4pZ
z3U>aUA}ZHzC$CU|XA2C!4TGj5Yj~>pUz|=}1E+xy{7m5CboWP*4S)&&FsXRRS<G$g
zbWecCz~xQ<@&31=`}z4^{(^<>O$CfQju4*1G4{)1Vh-rA<mS+<V#JRwiF)$G3PGPj
z(Iv~kGUS>iSzWNe397nq?wwJ_iR$H((7r3n+poqSOz|&~daR6i&L-r$)pz^XdphfE
zOzMqYuY!RyeiZRLpmmi=5o49>6<)gCZ+fU>2bEhci~NCYj#hyEQj0pNeCisWu!)W3
z=hVYVq5;t6e{c&AkDsEAF3a0Hbl!XBb=XU<{RN)|&wqt{ynMU*x`@XvOf`as?F)XS
z+~q9}l^_wPI!6-EA&0?!`NUaEG*pOyX+6Z2`jVV+YKieM+)Eq8G>d(xDG)~UM1c+U
zW2rdWzuB~1dBF2*!Q;aj{YuNPIHB~XgIFy4qhvt2zz1p`Xypk)Tt^O*HTDv+lRx+9
zp$5wrMca@mk{e@Ze_Is=tIdN`#dqZ?UFn0oRT<J*)mbC?w8RJhgdX+y5fu3F$0&NM
ze^DD))1f1!icsl7&iyb*vBkA7Y6Zz8N1g-@k~uoHF4!XZJ{#c4e@Uu!h4kWh-anTT
z4h6Q7FUH2MSFd??BmI9q)A>7Y;?@93hQkSCzw3tlS$3Y+Z<tiSym48Cqc_ey0YHLI
zR{Z0OXt+sJN+)!d)DiKxK_rlU#y_P<3i}N4N(C!EI$jmGr$JDN^;@__cNlmXMFlYr
z=tc?Z+6dD!Eq1O=1<PVYdH>qiu4Red0%Y_tYQ(Uok%Unrw*6pP;AUM&j57YKkg`%j
zc4l&F)PKnCy+Dr+sgw$`zN^1yp_?s=?OL)VE!F=mB1`n$zR7VPm-Rgs2f;ETgkA=*
zv7@{Q6yLC*8d{YJU1FX@{m1vfj6kp(bsR{{o4GZu7Ww>94A_JoqI#OfuTC9GfpPoi
z)N0w4Evl0g_Qq5bufzm@4=Fz?p%b-E!6dCXn5jGyZKm}vhYE9wMBWJnwJJ+Co=n?J
zHOE_Fs|Unkh3w%`F2M=6GrMd0I@53YWb7*_+OD+y{^4Q&c9dpVO7a)I{&`;3<n+cb
zd`k9WNMdz?z;1QnG)5Qj=f8mrAphrepjXgmRY%6d-UW$<!pooSd<(rTSq>2lu|bO_
zYsi4)+`8H)8Ivs++{g){&vV+mithlNPw~nw6_q|DIoxhe77M6q<3=WaW1<9S`T%cl
zCf(cy2O4RGk^n5*mHN}4g19rZYS{Zr!sjZhO=vi?d7r<7M@_&kI^bvx!0fxymj;L4
ztuMoTR>M?q_ynpUtJDrY+sDh-?kCG<VtMlkN4xV$tuI1uWkXTJ<);`?w|(VM$nD)U
zS~5V^o4R;!TRqO&F1*f|RQS139@qc6VzV6?;`?D&!=qfh@vicJn;{XtvD2r(UDHEr
zG7NzZJJj2He76yBbOf09w6qB#5Iw!O>#bHdbHiy5o^>m-G9O#V_-D8(eoswSx3-Qb
zkD4Txtk5z(2K`4CaJy<2NAOLAChrvywp_N>6Wrl2{f0l=+e^;BMILuR+B*J{1``xM
zmnrQY%<;h%VgqF&J!!DjBggbI6)4>VKW%*#KYDzgG`V-Q0XqN=>{~leU;}rn$QAU;
z@VNk!#mx_H9MipR*N@zW#*aP-7iS;HzYuwugki(hQiE)UD}xXro+z6#8<TFtAot5&
zzq*mo{DAhyt=r$fM5`V@eKw2pstxjMWsOo_$LVYcY6Rq)q!RV~b=E4bCm!NAlFoWW
zrjpE^j;l@rQ;r7OlQ^KN`s%D4<?L+iGcr&XRXJiZbFs;%#r3%(+wNhTF?nJgohaZO
z7IC4D$2o!>&F;^x^+%kg=i=^{I2WEUxe)JXvUx&0>}Oeir(q>*ggR_3EHg}x8CwD#
zrMIbx-&kuXuqH@(zReX<W|n5-ei=h8^cAkGl+DhYXUYlS<6#9V8ki>7Fxf&NFLf0Z
z75|T?^NMQXi}r9pkcc3mBcKqBbU})AY0`pp1*9XrgY+gtmC$=PAYDp8LX+Nm6ObT8
z0{-bBy-B~h5BI+1Vb;o8vu4hmv-kdeAKNGQCPTFnX9FvCgK3Kvq;XQYGchw85dm{w
zvk+}&u!@j5;xUl}S41MJzFV1evd_-asqzFzOrZUmw5#S!K$T8cJ3}$eJ4ZTdX9pFB
zTcXU%hm3Y^GaW20`Dudir-m+GRzkw^L5XifS^&a9ea8X<M#Ms#_pLi}q&^;-dx*7q
zEn_fiF5E3ln`^svdn-c+jf?&jWL1YZ|E*X_OXl?}KHcK5rI7Kxjl|0ho~_ju7|%4k
zFtDwwkApF7OSYLRNOiPUGPM}f34~Hf{PYfT$7#fjM%A9}olOp=H#WbzL;8{-pSU`T
z2EbP&5sqiO&6NgR^*8phlZH%|T0hF`D#{<b#6B@EoknVuj$>yW(n`1p`yoXT!vd84
zjKdVoNCR(!sPgiG*xm`9^+)Nf5~Ni4kGh?HJ6M_vQhlmd(oB>bP7G4irb?2{7oJS0
z)D}=GPKA7}#~4zRg0)i=`ez(w-tGhddqG+>Y2+cqFEIJ6P}(G`cmX9Q%SWuqR$TPD
z_F-ls_>5*k_7tB@1*9AAart9s`TQ~NqvwQDp3h~%W0^;Le_==`OUKICR(G%4?ZvAL
z6uk=)nWwpSG%|O*zp*jzAu7#f|6He`cIag1<ILgeft91Bt+!rP$=LbOLEkFSbapWR
zXMBtat3M!_vvBQ#-ga|zI7zW?eV3|RkDtRhO{)kv+m$vJxj~-E%+KST1$1EpJza&$
z)UfkXGvLxw?_LU-i1$<-lXC#iNXKFKYqP*VCqdbzpIYmwtmVPXRB?FBSsybrQOdOP
zu_nE3T3h9o{fv|pQYKgZH58Bu^q1hznL!ivFU=!^ZciL%U0f0(?|Id!&`|X&s46=0
zGxM-LNP6N}234XuH3ybE{<+>n=G`7n{M}h2(HF7SRCC|gZ`4?`?UIz&rTMq9DXjE#
zN^WO!(`r7|nT=%pepCt9N6}5W7|lH1U7X~{KWQ0(V1so1;zm9fS8~AhrD%IlAyz#M
zsyowlmoAT)^6$W-@qo>3*H+cxFKY2s_(8Id41^lCF-Jp9D)qZoc)_VLEk*>y8v#Ku
zcP9`&|BIvs>2N|3D$pPPT3sOAC8j}9S7JR`c&1-%Ek=06^>@%@*jGj-3k0D`vPZm%
zwE=zj;e)UQ$bbH!v4FsF^PJt!y_36sxb3wCsci?#WZ@mpLyg9qGk@5Nl;;M@>`Gv+
zV2CU!8x$7&iNSaon_c)OI=Q!!HVh(^60aOK`E_yP%~MKfvU%adABx^OQI#p){w5HM
zdc$XQ)4DaDRk~E_$yRtHnZb13wuBH(a3}!*7*4G!PbCWm1CH%!5W=Lbk`n<zRv`%D
z<W;0O82TN;OakS+`;Kp<+^LNMo*g@o6D5GZzrBN0TGTi1P#K&IY(k`vU|tYHPCga^
z;v{@S1WN~VCerqWE&TB<%-LDI>Y@N3AKi~IdKS%0F#Ha19fxPCdF0)Wik8?;Pg@J+
zG3l(2qyXLqxiS*s0iOSvRBm_Q5k!>m(kph2_Gb<Z45&|AA;;HFHTYlStuMb=M%V7o
zd^*KHE=F4-;H46$y0>KA$~G)d!WDRE*nCMq@(8axJIz89;`H)~Ma8M`WMK@3Z<H>=
za+1YG%KzWeMclTg=ESV^D}ipe=iwXCGscbtoPg4GU`^)hWcl*i5x6B%>v(0wJ~!LI
zfL;3^L!ChE*V-a`Jj&`x@Y^W`&p6s|@vX#csXC<Q#ZRGxwK~JiN(BngP;%$x)6)mZ
zbsj<D_D-A?{fe?ICf^X0!7roJ<x&t!>d8OH{eyKEtm^-j-mc8vVsCdlelI-AQ!iOO
zzX5!nCoTRvZM0Y#L$=Mi+ZW1oNR5#6%&h!|t=+T*1dx!#NCjLi{c+@PH4Zp~-+P6{
z*sj~!uJ3I6!BcPc$M1#GdDRvx))<;B%FGW9l&19cv`|OJIDP+ce*9sfPBk-V!YMAL
z7h!guc5C@tyW=KD#&=Vwu>&93?)J*nZv#eR5-gT>g8Uoag)-+Fr;7B?S@}KRGMkeT
zoBuVeaF$Tw(=0)6^2B;8Jmo{svRg~e5I(Ibc;5|emyW8vAtDl|?-YVKJR^A&A^+%a
zto_2uSUk~QLVzMZOsY=~repY=K}eM3@fgIku66H$92HmR_GO{jt<3FYch&C4D;%)n
zsLV4W6sg?OvOZ!n2WC~|G=XFs`5Qzi8Vm`~KKdBgZn57Jq=z?lU_`VQ<aKA{Y3v<r
zYOQlOmwG+t$pUg%+He2d=$RG6+YGs<WCQ|=SqZxLc3p-3oLjXwiMwDGWotMc%~CC#
zysLMgjq~$8M!Hbo%b6;2%+CU11OO^3C>%-*eroq}>iyav^M^Pm1$rq3vwHPpnCdo%
zW?t2>``%Jw8XprAp|5L~ASqY)vGjM=&-Gw!R_ae)(CSpd5uLr~5{hJS`N7AsoW;fR
zmJ+hj0=jU`=Wi9qC^NO;PYEChOWC{T>7l`)F&}m$C<$qGa6>ApaypoT8UWMeEp8^L
z{E4%uGCzs*rP~aRVpI7R+6$E^&!n!JoUHq%O0?qN=z9Ei{jHg{#UWoE>Q3S|=)ml*
z-MvHt;xG_m{8jlZ%a8i&j<W4@s{*ys6z(_yyuLFZ3|0&OZ-A>!;TS=f_=J^8j>x7<
z9zvq@Fo_KerxJJDPug~A06zHL(0o%*S-HBh@Tq3Yr~aY5yh7sIapH*QLp45lpcw8h
z*mCv<nPxJ|Z`(!CV5&l%{Wh6<W=M^M^0S9d`N`#TC5nq{hk7X~vCLc~S8%RKPpc6o
z!s|Vb7cT4Q2tmjj8u<`$z`Hp=$6@Bb4hAGZ;so|b%NU82)<36C;k_9tyK^bEBor(r
zlD}}j=UtFKP(Hmr+^*@gFjL!TJuOw$@Z2u0-O=dT%v_(s*SdOKq??=D{=q_vo<*{?
zqCxyX4t)w26?gu_0*xuTQtbETm`(fbCq#M@%dGWD0(Z^t77YpzR!&c-Z?nBJ5rw)>
zZA!m<x#M{$8&g*i0mRH_mN<(UhA=;Y>X0S9@|?_ur5N(kkD#3r5;FEW&boXXT_uqC
zwV(WSz5S|sl%oO$_ctn(xd>bA_^@o_D%O0t)HTB2@%_cTl*~)IP}6{;X}s&a!COY)
zJ_+~B#{Ctv6S4>q(_NBX=VovFWE#0zF5c0KefyugsU%$iAV!O5Q-jksN9~?Uuq=FZ
zmmrVC7Q%}Y=qXd$LDDZJaah&63-O>ehHdox-#GN=$bv>m=7pQ7MZuW;ah9N?R(pI$
zU_D`PuH3pWpl5t63|ao!b!YeY^-@>ih#w8UgKGY<v)SR{7p#EJzOR^00~yLW`&aZQ
z*u76Y%SjNx{ryKKDP;t7<LxCRV`M6kjfQg)vg0|ux!SXC)++v0_YCuzA|dZsq~XDZ
zkx%={LLmw%I3APUmY|MhmpH)OK{S7Dt9N#$ex^fv(Plc`TtcU40>|U;sqI(3=A4j_
zMwmUzN!XQn=p4I6XJh?Hiy94)$+xad72GUxZMz2A!g%zn7F+#4_%jlgEOf2*2@`$B
zdrcjEPwMq^*fHrokvUmIdPlk@=a@8|#3ZF}DsMwM6Xn2v#QQtR27^Q9A{koGpfI=u
z137*mFlNqaZ-0MjIH@8-8eRSRVJa&z%+j1C%yyy+`YoIx=}1F#Bw3?$IyFsSmXoGN
z`zuhNoLncEOqF$-95^Un2$AJh?SnmeUg#E*gemmkf)Gg#v4NshKy2@-pFXuuwkezX
z&OOg4CbnKk(bWw3Mf|W+$?i@Dir;8k4hFYzPdT09>{QZ3K(x8#j@5c^-OYB|5<==l
zbR6QryqKS;6K-AZ2qKU+BwNAJ5;L;H02EASgL0aGsxm1qxaf$*Hjq;ixho9GR3(|}
z5QPuI*n%mbARQP8UkbPv!3N<Zmf(NByR(Z|f1GSjICmWN?6u?5P<yRMcCEG3V#<#_
z;%=90Z%?(EZaRQdtd|?fwdtG~;QkQlGwcgp!#(vA-6Xd_RG|o1)zL5NQd9sHu4oJK
z$K)Cl@!K=qc9ey$6=Gpx+GSI?IqRB3X~s0Re1Z7@&G{mGbovQ{f#b7cGN|t!(nFOI
z<Th??aF<;dZis2HH;Kh>7nk2<ty~>i1pIb2QAy?%)}3zNtaO_ZoV*7o*3qyn<$FH$
zw-4JmK0Xne7~%R5y60W!B@d7?;WHwi3ZJb9X0zwpKOy2ujDXPuYeS*k7iKT|zpAkn
zS;N>ru6Gsg9N6i$!2R@uF(hCz5hX>oP;gb6#Cw)|Z8RZ>R#ZqVrkoQc{Dtbx?CN-z
z3l3hr!!qN1eaOQfv@W@rtIikb#*smP)O_B=53Cj@>;`)7J;6381Cjcr6||i7yfK=p
zjC2)iJ<7yoF4+vTN?Q4cr^9RUrx(iu2@wKezt!k?MtVWl!bg0h-9e)3OB3uoPgc-w
zzNu$tcNLfEVv671g$7=nI37*gFNlyBJ(AoK$9}~;idDKj2QF9^0<MlNywR|{<IC$m
z!Siuwr!B~-QVyd6F!vLVxsS>4>QgPT4`nV^fJRW57W~@qoG(zJUj3!`#hO;jAC=Nj
zCB&u4UpmsDI7QXhobp;~`=_Es=Z4SGt{nQq`c49_3GlIOafM4??}yBzxqYdz<PDFu
zjH}iNiFy;sNovnWB^S57{d+tmkX24<NakaSzAc42+-#PBLV1e2Zer<X!FVKcw<OVu
zc3T)L{W7OJLF$=I$P@VEM<)E+c!Shpb&iA>u(~u{&WCU8eYQ4G_)9LD7qC3G=2RjP
z2ALxuPVBeRp@Pf5Yw6Fc=S_YmBL+>1?Ggy1$S>!f!cSPJZTkymnRfVD1^~Bhds4eh
zDRv@YhD;zLLkCU_RhY=NbZVPGYv4y39gAtK;oqJ`X0ofoDGTZ4Uzeq+!i$Tjbjm!)
z6KVR1IK=NdJ|ZOT;>F~8Q}Mn@b03jTOb8ZmKkkIyWif%MQKhl3W~YZdQBvZ=Hha}@
zXQS=5he;AtOW?^2g%!<+50CJRTr7A({x0a-CxDT=@0hHP>Phv;MV{9Z>s@o0oKAsx
zYTaXn>PYHsaw{W!zOkvdf}hZ3cT`z^F(6%!i?0MO-<A0y?GUUOM|}kZC=?FLhY&f&
zb!Rec8oz0rv1oc5<$rd4mf^*lt)qhWa*px$x><?A->zn0mC5yXDetD=^@hLW%Ita=
zxV=bd@Hu6vq+1QlmoSl~HKO>N)TWpH(OCisc?V7mt)I-uC>_`JdQ;S2bPw9dy!8Vk
zeOLCUst%Y@_PGNT@wKX)VUpn$VKU4QBl^fP-5S3ut->~pq1F00r;28`({&Au)=@^4
z<s+Nj4Ak9Jogz)pkcJPBQ>I5Oe10AQP@<KMSIH$~ferOF8N~{h0cUx+cBs;6`}8vv
zof%=JtfQlt$`lsUCY>N(4uMXm+MjRntk3&PtnZlOm?^a&FUI}NK}y9Ly3WJpr0rGM
z!S2XT1oJBgPfv1>W3N}t>Z*TDENEWSMjWG#IMoM=7RGlqa($BZO@j{3ZZ^(<pi|r1
zab=c8uJSZBJS>-gJ%)0VgyQjxmk?6J=j@C(lEc=PzjeQ$G4EZ6AL1=0N(U-Z(QLD>
z=rCn%LefZuE=;XK@fgNKYTt^gSKqw+|3LwoNpgP^LT%xis7cuZqQT&6HEI8U#~!Y~
z(OMcDcy+nxUWJROCg8m4geAM{m+^BV@Xt|uRVb=rSy_{v<M%_;hcZ6b{cAF}$VDF2
zoWZb|qt{=S<ZIfR4(ckyQpcM%LiLoaqaZ(oBsiSXM<QN3u46GJNV4Ggyj*bbB*r(u
z15)}rb+|rAJ^zQpN0WVK%k*rX9A)6F^mccF!$%{5qe^~e8mNA3YWC9iaK4^i)c=p|
z1KNhssX*VLgRo6M?98-%eu~qJyvKhZoaQdpV)!$t5u4v~>O^?y6eTulkY;mObdtf3
z>llvN3rVf-fj^M^8S2Lk=s&JM3{Rq%xn}N*ntdL}7f56>rC#Fj?Bi#)m<u@{EmHEo
zB}YOWE!P^r6)+WIA`&y?v>MoB`dS{&$wlQCRhAwC4VINtVE^_cwVPNzzu01}fj3Sz
z+C>CukKB9w`>LyPF<U)m?7b%Fi61OhKqaDI1uE7*7~b~2T5>dvF0>$#t1pd49!{wU
zW+g13kCu4EZ=GEvJ1ohfazDRVMV?8_KN$Qq_{no{Dv&t{`E2v7^1d-$=U|8!*a8^^
z5fhx}wCLd@RnpO^T`sR0@tdpUUaTYhmXbNy|7)YNEMt^G9Hd0JETvKgnGGp=!5yA;
zkGS?5PLl4~Lke8}5&{(_&(k9tde1=Nl!5YT|Axi4(HD^l_XIY`MK8ieHjQ7OMx*7<
z#_}%YlK~r<UH%crf~J^O5&b8jv8?j4cP&ymiN5JWreF(LCe(0w`IIKGCbW1H)A5UG
z*E@0n1RYA|IFtGJ&&aF;xp%!YiwBBL1l|A_t7h!_x68NNvq8Rlg#%R)e||7b&n_sy
z=D#c}arKR=mI?2$TI8;muG<Q1IEZf^Pn&WO0}vbuHr<Hn6@eXU$4hXwA)N4|N)b`;
z_kI~Hh*{?eaq=0?u%iAH=a^Q>Ar9>?H|}T1ZfN^F+P|#b57QUfDRQnS6EGcLQ;yWv
z>+`H}Laql5)OmR|y;F4tonywBCPj|mf2>6ECgwc#baj7@eVf%x{clMv`ElN7Z~vE;
z02Ksx{w0lYG`pE*qP34~G*rdjuu$$1i4t1@hwVB@E<~99s3tzRlNY`vMNxGxe&pOr
z80WX_;ri41VE><IvxC2116s^vRL{EUVM=w-N!||2;|BrPL#x(}GMOEsn<tmSzT2lq
z^tQFl2XG6Cg{z_IW#l7`S|@{4h>XA=03q#}x<wv9pW<#XT8D{qoW9Z0{#sJ|79lkv
zQ8g6>g%-r|j+XDkl}hOq`S&gwOf_tP%ljS2Gx%KgHc2F&KLp>LzrkHyq3xXoyx6xB
zX-p9_76;WgzdKHmGUsdOeVfRX7~L<2f5l4mFAa_7aCY65HN*t`^bv6t47~uS_EEoT
z>c~c7A)Ot9{8KaEx^@~aDV4HkXw@%$jbO6xo#-<HTSG%vZQ1+siAw#=RpA1ZUmUIv
zEy|{GJZ2Zgw>MikXOFL!^ns(>`q#h7d;VJ}HPqp=uA=_AI>W>CYVvCG_;1?w@`&QX
zJE7gLsF3v3u}cRp!I8IUpzadDC@1hDf?Tzl*mtaE{`vwwLP-b$dC!or<-}6ItEhT?
zw#Z)^9Q3f>o138NhNC$hM`W?zzE)L9M#`pqrG)YV=lk_7IfIOhqt~uHDQ7S^^d4M=
zD@{j6kT0J#8J4QWP*awDa)Xhih!*+8R5~!=wa|KT!PIe!HOrwt5(b3$l#)4%wY0g?
z*x844h-4A%*qyTgOExa97@V%I(8d~0i2CeMO8tU+lQfIlGSJ-dr2v}5(Ij@mvKM^(
zL~X#|5X~p!Z{x%<#SK!~t0a?{XqjxRZw!_-ER*HsBx=CuS}!!9>P5m6*B2)7*~T3W
z_7psFKqdO*ls+5+AVPCZ#_nmex+8drss>(=C8?;u1<<L2T{L+IPLvN>dl?APvA9O{
zSE}5|?_fXI=eT)1^tY^VpP&k#8tUH3n3X@9WqwHH9vkwZjjD6wqa7&Nfc0_%^)reH
z=1pYbSSw`qx%Fz8p5Jzz(oO<-S+Z)5edkr@_XooR+|&b$x1#NbM{{;5+lRO39k*xr
zw5kF3%n$AtzaL-D@0!h-VMaaxj!*W|@h75lMR9R@)he3|=)C`njHMQK$Mn9hGW!hc
zm>DmmUD#liseM_2t0futCf_h42o~L`UZ6{+!T6GStzt%SW|;F~7e|)Coq<P1ghDA_
z=L%km-}(aUZkp?=w(cI5rKY1hXo&#dsZ9#@5f0m<xz-Jqot@3z%90cf2RiCq@5}xb
z&57uOyN)O=WOG#v8h+u2bx$u>yKko371@6Yn~5StFiYvxlP85BLiO)4z$0RyL<&i^
zkGQPkeh9}6Sz-B1wxM6&a_z{c6QwPEmDE#S)&}@rOwy^B5nGT^m6h9IB{$Efk9v@Q
z>I)sG`?k{P7l}}Lrk}koB_oxUbqy51RLoXE57z8zQI3l4fbHJUGWFtj*m<u*bOb~`
zr{zpzi~HFB5ooY>+{)13?&;x4Sew?fs=s?*QaMWLIe*l5eNYLoFD;DpqZ(ed<Lo%c
zoB{E+)?dEr>hK$tO*1u`-u}Cp``ap(SIs$DQ8GW6g2aj*<lXKLEyP%7S0KXP?^OU-
zCwc8nuGru&g-0>vL|iXBtpw<qJy6E6a-<Qzp;QVqKZso(ZfTVmO!az;F(W^l5Zz@<
zR(~Ukx-SPRQ>gR8UVsR^vg-t>?kM7z;sJ}W4`?!|lc#dz@0!<*SQhvn=qF(drHiYT
zR*7oq0B3&Hf6FnYmxWbZ?DwLVr&D^aUpft&=5ENcECyt<A6hjl%u(t<26%T=xH!@9
zX*Ay!Mx151IV$fr{gD{}nLPJy&txpD$BdI$*)eALBzKu37%B>$zQ+$Bk<6&K0<kH5
zR^~d%tVCb}9V!oBy${p{$)piW8>mJBCCgZ6hX?pqqw{BG?Hs;Lx%1fM;qKqIqn%55
zc?(jFIO*Oi6`ErC!ZZ=X^)_SHrzIJ%vQ&2i6BRB?BHKcZDX{%Kc8jcW(CW(z@S%tj
zOQi?uOx-aoMzzvU?R}gMp~4yie*YIMtEOs&i1({EPMY+qUL1<FUM7E?Sn`2SfiI?+
z6aW3oZZ<t5JM(V~s_w|v#FnoxM98A*ifq5iD9{i>X6tB!Sw2>?qFEEwFovLaujmX|
zBRLtMf|lXn&ZYVE`jElDg=hJk+v#b$#}e0LA6)I4nIdOhm`%kCSV-c4l*b6lC+0+d
z2!*Kck9Fcmx(JgJyDn`!Qf-ygIMzG<RRiaE50-y+&PdFrx!ZGQr{}NCJDO@y1K3$t
zsavLwi;BWtPvPrv(IB76o{6Ayo-_Rh+W)%30wnTf0*-O_^%EXT`FTR0ijF-Wd1E=;
zT;DDZT)@=T`1d@Mg&SOedczNh7VPp+IS2unH5S}GY3bDFEUV)hDgWSAu0_T8G))IL
z&kL_BD6yCQ^46xtxWu3|Q=c^aE!>DU`0yrd;5sfSwr?}VV1l8ljb^M)iYvQcz!x&G
zz8)EXEtCV?{qnuGUTWe`p7&7qV@b7H!G)d{RFuSQq*iye46Okj1EWUu)jI<W>dD;2
ztVC3u-)|n<TB^lPioDq96-k$E#yl)8h&Ga3R@*{;;J?t_Uv&tYR8PkAj!bi)$aR-6
zmqm+(PepW(x-MCKT0AfD)UlMl$w?2hTTEm7WzQbo66EN-T=m!N{0I*(L!x$O%>s5N
zmjg!Tt~~P2{{jmMd;IJ8wQY5G9LMpLT>9|fJT)v?e!e+P)nL1#;Zc081__=(;Ca2}
zUBfwCVvLfCs^om%+R|vKw>)-$|C+Chh)TAtg`Mq*C9rE2joM`BzIi4c_`F_s<Mb2V
z$Cs`|k$C~<F)TMRGUppO>KXwF$*k6ktsHBE){{7y8$e^F&1DuWe>bGuJ>SWAyMTg#
zvIw%ErfeHW1mcGfnPcgC<QB1Xt$2)}jw+ZTOW*oYFkY>#Ufa+h@DJ1-`Kuv5^{-F!
zb2D74KT%SCLxS|x{#pA4Q<k2n&x1E?ycItT_X)DS>+t_8dAnq^*m3jwPuA|0+stS*
zY1on-unYKle}D1)t4-?32%^x`=8oB@O{8emXEhsqdbrYNAZzQB7Ejrff`W}Y*EKmr
zK5y_|m;MYJJQcEHq5#2L9+|b1HKv6}NEU(yQ|$|tpEUaTv^N=6MX!37RzxU-4?@u@
zRE71YWJx_226|NW<9GLF<7pmrDc@a#y(anz_{g*e{JT2RcU`!tmARfiI%f_aOkIJ9
z^=gz_*5OSxT@BJBb$BG6go39rs`ZRpR6rKN;FbHpr@4jXOYAGKba;6UCbVR5i7<eL
z;nlv&*i$oC<%ja&Ck!U6`QVV}4eFL0DLC_=I|+LAcjmx*gGP8_LxZj^e+G|j&-`m3
z85VWcUak-3eR_lVbMeIPKOu?5T6TRl0t9FjBK90~A6k?ORj^^#qh#yld|u^9{ak^I
zR*+;JdA1T-y**X$Z@K)&?4OR$^IbALvb^*pp`$S>m?oSwRbeaSp}fQ=C;D2Xr*Go1
zw?tUY=1Oc$y;t_s+S-o*na2sQmUd9U&bn(yqK`jwNVqpEBfPx=uVuPAqn{_Qa=Hn)
z-Iwv*Ju!;$4mv8{PVGcxxVZdvebjc<s`kHRx6^>Oyo)P(+`qC3ncKtHB>JZl;~Fgn
z*<4Qv#6Cej&b?*VG&xq%c0<9o)=dZ)EUiH6pX)_F$tr9|+?B$*q$Co33$g%oir!X>
zA_N%K;U!6|tZ(@mM1BKqQf7LrUZtXpJ1h;nbh@v($yt-Emj64qG8G9?YybMUFeT<v
zGk!fr5i3IYj?zkrD%9B!1R>gf@yTPSxp*S;t$iE<64vNUCWm;TLS{+ih=S;3!3dS_
zmeCS=)$U?^Ej^x47A$F<wk!bd28{U<_IfCXGHRyfR!%RD+AfZMPPqE?-`!o)Ht&y)
zQa3k7u5IW2Ot(Y8KNjj9rAFZ5*!P7|>PQ;Z@zQSFpKrHD9Vhjx4yL+_vC@|-TA^SS
zHNt{|-g+w&<Ft|9jf3ZVS}R}~-@QNmN_%TN>srH+1MAp;*~e0!f1d;U?h*#OP6^(p
zN}pt*C$V&YHFKroJLB%didDbXRj~gwP@33@zg;l92*5Y4t{gQe%5{&LCa!;;uI+M;
zQTUu$?u<>9ZN%t4WBWp`^o3GDZi7^nt%BhVTuHl1VdAUtq*8hjbH9P0hI*ueizgYM
zFO!9F>WkoauY^*k_!L+Ht~75-x{-p#1qdv{%k}z{<q9Kn?m%Lt_sgY+WKL;|mKVD)
z`LF-apPlvUk6Up@0a&K9#g5DK$*u<T%@d2sPn~qWf4)!eNPo^1C{u*vU!lA`YX<=n
z{4wNrn~%&`GfUw9fxD@^?6YD*mu=d8O-_Nn(WF<gxOpX}7Yt;;sW*iSt`-05m3i{T
zX$=3#-JVSg(;2U7tk5td5hQH*nu`DeNyBU@=hMX}S%x!Evi5(k>I?CTyRS`_Kf66f
z@8P&p;Tq^r*H-J`v@l<8KU^Tc)|;ru%0-A}z#LQ9JQ!J(661eL38zuzXxgMuz|s3H
zK1HTc@N6>$%L0V;aCDVO6l*jYriCSG_|Cx6EY*Eh1QDgI6$1Ki({sMg8Ml2x^3htD
z%UUfx^8p)5Ptj!73;SF)+u9(#k0G>{(BK}7h+tx>8zifc4A#C%qY)q_<Rb2zPwWMK
zdoagJ70N2703-f$;gw9tDn{RGMq8n%{XD9o(WT?O58!V*UMqEN-S5VwIVv1m@<28)
zJpA7hwHea6zDnuPamBDv;vava=w`!RCvwQv!a`hI+hp4%OWx<)*7GdUhgQ|1(h|*a
zUlh$jU_!PKEvN0iKz9jo$+<;_g$|F~n*g6r@_lnswp&rK%3=2=F=^>X@%>>piIn_&
z{B+Z=fjgO;&!S*SwIYN*c;4a5C|hC}p*EKyjFK2KM$aRc%C7VpTCA0fuC#Q1LUh0w
zUSehbZM@!jJ1j+&aG<lsF1y(q>Sl?86s3BPzYOx%5w3EVvwhSB9z`8_SyuF?)jL;K
zJA;sQ<JrBCdr#6}y%6Gi@ULJCa9j-gNLwvXrky(%z#jpVe;S@@g>pt#hb54~yrzXa
zC!`6YQ>p6OUi`vh2g$(u$u&8r!@86OF_diP?@jE(RCMSqb8Q5E&gNn>Nb!gs6G&z;
z5(a<e>1B{B^>GSLh5gxO7_xtLUsUGmU*Y81+5pQU7Pqi~n=_$Z-Ju6gsgpi&XabL~
zFI=xI%LWbfl4<@)O^4!7fS3QR_=+(%*`=!Cg4S^b!3P@~|DiJ{^k*qVMFvcsEFJ#m
zS?{2qj=tI0TqPR>mCqB0f=2QCOJ=U?`N=<Kzt8P%XC|VHJeHg2Z@c)fk{$5F5SxP+
zHx6C`ZJE5PbMSlGXkF%~p=?8JO27lek}(e}H{GBL0^+2)lB47XDoQ(fW@%o@gq6tS
zDx@`FS*xd2y=2INybK1#dR-eg-tPZd0M2az2dvIBn@hN>VAf*GZ4o#>dranNPBiO!
zZth?UeXRR*s>me5De+|{|FlMF28Wtfz@Agi>4okzjjl;0-zkN6eBSsgz^yQDGG!nm
zl7yszK-PhYLbgoA?ED<?2w3NsRyJ$>c&n+Rqn(O3gwgt$ctf0wY8^=ya;fmW@qu|y
z+||!T`7PtziOC3K4yt=v8^R-(A3>W^_aSU-0y;Eop^naE&~FLM2$*VA5hG&ZPd{FD
zTa@O1hV;?Y&`5(3vNr%da<)DJ6|>Xw3S26x#p)$*v!B9sX}z10dpSeUl`^<_hMkSw
z!8herc)^}u1K;=lX2@r+D)Oo*oS-j@l;9nb<g3v)EbkP4z{E)>YEJXXs}}dlHQ4HH
z?#liLGTb;1JlL2_L03Jy16gguosK_Xqgrk@*Dg@t3hDK?dIAGb2BF*82(Jz!GO}0!
z*?mXuDCN#?e1vhPiL6txL>qOiZ6ZTYNLW+{^h%T)&7_<CB==5;Rk>2xh}mM5jgoI9
zmgB6ZpapO3&9$*#>psLMaZX0(EID<xr*{W`f~B|yUA$SmITdyJ-Fkt;?+~WdryHEx
zIuTwKRPqE}0m1-b=61O}%O{n*`Iq0@3e&Am_WvE@iPJ~ye4lBf=H?}TwYkT6Cfj-V
zd*cixch8Dt6)Lw4Rg>;HW3$7kL6E@g5@{l~8DV_8Tv5An@BX-W@`Ok({@h6fWzgcj
z2W#_C)TZ-#!AMr5BkitB2TvCfaWgMIF6WHW&IgsHzyPi_#me7nzK}N5a6hr+#u-7I
zbC^c0?lxPkwjfa?jS$jfsiZ7TO&+Y1^u9c&|B^HO$2?7i>esJC5s<(oM1eQvg93b{
zU=Yo$<p#zC$i1bT;iiuAE;MCK<WHDr6gg!RR5h6)C;A|5;xD)HDt&d;J*h|u?%}!I
zWXgXJ2Yg(|y4s_(b<$z_6xE;ekR~$Cr%N~2n|-Uu02Za##S4tj%?_}=2%o;{3X2Tu
zTB4r5+dIgnARim$sox+n!SP0QrKdEb!aB9tP2+06jU%b<%Zhu$=g;bBHT!3~fo^Oc
zJ)Em~RKo1(XH=Pbcp*hJ1(lTvc(R|!@82ESK=^K;l*Ajp1NeW-PVMNJt@8VOANRY4
zEZoOy%o!%=TbNA7;vq%7{pTTXtr5ZD^6W)oV!^Czo>Ztp#ZWjRV`NS^I&B^FHC)ZJ
z{AnMixl`S|=SPD`vL_V1PSlszdgd*ZV@iCUh5EoaR=^ogEruFy&c43RFyoK@57l@B
z6okJM5y`K6$2aVGnwEDyY;s0ItT<&=z_&%JtD_!+I#AKdRCh>w+h)unV_1HtFI_X?
zewl)HJ@ldy`{DHA?k<Oi${^FME8O4R-~IITY_V+K!O^h-J61XkcQb`|f3YCz?B2ki
zH7>iu&ZL9Zw}Jl{>S>q-grcC<Cg$+~f)k@22dy$dgE|QqWWjPeL&A>p=|lo<n4R{+
zemq_&NQx&tH3ytu&PZ?Y^wDu{Ay_RmmGw~I`DO4k<pjG_ZKM=QDo(HwXZM`lv0MU_
ztXh2djP3riY_c$p7?N+&k%Qm_hh)vlGA8P9TYf5x6P|c-<RBOm8VkqkuK~_@-XbR_
zqnQ_FYN}~4gWNNxY0^F@krE{+Dk&H015b{fe9bkhn5#e&2qLtJr<r|`j`$vbR?WY*
zP5%nK6!s%j%fX)E!`!>q7N5NW!jd3Um;ZQu`Jon9HNtX~vagZ%xFR}#A~aNy;neS`
zyBabJo;7}YWHfef`Q~PPXb6~|w6hdr_C~kzZgV7P25TozMbDR0jc*TX30I-pS{Jbl
zbd&AJ7d*IGqp<PC%f61ItCNM%Z-i=IwvK@it`G9cs*Ln0u(L^_ezY+ZU8`<E6EHA-
zawx1#(`ux-U5DZOhc5r@p<q>Kid~mPK425UJkhZ-rPbBd#YotD0&h@fG6j{YOj0VU
z5>Z+Vt*`nqhD@JHzGl;bQr{{ivkS~Ki#WHuqvcaj^28kp8R@9Ss?}3<PKwLFHB@=Z
zHc(QA)=+|B`!Tvtiea)LWXx;<Oo20X_Q}dh&@``H0?g#^LAj^RoFfefus7CIe|)FW
zDM(Lilz#a>WvT!i=;tjsQR1;hvDUY>`2>2e$zMg7UE!G})KK%*`XOd?FHJ?nnaEzz
zQQv<{`*ycW-(0Sd3A+Tt6fzh8Ql=spEKC~NzB>loq0y@4pt7D*EA@zn*Q#1DMHcdK
zd6`&SVt<P<wjW;EfEAZ{;nyp>2CHSB!q2!_UQq{9f24K-v>PjEe@R5!i3Sv|OJDZI
zS!p%dP*#+CE#B^)`FnjlC<XM?N^3zpM6Sov80mKslkmnybPjg&{*zkQ-zUPq2mh&Z
z>~3s*U3WP?uU}XxuJPYgT|bpO8&;BdOdggPT;O=js+)yqcXotiSl^LdS2T%uC-MxG
z=o-%&pM6>iNVb{%`o)m$^+tDTx$S5xKmXraT6I((R`+~Nxio<<dZeUHQf4T!Eon4n
zdD_vhk=9FP9(Huu^0Amv;~M?$u?!yb-g-1S=XS(>=W!l=ktC|Vcsa)UWwoD#t2XBq
zd~Cxq7B}G1xHuLdaCovkD|5LiN^#mC-q5MOz&kz;uzWU?BJF~1G{>`VKQe8LlE~Bb
zf)It)2`ZINFrq<uQIbdVXRVKG{8>I92~<_Ll4h>=akKvZ{&BniHNOZ#9U(%*#0EyG
z*-RO4y-k!WfT^2!iZsNIQr%$#D@qg=DZ`a>QXeAkhY*v}W*I{sza3%Ip(@HRieaaa
zNB?{PL`B4M5)|7`N}LYLCu^u-hNP5jO%OFFe95eZgdW=X{hQpEWrj>Rmrge2>?-m2
z<q4I>y=27Sm*M*v4-Nh$QIKeZ<p|-daIRYi6++53itR-mH_+F&$zdALv>$#Um|OV8
zjtofX?7kdimal_e{adN~2`6D?H+<5I!n2D3Q@^t5q4^S%@%4@)?3rclR+e^6dDTL*
zm$!d|(BsyVw!NT(*Jh(fyXN1|#w96=g%K++?krnKZ!@4?q*-5{8ZBb64~@7IY%^rP
zyogmkFJ1IKz*maV4<Nl5Oi=Oz6?-FIv<8dQ>y_bhg%E>uN^^%%bbatU`Usk(EJ{}3
zlZrxDZL%uICnE7ZA<H5K(7S-#@{+r(FsF({n@tCU_(tzKt#RV*R|fi&U51WS3+WZ{
zI)*%p4Xj%CCdX>~{SngN>EIIEQ#7q-gKDrR3fA~3V{n=*)t~}zxQBDB26sx4y+L|c
zKoUDM8|dy{;PhHs>xbNjiG##@##*WXR4cX<0e#K`{ySf8kh<AWs^BxPWty9pfzx#^
z)TLunO0D{!0FpK|a&bDl=#g8%;K%!`*e(tyuAFq!IBZKtRu7RrxVc4az<FPW)>2p`
zMI=*>W?nAenj-esKtkT{?>AR}x@I)hn26T6+TDfm1vW5oxIf@?)HNNY^cV27x3t`g
zWRaoiYx0sN+;cgRxtyv$P@4L5o7X7n*TCB{eu*(`5npgN%ll|!{YG6cfg7aoDufkE
z$fh2XOdFrrl0_1)^^gCc*D`VpEGd<2r9;C-tZqcadaO(QHb>WBt|Ra|ps{29^8#wM
z>ly1xR4_QBrc`uy<9F>YZV|X?U(^SD&kk0;nDF>$*i!2aalbTXidTA5oI3oBGr;?=
z+}EEo#p}6wvq51kN1I`7*9Q@{B=6B*_gVp=(hv7YlAdj5QN$#s=ep*0_}hEzTs0r;
zWO<Cw3^ir0a_fKZUc-x2KGPx#r#E}b&b(TR$sL)*YQDvnaeS9S29E)2TwZsHsmev1
zP~@Km1_0x#y`Rz)QLlE)y`1|qxO<wVr5Ey6!dh5bPa;tH&i{t{{pUtv%FR_)y0vmD
z7H-%Ehbndbvh-(^a5B=LXZUCqG&D|o&URM}^z;<B8P#b_khV={o5xP<Q`acx4aJoW
zU*P#hl0^>JUb40@auk4_v*{PP7ct0N5y*#GrEym;`oG`)Z?)&rry#VI^SMwGHJbxU
zFa^E>=^j>RkVgmz3gxS6&-bUqfxo1AkIxiNca5)|WVkF|ACJ3B+*Of7II&w%a|jPR
z=z94Vlw~z8&k@PdDjHaYq2G-hzQ>qN>*~iki%YD(p?dGhhXLZM@8qeioj4KYkCQ*_
z(aRC!&w8`+B`H;)JNGG+l*^iClPpeU+Lv%LYtrvMb?B($ti&EQn6<F4%U@>uS3?8?
zmHqR&|AvBLN(%8@+CL+%tp({avs8qtN0CH%FI<8eGDt#FbyO)qis5O+1x4jVh4Y`7
z@;xNiNth((QxAc;RQk0{9rLLUZ{=##7}+16_G=Pgf2^bQ*%!;s{~IRIm68;fmX-jc
zuFY-^^Um%&M{^6;zONR_i%X{A7~r^eX{_2_E}YgkYdQEkCZ8Drb9jFjnYGa~)ii0<
z+~DznGF#5p&0SF9rPzj<6w7uqthu4q!}pNJ*$>#6AM3i%gM3{-IvMkDk)VvWE;ZD!
z<sy=|Tf47THt4j>boaT4P=EMfXx_aS(Xx~!y<B}#BZ|EYNb3fBZ07f^Q#Bjgk%+0E
zo2fx-gl=zU6=>cN(V&tJ)$VY1h7Srfbj`KOsvPyz4Wc1|Y9ZqGsE1$d9j!ea$wOqL
zRha;dvXgdB=OkN*euyYUBV3C{u^Uo2;V~jdqaDj7$YW`##>MU|MDRleW5f0I6HG;J
z$*S&`u_OO(CB0&Ods72&+g`s~V2^G!{#f&^$uYY?5Z5T``a6?9=XN&_Pwd7QjbuKa
zT`n3e{(5I*#6))27T=j@6gA>?`R(uh@gp#O+RW>=dMMOgsYNNTn#Xji2mMRd;@ESD
zhu|_XA>aG>HaJ}}eY8f8(DUE?NAF`|#~VY)j^JF2bnU2jVm$p&2;6i{XBAoNYL%A8
zXecv?2?)NT>iI$yxpb+|R~Tx_^w;U9^veZK5j(+WzppHu>c{gNwfWiEM?i)wla0Wi
zze{A*f8D}!==q?txTe0}ksGw69?!#--VlAR&zJkmANXEqY$^t1DlD)3mHz2m7st?|
zuHc>1;SE8{s=S?4dW*m2YJ*8n*(bxf%3M5|k_AgS@^cHwL_~Ky$HkpdDDoEF{ga|c
z*SJ};%Sp3qMIT?k8U)F5HrIp5h#GFdt&+n-!`SK}n?L>9ebEgcvVs|rf7b6&y6MZV
z&Ql(ZgP2bX&E5k4!ou1tiB$^)Ek=w93E)nxlYU97Ah3#&B)!Ya%d5?<6l<_<BRWey
zbaZ~yimQh8oI#t~o{&N%l$DpuO`dL#cCATpqtL_ju;BakUmlX`mp5}~Hw_2CSb_ak
z37oKhZGNSH&2W3;6`8%b*TXW6(o%`All%GsOh!6h=<+YTu3LA(@RXPhA`{iuv!m(v
z%>uyr@UE5Te@~2MEr?*jMX`d4d`S<bK%CkF_h67d)Y+XEzl^biM}aJ7#IiiAtv-{;
z4&Tkw5&jAqd>1;rqLr(pH#`&|?6&vHA5JnLxX%h1wk7+Lsu!iG8YM0l&C3hQ{=tf<
zG$s6xhzMW*)~(+7+)?9kGc0s&PF{z<<Y_QbXRvHXJOBb+bX@b_jEI^7;vZ6&#Kj~s
z^WwHPodj~mN(Oq-;(s&z(lfr-%*FivjC|DSjNMLW`G~iC^Yfdv!hx@H1;mdy3yKrF
ztwivTil#Iw?U6K&Npytn2(5RVk|2$n|J~$%UOvrqmO!WQyN*1I*PW~e`OG%QB&?g5
z(pR5hN7$DxAqoZOqSVPD{1OdCfIWsi!a6_yB56$bO8vCz&Mk|)`Qk07CSrsMN8`35
z8csQD&zX}eef**QKsxqUopamsg<Z7$o6y#craHT}#;hzpM%m@C>1e?UO7RbasDord
z!XCIRQIq<I-R%OMzsaD?(XR|1cTNgWuPRCJ!(Z(1Hgc=V%G;3U#_9IId%%iTRTfZz
zX-7q=H|;|S<)g>kaSgwBoL}(4%oT{9lFC`XW2ivX_n%wn0GNN6iJ-&&y+HfD&7e}=
zw0-w>eB5|#a&l90FKBAVC;h~j=VY^#rQ_l-28-P`%JCKp)|5W~UBpBs-68P%r6fl2
z)9Cd{{En;CuZ8KXpi3l$TIl@<Hfx*qcHh;hkw$(fzsA5aQCP>#tV~XWVM^A|n!M4H
z62FCigP)BXfpW&ZY??~#j@yXY(d&gsj%BR^Lo=SgbRstfDA9t%9pm@XdHo~SOfX&R
zCSzVqNYF(sMc})@&2z`<#r9vQ)EPsaE?h)#uu8wMx>)W{O^S8yA9yh6#_RccmEJFM
z!hP(`ulw7}#wK2UubyJoslF%q*(>L(LY+C4T)1ojE8v-B<9zV%k02S^qHr{>xvt5D
zme<ngDRtl-<Q7VEPs-zZ3pd>xV*7@K5$mnJT3jNfC<`|*6~|9|8NF#BfG89+W|KO5
zs&>#BoQPMg21-jvTYg~(1*O#7dkA(;3KN**4|zT@y5;{PKoh~=c6fpRmLqp%jJ^0m
z!B<mO0F^iw>Ky}3K9zS!grW}Q6T@WwkeVt1%)vVBb^70m>h?eBDS5!z-l@g|j@n80
zc2m!0-D=IR_}9sAvESvzC|XUwXLqf57xO^ZWK_r*EE^1_ecB0u5=F@qD!qMd6_p$@
zXpeyq7dCOLny<g<NseZlkjzebAT4l5K!}tRc!c#9$kJ6Dc(G$8m^)^pz7uM>@oVvx
z1h_ifHuJrVlre`ZF+HFpdI1SBtY-`cciEn*f{0@))0B+KFVZGW85C8*6%ks{!*Wq$
zy%|lziUM@opNsj6@m0r$4NQp99+}-*oqIJKc(_{s^mv4SL1WVKl~<c*T)>6r**Fr=
zti*IRo&0eV$#EB)b@@%@V63KEmSN`Kw73VX@|gJ@Z|u+Lw}(`i{4FPsQE@cEn;ml{
zftwUf=>-|FuiQqds|x`X;P(OR@{O~8e%JrS?!}MC<2#a>Z1-XZs3GQbmg;lc&2!xW
zW|DZ)p7yNBgi0BM#i#iI4ni&gN<Uz4xL~kyEbgu>N<CHp)>Q(Jh0mwzlo~yBw1^%h
z0Hx{(^fM5azU6n6A3TzJnyEq0-bcg6THEi^NNsV9W3p^LQ^hZq9&WzB9+HvQfSD|*
zICQLHPjU3aFiP+CvhDExcVVoEM4fLK#zIE=IuX_48G8kur(fj3glPYWBmj|x)Z}yv
z3~biTHU+NtkGldV>!iUPrPN6l>Ak$8;jGD@dO|SAcqQ6`|0$;4HdhW*adLf0anzcV
z*4K~eFCL)}XdB1MTp!?FmMjZxmqTur%SfL^tjuHb$nCsYW7*KkoQ&o>`C~B!AvVG}
z4bCq|PYlgSM|(e{mu=p|=YJ1AHcFw~=Wsa0Sn@*{RB`+;R;84POVRNeNE{B%28ong
zw6=)m4e(!Y%FMr%YX9)ZA?@s<RQbH)`gQii9s95!GrwZAYTX|3dy+5}1TNJqAxS9m
zq&{AC$MN(N#S1iyc1nNP?kPAaE%cTk7nut!n-V)qVXlK`Qm-KZBLe}IR%$<^oPgIr
z0`}1tBfY0F3{jRuR>a2!B~z-R!y^mmsjDvE!w>I%Uvx9+txU}~35rtqthX7>V%~DH
z>~f1y{mgpLPt}?$f(zc%yNh64w?xQtqm?JfbZm<CQXfY1-H^KdyS=3PJeH%Z=}qF;
zb8!6EDzL%trnN>1pwE6m-n?*N@SIVAKgxVHleZ=8D5L6gi<_5-_^V-AmCUvT)y=d*
zHODq|g#!v7L16u0Wup<gqN^(43$LV|K`9r-p^)36c`^$D{>^8;w<~s*wvIxg<NRV;
zs!^{hb0Qk_&M>pye0#=iS>z4quI6WgZ=(sEODu$=-YoP3bKT+tj|DN0hC;cz2svbC
zttma%e&|`-e;E}vP}kK}C3RkwC&LBYs}vKc^>C?q2R_;jQyx$*jgj!ai*&u(%xgVw
z>A<3z6W8MqRB|Z2OeTVR(5rGL69Ot$C@E;jwpH<6mb%~5r61*CM9-55pK^i_KO1W7
zTMM1m4`)1pb$-Iv2sYW?!A`EH>2-`9!S{lm{UnPLFK+NUnp;^}@pRo=`uFeO<TZ&w
z`YMJ~K<T|M&2{|?cEIKGYFU{gO_9aC$ti7Wys3QJ-dOds-g@HA#l`L}UrRWGiOF&N
zg=*QVg#pc6H2)UZ)mXC8$#!U_WP{F6Z^EIX_BZ$JOFpUrbRCYkhxws}yuB3Vs^3HT
zx_QH#vGpUGC$dR;v?Rl<+@R<r;-41UqT{rR1Mgxw$|ar{&g7z4`*||V9}v(2I^zcE
z_Naii+%b>1p+@o;&frz_18ruvittns#yA0`hsAYAD0yww;@%O(xPCSV2py#OEw+Id
zuXc|(d>vC)ImIp;(>(W5jQQ;<+PmnP;$1enxBdR~jO*Y|M)#_@U+*q=DQ+fYt~qY^
zcNbbHSfqpSut=QWIw~^3TT~SA86Q_NyPC?dAiMsvv^6uL>65o+ZC|@}fx^bc*xD{n
zI0~-#mAY`fVo0AEkKnmnUheuN7?GPBOJY)G`vyDIU#nf6w<y(Nsb<)3liZC$o!DYx
z+CCg=Ft@Ddwt4yPD^S<;3hwcyqmFd4uI9Svpl2?NS3?6n{H9H(R)Rp&&x0*wUPNhv
zr9&(?>#G!aI#=e=7vUKHhvA3s#SO-(+{&?M()0TLJCD`N#%4t&j{KA8tr1*A64H(@
z`V6f(rnHW?8}t=K5-$v06<TbbQJsmmhO9L-WP@5~DYbdz6yAPTNv0859*nkR9lE%1
zl4v`<;8E_W6Z*x>!cPHcDD{ZCn4MSO&fyUgGc9|!)%YpcZvn?i)9iKR!B%=EzviT0
zN-$11+?BPg9VK+n!uDBEwr$fGG#ceH#~m~(Rx8L`J0m9741hM^xdb|^y!k1oce}Kt
zk!-gd;Alvq0G}*{c^z7fR7~J-i;K?JQX{Hxd~f=jYQtAEc>ttmaXh?_-eNe$Ij%#a
zii!|gy^&u~eruSc%&AD3+s(Fx1=RhFmbNY{34XOC9n$exuZ2y|)Ujqq$L6nSjoI`^
zN^)9cpd#WYcShG@4xHv2UbBQH)>!k8yp7dZPp`KRMs30>XwtqIXocDp)}guK6=aI`
zim*?d!CofxfS<R-W8_%rs?u%F;`y?E$JNZZxrt+8v<BN48|Rn0XecQMq4Bw`RehOl
zNG}2m>*!XC26x)tA)(ZXR)BSb3hLz*tXKa170yEfWgQxRGrZ}b&@W;`sycKCP%!7{
z?yUYj?x)h<8=N1%^mjko9nj2^R_vW~5@hylb$#OoD0Ud7v-j^Gl(M}1_#M0H)q2KJ
zI=xEx3T$BNyO%{>B~Q{1TW@G+zya%Rkg2r$8T~2Kl0K)Y4Z&Oc|2EETUB&w?#I9QM
z+Gmlj&Uy<yDVBuV(Xirj-tSog0d&=MS^P+}K^0{%_?~_;-TeaYXpQN*@Rb7nCeQLv
zYN6KvVNFvMbXCvQ*`+9T%X+%mTf(dg<jUhF0OBl8?T3WkFY|H{$@}7F$^gP}Qy?nN
zAAl5uv(%i@U#L&-tte{Sp%i&gg-04h#1POsS^EVE7Ak$+NE!xw%-PR0A05ibWrxyH
z#Te4f{5&a_c$io(qLN5}_+eqt!i_b{tpc2pGK>kb+NyAQHyjrcs@{>wGTNo_Z0~0G
z_G038Q|1Oo(E(gvoFquM26zUF<nqp9M5#)vKVCc9+S*QU{u@V9@H<te8QlEcp=a6M
zLALm!K70^7%bT@(hZ#$`zNR<(b@=7xT2uPF_s0OwR-bK5{dnKa{{S#S&%SK$`0DF7
zzx&RA`!K9HWy_OEvdkEVB-eX-=c~^hpVdtZyKQq+pBYHX*eGefXqsBalcgm+osFa3
zI2@g2#rD(V{Xz_mf_nNOE#^^y(9$Toxz$;QFjw}|<!p{xrIcLrL0xZp`J?B~&yHtT
zH;47Cz0f^BpU!*GRSR*ESz(oB%qSshMHQBRCP5~&svUJHEt;mtGeTlW1|T6OP)m%0
z>=^{Hi7g|iz$zrg!J=HmwSV~Li~XK%j!!17UnX<$xhwmf?4r$x`kOC&>B^Ox)%$nn
z(sXjl%I;#hI5<AmIu{j0ArLH^r8Q>ME(>F9M#DjGQ7`iD=az2!4Hgn=L5wkMZEwzh
zmopJRIh!Z&d0~^6+;p<6C<;SFz~a3ZmR?ofc<%X6KmOq8-W?y4lFu3hSLf_i6ZLA!
zE2Wr|pb}C_lk=f%n)z(jG)-@9yKSNiep$EHX4h^#ms@-P<9AQbX4@N$?%9AYgb<ux
z9jOsl4ez3ja^AThNXlBftlQD_?BsZVba-;uESJm0L@B*e$6MF7t_9SQq0IA$Su0>r
zS%RWtm89r`G6jiniY~+$BO(yda=F|;JUE?Ax*O}?{)2CS{ab%u5;BlAMXx)^ZKc^I
zA9&Rq2qFUiufI1Uh4fkX2z&-5{?(_5pcILy6^t4`;{&OPGMY@A<-E$QQ8W&J(X3)v
zD=R2`R@e&wDaBv!x-pj9#vdN;fBNaC@4oxKZ<}k+T+OnqEK66<JGs7g?Q)*!(d6{^
zzVVHR_aEPS?#93Uo4<bT^)JPUnYA${79mB6A%q~oZ*1*URdwmorN@tVkB$#cPfxct
zx9{Km^re?y@ZR?az4t%-Xc6##_H)-;)b(;wFQ<i7DZs{{cX4M}Pfqq8KT5$Rh=k~3
zoGjZ;r$fpJVxHyRN2Lvh*r~cHMk2jf*8OfTGdjl9_y9;-8|y_;a10>Qw(VlMT!G-7
zb8XuKz{=GW+62fcveu+EX=Ml^0LTBJm*cZ3Z|&;%*)*-BV6BY^uJs6i$O5ek8yo%B
zwIKwh6%qySB1d3pTQ8B7O(rXaMTCv9BI4R$tR<q90ud!nF+>*?q0W>&9Zk2lHmA#a
zb@W5Tm||uW3a|D)M69*>_P76#6y3OSqh2<D@+bf4U;O!Bu5GLX$b3F~|Gi(Vt*z}o
zdi?TBFa6(t^KS-&!S!2zFTC*l%isOMk2~G|Gnb$Fyaq$3+v!%F@yXnn`26jic|GZG
z4Q7jRZ&0lD)u;D9Dyr<!{Yl$SZ(QHmxw27Jw$IMZ=D4i8-c?LXNfQlaX-2cD%?ev|
zkIznt2owdtZ-4y{{?+$?bUJ-Xv&;$=nd&GU_WI-5xDhNGALh+}&RVrJ^z#UZjt`o7
zx6|ugxyr{U_jdj4$nEVPRz+2xoc0BKWkJyowbBfi%bsBXq)dVK1~nqMXmIT2kd;m3
z1*K8bf+!zj&0LE_0MtlKi6C*xFj~jC#<k33gKV;9=i+wH+@nwTgv$LzZGUm+!ny6I
zXX^KU|DWoNQsgLqv0GG;=;Z8F6p~hpW=4QW#3)&2V~7NLty?LD!Fdz~5b?KF4xUnc
zy7#as%BEf9RoSho@nkd`otaD-LJHZ&V6&TTB)ddWAPgZy-$I`Cdg~XjTz&H3?r7;t
z14?3^XGA0tTi+(16d;01jDX4zY6c|Ate!aMYiEEIf<(uE+=jLZE&)ifzIN@U7Y-jh
zIvI@(OAU!g5+I`{Qzb!Ec~Pth<$<MX9k4X8T+WuKrzfXpr_FM<Sk%rXYgST8iZS{i
zNJ&7e%(jrDFfzA{<63fSa%(z8RzdIhU^jRdLLgx8>n73e@tFhj<*$A7g%@9QtrVu?
z9a3cm>zinQEA|Fy0TLJ$C=vlCl>oKUZq>O>pb3A;PXncp6p%_n4JhW6NJA1}LPDcS
ziwHxCB6+9MRyR$<A+jbE6rmW~75zA+L_|pPYkwblv78<r?tl2v`#=2Q4`Ymy;zu8S
zR1`UfptVZkL_*ut8yjms{mD<i_Lbk=KX_W?1tMO5yLPR0ZCRBuu?w{T6QZ%%d^`ex
z!^6XVzx(9jgR&^QRrj^mzECfx#b&;k&o7?aKApAY_T~ro_c05F#ms^vwKeSI#FVKE
zJHw-svo`Aavd;4^l%;nrdY|WoDDBvRteVNVEQ;WKQ5gUTA*3W#x2uSlQ;I&%@|Bx@
zv8+vhmc<wsi-p#Dbr}+41OieRIWdT9UEV1YN8u1ruP{=GxLhvtJkRqyF)O0zW2Y=b
z_^j571(niBFk8%55+pNo@FLW>x-3f((AttRB7j6Iy)vcboHND%i}$U~w6fNDFQO6v
zYRw{}>9h$Tsj*s-QV|n?c;~fM8ZmKr_0>0jfA+JV{NW${{&3iLZTPc4`?KrUpS^VP
zO21eA^<V$1UZ<R$oo@687S*?Y|L?!^i=SV<c0G!D<>gnN?(Qo2+#06pda=9rxXSn|
zUwyNv%<1e%Qk&;ly_D4WM^E;jxpl)9T=hzmv$DMV=<(Et&HjeW7O-><GiUVKXd+sF
z4~E^k^+vJL>{^Z>vmk%|um5Ilm19ivxdAAS+k(~g?VX1y-Jee&k`gclkpb34%$M_7
zn2kmq>0OCg*18L#*N210CnrwSG{wR*B-V~8@}R0Du?4U?SuELmvNUz6THm~XaH2)c
zQ(FrtNhpvMEFdO@g{Vy0ewEi}p)y5(ucFFiu}k+h^`2VJr}f&o>u&{|%-UQRYf<xL
z7Ju;l-H)HT*qly|wl7@VJv~`WXY0dZR%P?WB86p&$!JYREvB^(GU)foXc3R1I7q)=
z!EZ~W7l4!`q;xu(?BvC8ZS&#dhcR+d6cR*{F=Xp`yE1x+2m+)So%33mThHEqetPfT
z^k}b>XNEGZEg*XD1Xj8|yBMQ@LQNz>sFX5@g;G%x87%-w7!g<A!kB`f=%U!Zd=cv9
z@od^LWQ|fvDV>ATRkz<Cu9w}ewi!YK=E-O@Iz4^z<mqHOpUkGqMIC(-KyB^H#vWq~
zF>0-av1vwU$D>}Eb-P)ipjTyo>%(rhD$0IE?9HTffgA5YxQ%Jn1VMf7{I#9)SDXi0
z#AF-PDh_(xZeLsLo5m1oP{;x-2*}KdIi;i(3lbm#f+EEXDS?O(p*C748>1%_Kuk%9
zP;0HV6%lE5zdQKiYj3U%HlG|Eg2)OLCla;R0{kl#I?L>@eX{y&boBmz`|rN{?z^18
zwV_jGPxp4OUB2j+a{#W&;_Br~lj&*O$P3RsZ;ZKc;Q}hz*w{EYIIF74hY%BsNQeQV
zPr$=|e`{;2)9K7+v!ciri^aj-L7wHzFqw>*Q>W~ZsZ6)~<`>?0>-`U^D$6q*^w*B|
z9$$XtxtevqT3Z|T?mszyby2mfMs=;VZrZkSK^qf8iZSIyr4_ZUpN+?zbK5C#k>|Z`
z*Nakw5>u2$WmYTANmfXvrfHbjTD#ia=Xvg;U$rP|B&}8~d#5!lm$kLos+y{*s=BTL
zz-C#~w)K3P8Qtx6r}Z*Kmu00OxEQsyZQEv91_Cie5y=Y+07_|pLzzz!gHJIE$clA?
zh@4m(wYsfhW*<B#lTs4Y%&@4v&Py$*kW&g)EAQLFC@qOJbgJ^T*I&86`w0>K_{TrK
za`~C6>PQ0LcyIN_aQKDSUcLMA$F6n;b=50lfaRhtdz()mKWUb8jP7%r4Jm%%`K#x*
zRaq<wV>bsn25ZEBp3Rn|@<th@@bzbR%Fa+UpS5kbyP2If@<8HIbj%~u9u4!XI|wle
zq}=MZUTpO{zFmOQNAvlScXi`xk-Djy`E;>p@|&+_U^dp)9xW$zV%zHF(`7!O>sPv|
zRoTPQ>D)DwCeQL`RCQrU%ki>r`V)z&W5<iRV+o{cjt=;LIM!M~c@$__u<@n^=K8o&
zMg>+{=_r|?m1G!mq~V~SYj)FdLG`7}-E3Z!$jzGBJ7*tI(EUx6aDQ(P*m%bTdf~y%
zUE59`9sWE@T85-CFG`b_)neJqo238-m-5WC-mNf=F@;65jFGq3hA4dP()qoEr`DLy
z?Rou27a~@FoemkjST1+>_RgQ%>Gy}n$47bI5fR@6qipTd^laYE)_)c33?V2o!{OG}
z&ZT-XQCe4JxhjVc0c}XqJhTpww6R31u_gkdjLsBdnF}T$v6=A_5QT-AV-jXkm{~#$
zBCT(&HJWUU;swh7df6S4HYuf$+Ta#Phr6GC^6CA5`}ga{ClO9UWJJW~U9B`A5Rq0I
zfH(=P;3$nh8%>e@rE}}U;ieZF&6>gz*%EERFi#Q}O*>t*V5_`8V6rTRR3?b6ch0-2
z+qHS#&ZfvLDRPLyB8aQ%C}svoB2k186)7UpD@1TgA$TtWgo+55SGg-9twfWEMA5gN
zd;aEsv$u&2ds+F&wT2ben3JTGRyr_(g<l7<){h=PJ{^q^b=|a`s<W}tKY02GBs$j?
z8Fahds?!-wPNuWP{YQ_7gT2Y*tkdsYf96Hr(%uK43@{}ZQCO2QS_cjnu3R3EPdC@M
z+U0UQIV|%khFBI=uh$z6*Dhba_M@M^{b2upgD0o+jq_KVn1Xkd)7r3SH1&H`mf5Pe
zkqyQ_`0>xj2Pf3+MV7#T1RQuU=+Eb~wq5qR{pDi5wl=WV5Mc^|Se8v)Ri(Dp7?l9)
zx&dW?0FitMf`E!*^edGlaV*QSO@R=tu?oeQUW!hUR!SRdNvWo8I-Raos&4AYvazv$
zZ8PP(FSE?I!L`e7x9dHJ=+w$j7Gh*z25CZy4pE`eDS=OslK^32jES@+<`h#D;m<10
zNm+$aICw`IA~JiIX=q%%*6nd{%qh=oN{obUs9$>dcGb^Qk}tmbJKz28_xru!%dfoj
z@WG?q-3M>H@x{sb?5kh<dW!tcTW^1V<!fJi=GKinAK%$NH}&;mt>1Oa&uugiVrg`*
zES|Y~De2@>{doVZUs#nHumx7Vv1_BNPfzMEef7(SAH9FB*E4>6q0Wn9{mJPlgrrIi
z7?_Zdf@iIb3(#LHN24VpzWea*u*~M(3mL$m3!6A4B^BrkMfc<Rq-9~G!8kpCtLwO)
zz4*-+|Ha?EH>uOm#4Ea5cdfE{0*YDIM9*dAfgD0pm5Fda#WW~2tNu`qh2J`Dy{X#3
z)0Fx-p3{05pcLpb6bu#xm#)Rx%Wqt6mvyJe9^4r<Ydd!b;l3;mPukh;)M(u}UMxIP
zCR+QzDajl{1aXTni*c<|O%qyw*D9?zBw<u4Qwow}^n!p&W4GHeAXPbB@0TYh`<-s}
zOWXBtW9eKWN)qpCH=3wkueY|g(J8wJ`-e+ETTU0ZpL?S{FSAJ-LWJV8)GMNLt{E?;
zf}k_|%gtYEoLiltl`%P@bB+@jV-tHq(UwTq7FGz`*w(&@(X&M62p|Z5j6w(oF{v!`
zu4X|(6(TbkQ+9Nlu<<(p03ZNKL_t(hB|%r$Ow!cN{@&iBM-S(dSyTp1Mw$>cCk{Ra
zNX)F1($+AuB0@l`bidzcNsHOpy+=<jT)dEXhDw`6U^~O^`sQRhYTM<!X%?+7i#{u4
z5+Tt><$10$N&t*fRMiQ8$LO63UWo#PRoTW$Ko=oq<OCv!ph;nf3?MQ3_{(f7A^|Xg
zD6O-qj4b2XwA|SG^6!50@-xpq+CQ8vTGXpU0stm!a_3yrG%Ih<Z}eEF)47kF7|XIF
z%zL}LgJH*qB|>sO-ne=5Y&@-95R}=X{>~r%(c5powRFw<?|oc<buNaGY%U^!qf^YJ
zfl);boP;G^*xr2fWVa2GDD=C%<zljTaJ0R1>FMF=;mPT8v3%yjMalCNQjr%1!RAIJ
z-hTf1WfP9(k;>ubxeE{X_GgpH_SX5xemt5KM(3GE4!uq(5)>t(<lLfH4%oSj6e-0B
zE=44WO%nioOr-UHcsynho7*-z0oEExBr{4}ttA<4tW|B(vZ$0$gD>(daf&QHB`u;A
zW>&kVZGFp0S);WmLt-*zjW7U6&?e&)>bgxpOXn~}YmBv)MVi)u!aOfP>(oS+$jnBm
z7#**meh85z_pWO#YnTV8n0rQnB2*N;H^_rBTQrVI0Yg@Q**l-y`Q(#NTJKNJ&d%*z
zB&Gl2FaP}3&F89a@8J0C+SN<8D8K*x?=hvHyz}<gzwwRt-g|F8IZM7~Z-V>Wq??rZ
z=kI<zKC8Oj-e6dnraf2X-CpO>)5FjTMEC0Lul(Y}pSNQF>TkbC48D|ascu0|K^o>}
zBBbinK$^6FEo!AjXI43hYwO(<eeg*jXq|6Yl@l6`>Zm!ABhxLc-RW<hw2K2@hh$We
ztz*}YW=F@THwJ|{zftTT<(kWcBrFQG)~f>(A|*gC2^3wUa_*T#?;jk#_VwFu#e?HT
z!`kHC&XM;gb-m7f-IfKZK@&=)ODO6DeP0~iKis!}`n1=*gY_~dGNz6g*en*U0_%}7
z!<Hfuwvd9W4I0NG%L-dilAtoPF<9$n=JD=st4JhPS=A(`l|8q%Ezxyyqc!gzoq5Ma
z#oGLq5D0*hxD6?|wDcj4r>seLJHxf%=IQZ7;-p@-)?{D!oj03Bw0XX=Hmt4)$D=dv
zoXre>fV6G>OG_vtf`CG0P)J(>0A>(Ts2G(}MjLCKZ>V1Q=zVlCc}__nA)*$=M4%Lq
z7Gr&khJEY6WSPzi=RK2W<lw`6R`2Z|?maokZAN*<O8X=T5CVg9AP{2|S=lB)ks`#x
zW+}zIDokc$zz;vU(;0kmZEda0bz5Uz_Q0Bd`pmW7wIOE25o6$xI3S72j4~>6Vgap9
zh!Ft{sg+tDQy~0o{&G@6T10~w0L7As(EFHD00d!Lu`NZ2Ip$jTd%cHyd#vd7FMs*P
zH{Ud!Zo(56IagqeU3r<Brd|E_ucY7*egoIdcsyyEb~YV1^-?%2W}}^*LA#u#7zTrX
zsw|7=pTG6vpNvjMCq+^G^FRNWmoA*Mnf|-){kz}!o$pwqT~jM9B8f#9lqT{n=Gxx4
zar5-#Oews6?V8c{U~j+M?GnicAAV?ya(jETb*+dTA0Jn}!R5=B!u0t1<qLz7sv@sl
z@-d!`#)0MaH^2DvkME90XWgnl=ysBS_c2BQR7xinYi*}0ORKZYG|q<<$|6@rt>~Vv
z4Nbk21V&quXfsP1V@yDziG17iH#cXqSy2=rgd(%`e1Xi0sMoDxiXs5Qnt)bx;S>l#
zD+<9Enc<X-)=DYoqf$0{muLBMS-0(yQA<j_eosVJb-+boga}ANbeh0wQChiw)`@@!
z#wZCnfuL44%MytWjsz^Jm^>*(L@9WY<h{3Lks#i@@$7Q36k+e$=bpVe7z|>P#e8;t
zXZzm0duxMUTQ_4#Z@%%y{d@Oj%jWyv|K6*wyz=<rqYLNGX;QAPKd;GPHGTDquV1~q
z^PO*fXZu|B{l8U5vnMB`aj!dnc>j}qNox1$gXv;HYOr>GU78RN7r8C3?>sYWTFfSU
z<FmpVZiCSo5TzKKw$WKWI$3P5ub-WsX>Dwdnq1ZCz4G!Gc=vIP%C~I+l##i%w!PRt
z7^!xdq6B>&>}G8($Hz|}^_5<P+C<jB<YU;CXpoo`Y8Hj4lB$$KRNAb6zx(k)VpDWX
zWVKmsc5S(P@@Qt|r?c4}krderOe(mZ)%QMnq<ZUn67HiMTOFEc5v=VVot|y)tZ%HB
zpWfdmR14>=HX?-RZ{B|9=fC)n2#(Kct2JvnnvBMPF+iT1Ix>-}%Bovb<MH^`vsb30
z@oYX)9108G#rWIW(KAbb1l-)&9Bfoaho`gI>`ZuLZH**RN-2PMZDfBsJ1fnAG-)!D
zR!)+3F`u}mwn_^lrz9dkT2M0*2_lNtS}O~Pt#_z^ImHBw6hd%|1u4_EOP5+7y;8LD
znIj1(U>4FqfRVu|iHJlh%ZoO&rYHc!HS^hWe0;bUTw`<wJ~Xj^T}jw6Mo9pP6{!_X
zoFp+4fP_XZ>Pe^D$t-bl)A?*!H_2qtDkVM4Dwkjp{L%*|h%6~dh^Y<E#nw_`vO?R6
zm2e^wBv^TB0w*CPRI3=22rCc;qEh<Ie3YV!Nd+-poL~}-k^n%I)`-&9%NY3L)#sjj
z@#Si5E1<PSWvvZ=u5Bq=n~6|jzdE3-3detwN?|=2&nKg4y{y}|CP>|GNx(KU0a8(9
z_wRqIb(+nmr)Q_zJ3H%}8#>nqhX+}fDMcbFw`N(-5=3Q;F_wXBW@n@E>#u$Bdw=t{
zMP`nV&Sum3=H}+c`fxHi6Tz&=PsXFxyUD>pXY)$p^xX4*&mBF&OP9`F+UPxgaPJo%
z-W|2HarxHmt#k9)Wb5MPfAd%W7Gmh+b`hN-YJKt{ia>BJ2ZOqZ&ZiWMqFgrZXH`Nh
zA;rANOH!_F)(551X1bW=hBzTfs=A%v+zJ$j0KvPId|7k>q^TDn#;Vr`BxV9&gha@h
zQJQeYU{HjAMjHTVeGpLAWNq6{r(<nYWUe|z<mA0?n^pkwJO>10tU_Yq%t}$@-nS5i
z8Df+GNl1Wz2v@l$g-HX%7>!lhP}@dBR2I3-Eo%7B{>eW+K0bQw)mLH+y|VhrkAC#}
z>#rZ|?xM(xw_gmwk4I+*O<gaRSFT*ni_XJ`50CbL_O{lBckkZa-roKXZAJedSJlaP
zAAR^={U`tExi|g@iu_GU@4fRcQS18YLYvO<=wxks^X{X=*?dOMWUQ*1H_Lo-vKN+~
zndd1zowlainM~)1G#r%9)iJojQkAJ#7IouNN^2WE7u+ZJ-rY`c29;ojwouRn6^9qi
z=y(Z#AV8`~f(t&+w1WeHG;4zsRxG}042vNoA_Gb*(V{X$@rfy{xO3yComG%JT2wgm
z#}l8$Oa~QU*g4hatr%z;Ulp<nL%#mLH18W4Mc1*a9CUoki@Metu3X%D`uO3st$sxH
z^!SW4ZV$WBh0~LJSFQ~Kb=|`L;h2Pt>2#}qHOM+%`-}@0rDIrJec{69$<dRc(Y<aK
z0VYjb2a3Nf*bO43NSshnR+go6p=lQTPaoIQ*=%yg(HWy=v+)lf{rf?+^VP3>gECA2
zN~z%6@ySVBFFI98AWoRH2BORsky#M5Q6fS}gjh;7n@^l;06;5)K+Xkd+7Ln%Cy+va
z6a}cYf`oubV2~qgBQa{5D~7VuUHS3#3MW~&qqEt`$*5^OQnOgh8}DL_i8+Rd0tiTg
zM4(U+2oe$zCqYH?#aW&iPQJ{tr-w&3Z$B?4S6X$}w^Z4mF8tYKI@sJsl9ZAUZHO)e
z-?nX4^|LI``dD@<Yb|+)yvpmau*4XbH7)@be`Tgf89@MTr?t1`;QDj7&+S}f0ZnQ!
z=yy8h__U5Oeg=s{Mo@^5q9EdLQa`4B%ahT>xi&?oH0GIY+gdmlMV1$~*XzvZ({8u(
z^5)Bj$HxG$vAOZq+i#QV^txSP7E(x*0LWNDrN|JO!C-wo@2w3751!neP0pN;O<OOP
ze+G$co}G;+N5{umQ4;0Xu0OMXc>25F{CmIn0S^0i@9@c|_wI+7l+E6L=dG7seNAh1
z<NA%=`wyQ!-VLEGj3xw|Wi~59+iJuj)4*PnVNTXoRwqRiLrR?Lx>kw^5{fU3X?<N7
zRksbtWVFe0J+IrisxC><J0|8#E1PLxe~L*+5hBEZilE4?(bR;MpdO=lSydtkGoy$I
zrxa(?@y5p5<;z!p@!rSd(J7Nz<y{fc1;2_BDy@jnXyD|!RXLu_g+*x_B>^FT00=`8
zj#W_<WDG&uECGyGnNkv>OVMjZO}!`zd+q9#M~@z9qD$w_KfZVWJKy@&rw{Khf9KQ1
zd|u}HlLrsG-R{P4{d6*3>-WxYZ9Vw(-b>HF@S`97NdE@c^uOV%DqD5z|Nbxj<_q8c
zZzRLKzkcWAr)MW8r<0hNQPKM!-0$_vk)L8iF{T^m2gxtDH!qFMgd}BLmw+d}?pEb=
zI(2SUp_u^P$#~K(=NlVU)lscqf1;*tx4T|gcE*j9j$UQvH0<5};};+NA3t48#Du`2
z0a%ztg9z6EUMorhq!UJ$m;e!@PGkvel$1P+D8dA~AZrY}7=$$DIGc}*sA40tCLf>7
zJd^RuYuP4*_Qz1~_bm3RZ9}AeT{p8unk^gBczk?VAYVI2kDfl+>~&*cf4>}Exw18@
zYcw#QFRQ#*+pyhE$IqH(F<}tfQ3Xlj_W8@tJbk)1o*WZ}$c?Skbm?Z1mC@Gx$57*+
z<1*TIs*RK5lct>*Ys#VmQ_QC`O0vDR*1fV$U`|g@_MS|hztLAn(FX*h6-}no)5F6;
zTZ58(OG@#oOHe>(cD3bPf4vJ<8;Bsd5LEpC@b)G#mS)$T-#O<_-}K&#_hOFBjGQYo
zD{HPIS!9Vtiln3#*`};UyJ0{Z!`*NLUfDn|v@y^dFT6LrvRVr_ECaOFY8h&^tcj$k
zVpWk<EY_Gc<eV9E#GAhHPREOvB@4Y^$v}Z8fY?OfdoOU~oqO*&fB*CQ10sS{S_nl%
zQs@x9Ak<n(5|Ri5xX`k<GEI+4>p5D-Nm`^?fuFJjthLtKqod=KljCwWYr3kou5Rnj
z+8Ck_0gwn_vBzjB7YA)XSj;mr27wf<&9dzBwd+D^p%ZU0?XUEPTOJHhqLcw9=Q?O8
z$%?jXOqZ6kc_K|ee@#hBAxIgGZ6G9IjJ&`yv-h4T0%+!lNQ5d9q6jW<=e&!Ikr4rb
zgpfKhk>k;9zIx`d3s1gK^q0FXpyD)3i=r4G9d%u|ur!Il3%7CRxUf!%pIN$@j;BrC
zcxQ<K*e9l#HdBBgq)Ah=ys}(XleTM7h;!%8*G=`4*WXxKe_87+m1PSkkt8EYlS!2j
zOOi?msYUqY-FH@22AdnJz&x4E1>w9b`~BYf=GM-^;ge54m!?^=w07x<FWz~ubNR~U
zp%HsG-+SvPuc<U2s4Q5xa^*6z(^|2&8*4)f-QLcARd)ywLL@0%aEVb}S;|BQZ-Ke%
zS|gQEiUEyIe>&Y-TN6XCNHZgvrXmtX$u>l-^uj%o8NBm@UT;2|B1D8BSy=ClNgL~Z
zWGSU26eOy)Y`UPeF49a%YU--Uv&aG16N2~k3(r5XzP|E>FMj2xKYPEdeA-ip1PDOj
zeGr5(MiGEx6hcW!^Z85?fJ7I%0Kq5$F+}IA??Uixf7cQ?lS1@?Bk-f$A_|11K@q&)
z-aeBUwK^K^?>_A6YN^PTz_pdp!S2pr&>x?i5a8-)^|c?r7Kvxm$^7JGsn^5cIJot-
z&&^5@h1k5bF&uXP*Z2OX)#1|p`+sGW+uq)|^Wa8wc>46_?#^uM?1qwdeAtQz_h<V@
z7cPE$f3)W&ZPkS%j-@0el~7=H6ha0@r2xT<=k@An>CD-sJj3y9>O<68<hibWun;nC
z4q<%tYft{^?I~T@yol!Z+xx;1cnF9gID*&#`bY+(h0tKo7==KNpn!bfzz|t@1{tA~
zvf+p#8bh`$yQVd1Zi5qu*{n7`n56<cdl)+(f7}WCCMxI3%;@>tj>lENq*wGfI<~$@
z(tc*X@YMN9gAaC3`h!8*!(=#Yn^~%GFi6%K?K<2YJE6&YuRNl(t3|rf&kmBPyw4Xr
z4@F&Q2U#vvQ1X|?y1}LXy4JdLX^dWyOm$LD<|4+-WNT|H)x6t%c+$3(NaSf|+sF*T
zfBAcN?@mvSm$Hmq-Bu+^C4>-COKk)RCIJAEgL4iDNy;=UfUxUY?|dS(R>qi2rNTpx
zNG$l?z=$Dw@aQeJ%~+n;#(IvuUT?I#n&-JvYQbL-LI8N@+oqe2X9qUKMc3Ca0wN(P
zsf<z`lIOr2i3kCmw*XNK={jGRvwBiSe-{-RWs*$yE?<A<laD?`<UxPg69)hQF3V}Z
z-!E)0Sm$Gi0C~TswSl$~LI_090T>py4I)B{41tK4IRr)|*ZOK+P3I*MBZ8Cy0klyl
zWmC0F%Ns9z<>jrjj~T5TqjSM1xpivuaBr_|+XW$q9;vh+kwHTUpG~hkIykbOe~sQt
zib8;M4M0><93AgpyLw@HWte61jW^#IE-%lkdUa*>)W+%K@o_n8&z-v<gz`)}$q`wS
z79bg!BOH$p|KR`lx8MH8%coAQ9~~WCz53W}Rvn#;8F4(Bnyj~V`powBw&|@*y*P9B
zB7-2E-g$T!P<I}M*7al@gu8z2fBE6k(qDc5)kKR?zwmAA{ZSOe4xIBL#Qq@9C=^DO
z<s2zPq~P46%b5`-NrJ1Tq(LvK%aS<)imGWw!;uoYsT=E^b)5E#s;&ue)F1ScBn0Q3
zXYf)g=UVSQDakQ%hzo|aB=pYZSw;X-P=Ao7z4SMJ{kz^(0Nfv>-~GqmfBgO5|DzZ&
zl3)M{#NZK05=o&DV%;>%*vqrlx~jFdb5aQbKoNp-)^#Az$ACsV+bAJ`0g;^bgMlG%
zSt66n+}n9DEPHzocM@Yxt*=Wd-+J@SXTI>v?!yO0>zMch?}PU*e(_5(Q9pU@$1gnl
z+<Y>vrt?!<n?;d-UWs8|f6ey}9<KCobu~Qw;?=!B`48WC>4oqA^-l)<-rU)?se#4*
zaZ{KqjD)a!Ze{z8dVD`mCr4A^{9O?`ttF9Ce0F<l_pq`dNQ&dh)MPRcH;q?_7(-Ub
zwP7yj0Vo$r9MqGouP@`0sczOLb<kLnkUEe^Ek+j!5d%hMX5c1zf02pK+cZ^a+B@E#
z^f2`dF<`_Pyqih=erG2oUK{93Tcfx4=KCUT%Fs;x3S;JCHYu-0Lru*{Np9H1(I7X1
zr*rlJP)03W-j`?AmZr!1x9%K2{lvvV+D!{~(d`|T(#T;^oL*g<j7x7jSMo-0P`9qr
zK+1RyD@DI;ORaHjf0&&s2D7$0a<kKCSLV~olm2CSg9AsNkL#tC;efKT9ETX32OlCL
zlvR2E;e8AkU38rzl7TUzK*s6p<nUmZq9?X7IFt~*lSG100%7FFbr2atj4_}wQb=Po
zLR3<mOeSs9YHjkg@XQ`iN@<i{V0#6M$UfK_b=DisCgTc#9yd>&9WAc}tr?)Ro$cD-
zU7n|maXc9#L146<>$)zNQ8@t^e_h>mDBGsBoh2X;D1_cRvwiXUlU?-vLCy$m*Lv@r
zb=7>Hr~Nc9Fh&SI%ZnIeaN$z}B>ogW3;+;g4ABEX1hBy&abzEyw@t0JWMp7eT6m5}
z#}kzlFMZ`#pMLI3S{oNbV2IWlY5KjTyyzVt9}6k9Qr`Q;Rfc(?;QR;Ff9SpSwzEXU
z?6N$o>p2mI=+B=!+w0{&dHuCCH5;272Zu+amDRRxH&!<u+}}eKmoHxnEFuGvltM5Q
zgYY4^u9{3Hhljhser;uCV}0%1`Aa|e!H<kFMbVp=Wj0J1P%5L7T(lr`#u!#NwqE@D
zH;?uo?mW0NSn98>uNM8Kf5W{UTTQbpY3DPeWvayL^61usy?i)o+b&6y5Mz?cK|ddl
z$7$AYoj1n#7<Hlmz*?)dCU8wtHEk=jYC8uEM6$Ea2WF<CmlBDt?bx;Xun!QGq@}^A
zvCcZrk&-0C7|XILtA?}|2n^8}B>)h)t(yy1FMjP?FHOdK&dr~DfA;x%_wQ_NtzW)!
z>c-7O#78%(-~vQeBv>$n;BBzoOl64>iinW{$0#F7L_$Ucp_Oghj)Z|AqzobY7$P$g
zDj`y%(==fY0MP3d2vFB`VoYlEJ3oKBZ98P1&u8EI#y39r;KTRcd++r2sa{^(zjHTQ
ze{g(sfA`^NZS||4e^Y880p}iDjw=52t-t=@<JVT!d;jJS{v^?TN=T7vN>p0dX0Dco
z1D*N0JpaVGZ-4b)zVq{+R@3I(`P2XQ&;G)_d$-}*RxH>zodsmsE0O~9gonNK-r;eP
z>QwrEk6mS#W^$`>w&5(vZ~ou+cFf^_`(MBOhyVL4O&iyje+FgSNQ54OSF(%jF|v12
zA_1G!_XqiS+O^VlG9(PhR$v6CF3CQCX1taK4$kwWjC)gGnJN(|Y@4t-7xWpXtl{A0
zv`6<QC9kGhp-@s&n`OcVn<zEx4<1YoVu%}~)!TcMEafCgCsn($GCDjwtVn$_Ym6#R
zCgmu}%JEUQf3(aF`aNxOnJWbHmo_&}j`!~z?XGVwtEofZRrC4k=7yI4APL>)v6`Y^
z4B7g$H|)s8SRGG~!8xII0I<&Hx)|hUZpujH7=%<!T^}6mO~*%4#OOP~5Cu5q$U$of
zg4)(~(aQ)b#u%)WLP;65N)zLK*LKz;X_F;M#(^!6e-MxCI1o{y48aBx)>YYHwKu8^
z(em0R852P*KzQDT7`RvT*Veb*eeXR*D1=57QfdGnLd3`jOahn#!-74_%<O}YjFN;v
zB2q#nU^I{bLLUMPQYzgzb$(~>h`?(t^DIX8RW*;^Ip@5y!FA5cBuSOdVl`)u-bW${
zi#9Are?(sR<a!LqC_pNSERZn@0)!w?kPs|HV_gvX#V>#D$>&}i4p*J;JP-f~hR%AE
z<VDes0G+iAyqHnwx^6M}86$tD2N}iKDG`yqcTSQ*kO;~*esgv0_PzV-=g!vAFD);t
zB%O8L=7n=Vzj0%<xzYGwGVQqo1Oy|3fJmMrf1sD-Ge!L@v$m`2X<&bHGFB!{vwUM?
z{rTsfogN)R<gB;c+GtdwfJ5+(akOzJTV4}Mnx<)P%)!p?&b@o1UjOln+w;j7h*#E^
zx6hu+6Y{g8l*>YPIy#ABI;oiv6tto4f=42VR-i2s%(TD|>ar#!F)~6lN_1`0Ip+X8
zf5*0|7k0xYO^_%8ch=U`9HZ~*xfUK{M<Pf~Ce=a;XT7VPtvaO$(B(_wrC)h^eDYv2
zKX!uNzH#IHh4T+?-Z(u>6KDlHhG?0i)SieXAu$W7lrUQQwvh+~Oi0lMM2bQ=Wmp(%
zqmT-uka<d4NQFkqqDUVZvS~G+9FK<mfA`;h>*=SSI5^zjJvkhm+N2~AD$^!E8c&?(
zI}dgu(c$>y-S<E2kJjG%<o49sS=)W#D=&qAz!Bgd7)a!iQS$cr)dvsny#3~_hYzZU
zd(EuUhttNfCjq$;7&=Y1mrAAe$@t{w@4a#JgSS?TVky^cSH*^7#L|a|f`D|4f9pzM
zhx6n5^l*8-XHKuL=owWHCTDAmyR{31Ev(8BmFC6|?k<g&&0b$m!jO`TFeRBV^-%T*
zwV`R4Y@A&=e|Dp)LhGZ5>ePDQ#ZHkhilhXm@K`~k!AgF2-n>>J^hfKf`5-qV50^<?
zVbLo&A~l5^>vWFfkh&ml-QR1vf9A}_^6Aylu$Rrtd73S?u6ww@-*}jNc8MIESsBkJ
zYiq0Q8wg>1a<DW^N2|r@i>Io#Q4$m7R|dV``0d~Nt>5~c)&5W-T{^dA8@IW&dFss8
z#q+0BY!<SNUq&#mj#l!d_~4xz_ipWCFc;50cJ}ne_4P9*?KiDEIho9wf6@gvFQ+!N
zAY)sXM~8dtx>PG<E~|Q0Rt`Yrd62>j;Rp~QgizNt68Y$aWMhohTB(F1^@>rJ_lcA?
zX=1X(WJWIzGr86}(TOohVtQMr&n~ZR<b%=Da9Qe{NQS^6EDq43R}9W>pEWx5J}&$`
z7C%r*8Iwq%nbF!10te5*f3ve823XvcQ7WO;>gK7(u07EkF8cs6Ab1pH@@y$d3XU?%
zdU;;tS#M=!z29GgN48n6swyIpq<LV`IYyxoq!3C7wRmGDO-RZZ*;?!Qk!9fG&)3>$
zW0E9u0lFx@{_Ed)>FeLxICVBgieW)LgSKtkw)Nf{V;&9Of2t<3f7Z5bTbAW?I*IW!
z{zJ&Q);k+x&{{j|2&rirW0I^WUi|7;K+@{!YM$jyQ?INpAC3<%UVBWZ*|-0*f9yGi
z=pNxlBOwEFWC)7H@@RB$bka24bI(5g-S7OPx}IyTHa9mPzxMd6ul{&xxWtH)>0C;!
zh$Jx)q7VTo&wJ-De_U)k+qK=phYwdrD?j?--^LI}%fqUw4~~u}lkpReU)|nZ>!&I=
zq$GOhtqVSa?K-7}_ZE>;BU6KkgwzPz)VAxikj`50E;A<2k|Z(H`OG>8z)f4n7>zN^
z-nNbRKF@QbRi2t6ON^0%fH|l{#Ta5_tu+yKUDb5>Km6<8f6aPINlr5J)vvw0wzcu<
z>p$K;yEW=(XHTtQs6$uB7=4U^(FI?3HUI;mwH9J%no0`ky%!`SLl<K}652?T2*NoR
z7Y^f6YptY^M1rIkuB-r2o@Fn8{cAsZ<!`S)_4K97k2&D0Pds^ae6qf=H5#oxcJ<oc
z!TyV1`SSAWf671mFaFDS-hS^J-~3iu^piYC-iG>*001BWNkl<ZMD+`Og#UkqkN#*f
z-aD$^diSn0Mc|kfy|RhBN0kS)vcBFnHt!YDd!<5>z-)ffx|$d-U%L!a&g*JvX_)7Q
zR4RsmM8ezCt^M7npMAP+e&@pF=Y#tvqm6%c;kkcyfA;h<DV}eI?gUt%Ah^}5d}jV`
zSG1HiW#3^QeL}%d)Wiw-p2VI|tI1NmQ}5oGc1``v<)J?s%dtNtbj~gp)Hm8oaX_x2
zh!}?}S-KRZbk)fij>gYZa&aC-<4B<8@LpZMI-NL^xZ3Ba&z?C=LIvyD)@#Fjkc(z^
z`1s}Re@hp)QKL+BqD*YtY$@w}-5(5mXY)MWT<y)u>4TF!NL!P{CDVWW{JG!w?mz$1
zZ~o5p$Dcp9apq`$Mq13L<-y*8Ns}O@b)TD2%r6ch#Mb6|k!M?*TW7b=E%lZ|OM~9}
zsjUl1Hrzj&Jlx&ytVIc}s~cDQ=;q~QdVDA_e+pzo%(A=~jMA(pq!OfrkVtBgcK@Ct
z&@r==N=eP6^WLD>AEjB}B*kJPT}q{tmQpQxTu55!Os9F04`h;}N_+(87M`gLKw6tT
z@2#w?ZES23Nnl(|5-q+zdK3y^TmXTIl$1sYB}fSr!7~YY<@)ulbK8-?#YeDsL{w4}
ze@d-2f@BuVAcanoTnH(Ih>s+b0EjMn6eyJhLI4sZ2?YWzd;=o_Y8hgzn`WW1Op-(?
zg+jI=v;m*};;%gYg_ruv>o!2^9W0uzF@6d@`_zRB7yuUV<f^J>v)Oz;pH9c0IZ`J%
z_{CFty{v00j$TSBjT|ly{_qdK_xzWhe{Ck?zE)?qw!F2Nng<VdivH;9U;S0*1+YX!
zfyADHm;r#9r1;A}`@Rpbyt4ku`ybc<%l(2QC?#8K1IiCSxqbcFFJ*b5m9`N931I=d
z^HQn9!^4k0yz%6d&u(pOuB@yWr4IJ@(=@q%|K4b{nx@$&pL{Z08WIu#6nSpzf9B+P
zQVjZuHeFq1N$PFwEhi>%w(067M!&YU5fL!*^3o_UAc@+w$S~;l@-%I#Qc5EU7j2C!
zP1?EvLPVU;=RydT7}s@8-2!Rn?P%CH3YJFM7ry+=@=E`~4*KZ6_`-7!?%lb3=`z^x
z?SJx5|L1@GKlVp!yZ`P#CPh-(f8bDr$gNuxguRqRQb>|MvesG(0gNF8KtN_|U9erI
zB?HuT9Yav0G;oa3xvuC9j!!1b>l-KYGKPRkpG>DWKK{u0_?y4^?cs2E<HH;Kd;8Bm
z^UUefr$7GqV+Q{5tFIiN9PRAv?Cw5v&VBgd$5*dD_IafS@2Rr`3MSY<e`p1OG^!^|
zRW()X@7>>(%$1D*fYt)@d^Srn(T4W)=}o1uD(5Dn@uXT`Sz&ilwaxic8_!#eKYa6d
zH?I{(k;G<lUnd6>sqFKcmv-(Rg9DvIgwP4M1n4<A3bruA$Y{}a92i7G92f-)9b%3U
zkO5@qirtTz5e<3}1(Ffkf1$6D@3qrrVAi+RXNSk#w5F0wRI8Mn(sZ_s$_4;x6qac*
ztI;w9j1FLT_u$<1Q>QMh)|2USuXl937fB7%-p<2`W0+R$nbV{3?!z<Nt6qgco{kSH
zlVuS3)}06ae(&m)3wLkb6DS&I3po%m0hK@f{I7lTv!6u=3cvzZf3`c`-P6j@=eIHc
z;&85e_q{iTp!M|?lNcq<?(Uu-LzE=RV0CR{c67qPj4-cebyxTEeq>7!ylq&iEX_PK
zM(<sK$iaq)Kny-cYi*IHQpy-Z2o6Aa?_)%z^>Ae^EqYbc3L&MWPmTAG7Nb%Cz+^jX
zeK1)eNz~S>)(7hsfBOd!Ip-#mnbzsX#@6B9zSdf&$wHG;mZfuX0Rxv(S!)Slk!OuD
zMkk5U^XWt-`NgYO`lIF6`cw-d;e7yzK02j{P%gYb8RWvgchT?V2sI-IkBWti0zqmD
zKo}OjD~O0dh(z$Ib?oAtu9R{@FvJMd*syWt{9{i%VS4>8e^789kQoB8{B%*A(Nb%j
zBnbd8bKAD_`TXSMWI<`1%}>1ljH3W~?-3b+iO_p11jgulgWlbpyWjckZ=c#)uglrR
zb7zP{k*I1mAB~1us1Ty>ytCF@>s-T+n!W*;4-XIi<WK(}8|CY7{QS(B(?ymJmwJPK
zhJqjN?cIK`fA`_7yMOt^zinFQz2ilBDYB2z`EFx#?cUuxy*!sv8l%@%R+a|6yw{se
zD{K9|d-wKs_l8T$hlj^a(=>I%%s>Q!j4>euicIVUvDOIeLh$o)W{hDD)01&8FOV5R
zbk;R>Q)C%L295x%wB!)Hb>2B^tufkL=bZD_NhuWxe*h4K-nOIVL4*!<{I&0Vb8U0f
zxGKxjYuBzJVxFgK8*9%!|I{mg{-=@LwX5egmvhNoDxpY?0Q3==RDe%cI(U;L5M!#0
z?<_<Xf))x(01!e*q5xowiIJsFgp@!OVsze5$}(FT+`50S>bm2}^kh7J;^}Ap=#T%n
zZQGY$e}4I?r=R-d=FOWoZ!V37MX$HB`{2bFzt}XDbFS;U_4RfBY}(yFG`g1W-D<|u
zuC{DrGO6L{q-v^)*$Sk(wkk`HrjxGYM5eKW7oYv|#>(2>gChYlT6=tO4AJFUI9nNo
zk})*(`L)&5t7OvT7J5EYDnCiOQ(w4r_`!~9f3aT-(GxpiD1%^h>?dRp5rGjvfYhjo
zQK6s!sh|{m2CiU#T8dFfmjxGuhJ-=PfP_HTOBKoK&H>n_06azM73*Fgzrsz^)=tJc
zk#BV65AF1!jDQgkqR@?{{k@~wmTPBL_hv`SXSQ^a^_B`7#biDoEcJv$kLt|!<NNoj
zf7v*ev(`8M%=*gd^_A%3tvkCus8pFm%gDm}7QBCO^JWA#8bQj|M;H0jsm*3S$IsGK
zd|rV>sBl`(KD_zyjgLOsJKUF2F~_!TmX`<Tx6c;6+=7!Sd9Isg60uA3#3WjPP^3sn
zQo=hITxXjqxYo55y9U{-L_jJ*3Mm9&e`E?>v^BR>sKLsB*)bBUN5&EhYbB$590h_R
zmdG;Ent?r0-8I2iA#^cX=HR{c-iF|$lIJg6I<<Y?1C({;f{(yJ$mo5vfdeC`zo+$N
zX6J3^Y~#ZE=Ju6~S0swAZa4&tfnz`qB7k7T#pi_tsT5TPG9jHXNC3e|jACVde=X?`
zM4B*$$ZbT60!T_B4S|Lz*)JLd97BvTf<Q0DN!gy1-C%Y7+_h`Hl@(+eyk`gt!zt%n
z2;q^gCITu$!nCbxTh}$dE}O3DnyQ|hjQ5TX!)KLlBtUfD2?XmprDWZfi7{s{oUQ%I
z=GNxH&ciHC-hAVyYwH_z<b%UQe~zSdKe{-ZS83AE@&q6vM<j?4+P2);*(v({+n?O}
z^Y8!Y%g;ZKQs)KYc&fF!eg9!&IqNTX9wLAcGB{6C0z%-JXPJ@m@}-NnKDrTud+(jM
z3nNdCkH7Vu-yWY#pLyn)8y|ggbUYp|E#-M$&nJu4eBCyTV7)Us)lw2df05?^v1z)d
zX?%<(&l>9%Ix9q>wV92l2mrkJijpWZrm5%5F-eRPSeJ7kv|S^l)KY}teAimrX{nle
zcHyz@fAKH>t50sdfA-SFd-w0=SzeWOe`)FFo!hUy`ig72)#a7V?Tw$l`Zt4~c7cz|
zu-MY1kRb%XAQ`=n-nl%df9Szz(}>o&4h4DVB%)9<dK(CJqL={)T?oLWb)vKx&t_%Y
zz5Bzz**<+HPtrSg?!EBB7Z3K25|dR`_4eCuZ*6TYkCub?w{P9@AzZqA@!`W=0C2${
z9v@x2^w`OG`gx^>;A1%t2;D??iBf}pPGGtw^pdPDtyT#T=1IZce^ymnfcnyNFZO%6
zotLLJwoBKJ=hlpR>nr__Zr&e`RwE+{wTt06kJSo9IdcUaC#A^osW+PbIM0kCjHng3
zCfGoVJUfA&y$u)uk(iNLfFlx!L`*~~GE-D028j>>BBP<m5D=l_aNt@NI-zt)(fa=U
zs&ZtQf=}1h_Qwy>f2=n%Y7e_P$}R>c8IeTb(2y4S+H#K&25TF0*UA*rQM!t4G4G?7
z2lu8zj{d`+y_OPAWiT>}&US6>$|`tW1@gXAZ6h81{DZf?HQwu{w_p9iAMelhf@BYI
zsb8c@)Z^oG+gtLN){G@AmJ*OoCpex=j!yhif2nm5D_B{(e`Lh+!v}^?dZ!RzHk$%M
zWM9`6A_oXnHTNMn+j$!VQWPppvKRu!M__Ovgpdmq8g(2!z+g0#M%Hy5P_jpeF@%UH
z7$Ola^d!unlnUGdz$1VRa0mfWGIQ50I&NT$+1S{4{QC9C;o;8x*=)8r)c+@Cjm6nI
z#@M#)qBb&G3SNEui6^wyZPVtL#ZLhre|h<#HiSO3WsFf0N=ZV<7;uqh77;^Ooa0C(
zf$OsKr_P>x@mIdNcIGU}tZkb-mmzpWB$A6axhNvV7@5IZH=E5lhNiA(vsqo&lgW5q
zPT7BU3x`5<08E6=b=Gx=kftV0(`!#XIj^dpzx#Go*Ct8t@9pj#9V{=8UVi!If4!Yu
zrHpHBS<c(0;^-n;2HQ3B<D)%}{#U;GRmTuj^8HueRM}wX;BY*f>NHKW;>o9<f98c3
z@9*sW`Ct4M0!T>^JpcfsAR(0$gu_Aq=;-LVXP&96>eA)QckVy9`{2RN+xM*XQfnqU
zI6m(6`bwz~BO|0)rnM$SWClbKf26dT*G<`WUEt%%{CGMST9Fhag>`N^n?{Z(C`R8k
zEdfODgLN@Rj1eLmV+3II0hmJwby+n{BN11Z`%gY~`B#7KrP|FL*q^`qwlc~4AAC3-
z&&sMB4pxr#Pv-OL<C{14jt@_r-nxGMF$CXFO)4lM@LkK%MF#6!h*1a;e|=c6Ka>z!
zDM`X;B?JLS03e9Yd*?z`wSZ`ngckrnl_1H{aK(52{dYeovfit&zFy7Be%@POUq5~3
z%+cZD0>q!C*~aE32s)V_e{%a{5On73d60BGoqS%YAq0S!RP!K6bX_~2Hg#>wxt-6N
zU^|86TF;&dfK<}!4coc_f2Dhw$&8#;lVW*Tw)J#c_EV#UUs@J5RnxP>AFk{FO&{F&
z&i0>Q^dFo{(~D=^ah)ln(PT*yLK_JqA^-?UB!!U*0A&;^l14~SNk%~!ISNF{Oca@d
z6r2&JQlJ1pZ-hv}oFl!3;nLhbzh8ZamaeoRw_bR8@8jKWdGIr?e}CMJPb7Q6B!xf}
zh><-qb#r&Jw=?J!5APq^hIj7n-o1No)Gv$_bv?^cQ%?6KMiNv`>y&aW^wIIW?6~ck
zwwjKHJs0BCHlW}y|HJ?Jmw*1f`v-TX-F(o`e&vZP1&2`2UwZDj`?v49&$^g@UeUD8
ztFQmy&b^Q7ZlX*Cf7wORbaqnLC$s7CWHu36BzY>09M8r_CkH}c*EaL{G(;DITUZrQ
z3<^*SnQD_JNn{2_CAHCsl3EIr7lqI!GDc=ndN!Yrrzi9Ics4nn%_mJeZ`-=->drQc
z8V&%&7$HPr5J&`=nba5q0E<MwsHZRd;Ib^++S<ypY=PBuf6o0U9GJxe7rRcHrh~y?
zb#--RWrc_q*i38f!mQLeTh}w^I{@&(d+$T^%-nTth%QDSqeqNVQ?EZ*-P~H-Aar7k
z(MTAANBaL>dXcgLKnMvStry$puf6#4uRrs}mqu%+KxhD{>(aSy@nkfebv)hw|NiyV
zTTeH`5fjtRG-FJ+>F$n~iD8bKBPTYTnZsea+n73XI%cMa$s>l}>+`$azW<(oUbpkd
zv##rL-7m-%R(h9@m)-Vt#9Tw5(M0(SZ~{)zgoFqQmSYq=A=q%9Oy_(?`DS{0Njyw%
zKs96j1$Y2*ldz2@k0WPl8<jsuzg=ilxL-Cw6-dbxXR8>n2U^9_zThorM$H?<P5IgI
z?n&Ypuqa60OeDB#de<C>ikf@9&UUz4b9pj+>)mm`RIjJtzcuoBbVc;VfyeX&PzG~%
zcjvy93ACX3_R-`w>Cu|MQ-NdaOYY4*5693}{?QU1MBRwN<2QLiFP=dj3qvY+cO3A0
zPeP<?q}9vlX~IJ@oCsfgcu0K6;)Kt(9aV7=uE9g;V}hF~wdxcnGVB9-u|fOQ-2!9M
zKovS)+;ToW-zDqm%-T78etWzMSki+jrhhHe<~#(J)zwu*0!a6#mR+XD%qeN>It_c<
zPiF|&GvYDW2xDl|mD2yNxg*It$okX|<kG)!Y9>BuuRf6OeFr04o7j2$xPo<~gLI0V
z>=BuHHL(eEd&u>UcUR!iH-F&;veoUIHuv@opIpa9@KT?enyOnU6M5SUfHF~7P(#PG
zXwaIdSvR_M3-9g8NsERi@|>f6Eb%juHeYFfEG4^?Op$k?IY-QJNJ-gT-rHTY*Q%uz
zedzJ1+giM?%O>61$p6MHXe9DcQ5`@xU6Qjz0^eu(u>t$}YD9O4^QxAT!%LBN5?+jN
zW3)J6dsAt^V@t(e1B|bLiV~T<kAVTShzzR8l8A}L#zvS3T~X$VwrMya(8^xmS7lp5
zX=Y`80vil2ex!}wD9(to;(hKhc&TFE^O8W3h`cc3Vo1S^38Y%j7j=ehmLHduGq(df
zYh`ZL);3)&@8{0Eyn{WUbmz&#E!8sekwKn4zKtZCC!(Y8nl1tFsDLw{<J~;&r#V8d
zN;=>={7_RU?SJVad$)JX@;yqAd)wclX2Kk-EdJx+Wgoit3(l#TmNumIy|3)g%6;z}
zk!enIjnD{L1=*DYk5@m=LZyb|6P3S`3gg)7#*r&iJU26%YoB${)-H`t;$l~$()p4}
zl~>Q+JjIc)5U2<w_89O$F0hVr4}Kl{bsq9#Oui<}pu<qcx7wk$8d3Mtf#faH&Kz7+
zH4;fo;{M${E{fGicKuw!Gc9^|+uhN^BU^Xq6tu?7%&b$s35RzD3yZ{Zvr-YQr^j0g
z;waC8_R3ho7r222#TyQxp1$e8lYQv-u{G7#`r=-mXJf#bVCTPuOub}Lq8I<Ay>!t~
zK|P}Qxzguymm#cXhWb2n;vw$!;q%kVCeMro+Kb@ZAwd#$PPVeSch4=aoE*)}gD?1M
zI<TPJo36&^Ywaf+74^>E5vQ8JCy&uhP2yPws36VP`1%k!<pkQq*U+f1h5=T5jZRdz
zGw&Szj{4vLcTrfAgyr>8bJX2wmiwqX1G&&}-NOYs%DdJAwe4smCOp_NU%%GuNi-H(
z1PUt=erK6`9!~Gi$MT+mL3|aQM0pj*hWGENS0hN$-kFW5lw~$}t@aK#97ed`HMp^t
zn3DOF;R~A|BmGe_JN)e0Y_I>CyRyakcU5V*x9<Wp2TUPTeI{CTg=L5;r<|w7-Tt>h
ziN5y!2Uj8Q9lvc%U9GJqutZ&H>Yszo>fp3bgQ%r^<WUFY)Vc`F_E{6|?j$MdLWPqu
zv;_W%!eKpCaW4(PDpb70H)I;=2JaG;Z6RW;W(ORf>Yr*FK)=e;Q>(J!1OFrxZgw7i
zbpv1%nWzZa6nL^WH8cc7L`2+Pg4)u;!NvBA$m@|T3|}F}wYJD`Pyfx0qx+pK+H{^0
ze!SM{k+R_qWFhIyu<SepSHsAdv(;bzMc9kYEUi;n1_4sIaLgC}QA6%skhlN!Wnuy;
z>3di*xd8fSanvT!dTT?F<BLmXM(c>5K;2UZa!S>VmBoW)a_l5^H}?bC5+P11EI;Bu
za;AeA_?Xdg?C;qm`+5e%!SMil8*L(7w4?`K*OAM(G$lrvit630;^$-75EC9JMaB~?
zE-;g2#{N{2$Y>G&4c~_YX^GD{;~p0aS=#1!6#+k{#>2tsQ1k65d<V>+a&DPSfRZRg
zD34k1^B2{r1D4dCA;(S_x5rh2cZWQw5+QR+-Mxz_gijHb^T=Y^qPwHj)Zr~EVQJ~Q
zP=-Z*rPaAXR!S;wd^Gm%xe6^Yo~k~E&lKNM6J%5hgIkjw2Ma9tdo;{XM&4u2m-x1S
zV}F0&Rw$&51cwCl$7#RkiVB<poZ945*G&~WMsk(=W@e}nH{<Mj1VR}<nd8tyggpAL
zBXKRKK6FNu2!N~8I<Qgu{EYYW6{7teS`<OAaURSx6;ComTiQ5f&R6lCbAy&<tdiEA
zxL2A%`OATy2$Np<Oy-IHGN_55c4n&GV}jXQ$Qb?q_6AvP+M8M}FDw8{?2vc^142kV
zS+YP9%}R`(f*@^lya7XblZ2k9%<rYSA}ia2&XlQV31kuYD5>SZ!p5MCL`2yA&y72o
zU@s=7<3`WT4qOa2__WA`{|&#19fjL(OZ@6BG9ey^>pw*EN2}pGGTkT0;*L|Tj4gIy
z>rEM!PV}M8gOgk0q7~2?5h$O>*Ju4XrDsz5T}C0%oa#hP=jlxM^#)(>+qv7kScMz(
zL)Y$qD9b3P!_%DiQ=u#_9o;o#>iti$Dkat;+fSIYH0v>At2}TDN$2Mpe)o5V7oO!X
z18#wVa(KZl&FxLHme%x~=0mjYWXrT^dV@bLu@sYy<?V{-9jbv5^CYU5M26a`Z{{?2
zCZ*Q#lz6rM1+uaY$oMdOo0_~>W71}E5&Whsh0Vg!n8x(`>1e+B{qvPg6SLvt$m8`a
ze;KKYY8!xyfu9MgHxi`#wpt-F(tI5IFPzT=(Uh!h*45F_SXaP-5zLDHd_G*I3hXZi
zD*$^-6_N4NSU~Qg40)srUr-^Jp@A|6;X>0~+d}+LY#cS>7CFX@dS#*L@Y7m<g{#uz
z)w`cN)%hJ4gL-<NV5%bn>rsDKn%CLb_CZaTvF`TdHz$M7T2HjRy}beJiZ}SlpF}!;
z|8z-Z&m0kY*E89|lU4EI^6$?tphi$O%^m`rZ6gai697$w)!+KPhzL<~&V(c`87r&p
zrm8{p$0fl1H^G0sXzx2gAksOkXS7CP!OaNa08$fr;|pa-Iil@{$lR7=O5VoI@P^p#
zaD3(dtC*^yXiEd8ap+3ia0(KS1{xl1hnywHPqaaXdEhC1VP8}_+?O^}e%tXLVX<UL
zV<s!}090`<af*0KEiYTYeHX41LhL$c9o%S_%CH=-bt*iis1+gWFZg=a6_~j>?9yP}
zXeaCazC@2_QJcjzSiDOi?+?>NMx&5EaoOI^nHhr&y`FB7WVe;#LEC0nhy8&sw7YEj
zzh^wNc$7XR`JhYrm2Ii|X*sHiHrlG-1@{XhYQQX?lQ+Oq|2#n7lnt{NAJssnjKkK$
zlSf;~rOanYkm<7A<bJRKvQ-`}3om~^JlQyk4R20MMSu7VmeDMWdK&$xZ_5)qtg3Ug
z-{E=wySny2$S0kBj)}_8o=E>*g}T2!$#h?EU2R2}gI`p_4V3BOM)@N8KblzMv}Q#2
zo&a3jvy8(sT<_l-W^l#i#eCb1sb{<TreXk<S6Oa&=FSJ{!DqvYP5#{**yOd3K#o>d
zwHu#5d&U@=XFf+F#!g7Uno*_<r^6)xS230=SEh<uG~#f?)oh&iZ!{(wuk9g|FRYiy
zl-L4yg0ZlEE5NUzyFL6^H_~q<;MF7Kb6_X2FujT@g1W~5&-Oi{_Qg*Zr~%%QTnBn}
z{r)H=%`56w;$aE|AKh^)=u2lPKWy1VJPdQ!fCY6QXI{HDtTm)KQya+82;<;gKAjCj
zQu^Qjlge~~Q{kWVK9bn(zWh(;VB@BuE^saUNxaIJ(d*%jspE&W?(5_GTcEL_0U#E}
zOD)|vof>YDpuGZ<_SH{j{2F8ta%ICy>MT))IpZNB$YtYnWr0q#LE-?dt+j{C(~x&q
z<ez6M$fjwT3bx1$^Kp0Rlpwg476)wz-8#_K(!fYpTx>#4s?is2Q<OZRx>M4lHwuiJ
zTk6+bwF}<Z|MrOVDDrwMCm_E7041J{qavd022s`Qvc1d2`rxBA5V!pA`h&tNtt7+9
zjZvGMpF_tQ9K(b0tL`N!q;{AzkCj;sR>jTTxnS06EFJzLQ|6k;tb-~l_CbxrXh!8P
z?vK7?zwLZP%Cff930yDPOACwNG>%`~#olq<AjbtTS|V>RFI(&57J^oRgPQL9u^Bz@
zVkuAu-p~V1mo8h0yzl-9kQOuiM^0|6Cdc{V@)k)%;fP_m`%sC<N}{fK-7pt)7ZtWK
z9p5<K*t`~I?{ANbp~v+kUzCuy($Brog!nx3gOj`?LhEAf@D-cUe04rcWJFWnr*vV8
zmt=!ggkq!@{Mp2e(fXeN(v|q#+kOktqc$(Pc%`}+(f8Hu2Ko&BiqRXAgFH=Vjc1Du
zVhS=|Kc>?|e`#G0pPcx#)(EX81=+(}*2M+rQ&~CGXgKW7bw$E4TQttGq+Tz#+1%|n
zuZKOP^nO%eye#SlD|uFLSypfU_HE6$jP)0aQ<(kTI`Kth;=DHnmd$U)72~)h{L;|}
zjKSPz`LmSTa!o<5={rjW)~oWe3%92<-9nxS(&H5w$uAzaty5}%?*Q#+C?DL=pbF_p
z#@UWWCv{Fw1#xRGVdET2XE`v^3NtbudZnT~JUp3J0E{Vey3;BsHCLv2=1cOA?(K#_
zN}Aq<%}uUuX{-PR$K$}jd0Yna+*no!oI;tM{S%>`Z<E^bQd>Uu^O7=lw)E1oH_~~(
z$2w@4-U;iL&r~0f`GV8WAu`2!$i`$D9Fd(+d0Jv-DTR8h=<fPItR*Ep)ODRG@j(cL
zU@FeGy*f$w{8lP95XJVMwJ}%8Gm^%h>OG7iaW`w;_#D9N5Bq9a$aq8cBm-ko`3-Dr
z3};NAfSjCDA0c$}G{hbG4w<z?Pyy4pnn@IX{Zb18d7@5+6`0>xcp(pStVjo!Jp?0N
zxu@Lfz=rHs?U0}g$Npz36;$x(K|fK>#L9ZB|HDz?L7GkAoNL8e=<O=IU{S=cq{N2M
z{XQbW2%yueRM!uWvmy_dt=ZE8EcI*9LJT{I=jj}@*>f%%E`vil)!vTRA1fHxz9K94
ze#eo0-W)#JEBd1Le<K4MQ^j|Cz`tKa6W++0Cu%>o&NE5Z)2ds;4~&V0Cy1Od3_-Hd
z+4aNd@`rDxQH8;;cVw1<Ztc~j`RX+ZK)k(G<Hfu88c8KQ`tLmjvIWwZr2gVybJFP)
zol@ZR5{vY(#_B}le7E4j<6<WENE<?h*l=ddDc25riGS!cp6@bQz3VPI9PRfEeNVF=
zf1=bmtNkhZcy0X}g}%$KT0Nxvs0CR+doLL<D5NZ_r@;oR)pAGSpyQtV1*`;-0I;dj
z+A&k4F47O@%}JfXpxFdzgRqglseOY8itST?L*Wz)iE581LWsbLddDX8F8TYn!8K!8
z$b@Y8(NU)iSSeL`c%<(i@n@|+{4ALWT0eAfFkUMRzc^ezP7@mft>{7e6vFYOZ;rZG
zH-pB;)Jv%_gE4*cEX4P$<4_b}hD<Rq$5+Cv5=JfVI(OwFcq7+qc;itbgFLomR|U^F
z2h$N^GIX3%fBqHlBJN^48~BSNH*PjIl&IqEXo+wBx6L?&hWUiBmZ8;wb9F(GgoV+t
zd_eCEo0_+|x*5Xy@AmcsaADGI;*z-)s-Mj($Zhf8AwHQgX_*KY0bq~EflC>BbKaVF
zQd@{LLR|bx%2a6eKNAwuEl+-y(t_$29xlW0D-$x)8R+{r>vQo_izeSdGJg6j!Uq3H
z|EU8-u*dbwt?F!Ie6zHE1yif`z4M3F#ZGv#a@-;7zs<t&_Rl)opA~eHyLDxZcKz48
z;itjjvB*zRAs*|CNPrnlWKvV}`x|R&c+3T)WQcxqN=W-Q-mk#S#8ri;0E^5Glc2nh
ztuYm!1e&k=@bc{?*!SzsN1Dl*n^%-uRAd$2V&pQ4P3x_K`DDVXB_v?7cEs4o?9+*|
z<{QGYfkiv>$&bd~H%+`}vNNbRqfYrJ_r4FPSY)u591oQ_HZT#npTkwI(4@YgUQk8b
zb$wqv=Wc)IuU!zze0DHp$|;vdIUFrA!efBaLf-KNP(Qj0s1sA=uRx5r3D6S#izY}D
z)`etE*C!$n><TzB;H1fRMbo=$<_%kjp_*$IRDaL~Ua8;3D$R+*BwdQEpEE-S#G}aE
z<sdjarCVk|shIHlr+z2`HdXzf6Bv=cc12_n|E^9lH!gUJjT5;?IDRVQv*Fl7$y5fC
zl7?$*jj6s_v;LB#uRS1OJKVhc(fAmH8;2<m^OYI_rt&0tz%IV;oU6T<c^?QL*wr!#
z)hu{S#e<0lrC`N~j!X3t8pOptAj{3oEh5<D)PAM{DAQrvma<JGdfWd;ffa|rjt^=1
z*e<gSK5e1;9~JRmeq^NG={vid04{&<-0)!Yc_n4TOtNs5G6xj=k8scf*R9Eq!rL)I
zPJh=;c|}GZjOEp*guRL#CHw+Q$xK2ZcpULzw_P*OwL;;6&C7Bi`N|7jW8Hbf4P4Gc
zH3fmHhm+zvF`B8J&)c|S1$IT{3E}Nv@-Z}D6UE1WzXv?5-;II@Mw9?5Ek)07!oo%>
z+d453$Jg`1*KYTR`^0J|8lMMA_zl9noA~?}qrIV!wb5fN`vML!&RmiNjF&L@-zO!>
zeLOE_a?N5>X}QZVJfXGb?g1JE!6-Uq3_w6gnbwa$gu<Gv<0RQ;bzxB&xR@Rreg)je
zHn%(qS8tz5u7B7EXNa?@X(lWsS-wqeUk>U4H%HpV_kU(r{;ZN76?6x<E*!ccM@w2F
z(U*N=(^tR77RFSZbN8ysrUiNb!}zr0Y<A5l94#{VkzafLj4VSK(O9C0WwTF+p*wko
zgDoFck`33YuTu!%){S&M{<~{C*=R&y^qA25wcGob3yX_j967(Js4L`pdgNwPD{J1S
zWsSuF+~I=5I@gYlg8X0q^YMS?DlY#tC(Z?v6ZoNywuv~B;@Zu7RWHF@RU{`j9wH_+
zBR$qezq;024I9>*W3iF;|MWxVYWd!dO4Ni|{${RT(7JkFdMLr(I@0RDBcB`Z`8m9=
zThF9EHo+|Ew*J*l@3G`6pA$me@;5p=FL0iyDY?Bvd*H#alhZ`(|1-c>?uj+@SFp1b
z;RNDxImweI1`qTRbzQzNf^w#pLZ}cJ#bBdqZ?0113r&3D@CxGVk^Y3whC|_pYKAuB
zY0HS4n)?ml8o0mg_F`cv_#hLG-stZ$Y-HVtn|j%Tz=#>xNn&ZkliPos`z-r!-Xg(l
zs9afE)0{+m{z(tf=Cs)>qlcuA=!Y{qsPGHjF;30S?Ai9!DCdXo&Tq+M@P0{F`OWLT
z`ZXg|Au-tB7`|*#Sm-~RW1dO5=p*?uXWkW4027vj>xRkBq}IkZvH()Yd6?+QF;9}d
zZD#86vsBbd?m&AdC~{QXFa-d657~puQ(3eWoazlLrUOm}S+WMfMw~DfR8sWE$7M$m
zP7PX2wcQ691u(G&!3bP}L4^|nUc24w|8QoOYgw0C8mjvxX)gH83CdfU!TzU|H~i?7
zz{-`7s)x|^roS|S4?Sg9X$GJN+s*S7?$dzNzKa8f1BOOU&1A4w9!Qz*`WrZ?3>TnK
zRL+?EzUeA$8hY;Id2tec+^H@VzHvizyb=-&n;nVoxKG+y2YXC+CTtegmN|A)(^Bh1
zvy=m53csVp=jX-6nY&MZ*a#Ur_xEluX~}`dRnS@Q-(TN|lyS1e*MS3doCRAUvI6c;
zq`S~LoWh?_gMADw01iLahDfhk8yk`kY!MfZ*25&Zd)Sn)3_;q-bLFb%5F9Y9K|t8g
z&r<eTjLt_%ujs>r_x&<3rtDQA<9jo^Ke&H?B92+tBDr{o?z$LV*ZSq)=;-L|tnTyl
zPLXc;FsPot-FK4-9#Ak~vAkTbO;l-NVZCdj8XDo_ko5vci2obOX<PW*{hh(HUdWCj
zWQYA+ZpYFnuZIS~i_cCGFsG{jC5s)07S|re4JRHLn;w8y>IK&Kw)*erRCHaBd&`>J
zFc2p^Y7dV~*DZh6{Z(krEwFv{$B!R#R|7thjORBA2??Xzdd6)51CggU5(?BVlJaY<
zzMFcq{&Q3SG+l@|-8Lh~bTRs*HcaltqiM<u=@6{0lONtbzvhgOtfd31#n#$_k3Dtw
zOss*ejT>{#^N;+D@S=~ruL`EK?pAMCgMg|~cl7Sj$FQ@F%k?VR2=9fvnUoX-)Ag(R
z(tf#(i|9*EV#((3|4kXEr#~yk#>Y@jgu`x^N@1-)A^}^(aIT^Tp8<O<Bj?T=e9klS
zFM_#4U#XL8HB=1BTD6Fb++LFrn`4w|Fk!v4n_^?dwS{U9w&)pn$}CQbynRz#A#&ca
z+`O*$Yk{XOeAXL1Y9NEdA1F2E50PCzs}ub|>wf6zS(N)>I5{A!t@<5mz%=EWiVk<g
zFPbq9P}-Zk9dquMze7(??HJwklRmu1a_A2TcCY+icW+m2=58q)d_4_!Z@*1&XE-2X
zM7OHW$RAOM){1t^iV0U{Rg6rV9x!4#Mmzl0?V)WP#ELh_Nm$(++51?O*kgin0E6H0
z#)GswRoJxpy7(B+%w={8o-Mf`5&SIs2tRrN;yCV<>F)%60@Wv+Ya?f5D1|)nrNCAC
zSv6i?wSK%4THWa9pfVwWN|gnpyamEs@Sg=g4jMTtuqT~L>A1+9mGLLc)FA<>==nKj
zGy5`bwWdK1>MnnjbS0ARzexg25IVvp=`Q}E|6Vcc$wzyb-v~Y#29}3P$Ik)``&58H
zQ?k4W47{{BSehmX8BEL5@#%A>)@Gauf1<)t_>0vF=2tsg9Sw(hghmQW)!MZx1%Z<k
z&TAfCrgk6KUc+qDK00v*>-PHFC3l+;kU=UgE)Lo-b!~0N-lA_~H=O0~rayRhF0y=f
zo=}jRGC%DcFwVv@z)|d@Jv#D`11v++NfKzDT1vJ?h*bPo``7zHF@#nEp_RvuoAt0e
zq#zAk?|F4AFrLqd8flGBpw4d!)t-euY`0q*%J^^VWS}pD#Z!3o3QZXsmK!ClSC_=H
zL-ae^fq{6JS@z*XeflX)h$LLYF8y;N4h}l9W_1syMQOX_;tr)MFy&4HX4J6Zb8YuN
zF8sVm%xHZKMK+C2=2?cZnCr&l`_qauB0ITUtAage?4wa0l2QK7h-=Wi^Pn9fWs#MY
zmc8%WvF^M)Z|sgd^R@{&c`^tRi(4bW>Eu&C{VWhV^)+%V;+o~*pI*z3KI`FX|4~$q
ze_l46IHDB!T%G3|_YJ^=!%A#hWd^eGeFkak=U!rr@JI(x6VRpjGj2kKY2D+wQ3b!d
zf)AH|L^UUC>~A7BH!sJ7y>xVRd{8VkvOy&4k?78e?wkE^32%Qjn~wWF@Z(4KN4AnH
z3M;_HkM)11AML!o4?w<gGsLHS%(SvP!2B`&<BPj5Jav@yK&GuIWdn6RQMuls_t7}&
z#d@fZzpGP)u*5|y%l+d0$;sWtW3A5Nq%w&=ETr$|W_1)^6vE|Cs;GtvOV+vYVj2IY
z*GF<L!>FbySAwY$9bm7pG@h`Le1TCw;3^ntWe!l*Xfa*~LoGVrL0nU<$9=oi<#Azg
zT59oSE3JykKs_@j^A44kWAca*wYm;I)_`dPKHe0$D2f$RE$@fvm+6`xc~tB_dlq`{
zKfMaO9<cnQKWv(9(#b}z&S#$s8=T@Hd{^fp>V`0w$$SNR9MAU<=YLLL&#`Pjfrt$#
z!Qid~4ofB$?fo?oTQc?a0E?o{mp_kK{_WXL%E>Q`0|}jByAv<;4Ag${a6ygQ7`N~2
zrf}cRnfKztJM<Y?yvH%{k%88T5}pEK#CN0^jA92m=a{`?mO-Z|m=tSXm86|4`(g%x
zfQq5aFkV-24M^Rp;L7gzHX0A|R5pmUeSg<O`nkNk*d-C8<>z4T|G>fA1CF6#At7ID
zyY6@Yo6;(uF_W$ZL#mJZ<#UdGG3kI*&=b?#y?#T6#WVGJVbywlaY@C!<u^><y+GJW
zB~bz;?M&{Aa|adb#eLEP!6qEsH-H3um)eA3LK&l4pF0d-ktz2QXtrSJ2?^Rx6UFw(
z`If5f@R#`Gmnux~lq$;gmPnq)H8-7xEGWD;z`~R{6;V6Qs);x(eN!Y|8UZ{lP)<ue
zbJpuV@8dQ(0z-l5H8-(A6Xq%%Jssc@o<qIu6t;R<11cjX#>S{Z`FbHanO%~^er{^4
z-J8z)qcwT<yrRN9o7~mq<-8BYuA&r!3#@#MTcpggE_0=R3wM7}o}T0P8~1iQptE9^
zt*fJR9#;4behsp<)}^EY>HH5igqhnQK<nyK4_G=tH#JKzH0JLljgP-jAdX@E@`RwA
zia=yU5}x>U!XM&LjVM7h81&n@L8)W3A?ojyk;VVsB^trWh9{hYR2Uo!4L(CDq@<Sr
z4L$s3;S#Ii2XFZYg-1$^;fZZLH#vgEYs=;ewIRPci@So+vEjfQVz0H32asFE0{&t2
z^6Fy6-khns5beKRZiV|{z4iM?-PfzhdJ6EATjz&=#qMe)P9o1_<m80C8p(BhJ{FOI
z4B(<%xeW5<H~y%_;aAQLQz$c3tMHO};Z{X0Jvju*-$hpra{ZORgZl)P%H6%tgujX(
z=vD!jdJpP=dpZ`dl{G>6G3TkAn7sV;l6#HS!|&w(VR=Drk&mRT|Mod}`pZ9cQT!E?
z7^bX>nJbdhr=H-;gKv7|i+qNa&1(LQL_96((4@=%05zUf_~KAgc&$J|0)2W=>>(gd
zQh&KWs4D=q?>)2brW0B(eXfdk%XMh$WvRyZcmz8pnJp(F0FwvvB<E`YCL<u5P3cP}
zLe_)PVm~V^{mxgtuYe;JW6R(Mv-zp7Y4iE{ACLOAzivtOYL6j~Bjs88k$NAW|2aRa
z5=uBM82LLNwb^c_?{&;Xt3^^DwBjbaw8kV9Rrcqu5M?cxjOATH_ms2OkP-h9O+}K5
z+CyL@m_vF^X*C?D5wum+>+OI)`mOKc;@N0pSiyg<tlWy@lZqFnKPu^kOzXL&oH)4?
z!hw}z(mGKSBXerk4AMqk^1HI^Sq7>rjTHs@TplAgE817m<pr-uUYKp^y<9WYDqjL`
zUJ0({3X(ylF0Uk=;lI0%{ZWkcI|Yqgqa(WQz}NM)rKRxm-Kmk0kyWGiU~*5|cpO+Q
zow8=$MxT~V$Q$BKp~Qb>)VE#Eh0LIHs!#UBjxk}HvS^6Fcj&N`e?rn(E35S33*z0>
z#P?vO1FZ^!1H`nc;b)#e>m_sl4tel-!xDr^4=iSVM*4Ha&wr1&df?c_6%n!M@Y|{Y
zCwOs1uIL_?Qw*~dBtqf}<(e^4MOYPQxfM<fYl1O)5OvbE4K587(6r`3nL-puYFd4S
zh7*HRSjk43<y_PwbvVDG$*@d~_fVH_4RFoRLL70Uxj@QZq`lC0g3UqG@=C8;7mq5S
zwX@yRP=DLCW<EI4-1TWl^ZL}>@6g8XX7Dtx0sB4FAFK9+?)U`z+F9cyVMm3m%sG;4
ziLX@|sV0#?a8Y2<9}c7RN8${es7WT0Hw2#izCHNT&Tk}jlf(z54a}+<>|o?r7>1Q2
zCEE0-tp_F#2f%4s)YYcM2mA5S96`X(rVBv#e=$xgHN`pz4s*@V&LWIQ7uwukZq?Hu
z>V6m0cvnrJCQJtT$HwpeCV<-&zn~yXaN4}M1pNgl7@@hjxUn}>XT4b(z?(Le>eayW
zELvwZDd!&Aa(?32%ae^Cd^yb$&2+f2C@Bi{TK`n2u^~#Tq2L+U&;9xwARj4>8N7Me
z9ErMJRm!8bxjQv6VU`Wqn-O$0%cfqu@9ia;uPN`}Ku7)Ejz8Opd`i&Pe>v5AqE8!d
zQWJS-|K@FMERv<`f<^b~i<gY4Y4MKwIE^g`KT*@nB33oEVc+bw5dr8#=V?XG{rtn7
zjo8rsj$WaSWcWU6ynDA1K<gEndCkT~*ZSr9D~0zfo2WSED;@4i3TSg0M?c?}e6CzT
zj)g-LJ*Lk&Eu=^=h5;vzS52eXC=ir$7bsN`G2oE2In!&5sa5oec`wYGjje%_)yjY4
zp<?OB9xWOnzpTYWucePaobIv-jWwW49w1Z1&}PJxM=)Sg#aOlh0l!-Og@lFGa6MR<
z!5cFF=`4kxk53s@(DeRF-Zg8Y_ErOjsw>1U@v3*60Y#6st140*{&_mST9hO8t8u;8
zIBOzDgDvkX?LIx=C}*&oT0T20@_cyoe6YV+mMakc<T2415i=TxW;!QMl3Hl<&y=qW
z1EO|Nr{}$zWYoR$EZj--Z7Od{+{-~YHX)R13AR+0Ri~)Rs5^%G&N}HrWaWyyM}2oJ
zi=z{EZkW?@R;;Scs)YxR*xGq)RVkMs_3fzU7+xsgx2EfaKx<2QPCItdw?l{h)-{<l
zX4xcFuVAfpjM+JPp_mYY+{^WF^r3b0c+O}jCL4CY5=+i;D7Xky!dl3mQ$>mQ(_gFM
zKb5qPW8Ip~5ji9^Hen`t^c=5LRTQff^Z^8)XdZI#j%MreYxb7;XbxD6I|%JoBNcKB
zKP*{V6zAuDPjHnUtLjItb|x(@USEInFE1wPI6>g-;0q;xM9!$Lc{P(F>!&GSWzM9r
z)D(i2v%mjA%M8GQga5K`ONdYYdaSWD<x#?NWjI6S-R7nf@(B9EZFoS1h-U+1nc<*d
zT5+UNopj0=g9mqgNruAuB_u0|wA_8I^%ZwhFuK0(y)s#)6t>zVt@azK<mt22+vloZ
z#Kg-i$zp7Ymsr!yjJ~C83-J+<sSG9Z`+0)Igoj@>9Z|sD)((u%`nC^5-C5|<U&`D$
z+x&x_-<jg9-=1}!I{4g-wW^aIaa(_FZ<|{LWl^g|`Gq2d#)kESvQT-Q8PG>5?Z4aj
zcsuSLxjC2;+;zS+Us>Xz*4Q9Zke_d65o!7Pc@_zuV_`iElij!t6sHYrtH!hme<#;O
zS!XSx0F<t!JIOP6BVp{<>3rdzF`-@R5W-;xe57%t-)kK$p3{GSl4e#rkNVCno0m`e
zY{R^~?Cp{IHDLCiTWNv5c$blh(mgeeq(5xBfOdUZ=4!UQ>#kO@-zVg6v%=M33pI8+
z<ypOIHpd0OhLw_SOTg?NR>Y?%=#%?z$=bR>K!8!B`Lpus&VaAs=sb55t_e$V*Y0KO
zgB>FQLS9cq0k`#Lh~R|5M?<$me4m83Sb;bM7#yrP_V_XQ`;+9O6`7VkYTq=Ju;Vbb
zMsmf=x!iZkf8~8<Bx2$gzf_ewk07*HqI&ksiqt}^Ta67Xmqm}AfQo;!fK5wFKuH;v
z4zR|TBaMtXlrW^}+pv@Z)DC#_v{DmRm8wsb+!{uqsqQEL&If+091cvY2&5X%k-V!b
zVDxsXU?-k*L-m~DqVadAm9B-o=kC(eqbVVnkeBL)t`UNNNb4q|PL?S%&selO<!@k!
zj*DIHXwjc<%KV(%qP(2mys8nNJM9z_0P1zSKYVulBcWT|(jwG2t6@Ig@&h{5|26w_
zRxPbs#?R^G*Bk2k^-{pE-@k^R`SP(rjg51RvU1|}v+fsbbs<*92>ubv3PMfS!s)h`
z(negb+Nun_8pPEetM=l!H*(RTq0ilG`71%gX)hqWQ@0yfIS9Oo4bpcSgw1Wia$+Mf
z8Gw^H2MTT_vbXxX`0ORZcuA|v>s|i0+aK3CTpFI0`erl^n%h@R&E=9~&q^szk`3`Z
z!2uigTFg6pY0LH|dA{Y9De{o1@jv=SD>TShnkV9G&rkOOKl5$6JoYyKeIZG<@LL?U
zXJNfkjvgwXps`<yIInH5{%$;nH?U}^sIUSyw@dQ?$z-v#+=P?{NixyQs~MzP<#s&#
z=nN&dt`XvwK~PxdHKle<*ZsG>yuH%?qlzt}Ro%P+dlE4UY}?8@$%W;m1=IC)kR21U
zwGo@J>UBJl#rHq)Di@;8L-BP`OZRxC>w5MZQUB)M-Ng^T1%{v{vu+C_xhztS3E=Es
z`m@X3-p<y&ip|oy1eP_r${|O6!|d;@Bz@mu*Jl%!he2%??YB^EPLe6ilBT!9c8tCx
zN?)hx(_#Kxm`O7vk6*Hq9N1roPB7VNnfNjqWd(lEg%wAf?9;4=1Rnc@hdbMcUeC%L
zcjVCg@a+yqPhszfJ_@`1(;8HsKnzUtGqrLa9kQr2woIzMWxM#japEhG%CZX9n+Fbk
z82_Qp2c5QVeqN9X4NhCC&&*_|k+Jv?tEATY2uAOmM-~@do9b&vY%rK2?p~Q@VQ=4Q
zt-wGsWqw(Y<4|g3MuicFMSul_dA)ue*5`7uHXgj=ynWe3iYk%^e7@#9js!Sc+PnGU
zi2r@%hY4brRZ-6y>U$j?vL+P9#%|3x3)<ZDf%cF3?T6#L^NAy*hu#E=2`wGZn_}0m
zG{r!bpuXQC*AZct_Xb`dAgG(budoEPw6_EM|G0InB3(@xQ&WnI6>j6<W@e<WjG3<&
z0_Qk#J-oMEwY0VQ?3pAJ03J9eekv5}vs5{8;bSUVI90m-2#;NK3`H9DBZmR~1k9FJ
z`u3KEg2X$GL?<e-%r8cy)Z7eW#SQFIi+{N(y0>SNYQf(iv}LIVR53)u_GuDsy@r(#
zx(Y;}+4DDS(g_-gSq0i^t*oh1Y@BRg0yp?s+~+uo-zs-JtgUtecTwf5k9bQYo79L<
zh`QW-9qJrz;xZ)-IHx>%-&bUc@%Egkm~2ebt68FnjwmfH%zRJC)%xLlo~!&pUizkF
zEXr?ZJRp}cL7%KxI#8q{e);x;qY=B0x|#kGr|S#72?=`pEk#P2v55k?+K3_paJHs<
zW)FW)zgJ9Y|N4TY*4VW2R<i_oKw2-$kews=29qBUcp8BecsaEfl$M$%-MM-1BlP_y
z;qdP0=Ynng$1j{Z7(&>L1c57?mM{A<2zcdmdkBUh95MPass7AY`eZ7K5E|t@#QHI}
zoeqJQNA>5>VuQ^|9xE!<&oAP?2$<Wah>u9e;=Eynf~6|dKIxz5%m(vn=g5<V2#M#H
zc!_j@SKk|6WToT_Y$qtXP;nKJwavN+ko|=Eex!PXnPrJTh+}WyqbtQmra~^#)YKN%
z(oKUWPlH3YcGao~oJ?!>;<Ifp{TMZy%fU=@6ZUZ;=W=#psX-E~D}<l?S{iiHH^#X>
zO(M#q<VH-By@oUn8YbL^^YNigjF}Jrc(5=5%;NOpvVzq%-S?;z^eIKg<WyfD9c2g>
z#+Q-6C5bGz<73kH!rS0udgOucIO|Q(PE}_xT33KDLWy8W2!FbYxTI|-;q%ymzE6nH
z<PizRmpSQFc6+EeilPkaZtnmqGtbd7LrPNo;>R-3<#YVcP5Y~mqdh<G=!hAf?3XIw
z?&A2q?VG`JM3~clxWZM@6CZH3>SWE@&*k487Akp9?xBacxO1bEX#@k>3MV2r8cTJ<
z!>3IV68~ptdjmT^KQ7YCNXzcCG)J=OJ$<SR6)PX)ernY+E5D-s2-~@#0^!uz%L75Y
zmi)t@T_pz7CM3?~jMIW78vq0fsAMb*@Ky1w-Qq)NkwFZ!e(D`w;U@IDYo{us$Rn6k
z;Wlq#rR`#^Vk^_+sX}8*OGj6i@S5A(BuO_GK>>lgpX1Gg7L>6^lV&XeCl#DfiFt{C
zlG60&{9Brxp%)eN@}L;Kuzu68a3i5}^CHdPuC9*%w2!`0)V<t@3b<kW@^*q)%<dO(
zV6x7Gci2%Es-#ya(;RR=qko_p_ZHub<9m|{i+tea3i1Q`Eu^fX*zqx@9b>9xEvG${
z0_Sx~CN#`|mAw=ejm-+}ON<&LdxHBm^_9U{XN6=wa;73~C`MZ?Li0HbJEmq2WwAJW
z4f9{M0e4J2Wi_N-91vY!;5%4hYw7o4;Qc3R;xZZ+N2b0n#}rTkyn%gITf1lUPW@Z;
z+=(+VO0B?M@D|j+7$3iQo#)lQO3$*q%0P@Kj;IxSS6JD=haXt<7g06h#}L;DMtWD5
z4q&Mrrz1OD^hkcb2nA+h!o%}&o9n^zvFRa$nlyPhip8T7uyES*ncO7S<)1L&U0t+N
z_rXb`dAoxurqOXSi~1VAAV)9GFhet8Q|R)75^HFv&`ugZMv+jDP9WYRMen7pYrdmu
zQMsAfZv=ni(=1-~(hmqkfiKKp1VRz(A|U9f@%}`}!)IT~H2Hn);MkvkzsLUZ=)J>e
z9E4OF8U;xw(A`Ca3+*yDqtn8}9mUSE=NPr)h9KWq&((5l<G~|4qn$)Lhvd<t@MVm@
ze;WBLzb2#|@fn%AGn#wC=Cs@$3mK$pTJ-DFN-<%oF$0*=L{Lu6aXku&k$(;!tA3CW
zA(PG(@qA0Sh5{WJqb*{68FS2}{G1t(N3@#;_`TPOrxQ|}1s3jw{bkuz@?l>)!9qN2
zJlI!>sH5HOU-7sAA#J?%dmIay1G7Ao!1#-av74y#%;xo?{35Rxx^mVQ>T#F8J0n>;
z{T~+(vh`_SZPV6=ikn(QM(p%2FI!kZ`?QP0_OxwoPL~xTZ<iX~6>d(Jvp&zsg{DDa
zB!^z;3L=`IVDucg-);^{s_&wX^xKz<3-W&}N|k;(eiQ?IKzyJ1bR2pZB&f507}UHj
z#auIwcc)~Hs58)fZ|eNtYa_Fd0kAfTxpI=FRXbyYXkM@w(n6WnUamND55#)JKKwK3
zz8|l#YKgc!oyl1`yuUnBa$L<>5Alq$du@L<wLORU`+Ki`=kk&?ReC7y*ZHi2m=5VB
zscwPv52j1N*~wYH<*Zj*F!E*_X^}E_B^Y`3!Iz4N4N+ogZApdi2pZ2p!NE0hccSLl
zB$XW>HL9irJFfc1{>?rLKBnDj5QZ5f1pHRsC+Lkcq*mAGwqW?colX6=v*`Gw<mDV2
z20luyN!i?J`YeAl&~x?(Wh{{NHufGY0%frdz0m^7*GVGQj!BJ~Yjj`eS)`1PI&V8~
zqe)3mH`1}b_Y~(ALXdO}#Vr9xS!uZRH=TD1|41Ly;Z*PKO&+V%KjYSOYF%EWBUGTW
zwc2pgxn@h-A4WfHPb@zg(nN;3`5tYjMTMOHNoFr9N#!`7H-32Swf)-yxVL{}?`{Eq
zN}iE(nSiMcz$oIzQg|pZ{{56Z*wQ8c#n{Iq_g#vU6B;?rWcRhcd!5l;^;5HGd-}!c
zu#<Zgd5Xm5_nIsz3$w*d>)k|-u@KBH;!M_{5h+z!!9PLm&M2#{Gg}V6mCkljbWm_+
zSVMQGjiKX94x4NEx|^ue|9{_V*55B5Ke%=><Gul)?8VV?k}I48VMQ44pvki1BOMJR
zW-scx@?FNQBKNO3y`;5oV!^P87KC<lPO3M)Uf12TVK}*jmk0@)8Zj?zY(km3V}8NV
zrNiB>dW9!enfG;1^~J`#<bN;io<s_2<&%W&ChU5cW&1K-hOMss{5T`?=aRIpWS)hY
z{Cg>2%GKCI+Cs*UkBzF>sTDfy7lIkQ<F<5D#;3P$@H5Mnkx+JX*D*2<HG_qUhjg6&
z>_o4D<>Y^5rR3zqbZ!sa52iOuSu>^{JE|Yjn7a#<1nLEJ;0aI!z_7sDuAjaO9!>l{
z4my-0{pKq!Ke#@-AvG0Teb%{qB+<Wkd{IE&zJh?Qdd~FsHot*$^#H(mef!qBIr6;s
z_86O$hAJA^<5M8r6|~IFd~em}bNQ_J+ORWE)I6|n@mG?n(f$PzDOvGN)6>llFM}*@
zcsO)0b8dl{jCFuD`@lWT3;26Kx7~IeTCQ6J3S%IUi4-_5y^fDZE3~*L1#s9)0b0JC
zq1bKhK*QIFLGrICrr@@blILp&I}9HV^l2H7j#-!(eOSx0a}cr;gN#f}hb#@7oJkRM
zVIA{()LB!523i>}ae`)#u9Prv$TQxBq?=H`@wdg1OQSJ|zmo;-R=n>@k7IlkBGKIH
zg|YGQ)@{?X@n$bSE-;XI(K^uykPUTAeGw24khe5L4mF0AcD2^mN<Q;%>7>-?UOhY&
zT3E4Eh&(l{k&p~Moozo}8=LN<SLbiq7qc)7|6pMgxjDLid(bT%e#!r`t-dxC9ZRYw
zPz=-2(VO6LZo69>epMcIKKGxpFcY<wjI)3<RkROVRue1+0TW+xc@C)nMzma?W6;{8
z=m&F;!xe>~%6s)T*OwA$B*8R0&J#A0HBh$~IwUi_vmpm#t!F7t^@u;4>S?$%zEl>@
z5_8NFZjG(it?i`8z~7BRWBWg?&g)m8VdNZsk@oP{ChbRzo-9yHy-cKM71hxdzJQvZ
zzAnNz!0Y@sHVt)X%Yg$$zcC*pYjaO+7yK>7Erunz<L&P*mccerWU-_%C^jQy7izic
zNKzW(PBOAsafhT_JTsyv=rDaybOBHiWqh)8TABj#hG-HDNc8W%U6D~O)`HR#sSBJl
zDM}roM=;agbSw%{T@zxmf+x{&WNOzu?1=`-{O0y;s=h#bBp@{TV>etQe&}1G_K;kD
z`wCG^|B;X&RqmTc#QZVIyIHcCqYL?vzb@!66174z+G%u-#FoFpQ@+%;N6ONgb8U|N
zy83<9>wEl{r_ygT^o?40L`(R|%HEtxz~S{-`9#;W3o4TT;%4_5ZP3Gb_ib^#hhn0z
z3Z){PS9#K$I}T8X95%>+XRo;|_4h<c@6f%K))_tDeWR?WBPIOWtl%nAtSsTJQ_gxv
zVNM4(Qk|4HbCKR*yo%z9JCEbU5zxms51~@l&f`0uw1vp+gVDXY40b_xUk_-ln!Xkc
zk^-5jve}FA?ZY|2=H<kubuDG2i8J8*Nrv<vkTRhLcqfwi#}i^p_@RN5u4qgt=V|(q
z#+eZ+v!#Q;ckGpj*UrxV_V$I*L@~BPv~I5mdeyb<F@#()Bt07f2c_L%>?n!^MSP`4
zrFaA62{>QA{P<vJnRbBuGN)y}$*z^lBD^Q1T!}cn`=ZZM3QLVKqq8p+s^IPjM@j8K
z<uYCYDSxCY5gp;Tb9e2*5ZnnGsw<F|q5N-35x72$ncEr&@ppB-ueskA%%>bLyj!*Q
zPA92Tt8Z;@?*^Y|7fX$t8aSJGQGxsCTZjORIAM$5tGB@lD@zN8ZN&YyQBl^A#@5Ep
z=H_6qK9N+u#eE7}>&<)ei2*hY`<@8xTq@w)1(>H)Wbzx7op6RSGBSQ_G|R58K{k`3
zi=%EvV#9-xV){rWd|c<(5OPiLbZJgYALI*#@r;<CF_N1Ec1$iNIw;Qd`z)K>4-D=1
zzv|P1B~VsUcY^od9()dB8FFgm8ry2aZdU4^kG5W|p`!>f=6*kX^S&eW{dU{zy#{n#
zD3z|=4ERrAKDjVAfG=LSPpByg@t?kOp7<hr*7@`A&+k%<EDWR_Up12VYEQywSFVPa
zo%IUkr6nV7_Om+gI_Ivg{`}TKw`?ZcdO20k9C&zo#y6V#d+?|K8<Zg>=oNw%s<N_5
zp2bFE2&+*>Phkt;N4OJXz8t~tRnG(ncdCrD(v@${uXk!Zy{n7lJ2|mymJcMO`Sh%=
z*KDF{stgU>SK0y(PBwq&Mg5{oKJyQj|7QyX(@bIo%7lYR{W%GqU#&#NxBpxA!N+>{
zS7^b4yjJwaDowS*o%_S0g035-gx9}WR&v;<z_>GB7TvV?Av1Y}RpkT8XX3Z&g>~#|
zw;&~s@`Ckm(0cf7jR`t+PNHsgC^l-IZb+S=4P~cK9rx)tNg0AQD@B9zo=hezXEzO#
z9g8k9hd3QHJfqB*;E1m=B5z65iLF%1YpWzW{Y|J<Tz=)#;G3D2q=#y9%Gv2W9$33l
z@$^IZp3crLW<Hx4H_gH|U#$`&S&r?P&N8kLv5*NL)HnKM{B(kkZe-9j?qgp*kLS%F
zf18?lCx!p%()zsZ&8J*D4PedAtsDkLh#lil7YltX5Ss1ex^(=xi$*odU^`_0_qUl*
znn`7S*x-_u5lU)MQc32N=D7+SMu(iiFR71XWe+Jjz6RSnk*dEKX*YQ{gz2Ki;el{5
zE%zs_(0)m>PrSTV=5V8o9n}niVQ4k%DZ@CuBgU+zCOLWZoZElONyd(5j}eHsuE3;T
zr2!uRhlxOa*DJ*sGB=SGCntNH2z4cnRC<HBI3f3Uo2A(6l&@Z88vMOT8SlI+WW^sG
z%;!+h9=y_6Q2m1aG47|v_&bzq=}29}V2M9YhGbc)p92YmO%$JEA}x-FoGou>lojJC
z{>xGrRaB|)cBp(g7KYj-v4NJdEzS+c$8sv*9>v;Ad(n&hKzg+9j}k(WZCpr1o$Lgh
zbL*ux4(2y<nwu6pjPVJsMuyb9+`2^R{)vB;`D*A8<v#(u?hy0h%OtJS?Un>i=)oFO
z8mciAYN<4V)X?n5exq|y==MMorB=l^I3+lT1U*d|_dJvY5e3m18`v*Zv|4G2!-0&l
zTJ^R^%bOQ%f;b#;nwt1dq1IGSI5@Qvh<*G@0$02gA`hkPE%W{xa`QDZIz;ub`}NM0
zOeYr6_^1^&7TG#x6-l@MJpfTs6TpVF;>j;(1>eJ0Cgn7P&gJHV72S6y7MkF6A*0;I
z?5oZ-+u(AU-9Uywx<{PndqPa|XTZvtra^bWec$oJ?fMoKsz6d$_NXI#b2X?tfr=~N
zo+w$nT_*e!&2KZ`h{^4^<3o?E*xsbp_`KnYlevZasoteIyp0BOvF*=J&2!D5MO)yn
z#K?gQuY5LI{x<d!b@`?drZ4Z;!G_b%)3i%Vsl@;u9vTIlYOWv^kse<QU?U^P$NQw$
z3JxO4UyS`2wcA~73w(vsMS~}?9{KRQPvm7=YvaYYc^jak?lftqIQ;B(9k{<ao3FO|
zFVgq+ay+v<Cm~j;a?VuxZV$X`H;GF;)POy%moky+ujM9-@6Hd7|NRvlWq&cF#aGT3
z+r#<fmA#UaW`;>9XbFS?5(2#pqe>ngD{Voaj=~W((E^LpS4p?c=(kz31N2xUrpU)M
zSV=??>-6f7O(?!Dfs#G)xdY71{--84ii{vo*I<N>V!(hw6`zb(e_Rh#wL%$rJ{1t@
z612Vh$T24zm%#*ouA)v$nW%vgjRhYtmEifKZ<7Omo)AYyI4ccEywfE7)#Hv6525N|
zMOKL%dYOd1k0~7!(vq})!?N=5_qhoSld^ElHq7mE_v^DU-jNP-80ST_K~%)7Re{3P
z-Z>p54NAcM!^D_GLO(NnZB+0<jkXEAtKSu1yL8ko7VI;V_7W<|5i&;ms6`eB<HeKp
z&R)TOTN_b*PTv70<i#Yyl_90)JcZ9WK!GN@9!2M#L6$&LrQ}}t3}v>%Uka{vCJN<3
zKB7WMm8s%?KbE!<t9kttfA=7ifAkG=7V(U{b`I9dBTQ2DY2?I2O|z5s(GrZF^6I*Y
zB$jVe>)EsgtQ=p|J<a6l6#6^lTn_qM{1gI7+x4%slY#Q5W8=lu5}!L*wcg_iVH&bh
z>|pTl@mWCT>5JV`^}Z_p`ExumHlctXKDb-+S(t!4hKgXk>%rWe!1|h=YN(6Cl2op0
zT%vyzM_v{u38yM~uA-ui?=yl4DTO{W+kT8e6#E^G?O4EpLkkC|iFF^mq$eQof;7Xr
z9$HG5BLIx!p$eA=jhM?}N%QMh-->6ZY}zBjf<A+scFX+0qsWkTs>>U&mHOW1-?)Q&
z>YD<_;+BVvg!iYNtDPl06U9JKgoiPqTE*T)Dszo2Ey;~?7>%HHeqLVEdzT<DrzkrD
zNc)42qw`Mst6rX3)ZG}mmyLBRTk|7mS?|Gt%}&o(kNc|(dm6+C$W^{*!T)2QER4OZ
zc^C|y@p5)^YFn1a3NXv&00rnF9!D5`X@Ne;f2FK+jxbj@xf(XPhmmK_%(7iC;#s&G
z;qZ8`=G?*p$j#!)9}g<^M#+NWqKKpEJ?CQSyX%~q<3rEJ@Ky8ri_^-QPIT|#0I3Af
z)EwqmO^zlZnU*cKDd2{!bffKz9_y2dx(8yyK6<h+S$70fFS<#-i+lgIsLI>ZDsp?V
za5eVG?)GGJD5pkNOy`;M9(+^}iL6HH{Xd${GOWq}eg6ZI5T%uFq>+|Rkr+sWFcCzg
zl^R_)<p-1j5(7bcju1u&6A=dz4yn-t=^i27{onV09KSu|*fAdMe&6@5>vg`)eo|h&
z`1s-BbmGPiS~r}bsm|hoij!3BXU=zE0{c|e@BZZ3RVw7wypv|mx(E}xi2#<a=QlAi
z;yqQe81%I7*rx-yt)N)Nm!79#k3g?MujO*cC<Ok~WiW6TPR+osJM%FWJ;ey?n(Ip=
zvI`a8)4NICe$S&QcZUJQu6eE-5MULQ)hA=S)^|vlar*Y`8ExsWK}Km>PDR?gr)hVy
zL%ge|&K<6=0l4r<{Pm{y<!`{@>aQCEh2vUX!8Pe}Bx>l7pL|PAO&uoU)!I(?pc6YS
z-ngprc4zx0CQra1df@lz?=9no2B^>O+k$f)!FK%C%68;t#h*qJgkU=ld#~J|75tf0
zX10JEtMUR}iUeLI&rv9VZw@oxvEOb-!TPmp=gMyB2j*ZwwptF;Z3mPeZjMzUlMOcf
zD(DNPm@Kqm_+|oQebIJUhI{NP2Opc~Te>-pn^IsZFpHMv2DL89=fVFNUe?^*MY^}J
zQIVhgVRgSRHZ_Q?(c1)UJ}*(-H<BvMFe~;Kl-N2xbWwooD=lLANXNC`2_#c~m|}^c
z;ec3(HK;w%`SmVJm<b>um|4DLO6Iax4p|~wscCKs8enzK?(zH0Qb#WSM;zXxm!c6|
zV;{A&;5&QVQ8P&_WAAkvQ3~6FyX<#5R>RWg9U`}B(cO)O@mVySDuLoXb-ua5d|#{~
zb+XXtgaMGRofz3-o@z+o7d6*<xcS|Zy_jjuKl}`oAf96BTfiWx?`kD9)0aF7Y=DHm
zLDnjn+q&<`ZC9)Pq%KrTQFayKMFh)ae{_LX4dz6AZ~XiEp^4hjc<u5clzQLIDCgB&
zjb(B&m)gHvPMUM-_?4_=apJFh6?S$ac?Pj=0+%1c3(or`bHid{{>VJ9Mi={WM#=IG
zDAi}u`M9+f0r(-MSe)p6`ZpM}YJ86F;zaDBZ!YBnn01B}R<)aj?mzuOQbnIX33YQ(
z()Jk)MRjp>AnkH!$E_6K2OzThTp)=f6&yWSnAg~I@!hZE2v2TDllNR)Da@`R))yNg
zF|~RqS6H2K(Dl}O-k^`^1HHHfX8Y4dZu!-Ij<AbeKo*_jP^F_sBYCd+FX;UGg!_u|
z&GbZjK({{2nSsn8Il_E-XUGYKhgpZnum;h7i${>$_t}~{)oRFEc28YzJfF+A{@&O)
zy>=i%A7|<lCk2rcRMDopt9(1Q%<PM}5n2!9mAk}REJ=!zCv|iB5Op(0-wqQ*Xt-*v
zM$I$=0YQh>s%(%%N0!5LH7wckZOT-zvMWNWIO9~y{q@KTx8tVgD~5tK8XxUKt|y4~
z=Mfucd9j<E={=*PlcoK6n83qzqMecsC8qVEB3x=+!L=lE?>95B6B`s182Gx1Av<Ee
z)FI?}FRW#k{rZm6tHT+OFpu?GMj;rU=>-6|x`u?D&a6)xW<98%xMS#(d;fTh>nE_>
zZRil(IJOpf^t*q{%i990{xdSd!=+?+<4mij`KV^_HBCIGG(XfMFz_fV@-(aEbiE=h
zY+Cr>7EduA&-XdI*Vzjtk2g~2fruvEYn8^a^2f-U)CljG<SKH-Eu8hSpHqb8#m)fP
zokYLp?qO(MBO3jfWNWH9f`~p50dhm;2cfxnih&nLea~Pawt{0^UO1d&ZphxU$t$^k
zL4o$?R}Pipedvv?MglR&-Rw{uI@xq-xqGGmgyAveJXvWsEPBJOTZwF$oNwb`C}rlF
z7~_CKlgcKOT8^5o`(Jn;V8*;Qfs|GTFDl{0xDTZIuVI3k;SY*`Q8$`GZb<1RjQmOb
zYD*w<g@Jfv2;k;;Hj?s(e^{u6Ioz{o`v2oXlMwJWxBgJZKwqS)7k^MMjwK~2>_{aQ
z(1D^T%I&vToSsO4oTkiNK&c?!Uy9D^hOhI5Smmp%L<9706ei$^&yfac05UC@l(M6J
zetI;%qCug}uVUDQGcGx5FwIY0MNQ#fie(kjGM^xVa$zn>c_6vu{u9$Q=J72*e7z?L
zoAT&~!G0I}Q+w1%Me^Jc?i-${2?@@y6ixb@eG^yK8QYkB^Zjc%7O{7oStfl({nA)!
z;dJ~hkT9}Y^1X<|lwKkyAg*jEY~v|>KZAh+&vO5;BSuhrnZ;Ew|MR=;^JPEGU~c(n
z$t3Ex^}bKJ>XXKnS*60<(0obc5Vl)jdU_H(+;!l;v0B?6v=x3qY*_2#Ci0UX&o{g;
zM`vd6mQ8o8gNyf;(hK%<6uC;xi!HIde~bvWRq%RiRww!-Y492_GQolRX6+nOY&nyY
z1DCm8MWhht#_hjS+9)~pF&{ueanSW5Uz%u!WISX;cp3J7Vc-en<N6Fz)1@hdrOy`i
zmJ$!>vyFv~C#52<D!7a;|B?9%-S>gcx^BGFOkQUTI@by5k8`!^(2dRL+IJJdB-cb<
zXj$*}SwZ;GZ35u6IX@kCOf-7MZYX$rzxJAn@Ht*{&A*<D6`Np5ERmKESdOj0ei(Kw
zD6?h@$xm7=N;}<}p&|Lv5VSKGKbFxwroh#Fy7CF*6*R!D7=@<T(AELP$*fR{ybvRc
zGXDa95WhULb>*5LvA#P0`&q-&*uq>-xaZprTad<k84)1*yZDm`j|w!TY4lY_u0|M^
zt%tQgqZ{4zS^Gt@l%%-(8xexY4|C@qcsC9WrStJ@CtqN%11U88YGgZb=4>Vi*nE0L
zy$=1H;ahF~CAhc1XpacD5nn9_rv+!9w!=4eshP2etD%1}mf&ZPSVQ)ks-KE2x4Iov
zT+fZ!1Lydll9G}F;#0-1eec;Hzga#R30}^-RS!1?9bbt#gr8DuH&?)?W~PD(n+=gW
zxMkm5qbDY_&2`{LHxm=u&PBQOMFI7IvjgICe|UoySbho|(q3Iwpmj~FN6nsTvm#?C
zy#^Stg6X@KC_dE(#6=3sI+t!f)x3O!`_4UprcdLrcR1G3RgpzPf-7|Y-237TxMXKn
znG9<RGf_WWk~Pt|Bm#kEIL;`p5{Ju{t8Ht9yu4c2CN^ob?~c7l#JX&OQMoPQ1@WWd
zvA$35dZ|wJFu8QU*}S8Id7WnVIehtKFBSQgzwU|HFvfsfoCE8pD0hC2@L)2|`Uw<r
zRH8N?a43DnNP17RiTdOe%T(>)5E6D90`Il8$~!;aa2+&|0BjySDHB`2GLypnf~#Uh
zwBAtNWg1WthyED2<>F~CK&kbJPSN=>=UvxcB_A9K30ciO4rB(E6CJ17#GAvR_~(s-
zPy9N?-&%USLHXgB?tGaFk+;s5xXA?^lj6dCM&Rslal?qY-`ANg*+Lgx0tMM>X<(l@
zRy|o{ZosJD&4i8Pb3C6Ec^t&wXkc@5`qjt2K6D!>-T$JEx@FwsGMLojZ>Q%r|HPyC
zdp9(u{?1qpNzuU4yIy^7Xkom0busm`ci+oZqra=xNvsc>fwuX_l{AQF^GpQPexUIy
zA^Fa>`A8ylc$G<7QiK+ls?!&^TPE$wwcI?G|6CP%^)5Tx33W8z*vJSS79!s5pQ6y6
zUBq*ko*+y|N=By`-6il)Fa>;7yWIZl)5dw;bYhA_C0>j>G8i>l$qBLfBb)2-AYTm_
zEKXV+=Mzk~EEh1`m&<m`1H@1n=DK1C>RH#7U&=Kg97!^k-$_7O{yl{m6TYHvd3kXY
z-|HsFJ4Z$U8zAIKOd*cZA>^XLJaLl2J7wD0J_Igl8kBr8h&KvH%EJ_YtDAU8Nc%9+
zD@p!%rdXQ0ub0%FF~k>gzM7_E!17pk?eP3K{?d`@KS^Ls@6H*}kMUV)Bya;YfrO6o
z^oH#z&nNv-1=gbQ*w|!Czaa0RL!vl{R^g)1QglrHvW@U|XIPH2vt8~T(R1H=b+x`e
zt$DH!g<4oxB*xQRu2)}W<r!;m8IvPwb^p^8W=;tl_<i2WLugLLv-X&xMd8Tbzk8lm
zf#s(5>!q=cgMi{<o0GPbc09$or0Vq)b^^1gmzm6AFZ$TAdW>DD+?NQiFjC%*RsuKf
z<Z<i5MaWcrL)D*9Btgy($}P<FC3BUY49+x1cq+aNTMTyb+S)HT4ZkkbV1aP~7rVrp
zdicrw=*CzTX0O|@t#O&Zr<6;@_Y!+JuwTl4xWOF>_;0nHB095L4w(UTu$!6Do4<Mg
zx4<vS9`uNacrJA`WD>Epc|7yX#%?Kh`Z;EQSrQP*gVp(mZyh+%Zh2p<cwc2^kdq{H
zJd50)K86fiBn3jpxBY)S@BZE65bhjwJy$E+Q*n$R+Wn({mn`+BDEcr)lRe!KE`$IA
zo8+@$gKj2LxM*{Rw~jetiAU>bv6chFmV^W$YvU(qR}9ad;j(p8p1|L*?G_D{7N7l<
zO^xgQr^II;pPJc-6(RL~A>fL+-N5>d9JkkNV3|2`gOT-49uBY2vah!cQF*rrPS0XP
zwT>qXW`@CA702}H4qI{7L)+$}PC)rWuCGS-=Y?hm_mKt#6?f9vmcWW7g{`Ua9;q5o
znHIZ1M&aZ2G8uBpW!NoUrTf_#<atnBu@yP>Jxa+q$NF4S3cbpkx5y=KE{KPMoL!#X
zzjupnB$_|N=Y2=97!wP~2SoERfQhmP?ia`U=DgN!nx%S6=+UAKLX}Ijo+$E``*tFK
zH}{+C`YLO|x%&e>8ProTr)<RzN)`x3v+&JmGht{t*yzqd(~TQn2CLTW*f)jw9fdh=
zQV8_GhgPXtI8M>vj(4ccOLHdm^QnIMPZ~61eTwg{*HIdMUttzAujT8+74SY<a03~o
zm0UlZZo65_*!HomOf0{l7?JvaonJxPi^}kygFUp(HxeYC8M%bUj1bjz!)*6-#4y%F
z2jO4k{~wNLaBA^T7d7MYx$(2(=1KS8#?!m=3Zn48y3RSKk_JNOm^`~=PKdGa{SJ2(
zS^2kav`5FDqw8%`|6a&VX(%d2hBw+tczbzec$%Ti1&!rAWepAWDuGr<K#{f6_Mvr%
z=zcF7GN8n=#K;S!gQ&N{+saw9TQC{8$cX8bcD>3#Qr)kd&QT6Muqho*>5L^s=u=AR
zQXyA!%_oW&+Yoj>^JXjb+UUnlHDxlorKp>r@XuaRvRDuuX~(|mQ-AwQAJ&cCwo}Pl
zN%<dM?i#+&lO?9%fLcs#?a^E$la9khn8r0f`DJOoGPz}cuSrs;K?C>GjR1<^!`R~r
zGYQeG0SNX~AX|z{^?a(-xIk4;Wc&F1e*pXzuKf_5hr+njUH+5p&r>}d&#ogF7bqE)
zz36Re`Sy2{M&7(Fn}6i)`@6iLU&1$e;zH((?j6%UQTRatG>!U&w9K09!Oqo{eddIn
zR|_rQhey#?O$A?Ebm>au<atCxX~)K*>0ZbsGdfplty4rl6n!6a+4_r(>uqGk?-c@j
z%sVq+rxku&^}I3stixb_SEwMu%PEMkCRA`03mk{_=w*e)MoWi!dNK~;`ivv4(c8HG
zerfG7uBvB1JPNW&gbhBSCobbXDe_`}Q8(``OST;2nuxv2^I5==Ym#-|<s}A#*;Q-|
zo}bG5)bNAiJZy4OOeneK6w9u$KkN{BIQo?M{eDn)^e<!Nbd<JQf?MOE+GtL}r(@M^
z+U6rP@vh`O_2cySQ~SvM!ydu@HYszgIhNDO91MuuXVIxpn~$!GF#{JuD$XMHgFlKj
z`UkJIPE3p({KI(>EF>=vCcT@_2m79J1L24nHx$ZShBuf_cX_=ti2L=SURZl-{VOj=
zcRBs@ym{cr^Lmwd`JYof11>kO6>hI13?Sa<t;((2dU$h5xL^+_b2O3;-oE?FrULH@
zOdRH!8Rg+*vc=LoIbi4;*5xBjH{=bdNC#=|@R0^+k_eL$A)a4Ul)9=ox#DjV^r`M|
z%=l{)UR2)H1XJEvpvMh_wAHPKgCpjZDCh$c52X$o`x|Bd5UIPiC~1$uds&X2a&$4t
zW?1^?Y1x}veqkSv<c5>!j3sPhaiC!!SzRWT?sIe2=Wb<ot4E8K=6Hs63wsf`Y&K4w
zouWx?u>(I-rohqPiv#l~$>rZHAPNtel)=(jC3B%8?8_ea>&p<DI|*AU|78sJM;v9b
zCvnH8sCQwj9nM#V=8=I-<*#Pg{TuZ-129_L!cvlM{q<0!tb0)4*^3LI2Wo(tMneZB
z6jz9*Nd@VPz*E5ltVbgF3z->v#ehK~LoKM-1^Y$3z%41s8oIo+YoN@E^_4NurnfQ-
zD4oV(SMpTGwx=3vYa^;>XYn6++!8|vHV)6G>s*Nihxbwzem{yYDJ@Bg&YPZYe0#Io
z)_a5sJaw#8YGhh%S#@eJ1+WUWI|Aby!bti%#acRK%ei0iuVkK1fbPH5G!}jYBE3B*
zqsx}Go{D_r$m2EJjBaRhcwhxez~AE7jZ1Npcub+mLb4S_r<Xt(&UMnU*euC{Xk*Nf
z(iXW7lZG$_LZ1Eeen+4-IZ1etbU%)%3P=h2y?CxpNFk3fLscU`0~r=8+nbwe-+=<f
z^X}R6bKtm;z^>wH0<6c3tPx{iQ);VY8+~^7N6ioZS(spsCOlDrVMLzwboD*AHSu0L
zvEOcx>k=}waTpjF_#^KsmA?~L&@$p>LS%yEAa^8pzBL?wcbJ%(7}k%H6_#41xg{sb
z6D13ij!lMUirNBSbn-4Y2lgjWZ2Lb2(*JzizcfgLZp9+rb-O%+zkPe^p~*gKUOl<1
zXsiCspV0cKl|!n46cLxvHUMD^ef`Cj&~Zz5aZot#;MpO*k(d$ztN-vr<Gp~x<CZh(
zmOZg*V7m!wfm(BMs^j3|;0RWKrYfhQHVS$E922>JBnezCyw$kc*T{2F73f?vL&$5W
zsTp)OjBQ;%Z8_R-xIP<$r<t+%t5EFk*J;gVbtBP-eSPt?_7}aSGZDv|a{cPfk&94C
zjQ?RQtw-O4&v=1@T`*B*zG!gqTQXez6Rf|=&L{b`>XaO$s%i#eukLqzwQ16Fvea<>
z?>rk!4}??1F_=q^!Q{H!m`o1lA0?&sQj+pMFx(UVLNboIyc+pe#H?|?Ms&nAg?L?E
zZUdL?8wxmkMFoy+fkvlDXQv=34jSXi-0>zT;UdwbibLs&yZK6pqT&;2ne{?C+a0`i
zeiZeS-?R(%LzQ@u`E<D+w4onAS)3LRi5M~saCn!IDNz7%@aYdbsp-@)SfR1AKjH|O
z;t`5t=pKOh+@lNtHKt)9DNt-ST^nC9dM=^%p12>FlLs*Z@g%v)=qAq-nkyYcf>bBw
zrG<H;2~>P#i)GPtx>3Hy^f2P8qUhnx8{hP6Z$yg^+iv}+WixaV5z-e%lCezm7V-me
zN`;#53W<!S@A3z}cXSn_X+ni;Rzb(dfnT-i?9C?3`?n8uWi2c+RNjQny`)twe%m|j
z_inj6v{!)xU7I^Ztoi^uB0IoVwRKI;QDKbbU%8pVAE;}CUr8r~>$z3I0_q3dEWbD}
z{t0H-4mi_ytB_w-wCFg?g?Hb_5(AHIrXWXrk?em`-7Jh#Vx)*>ZsmH2VK&5{ZIA_l
z?1q^^*y&G}%`Ixa4r+>wp6*<&qg5KzUv)@ki7gH)8dz-Xxiq~d;pC-McQ#tyL!?w=
zcN2WR2VHA1!FA|D9Bofp1$t!QYX%<0;tcYJdxM{!R*$7Z3PBvEh_CO7B6}s@WQI;9
zA=up_i)V_<D0G~nz}yGAHc|M{NN;7PXwPQ(uuwth`d_i9Rd$#Ot@?}`j&bzlpajjb
zVL=7qY(kk2vXG1@XuYinB8x8!sI`-*{0TF~>d?0>*wGbU1G->(E^-h~$!X^gr7jgn
z7}aiG|MRDv_ro9OlP8E@0RP2s!THGLVVKscEDli?eljp0&mMGrPAr{Cx6eso^bTE-
z?Z4JR4^E6v5Rd2DkbGM5SQJVEIbva=7IF3Ow{FB2bMMRwxB-J$#r6MkR^}_)0srCY
zx2QW}3g<aU7C21x=s+CGp6{sn9J&ww4E-zA&=E1~`pRvSN3IPJYnQ&i%6%y2n6s2S
z^lNbx?C(?ai9q#9>Ls!>p*LfP&oxPALWrmVed)z=JlR**K4UEM`jEWcsfU`DYwEl_
z%i$_r<FfAbG{k9XjZtj+aMLSjd+&T({qkTKx(}tuyG-$(Iy;=b*!HTP3B8^`Cz_}c
zM{NNwPZJFGWBY(7DQQt(uXjp;vo>fyZP+ETbCNJp&bUF#O+$g{+5Q<ogdcWa`R$)|
zCMG6sX8+YVLuh!^mj)hvv%hNViEW+R%KKDm!cm&bfQW;tuutZ1AFE7DHec)=Hy`V&
zz8_Yt9`!86sRbTiUD}Chd{Gw2sL^c%R~ibYiw{RLvjN$+`o*%ogp`y~3YA~rXMfgy
z!!*^@jB^rbj~-rjAFKoliOUn~v)0|@4kA*hrNO-v+%qAB%TIN~vh7=QweWO|u|ulK
z>j7D8MN&?iO|URr_odJ~d^Y)7H^+yeEvf2Jca~cMHR##w0uL@0g#-(y+yG+_WpI+o
zjnceFF@PJZ6IRLpPZVpsKIOd8GqJb<@<9*eo1&u7<^hPAb{>@M<5$f}@sOVk<m!sr
zoCq4!sM}4XTT<;RT@(poieBfLzm|%pgd&AzQE+@4KMmF`!&qQ2xk$c9XuvF<jFU&1
z$&%Gj#4*l#{jl~4>f2nIk?imK-V9zN8>v_wd7vGk$6W9!8!~GnH20nIUkCW{aB3I)
zONMq^6l0F9VP#^BxjL7<YLnk-Y{Lb(Zl(;hE*`X{N3W$(-l>$kwCJ_X!)3G%|8FSa
z7hxsEbY=EO02J_+fOW&uj2Wy%vQ|8(Rd<pex!3#N6Gq0cqhOCn>&u*8!DI!m_<8$_
z0jS$+Ep}g6NHn*od!-}X&#AO2yUXU}V5z+G$FvsHkd$b^k1iciCZ|`cAJ|^1(VwQS
z{-&`0>zlh|5cux|tR{Y=2>X5hg(62124vyvT<k8fdOJ|9-*rw_E83hG*3AszBAFXu
z&V80>z7j|0S#+{!dHW@S1KG<mH8X?R1Fr7kP8JqYq0GkBxvKchfY}y5ThF!lpJmjM
z9iixNQMXqbUuo-@y^}uUHmJeb)`#RX*X2v=vWlpWLD<PSSOy`?+Tp3=V$BSm63Dj8
znuA^vUousbk(@gkw|6y2G_m%?1Q!@J;-OH<OS`33Y5iq*6FY;?<XPx=mp<(U5N{uT
z4;B>Fpex}>yY0iXrY4o#T8mO+dqu$4!{ghG8$Fv1A*QCj8MB`^)+~lj@8&JyJD_q-
z1~DBx64nuP8YCsS?Q6L%4j7@W8P<Y-j<H*b%6m@pt7ZC>%ej(*9u**ck4L)c*L&R*
z*-QivA%_EP;T3A5Uj4n2;xOR98(Y39j830l%vhK_w^35kU?ajqD{svke*iUVY{~z#
zSlC~5y={-LA0&oed>e?by5NHyEy5gbeY4Em-p0;f6JhQg7L;fvvl>AAE%2VJi|Q69
z$G<nPz4DrTr2FfRXIn0&p#7VhoB#W7KBi(;3AKH?P(`*kZ@j~R=Ig;jc49NjtGWMT
z?UCk3s?FgCt3yM)u1_!!wejt4J0*Vwuc_DalSF}x#<{G<23_gICb!+yFWwu4un;#1
z=x4hMy&_sXj;O~_&Xa<%EZ=KYlt~Z*FRPqWq~tZ#k0o(?Z62Q?%8y3+R4OJ_&Zmw)
zp^oyxXsc$w%gX{y^+zxEN4jL~PJ9JrJG@k{I$KV=^c9HeEZ`!(#X>UCv0BpN9h(Re
zr!4oYae%UW+`6%HFpjC-j3AlU7zwpX^<x<oUhJUPwd+vwC8HO<8A}eLYvtopeDj3<
zyh?|})yR`NgC`1q5B2DHmClPE6>$GSiJ{{=Eln!L3VHxuJ7m+^HE6wNI!%3}JK>S5
z2qNx=GFYT24vd7MyT?gydB7g;oI7EZdS$Sxvtm@VoE{V+#R5E>d?b2Nr!k@a9wmNm
zCEW>w)^O6T+oW`PC3VAhvW+d3|Jpd{wFfGk=V1&G-xU8Hg=^h)9HGv8RB;>9c9W3<
zoXUKLQv&V^`8HV<bfo$1o-(({oI90^7=PVWm_2tw`XkL#if<wEp>d$R0*AS&$;rQr
zaWdMP^RF_D(f1_n!(}MD%|&=<IGM8ekn#b)Y)CAHbGY_ZiPK{V4-u#nT>1wknG&fs
z8F<p*9=lP3cl6LqVTGBq@@5jvOC^D}DAWIh!2pM%b~L#oOK9`sgG(u{B(YIQom#~4
z6{dX6-T43$U$G3H3lYE7E+vdW*b0gqUhw?++6&n{wRf!cqXfG@4-Nn*RYmyB;!HjB
z(E1|2tM$^Wbd+|@7guoFD%0*bA;C2p_q9TnUnzlVG$Np3a^zRo`xw5~w=ZBS{$yRW
zfcNdVw1vI)L~1&x7-@UT18vGyE~<myhE8&P%05KY&IYSwnu%QWd|K^K6WDUG-(p#g
zze;|ZX_6KR6snv%ZWp$EAF#8}=Ek<GW;J2*YGu&bShq_plt2|w->BF!vO!CwS@$=$
zprG1{XycCAh6_`(+|{4*a@}^g#I*o<Zad;+j?<+2RfBxIFe>fa46aFm_fWF`71w=0
z*(v)qG=zS(Gg%yG{*14B6~aZDE#SnJ@oyXwCHHjB3Lm<5C3Qz>u*)aJH*oFGAkX_D
zORv}4>~n1rQ$HH8?g{@Hay)hU9X6)%xh?F_yE$m#WJrKZ^i2u#y{gD?w0*TT(2aCK
z+tE9U_SrW;SN}Nf??j>?@8dDc%W!XC|5;}WE(|sSp{{>EJD%2PrxZ}Eq$aEO9Jub5
z*SJ0Qut9Y`faZtZy{wVoy`7oZ!%d80kNg5g<v4N+DJiGstzbYLDry6Yp$kJ6pFVxE
zt4X4u&DqRE*u^b>ak%P&*X;ss|8h&si3W*b;ivZDD?=OJ_BR*lX>;ph+&ycplj0mA
z7cBZ`Y@1M~k9o=vcq-$@?rd($nTYdH1ul`7@U%0Wnlp#UXrbLKn_yUbmt~z2xNmA6
zzEN?!uUc){9XTg*LeGr3KF|w(xiz4Twq+}!dQ1lXwPZ#IV`7+B$pTWRlK*pWTnSjX
z08UD;E&nc^5r5`uu;?3rmnYgT{(T-J#JPKNFs+Dt-eK|A<i4EYQSN5p6Ggz#n%Oeo
z8}|&YoH(3r4^aJzdi0B@@K1CU{~X7Cjy&<V^Zv36q;yG$IBDzEy}3&M4?VZ3NN-?S
zlu7Ri-_=!m)f-jk)~XCx(|i>lO}1@I8XaMrU<3a$C8zn;pp3SNH{r6!`9GY_pBNuc
zNY^<R#|?eiCdyjW#<x<4_E%(xpAX?kyTbVHf}(=9kDwcmOg0^b;}X>IyXz-;awqlG
z<HvMea*j)#{>2m^{s5%;?jy&%z}rKX#3~Y@Tg9i@zu}BR@v*>zn_l#+j=8?A5eqE*
z98vBS{$B?8_;Y=Fk$zP#T*aI$=Vi6(C-gn7r0-GjmOaUQqO;0j#wBS5PLg}Vpk%y9
z)=EzNi<AOybC=5g0et_Rv?w9cvLbUh*rX(f;m>gWLY6V7Mwil4u<gbkhvIUjr637U
zNxOkD3$9OGS03nBHgU-$wyRbpX(guEJ)-$#`ckPFk1ZU<e$h5cQI#GgE_&M|OM8_1
z(|orJgSMs`;Z8lrThsiMSa)q<Ck{;;*W5?9$l$nP%M-qX8@gq4Q%ZUV2z3YwEu`aH
zFPT{p$K#wVZ5~WWl`<4A<d37HAmilKytU0&6ho)=9AG(thsLII_aPlkRHdN@f9OT5
zatJ+Qod~H38;1f|r+ZGlE8*kZVsHYUOX;iM*+sgqDleJnX!pnn@ymAi@m<YZzbFy3
z3JO=F-q*OzQ3@>^Ej?=4@~_D}MVhRp#=nZJY%`Jk^+ad~w&JI%F`S#+o&TRXDu)n4
z7rGx0(9rAFi4*MfvnkAUC^H_l*fPi$lq?R892hm4`_gb%svWfN%ww$^G>o7I<#cwp
z7aUQ`4nO^yt>5$#rB8l4KAZM7w`N24552Yvtt;#>N72y2JA&@{gi|OHn!>4UTUI2+
zvp>s{bhw-%I3(*~rZ+dO5lU<{Cb_DtDT}~{g4YqnBR%~G&H$ap{y!*wLe2n|F_F{w
z$WRNN!=i$_4^mp)+TPl#|Ab;!@a~o5$T1oUY(ihX*c{f_Wo%MuqBiW5Y`Jbdd{z-@
zN!;^OG@8eKXm!gFux+%P9iL*Qv7pGH0+;hs=E|1ur>EpijZc9RbA5fhik~pS7XZaB
zxTF0qZEq=7s($K=xx&fEJ{fg<Da@U_)95?SgUTPk)*kY2e3(qGPBqS43;x(CDvpOQ
z9(tR;thmeju{<+^;FkFeSx=yT@KaIt9;d5--t@s~mVM~j3s;OeEjm++sU#hy)ct{n
z!87TBuoz{stj-1qCXM^yXG59@5d%=SrCwG(L6&5I==>9CU;)wbn;*t*==$5=Qgx!E
zuMe=?NsB^?R^Q?yqt9gyeu!fsRg8aI(Q0G(hkp>yS2AQI=k?pK5_O}bUYbm1@$)zR
zgoin#89mEJt{BCKfA)G0nx=h;Va0%k^G7dRKimGn=$5Op(R5l5r#<z`bXfsPX(juN
z&n_XR)7$>#{USvW5ka!W3_1(cf7oOZxETxfU4DMvOI^&#!c!NQNb%nB@C&S7wRJa<
z(GkclQ<a5ut6Ir;Km4-n!J5k-l}oAF{Jq9+y~9>JP8zG^S}#K8g+QabZJb~-y6gy|
zeXUIHNd}ICr<j-czKG*J;Ev<!srVS{{>L2~mbggM9D%=DrDESDFEoAn_Yn}Jb%os5
zj@4dsKCr!k;TyCCxF6SBpcpCIWlne@wO(DV!2>I-g@YR*73JcSGk^25^Rqua|0uvH
zhZWZ2fF^*p%q(egq53tNEIE9YTfdk}bg?!}kFhKw&bwYp<J(4nU^+3u8JbVr2t;d3
zSBpse=#Nrg;4_USr`&z(UH@2T{Uh`NWelI~N?1N!{`R>R-CIY2xSdN)pA>2ATME!y
zdlSkNZKkgAHa%iuF2fJ^V|oVPel8iY>nb+d0%d(xjc2`!|3N6Yey1=K?37elU0FHR
z(v;&NYm1RpQ%MA978TWjt47T94X4-T=6*P$zcr*yNsq-9#{RzLIv)7mFgu)P+e2ys
z8iXwOnu7eTRWzLg^0Db%!V0x>kqEo0jETip8y$O#79WxXyr!GSszxbkvyxG{HujBs
zgGr-JwD0aPzClYjGPRpkK0*9-BkEjMx|b&ECoP=ZVL-sH;@||!CqiT8)1QnzdZEOi
z&2J<oFAtNXXp)xY5F)}wTc1gRN&XgC4V4;Q!EDVD`B%<p)zbrNZ_IwI9b36w<rFSb
zlB;8G^(Zo^<-BC*3|?@$Mx+S>n|9B+?ZVC)nTau82{J1%weH5@5$fENXyx4et3-I)
z9{dO3$hT8JdRuVzuN60g#p4!={%&cx28DThI)!;ggb83;E;bsn_4IC~$(f|>Hp~WZ
zFYO}K_tVUj0e@${?NWF0dpa!T;!Y#JlHZT!^^g$Er+;V3arq{n$v~tpDw8^;(N_Xp
z8PXX;))jX$V2kgzcjdZ7_B~J#%PL;T`Yxd3p&MfPCfppM^-PR2`Nug5BDH>`V`{CB
zM!Fj(=nb-J%&F7iJ`Vkz?vqgI%2$uLTFI+^R?&(0I2koeN&gvFPKUSAE7yv3i;^k7
z>pZ3OECI-n;qocZkTIq#Ab}uQ2;`grtGqf1Rw)GCCA<M~utaKHFL4@PoEzA71280$
z{LV!aO={C&C#wC9lC*w;tzPq2QGF<%nLELvRc%)BzHm_#sH1Svi0-{e_lSG{F_^1b
zA&xL80NwcXJjKE+iIQH_k>6v%veERU_Yu<hA)BDGXb1${Rx?^wV(SH}H*zgq;gpUv
zb>TyFtlEXeVJRZ+8ZJZ{32Tvd05>#Ii5@K7ls88nI)nwc&wp2gCO}^}7Um%>DYT*z
zAj`Iy(KlS*48BHWkk;TKAoK5SC`VXrn-=X&ayW5lrntc2hD?%f>>l!kyvdAcSt@g=
z&L&RwlYCYg>D+u&zvA5D)k(nbyircDEJH1sR9vcfPo>0HWTve3gRD;g+Jla)dAsvR
zL62|wQc+J59LCsxEC<o%Tx3e%&+vFdLLXHqd+$kV07F~+!bG&bNCo~=_J}VBG{rWm
z#*Xa?b~AeogKK@o0DV9d;`$^rEJRX*Gn7r}r4JtI%)yvlg5ic(7GJvH$Y^CVu&;gD
zP6wSOSlZ=%9Fr4F76(L={zCj>ity)V&=!Bd^+o<u+*N`Kx7t^qdIl>RWsF^S_TLcg
zKMQTS-TRLWsa^XESd$ma*Ve#|GAb$>Fg=7X(WjBoO~Oe+(P5^`;X50#ad>GjWPsc;
zjET?A@Kt3qITPLNF_27BL0lQT0;tKqM(ND~Az!eKU04Gd17P!+cf<!V+wu^x&P=|P
zl(2IdE}UG2!^OW+$$WMDTkkrf8GLZIiK8B9hzLatybUx+y!5g7ne~Yux|ZPce1$F{
z)L&~2t|>c=@4&WHf!fspV^;nT|I@uDV$gQXd+LiNyJkn7Wcd1F!)#N4>!K^s5u@6m
zHF7dHu~{7nY@&mz!!P=;POneAfn(xsax`fe5*FCP?Mi<2Z`7OU5iM1{Rg6W`dWUJD
z|K_BC4~-Vyb>e1zHa=7cJLz8Dkc>EjM@rg4RCjm#!n`q;vECP_4oCkv(Pvt$t)m(z
zD>k6<*2=Nsl^MtJ>Fhz=L^n2iO9zbO`gbU(`wze>C4D(`xofm;BnZ!jbz&k*Evu#?
zzBh*r9LEd=rzwI`pcMf~$lY88Hi?Zt7e$Y^DBK01<-_8`6aHbbNPfL^aWN~=#xzPP
z*#;hTl2G|hv5AMA6z=$J+Kv%5oh@5x<w8bHA8ndL7;<B$?03X}c`T-PPw(;y^(-iA
z0PsM6m1=PF(`%K*lHNC>d=`hGD;LvT{cZc7DLA44N{-|?U=>?W&#jANct8dsiH#Xz
zU`+B;TDnbnJ6#q3K$n>v!NKW##)E?<nl9B*F~rkIh<M#5K7-?Jqf=Ox{*NW-R&us;
zhl{4Od>bB9Ql8ycle(Ki)m=#fypdPbl+*fEf#!LkRWV<ns3f!>?&(zcWJ0_lsmNDm
z?T+8p&345^A+JrZ>nl{=mY2q-RU$;4*S8xfl<g(?8Ib{TvpQ|_zthJ8!=)7&SqW|y
zYHV$Oj(z^ON`cI^U}WPkNFxFNs=P~9b0Z;u2Hq`{u9&_!|Gj}YUt!?_I(CWKG_Xhy
zGK~=O_>&X;p*a3yE|e$ET$JqE@k=_RHp?P}{eBYRG@sd&4|KvGGO5(RUnj>&<1|a#
zE_ExGxMN$^jp$yaCoztK7b}wx1zBNcC~M{^exC1i%jSm*zYU>KD6Ys{wCC&T>7%#O
zgG!}n^HB2&in@JN@D#wAOhd{07BHtIe?c;!<O%b`sdr09-_+EeH+xD#dLJgkl?fgV
zRZA!~>h?hC3KN+-_TRo?9-_AmajAUx4wmH@b8^iOGL&%?6Um_wVsKeqLw7XZoQ+5g
z8j7ItI3-j7N2f0bZG3wS<z>TjN|>~cZ2oOOrLMhNODYf#Q2a1Q*1p`*<52L_Y$MbI
z)pY&ha>S{07ISgco^+Tuo#!3C$Zltf%Qk;+O4K3#z(i&m>gW&q%dssJ`x=P$X8IVa
zR*Cfr?uZN#6#79tUqMHDFiT4OZkYxvXGc56_vbA4Kf$hn@!sNYxH<ly6SIlia!V_~
z9Q3xnu>1sgnZ={cN6M^Np?x0i<0?M7oH^MvWD;3?6dw%kXmUba1C(wheYO2^cKI<R
zbqKdliVSG9+OI%vZe8uxm0q7U%}%c$G+)mUFTL|v@5udX7XnShnQVcHd4+h(+58(x
z@wf2Re*4S+!62bu{^U_K$?k6wkzD=#sZwBl6X@ZluZDbr*eFHRPJPUKu5ACNKhl4Z
zWPVvLUSZ~RZuWA9yOUL}B=hQb%ZWxr^~Kzbb)E9%(XnLzw|jc+BkosclidFCdl*`;
zhc2|-N+D}W+*2Pl0$tt&eGrZjpHkBg^R+MU=ehH<>|xjL8766OWP$huAZwBIVdAwM
z7Y#rT0%@k%@eLH(nXSfcOJiSULJ(j_5C~(R0veqNj%3kUcH}TDr-&_*{x-Z}<#FhN
zvaj3=3JOAI;2d*|)yoH8B#E5nzxZZde)QwAygU$)BJ0k(abF&i1`|StqGdV$9W|Aw
z264?iNeVsME#;blNW5c+kK##%Jc}pJpLoegc34V5zfB*4Gf7*iLUTi5BVgwOUz34f
zj9jibB~Njost&t8{?r#_5=cwZwH29ffiI=TgeACF9>^#ACS(1;(TW979f~-nPGfL3
zG=ucU3#EZrORrAfmwC*=_Z>#~XZ2%${i4!jgu5Z`mWMOkjEh+h`jMf9ge14I0fY~P
z@BszgHe!gzqx>;T;hg~7;7Od^M4jQgw_0h**jls3IzOfgIr-aeUHjpgDoTR3T>q(-
zO*G!DHuuB+F<B>$Y`N<{!gR?boXF_m=C)tpgBj8y%Ud&nz(Aifu~I}ZJ<i+%b~Q=w
zz=NVbEb%d<2cOb5vQ~|+t24*}?_(o>DY+V%wq*?2W>U`+LS$GD7J>z=?vdFeA-A+O
z`k@cTk%lHZW+8HHRIM?;_AiOT%!Tpai78pd-LVw3;j@x(Fl2)`UF{Mu1Ji4RavwiO
zS(6DN%&xIXDc`=~BOHSBK|UF2+VvzLQ2dG8e*aCe<fY&3O0K`3a*zRV@J-PF*UxVq
z9R+f9Qd3heo1^tf5wbM@W_4{y3RxE3TGY;R>k}`o-^Y^h_lt^8hQ)jKQ~}G*C+@rp
z%#-sAu?u>~4I1k$H>H(*4nr~c*Ys!nTj2*6*~AKA6=H+e9#S%L_mn8xzj#Yj?W$oZ
z8xNKi3$EJ;o$0}VqhmBxv;Wdgcb5f0Qq>IgFo?K&JU&)XP|)+$=+jl$w!7XzeA*Qz
z@70Ul#I;TwVJ~B+G4|dyv$Gm3{<LS8W@!T8<xI#rqcwkhu8Z<8R?53VHe_+H?{<ux
zt#+BTu8S>Qs3Q+epUKm1N2Ob>woY7|w3B@$R^V<sQ8*x$&hC0U;8(b$&YLvLtEs0~
zk?0x9;-UWQHpv-#4DkANNnotMWv5vFVrDuIbGg*ED9K^3j*z(Cle~O?=<zIX@cN?V
za^n=~q*V^7Xz)D5Z)vGE1TT2o1O2soEf)$7Nfu8FAZJ^|IBob(W$T(+jdAW*j&A}_
zM;Na=TY%H`qgET%=J1QxIgSk)RhE+q-fF8f!{{V0Njz;(=k@UQ-ek*Vm_ujHq0nsT
zIhs5CB5!Oq<h;9&U&vT>>YwnVPbejVYFm5F%vQ%=DFc)LG3%yK_iq^;@A#wQXcq*`
ztX0Y^MPUpTvaMxx9|@<uy+)08IC$vyS|yI(0Ma8QFz^9gWF7va|Jf&EVuovq=*P+z
zirWs69o~F<%G&BKvZql#$fM|K-5&ZF_TO_Jj%);1&;AIt4V{jAPJHy3utFJvOuuRH
z;#4wX?Vxtq2^x<pun`F%=Ems1ii<s{F$e3q?bceq2NlhMo4KjX-jRZNo>=hei_nn*
z9}4eUN@<g>=cIo8a`GZ+5NDRDtnlpv%K*$wqsI7O(vS^!GhuED(#~Q9At}-qcAnx@
z<iXJ=TB@j<zI4h+5=k8qaf&O9F?C8Z$C}-4{~+y{C;Ief=rmLxf)c3+0f-iY@kzCh
z8K0aBOa7MOjQhvf4%Zt0V+`SQZpaD(eyrKMX6Un7O?K1C3EVOqEmO5hG3Lk)U9RiL
zV+!d{Ec#YY76$i)QgS0-LsID(H3ZjmDF<KoiEZj4%pNdyzd>Rzqx@3%9+NFKKlM})
z+bj_@rWeR%1d}A7$wAzCiGzxIc{jk6AQ|2FUS`_D_xW~uVzLy8Gho;)EU6CQBDY+>
z^e$&EH9+otcZM{$wXi2`irKM3IpT6XK5sgLB)`r`&}CrzHia(}zP-@@601i$ww(#)
zMK0|O?X|TzYU$)$0zWZ-)9;@Ka2-uT8$*tfM_W}hC9c13OM#+E$ba3|*JPr4$4f4*
z^4T=*-lu!08m877?$0%c6yuRV?PHCPnRj{Lwr4g{GEqf=Na#A$*<A|qXTCHYEpHvK
z98KbOA06FZd&?qPd_!oxXu;ZQ1=af{oe)k?na%QB&K|<T(K)wCHS)M}^QP3*nyH)T
zV$54vnVIN8!uPQ(bCev>7E4p~FIg3nM=6$>)F`p+2AEMda_|G@#I9h#Kh#gA;EAN&
zl-IW$A}Ewu*nXVwA;CszY$+u2dsEQy3Tk(u(nrL{``uWf1O#PBH0*4nd~w#XV_<{X
zuxD$!mg-+|B)l+axj?DD;@yltpYdBkXa8au)UHp`pJ7Hv5yQi?S3xbZ&%_ny`ZmT2
zB95{(F0YQ&_b>Ws^%kknqgEB0#M5*nOS+GQnHm!wp`2aVv$grtBFmzWL<jiu`_4<5
zgM6y&dk^YPR+1(&6L^UJ{+aCh-h#_sq3YRi6){7LU#qWjB{RKSuGkQ*JN{JmSGx)h
zQ$&m3v`T(|op<<#%D#3%lh5IjKie3am{^6Lo?b})N@L`H-$rhC>}C2XA9tnKKhFHm
zjWfi0=^sS>OCDyQnfq+)dPOE^>)*dyT)-;7NkNnUX;-O6Ae3NlPsDW7E!h+0-m?{)
zj6f-bhLK0AUSA8F!<2twvh;BzoVV7*Dkd=}qXj=dWqIRBgvf$|w#>6~i0=SLDc2Pg
zX>7*vVY6pfEz23`oF|07zri!>fgfTliQWdRyKo3cnmJ+9zK-#&i9_=j*ZtP+Um<6E
z727sKA8Mp5AKTvIr7Xg;w7xApHp>j9^G{ba|0)xXNQ1Ug{?H_!Q`PD&^A;YY2^OOH
zc?9e3T@bF5rB0?#(R4?%ELsE4JTR=jJ{j~W$@E_IaSREXth{6*ylktxy+0$qepeYb
zTw=>zP|eIWH<HNqAc9$B=a=MG`G%ByA>UhsZzZa)`Ztvdx!d2%T(z!fwt;~Ww4D@Y
zbhvx@6h*!hA&n7{Vv<>8c29PVW%ukWAW<N)<yXo*m20SxPkC;Nm_*=X*qb+N>R090
zWi5MkH4Gzyz~YKRR?q26_1Yh+0V?>3V&OCadTl1%XDE1v<A;pw8Ji}5N6$$&*Q5X(
z3#y`+3q;AAQG~|E@{=VJo#vI|JsuMJ%?eKi;!NMqsL)f!=R-@{q;9vt;qtC(4A>80
z)UjT(w!(flxF~s?=m0XmU>Q5{uVPu%N9TFO+P8MLu+#PBQ$JhpZ}^AgLPC0YxE2pc
z{N~pmU-26U4k4$b?TGZiuz}YBi=XFJA+BQoXnqFaq_MLN1l<lWWf23O0u-~IqD!3l
zCj3&8)xGhuZkjd&@*vCl8P?X>_4%7xYtsC>u9K7C0yF(}Er3oG#~QPk#$N>UBh{xo
zJT~cz`;ZI8ZS`0+n_Sf6tvSuD=EAnc$gWS8e8SGnnJ(p2t>to(sv|d2xPh}Zhf{W7
zKWz3AH-8YE`k>`<b6fTNEg^DW;kxu^P2YFk$EfKoC_QBnM@@;_8*guJPVQ%(g8P4Q
zTU-7zf!{ci#c4j+V66~;DLpZ&!xNe2JDS~6mAsC)m;f#_X<ha_FVyJQWC1WuY0uLc
z)hzo#5=CIA!r^*JQdPr-_2<Q&7&e%8VNfsMcs*%s=7QdNKArn?r*&|CZ51D$&;Al@
zLZBY=3jF-OswIqF#IvEPr}Bf$Ullg#dum5TDS1iv!p_e(o^1dJo7(|l5xCLB>YvR&
z(2ZgKE=k!;k1NK`&Vm5BJlwuqqDLyUWb#dC5)RnDY8^wzww|p7XWAMO%55<|*>4A5
zIPIgyW?oqe^QM0{RvG{2AFJ!?8g#w%?*Nw^el@t+vUh62<>HaMOf+8_vNrkbGL*h*
z&Lw#XVG`EX#EOt?!=!U}jO%=T{QP~q9d5De_(GUqvMTxV>|J@24QvK#mVOVjEn=A5
ztAXJV({hsN2d;wV9`8b!#8~%=#V7JsT6eSGXOI$0+nI}IetehiD~T8CVRi_q<tn3w
zT`ny*+@~gKRV*Y^B4`d2EyjeU;FX9y1P>_h-t@;Zq?l9ke3nZ#wN70XK``k5sbLi+
zE0~ly_JNE$I?6ecetbCsp=R>zEfN!nx&id*EXirPi2Vg?e;7Ew&TU-W=8z05ae0~G
z46`t|AsFbqqa~w49>jt97EHB4rR_ht1<{lgn5~}bY5a`Z?&>$>eTF62!~Q$Q!*!qU
z+B)EdcP+R>Rn7N29RT-WRkq4+nQJ;DY>ck4Q6j?p@YmkYLk|y`CGFe-*n%Jaodw42
z4TYfducvOG_P>8l1wx95-BjD?+c_{Q<4I=3#vT7rf;@*|T|F<$2HoABgJS4zzoXb0
z4RyU45xo8OMce=n<(}%}&sNUtWJPo_bd_A>FzGHuYNtiH!qT)j8G~Xr7V{OH?dyig
zzRzj}{bELxsYqK$rtr<jF=iZ(Q;mS(Q{42YNkYThR}kjK)7t~$%!!{L<OTfL(M}M4
z|6jYKJ=5eQv+{KE{Ri$s_lgGY6SMn^+HqQQBy{fZr%J9|MWJlcSW1TNyki=Ac-p1b
z#?q4Irwo~(@NiS_W9z5ZKQLGGt<eEI0u00`d)x=e6S#ATPoQ^NGQR&XBw-)$C+oOx
zkQnUoli5!6Zt#NhC<(|it5&;Rv|l7fpXyG#DJ2Q$F3dvhMuvIC#s)0n7gKx^p;XLM
z4EyW2`MfQw`5=hny-@x|4fDZq^V#`Z<7zL5*Q3gl%-ruhII%Y6Js#YJlbh;C<BZJT
zs%7YB8XLBGo(xMD<<{ohhG-1|slG!R4WpCe9@SL}$`M?;I+R>D4VngR%W1!z9oj#|
z`eL$ueg~U7Xi|dAbQU0{E}3*`S&wK1O-NFSYJ`oS!rx6Fm!}Je1a#eHjCl)Q4)H)q
z&+uq%qRP=Nhi5a}msiyLtl99L*r&5fs)N-(8>xTJ>ue$wW?7Ag1H%phLIH3!f!162
zd~eLdxg2>L@g3cGxbyc<u}8k~(A3ai@zh_HM&Ga&shZyBLx1(<(K><Tp4s0V5r+ee
zCYd^UMu%%ZjU<%0Y(%pfCsjCF4vr+m+C!}I9tPQjA0r{D4WXCQyFNwx#?=w03yq5J
zhrW))<w@+%zaCsePhz})`SG8>T<S)>Yxczs(O!H%+4FMgKO2$jkjuetZ9tvA%8UvY
zBpWv<C=~gmIE1L)Gm}WR;;5(aG=Tc}2VWe-7yR@)jCrKOk$lsvsB+A6T6YLAEsr)E
zs<_X2SHF#U|EHCq@I0|yaoYMhy#;O8O5wN0zqj5la`4=fB+4+PfKShs6uh`Y=QMe4
zJIfF7#|4vZ<~}aWRq|GH$`}IAYpM`1dbE@;P<1nA)@H84{5n1~lr%4>ylj|mGMI?)
z`8J!Ed^QlaR+jL&$v_KY8Z0W!P%LFd;30$;_Ggo>`|OC{11AiFAZ5Pjo3^&tA}}d0
zy&?jZxEt(ZIDixbhF&7ao*R|s8P`p^i?$A}iKm*`y4Aman2V+JFD_x&o0%E><lStQ
zxV#dIp+7}Os+Aj3sEZZO+Dh}-H=S-B@n?zq3bh-wJm3C3|6<^5T(3HS(Ttj#W#nkM
z{RUHhRK^5izgL+@U&L{4(6*mN^uF-hF8>b7VDBUm`c@TBptAfw&QaeuI(WR>C>;+S
zSn)o`ETdaxTE;OKNvq0x(|g&YIeZ<CHus3VDjt0z<Y}@CiLXTjx)LAf6E-{vY=>2I
z#X0Dbzf<FO8$)4PE{R?KI;lEZHcO4Ka(QUtAAuB=)CPaN{`Sd8dh?LafvrZ^REK2I
z&hk;DR%0QI5$G0;esC5U-G7sx3H02|1;@^7D?ybz_Bb2IjKIWl(uVF|{CCUQy?b3x
z)W`wrQKLRSZtw8de$D>h+s$#R-rT;VKBKi(<r<>lJrh3@#OjB$61j@Z1X5nvMi^6`
z)>%CR5o-x1iB!e)W8T0cn#Ip|VDgjG+!s6N1;Izzz+vRZfoHr<T^|aHGFIuU+mEH@
z()6h{lJ(%F<A}N?5~Fz!t4x4>d%Tu+G@*%rv*`Ulw%$4_sxSKcA3_O{knR=`knR)-
zrMpw4q`SFDNOws~cZoxXfFL<Eh_uAeHNepP?&tgbo@YJ3wSNE2x@+#8J16$pXP>>_
zudiqsX&O}A16UUNu9d>1GCaeq@Yz2yMl7fXGMR=#hBr&XHZ|_0tWZl~3OqD6)_%WE
zN%CwCr4pjQ{%lq?y4mS$@BqPVJrRcZ6lv{=-WQfpu!H@&4i%Fl%a;biG#SjCRyiC~
z0k>ywkw+^M*`tcy_K#OBD|Y^BXZ%v40n_^bKC4(GRJ%FRbe^FO|M3yk(0r<Kel(WN
z5Z}HGvc$bJ3>Q$C(3W%kGd?e%5YZyHIbX@vOyYYQrdx#xQzX=wfHmr{ao|+%Az}4V
zG2_PqsL7i`8+3C7dnl7aE)5UgINxt~Z}Pv(=)XRU0Wx!UyB$8cl)Ec2#r%VqSQHrV
z$}gK{qb+>8Ahr9uU%O11FDU#aT4>Mp{aWUDz$G&1);_!_+%u1to7pmz$CM)O9d+;~
zk$4gV-HuXCf$sExup;?4Z3dzxk?ukFgte+lg;&5wh7tGB06RKlv5WMRj2vzJ#A32%
zbFs`tR^dUK#gR**YT*;|d@mIhj`B+Cj$o>BYFMP2!u!E5o@0g)4l`#Y{m_CZiq8Dh
zY@0gDi^CI|hmEJdEgITx?U}Q#qTZDuC4gd|&utI;wO_#EauM14u5|{*)2ang3T(yR
z>Zz|6wzuLYch6u)#SLx5vL7`#1lyr8q_2N4Ozk+C-cImI-R*U!C2_Bam)~AYqh;VM
zqR<1H;xE8(&4Vj`mBUuS84>8C%D!sBVXR24pVPH`uUKJDgpc_-+GyTR;0vE+jbN#`
zgv61DD9p~%ZS>i>wUyP$U|+&&n`?~b%DyKaE^@KPuWTBt@ZGDZNYGO)NQMd*gxk+i
zYN(*gpajp|EX_T!4AjU{r>7&5QzX|?<OT3xjO)SZ$9$2mLP+i#VWE6`Y)^v4K?#=k
zCW=F$(xPxEn|0$l3|12X7Hb9u9&XpPx&}EqI&y<a|A9&uE`m2#H)W~{5#g#jonLs3
zVZ$DxyG@3vn(!Kh<4#YQ{z!NtOyoH6N%dabtA8)7a3HLf?*;>14kV9LF8m};fj1o-
zULJz20f&=drb7;H!fK~y(TZx&)fAp#=t^4Eys7_1-1KzgwcEm>*H3@EpCVA*XFWe&
zPdnJvJPGrYaHaTOJ@fQyBz9<Q#G6#(HaBAN;O@FJX-oo{7c!7Ug4*KfP+?4&XKQ5R
z;>GFM#2BAN|14%GqF8h~IlEhqi2!BVKgl?wL);1ac`Qc_SxSb#7xHKEZi{md<V0}e
zTqfY+Ca9&zE0zBej_+vl7H@561bjsEUdIf5|N44ldbh)cY%UAo%ad(YR%S_G)MIWk
zTt0miufX^^EVakna~4%O!+oP*Vb=!_aa~BlLR{9qU@MT#d2F0K(;2}rP72sjprdSj
z=+Yn1s8$M%%yB)DRN_0f@ZlxWW_DdR6qcZ*rlR)8d(TNfx7_aUFD}YAU~PaBTxB0h
zYIb$Ha5}Vm`kt-?`f^Om(OBEQvYsR&DtXep{sjxvPE?G7I=|iT;c_JxECx2MlgP%U
zgw_x8SsvUB+<13CcFwN`a01O%?b8dwfT@vGCMT`2TW~RrCk9xyfhZ~EeOHsoIgQ!B
z)sw|_;UenHEI<%#%@@st&Gy*-cxrZcHS0~im8t~U3%vOKWG>X526K9jXE-tWmJ01S
z44XBh#PhH4Nf_UxEk7AtIulLPFQ?cb%47Xbcs}#(W{BoL6!DOzbRhJ2R_24iX>gcs
zO#{*#g$nhXci1LtEiGrfCWov;`KSOZv9|sz!a`X|cE?ZRB|lsSM3w)26qT&%&rQ*2
z<KReh0At-Af~ub^$XFF^^GH^bLzgXw-lnK-_o1-`s@rIYSzG)$sCCF~ck>c;Pj7B>
z{P;A{O*t!0XM0z%=t1Sg%ckv7caz#PJ`i8&=JT?WGz2>sLNlQuZ>5hm%n^JFqwK{Z
z)K-_yCY_#6QjKFu_UT=*ywS6I8C;Ue4eL;l(cf}>4pC%Nl=Z~UK&+tH1IU8w+gybb
zwzsUw*l$@<8R%scrDxhwH9_{_ablrtjT#;v?$P08d*`!7QaQ5z3)s!}Y(n6_YrMc6
zc(n2OHXnJtGS{;Vf+yZyPD_5sIPq+~>*dcSvBJ&}YKZ<uj7-^X@7M2H&WxvntZq+J
zA<SXiioX4s=ycHXhn|Ts<KcnU;@N=hOi-m@$)--iZ_B&;u077{Hi?C%-D9_awK6T-
zz-%$HaL~8Ax!K<%Ksy}-%B1XGou4zQWccRwYZS*_(qL)5SvntN(_vwGYa;<^bxvlG
zz0h6Z%Me*Zp+XmHcEZXqSVPf6&u(_Y%lNT|!kclnh9VENp8e5EcpY;+v|v$0ZsxyJ
zn5=G%5GEN9ID=~sN}ipkxK>xsrwncGT$Rt@#n%2N%U6N#`plFMF+~k8grM2TmbOVh
zIgAWCAMQ_+kMmLK(0%0tUy-1=tVzL)rQ3>SFCnwU6@L+uO8?D$$3(!(a?U`$)q5?c
zC^cQ)$QcWu|Ai*UAgnNq<MVc>@W(T>3Y)y>Xg!R`RNgP9_zq7s>EXtrYM8qA$I@&U
zL=u**JsVLXsuuA}kcK?<mvM+3KA5_{cRGnp>loIPZ^1bl#OMx)06;u*)6o9cu7lvt
zJeB~BI%Qw0I`l6S`<IedfjmNTXmWbUx1P7wG1jBN_uUEWg}0<lk{72DC9^6Y(6r<=
zI=e3>eoG=tYV5}|d?)OiKT<hX=N92rqud663&VGr%aVQy<;K#+kDy24x3Ytt(um;Y
zAYxqyu3fyESgn8d@SX2G-nkx4tKze#mVUxa)zo}N&#rnO_`vWDx%0j7t1^p0=2p0=
z^_F=(FzHy6v)JJgbpLN)uk|;LpIc3LX31ZXspA^XZQsVpWl?VpOr1|k<N~Fv>FyIH
zBaMkge7<$_fe+X4P&(LD2#+o}7D!_Hjg3Ny!ag9j0(dErPnTmd#zzv}-<QkMzmxOx
z&8NM75dQH=y%GzZf#HNMlJdg3eYqB<K&=7+Aas!TvTwA9@Ay^l@bJKMf+vJE@?u8Y
zBfg4BHRrk*xwBhLXbK2li)dfQbeq0TfB)u}(^|ybGChWlHKjxZo$hnKT8`B^ME<hq
zKap4&#mUJhT~it^uugynP~<N5;LUy5WU1>oDWcW<sal1YfGU^;Uep-oh~J!S*ku8a
zyZJ%Qi@TaKPeX&c=fy-s{O;rQRs#@m`}3<P=V9Nli(f0Um0(+b%ZW<FB9?;=`4G()
zM#qhi`>UNtXm2;H<^I#QF<`GXFrg??gKu4r%*Qr-#x8Aa^_n<F8pVmOm?=J0tj}Z8
zd?))MH=mQlSL1bh#i=Q#4wel5my`v7ERwGapFX=(i{P{+lVci(&EWO)zH=9KzkqZ|
z4lNZeA3cfPcCP1Klc5R*jIg<_!PJ>QDM_`CWg`+(L)2eXlu~8DAn6&Kgp76!bh2d5
zIPy9`Bk6f`;ox@%OG`_8cRrlH{(v4<>fGlE3GtXdTB${i)54fUq#=zmHe=Sn6s^bJ
z`}!$I!NoYIc7a?~$}wF{wZCI?!{6GWT@}UVM5*Q_u#!}Z9NWVh7FYpdEjMms?ZlQV
zdh_omyeKIkDA2OT-Kogm)t{S&_HHNKSs}T!aeehcEhz1?m*1rQd|S=Vo06x+e+=!v
zlg$6nMp60u-12)GH+IL!%MH^qAaKLw_;M)f;>UDl0vy3U|C&?&*)bcW5Th#YTZlGE
zRxAqS1$G|YQr+{d`z=CL$h#=)kf~I=t{E%xS8~CM%;co4f=@m?Ro8!(MtmO_LW%V?
zQn-%oO_Hwy>(+qqF(-qeO~>2)0*?pVA33RA#WScOEOUmN#$p(;<iYQ=IDuSfWU9-1
zv>~}XmA&QiMQ0(SA@fAzwi%@ld!DC**9feuL#Ha$L=sVwN&>dyrPEyx{`{z5CM+`(
zyK}#qzM{$Xp{o)hQ)WfJj1MB81dF&HF1mrOzjf6nw?=T9$M%c0!uJawtV7m5FW2SM
z#5T~STIK1TJQ2px!GsxEVgbRapOcdsyEv5WFL<fPw;6{MHH#0rbThzQKRuoO$h5B4
zEMXliG}69ltE(yg=jb)=0|p<JUsd#QkW6neKuaoy!cYURe{*SlqYe7k_kId>TqFUj
zlCE41`m|t*u>qDLzxdsw4U-AQ3@0LD&MFutqE#ov3Lg<K3V|CkG6P}L$~FiOQ~Md1
z$r97jh&nnn@x45zn<{xxETbjT{ocRgay;)NM7ko8p?Y?&bA=8^2P(~-sqD1W!b7gc
zKpYZYbl=xlr1iO{kCfIPeay{?II7gfvS(0BP4WecI6H}4lCLxLbw_h>*QerGwyIb?
zM^+>IPVY3RQosETf}79CzNV(dT<RVsc|Ehib$5j^_f^OY;54(Q_1!D7Dcg8Fkh;Hc
z4n!<&3~eq`3T3FnQ7DBJPgUN%59d%}W8y-{u(6?M$Y?*T=|mh&5{o`DOW#e-Ix~mE
z7S?-UhembXf%_*O)xDS<?adj5aFl<*{Vf-K_d}qiV{1+=u3>_d*X~|goIRdn&=EY}
z;w$aJ(r8(qlgTXcd3MO(>rln#u%3mXYGuxXKeh_4k)fu`?N+%wRU`G}0#OSN%=fyE
zNxNQ}`LrsaGI4HMveCBepj!&XeiAp`s4r?{ydT{!e~1iq@Y**r^jOy6+^6yYNY}#@
zg}fxd;*sV}$g~W<h~?|SvK(aHNPX7JT7qKv*GDbqt$8Y%<~X{rBBu9B_7a}L<qB14
zx!X90RmJR6D#q9s+gL+AWqJAx9GXmFuCbQ*RPkO3xnnUL`SCGo0XtCuC{Hg09QNS>
z+fgHwJYZ?tCpXoo1}6LtnhM}u2<O>!%Jl+MHf*YKlI<CXhA7rM#_f&kmSxQO-m{Xr
zhhwH}LWcMCrin4KLYH1IN93-*9nE7r3312F)6zXGdLe7W>8HlF#c@uPhD~Pao9A1|
z(x_qj8jCS?8Bkq~SbkEMY^md*$d<xX@oM@u{RoEQ1EGucryKZ6huK1c({dL69?Nq~
ze(GIe3&~2m_wB9Jl~xAn9*PXGad~0n?esJWlYxzxG}$%CjP~mOjcs_8UhtLwNxU|1
zvSz<4)o&hn2P-yw$wj!s{Wjh_S<BVU!|(X0cIaSsWIoU5yh5LOC2`}vk<!`BnLn8&
zKcHjL{yj6Rob+)=RV*31>B|>p&#hySg;^#>{=nNqiFpGvlM_7Eb5a=KEQwrdd_VfQ
z3n~WD2JQ{{<$SX-dn|H(m^HiG$`wQA1?~0urYjAM6tYI=xL3wqu9WP4pWlyX0aY%h
zj*;`a7#ih<HKp#jm_8>mny@J7T&Oq96&AK;m|@q}5eOF4nO&{VcVEuZ2EemduZBU|
z-Z1-d$IaiJt4(NgEqeFOjZuctKFV8kF(GH=jKX~sWgrj{CnR6A?jQ(-3GR8;j5Ib*
zo0f^?kALU)EPa=(&o;U5UHC49edL3yo14qg-JFA9No?R{V}2lVAkI&7-Z$MQ)BPpa
zSx~oF?znhJqUO7y7P?f!{TwKsS!dhVSkE?yCHrlrHRH=~J!ScKGZE7VKT3aXZkqeM
zZ!O6m9|MoMkHQ}>8P!aK)T2-!ifEN{!_`uG<7ye!={Lo5y19|wy@ebUxoI0k1)97j
zM1q#n8_bBouk|GOkqz;5U~mf}+v{t!rAURMR*uG$N`a^R4&LL}hwcg^EKcRw&uyT=
zdNQmC$973|XSVvOFU@8n%dLoCY25z=%p50h0pfMAje}RFHB*Wz0EBr%H>+YXO%yZ_
zzp!vZBqGcpA-u0QR1PAK!$x+a%1}h&bu%_Xs>|hoLlLsD(b-8Gbi4k>{pzoM_tjNV
z_sO5jS+B3^c_d2X92WJ^uCBlKEYWkVU-RN8GzJ(?W+<GWJJi+QZ1I2GI$4R^m^bU&
z1qM~iKw?y2K$i;+pnI+;Pc8sUXWL=eR&0<aeHlgBP|5bH0<ngr304JoGuwO=B;V~&
z4|A1Ogmjerq9?Ogl#et=<tVfa!ym?E@Znw||HpyRo9Gf9=J!$CfB2)eVXl>d>8L^q
zS}zWvBs(elKKngXfZ*w2S{l^d$bz#N2*H0U*W@Ak3=F*3*Z6fWIJ^hbuzXF`_sSZl
zm+!L&wl&|$Z#wYKhsX8d=Z=ps7tl-&BvIv4FZ&^>&)XIc(@Tb;mr&IE6A8y9utRbf
zI_p@CCGy`vYlf4s_wd;W<E&Zm>GkWw5BV3TC(mayOfUY<6uE%G8cOGmH(0m8P6Xgo
z(@b;rcYCel!<(wfgL80i&>PNT`^0j^H)02?Q2Nz`46dXCE1RLqKqz#@*2vEztRRse
zvqdw+{isu6PjaGOddU}lxEs(mb*yYtl(KGXRe$<xgEmXmIw!FT4kIXPZW6S%Yg0aS
z;5B+pTvXraE{6HUhxGUQ-BB%o^w}ET%5FZ6rW-$#`xTY={YO?IW5vpnAy_|{SlHSQ
z79Peme%P(n(&S<IahjLN9s8+r(t)Xx#bD#iV_k60U#hEBFXj2u-M_8SeSdFz5T?qE
zSDx75p@cfZp<oUZb&1!6l_&K6)EpFN)E+I}Fyw5=JGWP{e)ALsYRwB!Aaya8O(XaO
zbH^^dybQt}G=8iF6Uw|NLJxl(D!dso7xMSir=_W>bi_?GTSHN-)0!|@ldI$x_s0~w
zhE}M~-r1UYMo3KIN5Q<m-~SjJ7{P30E9J`&zl!iw)DldR9{$$dcRpU5-H)VQjAaq^
zE$G|CWWJ}LCbxXrvH7vV;j7TUuuTL7vWj%BJy(~r0xzb|%iHslN;~O6ZXC*uYUpNc
z2$~udbIEjq)WuHECaW?9<Q1;{Ty_QsJJRdEiUI9RMnu^yt^bi$23OvqzErb+J@Se$
z!Kvo_VY}|(A70RT5k?wC&W0E=L*oR66G)SOm26eAunD0mBr5!Y%Q`ml!j?J5s%343
z5Kcvy>bgs<e@vb>N0iKdE=|^q8NwGYt69l9P!Nl%`V~2+Nk38Kna+;;GXn8%dHQZ{
zG=XC>49q_^mvcPBVZ%Sf!q*kXP^Nnk9gW}A+IiFbI9Uf;)}6Y?&h8|VY7c7zpe?&O
zp#{3E8>kCQAMs$eRr#1@(^gQ7<C7kr63pTn<06vtv2Tc>Ime=PN|na%U@vC%`Uo1c
zI9<(QEwtbr1fLAew$N_xbPu?X6FQTjzpIuVW8LCqU~ON1)44LFxa{OP%+VNr0UOZJ
zHf(%1J@l~`$A|K8W$|`!!=POQz;Hej&sY8Z1|R)%rEn_oZg!%g0WxTEH^&C@E|^Y}
zNtye^wq`xYw;oa=(E7SjzD8GRZZg@riZHAIW2UY6M<PRP$b`X%ktvF?$>%l(-~P>}
zUC4JXmp9MPH&fL3@Mk@FgWpf^hBtTYhk%SL9mE_uCzn)LQ>(S@v>F6I0306O^B<Nr
z63haTH;YB5ljfd`>H<tOaSG=%g!Je>Vb~J25I<SG6~D5UWKDUE9Ht9#E)N6VftZf^
zzY>TjwdM9^?|R^5F59_D?7}|5#CIE>Cm}AL{FxLblSAQs&6jwCjh5)2gvU{yrXk+2
z_txQ<KgHmze$q_RY#T@S0Q-=5aAKJHak*wP{$-D$(3qd2zy1Ew+|7K~LmE5}IAW=L
z++;Vq?$a~%d2e&glJ{WC7}dn(^f=djxB9RMJd6Xkk2g{oBR>`?Kl)aDtp3PN9T!nF
zz1Zv|Ts1$hYrVlIYp+sDv`!@<9DO9V`cNb$wED1)>=vTI9jTTE_GRcrc-feUL6<X5
z;dS2+7sOKc)!##1XGZKa0WWWxaXT$xPO_?qXu8OHNvb$D;e@;qbHRML?}C#+k@Wka
z;%{$Qeep-F8;c3>=GrNZ?E6wI-$Z%+CAVdYYayYbIp+HQgqbLg9f4ZC5t}gw=^Yw!
zB?}8V<5ht-VwFq)yZntD5|WbXIcDu{a_DX#!XViHh%aAKQZ!(vfL-~d24NKEw{+BG
zlXB{Om-6W5Iq~?<>>l;W;=V_RxA8y11Tzt_KUiN)4_TH~W#1WY^3`NqJ2b8A&-b--
ze}C0$E10bkVS`xTUFbR5J?&ZV=`!z_u*yoi4*~>fB?AtXUKDY}?n^@^e9`G}$X=ga
zOWi3vUMonRC5??;^z|+8I!rkTIWM=If~saC5$y_Q4~J46zPB)=`MM0)<Jxa7pJ4eH
zfoFuY;GtB;n{yNzfEGCa#2DGKwX4neRHi@2eV9yhk4+RkK^n6rOLLgaa+Kb}*iZV|
zFES;4Xmc-H_qdC^;_y^3ves?0p!j9#2U*@PI})e`KgqG9QF>%g6H~CSkhPzvi#=N=
zd=R$2_Qs>ZMz6);b47@}hG^Y)<d!7wtUKwK{wO99fOx5YfSPe@DYC3un=)>*E^r%E
zWBv2$uRiS`VYZ5Y1Q8Sq<av?Om<G>^TKN6C^!G-oX%Yq+`IRZ=3MVDGMwhd_=jL=t
ze0}|qS1$bBfj<zP`2h%T;Mi>WukWdOOQR5#!4<iRCHclHW$ZD*rdkIF2PJX_O#0q8
zK#i&MHMZO(fYiIIf%@t`eG?nJ@>XuToZGKNI~=Ar%$eiPW`&O)%fp1+Ma<$bzfa~)
zjAi^rf*LnsgV506sA2CAO4YPCNhd)VSj3J^^?UHrBz!5F%$zpVZ2CIJ*5g-Gs=Y~F
z!E&hD?#%BQZ=i(r3SvN(P2#JY9^RxpnAW39CXcuV4(4jSZ5p+rOCk<#F6I#)##sxe
zyPZSlKUlxt)~=?~wcVPoq;?(STuADgV)1uAW*gcQ*b?=FhtT==Gi74AL2brh#UAx>
zzC&%k|7GSlk&=H4Q2mVX9q~NFMrQ*F{xRCgaly$zYFb^v++RZw4_(#Cm-fjcZ_6&r
z#?BLfvXjr@VLn`0=Kd#rvbcJka;QYw+yq-)X!Wo69>iQUGzvUeXP2kkC@-sArr~(m
z*cm#B1g_39y6=w7f=;)ZSA!6no4<_mZP;u+@shj4;soaY^gFb5yTd(@ret7~>`tpm
zb#{9zN4mf$wcKFjol-BY4<4RO{S}5!14FyNYio%SjmwP-o&HFgz#UD)tou+#Zb5JJ
z4&S4q1jnYLqLk^$LbCBjD<{Qge?^$I-+vBPh~e11*p@R*XEuzW*&DMqR*UowGcPj0
zlaWhQX3?@V=ByCN9!p6f?2*eVwIHLVq9XEz;is+Hl$DL?qIEbjLA@Avk%r+sKy32D
z$DERpQ=jji4|ko*_kDe#(nCWZ9b&=V$NWO!qT|CE7qEUYzp5GHVRZ3^{rVqV%4YZ<
z>?qep^1i}DGM{tGsTnw&dK+=80K4{0LjAz0`h82&2N7-~)wH5IweVwkm+~cWwymNx
zJIm5!=C{RyL&*JZ!bLF-W0+hI2xy`@rlz?|12++?H)ap>abszC5~nApJ)LGZcU)9Q
z)R($Kk+BgQ>t&MnOSAzOl%+0G50iNDxVYVyzX6am*5fRmIOF2;;}Pn7r&{znx>^`K
zqf7X0e#YwA!L%YQG#*e*MxXNDt11628x!6SSE5%PBrS=Nw$*TB{bO+eLQfo;DYikz
z96hbe`EM-;ZMenrm!D0JMXayv$siUZ1Shwc2tP4cp*Qnn?Flo1CH+kh>->H-HGa<%
zt5FnYSqQN|A7bv$5xltg7hOY`!AaP+<PbihFgI-JpY$k9vAhEn$flIRDx+a8FEihD
z3>0Ye^`wKHMh(u~v%?6Waec*M#=e$NZQNv8R7|WP*uYwRP}g2(czd~skVJmk4*Rc!
z+g5=`L;Ws;-GD)!>nqk21Jm@pgZ!~e1<T$--#m#&B{M^Vdjnx*JG~&}MIEJ5t*O65
z-fQopWdf=+d}9Aa{hBzQIbNB!^lUnZy2pDZ3KSP#c%DKq-wfyjWM333qSfBBdBP0z
zNaRe3MCf;N6^>aMQ(wTVZA`;sx#90CKA^7H=FX<HxJ|gPv70eVr{=f~_w>1mQsb$S
zg?JBSP7(RJ?;GQXk9b&{w6^<uXK{W_Hh%RIUGozsCP8ynAdCT4ba=>q4+iK&GnK#W
z2E2%P!x}2idFIgyDBBDe+DmLGE8_0hF+V*zxE1z!lItHljQy%Qb>_$W=~NBtT##ub
znZMlT8|tb^uKJtMgv5}N?H80Cj|b+o-1<gD7Rrj#w;S+UXR!QRDnW=0H>_nX7Cb${
zq9;hym5lYt??;6Z`>JBP(PJ`eVwY;uZK)~j_SXgN=0nARTo6>B&owMFye9MzGp<(F
zc=&8_`cycoXnSo#<_iK;)*flBUNuu`?BH-WAsL{Mw==GKUy^JX$eWE~o=#e8xa*cq
zXU9(&bIno9oCbNyP=WtPisewAtSAy`^ddf}(TqhE;NgUrTRJjxU6`ryp`X8_x>Cxn
zx=56|cJC+Y!&TbAcFe`t7k|M3V&Jhcv|oKI7R}hX0n5CwGoId}2msRwfHn8dc<BZ-
z4Mm?l_3j-0sOvpZX2$TF#RdYw#RoayPlV<aD<;sMcsoJs$oy>5aGDemGo%l-(S3Bf
z-4A#qE1EsU?q4+iiK{CnSxiO#PFYA}*L&DRc01NE{~|=!O2sp~!@)N+RA2>_Ol~(B
z&C2C?P)=Z|lJ(dG8Ev;HqNqoW8JD*x8j3$@zrg&NB2z;!U|%&YIDA|ArEXwjz|HDo
z3|JO&ej#<`T|(|@c-Xw!c0_ieE|_d>V|8(Ja08b@$_*6i9j(xC8RsGNkjJ$dNp~o(
z-b5uIeh!1aoZsEP9xI+~0dWIfUgcqL%~;UU%Phz^5FF#~dVdomzZ=6y87&*>J{F`$
z3czuK^^nlLwR-^ZN-5*+4y6o$xp-VY>pr<wle#@P9h&USZI9()4oL{wwB-Fo+jYlM
z%3%-vYd#?ujw}~(gJ7+p)F7#X(b6EQ73EowZdwpnElPgr*9nEH3=kJ!KhWa7dVHj(
zX}g?!oL35^p%1IBZo;G&kinz}R$^VYxP&JLZbGYtU*FBtRbu#*VgLksFteM-uB(%`
zz&aAIEgKkO^nMIo`K>l9pnrdL15EThB)?%3Js!`M(yPk$YAzkw9sZQzx8q^xkHpO&
zAxVic@^jWE<HvssXUmweO_{MbNGfQyp!1cs8CeqjL`9HHxR=PYPi6JvT;FyPQ@b`7
z1;S(=NcQxnu;G5qHW217kBN_$r-l2A(N6};Cd`ixCknaqvuNI2ISN&T`NxFQx@!(a
zA1d4g>osnOl2K=85dq^MK2sS@6sVEuw_ouhRSQ<`w$)n9x8|y8-*Ew5HOtE}Jrtl!
zBC|<9|7)`wrlC>w+eq(q*V*vM>h(bL4f`@YTP^TDTrXzI3CLai%Juugy5=2Xa%kaZ
zc>83te)?0KLSInb^sZn`hw<xaVJD8<*HhITT3qFv&qDA<Z%)f1+%Gp>;QzcNj9~Gq
zN5osBJHHHv$ZU)a>k(TViJGpp?i<uK^NWyQl&I3dJ_SscDD=M=d136WKra~_r3l4_
zhhCQVpWM3R7hsey%HpREW1#@P{0^@ruAFUB35_hiXT|$mG(2@Zo<Vu6`iW-`KqGRy
zA&YvpbBChx@+8paXyM3f@fW`3%j_O9%)YG;UvR>fnCN6iV=zhDqDz?6d==57L$KaU
ze?}*Im&#I|cycyKjAURIqNg9S#56B`sVqcB>EP17C;{YNXJ4-9*w-Kx4v+Vz%a%i3
zEnDW0DNB-Ns)rO_I`?yxP~BOW%W>DpkyL=%k-Kyq19AmXQ>Z0*eZx}8|BKg%&Ol2j
z@C^y?lrY6}yj8%=HaIgOSDdP^R4^1*%fz~L{baqUWS?K<?V9W>$l@R4$D_d2z`c|`
zdm5xH7Enx|#i|uM`17?nf2?)LQ%`DfAu-bSTdwvchB+usL33-X*ZJU3C`QmGkDlZM
z`Meb3PeW0D>qZK4$+H=)(`g$^Envx0;i&s@^>N<j3A#jfTkYFebj)&`$<HIqw&w(C
zJt1i5#6;(?0&@O_+S^Xa<EiEN82MSH=PsWiOJdKK61n?{NFg6|eS!@wTDY4h*rHmE
z`qvGe-H(1+)%HBNIbUybyFan8mbJz&Op_C#cy7~fqG6u)r#!<51(gBtcpdg9n5;yt
z9~IU5!0lImCajk#2@VOye(^k$V&F@zs1(IO(w7nwzWIYy=z_B<8QB}5aqYLUs1A^i
zu`dL?aZsOS`cD!2i+!WGh#v=D#qMdbm#v(b0UGqT^J;+gs#Q#xZoE*@Byonx7tGab
zSTmf&=145y&UZZEM&SWC)Ea<hJ^!Lpw&?h2q1D}P#th+W3gft_st*_^Kz|+3(BN9B
zmHrGtZMrr#W})FRtgu&0sRK&I5ZUR@t6unJgA9q-F07Te_<KZKApJ5YsN4H^**l`m
zV5y-Vcr;BajR#d@ql}6AH`Ud!%T$FpP>b;}-W#NcF$u`VO#&aZ4{pwWK7?i&5#HXf
z4D4-BYOzNYT`pvhKOL3G4#!+G(3uy~olDPvI;C%WNp7)?4=+%+9&@92yZ?^w?+>>|
z_k6PK_S+f=Y-&%L)Vr9#%*-)l#C}I6SNY&);wa27$ol=mpQ>YtI}=9-J68)=I|XlR
zYb$pF=bZ7LfIon@i}%3R{i?*=;!YH#khciTiLo+ZFm&jhxG4~8Y^nF(nsHfO=`R!C
zYKw~-dwf`x0;@eGFRupkrT%x_=_$~Pkm|~`3i3Q;D$F%BX+J(eyR8%Eo;Cq@G&N4O
zu!F39*l3ckw8Fp2%HaT(0Zi$z-4Hc~k2=v$fi#2dPOh`eh3mo{h;%Fgy>YDkBc>x(
zna>YY0aIjAyJe^RlYB!Xg0xpS@%h3em7gpf@7s~Ihw~klPF(0`n`BIu98^g|#?7gv
zC~-xTgW6d5%8^q%)V8wd4Fu1y`|uk(%rl~FiFn@fa0iopi1vM2kLs%Tg|FBF8?%sP
z6nF)Z(}@dvr4skNe?VS$zC85ZZ@oRsPFCA?uW{-QH#;pszaq)xZRO%Gh32JSRET&`
zW%<ohI>l^Az6ZQyC-L6o@?Ci5W!&`4=6lKVZ9J#*(w!NwVK9EJ)=ui>6`g4zCF|}h
z>5gnZ-V&!huUfs5daNrY-+q$abGBz~>H!=}1$0|BcZrC%3RYXLIDa1y-M=TBQIP#G
zj4d?HNnP;k1oL^>8-rBq(}&w`qj@N6j<ErGzj{?{R_Ot3K8{M{=BK&#@k+C^h27E2
za1_kpGj3(6Vt$6=>*n_!*nHmYCUD~q0kR$|<b?z%$s@cv0c(HW^TCGuCov<S=L*1!
z0{tauc8r6%UT7%^`tV&n#pUh&#l=|K*7M^L7Rl2AxH!^k8J5Rh+j7~@^}v<je~{w+
z+T^q7zFk?UZ_^8AGz-ED@-X2@UI!VqYBL>5&soanx48k^3qq2QY-$0b*jYdGO#Jh2
zQp+TdTgL-?q4QqW2zDCx1*?PwcL(6&Wbnox_lOnl?t*R`D`@0JVr^qR(o4vzSiKLO
zRf+Oh`<Zsj)xD=cgSe`~eXva$55u3I*K16!<knx*U*2(~k#hY<=^b?YCu%FuA8F>g
zkXFJap8cybT_aduN9RQfo4zT(JXU1`mZ!<Qj%|n?5l>j_LdkV2HIF_fj|sr!+A~sB
zN^wtw_e{Co=5tnCUiM;ROIhRP&&ea1$ss-xbv`=|_U$kCf5pm3s2fmZD@I_-4l+l9
zoW1V~cThSFq+>97BI$(DWwnFb?Di6wm2eqQ=pcghn+I~mQ9PC=hFQxde*)PlHD}|#
z>qonODStsAv@aZcsGy2jk$eZ_LHCp*G7{eEa(v`7slphs@?P({Al%8x<S9gC>b}sy
z%>7D$`u0G$W&8W`ewi-K?Vr%M(fJR}<6WCpQb8A!v=U4Ctv6DSv_I^3&oqwh_J2#=
z_t(|d<PjyCD2dfwC9xl0UnW6kKMimiH`LpdG`n4C19ou9X9QIiW!ZHA(l2h~8*gQG
z<$mn@B59Q`e>KYOcaTC`1hH)k-H*d<74oy?m6dT&Q&Z~#tjrOdZ&*zY4P9Mbtv}rD
zz#;eGOzVFzJAd%EJ@j)T$_M`y&wtGh&P>ESjYM_-axCR7HEe%<f(z(x{B%6O$qtRf
zh?g`kCu#5UuxBr7Gq9@!BK18+Ix6=3v}Zj{h_(V+eYSe4;G}Rt!Fl8D2?yV6?K0N<
zld%25g)_xfSIPO^*ncl6LJX#Hvi0AmlJ)x^Q)9bT|FVE9)CFnIQ_EeUz8TD+b+}aH
z#1j;nx7ltNyFB`qqlaPNsS&vKxoV{K$o2;oPTr6>qww(%&u2PbAoG1{DQjDw#j)Ar
zMG=cC4Q-s?idHQBI}?wy2bXV^gGAi0mHFjn=lNP{rOqA=3=$b?G&{++IoY2hxlthH
z@;XFVL)}R>B}f{a@8aH@V!83-d~)V5+c7g1Q+LXI^re5BbtBrxefGNCYTGY=2CB7f
zo6EAa$AD>{`*I7pz&=bSAxm*s72AnJZdR-|A{3v(x^5D}ItTTJ(QhhSn}>xAOGl>j
zZ7JJ^NPC(TjEqjmO8Ca0jt*{$^FXpkt&N42%0IsagV5UsJe7M|Z{zSBsFp<3b|RoO
z2CFUJWHgvF27}Sb?aSih%;-=|%mZT*OGpT0U`NxRDjE$~ZWn$Gfs5b6xB$Cpqrvna
z``v+`tdK|d8He4UEPr|j+#K6oQ~JNozQ4}D1qB>>;m%#xqz^}L9}z&{b(NFFBmU0g
z?Kw-n`$pr#@5eiTdGCWiJXL*Zpc~@b^}Pw!nxgMH^1;0b?=@c$)LK8SW9hv0!d%kL
z^}L*)PYjsezsP>+HGHJp?D4Z9%8G>%{<N?1zzYtCclpknn1bTUZLbC@u`uU?-u;>x
zB!JK*7DT<r;qB`AGl~xRQvqt9$Y#fii&gLSm`~L0Q_NC%ed*L7p8G&WEVmwDI`66F
z9E?hExX{>~Jyu=K3H(cWWKvOozCThm>tbaWWCJYY7*CQv#{ferR?Rdx6FA!0>Ji+E
zW+XYBVhz}yv%iLwuVIfZ`(6Hc8CI=4m1=pS&@K@dk;nv4abP*Kwp@^2n^)NF;Il-G
zKjJ&KC({U%wk?|z<`nJ#-4||b6@3GPV>~~Gi{2^a$#p`O_`vmNMd0E%h&b3B(vmH%
z1w{M~-lYcFez8jT#p^58jf%tVL^kZi8!T$e62#?1(<w&w5HLz-lCD@)3R1)c<nb^V
zqfe_v1w}_ZflqS&93I$C7#)?|J=N2}^bSPqm~+B-kIf8?4e;){-F#Uvzb0mWC16Nl
zQQwV|-;^ld%;cNCR-KGIuY5{pEn5&}3rsK%Pw;@}yS!!6)?g-7vgpA1GVPe#OMZnW
ze?f1O&}46Y-fiF3$w6s8ZX-O1y$+#k<3L}wMG75BEqG#nWNe!FUg!B29@J=FtnY-f
zKk-$ld0s_A8jTz53xZu1diqt$GlWzn5#d~2@XbK)d~v_)sI=Lwi#qC?^E3>QO@s6{
zc7Dhl_c_Y%dK~YW-nMRZyV~ipuQQ^QhpzAajS8?R!#<Mo$t)_!chyD}X82+Kqm{qe
z<@`@H;-MiyLU`vwaNm<ZGjmtct<SE$VF_`-3%@*CYM7*)DToUG$%Ly`TZh?mu(Z^P
z3}Vq8v#e{gUOnnW0`T^5BT*M%ZST{^zIFVh9!L`XvSQt@>M&da9kgV^=XBUFtLb+4
zW6{pj$B7K2+&e3!D*b|=m+l`-cGx^|jn$bIf)-9QH(?9ztxTZsLM)x@TUS<vgZ^!H
z^6@>hS{Q1#vLR8ctyQ9e2n@a>%8Bt6)FQIV{jL6fwyG|$S3pV0zv~<d)V1KcZ^hSf
z%zGMtJ$!wTwQg`k!pDQ_rvmWfFyErUW1hLcF#lGBX`UqENoBSBRvV)6gYB$iezAJb
z!dq}vV|0#D>d!$<UEUrTVNGpp<O7$vqC0SKm8nv5DqSuFET7l4{ZqhN&h=hNaj&I#
zOUUPpBP~txM3~p6`j-YU1eVmjcx$;mN-X!Ll%lBMXLxzgr=t9S7W9Z9<kV|A*7m*>
z7PTN`UQzvm*jLgF&XdM`PZNpqtR2era#-T$pH{Y!p42$3)Z(r%0{zU>YVY>pgEJ=^
zz~Jch-NW_KXONHOW+I<2W2N%}=bNUGCuR?+h=`z;hK7JzDmw>2a-Da51mj&b*nXlx
zHcRe{D!Z8#O}ySN7@-h~gClJ>qun=IpYEqVYp#mYiG&f*&3eB3-7FhUhX14Qe1=8h
zp|-tCq}#HK1HmG-eySFD2IcrZWpt=FqIWQc9Cei(VQ=Ua_kUcyYaS&nNVf+iAx%B}
z`d-n=x!u0*cCZ92O+Hj-S4!<AFn!P^t4DNAXU4HrW3HX7GnVP5o5&lRGopzhoCc3+
zhYNGEP4vwcSrLVU?%Zr*;UnS=?zOBj4C9E2uo0xmLzsP@$J$Tg*rU%F@%iGuQw`Yx
zi=yN^(;^kx4$(E|E0&`B=GN94Oo8j71Nbs2a7nI!?q(op^Wj!Lv}>STV=|cFt4nD%
z-`}VdJPxjWrl+!18!q~I3KxU;zdU}!7Q*{g7fMt~w@I*}0~`=VOvc&L(U?(q0hT@c
z$Ot<NsCCz6GoBaTPu<yLPK0K?lzD89ve!G{;_2zwugp76KU3K-zj}Za!U*l@?fv)f
zU%cvq=W75B;svv!!GD`-TIXzyfom+uXY@rsro!FH)$Vl7vbJq+ZOhuz_Gx^-NV$xy
zEH^ua%hxgEuE(rSP`7L0Gg|HfZ!ha$JudOL&uB6fVm~|s8k!)|Z7O`r_USLQZmxNl
zZ`f#PX6R;~xo_ztw9d6I<hr}hQ+HGgcio)-mfQv=^t#laXUl(=U#@Q&(^#gB`(#%r
zsF9X~_Ti~laASD(n61LDmtR};T4?UUS%`P?;rS0nxZH8uC^-S6h{A_Yw%ED?&`1%u
z9DTF_BhF3|D|(%w^}>K(!i1hD^$hAnMUeaP-+{@t%4lPew{M;dyAAVJXP3q&9{Hiv
z<5NHYX}1rh*5$*cO76@`WP-DVb^gOZ&;uf9^Q_bC>gu;mr#JrP0SAWt>T+jlO&!&$
zgcNXhb*e>4Syj{)jUOVhUHN;I<Mr%j#_jCN2(GpEJAT>aPUmAkXCltU9i982&bB5V
zviR@fLSYvx-wx7W7Zq-#dySj!;Qm$m!U2+1Jq|A}zLB3t^!3mwvWk-PP%wz;5%mud
zvUCN`H)j{5E`U-7-XK2kPs7Hp;4;UE`3M0y<Q>vcrS~V{B;$|#kw|)+r)aX$s5*Aw
zj(`Zntg${@<@aoR#F@dI;6L}@?_4Q$kF{m6;HSV*xEIBhu`L|PTv!<4>MmS7oVx&a
z{+wda_^jEcTQP9&=*3%$kBMfDETghUw9(h+#wdw-9h#b$NkJRl_#U<O-yL1M6`+%v
z{4O9QURYXU`oiH3`Md8JTNyogxB~rs8@qt@74do;(J`1bB$5z$|E6ww$VBx`C|AHF
zTQqNee*TcS(CnZ#ui7Xs&of1xU-D!?Px#9`L&-GK%#m+VMK5vD8R$@D+T>9O=#_U#
z_^egEx18=_Gf(1#e0aXxv6Tvws^*HHMJ3#KBcAoU=ie9hA(z{`Zlo<*{E!U+t9MZv
zCdhK!yg_DSt^xIrrm`G?z!!ZvD$P8;%v4yWtSpfRzx@Y(2%7l^#-(b?H_>*$b!apN
zJ<4|*nMB`zB{=*8AE#koXB?nHM8iZHD*e^bhHU%?zk;>JKjcl1#&WfFbV;ha&Z>{{
zyHgcn1(RcjFrpJ+cwK++9yewgIk1!RdOLkgE8^zhWd}cn&IZl;k8b(Tt4gdTULLfI
zv)ky!*Dd=$CMi;Ak2ZdPL2<(Yu!~-p*Z*+^G3}i%tj3R+q1KL;L}|m{At@95wllKC
z#PX~oAn$C2gcP3la8A40y8mI^RGf-zuwnN2?pCksPuM5!10f2VZFtW34EMMax`~l#
zl**rGd++V<ss@|B5u@<hMth_c#CZPV>mJYU#V$iV+5V)h{yZZCe+n5OLs<24#*^=r
z4YrxJ3eG$AP2W7kPzsCQye`(8HzwO4)#nL!b>#pK2?+_`@tz-9Z*cHnWN-{eLL7d<
z_>v1h!*8py#@5!>`j!^=`PZ_bzeA3L;W<r+_02XU*Uibv!iLdMXV$!amu&2#qh}mS
zu4R5}A&1FRPg`9*A@V!mi29ZU?hW^~v2?U_^n_CASe1RW=Ef&}%JffsT=s|1jt;%Y
z83STP41+;OM^8s9)oNiEx2i2FNj&8(L@ruxH<C`iflOtn9kz6!sgX+ax9Wso!8P6M
zok3oxfr-$8FoK+*!BR#@N(y*j-z}4s>L=1tK4twq<H7geCjk5i`srUkM6=OseSL8c
zUgGI5qS7kN2d|X+q*^MT|96Eb!A%+Tx!6j-{?mwOn^Xg@Xo(O1XWUA`TNOk~|Ic&(
ze;$?~Q}{Rto>ljfcAa1>gFj?v{MQxa8X=e5)c_rxjLM&EU20!zkug0b^f%piBi%W<
zxe{XcxyJ$U&uAE35f~`nDl5U_LthJvbvZ7-Nw9mkx_I=!q-}dsL{`kWRL{-LjoRHW
zwFA&d5ZC$iOMp+FRuNh!sjjP23;K_o$fa*sH@}z#7<totI&Oj<`tK|w(N+^&)c~Jk
z|JB<(WCxsHVB4Pt2~t5bIyRa4`1n9Ao9XVG*}BqSW>?{6EjN2rQjgc;Qlm+Kb+0^@
z0H1+Cqc}YGaMQn2udiFoKEypd+#Un{?4~?YQjb@^r5>J{U7<o^<%_bN+U$PEQ-iWm
z;K7Nb;~_{vC?h2yF}?doN-N<9oQumX2i+d#-)*uSw`|0>)vlhG3gxoEqX=1o9^gO_
z$Up}bSCsImMb8(;gD&TT4(B0j+u6=}>WAQt&CINRues}@Mj@W+Is>@Qa0YG{fvX7M
zY6V0iOccD0z@={6EzLH4ILlZKL@ulO^b%?g8?BtB4{3E@7<CH>0Yc4PT@SG~;u~OW
z;)v-U6q8;0ubE)8?e1w!HIAed5egI;E=&LY{)W`UM)yrsx8J;pcW(ETlE4vg#sZwh
zf$CkBw6}kHdu3j&(>|=z&dtqDnbB^)b7T(?J%aZ#N*yhJ3i!%?K9%k@^!xq=;XQ`r
z$>+4RH1O@lfU&Xr-TVhONOg5}$K4j!WA5s8?xIus;eTA1<Hv_%F#F+Z#O%Q>4_*h{
z?>9@fx3u^yIyS2Wb#BCq1s(hjIsn0qU;~-m{PzdM(^md#%<EAbTx!uDz~ci5ag)j~
zeq6}EUpPBE<7{4<I`qm%<iK-GE}AI7p8)28owu|g$(tNE0<04|+@L|Wlt~=g+1uNj
z-ArZt*NlN1l&gRK{JHw?8D=*nW_!rnl(Lnp0U;8((-<(9>c8(jTqQjEBzJK>HXY?u
zdBa~h8ZFWce-(ex61LH0`8$5TV_P7>e`({j{%#2c(!Q?q@(}E)=R!*^z_RE3#qE7%
z<$n+Cx?Mh21p>b^LTgTTThva_XpxQrlB2|ac1rE%8%`OhVCLnmjkWcZ_fyHM9=xuz
zv^tLHe{e8MdF3?Td(+ocXBTnUe+#}eT;kZbv9VFl)OTxYYO1%lck>P3Z57ygz#|gK
zPrd%0(Rq+=<iyhmcz5obVBk^-s~rLeA<!F|H+CP(UcKACxVWgX3OaJnx7-E=N0+Ue
zYW5oKm3{tB=jo<zQF|{X1$92$?62O;HD@hENZqZkUd64pAYfwo7u7bMo6KAbH}A6k
z)de2fH@|cm>8=rj?=FX?Uw(7E{k(C@bA-j^)AP)!^I{G_%m>;1zK=1x`xj3wwt~n*
zE_p3F{x^&>GBQxs2=lLgF_TxkU!Qh;w4g-3kIVS@KQnIqe`ef&jR<z>_m&pnBpK_c
zyZ(HUkcmStGi2L*AgEkpU|^7W5Krg?8i4(}?x3^%r6ps{9MJs(oj>1KfwNVn7n49j
z(0SW_t^eZU$|au^f-mK~t@gkBB@mh89T#&(v_JM71y;OQ1M$R9euDGK^!D%k!%f|z
zC?wz-9F}`}-B~lA!N#=eM*4Sqj|t3ooR@Ij&oy^Jp(rZT5NS*c`+rS$i*4?CD-yPL
zHqD0-*x+p9qRIHQgtB(I=nMesprL5*zHJ9NO5JA(k5>uZ*V)~z?r>Ge)lmN3P=u7G
z(X;&cTXZR$jV|K<%it7(+@$WHK5!8E?bnp4<|?wOOyBjJ-SvY<kfu7bjrsQnV7BII
z{^NCi0*%BC_@*~tk1yODU`So{M6u&QoH|Yi5&+<CgZ6$$0i0=%4__#qgw;x%{id_Q
z35s_&NqaxZ1!`b{Q_v^*A-%a{>!|%W2yEDe{fn$pk#?|U9(ri+dQ^m#&%t`zZE)CF
zYY;(Bk`e+o+0=Z8^CbKsYtNm1`r?FEt}4_5_naRuotsrNKpXqt3{R^I*unt7mjTUM
zW6HVXb+aejxKm8;|JzDnHO|BC1vvNryIJaXVY~|*WRFM34@ZMZjAmeOK-NHD(_<f1
zg=4td@4yh+s}|?JKI4LnrXNaDM|gmixlxlFXq|q8u6!kn)b-Ff@PMf6I2-ov*r0A~
z1RCeffO~L2fc-DXXJlsfSdo8IaeaLarnzi^R&}B6&`@9B2a&xh&gRZf(FraHdD+?i
z4~TqVb1(@p^8f|MxSg86VIEx1*tP$^neaXnw2#Zh*~fnK?M*s?UUA@Ke0!dms;6&h
z<=@PJ|C-(Ua#*kHOsCFI2e{e)Z?)Qp6Y>(xCGA*yf?W1q-K;Qj?)0>?t5`k$Z_Vjm
z|3dka_Ha4<f7@j*uf-s;lY25`Sj<ANS*0t`N4dg11b~FRvR_fV+ZORNO}%9s4}E<9
zx8V)E8T<49uGTc+7^VNM+iazOfaw3Tny&{<Jpb=vP$XEVBy}$JU}T%sR#7#%vpKY3
z&MPJKJJ$AT=`U$V<v0{FI_Voc6nvw%^~}UCdhj8TNPH|!X;nWQZGn-q`wTDu^Jo35
zzrDTa9%uSOPoV#{+wj655qe>HCa(x1zwAewNK_K-+tANZqQLQR2RFg<9zvT;jK<PR
zl(@`JcTE9v$h+sZ{Ftvi38PgFh*}wSi6>bpP*4bFiiIc&-Q~Qky?$#-Qws1_7gJ=S
zGyPl06lW(>zr7!O$^R)j_@X}f<+qR-6k<$jH2-5NKK<%1_xd{VI85KzvOgQ(Fld}r
z<A0{B{5dR}Io=LfH7_QqJ6Ci1Ghkr9Dw<T+hCmX@I#Ishqx|vDC%qPU+{y?<2EgMi
z{+A=98~Lj*;<#%66F1elkUx^Ti&WW*lhb$H7W>sfaO8k&U2vj4EkQGZM3Zbn)XueX
zA*dRrku3L2WgzH&6N}Q}#DR@EjOw{u3Cf!ohSJRrfPjLhWfY%SS!X})Iq}eUj$;o2
z^a}eAO&n;A+*A-$Vfc3rTy}NoZCjF`6q6b{lLxWK9)Cop%NwGm@Cn6a)G5zrMUo80
zEG-QMT=j?QZRqg{8&gRnl65;T9_r2J)1t#G=UygDgZz@g9kSNJ?@MCSe^&MF<?mx-
z`M+170g%~=Uw?iv9LqL{+REx$4>!8i+COtIsMU?99V3?@`$Tr`uOWw$G^l|quQN^O
zmcC_GR&*NhIS=c9xrnN01Tlr`TD7nEv~<)3lr{s{EUYWPg#9|m9qB0he&8XRtNL3*
zmOuyPiOv%=K5jI~TlKdL(lQ?zAYd#2{S70as1auV>P6JxC=MNzh9Gz#{nJY(1_m!&
z;!$l*zvD&*#+M)CpYG&5A;6+YBO*w`jG7qz85<R_ZDFn>P$~bMydD*^{8Mt^Q<5m^
zf{Ms6G72VEvdl<)NH9UkGCA}!Pc(KV0|P_l2ZpD`=B%GSl|nGT1)~Q`%UauRggztv
z1^h36bz1tDY8m5;h<FrpnRDylsMlN;CJ@YW4GM7lv!*~Gqp{^LzggRwmqz_VF?>sa
z3L&G%N2L=67o8VM?XgU`+P`mc)WL0Ngmi51J6n{`_|%QwOpuohzk(xWVHuLnA7bBZ
zm@~b^{OU?ThDq%|zCuq>DB=YZ`3=AteY|KQ(ZhLTzC?dZ6oKbz*%^H(#Z->kAx?}m
zni33SN|4rJmM2dML}!L5_x8~pCjEV*_=TUZQ@Aw$fBf}^Cy@r}B5+JB!|+K+a3m(t
zk6G!{0?jTcPG<=Xx-4H+<tP0JFdXLjo>YmJnSw2^HJohT+Wcwd#%dfskeP^IqoTnl
zF#F?)^#4QCS;sZ`hJAY*HOY<c-!^ap0@4i<qf1H!1j*4QB?4~rNNH&S>5`D{1{o!t
zf*>HNq=4Yd^S<wM|9Ss=o%j8@&g=S~$3e&mLJ|-T?yv&&QA$Dx80XeB17SR44N{~I
zX^+b<g7bhbm;GMmfkD*F&tu@>608)vEs0zfMS6YYgoe|7PrCY&CD>?)2{iQ|4vP`U
zb82}}gn}h}e6W!vQNbCd%vex02t=I=%&30{x8gA>X1>AvPdD-;2!IMjVJ8-41QsN5
zfKgif8az?P^?@Jux9B~z7%SzKQ|8@I)R=2oe{jy*?DemrG>KRUMLan1WNsP&uPHLg
zB%zpqBt@!Hkat6X;GoDfIzZZlv<iX{=6ExtCNSHmM8k_J=MJ4fMG|XN#Nh(QDD_4~
z91S-x5KfWa*x0jmcSr^??Jgw4q5?3f1wde#3Em9iDlO4g`9L)@ASZ`&Ynn|86V`EW
zQ%8gfKxzY)54V}uAeA>1$ojvl@Vh<l2L}NFhEnv2OJ~wH`knv!N1la9+0(hF@7Nx*
z+eM8QHn&u92z>+rpaQu}Fa^y+KBtT_OeqS15*CpceL__I*)vV;0ViS`+?6&|gUI9h
z3{)Zpl;u0x(`G`n)1Md_jRI2eF{5NSM1F8Tqts58r-htK0+;<-_NP5YrHNb{LJ)no
zq3Ad?KAtH(9%?irl0rmm0Uk~a)#94Hs=OC(uo2Wvo2bApdC92184Co_pxQBiYW`o>
z=fTYVYtoQ{*CBJd7O9w@t!b1{kA%%Uf9BmemuVR-_?fhtyF|)Io{(6cuyXiTWHOT*
zz=4d8OXmSV2p~ej&;}iVqmm58A_cZSlR!imDq1gi!OR4$eMO4wQBepEErOrh7YRy@
zujsgjeVczi2O+dN^z_DnoO5@vKs~K~1XS7wzcw%EQ9HUfA3n9{(FIt#1O5G_R8zyE
za5j~U95(U-DD8JpA%Pldn%~OVVHy9w{AX3FIAhtNz$E2c1quPGkL<$290&jo;cceL
zOMRw`TWyLRA_U0WbGRXwD4ZOiz^-Y0VyguAi8WOTrYOrXoZF^<@fEcp$_R>$2Tw|>
z;z(gwATNq;5=qGOiWqQjUt-)7^m0qNUx`dnCB}IZJNl7CWWqlG?l+ef#MQr;#>9n1
zeD>k89~B@b4i{^fo|=adIYe8<lb-_8e8f@7I<ye!?dASIn#=!g&gXjn=u8^=3q`<#
zjo)DUrw=(p^y1n6H%u^KtiM)aoI&{c;58pGO}l_Ajak|?dkrn0^w+Vdv|D1ex8QHm
z(6Ud4!k(6H=Tt%I0V?$1reNw42=Hj_TS~owx27~$%Iq~H+uPj+Qk%+L{?|a|s{PGm
zE#!Wd(dg?{x5^>s@29ou#F;iv*tRhXelyXD9RqlrI)P(~`1bPGuVdzSE{n1Kg*pZR
zDmVk7@8;RwSYtld+xxAZq011iA}DGwdaQ-g|F`r7NG@+^!CJk(5CO?oGv7o&0fySB
zEYHfKqKy3SRxzdOkz|G`=?28Yo?}o0vzgUvu3%IdX{0D55^}dy@YFE^zMnvXh%xB^
zc^sS=1rk6wW*|pu89AWQU<5=kE_QmXRZ3*lj^}%tfq3MQ;esSUzK_w`1fIoXG*YK!
zsLz0%An(%B0y>Z9$lmS0Cj7@Y0rH0FuLC`;@Ax8yMn4Bkpn9SPdtM_t#jLSBgXH0)
z;c6=6U>g-2-nFg4ea~6fM*JS;l`i`HZf~f*WH%XNnLpep{HXt;zC$^nNwX+B7LCTo
zptyha<MsP3${ceutR4d68+0&)LLvf+oCqXy<Df~{>t!bcx_d<yj!1OhfOFX}X!VWT
z42pIXHw?8k4#(v?qLWj_9yIa^BFe?F5DqnRZ?k7cvK0vR3Kp_28km99wthYB#By4N
ze$FyWo8YHbMN`H}7rA0$43kaK*0U_vDL@EqcQOE%?O74thfnWJYdnxU8~-=Hd2I@?
z*s!1Cd=|-dx%BgNX*!Z23<qGsT9Rc$wb-2#QtiKlpbo@999TH0PM4@iHBl`MuY)RZ
zLFq0<`ume+3zCgva*~(yJE%~f>~2@a{&d44TMmBO{rHyps{3?Gd`9~_ehgoIR@DCR
z%Kz(#%@IU!+!IZ#sB&uT^D8$;rd~+8@kz2UP6b@XtjJ11^6z|D*YM-5{X9)1!ep)+
zI>>0CjGKQ|?wFye{<mv>f2m5%nZ(3MIaG4;VFMgeyv@#u+12fVg?8Zu*nf+B3LYxh
zogVd(OY{3~_?m}vph;A>Try5sjrsx2d88oPIF;Q{Ku}ot=mn{i$5>EMRJGehA5Y>X
z?-`c^PST%9^OZ<OgD65^qw(-|>#n6_U|@)DL!kz%rhaCbm>p0>+;N2^Sc~3s<i})f
zUw)4ZYpH6-cv^7sggb|-#c8JxNs1@nOlKXU3}h(K=q)lDSw+(?p)`xHmHRih+de~6
zal`pg>lg&jq9X#0MwZ2UiLyMTtSkV6IJJt%bQF#)aHSVGY)MgWuwLSsxnU41FoU@y
zul|J-aalMLpoZ_OK!adtQXGDt)ozv5zd&az3$wwfB+M+TO-Q7Ow(?>Uw9WDx6LHk8
z2VU}et_lOGSuDykXyVRJHno(6AqIl9O4GvlLu=qfOEyqphA|<1vWAAkXZ1Sz>rP{b
z$9%q+4%qkaF*3KNt>*VQz;3tFG;2h@uPn(ZnNwZb9&h7A{2&LT*p*%o(0yn|-M%r2
z39WpD5}}5cBtQ%7Nnr+}-y!*BK|!HNF*-p|Mv0KgdTC=beVLx#)<=y<?vZq0mZv(0
zW&t?>9!krO$BW4k-$fm)1fqE=(d|SyiS-|m#H0Yhf#k%~-<)dH48oqqsorM2^fNZ)
z7MXuDccf<$^(|zEmHYfZy!9Eo|4J7lIrv!qYiXNiG4o7awBVbg{zcoKK6L;NXMMhD
zNHowxmu4vXgF&yZsiCE%p(Pv}XKZD&W&hO64#hd}e6IdtbvocM^N#(soRyv~Xr#z#
zB`Bcj#;}I}AvPJ&aiK2s+XmsvO>fA@=ELT8y2L%>syB6<#mI?#oXz4uHphcG1_-K5
zoo7`BA^ix{Xu3ZnU;BJ><8cJUpR*lkst^K1IZ#$QXppdGafKG|aOqQN5)lIjJLU|S
zV1l(xDx^<~98m%S(fB`#OK0qk7YKhcR}Yjp-$@u|=s4Y!^(SV?;EK|T2^^Dq0I?xP
zMCO7u(#(rW?uWEyM3!)IAu!A=B^gEcdtEl$?)Qe*cQ&kqrd~bza(@q*T6_A*nsTF`
zxH5m*adzX!`%_<eP9uWp;SU2}5Y9%UhDoWuQZY|aS=MZOIqIsK9NAv+!{!~?kb`Ll
z?@LeZ?^F4<tpB7fLzCr<SM5e(T;|UoxXh3SOsFD9_7OzmeT%PrZe1`xwBCH?+ZLO7
zo1d=Z)fkC>bAI^1R;O6dup6&f5ts_J2Zx5P{%{l=V6-mTyF&w|`!SQhxmgW=(Ad!O
zeL`X!A~5nwZ~}?t#z8G(s6UoPdGhy5qlC3oX+{%+D~g(w^XL=q6}lemhkj4$`JSbT
zuPQs(GU0^YX~EG+=}H*faQ2Rg-3wFnqF}*GgDunzjkjo_?toOvH#t{}$Rg#$WesUY
z6d{3PhO@0A4H!y}QdYLme2FyZm$WO(yNCFm7BkPQL8vUwdKz7uweJR5OffQ=+25~7
zOIK(450|ztX8Is9mSUMfm}NXfaQtJzyhqV*6;h~VU%VI<APtlHZ#GD}{WCveUl*nF
z5y}P8(<f;^S@ko7TOB37u;xru$p^z;Z-aRW%gcEtTT8><&?*54i6BtIp_ApIX{|nR
zyw-3GfK{;W(Frxs8Z4MT9$16}^rVY#J%46}wD2YeMs`mK3IVaWs5E80dYN%O_-r@f
z7Mg+VW!m!+Ou;>i9a^UOk0*Dw4%`d6&!V4&9l*?phL6O0oG7AgMlEg5OH5|bhKz76
zbL?n$92oks4?>4ERB-VU+&%WO+qa?euOe~QT1ee|*vPUv{JJrE{gi*TzL+qP+`iCP
z3a8T4pS8kn%&t2-<*(BOc=RAE+tT#Qh0NKb)3twZG14R`a&2pCGv0wGP~^c<Yp|JX
z5umoArL84kS!7l_tgK8=5zkt#7;lqU!#n=j0WYF*wi*V30HM^AQ$Go!4mu;Xk|;Av
z#0qs!?E$d-VS4yYriASHiO7NpJL`B7Vw~n75ZVxzT~J(qlf5ID)RKrzfBtW3V_2U`
z-R=U@VHB7}27kS6QeQJu)+I>-aCiqR!=_#02Z*`oy!XkIOg8czzM5ruQinli`XoU!
zEoF&%_cf}yjJNJ}`!jr$@JUKBn)6PzRWH}hu-&`QB7BW9TPouUa5FgR7G=ghwQUrB
z8&OHN_CjM`^3)%NQcRj<8^#-_-i2gB;>$1@5Ne2$)DG>ySeX|aNblGa*(wA5SQx2V
zNfeIlimzZ76^vreOT%$-I$4rB*(fU^k#4~Ndq+LPBzo-_)DY#AtOPNPIaqB)#2dEe
z>Ps(88zMDE!3N2-vr7hYKQZ?SFxTiRSoBQYPCv5Y_gpso2u@KEgo#3GARo&FF>uYI
zeCzvZtdOL<Ad{%{XIjGU`y=(ykP^U}j??0X0(@X%XW@&*_4ZVbtmX)hi=L92hhAl|
zcm9fQ`0MXI!3A~x5n!AQw<Mn=W4XV;(|jnuW374oY^N^j@uW5XaqK?<0Q8<b`PaNO
zh~lvkDH+!|9v>3v@WNQql6nMzc_*05lv}1Iv;cS9lcy29ms^QXdm;(o{2{9ylh2h=
zq;3;|id1R5v~Hjlj4_&?UgIQI1V?5+@b6~a=u08BP6?>3^qA(^r=fehcPpwVfCz`^
z`+up`LKa2xf9uGq@g0*?#me{LwK&T{t@VIl>x0mAYaISXL=j)-(8hC2Ln-Q&oUuw-
z3xBjs*oEyFksb{k&8(X9nSyx#;`;jhpnMm`Lq4omkt6|L`o4ta6ZScNV>!ZpK@w6o
z`-D1wteI<=ekN*$cbt_FH_3?<=Y+zA%Z&H_=)M(1S;;4hEh8^2DB{T{zVhBz1ck)E
z)k#m%cg3Q#@Mu!v&}B>+e`QcQ4wt~0Z4Ht~6WfS_nLma@76ise@C<?_#PYDg)gNp5
zEiz3NJ6>PfmSRO2veS#$p(bOStkSB~ocban%0hSfd9=J%H<_bVE>2!62LH1nMeRDh
zBr=6p(0T7u2OtYda7VEB&(hKk3Yub105H<xr`0IH3j-ZY7?h!?B9j4YCKHHsItH~*
zG34AvB*uX@a#8G?eU<~{0%X*aEj5!TIxWDoPi^9uO@Ej)*SNgHyg_IqFVa1~3?M5`
zbF(Sm1Xcb_qDEpKGmIflMQC&~!|kI_gZV@k3S0vwULj3a59l*JT?QJrvR=0t&<vOu
z?q)cb-=_$|M8dF~7a2e}BG>rtu<TL5Z5{eMYlv}-UPinh!wfAzB;9IpDI0(V^iSiF
zhB=G$)|{`5Q$@9thC~MJ(b_T17sW*X=CL^@qBTrNEA6tt{OZ0fbn0tK?aCsDyKMa}
z0SP6oa*fer?C;rMT29g+dB^RiCR(=0FJmV%=~Wn{*bM|Et%4eu8YK{opfwx_iHnmo
z4#EvdKAlnzNs`yN4wo6<EkbWVI1mWI;Pkw*KE}Jm!2aJhPfCDo+liWRkppwuQ2%hk
z<?j{=ss`->Rr00eA(u|-go{<_Pu=8QhQNe0R%svd0Y}zRNDKtaIHX@)PyoXa5EzDW
zVI1>aWZp!HJQG&i)`|RTffcNid_b(AMM2$|=;;W`AYbQprAPz=d??{%%<@9_q+V?S
z@XWACQZ*F+xaM|sw_4xIVGNV{XJntV1W$Au@9~bL7{|mD>DeNMCnzBsqNwnw&@qox
zyVZvX0cI0_kyLD2dK#@(-`7frb(G^jOffOT_Lf(^3Yg!61|(04DPa8Q8KV|ebil+<
zFKSN^w5Z+qSx79>Ml0?iDo!xC-idqk__HF)>T84{-A^yAp%9a##$66my3~l!x=r8<
z>`>>kzel#+HD0=3U-wSe)@hj2j*L7|BtJRvL~B^UK%~88dbEV(-d<xf-hn1>F{(;^
zeH%2!DPDM+3$>@-YV3waM}glbY<}j<8zrWo5f6~WUNTdTAGWBW_f387?xD@RjZ$qt
zPI57wK4>LPnq{T6&YE}BFGq3GeQ*D^d#(PN5}>T7#Vh0R{eG&Ots`yC)8D!WT2;Zb
z3lH=2<wm|bT5d5a?dFaLP8p`TKE>QEk~pg&z5=ZkZ<}6nS68+<xJ+b0W1iLOWvn^I
z!GSwe%0W#_!lO6)o5n{w<_UIq*3WnKbvSkt4y`>7Xf-mt?9Sfvwu;UaMWCl7N=?-N
z)a3!{qYdb(-j7-H5P@br0J(M-qe-2vxjxz+6>QjVUx^8eXT$-A<v-#uxFRuxup||C
zEsuyEoi&~gw;<_-*PlsY3D=Wlo>61+RFFH7+ioM|u<&FQh@w^vL1r*ANpci(mXu)i
z;Eu0;2`BodQwIw%1?N#G@c^w5)*=RKc}gutP9iR`z6`?5Ius~EVO4oYH54LO-ik14
zM<83CW3a3U%bfUz2#w+1WMZBayCRLe=c&tm3%Z+p+%qRp^QaM9P%0nd99sX}lY<<7
zIyIW!9ZY3elwAt@@G#Jyc=7i8*ea#lr=M*U($DG|i*&U}x4G>SJ(pNQ_+*(M<cvGD
zQxN8w24?0nD;=_xNQ~oXt+lSrl@K-`@}eTS5O21vaKjQwe+w~5$ilo&Wu#Ey5AVtn
z%ILG>p3>C|E6@z<ycjg?H8TlzYJ5WpNL)c`r}2&;N&D{Z(KNE2&kSysh%~HomR5&R
z@I{7lrZ<>)-}!@$lKPSApN5+_XP?dt;~r4)0oUfG3J`t#Oc`0e#;IYDM0?9ET4pLR
zTx878N>{G-#0H~cVIKt+R3R`lLTj)PLnmAAjDOKI3rQw~?owZNUi`^<Hqoo<r0%*j
zd7E)dHv>_kJ|DZ*Dc;C_hq!tB*d)8oFM_lre33X4B3SF$#e-$+7eP?hi5qgl)GT+1
zEl5fGx+q@^<XV(CBu_O{DSrH|E!9`{RcQM4oh^a1kOR|=1E&#_7#fi4Yz3NA?Uhlg
z<#C;~an<H;jKnb|F<vnyu)?Vo&GJr$#^T$?*+3kCSS+g$Ksw2bG@TrfJ}-UFZdlPi
z`*r$jvEHI5@j=idwpNQrha5pVbjmZk;B;$FyAomr7>Z0VN-swkmTt*rWK>A`Gc@>O
zl~!4s5R-uffb!Eg9;}LliP0pum}xqw7aK7s4jubE)7i*vEI2EmI3nB@nS8-@gZM8X
zK8inWoqT@wIe~;`{EO-q&mS_<fr={C%G%P;hD~1|bVrF2YF2r`Ky&PZjg$7dMR9B#
zn)bb{ERl{|Fd`y*gSir7t0KpI<ENZT&BiR|wwTnh>(85H==2jDm-Y~@{Q@4kDNE*^
zw54S9EOJ87p<_%`n*cB-NprSJhElx~#n(6}lL7wlY9_SC5Y7+05Ty2_vW^t;Xo=(K
z;QYCW$gmy5TamB&Cp^vfKgulX_K+|ZYAz7}WNqDEModTcz5SiG1)65Kj`zeoU%pjz
z05fSHiYrO;dpX#JZ+;kG&1^)z98)GVLK322oSF;yh`m#$F-0;#H7i6-T2`J~mk@u{
zF)mY1*=c`2Pu$Ohg%8Z?W3I^eP^t?Jk8c$$%gqdx?@HTDTw1!$X|AgIQgX1WjHixS
zR%#4;568Q}!U#^UC7+=kB7$nGMzlF%^#3H>BiH)c{CRW8mIV3JoGF@G`(yh36{baN
zCADPNCk8~qbiy>s`tDj<`!8n8h?o_JgvU2)c^{nX<wYhkwX_Yj2zPJ=W$`>QJeADL
zb0nvDM$}A>t;-)u(@VLD54_75+Ih$5X=~GVN%9^q+v?n}pXfu065rD^q=?Th#Qf5l
z`912&K)1F@D4$Q{fCEepmH=3pBf9}~T{tWwQ~|Y%e9qS9tDd0#s320`D2RzXnd_Yb
zFvG^3;3Lvd9jb{s8b}x_%hzb=UzJ)&bff5Mf=Pe)r@PI)zF+Ka^%2AM<E6k7(NBLx
zE(QZKfYum=OA^=5nZlb{C6@yfW}+zV)P{9Vc#Me&5>*0ik%<;`ByG3V0dXnows?4H
zyon!($kD$F<ubVH%K`y@S~%*L)W{ZpPfWkhWd0SOzAs~91|QKsK^El<qptpy`X{k#
z*7C~kYh5g|EX&FWacoyPo_9Gyslj3<2u=trmzW4d4cooAH=%<S`>KKQcPR+_{B^>_
zOT4uud@Q3@NQk57Mpn4>g4~2v_=JHyAw6NHK;E2FX!*lZogI37pZso+&G+ln=D@wl
z&D)a#K!`Eek|1C$*!<^V>k9v4hNrcdtF{WjhQ_Yj;!1XOsH#q3+IFfSWpuoIP@;Q$
zxE)g$-u|ZxLU^zJzjuQe8~~`OlApRgvY$#TdY{(IZp$yDDo!=nG4GJ<J>2Z4{zPzC
z_bWf^zpc$UT$m>J++b#~(Cy9-Rq;W)|76Nr^So#jc&NEJ6^#je0ajc@vO^&nKfqSN
z0$yPbLfD(o=JHZFZIK4+)}bXZFyU4i4YVMeASA0Z`f_sX;VF(0;IyLuF0}9S?AbHn
zA&H+7I^Vm2_2kzpaO__7qwBzx|8{R-*BtSLHH4IdageImo+ox627lh_?CPW`rHnlB
z9V3kB?MNpMZk7fRY4E9$W63IrfO%WH>tre7#*{T*KdeGqYS3Wm3m+-z$CfHPD;bv+
z-8~x;DJBF%qD4#?-3yh?ptxLwRhPk>pKCiyBSe861vhjdlNmLV<@GAH0!HP<8qk3l
zy!va&-npZSz=P4@??p-151s+4C;q0q4`p8WP6*L%O!`@$x)PyrznvcIraWyFVp(eS
z7^J^y!^<0n%H<1qb1?1}&nmVInA>(&a&E}r1i<+F8w)VPM1Wic2_h3U-^(i*_VQ&`
z<`#t~*gq!km&o_dBXf8(7$zpletvd#2mS#7Y(%sll6XjJD;^U?1kuUs_8~v6EE}2V
zKeyb|gdqQY*pgQGB{ImueJd2Oig&Fr0ke?FafIt@99q_0S3fNf-%fbyuatD0@^=-{
zYGn1ZVA&aSh#Xw<*WE$3kDYsOlep|6P|0)_!qM(FDhSSTa~k=f0?w6~?yXSg*5h7*
z&$C}2+@I_jO;_{(_Wt{<u$-<zPRaC7ua@TKNHBPapHFRI_=f?(KXzVpm3KTMV#DKQ
z#O1}?qnqAetNx7<y+L-LsP#4e1gCMp(g<L$n_7%9j;5p@3iJY<G0q&+UnxcGPDl`B
z0E9X>iVQz*)J@R?MGVU-LP8flrBrxV^}2S>F~M6hl}H_?^NuLgcim-JKZV~Y>|%%H
zV>q$k_`h@V(8MT2kTLa|yi($wacn<gZ==Hj<KbJp2&yBXdzL)JCPQ6m2Dt3{@l4HU
zq7{Rgv)V1IQxnWC_UH)|B)Hfyo8e*|7!Eg+y?e92UHa^|z#z~1e9yS2TYH2bME}<^
z{#gsV11N6YGE|;W4YW2d9l`+~b9t<m-`*5-?DTj<ob9)7&Z4i%Vdv6$XYa;IT<L(X
zYy3*zR_<2aI>XDhHq7ti$2WD4AzjYKei&2sW48ZJ(TL8_CH7v|`52R`&kYVX!XJjf
zvP%*K3|b+fNv5{z@;C*Sp&W&q)EPm~me{79XBtD4kEEW_YzEEjxbefScYD_b;;bvO
z|NHsuk9SyB{hQBRG|q4H&+bkbU2lnqxVk;U$!fRmpLRBH;_X)u));z^I*agnl3$ow
z*NXMb{}!uyw+d&EKy|O<c>osteQfw1Tgxm7sP>eX1gpRga+=Uq_ZjZ9U^0Rq-*UU(
z6}GfE^i^Cyo$+r1T=t^*W|tWGm!SYbO{iL|Dz~nyo14!McZpOJNuD|yvQWEmT7aK@
zNtn0MUi+gsW6~K7D3%!7x3f-&UB@KwoKcktDWLR8qj1xPQ^pkSa7H*a11Jw1>u!@A
z1?CMK0cgYZYnlx$E!>##-t+%mM3mPq8I%H&K^Ns;+m&8%DmNMO+^>hQ;!uvIbK-Xs
zf)P^^ceq6d&M?Uau~i&h@{(fRbj8JUlc4_$$UFTar8DnFlq<iYAAK*fz=SWZbVDq&
zSRsdpV;6Q*gu)YBw5y-RLp=8!y8)znQKytE6{0kjSi+7W-X)IMNM>mP8bE^bpSR)_
z@3^{-X#-WYkH5WC`p1>>c91{`KotY?Ort0!?z}uGf9rY&|9i4?{ZVAEJYj`i4cFDw
zMOjx3L%c>YBBTgm8R^Ob82uwQh=D48If0A60PTKYVbT=Y#>!TS?->nu<sz*}c%>-i
z(A_vnZ@rsC@25#A2EK1={?)1~!Q0twJoZs~@=|oPU)X)`o%Xv_so!C*fb$v^uh}w{
zxg$Z77$6ElGphRHuQ7}<wA4zb?^w#Iat1%G-Gt&jgunA)4EM{y7)$gl$*7lBo|%wR
z3mc%Y%|Koz#P6RXa{lMe8OWxg;Zzpql&%~9=a+PZCO`{-JE6HosoR=D4NlL+Ej-LR
zM6;p2*2TT%O!Jd1plnd)qI8sQxCgI($L>J61B}xM)~Pu&R2o-X>y5QXWee`?K!et{
zK%-v(6`a7{gd_&6Trj;`mcE??21!_O*Vi{+td92by?{G&g*B(@W2;*q0@RGe<vlE@
zl)g*tpGTAjEwmgv^D33L*0Pk`1rOxStu7C!AvMWT`*(%0F;m}%(9o@)y>WkiY~wpW
z68%X51;l6LXY!<<=K`4%IY6y;SwW11EcGHO{ExitF&(y%KB`>kQMPtfHz@RQ)0nNx
zo<hdUae@xl4U)-y0=TlNQaD_PSiF_ndlD^GlHfOOqQwW2fgo)Coabje_6sfe5Md(0
zf*<TiXtxv@0y1U#_@4WR=Dn0@A>;kEbGCd7h5f>Aws{Q7^~iP%e+!r*ekXSUXUE)x
z(0}=bS?YT8@_rWkH9L&Tm&DV7c#73B?m3`q61bz?`PL8kND%QnXL_!Jz2bo9M1}_b
zy0ldhwRPxPDeg@roJ%tFoto)=&-+46Q*BJscF>Q3$ObpTTePltkxbP@fXVsjYq*uw
z?|a2BdQ_`xJ!Um8u&6?&*!p@Nc*YOitwLNG#`U*9nkRE3GUcQWD7~aw%%hZ}Y2H=&
zy1M>5zbYkAi<wNdE;9-T*@8^U?6<l4mEdhspVPBlkoekJp}4}@of-;$<h1@2xgfj~
zj5szY2v}bp!Ct(*{-Vx+O6Kwy8^Z}uL=16xv#dn?vAF(EVkL-Cp=LE5C=cn73@wCX
z*n^t)No+{L+^n?BRqLVCS~LETymJS34vQSqOw(r}@|++<?8NMWjNeg}gAxLPH_|Qz
zgBW2?6o4`AD3}ex280wi1(%A(EXg>5gJATC;q76Vvd6B(qJwiCaIT$P^%>iCvoDyC
z!VBg}<;&Age@ch2BNPHDP2>koU{<=Od(n+ijQSb9HUcEX8Ze?~ydlJAFs|#k#0x#L
zCui>cu6=Yk0&sd5q4MheviBbQTK3@|FZ8W5><IO3jrK^>TGHC<+2t*Q&;WY!!?h$8
zSQrTn%xZ6am9L8pH^I@u1AkQ6>WXrnNI1Q^a?g2C8Zj%ip`WL{^Y?2JruWUY+|}8n
z%VoRbpCgvF&THbMKQ9Vt{=D@k26O@R2TQM~EjmuFJ(u1S<acun9J9!6RkpT^5H75y
zZyj%4q2m*M=<h|V&E=q(og-((<E1V-htvCsIB>XHX)$fgRd`Hv-8C(M`U|eUj(e8c
z)iqu4QAM2N2_eht+ICqZ%<!?6_>tyA5<QfOlll*Qz&G(k1TXH15Xh0w^v0rfkN=JU
z65%3*tYh?#JmCGsU_*3IyzM?hjh#n@%E&l>p9z^i+nPSB-^(Y{a@j?O4aXnmmgy4%
z+(S+-17z44YR+V)H)A<7FnL=ytraP4NeY01AEZ=`V7q)0!tGkjE!{83;-}qsNm~LA
z{ZQHTpy)hRauh!icqJB$SYDpamcg_3sHt(3_>%0K`OmrP425bb916zLT;_9AL2+x4
zt)e_<Ndg{-L>ouu-~Wvg07iRZ@sVj0l^3h^nM@X804jZw33srCFO0;HoQKG4YoQ1&
z`=qRz9CNrMj|wY++>7tZuW7Z9g<8;mhd}Su3B8PfYr!RA><ktTY&uEgnSIXOiC&Pm
z9qz90hC`Ou-DQuWz!IFjcCJNHI8K5m9=S158r*I?HJx|e9*!EV5;PC(F;-*fXQje6
zKfIS|{Z^@ropMrvPo}G9CYRJu!F^8)Ye|p5C<iiSu>f6*=nEx$>?nI^Ub2nZ5%JvL
z0(W+FoIlA|_Hk|Zu$U|z<h$*!e45^fPi=7aI0-Y;+~H7mf_8NO_X4#Hb`y?vljlEO
z+zWaq|3#HqX{ojGKl_E=jb*D9Ve>X$z9kS-wWL$yRC6QK+spoygM%v9wVy1-=jefI
z%uU8C^gI3@tNK>rkzgGOz(<_DQ$i)pL)@f^zW}E1{Y|-puB=ff)FUMn@3NL}pXJk>
zEHv)9IA!us_<<5KWQQDN<&qs%AC(#8^CCGwT<@A<z>m`@M)O~_b%Dy%S5GQH>xyfR
z9U|)W(7y(Y$M1x%{ZDjR<PDv&i_^ozF`I#<0t8n?)(>W+7@n{rp5i|6XLN4K2^-9o
z;F!1#D-SiI7R!0M7nQrYFT}Jq9LkF}17ZUVUj>|}8oLf3UOW%_Ws=LNzV|l1#V?D>
z?cL^;YQXEudSBQdkb04x4;6uMT~X+kW7%j79zpBHf>aDjbL_42aWAk?sjWJZ00zIm
zQ3bJLkQt#(@g*(BnNKJt2ng<>zEBe4)b#!5WaCR_*WD%Ps|>$XFHzN7Rl=!j>T|6_
zNr>SP&&m`kk_OYfP_@9<XRjK+z5FIe)V1W0;a^upk2I=sb4zp28EZA6tbYm+IykX!
z4`YH6;iSq)fKh5pyRGEuiLL-zbT`0qKGq7ei?v}7vH&CjTa>$c*oECAQeB8B+2=*^
z-F(|+rjRCJIr@h(2SWNao5V63B~hI8ge&i>j00ST*|9eypAYi9C&aL1-Vut!W<r+s
zh%Xh&+;1!sY6@PTUa@ozI4yOCC7-u21>VFa=*~Hr%H)5OxymJx&9yxt$)223$ieh~
z@rOOP5YEW__APd!w6$0F?d^~M{{5a(0P}sw8r-8+Sj5xt*2wf<krvun?N<rWGL>;K
zN<3-G93=)Q($eDuMh#5v_t4S+(T9~B2wU6d+9<|SH3fvHW=ybgi({&(k*G89lf9AP
zQrdg87&!p~okjLARY-dZ5nCHIP=L84BjgUo0bzv&?>y=diPtZ@7)&%tswd`Idp}mT
zN@Wk2?0xz1d`JFaeh(B1?tR;s>;tR|&7AG5-F%TVE6p&B=BE;WDRTOV&U?#mRnv^_
z3wwOktVt*boSG<4;a8U>5E8lxHiY#clCc0;gw9kNb6!?DDKF#?f}^A57p9R&<fxW@
zyi@Ltho7ek?`FsA*E?bhn|<GeZ;j3S=)M~hlFEYab#+rP==ntd>dOLWcST}@6V0fx
zTKGahZHDet#mdAJT)7(b-br#US0O<0sq@SXEMD2iKRT^$kPgUwTuYZ{D4;^()jM7G
zk`t^yNj^kp)}7J=?W#2XfPq8s6BOiTnHMSuNfPDM@`JT-uw~ha&D27%o~f$Jkh$l<
z)WS<4A$^wdbznlh1O%;ULHH$}9taG@-3?91*jDG(c$3E?Y)g*x6H=N(!r%%L@)f+!
zUfy6>J*x*}JAox6kHm+}bSuZ^iilo-Sq;nLMFN$-aL#%8Tm^t($C54v`MA42@eQua
zukZD+NCnGt<|}iD@zaT%JT>OHIw<J4S97Vux$UnY`>nQHyz_U$3dXbU=xhj%8dzfl
zzI=OzFlv3-vE%z`t9+^a8(jGtERI58;pk%g!{hPi`Vy03-vlT!rJ<$gBJS0}RQGaz
zNv(aSri^-yBy_39s32CDh%TC4=11-)(l~Vsj|Qev9K#|k0a%ecYdU9F;N2sY{p$K)
zO96#)bQ#-(`N_R6&iZD)MdMI=Qte=Cd9|n%|33Oh9w>RY+wX6fgs&a?`o?^;fxSyW
zS5)c1L<4H|OJe_g<HliwSJv-%^T)`0KVTT+J%a~v%8Qu!y=u>yF6~q{i}Z1}!`0%;
zn|F@`ZvS3J1yDQJPrPOol+ifu2&`OS>tgQyg#33rGh*g~y!{^0zg8S`CFM!{{!hUF
zbgOD9e8Cp^JzO!MzMMHf_y=6FJ_)DTNwxe@&LSVaRiA<a!H7%bh(w{j?}K44C|eKF
zJ9)BJlNd>er6s+E!CBR6ERC#+RRjQpptaJtQ*Qn`%yY>Y6ec<K_&S>~RKcMbxz_H<
z1mOZ8mi;+G1)R2PR#whh{bfI-!Mi(8ZrUSsw}m-Y?!MyUYk{L8qmkdc*qx2bfR0vA
z%OKdyj;t0ixBzg${yGQg9<T3n-Th6uXdbYP2?i3}LriBeruBCjt+2{S$l!8)mB^Wi
zC#tqeRgc*5WZRx9@;x3gQdn?Ph$%7w3r1rB>+X)#;;eRA&Mi`0tmjQ)x}lxu!S_V!
z^vDafq}r9ZrPN;yU+pWF4;phWx764V+?Ng-lQ0cu+K`DK^loi-3U42U9FA~cKKo%3
zeMcu*Uv@|-V@>TNUfzBm(fzQBr2n-Q)Nc~8eEa6I*QeG&v6kU<ch~WtIszwf(N_T+
zGIHW}=>IA*RfcFTT`sWIN+yQNveP<<fu#{*k(ruTe^vZDX{Ar<10AflHb4C|qrc{&
zvKDT<Ko6xK6MX(+-r~=FU+E|6K%5qnGn><)1$LkLGA`RV+e6!suKn<joHk8AVkjd8
zfS{%a;Oe&$QcO+5c@?D!C|=xUy8-SJtZ$A^WQ2~<3H83IwGerD9%tQQH>27MO{W7d
zDcQiQ)7#tIbtGmf`&V`z-t^h*j(9Y?DTt*;%vO%~%}ofD4a>etu=EcKB*5rvGlStN
zt57VEM~|)w5d+5(PUwG@WxA+GB$QFjbk{@#k-+J=*VY{mw{U>a7)4&KF50kC4aVw3
zqN+-6y-;88k{1<c^ae*J;6mC|YAhuqiVMUWK2a$QXeqt&>QFlKVXya=U-Ni!;C~=3
z-tvV>ZmsGdfM07<?t{(+=8?J9qnEpvTD16~m`5)=_S@kIn$al{U27CRuu5s=zi@CX
z=IprqQ713q_c*h5aj+6$YMH!2<CFyGNu@Vt{)R*NPu2Zhz`ayuSQs;wAcHucxr{`m
zgc(|@M409+?@WRC`+9k__#P-jK}@8C1lAN>UnmhmS;~{ge5qcK>2Ziz9W0hs@Jx`>
zJ}5Hho1&E$rVrr4RZfsV%~IJqxI_P?GJI?4K00Uo`>z26#!?7^idTzEl{v@QAHCEn
zf**+pAJ?b+wifdQ!3A|U(zuoQl+z+7vJFMkv>CWt12%}G-C8vWBFeFBfC3kyP}b6x
z`r;y61}MprduAU-T#O677rO}XeSY88y|CBLR#aTGfxJ>kVo9Mzn&bX=DnMoZ=cA@q
z{$DkJ@D|%V@5!L(<0(qzFPqIERf}$qi_mb}ou=_bl2j+lvvVoci0uuSc&;8@7_T#i
z=gmrKPaW^MMl|6^VfmJKIp6idkBo3u*pJzJ0$t$Cneu0dpK1*>7^j_I?vDbgPLb^C
zu4)94Bag^Hohp<|)ob?)p99~QJ^}I4zxjZEK5fyHx?aW~l#CZe@&t#CyEwkzdjjId
zJUHM3;b}d(i<r7(E`L<UPI)vg(fywF%38cE*mzF*8>Yrf$h=;559b*iPY=;6ZvuJ_
zz_!pkq=(=xT+9yrPWYi(GH0G|&C7hvI=Uc9e%wf6MrjZ?Dvbo&%u?O8ETb-3<l!qW
zxPHS}e=K3zP*j6uNkhZETZzPSS&J~ly(!Bpd)n)MPpKG-pMLrLcpv5=^ZITL!xrYg
zc#KqajOb%piT?QxPIR*6y<`Bv;9k4CQz93x4)!o-nLYZ<&g(BHbjkD-fH%+3t|u=v
z7brri3j*X*lf(MGRB~ernuC5;e_f|B9Oj>(SW+iwS5SH{#^tgQ{`GYA4t9k(H@`nn
z+h5$|sE))YZM`)x9+DNd<k21oyA8t0qy8Y|viwJ^H8aQd=94zhorQWl;M{F1jE6a>
z<mKWP8iTPceQ_f`$k10)8t6MlN0-&8<mS;uW<b-nMQKUl^2*A=#dU#0j?M$bm)rT-
zt-tfnrLHDpS>;jSBz-j~RUoEKi(X;-L~>m;;70D8OfPn{xML%VBIG46(A7S0b7H<f
zeB-03*jnD-&=IeNEWPGKa-aZ0AT{rk`h7|&&pCJEM)#9d3IgTbx8v6Y2N%zIR6-@I
zIT&n|wTNC&Dkk;|CF?f{Xizq*(uVg`2~!{Vq~LmI3|F$k48mm*U}}sU2uvaz0e%2b
zC4Mdw*{slHKp4W&tcNHCTCQ#rD?q6&fR#(jCLgv*YkFnv8^!iSoHEItks?ET3PM+c
z0RjdjR0^9S-6I(Kbqh+#KkOk8CcSQfU}$=^K#K1*y5A(3Ez=b8`enSZ*VjYf8g^Ls
z41n4Z)*{Z)I7zQ$!aUD?_=dO8Yfk|MiIa@*Z6%t;*kPa9O*8GT>|-MgXdP`|;mW~5
zh8yG{Uz>YaXQ;di$%6ulDP!FreU9~@)mwMWcIUf;KjJ}>-)ux02{D1|e8-2M+~nU+
z?rbbN*Urwqr3Qda7#ElXNQD3`ssK1Vv<?I$>T+c~r*j8k2*^N?G;3-N!VmZEbpvH%
z#H7J)7B7sC8FzEhSll{JWZaW)rd3-Ci9K>Pb=Np${#={BA7k0b6wks=M-{?p_B@U2
zfyUNggg>_^$M2>c-}YqP`CmyC&_#=UiiA4la<+!p#)f(P>$|yMp5(&A9^`twlc&WJ
z5QUm0IKRf+h-haLXx4r8zW~G0P{m-d<-kl*SVUP7n1pvo4whf}AoyW9i3e1UCZUa_
zab%=vTFm5wCX<>hE-f^DbjFP1SAZPmc=*AIjHyO#r8MSGOD7QiH<eDaUer9qRqQiT
z`@7T30!fQPh4}MB0wADA&>gt9o}X=B^A{>9{S-4AmD$RY1`HLb15?q3lj?VqPY6VG
z5WD+Q137bTXCKnc<>f;Jhyjq?crp$>C_CjfqN1|VZ+0hsv2VU|v?wZC!PonsN+DD)
zRFLa|`}4QOp<Jwlj4@oa?2@Jnc%<kPPfjB@?U|MX$7$+z)68ob)gu9c&_{ot^$>m%
zD8*1X`nBF={?Edy*C4}4<5TFXE{b5XJOMhBtVQ3q#fkbS^rhO66@aAW6E%HD0|BC*
zAJQW8U#cTJRS|serO)PAg4y{DBvyPu)wxr&j}DQU=In3(RFsn;s!vyOiZ~t^qKxm)
z(cbyZ@2ev4jW2lq+f`~bLnlYjpH=vI8Ae0{GV<@4P6$cby2$nB+Kauawu^o1$Ga;(
ztA6g#YlWB)A`G_r0l&@I<$9LM=1dG%2pLx1nUKPT-eUz0pkoLitnaeVJb%+Epfl@d
zkt#7m%_iCkW&s_aBq?Ej`Y!WVKl)m1@52#WIGn(}WumP$)YZa>Dz1Kn+u9s?jxm}P
zmppn;(%H@Ho)F-iFT1pNZ^4bJ>y^r%F5{S#!reXc3PaMe!VH#*?=q$h3EP=E{JuFF
zSG@w&yv&%T9|x6CJvU2>)|FLS?s@`)ws<bso1>o(mKhFbwp#d)R4{fkZO`Eu^5g}{
z<FTJlYOuU}Y7-<1gfD8COfQ0seKGxfSr`jLPS(A{B3F7>vWJh_S)jlK2z`RUiQfR2
zmI(XnN&W2S(LnY)bQ(&8B@Z0Lon`^8#@oRK0AM&%hyly^wg;<{3zxp}P|Y|^85x<7
zbsQ(bFXvYXtq<!!cKY>D#kAIEzdHUr5;bxYtoJ~0L+zQjm=7TByD|+PnEex+YrE|u
zKPS(<lW9{`ZuMR7sbrU_)n3tMfp&Fv14wvCL}nOwlc!Im0(Q@oO!!@@S42>{K+!^r
zqNhAy=JazeanR<W4&+ngR)Nd-yVo{5oqJCevA?iWwBXas?c!j)i$+f>7$?szKfo1`
z1qAS;I9?>~;pUuwf5e?(UbWAN%P{87Vcwpms-9nc=NmElgM3#j``1nKEILrmxwD32
zO6AdiIxF%atvxF?k1zj<A7TDpHy#E4dHGKqBj$d2_DSwv=k55voA;f6hAVnSe5qPH
zwz7yCG%5lkBBa^h)s0aK4jmre8kx{@-@g`9eom7zR4pQpJDK<|tR3Se*vUPa*%Z&!
z)E6h7{(dxOU~lG#kNsN{|NC#0pMtaRtvPdzN@bqc{|tc!m3x^berqJ$6MFpmL~-&<
z#kW_<ed@=AbFu8JT*=&&3iJQ~bYi-8?v4R7!;$v7L@D%Qb)>B69w5^A;KCoF_+??L
zfi?(ONXq{aL@&ehRN)ANlPAoDRDCjUe{*$OYG-oMa_#3ecSEV5+u=sBE8H{ULHkK5
zc7&c~I)mWF)PE4y{COt{Dj<8FV+V|;3HL2oe}s}CR4<%VEsuV32>ARM_W43i{2&U@
z^Fni2CY+nBbz~A=2+M$IqE4GXjBmF-^F5I<;_eAfEaU+Ca9U!x`#?CW-B3ePst}rd
z7P$$Np3#cYDX-aPC=5Yj@xg!ZFs|2CmorE9!3nF<jNj~Rj~1iw_=&u*sIV{DwsICY
z<X-wj(nfTbS`QbLcFlD?8ub7;s}D+u)PzF)SUw&7rk3B?+RBY9%N>rcD@>PcwE?7R
zO!UrO-B``4I$``ymaQ8<wmv&yUzbs?+_lf0mnH4in_pInZIaJrvEfoqUNrp_W>~%W
z{s7p4|MK#Be@^@0e!Z*z@5~&M7qE=e*Hopyf4qEj@oV<o>i$}PLN7BPHNfivoSrhE
zV?B6cZ}{z>-JQ;ZEgTV2TDEqe`r#%TbNTXC_ut09U!8w3pZ?vIlO5*j-`{Wl?kE`V
z-v0T0tx6n|Yct&GXEE{+9O~7!fVun{^h6_jD|K#`0~9Qj7z;^d?aFoZ)8HN>dT5bZ
zpk%2xwJNoaz$)RfFy(1p<8MXG>tUKXrEWL*PN6ySekx!3_PNc`vs!wLvi7kT9(>?W
zPxKpRm7u>dICne5P?ta7!3%nF{^*vz<P(!eKU0?GS`_rN29v5qz9Xw0(cqnRR~4*M
zB`Q4yh#2#<wuJS`bAJBi@^$Q~OOJPyrFjZ##{h41Gl^s6-F_!`r;?&qf#8)Ek^Jdu
zT6a(+mmP)im)e(2<85?5^JNbz#m`M$z51}p3G?QdkGz!VAf#xS_@Cbj%q)+{m3?|$
z`Qk3w^Jk2#0L?_ynrx1xN^Fj7p=9(7$V6(#Jjs5ekDMt>N~5D)tnfiB?ehe|7c+jT
zYax<GLt5HK+ohwPPL(w7gwXej_m^-2uqKOAE=2rI4Y;0~Mine9O5su_rfq=t{7~$Z
z16IeNaAYk(KH1uUkbYUb*5G9L?qW9`;z#go6P-tQMUT*!5(;fKW_*XSQ7?Hw{vABJ
zzD60&rHl)lq_ztvHsMtR;44YwohGEXDwv+yM9~4hJ-Pqa{}F}#%F*})<n8gv%fFLU
z->i2Bmx^Duzwz^&rz;8~d3V2Qi4<F29|*xzz4^=^yI)w^9`JdUeAClR?w%Vl>Ye$&
zqO-t_q6izNu9cZ4lCdwJkl^O4kN<w@zWMJB`L5Bf$9$fCh&%G?<lpzsE4Mp2@a@s<
z_P>J-yT9)-H&YjP_BB|~ua#jsM9<m59eUz(^?6Zi9w1n0ch>tLgMbPW&!sxCv~0Qp
zI;wqt_2TX<ljF&E(!p+?hpEVVC%XhJpeLE4x^MPT^%h@{1=>U}n%Db$btQEH)@=Lg
zVWaRnTo#9R*H+xaUpQQ;3dD&5J1fTTJD&a^{f)!k3&O@rPQV&0OglE-u%5ElPsUm&
zZd5Qc6bjV3yI5mkObPquT=puu;tB*vf=whEg(`?+c$C~U3TtV7!*K2duK@3GQxfd&
zo=u5JhQ6>-^w1OX8*GIFpNB>f%I_9I6G&1KX5JZS(cUl?^_Vv5{P*%X^j{x)3nS5J
zI9ML%?g_P|Mw+V@9?H0Pb}kCiPwnlMuT%?rih;p9>}G-+saZ&B%Po8UAi%s)Zw$k?
zT0@Kq0l_O1=Z0y51FE@pUx^PF*97qT*-;tO(dNTu@*;G5If;8|M7e(9S+kQtuB9(y
zRP|C|-L*%~;kdvk0UVy55h3*bZRfX?mWMtA4vbOKS;Y=A9|ms@9ZLw+Nl6aw&ZH~r
z4;`KOnnxDB;bBEhMalXaPFn!3<!f>v-FpnFA_|sZU2@U9$9|zUk=ai6lvm9H+pcsn
zp_*LQ`EvO-_92#Ow1*-?*b?w(91~d8dewcc+AjSVBNxuzb~##Hui@oIHN>lKDh@&%
znstYOph33WK17UsvY9MXu!gTYPt)5@fu`C=>;3X|hN`c&XkWjW3M(8tRpWQn`-6%0
zwpBV->Y~g{8Q!R=sBw5I3Z;Xq!ll)mu8x<d1*>n@AOC%@+uGW?g>2n>(0fR1`)n_6
z!w3=)cs5-8;bv{a4s+q#Nueqfy_!Zv8AGL2h)=XWbNR{K3(S!Ve1{`WaBz&ck?ToY
z(2Bgj9L9v}<pm7*Egt#c(u766Vlj(S_S`{~0JY&Da=5UDc2492&5ZPf!@+9=f<f5Y
z$h-5Ko7uCGu;3pC^iC;yfA{2NDxOMP|5e@3;569!bZ=sNkk3EY*2@Y77DQ|{(V_6W
z<)0KqOcmvUqawCRjS~_GA8!F+wBZbovi6=U7{KHhe7z)}Qb1K&R={;%4?`WB7?i{y
zz%6eI;782O%>gF15<qXb6vCVpp5I@>lMgQ~$EjMN_g1*w3rhkGUR_q|ZVR#TrdnHU
z?eEY0mgxvMbId@yb8_~HoW8GtLf}I1S^^<58im)E8k*yGSC9za02x-quOHFRCQ_6J
z8OASu{jS@voIfa~xo`F=6O*(Hpd}m%BC?+}C*q8FVj7iC_96rvS}I`2J`t29<5-G|
z>0<xr=gUc;FFylIel@@H?SsxXchQ!X!H3p|RA*m4vY!Y2Iu>a-yzc!yd<%mFG8dny
z0SlA!3;{M4B8=e*qrLcwIRt`N5rJ1+*v_hP+|t7Cthq8nQ}b=tF|8YUd-<P8nsWQt
zEw+kw!HoS00W_tU#`|i^4>FhX-(I@8xmYNUZS+qsi=F*~q)Gi<`DC6)esjsNQL^#3
zHuR9#@0D`xMiui`>weQ$3yuC1ktx|=1VE5NihqX})g~w9EiL&D7LeNmvlTX7><caH
z=>^DOh3oxp3_FDXVFJy*5Yx{)B`^?a)dCGXsJtgiTyx{5_P0F)j{P6E1fCC<OSenk
z`+B%_&cQv`JLBRT4(7G7rjY|>X<eV4&Ak0n3to|1m_h8>1wI=Cx_e$5qOdyww_hse
zOk@?CURwp_B;Aos3+QP(IN81{DxGpx6<DsE-fIX|Z*`T*#;`+nT36y;ZT)B9#K$dY
z`rcr1_Qv&_iM6G6cb4etCw(#Auvu?o<FFYCH28!6s?DTWH5VfBLx#$7OO#q_fQRiL
zXJYYowY%Zuo4)rAos<Ybj@TR4L6g0T>`;TZgwI@V0ye(pl7S`H3X^53S%vET5@bHM
z$&vl%0B1-SXvRDgv_%{p0+F^qR~HX_YS5bQcS*=p+H;qO&gHSx&_~<(Wm9)Xwe9O`
zVTl4mhRSgExsGROloh9se1alw(Vq_=8lTPt7t~z;Eu*s`vhu+vY`7)XS2dWb>XV$`
z6)mWjBcR(324EC~fSCla`o*w#f&W9(RR=}&f9)ltrKCY%X%K-0r8^{~k#3~B8!nPd
zcc-M1(jC%`fGl6SL8K%XSmNFHkKfGxcV~8H@7~Wj=RW6&{)IqFRO~_Jqp~Ofnm3m~
zgVH9PFvu@AZW1Y$@l~L3*e++EyEiZx5T99oY4IC2@C~+;dSu><Hdxp5pad#hTR&|(
zhOW+6F+uR7DbP!xJYX~PT87DR%4kMV=;vyh$OU#HIea3|vr2n0EKyN+V(0Ht^t8w3
zlL~3^8H&pv#nVQpNfur7=12j9DCg&l&FrXVO8@<`N=8&`eAJ(8om;=JM<fCu$4UaP
z5xyM(eiHG|e-eRzyWY@WRtXV@_4wUBv`kR~Nl-Zu{i~?Xy|%X@A$ra~nG+JGmhH2Q
z66$++&_TaT7;CnN(TOBg3D2h-GX!F~9el>_;SX2ZKR8jtP<c>0-5dblme=ST;*bCK
zyQ^S9j<*X{!pGwN&)x8k-RygW*Gk?2By-;0uRjtwx%%Jbeopo)@K2Q5j*_@_OxiY=
zdf+0sdh$i?*v!QfUx9utXOsuop%^QJXZy)yC6yh=UiD2S%`Sd_&AuB7rA}X8%O?9?
zw;PwK**`GZKqoN&-QWno_)$vpP|U19&O2yrjP1IkRJ&oKrlzh$If4Y_%4C$3_JVuW
znj4(F>!GWKYPosS*XJB#$L)oF1#m>rsne^G#B{#Y$EMgoQDS08==@zE5=UI~Y~AqD
zvOYb@>(-b!U^r2F5ds~3+<l=q9hdzYKWdWCegD(nPL#fPDgywI=XEGW($p|I$R4H0
zi>>>OF4c%z?}iKxi78a9OFzOtvssh|I*Ohp6C$%<xM)|<|3#gQbnMvtiaF0l^)+Pi
zv<!oxKn>n$5H7U3>TzYq^Z9J8nJ&_t^%J}OAE%naRq?yy?uv!<sb6B+{^b7@d6pKw
zWL>nyg`?>ACg=eeHjH2%DcJ<=W)KE%)c0$*oRM}fxIxsw!qmd`D+{6b&qGKHY;xqM
z_x_rk+oW4`CKW{)mXl4CwieQV+QU77P%8+$|B1;hes{axp^*K*t^U%~b^Y(sJfF+&
z*Tz<;(jWo&@2GRv#&x9AWtF#p?t9b70Zd|E;1w?FR~trwZ=I4*5{u$I&d(8(OL+Z^
zqAJiGE2jrzGingX?sX`kP=@;v{Oe5D#4-g@)8vUG=nIytddzQwoZM_fNs3pC!Oljv
zt4%_|=L?F@e=q5tHkXjK$=CmSi;prA^}u`e<Wo13Y^W^ySpU?B_CU5E;XA939{z7p
z0nRN_Vg~eC{tDETk9}V(QLvaB(TUu{^f2E050Wv9r2$)BY!aq+m1UzT^t-p|oWftT
zP*4aPQ1ccR;4`LEz!mFrn@{1s>x#zOnvv;Af?DKSJf482%ed?2<M}ETU$Vlxg8Rpm
zqR61f74F0rwTYeqLFnpC9eJQ8S?{AdfVxdqc2KTV`D8U9;C^>3RQDFG_m^%N3Ozaj
z0|V4KW40>SIy*djKoZ(~&G7=%bp1}#sQ|H>)qed!R*2GTzI_rbWvABqCs&8G5f7I_
zQ8(P_C{>B{OVe$1Q=jll4!0D?&24Q3n>Vuu(+7TBoN?$NneS_9;p9_W?R#5JfaQ3+
z<13<WxH(13Y2_}-$<Z%^kEVNsoV*Dozvm3z$V|y-FZdPsG^@_1v8MDe<!|dWQH;2K
z$E?=0U2->4Y8P1}@Jhsj87`f)HrcUT;WC>)dm@9U5Pmdn=?{~;n*mh*OiJl;dP}t*
zuRCG>qO?=h0*0da?$&eP)zrc@fh;-bi<^cGI)<((J|ZFpSJXEYglWItUn2iLgj%h$
zx6Jv|7ABpMsWCy>wj)+D)?q&;L_V0uE_45-Zu0Mkg*6${5Y)`=y3QGo-d>5s8nkog
zR*Ci1(I1_3Biz0I=dA!DwfAwQ@ZqXLBH(FZ5y=xhW&1s8eI&MeINje2z{mG=o1fNZ
zf}bV`B<=?`5XW?ZH-}Mbg&CU2a_UfH9t`-<tuF*RcQ^CTT>c)0*<i{`6M9QW)zE)5
z!ofxjUr=KB9rFpI$D*tso-1?qA2MfVVe%(9V(aIWfOCH*iAI?n{;!8=&u7FcyThZ+
z&+0{aX9|o#`j$0k0hLssQrz$2>Ch8(1c8{}`l}v8pH$hxzqrP;QP}GbSuX_cN`PDW
z-rI-Pzk4@h3%3`oz(t7!SvkyqNteOC=UW5UvfBQa(Js~RCIR%?UK;B33QNxPO3X)k
zLtfF};OcUXZ0`xp^^~#qZ8)t87i&qKney<6Nl_^Y?KJQuF=PP`rnK4uzxoFGZr-2^
zwdXtF=8n6kt;cWtCCGvD=GxV(t>Zg_dgSZxH<B0$9dC&GG^YHeK*M!vKCVbJiyscp
zk23|@C`Ph<8(X-jY*uX@J?+4C42XfVqIPK9A5U$Wyq!{OOK#aQx%Rl5QEPb+`i=Kj
z#Q001_#G@z)FVKPvch3QS5-?{z0=?7>nHOuyzSbW{M}EEvRvmN%{HaEOJN*QE(JI!
zNs~V_`Ph1oPfyk6N6_KHyj}0cNQyA%RYlYa%^WX#CeOl~R*OFO8OgLYvr{Cyn9e#9
z3KmN13i%jio=F7vzK`B@#CyE*%j5~l$eG7I^%e;+O9433cR0)~tx*QwsC{kr4iep#
zyY~86(z|h)JCORSe<c*HTo@1W_|+#Ot44&+rdL%K!$mTVlR*wYYKOJ(>EHf-c{63d
zac6Fhy`aDQKB4!>makyozo&V+;EUPDqc^ZS%)+3XtDBbJ7EZFzG=Hem+#QuLG6_dL
zRRaf$ARe&8qCK_Wf=cVd-(U03%)j%x>>%f2PU;Qzg700{1wW(~KJJ-59+^J=UF0@h
zTgz9isY0XI^2pT=$YRXOHK@okl?#k;eZCH&qt?SZ4-inZ<X+mMCfu6sJj^lQ{l|gK
z%Z?Nbr{=SfqFvScC=bmZ9ne6JO(Kj5LCM1S`U1$ZZ#u1vE!v+O{DCTY_OONnITXf6
zZ2VqU7A0a1)yH(Zv=cTpro%@(J|4a!A^LhA_<VfRty^aTz6($*3_M@iI6eBc{4iEQ
zS8%Gf_xJD8$$a6%L+10nOR#^C$L;onu93II^C5R_t&2F~iV@Y{&!;EYUFH)i2?M6`
z1uqcvlDOG)J6VMZ(=GdlLylm>Mj~SyBN1v6R5CwTb!RYCc2hSF7GfrVM=SN6FNgBt
z#kHdqhxuYw4w3YCJVIv*^l-c{@RIzQZZ9o|M)@TIWS;30(*Lx#)9?O#g50Mt@E*~8
zs(7;&AcmCa2AO%Wl4V~_GTCc{t4N_wR{*-&UGdz{$7;aU%ErUPj#{ByICq2O*>2N9
zQ0vMXa*yQQ$?*dH)7n}ki31Li6z(e`1|t;}_`ok6_T08#WGtoOR4v~EiSUoa7?ou6
z4zgpzC6H?Gj9B4veJWGS?<qzyeWVOC7#im7juC4yRlYxesim=K#ZSDQk1N;iW&&&(
zw$FXTRR*oSnA&daogVxCwtyve1oxk!G?8L_oW{y>sFLDaMZMo7t3kS$lLaQ199E|_
z)o;+~_0pH!e+yP~+#ttT5mY-{(jTA&lcOXZ-~6pw%Zvu;iuuEP->iddUFL%muJ{R&
z8^cWbtWIV&@+4bqdIfg&)eV+r8lX8IY2&oR&@ikp$f>(1fHZx5-d$=WI5p|cnXc&6
z<sO;0qnaT~n|DWhIGmc2rvq-sqUavR=$?L#oUQeH-XA}%tgs4b4*mqssa1p;IYzK1
zT4}nWf~`m+MrBv}r#d2qLM=DgJY0;m7f6!znFvkSHlDWydhgEFf^RYb;BHakX{B&2
zf7;02VNDd&*YSSdI>&-hl0@%!mrs7v$`y@RKOV{n+^9gT%F5QhC}H$82;7fb!b7b1
zHb4-+^L4nPF2jd?S4<W>>sp3T3Q2TjrcaS#_OY(I+K&5QE0@4v29zS>0aDbT9tO8z
zRR6VdqTn$o=(o9ScuqjLfkmAkq3;VVhJRmAulN9)SOW8j;=$fNKCRT~lfZN11~RR_
z^8S2;4vC>6&I1uwhl|{2NioG@LC<5lGrdn?bWdnnkM~Chbil(p((2cV=&Tz+OmLgX
zw%<?C0SSAqUWDOE7)C{DkDj#I`bmZq#eRZXC>;B7G>w#m;7~w>_;<U?srf$lnQGMI
zc@??o_rH&ikBGBgZqaQ0n>3{Ar_SWz;$ARKld~8`Ova`huK%Qh-lB5|ixyJH8+YmL
zkq`jf14b?81bt2(yI0_<2j_RO=R*_o!Mz9mkA=^%!1W^VID|>}uzA!u-<TL|i!u;A
z!W1MP(3|VX0?=gLWQ3xl?@9*p<X7BH`1xz}nRGv#%)<%J#1M@Ay9$J%jDn6Sk4KCJ
z&*Eu2gbuHrT?*INHux+IxIhfDx-uXORXtiz^&1?In-S-BDo!vH1F^@qg`oU}EHYeN
z<^9@v3f`g@$#6XWMM{=T(*D5&ES}hSO63+9*LBnuP|Pimk6t6DBjm)YgPs3!sZU!F
zY<09IS~c=<MwcIGXDlz>4&$$Ya&C!AlELiX)+-}qPS#dCSL|p8sx#b0X<A9hbUt_M
zY%w<KNl&L)2r&`$5}N2Ntop?)AGF)}D&&Dk=q=gxc)k9)YRjEw$2YG}`m8KA$z2&n
zKIW4F;Mn>WoqjiM;z`k&DZ>ftLcNhDqg9KK*o9fX6n_`aH({?vxg#JL#Ssd`<^w$8
zxptTWp{@caHEJabw)2$Q8sYk6!z!o2Dv=#GO}dHGHWLyp#4_%%kL*zcC}km|el!x#
z*W|#pfW%X7;luGk;q&(1+T-d=59bf-|4@v9k}3yu9Qs!b&9K?87r&;0ABQq4bn8w{
zx@gW$cjPmegS<zRhF0nGJ^dZbCpBmZg?-T3%Gf}WpFw0`B3Vlh_so984#Wn`_chnw
zWnCnKgvA^OlnCh<8kh5zr~k0O1YQ5$e0|aL1@~2Njq)TTgMM_|kBwv3)po?{Fz5(?
ziMIu*X(Lk^`LrJ&E@yy$VRV64$-$4?L+7rK*RkZy$3*%nC`H%aK|yN)z1`g}Lh1^7
zyV~2kyB~ivpR*PMTk8_{|GLSp@9k%3WF!1+yBU4I%hd#cz>J#3RR@<r!|KX{Q#oj1
zrzY$12f+YfUn2%QJ)E9CtOuwO!T>X}5NOZ+-@UCLi6QHxMCPL$l*C`quG>)v7p0~a
zRzrJx9Jps*PHa6r(I&Gt*Y!Si2+WnkeGb$l9=C#@k7u4zft&dmQ@jy^U_@Y--`P@d
ztBA?T@Odr%PAyw<!p_HXDx%BP(Arw(!XAG=0qn4dW|3CD5@b8=^}Ptk0H(!HLPi<5
z{}YjHU?_v3fYx3T;au|I!QnQtX_^ehhMAhgm`&w~N9Vt9!Cp#xuS2StKbXB9f0kd$
z6`1fghMPOoh~;_spNjNg^FgIbWfkQV>)$j@4n}j(YO1EE$JE^7?tWxWZ+OV0VNI}l
z(?YD-qMdA+!lMX6mkQ|%0DK4X{l(`CVB>V1i0v;8%)7s<bbbEzv?B5vZ<c$nsbSa7
zqO@}4P`2A5_g=Lm4XF8h*J&ba1Y?U_vKzMWpay%{g*^^VG|zCU!q-u!{@*9htw#SD
zGN?pqIm_cB-DSMu+ws|nlh->dD?>(wD$--RFasN?rT%o!8tBPN0cQbaX)TzjB8tiy
zUn+4*nohf4p=f0h&JSJuosS|grsLw`3iI$eTYcO&MObU=gkUK^UL?!PW{@znIIT2w
z{`5Wm%wuri!(FF_6RpKoebTKpe7l>jA?{a0;^7(80K-X|Q&0fQsU(HRYbX|%_Y2LW
zNHRo+GKD5-@<zmq1D110YlVq}%!TLOWcJsH*GDE~>9IB@L0WN>sJ?yJkN}!v>Ib9M
zhWrYc(wprxl{VyY-aD`$Y>BLF`0>GNkb(Xq4o$c0OLp+F(EG_GQn&CVIReNKg_ea*
z>E@;qI-;Kys|X5Bx;WdGm0kX`O5*vmu61QZtx)vgxN$@cDCE*{pcAyZG?5Gsnk2%7
znr)*KC4Es~#vmsCr2MN>{HlBKUG?~^t>EkUGAc49w2#o}Y*Kd3aM>kA25L|2My@!9
zviSG1*n|*0LyD3{Uk;R17J6yT(q?`gYU;G<g7mDyBz@2e4G?8=ZyX!{#u=q2Pe#{@
z*3z7m(bK~Wum$G4I>@<S$>5ktXo7JP5mwC4I=7;s9MG51f?`lH!s_9t_Vwr?u)A5?
zcNA*gool|LWFZUw@QVp<;AW`;raV1;%Bd?5)^IE=K&!TRWV17C$y=L{nBbz5xpSU{
zt)WdPVIbPQ+I4>s*1(BrwGfr%q^;EV({>K_`eP;Kcz=ZCuxGMp#h}@{-$8$=7n@)5
z)agugEo6Fop+bNl_jopOPo>PCI~4@nBRhw6&&vX1c&al!7fY!pv$Kz|qn-KBhRd4+
zHi>nodGYtwV|{WE8iI<6p#~~Fkj>bj_xtb17x(>awlLk<6E-iphpYLz=egXF5Mk0`
zCA>lGe+NFstv<IW>$spi4=ercFK_Zi^c|FyQgn}2NW}oPOZT1Zx|m0fEIrw3e%Y9$
z$S^59A}kp^)PDMPA@C2F^?yv6j)h&J-HmRiAr0_dluoueq+=1;(#*o&mg)`X7Z)Sh
zl-N<_@6pcn$+-~!v{qk1w8q(FZenBelCX(X%=xIMX4pCKJrtyBKs-1luG`_Ibz~D~
zL1zJU&&R-L`U~;vZaxDah~sme#5aue+5PFIJ4@k$DVF*3tXB?YpxF#7YePaJ2?07)
z_V~!umMS8Mh_cmmBLDEN;I(p{8>B9RD;%Az|HIM4J#RAhtb-G%&rgEpr4WnX^xnyh
z((S}Y)L=-f&_QSPkK=qC{LOz)oj(fUj5OjKK1ix56%BZZKIMk!(<J*438RPRJ1U)s
z_Xck?nWP%vDlNODWeSw9jg%+jFx~L58VA$F+-1KKgVtlc_~7`LS6J^+zG4k#;mb7q
zz7EztD=;yZ@VmLPcWZRYNkYG=CGN}4Mp~K*`F%oX;w6SMpeQkjv&hMg^Hf}WBW{=W
z0)+t)3E|la1Eie5Z@Yq?ppqaX6f67F=lJ)-_D+<FA!&0~8dN2GTf)>?5G;BkJE!V;
zIUau7_!M^-tqs5UdJfT6ri?Sa%{&NXiv~G8+cKoNybHxQm347u_q>}Qm>M2d*Utz#
z%aUT>M5kwBaN7AysOjqYZmPOuU5LT{-WiZAP3!Q1pH5AwXya#0Tw5BwtSV-U=YK~_
z;Ntvg-nph(;_;6^e=b7&YI<odskXq#=)~KcH=9JxEih2@bt3<SwtjY}4{q`7P@}^E
z8Juo4;A+c;(u7KdC8*QD#8k;u+b&`Ma&V=AGq?PErIN$Rz0RAD->|RE1oi3`W&HtB
zTA@WB<Epe%L{T#mA0F6JcX-X{+<MY-dMehof%N}d<N$BTEFy~4jgQLb8>r&qdAWOU
zyi2!`mgzf>J`zI0-M!y;Y=e#PhQp-3r;7;-zjU7EcYen^dlx?BD7tb&7B<&bn@p%t
zftu>FNSu?hIY5B)F>ia?oh_WU*#joH3wkg5*QBH}e7Sj_ClcfnY4d5OHtF9@NupTJ
z>lWiZp(gVi$TE0rrwvK{Tm2zsN}R+b>{6p`aNxm#uAyg`dYcqS1s5p|np^)#&1-7V
z{^bvX78!x_!c1ZmBg9Z=??eu3W}$j=Vlo%ai)XP3#<Fjuec2!-+b11ETmc*xIy93d
zPhV5gYAD%hhGHh7oAdgsMaz8}zM5wKIYS$Kb8;3K(7Z%qWe2_Li;E=1wynvy{?ObE
ze`>5y14(HLiirn5E-ab=dkagRu-{(J-Yc~&FczY8v}36MSy}_Jw1rMgTz#^Z39j*m
zp+gAxtn9u=zAO&5?)UjEJz!hIH~B4Xd(f6Aax=db*5ZWlc)gPO#Tgl@U6NMbZJ-kF
zcNV2OPNAi@<4-3D!9la><sp)i+RvM0Wh#n}o1;=BX8h=FU_H$dAKvH3^%Go9Vi5!O
zJ^LAOQlio)YtR(atrq=-7t`ORIo(eX6%_|0G@LcJuB}M<S^0ZB82HSfK!W`x*>dr2
zBKLY|YJ2<se(8E@C$_2uE~if#hFxTlL8!SHa1aP33L%V;{K`ZqCG`?&nRmwTgk~jF
z>!`16)M!{MP$=#n$Sc9_sBLR#If}30FlRtrrbt9PsKSmzYC!D}r?^B6KCG;+exHQ7
z303^h7B^<r_Q!{4GeEfykA06sT1uAg2KUXjDc7M8jRAU0zqLr^N+%X*?kh68FJfjq
z`FV86txYJKnfaH&jCJFXMKTiN7CST=MT0L`8)h(%5Nf^znF0tE^3==VMyKhSnL@Du
zBrW>$@yL$b3heUk9gBcO&TlWL8cMvAU-$_Li6UniRVwe1>uINq6>lD)6fBiO*iqFk
z;wz@%O{cMkpWCZY!!!Sb`?M~YK>=4x@qbSIm*<k<Mk_C)N>02Dwb860!?bi$%Ppit
zppJHZj<QhP_nuZ(x#e$TP}A8zzOdBIWogauT1*X<Cn)^S;@6ET5oJvPHqD)_`0FeS
z9*UkjTvNck8OU7|>ojPYzf;7r3l3B+{{>riFB;<j8&hr6<em)tEL!FAfn`Lf)-EOE
zSwKg>$hfN16vhASMyz*U<A<BI9?^*f-wvD)w3<rzA>uK2wlan93IngsX9a?-c(ZR#
zPF?4)%}mvuM+)waH?~<ZZ=C_v8W^b0PR&UgqvjJa0IrcmkA5c^qnPBu`OhhSdRJO7
zOJsl9Aj2vph*3?JzK{JxDlAMwInwsfbyZxn_uGS<HS+de_~j-Bkk@2cp23m|Syy6C
zGCK&}efjp4g^Q1Seq9Isy{|pts^fYtJVVCpe)@#=`6V%u?uTl>b``Yf6|R?Ig;$-<
zuvvcgKKNoQiZ1AC_VDm<#ih$HIJn>?ahHb@yQZ*v!%|U_9J>?g9R(CxVnbF&V)rs8
z1U!>{sh_DDgh~laS6$|Mbn-tBMIEo@QAoPS>qv7{bYX9bmtw=MH?tJEnX(X%=zUnl
zSh!k}X}iT597Iy6WE#+RWwshDMX<KE28j;4WpoQo_MCNXtd1W}&Z}%9Uza)VrJvG7
z47f~^Fe(y_>0;dvZ)%d?h5hZ?@Ore@mGE~*YH~cf^^AeHnMWrr&dbvXHa3Ou=r2Uk
z>UL%NxL7>6jJBUh^@&FD4V7f$A{*3|u#<`!_I|PUm^cOBk5#13wl#QbW@u}NV0rAK
zvpFgdV3wq<`E#)2I#jbT6oy*N8RYh})kqDJ$w8CDCyM_~yT$d{dLn(_Q<!dPHB-{m
zAt}-fnxjBUmd1B(`MTd!_v>hq>n~YVht2RbNru~t-tPq=^GZh0O7?NdM-Ebh%p!Yy
zwBM^EN(CmHHZo0mdi6L#;ptcFZ>*xi&W9SS&L00@*7e+DV3miL%sTR4W@eIK-dWZ!
z=NA^_6|6f2W)_)Iv^gJyabuQceL+S0{hBWSrrQvx(M9U~h3ldvtBSv&|9YYBKlmd%
zPJX@p?)Z0m%ZY)*u`SO<4~xYo&~i$p#c~88qNm&EY~6kyI_k@7mQXbKX`J<R2*LHW
z#+m=*9Z}+UJF3ritt%?!pXu^_<A}2Cy7ln(Z8|{TI=9S3De<eeHrF5P9DQ8f12K&|
zW7VpFS+Suoiah2fK@4NE<tr~%JmcQ2-)PiDo1a}ik;;Ln%P?WztO8QM(UJ_M9<k<L
zUsu&l!#elvUu%ZuzU;bOpDL+9Cb8%SA3iC{%~C7V3a+iQ&fFD36^s`JbQf7!M`mb!
zpu0x88DU-P2V8>i_Pe-^eilkG=&VE?ug5fCHN{Gw)z1;WE+m;o$YE&7{>!_w+p24K
zg6s3uoxu8&a9P@qBnHJe==l%uw@9?b>jvGh)5gLELOB0PqH{zYXy@FwaVaf~<C0xY
z)~<%2s0(g_-%wqhJszlTNR}+r`|a|wvBOUMb_^2<zMLbFhC{ilOw~2aQ>kOhgfajY
zrMp{Dp@n7$+b;np6YiS#@YCMg4P-jUMz_bo{>9(t2Z^U_`1$SaH?OHcVL1}X=;Mo<
z*fdq**kWlk@$gySKV6s^TdsfdK3h7}w6r)Gs#wN>H4VOtmp5>3g;dk7!ep7OPV|PE
z3T<+p%1BWpvq~(j2>5V_i=-{g$<P5Q)wm@tEOghp)NyD2rTN!VB${?OW^v+nbL~Ag
z#X%iqq+kX~^Y~dDB4Spj<_2NcalNR%;D*?5n`TO5$OP!E^(Eu94}?S5=sW@1yqX6C
zL)RDQd)F0eT<eI#5#89m+7oZDoo+RT(k9Iu1r+*N+s`q@lgEy7=qVQ?A2a~6q8J5|
z`VZ_JBqbmg6o(j+)s_=FiOx2kgXvjMULI?8ZT83>yLi0>6qGoACM+mS4|$G3q0vHS
z@&`*Id|H~NB-rUK3g92~JRDAO?4j4-!3SsCkEH=+pQeRNkC*P4SsIpB>13cw`Yn}8
zva3T}gGlf9MK7hT8bOiP0Q2_l?XZ<GlMHW)8DjGlVF&86CI$46B}Z7xiFKp-{<&`&
zlXL5}rMR~fA6sngikSb^H_nNjl1BMN@wYMC)A@~i+xxcSC(E^G2oiJAoc4m$J^%T`
zj@ji%LJQ@>zze&~kD;i{0Mb$WP4#)bu2Q_K#2cgzw_v6scZ%gm|9+jaM2?M_Ge&9y
zuCh2dx*m|=7+ZXTz`-T+y*d8-=g*&gYoT22^{7hDk|U3|Z(=M23kt-1MLt}@xF13x
z|6Li*?NvrDITAD5YuP$6;ZTrl7B}IL+P<J^D$ZiMUpHtpsQXaXp}Wqh2nisFxZh1p
z6dG<__g5jcn!igF_rJP2*Iu0PM|-{?Sijpd_20_uxi9|m7UfO7L@y^AC<N}-ZlrIx
zD>A*@Q0w3ji*yX6a!AI?I&9C5hsBunX3B-3N)Q*&ebP=hM0AUUHfIlxMcpomW@dz;
zf^pJyO_3Y{d5PfXOeEvzHNa2CFn33ocmE`^@f`nrEP#yKa%tx7xkUzWJ<S6by^p)B
zbQsYqYaL5Yuc8h;`SVp1ts5KXFVB6ENX5LXC~_v$Fv1hIRUuJNEABbFq)rd~wQO6e
zp?x(?3TrX8c^k?@N0ZhC-#KrqJ?f(Fe60gSOB1x42don@PjKD<4WB{^8<*OwJP*d4
zhW|9rMWI)Oy*77FfCDm-u9|U>d%|}SgcdOnX7r4~oD>xlg|8Usat2?E^Sqf#CD{qd
zP)85d15ya>Ni>)FRvlZOx7O$YIHm-WYDISyiv39^4wuplTdIqIffhtKOD0|F64e1M
z%oVbZ4_WQ-H6sI(8PVhhnXo3~Jz}S<7k&1A;6V6Qx>ua!OEf5>hDFGU!+2ZRl+q)(
z460^rWlNj*+pW&mMaVCR%j28NB{3B8G%XMF2=YD9_4KsJU;*TxdCUm>iLb3#8gjB_
zl9n9PUMvNjVT4(T7drB<7~Y(_3U;aNmxlfLqtouw=_U>|Ifh_C%3GXb`&C&|jf5L>
zW(_)*sy80SssF!uSJm|>mC{NCvqc}BOtTs58D|JZeZ>Rc)>a)YXU#4Dd<@{`B#RyM
z>@hOB1mmI-X>E^LYLZ1{PmLEVL+qW@5@wc>+rq+6+OzMcd_@~zR9ZtMX`0xYXPV~9
zaWpA5v%nJC8<$<my`@*r(J}v9z%T5*sp}&7?AdvLJ}+O_{j^+4h2vaTM-C~k>>IKW
zqrp#`llz*rwQud`>ABisyx1}8o5Gz%!I;$JaTZKI-r95j29|OrFjaeFt~+g#HN_%O
z^+pp$fm<550fMuS+o~g-DU=xFAU8-BV~|o$-A?GMPG>%fza++Z-S2s-o4A_a{?QG-
zn<6)DLrlZhSDeE@4`XnNqoC)i-iIP-m@>BB)9QY$)AF9X-#dw?Kdf|kn53XIZ7Zkb
z*%m@+!Q(AnfCU;(81e!{BCSEv<3m|@IkpFmYOy|}k<7X<xew_$_7?f9j2w1=+tnU7
z3G^2J@TB^E&mTRIWlJ-3P#mO)NzL;-Ng<t1oGN~~C1PNg-ZaCH{##8vyQ0!ERVFIv
z%Gi)XM(K{kTp~mAOl;)pPouoV?QEjz^i0Bx8e-FCWCXZ$d9FG4yhudS>}29({}^F`
z(*{szlutfGY35IUaFn&N1~LMn3XN6r^XC#z^1am3x7RKY%V7cVr>Nd%c<<$j%kvZ6
zL+sw)N~*Mww&m4of!LXXq-OqGk}kEp;1lXzw}S^~?U#fR&?J;(f}(7;GD%5rdrM~M
z7oK5-_7eFrk^{2e0lh28s+sMPKv0ijT00JWX@EbZGhWc052SGa;Rkra{ADd`G@|Hp
z2CA?4MO`<3AH=^V4o;?Er%5HBmHmdtAtzasMu;tmpFt{1{{l<n2X1{0wQPf2Sz3=l
zP4La`5<K|ov{T~laK}bh1*^!>n;=~@S(76`fyV<14#&bQUm_PSUg@>&?H&artj;RY
zQ@%V<w0k`+Z0}#(`vdU)pfExv2-tzRAAMU(xnDX>&(XXYb&Dj(5<etCADxN$h6@TJ
zEE`?NJfI2B>P;tCBZ5wXhm!<RIT)Cj6&|cU)Psc4>oZ_S2JHXDSKB>WR!%@>@2H#w
z5D3<@&db-nZ%#U`o~*ynfh_5ID9YG87q8jBXxcX$<mct(+2!Z&{imjSXe)k}b923D
z0{IX>Qc~$#!WMHD^xT<yJ>|$BE0s91GZBlsc%AC@e%<*jkBpvkvSjSP3=DI?6l{SE
zV?IxUcJvhD%O}!F6wFv(F0m({BA%9pQcFtiiYQR~Y~LAh)_%_9iAI*&5eY;oXVi52
z?CtGNqC~SCgi3Y)a(;$Z!r3$Lzd1e%-mqCcYCf6-t=*El=*c8~gGGOG8K{6jY`NEh
zp2YpGOv&z>x?az%KxO}LpVq2b8!L~BA?0jS{z3Y--+bPbbT|Lk+PURwMDY*o>~mej
z5wrnDG@wT`7?|h}z>*+9C`o*XrmOp>hijLhoi%vy{q4ER(PqGz$zl-Pu=nmIcD(C&
z{_o3UT=Uwa{q#(e2H5_Sfctoghvr|MmK;F5EX~hRXjFsUeR}WLr<!-{c8K*1xetHg
zEf!Fv<o6j%R)(ky<Ae`R?7ddAtC3a*r{amB7D_2MDc>kQ)4tdMq!nA|mZq#zBjo&S
zvfrb?mYG()U-Oxom<IjZQiB~e<NEz02T7*>?NH5f(!j$ETk<qm^jh;;ktcY*ZtWF&
zn3s61#?pI*OeKs*hC|i`Jstr!Hoy(y`MB_H=9|xPrQtA!(}S^w88;FPb6F4>#@1kR
zJZ!3*EE2d%42A<D8&A^$&zCMwi_b&A-PFc|igEVy9`IPWxH9GYN%jU*B*{#!xs6`k
z^t2;!=R@UW9@BeX(+lB<nnrqr-~Y<AT51>fON=9y(=3)gK&(=vvOg=GSm`MI>6Q=|
zKkZP(r!6#XfK41=0Epq6*!0J{!iT$&?Jt&gH4FZxfA9&cf}d+dL&omXvZJ7gOx;*i
zMgWN`ogo|KGaPXy+TL*($h#L}3w+-_%1%n-4M1ffqGwWV&5-?<x1Q=pm&6eTA{|st
z_<~-1K`H+y)YMUjTwqbRYVX_<PwDHdRGXT{(65ocP-zlPPT30dp8*F(>tRf)Z9e<|
zqF*t@m?<9zL8qgCz^6x;LZb4~vDkk{0=!~$J=GWbjG=H*T3wZSa6d8}mVh>|)oBFV
zx^c|Edn~WU(4{x7Y`^F5hI8MiqSm2CTZ7~C#jX><1+x2aYEa@~z)v@PjCB_~3616*
z<G#+}OYJUXnk+l&xO6@JwNZ0{9liHZ!YkSObN|;uZ|>YK8$D+3W;UL6XUe|@AZd2$
zTseuEV^i@m&RmWgW~{kBJhW3fY#brrOR%iL1{dVV*VS_EU)~|%?Qn{(C#d$RFMOeG
z1~sRvLWVUu9iz1^{_RFoS`6d1?~p7x<AXSeOZWA{iN8v?m5_oW;pl%TQaE5TId)X`
zvY%$_-n73QTh=h?g0F@h{6+JD^f0$=+A&)y`pTA-W^vM?$$&a1`9B@b4$57D_obYr
zKZX$nSQGv4-nO^6^=LDF2!q$VI3)B`|6*g3LrDQK)<bTa>$>@KReBNABj7$Y@#>F{
z>jJ<|`tu)Ger%DmxLH}*ZyqeD?gf2e5SGCZ968z7qnZ5*A&^_g|4QBgEY?o0ba<J?
zn$Ym!KaQBC3gFQq4x@iqqx3p<<ay03!=rr=^s;n*j^S@AWPV?ziLJ{t4p;u^{W3s-
zLX;eKjOx-*lZt-?Z4h8u>7=w8>;7Tjg|8cPI@gj5s1=AI?~s?V{C@F|(k3R@vS;na
z`*;Er;#|TOuOKfEM$ToxmMr*8UV`c8;6vE+zk%Mn)+e2Rj}p%n&xrY0+|hcikd!Pr
zk}nME$TDm}Vr@Qs>;jto@Fj%Z>nz{!c5iBXc1Ha15L4nVs`pZ)_jbYbVZ5sD@n6(g
zDPjC<sRxeI_<!E0IvKK>bLXZ{TfKqT5xN_}J2lt-`Xm{kl#p5A1q0g<$j&S1g;Hw6
zf-6=i<i}jKAp1T#QS;yYDvYo-=S;bD(+0NVfF+sq-sX0U|2m2K4yr2#$oXboLu3SE
z{=m{sT4ZUI#ArP&v+1%r0~we~nwrZ$ePej_14e}v?w%p|!*zRdTa6R(L-M5=)KkOW
zDXed&B6H(;D_|oCfUgyb1rtY%l6-S27~}hBO$iMtE{X>?`!t}19I!t8^FYKt+`G~t
zfh4+5S%i{bd0dr_{C(=%g5AA@s9r{Ce`WFln=RJ!ier#j6p8!VzK96(0VBz)^wWm|
z&~5^Dxls|)qO8*YxBo}1UyJb=eolkfHgQda@_bIG^nL;~f5l@V>&Td^zpyDMOs&u)
zf1`cn#sR%Atj6rLQPz@|vUYJrRB1dtjTudi0u}8?&G63`R<V~NaQ;rGZhgqaP~#g1
z6|CTPqsaq#6)n1%{bIjXhss994S31epCMh^+a8G={OXL~i602cB@nw>W^F#)!+%cK
zBmj4}h&(P$tgyLSfr;9)v(?Udk>0iUDr^5JX<ko7Efk%IIMmeNtUn^Gtir0aN|SDX
zPGr(0A-6lUUAL$xu6WSZ#S=^6^Pea<gcx+oE$o$Fl6nH3e_pW}8FTj_XE$Vk>msld
z(R*>b^nTM!Tmt^pw2oVCW~Nd~5yP;xU~(i;{K-=sw$Mj7*ua@NQ3hMq`$+SuLfRp`
zD5-lR^0$3#wA3pKB`Rh@#?Q*?`j6X>9$#+P&cuN$qTq-5qo0FMbJPbq*tZHrL=b6o
zbvqa86k^t@%NltomLis-UAhbFK2y@`DFq9qq?)Z4pwY@8u}xW!l15B2ibaf)9RnUD
z1ePL;e}h+=jD)lJS;|M*VtDG?HeVg5njv}`Vb&uO3`*>$6EAV-2RA7OZ6lzPX4P)f
z2w?ErAOHHcu6C0;fyD{Al@{?x3e64pv>#|Je4-15xnvh4Yk3QS{-Hrr`Q=E>T%^eD
zT>E^r@jN!QurRczWjz}9hp^wdRZB@*JKbWCW+%HU&^rNrR9Zon87sg;KFL_?`JC?A
zO`8O%m3|bd>%B=Q2)^iP)}fwUmXt<+0nm4;Su<m&q-h{kjZgON6zS>G8YyvHX<U*a
zA@3!R!Z;?FuA1wfd0B}{NyOwWoJi~lzvbSa=68A-pDkUP(ssAMy`T9aaQ7OaCRn{{
zafPt07$|prc5pkB#F9-ylhRbG^`UelV=>bS*U6o;PbH2?(kXcLQMQOF<%I9E62N}(
z6GM~OM!!^3ZcOoSZLf)o&I(l%dn^hY<G0G<>hVf8C58of(EPhj<JT4;SW0$t2C^1P
zLjEG$ig+o{YtDAzgdX9XxJe<K1*6rZMK_Z~^)zb3Bl2Mleao&Ze~Z=2m<eWC3#HGD
z@xavocx03(Q>2mK;`Gk02KS><4Wm!9jgFO2({iT?p)nSiSvAjMuD^iUiY_DUes_X%
zBM0lEHuE$El?sJ06FU|8y!u`J+x0CF7AOn=ZKWZxX3TUJO)RR}C{K6y<Dj;7{k3jH
zC{N}G6x&2sFR#6BOtZ7cH4z^!u7dphf~0;a{lPyK^E+vp+S*QrGY1EmYPt+o3fS5V
z(i<b^jx9!FC;9R;PCg&`ehyiDjVD4W`(qg#{%ySR(q@K>b;m9K7x23?ZQGaM)P+p$
zz2l0vIvH5d!Ban@_RLX#cdwTdm%&n<bFRghSX5X%NBQn}`fkd--eLQE_P;yt_Kt5#
zGN4+AX0LSthcle2aKxP_ahQzkfl+xvTaQZgQ3uysl?tmBn7*$vPV_H&Lf*6Q@3C-F
zRgn*k=0+KOJ>Xa@5P&1prW05zUh(nZ%uDU5K(Qdh$Tkk4SHw?1g>D8ZK)}VNmho0(
zbFwihN(|L$ieRdmp#P%0O4~9`k^M=T*n9SsMvR@Y%TdSoBV*!d$o=L~$IUt>T}SQZ
zGzxOW6_$|b^?80eJY*-zF1BBpHh85aZT=aDND;DkZ1GvkX}lR6e&P(jI-H-#m0SDG
z&P*4oS$SFe{b;)V&!dTJRu<HM@fDHB-vE>{5Pf)&+s026$wl~{tziDlN>waLu`6Eg
zRN>QD;UFRBDeIim%6@z%ZaQ+Hz70d>>moIn5>EqwYoMpJK}XL*x$Ama8q<6!O5>%X
zB|e0`N+RgukMBGG#|Q*PD5>=ZioK#D*87S3zZgzS#~I#&)<u5?!))gmzCk-$aKUjS
zJC2bL&y9l;Gv%o(+)p$nmvWo59zF%6X#EUJNfAj&=<dg_z8iTmslE3FHe5VVr>pM3
zI^nM>vX0oxk7HVFCMQhFaUrEe415fBgmyvy!KP|YUa4#2JHA&Rrk&ZJKerU5vA4=8
zH?GkU5D@U1w|qc`fVH}Swgw!w4q$v^=9Q8NeLlqmX<6AK$7{xDt%FDRPJKs#FHvBO
z-^=(I1Bd_J{s*h^sXq}0A_UpSD`~EgugN|^y-TO=_T_~3W$xd?co<`(v3_0~Z;U7?
zWoqlcRC0I+^ZJA6a<}=qm?>0G1tNsc4uH{1^+=G_%kO~nv#7YZN!?=W==Wyt)|G++
zS}T?hO6+K>lyw3h@;p+<Y+Kh<D_%+SL>zXUJ{)_#qY8RxBskjuMAjUAu4}857Ft!h
ztIqQBOvC~VURDxoq-&c``ih!xW<;Bd=KBQsobS3>w0Cq&3RXKD{VY;ue>swx<=d!l
z5HWwQ9!gC@B`MH}WChIH>b)h;bTUS)r|<onEs-ZM5#$~Aof#%u@t2ofC`vNxSZB^?
zY!u;YDrb8ysqqO5u;<SeI!nfS{{mkr`UN!D#~&_-4rBVDrBAjlLf7xAUN)$${j##*
zXwCLhIn%Pv%2aJuT%ow%yX8qm?Qn7=Y+g7Ol{-~=n1dPzlbf&AiLGi~_uV?LAV@rr
zS-hs4b<SZ4=b?(vn4muB7_(jWdCkYjp{n_xvK)r_J7^pZfDV<_BMxuf2z>kEek>}n
zI10Fwr+YYg-e2Sxh{KnRCPWS2e4pLoggIc2!}U6}OQ@{j2N)eR8>5MpoH|{870q{_
z3lfBU9q+U&BV~v{F6#e1?p<Fu#>U1@OpGK-cyxQLLp)*>4O1Iskqu}HQkIp*K%}lz
z|8Hq&X-x!xM-I%6jh&nEd{P>s$h$ONw0>WXJoA8uN4~{hgg9~|BnP3zm4<B~uX;?k
z6Nd7f$y5?r;6bzI#8V6{R2f!s36Cp64ceKI^F$6|tAH+q*EJ>u#6BpAIo~vSHaWjT
zQ6heP#fPO@iXs_;O*&sQLW70Mp#-t3DKHV0LUaQc0l-=7^Ew@JxRZXTvVs}{wNRxO
zs!_+LCw^DkT$Bb|If=a6^)*t*p%0JIB#~o@{B`;t+5J-EjEV2#o6M~&R8oM+^D^Qh
z$o8-haQx5!oPuxLn<t;l%*ZnG<e3MhDO<+czoCbuI1v$|xUVIGnbJ0qbcG0PVm4{8
z;wZR0>E#rkD?WWv;J-~peNYVhZ<St$&gp<z9r7VtUdnjZ16G^HF-vvraN!Y*G^_S?
zXMq(YJGbV@pkV$~td3hG1i0>tl8k-mkCxo~s>_na@q*23G5F%o*ZO;3aqc);H5N-{
zjhD(l+1(cF7Gu4>zMR^>EGZfMb@OQ7?C<TSzt~Lhg@~|^Y{*gAYC5!1Y(|@C>Y(wb
zr+c6%9X>`??va0ISy)sV%{J!Kg>+%+Tcqv(hEIi!Pc3;>WhEhf>;zA?pCAp)E6Csf
zZ(h|=4MZ5705mcpKT}{_J_rsEgiwB7?m*(7Y0^{)7z5{<S67)#IjHo&ot*}WyuAF1
zG2eHLMS;Vs={0$^0#cmBgZT!|=zphUheeev^GnXU0xQj>o1Mttl1d^2jriVjwD)&^
znUFG~amlgXG5zgIEj4L6^a`A&y|9>DyF`M~{Mw0MWkskaeU^-gVS8%ILXf68z7u#p
zou8N)Z}@g3oa?rDUH$`TIkGR&)+U`qTGg?9PV;cm&j=4CerKmJYp^C@D0{s8p{Tdb
zcljF0P4)lEc5KOev3_@>-?}o%djEHcF&?>-Qb_5jZJb?j<bS1@C~la@DzpC$OIl&l
zLfK8kNE<x~3EwG=W+kC8hXx$QXY%Jx%}-3(vz~P!##*2L8WKz9@i6W!ye6u5%0)ub
zu@!0!Kk}iHUtH&6K=nS|FsfA77)W@(PP@8w+CWzS3zwLN?>@+aTPW&(dsq^+C?Jp-
znv1H2l9xG`i^>XAy}d0q%fcePTz84#Ye*cWSE`~Yf&D#fjLjXopnI5>c<A0hh?T`q
z7_GD@jP|jkpz`*6x5|8Fb5kb)3uf#Z)96QFCBI@wiu}Trc#!I(M4mZw+Vfy9)Rf1@
zat>xoq99RNYH@<a`xuWrSm8b%kUr1I2ml5B*Q@m|UPJ&o+ytAlLH<9Hq7&UPBOdk&
zj(w$>Jc}4{Gn{oPoNpkK6gGx_bauIzXa-!42&RyJD9*d@TdpLjMy6{Iw~qu|+_j*8
z$H$XtVHEQg^<BRHUz6hMKODuelgdE~;;N+HQ{ZJxSkW#IkPASfG70>KQu8{vprJ7r
z<w*abAN0T%`VSc$N6r~<thr(>=6_F|_+Rb1X=Gd@oA%f|-SeMEOgL@V`91Y`){upU
zfC^m+NA(^`WKwu%&%x$BMa#;ycO!B%_}8uZjyB$OnEaqZe!<t%)15Qr_m`Hmx?#i2
z{oc)At`tr97-6EF+!Gi3`@>#qmi*L-BVUg$`SyX*+}SqAm3`PXqAb&-t*grunaPoV
z>8bG3_MLOiFqwfJXN)v@#XLKRlT5dB&6_{J_*03LRov4#;%p(PbIB36M0n${A&Ym>
zOQp06Ue2kes`~Wkv$5o~IB)KEHhoyc6SPvg$y%}%C@d9HvV<<AtqrMmbYJrolr>&m
zT?Lv{j7hB9+r%X#qGxZCkz<ttL6i)67>Fc7&T{ix)9W}^`ZOji^J5uZANk7?!Z}M`
zeHU-m18xdA`efO^NyvJVquQh+XO~%(d*|6;P*dy!6{Ly}(Td?Jqc7H!F1-Ew_u|j3
zWBXFV2w$J8*ByH$h!?BayDE37jUtg%0GP=r(Rg1tr899vNQs`NIe$n(lsQ4lu3zgs
zt)`mq1O3+A+`<_{jID^$@B=G4EyKbB-LZt?W%#)+a6L6d_oPp~`AAT^;eS27L_~BP
zPwRo!Y{Z3Al7G3g#bdLzt4u~lwyw)Ygbm5C`kuBw$S%igV7K?#71D*6Mw~t00wYxm
z_AAGlTz}|GKNY}+#I7^a=0HUkEoh|28jhe4(QFxMItDN{PKrCYv>c|Sh3|J0gav^N
zfKemmG*>`SkX@P7;GOu#W2qaIrPh|WGYD!xT(?U*=|qf*2tB4D@GgB(BJfNsG#mmB
zWqJ9}FFh*EUsU3&G#+>`sc8(bjz|j=ytvdV$B-&NQ2+P!2Mkjpcj{?CMB-tp_a+yB
zpFR0J!;xL4aU`)Mu~?5%E_@yHi~-BOW>9WlFGELGoqam>V-2oTmXI7I;)HCXf$FSf
z#a?uWO|egsOt9XkmaTHg!VP|Ee72Lxc_}f|aVPX=pL^E)wA+OVw-yNZP@Fk6@dFo$
znI^&KHN<6^$JQS;HE8b@LS-3Zu+d3I9Dy8dgAgWA<R|mEAC5jN$mzbWu1?<!nG%jU
zV7v(LNr)D@__>J2$LR2m3i)Cup{;)eANqR!WOVkJ2?t)gwq&44Ya(r$74vvGNV5YD
zd{ks)4VqOtKeDH5li!U`aFfG^3AwnqKZI@7uGhk<)Wjb>Goy-7>40<Pq}I+UN8#l4
zMjy7~PvsSV4}8GS-(}E>nh~F(%IzGHA#MvRE<NpfT#H9u-D~!m$i84txB5lt`?+s}
zLES-Pc1V+x2u*w%tkYPGCcbI3_Gl)O)#q#tIh3tB>Vk*2TrWz4=iaGcLC$)FZq2oo
zSyIVqMcbB7F^m?x2yJ9roIH+=l-R?+#p@dxzsQvOib?_m1(AmJdITG|=r)eA<LFiK
ztMC=h)YaBmTtNxL=+r(e7K$_En{)-39=W$0rAMex(AU4kOUao@<j+01J6yZ|`!^!M
z2R5;=x6rs#!sigzu#6DQQ;pO{zK_L4?kqRQ4o$6HHct6}#&d36Az^%FSKNbK%n^B@
zgAGZHO<MscDEm$Kk@0~vJu@BPUT?rXv2|EdK|V22;nIE6pC90(TSfY>$;H9x)4wkY
z2Y8wXMp#H~Zv4nRG7|S}qS!LhC@ta@>wg0|eb5*}tR?$+#qc>x&4=;2;JuB<L$w!M
zGd=Jp;JT)jqeT5CJ7@BH?a_xew(-pp_!p?&O%0&J&y2!gg;z?F{rQ-1)~~l_rNt@u
zUkTmg@w}SIBCho{Zq#amwTTij*)P-I8fO6}$p;1#-PPO#zKpWO;{pW)Jx3z4lhGPQ
zpHgV3Gt)G~v%;~l1tC~`AC>z1`vV+7C<T`m9dCQuOL|{aLpn)Ad01<%uIFtCdM;YH
zqI!TOB%N5%whnD5tXA-KzkL>EniBg|qIJ;O@2%;TUg0!Rp4ff^b_Mr9VYFoQF(LTB
ztpNZb|9sOLe1u%mNpxS!16T6T_ok0hCpk>fsT6bT^3gQXrbP+H9?!>~{4~<X$#L{|
zSt(ddA0(yd3Ab-|Q`O`L>>zm`%+6>?KL9C;keKviM$0(ths#JG)4!5!YvRG;1)@nB
z7u3+!1C5*g;Se^qP)XSkJoh&h%j6nl2}}8LjX~1W$m675iUbwiAZ9R=U1@H`6Rn>S
z3J&TYl37T#43MCT?xM<khMdNK{cLN^PHQb7OD7ar_tCl;#dt7!nHPWah|IknQ2sEm
z4n^Et404W5)g}uue>@?t=?<j_Bp}ctTsPfCBv;#c*<X-`mhSb2kXTI&s$jHh+*i9C
z8*=hQ<fWVznoq11HhZo-`6^me4NRYF>b(W2I(k-4_>UIs8%-pFZamw&{BHlv-&yj$
ztx|FzfqbBLYJ!*|JWkJnCw1WNx-gH1J9A=wKGnZ9UuF2GpQ0UqzMGPq#xC1^PzE*o
zCl<(WmBps%5hQbKv2)h4tHx*CmcfW~v9selMoEX={q%K^KtWt?i@syadV53BtZ$kV
zqs@Hu*`uSdPIxyiYsJCQo+kLpxPnkMF){Rrw#!#w_`hrCM*R&j38YAf1&NqjEqOWN
zJ*_`>*4EjSAP<y{oK-56UZ4i>Z+8vS)hp0RZ@%BY;6)b~6KdD5)y?I<ds}w1e)pXL
z&BDjYVZ3;~CiN_5itE<Z-@Re2T2*x)bpG%V2jwml7s^r7^7J;#9`ci7*q4QoSfvLH
zcdRk1uBaG)LR%bh2Y4>U^Edq53<W9<DmbIDO?}VHelhPV9Cvs-R3l$*<utCQ)3wv}
zG-Du<9nKjOq#&-$Dg^xO1ziBS#rE`&l+e?z4V_<WkSkmk4QvPQH|laHgzWdgZnH2L
zJKH*rslHQZ7yfrw`s<rsAMJCfPAO@6#7U>>o3}!BGRc+_fd7A;C34dmi6IJ38{P1A
z#Pb6jct#)r0r$@Do&4CAwPZ$xIse;g>$OychQ;uP{fOF5LYbcNi0c*XT8Ua^5QQ_f
z&qmVJavC^Pd8z&%O=lg|<p2Kvfy6|V5z;Y`1}SNf5|A#XQ@W&kz)h*6yBQ$)CI(1K
zjF1onhKPWGbjL{P`t9>O=exgmcFw(X->>_+uIKf9Jo|L6L99?kEw29VZ@keTaKf|6
z1%58Gr;Ws1-ZYQiul*c%&{L5~IWeC=T>)p)8N2U&p0I(03bCBt*v(J2I94^+U@K(E
z;Zl!%!_(w?-o%82_%9|dF2rPk2l9$Mr)PBTw58>#N6rn5Zy(q%y!W%b-Z2!}zV_GZ
z!n&;*PWT4Cc32p_0mtqllqn7@JG}4Gul>^$Utw(t%5o7Dy4IpG{BaiZ)nQ7hIlu^e
z^f@^gsp)3a^ozL!OAu(i!-y?j#PUR(-_0|1?IPyhi(<(ribk{6I{GYRHpdGGJCbZ@
z_d>lGm1y|v8&+Mp%6e&8srxH_?5NO!*De2LAX)J$g<KYxjiVGqF_rKsGIez(Sdx#p
zwOJx3j=SoQlu9y)LhT=e2kScxfLP*BDqA15wKb&;QqLgCRrh(ut~QhG<=*xSommQe
z$Y}^aOU!=<FWlKm=cFY_?IrNC6eUW2uD$at)7|Iz5;zf^c0a91q{>Vt*Zgl>v4pc%
z!~t7Tb=|~>_<1&b?P#;!xO6@H!NYI-Hg|HT;Zq{}>QJy`F1oC}c?K)Y0{xmAU#csI
zd?a%<vn4))aYy#9LtAF{{BoNs=GrXys6wWV-&+JNhcl;9gFl?nmH)e{z4v|d7C3U6
zRO|5&|N8Z7=<)}c$!>kMqOBx?kMa>|_NcwY2|p~@P~W$~qDk`w^^*qOhu-uFmCicI
z44zW`MZRbOf-@P(&8o_)dw>(xz^+=qC%u=~en|;aR(5MTmw6=cf?5tL0CE(-wHGd?
zfrh2fEc~~AuqAIFl)9~3Mm4U%Di5NN3uB=~gBsx78s|QShQ4<C``O7~uDvq7|9Lzs
zst*EbPB8squ{k>Q=nCC$#YTahGrwwV1di8+_V0T7FP$cYnd{g9;42`|*{Lysr@I&L
zhak~!b#>UDoYcqGaV>N3iv`$$CS`2VR`3CAahfxg_486yrCLf!%`WWqr@-XQ;q6;O
zk}Bu)rdhr}N0?LRbzBjR__g4#FO1SOlQ}rwnm-?!u3B$Ibl>f}<!{FJe<O59_Agz}
zcSd~EPnXl3_RZe}rc98UeN92Ag-=YO0S61Z8Ft%+D6=WGigjm$>6$WjyiDhH^Q=*O
z3M-icp6uVD+@*P=ZcH*AvIzcqVGK2StPNUX^Dhea32rpqTv^#xldk8KLxfd@+nT$@
zC0?=-C2-rsJ?fm_wl(%UI~G>;p0kaA9GC*FfsszftH1%WpVXvqvS&jd3+^v#Z;zOO
zF~W<j-V=^b?Rtt~{IlaoOmJ>nLUpcwt5CEqd;wN5UG?pc=n>`sxF`JIDe=YHo+{!o
z`m!6YHt0CcE6PAG2zdcQ!YrX`y8Y`4QI0L)p)bh~a{x_2$Jb%7-H@I-C#UEagb%vg
zzH(1#1M;luU^7{(sd0wT`HCxoC5Q-Gv;i$CHQEq7d2tzhR9bL3yBGq~iltrI)=hb(
zqlY6Wd=~rb(aGgs)2cK{35j<b|IJb=eKbX8q~Ssg)}33;YAA=P#$+edK0xBe^XI*q
zgFJi+q}QEtN)i3mXS-czE?t+oT}R`em~MNzfZO8BD)PtQK%KX1L!q_LUlp_BjiCce
zKlmCv1f9u^aF)n@AZ~7z4#d8~I{hgkUZ+aTZxte&M&Js{hN<`W9v=RyO>RA9MIq<u
z1;Jw9=GC7~+06t*PZLwdO^(C|eaq=38*bT6;S~~mWA$;1<!gMKU`0Yq2d*5H@Cn-3
z3GhNb#u1O{06oV6kHfT?`#SX!ezI4ezKOOz5pd0I48(jm4^pgo{ist*HH$htjFo7C
z^EFO{74!h*%?O&1*8VjzN1(%5YN{0K;UT{DSP}g79pytDrHpt#g5eq0CNuF(vo}b}
z2IrXvicflyNtnNu`DtQyW+$6k9!FXIDgg%T-P)D(*CSH9b#LWEnbWOxKmrAs7(da1
zu{q3{qj1jH=6uc8X^`fsn;RAxr1d+`1Zgn5E+nE1;@vpWYN5;9Jq<d!td^Qh4@0+c
zzl{heUqoPMJ3&urUxy8B4Pi^5edUq!ogluFopZ@@2Z#3F&&Xk3{-9fguxdew?pbzi
zm0$AfXWNbhDgKgu*mkeml;EtHbWkZD?eGTAK#e(O(kq>#tzzD_?Q3iH$mSzF9kK-F
z-ZBgBQo`FOFVGI;yB*1c3z#SVJ8ia6+y64dt8p2Dql>cLiaCabu6rxd+ylnAvKGUZ
zRi2)&M%4<3%mJku@2PI{k2<%{uM1G(&xhBSZwd9_M*&4?o(rPQH1W#42t>n)_vTH5
zLqX_a&-GY-NL#y<aCEko18C(AML0_>*NH-PcC@Lc`rI#VO=!#dHlYycj0rDM7d(&h
zH?BZD*{YmA9OOz|Poiekp3cO0rlnPund#$c&V*Ulz!@Z!Ph0s>KO)xL{1|!`^it6%
z9Xre>mWzNKZBLW~$L05-Un-+Huk_SQvt~HuM<5MM(K6iX@7w;8%6WD8a#g>Q9{A`*
zN=@2p-bj^PM0An!fz+Y+KR+%7_{*4%cV2+IV0PBDT-^Bxy?Ix*xb?beA3@Or>yed%
z5+}u|j?fBc-bXKMcZ=p%SedpW`!6W}%<Ax@e~|-`=FVab8X>OLEmOzJ=oiCYfA<Wy
z%%>-<J9EaUK*FYcO%P7DB$ZTF)fAG>e^~UJOWq4*{ic^vGJ;OCyZ?aeme9?`yB^?x
z->r60Mso1!qeXE16rmOK2Ze8nM9NQC!&nQ`^v;5AFN;c+99kxP>Q)~Xi4ned7O98(
zXWDnhxv5i5BY>gSM&sVh=)MAcTecG8QXRT#QO{5?H*H<GsCY9<Q8$1!aH>S(7OOn2
zF0GPKxPynu;3>Zy7;$N--l=!OZEygdEWpC*aOOWi`drQ<e6^|YRtN-X3kem|crAG<
z-Y1;Ill8Ze4Xc1P;&^TI$OJCHN<^i?MT(y_QkrMyZ&CBFNtj6dKaM&(HU83rrh*3F
zG)(Lw{XPw2$CHTUhT$0>!CiKCoHe-xn0pnsux4Gn`_9kbWa46Fi1G@91yuk7*C3N;
zRyk5yk^WCuMA0lmzkNPN!j(6XRW*V$wg@gHl9`!_(Bf#0`8t4AdD|Tv9ANFmWfcpd
zyQ=2X*Dx5!BPeKg*QdIv7tvE?$Pt)tIj=EXPfFH#g6Im|y!9^?>htjQtV*Aqq4^4N
zeKYBcda>m{i#cBi*~A7m0N8EsmKMW|b?2rXnu$MUPC0jlvP_nimgcGNuWn@|OrK{$
zzKx!qE-$})_<3@tn%gwv)LK`l!61(Y;W+8^G~E};o6OEd<>4*WY}eE<XFBQ-@1GF%
zXN<$9$>il%e|Hgv^SMBc78q*5SzokSSJI`)L@DQ??cDFGbzapJE1>-#<Zz+Vf(zfN
zD7T&hG%@tYlq0^gmBDh}tW0Uwk1Wr8KEWMM9o5m9&B(|oY3N|=<?StjNchzIPCs3>
zvZjW=GAQsxnOm);!MbB(Z6lU!VxT6WO-T6qe_=`a|E{HmFEzHSJv~At78qnA3X-Z7
z<oXJq;!)~B>jMsMflswsWqjQn1^J6z_c$|>Z3$qAULkS^5}`{=x?6d?z=a&Hw+%Dg
zEML68J1>w0Y8q>Hef@p@{72}%vQJPr^Y>}NgC$bWxedmHhYNq)x<xAzgBm+Vk1Tty
zJfk1)et<*h4F7zMSnk4BT7Mavf&2V!pYj-pF>JVhWXm=HI0(HxR;rq)sltDtR<S}W
zOR3eYH2>#<Tz#|{YiIo%4+_yww`+2BZQ#!295mqp;plSV=!xi`T=Ya*UZ-O*4U6vU
zmq$mj>t3x`vX1G9OwiT#V(>X^YD!8f^3u6X2(jISC1nSAMagreXDO-;;o(s$2t+?=
zkb|JLv=)FF8cCw?kh+XmIYrG0MIog=tNUTG(A9Ctx&DHYz3Dhw3PH-=kHwK-1*kwF
zh!q01raSUH;Rgvl6I52R#<Pb{&A13aqO(Hak_0Ad3~~>PjfdtgFIF}-yri>{KR#xc
zSTR!@Rf4}1KB8YPGlD|lM1Sg<%W$AXUy(zLG{AHJ*VrN#^yLLEMCZRh6OY2woS93{
zZ9{bQ{3Tz40{(4{OC5AHNNw>)Ij$@5jIn~Epg1PRzYdx|M&`FPSMCNinQ3vPIVgZP
zk^VJK&WmY(h2Ktpu6e4%Eg(<q`<zqpNtRrap5y5UmXK=kH~9_-EeD}&dPYX>QiG~V
zD*-@dqN8T_Ke#Y8adi{YfJZguKjq7dBO4H>iBr3Il$??-C!4?jr-nJtl#&sjfI0`n
z<k0rzB^xQ|Q=;tE7gh4lI_iO;o|BknhlbhoSXNcb&fwN||FEatMm!VECeLj&EGTf^
z32>s(V;_<ftX#Q48(QrksErR;nEqc*+e4s7EZ&8-{IP9CXq}kkpZTWGs$`Gp$&h(5
zncTJ&s^qKLYNpty(H;|?wlX=KiS2M2?U@D8fr0sAzTS=geQWXxU5gw<4jf|<wI7Nq
zEr{68&d!fnNj1nm&ny;bbNtFOD4!P|?H_H@`+4B|qrL-$dPmkU;~XS4G#4#s-wK$?
z1RP>3MiPQZVG&ScdpmhJ0_)yr!e+;g&-F>1?@eqvWqg!VtE{Ym-7c?hZ?8|Lb>ijY
z9@Fi*VCh$a#d?LPI>t7Xb)%)3c{LhspE}cFXPa*~S@~8?1rqN~HoYuW^SI-Dcyj0b
z{*VMa!(-7|jc^FUSRdJk_?|cd<ZIbtO|6=b-~8nY{l@U_ALiY6F}yAV1r)3c`M}fY
zAiEd`pR`YCo}q6kKDBOn?r>w*B9(j}@0Pk=EL8_W#zD0in8PR?dhE|H6C%7-lcA%}
zRy2~@SmWX)!yXF%Mwm$VcvD#O`ZhG9bm6+pDcEC5gt#Q$k(N+J%}MzQ&?DI6;cz(W
z9(;mrYO#TjWJk-rgh+!U?6iiEe@ttjWny?(eeUh%s59_WB49Haq@(+dR6<jKLRbL`
zF?Dz-UTI2G{HB<y6oqOv)^Z3erwcy(vc|~<gBF2FLc_D_N_D-YCu>C@qH9+Gw7y#V
zo_5#5pIXPvS5UEZB_5zSG`WQwJrQjr0HAtRQ>;`3J-DPv_@q|gKDEzNze*z3JDmA8
z%9&(k?=CK54XLOm^l9)SQof!kXTl>SgsyNvMdBXx)JMO*yZsy#CiNGO^(P18y;XQf
zG2R^09F(U3wr5sMDZ!qHvL~nRe`O`r$jvjWW^9A&WF_1D!N7tOGKF>MSL6H-)4&|g
zAyyc2#iKEo^C=xy5aQdG(Hkz;iLdNUN(OdLQX<3B<lsn)7Q?kI20{`ImADkP2(sfx
z|4~FLAhUVZ@fG|DF}X84R*itK-+=jYm<rc&1+NR&IvUOQ!u!xTV_vq6s)7_f*O8uS
zHx#Akh)hBB00L@BhWbk3uOr=xAf%QkNkXgkO8M?P6sn=&gh@7J0ps%#;Wxf*JjZL_
zpvA)!e-=01Fci>a)-3rFQ6eb_53UoFmI_5Pf=WqhJd-xIt-owxo|F%(woLz!!gyKa
zYqLx0M;B^22hB&Y=|D!boQSEr8{H_&oif0xoWaQe+q7Yes*JO1@neYhIEIBVdM$@t
zokDGL-)CpCV-E9sif*2fv%zTO$2xbqg12K+TVEl;>=aKAv!*_eVbjy_B={%TxwiJ0
z6zb0L;OI2q8|cz!!P(ye72IlKV&w~yI$apsNWJvATN_H<opUOmH=bYM#>4YldxK9U
z=I|UqoWkaJb8|y$ftUHL3XO9^rUx>2YpaR1>{Yj@JciPTTTT07QF31(&fu__hp*#9
zc9+L7{T%3T6Xh+I_dAi1DAe$Yx2=co8(Rsi5bDibs<WkVu}lJ7WZrMrayy{T<Ce0;
z>lVPBi<)f*YX`NjPAH2PSwuZJ)W1Ea1c2{fB=;4;o?_a6L+2^oKWi4VyeO87peC^x
z^R9JBkP;d!F#WhVd9ct)(Vz=msdP0^Pi2F>h8cbzv$<Bur&>KS0%J)C*{!_8Un}9E
z)v77PpL(~VsXw%*>$#j-zt%8N?F}2u%JTUtbv4C*|Ks2e?as8%H24@ugm)fK)&lyX
z(!d=z-mwgIssB|73rpFo9w>2;+E{lDUj?N2a3PS{_OQB%VoL#Bjt$YdE#-g4f`721
z>=F+}ok#y(cRrnXh1-4R)hF_(D|`H&HS6T+xR)S^oKuu2*+nWKTqW6vYl?NY$^Izk
zCk0biz7bgbqXTnP8WisxI4nvb2Z+*Ki89eYbC!IG>bSkQJZdm>Fl-3y9jVI*Bh5uT
zVzxCSgVw~8FP(x0;yFqKyaU|tY5)=3m@V8U2{qD5;p_*=swua<#qM`E{KNzyHg7%r
zyu2sT6Mo5B?lm}aTovg*>RHvBKxA?ix{+>DB5l=b2d^wBIDb0M@A_5M1Fkziq~<7Q
zxV483nDLvBs|UU1nFB~D!d8Vi4HQCyN(|Ahtm??qX|Yx8#)vrJ6?8RuF7vgub=)Z;
zEHGbc^X%*ln+*&@<+o^lsp#{QL=@?pnOkB36p#qWFY44LuhQSPLecRLjWa0v2Tc|a
ze6bLjek=b!7drY?J%v=jw&PUm+#HsnWxi?V)0H=xp?0|MkwuBT4A$z_bAkRck#IRD
zQJ&Z~b3vxnH~&rU6;`dRe1hZ9@F9f=676`fq?7A2`KPj4TT7b>bxP^%?R)7gVfkbq
z<IFbk`_ISMu=eNtvbX<IrTKkKEiBA2Pp<G}<de%Dg7HC4)^HWTW&Kr>9wz=wE7!2R
z)}k|YbB)eEJ-ZQ6mU=X~u+THP9?0E?Mng#5%Ni_NWOJYI%La)5nZG_t1a3}if!npc
zo35)5mN#qL9M{*y<=V2_e(lE2u6DWg^_n8j=C;d7_A!~IyN~@6EE(JD(;^M>%VDtM
zB2G%XbRKhqmT-W1Y_5QzYnGvtuGIg)%m8Nu*GNU4jN&1ehr56IylCeaj@p{etA+7M
z*r$_ijH}S=4Wk5fEOzUzz7f0BJnnyYCB+9JC#s;9!-0_fysCl7zT@iXeBZqKez>E3
zRrgf+*JxoHk%!pN6VbO_1A3y#w4kB~n`d`__qorOUC9B#KT8S)rsS>@!X3pwE-$%o
zsI$n~fSZD@^CnBIPv!8D<^6==7>$t&`AKJGrS+qCX)dED9p;TJFbl3$ljkSWgh?Zc
z;U-G2RmQm`|5pVX=|JrniT%2cnrd%b?ne(g@AqK{15%!F&`-CP>!ug%uiiHMH8;bq
zvHyjX3y}RFQ0*iN9~`bI$46m2%t<~NA(s>({pYt6!aQzMjB1mFN!IQBYASG(b^{mx
zJNsFgQ>e4zpZw+Nk;})mUJvx$U8>y85~h(~6jYe4sYpl#NfqI*;P30THCXI_hJm3h
zEo5Tifg?K)&6X6^Ka*JzZ{ey4>#cvtEUWCD8P&d69n%WExtSbps2ZwwhWi?XL2&*^
zaeyE+P{-mT>Uq(h69b^lQ#HHm2eMtr-j_8^Tk<i!r<UB?9p?*^&7^7=hCN_4cTtfG
z5+1=5ld0Q*OAWASBb%ZH!#T(52|ryo%v784nZLQ@tw<?Bzu@}m$cArX3OkTUg6K|?
zhuwY7K97HHVl@#-NJ4p{pHucc&1A4D*X00Z{bG+Kr!wYmE_RAOw&`u(+1be$udI6l
z!Y_&TYd%|WVT&^WK}+eMe8T2K#2Ap-H7WE7k9J;;761(k5=0Z*jdK{u&4E=x<NQ&3
z1@LpFqRHju9xsXA<`b>*Rou7nnkV#8gTEBNkblNuC8*L|L5<c|n?0u;Ll+C89ba@Y
z<Q9Ex>8efRrgKTz^D{Xkgw2+;q+%Q2A1Nj24zCGe#zh%AbQ}Z2b}1#AnAB#cmXo<5
z%y`eB0Kiom6*Dl^tWUv*hp<Pt>cw;nQl<8<Uk0X>53r`#<!>}&`45wK3#i>Aqij5g
zf78sO)Xl|eIYO)4?{2*KyXHj4>6Gc(mIJWiVmO_-cr{fEUGWf*mJJCE^xbe3L70WC
zZg&3KZ9Z4VoSY25Yf(p>*>-ZyA_v+w3tkQbrifhZ4;MD6HcSNsx0#6;k15Xn*r+Fh
zU;1c#eFdS%TiO55KhV<#tMkV;4mM4pMaXo)pUEeTTq5r$?hhSmwY1+K%3KXWm8J6)
zK%u=FSqDGK4+h|Hvu6*WLJGeTiDy*F<P6`gx}zw9Ip*0}j>zk)+zZA0JMqOtpVb0v
zZepO_`M?#o;KUPhrP^(=b0(%bIVi6CCRO=^r+PT&OpFT)Bc*AvGxTu$!j=SVk!fIV
zfUCpx0eo^PP1KG(KFgd|^j_knnCu^5_<w!7E{yt$-CvV(@0xtyG9>kcBCj^5%_-+H
zIPmzG4Z_5*H&(WqeyCLWRFCJ)9y9QsD&AjGv-+``>Suu{o$HvvVu~kVaF`C)Z=y&N
z{(M>3eU)rgy4S+1LuN&`pIu@}w-4qBN@wy9#ViNJ7KcrDzOxrojbFgCX;fIC;0WIx
zzUDhN|0}naf<+55`@?*SXP+ssgJ?tNXBv89%X?QF|En3ck7i~j|M}^PMgi5toD5mT
zj_s6s51G?rJZwksvLu?V`z(;tO&Ma}NrL%`9jvTah!i-})E+8gk)3E!wDGP+imeh~
z+VKL;3*1Ls*|u=s^(Ibs*y5iupXTEQWjY^a+vDzhc{SVTJy;g~kg4e;TC`_Mg;Bna
z1NTXD^xI8M7mGB={r2>megJ>iM{=}b%12T}A@S_2ELH-V#p)T4@sTpl7~MSSw;p$H
zvM0|ib283Q<3gS~OYk4_iq^5Jl3fpefS_C5T1=;#PC6%@MKoWnA-spz>*q1x?o9p!
zDMW&9+vGkU71mJ3uon**r+y#D&d%JeQrg5lT<=VKEF%^m{8#BIKpBfKAi3G}r1J|q
z7rNG+uy=fBh(-wh(~-D_TQdjZxvEf=nqOckzFEtUF40lSf4g3kerlrh1NxoEMZ*gw
zwR_%XoWW<GI%HeFc%X@8JuQ5O6iuL`rE&kX@tR<MV%XH|X8$JeUq@z$R)%=2?Ld=*
zmohoZdzYW!@KFu}*w&I;QZ;2h&b@*6#ekW`e2G)gO9m!JM*4XvY?2IlX~|p1Uu$%W
z7Xu{s`-Hwt!M=X)j^(Qm=U1=X)j#TpjvgI7l9NE7Gwn_<dQ$6istrzW3AG`Z$@Q<f
z98T5&uaf1G!i#DiJ$e|e11(exJ@mMrIjJIB{%38GF|2P6;Fpy}Z_HA_@32n#AqW0M
zbKxubHOKXVFFXQ7;U}WQoewJiGC1z;%g3pQGPAVWy%g)1$s?0(M%Q{q&UHvQua36f
zzVm#xVYD0j;)k|^*xco6&CcF@)?F*d^ucRof(J_(siOK_hoiMk={K@>=l9;C*|ZhW
z(P1w?h(MU}fCq}Qyr2Y|&?Xaeb4zC=-eA8g%ZKKka;I50e?}~cZi+u-S1k0TVDI28
zJx$IhBt)WM4B5m7Aw3P6cMra;TFq^!4c<l``ELBk&It8_O}ea4uX<KZv=|6%4G`jk
z7YL^v-)B`ECA`x42DRdu7`2u_p=6|_W@N>MReuQs@ua`D?o&ALCkk^hfp>qeNyVVV
zD^pr{;z4ib12w@V(?4aq+tx{1`f?vzk-?E1%PM4el)S=V^Qp5Pzq6b5s`3Jc0y9m@
z$6_P$oD@GugNXQaph>v$U}jsZ*`Ko025EG!xZ~9N2Lt?{f^!!`9qK?uRx5R9*;uZN
z=VMNwiI(`&ErPiaWOlxTg_aZH!&2GzxQ)mYABN#Y=;7)rhZ%w)xJ?we<Bn{k__#vJ
zEfn+~kU?C9j?of`yg+f1T5(dJ+<jQ$`KIX8hucTz2mAXek9^IU7#R?jd6{;B_lJ&>
z1x<4pgQ}<A*tF7lBe$5n0OmehX|0LINoU~s#)+^dokMXQ#fwb4?dsRvm9EuI3yWB{
z=7b>@VSFvOF#9q4`nH%wyL<wxZz&JW&3We2u^*sbROp&SB<26N<u|U+sT0#DTKyF5
zP{ne9)shp?(yRUJ0ioW{!zd(F<8^<ic!t25oYhczz7l}%@$tekV-aV4$+U~01i&&b
zoPLg7P+`|kkC&i_LSIlLVW@E*Ng4+FB>!8){CcJEtoWQGvRwbgmT!DSLu%&rb?@*=
zkig$m?gsS3hg4Klgs;!*c^{M(l@qd<xTH@Vbey0oUH2D(E!6v`NdAz+b?t(eA%Vv%
z>Z#bv=SW=COkMfsd@SGm#>A`!aP69Q%USQ~>3rlSW7#-|tZ!Bru3M{jHp!;d&M<)=
zS|xHS($LtKLDQ!cEx->)lE1Jd5kzX`MC+?^Ok;CWU-AsIy>LOs3;No0-D0WgH~q1*
zCS-%8IEtjfczrfH>(8|bj!r7X{<c2fm%Hdr{!TwbmHUVGUfF!;{Z1bQ$h^3rtubD-
z&86eFE`?lwOUmBoV_(wZqjXndqX4<S5Vuc=rk;;C^S{W%s|H0h=^dWmuNH)^4kwx-
zrugrQE$=2|FKrgTzQ-hpX6PXes^<1bu&v$HM-PQUif<C<Tg&gt^&3qvL$)cK{e@j8
zleM9z1$PsRBPsv-PaFb)e6@<<dNaPN&@MxL?8k|lDz+b8c)GKmIK2M;-Pf8jL`F?z
zYm%2SmK&%oh0?8^bC2j>s{ajQCd%G4fqH<Vb3Wu`P8QW9SawjyD~}8WCZe$d)1f@@
z1^oxJZ7kM-T!$<~xh-6Dp9v}6uxWj$<EuK}g?6E&!@A`(oaxfw&Bbxd$L<0|gGr}-
z2K7(~XVZfq9F+h6EeLdEk0e+kUS98A{9Exd2ol0~j4T8z9rh5$+~l0@^+2QrBa(2b
z6~JHU4Ae+pZjlOB?)_!xo|_ywgMUb(&*~_DKJ1F@K*nUs4d7A%CJKFwq&ndDX*%CN
zK~BX>DKuEd)rcyZ2sYxQDOFjMprCM@SoqHY0X@*$0g#Q+Fa?0KWiMli^D5s7saZ8?
zfu;-`H{r{i_wx^xdiT)+n_u?N?F_CXRyzKXnKv`tHzV@(l0j8GC;U%fqKdEiXxSS@
z-r@Q$O``QRJOtPw()AK$Y-s}r-rs|BWK#hu01E~yMxju<!^3TE)eTC5S%1;-L$ip`
zj4b`95l!;p*8JgQh-YAXU9ld?Eg$bqSCN8JWBWL;d<$%59}OBJILKvJNevjdboCUc
zotK>@S!?PGYSpXA=C6#+&2H#0(pPh6Issn|nYxFZ?Ew{$43yUQOLf&dEEvI8EpzRi
zolZ@rJ#&6Mrsz#<DvkKY&}qL~m^*jI(8BU7u17NyzD;_w)jP9Uzrla!xADKtR0^?g
z`w(C{PL^K-*Lw|B*e5FVeBK|p=sg%Jx)^`W^vuS+?tE`@@--7v*ZJMpyGn!3FEtqv
z)Q*i})d2S$_8VNtIaKKQ&dnuo^T;}jpUyr_p`VhRj74{ruzj?@pOyiC>|LX5aA2@V
zr@59pd$yL{fZ&V?Zxspf?-d38`nOe07Sac}q|8YvgEuDq>uxL^HSZU6Bclc7L45<P
zq!j^zT?_V}9R{M~YJ{6ylu4ewN$-@u%y{xj0`J@3Lh#kVQSgX;Jq05$DZGRt&heKb
z*LEKH6EhFUqhM;3ERN_?n?EUUaDIU_T%YoEpRDe7-JxG^au<}MEbrohGtB*U;!Y(Q
z2ra^Ld|2|+=@`~^4g>C(?g|V!_;=fs?QP=Zj`;8Tfpg1yw7Z~~=48NT3M)@c5$7z>
zy>fAJVR<_Ym?Jm3&JqC;cPgjx?8(#1{tCIf9{)Sn;7dIL3LU4%M5>Qmb7+(5e;Y-x
z3-j}7FR!fd9CVeJvKFzbn$@YPuq#LRm5^sU{18p|ibtv;o^lFHDA+S|9#UD)gZdN}
zU3l|lm}bN_pHNRLryu3HWo5#_VH&_7P7)rRbDd@^aLuUZ{7ZNmb<q>rx^8GuHLd`B
z;A((?tz?Spp|P=lP~geU`p|Gg?5xmdR$K}qBO$0@q0aaZ>MU9++TCp9|5OHEGYt~r
zf}X#%CbJ<f@<2~fj1j+hlgDC}YXfBo8|Q|KFpJvbMCkN!+NNwMwcl<nuSNpBFn>7>
zW*uCzBE=+T4`>EAGejndW2qaG#U&6}5=IdgO(PNDPFJSFKsKlbB`a!rOj;bKSpJ_;
zL=rbM3%koLm0snW;`s+GZEYRDC6T)yKjiUeVv3ZQ<NVGm+pvndu5k@@gy6PUD?~*m
z3?jGp0hDTs?69(@D4O3EcL0XHI!*YlZ_$^Rty7iNy<V+9l_z~)&d#06&@(JND;f&+
zyE(=V`v-rmt^NDgb9;?)tL3ZGmni<ZdwLzdwYSGRO2h(QxRh)pN~Io88)g}+FXA-m
z$Lxu^ADp*iV9x}zElVHnGoZ}ZcuU?Y(3Zxsw3H(LF33o?EviKs$OBN8aEC0-T_p_r
z#ClP+ot|T{yPF#y|Mhh{3NioA*f22PK9CW9E!CcEoyMx#K54<nu2fYQpFX>XVTj|m
z3W7cVUDD9-qO~)P`}r(lJ$Pp4><W?bJ4i5l!)fE8W{Mx-t1@`^<XUd-QM$p<>T7of
zCZ>bEJ(bT`8>eERH{kt(UOQXgobMVn)sgpQu)d=uD{(RCUj_%-)V$>I0)xe6?32J9
zXLMRhcY-S7G8D*Qie{0UU?QVp$bVTLjZ)H~p-Pk`VGuK6*H>(^H{*#263ta<{L-`(
zp|cd%Uaz=G(w0_rU3jl!Y?ymyw@Z0!0Wq;ZH0{E@aWFa>2#KPbwsy{KBogxsoa-F)
zR|&<7%Y7EjECpdR-ib6+Rz<F~y?ejb$A~8oeSw-Qj{meKm=qpHFFlX0lD)0!nz=8v
z)O_+}U@ZSqk?^>t!nY{WUqJy>nL~tkM{>Uk++(e5+n7HlF1Z`nw{a%6$99kKkIr}+
zn7Y-*SAGIylLZM-QzEBYLQf6dYWVJJfz#{whJ)ZkHbhoKs-Z8cuAxDf|Akt5gI;VQ
zo?%t}{E$tg1GJ@U(b8D3(>VCwa$?ub!dz_GS<i#i2UY2MYM&t~ttuKw=V@UM{zPD7
zX^@a8Jkq-dd^FKt3&LrH491NFA#+Nza6aJmsbmh1;4<R4WbTaMvIB7(98}YpV&v*@
zp=7m3PSVl;8Tpi(CxYPNoT2Za;VJS05K<g)g+|6xPqj(|U$6!Et)3c=;Ajh4jVo1=
zN~xx?0xPztYsV?5K*FIA6*~f{;^Jb$_Fk08MWeliGn7)H9BW0H=RF`wj}YK|^WMvF
zx(Bu>9<HF&pR^<fG%_ww7q}|XDp=vD6}{EsRIm;E2!`WAxo<xhiD7%G#RJ*kf7pAT
z7^GZ$yY3YYmXea92ZIc-SFNW)6Y)aXg)O_HhK@@=g%jypH5XmlT+i#!;N+pNKPUGn
z)mC4hD^=NMt2@A33L9pKzU~-2@SRO$dfC?5>DQdWIRLDBc1<8Z)BYpGdcg>!5L%{c
zA_cYZ|K@gMto%PpMJphznsdsPDYDNLDYh>9FMZlBX2lj}X#7I|703$a;iaER24Zq8
z)TM9#1pxkzKg-lMjTP3S3kA4{=y3OZ=1>Sk)zOFvPkIXbJLa&@6Z>!VMGD5w&q=Z*
zb`r4-1|Y1NV^=ZW^Kt3(G=tv)10pZfATxFdQ#4_XbTOUAxLFdIj=5<%hef(nYPpw7
zQI8InN&U&zpd?{@8GL%&>iz@Up#Z+2p7(blps%f+zUX-6;@sTbhE-Xtk79y}3Ah}V
z<&bP2Al^xOLZVSKS-;FXS~bZc(x3W0G0`#<OZ%NMqLJV%p-`-OS8-7Jd?1{k6!PEn
zdmIo;(uqcA4PEK)*s+BBow>z|yar}a*AoKMFsXzRB>%48!%(rk0Zu@S@<e&ipF#fc
zfF@zLrN@Lz=TWO)bfkzE9o2F$Aw{Q@exF%VSieg*0zYOcIL9H`5n^@-g8!H1<?@9Z
z_zF`oW&QD;iK!K;U~Xo9e|3EToUdvx1n%nU-TyuBx}Ih>DuQ@#cCU3{746RZwf`?9
zyFd2#>p&h0zuzg#i;wj7MYYi{-gGgAUSN&E|NQ-!{sI@f_ZI~efeZWo)h=geBzOA<
zq5pcZ?z?&u`TxkCoqfVu<cu?DS%FP6F7nk#b&XD5!zo<w4yAFjOMg9IZ~2%E>>Ber
z0y<GaLdh(t0i#lI*2s5wJTw_>fp!QS|FR&ai6oD%27*cwmDMTWDjj7N0^lBaxQ#d0
zBOKjCAxg6k@FhAyFXDs&&Lz18hv+j=kbI-a=2jVFIEfrH1r8z9DeZMLum>(w(X2zy
zs&SRFT3L-*MK<8jQKqi%*b5=hltS)2#c5hb&a76iUDSh&h8Mnne$_+v1_W;m>VXWB
z_HA_hSs?03^|s)N*6Xj=vx08wy>COFIl3EiN-!pJq<$fR>Z()psYEHW^&(L7^S5s0
zRO`6m52?f1!YC9<ofu;Pw;W$E7U-7jJi($)zhk9JSE?&W@FXgxRIZd27mo-Hb1}C;
zkVIkc;38EsUR(jG&2XF#6lx*Nj3Is6?5RVy*XLts^l`@|#*bt5%ecL}zc6h)woXTQ
z?<wdbXE6r)>1q|dy~Q)G8LUIrtzSDjX`g&rxYlnqsm{<e=LSmM_-cbt`9og)W(K2P
z`YbGh7TEbi_PZsEaZ$Cd(=9>yfTNp+%V3KV<D4>uhz8NI*}$oaO6R7^>PN%tK0ncw
z_4|81?hYrnx8ttO)c+*OJlAjPE16`1-Zma*^5+v8l05N?)kv>F*H`;Ayp8x+__HjU
z6gTe^>nI&nnkaxsph<K`dY_Y!)Vfin&dmgWp~d*XTPv#!@<|*1P#3x&Gl%W+H}Pbm
ze!g$XX!u3AMdHU)GE+r0mf2t`H}m`HLz5Z<GQlx#&=!W+H)UjVwP*s<^ayFUnt}A&
zjU<uz+Sze|_P^Ufo_97jG<bV?;es|dH`{x$OoonIao}Cid4Y83(edRcrkAi?3^u})
z@Yw_%@f)7#2#_0XL0PNXG1JSJ`pQ|~+fl5l@vaJgl#RI9ydhvURZ?(v8i)nl6L@eU
z{ZdBacKUu}QDf74j2`sH1|ry8>qdSuc0}4Dc+RJn2p1$zybnzPE?c*k`7Rn}#_UD9
z>5XZ2ftLdA<#3EYDtzgI>s<YcBy_;XhWJJM0J~aMkw-imi=FUC&~lTtqFG^eVgXx}
zN*{qc!_kiAo%_9uUgLK=R2%9iGw1t^*bfFVr0H^tQ4J?AHqDGP#1>T7HZwO@cd;vr
zGA=(VD}S_B0T=(Wou2a7*2vskq{?14$kILe2>cJ&<Kg2WKz$0F7u-*|2DN8tMbu`T
z%w4W2K$B~)=8H^Pm#qY{M>2+gwONr(qU-yu?KKvLHHuSzOz~@>7o{XAOdb~gCjw0q
zz>O}{Uz_w<jw~=UUrU_TBckePG5`dTxl4!!w5aB9{C1e#G*8{BvE{VxbWSR1v@s!(
z2Zf$nPs`W480}2r-?$=xTZN$B4(se10Xv(x^*?1N`!n<dB7$*YKdnax2p8VKn5NaG
z;KRQ0)`x|QyE0#ae@i{Ljy|xP;}54Yx_aFGa1#!sfrEoU9HuCjxgR2pM?wJ=5cmk1
z`k;@A^J^;ZRsa{LGpDPtMWevD@&e2tS{u;MRvrCxU+fT`f(^+2FdqGjfQA!~dC#Ph
z<X}tvrZ=~(tqr@ji}8bfF|svoHX7M&3*8(t#gZ(sUKtDqJA_*IG8UsG<>%+ehhy6~
zFa7Cs{a+g84R);ozFuUz@d{ur?9Rj^=A8k_&5|ilp$0B!%7odf(`<I2{#4(%x!#Ob
zNONM`(Z@Rw@I5%(y4r9S*m|B><#@=6r{%DEG+Mecu2ue`)qGGmSCdNFDW!SBZH&L>
z*q|)l0!`Hubc{!>sK>jZ?_@Y^=UoT<+1)@XsNe)vtr|2x5ip%JHh9}yjhK@a*O)`u
zVM+w6+FHfCs4>kGzxz0KrhRoA?HjPpPpq0y$&CQYuz)U}b;{V>Gp*NjC;Nn!?<*&R
z{o?n^Vq+8_gPWUFQc}0ORk0nD3o}wtR)}<c)Q+#_gidXP<s=^$yPYwjz>-2KnRN&o
zb^`c%JL_e0GTrbVR9FJ%ju)zzpYHd^7VdtHg=uvDTQKD3cW+R%%rA)hK7F!XvUrfX
zB@IY*3Y%a->XOk(F$b)xVS{A4$`Yh&2I@$)gFN+=lKM0L`(yshDmJz7xD}D1=eU)@
z{;zI2tj_1dU{nvnIR)yM|F||0S$w~GJf$5D(N!bOe1wLju@JG;B+43i7L%cK9|T2G
z;e(({gXE7XI!+Ye^}J06YCoh$t19DG$$_HM-(|GKbe@+N57-LvNa4xyEcM=14XOCA
z5(XZgO@(fw&TIJ4Rd!q@aujM6-$|HmF^hMY@5<9v*-)>^n2tBNB5F^BJ3Ev0B-xqu
zK_H^cX{xUKPXIKG5Mj`wGFy?KP|LAOSFE4E9oiD<if}oW#sLqo>SO0>Rk?Qh04_lv
z<j4B+3VOV_%_=W`Fx3ZUoUqcQFa?@Y9KGj5A6GZW682x}u`{ngID-X0evIZh_Z9my
zOCzxG<V*dn39joKro?b2vHp`SDPp~P2u>12tWO;$iZ>efMP2NCv|3$qcT7BMe1gK`
zd;ue3vJVsy$$tow!pPDgPHw+6cz_Qhe8c~&gpD$qUFi$ih^8c2><z$V5b+C)zx(&N
z6e3a@E2POD{R1cD+qY%x_E1HY6I*xp^YP+O+WFEsXnO}Lr{B+_86Ps^F)4r<MQxeO
zG$o9Q<Q|4ocS{^ZIP|fqg2&0jAqp;^g~2rNkR>^26jTnEI1UdtDurGBDZtaqmqhrF
zRe}7c7$+XNk_5W_ZA1atRw}FK6hV<+I|E%=dP?dx>zFr7j#jy9O29$m)c5ZN55p&h
zTt#yi+x=^+D<>=S(DrDtqhlAOwzhV*f;E{7owZ*A9E2}`L!|Pix0~<l??E1aPwc6O
zYDG-P-lA?-S69y$LLEf2f&IO`K8cxYXsH78)$Q5j`po$FyEmqE|Je|^Tw+5myXJ_p
zr%h^t$p^JK9(ld{$xyc>8Kt!9UgJJ^J?^1hp~#`<WH!kc_BH$YgP#7XiQ%eYY3*vH
zlnLtgY>wTjATN(&i}B^(hEt-(5xqREa)m)`wf^7UN4Vx{x2RH48o>I#@APyV3-wz3
zEga<I;hSDovB$e3tqvT{AM7u}e!d;HXg8`h2-(bYX?&|`uC*yPe>Qme6Z1s*`t`~7
zsE_1h=C^hYM!X-}pP7!DM@FL&xdgQLBbgs%FTy3IV4{Bl4F_!-#eAB72zh%3iqaE5
zc|zjjoR3<%|2OY&dJUW*KD|basB_t<#)<uWjQa?WO|fN;uS&IQf~#r=%f-9=H{<=9
z**R<#;q|L5tj(NKpe8<b0p2hNr{QvujUEf;yS}3@Y4{Lgt?R2^7ihCr<I-MWic`H<
z_gJ!zm>F6GfrM*rD%Q@KKRDMFAdEiPV-nl_c)I-19!k2bO9|jPvWkY&GIyvl$pjiU
z#n#d@-j251S5EKl@9(|!J>B`YHdC&htk7=GiW?)$fFY*Ff5b)%;@*_zsUYtq&d(!r
zH2+D4=1+K6UMgiQ(M9{XJ@u&3L|?h2TjqlOz}LP(DQ9D(=Tgn(123n50Qe!ViW7d}
zOHaF8mN7hqC<KrN!sBGqE6y26A69BIX-w&R5y(V_xcLpQL)H7Yh(C6Zj+PBfKSEO)
zbN~02;wO9Gf4`r705OMyqO78c<#1JTxmdBaoeK?X5K#ngs`pZ|{Oc`~$fW<|BGTqV
zVFG);H8DRv@=n8GQUNcRp-gzm#A~mt%IY?7-$%ddi*f;|aQZ|TI~gsy;jY%|WI`8B
zjf&&key@8YBeAzBYQXrtPev)4P!9S;F}c^~{TnHVND@6of6s-VIxT@jg>t17f-n|^
zH0&-IuicUyc4bASUyqDvOodaICUpb675#zZa7{Ppo17v9J{U-)v<q$4xNX$*jn!~j
zl`d6^#Ras!XPzfKDQlSyTnM^YI;!N})oCF_4B4*WnAzB+XlFb%TTrC>fmluo)}(N&
zsjWeek%xy<2O~kpEWT)wKD%ww@Q0PZaB*o{D-}|<hPO39f7@FeBAGA;C`sA*xtaHs
zLc^(k?M4|qE+P4wC)<};A7@E!_L+x+jOCVe7jUswjIuc2xhWMC-}5)iQn4HUMTGlC
zu4*DTBX*NR480AZO=)o31nG(l+r2p{@<RGK$e5;A35^R0{<7uLd!Ln{Jg+}Lk~v|#
zeI013^TJYtsq0il&8J5RZJ4p4EFAAT`B_&<U8APKOnWm`_<#`E-uucc{asq+#nsrY
z8t{fc$Jx`@uk!^xdei1JIP8^8rlA<YzwzbUepl?KXZzx9Zt8%6G)YSfIrAE3jUe74
z=~0^{-=yVz`zrsv>)B-O-|N_}0A~e)ZyZjdD3s4S@9(?Ap5h)irik4Z<!RyTt%;M4
zxeX9POI;QEq)+|}2pS$&NTX!Z2;qpe18y^c)6B)Y-Gd8GUKVTBRB@1cIJL+piwSxg
zx0WXrbfL^QilSfQas6$2nos|os{5<Bj%a8k+Y^7)7YTc;kO-Bze?G^ubJdVb@ZXkd
z1a^#2!}_D1x1PmSG(v8~G-Gp~0awoFT_2}IVv4-M)ZI?)ecLBt42D(BBA>qN6u5&W
zGR?~dUankRo*Z2$SK#5~);CVwoh)8r+(U0&5C&Bh&IC{B`D)$o%I`0MamV2I53c3p
zi4<89tGhA8vq(jc?~YH;{Mv~dO*9&W<twt1;aID$jY<k$)_1X0g9U{OpOg)x1r2wY
zq<Vv*ytNNltBA!K(%lCw<kVvU$R11IudUYJ1Y2?sh41K}zIlfGLecJ6oauXLOTgud
zL`S<@?A*MAhsz@#3ZjRcV9)nNY)TXgpojF-q(+fcdr8tloOsVsQ3F9<k71ibf|uPU
z>d)Uqa)O^0&&d7_5$_T+el9B|Gb)6g_hr{+#6DH;aVmPCRQ4GArqolT0bcL;Ww5Z|
zC38OiKmj*#0TJal>Z%c$Y!W%f?fcE$hStvQZYh%2{i>XBGW1=IaG_250rCQOnmx~m
z4V-w3EW!t^y{gMVJ=GXYV;x{OQMtLNhD|UH)$eNU<-xFacHql;zA;SyV8>0(CSFmO
z_IVQCT-fb9+dp+Uy#qHkz(+v>S9wMff##AKD|HGbztojqthp!yJAH$7+UvQ49TI+_
zsz1-7gA3}bUa&k{>3@0tck%N14<Gn{am@?fS{h(>d}`1Narn0|@j$*fLa3GyMOjP4
zVO|H7oC>kguIse0c{#Er;P&19Hy9DfXLm|BW8K$uT9c|@x7EZ`3;6WuQyi&K27C0E
zpMKn`Ld>{$)^TjAMQJ+^z%-vJ-cIXJ{D!r+zk_nDE@OAe0;Gw0le`(DSjKt<)^rDZ
z=x97hAozL$YxJK2wffu`7-?+?Jh8b7)C8afi2{3U;K59=D9q<z0c~5~@RDo#5#F=Z
z3#<lp-4467!g_QJfL&|x9A3l8$w^n3^z3|o^>mf5ej}nwJa@he=2Is&n@#LD5J;=!
znNt>M;G1r6cH_0sxU)5d7*^U&wwr;&9aPP#r!9l;*4y2?=#-rF4F@mIY#-6!%fGFj
zR?U-GWNxT@o|@%vzDgg!-r{#AiTT*eEY8%YN-?nXE(sf5PpxrwMr>7Xs!WVvwnlns
zI;4rn^KQg8V1^Q=uFwc6^J#N)b1i=C_S)Ifv6t~%>@M)FqH+brsBQ4S9zrJfgZ<9y
z1*y+#Bf0LtZf#x*aQF$hneBaDrEm9`h%~@sc+y#bnwbQ4@XP~@mXcfT3H<$ot8|Q?
zvLN{KqG1f^a%`(<@s7~U`P6=t9gP2FhIgs62WZOuxcl*{Z11iddN(>Irh-<dzc*p>
zwUJqcgfbS*<XtXbg@;6Lqc~VtcEYV}tF;+XTr=P$(kO^71fr{-9&+Thl;{1@bZwF<
z_%;C&_W8Bqh7N9>Fh$qZ{`FWvey6wu<M;8!kZZs+Hr9Kt;yczsHT|BV434*D2R)ca
zosZqGX)|`-4OodY#<g&PZRzcL#}}|6zKeGt(matRq~MPoN|dj%=9u=0Tpte2WOYpG
zQXJfK#bjH?*FbHE?b)}o{S{W#iIPgXMC(?!vJ(XtZwG*XMT6(VLiqWT^!xg{Byn$o
zIG|Z+Li6mbi1h?r{0X><U%dLxqcHLNet+@4-}+0|*d}&8i)8LWU-NEGn_G?f6PVxN
zM&|>_W*@cCfW+9z&EetF?fGJXW^KMKAX=b}dJ%^ss2nu~v(aEW){Hooexk5KM)N&5
z3;(kRIz}#OO$g#ii?0j)CZJ+Xs?5p;0Y0d57Y{x_XAUE<+&L0P3kwUZo?7NabF{$x
znl2u`1)D~Up5t}2G$-C~&x?dG{kL8H>hK+uw3`~~J4|hB?6rx;w4d+CeSft%?~KYH
z+nm_;L0Qn*BnzTTy)9hcOgHI0XJKSuU=H{=2p7b|JE-1no1$*As4YN9PfXbYfquJF
z%$v7D56I{r!ty>o<cfgC*sSJZyVFG)(<w|nr=gvQCDla#&Wf|alxCH$13im_LQ2}f
z{gqe*^ya%e&w=K?a<OiCwJ19{@+iPe|I|w0$1e?EHI<4cKEqw)xOfVVmkDwj1r{KF
zby&u0sOWQlcPj^W@Oe{PvzY;8Lw*a3gfW#xPYu{dH#d9B!6S_Ht9_@5>o0Ehs(EAD
zTlQV!IrPYEENF_;ax~aE#<XxrA0G!xGn$#_56R_819=-<F#<$I-?bCYpZH0C+w;m$
zn&D(>_I;ne0-sye&39$kOVq1a36q%rC@)7=*&ie@I5w6L9gs!Lzi9_Z{^i@**t3>o
zB(_%VI-l!09qqb3ytw>lxDe`nwr;a@ihTA{jl5D03#cLfzP+~-NZ-(03I8g%)F?XO
zK_mG{Ri=7Qi;Vo01#h`$4lS8{eq#Ics@*oD@2UYFo1qZJ-J<_uS_SQ7I&PPGdM75R
z^nO6ocGF>J7p}%rhP!Wo7))==xu&NQyi(t!30=WElq1saeZF5jy~w!U>fhMq|H;Vt
zdGOi#W>}E0b8$~-{<@~%wE272%^v8pl{`TPd|XhC{Uh73!6pN}<J@c-*Ns$uyYP_D
zXN=d+m-QhVZPZ;mkIuYr`!&B3;tI?fg)ZtWl7<D!O}9^+CDbwk4A6E;{mXc{O8>Ou
zX(K%nwi!BABrw$9As+=#PE+vEt(hu&V-+%HYLgJSH%7@nI(qd!A!tLa@VutwL@@qx
z?USrp(O=2ezPEG-oh@eQF)fP{Ra&*wl9_4XdiV8fLi8+_qINsAjKkui78oY31$uUP
z*xq&9uH7kpx6uM%m0&Nc>h~=t;<1sF*6c4cl)??jz6p%wy(W`?W1xVi+}?~4=U`Uh
zNsm%6?Q+h6(rFn^2J8EpVD?;Jc%ZG{h&iRW-pHG~^RJ>Fa4D6mvq}bf-57}4#>JUQ
zeOvlJLI1j+I)vnpsOALMq$%D%RNa&b-Gpf?>ITTO3|UkETec|85O!cgO=Cm+w`ui8
zU86T}Zl`8{_U9p#sF+vNW{6Ey3pW{pD^Dj&&z|S-BVchGvmEuUMC5K}`|=Du6(9|M
zwT+mld^JES9pOLVsACbphX-L{VGi^mCZ%5Oqn`W2Ol{NQkBXxlKWEjY!CM{LTDXN>
z69O<*T0{NGu7*2}36dK`6WV0=pR4HYkaj3No>!8CY3fmcFfFl{7IBzFWOtG$S$LZ6
ze+V1}404-}Rn<{C2$z5ppX@V5J{&(Bjy(NITNP|BjA^T>Qcg#5?@TeXaI8p2r*I~G
z6S%MwT5O#%pTHI3ix`yb2-!8Dz?X-R0S{ODsNb|b8)$3hnPQKgnEHbmnhKWDUR89N
zZ?uEp!C>urhnGIHZ`XrF&KJIW277vrmn$0Y{-`I@6Bxi9zzY-n7(T~HvMP=yC>9_#
zjK$|5Sg1e?M6iyw)nFaxyO#H(r@{Fq{$7%(ymV{?<Vv+-J=Av2_tR^_Kt|3-08GkM
zH_c=MgL#Zf<eB9Ae>9zkKb8Og|BsRJP8?E>k)2VJ?bstLg(zh2q~q9oT|!yMo^hxM
zg+tkUgp*N6IypiPWgHwFTlVkz{BF0eKfulHI<E72zMhZA<NiQACD&E?LN_IFLO8V$
zOe1}}^&<!MP~|D*h@i0NRpW(+{Z=jcHLxFu1Md%Qhou@v^PrA)dQO?1?Tl=>L%mc#
z;9A#d%dX9!x$5$DwT_9lSz^^w#o+UzvjWk`8u>Q>>bz`AE=v8IpYU1uiWbmeFEMYx
zt8sQ;xKDF|I2I)JPx1WV^#+9{rCbU{<)~bQSnQndFKz3B|KjV|L$FjHc}jJ-$z3;<
zYA<+lS5K#Sxv_b+o~{I5Uu^d^m3v%bBv1kNGjQcWs)&BynB)2K^KrU-Z}ZZ&qrWF#
zkL2eC?m%24y<TTixp>OHjw1pwNn?$BZw|f(`%IKO6{?@Ul<6M3YM&+TRd8S#ejari
z-tFWiC`hw586L^l-8(ttLJ8^ng+2Xste!%vpuP~5WlRl;{0w)$J@ecC!a-sJ6?ytG
zok28HzCUq~`i;?AMm+EM>YfeZ(<P>USeVFAI3QnmjTbxr=Y8c3QV|o-r#KttF9(a*
zjph5@9!}{#&H{?vK3H5E(_5$JeKes&Mw#K9Vc7FSzEjg;EU@K$Ug8xv5fIN3%p4?A
zA*v_knBX>k312NaEyma(JE*xLImbp-T^B2SQ~639`hoI2hpv_zTmu=G=bQHr!ekl9
zoq<Gq{RLJbDPzQp5){ac^&eQg&Y4VgXqiGe`-8Y_PN3n_Ih|?Uia#9A*tj_{B8#qH
zmZ15f_#cDGT)6)5H{NJoss{NY0#-Rw@Y=$>6QRtvr***=N=0q^MThI+d(>Mmsr*5)
zTIKMAjr!iF;czbO-8iNk3tq<O-w*VBxKtr7L?x_}FELy)lmc8ZJ--$-rQ{A#b>U&;
zbfJDBo$WSsc6Mgt-RT<=Fjr(#Yl&uH4euy`vHFe&HodNvd0)Am^*YkF#u5aYAK7?!
zZDC8}eG7bO9SzW|q{x!e1nNxNaJ+RIP8rdGALb-SdXL0rhJ%wbt8R34W3G7$l|$ot
zrPr{Yd5U!&Q1-*xfb+{;o-a<YUk-w@xfLE}6Dfg)oO$KquzLGieIYGcBfP6s!8^ZJ
zYdR{hkpOO7%~7=%=I!lW|K8^!dUXm6#fl|WTc}NTm^`G3*83oq`YIcwj1<-DH6z(y
z_Cq?Z@Tfg5oti1alvk&idR(;u{c0Ws%JkSN{75peGjh+z_@)ZaVO7=jp{I6D9umy^
z`ozZX&%_yfv8IFHX{pHQ#br7xb_0|5<~ED5rlrax9)7t-(_vd%W8EH=D%%NE-P_+r
zJ%uG+J95Vx!5R_AmQk2zRVxFsyEP9d!#bID7<yjzIpM3fBjB}!ep*olo_Akc1I;%K
z-!|9*HA%wP^xlejKK!{z+?U*uaP(9C?fr-&g;{e`>YL}6{Mg!aEajw1<Jf`Haklv<
z7EGFp#(d%6uyo(LTQyww*njBbbA5f5cA0<48t3huGkxnr>h6DhWH>22k7@A91NH7;
zjp)hQX*fvxpVj$v9}mAPI3CD>uL_V_fDw!V#m4brwq61;b}082fy{TZ`DNIO95&M$
zOn`no<GC0O*?4mHGNIFex`y$_rr~Yi;o_gRqxQm|Yf+Qk@7#CM!7X$z=9W0U7_=Rz
zzIiLK1z)P(-(<Dz)7jj~inet6dq(fVMB}RC(}@qEIE?S~J(+l3ToCVBnn}3^<~*o@
z=y(=BHddC)kP4NuWDBXK9QEHsarb(@$b<0O&V4DP`4vAU9%jyr>>o1IH?GmXl3)7w
zFT_23d-lBf{F44b-ueEh60G7k>2*|Vd!t**f3QgcoTMaV&fy{W+$LFlq)#eS*J|84
zl(mMN^4-L8q-JH%k^f&GJ<AK4QhC25;Pb~CGCxBc)L^0LVBrF*ejUL$uS|MlZFdDE
zL~q_>a^Z9dEpA3uWwU927m*D1PxG=}x5w<&QU^e|Z(*A0&~9^~qusmTFIIlH@<~;q
zj~x8;sSR&eQB8&I<Npf3ES$hf|L)tXUgcSFJ^mq;tJ18`wb{nb{;ZFMxFhP?0r|<h
zS)>HnaSt4)5bw*kjl3DpY37+imzy`c?EavIO60M1#WS;M5vr`x!jAlRl!#n2KnQ?9
z*s>4o;PJvn^%9r*9x(jLkftL%bM%zsQ4QN&o(Y~Mw{3Wz9o2%&`_pB8jh66Z!fAN$
z#_zVqb~UH3h|(PSscF5&5x+nnD$sat%w{+D^7CLFRmTrZ)U4S122xjxICMBa5e~J<
zX56ANV)559dGqyMQA7M7;LGfH^><+5uZ5zKUBK}0_m@sTe#(a&FJ^t|>kSDG<{R--
z>;64CV|T?(Xp!&~rTVcAQ6Rz;6J=gIER*_oW_=yp{nmNS3nynWK|YHC9@7dVA0$T3
z`Z->*mK<o!p)YGQ;HGe5veUC_A1x@e${lcx+Es#5bGNK(Pl=`!VqYZoNhj*FkNuQr
z-bE%lVu2H{DD_Cmez}!oUZv!ZE#Z41!b)6MoZiJDlg+QL)7@t*se7Enzt)Bgl~@gH
z>8vP{C<yZI(El9vp5|2qKsnZEZD~2;IR6VAHO`*wReAV^h57obBc_fHlNW2dgULR}
zgvrAjv(utalE-u6zdz<A9oE^}IBKz18q;ej6k&dh)rh`!JNAWtth^X6l$NQI^x~IW
zdrw=SG|R$wsx)jnm?39TKbou6pVYBYxLq(o48zMjtg$?IMUX@Rjl<3D>SR~5uLOLS
z*ZCZ#v&d#r_|AWEOy-uC^!qBG)zs7+r7Kkhid}~o8fR^#279(iaOU~uVk~*3y&lmx
zeijA}_7uAh4nI*+MxhyR={M(8LkRwJHCOZ^eVYpz$48tzyFVVEBT%ir6|En(&pl?p
z*fTOZO?=>2>^!dmuG82%&3DA?bw^ZwyK^+Y(zR+qH$j33(DD<%2iixfosksi_%%ri
zn#PnTmtSf@Y14Ss>IlUk41DxSXfYQiSno1vt+!WQcAtLsocJ2lN&QrLJ9ls@9tFg%
zaM9E(vAePS*@BhirG>+-XG*8j-DhbI;pdjC18XVm2gZPjL&RadJL)8iK<umsYpwOT
z2D9GW?V6CgLjRgH>DQY2<9?jPwGB&iw-k-YAnWFoW*^G_`j%rhZHDQsG$ldhO~Nnx
zhPOFOAisiFPLMv0&=Ob%Dn>yR5aoXF(neS)ziseVrzlg0*|PV>m}O7fw{NTRUWAzX
zuPRetiNNtQ%BSEPkWHM8i@E!1+r{CU>PHiy^MkRpUsTh{dqz6E9s344{o)J6aKbBN
zZl2WO;C&X+cKY`4_`Wf`REyXMAxjkdE`6ALx_*gMO2>KmTF*%DS8UVUFPkTgGq38)
zo1GV8(y*nTQV;ZmXgCMqso~5GJN&fX307oC1d!_n*0Eb>_A{MCNFxeYCx`Hh&ie~n
zc~%AdEt;EqVS9hq8lDZe-HJ)L9HIp?$iRMr3MMhUV911{*z|Z%P6_A?R4Mkcy2H=P
z!rF=97XJJ1m^gjCQGxb|Zz@On%#Rg00P?qq(c7e|z@f*ER3%dBX0%!@jg7Zxy{ANs
z0e|H04WbqZbFb*E!&wOuYK!V5I(Gg)OU<7*+sw^e*Qv$!R*qTyz>T|^O~_KQZ6ZyJ
z78}?^q89$CQT~T@MMg{i$X3Y7?usv)DB6GXqpjq|?=Jkfe24;1TATA*^CFPMG}I~-
zMcU6$DEHkM7!ZHU{7!SzgN^pNlpgTK?2WXkfeJVU3tylq3oOQB!{{SGAz+G0-zINF
z*=VQj8TKF;@-GB_bmV=uU-$m}K;wAg&~bNq_Frs`eCZamM7dbs?Nk$yLofd2ziR^<
zxQ({&)3NI{RSmL-Aq$+L>29X(hq$v|&6fiDu1|%9+p3rhIpgeRd<BG1jX*Cvz!r-&
z8(#hv+!fSNe+m*rEp^SWxVd2VEn?#jJwr)^(#hSVPfHpn$1Hcc7JI;T6Nprbp`Vjx
zh#BxydxfHkumeA>VpYVBs&_%|pq<fe%$~2Zz*Tm)i?DZ-J@1=7SXO%MNW>RvkzeYy
znriR==SQ;T!$!^rZRW565k@WjXTSd4u-MtoJ9oMkFJ10^#rwZ6mIb?olnr^G{+cgM
zg&>jbjzhQKj4C%rN!sd0TMcj<@{x6*j+_&9$`9#2)K*DL@F)NWvBJ<|U~7Qmwta^q
zs;p1zp}BU_r1O67LD%sqUqqS^2$B6zE;eCVzvAkC+(!X4Q1jqIdF^@UDX#V@2H4*+
zxmqtfoqJ9JLU@Pxm;Nmi#yV}<wZziW(nfDD1<ni_JellKI#@X5^iO5g&A1pdx8`bJ
zGof)hSwC*LSn3Lgb?jlM^*%_Y$2@N~^z*7FSj!q~dp3Mm`&^!)w@qu$mZtmhk$I<N
zxn=!nc&x)?)=i*59)VhE_KD_>ud7wq-rSi|>>g{`3&h9xbElOWiC)4^s+;z!d?Z%k
znlQR%C`G47qa1<kjmy{W@Vn~7p)Tj~e)>?fV9StdfSkVaqK50TUTA?)W<*1M_{YY^
zLXpvtk{#w~m^9rodW>MhYAA?J<m~pf3nc9O?sqjm+6KCSZU)@58M#1n^S~LH%Wu7$
z1u9L3jPh66KVZ|`=^l3A+5ZSkX@L6r!A5Pz&LNnfJKNr=_+lbCI%;BEoU8k`JxG<;
z&-pgac)Dg|*Req!%G9SBf^x}y!0*QM!h@^K3aW<-{aKJeEs(g=VtRkj+^P$%VwV>P
z6>)+Btc)a2jotO`h3?I!?LQ$G#Rq?cwc3iiC6y;?YcW+hUqii+$5N-K7b&zK(`&I?
zUF=unK6C!`o(%GN1xxrD8VKg*Vc(Gx#SZph4_h%QK`d^$TTKfJkGA;G8Ec2x^HEi3
za}!22K4+hk^Y6NGQ~VAkvMP|5h1=Zz)>Ru|zB<2{+2s@A!35IA>ic+DP$<b!o9XWZ
zTm{~Vi}88mYQqlxP&w7*rIzyGg`L1@CHkroj14BoCflF#qH%S5cOXDTy%Q~S^j9&8
ze84c%7f(TVz)!bF+>PwGXY}sAE3!1U|GJd*=RMkCeS?@>KjR3kT+}tKZK^}jTU7zb
z3eWr}SO~bwQAzHx@;0UwzGlt)k}At+#4TNxSHkkT%ZcN;u_9Mf@r3yn$N7>2NLmnN
zSm5WYy;>zaqH((L7c?kT*ni9)^_rE}F2VocGFd37Y$`uqNduHSmHAS!9tH0_tj&$s
zMPjMM%Nm*jXQ;6EpCrKTe)>Y(tbqTP(q5|t;_58(Qwd(wbw$CQ)}7r5Me7Pn#Hmrv
ztVE$@!nk6}dHHVl=TE)lbf}K&1dW=G9F{@A%lcpKPu3PMVKr(O&&z!$N%ncNtG&5r
z`5_Nc2@l7AwwdGd3q^zHCIi!`Z&QiMVX32o-zckbrRkM!EiMim9)fp+T)-@ZLf$KY
z`?}`-@>EO9$`{yKt{f0#(sBG*@}WAgdoN}7uxR!;QuBUh!w2&E`XAsV7_gsKxD;Y(
zDIi)Qp9gu9mb#C_DJ9W8kR(h!ZwaKV_wsvJ9Ua!sk00s^(?0$^GjlH59hQ7Akx28{
zL?WFHWZzSV5x+c>p{3Y)11o(;k1m-F!|d8|T<g^0k17cD1Ao>XVtg<4KvM|BRtY1~
zth<@QkO#SLuc@D6)=7U(H@4ZTcZk#m;+RIgLf({#sc~~|1J(;~=Q_mhVQj`*i+PvR
zIqw8seDa~ePG<{0S$$XVLk=ATn@?c&krkdp$qGZ{u<wnVQ|bwJK=HRmXRUN2k?=zw
z@@~YFTWouhe9PCuNDXFa+1rE9*3ow$?poaQMM7BE9TG-f@8+T!6+E+~vg(ToFR)_+
z>`Woj;3VG3XRXzRA_>$k_pL0@{*L_#u#!eSaCKplxqxXVf-rpWnOTxZqF|*E`ZfLY
zgwKwR))AyT4nPk1m}TP(ESPcPSBk!~X;bP8jYD7L6hTS57j4M-rQ*W%gkxN#eiNs;
zjYV!t*6!A6j&mZZdS0ES8|eTh|9zi*xU5(;!a4K-v1&kPhnI2%ao_b-D}C`Sv0~sl
z+6l(yONxN%*-dcbtr6t5SEC$mpEk0tLbp~|D_5QYdJCQ0??TOU?JQ-naZq2L+6l54
z*NyRE&0O;rkf|f#p`=PyZCZ4loKW0=vpS4vq)t^SbjxI)nNQ>RM@UBtTGxqxu>9_u
zwkMtTWo(og^(jtW-A{s<$>Vj10##ls?5gvp$2FZ_#-whqddUp8y5->)+kS}Nej;ex
z?q+8WT#A{VnZ(ttb)NREc%BZ4zGg2o{F1fhIuXur$10#o_&(Mq3p$DCxQ#89yH^)D
zmVV#FJy~LihBAzeV;Y%i#NMQj9n1H}!K83hW1+%BRQaO!1b1t82D608C}!=q*_C8E
z@s`MgvGGGN4bxd9@`s);!sp}vd}OgJo_oMlX)sJ(sns@Y0$rqmXfOA$cFlbY_Q%<O
zBPAaOs1DQ5uPq$9U>g0UU?|YsI~MM6+=#*0Arqu43oZWRH1umTa4vAL{fX>_{C)yS
zW`-OOlO;}6b+=#QLLTD7UFY3OPlxvmwa&vK0R+)Dq{7R*Z~wlvu-H?bdY}Ei1YD~p
zUA&C+vBBo=-FBi4UZ0qHt^F77HEqE;Urj~17P8I#MUG@g@r`)6e0;+Mm4JULXNdo4
zJuPR~&A12L^O1aq`ukp1`OToR6!8Wo@vUjtOdy&8+P&91+jZP{(6zPa{nQQ3efF=E
zuk-w@wn#nvWT8IoI`x|?nM(Bi1Aq_QtjLDr_G*omNQeBFQR9@zB_mPBOu4N}5c|*?
z8K61%Xkzbmx@@I<!45G!+72+}ttq7r!}0SQ@=E`ly2+a9?q~mVCt22PmBWITWNXT9
zgopGvO8uIizK8z_1Dz>h1MS@}mFCQs#`o4YLT7NHq~6d5=Ekl59tPK9pxOnOMZb}u
z8cqN9mX%R)#)FRPH)5TQ-+#7^;-6YC9zURQ<$f?$>W@<a2~Ifqb?4R%m~WckUAnuO
zDi?KlGNhj*>{ak+LJ%nm!{M`hv>0h(xbv^7UFfUErdf4PA<X*Z<iKpij?HMXy3r4i
zyKQXd@$Ie@L1kml2YQDV!0bh+celVzvS=YNYSH1H2=`z`rdT<ncXRcVy9?my5+u)K
zq?fg>nOTqJy7d{x$(i=%Ry2Mp+szH*mhh4$eJ@&&VO}`p@1w!bG;Q|ZNlacHj{$Lb
zD$U0sA34s{fQQY<o2CC2)*2p9^f_&)@oI#Jo%fOtHZ;)Ro>aI4BrtDzzeOtbnx@?Z
z{;)%0q^hb)iYdno?eF4QVbqggAdy=omek=-m=a6MHtYMdxw`sH6$D`9%2boX>>eSf
zUuJ^RS68>JFPJo@6vLH08>=dB@TI6}8pNap69(9zBXx{;i|}WaAh@Sp4L4|he4JJH
z3D*iw!aN^71P1DOQB5?sT1j`0V*Yu@YQkehWAX0yH4RO+A3-e5=1eoJRjXLXw=Fni
z_IXH<((wZR_LGrGc(T{*)GO4((!GeOfXat+FIN~C8h1oV$)K$&MH7Z-aAJP{&~E0v
z*xn$TUQg$$lmXpfgn2VoHl8Ve8I|)x<-3za3ZASBWROBTEmYdu!4Akh2jFN$^n4L_
z7S`yDnrREHLXL;EF3{rdskJ|j*eqh<?b<SCU(~hACSSQTswwniaWT~DDmqz_7{dO7
zvqh2#tP)S&+LshkOc1I_sY!OC9tfNK`EX7xv@BStr?_<)L*~4tIADpGDLmO-y1J1=
zewh#WJ|uWM+5&1i>j>anWkNG$=Ho{PPV9cxr!x(Q8HFQ#J0Z5O$e(+K<!m)Zl8Iwg
zX{$TdIcCfqk>U^I3N4>@uAk@3EWh(i6AH)O6*p(sh0;O>VdOid9S36YJZfmE6fK0*
zVfJ5seNS~G3eLt>@=Rr>$(SsaUbL^G0v35}1Bbxb=0Q|*Rb`MGf_y<{4uu|T!=wc@
zr7|CMhn%csnW*0pQB{arMk`>L1$~d7QV4&ZD_V$O@!9AMCJsxjwJ~QTEQ_?Vv%-);
z-&4BVcY0K%MulViWt>|GHR5@sQQ(y}5EwY_{z%7EjyLUS(i2Ln*t~tcWCmO||B(;p
zqY4X_M_Rin&!^~<^XqyV4wh3~aQ>q@vj!m|%6%}YxzAdjcJEQHhCi=)tfTkrrYoeZ
zqKhBbF*1hU9jj)lDG)<-RP_4D@FFeEXS_#4)}_BDO6yR<++{a^-7%~=P$#qGqNH8I
zR^JI>sT?~cqyF`!@oI!-x>*$fuE)pmI@Hm7pQ*9=-E`F`X%H>;<;xde_M?=8CVI$t
zAo$EcLa9}zYn3w+eS5fVexYdPmPWNdLAjU4`M32%a8G|If0^%_Y(FBs6T9wRsGbqt
z%v;#_w2)^1U@i}Ec*{qkaTcassi_*NtPgXqljr(X>E+M?pPu9*1OoNedA?|V7hS6%
zDQgJ=I?JDzwXn$u4;DUDxmjiRx6YMFs%RS28z63#(PAdZd#;EPYi>Q(klSUR<Tew-
zg}X59*i$K#^M{mIw{oA)>43EmHiW<#11-Zw;wRU9Ox}QvH^?6gx_XcGr6AOxxI|NO
zN%p(m&C3@|tt`lafIRG?WD9<O^;wr~&DcY$DvSElQXA3R#kWWHp33nfy!h5GdX!3?
z45dZw7xA6RrLXuLNEy|at)gU|<QU@qZSR;geQ9p|gP-*CK32bQvM`w}_AOmMusv!1
zy;*Mj5S&+CL+zmdAUH^gkzP|66<gyGo-Dd9@%X0h*9%IcEdc-6mc+e=jqQkZmcL*n
zLjajLKdLT|Z9-lz`(89T3C7s?PQeB=v7Zs2LV~v4c)IQ7ue<n!26EDtnEsb>ZG)2M
z7U(OTQ`d$zU-{m7NGNo(Vy=H`l}#l|e+3Q?J&|Q5a3!Ny`?^9Jt*PDw8t*;_CT1p$
zdPF?0avoVe0Wh08_}0Ql@Su6~Xl}Dhf->5Ff4?F}gQvEw;ZRL1ox_Yl`k$M(PL@14
zWp;aXx(RSb;`GFCPi@HeFX%l@iIB9;##4Fav;>@&lvw<C{0&b2>3{Ld{0!cz_E1)6
zg1`86!`?(#hA{K#yS%L?<jrwL@fT22{?11F@MG_8;4nx`SjY>FELb`K&RLp8=2`MU
zXdqMBgr15CPH9|0-V}yy4#qY&0i~W(#vU%!!3v3eaOSc(J#j|~cD4?BEcEb6JWZxB
z7390O-=p74I^!d`4Fr_`^Y_*Cltdr+{iKr6lOYqpN%tcGUEJk^qtoIq>rrPJRsy>x
zhk@3pmKaE+Iw}w+yM&%URhS7xnqJu%-}X~*Fb_nab(vH1oeaLh>-bf}Mj<eIc8s%Q
zuMCr1t%?QVjfC7StM$&{1Zf!-+^nO{+0?x&3cJ=U%q`D_uWj)4gqat0!R=~u3Cc5p
z);H6P2cw~3A)Fn-L~bOEnxvZKOj39P^q~DQo&Nl^Bs5wUf9U+l-rcIHy<QZ&IcAtm
z!6{UyHy=?`en%*%h!7|sWSWA6(H27GXiL96YmErnK3QF@L$s#>M;@@<L^CMuTLSi}
zkPd&GjxfoSCHcj3PZcC)UD>GrMLH^jz;%JueQoOf->@&|x-W`(qU*8B<G_j(YK+_|
z(cMw-;_m0T$F(=d-?xdC4E?g=zB~C7$>vTDG@Y;mz0%G(6>sS^V9cCjeyy&9oX!-c
z7ZBfq2IO>Q4L(U4^tIB;lx1%`8oU7Sw!M-(;0y$N+x^s!k2lb&3e0?Gjlk)e=TwmJ
zL-6S-oS=Y`D1O|y*=8Q}|HQe?gjo*3fAL=Gs0%F)ch>#%i^$7=$8qVIZ>qX#m+G`e
z+Z8=4_bTDU+#Ox?0`Yj=lvmST8BxTCq-ca3;~c<S09jPjroCNT1b5uDf1w~&rB)5t
z|GP)eQmhtvv>8>jlm$9VqW<+B<l4*&8>>3a$bTBltx~=(wEi5}8Wh@7?6ofJrzz!L
z{BQi3mbPCavl0C=Oe<Y@ETEC=61~yHRddD80!#CD$yFQu=9F=x8Q0j8`qyWJQU=X*
zqxPy<X!E_PuzB>YDrNpI>l}~b+N(S5!Cs=VMI3oMBhKTH*Eu(G#wJ&`#>Tp2lMpSc
z;RUpRDj&;07UGzIR))`FKSpxKMbh6kGk-t0(&Rq1>)~7*BX#QBzOq>4p9|?>bY=3L
zx41Boc!eN%IboSoFdiY0O$lfg{~L#$?E1G_LjxG4ZxXD?og}mUAS)&wL4FrAXkO+$
zDu~W6@paA+8pn*w(p`)TPE4&-WYumx?xSv_Uzc0Pl;crADB?I|(f(^fLch}dLeEXa
z^PQ-heuSW+AJ$;~pM0dOLFZPDERmnZ)i56We`bQx-lw}Nv;T@tJ83xeP$83G-szMQ
zbzF#SP3hvk^f&tBLC?Cz9hdv~0H*}U8L<@9m97bMnI|b!rHopJPAIM4YeA`|Yy*TT
zVgSGdJ)U_)t~sb)esn8N)0rt<R-ChWs)0zWU2SJRF@C$6ff(DkbhJ1QvZ*~*OT@tJ
zgeg06`ac2tvzl2H);paC6Ns1N;K&1x*IAy_gZoo={A7&me{tDn%f`Wm!NV|?sj!O)
z+Ar^F%Ph$}5!B|Tz8oo4&`fx`UcE;+0Fbr*ggr(RHTkQmf>0S!n-r4{l#Ro%s;)BO
z?T#8U9yR95nb)u6PvC51y!!=l3I^;)U^JTUMFcfEh6JAHoa+Sg=l3LEc2-|RXZrK(
z5)UrHOtW$iE?O!`fZasXdRV7@Z5}rj^&LtlGzK=Z-5b#*M%b&b80DnC<ER6~T*Nd=
z>kU3giWcB3RU&D@`hC%EjHJ)7JN<a!YfCg$DFh7R$Qu6z-Ny~x=DWFT^l<OnJ(jwQ
zd|`20KGi8M>{wsR8bckCvb8oGtWboCLy)PjsjA4|Y_Rypn{1GpfQ3lkjDwUWm2jr*
zEcy*8nPk4Bz>`XT6?uAK<^lKyOKdJxbfIxYyUn!S#J9W|u~G9;wmbXtY3eajgk-bk
zfm0iONR(Z0)@dJ!^w3nH4PSa6wr(`Nwzk0_PB3b3R13oG1^+#p2?z6ttR$IHf7+?U
ztkLcWHuq4>*DTWLY-oU@!4Fkbyh{Ag=10$OM1re=9Wo7@yeN|h^wm`lkN;ghyQa50
z!xt3*{zNpkT{hnS^KWi>`QN|8|8Zz?7jLA#3YHkrn|!)#_x(mzxc~<><Q+r61l6V#
zYP!a6xmw=hjs}S-rbkJUNK{q}jXL^!7@@Gf%9T+gNwUeZ6D%0DBxorRpI7}|*|`Y$
zSAV|a(0NW&>5BlqH$4llL*$58;*=%=lJ2%gYzf|%scP#S(lH<3s+Udu78WGKwXid3
zE|!tO&l;o8Y2bR_a47Ky{6-S>8_N%27)9Hms0HHr?-6J3o`bGLItWh5Ct#VHZLWhr
zv}+gfe3U+z4i}Zw(HtbVDPWa#jI&)mVDv_KxXT#T3nw6rLjV%8Js08qD^3%oG)+%t
zkfg?45H&U%l*%+zAvF*ln=cjExnO*WZ-G{qnMdVkFZR~BD4qKJ#?gZbS^R8qOdY|H
zVwj=$wHRsZX1jA$LbUvotf{BBcDtAzi@e5k@gw-Q$1)X5$%_>Z5mEam|F+yU?Cjn#
zp@J<`NB}2!bib0EmDWOb5oxHyNjVQcM%;3*MovkLC}ZHP_N68X_GFy(O<GQu#H@N;
z3wddZTZe@xO^f3RNt;XLPld}=b5AzSfG=u7p;P-omq$TW)lFQeRc>Sl23!l^`F`4a
zd~z@z!!#l%jNI#4x{qfn!FWNSjX64l|1W^DaY22W3L|S-xvSFoa*24bU+?=PHu~~z
z=4DoC5DZOtHm>B-IsM*$ov!$RdzVgClnvc;3{QNQ?ZgufLGCU+jg4EtDDsAJ2>lYq
z6Z3uG6?-ZFRe#BXxzI-A!m+d@p7hl=-8rVa3=~^$s{X_bc2I`YD*j^1Zs37g0mSXA
z#;1<Ig?xBu$D<osK#Em0@+*C0{?oduIKKz>5R9oNCPp=Ue|@V&?CW-bDsQ>RjWVVf
zA5|_J7aH-AXP8IrcMZv$Rp(Nj>uzsJ=)IGqsL%LDf;GoxkJYWg`ULH}pjQ7nYQLWT
zfDcI!c<2QC{9o7~+lP7D3FRJOKhMXedP3Tl$?*UMt}-kP_lw`?O`7Sx@qc3~&i@Hy
z3bSfA=pit(A@i7l01Hdx2#VoOd4O1~Q!y#FgDN!YEFhmQ+5AtUrJXD|{HHi4gdgmq
zZ{3#Un4ZtLP=$%|4f`+C`;ys^st@h%{{U%GVKOkLf;K#sZ>H;XAdAEdfR-M<Csv<h
zJ=<ysGs(C4G>UEw!})%<HmVHcOH2FkSy?M<&hCg`*>2D{o9RT|Ph#I9OKQj)zr%f^
zqAu8miS$L=qwSQ~RgU$oWVQy?HjNLU&y!ulLbp_z69tsqk8eM#6(?KXduTZjrprX@
z!ZI6HDtQkt?(S)4;oW@*5Gr0rO$A;XdnhQ}-+jEPvgfzocrXj>E(ey{u@AUeO^y2j
z=MS(e2Vj}hwC|%^wYm5m2_wGWY7yIqX;VQ>Evj9>`Pu(=N<Kb1!g@B-iw(*1Fjy1v
z;#@~_|KJ}wjaH8!FYG%-m(fcXVs^U9bah5EU{;v+nw^ku9+SXF1eCQbTQFTa!7S(W
zO(MO_f3cqFVWdncDZ|S$l(*(D;~*4_bix8v92Bc0%%)8XPLIO9od@@QbQaJTvmp>x
zT3a0Hp<@EmnC#Dun>@v})$w`S82c5m>n?+|qP<GfU*kbBzI(wGO0CKBg#r4)iD#fW
z%}d^BXJKo;4XB}wS50Pat`Dc5=ZiY1g^!s4+yA3ymcp5WnwxDc4a9r6m$VS;8GC>2
z^_IWxJ?(96BEwh({r>jrW?m51nFu6uFYmB+QpnZ*HDHQ;5Na>`fGRqj;khghwqLW&
z#N!fYfc_JbUu0){_Bja>9SKZL^GaO3NQs<o>TCpV*(aAcT*LmoS&(=p_R+p}6nEQ>
zPv)0sKnOif1<%6PBfG(w!Srsfuh$3g25-Kg3q7co?t!5aqF2S4J29S;;(-qSF0@7r
zCv{$Y#}xYpStylVi5bP2WJz4{7l+ZfL(uN_o6#4kmj|ZBgjuggc5ci*nGHZ-So4`m
zB!CgrICA=N1x;o`^~0mH(^v_ks>;<<Bur=*HY{aQulkLelU64Exo}AnvZ`{bL7%{G
z?vdipe0OjwBw2{<Q`?6I3j&u*06xHc{Ha@YAf&K_OV_Ewy2PkXJJ(H{Hnf<su`Xv+
z7JxvHS#u*%^1CAp<t1W2tPgGwDl2=q0Lmwi8!oo|@0_}Py<|8A>K6T0)KW|1{Jm1h
zi!?+P@iH?*aUojPxBB_tmb4vtz7VG1nZ7(TC!KOeMxb6yoSr6M86O!(=fhQ0M*c1k
zil#{%qbkV4%Bkr|g7s`?;J?4lS9#YMdz0$Wx0Bb4B3kn=(YJa{mxu+uv_iW9yxBB;
z;H$OLaf9yD&;?)CoFYtB7y@zeIh<6rxBARf4)e((uj=o{SZ`k1YxQ>Yt=cCl(>9+N
zemo$yhr3G^tA-xge(;jxLd@twyAPKNowB30{`uO~R9IS?X5LQTIP23lNj2FIYK8|N
z(@$ge3CcRDJ>f<&B-e%&)K)Q2(6mD|=*9TYp|k61$X&LpPcF3uAwxf^!9%w{`1>5=
zdcW`isxYRmh!ZQTTaBwh&7!zpTD%k_QZ2+sG4t!I+>0No1$VKndavBn_!g|pB{+W)
ztHDWs-b7s$IGZe`4_|C+QRM)Y4DS;@;0REQSl>$?Idly8DVNiT1z;}{h*tC0c0KSD
z;+Ib~8*eY46_`w#sALKEB1Fobr_{df-sb}c45WtWYK5*fSW<#%Ttc}*H#L`H*EEjD
z52E%Elu#G*2}ZWchZ;x!+`*^x+PLCW3XDl$eOd1N>uED!hmLt|waWrPgPQ@ajBw4l
z<Pie(2c0PvKmZy8D;hT(eqEyfIlBz;sejKyooDt^0BpzGD0*mU<<jXt*FoRh-V9KP
zouGkcA=E27h+wSeLiZHK>!+lqk@Fd8^aK7b>-EWu63P05@rPK|^yi!7+J2I1Le2zh
zTJNIuU|vduFbxU4l}gqBe9e90bS)kBy!kCZu0`7J959s?VA`7pwdC*Iu?g7FwcgA0
zQz7mt^rNs{Ak}|0bbihX>?7SX2RtyxR~O9ZY7WF22tjti;M%Oo?7*<NeJA?;bkltQ
z3n&`aLe!<R8cgt&jAIg7h|$SserZ#xn(6xJAtq0+QP=>=A}oqcG2>NW>`bTzm|I>A
z#9Ri@7X|5!4gOTKn#nlhtQMKV$mk|qxfN<TWz2lsN3uk0_{UeI4={%odNWstRQ?+w
z$Ey*x`#?M}G<0iW$i-bv^go(U@{_n7)Kg--8O)`~*Gx)~8DcG7Mmx_0iQ$Ja$D0}c
zeb+!2*W{5R5B4!3;5fWQst_*m;@=pwKQPdS{swyXBRX$nn`s+h`B5=qNq^ynnPMY9
z;KoY6n{QICf+S_PN7GHGR3)?dfYw|3-DwpYH!DFwnK|k^lJxV5fi>?7Dl6R*k_VKH
z9=`l+eZ25*$We-uSs(Q&Ha2$g@86Xzj?=eE($YN4v@Azbm$APH1Pe{EG9TOvZ9oRQ
zv=ds)626sf_8;Vj@V6hol@xiIEUU(!wp`a%|GKjnp<oPF2p?b*x3tnlUIV^s27>>d
zb#|b0oA+5VrM7X58I?5|sx57826ffnYaZuziIs?5%JR)!5@hN;2(+VYY}iW>%qtjv
zZfpb<vs6|djXMDAqQHUOgo1b19smaMD#%~+UPqRw4MlsieA!u9oc8X{A(f7Hg`fwt
zirojE{du04P}%lj*jWxIIHVNJ778=Owa!t65H4H8=IszQlJ2I<^Dxbi{On)rYsl}i
z-K?6B?ZvMn&bGc7S7o`IH~^=O{3u}alm!G|hRLswWVuEK%LKLIkt&9OTVt`hMktyk
zQ;zS7n;Iue?URuj{75AgOI-1xMc~LE$H$mLDW5($zb2RE(NSDe@c!Axu|`hJi0y2v
ze>^PWWb2DZ?)#=jB1FcWs%GWCQ(c`-k6ZRt|0$8whFdfAqo?^{I);el4~%L9d2m+o
z(&?u4w>Pxua3qf-jq^0%ystDonCu+r8#GmIVe#^H<WVU8y<8$Ol=NEO=osXdg+}u;
zr5v7fhS!RW)RFIdBD)WKC(0Dv>h4>Xc50P6G00(;wRp@f3tZSMVOY`azVr6{4MQrg
z-jc4_Q$gViAJXOe1axEW#4OVlTxzJ&voy~eGW4!d@(Nd5`>G=ZXy4h59QsP$Ci;)s
zg|B-@Vzp31Y@%^c=bpI72u=a?k0)K50tVt>b(}Y`r3+{Y_6c#sTv8V+V|yPG8s*%=
zAHDNa|FT}~&u-d3%jipl@KH_<^g?CCVveN(K{0MFyRJ^Gy3v*qjA#(K=Bux9SdB>Q
zS!DJI7(5qy{z3ux<n5Y(*0f+}m@oIp{@Sb0&Z?UreYK;RI9WpviSd9yv+pE1rwv-U
zqo1Ix5xU!>DZ(!kAhKU0H=!9T5-w=z4Y2dGCtsH#?o+>iT0+|!PIq-@GLMrgzap$A
zB1@aa0k1c3?^LJm4uiuEyv<p;EItA`st<X~qsbs}fds^{L6Q<3QL>U{*@la{cVE1_
z@0XKY4ZU$kD|qkcF8f0X_&mBoosr+Pib0TFN-IH|nv+!%r5=*}$)EZ1LPs*k)vDg`
zdv|}3l=*l>Q4gIPs1l^#F<$Ox5D<Q?9s^~hg1<GFZwcMWJt&s$YCryc?1<rr!I1(P
zOc6#51NlJQ9Hvp*CwdvJh<#wBv<VXwcqE=f(&ZwB@HG&iLZtyRu#`$ID|SOhEazKz
z|F6K)5IIhk(enDx*#nqMLb0!P_Wwi(**HOJ6LI%!@i{khzoDXFju-aahEdJ&tf#A!
z_5XIou^}?AJUu;8^jSMQm)#M3)Il%lxY4`|K+?`r)#8aKa4se9eU^h3v*LuDl)|01
zqHpC2a++2>gTJCYLeV!rfRkLVZtn#}tk35Z>3gNaT>1=KZI&poz`2c5ULgvW!6>M>
zql_xnx3^_71^Fb^errf(-}oR3eL~-!1Id5?;lld&l$5=^jo<y>aqbmUto+Z^6=Czf
zz}+$U!Al0c+<2h@UCRBL7ju)_N0{26*K$){H8oT}-Smvpif@(gZ_QU2ekcs~x>ak^
zp*YH7vJ}Y=+gP(jjHY@cYZ`ZsC}$fqanfL1VdvJ{$HJ3;R!#z%7~X&k&LHxXym!F#
z*TLb&`%j`(Kj*X2MVB6^SKd!OTR%Uo1?)S5Bgsok!&ColKQ|rUzv?gfSoLsWZFf0@
zMA#euCQrK4{?WU8Z-$;l<9Lpwfp``f;)A^qMJfu;)w{j6u0|5{Pkl5nT>qtgqGn(@
z5Z}<5B6{(7`{n_A;Q5#HZ4DCwrKZ{|yi?4-+<!?{OPf(lYONMo6LIJmJl4k71?>LC
z*fd{F9v~9^UBb0lMSq0PP8FL;mfcr5Xt><{+3600Wdi<6z&xgzaNQ#AATcgWH2@;J
z=CXWD4Ee%n^lQ~sW|H_j;$(Ad)h8t6<MGOEjj;W-&`n>swf*F8;rf*g{1?KJjfzV6
z@o}PMA46-;CCLcKt>M*YR_1-Zj(}UJ{KIE(o~FoYxl1m)VB`v6P3atWzz5l{hI6_w
z>i^BYuJO9Y>;0SsxA=hLV`66{X`@I5Ia6IzV<-SFqoycyl&ZRbDIGorfjY6HE(dF_
z*F5WNnnpA4eFfZJLb+s>s`bWRKzj1upndI~^;{Bq!%5fl2JtYoES8P}Fn(5#gQ^KE
z2%ZY_Bhf5SAwS}iR}E7lbh2gH&;poN%6`$fQEArl*KiZqoqRY2Db=K?4ULYDjw3uU
z#GLCfK17*mictk;i`YvTe~+kK<Iu*`VlG;4?k`?&LcN<y@;83vs+v`IWBWd6v)%4^
zPj~0ll}CdKAAH~9%aJyKOZZGHgVuaig>$sHHq0qWOe-JqT_=n&I&&$OUx(qp=QcAA
zKf~CyKgM0uewu&_7lm`Di5k8iDEUOVo}+1K$D(e(x(O|l7~N@X3^YtUI0NM}usPD2
z8@0)C02H|yy2&`H_o|R=Ti%e`7B0rI?h{Hu79z?QtCSKUV-Gl^L3G7@X1eXL;B9z?
z03FFQZm_aaVmN2f+yw=8%&u1_iapR1yTjnznIDYod*z4{h(kA971Q(#w9FUwr$WB(
z<(Q+S`F}R9?^KMY{xW4ScJnaQPZFdsDG)9jhsAfW^eT(Pqpv+0@a&pHlR~3DcBe@%
zp<nYBG10#`0HTm?R#FA-brfrDF>0Pz4%?;Bs33()0w}%GpSM!4)DEkztD~gz7wjgS
z;&Suy3fpShuSSc#49Nw50X%KZwiZ*9lC-w`FSN$7PR{--2XZ<$6bDOI58<fZd&OQq
z@W}sV>3)-Y1CwK_aawbYw?=54A}0n)ZKQ_T`%7#F?9Wf|CPJV0JWVI+%l^Fo7%ir#
zXTnkc?OWq*zKFy5gQ$k%2opwJ*#i0!_GaI8-}4dr<5507o(*I*>Ph{MpVU~}#W9~7
zkRBxy@$ziyZ9;1^@o4OPSK}m^a=w|;+;wVZv5?Q8=Tp1hQOsuOwYGM$FnhYNr#NeC
zZGYYl9GsuDQ@Z?JND5AWyN~nE=W#g9DCJ!F_Dj3zDDt9ZEfT3<glW`bU}k16swYoR
z)4PwJ&z~=(NeO`WZWGa)i221uiIFb|m(AqmXT~8S)Tx>b;z`cON5gq6X<eu5&7CZ6
z#gUYmVD+PAp=r^OYqDHW?8?0;s$K`K@p1qW2VJQu_G%=Uq15!$;}}d0m?1DHJ@M=B
zf^o1|&TnN50wFGn3KLZwExtQTtJ|v`_HstCoVVjkCp%s>VKm_U?BF+`3P)<Hi9<{H
z0rBF2Um2UJO2}(NS(mTYUzGY_E=<^b*xDNjS#jr%J@C$bc*x_#PE{5v)7J3B1wh>u
zmv(`@g~qJQ_b0Gg+gL||21@VQR=Zp!Qnhz`+1O*LjiRdU&rfA})5ztF1XgX-fjSt>
znggu@Lr+_?e@|NY+z|UjaK6v9Rrvsw+~)P9NaTOmAZfa|j5O(QR~WHIE-%={W#oQA
zc(0#BNLufms=$mW1J%+Lb9@!RDUi4;Sf<19W5OPxMtr30BkqcFbkTv>Zo=M9STmaE
zf;R`RDF_un^Fv-w%<@OP<1L+drLwqmnD#EaK)=`pTU)C!(P43sd{#rwd9*K<Q6QV`
z1KSHIKi%C>5w{wd7}nl6@MU|#6sLJ(6qXlPmjEq8K~WF>WAMj?y91F4aQkQlF+q?K
z2N5VqUe`lGGW>5aAjwBhWnhn9nRV<bdyrvnZdSw&-<^$%!R_Usfn{rBW2Xi$Kn?kv
z|3DAw7@h7`IcQ$4QyyZK+vYGm^}C0mpga+Z&2qA0hdIy6V^tX5-q=A*bE`B5DJT-~
z%OxIHVJ-rXbB82=QA~ZZZe&ytmw;m&fB)i^tA}bi?Qu`_c;Gagbi27y`ky1lqYxXG
z{J<40$V60ezjghQ`1LZuxPZvlV%>4+{k*BOae<e$!KiX#v(&5mb@$nlrh~d3f7RAx
zV1PC}wbp(#6Qmw}@KjliFIBXD<yGc|YnRekWQQM{x^{Zi17=Dj3row9y0z&ipB_u3
z>7#eDvMz&Sxx}|@u)|JJ)C%FB^U&}z4OmMi1)9P_i$Nwg27`gdY;bShdfC#{)e6XU
z2Cgr;szN1lpBW{-Z^R{ke%8nE99HG`UFFe@5A1yG)aedZZls-~4ubmrM6LHaUs2cD
z68SZ7Vy+HYSzF(9#i_TSJ0c?yaG1GLbNPsp$Xu?XMby?oH&=FOE=l90lQNy)&=Snz
ze6rfhq3CVv-9|U&gAB)b8*|2%7p<>rblauEzSzEj2?@~J{R?+QTpf6?tp_%6@bRui
zo%~CVFH}?<`7@}@fmYK{0rn$;WjH~P_ReBn8jxq{eb&o&wxe-2f4&5EY_)?MM9O(D
zNDeeP-x$bqP*+yBa<Cf8QB_jTObVUwmmKckj|0`s6us4~G(d*e%SmVU2&WP~zcF7M
z^KF7aFjBn8%82FfkJhAb{?v80IcAzW0;LYwpF-5STJ4`5Qz+|vZ1-FL-Rh$`nf?v*
z>#((G2mZiB&e~8U%iq8%it#oR`)Y!;m)EqH+i>Bp8AHRulvV!sS1anJ_0DwG%knT8
z>GU<G)bE74OEjE1N}-}(nqOD4aj;7i__|bIHr9eHfXzF?m!}kNZePS}@d#T+dLMQD
zv1P7WCC-G82EP`HJ1HB`)=v4sIS5>6769`?Zlz^q!QH1gR~gPrFWsyjyGMSUQx$Cx
z`%iD#Q#}@1RZ>k>j%+lepx@I^6~IgvoB;(AJ1aswg~92AOolAfnHX67BzF9Y%U9vo
z+2&3~c=h)ZccQhKdg+Hk|Bf<>qYseNhT@#=b`y*^P8Z~~-A7yJ9+*EC4-ke!Yg-_Y
z9C0L*NIsOBA)LZWO^TFfWylB{^BaC_^&d{?WFsjWI!s%SGi4X^!Ey?~1f`Os__aN_
zF5aQN&o~hGKzn5XjrNy(k#1Ib-@=@;LsriQqhu7dCQ~+KHXQc`C14nxA;!s|^@=?-
z*B@+D!+|5t#a>Ph2qXhY@X-|CzZjpbQ%pVVtO-(eVjkGk2YouMV!2MMFU|-d($A~9
zI3p0pMa|voXgdd+(2k=Puc^ZhbtSdgX$@Bt&1V4!_5r&eGSDv)ft2ITQ3AF@Dv(d?
z+BKs=XvGg#>ao#s(MG)3%f0a&?@piN7aSX@SRn+M^?lJl0`$k33A~Vg$x7QBLV^ft
zg}}v<1m`n3U9K9bIehJ4NpwwwG7CILG>i`^&W2Jj_udN$uuDGP5vB%1V3f;^{F+R&
z@4kn-)0+>$Y$UGg9k^GB!h|H#&2HSR)SaP=lex~Z)4Fmjcsp%LmV#yzhFoVaD6g$F
z6c6?D6BH&N1H}TK7z{`9{kzvgXA5ia(k0M0loY~6akq&;X{^Ku%q1?SF_-^)q`tI_
z5j_OC-5I-Fq$|Df3PlXSD()PfoOlMBy?L6V3t?B)P6(|%4icaxAD2IP#BBui<J}hA
zA$+)>h&JXHeZb@XV&z*3Vj7Rf3tHYLMz*f-ZCv94Fqr!=*GN37%1_m*K(XFQKv)#F
zzq3=!5qlaN3$pX)V<QcJeNzd(*3wm!XJ`?7->^3!&Y;q7uG2lv2B9PrXeWzR8sTF9
znSlKLTS-0Q<o5aT`JTr4?X#yIjrF}I6%YBq{Bl!W7btDaoQ{THD;PM@I4wP2GdWu@
zamPFE6V|8grcXg5#>6za%I&SNu4siQ(kQtkz}i1aDlztk&911%F}XWx_vz6Qw}!P{
zt%ZdsnHwau3Gr;c;R@aC)rL%XT<Q@cp*1J=4aIMSF!;LoA7i}l>{x2Ny8SGVMVLkl
zh&=JIU&tC7rhV_+VTBeSU}j+%e)hg``uFdRcKZjp7u_1%?mtGLnsiN9y67R2I(P}~
zP1KgHZ_i`h01^R%$7dZoMEE*2+`Bvc9Ri^q+<PL&JJS)m10MHuy5dR|`C=bpSfJFj
zJ#oXRy-urNbP;peHprclwc}LiaQ=6|nkC)^fke%5PIW|n*|jTn*F&JB1mSTB5V%Vy
z7o(3?3WiJveU!7!n)y<-KmFfX0v+m;eu9*UOi6Yf?vnlm$d5M5!|1t-1iKQ|tGaLB
zsj4fnR%$Z5fGIs><v)V4KDg28|0=V*S=!o=jWhn=rcF8f`I$BgXGLzbsF(~;psmdC
zSs+Qc0!v$QJy+%>t6<y%Y#lAN)QuNn!+ugc_eJJw5F9(9)6O6+QGAw~)eyFrU9qGV
z=h&f3rT>)`+|<1kd(hJ(#m}BY9|yVIBkW9_4-s@w9O?5}N#N41mHLMfoF{UkGFa?w
z5Y-pJQPsje>d?(_6FaqvDFV2$`JnYVa(BdpS3M$OCkDzDL!S`yc1y~^j_baNkltrp
zO3LjGfGgI`mWh5~b(L24{GY%}N0`8b!i0U)7A|?VP4wMItP&sf8+!zWhCf?4Ata;!
z|NH=!VUKE1_b8r!YwzWhy-x%G!=arQC8;%a)Zy9&|D6f50}pYG*BLaqv$sDbsg8h4
z{l14j=CzIT3L&q5<O0EO&;RA|xmj7T>&6(;<{`)<H|DJml0!$t?3TTP)G_2EY)WCf
zNwG57JXhZ*j)7t{8w{J57-_3n54<OC$7}K<0Hl>RVg3$>e`#GiCWdoR{|uawGcq@?
zpOhH&*ApTEXja=)eM527T{}gFw>8LjY`Ph0fM1JkO3qgk6Y{t59NRyWyPD!*KLQLF
zP(R9x^Im96YcmNv``~JIDMp{5&j(-1RG#5pPr>nT*Od$`PaMB?qYl58VRuE9FYWuv
z%Hh%vcq$gnsx5W>DO}7V4?9x&cXJN*zNIO$ld|^)sLX$WU)lY))^fveh(Xs?&Q|^F
z5(8YD$0<`xj5$>NIyTNx2?j;(Jqb!l25eJBUq@|w?42AI&PKUJBB$Hn-M;gEEE-|p
zty`~jQRBE?<9wr?B5})Tqhk{@3bI-zM}eSY+#YyQ$L1Yqv^U8J-nYn9{iSB1IC2=F
zTnM~25fLIq<W6cFk!Fu;!Pwv4O1y?fr@HX2k=F9Y#s&!QQ3ymjG~<*oDwpR=Sg3pj
z@S1F9))zSb_tn3rNc^|i$D)FGu&IZL`NYEOgw|F9>Bv~9s)%rCycOlBE10OuAy%F!
z!G;d*eF=q9^YJ(`sLSAT_;oF6Fc{!;IbP#;Yp5)>EBYGWIw%W>0z38Hf6%2+U#6S)
z?j`Wwf1F`?uwgg-cox+wi@iVVaz7KQp~RyT4b`4KfV7AuwbhJhfdd+UOy+7krK26J
zp9Doe)tQ+_9CIazH}oy8^7N0KFSB&uyc7?0F}JWlxFfgw`x&ZK$$@-;_5xZcPE4d%
z`MSAgk63A3BQh&yrQrZ2GgRhpY%qSHwzT@6afV^2i!?S)8d=>~%=pSHg+TpgS2DAc
zmUL2Ch?Ctw@8Dt$#M{9`T42elcuiO~M?iDu&{?J!0$PW4P3K&nVPX6qaZYiaj0_K)
z`^RSACj*KBvHy>zvkq(W|GW6;5fXwhLK+FF0n#N>A`O#Jx}-}wZUjV>E<swlK`F@z
z$S4_KKw3&VM~~*Ye}6oGbB$};wfht2ocHTc4yJtv(1946Xvd0pDh;LL$Bo|l?vLUr
z?KGiXJiU=nHpB3BdLKcyU>I98rQCVO{eX_+TuU-ID?U|WG}RKUpnk#9Hp!c=#6FVs
zC(Jr*Q;##zGMh#r#l9DxLo*qYw!}j}%;}IVu9ZOiqQKdq<M<tArqKoAY*_`9H75)V
zC#M6v1M~!uO3@25g@rdd3b${(T}2q;GMnV2r$t*!@eMTCqLu_cy}O@k$8O~0YWKbK
zd&9RT_G9)vXNUSnLx#~1DAgxoT$*Rl$2_XxR9p?s<WMTU$AiIQ^k_QL@f~w#8?BQr
zY3}0wA{q@5l(15Vcvnzt;%^ZyE(bEm+r=%QhLlZL^Pl?_;q9~cx5Y-t2hUQKhSz<h
zm;`kmg|p53!Tkr$w=TNzowHw=m;ncCwR#>aqYP}R2C>g2C1v?&ByvX5B=HW!tN5$$
z4ZbF!YTtNmP3AhKH+b1e6x$vHjS&6(7+$;MrQ4|L@GgIo*S`PR30t_37cKk5>@MaZ
zP*qzSS81JShyIhIXHL%d9RKT6J7W<jEzibFyJ$=8e|7rFN$>Zq$#Q(h1A;m;C_aq_
z`*yxrXJgg)u(G^4x?uiBn|K}d5iF4pb?&GhxzM;cxOEZcPZ|+^fAQLY@3`G%M_g!V
z*><b+Zr`2q5!O8La5B!7uFL-!)$ZSxZ_ulgf$M93yMT1L$9VL>w~(I5g_A2ai;#mD
z9k3~USXPK=rLj-vj%l-Y;by<>e_U_mZL@-jxoN<~tL9^+(2Ii@3gGk*yKsBvVeCn0
zHuopy_O{gGQ`c(Dhyu53n}X$(n|Z?M?~x@`6O(`|ZG^@Mos+J4ThoQDpI?aG+Un-{
zfhG2Q^+)5krZ=)SSA;gv)O?YfJ_*)EP=j*wkgekCxH#nb5K>l3K_#1;ugH(wMw91W
zms)gKUE+P4-E2|QK1xAX!fUQ{Dt<gVU@a{T17N@s(bnp_(S*?Rg@n+P-UuR%wS^yr
zJss`+;>|TL(w=|eLKJPuut{2WOCh6U511iifz(RsR<^3nT^9)Y*#UR+xdnMSnV|ED
z!nL)v>Nu^ivPk~#uNfa^)9Ig?J$|o#k;@DG3i9jae~Y&A#W5zpK)V-aMTBO7k|(C#
z-K{k}8YR|X9D}WBR*P3_KVj?9kA{=ywc<nHQ=7BRf!vLc=nTG4rRDdO19e<h6&n4&
z9<rxPsilr}V%{f5bmcq0icL%&3~<On{!_Dx9hc>ZGrqrnqD9h?8ZPZc%#c{AH#}uT
z8mTKiyG0&5c0$S(N!1No%a#E_S~Tqty3fQPKjZaZ*%~AWX2($@P*N#|1nD0o@I{$~
z3b@$_q2)ihtJoom%J^I<H&7S{*SV!`T7Y~G8rBW9QYr(pF)2S@>3irSV!hMkdZAJd
zymq{vlGe6w6|%JEnyA8e%(+=`Q7|RTs&YC~=aaY6u9=?=?ziu>`uYO$6SELi^n1>6
zQmWusv`RQtPjtzryw?-4I7%&0eP}FUIO)8f)_9YdpVfC&$CA{<WPMz=`(skhRAHK-
zG_k#ub*cDYi<ER2tK5?*w2^@)tLwu#=5xy)l1#DLTjV-A#s6R&5O(~&3`S~l@&pLm
zfJseRj3RHvPVdoJ(GJks=`O-m<LbV@$6`vSt*t#4DGo`sEB}rQiPzDlwiabd<?GXg
z()GO6<Di)$v{C(9!{{uqm-&5(N>3?Fx3#H7PiS3g1<j^m2bpR({%*=}?*-}~Pr7S0
zotA}tvgmI|u}OVZDoUkI)aBR_N#MajB^G(k^Y~LuHGQ#AXC1&%e|_NLXJ;5k)H_h+
zJu14*bP}@Tv`P8B-%+`)=}&`irj8VWbK$AKw@r$JQ;3bt=(bSk&D!$vz<irb<Pog<
z_PBIs&W}CPoKfMGI<HNbSmIBZ=<`N5E(0o}xZi^)`|<X~dm`1P$E{DuDXmqcZm%c+
zpR39FDGMO(h#T;k|BD2WgBNEg(_A~XD<Vwi%~{T?scfH(ao>ZPedI+w(*G*Wh_skO
z$B!lSviCIS$;!6GVrb>oFUtn&x22pPru)J;y1BWz*;=Q}fnNN-();Xkp`AMLN!Rpg
z0(`b;U0W%_lMYPqM*UTu!ud88B@<3N1rNb<*61)O7l2|I&xj1+fM8_9NoVbp+&dEJ
z2HiCZ36!&&$SJ+h*0Id;f#*WEa|*Wy8{pt-1WWMU{{;J<L#*0V?Y#fxwN&Z#Bsw14
z`@T$%<9p9H!g%a{Nv-BQ1GP5{6B`Z3Zo8JURN+0=Xt+M=h`$~k-~gwxOLBL0D>9+)
z!3QAi^<*|N5l<MNl67&AP0szoZ$OD?gh1h{Y>|GsdfV9?8Hxwl$Ek!W@F$b28aP}=
z8U>YM@iE;Ga3GPCW>cQ%T2%t0^>avK0mGOsx4;7gSPLGQ32^G$h@`a@yx`9F5r%OV
z`15TJSc8K>8<Ux~#>&vGR>6A($QsI9-~(yM>~OG|N>V72@5D&;s}W~b<p&F~t5lK9
z^~2RcL0<D#`cl>b(&BH(M9N2>r_$b$v2hL)3Rd_B$_Bd*Cf@!un!=A&)PU2+srGb@
zmyTFNqd9%H-@E(P(JB=Y`zu@Czse!jgre~Ee!{3?KlFf4#vfBg5c_1CEk!2`bfYhN
z3*0g%O=5_bmB0zk67flE4V)DiTp0?=3D4ewDIt8ydz6LP$yv0q68#4I%nW$`Q2IEE
z5t@9iL1_=(*>Uh31Rt7wW-=GLc}-u4UP^6Ig}!sQ^nJEJ-v&a(-`4plLgl37L(iA2
zJ`c3(YLweP_Vjqm1foXe{Q>Ou3nA+$YcdU<SYtEnRnQe%k?`}4j*VES@fW=H1^IOM
zjAZ3xDB$9*;|=p~n}o$rQ1RSDsbe)eu8Vktn0C0Sol|QD9(#r`21l1OwQ1JjOL8*R
zK}tooPHFP;(=;1slRIVD>L1hb6SK<st&42d<nsGC-%uKYvf!;XNB=5Jb5VfSd*_CZ
zg1Of+UMD_Y8?CmoHa<Q-bJEaX&pz~a&Tk&q6LnlMZtAHj%n%ELyXD}5wkoHGU{`wZ
zdDR}%y3^govXH{%8o1LocCe+pi{LQ>y#-P^+8gz)d#6DPw|yuSMd;!3(a|<Khpwdl
zcsYNY_OyuMFj{j?XGF(5=;qY=$FC!{mpux1LwSZ8E?TPCrY48dQs$P0vuOJrMhUQ}
z;Za2P{UsXVs2bOVgtKQJ&XteDJ|;}J)nWZH9V~YS;oqL5zaN9~ayq<z4yR85K!^LX
z6N|oBvLTyfi2Zg<wW6%5J^2p)2qoTaG}#4L#XM};6|p8ceFsxg#($u~3?7^^1929R
z7u(Ju<4`t)ITI7}W=-=XvEUe=67;)jA1wPxGAX7oRm9U>kRygtRWdRWRrK-FvHA(h
z10}556-1UFsSK(P^mf^z5Fm{5ePVejIU9sE5^g8<BaBL9QUo=o1!H65QtfAmfav3B
z{P#UMnKC(6ftbya<sMtf_suGhN!uA(K4|e^zu~|GJu)&fP9~Zouyy<KPwIZQv>z8&
zX9JTumiH68-eJvZPU=|9IcF?{E&QO@I#$>ESrla?Wo`S#q{3^ex4>e^fdt^$n}lS2
zaoe<VGUY>j(72lvsq6sCkS&9TW3T9KY8njTC|l~(1y+a$nzv>960nibGB_e^>6cZ5
zgpHv|iw$|m!dR{G2c$#;*bS7x?w7M%j6%pInN@Px)5k+YAn#(9v9?1)*b5!CBxOUv
zBHwy<L1!WFUt|cNB-Ow*<VbdIVc|LgIJ=T?Utl7O3j1%|VI(15r$Jl@c|GPYw%rOA
zX@6D;-I+vFbl!oLGdab@#j-+#J$ibnsqjYxcG>zU*n@0Rv{WhdX*AqFf$D2293^Pi
z$!jxYY*RpyEgCH>r;q?=jZE>ZqER9qc>fcsf{EL&2H>qbvrSJ>Ma7;$8?RCifpeSU
zRxIL)2l{u(#%RlM-0<-5y`9-!>!;9|QE*V>>RR}haT*uu2OmwBzjT&wM;>~%(>W@l
z+l0}^HuY_X;E+fpNc_*;Fq<-QPI>UhLsni=vTjnX<fv<DH+Vl=N>0j*9?_10E1A@w
z9mWrVG36|egST{S4Xq+iAsyfP#%JHYDZd{+R$YDC=}Z^nt7{+ebS;}()5+%5N*N`Q
zTHi)YP1(PhTPoEI-G*krqVHpy_TR-gM-dU4P-~n=E5#8+2??F&nq|82s)n|L$(vdT
zK8$88SV$yBI#79qRWO_T!S?4rV;QFSs*ilNz-OZ=o=>9<O)?63%yFN=Y9Vlz@zic>
zF0GSq(o_R#LxyvNChd|LCLXba&&(PtNy-Nut=`t=b)L+=a;RUooz0!1un1f`o$pJ)
z+3wijcJH{|o;*9fi6}*$gKhE$D|eOwA`DN8K<nv(+hP|u?FB%vTG!bd*?TcnRZKu}
zYo+_5G3YQ<j+TAuZ*<+M%0q8Fe$f+6*$eG`(<qMS)?`hiz=SBN+{gcxRE<13dVKlJ
zpjm4Qi^Hz$fd+G%zFZwLHe`=@cuV_>S61xIqS=%+JlQF9(Xg*j4oxU-eyo@3V(3oX
zQ-oZ|$@%%NW?JvbZaW{}0OzA0Q}^ZUI0O6W-z(C(PWhPH^#*tTNT-v)GY5=H;N{@1
zTHWaQ^k4P|$|G4L4FTGi%9Xsq8-&}`1B<d6XI?w$9DT4_Wi8wxV1eI>x7Zu~>IKj9
z{n(GcQ$DAh06456#>nIQz}0<015}M0S7eb$q~Q4SvKL-bUC^MhS5y0+A@d@?URclC
zxP_eXFORSNFSzeRN=^p<4g+|svFz+nJj?s~Cdg?1SUUUO1%jP|gpB`+2gWXETa(8K
zR)e7sjf&lZOpBsTh@y1Z)|31pL#EiIM0TN11^hU=dNle_RS+Qxr?qyOn>4K)3<d)c
zuI@qt0x!6zSTpUyh^_ucd3@#VN&JNmwLTFvUa5`?!x1j}F-sYI0Fc>)MZ+a+-cv?f
zKIKhU(qQ`miA)};^qh3GX@Ebv`7dOu)W*Nj_BLMWuHNGMdh*m+=x+-zbPz1yyFOX{
zI~}N(7sT?(Ni6Hh<HwJ6^5kr)jusxLX^q4mS?jLa`i6GO(#NazKt1GhIPk*_Y~+gn
zwAPi)!EDT?U^Y}gTY>k67lW6uk1jqTcGyZat;IFFE~$|rTs7S<zeP;QP_sW&!0G{V
zpT#PE9j`Bru4m5o%z4FWnEVzZh)sU-4Ta2&(7Bq^$SNYzULTZ>Bw*u5xL2sS&u=Jh
z-6o~JM)f#B*~*l1MszAiEip?M?KM-)O=xZV4z~|I#Wfh7X+SBtp#O=YnNUO1Llv~0
zM2&kw`XpYm_>#(;q4Uh(th8#X9=`TN%oAs4AzXyn+=HG2pV1{QLk14e+ehVV0dKt|
zO-(zlx?has-D8%&3>Fd0;7`zDT)0{c!sOkqasyITgEc-=e&(~Yt#S%-vw3LloyT{}
z6rs~T9X)A2z)WLxo1eH}r{qE@jYA~Ao}Pr7R5?HQ@a}(p+@p-8yjD0WED>{ac@!Ud
zY}b@^8zwpV{3^)dsI;^+(9VM7rxNaTV8H<R{PbT*NqMlzA@<n*+jPk7^=&PSxp7D7
z4aWvtm6k)B@nC504B->a#u^SQgxa8tYkk<ek%z$GWAHP(y1UAS9IX!Qk9$d84=mme
zgUJm}hx)B-uctiwZRFHo@NB9%;PcJ`y0m`u%^x4LIWd#Z?<t{5UlP$$T^W`x=0d5d
zvoLKFTwApsy*aUf&pu|QMgL90QM9K@&@NsbwJ?3B>i3jBQv`Gb)PjN0bPxiRWYmB3
zS^&H7V77NE9MeG&D$L}+5meZ>;yLAG<oLS$z0RrNO|Gub#$QWe^>|#R=*V7f+Bkha
zErzjoK4`9QN@Gr3=fymqDsy=ekJ<1o-}?I-uM^O*zb7gZ^W{}JVMAqv<v(T=Il9N0
zY9WNZ>&z0gin?QodGL1hT5Y517)_L?&ITAYU`b!Bz4-V<3AFx~^WIVZQr9<?fc$zB
z2cBo)OJN=mO&gI-l>s+-R^^mM`?rFT(Zmn$eQ<vK_{(4>y67W8R|cm=p+^#zf8)eY
zC=p9I1@tQ=9!i<yBfJYn9X|dE&0~`)g?Tn5NK0D%q=KaQ`?6)3VHBJM{;0I!mj=Kd
zd0z=u8hJmGm5?qcF81jG<u{5^=?zxv1Sl#)L`k^tms5&$=d;h;CH!5S>`?jhKN3{9
zvG-$nx>7AwH{HV;-9=*eF4wmf>gOFO7O%Se`g+pFQwoo_Vm}(weCH>u=ugs>U*aFC
z`o<D`aJ;wYtfYS63*2374+;zMZCnFKwq9PLZgFt*eL<PR_mo<T^ebLz^wq`uRgHzC
z|3ytt%x~lYU#~JFgoeB%`H9)y4@Ry!J3s!ZO#a72Qc6Cs=_(s8Qp3JHDD^$a-&8&6
zopTWK4Q<%K{l{N)WtW;<bKnBKhM)Err{c-A*R_Ns1D1>3<pNKP%dfAm{}NXKoNlf1
zLn7fOibbwkpY0IxzfzTHKcrZ{UTVt8n>1^wijN<8J4c($%Qd1M&mt%1e|qXpQRx&=
zHJELsoIruqCeL#ScN#ak_|Z>D8L=M(T4l0pv^a_$I6u{BDA%;=LDOmEJM8fCQnlo~
z7I|G%GuI~cw+Hsb^*Bi3k~<t&(viPyV5(W{I37DLrO?qio2bJr`CRe9raXau%q?~`
zsJ8QT`^DThHzNJ2x~_q(t*+~j{e1Yo?Oq+u4iS25IcE!xRpJx0-7&9;FvOoW9Jo6Y
z7Qk{os{G@vKQK{-gsYk!7)6ECOCd=%=-2NR5RQ$%*3wr?4wwx-j?{+q4LLj>S+E>#
zJ}$2D?pWte42JaHnQU-$UucbBmENeRaB5qS%Hn;%wpp{1>Zc`9rb^Z2BA%ijuUfdu
znk)RSGW~@-9s;PXtEeb(ta*Jz?ab4W)ZkR|&h--YD8ZO5a#H%{BYe`78^ZnCo~z0A
z6%7IEhlDhEd^xoKramN&;<<Q6sR&8*|INEAbMF)K`Nz9f^?xL-PtyruTLKn1Sflfx
z_PT4IEdU4CVfev1dHRq7Yp`8*I!Va(CY92v(nrt3h6TXH;U0Lh#i9O5Q%9t^B;v-z
zic6>gm88qjNn(%q{+dAt{86p=L{b)dPr|AjA1!tGumx88%r&)QsKV-LAy6*JJJFZ6
zCLZWFd)eSYaS-1@uGn=>(qSX!eG}?}5=Zze8@^2rsks#mIJnM;L#^6E8L7vF(@Pn9
zx}eIU4^U=e7tD;CRKmY4Ws5$N)X4)oRW3ZN2D2w$g?m1-%e7lA04P54ADo^Z`OyV-
z<I}C3T%U~GZM!3n&t43zBn{Rj$5MX^_Y0v@axqOH4!E>+H(YZ#4yYr-Q>m@3_B4@Z
zp(ynxb-`>JBMA6uXZ`73&oxse$<|7{7V!s00E1qu?|4*9lsCqi+Zp53q*g}Tte&*E
zNYw2FvDZ6^@uP;+N;f{JISj3M7U3^5WnXL$q)3?GeKxIVfX7V+w=SjXs-jW`--u3$
zkB}_pBn*1Nu^B&?2a;AV{<+B0Bi8c4+o3U1GADo$N(+won=z1c_aw&@NL-#bEZaf>
zd9owfKuJkw?!q<CM>{En)j5ZtY7xZ+rXBI8XKu|6c8b>ytFvU}->^LUWLXURYFWM4
z_Hw9dv%zOlRMQ3uE$qa^RJdfKPRu;ab@Rja=L5HH>GH1kEbjKudqF8_2wv|`%WcT!
zckoO!UHzoXUi<V&g2L5#*WMgfG6h%wxBh^-5Lf4CF2r;9YOCeufmN-UTqdg5kG0?P
zg=bMK<&4sBuup?;3OHOYbQsa$UMaq4kXMir@`7i_&UJD}dhl-7JR~Em-Y-tY`6@dS
z2e&gh>3I2Ndyf1;@oGgqt;$li8l>1T8lGd7dbq@HQ{_5Wgr(ACjih|a2c$$S^Q3Ic
zEM91Xz`MDHPM>k7!mP7-O#58(jfdDFTo}cwTpLt_?;xi56DCE0c(R@4E0+D+{x}-^
zM|*9_qpP=IX+ma6(re+a#z{jD=c-XT)}I;vEH<%`$>_~0p5OzNaT@6IJ6D5}fweT9
z(u07BW&zw-I8G!^FhnLp%c18G3TelpV5&UJv0m#TKU4qL+l0STwELo8(`lluCPiNa
zrGL5ZAqkch8Hs&;JjBXt7rwPTmTsklNB1x-PgJRJv@BQXeu|m~Jx;7>Ed1@LL|=n{
z(UbM8^fV>BG-%sMuJgD~sc5>$Gdw+D{=NAs!u>D0T4wLxbv;pn5qEv@fWMEOgPsln
zZOp~LqC=c<qJ-Sdnu#VgZeLmcTK9Rr3caZ>=4~Z=E`KDNnsem?K+HN}!^GsASDoRM
zT%ni{k^X1Xoq;#RDgJ?ht&JTw2T#XOrEO%{cR6$ig^Q0{pF$)0Iu1=fld=&4D({8O
zGx?=huse<$-7bMI4e>D7vx7ysIZ_G3-Cx*=BIo4yfeU{}<I8E~A{`w%GE(#YtzdX%
zmdW@76#~PDO!7r0qi6F+N6D<}*oW!u{Uq^BQJXeJFd3*sjjM@`btEW7gZ6OsPBSmz
zoWzK4&M6p-+1T3m-k2Tm-vmNnvEFRz2M?`yeoRYavW-lLA^mEC{^MdW8_Yw9hxWry
ztrI&tm1a{NXO~&8Udbqg+-*=W%byB@uGPY9iyn<i4dGl?xB<V0NY#HXv%d4ZN}C}1
z|4t{wixtzcq;-Y7!2rn*B!zURBzmk=g$>yKsEFCBM4`*D?#AjhATTuau72dG!bMsJ
zmY|l(ZOnT5l4jX5$2J?{wC>|Ha@9(~w)x*P`~wf{{(OS`73VKj{3%n$d1|Fp%7vhx
zW&AN)E(|t50#;CiV@KevxwE~&&~UB2=BkqpB&BNB8i}K(m1#d3+ATV#J$OHu*>o>w
zIr}P_EgbuflD+H`P=m>0wwZnFD#c?Hn=@)+@&gup5L>S@C#(bpM$V%lIQt!1(c&m$
z3EHqXjK{&kc_V51c1eCAeel{Sg!_r5FbSJVzGavHoY?DV{P4%TbFDbBMXAw==*-u=
z8%&guJmp!^Y<fzAV(}W;U;g;n#HI*2UW%qI{V4_AYb$<$xJ!qp9h@w3)+pC^63;5W
zTo3~{aN<Fu0Q<l=D!%XSc&pTfVk5Ee<BM$gq?3gW@c*Xr7PHN+4J2$SB0;d%zIfCP
zx{UkvvLqe0^5EZngE$={cEv>NVnHDz;Tv~htgStRz7X(kj62yDilcG>Suohh_NHZn
z0*(iGyGkpW{Vfp6cO}{GksK&0>Yp&mTdgXs%)CelTM_S5k9h0f^e@E6X>0~ntn*8`
zH8I1m!l1n>Ez^GuYNNAoQ2!)~{wVkN4>)HYvM93=A0HnJH=E+FD#u)wxYlcylo2>T
z?8dk0CfHEYMUQ(%v;B%@=L~uRJeSr0e*Xt{F;d-Exu^1Ti|2z!K*PT^ZV!Se6sKJm
zzEv#9f@nFRy4zR6u*XlbM&donNRL4~bc8hd{b=kgXhVI-cz$|)&P2TpVShySb$E{v
zdA?Ta-g!2AyRcv>Ht@Y*JS5GN49&3ex<K>kE3Astcli@qVVuGGDyBcDF5<O#fbvG+
zy|2wLq4AX3B}xlKC5HHau{X=ho`ZKenxBcErXI!(R})$AgYw;@LeH<itUEoD{7L1L
zfg2-G8klA_><KX2nWRvSCH)cLy)(0Ne6xbbp4xP~bnr$xNS*j=_Wr(Hp5{v^Ntda~
z>S=IOgY1A*;WRx<dAE{?)pZ#YK*C0Ew>PuZb-RCY6720-4HKgOpo1={7O(BR{IoMy
z<H(Dm6v*E+gZz3XoYM%iL8&MwT2CLz`k~oqpzoWuhWL&r^`ctn9zO=_G_UTeT(uVG
zn%vH?XXJSg$@_Xlj9fkRNg^p%NQ4!nr0l=E?mYfiyRoNmxg3O)o({PN&X1jYErrb=
z8GAosRqdkhqGEG-#+ktymTu8n-|l6O&8>W=e*v=EsLf;H=*Znx!8k5zwvHNSOvq<D
zUbzX=SBQy(#jC@|AnW|FHuQlKETIqGa(a3?>QNS_XJQ{aaP(UE<!XT0lz#Nvl1FW%
zpT1$uJ)c5}#Sh297U&XSXX7~Fvt47kNvsmITBszcw6rv3fKw}03M^J}kz~@~4S^2G
zuHcH|{Ao7ROGAM<>7<R0$w^amh9&=lA2Ti}DGIbihqmuV$Rf2W^;<Noezw7N2&$!G
z0HwmAB#%9Aat3#rf*P$_dK+lr<}*YpT8a$XyC`}22&D3y02PC!k6L$Hl`~~!WL`{B
zDqVK6kP!}uh&SFL5K<fWABa$ej+wLRty72R@u~=Ge|wL<ss!G_Z|``5>g@);uS)-(
z$9j~&_EW#4eR(1lSb_SFh9FE#slbx;P2iu1-<%!W$dljhiCZ6&KZYr)!=P{R<P3cf
zc6Kkd{h}Gi>4CF_{TiZ#=s`EguC=kr%<vCS4(;sd`oAF;M{B47CMe~O1N+R4udHQ|
z-y(eTZshOkLneirfp?b84c8eexd-z+K`Z;jR6l7W`?qu!_6|O=il%3$2(WR|wVU28
zDND}{w^_y@f02Dkq8YevF8MZqfLw+{4=0iyhg{S=tOQW5s(@t^>50E+dF?f@B3$J7
z-G-w8_AD2*6pT>Wsl;C0?w_6A-i9I<W2_a(_z9?D-3GZR#wAN=a1xnfk|T5pHsk_o
zJsKwY!}%%(*^6mjS*;H|$RfA264>mx2P4`s_Ln-eyz~)eJoxVdk4m0){KNQ#mb`(*
zl#f}T2{mB-g9qOt^!8$?cH#P<8OhBxva@~tZ+C6)#_MHu^V8WG-zn+@i;iGXc*xPr
zv0A!APbx+nd*ge$P^iWtA956PO~#a#`>ZNPH#4Kcl(mdB<~U~9E%@rpchK(U@`;Yk
zUK=G}T}MR0N@_*Lk)l$K+pH(k4ou;8!Ed0Y$q|sq62N<KQoB%-E6cnPgjufKF{-=~
zTl;SO+aVoux0aw_Cb*#k{-HN!q>=}D$OsCx!JV>g(<Mh>eMzAivNo$JQ$RfH!CJ6q
zz?Y$-aG$r148~uJJgv>LpG;f?JPeV;{H+S{4|wpfjZ#RcMxS@_Vp7c^Cd%Gbqjly)
z0l3~<bgKOE(61-7u<UF3+{G6T{O?^@jiS{e<<PeI`=hCZ%F1Jo3>CALB2=szWoA3L
zWtdEHt(Ji6ljWJ-z4>A8yzAW=-;mINg`DB)9s`Y;y;7{g^#bNPM#2C5R?WBxdAB|j
zc(Xk|l4mx2vXW4vNdH`bgZ&T0mKI;I6(FLh75qszX~nPnCZO0mo-}Jp7cC{}K7_1V
zajJ=fEM?2Q;$I)?ZC<$hd3KK~shrt57L=>7KI75-b(pM*re`@N12Wvjw&NLiC8^t#
zHn9c+KdpJI<7jws)4g7k^F>8)e^~+Bv+V1M;n5Y1(mTfY+B$0a@RYOKwyxw~0uz#C
zH5F$!OVRjs)r>ESH(VXEB~jDZ5T9%xTu7GGYr%9PGk(3dk&)gCix?~+Q;F6;jsT}V
zLjE4%C7DTguzd*=gs9$W#RP#23vdx=c6BzYy{Ud;ljA)uCO}Kd{{6!E#hs3~6?bvm
zw`XS&5Aj5W(mz=DYK|RBM)tM<VG#z8r+2A;PQJTo&8q0!k*w`gS;GA5oX{+xl7?8I
zA(S?T^_@JATI_B1?rq)+ApUO-pEj(qZ8U(?o%;bE6A;5b$*Svx%MRgZy=?s4CqozE
zt(;$BeRU^>k9;lYzCT&aBFx3OeSh>Trqq#A@Y|>2R4||MEu-~$s|ofTI9%b@$tAzh
z?fn#8Q|o?qB}X^e^G-(hxm0}}wzE77_80=|`!j4m?DFzSb;}3b{K2O}qy?|Pn~{I6
zj|fAEvg76#^FY?KuR8tWUBb}X3{*3;&tL6FLzT%DT~u?vATD%$ZcOv<CW&<NfFpz2
zjp3@9I?l-pmqU<bIq(6TrEeb!kXFcB+#4$-qU$l1zdETG*>`t0Qdx=-LV^U&F9I*D
zcw5T<s$mpn_OHO_WT(}>D`DX-F8htepy4XN?@h*~RZyF0dih-saQp$dbpXD<&XsVT
zYCzxD(JQ9(w(__Qt$*$)?aYV?o^NtwiX0M--QFI)T;H3wZ;}MOPhFZdu-G<-5eos9
zcC64{o5Jk;;$nM1fUl3I&tZPv;N4Zaf}S8nZgAE_ZF7&CX8jsJDRrfT+YWS4YOPc!
zccEq<{9@Ps=*}U!2n4nYDUd@1HG<LvRj*PAz4k(uSJRlyb4x=vST$Gd7i%vaf9f{Z
zAHx_G5&52*X@HVh^Y0m^Jd~#gnyuRnnd>L&#bau$U&Tn2f$B;bNj{sC-PxZ19C}Q&
zi0ll#Dh1A1f;VP;&lU<PYHI#xBCocw0HA+-E$Ub?Jm)Y~Yk|$u+LK$|+TVM!B+~G;
zI|{0T165Zv>$lE@&2SiMcW|)d(t1C5Ig#u|W(5NX9XJfpU8CvKUaF8cJHlE9=&gS)
zR$p8Yh_-pbWIvzY*O{%Ka{p`5|8yMvNqlV-d$V@&{vz{fnEoePLC9~H&xeGR-)auj
zg);sdFN)-r&rGC2e9^PYENOC$TRd36jzgjKH8NV)GQUj58C=#pMQWMHYt+%Dzh_Tk
zxa@F&29U<M{qeW-v?h_sk1xYjElz2_B;%thc#E4{sXXgves6fU<q9C10t;MU1=UrU
zwpgO?71NM1a+{!eF(oS%pHn#AGH_!$ivKGx^=e8A-wPDY8`jt8NsWdMtnpH@v(z8$
z`8qGJo+u&Zu|eE1ec($ho*z_qSu(I)b+H0he**2OdM?J@5hqKDHlHOhRl?MFrwmHe
z(0AVXBpT|r{1Q--FrK>hApgD&jB-VRm$Cl8j{4rRCXt5}6yhICVK~KJfkGer=EXqt
zCb(Rn@=~)UvWA7p-jlY5MHbm+4OF(~3VlW(E-3k1e^pd$OaJnJ!_7qdGW;Quc?Bp|
z5&rY62Z1o8u%q>U?#NzH;;213D&_DElWLyGp=VH2AosCf;)wT*LU2|nr?b7|Y5^-x
zo@Hj{S8Ifg&u^W>TrIEcKNH*!4q)3^_~3tgaC94W2^ZR^S79_eF)itHsP9$N3H!Sv
zF}1TZ_XlHrLhl#?+$h|P-0cMZ0aL}J3F<0#OyYix5&li*LwDzQZxPvzT2Gw4a)rH@
zZ>pT5<|n7-E1QLUvs}MKdB`)W1jicugTwhI7M@NahOg_VBy8U8Dc4sf)GQ<<Fux|&
z@j37zvqA1*dwMF>5M@)9r)Rh87GN33QT<4sYoC>-6rH*F?FZ!X$kdK&CP28Jt-Rek
z<JL^-(<u8&yT{|~*1$p_6Xy8R8_~=b>*zA~XhTn_tddVKv+Y!z+w64*^6d{3*RI|g
z*Q*=(`sfvV@yEtY2X$L!2g=N!R5uJpjVI<hh4RxlD9M!V2IT`0)AtG!lEM()F7ur_
z^~c8_5){r?$C1}F^ZUR#3xM=HJb33_BcFG#=g+69)63zZJPWhQDU12+ysq9azG<me
zztD_~A0hF|McfEQqD|_;Fv<TEYoZx`YxTf(P&Kav2@)7Qo^c@>Bj0)>ReyD2q|!lU
zo}!rhr{^N0W(1FDgt{OBgp!y_Rtk6b>UP!G^0_oS+Fo>=raQ^F4j{uu3awMmU_<^F
zXRoNhfA<A@)uHvlbjxFUM!?0OBwts*$*~Y*p!Ql@-L6FK5-k@BLqJNaRtkZ5f~|sa
z0pIUpnzy!|Xq->sc&XhL|Gj)dwZ>~=?R?Re1Vc3#hrw^I9NQrf{uzaGJ%Ok%KT+1D
zGJ;si9AD|#PWA%epJREhac<`##zuCRqORFpR>xq9+pRBN7qKTBj8}H2MvLG1V!nTL
z_+&na?7xH=qz^NaFmP`Aee}`<V-`^HV0A~KxtX{^5kFpi{;7~oRN~nuUw=hQKfI-{
z6P1>^5OS*O6wQ6gPl{bZBVAeCNan`R5ZnB}Vp>y8S_VOY+V8*E4uNMv#r_+Il!%`J
zNMk`3S$~+d7P^o~t+Rp=)6+&>5qLk7dG_9<k>qI-uw|I1T^CaIJJ>WS2CDk}BTraz
zBJQjHhVX%f*;=B@`rhlU!agC3j$Q1H*SGT?;D!hQZPbpoj-cDiK(i|~jGQMTcM1fd
z9_QWsBL!%Av#&yfe0<_{|D$?lo-Ghlv@v4A_YXwg?D2c^nU;7SV{Yf~UJeVs^s8(w
z&!qIujp`&SW$C=T^;+nKAz0}SnVDKAoI*lFJLmVn9aOwh###C|-H_*?U}WfG&_>b=
z1+p41of_tl%bA#%p*1g0_|EoH-6zDN`>coHe}MeezZm4jA>BqyeF8Xuw`OEf9D2Mw
z{=t2J-)-Lwd6$zgt4JXT7T`{PnA^KttE|18c*ME}Zz*}fE3|HgY`MGbW9ew`47glY
z%PtM&Vf!&WeABxo#e3NA?rttXH?-C(f{|!zYwp<=Xk8K<aTuk@>QLM8_GD9hn_|6~
z0Qex_3Cdj;t+sAiO^yt>rR2O2;!ES^PVxuo{7Ya3y#bZS)3SE#$qk*m{N2I7n-4k)
zmYf?ranCbv<|bn9FgkY$p3`wRXD;t_7AongQPN~mTV1y2OV9r$DHc(d=lXIKjvf}%
zpo?pE#!+Q|HkPu#&TS1j{0cd+B~cop1O)I92$fCy`c6!r*i1&JUroi%yc<Ri_owwY
zN!zZbhNqmD`A}607z2a8&WfQ?8gp|3y`D3}g9z<PVZqN1yj&6M^KKG(BL=+cG?kL%
zHe_E(;94o)`8Q&Q1IN+Qa?)kSmIn)k-r~l_Q>S$pc%(66>&K8Ea*^dtN@_F@z`~{a
zwqdJzrw=mA+K34Xx{5lN+9+|X>asL;&v0W849W;0DJ`@7+W$r@W$u+tQ&U_-H$FKe
z>_1p8tPrmB(zW07`+&!ye2&4P)dQ3S&LuVB4M}ER=x4t(k=bGN9=vO#u;2)WpHcnv
zg|ZS_-b6CqWH~8WVeLFR8d~-fu)bvsV-57;CVXZ;cM|tDwIrMTN;s1MH$PIi*paC(
zwVHvH){>k}9iLnkF1cfWe9>OcsI0|1UprMfVYY}Zrm<LliYbIC`FMJsu5OLP3b*F7
zN{A5zALt3L1D`Tdh8a8KKnPAGr<@6|)@L|qOK5)Gnn}*^c@x<2o{3Wchq_On;{T!z
zpK}x-iYd1sIvMx9s|~%w%HQ+`@U>_7q*Dkk++prEPls<n5Y!pi`*#i8^dZlevB;Z4
zitTCN^mMwp`Ot4Jw8unr>E8IKx!bKFAt4-L$H!o^Ye##qB9r^esT;T7(opKE1>BZ`
zVHzjiF~d=bG;jIy!7D(Q#2tIJ6cTIIDM|Xx&Ea%JAw=9x-U8h=gWFH6I}pHx>a8N+
zU9ieUiC?;dNVuIHALhwUuKUr3R&SrpVEQ_V&nMP>|NhmOm6Z&d4(vc8tEFzx&GHc3
zv}A17UTP+)187`&BpY=5uPd8@!QH$X<8zJn>%6)gLfW9N06Ng;dFoR?BO82f`D2d6
zd%~xtNCI@o8Gccd_lRTXDk^G#!*vx6DTmEwgq%A|@J`R-?X#*pF~myy+}?)@&tyU+
z7VFi50;N?O1EPF=W1JZE?_;U`U@PuK>x`8Ti@lM;^#<|<A|_49PQx$D^XhEENa12;
z@$Q2oQtJk|l|Z)rkS=}X_41vUr9!ms1Fb%7i1u@uhfIQx8J&Mi`Dq0WauI*4ab{^Z
z8z9{L{KOD3LD6;*oEz2Ex0EAFO82&iK!0Gbs=m@pS+}uB%zZBPQ}o;V8DTQBc`|8R
zQ9@o9re?EgP|W!K+!Z9Mz)>ZRAY~+2UIC}<sg(~vhNl`9N%?tbO_*%BO7y(4qJqQ<
zIZr>_x`=Uzxv;8A{Vzc1cU)x%)P($nTaNc;RiXOJBRs_^A8uy>QbG@=s4*LPyRG)(
z-wQ#aHQ(_6Dy8d5qZVMkL72Gi?`-Au_4V2Hr?S6Ccf~ZFDW>;)V|v+3$sReX!1_ix
z(z;&)?)Q2WmDQ-^a%NlB$1U^nAC3*2tY;}+D_r)p-8H-Sf9!$35#^S&hs%AVI{Z95
zJM}TT{~>Mdiry38%`&r!Odz{FL<L4eRW;@GT#koX#DRdCt%8IDa^E9q;4`AQs6w5U
zZ)bSjt#Q^(!tB)?3o|yeE}AtlKWy7&!KYpj`1SL3l+_Q5>dBqc(G+r07VBpdh8uKr
zU>feb4#`I)vy?mOGZu}+9tDaYz*w&_RA!-*T*v<Bg<Jp0?M_N;G!GAL;1?Cw?iUmU
ze<YSS%VV`mFu_+hdv3@%ccUp{N^FR4W_rwp`HL^8y&JyRj=I+ZPJW|?@VBmV_nsL{
z0JWWg<HP8ttI*vKR*6IPeG;|iUV>a%e_kYp=~is^dA`XEm=K!}rhoQOse$71Xqzhh
zyn<G6frwygm~K3^hPfSkzGc8t;kvkJwn#qypK9}_=_Etn{Em?P^YTB22HI^qIUchC
z2h7E(to>L&ipPA86=O|R`nj$2ic8PkrhujV9a=i?#mN9BBS}}^t7eMKuc?|mF~>F4
zm6<h+e`cSmMbPT;554T$Ga}C6+nuzQM|I)w@-rGw7POvvP+QRqc6REK2j3hGb5E!t
zvB)Z-v(8W7tZ6rT#7?6{RrR_zE7-o_$`quUK>B|D=At9slYKedFjM2(wzkHhN(3wr
zJMi|bU(N_G+z#YjV|qZ4Xuw2Rn{-UN1N=0L-^GHT{W~U+O9P*$L{;YV<Vy}Cwj({{
zS9X;Li%CH-(RNkr-(L%wh(KU83_l=0#kAN^B|9Jj@b~GcFCkU)oS8v0TjYe3ok@k+
zvLaM1yv>mjVTNh4zX&H8S{qwe2OR3Z$gM|ip)Bi!*(<uAi}<s?Hx57k?c!a*r~kuH
z%Qd{9zzYVlpbhMk3X(NcA>J4^wc@%@e|%hpt;+p)iyax}I?eh9Ij7C0-j&@`g@wQR
z=VH`$N}+k-#u=vaQnE?zH+PTy%eCp*Cp-<*q20Qb#AA~5v5y47XoZ#7JD^0a6hX)x
zM2L7OK*+)H#Jy`Q_W}vsyki^ekuhn1xH@&LrIsQh&z^L66HJCDzPFXI$0Ds~s=Ug?
ztZKfv$IcVO!z^~9sEtF>R(!ykhR=H3%gL#1yBxBJ@ofXEoM|XcY<&R0+dBklN&svz
zl{pRauD$*d4K2uLXZ`H})YrIvEVG@#UR|9AA~A^G@WfU93`B9oHu&ToACnpo2O>yt
z7Qgw(WDEX$2D`j;W6UQ|E&$#LxrEpv#z`qh%m00g*3amwXs}6r^4X&EOY&fId}-*g
zdfT2;=22*lx8B=#SEodDIc;yyw8w%<WItX!+lh=;5BPx;;+1OyI?vd=$kS5bTmv1e
zOx9n9aRa7zOH;pXncC|?2{NpGPB-P$Jli_4%4H&1ew4w~JB}JLZc)n5&?-v<DW&zj
zu_~phM%Xz}TzI}mIo|N*U%}WHp<`+7{+(g1@dmTl&h8~W4gC>a!d=um4c5$&qzn1;
zxQgAQOek@fYAqJnqoEQTX0muad#3AM>#I)5Ua>$ez0~N}&noDztIvgzWzFYM_S^O4
zfCf?Vl@sJ@P{azlK9rsg7OxJ?oxgg34z4P7oONwRAj-RiW54F^4gS(#w$gkbjZYc!
zHrHB)k$awuBasL*YvOl$@M-AeXgy%2_T>n0eLW=9c|HS-gktkTg8p`Tka{F?KBjV<
zbkvyCkXs{tdBZ%>rWly-rg<Y5Z}UJly_+vSd#mJL1=&{!xpp>sx2}{o@~0G=^;}c;
z#rk?lwG2@_6Y(=>XotzLeSL}Bz5d@MsPA3~0zJjQ6d*}3Ygm<7NKnwXZ9dgNvDvhm
zpZke#cf2<spQqCk9Vu=dosZUIuJ=Rtva-uF@8a`I-n#+4CYxBWs!CEpfeum16GpD9
z{Om!FcZT=8GG{ibJkBTcQ3X3e-h29n3cd$380^qmGNSp$ZDClFZf@D3fw<DRkt!jo
z<E3xL0!q5MV%8_;fsU^@|5cND??C(Sv-kHOzN`mK---?HoSik1j9S~SwNtCF(6qXV
zXU)4Cu|;m)?Cs4zWiK{0^1(f2a}j5BmQPrDGo*Q*^{!hX@)S6qv#{FOyM3lG`ALy>
zTqKj~v!9@}j`9ar4qiv-hoiN+N6*zq^{4S{5;(iXJ5)aW$%rk|!)Ilc%-DG#hz?L5
z446Fy=7!xwOGY!1EvZ*mZOv2e?r!euF*P+xNwTpagB<B(tWG<ox*NwW1WUimG1#`7
z&s-mtRH&%FaHWTbY0V9<$Fw>c<0~o`!bLH;EPA?$g~xf>==|JUQDe1h=}30=Bw_3d
z^ih`*4TSUqs^`e_<x9k$$YMU+#n{#vm#G|}-*@`4G{tnIKv}%Owl{L7YI_r}WE00Q
zDzw?Oc2MjOq9a2<iOgtdMz`c<zf~AdZ=mh!N(;$SVyHi6H&v&rw>@i8HmhG(J*N|T
zIWY5Y_ur9I{loMjf<o-S(vLf>ICOpNbMvXCV#xNYaEi{r5O3}>zTo-=S=m?pb;x$$
zXYs46!;;-gQJe+Nd8?-n4DqS7nFn!K+wtfqtsdILaFYrtm?y*j$t8JL4B&fL5WMy<
zhD@~O7r~p@Rnk7T^T~V|?d6Djl<T5Lf&1p$7BhQJNh<2E`*B0ULVZ-c()d58)2-op
zg&rb+bKaT;Ru4V?L*YonI{3V7`x9;oa6Fpu<yjTp1+NYndEx&4hn)#CJvorncmrb0
zeAdS1j7H3n=e=&o&i2=T+e2QC&i4#8d36UYb(2i^Ds<1aRT3F3lz(S>`=$&MtFFO{
zA5Ehq{ENQtkkz<q$CGJ1mGW?RH}(r%%zL|ksC8<+2ZfG3#JY&MkXvwGvL$)|z#k@C
z-QDG9)@nT8EO7hMSzN%iQR1vM_;F!IHi<M8xghQ%z@9Y>vL_%vCF4ups}cD0v#)k|
zveu}4FjV}=6K9}N?#KWUemW$kb#-&Qy*&s!Fd2_uc{2-h-e_`_$r}Lya}Y(9T}WcT
zu@`=Ze&(?KZ_O?vtsXwy%qTd3+DwdXamqw(Q+Ovzp#Dai^?*q4x`48Amx_1aTNtvp
z@2^CYo#;4So(7sB5C>hC)Z(Ugkt5G19mCv8)BB54mH(9ZR@2uR^YfF0dT*~A{m7)I
zy&1I8PV4(PJhc)Tnp+~qb-cmGF2$#NS_UV8P%q>K=RRHjFQU$@o-ZE=%qHj39qoBT
z#Tw>ltQ?&%GN2#jsm}Xxu<NfX0;j}rcF!+s4`eCee%1%AslIZim&K1+{lT-LNtMlV
zw)b<XkLvSc#{pm2;>uyq_n8FfVwxNwiWyJ!P9^24>?bOmF4Z1Dr2Hv|o>DozkC)25
z9ix}eMO@kEW-6Ax?qUFBr5p%`n;Q2ht-pL(W>NkJ;BWG7W?-dG`E(~5SHT;Q<8C3N
z&`71pRO7KpEb|j;TeV_!ZJVv<O5{)hs+{3hv;(|Wdl>K9D(3P^R#irR)vvp6UyJ>5
zcWaw`tWG6ZQMp$lALQOl?@;f{^qnL(PsrWPoQS2=Dx>+uNoXe^3mjg@yuL4;XIx>`
zcN2dj98=E4mM}PgBdNrx@*fP3weXoyK&<itVvmW6)Vu1j-UD5BW#J({Oo(~`{N=b<
zrg`{#rK$r(Pt`(6lKhrUPhL>*P*#PozPD;lv!du8l)CT8Z1$TsC=>G0?lhqJTJosZ
zEFp#P3rrGD?m7-6aZGq-X#Dzg%^$(++rWEo9Q3<<7qO0ia|IaE=Ch0}YRnb94R3cx
z?oLLn$-|5&)>m#$BZf+RQX&jdY;@}RRhPi+qQZp)aBZRBFH9#}VexYoTM!){sV@HQ
zi+5*hD?{JttRLt_q`w%4GIW3!0Wu*F!=4yLjr@x_11e&pd5eYeI{YtBwp_&LT*RZS
zO9bE~bLRb#8HATw<GZADYjR`}tW2qF)cotmp-2huzD`U!s4MjLIyK%;Dyfi4FL6}n
z+K5HhJKi@$ft-t&-tOFSlK0fVjJG;wT=)aOe9So1o7+2EHMcy8q3E*3-1>j(g~hnv
zPksmiAgqw*%1;eRHewez3LO1|9}M>UAm}w;**hVk4|k6*{{E5C8lTp-7w!1g{Z~w#
zdrH(Gz?6D=9|tmSFtTxM1&ex3cFPal(lc@Y;Iw<jRwjx23l^df>&wvRsmXgomwzp^
z0^$%ydt{CnFYJb_sg}EWi<3puy)<1YcI)K;GO9JndzQaDxvrF>HQ2kMByZZvHE#-?
zbZiyfYaIH!*^3~>sh*lL)M35GPo7Mt^0xH2mp+f&6Ta-s8~gQ$f<V~XRXooK;p3!{
zRCp0wqeRkn<5Bxz@8j+6Zo4A`n7uap630wv+RGIE9zlRmJdZy9`I{tvU()AL8*b|W
ze0hIm1+IkA%61IgoYq#DU|z2DDoa5LTzu<ei{Fdt=29ol)Y}q1!vV9qO_|=Yom+Jz
z-?Dmjr4qQbdSZQjWY)?+caSuxp;|30^oKX3+M67UX|!W`n8;2wq=+v4YN)kV;ZeUb
zZN5(}>^r|}ja=ll9sZk;az7_e0QCz%o?l<=?QP$1QwY@xDxJjCukEi@&P~+@Xd>p{
zHFEf47wlvT$?#-Wr=#Gc>H1StPfbV~|1><r`$0K+!M^Z;GH0k5f`eOrKY%XW=smms
zgb%vBg07f`CEI}JC4`ex@SmBj{iA+db+jHQnvY&%u*VH)Mx<zYrtGvs4E+JvhTFri
ze}3NHssx@nxGx6#UhFNlEw<R7DUR+<`q^kpk|D;Eu&xvFJ!0P=g<dToH(t#hLUyek
zL0MxN$8{|Lh{LJeum}zG^E=MLgp^i^!-Xk!5^~MgPB(@Fe(ICpiQ0~4R+M~3UdJG>
znt|)uY`mX9(f4}a#Nup1O+Y-r{fDKsIb-zR)dHslZ*eE4tpv`{_p>ZjBx}N9BW=0g
zs|6qO48Gz-rt13VKVdz@y=weFBA4a1kT-La0Yn8a<@1ymQ8H6La&hC<Et{o!RN7{*
zH#+tRK|AnFHF`6V;DhBw)dVFiq2`qTzj2#b`>XdvM8QN5KR-XT6rd~FqF!S4R-JFV
zBTu6V{+L>R_BdmEF(V|XW!T<E4mE6FSV2cecXh)|C2%OAD1jS-T%05Q@_js1Shc^v
zX0D0LPDbf5>u-~;qZ^pbgjjd*W{Gg0N8(0S@*vAf%GYfPaLX7R8m-$+@03AS)uykU
zthWLE-HG+W)cM>FfU8N6kF2=mVaCeP^Yn6Y8e4f%57x$n>)-DhaO?<_J)pRQ{JR;^
zV~I^|d-lz)nKRZc*+T1KKfB?|8v|0@ftwVsh-9>$D`%7PIW1ma1e;GZK14R&2%$$F
z_VRjaxC`vmR&4+K2hQTX{zST5mZ>7y>Ye#kRhhJ&q>!XS1YkL5Bn{H0ba@jtRGuKf
zSD>HK=cp-_;{CfFgN@F3@zAC&8k*0}Z>b1CQx>6{>HTqZEM*^+-U$wPCrMwHl!tS(
zC^`zL+&_s4OqL`_<hz7L_{skY&n1DfM~0z&)MF^|aDTvP{tKwZH#o#zolhyWAQydi
z;6G;8gj0#=03E{^HH^&>Tf6^-FDz=Ns!bB<^)&}pE67~EFAgfi$Jxgdy&$4_A^o<K
zswyJ3t*uGWl`|m`Zm^f?m@JYv;)GG7ym#*%b-ub*lt|G(ya!J?U1ncS#LSF*<#^8?
zR`dLQPNhj>c}4P&9yU%&l#-<K_^Am3Ha6`()wACT_%018L!jCwyJOvyU+NPCaO+N}
zn%nPe+b-ERuIQYR?U(4IBe}uW+1<AK4?H!CZULS4Rl(z@JMD`#=5lwvZuR!huWr`Y
z*TKQ4wyb8l`m+c%9hMx046iRGJC#E*Dx%IewBKH)-N5v{Rt#13qPDT2nC9CX<lXU9
zty?Z2E5~75Unm`?!P1W|4F9k0XHldoJx9UKpbPZgZtU_m+r`_#)wv$w3}QO=g~i3J
zrFsA07_O*~MD}?z9uCvkZ#Dd8Tp0e<KQ$j0=5|NQyWV{3lG=z^bZh?l$>4t&Tv<|T
zs{8PN>$l*77|hV8|D!tVG;lA%f}ggzx%mUoa2e|QQ5=%pN`S>;&rDx>uj+@Fmi!-0
zXZg@%`}XmT5)ffSVl<;c1V)dN7$IGgk{l`B%>e0=&LQ0;sibtLlz@cNtx~rj`0V%M
z`2((3yRLJ`d3@tjt;S85+^*i)V#piED7f_2+%`cpl8puX$t;D5f8x%jEcT4r(wH|i
zLdlKtsbL2ufbEY7y(4F$CLOWN|4ftvzTRAO8wxj#fKuQR*>#J39XX@$kDE6#2uBi#
zEe-s7D(nI8xFUiP{p-LVm$UMP8$oqIdVlfN7*bY|Xzp%nbPbc*E=ZnoQ6C4xg2g^c
zZLBt7!F<@pt`xL#^IKxB^HGBn7oD;as!iIt;LUg2pTs*~^JbZf;R3)uq*zK~wS@S&
zsn-s&SJWsB2QE*^5(gFbemZ5zG?&(Zk}i1eA;%<EM)4c{9utqu=C_66*$zX-z;MS9
zIN(@@;cda6o~mM<BS5urg8ri>B|aghkI#%Jn>`b&nTyNZRL59K@~v&<^pCx9q493t
z`+p}^=AH9D+9MDDuNPWK#uk~Z>Lp*f1Ags(AeR!fZThq-;^);ZUA2{xkf;Ze(o;2*
z7DZ`AUqOPdlm`ZVtFV=Eeep-}^XiE7exORtl%s)n-a3T!>b|nrAd3uDy9VKJsGqLq
zoBLj4Ff$_e;=iSl{EBs3PpZc9M#oYAp-^?07iH=oXw4_{aZTvsMx}SyS}L&I@79`T
z3dPENz1cxogluo^MmwgdSfaSZS9GDVX=1T{3=@wQnYE$qvOl}C4CZ7<_ByDT8dSIw
z%i(#Ahr`&$da9Z%kOsvLs*(sSVc>n<mn7Hv`E{J*M{C@|l%E3{Gu=O3$Xh}e61G2W
z{;zGQ-~Xs%WiRCFzqyhu^I&h$JTc0%vuiV(z|T*hUBQ9fMD5-vUbvP+>pXSCgzLl4
z!!$G*(=+^x>l5BCf6t4ZG0N1x<I4a_`a&H%Q5VlILv;DhTy2}@-?@lBrf6cN%FcSv
z>D{Ub?b9}FOET4j9~$jSbKnp#TFM#~K=6h$wngq8Hy=r4;wR7^)CPWJ_San}>wEeA
z)~F6LOSj9@$0i~8U5qy}Z_|lH4sld1FM*DDhANvBJQ41(el+i$Z<B5d-O6Vz<<=q?
z+`&x!klfd2ixMx7&H437e@XXqnkNkavQ%({8(5Uawt!Rmz2R5Ha4B7@mFc(n&Nkv+
z97ZtBTURdEKduR|Uq$WC??tHwzNBOEBguMAVySxagMX)~e*gP>8mb9UE@Ot!IP;=!
z=U$K}?Op6Kb5Fi0h^4=u1&tLE+rxfofbsF=ag;M{i#!kL;aVFhz6a~S7+IPqXy%sC
z^Ol#CTNVv(Lz$Dhe1Q~Lg0JqIHBluKs>jjYJB5TtAZhrcgV`dh&vYJE+BnQehrH{t
zG|uL>ivw9a%IPdSLj571zkezsBXiJWdX(E%8zR_x5}-X#tW?u@*clGe1`kl;C_Z4B
zzLo!e1U&d1aP0OUXG{$7<D%o$oND6u8z9h=?`^^>4Hj}wOVJy8kraE&2Lf@jhr4&=
zp1?^JRZSo;KY)elwcAdxc5l7okxPo|wDD^aZ#|)!jAziLO!mq+k)9zj#dNfyorWwM
z1;25w(hOf-g_fO|FIK0%K?nZBS?u-`(XPz>gea^xkqWr{@4kZRCAaL^+UB3WEdKYD
zMVk+G|Bn9l{n@*kh{L&_;ip2b&;AbD5Rug%sef$UAzNw)OlB!!O%WX5A&N_=egba*
z#F0{9#6HfyEqGnaGGVA1w`hs}vzb1emPR3z<{1kFR9Z9M+^>lNi7Bq;xcfq0v^lGD
zM?&I4Jr8Qi?SILfsevtaWnMu&iy9|?tYj|_yi)5rQ1gB+W;5>paoslP@59IsTPae*
zwhQ<)KOb`7&(N;zJpT;8@Mc%Q)$Y}VohkDE;KhA#^C!ryhJgVS`nV5$Uvu1ifq!>a
z)7j9o-EeIzQdHhg#`u17vjHeo6qo7iZKm|iFtgi`x95F5N0StI)BTf7)#nNaV(;FP
zyg1*U9bKe1H^yJw<}8!8AmKGH1~Y0iPa)_@A}9M~^ZUWbv%+4S>Q^+F-{t^;Xcbn&
z8c6F0?w9&kY&R|2VjrHN%K*iMbia+mozfSGRuZLPgJS7YIb((wg(j2V?ODGqC8lge
zL)_?ph!KjqjBo4_#dp>>SF}DlaIg5%`R~Q<(*R6d>GpJkZ&~IAwn$0SBfqrZ8e+@g
zYlt&F`6#OHco=4M*+hUX04mPyS&8DFWBtZt@o;%6r@f%CHW&3CjR`a0E7ne)73Hkb
zz*JWQ>@?vpA;M(N`CJvSdpIf4Xpx*0!(<F4=`$Q61ZWfrNAA3(O~rw0O=B3lFE*kr
zA=kkLY9YT5q^bL}httaqBRk>>bv;sbRe%TvavtZik8l_uHuS^$=QHTzZ6@A5Pn?C~
z(VtsK-v^@K*<GNU3;z9B30aPhQH|}(FPI{GtVnp*Xt0oLg6p9Tv;oN4B;$B|&9Uj3
z7IpIe`SZg?MW{+)F$*tmZSfDxix-lcr)@?PR*h~H!>M0{nL{1($z&>O<9I}G>^99q
zD)R2MaPR#&5o||*Plwey^ih?%^0k@GkBZ&|?ztHnoGap?V~rLb=_<-{ZsC)9`f}Ut
zGdK`Nl=3PbNf68Ng#LK_Bl<$(Xqx@ox7HVTX<EK_KjgE2azyNXlzAvwsja7O)!pw7
z^t`d{4!OftSacn%et&p>qmX?jS@7D_j+1_WHURIX4U53N6ZiZ9ig{oAE$cf7%^3bU
zt{}D6Mo&*fIyw%(0;Kw$3ZTbzn?Au9EKcU&G&K>;sBx1Vm09F^sfGnv`pjuzrVP9w
zdI@~)RkqGR=nU9&-0>X6LoXkRIc_sL{ZOAS<Yl>fGjr7a^BRtO7PX!(Z#MYpuf4yh
zLFJ3k=Db*+l_7_V54v>&Q)_Dv6HHn6(lS_2X$1WUdVTcHA$zl)Zx5Emz2I!WMLn}2
z<^T6Xc$F+>X6<yZEgdMATu4X=w{(-m(@j<^bzZ}&3GjwkTIY}}w$$$>iCJLrG75=L
zl_i#-W>9b)NGvjRY*A~ti*tAwG}4tq76_RfsIgXfLLHRQSnBk$zLEVJ%gF05tI8EF
zOI1T7kznCndMme`Pw5@iH+QB#!*bdw<abKk9+(}$V<X<R8+U0$31y2FU2FCjUDfr4
zGKxTRByuGoDXd4%?elshh}UctOuKI9I&M!DNOW=Vf~GiMGf78+%0Lr8+2$<jf#U~T
zaVn`O5tUA@E4{qo<UV%pT6YTZNQI1y;h@zMTGu_jy=@8#GpIi^?k#SVV?Buep?}_P
z4JCLXC^1j>yNpOHahlWnBjMDxjAYI${p{2Vf)%7dq1;BfSHQi(fDZ&90^46hf&9~w
zUJ;od@r)u_aXgC67C;1mihEnQ$N%nj?<`YsSXe|R6<cYGlr8P3m-E7upSo}sqYK%J
zw}oNxIa|ubyB@!=SJ_sLr(Idak>7lp+fD2KJz%>H54A$Bj~0*9Q!jKPa+)JKYPf5y
zL40Sk%@zPWWHhti*HZQX$ou5CD{mP^#IO;P$GLGF!rA54r9ESKRtb}WBqeGqM{2)G
z!$2`(v8bR5&c9kX!QD;_kvL{^Xq|PwpaUvX&0wA!`6n_26+U%wJgH=j9CQ6Y6tq1E
zAvnhE4Duk#H7;vzMnIv$J2{A|?ex5h7`A@(#h*2%?g)~{=NMeMT<`B!Sq1HHD1*fc
zPPzNd|1K7oo0;u(-2IL>RW{nSg|CTuc*34#*)^B{o=&+M0Wss5>xfvJpjWFRs`b0b
zstMT+aKi0g?RW7os^_<Yz~u1>n1#UT7~NX3ZKu(JQPF5NOlhgqRp+9&BAX^8LMp~0
zK~9zvOr2NdN8%0<lAlQM=WicP_ixrP@7nnLcXz{J@nS|yi>X)nUD#t@J}Q!y)0?VK
z7x5Jo#-S}6WLiuz>B1Oa^uFth&pkA^Sz<K>54#2L+nsAs%TwVMQ!cqvc6W={y9UQM
zGG^&5S8~{F;|TiBw;|-`Dd%EIB-k-#y#Lpy_B#WYdja7>?=`>uJ4^e{rF08#Qxlmr
z;lv@=JX2L}WTP$^kNLNrDr^|!4v7*>dL$(w>hRpAjl=+1x9<>uhD=N`v~U0I$Qj!b
z*EC(v4C?ij;Nx!Pm5JyjLFED#{LRzVEYE{0dAOZ4B6T5Qs04OueCNDZ3IE~Q9}SaC
z=&Gu;FkBbaV|-K9_{vTtq>V=mAN2<ocqsKl27EeM2SHZz!YEBIjEpHSRp<Iyi{LWh
zGr-sLn}<)HN5Zpc>#VkXCe|NypPfaLcWs>Kn|{uc!3Zq-LScA9B|?<tVSz|wTT@)j
z_1%X53yra*9I%d_7`&p~0aYowmRRTWd=jDmM_|vK?2>pqdZ^wO+F(v*R{i{^kVE3s
z5f{jxM8&#CsFsCi5|e}6Nv?Z;txlAxVkskRO-4Us>9>Rik*NnMM5rkB)$pVWtqzyp
z|CFU&A?57pHhr>fxR}Vr#mptBrkP^|s#4iF_wn`p$i1I_(Zlqr=FSolClHhDsj<d4
zGfg49TRI8ADhdTJ{{7~B@#k)Troi=M9ta2LL%ObDDkJ@qB;pbWdMjjK<KIXHG5NtJ
zuN2+5D;-j_vhR<8+ALsywUw@Y1PB+dT6L+w&(rqGjeaCVoG@;+?Fvf2+{a8Gbb@A>
z<*%<F6@%FJ$OQiPpGnHre8Im@Z}JLXQX2L7%kOuQWu6Pf-!Q7+CC}or6dBKagDW4E
z7stGrn8u#wvNGb2bllaqJvO8Q{QgL8W>sV<sES{sQHGP(12F;naT_vsbEdg%d6G=`
z)#f6nMtrp`HP%-Lb0=fJofvPLGcg7pC-g<+l}!Qxn{9P<=5Bqvx)e_3fNWqGM86c&
z{#p<Kf;I=+hL6Yf2e^e>@)gHzCBPqtdGQnic1)$UBFFB8sfx4HzSVn>ub)Kg-daX*
z%UPm9T?RlmZQ_{TkkDni+rw6=+pXW_PSXHYoPA5IN)qz-;Hu_sh~CvK$Rp6x^WmUG
ze#)-X(#PlFhaLLrq*)7FVdG*RGN*Xm-=C^mpLZRy=DO64JN~jQ&oYFHrrb67zq$1{
zcLwOI?THr;2JZpdF|+LobR!+o<u7sP(}vIhpE+n6E0vsZw5D{GvXgo07+=08o^Uw=
zfs7m;6cl0}1SNV3NWPqqb826A;I1)}djB64inDg&2sC2ZvL3o|mrCZ@?$&bAYV9L=
za{Nas>eGq~Jy@J~ylnHMhDluUK!?jn2xyj@b0R*kT2y^!yPga&$=(TyQ@NxByQ4Sl
z_~rsFgilr1+9tmReBM0gwB^Q^Ke*d~FSjJV3?P|(?qyVFRZ-(Wd#nY?@RE4`sDTgg
zbbYN46Y(;0fi+YiP6L-iK>6ryqyIV6bM$jjNInwlw(m^TnBq1$9ghT$-4PPI(I0+0
z^{J-q^4<AKcRP37u%Y=`;*gax42dRfE{Hf7rQOcFu6Z;5QfqJc(T3TfKoJ|)$rNWO
zh@tzA39h^WKTkdYWK@?TRAiZJi{lC-A^^3yYG={F<fHN{85|2fDU#s<N^on`$oe|Z
zu9W(AKxcb<JJwXmr#QE`v5XYS^PaT9d^H}+JKfu@F))|9`*D57vA^JBhbGR+@qBr$
zgc!n0Z(|w$@XPdtjO{C`fg2BcWb-f%sG44V(di&IV@2Ylg=A)<+xI~q8lGd9&qDVm
zYi$6-#c>`^_R0|4)D1l$YL4Oq)ctNVKM^DuFF7c33uKz&#Q1ne1vv%;{n@IhBCJeo
zk=#Fx7>0C<J&f)Enfr~2(J2>%hUDsh2V7qSNeqN^Z#v(76vLgbuQabScZtdhqO<Y%
z{Of1;*Dt5H4<1IF8?|1>A}$6#&vKE}@rr~r+UeS4f=1Ceio`fYzKVDh$w55w8?np)
zRvwA*c3JGX$*tQIrTq@Y5~l*!lAn&ZAHzX(2Aels-p}5=R2<1qLJx>1#is3%<jCRc
z--z-6&jj4#R0y<jcnQaQ0lbBPsjg>YCd<O%q0_7^R&sNl$9WS*yoIjjAu?lG1=4}n
z>zjYx4(<L?q%O<Us<=O$5@`5u?$NvY0@pZfM*E2@kV^Jbm!XU6^2^bo>grtCpZp;=
zV*hSm+)rI(@yi7K?n4V}v;lFz&Q%kKYZj3ouDDh^UUI`dWT!*N1V!c@vMUL|o%L%j
z9|OGF1%HZY#9XmG?-p?=FB7<$lTOLO0`hu#L^p;%Y*@_2*kVC3Mhp-Kml=x|ae&ju
zwOy+m&y7SA9K6WXcn#b9B+uOPj{0!2^6VB)VZE0FM%#9(+-TM8>~^<Y4!Qs<+o%7&
z3S)imc~iry3rofy4qDevNpg{k6#RGh8yi{ftekus598?q*zu|2sg2jtOJmQ}0uy4&
zE-}-z?iO!^=5No(0-RAJT(#*!F<5>d_sJW(H65(GeWbDaI9+$V%6S#rYc!oX;fizl
z`C4hjy)>;i9i5aZ68(mNj4{!trQ$XriNR0GI6Gr|%3^}yqe~Hz&>B@9SGn`E_wE_P
z2&+Jra3XxL!y)e?zNS9!o!*9%@5dT%9^Sb27?wu+aU#%&FgCid`EBDpMrO<sIkHkK
zbonYp_qT0xcXzj`t0=sp!8xk^>wl3khSZl{JDjB36llUt`~Z}5D1Z{^Y)J5#G)FnH
z>QFuUexog){#v*)Q@3rcj{V8xWHA&W<CH1AK)MMyFs+Fw+i5ZJ2UD~7)l{W3^MK}x
zzBS`ns@_fhnC%P=HPpS{!IbDiWFOm=jz>_9KnX`!*xw-zRA_mZ&4ZyeVq~j+{y;M0
zsnlU%w7Q*sq4DeY=-(I1(4LV0cE0aAt-ZshqXJK@<`4~a#|peSNk)(DJVsR%nLlk2
z4iT#(5LszB(2i;fWJL9watCr-N=$idjAv@di`|I+kmeQAfe%Km6@&gbU=@;DSwa>!
zuL^)7v%I49tKVGzX@aDQpU%emmxp7zzyJUU4c#XqGHfz%)5Yt<4eu3pTV7hp1V1i3
zHtP;z|J9c_^YCzG_afEjWe;|Y{oCgsUCmvgE)8e_>lf8i?pe-X+%zyQd(+i$3?F^|
z^LEof-s0e{=QQ`N#CN97{3XMQlvWE{*4F2TqGN?NwZk%E8YoN?JGianP5bLyzFCHk
zXqgqUnVy>^vX)1mL+)t->M;%h5IS=~6NnHB7G7NM_%y-f=~ZFe%QScKC4Lc`BxE9|
zsx>`lrG-qv`Ar~dK%<QiP~m%odDef?rWd+(|E5)y8wmB!8X5a8m+;j)zuR}9Jgq$(
zd;?R`Tw%0T^~Y`1OK&Q_6d>U)!&-dMdMP_eGVYvOVx5K<PZpA`L%-%@sWNM4LrEc@
z0AO#W0QG=N_r;F=$BSwIKmS$&FAOKXR0sLhU6BT!drnK=Y;=fRI<ej#o3912pw>cF
zqrVC&MpS|ZIpVxjWF*6fU+7yw5Rrs&V0!Y~3Kexa4Hb0~@O?OJoGp@&Yia?FBdu@X
zyI3cQu{^&gbqq$=cKfbZ9~E6X{wje#%Qy?XUb&D7>^=@&YJP8{H*I5-3Nzybao1~9
z9Qii=Y-;Dq`SBM21KFR9hmX(9DW~Cwmpb^ycmZSox!BkY9bcQ;b@Op$PFfk#jj8UP
z6@U|!@BpEaLXSbd#8ZHUcU&y!*UMAWXJ$I-j`LSz)>D6e^v!pYNx2<O<}e_(<B4a}
zHxfCve_8*SD0VnnzwaLM01{3I%Ic-t=unulKUcLi<D<6ona039Iff6qSRFA!+<Me`
zj`Vd}7{aQwb-BekJn^lMogs0#Va)yM6lX`ks_(uf3N{T+dP&vUx|0s*_Sa`8Cr+-e
z$7(RsJX*&!2r!h4(9KN!Xh>C5MVuB1<#g89RDpTg{>lX6B_3C^KnMY*MBJQI3Iuow
zPy$<^N0O1e3ggNxOCVh;=&h_$62-Ak&*#8f{WovMr<B+ciOjjWU3AOP|B4eZ6?ETI
z6DNAyZ^q;`L-xJ1E*a3QX}*q23BY0)g=b=vF1bSe2R?<Q-X-dO4g$W3UuR8JeJf6|
zf@SDnM_4&gy#GA*j*Pe$2QW+T@Y4V!zw2h`;RBR_fZ*5m%*BO3ibC8TM#<Sa>i0|e
zLUAgppw&?v=d?Ll)j|gV3P6*270ydFJe)4?M7>rKceId`hW3S>C0Ken9e>fMc2xdI
z8>SYgKr*y3ov#(tKNrx*OCNM2)NnU~zLm<J+uJi%RVPqhEOn5#_szOkuc_}K0g%6C
zCMwA!6?!wlNW=9m_Z3p6Z1^)V0^28C^FAyUsUN16UkI}m%=w1?cb}J+x3;!MEMxz=
z<e>S?x^i!M2VKGQ(cVt-2^JVbM@Pqk)pM|QcT(7pJ!J_dI5A~Wj#@D+O8BCRt~yGt
z0@1)f8}|El%(~d?p`#v)((v)|Sz2BeL2hB_BtKVX9dW&5ko2+j9Z8*nQ}<{ui?B%@
zcPtsjoCO*2w-!2x=J}(r59#~qnj2pIrgg*VT@gl8{+N}s0V&RYU;qdMmcRGW*1(+;
zpKzv)ot@QlcwA|c&IKEM%e%O>)^O0oz;_C9?UEX6Rr6=<Gn&}9yBo45l&=Wgl1fG{
zHj?>VzfudxEki|V6j%rVEe_u2gQXQ~zVFRCidhcz{U!U0>~m5A{xv&!fx((^J4<VM
z4t?}})B6t;dYCC{EPhKliJs;i=LYqp$L+#l;Cc7?WR9uAh6F3l^xfF1b+G5U|Hok4
zz`IGB#k59i!T6=6#uyPcF4%YwY;$F0WqoY|o?$~2({l-dG`^)5Nm7#Ey)Ow(8dUW-
ze}=RH_<Qj<0fan{ymv1Nycsu)PrwibVf0vzBKiJGP>8MFKArF8a~9=yjMxq=ugG(^
zqWXv&rtwJWh!qm;TO5$StY$|`J?CC_GgQ!8M+K1sh{>!ilVc#NON6t0Fk2|d`Tt@9
z<{025`48TaV2{jDg-$#^XvOY{zT{dXh=$e<m)!VVo@AJ&d}Xx-kq0YBLv9jGrGBl>
z1(|*PC4=IO8Krqj&C8XbVOr84KkA%Db#Wg+vpO2ft+cV3KQijfDi?~cKhjTYp<Rk!
zi$5Lu^}B|GN{HEL{C-RbaW62sJ{}_M!Ff2*(-_FPAV?~0jz<)i`Xx1%IRZdvFQ=Ui
zz#jy_xL`}S;nj1y24+|2v3KbHt%r=PKXr8jOF0TA%EOMwDrsYW$AiH)cQ_6Iu*lcj
z41RZ~`i`S7b*$!pPWF5JcFND$k0YQXF_U&}TDYTgCCXqjO=$@up$u(d6-@i(kViWk
zV$5lbi@w~wM)XE@@c#aO$jbfXYhzgDa$B9K`xG3Doxo=SSh%5;j&iS@4aU3`Gb#Cp
zaqnYWRJ>8B?ZMcm*wCAOiGRiSZk`Db(Y&IpehKpA^e2T6Iv?fa$hS@7DZFK#(|h7%
zj#&%lO|@vM=A1|*C)DUnaxWytB#f)x%Urre2wF!80Rfci*ME3pM<pBZZHSo;*_4zS
z@`(>3-sVu2eu1L3`OUZ8rI?=^EPETJ&z8^!oW>7fEA7?JKbKp22Uyo(2O9o3`FKSy
zfmx`vFfFAbEN@EK(jozP7U$hL9va^abYy%8mAS+dS7Qx(xLL6L6?_z_uTCD-x5e95
zxW#y-I`qg{oDV=qjLS*N&1nl|W3Uky0`DbEO05;l*rJ<3^6=`{I@X!xRp&kX8x8+P
z#ICRR<`2t(D%?q>Ts3ckgQfR_0s`*_$I>P`=7L<O$BY$}l#QYR@clm?Jz{I7ytb3=
zTo~_-0A)NDhJuX1-x<^58U_GM-C|%j+^?a-q$EHDit$}6Uv|!XzhBCtj|-^;f3+5D
z-@$Gy(Cs_qQ-7K0J@}Nq_1_j-1BHhRDU;KMn5)Tay+z9w92#kahs!XLW5#Lt9@E3g
zghf@lac~Y4XpY03o1Mg|Nn5ST-_3XbVPRS&n@~+z*4~s`zUQaWs6L5$c!=<-lL7&B
zY)QBAW9L_CY@Y7)jbr62aRt)>796`6BJ~+<G?Ha!g@ccvg)ki?=u>KeX8~nDxc<Bv
z-NWk9=OMK`XF@}uLjZWikn=Y32bDxqhhkH>6ir>9v7v_xNw~MRvpOhSVFF$;+Rw(E
zcDRHa*dP0jG4J8nPV;rZNp{U}&rgHbA;>gXISB50F70>OlsM+4PZ9F!oGxA~@YeSq
z`hF9A<Z1rju=l?^THFLc{QZ3zr<R$|3j6ji1fo+5gO<Zpk=S~hYfyXKL5qa35A9M;
z(p2xa;+bRr4p~G<7aN41)(rlElqj9y+Pz-ZFTy!^86kn?XUCU2kTS{CYNsAcR*_zn
zVNdrXmc_&)e6HU@IqG{C?E#Gt)Q{Kb^fH3&jdmR)2S%)z_qk@Y&Jh4fR3a(<&?)3I
zx8S%<1YW#|{B2EO1Y5t*F09#lM;@oE4vlDS6_Lj9u$-KpCXS%$C}&r+R2@(#l&h;2
z;_^&x(*$X!vT>(Zp$Xsb@+^8n?CfByb*QzF<$LIbXl9l8DIEOSXdmM%Q-KzqES}X~
zYuo5ZPKCb7BLIms{H)|4E3+rq|2sGX7NPq+RtN#xA>5Wo&tnOF%U*?s10rOz73EcB
zRmKipFjG*9f?zqbp&U_Qpky}TZH`GC%*E9;8pJW*+Eh!Zu@;rTevjqH?f2Yi5~>lV
za8*#`4YTOLX{V^-pWYO7q0a|nmC8bat8LJ?s<=w>n$;T$8+XOLT{kbeQ{Y*V$`r13
zM`1e@u`zl-(i1!<x$WTDuCfRPykV@N2M9(o9goVJMh1xS#B4C?V?p35nTFXUxBbYH
zFYssrnrwFJX*$86Lh8kVzUF<u=YUZ{Oz{;dAfef4O_ZgH5HyT7BV=m*oNH|>`!;l;
z+=;~SEmew19JFRr?VE>FyUTA6V7zIT6*7S>EC$wN4><1CV1>%9mEPMp<(50#tN}D|
zokJ&kRVA%JQQz(@2mx};c_f5Z$w&qYp_L2zj}Pqkl(~GhASIl_3kKgKARj99&q+ws
z_<43%WnL^4X9e9n6eiD1sBV#f4@4!3P@Rmz?M7*jYkw^<N$Ah-o&B9J`17$fuC`3A
z{*%ODdEN2K!+TkV%s;6$?}G1p{uB)?n+E(TdzssS9{6_A`xk9)UI+x}{U<0c?*1c0
z*1!L0y=WAmNrkj%ebpHK`{r-M4Ep*RRK&pCpMA&gX5AdEJUGGsIjHgtjhR-k-IX`;
zwTu^Ut|qS;Ff6@U*GB(BE$|f=MM>|+sGqd|scjM?xh0G}<}!GJpUmQ@NI~qB8Ltc~
zdv8F5HNuT&ddkesgK@#|W59^Q=K4!ktAnyCFPFxio_cjk4RQ*_BEkf-|GLbK9jo-B
zK%)t~uohZ$>PQdpV*pG#k<UI(05L)!P*xh&94@fnNY2Gv&RV>Lhkfivp+e+OZI3dA
z5}0kJCIF!E=O`?Fxa4iIt;l84VOOacFBmq0STa1PeO_{<nXSm&;=CIVK_S!Yf5ykZ
zbs_Ee?`_gU;o;b^eQA3#^a}+Dq0JY%T395(wT@1PSrb-O=<^utpS)ma@_8h`NcQL`
zkNf5H>*Xu#2-LVY>oozN;nkQwJ3W(j``vdcBsuLrZ8|XwNrUOXT5PJ<Dp-*Tww3gZ
zs{iNbF%7o^vPXav=;R&ia@0An1E0n`U!ExJsk{*t!lKS{8W#06a?N>70^qR5mrexT
zL8p5yDI%g&Ci*gLa4Qj?-j00*VnL-JvRPQ--1c_5^6ItT(MyNNAd%NsAHS#-?|!S=
zeT{caKwq6{OGhUdHR|zV{lk(iSacL4*FK5Yzd+Eov_}#n8rk-qzq=us!FGA6uYHy;
zZz!k{9l-f~ZG~}!@FJkm-nu;k6be3KM~jnb;Gra#a%pBSSAqy$U2_yeC?*SS+?XQ>
z!xvsk++4`Xh4xU&{WH4d+GO_qs7L2->ZTR-9(c6tc_T@HA`*dG8xcll&1a|9vNjD%
z9(TLr8=vR<P8&NWdgClXv?^hw<QdxlzoW~~|L#6hThx}!4O`OTWYeTQ$~E`@gZ?gg
ze(x*Uv2##npfJ8oBy~#jZyCElcD<4IJNjDGDCcD3{x5cZZB267>~8STtN5t^ibyA+
z_1|$Wr&E36+3mZAzZ0Kjl7*OLy8f&$2Y(b@Uy6BvIwzaa_#M-R6hp6m$5g6jKHBz8
zRPY|ImND?g_tF$7+@amktk4^(x<W5iWx1}sYIR>gI<)T3zIb;RGXGXA0rqBN(L^~k
zPNYgjbMge^7%&I1bwTyk_p&_&yKS16aS3|H%)u)j6dEc$HjJO;w2stfssd$IV{2h5
z#M714xwfT^oUy~@$g^$j=!{#j*Hym;q~~4;i!N^&D3^Q_+o4TV>g?={nnd4vRsZ@F
zGifV?%)|QmBseDn6fVu4JS~@t)f?f&^IB>x@}7n}QZ*Y9&>|Ou+S5VRO+ota3SDRW
z-c<F?k*RFSeeE+c$=nUq_|<E9fbtXX?S^%8SJqh~=CCC+_kTk{VQ`|j#+@$wa+3bR
zm_d34Lab079Nf{<R<7Arh0a@fc)$|KO~tn7J~>b_w%lr0H<7@m_psm^B82r59A-;w
zW6)}wb7s81{V^om^K7<AA<Hy?`|<h7*`me(0EuMo|4Dq{S39FWra=4X(ft!v7J#uN
zHea1f`xddcWtwS5nRMqPOUuMnRZ(4y8c8}lH$s-iaFs!XM$)6u#0k;#u`@z7o`%P-
z>Eu;oW56l1<>{VD!@243vIK0zf!u0-2QqJEHUd$IsbKVTYx`)<KOaP8cL%}+DWrws
z*8(^QNt?bDQ#7KuNFpb{hBqby3Xj7oQxCs6{Pz?f0A&X1nVW_t%Q-RGGpTP6B>#)(
zJ3oH&MR^bR5kLan>lX9P`B8PsxL&v%%i2GxvDK3dmhW8`iaN<T632D?p(I9XI}uWc
zev64;VvV?~I5-KIS>+M^V}W*~G%xDEWvkoV`vq^?cYh4TzQfNM<KQCBT+*igyEd@9
z`8BeTKWaDT9)6C2uCBoy-n$uDOcE^)3}wAhvinhl+J1rPq1_Qac$<hhAHY(WF;RQ*
zvj`t9SP?L@Z}Ft8=%`jw33dLg%ylO*CYM7Ip%JFrAXE8rPE-*^V%&`3k231_liFBc
zH#IjO=^4Oio-Vq+VXHZB^9a-|kdbD5M+wwcN1@T?U^ZvGn2oaZh{#%_1+_U+I+(Fw
z02B+qdWo)ZMJ86?ci&fYvF>N<y`g!M@{wkcDiY<8)q2WFMWXwvZ2SzjH>jQ2E>eK#
z@%P)?TaWgWS;)Db>b(2%{9mV)=b@zJ>i}G47S)1BfbOwsRxN~cCFR~1<*fwp=={}7
zlSk`a&UCUgT0DDfy(Cpr+P5t3nP$KHYEqj$TocfiwieJ{cjO^OUuMsH{=}PI`vu`A
z5UZ392gXI#OV=o_6_O8W(g&MV_YHs;c3R}FwG@p3QoZqs9H-%u@qVocftN8*w)|;@
zP8Br;jF5|$n~0@ibnF8Zild-f+7dO&k}`#n8>GtvCX~60OvCvzFXk(>5|tK<>J*NT
zkI_9``0kChpZ_c4pWeT-`920=%bbKadk)zPGrk+?3=D)K>asPs-wFWk`ZYMV6R%qP
zbIN}?&qsV^gnMKw27TAc-@w`u8`f+k^6fX%$NcpD?1&VZqkOniSC<S~ap3HyC#rzf
zB_&X7xtQl3CdDFocX_U*<nU7D6?Yj0E;<Yp>g(d#Zb0>LR8z0Z9)MD}r^89t)D|tL
zxGM8ouzQpDHKVS^&^V!;$N}!Ucj1Hf1zT+tME?n*F*f6s*Gv|S05ETngvwHl?tfno
zu6soVSaWQ6a>y_+tfJGa{UIkXQ;p0Zf6+z`kgl2W7^sLXK^0;qtb%|*;c@^xG%^T;
zgE$;DUAQWhLqM&;&QFK|v?BHq89|42Lel{T5_Yu(v?d3VP6^QC*?{?*7B&_<F^X_h
zLA9=coZ>AY&5VAZZIZ^6irGCAV?Axe3&(3JY5G}`Zh{ABzZ`e$=``Z?)m(r38q)7r
z;O7uGawgf>V>sQjJ)3g==|~cd!B5f=q;095MaUa_rD}&-2v+lk$;lq|^MT1NL9D3&
ziC7$hF=o{l%i1b67zuVY)T8T|7<;4Sz1z$8vJG+D7)j5)$-;q19HjX(>=(3FPV;V<
zct{_KL}uu|r5vA~EYf_eYmO|P>kfx2h{+&b=3q*mpEc&WY!v%g%+Vd9Q7cYLh_t40
zY=YPlifUo-Tx&BiF<~KZ$>YPh%<s9H6v3sJ|1fN8qea#81lZEDn5_BwlEYpi>sGXm
zcHZH^)7d#O#<n@9jcIOCn2kT~$7l+x@h@wqWlA5A>1^le(l;xak4ve2>Aykn+a|Z3
z+Pv})ngfoA1qU&58}#?DN(xGzJnJAAQ+wPP^(Bs}CYRM$bhRb$wO?l`36d<Re;lMC
z4^Aicgo(Z-O+1ga2;svN#crZY@Ul<OtV;eXItb+KH_FCRA(v=Tb&X4gNF8lNbc^Bm
zC(V9x*%m8`d}zy3;ETK|g|)qOQ86(wQPB=RMdqQOy<>M-6KH%E2#zC{-Kf#`U-VX(
zsXdK(f%rF~007i~f4two=4C%YA1IgVVwG)(&squ8S$epd>m7yVon9^-osU3Ju^8Ae
z7eNxkj|E?1eN`|^bw@7js7#LS@nd>`)hE^=JMS{YK^>*-{x>hdu}N4|y6&UN5ehvL
zXF}MU%=0-0^!@3-{eKQ?hRg645NiY$c-UeF#bLIHYPN_8g>me@&WIwsosWPPG_*W}
zXL(3MSlFmgx6{0|oxZS-Kb{3wuG{hAIlWs40+J6qs!Cq160scsf|v!OvcIMRWwBEj
zK5xQB{m&)1iJxPzk%qARat>=QcZw!uVyh~_(o--Rfde|tkofy0GB%Kpf(+kv|Equl
zrFPDu=OC9vkDRK%U=C)*p^hn0y%jU{k+kl-0LT(gAX`}b_i4d@5!uE7tQY2uNnb)}
zxYhI64=}a}-W&vA%s3*x+$6^=LM<&K_s7$e^b7lM{)Q}H_7Kf0>W|>hujjtYac;$k
z^E_KZ@42xMj6nwA@03bJTbAJY^p#F**9e8kYLTnz#t;3Mv<0N@jrCQ5OF&ekA!u%O
zb+r&BR4_9}J4FmZr5lo4{fy-7?Wo88s_9a>x*m};@T@jTu50;J)rT?e##B}#Ok-U1
z!8aSi-i*0~;A7`QH5QRal8A2-eAw#e?ryZSw6o&Z?i(B<vWil_C(KmwigUQs#M1`=
zZHmaoHU|@HA}Hj9z=CD9U@ER!*jn5s9})>7KF@(i_QFjBF-<WEBPfRbQ8S8TB~q^e
zztI;lv&nViO}xYk98sXvUN{Y^yE%$55*|OEcjZ4a9lf(b*vLzasyi>Am^P~EMJF~d
z(W*so;~y$AcS&4g_L_~lYrAih+uF*viS{Sc<<+k|8f&SGoW(g64(h;da!nkf*FIRM
z_VsZG*b>zZBfzVLMK({<$~O%MI7TwWm9ijhbSQgxg7m0T^^g=xMhEekPsqXk`*BV5
z$PCXUk|NA=9!5aUWUzCZwE`iUdE9I(DT#kFpNH+zpF@ESv0?lr7T$bA3JhhhLc8Wc
z*4kpBQRzPv<*=;`s`EBMi@Vq3Gcf>p5@z0D?RuoCkw98Bz-XBruMju!v55R<W-L8~
z6-Rr2HHBH}!=BR{eY}|Tn<u4>a6lm;=Y66Ins;hbAPg2U{J+8O9^L(4F|*|Fo)2w5
z>v<-?5_b8`8*<do4F!i}hZbY#X@sobzPhC&wR?#QwP;oQ3b>}fJi&eAe{BQEtS=>Y
zs1%djSHyv0?V$(3H|q+BlOSZByNgyqKrhVz7xP!kp5)k25dF(kr{peg@0ei-DuslQ
zNFa_YrHJ$LW~pE?3@s&c1c+K%#m*2{+5WR~Y}HDO$*-{e|JA!#0HFHjZ<F=<CJp`F
z>^qrdxXr0W)d%kc?m<Ij#?mZLcI9ym6H_TH@J4R-rgTx(Zwn*7--3Rh^yqCkO4R|o
zxN?6#vkwSWi682sW3<}(CbImhCvHRczM)@kwj!pc8XHAKn}&!te-L&q@szPuaxCA*
zwux&Cs#9m*@MI{c)F)E|O3j{VJO=B3R!_lTVYk0ZzdYNmpC@>;_{&~@3R^8@b%rtc
z&!j>b7ExRoY9mun;%%&2zLm{vZ|Xk&a(-K7U4PkVV$ylu@4RNt-WNSXIW!az!i-a^
zQQ0Z`(EkF29s~I~60!}o)gPDF6U`YZ0Q#|7iu@^lj7lXBkxHmV`bgqvPKnPyS^dv1
zC5V)TLz(3X^x|@8@K(!V_Tm#BMlam{#dfcoqUxH5jo$Wi@xu2g2nQ=Y8ugMo-NQIa
zsy}4v#lbwNP7T>7*?uH&tDV<<hlsOVD#<GBB>b92BNbtM6NX4OU5Uq~#i&yc_{SyY
zBHte3=OeRlrsDu98px-KY+^?;1kNISqW@(inJh#RGzbEves|$v2QcgVDHCF>5}ZYN
zwAGV99!l60JDOlRbw0AI<MEX3N$ZS6<((|shDIk~(dcSS{X`i@O7;*ERN$2hTc_U&
zZ9?ZV7su^53cE`0;F)fk;{hxHZGZ?8i`qF|1A)~S?}eoh)wMTEaB~XmAxb%&7-R%F
zPiS3|2v32c5*mkV1VxT}=Lf(6Z46&=Lce@?(_lpa0?U6A_z4%6_BSM#M;9%aO6ERJ
zaqm`KT7AUm*9+lHoz0v-42aX^zs_FqLz7#wfi0eMD67MjLFH!~kI5?+2tuJ0Lw$q*
zZNi`5ji(P5)n+<$aulR#uIi9TB9-~JRX1e?KL#>kw0H{;?*DtINFmw_19|_jRHT3q
zE||&pd{tuV=rN7Gd<O;w@}vp)i7ij)%KW$WcK%SGh?BS+D0^md%srYGScbQY#-UU!
z+J3~Mx+fY@%zH_`7;N`yPFiUYiwp|u*iM|PVjX?PMhqChQ@k<CEE6gB>Zz@^(L0t1
zoMOBQZun&2O$EN%@-`(~v>s<+n3Zxas2QX?iYzQLsAu}wTl6dDh7|0fxtb*vq5Wyx
z7+cGxWMdQfa$*o(Uhm(L*+!#)p6q(fhn0M~nwy1a-gzg`Br07D74xdfW-n87>S4b0
zF*j;mV@$)X26yODYrcEO6xVg}<u+8i70uyQD$Jz#y<Yl$mQ<7-2zi8Pjz;FB1s+0M
zge)M#S{^Tg{%L0fTch*wuHdjCl}$rqe=T_`-hAsXFK2p`V5!U~ZFnpCm!B}L|9)K~
z0uP5<iDJ91s@kJN^!bF~P`1@1Y?s1Q70&*WM*L|wyH`}YJR5-C%6@hDc^z!j3d~9a
z9)pQAAy^a^#=xHZA+zo1SxxR+e6Qaxav$lSRdEyEA`Tv_Ekw-n9%H8l9cd_@8r~A2
ziD*vj;A>2!Q8{BO7@NO*788D7^9-=ga6Skilnq3s0ZM%OPs!#P5)dkdHRdwO{^x9b
zD#-w_VQ9>(g0RI{DAbh~{&;8Xzy84eU}R=J@2pn$%{1i_?YVSEiubg{2w=&?5i_1>
z3#ObNOb5t@%0X0;pA)~a@9PD~W7wHaJPz-8M88~|M|IUFqC5GH|5yS@sLMzeKeO-q
z#zdQaYjnWz#KRM|m3oi)w7-Ed&N!V9tmb3(l&2UQO9djv*1s7^y3jfTK=@$%2%;Tu
zlrI(4p_1g#Q-JJ5I)%wBm8NM;ZJ6Ff?X+STZKBp<^RuQRrdY00@q(%GVr<Pz)YhEi
zV0GTHv;J^Ax}O?0tPo9lxYrR*&>W|&z`s|&-_TEBEM1e}XM;zHTsge-FbMep(AAu#
z<}N$*%2hysQCx#AJcUxh`RWHoMviL~iLEh}vB3U%X*C^+(<TDIJ|<4D`XnKV*nunF
zZmDLXyQ}#X!GY-)k7g{MYEq{{@aX>h^S=-cXpAh{=(g=iT)!jEfN_={>BIrXG4raj
zx`kpwe1b6T>M)|Sj{qP~*U?zSmD{%HqKAz<p^__~I3M!W|5A}L;8a6Qdtc&Yquv~I
zF!Sgj5TWe3%(?nIPR||wYN@=o_DM=qlowTMij%O_9J6`HQtAfc+H8tHNK~AT<po+1
z!&cF<*1GC|t=hVAa)gH_b1L*P=lRJTtBItYk78lH!YkL<A#364@$tZ^AV;-K6~NAk
zy|8hCW=-)V_bzsf<Q|6TIVq4hYwCC`Ei`)k1|c4OJ&=ZvSbOQqMvT(0TGY0UHXSWQ
zmib=d;B82h@-;D(|BHLd>ZLIdh7U?fxI*vLn(lpDn82h$%a-X{Ncu5P=vbEkpOk$s
z0IETUO}PLI^$Hklp$Q69h%VHNZY2Wj0D~n3ww~hPu;u>+Gdpa1%1(Je;;BZ{F3Jio
zLUw|<ZFWt+zt+iMHP_!EK2qUhm>V~CO)N5F%fEdoe>3sa`Kgq(ds*}d@`cKPA^PgV
z-ut#{pqVgsFAcwRg6hy`9kR%(ojvoi!<Hh-B@OzC>CHbpffDMM@-&9{)t&3M$#|)9
zZEbDW*S?1O;okzqB)pG=)mlcUnfDVSrj0}KXz2aJoDzUd-lp+vgE)=2x+I}01oE~}
zW-4Y-=z^VyFuN$1g|8kBtg_2#3O!Q2_T^I(*97*!OH6zM(6;MnZzlp2o!?0m%+$3;
zlQM+&3_W6H4(HhsKd<v`{xfOxvAw;oyOQSWs3-0}ve|!rQhXC0Xf(R;2Z<ob(HYjA
zq>Nqh4G(*?>o?SuOoJYxL9f4~61IPRa>>JL9`8nmjR6ZL6e!+X4)nA!8c+W{*2{oB
zsIk&{8_VV8e(M+>UJJFwq=ReQ=T}MT@RFa?z3o!oM+?KlQK|?oGG!qq+Pf^9s+E%>
zUCZ{@<=?omXmt<4oP$C92<dE1>Q9_7zO49adOB1q$Fr7GYbi6{tQh;c!=vh3m450C
ziOy_(d6Mlhyt1bg^zTN{0Bvz}NONPfAl}cV_e?ViFPKiyr`!?W;_JOizosAUe({2k
zh^92NfjN3b#qIJ%2ndPBXi?p)2*mHmhVdqz8V~)9o}Y@?jnR^kU1ICBiMbz(@nlw!
zbDcNr!^i2AD%DM4HF-Yb0&<**O5Xa1q#6&Z^Z2KnOisC6B*Vt&eUiAHbaq_gRFD9u
zu-Qfb6;j7<?Rd=BI}BP@$gY5QkO15l+~?&b<Zf$pF7_oDgs7*mSSGLmQXxPHk@g+_
zo7<eo&fkB(1fIYCF(}K^yzK3u!%2{@%zlvTgj8%VG$MgTs!Y=q_urK-g+xd!pZ~os
z$z>>hc-&l0*8Uo;g(uu`4)aC&h((dmKs}~YV1>X>SM@(P49pa%@}K>!<WiEI&M0#)
z`?_OlStq76e<c7oke;+6{rhF7XpdXERwhV1u-=YZiuc5$$n;xx;H5OClWCuRuP_~9
z13Z#F#5{IJjjE6a>_&t#FuzrqWq&OGbWW~FVwxtO&Q=Tk?#-9Gz`c5M>8q<-WAQh2
z8ZZ|Z7gA7SRK&*Jp_fG7)PkLZ%TsWMnP9#qHhyUm_T^4<3qM6JxBLlAy+W^!8u``K
zG;i-YXIktja?*=gCn4~_YIHK1jH&)O0Q*`=F_NW9N>m=G6e^#HbH-p-J^{@bUpPnE
zKYoSF4M()WQwDMhDg&z~P1d+#L+o;_qY{)cI}E149*@NFORy?GPQV}pixvFc<~)Oe
zH@CMt1X_w`?CWb*ms5SB&it^%!}$g)yM^)Rzu3i)IcsLcM`CG^6sRxl;iQT1vSP0E
zlZJ{n7-v0w5@H~c4q68~T+BPX?f%plvGA=4t@y_Kl^57d4zX&(dM_z5_&fMD{-H(y
z(f^wX!b5|NwZ5<Mk>z-h%+cymZ1`v4dt|-C2PRLhKo0h+M16L*6}dj85#o+vJd^qe
z9A7Gv;2hJl@MX5PK`074%)$YlQq7*L#@iTtkf5j#UFxBAS7T4x$Ir@uN&9^Ce9pHZ
z(Sj*E3CR;F5>JvMGrnsdA86UEC@czOmC)#ryX(=IJ+oFm4O5wRwts3T$WchNQPD)P
zABu12ot{I(=BPqKB&#PbVKi3@2p?>lf?W@_tx_h7xEIS3UDM9dN(Y|jd&36PQ#owG
z2#;}?L;iCj+Ab~w*UkUd&_U>*7m;R{vMdeC-#W*ly!MyGOyrQBLZnZ~xtm;K=*fA)
z#xGBq#vlO7bS37cEyYic_1?2|jtMessF^zrM#JsDyZh?GGiSS)bqg`}<g?|#Gi`y|
zSUVIgAeazm_I|7EMW>6w?9i9m^9WL8Y*E#~h56qSHYrAGWl|==;Y;+0+;$BGD?m*X
z%T6cu*u`8;3)r#L=*fQPm2Lj>5Y&b~`Kp7OP6{8`>Q9bV%RCca_^f6-$?H9D=6jl!
zc=#qsWHG$-o$O=(%Z{PI3jG%c=487+b<PnnT%NB2Xu|Cukvvg*^oj@o@_ueEd>p^t
zU>qq-a`OW^n+0|l_xcZgw1tjVTk<M{vd)&}rwGRx$Rc+aK>&@<7*7HKN<)>Qk<&MF
zQw0GXPV!~8iUGm))T0U?&oGxc#at@6^JB?v&_mCg0IiN6+$ci`Ao_b)3`lP6xx<yK
z?Na6t1AP4}gf*pcX*j(W#_em6Z}laYAy`d;WK~si-@0=9ADZhG>8!WS7aC{uAIN;k
z1bv)}$?})^XJx+E-PckvZ3J>S^76u^G1yDAqA*pr?~+xNdJc_|p~qPb6+SqU5HDeF
zcG?eXmzXH#ok&_zw{EfX%hDrfX0z1HNl6f*k`=EI6&}|9{mL*w#QLO?w|xHB>cdw_
zI<6Q2LRKsLHFSQykX_r1*Gp~`h<V76KePFMCFk_iM~1BNG!BYQb+)<W?plFYl^mXH
ztXYE{HKwdcXi&-QBNVH%9UQ(47GW(3y%O6AawSGDwLPBvByaf@Q>kF*{9$ePLZ3Ko
z6lCX2Dn~G96i*0?;QO{sub=|6-;J<s=jRY?<^TJu03EQ9@v5yUBe*kLC8<h_>3!MS
zPvF@TTyiiC&Mc`kK&4u^)(fDf1md;p!^ov;DiD6<{*neY&QmX+cR(WNuMTg$f}Nc;
zyd~;dX`iy_=!hT4m0$!phgu`x6%;}hze<e@>J?vAN=0U4B-=krvL-`vhl*|cCipse
zTc+NUp=JKV?&0hL?=a22X5y_H6V9EODXQjR7gNGI;>FU9IsUziSy4DUP?IHebR|MG
z0(4~4ZB8L^z#<z12G=y<o!A;!cT|U(jTjm>wb?@7MlhVk=cg%mbE=M=5+{J+b_yzh
zMrW%vn4Y&WGYd~?NXo%L$;A6QC!1$?qJjuCYO-&oPBXc6gBpOxR)NtYhtU9@YL8DF
zAM9PP>O`cyqQBZPJt$sEIw*Q7SvE5a^3JX7GMw^EhO*7lunAH3e6SXDsCo+^1ZOOM
z;(5&el|2m?3@}z0PWG+8#J8JfiI`pRUXOajDBk{JZH8M<&L|ll*mrAnJUaJ`uagbk
zqVYeIK$~+`{E!nGz1+EL#kk7hX|^Ep_^FYCt$v2%vE%XlRTD2(QoKzs4uv%FnriWe
z@_nur+U9%yj7TS0qP$8rdVxeN`hTnaHl&YKx=kMjzx2uTiqbOj)aEY&ooz{eq<FvG
z+B`0wnXHUY!%(Mk79m_rbM~}y5$IC>$IGo9LA+2Xw0RJur^#*eT=n3;VI0**5FRq&
zD;ktKH%oXMD_&jbIe7U!6?=^Au(+UUR#7bxnQD$mLsek&FvW%mi{q%tvrs(HY6q#m
zs~u-cVmY5K75>4VcV4pSJNl}ui#9W=qA~xSZptz5X7#aH>saZ3Z=2ogE75VSht?Of
zd-hU16JPF%wHykmlpP)oKKu=9b<X>sq6A_q?Qy0g0W^7dA--$G%@KO)_{lFs_yVSh
zs-FHnVT!Vnr?z~i{JANk?u1@JUPIZP$?50^^iFZf(Bn}L4giuHP=Wbm0|XKDlMV`4
z6D6jSP#C?c!j<2_h;!N{-k_I2<rDy3avZtus>kZn$vKLlqZB>Xw#<aw0xIK^IJ`zB
z>$6IB_twHtHcL63_Zd5owT`M|d>nCCRD!cQ&(+V?r#I%_|Hd5aJ8c_Zw_Ux~t^Gf$
z-a0I*FKQbeLKqNXC`AG321%vUAq5Eu0STqMq~nkh(kYSx1JWgoG$_&`-9wAy&_nYb
zf6w!N@Adw}1<W=3?6db?vDdoSy8C{JU?);h|H(`@iw<rdpfNP@*|yiGocPMAh~RC0
zcW8buVbd*yc392ZrwpL|7J4sXWZWjr=Pze(LcC!2bWN3@?xu7mm`m@3{V1^6Ds-GJ
zbePlC8sbMJzoRzOJE)5Ks|SkgH5(5m>t#3#{^qYyy*Et6cZ4DSJCX%69mnQ7;V#dg
zf9a|+{F@>g^&^RYbi`T9QZ|J1WAT~0NfZ4fBPmwM)0f=zRE&Tgm?%gNJIbUYZE;(-
zmiSyI`Q6{rsot8L4NnL+Fkv~H90UAu%PHl=gqBvmE+y<4AMb}5f>g3~Y8szk-6w_Z
z;!vAe&(#w@6~lIbSf>Q`l-+rgPGU^qPN~_M85R_g8;5$B9+sf=p*6%mKe57)l5Kfj
z@3^W#CPe0mz(e5D<eI+>6h>daSy+tJmaBvYEa3O5$(rVOC+qy!OKyXhh*@jhqnNL&
z;d!)HwCty9y>(*nTMAip76v6sXC}wEaqT5>UqD`S$*=eGUuHRGOVrHFb)O#kkeD^e
z_5C?Z*wc^6S}I$>7wJ$n{#w#?fsWrx90+?3wgEgM4?wf6YkwdFmg-85?=;VE|KZEG
zsdt-rkRieosGF`V&2{$zooJB+H>t7wf6{Zq^(=JjLE{cXYEi`~ElGg;@5d+XB@e?R
zCP!rzE2sT+WrZl<<D;IES#LL5(vW?fv}Aa+>L!T9tmkkZ90~!*qbE3@D<D}BvmWQ|
zy-B#hawC`)@^af^%a`(w7Zmg`YWl)7YO0tc`FcCatuT|1!TRaDN*`C#uCaxC%mNW1
zA4d5k<#2IBA|BHJd4K<C6Lf%CnF=YMD=}<*`cV0aw~)wj{lDXpWjBr#m!^96oGyf9
zX8bO#|AQ-okD+m&8MXIfHpMH6TNJ<5Ft`CPMSY!O7-VWDOX66!O4i;#w-ax8WvOe)
zT&qG}D#FK*;2NM6tu|JW9RC}(Zj1@8!X>Dk**V@E^KDY>4tqtrNfnZMZw1Spsl6gA
zm6y;Wc#BpeMWttJCX(n=YxN_DOl3);TWUbnIH??(&Es7tx4)J59{RRJS$*8L;*%>d
zF#j>|kn>^FKLM7u%RiWXVSMuW@fT72)J@8RA?$+<g}v!GtcZv*9mA|<!qMCn_=1<I
zDh5D<-PkfRsCAy!=%&1rU2iXY%6VCELkrRN;@l8=S<JLR{yEbjT1Q2+7D_>#HdkkM
zQlrd)^-jmV{Ab(hEV_737ZP+K_Y5%8emVN+`=7`~@6Bqau{wf{`Dbb^ZzkOZb9c91
z-RTrH)X{nRIL*RFdx9oD+L#iHs3zX|^i<g-xc%W}p8l_8L_XCC%NNPC58{76=x6z<
z`dejmrnBU**heX8O8Rgug@4NWXKO&k`&iw>h1Tkyu?7FJSo1&X80gbI6+x!to6Piq
zpL92r(_H`iB}H<`;eWQVe8<J{fApKZMVy<V=*7QpbNOb@^^H{iH}2}ZZfwB+JYeI0
zRs8=uq$Dxzb-2{=mkmj@_wATfb&XSfdG;l;2>kJD8ZcY^yLfwpsVpxy`hI1E_L)l2
zGd?$&hCx0THms)_K6~~oU>_gY@4%;PK1*u&^V!b$q(sl?>LMgd)a}ptFmPio;N2yd
zeK0c|p!=W|(}MON_C?`97!Ui~uKGbhv-fCHHpmcpHT<Aq`0WuG!hX>5SAO;2>;tvq
z*<o56a24O}gFCmd;D7MZq`9u{WHhbyxT0zAg7?h=5H2Z|ed&!k{gdG&^|iFMsY(3G
zzn=#wv4sK~sCAqSH)Y=LO!!|)c-AfZH+Wq-`RswuJ|YenKV6SAy_E*`;#}asuq3AW
z7{Lod1a2$P7f((;xEn3{9SwphPN#sQ3;@g2%4$O~U`_GgoBj<C_}hB9=NoFY5OBGB
zyH9evUmfQUG9(JO9>l~|{lOuT*k6^tS^f839e0hLug(tQfU8v}ub<SArBp-5hmO}d
z*E2biTVDV)*yiW1$=MG@sl$Yrm>A*a<$K+6bil|*<Kv5|wwo6w7i5s;+iL)Q`?2lp
zBT2x1h&0Ac8go(!_)l-81#G4jj&1Jk?Fr_b51?;v&~&Le$aC=JvVHd&Hj;9#*8Xi%
zp>WG)vzg=;3UykKTmS*%0LBx*OjHJ(AOp6Mj}I8GNT!j;ZqjJfzm1U$0H?d!?bLey
zC#`vn#pLENE$0|0VB1VYww{##)Fr_b>@^<^g>2_|ijsJ`mu>{0E)SungWe!dLpb2I
zpC2{*6S?CCMAKg5OYH~k?(VL3$Eg>ME#6));s=6%{u|ZFXDkDCn2PkfIm)?>Sv-pY
zv5%MFOQ>p6RQm7QkvM{df#<2nW{a~kaAkKcfmZZoMcd8AgXX0>IcJ@0kC$_iivgF4
z(ibD(=q4_|D1l=|K#2P`MEWcQ1THN4Y?p(lM|9YGC@fqG9WiABT#KU)VyPX!FLS7S
z_xVXJ&d&Nwlp4G5)=h08>CSr;qw>xobIu}>ZP&YcZ(M#=O;ld>RVG`UCrO_t<p4Lw
zMZkt<fYs{&)E_X~!`Xkhzx*?)cWtY*QVRV#t`RxadRhmZrvakrs$iGy6Yzoj!z&p!
zi=)HW&Az_iQ>3mp)j|CI`Eil-^}lugx9cbrD(>cQ+|9W9_4qG|;-g7vwCm%&EnV<C
zn03;fjHCJCOA-5%@(W;oy}AmVh%|_nD@g;oX+4{-)_HmR54!_)x>4stp7-TPgw*Eh
zE%aN_{^-jfy0f6Vy1G)391g@GftcE#9~cjdfQzErL*83~oue>bzvXfyijxr?fN831
zKI~HzxyZRa&%tSTpEAPy5pMJL^u#PlNgqV2Q<(*Trv=c14~ZX#DQaJi7fFF5pJZd2
z6{S({$JCGu9vKnMI~Kf6$D=Q#Pm0w+x#jg4a?5seCclnsyO^kK-Jkz3sGcp+P*=C!
z!j=barx}TNnc~9SSkBFHNy&%7+wR?i<N4txq`Lpst3~jPM)u%KpLLjA{W1X$nt0Ah
znH3;;dp<1vz4y7)|1Q$<R_kC+)9ik$M6kbG_T}>c?Mq$1e}a4)>)E_IY3Q^tKTo5;
zAc~}(@sbD_v@<@ctMqpO10A3kFV!RGUHf0<T#PTTz89N-*jB%`cXX8cA!~9FrRR6{
zlP;j|RrcX2coHmleP^ng9p<Kh=08stFIyIU=NyodYhQ-N86oqa=>O9r0lQNHyH>!d
z)v$OU7|*NC2B#~r`j9u^;3K+WBRj+-3<ja^I)k#E{}br+VZT3qQnz6yXJI)vn>hiY
zW|C_tP-{44f=E<Q?m(d|1#UKg8&cw9>39E5&kx_r^kKm7Pi@Y1*@qx-<L(02Q^0lO
zc7-pfTlP~b{nnQVB-WC|oUV_aY>-y$&0RH^H2+(aFgpDI7Nt7>_G&{K_<2uw-eb<8
z=w;R@3qvTVo}^o^Pv<fUy-#a%{8wXyZ*~`NcER*0rcbZ+5L`jY@4K{~^DcZm-m5Wy
z@Xa+Sz~RzY%WXH?ZLP~6s6i2Q@>`{r-VY+VI(51|9ljbEo(I4F^YEXZ_@_1&PbxgY
zCdaTJ|A`=QBn%wQf?@&|S?lFo?^rE*Z)9!Iu7eigRQh^M+8<o_*40???8D?DqJA)j
zbOg%(`Z_Og4$cBN_@_aaUkYsafyC*c$S;Bib>Pam+vj`B^&h93KZC-HS0ix&#~)3u
z-sz{tISbz??igM)n0Pr!AAce7AIA6%SDST1mfS0S7h;4_XK}Y@ao{`v_upy~^Z&bc
z%N33%^Sr25QE<`YY!^$0uWv64WQbMQr4E6FIEiHfb+1EEr_OIkVL-7D*brZA0OQM`
zK)AcP9ql>+O%XeLdpG~A0SNzkVSJbfs#?s_H#@LUV-PeVRl|)HCg**6FN?;&5gIUH
z6AdPsYqy-s`G10t#}1Z`l>AfhKcz?-S3SW#4jZ%=KbTtg+$*Ukfi}>jzF+1`8~1t(
z4)<`_s|z%Mh_|J4j!nCuh##F_H4OXjfjgX=wSZ6k>=s;Tk1;jj7=?TgCE81O{a2Ch
z@O4}j^52vSGaNyFBO7(!*?)U^Q<U=?i2G+k!!W1+mYd7?fbjYhG_H*YUDRNZT1|N2
z7>IH5Gn?L8&K~#v|Mh-_2mkLu;f)59GtFJY<AzF~u8RNlG#-cFdG*hg0X>wE|LbcA
zUw+zIcW67$2AYmOhH54Maup?)Wi*RnGlMg43=BZzp!9*!L{MhzhoLd?P>4s9oZPR(
zi#vgV`nmnU3eGke1onG@fF7F(g3U<JkmtxV>cC7dL;vGqg#OVg<ahiX89dnAKiov^
z1h{TPxE!)v!9NrrBKXXi(O2XZ^M0eA#+KhjYmv2rmC~2LE-l?DoF{b|Au{7e%(kaE
zfv~UQW{?&<`F>dVul65}{17TI8@O?8ErF71aVQ5)!YVnDb=6Xj(vo`r-UckFOS7sw
z{Bx&}k&@41Zx+#4huzWv&D<^SRc&qRaQM*WX#C*7k4!D>o(L#oJFZMHcROzMXSVjC
zb%s282;{T@2TlQp$uQ%>-P72&4&~wgtnY4FNU^c$p^!vk{OG)wlvpBG5VHHQcUYEq
zzyKpruu{7|G%)@SHk5!NV*F(E($%cx%^vONrPlTH-b=J}!{Yo&-9lx~G=cqHX*r>V
zbb=S29=J-*r#*oYeKJ4y$g!)SQ0NwWka_tLYMnc&uCB3Su0BhZRmURkvGL;keDh8X
zQ(Q08zl}THHH(YMXCt-i@%@<@ElUc>nl7t$CGaPt!?2*FFkB*)2fNseSXCD?3n^mK
zaJX0#E?*=BN;Vt&<>NaUT-nfwXh?qS0~v#|Xxv;TJew-@f%NoynhNyaIE@+dnDB5w
zs@Przi_lQ)Pqg$ntPuV+Tu6Kjf0OTAO7VmKg_yeMh8cc`$8Ykg@$_*a@MGX>a6|+)
z{Q%#9;`@T_E*4zK{lEZg92r^o1~-JS==Za+x?T?=5Ju=Y_soqdrlfl<;lI$$X?1tk
z$XzQd>o+>$;&aTwO7yBf@E#_#(_=-!0_hV2BXao|VUW8F48#aSk<Lb|TL^P0e+`!V
ze6C9~1dFOuYAg_&9<Cq^u;mf|ls&2H9Kh3zCdGQ*J}|%trH3xO#_IY@Arnp}Q~d+C
zL6$4tZAdg!gLn9~87G=ovhiSP^LlroogT{g4&waD9D0|LoDAnVJ|h;H3KB08niosI
z@Y*W#iDl~exBl|-=O6gKJ&mxLdxi-8{!aP7=?~9<em5l`910A8twqA{Jc;Ds`ky)>
zReU&n3{XNWEG;}s*w#Yh%D64O;4w8tr=kK`Sm0}zhEYv!qq2McBW`ld@FOZXPERlq
z{iMM?wESzhmWuwkyy3Db%Tu-Om`EZbXefQ4J?Xvl`w82P*~8TBPwid^RUdn3kLww=
zz8zQ@f5}5cPXx$dw_D-U$xtNLewR=RU(SaRMnD)g)8OXqxq;TB^cw>sV&rnzAaVw}
zymYPqpXh)5xAuZ=Ix!Fit1jZ&-%6A*)d}Q(-&F;t-y!`3p=EgVY9o3c!4l4H4EN!U
zi|CPqD0K#6L9k*N*yOO~m;-}la;ub`0t00raG(pqcsGKN%>8v+E~_1b)qPFv>%t9*
zL8jYoz1vk(lGS}IvM&*&6fZjr<bDoOI>cKrDNd2Bdt__9GB7SZOEg=3wtk*xgCj_O
zH<%wgIFL3I=WFp>m<F7{H)oM0l_d8xQ?wI;{iEg+Z&V)C;qE)vmEQ_GNe}<iVjc;n
z0F|sTy}W(_ReE}QFAM@rgkX_kF_1#P$mkLhR@XXjCJYYvVR{h}K~l+d5+|_6D2RTa
zs8uf%FEsi9iW2#mi6xK=;|yd{%3T@hq9U!vlNTVxqJoUwV^D%z@Nn^3Kb);OoHBBo
zNQj)@FW{28o~gWAP?XL{<fBU9{w)URLCmeL_fYRVTRptYo>vhJ+E>q;(5b)C|AzG#
zGSpz>N*v_>luSlOX2iSeJj`YzU>q*6HO<O68sj%EwETA@-?!;9l7WSQVg2}%MD>4H
zONG;E5fDeKdEd(8feIM~M62fx1R@*JlLv{SdPWohdm>ryvO3mg=j(a92Aqa4kOox@
zD?#DE!x|S};?BIrJ_lxe$EH=rwpDjBJumB<lDHFpaZVZ=O&Fd$ImtsTIKHS+NVT`K
zk7JlSVtPozmyQ*DrED2U-%k$>riYS=WFLqmd8PM|->F}(uKPC5Q)j?)qT4b)Sy%h|
z7w7${CZNA5Gl?!uC)EthKlYBd{ap&CS27O1N<GTL%ouZo;@;O{PnILZ3XVvR2n=M{
z`OA}{K^O+HOt!TuRG=Wr4OM{WMGhrGOjkr>!?fSrq5B41eYZOOMc4B?CaZSpMdgcb
z)!<JliJ%?6Qt<rFw-{Sj?SEIQ>f5WB*u=;sdWcU*l=mSWgp}_`D3BZ|?X)NDYQ6{8
zm^W&L$Uv9(Mmr6t<w+T9*wDQVOOgvouTEDlT&}<0+`ht;jND~Gl&ZKWQpXZ0F$coD
z^AK=Lf_J0NOf1BEISWt8?P%Y{4B-Tt!XX{aA)c@(f<hiyh;lK7OMK64F0A9+Pl(3e
zzT(2PoCn6@Cc^*&LNyvwpXP};jL!*_D{8LbuydeoX;QRkOb^rI>EETwLW#FC=13k(
zRC!;z>Qlx^HhX@Gt1uo;iD-U6Q<lgG$&a_oSe=L&$>z96F9Ol$oq4ijtM`ivtL$V5
zKbWAmiK%d&!Nk)*R8ZZc^js5$$2KOBz($B0^swMJ%9pVZs4<WlNay#u*cKE3-bh!_
zO~q~VPPNz^>!+#^v~#olmEdPoe=p#vWN?1~8Vw=K)qR0^qS{?DnkVpxKS!xo7K#NG
zmrE8GcmZb!d$(`ky@4MBUIlk~40VP?IpReihoQk1*p{q!Q8BaS#e=qAwY8nT6&88E
zpd{tblc50nU)&K0k6e2NHHTL8UFHvdJJ(1?ET9y`9i*=RUL*BPcJD=fo781*X2X35
z>9!C7HgyG!2lArrStuKAr}u=!?%jJ$7sIvP!{6xa8Ms0B#S+c<%xmL_R?93H1se-p
z`1Bi3E+UWHval#Ee2Q(YtPWE+0y35E=g^3~fB$I%a50EWdOvM~cAt8#BcJf_!87Go
z9PN$4J!uvMfd+~YG9}gFokn-N^|R}4@i9RQ&FnVk<q^flF?pXvaPvIXTQ8@<Q|!VW
z{pg+|U9;zDaZxsQ<dg46l;6>eRnGJKQ<c0(gWV+Qvm0ksMLMsmnK)^2<ZoGMeHn~B
zl%5n^e{B^3H|P5i5h2Kl+&0NA2Y2Dgj&OFHUS_**+gY0rZK9tM#0xjxXH#Rq4IvN4
zHZ;_sDp)HS@HOBtA%6bULua{Heimlv!1yO(n$ZyVPzuUGib&S7?tnw>OH~Nqa1IN4
z1qOr`JahnsFZ1Qw$gDP;20PUBV`ea$38V_pBGZII?mx0%ZqFa55r)quG%iEzvWk_a
zP*)w|AXo8=;oDBPO5j#Je*^?{F=a2U4hLNIjV4ZBw72?)IHDIw)cwJ?;?O|1VT==h
z;S<yTR*CTqaRN3b9vFKaOgOd3-MhX5eN|Mir<qwSU7K_w!5zi5i;#|OX}`+@fGx)!
zX>hv$oMXCIo2RyI%WKR2WZq*C*_hppb=M_qxo!6^8ut5LX0>`CV<%c?Cl$47_hO2)
z-M5mmUOdaks)T<+-=t+`Zk7n~AQ06sd3+oaC+TyulUTbx^3zwhf>$S>ECMk@`T)vM
zkc`a{Wrj#@@8!H7J)&Q}3$~~PxRbq6E!E>T_3z8t&g|L_&)kNg(Kjnf#}A@DK|{>H
zdpk6Y*jfxWQjC}Y>x-lwf-a`4GI26sRHLIRr|qO`&Z+g_=aOq7A5HA%@@G}dv!ZYv
z3lNVb8W}YremMJysXKPQ)!ojwX&B0TGhX!ZN9)CI8grE?E)SPwJoj?|CZ~$p_J>Sr
zQasHVWKxWM^WAwww_?N*XT_uP@`rkh2o=##rbKaphDyv)f{g51geLnKHIXNJ1_`*_
zPL%4Urac*W|9S1JrjwCD18MijK*i?5(Ixmf=0+T?E1&&~;B!y^T9Wj^`9f1HDno&;
zo0MEFnT0GN-c$AE067pADFf?K;tq_kDOrf3oc4<+30KK`HjCd2JjuzC;bPZ-G9X|I
zVqZ#T7M>#3bBBd1Hjs|QD<UN-)XJ}Pb#p1$10ROrGDy72{cUYNkjJTrbTrCIUr4j4
zTf&A3j=+CH+lhkYB8Uc6C$@Nu@Pv_8F6Xm{zdojI{2`G%ZU<KS`WIU~)}yo^(MPRV
z`dU8et#RQuZmm1-Nw9Hc2~T{kALCoyp^#@OR@b~A)}42l!ZbaCePEP}w>Mn~S=#%=
z+qrf}N@EGu!^B9RA2?$D+p#VT1{YV!^mXmKIibVrC@%UxSc0pZZZyzd&8Ip0`LpcT
z)XL0lC!m7bp(}+&4&CQblv+5_;I@&^&Z#0kk&d6Lvwpfe$5b%cFI?^pTq&BcXlYX5
zpQ0|oILyhMUCXau@_Z4w0UMascZtuGEkkclJX^`83Yie)bpUvIe*$<h6c0;Z9B;VM
zIi2qOxv(u-I7aDORD@B*D0ZB)kM3l7)^oc8^i9<sc6MfUZPhGFPq-ha&n~=wS?1t)
z-agALrn0KU=EA~q#-~6}t)QdHB1Rt2Tvx3fmz0OpJr?fsQ&O@iC)R+Hxr}p0@o;b+
z>gF8=t%zu<eC+&Qf=vvQz<J0}jv>;LPk2Up32XM0VyBgN`2itD;Iem{8AjvE>YTCy
zWX8qGx_$Vn+dQW_kj+)%siN({5EOjAv~c$B;`!bja9G6abmGGcfb6vkG2xHBS2nGr
zTbpZcDsL+*rH*!Lhd?I6qY`VDBRwaZ-?uAB;1=z-nWpW)mAp3~j&8g7V|Ym-^{syP
z_9ri(w%tYNciu03dnE0Y>XiSg9SfLgaX(xmL3#O3W#^Y><k!vFX+ep<O)T0XXIGbD
zq`l!Mbx(snP1>?81{`%phEGlEtuoiutE`-@{CYv7#1Yn6Pndr&&A-jAB_|M;N2LOT
z{zf}w`)!1)OB_TZ5dHnasUJt$p~`%l`*L=Y{Jr>u?J4ArhxZ)dP)!SDK2Y!^Ou}`o
zXZuCkhy1sAbtE88HZo2UBi#5oeJ<<eIxT5$TLPy;@TWu>^FX?*{g|R`zdu<>)Jw+e
zUUk1KJK)rPyUblXnT5W3Lph&~(XO=SYkYz{{aOe<$v?R_bI78?83E%iDgW3T$8s*y
zodR~Z&%FwoO?!2gDI$R0^8Nw29+)3ZH$E|qi2ucVC<C9tMpPbq)Cw4HAo&-m%k<hg
zNMc*mWjN~_QuUlfHX1G%TltjCSU=gDxjq$lvJ3h-@;w1D^r2?Hq`Ps_`nVB?0ZCH`
zGIqP1uY_6Fs#$jtPd%%TqZ<KjA6I^t1?i`Pc(_P4Pqk@0&;!M`fIjikZ4cyT8MjkV
zhQ<b4*|oTFBNX7DziL6!HJ}s`3XD;F3xzXbWlz&sm7h)cjeqL8u&iWP$!FGgaj<c3
zLbfzCT56=lN^+)Hx+K~nzG-XO(oK_#i}${R4#-f5h%j_q3Jj_?<uh3zv@KZF#GdY7
z?S-e2r2W|k{C%abuU_d@lyO%&QQUde)0ClSK`~YH=Wy%ncZbp7Y4V>E?h8Allf29G
zQhUssQt3Vy)s$N7Y0*N63IT(QqRKE)gaqlx5iuzPBD<=a{7x~hsD6GZtfWL`ls)RH
zrs;hQVLtMSNkfB>q#zwdWqE92kwJ_?%cGO5TCJ!XAb0VlpmS@w`{~dymYLHVs(i(t
zT6S(Losk6YWyvKIwqV1WCXd0!C78?Gw{LB1Y!+MnA%C;_Ys$F$zup&Y*S1PkVn<-l
zywH=fpR&3cN^AY*=@R>O$5C%kVQ}ULyG1(QbpQ=@JST1uvu@Z)y}2trpUt;$QKVt5
z;Ky*ga-dO0aY-b!(>h)4VNwtUR%UGq?(q<r28BFU*<CnIrYFKZ%-w~iHX^g1GM=aV
zS8kkyu2zHy9-4C7l$PDY;z>NNGZb}KC&j|RNp9a_FI|S-t{g!&q@$9)#I;QWnV`OB
zn9Q19@g`C#6xM829-~X9^G5qIJ(b4};k)0O`M~1#^$PFR%FHV^>C@^sU^~MEbYlU#
zlgWhq@X*eb4B^5;ac?Opw?D^e0oQ01EctAwllqL^!rGUROi|xI)!lSDq*6!KajmBp
z{kL1l8~OO}zL?B-m_0!!&3p>i1F(8q=?y_djSx@N577q?1s@TEOf+3x3Ka4~4HCn1
z!0svZh(ucA@^@soKQw>D%gzjkJ#;BXGvY7JosGi-FJC*ZNHq%iFmo*P0{#R`^b=#T
zjLw6q5*nBN6uGL@eD+ZTU-WY*jl`nzUH&e5b`A=DZ5{oFQT!Kjf!er!%UAM*@2x5)
zZ>@H(62-BbMrVe`wJk~ks~1hH@VBgl01-Mz?D}k#PDe#eT|GzKvwqC}8jPj){Y$z~
znx_-A^MC9$>+mIk5Ya@y{{9~MiMre!<Z;3GN?AgQpF3(c4hP)a1r8N~o}rrBgyf-D
zP<Uq&zVT;Pb;*-)Feo>L#-wNkbtEk!olLwXd84*=g#*rhM|Mk}U)=>e-~%Thp~IU_
z8y@%NzX_{aeN)EQsQIfsurDTc(nLlG$XqoUt<7#Q$nlPrZ4ku{to2$Y6;!yFOFv0s
zzsLbr+ggP4?8)Va#33*_M%UCrh)ieFdkcDORt8)geAaS!QF<M9F+2iPC;3fLN(3r3
zsn!&nkp+k`&4w!BtOaR+R#;5AneWl~%dCZqQ_tbngUN{`bcG+ft{-Hf{kcwUm`A9H
zQAKBsgo6Rm41J!VWR>J7gbIWxFF_6;9fIsIMB6)4;%GwKHP$9(2bcR!Rzfh9fS1j6
zp9=agX3wYXdqcy^H?)!>w|{Z)9~CG5P+n%F^qZkQbvnjFRse5IROV>k3NeDGIEAd2
z{jX5cOD>0TGZ`{$)=Zp5h5)u4L43$X!h@XwadBiv7KIIb<4xP5`WSO4ECLHX$nl(D
zNR+N`$u(DCI2<zXC`{VG?G`j`Eo56a3&H~AL^irsM{olT-n_xZg_o5OQdwGBIxe>E
zw&ZXRZ~XbA3+U;2N)R#(3=W=L`ZKB|u7SmiIy!2s`OD7ZU=ItoT74t(2T%{8El#Xo
zUQ70l->P($?MdJ0Q#?0}g~?g)kQ%7}=wvJwO%8>9x(p!=jK~w?{7Htb*Re6Q4&!8w
zVhT~=&pi=RPN&+iun37)P<g7aMuLxHn3A-L&Kffi1aKMqJMi?*O1?%Vgf1B&QiQbl
z!j|6bQ!-NLhqlY08h>2M5k|$hpzfAizq?A6dsM!vq{7|z)-+Cq^Synzy{|s50$(E1
z<0?HTAR*n@Ygyhh2ohsCs9VoYNS<y`Uf<F}Nln{4H#=K4@hZN(r^k7<E9Mu@N7;D9
z)&c%w5P#X~`xZG)7MP-dw}{51RX(AW_+F{|QAC9Ji)Japg~o;l@@#VQof$KucV*~b
z2dCkeNPC@3F?S!gQj=28!vprwKK9;@7i-J{Dn0B`!R{4h!AEj79N%>Da01OB+1cs4
z@BgGz20BwDYk?QSb+{dU=sN{C0e!N7fg(ed7SKmU{_uP`?8pRU6U;Pxy&r_okm~`c
z9o_Vr+u<Pisl1q&iONhNe`IoUGIYLiyZ!csJR}-ARbeFQ_qR$24?^--1oqH<?>M~@
zShWMtQ@66^r~@nnD!rZc`HH#!2ZP;(Hr|g(8B<QVY*>u16;XD@3FqUN=+?tNBybMs
zwI$cjvQ}tnBt0?Fc@mXqQ0@XMe?hCE<4b=CltIL*yl$$}2<Oi@mFwR8=Bu<P#uuGz
zKprZa(Lz=0)(}khO7JfFqW$wvzE7#=TU**%3O!=eXHub_e5)yD`Bs&NM)bwTkO}=4
z46WFAr?;)CeL^FcuXfO5lzQe@Rx+U6!+*n>jGP}Lo4Z7dbr;K}fZ0cnXvWaW0cvdB
z0S#3R#?ogb#ARW|Q7Z21Git35*xH%J7&q(>Wq`iF&jOE0v_(R{OmTc+daojJeU8Yi
z*UHW00Z!dZJWiNMp^Vh)?J`V<xOxW;IU|D?MJG8`D6a2vrVHx*<$Ags9gr>Rc5@LE
zQ$x03+*(uPeWV-5Amy_*y;HHEJWIhCt4$O=2^zPwN#Q1jD)5T`_3%fEcR{s&+-LW(
zuibLEfr0Pj4Eb<=hkkAu>jsV!TLkyUXjN@C@a3S_0VZ~m59R+pj&xjMKa!dreHUmz
z<F-Fw9xbGxqcdUq^se61B0!8;qdOs5wRVEDKQX;-5-y&(Uy#Df!=ubmH)*iV-G3;7
zLD`k(vx|Y*Ys8DRP*^=DW+%({a##4Mx!GrZ)?e|l*Tv?BxZqd{Eu&rng}%8$KFUi%
zhFFLHnM!=+r%cmIX2K5k<lnQtIm`$1$J$JSj3RdZx>l?e>R)=l0Yp7_LN%+Hmb2ur
zR1NIF4mLq@*&ABa40R-bB@%g?dEV&g#7b~#b8IEctIB!H3@`Q1Q+hk=RI$vs3O#~`
zE|ufvErmj47zw3bgOHRIY;0`3dI*G~55M33j@N#w(WcScj}oEnJv8=2PqYnNcf5t~
zTdPrIdmWN;tP8431B$qsG$eQ}j!t$iE+l-E>mvPR^m2b*Y)Jd<jC8MXM<#ur+K&17
z<`F^EnC0>O5gikOGSi)oi0CD?0>_Fn>v9bD0LPu#9@)yAuHLJiOh~j^GQ66nROn?q
z!!u2?bYsujnX(Dnx|xOr<?^?xLZTz&$=-+1DJ>2}pgtEu2f)q^ggeu4msdC*hp+!+
z%l?$HutB5r5DP-8hN70E4~`6oewbceLKA4(rCVB84I-!OFWPSR(7WQ!OT}@1<6;{N
zZPZ_PoIK~UV%yxzn3YHEpCRk(<)8p!GgMS{uOlYrL?pJ>hny|em!U^o0dU*xa=Mw6
zzWp)mf6)E~Fq?dvON(KISeL6&?9Mb<VfMmaq&*>Mui;Ktf?)GwJLkgXob0aq!Mx)}
z!0E<p(`huJVrT@Xn%#-d(f)FN+7lzNmxz`M$NJiNhw7w$M|5XYJEVNUK=gdi*W_y5
z<muq1KR$8geB)D^*jk}1N6*3{G3?w_3w_-OqXjmAzfS;=S)a!qBs1alIz~|^W#iX=
z3|;F*U%!bzW>C+mqHG~UffpvX5243}SN~3+H?=HmXE-BR8Ib68U1pw)Bpr$kL0ag4
zQp%5iThvcZp!gR-7MM6AlH<`N@>ro>skcaZJTx$Inttk)YEn&73JH(G<?Q(zf1$<0
z+IvXAvx`czhR#sDCSa$S@f2Hgr*knL#vPA<8E~3I6!1Jy*&+w+A)u0T-^%h!PLrRW
zeeCaLb8;MZf*CH4?v{$Ae)a0E?}Iox(bKhos}c5vr;o!CT*A%I3AOT~7*`$AA8M-W
z@T8Rc6ckuF6G=Seu_hZjVq)}Z{W<tj6^9(S@Ucp5`^ZGL?w!p{ko!rM5(AcF*tBr-
zkcPr=`${<RLyV#0w(f;R;>w?+KQi(K!mwT|Rww(2(AefXm7}6oUn~6@==?HcPaOWU
zqy{d_iC`MDM;<nR<;>VXOdf7cl%nt%I*(N^ug;I)D<Go}iPZjd<<)HEp3xg?q#bZ#
zK+#PYR3KA+$2)^L!4p9NT~qui3(XG3P(Wm%n(B<Qlox$E4T%Q5?4@-Cb;-=RD;kMh
z@xh5yHuo5CO&Uwe`Z#%A9|Q}Dl}?0e&DkNrb}}V|!$B#;Uk?*Q`4Rb}jm>3!&A$!2
z?MT0K6Y{LBy|s5h;QFE=koZ0UHsDm#{F~O0PKV*jQZQ0SKl;O}Y-Qi&7tVG=jxU_u
z*Qop6-Aj>OQfud5*_fR?r|J?eNp>nHwn-$<_N|4)JWmJK`ybHxUy*X`_aQ!Z(}`Gq
zds{eW^-}We_w|r&$7O#SXi(qhg3YC>VzIkOnKC7PpYfKO8j;x)IG^q`gaF|r1RTMh
z@Cr>Q;>+ujKF`6y!0|viTb302Z=Zt*5M*-6DP|v*O@$^t%jvtlv`ta9d}>9@s0F2H
z7MO(>KO=-dM)W<fEH_RmR<EcR&x9{ec_nUFta#+HEZ)*W#|;Q^={OY*wHr*!d2z_G
zifmSlEIw9~GTvWlTJ%1@`UM;fOI{B2#aLKA=)uDM&<=yLn$wet5`rtCVavnBLJ<s)
zC-3tF&F#-+XAzp06{iGSu1F(q{8i0{*7_D-jFdSn5SI|~c05Rfan5!BvhB8Q7gf7I
zy{ewZED@m$m@#1Xq^s;Kr;0I)Ic1vPv~<?Xj*@oBv1?MoU?uN>E8wbh(SP~XF!0AJ
zU=}AZxwgp@z1Vg;pFZ|;B;r$8GCPY(I&(2L2Q*omy&yKP&xgTfy(^{iI-~NjsF%5!
zgS(rVyNk<BSdN#mB1hkAB_?gHN_7?+wQl9NrRg)jiql&r!(=9Yv@X<h*7n$!SK5YT
zgp^C4<x4jT$DqW3?g3DD$bVK&Cclz}Rbx8B+oR<;@XXhIcq*o%k*hV2@Il|Xg+xX!
z5H5p7C=0`Z7OO`6V9&BmCyYiE<0w=ordJ46Ox4CnW-a41{S5MXYD8XLpmgS;_IUaM
zt#k#3u2VOh6eq#VqZ1cf11iH&pd}_3gv%YRoj?I&>;OQw;I|EP7D!8MwxV%i>nrqz
z?!6O6Sl4Nhc!F<&_|3$&PLOTPpBo(fYz@nLX7X<boju&~hnglIe^+#8Sw7+@PLaXI
z`$Lz`<MHiCeD7_%7MWJA@~5Mmxsc`9DP#$^icR$Si?p3vOrjVb;^U}5l#6Cf`LwS5
z+Y|x8EDnG<FSn%6$?1gGsJYR@*RE9DT1}l+G~nB@&iX@*FtE8`dG`4B$#SlSflpQJ
zZnpnrs#6(W5dyr&C^&sD4<jwWRlzD;<b8mlyIFYvcw>T?M_08k2A~dsGnDxbVUw2<
zUL^Dh5(3m_Fn-vEx0BG$ET`ue=5GjU(%{kbQ6ODL+rHO_m|jzCgIfnBYFbk-{FtX<
zM5T{%BV@$c5Z-5AIB@AbLq~kbt+x7!`9p{4Qt;qW9?gTFs2=F9#OoJ9zwh+)laU_z
zz0vW8>gdE^1!Ei~9{CvBGShG4ykI3lU>~ABf5VdtEpc9zWml}Hnm({ZHHFm|C3C&G
zLkRRHXmSL<q`p^bK5jprF<6xCqe5`7>*T);!i!N)#nt7%yPmEj;K%9*=9Ty?#LR4_
zzVpam8ONuSI2>ov%402nSUoWgY^5!CY`d7nxMymt<6WcH*GowxU4QgICE3gClvfI#
zQ5|QwxhmeF=d=s|`IbT+uI>GZNKu7@5daQ;oSR`5ZwI=i8NMCRYuLz<Vyi*IVeEWr
ztYdv*LINsdFY_B*uWNWND%)-r`t^7NcFs4{43bz~QRE4jNqqIS6fMYFruz_oEx*t7
zb}fmGd2-;=Lb)NA^i>B&J;%@2&h8aGES$E-%}TAmGbpIOUifqQDYF%>EPY>7EYR=K
zqz3nN`PHNskevpi%uZ*wL69RKIV1Y-CB#DW92SW4B24a3xb3{I!}M&k?ZkW7bcF(T
zN-IJoKhjWZ#pdaj$SenC(582F#DAZn<znjbgpN4i2sqC<-J3gdYuo=A9y>2AF6{Yd
zAic>Zmi)B|G-yCYg`uD>HlnW&VA0PC8s+|+mF_BE_ucBTDAC0NH5V9kC`^|xt$k5K
zg%D6?q?|NVBbV2oWYwrWIs$`|cEb|aE~e3|%!9VKGt%d5(w}<{(K_-{D3@y`TOo3U
zM->#sAdy>UzdsPDCD)WbSu66A5hkbDB}R>nuZY5{flJ|T*lRY{ydwr8xI_6kD=b_l
ziCx@YXFQTGUtSq04p*P|TnP$k0vn3td@;SRFqOxy{7(1X{MUo!=S`3OoWD%av(P@*
zn1v#epUpDQ!d+57!;79WxmydLf%jy|>jZMUZLXHRNx1tj_DC7dPg~5sh1=a2Br!VG
zx6p{`kqvwAfpZVn_OCi%R}Se}%uX;+&)GCbt-~KB#0ez~k)u$RVJ5}pbNxg^>9X--
zuEUufCb|MqVuz51$`FV@<J#cXj9C2?fu1nt*Q9V4dlk|l!|_zsXhnpDk;9@~H%aT8
z_+Z`Kb{F#{hE9OBuV!aCoDS9G=|myF;HVd&d?*DA13L!yr||+J8!aTARaJbz`6xn!
zfA%Z;=-8{VN!>oBVjZnGUEL=jwlF(mf%3#D$FMEn;$j`6m-b}Nr2^5X$g}}AMwsOf
zCRq!`m3ar9Oj$EZk7L{0u3<0Zk|BjtsnS31fMA80)|oSbh3T~m@4iW)-aKAtl`v>w
zk5Y3s>t#ui)CBImdm`Xbd`xA}|A+(8f>A9%MCbRUr0CjWi$H@IaLuCY?qj<nn^3;E
z8rTUBBa+*fXhLBeyDf?oIw{=-Dhgqm6jsWUFag)0wDX?z=bT*=xpwV130jQG-((V1
zl!AtIYmeMO^EK6H?_|%*%PVYn&tYu>!R2=3Rz3N%1E7yG!-G=DLAkiNd@*}Luul@<
z&dWO>_6cFqd>e!hX2D2aY-s!vc5+}Z-|!sHP-4H+ky0@Fc2i<?vdgk59b`_)`)#17
z^OS~oOz2KmZ?77ezr*?Awo~(A#|$#(y0f<S?p<PSQ=*{$nnXF#jk;J!*ximKD`MaK
z4^k_$fd>_d<|8S>mBv>nE?&v=@v(m?3WJYCkM0PPVdagBjf%;2zAOR>1a>p;Qou&S
zVG*IQ{Eq#@uoq5k*F&Rqh^DD>NIaj-h4c30*43UQ`R%Gn<S3}JH#axe*VpOPB3y@B
zPG{|SThCTH<JmiN+74nvjLdEBD|X#K4=)1LO$-|kxlAs!DA+L4XMC@|`yDWnif*im
zq{`Wr$M6sAG%vm_fnZm=4T<A-wq1(1)q?FTlfjy6BECarKGTg~2GIPcq~L81?XRN~
z@kU3U%nG6cb-z5dHBS_I33KuHFCY0jLGAt^cOqMV9*79i2<CS&C?!CEO6dHq9uTnU
zbh|KNxbQPPwr(z~n+^aGo2~a1F@<&_+qij&>|w$QgeXFIUKaO5G!#)EltNags#cd|
zU>QBBoXC(@UIt0kvM%lf{j=n81DnC10+<2;slt+Q{3IpOA5Y<&tH0!NQs<qmVb>gT
z%ziDTipt63g0yU`QMCoSF-PkFqV$&+jKgo~>`R!@$;($rf$Gw%A&(l}LapbKhl3O&
zn5{^0XXlGoIp>|n+JXYC!)|t^OCiXLf&THo&o4cIwS!9xuPW{An9tU7gVp2Dvtwnd
ze+k|<kZ@7w$F!<Ha=`9j4-7$uG3$MF7yOt}$IhQ*{5D+@g5}EriI4mNFdq#1IytPj
zWcXRI`Kj$#t07<!8Z!dQ?=D7o3~sJDV^@_|4&Wg{^sH+WaAn<6E6sw7>(=rg)~}N-
zy!V$nd|E2%P_<UWhlj4l(a-lI%91~4Wi+(Zb0SWq_}ua}7pLFc?cibw4SXI>@}||#
zCUb)`5ko6F{dIGb)e$8HSnN3LJ}cPfsya6{@8^6$#>D@WbNWX^HD<Cw#qQ~&d;K-e
z^(B3b&uX#}D#5VXPmS?-p7(=gk3K~?j+hK0ojja)7v4yDZqGiPGkiGT{N3gH%?McK
z6>pQBQDQOs;QFaD3uV5hXRUre^UC~AG*ZX8VxT)hVh4j27*Xbi0xoN~TWWvG(xo%Y
zk;%&xgheF3LsJIg?zfJ*49F3h@0;DJ7%5|`)6J}VpUGa#Ch{Hr7~?DOnuVqI-jlQ_
zvvNNjj&hl(KiXE)be^BHgkBWL;q`ShA<Ug`z^ey|lq+_fmK-sjKSd^@SXYkEbbqSS
z*!OR5cfdt$&b`&c7Ql0{JX6|zYXEHG?OsMz{H=k)dFN*J$oD>0KCEgA&9jq^L8%`b
zCCPh3nM0Mv%r|FKgaijBPA4z87SB(|2ac2I&cBXpLOVo<i6+Gu%xSD<sp#aAJ3}-V
z@0*3u+@<{jCtg}Www{ymN!<26kVvp<&KIE2Qem`xM!k9<0)Xd<=MEu*>Pl-1Mk`|P
z&t#GG3CG6#dVgBvkIpwgR2j}rPEKYTMDa3Z*nx(TMamDf5?;<WHXOr8dF&V0Hz2R)
zKXA4(M;J^w@#=3?WsHl|8DwGCQR)#v;8H)y%hsb9r`GrTf2(ht-ye0k7t)8y;M(n)
zdpksHs&H@sEg)R=LG#`0^b{?HV&<q6q3+m);ZArUJ3B6`9E1>;Xh_sd^>d*!{*T@F
z2KggYvZ%&7v0A%}tqFuj>6o?9r-yf@9;HXtovcaleI0E$`2KD@P~=fuj`u0~;N*B;
z-yi}}%EcutE#b-XB13CgQl-*7J^~^YYmuLD>YITLB!rn}YfOy#DztM)ZWOBV$TQn<
zaiz#PCyXR8kUZAdv9Bn&Cki?y`PtqjjDx%BYx@Zyim{KUYn^kSwtPfHQ<Mdqbrh4B
zsDF=I73vjG*r(uwdid@9u2z(4^mM$%DIePNnWc!P`4mwwuEhb7Wmx+~^*c_%2>tQ9
z@%H;dfKYEx72Sy5UyiAB)UUq(=^^d5Qr@F=GZS6?P<Nf(jFF}75Y*xC{D=(r!KBAq
z0WF)KC&=_>ZCq`ew}l2aTNERFRd*p$xVTsmaM@fkEr>Hcn<-oW<2`#R%Ay2})L9cF
zt@DtYdL6RKKhKx^6^)#RiiLiCYpayBbk1Bez}>A$K`BovpWudwjGmnJE#VZGl#~<{
z6tuwyfAmkGELb57DanL|MMZcave{{zlLlO05L!Qc=~SGe#axf~F|C&&&lO2NxXdih
zds*K)()z&a>)zLY&79&M-Ar%s)_la0#e;=IDEufkkUuJ~hC9p@{?4`iafgT@tC$ub
z$MJ;`EAj3l`_u{8QTgDYyVZ+8`a7&A3i%aLyV<nF2_XDIi6F+{PsvDC)8@ywqmC~p
zKC`~$Ld){umRZ^GKb{k?I#kJ%<3VsAW{N`TSxMTJDU{(HcW`WY*df@VODTy4D)?X3
z;1<R)jruLJsxIZMq(=PFjIz%>e!W%!SY#=@Nv8-zadVnn+20$yFxwD?`R!RrJAFLx
zWm^>BxzkQfy{(8AFH436S`BAk4{S(<$T>qF4ChFD?6fk+a7@d*H?wKLCajk7TO9Ep
ztF81~MUX3^n|yY`GMe3~<fsUm+WnxkC%|+WQtEKL)~W5ta?k6eZ?wXwDZ(1ai#b3E
zXE-GPI4_x+i)AFg`|7(EhaWl%Y2~r&^Cvwn^L%N?S%MW?J()RuT>Cim^#6bfx>6<$
z7eCvY7`57JFywYCC(@JS<vMTvJlZZrN}vBIYP+2!DLK(WM9ZjQZJ&IAxds=IYg%wI
zg!1(X_QHZO9lm%X@BacC`l(2#MO2+#8cdEJ)3~s?X~S?Y=nUDz>uH;3DDJht^y}k;
z(vn91%Tx86HT9d6RcQ!4HYpa=3db1+PoSXqx?uZc#|WfPMwFD1Ej_H8WWd!bs3XQz
z;a~`5J4E4|LNrvES>-xQ*werCl#0B3npc|wkIsvTxMP|t;|btCAea4|tYx!~*E(h$
zr+9vYQzz-xB<*#Z#XQmVw}$%}vI$g%{c_{mS<JKpLN%sebe&QyPL`gtJt_xNX{-AQ
zEf09J?kqWao(%^ancQ^WZvJ<(7VVZ9c%uAX@<joKigF~VAE6GV@e{Ud3^A%jjsm-i
z*9kms+QlS*PKkTT!gK?(N@%;#GyTt#qhBYK#LB&FIephU`BQ&egT`8rx>t^%+WqLT
zA&J*+k9sr24JA&olXaH?zj8JvtRd{W_9Oo6*j?=C<i;9G9A7$C!AGJIYP1R4uXxaJ
zC<!0@2m|e7rs&hZ4t53XA2Z55GFeeS5nRY2ijZ``(xLv<nD*YHFm?Ak+rDGnrpT`o
z(bb6Bbb=?I^l<18O=lMR;rBJNL;3uFbl9_Wu-WNx2$S21=<ld7LKyZqaXXm<)uKuc
z#hWa#7)>?TB65>I+l!r{O}eNkBm@O(-=7FTD7+v4>Y~aAJ=q#=!^L@fp+|-p&&%^c
zhyV+qOeU{M@hd_g6eg$iDg=u1HB&85CBDcvni(*UsrmkW<08fmbQMebCTey-JW54b
z>*<NK{$EwPS;5L{_UT=hbq5&@B~>{B7Cxj$re{ot#4sr{p&+TRYNxjh1$9yv>nLZe
zTBfOvqe`)21~oz4;huvqRmx<2VWH+0yD}hQSQ8+zqbO!0WK}as0i~i`)u4>doOl;&
zr{D5eJ^_tkaxf_WQ=TDwWNYfMZqITjw=4w0VE(zVqR;;=8(i#;s}2hoT^6MDK}S+z
zE0;&~B;cb<nuo<+j*f>18u(DlJW6@Tqbz)h14~Q~%5*cUs`Jruv*5_b4ShY2L<7^+
zU<L5$i}PIP$!sbAEwIioS*z+Ebw)<krcJUO%zyvp5(G^Qt67%4F1<^aEqyS)DwU{o
zHUkQS)775S?ExJS_I<YHJAJ4*n<_Je0qtO=c{L=CmN*$H;^j4ZMD#(&X71=8B(e7m
z$blF4xCy;AO3893;Xy8u?Y=PpDX%PV6{X7ZI;+krNlmRx7m-WWCkq>3={oi`LT6F1
z%RusUeXuieeCYL3fAv>3U!1*CKfme`M>m`K_lycDuQ(z<Vk45V!(e$|``XQupCzQ*
zxAQ%|H)g=2HYyU^wBVSpTWUXIEmV-~$#MXwsj=EYroq733*MIFFkZm#ZSb&awga%t
z7L+7(FNnR$O*Z#IkBYMLw;ETsgZ5>QCdqk&yA0OFxS>KZ$?+j58AaBpfi+<-LWH|Y
zuSjh2I72A+5N=@nSut2;<y+~eNNB_Sf}>6o{stsT(hXuCK2oMI)6vn2h<tE;ldsr@
znJv|yKkHG<e%<K7j|*t2V0$6H3r;u+8AgmOKgTdd#0;K*g<y(wzJ39itHaThauCpT
z7I%{m7V`nR_iw@;NF5orUGt)i@=3C%Im<!Rfj*1!mNLd?4!oNsKkMYQ5iXj$SNuDy
z<XcJ#4MSA|d!_?a|KpdjUR!;MrxVsw`DTjucfFdB?yHmOK#F7KgW?`a*{*|uY{{Vc
zix@7U`L7&u6s<KA-_604$-r0o``TjSe=)-E!Jghk+#1kgL}CQJJ}x<z>F-%3j*pBR
z)uo3wlNaS<=r-T&$-&VEOCB<Q;-iYvOc|qOvy98vicK4{nL>``3n;4TQQhk~J*$mu
zpQ{7K`pL>fAk_4(NleglooF4R7$0tC_e+i1Q4(5GYeXidGgM2^B{wh^Tdn?a=FR|%
zjI}v*sM|#@Nez#*1=;yF7!Q^7%JRsiMUv+|PX*HV+PHKkiZTCNWTKK{VM)}Fh4{Q|
zLtn4OAfKW_CLU%H6GU<PXGd<6CfCJ;pPDq}@ki@>K+BPE1?p3F`cs$acCf4z6oJtW
zoN)8tA^lLk$FcV<tzw^e5br$0>yyR$OS%xzpf_8SVfpnNvSqg}MMy1FStgLohmXe6
zrcc#ElTU@Vmz#VrI%V-u<(s%#_GEoVf;ZBOt)eK@-T;qE(2(zY3&x3QT^9qkJfTvi
zcvenDfXQTyjJcROCM-U4@)b2UBsPjPNnTIc@Ow#3;<i48y;97}Zw%Xb6S*VSP%QY6
z?h`!IXSp%!Fspl(p-aJRW~!u_8jN)IY~@D52-3pE?XVhEYJND!FM<c6vc#=&Xy(M1
zGvV{P#C}1V4%%B`=M4kqtA3Xmz-iG2@0*(w;C3HPplvueXTw8jzz2z+pPtWNK$0X+
z3Jmq%G~C?)meIHP)!61P$JL4(H|D1yBtudcBW$m+nFu2jDetoIHM*?M&R-4qvH@Oy
z_Ycx1kA2bEJ6=OuCozL{eQ8HWNBJEZce6IUX}W(3#H~_kDB+ryM6G}|lG9DVZIU_=
zu=eBEM+%vTl=25XsSD+foJ^-XqF&G-WjnPVdJ!4tbx4fp`2~_hdcQ~A@1m&fVDv?k
zcvgogmsgRw&`7^J5w1$07H6rIMfRnJu}P!rRR=jeGq$R`7+tnxtra?N@{#71nkoyq
zngGP!v0`|57^j`t0`^H#6arI%B+&v<@S-tRSHcnYo-H*5B2ujbCSOA3j+PEEDb<(u
zyqLG1P8&@_B*_WVS3i9e$^PQl%id5k(Olo49~wxyCmyhK_{;65X6tSHNY43>;hO`!
z+hhxQ?2A>b;IEmOw~D=^)<TEP?}I8T%k~zUFbCaq-skISclQ4{A591wU;YKi)sGt`
zIVQY>`ogbI@_es4!tbBnP;3cJZuVkOl}{6e7yasHlqQ{=ZdZXTlmCyVvx;l-{onXN
zVkpW$q(oqV5+YI}U6KPyX%N1oFlp&{=+Myvq+=jCx<f)lL>MJQiBZzhCH;SX2md3^
zIN0mH@8=WO^}a5tz(3cy*OS-vyn#<P-%TC)7h=fSZfd-~&-Vj_fr-5QufQjtEg`!}
zH>zyIKE^;*At8xZz!-_F$;hmf_IbPauRHxxZ@-^!+hrxO!|M_lYyzWQXPuwkc7C1o
zW%_hON&NNK?)|A#*0-+%l3OQDskuoEjScm`=H;HKc)k5i%EsE$|2tn$A|?sc1Z<0B
z#<uDTB@BOH4q4)kD^NvHb}P7Iu8y%O%iAlDqId<xW;KzeY=(}O-{q^vO*ZQj8&~{_
ze2%iI<|I6QPghm|v8N%I_g&i0ur7ghv*{leCGrd;k`#=P<yKh?zbi6Lh-=&%6nS}W
z2=f-9qB#;3OB5lzA<}zs%b*)4`2_F(m}HYeg8iNlTLJ$4PGWoQ^iF;+x9*k!sQ5Ka
zeV8}W2VoNy&DbhkH=lK;f;5Om>+(D-dWE1FU0ft={+I;9>=nPWjhQ$NfBw3BfAS@5
zGF7x#oBw}Kf#y)-6iq~f%bZM6^?2FkpMznCuNb#EI^T4vw}QOimafh=@B{lLnek-#
zIhyu5ZR1(#j#~E@>WqR}?V0Y^%*?9Y!8O@A8adC+GzZDwDi(;=)+XG?&hh$B%W2@n
zSI7CV+^=u|1+Cg!nkQCD6U~MOOJaEL=Fxt~BeWS~nU3JvR}*|~@0)SQCQqCFGLn$N
zzZK;JW&J<4y+G|Bd)$%{)R^v}aEktVcI;xwgd@@#jbHZZuaslETg;1lmlY}3PeL0U
z6C5#SY(Nzva%;VX!Mo!Jky%bWsoEXx5wX>Bx36r#!I+<VvCKb94-d?QgL?kc*R6;1
zH>jZSc8?#El@Ckh>vbXGF4&!86}{{@62%xDCfW0v)%Fumr<o@hN$n%_>@oT#I?1r=
zwUY=F7`oin;(oARoO`#V;<oVk0j*-jr}FZ0Jv}{diopWz?K6eIU7S9+G7i|pKG)0=
z)2RX`&A7jTE5AuS-i(og>ZKd*g!7od2}yT;lrl%iBB&F|75DISbE|pIjOru3LE~yk
z9FiQ5Sl{z$2V1xrq*+VA?<h5PsNZT?nNZ*EwLipT@psVW(IgjL!uPrLj*}24ww?B;
z--Kh|8A>`eCAXYv)}?^cveI{TInpPCD!_2g)%>zcphwy4|NN(hF7Gmi)+xlPi~ZPB
zGpq0D&{AvXtcG=W`2!c9tyhNwe{Fdk?ESkq^{5KkNy$CE(88#FOR2jjNcvSqR2Ot?
z;99ufqeaL;!w9ICcwE)Kn5CYi7?~H{dEGI2heg6jqHU*39!{(M53df?`kn6s<E-LA
zhyDLyCq6t4I_u@oBD~(b`a66*2JC5FXU0okfA=eIK~7%eUalL)Kgeu(9{c*Hfv!NZ
z_wx17RHn4Cm$`iqa%)e+d%(ABU=Ts;m;(ugk;6#j<zA|BLRC4HzB8FPr+bC7&UmTF
zZq6>;iQ0mC*PA%1H<~(nl|M8A@}fgP+YmQQj_6r|wukMI_}sdN|M3^nBCa3rVaKwr
z&b>0;*&UN-pLf6+E4*S<1L{Ug6dIPt?WXWi>`o{tC0Q6bF+_x>6kR`S7aQrK%nO$`
zAiz%@-->=DsA=oyQR}s}%2jQaZEnB~UX+zb!^w8CK5+(n8>HbJ?b`u;yf|i~miv8K
zN&6ifnfc}xcI&9&XxnSpaaCiN&e3+DjQP;ZtC8j6Uu>F%-|1J$CIj!6vDv>m_ghq`
zn5R3d#YWAt)++c0MZg8^k){78b{vE(25v1b9}T^B;)BqvD(r4!ez0CnSOi|54YhU1
zvvBXpy>fbG+a(d2KMO?LJz`4EPMwcAQjiSSk&+ebmdd-iCgnrlPJSpy>_ZhtqcZBI
zDYlBz-D$^13frNs6dT_bKiAf`_53QHk0t=spu}s4RHYZvZ8E)cJ9c!)uCbxvp^^}d
zcia^x7>V^oiVLmcHq}2B{{4GtdeG?Pr#MB$1VR1=oi9@4Y=FMb&hNT^Q{_*GUTS?3
z6!t6m5hU}g@`JPBns-y1lJ`i5cRABLuD<E><L&i&j`B0l&k>%vW$<UdlA$%>ns%SR
zZa7}rwFG$-d?#^3lbId`j&4=WHJn186sc0|Aq?ddpdrEo)UQ&!pKnj|QduUD!Fg)s
zj7B`SQ9@;wYhA!k!hQMxq}5w@uSg+B@V6(!pOxwA7dg+hW`2ro#p8U7>b>PBJZXQx
zr1FYsU$`2i85kPqD=5;nXh@nVVyGIUi@f4uq7V9C;WmSLwVB$>D|p0S=$ld(MTdvH
zn$pl1cA?sb`Iu*(Mi_$@>D{aJad3Qe52eEis<OnqfD0&`pFxQ@z-M8<XHkeFoQ5VK
zLC9}yp#wFR?9!p1`Vb{`mQ~Os!vNY92WkhIMsYZN6zgjvlc~B|dzVzv>8+ZhBTt@X
zP|}+5Nf)s?)Bi^*<F&+UMP0OZ>Q|(!-!rv8E61%u8MhWfPvlgp|0HI<LT+8YIb^68
zh4h*LT;k(pSvfg(<mwOjgOB(jD3T*0*_DCIi<Yaybz#=WG(_Cibsip|3>jm1cs>2^
z=5Eu<K1}&$5xqsg-sx(tY{0)=D&Vls%UgiY(W(7Nj89+Rq&)kX-P{YRB3;)I0*`eH
zxeB+W`@h}OI{WO=tWj@5(D}y1<6)`u4^urr-iVTZs^{_phOLS{{Zf?Il&|e=_@Bh3
zZ^Jn<v(p3$QEkq(%ay*V`?6OHX`TD``0p0>h+EQTH<?Tl;*-9A=irKyt#3cXXSy7E
zH+t$jS4#I6x|Wf-g4~2_Z)C;8rRE8i=LS?~rrs{*2zFU~s=?&yaCRm`uJ7OR+_@=W
z`|`^Bv&TwLOD%8q1h6-I{Ww+@K5-k~&y9SRx#cD7^G8E8ljnscDgp*o;QCMaiTErm
ziQ9!2dlL*Jd(X{sw5l7E#K?qj@t~n3m7~%$-KoGAozQjNFGrY@F(tt7GD8KriQz2y
zuNuY%NReT;;6D-iUq{v@hTFbYKW6|AKa`<$I3v2|)pl{Euy#+pL_>qGi^ta0M#B%K
zldJ8BQVx|nQhtjg_OhPy9=RssUN%>Cxw6h(R6h{+B?qJMerz%5Y_UzYRc`JS8pgJ^
zqxrERr0HpOeK&lPrT(Z)RDPNd*C&~*pr}l~^U<T{Ec$~1J_D^2whx6xz<?~lZXS*<
z(jr=EiA^Mr_)<TKP@JbYisSp~oNT~Z6#xC675v%O1oG6f_!(Z5fUR;7+ia&&o}O;n
zk@2Y(7Zh~6=oPd^dGx-y0ddH1$1RCD+YAlGZkhZp5bg7`sSy6XP34C(u?DGJW+pR)
zshdREVcA>Rnv%$8G?Dzj0F!$!%x`D&b1w@y_NgTc3q4QHn=|Y5uWa-3-Q50qb=Hs<
zIrQ-pu^9IqCUc!5{5jH_i8?J5BB3mG%OvHpwL)5mrASU~H#OLrs9RE>4#esp81F*F
zo~E!+M!kZrh8^(_2@Nz;g$QaglFRXv+|5!+x+PD+NGggNYcAm42l(Z>oSWiR(PJD%
zErJLhQrJMPB5&A%f&Q9Pk_hw2ZVhyTS~J;hoZO0;%tj|ehsA#rDyEzztk3L*0KG}#
z%c#5ESErp9X)vF<Q7Ny#Z)nvNcCW(7^^%ig64Yod+Jd(FM>$BkCQX*Vrh62#Tpn^{
zX}e?}I2+w_$&qpZVmsW|b@ge3<;1Mf>Fsuus0m$Vya^)FO<MvunN2oM73_n{nNMO=
zQ>BLQ$M&g?7c{VpCaJu*X^r-vV>9b?3-t0zL(@Qs;Zu0gU%MU|p(mfq%E~HK&X`H8
zupc-o9iItHH&j=5*^1PD01u}JxO!GG_#xm|V&LfEVGX?N-<5&&vJV#*7rV@T#x~r6
zU9sGrn!D$>dbMA5<$kg~iykv5Kk7}^3ObR!EFRk0l7qq&C{BoYm(C()nMwv~YDH{q
z-?uXzMxL_}{Es%48;;2J%_ltd4ZgNe4F|X$|MVD^&9!~-s21`zTzq26hZUOdmQ+8r
z^)vtknE;*FpUWI0!{G~PcAuL;9+y`q78m7Jz~yG0M_b}RS2I;ti()9MMhiHr(JVfg
zLRO_J$<SbZT|(sWsub)*Vjdf`O6<0O%s=LQ<RsdjYdx~RzKS<I9+Wx%tEUyPck6oZ
z>D8j_#qZ9vEEd_IlJWPZJb{1u^Baz^T5F10)IgUWqS?~6l(EQxotQ+K6f<M`jVJr<
zbCMd7wX(HUB~9447+uz1*>}6?78heVr!^lpj7XHtWSC8?RfHhejrrOB=L6#2eW^+d
zSs1h!R6|)vEqdNeRe`$u1q2yro>^A*!zzJZnkq*K$3dYAg`<R!DWd}n=7##NWae9&
ztO2d}p6)PB*GV<COLZcht-Pdia#c_GhRs{tj`}t_(pax=&~Zj7X_n0E*oF9X!OUdV
zT0s-|H_4%xm`0c?1<Zy)5&ce7^eoNz`UX3o*oYEntZQiLsG<9&gPDM`Fj#}lZFtRs
zYqQ^`0rSIKx7t7mKJ&Lt<bCvLZaFH%<8yO!z27PeoH$W$AhFi`$A|Lr=PIwCE?w^D
zd{sB(r-}rpIG`E#`-yBjZC0NpC%GmYLvc#g6E(@Z{w##@32myTp=wjlrhM+c<W~T~
z@nuAJNh{ktuepnhi*R3i)PPu?5oN2s;MxgU5$TJ&siHO+(b44A@3#oJvoQIpTWGC8
z{zkjh#pC(}HT)WhrLLXB)Pl~VWIlu`T{;odH;N*(vJmoFvT`I*Z9R{YR1aJWrFC?E
z1DWle*YGf$Sk9OZF^T*ApA0N%uMohZo_RXxzYHU_$x@lm9(|^ZN<!;|gcohq;D1s&
zi7u8<u+;OCP~j=F4RwWZf~kWc85XY@STo7kzyJWEi(`YPQD<s}vi6F|Q?DvXQNf5D
zih*n$@D@yY@4|`ud6vU<5mj^Tdxv^jzDrjt<g#*#o7uV>+egzw2Fg6Q0iumo!i<@4
zqho_mQMs_a>~1%4&*9;rssU@ya3%%#5qV!Ar)Vu=9WMBY^F=tP+Hu9-%#RsP4JPP(
zD!cF%5(r2-c#A#^R}H-uD;#zh;$HV^(}=b>UwMw<R7Rlc@RIcHNBrUOFGj{hj=J>_
zD&`zJvta*NgPi17oaq4kPgK%mT7xdn=B#P2O|4H$^@<_8i7G6Nw~qxAEUxh`?N^xR
znjI(IdAU@Pr<MYP>UwXCBwQ~d<}NPnb5E!2gBHF=ov>W=(0_J49aAZ*u!7f#{iiCH
zpS9FMUjDS>V(j_nqs*Z`&OB&{PZWu`eIi=TzvAFl?j>lpYZnF>C@TaF?j!k#=4mNN
z5%TvW!B46zuo%Nd#Vk{K;~^i@P=PGStGXKs)<T+Xzmv>R0q5PmTQ6VV!%rDOs|H(6
zv_KSG;OV-3?9I&0S=oz8;N9g^%h#gyVcET3WBXa6XEiEX7t_FHr}Sy>Q&&3)#>)X=
zi;MMP!llj44#0yY<{-batd{p>QZHT~CbJ?z6v0iMtvU97-SzS_(_x^_&)Fvv9k~g$
z$$k@^<8vk{iYm-ZYJ8-^t&L@iXrI*zH+_BI9fl&>QPbj$_c}e|O*o>WaG_9iYyb<I
z{ieN|d1-f^eFMB7Lc!jxBPs(AF2PRVO_w&@*WL!X0z2EoGC>z>iGg)><JMT^HtL7W
zpNj3qDm)I3O(62x#M<&aFP^9NBu0H=kreTNYwP!D>4`i|2+FcL$fK^UW^|kmOTL+P
zH1$h@M0j*PU<=uKQU3H|z=sRdeKU0PWJgcVh4MBVPyW1BB*ixanuoeVJy3X`qxGe#
zRt!md0|5WUeP(l0+IB`P_Zbr;#A9>=^)>dLwST$n+FU2l$$NBoxSf8g#KKi~tIfdB
zg^exJ64i*4*s-^>W0f+(XbA8>V1xatvp>aenD<6Q>`?=pErBPRNdq-8E2kgqS^N(b
z+eXpo&~Ka$qSG%Qk+?H9f6U3oMVWpc;CL^Y31s$u`O}NGBo15tiJ4@0YfN`!*#(80
zjfoYRnP8mEGj*L@PU`Ug{#^77{^i8jG>^}{o`cwrgR>NtO%07*PR)!bTWrY278H*y
zgf@goU9K?XW{=ns$~`Vz0Ku~aC*dVssQ7s#FdX{~Q92JNsm3Gtd35c&f8r)2cD7(Z
zaANDHiiTR8xH<g|9%EXS{*R3~CJx3I2!&m=HJiMsE`&J8l1;fh$UBp|y0OPHP+O?r
z{Eke!%Qd`wFw^C#uyUfI9PCsH#ZF4l7QdtYzlY7V8m<mb&7F<Z#jytEN<2q+anPvO
z>DY0ZLjO6)W+&vwPR{oaJwDk2ej9*&F{9QJ_d^m6{w^AQ{8ILv)!IZ)yH$g+m7hoc
z-bx(U?6JoEJV&hZI+Q3M|6U00L&=X0lX?@2bKX9T1ApFy4)TZU(Z!L!#wIiV4LIDD
zipyvMSKu8KmLoB5?1Hxc{-y$Zlf8k*Pl1KfVy;iC;&<WUmM;a~tK&7s!+~+KVydTn
zd{2jx^I(H>0oUi`R1cPfY|XR>G?``pi<LOVXHMPUs;9t{U^)abBdWy@{A?#$sGgd2
zcrBG%_<uI&`(^pSuUfU|f5bts#%sSii>IJhR4rtHZpbCql-8I(2E_@fCoIjNRfDcr
zD9)DVgOP~jkk|(+xDZYQ_<bHfcbQzAd%5^IjQkl?4$+<#`c2L2?`{p%aJF|}uNw*_
z0miS>6BlI>FZb*9i2s~BAHM$f6dV^!&z)f*QV6F1p{&Ykr97@Esehi&XBDG;Sr^49
zM1Qs<Y3Q;qitXWMn*C0DqKaA1|6<f1f4K8^Dy+?<9AP+}>1_zC+~<mWa((O~b3H42
zi9yu_Uwz{OE@J6}){}K+QNJF<Dr+Q6dWuW7{UE;!X|_b<d&J$;u$qcCHby7;u(rFt
zcZ|x5Jn>UD)LfotCWlgzVnP$k%9Iy65ciP}iuqzV5UzjH4fG#yQ3zRKN!ehoBv*Hc
zI=CR(sT0+6!mWU)van{m&H#zLZ(d0~DSv~fr~X7(5MVEzS~2_PnQ1JalR$kXKLdRI
zSEVqk*_bA$j>%(7+J|SS>Bn_i()egK!d&I!HnE9y{q6niqlxo|qc?n@h31YG-Z@%X
zfjYqlDsve5<()_1@Zw=|uH^CAri<;%?+#Ykv786W@dChd4Jsqn0RLvTwSzFMj_6{D
z&k2Wwvxm>O-y$4naT{j{t(P|9=~D{E>if91XF7g6f7;AMPEU7y-%F2}eXLe@@2flE
z_pM^tYLmO&(h^<sQX(c!5bgn92T`6S{ALaASDtLWr~v%k?|2`@5Fxfn{@1TAt?b<2
z3oo1BQ>q1Y8&JA$xCn{gS9+^?#YtC=j?ZrB9T$7;&<Gm7mc*dOH8Z)@U@*_30W*IX
zT9v1~wI1R5+>6QBKPV`zJ5Rsma}CQkV;U}@nA@-fb|(=e_F=(1XYk`6aFuu6LJG-t
zx7w-Tu9~5qw)$iwE?ariFBf|qzp#t<M2PPY!PU~dpd|Nh#KwE4#X(7gfa-s1UmCb$
za5+KO^cFtKsyF*`CH=P(RpKE`tDy;_)5Qc?zrW!y9hYD633_?-xLo7k*D%FJ!<@_I
z>r>#u5Tw#kbsisC@U~qdy`3xhnn&TY;)ePu=>O%w&CJa7oq1&`wx!Xn(iIQqO5azB
z?IE?c{OOpc8;6ER_lg5)w&}X>g+*nbpO31nKHr{oc+R9HeYw1Iee?lAkFu_gTi~AO
zMX@R3+F-<KRu$vFOn^VGuLS>p#XqlckJ3cxb>MQ!m#Fx^zdJU++*+mZQC1U}Vx#Gu
zihNxvIpyVj#+r4x{)K7>lqsM>JBZlAt55O+V=IonOa=k}{?O<A$+{6rg(>ua!nF%Q
zS@-7yLfiG<+$&)$N|1$G0kWh4&C@tK9M`gF^SZ3M&g~5PyY~&qDP@F&3N?7zIR!4Q
z)u<R$S<sj56YeSD#3$R2aSWKpMTsMwpyBo+eK;^Ez<p)wIiSIuo|8^E*uB!*=_m)8
zyD}O20E4v*zcC`V_Bs@tMm#=Jxjv@4{{B=NM1%%nE8nZr^Mg_%S#1;veLwpucaa&i
zKGmbL1B4%^duSpYDnKE=Gv0aiixtuXlU@XJE~eJ02&{)5=DgKMpmhvZpN7kB@jj8E
z*;gM1?{_H^fm|T~j|x@Uxv?~|<#aa`aA3MkE{$uMNU7-~eOmD9<|5XXMC^s)i!iN+
zMm-1lwRq!Yz9JpY@C3XBPF%?f&C>9q+h18#Zt#FP*arKFD#Qa<TK&nZ`-bj-t)8lY
z{IaAZvYbHxCn1#7eK2)~(C;&`J;#S&&$6Q#i$w8FwS6C9Pn0<*m0thMyg&O1s3J?5
z>u6r>o)XPRe>__wNaxD5N{vDZso3u;0)_{-tdX<(Ab($fYq4aMl{b9kN4{Z#Dh-Z;
zBH_{cHXGt(6x4`tnJzpYW(HB4FDT(_B5I++x(LSNa3LNm<lb@`8J;fO#<XL~bcp+;
zJD%hS|52xNpS2U3*6UcbUk2!+>DBO&!Jr5P7m;vq>z2BIYV)7IpIM{I!hyS%Am8uC
z6eENwSew`D$<SPD$Jwu?+;S`pN_gedC$35S_Rv<w_|&7&VXYyl=DMKcpDNTlKku!9
zFoTecLJCj8k+=qv+bp*DpRN2FHTur8$*Ntv*0j1r2KT;%G)3y_@&GT0#8#obZ=?~<
zBYiL)i6y;$qjV~D0k!h<!rS`gC5eJmW&^F2g(&F1R|9^{4?KFrgjot{(b~utd_@NL
zl+anm_Krn@;qx=UjHXFwC`ysGvFpw|+RCJ>IMdWZ-<%f^{+88H@;(a7Y8}oIM{9k;
zY_vEL6NKWPP!e(W2|%cu?>0rwsK#qC<aeQ?4C)qO(!nPL@+jH{kCdsi2@C&)pVL*T
zU#>Pbh9pm%pVui1a&yp`xA-luS4nMk7&f|h2m0(?cwbFlN8Wl{R61G7zj;9aG~m(@
z(2stz6LQ&RU_b!zkh#7aAD;7@1%m)RFyO=ijHbT{!SwY5ThmQ;_H+a8|Mq_rGhvU%
zEzZ6ktPWQZ+Htlbg8V6EF?s0f&Ip1wVq9bvJna*Y4F_u32*=)yRTgchf6_i*jHcI}
z?98s9)HNX$RNW*(H!q+h-#$H(FJHikTF)4hlRVf}XMO!KJtqh(u{Cp~N@MTH!|#T_
z{~i&s+Ia+kHlg;_#UgcFMBw=eIb`1>>=l$H5r-RkE@3@)PO!)obF*@Ah=EDe7Qz&&
z%v#RI)T=HCJD0Cr67DLnjm&p6<hWbaKMLn90_mm&oZv%b7{@H0%Bcp0lhoC?dD|oH
zeV6mCw0mP??2^$F+Pf8%>U1hZS-SUZ5saBI;JCZ=<W7#yp9q)0r4DaGdPY{j^`Jwz
zfTP#tgI6D5*rU1AkQ)q`juZGV@43M9H4N7_DDP`&u*H52W(Y=TQr?2R+7(kDjndgo
z87rb9zM+d#No9lNp<ZgCEZTi)m=8(s#)@q?%})5b1hjiBFN(W#kT*3r$kyDIUV<qB
z0V5vhNRbb%H~N_b<#A#$N`s)h?w!=}z2&Jd+si3-856!*^wRG$f<0AZM^e61cQIIF
z<JozZj2LH`HW`Y}W-A+0A#Cg%Bxc%(npQf6f3P=MB+4)Ki59S$)A<6D9oQp%>|;!s
z*nKW3H<%#fdb~UgC%*cUqQ~z>vd00SQx~Q&Ahw79hAkyiWY07*v$+#)P{#d(a`UN7
ztKlyx<<T6AijK}skHB`NN&F5~>ud~8qfOZB`pU0q`e_F%<qfz14H;2apTS<~QWWV-
zEQw}Qt&v|}DT*nJ!o`SLrK`jPj}HtDuxmZCbmCxi;iwBEu&HyU?1yW!ju-(ium2;-
zzMlJ9@Uuh$!ds*u3t7nC`+n|i$e3TJrw2;T?_^co_WQ>3yQAIbvM=^a6cu$(78Nw>
zR!e>DKB3J%j@ciKsh@xH3p_q|BB#$y{cRsnUmDV<VZfHpBB*g7N-*m60%>eXOf;4!
zyMJP9PnpD|M%!ZrY3i#_g9EZF8MKtGb<26bbNdwyu(J~wmJ;IPUovcZdwGc!sr#SY
zi7MKYALu>xv$MmSaM@aA+v_SO_8*Lu@a`OObc0XOv&`F!x7Y;ylDqs>+MhRg$Y|dr
z2!_IaFpqbI92+`3O?Zla_Vg$;%}xLQospAag_g*JNfD{Q^O^|UkwCBU^c-Q#7HChL
z@XZwu(|oKCckhmP7D-)(b%T?T^!WsL73HP=o=;@6uc+KCq;7$rJcuv$8<4)849NIw
zsnc=p<Rmq=Lf(!4UTbP->M^}D)(;ynmXsl|-0?Sv3jguCYMS*OeX5tZ%MZ)kfR)k@
zz(rKCW=^~7_TL2+a4cS#m85%oei;!F0S@U|&5E;0D@?Zf@W&GL@BFQozJZkK!@<E~
z7XORIuRmmra7yNe7=yM3yKr_RBO`&nnHG<4*4hokY_ZCl*-e0-<<8aVX-j#UY<;Vh
zRM}<{J56_%w^Wi)><*zh>yz<r7dHm&PK&B>ZaV)TX7?2MS}b$6<}SX*3X0jR;{m(L
zvHd@Xe>)_f8yfChO+D|V=z?^|^8bZkQXj!N-zTr;01Fd_79BwzN8h<}If;1nz1wC%
z;PjGr&}uFMjQxH<?1-g~Inkz!M7bevlPvOax*h*awvY%syDVM*u4(`GZ_79{=&F5j
znztX0(gKDIz7YM?r?tnCv;K`%f60x}nAHCY&5`t0-jZai>nnO}2$QC_qD6}cEhBh{
z^2*8*<4X9{X}Gmf1yFIVItHR8zK7^uH%izm{g;|Ak=AVCwPlX_#X@_8qV$hBES&(5
z3^t1y)0eOH^|LkYe{Bl(b(32>)+nj*q^VP;1MY?!_5as_$<EHo&IkU7k&mxp0s+wK
z9o=E+zrDOoDddao9(jA6q13u<(+#smOB;VR32`aGlEixDEB#v#nM)2Cp{$8&5Q?}$
zBNCp6uN;0fM<@46OurQ+auZ3;4=0171vzOGU3s8IkAl%Px@6FNGp5ysvQ|k`WyvDI
zvFN3KX>A$w3#VP{+FbkDXc5&gf^5r9LR`?9T5{>`mYwNiTeIl<TPwX;8eG^L$`qKO
zf(H+#B{LP|!<^!W#|rDjOURUrUo{Bk&!&rkzW5i9)t9opihpNiXB!hbn%kS(y=`{}
zf4;37-Ej8AR@y&eq99+49Ut&lhr;^-sQXSR)$55z|Kuoc*gSWZW<x#YfBM!LtCAHJ
z%yA624*2iCD|M~QFaI2YoioGh^IS%Ae+-+Lq6q$*n(iN8=C7b7B{M6_YhKXjeGZH*
zG~54rW^9Z-t!ItR_NJnU=@Q6%4*IJXuc~(6xE|?tGWsRzWZPWF<Rt3UPS#!xuwU+~
zxxsY@H9Y9FLs^zo_iiO|IMBgs2cc9Kax-!_m`yy`Ftk9h?;H}$0-?_4-?V~uFG9hD
z(=n#tezA!uC7oW<Nw|o_6UK*GLOc<=W!#RK4sZFJ27kgV&5aG*n4g=zMBXOqfnxW2
zCJFwVv2J)uKXdaGL^-HJB<2G8dG3W0E#q2akB(H8Lk~=Nc#1}+B*}5qz4Iu~XTdih
z#PjV%(IEF&l_8}R`#6|wDSWS_a)Mj-gCI>P>qA3HPEpvc_A#OoR1#$(O&^6V-hZG#
z%<*t?vbAGKl4@J(DqU;G)K~16>-yro_}%`c+{G;6KRa0glk#F${cb=Z8?;C@2xL2R
z+voahe4jV(8riPr4{+U>TK-_#F;oQ{M8s~UEs3$D<nQgdSy#ZU0^AGcc5eS@A9uKY
zK5N%j^Z9dP+Qb)fmhGLt4Au~B9#E>yezUud&-C?c8UN#oF$tJ6%GhMaX|s%<EX~F2
zzfhU|QkCaTjj^}8UnBv&ZW+A}waxaYxJ!S7`>YAKV#0}aA@M}nC*k`Iu|r>_A4kQ@
zdL915c-(4RQl8p8`FxT#cYe8ZVHn+eI<L?b|Hbk@8v#nl6hr90kpV;^z02icaT|SF
z+bjWG*I!T1UEQ{>(48{c2`E9{BOaYPoyU4Y>-mt#Y=Z_JKWD&s1&0M?KG!<Mv%l)v
z({x`?K1&-Dd%@-kN0+AO;ryv&NZtj}enW_3&?9>>rQqS|?C=w2|K~x+&(6LY@UEa(
z)`H+q2^K96g%C*v6?BFMi(-UkAeUngM<l;B^RM?GH*v(+s5yd~siF?iqKqdA|DMjc
zc?nih@hswiu7l(nRkkOt4fluSp(VP-fAuq3=&3u)XvyfAjpku=f^Cd0;<2#soKD&P
z9VOa|T3v+ve+BpQCmja#*<S4jIm_8>N}k&2*{ONhkNha${*RgXS?f(ac#`N%Hdl;H
z8J)nxxrKOKb^AY!Qi(VlS}sgx;Ss?|R@Ze^$>|HA{fSY2gYOq&w16p>CV4!WjU<UJ
zhOkCx^Sa8$W^Wa>;06zW9{wD>5A3_V4mtw!3XWaula$Tsx$JE570-0YLMO|9A)+;^
z3wcMCZLG(Y5>VxKct*;zWbmH}tH&NImX8Mf>DMWGYo3|C6@pIm;D1V{lvDMj5=*#v
zGXj{#vo)v^TjAGTsbg0{@#3yTL@)~b04sN?_lsNqQDoXV|5w1|W8HBeN^<e>;kg<g
zi`r}D5D(MGJP`>6-0nC<=7)NE<NY6xnTSSB<24b~N_&_7n?Kz8(&h3LIO9g<&bH>#
zGjft>P*0o*?JW(T1(_*hZ}~cL53$TV-XQ>tjS{hCe^)vlh;VM8EPgk4UU#`?6#94W
zHW&=LWC16HqZ6*$Mrc=Z{?TP{4#}$LZ2}Fn*IzkC)PzT$&KCJrm&fTrp)qe^HB^*M
zj2u^9w$Bot0Jmf+rVS7tBKyJ~XQQ7IZaMXLtu_ym)H_cszgSWE*tY(CAu1U?HU<0~
zVEDn9A{ENb!mR{j#Nacylfo9*VmNNT*(`gNGEt=A!<M7l$B-tpqSmW8^H<}T?1lbO
zw@IJOqrO6im5v%w*Y2<b8C3As5Z6bT=HGAqb|{v^VXM;CJre$hzZR~ZUhl69yF7oy
z^i7L-$?PUEJ%1K(agbZn=kuXt0h`P#>%TDh8YZ+z^G;{k*LUM*DgEcmCr<SSM`1@t
zwX!qPvi{2hUpcA2R#T<PGLp~%9-Do=|7b&CDiw3H&HFlm7rXi7NaUn}izLY^7EK)R
zF+01(efd3Ah>A{z?WMDezegC6HdOb$X_Mi3ShpUyp*lUFkGgyyFTGSDECUKN9*V2<
z!PuT6_&HaJ6Thm73QIokM7r7T5K7n6C#FxH&NZLzhTMQGXkE?bUQhwoixx&y(Q693
zIPp#DU2E9Ex4~P?wl!PR-qo{=*GsZz-uT};0jr&ruZFm-Q7KCL1`+dPJ}-Ex%BX2t
zh&gR_`V!yaj8}IE9p1b1;FCrXTJrK<v9U3o0(4ah)7$#fIF@z*461W46!}j<n`Gnr
z)MW$bz|@wY`7&!%l>>mpzEEgrKyzTwzAlk3aN_O*Vi>dzJ00C8mYEJ=<~zW0OI645
zz+=04Mk|OwUj)Rg7iY3bncpLUdG!RNcwbstX12d}v%LIP-2p2I3HrC@mToikt0Ixo
zPde8Q^!2;cn_Rx8r`VJGs3DWDE$pg0z+MGU;9l(Z?^7(#2FjAQ->y3JEZOmnF{Jmw
zNZ`rYZt1ZH&G}{dt0Dc0{@KwbiF;Q4^*=(G2D*HQiV_?!lhf<<8B57}Vb1i_OgbKi
zz8BNVA16V_561_e!q}4{FS(?`@pQ9~A1xNuc*Vb$2f2(aLpoyH@~75&e`zoj{Uj!R
zI(PQ{9E%$OfYY{4e!;@^!nZa*|Cy`55->k<ZtK>hiz%Enh`WQGd@Wy1lMlxXgx+}5
z$W_M|27B+`hJW-@S2?22q~0X!Y$5m?wTFT6ix+Ziugc3RJeI>?I-KH|Zl=3Z`H|YZ
zn2*--qK%EoMwY6uQmRLg8s+O7i_%x}AA}-N<Tk454Zu{CpFdLAF1GBDn}&6d>~1CR
zXub|p#cbXF2v(Bn4y*6ruY^b07_gFDpa0whoSmh7SS^z}nIf|Sa4k(q1&!jiBENEk
z(Raz<Y`lh^lP+IewI`;MO}Nx*y=_mxA*}a~S>VYRZ;RJ&;?#<>c%(jskZp+AG6Wl~
zc0AY%1lS`JpJ{OF+F@HO7gayjGSj`F6p$7+G+`4;OtKQ7d|{rF_{YQ)L2)ljX)vml
zS{=z@E6?+w$wXsv;`kFB<+B_9Y%1Nmk{5BGtz@7Caq~~FdmJU{d(nH80c%ue@{By}
zks7yY-Q=|w8{~O>Y2m9v@6SX&p(I%=bW~g53_x)3b3*8i4IQy>=JO#3BJaZt`=+^9
zh;FmnqsfA&?i3g(1gA+VJ$?GkJ{W(LJ+V1_dFIjgN>B(|Jv#TR!<)XqGi8~%HRZ>F
z`|#p6dDC{sVF$f;tEn$`3XA(Qm~(X0AbVwTHFP&u+zoIF6UG&ymYSdU7u(qz8oKZn
zB>|jCp_BR#RQ9ST$4yjFpRYC^lgH<Z`ThO(y-7<~vjN{m?`3Ko^@W*{oJ21z+h7Hi
z${re=_m268<sqoGJDl+t_HLWD*Xo>hw(EOiZipV^8>;Xi=2>!gwDUSy9N7N4T%YJE
zk*(Y|!Fb=)#23`&g8>=Wb^JeotGUY)G@y5%Ih;ChNRqz74XxYvu_0A=!>lBWWg>*r
zfmX@$-|l}3j5G$0*vK!+XtzfU)=D}?rrPjWeI?`ALkx})`;2{i&U+2i`C{#&tymJc
zi>OOIrhHvc_&$O6w#MuVv}#4PO9A20Hy{1Y$nEqy+E|L04LVr)-{>V0Ld!@G42Gr&
z-I;%7{7B(Fcdbbq%!pHb)uC?O+0%KpRSib)JO49&{f8bn+=$;_aHW0Aqhy7`aG!nc
z`(;!-w{w>l)o|B(X<Sc_p5JT)QKZhkvt{=AowMyu2zG1R(jQqpySa0S5t;r)0LN*j
z8D>216%^OG^w+wX%3jSY^hYxSPAcLXIKOUN+OKA+jB4fXUE2Df6+x)-3zTNi^Ucs-
zoiEr{1{j9=hPxy7L$ZeHeW^qp;{VF$rqyZ<kX8yQq&`|aH3Tve>kiJrDeZhjvK`$w
ze8iMkpH2B}MrJN+Zf2Y3X5@xUbEns&<V!Z~E(2nrA-m!}u5VlzJFi9HSK`q5$&c)Y
z+LEQcbtW-#vNi4x(_5vdEng)bar)4Q3Ly+rPUNzz&6Gvn8~st#<u(0-`~sqz>Li~v
zAg`~&Nb<P-BA61OGOfF>lQ(PLQFh#5!=ILMqg0rF<C;l&f)#eTl_)t4aJusKHB+uU
ziR_~Tz7Wp&iqr57DHs*tMk|@pJkPVIq@yfkJkwC$*f??jo#|5S1kzBe^Mo+4xcJ4N
zRl!ovT5qz7a4dI+MY6W`y36i=Jk@I|)6J%^(h+e=qEHS7ZlZWNiB_-}W}`J&MwgRN
zGXE9=;f2!6rQa!A!=Dh^FYB6Ki@RG+?spHH2V4)G_Sw+ct^v<7Orl^6FE|H$Kd1`O
znfVkB86Zeap2@BLXis%ZYWNm*BA>t)djm$C17|X4b3^GeK2$Q+c^_L*S63q+Fw$y+
zj*?Sh9L=r~RIXuOS(6rE{pObR=`v^-A19E*S(*@fx4?WpoPr-7&Z%ib2mi><%4E&W
z;aO^|$b-SRBmzT@;7t+ZLlME*<9m%&x=hS5m=O0uBc6g_ZP!=g{0_{;wmQm!Lt}#+
zj^KysgZhti5Sr@0cPPoFRmmrtMRgt`m+=rHmuUXk-PYZGMtl>hv|6M}IQ<<@@3nQa
z`*P^yb-Ta)JqjX)%~r?$R{~>Os|Nb<IbuNzJ*Nu-z;hWszQM;uAbq0JqBBx8`*z3n
z%Fmy#+Ge-C+WHvC7y6En4&?_k9v0HxD;nndQi!VA_V0@9;(j~rIwlU7clU1cEdMjQ
z3k3@o<)DB2RB;a^TfKL#+WXri&H34tjWMIn^(Mk~2fE7OFqk_2(Ic)FS{+(jtR3^W
ztb>*oK)mQDczYw}@uY_WOf`NxD#Wn`+{&jG;V^hXyqZk-PtWX6VCpKOw6#*21T!Q@
zg%ad2eG2)hBq(HKjAXQiNm*Bvm4UJ<sq5V@RIBuM$$ZvX_pwA>3`-T@XhIVtWtoHF
zx)9P;4u<vp{V|XqO*H@Kd-Auq2i9gYpMg3@vkBED1ic9o3cIx)n1?3o+o)wnD7i^V
z3I{H&yrdzfQGi*1p;uK`1BMns>qW8HmzXxSv0qIt*DCZ+y(O|21d7o-`dV7U%o48i
zDQOlfS;Bp2T~$>b%1ZOgHz$?0YQ+cLBcv)&ai850Jf4o7ku{g$e&#<Lzr(7}z}fx^
zzOm_Juay<=CY8Ig(G;vqnWUTNuSI$$v(`qzeGR9L*2;~AygAlG)9gm_0nA5T8-kxc
z(H<BU&YY+UCsG@G1EYCg05f~!u<~NW#26b0#rwWT>%^dpS(!;TpUAU|pWr4<whmj7
zn^dAZvrif)Hv-7blq%TgGfOD}sP)~t$LDd6hVU~pJ^T9wY|Son^1K|pob-c?hWGA?
z6>sK0i*%<>VB85iLc?%k^n&qL{MJ1W-A5fn^{6R&Hg$y`LG6opi#QFRS~LTfJu^l$
z>LH*^q`j`MIZ+dD#7bQH>KOyt#G1|RwSLnMsOCG)DmH!;u1a>fzxoT1uLY%);4bKD
zsp&6pW^wJ5VbfqhkofLko-$;2-qno>uWdWo1|RRh>)EEa*vQL;{r!QKTWQ;~=GQA#
z*DHD!LC42^`14g9)&9&Ib4st!iQe;hEcx@+dOul(GEyj*jY5P)yrZkQ1A(zwy5c8V
zMNPypudvQD?UxG;z(i38IT%cx99o=(h|JN3goYDq_t`=P6V?3PGq@5@B*boY`>%!w
z@GsT>u6((k^+CMB_V>;<w?45?cf>)eo81#mhqlR0q`9vP>}^tG5t@TP@g#>M^hhY?
zhMW@`mPDOIAMs2yDamyui#*gI)6od!-B(e%;TNsS&ZzSUSmabfC^5%yi{ilurF6D@
zf;DvBiEUm@o0G&|0}6@-;Q8#grJOxK1y|Sd)m2a{l{Vmg>7kJP?#Kpzq+zAS@0rin
zgTliByLH~R292Bf(v}@nTR)z52CS`xET<?z<w7KG16Q-1|7PySt0a9s3uC3}HE08=
z8=!M%xeU-LtJt5{myC7BR?6(JaS(O+4l@upgKKyV4UKS!tued+X659Bop6k0dW5Qz
zuL8sD0(VDrtQWS!JlT`?*3<c7Jk5=)H_AdGI-KL2q`NTUe6(=4*ygd{R*b_^KvIL<
z%8LD}_s~-rkxrz8opZjhjG&;*s6&IwK)-q#z`y1D=BvbWV)JZC|H?$$^W!b=D{Gq7
zlIHlrM;yLuC1W}RkzW>6V^cq^M;xpsbGCS#INQJFgTN4>w=nW0f@IVo3Dn4<(Km<Z
zftqU25D!qQdI|0mS#vWoTCe9>^3i${e%xG~KJn1**q~ctoBG%aPkd{7`fo|^o!Bwp
z4~E;s!XiHT?a98o4rRkQE@R$&&>B%9N=D%Z=Zv*WvU~iP%Dml`s9U&?+ZrvXu)ZA6
z4;%~U>P)?Lak)iJH1C=aHD8D}#;YbBf}v-!zDr!FQEE!AA403yM{(?Kcty9y_VyD{
zpGl;ErKM4TO1obq9(B*(-59-lhy(71hG)R?Dkpv!eHf989v1rTd=p-hhoWAD!gh7F
zmtLkSsj|gbcgZnw=<+H+NaW*0wE4@*1n|6AB|NQ?v7r(TSFBwojV@&?A(C2G%I8Zs
z31^?lBOi#plWU(Y>wx>4GxMD+4mLFyN^ZB2pQos874&Tkd?<J(T~2&I6R?nibH1S0
z37#)2^Xe<BEjhS4>+|aIP5Npwp^q(L3m*x_M7w`u6F~T8x$~D1s`(GVRlhg;3v+9B
z$)oW}{;Yt&f`?$}^BY`r9tp7y{&%YtM1Q;EnFFR9YC8SPtj+ZjALidz2McZZ#An`?
z0x?-tF0&~|pD8w0{B_|31k-9bVNTfD6^2h-pQf#s*flHL_eW#P)1qcyez(FLP9V#5
z#xDArg1mi)35G!z$qK*gXLY7Rz0ox0rmk^BOfD)K6uT_#^r{-w^P8633UXCpO$={b
znL?Gcpg#7<cXZUu3GJrWM^prfc3^#QBYT)N<K-gc2*xe3lnw)+Z`uc^wru=(ee^2s
zS1SvIbTeWCAqs3hy0n*8NTj=!-&?Csyxx#@!8i{hxkB30YS!Y?iB~kpf!8^!!TJ2^
z2*DWZTOb|<wC^$IhikJ{{C1=dH!uN1l}jF%7s$S~`Ryt;cyN=M$;o60O>OYUc1DA7
zm8i>K_Ga}B1qcu;nxXe!ef~U2F<q7*f3;=Djg>x$2P_nv?avuRnFx%w(#<etwukk<
z-&Q$ommuy|MqO?l(BlG*PU&qpTVBNd*)RI>G~iquS6aqrG}solU|GF0@`bacrlMwc
zuEu~>#(nK8y~V{-FdcKw@1WS8aBa+_E_c0fdMhgaL3sl(OK5#AgO6d?(VtjcOwzr6
zigY<c{`Sp2#WoNesO;MhI-XUR_48^f-BU|sSNg!NWE%FsxOO@O{`*0f8u<fF^ILNN
zuN`PwV58W7PNFu#7#u)Bd}Cc?gvF}K>H84g5}e}ba+a&USE_xlgqNB-B?$lB%Qlc<
zF1rRoDi78JB<**+tvhyJ57!07W6Ru+!w2O)6TyddiRt>mg?!}$DJlNxq8BGax5<~b
zOqXUov7QCD;qoZLd9)4>Hkoh8!+OjaXm+0m)zL4<UsvUFswH`yU9x-|Q4>~pByXJh
z%;~)Mu2wj#UJxga+)Ct7c{_fQ2^_e5J_9o6RFg#d*ot`HTZumZ5M5{GgjJ$7*(@sy
zhuPz@h=Rem*8|jUGMH-jFO2IBi`1L)Mh7NHNf`q9AQ56|ZH<}U^4lq2rw#sJ!YUYB
z=drZcJiW#L<O$Zr*=yhI=0cWrobc9E)JP-NJ4Dq>LfQ^@5a8kApX!xvo8DjmAZJC6
zzJ8ddlHTwCeKMz>?EEecQ&sk!1fKZQPptFj_gWAQ`gJYLUaK&i4Nelg{?X!Map)Th
z@K%7qtTV{N*Vk8qN99}+p@vxxIcN2`Y_Nb)6b?tf)z`nLR65?yprvQa?_we5I#*M5
zx$-<{KfX%9(%8V_{LJNQEhSe9xLkfZ_t^g-PiX`s$#&c*uHrVrdS8U5BuMtbo%3i^
zS3h5Lz+1SHyujLtsx7mVaJse)`aP7PAn|5c1HpG^>eKq_2Va-UU(=;OK0Qwxe8LCI
z$4;-^_%<S<|H+eUm<bse-K&Fu2S*2jO{>_Bcr6>EUI0Nea5?JZak*Usw0yDE5x@Tt
zBPyXTqWq&$M{UtMD~{{$Wuc?^1Fxvr5lxk)2Q<9T#>i}R`}wL<nO_y$a20-HHch!2
zB`P*Hqdu`&_1S&<@cyj_Mcm3eg}TIv7<DvB94e%ML(%jlchAMf)WVm8;yJ+9!v#DP
za-qQ;AEFt$;M{g(cGEMNfP#gV*R;GAV@FF%=9@pFjo6p9w46tNv)%+>^}Z>0VFtd{
z+3hrov$Nh-){DRPoU{dR2k-{Zl9m`?>~fGi<3JUN8tq3?MJmW>)Ci2NHVPKZoJ721
zL=#!U_D@?SF1pd_Cdug!4?|X|*251}Rh|DL$dg5aZ@!VfL5(Oq00P`nT4&#Mthb_9
zGVLl}@cnRmr}V(As;T0(fpd7F&D3|XzgtfBG5cMKn2?chw_@QEr0%G2V+9(W8*w)R
z*3WH!|6OH+0SwNTB=Ts1;q3A?G$hf<*!s_9w)rE)gmImkY5`78)=(jZ1ehWmtg4~i
z^9t_-$#{%(VTywk01Y>$9yfp|zt1MNIZeB0Q!gniyOU)tx7r~P^HC{Q)tL4kb<M%`
z!Sz2`RXQZOA<s(`eUposR+0wdS<YXs*7B6oHH<JAyntV~V>%jr5!L9`sEoPX0j|#l
zBQYxRuRr@x$PY5zi9jB!ETuHuz-w|uiM;$!;CSC3L^(+U-p>9z$(&#mEXZ?Hqkxv`
zgcu__s+;%O)+f$ccJn{`FX@qJBN&T{V6b#+k|O`hZr33Akc2Nmy|%t_%T>(G*f{;A
zd65HIV(csByYL&X=93ONL;pyo7^WM?*AGD=()B6pCAGly#N5T!3o6-n7ysN8sW~pl
zU-<nmVix%1U_z^u#EEY_KgUeI_TSRF`e$K;<k+SEyc_X+E(q!3z4H^3TfS)m9o_QE
z_D}`Yxy{XD<>#BGnV>)h46yj<`+lR<9e30^`^kZKm5!dSuT4_8Tw%Pi&PJW*Ctt9K
zy}hJzG|twT4C;^(32QHJ=&NJR3A{R`AI|Ze1I~qN>7%J(5k}^w>UD&M$7!+njJ!(Z
zw5XgSj<+bYOi&p(9E~E!-;ue#sOkXsI@Dt`GrJ=TUl65rJDocn-j(0SZ3GxVgcu#@
z2RInT@>imiO_j{&TKzl5*TE|foX>jCNrQy%xvR?)vM&p4Eh-<a6u!#|D)ZEQ!~=r~
zb`s6p(3p$BC_^e%8D<4GxtQ>NWu*1H)FNLl^*RwX8jOv;K8i$_V;us~WwJ&m30s*=
zZF;ZJDY!0mIIWtAW*Mr{WqrV%!U9|sgS!wnc^<mLh*2aIn!U=QcWHH>2t=-#xZDp(
zc+|)6?D=2G)mY1g%toh-vRIi$PrzK`6N9z+D_Q@e*xb)vcK&148+HfeHlifyyx3Tp
zQaa;Q&TP#jZ~qSYrpX^|v&-OO3OK%Xa4RhLdiOl{^Ff~izrtIH0%cB^VG<|&zgJ27
zioBFJr6L-i{1|F$YVsB>!t={zc!q0tG29TS`=?*=(4)Swg-iXBlI!@aoi|`kx+71T
zUp3dV6)-sd&f&ith*bh-GKNa#a2|{sZ3$RUzC~Y@;G3Z)kFZK&U{@;%F7obfC9`6i
zXH20NFJPC8XYL}kiOEYR4!!#(42$vpd#ejJ;@ha(&pBahIyEoT;x26SWAeE96LW2g
z!OFAZhu*EoC?$d$f5F&gN*M6d6za$D>iHJGU58TgVSAG?*ACU{hxJ0n_)jUDcMRG^
z-qfEEF~WNn^2nLnUDSH#pRC0VkoOhZ(o*a7(8|ND-)1=0Tf~Y^yH^yd9%wk~y>x_-
zd>4W%vS{_K&*a6_)Z;nTa!!oCcg?>Ag;E2L{y7Rc95-%V#Z`HU(?tV<Hzz_XiC>zd
zvs*Z*q~6vi20IiuO~aBoXSXP6RV|wI#qI7~F9S!zR27vxZ0|+POi;|$*5w7hI~Wft
zsWVl@(wBzcX;s^gj;FO`51R1`F^9chs7QMYr(1kZFF+rTx&Cs>hcvnQ{gCrE4|QHG
zQ`T@z`c$fWDIOs&4hcZoeebIK$$xxR5Gl6C(^F_#q3Q&dkLX{7+?7jlCo}O@X~O4B
zX!fJ*_VFDmdOAZYz6WCU)m}aeU4b8vQFMlrX9r)p1bu+R{>00U4l4<4`GEe{i@fR0
zl|iEk9BFE4sne#oow0#NaU=gVNLIK4H;7(}AMrzmoyQYcEC7zU{cl4QpeC8dU#UXu
z1DK-)7`OZ<FWPp1o&UWBsJ`)9M=J=?;^x!9m&tzTTTQaJj>nm(@BL?*(pXtg*WGAX
zWp-R?Zjx11-qUz=MR2ZvA|}~C9wEOoL%x%!!_sxC%1&`7Mp4e*y+L@2Ci2~*dv1b+
z1JJXAfPUr&3XytUO$49I82en73akjuKS^SU)n|>G)@~Iy6qNb~9$ko*h#Wy%8)}*f
zlFrd6W0xn8^gC~7E@E9S{O<Scsj#O1Jc`hIs?xRG@D-M=Cc>t2C%C;@EY}Xg0Q<Us
z!QS$}Bzy3E7&?lQ^*QzLh->#ex03LVH+xP=djmLsBmBu~N<f+Q@0%*7EjrLB>NxG_
z0!+5hK{@L1yk{k$BPN_NUVIx4TjuJV49yNt)QZAU`GO7##6(@>Pto7cZg{;ky{E5#
zL}&Q!8$LM@X^QT7=J_~$OijPZ38lyl>!>41OeC%0d>A2jXrdl7aGrNi&;N+fFgpE9
zbOLxMJJi~;4(V@fz%?#k{adZNx=3NY{CD+D^YifskPO-B;^Z66O>sd<6sN#OtyY9a
z)G|l-SUR=HGCNngv>z|A5-tugpW~RZXa!y#QfeXBj}M^;;*9+JY^r3#UJ{jF+cOVh
zkHg{>X}Z2b1#jA8L_bkbFVva3xul|erR4sPrn3%f`hEZQBu9K?gfc>5Fp!1;(#@np
zN~8qoknS8QARyhPbPW|4qeE!~1Y}5~bW2M=@9%RQKmO+6j~(N_ulu@Q*Lltw8nY>s
zIf%?Iv4##8hFlK}Mh|cO5~q-PEE#gLc|$SZ<X_(?GaOy(5DEDXjW~whd+9XtLIKiq
z%{}*ob%H1Jr7!^;?CVL~7X0^@US^~)0V{z8s(~T1TA08Z#l!JUV;DzhO?!T(+}=G8
zg!5=KGmf;BNrdoM#5Tk|K0EIWb(`7=OfR*46B?x@fhEZS3<}hp_HC~03sur(*u?1Q
z;zB+kp$g{cf#UJk@RY@FL7gcJL$)hVPZJ~=c65j9T%ga@W*XAr4I%&r`;+<g>Q_v)
zVaX^yEh(xPr=UgvM9kzqX<%J;ufOK#<|{u2cJt~^BABihZlW`f_dgeIz*Tp~>INHd
z1<w}uMgU$%QsKsEe&ALq?Odabn+3)phYX0#h8%slxw(N_NnBay<C<sg0a|rhO7q*>
z^A5m^9u2XKcb+qL($1HdYTTnh{ri^sM4R3UW$QLVXXz^SnEPWS<o~O_zmoBP50J#o
zw$#Q}@C>_PCI9Do6ZhQy$yuszU;t|o*39=*ixXY+D^B%&Ea~rIXZIu$un>R3yhrpw
zhwB{YZ2kQ&hSr9kXys+S@B2O;{UKrZBtTM9vi?35%O&()m$)cC%GUpJxOS<qoGj}0
z_b|X5+G*#xTb(2ulRA!yD|}T^{SH)v0SUA^!aifX4Gdz#LQv1AFe>`s-+y}J;HiK%
zdblv?$h}iG>lBlwdlP~x)BYamG}kyh%-x!60(jR|b6TKphn-di{W@epWn=7(G5g6o
z?4W**$R88&xhw|`>(32eId1xBpiHK1j4b{ll$dpOpMFeWY{S+%zx==qFeL)|$4z#G
zQ%Gro|L%s$GgIouAXqke6m9L=DYQ7)s3M$Mb=<(3X1)!;GM4w8--g*=Ku!BIdNVKA
z-=IMYV5!}wv~){7+_w*0C|PLVlVntI{c(CPOL6i=Sj5ETv9^qnVed?$;8>^UC3;LY
zWcy2WBBSZHcH;p}u^Dgg*9i~Y-nKApeu1jN#DUR)cA@z3Wg^Pdd1))NU6g<ub#`6`
zBLZC-1b2vbvx4Nvb5`fwI6&Pt&&<r)J>&=cj&Zz8zeX+oHTM_J|N2F*S*DufzWv68
z62^Ql!kgm(R~Q@;mc0^hBb$1ZQ|8`HpMfU)s<NE)+#(Ws508aRA?j_`c1s^L62dG#
zCK+Oir13p?!~5S`7Birlqn^-4m8Z5in0YCH;g#t*Bky1L@I2i$)q7B&+ns=}9`|Z#
ztE+1d!7?6v=jFwY9we|=(mXSZE0N@UvuMs~GIUJdptC@z;B_UME~q=i^-_*GOaa7V
zvtSAiW44SBEvfryGS?85BskKb|1McivDc3=$Yfd%U7P|KZZeC757^jIujRxUTN>|o
zB?p_(ZZ`e*(XApi<kK`8%(i{Kh{LiZ1ohVPT$l9zJT7%4!K6P?I%xUFP(M>2K<2Nj
z*=d<m1uO!&b)t+VMictp@j<`m`?-k?Q+b?n1~J3aE@rK)&jggpBsxR(0F_yXP`(ze
z;ibdkm&MT$USZApc1#Eg*!~;8T}-x8?l_wGk|^7eM6s?Q={>sUzu16wrKcFjhc86^
zzMlPEOtFBv{)d(nr5SHMxIOw8kyv-`Yz9}gj8E1rb@&P=iV$~l&Ve7aACIc@2Z;-j
zP^tB}WJm+MwZ1M(kI9MJU%9edh-fOR0U~$0_}{>A?3J<c_9-fW+-+m@5sP@Wy=sco
z&B^4aj>~Oe`G8tmN@v=8Qn-+?w)bBj!>6$^UZ2*5goOX#*)~Q71^{{}b<nyIA$9y+
z$hG|x*BwDj`|x1{q%W(ia>;)qm__2Y3xDO%Pw82Pu8WJy*-=_#49sr>;_mJea&wk%
z1#mc<J(U?)2~cmG7xOv$7>m96p1uQ@lx%s?<7JLJS(T*O7@XCujsO^+39?T2@a6AY
zla*8~e%U4UaryF*C24BDw%yHvfmiSfXD(31j>On;NT=+aUq)u;!D6#FF!y7PrdVhR
z5&D-^dhtQFX1m29hiBLlfKc|P+oaove*Pcm_X}DZHKAo$7r4@LF{pjlxscDxlA)kk
z?u|-f=@}d-BjMp6W0G-uX>oE2ogw`T{yvUh^js~Q;5oruz}$C^pSpDNx}<Qt%90=K
zBdu!XnkMS8#%ype3lSv*lBTYb0q>zgQW~sIzh!!7<WEwAh)Q@QDWVA{nCdxtkpW67
zKCZ1nZ9~2+?Iko~3HzZ%PP6&>d*FFG@+(izxHxX*#77QQszp=3BdYm3{$78`J#blc
z?rR;fo_?Cn%xV2choA>{;gnh1aruqut)**HsLprB^puq6R8Y1O2*6*b2`|w{hQB5f
z#t+@J0>QTXXl=L?=uV=G$Atq<MxQasN|NEBYHMqQu6#qTWJ3mT%TX7d)O8kOA*%yJ
z^Fc*vY4&EGzBgKWKa&p+%R89UXBKDV;iYNg#%gro#Gn*8Bu5sroL4ta;d5FHSdD~}
zBv~<@amN2sLWUa4<#H8-5yq@Wr3N85Q2fkTe11PnF&=%|72nOzPi|JBE#=p6x>kbv
z<*GPLsiQ(grTV=_j2T`OuOe|!^kV|9;&BH4_}qDpG=GLUW%RSXakll>RaHAJcz%&1
z@iCbWXR5DFLobf^9S-N+jy))&mV9(kRPkgoRA^HW>&zM(l@j?&7>)U_|3iZ%6ZiuV
zq%NVtg9A=Q=29woj!4_C0kEzTL;*!FUuGZtx;x^9s;{m=`(jr#N1Ez{S6L%I0KQVH
zttKT=S{aNYD>Q8nvbh<}Z*ev~aU#C|eT+Bg=&-klr{Dr(H`Jo0sqpabb}8gw8nwe_
zd0X9i9rr)g;!t|E=42IIvkn*^(7Kwo6Lm$uXB^QGq(5+VS<EkFXwV!+pzL{FOk2NT
zWdl}|Q=jQf*(9%D_(?swAB{d`*8hZU@c9>Zqk4R`bn*Uts+Y`e+SRP^;5?iHov1z>
zHw3lOh|%F%*jZOgdGTi@ag3L5l<B33Ce_qW8*FNzYiZXj*U%VIm-)@}tM8@;Pf68M
ztZh%lgD+1Gf_6;27t#!D@=@w3UaqeEnv-ZWFN*}=iagz!23Gak8yjuW7~Bhk)8El{
z7&Wc@fC+I!n$+75ot>h1AQrVrFA?XJJW1g9fkMYFGP{b)9Cd?v_2~+j%q1vW*%mzN
zLu2x_l$oh?4blN{$n3J!-`Bxb!F$VLH-}AJ><&dG%q(wmCB{&?p#xnpa^V1SafgU9
z`<u1Y`lh$#n-~gUY<WKF{H!AC-Cyoa&3HA=jT#X(HC1?4vol(lYph;)xa?WF-2%`v
z1KcbjpE@xZ@lD_J^K&bUg9u37+#WA4uX(3`J0CZJ918@V5VRb&GF}g{^t1{V!~R_!
zd~z@nD?>mG&774XfZe;vyp|ylKut3zB`%r(h|Q?rhZel)uWeZ<v%CA@Q%DYQxGx?I
z)^svX2X_Bu&cR5rF*F<gz2%sUc*3R1>#YJm^rkT6oMLDku4!PB=xW}{oo;L@y*rz6
z53i!=P1R>?yX@@DuZ50QzBz{sxgvtF`*<;O@4$>bJdH16!T0DO&@hrLl(-tpK_2_~
zOUb2?p`3DO#h!Rdh&an{yQU60>#G8u>``448fUZV$wPc6r*E&yL@Cd|;ig()%O4f{
z9~$0wajCt6AjH;;59skAYC_6rI9GSPOaP!giGhShF|R|dn7<;c*^MIw=nm9uLd7rX
z6_6j~{=;H4XpqXwaKig2zHr?kc;8faSU+@y7*HP1b)HU?)V3#p2UHM(kf?!h+gMe2
zq+FWrP}JHhz06X9Fkw|w(-zkT_eqJDVnIkp>S<T1w^8Xf>};jReH@5wNH?Z~*V&)}
zdfY(W{^(%O@t(IhB6e_t)^twT*(O)XHkQKKvaYwc7p~)NCXCXI;9+~W`Nf4^ZL*;_
zm?4HGmYEPj%^L9&0jaYL<rwrdYY^G76GR*@@Zt9lEC@gPKJvh3kg-I~@&p&a$yT&$
z-PH6Mm$4^VzLdZJM(G}Z=k?K`G7}1T%9;CZMb+_QbP;ZI4Z!&F#;*SkkO(hzHhb(J
z4Q-Q_zmfQcK@~rFNNafJv!szCjL;oe{SmCJ*r1EFeZR<cKlX=z(SzR?o=mm1kHowz
z_J-2A<8#JKE(8wF5~cqB%};1QxSqUwgR1jAzI?)f<>RIBh*iikeR2U1b%uuC;pji;
z48c{YfC=`Hc;NLv{KZx@S_eA^tc(iXq-I~;05!)IYO^>I5W&58F<paMZ~+Q~*8g3`
z2RDlcd;7qQ{suWvRjCM8S66RsZ51jH(xK^3WGd7u2f$#mUYkj{%cC`QZv*MW9zA0R
zWI=8r_b`{{WHx69&oI;~1r8L>3NcetQ<A>>^G(f?^a&Xm4-Y!;ZjH*laIIKenL|r6
zsySN)LG}%r*so&x0Fwm#6Fr<{84(Oav;^4Vg!}vs2=PFW<g~h_gG6P3>zW*)<9N1i
zsC8~;Lak#UqNY*_u@#^DE|mUHHnQE1sR1_xW$}Qu{reO5sq^gYcuW?xKksijblm2`
zt~+E`h&oFUT^u+L5@X~ly=-6X+iOKNI=8mJDxbWVI3BxOGr#fXe<XAIQ!iiYkzpy#
zd5k~9SDp|2q+u$%dHIA{yx&h1PXHhb*(3Ph$QnX<$L&qL;!wBa?X_>`@|FQG4+4F5
zR=9kh2vjFjMZ-3x!HMiFUZ*IDlBBezN{u6bV<D<^ihXa_b!iQ(K!tS4B?6g50+ewv
z1{TH7S}bZGLtz0F#`c_0{C^@39CvVQKO(CQT}S^5$p3vbXjfnbNcvu>Xmqp%=(9Jg
zc%-(Ph<<WjX#Y~*T$?`7f-hvZUFDb;^>$}`)LLq>$y+h=og#q<Jm;cAG^%5AYwrOe
zxGmT&*m<jK_BZC)XQq(L<a4c^ebz)7`pi{D0zQ;3TTF!8Q)2t==cLgpFr`o<LOVP~
zq;VAlB;ew@zweUxuIlg}__Ss?CKSn`a!m?T&~<{9=yOdg(yH3E^Jr#W{F~Sw9sOmk
z2=O?5$!Bf9Ln2RbLS9=L#~vzD#nXp5k4?LW;BuUyHAsJAY-bez%5fYQKhTX;NX4eH
z`=jQz#|JW=;mgZ`r%EPL(qr6Bm0<N1X=Tu7W(LfJ%m_!wYS5cUb)jvks->#0<Hs@1
zJu|-qC`n5(x^E+sa)4w)|4<+FV)M%io*wT5SoCrNLezCh!eiI1${T@>q_mU$x7_tR
zV%HxKNf`*M4Q#yn_vfN|N7NO~B;$vQM`=(M0$1x}@rw)3gPZV(!qU$lQ#$ZKAJc1x
z-ODGr&w!)8Ut9VZph%TVCu;-(^%CXQct4YD(CQC>1jQHZCxS@JHU($fw(|@Vd(sJn
zc<mQc_iL{{`o@?>BP)A@bG`0!g(&3M!UrX$cTQQ{aD(TKXGg*$i-ms6w?MN9&7CPL
z?Qybs*0;cPbCr3w+=>E?r~YW2lZ5wLh<P<;)98URVtu-8l``WUfPmO3Ur3OP1l_%3
z+Jd{MUIL4Y1Clcp&{7G#$XJ?{!JgGe2RDAkB@R@1%JEO$FWn`|wgzu@tQciv*jP2B
z{L-!PVGk_PG@5I33Np~{n?&U=w99v@4^%njbR{F1XGZGtfkK~zd3-2PHZagY!PqIb
zqSi^bq>I@y16TuT-ZcOBhc@J-m?`*bZYc!Vq$^2-{rWl3RRAkfO4rQyl|Y1_>!|F_
zM5#FBw4am$DVK8D830Ghv?p`bYj=Kj<~r}rJ{UK665+X_S+KN#nit3x_gIZV=g=+h
z{`jZ}jG}1oKHXiJJFxCJZmgm1TsyDRKk=V!F#nORGyD@z{|=!9qZFIP`AQgl%2cLZ
zf4EMgrNF=jbE8R#N5I}f6cK2dppEbIUuWlsj8Ov5)hJ2s-(#kuEiS4gx_O<W21=Sx
z(Rcn^dkWApIn3g%&x%j|?U-A>HRkj{p`4K)B9Yc;jORnj@I*D$`!anHXMK7M4&f&s
z-ksa!{M7b6@j<Jr;OsE*w{}$rO5xWg@3EpT{>e!xwOJx)JyqX}=Ui#zlx7xR^JX0z
z>M^Yy?VTY{GF_Dk7%j6!Lwt5XFh9|VcK+|DPey9%6ZE0t5Ne?a5U5Exj{yWKjn|qe
z%i5h<yU4|kY+p-mj2>umoqhIM>V~WODz)yK#V4!C>38>5bU(|W3byci<v^kmt18`)
z?u<?_-Z5Iu+*2mtX2@Sak9&RceoD<TNEDwu1dfeC!ZFE1S@7&fnL;WA!US_mv&@-J
zDg+7uKDIdG{826qZ_!^7QeAR_ND+aN_^FBVJck7%=t)h)rp2$FYut3F&v9RV_n*+g
zsb`y;_lYHrZ(GNLSJWfAuRkiD2M64)0BxgOhUyWo_}aF^E#ltW`%Zp;n#cJ0A2oR_
zYOp6@UV*)86UcQP(RPz<MJ`qQ^GA{g9f8-VZsK!gn@;R6)p~VNm)m{c&cNfowK>sG
z>qaaw$t5D$Y47MtQc?$3{~Z$h_+U$^zO9vJnulBfYh8K2@8y)K9ls&!Yx3DM{xm%d
z@XhS^d|DlH9vWN>bAT*4|8InRzQ~Z+Z+^9Hm^%g~i5RqINo#QCj;x4qRRkvW4k;S%
zNW(CpoB5dkNPx{Ze&&8l!73#^7iue{XsWlF6YOF1!mZn8J3x+O4KH$0892=~X-1GT
zPfpC?S{mgw6}Dua9WaY+C~S!h23DTl>x%v{tf5?9JR$A2`-|aI=jFVRHZX0ug4_#>
z1zW~^U%V=Fp5PfiX3TA@szMo+pdo)M%L^oe*U?&{S62r&d$gVZdV$`UfG4<KVD4px
z9z_gpRG-DJ{haUZKeX%Z+0Li-?YYKWJIaOVgM9zZFM0|?R{czu|B4gYCnuVHfH$y{
zlaq@}Q){av<9()(f4|ey(`B#fWhYf1=aGF(K)eFhg2SSja&3YaB2bC&(h}Qvl{Zwe
z8bm4bZ%M<vIs%q|*l*Aaf$?m~2GTg*v8-jO>Bu*Eu1s9^T^_FVM}KSXzHSp%8~Pd5
zk`-SCl%XjNa|TLS?`&YJTR7_Bcy!}zp_U>;S_V+M8<X0Uj4L_CATC7$=7<jPiN?HV
zdS2T27im^eYi0>Mqawn+6uunpVEGySmw#SQ!~**T!m%6^`l_X(SS<ET35F-g%IuZ|
zhse=CfHjOaCY(Cws%*l4jR?sb>-^NMLiQ{Z*84K%MaNJIC=pS{f=R$e;7|zLYf#1T
zSAVbjDqITtzTcZFg-S_lpveR<C0l~BK!fez9pam2-Z?i^=(zR4Y<e|NvUE2}j9_qZ
z`?C8{5IBkLS+|m_fYKpdEJXyyze=yz#aT0_j^=A5Buy6-8q-z9rS2sDk0VBv1iqH<
zE%PoiAisH;S1x{@9uBvk`ypar9ShmmAfBxN<_C2gr5Evp9FbX^hC(&DNkag^zF-UC
zJf~JkfCA(Ky0hXNiq!^|J7uSAx^f3Q<L{aFcpq&(OvTIF<Dzt))L<w1!t&GTiGX7|
zyXuSm15M??y|Y&t(BgDCK6W<M)zFkEL5csps`4AZcz$b^x3t`RqDrbDLf+E)@@d}S
zXvJr_Uwga)4zKI#&iqrLd#y&sjZtXl3?HRpl9x%A@15r^H-&8?BX<+D`d3I)Ce*wo
zn&oenZqM%){s)iv<S<&U-C1APa{9eqn=~xu9o=?Ihxbk&)9pA@QGBAz*)p-+B7{aZ
z;HqJFOr8dNV9Gyx6OnnSK|qTtZn(euIAHUm=f-I8(cl`Z^1;EW1lHC0cB}Aq%g^p=
zsMm8bQ;A!iG`Z$fi4_lXs!z4<&rY&a_wP5wBI+c$Un2iKMnqHdZ)27O+nSf)gxar0
z+kXE7?Q%UN&u~g)o6lY!K!%zSev;z7AcoDgF#>;dplA6l&kRF6SgAxg<%B#rbY8)q
z_HJ%$xm?fvbWt^`NrgWC_=x=Sc$gdZweM0X#hWOid&FdV{q0R%#gfELBW@iSZJ3%#
zxgE~8CKgIOczx!v_hcnm;mEsM-w|2YmX+v>-S9NiRI)LlMNE>&2L0*j-x%XNX$7<z
z(!g@e)Nr9lFAZzpMkvbKY}zPCwKS`_MpiGjd~yr0zm0Gk*8VD!N5&lHPnH+@{fR~A
z=Z`NwPTKW~um6v#7n*Iit`lr827s#yZ(2KV7*y@k;LBx<db9rL-{<(-P7<{F&791V
z6u`+63`)holLN3{qt>|1XO@;w04}$lCs*Gf1wLo8f>9(!4Hu0b=f&(NK(FczQH`CQ
zPvEMYX_lNclKbatvqNfsKeyC9G#%IS;ds%L5ta>Rp{}oYb{q9y64w0dk^wsq2Zck@
z>`S9!x(zX|&I7IaF6AqxCYh5hSt)R*d<vPmr^ZDPJ_0#dslF};Z)uW(V`YP`#JI6a
za;XGDg1{Ap#z%r!dZ2BXTi}JUQdh$unxhpAHkjN`=#y-1MOFd%)97!)!x#h{&dlae
z=3SKb0^EDMzkfM`jOqb@&9cvC*HsfXW)V`RKzy$eL?Q_kz#wK-T`swI<T!CA1t&#=
z)|i}z`^-UIRcD$EVU$_1_8Ia(Blo)XZguVsb*O%JcfHfO%l|#Gd~!MX@7~$jdPAPz
z@zD@zc$WFg*+&U)oG8cb(djIJAPu4mPEn)!koG*Y_+|lZk-HgW{daLpOwY#1N22}a
zCi$FF2l9Gxs}hN_rDx&wqL;rIz~Q>?A>KmM1XQ3iyd%E8*y(B4Po01NWU4ZQU@WAl
z(o`fDTC+&As3GG>h-kOwlB~2r1gSv!Uubf%=hHqP!Px;SgMHVxhaJ`1HKLuOoWwU_
zgzq#1kN*(f7a8$P&Vs-+sjXjxhqC_r@zdMPAa+8Hi%DKB0j`n+fwBmZm1<f=sMUp1
zrAqDbv)@FuwKg@2zbISU;>~HG$g)odq=4@U#!colC@p+Z6WJ!CNB31@;Vtc2V0rGT
zS)9Fe{ew~?A{Di4+-zOmyAPih|DBvI<j%DQE?q<%*cW3|oOb|IjD@GuP4h;fB6+ix
z&wg!m8QsA_LL8^(=Z^Of%s9n7sz{cxK>B9!&sSXp@5V;v`o?<jUvD$z^|~rlWJ>(M
z!zgWMhV}K#_e`k2z*BbVG9N`XQ+T(TS-g;-Ra6RbNa1*T6-@<#M9?Te_`dW1^p<A)
z&d<kp5@2@l2-E}3X`zHm_!C2`>e3WMHB;fSo_wlQa28lVdjTYz)d`*3NJzk$MpIW?
z)rYH0u{UuU)`;_6yuP5g{f!FjZFp&7)O<a75utvZ;_%}@)R`Ad#hLGGqf-()sMj2_
zBZ0LsiFaLU3fLQ3Lyw#Sg??l3#zVtNby^uy)0a*49qv2VCX+i%K21huS*#y@8-0((
zszg4S-~E%8T=-_;?drGrqnje+LW^Z^E+efi;3goaY@u!PmyaN4!#D~?q;NEf)s%dk
zkd9NU{2vVxkWXIj@7K!3z`Tx3RlhWU$@Z22GVERCjOilvN?1=BrP2E9=8b~DwQ^ZX
zu(KI{s1A#I>cIDS!&=kUyW16SgJ-lg-p>&3quKC_QDE{6SZ#n~r}gva&sgmA*7^T7
zYK_s3vYhXtwPjFsO<v27XSgXrMvy2d$$#N^<350lV;0GN>Sv<PZ*$^3Y*Cs*01*S<
z)3q$mB`u!p6E*PP!FkipyS?Vz@vaxPdRBl?)v%V=rIy!uuXqA4)sP%OuuA8@FE3W0
z&W2C18ya&zT|vHGzWKiO2N0T6{)jZ4#-u4Y?J0t66;K}uI4K2Lhubyz%P!^}^~&I!
zz_>l*MJkV8J>Tafgwla2_;D=ofz`Ka>H(NVzqtW&5<GY2&}>e)7&n;Ui`^Jw7$pd-
zswVDP5-Nwp%UOHEEZ~He78EE^ZE0>be8^UZjGTE(0TRlYkTwF~kiX3HI-5Hde?2tS
zgj>d=hPj7rj7S9|H`@!F@njen*+>B=#4_Vms6yK07iUsyhP;(fnak|VE^z?cagMI3
z&6o0@FqTxOR5_|m>Jzmw!;R|rewJ{Yzs%2_Gtralib>%haH<P(f4%j0cE0m+{QO1c
z{U@sI52}!=l_hHFTK0|p>f#X+Bd_>u`^4a=obD<GujB{|f(>R|@1oMsH%s3!kXGNM
zP4Y?ReP2o2)~p-|$lAa-89sXTlfy&{(b8B+g4pv?iP^FiM&|<Y3|IgYi4>Sh!7K3Q
zqB+8O$JC-!kC~--(E2+UQ8N323zi#$2V#R-GR$_(*X2jr`?5ci!Tu1|+_{<mLxoZ^
zn@-1QsmUu~oU5txKI*?z`)Hj*5BUppe{R1e-Uhk2Y=-Z81*la9%(AOqp3S-5uDb@a
zjx@NMVdwnUoB4we4sPFKR7<y4FGj8T$BvtC|4`g5B;Li{T^9NVET9+KyNW7|Q++~h
ztfAVZ%)f(T5Vl5*ZC-zW;ZwZtr$$-*+)LOAFT$pl{$_6CpvH^Q)y-M6F*2f!l?GL~
zVDsW(|DvMSvb$99^6Qh^AI{E+aBxDX&`F<Dy{oEnD8V&x5x0!WZSFTO$|r$Ehh;yh
z5y8x|?nX2Q4d*Xy_1>B&XDT#bY7?SQJo6cTW#?vZ)0^m%qfayb0XPQnQA`gr9x6Zk
z?L0n49X79e(xc_uC?b2anhMwKxc);X>ay4obY^Bq8_{jC(&yoOku^+<^=bSR>~VYD
zS{@kZMCKG7axqM1ua5q?zi(Y`dbgfx-r}~O$J?*@rgW4biQUK9EdBKQ&pl$IU8I_7
zcuq@Ym7~a?E5L!UlWzEG3+2=L3_<9M!PVMumQv}bVZ49;=#CJrHLK3GEgR|^0D@kA
zC&{Uw+Kx!^>|#G|=2kftQ~PUi{r(Y-U9;aA9Lt}dS(Nl#7cLU>Slx^R)RS@MZLrAA
z(|~6>HMO-a`+L?7`^G=G!dmFbiDJ$xloM4#<p?6_=}dO$BJ@36QLY<ZUvpf~g0fU7
z@xHjuLW5I$2q>d1s5HBDdxfc~yS*nfdQ~_H<#L2|$zcKnrK&2n77QkGN@-n1=pCs8
zhg1>oN81l6Q{jqK&%{YO<UwZ{3<cR4%wKEtVMV$UO6y>?ZD?^C5}M>}>lK<6lf;o#
z$l&u<U1HB3U8)O0b%jETkli>zxfq;LXMMXk-QKo6g5?<wPweG%K!z^lsAiv<3LHA-
zC;?)jERus45NhEIhU!i{D*-PWk*(%Y@b%677dmq$d*$KU^Q*L!ZE~;nk%Sc8!E1|+
zT|lEv{ub-9P*9pi0%M$;8Kb1^f#yF;?OU$4l*o^SnpC3NU@_#2#DbPUpw->oU5t@k
zl?vu}Fzh(-^FI<xFoCL9I@+X6_+Odqw7KLzq&-vH(bh%ov}+HPg_zCZ_3MAf)33Zo
zrEgAzzyu&9m!^cS6Z$FT$gCKk9Sk@6la+>CjX2*WnIhNa$mV=Ic&o+#bPnt4_&Rd1
z<)$~ZsuT6up8U)^+)7-wb9+85YKXeoePi9vt~HbWV{Oo1)y_i4nMY?b)u-)Y7!fzO
z(I`k;R9S^$q!y}6fioXTYRm>2fH|tFyzWy^>|kRdc=*#;wmD>_jFE@U|H-}aN93ir
zz|=3QDX)B!Y&kzvR7rr7*|Y!q9=d2E_vX#hyd4S@(M?S`pfGqnXLL!FZUe|f_W*lk
z?<NeC67)AY)!b>0>E=ZF2|&lW$~SLX?mXJ{)Rw*N1x9n%u9}nVs)G4HZH%N(ez3+o
zr!AhEN*@RTN*o6ON}4Ep^6TsT_IUC6r$-V%lHw+H`p(_iy+DlUAnFH<pvS_QnZ=fZ
z5mi;Sbxw?vV1Xq7mfU#Cca}l_I`CtbrN2Wy3VnbIu`b`<hG6INzj!IF0Jzd&?l|sh
z^}-O3t)q1ACy)JO+Fh$i0^=47g6~dn2|FEmmVYH%C0Ax2iJuL`jmccmqHaH#i9amY
zs+Bp3Jl6tG2B18xs;9@UzIE?1*EL9CQ6X1Dz<$Sh{~GFO@6-PN{$jg3+D?D?=yFP3
zJ9w+Nx71(zvg=E9$l=kFJ?-p5F2joCaJBOJ`QbOR!kop4Y1BE;-<N3@-hF>|<`7~Z
zDWb-vE9R8_vz(S+`ecAr6006>Dw^OsDp6j4(h3Nb3-2xp$H-OcYlix83vK17hP)a4
zxjxk2?itzZZ>URJ)FG|SI4H7T*4I%W>#E-{1Tbku9LETViq=e83qF1@SvLo4@`(up
za?O_D9b9GsYA|4WP*DEw*_Nm;7Au)Qx_h7YS+cLhBJumxh)JNCO5~PERF?J-C_*`t
zC6tJc6h4ShId?t3lECe~j6+#8oakU1qXq^TEol@a-o&Tm5V+T-^Hf)5#W!e`aYVu^
zK?IGp<6yW~%!dYeQe=-rBxC|U&cOg*Phyu4w8B#7AtN^x-R?tUy8_}u;Upj{Bqv;8
zjM)?*;k=3Ah>Y1a6@S}A6+zbB4`roRrcrwf2_|1uK4^9ZUX_+8V>fbyFw}%iFOoU(
zANf2qIg&I~1y1mcfS|;Div2yM?2h>IZq55p1?GARa_2amE%+coGj0SdDip#bAh#Om
zSwxZbL1P~n{;f|9PqsXSrs+xsN^%oZ+;e(fRk#x%MVMrt_0hYc1Vo0zKBl}w;TG$S
z#&#qoO`ZCmv>l>WBfKEsa5#uPnMPHIijDhuW4JTM<VCf#<J?+7Wo<!D&R^m(M;YxY
z{juomTKdGY{8jj1hK*`Q7CihO0Th(PEa-M$>%MEy*Y{2`ZE3f1#}|aI^MUBVg`F0J
zQm=o!m`BIV8{S7|--W3N;mRS5`Eijl5JReXs~*j@CM^^mhMXb5wE}7*Pn&-De8KbJ
zaBKg2Z(L@TqgzlVP+I~pF~08_V}(I~E8^ZB(`0v<6P{d4)F^Y2#NU%c_BF>Nl*m=)
ziSZ6tji?>FEy*jM&q*|tr_K^00=p`C(kG~#Ru|k~`#Xu~YivDiuIp^3+wlk}69RW&
zXI7TI>`-b?%gc_prgo;ky{W3R6{inHgTdVE|B>Xs2CcqQvCAgv1I9oPii;q0iYeq!
z5K?qVPX>Cr)avoTJHww}VU1Ov&1p!BRkN@3G`rdf*++-pWuN!uap-I5BjcxSWF;#=
z`uA1mnTvP;>jU%Z;dd;ky9r{?V?8507LVc2F(}qS2D_z>fWJQ`Cbea6W-q3>SG1*X
zRm$5x1N$VUaR=0|%jaH61U*-Qk`q-sVar<*>HwBp)ks(=E0?pd#6pxJahYw|-PzTW
zb9q%~J9ThYVeWD5sF!0tv6pHY+w^D(GTVkuurGanW?t%ZQxb=P<lTOhI)&`@7YO%C
z4Ju7{yQFppKAh~pmwwuPxeT;oSJtf(r7blm!=-{w{#FRf-ejWuug6m-az#Tfu4jjw
z>5}vC?MPSNy*uO0_dBB2Za<7;DwL4Ib|lyVX%os1R_6sNf^>$XZcA?}epht1ep#uu
zbiJyrCi_^w(Q#RNI)X+^PNxXy#vj?T0#Nv$7ru+)Y$^EB50`>YG1@nM8!~=ZjB%_F
ztKD6%&bIHDE_}MKUIM6J(!z72xg+pa*OxR=#W`HvS&%lK{yiGkwmaNL!PSI`-Pq}+
z;;4VU&}r0$%$*(T(9W2P^p*M)V9isT@^)5;7f~6cc}9)a`saIl-Fba0db|7yb$boe
z3{hk4cZ*TL#Z*pG{*vrt3<sWi9g)EAK*^Qvf*%sY$3R?1Y%{<8-`OGS`KN^rPhbFr
z46qiE50u!*8Cm~OK^ft`R>3I5tK$~-hxIGX-L-imRz%EBt)!b6CZOq7mh0{+^td}S
z#z%pKVlg?9@d8=J{!T^5ja?kIlE>LgXNBk)cY70WL*9@B4mI7$drtYfh#IMZq2fvF
zWfqHzLHrywz)kh(cN=lCv5>)7V1zYj5*kWzT7^nsRl&*zOt+X_%{gocZe_|!o9D?c
zY-n7jS*r`f=vIx}t5t#pkTJ=5-jc2D)rPe-^Udx4=Vg!f8gy%+s;Kw69soAP1I8ZN
ztSW`JzypC}5Pf2~){Lwx$rid_SVI$+@sTt=27C6NzG9C}!ea*so}4n^napGVwVL07
zg)a-QugTrwkxtdzwN-RU-lP7qLXw3WdpyGoM!22-{-@;Zdk&73qlxgvNe(9yx~A)@
zt4C`U1j(tr)^yN7@e)h}U>?oz$JsytVn7+WD)?K&-#i-1AxM*>y2nButr%mF17ODD
zOiZd#SE4Aa0QT-<^kZ@F%g3ChY7G!wwcZkDf^_@J^yevv_c0)F5BtzV4{PSOa~dgm
zf!?m4Yy_R{%`(!`3GuY)4qgCq$Alvmh*^K}j-`Qo97nk%S5F&e<Ywc-6O~u0s^}Bg
z+5K_*><N7pJSGc#w0I;eQHO4(tuI@D?P^6VvruxSt7fE9jJ81AUZkjJs35lY93NsV
zZ@Nd<C>C8CW;sgKg!0#8wH%#FCsHHlT540_2FCtOB@+tvOqMY)D$eu<<ArexFy&~6
zdtcO1DJm27rRkt+e)oFmZh@lUX7|b2f%%76v)IFyhP%J*OhLc|spax^%6BR7uIo2G
z6Y6NtlufOo@On~vUp&AqB;;%;oubVb@z%1T9D`?}z)?E6RoQ%_uDn6}a2DMM^bk?C
zJC^3yVUXnB=qZC}3QuK%1F+vF&bbCahOVBYXrd+@0|Q&f?HPZG9^_LX<H*PsiP6-N
z1JK-tw?NaQE;U}7b1hrQ?y;2rtWoB=_;i)*V;f`b(tqpR{2P@LsC`CcpYww}I)l^2
zrft5*0cVqiEq?nC2>H@$7F-%kM$SK(Ry&GHOPkSryIsr>`isk+u0~IZ(VzM#PanKN
zg<OhuS`wMKH+}q}2j`U{5%K-|m+5*Ol?%@Y*ttt0!k?5)&hw>Yu%8>+#5+8W->cLC
zPY09So$C5Yz1EwsBtcanxp;^b7%Wgp8Ox!WGMp*v2ds(CEbNVFXg`wiTmD|%$@~t;
zdBn@P*ZEF3iiG<b7-hBO)R4dLvsyv@S#-Huy`eDmI}i8Oa$_gdrb_6Ff>6(-B5;nP
zRwYF|)kFFQQ`=tawC4skGmgTSfszCzrT1Ykovh+-LJHNRVt~dt?rQYs39uZ-l6;|z
z|5T_?!%^!1-zwfp<0+Az(|g8Zi?H6Vd+Zlq<G(CRLj>8y)Cl_M^wdhTamFZ7dZX&>
zwQLXXFZG!5gHsJ>0Bn|8iT_L|vzCwR9UlR0TjkCWy;VMTH2s-QhLzp}Z=)%Lrw9Qf
zS2`=z9CrVVvi~X@j=$7MlXgcbz?wC(RPbNJL2@hGRnTnrB^p*cuA)r<x+%ug*Xw8H
z<mqo?6KH3*w!d(`*xJziB^hO3DaQJa`VWjH#{7mD3~rX|BMH;?*^&uj&z0G1&(+`2
z1rPccl`ek|G%-lWK7M}9^_ZH45Xl_Fl#Ha{Xi7Ue8_A7mf1UFb{~n&=a&GS2q@Iw8
z34jFwF!KRN9%28iyIVNk4a(&BdS}>B$oK`^-QFO3#I4?Zz0pbCAxDK8ct72;VBeQe
zu*kz##H(ARGoJ(F+zDPh42-_7d9YV|JF;}GS%5WO7%i7M9vs{a_|p)Z3~ElwapyGn
z8pU#*^Z=PU%JXM+6UYvO<QC7cBKei2TB%A``?y!L*#$Gc&+Q+CbH9}?f{3e^m!B_r
z7Lm4{1-ai*W(x;?n?SYv$j`ygW}Q^Utvle1U%t^A+g38M<*oHhi*>Z#yQmCm`ADxT
zD*R;=T{%RlIQCNRZ+tjs0B3z()$RUl`f%ouhZH@<P@P;!6&KKM)3Vl~q5kYwuz^t8
zxQ1)ZW&xoO;VE2Jd;3BQUnvsw?%q%leFBiz*xldf`!p{Rf*Sh*!BZTCf`8KTn_5eW
z9QVx~m`(`hJ8JQd^23|2Po5NNg<C+*eME{yD<5+Ey6n$kWs!313DJ`j(jt<m&olNT
zW15}U6WR!z=R(<sU(O;$SmOJ;WaAXdH@`nK;r&O<$^yU^RaU=WYoLBD%`Fb40fEwn
zQK+7=N&lues^4>Ckwt+;xX+*INp%(a#m??b{S)m1#{Y<61S>&08Ji>NalpPNziiFr
zoZTzG-`LJe%>GFYMTf9YwXBB8!dUQc&uRY7T%ke_%hi`||BWz-xsUB%E{MMJ-}-tX
z>S7*z2e2OsFaOYP+}~sJtJg_0|F_^;gGTSow`ccKp|(44g07p*&DlUg^xWs{Y|rF+
zz4Nj!pMuYIzJ2__H4$Ix=3+@_AsASlJXmP*XwI%$Kx^_n$P5sGf=4~jbk9sOd+i}@
z1}O}BIaSxyotHZoH^3nDd@*1rNU?lGn4}LN09LaSj&?~Fr`u!c7qf3ITmzjZp_4Ay
zvT{^;;^TH@&b7S4pJk&_gW3Ts`lPU!ikUZB%Q^l9c1C=mbx0QCmFp%iPfs%7+ihcn
zom)6;LOvXFMACqmiIA(XrbMNd!$nOz0sKTf9qs}yi*EJz5IljL&lPyd%Ph|ywIwPb
zt(ny5Xy}x?9|(*_dGHC?V*b4UXD@6isDm=LG}tT|AIOb=W|Elqbkvw7EFOPSf#mJP
zlmR>%IUdMYg1~sPYo_uHC#fR$^RRvNR;AB@D;|!ogv1<ECda@8*n0ea0mE_^n=to}
z`ry@|T?*xIhxT^9!}_OalVRV?%yS=K`j@XW1XmG62kkTiY`NAm8bn#TLz~}LEi4Ne
zb2?wN2ZQrq_inetV;GY;2$(FF;{otn93`bb%afOHSn|lN?{=8<=k-r#499#-oMD({
z&Tz$}GLgXc+pk}5yDj{VR`F^1I~Z66`VavJ3qACV`T02oEx`CXHgmuZ$h8+-X<cs?
z-rZhom=C3DE!v}Q<ID$gNvvM~*%>&_1B<7svE>azW%JDNtzbP+lCWo>zc8^eaf-)z
zo_Pvh>G*1s^Ur?y{^Lj`F(fsOg)sjdrDyk>&9V&DW2uZh;zKSMyO&NhD%xxEdzEh_
z4Z0)VQR?HY=V_nydS?hnXrX?7&a%$3IC)DV*hdNjg{R?h7z^Q<U})xyf$-M$@96PX
zdTTt8S}c#VkabYXV<0PayhsMi9a*oaF&<%e*GYV5Ta9=;ZJ^`Uy%=;n?`Jy7+nBq6
zi>2=WWZKbw*JYWa*5@Y0SEr$;fx0*Xl5VB^HMKQ0E;Uq<45BV4sK(1ZpWN_a?*h93
zJG*{Kz4R5}>wbC_m^a)uF4N={Bvt$HjGs|Dz{~q|eS{&?PGX^VhVG#ixTETw)L2BN
zcuzt?!q!&ypvE(Y*Lhz8%dptMgV>e!_E)}D&KjlqU&RcKUpmR2ZLS!~G&MFfq?J`7
z-jbax0z%34Utg?ZP`~mAxt&-!CAk`1fX4`KW-Zt7SgBZ2=scqH>T06{h;GFl%U&L8
zo0tn{JN{oNV3Mr!B6n;d;EvZN8V@hGp)XI}n(Ft@U}c|xwX^)ahGdz;AA=mr0O#`-
zw{)}2Jtl)~sngx9c}jO*312CR^o>Z_`WZi)gC!b8_g~;5%~;T%LBI%iv+J~RrL%DN
zcQe*b=JY6z-`kVU^2Z3>hnEc87)x;I5Bt?;_q@$atZESP2;Sb4`3@gDuHhUmKg0A?
z9tIczan|+Plhghl7kwH+!?u}yxM~2<waFgc?Fc5u86GH^_k;V={BZ=KvGpJj1o0LG
z!l7pOf9h9{GoD=TDdzhwwfh})6Z3Zj9_}w-sGsbco0!ZkE;?gdft%dF03Sp-Ix<31
zTf5YBOWEuKCZN{k`sl~<vdT)&^RXjQJj`uWTv0&@K}E0%ts#-eQ`-r4drYA8PS<_}
zv;0bcwTbx;-_hRU=Kj$3?lt*J)>A<~RTPQQgATsoWpq-4ZGCAf0ZDlEyK&|$!IX2@
zPwxcX)seSaX*fD0V?taNk77KsqNFO{hJ#>3Mu|<u!9@dP5nA+^e76b7M|<{2(7+(g
zMU2Dc6VFmxaH_h@E=52)prlrE^~FkO!jgSoICLlVLh0Pz-4l2#haei2^C0g5qu43?
z?!`1{f#uw6t>PEe-=8VwXvQPlMisyq1a25gDGp}yt|Gdj^dMr=$zX=Hoy|H$pD{@R
z{p4Qt%n6?AVM@X7Q1qgi$;)4z*jB8U-<_Ve?BSUSZ55ZF9f%{cOWgHoZZrk|RN0?!
zaV9z|0v+&Rb_}c!jL>2sB><r~O7jF#jdLxLd7y<R>&i;s{bswe&Hrvc-JPI-K%TUE
zOZ~CBw$$P3?_%YoX!GGi7y(+(lAMm%nm~m*shom_B$-;CygRi#RCjw2ESaSs$PqVw
zV#+Ew^Tu0>67ksT$2&##bgCX9D$}BY3ET38r!wm!J~lNEO*sm*#|Tk>+D8^nF;h~h
zM~gf~>PIh@rT#%9#=MHyu(;P?8vHPefi}@pO16zO^)~m@qUt^=_18z$O#eNk7}=D{
z*4Pf^c&WakKuC(<N*6ou_4SR?3TT%BkFpWRt5v#)K05X3yxDWr_LZ<(3h3Z?8(;6~
z+Ho;67LvhCzk_k~9DRW@6!F3O0zs2i?U3u^I(q4w?|M5XbFKx_079m)E!fkh8r$B{
zK?*{@P;?S!JiGU$%E=<Y-)3z*Q95`jnBvXuy#MWntXWpBC8CCJDcdNEJ1zww^7y`9
zDZ@gg7-6vvVJL`4bMm#p!T$bZDZeh0Ne9vX72u>pB($l@fl=UOet?Yd>613_qp3NC
zL!;4bswo@(De$t#w$1ea1u|e=TueLC#C^TC{_dr(wR1Cl|2}s2mv~9?VznOF!pG|N
zzR3JUDimoC6$Bw+rMlrl?}(c6@?}F#76Z*CeP7KTy3O4jkD2@Lj`u4%mm|SkFB$sZ
z(#M7`L#(S&GZ_xXQ^k&=G;1!!GuL~cf>t_%&ReT;#knN%Spy|a^k%+(z3xDi)IbyC
z6dbxs;#cYFDl02h|NSiSeIuH~r}XD~^Kiv*DI`LHM#HbDrHXwtfP68Kzf>cghjgbs
zGo{pl-tzh=6DWknP#@DJfRQ75hY8HL7F}kL$9RHL0P`iq{UZ8X@UCc63jjtndEMq?
z-#>1*v$H$@t<2Qv^PBgb7?j*He!)c)d5~wImOZ59ZRg|TL!Xey#h5n`9SU<K=nk{e
zD9z~RCJLWK+T14qa8qnX`;__UC`)jcc;fgsxC9*?W$$LG1p$Z>qZ=4_gPGFggm2(S
zZlX$$x-}zbd7LWYB;yJv7~WyW6BsMCU$s!K2!rWgXasuLQcSqd4aUB9)&v!2C{as_
zRf&mNoZaMv80&e!<6VS*EzB$mz?IXNUe0Zw0vEb(V<T3>Y^l9<D8{cL)`e_TjH(yZ
z750)`UI2+l1hI;FOc$X#DDGfx86Ojh8i)ahiS<M9>?+be!#H{w1y|no88hFDr|y9$
z*3U2|C{eGBC?D<CyxFgDMx8pDdyX%>?6A>PRd9>vd2V)psnzSXyZi6mX0pP|6QM%X
zesle&_Qnt_aPu<R@I73&YsVOKl0GJUVv#IHB0ofEI6YcKJSr}D)ORrp0<mCG2LC<B
zCx}Jp{QmdltOH;9Zkzx1#P05TCE(3*kQUoT|Kqz~)Y>xd=BK^$8|S(nd}L;i)B<Nc
z<!j1<lL!!mLhmKxTb9I{uH+{Uz>q6_5%Qh=ket0UH`7NiB?i*1#|cex>Yh*&cF(kw
zmssjlARV%2a`+j=1h-vBnBs~TP?%<ya^Zc9?~-YAfR*RPzU$rD0j`YwT~8RaEa5Gh
zi`T~JKe}%amW^<-2WI3lX}w9|CZ>5IcVA7W+_2-0>3vZ#-do20Sf6CZ=G&M`uf~}S
z4Hb0o3HrtGOI9El(GQ||Ed}){D{G^mLFhPo3&&dniiX2kCt21HQ<B4|YG2nS&y!q6
zv6W4;dpfsA-;aqi9x8korsX4jRCv4EuN*ja-f>4!c(Dr9*jsjcH_m#A=+l(A=Yb?e
z!moWCJAcwQ`>mXsGfD{@uXD>vN(o6$I^ruldQ+h;xlc{GftSS?AKmk^s@Gos7n0_!
ztFw!?D7{3|@k7siPP6HUD9f5lMKpC@MKMjev1VW7=^KA%>7Ogq%J(1Kpnusj;EC=w
zFWMfA+Pv>Of`prAIZzN;20G(gcU;sL2CX^W+3sz#J-wDI=cOi;vj)PVqa_wG<dH+*
zaEm{KiL&nd*Wy!o7uWTLkEU!r?Xc<Ao>!T;Ojp7k9)b*&Hx)s)s7B#<cif!uz3%r?
zeq5IW4B|}{YP36ld^=8`_`Z1Y;?jUynf=3z3&BAwi@(R>uPOyR7IMlXTfOo@ulnYY
z<D;Xa1?9&fe%{_TV!Qp&Qs(%3;!4)W9Cqy%x+Lfn4*@PCUDOdZ0Cu=|P=}vxSM5uD
zUMS(e)m}YntVBplfT9t%KUxm|txFyl-1o9|M#g9L7ulFB|C2YtY>Ulq`@f?PT3g#l
zSmZr=IEVo2-<wYZZCfuZ6r1d}786xkNFLHL7uDnNt2Bu8M0INk1Rj!DkteHDg2F;I
z;fAF2FlG|Y{9LhNR>AH_<dYzQJd${MkBrh5!zTAF5Lqc4ABEf3xUk<YibqJ;)$qPS
za3}>E&LE%Usi!~HUrbZoui@=4746I+7NjH*OU<E7K9q^vj7>t^kEa&UGdYmXtb@K+
zcofhsouRqVglVbA?6k2mIXk?_IDmwG5t)v+{AuiITas6zBdG8J9Lh}K%ZGOlhGe-;
z4<`r75cGwTp*Sgn)>ti=nJuUv-v^PtH7auKcP7wYj(P3@D|z}@h9z)q2AJ>pZ{ONV
z5{Kniu3sAer@`XCG30~Aht>wUt$kE}d^M<T9<q6YkNS6TySj%RYybO|?3s9uS(a7a
z735h=YBoMItMx$Bgz5w5P_dc2_Xypd*zO-!p(=TL)u^eQLRp$3)cMz)yRADv2diK-
z*L4_vbbsm1*UW;eUse>9QBZumR4&+Pk|!VBgD*QLzB42p(UU%rnt1YU%7`7`bCK>|
zNGtYt{FYa%8U_zPT{s&uo_N)oJ1<$aYab}-ffres9SkkrE#cD6b<`@Us!__{YHrU*
zW0!(aD5o~*r?RQ)PdWOe1g*&F`opyA6rBu(x<gIP#5tMY&YIAlDyjXgd=YYZy(7us
zq~@7Km9h>fpK>s%ZQZ4|8usZaKW&_i(LKia8c-AJ?$!1ZqgNRQ<qPu5ARSfC5-$WE
zt2IQ*+g1YbNXhc?e77|cO#TT`Wd_8q*t<<s;w{jG;Y;jW*|+p{dHs4R%`3gtYwIg!
z$M7!t?of8o?|ftqyZ?QRf8O|&R?5k$Vxn86I4yDJeZ=5h;oZTVvucNzg}F}pKxJJi
zGAq($8Y6E#KqzP%?|?C(ya3pWcI76zo9;RSVb;xZ?2(|$=cZa|y8huXf%_VRQyA2m
zQgO2HT%SK@Ddu#2yAIwNaN|+&c2R622+mDuEAgL|>F+;ZdaYev`#3e|&vd+tiop<f
zeTafEBJfC%Fq>J&!G<qFP~cdR@%raiP%cCk7dR3bSnt2lr_}jIRKi**TR%E+QBo2M
zDQ3*@HlU3o6MFaCMaNjdYGq5ADG;?4Rk&??KBygZi915;m@$};jD~Rp5F<QbAVPT%
z=-``$m>kSR<NlRN<(~s{KPN|@{lyjFL|p78F~Bpe$RF{->(Ypc`l7o^e4nSF>F#H)
z>vf9G8_|<dT!z*WAld;IZBhXiY)=vuwC~9(9%+FHPNYPCPid#;=DhO^watzDi8D4a
z)LR&}5b*n-kdjf>fcEs%$uR!u-spFGF<jN*iO?AEmrpwI)xFJ7!ry+yykcwPQ!}I+
zghWs?hDx!pSRqodB1(UB1u9F`qJAKHAZC#<c^4`)co3p@8bbr1?B)30Wwv7BHOc`3
zg0k?$F1FOzV+4|{`!i~}Ysy+Z>W2#^r3N<fRz)|Bg}|=r>zsrHF@nWy#gci{CS#Si
zvR=g^o(y&wgR_c(++(p?slquZmM0X|^cNq#a&>)4ety}$w;=sqVlvjZFR;2qMTiio
zrq+D{=NCpEg@vHUg*d@<f?@H`K7d!P1aOg?_vFE`1FZ6LN~i&-U^i4T%Td%;(unnS
zEb^e;ySq8d_Ci^ul`g)zxfdUEDk&KlM@Ed%!VJYeSL?uFWatkh4&D`+&iii8p0?iY
z+?{o%+F$?fyaNCoxu~GcUt3)aVNaM-1_-{Wz5?kIAb;wzB%hD|YN>Lfm3sX{bt2ob
zFl4hMwOpGj5>+UDSATaUd$-Yfxqn`}WGb(J1-uMZ&zNpkP+}L$s`+JIbTHS3R)^wB
z{UWcSDCF74FS=@jGNyD-Hs(KR`y)#Rx`_?%$BpwTRzvxfmEix6rt^%a^8f$%A)9iH
zIHEYR35T*roe+nTl|8b{-rFT3>&Py$ijZCQ9$$MM$;vvmW1l!i_W%0b_@BFOIgjhQ
z&gXi+U$5tLc>46Iw&V>s2N~W;I@jNTR&sl|Y2x_oc-~=U$qb^I6e~1gs55|0^uKNq
z)C2yOrAv=A^f7#_T!b4Wh*@b6xu)8t76tz8NY;*_JFDb7_EtXJRRT+jl|KV}@8}rL
z%-uC@t8u)~?e~iD5w3D#i%SiK;E`dWAxrN2Q29M9UPdj-@*4-YK8mMkIrG$_tZc9C
z`rD(u|02KvdTeHNrI5n@P%>tkON~w%(gs{~Oz858kBQ?J6wCt%-MH31sX@ZU;pW)F
zDaPkp3i++O0?$@n6mAD;{N7ny>Wui=)VZ{@@(hHoz~Up-a%Hmkg}GJDbnh!E{;U?r
z=dm~}F6P<swc4?ld;7>3shYmgC&iINVp(nUKxte?ZOqfh#V#mlB7h-}(2AWi+6Er7
zl04z{%a&xs-8}RWX?w1`@F8+F^Km(gstPoWcCxV;-*-kl_qWJSVbg22M2`3`%$FjO
z7;y`Nl`iwj8#S1#!;kjn)cfxLcj{3Ae%Pe_b#3Ar@)^{8M!&3zZEHL@eE)TU6+!Yv
zRK6)et);2eZ+&Qbw`Ezd?c^M|1X|d4oSZ$4<J3QiP-Q)|EAUF<$o2SOUl}WI2rVLh
zQZm-hT80_AFm`D(O;xVe<M|0Y0=-xVJ5?hRcW&Z{liwE4FQM*#szLOThWf8>YMQqB
zHwZ`CJc7}j8%{#el7z$G`P7lL>~B|}<I_$c5bP8PE~0R|?#Ktr?WPStpx?rO|0Uj!
zWchFX><oL%^mOB)%v8Bac23UKfj%&JCUxf(V$s9=;&}|(<M#c@n$z!01&>jNdAHeM
zq-CCey?zg**kbjOVl=R)ixie1+`y*nD~!^RixEk}4vD%SLu=*!I;j~$>hQ>8qK>gY
z;2{qehG|91G^HFy==!P;z=*Aq_zRNQr0q<69p?WiWAc3Zq{osHa*t{rX(NgnPOMAg
zL(tR~p)l8X7@b|Tk9DBXz)YTp)XTVH;o?P8ty^!!%iq^%b3XS&A9WdVK>KA!z7cmw
z7s5D|g%glNzsJe0mjr}JIle%N?A=XzVbdUeBvfgg)9nMjXd)Vbk*`vWo(w9v^)z^R
zHI4E2lCH21&j%iK6u#1bHYgZ?l`UJF+j;l4xytw^p9;=I?^WiVEX-k0;XNbMhNhN=
zeV>JE7CJV?V0_zU9RRjA5?khv)R{+aof*T8ls<5TxN&py$ow<nMe-V<q!CNwHHN09
zL09?AQ~&!&$diY)o_y2waC=R~q2Ri;DxTwuvXLRat=CUVRZfwnYlP~KAQ7gR?h2Pd
zY(3AWH1c*Q0;f=jgCl#|ov2UO0bo#G#&9{5Vf695_~j#oDn(*D)hzD-zoX-W1C+=0
zP`waix4W)ftxL|cWxoN^&yA~Qt`5dhdU5BBE!vx36r(m6$tk7g=9~R4QCP<l6&shP
zC^U1y=j?1PwVo;=DCBF=P0fcZLTX8taxV^S^G3tXk5a@8aPs$AIK+Cw7vU-Z{G0Y!
z7;Sm=oy`|62Gjc>V}Yd;q!qr@j{0PF*3F<97-$!8eteKsIxFQpBA@Buu#>E*aC#g;
zFh9+zIJ4lMvN7FK64}c!lXG=-(cQ_&m_j)_mX=q{^*_|qE(4prrjw3HPnf-B1Q&5W
zqdp8$J-O@c1G1EEkS0rf4g;PjzHg_z`FX!yS1dhH7UO3Zl=C`Ua<I4+_h-LpunrQT
z8%^-8uM1kcRM)+&tIx*Ctg4jR54*vay&m_uA*@irZ*9(SYa!U?ba+fN__dEiU7LT_
zvjm#xP!FQ$`v0K+f!Uv$*}*!zGw4&t*)lVk`|fvt3|Fa=X8QwTaDWSNnm;A^%5P+(
zW(C~X%LeZ}PO2+ly7g4v|0vEwq-}qXx%EJ{&3Auox8`UBSf@;V!=JqWofgs?f4AYR
zgB-FN#HdZD#D0UBo+61W!+g%?Xn;KQoupSWm(@@JH3Z)rbX8QC<F}S7u8d&f<7}@N
z@=FqcXZ!9HnK%El1hNN<UP-XUd)InzxlO#SOK>Zi7-rSKwPy|%oV>ie0$dCx@?hO7
z@$#?BTHJQ;SZz3!HAs1{{ylcaPY$!j%N^KPNE)!0T`qqTx#(fe2LDHe>rVEI4daE2
z_4Fx>iMgS)+m@B~z<+UypTiVftn~~JLrIs>?-;+6fYF_HAXU4Ag(PN_+^SwT>KlWT
z#%)U|%2G**DK{Q2+7r#W{LMaNDC>!6_M6e!`R<USJzw?^q-c#G^~zh-WC&D?gSkic
z;FT!XwpEn205pmE#l5Y?x7*>rP`qAFQdZi0F;GgDWR3=EIPse-8M{mZ(q34sr`|(B
z`<rU$zJ|I6AYk4k@LtyrMTxtC=(tTD1y@42v2ZHkVZ>S?P@chx<dgsYaG6p`<DmWi
zH<U6FI1|qlio1DJH1CMk**L4ry4*^mn+bj&sX9)dg+Fr8^y1yHxK&+OmB%<+reZX>
z?<!=o%;wr-Yilc$^-<#%lda3=nvj132M5;&ikBl<fcME^CvVeOncVGp?*MR#7yTkW
zQ)<&G2!}bUsga$X-L7+9z|Mkhw#~TSZdXwn#JJ5Hw6aAOrg~Bal2w25uTbGq<EaC&
zV9gIK11ZUKH{`Z8Qs8Sf)JQcBt9ka#T<@=2CA;jt^ZP|<!srJp)5m|KI2|<n5(i5A
zNyx?k!4ruF4jmeT9A!S8jK|}|z=k^Md&fTdc+9+q|5bY;xMH&(96r($ULM#w5^Hb8
z1m+Spi%5w^1MsFCE{ViM^!;4@+*`R*I-Evdpvb8GuoArZt=l}tdNq>o2R`VKs+SeS
zj=z;igHF!Ce!aTA^kv$yz>vvuT!zR(y9f%7kjy9wsw+EW@gXn{6u+coZLNe)>zVQO
zu;S$*`&nKVsAUH(Rz^f0TQxsAnmL;UwaVue*?!&R=f9rP#wf|+NcxQHob~i3vOd%E
zNin1M-3;%^`8!$#1ibYRE*`tPb2Pb!-7naZsGRvQRo7GHq0{|kH9KUd;&UM_^AHeS
z1+&qc!Za2ik7ccAtu={P=@bb0S$jD_pj^KAt;P?DTQgdiOhTv~>#XdF&|R@?vQC85
zn8vt;6bhzqBJd)VlARN7hol*N`*N~ZQ=xg>xz&IDG`Fq6cjWhky36noo0R|7zU(Iv
zYZ`=akKZ>nSjyn%EHD(a;B&^G1w6Dy&xg02Uc~449`uJ1$Xy;R$OmtI-TK!L9thbp
zrX=lTEnV!6_IY!h$?l{c$$qK;HXY2{+>h4FXcabA{rz8mHHSHD!9p!uH-C5YDqihN
zngz(K(d)A^1S)A%HcvYu;F6P5g?Ac@3j6~CY<3fIYDuD)9-E_sL6L<3VBtp7>?{c0
z>XEkK93-4;W(8lyiL3qhF!Fbg3U*X(7r#dXuWig~6rfnbY{dn{=g5`Dx49jADIHIb
z0M4hFtVNR~J_mswZmlnmwcYvufsqD<<Gz*H*z@BvSrAsW43-MlT!INF*GueeZB<Qz
z2P!40<oKwKJn4-pt|Fi~)G8%2#PLvrM?)}Ll_DGst|z$M$dHisZ!FaqjR8c@7Xj11
zw_rE=Q<E?x!X$sbq<Y>XXnat)Z`VR?Z5jfN-A|KsX^9&Z&(D2{b~|_%T4L1>qimPO
z6Vq8m(L|;Ex@Vv_fp){2&?I_fyBf-}n|LN7If!{u29)op8UV|EN_tb+bIV^RdWyL9
ze;q1+FEBD?&bN^Vdqbr|51)oD@@^4ZDsl32zeFQY{Fxr@>CjBLHd>0M|9!hO1F~xa
zW`z#3=#vuTz;H0|kr=EL7&U$uWQXt}(Mg1OR8zxA7m-pzXa%<6a<XACUg2u-oVgAz
z)qk6sVXkQdc=25Mugq8#^?Y&tnqy|VQ{60~^Y!W3^ppw2jyg&OF==1>8{?VOxBp#|
zZHF=MT@h7`b)1cceHlScGpRJOj!4>S`7jJD!|g{hGp(~E)#VdHGJm!Q(j=>pYjLb7
zQS`lob3VnSg1086%mWp%o`-jCa*BHCrRG~dB5>aaT+nL9rZ1zZ6F7~;%l#dWTt$Y(
z4_?{w^YH8Vi(+q&zRDvaa}BogaF|lmX6h7^()rl(P>1}WsO7(=83q3{w~j|~XPqs(
z9j@^nLshz4XwLB!x4Xsm4lYi=Xjll7F!TGL>L$-^bW=8pzt5CqQm7kOEk)_S`)aL`
zkb?XR-v2x{KeD5ce;L}@;r3-&9!Rw!WkfkN-6Pn~hTCD4S2O3VyF_GfCr01QZ!MZP
z2dS(57o3ue?Rq1lJ_(yVwvJc4QaT)c3OnDLm@@ZK0QaH6hrPf7erx8CJZuEm6r6I#
zI@ZioiJMyW&Ten#Et$!>lvd2|dd|vB1Q-LxjFyGd0xDo5*1Ksl(+m5=HMGCO)xZeP
zmak@8>?Sx?4*RsSKfbI$Up%#)?4|9%{<PR8+Xlt$wRYZ(IUI6qZ->1W&BGqgtZ%hl
zfr!<sUd5x%yDt&-#oWqtq0#k~jq-KCW};4^NKSy;bzm>18jQLrr!E!T+i}Vc29}Fh
zrz#rk>D^JT?4@<<Yx<Z>siT*6Zw1t=UtNp`ouqU=0RlYyoOTcI$F~u%x6>usjv^PX
zV%sj%H*Vfj^pi{MpTSIm$-9c2i$fP{i)2f})#aF`+|hVhoRe@0RBUR-C@(^eu*I6=
zyRL&b4;;yW@~6~DySlz{@o6dr9LU<?7T=0q_?V!5e9m`YW!t3Fvq39@T~EJ8viE=1
z!M6r2?epGarS-;K9z;A;>o}KbRC>ew1ZVUZ46YsdoE)f>r`G1nUmwpFvIi}9ZR$kQ
z?gX4<5t2Meh}ruiq5V230&W<uLhvFV>q#H|!JPT@tLIDtCrcca@DpA@u7@uI4O7fr
zOoTOyYl~vOC%z%kY=7z+oq~|!wzd%Rm3;k&I5#gOlh8+pAbuX$h4n)+CQ-wimD!th
zGu_s$`W~v^cBLw)jR>c5M6f;La)W1)6UQd<yezROOGy$mHQq)-kWZpQddOKhiF`0O
zAU85Yx-)qhlS+inUNQnJs!+~9Fy8jH=J^8%MizY(6lu3-{<hE{5*ojk(^6aMVds?5
zB%sa;yG!;VlYySoy(6qH^SRUmw8viVCzR`B49`CZ*cc~PS`H<B22(@Pfscx^XcNvb
zDaHA&*5RIxsW~O%sr<#AJekd)yW!SedM!sir{pg(`MmvwegSjKOC4Vd_4*jPZK}q0
zRkNTC-q`H7-Y5Q-i>0VBGIiFm1~0}r(45j-HqgjcdW+@dx5ejW^{ald)L*&DiF3b%
zuhQMRXEfJ#{dmSvb9hLSwcq~FAi;$1E|HHnd7SNsRanx`4@F8_@5hZ?g&V9x-of07
zOx$^*LDL|0s}S%!SC^Aib#iv5<0Xmy*QEBEIL6E)Wko49uLN!CEbfN!H)J7w-qpR9
zCVv{~**LV>rej5}rA8mDSQ?`5@<#NjOzh{;vYE)d{Bx&;Wn#_u&oevJL7exkqyQ_p
z5?LBg`X`v9foW5gq_K@XHy6u0L`@Mv1xr$%F;-<%sP-9nrb5?ZE%P3Rgqm<gr*TEO
zF_x*wk}M|hea&=}AmR2hoDxQ7U(PX$v;cmNWP*Yv*qGrgr^P2cMXViQG!<AD*9^K2
z0Ge8ACLC+7PRSM5k`O%71$;+WPa~h@jrv?4G^tn3$19#5)@)Bz**ZAezH+T^s<j<(
z%bh5jl>~0F%e9l1Y~N{S&RnSXyTn@Yf>2=#DNH2^i~AyDp01i<I?PghkVT11gF;2E
zvN_Oh@7L%<7gWbvhs4)ty1x5Q83MUnoLZ*py1#Sagj<7xYtt__T&}Lf+m7arw(;o`
zO&S!DX8yq|m>0j(1Gf5kE6k;Mk=#-v2Z#GQCjj{F6wm-$2v`W3s%TzF&rU{@$z~)W
zz8;-VmD}y|e_VapV0ZEV(E+~Y9mfn(#TWE^y=5u^ZnjHvrv#0BJ?RbaJJ<7V*DEL3
zf67+=n6LlF@E6mCAoU_YqKGW@d8tIL5AIlGhZtlkXO&oT5#LJNTHVQTnFkhtAj#b$
zf7m_W#SQRG0@J!b(Gfr)q|z&`nLIAr)>I{%(z!X!53xp@`5mRkD-t4l6|Tk);#j?Y
zmL!rWEe2K&Ze@<4B<}FT^=epd_Lg$zo^<BYK<${Bz(9vwq>RDaM;SdbARGmOX(41e
zj~TdRt%R=mDJaU)B0)>|5%8}A92|4l`iZN!*2L2(+#&f-@%kSgUp{9G7aH<pAc_u^
zyUj+)PxTJ&8}rWN1;pr02rTrZW0dO=t6S*K!KgSUe9<Io@h(x9D#IN{Fb^29O81VM
zE;MhWS{o&)^qFF~-{6+D0K*MRc*KkJ-<gygB*b`mIIjgliNkeL8Ng+o^hoii##_)B
z;>x@x$;pSvs7%}@UK&2*8J6{V1)>0qb#^s@_36Bmc>o`s9P?DNW17aDE&@VA@3^c;
z6hjvRBW7k}A@1zo6pr8^iF&v{p;7yd0zz6yp#UZ7QB^>qzc5VLuPN&(LtEbM-lWp(
zg1~=|kB?6U<La(~?Y2txmP3DvU<W&ofB*t$`Uu#bn{E`#xfm~uNI*#C-)8W4evNQ>
zcwbrRdu(y3Kv}X%Rbqi!)$};Z&QL*8JlRGqn|DlHJkIvMmRjLFlQx&(Zzsvw-Q(8o
zi2Lf1?Df~!WLmZbzyhvTbEM$#urce{I_qsEF;8kz%#;lfK|#miQhgwY%e=8z(ZaW^
zzsq68z)9n`O-I)ut50dGA#aO*#5#uNMv&=G1Z>a7afK~Lhk{l8MzO)3fS@W#fvH-L
zd{g<J4LI3{uEfxrDBf5wVL0;O#mG=jAUJf3#1mnN4f=hnrguD>)UqchZw<gd$&G#%
zWZ|a|;RP!5%eJ(V?9?W2%0(kyd6lKMZ~WFREc>W!q^if7!9|G^8)iinsgm>|(eN&=
z=X@Z|2A%iwBcdRQON9!-)9_Zps6gPnrDY0}`9MAI5nCJHxA0M0-Nf~NeDJ@yPuT(c
z5zO0-nOQj_#t(!^p3+%4))+^P_Yq)U#vF-0#ewf%8;Mq-Rb^#ISZuAe)DHeN1%^fE
z6C-cz6$#^sf55ymIRvbm*q7?aaEBJN*}d6nSy6>>kT4^u;F+|YOW!Bn_NZdca)lZa
zI+t9f?(G?ytVRhZcmzA!7LWBU$O96CdVrl(?Ysim)}L`6F*2I}b5H3q9T&X4PR?ut
zv}CzMDKf>$s(}Sg=B?lP>RCelTqTZD;7fDB$}(m??r6Rvk%9N*!QG>tb;CQQMOy;&
z!fY~}mENBv_vQ@OLs&}$on7!<{pXqV3+Eu*Fm6Qd6Ytr^(v2QnwWdcNu5SwYVk=*v
zBA{epaJn~%FVUE32T7M5MPJpoKLam;x4BC9$I!&@L!ywbxqKuel^`J<m*r7A`WDPJ
zusReL+E&T&s{|2$;-Z4^s!l~8t1`CuXuknP+p}f6&sPREr%zP_um5Wup19c$f3B0Z
zvmr|k<yQe~#>7wG4;k2%ts(EMv|Sy6_2U6STKPbCo1?woqXD<r8ze`JXJ>&XcCEir
z&`iYHwP8R!&-02y4ZQS~SB|=I`kFiyAEOELuaEQ3CDh!c@@zdem|~)#x4*$+e$m1%
zArVNrPKvVX-*1sAe{vm??jJvte(G7pij~kH`c6Dxmk$P_J_Ox+py2xTh{J~^u_@UN
z6M*qpn7W;8k&<i2k@+-hdo2<uSsa2}swx`_|BNc=)22d@KtzoKTq2;+sseB>O@3E+
zhzFMijWqd##|JMD-zV1+%rZ1+Urma4+@4zdVlW=5Fl0*1O#F!lY6XYCTO{T<U8G2R
z$-}_eJ^t_k2bAcBFiE^BMXAT+8Pt;NYiT?u9(VaD>t>y1CV}pt`SSx{OA>qbCF6PU
zcGH4%O^x5#1w%>a&2B^a9Q!<m`zSb>=HvDdNO%f@v?u*zXhO(O?#m|v@0HS!s4kMg
z?8+s+{1^ls2T`hCzIz1!nxBMAnYP@w?^VAWegroCLw-aPQVhWM$0sx2J?Z44Q^!|&
zEmp>}oi!8eSMx3Q56G{9j&r<|qmwv)3jaVbuxJ`Zr;~uZ!_VbfMp)r}A@=^+%4Opb
z4wFrUiub-(UQyQbi?e|7Mm}E%>^={tXp%Ms<;wo({_b#!_Si$^ME?A=MboJQ#;mFV
z-OT=iH_>1Cs4%4MNv=P3w8ByZlozR@o}v^AFb$uPg4MQY@gfB9cU8!cc9J?Lyq_3J
zm7@M8Wy{v9MVvpm!>@#M+mg4seS^hE`%6;y7jvhZnslO%?^*5a5C_pv#N5!SY-{9X
zjCbZTL?NVco{8c;K~wZ_D~{W`m%+n72`?N*g@_AYrr2mP`eolARb<r#^TSav9Yzyt
zzt<WcqD=l47(yQd#IN)w#J&sUmu^q5{*bS&F|<8cxSGA*1Wt3Vr(U=DUQSHKvo~F=
z%et5Ztl#kiZD=hQwAYb}50#Y@@7}w2wa<Rtv~Z2TnvIJ)idQ^a<rTT|Jen@F^CE0q
zg_UTU$j{BDKmVW-7Q8Xjzq;{Xx_HNQL3$G?E(%8&x%U_VrXu0sgX2f}cCJ8-+yYou
z#LkHI3^7%eZ%g5@#%1DGG;VaHO+I*R1T{0y*c-Fs=p_tx4=gpsb=)sF_2hwV>rd<t
zeo~gboF2Iv`!{-*=|W3Kk#OX4H5Pxh|K+&tVEft<ceRTu5MzdcdvN+>E>dTwyR32V
zC9y$*p1|;NM2XKmIXS=0nGtLtUeNq%;B5W))B`k<H5H+4A$53#U&?0Q!qrkPrTv4j
zmX!C(0feeq_xgiFf_VdBx==9h%C}6>#L#2Z?_2-gR}bM4f}Dt@#^lU&;A^pIubVe2
z6;$&}%!<Bwn(|7;!iDbUWD-(bS!l$2hStZP0qnu`KO59ccJ>rfEREl+3u#6}6hv+m
zq#!A1R0nJTiccg1r9z}D=BK|s6odDB?M5=|J<geURQVy?a26{eOQc4)hU-^!<F)IQ
zwH_dP7KabQxqv~Kg&^E+-e$iGa4!3W@MOYHRXz(qH1+hHky$xPvE7QiK|x@YHDm&!
zy6UO0?w!)i9PLdOm6?>~%1qVj@WlgrWD1gxEIf)bU%MUll#{cSge&Cj+XT6khPS2M
zJg_LpSM&8MKZmmW+N%6+l-eB3szUEK@@Nra!s#jMp1K+x=OIxvFnW%V6S{CJ{#pzj
z1BJ%wrR5I=6WP{E@4%oHgBF971~mXy17%-m_Bg@jt-1wyIR$wU_C<nr*CX-ooq0(A
z@y#&iTPRT`(0?^(On_OjL5V0yex%XUAuxP!=z}G4lE)yIgqib`wi;;{bd)bkJ1m4Q
zkJC1b;ktSH^3$~jEg1g3$Am(kO$^7od`o-Zsa2u4qc+L<;U2cFgnPb)089tU8O~$r
zO^)B^<@I+_5P?8zZsEuIi;h#O4-PolM`qGq^7l(IBZ>N9^z>rH5F`jegU|=2w`i;X
zX{mYTKnYPe>$bd?8TT%n7-JI?PHFj=Ec#6cg#NraFp?6sk|x@d5X!wnf&816#upQl
z7KL@dQ1J2C_g2;D^oIWeP}@WKdrK{Q$5|~I>^X*pRy=R=pBy<I3RibX>c#gu%P7vQ
zWfCRBziMvGkVbOVP$_s03(cqy5xsA8Fv3*{S%pR}R*!0QSiJP=cr#pr91@!jOZqvE
z>Ag%%<(@Vd=#2P_Jt1$&PaceD=gwz@KH}`!IKS-3b<R&1>|?S5gv80Xtyxt0Q<*Y)
z!P8#6+OA9RK@sh74Jch_zuur-2tEpeuT5<0d%i&<h==2`XQNv)e|M%CAQugVIj8%y
zfLyjDPQ)Z|E8XsDf&GH}=_fl4ci)q+laU<h`A0YNRipTMW~VKhC5%+}f)+F5dXqsa
z7Rg0b6RpH6l^B59MYca(m`cfURwkST(uG8S9U2<WsKaE+FVxP5k^4wWN;>IK_lMQp
zSv-tppda&W#%1)2GYjw3#_D5%?S5*_*81ysk&6S7?11I{XE|6t%VcU2+o(h)dLqw)
zI0S0Y<dsO^X$c<qH)>bsoQU<?4U2C(i*unQ>#eI31!jha<2E~5&8st`-&SwY^^gZB
zaP@dimV7>4#_oL((yt}!4Y2V`+jBRIh%#YDOFeB%drrE-|JCCK#n+nb31^A)_%r45
zp&yc}{FQOL)1GJ(x-a7mDNw7f6qp@W7QAon3B`M0*4q9}n4fl-Uw03DH$NIIr*_`=
z(~|4?18g5<ZigheIiojE0<h<`N7q~C!P`fl%q=$*nvSlblX0}iCrFB<#FJMJq^jC|
zDa1EEj6Oq`Tw<q!mu7)uJK(33484%Tj34Yt@aCZTi7?MXPx+;#gg-kmovux~C7<{M
zB04SD&Br{0qNgMYYuw$83}Q+o)`Ot^9Hr$7kUd7zMUG70(q;kc8t8Y3gcw{z^ypD>
zVjTP*Sg7$QbFyO&eO*9Rz=!*S9ye;5^fo1jw5RGC*-5+7<(&|vzYGtjG@&4ZL!Oan
za#gaay&F2q+Z|mmqK-G6ka5{Qb-Yk5`O&KIS%b%KqK>aBP&9j3>-W~yz{&_f7^4Nu
zo;wR@x!pqiV~kek;9Lxiq8H_ip$O@c;v&}Q0v8-8c)k*6E;p7o-&(yd9ip|mcD0ll
z70zgw?ZFR?j#BOtvF^tPtaxoVrgkjyUT?DVto&s9uff&Rv%0nAG!-Pj;IBICOB`@>
zO)hPOvZJ_f5&eYbKISU5iXj%ENPtcb{_wf@z<Z<`|3t^=XwQrQC&W&h>Tx2>DwB`f
z1i27pwTwL<F1rcuIv{R3>@!}7*vQ(I;>)i-QKLwte-9-}gI>3l(9pTHdNR179StYN
z8<&j+HFVJ0zEQgNPVQ-UpL;6eU0hVZ;(Qcl{Rq!}{w8ci&lo79_`epR09?4W5J*q(
z%+{97+2j7I5lO;P!+58D`%br^ILkuQDdO~nC|Rap^37*?Z-3D$yuj&%g;G?>jj6ot
z9T2la`r{8>t(#t0v(T*|mGL!*=VJm{#tzT3Z-ms8?>3q}qJT?0?}Q00xaRk#zIs0O
z4%#bG$3E8*Kc1gbuCV`S8*phciJYEwX&c*4ZFzD%JhgCu1$J2%7LA}caR2=enj_nG
zZkkUDUqLE(Fa3^AE(HDGJS&~W`WnY}X}_)A>-`)%`TC*vztD?yr`Q>qPGO|e9WeE~
zGsSm=)`u}muYq5hl(V({R5zl^3CKR7K_&uE_&Dn!VW=xyrDzuL*)wn`sf21vIL!;j
zZZz5702%2mCzx$%zrKn6w^g5xf)5BjdJ=#BC~#9-DHZ>s#Dw*cT?%x46YnIIOkY)E
z<LbK-pECc~KVrrou(T|04q75O^Zt~??0e3!9uHLdd!E4c3nY!oW@`rsq*6t?;0c0-
zzN2Pm^=ESIsW0H=<cqtyOfD252Fcv>!PlnuZO~nynnI}OPwj4X5qW;1wymL=?KC-i
z(6^h$rVC(q`Q7So()bNZ{sQaj4|F6}=?H3mC;buTBbUo#{ME4e^<&Y>T5%a;b5li>
z89(J?L|3f?%j2lT8|tcF^VZ7ytZk>Wif2o}D#TA1258c<&nX6-?Ob6$_>@gHmg7P_
zvevZ!HahgIY<f(Ye?M~oYagh#1y=lC7c8V>^xuyKc8Y#}o_CwEr*bq3(B&@+yj`_#
zK9w=`q$MOF%AfP(e?SP|eH_h4gt!{9=*s7fN8~d?{PEf-P)aOp3JgX%bLX#SJ}EAq
zGsMXQbPSWz0wQN&S~ay=rKPD8FDa7)AFFbt4^<T}Nb<xI*~Jz(JK|PyYqD;_Eu-F0
zj@*oXBvG5q$!GH_=4w4+P}9jVd*~n#q3G+1Pd59glF!(l)twa(PsJ_hw4UtW?D~B`
zT|Z}V0FG;(!rxXlI}%7{#jr4#*r=eMu+=jHT?WJy(Sc~V@*>)!W*n~u7tbU5M5DTl
z<}Y%@k$@gwxDp!9jaDb63`D5c`dsTg*}>*rcJz3Uvim}qmsu-X^3$;<VAl9m$~b?w
zrptWh-51J>sEiH^<$WQt9=(3Ap47$A+-(ESh+Oj+YuWc-!Zi@0FT13yNJ10npg=4I
z(T_N1Ey=v4&lq${CZE@?rN5^|ld!K+f(CzR+K)F@W6E|`j$udrwOBG#do;)E!wzZu
zr_D*U)|-D+Q5t*?zAbYwiW{Yxq@f3g`EFtT+41=IntRofF58p02-wtEP5S$W))QOB
z4^tCUq3II0ll@Rb_T|cdix^gKQv)dIlo0;*TpxC8X3Z|U;c(}u*X-Xxvia%OgowPX
zk-6HWL^1-yY${$QOb0_xem7V*aWG>toGwX0<oC}M5G|u{exs5-tI9vW(F<@(J*Lt^
zf$|28XqYRJVL3xj_&bVsp-|$--b{%EFlj^I^sWAMT3y7f?P_Q^UO`?4s0X&@ejb%2
z{-!te$Ja<ItFrwY2|ljscI=@r^Sw&1!U<1%?}H?Dw!MYfIfMQfv)P7VGf^5$4Z>G7
z?M-sJh8<bTcV=N774h5+SzNr6)9w7kh9Q6D`SEfm>IVp4<VRepgw5F5zAyb}<u|^b
zd+CKa(ye|%p0#~3(BDgduYVxHS6_1h^sJ6z5V=pvrMQ0lbP&#Jvnaj-x9$yY$8Vh`
zOZzB-s*hZ~ye?O}Ym#xP{JPmTVc9~SyN!Nhj5aO_IKuP=7K4>#50*19uPa-aSDeY2
zI(4;Ca6o9Y%84?DZcUE{dQ0V;ZaV%p?>7%xNu77A@;bgSzXW=(fB4k9lzCyubHtK;
zvT7!!ancN(!Ry2mFS6PKcYB`UwTqpq>V0rkh1rcv+BE#9<u|xg8BK?DnZj};WBY7s
zYELU#Pr!GeEAv0yx*VQW1sPLS{lS3M?7Qmd`epPpTt*twuBo$MYqX;EJYyl?XpMc=
z+28fw59d4mHNfiC<S+;XuX`_O9FC$GPUZR#IhJ>smuLR3+k9{9$F(bYpzMYXPB?4(
zXsRrB@)0U*6-}>UJ0?P<!&Z&Z5alt@yA9!LXooe`zCb{Ja*m`<Mt@9Y#6*ijkplE`
zmdlyq7NP3!ReEgfs5tUgAIlAc|FU3qbWip9cns{BfU(+}S-oA1D`Kw=khwY(!XQTa
zkx75R)7HRCDl-duXf!=ePQ~#vC_E<-{z;O4e<gii&d%`q!fLAcijQYVe0oAh8NQ-n
zhoTVG#uSLZ)+PScc=){3?m_a}zW67{_VVqN)zC2L%bPLozXZwU)1&1imGi+|$Xzj|
z9uuxr;P?I7$pB^uygrTdEAK~W8f`!CVj?2d27YFv#V6>_)5WZphN9eV488zO3-OAh
zwHWK}2?tH@j<7}TZM&SH59)M?S|<WF0AB!&V9ZO^N@R{AX;1ABPm&o7q>1%^M`ANw
zg%M_|!MF>riXGiD5h6;ChC*Nrl*yBgLLLBGOLfn2_R+4TLdI}`>6E_<k!1|H4VdL2
z!?iqBB3VX8ai^OJmZiQz@e^0}k0fUVNK$vubAl&>r`$LhaldJu^q;16KJ{&q*o;}8
zAPkiaaX6#E2&$aX%_7r|=NV^?D5vM7dXE_Y?aqxm3lB_W<IB7MG?XJy6iC&dC4fUu
z8o$Q0U_f4nt5=s)VcN`TC=nlbJ{Y~ADB7Os?Cc!1;LG1SIp=?!jT_q>m+ocJ<ex6L
zLO*(5_0U=xpQ;U}O3PX9TSBx4^1rK4xH$&I!pzNQ4ThMfvI7sYe6XP=f!m*2%Tk%T
z9<6{(iu;S3n%6s;3%-P1#Vem{VAjE<xz+!TY^A4b{rM&@JMDpqcp4j9Sy?FsyzHhE
zZo7xhs%+Ob_Vgfm5+;66^P8|GaaMZG#3OWRF<I4{l^Y1>b+4x8W`TSP?WsqY22~Ob
z@b$XNw5hEvU~9Y|WK>lO1;9+*o$Q%De+}s}WTXjbS(-eKrgaVsyqcQ^rh<>-O?AwJ
zw$l0BUzPu?6uaygSXVLAyG+$fJUQQ<&9f>TG4(t9Jz<wO5T7IEcDysaJyj8}@;|$y
zlaE|OTa)?CNae&6svG=5+ogJX(m2Q$*kS8_>_Vg4O3y6Vm|Bj%y2>%T*e6U?EVKj^
zmOJFv5X#@+PzX4LV-gr<Ogy_hn8szFEzw?kyL(0<d}rO;!3@D#wBV-zy?0+sR4FXN
zDw-Pi8Iz367tRjEr+^Duz+6B&XJota1L78081%o(tbpwd2O~iam~NgNdwBbYTZM(1
z=Due$RTa&L0zC#FyDfg@j1(sCr}0W`>7C%`qxntzwv49DZv(|LGR4tdC#~ld_=T2*
zadFJoZ6jiikkpCWWnw%j66r6KU{A{1I66}vVtb*c1th}f*EA7OlP@Q?qKN%FNa<gr
zv>nKXfbuvd60Z98G)uHb%$hEzZ$<rLdwmjy$Kin%BRw?XDce!i*=YS6hLyEug7BHx
zeIibso*O#9xd5Jb{VCeUQ}#DhxN1{s>8)s@9Gpzhi5<-~d|Z+flhI?!4V6I$mO%CA
z+sc1;mt8B;<HWnA`G~RJrmu?pOmQ}wf$dxd%dn?o1;sl#QeJx<yR;#c4@fCFmJPK;
zZz;XAq@j$k4jD?DYp^eSwum8rkh&-US9xp!wXzN+T>!MfKWm*?v_cM0L{?oa@rZ)M
zZn=5aWO!$))`HiA0?)0X(~hl~Z+neUje-}M^71AucTdAs=i+D=fUDDr7^H1a_{O`m
zBq|DY|7@+gsuRxtGI7!3!2Q)!i`i{aN({r#w4@Z`L^st?0^@6$q7rwA9L9ExF$jeq
zVGWA|vU{F^)T%zWhbYyKl#q=F<QVt+;U;FzQG9SWP7Q^BJ(5;a8p#f$&x)wc)|1rR
zVWT=qD&a9t-_&BV`X|rRN3Aww+J1PI(iUzDKE)%eKQh9}g<)Y_-!LUf>i5IZ{D_Wk
z%NFvlIm7#X*-V&A=bL0lQoJ!o=Rr1H^aC0IFaF~*iu;SvQ3!lkg5{>L`HYVh+TpU*
z>jU+Km+3UCkaWag!0rKU3TLtncqDJ;qsrk3|B_;x1TV<T%Ue<{OsQ-Jhvn6=agX;m
zmRdldM{+8-g-$lIw%Fe*^f{qrmv+6r#9xo@uM`5u%Yp!*4$yZ*9|#5bUW-330v@^}
zGL8m9Z`{4hMswa`26##-RPu*HjkZrgf~|?`=(mS1LKKM7e;rrOddJ>A$HFG{q8Tq*
zsxrl`h8$}lf00odgS9s)LSZB5VBc%Z%+>%F>ujrmjz+XLO8jYQTA4lNm=aRv*fepJ
z_qi2}uLDIL`oP(i=G77i-;M!l6+AYh{pKlLySL8j%d3Q*8aif5VT}!CrPBV#2^w1C
zEinV6*sdmMgI%xwi%{3UC(9P^<4xOuq32;X(HW`KU}t+VH96^hZ+FeF94hhRZd(8z
zYcIM;f{u>#!3SMmpS1xeC&6~FL~ivy%a}DzS$345?hs+3=3+Y};{{%1*Y7Vi6>2);
z<^3Nj$QhX$8_Kwt1G^J~@uvGNqj_0bO6p1Ku&KR+Zb9DK_@J}?IBoTpHilcLfqN~A
zQc{nVZ2!cu;)7O7%+FUlSrwKXN8{{{K-{nsZjP57pttIyVaPlx9<Z0s89DJeDGt{B
z?5H)erw}dzTpBN21IdaluKUYdGXsN)0YyUoD*KGP7Jr(2u7lB=|K(FV(X2GWUt-9e
zS34{(y1N9y5K*wNXSF7b+v?H`L-{GwkDmns{q2qs08zoLc`RX!VT+lfZjgN4zY1Mn
z(1uq{B5md$`_t-Idx-3_5IpbH;hswLNxcb~N1+}aki;m!tw(BwDoWC8w&pnnQ9VyW
zyOi5KhVCeJeB^imHGjyxa_5{@Uz?znfBVU}SCbM0l`*5w^zryP*_pNq*xRy&)5Vp|
z(zQPJHuue$Wc`#R{#>XWw}aUGqU}_&QNfNu&XW$YulXsHGb1f9N^u=~r)sgxpUr*W
zpepK|F_aAe^xh3IRT7SRjJv9RMns*{f8+Hq)qFlF9oX2<;bHt$4{&MTYJGivy^psW
zdqdUg&{*qL;;?-l8QJpIb$y86=mStQ=jW?vK;0gj3mjY=FLhvZF3uGj4*oI72!`u+
z!HAwNLQME>(unI|aL+%2Qg_Bsicj9~5bjtWx7WZ{BN!GCcs<)d|DX4Bs%#>!VY&PS
z{tiAH?er128!{cqnirRd6gKX4d}GzIvGo)G;n$16Ad(JsrvYq;cq;NgMs6fAN;UMS
zWBwNnp7ok_`01wS8TMTAMV9<s(kK%Wnk<KRp+2{W2Hqy<JuScK+Mln1cpZZvB}%Zg
zh-w8K!nUOzR|rMB7Tr8BwJWoMXusk7)ieIiC}&3ap=mGCV6iT^=fvcD!>#?``PL@(
zWerf(uBg<O0c&YaY5enSH(?0J68#0F{qf(6GRXx)!(FgPH?UQ3zM2=$Ts6J4bd{TK
zdfoG>LjEX)gfj8^$oyj07x96n?Zmi^zaEeR)-yRy!xYbFcUL+^bDrw6-(!<Pqpc2|
z3f6M7N4EX9yC`$yI3GkC`F$k%om<0kli2F<rI>ya;#N~Y6N#gj)#c(MgtFL-pJXcL
zDGJ5%x-!i|MnND}NU5UcZN~9Z*t4l}`5NFyQ=7-gkCbMxAO3Zg4|~44H^G}zL<5NE
zHTwFW<6PQqTfQu?K$c}m$vVtU?(6tXj|{#xyJ^h6r#RPi?W)PVHa22<GO2L=AGJ%%
z<*}X5UTH;{@vOILQ)TVU!#J^Nv76w`C&i`#F+F*_f1>EyR<!M{#R@`-qv2Bc^{RA<
zvp>gDS+<n_pC$NsB&;(X2(%hZ@=i@j4puIB#jSR+90i2IRvabzD>Wz*ZERZ&q;ksT
zkH#^3_i<0>Jw*bKN@~ka?#aDm3;OX~UoiiZ(AB<gY0ALuc#LA;zrSEStKr1nMK0@d
zd*K2*vJim(Yw^jleM4G3eRR-PJ8?}Re}7e#ze8$>`DkhMV(v8738?b*y}pI}PWlw}
zxG(kd3*2@_?}3TppXssF#pn`EFz4grB!3<1eyDgc+ZkpGHz@E*v}+g*QFEtcOnns^
z%@{^cPazGtp`Aojz(`ET0$v@6Zdjgfjxl%=LVG}XLP+rqAa)~pEvh}@0PYH}Y_>2l
z*rtpi5Jg48vQB_<uiDrYRn>y(ffzHbV~y<lNwiPicEV`k6juJ76P7=&{*8&0ma>bS
zIXG+^D{E2me+%Cn-Rr<)tr|ao&;e&N#vv$oxAjAC(?D=|%!rH{y-Ch;qu0>0b;3w2
zgwYX=5D_fXYe@@W<>T(i@265J6-|Fd&ktAWN!PWd<N{Vr1azFU6t4e`O=+6f2QD8@
z&!u$-noZs9tK3ZUFcC~f&iygC2DchS0v*ie0Vf^vvcrxsbPR69GePy|4sB-+hPK*N
zp-+H^_2<*i7ranud7Y<Lnx?*|g(Aa+glF-`DVX9bNHmj|RTld#eI$+H>&ntJ%zEci
z0(C(uC!k8AT211u(qRzA$iOCBTO=+VqJ0no%M7X}{9Sa_SO19Xv423q!HsE<Xqark
z9OUwWpl5|XQL+${D5#95PPCIFGfi@WN)a+6#)Bw@5xLDKR!(w%Se`bNmsKX;Z@yL|
z1JtgK|2l1a!|YUL&qPSkHoRGhP1w%d9!d!UEJG=U>sL)6Uv_r_Wir^};(mvXa%gSM
z@JLs%h-W@{^Es1Rn45C1K>;=rB1j^NvT68$pnEr-juEpk;m@xIFI3jf2ng}EDTNll
zDP#V)K+m_Db!{;WL(;FLr?quWnycX!6vU4&54)Mk_1t{xd1<b(E<tsoq<#`CBCXfJ
z^~LE47{s_$xMG(bpu&$~$o_*xDL^NK?sLDMDB&@!`d*mXy(q)qknq+>r1&ZX5)wkB
z34i2NR+W3Fl<Qlqp~VkvC}d5UB}q|A&cu8{0~Vd&(Z$%Z8~z)~EUeVu@BKXHiHNBW
zT!wwXtic0Ug+v}^r#p^zv*yezdh7%J;Fq(96Ob_awbm=Vi{B0Q-=0hr)c<a<B!H(?
z(R*aA!TzDw<@#dxoc11@qT5!9`6U^B)*mbt!ldP_YiTsGW|FT%dWW6u-mHQ``h-oJ
z`G_sNOK2oES58V!4C%jJ5B1Ad{1Y@@G9V_B4J^qpG<%p%Z2LHwj5|eD{{$M=Zb&?5
z$nz>+N9R)eeN^wT9V^=2(PpDUY1mfb+Ah}Wz@Iqg*3(_^kuo@`L&GGR&#jS2jc=Ls
zFFF%vBw?o5nE<<>ORQ}_YHChr1VwH}%Kz8Sfg(atDrrzgtD=Z{pi7pB{VaL8&!&xm
zM+a8=!B+%HDQe~Nl;v<W;H`2yE3T^f-)bJ~EHB1d`T8PX&@kJ_E9<iepI2u(1Vk2T
zSlhTnESsoiSw56eNATwv&_r>is-`*qKoXzZHQghMO{-5~45ZQW5DwpHd_l2)YC8fW
zLKKBiT7FMe;_T6};8WVWL3qdH<QYX=YWZ_Wti$>vPU#&74?fKjc#NjbUVD!i<$idX
z-G?Ga6`5_nv2se$@60WQs*2iYylG1FZGyzW&-y(_YhJ0P(;uIw{%+iERBkc2dz8YV
z5&dM~Ae80)JEnfymv4$nokFX$?YnK1w6&T1vMoWT$pR|>o0zbBw^#N_on|(eur>WQ
zLn>QKzp-OAQp<D<6i8~5_dhdQr*Gr)<_<CcR>cdBpQa;0&E{wYZr&$&|I_5F$VtO~
zKjQBLTXPYErcO-}+gqI{Sl^YbsRiHV0daimNr%YgL9K=F`Eg6sT$-hVF=GQIKPp1d
zA+OPCKx_rA_7&1%-uROqBt$y2)p|Qtw|=$}IQ3Rwm30A1g8NSl3>mfiV4}Y9luhL~
zEE+aR=TFNiZd1WYKZPfXld-+y4do)mg(kmxK5t;mx^U-3rrHYuk{AdZXN7wgq%TF{
z)xr0!t)gUWft<k&Umc&6T!`vpKRt>EnLZK-(_`O*kf9CY@ULG;w<?vUJPSv%y;^rm
zw?7++PHrCo$SLkfq!@n<?K`z3l8BEP4kgr)L(k9>-5lTg#TY^#3H5HwexxV1DqU<Q
zwDS4)lXCKvWGA@GW!-${U}!d8cH7dBm8*R5;*=T$m6_QxLW2Kw1ei|6k#QqrfTf}D
ziXkbmzxrmugq?hGa82iXTJxzea;(hCLE~?!_Vob}(01^9;o3qXi=v1$j}e*5GVA?V
z?*3g%)1zWr1zNbX^SunV=S;D$-smwIdJ?qzpCH;3W+FfUZj62l5&K`mmUdW4a0RGS
zrcSJ0m8lnAF>$9;^L3#;IgTdSXKXsh61IMK_ZB>he;eDMpCOyvu^wF(ckh9BTv8J2
zA_M^hZf-h^VOlN*21kIF`m_9P_W2fGo~oL+|E`{~KiJxvI~@d@G<7qL`%WGX*RK&U
zl^*7}H4PzIjgqjXuqK}o^JaHKGAqr>`no?#Dgzy$H=_0<tJw#SIX(t+0T^vn&@KJP
zz}Vc_+-27LXBu@`B7wnf1oPn*qo!ZJ9Z*&KtJ_h+1}(^)V5wz&{%j>O{+@jB+P?Gk
z=#v%Rxa;4a%zaPhr*113DUA+hG$x|$3UmdGg}|$u_B^b78zg4*c}e_xU6Etdd^uVu
zuk&zB+Kw>@QtLU)3Hmp}G`YedZ)6V_U*yOH;i^FFLv@3fr2^IthUgly4!4v5$~>Ic
zlRHiJ{eyiE{?h|@T^~8;+k&oR>ElK7#krr@t=S*_v1DtS&dmY6efk1$cR1|6rDIV#
zl{)4FG^#H`SGy+#W);o{Gq>IR*hcS$k*4IZR3IZ`L%it@e?{rJ?Q83~GWVZYQcOhc
zC#|k;G*gC5$>{Q0)8PhE`hs78$slGXSM>SV-Mc#_SQX%5UknkY;n6v0diiime0Xcq
z7a`9<gNjjl`>Ia;2#)(!vmEM1q+Bk2A8n@A6EReYzVS@IH|^U{qxZc$$#gBGR*Vs<
z%tmW9@|jm|eQPV$b2s-V@&sd6U)|U`@9B2({Z_Ogdz8n=zn83a4$uArR$otoX%9$#
zKOu!P_=!LVE%yA7we9av)_Gl^t4(>iS-{@!wOEW~G-n9Zge)h!{&<L-Jvh8{A`~0A
znQu0Ef3>q?t6ztgWifiC(0Sgkx~eMYcK&3bM46QuYO}M`utNViF805^jgAJT4P5LD
zxrEWIoX5W9#7I`zY&iq)n`@F|^axk+4CH7IHcVr&W)yCS^qxL4x|p|Vo9HOmiitg0
zB1PjbDMd}p9c$$Zc1{@;RfJMu9LejXZ`EioH4G=C00ULTB+~u$K-^o`Y7db?>wDA-
zU;0qGU(gV6%;4mY_!@s##q0}AS00LVV|Z%46C#dayG5VqaRZ3aq8y$4UHhOX945`&
z;T>B4F|~_?l7fNz5F1zg&f}H%G{(+vqNG>>QEMQSg6LfO<u}fwt{RnQ`J0D9fSPiq
zcKgUb^JmIS>p&;3-Df%SzIzS5;TD9=dOqsytb~Jzioou)N>d>uBM7^n0p|etxc3uZ
zwpvbSSg$840)a?=H4>pEM+pMF^pK2vCC`eL6+6iWVJO)y;ESuetKSu}(B|V36vPm6
zUkSq0jK4|EPbzLAipYAg+wnw;QHc>QjcfJ+1w1W2CuXS`81LEM56XCxsr3d4m0@Ga
zbd7+9=_cT0@^JBY1U0IJ0MeK`*)MTNM>Vm9ny+O6Nn&oHnoe{`RP6BXSTDx(vNz|#
z)2t3BVwx_wYi8#Ho(K9tysZ4GuWXK|6E8H=snBcQ-1NagphUX*M{zQcCCY~l-Yydf
z+Eix`a$j+kdxSbv?}RINXlBW|ogMb4{02efp9_9w$xBJ4y(m<wwtvgGJk&tFvj5HF
zraSOUInj>ge~}ovL}<cG2BeD=h~ABS_Kb1PqP{$J0IhA5V*7)wk*}<ou<I!NZ7nkZ
zUseNx@(!)teI3sYfFBBX4mGo+YdrAcDs$5x7-A8Pho`=_(kU_gxBVnGXg~0X=WH}=
zPX%A?&4vYoGNjtJy5vwnY3Sm>fNP}z;8_JWBnG^+js)Ocg>}yNj18xQ4*$kZZ#>n=
zj1G(@NTgzly@UUFO6WiCfhuF9wNsG9bj)l-EOv=6QD19_9`~d(Cc(R*ELKJioG64l
zM%t>&mjubVxcZd$+hYq9t45-S`=o8l(NWoM3wd47uGG(vIjffTG>E*N2^9u_G&_7Q
zFf_}ttC~z2LdQoBU^zqtdeUhzXA&~k_PBBf2P*h$cZr8J_mR1L`M=+b$E0iZ^L<eb
z=jTBpDV22VOp!y?*$waM1PN4#L@0>{-{>C+qSA5~*A6Gu*<Q)1I*j{-Y^HZrfdlJE
zf_P7c?N7N8;>ux$Ui`hlG!B3Vg?^x1x=C$ViS1c0Jn>O+$RL+(3g2Iiehy;3Ae}b(
z8T{SSib3l`PWbJ8T|uvl)&1Ezy!^=pHOhwMUE0M=U29FS<9U@a`$FUCk}nYaF#9H1
zSjytn<^0_DSJY^%_oPgTSMb4MIz!<LvW2^GwAz)I&BTw()&_noJOT)2F748>2>$$o
zT)8{!m>!+f-uX*;O;D{PhQ(>5_}x+>34OiVPuC4aehL*5c*6FS*N_X{9{v)=_a?<6
zR0K5M1StJ7eD_)RR!RNOKMOab(-tR<+=-&L?6uBzC(4F(aNqg*3;iYHTN90Ls8{mI
zR|ZHbnz9^9)J+ant^%2-i&WyCG=|&Z21tpY<z==Ec>yxf5!_SU$!gnyA3^%R9<gni
zh#<Yuon&d8{ZCj~X+REg*3f}Lv?t;O3rohGCY`mzT;URuk_<dtDd_7AfUsZ!{`Q{(
zXTb5RftHpQ^-nss#dX-=1?#|-KQfm7OA*_^6%Ghq13|&!b}u|aBJ0>GZ31WI|M=OX
zi*Y&{uhmlc27fp5sPIOJ45A<!@wfE#C)NJ5l)!_F2=1<wO`)D2wY7~yX$)GgRLiXG
zcJ1n1TuNOO=KNmjDbEjaaDN&160Y#q1JB}B!CLaCyDn{qIoIp?Iae!dyxIOz*`~qA
zrwYB!%~Qb2OTHU}Ki>ap{C+U1B4J|w=U++N)po_@zru4#mlH{ela?mmHPqC@mClym
ztE1;)Bc@ZrG9`n#Ma;_C6JB?%U?2Pk$@epIf{G7<r5T|@My&rw(^&>I{eFFT^bmw$
zzzBgMDM%?uNlZ#gLSmx@esoEPe2Edt5S5nE3?wEwkuDjfqlbXBh=72AbUpjOc(&Jj
zv+dr_ea`1x*Q4$|>7jjpg`04(X{3m?SxA?ptsYxqk57G9G1tF}Pd4fmEuOuBEk8$e
zMX*5X+_94sKgA>U)fgiJ)MVvkYx4<D``P!TWMw<*kr1LhUn~?WjoHJ78j1+unchwl
zcxLW+zt|nQSR_{%O`z{Ds!e{6PJhbC`s5i4AwbC>!D?lT)zhNIG@-n+BSJiT?|-s8
z#7a-EO`a_}bS7r})1C3~^pd^ffzYAMaKL$WaXCZc^iyNVMz_R&MMHYxsp5lfFZRCp
zIE~@2V;LE4M7vvx>D+?9(cy;+(C|@6imY6dZ^n>4sn4!`c?L$`bM=Q)6CGY9@Z&*W
z`sgFC`|vt;Jo7>G-fJq~rMdR<PgjcV#m|(61J>rZl#*$_uX*2pOyl#_eo1NcE}8;J
zz~d(xlPH7-r76EYfAnUWDAGTJc&Pk+@7w*RkDkpXdkxzvdeo*>mF1PLf=G$dK5kxL
z9nhB~RrDu(Voz752O@I7-{8w<NXHku;UHpTLhDEqiI$#Dw8zEM<U?(Z2tSjspTLV%
zA*~cbV0JHAi4avijDcVwB%uw;Yl{Gz<Mqlo`C>vmIXu#uIollO9N;kFGInfsB`R}R
zK=4bQZPUfyi$|^su%9C%D29;zX_?g`Ll7uwFM}+{3^_QGIG=8qSz9VfR*v|)&^faB
zc|**BKlZ}_-z3sKZKZK^%H{t0bYtCUpaTqzpPgPW8NuA+j*%EFCr3`pQGhYLY-GUA
z=-_&aRc5G^0Fl}psUvWiWqX$<V{}?w_aF6%<H(RuN!6RxFPTL2H7?i{5#zs^h0GRL
zMgNgbUb~Awzl(dl(A>DJy@#-F2&@k<ew?JoF}C_%>P>&e0-In3nrzAm6T27s>F{^a
zq4HDXO3%vu_!k{0T;klds2c$7by-{h6xI#7Cx8|UGp+id`N-YV8TYhAh0AxRF-}CT
z3SG7#h;q!~WQ^Ca@qyJ9BF)xuc(hbmYTW_(QhKyT-V>h2?|~je{H?QBHLBLzj|nsL
zXQV1)WAggxZq?a(1LXGez+ZpLob$p?BrbQWE*6+C^1oI*unl2&3fwNfB5tgM85tOG
zk<F2~>YlX~QiWW-ULsIXQ1`9SlQc<!T)!=?>DO6LLh2Ux$;s?@A>Zj-zmBj`QjmLN
zT!KR77o)lLb8l09;g1Zkeo?K(S(F>TRAUZY?CoD){iuAJru9H2RVvf|iT1~Odt;xX
zED`et1Lf&F_2Pxu6u>+$c)wTWwU(iW2hQ80>G-TD**oW9$TR$Wo_;TJx|aNP5tEKn
zvb9uHP&oTH_JlkQ*f=6B28XZzk&XN0_3I6v1I7gTRZx>WYW(@eczhaK#Kypp6jzis
zVl+!wnCba5+OtR2oafqMq=W}+H<}>Pn;bX+3dsV{=YXvdpmPLUYku>;@M^#1Zst$P
zdB2;s{w_pH*&Gbu@ku+uEkDRFVS?%XN?9COpFWZDww@#*hl}Ew&CJcnyTBI<TU5l@
z<d?cwJx%q*k`*gGwmstMcK9mWeCX=z&YVrjQ^Y}LehgL6($D_$b1e}hPvycjlcu1c
zc*o1K5e{vj8(gwn6w!WC_~ka=J$ax0SWUsGmmlk{o#%grc<R`{6ukXR<n24YIBM-{
z!HjRKCy;k_cl$CI_(CPpesV!XiiOz|r6?UhV6<qF3HsF_wbhlS=&FxTIlKK@(^87|
z9Ykk@yyio_KLxelMvsVy^NmT_mjz@wp`)(7yI0c=0QYnB5d99-XvrZobEmMzYkM%}
z!<||Mg_IRVy0P;eeod>dY}rL~4jJO1mVP?X<Q%Hha}k;hsFfwp3er5*<}1N;<n}XI
z;Zp_Ca#A43LvmX9q)4Fb?h(`|PF3$G@relBzt7N<Td`4(CS>LytVp9dLG8QjwLwWf
z-VU~ZfI&NHdAsGj|H!T?!xTmc0xm8Zbu%<8UJ&>rRZ&Vqw+M4~pPOoIhE_9;oi~@C
z*hrLf*M1Pt6|NuFpLL2IVC2F5%WVnV@9No;B9yq|4nuM?LPkj6{G}1&0ZXa-<tSHA
zPyf$V&+jIEqbsua@wB-k(ifN`+xSLd**IGX=;2LzduOBaFp!0u-0ieLxahE;em1Gj
z(~Ap-`YXq(etDrNwE_w5j7>%QX)-?cAzD1Y%fc7R&vwJ!PLJIIYiG>oY_$U8HIdB1
zH`s=Y@I>s6Ge}^5DEB8oGpvMnMFh%eABAlFA=M-ee?1HeDjB0E^s(R2q05yvvE$<d
zd}NP)3pmkVsF3C=CTnjG?lJJxY)<vg`kjBvj_cPGO#GzH>S=oFYEJbvpGK`io0&3H
z8TF6PE<X-LF`5E$E#WC<4@j1zLa!MGnPzO>F&laMR&TKW!QM$&^1SnDb*LQIZ>swb
zh-}}5_((Q2B;fo)-Az<IANQX4D`vtJAUsq-E2;P}se*(ZjMJdY*n`Q#aGf$K!d|qK
zU_E9>mSEv5?__G~igOzC;kKyonr(XtS^A-<tkA#rXQ7`=2la08Pdcd=FUD`p2F*WI
zmb+lSyr`NBTRjg;-m-44L5#2D&kxqI(|RdqvoAv(a{m_JQOq$I%p7rf#`_v5QlAL8
z(StK9#6Q({K$Co2S*q5XN^WQwS=>aJnnsLE6Il3`hU1B5PD~=*neIhdTJ_~Da9^nX
z0;WdC&2%s_(cs1+Lq@m<dEy==Bc*Ax5Z0PY=^;J%NJU9r`47%>i<#$%X`pL$>!&)!
zEXN#!cQ>7;T+Grs<TT3yn2^A2Vbg*8`j14eMt2-O6mb%A&EZxO1n)P^@(p5qsSB`t
zuOZR&l+Z`0CAr*h|CN@Ou6B<kR<+|V|76vj9N@A!EfoXOe@%7CPkJ>}lrLc{iYf!r
zM$w}e0}D%j5(f$+&Uq2F_(n}|u3m}S4cqUn_w$IAq=D}OeEK3cHAVnO>CRNEu*KLW
z7Xzk)p?EF#7@LZl`a&sFO05?IgF#zAXpN0^25YUN&omLHq>~2`dk?nf8dt7EJu0up
zFZcVmNJ9;ywR(kcbG+7QQxkdgdn+bdPT;8b(#NRJt6>G5h1P_QRXLx~K2oR06Jm|f
zQW_{qGzbLdL$yms0Diir7|UNyR=S($9<aPSJ{@JJLodk+UB*NWt5Ik8?V*hrK~~eh
z)gq-d!4gtWs3~BE5a}Vd82o<cG>JgOd$*kAB+X7TZXV6<e`t{X{$uug<GAHaLbIq*
zbN54_#sZUzd9EWI7ovbSLt-H)XcWJzsS^dpvj-R0ya~`;T#UR8gqVD4Q8Z!1rBfrL
zZ)q~|B|>~ieF~x#?4{J71?m%26F{n)M7L3HVVaoy*-adc(E~T!k-a%Jy?V6aJjhXO
z0q)-<sy6zTRa(NOXVdJ)Y3_sEl5SVz;{`559m|cjl4}n?hx*FUC%E3h7W2fWAWn`~
zAK)lb!2pWvVj5(?B^jTP_0}u7llSkqHe@$xk_8~;O@Y%TDOqm*CKV-818%<-Ki@Kr
zYJ9T9FH5BMO%O)ReE%`mI&8Zs&j@AH5J@?(^K__~FnHDSYWa)2D)@S?mI$6m32t04
zYQC$)4Er4t&6`Oa(|Aw0knX{l0PWy1Awpx#fI+?o<$+Bkr;kpLBsIatI=lpue3_}~
z<g^>L$henO?z)E2gCw(eQdcD5Xxhy!JLjO&U3}Wax7OB}qV6J|v;FqV->WKtI`$wg
zkkfnTvE#*^MgO-YP6g~qBy*&@sPK&|jrD!<4%5PLkyUO7MnyD;mN=nrYR0fNaxtI-
zD6UZ({Rz3a#;GbSL-6U0;muDq(u|Qz)v?!0r+D$WaS20J{ZnNM_e~$v(XEv%si~!<
z)!VCuu3L0TU{RKnJNdiH6LC7j!|Ry#F4K-ckNro65++k0UhcPF=(4<tsA7v{Avjjo
z&S^hlKa7cle7`SHuPQrg0b;3Tl5=4PRv%xx#r4P1eaTD@JfgFqIf4I@NhTXy{{>PN
z4^J>Goe3bCng-af@)dri{K@N?iu^iOXz_2_6U%YMPSlN=^ngO0yjY*zelf9J8a6Xz
zEu+V~Xr44s+vT8?qyL(T83LM8x^6Q*Kdq0wv%z5z6#o3VEQfpotJ6J%qb6r3&}vp;
zNeU2f7fguxl*Yf>Z8Ua&&4b+Cv->kORtl#^`8qMGQadc2a>FuBp4H!ZRVZAMgxh4Z
z4Gq~|EaJ12b5MfS9;2^)YQx7sE8b5d+>22V-sla@q*W2;Inv3%GX>Y)oso-(wmwr6
zHVp-?T<=W}1UN}XS5P7HZ@3cBU%Q)@Ve_8xNlsa}Hu+4i*J2EeD{xZ3_J)A-r3YSi
zgfu)!L1g9F2TgxYn(>}*dNJlgrkR<Tw%R1ywH*F@>E&qsQ|n1y+rc^~NA1r2!xmCd
z+g@khWg@Ue)A@Wg50fXO13ZkfIkKskVCqW$+x1~0DnXLS66a(sw*jDQNF2W->t0`b
zR&4XC06z)_y<(~MYD^w6?|ZgT8N<X2qVFQfqv_!ZkZ37Y@Uv)<WfG+hy(Lnc-6@9x
z1l~HF46JJ3CGJnEL`Imy4L<)3&jZ_tKFq~-k2osj_=*b)q74^7YKHwB@(T=AeXPqP
zePYZ|%~LwM<$iTbV7YP;9l~8jeqW^pbnoh7?|{%WX08)YHFAcEE1hORrBPHwwSgAl
zElCz1^3#|b(I>?K+of=1f;dFQWFAv2co>UiN_>Fmh)~}u>O1_>Kh<AZ$>l)oDtSwM
zVLi71lK$@23+W21+B-P)zE^O-J)gxSR9-WU?07?!X$Wu~`(G|*?7O^6s1^|{)k-G5
zPEHe}zXx8Q>VG3e0p=g3(8hHJV54XmpHV#i(()`_&|Vv2v4)@Vy%8T?ov!}!gDEcO
zO4l23!Cg1JIsIYT-@?ELUWcHPfkHWMJ1)rA%1WxJX=T@V6rU$9>bpbbW|*itzg$Ad
z_g@hV#!Eou9p^_mLy_(vfAGCXc78?$RbEwfu!J4v9+7L@LKdchxrW8@;G0F=&vOzt
z%xygaLeKGbKK-^+n_GE*9<~7Sphr15dat+6PlV$0&SxFBZ4fs!ijpYXcm!~=Y;evT
zHd(iQM#3n^YxbL|&lzD?&Rh<6{W!DPQi>nNlJ_65U;RL?o*2I*;{-`}7D4nW%ig1<
z_;F^6n;`Pm=a&vSeQ4YbHO7i2E9+ZIEr<F2E1Q-<0&ij%DiW2Jhfg~%PaDF*Y+H5?
z7zm&b)h4Q@@8F6SYZ0fdhfi^g(yKzf+@<7n?!w<-r-19v1o2%*70T&a55>EZzA>O`
z*8s5z5;|^Gjdxc`QBElzW7I{}PIK3I)CKo&$UWdBGpuf^()Inh^Ant@)i4=`65Ph~
z<t~{zjp0Q{zAgWdUF-F}vK$-W-LexZFK=GA*gB+qzEqdZA+1Wa;?ijGp&ZoY^-3&9
z;=i#TF&u3L8?t{P{J!IjPn3ClXO7#hra*-bA`+~M3Edf(W_1!<^S`COS~NwB)kUNu
zDhjB$*+g5FsXz6xwRTPy_GLV{Lk;RgU9(DeDjnMx{nK;z|HB0TXcH?xN38I5Te#oU
zf(cl--Ft69$I1EQ5Uh!j0$CM3MLbYcY&LqQ_k{U{CT;6ne!BwalRAGUG(c-G-}b`Y
zJ*d}S{WCAd*$JBzwv%R__b`0H=%?N-+Lp;pFWb+lY!iROU9f1m(+LS6<a{L~mdA}j
zaO~ePo0nBfbx3&is>!dAXa@VJm=5yR@}`IP_V%_rPIy%1c<0I78L6RaKG^fL3kw{2
zU-Bf|gQH^9EeB&G|G86^0+OBh+tY@02~TNFn2m^(Vw7KLG$JWg-)p*IGrWi-%E;=N
zKFaDZQf|0e={Pko8jqy%d=rx@=^ho34sk=rGQmJNDb~-0!o{gX3sae?zaGk5283mu
z^oYI~{vU16@y)(iBR$gqd6n#B13@9Jy^#=v<M~|vm}>*lla2;BY<e8lvj=y-OC9OU
zf#*|U9MqJZ=E})sUj@@X7Nd@h+P0)NFgOrODKL`JqP-2GFf>-&s2=?&;$%%pCsM=u
z`SWawj`Lm(PaMC?V^Eqe_GwzGu^ZO&w*J`f4-*A8@BW?G|Gu7|=#LzvNy>Bxmczj5
zasnVQaOdvfz7iAgI_^5UCiS_KlauF*qLFoCPBRI*z(xJW&X6^6<ap{l?2wBZRa^e?
zvt7{kk1mMt#JK6j(IS-{7g=RhnrdUP&H(F@{1?7ON6FVq6Wa#^vuk3EhnC%ACUQ1$
zl8lPyb0;SXsvjmdSk^-#mX7ZZC8iu)DX^KiF>9mf+7Juy#2PisY^o)U-v7481yY6X
zN2?xDz?9kG)KNS~QRGh%Tbc8A7y&L2qFkg&kk*qI`i@wsjP7v5?eSe#gP?qWtow&m
z7zHKuwtkq!;RTd-ySaZ!?Q{##(P5<{|E8sFikjQF32ofC;t|OkO(#{-1tV=;B{c*l
z2<c2{7XcvZH%3eGSFZg!j=^7bU|IRWOtlyWxxLD@Y;{C&o$nRjDz6iop$5hC#rIET
z4^+2|ERv#hkMi@8ztL(@Y+j`t6cHhgZMj^CXdb&*h+&3w*+V%UY=HBCgEaO6Um`CG
z_65B|h7^4kzA~mQu+bmPFX!E}<1fvgta(1y0m%Ob@uM?p%QjBV&M<z;P{x7UuWX1c
zwho2l#2%$@6|Z6o0%gIdUY8t+QGYd&tKIoxtN~d`@v9viM3+z6$xKWqt;$`M<=G{!
zl_MIt;=E!0K1T<f-eSbppG9n1BX*L!txlJgZIvSa+Ie5{uH#ZUl2#E$q@+solUvIe
zpaulv&>A7EPlqettz}eOJ+-km*O6s#x^JJbYXfXDM_e{<2CwvNRGqB>r=_c<z1bV0
z6&0d;$7hS(bf8GF!SMP2J$`)W(ifUWP#Fni-xq!IWC)4-aL3R2A73`)3XIY=+;`)j
zzePHY^Ph{^uxR?P!Jc2mzCAH8kgx@Q;9h-|!KdzfDCB+<#Ehk-*&u$#pPuT%`ZI#I
zygat<zrU69ZK?S52Y%IHD4EPOAC%>WgZaR);H~u)gVKjwOm$+tVBp_ivQ5Map4aW1
zw0$;rM9z|MbK;Ul%v@2tY)=P}1Jj4Cv$L~P9%d8YT<$B#D}g#@oAzR9IsoR|a1qVn
z<@%3T+&3q$|Lzt(IES#riCA0CqnA)VdC!@zdEl1OCOWZy!7zew0!xRT{J1x!hWHH4
z!Uw(<pC1FVb4)rPP~?Q6YhpU&AJKP0If`=XG>q(`!JWD^)^CTjmiMsgYO?RH!ND8*
zX0hP<a_5w>XHt9eGjrd+1JgfV75cm!^{T(JY}P!sLp84kW_e4X@y&h-HraaFv08YG
z_HLf-kGbH}&7g~;=1UjP7nY=()F9QLyj`iM_FchWTZG;4Vm@h4m6KU`xepyplH#-y
z{lCVlCLpNLzMAaLce|(y1RB1;euF#VU{BMkNzGx}w-S`wJ7Ypl2*9KZZF~uYU%ZzN
z)miKBM^sD%GR>YXk-tH$u_7UVk4A#;KPPYqz-lM>&L03f8GCDJquiU=zvnI5hX>r>
z1WlqlkB>(rfP*@Jr7?WkD74i|;x{G+a@uAgJ@~%lIWg7gHH4b<Ypw9(U|{paCiAO8
zc&ewjwM%MerVuUyux3Zo6m?<~&0+9?%#im1)1n04#SmTRzu4uEG`tlXwKX7LhBBIu
zuBoFOk9aEz``I8C23fI@K5FWAFv9p=cVzHWcKzUvGmzbXkR3(Qoc+x$f)mSrs$X=X
z<hqjLdq*sbn%w&#%@p?ka)0}uW_+}IvA0Pcc;aa*UcYe#-XpYIcxyqY_7$_+9L_mZ
z)0xIH>Bs4`xU?j*hN0N}XD@m2R!1nQ$hUuAU}Z|4xjPGq-oXYm&oYszJdyN9Iy7nt
zQWPNZb0n;o@VQ((Fi2t?f3lRA9g^F2ak61)Y3Z8^iYE@mlbwSxD`gZQNdw{A=-tKR
zp9h;WR&M}*sor5@#=Raf-BP4k0UI|D7bjg78#JM?I7Z5b;~8Uow?S!LQ|+_V0XBxt
zOuMhlF}*{%Ue@2Ez>l$Le5P@<54jR9CUd314<kau!IdQa>ogZGa{ppieLICk2aMVM
zHu~$>Y(s8l#eH~%)Q?*G(zzY}*GBdhaJ({0mkjX0zZS<Kra)a+MCkX>G$aJI{X$}z
zm)4-B%&y(#TK|syjU(31u|;9>C6O^2^=!9c{`8{#B;zt;!C&^4qEJd;{U_y2qGr*~
zl!x+vsTk(?R9|(u;xiAqhe`_5b5mL0{=+N`JA>B)xhk1y%_VGQjj#5u_V6Xe_$cdk
zn*m+t-(d$!Ng3INe|ll@Olqnkh>}ap2p*4@L+2__xvwiQ6}7fH&opP`0gJIa{?Dr~
z(bwK!MXE{cI46ht(UP)_JFY_6+1cHDz9xen%8Cy;cX`|V4s+`WZdj+HjR9yZZ$}5q
zD_%-t63PzrqkpY`eSLkglLFNAF7{JC&``U?r4IftF}C*u>14x|plQVLmW6iskZQ#P
z(h&2(vb6D@jE04KY_$OO(h3ubNlhIpMNo6U<)2_@kU$HeEX98!s-eNy!Te83b0iKB
z1l1>>$jeV{uajCEU_G9s_PrZQ4-<B%HHI#0DBROAb258E3@`5LU{fDAofQVCShBvB
z?zMFI%2{l17DYXc+E$BBDPCn5***}K9_}+6eW;@1?fBw&0GqU*2#zXO${D&zUnr@a
z>dNc%H8BS#BmF(`sp*IEk9XG4zbNvEC8#%YucZpjsKA&yk{TE^&I+f_!DC>~X+u=6
zM4pq|o%9-})&jFXRV)?yobCXtAk>0;b{0p{e^3)x3;AMwddOn~tPu#LUnz8SWJ;yi
zp@Q;W(pd5qkfkp)Z0D-?!_8zdZnY?|`Fvq6!@KR0+}{VjNS^F4M;r;wT;61AG#Zt2
zN2?L-QR!t5L3YQTA2+UC=Y?LSzPo27YicC_>zMFe1L3<!KO~wM4TLGHq9iHl=?ep)
z)cQCUwLZSpAG;SugX0uXRGS~<8Z$+ber&)R;t3~JC5_pgtgug4Bh?a~kvsOhAc8E?
zJo36Qr5<}tJ}vYFV>0A@7(+by^K)%&v2G&Gz;rwztigMB7GYacBCinI)Y@877LyQS
z&lA^HS^l5gy5Ob~pzmJvHYUJY{1weGEr*bBT^EO5bIiB=R)h0QU#rP9St5Hs4O^S}
zQyEoM;xGdxp|yC7zIzyIK9Sb(!Sx0~T?9Y=eI>+z%=4gPK)=oL>S{b9cj?pKJ@B%6
z1orga#@xCJqi*n;JHyRiFu}oc{7#!(sf3HA?qqBwi-s{!Cj1KILyS@)2llj)If2a%
zk=ftI$U&|9mQOWCHuhN)HK1{75U^&KWaUO^aKvl+wW`W`c-!H>m)kT6p|M}1-xRwJ
zu)CJSeYP0h@++2-$s-dU9v-G4)8mM=()sXBJelA9BfL7u^-wzX8Z@DR_NlN-q%kyq
z())*;oFZ^~RH_oO+B%P(&M~M0WV3GWQBtsdTKaH{qi9#-PUiQyIRzDglp5UNQlf-1
zu(C2qlJHPD+nwv&**U!UrJ^3yi;|ZM_g@*n3h$<8@DE+LhoUu?M_j(gX5JrVj;{0o
zp^^gKIV}n6!i+E<C?yq`P%P#L`nHY?%9UU`0P;LNy^xXHDxrUlZ7}|?3U}F4OzqLK
zX17rwpVU#YK?wY@sF88<Q2gFhVWq$qB}=^-UsMhS2@I24)i83wi8!OKp~T?dIk<cK
z|AznRo%5I7sPgEm<?s71fFkD}Np3k+8hqOU=V$0+>waBh<k;iCq>#hSNnp1wuYHFE
zY^a>B2%V?NhzMM#1&zIWL_D6v7B!?wmj%p@V;Cgd_Nudow7$!b*fpbk4Mb?zSyII}
zn5kJ%kz*Y2e7qX;ISkwtV0{}c?4V1hjFkaJ8kXb7kt|>`C4;h5hVZ$}>F!lRrSKrP
z=CrUo2Ey%6T7{c`-#d>TrJlO4Zvx<liSnSjLhaRi`f}7vEIPt<#B1<KvZ4)ZF}i7K
z>b$qJ&t@C3X6lH4ma3qQ$FaF)5qmdzuSCn{^SkTBSJS-Ng{Usl6^m`$&+CcG_%8$!
zaN26>P#{Y!7a2vx^d0+xmzqSW_r{OaVvX19X%dJlSt|nSuW&1&<M`N62Pmtj_$?||
z6JSSAadN5>A0H1c2poN}Nv7UBA(AI+V-M!Tm;c%Q`@7ifE};ZSHIjF^QsS9$<Eb^m
zX!gVTknIVc%OkDwcgfsJP5z$ef6l(A`?k;G1{7RPK>1QlA30y;ALflTL!P4y5$SCb
zl@ru3G-$O_q=F+d&1MJ++*2ld=L0{<7Rr^EUxCSGf2(Wx?Qh1q7EZp7h_RN44`^i7
z&!|DM2*jjJ0Scm>#f($W1feCd1C5&L;<iT*iyVu@t{}NVnUcx$RU=nPu`!?OK54M9
z1mGR-yv&b{DfR`)nX~z^hLYi*6_Yk+BSV1p+eO%9Ied6UNDDx6CK!pn#@8?wO55`L
z<~y0JgjlDpg%j`Yg40atrQquRCse|GWI$8~DBA3XA}Xzh_Os+`v~Q~K5Hkn1-ghtL
z3akqyev4J)x`xedaGEUCXaPb{Uz@gYLeET3pX|=z2F=Ake*9=msIm^T-(FNk{k=Fp
zC07I1C@@*!X3xuMBDSs=v}LH{ocMA+$t-lXP`FYC=Q4rJW~+7OP`UDAUSeIx^%iwh
zs<i&tU<P>;6>Ur;*FZu_<gQr1ee>k&m=7PEZd(JMVb3GJKobi=$fN>4qMSq1<^koZ
z7F~Qqhm`M=JSEI)Li_nlyXAe_K{)RhsSXGy3R?q28#>+<2yP{J@4J|RlOpey3qfzZ
zd64+6oTNaK8F{u*^kFaNyfkXt9<@>aa#_3IEok+muC+D!a!aUvFTS5)$aW_@bdUJ0
zrH^&5K1H3oJK7hXKs}0Nh$P$p_^F*VG}tB9|C3Yj6xb{*H>VY)rbJ{wk&JK04JmQ>
zE`VCq<wY_X$eP2&Ew9H|u%XBXR{zGArmxSNe*@NsO(nR9aH1mdX0q2DJW|cdzfZLe
zZ)?4!dn>c%{|tL?SWLe;eTipS&RrkMPcG1@;c|b^C&{-j?~4PiAy4mb#9SjdOEm4H
zpKSxrcso-eWp3^u_38-&vPN}fNad0d0Qv2azCD;fV_g|%4yRVl6_40i99rG0qBmib
z(#V8q=W;6vU|@W~!>)%9UQUh@ujxp2Wg6lF?h1T?@=;0^qFLZMRFK6LR<^cYYAqCR
zH0<P+Mf;6ui8NUflDfc;>9%7M4a8y<!WUqzGX=%L{E6gI(rtm2@amc|n<-#-$pG{3
zOp=UTD6vPW8-0<Rq=o$QWl>W{%RnfmQq5$Dzsm<GByT7#aEK)K|12)ktj&elr3X?V
z8bRRw)9i$!X*u)<o_aZklU6lIDAOZuLCfaJbCM>N(ih4uc8fMhG^=U?Js%?<Bc@I#
zy@&FV`HigAI!UHRA;R*G3*hgChGlY4VW<YV36nQ}=RJJ*CBb9c<I_8z!(CGUV%giR
zo=IEO_}s}EL)OG;SXfGbpPC~CKmG{4Pq#7Y)nb?>&+?ZXKEM?ZIPQ$`MDRRKN&^{t
zbQ<?*mm?>plg56?72{$clpqm79alOZ2tF86bzq)T11%K3;+W$MXw(Ql2j_U)Ddj9$
zm$v+XJW+D(u~tct^bKk2NU%?kyX~rA`LzzaW#;C7E1F5w8C_oK369fGLs;K2)tRu<
zxeIFJiTBNmZ9O<ib6iH&hR7r5H!E!!kvVssl&;42{6ybT<WBPb@%e;o4(v;8%+;$M
zWATL`9ZXG}>ZXo?{y+c7y6@>%O5#3uMYSQ!e1A?^iXc<Sw1N8=kjdA$_%Lb;hr31U
zlFZl2iwjUPc&3<4aI$2a$OMwg`SW(7uwz7i<)c*j$d+mvNgVdR*VMa!X(Sn~R-IR$
zWaQm`$Qs=-{_grS(>l`je|2Re#v2@O<9PcdFaM<sX%^K0)a6mTxWhJ6)2jqB)ecsi
zQz`h^U5PU%mzxDm`3vFtD>iiZpO3TqjL2F!yo?sUirG<`bJNJ>d976^CQ<8A;MjQi
z2Ue097A*w}{we0}W}bEhHtN-o+a8z&`NOFH4zCqPF6orqgk~X+UCO7z!TEl*SE`Ft
z-AvN<DX#<2<E7aK+g88RMBB@C^Cvti;X9497gq+yq~7o|(M61VrUrs2#Xe^|O=_%e
zJ-_g7*=yyw*tb3ZW_t_6b$CL>w{O-E<P;VUil*RX7Seo7jY+@iFZ;5H|KHOb9-cpb
zn$72P>u&Le<~Pcu-rL)-3KV_3BW-s}vO_yDwSv5{=HaHdLz*6#9vE00c?`Uml@er;
zMsY3+uNb?n@~(2TDVOM}zny$_#d>I$ks_mHjo<noNmdl})`Xe@oD(&?cl%fDo7zgT
zTkz>7xwW>#3-9w@;M7|wpP4x~E{Edqy5~#Zr;5)QZz?%zLu0s#BVSgIT&@5+Pqu!X
zoUL^?14SuRh*X}7B{tA)PcDBu<s1r!wtdor_FPqF(EVvLJVr9_n=yl3L;ISBD9MOH
zJDm-3T~B_Q8)n66cWazhRa?@Ds0PZf(KR>dK!Lyt)#JhPa4_tKs^cwH@FVgfB1n}W
z*G26>OF^0r?qUef^F^l8lv;3pd&sT+c+?iyuG~86=S3h^)c%|<4h-mUX+&*P(OLQ?
z`c+lBcYQGa7A}p!{9zlrot<~}(Ccd6titT+*}_FBv-ibp&J!T?&|Kx==gW$ozrmhP
z0b$OE9&(L`D@BFR`7C_LW1#F*Qu~c;y|@VjAk1?^US-{U)JWpWj5P}i#gLG3?yL=f
zi_nN=$;i}&VF%E<<gVr8(ai&4Y;vA*KPx*wHH)6GA5oqzBcP$msq@#el!owb%+B9`
zC-~Ly`P1VeAcA}^`AjCdajS&>T{ys58sCZZfoP%<7}x^ACFJIqkP?zaRu{NPS~x2Q
zpZz?ry%+@2UUSaMxl?p~_HuKN1n{pJVEEjUy+YwvO12SEFH}H~um~{d1E&D$H~ES&
zA(N0VRr2<o)YfU6nZB7gh&IKak^bd7IYz1t3|`OxiP_@?Lmfy8WmmtTB$@fplK+%g
zz!JJ+O?~BjemZJDD`oUe5vySPgy76D)_cDip!*eKd3iaG&RPmNnC*uEZS8w~A`KOJ
zPyV^$KeiO#PZs<*R#c?)`l-#&Fr+SccB=oa9koKXGeh*-{Q8Lv4zL)A$zoR9gVTl6
zbTv~|Tf2FzpEY(h<j0=587+txCAIPcx|imD%NZrDBZBI4xs9zSr8&{0J9jF8hRN;J
z!NPjt-03biIp8*IYX-;wJBhZA2(Kh5v-vhZI@Jn%9P2>SF<H@5C|v&j4!GrTE+Rgx
zOV03tCVO@Ud*RQeo<yS?Ik~yYp+|4-$z~-<JPg7bUz2j9|G2Iox?9<YH8T=r&Hbyi
z{qv`z{5KUSqZ!F83o^=)gh`+rVCRFfphdF2mgLwpkg#fTK0u%>`|Ypf36>f1TD|$z
z!Z**|3q}V==|E<_aqpW9#->(8GpejwLxZ0y=dMKjc$sqG1~|C~EOSb<1fLTanlr?!
zDsd|zkJ%+&HI=^)nhRMaBTV(llA0Z-SZyrhO$<ZJCs{+Ci2z)x&RBhUd0{!v_9i`y
z^gy;Lr`VOl0GJvjHke)G{e61zZi^TD1?mvj)Ji0gG!ZwD?}cn5rj~v#^gGNSQY<o!
zta)sK{6Wv)ur3HynY)Mp{O7u#M9bwCU0%f5+f=;FdjDLJ=#cf2J>8zFTYQda-MEZC
zKm3(;(NkRZ_Sr#@*NYu?|Nms!)trBgNv(Q#dZzgq0AM`SjRCI#8Ldt2TRbHDmdSb<
zYE@5fT#y2;M1)>oGf?OiA9lR6lMEkrKi(e}1<sFiN{lqp!`ePaoHx;U$#Kb;Mq%#P
z@)|mjYXKiuQ5xRJgnJ*0kp|Qj?&<w^d_WG2=`x}4^jpH3P?XP_C}lbzEC?Rsbump6
zcLy{OyjOQq)J83Xz@c#MyWMh>phqYt>m)5_D2F@ROxji~b==gNG*BPP$;tMS)+Rq&
zLyCtOueC~dXmIG2#N3X_d)QAn-CG>1i=jLI$)AH+O-TuVj$GLbtiZR091}Cb$XVA*
z@&@=n`|n7EwdEmnSkrh*K0aK9pHK5XP#b9=c#m7%Dxpw}vy6*CJ}6+-VT7(EB~jiT
zR-L<gtj3Re0_7YM%nLv1_mC~RFD&y_R6{C&Nt!_RxVD|nsvMn4Oy!-|skryzzU|9^
z>>K9IiL+rxyjP08;giS!kD?R@=OkHz0#sv@%$8J~5AALw9KMtZy<98kT?T}`R*u!n
z))!-(6*tT$ACR~tS?@TY(@C8x+44A<+m2I(;{<8(N<VZWl0=<SWi@ik(D<IXBn;!b
zQOTnYR&?0p)8$0aso7nH<f>$S%UZ~ka=0t-)V(KN-O(;VCUMP0D%*wO0}rf=rNFc(
zd}xfEvXNyX1(YRqqA26Z`M*JbGd<#Ko>}Z22~Fo7M`(y@nHAO8T0n!Wth@u^H=9=r
zM4;xa05#~5g&B>r)!=&;N9;2w+Z5jw%3e8Y7b#mGJ*V06TCYR%?k!|kwSjO06D2qc
zVl@77eOfNR-iD}YkA5-l{Nq7j<DnF}K((5@2{&}3XX!+7B&u8hD_UFyMMW`BfPGQA
z&Vl^-O@ByN`N;C3qnwh+;QG&>#jfuMGd>#fNBoW9c@c7H&(-Uca!CkM{v5d0*i3GT
zh2++V|GQ21Ly0_$mLw!<{NnmC)d(RN-K_{zlp>@DB(eB2sZvvbLehmE>$<o0J=goN
zO_w@SD7;&%jeHpcyzyZTju^WSR>Shcg+=Q$vSezBqf0zzbKB>ehZlliOC|6!ziV-E
z>TQ%H)*gBr3|%ID&QBX3xLaN_v#A2?hp@yw@SdBS{=tfdf$4kH6L`@`+(Bz|16kqj
zQJHmjfG#;NFl;tuxw|`MS=2l)0~r=&n;J_pmes$ZfdbxO4J6j~cPmE;_02iw&UE$*
zFwm<)-qbjw-;ttwG;CHnJ<u^t%Y}g^1x3a7(3Oy~r?S~TOTq$9rcy(txyc2OpdoGX
z^aPZ!JA_c`+?|xP84<K{kzvbWsuFe}_IkN^LP6!>%=anJ)GjarSJ+XMsg~fp_Jbjs
zbN;h$Fc5zK_W=^e3bF%$_E|3z&0Rikp?;+tMLG~a<lUP8#pnYYW8)`6*Qo}ESII-C
zZ@~)gjG-3<-CUjrrZ-QQMq-+^N>6Q{<X`?y8PZz4jrM8NmUo_p>Vf^u2D=8h<X%El
z!4!5$Bk0i;E3_XIED~gg=5%18hQiQNkzhU`L6`!h;4V@Cf^T3_uc#$R27>syK&sPS
znr?#8#?!nAOGLYBEM4VQDjt?+Vv^G`0m^w8xKNEBIg*u+|M*d$P%L2v{I&$crF0Vh
z+`m0nSv2m6RRrxi{ZocXu}HCe@wXpqyi`pn?<Tnk6CMXWa@*~%YV%P~#a`31YHI;z
zKD;-)BGsv$npC8Q5x;UbiAPXpqr??b0<u#Fqd*oLsnkS}JEqRc5Jlf%;hvu)n!WV%
zXRy=|n&4XCMy3LS;Do8^=>?vMLo#PZ_dw-*>hj0N!KVg^`mh%gUIORD#gUQz#gWtP
z)}YWpS+2Z*cf}t>=?ur$KQhab=RebgFY-b~!of!iZ__`1s4!FXF~}k)eGa=MBuZQ*
znWc^q5v6zegCB!OrlD8}fq=v&J$-LT%lp$xxTsd?@UtrTvFDQ6lCQGvD%zON3fe<<
zEBit;qv<q^uIY5kVTk`FjUuNE-l-<kd-D1ZxEVU@w@KE%M&1%r61aL@4+vHURlAm}
z{`GikHUXc+8(l$8lCu~3rn0%WRPL>OEfTiW6S;4kG4Pug4KB>>%`TbyJ_l(leQ&HQ
zvVnYDRy=Mjr9*QFE`cnSyJpV+`YO6hp2vXfhBretqprW~{6VQL2}ePf!A{1UDvTKl
zPeC2e-qORw;L{4|GLPaBfQj-$v12O72B-@q{i=l-*iFIh<MS^5^`=aHp-hA{1#aGn
zmV7xw7Vq3KRa~gABm7QelOV-jTpR+_Qep%0%->ZmdM^_|@4gXqDB!_G@5{xNhDuW@
zL50HN1LWV2OldAHBT=;A^2_abCb@ykzu{n(F8(1uoyVpRA|PLZ)xGkM4>cdc9LHQ7
zBngQ3tWNBxD4r805Qt^9Wnd#GX@6&}e~^4sP7m0e(ghx$j;2*O%|f{USbgqOhv47U
zmcRV_oJO;*N3;sq+ub7CUiJc7>>+#E^JnB_6rU!!r%Be>+R=#=JeA!aYi*t4*KcZ_
zTSs2gV&0n7MFR9#57}&Qug+=g%>Rc8ycFp+np}nGLDl8!0`DSnGmO`AW@pRAYLCAn
zVqko!jtJ7`;a;s_ZCSQJD$jMeXj!h(6mT)UUoK{81l8#DmnN4QliP?@2r45`!VzIJ
zd36=8BQv-gv%LIZK3G(DOj`Fv=K~TQ=Y~6w%=7TyLyAa66&_D-ZxLKYc?tT_KuKZ`
zp3Tc{!O_$nB#9zrb{x`*o~JA?my3tbi*Bs1<Jk=2Z*PD=TeF9Avthe^>`D0yRCD98
z6Vvia9Qq-cmcOGF_TzCRn7i`jpCX;RZMmcR*5*z|zt)!iEjY^Nq9P1$gg|=GrwM>$
z!b@LOj&F?%aOlGQY7h*?q7LRu;Qa!91itG-KqT;%pp5UJ?9ebq`W_}q9g&hR^3C@R
zZ4X?R&6a4|c&On%Fbnd}x4mZLSi01&W4_l836egM_YA;MnF<Nr)f4Uw>@Fo*LCo9d
z)`8)5^6fL|EkBtnKCk~#HovmheFN-+3p!vaM){5iY>K~B_BSCRbMdP4{=Bmb-AM@o
z<9cm1EnJLuUy=c`-&`@n_q)E?ot&L{(LS?TSyi?D<*{wBR1tSI6T3zMZIe-LEy==x
zW9o)R?#gknwEV+cxy)L+A_aIekxecX<QXHyS?Iqs0!$*oa`vXgx=SCV0C-^2k}F=@
z9Mt|}YIY}FIolv?Ho`kVpR=ynuTyyHY3xYSLjBEHgmO_I{i3P9DXWlHJkL4v<t&g_
z+E`QlF|7_Mvcfg4BU+Y?WcREsgri&;$~EN^IB_ET;Qt1obi<=`RtP$l*M(mWi)x<_
zbZ8<rs%ctgENQ?9mvhPAz@fS2;~CJ0oS00}NJjohNphx{QEoGpFnlwSCMsWKH(WWO
z+-uZ6#o&E7Qg3K)3X#QB+WqMEU{k+4Vo&Z5(@BU)xhAcZnBYA<jL?zYun;wSQL6Ev
zI136yHZGacs;yh{^X`KapKeZXs-;%Cpc$R|e~~*C&%fK2Dp^efbfFPk4bz#@5@4xC
z<c8Q_x{D+jM*epwOfnkv*^;NLG$Y*617p9!<v3n*D~I3bPCjJL_4KG}KN{6_E`8*V
z#=ny8lJf&W;8KLTDY75rsD~}cxh@0}>}1+<d2tFqB1Mk`zsoRfu6xH8V8c2#bu?0>
zg-B1*yiE*lm;qd0Ep^T|pFYXELS%D&|DXI}I62`ENuSKX4FCMIk_E2#I85rjSt8~c
z9HQe~C+E6P_hQ)tVQMUb{d(Nb&su=n2oDuZ>~EM4_ry;gcOG6I{PXxNBr21ygedTx
z)IEt9y{CQlCq*K#$_Cg^GuKPI9Pw_Wd*7pSIe#(vU>*oPo*J1Gdf*Kw-1=gFBP8PQ
z8Gd7I>g?~){h^In1~$4FJaF>l@_ycH%c%D0&FyU-JHFytU1<TVf0>{HPm-CLIiA$4
zNDe->j(P`;tub0muK6G2Dq9a=Xk*lxXKk)Ayk;7*eX*HTUfw-CxXVsmGQ0Mi=>6hC
z<1KNZ=MQkv-78wLF&*76mh*LVQSLlx`)q4ugauBT4wZdqUA(vE818#WNlgxKKd1b!
zs9>n|!mj&Sgmg~pE+IRaM}^hO2cG{_N4B<1Dw_0rA3sqpc6e}*YD^|CHHNK?{yH#A
zRVUjIdLalp{vvVf;^D}S#3bL2d|}26ekr(%2|(t>4lUb0(Y&k}@|O6Oct;{p>3?!-
zinyWq&o-tS%AD)}6>QLGGC8P<{6ZV+fc&9w6fb;OAx82U#*&;v0mF!mF?O!BZKz-v
zHTrx&g@mL6<zqM*Kwk#rP+3vmpO{fx%>*kv&VGi57$-%OJ_z2*Q4(w#)i=6<K$rrV
z3`2V_<{iGn%#9Tv$qE_pnKP_oOo#{&1vMDTkTg;&g>(bmadmToM%n+QscERIJ0`eS
z0n3fI@`P1>f}=W;p*M+Yto-WyYlV#gd&I?KpSMKpQyoMCI1WTL_S042)<i%-8a0HR
zIN`usq$ai~%Rg)nS^YJ#*mloM6bP2R%a{*laj5@$bTYlr44lb0{=57GoY}S=6SB`I
z^b=}s-!K~92xQJxZhmoiaD4o@YEs|1?C)98^4M@Pxm9DybRR{9+N^Vq586`Vg_;;7
zw(vtpJ>%<qq6N2)CLYXLi5bUnIq8WCkT2rAV6+UF5*NTl4Yr>ZF1pzX=*iT+y{6`P
z3*#(O{TjlTCYbu9L_}TKHRx5L_+T74KOsZpQwPC~2g)leeW7px_ae^9E7t#+nG!#1
z&S)3~D87Ir((h8Q>YJ7~Rev(BeRalbTa084Xk38acjE4WMZS?zP*&6&R?`k2q65=K
zspNbMpBGB~&}6XcH*p^YKu{3Dudq8PTV4Inr8oa-c+|TJ^sfci`>tKt#JO`ZeF_wD
zA}Tmguu**Jql8i$&Gwal#o?h2I$3^9yw56^$sk01FbRpQxCNR<M3x!-mADvc5B|51
zYqnCPbx2-y0#uk%hE$sOPafEkNk^&%$?>J7LYetZWe?{o%E=p&xYOzH8nP>xvy{tU
z?Ou2!GRv8H`<Q#^mHeJu3Ce-w8UV(+rc+Vl>K-q|+Dm4=eq!*X=(7g>LlwpL>K6s&
zjz)vh5~X!n3m#H=)-BaePj59K_xuA-N#AC3HVsPso`*|AD_26|*lEC+oIHGrqw=j^
z-D61x#^ro@L~&rt<8p5Mf=E8(Ku-O2_?hOEf6YQfKI741vdWLoXFt3z2tn|?2l_N3
zoRwa{-nWK}!^>scmXm3Bi65UY&bNWf-L#5ap2s52d;g|7$M|2%3&q$p=cp&iMvMZV
z8@#3T6s(i?7K49l)o4BZJzOZ9mxHltKRGjI;}`A$QvlSh!sN!36yfva=5yi`17h@t
zl+{s%H~$?E<TF#`LrDQ?{C8i8_z1fSumy_iX*%3p>0y|0US1?^VfRKd{UTibimbj;
zspXiq8VoK|QgR(dh<d>i<Q#;0d+op9;hflF&IDu2*Td(MpIMs^2y-7H<)^fx%l@`p
zCgTrJCIGQk;u$GB=xjIW`3IA-it-8tAweTPnH$)*D=duNky6zJs_qa_H}t5hgOTxj
zq(r!Puw6%bWsP2vfMu$S;k6!FjoRj^s9Eb<<>jI|3IXfI+u^>e4^Efr_nGxJ6JIn+
z93-|@A0UVJ!i1gpcukFRFqqI}h0VD?{?U~+W`Lq3YFS5jqZ8qr^zGHUf9}34$$XL+
z@Gkf3MvBlO_m6Y`7k>M$Bcd9>aE7Ud9DPdu%r+nDU43lU8v+$MhNaUQKnqf2CWh#B
zC@yz~IHRdFsSCb`Y_tXtDK$YptddC}C2|Qv4Gu|WKjaQkhZ9pZlDyEaV2FUIz|Caf
zRwv5t$^ux8aLgys#Y@>R%n>bl!(J?ycjL<Km^Y;IC3VgU3ea`2F2ezWoQ9-8f<m>i
zIaJcgh`b_TiSF}|jUaD}2e!{{9!oCR{+aV9uK1BSKVFAtEjW7-+>9=cw1Crop}gGx
zrbY*L2kf+KUYt*`g?cLUY$lV@fTjQp?E?v*9puw?oBfG>|8=)pNT_Z^nCbS<vysAQ
zR=kOlg-r6G?beVxTCBC1JKFBIz;QXU1Z`ZEGH~uMRfC*FOs)HqO%S4J^SBrA%9^wk
zj_bi>o)=qUcy&u#-!J<<x>?drF7@9z?(@Gxm;V&oj}<o~?#yDK`s5Wn5@3nNLb?=V
zSd5Heyif;k{Kw)$<yVE}Paj{Ccdu2_efPP%TvG)5p8q9;&zDP_l>zHOkSh{<x%RBa
zQ0GorHk^6-?smeM0cQtV(Vf1_;ErGuNIdV!h9=%}?Z_YgmeOpELz@4oiO9r#L`jQh
z1M)uaez@7U5{&U*T6vm}1;Cr~67{*XsV=u7{TRr%0g13jKKb>&nHKrAJKTwIWR&%_
zcty6n(%G$1tUD_UP9Jd}%|KoGP8Lk5SypH0sDb&S1{n`{5ttzu6v^HRH^hpf-q0ko
z|5AF6#B3#Bp2VyC*)M#e+&;C5H%mV;Gm4uU=}gGATuxw$*Hy!y0ij-<G}N25U_t!)
zf#P25@qYWEKII&zE<JnlYb?pU%BCk3t*K=$VNKqDB<@zuZ2r=a{c{*#?Mn;{BqD<1
z3>;f+yp@OQWic<umdJ<rtKc%j>X5uU^6}{UKjbFq<m99~o>>HIxk?=RWwP30{NMIY
zkyo?xLopnV8}l0Ic;Dj>gJkKCiDd()8{R*fg0}sejwZZcG#foPda82NY5VZ<Y;V%b
z%3VlGK>JIMS;h5UU(-S0V(8*{X8UXjxIDjHQMq_>xu<g4nP=M+jGf##@44J_-ZWFk
zaNgRq{xRiJF08GPA)&H|uX^F>AC7)=y3u|z*y-c*T)K&zyzA<`NEz{Xe$mwv-saLU
zgE7R_%Lrb1Wc|Csyh=}yY<f8-TiA8U>>4HcEdS|NI_B-?L<qD3H}UEk8Y6(FhS8~-
zntFPADu?X;jInKN2)Wpe=ZW}@VOG4`$(|fD<uUDwr)V^2@O!RgkIpHW6u~MgDw6p$
zN_>gNfk4x&q2y1`T!|&gQR3^;>wo?tQstf0-B+$sE6IrqtxDKDM&9)i&(pVbtuVum
zaL7N3v{;1}>E9M9)g7!*<3*2Z$O1<TA8QoAg{5O7?U$WeF^kujgG=X+*C*%g3WiB0
z+}a5J(q7G*uw{}g=yf9Kc3cgeH|E{~S5F_69FPo=Lsjk(<aLGjxI1F%4#Y?(LnbiG
zfaFbDVY}!S_w{KvSKKF(%=^UPVNP)ty@#r|u@sL)0tUYJ_S)$z-?*)3_zH8o@A($k
z7vq2eFQcV7G^Cg*iPSKDNc6RJ92Ik>;a5uMw+m5PZEr`RY?9oP#=jiOc6qOn)NVB7
z092}xEolKVD!0r-WO9E6k#T$B((7V4skrvt>BOAO^u(0owF0;D@=8pMBs!;iY%N?#
zIp?*|*w|asYcUN>9Py>G`@$V4g>8*UbF-7ng`kK-+si)km3)o4%C^1cQ)2;Mip<nz
z(ZmHR<P9Ys9A?*YFvdF}4nkwdZkibY4Ew`RYn#zysd9>V>rA}OQrdUgRNkRzI{Dqz
zVPtD7Wjp+E{*BH(9ZXPwDHSiA-hqpSOS2|`8?p=5HDz~)Xc&u@6-x~uk{}6@GWxOf
ze3|M-Z=CMA2h64}?|fIVWKA<>qPQN7@NSy5Nxuu#QdOOaj${(~T6)i&S6cyKwJ+1R
zxiJ{ymGw<8<Q1+y=^&}B_T^nxt>nY{@};K`wc@8K{lzo-peQOkI0T)X{G{Vpn2|$d
z2=xd~O;%8&CG{=I@D^i|qTN%JD~bgaC#nCc+zsb};D1#MLti~P7rsKlqO->C#O?5>
zb28vX3Fj~J%!U>&1=B!;ssdmK-tTg7ZKa3(ms7y`{P_OonZ#tRCv7|XRo<SS46O5K
zM`UR4inBN@d0BF5Cd}$ah-rrF^ymO(SCUwF!ui(rMa;d@LNh@g9-+tK!Tyf8*}1uh
ze@EJ1=xOtu6LEOcB7Va6kkWejnBH$CV@g9y{J-Ez^5LUvfrO&5J%CMI7~65GInd-a
zQ2PbSOiP8{N$Ku&<uK8ytF#jv9bgQ5SeBh~-V0MBx;tLASWc>KaxFEINFq0CcqDM1
z=(^fj_VDG6d4A26#ABb&KQ@^kvgnkH{U1%|8P4YazJE%I)~7<XV$|N7s$E*OYqv)1
zQ8i-kRkxzGf+9u{T8bK}+O1hDb`%vwtp=$XTT%S)@BcV{9_2XXNl4zg@9R3x*K3H5
z?nrKGin15@8`p7pYHr(cyf}Zx(LO~A6yNlybeFo5y3W0E{2HLFi%`)rD`aD5+uVAw
zvmJi<m*e~_nB!OB@N<02&Jg3>w;yV1ELk;#cKfNwU64l6r){{c*+Qbmu4}Nj3=2(X
zQ=?qsbPgv6mGV1C>wVS&2DL89kL$v;F#`cX`xB$ZyeH^|i!;-j9sFcqGIY4^D$^B*
z&oiN<Jc%;F_cg%w?nOq!;eYh;UQJsVysj@hbPy~DICcJg`=nNMsG=flYy0$c^-n-m
zRkGFE+S(UYHoa1(;tz6E3J9^NNzK`+s?yNV^WG`NxuxQ`)Oyj{>D{%z&(440CWH=D
zPfjEtH@AX<Uw9TTY=xyl=#>8o(pD;GJ2U$Z!ita<2pXUZgzD_#O$K2u5QmMI#!KYJ
zSejctN3ueJ_%M>tO03jg$)|#Xt8uJiZ8Z=sqOs{L{6<WJzgq{Z&kc4#t{i)xPvw=p
z2S2R_Xx{%+s;u&Ykn57&jDtF&xR63l#xN6IdMtYs=Q{d{2|1@wNdwN3HP)eQE+PG?
zP)7EoJ>d2La!YMl<1H2=W8gtRjcXJn-{4?|Z*6*r{nu=Ulai_K2s>}-xDY{fH&&I_
zO()1muN_*TWX9Pz?wB_y$SW!;C@6qs;OEy3++eX!)id<aIE3){-7$L2puoVj;LDLR
zqd%AbBqH~Q8xRnOxNl*kHXbBY*Aj!D?~Y3XGU!)ZC*4PgUmHM_*0GETUEh9Q6FZXW
z)^B<`C;C{8G$&egkV}mP#Gn-%ePk4NoNot55}GJF1?@31@{j2s!j!onH?bPfo~X|?
z`mk5qqCrS#VvQVDp)&IT$MfK4w1;qIW>p?9Qc%h=3l){s&;+_|q%;x5C|;ChDuDVV
z-&-+=LOp~f36>&3jG*<rLR0when#7w(6Oy~(!^EYprF&AQ-L-_jDctonZu$(KkQCn
z0Vnv+L0;#{8~FQS);;dM;GVL`HC?wk7~w&W5H}<JwH#x1_arGNYo87Bn<<j~Fd>P0
zieWW9IMhee)$}pe!e+|lneMwy;1YoNMpZWNvOC1K?W2sQWo~e>2ei^e9&~eD9*oFI
zl?(3Uz}Qx3$>chNo`Hd8D!Xdy?k|5B=kh=NR*U|V+Ztn}<C3Qj8HFeR;TJBG8yapQ
z-`->AIoLTW%dY_o0?dZBqKFS~^~y8~Fx8m#Qkrfuu4Qt0qRYEZGZt##p$}=kt)e1?
zQ?q#N=&L6(ZH(%s8qIwx9(uKUyjncxwHv7^?DuB~4<S*f=k1GnN_Ul1+_mV*?00By
zl3M*)TJ;H035)2wLa?U9*5@ligRV&ONRbM{?{eA>Kd0m|NN^YSp1eL)eK}Bgu~sA+
ze%|SQzUqCx*LIu~QPT|6Di-Q7JL{m>**Be&1~U3mXWg0`z8<!zUd#&=zQ`(F+uN%}
zrQf}+b>$lurP|I(M57EFXz1}Ma5jZox)ib%dU$wMRef?ygmk~O*C@EFd4Dh3-*b4H
z%@o`|Kgby#N;j%$XlpzE!*vw&rQQRL2%ed7)srM#2@P0%om>QH3$4(1A@9$G2OO^7
za`AjC)+pP=KOxkWV{8V(zGl;E69}%Jp5W5KfgXt^RCqNuH7$e_J&`Z3k8K1rJdHNt
zx`Ipy7EdH(?8<om$YhnEcKEgR!C8g%4yiE3(7dU!5cj9F%Drn);EY<cBC;cJ`<H@f
ztgq$i$BtIM{VHIPPOTFoRb!805M$MqbhtvMkV;d`NWV<)@(Z5yS@kXh7bQEneY7M$
zM1h2Eh?45<Q6$h?EH{|o?!w*55y&X0I!!4=i5T$YiDt4XyD^f7Ae`7ros^)}?+}fq
zh?eS%g+9kzMJ_$wY0@GC4Un(W8JwyzYb~=Z&FkHez%5_Di@m!Ke=S8@IdzfMNqVib
zDCA||>iT4>rw;4mYWF?w!hK!QWbP6Cxrx^*R5Ep*j`&=Et1&!~=lTicp1jDqhAqC~
zwpxA0ZO|^pJhIpNQR)WS8-MX@k75yI$4^uAgjQOE7n6h8!Gees)sy<#+P&es*-6t~
zusq%YN5J+MDE9+*`dN-8Rq<z21*{?7nH4m9mX*?ZBgFkLbB9qaYn+glinkX`m&>Y-
zIUg`R;n(e@$h$tqtBY)L#)h|%gP|c!w{gGu8`O{8a@Ai#qZNsgyQo^hMEUo479!}L
z{FBbv%dCZ8*PP^UBD)a}Zc?6P4~O>|J$Um0kh$Yknz75P5ky*I@`VxWc<hiHYmo*@
z*xX*dGAqZR=8k4RufP|4>y7pP7|Feqzw!?1&}TwjNdtO#XNK3t{cV%>_xu{&p29aS
z#&vB6mwX;DH{Iq^kCF6ahyE8O`OX3wtIU`*_AZc}Z=lL6(WEGrGnPMt1`IuSOFo?k
zw2T8*#|jO~Vz0vBn;ylP8Oe>Wzi2Zju$}(tKK>^0`f*IgSm*k>4+3$M4AKkU51bz0
z`^X+ZR%5ZgT^4Wn#ij*yPX-2-?0<gsCBP9FsYGK3q+7+zRdH_Pj;W#AszIiQ&>Lpv
z%uEe+$-d3vvvVmFACl8eG|=@Gl}Sv%sA^biM{WCRYqk4ZULl@g*a|tH@pqYI=x=16
zp-vR^aXKD{Q_%ZP&^mjZT~Hoys}!^5NUzC8{t5XXz2*$}hrH6S$BPd!JF7=Wu9r2!
zb-j*Wf%j2usoV4O^It<Q{!J=1c{f=~30Ro-x)er?Gp=-;g<p1}=gx!YPrg|I^}gFP
zjuH7sr{H(K4eDfX{6S@|^xp~9vkBJG`>HI3)*WFOIhd;t+CYHw2hR(BH?hQ`sf{}-
z)1<>NH-@H(WkRZ|E&D-VQt!nDfywrdma&WUo%?5-4PzneLc#M2gK1CHBoSg*Hm7SQ
ziJZ3j;m)RF*y0hUPpTn1JA(~?%P~9YwF4!YE;BJ7#lXc#`4CFEKiyif0YY4q{^w>=
zQ3+c*^{1xR!g$d=iAK91o#8?J2H?Gh4(zqSv^C!E1w!QaW*%tD>YY=vi#Xjad7Mq-
z;J`RSy!$MGO*1o#y_ho*q~>qZkLmL`ujKE}%X1o*eGez`@9)kNfH?`3j<e2i;LPq~
z0yzC8B7yK!hnb`Wz0PAcF)$R|)171_f7qiCBPEH2IiXC{U}eLuac{D9xt|GLE#c$N
z;T{y!TEsLg-sk#THOC(-qaP!+3j!_vlc(~T=ehm*Bt#s>MdoY_qYf>h=o6J(*42F}
z^&4|!7Sj~_xr6W!P^88&6WKa@m}8*)v|N}`JzYqYG2XQ=#6qa4xn|DJR=idZ`rs$m
zqORrV!_5PEh?_uiv5bbc<)f}Q6d}bM+)7JEw<_MD!g<NAHjMisjPOCe5!~u1-tu6a
zl?(C-yXm0$IGi*3YJ)pvDysb7lcL<Yg%=@_m*cos`?2-_*$&QF{f>q1nq2#Ca^}Uj
zmYg>`8=I&YCRwvsMn71*>V8=L<<cxk#5sfafRab;x1^CY*$xV-<H+#Zv~H2xA)jYd
zuoS$z)tQv3t;HGXt+%?ez;V~HJ@H0P?f<B%8rZ9j@!?Z4nX)WgydRA!6#NB@H4_~2
z@2Ww&3QPdTGT#<-Ddx6jKiB3R#ip9tIu`4>wesil3qM2tjY-tOgLc=G*Tyo|mUsl1
z;9eQLYeBQPq0C8JZ@2hY-^iOxpJZQy;!Un$`TKaDftrT|-<Hgf%Awk<KsT|EwO$Qp
zx+jurDP_~|7CLi6H@-ls$i*2jA|!gg9O?fmayWqP(!4IW)MEQIXXK823YZL<Q67Cs
zjm=VnI3Xe2oWnf4nZE44^kG6mbPe2>=fLqo`(Xx$rghj<8H%|$^Aj@xw^Xw7{@5!(
zTBW0{Tz`Ueybwur@nv)MB*=3zww<nv%xIIKPMcMvl!*`hdleNqi}0+Mfo_Znh{}pC
z0C@%Lo6kb1I>x=chqokD-GA~9*zILPAT|~~ZA^`!5U%wK1p=-hKOfw8WjoX0#3Upn
zZYL=PdpRJ%ugXl6Dcq*bmw@g#N#lAzX8BamDb}9-O}&ct>?p?j@;9bX<@7h+`{J}I
z!$9KNc*cNyLf1s`i%D4Q8`F^@3lv~FxppM^8)VmpI{m`o4I28B%f>l<1QQA80;#1L
zPC|!k&uSamA_D^CVFw$fDH(q{w~G<k5{6nCIKL^_R+}RdP5%}=Pv4H#i7Gh>28FlT
z28WR5ZI;iEj<7M*saa-IFimIUALCFuY>s>iD0TzEmV>3`*)8|gBp)3CAhTvp0a#nw
zJ`ZY{lhMx-E0TQ2Ud!#>Gz(8l?IUOyf5LoXK!TO+!STgxQhP_6|HGa!xbSO8k&#ge
z>c^mbp7Zq4%20SJyHuirMz0KOkP@m3Gph39=yX36e)w7>qx%IR0@k36yl7iEYiq44
z7Gxm*H2BNX|7hqCbq%;5JrJObPA!6Quy3n`_?%5MRa;deU)4_-2i92d>OQxG-?(ih
zCfEfdgF+L9n%E^Z;}_pZVf2MV;O0?eV6vzeX@o^e;VGPv{tj$L*b44q6!&n|G0d7_
zm-n_?I?GlO@28<*%ZaBZy8d276REJ)96h8l9_7FW9n5V8x?l>GHDxYet|G*!Ql)&c
zS3tW++JD?&JOlFlA=x(`?8H*7Q$()0prDmb-w(@Mp-PP0x}XRoH8rIme91ESd(AwH
zjp7_^6g;0;sy9UQ?#*ze6(9vBFSHS}+JRwVmQ%@fiVzqNSk;TdzbbG}_C=XEV8CWp
z_{QdhF2G*C{IAYX)y8VtU?#!klbej-P03aFn>)*IH{=v6V^sges9HOy#1CMUIBc3F
z<-Y{aPL&NS*illEL(=k&_V%#95A0=Cj9fqTU<a1y<qe^Zor^0|kz4;>xpbV&kzo8W
z-?`upUF((<M$QovA0s<R<nDYKGUYCBpIG|1R{^^?9Puu4_u?WFo@FVtkuP)m8v>JQ
zg0=8T;r!+RA!Bqx8Z8;8Ow?S%k6-*7u<0op@jv%#wKGh4cK;qW<dvIDyds6S*kSuB
z_w$#-y6{#1X<nrBevOwd=5eQXE|oD%WKx^2!5{yY)3p%HdP3>sgIlvQjejUIQKm`d
z0MOMKvJbwR;qPJ_W6jv#_b%&r`)gU1BquATE&fu-!98DkM=aq*IU!L;n4q#@Ru{Cu
zBAHcBeP1ap-S$E_m^RRPpF()%7pBIXl_{RoJbP1=C2!C6|FQ(ngx3A&Rll6sBb+j&
zf5I0YP(c<sf0q%uai15gEO)g5qv+PasBiKv`8Esl`37aCrkGEk^g!d`)JL~2Nn}yK
zw(EbdLF!SPEgdZC@I<*7jA|6gC;V2&Pev#!0vBqaTTu6RfGq3Phl_29T&x0^eW-0(
zSgX&YYLsgP`73lI7a-34dEK<?Kf{ZG%agv##Hahy&!jhd%<jhh=>$T!lDD+zT{fq^
zUu?wC&08r2FHOp|tDwc6wM&{Bz$A#*HC-oVUVoo$Ion_YdnBS-#s_0s95(Vmc~zxg
z0+GNLBk)t-g<xCV)T$g`q<wGFHK=)Vi~Wvr0?ox{Hyx<0KKk>wes)9MLA%pj!z-Cy
zZ8<anQUaO#8eSMY3k>#y<I>O)Z@>?Ma8c@cW}5|eDdAbhn{5-Jwql8nvZGbOZC*;r
zf=*SlmaO7|^Djs-nVF{Q_7RG6zkVIfm(tUve@;7H{NpMKqqOdMnS;)kC+=C>wuUS%
zE-wE4yV%t`^y)`U<bu_cAJP)KgkO6z_|4r;&Ujd_ZWi|vkd+(xGqrm$BUR-~$l=@R
zE>X`yaGq!j*~IQwUks7X2T1>vF5fVFncgvS#CQgWvG0I`h3M<vHc-j<0pxnGhYV|~
z1%r??oYSnaQ8LJo9n}oi7Ngl^G=2ahgK;8-_&<0FJIM(2gpWc~i+{$LUqh5YLt=N~
zv<^@oIx+xe<wTFQj>Rh?U*tnZUUIwY(!3Yk*OR+SHA<7D9wqtLNG4y9&^lpSLg@@s
z#`aqh(Hh?*$<zu=RxwJCU-;oT)iO=E6NO8ZuZS1b-$k)or4AO^&Oyk1hh0_f=g)`d
z(`E@5e^L&1+WK|7NWRd}868etQcqZ;xgwL7WePa4l9b^vr@*b-6QPY379@hqR{a^0
zyYeaA9RIM`8p9QDIg_L*OotD@45pj`&XfFP*;Bz(3BlYJ5JU3tiQHBv+tz!DKzo*K
zncIf0;L8fZAPb3f|6;!I5-jcBydh5*B;Tb9MSMt9i{!peHbg_Q(R}_sJ>4O;{OmV?
zYnjO<WDIO9{GCZA7Cek)b-C(rTe4V?tF>UHp~*z(!v}vAWHfIA3q0<<cC!u?tJ8#T
zv_#g}R^J`Drlt*htzrBlaBeN2qSm9tZB7OYi<YXJU(=xM{e4hSGPY-FKluCa0a>}?
z?IJ&%VnP}dQYgl`BqEK=dh31&RN5A(Y%c0|7lLgcHd9h?eQ|eJhuj=vc1POdlgHhS
z88{fj=0BFw@X_%g#Q5_Q$k>W`zr`~Xg@}xSZufYS);RvzEvQ!blnd457|kF8_bT+#
z<E0-8(M)7DKx4^oQgUV2TOq0QwM=;yPS?UYI?j@U1P%)t!L15HSDmr+cpKQZYr*4~
z-X4COy`vJb^A};~Pxg%1b&_oxd9vMRWunUt4#vU{*C!vumsSb#qZmqtJ!C5@Dkj}E
zjO#l(Mq}a`_Q0(6Hu$3&8L>8AvKun2AP_tJyp9u+pFPt|AP~&TvYnHXl!#Pl{4;5x
zJ@@dlUt_xt-ajWTkEtF3<qasm8bQJ5Cq=GJPCH@FUsVflaYX)I8Xq4&KRc!=u8!Es
z<7ho=q}N$;Qy8fLm&^^fpyO^0ROc6R?Wf7cS?AszmmL>_)&JaCs2;W?3lV&!|Ii84
zos<Q)e`oi`czJnw@{}!vto3qSj@TW?b<67_@fZUu#i=gQA?XLe4>A^sy-AF|p8bo!
zY38Iizy^~yW;>tsC>Q`PyKu5x9x8e3<z5Zp@oCg8sf7Br_F7#@O`L+Ew{rNQkBp+c
zd<b~C1hO281dH@S`d0T(*4ET1jlWI5o28<IDN}J!FEs&MR^j2{!5})%mvAz+tCQYi
z$>KDLv1+mn2Tn3YhnnZMr;Bl2sMccy0x>h4C+FS1^Lwa_usc4Z8_G@H{ZKY9uc!+~
zueQ1$%0i82nbxL%E$V609&s`sLoJe*$GsJtoxQ;wPC9Qon|oQODLeSIq{dbGZQIVe
z$n{qR2Jfk<aUrx{GO@9oq+zhq9<=dxh9yopMA^dDRARO7ZNZ5&kM>PnsoohDC)B+w
zQ4o3-8-zFgmG2K|_HRJ#)jm?ELRGr=6-l3|>2O?z5OUfmDOq7?|B1WwOkZcZr&Fxd
zO)%C=F7P?KdA%W2sxDg|dX){J{lcUEr|Q4xetA@VSg1`Ix_$8%{A_3bd~$L`+FNiX
zt~h}&BqUO4GsvG1Xe(24B>o*QRNSzlVLYPyofC>}Ha1&}NV})s4M|oY9W=wZDoe)4
z$;YGklT4Hx$#|{|{)h&d5{F2mqo)R&CE2#8|5k|{5r4hP7QRNTFjoS!*W;ydut3d%
z_@d=WbGy(Gl5GLeHy>B$80g$E!73?CW*n{TCu=pFN>@O(hn*g>ViHIK313lh^NN!w
z<Whz}D#!f7*_d4<APG}9W-Bqy?K#gZiEPS%uxJ>MgLjP0q6A5{RDnJ$nLIY$R~BhC
z1B1(g#+zgc&iqk>+gF`_nEK)KZ@hkNYDP|`p>9?y?e(2i=4;lQ9`wBmm*Qd@&2J8I
zqR`k(Q-k%*fSIQ=+Igg{tDwUGQD%O2ryMc%A0~_OCWeJeeXKwGJ^fW~SSb}l6!q&$
z^16ld!OF|-j(?qmw^SQ>n%T4TS2=)`cYiM0W|K~0X?%Bo^C1?;z$i~A+h<cv8upN`
zi^b#1e#LvLHuA)fewU;H`=4in<{j(#B17p{sK`k7iT{2bos_zL@X#!t@G4V%Td=*E
z4>nt>KhY{^6cZ?jzh-7;<_U8vD=VetALA>P5y9c1S>9e%zKQ6dtss`Z-D03E?s4}@
zZ3P%7w3E}-EoY$s)$?<l-!vz;g5;nBdwY8xvS(nA8l0g47u#nYrz0K5ptv)q6m}GD
z47;}T!|Gc5@Sd#;=Q#ee>c6qni}m$MME6qqQ00<O`RAQ^m!^%<_K+3qA<=6&GzC*p
zxp~M`I6J$b`g36tm!-Q6Sj$V)Nx`4xd^hxd(eCRdkWys8CKs6bL8ol$tIyg$L*R5j
zrcH}DS_srP74i;%Kvxgh1{`p(>k_lF64}9I;)wdRk!#EX3EKVLTO_J_wpj<hT){+^
z5^(YzqP&U>RJ=oJE0O@+GiaL8foHMbn+p;a&%fMU?ThJTa08k~x2#(OH|n(2E#0Db
z;1}Ri@H1XbO6Zgc8P7N_vQzLDV$1qV0P!M5CefKEJC4F;a{gw?kG(y=x1+D1&ZhXl
z5qg?vPMvXZD8@1txqF&9TL_eX@F?+xy@qMqR8|R^%Z-w@?;K7Qj!Vlfq8P4GNecL)
z+<9nf4XkK^pRC!uHC1a^{3|6w++gTR0L%xO@Gcg*iGIWmqr1ujtMWI~-TF5@IOY_a
ze-Cm}N6FxV*&n>(6D?I%o2JRSgR@#e4`JilN?f{@6z!?m#rXT?ayEp`dq0Y%3ThAg
z!q>DRu~2M33mtpC$-II(gp8BU<v<3*_Yh@LiUOqZLWJLDO&>Rw26KO;aC$$jJ98y<
zGSQ6(ss1;^@4A95yK(RK+i`A+L1Vpcs?DmmF)(>k#-lGD*UChxqraJ`O$4!AefbV4
zd2R95m*!8%!~LrKTjTQXZ8xv*e2(?OehH+!UlL!=WE<)$m8TN+<#;!aL>D(?%wT~S
zy=a#Ls8Z9Ous9u@vPxY3^^Ww5JgfDph5XpY7_w|3HBi1K0V2z~Sq^LopC>&7`!X14
zw$tdiAHpBUM@g|aKudTNTkn}<2Z<w!C((u~^hj1j{;{rzQ?gtABnwvnmgDM!cxl~)
z1_MZ^e0Tiohp;|hMh&D?tT9_(lo4l-y`(lU@ibs&rPVaMOe{szvHs=9dp;x9>%&}-
z;(lKPnlxv(O!3;#0`w8}`O1p;(StXS7OS4%L5vX_c<|CwqNrn~t3L<*R7!>}aWh(T
ze$$xb)Vi9Ap&&sVyR;ucx*^8ox4E<wKEqI3ntwV@M=!elC0>+#vk(sSQ#Pqhh>ik6
z$i6{tKKL2pedl3;zCkG)xI-8tb)4Z0E1St^Z+thY<`KO4(C2vD2IE;acb;i?V!DCy
zEp^+>Q$5{1+!zgN9snO%arK<8<_73{%YO-28-MG=TBXGLohSBr-B&5BB?4(9S{u}Y
zL0MZ`T8e_0P=6xM8Q$2^GP@SA3hcwd;hE@!ZnkrKO=B*v&J(75e<5dR{K!ahaWRiV
z5IDX>BL&=M6crUqqwrHxRNWUIvNyUzWYJ}A=B(i8f;y)2yd{{rdkKuctsc4*-YI08
zved9wwxHkBNW)ZBDUm4X`?=4Oz0ch}hb2@m{<L%)D#>{-yoiYKK;8#B+V=nLkHRNl
zlMpo_hs;B6q@`!@Mv_we_n?;hup3Mb?JtgTKF+2-24pjs+nfPDAwM?`*`m$uA_v00
z1e0vbnqo_SkLT;{PY3+(1SlyfX&$EO;!eRdV9XFT7O^lVq+P^rFW6T95tuCKkYBd-
z?1krutLZYFp=rJODKO{^zfH+Tc31ZmRHFT$Ya~^bSs$xWEcu1WHsl6RQ}ZuX1!lBR
zaYe<rB{)T~D!cQ=Oewj@KKS^8)$a11H}c7$hb*5D)t_E{I1ShBBn76{v<kCaqp__Z
zet4s__37fdow(S4|8eUCYwQS)N*9X}d;uZ;6SkAG$+2(>JOJDJJY@`%11ASav&~Rs
zloXCx4u6-v)Qz}fA+?&T$>0>7b5~E03U}3hP>8y+P0rE^!>St}OvU4KBa|q)m+y$E
ze>g*-7As$g9>ZBs3cgd99UuDYxoWv;A?fmrac6go;&-A)nD&cD@31(?U>gPX4Kj$v
z&zr><f2@^006DZ{*cmlK*jM@DUiM*vnmp>`pE%P-Wjo(197&T6HOtiC!8PEkmCc{l
zjkS&k;J0o_hLqUMq3T*y(Bickrc=(jl%5X--^6PzDm&O#QoRhaRC2*cTgvvw$TntC
zN%9~a)TR88S2&@@@5H!rlrwKWLj4L^nlA2+0lhuIuhx?R^Nx$=hTB~K3y{-mD!=B8
z5;2zm8=`*#WDKm@1gHhL-4b}+1LG!6V&xQe#rYBr+h?r`^9|ZGOcd!$W-JXO2p@tQ
z@6^{$bLZ)kJwPl7K88d2bzs4^p(mLO=S`QGB2hyFt%op2-NYgb(N!IW6wTsZ4+A00
z`V64pf)Pu~mAVai$oz;*cfhxujFgE_O}QZ}E4zL9*LEh6C*`%xOv?8||Hs4r4A?I>
zs?wB^U$phaoF_RY^%W9e*C{$-Zn+hL74^4xVxbSo7Vq1uhrq7zW248j7}9vl&3afg
zHQrUxbz8{2e)7Ji52-|r*H(kjsl}Eb<9LAR;>O}1Q`fBwQb-rI*YHtBHsuf0Yv0H$
zibC}$@nMxuQT3}C|Lp!HU4|YfjaF<LRbIvZxN?OOMOSmEKM}H{wo_YR-Cp~WMv~sF
zB$7x}B?V2$weOv3aYUpyT7?$K>F=MidPgMO@=bY?NRD2DfmvVhwHX)~z?`;%!h!E3
zds^HdL1@W6=K>+VrluzFH;LTftdJUo9yV-}!>7xS*J?uQjpvng6a2$GJ?|k7&@IP}
zjg9PEs<PNBj70E^Z<-jyv9=bC{~qT5`T)cyv9hveOc3nZLC0Gpq0;+T%c1Ww-*s3~
z4%7s3e?)68NNLgWh0|3*{~X|a6JuwAL^wy1IL@6!8lEKDmu;lSN)=U2-oazf7l7Ru
z&>fxJN0=bQ=(+2zo#Ju$Q(f)Pb&8{PwjE8)jcN)KgKD>&0m;_L^HU}|cJeMm&ATMg
zTMyIKYd8~Jl{TfYzDnRK)j`B-Fzm&P7qWS;9Kc~?TavlBx6RKiz%-PKcbN_SqA{P|
z$Lz6lKBj_&=Q(nSY1$`rCAqpi=x+#V*#{_JC?5)~o=Pt2hxs?gl`%$3PU7ycv4JCl
z-MUfl;0#=6Mju25`mD|Q8_tZg*7b}1yb>3sJ$3vw4(s*^p=VVS>kPvkr^$A7RuoS6
z`Sh44TyNHrk(F@)fZ{8aYcy&GtT$;`G<?g@GKW=e*M!xg?}X4nSP9A@U=#Z`^oSi<
zlj254+bWPluP5bbl0{}Qz#{L9Gmft6He-!d?$u`YY~7E7I?b~nc(3%p<bK}tMh@R5
z#LTbDzWM&=b(GtnuS%%k=M*)sr%w4PT}JmzHn1_hFsM90>!Ueen-fBk>Oz^Mx=zR>
z|A5_0C5HwrDmW!qNxgHFs;f8}H(5696DVH$capt(@|X!Zma)6G=3073+Dv?}T;oHU
zxZAN=xaX>`fDZf{?#-SqIB4=PeF&+jn1QXd(iDG`akd?_gnQC(b}Kw*<WY;}{8&*Y
zTdb851X8ahfGt+<i$5cVUKe}6u&XmiRR4XJ)s=irzx%~E@D{h<;ob_}pw$idu!W7>
zxx)f%YC{}@QB?}71$LGfr&&jD1>T8H4;i<a@c}(pmV*kQJ(GipVYkPLXM@&wP_Q>g
z+h*w1X`bBW-WF(4-5VSv^o0{~cC$Z!wuT+f904=3U9O~yha#3=eRkE#rZx_mqd*4C
zq;65XLD|F!HzIHbjG9)B`8fx!4P0C_MH2E%O)0+g{>!1SeDqwH|IO-+sM~6>`I^(*
zj~1OL>IpE7r|N3>h6zQ6D-=g$D4)cS4OCCo+cL4*Q2+np1SU_zn$Y#RaD{20{_9tr
zHei4LmNVD5I>e!M%53#((F--^Kr)w)R0K~wDb9N{ZI@aLm&N%_^}Sap%Ay0gQ=_Qg
z``crZ3E+FOBd74_^_D@Z{C&<G8C^IL>09}DNPr-;!UzenXIF_#=NAg%(-gfnajr$(
z-)2?5*RWB2u@k3i0SBdqYuq6|GgRE~095p`$>jX(yh-@w4h>hU(w{zGN%(Mfc6MKe
zFKow439res%*<s)Y}6#I5n_s~DkSWhQJPW$^^=CFT75ZO>Yg^{5Kc&sAIzRrr8T&0
z@{V=nYOY&FsBx2Jf~*PRt=wH?!c|X<|Ga!`v<%JcnyZWwT1@dgGe?V^r`s3kDHr$i
z^0-$h=uI$~HBP3<CGMB392$799gXB=SLgGv)t+$cc2)GdOfYLMpVQ4pu1#*hT12`Z
za{vcl1()xdf8vcE(2T?MgmRR0Di^BypX8#X8FqD?h)K*VA&Bsm%NP5YEp(DZ>%a7g
zLH80QdX&H0L5tZO6<9f^AP*?dSg3k;QkDOa2;L7PbZ!$4x3)rU8^-Yw#^PdPh3wnL
zU1|*W>gz=*Ru2CK=9j{a1GA@;9-ph8%wO)#AgV`_+(SL@iX`5M3u^KY2>7uTlvg@X
zr(t`GJHDwgIPwUuG}^J#wWU<JNg{!jwBkv34mRaZPAC31$2~WrH2@nX14NgGnc2oH
z(HrpXn4aAd_he@G=l}MRpw%+?QCF_&&fVKhZk!1(6kiqiUD1wv|BiezRKxW#C0et>
z3?;vTmI_R}yV-irE!|#AP&4+5+HE!qo(3U>6!O-qbPx#ksOm~Mv8{{}VRMBuI$rBW
zOkaGIKh|{zLP15HQv;YO36=;wr|u0ftf!)l5jD`ex7x_qf6sXf&7EV3U@jZXzvqpE
zZQWPfDY+Q75K>19)kQ&i*{h^*O0iJ)hTlc~g10KNn%&;i;oK;pep!#taaP_Jn;5wk
zv=^f~KFsMpT}b6@Li*AAo-`*L^}}yhbjxKEsef5*)L$Vb2LZ`uczMrO@<jTp>{?<1
zPgff1uvW!w$NdhmY@C#T)&dSFKlZwUAY4{~j8T<;JD8@(Bi6ZvZSi1>Z^$1glXpeJ
z6=0zLEU5@<jF4icqmC&q&Zw~Hp<x{*nr6(Fmgc9HKGCx<bCxk+6-y8>77Uy7j5W^X
z#lvZThG{AA1B{3g@^^yQDls4k?@nQg6rsuc#p$jVN9+EP;i#yF+KzPIYvWi*Y4lB9
zxhV-T2|A|fzt}@o-^v&M(jY0ZC1mZltJI_W^*`X-0poLXZIj@nWGe~6ELItoPbHic
z|4NLR=2Nb4&ja+${{*6p7SRurO?s4W8;R!x-e|oGGA@>xc^G^()JDR5e_49LZ+%wa
zR%W67`Udh!ta|@SozbMO(X_$DS9jFCPh(icMcRpPF`YduPRvOy_a`{s7;Y<nq<s#t
zA|z-%gxde6peIwad@Sx-&0&0-XW?+3emT>mr6nAkUSSTtCi{cfu{7zuqc&>WJbnc1
z=1a&C2cX&>OggfEv4Xa1U1>cZS@<W8?MEa@8S9?gJky7QH%n^T6Hj$&Z5Wf5#{Eo(
zB-cW``S<gUO|qO(_B&ae`75DiQ&YOBmAf<R>jZ%_Uxw7OG95PIVrD0p6Lk~>OUXsL
zy<g7xp<+;y5nk}>{i3xI=PIr2H}we_8STzgz%FdJ*U&5S@A3uHRQ0^Yg5vSuMl}nc
zsbyYsdado98&D=w(=iZ(gz>COPU)+$FKj<vI=QYC?}wR-p*~&(4L+tAUeB_GK>SPU
z9n-W!B{z81A{swq=~T6Tc)EYmDwtUGQ8GHHNu=Z;{)3z4?UO%$-hC4B337IW2?!Md
z(V)@^G+<cU*j5aMa^}?W$Nw(!5?A$KxnK&;Yi_I@GG@HzJm*d~tCLIgM|_C?8Bt)2
z1Iub~xtK`vdwc(IxKR)!`Z!x+%5)O^I7mK#1S;C-xz4yth*h(t8myOoz1$hY^$rD2
zfT5>Ar7b&9_IWqE(5)n)X+}Gl9@suQ&G7JIyH;#wyFB&DHuUq^AD_#ySF94dUo=AE
z^ZoJaZk8#U_W2W<>nxHD>q57}I3YJVUc1CD8ByNR<cTKdTGX>3Fuz>eO-tkBi>)Pd
zV1uNzmG;EQ%~UiGjDOgwCtoI4&=rbcen+I~wY)d)KpiC&gykgjVF%pa?*Yg+oDeq2
z>!rdetJH=2p{(FQm&<77GZycthE!!ea6)n&yu5cLnny=ogN3^YYT$YF;-#qWu1dnZ
zzj15kJen8g+j!ku)$gzS#&?n2K}}E!_0twMlJt9Vmj@fdrhzeg3}sIOn<H|wnY=Bv
zJx40V`YLCKN<GO6UeA{E0lnw7>388DD)Q$3`*>&+Wf{eJ{#r#Xik+`b=796vONQ?@
zf|I`{tAx#J(}(ra-_kShhmew$PL*?}nVZ;^D-X!;5a)gkZ+#*8<17)?B@%Y+zro|p
ztWv|LaZ<Rxr*~O7WHc@RE;(m3L@Me#yyAJ~>9U9Qq^P>zr*v)#b>KC(=4YHdi6Ekz
zbLHF9wOBq<``Px=#XezwVz#N{uh((Qn0k3;(r%p{;Cm+5k&GYuz!h-&3i*TLw9)K3
zXeebpzid*^nO_YX&l9$<ke9>P2BpXLJf3CZGoEmAQmz(1;`_R&&gf*V^;!oTnEN3?
zuhy(%!xHnna=NM_R|=qg8MZ|ykl{;P)1d&m4%>5;|IShLgzO#I7Tx8qQw{TlQb0bn
z@1sfAxj619FR*R?JHFgr*dnQtj&i}m&O!+B)Kxb1$E2%sMVkH-v=1pjTjN7+6~-EG
zkFUzfQcGI&pnZURjPdMYc5ZH%y3^|Rb;e6>bxcRxB=gNaEdY<%-%TPuZ}{kv@i}Lu
zJY9t!vM55*I576>eE@Z5v`426UTk)p4_xkaL;$utS1nlnf@Arzj(<a{U=vc+kqilm
z%_rZ3QE;1((W(s8Y8E?G?5$<=&8yxv;nvFhXyx5{v`xLRNS|M_UkCnO<FHvEXY#C9
z;^LxV%4-|gYQriBd}!#OgYm>Zz_~n*(bm<$^$rZ^n2MdnbpHQ(ft2)b9s7?B;)&7F
z=rYkE)MA-{)`oq*=tx#&FUys|edX6sCzRqHS}K9EN4T#Ho0VXEY5tdzg2GyWm5Doq
zKB_1+?;KMrT4{Q|yyno+k>>MvXHKr$5hMh*0bbq&pG)Vui8OdpEH1Hhkkzd~sXZbn
zq<V783*<Z=o&o=~E=TD)&d@|d+j=>wGit-O%8aMnx?FW95pPS&$4$ykS21$i3+L`N
z7Om$;v*6}6Tfz=+$uJh6_#an5i_Hglv>{WHGu~C0Z~jZe@lI#Vir)#E6yYMz(l85b
zW$6x0Cf@ZuoXMPYq#5J7rhWlFV{R_#tyzLAYsSTD9Vl0>D}+0_+|4NQ@b`C-SAH%k
zQ&KR>6ljBZ|1`Hz{gLj;QXGdRLKQ51fP07niUvzeV51IX>2Jg{4|x^V6%F6DalR*$
zjBHMuTB`q+R4)js+`kcWpPCdsL784KJM^uQnoHMDLU!CDf?Zhan}UBJpS-H7s(!7k
z6Y{2O5d=4x#RJ^C4e2=93)YI0cv21}dnlZsq_k!Bq+#}aCiX#%cy`k6(DbywW=XnQ
zIZ|pQ2PK`Riw(@`pL$;`!rcdbSp<!~>w|{~sl`Hzpk0=PHi0M@4?Ua+IC3U3#?XB-
zxa!vtP)RUC*BZHBThzUFBaMfFOcEE*O{V=`-l?-i-9=YK9jazRDb%1Fkj|^V@rvvV
zs{ffMEjcK3N{>SNlIwkmmkq(mo9Jbz>%Jw9T&y11BLu{9+miKBmTUDD=DCl2N=1qD
zEJd!K`syavo((Iuq?kMe#O+#@LqnBwhm)_nR^8%WVvC0u(NftUDz`>Pfc-k)&`#C)
z@`S07n)v2{d%?4{q;ulLz;YMPt9tgBOOkJJ-1mLh`G`Sv9Fm_{#`PBjC!z~eQR8{D
z6llm345b#D*BXK?k$SFBXz;qM)Kc7xaN_4-qw}n_x__$1_9alc=_lSf^qgt-<p1{n
zup*n6l@>CGe$R*CRTW`1%1X-l`GxwUqXNWkBDI9e#6W;zXsC~5p207_UtYXzL6MQC
zd)r$QE<-1iR^f79wgBn;ILAJ=zD<FXX*9*K@P6n&zstqm36M>++v}>#Vo+K<fK2*9
z|6ZHo0bn8%nw~4jBaDL2dxSnSapnzppf625XMHtPlF-o>{JQt*<gZ0HON?V$uporP
z!sfJo;+5v3do1vlhL4Ch*E~K}-U@ZG`A?<F_~8w*dE<aF7W3XU-GBe8FPE4uS4fwC
zRR8tG*}bqduj9R?@#tP2e5{{2eNpuB%^M$4{d-lwm|#7gaccD!J*Vb1u2bdF`4_|y
zv%0ah0HrNlwd#p;)701FKjMVHabPQ9D`>U_a{0rmDMfBq>I$K6lr&oU7}tx#$r3%3
zS*qLMpir}N=Ei7TXek(~E(e+J<Dwa~pwz)+kdQ%`e<zXYF3SSxq0db3_i_F;Q8tP)
z0Y-|(1QRh^onvEQ=FY9g+DuKj$Ti<i)V%*^=Mc{uX+w3Tsr`Y^9W^z=mG#;j3&2wu
zt2iwQ37a-5u%T>bdcJs3461Yi(kw?HOLA|3%%28t;6wueKd9`^p_-jM>5JTxx{yit
zz>dG`xTEbO#MCEEQlqRK`~8TV)&Br-UX8@Gss?yg_KFFjU3ep?<#M;Lqc!5hv2^*}
z!tC_S>14Iaz+Uv@8VhDhOYz(3{+cz{B2A~Ox0PYEvJJm=7vD{+HCo0*UgDVSxQ%99
z<?{5={E9kiv1Vn*f5x0j0t-r5(VrNj@8Za+!NsvCOCfMN&FbB8KUXhZCJ7)@!D^)z
z_~4qTSgm3@PsZq}O`;#;gksf&_<GvH>1Nv4T@|dv6`0uG-qVycQgiJzFf>p+f2cG%
z)Eg0^((1Rx^5j~p2ygo@KqW8v14F2BYwn%<OxFbc)RCRDfocRl<%k03J81{zfm@!p
zZ1^nRAh{^GKA?msAkSbFFXI8|M%Yy;D({#1Zw6oWGdm$)LEkvh8gInL=NMJ4SnDoY
zbJI(iXWe3`%=V|{fBd9Jb6yapReSFWQB4}#R>x&gZqgYNPl;TC(?=)Ge@b5XkQJ@N
z*QqE%Ic4c9#C+2fu4$-)HwwIkM?CI1@o;nJc<3%>6OU8$wlTHZLp1=kygI^8dGHDK
zgUsEYv^S7|^vueV{l!*?ev0>j>`^ptxzpq#;g-vmE>|rDa#lNqOUO27N<v@kn@ih~
zeS>xP%fHi`*r=XUR#9rd>}@ktxo|F<G)-otx6*J7v*C99W2G;pIvW!DrNcrL?7=*1
zzBW(Oc~vr#7h@~&&fXdzBj?t|Wj-yi2Cu4W@n)u1p&oGVBK&ep(|jm*Nl(DnbQ2%(
z58o?I2foHLPVmw&e)()kuAZJxZHHNXhAW*dy8Bi?F-Z8Sg+9*WBm1wDs)|Y(oudD6
znk9Zzq64VAE88pAlYHxg-U}*+9<<JSen<>t8EAMsTYg(s23gJk2t_&kNHBlfFfR?$
zFDqD_FOCt2ODZ04v=~2ZlcS$rN{ZP+qo)=kx?u!&|N2X5zhM-oomREQcj4xe{TZ`s
z@MoEKGkA4)v!h?qIs246__uJ`ba^=4$!sLb<6|4{jbb1J-+F07))bGTAXh_4d^gFo
zJB@i0LGWN~)PtwM)MG9%SnGJcVfSM7{&4AIj&(3qRE$R7NrX8a;Hsw&8$L_NE=Gi$
z9=N3iPvJJfA|SC(OLsk{yExk#Qr<G>Rm^4XeE%&{fk^A)5{{5zcZN5I?{<yt4lO<^
zN*A4+GxKk-UQ_2NEcglwYHR{ka~Suokn_`qX%7$(173f$O<@TMy7f;hCOuTbccUp0
zw{*(Vkhl~_q_j|wxZV8EKi&b&z?nI4QLOqmCiU{dTjlR!%!%V}rCsFjw#(gP;5%yI
z&&rVKtpR(dGjLK{c=?-&n6bMulwGGTaof3I1!}IcV|RW!ae>z?n}y@bbQ8@g59dgc
z$4BPC@%ArJ{dC3wD%Wf~e7}A|s4FW*POo{q2qexoAg1Kx<kUQKHEdbWzyuYXY5RUi
z$RT!QC8BY9Lv;1}%fJ5)BEE`JpFiX=EViqGObq-C2T!Wqt<?6wprA<N_LlOckEu9o
zXk>tl&`A5!^})=_q-hi<w}wkZ<z#a>Q3T-XX;S&S<9)f!1YWfqEmbX$I5;>c#&GuU
zcneeG%GG^dmo-<oCqC2}6Wp_^?(jXEKNp$$bdAg6omUzWUYny2<Ba;%QX|LAp?W7(
z_hw?}E3+n&<ZYHXi2~j(R;#STkJDT2xoY0-g#5=1j-tBc3C~&EupXi8t#tTg%}0Qe
zstB<gicI2r^IehpAnKdr=f+i0D%RNVjKzx&=#d!t#6T7I_%tovM0Mk}9~ymMwA0z5
z$hnD~IhG+jP7k6j#P1|24`tMI-k=?WZLD8=iI1aDFTJ^?rk(zwg4<90>r+d4YN(M8
z&4XpMKfnHUL70;$RUh+{3N9I~<@r9Kbu53>d>`(HCg#t&3L1U7S<XzIukTXurt(*@
zw#g1TtA=LU8N}pPTG4TVzj(<T#a$fvqr`$Y{QD~*%`n5P{-nPRRo4`l=E)hz8SJ_p
zsvV(ej3$Uz3~x=qeK?(#0Em<89#uX2r=)t(e!ScD4)7&j>HU&}PQ@MBeQ*QPc|j*B
z7z+`OR-FerVWc2S%S5v}W{5{TR-VXa-)x!1#z*ghq)B7c;Gq-=xH|ym91EPIucKE(
zvjP>-+EBHAX+Fw*94n%)+acE(w(O@;s87Vf5^JVLOD8fea0b4oCc-}i%WW}(3Ua1e
zd#)}~(RC2qRg&2~zf}Q1Am~d@R^_d*z9@0queO-yuU5U=+WX*7-K3N`+5<NlcK>{t
zatl&wEAE~c{8DyCy+4iv)mtW>Hoq=5Gr+HsAv@k$p^|@MeRh|p9HkTG>RIu6*UtC;
zkfm9U6isaLF50QIoa2(w94`*U#BOThSqGR2`8wnU>)~e-SFC6Nn^5m>PqO6H7NxKa
zSG5n~p;ef4WpY|9Ui$$VBM)~HBlPM6%GgDUE1?_;3C)ebSrg#xKEZJ?6MB?U_%Aa3
z4^)tn3i;xWTQs=AM|8F=Thl<=S1BjPLO$0>u^Tov%08cf^9r-*mp7LYZ#q@9_FE>S
zTIc=*nXL$$3(o?7ten7tDC_j<*Z^n{Sp0iXWh>Ss+a#~1VE5o1ui<lW?0C8glE6Za
zq94CE*>d`~e(_=;v}s~~e%nBchq3j0XuRl{;@|IXjlcou_FJ7W^xB+j4JIwk%{F;k
zx1FBmRSW!KN=o=KewBkB48fV7kJ#2_&b0&+K@gL7BcS-L*AEUt>Dr}!vU|{)yhW`@
zW547WzIsDyJy{*t9dIcGn!h##k~#@q)l(j_R(fFhMq=my_gunq6m}PTjSJ2zHaW6k
zo|<wrihCWfzluLPG7lz69WXUD&Lne(Rqh8UqDvJGGgEMeE_KiEUJj(T?+XBXM6f4r
zxUC9=S(;W(9f>DvJ@WSKm{L{ckhRnodo}#b<U?pK3gl0Dx1UYUwYa4h7N9*r6{cTk
zup*1t?d!+@Ut-r#clT}p-o2^8Bw5o*`t36sD+8ClI4$pv{S^O(-BRVZkoc$%0psrX
z!Z5!yE`EEIqAJk)^ZF?TuTJ`BO|yui2yP(8{K=oTn*6&cM&Uf7Zq$1txg;-MTHHOB
z>9LdX@W+gpK-ny8bCIiXr@}thPu?`zF{PojlSNpP`b#PHM`ZD2!wP-O0#_bT_19H(
ze$`~8ggT=6HSXIa!W14HG!}%<{`pvxU>&{}#mh%w*#C@DyyOE+a@?|H`2~x$I;9Yx
zL5bwzd_+u+7w)lO{Eu{pY$zSc%}k12rZsLb%@wfTnIP={p?&m5(OFIKeKK9A5}l*;
z+b?4uCOAHXQZTcoYQ~I^LnW~}Kh%A{R@9qB?`!wGk*K)(e&<_h<*f|skBnsbb!t9X
zg-0@_iCmvBZWkMI3^U;pQTzRTWTOYv0LP27(aQ-AAe1O)?!9gE)%Rx1O3J1O&Y*A0
z?M3kvrgWm_X_-qg+RH_T%}s25QpV8qBQcFre&`B~{T~WCTVZE3y)_RfxfpSx@0cTU
zqh+C#j8OdKDs86vvLTNYY6ErUip67P2T#k`D^O*6sUHa`?YhlvF4Se_pW5j+f%VP5
z?<!WE;doO%%Ii{o1tBD8@$sjz;a3xM9nBn0obg<-nw2Mu7dGV<Hd&)b$KmKk0ZFWK
zXym)h3bUcDt?SfrS|)3QpnP<{z{nbm(C!lY)F9rs>(+FIUi{s>IOS4!#HT>dr{k`B
z+ggHq*!@hdaN+!5n=ld?9{k-4xUG@SYcjavr(pN##%N#N8u!3nrlCLo&HoZ~huobp
z6XmtI>rc!3<NH}=Jp@$#-n)?!Z)(%Gc<V9Br0sdeOUvO6_4T-8c2^zxzk`{tT!5o8
z;K=)O3n<GpQ){W-B0<T}`=+KMB>}mOmj7XM${x(ODJu@vR?UIgPidB9px}PEvX6P@
zo5XYsMxSadGzn6UC$q==hqrAD*j?La68v)Pk~F#Ab!5;ob};|8Vt5=xj$w>Tr$c^_
z8is0>Rg{MlU*8Kl9$YctQ3?(H_O(CnKNk;EOs!19SRvhL^Bm3)*gFFb%>f>kxg_&X
z9<;Io6A3yfI@un<dUivAU!(FRC9<2E)De8PJ!b|W_Pa{@&a{+fm2`EVe3IsqH;0$j
z8?@aug<pxaYC5`+YLH&Pqf&4~0K5u{>3lag;OWViZBoibwMR~7R(VV7OX&HaeKByk
zI}vASXZ<B)&9TF6Be95Xez?+75~m=CeQMgPfRJU8FifF}Nwxzn_6Vei%kTz_8f{GH
z;&^Y;XkpvW>3Lb~hftPBzgm2^{_cD<UbY@uarBLp{kK!>P0cYlL-p->WsDTk<%u|D
zs5y?wwsJ7%setV1l=pmVz{;t2<o-!=o5(k#=dWz^n7?Y6Dst2_yeuCywVJ!p^LMXT
z<v5W4Btq8BLl$}!=n@$p(_&h$5KiPS5@7lOfx4p`nVGPxkoIYhd_!3DkLLLH8(di{
zGS~QFQdC)2IJvc7tB>($LgpXe=JCP7oYIszuS+R1m1skG3Az6FIK<@*!<6O{7*{rX
zL`x?fv!0*0D@<&?Hc^wjU3B~@k;^C+*dtGsm46M3e-q^othS{}`8-vRg&_DJth2q0
z1)bxs$>q~)V!uJrO~{X?f{l|R<Iig9_e*?LES^xSQL>{8_4)NmGIn8aOtGNoXmFZa
zqTUSid<m@|@*C}shxW$zmt1pwp2R1mS1Y5gJwf^Mi#^wC*jo&(`nByz%G9bMEy&Aw
zd7@xX3F{QF5PI3RaIwA*$_a?9Z3P}>!>2*HTOQWSZo%J01O&iY!&h+z5y?_VdIaCW
zz=^9&uE_gJ3wm;F|Fj}R^b{s9E$&=2apKq+tE?QR^Nm4>&Bt7ynP6FhwZQc@Prj8x
za$Vzcga|$9x9EDC^ztpkEo`n1zb$)yQ__(V=@NMAEUj6wF{qGD><y5@RE#Tj!?ITV
zOmTl#u_4O5*BIV4eGO~}aIol`>?=M?3dFSKy6i_7-ky+%$m&nR4-TZLxUaPO60n56
zR3}<on!kaNo97+1cqwVDyEmc}>I)Zp1??U%f~@HYG2`!XjF!^TnLNvnM~!n);y%xM
z0*UqX`r20sdJOvwNqT>*iCmQBCLCRg|GoJ}{*YG-3gv-<H)xasWQZ--R+bd=%Zr7}
zg|GN>gT|&{L#;=KV^5^nCNKLgPk_J{^yM0G!I4(C7y)s(laT4mOjbWj@yrC#`K}f(
z%tvnwaX_vep9$-*D=Wbc!q7qfis(QfSD*4VvU=^8X5B59I$oXmunkO{A$wtd!KQ8N
z5BH<&p;z`l+#A3>L)~N+xb%Q$l|YN8J;Xsmf2#Mdh-9g%3gO4B6Kes=1zrbXx-?Hl
zoM@5z`!z2jcQ$wT`$!k_bUb5K*3yaHYp{>pi7_K1K`s5rf>B~1B+v%f=>(`bF4y;+
zx8$0F?OG0J)|Mhg9><i^Xxw^{g(;qdS1$yqw4dNN1Oy1g(c`aAL2O*K6K%e~8a7>~
z2A}0z-eB8p&DA416y;bwx7LuS=falr-2}op>IA|5P-l$R;s?)!E2P6_mok^LQTAi%
z^Z!TFd4IFvzkN8=h|!8sq7;d(irAw?iM?vp)|#zdvu0G)jM{sBN>Q;|u?f{yYR0Ol
ztu<nl*lIud{_y+(IVa~N?{(j=`?~H(Gxxh&;j7KRYQKyOl18`|s&7{iz6EvM55N<>
zf;un`*UR<dcJ0ph{C|JN1^5+fk8QX7D0}axO3_oGPlvVK9qN@RF_aCvBw>O4h<Fd3
z;>uwQ>3fZV=^y)EQFK{~wMFG<QP0{GtgdTYqAiCwI9|h?|K1Ln|GxAZO3P{mh)?Pj
zib=9ce=^zdt_HEV0bDmok;+sMWU(f#5)Z0J5@PahYM|6V`K2dgVQvae05}Dlr@T>&
z#|aQVY_lm%%>;{WT}z`nJ)V^Nq4k3QmJkJ@C^+z8$)_cDr<FJLZ_r}>(rjwNt7RH=
z9@CD<M|xg`Y_>d6+R{8$Za1P#NF0|a`cjhzq4f$ThX^<iyZe24HC|$ycX0DZ21BjE
zQ<QA+19rBz`i(d}r#Hc?jzO|{C$cT#4w%SR#M4@>rD#a1T;MR+(t~{&8v=sfzhl#G
zJUNynldcsTCG_DikiJ-+P3_|oNtZ3JjVTsjDO!LDb25CA!oYwtdZ&DKinv<uJ~f6;
z_Tgb82<Ot7hU;&5?#_Q@*HO{ciDE4C%LmM4qdPvErB!J1TqKLrA&VNla;$0$5=MN$
zSY)URs?zGs^9A6s<8RlCK1;<S0bngPN%L)|C1lu^yc2!~9Bra2d{fm#<o>Cn^5Xgt
zQRQqNWn1yWSgGJcn@Ivl(!i9*M8sG(DCr&|vzLQy(a7|Ej6_}JR(9cVg-hEm)~q^9
z-K{$Gu0w`fLbXL&zw~g<Dr#!+xiqcM)^FcT!dnwb?0N)6pBgY;6dO+upds7X67&8q
zK9e69!vgA|i~;s!BvZXsa9eBQ=Z27zQ*qx+0jRRDJHc6?!1s9VBv7<J?lHNqa0Vif
zn!AA~M(7SH*^g78e8QK&x_^b~fJ^_Hn`?hHJVUy;GBh2$W~||;z~5{4zXdQf8mYJI
zCiPM9)<9Pgblo7}i$acCK*x4a<!a6Lkt|xe7Ij<V7IF%IstwWf@Paz0^C3UC*Ew(%
z9ky?qw|&B~gtS|V;<1Asd#mlw5qG#<(Roarb{}A0pMLW4va8EZny#52*qqU;4ga7?
zf=C4ix5z1<Urcy+`0tI^nL&+a1I^9NF>TZN!xdOF5|dI95fC6<G~66iFYi_{p;I=t
z1gs50PtSccz}OB`zey&o27Mm31hY+&Lx;p2=-wONSZSR1)+>VXcx)l|UUk+Hw{G1o
zH%tH6Ae)FzwkIrXFUo{*a9nn!dP;IuCg7Twps$E&jK;b8qz>HjG}?_VxTU@JPfyXT
z6}qFPMJ|>GWFS!HTDFR^1{y>3CK_ee0<I*D^T{s$Q7>{Q#|9G@44Cdl6B{|SerP3Y
zTG(_Os@1wXA@eei))1QFlj-XCkn>v7IHb!q{=dcBQI}VmpYX4r#)#;!9ZW9R5PuGq
zqh)V%Mh92m0icaZmQaGb+f61(icFHs13tL1BES82UkGAiUI6O1CfJoTS2e9f(-@M!
zmcml<CKmd$J%Exw;zN0qu{ag}(1!Q50nD)Nz!+o*cYJ3PA~1V1Bcsd$alW~=m|N_P
zY#<?-V2t%SG9l!E{@ZWkJ?-$hO|djos)N1I;dBruP5<<in5LSqSnGkkPCq#IX}=N+
zCwXm0iz5;V$&lpvX7(7a>RvkIvTaJA5t3|AfpeMB<*2>Y0NQp`WCT7Dv!~+LB9v*`
z16hVO850mv6`Lq}nC8CiIw;(VBTEL^v@lz$%_HGa_%!kT-U6ny?pOVfJA9}4HBMh|
z>_e}QPrA=bt{1ys{oC!C^ax|~v5j>Ny~PM^@XPXQ>^_*_z8o&O>h<hA+c1j!i$Y0y
zcTDbhdO9ZyxZG4_1-6>Wz7j6?FO~{W7Gu*JoNPi2c96vbQsWsVeg@iU`Ne~s;wjX$
z;S@YL1-~6g44A^Nzs*9CQ8<+@l5~YU<xUu(xTlnNu@*;cS5*@f>CZm6L*{ls3Ae06
zW%PETkNnyW$7Bk6`={Gdkg3)hJ*lM(eN%ACZ{TmGxQxc`=`o#jc2f@jAY~@-sY=Kn
z4@50mzpmdG9!{#SwM=cDPy<*TZM$`d+cU31iulBOZlo<~15G4)OgVTb$ax?0iKXqe
zHaA&6(Zuzh`Qw|o0E`sI6hL}LsDe)v1tWiciq-_3lG2kp`_6-2Hryl+_~{Unc2**b
znD@4jaE}(@+mu)!q-O7Ievs|mJ9}?$g!fU$yx(VA(d_LYJ>c&`C*-Qey9^rRp@xf%
z2?7aS)-q<7)ns|FYJxX!vJ^E-xAvH=J;q|OzxQ5-G!CK7ZofTvj+~yJR!~tfN{=C#
zKU5SsM1&`#?-rGiWG>!bJF%*82sW1l$ka`5M0u7?oX>C9FOXP5OBO`T=?ubC`Fv%$
z`+VV~i&!YhIssRS{MTyrAzBLFh>u+=7_I=#Wgi%+T%D6@Ut>b~vR{1n`NrB&y6{w}
zlm1Jwgx#4Y5BO3i2?G0v_I>cr7q=nb`3z=ij{_7}3+p@H{+Q_h-vAY|Fod`*M-qB=
zkXkTU8z0wUdzYKB^3T@}5WTho&awq`Rit)+35!OyPVKLjg=_4xn|S0A1LAxqjObn-
z@$`oq@BI%Z?rqk?gIOmFCPNJ$CI_U34FtN6!WRFnOqF=wl0{mTCma`Ni8aqAcg{b2
z^|>20?^>_5Qd?B1@#j9~61PPiM{QKBjCtwgWBtP(t#o&I%?0EC8!=O!G?jCp3o90?
zV*jAnk8bQUDScwLp7!3v+JCjyz^nlV))#OpAFLVM2o^hC8gx%ULR;JIEvOXdy|c@8
zUjng|L_5{}G1J`zrF-&PRB5U4F_{a@bV4yulc;y`@lc+(kLk#)P#(+KNe52v=97!;
zP2ycE?=ToNrP4BfL!fMZd(`P4X_cEH`I<g0C@#MiGB87#yNCj4dT5|dv|xfQPju}j
zC>;Gf6+o-3$<T8SBxh&V;S~=jRm_RDP*bE=bnkO^#z2qduTTA}Q5FlLOlzjhDfT4o
zWX^Vwl3sv$oqTj^x_|L1CBV|dC^myl{Fdv%WDuGV(R76vq2`WxtScBB2MP~&Ddn&X
zqh^e_-$X@Z2kAtvo@Fs~UH%qFTn!pUo=z`)wczyuLON+AeJmTE*w3CPtlXS#_1hkJ
zpV!rT@;H0-_L4+#iiLWbvsK4PCV8F`y<&f@jOvn&A8~+z-1n(xGC|v!Vg-|>3pxdH
zrdh_(@ZLbcHC4qd`&(ON+rNO^*J}3~_eqvR&$SKokCZGNJwzaZa)aD;`vJv_r;KAv
z^AXvU^=)n^mS{VnoCQTEk=>pdMXD5i3c{OUn0(KvdsDvn6Ga0jQfVDX&kLxrxCi&e
zG1j(_Lv@%4bFUs0iPJ+j;Wr)ux$+(ffSp<+I}o~lSf(sE5TvT+sYe6iRf~y=vfXG9
zW!$^C=v*XfxkWtyFli)`*3gVh%AIceEH8?>rmi5D^M&~r$kn5*9PWkgH=>O*wG7JP
zCwPx1lus!7Qs1^yJ>43-o<7Wsw;q0=H{othB0xEKdwU*tV5U6>6oPTGcZS){#iLu2
z5m0};&#klS)y@88L7UlGlQo^vw9lk=DvXnxn>5TnKR>8pZ2JqN3Q-O{1VuIF*J_6=
z7OaMN7Y&v_P*x~1teo=xyJ|Hv+Y!0%ca1BWWfQmWj9eaYc1)LM24^2+>5r3CQf^jK
zBXo=w$X-!KBIaPNNE|Wi?$8-|<Zg$UhWdcf<GOdY5?z-U%6_l~i@PMucpi#v|D8MC
zRGvC5ADp7ST@J4ew40Lio(uoC{xW8qQl7imt~=;*W+C}O1_Sp4_GH(f>TcLNpEXIt
z)+E?-=j;9gHy78fGfZT+p!`--;+m3=s6}AegooU$nYdQj`-A3rZz-JNJEJH>K+Ei&
zH$M+_6z59(^S0I~^72JX%jYKay|?KJ4LyR%7q28)Omqbt!Y<eL#-8zjJp)L|$200~
z#0hS`<AiG=blajVuZcPYxc7U4ttYN@a#fqLbo4=Lj*cTLlz29I?0Gc#<_CB9?r+~0
z;VWtt3|tT9!<OHGo+&C67zvakY=E(=j*-9J?Yl%ybYocxmjTwB4+ejrdyk*mB&L2`
zACmMrV!p}T&<DBCR0`rb)5fPgCAS3j=wRdVip>2KrY3imQxjyk+AW6td&s_oQL6%1
zFuyrtqHOsz%3wCb0#bmnb^FLl@1h|?hg1icCQfmyh$SCpJQ+(T49r5L9u35RI{n`%
z#3Je5fysD=$4^n2Ci(6zs5{1>M~6MrGX=VSA5?`hpo4fxAe)^;k|hV%{d755Ch<E(
zkK8`-t`Q=2(orDRtXfRo6hBMV-)7n(fxgGWs~9b3b33+Viw+wMry+HLO-8>6)gL6R
zU+us38&2Ac0C1SODHieT4%9MOB4ih!51<vRn<qs;4=TU0Q9E264{~2ERaWWiw(a)3
ztPmPsAO+yH)exb|d_R9^9Uv~pYTu21`<?XrmC1;qE@jPZY7wgxYjUu%ufT1iN5KBJ
zz@;w=i8RY<P`6^<YIRTovQNkl8b1d60N*7=a++8{og*RX6$tjK50}5S@VOzYvQ!-N
zkt&*rsxQ(N-O5uTnzbxs63$!o!sUkxH#}oOjAV?-uovn7Kmb)=dvh76ybrRf@C5(U
z+{+?qdUv;X4Zrp@OkIa^TS<lO|FZ-D95p~c{qOpU^Rv4WbqAb`yvBA@NT=wkOuvW;
zkeXL1V#%@~(SNNl3G$eai18Q0E~$eGX>e~>0$*hNY4kwTE%N>AGR>^%dYP-n;r_7+
z*}B5q(P-u04U3z|*k!j7`v>6{om1P+Zx=ec=+c;@m!g4~6SYrAL63mHF()v9Q}teB
zotLpQTZ`Pgcek*-5l5SWL`-V`^r^F=9SJ6XS9-#EXHAFHteN(fIXPj-A6~=Zg3PCC
zEe^uvLrD`oIGiC0{wd3K{NH34vFqyJ9(v5T^I@xV;Yz*|iY@SH|L_8v%wemF0c*U~
zz;&P01{0OT52p~POWP-1ZS-|ELM)r%(o%SbxpgTAt*;HzxL9d}$m=+j$o-2`6-4;O
zLCa#OxPer8(zNGh#Kr#An;x7+OVR6k`OV}=2h91-^z{11WrtEuP@689BWjg|W=*vC
z_V{96u)ED~)w-;RA&$DB$n^}n;QYDuU=ow@Y-_VTDWo{cUoF9mU(l^GxVyat(=<su
z0lQUC4)BKU?mk<1ewXX67twaT5kV^)P&S<>##*`DCOCJek&iIuvHUqmkN>~AK?aoz
z{H5Tz(Xo*O;?i})0^2eD{Ch^G-n38f@d-nkKg;RYUx2Up-!sW_=+cwJ(}T`=yUFVQ
zSA_yP`7RLbopOJWmvdoxf@rc-aUq*c0h<F?^45&@?}wjVyXkZ(mPu)CaYRa%)!2B=
z-#ycRp3CFIPkyow0u>lb8PeX9ZiKr<at;MzSWm?lfu{d8))}@rt2MDE-B!!v(^Y#f
zQT{Z-QX+9os>-rh4hmRK$V{(X$UFeW-$?a04y_>hAfA>5BZX-pv55!GCU)@_pqER7
ziLn#?<we_c!-g16gCFN(H3*A?7abTKveG0{NVT=u>xksYeczXLk|xz}#Rsgb6MA*a
znU&62|3kt{bEKY|1-TVa`@4K;ZFbrpM_7SAPR+~ZDIWF-X_{V<Tr(+9m2Gb4Ta~L8
z4;N!>$Tt!k3%oPscbA5&$2h*M-`98=w=}c)`?Dd3Mj4|T65=jagAmke8XvG>F9W`g
z4wKm*AeG3~V~)hkb#+)MRFPfNLtWPx;?>bVBhtu7#Oe6amYffdI=1!b*}NsEQ^1&2
zRBk2gMO|py(75Kc>qP70$^U^eIB_E>wJSB++H?h>aI6)r$=UNfx{w&2>6swKhTIQ%
z4DW78#=_F6$!TE-`9ynKIF!Gf#@IL3<U6>0srw|XyTx17)`v3F#84NiMpPicG=(W6
zA9mSK;>FnDpM_{kg|4gg$|k)-xgAGIH=GzC06#P72P;7~2Z=Y?cwrykOJ>P3fsHQI
zPGK{ubSE^?O_3a1`}cwByMwEVQJn|ap9^uTemd|4B?O^RibPM0^n{Yf>jGEg5zP_{
z4&9*{QstR^016&*Zjw0Rz+7j+$|+t!HU4!}@e}hRw~dmE{t{nYb^F<#QP&SJ6Bz;e
z9T)>&DZC`KxgzsW*;Qg*c^`DPJ^Fib<waw@d%Z=o9UUDZ#U+I$UUt8>8QKR|)9+E*
z0{cr5NmIw(x8lBu6BjzdPX|t)?}uN1nxF4WWQBdU%xqiZNS>ocw{-Z2KadVug{r49
z{w4*m-VDDs8JL|h3k=W)aw(8F!aIL=BKJ4kp48cn`F%otLj_IM*)fAb(fX3Zx%6rA
zIBbSt_W@3B;p$It%gM2KVEZ;;e|7rz+j%{Nnhx`S)6aPsq|p214fb1?gKXRnBVQg)
zzR}AO3(Q^8=TuR-GtAnu5PH42>Om@+m!4q7T2n8J=oUR2+cLY)R_p8R+ZNnq=-wNp
z{5U`VF|`;<nyC_JLx<S$YiPX+-aA{kq-KzlZOO;<5W6nsIplomYS?eNy}-}5|4N1<
z()`~$q9-M>U{ltjMv*Pj#|1SN<zLZ!XLrk~?_3pzI#$nqpPmbA+5M1#RMThvT|@LC
zGEI#z7#e>qzc<xnhEwpd%XV?GdJ9rR^29LBXp~g2PxZRDxl6P3=HGCmsm>l8x3;+-
zy1n-IRcxVSV_XV1{7HAd8h11uEHe;6nbuj&I_NI%SG1F@|FY|Whcq3W#Wu7mSNBa-
z;Z2vqFmHJ}fDA;zjTeSBDQbTFd?}q-l0>M<_7$LL!NP^gPFNoH_O+@Bywy$TGZ_Dx
z&#t=rzTJ0k#@D_mrLtD^MUa?`B4&CT`x_ni1q_ks!L+={rcm$0W0sg06FLzx9*_I@
zU-T$Y^V~vi#cX{L8(e}?><CN70-Y`BaIe}#2nSXeYpUf3rd)5cHA*d09sR>9r#FHW
zli$QS>@`l`_3g?%3bUDX@3J*Sn`v`x&RGAjxYuDT2FDv7U|+Q_nb4Ml*r85HUMi|S
z@P5<a%{NO9H^bj}1VlpsemI2uk0pbEY=4UXlI>5vr6^x?bf5m~zPJ##DRLddC1_Of
zNmgXBA`29CSJ-?O{`I6pB9mkF0u<`Mq1^e}%L6OXJvZ(a-8EP{0J%Xk8iZV`*&&a5
zz^{>_nsiZHpnrjP<R2d{Jg+5#8iU}kc{3Da;^SAQe)9K%<(<^DiEj{0TZe|4aO)~i
zbh-)fv$xo}8;ycnt1CH(H07f=!Giw*7j^{dCVolvujBK>dDzQwBwux3<gq~ieTw}=
z8B)s!a8lcpPJr72p=uD2V<|dN^C^1Pnmb*dfVZw~g2mRB`8~h6dVcihL({FEESJ2N
zM3gn=7+<iE#z4xkATaeP-AA-0uc~>dRtu8_V%7_T@zwf<YZLR^wf)Ww9UcA~<1WIK
z>Nf#uzKQFYC;QhaVm#1VzGbMYW0{$*rZaxVwZ)+^jNnGfCZ@CmH<SMAI6-w%R?~_(
zb44KCS%0n9aN(DVqJ@b-NcHrrYL&6BunDuROf@?}RfLGRS={w2cp=%9hOK6LXRX#^
z+t!Yqhl*^lK~mwqN5Sng{8kI;u+!rh?|kV(v?z}6pZ?t-{rML1GpY_P>Yr7$#;?(O
zbv8V4uvdDF&-0L5gi|5Oc=n_zAKx;TW13&FDtlEQOq13b=+YK|yXCgZfVg&eQCn`-
zp_pM(dufG8_D_B{W$l1DV+CBEmR$E9ZS`J_?VhWg=7{K_P@A&WtA6al+CF!4d^%5*
zd3?O-yH3YWh={=Y=PWrV3)XRZ%viXf$XNC2IFpf*4QXH7qGDUm9&8Zf*!H$qDb{%J
zl3qc_{BUYQBa6^Rwi9iP+$en?RoC2+qt<z{zJTa#d?HQ+q^*4I{WYtTIqrd*v&^WV
z;RP2JoANo#4%FFml{6{UHj&>qSb2-VFuz9cZTKl@P?f#;_T)p^$)3kYHeM(A7rUW-
zykU%bmZ}vFY}yp4rsF6&sJM`GuO+}SkDs&OiVz*#1aS&T0zdgE^-hhp|H(oWgVoqa
zJuL)I%@<4qWM{U*AN6p!r8jC)V0&KQU=_FhWyLZ)P3kwHhczPuiAtEAop0wQAn3Bb
zJ5J&Ix&$0+`JpLGEdWo>3u$F|BtRqa4C6E4Z0B0EtUYjUA}m1Rs~Om2J>(kXQbq|~
zHBL)?kuKQh`krLA!xa=Yu6yp-NBCU_A<8D3;2JYS4We)*2mSWa@((~pC`BHhQchiD
z2yw&r!St_<tjToP3bIzA9Slr^IE`t!VGee(^r!PP0RKrF$P;iDV?XPPU$e-nr8^Tl
zPw*33Be#mVuaD(QuKzk*bLeI3;rLTvF(B45ox%dTc`~YoW>>HmJt`}UUY!>4Ue-W;
zwsy+~e>Ul;%$f6T)$bK;ElYu-kcDKnBNHP?fy__%HnTFt3VG@6CBQ=ngx4A9pw)^n
z5PS^Bf*^jHYCH~cCQx3At0o8ds4m!J%3FKOTD(;i0_+Pbc}~XSNMCrM4bx#^gpq&v
z7z6#_ny*OAH30%dsfbRyhlnQC1PBb$>d%!BeS(DX!|5TTXO%4FnJi9r=DPhyO}YXv
zO1)fa&^<QgS^mY3Y566yBbP5vGPB{3n5d6<Y8EKDD&HSYVG7|_R@6n#3vulYx&z&3
zXq9_|`E(ckMu_hU*32L1Nxg#`<jiW{*Rxn0IPw`Ku=0Ls0TcvTgeG62zU33%RQ+)l
ztIC!~wm)xu$X4ZS{ls`f8=5UU!%{V$f}gqfq|UvFf|yh7VmHZ0nL|U)Fc$WqG4E*n
zVt3cs&CSjDKDi<0gcCNQa~InGUH-8~*hY;;<Tk!m(CA79zaWZlK9FawMj<RGigC){
zFoatx{8fX-A4H8=5iZHiN?b~hKmnucd<@Bas)DLqijPKBL0=d`ZjEXa9^RC{vDDhZ
z%s1aK>21@@Hvn}RW)V*N<J&)|RQ!U$<Kns-wf_l+k=#!QtcCviMIBbubZos<cvjtY
zwt2n~_GiOQ&MP>4#+uKO%OoWxZ=^W_@RqL_$PTL>aBu&JM7&7XPb%HAY0M=5p`1lx
zC7yiqZBS?~u7SM-d(XWhe7#!87<mwOz5WV#H;P1xzVBjcAS~{u2R>8o6&Se;MTZix
z9Vrh&18sc9kCCS2X?3;9j{s&-t2+xq)~UA@^G1%3kKY*@4K|SGlS9)%PR_15r$zJS
zLO53`7WvLk2nYENuY!^AW2wdKJq9Az;uUEGYnacIL=_(5bRxAFT+F4tY*;jmSsYFW
zo)m}ev~{0zU(A!Rgo|NB3<s&>c~Np0fd~z4x33p8U%iF#_XfVsL+oC1&td#YJS6Ui
zYeBllqEuXGhUS9iE|`wc%e;*}+%ShvY?)KaF)v~tdE435>=C2HcgKFiEvh$-Das8}
zhfO1rk~ns~=Jhy5xp&gA6ZMofG)k!SzpA^@Evr8b;9vfOEZ4rQBC-RLVcJnF<Ei5k
z2g>*E$^4WS42(#3;V-?Hl)ztBgCjK8n1J*D6AbJ?hy6aAC}}<Fv56*$kbg+fKhiKx
zhEX9@AIpr`E}pz**4$mq?EC(Gh|@$3;0B}Dp@^cloIQj{ye@IUi|!A&Xvh}Hsdsy;
z>wK<z5g6o0a`Y@thpc?wD%h@2#`t&@lgOj4Nxi1CM|)?US?Q?Z8RelMhJ$`wGJ`$k
zm(%cKLL2PEe>eq0QWfrlumCyq)zhXUxdlWGFX1NO6DhU!HJ|EFkD|q4e!BC&mdJxA
z*Sm;wHkHebEhFXAza$+|Z$4-gtV)`(q5i$P<88%4_4fQ-uA1&cYJ>hTv|j*gENb5+
zP@;smofuGnj>Oo{KbUqLDc_d#K+`85OjM~^dv)}Nr)aB#Q%&M)TkI{Pkp6}}x(H5s
z2-HeSGw5(Z&cG>|oyPq2n|Bo8N$^CQZ+CnXulGAR*h%d%l<ApMpMp}*2aTU@tZ$2M
z0O+5oCQy(A-aUSl^gvyo5T7a|6cz9EBl7!O=tq7j+yf4%lb7p#XhkM?6i^y3wS+7&
z*U;_HYy%|lQRW)x0-imxVTqTAk|D}8DVgM<S#Y@XFUuq=i`#Ecrb*&qAH=c;;%vM=
zJkrfAJKfpph2{b0cL_f8BY3PYU+fzZAnh*Cv&n$99)o4j<6brmmbTz<Q7x?ynS>kn
zfB;J;3Nx9)DNSr*+N_oqlq8YBZzVTjbyWM(Y9NrZ4(g7DRWHGejc8F;!~`c$Du?W`
zYY~Lm?9r$KZ-T~(+PtOK^L_(?g^3zb6QpY<7lj1YQt27H<yTj9-9XWsZ}(1i=C@jo
zJezHB_f-04(QCW=E<6UpGG1@i*U+Svr@@@c(dR9d(|#l6pvyc0ci5lf`Pq)aH$zWy
z3Bs|^lLJ)ZXQ}5c1>4gkf5S-ln$#kxK1rN!I}N_VbqD``ia=Y%)hWsUS`3&kQVM1p
zoNvgtf6#e2(9&JxcC2zveZ6$}_psz@;(7ynu5zt)eS~P5_w0DTScA{0vsHYkOs8%u
z7R*R`B{5vDv_x(#o|YR$Zsw?5zV({DxVqj~x!$b5oT*efO_)L;_G1q;==;HdROpGF
z&2=^66k&%64fF{nG;v?9CzYJfIdomrUhi>VFB6on_NG5<&Kv_9U(7&9k~&FP{sQ$x
z(o@HyJ(lc#+v?5^rsIV2r&S9jy^bFttf|+Tu=4MDyQs1V7slG7mh#Q_g!J(+IUjEO
zkeG9g^%o|rkRybc_(8Y{{u|1tOekq(<q<@~<xxo%tIScvW*j(rm@E@e6TB)xKeU9l
z4TCI`6|~zAS9F-aBqLBqSIy2j83MU_!4#7~fb4;+f8osYukCsrVL5r~Pf7r^40lDC
zLd{U9s+h><%{7kkpizQ`MP@<<#Q7Z%{Ovz=W_7cG48-Ysao0Sa(?oqPD@y(1Cqt-+
zjC*&xk0Pz7I3bj46oQr8Z|PFp1QtDI)jzT&ZTL0!6)ALu3Htp2k4<mjdWF`8TNdvL
znwGSIEzOH$;+G1I?D!;N_-FyN+-Y;eU`-7TbAXwtlY}Cn=B4vK5bu&_^=~BN)Sr|R
zv%fm%M*KT?Lr0y2DGwB=xCN`lyCwKG<XX`U)rdGFzsEywC7UxPJw+R2vYMN-O;qg#
zto|uasz!%~`E@FIY=+8jeH=jcJnd1UX3Fik1Au#Q-pR4yr3U~$JY@`saRdP3wM{?z
zrNDU~FW;nbTnx6#PWmLnhhnXTsPe#}(l`!E#BEpp+bcs#=Fh0q{~K_Ql95E0-brG;
z&qyN`5(g-S|A+b@I;0!(A>HZ*0EQj1)`^Bj3FXHlB^8ZA)@oxW_vSC(M99QYnLm|g
zaeC}l1o?#Lc~35Rj|}XgdYckh!*9Z<-iFfOw^dE3K{1W?(w!r&iPu_qNEApClJHy?
z@q+Ku4S=e%>*SbzY_0+ggHA;f>Gxg;bGuu))Y%qxZYWc7-}N9DJnFSI{2?ES`-~D8
z?9|@R3pw8|;kNC3usBU^IP?--MZ$_C>R4nkGYUxJ@<L6~>(niVT$n3oD{~^LhUVTh
zGy8CEcD5cxOBTizGa+&KvB9o>K7fEo+ZsuXKf0n`ytco)TW%DA4_TBB1ol`L@L73u
z)Lri$pCFc7BJbXe_`4x^-W+ktAwD9``QYVtLV(O&PFvrFK)bm3gN;UFTz&ZJy_AQM
zmw74OyVK&48y*hlGcER)IG;<b&(8LCj#U?m`Zi^1^R)8$;ep*O;;0&N_!Es7!R-DK
zR5=~5^ojIcX1F?0x%lhm5V}#GbE`zbGvD=kpyA_jM0Q}r!N~RLH6znC;ks94B@?x=
z&wZ71y}nb_LCP8)EJmD7h<E+_lVsl38M16|;ktspK2Pa7`^J6wk9di>9(C!xhd65g
z?0da_#n$z2tG;u8KF8q-n_+}dHeE5#cKHU1eLwF_f(V}C1?~Sin;R^K9^4?@HD22f
zZzGw>=7Xi2>g~J8QJ1MZ^WBq;UUG*i50C$i^^5Sky2jr8STw}yHOt2)=hZD$((r3^
zs$eotU}#7ybL>Rj7hlN-Ni>p$ag`LoUu$t+em`3&I-ea50wxvFW%wc56kjmO$0hWk
zY5AYsgbTUX`isfgO)*?n<PXz4eEqUHN;KNqry=TY$dlZS(csx5n4kzD>4gkOKDut&
zVp8mf@=p&_p=kei?RDr_crN1%gmqFs;4v4IL`9+Li0Ie3!1>XmSRPo0Oj6NT#e%O<
zA|l2#=kL|`!=|`qr#n^<NiJ*&=Z*8t!|e~1l|0Uyy{TCJL;Yx<Cb<{ieyc67)Zrub
zB%94P`+o3=PiOK$kyW-EtfExtpwpxgqq|XYu-e;EdDpmtJTEw19t(0}G1h_kb7jdV
zn|Q93jkl9Yi_kEVQ9RBxi6289d8^5)n<qHegq^)5EL`v6lzci7=i8*XqFf{)I2oM|
z=8-vSA2=9LxD{c38v={oXenw58uER}1W={=CK2^ER%tkpS=NdSiwkBu_{LYo5Qy<k
zv4N&NkSK$wxq*7f8%jSsPu76@xqiUV-4i@kzVS@jwXg~e2R!{DEJ3V!B!N#g5QadY
zYaz<F6qDJ5mnaZGN3J<C1Dry*Het0`BQnjbo%%Q#87ox2QlsvC8%u4!8L|eC;Y@pp
z;v~tAGlo^k_}v$Cf6kpgMR&exDl~f@Z&g^{C+Qbc=ntWw<RN#ffuw)OSwDSroA856
zGEm*8R-Mx=r5D5`EC|CQDT+bUcXFU9RS}CaxY;3s6GHBnH8s=C{N@+uT|a0|_07||
zuXmC~NA9<GF~BvYEi`%y4UJj`SNFq1tq&`GE-uD@|6bU69mrm}t`7_8e072ec767&
z>ZzfX>r91BE#Bwyg5YqqU(Y`Y*$?RQf3-JW;B!UK@`Fa+VTH=;xw@5e{q;(v!^M6M
zoGjJsFb5I2VS2K&aGs%ZK9kdR$WM@@V{xJ<ZNx_&Dp&pW7XyTuAc?%Hj&`4p%HoLQ
zOAfi6In2o%nwt<nI;BbNwti>71I(<$<z`9e;kaJX^iBtHZdwI+PR%C2p>pkZO_(rx
zc(gvLMJ(~kRyt_*imiLt9!4Z>4hLIPC3D?~S9|A+=f<-wz<>XoxX*XLb&$3V#I;_@
zKmILG`yunyzkd(wyUsVIMjrV4zS=lJTq7t!6qsNFE;R6%L{aTL+pPBK+{2Q9$s}+R
zd_swcY56rBv65C@-+AUmq$Tcys3@;+4iO&8C3=aAXV|GqbcicjsERn7S=w~$hEO@D
zNi%@?(SIAwZtDo!5-7dPHy6XAwv?G`W!5PZ?6b%jF{=7m$8I*J-?_QNmaAm$=I3o1
za{q{MZkEmwmLaWj`xlHw(Zh*W#`pg!b{%I-)hX|PFaIX4eEydpThJlr9r4MU66)MH
zHcSDaP<vh@18+`IHz{N(_ZQH0??5zJowM)Isq06tFe0VAqo^1U8uVdD+p!r`8-x8H
zL|pI0vb3khW{xlXc6xqK%f2&1UT6jqcqfjYDUDGAuaI$C;SFf5ntHMBNUsZz84x}$
zI-n#G>o`M)9#|jz^B3Q_Yj4vMe%`8}Y|VYQ1o{Wl5gZ>te0uF0;!D6?1|sIeo;<Ew
zF#Y)}P7@-$>)G{+xHH_@$BG`X9)xHoij91ie`aahdHg+NgzIwB2vqbjN*nEE^7M^7
zea}Z3?6DW2=4OKVZ58Oc05;czhe>a<kTFMLi(fe7$)|X@^H_ZFzj9LlejT^XeYtX;
zqP+JHE&d<{AiDp1dv+OYh(!GNm}I4I7D~)C(e1ZZ$0Kj*RA%Shp#H?E#&MU4-(**t
z%fKc0WUtixfr8S4cR>%05i&FOK^xfCLeubhgd^EAk7OZB0)1Lk)cCJsiYV9wciBN-
z42&ubNkZ?@0ZN@Lj7WuXb8{ZD$8I+zwCOd9s}dm}6FoC;zWS2<dJpk$L6Pw#*2=@l
z>@k=>9q?DXlTnk5QhFh4C;?g8@Ebf67#8ZceHJK83r9k%34kxw1IQZy^OxDV3bfS#
zb~n-2iT;}A317g>PKU(+N(p{e0w>Q0>YJO46A0}J&b9-*8(c^izS3zlMEkgYq_9aN
z`{r<SzWzeUR&A+JDk9{$jg5J<iQG-{OqUHBqn-JtTZMsIT7p&<w+vbol$eblS8WMk
zEzNt^Ji~~`mkpw|O@e}gHQ$y~9yZ(8JL;Ee$J9q$E0Gr2&)Ndgdx}Vii;Ig|uqIuD
zyZpDlxs^m5Uu}2P(4w5z4_5-ZEN$BZHf$Ep@T92}!%C<h;{15Kz3$NlZd*y~BxN6^
z{bhEzdBOA6@iE!bDoM}2wDeT5sO#b==SlVC)z!iQH$s+vzLr5n0?T=K-gt%kaw%?#
z8}Y}@=i={Wu=nF)h!CU5!8U^r>4(l34mvL*UVqw9;$zC1Yds3JTbaC9A1*l9(7c@>
zgnia#yU-DVAmD2mtX`X%Fp^sp71g~<$qT_KzSta|p0=W&cWnE+HQeFj1FDTD2l)Rz
zTL7DJZB8<#$(R&!aVrr?0;8zR$*pD4LHH~Cwepj%;m;I0{Lh~^Jv=sKU=ZK_`}ZR)
z^p14Fk-*@9=pdu;`>wNrlg^%BG8v@z$zG+LTP5o=#MwV5V)y0oWmvFs!)xJXq&TW`
z?q>gAht>w~(|Hc=uH%_4hqDVpRt1~bP}bVu{O>mm-tX(z1_yt%G~0pwzTx6|L1W)P
zr;KGQmt?};=FYrdlrOlUOFrBW6qNW1$Q3#<pQ<;R+ihlJ7g(J$Jdq8Y2;Le#)6dF|
zGoqjoK}72SB`KL5Z;rlEL+UWU@wHmBGCU!2E4@LfIcv0j+^?uATWVfdmyov!CD<w3
z4ILNM*2uko(Ug>VTTgv$KWk&*D$+iid!ci0&1ms2^>8p7o)u3J8)0eOYH#+Ca!YyV
zv;9|pptKQqw*{A0dP~QkOJ2THAd%P^_69jDh0p^QrVI4%4_OPQC4g#a(h{^8wKPV$
zPqE|+VKHCc%Nfd44K~EzflGhZ-RClK=5=**BMf#QzpcNjR*6{6^08|Rb@^NfG@g(J
zJkI7<1i4CU#+W7oBxtFSzZ3EjqWUD_Mm@JC6)h!!!t@e{6j-Znsp$gO)yR<6#W(?3
z#O$-#LQiUTo?vSMl{^|jO5TqYx>V+=pz!R3zK=bUfBs?>U2j2<g{*M?=)J7GS{Wb}
zK+}uc+#Cd;%o?%*?7o8o7)=~0jCJrJH1crbb6GVHTsm@KD5KxmGdIOi@;}}L(xV@g
z5VDJBSC#zm@nmPQBOqciRIcvfKp_s%Em+Ku^wIEulgvB@kQ({uHFqVM=6^5d7Qs-y
zbpNJO|4Ywa#&B+8r%)*2?duGlhL0j`qd))r4GOf*nk+(R%(Zvf+c=%ON=e%oH{p&V
zxZ4Fr{8_wa6=T*T-f{8;9Bn+@ngf+K?fg3wdaFbGtZlAnxY7r6%F0fADJwIJuyOn0
z{7{hW=g`vd$s$R=Q87zg>>JRbf4|^;zh|3aKR?>w|HA|e#(c|#e9xY#S(k90K44o#
zmx(nVg<_uJUswr!sbyH;fAbvJ$D@X+Qa9z0&JG9+z5cgQJKA#I;U9Bc)Phiji>kqe
zY4gdATMBMQ*I2kse^{SE7|Q>b*!1e)+X*NY$$XdN@h9Ez&``fuuYNTf*iB`)H&_n6
zOG&Bamt615_)L)>M*UyTagIvlWmEN&x*ej*KGtaQ)zxW=(O|m5k$T|}tA_Dgc_4c5
zfpEL%`C!hqpIfsp&KqAjRgdr_6~oA$!Ce?*-YP))+yn2ZiyjOJAt=2C&6`>}l<03j
zUu~)K9f-v$_bx?fW-DLzvlR|I-bsa<W`Z;Ik$#sOZgmfWZZ7ef)SJb0*0?H#Iak*d
zX0Cqd6Ffp)F1=w8=h`{15gBla#<Eqiq3&HabIwe(8sx#!t*IIp;zonj2a?9Y(kKA!
z5Z;t;(=;8NIIiS|aMb;}j+|<6pSev>X72Bl;Aw6!Q<wJZ=CH_d(^1I8)OT<PmJpL~
z`T7gbnyRjBF9=(~nJPS}M%ibwcoLCGH_`UVa%<m_Ty`Xpl!9b^Ab_uUAv?S6_}L@1
zV#4pUrRV0^<%!Dr8Lo|%{@>Rev<MCKS)!do3FhecED$4uFm&Wo=h;~Cyc<ms7<Znt
zA1;6*J=k8KOs_8r(Bodgv17AjUn~a+4MN=Pe&2@jVDb6mUm&lR`86e+M`@E1qp1k{
zjK4!R``va(XLnfr)yfv)Dw*WGxt_1vn(4YXNTOAE#VS?+%Q_R}V44IBkh~AbQyj<D
zu<8gCrbFn!%B2G|3tyVwW}K**fyy{`{!15RJLj(}7{v#Nm`=eS5nN+IF%)A<A>WzI
z6JeyQv*z&ujR|{PiH5nxxcWaz#<=$!F45s@frWLfkG_nBgew-sj{56j!Qo3k-!HpJ
z(3J+>!b(qr0extCMxfcFn~c%Xi6YS`AoE9mO$3c8^1n{RtEyZaYUBHw>}Pp7T8@wW
zjqy=WfmYvn06-B%bznU3Qevfzn@}9Dtz%BZSi}!?0)TcQRtNPHI%HAt4V^$3&z;9C
zt{$~Qc3lA=r<$$yt>y_7DSwdB4?({ob~G0e{{rqf4bA#=+xg%~W~on#4<ip}%EH1B
z9as>wTw31a^-#E5$ljj?k9v%S^VQ0~V@$`!MrE<$uSJ!I?GN+Js*h$Ai4k2@Er8V&
z#Kkg+B5~i%l(8-1WWiH1<ap!908~XuVPSfD9^V8Dk}aSeoF#V7%Wf5|8cs+@Jha8U
zk|@~8bWw=GD<o0i!gAQ&b34W(&U_G8D{l!7mOEmI-8bCIM?Xp7NM&XxU?Hh$9Y3@^
z+hX!3CE~9Wt!5VKwK_;~R_$<g(S5P3LK;{eKj6N0xiJ%zER!9N+OUl|10{CSzg%zM
znO(dJGIq7fet11chs|xbClc?UoSfh)rRm>x8F8A#U*qeq&*ZvKhqqk~0EoffucfFv
zdRRc^sDe8A2N<oq-@gr`%l)ErlC+OlbeY5!6qHkwK5J`bMrmfMtEF|4(?N|smF!i8
zlj?|1xDhLzbgMXHWe=}s;beTJ@hq{)GLaqj<{6ztYds;6Iw9tD0V<vw0Pf^(KU*9^
z%o!`ucct?JyXv+Wyyprn5PHy%tG5K?0Th`X3qF%Da~YGH{WbZe2LvkhZx#_D=<yI6
z=B%m7fo3c&#uhr`pk_#rmBp*Z^k0@+2exF!>EF9v;C0>v=0R%Y?oq&(lTd1p->C!d
z`!hhZku7Z}Zy!lD4v)5ExQ}~skegHZ;jL#GieOzrf-p^=2#aBa&^>f#V)PLxUs*tA
zJ~e)*ZyG>B89|s2ZYf|DDiLQ0Vh|t69&~T`wWyHG;(}4MFIhm*=4a+<ah%O4GBZ<0
z`8A{GI>Pt%oebEzuabRy@U;;8zs~Osp2cHxMT29247z@d#$S@y#4mOhuadh@dC%V<
z_9j<x^zxP_OCJ$Ku|nF;j3%Grd-<FmL5;L|LVaHnk84TIg-a6Xd3w4ef_Sd-D^IN6
zp}Uwb;oTdv<=01Z^a)nxXweqDBwZ@)dvV%jQLy^L+N!;5-mGW->{kJn=;w{5H``Js
zB!>m)282g0JvVj5d6Y+xu#d|Ms^&k)KE8x}#w&&uf4YJAB9Gj}s4%3V@Ya*YD<MBw
zMp>vmZbHpmdcN{L+3(D#a!)2RS1$_^&VTs_@WqO5S(Q2ExVKD%{f;_NRU6AARQ-v!
zTBmlTM{$6zU^PY+%cG`C^2eF0sXm4;W!6aHt@YMx`PI}uEyRKw9w@1F-JvLjRr0^l
zhSCFgxylhvy?ayi?v~&*)fYp18jPcLDc&Ij)}&L3!?5$gQ_nTbOV<2QjIEf00GatD
z3Uz!OM4W32^Ci6n$N5VXpx{`0^NNgw;B4UAiLAk#5ktfF<~EW?=Bdxg;o)~7wguYC
zqTcAY{=}kSqlY1vM^lJGrm2EZH=mI0F)g5Z9zxvy0r(8ZF0!wmm-L|HQ}<%fb{zM@
zgI3>{l6hR5wz(s&v^pkG^2e^(&2>^Gx2;H&+auEE`V^x)J8Ng(aV$jcyfEU{fn#kO
z3_cP2d;d7Unfcl3;V-ADow@eF>#KQp#fS3pxrTzm<)xyT8Gl|80@mYDHGS)zvY61L
zO@yh@!-I9+6lf}6<mTB?{|Ct!eCmgvC-<3NZ6RgI&we#|JmC%@b*Jf+{jsScB)K<>
zX|_w$^U;sb*X(qw`jUEafA?1h4^HXLV!JAh?<P+xRsKg}bmZOJY^PGCEF2-&u`!_`
zZ=T0c0?ecOEImxqpRHrcw?nrv=oomPFrrt2MdS5<>Cqo|wfNoFB7|oxZVODjAjPk&
z0HX`7AzL_1SZ^brx=8x%HgCDD_WJ@#1<TM|98R9x=UpiWA$hB9{u<eC3%@_vNyX)w
zSeRv;8Ya$mO&@!y+!I;0)28V&6iCXRXs#tIpAHalua%<sHrnc))s!`u7{jacXcV!6
zWqr<>JX7lk(x<0kLBh62EpWHKJ*}lnOOS}+N-jCmjex9VM2n=tp#9c2gs`e|KLsm)
z_*`9VpQyC5m+gq>XprU-80tv7WQeS=E#7lv2CbBH5)2Pp($0=&F-z<OI5<z7{w~*}
zKBrbeoQIVixBGjeQT|U9-D=B+5XQ8RXq<EbWplTM!li2ph>@2Caoy)FtaK$j6r-tB
zbp5SNEDa6P*eL@Es^O3H)UpLbD%Tz**KUa2xW%($2hoadJ4b~g+-*xf*N|uWv9t*p
zWXnZtjAM)@s^*Qd>hLu(CQzybgB|hiMk*4f6OBUMvEaFpeCwfzynMbcBHJJx2mqR}
zIK{It>G8qgi7zCued>2i^exOCcX2fSJ9MS6!#A*Zc|~X^uryW@eX_t*CGx!oDrC+7
z4N`o-o5td}hW_lR4RLjmkqqy9XYh24y7i`@ES3l9MW^|F$OS^NoUGm8{1D2}6Kd3l
zN`zU%3G=@Uqdt>^zVIQUa;4dwp4K?jaL_ajkgJL~7I6rF7^44=mvdxDm>;UN05YVh
zm2v)IP58a68h;Bj>@|yL6Ss4Ir(u<d{dLnN&qi|PzD*(w$ENi0aA+dqGoN*R^WdtV
z?ETo$n~*G*>EH-+7U=$klu^l+va2#$pf`2(?O-*eB6U1KcB=#Ndu8RS(*7bw0)trI
z2~c$UJ>*t!)V`^!){w^?`%Lr-us8Oj?%~-lW^LSR8)UZWp2y8<R;y=7ogwqAglKUQ
zvWqP7tI)qwR@s4e#LgR!&F;>`*4~8hIGUSl{WuYr*qATtD7pBVM5^%St{t?Y4dH3F
z+C@#jNMCE_)ESTP<|z1$^2UgpYQsK&N)N1_vZjgpzXY3s$-aSv<>Il09vpc?ZZWa(
zqtlT<tQBPw)|bhNGfhYMFoM((h!sUww_DscrlQdBQg`d4ct`c64VgFC2~?NBDFy%(
zQR;V`&bsa;mJ;2a=-uzMOBb1W&DL!~%2zO2QC$niUWo0Qsx2%lIz{^@y=dBvsIFNd
zMRQZLBu|spn(#|EqJ7)hN0@L$pJ>L$A6y0hK!gl&qJo<|QEjD}qtCO(800qmQ|zK$
zd9(Z^o$;4sb*q>9Iu#+Y6p&CD4YOhakdE3ylMclqvFwAQ)s6laXvO9_`PK`0{UUbB
zs1K+_gdvjphHOb@#`KE_&OfPH<@-^0Oc!5D*Fc@s7kh(xr=5SuOPiY{lD{@Li>-hd
zY75g}_V@RvME;u@E(iuhSY$q5-`!vJNO}0ucW*PG`Nj~eak5e0OYbYsXiE>+L<5S>
z#a%kL2#I)X+xT=j^>au%OYpN4biRR(fIvtAL^A%FzRqI+@&Ty#lEa%AzKff32y(di
zS{dQ0zXq44YUZ<oG4?>-GM>c4-s!-Aw(WkabPoT%%3WQEckS0-KL?NhLM*e~WOwCS
z9uwMpNdE?MAko2DZO5pC68VfmfzP_d(mj#S7>S)I;18Az7#uIaXzp@bGPdUSJKjVD
zY+#;4k|JN8gfD^skqO)2c@UN&H<zR`$09Sn$h$9rqZzr`g)30bUliD!IJ-Ch)hoT5
ze<y+jkfR8CwWua4Zu+L=2y)k>1%_`D_WSYODLD!aTtAiz{kM}tKFZ7PyipYv0FVyo
zZ@MlQ^VCW={KLQG#8F#(T>ZE4u>?U6Sgud0z6C(Ku^SbcfF3&`qhFQWcx=j`&Rpmp
zX%F~Q;jvWsWru7e!W)HBbokQAo5>RT&bWx?+fwZjF32v}qMmRNQamDCU|^h$&qs;Q
zR=mWogwXJqn?F36%g~YM{?Wry7;La;NH|~&yi-Zt7b<DNc$+L06AXzt4M!;3XPQ&;
z$Rku)dz>@MMgEGqi9}J5LajlM;NKOVOcl7I9pAw!cv6kl!d#}Cz3ir(lFEeyA{N0;
z9d)y_v(LuB-vSitw=}<tJU3p!%`|%eoIAET**!2KiaAvxjPH13c2x;&+`qu;%Zr^R
zQ-{if2|_dh>O-B>fCjM!LqvuDTKx0eX;3fN=$#R3lEz4FIb(TdB)G=Z^%13xt!vP_
zHQmd=PLepo3uEPjf3Z|5q^Xe6#cgS;3;Ag2LC%0K3Rz0ghInYhUeXFDJe|dhl9<GQ
zksW(D>S4EY(WyvagbL62j$@^jp0Pf};O+S1xan}g7RO|HT9MiY#9(-rttcs6ykS7l
z$Cf5$#`0Lj_n$(E`HDfGciUPY)Qbvna@QBM&(F?pp^vZjgl9ZI2fxqVLEU+5aVv)6
z>8PTU#P>&OYf5pgUUCHtz0Sr$!f!qeHq=$gy!>qQde2>t)!93)IRSY`QHN=IG9q-F
zx_&YII-9g%x+pB>A;hu=FE8EOeU-J~#y}XjG-f6+`4i<MH{>84!N1X4(yp~;{Y`O=
zESB&*`E#yiY7Gnz){BSgWImb%FN2xyMVOiN3Sh3*IPxk<opjUeJ{JW*n`lPbYFxU>
zK!fLB8>X~iuE!1+t41VoPzSN=;DGL&<Q8O?B9LZyhP@EQYWl(t>HTR-8Kk>@MDa7_
zGJg!P`!N(VFeKb-T%6Mf6m*KWqKEvD*F3H1yEy|c(`~@|0^w2QrC)0RSZk@T6Bnw7
zYY#@N$%W{Wbc@~mZh5WWFuX+uB};AU<AM5edEnlhOVDd8CT3p#Q3C)}&5*D=i1#G;
zKJV1?4}OS5ZRBU6#Fu%TZXH}u5#ADNMsp?r4<kAX_MfXQ;}85!>+@tqtdldtGk~f>
zi6{U7ULO30Ub=B~YHaM5c|t?7R{WrQwl*A;wHh)N*d-#GH+fdmS3T#NH`3lLh4cM;
z;0G;*G>qsieVVQz`wt}o(5~7#Yeu2Ej0{Y}hzJs{Pq#``5Afij(r{wlp4CLtx6z!o
zIO%{0&e&E<Q63nR5|067uwkRdGp-`vFkO2hFr-*e$K0yQ#57pLTsoq|awzsuK55b2
zamueyP*e=PNZ5J-zD<)6+y-d$QYslCi@KpyQnee@F?<ltGL)-dsnpRSx-=-FG41IN
zMOcd-Z??Oxi4JCWP`g!U<c?o1^Ey3Xys0jlSeVx`+deeF68c-4sv4@7cGMHp6?p@i
z_KAlv9_kK^$acXc$KH3o*$XytB=muqV~Z84P)*K7#*=-Flk^cCb9o%<3m_@2cSjW(
z-+!J^_=ftSp!i=Ebq8N7!E%+xaEV(`gkO+_jR)q$CGUKQA6v)pn*B=@M2|a7{BGKD
zgo97VBr0;b(nrQX#rn7JpN#YBC*7BCTp#+5&aD<qVil)u-ng@6oA+d5BYbLRD9nwg
z_qfJp{>9Wu!FCn>x6VRuoCPP92PQC2=I2j?*`I}u*)F9hCV`2BDB%J3&$h9kI7DQw
zksw-UtYvF>jY1XWUZE;dnm`5`<NICHC(db4wLdH7En5`Ge(x3}=$WroRVt;dGadh6
zqA1vDiKX3jBG}*#)l<yn@9l|+WPca?A=Jj7rST@g;&AU^V)}v8>pV~(?EgAD?{BvL
z_>ae|)e2Igv_!3-DztXdh!qsEYS-Shwi*|;SB+S$O=zvwu3fc9sVZuP*50)@-+aF3
zeE)%Oe$2V&<mBG>eUI1c`FuQcD2r{3dce~~apur)7!4697J)yN{*)v)8dD44@0jV0
zc3Fspqq@`g{Sy*;q5TA4qHpTbP}B8r&+fIY-*)ma*MA_*9$DW@TV8hJAkr78U|ofT
z$i|9kmwJ#;ot(kfhE4CQt+pq#89C+u6&;{$irO*QpB<*(I@mo0WnXNaP=W4~jY_Pv
zkR?bDD_X=NzzYe;xBa<<Fng<;2K)bl*au-yk^_(=8<Dnc91I+9Nq_wL(lP6)rfshg
z315GdTu)Z{Q+aX4;RtL>1P@=dh5!Cbad&@LIRPs4qy7ZqKMw`Lpeg~D`zJm%BLhw}
zqyOYk3~c|TdX_kEwkZ67sBqnU^3}I3k2iYvS-h-_+F;yMPI9Y}uYhnY<1NdE1Wd>c
za*~Ded-V~j)O182+*dR2BIi7_yc>idVLOeDeEf84XNuO)Gse&na8euxs^+<XVh96A
ziG{&4f+K<(-q*MGx@t!1%Y_}sd);b_x`mkNSj+Jr#~j8fWcdVI{B+T*litSR+T?EE
z<gUtuC!#x%`x9tj7PxHwP|~L(*W}`O*E7J3F&RvnSW@!(v-Ep@{ZFFmBI{A{w7hOY
z<FO<*-6P_q(iv)HwXo1h{|0ZL$J`2?flcdmJVOm5d`||d2uJMg9X62U3bkAfJ|)@W
zR|j(d(cX+Z6@|skkA<y{bRIMxEv-{Aq`Gubu$O=H*ff*se*<E*ZgS>J?3Mb=XEw=5
zmb3DabR%|3MqfL<EgLep98LA$bov+Pq3yZG{j2HAWzI(%qfKnt_$Hh*i1y<r`KHQ3
z`uns*3anZwOc~YHD%O-DI~LIO#&{SV^9I!d6^>Yzk22H&Q{&9Sz}`5U)$-koMI!10
zRo*+<tNtA6Q=n3*{guIO3Lk$D!hd^P*CU6YCNxKmoL0|&Dg^6sSfZ0Gs3xsMVcBKz
zZ&PoaC$Dip4L|Aq_RbjIYsOf%b8?8(%-ph<H0%7}JpHVMU_x=;xm%n^@~7-yi>EV-
zT$nNUN*&`Zt4u4D<{P9?$c|OXT2gN%N9k4pduQWKJ^)eK%{y+>``NHO@M?Rycq8Ru
zk*}#*>Um5{N1y)$Qwkydc|c+E!=q{E>0c((gcL73*4H~kxf_xZ2C-|KPUms@X64iE
zlCH4`yO@0F)KnafZQqwZ$|{bF7;Pb&uXOZRHHY=x)vOr)Y{<J+EMs|C`kFOyEj)qA
zl;}2W9|)vTv9z!=;r<%sx$_O*BfGoXcG2zKdOo{j3hcdd>l`e(m=NiYs7Xb9@`oNP
zWj(YUEW*DyrS>4E(}kc6xZK0FW6g7Sy`uU_L<GpV!#dT-v+_EL2Q5d(pbSXPe1QSQ
zO@;J<DjHa<p@gO8j~k3-R6+PBk>Bp=FpzK%^rur_?!f<DZxlT#G$aNQBRV2oo~(xT
zLuI<3C@G~?qM`SjjxFnao>4A+C~#Tz-Cc~XtdtbYpiN7tMEOwMnkOMq9$7S&!Cdvc
zRebtB!j?6y6ftu@Uzs>)mLOg`-X30Ym#;B(E!Li<J$$_on^#?3aIO?!pyP6#FWmGE
z21qF1<rPR9IDGviG{LtAy}*YYU$OQ$^)Q+k?Fxj3=L?L2$^KdiOFU^6`?&YJm}6MZ
z7ud<&5*>s&3=ZTIg2h&EzmUv$<eX#OZMb3V_4+jjebMRQZvo1p&VUICTSBqK-Nz^0
z+5in7-q)HRg%bbD@9mZ#fhtk;c7m$}04Yfs{PgVinDrX_lC{jH*Q?X5<;+#5PPZt=
z(-#&BO5PdSmRB02B8ppCn!lgBtVWDW+(Fugf-(_y<#fTFmG*Zmof20}=G@V+uEe(D
z!AM9)T?x9wILDb&MY%o6^0xIKqLg-a1IP4*wU#@S3Y=Zy$SiXE>e<Lbh~=6f0PR$<
zVQ{x|a7`#PU&j8refu>g9LyoYQaY&eSoE;pkmY{9r!PXGkUjqeNS1@Yeak@}gdkmp
z7TgMWs|??@tMj^EJ;0ApzIeX*EQZ6Zd|Uwm=ZX2g)=f+_nWzGbdO(pEeemixRp|G|
z{cg#Etz{?4xh^*zmheZ9dMVQl0D6A$uaj0&H(N>ui;}C8tBdUWp3}`#h{11Z3K18J
z2a%>pL?wOM5uJ9aZhYwz6;%GNVmr2$-ICn4e!5Pe__?~SiUn#hSVD++so_(+#xpa&
zg<{)7#k*5{552u^u?1W-Fv)lx_FMWIEc*^+V()(xQqqF=TgdbYB=NbS4S+<J9B;(k
z**0L~>R)-<^-q1j{&spw_m;_@t6w-(bai}Dp<|JV`+a)_lgqxRe{D@SwO?NooRz2L
zY>=g)+8#FkeGWwh3?>c^b@cZVBv-Lhq?T5B?X@#lN>EAQOF#J^$F!Y=ahCBZE9#8o
z=F@!id%>AI-XP-<$<C^i1i00kl=nA-)$xX?$9CtfV%+-`a!Ft*<jaiV3kmX&$Z=>5
zx1zPVV>utg-`^QJbec=pZm?{)iQ|Jr7S)J|uLHfkTH-mG(Rsr$Aq;Qpv;;;Z?uVc(
zDJxQ&Bf<2{7B*l_B6i&|_rTarY!VUd_KSG8Y^h?b;>8L~{T`^B;J+Mp-wB`~0vYQ?
zMMf_##b!j&R157lH2)P6Us1HtN{wUWB9*L?q{GH}f`qt>X~PS<-C-ObtBMxOxJdPG
z+r0U$WKzC?E#wPWZQ^I51^0bWmuGnVIAEc=-?|!R_orOBY6xc2PfRiJCI(&vmg3s{
zCiKuv-l}=wM;aattUlmyL#dGFgVHKY6*`MPidYhKBy0^jzib_CojRZYR)Kz*_S*VP
z_R?1(Cb9g`(<IG0%tla-JtQa;lX_dkOuBiHLjatVPDA%J_$CLe5JY+6Ulx1AYe+Yx
z5`Q*mYfWSRw`Ml85b{G)OV5hj^;;djD7`oRP~`!Kujz9Tup>>)e{8+&E^9w+gHcmB
z8*DO%3K$T3WWeTIoyVs~^H{=Lm3UB`h%gwG!!g$=gYFYc*T)5tU|rDX@gsZoN8Qhp
z#c}(XDBYH00pSPDxoqKi6=Ig(f+Jyt&}6;@&u)iri&RV@xz+i^S}7y%IEr*qW0I>!
z`x{`76eL~&C~ucJW5F~w_1w0@s|ed3+tJg^<j<hI{130<n|Xv!Ydj1@%?9hfI~>3G
zl&;qP8&-9UBO{|+rQy~kdrGWJAOBH{;c3=x#<1F~Cv$}Kr__NKzxwrx$rFM5mSHzR
zM*-B)hp!zIN=ob^g8f^o-NDfHv69}RcloftIz4v6K-=p3tHssl+S892;$U7+pmt51
zgVkLqq;b~0rJ<piC<+QncezIfsrtGj_5S{i>u)@QO%Rh(Uk)~@b52jFa}2lH^iDg^
z2iX&gpq4*;Jg&2A1z|@4C{%os^`y<{BG+$!{Lzyy1Xt)<@=LW)Pzj7wCBTOFRtSh(
zB?`h0;B<7z3jh2mTb_Ga%@Je2`_oqTuBCXR=Z?np!7xFV7I;45s1cjJt1SN7BlZv;
zOCa4wG33>G4yfFExc#5RAmlOQOV(O-)CYEuTRby~42ZGQfRxyEzg9R1+^{d8mVbN?
zYY{sR%h!S^1pd}%Z4;`FJ?Z#jRq`!;nph5?FyAbQr`+pJW}c2b`+Bi+ZNH;#!8uzY
zQt7^ZqCuCKXpoWexJ}=cldYJW%faR2`D3mOXOe95FKMDPM??R<eZILv<FgXx{Ppt)
z|HDS6amDbP+UhEW%F*gkZ^UJiKCO@aE2#?4ynp{M+qOqMlbG;x7uQWaB8P<22UH&h
zdjX&k$aLTKZgqc#hO5pGv&bXO%UEWp>wId-n8RO`e;;|0J+M#oB88D1ixJjBPcj#S
z-6SCiuUZ27CKBu|#6SJ)NBI`97h*w3$5-1>ac}&f3Y~cIo(RKOoiiRB`gy1`+H=e0
zi1nIvI@y~v1D%8{qrZdUPF?65-`v~;2;T94Icl<>t96Ii7mgcKM@x?Qp2iFc9DO8M
zdC8?R=&L|Y$tEXm6}fU>IHqnZRtF7-*Mjpmvc|%bl|YsbAP2>K$Cc@{PXtaspOU5J
z;NOEUtm-(OY;%8mN8tLUYX3I~{6N|MRUH@^-(8TPto$cSOwl12^;Gq-hbn-Itq@a;
z0P7kE=YmU&yrw9Aps8qPGit^UV^nbG)$-0*T9P843ny1$2k}HBRGT!RxXu#SlMx#&
z6D8gW$x3*`k~*uYi(@g?N`>RF;SdHPLLKPdTls%#*a)d^a3SU^By$A?c|w${1@9`d
z7ZyvgC<OoWXGNOItlTIEoMKfRvy%E3Z4FZKeiNRJpM4w7TZbP4c=R&BpjC*qEU4Is
ztwg*|pa*HAWx_9Hcpbrr@aK@OzL9p-lGxivi!aCPJ3Jiuw8Hb`t#-&^a=gBW-D4fe
z3tRm43rRbHFWc<7ZTBm-GS%94*uVXsmY|hSlVaPVhZWGI!0{19@v>`1$Mb@CqKYcS
zZqdFKRBuUK|9Lc)ocyM`t8e^htgAV05TV%q0c`1p$!KKm;ZzI8@ulud>?Yet|5r*X
zBTpD=vg+S$$G>X2$gT52ynkP!%x=t9LrHt@$q_8F5%vS|^nh@yFih$qMVu?<%=(wL
z>IP0~Kz)T-i<h6@ii6hUafqU0upH}p5!D<1jgdd?*}}D$Z#Rd=ad7^*x*|Nde?9lv
zv~|(@r~D{oI3E|7RIbj_x04S|IGw+g2~VG@3xKO~2R)uY^0?S&+nxLru(iRXajSJN
z<x}90sF9Su4jeb0&JhCXZD-&JvB(QWD)i9;9iQEO367pO0Px6OnheHXC<j*|iX4Q~
zD{(Knb1SjaSGwQBtoobjj=5FN%owDk{Ppc2wtOtYY>D(hB<ZFv{Pk}iEUV@C#=5bj
zvHaV2Ea;vO7Z)cQeml%WjHK5|QD(?c303n~xA_=JQF`+g1>$+~(xF5n6X^34Z>9(U
zGcdf5GZYvK1zQ_fRqrs^*sF7&?S@U^oIZ{D-p}1<4SJoJAACLG-;=5Loo&d*B0%ZX
zYj0${LC<EW)g8BJKD0Z}_lASCKv}KA@uAry7q<~3&7bWKsR=Ky{(NRuM{Rxu<l^7*
zLQ)PlJg5sSPorl9?;N+{(+q_=XoG>S5?%``xkt=n?dOo^RLUdnvB<Aq%KRp1<=;}w
z{@B?z-pyk1udl8irH$E0^A0(-w@Z_kolRfw=PO7Z(B?|{HN_X(Aq+<B+etGf1^kOi
z7J*w&cv>5DL$SgzQWP&0OaAfYS1bX#I^Qy`ND;dj^E36%tXnumRIk!-t2>}d41#09
zRqwsW(<d#oAl5jThb_dy>S0RAYVt8la$nR&Z!O_aF0Urz1Lx9q$x<gCfhK?BGZW4k
z9<Hs%*1kZzx(Bbo%00p=6Mz@P#H5xG4x>`$zKDb;^$42=;|c|oM3B(9B3gzt<E8H1
zd{2tLh}>AgQU~dK3?uoDfNgaKbstn)e{2A8fKD+?k>9zM!#3uL(?k{(1X22I{2Xfk
z)4sh{G=BD<YVNx}{x`m{u&}$6oS{%5FDa$5rcRhqywN-Z^N_0u%T7mmFOG-O=X!pm
zaO#7SQJSEcRMy?Wp=UhbyVoYm1RTSZM~->+Qza7^QI$?D4}7J8Nt*T?)?utpzbFA8
zYXS1MU*Tq$^Sf@c5MIwRJ9@%bu09#zA!*ip)NO>0ltG6PM<#EmYM)lH`TL()xnBSC
zotCohAF(<=izB30wdjspIy$2lTW%11IDK}HBxR)a>g+|mW?m9wir95(eYeK+3F+ZI
zMH%$++Pdztolv0MetAon`nnKHI%OC91tV9GaA;TM)KqOS?bX3wZ~9EQ?zocA(Yh8M
z^Pfr1?DDUEe7r{6VcOi)%+6JIxe~F0<l86u4Vpn%iY_$2DWh)w{7jG`cQk^jhw23g
zjyPctmCNH*#BnkxKj4+Sr@1tnkm7v$J^^mCTn7Jz0M6#M4SqS^;@o)rv@czbDi-0u
zmA*fZ)veQpD>NtRFJ}|^$k<6mETZ9|Gj{HW!590UMl5V+heY!K#u<q_B}o~OP_F&M
zvY$z)@u|&Hz2f9`h46r^ccx?(D=k!uB?w_0#BE6Q+UoAmitL~gAu(0{ymr`c<h0%Q
zeJGPJu&z5nT013D&dMS}(&S~Db~6I@x^k1z(<ET8fGrFX;Q}4IeJWby;WfqgAd4*&
zLWM(2LHThBK^S`aoCjySc7=sBdHk`7ANM=h!L%BTiZBj`(qLbi94(F=!s$)4L|8LY
zM*95U_J{25PVXZ&$2bfXBr*uycO&9F(LxLO-F`Fjn;{{&me+y;89Dl9B?zotvQ9-;
zURm)(H8Cm4T8p<J#E3V{+7h0qt7-Qh$!QZzihb8FxMC{w3N2HlWwMn27woSE>0LWm
zgwRbZM4*ivGa`E%yzC1x#8XSglrNo2YeJO?23(RiZtbez-b@f_CBlXRGJGlqw+3hs
zj;4E}f>M?^8hLtTm|SG~{$;@2tBO-bVmiV?l#yLzZxQPosZP-@(L+F{-v3|@4NX9*
z<avlx2Gidc+8-H76x>{1<_-qa`%knMhJ4J|s|)`C|8hEsv4<r#yAmnq`dpm!jW2%h
zc*CF_>5V__oM=5`YCC0V+dcTx0T@kx!&#27iG~kU{V4j;PeFu<3RLOk42_Q>b^F(`
z>>*DLymn)ycX8fvaW;{Vl5%U!Q4_U@j{N<77=vy<B~>w$d)AAfH)(mb?}7iWF{C+;
zf0KmzMk-J-cXeXnyF9({+WoyYh|?%O3F6weJHC8$xHQSET8`baeZX@d2+*`YnU!rm
z{`V)@_6A+2og}{DwQpm~M>I?1Guk6~#kd+@18Zi*x{8*QlX}w;_6-BCy4LolsqW#j
z#VgU0b;k+Q7Pnu(dDHPzdoTs&PJw7qL2DL~5*!L5Ro&^>-F0f+oCU`^N4s5MQe&|S
z2E3Yi8>TI%eAgzYD`sc=z&Y(b|D^%5fa7CXGmU6D8QYwix{MwoN?7Pp)z*2S?S2)w
z_bcK{Ixz`<@L(VGE>i9(W*wW4Xi$U`E$y}?XlK!r<rsC;^w)+QKSeX^k8x4GI!tU1
z#>rvQ>$V^qxR-eAz{mH6G`s?XbD13TBr~VI2M-knCAhL}S%utzY)}^#C3C-TQ;_1H
zH1ELd%6=#^YdDG5TGx9Ss~P!?v2<7Jp-ySV<;}0r)crYHXzR(7cR76<NuPL%o%fRs
z$|rn{a~uoY{L`|G8MQ0+@eL>$S+R1@3C~lz(`7M51*+`>yKn&~^Qi_`jkjfrsMWI-
zg|U<LnpBraOg<@qeUWHQbqPkQA=36JVb7)({2Ly)H#t3hVE~EzUGz0`2*dfzK%z)F
zd}&@=$FK(~P9Faa9TP53fPE1qWn_F74pHt}RpmknCjE;~EO3|jUV)`g|9#AhiX`;Z
zu!?=+0`Oe-ijwH;EJT-(2SPEJw=gyclxS&1Ztnz}lmJ%bBVn-o%J9g=2BBE8JK8nQ
z`VdjsG2iJs1!N>7;FoAhjIcFb1~L|dPOPe9i~O+Tr`TXh@JLX#XA(||MWgo5x{ONG
z25^s;@{L5*-%t#7W#)Si-NfG>vn!`34_b1|nA^Wt?!*@A1<Z*ZT3K3JHaKZh#S(){
zOQ;ghfYym_)_eDevtzkOO9UG${8UQbk0_4CnoKLDwtKfrz?mgpQI!a#|6&7nHDs<&
z+eF@t-?$MG_8@2TiNq9e)IaXkCjICt;dfOcyt^~Ekd!ggiblnrC>8}jA#<<tvUDHV
zlm##~uif|e>^$rnr0;XHLnHlGY9FPfL?JLJ;1=H7zB2v68Na@1%kk(Z8bln=Xhjyh
z)TcnFY$J}pm(ybryk3-q&ev0TOhtE}|09)>kXbmEJxEGA<8Ic_N;wPt-TM7!A<}nd
z*XPjb{!}t3D;U=al$whOCwvDK$H<Z<n#@uoiZRL?9NV>-dsvnfWV3;BxA(KH`$riJ
zkWDo<)KvSu4jJ(#Hiwe0T)Is@NF{kwqGZ(lyU%JLCy)W=rfW)AuZmJ5<QgE@M#ZiU
z9A(dTY2VrG)dMRg<_a8pGE}ex=;QECjH;N2%Rg%aRa_xt=q6fqNB%D-+3OYN-*(CL
zByWzjD6p7wbMT!v8)PEFyv`idq`^TzkQ@q5`Vq3c90T`V0rM{Dkw$`E)OfjQe#!5R
zQtnTg;*6*<$V$2`j^Z3L(AsIP<qidPPOW-Xq8x*uq|{$geSCL$Qxp(1Bp_l6_~U#q
z4Lun<I9dFYrQtr$fPzt~o?`*lhMQX%(Z31Rdv}@QAR;!C=Wf|@YuYlF{xiTvUZAT+
z2^QH(Zd(MssC|k;Tg6f0N#v-s(0xc{m_;OnfvU!^2Q}iS4N<d-v*rM48uH{<2|T5U
z)<Umk?l6`AtRm!t$Bw{L4F_6xQs_OkzL<kCm>0?_dFAR^{<&cCHZ0nX^vE<{B^!&4
zQc=B~#5O@NYxqY*%4M1W$#Pbm;N%xS4S9B`%cFi{x#((L;(h<V*Q(gSVe5F^u9)e2
zpi$yn<A&!h26qP32m*koyb75ZHLRU*MmU)95(7^$PQo-oYxu;$DjD<)A&e9wJy18B
ztHpoMZ*z&nbSh|3z{1VtI_sW{8cH!WA15dBrs%P2=4u-AC3^$<LxgZltrB!o=6AVh
zNC-9wgo!-u>YZ+{s<@c9fRgfv(O8=oCEt9Iax8PkE~J8mNouDL9YQQpBg)Y~oe~!)
zcLMN>OYH3Y#<k;}lq$Ip!EgmJ_I~_%qYF73#lOu>n&+L+lt-}8#qhYSElft5<!V^f
z0s({hu&Ikx6VUvLxY=qZ^ZH-%R#si*m^T~VY)H10dNy|b^e;^kaM+JSlJ8N|AZmVb
z30YHO$RUH9{=*2Hr@mBuWOI*3D$Cv%eJnj)8(c54>t^`KI3R+e`<8!i?6T^1Kb4Kw
zGb8Jg^|B4wFZ&E=^yD%NaAn)A7rD1Sw&g8zRp&ih1?(eJ!pQIA9>mgO<qmI=KU0Ck
z;C80S&|KE;3vu<O3(3WE7dCI#`^sbu1Q&m>b^ZDMV0#X@zE7y9FB;{{H2aSsdaHWP
z^&@q3xVwKe70R80;%YHd&^O2Sjg=#HSAF}J$qORR-5h+Rx8aG5OLhVlmEupfm}V!d
zHb~&wfarpju8z7$wo_|nq$yz+7^0b|>|hz{e_r5vn0G!TdwxVH#+S#r9ZVJDb#x1y
zRZw~%LeRT6#J;$=gl0YSCfMNN?nB4~3nGxg5=U4lDTp#?iZJYP<EeI(eF_sPU3zNs
zQMJURPw{XrO?yxf3Ly^1ja5;r@byK8f&fs6im(U|R^q|C(BAKU--T#4lo^HN<U0|>
z6hxX11))jw%BW|rq)X%}ii|4-lTf#gf<wn{i%|bSQgQr5i;I=XyGc%<vrF%(+9Z6V
z!15-LMSu&x8uO?j@-QJVv0?M&g!~$>EIAt+eue1PAHq0gxC-uR|KL9x32y6KR6vj=
zeb8RB-ZLv^92U;?ER-rinO{MM5>^O?m&Oz(DEjPVh$G}Uf`yCK!p%RFU>xH}@#n^p
zLNwV#H6w#%HQYt!P|^+|58mgWX<&rbKW!v5!#$pJ#})JfN{+*#o}vUaDMq9|g%Jlm
zzSh&#75{`C8Z2a+c?TObyq{AFSVe*>ou`rL7tt_^#9DqVhraGj8k32o0He8MUEMG+
z2um!eL^x5MstD6z+=d>jGX#7dr)Oth{MdZzYlEDT(1(PLONQmvW)Is)xb(skG6$8G
zJME$2%~8LQnEdpgjk%mVHD}gF5Hbm-X4iM&nBI)M=G*Q8np4W<pgdsbTx<3qpiM9;
z0ez-S_}vF4c)2A)r?UB&h7}1_=hl5(sc`7CbGqrICAM&SF$f){eFHMk)is%z^d&@U
z*E^rPy?b{#so&-&Yy_$MQ6~qE5)yNDGx4jr{&gHvC)IR$#^R@ueJ?wE=5$S?UMyzr
zWNBjbua>b)%8&{K0?4o4J&_}TWGxONIkc=eqBuC|(e$U#HzVCQ1v@{~=r#VRtFj(>
zI2^<)%x&YWW<j1;C|Fo$BF4JOX4(|Eu^eZFzhPq@i^IB@wk7fU>~tqr)`e>hr~Ww*
zKucX-z1;~ob)1tF&lcI;Jg6yAlA3S@&OfzYcd)e`cQoxh1UP6D`pZXpfF%Z_CKi`K
zi;%bbj2XBj-7#>Y@H^ghq)KoYytWF>U!Z0~i=?R&mI+%}sBY;Q4({9?(Tc3NGhI^m
zo`r=cBH~;ld+^y1g}Royny#~RJRu5EOtfg?(^lYFvg_qx+UDD<#esy{$3asNcvSue
zZ!fRJ)yYvFU^r)cOxN@BOdW+H3XKwn+iT7z_t4&(2{@_rmU(6Gr%h>v?iJ7~YiiNg
z<-)}V&u1>#+9=+_pXa@mf#~9}bZGhUoO!;jOy4BkM&aapgM^T!eSPU`#zKgp5w1qO
zhtDgO9#qvZVe$CELsy?D<;WSH7dZJ9-KO;01xav#{S}@N!yy(99*2#gqppb~T3hEI
z3#YxK^)ww?*SAee8eZG;=Z&)15-r4b%B`%8UT`seMvqmfoW^>7^9^u5Wmhx;^{(Fs
zDWgtf@#jt!7&Rr0d+gtQvtK+;<D(1Px9JU1!-68CENIl=gq1)!3i%?kz2T{&0+ybO
zO&ZALSK5bbQnBf;ZTB-GLOr{srvIXNLkBH*VTE$+=!PCo!Po3Xts8Dk5;lCXLs9Cm
zPJbC%1iE}y2*fTDI<HNmRv4;-JZq=yG+)hPscH;PPq2rQ!<VAZ*!xEUC}1%KZq7=9
zpW$3q>*w(q>2OlBRNbP1z5JDDBts1Xz`9u*?OvGPM+4rlmAeAMmvvqr!p1U!cZ0DQ
zEPZxNSs@mI18WaetYEK%XgY)z(?tDHa=*HHt%EDe-tAh^Tk;L?g;E5ypEs-bF(uz!
zCzKK9uG^1ot4FcV9gW6oE6L@A9oe}^Z@HMJh3)^r&qnM7T=#s!|GvD~326QaDD8L9
zvU}hV988wnXu2p=tnWpAE@X7RT+n%JHD7NEL!rom%+9(SQ86}d=Nw_JBtv6ToPsE-
zSdV&8Qb!%rz#ntdVta~NTZ(z|ZR_E3oTL66fhNCIZg1{sz$l=(YVjiagFo1)=@Y8z
zE!$nfu2EWCGB3f!w5gK77Fz*^-rjrC*mit?u#End6z1qUJJa0Yc{#?S?=nkTZnNc2
zaf{+z3B2?jDFs(B9YJ&!MM7XGV{_w@Ft9r>HpBlMvD8aV1ny7ILEjFPraip5<-gNT
zK-|f!w@fJL>Mr#8I9d`IQg+>kUjJC%TyM>t9hd=qtprmjC9J#_OYmQ3a1=Z;Tvwf*
zezS2;rcqY*$uz-5F_Re%kt5<nCK0dwx#69?vjyxi<YcPx=ZAvO>kv136+WD$n!)~b
z-PV-&6w(owplu(6;>1x!MfKj1M-)M#o@Y+z+LxzbItBD3HFB{**4&Tu#w3wcl@2RO
zZCdDjxkv=N5Qw@P@VtL#B@-j8K|WlcV}Dj_M)Z7I#t<+G`|8;OZO*m-v%gKe!gfAM
zC(IJRLLxWb@K=DU-x(^*GB7At+S-3g?QcqUS^esxTyv^VMO4A#r#Z6Ktzm?rJ!CL>
zF>AA!{xjzG03y7_PgWiKLO1IvmRITBX-IkPephBj3Kl^Ml-G_M2!|rV2|10nOIMQL
zsX>z#%lv68qjnRGKbwm@OjROEmlz~0)evQ)zfMVJe~kJz)Q(Gvrv=y^Iba-}optp(
zgbGoEPGv(xuG4i2TF;FZo4G_B%b6pQ#<p)PVuST?QwAsR&F<L@_0d?yZZuugxR1!k
z_DR$qtrGY@pdsIp%`W#n(9r!@f2NI}25_9*^ND^!CA;+|j0$dPfrGrPkV7}bK2ug{
zcECO``@n1b$xT<6MsBdDry9j-L2nq5e%79sz-y%i4d&R<+T+gUjZTxFC3x$8)1JRi
zu5Tpr4|?6Kh4q}yZgBZ?BUsaCZcElzCuJ-}wO|goI@B<m6`XN1&=vlYq{96a!xPM|
z38UiBX?#HndIRO4>fYojuZz}<e>WusovQBgPfe1Ql1VLmzuP@;Vg<7AFA=!$2dT=5
z;V4|3w*FDHAa371kX+Eh<2T{jf{7FDx9CXkPTFsEMPreTOMR}_Q`2qxMmaQyN1hGW
z9y`F*ESt3KjNi#WdCM+}SMIh8+kO5B4^<{ZHqBz4_R$WWZ?PKLv%8qUnvYav?l({t
zLb{e{Itz+mYHQp#@VCC*HM>Vo-h0lSbbD{l_<liToV7mjKRO{y?X#H!ZZok|MtxNu
zijq}yHL#UR5yb;b`DI>a{Ygvf0UZ7JX>b5)&(vN!rC|8#?9=enX?2qyaMlA{oDMTJ
zO?bok!&0zct4A{+1QCX(KOG5Wo;KB24uw<<2fLi1tr8AP>{<B8p(th&#+PwU!w++N
z**`eWIO%X95T}yawoai;gXqG35TGU{2RFDY+x&rP^N$dk6rd<yv7ikPyE=hCfb{dB
zv=*vd)0)_o@4b;91XNef4sWsV^iR;7^BqXl%S?h-G_r5JD&=&j^EPzMTbzF)veFO^
z%I)lbNex3%DCiy!Sr4S|C}+BhUxc*YGfRm6<|}!YbFLf6jlDZSJ*D3CH=ZNkf;_)B
zc9O++M|*yh5Rv+TqZk6}ZIBU90x*1B@@fsYtB*TQFWm@Iz!Cv@4h~1>`A&*=5?#Fr
zUG)~H>5G@?lYVA~?=ih0hRawC=5_4`JyO`+TxYs2jF5H>8eCcb-HeG%Kkkf7Z%R;P
z6aM(5ga!S}OVwt)EK-|;u;{S=ee{Jrl${$E14f#T=JQ&_noN{4i>`zNGXql%fBKSJ
zwziz%>ZZFhvP$S+iVWlaKgcU*FVE9p+n(fO%=lE`>V%hUsAlSIDrPukIGDJr&<bDW
zHsMLiYR~ocd$7NZ>X3r(PE*C`xTtz{rIuILKAeL^8*-Y;LD!kSA?-G(qkFrpj8gmk
z`{k3G9Pb^KR}BIW84&k@_-;!_&$-l!4bD6pGD^t&+t%|IMKj;au_%1&R7?h1k`vyq
zNDUADZ@X6eAQoXrS$bP@wb6HOm`qrcQ}#>V^vo2Mx`@=1q)5g0^L8)usIB;zwW!S1
zo_u`H6YZ(ZmX#Um1#0*Qw<EI-R%|S0!#C{qzh^PhSC@#ROd9|ucdrVT`mQD~dgomx
zV_!t_JP4a_Tyk>VSz9xhZVYrU`2FtD==)F7o`3t7<Fd-PW*zkzXRGe4N$t$dm3WO=
zTrb|3yZSYqla4Kjx*MEOb_j09)M27|6?3qp(YyR+S7*nq*H@z}H>T{YBne9o9sBcm
zo`RA)aowJeg$s6pLf-!Qlhe^lYcapQ#R~9^ht_?O^z=77a%+=*D!0?H<G4xlV&=}8
zcCN@SCh@a`cSFhSWM&6&dF^|j?mjoyzk>r5hXb5cNHbqq;%Ly^hmi#Y5=pa`py`mz
z9W7Bd|5&S7OLc}#LMS?_)9{a5KY;gT_BntkL^$p^Kh*|ewa{NX(Q6)fLpQEXb5lof
zROG0`2fReUfr##3Lk5+Hv766Xwl7MZfIr5l>%aAqmI1yJ5EGFSS!$Ev>+eDPP~fs-
zQ9eHAKW%9*9q)dlN=8r)+1|@-k<JFNT|o5VyH;hj^0;T=m=b3r9BS?uy+Xc<7}bH1
z+l(vb0NF!zqQz*-X+dUiWTB41PsaP27Uk6=U72TILt$WThZG%NQrsgN5>*%}J@T)e
zPDY9>!2vNYdAv`IDt|;0CWjyr35U{ajXrXLmNX8*MCcAp#O}p~&>|F_J)Yc&z+%)S
zP@lWOn)U-&3`KTEX;B~F@7n!=1?wt%!RUP8c(di<00-$&pj>B_K>Ab1G0z7gMg_t+
z_tW*1iVctad|tW~Ld0kH7YAs4pfAPqJWvqpxRvX>xm2j7Pw?G-ubki15M=3vxQZ4{
zFh)4+LCR*tQ`Hm`dtF_Fq4(Nk<A#>82}ouZ2G7eGh^M*E3M+^Lr40dr^V!t9GkJl~
z1p0bsse5gOqFwc_Z6jv}bKV+9_wL@6Y4crRpjT4rDYljnwQ0ugdyog=K=hwa{{0#4
zp$+ue=suY{6_cs=b#~mjiYWU1Ph3Xa2x_gab#(c)D$A+22=TR2*h|OkuD&)^O#6fT
zJ*ywK0(3QI+AC$V-sh*Q^<Wq`N5Pf=N{3fR3(Aq7POtvZWL#nP*HZVFquKnA18d9S
z$1p9|@(&K`eF7R9CKE>k$=?+nHZ9ngRF5z0jxO1*POdJMdffG$9wn=o7=EcD#Dtz7
zYB=LBR&4|2_4;;&DpX_g{fm7s{;U;DnkTGlw4JWBovxXsgdwKfNNwhO1$esvJBUNM
zX!rLt!j|YXB&N}S=f+&|DozNkr}(~4QN*F7szQ`o+j`T<tMU0DEiEnER*RCf_Sq7r
zr51XJgoB@sS(={&1NJ4aaI!8xJ#0JPmi4^u!B@8cM{%F7w_UGW%LT^2B^Cs0!42%l
z?;`1uD<fr~&OwZMtPGA*s}uNY2uG>GDXjj6CSVgvkD$M#h7B`F4bfRuq8+~F&S=gQ
z_FEJn;*9iQGI|jl1RN%sB;krn>}8+>QDca?loWB`AcgVeA5+Zp4c)H}%@WW{&xQ9T
z8YI5;t9G4K2UkigD?|<ogn^#%Lh1Y|d{v)zqT#u>O<M76WpFwhz~(Qs&+q{$ay!3(
z0klm1w~W0T#zoH?tEu)iR)84?g7FHYiP(8*Tj)R3-oi7G5b)!mue_0#fqFq;jPx;S
zMt@JyN2Dz8gDQsaxb)D$$l%c*=MFlwcvYodVjK+wn)l<#xZhrzlu?|f1=Ih)uRP`F
z_&IM%EP~mx<#-c#^XsQVMC84@6bB>xxT(6FNB7}4jZ)!d3arsnAv9akPj2p&f&qI`
zHjC7Ux(jqf=<-6nA^`@4cGi$&%!;(<3bArxh&~eBLqHsM{6@h;<=`+1?n2T?`sba$
zRij|OAAj(Y#7eMICTdp6J^ecM?t<niI|W7iQrZK>wj(xz{_kOFm1jo+(t;j@jFf}n
zIG+pA@eJKwJ}oB+!qTg-&*wRpe-tr@Kz4$Mmfdj>qK$pi<Qj*yyfgfmXBL<FVAYnf
z!AFH@4K-M8b90OPYyX#Z>Gg@CMMkDo`y?S0Q8mAi{p}HivRCb06G9&DLJ#eL!#M5v
zaLzSuYyx<HmYT5i?Rd$@@a=oAZR2_xVol}vt_ZfH>&xpCCH+rpMd5GWxDIMCv9K5s
z7~-CfUAmL2T#VadWaX|wAUco=>Zy+ZGjnPCB9?{ISaggcE?w?JBkNd5n9l3whG|cp
za-&-}E^%2Ko+7tmyejRYNEv$OZT2PdWh5MN3^71($u76yW)!zknr$%;uk{nbm~PVH
z{?NP}=WVs%0?nJNe2v?EBguqNbi4l792o&^s}JLN3V!0Jn9TyW4%bfl2eS+W>|b~|
zczGCSU@Sc?kIz1D0tf9HjjJ+eQ`5L#xW|~kLY<S+6pjg}f8IoR7RY|_F5W2Ezx=>_
zyABS-*gVRJ?e6lK;r%#0p0&kBC}{jy@ay5BWo9xghItT{Ba8+7c0T7{HGvgrN_Js7
z5{+V(o!0a1oXZgcs{Hi<L0zykJKVw_naMI$S6zoihk>gZU#d6IbOdoEw@H72*_Kw}
z5u^_jNVQ|ND9dZ<qO|k(v57k9mEE|E8vu9(r}0@(Npe_VDaXh%ECCY*g>l_j@sI#5
zqL|x$$hfQDExTsjV{HXTVY0hQYmpGgXA;+x6)}EcFTP<dq}gVR3qhv1oxNM^Z$k}o
z9A|gAuR}3d%+f+m#A2x_MX56pi50g(Bv@OSh<T#M6|K^*`^GPY#21y1{ba5HNd76I
z5)eT?R6Kau3hVW7Z~fG|6F0)1kU{$5Wou2>5ATK(nSI(U@5$k%YP*)7SDz-MjG0MS
z>{kW@3>>rEPt^x9hG%LdD!TcCFb>dgwK5|kL+#sP_EPm;3FtLr{lq0*iNd?#d3@9`
za8UYwKgJ-nU8%rhPSE|kL#s9ypc_KkN!)dY{^?7bk4lUb2ah=P32XGCcKvDJc_Y)G
zm*}>1NqW2-TKai&C6mvY;_26D7;AIRv*Lu_oANKShi%C`P3k#jh;Rw$kF7gB3$LC*
zZ0m$@R<&M%U4)J=oyC%lu>6>m*gLg%BR<~8>rOJ>f;G*`aCPovaMI&&AjA@~Fw%Km
zhEeC?>3P`mBW<kELDXfo@v%cm$K}fT4IU8B+w3jJTED|H7M`ABy}r2en`RROIF&#O
zmq|;0tk1y(^b3=w2(J26Q#2-DaG?8_`bYI8@(Dqy30lR{BEfR&C@o@)#r;C>zDH>X
zAL7sbsyhlC7BxfHd`i7#0Er#rVH)#Xuk>Ntma0bGK+7uDw`Fofp`>ns<2Rphtxvb^
zS2{<Gb2YzSGMsdJCTK!^L+RgD!A8B=`|t5oldfu;wfx#f^ElNsR3=>3F_hW0E!^(y
zX8T$O3*)fN?zj@(ckAvQjAu38@5tkNP2`WAx1+6O_AhmWtdWbe3(%R+`Y%lOlyEes
zW!rhe_Sn+VN&D@)SChXybwxc*#0t>Lhl`WJa=QnoEj@nTm~<v?dfx|eDP5`tRlv^N
z#ROTulQf|l9+GZu&7T;wSM~)s_XbuL|B)JO8snMG-W5S)cKbcczw>`;eYb*Dg;(G1
z>slJbHgytW+J(K{0J%_<3Ofi~9ExP`pV7Zd#-WUogJsZz3Q-M?gM8v-L1W~`i(Zc+
zK@?~{j{f8UdM+4hXtDXTxmnn1XXZ`b{b34l&`kW5t@B-Xytnh@gcL~JTj(p6cfeB)
zow!1|oM*=CRv(>HD`a$@lJn#h%BGirl0dY>TyfGZz2XA~Gap)}IDc=Se_irB1|J_e
zahx;&*I)T7?G`gZ8X&0Kd*g<<-%-Nw3dSCHYCP0|O2A!Gbgw29^W_IYn4eR9tRcio
zdn-gz5Yn(ne;G#XXDdVAfO++>ud|2Z5#5btLrfw>`Eq||2tn~G4CT~t7}xM%Cwth7
zZm5O9ZNmIxwm%(|TT8x-<q(2eMloaausl!0)AMT;T-Y1eJ%iJ4wjAFA@h1Ly7ikd<
zYTz>x8jYS@kS2ZJjRg}CMVRx(l7T*2*<66+`0rZuZV2MsMo99t3^}^fg%QQ1Md|7E
z96W}*5v6(x%H6-Q^#4i}Igg)wCp48%d2}$YejpFidnr;F4Zq|3oC1rap7jb4c}FaV
zrXfyqF0HNVr?7&o`txp(gW+1y58?T2{&R-`u~p2@%-T*nrUn9%-IpC5JOo~mM)Ab@
znHx<u>{$^@dodWTJ?8Sy(7x!axu!)-gOtP)bY&=Jps5)FW2a9L*R)#4u_q`BX(>km
z!TukyzwUg7y5iH-D%kf%FFST4-xaMj{aIb~k=ePWA!$zqi&^m-mN^_~J1BPr<`~xt
zT-#c8C#+lUF2_`l-m(IPf*-Grj33!NZe65kyIvrO?tv4Zo!zy~7c(D9^|?RI;V09_
z)4U(@d<)KeEFg{~4tC2ItH3rbc#Bp5+op58+;ndsdvQkf3!Zx|JE<hx#)H#M-aEjx
z-`x2?+r{|E$l$x(p`n`vx?0bj)pVyD?=jWI!e#v2STp?7ftS}eVovFh4{^iQ%G&6M
zj-g*|WiLMhm9iHHJ@`O4)63vCI{ym-HTtUP;U8jv#79N_J10lJ9@iT>q>B@<HjpO4
zWhx+(ki=Bo!*hOvfS}yX%G$CWrX}j~*hq43*(hK+b3yQw^-6_=&&+2^NlW@yiTPdD
zvPh+MtK)EbZ#Har2B0L`37B*wlcKwU10yd^YoQ84q4hfuFPJ0aN1zOh6?9a;Nv<C0
zeunZ~xq72->bup$)_h&ibPSM;yy1zc_wp9h5EFEEVr9*KRB#h(bN$l1^t+9lwak}e
zeA_-=V>>gOHTY%I^r6yR^KPF;v!YOr)2~j$5GSUg8ywFRKO@2YruTM4^%c=Z9D=I%
zNvR-|qX$1zy?3uKI<7AXe%%FHS^pxesN(yoY!?&37(>T<aoD<7zw7|6m}G50T1uIS
zP1@hp>nJcf0<DuRu2rL{Jhp*^==#7_y@IKY=b3I0=4K;p{jSJJdK{;2;jdp^ga>wy
zkZ`_fC1*HZa5CSqbEn1E{=9o>cHqrM)>{RG+jWA@lZPe~^_yHu>c&J}1q<e)ev&0Q
z>O2m%P59Ya=v$W3a^NlN1!$82ryvGl#8yQW+7^M;M6{I4S+3S^>~!@uva&sxlG&Ti
zIs28oqS1QUk;7^2^~n^#(`o=GriO!DTn}_+A5AeLzxwD<a;2HnZ<2#wpv^hQy8iX^
zTWWJTJ4=Wf#y7=V4|}>de75AaioKlI(NZ;Wu$Qx#g#?JP0<y9v`2RX=t^qsSEwmaM
zw%E=OlX^e%m-Bxfeh}RBk!F;d_*Opsp?CIl?Xr0ISvG5CcGi|Ux6*b0%0q_7z~Og7
zte^tc!tz_&k<q8P`d?e`rwN-D!I#0*7k}rMps$N4<c7SFk>F-Tw-jIW2yQ@s+4i{b
zDM2mO^@YIL{<GiYz1SmrH1MfquN%K`dAjHHDEm=z<+01Eq}ag-?GP;uA4g8Z?|Vf&
z=f%q{Eu~GPV)*8O%d=XWX>eHJbIFOE<3K9b!sBLRNyqXm<F?1GjmHO*Ya`}}$&0}r
zS>3qzZ71!=sc;vwwxjhP*>%PNRUE<Yu(Q_Uw`DO2gx{h}`Dy^uLcfqGd7t;*)wJqU
zo9|ZnTjRiu`}ONZ>u0_m_jtMY76`PO8WYH4-Z5G&{+pz6g5~8$N7)2MRO?;@a1ueg
z{X`%diPqLDC|4CLuU%-wbeAA?YRgM1LnYhWn}D5#g*c_Wx$BD+uj_&9zqGB#^7^Zz
z*GFx@skhm+@~oHVrtCiLbqC<_`U8P0y{HeIn_c2ygJSQ~G#UN=cK}im2H^Spa8oc3
z$l@95+s&)Jp`it(x3g9CjpSbolZ3$*XMI_p=KTJ>;cu#Uo%s@_nVC5V)fsjAHTJs!
ze@bwqnXif~L}z~iL1kZWA0O@9UTTT62O1uid)E~&ykhpp-hq>`eW_54JsO&A=`6%n
z;ac1dQ{MU?Sn58O6o0ZA@$byhb>{W{eb$C?@+U2?x6W*|?y=n+PK)WXkojVrlYiTg
z(k)Pwd76KE!5jbDSy;E1-P+t^HT6mHJvnBRc%^Ai31)vzrnj^j+s(i+zU9WT%n9(1
zaE6uN0lji}SF+bmt^GL;Q7~Yv9jW7UGNCM&m|vS@{c+hYd%D*mb$u|I>~}pNdp3c;
zyfnH#7ywS+w>1Q8EHAQM-MCJsK9s#OyDp!*_&#w@+739Z=Qg{J$zi>FP5_z~^Q4*f
z=bOwEGo2{36FMXaE(_7|BNLubFo3ji{_4@Vd$v(epejj4J;}!?7DAq)aLg|VTMoiw
zJ<&4R^xsQvt7`DwXumw)KXsGNhKz}KuU^cb%beT`J`+5_1_hwov%23bOloZ3V{iEO
zcicmI5}Kkb@g)N^gKrId>64q%`1<rLj%6ZLb2$61|JK9Sy5|B{<F<j;fUT`q<2<AL
z0Oga}X0OZT!wo?c(|Nn6bJJ%-=B_wh&yQt`$U7N2Zv8j)(Ato(5p(bHF#WkNV}eUe
zZ6`p$zSP{+{Pt|ef_8c7x75g=>5ZS0jqPkp36VdL6ruj-B7%zjACd`T%)YA+6<Llu
z_|uWU2N}e!RQA!|X^*!L0Qr-_hg<jV-Fz$4Sv7xId=8xN_%C<Ax9zf#mS}-o5Y(s3
zan$m_e?#z^)|B4U9bbY(_X%Ks>grO$X69YzTAgDFjnz4C7rCEEWp1u8<<6iAUiLl0
zq~h%~c(A@bW>oJ@q%(c?yEwn1XEX|RG;}DxIg=1lyqdp5rUk4@EkY$_j9&YAa9E*2
z*R0(9k>e$hQ~Zs$M>%(Ix%~Dc2sR?TH$VI|boi^94A&pqd?^b$!o%`^p8Z_?uc$Vy
zqxV%4S|s#H;T3@z=iLmvWPBhS5V+^G<KgQ1su9X~S)bZ+WIBzxgX{jj<nSmFqxzAs
zSyzGMgr}g|Kq{_Q0rgVm-moNFz=XGn7y*3ri7iJ)CkegOGuj0ZhE)G7f%v}2cKwSn
z^`q32Ep8(t0;iephl`8)T}DXi2p9Ku;}uR|l8i9rnXxvr%I9pjAYX#OZ-&XpF#R?<
z3?tv)i^Y_jZkq)R-;Boy=-j<8!BnFuAYB<FO`!m2WZ#`@IMxr#3f#;ZzM3KoiR<~*
z%j3B;q{}lg9*noiv}gHF1e92qrikSXPxcUnhSPIWz49IO`I7rvRR!?{eB*u!tvtPv
z^7>0z&H7D;czM80dV)M92_mnX8UO}ya*}ukKKHnEk}w-wnQd>M6zW_8GX+@u@_ekW
z5)AN#>Xr!T+P`c3iD9b4=n+MKOC8Luok@53<yjufYAh}xX-Z@6?&wIdAMrM@%I>Xg
z^_zuDrG>u8zNV8!+4E0<JKeSV*9+IHK)go*Z}`t=@3l^MH2^EWjoPe%uDvLzv2d=S
zb8NP>QH70-G2spT^7Y#GLiV-w1@%HZeio=R>xO1TVfdnn>2Mf07up`02`<@aTu5dR
z^N9zsgUPf=6C<Umh;j>m3<`7oVbvY8TG=CjT1*7fig|cYPJ|xY%YNJ<B9vlKEsB>L
z-8mS=P8;0kt{_2&BW#32gt$1VC@HaWaG|b9;z$l+)jW!HP!u>4=Ivpjm`4QEfWx&2
zy<>PNF$8lTZ3BWo5~9Wq=9V*-{-k{n%8O<}uAXNTIcZ6<F~)?kq>=_gmtE|%X^_fr
zJ6oQBD+0R{Cwq!-Z>{Sps=Da&K8VwI2{=|dS8uqi$LHB9y%J@^2dd9p=C0oGJRj4q
zFLu2m^y2@!3lDFZ#^<$pef@9TP+zFVq-opu{`Zy<=BPAArtqu(t|1E?dS$-)kZ^(V
g{0i&;=gVDy#8KL?rs<M=2}Jm)C_F<|AYX+154o70wEzGB

diff --git a/docs/dev/articles/transfer_learning/unnamed-chunk-17-1.png b/docs/dev/articles/transfer_learning/unnamed-chunk-17-1.png
index cf2089adf418409e903c35790222d35b7a90f29b..c007a8aaa66cea2fccb915a84f878392bd4d7ddd 100644
GIT binary patch
delta 322071
zcmXtecQ~8h|9*_tED5SnJ5*^=u{Vv_RM8kIHB0SPwLNG`j2h)_t5qu+6tQCOwnkO$
zUBsqp6~(Wg?{)poU(a=}^Urh6eeTzN->;LF$1;IuNkjkv4bM$c59FKV*3d3{A&-1F
z?(-(9lIPP{V%x8!Xv`BIU}>LWO2TsI2z}=)yG7c{7s2;737Uotf;?Kc0uJ%i<m9|D
zs_R6or1m|O@6gmB+Ghyfdv2G5d!l#E=$7|V@4<Piysz_~)Y@&k3b{UxWtm=!H@gf9
zlN}7Ns$K&#qU%pZ+W+^-xxF~v6Q95EfAcf4ujnuSZ+5n~BhB*FR6RP<!*t!3JGke-
z>d$mSXMmsLFPXNBHIF&PgHXTe5V}{rt8JkxxyoLgyK(`n@#h!6mf9`{*IF+Zoff<<
z=W{|{QPra+I}Ctde7RP>B|A?0ThD?9fd!=#KdYB_W>4oG4cr4ig>1MDpI)Q|n~658
zna@q^I8*gS?JYV~?Z@=(2+W7fFJJymwmPfP<d<LVn^&BD_iB|p*DD+obh)XyHE7|v
z{w7VbVJqs{D8T$?u7ja<Pz=~Ok&+FVEqDJc`GC9Wa2Iycu$`Mgbog?f8~7d*3xqBw
z%hVUV`sMp+d4!w$2$!a^HTA%M&2h&)A@FY@SKvOUpH#`hatCAMl~ZoWzK7D@&Vj_x
z#=`k@f|Ot5g}kWpNxW(6;cuCilVz%(mp%qrk6Sixwq0hMTJAs4%?t4hXn*7QA*+c8
zD_AnW)4>?f{y8_eWAATIkm=TCM;g$!&6&G-^4=#EIQ(xI*w0d_{<Iap_C5s+ay?-7
z2${0F8q8WaNP2KywwB!}V0SBaxcR8^+3&jHudSyuSAT$J;{3*ZeXdQ*aqhE|-)U{T
zbGf@GIYGzo8<x-hDyU!nKDWB~ZrZ%dX&UO_6L`F1;7om(oThRvmj*Zu_xJ%yZ+5H}
zervW}eJ}qxC#qPZ-a>j^891`aoxSIzd|IFqFf|q=+Im280nQVwe0`@cR%I^7Tq^%M
z&jP!<Tl4#ASIelq#>^r4s{u9&(4U_*drf~=)7p->E43Zp`%GiVw&^_nw5a$`mCVeH
zQd#}(oB)1J?=r0AG@lNwUG4%_ts8b!^u(FD^PdVky;-&7AhXaV%S!+K`&XB`$%M-0
zvp0LaR2SU_iYqD08jZEDhSV8DzrE>t+FIc+gW?7)3S`b7v9y-K>YoRf`(=6<)9+n%
zaW^jr#f=jG_Lck{>r3!IyhwY0V3~A0dw%KXbow#(f~W`#11(iWwI{$UKUrTy?HV!k
z%I>-U6?gE0(ksYL%Y<F%h1<l;{+aU6kq630+vnF@$H$wBX|T0h5kU)HnVDKxBG;*-
zY5B}j{nhL#Wckd=aoa8NSn+pbF3NA=?|I3B;|=%+Nx7|UWAoj=R+WNXsNvHDN6gFb
zjcd8^{Nsq+&@Mt>%MrlfzFeAnnXI|ZyLr5$;;S$A{)BFC^5ve@1%^#};ozsf-|*h)
zgEi0QD-S?pYwhYc{G7r_$7x|_DAVvj>B0QA)p_@vR?X4T4}JuFJ#j+mk$Yx0*1ONj
zf8zUGMbPMjS-ajEiQ%@c_w9F8eln`aH>|Z?Y+Q7g<ofl^xiJGY@VRAk)8Mn;B_V4&
z(A8tci>bC|97pAU4^;lJGfwOFmuc6G8d#N7?u|TXeed6gT1Y6_@VeP}!!C+nS^1~B
z%F&jJ|J~viMqtxJ`Ldba?fBIns}hwz)hErDX?r`St=o6Fh1RaNT`JB7+JYAiLjSbx
zCv%@wTO8)*dVOXCx*bFIp0^!-Q>t&r7t`eitmdoj1)a^E{wg{A1O!<h*bRrOPWlfk
zR17>;*$*Gy4rn)icGURbV$1RU)%xzYg_fljpQ;C~f2tQQG=nF(LtU4f+CqM|c3lTm
zot{|*k+Mh${zEDuzn63UcMNhaqgBo;hflRyd^H@X7KR#*fiH~s-M{Gzz*Sn?pB_KO
zBhlk2nU+&#841Qnzl$M>D^hv=)kb~r^MlbgzqY2@!iB~6ea(M#4<59xpTF9;)#ksu
z7yN1~b3*Zy{;HEWe-tp5pmMSS1dp*91dQHQzAOX&UhJV;cfMO)Trnh2c{OKW{b1xm
z|NR=<26&uK0$KC_h^9ATK|!@?!Qf(?G54fwxwhKH>eW<b3-K!6iuikRe9h|D<BXa8
z)n=O3NdH$BNB>S>;_*sdG^ZNS*>0Kex6!I;CDd<D<v*XjkR{Rc$vwsOS_Q>*HtyC@
z#;Y$yDxTxf41zhKBNRhU-!Er!bZSE@uRIf+CV^u%fc%c*YtOc;-)qC6Ki}71&K7d{
zj|?}zTB^W3Rz74jt?0bqKFJ-@t~pmc-9MoL<>dxm{l_k971C9tvTtW_`n8;8^KRvz
zlEInZ&2t_5Uek<!$9?{-d2rFO%pJO*dwJf6E1!?M_))I%E4;GZ{Az+x`Gue20o>qH
zCBBR<5uoGpzfx-6`Y7R-Kf7^xqJk=R%xOtzrk|>>_}h-Uc@Xvf<Ur*(x$P`<i`@0^
z^355A=-Ji$1X+cVR^r@+-f>T`*TA-dd-+(@d{7U};xFdzd`t2u*U^gZBVy70H5aUA
zpYc^gX6Oxd_pC$@0_f_TY{s1!7xw6CDedAQ2e>5Nc@VO?S{Ztft9&sw9CTW;;MY^+
zr_`~V82{%2DH9Je**jZ}Qru1;-BQ}`yBY<EYuiF?#|$7)C4sz}n!JIM4D>6z#;)Gu
z>hgjJIab-oU?kP-RUf>#;{3U}cF}$T{GKbZ{Jm^&Hq0n@p78#6#AR`s>(fP}`+3xS
z0LYrUZ-^D4C=N-AC$v5Denq`ivb)fhYjbs(+x9ES@kH*;A{|K6@xeTQj8f?R0XYSZ
zjKA>t-ygmY#!&qa3xyklLf$hyZszq**+X}>a!)^41$Xpz<z9RN{>*A#p1-!eJnU0W
zu5bIQcB!G{eDU-Aa*3|)3f^|ymv*rZDE*r5S<iYyoZoVXUv-sVl45;*$8?{)QYQl`
zpX}bMeSCZwl^ZZ#H2in}ylAUpuqLFh%y_25jgvdz<XpP5N9F2n+r^cl^Kh?}d??m!
z!scQY#cg%^n|bf7-IN~mXFKRw!-Vox@7K!oEh&%S{j7?{I|DSkty^<LKW`6R0pTT<
zfg2A(hRwgaRC*j|9}=q5Q+?z@%SnHUb8;&uEi0AEb=4Fez{u96dYjkI*Zm(V?~CtN
z^k3~x!Vj+w+FRDLy?18bpBOVLo;TAC5!cK1s%I|RnYq4`=hR*umgfE)<vz<WwK%vj
z6*v!!s5^F!4L7cAOL$zBwlxUk0hWG)zF*dqPuUFQ9WST3TShBFZe8s<vQvT-PvHi@
zX3y#KtCMSMR-RD@4+6$rS_7kqxdBh#*C{lx%^v+cANoB5EV(=GwEb#pS!kE3Y;?G`
z^lagx%6OZvZNEKv+~9O~4UppjwXZ<h{?t-cS{^-Rlz)D4Y!GM+WN%h21FE~{HEq8e
zBvgW@D29kz0rgk$+!v#adZG;nD-IQJOLDx#k32%BPyUir<_=`>>kJ83Wx9{h7Z(q#
z{J)=>Cxje1oal?GFEwm(8;mj=u1JUH{JA8B5k)VP9G^9L<<7tt59d`52Xwi;5`3Q5
ztwj$vAGWu+tK2qJzH*%Z1B{%vT>1i6(OWO;u67xN4!ee})WR+d0x#LW-=7EmY%3Tf
z5Cf-!6ffD*lolfne$Kb39Dmd`y|q`eqoBCGyCx2rU4<L8ZM*w5yI^*C+b)LuW=~Mu
z7TNplnwHJ}gLLx&i$4bM{CGK1BMQt>ES!#X1&KBdr{x|mmT)qV0mUZ)y)##*`IRBd
z=dD%KKJ}W4XG?39d#+1YTwccptv{6<n-4{-UhE$C>pnVPQz8Ccerw_3e|(~{*WwZM
zr@XSkJ+M+!Wpe({1b6x@!1yG1_Q1-ISa0P;$UV;qdH{6WSuv4+4oC={e*TYlI$6Vp
z16PKvn*4SYUh#hW3@~1Pc2xSx?SGFf=d|Hg=&xM^|G|aOLsi*xvNQ|two&G=xy!4+
zJ-P;SK_l|yi=Y2*y)L)s?KgD5D&axfXqNKC>>j<`L92}0wGySJBJRdT5fA@0R91<~
z_R_iXuP=VfuOBIex~;M?o=-XIk5ZiUCb!l3zuesaabi3SG?W$hp;n~n4|D2G=h0>%
zmr5Szzrq1#rURjq;O-vBo~zSm7wzS_!IgcjDO|t#5{TzN_n!Uzo{$r=p)h|G7qn1U
zvcb^0QObS!mP+{-Tdv|7T;=y%kc!u-RWspgJ}r1rJwfj1>F1ME->bi=z#jx)_ZpMz
z!}aGIasJ&IK*;K<&o92(`ee86A@!%1>e-(tkH4L5`CO~Bno6rQC)=(0?~fLvtPb8y
zSa=;N<^*0+w4HonZ0T~7U!U#3zZn|?PO8(|mX{|s&ZbU+SM1u>p4|<JT@JdK-3s07
zF&*-7O?+|Ur?}zB<wcTV%?l+uowac(M21rR&w;oD%d+7)gLC!L47@NH6NDEwl7zTR
zN$E)ywjZPlQf;td-`{6%*l+y_qF(nLK+WXv@^C5D2R#Yf>BQ|bb9h96;ynlYunluv
ze?H$p@DvagW<9$;<+@F|C;AYV{HG@SJ{i3=<=pWj?o57}>(Croe37OJv@M0skaM^E
zDSZOSNP1I2Q5xOs(c28N@w@OZ94$ig-~O(LeH#hAn0y8T%?>~kEMD1{E=@lA6&X{<
z{=QRxan~P7`)enOBNxswcKV`?0Ti#rkmpQVB=x;=cPMrU6+U^Rxzf$<1gff&Cd!qy
zLK@C(mHkc%*W&tyX&^LxQ!0<@yoNYOb|Zo4t=0?TT$TJU4LeFK^>W~C(>RZW$r}K5
zQg2yQ-%ug@c*Mf5_V^gj+6%~j9sANqZq)^}t_K8CnND8#Hk#xpU#EEQ3~sO=DTSHQ
z-R>>_3;k~?-O1P9eU9f(QqXKac(w|9wS#+8Xxi~<lRIe3PE^GtjW5CJyo>!hX&tC)
zIB(cYV8h!zXdG!tP?S^*Y_k&?o*YOJUIXl$-6el({%GI%>^mUKog=I1@I-H?Es5l^
zNU3;sI^OH|9{4gnZNKpTDEWA29ylwzo7-H!m6=FRMOM(dNA^NIYiGKp#7%*JH_8H+
zc`jy8V<VpN0wtpcvNakrk}9q<*#XX7HCdVI42nnHzh`WOZ!x&Ky4G3}ErfrINxHoT
zRg6jYsk-qco3!80zt%D49UfNWj?dD|ET(~fG%4p@v(QnoXUKP&Ud3762pM9F)|es3
zB!B*}B@goSfw^YYrHIq34Y7fq_!7`GKBJPB6RSo8Is6QbrEIi)#Y<2Q8vuGX<~{Ym
zR6hnz$!du5mGYZ?_8E$>hp>`$s<&+(eDlpIOU+;5Wwz#_X%8TLDSys+HdULW_YPSW
zMr)0Ijg#XsTB!71SmRAIqAAl(0vBk}7NB*woCldYr_U+~M|1s#oaLDrh2iC?tyCwO
zQCpS7N0<_qn&d}v+GgGr0DIQc!dcw2CvyESL&!^=i&A-uy7fR{#N(QGbi6ocp*z|v
zw%$<IuV7o}+vPd~9Q;x@#In>xaE-$kxV`hw6~;lj+^5~QDwV`kf?vojSV;0-V+EZJ
ztW~t0#af+xk(#(2#s{{E$#=_in%G6>iu>T3Hc>Z|Rzmhdwp^uKfHxASsdkNKikF`j
z+8P^u*T3~1QOS^E)$vl4$^i?z$tM-LwJzh;eo2m&5;>3TVzfRKF@&^P2+O2(HCI0E
zVRRR%T$q&i_ctzol<9DCc4q&b3{f^VU>jF6<$9IjI56(OAM9_(mnD?s!$V^6pZ=yz
z{P~&KslCN3aLZ6t7kG$a1L@SYKKS}vWd)5hrJ{N~ij%s|(E0F#3sIDy?+oUNoctj0
z0c;Bj!{Y~ePe=lAd4Hh*7iB93wX&O_+S{<mbR)eaH6G^zq=Q5Hu#Cxttzbau7>6N3
zAw%E6WzN2-u~BNJV<3HU@S7VxOY$K(Zxaz`vJ+L!;cU1JKn=TTaQDTv6B7}rYf*ws
zX4K#G55!WkUGN~#?sBT87AzZ~%8#EuX~~2b#YBhexonuO3n2AY9I%tpc&x1IP-BX1
zF-R5r@PiwgR{%z7A)0n@8tmdi2eD18aLEwEet_E1GbFnBViJw8gTbxn;{ZW{{Bqhx
z*}RXD)quqfwcKdg?vzXi7Z;cQF4yS_yswa<P)N-8B3UPPfiEGz`eOdOnk=TJ`4-|#
zh5E14@!+wOBG!#eF{cLoE-_)!K~M1Idf!DGZ(Y5oKyH|(ftF)-%heC#NAf!<AzP!b
zPX3{gpsk?&Nt;(4!WQXDf_=)89LUz)i9!wC1fq5E&J3!OyQvP+XZ_tpCVsT=XbgH2
zw<jZWt6aGR|0a{6Yq4Lg0@aZ;2<afBdCZ}D)0@!V{m_UB$@D>6K^=+rJ?BYsqAx%V
zzBI!}hI-aOJO4Cu<~yMw8dN9>Pf#^lALX)Rszl!KbkIEWI%lm3uXGen+P%fS0(AqJ
zZMT$q19psKSgtO^kw_p0YX=6gr_A2>?@)vjnGiy3phUPN<i>3C19MBcUcyvuGzU4H
zkK$e)tm1b;<{QE82&O8u{nRns=J4>w97Pfai;Tm>qkRRXOrO&{#Ar1qf|=AtrG$!(
zjtECbmS0f|YdyiA>}iBMyC3NxKtQQ8%0=jAbaw_gnl0Kk28$%Xe%g!s5^TnT7GoHs
zh__!;*esm%+k(^<G?(hKJYSf;epc`0Z|Bmu87n90CZUZl_i%L;N&9L8)p&3>@|*s2
zoj$cPeR<EUS{#NfJX1i*&F{Yd0SFWp)*1g~uus#;-9eQSu9b{n($hAY0gCF}GLgtG
zu_q`V1j3&GnxBl6vD6sco)@XggMh^t8qp6(1_lZv2-EA~8Rclvs?D##H2R$k76V}2
zF&$9x``8<eAJ2?%Wv?l`sQG#2X$r++u~<)QTZAT+AbKmWJl_=m^k7i=#92C!D}E6(
zv=T`V%?j%-R4^lCk{1Ac`JRa>u-jrAVo#3Xy|+j#wi9)jbRQcS&)qOEGMy%=J)Am&
zPdie1&8Q<i8$)>w9Ns=KyYqs@Te6=A=D<~GO$ikdm!)syVPewO`Dyp<Er}rvmPCW0
zo<)_TERV}a=%i%?Qo+4Y%4m*uYlXP$NSgc&G$l_beABST0*GnkEubLJ&&s+76Eg(U
zck*gEYj}wKW4JN0fg2pyNqPp4As-<(c0@ar$w;idogOK-_D-S!?xSLFYfHjyx_9?e
zXC6GhPN5nlc-Q2c7cBDq#<tI}^75gvfy&W`3W$xC&_}$1ak6a;%*bWT*#T^9q;|h?
z98C~k6r*1R@^FUakjC@G8b0w@vdlsS-(<GfRp$`k#U~4RQg{^j4yDgs8ZCM)=gsSS
z^w`L}$N@@KhHP7QGh4j?Xf>;FBMG4wUgY_?hy{O#5y@dZD!Btq^G$w)I!LsM6B@X0
zBH`nM9~+z<9BiL`;-m862aoM8)dETW+Y`cy^2tMhzb7%X_(!I01p&tf0_A6vZAzQ|
zbaz-B0u6mYll7f$w7=NxDKCs-aUf1y%Ng<d0Uk-DHYs-A6>CzZ$w~n?lgRW2@h0jL
z8uqOGtoEKC2^A(GxjxPcs{AlMkO1kMUQ>mWK(doFu?M&3^K3PL=GZEe_m;$c^)Y8r
zsWm`RTv1rh0m07;hA_i@WH!8q%JZYzNsnhi^zHq4Rz6Rvfjd+ij|<8y+u&EfrLx$w
z{vnDoyVi}?i(uyVuUZeTT_hPh!rv$nCo-z6L~fN|d+6r+&d7SD!foOee(rYjAobmk
z#cD#0tWp%7NNlJ6yJDIo!w4<CGN-IM8vy3>!6GagW1jP5PM@#|)sq8}OYo6_#f&Ei
z$xgS_R4Lgr?(>UL`T9wOMYLL&vl!6Oi{Z%W^}-g>A^pCBnX$L?_;|_Sd0=L^JTHWq
z2ExoVj+48^t+BjHAPx37sL+q4s6RGCVD-fjJTMSVc%n`jsyzM&gqKQNi^C4EUF3;O
zDJOp?M(1T6%@KDKiU3pccxs6I_}K}lO2Tvy4%8-d;q+LJa4=}r!X~SOoyz*}%m6(^
z^ZPq+1{=siB#m5)w-?F>DSS<`MUjQ6fnmI!GE}Bx{|Ety#r~5juujGK9#H$BhSVF|
zEZHx|3~n(XYm$TWJz>~SfWd^|PbMQkscHj)<$*XjwFjekC!9eb>O#}n^ar|SY7EJu
zQlDewpWDgn+<eQgW?5mvz-ru2cw!8*FDUR`H|5}CmJ-VOmaRgjmZ0VM-nZAaB{|;i
z*7@qJ&eY6qYIoQ(Nj_dG1O*d01>})_3OIe)mx>|9$PbUFN(Cc;xRJGkRp&-DUZ7@b
zJXp#~c*gzvhdQ?=maJ9pBo`W?w=P<?RU<}dOINp4Bci~&D$3Nb>%RYOiP2leK|h_e
zrWT-4Qr+#6trlaDFfa`^8us;rTeSJ01)iZXl?jO(LzfdJeCT<5)6eGzUQ-%B6*1kF
zt`0dEd|RA@DYgL|e7A0WrI@gi$<ecxzJa7v)xtu}%rFt9Wv}@RsUxrh{g}-m+%;#Z
z!A4=za;PzN608vX&fb*uA(xW?1r`4zUewX7BNh0k!@J|{spaA*MotTnx=6(^6W%ys
zJzgz-7SmCqs8F+yChBhe6m-n6Fu3Yy9WSNC)1g~x3q&A*npbMj)y>Uq;(b>XqlY`k
z{PZ+`gvEh(Bkem|dxSTYkFl`^eJ8wM%fJw<rV2uE-NN(1Aas08ffQj`2dkzPGiTok
z2NYE2vsw2kNlr62%0pdzw!STO#jfH~XhGO-9~y!t&N{X#vpn2{(_&fpcqM2Gx}li@
zbKcdzF~A~2$WUx3!AwNcBc8^1rH;mu*^a)Vs1zloYRk~6mrX{4RZE6duUi-6<OSF@
zhD;)~5~;cD%(4c@Cnjd<3{SFs@Xolxui^AzFE-VV)%WxbR3AITEsOgHT;oIVjV5aR
z4z?N&)}ZPgKe{}H^w&aW<JD5R4@j{VUtUV<0{XDuw&;$xAZjEOQ*9!}J~v|&o5k4J
zKmjj--ptFqfu?zk(9u^DE_xziO+&v(NNt@kqS&w9!B&lbuS|Uo9gr+q$Hohmh4;se
z)XF19#8-IVKwpD+_#ouzI3aTIzS!ZNTMqLK7Le!{E9HaqY-CC48Bns1<6iP$W*#UY
zEAgr3p2t%%Rf-9(18SRr`m(IkV^$_39n88ceb={>uDqE`Wsp2i686a$I)-7YPlWYz
zZHiGjypL6|)^e0lZQ+(55(bHfe@OZ95{gp*d*R2CLwCf`GCAYfK3r$5R>$Q9cm8|q
z?Ii<oaFDqh){cbT4IDqh&?|l1%k%|CQ>H=WUYSNQUf(~YQ0LK8aD5gfHQyWa5fbhW
zr<N-QV@i@4Xa=2yzDXuQqu$L3z;%-|v!l?GjCay*8)s!wYt6ir21!UT3-$eAg3zs@
zK_TMid{62>Hdc}v$CIxAK<g9!Rj+$$WsT`<Om8L1QsoaRK&2*0s$tsXbj$#POi=EA
zl0st8Of4@)Knj<6P_=2gtAR|e^Ah7Uky`ug$~r!QEG0ek*+?r-H+uKa8KC%oLJ{;A
zV*l>LeALE=t!Wb%p`r#C!vi#dMdZ1sjB(PpI!oqcIoFS=k6l<s&>e=hG4Spb&?jzj
zu%!NIaf*+Ms2i1^r#-KYJ)rG&mjg97@a9gBnTw?`2}jsMgZNHNjEIgNGC9)vu}I$H
zimZ29aSzkUnfcJ7Ls(MMYxZ>VYcw22%vmXu`77B``LOT|v2r0uY*j|jKaK0Ifi0FU
zJyM%8OQDEL%Q~24#L-fwAO-Z%uI_Gb?ws<?a`XX*o@&+t!i|*xq141U_Fd*=HxIU(
z9(2@@(cnQ2mKCFe^D>78u%NdxkDO$(Ug_%)z<fVn*hh2ZnTVji{+Z1H=kJ{-DQoci
z+(&ahj1V)H*(i3YaTn)ZGIBO7U%bvoRKs8x+@Im_fitt>#e$qvPU&GZ8KpS+2)QaS
zII}$C9uGPEYdV1O6C>dA)CQ*r1t(~WB+z~TYbawQ9y<`kfn5CY{T_=s5ZGu`T7QOc
zgPa-yc})>c=f!Nq1E+AX;UBznq9*!Zx_9(IDf&!tMLazN&PL8Oh@ps<_)lF(&J>SE
zfpZeW)wRqp{hkUpgkbKW{4;b8CN9ry1$L)j4LoU@0@4*ZjYIrDK-)tU)>ItiGTvm^
z*wfi2LZ1vinY)n~M*XIo?TKa>4n2swEk+At(xw;Q-o;7xzCbEo=j^E=r-&eq7hL2y
z>tMo=dc2X>Ee5@Z&;*mGcyUnXbr6<*x0_c3LGiepZ|va!sg&U&(fq}CeWNUneTM}T
zKM7f0KrO*7%ZQfWtOX+=S)=`e-G^I_c65_O{URRa+aEIX$j%kJ@*yN>&UISfXUIWg
zedmLV@KJy@gZ}8toCmBSY^-wS8NDoa&N^kdWornXL=E=CkGKlK0(jJqm&NAg68D*!
zg34kRk&^OhyTyd2cp)V)%(t>=Kt^Ve15`W;IK_k`IXJK^d8BPn7=}Tpr(#E74vDWY
zvC#Qvuf(_hS^EbBkh5)!zO(J5?90*=a_78W-eRzc+Hj}P9b2wQ3m0{iK4eh!#^6Zy
zk846+g4u(5E;mU(y6fDeCI<0uXDge2tc0qD^}5PG)h)BJ6*HBYDmDxUVb~C${9=IZ
zbs5;W{JLD4#Z8D^oHTtg`nQ-^rcjeHLetU4);Y_B|HBV?5K~fCT>e#h*>x7ya&#n<
zR(f_0yEeIM&yT>58kC*c60!|^P!R1Kfe5+Mn<8VuBNJUuyYlfiBQ9mu*Tcoc;174t
z@u$JcRw+2U{D*Px3dH!r$Yhx{X$%2`y#|6(WL0)=_pBI1W}8x>?=s5jEc#kApP-G-
zeO2@MgVSxgaIog)J1mDP=MN3WbnWaH_S0i?WF51{r4IoRhzF#NeE4LHKAK-A?$b_F
zvD+PWDLknspIjay&lD$pd-mbY$d6BBrwKN0@0#i!_G2UUQsbawaS^hA8W@1V*|2Od
zHJPR5Lps^pP|b&6P2pndMM(jVK|>e(0Eaqe5*leVt4X}oRIxLyJtJKqEehifvGBc)
z<cPlIU6apHl!CsQ6R8MR8x#Lw@v%uEoDT+nCnlb8zaisZDXpz4i|Q+u_J3k@_t}^Z
zejp+m`gu-o5El<`0>f)ESAha^eI#abCz6~~QDB_C=6SbdE8XJ=Eo$MPj|XpXXi7Ky
z)5dTqL@f@r1=UGVco54oqV*ZJ2qkNHP0YaMQ{<a}HjiqZG?ONyT!e|U4%7Gu^$dWt
z${AntrhL&V6}lFmjdl5q<$;F<;fqqTb=`_G^qSHmnpTb6MK-V30b3Zov9spve##8l
zBA2$nonRlIX~_%krDDs8L5Hm{|Mkahj(hJdiLZJlcV~z)jB0J$eHal5DSc!8q7n;Q
znAqs}5~^)eR|-;wmZ`6c<Xuxt4;Yn#_ZDl9Zb)L*9YWe>39YTI%3A`1y-Tl){Ml`p
zDH@K;)hD`n^wtRgiwyhe?Uz)bF?r>K*Hnz;)-!d<#rZ66oS@ZDeeoC0G+HoTY+eS6
zfPN!{4SU*l+cs`%)QRo7s%xe|gb)6qRCL_z6Vwq}yW9CK{H1=-uW$LQ-ZnT`_2e|#
z-K__VVS~JR@%>ddhuN88=~JFZ1N!t|w)NN);WvsaXgfK9bm9%Q9n%Xjdb(Np)s^JQ
z95Q0@K1;OqcLH1hK~^$Woy&Of_#SVP=cuGx@~rPzC4DxQJnCZ;A)T=iOU>NFqdx=t
z?%YU=b)&X)ju|f?WHN?W=@M|wwu0kfABr<DEb>o0)ivAkMzvu$96MY+Nq}F%l!Ah-
zc8E%)xfCexD2J6L@id9t`Z{<T{K<@DVLi|$7xT@2Iw$x`YZp6JbgA1@4gM2VEO|vh
zV}Jd$7noi+XktR4yDKU6KtwEy;56-<{F%R~djP?bZl|ZU?a<9(>pIN<{sGrYuhnPn
zsvh2TFEn{*J6dbmB@|HwJ46N5kXv^zVp}a=ZUZ1`_jKB?D3P0qqc5dLsvqxgO8(n4
z+{Stt%Nw*_c?Rn(FCHopLfk$jjjvl5lS%nQ%ZEE!Ws>=7oZ=?B?QjkLSGLwGU?XP)
z7V%<?+=qv?m)3#s<72V!0F*aF-X0}Ifw+6ynP8;D`2#f)mS~@)=P1BFL9J@bO9RJ1
z0dhWbls6C6b$-Z6bt4+?6;tMxY2-bq{wYF%8B$Q5nthJ48Z_3i91Wk64a@6A4A>UC
zU*9d>9MJqY^{PlVt|%S}O4KzUSaXhHKq#gvGC^YFEJn;!&@MM=BG%(!1rE>Xa9OPQ
z$Zhnj-kil4XPmi4&mByioAU_nKX3=ol3HELHiF^2BMFy+XWWx020z(yZsdrhZD%qX
zeZ*gt$(xhDq1?f2iy5QGy%nm%HEo_7mCLhG&?eESIqTqY(iwFF_Qkj<y}20lzsEiV
zG=2+x;xs02C8}G-+YL3I5wm;J2cE&J|BDw(dvt`HuiyWOLiy5<4>Vid_zHY7kFk{5
zXXzJDny@9c^A-z;Ed~2@&A#q`DGnbqipfe4A_%Bc@nG<S`Vu=$CIr&EsC+*ZD@*4m
zD8glGdiKS1Z%2eDSMy4<g>c`zg>NH4ySw{mV`;{JU*Jr6S#<CIlV6xc3zMkDvh)!D
z*xiMUN=gkRQ{GtR&s;Dz4XmNhY+65U;L4mzlH+LxpLEa<d}H+BWOxUYIp{&S>`YTS
z=uH-_SYuT&`lC+>&lmV~$Nv>$bD`H+J<v>+K5QQ-8(iord@C!q5&Sx{O7HHd<rdLc
z4^5j6vDFdshw2=6+zR+08icr4{!xu!xus7ZHiYQx?zYqm5h<nxcE95PC9(!nK};ZC
z+Bc#umh_2qsm{d=4usS=5Vq^FGXct8s_{Ye<@bO2pC6vjXo0aSF)I-Y7em&4EMh62
z5!laSM6PEFd|0up8UJO=mpOiPbfiS|+d(hP>C1yO{a$i!=|kW}y<~TaEp((D5#)Cz
zH8eFKi0h876D7d;qxm&eT%GW9=>6JRw~txmwL3Zoe5j-N-CB#WRh&HX!@%RDsuzZS
zs&^xtO}qsPirwtu^qNCbyIY*uV!gvNRDWqr^J5~(rs}&0{9YmAS+6*Ghw4dVEZ7KY
zc~z7cQj_QLHFjeflyBCds^5Kw*SG>ciE9sSuIGi#SkMASIzn33#hC}P(Tb)5^uli`
z$#36xp8dFDq2m>X+qC<x8V3o-Q2q!{)abJG8A12+Mk86TKlZ72#;HjsFsbziJVqH`
z^AV9!0Fgt~X^Q1dG~yV8|H4TcWAw8$+6ZcR{!<|}!<(cLXNo+NYhmy&*d#+6K@*p8
z+KOFLu{*GcqPs^!&P$P;mBHJPm5J*GBdl3fV;Jn>FvxBWNTXA<Tro?Nc9zlI@UgdB
z_>GYSM`f&>77`})fRN~kH6-Il+Q5=p_vV_NvW&iZOBqNpsR}(`zg524JWl%zRErZ$
z`)Z*>Pctnlrp?YFB!$mlP%2`v*(e`->{3Q>2F7^u^#{$`Yb=DL6Eg(dSZIqJ{7N^+
zKH@T?1Us3m2P(k#3o`B?QFwc;oUHO`??y^bgz9gj$ZRUlTLpr!&P9`0XFQfC+>ZKn
zRoR#kH4Pg`kuf9DHtre91vN}#h_U=HNe^#@Z@x{X%@o!@07_Dy)i3`CVZDDDJ!>J5
z@gMrD*r8TuesmHft^7H*BTmq>Y>XDBuH@Y$u*UHQ_USg*h8hu8{A?^8j5R~pmK9%o
zD@Rc>8urfJ9{G=FE&q_B)Tu0Dmhm;wpeb4bjY7o0<)C6F#<xpTjBxI0skp*`i4RaA
zDR^-^#t=D(D--*#Ak}B^8c#}$T0U_4rH)nV{@brmeO*IIzY^STKkQ*kJj0LHmZ0L~
z&m_{g_mm4tn+3|JW^a67y&Sio!=eO|{*+M+?r?}sbrZuUPfb@G9c{dn&v`UASc?`Z
zR8sFQcH-~-(1boB;K0`+_<t0&!=>Dur&+RO=cHzWHb0LAFaILhj;*2p<<E5nD{eK<
zW@6pXF%bE0xXOcoXUAWb<9rA2n3gky4dG01z1_S;?&7ca!I>GUJKbbF#v;_RsUWx^
z^E=t<h%C({ZJDW-I+yRh)bX1kTsWa8zA`+lkWyTp11WFJdCazW+1(!7^RbkTmzkL?
zk6D8E^~{ToH|da(yzzpf2msAP%Y*=H+}53&b&GTS@p88PbyY&PDSm}o%PuLQECXeF
zC#s*xn1u$DXY%lVdQ+K@psI-Pj6E}vSO(+GBnF9$Y2KxU(7Y+t7UIjeM>Y*TjEir3
zt&q+Z$L~$Otm0DENZ&|Gzm0$%+_T;#xzBK6orI<cI@BF=gD651@K)$1ouw;LgFJ28
z#cEn0GgCo~9m?ZLt<9gl<o(~%7m1LK?)gVSwvqEcY+Z@w-mkdJ8`FvIFpjfTTF@Y`
zsim+f<)q}!__zS?Kyh-NlM83+*TFg(Lcw@50X5K^+O%oP7WDPGcn+4*o@Jb1?mUxh
zq|iv3-l}XoK11aKx)o@bWTTXJ%tJ)mrVrO|v5&(7hNUadXONVG=8PV$<7oerz)iHw
zfl<A2C`f;_!NZsqP1uWSbmlr!eyrDlnsfd5?h%~lyBK{D3-U*15<)M(#-zL#ij)Tl
zs$G8t5|i*eNLuvqsiGM*%Z%8y93@A5elBOlih@iL{hR?@Y?IjyC1wKF&AE7vzo?k|
zI*Kz{zMCaI>jws`xpzw&Oc7pNQ%15t)svp;C$jKY2C~|AE3!~H8|%W8C{rh2&h?Ud
z$?8{wzkU>mjbF#?iv5Vw%kHe&6snTrn3vyP3>kt;BH}1{uTjZFJl2)j;Qy7lC6?l<
z%pJMZ?hEKA*3oQ@x{Euh(dOL-BN49TOOzbWBNo;!*Qlhw5U6fZ$(&o{-wwOyJh-Ug
z5E;h7&jV{{WGCC6BJ_PzzIp?g3#Uf>=$^dpPF@VUUV43%lG*q!<<ex`j^uCtpWifO
z+-5eWUB|mw_(2U2l&2~CNhXUNM9zAY9v8U&tpgB<BveH6Vv2P~H|#?~LWZZOcaM%*
zh#a1RtBW0Kq&iFJ^<Ay_b-T|4(2sw#vVMzA_L=~5;e*;Pu%b;{!TVIn(ECQHucQci
zmPCp?I60sF`dG?;WLml+YJ=(>ijmOb8XiFiwUcz#4%#Yra6G*<^}FO-TxxbyDEM9C
zG*C<<x0at`L`!Q@+DIC~m1Q*D&!d1<v4=IuX|OQCeabgUFF+p(CAiFBbbK+?pcR|`
z#?pHd*I>2+onl<PUS=YuCcb7j!u#IogQTADMG4vn7da7G$#jQ8xZ0nj^Q(g-jWzgL
zVYI?UCO)z7NH92$<hm^I&vfv1SGe^7{+zSAqIT1Ix@?SMqtX-VTGXUnW%>)875#HD
zPbzVB4S|=QzCJJkME&$y$AFsF06l}8qv<t2kdA&feh@)#Y86toGS=Ec^x4fQWi8sr
zHlWxL_&XELeha(B`UCz5*Hs#5pikRn`@?ripKgFEDo}$vzo3UbP<<CEsZju2AV%gN
z*D<Fng!}g!UEhX_UDR?JIpqbJ5Lh?8PK(%aBU08=;qR~Jgi5k42G?RyqH6S2L8kRl
z<EG_`AH9b{@Ha}J{9>P|W%QTh)ynR;=!AuhH6P#UZ5L-c+BH%VC`bAJ9gcd4v5RYb
z0(<2%7U3S3XbTqbp?^pnP7SzxR#U-1QDZfx<qljCwQhDK^nX}c4Ui{jr-rN>8a*`@
z`Dr{N!_5107%0xSzi3=%m50+7_G^EeQLcU5R+$y5*Yrrg+<BLZN3!067Q|$X^%CqE
z023li9@l>ny&0!}Bb!5yGc2MT%v_xpUTBU-Nbbm!mlJ*%-RTzK2N-Pgic&0et;ETw
zy<!YApUhSeN#)k5wiZwS>8r$c&9mgk1<+(RlfLH*9(ZB=i&k{S(T?jhypS(_!#fE=
z9X@E^8bOhD^mXDSO6J@snBUDpijgAS*R8&_^*qysaWxxNPOUFcW4R?Eq0~%hY?L#N
z66pCM*{^C3=kc-!lE5Cvc=`p|)QfLOC-S~mox<*{2=Ybr;163C9&bE~v%{P`Dol%H
zJ(dZbARezM)ku5$#?P^2rP?yt;t8qW^#t@MKGc1}ZPJp~7p-SBxavkfRfT9#4_f!X
zdSUnckx&K^rD;zVEtvHAHJ>gb^;2he0*3)o)h3=d!|pSHs$ueCw|xiBig(<C_l_Ko
z;X-x>KW@lNah`=Xnjxd<B{B14(O5OX$FYI0_yqWu_QCyvztceko$fduWY`nVnDL}f
z*E?AFAdTi+k*#}f#CLO?HXqw0e_7~?W+c9&d})mo^ZB;3vS;?kD5hK-7Co5=A#;B2
z$8XE!V)p}>Gsc!VAq$*$4ibIHPBLrCVT=CPG2jl4_QX<s{dak2OMLvBIRe&`g|#@z
zY>jV^&W4TRF(1-b@pgn;Tc`c`5#6%6By~6_J8}9I>7Vmy<(?gLE`p91k6Mt!$l@GB
zqNML2#k+cJ2^Jz}vodbWyP#KyfHLNQiMOX66hKs_blqqqA1mHj`eU6r9mHgGKikWh
ziFf5)`R#_ReCNT7$Idpz;PhxrgMOJCk!9Jo6)l=3m6i3>2tlnE?gfp6T8E7H{t!*;
zvM?zRRh1mwWN@3%Z+a!4l{&j-bFH5wIKW$so(1Wm{2W2oQ+ardbU$}3R$QsL?R~e`
z1m4+iY>W=f?Y(U;tgdbfVr5yP7wXEmx>6CW$xpmLBw0kP7H3vP&IZj5>Tz47Q#q%;
zQEyry52H>b!&pg4jrp{`)-pc5JCNqanKK*M^r_03EN?*0k50x%=F$7(y7af~1y~Mz
zfIJ-y9W3C(i{TCR2>$Lk>Jz7lW1BFVZ~$7v8Z>j#n4NL`31S;h24WgRC%uRZCC+ut
zb@oAV@|LoA83`&V$$g1!oZ_2!XYm|mFON|%r9pRgxrobKkm%MI`tu9rzy|{;mW=t`
zzAEKjs5D9BzgT$$_r;Od0el=?M0@Lz)&XQgwPgP;#jLj~CSi99p8%cfYZ9abC{$%M
zt@SwVf<t24LfG3vmE0L6zQ%<$!oEIX9YYN4x=Vu*sNBl5w!eQzN2cp57vc*c@7A4I
z1Yo>O0=6utuk8;LGeRbpUwo;$sqw_#GlfKkH~3sB2G{oF$t!)tXwC3i?|JL9x?E)-
zu<2QN`Ny<yZ=}4nM+mU}^_5T&AUyN^SZQEMQNyRgZlUEe;4FuXuj(Y2C}ns}#};8?
zLNd~(b+`v+GC1PR$(>HG4i+#$A<EumclWzk-YWZA{R2JjCUG)9<JERDkKGvUOBm?X
zORS;QtbZu1s@3hYu!`*$#Gxhm?rsV4%|;O=^1(C`$X-gj=j37F$}JrLE8XkG;lP?9
z9D$*x#2U;0K-2l~7RakVmkn7THRo_Bd^>X6fr%LwNC`ty=8z^ftf-MjNw9mc$i>}<
z%C6}_;ApHTgdY(z&?v@di_mL7nd}^(*0cRNmECAgm3_fh^P+4tn-yMk5inG9dADjv
za*Ed2A?DY?fs{m$iw__KKkJAil!`G$NKolzO==S}M`a}T+<2^uuf9Fo?8|dlXl=H5
zKmPR~FyL9>UdT(;d+NMCj$b?la9@qmZaa8*RGt@-eQ&Dp>3k2;RG?lzJ=4>g*a-}2
zI(FZv+9R?E72*0P7TT(`K&pX|?oI+N#wL}KyRNTrFF|wfavxZZQn@%BHBh-aUENBn
zpYWijq3%9h>RN+yQD6QJd`*1^t!thJLuzrb*~JZn4cJHrIx;=)s0!tHjxVBRI(BHu
zdfQ0$e&!VzA{5Bcm;INXeJMXS6&*dP^nk`Lu7BH(rw~_-(`;aeARW+8Sa}(=Y#|UB
zmwUJ5nCa3ZB!H~76ot5Wv?RpSV9az!j6m-oLM_BbM<nGRJ2Q43(3^~;%qb%tMLf}&
z22}fNMEhUso=&%utTyGIuv^^31FP_oB84y-aQNxbrI^n>a_SDxIJ>W7qsPnPm@uF5
zFc28p<V>$v-*P_UCj6fZ%J7pahAG~oa{1)e`Ocs_aI(#(#MB9Y&7*yE(3hZV{_NQ^
zv;qi`j{EZb>HFOGm*;JlTlF3bJOwPMBqWcaZGMKV&fUE&#z694$FdqKsW9VsQ%X#;
zU%d@=6`oFlk5$!vg!0g)&ns@8%yUjOSAtyi%kApftRA}sH;$m7kPu?1ze918iLK7S
zKzrDMty|Y#3r{oiju?mQ%E<PM5C5^E@ThybhO9Y5GoA{cxd(i#*pj&slr)6#BH@YE
zh-jVW6!a5y)RU9()|S@L^J9@N7rZ(h#sK-BmJKROiyHX|!{x9cn1VFg933@o7=DmE
zE|vi-6pSCk8Aw8~yPz8iJBJ5K_W#|iGt%*|EeyF^eDQjiad7?M@)fadHokt%S0p0N
zh{>j!EWPxtyB)aG=<CY)TN_9;($_rI%?>76Z%1DPlb4DAhm+;eSLSne^aonVIP8Fh
zOHL?er~2j+0jKVe&M4+{aC|hg*A{Y$`qyTu5~YiXeT>h9JI^Ttmw&D<POkR7fu~R2
z=ZYFkxZ$7Ah*H_8Foe<j$XL}$MT5grifCzqWW~n`uBI&uyIlMGqjyzne=S|dyl<@U
zk}xyZVcU&qwiGo!P|z5r@7H##aheiv7+~hT<-V~C!3h3WZ}iU&D4vD+k+ri>umEFI
znKxOn1D=MEf^aMZbu;C6?U%LY9luy8g*InFzyo<_Jj(vCE1tBY)gKXj@n@qXH?+F{
z-|c#E<l;vcMtA}lGb)}APY2HaEO`+hHkc@#?e3!a$;5?w)1NW<S=$=Xuh{KTr}V4N
zM$VnsK~q;Lye^I>t+*~)HDx@hfM4x;2o@34N?-{2aGFP3t(|5+8VL<j2)&qjP@gst
zT6@T|0;?_+AIgj(Y@#I_{(2R%8>EZ=vC)uB*D|BsFOVN!G)l66%0k0zZMGexGVN7+
z9h;F-EoyolVILB=4nL!&n@qSo4)lpNdO2ev8bQ@-u5&r$p$EbsWpxJt$uGxTeM?1%
zvJ}-s@a{nl+YYn^f+&^GeIZ0{Yj40khiNJh8(O&8H=_7mg88$Ad2EV!y!2T^OREqw
zvSRDHUVp>@R<;Gg-`%0Dy7w`H+1!DPlrGe*Bdi7e#&zDxEghF$nfBG_7Q5Xu1=4;y
zrTTwhw`V#MY>!fyn0x$zw7*fB*#h^l*<#fL1Lm>OAZ7@i7eYgpiOEak8d>0;jk0-S
zoS@#4aqZ|^EJRyO@nb?AcZ*+`aM9w%Q?=|^Q$H}7|Jtn)r_S5EU3UEbss%h0k*_sZ
z$_jy8&7~?;WZ31=*4IPE%u~wzM<xQiJRs(3+Mn9T>PWARPU-7_L@r{OJK!zwu@IQ$
z>RX25?eeH0`I^W;4ZWVTl;A2cvL|?uN66u1c<epX@pobkJiJ9cRb(P*z9YCIR-c8F
z!lR_bLV~iZd76+fO5?7i4#tAT?xyWy^^?oVf6wv06%1(A+(6i>GciKnvs%+66vC6K
zB_K6Ch7ET=U`v5}A3(g<Mbc0c)hfGTWU7QGj50oka6iVq(Ut6sASPyT4vul<CL|iv
zVfQ6D04)b1%YaKT@xV&n=lKcU8TJ7^PRYQ?wwKs`Y)iiUclhIpVle85&W5baUNyKk
z>SRv$xOr+$-pj$hi-WbXFbPHQ-1!T)lep)WslJWRDgc9ttrgE%mC#V)Li{|rm!-`3
z_ee4xiHV~4)&zHlxnQQZ6yVW0yt)ASy+W*NfnAdqL`e6JKs`q)X(j4ut%t|Xan(F;
z+`OAQ9OZrwfqf5g$MEQV#!@fUEpR8Cw{YJOzx%>Q=$@n@+&sOOy7aPN;mU59=FJh!
zrz|BNI>5`gCT^;R6~+HDp)jF-=HhZPVLmB9Ivt^<wGp7wJ3qj|^i6{!$U~E;XqEdu
zcN0Cc?8)KN9h-3%^sD*ruz@d+r$Ag%i{{>oA=~J6Yxs;Gy&bh)SXIqOGFloiIS<Bn
zoBrcRFqrFGpJ&ZWDm}_DnhriJ)Xw_R!$MgHYXH{$L4w)P{9Z|L)YkF5??01~dn*|S
zN(p0wM;^^b(Zr%^;L{_p;M!C!<cIDEXvR9*?)t;`>3POOpgrLQY;x%9wUx(MJ_}^+
zx}v9eG6n*Fs7OHu=IvaAdG4-Zs4>CTNX%O?F9#jh5Toke!~54p4{7L_niIT8w^Wfp
zQeq;BL5t+XKt?Ut&*G3&!bu*MRDA2?Z>L0!pA0eN<f-;ehz6)zE+k<fTYlrO!c?Xt
zvv@`{4RFYH&c_+j89K8AU6k8Hr#d)bD3cOq@Z;X*bq3rj<LK{v(x(5NU&`+K*~jQv
z1&jDv)<yjNeV(Swz@BrHnLLaM#HJ3&b$BGq%xul9dZWhBX>R7neLMU9*ivyt%QVM#
zs$?Iwp6}RALKEgwVS3-sdESlp{wsTJLEl$5y6RH<X1>ns#3$+sfmM+$e*E*IvE$@4
z+UrVnYY7Iq%E6~c6Lt%H<kY(Ou#26Yu~}k34cyh<!BLbD=&+?`PPX2<*9Evq+ro^T
zg$ObI_WMWkUf%mmw&{dV_^1ALmoLualhU?)M}FEJDs$Bnv+j~ELV<vQ07Wj4XUj8;
zvf#7Va_XNDWhh;xz3oJrUpq%{n9L+^d`QuqYge2-?pvPZu0I#)%za<6)$||wsf@_#
z=_#A|)p5O5Tj<62kCtiro8N)fP-X{#?Hw`pmii00jv<U*#$>+=ej9!(=kk`eLD$FX
z|BTYmEmAxctDWZ0=6lmT9#~nSw^oK^%bSD#@Dyj)Oo#LhxIG>B-Do))DT#X&{d9{t
zrjPPVZ7B<x4O6l9&uf>jH4YaR{$d7B_jOgfpzCvP4Un2r(XR*${v-{kKi}H<hxtOk
zSS=7gxHsCpd8g3aj46>nFQ9d9cOqf<pFX}^^=MO`=UVp4{{CenHFGTFJD)+k{PV{S
z)BSMom)4SdO4ZSi4a@F*q~k#%xw}H@qM7mULGwWdQ3dfdOG_ll4-IT!38JPy)1yra
zM_YTdZfZY0ekRnNoB4q5T=7W#i9=U>9|Kx!w6_Y*lT4FLAt-Q~;OQ(X#22`Cz$zYS
zz&K`P);asMe!Posu$=;y81_csw?`oPix|#fji<#nXy4!Q-`(yhtHo9ixTWe}#AVGu
zcZV9PXlu}W(hiR9?khGY=PiuJV+H@3Vcpabf&<8smM?Qj(I_BkR7rzUKtP;H?EbsJ
zcPjI<5=_hikA4wY95mW#Ot$*qfnW6?Ezd-Ygb%j33mXlVs`>1{oeq3*MtqmtrZ!mU
zRk{4DrOEg$lkanr`&Koqkm|md%U~2)7i0)iXcd3nQa^u8S{jg?je5IO$HUd~c3cjA
zjqnU9NJBvmakv7q9E5*wj7F7|xOA^1E^mGP-)P<2YGUv)I{kBcmnR;1QtP#y5JJx!
ziWAyxw2%(FrX}F}pV(HlP~nUH^h|RNTUMLrbjBkDS03v37olQ81?u!jLv_ElIrlCl
zh>M7@;+tPw?n+|OtT_j4s!zs*Zt*#2{CvJ)x7`OJu>l>JqVXb@1J%wJ?z!GuM;rgY
zZ7(ho-^Ztf?3wngd3e#>ul_`&5vd_q1HSk56Zg1g`VpvJP|A!FJUiF>0{@)y%(aK5
zq=XI;Gj7IYJ2zgk<|y<}wO1CUGzk+#fy;y8iLFtu?e_s~+zItTh-&Y+T_>d0y`q6e
zVC1?UO+Y|s>;I$a+W(pU-}kT>Ic!K%a+pKu?W}}^n6sRfL(b>(IcG0QOvx!>&N&|%
z<$Rii5R)WR&E|Y+G3BtFzx({~{R{Tk^Yy%+_kCa2bzk=!rMw;gbE?%a>(_6Ez8BD$
zy=N`Gq}YJD-rg<0FNvzlomMW+x`xjk!Rj@YF^9Xg)Z&cxBP@zD<A@5%D&+Ix;Y>J)
zuFY<0usQ3S9N&1-PbGytdWFh_J8kE0DZG(4y%pU}Nni4$h(5|8>j@f(7_d4D76bVy
zs>geqrw8lC0>zCJ#sPEo{EY!<S?t>kG_tvOs`hQn16%stt#Vj}j8HkpdG}@3RH&u;
z?}bcQ2CA?xlbadsgpW>ZD9WT0M92{3x|k53a~?j`%C^b~`dlH-ia<HCv&ji_$$!O(
zT~}|tb{+{jXEr3xwBvYn)l8qh3rI#81(+D6CafV@jA#Sv<KxVk2@t2mh3+E9XFp)u
zv{=fhw!q?P=a?<`#sxut0AGJ}x;el6&}bv|1<Z1-(3Sl;rx{wvg{6bpvG&zJ6{F8A
zC8b)Ko=r@Np4JyR6m8PW>+3)X9vs|X>-L->-^;?>cuXfXrtW8`>t1I8F67ZbK8d<l
zdA1Y>hh7<>@d8U_*!A-ZiG|hC$RX~_qL4*_jnuu;VNyW4wnQDb-lavCe<(1N^DZKO
zIBqt88ms162Omecd{;cYYhIB1R-*o!jpZAG%ujhdTN|j9Y%S#TRR0Au9hY#-Zzg&h
z_|uksP&V_PhoKnq)%H3N>)I6k_N}+2TcYmerq_akAO0xHrB*Y)zAs*RMUuA`!mCq}
zF1TEF{-xowk(PJ3YuDVXlsFP%=6aL_^P>wE!;$XAPr@$1EWy8Znh!Q~q0Y{Doci0;
zEmh_xs!7j$$x->GP}sYiOEBEByMeg=OHHL7+r$ZgPb2?EFa9zYAZ;){0#DFT`cn<A
zk+8Z?4~Fu{`waBP9Mzf82(rB>?8C;~&ERh0{t|rPLZ_{<;k7bc-@57r^SZEi5J8JU
zybd`w)!$zudt)cbV31Cti9SFlx?9axO(jcsTLM5K;qlD;d1>ot^_DkdD;u>hoVK)F
zS-Xk+JnU-_Q7E9EsK0;^)PLwhYzwnH?Fjq5tTVvNe<KqNVlIX!q`Ra^-j?kgcJcPO
zqw<${08Ua=+Uipdf54o9UX6TD31~msTbkP2@WR^L$DPqLd|0X|(*Zdn#QJ%~_j_N3
z88@@)XiMhPW3Q#iJa;T(>hMrReg&H`^`Ss&#;|k~Qve|suHENq$W)T;(^&CFzaJTr
zl;k+z=W3;`sIH;EYGK4R&;4W)e4Cyn^J?`GvVbw`ZAFC>&TpSv>=_TVD9KX3S+<`q
zq>1<xl#NDTWoExRTnx|qTp(*|G%q7B%TTrk=F?Qq8(`(g<NC&It;D<^@{L`*AE$V>
z9he+=0dUIkv(~pnhTh88i4l<(h9RyZ>Y-!XPW-Fl%IB)A%yj?D60~;R0ZAgtP&b?$
zn?2A0*Pm-It0)Y^nfZCmBeEB`J{sOg2_^Zp;8E{#I1@syQZ*a@a3G*2?|WuXH*@Ds
z+t2>KmqM^)gN9Gx5Xf7}u_-s(mRY%IA1YxJSSnF{IV!7+*q_{VH?Ur9*OJvkJ#Sm0
z%?Z2}h<4{R?Nd6KFq{DQxz*OOX1K7b2R1i402705*y9!HvlgkR=;z<s+uMg-9CD+o
zu@YLAtrpsW?knd8-ds8A?S4`AH|VItUcpo~s)Ya0X$TRb-qr9QOm#mx9W~-+{nmUL
zK)=vJv2SlIdAP^U3Sf&uClShCW#!=wtu@}_=pU{B-ey++h)OdicYSkX25aXz7lc_C
zmy}rUt!$19gEI5-YGB>zHub1oympD}TaY1`MPBE(JM|<eHqORdw9F*{(kG(ID8?nn
zZsK<L8A|#+(0OM+cjf_iXIqj4`H?t)*z|0Xqu;~ZQvy;Ve<;A@1gPCY^F2N3{!>O2
z_2Tb$sCZua@w^DzHa|O7k$1ak?}6yqt$`xa6DAOe;Hu85fecLKx1QCiXc&9c$%hf@
zbD4V|bE1rk7i$oLn4|Hn4f>-MVRrb<CO_c&IwLH?8T{U0<?|}PYy$3Z0Ei|C=yJ#N
zpwC|NFS;($(*+i*<%JNG(#Fg88t3eD#|}4)OB#$HSl6XD^SKOtyF{<vAN~&^`1$N+
zM0~pI8(Yh|YxFn>GQGUre_()>9>lQPCxYwqeb(WT$RwyJBJ}o4Q}%X?le=DjSwC0h
zIJp0d&{VQkFT;R01~l)20(j613YlDT16(X>)Z$E->@a`E&RR<1^`>uibs3S}SR#cQ
z2wo6>`o=694I20TzD^W-Wai<=m=|31<AG3_N7Z;AsD$NWxWO0*s+|)O%S_qqhZjYd
znSQ#`!-0a9+4$7b(1|cR=-+n~iu(5*BOyf7H}EcMI8J5XpIr_pkjx%?)LK#gVrhMt
z6xH7K#$2Bop+c*Jv;(l*cHWpudUbccVg4T;i}XT?>3DW@>b;6jx`EAv&HHsy23i`f
zeF6ou$MdPZ)-!c8<Vy>Qof3OZeu)P+X(jJo#O%|Ilzw0P7=G$)mXvMTB-^a*s2{kK
zm6iR!`%RK#Zg#+?B8$)T-wriYr_?JgITaNPi^P-;-GhyZl6Jo<u~y$JeEbNsSCIRH
z!}ZL&Wp$j&xIXZr>>HlxGB;vQotIzo!#1BJ=8*Qc?K;}Wrz_zh9t&l&I?`CW(wyu4
zbe(om<Tg^kt#p+%m+uLwy4gtop^?PU&WBOtfX2p*PYQtaFPX!L$jDCLT%0Ry%1gUt
zBRcM^@IQIXe_S2Bg8qc-as#^Cictbj^7dUy1StnWt+Fw2`Zr;s2u>#_XTA8cORGj=
zt@iEHI2*G2!1zEBCpRp9)`i4m6^eb>U_4<qIvE|{mkKKeZ^y*`Y-590jK;q2RGQ+3
zv9-N4%m6^T8fLBZuS7j0K5h02wB3UCb_fJePuk~5{M}dAqgzMiZMcMtQsg2usNe6%
z^wa4^V*T-KRv8~XjL!E>rjahozcoSd#qwd&odE7r5D2t5EB=-BR>AxI;-ufQ{|&Rg
z*D6lKAK<PGbjl4hXP()2$jT%927`8E14=k^+yJb7W4ppE-*JmBox0I4eD>6UtAqx9
z$8Zqq$i2K*Uwzvl+)=TedrlG`t3f<;mR<^7bd}j|cKWosuANSo*K742vc3lHqm-<m
z^vT3Yp^K1*y=D0MD7!A&q_V6|LKUat$oNann7Tf=SfVe7OJ$Wwwh%(*X=2K}D-FYz
zfWq~AMi%N_N4}rYrYZG2?w?r=YAS~wuF7jA>FhhL66XxcqvtmsBrAVXYsJYYJw^EP
z0EkrdSic4?Ln%raCqC_=6vTuC4}OB+R!P*QQ^hoYq_Tuds*{~<+d}$q5b)bY5$3YE
z{;1_q_Z)F{!0-DfR<=e<wCJ0sFdi!)*WjOSS(%UN(-D%6wx?4LX3y@qzwNn^TA9Pw
zYsv1rN--cYc|=6g-6k37$~R+Q(~6)(U)tOo_i9<0RwcJZ!b#Bgz!8P0fRmN|>nMBy
zH3@BHs^!6^oF?nSfeN(CMe`(OlWi<rS;^%Bn;mNWDlZr}q3$L>sM?T(FUi1_;SWQy
zMKgrM&>yJp33uYHy0s_+Dm2MMui93olz#3QPo~aoq$P1xvN^jy(S`L+T9W6$*Q-t0
z0gVBz`QGTCHIXmQ8KNZW{P8vgFg0cwx4*y5d}OIkt6N+FeaBd497G4{Ob-cv-dD@6
z%p0xrq{r0i+4{rD@#ivJ_wc~+`jZr-t3CJHEh#y!l+3?{cjz-Ax0~hugHm*MNW*a*
zO0F!=dG6^OP@VjZ2Kmd=6a0)tj9Vb-f|&@3)}$AFt6vJVwyHSOv9Txk0FKN1*sJ#<
zN*;@@hQb#4tZWhVa+79KE1K9h?@#~Q-n#5QMbXSN5GR{Gdsg>_^;ij5a{}gH&h1Ub
zA8ZvX%@hQ-)tJvSajUa7gFh$8Z9iQ2Kt1c7^5qnim53&%533Gct9)2evfp0NPRFhI
zG;;3TUIcyMHus7;a^Go-xexad6AicONVWwjf@B35B#=9nj9_t_9JvIZRfYGG$P6!(
zL|J9cTPXIZLT*6VuL7{8X$D{yYV%|=Gg6W2&;#l;eWAD@;>JA@+CW4fYbI|<@eRUj
z)^wcUU|Ck6G9>tV6Q^KsRFq`}iW`QMQ+rtr5r?Tb|Lu~TMF=D05b_J$-W-=fwGxIs
z61aXO{4Nfc=#-#Dui5FUp>8zHFHbKvevjV(6p<K7|1mHC7>KFU(Y@nnwJUpJSoPx}
zN*9Oz60Nph-p<2|D%B^~8JKE(zi;}QB!-t1AMd$a1-ZDA8cG(l(zUQ;Nf}*fLbEk*
z;m@uOaDyW9jz;(lH&lOScB^uh;-j$kbezml6pc4^<ffQ^yS4s4GGFZ+XqZ&*+!poZ
zv_owxw+jwTO}=}^`O;JJKYkqowfq)h)Xh{<?iI#eZ^S0T+ME0gg1Mq_xOA!z1)OeM
z=M8znJh#$gDfp*obQ+TtMPm?!KU$h3&Tnr1J6tL7NW+gGHf5WEU%_gq{0nZ@N!?MM
zz8BNu#h+X(%6zqug1WZ5IX2^_6n8r25EY<vdlrzi6#6i(K1~!jQcbxBGHH2>s_`X1
z;?tqTwtOBfW1Sj>$F(p(m$?&)+wT?E24yhIgXy)nLFkUXG&CZMf3U+d*6&V_u*U75
z67etJJ`+)ARgb@h9L&@u-a!n@P<@Zy%>~4@pm$><yA0*#CTR`Ws+g9Fybd+c_1y?w
z5&&6Q_*}4b;i3NGWn8_%{%`+lZ_JNz@oMMoV}riiFJPzk+w$WGA+z;rV>Lely=vUq
zUq8y``uG6dOQ<T)6L-4ZGHuoZMNG8*wy-&Mn&7*?EO8e&a}`y`Y0gXN!R2js@BH5V
z9m#cSzrNwSw$Kex=8C$UeiX-d&!S<%7;uO?VN3aI@onMLqdxN^MXkM^mOtPS%0RN4
zrM9z7yOY*=dMy!Q7E@Sk-AYq~JeLZjC<~s<ajB1fZt0MQRp?75QhpwH2D1%>8XC8U
zdgS;KhEmBpg6gQ>3S#A)oQNm9))tmdyQJGYh1tz7=s>!P1Kh$r?f9sjw+NPaAXABV
zFQoOxEqmPcV0$}Z1kUgK?5?15As5*YRQUQ@aY)xa6x7MxKZo}C$#F%4JWh!S3H)u1
z=djU_xW44!BX%RCWy(jcpKwgNf;JAr3^LF{i2NxX0Sl@-FLVWpxQl$%Q{yYwWQ5g>
zw&ie47#mE#?ZsF=?E=FD?*;(vkNB87B@Ai=5zcaV?Utjkr@OfmxWbI1ARzkk@MPTj
z%~R8;m22l_DnYJ3!%O}&i1cvR5K&Bn^<eL$#@gOkR<7?PuX$K|H^WwkPa}SF2x%cN
zXg?kqfmmLc?mON;R2YKhq)!+_n4uUp@V4RQD+pOVlOuTg5v>eyquKnS1H12V23kv#
zn^gEWPdiTjj!B;#Pn?<g+>NI(fSon`v@1k8QA~?vnmGG&c1k!q0Z!D~k9r%(Ycy~w
zHtzKJ;2_w}GPm|$s}sMTZXM^=iDH#gweD-StL)}Tl7G8(xr4n!G^H{Je~FG~GI9rw
z1&;Rp4lilBinjvzN~*s*ruFFbi?V#<%#vanDKFBR$NAA^^Vce{DiKz&|7RRwMeZ=~
zl2>M@ID_Q5JRBPW)tPg+zyHWchAsu&Dn%aBReH0@B-Er&YQEx^r?yY@`|c?_G>dN|
zNjU$5lRS1K1<GIetAaEt24nV;r3r0sZuY}o<45&|2Y_@&?9ry|R{Fw+x&MTV?P3e>
zi5KSwG3w?oIUa6EtHnmpsI<RF`#p9vs`dUrR65;);}fCwFifSl>4jPwd>FdBLm=wq
z*`Eje!OOc9D@V$Y;=WA$1-5kNBE{waMsqKQk%Y(BLf$HSOb3nr6hdfUK60U80&?A~
zbGsUrCV_sw-x97$6vpk748`{>>7I8;TSsOu<NI$fgGO6N>hyI-&5#%W_v391V_hcG
znZ33F<#$CmWn^Tq=`S6xqef?iyo=fy2CcVhW$IBmbrNYw@rm5-bPyOy1efUQpXudd
zIISdn5a{H-sIDJVawJ6G2VeRVk;m&$BF6j`$cLFuHu>esCM9!y78%bgc+0!Lu?FE*
zd4g``Mzcg#;U6{2hX0feapW@Ek{(-TmX$4+3s$euFq4y$D|1Ve-_1;6=)+xQbWuB;
z9NrrU!JL<6U<=2&z6FCNoBU^Z-~3ylHcqLjg)B?kni7#$yX2({d<|gsd0+f?ZwUAU
z{@C!HyM}qH1EC)Mo1#*L`*Jdo`DY2d)~7O&qM(|-VUC#+;PiMm)XVQZ0glyEvx37b
zudCx&yUmc5IW#s>-`X!#hQaMI_v)A%7eYqBeVnn>Dgx2eRMU}8Ivn$PL$0eFd^pNc
z4qq+K%=IHm+0yqCNpjelZ`B5i20$c=Co?33Fub0pWYbJ~e|$=i22R|5FU>f_9c|78
z=(OWOr5)vgKiUj@A$^eSKo-UxTh~<8C^Fcjm3xwMD?qHUZLmxuvH$DecwOg9?gNs7
zMCErQwd14ySJa<AeZan5*l0?ba&g<jh70n!he#KpafTx<Z5rw!|AFWC0oDIXet(-9
zkt+bwQ<g>y5j=x>t)fWBBpawBYxJ7G&Y-vtMFB1=cLH1UelC{4GdsX_N9ayoA?lfo
zYjDz@dnRfGg<ht^3~yr8_*eI_u^fpJSIh1>!5o)=4`nj*j+qs!sm31fKD4AzZ>y`?
zwE>3*#xr$Co0v^1X7oNV=0+$kR{Lh-{;PE~{f`Z>`w8A0pjM5HEVd45WQyF%4Qepr
zA|0L97C0(i^Z=_zJX(4=dvo&!od0B<N{F>1cfA?w+S*f1OXJO;NF;J4Z=Z_N86~W~
z{hl5-_ZsEOF7%+k)u_1SKE~2#I`VqmHBnJ9sIc;vXxmt$Z|;V`CT`wU8)k|IS#E#o
z9xjt>|AP|5DY-<>UJ?5kt~`d2;N+BnLt${w?=r@Yc^d!3JvXihHK)tAHoLEG;gi|S
z#Xl+e8n!q@q#J@A7a#FunCLU_#Dj;Nl28cNov)$J#=_2s%bYMv<nw3OIR9hjgdtwy
zrt*;)Lc)-g_)owDQJ+34zEdF9N*yHHbZ#Vd-3<&5N9HSi{T`*6kEW-CqhzO^p(^nR
z4b+BgowMn~XGz_+E^lt5xql~t!My&P+l9H;6H}cYp<Esxh;*or2s~Ja{n_w3I#`=H
z`zs9`{M$I#e9_fS?M;93*kpa<P$JPuGb8UE&!=sL-<v?359TQo9gDME`RFi*fZ*}w
z%HC;&h>3mPbDlw)Z&T?c8q0$)seIUg;v*)r1)^l`pB^kZ=!-}DNC!NurwKr+*EkYn
z`#D4Ge2}|%``$P^{R^w4pR`N7^8TdcY;%)v8JydMAv1|YcH6e!!t4cO-}n@Xyh`aw
zT{t?C1#(xtK4?9^-luoN8(Xq+;qSmsh-(X;H>i&D5T2RR4{ot<qoK)$>lgyOEXwYd
zWK-M9JGC@?8VMCCUf)d_R-F~gE<iIwxCUbQ?0kffWEcLz*h1@18ICR{LoF*EN9%s0
zKy*N5Hmm++Ji8-o(a&~#c2<wqIN-r111=e;yqaPC?;TZ^)9iUhAg8ip7S35yLQ4xd
zBoh)qZC6OuwWw#?VoMWtuqV8`Dsop$9Rx1=GfFXh4}PW&5&?s`?WG74XS+LHQb#AJ
zWa;;53dK^n#$!!D$84`;fOPN={2KWY_TkxXb&0p6{BzNPzeP=dlT~Dm_71JdbM?Tn
zMQ7|)Zs+2?yg<?-?CROB%Gl>iC3AkoS65}ZaR2)pOT%mA{T6>RCR^+_Qln5Ju~X>Z
z=elIB;H0e|hl=bf*P)8u5)}k9d>6N*57E2!x-vz0yvthgu?ymLq5KE)g<I|2Q<!aH
z!OKv@P;meCXN$WOWDVqlFH8T2a{#kk|11v23e8X?jV*kN7r81XBs3^HgxlnGp($tf
zC#R~l=<;Obx2VzT`H^Hl#v!@Pz1Ik>o3kFrZwitJv*!}KuV4j|f8;$a&a`NeGO)Wp
z{TONU{B7|URG>VUrQnTQVy>>?Nr}3K93$}J0WRw?o0<xpHdSNE!&BpnXkhguD7s)u
zn3~{-=(PBXsC$@Qjd>ISp>vbbNbG(=%4cH$TTUV*6wgCIH7x(hYx>+iU*#k)Fe@Ch
zT70%N!NDCRb~tf13>@xOa}a6r8Heu9yR!tB*KUlhZ+dX*y)G>|!&fL3ZnusO_Yd3V
zk_%*jh^V?cn3Q4r@KRhiP|)dlQ2Gp+Is(y|E4Y~q>R%vfYnF~IBLs@oi2eas0rfxv
z1^q|hbU@87x1I%YIGk==R@Pt65iTuI;v;E!p8oXj3~;v62%OMBgF_dT1o~roa(0;Y
zwE4a;$$xXo!vd|+AM=~)@*7-Q$vsUTK96O1iR0Ar!3;=3p8;>5rmOz!J#S#@mIE&R
zT-<3VEuaR_s@etWT-;@M&wr}EA{FRjK}c;~{J+CPW!$Kl7R^B`ihMu{lRh70HkU4b
zYZh<0T769(F|hb;*_CCdt${a4Wov;Emj_c&5eB8ZTy=J}ksjmIj^J}?aT*`6MrY2l
z$q~l~^uStKfQRxq_xa;5%+EOtn&BrMlz20JZHdj&rIAT_J`hw6L@$Ie)>8!GI#sX_
zc+)Mi_L@$2)J|45_cjA)dFnj{+9^q>Cg%tg^G}e+C_R#6CA3cwimiL*z3{BO9iH*e
zMfxONg;|MWMtS&Y=SY??ROo06DrC7#KN!Ol1Ohl+H5`Yisz5%K!AR#s1YLQsv<V}6
zi|3B*w@Dps!q}NTGB$opmUKHtINUH|yAVPD?*7x({<dG)3r0CPnns>o6Y#VHOU=*1
z-k;M}N3>kOQu~Fg{|i$#mhbPFNVvtY8mVcJ{~R1&m@m>-_z2o8qju0T<<9c~7stX3
zssd)Ot>V8Al-D$&*p`$YPnuPz_XqI}ug6>Exbd?NPhLxa9<9?k@hcSON&J5F9y9he
zeJL(lQ-dQUT!Y)m(mKbj`=LeaUL3!+i{WK6AErme$%9iXHmx#aC#UNet;Hrgs|rr0
zVEX?u-iN4Py)!p!OXiP9=K9?g83{}XM5Y3j!KPA{GhqqxZm6owhIWN04yw&4x2}Yb
zUHemr4y8HanpURlSL#V|-XLF~zs6#B^`by@T<I6CDd4p6?9bHNc@vwCr{@?S2#F>u
z{uDBNK6<>K9~u6{>5d#d#2TL0S=V4tq;I7yQel>ulJtUP+r<Dz?2?Az>U;r><Gz3$
zp4CI7y7YSg%~zE6iz7X%4k>*uEpdkDzjeJby!hIt?wRXy5YW@x8xtMbImZtlQ8~Ei
zA+0@#JtAPWWZuc$j=zV@E9j=k8`iB?802?qDKi=+tt`bc4Q}^s-13mRe#GvNC(g`*
z1nFqKv%K|pp*QI#-;dy{3TyHb2zg+?Is9^4wm!4!&_h{S<2n(UuOlt~3cj_L6Mn8;
z$!s2F=w+GiVw>!gM3cP4OgM~L`09k~g4<{|G5YblFHLyF(+p%{9wDfu+(S_e=A9TR
z0z!W%bqv+E(TA12)xGv6cWIvF3*-^R4bWF&HIQcNhLiIWyAk(IAc<pBE5OuRI_;{B
zy+1i>1dcGb7>w;{QmE@Qn~~G^-r*0A7lfM=&t*&4@V>jcyRY-~2?Ho*?L+T`UvGMM
zKSLD&3HpkKdO&yi7Oy(0D}xN>_kWKu$W91MDS?95b-J;=cU0}wRMcXR{z^`?J+gni
z+To#{X?SV$cn6k&ka-NT26*e8C2d*+-|6rxFp6}KW4Bn_j6%qMtS$&dzCIvo3cDrX
z1QV7li^hjs&7|Xc7xb%A5W&YgVKJ?{9SGl1M2J)kz6_LK#Xg0;yf!N&xYj+-)tVCW
zUhYgQ!307}WP%Bd<y^4oN~86ulT8ILD31}+FpsY8SxkSXJf9p;pJS*Hu4`iIB)S~Z
zS@cDx-AQ%mTVQvm7SdH)aJ#J`1Ee0$2h)8MW|{@dP>-KY9=EQwPL)mJ5SB}cY+!~e
zF=+3+E#-B*d)FX{$~-G854u*zZ<I)I(;U%bHov23lc?l$e?QN=+V{n(G-pxD1HeSb
z7<u*wICFr$5Y+=Z+OgL4wijRm_@jtE1|O`%`M^OX=TT(c*g35`Y);SKWiZnP+2vG|
z&1+v-MyuKP#*(pg%PUQLvT6AEe{p;(@6I-+uDZc;+Lw$ZTE3}!`l}8FYCc~7aR+~M
z_D+mUB}&!qKH?`OeK_v+zD|H2>qRq!nUDsj#%u1x$)gkCbe%s^F6D8hMFn^rvw5_z
ze>S!6elKfTx2@P8gu@YCK2_FL*GF&7{-<$zHmC90L{~Ma0k>;mDfO<ifEQ%<5A^(!
zs*zcR;i8a(+~B|!_G=as2Wdq)TF91L96K^MryG9l?;qxeh>0+^H)uoLnn=+;7%GAA
z3WdYr*q;OZi@H@<VuL36cFQ^S8`96yo_>j1P`T70A%Pf3zOZVYwQo;|vwfgek{kKC
zQn0kX?S6=U7FTs5Nblc`37A6OhF+d1+vV2`9ZbqyR7l7zv1M@~9t5u{x@q;Y-g*Jk
zV_~JJkn>mj*%?<!dolS2pXa3>(zAT2ZEx6Y!w~%dq>oZO+sd;iM^GmE#XdiT(p_j!
zW_pPljbazVITSlM6q9g6XPP4DwMg--b4n|PV(p&MTsfVFN*5jw+to}Rd3DHMHBYp|
zj{_h41d;d~=$%Cb1XJOz+XbutuO%od`jZYniDtd++Yo=34x&q8xfJeqc)Z&qu>#}*
zrv#M=-9+3Lbm{9(@4NTp$z<&$$A~Yfed1ik{aT4ue5J{^Hfydh89YGh2J5e?<~fV5
zS%|4SyJ|a^zFxk|6fSS!^DM$w-kU*z5v<#Xl1=PF>(UWh!Y}A0r^+D^p%I7kBSc}u
zHZ@%w1bfO0)#;)Ja_N)`)z)RC0y_zL=t!j!Xsv%itu+M<f@Hx{IKk%yzE%X%=LIWI
z?BqpsD#KWUz@7r(7aF56N2~5tJp73*{?V8_e1<``hyRqhJm{CQL(r9m(dHo_*PH8h
z!$Ci6D0K_`3ki4OxVw;%Wn%YzHnwogL*%zukzczcsA(RkkZVnaMUD^nQ05-n<E43j
z8OGxXvb<?AaOWy-g!pl0(P?tjZlLm7HjCLe5{*zhIU<j(y^o2axnfGJB?epPAUJ>C
zy16_eemjNHqP4Oj%WYKLkV(*1mw$_&UqG!kzs`@)MQW&09)J2&eD-qP^&lBVcV753
zJ3E$6-P!(g_&f=BM>!r~{7pCMeR49^xHcu~!EAC1nd&vlkxpIP;f48ZiWWehMrur~
zPu>ctD>yqOH%h3*9`4GNgiL-Vv~BuGO~oT^J6fxLCT}TLF4BFvm(|f(pYKinS+HLP
ziU<$a>3<I#G;NXnY$tNh(|c0NqickzptQkd&==Cqjqyk&E*AjV7(4eUWa43#NP11{
zeGn|DeK0sb7uT<=F-z<x;><pauc*9=wH0zMe&!Vpx~X*uVY(`UZ{5dxhiyVKb<#u=
zqu)9j)h=gcNlYDYoI1_zhc&j_yrypZgFk#Xbcik*Rf9Pa9_ck}{e0vQfW7}5Zj%Xt
zB)-1Spcci>Gz#!*KN5uI2A>@aNCQU%BJqTl3Dg<_>Te@hpj+c{Cx4%+0v(24ooC-A
zx#3UtPh>vZNnJ%Kl}xC$te>5!lwVX6dnY@)$G;^TxPa8Tnta0XgDt6?3vt(E{`=1$
zMc+A}VAXs9)d(FLi?)d^G8#)x7qIWWbH)4~DB_~!5Fj8u^Z`nzd!<{uUW_^>mphe3
z!u!axEo#0|P14`0xTNFr(T<G^(hTbMCfL5!0V_gEt&#He#Qte+`hgF)JsFLrK>ZP2
zl|y}h*|!6l=Upkua~bf|V5N5zg)X!CqW8QWvC8|kM)3_q%GbO6h6!qhofOO6KJ|B}
z_YliB0ALFcN)<#lrUFm!Wb5);yrIXg8nuoM192M<%oDtm1HmG$$PY5+g!Z<E2_?+h
znD|w3b964VNrp(J9@ct_)LPFE)E=N3ZB}_@KrY2*7y==RtEw%)Z_dt)@(dzgncWN#
zrjv!yjp=@7UEE%G0?kJ^V(V=Lxi}NPEC8FgFh};(I_&U6Go!*&9PYc^m4ORvYDLC(
zxap&Oz=dYY4xho>d+lqn1QkJ#vY?`y0z0)x->N6r%E{`L(EO{pCgvIu9o_OXDl2mL
zd@#|NT33B~4O_w_16_6AoBD2#oS<}kL+(x%XT6P2LEWmvPFBbQB#gsZv1HA-1W-HO
zq`X%FfD<Zk(z<!Hw`Z*=%+;AX=4H1>ro10-E;MqJy6DR2oO8pO7xdxYK#`SA{c9=S
zwG|N@s5ij&vbE=y;ltoG(@Y`3$8}prIR;|R9BLW~!XlNLq4&U0WOmlF1;o@x!0RO~
zq(I-k%|-`oNs0B#{as*sRXd{$EdF|~dnv^e=C6?P04hA^_vO6E3O#DeA6u|*q7`py
z%T^SJjDVXJPBcGM<n<@Vb(8G<AEd_BD~$elJtttRXg(VG?waMFwqE~e>W8^YC(jj4
zB!dh)=!xlZ@-RrQ+u^^15cFv8%p7nOh23<ImoeOnBM@?{b8YoBN(T&qq{IcSiUQ$3
zDk~xYqUxXeB7fL!cZm>RzjI%bqi^4mgfQqTDx9c2FR;na_wVGcmn_l|gh(04cK!LQ
z*YBw7Rh9rhrBOOSU!P*w`eB040*V45SpN!g$T^<Z)*nQpth1o!rf&ZUnpOKoUrKm;
zAP2RIxTVd%`5eKkL)ZoKDxk`#8#PeVQSEO!l28i&gK%>{7iy5l0t#-_H(CEmq<kDS
zKo$DU=n_;_h+>AdkR96u=I7w+TXr~**Rtmpp~}t{!w{dU!|IcGJCDuH(=ksx@hWdW
zHlet}srGU)38v~_M=7(ScMS2B#M=Uq)=V|=ylQa~|Bk1sfs)gWtK3@tBN%*}|54X;
z1I_IO9~~iMs*f`q%xKO-B<<ZN9>Pt(Y&Z2d?B<y=#-P)*62NgSEqxp;!hxKwavC}&
z?G3S)5dqXGIf1}Efyg)u9W~FN@rE?v<U^@yf**!JsxMwug5ow0Ha||MoFjf8r@O-{
z(W1jPkO=_IlvZ2ty{QzmIp^R;^+bsTE942p-r-wlmguWgLvmZ!r7M#x<KIZ~k26z@
ztGy(eY{J!JB4VzI!;&I7IE2L4Hp<o74#Rz-Vrf9}31NqNvarSftkgg~QIjJ!^6^+?
zT*qyqBv&buv(!`{nL#~0ULEn(uO1`1v$Nt~tnb{|osz^vzwfV<JGSBsy%!eHn*I2v
zP!IKUu$3<=BQryk^vdzcqcH?>{nt@frw%82bt9^K^GMZOG@QvYo-SpOPOB#R9GLS{
zC-i)jr{DY|R43C!0TLOE9=$P*?W!Xuy!`Ea3GlbFtXF_$We)|KiB#W6xTxOk=T}eE
zB*ngEzy;R~oqa!cx>57AvSKCpaLggjta*(hJ@@{-zpeR`OhuUguhY$yP3A&qW@1L(
zHLeL!A8-nI5nUx<Go-ekntA<rO)7l>%1B|bR_OyT4KqW8*54{znHu}^j^hTe1{CQG
zXS=RT^{lh(rjqu%Nd7OgT+Fz!ak~w_dT0DN!kv;{-wvbFamCHN{O5uIDCA$>$LtBL
zmp8mG`P2DA{cA9=Qof{t$q^pLh}?kuNCAY=C0!H8aBpHj`{~~%T42S-MMg$P6Y+$(
z(|u#RJH|V+Z)(GzG|+I#sO}#}c2YQWXP^ccrk}#A2@hHHt2Y#R>6q$lU)5>#9WKeb
zcugIYcA4NLP7}r<M*s**|6MqQ*wWS&K+E6{CEuZR^{tVoZ}?*q|CfSMBYIVCIfbiU
z-mMe@W~-e{Frc(9BN4gX&ARt@$X#KUAXRCuB!@ZDo9V92qu%E1hdkyX9Mp!N$Ow_a
zjFF`Q7s0Uhg%YW$H>>!gwubUiP;2JKXYoCfkEnW7HxWR7J=j?pYkdEda6;5V@_l)n
z<94{a5!BsKqB>{iVDB&OW96|cV&W67o`_^W@{2y*-wm`jT8q3c53BJaGd<%}^!CG$
zNjBz^jkG`D*G^dJDKx%_glpKTL*0`FKTdhA4&9p`cz6X=RhT8BG3!J0*{m}cw}xlV
z#?}{X0T}-aP>~f^MpTwOvqy<xYwKZD?D?I-;D8sIIN!lf?wO$97z!4~;227#bEHF=
zjut91SK4-4Z#vrRid>{Q<<dMz4dwW3b-zWuz?)|XR^h^9vJ%^4JN;<>FYex3(5gF8
z>Tj`_b_d}7R<s%qMc_~Pfbg?~yTZk&p*z|DJmmV?)KYai4Lm(NIXhH4qnsTyo;|`g
zxW|uxZu}&a-yeb~2meN8Twf#|k-A?Dgu6PODNWQ2_R;q<9nOpszl{sA+f2Z%5ve?}
z$H1Ykrb`Gako`b(^5OEr*5ANzp`6&pfcCvGOZj?on|*+|n>Ou$#7AtaCq=)I2CluD
zgcg2DXNOuPMuqAyX@3$AS#_RQ{yzV+)vM;k$tCuW6Ch0%AzQVV-R_rx?>^HFSp5C}
z=s^{PE>5oVS3ixYd_yNFuCkRAW8f40)L7$8@jsqPXmp+2I~k$H*UnEN=JJgM{C-i&
z=dpeie3Itttfk1M!5K&ozPxddF51G$g$D)(g#=*P^n&WcBI?jDnQrY7TGj5*w7A6L
z;&vN`*M_`v_IQBSJOQ*Of`$gs5~&(mXH;}8MyQwMPwv``4r{K>&T=dK<KKm`4d}L>
z=vS)DlAm&12^=kJAbuHO<z_{F>7SjQ4bdO*g=T@z<qtQE_v<W%%&4YPg-y?c(hk0q
zRCJ}7n$+(Xd?fpzSyEc62)+JfMi`9+bu!1ImTODTg|`Avh99rXmN9z!*lorDQd7>W
z?YS45wZY=)sm;Ra;j@Z*iq8wHKItl3K3J1*v6I?Sxo0lbnVp(gXM`B^<rfj>lXMgp
zt>k96yDd*57M-3_-{Wmhl1~`9TP=I~qX-p-J4#P9iTi5-b7vp^{rhto42XyDvY*2o
zS!AIiB8mYPneONZQw^<c2@;A$M&jW+>)z`ysJQ57J^cnO$LOybe@@e4F>Ypr?~4#I
znBa{&timoT&l8EP5ay?e{j*Q3MAm$X2<JqDv`%Tk)JxudHBn@qRgg&P$e^dUUn3qv
zQXt;$2w5FO=z-M7yDq`5+zkS;f7dDH&P<KZ%mQdSpzP<qn5|a-);et@e0MGPaD{9@
zUR95C6+^II+q>qFG~~6*1$mOmIR`|<#2mlH9F6Ue$z!cI&v`t~Br5z`fA>$@L4|bU
z{DrN1k7~QAuo?&SKS5hz>kA6fqP|Hu&(m#VqiMa0+Uhub*Z%W++scff)H~>qtf_}N
zfW$fBo|JHAt*I)k*O^%`v5vNg>QWtRW*)-_Q%<J`WOt%rx*#`Ieb4jPc$rTq+lTO9
zj~OCMpMW_zIbd~6NN5)HxyVgTCRC;{2js?T4@5Qa^}IV?FVjRDts-K-R0N6NAwJC*
zQV&Z^N^?w7SJ!a7^o+7C!(k-X!tCE5P#0su4ncn?OcWJ$!jrn|I-@cQx(RQ6L>F$n
z1;iR(QoMM9W(IY7$BM+yt0#Lla}8m{C_noGuUmn%3YEOs)q753)Jv&Eo6p9FXrZ%l
zmsLCn{k5-q-!CA{d=p0VQx~lkkr+&5rAq?S*D$tnY+SAVk+D;M30il1&6O_NPeE=8
zg~L!+{}d01e)G0FAZGt{h>LTDpSRNY@@{XnpRomsgV9&r{jp(W`1}HoeRmy`+}dER
z0?KjlF8W+NNMupv)5%hDq>>k9?tL5^Wz5UD7bOGUZ)k13e;MkYV~*a+F?Ve&q%rS-
z$~uDsCnpENITElhQ2|}m4^cI3#cMw-3Y4k~^%Yw8+Uk6y4jSJdPwj=pM#qw)X^&j&
z956$O>nh(!oS$qv^u^Mm6H^j(_U~W`HE0_Uo%c%o<w%djIrZr;15r#BKa~wicp!pI
zWN&pNjVyUNQPg4;zlb3&ARypcp?)Uy{;AbmY6S*;iN2y18X~y<eI1?gMVtNkc8qf5
zj6LB4iJ4+IE$NB8B22eM#D!(}=b=-8){U*(bZub*DX}g{-2_DqI0M|v^P`C9=sEOI
zhd6zTV9+SE(#=_92=w01);xW?3!mG*x#BlH@?`L`%xXiZl@rIn#D~BjZC#xC>F(6p
z(a{lbd<3Y(0=3cMmtE~E%8mG0_TC-Oue&6j11ab4;cYPl^4`s3<Jt|F<7dEYC(3u%
z<(ndshiQ;z0ddD!0svo<tXI3{9`mh5Bt@zE#=Qf7v>0Q6Pkq>&Es^%j%wQC#HUz>-
zYj#<yY%qV@FA$^3_YZc&Z$s(@*Z8J;!U*0D;h+0IX7Vx!rj>Gu@Mt?bgJdktuVjo2
zYQ(!N|MHei(qe!l$s~qs@c^qyESfBfFj|Rk6?`qGx7z?0Mre=qSTK9`q4fDMtbKIQ
z@HG!S3#?M(<1;OM?eR+}_z-3IsngYo9du$G{1ED*Cu-n4B$1UT>umeroNiX0wPIx<
z=##eU%S-X*>hbjR9Zz1U8JJy)^Il$N#~hIfZT_ucRFEYFqq$3J1Hv{NXL83*4*m>0
z$=6Qy@1@-4ggJ?)ppC|Q3dItw0*=~<vxkA=tXi`4^pBIT=x;ey+=4T4fYBU3IBL*O
zZ*BIyfwhM7KlUu`qPRcP>={f2h0QJK3LYffJQKbXmFT59&5y_rUY66#ffVH`<yH{e
zYzUqoL4pS2D};6mu=*$*ShK@~GwDvnbj?nTZ3Lge=?HDTJt}S=y2s>J5+%BSOzrG+
zksyM4U4~;z2Y>%;k-B=Ldg9LF2nz0>#OoiO=hAN{`w{BULLuPL>7UW{nD!y1xjDPl
zTEQKN8rpcx%bIn5rL9c!)6)#$FI|t_Fj&iO`^IPr`@cWH)P&MR|I7OxV$R|r{ODJ1
zWm!$(wD916`MPng1ZSshnZ+ebSI!{wa%B0&vrzq=M0WVXDu#-|_wN)M@N8fhURH#v
z7BQ51HIiS52RS1L16dza|2aZvvQ_2~=ekYMdnFJ{a`)TiLUW^9S0qciy`SRDry$pl
zu+sVm^_sv$v{KQU+ri&Wa#urT_D=BiN>*2$nKpcEs3a~Mg#3DAe8=~iV<vFcg(W^x
zPbvjn(pvDeJidB<)zfJn5@r;lq&9nYyklHFbpV{rou;o<U%aJWx`249=3W{I`F1BB
zs-kVl2vX!!Z>u|s+fie3N<neta#{HhqHZkE{Q+Xn3nL&XAyQhP{1ZV(@cf^;grsIe
zE2K!Xgn4gaPV$)e&&~bA-Q$yu4Zqmlt%>%-u4!V7QjuCrL{E3&#+M3np1VjVaoS<f
zl@i*LesvHPUq8aq?pcB1C&iK6!X3!w<#nCROy4nI_abeg>h^~;cF!#=#qN%%(|&_0
zpsI3}Dg7-SYy<NSRr~bl!+GIgYSk5}Vs+%;tf<npF#iOZ7;Djs#Y0x#AC4pXc+t8d
z#?712p`wcx!Nn}}+5vgXTQOhph8L<OkaF2SG$3$(rQEDBm3#cE-;AeJ&@-_Ptv;XJ
zP%#;LNJYJRA}+YtOiH)bd$ysSGSMgvOiY9k2eSKkh2eC%w|7bRj=CC-dUgJRHuRk3
z&x1JaR}Vt-*J|?=rtc}}PT`Q6C5u}}ZRY*2jWonG=$Yg%w|~Wt6H~<R{dgEXQ-yip
zQx!dP<JH4w>fS%?EaUUOf1_s=7OWPyfIbmSj|*@IUE_C+X9ZuYtzeyJNP65O0fdOl
zFG5D0or+gGhs4D1?WF2v>F2Ks^Y!`Vv)rw*_Q|K=nN?oAOi3Uqr8&uJbF=~h9Hie_
z&uRKeY2}7Sf7A^2{>}f~5Io#^NL|B}F*&BGFx$qXHoT2xrnP>inZy@dnfTMSvbtEp
zeP>oy^_W;|$v8rO?CJ58$^^j=C|V5-2kfQJCq{w=^^0Yn+JN)An|+XibTT2M@n10Y
z*3b-Dcisa9LT~z5Rjn;pkiAHb%eo_}a~SV4s`M(xp-Hpm$F-JpmW!q+)26o7<uN(w
zwGeV_!0j9QnMpUU>*ZH7!^UT)Ij<B^5ORHQ4OPBQw$Xe{A+?*+-?aY%FWcDZqA^cy
z)Sm<nI;!jSCph01i8g7j=&E;)2YrD(<D)1D4=VOj%FB{J^yL{<)0C<YApdA1`a9IM
zSd0@xL>L$lfL+mzXI+m!C7lIDcrYV7XsG6~$-v+D$Meu9UiD$O&Vw{X1_lDvy3Chu
zayYB@)2!kCA`8cE+%o_cZ|njZiYqPI+04pG(GLS?t_dZ*Dt7N3$UkwQcFYr<(1Pli
z`B0GHV;abPKpU`EdS3sOS1?WE`?r0pT>wG+2+=80_9q{N=DI2(f((Tm3uj38p9xZ7
zFGVb`Jp4Uy8w~a5aeu-9r31O-s@D7|WMqB8l+US$%t%2`-vMwC`Xp*WWQ;GJEb`!2
z-8`!<EOvc+HX&)2iB5LHTzf>x$AEO<BvwTBb7Z^B+MkB19!j*2-5vUxwzl@IRfz<Y
z&hixmy`F~1<(^NVhXG`o*-0-=YkzoHW_ee$Pf0=$<#m&6cZJChPKOZ9DGb&l^lGV(
z9t5`=iqf`mIpt>+<%S=KqveGd%JaG8<;=72wvG2J)8SSv{z(hWa$oxNB5wI%TN{k#
zA-sio`uS_wTqZSMl=2Jz${&0eSel&kk2Yr#G~$h)4eE>5LP4W9Q9U@1Oy{jI2XjuR
z1kIOLAu>QL(Z{b&IeyZW-?MT#J0vA*?bDn0y;uz4XUlC3=u=45??FugCY}EjqbSQa
z7Q?<RZXE29skynin|~JGg%{+`a{mn$(oheFz$TPdHhwq68*#f#+Q2!{wtnL(e6zPR
zUa-27sLF4u-5F7XUoC9|ke*P+NjuEe`43GJV?BH=B<$AfD^O?(9PY)f=vPd)7^)nk
zKePAfs4MNIaHtrt8P?e9mG08x94c6@AqbZZx-Q-|gf~r>7Vh&j9`<ycrwcwSSJnX`
zxU<AdC{i>|pJqEkrs0EXZDH;HL5W$XuUWt#)_I{%th;n9L)6PBLE&MyXRB$|GAf;e
z!~LgC`;)V+Wre|<)g+mr`owjpsy%IN>Z{o(aV0qX3=rJq*G$ex%*uc&KCdWCBarOp
z8jPiVZbGt2MVX2j0#cejiV^&Y7NdSjMbe?XZ(D@;c)=+oXFIBGLN|KRqQ76m{JMln
zX+Zgk+~=zrDy$Bxg@3pIu-%)Kz4%AwR<g;=T?B743U+XjPOj#II9*vqsK}`-OcoIp
zFn<<_WR676u!9)XYAG*G!=!)?9t8V9Bx`^mEJg5LG_C76IXK#(`LU?Rp0v$1cqY)%
zqwtyqOrA54yxfCer6Qwd$TO4AoY2*K<0uzbbly#>R7zE)idsNoVRME|5W?N>0_cfi
z9-)fdhTgCcbH973{EdlxpRhGu(xvc+?9u19I}OKF=1Pp1t$c_x7XbPs!Oi6Rf72~@
z?J)Jilhwuo1&?l)Qxn5^gbd+ux@e*c)vudGBr^2{IP^O9_4?8D_5uQGq&1~AXDlzh
zdZ=Y$MxnK<50=6Pdb+Gq((2Fch!HoKyG8b^c)M!0wY8Q1Yix@4f~?V|Lvwd(>#;r9
z2J+#K!)dUxwBauB_wPd=3}Mx~&|GR`-z>b@mTXxB4Ur13-f@GDn_Uv!3AGu*HsN0C
zlJ>Plq#Ma&<W04>lmD4)*gxwe@e@u?PJ-=4+%$A#TF3iw<&XTZq_$WkN!T}Awr0>n
zv#5xF(1o8DN|WQQeWNA4Wu7d94PBZ^QPTzCm+QRp)__j@C2tKCv%F}Z9XSSP=q6?h
zbIzHz5o|wJt%ye>v^7)r!fGBvf4?9O6d0-yE27rX{u3ub54OtVG_j-d^DKE9p>rd*
zwZB<59%WS@?G#yKN%)yuLa+6M?b3yp)rvQsYNiDL-cJN6E=C&X(kx^I;tv1)AyDRe
zffGheK=G^>z|NhUPw&WzhirowDxT;;&-6^sUSBxV<Uqtl#Heh+^BUR-ZImzF2PP!~
zE--}H?j&BQzwoP9u^JF@e7sT3kuG4!*3Ue2`b4<%#vUJvn%-y};1g-La-X(XY>WRv
z*<eh*hgfK9?A*O8P}_EChu+TNNF_?xO><}SmT(6fOxGqx0hUwPvGr(BAuYq_E<L^c
zbHVc=Y}>MSeM?bbhG$Svxx>W;en!YKzo+ZMoiv#bwT6-TF)hV($-1d2h<+bW3WnOi
z0vXH>xkPwr7$FE;dBT@K!qh*S{_IUBkpdPKWz2XuG;MwFURbK@lm%lyyr0}g8w~6{
zIo&(_i*LIV#YDY}-tMK<X87@bX)0|8@tDKeDGkS)nkf04s|5yydLp{F2ywTIo+-6s
zwQIVkNyT&a)scGDZi(tLS?Tj`xBPmk_LyW|FJn}m450NF^1SBg=TCosrK_%Hw_G2i
z@cFC*9;4MNk)I#eC?o0A9z8#iy@IV9AIn;cMA?rQ=Eb1KTO_9n&H3M&fF!24q1e5j
zHlJ2PJG;AqylO3xRuZ0YVL)VU?Bog}$BPAdQ$Ah9I?<%VrX7zb`&iw2;Ot!osi>ME
zht-GG1JjbA0gD;JPntWPFQF|naXOmdPmaQlM;d8twQXHM(b&PCHs)VWywJhoqH8K&
z`%y?UagoP{gU(xaw*Me++E7*jVfieAJt~;$shy3OF4`jRZroT_baDd>6<Z%o>s+qF
z)W7(TM`2>S{c5wbIpx#4Rb|)AaGQHkL@IEHqwB}md3vMDRHeB*Q92O)K1iu3<TK){
zEfs?JU1fOV(oKdN>+Tp!bHA{wgV}@kk}r!fR;wzewBUOR0<%Q2z;w~|RV+N~$z+df
zG275e*=$Zbd1gTdvFG1j)R~5Vr90#A=ZfK&f!|(*T{03}N2Lq!tEs3SpOOP;gNHyj
z7E{F@H~WsGdL2d%`n^myKVZul;rZC{+irQs*=x&VO%5Y-+R-Ws>K&Z!9v}Y=#**GV
z2*0`a?tEGzeHFvacM>}TC&Y;o+8vMSxm{v|@gI%{Ll`v1wd2NmB*Ljw0emdkp9Y7H
z<My%k4vof$YB#3^*;$1hJS{N;#<1;!9VU&e!k4-nGJLSF3u!3kjO0hP7Nq}Y!%MHs
zzR}E+d&%`FuoQh)Q7CH!qQUDU-Y^p=p!t3v^rGynO-#!rDBV-54~*4@S*%UxQemmJ
z|8@L$`+haF$GtucLN{q+OSQ|gQiR>yGE`9yR5`w<bNqwusc)+_KyfPn^yIG0;GdCx
z@r<cHFKBv!)JwjkK_ON-OCNW3I&P!!o|#x|!%RDYyt8+((b~`*-i-giCX_ewX0NyX
zK80V3Cf%$)TLDf4-XCvDrwdTV3I%5eV70D~E^_!_%bb0}mS2SLs7OD2MyIHq=+#k{
zjP@ezw@%k#U(C4x3z`|)h9WJ!zJw^PeuGX!GaMU=NlpkiH)qBTldcrPZwXKRd4GER
zU7NBn)>KHLG_+r)SJXoO{uv!W7i9mr(h#D)4ieI{@oR0B{aBJaH88;N(sL$^R!HbW
zt0R@|qy|o!@yfW5$ZHzQ@P4N*OufxocKmDqS9-YExqSeuaPw*=sKvi^4;0D)o~7RD
zGhdXQC$-x|T*@??A^&2}9s8ly@;C;45#GvL9u0-*A@m9pA=45=-&%fUTGZRL(?s11
z?Ox+6lCO>rs?!D3h-xGUi~j-x7cqEj*v<8ObRW7}e|GJIMy6;Xdb?<1>3pj0l**>x
z96oHW4Oj_9l+Co?5_-J7VBy({{&~YV<s;5HB|)-#hz&=BbSEm83!|_65R}C|OZ*;f
zAYLiEA?oN#);APncmJ8+CWaGTDU?_L2d%c1VQ~VN7HB6n{~t}~9?tat|Nl9}$Z3=)
zF>|PSJ0Eh$DHd`phm=E3kwZ>%K5k_(l2glB2gRbM6XwtmiWud%vY10GErwyvey`8(
zy1v)7%U}D$#q+g2ACLR}e!I`tyucyADjl7Ig4fbsMMeF!&oJ%r8)aJdZ{(_o<@|&_
zTHo4wOJ*JJQ4UetOZ~()_V+t`cb@s^ql<cY?s9aU$kpiF6Dp&;JH7Yyl#%+7ww-XV
zz5|&ruO%@qcS;U~#>!uU>pv>Dz7+Mfh_%=?H<!_q(K=FN)hT<G7s?}{ZiI#0R)SZ2
z!}s*;Hf&6UhS!QuY&3@1V$uCUkV_l^cJ8m#-h^`$I~ky!^u5QyZ*Bg8{O$6%B!uif
zb<Z!pI8N|Pw7g{iw^N*pnKZY8i~o!~0QCTMaNFv=7x~g%LStQWvgWwp8H(#TNcEKk
zuZkQkJum7Hx$v*H;cDol4UYxrPYTr82kUgt|8ddKqC_)6M8dp78+JCmIrH<c@<3=e
zBu`$!S1Jgojo(b*Lud)!22Vd*zT&RSbrUU47-ul3>^HQj8&l(q@zM=XPl9d?LF92k
zaV(<P`e1!~H5_%=Z*oX%Pu<${(f<Y;ZD`g9@o*lk${XMG5iv*CU3D+=$a}fD5GkJ)
z<0|}yU}R)cqGX}Y19OXy7s2B`JT1P?OcB`1xMn8>VvqAoy{9yWQ(<r*Jy&Id;Xj<G
z_nb@gwZCL-tx~n}-Hj63IvExaYfu3;#S~-Jf7y+>F-ZSGDC2d+GON_&cn|xFq9!IN
zTG&G<>ICXWzvlVb5k`N|jRnIHU4{YjK&pcaB<t^!ODb(tMO?yMcu&SlmSmk;4`xtX
zJYB;R9~WT|%5vAoMuVBJ3-i~IzL0KksGrOQ@K_@yEZAJn3wgmB+eT?19;`1>HSs`*
zXjhnBR3Au_bA<ahZFQ1c01UT>{84;D=|D7!_Ui7X!4%Ss@K>$78wMw9;q;)=hc6h6
z7%#SCwZ4PGEyjM}Cwlg-5G2j%9KIg6w!U98mO>kU*L)u=d!1tU;EpNGDwv>u>qKi?
zo8aAmQpF3tvqPArxgS=1AUEb+0Ssp`mSuWWQQ&?{ncQ-EI5i?pO4RD;5v`j%zd>J6
zac>FcA~V|Go*b-gFC&M>5B6yXtDDJDF&3SsdM|>6d?dIuOp3iCv-Nl{%dT|%I==&P
z+3~x7knwoxJVb58xvyKf`<G?k2PKgFr7AMu8Hm&O9K^GLrSC#N@T0K#H$tMCN#hxt
zv<UzHi_u1Dx#w=5tgptVCQMG;*hMwQxiCp@sK$0l00c(teaUn45YIQn4YnH18PHeQ
zMGv9xwnN`6{}9|<2nKUvI_6R$<9HF4St}7by29DrW{6G?c1nZ$tKGM0#gZCDj&9)*
z0?@$l2gQT*sBsT>YNOyUMkA?>R8pp{=5J8^8Ebx1yLK!t;HJPk4OsDYR6PeR9hZE4
zkjufip0l>Gj*>#Hj}x$h&*RQMj8y>H<i(q^t<Gwk!Q4sZ+W0f!EO-}bAKX;8%VJ4O
zlg&IR)TQmk-6C4*8{%XOBrAp(K-g>vU#_^vH68s}4Ils=Fo97g{4l!tJROFR_``9-
zeUd%NVKF#<gYieLBmqU(*hu68K@pWui9J^xLA21gt;_ifH+=+I3)B`!WJv&`7@sLp
zPgD+4^)FRqgD}*k@sQTihx%m>1)T$)9CAe_xiM*acK+x;?|YY)LP<B0n81I=AMlW6
zZ*8u1cXnp?+Lr4{VqQ2~XJS)mlcAGW?HvmxI{{AKTOKP^X@O}04&nZa@Mtv&^JzOe
z&hk<i)xLv6&-5c20b7t8{}Rdll@}yya2F;UiI6aAZ>ys$?Ii9VY%J~|Lqhybq32HJ
ze4ewGkmId&W#>rD2w`WyGi2O{A4ymRR^-oP@)5UN-NFgfjYc(kh;WR0W$~Hog&)Z$
z3hzdH5_M2?;Ce67%`OZKnKUXsQUbW;)1z&Z;-a9wGahTW`>^qtrLM#rQ%rCvvJXx#
z&y3LX#aI<vs+`M`lh=^YjKMs6df9)LRd7x8NPNhGg<Og3CDnUaY*mP>E^}eBTqUnG
zBWP5}J=cAMYK&HZ_<&WlUlUBzsGm6fZc5mbQU0ojdu%W8G=u5Irq&Yf2#qKUevRt>
z?$iFQlrtx153(|C{>&u%W!hVz8PwE`Cypnb=^#_RX6gp8q@%>NCR~QkA1tpe&f$A9
zcKZ`btl=d#_D+|wr|atYeJ~fm+Ir#;&xWy-y{7A;sZHfBT$dix#;GmD>Vfq}X2Rp6
zg{?>5N5`<<zWovVuscl&CHNL6%DEMB$O(N0)OuPk<zHBq<{Hh9QmraZD*SDKJ>EVD
z8uq6{`|5+HPzIZhCo(qZPfTe2p4OTA5Kb940iB~Ra~d;#0khbM`z>Qv{}Y3FL(ezY
zgrKiykkiyYA(nr%=70j0NaH)aI>fQ7M7NMr4Q3&`iAk*_QYg{Pj?r|X*1di_I3&qw
zY(WHWV{DiE#Mm%3+0fHo&V#!|{@ukq39JZAGyyGB)OR6e1?=h~Q>fCZYIscYLf}mK
zXW@r$Ffu4+lGa~t;|{gC-8C<-FD4QS#lw0pY~Y>Q8JR+XzyCH9Q&6d?rPZ2`;B@$H
zJ#h_sfnyFjg%&dTh8WTZgWo%q!y#na*9;Y@6q%ZsN*oBfRHYt8Bj#T=QL%~k_LeQ_
z!yzqR<WtYv?oB*G?@k?m9ctY{*ko*M?lVTHEEEHHtdiW5T=W8!Y=3tv-&K}d5r#Y)
zn=CHQ%Y;CFc?!xELnU%;H6<Rcn+8LhbjIA?P#A@eY4Sx+v48MUnN7bnAAMX8t1Xnz
zFn<-=@%a~tEzO}E)Jy0h5w9YB3O5qniIe2&gp=Q|9$Va<3f>rJY4ZAxs1}&{g4Sit
zP}d4xX#aQ%(vas95kL8x3#ajRcJ9=DMD2u!ASc8!OD$LIG#}U}f~E4YjBBZ(LAs%z
z(a=VS5%G}Y{H5*zz9%7z;iVXJ<a0SYl?*RNCnkS9T%(42p$;n>E;0j)F}z}$z9KXs
zS-TDyLdo-jo;!(>Wh3X2-IE6|QludTs2Ap8+_E+o^xjar7}Wlj8~H21A>AJ7X+M~!
zXSDVko(M>f=rjd62`)WL@eY$1lJ#xsN#C~~t4}Kvd0hREed~qKt63uBsyo%)ojoaw
zke@gF-Woif$h(5uUOwn7IQwApeDOusz{yDkdgc`fr$nH);+#adFmVw&FZk^=DqSl-
zqqU?IGV}C%cAk_eSwD9l^&;6}giH<&VEqfrJbR{Q28{&+1Mo!YPhj85SXzoD(bW0@
zyPbcjMGNQpIpye#$1Ee~ePO9;cN|XQ$lXMivBw0Md*;})acUUP?$jF!Aq;>|(AXT!
zw8jBvUhzCISF$@LrL`eN7?lQI<NCTVxBoA|g%IL}(WAHaWXvOH%hUa#;Jep3j;Uia
zB}CHI6C?2X@Of>Ak;ZTmBVJHk2yA>a5afqJgSv|yOOYpahk{BM5{%n&t7}8cT?8)6
zd)nXnAOTZlfpj^~NbbJ^!UDIS>*|h4L+Gqb2p-p-yzNU6JUNRRVZT{OY;Du<DYzLx
zA%=vMvTpt?H~jNfm<K2*gQSCNsv}!ads`Z3G9wsv?tDbUqLIlX`KkK{p-U?BuL<9_
zwdvO{FEf{`6Zo-oy6y)sh#E|rSPD%^A%!Kmu-uYfh0^{mEj7Mzrv?)$^2O-E6?q6z
zo6<bTnS_!O4U`9)`am}GnAsh4*wqk5t+{_TrdLm$0+X9r3^XjYV5CrW{JHprC)roa
zmR!_|BtcQ7A~lWU^<qQVA3h73c4q}ZENz8_DzXB7YCVy(-xJ<m6Ra8(o@5Wh&_`xH
zfl!7!dy04!vRYrhm;95W^NSwT$5c!ksyQK_tsaWH5Uu~jsgGAe$;{>Ai-ty`!AVB>
zMa6R?VZFBSGxnO_x21=YyxuU}QkE7eZzy!iVNs**Vu94xI&B)13*@53l7(qLG=KNX
zffc=f=rz<;dqy?uqUL`kl91)FumA&7j(pt>v(c|^!zb>9DtCGu=g0Q8vUjhZ3b5CW
zHje{TNu`8(Q_G*cPRP!Nr3^MeO#H|skVqdZZrrGfIRVE_$=n*6O5}2Fl^H?ZkITmI
zwL@lP+lGEjPI~rOZn@9@nHju|zN21dt>Pezj%X`qH?pS=|FTW?2fX%Y*d_;gVq%V<
z%_$FpjU`2$Z>(HXP7hNC^Mxl&yNX3E976?62Tbb>bs0WT3<XW-j4q8&gwkYkg&W3$
zO6wSQ2pDI0zQgv`h6???l(?ca$wT*#Xh3hbz$dxk@`B7uOm^gdLslClUnkEZ<35PN
z;<>r^TWRNfZz_o7JH-lv$ncTrbQN`M0lOlrrhbXRV9dvcDB-k<qLULN(BtRvz|mTf
z!1@M5{3M^~aR@||<V0x7YT<(_bD1WZ1y};bUFN(y#x`(<8tSg*)ngw|ZHd+x1Bi|4
z|5c*7s>xg``Zpdbxacyds)ReO&oY%ng>d<t5j7*L@$tzoJOn<LjH>gyj*V<IhPF04
zG-$HW;<92Ic2P!B;`)C#kUvq4p-fNSztXY_SHT*<LaI++&B=%yHIShCDAT4cih#XY
zT{+Gc57V!5wtmbolxaNodn8}|tiU7|5`?R3pWA8UHPU*+XkqRe4oqug{~O*ViqzFG
zrZ)A=LrA#oL{bYc$PQ*<xWJY(udIfz8alv*AqwAE%EN$m5?*!*MJKaXdmFIDawX08
zMGMTe7a46?^+AuVYdpJ^Uu?H+Qq=HX2YZb9mO6K89G}GfGoeWj0kdt5*t-Z*??HBl
z3La-84mo~GE*_!TQ$4Yn+PcI@JYDW${~DrSt_6DeD)*fpCrshdm9pnoOax6A-`yy?
zew(qm_12J-q^D|$!v9e;^ae_^F;ADDRl}<aSaV%-jgw)NhO)w%w~-GIwpnzWg!A^S
z-HaYw!F|;RMSH-etneVNL$<Ja+!_lw3#d`nwuYU5Y|9<{^aWo@PBOs2FH6=QI|(Yr
zM)hT8s6~&K|LWoZ#EAH>*GB;(MDucZI3tmf#xVy27ImFY@cxe?h^Ab@_?N+38;Lq2
z>cA~H<zRhrE!J!!E@OT*BA&f<8M!=cVr259jym;kuu1CntJWa6dijAMdhRg&;G<<h
z>hu0dOGTjZ7^Y=nB$c{1)AhY~ACQo`n5exOpX;OA7J1MU7q$5Uu0EQI<F_>9F2)<o
zDolr=+<x8q^5fhTaUk)Qu8`fd(`-Vm-9{rbiQ3-wk<0>K+Vq&Crww(mNUD9&#!Mu=
zoa@;u^$Q^rv7rNrIoaY8)jWi<(pL`zzBLt><RR=X!d|a!NlpRadVL+=yYMrtt;RUb
zsHix5Iaw~h{{q5zre1gvfiGqDlz{iws8SR^RF&n}hk(-^KaaN1lFYGO_tH4RV%d2d
z$Lqw$%kK9!HZT&FW<`2&6er;iZOn-=W#zX-4XxEin~+v)PIa8B<g?_t11><yuP*$e
zot&2d&4~qZTmC77--WCiu6jT}G)*R^6)cpqobLukWHO&gYc;H1$$nsjX{~7vMm5pP
zkpng3VMuD+Qe1n2WS7>0MDE{+MYHbemhj!w)_Y}-JPg%a<(hxSuflTDtj}QF94o46
zSvN94a6E7<LzBKHlvIPiPs!cQEY2+M{J1gRTCbR)fY@kDLb^hj&lP+`@}p`S$Rhv;
z)AAo6_&*su^XLo4iQt<~dX<POxfLad25ZLwi2Se#cN%{Cdt$i*mF}Rn8?(E!z3p|l
z^~Rk+onkAJgSIA|dBTSV?q=HD#c7_AI`vgnH!y{l!RKA71k>5*ZvZ}~8>bS(W0a0(
z@;pGF)wp{<p^0DM@n&%=4@~~V#cC+}>6FJV!4vAYvr;FuU@6ztCWdMXlrcGVs%3Wt
z3#~M;+x_u@f{fRzRf4ruObtSowk|sv*lKy>aE{hO8r8QNt|li-c=dFYwFZ?cAvIq_
z_|29dopQ($<S5A$<3jZ5%VjMFZVK;QgHBkvTO!b}IJl3CKVT5rKQ2)kHWqw10s{Xk
z%?u)<D(Z^*W8JIR<tV+p(WS<4AV*PWw~?F}LfyQK$2$h3bMC0y-KcI<AqdZPfOA0|
zPc6Ubo{5{Dys>72oI$<fP2qkSbdsYF;};lzt@?>^YA^iWE)%u0x;WDdephA&<rd~e
zi(LdYT{?k{+<Isu8DU{|R$rDZhu{c%e<h<=<WZzj(7TBN1S%$cX=#14OISv@$KTD`
zzzO~8y6LaR*tXY{jjQ-ln}TuKWcQG8%f+q>H9|b*JigL1Vq-Hk7b`ont$rjmJs5Ma
z-U~0x)zIdtH0AycBv*z$A@Af6aGGEC|4vlh&*IWvbPt*Amer#>+5mJ(dXRlNR<w$H
zxoe%9!=7Vq2^EiT{{$YF7aiMQZ&|i$B+$k^$3AJ~7uVh|h0Do`PIoXt+{s5_)IYB=
z&rW5`Nz@DG-Q*ViPgEbKF<)@v8X8>8{d_R0RMAQxGwo|vV~VhaYKf>TEG}<+fb92F
z>epnjO0j0PNb&AV(Gv?@^|l!a%+tS;sGr*+w>d2*exOEQGeemTgCgDwq|6}V5HyO{
zJ|Ir=RE{fOy8BJg;GQR5m)6sANk3_;F?_e}QF+K<t)|0>wu8TFuY_RB%bKL=nU%{S
z1I<!Gyk@~E!2OQwXy31CVkjMD&oSj@?!B<L8QbDv&j_^<K-6UJ&C|VtD;DNg)|V^d
zY1B4P)Va64y(1o$U)z?}VkZu_y$;rZN&N6cbRZ+un3RA?EarthXvhrC0CWHI)xUME
zI7;%rLdT=f`ZnfnVJGrZR7LFbe!~75@&;wVj$VFO6j_N&S4xMO?f#+Rap11!g6FR_
zEw=G$Jkox@GTd^HR!3MWh*x@qcDi=r(Nse#>NY78Q1^>Uk$jjIGES0jOQ#|9Ke*Q*
zzKv<Sg{%!QC{haR4!T}g9tCe0cZ*siH2@!ur*3KXYl{`smi|TOW5F=ZD<Mlrefrfr
z^&(Et`=FwnEJ^EWk;ge=2T)re6=AG=Kh@Zi65I9<#v%UOKH?k69Z`+bGXe&y6jaJ0
zX})~<^>{h?AF?GCXn0CGU()BmlzFKA>rLq3?&{@g|GEuzyUL-K?5&zso?l;!*mEV}
zP|SHD*;J~}>sEbk%|d5UKi?PKCMn5>>+u5YE<**_(&;W<ZSTEI+uyZ*le{y2DX*0f
z7(}fFxav1p_t9gPb5`6DVSDYH!Kg47w}${6W&>#?BfXGN^4BB2YFvlBIqpDbY`>Ql
z%J9EY?}hm7hztox4*)BQXG!G92Zv?ZmW#rW)-k2X!Juy<L++=pNZRw|yM=RUn}Gzp
zhulL*aryQbgv3mD9>17h5io!gfYLSHG}DE0$H&!<AtgY^_1AkAjnUnxct<mNE#?qB
zPze{+SEH`L&-uV;r|m|2+hc_1Xw|N_Gh5w>BhE}-=!NN8q_+wfu3=lHcfarDj94ny
zeGBfLz}GcDRVQ7<Ad!8Kq;23Vj8nqi%>GD598a9=NzkuNLfN(W3dMzg9Qgjv&LIR4
zWZAbaSbyF})`854wz`KJab$GOTth|?8;O^r*p19bPCkkf;HpsS#9XT|EBR&2jn(0c
z7h8CpJ2bUApo>&zq_4a!%x++`J@?Lvr={*SBq2s$xloMa299`q3$P2nd6T4i1Y(@J
z)x<OcE{#I^V+A($KPOw)0ph|xQ=6}r5%-}1b^e;dF0G-&P!z4FFNGRDtiV}xQLys*
ziHG1f4FR?`B~zzmHVH<kgnc^dApD8(lhp0)`Hk(9{KGp-irv2<!9w1Cw9_jp`<7Bx
zzIQx}qiU8R!JO!xs`uwPf4P~^)*0KINP}>qpzLsDqEi2y&*SP+KF^CCVn_Vc5djD@
zEqwW}AtP>&Nt7Ey+}i3G2bS8fzsD+4!?R`<Ba%G|%`f(`kR5IJ%c4Jk2jP!~-B=O!
z2!s<s?xI%Wg1BroPk!)XrPLKX9~9A2o8DSmyJM)P@4xt<t)%GUaSrg;NlO7#DgFY!
zw<^0NB79&nKOb|qL#|t8nZXAVyBIh*aQ#75(pg7S6c04p>Yw%_!{#2~(eGyg4!!x;
zO8R(5<k#bdJ1w5$R0?GzhE%)Wc!M~?*jZj_Wm~T2D$slIGNTgcL5jQBoh!MdxW$>e
z{CiMv#=F!H;2vh~@xzQamyh2(_eJ)}sqXCze}nFu)pM9Ek>bmzm#JDVkLpqwyNXIU
zF#{e{(*_60`@RI!M~ch;)c*5wBXaU!;GoZSfBp{<=bwDj|DR$UGBD!o?xZDirGI35
z4G<Ytxx|o^RMM|HN@!B)zTP9}kNs_vExaOC<?vhfw&eczI1iCbr)5&ZI2N3z-09pn
z-U8OxZ7iY1m8=s+7(O$qO7D8c!;)xB7V5Zt2M+Hhd$ua5RG&7p$k_>ZR4vz%%TpH}
zdK!o;b-G|x>pcSi2<zzk$cZreWPbYlc;#OmpKLz~f{oG_LHX$qI=Irw*S|O_`U=SN
zB<gp!po{OlG%$t!6hU#r0RPKO1i9QJx_yZ>(5$X$rOWPhG30z!^QAG=&!S8eC2T=u
z^oMvY%=wf6u^PH8Ov?$On1jRPzNUdPod$L21;s-Guon#@RED-pVON*-w~RS1)0f(N
z)#7r90JAS1HkWL{u~Ct&Q{%KOOJNY$wwJ75^;-!2PD+^CNZSN>r!Fz2)m^I^bbFkm
zQq|S!-x}#-><$i7=pAcbFiKfr{;qiNWEY2zo=;0(V>rem*u^zuBGLTmeB>Y*e#KhZ
z_uIGd!<EDRIr_IE3Wd73VbL3=damu~<&9Mz*-v=`<{&ZNyq8?ptP9-Dv#{Mdb_qNq
zQ#U7plXMXKGl{ueC{Yqwll&Skmc8B`(JO0E>C~N<E_>I!xD`|5hSCxXGWX)MCB&rY
z(~Oy{uC;@%k-3qXNkIDPxfjK0*2gC{z;X=MI8Ip(gI7V)YV;jgQp+_rWu_z-Q)t$z
zEKSJ6tS|uQ0HicS2x(_~FiQMs3bS2N(|@e&G1^(DOtsfoGwHO5q~}Ha3y5BDw(Lxp
zY)(>!;xm5lJP=Ah(hTW$#i4w#Xk&XD&}VAcJTpLzU}izq4b`GI;dmk|CcL%gpX>ct
zxe%?Mkr6rB;|pER&SPW6Z~0;_MnzMS5K-AI%gS+`FQ)g}!qqoCi^{Ujo2}=+<N>40
zH6ZSJWfLoN7k=dic8%5BgabFpfLdEHc13WzlWfY3Lt%Ld6?tGiKS$>-Y<W=T?*+w&
zLD55POu*{wL=XWwP#*&6qYtE>?e}eM6;GsPF4}e@#z%rkAKf?cj{J#;qNxE{e=EM?
zvv?fRGk{a-POrw57dUU_T-d1jK$pku`cFo{)yJNUh6oAORr?NCC1C(43tFqHyucC~
zrG#2kC0x|jLfjQ7Z&FV{zq(?&;L+s9o@$MD)evKa(Tn;Mf9@z=;7Z})!hj&t{L|AA
zIV*t+IA3qVkczEyYcS&N)A=hc{ibZ69KfZ3|EGsGUS0r0L~Q>B?}oOP*TpDd^YF)V
zI*y6Y1}_i0Tm+dSEZ=n5HT;AcsWfOhq2Q9f<<m!sAM)~Mqm+?-i*3!42UhLW+Ph$<
zuOl-e)4QqPjAVWLf|4Q4I7dxPPpp#@W7Ts%pPBTXSiP%CbPrM1A2A#Q)FbGG-LoZ_
z#cdXOr2bdRC91S^z~<K0Wov!uToM!humoFH8vmo<sDl0DYJamFdA+MumHZs2<`*^M
zywo^X$jCBwcfsUI_><I>?ViJ(9@JI_%FWf?)#wSkOS7mT&#EVm-}uMrG<1wtj%@C`
zGh*zi!71VwrnIn43aH)Yb=!^2B1T*6lePL$L;qyAzt4~tHc|#ga7ZDmKRg^hbucO0
za*!04nM>>XRNPpBwUXA_Q+NpqQZmC}&CmC{*DRudY2$z6Z);B%@>+{cUDc@ORbUQT
zYOI4RKrY$#WXy%8cvsgPQ*2qHCX&3jMu4}nnSY%TJ3w>AOoY5|9F0vHccYhpV6iSK
z*Tf+|*(tw6d76`=dmqSWUj9Gfe?Reb7S=7q*IM6>^#T><dEEDVw7Q0l4_-jKB0TYW
zD;#4kYiz5p(NNtI79L*Y@T5I;XZ`vMVWH7KVxp1E#OBv;)bKIX^ol-@mb{lST-IWE
zUsBbI!kKFe@FD`+J;Xg+bmqC*JlEpifY{t~(|Ueegt`c3T_vYTSXF?WD9~z!K79N_
z`l8lY_#c!mT6H)vPJ>(Bj6Y77I|5u?V3BC(<7(si$+V0=OH9-8Lb>uyPK_TfZ>YfH
zbKkdJH6z3EZx`47(%Qt17*fj4-s|8De+z-JgPp|Glu~2E!er1U2yFja#IB*$zs^Ie
z>IID2cCR_DsOANdND&44!MdnZoam^^Pbd61zo3#Nhkc|5yeJK${AfLKue9OCe<H8~
zq^bfPcQO6!=i-Du|B?ms3V7b#`fVVv!eUVqe>PI5+|O`{#Ay_%=be`xD2GYh`3|Yz
zb^b5C!m3Ah@Q+=VrEbww*T)yP#YKv3TKr_8)?d-ep6(1-btNS2!-6bSY1w#+Da@3q
zMpYF$gA1Ak!<Wtqq*=Pka$7)e1|bU`+K%XY89>dNHd4e~LXBjd;JBmsV|OY8JWJ0)
zjg5`Q+$^%rQ}`u%0|Vq}o;I1|#72676MWcRY;~nH^~0zg;x+M<hD?qy$JCB0rR*Y<
z^}nNHocga4Hc@IjlHE-TX@XuE%q&JV5&(3el#c?Q+F0G)|55!GQDlh`8ah0wg61-$
zB!bk(-1@c0JchewutGYwPK+!Btk8k<Jsq{vzZ8%Z6**m=QFEMUA{N7g<@Ei1XhEDx
z`Dcw4<88u<RFS*>d*bXYMO3sKRTi0_Q8$vVj_WZf@(_)BEZ3397+(H&9c7Qf+?g>C
zJXc#k$fpSV;{E;Y0?({(rzaO${~aXtVEf{jH0_&%48RgFl&Y2lyHg$1)Rf$;OR8&@
zT1!3N+=_eEU@VnOY5+D9R81FNlL}f3c0C77VlI0FJjLq(U=Zw+Y?;KwTnjk@f@Pso
z!n>|NvXQJ$K2St52y&<y=f5L!a~p|Qk5_Bk`5!-6LpDz(GU~+n5=!v+IixYbY;<_i
z_VIPddCzgxE~k9!8rxng;>PB7KP!Q>GzXTfyqU_5*aO@OUi<5-%!7qDsfQHQ?s6Be
zg{rhmbRA55va)PybN~H~LAq;@@p~b2&|MCM_Jl1hW(3u99Q8fvgic(Ms{5cq)XA+A
z3!g%w!B-yUcYvXNT^kDv0n3np3w|KpUT$0TIEUaTr-~x~tF~WaqU!62gqp18Bxt14
zaiInAcVUcq<R&`ExZ2r5T@mWW#Iu@ho-kH^uo&BWi8t&iRZzRzR^?bndyi_j|Dx8g
zKt`vwwyCrQbp-M7+_+hJ)F%T?3%MBc%mD$YsF+Xjo%>BcCx6-?$W^#q4jHq)1kB1r
zqrz;96!^QLI_PxZie1AKdq$!g+hyv88l4!^+VF0JjCd(z&?u->7f^`hhvZ&6JZme9
zzz8+V<#FXcyfn-GQvB>}QuyT4^M+D%G8$C(ciA)uMNTy``*L_QehSiJ^{}p*ums%6
z6A#x{qQSiK9Ua+l1qpU}s~}?y%T;_Y;>;<QzojyKrRA|p{e8uNz9UiRKvx>T7L~ZO
zRc~eczI-~_jms0`6j;3qR{*#2PlXEc<s~sallHF8LePYUZjWkfjJUD!{h-H%K?cT#
z>#7V?;`s{I%%fU1MAa|9tHia$PX7tIvgA3j8eblaF{e>OX@f-}&<J1OHR!B|5`1%N
zCy|uAw~|^yC&M8`_tx9g#pK=oZ{JYB<sp_;J#(i<-c*7gEM4#33H``=Ct{SY-2>8>
z<w0P?ELzx)p`4qlbZRUB&I;zx|BTyiLXFO8YWN^7&rVG=h7+lwCAUfy`y@HZFy*|h
zoZp<c?%tUV5a50Q)^hc8x~b?Z)I}`nkMWeD71CnnafJ!h+;w5F?Bb}3kVlQLuzXP1
zEk$X;2yRYW<*;!2!%`p6Q!2U$mTeT{D%$}$2;MawGjMRq*@{awxY8GNU0mig3aBP&
z%^SDe&B>I2W5sS|$p_>6Q>o53f7j@1wC*k@Z?A7})4kfUW{lx6ER3?Vvf{0j5)M>S
z0Jl~rL3Cs#$#xzuGj({70qp$OyLN#^j<E2^$M3z~N175&I7JDHP$*6uliynY>(I53
z=l)#mE-*D7H~IDO(|=Is7jY;^Lkd%WRpi(Mu#Se`EC?tyIoVU{GhM4>$2&Rc(fJ7?
zkzVNUZKN*J(TpotGh{7XE1ezRnQ!b53q}Ne#L)RZ+q$%ByBuFcW4KO9K3z@<uWbo0
zkE+#aZ#z9TD9jj_p*uz?B`7IG4&rQJSrH?7yYKGEf)s>c?a$Nli&2_GZ$YynxOv0!
zs!zuxmg8lPC0thZQ^_UIeK`eoL`z8iMR@4>DZ1Q$sy@0fytMw_7?qlkVjNo3NPEGA
zdG_!_HS;A~o?BjvTk5J2gtJb+OK^Rcq?evD7tRCH@V)dExop>+9Nv<tBYe|%utw9g
zDeK6|$o<0EqgP}UN<b(}{$CedfDPj~+Y`Vl=neX{b!2|&e$?|maFop4)VK$VI+}gq
zd<=G~v?jsS-q4tS{;SkoQM@CFM<vtJ{GOAUb??tf!@Ix61R)IPfzvKrd3s#u_LrUp
zK!%cX9rfrzCHe5s<W_>_(||CATdzj?l7JRQb{MrWH|;ew_C5;BmYMQWqmRewsBBMR
zNRQR<{qyf{2={OS`H*Fh`t*M*d-JXgl@XeWk>PLulA<FOUS69fhm4{psDzY+X28K@
zX6qqp884uy2Z0{BE41S`y<&O(=MPN;f=RgfT0y0sj3BUA;uBi;cDDUPm4x8)aC}W+
zO`%)XrPi8a6g`wmiSbl(Mf6p1zWaOmB3F?n*-{*hRj9gBd7x-|gj?$3Q43^RC?j@g
zYV7nRx<*RojIZ&aWt_qnHzd&JJ1Nv(5S4?aruubz9OZyUiv0}N_3bN{GQ>M0{RaWx
z!Bp5T<3A47eqd`hpPZVQTuxV@1pOdj_2vy_ave|(4%AT6<W(J&2KCw45TWmYx`9=v
zJN73}?1X*@%^m-xEKxIF8Stdt1IRw%+xONMSw$3QI#}JQvzWe}A%jXjSes8wN=irx
zfr($hG{^@SDv#vK&<?k{GU6PBt)A8fT%|iAmGq08$AQ%}w|5C0-}QR8HgrUFxtDtm
z)%(clq<0OoWwD*1d1Y}J0rTdS@5XF})3u5d-i8E!byCT^+`2ghOt09VfuvUuM0KX+
z1H%r%3h-%ysGMJIgKfX&l9~S_LQF%qDFUa*`>l(b4za4OO+0n>-32cG4(y*w&zn^C
zydF>_ndFLBaM;%S)D>wF5W`fOP)<jAD#4(APirk-zT71(UBYC7d25g4i)K7sO!=oQ
zJxR)x9nyA)YwkhMstf)5a?N7EU#Uh@UFdCN+R<=Ag_m0U-uy^^5s^iW@jCD4YyIaR
zENu2KJX=l_)&xC!=i)SU(np;Gq}P;tMG!P~_#eVzp+M~JdTQH54DED;lZ`%V?4zx|
zET$&VROz?Imx^QPVA;y$lhbJRUVdVX3CXiMp|of*qBy5#eihh@7|PG$^zqLM{J1Ye
z7VYX8pdnz^<revy1&GweX&pST6r@Brn*rHGnKuq3b$LF>7z(MDZ!N32sQu`Q7Zbp}
zSS!ps-tJ^_m!&x5NUu)rqT)sj9h?*@6Kb6X_+<T+6z;wH!viuoSTQc*?{IMNVEhg%
zkD%5q+_2tXnc?yluymwRV}J5Y!^t9Ve>+wy)iO3FlAc%(8b)@8l7bk-#lH)bS9{xZ
zwZm_RyAI#HTr&_sRzaM;-IdVUHivT{B0%dD>WEB!K6CRaHztpT0@fKTGm|6EnxGp*
z)tm%3$LJGBK)~cql@xRqG5n>P`f$qb5t7eWqy6<Mh&tXQ7@GV12<Od(0#RXSpQf6u
z=jQf?8gd-zyhlo$X#Y0)KPzL`a@!9h_3cTYoULVfHs<PYRyy`?MV9WqjH7#a9EUj)
z{@yu;6UF}q)f$0OQ3TVQwZ12J)8R3IO6eBz6@m5+>zQAi++{b2cSNi-GAoD4VWAGq
z&=0S4f`m>@uD>4v-Wf#lB*s$t)cZa*Y%cb*=yVEeqncyt2qbE@r)O?;28>n1k5Gt2
zR;>AgK0ZiEIAePfnC%x=<}&aJN+QpKhuvsI#1A&AwI*koSF1+RK)O)IIMsO!4D{Hn
zit8dxr5gUy_-~893k_i!jQ`U=fe_WwmZ^IFBgs^twC#-a-uug+!0LR9<7JzqG<V<f
zvlPT5o?~YPNX#r%i``}OXK99!E?>IIGl74Fu-(nJmKO4kX@rgQZ^2S`MLI#2pU51W
z<Ka|PTiE&ORu8;pesQa|&P9s_L?-0jt8+_9AzVi1aO<`Lci3Djkqb*<X(0nrA^?b^
zgU^ZE)V0;=iJoBP%LiW?^tq+0<$Qj|{)jX30#wV$mR?l~Yo0+bN|~&xIjT-;R^c1s
zlPK2^z1V`^X7+t)Z3_g|YP7+7m$j~Qt97H6D_c(d?#n9?^&?&$W5iTA$-k6*S6IN=
zJ(;3@?mO4$^RkBVfk%?95#k{OYZH22Vk7TIGv?OTm$zC?_AnRpUA%3}gW(Eiv_c5~
zLPLhNqn~cd!Nv?81Hs1*fpYA3%RONYk<;GC6h)eCoq*h@BOd;!e#xCVU`V!*eF8#t
z{3>q3S!cE%O<PG3-pPla>BL>@lDLy`eTvZF&a7}UFqX~;C;hfXz)0*#@_^KkoS{Np
z)zeR*?VhR>TomGWxA{HmJDGY{G@W^FRCC#haokIy?)Fng=xP+yL6O(Nd~W;4wxwDP
zXKs+jpR?FMX!IYn__8!M4H!_?exQtzrzGxRy80_JuZ@MZ{+m5%1<&QJx5WMl?EynE
z>Y;>&lv~wt4!Ne6j9Th(PMpXrY2S0w{C9+YtEU@Ei-*NQ<>u8jGG_jGn~HhGgm{jw
zFc^(e`0`o2j5N(eg{(JbZmm3#5qpR0-j9Kdrc4e{<S}H(V7V5wS)bz(4%MeE$OwqH
zbsEs`x+W|OgV1rv*=~+2biNX0!_EOaxz}+{tjJ^5*N>D=hmKH+SR2WY0Fq@=5z8N-
ztt1(FjuSHV6V>2F2K&o#Ccg{3rfzsmP1Wr#h_JLz3$x~YQVmp@GDUI1EFaZB<L)+O
zm}er>woBz6hr+FD@^m&zowByt#VhfcflHzgHpblm-y+s^U)&WG*iOHi$xQFNa7!{I
z>Pn<8Ok2^F=bNfg{qz4`VzfhA>s5Z?(3ODh#2tn~EOU%L)BTjf-uUSGCdoCLT&zAO
zb?)Yu(`SKnnHX&(OnaOe_tBlZV<qWA{OzQ6k8nDJK#f2Au<HPQY0%x|1DzKDg-zWQ
z266IZ%aQh05e^N-yLxR!90(b2eqGCHM;q&|COe#CAN<y-EeT;4V%ReTny|s5E+zl8
zr&C+baU*qpvZ?+@9y+`dM+u@1pJhC-UAjzdsVPmmOvpwb%ekyIp!2N(;Tq)?PV9U9
zNqV6?#NtKc{n|K<D@&E0uV3lvX!)7TGJ^?{$Wv_}YJRgEzlEO=%UdQ<GjbjD-<=ST
zbP>zvP|7ydan<54w!Q}ziCm1QSe|L2wt;(cht9USe9zA-hQO<AM^Q$cUnT}E={X>E
z`Z4g+_MchCFkn*^qGDud{EymlLv^=JOia_FDfmh7J(0!93gmT>-hnXUSn~RHHf!El
zv>lb`);H{K^2FYOIYkMw)`wJHI)N!bCMHLFz7BazFU^OGe#k;?BX`0NxB5*EC?=?c
zwtu=uGThY9V8F+@P?b<KU*2_zVArQ0c0T#kmQGp(Sl)Vh%Kb!s#EA{h(8=A<BHYni
z+{DO#a^fxL5b8^SBCqd-Hy|Y)YN*%vF-y_CDic|dW5(+gsi#5#q9Pq=5Jn_S(XI<7
z(VQ)vh+zZcUxjKY1I77;w!gejJKU#zroEWcQ0o`?qlOB9Gve6qtWjnAI(YaLU9<k^
z?o#!+K@F$DM~0>tIM7pJ6Fz{z2bBg7z)RLwd&n6IXB$%#7VvS*>aJvldpx)3N+Tfz
zi$mn*I>Z<tjLkk!V^aVL&7M-;w^#qG5uonw?Ibo&ZIUvZ<$U0VYV*RV{aqkKi8`2*
z@p3?&iF>lJI1{nAvc7WQqdz!*xmjI(t4$ZA3~#OpqU1E-%EcyoxU9gCtH^_`gB1A4
z%+|{CzA%7!Y%@fK+U1`$&8YmN+X73k;j*6bShjs7vE*SU66fT~Mm+oKV%HL6e&X%&
z@N1@O<wTO6*wgzBR>w;8{RJBqKJtZ?dD){VBSB1U8z^#C69Ro7!zaRV#y%msQGcGW
z5gIPHcz%2X<k!YQo%&WprEVjsR8`F&+sv~X=`l-{o#YYd>?^=mU3V8-HR{iUxT|4t
zm7ydp4LQe?hUnlH>GoK-i>(DDNY#eX0%X{*ikXpHM&q4-ZFRd8I_n#N=wCykf~kL4
zW-u~_`$lolYny>d!SAxiJjf@XMx8__%6khiY1b?vdIcj^rWlT=ToILbC0K?<nsH;L
z!(Ey&_d8;#J5v)!l+&)dAr<Y9EfX53Q%OWZ14-CUhbQkETEA9I;ulv^q+Cn+II+lv
zx*y#}LKz)?TOoFfJe^=avr<N}ejgKmK3$s~uO}s{r4Xisb5vCr`zswB7Jw_x69uW~
zs*Hk0T7jM~hTpz8d3jmkgfEY&{WM5!ep{x!%|W5*@i&~F29av;@gZWp%fY0&{9ujc
z)kq3=KM|>=)@*G2$!d4<+}_M&?!wN_j*D}LZCTj+bShwZN=!7A5<+$F#TskOcFp8j
zd%j`#Q>#KZ)W40`a#L748;nJY7wur3W&-qU+_%2DQ2tJ_H&evpY`3`_tOH~bc_(99
zXIdz{Wj9MhOU`#7q2x?loUb>}>;<K;%H{NFY)um8?t|u!At0G2Q+6>ED!*Iqm2DcE
z7+&ENyCk7A<SYu|DnM~@i$Fzu`fSBXtuJdz^+iLJ?%51xBT7_s?%B%Mq7H!XrYBwp
zj9lgcAayT?A8cl5&iK33-h(+Ch>1CsR8N(@tgV&h%$Hq)O?DZ=yf5j`kGjh(+`$W@
zNyNrb4A*Zd&(EiRI}S1)iOEG#Vndc1i9RQwVEfOq6_f)8s;D3D;H4Js3a<6<YkmyC
z8~^_e37{Q)`Szl4eJehmJrhZ#(2MA5UQf!_6H0|M=HrX;wtVGwmj#`yuSw_j;QEl)
z7Tgxxg!_kqhfJAexbC?8o&DgxU!Oi6_`4#Fq=1bW4Oo88&o7NlzE(=s9b}qB%bKc|
zIC0)ZIa=4;bCj<C%rU{o@X=hLpL(rW7IKsb)*X3w{$$@z)mcbNkCCOn#B;Fs6|04R
zC&Bh$(vof@*Pr1WmMn0^3?;cU$Fr}|6A}*(_U98x1jbvTT>Rd?u!Xn1=`TTh#%3%$
zyGjchkHM2k>=@eSxZeD(<QbS|kz3|N%MR3`x4g`#UQF8Vo=gc`HW4I7FW2)&{2OR2
z!!`%6FVjDyrnDt*CnqK~hfzGsL|3MZ6e<yUU{NkPZ^X-D$tH-wGqIC$D=$xDMb;#4
zKbUQ3V(DeteYN#aHZ1+|^0ur3uhNlGDG^{e8lsL{u9}fs(z&=45dpCLwz^l?`>6DU
z^*S`TcRALi8p}aD!VTuR|7SQ;jZx?2aKiqdqRGM5wWCqK5(lbrN6an0;IMa!k@nmX
za#wV)*4OX~$ANE%%99X6Os>QyaM*wpPv*-)gTHsubd9U&ppKgCj_<(wf_8?tXx!=d
z<)z(NJ@SZQ5fD1y=bjb)f!IlyM4j!NWPL$Mb0|2$r8+XYG+@!RApyu=?`Zpx_hi7e
zdpWxF7Wc>I3?;Ggf)Evv$kqmjCpQ2z|2F_RfYhH<f2&w38jw{ZuXt=OhFM25e8rU+
zLz$Uurg|SNvDFcG%0-jVP3})Qu3>UyvFFX2?sEl5lkF@q-3pXW1?DU&ELf?sWv);(
zEy&I)FT{R%8O!rqLlpc;dS+TYlu@~SA?D;Cevm0=pOq;NeEwZ*O49Xjbn4>Al+^96
z9+}*34Wk#ozkXtM<Un4Pn#Xl-N65<yyPBTd=7K$aM_9!TdxZS@*~S)-5>>EA2g+gJ
zTAsWy6n=uMz%Raj83vH(HXXdCjDht)(s_rNFuS(Yl>N26wKZf*U0E}iukYicL_O7p
zg>PQQCMG8AMv6yHREnXpaFo&L$D5^Kr7m3j<LaOr)dpj3K*GmHpDcEvNl8wz8l)b?
zp<(ey<O))$S-Q|)Nk`!vQI!yKNXUAJWf>-Iy&R>J%O$7n*L6^<1wh4xo>}qN>T}9z
zVD*;1;PKC;fxz9ck^5AS&uo_mXFlhCn~4O39Py!zZ5Q2oN6Np19^n}EVP!@6<XFf;
zmm$Gvksu3$JxHXZFAt|U^%upM5d2HNJE`-?%J=1bcj@$l*;nBVd>rLf8**Nwew-FD
z5)#F7D)T&5R7QbW@f;!?qOiQ!P`gbMdwF-6Vb|W~f!g2LDDrA+OEHR^j(q$?s0Th@
zDmM4_!pS>++}Y!K*jKP?*2P*k6)T_S=bQ6(gQU*XnSZiGK<a{Lq?dhV7g~!aH<z!B
zaeU_(_(JwAPrX+3!_CY6jgtz3cEe98So|1=xCSh0;SOqx@L)gujS(Umg9Z>C?hcfY
ziuc!DgHXG~_N4f0>N5l#T@JMAG}$(xXsV&CrX>?z?B-n;WiKZI2B9&E3(~ePgR$0D
zeufYsc1`PtoTR~&5<bI~^TfBb6K}YM*G#k9m=3wZ!0iCnKRz*@RF^S4SwRl{Gp<_1
zS^~;E8&vz>u}?Xc(wl$(3I~9#TqDKEt1d7@DdG0RE*a+-PYM8`4ZZ)?zmDbX=o^$s
zy8S=VZB7QYzq6rKSrXKztw5gcDFyOF7gb_|CdTra(F}=?8htfQ&GBh(!6$#KnXQAf
z)rZ2K1x&uU6xZ(R<3>?6z~*9#&-$o-`ier^+P2h|=)I=vXxU}Htu8I##hm4aDXd0B
ztV1&*MBY(bGI{WZ1ILHDE`G*}MxJg=u)I;dy_C4Wx4}5vMD5WIm)Z}vP4?;SJ8D1z
z)u7ELrcOV}+QV8E&P;mvf@5oWt|!9ei9-jbTq1%@Y{8a}IWn{xLe0$<Yz0Ivpb+Iz
z7-3{#bHb)0axkIfmXS8KtL4=9pvgDBUT^-_gTexoB*x+*ACcfW4lGisi|?}}K`**v
zlzITCNLJ0>q65l+RKdTxJHN6V5nqn?R~2^B59sr@8OZQY)sqtH$%u57UvL|CUc)^E
z%;-JYAtCOZPY<6o>MlK+>qQM~e*;=!AdzY(E<JJ@IXsLTi1v4>me1;c@OhFvcuq$7
z<?W|;E<gg0aHq)1=Y5o3UZ8(uvOsP(LGN%aU{P^dpX7f)Wg;!NC)e}K;gT?CxjlV*
zBO_Zf)Zd3)PwqZ@4|-I~-l3+nVcgEr?8Utj;oN8E(qm#mxMY>^yPYWIm=aO1;>#JA
zs6F7*DX4z|iT`q0i}iJM<I!LA<l!&iQ*=UGIe?6$dNkXL!iL54`K>bN8yP9g{gmTh
z_--D7$y@l!&pH;yaQO)GcDog2zkPPj(+~u@W<4{FEe_`vHzrZT(ybuB7ebl4&rO!+
zR=*QN4)<aXmx$r*DAdm8e7_Ev_T(Fbv6eV77(%e!O8(3@B9sT_<M65VyQ%lu<Aq_0
zz6Mtm2oyCozbsx_t2O=5XQf^leI)k2pOpWP_f5D0^xvPm#b5?jqD!VYT=z!s1w~;?
z+tki0$n?>rlJ~Y|P-~EBY16@XY_p19X@E=<*M9q9V5n>8A?kqH?g8|%1Z}Tx*sQOB
zZTx(pEtAZp9|O_X86H8=8!_iU;P7Q9@)7@H<37}=^RdxrvVQdG=L6RRFu6ltB<>xR
zpUuVmnXz1abnj&r&aKc*O^u1Qk%=YAbnUOLg_%Te4dt{s=XV(2uIY3HYjow3CYR5P
z@eRGN>&!-oqYh$|lIls+histT&-idvQqy{7|3^<Cc)f`T9P7i?TvYPTK0BPCz{v*`
zc6Tb(i(fB1a?Qfjsp+(E4Gs(W5YUS@#7inWT4B)l8bih4!9$iu<!zdnJ#8k=yL%cU
z0L~n3nQkD)dox(T)$C@st(Efi&(S|FGj1esd0<i460HxXEoIV<zQGel8%{reV6S|X
zBa`gkr%d6Cf{0&$f^B^?UlC&<Ku^!47|K$4>HFa>jdr*`5_{3}L_&g*Snp!=YBslX
z?$3wi%DxOw><Reujh_?IY{#r)kLRxi+~Dhx@Ds4O2_YwYyb-&!2~5!S{gG0@l2)u}
zY8nNMr9=lW<NeinDs^d7@7I1HX=BLcj=Egq$WH$^8NS&iYQx5UWb4t@#reGv;89G_
zf0OJMq;gr}TwzR&lWA6-?KO<1M^ch^CoAt(&)hl240ZnEUtd$3>(?;%w4T?+xCdi6
ziZf=tv|2d2`}=Lu?}apJ9<{j;(A6Cw0z<v7I>y#+NYx(GYuOxc`>(c(Q#(>Vwtrlu
z^C;K-QJva3*gAPGy;bluLFY6nh}re!*O}%dZp<0Xso_7P?*G~1V}1s2yr^-)1kIaT
zfrOcU#Xvo+K;ylac`F|r+bitJ$_=)((9dU{O5BC|{`LSeH(~S6)>mA5!Q}EAu|(9+
z`Jh`$7{ivvuAGdN{p3Bs9{@l@_P%+w|6*FkC&#C3ZzTe+34Z7LyH@3QywwSzEbh}K
z4)d9C0mTCNc?T=#Z^@^<R`>q|2TOjAlGXY@>wuNK=)*7`K19z-kA|BJ?0rRL;SsE1
zrJ&Q3B@sxq<5A>ThtGv>RR8<BSHu|u4;@%1EX5*z;rc1$Ic?FiFqOxN{bX){;%AC$
zm`3gOm|$o2drk;eNM(-;pLT1?eYSfLNcwuA>S}%#n4R(sA)i%4_1Ax5pVkm}nsCa9
zWAI!(acUy053yM17p-$#6?405TJ1ku9~GNNEd)lQ@mMGe(@1kexjuOP_c2)B)RDq+
zJo_;`M5Ri=#q9CvNK4I*J5A%Tiy+5IBi3rXh=!?s#9pHcKsTmd4(VB4B)(}s*qQGy
zat}Ev>tL?RZJhSsb&0Azw|++(xMqkl&m}VY7@wAn&e0FBdgZF|dsWc)mTnfh?UOPj
zVYEC?Kq8QKmHZ+2Q0pTE6_6RkFEwf*!||zu?`5GUVS_`u3fEu_sooycuuS9b%oiz^
z$aMcyG_sl4oJ^WB-ro!r0B}oel8N!_|H{>lYZk@88*F>}$sHvnRi*DYuVCLdy^@2<
z`YVbm`n$R~_1=EodS0xpEy?SeCM@q(GSNyUh6DF-IHq?s^l*J1wUbLb*xDbV`!PX(
zAUpe8X&5W*XCt(>)KtT|=elMF0Ge=d4QWuJt55hy=p2D|b9xl~7zp^R@vC!ER#q7X
zD#-hlV%!C~g}+-;6397FezCCus4>sA3@5PZu(iYWTPHm=wQmbtRP=#cb?4zef%&BP
z783r2MmM0JMMk3-bMbBayl$0eLXc4sCyR4z9UETrcwk46cc;c0a*d<2=`u$2zn@kY
zeUW3A6;GRIjl%Gu`YnuKm!Tb_jiQDmSf)^U@B7zP&~HHkmpt=Inw`akEaPYUKyZzM
zl8z@~hQgMeRpJQjiqs{{;~ImH!e)>gZ^C<)3^p@R`z@%$)t(JDtu19A`2fpmOpVu1
zNj4(gu5vu~{%hx<(1~Q}=b3qY6&=dtiVCr<8eZ2jv7Ba+$ZyY51x7IPLRw^R1Vs9=
zkZ(}6&+geH#rlxw;Cj)36a$cssS(+{_#z@p#0T-R)ea*nFKVavxBW79E2Wc%FUm_K
zL7Oy{vw>nQ0qf7zwdMN9{4U476ga)$UqQq+HS?$!-Nm2-!@6#Jj$i8a>jrucx!x1y
z#1Phm<4}4D?N{#HpZ&{DrV{74KMx$D>VTR_miemnwZE{=s_|TtLr!HjNiZeKGOB%2
zIzaY)*sau*!!UR2#P;&5#MS?Gs$PdI)>n|{P(O10Oa}3x?nRE*tSe0Vsrv@B^OT0i
zgZLPCqhwS)m4G@x?(faVw#RXo_y}-knB>}Cf&XWItn&ZSbe@51{$C$XV$>=Us;C)i
z)uvW$shOfwYu4Vi_Y4|Dt9I>CwMwm0VpIH#id94_w%Qe1Bep017ms-Bh2+kCf6w`x
zb6ri}*S{)!(By-D362T-&6~{=%=^@A_T6Jss<)~iZtq!f5;PV}Tf|k&X4N@tn4MBn
zfx!Io4LlMFRb`f!izxg3SXZ!D8d2HWh%rUS#?FNQjAX0X{cuqBuMIW5n!*Z}m~A$G
zyTg)NXwpo`TF;`ZyYLA=|97%}n8)Gq=ZVD)Zt0~&x@);sF@2la>F=iU_So}jEf#@|
z!|J>g4vxai$Rli6JhCi<AXc~k@}-n|ScM0UGi+a@eSMqPG2)7t$A4W)MtBjYmd?km
ziK8l)@1rYT;?KsX;uA4!EK7@Bc)}xw>!a(<*^v>)1#DA>u4Teohf+SO_un73Xb$)T
zx3<R1P*r&n#3KusI-5}JB%0HI8u2VY-Fk>(|E1{7^~ks~gBIf3E#O#%aa<e3LDAqy
zJ6jFV^o%#5wAM2>N(3~D_z-|#<8h*g<fPpG@2T|Jgv2ykN@<pYwyxlxoGE+T+9knK
z!G66OTtmj=S5;QM!rlV!Y#XQi7iAjz<_{=sk^QZOIlox688g~EE57H*5aAf!zFTg#
zf}Xicf(fLmIc1C^Njbcq$|*z4@7hYr>hY&}2aT+4ar9>(E`7R}5dPAaCjs4&VT$-L
zSanl)u+q)t^*?;aD>C^4ZVGV_bb?2T9>N@*pu6HP8T*S5Bt)^o5%Eb6WAeLW_H)R+
z@rDO5$r$=fBS7koP=dxWvvN3`vc~M^G?pQxx1NouuNRt7gmBB3!-v!%@b%)|dfS^;
zk;*R7U4G}z{{~kOE|Klc28fQUy|$Y$Rc;8MAUvjVa1$@g`?&nCvu6p|wN~{FBW$QX
zJonDW<GxXMtH;@I^A9rUt0F!rpDgt48Uc#l$8ff#AzWCfS?7`#=j%NyvR<ePIhxJV
z%R)tY`JA5ezhY(3tDL*oBp!199TKmGCB*6WkGZWZiSr4BG6rKco;A{3^Eh~k&?H(B
zuz*vh<Y*FZg7p4vIlfxdqTBdqrI1P=9TQN=aNzNai>N-Y6r*d}5NC^zwE(~d{=8TY
zM?WN7ZDD#HA9B_lM|e^8^e6ToR2N7*Rfi`HSnhrbAr1Z*qN+eI35+8trXY#R-HWH@
zcA7SC4%kK%qAxVq=i$r-xaOKWl+z|8(!ftq{8j2sf|kgX9*Mi(=;|uGW~Z*caC#_m
zTm`xhf;-ol#fMd8;@td<H3fQ?^(h|{q4o}fccdzeQ&9Y=A!f3NhEn$w5ZlBXi<1K8
zHV~;N5@02h{0HArZCyt|T%K>Em%0;IP&84>4-p3A@ju`-M(%?+H2uesb<47@Fv4}m
zD?LF=9gT}lD}v=tsB*#jrElmBzHj$zp|Z>KE&cog9}yeA5x3`(jdJrjiYB!4eQ&_F
zVJf^Un<&6A!SxEdI$^kq%kv8_L`ni=C>jc(q};foPj#_i%KQ)XVf*QwR+_=ii^WbB
zM54rGvgs}>aB#@Uo0+!n)k_Y0s|(+`=cSk}g|~lmSm-I!Z%nb~cX<u>wA|-By0mDU
zpfr@RX3WkrD<Rif>Z@n=M#tDAx#|G1dlE`oOXFlbAjzkO?oUkID@`1f#ZporF9e0(
z=dch1DfJe}ouK&X#lfpZqST8sQt=f6v4yUi4gC5&{Q4B{yQthzXh1z;Y-MjeytUyb
zPV*29Vhgv0>TXaT3@IP@d=*0FVCX#y1$Sc-qPB9v_LC({64WX2t_FeJ3h#Lj4C$h5
zDL3$ViQ<WLeHe5oO`FNqzzplR#tH^V$IF}v3-tI$9vpGQT`4fK{D~AM2sXsI3^l!5
zR>MpziTblS<_Q06Y(TE#YhB{TdiiL9II#CCp>N@d(i9K`Mm>e?G%UC2zQ>+KATo^P
zOR$nllG>;7wQHPjU+9~U9+aZp8@U9vO_3E(;3yO+jV9-}U~}D-29e#7GbprdI&1bt
zDLbT(0i+7ce))22*F2%8&j9U6tg^d^tF-YUmwR6ObzVePqtDg9vu2ezT)wa3=P*<c
ztd|U+#1|^&u;|U=;{Z#H3KojIajZqb->M=#X(Q$dCqzgh507i8i?>_)?f+O1%5ia?
zBMr}NbKVK<E9^FK5gBF~Mhssn7uo~I12j}qRg;C~ce%-tiC>w*ZyMOIOq96Se&UYz
zf?KNtxT%-FGr3{u{RAJ^k6km8K>r$jZxgOLKJY#m;<W{X1p~|$lysFmDxm*@iE@&J
zgMa@ZfE|6OzP=yE+l29@>oeT{Bu(d9yClw}c9J1QUCH@DG`6BfW1CVAOqETwmc&6g
zc%$J%-n1s{)b4uUU{==Qh+|k#Roc#)w$g^|D=gb#Nbj4F>m3cpFAcU1ZTEO<jABzI
z1lBsR0hLFyiJkyox4WYlrG%nE8UhTE3N;c>y<IB&%q`7-JmPdR;wR!+vcv5$*u$-)
z+-fdc?syS&V90F<%#Qk~pY-x>6sI@{BZ--tsLiBAF;i}=tRxyDsko;uuYMEUGn{F)
z9u6E7%TG7)`ixEL5anGtEO_Ms=Fa%XSRuHuBUI=xWaO=FBLx|9RbRpK-Vysr*rF#k
zaA|)pGGc4;!OD540>WB9>oD+_TbZDKQj~TN>^l+dl`LZ!K+jy=1g-b3T<ya|sL+a8
zvgm=?&)G%f`G5RQbyEfi<j~8f&-<J>a*OnU*1`ZX;V<>Gm-C1JeziC64^kXEym^~G
zbfT!L-}lYWNQ3gPbk>>_hR#4WgpYQ#Zi%C)d!CqQ?p`oZBZ9Lln<KjSFT@R?f)6CY
zlmN$=N^Tol9rYlkBn^`%MYLj2u&ru#r=qF@lVHGtK3E?fu?2`dJybaJiimX4EIJ9l
z{?8uqr)Hc~+Q}xHRwN+ja2rcbF+=Q@;@XI(OKWL{gpHJklEf0&QILqj^QbAaabg79
zouIDdPD&6!h+b#c-pyH2cf{FcAREtG*zNOn^S9+0>@K~c8v?IW?}DY4Mvau5^Z59E
zC;>fE(*=)nLmID6eroMI(E<{sT{FhhufkEz0}RC^a)+Byw8H?R&+kQM8RlNZN%3Nk
zlc-7d_T-?%yd;nD);83@?l|24Ds+jiu_*iQq9MWo9|g6ec^pStoERilp{m{=p`y9F
z)MX`>MJ6!Pr~7^n<8;ydoICM%FESx9TifJ~woe!-07S!1?PaucM62Jt|NOyYJi0wj
zM?`FYagT5%7y9C`)G2s(u&!uHIifZ4+E<TILGrtDKMxkkg*4wK4lLc^n}R6yw1K}>
zX%OnM(=d`sOX5iW%f-%)E)|ak3KQK3#q%}Rs}%ECJ&3vTQu0jrC;kbHr|5_s)5b<2
zu9pIavMrfD?x;bSbU=DWpA9mT_=0|U3+-;KFw+-lcnb2~DRY^SSX~qbV0>uyYojuN
ztiOAVBZK!v(QEdNZJx0|)kx%1yVl&fh+$(r>|jDBE=>FmFOZiGrR_w|XUj->F`FIj
z4cRZe%zQQ?VX5a!PPu7G1%$V6<}+5dxxPgPnCV(kd`G6en+*6==?H=H3^V&EUbbAw
zsopcMSX<;EuK9*}4_q9ED%Xf*O|>!F!cE|QbTNrK`%g%<4O9kvzQ(P@sJ>-Hd~`Gu
zR*P@O_OfI(6l5MHP)iydQ1Gh%*1jh*{#X0p>Wae`)Hdq7dbY6(hJqYnagToSZnv|7
zfV_~=riCd=MIBa_v#WV=do=(hNL!jyKrIUOM>}PU*1AhM^5$r7YSU|9C4YLzDt`Na
z)S9LA(%OnP$&If8PhnL7_r4<%%CuWf7L)BcBR+-j)js&*?Qk?1#)-*D!8CYQP<ndQ
z#Euh6Ia?ADuUPajyRKltLix7blX|<(cY#Q)H`=N1vdS{c4%U5HYnJMkB5(F)kN-hb
zDWxet1H8YM5K8EXc&{?9jae_R95dJs@%!LET)K-}8`;eH&ay3F8HJFutiUJ(6^-~y
zTnigSAjQwjM?e7kg9&1rJD<f&J6m!A+V&{ae|JOq-Vw1U^Z0j2iI#f47}X)Agmbp|
zY+*Wd1OWSrs*ar%GNOWWgWlVI+flo8X%5=mpy|;Ame$rTBzY2Q`(B$-O&SG@#D3d!
zQ(@@k`SxJt;;>nC1R->}om>~TOCX|j)`g+C$B|dZp{+wB?kk54L?c&-dx&+Z<=AG)
zJLNxnk#7@)j6SDKG}Mk56$J;UR$`CfzcNiyP;t3hd@&kFL+JSHtJ#YyR6`-+e7o-|
zbl1bdLH|3qQMof8*Y(j#-$jsElkR}zk3NBDs6>>0meg()!fm(B$LvXylILAZ02yj1
z?>FjMV3dD7K%j5}$Gnpoh4E9eKzby$zE&`2h=CHt5&NA4&i0)fToI|9x!)ArG_#4(
zw){gPbVWH1q{F53Z8(AU6-vpLlNPx#A+qI~PL}KsxP|K1h@R+D3N>rKeLC@qr|o<?
z8}L!X|KVH2!8Bzq9}}GFmWf|QuK<m@V7_Dm`}pfk7!5UPu9H`JZk5{hw#?lJlz(dY
zKRi~2q#emX66+_R90>{VGAiRxgLeQaW)eJI=i>R5LN$;k(djA@z4r9mV&|*<m&AMf
z35L19;~RrMb0ts{Qr`n_e3JW0@mzJN=|DJ$>xTmZlsnkpM54}VFZp?NMQ8di>REQk
zVK%=?%CGStt)cA~*>qjV%Il`$FPwv=Zys%sQXQsqGi*q-&I&`ejrrv6IT$y<QW}ZO
zLhV<Dea18Rpqr~H9O7!TA}ecsE#6qrW+mG+L<h1bGhn~a+&B&WFRr<EfnB)S!;z2g
zq;d}NoM&}4;NJ$H*ZhKyuTfFnnS&_uGD7WS>FP3Ds(IQxt1M$x;Z7&MM1z@!%O1=m
z7<F_W`U_j;z1SV;)AY-HqL_H)gomj02UvNQSjcJdVOkio0s{Q(%+U;hV>}v7Bv+y@
zJm~WFC!$hVD1p&AI#KCH2^_8lR_m9a?GcpK|IL=UA=~ujKt^#D{4|HKg8ka;C`zR1
zo1w2|qhe>JYg}LnwckPfOX<&mQ3kO8(%btny=h3Wvan4H!yndUTLphxNvDU&NR6vW
zO#%%v2F0>Pt4#fSoaQbA2ofs3_WZl9*n^eX-L<`wV9R}I?>}0u1iQ+VvFw;(7y<;q
zb3=j4232`^w*{L-o3M^8+`RYYp**P$J-WRH%MYdAn8K!7*w2r#CmR}lmYd{)w)%JZ
zpN*KgSj;AKF4noTUx|-XsHe46Z;rS=5_vWe^<y&KVwY~>dGXpuvp9yWRIK1T=j#6O
z7TMRHsk{R3C7`_lE{}eAbR~6L4I;MeY4YZp5l!aT$q30&=>>`OqB$TxfG>u|?-o0)
zA||f2&B7^-H{pI;S?DJc<%l-<d+G_yIg~$1x|Plwgob_;u*H^Uy6KTqgwjJtFMcM5
zXz{_INTHT6atSrTma=CVofjzntVcjfmT3R2s}p7w4|7V(g4Hu2Sn)<52$2FARtzI}
z|M&aIdA6BK-B)g_UgBfGBiha-+}<JwdM)@UGSTOVEc{m~eK+FQ%druo5dF640w)VW
zsmsbNt?!k~x{ZkwL!A$R4#)$`ALUhA^Z@!|n>c4q1!3XI=%UxaqzZX)2oI(2^JQlO
z4UjULPZe09ew&?p-ojlk715hdnn)2P9QzYxNvg#C*Y&?qQcO?lLm(X9FF>Ow7|W8-
zc79phM(p~obra)#I!=A{GyK9pEIPoxinS*-+LV8dnF1=+Yf=ID#Hbl(u3P=!3wHx!
zNCUM}9IMn_zSP{wXf>G2qtVuBYt1@RQ#V255U$n7(tTR|QP6*Eh*l}*RbsbzB*f2S
zkI!f)u#MtvUa~}KhIv+i#Yq7)?eR{-s|w?qw`7=q8K^|3i;@VUlZ*H}pXA)syZK4p
zN7<Dun&|bMR5HI8VCH6)O$JCYqN=fAv3=sV^M{{75Cq}jV-WIdKBiE6Gm#IuagVyC
z4FgD-DWVUjSWpNCJyLoO;|JQXGAtKsg#U{v!?8hB$>*f#$5r)JE<F-|m^ZY?r{`fu
z8t_wF^^c5#iLmA8Q^+0S?qKzCgQ&#$Pr{>N*LKFrO2wpU^0_G?uKL_xS6v1AWmJ@I
zdiVM~GR!xHRQjM0U`{KeZ9gE7Hv*cO7QLL=cd@v}+=`IJC11BJNdbzM*t31#Cz_7@
zBkp+<J0hu(L>sOJdDiz4*Y=LEU;4XyKAfypJRh}av0L|OIh0yip3<OF{oVme`#5X%
z7Ut^L{RJODp~zF_!K*PS%Lg~271ueGt%lxC*;e!B>cLsO@=DS+X@aXOSUXCa9=Hs)
z{}f&MA3~fR&kF1pATxb70|HSx!DOLQzIN0=h*|Vvpp}6TzqoM6@|*Si*jZDy8QZoD
zdl+2VTFy~DO)dBFbFMM-)@P~f|IW^%IF#@8j~BD<Frz;=uQ@X6NNU{$IT1PIn!9mE
zQ9nt}JqNZ)zQ)m0rc$eZbXDYN<~;v-u^0I#46(%N(ao8R&O2745UU_=)wq8<H2*BF
zE^SEn`XE0Yo9G>%-FN@Rc=5PheAtmTXeco@B%AI~hA~GmokGaAjbOV2RuxR_hn8jd
z>op$i`>T*1p+de^sEofvNnr+1(Z9EAcFjEmqkFaG9@YvAL82M>=>SnvPfq7^c*iOj
z5gLxI)lF{zfcvWCP&THElACK>))J!CQvW#Iym_km*$-Ai1Y=~NL$<4(lmG+J@>oI>
ztlXWS<+uWMWQIN=&k755;H63!YM8!1&(b2=P^;&(b@~NAl>E-~4eE2-cC)8@O@{bG
ztY4Ls`CDbkfy#}2M}Biqma(n8*XE=qVpamfmw-s?cdvnlE7a^Sb|WGpw1^UALN_{a
z+I-e=hJ&z6{N`OJ8+2d&ZBeo0R-Bre2lzEJq}~U&Us#^N)QMFdQ|&wkV7uT@n)L?v
zoL%wo%!-H4X>7j?h2kQ0$bG;W^Xq~f5bPNYOc97Ef4h?NhmRf#@|WuU4o5WFL-5h1
z!+i%X*Xx0aYaa}KD-)X-Q@^V_eml~hGV6c(J=N$9EEvRi#?eMaVo)xewrMP4i1WBO
zIzgSd%%X{u$<HlO16LlxjuG-Ya_tt`n)7-hWtmfFuq-LwqZNE{6@lnp)fMDaiR6Xt
zTE(j=cdkSkW}1)V&{uB15Hlf5$!kBO#fvEKUFzQkMFj`_x*oha);pIw!QwLL_}V`I
z+N`r4*uM%ywD9v-H#gLNrLa$ARtC^LQH%F~NbNxFrg<T8UJoCdOs4=iSg=U>)2j!c
z$Y8>;Po(mJr^6AaddkdkD=Bzoo=!hj@r8g=r*qQxF^{EZz=bGFM3#vz3|3k>$6UU!
zAjN<JQPAJs^6qf=jH{Sxe;hob*tt`m{kd#y3w?35E31b_MJ2*N{^s=|cz<|PW@v>a
za3Xrm*OgDeslQh`35V+2TReEL*>~gXmQY0&u1O^-kz;?Eaa`m)XPB2R@8hH8q$r}E
zg=<pS0$s9wHMxlJOvdeXshph1sYcj_D%EgWV-LT-<+WFz!~CWPQt#Ba-VO-D(DWp-
zZItY8Eff_qQE}0+ZAX}L&Y4KarM`LHUeO*3qJ)rqp~&pX{ouMae)Iwa1sQ})Sk{h7
z(A)9>oq9XA8DP0*hGIql*DfIm>MuBEj`5}6rO4%J-KPjBmopanzr^Ha?8IpKQ&I{x
zNgj^m6C&>KrRd1W?g{?Ny^>6|@6Ms?a>7BNwCv|5(K45P^zIDmrS(hUdEo(OBO#me
z$2DK4_r7*YHloDG`)VY3xIdjSU&TW*Rpi>ZK~w+%=yuY0rvSIzs&MOl$^GODIgUML
zC7QmtV+2-27<_$tbRjGretO{Z$iv8O|NP`6GY2no@65KHZhn7|I1qX@!fA~wAsHe(
z`x*W&7q6WisAI#F=@^(orDOXo3I-C@ZAuA!szDuJ$^7`^nc~1^-a2mwrsa<!SC8K<
zYDA3abw?>68knG6$nhj4Co_i4zNMYae_Z03F(O$;+)rfP*L=KUs@c;t^GLd9D5;n1
zaD2qi_X|Usq~#!7T7s<jPN^>J-c7RgOMQ!oBSQ~@n6)wP%yASc<dJ&i4_SGS%=w^g
zwqu9yff~iU3dYakzHkC`mK+6gimOcCh_^q2@HfcgNAc|(3o}X^PZ<lfDs)WwFnb-H
z2uosm(7MENVbA$Q#A#T#4(;W%<}x^`U@Y%(uZ9;D`@sle8Y(RXf9q{BtNH{lIyedS
zt$gUu%RuGF8W($8AT8C?scIvW5)MDqIwqO@n;bDgj$9|G%0Q{IVC0aDbH-0jRojN_
z<&9Ah3I(edj-{Pw`z20^{oC8A(rQ(RQXQq>sE$`z0cI#ysUikSgBBGI%it{lY7rV9
z=&;ekh3T<3FN2X<zZ^4@@thQS+VT$ng^Pl(Ny|%5KH&pDCA6Z7=m?7yDSicelQXH7
z%sjsRJMk*GYSSkXXWvc4u;6MqZ(|mTglUgonQas=G{1!_lwth+_23x?!vzQswKH$z
zk6s3@oD>E?YBxI5>4hha(ww&4Z5u&QG)s=5X^I}S_*ZvViJ#XEe0p?$o6xq^xbats
zfpeJwnJYg%$6S`7J(Rd&n`$hhaC4hfUV_>n-2$Q-?Q8x$#H!hx!cLB&*KbgFIko1U
z7;zPSsfbt{EsU!qySn9d@F5ZfU~+^iC#h+Fu9Lu)i><!G(0lfke&mXEO#x;bqj!3;
zzB`p;Q9`N|H>TIix3px9D7;yb&d$FU+-v-xi44Mh9quaTUIYOci>t<HC^Y2&msO{=
z-|ySnfw@@??aWac%!5J-BNBc3_xAn{?jz<GIqb73_zaZ`pmL2*5QUX15@gjzSaUN9
zmKNU?Bmh-$FEdsI!Uk9kv%6IYtj<@0XaB;UDaeWyU(Nz<6+QapeBs?}T1n5O$?c$N
z>j2%e8Ws8A{--aORjtr1LVh;6F6*%Qy*c~+>4&}kj4LKB;#5+SB%_)-`Zoen9&fow
zqn<+Vzv@L?1QQz}h;>}&&EeD@F)I4|$%zvRNS|NzLR@&u-vtZCG)+Ux2wH#-;wB;U
zcr@-dm);6|Vlp~}4?^)kDol<mbh!kFb`NXksHwa-tXpOAAOmQ}(8mOciDey9O3qEr
z`6L!cBjGGQ-TO67*xahR0E<|`7X$QfdJQRTdgq_S?@S`T-b>`VJt={#%#=KSMM_(2
ztHGV4M~bX;ebEU29gtUp>ZUQy{~$8etNG%IHeazLS=w*d1}2B%K>67dsq}2O8XZMh
zMfmTLUbqw1WGa6A7pH`;POadb5px5vUDx!pHNX&jJjEPhd&JMAsbx?UpDs=^BOU9(
zz$bZcq`?aDrsqvH9>3rr`klL;GjM)Tt=DFF3tp8;<_g|0H-27CDK9)%k1ZD(_HSwX
zTyg6wmh&PKF`sk=T-huu#QyW0!%E)$u)>H}cw}9I2B;^|j10-cwfLOkkU2KBWFSk1
zH`*d)keEKp)!Z}fSs3I&21rm_y;qo$!o^r1)em7Qd$^aAsw!osCEdM-SlmqK;?5Be
zfEvwMya3-xz_{?tZ9M<~Q31CpLf+KdhsLsrsD2vEP&TMVyfCZ5q%x63m7Y{0mBDzV
zI+s{x?atzeSW7he7MHdy_@pAhrFu71!Q*l8MnTWQ*yd#W&Ov^Qj6-Idu%4AaBDkE|
zxV-gqJGK`*H%HRLkG-zTog`V_UBCR9d?n}UHCw{2<=#oKj=Wy?Z6%EwA963HVwis1
zAK<dj(I(ndtGDpnE`)6qO8=>8g(*iulXDFKC+&r4hhCiT5sxj_=-s{uEVyEwE5<I<
z9%|2;L1~noX<kZ2?R&?}K|1Jeve`tC!Hjp&9dPKm4q($OZ~%nVrr1me&a)zf-QSTa
zoo_AhJ&^l%eEjeFYe#K(E=Okmy)%04XaNk)Xw2P_l=rteMQ?nbpgI)XLhF%Ke=w}s
zpe>@_--TWCm279PA%@m>-{!xPWIw?NghiMwbv`;|Yephh!5b-PoDbqCXkW~Tncl-C
z%{f>wmIBHK7?|)>4)0Vsx<)Wdow)C4r%Bx9m2V}B-S(ZM8oBcM6Y<qk!z9z_O~&JS
zKl4`j=&#-y&i2_@2xSZ^JKM$BZuT^=eKC|q3u+6E`Z1$TO27H}<*pXn;N-)!IdOI+
z;A0`F)j#4_juN303$MN|dL}RMyd7%K%5qc!N6{VeKKvO!-sWC25(q*Bv~YS2PNuk6
zpv$lUj|dVc*ol04H>(z5%bhk4v3-@U$jh_!J_B5v_^K4g7B-+-FQPU7sv;TljGFyI
z`JCbyUQh8*yf0+&dvjkem$17?)5lwJ1dP_bPfr_A?fV|!YIrY-(%i~joQ}XnqJwv{
z=S1ToRfrM*FSHgh73<tgKa~3_7E`x*yrkUKrBb&UOF9L&Dv<DnzUz|&a`l?fNA)N!
zWMrpO2<YB{B-jT#Xi^xTDfjb{^8SE5`ue$Vv1&r7gUKnGRO7Pt#o2X)CLpSHbd<{`
ze^dNp>$@wCT%P=?IcIT$QO=qyD8&a3R@>B@5jR>X0CQ}x2%?2bG=|K^v2@ekBXmir
z2#EXD2X73X$7P8To#!_i9<M6Mz(7HE&skZ3ToLhY_`2pT8`Y6N-sMrX!}5MhxWR)M
zV`n<b6NR|g+Z=?zKgn1N!H+VbUa5kLB|i~nNgVSRUzQeYnIwQIyA2r$)hs-lf?>IJ
zz3r@g!nkt;{^}7gpnpibF~DrG^JdtY^|=WtxHzL1{#aMx`g(l)=5*ZpX6@$e;wrG^
zR!VR=BI~XSt|sR+JUh$|!f;o#$=jBWEwjdb_wUbR7dO`HO?sy}<;{2ZfPlxJC|;Dy
zj}AGjP+cF(5H9ci*SeWA_+ZEeX&zi-^p>lKdG6c=Vb%W2;gjTOqsnt|>MuB4b`bOU
zQ~q<?>R#A5w!o3bq|wJ*h=EG&bm#!atYKUa02!8Q^qP;HeNDPp!_9hmB+jlKuMH9s
zC0{eR{tR4KXOi)lXNvr~k}Cu0ZF)y+!e_RL-k-Xl7+OiloWHZkarL)eLXvIMJ}rd=
zvf)OJ_()YSYmVfx`-In0Ip5kLj|Nc>w-s_5wkP9rwri^rbYkvIDNO(*efO{TimmZ$
z*m=<BQ}YLqDe+2faG%HtyBWPJ5R`Uit_crFoqC@7ljjC1FQ*!Z??mIex)+jApUN0@
z2ZVE?odg#Sn1QBlR<kC_E2)!VD&~1{4Tp$rUT9WUkk&mQGdOFcOxWEFn>9;NxxsZu
z5Vt|6){SBXr6|CXcDY1rD(`GrVe@h5Rbb~LF0%Bos(DOk6<Iooz>s)GlH8}BUiARF
zDU0sp5|IwC^@Zo-I3s3C27AyuFd&;b_1U3M;_hPCOU_xXurM|+LT6=OyB0mt5J9z0
zoZoJF#VNL1`9ZaGlbO{MkZPL~M~^wwi-Tpe_Ko(P%5&wgQ+|pic@w<a(d{g`V#JF*
zp7~a})ZT2Hf<eU&VZf@$E<$D+;MCw!uBYhm5pV8c0CK;SHR^1TRF#ez@HVlSDS;GU
zF4YHPl=4)x4+f^Sh<m+f0=}_9z=u##t>1G~@fd)OI=tWCWdToU=3KM$fxgU)(FVJ^
zb+S`LkzUw)?NKCGa_7Gq1R(j*Qc^TOo?V^GBw@{WBNAl_AJ^XrF+-P$s`!uIFqfNZ
zso)8Nd%9)Xf@KsSNNW4>Vv0Nf6aE}!qE&Nz<a>=qT;x@93`Tao#33ffvSy2vuP^tA
z&7a^;0bp~cU(HRK=DIUZXf)32=JE<{t#T{_A*JlS9IxkDHg=JxHl{Ju6Kvv=dY_B5
z+S@vj8;Jb5R_t+3xomwUd)X+Otews_EFJK=a639mYWvQOQ7Rum(O5FwBOZYHI(_c@
zE4ydGf|JmRuHCW4Bc#R;!v%w7%nC0(9IAU84#E+Xv+`hRbn!yKgit`C_dEpva#2rl
zq$<$>rxU^ugS;(fVx2%7lZ-<4u(<T$lrz7VHB^*vNrbdi@Hn*-jF`VAUx7F8h(34<
zE4{^Q_^<Na<>gr<+YULHo^}ObAPj?$7K`<6PW2);{NfQ##EA?|S)rEy%y(CbwouIZ
zpL0zJ5$To3N?uGG*50JDm}DU9!HW8=CTtk9VU>vasJ3d8X~a}%#{5{0D<{Ot!3tUW
z+d8P|RyNln?z#n?n!3?!VoFU7jmkSTd=+O&{Z@E42Hn0RHz(nr9kScULoEMquGa=-
z$N`_W*Ac!io2MEIi66j-fy&L%`MD9&`6_<pZ21=4iE%|fh5^hgPA4_X2Heisu*uZd
zN7^|3x)=Q5+c9kSU(362#jSHJ^PKtM2v%%U2*yI6DYcmL(p|*pncsfk!bKK(5UtgY
zziK0_93mM_mY<z9WCm^S^@W6)>3)L~vHBV_2;m=6G~C&~CB1U#OAZ@3N#omZNPhRz
z-L)+>0);zmK~quS&qHTlX+MC2kK)0(6+R8|n6CB20_Z_0#r@OjR%QN^@mt>rk1pr2
z8;w85Xf~vNyBVv(nQieLIohy<9F{^-0aY^BJ7%O0v@pEn-j`xlbIz$D&<|i9qjt`C
z#Ev$sHoy!^1z53lD{s@{{T<1sKn`j^(F32r1Dt-tg6+mU+;<~_OU*t#CMpXudjH)_
zo_H$x-Z#r!?8^BiADO^)_`_s@brb+Pzq#RRA}x-s{9eUN#CS)FIyH-h)~{8AkBk`k
z=qS3x8EDQWWE(pP=1W3T{$2UwD!iqL8wpZ%+~*v~)LE!+GLc5_uD|WCoRtwxK~WLo
z_WECOZ+MsG*sPv=%}LknSQ?K@NHvLgr)YBR@$s>*!j<F`vV0EQ`E?*JQ$&%LfIJib
z@3!TC^&Rg%o#(GBY!anj-yhp=4MZ&9E+VY3a|%alrS}YjVVdE-BE5ZDp0Df>tfo8Q
zn#v?DI}IeWc<){MVRn)}F0K?xz10w**^%`uYh|%y+}dALY}E#>7P~kf6;zQ?nbxK6
z!?*qz9r2Fj51(;^X?lXLfkFLT-h0Q-eq^M(B46j!S={$^a&Dgdo=3%C{3xqP*$Qz%
zG}z9HB|Dy-Zl6q57}SZ)=DXw~epo`l%L2J9JTwy*&s8EqLs|=KiwvTio;;l95$=f*
zKl?X9MEZN}x)H(ngWjuA{;yRhPB*gPiK86>8v1FIP42VhZTliV3+Tmm9?7Vu!8)<@
z)8{W=r_i5$`P$trg1891!CedpSEOx7KUMb0`Ik}PU1K#H5cOp<!N8ddk(t4(L0>HX
zI!H6^pO(BmiJeKr*XvK_9;h1c7kk*vi9~qS<m7vF<O>_#qXQz&!6SN$zSG3+TD}Z%
zc35PwPhm*#Afh>Sb1JfSLqu8Tn*+pj%JY#r4*n?0pl5V+52@)gq<wU=nKO9o5(S_1
z{Y-z<B7d~e2DH(qr6m<d&?Z;8aBx(4=l;vc22ngk8E0lVSW04=j<H{x7IOosIoY*U
zRaYy2&+eYBhboUUfU8axn@l|LJ1ZP(t3{;$MU!yHF%|lzacOQmEy=EIH(*F-Ff;uS
zqPS4QEFMF4yLV7&7}Fj)tYehM`)UlP!z(OoCl_(GRm*Yj>S-z`g8Ht{7A%hB&4iYu
z<i}PzlM7XxB<%zAEmfynJ%uM{itp1vA9K(vY2ga8w0<O@i2aYqFfViTVjc;FUTu?<
zF(jwLY&0%L6$a>0GDmlws4>a^Q4o(IZtxQui%Uuq>=*#b_j`x}u&lPr6Op~c$HkF+
z3sNAqnReLq!kZrvB^;jnrVj}tp9~edKP7QIV5v)VfP1*SNsXs}ZnY3lcMnX1PR$b|
ztnYVUT@jWdyO%hsG~(aVE=vsg@MNzHIe-A_c0|Mu7`E766V@p<viGIQ6HraNVY4zw
zJ(vn(qh018DIC;Nr#zpKV*o9?%yI!GBq+aRw?%`g)DgTZQ_n=RrnaRwGdwqHT>Z4b
zAm4|)nEZ0$D|=$ze1C244SmUwp$`NWJXQ&pcdo3-%GYs$Fe#LtGc5|AY((jfr+NT|
z%}Cesx3hLTgj<`}*RHU6-3Z}B=}|vrgY|Hsy4gclA!D7X3}gZe^=5+ZK#XT+kl*Oz
zd5j1m>WA;rQg_X+{p5h>rv=TI+ze2gv29Df3G)U(Y&#l{!`?#0W()qL$!uyBor?pB
zQ2a$bb^<b{4XcJWigDtW7SUpaW`AMTRnt3s%R7dg58m~BJ#bYob=9PF$dn$Jk^suJ
zNZ2#!>qKJsPRHxcqQVGaO*qd^GO!#>Vh2GV{f!YM3S?Zgun3cW8(?PP1+?7&f9oaY
z!i9VQRREYno5dEQK8xKIrMW*6B3co`ks#sL)?^-5SL4yNgilUR#&vmxc`z9m-v2<t
z1xyMcGfm5v7vq=_nfKMrHo6tHSn@hA=q+|HG*`?|KS#H>GKB(2+8|()>{343iHMaA
z@?(FU12am)7#I4ECFE4f+}lEaVW^CsTb3>!l+&UBH<YT92XG28l~3AM*jL9a(?>2>
zK#5Ck%!-2_ZSii{wn!CFy7t&`UZC+CpVCBgrR4;*NH?!IU7CbnJoIeaQA2=#f33OR
zxiA3uX?PvJ89C${38q0PwV0(25T8g4<&q|9N}9ao{!4|b&ddgMFJxkAqGqqYkyg02
z6^i}Qe50D->seE`>&XPp{s+^L>LmhtzwEz{9SH|P9)4Qra0NgeGe0JUNfA!8xapy*
zf!$dlLVnyZ%^{3qrZ5zi0ih1^Ggd<Vp;vX-NFJopW{TqNTSX#|NNL9q6=%M!n5p@b
zz?&At*~KinZXegs{uuCIy5}0t--!8i;1+YuD)2Ko@9@al$;p@btg;ghV1Fr))fU(n
zp!DOIHp`Z#PfEF@@EeI%T)9e9%i5d(6kZt;AAQLsAa7l_IpxZ<V!r#52z}yBb@*Wa
z+gew-$O#<dJ@rXBa?A!XhWsWiY5_!Tj_4SvLy0~!{by7>oKLi{sPg)NTQErmHBEkG
zAqqvcRw0kbH%M+;#N)!cJ0pa(!<9!j1+Deht%-iJqL;Nq$GK6Vk&sS9DGlYFXEihU
z;+(fZvLx`!7NP-fe^G?S#D3vquBtsNpY%X+*g)*?x&<ibi^#Tce@<AuO=Ek?w9qNj
z)D<7!$^&8$=Z26OeIPYZI_N@oN3<haXDwoBF+EByve`7%znT@Z)ktIIhy1>&szG!d
za}DIFk*-UWTBlW)K8-&uUR%VL#j(=I*~8;*52ZwLNim(2Zdjqq7Mh!D-{vKtn&XgV
z@VdG>$6RxF1YQ#*&+p^_`d0tE^KRUBsC=ho-Jx{A|8Bp57MT}UU9T>>cK5!7htFd@
zJlb&D-(fVQZ~xw>2y*-==v-uAo$0})dY>O|me9wBneWUA^1H=r%QPna0}K|2{Gj|c
zD5kDuh_;=gO@>Nm`3qH+`t17$Jv0vTv!OPmyUoRgAZG>Rc(S9JqxuX*0F<|E`r$lL
zfG93d+(PTGo_Dc*KHthVpVo2IJYb+=AfiHZ*`{L*wN~g)J4G%c3Pb02Z^=cuNTa%B
z1Icr|=dZ+Rv^3)A%~GPvb%?KCjJRh1*vE~kf+T~;%ih0D%CwBCFX}AX7cwLE`+mGy
zqm4xXDf;vEl2kVqbG2Lmy1Hri?+=@W)&wkJJ;JVz6tc{IOyBn2XKD|l=LZm<Id0?6
zrV;<nzP-Eto89M310GHjbrhA~Z|dsqw0c=vnW0ALKqz7ZeqlwP+&4H3SFo~vIhd5=
zCpk;NS?E?~Y#(kGDc|^C9TpL-dB?{$=PHPsMm&*4_4akd1#v8p`+XSKe_MgnjpEc`
za9-GpHPZd$z1_XZP|ij!Kf~n71<tc9Z5Jg_zPRS71O+bzm@!&fXo3qy3g!@D1+yk3
z0Wfg0Q{P;b&Mv#tW)Eu)Lg#a8ZNXbHgJ1M>5-2^gQ|W2mr=*Ebr7(pcsg1w4!V$(E
z5evK7$4}&+sQbuC_P%b%nHL~o1Jctc^#{sD#K+3UZU_+)fb%d5OPqM))>VyL<F<L3
z)zoPz`K|xOm|!mc3$gT9ZO^@6ZORGI(rsRJvuc}wGu9Bj!x6h1A-R0fBmG5U>shmB
zVmE}33wrv5|3Vw_#rV+aL7vmhyM4pGH2H!kfH<bp>=jjOY#vdIO1u%I=iws}S>4ws
zK4EoDko(5BdmI<+44Zl=e}UYk9zX!#0pX1Fb(jVF0-61Wk+;HG!^7jP=OVf1FL76Y
zgF7<}W?Q#ig)sg<RX62%ri!ib><bKW79E)FsmhN0a@g|Ry=*s5kUzD!)-ixh;=fYb
zsC-{F<YXw>!=OQclasA1z9tXB`9_9r*Ubq?!SmJv=KI$n@Rc*tKE~WcY)aADR`5w@
zpyQ9HZ|P|MMfWjVTn1@j!0B+H+I8Ynl<HA&Ei=Ifz*Wkdk)DB(H!fth0&*v)(#qD=
za6f`7eBcR5EJ|6Gr=%dA`iX8<Y{vfCw7w=&Uzpqb>m6h_CJ#2!_iaU3n%_%kDl}3k
z*f>1%zX`?>4-_W`MJjJgJ)@AqDbz?3hMd+M^2t-Zo3qWqxGyGQ-Am|*ymI2xv(Q<{
zIE@qu|GIC%F4uu=+HR}83s!en{Y$h?Fj~vAbH74v9)@!-AAea`^dc6TS8IQ-1DDV;
z!r8Bb2FG>9peYmy2?)6_zw-?$riLIm^o70qOs{Q+j5HlV6{Zs-4L4Z2aD8nJz-y5M
z!wqcvD`MHag*ljC{Cy37r%np4dA?VJjljtqsyaDrS4Ej_Q1;UIhLXu@s6rcoBP%a*
zj3K>-QX0zuUXH||#vZsj9FX6bw(q{jv>>6q+J3~q(`&Jdn97i6;LYKCRIo91*0GdV
zS^3D#hTgto$#a>#=KepAkRrWY{l`st7wf)0vUKB9C^sNUn@0rhPuE;hL%T}oqs}h4
zeA$*5nLfm`GlP_VuIAzH!h9U}50HJv6S!o7Gb~D+{K=N_N6FEi)<-5K7anKCf3!s7
zHjTQPe<?=`;^F5V6rMJgtiW!zb%@ODdwzwnpx>Bdp|j?G0oIYFqlD9Z;O3G6$aznc
z?>)aA$b{6e@O_2;uWaUG(!w;WjmuSvkThwto;CtHnNXXlD_^j|tFFCyC`jFYB4IKj
zSg>&xu3VsRJO%+<cr0~wUR|&65o52&i(M~HG#ZO=(>Eraof{R_T+O-lR4#`5kDHE*
z=r9-_u=M<Mn0Fk-Lw>u$@jFX&hPVFvz^1~HjMGr&n6huT9CH*+L-PJz9)cUU=GT0t
z^z$kJL`HTJ1(4^5yE7L~5p^|HZtiWe)DUW;Tf;1xGE)*4zbEe?)E|HWT5s~>=llRN
zl1rnqU$=W;F~CHLY>MPZtZ!+$(yHiG%x&|_f3#Jd0R;#52S#55qroSCJh^(bQFoyC
zS)FM>t2d|V!~Z6MoauRocrt_>&y90>(vc4N()IPlQg`FR(R~p8ez%vJ`$XP=d*S-}
z`r*yLbO-@TW_`2UZPh4G5Cj3~8tu>IfMjPyH)pfyAI{@wjyBT>|KFfb=etM-4mUPm
zd;pT}p}QgJ=2v{gzlgqK2a@i0-*7g5yjBMSuhZ9WYTg$DcM(@3Ui)6M9!z+Yv~MlR
zB>l>8>@8k88oGV<1d)1@i<Lkbw&xBX{w7RuQS`cWBrauZ7BfZ1(?A&f61DGg0_4jl
z1<+eNya_4up&vCa*X;eH#)}yYcd^q65H1+<c&w6FP4kv4dIP)gqy=7<cwdfht0c`y
zlh5u=eK{-a6%iD9+!>$>r3u*W%=s*?lNJypu4l~6G<?4-5wonyOv;VCKO(*EDw=kv
z+8t}*KG)QY4{N-LXkTb{^h$a)6fmzqq9BFgSG2avb^lpw)pT=t{*OP#)na=7c8V{T
zd|b&bi1<gt9D`Ok3X>1>9o|+fD!26gKKC#siY0q<iddx)8ynl>H@eycT)g|ztk!2H
zCSg})V%TszD><rNiofqH)wG!Kt1q6}tZ8OFwa!^B@nEZFm$T_V6;95FXHR9gOaWv8
z$2g-bCo!@Kpaf*VtUfiiGN_-K5=IN|&oE740UM2NJe{#v2>YBoSVCwPzG$evXtt^=
zB0^{pe#(1Kq7Bt~zvQah=(5CE?){1)<vcD13x>4BJd9&1f92^Zm^eYoKHtNoXvq$k
zm}F0|`DCd$9gb1waakT~4IoZqqmcPu(T-j_oSPwI36vizLtwgs34o{xiYzukWpVLl
zo&c6!@M{+0oG<z10qgC^>UTWP!VKg$Ep>Ac#KwMe)rM2m%RKN?CP5vLtL0Bj%rIpX
z7&#2)e#Q0({=0;Pe#zNS9IzJ85{R6+T^q_JLH{Gzq_0c|Y*Xy?^$!5q<ENtz?<;(u
zs|4`pjr-=)<aemHOa(v)@1n(s3kqGGk;+;BVrVuvz^^;RHwHbEv|lgf#CeTY&$c0;
zO>@z|01?ayiu7!kE<d?woqIS3+JGMx!-zvGlHyiXZNkamop<m=!S@Y>g$3{z0xh7G
z*FC}-J-GJ`HucOm&AUQfa6LKMq9(u!yWVwDQvzg%gsQ>@SJHix!58#>VoGhY<Uiv}
z0-m8@^uf&BfP|lz)w2-A2!=sa0fPQUA_iU34j9tDOiWeOY(fETkr)zJnSV80N!#yQ
zZ&xx)s3T(o?4jJH;<C42^w4uV@%1~ay?b510?;u<!UU4elz=I@v^`{C9~fF$oE%`&
zmyiXiPbpO_qVeAMSU}gZbETn5k=kxVx-<pqXLB1qV1f9&AY60_2kaw+rZBD2_sBz^
z*~%Aehr7b8X8OohB;P+M`PzB$si|Qgg$<BsO#O7!-K1QmNEl>#>Z_2IL0-NDVk>cX
zUL9G#tV3K&wNOU8Ur^lM)t4?3%gUPF_6H)d7Y()d%QAi_zBkgWW;t18{aep|+eyD)
zuH-7YeUh_aeiIVIe>jGqIjPcX@(3gc8B0IdVrb2gy+y?=Nv94f&tORSU@SU0Gkpgz
z??Kba#+9}08*z0xwMmp_$d{e1bu}-}zkAmvL!lkAGXvlACWOhs_6xvd>E`~4cpK?y
zkJ^>Q4_oBQZ&v!K<?`=yX@qm!a{aZ<bs3X`2HQwG+N$f@VnY$QojK`^G+juOYH7X^
zQyi*~KB%S42KE^=v9jpMy5y{t?7erTx_GMQs7WCkYh#<OtKRFRNwHF^fqQNL{b92#
z)rS&xlp2zz2jHXy;sX03Z4)D@uGju4|5jZSQ~HTobH6#VxXowK!ebaU*zk}Jw9K0(
z`T8EWlq87k!v(_jt&&3+o}`Bw9N<N`SUd5V@)UF&mPQ>q_3gp{y!rZ5Y)-~R5Fd^A
zG0LT>dGYl`D(`x9^I&SNI^zaK&#8TQ1?SFW-<hx3>`Vk_pjFgCJuX56c_wXL`vN>%
z*1r}fT(=p5lYt2cdW~$jCCC#A%xjT)t(J9J`6JD>h+gTe5PiSY`O^5wzvLjc$y;Cs
zdB6;fkGA~~uwCmRC*GO#(aYZ}l^@sQm8;M`U;BP)<>)ZSy7pInV+dt5*vQFwWcZyD
z%&hLP#eSWtZ-Zs8&wDXb47ev{L-5uBG4ZO^dD0hVS3E;jVy(pG5`g=i@1`sr#8>;-
zLE<Tt2nHX3`HKhg)C|FIxl}(jLB4p~4Xc9k4Zk<<d2tem85d+ILsyq2^`v}vF%%cy
zdVaOmjQ6N(dllieSRk<ua8;@jihE?s(~5Ccw0dq5q=&liJdKZ2r_rWmOgX6v3*z$i
zrVu|wqX{0BStAdL1Tn<|O*GNz<*U2kzaN3HdY%e(J)i2noI4o})2M%mUJP6%l8GZ*
zyE^{VEN~KtKXumK*GE&E#M`t1ago~^vZlD1@z36jL7#ZB3nxY-ca_*6TTeJy;a6Ky
zRF}-F?Z6j4vH;_G87saAjayxxWs(G#b+_<Js(^+>TW-k;wR@`Ss$^+kZ94jXgnwrj
zQEYzFeRVlfWF%TwR|H<)yE+N3?hKOTbyCePpL)XS7MXZM&+~D10O$univy^m;3c?M
z8zREy8ZXpsG!MSbFxMFJ!)(ZtQ_TeWiQ8J>#kST5U^~Hn_52SOdke)?t@>R3kL9w2
zNw<jMqXS??Ra=`ennP9M!#73*o&wABtf}4pLmz-L@gFOSobFqXRvy$_dj&H%A8Z)C
znQ;mjUG+h;d*+e6e+`zvJ7B;PF&go7*FV`-FTO<BFcg}H!k~c%B&h&Y|BO1?iB@dZ
z4;a|Uzwif@bU>l(R>Jd%gnj@chQVRsR;J}?M5`hv&WRkAU)o{$65*fQ9H2uyY8f-R
zHo3nbEayjQD;cfY(y-53c<)c%AgWh5?G}1o2I=@J6ak-_nA99AkKWIMlH69#Z7yT$
z6u0wRY9_*@BIK<J0R`jyTy8b*UDaVh2GH8Sr!7le3%hL^r*fU-+rQ$YV_evZ!CDiN
z+-^MwCibMUhbx}F2-`G;b3cH=9j`>(Wxe(NVqvbtTk0#>0TQV+#_{Zp_xi<2^(4(B
zZK9nvWT1LUk(><dBYk&r$i1-7EFPLrQ<GleF`8`3?zD%swv&Dku#Vud=Y~rX(`)3L
zUPaSEQ%NL4s&-`eLYtWs{~!+At=v_;+g!bi_0OfQP_Hya^t{A1@6F*(%RSonX0UU!
zb2F4WrvD7}kntLTGT&C;oETR_nSc=C?kl2WE7Gk(u$9HeZ{E!vzh)!>k)iZnvPjc0
zs=Bz|+rxE<1rSLC#H6H1S?Xi_Xy@bp=BB}OtU0<%DmqX^zPqby<9}EB^*xE3K_Vuh
z*~95%8!=VEEBUoGD{J@o{NMJ}XA9hkOtIb)l_|6li}kQ-9=++|xRRK9#Hg+{rL3s>
z(5(^6qdy$I0PH2>nxfED@3H+R8u@!>xaR@4@7Ki{JmH>>`l`|Q<8>3G?eU{dMFalQ
z#fz6_ia=6YS6TMC#-}c%QSIpR_*~KfC_hx*8}YeUDRE&yn}lxaTj~RGbp;zXD<@A4
z5SU0*>3q(Y@5^#z*oZoqdo~&i9zWdmhaA*%JiX;YE`ed6Q8-L(Cm(_nZOUt!)KTh;
zKn`VAR_$z9Qo>u@Y2rekNZby$&GL9p(N_9CfjGt~|1IU!GV|Bf4?DCKdo4$|e++-o
zN34E2>TY)yFAT2z>nk`Q0brUC=5Q}VC8}bCdmW@d?kFamUr+q)R~cKbMJMA1F9z4H
zht?SYEf)=n7@O0j<jANHQf=L=U25obXJq(%TY#~ugN!p3N0p&1=}&@H6Q}Yr;`~ev
z{^dk^T{7!Hl{KvNeI=$MJ%D|Zi)bE05I=n@ob(ex{(gZe#h<qF&H*nPqts%a1FCng
z_x6{tgOcV-CHZVMnYQ%mr4oLD2gSX1Nc&iTrocxjsb#6puV`%6+Q|tNGzk>yj03#=
zjIT)$r>6iEDx2Pwx0<t=og&VrcV%4JC6@^m%<QDaA)$%<hI)|*K8`CeHF<)VF&TWU
zpThg)mVbPeZ8;{z59(wWr%baxITI#(D4{@0O3w#!NX_7XDeKW`7L0CgP9FSNwUWfk
z%VNg%A<zQdT-~1BaQDk+kD2E3nQ(emK=NNVB-!d#nz4RFHkFCmqt>|3?K$n_9!iT@
zbEz>~sQ)BLQHHWoLG4RRUP1hbfxWB(`J%DeL&bvc{sYEbEuA8S!IR)%uf$?cL5WGr
z4PpZ_F{2av2w9}>hx}ZXSH>&vtSUL>^t||qgO2*piAU=aen0u)b^W{Rn<+}AOLmT8
zrlymc6PcePLo07qK-sDJER-y!ax8{yEB@M$*TqYr2t^JJO~dcfrUJzG-`<J`kTar4
zMD^B($nFJrl^9Wd45m^B^zgLp2*{@vYW=9c3+0O8RPh2p*<~|*k^^CpKdiaJnac|3
z2RzycW3c;<d~u}s$uQZe+5$@7?;COrNv4{IDxl8K%{&F+*;H{X;-LB<_XU9WcQye2
zR^EyBA=@$9LXKu4=^dhZ-net?Z}<R5L^$aZoybvflzhLG*2qb_9Y+scfXwQl-3=S&
zx_frKcu3AN<>`zv$GteKw>{jkJ8RcRS6b(s*UrIh_>!6Cn%(ACC)>RV>8SS4ti?38
zL5jO98`4E%X|Ny7d+jqM3zv%jN7Gq8G}*Xqc!adHNJ)$iX(Xf_-2)z48U>^~Z#tD0
z7;P{ZNGaW&1JqH{AR!<OsnNXqeRx0YKiKZw*L9ueaZKal#&2J144){z^~VE|d*zoG
z0!Fp(l7{kv^LL<Q3nnI(Zq?Wcw<|2H!}A&oF}y<y*Onx*<zX90^CEMJK^NFvs7d>Z
z!eKY|)Y0HCH@h?coG($$TcSpk@(<PdO^qrjgfg-njAO{y8HzESeqw{y6Qw`&153;I
zrdFee#`f$4-L1V-lbGqXcYF&;@6rbT!^7*DUQ+Q<liv8K-u65`p&-EpZ{IdqNUlb;
z`ad>rQNLG(l*<LnIZt}^ve<mtV4rYc>T|CdL-(zjXq3plRAvxS6J};3@Gw38?rFnU
z=l>L@PULh^6524_il(ju|5Z0M0r@9?^Y-d$v@Tu}EK;d+pGbzlLCbtou81E<iF<Bl
zCan4Le!Sb7nj`kPL_*o_VmRz&4UR#mR3OO5lwfv+wYgxSYsN%(e>Q_;^t9XtkhRtm
zo}(eP{C3v0)DRWeYk<$<=XbKwzIih)tl0cAcdH)F)H20I^MTV4TjUIXx|t*oHXxg(
zz52R@P8+_hzt3Qjzj63PAJYjq@Zb>6p9k19T^GKxv;WuC%7-svXJ-#l9y23@;7zFi
z1XG*uFLRuSyIJFj<Ps;D{Jt}@Fr`?_)`rDlH-xj{mXBwi;O33M`hJ8P*S%=|!#SKO
zN$>iAgBt?!<;GWen1#O2q315?b#r9q1ylHnzi?)pFgU1YZ<*5k>1zrwQtNuUj;|=N
zU~xtKeVNC0QYOoomBx3wLLVsW@p01$15T;~#EERVnc;xRa}`16zU8+3PX+H3T;}WK
zQzvMU%S%MOuUob#SS5#ySY#L*e>)BSqkxx}DnusY4u{>gDz;zZ3^9arGUK#F5gppH
zOxzL(#1R)KGZWF-D=Yv~-lykBG8f@m%`mQ3WvpSaNN)>!-PyVc`l7+6DNS6-Cd$S^
zT3(ph5(?v?4oHji=7=GZ5|5=ElX(TI<50RN36BsI59lhKZm6|Z6O2z#@96Ag|LEVj
ziWPKB4kEWvyC+kMQ#;?<n%&32vx{9dB`<AYzjT`2QxgBfx<!H8<3->eJ1DM(ps^ZT
zL0Bw~<`Xa|hRrKhCm8=5r6^u&d;G)fJ|kQ}B=O(c#WfpFe~TVS#Lj+FGc&z_rTX!P
zXsarxVvVFjz83b_UOJx;@<+Vn(#!mbO#o|U;J!6w9x=PcT)5|pMwm}Vw*T5$#@vDf
znq&U6Bljc~;Fua|FynU8xz{Eo$(4+C3hIC=JF{ipCml~K<Sc%DsMcz(vsLJuVQl6D
zYMjmWudXgg5Rqa20M|7kZ1u_`3wLzdGx!POQGfG?(G^6eMM<r<fJ6|SZZA`D<)Vla
zk=7<S8fMfx2lWk3tw0!4uVCiS-%Vk+w{AJnALjOfQ=cMHK_1x&8PY(NSlhnBScXI=
ze3thJ9V3MQ{z7>n?DcBX;xOTBfj%Pw<tf+l%vgJL_<_4P`bfRr0|dJHMR6JNjkTYi
zh}6vOI;bam>nvOP>fFq7If_@8hc9-7N{u-dl=`AcVtSR(j1=soiP#jIgD*B6Hs;4{
zhP47O%XecF_rF$9fo<{a-oiTY<ediA$>S$Q|MUz9_R8InN83iS-2tyd=Y%vfIiCN`
zhlN5MolOEM2uLZ-gp{Gw6!-TfKd1({>Pc;JPTMQR#KB7kA1JmB<=d~mR(~3W35%nQ
z-ovQuJOU0B$g&^F63o33#6S%njOoLo`AGqGb^>2n%hQ$QL=D}2T+nCcPV|IAL~y`2
zsoX6)L=+o2pX%Q97sTF*<n3RxL`qjGPP&e&D|dlDh={m}xKWFDB94l^>p~m6wuFf{
zPuz-xp7Ff5+)F9ml6b4aN#V}{zKHhFct|W6mS5>BK`%3W#YN8a?jj(AVm=jQiv|L$
z3n(XjAEe*mOXA|hbWwZpxl@6f|L8Vfcdg7GG#y%ohMlw^Q3X<-aaA(7^&Xh;)$-t)
z6C}9O748wDP_oMw(n1Jg7n5wUL@(ZYctAJX4>!=ef09^OSYN&@ZtnV4N4T>)XUSA<
zGa27E?n0H3FsyFBaPDr6)*T301kf0ZaWP?NGEqTXRrt+f!W655J7y-n$Qn97n1qz*
z{_dq0=*d|8qAz2=@JM8PU6FZMlErq9Jwm&h2uH;#nYsOX?qBN9H@u@6ER!JdSVQ_u
zROG{DOEVGr7}zuCA)-KW@u#ZCE21eLE7Ni8;J(ehy$|xHOsPF$ra9)WfMi;yKwEXY
zN4tCb3Z{rbyBZ}i(?@2{XT|;S(6YPwperEF*4`Pm<5aw3!Q=d+W^}{A7feT`zc^CU
zX^9{Imw*KuUAFEoY9(Mt1X-)t(UD-OxlOeK96gRpZ)4&rgIZv{y-1Zl+x>;}=cR%Y
z!rD(yx*GRS#&dT$3>8ZNxA&A7i33Z$@vdqx6pQr27tPHdS@f~c+d(zeUN7sFV?_?t
zjr(V+D&xU5M)Km5(dKKhPa@ny+?sB`kd2$#Z|QG`ffzD73YG)qR_72uksqS2o7gvW
z%(VCfm6@Z-E7}nju_Geirz}rQklD9fG$))`7s^3<AW8yeu?RHBWRh|dMUlveMv3nn
z2N~6V_ZvHW{k?seX3{Hd{W#A}KugkA3zkPtSMuCxJey!&>2zhsrRIH}_QD9~l-7g?
zMqb9@M8eTcfw}t=%y1ue4GKC(v*CD~uAU1eSc!rYd;0Kwru?`Lp2{0b^z*(%{JV7F
zfQ>>5yRyk}kN`)g!^v|^_EeNz0;K<bO@vR=?6KT{QSGDOXZ93#*juhaTD8lwbQ8zF
z9w^HjX$cPUyNBr--IcC$E1{kX1z$(3UMVV1z`h&Hhvj3f8fMA)E^hE;UBa<k%_qpX
zL9=R%M{Ra9+RdX&pCL3j>~$MD+z53<X)w{wV7ooO1!PvS2l6;0&O$<eoljV}5urPq
zv3N24?jTHmZN%9=YU|6g-@c7vM5{tK{tu#YtpIgK`GPqMw<4s*2Ygy3Ge;5Euk}6~
z)R67<Z6Zcl#EZ+DJ-<S5+q7Z6EA;`4&-aU5E|ev4#2`_N-A|g;>AP|II~?37KM9iU
zrA3wqK!%Kh+@8=f(}rr(pnukH8B-nTD-1r7+bZARz797^o9(EQI`fFghr7V`C7eFr
z8t+suj$=Jtmse+7E7K`NiK=4~x+fC3;3%_ZEP?bl3MaBXaHbREL81jT5A>6XkW5%E
z;RtM&YYGG=dwK4|r=%>=diC0{%}vUOhtGftO0r5Q!<TBj;>t)OIw2$L4)w+J{WGUk
zn=h+P`?<iyf#+1Nl;=t()+=l_r#s_{5QMtEcZsi4Eu1Vr*;+b(x3t<B?)#FixoZVY
z^J{Tb=;kN@wxZX^mG3~<=@<Iz>du!#>QfS-S_bQ;0wMFX9QL3&Z0u$|dA0TJumBK&
zyA*3V@lnPm6f}i@Q-X0*{nJ>Zfmd2f9S%UXWs3ArSU~;t`7kShZ{N8}8k4cEyTq_&
z`;pDCh0yJy%0wHiCC5mK8@=>GQi4mHCAi3^L`LqQD85sO-2OSN9Y3k=InJNX0w$t9
zs1^fZreB`h#e(0;mDsKoL*i`X8L%?l7F@~$KHzTTgg=G6Ih@+pf*55u<-2Ae#;hDQ
zD%w0k7cubP?sy}5G6mThJQjgcV9gUdl6AHdWX8K6N#&dJpbv>(4$W_6<_HO;Hr7tN
ze~NtTfOtTDN)Ij<JUAD^FEKbKk)_=A$rxWMypJaQcM&v9Zq(kn6pB)^2dF9Oq*B0f
z!TE6T#`gYgL}%BRW`Rhjmn^KT`sGH5GCHc*HklvBpBDoS1)|Un%Pse3wf8>V%UFKp
z2119-WAX0%U=elsHnFpD@!;z6<dw6Y;&|u5;aE?5vnTIOL^xkz;o$4V^Gnp6;>~ZY
zz~=xpK3=&TDn(&8`X84Z5Ek2%S73zyrvTtYz~cGj66&M_!2R&x(|@%&&g^}0Ep-NL
zx-vvBNsOL4Sf*@h#%@l~e!BOb;*vA&6j)*h<44A)fk5Jtn9s_Zj7gCayz5>Rp4DjU
z0*mJP>Eq_t=<BveB=+>*pPnzb#f9XB8lq}C-IA4;{+^)D#!uvd$7yBLh#bHEg4Dhb
z*nb7oic9*Yt#zoBN&k_-lL@)rxM+Gq0i{Wk9s4+s@YUyx{T8nh4EV&_T?4lxEj+8+
zLtC=&$j_&Fe5&JKk9jnX(CR_Cme6y#<>hG@`PK8`|Gx|6PQ`5~-rXu@yNG_>I)*E(
zdfDhUPD25=r$*MHs6>>@rEvH4lRFe(Kj0a0D?oC;Z*v#-(S~!S=yEvW@&HO2{9-BQ
zkh@uaRxcUN<j}o;Hs+z>%(W%9{k}u_GgWiva{Egm3iQg&!t$YhNuzssC+UfCrFjh3
zAb)kTM3L9jPD%kf^qW<bPqk{B*?;#<D@gaw)oY;3V(|;w?%@Y~hfQ)5T&@XGf^_K|
zV;t5zN%8m4Xx=z5LiN-uMHxcXaahvj{(+$O&ubrFfyL0@NjAS8HfFPP;z*uduOxCa
z3>qcAAsd6daC9AtG=H-{i*>Q_4F!Csrb-9(_RD1qegE=FlSu5<dV}zX`rlmR<ttRQ
zqe+hemR^XooB-z!Mjr^A4rlj#Pq2RgBt3u1ARj9CJRz3gQ$D0enZT9Y&!{omj6ltG
zV|_gMd!CX<z58`j^K9tiVi7uYQT69&_VTCZP7F(AleOCii?3p2KQmxWT{IlHjUi|_
zN%Qb{OdI0<4ShHSYhL@cgfq>r@;0)S>W49RRU)SmKO)OH0BEI-lDXs_z!E5p4g9|S
zx%zioet)TWx4S==+nI}I#(6@goLlbcJKuasHd?2<tM!$1iTF^DB_S951V<VO7bdfp
z{uTu-EZ)o#6niEjYEz)F+MMgZcwWpEQE8MZ=2p#iinY7fEO5xZH{bi<T)CI9A;lA{
z6)Rvxq|7`p`G)K_5xIJ7Oc6dr*~3l^m*iu(CzdF!y`@WzlCd$Vg(^WzoTJ(P_{p;W
zy}CKRIrWE)cVR%5N)t0aMr?p}W8pSJ3ya@CqE5Y93%7%7j)P>}Ru}K^7DM27FWx0+
z!pA_88xz5Vhr0jO@=A8K4_;!r5{EuZ5w-vN@tcWZ7*F@Msqtimb@HTtF);4m@fOG0
zhlk;KNr8u!zP?7LX~FsBE*9IUuET|iiQ&!(otWp83zmz`XIsJD-G6|tkkxDKr(f`-
zEtb76$C~S1rVfjS!9r1VmXapfHYAnMSzVp)=G<?`M$yI7rkpd-bY>-#PxAUHm><ea
zyUnI&_DN(8taKPln#*P<BG*x6DIJ`*&{h7f!!6dI(&gC{gB7)MIQ<K9ESK1gRk>u9
zt~iM{Sh{>vt?zU*2k0l0Q<3HiOUS;kg(e&P8$r%Hh|1?l53BCT(SOCJCl!?#|B#cF
zQTS}(e7+({U@&YlT4$-=Z8}(NDrl`#3(Bsrg7Gau8C(zHoz3!Ax=+8+BvEdA&>bG(
zh4hIiX$>9)E)T7m{_zx^bAjpaj2E1!CSrBz$9=-%KO3e_fSZeppopN?mHJrJjb2J3
zrM<P3GpEnY#N4dP@7si_`+e5?VC+LFFRF|?Rw{N7Bx+gO?(WQjzpe83kZ#3#X1R|~
zI!R3w{;e0$x3*~Sl?6_W<_dizLiI!7iFmETU|n~Dh}bZ4;&|Gtv_Q%?c5Y0iPQrew
z$gyu-FiEfsaI1fizp&B{t3u>6mEx&wuK!pKc|}g}K3aAL(MPr1R)bXijO&<ceYP9Y
zjjqHI;!1jfk|4R2(A{ox59nUa`cLBTJW_XJ!f0Md6@13x>=)Lr?3E6-XyMqi(u6Q)
z%6M+OyS;r2dfRjmiYC<0qe3Ff5{=2HBWV#vG8kvT`R?L;;sE<naLEUT2N&zTBDH9~
zz48gYx|(S)>D}w3E9O^RZa(5|q{~2LW3^ygQzh8W<hWINojN>_OKXo4zQEk>@XFkM
zfnBU5qDfY9KC!iD6druk`gQ#Q^M|ON>oI1PyFfI{iZSv2pXYJA0S-bxel$I-I9w@k
zK^p<E<mnG_?3oToLY2Hqk+|lwt}eM+;VF;3HZ-QH5{%p1I4c)LUc(6%EX;T|n6st8
z6SiZFOi3q{dymQ9pQk_lXBu$X9kO*oi}2Z#)m0m2;OTpx#Mz1ZKHW4cuM!Ju)%kr(
zC5T%&v@J4MvoM1+;&sSKdcH@6Q)8m^`4jLRwI3l-;-W_M3YC{7#k%-gd5j9~WftpI
z(HeZT?#yuKYDcWH5RiWSr{brvfWMc_7$r^5p&m?ug&>>E*pT95X`Mt*?(ap`mhavn
z@d*!f!mTma;c*F+q@)i>G^q?Rt^ctvVC22!_cxt8r4EW<Me>4LA)MK`R2+v<{?<VA
zb%-Tektt2#@uxB*mK0CSvVfl3Z_M^-wQ{X?>BLS-49s*YOHIxyjXC36oXF|}ltc+D
z%j8ai;_doW3AXOG`fNYrJfP7yozt{vZT-(S<1`1dzn?GO+(oPfkG^R@Vbh)Xv-mg5
zNGLs5rjlcToHH4VDOc_2QMD<U5omoN6G6!<qeFFAZ|+A(1)-<ic7=27jTT`eqtwKr
ziDHw|JnG`_dZRc93tRM11A>C(2)&bmxaO`<UMk&kMjZQyu&}&x5*4`SQgg&=TU41&
zr1Rna2MV^pMU4Am(@N-n!C<JQiOKVY;0;bLuTo0FpB>*p^dDOdpZXUY0jwL_O;W!s
zMQi;ll^xA6*0*aO!)uc0N}qfd*2#fwsyLoW`j+6kGvH$8wkgjYxI!s5!m$=4ijwUu
zzwTwlj{AZho)5zb;~hxlKb<n7>bspbp2F4D+K&#ZMT&Ng&2pQKS0IYICONG$m*aDb
z2|EWTs3v<?YTB*)!GL?<ephu3*_;ye^pt9SxBGb$W}OP-|0?KFNQ+?F$?0|iYhCS1
zpg7MlY}a^nI`$(;*)z@0NYs}~hu$P2DL8LmnTU~NPmi$~9y1Gkw%}Dmfb=&AC)$mu
zWsTcobF_yTpMMbI0mr4~`%pYrEcR4fzxGR}vdh|G@ekukNEs#YWY^kxA(Tc)I-g&`
zBaWzIunjZINGZsbBFzikWwaJBwBs`&X~Ea7XjARM4KEd0c!+dtr7X4LCsc16*c1wu
zIAeQj4awQ1s#1qgLPFk`e+!J(L`K{BIQ$CM&e*VH36MxE*3NR=xa6Iu!Y3gR%<W?%
z6**>Se9_595upztC}SvHwtvrF-u>@4--uE36DkAsW(#<h>`CZ163roS=bPo(V=c7L
zU-<=zQD$g-P+GM_FNCVug$?gt2c7!ogI=Q-n4*s5-**UbgqRvV7?E$lUSnK}`|3Rl
z<gNCMbD&xoY^}>pz?}_nx8EK1W}=TsxiH|%Xo4o;8bEaR2e$BtYGj(XnTWLcN0t~3
zj|v3|l~5nj5nszLCfz`VQnoZK`OT>d(IAzFCg(^SOgB8DExWz)vWCx2)7^pWc@IcV
zxs&SE@bD9qg#)~Gw!Y;Rd3r(B!*?7}t;Rx<d9>99pWKZ7KY?0wmZwc0t$TF&@3XM+
z#R(WaRRy|s(|=$CZgb(QQOcFXJ(&fPU9crt`aT(T$FW=LAp|jhS+)yfcoiX%uLts8
zAT(%xMSL@h+#i?5m$RUDb}NjS;-kg5u~uuLzT(}ourNhRFwX~8EOfS5KXf0t@nS92
z;<L^xzB4Z_2&Zry6<MVY6|V62gfR*z9cgykC!m9OMX{E(>&;w&Wd0Wrr{bzm4y@0{
z#cFkAPP%T9(=q^a&2pwEydPYHT`n>|Xi{AbOgHU=euQB*S6kt0s;*BPwb1HiY5Xqu
zF1n$f?Jfxj%wcD1CC~R4kVp;LfXE~{YD7Rw22;6FUIL^pM94%n*+|Um^`CL#p<cXc
z!0`!$B#^@)aJ%>Eden3kqlt`d2|Gk<C4{#yy07n@S+5Hp9TAUoDCIW|9GtyhL){^q
z(i-v|XQzvr=laau&-<ZXR-m`v+&AMnqKo3KNPZbT=r>}D@x(Rzh}<FgVL<jRF|q79
ze>C~dMv2Gr?)%^u*=`hI3ZLjFp7H6AaexT2(BM~;?BYgN)`61#j@tv-X|+frf8UwI
z1;YnKGdU8)P);ZfPQz?$>#T*o5q{4gt{UVxyzCnL-D{j}LJNm-q*(}vcZXkZ|N94Z
zTgHT2qxVBh{kIn|5uvRLRuND8u--{axAO7JFq!u2+otf~`%BFI4pvoGa!;oQID;US
zZ{J$0cvY^AOh{UBJTe+!6BmztGH#ko$&HuJ+iFm)gMaQFL#MY2@7EOBc#Ss{_hf?d
zLEhGy)T?S_f!0Jpa5GXw*-nLii9tiSS|v97;BGVEZ|8fkkA*8gMN%N?+-yNe3#1YH
z6aF;sU0GF#lqF{KO`~3ul$1am5Mve-6Biv@G22($FX#(tXj(-#&oY+psI~pOcr8Uv
zrA(6iMEwH~n{AG{6{ltZnKMb|`USU1a$Y|Y*<Oi5|5Fldud#07%U)k71Zq9$zl4@!
z`z&dmthp$B`08E1qD~|7#5g(eNKD^zMK0a#5nmi4k#IZsmj%X);xQa>s?2p-Q5j@f
zltqb&i@%@TQoX&My*u8$Ix8#CFXGX^TAVu1R1s;a`)<t1@_E@amx)OFLxJ_&!ET8B
z)Qwk{2uVlF8IjY#fP%PeD=v|LGaDsoien}-k&bb>O6<yc_@bKCR@>oqQ*D)7p!R!K
z*?e_7Dla?6xhzVrI8k$8>r^;iUA&2{&VX1C*C)1>kZVzy{=KFajyW2^;q9%+T<TSN
zE=uM|_}p8XLM_(3LY-YDooOkwt=AG?oW#hE+L>cA`-%9J$-Cbpp(=hxV)w-Bv;)Y)
zzBi{@#pC`~$08ih!=c-yeoXq*ejeODwu*bt;-_!xG}n};okPO`wWBMUgbL|j8Zo{2
zgo(XG^JdJ_sw4Bb)bYyW^G*~#jv4$W6>MvV5a|*<M47Dcc&@O`;oN>mH%%O^_Z*am
zvOOc);xlM|PTp_=wcFQcTuBq74ZDa$+*{Y8kK2;=$5WcN;qKO}UBRKPa<p^=E?5Ff
zV>7oPv@G!Pb|B!n(iM8`QH))Q;KOHSU;ARxfZL0N)%*3!-OA;erFUQ7NLLgDNTmm*
z%gXkp{Q+@j(n<jinX0n2Vh<kIn7|Q4zAg14;;~|1q`^P`aHFRi-We0K;6v!Y-82{v
z(C2@j$q&(gI@{vQ9MRy<;1<0mC#Qt;!cJhZq~&184&C+Y-4zMB&qP#+%Y}B)`dsS|
zw*K7;tFhvN4pT?`zVYcilG4#caWbiKq3Go=wflR;&!h)YS@Rjae?*tgB0q_rVF9@*
ziOe_;mE?_cVf;8;0D6^U$ryV+`i6WDJ9l56ZFehA1efWR>AW}VD;zMhA{sX)O*lKd
z3#;ACO<EE1w-f0T^vj%@J7c}SxZ7H2|4L1Egl=2CSL{qx6HCDfw;Y4o6aU&U3Nih1
z-Q`9qGgk&Rc3<q0MDeZ}*=d8+q~^^184N!*!lU?=R1y;b<mwW}y|Z@z<fLhZLczrG
z{*YzlyIm0OyBO@tMzw$B(w3h+yD7$bsfZZ{kBxjnst3+3<%<T{r9%BK9<d&Tb2s7@
z&$8M<tXzp9d7U1R&_I4oUd<z64*cvqt@mThcCJL<RzkFKV-j#SY}fhm*NO5jXBk<^
zzqXd-*iHcsg5agE>7hT_mwwp|=oH@u;s=dbdVyp$Y_f=29s>OKinHabvKJ+9O8?2$
zZM7;?NUmqU<4{$R8Tmw9=uZ2)fcyRtLp-5e@`#K1N=NYYY>3-jZf+MEn=EK_$NG(s
zol^S??LsGroUuN4BR{=RLw#$k_r8mw^-|~$w=J+>653+QXMNq}F7yCI>NdB+BuDa?
z`X6FpE3<WByyMIDtR-Q6m120r-8k!gTkVtH*v#ly{Dk&kDGs&mv4Vb%34R)J_4KWo
z$>@S3%V!IX2JDX1y&3|eRUoaVP-zjlie==<yc1(IsqBAtR3iG2VgK|ectliVBQKe*
zLEHdq^U{&Y<O9<dZXgx5h6PzYtjG^lp^JkF+MBxl=vf9sF^lIfIjwAuz-5U&qs)5b
z@V;MnY^E;AH}LA_S<9RhzZH$h5#-lDtj&Uk$6Xv@_<UCv9G57q3@EMgI<sDecm#hH
z67_);5*Z1%NF21U7|pBD!)lHpKCOi|z{}#Uu)C!Ng&8sZGangGJM8`^P=!c2aT~jf
zCr08>A43H|aI;F|uhcK4>CS0TxJfGoL={V+VV%vwQ<ih~y;@#ZrfJ(#C&?u@?;J#s
zeI+$I^Z}W{R5JnYdm;I%BKuw6qO5;~(S5~zy58CUyjR=$_|^Gr{4cSL_NpNPpxeWG
z&fU6EGOcv(P84z6OFLSwxhF_D>g|2d>6c=m)gqO!EmKTqWk;A5%XfS(Ks-xP>CH@Y
zpl=}X_Ox7ClA--|dC6DNBn|w@QP$nPs6TrakipnQJh(H<0NMYzuVeH1@aBBiuL`J-
z$sC&|i|^lzmD-)XSdYvXKQ>eVUJjEi4=u}A1~b#~fuP*a=mQTE>c{kJLO?`{R;e`b
zk6vn-8Zv6HO*k|C{-AipNJ)o#?L3yRR(K8EsGJ>+k7ff@bQ*<Dpk{fzJ?nG~{sY-i
zPCi_2f19}pq$PUvb-=yC`!x)XJU&l!qMV#npF`(rz{dUoKlnt#(vueG<<csGZHri<
zF`WlH!`h|~yb(WUBSO0r!^37-22@G@l*YO&*sJ!uSf25X8fD|=OQLoLGP(jZ#yN3f
zAZ$9VeovZp(i6l9bd+faq(_KTBP+#e#2>S3c@KK~5(<Kma#dleRJB#s`gKb4{mPR;
zWjc<IPO+AVZ$vy-waUP9|1tyRHZ;~qHZN&}%JvzluRf(A5fW*Ym_pO&LHBb1`^XQ&
zEtq+ARVA$Rj;R6`jD|aZLQ;tAnQ?AM!pI9b0;3Y0^2^(Fijt@?yZIj_P5oI@m6$2L
z?98`GHae?_v_4q0(&}a3C$Iwb2m7d#@!NpsB{|J$6#JJfHNawsr+Z0xhzyn=W-*B^
ziv2qGpwf|>tLVeE;ZGVy2AOS-&OI1QuFlfYmY=fmJh+@~KS}?*6*pAXJUVsESah8J
z+?j(}(`cuEMhN^fP&m5nYNhdPXh+-JxxFxkr@Z=l2%=qQvb)Xpct}pa>_0m0HMTz7
zNTL)XunxAj3)H~l{WzH%#hj@>@S!a!Ui|k0ES9!%wYT!rH7?Pui8`MB%>-K`M3^u9
zwC`!Xn=`X|SW5!)BwhlJBbf<%dW*!9(+Rn7-dw4HVD&&rjKWl-D7~oDG)<2nB-Kma
z_zsc1-jF5v&3kimL6QJUOZa*tmE*CHrB!_STZxjvQh?*F+?L~TxKE^KM9{BHO2U{-
z8!6{+bIuB^nzA{Qcpr4prAo6&QCshN#inR+ZKW@c!aeuTJgjj}SCA4Z=2!l+&|XI>
z_`5`{B3A_cA=v*V9jG#(eR=6cJ4p&r^Z7UbQ=B)8YEdrcLOZ^m;nvOmmZ}$ivPHt>
z<@%U|0KgMD9+ru1w^(qmBZ`G7Fzy0nUEw$DnERXSDb}va-UQ|XC59gQ0HTSXU!6zV
zla#2l!|ZS?+WikQWaTp~AU|bXq&IoCo;_F=nR%mhv6f8BfGA4%9-Eq%)e%Kb{=95R
zzQB%}V5x!P+waZg7;~jK3cCPEo5|bJo=$-fPhi;`L1|a18VhQ6ATS9erOXXS!vidG
zbU1A@<l2|kz1Xxg3*v3rb26s_nhpqu@-SWm1h7iW?SFLaV>+>45(pEWsOVZB^##NG
z=;$Wxs49QZT7Sq9qjsiFr!*BZ;>)UQOE}w=ctsZHRvWi+dxTx&OZoGo<Zidp0OW)n
zQ0-rwgAxnBz2b85DnAL%f0c>*jGX8PR3OgBkN9aXP4K(_?D)K!N})Mh>j~xEcc@%p
zQ@2(xxd0!?#jg$zBHlih;>|PLWa*m;b-BIf)W<l1epufF#(hSrJh-xac>L14OQQYd
z+)nwc{%663g}O=~bU?m*{xU@tZk<x%01r1646d1sFQWH`nv#lGlt;WJTXsqt<Z3;`
zi4{gU^5b$fvxJtj)%>o&P-U!d*NB4CVA6s1!PghcqL6ejNtTF~X6@fM&zC0>WeDe?
zS`}qeI-7|TnX|{lAs6dI{IJC_c0zn%l?T7^0%r2rUmQUPkp1sk+hTYn(_aV!ms#G<
zEtD`qRX?w%83V2LK2I<q=SMM08TWR1^E@PImKWTjR?loqMgCpQtveU$<>RT95)lUw
z`PvbwL+AqS*ZsEb$q&M9(85!~&HLDGyE%w0xPq4LgJzDc(xb}Y_sUSWBDlF0ESJl2
zvAQ{i6hc^h`lJ>e?P1?pU5*08W?YJQZhX$eopPY0q{8J!epvX2Q?aU#JeYT||EKp8
zx{_7k@@jW<T9*TF@XaY2+n(>{dE{Y7#aY`bT-+;D3zyn`X2mHC#QyRQdu3DXuz=$f
zo^r!-vVima)g@EgXYare&P~}rvDqjKsIl4(t&0)!@1Y4H4gq12AvJ(>e~)o}*DAwG
zo(W)k&Yj}-j$SUOBgII8Fgni3f|d2b1H66`iokKcKTJB|-?`;}605h&MtjW-l)fIh
zmy@nYFEyUlgypGFGE;4T?%m5Cc%r=r*tZyy%<)HZ=<w3(+0io2iVOEy$Ngy#q@Z{_
z*M-dTcegt%f>8_Q=g0tw7IQWKLA2J(G%7MwRkI^wGoJOtaB8$5<B$E8i*07z_i?l~
z$wyo+XUaET1wW#a<oP-+s(QC4_~xc^X*8e&Bz89IzKCT=LSZC$$5<Eg<8GPze-uES
ze#z5wh9B%W##B56$N1jjXULxM%H&Vh+nL5*Iemhy+X7`8ur}aXAbwt1HrRwyps0#K
z1D9)UKuvDR{PxHXU)o<FXb|CTE3r-#;G;iI!AT2ydJ5|1+i__Eg&vK}kQ;CZh--4|
zwYG1PyytkRojRqO9=kAcUT3Q>BSrJKXYBR-qfI7w3`S=J&1f5~RzcXG)I(iMNoQ!W
zIFx5a_~*3pZ8G4+nWKtNK*(ONANW0J$qSPH`71wzA`V$munK{Te-AB7&h?_kuI4fR
z#NtZLl<+Q}!H&wON`sdk&N%d7r{cx%Wh`EEzke$#v5Zzh=XM6$Sb6vp<%?<e_zm2%
zKD4!LM)_dxx}XP9?u&V?vmY0BoXAmU5xLmkW@|#sAVB55nkyT3g2JA%fa}wnv%zVU
z@^Va@1uNO`I(5DjbU}!YVC!nyZ*u~|mO}39l!LFm)c+;hk8MqG?bDhnexGp??~@M<
zqp>(&o5V%FC~vH3VzmsANTs);1Z3SxjM@*xNgy8FHlul>13o#Lp2fi*9qQ36UgM2D
z|HjwW0nJX^jDe0;q8;(Id`?4I{3wSqIW{Avh1S!-XZnGjn9t<eSWr?P!u1sCMWmIP
zkzPlXy9Z_zI?|$~IF2NI#pCQen`Qlu*@9T*hsm-Iea-TMv(Y<*>*D!Y+e_^2<Tk=G
zr*+Jw(rNJ-hMb-vN?QZr+J&yh2#s5rZL3rQYCm$R-!Qge4hWowpt(>uYT*`(QWBnP
zwB)w5du@zc+g@9;%LVSsQA;Y!b^CH9a$`tQNv9S=Y_i&ZSbbm;`>R7yLGI=jeb>C$
zgx;GdcFR*I-bQG6F|NnaQN73K0*l`-t@qkysL-;ci`8pHY9XLnAZu&+&(mTvh*?0B
z6C*H54N>95*H7u{pONa>$Y{vYvQjGun|>xc)c;S!Imq`}{U=KN=-yp8=cHC_y^PAw
zTywKz=?X`VLOyDV3clJ`svdZPXp5kygn_;w`&7u@@clE&**T2&d|{lrAD%BQEB0+F
zJ8WnowQ+2o!%@~`FG2LUjA8WD7X&<`oa$eQXxUJs+;X<F{fI{sBA-N{p!K_$bJFG~
zwq$OFo(o;gS2W9E#69RlFr;|#ah6SrAMRtpWK+WMPw4910BDLiW4J5Hz82rAKMG*&
z{Hnp8jyTv)1&KD!CMK$`02g)3hbw0df=;mgOGu)Jp6lP2ZnqbAyO{eqM%J#dkd;99
z<%WJN6a&2%`vZwf<#?#6uSlEQD~&@VM6MR<Goq%hI12$Gu{|GtC`2t&CsG}?M*LXy
zPGJ054-7$)3&v$rnMj@g>Y0lxgQG}hxa9&#0eQeHC*Bz4>dIiJ%yto>T;J7M-Oo8t
z)t~h<76K=n5cz1i#0rtAos-KjZ+%#^$Krns661o6UX<6GBMqx_`>66XXnBGZdT%RX
zG5r3f<8J&~Kk%V;7LDxQbHkfAExKLw#(7nV1xy6~S3EJkE2O7aQxXjij3{050t8M*
ziLoji#hc@%UH=3OV986PjqcQw)6YWC<sgicb)NieYIe6?Z0qGHFbCNZe_xT!YgBie
zSTYYP-U%+=M(Bb0I-jbr^xFM32gG{_!B%;=PCXNlt?v+3Sue*dro-T-x77!0XO6E_
z);_@>&QtR}x3_nGK3g;Zi6ee1!K{8&o%(6u_u{D%&I8+Cfc7_#Ch{#TTZu!Tt2;i7
z!(lhXYhHeYd9T+w$htJ5U8BF4*Fh6J_eWGDwy^n$7u<1;XIJn52?pQ#DbYvq;?Yf%
zfA{)}n<~u9hNxl}541bL!tc)A3B44G5)s&J01DOSzWovo%UYJhZt7WCSacr7VvGCe
z>s>$D5`C=IFeoVhE`q5if;)sOk%fh2$0gvTd}qh2yB0Hh-GRi0h32|1C(1Esv~V-;
zS6GH+5%+#O;@e<XlRmopW*w`8+q#_1Z`}3}NqERVLSK(d+j|%C3YxLE^Z&Q^Ez?}L
z7gJfkU=>8V(4e(<#Xu&PJL%ARfvE!sm<M_N&28v^X8^ZIr3se4D#N-xaOofP$8~7H
zv|!1xu>2~ExATSGM7OlOtUdu5ewkwv$IUKM2}Bfc%Pa`!eZ0V>5TDP3#P22y+8&C$
z&(v_Kp(ALNSf`~mq!35D%IDV6lO{l%XFARQ{K=4$9~_K1z9gRa_^GV+Bh+sYXd3Rq
zhQxoevoe{gOh{Vnj~ekh*+c9^9<+vds(?<6owAg##!aMYVz(L!g?CayzwEbFGcDO_
zEsVT&es%sDEBWZ^l<T(g^aSd?ir#NJx$8GykGBatj$^hIzZ<f<mZ>kW^5fAiAbwW(
z7uR7fz^i(gb<+KFUAFC{#JkKIfDwG=gP`!-^m~_kNQX-z&K8SEY@yAzEzNTEj9BV^
z_TP@<fCynm+o#6OojMJKDnujJOk+Mk97XgHqoFNm+i+I<cvGp3qGT{e+?lV>E1B(K
z)>}XpKZQ{4#<S~C^!>JGUs%s=#FzVC73PHQ-=%V4Rjt3Nwld26zB{IW2JDz$wcY#C
zX|(ZBCS03k+8|#{sQ<VfZPzi^WFDr&h4j0ZkneyAAfR+EcRU9C?CqE4SKms=(i$#Q
zT40t?d5w$juUJ`==m%H~PW;{V^{4c0ux|JQ8@MrepGFB;I9j^=F-KOoyxCJ3f|upx
zR$V5E=j)32*?Ra&I}f}D93L5!#a_<s)LxQ@pmT?ZrO+$4+lqJcG~t_Rp`MtPuvKAf
z?abe8`uF9@f$;tD-R-JEqjPn;YaT0H70H1gl>T9;i=!kzaS6-!Y1<(iGigcw+##}*
z7^kVlH-Pv{;cKdxBS;vlMYys~Bt2zh7hk|KBKPn`(b7_%?}-fXs0J&vkrMy5^8WM4
zznhMxlTXZK`0-`56g_9J<<kmM=a<uYZ7r^T1}XXEK3c=+DcgN{Vem$~xAfr?>;Cly
z!srkw)QXA8(~|uL|93cGM0vSKPYef`dizB^(U}MTgiGLZ;NpLOb=$tm3?GGR3Cj8K
zl)6?X8R(XgWPti80g~89c0z;3u0n`@lw!JXu-T3Uyc2WQ<*|SJ6}c%dTM=4DjvcFZ
z?~Xe>1<WSX&^ZZ#Pye(v?$18@O#aD>u_0a=A-xt^va32gB=su)lZ@JIw0C-p{cApg
zE~RYqJ(1Y7{4og=b2=HDpcR?78stMSjWc+;)+?s33S7o6fSC}qhcdh5h}bovfcl>X
zSLInEHL)-H*(NJddHSb@R!S`&%3+=Sm$vL=dUN8Gh5^HTm*i7&JkObJzy5lX((v!m
zTOj$!3#_$+NYo&(e!G)6nngyem|u+ZGnGaR=3IzA@Za_KTcsWTI$>lKSI_uRojTQ_
zP^zwfbjQRB1oXsY#^*eW0y&TKmpEhCcuJ4#jtV*1>iC0MMAh4~9u27Yc1#vAQ>9Q?
zhJhTm9djsqW<L63m0!F432FEL?iNW)<oMLUwl%2fS^v4NDy(q?F^$^wvwg+TOLyHx
zV~8lrrlut@LZKJ1IR5NXhWC4=xw&pnxj>S_Ctucq?)%&O<tE+bp#0aB$w95l{%Fk3
z{v4*{{$^Mo+m0S~-|uX9<^DuYgU6(q{+l)+(kzj)+<2<*Cx$j+PHbkZPmXg~V~vYO
zn$u8>aQcNWsMIq{`DG`wX*uHNYOecp9X3p7jY)hnL-G3I?@^Y%clDCGO`#(#dB2)l
zU{72CmY}DY%#LssK2>QxwV2HjFx8520e`MmW2*kfI$TV*@N(=B20gB19(_eDlhCCQ
zygK_T;}Zq$=){PW<Qq!zl`!cMteu#5=?h06La)j4{>rD_+QD<MbHuFc64VpP*||HP
z|LPL8ve?vg@bV3Az0Pe!x88Wt8vGrENR=L7;{#*Kj?<NU*KICtwEszWcn-A}3WtZf
zG?!nW4dSMuBQCC4?+(`!Y#3)KY-+KSd6N{6%(O4z@Pm7j^xM{#d{3ah2X>5h7n<Vl
zhvk<N2p>O4o!1sOjmuxg9JH-WV%#XOC*I4!wC+;W?d1xZi6gd<YmAd%PK0uIr5OnC
zTzxrv?Y`LTVJ%OZFKtEeJvY$2t&HW<U$(I{ZhnL%(r%kpZ`fOvjY6HC(&hST=4cHA
z$v=`&9#QsY1#(+ylg-NABYf#0Pc<QX4$aa}s-f~k$J@ku1>U`&mffCj!@`faxkH@t
zue{&9tqt`%45sghU_PpMX!1JlO~L_suyewT%gA3K?<D(yx8=)B58LE$bc(T%4O)oD
zTcSqkt5f<OP#+Svm&2Y@qOdR#abz107qty(Y59HSbAPkzviywy&AKa#j$k46#(KtU
zQ<BebHyQ-dE1gx1Vt<JIUWD+G#b01Lx={P@|EOSlr;DpMv&%E?!7UxvTWRe6`8F?K
z#;R_cZ@Zvx&wXZU4xGZTj?QX<o5gaBu)YmCH(Uir3P6e7Ui25N-Xz%E<=)>=U>31w
zug{B3t*^?Z^FriuZpBn0jOwcoaqQuRqF#bg!z62UW16XElRd$NC$=}+X8e5%)+PyW
zB*lUYefYQfH-y~6Qwd-6U8A&_L`R;v)OsE!nV74b_KI~GKn#?WT>3ZH30I?u|6*a-
zF8>ajfaQaLT1b``mu8PBfJ-Qv)VggY@;#gz8z#ln{_h=Opzp-T28~0T;Z&T7%w5%~
zMs*aH=^U<vI!Ke^%hpeU#oJZb^Hvo2-|rr8C)nI>JE3ZYo2RCl7mM${pQ6I&+^Wk<
zd}&B~8r(cP4=R_f|7EF^G@M}<^MV`{5_v$@yU9KrG`#)rC0(mvu9&dm$oW~w^~L4Z
zi8{(z1g{FA=GIobcn}e~;9=@+*@BCML*@NQa~OX;U(@A9XB713EV0I5%%(Yo9knd|
zVGkwf2^%^u-vwlQ>e*48^zGj<o+>!gZ+6X=Rea-h8R0`2L`e-j;i}h-UVgNCo+_~N
zgODYY!-e32jG$b7WIpk>3a${?E!t%#+y~)Ak_So)N{+`&1uWgk&EhDsYZ<iJut-1$
z*2CZa%Mr<Z&7p*(WLR&MzPpj)eP+n9$$&S>@KjuVXmfyX8k;~r!^#}4wq%R+{EJ)Y
zq)0cGs+>+{tw~ShraI3WT4W|jserQ;jQfdK;I2o!<7;5p{U)3qn;lXe6BF*fo$#z_
zUTkx(_*$sH9S|A;+YXj3G1wo!+S&?P^xQ{X3$xxI?|K4PEh`dbz_uM8VVHbbchPQw
zQ;~Gv)7`fufp9BR&N_0)m0Oub7TzY6wWKL8-L$-5QU7t&X~zuv+dxki;Qj=Cch`N(
z+I@Xi8*y{AetUw-owDhw<jT=tm>JR$)yd4n-*sgm&-i+8YN67xA|DekQGD<9?FT-H
zN-pOjTUy~kAF?+Z6il1ZLzI)ne3Kw5x0R)vG=y8*c$XI?ZR-q|!7xKjiSYXS_~f1l
zxHx|xdN;Dy-#3Da#HN&hS!{dC!bWf0C7;xH+n<}sNCPhGW*WT(<;SV<aJQ5)BDF8h
zb)B`WDkDMpSRC)#%L=_LN*pv<4q&f}RtZOh!RqRyQrnO_3F=!~CuPN~3cv-*Mlme>
zm*2kY@D<<V;1LT7Hf-+Mn+CvCVT@U&zOlIAB(U#F3ng~{7Yw|ui-~c0e~I6yVMoa}
z;GaKrV#RvB^(<;GxAxi{)%Coky%F1&2xyP!?38P~PQeuQs~?tp{!dm)asZbFtep7W
z*zvP;ySAN?-gDg1ZCE<;yP@Q7|GUoso-@j>jz!7if{?!`?&@qP;qrqY17~yMW;k^j
zVZdi~IVv6_SpcYrGiBpi3VMhs)!}8}){*<_dij1RX93KpVLl7V{TpfK)4RjaAd+}=
z;S3QewR&#Aq%dOGI&u67#J+lCCfI4Q$wQc(M)5#>Q$I24fc0rU9=UWzT7>A3f^2`P
zV8eCDs$#^&6*gy`lb{XT`Lir&Y@^77n{I&}2r%IVu0WAi(Kd0^v}!(+DE2u;zi@Zk
z{10Py3MI+Nh6?YDcQrq5H#dV-CJOGH{B93)Dn`3M=8MxN1g?%QZ*3G-YQq<&CBP$u
z?nsU(OO|~B>?WJDgnJ~tuu23^&IIdWO_TYBd}CnG=aN{L*ddvaeL`R58$KD^`u+Fi
zY5VFeu-$!+xxZSy!>lf&ry74QHe<84ovkYc{P8$h26mztW4*Xe_~=@spoW>&)9SN1
z{`-edKH~Iqx(C(`TcN}s>QtMXo)NXMMx{Ib&F@3;=Ea(ks8^hQ;jYY=ARcodUxSAI
zwH1d!LrwkhiB)N_-7q<lPn3r28Loy72I2j{mt3(K3+<QI+=+$z4bz&-W?Gnl_S%yM
zo$&_2^4WX@A<`OUxby^H3ghAGa%7<G11uW_2~u%FTa8K0Ru>~00|dTnUVNDSc7b30
zdT6~*P9NoGSlsmzjaF0)zc~KPu3)=p<-q(WEN6i^vn$K!1`Rw=xvuGK<y)ePe?beJ
zWOWBO1mUQ-noG{?L4)|u4hovFf6eRb>vnN}l_$a%4-fZz-#aCpt<28IPgggGhu?jT
zxb08`X09Jr*SAxCxKV8AG|hUWp2@up2B&J6Y><GeokL%@Z;mUZ-><%zxR^Bk`6d8f
zUe2_EYiexb%<^dcFzhH%5J_|WlZ+5>UVbg%VUkRFSOS2Rh-hI8{`S=$N!dTsx6v-)
zHak8w^ZgqZrqURs?&ZeD^DV|9w3*D#^Ifm-qfQ1`WArl8SYpS@az5E#+JUCm!TkoW
zzfxQYRO#T#xuXS(o|=Muk+lvDep4EcU%5e3wa9hqR@W8IRl`PQXYa7hlvEBh*yk_g
zF&AX&Rc{y*NqT6cxbmqQ-v$MKP9|ca=w5jqwLb{js1!gakIn}xZgNX0Skry`xV9V(
z#{a1Cx8t?R@~lrp>)p|<enNr+Jyhd~L8QCo_6&{Fgkbz1=+26+N*k@p(X=^uqsMyA
z&~|B-tJbD5vzxq4Q&cR0J_Z4(>Nzwoevw{RnHgH{T1B6RvE<6hTNg(JhunpAbqFrn
zJJCp)1!T}R6xS2rr3*wal~q)TztZ_YzH($5^6&&|V^Y^vUxiB&sr2Y#<`=*5Lx~vC
zjl2s#X6(24ei{3OJq%>W&45GNZ;vj5c&=`4!c<>C^lfVhllH2W@X`SKnoI!|G+%g}
z!OZ5BaJPSYV`+`AvpSh|%R`NUnN|wQ1xK>I^wD&=y57Q12tD|hGX=Av+jy{B=bk)@
zY)_Dwseu}#m7=Ex!rj=eFLwqp%5_7<l0{y#zIgm%t}Hbu3wOX=>5-Np-dhA7#OU9&
zvyqFtK!rJxT<VMD3_O5?-Ol9DOM~wrf)%QT8gF=6f<QX3rapN`g-c6gX^ImsaoWEe
zF?JDF9C34c?AKZ|nA8;Ht15M**2(Q*ytLUIvwHwHuN7yk+v7D|W@3_Fvnlx?km5AX
zP|@ELna)MC?hR;Y<+z(~1zR(S{2lUW)<5w#e`TKAA++yjt^kC;R0Q~@rdkBENedou
zJJuwz%&1c&vWfZg`B`hR|M#_?Kl<Y%{D-ko0(XB0tA*t%{&*BbHQm21fG?aVZ>2+-
z{)^5Jam3n0zD#&oQ(78DOxI>w7Fh`=C-<zv4gcdqft(K&%>vWywV^(@GKq;LXS^HU
zTMAHjUH|vTs&Ifm+*RrWXdu9zGCze4(pZuBTLVtOO@L409&I^H;4hxi3_?~Yx$Z^y
z6q7#PJbucW7No)=jm``MfqEo2`*y<WYA$!j$IH`f;!%e|K*N}hm~-lzGNMEmL%zNE
z7Yb>Q9{1&mdh1X<N+EfFP(^n<r!2*e&#e;cK5jc7mkO{mzMd7_ds)GZC5tR}V5`>&
zC#Vh~=(DN}Gk@(icKVxS>V3f%xpuDB^xG_t6P4VLjdx`x$h(rwA814^g?#Swdf&;F
zAxM25#hSV-g+>GS37*~IfMTklKMt?xPmk5r-1ZZvh^rgO2#XZ@sK}*E4-5kH@kd_X
ziAGu_15pJQp^~K@gT!k%d|}^^9%|58UddD^<CgNL|8=;FbhCY#L=iD>oGMeZPus)j
z4Iy&vmDJFrdLN~*Ssb#O)-OmAR>dILwl5zyr)jo^GHPTX7Rj7I<tu8BlP`qIIms9I
z#wN@*6Ror+7xzT}di2m5F%FBCnc>Ynv1QW&5c~1<`=R*a=Kh`MoWnv-u6uYtlQ<V_
zcClXd*E}cFFR3Y+5(ch!`#V-Tw5wjiJ$(FgR6FUJbT^tqw0;>cQrQ^M7bHCu>v0#c
zw5iO0^j`9DPgEX`yZLseFRRhV$>{2~;-`4te_k+OrOR^XK*u_=bG3s*n#v!y=^$t&
z0V90xhisxVK7>T<5IxazK7T$jnhKc95Bp4Al6*?#HP=o@DH=wQQnYixoZbe38t&4Y
zaWcls>ZsYt4}RG=9Yz711MHP*D(+dN82VY8{(_fh9If0_{Df+aKFJRmNB;(_bkLpO
z`O;TK6PMmmQu#;9i!AH8HEM~7gKAw>042`<2KPI1GOIp+Zu@9_DPS17sGBSdHpzS2
zgRjN=r>m}VIuG9Up>Ws{Yru`<j{RI}ERW}AYH%6?D=85g-f@#6C*oG6s!K@{{nFId
z9UuDHSm&8UyZ4uqop_o%9^}FO$)zkQ{`*YR(S14*ogM#iF&08Dr5EHkzta?B^?{Q3
zh>O$p+>*Py-w9|%q^SuzqG&;&W&NOu<5??F6e*n2BP;~3;)P-D2T@9@KDerh9PfZt
z|KA8=|6=@DTHGjFV~%gwq}5>i?Zxu-wc_n=AaGNzI3RlFK$f%x<z4eod^6Z4gdcB#
zPn84KtNWLy<SYi_hHyBa7C}Et1Ljdx^0>)>njRt4hq!XYJ>5y;P*L;mGZ}?#YT^$O
zq%1reC!Angwf+Ei8LJw$>P5*#Eq|rgH3o_AwUr_#bwMxGDDn*+xjp~BOG;pypF)gV
zNt^a^{A&iM7#?_IL^(so^YU_QkYg>^=>R&+o&zO<^(5A%@I6SQCso=s0=8<j!{^$)
z330xlV2;ayK^nK-I@Z{pR)Wx3ILVKI!1B8~;mSFuE+nccNn@h$wcFzJCd+q3pm-%6
zT8?{X_0Cpnw@1E9rA+PrejvKQMyI=Ft-Lm6jE1Gz9g`<wy2IBFe?myOv|EI*c$R{c
zHL5b&5iZ)@XFtAhvN0!u0J5*ayi^1mn<2@OzrUB!5=V6m`@iV<Kbp?+p~?4c;~NbU
z8|4U<7Le|45D96N7!A_h0yiMSC;<gVh=g=X!{`?2FEJXFl5UWG_IvT{FW8IwzOMT`
z&d+fix-4#=MJX(mvcDG(WLlj!=BqJQotgeXXa2_-Tdsd4y2N*FEQ2j{Dj=vNz*H9W
zl3B@$Xc*DHsl51bX6VOEjvG4ju{rMO{N;@kn2&;8)P6LU3ADxJNnpxZkIuU69aflT
z`yLw&Sp+6CJLO{1I0b#+8gSm2)0*qnu8QwFex|Ro|87r`@~1tI&I@ETw*4LPZJVhS
zPa)KX;Oz=Wc!|M^bD02-`*$&|A%Py-<r8BT^|_er)v!C>%cO?RFyo}OQ)+BxixnRp
z&nB~~GV40W4k{HgQ0X7`^{p{21ivp_^;oKn%NF>=k?(3jgc=;$+S=0MUC4V%F-x7o
ze4?V%`?gmaYku#+Pyiu!SG*L6j^aTes^E;*<7*8#(=V9`Ix#t@I%z_Q(u@1g4lzLG
z;lb;*o;44tv&n|muZ_Fm%0%STv(vDaG_$st+<225Ln$Q~yVzDwqr0_Va{(SFEbYes
z2XHxXP_Vw3Cqeb!MaMkOp6rL4xA>zRk&yfYKK4>Gi_oGDCJ>0CZ+1Aqk%dxhX^w<)
zQHN%4)ACBf_f$)xf^8E7)8oSvP2}J#clELsCXi9aW*65w-TSDbeZDWZ4i9&j`0Tap
zx$|BnLIQ~AS5`RCwd&SXw;~9p&0_@j0Z0Bp9V_v0s~@(0^VY;#$g7~N!T^f%tdE7~
zr;ZjLlDkgLMp6D9OW%z<i8C6;%n{jHh?5tT`GXr>t&h3#bn`>g2d_4^x5#AoeG_^P
zK*8BsXN2Y<>4mnaVdPL?y3I0!I#&CkUr*xmYNye%WaDK<&sECFalevS`szapfWmmL
zoITugHPn)sy_uBp#o2*FU4_U@=|6nn(fEH!JLNst@6zfFw(Z#L^7F!}%Rf2nGF=p+
zqjXd?Tue#Ne#ySiy1I^<vwL)YhQ4-4FHQeIl_CNkz?a#nSi!;aR}b`3wXe~S@&!L>
zh{$~3<~Pg#Ur*1ryA+M~C~kYJDX?Ww%7hDE6d%w?0j11SE8A?&cNa?}_A34P<K1W?
zD9nuFulk`7<C?W9<#u-Tq7kh7#R~HQQR%c^molw^jZYywu{r?NW7sm@a;Yf#?j_&g
z15dT(&NLRZx{u1P%6}6nr^>oJiF5$=P}R>11JVo612u|rcOn0_H#UX_fTFFjcij00
zk?-lU#372R8aaLYZJZx<GR+$T_TpAnUFr@(-WISoG1Z{RI^)A9{A0k}p#qLXKX(6!
zCJeZB1e_k+tu@iTPVnn#=dG;_7&Ljs&-9jqI!%`t9>+#<ULlD_OMqaE1#&nIcG*z6
z>UWFShNIMCcm}-O+7*=QK+BUF<1yFPc)C*6d})pKVJ5pmrq%vZtV#=gJ^g|EuV;}4
z!36o-5nYEYmCm*l6&cC%YVT{c59!9{D=*yf@-6;^@Yw50-%*Y$5Pxb*H8ma4RtkzQ
zgXMv(r|VvI?T{HuH>Qd8C0of+2yAq6c=NxF6MA}ynUQ7rGk)cX0}(tGx@J7+Q=e~0
zt9}b&Jc~aXq8}pfX<##I&%mXAM{bu*g*B<C=jDD!-H*nI!qD6Yk0$d~^(51hJ=?wl
zS64kZm%vU`3!HxW;6pUe7E<OYMw^%%q8&$gO))|w@6A0{YK*~G1w4s!AC(-U7%g#P
zh}n$e2&@qNY)qR#3fkwmZ<C}J`DOXDRpzi>EN)erWt#vLgouu(`daWvJ2wDgXy6ls
zJKjoLyD7Pw`qWHDh-k{`N>{cIWwL8?Y52+5i_8#LL6kpmq*}pyQoXMK%EKVKHnVFu
z?smA(SR`QZ-N;E7!h!zJTF>_JK3!gui_wVZM8FplLc@0esa^QN<vUYZQ4vAysBus<
zR<-sTPf7YaEa8%`lk`(Lfz#F>_n;ZVbXYNP-qUj_<xfnw;|c^uhBzc~@SITljCulg
z#&ES$6xGs<kJ+ZJ33wvw)BW+5(_2x*rQc;lYo%wK0{XK0Q9(I+``R-=-mWGGSLXH~
zrms#m==vYPLZenGurjlS3R9xGx5yNoN;IqJ_mDB)gt(IKg(i9p+MXudOiP(bZ{r4O
z(-2~Fe1N2`jkf{sAMW?Jk4JvrA0KTlTn@D8$zH6)cVwq-Uu_eDd3cdNfp&F2DY2U8
z$KCv8v$30{YxrqbBEvs@nzDz2zBH7*%}vcvMWmQ5>^7!OvT{B_a~y@NF8imG>NQK1
zQeL_$Lc54Yu?$)#ZUsxQu40lsQ=*t43d$c_``E0`)~%hVzeHvlLBgT5@+8)Z2#$Wr
zt|{Ko2JL_xv%FN@lhik3buT`aD<R$9fEfVUkW0*{U-fX;0X(3AyNv-L!DdpT&2%e4
zix;`uU`r9)_ij;RYKzF~?|p97t=m7;(~F0JNxR$s>^du6m3GXV;dj3s5MWY}gKGpQ
zObTtan(<&Xsvz3PmbImjxZ{1x?7bzKs_dKJUe`y_x<;wmHbU<OObg3eRcCbN!T@aL
z!d}QLrc@B6+%G)cpX+?2H<t7_@+5z|(Yw<DB>G!IIClIvP_VEs6$2xM8kA!O!iVkm
zU5VH(uhdl``fYqP)%4p6HMBBOF-I{wZ77r?uatxKZ9J!7{X#MtzjPdh<AkqPTMcqE
z%_-y&C6_X6tIQPmx3#xscs(sN2b83_PXC_wD5aYv#5%rr1s>e}gTxA5`9d*~{Ak;j
z;y+~$<o@y>2|5iW`+k|#EcB7O8vY)rECn&;>8qqMj}^{xnz$@z>U1d_QA+0zoo?K2
z5XXdl)#<YzM7$+o-EuLRF&O?c?x!fKsmC*y!TG%4$HsY~-^~E}2yJj?2mEjjSh86p
zq1RSUdlfT0N#1Q>n8v$n{%~@+#fxk;C@uQ_BEG7?PA#(6`!p9v@i3zE&_=<Ej2%<j
z6{5MA_QPhi-BKF|jKD^jNa!#iFwZ^ZzmdCpvK8rzs1UzBt%QM_D7_B4q(M1oGU4WK
zU4>whbbGU`EGBF=nib6nl<)7)BPgt_n3R0vDVPlwu{7#cRT8p<6WJJ?x#JAA=zTxX
zFh!b7TpF)UG1IPMe76PCHj$Ak+77D4y>X#)mPQpcZ+(QeXXsUYX7$sF=Q$zR#i_L#
z{_kst*<Zg|oqzf9r}JBgKBFXWxMHqaEmTMT#b}SVb=c>|uV|)p0K1QeU*P?-3@i#&
zYHr-nUTsDC!-J<CH=FmFetmWJ9L7r2$|hb$4K;QnxWhlU)!U=#XS`JOHo)WpD}_%x
zehW36NYOg$`vrfLz3uox`mTBoyVxPA)n$Qt_K6J3*y+{P)#>)Z((#!H16yd=(bP+s
z1HZ#{8Il@IaU1LN7XWfw{F^L(9Vqb{9JP!}!v?9#tEqB$#tb_;!2aX&UVA-Vqbf_f
zg|yIj{bCYoQfC(I!Qi5j&=_FT()RU*T*wR-KY5N3MPx=F1f-GWWY2!T{9yqJWxmdH
z7l2vR61nk7*U_&0Ag3VCzteMm)A%s%hz3pjOoB8t=_ASFfRG*LoT7e%|8N-dplN9g
z3M+m^llQ-W*R7ECKyGB(0;EE&JOcv*(>HyLZ7JI3Oy2~_QBn0e)*eu0Yq5bq5Uko3
z3x6huEEWbv=%0^GJ^#P|&lsSY@rz4INI+yLo66zidZzB$-fthuE=li2!SPFb|Gr~N
zUF8S?mc%X;$A+7oPo**)c2Jdv3IwN8$glgmScUDh-=iFi(Di3`pqzyU(BB4ETMEn^
zzr~yV4fuBxbyZU$Cwx;XsSMPcyr=<#WJJBK?W;7fYyDVdblv>RqZfJbtcA4hK@mSI
zZ@WTOi-WuGVEth$Z7nyO3orlL1GED@5J})uG`~9!O{dkW*POl@Lh9h!1AQ4hZ<l@W
zc6UTWv&*y=O&O}Hgy&jOQkrA2)35}>d~0+p!{2{gHRK=_#>VO@b9?!L^RCxzhDnX+
zt)giGrE-VaczSnO5<xl1My!~swm3VL<j4M+3%#N&z1eHL%iLqhkD+PN`hx@P5yQY#
zK{4KlyqU(x((IC9$%%1%MuT>NbAZW=504?A1wj~z9x+Fs-Br<Of?8PA<lXV{{((2L
zpy?cHk7ozgXMj?c;BY$|K`I^#fRywUtZ~K+y$`=vF){@8>3PV%;|^^85BB4)YvTwf
zZs)+?6pW5DC|)%b6MEpAhif)!UI|P#C#TTNC2P*w#7!MW-H7EdW^4;oAbHn#t?5xR
z<1mH!{esx;d<#|vK0b;881w^{C!Za05?ELEYz@{js^-xO*<GD-NYw5s_~k6vI!(LJ
z<PU!ik}*?J62vMGEc1PoQyLSlAUt*~Uw}T>{lbNNWN_wvqk8wzp`{zX4BT<DxirB*
z+CxFiO>lj?(72yr#!#a}^A-Jg6K5Q}h_%@#S~rtRAuzoiF`v4X%LKBhqJgyyPvK1O
z!M;x7dLm*HN>3dHbomtbB#Xgx7v5}RQBQ$_!NMoR`fPbNos1NppEULJmpQqc<>;Z?
zo|O=Kl<b~gTeOv&6&o`X(A`>+KG_<Z>H|mmfNLuc9cc+`E7W<+--RC;;7c>%>X@k8
zS&GayY)lr7Z6Vb={pXj=M<*qQyv<GvITO{7#Vm`}^L3x+=s0~fCy3xl;LL-+WW;0f
z{ZApwey0A<kq8r3ST!ewSid434h#yd(+UD;wFd5&Zla<t$vn`2&zc;x$cIQ#;tA;j
z$~Wl-b|NKcF)9J)1vSX-jcQS(h3BpRQ4}>e4lc*S;^bzk4LP;f2n|z~?;_xD3YKAn
zHYad22$m|Zv{nX`b1}b>%Vc!4_7bR#akj8s=+r0VuW=yMU_1v^8gZH|H{@B_806zK
zx)ZyUPh8;1#b*FDDj&X6+pfa&R~a0w@GLZFvCrVT*oA+j!hMFCou5KU2|DHEpo;u<
z2ABR{S}~c40fDbtcu1jfO$j9ls8Ji5igIy1o2Pyri<W5lIQRe&Dn(dj9isRLhM&X5
zW<?e0X;-Xx7BWUbo=Z*(g}i0ZVtE*O9eB)~)Y(KGd=D(WAqXS>+M&}^sR75^o}w^%
z1BMvzP3;D?zaBdce1eY$Nz|Qb=6W+4fb7quOYiaQ3W@ug_@4e!QilHA-<ZHquZWD#
z6NiXL)xfBM1fj<Z-%&spS)vmpUnr{^7XOw6^(+0fH-33hZ(d>P9J6CX<=mEC$_|N+
zDWa9&s027T8484WW78QeEl4$Hr-FP`OF5O<S1Ud>bNjmw&%dBRrur+H??=ihrE!+;
zW8uDeVz4ZY#osTdET<G}_#$~GivjhPI1y|Avgf#I;aTJqKfXm|^528Hn~SBTz_v6>
z?toEUVrfpLivMis@(35prSIymQldjG<jV+?c?p5C$0<dLWrw~m_u-X5ZN~?n5#-SA
zF_QOiMf&S&$sK~EMDf02lSm&9(AbtBuh-sxJ}B3P&5*~VzaG)+yQJIQWPG_VBMlrX
z&5dQIvF@YkB*syj2rvfjuTG<~Wkx3(Dp<EZdr|8g27zwde!NV<63aIn0U-hw<6XPo
z;TZw6Av#)vG2Zgb61bcCic27!!BFteoWrYwruR1&_wIWKD~nH7mLE@Q{<^&-S9q&#
z)<^}tNmC%`^#%aZxE8GrsjTehY@7w_T#j#J(mJFBg;O$^{#ZF)dS5n1Out`-#6(jX
zDCdob&mb}e3)E2=Q@y`!1<%#NuB=YnsNX=Op?3Yoq%N^Mhr3!y9WfR^c5f#O&xVNm
z9iQNvaF~8-b$U2TDqe}mO0M&(R~&yYh`T2@nTd7F2Y%E|pXT2|w+Ne{Pw*Lh3}Rx&
z9ij&@k*7K@Fzwd8-U^`ACB_QVmVX0(tlK1;b>E1gJ`xFt7G@d!7>6#7Co$TO1F%f}
zX;3M(et6O_M!fN`A0l9{By~hFlQSV;wF<*Jb$eOWT74FLpU{C7Ref=}`q-u7c36%|
ziR;~<5kg}mT2O3+7r7l^#>;<>fVCu}Si<@YCqAEY;WqAbHt2-^9$#~P?&nt{JDTUU
z+~87g#pc-DYWm#BaH!yLKyn=$0u;#Ts`k*LpgnmZEFWm>Hxe1@Wl%H2_pbA@)pq=w
z%Vw3!3`)#QW#U3?<w`Iyh(I`~HBxKoC?WEFhpub6Z|3=9BZ{(?V-M+!x$piNRzt&!
z5%ghvSnnTmRMIdrRSuLl7v}#_cs<0=QkeEf?4l=8H>Ov=jw=8f=Am)80xWIWics+d
zqlVRI%8fxect?T6(3$VFjvpgIj=x^B+7b=x2BbU2Oeb_0i$O#bMV<bk+*B!tXf^|a
zuTM^X_YBM)|F+D|i6sbasu(gR^5gIS$<HJiRzKP$t^Hc5b?KpkGZ1@LpIv)^Q0XMq
zAR@Hi!@R~NGYC^Yk9o>>fv;Yv81;A5ud0OaZ(a~|0O!`~Z1hEe*r%2>T^?KGnY|1A
zX^S2ToCW&G#3U?BInQObaPqgHzv-`A4n{I3{W<_ZjMVS&Kq~=*@URT!*_zD6`+uSb
z5l8XIS)&W??%k!9QVcFy;47Efbn>1c^1nXnVpWqq>bY}EQ;O*TL0ob#x)8}UPegOL
z<jTYNN$YAb*O9)ow$0$%n+o#aF^Dm-elPUq&vP+;r&LG6E&Ya{#-()r8w>6v%VG2H
zX&DnK7<23YV#J+Oh+}Jt_2|GTU0OekIja2;D63;mx0kpNq&aaq)`<l}V(oHH`UC_R
zNAVQdQ0m0mVI=|Z4oPX5`{RzgH`_-wQiB_UDm=Cr#9_;!nHI)CRoyz6TBg=b8w+(^
zPq~2qqkKk)t>;NMu##+UL<X^+>*m?4&1@LyX#8a=50W$Zf)!%zfiUI8#Wa#^B6n{E
zh@*He`1>q7KG=?g+a!Ap4z+Ue%GF4ckOfK%{P1P(f&gKZ%vL-LB2^1@-?Gh89dsvJ
zS_YQx2`I(S>C5T7L=VsoW<z$`AGqCkM*=#jh;fg$wSv8TmJtU>Ispl#oc4&k&>S3A
zKE|Vw1^r#T0itvL$AT}*pUnEk4b4aCwP<wc79UHLMB`Hh&6B#s*5mHY;H4zSX#=ri
zS_=uT<^U$=)(hRYH0}o{PO^HOq@$7Kuy~%tKK5nZ*-q&w$f&R)F@uTSf0n>L(^3B$
zk~S|1g`ctIWw}55<d2VPyQLUnI+bQOfln)y9K^CP6C{J#@BjM=ee~-osq|a^rC){`
z9DP${jChUYy7g8ym6>ad2TfE8|4P{|cKNQY(g6sTJcC(L>u_yp@Xfwo#M1x5DRu)f
z{YFl{?1d?07NwPhob}TR`Kzl^Fuo(cxSJlvt<OU!R})`vbE+Nfn&P){dks&M=9(k5
z#Y_7Or6(O!6E36NFvj2kSq!PRzvRF8QHH}~2OlRWAmC}#Z*SE5fWTi1hE`HbS-NVv
z0FmIKZ_DI%vaT*So+USHZ0N#@pkQ@E5Iup%wrE<dcYNYkF}gH*e{ga05(uK8EjWm~
z<6f9I=xC@G%@_%hcRG1(G1hpbiK~%pB~;6WtFOCftx>?)6l~nN6Z4sgU=uadzjjhl
zYN19z$zEs!QjU{anRNbp1L=<MlmjgSsudh~e3La)Um$;6_Xa%t#mzSZ1~bKq=%<;y
z_-r|#AT+D&!dLc7={S4>o~iZTYT0ks0|R3VaU71%Y!neWCgR!INzSt*j4DvRHJHQ?
zO`8|W;pT|V@Q4MRx@8k^mo;HiH>hLeLseSh;TqkEof)Ksx2i14QNG*|{w@hnY?R&8
z#bTgJ*=5@Gl{WwOx=55mg|taBD{^+*VcItM{%m{W`br`z0I?EuNlNzH93v}T4UI{R
zh4OR!hx#?C9)LlEShc(cu~ZFl!W1+ZNfA?r-iz4{wi7j&VD(6g#%dA@|F6v=gwEd^
zSL4^$@2Loi4HtucAPn`egr5VA_`Z5yGrk!vxKW*GER+V9n7R2BZ)ejqJ<b&+`RS4z
zAX)wbOD`o{H6;=cQO_UtYrzYQKNZvbsoOPyUq{UYrW-RH*K%i%C{4Ap`aaw$yWhbd
z%YkQCr^RNgtQTRWT{mjHgY*5QH906b3@c?KLhWGJ*<dWjZark=;7ARq2rH=%F8KO)
zWOYw4QQ=2;!%NK>Mg|;_5^>*ULiV?V8&?~)@J1LjgD-rH{>bU|b(Q2_dB!cD7vhi<
z)3m0;KjVjm91rI!4;Lpv$CASae|m4P&od07i#ZLwR8bw6tTf;&+Q)3>^h%lJHo3~{
z&CANA4J5a}F1+}UPPrsN1}jf!eRgvbwblsSZjxaL$dx-Z@JH_Tnqe@sn$ijx36tQn
z`m~F3-9;{8)*t9|>48@xHOYxg<)6KLg9s<mngeCTN~UA;R8%VW;B14R8>On~qWs^Z
zl`7nHSnVy_?c~-Kwn{(LU-P~!q_iT|z!6>Wdnq{)C*BK*Rsb*qc5G1$Mfx1dmG!av
zkI!Gm9H7-yO7$<R);_Bl2!W0+zC=&Q$=xw9=T#nJe(G~44I7`q$qK7e@pi-fyaH7V
zDDF%<rnoTtesHLUrP=mI#=sUw{N3^F{sn*G(!GD+O#{&SuyNgu+mAB<iGq<YWQvlR
zTQs+pBZsDe@-d_AZ=(whjB@U3-YJ8P(d~!+f`Xm~<T%W|+E403zTj8qHZ%@TM~c6J
zj4rU|8*jH<L0j6HdZ#eKKtdA&ZpiJx`+q}PZcp23bjRowSHObaZ5v<bM7X+Hxq2VD
zBzG)Pi=^Si7&^_%M9TE>ioHn7u$idYtRZC5Q)SEtmfeAj23w<|S8u>nHf#nW%=RJ&
z`*)qY>$N&o^j5++a%|YOj#9We76q!7guP0{vvTsXYq&UIMVHlR1j5bI{(R;9MFELW
z@o{3A7M4f|RoZ)%z~v%6k24>oJ*787*o{QR@d6+a?R0>HY(6GLBh&AaBgr`Fv)+nd
zOUNRyufxbohp)y_oNAkn_WArdyQ)lX1h?bPqS3Nd*mX4CP6XY|NdxnL9o+M{D+8Mc
zY#^NuXjD8isAs||`!*&A<_PDbOf>8FfLS5HB?x;CdWAid46@bLSV)htvQF%Q^Z_Gy
zk_CQYs<BlhrtcZw2p_5mzTG$_TY2!8)&iJ(&WJz}>$})HMRuj!;1mwN|M_>W#(?`9
zGTGs0A?>dP$v1TGA~~<!c4i3}{hyYY)M6~RqR~%&d&ss2x87X%xpmx}9<JQ~zVioW
zXSmB!O*aF$o3+QLBcmwvOl{0&Xke&HGR9~Qh^K{(w?lCT?LYN%YRZu~NS4+%07t2F
zn*NPtuq(}Dlz3`uz}^+XT{=QAhXm(EEmAOsR#_-@uZq+~<m{}G_^8yfgL>0$7e<=b
zO+AzTyb3E7e+Q{>wbK^QCE9%IQF#sm*_X3Aeqi+_!eyi-Vq{={yZ7J0580)aSA#<W
z<~&GewnA0vYb8qZGH^^P6G2`CK%In_XQN^<!ERga^M!kjfZzA7DfUL1<j(V6N+3Qs
z@b6{Ow_J;u>0s?CwbXzS^x!s*D#Ae2!T0-{f4g`rd=sA^;S!3z>0Vj)SOl50EieR@
zC+E=3_<mTpC4>I^`yZugwVcf;*W1nQ?V+OGjR4W7k%{a|>!CXGs3c8n3jlw}YyI)U
z(AmBJU*c~(ySfZ5if5Lx1LK+G#pz~-6JJ@8obOfPFD<1;PtOAb0*xEy{1};Q(ie#8
z`JtUo6GZNcc{}E+q$3?MVL`ArUHXDsYkC{-=HdC^*5<J3eOgdeOflA9#M4CWS3B!n
z?C}NenL10ZSmkLli|k{8lECqj3X${;1TujW&xb=BdEmYnZ$04LOb}iG^KG$QB*`(f
zLWHTCZCy#gqimN`FsW27WoX*mrf4|pnts!=EXPYFD1x1jAymlFWP_AEFLQ%AqbV!=
zRqgb}&s5zY-DW15ANla?{!ceH*pbOhe<U=`UR<uJs*;Q1u&`~e0<u52$AGgdwLn`P
zyv@Y%y&tlX<Jn_F9k*j0_hZ1-Wf3p$Y3usp?Y?V4`017bB#cmjfWHvN3|(50zP-Qf
zco@ReTC{SJDm^x6frpG>ClF$P8B$&t^;Dx~E6h!BTamp#PFT4+&ITU-6<xMO^;`vx
zEEqP_R4h!M3L?S)D-K)HgNv3sDNVB=c;y#bjZ@U9xbL<%;HhA}@OuP?w|g`ZaAx#~
z9~TFUxsp93{=$I4N_adkU<hO}f%V)<V3jKzG;KTHLlERsR_iUrL156Qf6@>@@CHhi
zBsW_?n&g>;xj-sxpkO&Bofpd3Z{>*TjlmI{dxjkL{W1WX&}yl8VW3@`p~1I@P-UOl
z{ex#uFN&}!Dt2S|LqbB#PVnp*eRH)K$4kWtzE}(zv%JtIx`XeK!e6%VAH0^@6T!wo
zFv3plSbw5okC!YTZf@2xd5B+)@m>vWB}@j%cY8EGoH;a`ZL!riFBx`tJ_$7>2dAl4
z6G!s@$Oi$C4o4bbTJJ@md-2@%!o6sS2qzOQyAa8WW)81%ovKgpR86_wpb6Ayb9d-;
z((-}t>?VVz6A^afAE!9~<c3!^>DO1D-yw~3p_rvXYy}pR#HY55Z=}*830SDXxI*K2
zy-+a~wLAa6eo;cilei-c0vHBuZHLVHHBIOO#~~1}i+i?{YpC^Jp!B~Lk=H(3{_brR
z)R*`6&&~V77IF6R?lMizn`~EI3V#<#g<B`?CyO2#FFp4<SF4bglcpjb&)S$L9xFI^
z6~@kq8_s=(@(^8b^+2qN`NBod;7IatM(aHoH}Q?Dly?u$;!1o5rULmC&q-cjmbmuH
zCKou7uXf-tIk%ifk%N0PCi7MhQ;F<)dO+rkVdT{)B|4Z+Bhxa#zp?af97o~Q`Q5C~
z0p6s~6UJ9)>xm`4jJX=Fp$2M>3c;I~*;O?C8EsP16+@<;Pf?XNut3%wnc&-#yA>df
zcL?Pi@WL_v`KPLnQZR_Th0+h6zPsN-Bf!w5?46X$19~L{v-A|Ook+Ev9QIh$&SZxu
zyRs$e5)(tjERfwHvNG+>H>beO*2=^7!dm8cytGXYNYyTiYgd3*iQa0pQyh7Eb;5j`
z)<}<;G5<|QJu3>3NhRmQqQyUmB+cdtFfph>qDUG`o^WDKA<BLwmwb5a7g>!5=r(JG
zKHwC`yFj5fpOhS_p^C(62T)s3%v)wCSYA4b<Tio!d!}Cv!UdyS8fJ|Bpw=<Ki&IJx
z?_J$URTM-ZdM(0LwV!kmPl@5!h(x$7DD_pW^%5)0UiKXL@AN+zVBy3CrJ<@tqq~K+
zYBokhxEs{?;=qPv&&G@~{}izFgFSuLP}wq?iP&&RHyiD)8%nT7hFt#J9B~bTP5ulN
z$Kd`*^TKV%(g^+DRADWrWvOhEPKPYRv1=kBbe<UrLD{&1=b*{viaTt`-(>h4@#cBW
zBE!2wAxh)g{19^0<INW)M4@&>qc5-%ZPOcEgoR6AS>O@;4Aag3PY2L|F-5|@Fs*h$
zHc1-43i`qk{CmEH5QOUaq#Zba7(LX$v()r`_G2Z6g`)HO`O4BjZ*@L|fr1W<vT6BN
zX7=;ub9G=EGtAh-v2M&{a9Ayg(vfDXLL6MI2E|d(tpFg#lBE4OI1w-@6{6kpvOoJc
z44jqN@5r3(nISmt0PEs@WK~1i{!=U^Ar_V68!F=H;8r@l+5O0ymbIHpw;RUmhiqLN
z6$AbspNDERB1H}c2K>mv-u6{a-g*wAjiAGAvLEuoMIWv9)b`_0z7X$U#|K-4{(0V8
zCQ$vyz+aL*sL#TS=;#EkHskW~^whR_)Ea%*zjW*^b#O7{0~o)Pe|-N!;6&^|$p$WJ
zWmo)k^x@yB&q2zr6z?e)6a4Q<A%jWM{?Jy@R?#Uc*;LI#bV;J2luSrBS5ICK9pAF#
zV+3HTb9)BHwj6#v9hf>2b&f=ZaR|H*osW0fDtnImwBP&E{oe)IO6yHe%hH-<gNx)l
zwt1hn7&RVUZvZyXzH3du&qmd3`dm+5Mf8|r(v+x(z#c@4&72`w4gE;lvw6m*!^V{V
zx0;q<|B;iADg0|A9ww6^zE6Ljx^pvG;Bj{f*1X31AJzZ9c9pC;i{Mf<!m%0rG<ovA
zS*=xwqFB<zaY&!~-@cH7WWmPUJ4$Wjl>DgV-#EVjQ3Ig$I?3hzgrbnff!T`54HS?1
zC?!U>Aju8oHjb#l#bque%_l3Ya~ZB%h<E$n^5UYU7CcN*C`8(F=ZelEB^&5jSlhna
z{+(>470Cx&ogGI#yKfkMQ;P)>Uqhw%d@<FJ-;sYX9+fo6x7(eG8F$5(e%D>~hZ2}p
z6#bS3xV;9z`<Lm4SQ0s4G2O#RZU!cCx80h4d&)QjD!CXIno_gT8be_1?%q8pr79D<
zt+2zv65CIZHE-3=nHQ78p^XVW+s~zhiv}Cfq@75)oF@C-@F<Zodc$|{KV_BctwGXi
zt)K7-2@BXTIUJms#VbfBWTk9>pV`T_-JAhT3{F%LaWF_tMjqmGu?L0sR6r`Pu&0vA
zRQ{*ZyBxRJj&0+lMT^>zcP4@)hLh#Q^2g^OTq~TM!qy7c#@;D1jI&AdDVd+&KxKks
zjY~R$Uhro@`L2X{m2L#nvx84Lsb1-of}J*~7|Ih1B3%9izeb{1<GvWVkD6tKvP=Mf
z`jUpbhGum-ZvCf{5Y7}Cq7>~ytIw7aylAA;T+B1pr$>JPEjcAfDE3RxdM6mmz+?Fh
z*ILyFR|rD^CHq&>%dnPh)cpo8@cuq%#;RE;?^T=251M*TCO$?%elImRnyln7cJ9>H
zxRFF0ToYDD*0<Oc31#;3wEy-B(}k1_!B&V7B0hGfeYU&;oEEvECXYTBbpZdK#_Y9#
zpRJtTjTd}M#0)++mKJ~7cQ+j{m90ZVLtfuR<uJ3|J3G5L##94+bvig_XwxJZE3nvu
zBb?uJpX(<W?=hB~&n6qgX}8f``DJA2;5*@ZPf^=m7($#&yZ&|IAF8tA48Xx~(9wkF
z^CJW?25+oa082fLi7GP43G2o)@`($EgT5NLx!i;o5i-lIu9kD@a2r06zP00#qr?EM
zjgrfkE!YfZn_Ah~-?pv-#hzI2^!5}=M&hDr*gukDG2r!na&M$tX?=gA1>Ex<-k)u4
znfjW4vm*9$!@#JQzLzbRz~xGa-{n8oe4dQZVH2~EfA_NY*KQBwd1)o=l=01Pp<0zp
zW#94o7Uan)-~9+|Hf)G0PS$NO&b${P!(MbJ=o2}TD=sObdSH+P+pM}Pa!q+JoSb_q
zr&`AG<Fa7s>l~>lc@~P1TWR{dRP2;vERgZBRt=`+r_cbu@x{;l3N$LfQ2vc{LklnY
z?>AVr$jhB~gWy{n`J`QG>DJHm%*C$WacV0<pef^N41pS&a34GWTDXE1^h;)aza}-#
z^yjA_1c)a7RX!p#67{RPKVENA1VWpy4gWaZcg%@rvo6&~v^ZOp^Vy^|L3HXq_7#=p
zu9I_RB*EWpPajw(0AotHS^Dig`wXL;Vy1kGiPTdm-sa>0j5X}r77f&j;4VZvsdZxm
zm29<8IHWu7CY5KUvKfgI{_;b_-LAk!3m*hxS%-k=^*r64xwI30x)x2~c`tl>i8V%v
zba;Q|d4F~~F!h{+b!$ZJlWK7)JGl<XlTimr&OjhpztsqI{f+*M8Me$*ZQ+|fCYaiK
z;QnZBpyU4hCLz0>%Vbnn1ZtqzNxRBq=lX?Q>>mw<mB_!dp!=Zo>~CXE&{b8&;+W_e
zOud_!-Zyq~3EEG1uDP0a*3bNIDrLRNR;0cyuS~+Sk?K5jY0`^Jhp1)Eb?`VmQz5Qp
zDZ~UR>?YzgC*tE5^gb}qkXyrhK3ho9CZtIg(>ow8Yd+DbG-69MO+0<YtETGs-?*lb
zBJP43hpr-7DG~?((*p%FnijuuDv<nwl=Y_GOP;EP(RZUSSLoeM^y){rSe=Hnip*0~
zm%a5v=ma6m)!KVE?F-K?{Sxke^N`+fDNkXpu<f;hglw4tk45Rl%K`p)#ZUPI(qJE0
zcTbm0NL{viT$3y@Mt2&yIcyKc2?_Ed(GY@=@UtF7OY~So0>zVocS9X_E6;!{9h=>o
z+4B#^{sDuhb*5E<p!xt4<)ZNy>IT{+wK{Ux=LD6yCr-+@3sT-UU4Kb;=NI`hT5v+a
z*a7kC?Pm4y-Cze3(N3oB1Cuhl0_VJfaPl3y|A-h6YLQ{q1SK$TWiRTI2vv6Fx?~Kc
zT=6q34Rf>s$?L7ZZJYx}hm88aeR$Wlrz`5rSYo4O9p0J4(}d+wkIzh~(Dr&nA4k8-
zdc>_(2V`IQu5n+R1wTf^h8|X6tkO6$1EQ~fn8vrqp(cpsti9}-E+2DQQ0`5aB1w)?
z87bd;Tz*{P0~t3n!BwI?z;G9FB?%?!`~{Yx3=1`Y^Q4e16ckrY0p4dUBqg*afU5=n
zWqQd+UEnflYrLgt5CMS^Lt)zGpWR&LK87U;<Ta#;_jYpD=elX)$+@IEhV(nJy3Vr&
zwf;Xf&|cMr^qupT8Pl#KtB!3QwPP}wBlj`(=9$y({Z%u1RDUD|3Sw0u%a)Oa9s#J-
zYz!y09c3<adxohrzVmDzZdRS3ZP=V)x&QtB#@>$uhmBO#a^XZm%OuGZ2O*BjVx=wQ
zc>@B)DCJV0%M}){(qk*dMw3MyyNgCDE_R6f`mL$3)?|M3YgwvX<I_5A9sU`HRm5|?
zCKlI&0S~H%A~YD4DfKx{Yf@WP0MeqRXyQ&Gf%mP^f-amfGq5}DyB9gv{ApO$xS9IO
zR7vm#w$d=dXxf%xbld3xAD+S%vw=CAS3&k}l$(e<gF<1}DK|9M7&^kFJgs~=s#5ae
z+c<~)=G{kganh=p)CO)IZ|6x{Gn=20;3m2Yzkz|qBGJRBo+p0rZ_jmx0H<%GXwwGo
zHl|s7)`RQhv}>Ni%7l_3>)$oyystcF3tC!wmbO8Zm%*Gh6?L}Bst_m>C^y=5YX}M4
z3_YCQo`~asv=sz;e3nvehfMHRePgid)|S6#4HM>ov|YeZsDYZp^E<b*y4w6|Xhn+k
zd7wvc|2u&&F%T?=(=iFiv!KUH%6L0E2a1vKuH9^<n$xFB;`HdH)PTitxc{ZJr+Ql(
zZbZ#)Wul`n#5CDZh>>CT`=2uRk8~H}#%?y3Q&8X*R;rimDrm6s_EKZ}4EM8TZfI(g
zzzf3}W?>Xm@2H&Ck#CN=()kX$<|C;QonZ<EEc#*(vH<kcD_TIj2+M#Z4is89=Mg$j
zvKDImLtm&pc-7JHdrW5rNs~ba1by;Ll#`B~|KP$}_Cob-zA6XWA!Bn+>x%L-4#k@<
zAoo=dSz~n3)W>m`%wY_6rokDjf0u0F2fg$FhpGv9G7y&@WW#xewCChrKX$f;9BLWR
z3w@xoa;q2S?gLUW+V|LyPm&&Kv{KA^UQYEPs5)6db1nx#<w&Dxejd~(Z?1Or>|yO8
z=;k7mG75NTeDJ?M2@*H2I&Dq4%|6x#y)4Ei_tGeNhOcBOr(+WP+R3bu-pa-|0xvCE
z>{>e$goPcJ^xm)xiy($vo}hKbrzooDw?`GW?*KTk9az6=iJFA92ukCVI%@0lYH7h&
zIs$H70jUGh_oTm(W7#u4&6sCE&UR=4m6uECxq9c?VK!5<VVj{g3WA9LOD<pMD$vhT
zl&mt)uWxGn5Ju%9_>RKhCsTncLYh2=BYDTRh^QBqrb|R%igWAb)wvwz6F@RsiXZ%p
z8Q)L|&^}%a-1<$cA4MP9U`bG~x@2c}meII9%o=V-HI!?N1X=Ox+x!WU!nDO)Z61t>
z-nruElzt(y*Li{KWyn#R{x#e0GOERGxtB8P=S?u$<fp)EG+Y!p6;-)L8Alkpce)YK
zU*RsAArMIYZFJ~tV;k9VKhWV@O;1{~?(z)@xb$olAzhY2Z1V|x1mDm$Ze0kS^QGM=
zrRS#S$z5^hX1@6~_yRVgN+}!?nuzuBQ&Uf(TrDGLj;$cApXY)Qtd`oBQ`+ifhW0T|
zb@-b=DPf~aR>#ptD8<N@|CTbfys*J;fyjoR6hR>VxS&FDr==-2TC&%G4Ln`Jg@$1P
zv=Ki`)EQM+*&3!MDy{!Zd~C`#*t~9Kr=3B*aQI(_m^EVL=Y0GFlmy&NlP!_=XMwRl
zv~s7ISkYeF=Se;`jhDrxq@Iq?QjX}mPyd*qjn#bDP*dL1!1u+EN5)%or}*jF80?01
zHLGz#6)`?~k6E(5Ur6NESF~+<fjx0B5JRtEMINrv+iWAG4u9D*9>3;3i-9JB(=omn
zj08#<fGMqIP<7!qT~R#9)||yVIp|uiuK5%HNW`l0n#q-tiOB*p2<B#NRA`_#woivn
zkwumoJ(Zc2Uy}d(=qd`r703%{CU?--lRb1o9;2@&WTk=x$4-XEWP$6Gno+=n$GEmc
zSuLqw0eq489g>J8@VGu8E4CObHBa?x;25Q)9YVMr8)WP)4bI5(`(;m{X3zi8LsqIi
z*#F_?GODrY^WM>c4C%H6CI+yUt6+UsieLOe&_g5g-|fSH7i=S+S>{rfv-GF&o-M-^
zRSa!&iAWMHc#Vx!8I}J%JSYK=RE0u0DFv8LG!I-qlrFz-);?9`YLesnkA-!^R#l~!
z0<d!|-THQ#by|XY1yOw2@Awgx&=|#2nB0vXGv{&Jsxr?${c5f?x$BUD{x73b^y^c(
zg^5_9fyC*W`Ksp^yDvf{t6WpGG;v<yD!d~i5#r$R{O;5^^rUJ!*@TJ~U~p!|A3b2A
zf5pNlP%@W+HZE)K_96FC&IL=9(13)OKWAo$a;3Tbf`S>{E!eCqqvex`LBugA=o8<Y
zrIGyuf4Z9_R>zS--vG(1mF4ne&@#oR9PDoU8LdZ>3TC`j9beI>VcQ8EH%BmEvVohb
zDsx`rmXPO+nueAHJ->edl9)GVT3`KsckJPs_u)EWct;!AoT_^xQ~GB^YNOuuSC<dZ
z!?RDnK%jOg@q!}mf$!aH`#w)dS()@VJbSqX^T<8b@y#R_xj6bT9K;B57kxDDlc0>E
zyMstXaLLlRL45#6fp}BVkSiBcbUJ7b49|rsSgouquDq+5=H~@yRFhudD^Gn!qS0UY
zFl)@pU2BHuZvF@>y?Wb#SdP#6Qw7OZaXQKo(s(r2iP}OsuanWOHd$}_l({o!oUjRV
z8t0QrT^Zs<NybG+H`hmY`)o#36G%=bLU`zg%DK_Q<*3{bqBL8M1~_iM*gp7eSe<nu
z1S0ih|A?+zT;aeTx?K^|PGHY>Bw~tZ|Gge!k8yNh5ti;QIJ9dZGfwwM+Enc`Kj`jA
z@zUFv+1c!2F36w59-{Qm4xgHhA(5w_cT0u06vo?S<v=RLOs*D&22>-7G$2umRB)|#
zuqbKpX!FdUi-!dAi}<`n(xtzlo7$ei?^}ZIZif|GdH}I8O!HFIRR0{k%_^KAO+shE
z^zVuf876b<KYl6cj-dN1WI%?=vo2w*r3+Kd$bJTT)?Qexybf={uv4X@i60XMbL5M-
zQZV!U@2Zix?NY*qs&odN<^8C&rQ?I1rID6}W7jIe!)59G_RJy&8Lfh3j}DW~f*D$C
z;$ccLCjfq8yfo}QDwvjHY(>DurPIHZhNOBjl3UDc(9v!-s(!4Jy}YPDvUM82h;tNg
zfV1H8E)yP}INQpqzi>n|*rnM;?D1aTy*4tq4922BP8e@LVa91c8p=gZ6J4_JSgDe*
zd)%xUmjsyr#gsO~UyxUlGg;K$KEkFrikx;O0`!@8(j}7ef4jrnT(9pK^PhG(m7HGH
zXnd)R4c-4jALZS#@+9G=ayN@+v5m5rszM6G3nh>xv@G;4mGIP$vb5aWEj}SX<6qvp
zwzRN3uh|?z7j=YOT|dred(=#~Zi?(f7OY!2b}YAT%4I5Bp4wV_(K<g(($sJ$d|tca
z4&*lOGUjPUV}iFr!8Gikj}7hE{0uJE0W6^BZ20D0hJtoP<fy3<$sC3nlabxlP1fE|
ztYC6l>^JuPgRuU!>P#Mw3<Iz@bF`|`l(AEVp_+W&l0XzhH;$38avEtoGrdcUh)YDJ
zq26wFmMi_l;ZXVXH(HIO`MQoan_7|^=<~df++Azn78Vp_QF0o2^d?|{7@iVGrvZ)*
z1$RwUQk!)f>a!l;nKjzhR4}+3qy?+@%o_?>?x2osezY0V$rrk0@9lJixjB|L2Hh2s
zi@td7%kim2UZE`R$kEYm1ZulkFlB3`R<bF#x#_ArwxwPq?8BjpeiGS2b~zLP7?s)8
zo^Py|PS3l5;+x%uE65u;)HMqnBuRwT`SU3iiHD>rV&Rr)9ljY9lcMqEcNh~LNsRj-
za#AV_4O^2yLdwC~nQgQLjh!7{A8yw!3B*>mz5{{Hn6Oo??1wuiwI|;fs@zgZ2*;^U
zU@ZMG-Hg#8+d*}nc&ot(lO2Fiis+>!<M`q@!M<v%$0&qkFg=toPkW{4#;@`6+0z6v
zzdH{C>z!+Nz|XH$+;|$eefD5IeSLIrymEi*<}{L}N_SNi5Fh+zW&;xeJ6goPcuC04
z6<{<oH{}}@EE{LbU0{WKe;a7DwJ7?B7%G;=*|diDv`}TP`GUy#iWCrinHSniJ_VWv
zarHw!$~jHPScH{qSTBT_E=^4>MzcpyT4Oocqjoll<q%;;v{oxiPu}9hHpEaZE{o3G
zlL>RzPn9_|W+|^qxTxpWyVm0p4eL1h$SZRYed4NqmeAO9pVqIVP#gd+x;o!(9<2?1
zmEWTHaQu67;4&e>&ku<3l+l_dk;fi%m7-6$Sy;2|SST4PW%nCcB0D@-c?j+9w~wR8
z{W&!h5Ev*O+^jLyOQ&SsWa1^AJ%mD~a)%siW(#-D1TPgr4n|JU%U?jDax1B^7#+E6
zVeO0?XG-c>IOW&8@1TZ{?V3Kx2twj6h{@yl(bI5<lFbWjMnGPTNq?t?3!{Qo?-KMG
zNwYJgy|IX~e%8{<ucpJa{CSDXj+3!m9lp77e$L%D>jPhTCCt5}{KQ*W4*T`1>)bLg
z8{PwB_1xs^WGsA9;Y5R^_HVJ$TFrqaBL^y->qPaxUp(=M{kKN7={Rc@Ji?(ddmzeE
zX!v*;Mi7c~umd=9OT>(ez%p-r&yj9m9^r*E$J3FTcID<w<iy`MFHRfNjnuW|Z?qH;
zVNs0xe%R`#vmbB{qU6gqN~r`aYZ(GGlG4cER$c|Q3+<>G^mN`tWj^Daax9p&HF6m|
zAVK^v;=uk;9qa{@aBgPA?<6w72KmCb3hR?A_<}M4F@Q-R)|DiVblHEA;g@kmdA(gN
zQCt@lWkss-uV1Q4rZJmVN)RmbXHkbu?bf-Xjlgx0Y==z8W_fqyo?WRy@ES>=-hy6S
zw{s}6;46l2k4+a;kK^}=rlARg7EJ6f=CI~hey7Kga~9$!2H)gL?nOmKVN9&Zj^KxF
zMQ~Io0AoGVQ4a_RM49q9{_1usnnaq17;Um5J;mq;JmNS~K4Y>2dA8U+{rqkYj*q=1
zg70syuC}+250av1Sj|U=wzAD1ZV%1Jyk_#^ASUsj93C|ZB)Oy;VTH!09LhbN&l^%A
zecSjtFqu6`oCwTprziZfXumkNz7&jNPWsUcK&%qqIw=cUI#34)**ZRzsVFUQ94(mH
z3FYr^m?LKby_qbXQCnD$Zz*YO-L~i_z=^DAq9=^ti-@s`+!?Mu><<>*v)UtgRlj~E
zkY)B-t78@w7e;n-T%>gva((uDGRluYXlysbBbky=qGrtl9p4r){_LXV#;<2Yk)Esv
z@LvtF-;Y)$U?FW7E&}eeAN&LF=FbH#8vk1Cu=XtMqBdKHjmxkZ$lV+_XQ<I!rz1Ln
z2ERbi*(}qAqo+rrN=v3!>n;@G>vEs4U|+E;Kt%CJ;SnIBc~k6tYy2OnBSw%?f((vF
z6zYhuj4+&Xn#IOfkCGaHM7txZpGCO>e~V80Ceow%m|*<sh#}PGZkh*dAS(_l<7a|m
zvd1t(QL``nr{XW9>W<+q%I+DK*I$%U#7}-YUg|9(ull~_%{<n39iP88e9^e^NOReT
z&I@@O;3{sq)J~cfX10Lrd+s3BEh4=>f;6T}$D}4@?=v@M?2c@#EawF<{lp!>6Wx*c
z`8#(_+@{>$^>P$R(N>8v1-V+`iD;iuq(eGbl?5U8Wk4=BA$TPUqVHpb3`j@Yf}=jG
zmne^zwskB8|J&BXm8y6o)z)2?Y=wRz)^PAduq=N5ZzlFMWV(C`#pN_8ju#LV#1Mf=
z#cVh2T$0NjIR>qo1V3$Rc{2m(BB+A`Lg;+&emC(pmv*DdS>IWzB%68?gs0ILPYI0e
z`jv>T)P5Iz*fsC4=f{e0>Rzb&okN{Mu?e#-@fi##X-{?}<;Eko(%|O{gYbo;iAq;h
zNSbl&bI_;$@9Bn)U2l($k9#ot9zQ=!ElypB_*bO3iBCU+l7;b(fEK`rFbEC^D5*!?
z)NsMkq8~ls)D^<@<oa&>t*G+dE6?qgcC9vWYZfo2J810TYz#<88qB8^EM=Bf#we|*
zY;Hy(Q1uK}f<taIJeQ>|?<N}ldt>Ak%N6j!M+8p(s`3MiW$I>jSn7uB_vg;`Puv>6
z%-~)6b&Ax^@4a)O(*>qLUuhbbH9+1LIPzqF!$Kzr!#;|$5NGL*6BU9~8A*EXe{?j*
z!?#?0;G<PqNc8jKT}#B0|Ed4#3J`ROc@KYY9-Vj!RB$DueyDpQ1P<_Gx>zDCXO~u1
zf^Ngs<mkbTm?OkH)HdDx8%7c3eF&0}#t-YUOpvp!r<sNWPRH}f^O0dVAfe6%<ccbZ
z`?OQ#g1jS7r%U`5H)~+j1-yO=M_eEggqObxgUnWKJ?}OwcU_HfwXm;G6|Rb9FxJ+t
zsIcPa<nx0YkCeI@&o|(SM2J@RQr=kpy>1WwoN3tzWn%4598Oj>w#T0Mp7?>~wa1>}
z;}78n&4;N#ZmrQoL#aB;9?F&e<;!N96{>k&iFe$jt>u%(CmCSchn$bdh6vq9%)&!+
zcF-Db^=Wos`M|}lqn#@4XRFbq@tbSIM`|lU_8JvKs?(jhG)JN?CmAn~_J>?_7JA1v
z&G=r<%cYr6CpM5oW<IVMY9$wBie+%FGE9cGw2SHi*u3@nc5>s!UffQDq=^cTFSh^r
zdtDA`RyNN#W>g64fQ7*W%ErhX5ojUF)5WqKV<P4YxCH%0i^uKl6(;DaA^nZ!u213n
zdMcNOwo=^R-WfbweI$B&(I(1yD4J*w3v%D?O~;ZIjyEAi-lyI>dbB76X`!*Y&q8;4
zFAg10qrA*>>R^Ucv4#?oD<raWMT8{|5^|xPjiI55mUM&for-K2p*#KmeowB+-nsz~
zqQGsB>msA&M2tPbNL1wNe|3jx|6y{)WMzwB)`)t;3S&<rdm|)0_F>f0_N|E4=_I<S
zGzDRW9tnu&_&BAxdyVV}!JKm^NI3y-4xpfODp2Q<>9n$wq_84O>Bx=3iIX@{9p!NU
z{c=C?K0<L_X<y3Gv49eaWa<7BI;?;phi60(IA|inuvT%YikVFVdzaq_mG#aN1pyVQ
zWz@oi!@U(tnVsqEi7!uB5^+9ajuiAjNR)?I6{<KxGAixba`Bn#!2x@Ov1fZLkd@uR
zRfMcwTk`jsTwBw;bf>=z9w?pMkm7w2tD5h+cM<i3f{t8mwGwHW=wY+{WR}=;KTxjm
z^lJZft0Uy@<gor5=>Shb#Av0O>GThMKNbcY1~)twmYmedxlmowpy2r`!wF4XIk$%M
zl|6yaWrq>%+-Is->=BCu%?L;YfCEib_oySXGpnXHd1tQI<CN#<B02Z<g+D<JZ`8DY
zl{%ZNuz@O9IKdL%S&!UeOmZ}cia1GGK{oF1Q)>vdAKF+i5%LrBnZ~5N8tUqljcPaf
zrSSA;Tfw?6dTD8CgXp}@J1e0juso9KuI6ju)y9?bE*|_rMPPZ9?mwGLpfJ1cV^s-z
z#Q*VhF8)mZf83w5<S?S<6y{LKVa~^#OAg7HGUgaxhnx>X7?N|OV$7-JEOVUmshlcA
zF%#yHOpK|75bo=DKOXme{RMkGu3ewc`|x@_U-~t3!G#rR7KU%D=z4Au33kf!%l}xX
z`!BzXvrU!BKF%#Gua)iVnV-M3)V(?${Tj5Se^k&xCtz)FPA>FhT%hlUiVeQ!F>TMQ
zkp&Mwt?>9->MZBpgP+pzu*r?Ic%89A`L!S2acvhKY)z2OD30rkOl--q#{?LO|MW?l
z`*t7V9@XAgQn4{gs=x2IzYP3k+8o+r4(1cTHoQ}kqdw#Q8__PN*33ty%iurGd)iKL
z(tv(A5O(hifC#$tWP8@nRk)3N8cM)&YRvIdM{0FX3O|nV7{gOg?1rW<Nu3|QmB;bJ
zQlHwV`wN_}$jlm9IT-;+m3<G-2C_u1#~uY92ckpTBOgB0)hm(5+RV@!3x_)QMcXwG
ztPERIaU{T5_bTVp>Cb@?Jk{CRd88Lj96h`yA;v7_L}E9sruO?ek#;>!j`j$)x{F&&
zOG}ujgNPri>zL!+KSvY_ZSzS3<EJ>|yVR&-eptM_xQX8kUBEQPaUoE{Gd>?oyq03C
zzy~^)|ManGdt_u&u3Gpsyz23f{;`?@QjJNEt(=JgR8*DxWA5$aF^g+Ld5`lN$B+1R
zk{U$5xw4}G!^b-wmW^;TY3gHEaM8VCvOi{4ygjFU47>N6K5X&e&rcfcP$@<D;G}T<
zaC9(n7Y?$f4?tihVO3gYKY&@vFTBXUVPj6&YcYBwUq@I^p>s)bY`B843!%5SJFF4y
zUfXIB+4*KX+hbN{d(P!s&5(F|z?DPl?H`*)iw`hMSf;bZ(K~pIh{_+Gr=l<(2A&RO
zKp#1B0)c!Cj(mvwsPhP6)9|t4J`#~%x3Toz!r~4!ssq#gX-aeFKO)1bTH#RzFSkva
zHzaxB*UR$T*E?{UjqfM3R0sEyu9kn=EiBDg&2tOp4e^kp0~5&S8dKtw3<DWMs%l1+
z?mk9_)XBy@hU_PwMCdFcdUO}Nd%JJ1f2j2#2R93DzOa|=dMHa>-T&zmf8)?Q*i4p0
zcO?Pxm1uz>I`H$gg9SmeGXog~L)|^ffi`6RzOR2RIV|RlR_>nuqs;XQNP@Xj#$seV
zK8Dq))%ot%;8!(hS;=JAoj8gUF}JDW8Ou+VKPacl$j-uvnlu`Z;d&3Y+o$8j*ZxkU
zwl02b<mWfb{AzjYH(Hq5zh%@rvMAS&|Mw)K(*lO+ehki)yqPK|m0xh>Q@zWZ0RG|D
zZXP-pC+g_v{PuPhYGi%1xu8Q<S)e06YCyAjWN;~L9)s=S8SXvLnBDLe+w=7%cQ#`j
z363Sau<QTCW5^DLB}#}fc`7PiVpOcH_%ZkAGPiyi7+~}yM}#N?l&Qa6|8vH#CgZI}
zRx!QisjP3u1#~48gxb@SRTGhQ=VA4c`JRgU{a`~8H+6Q_hHJZc!RhE?irRDOuvqO9
zGdW{+{msC9tU_Nmq<=(Y@#JLbbmO!k8GgII?;3V=MwH3Rp%TvU<pJIV>hGRcGdFw`
z=YnUA`!p^!mFZdF&C*EUUI2zY2gAL%Ft7as%tM;Z5))~5%VyJ`{Y3kX*%Ogr^W&Wc
zf{w6`u6(;d5)bP!Z$JRh5AGj~T=tU;HKf0t7KERL{3mf0)u#QwBl^|&ly^_JTRxiG
zhC~*LvUaJ8Q>^oq(|oK^&z~wSkPaFxwM!3+KmMgi%arC=j;LFwgN7R`qjE7D(Xj>c
zSD;p~y_hlI%rD%_q4^Z=Prj{BK!Hq>twsN498>xhF@lz8_v_z=U&bv3zq(l&c(Pm2
zK3FIF1Y%#69wd~s8ianh@R~aMd`<b7ZNbbmKJPCsThy3`lIOM=1*Mzx=y)><j2MaW
zv(p?BjOT&z#IHb|Yg5lig>skzY5G4g^#Z>;Z)451SIQy#$1zJ`gRb~HX7La1DKiX0
z2-G6cdzr$+z}_d%{-F=W+inkRxS5>C!$ZED<qSW#);ESVcj}&IHuU<vHk10-FKgni
zU1Q{sU$}o9tN)KVuNq<;zpr<A1mkS?-@ESf6wk%)o3$mBkBz=%isG1;QWS#CAL0)E
zbkvG)?P1dkY2Tnb{k(?kC{ksRdU(uTVwznWvgM8Hgh#oFT^;dFo=;s0*#4ddT!Ixb
zey<wavXeA=XZ;(q|5vkudNF#n`K*xhb{+8@Q#Xi=-D9%f3UKyBu+8qhyH{gP@*k)j
zk3S~0OXG<%kui}`0G~`JlssFW`IW4*zLj!2<C@F7*LoLib!Yzb&e0&gWv~0LO+YNj
zNIq@1bP6t%?fw8?4>1ID4kq}yP}rBZmoh;82{4yUn@vO3=#VbV*{@$&n4^<q5xp({
zgUYE?`-K?PrMetY;s;-*cLR$1VmUs8(V9iOP0Z(vc@1ZUxiBWwPW<}DvpiLjhm%RE
z9qRYd=<Wp`&eeXX|Gio(b6WE0NR|jdRrsqtee~xrL_mudUx3EqJ`)9)&px`lQQYjN
zryuxUqW3bgeo0AlW%N?J_JCPc87(O#bw7!>+*YBYUCg*etO*26mejtiR4iUlPLFW6
z{*ued7;^K<PmB+7tVn;di#KazpzB}<^Dpp#7VfApflTMWKeSs)?ev+-xN~&UuDzA?
zilK`7uhq$EKmN!_fDuh+U}COoh27N=*H<6h>%J!~BpV#~a<>qnoB3sjbD)7ss7?iY
zgUatA3a?$?oezzy4Le770Du`fr>Gq!9*a&ByDB6kz^8SXYb$sCJb&?Q)6ZwMON!V}
z82m*V`QFy8;lzYs>~FOYvj%S-{fxjIL|`7GAI|P1g%yANC%~voFAggOC+7zFsO#CV
zzB2a~VtMJ6<lB2~wERVm<O7YZx~7i{utY`n0slVC!x-(7Sx?^*1ooL7z8tUVD@W5}
z7Z$C+yl~sx)3Mq`uJdooxJWY!p`|Q3_Zbe1NMX9A>)oHQyGrA1**06ru8<_~z)GNt
zG4n&Zf(7pG8U%}Ktz=svy1=aU<f?MMclBclt%L;nKnBS{QHodH2OiY;q~JkB*JHu{
zN3FLq{8H_S`xU@=eICbI@W>-1s`bBdNc;!PWcpXzO-U?qA@$27m-`6xy={x(E-emG
zZ~kFD5@{yy`+-pPur@@dh8b-5qbszy@*A6QGK?dc&Q3Y3&dz7Qwh{?0Ty_v+d$7Gk
z<78|2j@;83ANBun%zI8_Dxsi`7pI(eYE*zPsCJVM!%?v7Kx`1Um+P1ex<?Mp{||HX
z+fxbHJ#89kM+q0>tf|paTI~SLATBce=rSnsX&iN&wl;)PCRGWJ5IRfp3;GLH4n*b(
zE@Pq8tHxiqza(d6?JVlakA^<P0DUI;)U)k{$7$1>UHEaa@Z}v5vwQFqw0axTmxv4r
z3aM39Fn`t{I07e*<rV_cm_lwbOxe%TQ5#fKtvx9ugYnPd`h>*4CdFm%|0gB@WznRb
z!8cFmF~{`=?||W)a1@F}c8xwb{q-j*DmK<8&?c->O^Glpe)F~5A7)Q{-A2-zpcm(a
za5od~5Y6OLc=Hgxqkr@4e!d?Ay{sX;$mNBqoMW^jtK3M1L3oImCut7n6=AQ8n48l$
z<E5E>ZY`@-&2A?f+}GQ0OCC6lFhx!99Jx;8%<+-<(Wc^0=H<8H3y)9#2A06DMmBx3
zU5EW0hlJmXdb?~V+b6fTi=DuZ@)~9|u0LhuR~)a^mVq&4&prY6{ZLYA+7pHM9E(;N
zoYyS_c7x%gMkQ_-)@4)E_1%`+;hGa2L3QL;J>ms^WWMfq7aKs`xgY#ujLux;+qMrd
z&G)W>mRlzSaEv?`f>>RJl1E0m`i|%I0H*J(2|n_qeC)fADhZb@kaWI3JJqNb`5#X<
z1PaCF0JH!N&NfX27>7lI-{w5$HRd|GD-{xAAh&9d(HV@m{Ah581r-VIwk@GJ)-p6D
z-l0k>*c2-u+2f59tM7^_4020Le1926L|UFtP}Yj8&MmyeBKaMEKKbN_hex1c32AGS
z)~ovNf%cud>WBwn-iwM=^LQu0z9HVv9CWtk%4_Pr(an53+&t~o`AF(#G0;n?#sK9S
zYS#Q*^M9DTSKJ-494(({XvKB%HIt^+mSRtv`Zlg6yc(95uOxVNboHf6-+GE1RY3^_
zE^G&X=&-x*$!LS}5i>Ma7+r|%S=tG$yFYCm%wQucNkUSIYv!OY_wN`tFnckF+~i?+
zzgc5qoWOcPRQY^?+I>gm@&H7nrFb|Rnzz1oxUq&=JKQ@x{TYZ3y^1hD-wo!m^$jwe
zH}yUHN<IFO!$@=4v_nZso`Y40T^e<|rDzQTx#ji?-An1`dxNWftxw$!qyl~iokTDo
zOWHm<dG+vYW#tzDV2u<+50J^*4I&j`9Tu0wT1$65io3qnu%$j_C*SdL{T^0?jkKQ1
zmBP>`6L?jOxNou=96{I&>olEn2O832L<aswoaOljDKp+5#HS}O2}>r5^3W@Ph=9zj
zDQ=C-mRI99DV~W1oD2>5U?frh`A5FOLNIwjNSzwAT3xGM+I-B{y2*Ll;+>;;b%&3f
zT)(jws|#{okTrXufd3pH@=;c+Cg^RrFD4ZX)BfBbm=pv_DkdL64*pjfY=x6G60fsQ
zg~mq3N_Vz~B^ag@mWhNX!z8(zuHUgu$lx|pC`=5k!;T&TQ|ajIwNWX?LX6zEB&42k
zN+wAp1BTs)OHZ^JS}sOiy#MS5vw|wXP*LiU0}f`U1?<=JpOqAro8*f_qzN;aC#*kc
zn|6&J9^G}v5k(zN!Wn(^srB1irw+7DZuMx67Cg5PXVXCJ!6|vQ<n-^9$l1@Xu6f<*
z-+xB$(q}b&!1>&%#ll16A2Cbag0;41)c_mZO&n1;cdZ*grE!^?F?-<yUA-Uvw(*Ig
zP}aysbVuQ*3W8;-SzQNjh>HY9wFGdXn}1R@?)mLz01*ggQXIVMm96{QShTv;RvG;t
zqmg<&jZYM;5#f=*TfXxwcY)X7D&KUPwBtbCveylt2tFuC{mA^G-y_*}xAXdh&vO3P
z)9SPv&RrtKIT!qLtL=Hm5jMFf{oRGafn*NnSOC#;HwA9Gu~mm8kV336C%^yv{uPX_
zeJEO(8;Pr|3BzcV;m<=8Z2_8d^>&NEm1ZxLuRY~3a3SRB)S43Q!25@NA_sj+(rt4&
zKqISTmG0r24WbV<_GSTx?%v4sTb2%Od!r@ymHb*U?*hxYCu4O!4#!UOO7tpoMHGSs
zb|V-lFzbe<Rj^@BS9mb4=dQWQJ^HGIJ=>r@REYx7E6|24xjB<h8s-b0JK5Q7>MzvO
z9Fl4S3=L%OsoL_gXIp>U9v}niEzW^JjjaIx{nb2w|AXU`i0eBkL6D;p{{UL#hgHRC
z)%8f-o?beYQtAC_E-|HBPP;mmvpQSloew_^wYafE4O#hK<;Y{?rph8sny16`PX2b0
zb~i#b(*v<?OCu+-S;kMsF54&Gkc-4Dl)vh74DRxE@GL7*cL3>A3GbrKeOtwg=*3|E
zX&sNmJB?r-?{Yy8>f$X_tjs?UrzeZ5!)|*Y@3MYiPqT;Bw+5B8*wJ4BQ7#I{^J?rF
zsdI|N{%x3llBj{Py3m;nKH@!+=R|dDXQrI)9frvv=%ghtWxx31FBA@tC`8u6Eple)
z&)Z*2VT^FXR{q~7Ob!+XRH**%hrsrRX{e>Fn9Zq^;{$aru4m5}a2<`xV$23?nGM&b
z=I5u@es|)rxVyDAk{4X;eV@J+BEg${!qA3y8Pyvw^iG12YcdG;=I;-}rW^JSW8BkX
z#Fziwz$QM?c|eJifTFMf%KP-^)Xv!o=I}VV%ZSIZ;(T6*Pc6rT=Y>AG%GZ=%pp5&$
z-*8xFBXtsT6e**eqf}4*jvp`oT4vbi|2HL4XykZ=rEcxdQ7a+K`w_9FS0H#q73+<x
z@Rq_{)MqzvW9O^krvnF@23Z*P7Q4G;zQA@M?y<QO{16wDzzq~LmoW&mlJ|`i<4vdx
zlUb1BqUc-3nK+AazGmIyR(>bJDE@eLP(r|wafpf2;T_Oj`?0=ADM8g45W+zR+~%gh
z65?YqQ^zlN=f}5(C5~iqG{fn?wT<n)=zL6t2ZDvi>Grx~@c$yE4@;_YkRMehP-aj^
zr58v_B%lFUC0nL6cK6Z_jtjaR?$?-0nVd)2n}4EqTGW^z>Z0Z&ezr0gK<}_DzOOAy
zBKD&yCD#jWUdWq*y;1K#Rd)525^=uE^JBhsno~M*Q<2w8Jo}rzAL<mPN+&|xF;z8;
z{Yzh;|I^(!i*qD&@>{2nS7@<qdW8zRI8<4_KK=rIpIfvozwQnCx>efGRV%&T<G-iB
zPF7C_in`Jo7z-;doTKfpb{Es3$R*BB3;!DCchJX&!5?;cB;HVcYGV2>*qhG)iMJ|A
z^tS@<+bjP@5}Yck&Ap2&fJy;^;dzl3SFcb%+8y}x0nNgoY-VhBd-!@-(;I5Hg|z7s
z*XqjO2=wvq(W$<FQ++r)brVtbFp2rGJIkC7y9%PpN<4WvVUi%q3|potBkOF{^Nw~z
z;Drjv1?9HIYGzX}7Q~IJI-5&hh5XD7RU-tHvK^R|Z?GoHJp>N<<oWq7%t-Qq$f7yx
zs7~dJ8SjtOKmSGr!i0YQd@#BZEsbbKhJ`LW`YjBRUHbqzFshIG?rN{CGy)Q+UWVKQ
zp3wXmLRGwJO5<~ID}%6x@~x+nL%$l*-76U-QpT>)!|2b0fB?9|f4`6RHa2#f9F42Y
ztF4jAtBa@^r9PCrZyjcG`T>1ynjM+FyOb(5&=!8qcoFOt!89|l|MM*OBM0;{m$qSW
zvZD+PU9G(Za1DV}xWC6sJ$-%W>5;jG*k`0?A$8@c%1C9g&STuJj@+(Lml&?Ydj_s$
z>HA@{lXg%&<6-|vQ$b9ECrM6_R&bdfd>ksWlmGKi_(y7R2gcE-I$#}$E*`Uhj!Cdr
z?n?FmZvyPT+2w7&z_%b-Hp!u`-geNYp8CKgr!$(Vgpyy8Ph=A6XVg_>zH!ronHyRP
zW<273#Qu7{#@7k!ftMV(=u8iqY+VZq2wi=W3(AZ=NdTt`cy>g2k3uhKs%`doKN5Y=
zH!zmr^}6%{=GHww*n+d1-Apuh;?;$n<oS`VUzFS<-o7&I6d-p1SUv4!ZXd-M&w2J<
zGZ&=(sb(lsT_@6UV<U?zBAs%PR29oO4R#=Ms8U@ZNsM_T8Lx4f#yNiGZbzJ+nJW#Y
z)O`fw=%?WHt-^Ni^Xq0xlT03)7^aiY!)I>f2OeOMZC4w*#9zMRjfv4b`MtN+uq2Xc
zltEsus9=4G@Rdu<I2D|ebw&8z$e-QYRGgsd$RTCh<B8*phF`vGH`6=a5@k${X^&|?
z`O}rvmUY0jbD6fYGZV(F2)*mfv8CjkW~b~h)A`Aq6xQWhC~q@9G=}w1Md)8cV7r|t
z9g-M}yuZQ4hQT*^bLAzK3|8dJapvKnXsJw?9M#AF+;RDfInN_}R}il!p^4&+Qi#!F
zH5xb>h731e%BDxA<2R)fML~7P=ce1<m<xWfm(}1~5j8MI&SN`j-C<KT_#9RVQw#2F
zGim=M&*^*f^C!Okly-zro}HrP&0yOQdJH%rC6E3fMgZ*&^cDg8>gvgmA@W$*Yhj1+
zUEFiu)XD9834@vLVm9#GLN-eyV14tOC-REK4tI-~n=dPJNfxHC3cO7jkIj`joY31=
z`$BW}k%HBRrN{DO7gFO`8x+q&3+9>}D>nS><&9_1=N^9LAS)?Il7bd`igHJFFtY7p
z(r{~iUY`Bot;H&JK{nS>Yp9#KUFWf??%oshp&)APALexDtjBJ=<J0ISGc57M|6mH(
z?p~%H^OA${|0^VmD<g3gObe;Afm8@npwH2Bh5&LEC&-}8GIac7lhaI?nRkj1j(?yY
z@=tBsc*Rw3QFrm|_nvw51u2Y~biUU4w&`qERXdW>c;<F{HSMBA8L&TFB<uBJq9bO*
z#O?vq&o<KC%n6BXBE6PaGuY<5YXt8XQCOO<{!g>_abHq0DryUSzIr_aVwK17jo~c;
zHn)^PKKD=|3O-(D>=qUlv3plr8xjiwjPotW;&W@fqr6|ctIW7-U%@DgjVNxRN4g5;
zlMBe>9OW+bqH%*Jb^gS57aA4!+4^3q0hgN#x2JIiOf4-ZWg2Nas`$rvviQ1KhV?z&
zMUjzS?aTX|NeT4|>mGg$*6Lonp8d7FkABRu-(qEQ<ePS_BQ7M~VNEqknQ3C0hq3h6
zZyO=@LbWi5zs*TkMli_(;}JRwW}?QnfJocckGg(j<-FAKTHT1!pCI;WR11L3BKwgE
zl^_?Y9%-`zK}{>OaD{XZ#C8LMnD%*dV~GIvt5CN1xMBD%5QbKRtjTdCMZy?v#E#2q
zfxmTMLz%&bBFERguTMOGbWrX7EfO~#X9M6eTut;T5JOD1g~?-6aO3X!`W2?9yge^~
zVzgQ;u$egh%)HpLnYEqdfgkl?Y5;s<XlW}+2UN3C>U-wdJI!u0rHirU4KQooR58}`
z+y(Aoqct5>xCDDqA37ya6~TYs6}{N6wWt+RaI-GV4|G+W)x|-e)KtREdid4S_DHNc
zr_rUy1PrAHSYlSk-<2A;xIV>W^hUJt?#8t^+<CMSiHm-jYB9$IuI=C&Yb4>EovkXT
zj0a=XMa{;_IT%tS-P5==47a|Vtgbf9KZe}O%ZqP&2VsQQjz&A>HgA3@2Ya|AHSSDu
zKUfhpcbb?=9ORfp@!f0pNB|*E+v9xQ9q9atm~c6-bW)Fw%Q2x9|4$05c(!y#I%~W}
zRgv56rVyaZlU!T}>LVwascBJlVfMM+IC-Fg?U<g6x}s+AD8&$fNkI&a<Qdvi8d}~-
zt0_>U>`CTz$KVSO{EbxlI&X0o#o6dAE(v8OZ-lDt6(S_sAIkB8QikY)t9Ny@tyM9q
zXztXL!1kVwm^0cbIT;}5oJ?AyD!`rO30IB8P<&0Mndu;_e+?g*{`ttAuGu6>f?c4|
zA4nPV=Z}Ex0WmqP8mRhQZt0;P%TK?^q$e*~5aZmC$!sQdNxXfaKsa$F?M9qj4x#5_
zj8T_kPb%{dfSykaM0ZEF>n;i>V+L*?-T(f6mbKM0E@j+sWAjG}H|p&xXi$c=1n7dY
z)l>`qU4MmV4Jys^e{ag=<Nv5WMM7LZBkJum6h|f${^yv?<=Dar_&zAUvF-Bo;wilX
zE3G#d*Q{)~G$e6NO!1XZ9UtEN{eFI=YL~^Xg4Ki&gkSRgO5l$p(~KI7j1Wn3oe+z4
zJB-ne0$c0X(ss25-|N?f4vY)zFoAsJoSLzGn{HJblbEAFrxcn7t+Q>{cZ_Z_MTb#q
zQOMsv6FUm*%mCrK6_1Tz1jjK0v<P@bq5gEgkRQ23^!{|YR+r;yan%=sc{6zE&9R-Q
z4JBr3jJLdtJyj~nnCom@sHJihETNTWRaAHHkvr>`Ax@6Gt3xS2@LKC3_A_fu(7!7U
z^V`Qwyn(D+WBx96ij|tww}lXa^mu_*v&3v<d=-OjocD+Fi;VhcZo_Aewh{A#>JLoW
z<5?{c2s**N!1FfXSD-R*XDkq)%WaT&<3{M~Iw${)V_`8%3~L2{o$>zQH4C3R?CcPY
zt53^#yHHy!j92Q<eI?zxq`{85#QLVU`+8qbQEzOFJki7rsP;MD7rKpH<FpKb^9USt
z9UtsG&X)uy?|(9KTVQLNkFVsBQI30LuUmAxoHqm$qiebAh^W&Q*XMcm90Fz)2U~;W
zUB1>uJXdP-!yJ27{cByg5a)M4cZp)y{xwWZLYGD`tfOT@Us2uI6TE0uJ{{A5l?=G-
z%X7cfe5-9`u-k&S38IgK0`9~Z^**Q;L<h<e#xy4go&gzK%rU>WdtBs7c;7Y5lXb7>
zO7`jYUFPfp-!EQAHsv#z%68_HR5=*Z%kWE;v6ya*?sXt+H+q(QJ>%FtjV`V1S-@qz
zv0caDPlWFqa02t^?rTh`{qU{f>#F13-wD&f9gYh@eR?WXotDpHk>@+K5^yz$;N0N|
zzozr0{O!K7mbDaSC)7ABX?ZsMi|92&yI+D00E|X}vWORN7;X;=W_$`6K8e^34C&U+
zNEX99$?ob;<!U=dHb>@HetkRRQC&_;ijU{7`fm6{`TR1gMSP+WJM`J^W{q9VTIaQN
zAKeSg-1PKeSH6i>i=6&JW1977`|8I)$k5?`Cx@C?kKHq*>^u7IFPd?n#~{cgvEQr?
zSu6NvXxzr-oB=Y^jL<`D%`b8PtffHR-7Ua$T&f%T6B<j`@M<S3dA@t5jns_cVp{1+
zYD^J>s%V3z=L!i_B8SQB9Db7x$MAMVBh&rhbamCL?`$RT;Z_pscLfE7QJpm%&Xf3a
z=1vGcu<76!32vc#`Af^SsSR7wf`zxgTD@<c6w{B+;(Tt$drw{Z<{hXr`i74HF(1ZB
zfs*Ui9vLx4y#@<)@9WdGs~cz+<a|gk)sOtG_=o5Fv;XdIn1HV`-+JGQ#D=yB`UmWX
z)w%S#5oVg6;F&pM7K}5PMP9cS{pSf~Z15p=^Ac;29Fw03wlXFTN~Ly0<Xgc_(D<7f
zIKl*yAKzf<4Zg}c1WE2Smp*6noE}pJR5s#D4lW>2FrhQSllj68D4b>aPK_Mjw+NRs
zh}Qka&Qi}iGU;oejqs()r+^qbcQ;b2NWSC8N}w_TO*#$C|0E|oVj;$SYvtETSJ!cu
zeECJ1uYw#Sl-Dp^xwhkbWN+c^0EMYP@Q+&+iMHt!H(JYB2jxiixNME@4hGbpFT~g&
z%JwepDcx8b0(52s={5Nh>-&q@QTch|)w~2vKUTvVb29JW)a1Z_<ILp?n3(FzpnLF|
z>eki6eO|h_)m-(#iyuKm$B#(gUR?BpkaPwY_Z9FEUT<+heqD1yvpzDscARf2YKpAx
ze97ghMpBD5aBhkCU%s5Yih*$$v2tsI-2fwO&skYbh1U2~_qE9UHhS}Zh>_^ITb;Zl
zRmVd$K}ws4gae|dT9e;VZqH9}sX>Gz<+aayPF7!eU-|Ny*N?O`HVHOaLXbAroK`a~
zeU^GQ{KutD<~jA3b3(TU!Cip|fxA;n|7iOwYJZBvGzK7@{uz<Qm~{{2-C9Jd2GgdE
zh>5<OLTyLsogdHQnxvJhzj{h41$<t4zh%4~fT*)m5w$jpDLJ}+i~V*(?l*nV#FPi>
zE|VhT9ON!&HbvU>>Yz}qCX}xqZiZ*R!yPZ4p7M9*U7Y2){q35XUEIm`n1LyyeyR&9
zrg$sZ-l^DmTRoVpfk|lAX!#G>GE=zQi8p?iL4Tjq-;<y?s}Wif+m1OqJw4&txgHy}
z^d!cVdvc;uoyb-p%^X@HzwXFJAC79fNR(W6tnvBeVQ+U=l8x>h1FAyWH}ym@b5)TO
zyUU--2rw#lyAwM2nD&$a_y0QBsJn}Qxbejs3b9U*MPCSw(O&5v^inKZf?vw^md-bE
zOR-}MNs(T==li?YDtXmk%Zl1Eb9DB3-_tT!!akgWvL#+za%19o|7KH$vq0z->rKzo
zvp!LesEV)S(ifYRvi0dxSkY}Jh#e&^pX{NLVWUocMuj-j#DaX>X2*K;h~u0MBip#p
zD(H67CJ#RY#`-?zm_^ERsZ-4iT>Wt)F*QY%Gr6~yp{kCbwFZfpmJ@Aja4vW+E*kP+
zYTt9=W?Y_IMz$N|Xc{fXEJ_c*qQcD70s`lyiVw?kK9njS3g9$S{a6Zp=Vg4uMamzk
z$@oolb7BG;`KkOpZd8$tTZA)NQo+P08wV$Go&2I<{?Ja2yN(}9DY0M(v3*iP)(8NU
z157D(db*?SDeFi(Wh(q`ZH`Y$5GwdXm~W1k8oQKnxCDCNPV#G9zv(jJ@wi}N#IEB}
z?GE!lwBrx5!a5c&codsa?fU63jIg*x7-z6C<2$sR?57-u0-Kmm9pA4JG+KDBT>0kN
zVuCTGT&63dz-Qng=|z1smx3GQU=tyDn_ltn?au%86$>r752WI%WNO}QvnZ{Dk76~1
zZjVfjZ$y}_>ce>2`1l4Q+9N^edO~dE(ZAr>Mk+viwhhw%*%<!pS#r2ui0mUnv_r1P
zK&txIwFmb%m=_kL(o)14s?zdnvb9BLFg0PmTz<&u+=G2J2Nw3hc)@e!eh-8f2Cua&
z2!ph$BI|BEFpe`Fm?V>zqR$4pVo9<`g?H6gz(<ozDhH*8{)4c`U|<BPfitzw-Av2Y
zaWT2cZTyCiC)moUEt+>AO<)Nr#=~pve&vC6hQT($gEjD4HJSN!s|i}Vk6#bQZ#$2^
z0(3)Y{_X&C3VWOn9@)&wlc+Crm2q)S`VaBrx0xH~5z~~6XeQ1BF7<c~g}qVi#QOT^
zTJ&LsKvF=qzmOd_nsscF?T}J_k(`*!oCA<GhQ)d!4}SKYRbrygP71n`=gF~m%P&@I
zUIMu+4D-6+5pQ|P=c>&gY^GqmYARP(Jo^AZ3F*C%`%|PI=G3unQBUWO5ou3be+-(?
znv(;9_IqQVfBB&wQoQbc`sdec9N31SCAX`X*+?z(`jAdybMa~VUjD++P1N7Kh-&N;
ze~!-WCFC9=<b!7~@lroEM`ixULsnQ4gx-&)rw$6c(_F)*?~t9qAB5RDdNnSC;1zbu
z5KFpSBcfI$fP89ZYr>&6$fsN~@t?A;P<TuU{L-;2<Y?B!Gu%cM2!M)7Hlxc(?L65C
zJtJ=y>{tb9oVe6}t5mQmxsEyaQhG4+D(sK+-oPDmgXYn&qrV#=@&axX^i(9CB)v}H
zn5yumAIv@&&{!Y?uItR3AuJ1Q>L7k0OG}iG+-UIOzr%si(f@Lo-ll*H%e-4MN1Mh-
z)z@U**JTJPbWAtCzr3yAdl=%3a0v_z*!b$_axf=KR!ZwzH4O0Vi~b9sLJMzS2R&}!
zCOKU0^K&qfScm%}Yzz(WqwiQ(P_8C=e4yeiZQr!k`K18B8O)13+f;8+osozioiVJ9
ze0Y^h(yiTb$~-~KzXBU-aKK3CFsegQCcb8-+h1St9qL2Rwp)r&YtPL`W#49s9;G}z
zA_r=0`_sZJkAMA?RIriDPoa|z3~WZ8=MGTgDMboix^za{od3BMXG)2CW6B}?^`k0i
z{FQ1vJ@}@++9gb|F_iuV^0BA}3!@ZU{G<Gocv(?ZZL>#@)$H+vcRsQZ*ULvV3Z;FC
z!{N1-W_FGiiz$G@y_Cyu%Vu3`!hDIZ5w{!NlOz^aoh6MpfHzlTJqSqtm+943!X2+Q
zm52M&HF0RMVKC$~0+Gl%LjHTowPQo#K#8nc89Oi3W6H0Beo}AsGlJCwq(K6&lpgnU
z+_N_(A-N6NEZ*`X89%n2JH6P^0JnSq-vzL<uKA>T=VBZa#X;@QQq|{lgHfevc#EC2
zvO3~boh!tRBog4CQh3}8y*{B__=vkMlCRn)w=<|uFf-ZRDI|$+L7=*FZAdxFK6j!o
zddsa1BNC1>qb(*5-Vm%KIDP`e_P__Q<<b=teN2%nFe66VZ+d({*s2T})0wPI)t4B4
z?>g3op+aIKH68fsh&~1h{zB03a=SWB?LFSFNJVCi$194(G%IqlGu^A2s^<*(M!N|o
z6|Vw}-Mqfu3NvJpaVt-qOc~7h^u1j$d}IKVomwuVs%@D5?X6n8`0Pw5G06_}U4M{v
z$IRORg0}jf)HV5}*Zi|1srzNSrVF(Tgf^xnsaytZ*(;}Aw7`&rz~bhh8Yl_tL%oJf
zo=uxdeg*^vtnIE(Xx`;kW?|!gI$dXGV(6*1`}AA8pq09MH5Gh|ESPi-?7$?Uh3WjN
zQv~QZ?|qz>-1_2Q>{ry|rK+xehtQ3Or`d5c)TP}<jfgS*H&f~$^8^BOq1rc7UfZEn
zW(C?>iM^UCv2EYD(PxM2T%$OXEeSRDuIW9oVKp;GiF*c!nQlCA+IOc3&Te^gmf;CL
zbu~T+iS;H*M!sQ&^bPcT-`HQ6daN&OHE$R%sG;Z&Kk@Z^AV>|?<GK3&nSz4VZm4`%
zF}~2;`@2nbn903df?$XO+?Nl5zQTC=(BNM#A^k+-asX%BL6?I1)ToX9C1dvZg7f7=
zpWzv+?b?^k`BWc;$+^jW{l2W9>A9#RJPv0sIQh)|a+?bbPCbsj*Zjf_(@q?3osQ?v
zZkr7eU*;(_?An{ZxuX7!1?*0;wouvn@+%zk@SzzhZ~c6O%eSN{lhRyS-t0py2GI*w
zIN!cC{)jWFx%(%w120>_Y79k^GHQ%0CX>v|aaTJJf+qA)SOfe2uO%=wTB@84Ycy|^
zh3|G!3N?RjP-Zdr_dJUVE3TO!98JjVJqG%GMG$Ev`b4SiRND|#ZGyI&M1LLWy?UBA
zbqqm}aT==C`hAzvb|$*VqT=I{2KK1+@ns_kw4$tL*N-Hbz1GHXwjjuH*9dUQ8~#xl
zCn=R+BdZZUCW%ko^tQr}jYu5p*n0oD^8RcFhByW>94061@b|{u8#0yiJAIfEl9JV3
ziJc=}Xzj5IintZ6+4j?V^8(MOso>b+PuS|TA^FT>SlLT?(uRPW2Dzvvr7Azt-VSUM
zc>hj9rh%=wJR(jU%)-VCPmvz%9W@Cb(H*)YH!mA}+d)oOH12|n*N*QJa`Ej4kf3h+
zwwm)pW=tdi01s09JmcASOT}+wpDn=yY6xFav5*Q+MwZX8xrix3E9C!YO5T}_aoP+;
z30;D^8A|7a-PJRa=TEj*Pxr@9Y2!4C)74*LH5B;WDiZYwZoc1mUPvP`4EUH2e;@AN
z1^EcD!?a4PC~#rXVN)TR2u);3XK|7I7>p740{%zvpwEB$)}YK$-l{|oa|d-)?7k46
znqI9!si;Xw!*$lyrqte`?I$<R3-i9)^^h?kBB4p#9PxtMYpUE`kr!~%^_Dw*($K$_
zBFe`<R|Yygxe<h^Tj(Eykf$jQ#24wNpCZRCUrE>h+Q0iO{!ta;Gn`;sU0Ui{Dui*%
zxHVoj;}L`0y`aCa6A%zUGxW-qomGsIt6{a^tx9*fzICwbzaSD$C{9tlX3+K@x2*~m
zZtnNF2AO7d-Y&pWep{2dY<6#TCA7O|N!Y59N@N#SG>+BHqmLj--sEY76YS)En_2Ej
zwa4cND5}lvzdF@YU6NS`r&gJxu$bQhX>GTn9Da-wV?!Ptw;q}>5|-}QCXIOInL}B#
zkHs!$0zK8#`1q@x(rHIon8}D-o=461A8v${d}VG4_ry-hGi6xUVJBcIjnzxS*AmiL
z16FxLLtx%YoV2H@M&C}_uax!2$8X|Dk<ywrtf-R9mv`R2B)}#}Yf<_27_3MBc$+Xb
zV5rJP@K>l!eYYJrhxh_)qYrcff~?>7T2g#u<SWeaCNFT5asjZ|KmUC*V;!MPyEJ@l
zPPf6VqjsqnoONu~wYy_uKlxVJ#Yo&mx2_o_z704eIaXwKHY_uuxPmY(AEd|GR%wJk
zTnpy54ox@CIzW#tpkpxqccHowrCBr_B+}Juqb!jI>wb79A4ZPt6WKr3<@q%Vz+H%C
z+}3KZOXmkpVl2dl+Q)2IgGgB-dKk~d)nEETwJ)L&=uJ2JpVYM`FfDv_d|>`0=s#w+
z_Ac!qa+Osxj$!u9kdF!B`u*v<3M2li*By7;j*fKlv`b-+F#r|uX`0JJ;-jhIYt41k
z%SEYk;9*d<Vb@{@uQS5JaG5+CGTUSJx0tjS+)+HS?r6r|E!fqKKOWfl^K+jpa{BM@
zX%^Q~&k|W@@RvUC!HcnsH!jz0ilo2@hZ(*x5-`;$x7#=plbAqIBJD%M8>lKny!#^q
ztMhGVhi88=6In|=Oop7l!k*6j3!}^Co^)K_`1kK0b!zG+r)(P(mlvwYGJQA~^_?ME
zKl>Jg4Vx+XCk<G@uWDERoJw}q!^9M?cU#+JkogM>XA3yL_2t~S;PN~)b|mmk7v}Hb
zfQp<ZRz^mK^D;wn3WuIF(tKRAo-l(fw}U;ec-D6jBjk480aZEV2!e$bF$A(_>RXt6
zRIzd{!NZ;a(ys9qj;f%6$kJ82>Cfyerdx?_Y^=#ApDf%nXfK%wKGg|vV=8}os?9z?
zO-#ND<RmTcMT2eBX<6#%tmG-b&6}!d#Gv8e-?X#sq+%m%dg5+QUMlNTkY$_Mb@xQ$
z?w}IP!el>KT{!l_BCh#k-qqo&O(qg!C}QFZ*VU7FZB^@xs$Xx%=u!)v1jAVKU+t{^
z(zvd(Ae04=hZc5{Vt?f5n{?uD-e*MbLB6@7uM*S*1;PCE;B8J8qbg5`+fo+qqXO#`
z_SvE7A%G7;nIEtEBn18ZG>SXt+T9+<N?!!#iBP(!X}9(kPFR$gdD%*&@86#`XAl^c
z{-eSpUiIxIZrvsWMi9%3ZwLV2B!A~pNBU=`g;1Lf=L0op>kBP8*$YG9&i>1|$=zdi
zcREqgG{Lx`7ls*IY445`lL2*bv6SiKE-c~VO?_P28;>svd@HL1CmW~xt$6^C&{Gd{
zeW4`DRp<y}6rrhyh1>2$XfV1l|04>68K#;(d4A#W#w?6LOg7h|bcL)|HClao;cIKr
zV*MLh`Cm2m#cpuML+R)=b#)fPP-ZANpH2E$+sLB=<_a0r_41nX$)fO)6W41nvk&yY
zm5N|+QdZXhEL_~>XWat5vlVXgC<Wu-H8qX|1S2W#u)8<I8aO)Vioedt{f8_^l`+Tr
zT&S|MeLu)r<>Si-JJT+|B$f!Mg8-@MMg$PUlke+fE={{^$cUCQj+^EJ*7IFJvUM`%
zpA}cYQdV+7i=Nh%Kz0N4sy<rXU=e-axo@xuKlZ>AyIXE+0KHHRUG81N>}aK;%#u%>
zP{ZWE$$Do4;{zx?PPZCnJjK}3gvpETV*7`^Bb^BaXfsmk{vR(-=-lKa%vfv;>0qI+
z?)sMhgyyM!`8Mj($gAVuf1-OWzI?P}C7;W1qL$%ENx}=#8i`U%m?^)10cY(-@{yfR
z`Cmakb%?>Yw_~@CFHibn#>>AK?hbDS-SIPkyqZk=XfZvfAYc@%Lg>6^F0Z*R<PPQw
z2_vaatd9aZj(0+aDIC8ZC)SUzX6XC*X2u7P6DYmuBdf7hKAfXSd6K5HxgM_imK<|C
zepF}T1kY%r<m>usY32wF+z%n$J%R@;=z#5;)VaSCUNS<@-5W5ZJ0RGjIp41{p4Ko7
z8Qo&3fgDij`8dX0<tKb|*5w7K3#*$Q-{&Idn-hW~-p8AC3+m4h4;O{zvfUwF1FjZ;
z_5(NmDnisCK$}a|Txd9m@#drPgN+LP$*h5GvJt{Q<d$dSj>a(W>x%}>#g7Ws9b0Bx
ze{csm<;uKkG?KvB7(6%-KafuBEx>99vWl_CpdX^mq$^Z}vMPfXVt~#4@8BM%N+jS(
z&1IV$4oLLO00Tmrz<LTIt-2%fd))i_8;>}qE$i1jD3nagzg|67eI++`?(;Zi2MM9w
z*|~XtAF3n`{8C&5GV$P_((9%+myKVJMio_t6_roBAu;3Te6^9iy5Z6pTx~H}_+aH|
zp*g46J}OPfayvNIS{WLADHHf~IeMyZ?*iwd5MY#k{qTHNC&SA_9n13X1Fq_{8xl3h
zAts@40QF`#IN)DGdeb3<)~+u13j0aHx^S@qS&i{#Z|xjW#*H|EE*5n7zzjNZ(ISUN
zdEbEf+PAP^Ntb3|T~8TLck^RTPkc}(B=i;m)OC{%F&U<+es*u{)V(^==)kZG22`-s
z3SjV9S|TqN55GEZ0k6&H`K<AWU${TtNN?-*LBP)9$;vnvrn3k0eU`s+lhtQzx)Ko8
zd1c8DSpv(N_^&l%qi&wY20hcr)E|L5@tB9C@_`@)DL`VdbM`l;Pb7|8m~1((+ZKNv
zd0hXZ$+6zb_j<v4r_)<#RDG$Ru!R`^g}fYunKgH<o6qI9&;fr_E_CHUELvkLT;%BN
zmmUVVrdt}=gI_244;j?c_Y`QW1unE@|Lff-4p=Kar7wShG^-;3>0=|hH=Sl8G2_0Y
zEc%a#4-8_%DrcCkltL+)D3Vwtcjs*zIl|1`o-T~XCPkvjN#1IGm((kQ?)o|RC$2vO
zf=eVEzG)O&F3RUTF4DAx^+%$71>e)R+gcxZ6i+B998bkm(&Hy_>&__npetHq5umdz
z0C`h+^_zIYz3=Z$mxo=$IAt(~#?s}}L}X2VK*v=E*cG%7`@UcMH^0sMG?yRSkY67q
zf2(<zC^FC)p+(te=dhsG{hv!<@bdCDFZncW(A%_YwOw!9<`Uh@i`71t-_!Hc@n}V|
z7k19dG)})U{Xi9PbNNR8b*DCPjKD7pwh<78;8;L%CTV%B+YoL_q#CAK-$1YoT=*WG
zqEYr#<}S-6YYCy@SJciRo&6Jtn-=i1V*sN-Y(9Pp@DYxVGTFZ66LRE>gGk?C47sCF
z3PtwUMwD?ha5b1D@QW$=8i$Av4)V^;{647VVFX>(e6<r!E;#Or!6s1lJR*-Nt{UV*
zIZNkVQe+*3o<Uy~lW!t>=f?pVxkpj}XeQ?$5xXn&Q(Z~B+bheJt4(;s{o=dE5<d8g
z?AZq+Cn^&NANjRlP3KQo$B(4SO1Fz$u|nu-z(T-4-0+#0a265SN$yoA)zq3Hnj@7;
z8q#i0z3FJMzhso%IipmVdD0#NN{R!mkLEK}$O{dh44hydNjobXt^&QD?SfIA*QVb~
z4;pPQGufEYh1CUABem(~#PPJ1^NC+Py;Zi+G6!)5RA<WamT17c5dWTT8rALny<d!d
z^`05?0mR0_8#4<;Ksxz;?o3OCT`2<(ghtXG1gX<9+bWaW)U$Tl1)qq+8}5GN;n5O@
z5g|rfa13eZfaQZG$wVD((#)d4O$(1ntIkiL?T7%7Vy?;&ZO_Ha{{<jYGv8ab!AQ^6
zH~x1QNID|`>8X8kcHAedGH7^ud)O#{0*JN;h+Hom5yA*QZq8$2@iNn_fJNQw*DrI8
zL^=TmJGYvdX&3ZQDYupt2|`IJtf_zdU=@SARp*Z{9c=<epw9>^UpFoF!lyz%L|Wek
zd4<gm#=pVE7pP(F5WqFzNZYYm#oJtB(R2RPanTFa8*j*7x~GN01G2@I{ejEb*#b^5
zQAYjno2pelZpFimA~$oo=bsFAU+YM_ocYPGmlw*tRyewcEb_nHD|pZ(#`bRqg^_>1
z@MLv+px{97->)ey%)fuTSyP8hU|V15xMfA>qJ*@})yXY!(kO8ZbkE`B1q95~p4QvR
z<Mb-T?>aFpu8sNe6R&>`Lq4+hYVY(N;w~<bTZjULlFwu!-CIb0S46iixu>HqCPdKP
z3ml37=in?qXa<|5cE=GOKDrtxf??5De3Dq@Ke)hUq*8*P(!%lrG1d5V?mz@~8KN<~
z%&GYw3;P@UK_Lr3yCz!d!^ihrhL9GDf<iOJ_F!+sI5x<!=uTa7N&Wh$>Ivj+bjK$|
z4N^6zKTZhBlu;LO*|%=F*K-x9RD^8JtcflHPo;w_ZjnDfAwvwvebN{?rX{obg|Hxt
zP=T)G)dq7hEn)<^vnM~}lnTJxtz@0}YS-l5*um~MVXZ1>&wS<J49|&OtH2Jj+RG{T
z*#+ohumgL8pU@FrmLO`d3}P*nsTdTpfrkW=koiU*d>;;R>)-F}(H!cJyPXoV#6KIX
znf?s71|LO2zfA6RVO*=EtUm_MD&PFpAhJISoVS-$B>~{;$gHfASl(mrFT0Ze(FG5J
zW}{T*b=iM#)_i33cOnj23(x0=iBsv7+}IZmQ0#R72(cH{vO>Fl6?{weggP0U<fI6_
zoSA!oebs>j(o^L~*d~jVypZY|e`GKxEXlw9;XfLg3lnvQN!H`&LMiXG_Sn7#$s{DC
zSJia7v;6{Mng0B9qiC*oQ^HWjjQYS(f4*Qowhs_z1uXV-J=|rkR(<Qg!5JDD&?n+H
zfcqHLaUE6fOq_XT718NhIun&)ECWF50QP%_Bc9rXqvMH>H1k%^JSJZcW;Iq%E)ro#
zZl`SioWT5U9vhv@)d`$`(6V^(=*&dKtN8WO`tuOai>08rIb(?O=g#kk5nA61YuR8y
ztb^The(x>l8nbiX+LBbA@gxt;7Wq%|VKI6jw+qxae7DR{xf^T$qC=Fo_thJz*$##B
z`n9#Hj4aUGD2ztoW~fuKG6<~o9Gy)t_XoY7p2aAY`hLK6WQ>#)eGwlD8X3&MWbt2s
z9<0(Rr@yw3cc{^I)WQPQp*AJQvY8_~3n3#@>CXITHC5F$_P%0qPxjk{+gcmn2zH2n
z8GcR<zA$g%tN6$dOviEQ=X~5}_2Hq~^wthF_azY#8g219pnDQ`v)^5d9yb|8ea;bk
zDUNxS`@l_V7#3_c=;&iaD@{?mp9(A*-^OJ#Ffj+)NX@}hU{ww&=bmaXTGy3d8&$Hp
z@VH+r5h5K}y9RFS7h+hARwV^3B~R`d^=WVU?`|vp=?XfaDwOm_$I4*yp9-;$f#q6y
zJzhe1;<yb{gm_Gt^;IswBn^Xo*reZuM_}OD$h89h)qwfW2XttZ#&+lNz@&21^RTwd
zot4<N`jTQFbknT1DW6yw-dxBO0pPRUh7k5q^kqOC?D(jT<?d{CT7VLhxM(|zXSFDy
zW&jsb&6j%Q>NWmbY*5bRd3x<LqWoU`kJD4NuxwU$v#o1V##I%7lcgRD66yl7bCa*b
zjHmm@U=_WBw?{63EOrT$cikdMZ`mzxCE`g(|GEn&%n>P!ahZt@ep^&An=zr!W3_5Y
zQNH~QadujO9?fphFh#0HYI$b=(`l!f{XC@#L8zh72FENsXTqv?x?F#_Wi02CJc=rx
zk-)1>@_>G(j*y5KIjBrV>tVVh&yu^2Z;JlDyDWDqEVx#qHL;6pPRpC#&?=g^q&^C4
zZ?lmgx@?~hFre8FJ)?{d6tpEzJ!yN|))}x5f4AgTO!HB#pV~VLb1EkL>V;<86Kz#0
zj7<}}-$>S%e}_Ur@hSZ(<^DfLwk7DE=DCxUZ?eS3UU$atK9eQv0j&YUm$nb@XH?rF
zuVE^~7Re@Qtm16YHt}wpRy!^Qz_+p~S-VI?5fz@c(f^pqSwY;lCq))_$%n?=|CUsl
zOo1x7l)~le0e=dqy9!yVKUg<dos-n|Fin$vAVT9ZiRu~~3Aa&8v50ZAnKJuEtMDsS
zR2X%{|C^fNO>)I7pikb5={uy&Pb5R8fRvHz-7)BxBOs8k#m+bEW}-~OcN+WqX(#8g
zj|*nbbn^;r<97se-_(t6kNQU62waylH%)v({lbo$k?%}UcC#sTI?$^Z-}fvfN-c(%
zo}7kiEiAGh_e=v?DqM=0U|@c=En4vs0cuCgc3XP<q6YIKT-8jPj=@G>|L#pudgx6R
zb;ONt?@=Y}{lZ=NfGj{68#cbH;#^rRZ74SA|3lco!gr=$;-k-H_c@X9ln2bH9q7z`
z`9L7+6j)WNgrS+l>2E+YztbV(kw316FI=gJldMOkb#S-*&QPoL8osg{(7n=U-B~Sb
z%%!m)9FEy4O-MLdDz4YlRal8I_s$aQxK;cKZsL;xqi?piyD3|PA4x6%uj))Zoe=)&
zD%5*jH!KDn@^HRfem$)7whZBWSn)bl#W|z`(uLpYDnmevxU-E!OC3KpO&^*7WyAph
z+ib8x2_(g$7o71}iH%ULaq7|U;M0E_yW1O5m_UvCbTmw{0ARDw!WF|`Ps<>T@8_@o
z*&h$WXqAN6Z?Acu0&L(zY?d>AYEH|tsJ}WVO+ySCZ$s?}R;_2AiPI}+?Hq^OupN_4
z<78wAIG@+G;_ZISO&k(;hHN?L;J|G%1j!C;TN2JZUR?o0e8)@lPT6HxU3?<#Ycby;
zrY%YG<nGAh{S%Sf=!3}@0-|VWgs^lT&S(6PN}AZ+gYN=Hw@xQKFu1Y3doI!S2kR^9
zmyL+1UCQ>cNNhy(S#O_a;3l1TT+7=O{=Z996uFzS7u9neg?IKXGMeP=jI6ULOyV`v
z7iHXhE&se*XCf|P@<%a`yTmsv2D|$!LS+hhK3SvePx8*<>F)t9s|l=v)qlo@;^^0P
zJwtR=1P-+-t>vAyPH&MM9{>uU0)HBfv{4_DuaJNI9@>NoWK_O@xFgO8^#>o5O2EPL
zB<~s?4Hwoz(HuudIy#$8+w}iK(|Lze{r_+L7)i26PBx*AayZ!|yU47=QBJlfgk&Y-
zjl?k`vS(zSV}%gMD&z3spzQ3uM>fad_x}FzyZFOj=UlGq{eF$-ecz8;hXrO3DDu)P
z`fP2G8lvX3HEjdP8_rJ!0lHWUn+?AZR581EDcU*ggY%HgNb-8gaRoPHtxqhe)zLZ5
zje_>F&jK^a2Jkp~>v8cm;9M+UCvi#kyLcV7;C(f4VCXI9#xk)a<;r+mdO@nqxl_aS
zyi_c@rw1G5T;}3zT^=k+^~gttwt;M=<>_PGDw?SH*P+HKqBTGt2tfI%H0uvkzszSB
z(tbW%If#Bkq+l7;0(pZLyrfBZvP=gC0(Oya;?KV+MYq1}S8l?U^w4f~VnXN2Ty>P`
zE#USR!}e+B#5T;2$d@bxuW)8tY3k%*-4F<9gP46?gc_+%Ku58AL_{1Fl4svG^o}i|
z1}AEQs3~hZJ^)GgBVYw83}78=M0v7aDP7Rk$ATi)kA)Qahz_^<@Mia%<d6r?=i-aa
zu9>l@SdK^(@0@E5Qm`udMxYz7spGcF!YrESo0!ThRB^u+?`DR*>>Y82-zVxt@45Os
ziu1~HhNIiZ<F%<F$;ruASSGpxpdBf;s7+g)U`Ty>IxrsyNbZ2WABskJ`@51-^UPuH
zKm2}Zx3!*2M&p_1yFFpuuW*(J;EaD<ZaU!jxO?$U655-r*0T4#L751gbQ6Gc05~hZ
zSZoGP4sOS1yB-B|JRWSjq933KC}CXx(K?eH+D0mh-*XGLcf#&_$gw2Sxu`BWgpHqY
z{7nKdlV8%(%xHN)4|eW@hxSvpwGINHuVpCA8uh?4C|quuyt52kEMIJ@Hu+w(x&r=}
zep&lDv<z13cCFtf4pf&p)3wd-9E?~I?}YUiQoFj^7R<|UFoAE{(kyX)bAFyUf5V$4
z#d<ur(Yxl4^ZQ18n!!kv?%4LJeWXxj88B9xbxpVp`j_SJ4OO{2#VJr`NU;XY%Mv_y
z>A{5NXxMLk*f|7K-P74tSrYk4BIWCf+dfKC;7o+sm?J$|h%WuXq07-fIsb1KuRJ(}
zY^9QM>uKc!e=G{h+_oYL6!AJ7SktIy*WZa&4u?gR+RIb>Snbh`1>O*07GOQp0Yq;_
z%9IQFz8_XA*1N!e`Nt)qMkVHyypFt)OB(e-gytAIEa_p#)!TR_L!SpNcOMX%$$2zP
zOiY9UgAp%=5YuPddivqxr19|Dk>qj^0Pl#)7f_3ikk`;@slY^Z^4x^H%wd1)Wy5Ma
zKd#31=i9yOkW!?sQ~F=R&a(hGAPwShYQCXfBOz@gU-ld}ekO_eNZx#WcM;iOszz6s
zKAtjnve=#$O1pwnSQg7yx4bdX56&84T2iaVbb|jqOwouAh76&!*3wElDz?KauEI>P
z*(ziq9L%e59-Gk6BmTZ8nCmYe29XEVkcP0RQ#;HtYtROL!O?CN$bk9-7>wln9bf<2
z4`62@AJ-*~wOe=3JnEj2hbX)^V|x<^wfmKzuRoz)o1BSgA8jwws~{EN&&Q=tE0{<s
zA7Md`9I27YQOc!vuRm5zWBc*h7;}Y}=T3Qb+^X?LegA!vmPa_abKs9UIr_e2Ga`cv
z(M$7t>!GO|6HR)=$qBI7qh%2tJ}G}_q3>=v0{?JrrwNbWB+V|q!B(3Lr3euYITYEV
zTZ3wS5CxxDJA#}g*=4LjK_#Wo*K*_oB&^BjAh~cIJf~fdn$Hfh&$oe-bt_fs-1LDh
z&<*EK%lNL&(z402U>$cqUCNq7^=(Ly0v)yU<Gmj(I~VZ*fHu{kZM@{db&QpXdcFSi
zLR@UFKgrl3GT6@A!CM?Bwp5+)m+QtP22rqoOa||VJ{QMi{)@@gLzP3#bmwZ9JUPC^
z2uF(ouvy8Ns3#Egpxfkr+oZhYjUN$7;OZmKQy)0NdpK`1&CeKC<RJQI;$sc!>}So#
z&uJrI{&f$iRK}Jd*Xv6hu4$Uw4W-+n*6Kx;Cd7LT(PU@~<cT_ee}vV&$}K2JwM*Wd
z%nrO|UkVZn;CqVvOn#_FO9`BJCdgow26rZ7QX{1?#9EYiIMSl#Dk{_+f1+3j$1%3$
zIh8Z$Zt8KkW%+qakr;rQ#zBhlz)=XjKp0dL@C;`K4PN`ZUTG4z_8c#r=EyrcxB#k@
z_s~1{v*5<sTVnX8b-eWcJqSj@q;w{T&lY;C-G==DG#sx9<U%}^`7|RLAKm}SqrgZS
zm|X$|dBE5ATAXQJYqTIu%)>Zd>HJO}$ivO;gqjhulHeD&MnQw&SG8H!X=#I%*!}^s
ziDdCD=PBkyhay8v%t!L+Uwr8Jp%ri2H@byaxjG49f-H7$h2%9l2%iRqbudqG^S6x-
zBD94T*3q<GU-m-ib15n53Pa;d@pjw1f!F#LmF-c3=8;7mP;_a6j?AOy&trOWx{~tQ
zspR$_B)Z6e?jI!gUn8f?!iQRR^#1_%)Ra%V^-dh0Q5Mg}q*;HN#zeV_p`7f6UklSe
zuEwPWyW)#mKuOtVqkh^g=jPe@zw__>u1t2Eud%PYcK1Y#BANZbQhRGLhqk!OqHW20
z3d{5>unslIj5A!15p|D28v&9luZOoG6#g{a3AP60d)9Ly|Mfu7Yw=Qp?*I?wEg#Z&
z`n-)CYc9?c)TaYzXScRGC{+E=KBN*RxBd~iH9KXj3LMRbGpdFaEeh9Y1e?fP`1#p{
z!VkPoiOuBw0citFe|~~|LUoDW`6>Z)0R`4&lsnWnj|E&12p7lv=R4;lIYel->e+7j
z#pVU+gupb={i@)PQ1>%HC-1`)6mq}af>LwZOE45!9D&263qV32$Wux^LD)LP1U)sk
z@kmsIaZtR|5#$NW**<ku>GR6G^ie}i3ne2Suo6p{C-H;Lf&3T#M@trK+QNY@1>r7A
z)L}E!Di#vW=cm{^RS6m6#nppI-3EmiJAO)R0@!p&@7zllg8&}UHhHB8K@@5%dB2Ei
zDkXV=-7H|AgsNq4_8S$eaw}m@2-N3mI(8WD^<mK_CHw!u1eN1n^+^&|P2hm715C)*
z1ZSt|QuzopX0g)-8HO7l+xd9Zi#&K0lP4~?rF!1S50K9m*9W=_`;5TjRkznUxn{p0
zG36HUsB}$K6s6A4^IOvzr3X7LZvJ+kiy{L$#9Q<6Dn_Swz2Oe|6C>%pPMu-0Z$0aW
zAF(hm98FMnz5m(nKBzx>Sa@3!sjjz9KXaFR!vl%<+b~=%@uWyWqIJ}3ds-#Njyg2o
zM3WxML9YS*^~qcN%dItj)quOzhhcm^N{s3NjuJu<?y8SBu2pTfAq*t$BCFFPZ7->P
zU0GY<u5GnadgHHWk}_XYSCO>uc(Y~1?b5dz;9E}LnOFX_4pZ4y86UL5rX}6$+Rv7Z
zeir&igUKi@W@YLv=EP5$(V!{i-fMZ#em?mvY3O^BFHCRBOEbORTXV#REV*wtUEU3#
zhc(nF?1ZC3g+f9}tyy+6?9Cp{mhftJEUNnM&p^4PYK-YK>xoS*FlWA;rWr|^Z{j9E
zsrDWW$+6H4=I(;}3wJERF8S0qExfD>eI@AZ<Sxf!FHO4WQJq@_LTZ@XkgqssW$q9B
zj~mA8k0Qy;$jkd5S&9IXsz@7Xx$wh;M!&Y$+cR(KUF=4x_9^;p5+GZmC|yR{U`fgF
zmfSm_>-hJt{!23gEfP^@>#_XxYH_QZo9XVf$YgNEC%Eh2E6QV{RvMCUG1zta6f`Cx
z#h|n#HPx1`txc;{azUmx#VERh|FmqozzbWQwi-CWP|<B;U}0Hoa#RVFEzk-<l=}v}
zQt9<4N6f!R4lXV#T$~L^@T&#}Dp*m>PLsi>yZmBLwSjaFr`8Apfa8IqZICBkYjwWb
ze5PszkSq1YE$6uRE<LjOY$p$LnF6M7*uqdw>T(?;c{R21FI`$oWRo~pjQ#F?x5c<m
zY%+$=2R1Y5ufF{z2$&Jm9e-jZd0h?NrY>fQM?iz$UfDE%Tq{Uy9P{<OX|?3900f+#
zb;knab%Iskb)k;|{F?O{g<wJ(Jm0=n-U?F@tWRGQ8mNSX_cPFx=j#|w$tH)s#B@@J
z(snBbobJ~)pDkwVsl;m1vd=M3*kp0WMMC0i|N5+)b%z7x-DvcA>)>R9e6_~k>CRTj
zsG$x<ExNV;Qj?&!md6t+msde&uOl}|72Ngm!so1KKF?Mf$Ff-vlJ&(pEL7D&E6Hjz
zWtx3G1?5sYqu(Ycg7z30;kdd4+4f7+KF%~!!mfGJQp-@%<cy%g-|yzGXTQ~n)hvAJ
zsXR7U2Wx?>mxyM|rG_%Wo!PhgpuJkKw^)(u1D~MUt|TM=-kT>wmX?-TmX;<5ibm4`
zM_cE!O@dv6SUo(b;d)*if-ZzVo~6E6nPT0&MAWll4GwL{6zgP}!)AtU)BR`UYQ^{H
z?E{+IQ$d+QP<hGQMb+i|G66q6?h{#BhUDP1%)q0Um;J;zcxDvEvsD5D_Eyig&Ly{4
zK&PDp9R%h@B3l~#(#6GNWb)uzEEdjPXjJ}!iezo1p->DWZ?wH-q+2E#q>5tday(M0
zylD$WiqpMd!y90!pNGq5s)4kTp`R0BEh#C?KTX;rQN?)$*T4c-_wkGY8$c&RD7*>a
zwmrLle&}d<U!U8D2QsN|krN_WpOa$ha{ud~<;S3+Ak^Ti4^M<UoB!>((1yw#)yD5t
zf@RG>>h3bOp~O3@+ldwm{jD`@Z`i|p;|2?omL7^pO@V1?_J%E&99BldHo`qE?wjns
zBk#{9&j<d%<5Q#b%?heb;!;34dIgZ(=#ZxGnHU{S(9nmyn!+O2lU7&i4GlntmKO>u
zho?G}6uiOm$fjDb07IW+9GAaPB2&;Cr&Y{>9b0w`^bLfI90e#&ZC&gyZ=avVZiA4n
z(qa#}jqsywS0lpSfU)IB!?h{~8J&KIl8^2b2yg?TlKN3>!eK+b@KOE`uwz+oBV03#
zGi@51s|<qLu7u}o*RxZGd`QU1QPTsBa*j{}UT(UPJj>5FO}6o=_n_nEfygP0OiX&F
zmGcLl?2I&@b%TGP3zI#4273Fo<YNLjeWA$dcK40bGNo(@4|v6T-MXaK<Zd(Mi&|+$
zgFy@}#K}@(xzI}W0yyg?@c+EJM|6d1U>sz{KVuL!9ErFsNdsrtoFS%UEryR(DxAh(
zUsR7I>L{or)S-s(3?;sr`FqNNf|f){d-FW%yt}$fk|aVP%v?GQ5uN5J1M2=Z8>OWb
z-w`jNzlXan)qbd%bc0S4R9b2E2`e$~R!3-0OiF$|RlX!J!}wLeu5_%V{&rB!d-k)W
zF7F+2@DzY2FZ$n;=mz}|4Zi+T8{UNZre~(@wmQsXz6)EU{H~JwDDmXGMgxU?&4%67
zp%vS<YZMW2lI`Ffz1|PR7(OlRzH32WRxpQx=nOJI>(-GvFxk_y+rAr#oSWsT{?or6
zZ=&n#=m52sF*$1LI)ktL0)D!wpEH<->p+clg7GSw-kQ<Bqb)&=5<B914t4|16W`Xa
zPW#txE57vPdTsrS{p)pI2G?tv(NrwfD2_OkZX8n25Qbo7yBvcE9d3xAc0lV=2%*G<
zHd}0k6v|ho+SQ_iEZ5TZ{`{9iY4?0B1>b^I2Z}U2y;#K2*y^JKzE4i5&_6yFYxR@h
zzXABA?v-2h4OMcuthQd`mPL8of50?hd#Dqw)1P+b6Y=q29?s>s0a#xizjlgm-J(<J
z>-H7Tmgp}3Y##b1PD~Ncg!QRvxWmjw)uuMXN9m)<71mVO=?phUM1ND=3Y}fI_dPEJ
zE|Nxto~8CV!%Le3#Jw!-(a8fYwcJN`oQC*+{e?4IVvIC8<ZtVzWeM@DY#zFfKJv`q
z{2FWqP<U~6;pq4IW>@@tk1^-7skyC+MKnr{L!Eg=6z>=6krH>Z{o;AAGCc(=q{0N{
zu*g*$PI~t0@sL8w<rceSo#co-9lkl>)5+XDDVd>dQh;?j^qyalV)^Z~`^!!5PMii}
z(TiK>Tdo%ZdJH{u7$=xSulw|+fj2eabUPFD9i1=sNf1i6Yb&1qF>9mq<B=m*EkDUt
znNr>={4T7_QLhsd^sq&(MSIIu5@k&BSX)h8{>S|!NtK5&;|_x2HgzlmI-L*zqwto=
z8htg#v{cMYcv+eaZ{rc@?R9b%C1Djn_IXm?L{D}=Bcp%UDq`=8VlaaszFW?1>TKOD
z(0F*&e<LCjvk?E=3B(`fY*U7$B)QAu>l%qDOY%m#>jW-NOo$l@xf_33viJ}BHtz8F
zE#R^DZ)vIZ-|<W%>Uz*SO@O(IIoIwo9W2qSPQq-&q&VZ<M{#I>c>mlT4UAJ^3a<=G
z(&@T|!Ctfz8U#6)!%Fv*RVSwO2f$`arfKqGYd903b++)UfB+<jOMvZV;;;+VtOE-K
zf_hYejO@hb$-F$1)jCh#Zoc$5-{eg%ga(X<^%kP{>QCXP%wf!c7N-NXbX+bv^>_g)
zf3jWjL8VB2qyC9i13Z7i|NQKq_r)gOsP%B;WL<(}k^mB#=AmI#fusg*31WM-E<8WM
zAGr9Tx?!=Cvt5JVgEJzuP**Edm73oj?jE=UK7fxnP2JB?oKOV_u-4|+!0Of>jh+P%
zna{GfjSa*B;LZLh42Dl!f8Mi^apT$twu1dUtpA#f5SLOU*e&@n`3a|lrAm}VbJ@w}
zQ8p+&ogA;u64PsgM5yggm*$)#*`y__k&%PA0-?&Dt04+YVri(z!Hm1ZxyY8&8M~db
zb)nA#I0s$omn$j3=;*?w?@0;5VB$fIcB<ki5E>4k^L0f#0x|GMBM11+g4myagq~&m
zo)?jcDb^XTB8eYmNS9fhPur09vWUenkqE-S5XVjH36O!xJCG2a$ftV2bU`hn*QJ;R
zcO9>l(_x4yS5OOu1HaS9X59hvJ;83fnt%#yTIG{xc?(W8rjC<N+vYnzeW!Wew>;N}
zh5#?Jiyh(|4dwPneHou;&i8(3@_B)G5qAW1R7TTu;q{Z|uR3@It)?eR2NGmkT~|7q
z*ueRQRe2r#{4HGdRO)rXI3I`CK0eQ!w=~{*yH{@#k`@z2UfEjg<f%D*@%HyGa<q9r
zzV0k(ZCgCoh*us5`x8h2F>;3ddf1{-9*BD1C&Nvtq!F5sEFvBzS#EDh2V-T1n7SNu
z*)B1g?`K`FHZjRJ?wVeZzln$@yYM^uMGT4As_APD`b_rwgnrZ4eyGRU$x@V%Fx>B@
z;yq(GY97rg(N|_!CvVy=^1}6xf{?L1A6RwZQ|nWk4GJzzb)~3^MaK~0UP5vNpcXQ(
z((ka-%ffPRi@$iq!RK&j#LBH#qSJTakV*JCjpSYJlDwBR`3d1<^5$kCDSSPUHefr`
zCw#u*8rT~Iv%mz51S8^Z44Z?3MeZfr%boccGYWw@6Q@h^AD(`g`Sk7n6R=S%t-Bsh
zAfL~^l?O?x$oggwdkgY!kKBOe`Pt{<=U1*#hPAv|GnDqgB2u|h)Gq0Dhv*o<t(;x_
zvWiko1w|=E+a8n}7gdAb`H@HPtDG!V;=4)-h&;b+dd{{I!81g;=sQI_));WwyAsq6
zZ+b_@4X3?nYQzJifA!q5^C!NX6W<iZnJKlWv)eZ>)d`}i(s&-z{sc&peS0AIu)tSE
zsrfB=H;S{*O7fL#BfrlJ2}{7&XK!`s^hS8^aGv!9`^{*95B<K&^>DbeVUNH~k^oDl
z{>e?fJIZ6(S#*IYDq$_}1<F(i_cw}~jEcgrCo6O9izjl;XG;f9%4?RN<qi1EHpt}?
z93NNtqCsV1d;4(8nIXDEhYqQWc=_2NE2%oUag{<Cti;QBqw5RK*h1yS^=O?OR8FC}
zK3UMzR#zp<jBGE&Lj3alV?CFy`;P^mq!;==^nC~tZ-$=}%y4|M#=Un-$IMr1<1+gk
zWnoSNIzp6~d=uIN>wGblY8i$WDHfywSVmHJUgUGGy}NQh@>lNhCr=sEzP7B?XQI+Y
zka#;_%a+qtzS@&vNbOl)Vk4!j*g_geXrYmg%xZ^be!jC;UypV|2~4*$gnXt0X#g77
z$hM8fkWvMby_X<k`oFgUJtEsR>88Ju5}@MD+HYRH)~y$$!~FIg(U{Vw?Ln|1uBT93
zU&5`FY_~3TFR#2`pMR|qe-mLboS$TCy}=Az(OPdPFo`T;WrbqRhQf$4PDnZjN#~U&
z<L<8p*8fx+&u>!POOxIDaGe!_*pp{melflh@UT1W#q9-NikFFh!(Dh}!A0^#xrAk4
zb=_M>o~W$5$$6$MdNA-Fa{`Kv${YMk_N-Oz6Hg6#ZA#m9kzE#^cFUv{e2YqG!a5Lc
zG%~@E+-Ic>oJ7`(?_9}daA|lM_A-WoId8!pbOZ%9HIk$aVi9xwM;0JwStU;fo8klN
zc&_-{X2!Vg$~&-S<j^Ntbv2WR)R4Hm8g}bPwy!FbFkf<WMamrwx}|v;!CIBt+Spi<
zucg}N>CynWJ>28rGiX!gxOcG*aB8TwWjMo8sI*rZ*VPTs+Om%aAe6Z#zBogprW{kO
z?5Ew&@(1Pd@rTAOkeGgt&!x90gK@ejmc+kl!CK;|bjX*AU1o9b+1)jvWq)}d75}pz
z*-3KzDhKJCoaNcgKk6s4Hv7HG#{YdY#`s#Q4<QKGjLX2Bg%98-7n2TnrInh#{r79Y
zjFY@IHwH#oM4KP3<B-=8-)>Nfca8GQdrKMW%_ZEH36y4)%E&v|5ia|_gten`Nie<d
z*`wd9B;v?W*MXxFpwj%pJU_G-<res_DjV2K+&;@#5=%@S+l${LG0BCuelJ;XC7+*f
zHAU4-o`}A{A&q<W^Zo+1tfQF3PjUA`%hn-33E;FD@0^B(7SXXBB242N;Nml7kvJ1(
zL|n&)bPMvfKqZ*OBv8|c&s1$I3Vorp7Dc9J+Q|m_S{@Qk{>c{#Ig5d7RXYdf7Bty*
z&HuTP78wER($%udRTL4oOc4)U2D1wl!Zf8S()0|>oXB0kyM2g-T#xD1E;70})`C-a
zC?|^IXO(kmr<QUwl~Pv0S3xdoaT&2RjnZ6O%tFBH4We-VD-N~dNPVd^Qx<*L2QZu{
z6}Fh7;dyav1!Rt3lkD;1zDBlxnW!#L$EKGzpF3o|@GTkN>`;0zAb7mesGW5bF6i0b
zV?uD3lroZnzYT|2SB|^b!Sm@pf5s^!N|%ZmaJ>?oJ8S?BhPoB<N#sDY;rL2B@A-a0
z)+G2XMalGA6kNb_dM&4pE!i3=ms|-~ys}j0W=&sN8Fcoa3eh<Y=@m1IxO>BIguw=U
zu!rQzn@$}jEH;<#*djc8%$-vi31HB`t(mk%lA8{P;#IOTt-Po59VJ#j4Yz;Y|Lqno
z?_Ck4B5%8-X?d|r#$W7CpA)(z0&6BSuYSN89+dMI07)_Z8)->NCtK%B^8|3i8L1Xs
z&e(Isk4;T)5e29LFZYwPv8XEISY<*ie|h1mYg$smr7l5rq6p`WH#^cQfq#s+;~(F)
zQO{p;R8av_B3rI*Zsp*WjxoITZr_)8;PrTbaP*V3V)6RIF^$yhDLCh<#8+77^|OUO
zPWc2F{>c0rDOCaq)7r{wbaZieV>IyE*(}I<rql1RtwdC;-6^8?{&pJ86tiBUeQLq#
zw;p3yM5mOX%zaKx?uL|zU}kk%SNwQFJQIPluiQ-~;Oy+*Qd1|tN2!Hs!};9R5f{l<
z#8@-vL*_0e&c@hYDI0VW$b@yt9&v=P>skXbkiMD<?!N5ChIju?XSZa_uNK8Sa?30>
z;AnG{RHrv%2fml&*C0M4q*fljo)E3P$u5#f9q+<=>4E-21DhlQYi+Gdlzi!g;W-0$
zj+1uRQP2BQ740(}?$(6Z@RPGQRVB{7tFyDSTdq9s)VP9LFwCp+ys1!=+Ibot3S*!b
zd%_m(zW*~E8glgVsk2KRt@X9h@EOFo=hFg59Lk(F8WSTJiV02~BPGP+?q)YVq6QHY
zQ|V@18klr<41#h8?P>9c+<HELFgbbtlmnRvTDn`^ul;tPpq^%VC+OCz6uLPk`*{gv
z{2TX(*z+yPDXgW$h(`MIm)v~ahk%)0#X$0jDn3yEe(XwkMae(~TBDYo=$n{vPaR`<
zzw2?;2a{5yR?cD<Go|0*m!!S5)a!99`_!Fb=4<M@XwEmo8JxkHH{DiZ2>HX`drtYA
zFN~+_HVdTQn>)cksJZ-|?$cz|3wuRF$Rj<$D(UiYJb~Dqthu2Nz5_po0mafN1f_t2
zu#mTKNN65P7jCgKS95cL_(KZi;1KR?QSRnO=Y(mzGAbs}(ev|?3`P@NZBN5(cvPG5
zfelTn%Dn{GGD<Q+YE39Ts3?U!977%Yk|+=ZWZGhX!Cs_|Id-9_BN~p<i92&+zVvW$
zZ4>p+&^i-i4&!}<SRL?hV>cZ*E=(?OUJG-HbCA|jV;*ASpSl2hq{weGlDtHgy<aRN
zX!?<yyD>_U%XUHaxd}~C<C~%nn#<Og`+WFSrg@jXvm4lAxn3J1vodyRhToWfG|9*4
zz*f)C-(@i&PTip`JMMt+oI4Z`9cSQVo(7JO&s&$%@gy-_AX?|CE`n;iVJ~h}-A}%N
zhVm9zj#GkSj3tDJlP5PP$sm!G2vXaY9Tf|_O7jykhV9qvJQ<dbK7X`jnb~;dY@Pr8
zJ9`UT*i~8vef=+&Iuxc;Vvaq3*L)7(?_$4HVtT_?<F2tSYlYYc_4j*%ZO|LOPX<k5
z6iP|t*1hH1fbn;>HxEr3J<1%*D57Z_UT8Is#o#Bst0v8-CkzJO9KAjB_XGZ&EX{44
zZNy)eQxv+qaxa`&_fTI85<NIkQI}xEugb&67kGR)-b`wjIG=seeAW)S_nFeGx|heJ
ziW}fay{s71+bOynyPZ8h-8uc2Egp{L_)=;=i1}GE4E%ABfBVN!mI1Zbl@`e*1dUL0
zQPu{LPqaZ#pQ$N)&WYss5J|tl%o`dbk?j(AQY=9b><z~{mWB8rsRO=tx%WoIL3OG_
z;t{&oFLc)8yhRqdcJx`#tPf0(G&N1zVJeX(8afKKKk+O%S;+LUI^V1juOYEddedGO
z^n)QGfT5j_kk+G4nLZEnh^Kt-(NXNk5y_;qNWAWvCL#mbtD*n0!|4x;YhUl#%zt+`
zj|h*fQP-Y}$U1Op+zvG<PW+!a`&@MP7_m;Mc8wn-IW`4m#HSTuiyn$lQ?jw~=U!zQ
zOGFcccx#uqL$EJVpODH3wvg08Co!DHm-aOvk}|IseEs_s)bBVUX({Np-S~H*S8UL%
zESR9!_a26Aha^JA-H={_<E0Xq9yYYCE&o4HWkb-QqdBeQpr)@dg4Q*JtYpNt2znw8
zX%p;NP|N0W51zg4?RH1gh!mrfHF_8`=+dxPx5kQ#o3<^L@eKh-pj<K@o6Y~{0LY0H
zd^Dp&N0TemG5BNDnufac$G}Z5G`$K9)6K{ate=mR_y4W!PAJ`34?*)#`MO!(|I`rM
z>OE><-gG>-ZgsXcH>S6Y7fH<Mjvr2Nz@BV^mV>~HOfbr*s-&da+^`Wc68{J}8_8iS
z+!`58?)M3BbErKE2SX>v=X>DH#?VavqZYSHvx3#%OzLO3dL(}jGp2CTcSDRKz2-E7
zgM43<9cRO~hH!2M((qtwYYRl7^+lx$-etu>Z``;nq`e=C5i<E~{-@IL>YeTc*D^Qo
z*R9NrkN+r^M4okTF4OBWb21ybu5U~G@)z~l=Lj%JQOm0<-d$M+v-t{?zn6?>Wll*;
z(=E+cIGa&*dF`FMtU5I71Y1t{EWz$?Eo#q-of$>2-ntPq4-Z4{f*;vaH4m46hs>z?
zA*ASNem%7-n(L%;T}Yt}N070YzW1u5Z@&NbyU28d5e-#v2VNLb7;zBI8mnR}am*?t
zkPiq)>vt(SB@eUQM}ZA^cJqJh@dL==`|GRkFSAj5>1%R^sVsc2jJWARO(T4uwNIt&
z{*0S=rG;zD<Uad#2<nUJXIxPWW7p>+`m9rfzTZ_ZyiG2p`B&+B1(#=S1|Afb!@biT
zfx&oqU+vE-9g$rfCt-&*a8r3OO=XPujpi6+@eyo;fAe9moC6w<dWXz~Ii;@tyu7X3
z8@j{^uY0?U;xI6VfvI8fx<T`j;ki*Zv8qgsTg>!M9C2%zRwMq(aM+?3k)UvoFDZZU
zkA8w$hY5W(S8j?r#U-P0!3-Ckn4e$0u5;$lyK-QIblZBlLrezHMr!7Yi@(oqJ^U9t
zx-ztJI6FJJ+y`u{PM1R|x>z8=L+CtQFjEt^;`7>x?fWir`wE;K_rEnPb#!a@6^Z^;
zTow|G;U4bQ7lv}dBGv!PkGwJATCeiM>mFQBRMFca+G6MCKlW6*T+CDPE~g=6mq2|G
zjZoT~)Mc+StR%fFw{3V3sWs<1@+%Yh+w};dPWx~|3;`JPoo1LHTeQDrml6wGj(4)~
zu;`YgzAQ+bQ>5+~#lPUR@+U8@6S4zN$VdFAs*MH*c*p-@elfpGx|H{D?eT)i>)Vi2
z+1;;P7>vhce|rL3@LethO<w?K&HNYpQb(KO1;yu2hzUQB=9kXK_C{0za+OuX>00{e
zHgecEZ3(bH@!vZkGEHwEF@?8+mO1<;c$%@e2e|tllrT*>t!0#4zFp;*6?f}_YBaZP
zGoXCFNF=F(Xk`b0KHz3h=Twm4R8aHQZdm`|Zyb$RDu20Zvs|g2>MOZ>Qpsq&-g(J3
zW}eSjkWSHKtjJS1f&RS7>yiRzY|B}VY6QFF<bTYV*r&zjC#Bf~7sp%SAS5`TzOG4J
zq?lHN&3**^MtlrG!!P_q!(;2__!>(`g_W_QcOdZ6)YRmX?VF%B26U<DP|7CkW`Vd8
zBFqp#4Yb<4i8I%>VGD`Xui;}9H<vEIjr+t@vFF&#!$?|viwT^W$`($CXhAv+{{992
zgrTYva?zO{S-e6^NH=~y-W$Oc+jHM|WtZ9I-ws1{>3=E_dX*Tb2+Us=n-@Uuz*M83
zUW_eg^V!@%|53SRK;p8a0^jgzk%71<1ou*xR&&}V3{GUj(u4<B<g!p`;EO4PVYig;
z1v_LQ{UQe*mE|Yk_~IJYJ0znmCV>6rzTWzrC{51#r^dfakQ}tN^GmyH8~*r{xf4|C
zlcM~=$|bLd*LCRW?`wD>cDy&TE$(T9R7+h$q`<poUN8rUIhj4al?!??IxR{R(#53O
zLC~ePwDu238?yleaX4Yg_hJ4^j=S~_cfT$DyWOtE{Ag&&F|`1@UGGH<2j=Tdd+sCS
zM229=PMF>k<_u?glTRSrgTcU&8%E#iz^3@eE~M^{ldCEyN{q1^^gz<SypKRYC}5w+
z&vD!+@*Wev>hxB!{ZLE0zSB^p6AZ&#FSuA;^l1;GixJ6xAoKEZ5u`}S8D@<`8`Z(|
zcSi>AXiBk(=RbW@78M96?Sm~)66<f2mH9@hr?+1BVwgjoV3AP?iC4nl8{SHHJG*y}
zcLWMh&(lP;ct{lQx$F5bPBNT$qQeHLmwSBN0QnJ2pD}{&a`d_AJlm)}Zxu)n>HWg)
zTKiR7(2*z^b3;L>t;(GWC0!1B!VdbHj~6A%vo6N7FLubY!1;{H&_elh)&N_<ery&B
zAw#u2Dy4`Nd0oUuybs<!4mWluq!NU3tB1b_<#-GUysDWWvu{=c$a~Fx2a6*r^o05S
zZ>m!h6HO<x;c{+G0BKA7i^B40Ae7|3L28M{#lSicf7)4wzw&%k65X)d@Zki#4(EGT
zCAZ4TArL?S@=#0NWXhkkwHNrXY#m&U<(uCP$8x`{>?rK@@u*6twU+7Jdw0CE8a{o#
zWp!oObwHbwp$94%l$!5lfie&evw?P$JsT}0g5y1{UizU&_pc1G)X&s6T}6nE&d+`t
z!MB3Gwmt9^M?SX!lzE5=*Bd?7?<Qw3M_*w!0YuoaJ}ZI36{|p?pmlU0wBDVk!!gT-
zXl8i$WMSq9ozDAv!}J{tYJ)!@g%!ecY>Z#Mc#G2%x5EWPydATKzlZ3`ML##sH>IGu
z^qG9_T3h#;TMR|ag1;{~_C!;vmC6*-hv7H9tFAK=>U70Wd`x?V>*-ZhvFV#*Kkjk+
z04ws;Mn661bGGqiQ5MnF31K2%u64hBvv)G@2I#@^!He(MP&@b&1qP4K&rS6n>!8FS
ztkR!Ez}072DS<0B42pzT-;zPs2eYt#`}O(WwV85Jrf_!U$P;FF{A!j}<~{u1O}K2U
zH?&WV^T~_JeykY9d+F0)mRz4#L1V!C%7-o~DpH6;>aZ_dn8($5iTYPCL}lb&;eGy&
zQNdZw(mE6At35GA+oTL+DcBmV#SpRG;0(De>J5i1CcNPA($BdH_wMz3)2pD}<{;a*
z-jJ6drq*TVQEd-{MM1^BKqCb;*B@TyK#Aw$pcFHLL%XYQ`O0ONC&ygJQGkv{vBLI_
z4VuyUS}D!a<9X(<^J;m%Hbh1-vKB41lppEe;k#y@rO26Ltj;AYE-@k{ed;fLLPU?6
zE+TGV@JFKDAKWu4$;X1Ct!&%uko(W-Q<88pK`b`lDGlB=bf+B}U3mX*jAjC@N&bS@
zl*=jfrP+Z$US0Dy2|0A94_JQ}=*|+(caPxC-jd*q;jOrh1Dj^b+s@?;w92-_qN1XN
zhL|I11x{xkSI5Sg%a3QktS=Q?g9}ipNa1ycp=eK^noB!<@IF`BO#Lc**cV$m42s=^
z0z7cWm*g`F#`_EsB=>oR_C>hxt~`C-^YU7f6>0i!s@~rake7P}^hsnjHoX1U9`(e^
z)teJG0$LZgr<#2IF1}s(djxp+k#=5BAGrH?4~NSilKT&S`+<xBcTDhy=Z7f<l^EPF
zvJH7}oay4caC^#P+RYok4Ek$M=MGNVKQPTcoW6J-s4!cIGeg`0yR}<qy)6qXJlgbo
z1#=1#oUrXYaUK9;1hM2HG;3zQP&FSTo6Eq;D!;A4iNOw@9=}!IVDgbW8p(EL3ZK6#
zp&&wiqhsw>ejbwNMhMHTw2zvYxXf05pIEZHUgomRO3nD=c959z{p3H`{7BFD>bbW5
zl?V{kzyH@gwtwXT>bnUF_1r;QEn(eBjE9nC7@cgc#|k`1jfqt#C%z6fl7I7v%H)-L
zT9Co`O5?bzTrwxfm|hYRQY>HHn%y~{U$&{@?r;&W=9cRT?)QIPaW%*5c7`#ELEV73
zRu3QWJf7W52VIg)v8wAG#P*>*Qu?|b%-&)UefLcqH7)$DT+G)Y)`-&;VvD<I8{OYQ
z?FTO--*BI9%2D$%Lxc{0eoS{A^|tc=x6!^=wHHPFv6{$Ad7HJ(Z6iVKhVNWwR;uE1
zp}pDqLZT6+2pGZrPl4lM`x{~q^+NLZ609lW`L=R!4#o9PVx|<13Yr^18M>KQd8&ki
zEmXByof3k_;ZIS;lp&Xvi2bn$EeEqMn`p*}z_gE+#Hs0fQ-+I=1p0BbVQb91N`y8o
z_`6+-pxcq`EWZa*?Hq(4*0{l?EiG6f`Zk5aG?Gh_6L^eLE23VBZvW%{PJV-G-gPvZ
z%v>)oBBF)5WzQS(q9sA=vTiZYAyWVkkDqNesG(sO0(s={r9z&qw}Pgg*!pB7k<Y<w
z)KSH4)E3_s!KP|*BkLZXbqH2EQ2Aq~!Na!u#9QvTQ!DaTsO~XA_59>;t89CY3Tw9u
zpPHHiRHpg{0wrJ-S3a?`Nm1n?BU4#yUe?#&yhX=CUEsdrGoC4UrS9b@!~CdEjD9z|
z8;3Ms>AXxIXQ7PuKirMVZU&EFlP6M$WB&m1#kuMQsA}wu%j7fPP|()OGUuOjfQFT=
zgM2M!1b$||<J-H2zYI-Do)<fB13W;IBng;4o0R}g4_#2?6N!sikn^sp;&=GUC(ygH
z?93(G>8YO#^Lrec@37W6&1ZRh=F@w*Ur;-^Z|V!<$yXgVP@Zr{^fv$DfVAAd_pn<<
zIu=RP?4Wn^0p<3yFR=v)!(NTe<I__r6Vs{%GKulkrc@ttiyw+#SC6Arqa`Xe)dR}w
z@0f0>{#J3bhTYU!3EwJtrI(c9L>ya)co?Dc5<?>j7^phV-WW~TIel)bE$nno)m1l)
zCg!lxQDTzN#5-8NBvh1F*|Q{Ae%<h^4%!ER-9>G;gpeeK!m@4jc<DVqZy5>4v?NEv
z{7(0W{3;d%kiwH4Gtj;`<lqCh1+<uLX$2+aY*hquJ$l2RI-4aat4>|)v_7-ljVXhB
zKk2iYva$lpkXHIxn@pU=IsE%J>dVSgd4idyl7H#ntQlKf%@j_k7O`g)l>SKH)|p(M
z6>vc0Y(6J6tB!jKvLc{gbHNg}h?M7=>a3Mc_Qa18VYps*`k#KA#83SS0P-`k$}ajh
zlZ7_5LnLk1SjGztN@0l6_w`e034%RMHPL%LqZm0tjS6MX+*7f&E3ZM6ILm)WrgI1s
zt8sV_f2GQ`H-Po&7)DpkG?ZAV%1=w21R9Dw@n~A-x%NS1v7qo18WoMv9HH_vzGPJO
zM{vE0$jn7x_ci;KhAa~T=wci({uS5dEa>&U5@(9GfxAVi*LglWC<VU)SEpQz_I8RI
zwN{W=29Aqupdehbdhsna6`?1zATLx%Nc0OPRIr{|1GHuv{<IcHisxV*ZNA0Nv=(xU
zz9XM5O72hPM6#L~L3rrN84IU-2DF2|??(BDtZT<JEG#EMng+nnB<P)E)#a-x^ou;-
zN+ecCCM6WiBT7@5V+mw$sh;&WM?X}D2M;n|q-|}vO!P}u+F3tXPMP<vg5~Ro3(9N6
zYc!u+oFD(2JXk);PHI&bkvskN%eY8XNa=FWyZCmrv&#2}ca$#cz&(3U%3_fN|J);A
zDU=FPGE(~O@`eIi>CG1h(-)`B2jfvj*<il$f|$PDL<UL!o9+O=?J}>@cHAwHCmYPg
zc6nQKqrQphrxe>UEtg5d(%ELZ&&8?R|8kMpuG;}47iUo;(`S<-5>|~9uR1#FI-0K|
zxLdUtsK4NbknuHVe4qD<mn}H+y%w0|@R>tvjdC)(0Cf;*LB0=`d^lN*gG!sd9cM^c
zF+mN9eO@x>I~q_fMDDM;JN8~tfA5oZ<tKA!XHEL~VpMl|(>o5oLNvGfO5sy*z9U4K
z?02h0;dHCbzkd*@*-Q^M;dWq|S#2)Ze68iI_;+dRNX2tfZ6oR_%7(w5A~RLc6eZ&^
zMRlDGNKOM8<4JYfHCp1SoE%)8rbe}4&2*|g8T0Nd5)!>}G3#l#PIpVUExxcL{$YCs
z<*o3X!hU=|+t;fx(4o8!v#CzB>cZDxk+KHi{8J5BM=}_f0L(W!Tt@}`92w+D#h2BX
zGd&nlE#KMI)wJ7;-u)Mbk%VZne)iU7p$q~LV!@RWno+@-+G-r7LDU-|Ow@U(9I=DO
zL1bRht*UiG#oliUv*8m-xsMr4Qze=Ex;L9E*ibILmnelws!ImUuk2+g_cIGg1v`oR
z8azx+#Dp+`cR-IW$BoN}WlLf~E9NWV=E9}jcS|)#Sgv?reMs@^X={r^QT;A7a#H}W
zs>+m+?LccnfiM=0Dt5YmXF2j^ui1W9=ciGH@`WCXcPfJ?fn8S6N}L`2RE}xa!#_9(
zHQd^#Vzt&n_>ub!c_IGUn}1g*W)LVr?cNAo?xDC8kA(>5L@&koBb$}RJ?f{G-$xZB
zZz;ULE-wM;(aL(Jsiq}pFYJ1MzRM7xe2QB@LZVxEQj)x~L^Uw2G?{X**Bcxj^tD)r
zD<Fgvl5v#KZ{tQ+KfFHQI5~Hhz<NI5Y&By^S-KI&8%y|JmOGcZjX$5;RX%N@cFgqX
z3R}A;b)!h9m$lMFB67iEN13tJfyY=k|BlW#_`%j5*kd#wz8))QpIPy|=uISWS^Cp8
zhrhOkhDp~^C?jY}N7Icqb2<UijQrcaI|`0?2qmOaC@w0!syHgSJiGDij}Lf!-T@oY
zj#SD2Vn_G`Q(|I27D??lGO<L2HMxE8=d>F+VDP&hVmBz}n_AFH-UQFvwQk^8b(18r
zm@rIqI36m$m|dPe+gKZ`^8nz~+Mz}?+L!uel}$@AwqgYw`Y9DrKYgF%^D${luV-+d
z+E}2U{#}(zd1IE8zwlSs8=`ZwnH|mDepfKQsLUDp+ujnKILzC?>w<6*gYhW3%wM)2
zduWMLn>FO9<z-IJXPiHr&R5ZLor7Wr{U0`6?dV`A1vo0L`D4Xo4p3q9A)8eu>yGiM
zFg}<@>alS}#%)OftYF0v^&=0k=|yOz(PtwC;)KNx>x+-RUnxHkyZIL(jRft{nyYsQ
zTx$_6+HLdF$v99WnO{zi|8uhS<2udjrs*tF;8#%^F)>u6(<Q2HbK1uSqj4q{1@i=^
z15(xIzd8T3CDRE9;7bBqYNdzF3+tWWp%z>UR&ystnT=w_>VjnS-`EllRnUN6zDW7r
z`T6V~VI5sd?o`Vq8b&EI2JHM~y65x>Io!Z?1A&LukABq<TZaLk(ktGlJgjC_=dc(?
z^`y09DdtAJM1HI3(=a}S7?|X9*>;PFK0gdJ8%dDj`sASl+%b&;eM2uD9ZF<gr00Yb
zT)(rNR%znD{}}gY9ywy}kw``+If+00Zqk(>$@Q5+fTgz8D0@ey(?qIo=ul6PDW*T2
z`BrKWyR<-<jx5a<RWTH)Y<kb|q|kN}+Jewqp;CIHD=JB&rg<0__r}1(^49B(x~Rvz
zccM~nLFE_NfYC0s`=j_4iqzjdP8P<lt6e=4?EXPdoyDthPRl$9iG21;Lh`+iehwSf
zfH_oHexrfyz55EhtSsZGv;U(qFTe5c+^6(WMH~TBZs2L75q}doq@kaGO_+;7d}0;2
z3enSb%nWv{8vg-LaZ)sVP3M5Crb|7tP^GuywBu|9m?VvS!3fzozHfVan)k?ff2mFF
z`>=5Vx~lKHUUnAc7Kce{-CiKiI~k?ksJNr4Dvy^O23LqIw0<HcyYLuLsVC0^7t8Xm
zR05k$x_#U(7Ol<~C&#xR?n!Qe%=aAbj*m>iso^>NVEH5g>~YN5Cc)cgc;gcrcwEcC
zk_*4R&G94tX7X$+*}bfq^_MrE45|)$?VJ)F?##buQm+u%9~;t_vN<#eMVeju#D@2h
z(h=`goSs&d6xi(bp8ru6CB*e7?Z1YaL?@K!bl5iS#JlQ#ML8|Vqbvc*J@>%3Z_Ss7
zsCj~Q#eHpPM5#-^L_Y7G?LV2UQB~p&Py?QS&F+L>f<RP`%`d50c;1kr<SX<w5J~=<
zTf;R#`P;DQKUv(nU7)&-IvAI$LsXF1i_EH&^i>;rFwI$IZ?^oMas0Q>J$%6u^JB+X
zYHF?&I!4c4rn^*$<(bD2<|})w@6z+KV*fUj(8!mcENxjji<b5Zqs)xAycP0IoJ|23
zuHEFBw98Ib(NPR~>F-^mi+|y&(VELw)Ca>|Y0$&^qUmva^-z>rvd3c<x4ExAJwNM`
zI9}ac_?Y>M&nR?pNf1%J`ntpcHTwcM+C5pfSkWJJszxSQsqojAZQsD^!d&y|ZDTT-
z7BU>LxK4fQwmgTdQG!4<W*9A#TnDIYKkI#lHS-8Vf|KM>O0%e2TnLKK!3xXZGS>R7
z%vn5tGIr<@g|5pA<Lh2Ork{yvd&nqu$RiY6@akSbeE1z&o;FF2dnIYq38S=aP8{l?
zNrkjJ4x&!T|E`3*ENbgoIQ-G}+b8<Jb~df>V6`D3bpku>m69U%*R>w!v**AU)!z@8
zAPe{EuN{<W?bp&%KJfgp*&!cRWwFf@IWuNvRdu%>?#T~IPjwgpp|32DF%@YL8#U0b
zpqsOvHk*<<er&4#<wLOm>`P=>{=_!_$WW9RtH6otC}RMBNcRxm(NWgF-MNj8wGT&U
zC$2z%Usawye1g|mcYShGdmEUU^C!)X4Hb`+r}s5rSfvE+_<CuM9J!CuJEk2C?2~H&
z-;=XCa|;=*hsr%Duq3V^iPT-|Htp|s^!7!5iigTuH7XRO=df+lxxm-o>07Qkf9oj0
zeFO6?aslL3#85oXWMXVX=oe#h;$O_jzk-8%$Xszbum`IBAixedJ`6wd@sZPFw&S#Y
zb`Nnw2;sX;Aij3@oni{0EsNs0Aqd9!NceTRY2b)7eQ{1Ctm6rFCZ%=eS#A=&eSdA*
z^f8@N74x6(_LnGVpqvx`%M)@mW6D+yXuvmKgHc_kt?=f(;q_C{W<^P+(;<J*5gikL
zC8s@DNT}F}r5M1`&_k?)Ix*%|B{1TE=UZj?9@0ZZ##;JzC|3woRs3fk_K<PcInMI3
zvNG>p5R29}DzB`%W^n$vxZ3om9N)(ez3nS+mXJ=}efJi3W&I?+L)o1zbQBrv$Jy(b
z%>Sfay7WZz#nkUHhmCNS)TH<?C9r&Ng~#E}X2#-?{X2l5xHuK|K=Zb)ORV&A!yUcS
z#8g|kZ+o!BW2jTLSbfQ2i^p=O6o^v>9ZHc2i<q<uQFcw++mW=4(X{V<G!lvz68<PY
zD-Dh-Jss5}3gpWi{HZteko__fuo04m7?AgbJsB|3GV-4^UUYPD(sB@^6S#x=F9EC?
zfS>l0mVxrHCDN8s^K}CJUB8nUEtm3gf<g4RZUw1+*V;w#ho#;1CEX<*8>Bmbye8#i
zunYDU!lhGw#=kYtdcToD4IV<G`DHSN=pfi)h;aT(m}zmGvzICrL?FCB>fJ!RDa`E3
z`!t<IbTLzJ9dpCyJ3{eV9<Y2g&P%si3=%%W0tC!F)$JAcH`dZk#5w2~25XrbmvlYV
z9?T^CG~RZU@}yf9^tK+p4AiXh*p(f8T5B3KBKTMR7yC<zT=bF80<&SRl>EDdLbc-3
zrJzF#xtPh(%ltwRFLkQ$mN-~XIN>{dJ*^)>SCG(;xa+&^d{s}ej?Rdx@S3uNju?oT
z0s@0ZsdgLkye#qy1m9^l(e>zMEu-F`mo;}xhT!G!Mved0iLJ9+v;V#=Y=(1F57ckS
zy7TyXPmw<T0Us7uf;fHSjp%?4@G4A#e$Agt#gDc_1EK|#`Ip|9%MJ*mM-o&(<JaLP
zvCv4nD-$Y9fxU&fE`UqL1;Mf-p9%n>VP!j@P)5k9HGW3@jCXeQdkN1|2p?0`dPkrX
zV><6Azfsk50&w<^sriEJGaUd5Cq+Zd0Xt1#6qxG!AIfOna40E<ze7-q@Z9uILP7lS
z{Ol~420g)VPdAV}18P;v>O$0aw&u<@c6T+Lx?oSRQ<ohc{JO@Fhf_#S0u-&Bx57XD
z0f{i<<knJL#-{AJ`yH72e^Uq{!@b8w_DX5NCgOEzl+qcBwx#IENZDgCDFx$h*fRBi
zvH@Y<Kf?(t%3*@<Wl2363>njJSCV6#Y#(!g3P~l!4>ue`3B+SN%X$7<+L9%iMP-`U
z82{x{1l;9^gK0Ym`lbMk(xl{FUTsDLzY@MHYb)eO+i*B&{6aeE)+wxZ)iK+Zehj_i
z;QMerZ6uc!!oVvHXSG!_#_+_mA+TajKmC@?M~*B;B35SQc^UQKX~jItZL|e<J%2w5
z_A0H;^A#IF@Wu#5s`bQVDQX(zZ8KjoXfvtif~MVA)_jo&VPFAF{;Ov{wD^k0a0!OZ
zykAbhVq^zy%?|Ju9#jO4W8GSWJ}r3ysxGcElWOS}(?_5i#kEG0zp(ZF(BFDW5fMSp
z$qS?LbQy*C=VtGGez()pv?ODC{==<(@c;6t^WwrNOw;Dz^{BB(W@22$4<*RB$pPH|
z(cAq`{;TXZ8fSulr4Z%o6zt?;#x}jZ#L8RR6J_3?5A({#|N9tZJ#JdJo-E(CvD5=)
z?wa7S_+BG>+^?nU5_9C1cTJs!O4L-%!9jaFJ~_a7eYmEv)XzEpnR;h*dP*HlD|0Aq
zt(fyrDcYlCt=cS6CE@CzJK7?|CLrL-(RS^KXsHZ`=PYCdpdI0jrHF|O%dm|euzB<3
zfqrnCC-$_yS_0p+RnK+dTEZQgAoVTXc_j35?x;mTS<E%ISi3C7`PsL^d6h}AO#DKH
zUZu+euQTT@qu$GKz?PzYc{HN^U`u=&!b0JE!%IGK0bg?eXp#O#XhL%TRDZNTm`?z$
zrHS>mXtcOBz)B_ltdwS68rL;md9k~>dt@}-!NX*yXymI%ul1W^**CxsKT%dY@UN44
zAO*C$d!}``KWF6n<(U$A+G@lU3r5kgIAU=^K|Z?xe$w?~H~Hf5WI(wurfh1NfFr{%
zZLB}NRND!qr4+hxc!$_2&FIJzUK%M#Wiz&HfVKuyL5kp(WH|}Jr@__ne>9!@Khy93
z$7fDU%$#b;X(Kg-%^`CZ>Mh66^loyFLUW!oLt-R{k~x!{HF~R&vvR6N%3&)rhny3V
zQwZPJ=ZEhfV7J|Fc3s!=dR&jk<38hAx8iiPRd=-IbhN#)MPN4=m|zlN;6xn>@|%nP
zT)`gYF8y8DZqc{2)rU`agWLKyR<^c)iSIWBIB>$-2qn-M%#^<buv}*n1qT(H*CJ|w
zzn$Q{JJw}wU*l6=LLKdTzZ|g7JX$^iNL}yONi!Q8FgL`Izo0q2!I7dVRi4>v(A_;%
z`MumBVQ3KST|F-W$;sBN{MvjENhn+(IN%l8s5M4{4v&YCif8$|*+h(9zF=C~SrCiP
z;F_PiSYSCE^jDGj*EbXa3X!w@W+k|g4px!0>}qdh^b53!w%kee69l7o)^IOmgKm?a
zYiFN2zjfgZ*fPT{mG*ROto4HNA_(Qt7-!&7%h;|bxKpVD^U#ql`ko*{X1j)4TPHxv
z64eVA@kj406O|&oZAK#%T>-eYU|Pj}dHG8B^YmNutl*}Gq5gc-b_;(*xmSOFz!VIL
zdX|*!6)4LmI&QJbFE#Ago!yenL*kl;fqCVHK_Elt_?Tk9$w94`vK$y|KfYSOyk&|q
z6Z^0D_gx7aa#uQMAhl(b0YG{jGBu?u2$L_WK3^=qTjD;s6U7O}d}|<nV<G{OyuVvh
z0CUaQmGk+ViTSZy%@Kx|_l(*pR}s$vgV9Tx*u)snSm{MnMBA;b<yHbI!g~P6fr9a3
zUhr9r)POGrRfBPD3RqtC#j81@a7~#fdGNRyHg=>gNos7UWge-Bwwr2t?HzC1orow5
z)SBNkxm0@9yR4z?HPfQ~!V1%(Hd&1GZ&uQr?`gJ7Y>v<$HL2c-S()nuK5=B4QFwId
zA7C@LHEp`L&s?ErhLYMH4|k3>i_XU{0i`7o1OvG3&*22A_aJh&k^}iUU-U`0%}2B8
ziSVxGT6*g})<vjM3TEyYn&CmmkG0G0sTEwWHfg!=z|rUI)-?LMdtYY4{pS>g^KWN1
z`YIOE%-4tOI!vfD*xvVEgl}Vb(1!%XyvyfL`mQHl(MH<czLCs!=O{&*Lri5~Cp{;J
zzRou<yyZC?kp)_uerfMPSCSFN(D=m?c1(zllxwnUJJVw}fy+mMdtfFr$&kE0c=Ts5
z|4PNLr46Q@1wRI;hQ;>6)3w#qF#il&!mLN-hg>Z^->(8Ia?s}DXFGfQtS#2^@?`Yk
z+*UZ-i(uhK;uow~vI5piMNxEQBv!f{QD*#}zyE#w*Rr)K9)>o-EJ>N!Xw`A}dyZZw
z2mndZukAiY62C}<I{GZPS1lo~7+%j1mDPG$;$hN1b$jBQ52*9&VAQCMDS?(5HOav9
zWh3FwE?GTQ2@zAh90Ob0-Cf_^^_Q?E*b@e700j1bS66n8I)>>NR0L75@J|@`v7IO#
z4<XsxoVLr3^mbFZ(8knG?dPINs~flc1D%mba*AdMmzJ=rB5l;e1Jk?nrR<^S6T7mc
zn}cxD9c!RgJz(l<#B<p9Ab8GA3!`F0#qg3}%crs`yTC-Ghes;?H3j2uz%AxGx>E)q
zNY8xd@%T(sKnQs1qzt2^uXy-SN24eWEw4=qIa5*SJ{LdN)LjEgz$q}2blguf|M$9D
zZ@N=^lx>x7os+DqjBy`f@*Q>_J~MN_3cwxoTo2^y8+KvsgyncEzVxLrL~SiO!T+Wu
z`7d6YD5DMGMTo^+L5@2^xs~sl@44Fd+neW*EfnKluI*Fm>dMKYiI_Tjss#=&a^AJ<
zs>o8aRNzJDj(^N$!2XRSNK0OrZ<Cf7INcCpZGIUy;w566VcJ!k+VXDfo;RaE>8?3y
z?mFLqR~bRsY`2x6TocI(vKIZjWAf_g{j!GCX6;?@>Q2(5yvpR`XYxNK4*Q2k>Kttj
zTWze^oz(6&@L9|XG*0>GDDfpF-IZK|r;fF}gM;?{%SyU``Y6bg%2CG9XD5x<g!<nj
zHhrQe^X#}^!M`EY2xT`30>0juLHnlhix`RCs;814HvHa)NhB=!VY}1AE5AAhe*qWA
zUn{%i+X#29X;#e3#^FSXkS=uYw|WPcg?iLXtjrq*z9F-6Au_w=R4dC!Yi+R&!IRcM
zlP;LZDMKhD6T3)ftrDLV7n`O0$2yn(TsF@o(SZ2M{5CY!tvEUv2C94`-dv*X0^u0A
zZGXpiZfj%mk^<1wvOsTbcl2R~owwk)3xDUSrK4X^RQRcHMlIFIQZr%XewxtB7Lr!L
zDWP%EV+r{{W&Qt}RLG2s_-c!?|1=S5RR&neG}Kvw<2ETu028b8{}lji!cC3QcFGmd
zW_!Ak;Tt+Pj+*7_Q}&a(T5jkd5lt!9;M}x)h4_?|HvH#Ls8&%$ofC;|s=kxS8}i1%
zw_~Z~$U>PnEkp8o!+bdch^lfr8}Luj9nGk;#0Rn3+nb7irK;Cc+m^CaiJpvVsI8nK
ziAyln71ICH!{#(1D)ew_wd|rb=dW@L3(!=d8rNi)rvp*sO@x{e$XH#mEL&4ZPq%Q=
zv5s!%B|?1psru%QU6p$@Ru1_oR>X~C!w&yMLCMPt(i(stdXqYg=!HyCaaA~=m)0#c
zR6-;%KdI?KX3SUq-0SMgu$P6I1dl^fAvh)U<k!5Ad-3aR^l^(wwz$*mihFdpj^&IC
z>vb01?6^^ofhX>ANd#X%s>)=Ydi8Rj&TamCQ{LYq9*yG*7Mk22A)qK)wL81>MOsfs
zTPqtVL&~Dw@>wQ)Bm39k^W<k}aMv5Zb{iU5Hc4)uB-|Jd4;!ej%Negn<#D9$fwkiv
zejvB@Km~GTM2ApSno>ud7bpY#pwq+aE}08%lY%${xh@%e|Hrxc5<@Y(w>^P^1p1eW
zgu0&jp<wQ<Nda4a&?1l5)cid@uhlZK<RFu--5{Mp=x|VWY9F(1a;!lz5Iib6BYgxo
zd5LvQ9McurpJ|vL;$lH2Wn@oF`rKKKg}k+(jehLrXq_}O7Jb8d^?rbcvlS_UBTrE;
zH&7XNxY3>Zq(42TVlt*-FKGV7E;`8uRIk)7^eAwi4=x+t^-x4l%;)9!vMgbSCYUQ(
z-SdcYPN5PiD-ZR&45N3i-SoVCi~r&WWd<FoYHVm2eg^^rsgQc$0UuFi3Is*%2NJF(
zsLS=;^9mSunOch-s=u10?1OCA@0MS9`}9(Ypz7XH$<YY*BenSV*BW<Sw|-nC?dNAF
zxcuGPpiq`od;yq_bIRX8e$iN~h2cyae)czUHwebh89KG->UYwMGq&Zl7J6oGWiB|%
zB4Y$kW&c^MCfV2%kW1*o!Xmb36|2Fg&dHa;8ZYXfJOv8b8D;C-_i-$(e$7bsyo}}(
z<y39I1Ef_AKVtgQ1F;gnyLSKlonBwr8x0QZ_RKX0V4joyV#lS7{9f8@+o&uA=@fy!
ziZ08CB}G8kN6V5gQolZl0U4eJ8d{Zq545R&Gt)LC@$e4wkj?!~zTnD^f%NqBs@~W8
zP>+FM+nKT?Q*wk12PYCDx?<ub4EYhl|9N$|nWEGyrq`#Z{r=?EHWHUD<lnTM$oL_z
zQ<|ORW<G-*A+X*?Sbbffe{Xqv#l^BKJyF!H^{fLk&g0BkaHx)SU`qUZjLd)48eBN;
z4y&7PBxf9Pxw5A{T9#MsA7XLz&B{{>3TDsi_g9)7Cg&>>Rk5ZbpzrC~FGsI*XBhDz
z=Hy~&cAf$mHXl1h?fEd-|EQ`bHqzD0wD_MCg--s6EU-9}2ocpg+AABmBC)u#9pKQ|
zCS{c<Kf!hG8p^8}A^)O&rQS~P?W2L7U>a?yl|+#rTMfg`L%+`On6uR@MO1}FoClFV
z6;MB8&&`AGP8q|I+GuQ3L>w%D4`p4PV3W`FuLNG?X@gzc>I(4*#X)KGM6AOL^%e7^
zqd%Abo_^kq!79iQF-0vR;Xv+sG5kAzc>&0*D4#s3Y2JqrFj1UUBx~N=Z@B*!A+4uF
z(zT#D`T&hhOS;~FEQQsVb#K(r7^cQN6JEp(OE2CKvE_+CV>#%L_Ej8%HsV??0psp0
zZIA@zfUynbwE;~7q@&HSoI1@Zto$pU+J(NVs?2(xa)1oykNvy1OldbcPb>EiGR6qR
zBN9Yy`xg%ElqRDc7kXIH<XMHwI8t74=95!ga`JcLm5|BCnh)|-F+T3T3iE|-@BnDW
z0(CyqMk^uRXL&o=zLtJs2%czUvKyqAhSkEe8XD?|PC{C02^m$Y!b9m_AKq@wRPTl5
zT(Y-rCCt&6-6wM4n^LkBzXVB#D6St%y{zV};dMVOOx52&PCX>>r-Dfc$?wcBJ34b^
z<zmMt`qs*CaW8mZq&SaV-sD7k)P8s<_UO-g*YHMS*tdC>op;^>HFJKlMsg?OjfwL5
z>Yzb{{@m#|$Z)yauM0oKL(JcESoL+GyE{Ymthx}|*_Erb${*{YIug*~&_9wwkob9q
z7i(DwGI}rj^+lpzYJR{+l4b*5J7~b-KVH$(vI*@ge$0r=g0((g9y}#JxZLZiOZN{l
zOHQ>$?w@dzD}mq7t);K9m`>7G`a1dg!e93P3p*JAc1}U{eq9T7O+_fuKZ>SSprM+*
zF_Wi31|g!as2|<G3FFa&!<rX*JpEnoYg3DTA-Q)&P-nW~t2{6qp9pGs!po?%Z^NJR
z5D7j<dO<$~Oxj#>v!c&A0Sw6cbbfi_2AUsrC(vz2vI<u<<U?<uh`Z&Nch%R{`BIj*
z9$>A8cU{ig0Jj>=mn}JC_y^~%y|<^8Pw+Dwyzf^7;u(=BmoavGayyLB?ctqKH_2}z
zkZWjViPPBJ*BzBKn7U@)lN@7$2nH24JtS4<5eM|&P;9`q=CL2jXnJ<Sej-Aq+{Id9
zCK$mmGFN-JeOna!_V#CNOpB4f?dM4mTYhrC5Cp-M^56OYAdk+We>Xg(Fwbi}3+Fo~
zQBw)(c8N3}wvo_}y4>j5uMfOwFE`!@k?6JVH}$5ed#vWHD^RnByev?zeMv!|P^2kI
zA$RsAvlu0S#50cz@^$s$=g3vPZ{(*c_HC#9NBqCJ?GC=ipB6Y_EysoRbaZnV%ROF-
z%MHYymX_0Z|Ir=A?=-id-U!Tuwd}1B92fE!1+AFaiKmv5W6%`xqA{EdwuS1<2jxQ(
zmteM$Ev+m2^Dzs&pOTWWz1_x8(&oy__V4hfelwia{WJfi=aR><<h$P(4ZipgLqT&?
z0t}2n$bOSMwxih_S?tdjmc8K4X5m*F7z(S9sf*Cv-KEXHf1MTRNZM*yCPQ520nEr`
zhd9|`X2KssHo2_Cr0i}pZl@WwCdyQmM3}7q@!N(MK(f8P_Z=Ik$(y9@{*HyS?RTzD
zriE%)a}NKICnrqBBD&kZsv4h-hlNS}-qk5cb+bfVnE*bOVv|fxHhq#6_`_#na%K+1
z_jv<hYie?@I^tU+mEe9-BsDXzI_gBNSBU}iYvbDHrwbk+qnAqWtW|NmrLeEJ%4qI4
zv*WazcOzRYhSgGJ5sc&bERT4=``Ja)6{BdK0gIC?64&Rpl0zQwf*;?1$f|El&Vm(|
z&rE-@N{DVsnyePjeXI?;R#e}VNB&mObJl|E)_~zhzRf>=dx?T+n!j7e8E5Xwaut;W
z#CXei$mxG6L)zoEs0iA!!}74Y*Z1<fNeJXMwwWmkfJVfauy5tsIR7v(C)#EVy;gt9
z2x6162$KVkG%Mdaj?vv~w@f1?dwmysodx^WL$iVE9Uar*wkC!`LaNI*_wzQ@=hP}b
z^*mz{o7nU2H;s~0|HY+qI{S4hl)LhLA_ET4Y6rck;)!Rz^Akd+7I%2%Kp<6c<yA;V
z&as=c=5w3ty2W`YLv^QWP$Kp=*M%B~)2$@jWIh+@Gg7q&a<AiLIehhDu1Rb@#BjD#
z;;}cU0j_x<f|`rgv9guJGX|0#z>UPQ(!J_^CSOtkB2CwBK$GS0bQoNDgP;C=wfmk2
zlIn(%jMM?B7YCuR3~28UfOO?%;CBP(!DB3AW8fcDQ4Py*bN+7(kD@^}F|Ra9pX@aY
z(w6*EjNb8(RixyZ*SDi4=_?oOmKln1N{<tTIPg!Fi70A0j1;;T8n(YhUr=^$xpe3F
zjHC}Z-YWrQZtN3_E`tdr0F=q~rKRa0J?kk}q`!uocmNQ-6k02LEziOq9xg36GJ%~X
zo0llMagC)ih{QK4(6TQEkOXd{K4-g{BFeJ^W{u=To>X@W3;XTuU38i!ob>$!HF1U<
z0{`rdB9fM51tT~k%I9)v)j635>8=tIoB>A$&~gXK*itW`;U=ns<Kz1Do3*?=+8Eh0
z8#=T5r-2%G&Hj9ZVuZJa<b<{6j%1G{=kSBqKGi-2+9|~&MGrYRNzbcdkH>MkrbKE<
zN=cOudw$H2407}VRBj_Xq06NWdo9U}H9#^p5Xz5$+fa!POd^Z@NOm64kZZ5*DUW>5
zpkFiSIF8f>jp-&7tNP$L&XuQd4Q=DSQeq6s9lgyk0$~PRPwnvBp$=JQ-%g4PyBWxY
z92_p+F3<dqo^>aA(i-a5_S+x6Z}zDG-1FjY{I22BiIaoj7hDG{a#FLuuP?=4MdR*Q
z|7aJ<)ka{1v`Z%b+us>2c~hnSX1v;n6R0Hlm^dRd*6ZSj)6gwS!*ti*K-&(>mwdi@
z)x&aIvUsFM!^4k5?BsD#5y7b}AP3N!`48xcows)?Z2N!|+%hkI*&Sd_=O$OZZSY-g
z$U9}d_?=Cp)=}Kge>XwJ&t%(2Toddw>O^Nn-K5|In$=zEi0A9GY^no3O2on^d{}cT
z%FDTl9GVcUFh~;|J!mGkdO|3;s@luzuEG1%Rg2YU9FSTsw($zPljwVAHDBAC%9j9&
z_o;?NJR#BR?R3w>WZIu@H#g>ma)^eY2Vx8MPAu-&u$>`cSbmAnZx_Xk-n=?rx;uv_
zqlHz^%mb*F$-_Nn*RZM}uRp$_ad|5||94;7U|L#Q#5ai-3xiMb#^NGc|7tMIy7B{)
z{o#F?kG1<Z#uX9s6Qz6xCo#`BMNW0Yhk?FS2<c5);)m{$m<T1|n-IN}M5Tf;uZX5`
zSH9^A>2RN4fTXFK{koZH1JOO>+`HS{FjU@!>iG#btaKOx*!6Wz9g&CIO$>D&LwUbA
z?HgTR_}{`gz66+8bWPcmV@Q90ZLZG*eymE?(aBHrpM^_HOOf`L(hveb5@a&^AfTz0
z)YjJKxNXx?aVzWTwy^6pqNf7xSnQYdl-25}v+<(8(uZeuH#eudHve`pnJhYmV$lvr
z);!SLQ^qhjGLiHba7Wf1(T`fGfpgkt?`u|WU-p3_PNRi6o2xn<hLXK3+6YAAP`0PU
zFl)}eh#9}>r`4Mui-O7qWHTi(r6g?mo0ht@EuAq3D+blJt#HkQHSc2#d(#!)Nj*+k
zeF-V~`^!(KDAo9IdpkhKLgIdGBK#k&e-<3gYU#Q;dm*7KL*bE?RXhfu3$&+Esszom
zcVwN1Z#^7QSaqH^PWYR;0W$E=2o}T1idk1tllk7EXTKO;^E7@@qCg#dXHPocg;qht
z#?kOuHjLqQPPawFpGT>2DePog?ebhe6B|(Q-h_SX3e7fzBzFDFQm9@_GYaaQfQ#N&
zZd6P(aDjiaI;&x9|HVK+lH)ypcxsI7O~Lq<w0#%0_ZPam<KXbG3hBZ`lfGe8StZpn
z(O-fm?}cDsLsS<`;@0(3Yb0M#)sy3^Fy1FNCyXB;_>JoG6SIY3&9QG*hZ^dE{^=t9
z0}0tc8D!8u##g2oxvv?<iC<j%`(~bZEx6ZEmXQfv>HSlFF#JcmdrmvM7k%wH`T*5W
zD*Y=0OU^dMdAHnbqOus*V{;Km^8JgKu0zK;%RrJX)KLeVy?1#p$;4zaW;^P_l;q6P
z<iTSMI!4|4$-KOWjoEbeo@y^-r>*VJ;hcJl9p;GX)E*tqswvr&k<B1GLplbOwoVHI
zjF~Sn=}FIX8i8$R)nfCauTiRzct(Qz@NPrfC)Sl$?2q`d`kxAqk$U?2LPL}K2LUZq
z0|`?U@t=VC;%YAvh>WU$yIG<ZQcS8%<9FVsehUHm1dGE`K8(hU<V-s(+IU^!eaB^q
zD9sec5L<;FKOJjtYsGx;3e!sQJZsjE;2#oyvpM=@cbeZ4Q(jntp6y1Av;krz9q{xm
zEDK1&(>Ww#(uGBM6j;ftsHyD4rSuAnYAGPUlYrTqG&A!}BGMrdQQ?vj=GQqm={QdX
z-p}{<SuEn@q?j=1Zb1g8r3v(z#*bP*-pfQOxVBAySJ#GW#WRJk_{sT|`YdFzK@9ur
z_Y`a?@Oi+Q2)T<?zXJ7FD2va=cKgGPOkw~(yiju_S9oT|3bl2>x2WT=p6T$K?Zb?k
zB#o$-Zf>sg{u9~cAbYI}K%V#PP_qtdkEyLWHb-=4g&g}L8Tb9*W>5V1Nv1=aZ}K8)
zeJLnBJQ`RM!)ld$$V${Xt&*G`(Y~%?1v77Y^O5D{HcqTH6=%w^Q8BCu4g2O~!8Pvc
z%HK_1eM6vK7qV_Ui{8@}`y2H>;AnS(#Llkqnt~Z6ON<vJ9cwBo8xo&2Z+jd-AhLGK
zM!)wvXdgM!*}OT(dcwZW;07T$#4upTCOAa8FS0=6?#|VNf|pB+CQ~nup9_`gh7G&q
zzQ9yA{;<fq>s`AzMGiTI6@J124^@4FH}fX#?^2q9j8$ryV1)}@_gDIi+Pnej-jEmD
zyT-xcjt8!Dbd;lI;dqR8x@}uEb)$YwP*8d>;>k+50HVx9*ZT@?VB|$lsM-#}VIbyQ
zLOO%sThfo{{rUCk#kcvn&p4|B)yKL0P*d(rinQy5<oCFcG4;}0Wz;eU%0NaeSmp*M
z5L9YezOnQ6<OKH#D72)ydUb+?U%q_j27lh9cXXnZ7`w<L)Jc15_O6Cpd`M(QQ(W>T
zf8#41>Ia*1L<dIo_U>Hpc_sKL8+WVJ$+6?u6~PW-knE3=PusfOZzTl=%Dy$=G`?QU
zp~&xRP5PM8R+{ZLmRncsB?GecbW*!q9GHVe9pqeD4rSf8G)P|VeGwpnit2&6rC`W+
zy{T>iQw6SqBO9Zgpf`6~a=g904t+YV2bp|1rWK=IZ6{OwesL)Gmi_41_7*UNR}!!m
zz@=r*+FgS=Hr55`<K=ed<LvlTwo_hv033)(X>5M_9s%Q(1KuZYCmXlMWHS)om$C(o
zqyq8-4*yI8&s}{>B(x9B=Yfss`y|7??caM_D|^fIxjKE)iHv=f!8N6!)OuIrE8P&G
zXDv(FS_bOz)387Z%e0CG=>`v9@2&8t7J|KPY;6MX&+FLvuI$5U=|e2EYbD@Y10ffR
zDa#B<(JXU;%g;xK)Q=NY+CBtf>>K8`e*d1EQ!S}9&A9{M^f1uArOiKYn0QA=@$0Xu
z2k!q4u&20t4voBa(}hu$Zxl<%N<LqWVw?7_9UJoT;9F&f?EqhWVIgq|9pAZXXns)4
zi&T9mk?0>95ur#^-)$Rt3i&L%$H0F`$+r0j>Wq)i4+Ps9auxf3q<O7meVj<;+R)Bl
z^*PlQQC>>QtA4GZ&Qty=|C+U-$Lyhm+OYh<04$=u%kd-rV**NLa@TpS_r?CDjx|>M
z8kIJ)4mb3>Z&R&ZDw?zUd3I-zE45(uICPfphgABxx%E)<S-b59-<S7>{uvC36SqlA
z<yhp+*N$N=_}1CfjOdmH>G7k=;a%x7>DoJ?>>UE>`~bWEwf@Jq8A*FL)27(F7vwi0
zn+6v2KW{V-mW^iq4iks8gyPr&$Ba^-c3y#r@Fb2M8)FxVK!QA@`Md%D0ID=aH1{38
z^!l*;a%-HXEO&3cM#yv}xMB37?`TGPCYFp3(az({mKBmkh#<OjjeBq-*9ZFB5*-2)
z{c=g0Rkza&@1$nPBDJD;r-&*STzvgC$8}|8zOOF!o-1kx#)Bqu+lzWlZ%`l+cg;WK
z_B@Pa>%YOeqM=C~)E%H}LETKNo`P4qbtKcAx3_~2tb8?|fStR^3g2b;FufZ~zz8CZ
ze?_c=nn>*Z6znv%A8vjsZY18<mG0_%n^Xd;9EoCQg%xVcUv-e>aW#ZM7U#PWbNu88
zY8m73VKPEECQ1BtXDs@Z2u8-m;dm}jeBDO#couzG9j&J;arA3uqv&Y!nNa@*tG?EQ
zYfw{=mr7m5c}1}tYn7QQb&hASPz>57S#5T6y(`fLj$2t#AE{4ogv+G!yhf0K;~;G~
z-Ns9SoIZ3-OsNq_RLpfmN9XyL5op_9&+Oxxx(-^d9Y=xD@36tc1qz`ip4dXmH0i=Z
zDQPLG?`l!I+s(Dy|JF3wF1{1wGDaJxWmr`#6&<cy(N+Npz-ooDusc*UZ6=hA5$P*b
z*z);tJFo`1e6ho^y^ZZyLmJ8dDv2NM!U_OS3#7wsBD-OE3tycLHp>70x|Y@KqcDa*
zw{0#~wzt%_c6ket(JDgE?C)%fze&}UsF<Gi*D_Dc7E!sde2YF&s&$5wOI1cRV|i|G
zX?P*ZH>ShBd!)c=s)YJ;pS~3kq&hs~pNyEz|Ja`AWfByf%kOuSbmN9C5tal`r}n)C
z96%f|^G1=rb-u=rMYzjN_)QArl0d`LASkk4$=(*zX@IoBv-<Hx@ITrY1VQ5t#yf34
zBQURk$MZkaiJdNv3>(vvGl`x3r5;D@>nGQ$)?yOK<YbV<JG4{+7txo_KR>Wt9mlga
z+fXRurxCWJTu-{*M2)yD{`;YRp_G*@qNfw}8{6HL4mDD-#RaFW8Z7?kN9D9MJ=Tsv
zy=hTe{u>_F_G!=nW>}&(cIi$DKA3Yrv(3Q?XE6Zi9ici4{(Ey9^k$~VA9~%S>8rLe
zCWYR*YN0b<oItm`{wSlfc;!Kct+}Y-+2U9`?z=qQH%aWv79SN#Krn2!Hj-<#YUb`_
z8pB6{%h|(_zdQD;abD2($ekhB1CWn`As40_@O0#<J>_w&Jdqw#A_FhsFu~dSi}1*H
zTiml2O}#r+>l{~3hQmqpNU5ntyG_kIhAJDPs*}3n37;L*ODt}Q4Omn+fej^8(45Z6
z(~;qSfElC@nBr?$hrfPB+IwVp)Ow7G@B;-*Uq-yBK8k4ojO6q6y<Z4pd08SeBBrEC
zyAO6Q9yZWMvukEZaUy%9Hn`h;+?5}I1B{K;wnT-So}WS|e`T?5HAJ=dQ-5+j=Dm{K
zg>`-(F}%FjP`lh-OPI8F>ByRE_;=W^=wc}8wpQw{TJMJEd(^=n>`zwn+N;+;%KOzy
z(X;8cjk_h}gWpRV1ge?mq!OKQ-rn}rPVr<&w`7}}MW^KFmJF54cEujRBXF*ip2R0B
zwbZ=Ct%kaR8pNB^o4YvW{k3V)g<)XX*`oHFm8c)oB%65$Bm<<NiEbemgNOAZDQpr5
zS<-Y-u|!O0JU>0rCFoWC4O>ClC<C${sde*JGtuh?PAFp^2wpA|b%V~}@0=p?nbyr(
z;5w<&2Y|i|@Fma%g8pQ<2^+6iV%U#No|ka7yx-(wFSMQs9o`Jm5IHzH2mn=yy)}Zo
zdzJ}~85*1Rdv64d53Hd*g(ad$P>)~RS=|H6ZLw$TBeunYLovC}l5)2HTxT-kbCO)i
zDYr{(PJ?oD)T~49Tpu3VCO9!xPRw-Yudem82XM5j-n6SCh7yU>ky?G3*!aOB{TjaU
z4flQSgrhnuW}Q}xa&j!zPh4==xNjJE(g3a$_Vd(HdHE=>d4rhLwJMTJaVqZKHx|X$
zw^$tV^S-Q*2u~dR+~wrh1PSal%lqOSW!VYk1ZDThMtQz+zRzAa=fm133ib5nFnJTF
zFUhY4pEsy1Qx^(J{$gr-*58$1h%Dy$ZHGy(zeh`|xkMeh^Pv0$7U@h#@VRUt6BC=0
zM{Yi5Fd12O)f-vTFA+d`sTX9!9b)@>U#C2fPiTBVH$H}3f~3*xF8APs&XxJ;S-!vP
zoLU%2&gitj58JigH1IGW`e|^!Y*7)Y-n}ITN;Z5B0*&j^Lp)V@=V7}$d{URX^Pgnr
zcz(*s?jPVtf;Dehj$Vde8@PPkTns%g8kqL#nLMCDBIXy;q?sg}OAnGp)k;HkihyF8
zndS?1lkJtJ&f)<IjcvS5+ybE4)X!dPgrsu`nTDmX@M{lpWrK&VD2u#hmtFM+b8*Ut
z$ZXA^xs1J&;dn)NCts$8xBCcYqfc7OO3?QG2lR!A765wXv_n|OmsB(~5GTwZfjY&6
za{S*TR8o!=4I(+$pEY>i>}9PqtZ|Q*8v)1j0~sE#hNs&b%0*5}W|-_6inS*AH0jge
zVApQW@BcUmxb6Do0<p`fBm3d4jjVLn&`_+du5RS+@3KrV2q8ZYXw5_a1G)T){ya$`
z?)<~u+?3Uv)tpNc6&8u_?weVkn=8xuF|l?>FQsce$S0!Z15Hg#NbICFuW@tBE_*-V
zw|4CTln}gpwrx{lNEGls{GIM2EnwyUoA)UYi8D6;fJ?&qrJxviyx2QG9lfJJGy0#`
zM0LI^S%75X;kNoB+xrOu<81r$)Y1ID%n}KLmH-rF#@1Nz8PR?rDDg$;SieO^M$561
z<NfiPo9lemHDY_R;^{g0AI7)qsjOS9az?+{R0KJ{s(QrER7fa3Rr)=t3k_s_tJ@N@
z{_6Y8?kBcrf0gKzSk0NZf=S&CE!kKPf>-=q5pz;%B-)1DcLbOfKMKZqpp`uKLZ5Tq
zR$#WQ#SvVqDB3d`%xJRh%B*odP!Yez3=8vLGBdjp_@#MeV{7|QkR?Vf{cs)|4}YHB
zlDC<#v|2%p;O4z@g-EiQqW*B7sLtl(2&FzECw<W3I$5a}qAsrinOr%j_JU-;(GDFL
z9|^x8rdoN^TVOA1^zGXml;2T{j|cpD#7SG4-gDlaPW0@r=_LspG^JYISS|opC8kiB
zixk<@b(-iDD318p?Ly#*F&#l@LUpNRc;(Srf5%GP5kX8>m84E65jI3?1PY>fkXPa}
z>CFvMH-rqPVA*l9u<1~^Y1Ps8(i)(~^!JngELCiDbtxB-+@YvMX2g@;|9}7YDbO#H
z-j1?o-Yn~Z^=}5Z0OAYSzkj{qMOS*DxUDzsXYIv#L_8>!Mk6|0Cgi8qc8a>DjQwa|
z+b6YGUi2tc5WNP0|HVpWb7HTGVf<z@RqLzp5MTx{(u<6WFq-H>m*zKd`K+TYx9oh}
z*^D6veHsHn)JPW5YOY*G%}%FV&bWb`IAuFGW_BZUT~mk@Voe=|MPU2dS0#Y&-3?|w
z|M!S*?V0r2c~gvRcPAWNC@uXvNCiqtPy6U!^W!tK;NEAZeH7<c!OFw5&;GemQlB&z
zfV%c0(65Ds1^K6(6{2Ysz)1}4O1D9NUlY+c=9lO$9hqY)>vyz8KgEBlaO01J_br8v
zGe+5#eKRh(u>TsWkqX?MpTCIWRdHuMO8ia^cRUB}p@s(~RS)!fUCLcdDmF-PRTQ~K
zqbe3vVPttV_a{w9N5>fIe9hF_+V&sv0x$zc?@5y?lG~lOR{3HI1i$E;QN=pRF9+>m
zmuQIHt}cX%$Tijt)ey8X4|?${?+-}<F3S`l{=FD~CJz5WGbGj4>IXa_)-|ZxeG)+2
zQ|zVoRE}%8%%p~^6R0aJ62q%U%GZq#got0&3r2re@>!ZKIPyKUI1hDphC;oEjpAm~
z3sd_%b@;>XxsDMxLvI*!lxS*Y*J#J&YG4AsCSS0>xq~d}&JWb#-^D^9KzWkjZu|BE
zb)>Oo#8flDTdq2m47wFmL}{d)<h*K>00-NO^YCB$*x*y6y`4!S6kk^uX;5Fc0Vc${
zI%Xri^@hL9eD@`L?=}k#SIpfK+No@Au96d7l5<0S-e0ZMUbRaF?%++y)&r(g%$3_b
zF?v}|d6$}#`csm~==jecW{uTmL-ig3yRKh!cKX!ZwigiHpx5TI?`;^S*7_(!>DT)7
zbHxult0t2ElO<)>JXDbpIs=tF3wFpml8!#OIBJ2)%g(h|{l^1Y-g$Sby!NAc_TwNP
zMOX7@RoAB4l}}pxS)d#icGo%%n<-;-4!>+)vl^}W3sjJnDuKl??$^2EG3yVsn1h^s
zx`FKYP2<a@BjsN6N=VS`*ewft0<Bs+5R#F8YES-wHA(b3{=7S@g_lnfQ|^M57}yP9
z&l5>)w}2alY~-Im{tkC=TQM}RZ#^#-`nE!)Z@+u5Gph#G4bTL-w9x`dmpEklK68gT
zwYMFvOh)f6cSPa0?Ye84K;s&E;~8g0c76k|WBr-7Iz5qv2^;Hx$pKms*!7EW^HCgF
z+}i0WNi%&1d8qsB<f>R(VyCp8p3pmrauJibGNq^<6&e{D8fqM$>1ukagBbw@EzZ+U
zd|6ppS?>5tIr^!h>XjS|+9udD#e@L(1Atng(xM!~A3lOL5GEVT2PSYVp#AYMN-5rQ
z?KUU`@vakY)URpLH@zNaYFlEAcua2E9yn808_}e6yzHle-CS@Bfu)p8`a7c!Kpu{)
zNQWCozans(WQ-MMTOWLxo8T0Phfxd+z$ZzH^jz07hBq!Hua%z(9jpoWosnQ$OPIb2
z^UdkX+(27O>lsu%<=i;!ki`wl(2D&4p*C&%IF3X76j!!XNQb`zv}R|rKP#eBHId=w
zo)UeT^_+}yZc+AX)vajW^RZ3InQgZwCaRxe=hyw<iHL>I)%PMC8ifs!ka9I;x}$ec
zosAdE)S;pNUYXiT=Dh%j<?2D&&)x!z8ON#)&Na|)w8ix4=l*aT*>S{)iR+TjYR(E}
zaHPM{xVV}MY8$Ji*=W4S*!uCj$D|lR-YP2>NtNsC>gcYX@k^BmW7(-Hmr>~+20Qml
zj~-GiE`i}B(i8A>qkC_borDmC|H1>&Bj2%!Qu|cd4?~$(aRk2|4Uo*S-t>%{xU&(~
z2S+6H5L~b^&wVbA*0=0|fgo6hW2NpDMzsi0>!+U)h&R-ywp?BzU@hcK$t8V@94}rc
zw2Q6bJFM}&(P$%;&3}~}>GiV=(@(-*`$vnP+L&^aY+WGl`d|C7nhC?xFgL>z#F>yE
z!3s!uS8i`A!^WF1&zS7<DTfYJ6KB*x57W^q|HV5BP@C}Hv`j|kF;3o-0-l-Zx$kug
zqn9Ye;~#o&KJOMSN&lC&O7`DU;*36DwPz$|lZ;7#8^?Ddcnn35$P#tRg7+caiL|@n
zL;|>j!%S8=?pbxn-c8@;2L`bp*x6Ng%KBsyJwuQDcKk@KEgjn91$-Yj?0!@KUHgL!
z{%O%S^)uv8V;x@s#4U5!U+GMTj=n%w`bKcmt&z;^K7ZgYO}Y>MX$VIK#Fx(nM?{71
z%~f<9vOJ*`UFRgO(;m`>VQ!FWA}i8?N!pmxd87(&`O?fFoVTE=z8}~=3kHFLm8FH6
zbDEHFp{tN#EYwyjKVdgIH`Pc2{D$`U14rL~Pn&BSH|kURWM#Z+n8X45NB)O^pK4k9
zXsM%(bhIuVZ(z{r-@bfyQ}kW>V>Hvw8x{5Gl#yJC%=grf=i>U6z($XMetmKFHfb-q
z+px2ZAL*yYeOW!^gsBnIaG9Xs0R@{BRJa>Zo9;W*F39{-eyR8Coo62N9?%3bG7ci?
z2iPAxj!E}<9{%0+CPX`^#K^+6R*LwyJ^OvtxI=<=`25a`H}zwBjl>t0)~79LIi92Q
zG1pF4$=a*|!R^xuG2nYmZMS?nNReTyW8b^Zk<na;O`w}6Of*ZcRxaqL*J0`_HCBbX
zg;~{d!$rFn_89L~9wSA?8-xL&TR5gVDXRIFZ&~$IMnIK))xg{E8~o{TE>Ql++mvPp
zfpilvTqu<3v-J@2CFGQiAb2HY{?Xc<RR*5ucMS3?t6<vFn5z<>^ly9)>6@g44nH6|
zC;z+gooXB>m)V<H?k{2UX)EAn`sYy)3UdER4x_e-OHL^0tcNI==qD?qc9!?Yp}H+?
zFU^*1VAa^!$){)O#*b1yUH?QCeqa2V8LaLl)6ckH)bLtJ;@C#db#+P!72#5JQ=kiN
z@!b0u62<dMG8Mz$r8Wp6y>axl812U^O2j?4R}E1OlH;<K<HKa$x+NgV9(^0r9DgcZ
z^O5Yc1)>8bFHbcv9b1Yv_B`{_gb>;^yM*mVSpQ0YoIYO5L_G=(4~f<?GP+x7FMHy4
z+-i><4j0qzWCVjEr#?ZHMBP+Mm8~cp0Mm$46b<OJbM=`ozKW8rzYWPh-gE9!`Y|B~
zGT2(Cuy+$c5_J4sN&^Ic7l|7*%5wNOFQc*j>WLS_7lW;T`xa0fSrI8JR-7eHN(%t;
z@aT|qxI<Z>bR2Fn^UXe$sk^U&zZ_f_1c5FHXN<dO7j@{64tmC_aSrz(d0wA(=jL|z
z*E&=Q%F1=siX;l&{-M9|%q;?)mRXH&(l_ym5#pG%{QD{1I3iN(ICq8%xnkcf_gyH&
zpRaHKKDt9szeA5C&^-x6+Fy^_+1GgdAj2-l<{L@~0@xm)i#&vclsd9Lt^0jx^Anp*
zZ3%9)Y3Uhc|NEvdip+&O`4~|aA{b(HVIkeAke)R_k|_4h*h%@^{w(LBDv|e&-|TA6
zKceL_mc^?k#U2xQR1d#5YGK1HxdI`X_~rfj0GgY;T5c{H*AFCvh5?`r>H%DOg5A68
zoXT`>x#d&+YrgWYQv?>O%OM2fe*<@X1t5lGyAZChH`*ywNk;3oa&4TD5IEMwR%0<#
z;yr1=SrH&TT=wAnJKWc2R0@Z`waLWv`l#64ua+_@m%KRi0X{Wf&FD{mV4Q3J<w_Y?
zMN6g!_BX)sQyqO{<M-S$ao~rs$(#G`F$SOqprmY23_rEvy}If@a#IEcJT)z3T7%<=
z$Ya<~K_>JGA7WG2!T$UijnU|HGa8F0z%nEwXK~xyK5HC1Ym6@@5pG$r6m33(eqApR
z%W2CWfR(6avNnDR+g93h?qoiG^Q!v%sb`#~Df{=$q0Wnvjh`DKew)|?&1hEcQ3Jzb
z_(TGhUzV>A#(qW<zlk+ojE__W+o%(3Y73?+Jy^}vHvx1LO{0VC<voY*yi`fEA$1aU
zB-d6`o%BfZ^ZpVy<8qKwE`>vYmu<P|Mo8dRz1Z?mig?mXlRNZl#pk9QDurKiE%xFX
zSNZ#KjYA+bQsjxOcDkn8#;prC9Pp3au>HeJe}EtGGaO9R-*J%jJ79o*LM}*%6U058
z>B|2sLMC{dyV>)!&AtAF*4qc0q{xHDj;P(upfHEu)Epw@An$tNDOZ?>GZ&Kl&el}0
zZ)WCz9kMnN_2<_m8~kWTA#;J)cCW0mgWB6RJ+}WKqVGhjncQ2&GV~x2auEUlJp>|b
zhb?RIyUn+<<`@sSjo)1`nFUQbd`_XMzp9_#nEU<ug0eF}`((41Dyan|<%O+=c$v1D
zd>fS~GfO^j!Lg~|%@aONo}6Glu(@R%93%oR$@^wHLf=gd3y{z31hl@s-9J7+p+m*K
zCJ(~~ez3kO_a1^S4YJCdNA>(`dXovX>`Lpk@2}73v)THlWRs(FVP}wS5j)Rw8M(*A
zn>4pjPCm5YQ(&;r6e$(OV-PQcee~>v!N{%VV868&J7KTGt+=g(W4lcZT%AwI^p{~l
zn6%;X<0~$3^Vl5wnrdoZP)lo_nNxUP5U^(_46{9R=l$W|)8)7V-)I*At)GR3jz*#C
z1V+T&mKTRq=EBjlY2L4mKOZh{1iZQumt7)JvFVrZRWl|kL>fps25G8l^eU@r!KvDI
z(mpN=57TE3Tp*sXOTXlV<76^4VV*@{7e~iNCge4s&O8NUzR&@+ZaLvvCeHg9mu<57
zGj;tU%5weFAEI&b2FTb`Wd8qlr9A2e%sf3H22ot@K4&zLYtMao(IEA}prD_(ihRzw
z&BNZI<vcLMB>1}q>W?;Gr3*tx3Fi<!CLmP2sF8PBW-9?^oS<gDp=0F<PdImr7JKu&
zdiS|^Mtvfq_EdovAQce0gk>o`KKb3lhzy2jz+xb|8Zv2^l#O2&!gSvVLvt}U=acx`
zQ0jHtGsoVQzN2oN3bac1j=a=Aao_`niOA$O*eH~dW-YuGG2~NuC=p>*p&K{@(PI|4
z8}6l^63|2=r3Vmi@39y6SgWSjPFEP2C&6r!+fG@Xa!8V0vjxQ${;~7&jsnkr2Fz)G
zHD%&W=YvU(FMv8v6WJ+0dnP&SMqB%M7NxBAjETX|D)0$6i6L|;Jo}V287%T#gNZ47
zw23Ee?lwn<YaRU#ioW@qdY3lgZ6we4?!q~K1gs?F`nSyQY?s47hyOG9PESYYSCsu&
zR@8T323Fatt1k+)Oy<^*d|!iqhMo%XNQrfu0#1X(#Kbgdpg1qBY!<*m*LoKr+L|QS
zSCo;)D&9@wS<C-h00c@W7K$S04=!wx8nTyM3LoU;RQRjyxa2Q%v4@$=e0z<baI7g)
zlD8WqGe!Q|tj#n2qUye?tg0X|Phc!O^(+%x!t#5GO)nM#+a1$pM9*X|sXgM9uK(5`
zHh{wdzx~y|w4=>F{l64Xy?0QrqknLlX=(bmC>zW48_Wcg3uL%Z=k;Ts1<Xm*z9%9w
zA_{TnN1UEUzr5Pg*r*%(&!^c$6vE=kHPG!d2-k~=ioO;&LwL)FJk9b&9-mU6LggmF
zO&dGq4<d^iI)C@fE%(flIq1Qgc~WB{Ao;-`%iE`wIfCy-hs<|GIU0b}6201+3z+?u
z6%fR4ot_q*;Aj<fUmk%FZIU^H7+?KaaM7E_w3qrsrxSG_x#Y^Bdu0qEoAMfq_VI0x
ze(V%XJTns#h-jWp=PCd7x^<qVsPZN0`dd-&jj<OBXjyR~vY*_v7v4*~?{M!!6Ef&^
zQ)Ls4fj<)fn&i0IUp!XyVWNNO%cb%C6(r`00ZiybHLZH3oXZ1QN`lYm3>d;6Jh&?6
z+nBC#%0|4;U!vkO>HalxHU2|ShF0PEc!T}~YPHJB2Qf+Q)Sn&dG;kvMLQMN4+-=(5
zv|{G3h1uchuD9%GxPhE><}L54&4eHbA{&MNH^k>ujG--+D)a@L+Wj3=U|2_0_u%3U
zu{G+H!IZt6t3TWeBhh<6pINyN87GXe63|9so6_;P1p?jM4kr)hr7HAW$2D1PKvtLQ
zLBUH^89_|$$1j2J4)Do~@ib}K*3s_&j2F&D>0G}U8IEVB$jMK9+zkh|^2cp>aglmZ
zjHemukx_1f=zdcQ>X^_Qk&kFmU;5nM@B*c%qqSAZzLq4J`XJ_O(qA{e{`XvlAZM-<
z@vHOeT?gx3T^NDyK25ZWjlE^}w%ha{%^SLq*a>_)X^r0Tr9`4LK{W1~H;o}~h%HP#
z7~H~*MRc!rtcXRW{Bx<)6UVA;1W3Y7nFOtV%>}!UTLBz9VY#V&Zpv@dZEb4?+9SiX
zI&M+iQKYh2ycxk#2LN2=9>>Nk>qRiUJjxk}KSs$LI)jPFIM0pQ1rS&ZMMuG5dtUUt
zLDJtR{4;j+gTx)t;jkvQPng6!P#CZ;C>{Ox*J0nzTC{<wZqz|u);j6Xu)A*`faPx5
z66_iY<b1FyLIjiXGU%<gh;4;6elg=pEVlv4+=X8D>h#;U6S-TTpH^HL4uTwu9k|wV
zo|{2St83vglk4B$;5YSYEE~b++WqrN3BNKY{e*mN)&+h)-fnrNLB<d@*H8;=zY{08
zPTTa`k7}q69);6i8U=2%PkaHL{#@RCF%Kz_amww3zm!|;)31#$d{R@8#&M}rqis#6
zNR?r=;cWotin6cC9$ZSdfta(8KQaLMGz>OZo=y+|jf=w}e44z;iEg<GRhbP+%PZlW
zJFRvgB>~Np&d~w>s><h=7H?&Fgw-UB<TA&9+Zh@_yD!8{=!e;IZcZun10?V0m>5H9
zZZ#sNdiy`FVpY}NDIfB`c-aY*?n^TgvS2SoH$6TSo8l3kUE!Jg9gW3%{@{bmdbrc;
z=qn8qxeCaaFT0%uVnXlq1pAY<vzt%HA?iiOSXZo4W<;$jcG)v*4Qp6C;kmTi!h0`z
z9?ob7ssKZBpOp`m2Z8xy$Pi;%gZoW#PWDLavbk{_Ir}1EEZ`U-2w|p}OQHwczgj!)
z04q~J+8tg1Du6X9HfHLhofjlJIa+CqfDh!L7RB;rg8iWMT^htRGdtnyf6X9gukLsd
zL2#|tUkl;Pp3I*8;iOlTm+(Bp+?-|MRA(nTDZk~c?aO%mjFsuir(&Gb*S$D)eKerT
zX%HB9!giOK70SvK^^=LDRa_5Tt{2U96;qtoA{+++V)JkG`PbGK{8Qd%S}501lqr9k
z8}5~U=1k}>aAw}wk>=XICOE0?6r<IR;W{zB28=t0tDo4ujqVOi{0_C@Ay7dnx8q~=
zs(#$jY0ycDS!dsFr|7Z07j7v$?<w!0s@#;@!gUpK$!hfTc#eB(#5T{d3=q++QC6$|
zNnUQQLM7sNdP0|fcA96l@w4bChr7n9F^qYCaVbP%jjUQDHq{<Qq}Nf-_--BShmigr
z6s1;#B<>cTGYAp<qwn>cw;R^GKfIr~FQ|X8A2?9+Xen`xa=`|S<mjMS*?{pJA(Uzx
z@Gq{*p4*C)LR+vb2b1#7790jype$Pz6_KJSM%Td?!AD;y5!b=zIk%RWl&b4sypCQu
z_Em~Q{idMT*x!cvobT-Q>4TXaz-_j_5>D6T#H#-H60Q1}<64N4r`Y#Nu=)^M0;}a`
z>pef8JVDXNfzNZaQ@n9t1yYEglMPhTcj;RH4H_;g;B0`aw6HW@bW7A&G8S$U;EB0c
zKP8}{&|>LI-`Wa~&{M7p+3<&41%|LVG|y`L&50J73%8?41|h60f_=cikKVA2A#?)g
zklpJE;4<vplZyy&4v?whyG$jNiNi$dVPG}THMyZ5)!Z>)nPVcPHO0E<p+`d5Jzv@`
z1zs-iykK5FwgyAScN#`=+CGQhKea|ipuNZNh(yD?7j^U|`rz5S;Wq;SPrcMaEdJP{
z!3ED3d~_Qhdm!%_cOS+nvvpeu3jG>h0@=EzCC?WN8xWPm{zjMjo6cbQyM9@m-#sm0
zJ5ioKk7_dh$GVF|z0MV=3O=3!x%FZx`cG*+fmRj&d7n@BHs`a3@^ZX4l_r(e2a8)h
zVtIM80M;uO{V=yG!kkkWF;`EBkLP!|?FmgCna-rIjJ?h|^*v7}cDvOM?0gK?di9b2
zS=Yq$etgKc=F=+v{8NjjqF2uJ)6z8^Z*&kG^Qw@?9#)br28qOVWu^|p;EA2K{HT}0
zFV5ehtn4!MbRD7(|Asp<%H_@DW#VHsxy6xI&n4y&b=Qgoj7&2U3=+EqLpIj20pdyv
zE032oXFNt?b{1Y*Ifi`G`7`lilG7&~_bF~T6TqrO?<DH~ofZV{sQRySl_`NA%IF=2
z5F=Amf(B_2g%G}W(%~+7@fzuHvyr|fHVz$5`EZHYq{I+^Xugrz`-P)HQJF$-sI{Ls
z1mN*(A0E9Xj4(&NCP+&dh7Ia^q#YtA{sr5&3pW~4dyGghSNCPT+kn+Io~|C4fm{&i
z^peyBrC%iN9E3<ukn;OP3>E{P7Cu_H@cv`3OybI|7;foXla}sy)i7EgdALWU_w*|-
zJRD^#54)J)0R5{{2)l=tn@7m6aJ*ctjW|dF%;*<lx%>Qeq|!#un>m-;(6butp8kvX
z_DFA3On<{_o~UJZ<R<DL40i}~v7Yg?1oDzimFnv1!dS+kh4LrbwF~s3|N3j?7dk1C
zOxCwrmAnDJ=;4RMnF?c|=j4apBFA0Gqhjt0_xO`5ky!a86N7%l1FYun=27Ox-0s}p
zkD)hpD1|TXHmY|xG8gD8Y6+s}jtLn$r#`ETOEp<xpIz+(tea&7?>40~j(mJt>6b2;
zE|j&l7k=Z;;P{esOz505=PP`2?c51ll6ZwRScCI|DSx?_7wrUhMX!OjO}l3eEkVX8
z8FNL>E6f6O73T7RWqKE(q6$UjL;G5(i&?G^&%0D5I!$9U!uu--Z}sHTrBW6_ky`M4
zoPjA3eCLTxQM=wne6>uQ!wVftZ*|L+<j^!myht9vWx$&>!Z?K<Yll9{*n3Fo?r)|a
z0C)bbD^~HOy+6?siLfE6nvqTMxqr<(FXsW>&}!U)uRR<a$2OqBIegQpyuT-$UdyOh
zh^SI*a(|%ZSjP~z|BtWt3~H(i-@QXX2#64hib@MANbgOm5CoAXReF;kz4x*K5kip;
zp-Gb(n)F_j>O+^_d+#6w2<7Db=RNa&Idk$Uvy<#hCTs1z)_q;~@3P;meK;<84kA8K
z>Eva$)6+MqQ||mIqga$?frQk|$~m%Ni<qjC8mAu8>W%BA{%jtg3kxWjuAF}j*z;sg
zf@oQ_9qN2FKH_OkJlhO{zgPq1A;7A!sxIF3VsmsouLKC9f*=_#4T(u~Lk%q)v%8h4
z=1@c@wWp$0Mm|o>Ts6(ehrQG-+gx{kgqTz;(_kcy$zI@E->qr2cdyc<wN=g9ekQ)^
z07ozCJ!%cD46uoD6j#CJ`_2QHMF3a*7oj7sc}wuof-aP|>!ZuIN|zHYoFD-7+NRNV
z-PtfBX*Pw1o`Jz^=~c5~xm#b90xx5ZwU7T5IDnQ3kAUK*fNFu>Nu$|e@}UcK&RAv!
zJ+YF{L`x(&PZx#3S%<9Pacb7G!aB-t{4qGcNG2|(9F_NJ$4RE48rUffdUkW`0bCt4
zMT-g}5C_nETck0-)K>ypk%86ij;3fqZq=c)b!Ryu*iJ>+@A<jttu-<y+7+HG4bm?{
zd#7<uNNa5yon#PYV_L(#Cln1rTEpoPfrhBAaSbFTWqk6bdz0KF0;c?#YFnJueOngn
za(MeLN$td4-(hkc*g(TNy(H;XdoT==&38xhzV#8?j{38A*L7`_yw`+Lb6Ubjk;mdm
z|3?I^-9C&AKQQ0P`Kx}`{OD~^z>g!5{C2&u2hic+|BhYn9ACEr^;mH!KNZ#3W9Qjg
z^)xT7rgd-KKD%NIbs}m6JX$IC74%p|cfLu^@9h8p$!{$?yKKg5dNN^PY+SpZZzJa{
zZ<9aK9sqfQ*3&v<Q`MvzmmHf51qJ)OsU~`Tz5zC2crqTo#Sx_B*TnJyHo>SwZ_e?O
zTgT*u$FA4GYRp(picyW%?moAYQ*yzvuTi(a+`NRZmv`w->1A!rPDaJ}m0LPLW9HCq
z<wid6R7P4ROF|(~4L&u7M~D0y6HsAwD!lWM1{5bYKk>WzdPC+~)o8K1@$&e_D#LQ)
z=<?FG_rd3Z8t@Y$nooUR1Nv59W7D$)_g}lu)wGt)Ad?}o<oXa;cOjxwu(;;&DuQsT
zjziW~I)sDzZ-%Y>cbPEs+k?pZ4w{90On*9{?n-1=I)3%2atV!gc*V-JlTpsKT&h<g
z5BcoE#3KARU%9e~)$U0epA#UWU&=ZRUG1>)Z_ZThaUx9s^T9;ca472Xaeu!Cf4>&{
z2BzOoSxU*M?E_a|U(bu%&_lA*7m>5I$W0^_?DIWMT1qi2;lB0$jH^}h;!bf3MH>)G
z-%M9qF;s<h--HaxbU?{*0ei8;%EezOO_9Imd#s+~B9dr>f*El=%fs?}gcH>UW!&7G
z+?%`XSp^BWI*a<h8R>6kJo|)}170qnCL8t;P}FaHSYho#rp#RSH;4aTx#=*I#umt4
z9jvbM#KA3L#6j>-ubm4bVs$N)9UvIM3P~;P+j*#w&Tz0OCEm>Nw|f1N%oHo*-Q_yA
z&4>4=8tvcZ+k>04t7`j6)Ff@NLh#20l8^-|0vVRKObBWoJ`E;@g0WJ&$#w0tvj??S
zeWI$PS=KexIZv#4C)E|(+hGJ`_8&{_>z;}J3_cLC(K9m6=w9AGin<9gXGM7$;g%r<
zCj+)bhGw&EsB9zf3%qz#w-NXQ-U5oo7|e?|k5V)S^W9rqL0#pe0*%2mgxqbuSO0Jf
z4`VRnJvT$pl_jdv80-L`aE3u|?+^T@^zpRa%g3%42OZdGsq3SrML(qvN$V$4{x>##
zHfD?j8|^f&q^a1)Tvuw}^nLu+j_23&_qB;CyW_Qw`MAefIsdXqcZXBzCd{F4vOLqK
z0q<js2?B;|qK@j3Wt|lvlU1Cf$VBcO*lYD!c=>tXo_PCf*Q5h3q$iF*UH(=+Yxm^l
zrUCj#0dI|-ODlyVXDyc&yWVq+@Ha7d3V8jb%JXSKPpI;OM`;fka0S9u@;uJEGXv{y
ztaJq(JY>7xHZ%S=SC<Z%fR~@wgAPYe-xpc4>-Y7W+px>Vx0l9WH5T-2x^2mZE%p}I
z$gdaclfN`|^9%x`+Xn~pN#%L=NP1Zs^Br7I;wlt^|JVKjzO;*nD`hJA4}~SD^PhpI
zpn9)`+P-_Cu<lNz|NNsKqnA{CU$WWj8?`wt^c~nEIz=etU2aa!P)*p-gpft%V?^zG
zsr<G|hiQf?Ip&a_VKGw$10(B8%`_ew;@&jS$COf3QUNzDEN%McSffSd59k4kpLj)=
z;y>xs)4QJsI6P4*s+}pZlI6tn%OLtGrZcWSR+0%x-AU>@J#!x@sHt5T*$#GsDFhbq
zs>n93*J{mQTDN$AF5Sqn(e5stI`MAe)VH_py|H9Q)Xh{2vLpA}zB_q1FD#79zJWW+
zyp4m#Gyv#xAUEEvU-N-<Y`_&Aqa*a&8<_;C8E2lY{z9!2ZP=97<m`$9qWgscR3u1N
zQTAy8x2o!ybC~GiAPD2u$NAWSjrLP12e_X)2<d9cgGWfb+nBN~5|gSU=;5#<gX~2R
zB5sz{Qj=_1zjnOhja;mbY%7%ql7lNMCdtWr00BCM6%50Hl;4%x{6xUEtAaHK7P2!;
zs@^X6*iD+aW}C#w=K0Ye!xlB0$538<5%iS43%o%HA^7-yKN4nFio-eVSL0I>!;o+$
zb-RrPIdcV(sMRKqLn9+2@7nZ+Bt6I7O~J!llzIf3Po9_pnJR2!4;ml%_26O413WMg
zv}wd<!tgyHjVJOQnMR{jWPz9Xpc5rI!N$)d_vx9*K3t_^bF!~g+hPOu)e6$XD!Guk
zAT@_`?FYFx8F6zDtqK0*hXvtsQD`(CiRk-?It<k{ZcORPj0%VG?!{{V^LQZ9ylj4D
zD+>7@b-H?kgIBK#d=uPMt+V;#rC<)AM91}-Oe?9#WNAQ!f-bfjrB=M<rS@A_2gli}
zZJ8OFLw#B{_XtS)xk|M(Y|t-t-hwvUiL)r2AoKke2ZV340!Z=)31s6i3{f4_J)%jq
zf}AxQxEdV~X^*1-u5vUsqs4%0k=RUWHxvx{cYA{YE^ef5mdWqHjEF%F5P&7fbf3GC
zVv~o0#`MS2`|SGOv+CPv4S4gYDE+BEgG|pnr6j|x>Tqpnct2u@^>rF-MSD86iUj8?
zxbm$_H?0UB!4nvdp~MY67d5r<pKAYZUe?jDWPM)nUJ?Q&d)esOu;-!mAjht+=NF}D
zRWPHjq*eq87G5|1woi?3@orFcpTMV&nYE7V&bEMyg9^<7xA$O;7w;lzGNtNJzbwE(
z5N}=J(i^woM!iRXg&9KcZgl`n9eGNUKAvFAA+B_TVUW5#3hXt`z!{vIQP~KvwhTX7
zxu$3$&c&v5+`WbC0BWu%>ug7G;&^w^(ZoQ1AQjosY&TBV^!j{1xazTE+q<{7K-=7I
zDtGxNZgRG@Hv2cAL@mKlcn+vuORz5KB41y7KxyomLfg$Z6z6BKsCZkDUn@LP3Fo{!
zkf!stohY8S55f+-ynleknF#93<+xF2r{&zY7@CB;!`(C?@2gGVa#dC81{bb0iOTs5
z)+9JVwSiDqpTUwOZOyl5qm#HCA}f@P6<82hl#yR9?SnKafxpwreaa}n{Y3J+p0?Xb
z;A%Cx&-s%7s;{p~P{P>3tM7Cfw|bFc*l5CHX*9yE<gt`-KLuZT-CS=M&@8*Yri&&*
z*z0MK;HtK#qoZfemq*JT0?cnd5OkVtK-9-k7W#F6$6`9iyfG00bF}@*2AwIkEkFzX
z;fhF}%1tD!e{^pHg|l(%zE(incHt>NKsHvVB7S!`A8hko@=CLTLqI=AHSp{w!`%X(
z#2qgx4Og=3<8HgXL>b@i?@nf-)*n)E{G4bUe-yOGJANaII54>Xh}X@7<&8VhKRy0{
zQAt8qnQ#~mI>ON1VfrF?wsJxp(ErTz<^|(@hbOc(`vP<M1L-c)S+YUDtfK9=Bj)9P
zo8q>Fk_+=v{h=Dj=T1BxFYsQtlne}ndFt>V%=||uHYw~iI%DT9SL*F?<lH|!xDdQF
z@8n5p_4IFTVlx6fJfUb*H%loK`KH@Rgnlz;roYa>+NNfEboXrAzqO>G26)9nnPU~n
zr4}2l|HK^X&3|v$f7ruL&4s^|v8&hcJ-dw&D<Nf>f}sP~Z^XH~af^pRjGCLnFHC<7
zq4F<yv135l*P0eur}4k3wG*ctDer>D^s0*{NF-!Q7$1B=N|!ZQlapzVPhc)<8D8mI
z^6rn*Fqu7-*A&K;LSel3zXI*BEF{eRqNMTM)pDGf<CVl%7YJ8U<^p&;cDRKo?5_F>
zhNSNHAJ;fAJ_M**{=Vnz<Y6el^7NlZ!pk=&Uomj^?fcG?jt3;a%zYB`Uc(J!N_Mi{
zXX#EZy{rA+Zh$mQJzOAawlM_SZ8c+DWvysC{Q7WZfcZg<;P@l;9UxDCL6PKF%t@-@
zH;{+Dt&RPH25k(_zlui{R)C)C<i+!^G&K_wZ1_P?JeNqAS<a(MLbreP*7KQ4RNb#!
zX}WQcbeIaO(uWX6rb=z8#j?BeI^(<RQxxvDhkgqW6n!V6w2FgPeBbCWC3~aCUd}x<
z+1D`6M-Uab$vQ8U0oXGSMciJyVtvc{zNw3(2Bl~fhpfSw{xVF~{XP>e3O;L0N{0XT
zSOo5FdZIlr<VydRZ!j-$W4vKaJI%0s5=VvlpzQIT-Q9-;yk+Ml(dKXEMrCGKQ*(aV
zx4G^vK}WvZ2$qJx@@O>12;cp6YQ`QDO>5trT%K)P`K_J<V9=N6uNYyC{8s{BdpuUR
zC7+CSh~Il;`iXSw&sY!~_KtVp^bLH%ZT>{NX59MJu^za%^=QMQnpQ)pS)?O*e-b<*
zWO|{gL7=5owQ)zY|E<CA>f{i(*#>Uc7w^h9%Ypkis)m=o4yT(<2*@}2tCtdPO*)wv
zh@%K6u~7qa(!~wg=<$hLd$|KQ!S|$Zl8FN$AA9H0b{g#z@;nw^20Nv8ds*k#x!DYa
z_4=&6{%ybK%Hxv6Lwa&_ba8a#`1*T@=_ld?O>143ea-@e;mRL6ANeJOgh%IGe~lGu
zscq@TM7W-t%PHB*?iplnrh|97Y2*sbsUMk^mJ|n|*5g7bL?dN!k|d|wydspj6J1*}
zi<=*(r|YIFGOMQ0x77S^pucHEU&cPu#tfFa5DAA)B&QFP_7cDB4J`OC3uNu1M%EJR
z-7j5JB<L6M_@E<OIPX=p_CWs>JeT|7BiPicm$H_Alfh$;-iGQKO|5w(ba=#%U29CM
z6actOC-jQ(DV-VkMjhz1G4ztsVrsp_Fz2~%W)7rpn82SrbaaGe@j<RKAig^vZHIC*
z<dgZ*q6tK4pu6>!aLJm(?ntqfQpe4g$W@O189h3t5v54|QEN3cx+XR-c|a2?n;;Nn
zk<lyeG*G8yLztUv2BC;mPE&v)G~D=kfC!Gkof%8$XLOBb=|Dezm{@X}py1e!(ff&f
zB!PX~OEDe{KeQ6c-{ey|W3912KmLlBIJV)3k0M13ZDV3z7wj|n?in4;M%3qC>KFa8
z;!4{*`MioGHx{Z5)1#N?IN{`a|Mc{-kur`)J{Hz(MXS<-TZM{{5zpE8?Vqg!R!y99
zR2t^UQH2fzZJ&Yt$<6&sbOYrFYddyM;aO6!+EPHUL?XlvCBK(M<_O1ULJ;4icw`Ag
z;=(*@G<BMdwCy(YahG;+3o{Z}NX+?P8XUJ9xWcs=4c$%IPtuUtyEo0_)<=Vzs5oJb
zXE?bk#4AGv*MD8ZJ+^c(1H|@4ZIp>+#zMR~#N^(E{-7nd%E-wkpl~&LVe)HgyxK_d
zrSw}8%Fi2!iGDmUNpVT>YeyX4A|a}+BG3B|={)N_YXmxP;)Hx>u6{c64wpBxKALHq
z0dQWrzZNHz9ZUbg^L#R{w5Duo2XAwOaL_(~fbLf+$}Bj?dS;JP2^gabfckj#G`BIh
ztL3CjtSl+FB<Shy(opZ5jv#81;#};;u|bFYBp96Xrc54<DHxp>pfjUY$az{==T^L{
z?J|C=yni4^<k}NuvIw*qoa&v-C~9-=Xep=Zevn`K+X<4)>=Vd^5q2X@Ar-;sf#xK2
z=cwmhuSNw*^jqNoG1Kq4F=xAS)E%nDgS?KjX$Rb6KG;BybsF>=FJ(isB@)Cj5`dAx
z0RarmCO|hN7&+{pgUs4`oZB^Md4H|S&!lhi)r@qb>?sNHkSh;FfMn$8vjR8>2ID1&
zla{cbpY+;mnbdQMlE34{h+-BL<#tG3I!?D~xDAQ!?g2Fu1}@(OC(_NMj<I%J#$w`a
zpEvCcT$Lfz%1q`awq|rZ75`~F+UaK&EAD*6AA1C43#<!l%AVx#<uAARGP~>Kq9K2j
ziqt5M4XkC~$PYoP<;5W(4P4UCU)STU8BB=S63gS&u8A2DU*uDNN3RN_)D?@azGUfj
zZEWQ$jR1De!%5qCCi><d51q!dOcBgP9!|vq_FBpWBXiY=2e8TYIz8XOjVMhz6T@t;
zlX~S;vrc~@b23)J4sVGBVyEJ>M`l@So&SyO{M^Mtrv#}e*|JV@?1*2TxbE+0x&MGZ
z@{^}kjHRK0!aL<kaH78*ep35<tW5}3B#Mco1|WPAL))(`SBm`HEe$W!vnY+BO=a$)
zc@>lTT3SmrBY)4@YQptKJTIZZ;BpbGyNP2AMttOAd(t&;NNcSqSr2>KP4i?n#L`e~
zH&w{l1e|N@U;Tuep@iOhZ)SbB_+5)#Q$UQ_XZ4*K0b^GmAAUl%RZm}6M_*5*iQS?T
z1gNlS*qzKMQ@z|#ET`Q#TOyl#G06A8%Ythpuy-yJ8WuE;M>s}GEM4x=&c55e&aN|L
zFms6YUl6XktfYP-bh$H*&(AaIgp;)OxG<^t%0+KhtWgKI`V5|nfY+p0QlERnq}avr
z$TFfvuY4lE(tzvy{5+L<wyfw!_?b07P&#SxH*VQe5Ut%ssK$>z(;)~J$`tIgn8b9x
zd7q*_3kpC}=6dfyZEN&U>NENJ<-&5+nP+Em<F!>SH%;f~hxRiq$usQ)CV`Ol@sb0w
zrH!1|@1A7l%su1Be#6|%{FGFq!*X#xK4tFuVl#>G%ajophYH7|oi$d2V|m^JOq!=w
zsVBLj`lGJu24gO<dfHW-6qZc+ubD>nJ0FhB&&gTPoQkM0{qE1<i|qKdoGo@warcaE
zb3{qjBIT`XXxaXwi1pjsTU$woS*~vQ80fVpeRt)8Qrb4W5pM~u`rn6Q6t{MMTxguI
zD0!Yzj@+Ac>V7p^9z>0q<F<(+P-K;c_yy}O``#d`q~RW_&eubuzSB5X%Nqi1m}m`e
zC>L5apCsen8{*m<3Va16s+FcmR$s!6kdHfTtqy%A(!b<r*)lEWrNh;J&`1-wWxbij
z6NsVGKn7X|_%zw}!9N|K^;pz3BREr>`p9b@O3J?=f>t@P2};Gqyys#Ax(3>(^=(rh
z1<W<C<)9TvKQqZBszq4eAd9Res*6_K&$W9jk&ym=*i%y_3W#`9yU@s=TrRpv0W~IE
zqU%?|baLsXE~UA#3`KczP>x=B0#i>Yk$fmo$4|nIlB83f^;jl@X3Xz-4?}5_SgR>p
zIE6t}6wX`>H29uZWrzYwFC#Fg(<19OljYGr7}eSXln#Ck*t~>?6QghIv53@g>JNuJ
z!O%HZ%2<wrw@R=VjB5RY65@Qy9UnDQneG07$8)I(B8#k=bgiE{4y~HxXTPHj3TR+e
zFr9>CKv%TY+vV=ffkkO}x?98isIG(tQ)6Tt17N!jzDu!~=0hNRx91Dh%MF@WjJfT~
zB)H-gv|5SpTYT6aCkp0>^Pl@}6iFrg6dl7z`u<x8Ei;mM3MU1H{E;IZp%{J_6qk&Y
z7kg4inLhSDn<_|)ot>%eY`6n!d7g3EZ|V3sPGfce>o1u$Fk2=jE}8B;m?vGTn9ccP
zRNz6h_{prOLlV%ZyT}6XU3K(zr8b{iT(r@6DwJk5)xS|Sm2saI;??l6%(hChQJZ9g
zNsgMf?rYQ%<n5Nyuz>ik#v*@=h_zzTZezaHFNx>|B*;@MVEeo&fig5e0XHP+g(KFd
zT{aI4IR-H|^b|0fpx4stM60RI2((39oL1j{>T!JV7a;AMU5B3@VYgMP&~9SNdfe_#
z+k{RbTUPOufEB%>qTTGE;)GN0XyNEqsC@D2>bJtc=M%)llX7kzL|P1?=><W(MK*M_
z6r9=N4Wbl5^~lJM&rycLiWL0CvLBcgi&6$2DThb2wD`A17RjoB-0}?N_Li2mX3l<$
zA2Y55bOJ1aemcA#;&UwG_}0V|nprJ0IX5;G4%QNxsSNPaw>*Q3ds0P=laGy*stENS
z!h3bCD_Hp&Eige@N;Jv%p(U>$it3RQ^`zlCUBbJi6aA&`m1%ykBk6*lXuHtT{-V7P
zl>z-1WzA?txIq9#f?-*rg%Bnlje#;JN8Y06%>YzpRG`SJ-?n91X;l*+b1<mHmPq~L
zxhax2KdG38{kWDZ^aC6>Q7A>ahdZ|r+VBF_CLddoZj(fnVLBXHfIdj^!}z|YIhIly
zB;5%W=?VyBp&&9Lv%4?z5qiQ%7^0wck!G$?R9Z6O#s@hFrZH=`8#@r<xI8Dx$6W~l
za`I^Hj{oK-)5Z}W$#(_2k_X<WAhV<GY<)<imn?@kPdSPT#W)beqVA{mWTrHI#|VOe
zpc#VXI=lc`6U|4OoPya$@K1EoEeQFq0n!SQYmlwa)(0W|Zb?)!NY*@RjU~1Q1Lwef
z<Xj#3J$^=Q(++jT#M@-e{xP_8@H{0TZAJn~+!{3eMNPrf!}Zl`ad2iVZ&)BKxhF@5
zT!+DP?W7~xP+$3X#d3S_aNWMy!r%t$TNy_xPTf<1cYl<ruF4#+U%TyJ`iftknO^O)
zubLCBA~?#z8R2F^4eWSKiPG^&B`{c}hO|sUyeR}^S_%#1zyUmY^P)roHAaDd^R~Qc
z+}P<oN4FO_>Yb#ZFk~1$rON8X(ep!w9xWxj;mO^^CVQc0FLqMPY@X>nDOg^*aI9$S
z3{jR)53hKqylE1kL2(bnL<(7=r<tWrim-3^JS!vu(VSLyP2>{|?p24&$&f;N`*D2k
z=S*SfC+=035*||K4o=yJ7*U`al-N!Fi;f!alm3GCi2x4zvBoFiwp+19nT*ca7P0R(
zeZ?ediHV1zXD3FA*9=II4@svIZC4qK$6pDR2+I!Yi86)p5SIP&a-m`WU`(+wEMPY(
zB)Cq<v#Y%|m{$9tu68!rbk;2~FYaffiF8ha*(?q+RcFUwmeZKNItx5qAGdb#t)8)E
z&V|;;4HQ|A#2I!fCh@I{uQmViqUNmO=?J6^5>0dBoUkN~KORw7R;WpKr?Cvg5j8i{
zZNiv%ajDon3U*XI9CBK8V{ZJ3#YFOxQRmtnMw0~1%UhyyQTn0=8ncw%V_@$Wk)7Qm
zfdqbiS`J!U55lJ4Y``h$J@gxj*qs9Q^|zl>2`S#k?rhb`Vs{j3-8|r(&75NvG8|0S
zAusZ%Dcnx=CR1Dn-8iBP1;_T(cJa3fmh!>8-B9otsX)7)(%VF}*F0)3`KaKK*gype
z^tCOLMxz}nl1~$-)YXsF)l$x8_Rjc3Hcx0jG?FRm;$E{r3Us_xe9}{4B&yx;U45qZ
zzMX2(yls>@WRpAHZsi{9`qt4_f5TT%)}xQoS~O=Do`UOfG^rNk(DsH;e-656W4h3u
z@tSBg(rhcex_C5(${jVsrS)<JDyzhsPm?@bo9Rm<B&!@URj-<4_<s~K%0@;|g#uj{
zG6gRiRp|p}0kY-GC4Dci#($jU3LMIqFvrU+?jH^$^-T#8)1Bs8uvw(S^M&&A-&Lg0
zp&Z?Q_U$@*X>l<mDu)z_e8hW&yvG>-OoV*{Y#`G?i${YG#s`Zn3q2;nYae_0cz#x@
z3fS-+Jy!^h<aN8Tl{DtqJe8}zN^UBi(Jriwdk0J60}NYFwh#tI#VYF;o@x|K*0Wrl
z`NUzH@1;He&EyT>gv1O2S;p|HOUlo`J8BBs-S>$hMPbd!b(lGqR=baO2BlFV(o{K3
z_7gq!KkI<*0SdIFEFUI;iRsZeFX}^<ac2{^WNBC*wU~SmP77}*RjZ}_jvf^tOB>*^
z&A0Lf!mq`_Jl=G^5!(r;$yMP9igv1FlGCS6=s!72(p>l(R%50d@;$qz9J*_8MryG6
zY;LWcdi<Pe%M;CrBF>E)=ZeA-?l%LV;%JJ*S}U1ZYC#W4pM&4GDOHBT#bFF=W@u15
zEI3(4sXsP`&5e|ia#SR0_e_NBWq#>O2r>!avwG4(JrJM3y6N<vY^<%85B?Kz5Oe5K
zRDe=4k3jf9p)b#cfD+%2s5CDNVuFW%VsH`f63`mEJ#4LYW9k>AvK+UuRI{(0m*;YS
zirnKAu*TU5Y%Eu|m}%|o(PWJhSdljUIgi*=l4PCKhdt80N{{+;^)!`zgV4Edm8ACp
zwRtZ5pcLZI<1VDUnvM1#&<o*qJv{P{+fJQh;~+-c$;)tb7!s}RUB;H(nCM4lnzp|!
ztwmRUh+_h>QTeR%%>Es2q6%8GMdLjakgaJ{2!rK9%_ct}L7Ls9OIq4p#BX-lxQ^_~
z6z4;J=!|D#@BIB~guR~PrN@N#uJ-|`G@P!AS0x^0wm>o1n9D>i_3ze)YoVp?;~w|H
z)9z;a6YKa;zPP>;+@Qb-KpiyZkrghR@1rG2?Cq!mgX(8;R}&tDY3wxGPpxr$ea;h@
z`ml&^r9Y5$EG3q|qL_h>;m42gCi4}z(hgfrfM&dqVn(+qzkYvUQGL*>FDyV#h$G@}
z0H2KVU%a2v(6CKIgZ>%aMb}u3cIg15nd83W8GGA#20KylN-r$n3DHg}HGa(l?xcCM
z3X|2dW#(v`D%PLEb}VDGc6%+TXfl5L$Y_%2NB2ICwX;1Se;)kbplMc)*MBxIoI^p<
zF)D>Dbp}aBvQKwNUTtWMC@MVxM9$}J8w|KGNy?*WHYO(s%UDD%v)o(hCV$8MiwnKT
zEi`L4Kh_dd_-VSN+w`L{A18Kw!&8Dj+9>_;hj1L&kDj(HN~eb<wMGTR>pa<Y+5CMV
zwKdwh`bim;IWlom0P0MI->4CZ%ZjO!GSe9Jn0H&w4yE_s`_B!p)$<sz@>o1;cXrAs
zlKWJsD$d6UhCnHX6AHv1bRb%c%5{pzc}P{~kqb)Hkh#-@_Y^p$RyU1IiKd&TJn2i-
z5c(7)R!fGEA7^S4yH+i%HImJIWsVTP$vw`^zLse-qwPB9+7**xC#~wv;b178W0h{X
zaPa^_VxuUI0R7vnLIm(Rzm~xGPTAHvwmZ>k@17mw9y=m=Tj+v^w;)hJ-rBe~zp|V2
zPvy=RXhH3lsz27>oxJj5Q+NK8Ln5s}N&|iF=^RN$Ez&tApv+(fCJTyzm34BLa&pP!
z1VGd)*^n|~s&=#IIcYVyifY?x1afDeUlPfu{W#-!c)y?55lHL_Qwk<Sv&nT9dk@!M
z&Gtt;r8a?Z{tzx>YZ&v@ej$+8C`p}OXT#91sFxQO>Qv%k$EN>?2SLW$38e!egTAC`
zZywm11XmA)p*0?AxqMNFuLu9Q7l53tE&ZW@U?QWw2;O`W_64o7`q8SMO+pZ#5P|r_
zgb#sz{!=Db2{e%;2kUj-Cu7pq>~i_QPKlH;F`ob@zEd6;`&T=WXWucBhQ)0?@zRIu
zs-mGXlVM*T;lVk(@YH5?7fql2*e!WqB`L^iYFJ@WCg4v%ANB$NGX%%LheuB02Z|EG
zWx#@*2$Dc#VdX2$vM2JwRI#kx?SWyuc?uQk)E~KC10_A}29}muyXeY@F)LdNR=NJj
zWFA7TBZGDcs5|n3sbk5f-rB^bFTK4cVMX6yyaIvdkqW+!KmE5Virdb%63<>Jz(U#~
zVO}GcQ{7Pv4u-1D5pAIty`*7spV!e=hmziKH}P9(BEN|cA4w-M&+B^89DY#aW0!lv
zc)!FafJ+gm4n5zeGIj;EFsI_H)S##J6zZRFJXW`P+l8Uc(`j2t<>!><T64sK_-UI=
zIDG2a_?0KWt>;N^e`nt3*-qw~zOJ04T&J}ktIuQO%zosg4d6^Oo}Sh81>3=O;N-2n
zX^Y8x<{Vu-61&;GOfhpVxLby@)(?D;riEKGK=}cV+rZOwGtCSpVHU7Uei+G?cl`fS
zP4hs4Mb*w)OpI&d57Gt@{0ZsrSw~%cWd5kqlc_0oCnJY{iSY0-Ui^8@rmy^!|FK|V
z<M8UJrchetlRlpR5^CbKCrqC43$xbP?abuwKaD~{VHcYDW?hvjUs6+KS#|RY6$dpa
z0Jp5<nM+)2-`-b#?53F#n-Hm+N>xUPa)x&-DPlJA9^8eYV0-Jxk(Nbp7C}4tdp-V1
zl_Kj)RGggRH~i=PGE9+Cc#L4e9H_dY^0T!3N5|}*c=U5+s{Z-rxky@RMOM+r`0kM8
z)T!DEJ2NWTkx0k}&!ZOiYVW<BUtbv90jCK;C1{Fz7PB*nmXG29ZP7^<kBDO_tGXK#
zJc_rTDVgato-;DhS6QxFeSV^nklDPfa?m`Msv*%1`4Y-s(Kt?<%=V|ydiD%Y3>JgV
zK4N>=I^gCcY>p`V59zk?D-qIXqx+I^EVVrSJ>7lHhD7O)+X2w;6dIrMA@)i}Kn+Pk
zHnVzMQ({gOQ=_5M^H1H4_T)=~#s6v@n<6S^KWjs5O=y0vTwe}YbBa<i)o1HE9-1c)
zyq;Oy#c~k}>A-E$hTRPfxt%x}X9%*mMAfL0Q?NoXr2|67rLNM>UYcac%g3%P;8>+k
z_9R^JKzx|jXlHU<(}PtJGV?!OKyU4OAM@;>EHjflO~Ho<#@OTvF)%v8WKfA%B8Mud
zaNB(wpGo5dDDl6+5x8wyj2SqJ+_?3;HI%-omHs|+6mKw;qLJntw-6}b-Q*1QbhS$q
zeonpZL&EE5oiJ;IFX3v7pBwZ-TBR4(J*L>-)vLv(oHq_<Q*(Cm$PMTP;G|2WD$uZD
zV%jNC|Az<8A1I_r0@DKQAxaZ&hEltj^CWqBLAMN9oNEV)PRDHgm)N%|z*p7ejD+Lb
z>Pj*>A4S}M%$3j_quy_85dHN(Um;XZ$4^nC(h$>xoSK~ws76i`cSdw<zLVi6%4Hnv
z*lo!0f4fb)sej>x;l3X)#MGda!CfY5d8CK_S^nJFX@tijW(j|&adQQqVEe^Ovoxb>
z<uCj?iB`129f&GjokY9X*+SX|iF(yhI6RDrNl4lGXL<oAN-R74A?vDuBA2=*-&$nH
zTkT?}-ng!svaIJ|*z=J1EpdK6bL3p^=dr2Sb>0X?87fXRA|F7inRj;w)2Q2?pATk;
zy0ji{ng}SbirD8n#rMEAaTcGrsx)oyrl(EG-N#2;pZk9_P*4%cfE~rwa~7FjOdS1e
z!lU5%AfpJ{>s9FI{kpSm-ud(8pAP|lmOoxAFa|?qnRti1X!33U_BV|c<bN|vYfK1X
zeGXzOTE*?dJE8$YI%nQch0frxhOoJOXot|dhqM&dy$|Vrf?}Q(zGgQmbkL>3Kkv;(
zLpQB7$C-Zr2U6*W1!cdZ;D0#J6~hRcj9p85-dmS<s%;aK9cS(HAQ$;uhOtcGg?*X5
z;7$C`aO9i3=cS!~0cxT*Y=+?t?2LT(asx6T`1s>|R4X)qXA^ayTrXb?yZJ+$RgCRG
zIv`e*Pv>~#^<!QSkM(%uMttYKh&RTRrkKN?RoAd8$tiu~blQyf*~Bw47%pYVgIK&!
z(D2id=@cz39d4%DDbYU9>8ln+koD=Nsn^l7kX5vPF|0zeLgNoPz3m$-K0msTr=eI@
zk<bqrtZoC^3bpvvUh<SAsCN&jIh!J6$Y@g&nDK~M;P>!l!cEiP^1m!d|JDNf2q!fu
z`Awjtw!2~d+MG)ES)s${kq~~LsunK~cZY^OEVY-W&mj@L*P(+1IyvUK6EYCnZb)VR
zJ3pf69mIG4#O;+AFzpnXp!cAv@40b`V_pK@*Qo#i6d;Q%iXBUd`}jd?CR|~%i&jpp
zT`}~%pjMi}cSe&wC4_dk$`4w)vQSg>pEJRgY?aQ6UF^DZN(535$(qmHja-wLC(<c5
ze5Y)%)NJ?LS(IDD{jAK+vPzWEsKxzW|A|0p)Fj*M%ptCHDHqTB*OF1nqh-Pxblr?*
z@hi@#zepT9eo+dz?Ko&Ue7<qh%5K5x8eh^Y#AKRUaC~vGbhLD3?Iu9_jAEK}6u~bk
zm;H>0rXV{(JwA6<2=-fqEQyj4T=AuNTEId^=xlY>scOQhl1ZD|X5_u$jO)L47@r)o
zqagbR1yaPBH2dAVeD&%=cC9WvSF$lTJL_kdtKK>+;Ip}LLzym@`x+yzEULOHvSAfh
z=_fn(@;yG3YNTJO#x>V^Gf>vVIf_Od8J@3Z^^VPN@mq1Z3v>p3|EpS6uMJz+bBNhn
zv0i${J(tjt&y;oBTUP?|d#I|h?)-1cq+y~&Leg@*JE6PFCp8^6UX%SZ3Yy~rY0w#(
z+1`eH;66yilM&pRD~%AW<X_k8-$jRO-UB7emdv@J^_*eduf2Zj$;l(%T9a)AYoe;s
z4g28(w8)?zX^IcNO#J2LTq)~R3d9dp{EzeiEa)biHQ|A#<R7-W5$C7Kf<a-LLxY8Q
z+21OZDZ=X5iFtWM0;#gtHi)0Uo*1^azW0&#VZ>F^#M8*(u$UBRItc%0Uu}La=?Cgt
zIQS{^-U8*o@VKVX%km!Fzu_rlJk~@7Nrn48HdA2o_Von8mDiG2Qcwf9_X2_(KhR1l
z_FOwmCWG00XFgoIn#dRtF#1c{eS57Z6-i4lZ=TO{&#FJliy;I_A19~MTy>su9So$c
zml*!Da#-l<?Md7ALvaI3BQ}%T&mBLlNGBwpZ&X!TZdeWuV0FuPb~`$pTDB=`48!e+
zDi~UPY*)U>Pz6vvQR76&n*n6JLZ#@m3GXn_u*le%3DM(Rut?(@f&2EPAZqA0q^3c5
zS_<{OP?}XCy^nuNMf3C122g_XHFh&J&nXoDgoaL*8??Imx_&&#GTfQX50Q4}N;Cbz
zrWAnG)6_Ius}19IhG@!4Ep2TmK*2rIN>H$n9h^itlQi+0HN~@>Xh1_q>n#auUfFuW
ztE0joCS?Vz<w2w8@@>TCVWt~4@|&zhtV@Znq`kfNiKnaU(UdtVimaTDfx%A5=0l&Z
zb5cR?$aCq)H{-bOo0+i}iIvkOc#OetLZ;WsUNtj??E+Ifg^|w-!BDN=M|R8i!9hg%
zWy%q9i|#OH_s1TDJV4HH?u(+aMv#Xa5sA3XQ)WsWJPc{qG-sT#i+w2)_8G_XVcld^
z)yU{(o!pz|V9NR3Yo_h`pM|p;0%FRJAYKZjk$GX0M`oj3n(v>6|Fp55f-!J3>D^X!
zW!X82_NWYgIw`aH{je{N{o`f%rs*NHGFy!ZA@&Y|hc2h!L4c(3X&<g+*W~H$04(4F
zKX=h=lr)WUGI)jY-%?4R{?Rv+%K#0ObQ>oV(<*Y&cDUz%CM0Vt)s!aC<o~3?n?>?S
zuqFVHHn*^7a-}3bhkurDHF<pf>}fkJD1bW5sqZ#gQnxf%{UPWDoY{4*HAy{-=E*lm
zATjv)7k=*!Q(&ada8#yr8y?B)XmL!smam~w6=JhcMWq_9n@#iR!$UWISC*5~ze4TA
z&;E<=+8?3L6mvQF`y)ECqn9FWID}ee?b3Hbuyn1N*T!thxW8wB`TZZ!@*XD%($NIV
z_h_Z~;Ds7fhiQ2g6qljTMfQ8k!2=IvTN$5&ez4NK7B7ID{TQz6K470$88YVOtC(#?
z6-ffI=6j$%Ude7D|9zD)J0+@PZHe_QNfGl}k<2&EAd0b6b&!w)NDs~oD%Oi1L<SZT
z&vg#xO?edAm@RhdKBL$+sqx<Q0e1iXvZh4ZF$PP=(4bCl=cZfuJ@)ZNmd}0^qsmg(
z1k&96<N>?PqG!1c8D4pX!GvonAP~as1yp@aU5x}3!2g<2O)J}FRGnJ9SPwr22_~Xm
zHrdPCK@-jjXhgA&=SHL^E7Ho?&a`GmAU}^*^Af<*PjE(I*PEFd8=W$h(MG%H4IYL=
zooSUR`9-_9dMM2kJdJpncIcF*wfh6uuRn1z0I3oI;vR(=2@cR=rfBy$7wJ>`E-ETo
zOQ4G;E&tFG)@?xY6>2uKDCTj{Z)kabs5dwyY(>%VW(bj(RMuMAuaxiP;_B&nd9uFx
zV;6T<>FHr7Dp68Z9oOTKnRHtf!IPHSon?OT<-sb{%zAD*8%ZW>R4TN)Do&QcB4iE+
zHa@15G$Vq>f~VTYY6hHncdxcy?3gjz<B8|uDmDYAsfw&%g;RgY%mEEt;E$cqw8WsL
z&*`Pn?0KtUEl1uOXX$R(YyF!Kxu95d*r(V(&7hCU&1Ek<l_F_*MUr(B3gu*YbRIjY
zVZZ@Pr0uPV3KjA6trj%fpq;{iC_GtUz>N|9s=;<bR_buJcHi=^Lq6_qFUaJ<P-WEU
zeB<dS%t)9Y?@Cf)PK&5~?J0yW437-vy!TS5lXO;EOvbf_PdQ3(1FZSwnv5-KYpKk-
z$jVHS>r)0EkJ@m~LlA-Mj`-eMqD~l9E&t=UqzbdSj|mWz6oosZi2m6ULLz?PXL~Y}
zaO?;S%4f=G5~MKI{WIOoZxS4UZj>dZg=nmKq2@R?=UB|52$Q=ebmarhL;{rw<r3Rz
z+96tZ`t^FpwIMQjN{LcMT;Iko$KC6l8+`+-Iq|!nt+#uAk)mtfkk5b`ufCJ4&(yS9
zm_t96pF`+<#|1)C31dI47I6ahz<k`lU+yxVHbm`0EjvkuDz#%y{oAii9+nh!f65mw
z0en*TrNMQ!0*P-;o7NM!XC2a4yBAncKBsP9q;53|SkQ|qskO>tgpN!k{r$eXc^Ppf
z7gl0*79_0?NigxxoZb(ONnT{pi?Twey({BuLAIgto}-&}I%=ZQ=nudc<+WXvgZ1Aq
z>wlrCu~(~(Gw;cU1?pI`FV|vE-%alqC2HG~1yC)c?3gm|RGv6^>V5Izzqh^3%|hPy
z$G!fYiwX8uynyNT(t^k(`}etSJJD9V2jld2ofq48!mad^3-N$G?NQPIOWSwxxq_Fa
zt@GRYx5sIb&+w470*-3H_y}5O@lm^^6aRJfc!6cP#I3Q;%l3Gd<v*9A&D$J5^Q0Tj
z%B}|Lw!S!AtnNRS>SR#Clor%C+&Q%Vr6Tt<AKjj|^t*e(wYBD|YG^b=CbsvpgA9Nk
zjghb);+~1^9bu!s|5!8)rE{<EsBroBKd*a;xJUWpj!Z2BWIcl;|MSMV<tfoC+#%CY
zF<LSCw*UQ6$58XlEf3rgx&y~O+MEA5H{Jj1Z2#AjUH_-kdV*<sAm$6*oAFsVKVPK1
z@)>k*`(c!9cm<G`Uv(ee%3buTdOEs)!MoiuygR_QVjehLC5A4q@E__glnES>iCts5
z4`(*kPq^ETYs&pr(}1%I&&AM1vCV?o24L#Q^CpPymeJ&_amdc;{GOD6WY|!W@oi%?
z@v^z`CA)KF7gx)2xu0i*xLDO9C8rdwuk&(?)pnOm47fI0@~}L;x;?yV_QB6@xo`P2
z*Cq2(`H(}^RR8YCR5+*Y-sQ5fXUDGplKjHIh&F!#;As1?9@r1FdspY+e<N#|c{lQS
z+8@hud)DSVFZS0Z(*p?g-nNtaH(w6S?$}}v4sVkVeeR|=fwhrmZOeBT*%)C<`rC>_
zuT^eIAoT!y#E)A1_x%n-ezS+&4&0n;GXa5&+b{#=cej&wO~#F5b9end2vU<n><DnI
zyokv-*#1{L=|9=#H*0>wpW!z-0{D&#`)@|Koras<RjYd5Rt+u@?cW3a^jb^<mIS3P
zM#%?~qMMhS{0XJ@`&#$-xwI{QFJ7IuUFHGh`Yr$54zIpCVB`3uPB|Bk?+mWYjc*@I
zZg|kM>>tTfcmxmZ`m~<)8Qpfbv|*|aZ!3Un*67xa*@dIB`<d9hOn;}xDUw}>n6wRY
z=$gdhNlq2~<`R2gM5r`$`=2mfL$t$1&b?Z0uYuU6&CHw3j6d7=7LRt>+AhCyw{QbY
zg32zpR^%Loet&GDx`l7o`4w>bYrY+XS8{(<>b$Jhj)vE_Q>WEy)gA?0vrD=Eioj-`
zMpnfhdXKbSjsKCj8`GJS_dTqVT2Y#0ma&yQw^_W}EeE<Xy`P$UURmDmANtCB<+S*o
z)|TH@y9Ccw53DTsV74|!zBQj}0l+rD@kL^J+oyxe0d62>w7`GWVdv1i=irFwkw^Hz
zY@+d%aJk<+a6Pc%e@$h44jMB0hhAJrxmL!_7LBcrMXUvIqG5DizL3#!y>fT*0Vclr
z(1BIe_Fr1&l|H$n=fvU_W-_0=dBXz3m;O{jYXAL$ryM;tV@OQab@{Xvz-BuiIo$2|
z(MsJiw8*Z}-EC2sc@8e#OxN^bTQ9%5yv9Cmy=z?0yb8x6xtiC;59!WECR_iBcS#+e
z9VR^+ono1L&+UiJG#uX3xc$Rj-q#^@fuc95w|MG9(sX{<wxSCY-z^60uMQV8#jbz#
z&>>FUFGJf-?vZnOUb5T)m$M#^fcx$)X&S>4+}O?Xt4!mgVPU`J`%=eUcUZA5soc57
z);mrIF9Q0DD&Tl@Fc)ilrE}Zg@T`SKa!141F$u#^4={98865>RbzBD$4aK8avB%u4
z>vTalr}51R_sz-O84H)Z$aP*_>-!xeV2|JUpNZ^dCYBddRSxLvrhW3wbSz&+0SBeK
z7uVcZeGH?aKDTqmUdIxuuFp=f{uhV76Yty3PFVZ~Jg&cN7A$r|%`8X#!u^sdiB*21
z^RJm_wl@vHl?N_*B&Q7}aTZF~<Ql0lg*CpZeC&HUG<UkvcH0H4^3z{W0-Liv<hT8}
zR#b=yBoY(=@Gi`SxytU*Z6oym#Z}49HDPgh4fFcLR+QAC%Ol{L=@%9RLZTV3iN20x
zdUE%RikZw4t1Ct?4$))(l>2QPUt6}hF5fr48C|^YVi3eSY(>%CLYAT`yxOj#gxhXT
z8aDlB-W%OUOJN&`7;bZP+Ro({aCe5CCryESkD7oe5jijH(Dk*c?`>b(`m@`EOfer%
zD>3@^vcnJlt8{Hb7aN8S{(a90LDx$?5{KzG8__MdrR**L%*!v2-!qs|yWT(QmzToe
zC0(SWT2^*j4~mUX+BYhS_owgQ)zI?vHHaS@?Ym@PV|nkcZQ3@!e=)rLb9fs|KKI&l
z1V@M)zq`1+`&Wu-y_WaC5_3254<-L5*1fSl*>>~Z=-+SDO?;;4zvY&@?e9d7&WAYM
zr-_7ry8J6ZGTdCN8mwhr|F<dWxOjSd%T04JlJ_3JGy1MX%5P-h_AJxe^l-(}Z?=O1
z4}{>pLhJZmozUG~k8&MI+(ZDrwiyX>UsbPx*}J3r2X{l|*9n%uzpBjJKfu2JA|JLe
z`dUpzWd79DW8~tlb>-#Vo|NAl)t=nY#qYMumHQp4cP*QSQ$9<li;iQ9&d=8A7EjT_
zhM4XByOpnxH6nQjC0ElHFrPB|=xz&_qbe>h!hoA%DeSe+?RsXTh!~dF_y%!R-*y9x
zXZrYWZaJ7dZWO&Iaj`CSRK;-KnCWSZog4B$mS{TxT5l(39LJA-4f&i6-QXtUaw%$)
z;cioE-ha<$k<aIsF7!Ic0UKLzv!=gz$Z%UQ*m5hEc{fADDvlj$TS;Vax;uRh@?WYl
zzIEG7%5097_VBwx0sE6z1q;g37y$U?(Xwyey0v&cR&EqV?*Cte@%1~55a|5h-Ewj3
z!GQ~4Mviw>rh0c$exrAj;PXoIkfHS$n|Zd{Hsb#`^ea~C_zNb;r*+HC;jS?AV&v{3
z>W9zWzxzXN>qG$Va(_V8*K^=yd?8qVo&M~Ns{c~_xyReK4x&}u2!#w?0hJ85<C}L|
z<aePS3}@t$cSFq=ml(me=$k7d@~wlm(*pPUhbOp*5)G*{hGuLE_f2u8+ks*ET@)bf
zKWMXfGSA)Ov#I1!Z(dP#DK&<>DBis#J-j1d#N0Ss)$}!DmU{foH`~@$x%W2Sis9B6
z)7K0}|NmNI7Vu<g!j6o5xJf}mh@o>2vhv^IftbKxN*IjsJ3{4hyUy<I!wLKhdQbg2
zx7AOcqmMyM_T&$?mS$H?C4QzgG<>JM5Awxcrjg^buoT{<SJ%-u-MY=b77_ie+?G*0
zF|V2m{3V$a>RR52pRc#Bzf=wvz6lowPDl4WM|-aGf&JNxru8{t@7?3f;GUK?)l{z^
zsM`6e8xFJig9BlP`FYjo$b!0y{4cmB7f#N`vtZ9YV|(kI#goxeOgoPGZR;Z(24z5n
zhZZ~r@z$t%hA#?=t9BHL&fDe*3wO^e<M4NpkT7Jjwd(itvp*Wn9S2Ly41Wpy{y%je
z(5#ZG(LU-iB%Cl8N%y4$LBk+;e%P(KPjh4iwSbGpc8I9Me*HQ9iu+<3j)s80EBw^y
z^u{;6!Y^k3*J?&=n$Ob4y#4eAeCQriy7;@>`WuzD`Mp>664#t%EB~ga&%IiN#P<mU
z;zN6XIoR#43STG>4Hemm-|n%H4-ObAYaEVQ$uEo*dG1;+ReuxqI4$BQa1=hYtYQ&A
zqgoN7y|J$GzGb-k1lpQ8&+s|G$&WIl-RwH_F3gd{X(1Zcsr6eO_#TIgKCNw@%qWI5
z@MA*H=2mytYLs{y_$APrCtpr)V7y5YE9Xj+f75cKyI26lCd^Q$L8~K2tEVH=iq=Zd
zWYikgO|_Y4RhJ?vyJv(cVE)$$R(gY$eXT5xe6k)|&sd)1S7tcdQg%TIRyL3oVX%81
zY57(Y1M0{vdnP?a)_Xe<y>xhxu$|QhAW&zd0um0&-^j^|0+#S9l;7kk;G9LQIcT!M
ze&yN0tA#pl;RqNXg$QF#N)a0<*_!p<95kT)*S9dUNEjiU4+Ns-7nD{Bu$7Jn_mHu!
zmk{B^M^kVXG(M$NIL(RnJu70Ow#Y$aQ5dJ53DQ1M4#znGtXg!_i23sp(EE8b<|MG6
zfnRB4!UF&Z@@(codw(Ump|~NHeD$MhCo&kWqC?+sJM~<L-e*lwYIktnZDRp@u=AKs
zPF)xz{kwfkAPn;B<}3GYD$Oj5L%%BJ%V?>qMjZ}R-=!}ZM82nZ+&bO`CkKuRDa$Iv
z+4+e#W4|BXe8V)~)kz)oevR0q_dHsN_8j~YQ3c#rF{myi20=9xV^d_7bQc9#13S)a
zh<4B3!W(rn)k*rZ=t0xc5j>&Hf_Ae}tFFy@baZ9?62Vggjo8fd>e@_0F%bju(GhDA
zk&o*us~@{z7(c09OT$Te2D$`wnXD>H3?`{c7p27P`i6V;=T4(Qrdyy1t<WeBPel>B
z5M}fo48v<e1$+ln1B(&X2$0d@;z1c=R+t_fit|`NV|3>W;syqOQ4IIb;Swa2-j8tj
zjrPwVCIm@A5Vv&ov2LbaCu$dIlZ~?7nOeEv@#E#=<7I)8c22^B*;<<`YW)>~$Wfkk
zj&S&_0KhZCw?T}QA!Y4!s){vG8q=E1H{?o+xa13l7=>zr5ePz|$PVwh$kmUUT_rRk
zS=5Gf4E&y-lwuKnqcyCAInsoTBp|D&$4iXHk}<pN5>K0vX-|_WKX!}>4=kAR+6zsV
z8{Tc`0Q!qk(`(za5tWH3%M9>KJZ99l46q~c$;eU9oD{i*$X5^5vrc;NG^-*A*^!&P
zqoGjdM7^uIvO&P*>zKgW!#nZPaKp(7yrvYG^q*-uXdne~U`4B#^uT`t>jwSQzfwe@
z>MgVgCT9<6OEOv}Wa`$RqiAyoNHg44XjQ<1^*5|B1)Zo4f7q_EqMNyKYA<Pn6$0`C
zwrZbFjEX3-k|wLir_v-3C}~1sp>$aa6duk*0p7$#-)Nb5o2Wksx)8Z)IFfe;!^z6Z
z<o-(z9RM9vwI&v|H=BQgYijX$2(VHnCudf2n8Fc+81+hYl9ltgG;X}%%I9zo+K}b{
zYH;r4Xq{bKmH$Q4xyLj4zkhs$Mm}Xup>mi*jB;4gJ~_AL5aqBrpAYLWXOZ)*qGpyu
zkxfDqF=ujEP9>QbOR~u^iG3vJQxbmr{_)%6vH$kZeeZr>_jPz(ujdjNdD54-i6YXP
z0qPs7q?e_3lDIGU-7rZSf^5j@<mza66$#JE5G#>|OI(R+Bdp|>`AIG!!M5Lu&kFjk
zHTX-Zg($u-=F*)K(7P0y;co5w)cLL-Rt`wD94T^ck~!mU3UB@4YR!l{=kD0s>O#-k
zhNPeD>cYIUwzS3goL_%C{xZuhEJRHQm=zrFM;DV0g%~UGX=QsG@S1=XKJ+l=N=mps
z2y+QPOY9dTL8bM|aCu|nkaH1^mkx*Vd4ZmU%2<C?La9JchjnS>UUASf10_QFRnF<D
zXh0nT2gZ@Q$L})U@-gt4_Yox5vyZO`uoo|OpCXHec8pIQWq&aR2aavJA4`J*P}J)L
zS~m1?@ejM#0~?8|N-zD{s7`U3^@|0&AwFmbU$`da8{?myyjOOY)lR7W#HKBM!$Qw1
zlx{UXAD!qH-#UM)O;xnzK7k};`YX2tr1}`5V<Hx6n40RG7x5xckldyH>@jOq<L*sc
z`w)2~=r++>Y_!Xr@_>F*H1lRJ0EL!-3bASD2<NbcZ#*h32f&jJ{T0j7gZ-qkz|r$H
zhyE3nFQaBbY3Po-NffBclIyjjT_3Ez)LT*R(-ak6ToltuHa?nP%y;{+d^iRJu}AqN
z1|}X7%oGtBXS@>(#XZ@aI2wc<9u%{E-bId<_e1iDLq0kq-&<7lS-<PDvdVtT;0{lk
z0#6{QW9O`kv$%6WxI!`rYkt+d1PhJsow<<c>V@T_yu~Z<VO3NjPD8oM%0e2MAWc(L
z1QABFBBZ+YCG33#%Byp(Qn&}}M9ysrF9INPJ)Ez}f-_7(D|VJ@s8snV!cA60Ae+Y+
za(wW9(eS&$7z4iuJL9*NMd_L3j~F)$=AXOcig1NDxmmnHx<8*3^IZ|-LCBLX%99mI
z<+&l>8iNzf3Aq<TN!_-{lc0C@FQr@_YF+K*8!82tKsQgxDZ=?7WBBGdDS5sgN{&Vm
zFBllc`8-#9bCE)jQJ}^~EIdAP?okR=GUADT$_+3C>}%ydwpA|;N*gML4Yz;#=cKX!
z37qL?JI3O{tHp^JWsAp())H+bnpxfx@`O?gqFzd4?%<$K`msBP%t1Dj=l$h0>OQ_3
z&~SSDv`AJgTl%=3qr?vKp2iyj-Y+zXE(_d~cCEu(!HWze6SwR^$pt@sFsw<RWMh<}
zN}L5hwN^;`bW}2P<;*Aa@Y?J&?kXfrSKS_;B7~WTIK86;@RE86E9Q3!a2SERXlaby
z7gR?(u-FC<j}kFKGF*^$natTGUt(!qcomJc)>EM&lDp#FAM{GFNyV&*^@ty^&Bu_3
z&y^}!S!vl<|DeN;r88jYtcOo_BitmE*LKsErnX-dsp=@k5fzAIgY;1!7+%56dLgog
za6iw>C+6o^fwfUVI?=DDTGz*b?N{jF*VQI?XJM~d@rT+>ZdcFse#CaB6ne<lABsDB
zu)6nfH?E@qHtrg}z2avYOb2@Tt}Hx!SluvQ65%^*%0Js1az}FdOqzr(_Htzd-ZBf`
ztaI}i;~?SHAuGy-tH!=jG0MHtuo4jwRrOK=kIm9QN4XYhGlf<lhL!DZq}^QQ;tw~|
z|LCic9rklF@2g+wd_fyy_zPVxj=q56D!q<U#?wzydc8mx*c&-eZ3&<ScbeTSD5FwK
z6Uiy^R*a!NME6XtGw)^FGaUp!{v9Ex){*;7Ry|h1!P~og7mgJRp={5b<+1hGNhMIs
zQE4;3*Y+JS$1j9;r1K5#v+BaC8me&pYS7&+FK-F*IrtN(nrC@+BC6R6o%OhizL7oS
z!W+bTv#F$)iJ(WE0lxD|9Jq(=9jcmiu|-|LBaz6Mj)~=_$#iYbq)FM0XyNT_=>`nE
zCIsI9Lstr2Dr%g}FU7?5BdQDrFWAy{S@TkHBr4Y?2;pFL^!81bl{GzNyGO$mqkQ+7
zr9J-@Ni^u+%c<HvX|M2;LEBADdKeEJW5G5lzU22)!KY$Ck+b@cTOXa+VEl#gC7|-d
z+s{r*zHZVAM9PMf?|`nEggLU%w6KC8o;^wA@tn?XyR+Tg-J|;~zO8NIZj7O9JC^Z5
zKue0kE9#Tovms0v=?uW4FQY~MmD=)TE(_YXyQPjFA56%SgdH*YMds5gf{y1@lQYpu
z3}I1f&_6&{>5oxMJ3szf$>BBKiPX}9)0n%Ws($1f_ALdR=t-~>j%FsQYztAofg&Ob
z*t60@+b}EM2X!^A1x=dUyE(g$_@NmpNoy-~jIpyX&cR~()8*8FmvL8R_E(IR+0593
zCQsnvpU{>GfUbXVvEPY+N9mpxD+V)O+tljU1D2m89uAg$n9WWtbp3YmPwn`r0VFvO
zsrXk*ah4)ed+T@a3dzF6jv|t*RL)g}J0au#{uY@FHDKjtnV`rt)|)DvenGxk1)s5v
zsFfcjrlw3w3?pKUcD876EbbUn_|S>u({cj4CHHD%80V%yNO)qGkZkR&l*o(m(<lg#
zsgqixeIk_rR?>mr!&m6qW1c#BWR{A~ZIlY}oOq9J(YlLl<`W(@3wUsF+Z{@cyV`3f
zAm|8%#-;I_`dcux*(B+~7+eXA0(+zU>`ZhjRR}4Bubo@t=sLZ*0m+Q5yTx@UTaSLE
zT~hs_v){{ene&2kbKCV;fw8u2*C+sOp^h)<+28DUu}>T~P?u)(YAIy;tYz;ClnO&6
zB?U;d?MQdWvrNXC_&w(_Q%Su>mSAfubY^DePm^cD-}k^?YME&BIkj+~LP84q$1!c6
z6ufy7J=ZsF{z{>lQTj|N6{pXU%FL>*S_pM#YCm{|domMXria;>jKQI8f%#(vt#ZEK
zXx&jB>6~J$MyE7Cp?CIZAyk~fcg`Oycv=DmdLTlO((lHD6HyHflNAyW^HQRT#1fp8
zY-8cfhjpm86g`{ylVdI76?u(^FMyS$+&><F<T7>CrHUrYR9DoAOG_m|Cpa%*618ob
z_n?1c48IVQO)wJeR99dIrtlN{utwrJwfJ9XGZ@1NqsCzE2(ToB@2hc3_dPJBsxZW~
zv~9LTt)t&`19T|$=T_u$ueAc~!|#@07qn(Iq6iunjmjqIzz_X-GYnNskvEYCCv`DK
z0#es@tNmdFEv$+Q4ig{qV_k;)0lKH3QUk3<6od;GjAe-d7+nbn=Az4PSWDX}&R~VA
zApbi89y)aq-5qZ2uY0p;OghVPg;POqm?IFCw4CIt>5tz7Z1(SM>kZk}H}A5oG1f(d
zq9k5Zvl&;8>%-7{6dhj)ZJ{oHm4IC3^6@r?yj!*+i#dFr`q)OpW)_CAAL_1{<__?c
zVAR421Hj3Du9p-}#34%vDKhwQ1d_yK+d9G*^E~QcU{{>S%g0<S+WH8T&(GHXg}jD}
zijjLL#17)?7llFYgjEZRgf0k3KHwFSoI5j$7BJ5)Ci5#mB-B8**i2qwske}<kC-<+
zsf@M&Ef5u)PDPwH7CwiezO?j|UE(S1W%mML0lCV8>w54{jX*61>yUg)d)Ff_75}y(
zQGB`~af@q}gn-jG#WN>=wVo=u@MbbNp2!HH=KAX3$taqx0dq1}w0PG=TQNTM%gd8i
z&a`oQp6^kA6^c{)T$Y`fgqDAa8HzA6m`e~+ub2F0;S8>3a{kWl(%p_EmJx*R1D%}Q
zC#k>Fd4jl>m(Zlrmm8Ln_q5812)4`g6O?-V;^}uX+5V51Lur?`|BII0`?(iG71<JK
z^sKIKm>Yj82ftYSkg3Zf6sEAmih)a*bT1UXf(aGMyY=)G3i0~Gh8sclv|PJ$ep*OU
zpRcH(SPD3_EJLVFR1tF{+NNX;z&(=B?7x7{re=x4cRiqRsHB*^gs}>RY8_E@rxrQ^
zf1kUk27{{?QImQ|2q-<nlK)<^puEpLs)C|=WrR=SbNYzW6?hR9KLT9(;h_2)OqI+^
zzPA?I@!UFzux;Ar*wVxBV&yD(Vs?1Fx2Uu;-(IdQ-o$V*Rmn6*O#q~EYcAYQ;C|fI
z$v1Z0c{?X(xR7g&I#wz_qc8rqi?l~}Hp6C^#bCGuH_gOdOdk>^XEtoRSLan5dD@bt
zwQhFqX;sHP+nk-;kL;Nd^hEP!fiIEO!_!t<B1fhoJ~<|p<8$-48_yG8s%k-a<E*8_
z0PX8;<tca4bS5MdfH+S>oU7f-jZLcWOocN?yBOKzu?FwgB~T5DhG{afVNu?nbPKMV
zNf_C_l5`!das6q*7xeJtwrJN&WasD?)$b81ev7At<lk)i+@s@cM+%y2KyH}bBMTUw
zp20IpY^fO<>t%dMz335zrXvgHo_!N=AGhO=lL{Gk-5^RpC)y2(>Z!=>GZ;6>!`D$8
zy^WO^$gF9|G!7(^iG351NEmrzh)MjMg<4Fw7~E99eZIRCocKDzmOtw9!2D1=3g0+3
z!3fI#aka9(R6E?^o`Naz3Mq}%KJJPeefIo%wbPIRJfCe(3U0FJgMvQ$qW#fTVP>Yu
z*@wi*CcrG$Cc^F>4F>G(yLQOZquV<=)awadVQ9h9GOCQcbClbRwm?}4hY*(*YVeTz
zPF$VcTa1iU;L=|^+VnpgnbUY6r4g5&<YP>b=i)q=xJMt|u;^@}Wx@3Q<XF!7m-UR^
zhB>G5Lw-~&ce$6Lt+SQx^CQB*@+HaOGfn`=E=|lREy)N{;l9Y^uA;~iuA0#YYYw*I
zRlSs6-R|`_`n-HThFo8bdrD%NxNg51f@-~t-mI05uNaZz%lMcTeKCD9%d_1Rt0nH`
z9fZB0;{(YAgFDAl<SR<PFYWG@V<k>rg=%3$P+hX6<Bo6OM$TC}{zlJIRm*iU)c}SW
zj(u`nX4(YfH;Pohry74hb_rrQCm&*~U4Pr3H@ButP*J_S*)oyi6drY4Q4%jIS#CF6
za`XIO>l7G4<FX_n$vOGO!&^vO2;oBcV<{(gA&kFVhE%}i?K=4NiQ=9a5@0?LQjR%>
zz#HGk;P(F8V|9H|qi<GN^5R1)0fdjTJtS%DMa))F8IQ;d?Co>o{{rV^iwI<lYjH}2
zG!ystENzV&ZBfNk?o-UMO{$Iam$Xb~UTSZf+TYmRpH0xBKsiomXENF~$r@_P&x0bd
zOR4j?zHcL=g2oy~p@X2ezo%^4$u#{6a_+?PHr}nTxa?fAg`kjlCV;!Da;vujCL(cV
z)gV8rQvP1<`HKTZ4M$g`>ZwUyR9^J(%qpH(VO0nUip&pwY}%u%i6?x!v3SG%twg45
zN-0&z4kGqXy3@jR!t1oXy#(i$KDxL_>PvBqc9c4suXpsx5&sy-mTf^MHs-cH<lN91
zJv-YZkce%4udwI94<KA5ld2OgS=nGH$JTY!0rrLp@MZ>%Mn@|cLa>j*a!X`kr<v{T
zdNr%EutywKJ>RZ`82?|_1nqiwk0()e)KYuX(p0x)YFj_6u}jkN$PotGFYT_CgvBu%
z!BQxCBQoE!qg8Bht@Xk7`#)b5e)JjWL!VyrguAiwF#cE$5dAFNZ_jpk26K*+NGPpY
z&0ET^J$vmmYMCCLN8Wf9>^V?>>R~=-_1eM4E2D$?RB%3XwofnhBh)M4K*Gu@fZ8;3
zgJS~o*xB33o3@ZHjd-LxN2k--O--)B+H7xY^2BqHB*ZU!dF6eHgyP@T9U4n?u;O(>
zUhpzsswol}py{xP$tC1m6v)v#I!)T}RQ6jh?0kPN235%zzZxABeh;RoqFiek{N`pI
zV`R=W>K;$Y%N}!x<b27S)31mxVZD}X8$?D0+8C)8&Rj^R`&LLVl?auD9IVdJOUi82
z7}nX&ThB$7PHfDx)EDM)#7_lghf`EjosXnf@B+=JUTMJx^FQgqOtP)~pkAs-C*z(2
zsq7Ig8>DDIM`B4lv}LLyf(NZs5Bt6oB@&7jy0Y;6KEW*sCGD0>+@M|d2WO~Cs!@7Y
zdZD6&)>*Q9T7x+TPp{JW`jnn!Y4cxF=pM!BT%MCtn*iOYoCxA3)k69Pt&fdavwgce
zz$?wB>g|~uqM0Xlj`Nz}YvR`_l(^~BhFFK_sKpx+^^XJYw3lXpR)kU#N#tgN&|e=)
zci!F&>|X1Xfe+CF`qIrRGt+j-kS-2=DML~yIisjUhiwxz@L;&r><xFMYB?oo=f5@>
zr&V^wrTsC%+T!ICst`&CLKPnWvP1{UG_q-7M!hxKo>)nUI7-8F`diMXb9KMLxS$eJ
zWRgb;A|U}Ty%o@Q@z67uk0G9=!plhIPOK?fdMO`Dp-`Vfa08X(y8kY3I(Dtgq;sMM
z_y?;+AYx}EO++%Dy-R0Qe8yr&H<Svp+N+$xuVQ$1T`(4CEJei&T4jnw0fBKZ36<J9
z;fB=^w4e;GN>|bIs4FHpR_oQnYUkSLSI#vQHO>6DBnv@Lw#YuLg{td2F75Tx^Zwc@
zL-|Ubv*>@~J^xHa`165yuNQOe=<8H5$Y+E)a(mlz&yKc!-Ki<A56wbBM|`YfHZ&J^
zHN8pCQk)MWAx?e)GYf75w#Rl|wPkVo@7W_$13PhsRm_LAHqCN}sWw@Z85y5ji4Iuh
zS*GPS^epmsuDY^5#rVuou;8<r3SC79|1$X}<xUvE*?e17@ZBm}Zduv6v2SNFL2v4k
z{OgU;3|A4+lu`zj5t1HN25C@UhC#ILu`jH9JGZyr->C{XG6e<72;-UxY+dO6Qbxd~
zY+vDM)O);1H`R3dd>+M)6pL#bTa3}KqQw!<_IimsFHmEQg4M_2<>pyU`8zwS6K2@v
z*boVVRgyEBs31KnF*Nf*u{Ftq%1t|3c2D}g8b<qRVsVT|FuCh`Jl7==@Y^SnN-4KJ
zF|=c=-Mnwv0FjFiz4tbWjem%=Erz3>{IqocC~p4vUfpY~;hxd+tQhx>g+ls5zX5YS
z(l;A3Ej{{Mf3)D>@BU(6=z?9Xl-J;%tv%+q>7a*WQzB}&<VBo;{=p@gm4*x0_AXfn
z77r71lz_8xPdFJ26v?$Y8>Q7JV#w&FK1Ghpv~4C}-NnN`C)sehwA9Pef_jQpK`}>%
z-i^Hd%7YST+wTUK<Z;h!4S2PNh$rR_XqeiB1km(-k9=2oCgv{col$bxZP$JrR#2V2
zbV4pwLsH`F`)G{v6DM3>+bw^U`cKS_nukVm$1t@j6wBAezwIQ4N^YfdKvcakm&j!^
zQ6>VAWlm!(%(*~RfDrIwKv4%;g%iy#on%&>@IQR;dlQjgAne9&;Bqdc#hjicrr3xO
zHgL^k|6v09ql6(ViebX*K(3w3ou5SJRQI69_^M^t4H6@h6}n=BA)dH5*KA5{;3QF7
zznCt9Q#(uQ_ut>=aP(+NRg<g0w{*c;;0C&X;my<2HM)ObmRYX=!x_X`2QFyrV0u3?
z7U<ZUEjXCoTbPKc|H>=j%+$q;K2pw30g0N_zjra3T7t)6zN74_cYGhw;d75!oRe0=
zNX5rx`Z{VLhioEWW?8M)6V-)5*XHX=ww1cT-`|g%)7KYU+iXz}uA?#K6|kSc-ngf8
zpnJ@v*{C>=aP3kmMtLw5=ND3+FADK`ciI23M_GRiB<nTT6V~`b-#bl_gzL{S*?4G}
zdu^>81$@UNRoY8{2P1S=KngV?ur>wE-<JE$7;mu6yngO!)o1I|^^S(@4g8)>D?DvI
zp=Dbm7grU6vqnSX+|qUPas7bA3x;(U<4f+%qOP!x@q%z!&R5rmT6~yqqP9FY4uH3A
zwz2>A#l5KKhmPz!VwYY$<b7QPm;V*_-Ua9S+UR-uC}wpwx?cnW!3<N;C9X5O?+TvF
zS%%o~e#7>U+3g%qHxB^M4zX_Z1i6VK<Ew-zJUTd2(^M~KElr!V!2ve@D;?V+zr<Q~
zZkna_ppn)l5FT&H&K)OM-%PO>KFj>x%cjZLU@Vj48|31KK{c^MD_3{+!jr1;6>AdE
zn5{z$i*qa(2Hv<Hskb;UYsij_7(JQ@`5@!V;19X0Kj*~cQx-F(ygAsq&794Ocs*IM
zS&3rMZc@)$WlQS<-4e2$!i%afCe|T){_NO;jOXiL)2toEC~HoA2645P4wCXzDD)q*
z;B+RF`lh2zLb+M!MrDeNvih@|_Bexd>PU${#QM&)+`jsr(RUA&1dy-ceJ<_RcZnDz
zqw7XC?v3Mbmz-S+g=hn@hwwCmv*)=E_l$-(Ye#oynhx;u$yraPwrqF}H8R!8;b#A2
z+PX+KMOQ$K1c5wLV?ed2YwN2~hH{Xz8cp@_yO9TLvs+Pdx!?82cE=X?_9j~ls?=ec
z9pDe$4=498{Ct41s)YBAAM3O1m1x(Ypx?26^ffgd>~<V@hF7qDwCZUBL=GO|i&C*D
zt0~hK1~7SJw|wCJL<DF3&D_a$Y}Btk0hrh5%Do1GOeMlNt>g9`$Og-$iFihHX0Oi`
z7dU{3kHKO{o0G4<UY11^`Bd5pe$rLxOMPvM+M&Q(j2|gJe*9RZ!AwNL{;a}7`|*xA
z?-|aFNuP77Lgw(lr#KeR>cLa@Z=$i{APHMtKt!jplJU(PCg7p+R8{H(-U?l@(R-@e
zKeIQFy{S{$Og5i$kl}n_p4oPRZ%g7Ze0cq)(3XxoxXv@!Jyl-x2(Q82WR!J90}G=O
zr&uy015efOWR&t+A#aag^-hT8eh$7v<O+;reSaZ@g%kMS;5}BowAS+{Nn||B1qgXX
z)y)wpx2%*u*2`<}4-vg3Z6S&T9}{Ss*lp^QWB%UP_6>v}j(82S#t!D_w0$Of`NDsm
z*G@wqWzcQB<$vXi)7LKvB{YV=Y{^;a)2b#?Pob}f+${_^O9fK?f7`vXcFiRfq-8d@
z^s|&=wN{F7Z$=Y#{15U0j@rGld(`y{)`182LBF)xum6(T^!P))!O<pw-=rLsk}>uj
zAdcUH&#EYN<{W<?DN7_+6IcJ#5x^HF?jUU{;2^Kc&6vc+!oCHI`p<{Hsb024&6N!3
zY#*x@Ox2S5%orUc-wCL8co0i$(DzMMFzwQ~C0hG+Qt@;7WB$cfz*dq!_=X44CHTag
z<lKmNiz%cl*GX^i9-60uZ!*^o`3ov8Q7R7~V&^fzhsBO>{KqW<%<g+V6Rs|`8(tgP
z#X$UL3A)_aWyon|`AGD4*BL<#)Akec9z}RCNwU+m+416ngpY8p_9m}Tr(V-@>EL(z
z)c#sdFq3l$vkYv{=*KRXSbvSK)XgbhIz4wSbn94iN%d}Y<BUe;H10QbW&VjyOW7&<
zi{~ijn6xZo_}Uc$V$Sl~S>9+wu@(}YftrW3vtT}QY<fg|TtdKl9CNo{9M-tbK!Q$V
z{3-%F4rL?%)?@wC+2Zs;tB5nPM`?3T>RAfBJe$}~Ks5i@c7JEoOP!?RJ4HlM?TLR=
z(^tn0_%$fEC<N;w@L~Up!IU}&>?5jO!`xMV_&P&qo5A+mGrc3;E8^gw6A)bw>h+s5
zP#59O+MXDvtgVnuc(tC{VNep(Nao=~DsL2WLp;L?miGOu0?0=Tw0eAnv_-hpox`@e
zicmiQ_6be?w>0;bJ?=A7#w7A(@OuOHI$~}lDo5~mVSgGcsOed0T0@n#!-Y#-L_z#`
z@&z(E`{0V%r>XaCm~08zaAH_7c<l$2r|zfn#IO2uwWq#K{#qE@Ik~xtUs<3%am(x)
z-(58e-KbtDZv2)^Iga{4+Mk(;x%eZ1wfm170MV{~K?U9DQ_VK9AoAi(>V{7pQV8p?
z=F_t0vEmn^mi8EnDyTgc?2WrPP_OHMekk5>sXDFne1?Pe?_a0y=RDl)g<;Poy30SV
z9heL<({u35&ji`eUDY$@8$0KRshg0hheuau*}qJ5ZJHeA;*&Y+^GlU_xCdUm<MI6#
zfURZ4x3DQW4?pk-gQQwUNCjbTwr$R=tT?@+iq`5>Av{bLH(lJ}m(b26sYERXBW>)W
z|12dZJxjT<!4VpQMG2HUFYu~3JibhU%*t|m`?+l0&Z4H0-dm$qiqJeAHxjbK`C|I|
zh1lgv)l{e;V&YYLZ6o_b#jQY~9Rb`aHPF$A<%L~u<U^COkw!tUzD5~Vhd$CpW@zwT
z{edStD{0|4d#w{enOvXIDUO@-=2n95B^$dpjtsYakIy>m5v)BPN6g;V%x-8$QxM$u
zHvvzZi!e}M|26R|VQc+@YTPkbi&9)aCqm->Gki~!hqeLP)*63nXT#z)1XyRp5myma
z(gUyw@z?Xm0;#wvqSLf<PetwA*Vu){gnSImkrH1MV1C(}aemoJnms9hskO`w6_YC?
zH6}I+m#!xW<><WOCjM>(as|-w*o(|bmqS|J=WyLt!s3pm4V4nej~bPm2s)i(hK`vF
z67!y`{Gv`mIW*r|sAK}O?N^SKlE|3zF=EfO;<Pg?G9f!X=WrC$XQ=|TU5A6puVIJb
z1X3ZiUxR{BR5#P;Am9eH)ym6k^1QS#j+nfpOK5A1lg&xnCB(R5s&nqwbf%o|)x@--
zh<cT^gvnvyx0|i`>71+jH*kBckdX;?E)8y7d+TCxBGpUoJMgXIR279VJ|A?IH#0?4
zh@nC4e|r)Iy$2I$n@U%8S784h)74CmsVeK*yRcuH(oUFjD`0ZC7t^G<9K5X85Z)zR
zot~a*tGXTWcRux$HOV>q10Rw$MpWI_Z#lzqO!!#zRBwnCD{fG|N^*?(Kxpe0GQnGS
zktDn*ldc%RgUN}AKYQm|N4Hdab3<<SbSc%6*;+L%%^HpF&^AF2R~So27zmVx?YLd3
zL=A2@&3{bvDt1)Ka96JXtO)PD<V5;UKe$p?pgW=r3@>kBtJov6>Y6S=zy3CiM18&t
z^)>UZ3ALlXbb<86SIxdRV6E-dQA&-^vBNxtPgVkDWI~M6TgIDH$0Xw=;PU*i5}8nU
z-vJ5R#cz>Ylj-*)lpaIl(q*93?-$RO!b+f4(jQ6<fb;cZ(F^^^m$1iwvyDI<Rh@1L
zkEXL($o{CMbdP^Wuimrq7ua!OM;MF^)!Yy4lE>CzJLTt&`?yFzEWEFf@Y(yTnp5CX
zYBex5=`utyhx$BL5%!i>^xcsQ9}v6S^=myQ4S2*&d2035P&*4b+uW0J71zxxkYaAD
z-35=U@a#Ac%3jx>nrxTd`v&zrk0pSA1huRL8j8YG(YiQJ=(*UtDQqr(-2<aM8e{K(
z2zp2soypDF^_f$$!|=u-bZiP=GU*&({f$FZ+wEzci#mPp4O&~DwwRF!WKhJ#V++<~
z8)@9u%!9o!`OGYd_s30>&%w%tP>j*h+8uSR>vQWdb7|!(h6%XXPIg|tnux^prxm02
zZ|8ndZ_&i=gCwht_rFDd;Hx1Cr6P;*_io(Zc(ka0wLZt_&2kamYmPNqqCmP0yt;D>
zf!B|t%Mgnf(Tz9#V#IAL@wI<l;PnK2@^VvbaDO}`ds&g+58JeR>Km)Yxn<60n3WET
z(Jx@*Nqym{e{J|Ai*tCHTD1n%*_k@_pJj@qo>`6-{YY4UYlVX4w$E*D_=)Y&jexze
zV0L)dD<<N)De=JgQG3bm%qsAAe}>zx8H>I+V6Q;JJ?iUlyr(l%o>slQ_+Mv~^ajSx
zmE!tDD%&D6w)C5Z@=}2NpAaU-SB%P~wh7kPox_afd0|jlAxTEE1xAJdQTtwR6Wr&i
zl{6n!LfwFqE8<W;qpxN+9QmmsF8ShpALjF4`4Q0-M4!Z(IFPdN17PX$S7NWFII;`v
zip!>9Dy)Tgy(k-{6{jjjuAE;^UtU&jX|T|>^t+r$!=G>XM)tjWjY+J4s+3H0tiuJW
zS_b*4fjO+dSy1(`#aluW6P1DdYNNw<x#?1xkGsbZyXxD>W^QcGXj`IiucJpC1_d!L
z?gaTqC$ZH@Nj)RL;yWl2T%DxpBZ@+bWD{=G<5#S}!3R6*t7Uv^7y5n9Y-7kRL{+@L
zkAEVKaBt0}D8-3mbg&cO)VmSHX)_U#kYlD7{u}hLMpSB(uBIoBe9vx}n@ZTz)j@6U
zM=o$zmVM`RUok(96dx9ui?<`)45;~Z$@wVG45v+3w*h!n-c{|0QMzl0EbdSBiYN$<
zqM>?yq-<H^(@s9Zn!yn~{_;Z@Sk}TsoLrl(f3B8#+tW)Mx7(*TH!I6<%R>Jn?<p%;
zoQrfdD|)_=UrPZIHngj2#~D&x_UJQ$&|XI7)ucaP7Z1ZEB`D+n3Jel@DzGQX>d-P!
zfV(Sz7Q|b{MAx>w)hoMkDb`Ktb@)WUElh)tgEE*)-jfYah)YO{MPF7qQm+bkP=S<6
zb!PW4X0O<&+PwU)goN-&&@uBzD-Q}!Z8wyadWBVjc_E5r6qq{Z_4d`IrP@pPakNC$
zPR_>4OFxq~CF)+7bHe2~DPhV=X>$B0Gi3mb`Nrt!no50nM_>I&4U3UW)v~ReO-|Z>
zC^>)7wHf=QcE7c1Dq(jum=m2(U%atBW?{>gVG<eiwRvsD;Hk;oXgs*72okRC9`0rs
zwXmJp8347^D0y^(cilH(|Ig}J^;AyvY&kWE{*{r&((`Dqzu(T%&F5M)1H1b>fNzCG
z=Vz<jP-q!J?S2*F^ALQAGZrn|7C|^StkO*}aYD_jmy7;rcVH1V)+|F*5NMLK$?HBl
zH4n`D7Kg;zookaXj(z1(r)Wz?%M<js^!XuQfEqm#Cd;F;KP88{F8W*l^LjFC$?5PR
z30w3D)q#qhf0-q>G(+C8-PUsfuFdMxXD+xl=hLEr4<%C~A-=Xgr^4Ns3+prCjD|Z8
zsXmGP&aFozt2w@I+%^<r?Z)S)iOu)ya!s3csuAx2=w-jBw#XsJXv6B%(u%QAS?5^@
zA9t~iYZyZ32IYnI^~DRdYFO{{<rR?7e)8Ms*Qc&@u(Vn>S-JDOdJ{l|3HHalQGWUX
z*DSXv>A>QkY>GMg5B=ct(SrjHW+U&uHR0#XDg?53A*jjM)h65>mqzfa*$-d^nk}2{
zat_wma&iUXf7Vx<3$7deVQYFO?63jnX;-?<Vy|VNEz7u9$*OMbIY?C3_UB=zFSpAd
z6iz%vw>v;G<xB5@Q9yfrjPvwlY|GdqOhR1TWpP`A4#9tQeTRg*qj6%-O05b}KrUR1
zeuQ(+3BDQIvXr@GXjLq?m|&z>)KJAjNBFKIKXg`ZTR&gSY_IA2RZ7tj)Zh=8cR@G)
zTTK<2w64u`UK)!jrLL8ZFG68dtUFp!@zwRgvl|zJ^=MTKz>Zt{XGI+uJ%539($Ew9
zh<Y8bw5XC58jmOwN8&!Db(jj<*<ijQ=u*eUCW1M*7%_qRMoCQiYS*#6JuO{(gR~;B
zEP1Ji(nBb;N=)*bYVV}DhJ8>!tpu%JC6<Y2H|Q8=RD?Z-`Fvr@#$eB^2)X$tFPnev
z{v+50xaj~-@i(IN^Vw4gJKVX_-#-bv>rHn&-^BIz)Z+6^^dqxAS}AS|@5UUSh_`=W
z`X9xbnA#hj*1+224*&ONo2Crq3TR*d9>)~1a%+rhI_bWs9o=dUS0G@M8N?6srUX5D
z>2o&SYdUthY0MjnI4vgT|LH+4YfYE!)ROoBMgV*ieX3=D{>X{guUA(bGPrjV8BWZp
zu44tUvMDZxRX7hvl#o;GRKaTZ+-F5X)(5n8<@c$fqnHL30pFvAd-xpcmSxD*5AC|@
zsfOoq&uC-jE8R}tO4^)T_yDcjDO`iX6Qgq^HE8M}@|7ytL!%LzQraL%+kMF1j1wIM
zLjT?`!}}1`<Ftns_a$Ysbmu@Ie^vMqZL{CH92a}i@W0w}*QbW{aE}m<PsRRA=s1NP
zyD`RBXPg@$xvr!DmhzIg3v)G5!n-UO8Fn1}Ufr8%k~`Sz*lQZwsXR6Dgo@{}{r=<P
z5U^8ypwFUEn9OvwQRG3PUjPL4nX?Y`3PneJSalofh`n5aJ2xD($kJ7usAihCF=Cn~
zjlt;)F~9RJ9^YEs(WZ|C#@Ky)oqEGEu}HUsLfnv_!)~!V_9H5tAQ&!JFrB_`lGUaM
z3TN=5$A~d4w6Yg#zfSAAgs;dN#phZWy_Rcg+IE$wU&-O@=B2(vNl!~ZCISnDS9B>y
z6p^#3c`yU)|5*qQXC!QmEnZ+o-d4uIh_#LHred6MuIj*|AH)y}wV@r+I*Pk>rj6wW
z34Qe9SL~`n*~Fiwa9LlEX%EPW)2E+iIr3?hT=zrCHibVsg06l%qRW|kHy(fS^_zDQ
zBKJ$j6%{@H{73Xb>tv`olL3BgwhC6cnok-TS@<I+t~Zgq#8EvY_~#^hCT9Q7Sa8q3
ze^zG~``DUOUx;-X4LD8vh?wY-7GIxjna%Epn<Z~N?%c)>4*bU&Fi~1vaLgt7wcKr3
z;3e#=5<L_24>s8gIa7w+PJA1z{j~PKoa(6)BnmYbxA*(^-~GK^IzW&8T@a<Wfj#r;
zWY&-E*zur^eXO$^N;Q|=P_;1p)%BCR($o3;^<3WyjxMmXKND*3>hKY@`qGu7%Iad%
zi0(Drocd5(0aY?I%s=D)TXaO@B`2$!{yaC@?%+(b`jWF@&_2$_?Q0z@f@8g|d@B>>
zEuh-TZzDN@A}-+u0Nvw|yEpJ?B}^say}Of7CEke*zN3OCqy)$sC~k%Gew&C4q}VQ1
z^QMiOd1D|&+f^lu6(<ixMfeP6vfQTyVHNSKKK^RK5L@Wjwoo+}Ki3^mxsZ#aQQO-l
z_NNQ>SDXK7h#U{{`UqTu@xh%yg>sRTZ@@ZqosX(;OeTN_UxVMTxgO$RJJ3+2%aLN-
zz<3Q*6m;<N!~1?b7Yd5iORrcj2|#DGsR{aaEcKQKbS`uLVMkQb)YI|3qp6(3Vv$N6
zs_e6|^rmXhd|$%?UpIECXvm)SX#-=QGee5ya%OiyX_S7Mo$!xp<ETnhQmlUd@kcM>
z^P0v;>wpY<X4UCm3{CF@PPFKzCbYUITeiUy>hFM}X?pTCvZkh{C_`bMxc$PZxPN%d
z)Dm(18b0aSAE|nYsrT;a7V*5EYp&vOl^HkagD!f;&i)nA2J~wJ)5-9M)!f9g<1Z7S
z$+9<h{%!_$#2>6qs}Jn%Y%c5;w6`r3OkI-#Qqi+V|D2ZKfm%_aBDUVerbx7Ss@4(V
zUdw-kJ$$<lE{(P5BEpoe<VHpu1CgDF`}3FT>vs<hh6jdz(R7^|y?i3mOWggb)bC&E
z_RuR})Ni{^ODNB=aNXf+{U?9r1!n$CQ->m(B_)o$xbP3sDp4{&$#;M{q;^>@zsp$@
zNGc%rBj}+~d`FIh8dyf8%;_XPqM%epz11_pZ~5lX+dlnb(yI8BQr(Fd*Y%Z<8l>8Q
z5(eAJBKY5|J@v2!-q+|ZqBUL?!ul#EKPQD7URTpknx#GyQWWxOD1Vu=+`8c=;O+mt
zbNG4a55D>fCGoq439(ba-d@!BD_0>PMQf){d+u8NV_>PUI>JPS=o=qrIN1Rl?5}2%
zgwWrP9{mQAi)}rs^$U`hRvV|k-S7Y}A}^-q-F$65l#$LYVSIsKzA;t8;y(TolG9JN
z(!O~w7c5U18PV%BJv-cjPBaM^7C9*6rEl(jIsS8u6}xR3hR=GNhb`q3YzzR1wVk{^
z$rqJHZyDv}p=aVsDlI^!Q1hpdx%Cl_n{q>6^r!opN4ELWT?Abu1hrGo1=?Zr!Hh1D
zt*DPOgks(UKWjUuV+I{&#TunF)Msr-ED^gbt%%4(aY4+@y5*K<$p|GaHOPapy>M=H
zTRzw+;3mhL?8z~s{r#o`U;$CB={PZ}+3~l|y}Bm=i%SDfHL`NKy`!<k+0AY~keF9L
z;$$l^)FNSbEjN*SW1eD<u8tS%N`)UybP9rKcOxl*EpFA^%%j~{Kq1oBv{x8<i=ERv
zsNH`?>xJ%JEzj_Q+1Ixhf;}fsPLpP$0n@p$g@wM3zrMH$cs@Xn{^MxEuM=t?n45z8
z6F-r@r&a%VP<iKI`-o=Fn9KY38COmrp2?G>Pc$`dMt*I)lI+D{cpui^=4eB?28?pC
zuM>z{<J;yVjeU!=5yWUfzhGBE<JtMkq+R&LP(MTeMMQ^(Cn}pvswq2<5&h}9-UI!P
zP2w5v6ut*KS^yAwtwEt_d%H8?EkP89DM-OGuqEdA>8XNrgQ7J<kaRegaj<-_aInt>
ziZmU>#px%zDjixSi40i%S^H!85_aaD{!`6#De9shn=5H!&EE<V7c>SlRH>B>Pi1|D
z;a85rF4dbSU%^QgAlnnAq{_IA@A?&AYvYQb43t?Zz-t;al1E>+=nNlV?(Xfx@ft6?
zlhAmwbV^3jJrw+1e>wd$9Q@YdEc1%!k6ut_mnrX8(U%MU_cvxn6irQ>U$aa<lG4;K
zB{_`J_bxS%$sy`Qm*HUL67tv4>kPzP?J=QW@{XUcQV8l3@_lOj=EfvYUBhII|9Q4*
zMtmG#IM=ki3naw00}ML6*NyG^zKgCcm(IENkDgHuvCX4V*jOGC_AuU)8DCm=-7iX<
zAOB<ByQ58x9e6l6==RM|{y0XW^Osb{$Tml6VSmJEDSg-QA>wAyJ%SdmiBGO(YQsIn
zPVK}@TdY%>?W7jOKyy=&uckI3I(m1uj}0K^LQm3btF_~!FF!tfL(wOY@KOELDUnms
zDb8FtjLSkx_3ITYA9L*GUoiI*3W|fQpS6Xgkeh9PVf1#U_SHhgD+cgeWL}V!l}og=
z4!mAB9YIfe^3b!w@=8zVW<O$S`&yFt4?jO8g?Gqj=R@l(-8TX%H<E@bVAC|Y{e58n
z@4oh2C(BdeZxhfC9Q+=s+FQ`{6j=qpo<U81Ki$jXUbcA#v#Wt<-|iYU7qgFl_I7s{
z#@Oxd$nYf0#hkCBoYnok_yKO*dJ7}J@_BE*x4_==z5aS`I-KWa4Z0WFw&C*nw(4Pn
z>9E*eyUW~oZe(a^Xq2Yy>NroOAaMinRS1E7@YQPnY_!ib@?5&_uc2#OT?HFZ(ZHVh
zl=<TKrb<y*wen_-{N0da_0aeTKP6i$oK1Y?5Av=ySS@qlSkAl8yeD}Yawb3X!l5%$
z2Ya3e1zbJV*qwsfK&}$g)Qb7uNN#<Vq^ZfumH$A?fe-PX;c?0F9^yUE!^|NK@oE90
zht5;3INDO<;wy9aD%?Kmz7i?l!uRlnH!pAPa(;7mSNGbtAVJ27?i-`KGyB{z>fg<#
zV1Q+k4%Td1IEi(j#ocE5R^ZL8;CDLtOz!QiY%>Qp@+Sj(7n0GG?xkEF`62EF{Eqvz
z9Z#G+s!Hk+BpQ=vFY><DS|bbg0Hu3wjc73*$el&P$tB?tC(pA@?1jF*MgNu3_S1T<
z*hVY-k-j4vho%c`ZHDT_Yb-m8wyqr|OL~9!>pK~9e2cGQMEv=P++w9JhFW55E~)Ux
zHdjgz+$;1Lq5vZ!=7hHq-7Z&9SHy&Skn(u?dKDBS{DNv-Y$IjGFmDe4lx`?NlTf2-
zKNLb#DzlldCgc`qyKdjoytUI1?eSqn^!5GO1@>6qW=u<?3I#o%!SY*xS}rvPT>e*-
z*Za&F-gknQ^L5a`KmP_qM{OQ<|JDf`Yt&wC+`O_}8R0#E-&=K>Nfc(g@WcK8O+s~t
zsrY4<w6ffS3etD?R(GNR{WNW;CK3-`m%<HbO}Uu-RD;rM;?qPoGVs*1*Y+z|5ah2m
zP1m_(1XBLqZPBX?rNhQ>xeVExvgW47QBYZn6qK%k$VdH{U?oCVV%qKRN^0=_RP`B1
zVnyIk``0?%>n{~~;asAslG{?<>xKbp7t&PLkk!>=<@FMOg97mLyUkgpAq~xVNn<ms
z-gD6L7Vwp*3U4N7Z*OmVJGf(+dEZ2lhZm$EIuSY$r_XUATVx^mGK2c!$b(R%s{_;`
zr$htOs>hMRnf%DDCy2fmtLOa$;tTp<MxZEag%27)u2o3s2p-XXQ~wM%#By1PkA4@?
z`d1KA8P(GXR4|Iy?_Ni}70e{rs@grROo6|D0u>XR4hRl?m?+$YFb4~s5cRA=eu*(+
zZS2v~Aks^7(~RrF|7or3b|<LzZPraxIqDx+`uPul4>gUjDfA~I!8ws%|3u!T*dfX^
z(%?U8jnvku2BL}|pU5c!9b--DJLA6ne24uVR7@@a`McZ8a$J$h!TQmdxY%eeIN-N2
z5;^b^aS8+q|4Oj*wjs{TrIT0ZaaDtVt=|aLD{H9!aO~1%g%_mzrgf?aJ^^|0C7l+L
z#HsZk*C?&th@4lF*&LR+@0=y59B@@2hBn#W+U~B&=F=D--i&ELaI}Bw{eX~oKoUA`
zSPMX{YueV;M?q-aNp?<z2M<q-Ii6X$jfz~dg^Fs!%X&x**I<$SAgT}^GT)~lnrZpS
z8JpP5n9B=tFd^AJg?D!1)iRd0cl#Du$gG{uqb%3rxW27+hN23Rx3Hf-lfi2a`R}f8
z_Y5LkddN%Ka67pc?0a%3-T!9C{(3<N@OQDggZs@1R5T5bU1ENptJ0mz3zrS*8|Src
zIs<uFIeQtIUn>-!W^4+PdKx5v&k2mBd1z19xx3hV-?MT(pC0|IWz0@O)H^VS=EYCV
zac`V#>SJxLMiHZb(c9B!q;?GjY;vo2Ptk2}9$y^A(K_@q^@l(b_>5Z5W(+X)U%`(z
zJk4%Zt8rOQR!CE;9^HrCIt4=vI8;>sE*haID4Ib^l2If%3xH}HtzSQPePo3UfK3Of
z^CUWDIYq=q$K&EJZtX=)8n7B87JFe9;)!j+o(F3Yr0l!wBVv(PPm5ro(5_CmGkO%!
zcIY7v*WI@YI=1i(EQ)RBMt9=Og{{2{m!obemfXsyZsj}tjr&*NRjs&*+D+enUt?Y|
zGd?jY7=7ic!vpp4N8|@z!>7s(pg&FU;<Dk^SGI+Rhc3F){sob65`Qg876?<8OWSeS
zhTT7aua!K^PFBJ1f|#x?Z_Ul~jyw_|n-ko{CNq`CJfb?HTD9O50Azx7B3WNddkH_g
z6B$c$)2!T{ngpf_m?zZ*si>*|)X|l^@JEQW=OAA9+#5ql=3pfyf2^?3wPx<Er3VhN
z$7ahr+|HhAu35XjFdk+25gP2-`R=~(sIPtp7kB?Yz+War>4<3Jncdc@sSco(o5zVA
z`%~>Xc`NgKB<y2O+gIL`fc3`vKpM(n2yDuea0YajB&Z{3piy$#G+9gex%t3WqPfbO
zjkLYbFWbgF&gNV}#wASlOKJz^^y#NY*|a>?$y<Nw_U{5l{ki9-I9Gd^Nw1xZeN2`+
zjbzXzCD`b|GoRr1Z2cH&g5e~^JLk4e5-Fr`{)`4pZTl~1WDs?9fwSINZ}W_j(b4(7
zrtFZ(*yh`EKBl+MO!$u0`O_0lGWQ}|QqDuK9zQmo&Y#$+k>rlJ^t07M&$XT9hSP91
z&{X~Zar+6-sT65Dz5;r<(USQ*ik_6$*X;%%y-LnVrZH+SLu39Gv?p+3Ex5|f5(ddQ
z2`BtM*<9N9Efmzs3c`-MC`#VBd<&NUva%jl2YL}bqLc2G{m2&WoRy(_K_8-aS8?#l
z4obl9SkGH%pM&?OYsGxPPs|C`0-Vuh_3IQNu_NKJ(a~{nX&mm-Z)3!;ybESc(7Dh|
zg?cUL>=^GUfU{j|1YZBVd014P{y%NFFNjhaFq)-o@U)=UDauJH#0yLKYnyuuYrgVl
zwzw=ZFhdpsdf{HhUW{xzzsL-2Rz4_fI>-Aiaiu{6WRaDugIyGx7!OgVZ^i^tlWYLQ
zZ|5loaA(fgW*CjMD>KyX^ZoNG$E9{|{Q}4Y<qF_jmm_+bAHFKGKg2(61k%AsyB6OK
zKgm-w{`9%~If>3D|M1HC^|XvnPxO1neaPA9?|lZ(WiVIy>$gz8Cm)^*TnuaTXzDB8
zuG~y$8SmCk3zTb@suQaNL!TX1$O!51kUuGX<{u5Z%n;TXguy*0eaPL<n{j&0^c{_g
zZQBGYd6X_W6wP5SK0e8e*7!X>vz@*<M*kZ(dhYg-jZ&8;ZTKNWLpdgAr2wrC1WAv%
z#M0d5xiT2zZkP27Y3(<0i$aKrMU8MlID@Yj)TDLpMpNOWVw}PSF>&^lXI^~p6jo)w
zq8Lv`qLjt06Run0>3n*(@xm*JOERG?ude|16OD`92GMS~k>T>4J2+U=!at(O*9{<w
zs@JXj4>9gBCh70@IxZYGKOON<NkOChEI;+h128QNG%ay($jUgRi8f~_AQ5r>Ot_dR
zWWcI??2su&ZX_?|wA5i~%#Xw1W4S`!E?=&NT`l{4AIF;+)M&I#ER{hRoBGLbN`!Mo
z*Jf|Dlfncr`*licw?-Q)abj3LFvz%<_2FI)on+Sb0dv}~PblE#bl)SAs8=#Zs?CA5
zGb60%gH^*%X=A5<AX^%_WeT=gvvmI{*e<9`z4fnvxm87v9Vs%<S};W_Ltan{EXrfx
zRdo>Nl6v0DRK@t^lj&=K!+Y`XtWn^}1p}6p)H}RYmqyz>7q+=;C?Z-I>^gQn)obRv
zcs4>Aijh%(i;CqqJWK*_e(e*ghSXZ(Dsc;#a&+-kWE6pr*cU+5?oI>;E`Dyx8GDpD
z9<Y4@hoR|zv~xLJuJrQZbrk_ZNK!ft{`~3Q<mTq$EEmuDr%w~u++6rKz&Wkx=5eSA
z-RY)R>Xxt3+|M>e6&&7GrNTdG5={n^Zeb5DJ>xewRpn&0XhU?BUp6q#T-8DkjtDB&
zFQsjkzxL<J|L;^R?@3>+D|rT8AZOAe;ieGB>_4hG&nJ101Cg&b|0}GYam1X>S=?Ow
zCjos|H?V0TFl}d~K;RwJ<UgLdQ0#T7_dH+?jn5h^XJd4;aaWhHCH-U&;^!4<y=7`x
zNtuoazlKqI=+ye&^ub@lgo6vREzd4RWyBhv?OCajpw}|QQX96om9*cgc~r1aQ)>V8
z($d2EesnCY6=)rA?AsdIj<;sw@5v|=e}4EVbZ6xZL@X&$97Rehp4U<T1k<L?)LDUS
z2U!?C=4X{Ys};GpJ_*A!cLi&j626`to8=y`4j_(OXE#kvO*n{=={Kl&Rcg86$$2Fu
zKH_-p<p+-}KR$licNi{WkO*FV7=y=OuU68AVg0axlFdX5jJKU76aK{lBeyWg4FiaU
zYWeD>6M8&v>748DLNYR(U+W%)>v_VZEJy?V>QK|mF)=DOYi<j)WhVU+QY3opEJLU(
zF^oa4l{;@X^45tkE%MZRdg+Dr)Z}D47ks_7RTn5E0^#AK-kw>gi?FOYo!IhiI|zH4
z{}aG;G4-x;U}{wuokHPwK_I;o%)QkporTTW#l;0rG|Ke^edzs_XZ~2b6n=1sUD)fC
z$dgY{{Mnl)CnrQRQw4A2XJEmSg66Ir8DU<T-><6fUyCOKKZ~JuV((m@-b_l0nr};u
z9JbVl56#wX_RUHaot@#SVVzq~8Xqmr1_9PiPSnX3@*`iblK$B_*N(#TpC!rOeF38H
zXw8RIZ<Ugnv|H1+7xy+}cGvzq5L}l+JoS|?q)K`=24+7<j`6y#1o=Q*;fFMhRpL%R
zz5HjkZ(-_SwdsHT;h7z7CvATrp=14wp7+&5=}tjMmI|hfmUGxmsf`0f@Xf~Y2|!1d
zIbq;36w<enRF8Th0;inComWw~X^$jWX(@RlGfUWG|L}AYG?$8}e~fT)1wS$}Y0E98
z=a+SFE1UK^>BibdQxwXR;fk0O7|mnw{EIZz7Zux*p)RIx>x8G|yVMMJH)Gm7g4sDa
z^qJapUvf&%W?1E&U|MbO$^{lL09G+vc=f`sRggS9_;g;D|N2`#B@3RES@jw1!}>!*
zjg=6Hyb1K~<sPk_y`h9%H|_)8K1EM(ivH6%p`#kETs7bKe>7ctJX8Pw-&l;4jfirY
zk&yeH+|4D8T+98EOSufW<nBxj6S*Z!Zsk_)cex9p4239jNtnyr%7=yU+xL&(fBW-1
z9y{;%dA*+34e_H}ySu<yiU)a+x+BI63SfA@6snODMLX@Jt>9LvvpO=1R-B>ykl>y=
z)!nVs$tiqik^`YG6I%#)3LJr;-l>!G$B*ZlZ(nFgt@}f~u#dTIVAu7~@p!-4A%lZ$
zHZlin)@?uE6-g6qX~o_Vg$aP@KQ$O$aWAwV^9Mmg>?%6spJazAicT}Wz3&8S-L#2|
zFbl3Qy&)l|oE#Hat%qtMNyw?^5)#i%MGU%=lbf->5B>yn&NZ7>(KdNNgAcET(|Ip0
zjg;G*{B$pSOl4R*<tH9ZB*&n#C<;ipVPjMDt)EeUX(;L7Dpe)FPB=S_aUl12uTFE8
zJWac<G|$F>LmGmbk_BC&L%`df`%Ryhk!ZQO{D`HYX(u8tOJ#{p9WaZud(+=|_~2YA
z@^s7TjB3V|K)ka0@*U_1)YDtlLfq`E7_z<5yQ)mKA9FS6*$RM@=j{|gj75xLC_Tcf
zhwb)I8=DIJMoQ*~CQeWx6c$D&{Oq9gh8l6I3F9|S%AzEWybN@~KDT+CQTjag2InfD
z8Ja=ILr3JRWTV1Y;Dtk_adpcuiFU}%grTpvuDUwgs;Y)=KRWx{0nCz}7M9-xV`b?o
zI1`kj-?PBzV8}7s=DMBW!H;+^Baf<OsrsR<m6erY(lBNCWbgFww>DpS`1&h+>Q9XW
z8$qa+ch#&i*-`jO+amif_x6ly!-7`Uda~wv$cfDLuy4BZaLKN~a(OdYuB)}JyQRzv
zu9m9A-JCtmvlGqjv9gGV-hMTHS;Gfe-T%gqn|F@)lr^X+S#Ou=ZUB)3$Aa_xr%_ZH
zb+wHb{N9pst@B(r#;vM13y%BGh^_*Ar@zD69qlTVYNHeU!|(q2xA8LH>}?`0sD{fE
z6|QgJvh}ig>Ia5Vy0B3PVxUv6S81rN!^x(8()x%^E+G}v6tb~7cSQkS*HrF6L?lky
zpJ_ojps@)suCc35o`9%lCgEE1-O*nLDn0L{r{=AeJNRYNBOnlA4<S0vjt2uo+ERef
zv6`#w)qUU^pvqQX>va>Z<o>pt%3Sf6rupf+fS{P6u*O%@P)}Y73JaL#<4nw&RtWps
zFKm)%`0~#2l1t#g6__>?3%JYY0z5pb_4Lm$?W)M?QsC49Bz3&rxjo>lOZ~p;V05iz
zwg3lulzn0y(D9$K?cRUPkvnAWRGY)gCP_oZpLN`?<zn?Xq<6Cei85*p%NsNPl?vWt
z^D~_5d@jCFA=Hp%+Ie{CP*~DQU22^|XQ%yO_VCEaYQ4%p_FmbGCubumYCVtCR;?Y~
z_;A7jbIsMyvhI-S7-4CzHUxXgIj(>0k+xrLO53tL$aIPTf9zzx+Q@)y^QOV0tNERw
zgv>%z|E40?;$?Sa*vZlU)*fkamhnEWMu81)7K{6z$&i1nlc`BxT;~(62i+9@94}ig
zC=7olCi9KjeLAl7StPZlKta{tvA?_oa!n71s|&CYh<r@1y~GK3HBmhnfAIMfWtg<N
zu9;z_Ujjgc3R%uvVgwMDsr9Z$0dx@Mh*akKyQdC9&p`a%+3G1p2TS=0{iGE(cyL{(
zX8udHxStKbDzF`B90YZ)eCV7f1EJ|R9)!NA<KmHqGZ$2s(F<bf)?7M{kJ|jL{3m-o
znhub$z?a;f@ZfY!YAR1zTlKBel>16QyA#6U3xta~BLi9+=U)=a!ICDI_P9O8pUZ;#
zp6xFC)%-pXdy|#4vCz~+ZsZ_{w5joaVOM_*9osLEre3)*ZBHv1Zn+`+<bkys{~aL|
zJg32_deMcPYS{m+`)NR&&&UOG(5r7Y`5&9-Gj_$Foz=wh={y$#HF;eZNXYf&;|fmE
z9RPl9zWR~<G%b8b+tRzb^=_x@N}+c-R(>$sH8MJS^yF=CjDNwlcXE;zglK6srH^7-
z=2Lv&4)qpS{<IxApEDbkwEun~<v4IlkT}b!|N14b4D@%vx}P7-7Dj3om7|M!O9$8U
zCv6;Sv+VEhy>Jsnb8UMk8S%62O$D)c`T&=;i0&F$yool}86{xz!}GU_IUDk$bC9T>
z@5lq!`t~+K*Bu8OM#4ky*Hk!)sE~4^C2m+l2a;<-enaw&F~Z-ixszlkLaus0U)|$!
z9o6)H!wBl3^AFcY6xRB<+t~&ZJ-a|ys320#ax?)sBBD)bFPGK+tDN5%-9P|<-MR|>
zH%(jNoz+cb9_nh8#Sh!|uC7bRL3-kWN|!lZ7MmqP^nRgfs9v8j;dKadrg<x1U3`;|
zwhQNyB{>#u@7iA0^GV}Hz{NYtb>`<C&;DddEp&GPAu0dKF!W>umiAd|94!5QwDX)G
zC2m`0H`mHu>uXI;<0iRNK0^WL=BeD_OsU&KtI!@i3hS?VzzbI|svMs!&np-ITH*ZQ
z`U%IQ>XoIpCcAXe<|n?RAl7<WxI5k6kJBOXvRiTQgtvMfG9AwtzpJtrx&>=dcMa{f
z^M}w#t&Hnwys;~74!&8#wsP?|>ZIuzKyupNVl(AYdbWd0gyZ~0F>wItTmANnHi`)$
zi1|>zi(C91N?x6t*PPZ^w03?lDkS{^A5NLZ2==IKr>q8!GQn(*htCcutHrNYqdPjg
zo*7X#7Loss4vvnF!aMER1mHy`SuYG2&30Uo=WNItzgqmpO564?Ccr(TBQ>^T5S<~I
z-c&O7J`KrWV8Oq+3m{BAB<+s(2p6Aou+K#%75#nNLmX3Q7xSBYh4s(hRKC><HK-(2
z#YP0%T20X-XVnCx(^GRAf8OW+F8ct<ajoY{o#4|1%kS#q#9phHXIi!4`)k&pW_!)8
z#DvuDCCO~_^P@x8PsZC3?kKJZh>&F+hVshm>M-%ZRZ`gVIq)(&0sLhajE{x*Zw7X2
zYA$FdM}^3VGx=@0RQTMQ_C6byqG3DJ-LsCqA3r4Hw0QVZWOk0VKhBrQS;|fYO;?4?
zx4k`>Z<(jm)7%BxX>IGJn2E>uXNo%c6D(|UtqPBn8pINUl*TfqS}08zlJ|G(K%$w^
zW%&NQ_kIo**xvT-iX6R8N5eU2BKUimQ?_lM6B{?#;O^1c>KdAfQ&7Oe`58g4L?PG?
z=j=^0wf^-V13{{%{HUvYE*`X|>9!_~6m1-Iv6(g=sz;5&vnsr9w4{oG3(KFC8)Scz
zVgx-Z=<Aator4qmkP@`hXqd__z<6jsc<R>M1bE!#xA2Z4`O+fYLTC9S9@T<qT^#gS
zLWG5<MBni~U2k7hr2<(#@N1knXI+X6C{|d=(YOBs?8|g`Wc2Bt=4dC8rkWPyLcNuS
z>;IB32o!cBDWcMH)>LMB#uH5SEQR2D9}C7@0lvV=>3Eo6Ky|B!=cD0Sw)FEHgDiL;
zO=dOSi{rU{Ajj+ryk0EU$99sB#0#$AXF^s49U3SoaA@Ev2(SdJR*`#Ym#mUwt}qdi
z&kc$HmiAnE<>tW7-X}CZaQ2_k=QNFJe<v$$<P6=HIh#}>+hDyYoBjP$Cb3Ed`OnP)
zAEKx`J4;J}>QrS0$-doRZVWnX_x*sam4K*LN6mT7e438x(U`yclY*P8Kmd0;F<FLe
zD<ySLTU0BM!b`}q0Wp}V3!HsV`R3Yy2!S0QW^YkD-}$Pi_Nw0nF&Bw6B_|Bg;wFi;
zRz;?{hT<;EL0q)6j}$0QviG*tPWVj3L(e_E4uP=g=wqLp-b86xZQ}0z1InYjc59os
z5`gUYy*^bd9=%VDG^MYgGS*FvWrn_^UfS9SGJiHR8_MCT6dK=<tmyiqAgGHa80uvl
zMiOCyy;C>g?XidU+rz!|8xQ_{4NoTv%)nICx@jxj@jeYrk0$MmStm0ss+5knvJMR)
z)OEAVF|8>IpCCHe1iLK2{o5D>1ouq~ws<?NO!sgiYdR~82mYad(Ib$VJsj%%z-+6Z
z4m6xs?)fKtLiw=9)db;{<xSy*4JZ%T?&oK#G*m#5mP@#2`rhbE!^k&^KcBZI;8U5V
zO&U$r?RILUq`Ec_BjZ|6!Wg5-Cx__7$_*p9P;FE4BkNE~RXvd7IuhJ{o<D2&Bbiyw
z>4uz~a&N)$MC1=!{2C`-z6_lH(3K4mKt1_BGL`+*3Q^~yzMWP_C&q0q!ytsV8vRd&
zQ7bdc=Lee^t-;5&QiZHaMl1i2@&1t;C*Ktr#{Lq*Jmi|JFra*(Qr1fU2x-$>$0A6`
zPPeU*GMTZy0|i>nWf*>MInUHKc*Q{8qZHTn$SBQr!1{c{K^5_dmf>Xhi>dO7OhSNr
z_J|4?OAzRlc&$@Ya)T`&B?97nH_0UImd{1NR~m~8u=T7L;M_FSm8{=om80-I>FMQp
zvw7CFpqz3Y7CvV%lMqYADCbR@Q0geB?k>j<!N7?DF(5`JQ!ljjmb%G0GoDQ#jcy1m
zZc>$<hSRf2l;{c1YaWO^J)ogS0Ve-BhY%YgH9+-5Xz*(h%c@r=Z+Va99F)jiDRumV
z6jGu{%RH(1HGYNF@0xi6f}`g6w>uNau-{N44mgXJ{*@(F)zo`A=&s}?2vUHcKGo=+
zEs*gNV7vu;f<XLq-=<*u?_YjUst0;4Oy}OZ)f*F=Zlv4?t+5n#N4Vu}i`k}n>py`y
zowxa%o<}o3$S}Bl9suND-8<)O_F=Z;_H^mEJYuA>whxtor)9+?4PpEF;T>YFS#<X=
z_hlH3UcN@t?#Zk^O7hId+Z5d_C}p+t<s5zkT))2jeX!>2-vI^qcW_EN&he!zKgeFu
z@Hu3LGBrUE3}pA@KytAql7|1B#M>$)FNxbHWpgS{U)dXvuxNFZ#61lBhSOM$;ZkJO
zmrpBYCT4!LXVCdk%?s~~B1pY5jd8Qm`c5tSPg*>0P>mfaK09bH)iOOWo*e=Tp*8_L
zGIEwB`V5@aN{^E3Fr?wF0qnZQ9qajm;;*aJlsnqv8+xYhl7074cYov~3L}!1v~a2@
zGMD3t0TiUN$`zQy4SyBNFzQ*JcdH_HkeH~|HsvPCqbeZx52}ATLvdI!^Ib1t(dC%s
zZZO`)nk$92MibQ&Ep-w;eo3mkivoPIx*SxIJbTuzNlUe^v3hi&Pp%P%>Uh{QOjEs(
z7xy^2e729-p?V}56CAz1*t#S5)Q+9=YpI_3vwj)<Y=`FPPM;Zhc97Yt`e&^#1kc`{
z;{OQ{$?L<Fj91Olw&3uXw=(p(RT6rK@#0dt*S?+iv=X-X5<RoAWX~Gc{talB{%5&7
z$6Gt|xT5>jYw1T&|7H{<$;=p}Xdw9%ikhvyd{?`_K43<+CyBT++Dt1ad>3whErEX2
z-EpJ3G(eXPs~?$n@ScYJ57wrlKlNxPhukq7LMO4G+G_BNeE0t3q`808`m`$tG9WuP
z+t^ff{ew|UkC`{w;>JkN#A6^R+Qm!#d^W<Bd`tyHr>`~}&kR)EwiqFz<9BzN7V#jn
zLOnyFcni_tA)|ZwoW5xdUQ=mNY(;NPDZNB=T0Rr@7E}wW!;M2A5a`%kP9dd=P#0!4
zGa>2pdXSs`sEYb%>>Vp3foTlOGqdYr#nwdz)xjR`hnJpP!$XeEfd9lMZ9fE><U|RZ
zTg@{3%zexwnA)gt7eiX-D?-boSX|6I35Q%`ehV7W$6M6ZRXO_cadG4<F}cnNr-NXW
zM@gL#BqGR0)AKeNkhwkDyWD+Rj&uhHQQb8LYh}v+l;{_-+}q-N+?}FuPhSMi4$1W;
z&KTIl)l09tCSd`$bk|MnP%pL?a~YYx1Tv&!q3$ZIhN=~~n22y|gLEddySq9&ySik)
zMsEftxACT^thV@>N9lTaUu@AaK3iry>pnD+4rD7AncyFcZ=O$Zlx&%il;%Y~i1-j>
z@MNkH^UC-6OUz%H|J5PKa>NLN)B0vd;BaW-vOPu|3Gno$nF$jr#N2SLV=}8f#43f}
z7hBJWisQYliu!tr3@Gw{jkQgm2QsjKXA3wJp>sZbeXHi?Z2Lolk&)YjOWAUT0_LJ_
zI!s|ajIv1f^E9Cc;w-KbI2&BB^IoJY6tBW;q~`uWGWkE3Ar4xXS7DxrhbcPWi)E+Y
zX2p2|cbgQsi$5Z32x|ss^2qvHMJqEMmOX{L-^Q$F>WCw)=?SvqSv3k#qf_COE%aC}
z!$+b1Hk5ug=}E)-+{E1g*+&#F)NvR6%POJOoW*7`C{V|8QRe#h%E5*IJg-4r-Iiq@
z3o-S3QBax94y}J4t3>h~*Qr^rW#XfgfarDrh*-$75OO#Fhh?4KQD;xQe*sDiYIhtQ
ze$=f=hg4g;F8D+<?8^i32G^0Oq-oktdU&*QbdZh>{GnE39D_YH$f-KbJBxh*1~tr1
z9Cp#IJu}qVSR<SHMJemnPY&It1Pvk~I`=(_Us6O+X0@=Ft8<)%L?)5mH&$1oy=(yZ
zMHqZ2lCQ!ZZ;JHAf$o-cMK@!m;m<NAMmjm~8ltpUEAGCLnVnO(G{z|%)GIEUU}L9?
z3LLXHPDJMQ7K_^k+I|!Rm9JL?Ic;5KqYhiy+HIG~2{;pvmo$<W|3s<fj{#>bK*ZU>
zaWoN%P9Z{Ynnn-fM>sTy1%Vg#fR{Z)uyf#cVB_#6Nx$EN_9?NyA?S58D)YMA#jE-i
z708&_E!V73klHF%yy4>a9Gicli?n7Pf_{RbvLM)n-+=DzW2!lyh<esY@SpT)ndFm0
z9OUI)*2hY69!gi2r0vF*Z2p4p%kC|Oxy9AXe@&`<GNH*7gT>KE;GG9s!1acgrJ$er
z$m|}%TE0R$XGHrI55v&bL$t~k)YKS#&0i$p*=br8J?@!;fL7CNf;4ECbs#)fiyI-J
zze%G8Avk&%owSu!4i>5>%Ps@1O7bOqjCw{?xJ(C~u=T#Mw^DqC+x7cNn6h)Xy`!UV
z-rsRs_D&k>sZePYAnXa`7>-4q?oCr6bIp9d<jdpHQJp&0pFWErD^o$-oMFKBa0m3K
znL-n)IH|Xd2rl&5<rD9<+#w}F?~8ey{D8Zdg&xhQ6m&C($Z1|UZ@t80h6HEC<?)p$
zWRw`_fn;%gg!9=r)8`uNW|yziyiZ)$^CZTJ5`FNVi4UJ2zyT(1Q%vDX!Jn15ex^pX
z@^2_RsY>x(G;`pr0WyvhzPU%DVLP5_=E_u6;ks!5;(EGkE(m?$#Q!kuo%ZUpBTRtg
zg;R?IBawo?^}&^&92`%NRwxV6VQSR<M^p>#+fP_}_F_V#+t|x-7une8{q-~o&ZowV
z8r3IB^Jkkt0f*KE*+9DNX%*{X2md?L@%{{KX;Tpk;VWN2>E&J$FjMBtK5?Dvl&`W-
z0Sanh{NbC=+cEU1^c|L7%5%2M9xqKR)u=ywPKyxY#MukP**v=+T?yF&bZpkBh#DUz
zL3_!obXZ0?IRy9;8H51!$=%L&^PGvbKF{xWoDW|FVEF~q{o|J7B^k9}drM0TrzcM5
z60p`koN3|4phoWz-@)HpgE;%N!Ft!~#};Vm#3;3_2shbbTa5v=t5V~ImA~+~Y+pWK
ze81b`AWuwc(nWzcI4FFO2Rc3z>vj_?T4s+GvKK3V{(Eq=RE^D7<3O-imL|ubX$6vT
zOt%25o>Vj5hLu*e0mLF{Yd}1<B0JE1m~KVMid-;p<Lp=pb#%ONCX?*C8UG_^SnY|S
zEJ$tok8O%n;%AHvw;^h1Q|7?>O9wp7@O*4(XVZLRg1n1%PitKJb*u<`5!73zAz{uz
zEDyHGP0r*VZ)|zoI)AWa?JB7RHtd;FiUHb2#Mtznv`*{yBa!qSr3tF$2;Ol49^QbF
z!w!DxF+Ir1$?>84gZXFL7P5kmrhbmwA%uROL?gN?6hYbK9!z^zo`#(Z=CXcFC2C`K
z5Q1Jla<*JV%F-)*!>=Tx2`UG=w(M!yv_Yl)?wL58F5mp84A|7tcIUW@WSiy!xTLNd
zq9cF7$}b>a7mmYYdS`Opg0$GK$%zQ&WxbK};~-=w^gO@%E<!;s%aISuomMflOu%dP
z-b&5Bcpd%X)3a+D5_mn;I%%XZwlO~@UxkzKv2Ki-a`;;%g#eCZo(s$W!=x~Uq+=|i
zmt+;rm};g36RKK&UjO#0KxOR==#G7$Grk<q{%?yPAn~yxW>fm~j{=6&ejkCZ&F2<5
zhB1FmJNN!BUj)o$e%dVe|9SV7jL53mr$afgKP$6%hRInrFw?EnOW2sC`u)e9H1Uh^
ztML4hKS4%_HE;LgcIteqf(KnHao7JzF_Aqvp#BOLWGTcRX;lX}3t<g52VHy(gF5FL
zn`}p?&&<Dlr}xjZPcY@OLR{>@BjBD!sBsd*TjNpdm<VyI*9eDBoT2G`JvX*~xeWU6
zrH3anv-f!)pGDFJgp<SXf?wxu&1jzf)*N<%5GY(OaiGdku`22`ewC}erlq9?^6zh`
z;Zm0<Oj{OGstaiSm32yF`}{c_{?d$1fY`q$#)&*SSUD@9h29FO)cbTs$pupJL&UcF
z_z6MY^|LcxCS@+(JkNmdKItFJKi|f+zxvU&<`Pj~(jp3G!Gg@rv2myIc^ubIwan-4
zY~!Wr&%;4hpBUIU;RNAKHvBXtIO#R+5L|u*NhnX}13Y12A?4+dS+CAS_(jQ!mQ3Ax
zmOENDn%g$s_&Ne|4whyJ?(w{vW{xsS+hu|~CuvH_WilZn*SM-8Aa&xyqQA7b-j$TS
zes71g-!@6iFxvk5ig56{%$-nb{#pLcp1j0fBOVG#k~!KcX;Mc@X6?!6k=@zgDU;7l
zL;uq_0HQF*r87*wwgU*67KBwhO2kIEO>D8hs6AP=;Ld~lcWIDGf*pmPUH1794-5Lk
z&m=xtUjJQhjQ8~F{>pE!s(YCM!1aFaFcZe#Pu?x_)$x4h5N?v@2Yt~{pYuIlTG~Hd
ztj{g=tj3q%hawrK&C9rnXmE<hwU$A+miS&j@Dc`{%L}wbWyrpMEy&n_e_K938CUMr
z{H;cG9xY_362328GO0Y7)2j7D>{{d-Lzl(*t2&niAfN0-b=(9zw8>8&tb--gCap&F
zaY$8Jy;9%zueedB5N2=@{W&AAH9`s>c>G}zuMO<=lxPOV+QQK<y!~kS_6{$CE8s62
z2s76K2_dhY&8fv11YXuM_=$ASL09b`7*Blmi^&12<(5qx22uT6bIJJcr0Sxv;eWf^
zE+i^tBJr3(P-I;mB&w|*U5M39dwQ|QS6Y9w{pFLUA=w9?pD-WViy^PmtN5A9A)xx%
z#l&FS(ze{yddq4$h%_DhzkGCR$46ZN-jkijU7XPiD=aT9-_GTk(t@(=5#{B2>?2W-
z?t%1M8ozq8Cw0in?`+;*614^+w`TavjKHk?eY4*j)Yq(5bahg_2Q;)U@W3h!Ieu*>
zCKM0hMEHuCTe42VJEIQ&Sha820pXE9nvLTHzPfxtWZDu`B<8@2<y!qu3|Rn><!u>`
z;^yr;PWfCeDr2&b5H8ZH^axd1X=dgGnBY_CdugpxPj%nv8pq%0|9-%^ZO30?QgW%<
z9amvTY1bhd2IqTMm0y9CWNaqN4h2o`JEQgSL-NMNOKXLmR-d+w=RO2Ma2doHfe+?t
zsoztO@$esWN45CT?8ea3y>Z|}-%Dwfh`qJZy9vuOnM&%s-A%lDFN@E!DE?zEr1|P0
zrb^dpP|M9jGB!aOF7t`mq_BzqcdNl?6OADJB4^YZSCm6dId3o5@3v~P3^D#{f@!va
zwQ`w~)tu$3dkl{QB<AUbh3P3<{&Fyuh*~CB?Kd~oQKROjNnMUg7(j3`n=6gJ;bvU%
zxk+=J%`+Y=JKP3Mee1u6nVYt2cWjvmv&np?&~86EEK9qK)pSm>gNro>Cgow$95}s;
zmddmv!kIR_JehwVwGbiM>Gfm`KPh36`X|9C-F)XnVP_p$TDn|Wi#V_2M&~=8`Sh?U
zaM1D#+pwvH9B-ckAi(k=K??^u&#vqI5N7TjvC|hXfS^-(v-&$Xk>wLkT<I#t%1wX`
zY92ek{&<h{ZR%Ci<I2qBkb+AEk;{%8iP_tbZv%MyuzXII;05Z`WY@aPsbcs_h@Vu}
zas2QrrQs9axCO`RJ1HrwJM)gCWlB5qNiyNy8uCE@biqv?048t;VAO)p`GaJi&)>hS
zS%80}Ar<IyjR*Y@+(#=A<e`JIKwm@cH~u4U2ejs{PHfUW)9_jLQDF=t82WwMdRF_|
zXmiuO{DY1Gb4=M!dV1*4R~s09#D`_em50k)7`|v;3?-w~7q568Jzh0ut5oeJ%J-&m
zA?&(L_g;SndJo+i3`75$s=PplK-^2z5_$Mzf==)ybiL2XvAJxptgPnF9j5~QX^i*S
znABoShPzpLiv^>2;YDKgTRs%jP&)=63nMgAjv{X#3<TJa-OJ0Eo_<h82yt%ILf9{f
zmJC@^>d|`2v(yVQUoL6whuc{cM>4-9rgLkyZy(12QK6t`(&s@rSNUYb6~{l#AG{V8
zC-C?8mX>P-+`c1;lAW}gmcCqEV)Cx62Op%|{xR+TWw-oM#u}t}f62eF)qh+Py*Tb0
zBg39K50KhE9*QvfG=f(ZQJx_f<G=yUIt*$N_lRz<>QvIHc9)%y31ZR3hUI%nY?EEM
zaT~7zRV<;{(qqG=h>BUDHXCqZLufzzD38~RrkJ>xzJKfe&=>W6x7n{ENA{k=`?-TC
zZw!eR|KQgWOv;(@Mk@Aj6>U}d9d|W-jk{Qe%>b#!(062(!a0{EK{XEkWTV#dl+yBx
zSPI5zx;d9T;W^*@o4ggecd&RGEgH-22CTzikFqW1=mc1L8vFZuK|fR5kX$*{F03dw
z239#;Y*2RD@~yewZz-UL_lmc77(%5>MW=7iAhO|~#LXT)13JuSb-6bjL0^kfzDDf0
zT-ip681@UnTmK6G)OB%jfxEz<4C{yI4vJ@G13<WttxkI8N9QUrMVLg2)%2?-;6#hz
zpG83^4}no{kt@W2T+!#RX&hWGn^&r#uQskRP!KCB+<Ug~q+!oDFG=+)G*NeUqS8E^
z60n_xjxg%!Rx{?fEqQ8fKe_jBQJkh%WSm<YFW?u~dE7kH_5EF@ks(yEt?Y$FNj)_7
zAzPOnIc0}$es(Tz?EJU!505FpmCpghv))|jjq_xxN4Dp@aKE8%ej0kQ&|``*T>6vR
zqh)kM6K26+rzFGW8P2@Sp7A4PLl8FpK_IdbTD8^wIPh|Me)08hX#IUCgLrVxLojx#
zIcmq*F&qA9$6+L16|7rm2m96BRFfh_tE!sVKWv+yI>B#Is)GBSU#kKf9#Ek{M27e|
z@s}akZ7%;L;gJz^Jf(}~BBhb_?vADF@=|3bqZXp+hG?B!FO@v<#9{vfhSArAmeQgY
z5+)djp1Q#3>?TQL2V&sHgG22)R_Zggo;ee_d1y%iB<QMclOM@(SP0p=5J-JZFY5j{
zQ|~_IDeShyMEx=fh$g*aWD_iR9aiTu%43TRn0w4wzJKRDBdfJv%RLa|?P^Hza`Z#T
zQm8Z8_69b^4DU>u!usFCi6DnNgnVp@lOO$;I4mpkqIJI6&O5?o=3vCW>XH^h(j4;X
zcj0Z*@h%zy4O9L`c>ngMjExBRJVGl&Y>m+8Q?=D%Ow(KdJEY(V#jePse}4kz>?nQp
z<YHZ3<wjQC4bIr}U}PaaF#~CeO@6P01)1`8u4L}9{+CDa+UXOA`3G%b{wqDMrc_l?
zs+Ueye8tXL1^RjR>0<?THpJ)D?@@_pp;o#^T{XN-sP_9fi#I0TOnB4SMG*)-t<g?L
zwEpL(Z_<Esjxg`c03IwP&S&VN8h4@R3)E1A6)AXIbV8o}HpRGohU>Lpl+e4i+imo6
zFU`wj;TA9JGdxLla##6E$v=mjVNY|vWWR%*TP!7mlYQVMGG)ZCVxP0r=G+6aDZ;t>
zeSQC!pk}AmkACyT;}GiV#C_L}jt;KBQwM(G>D@rpo3m3XS|_F-<<Lbbbfy`@KXymy
zU5fb1QV2gyzWZQZZYCl~Z){jpI}ijqEq{)Hg5sT7*9509_BH>VZk1TF`KYV)!kop<
z(c!`*_hk#VO;+RE_vATxOt+XlTx{Ih-5WfJveQfmA8TxT?5m63&wjx*><!<W8k)J?
z8w1qcQuxRbenaw7iBMK6Q1&X59;iVZD@Z4_v_=SesI(O=-4XL?c9JLvQAv2Z?d&ij
zAA2|x4~w}}CVl&IfeF&3ZvTX`x_=lJ72c+<?ig{k==pq5kkz-%!~p4+uS-CM`F{@Z
zlR<(d2SGfd6S~H7RbK?Af^Xjr%)2gntrPeQ`hefF+}_#Q=G&H}h75;kIAylC$9^Hk
zwF-NE>wO9$5N!Ge6vO>UUFBY?FTV95k<K}#>7$3za;!?&LY^olwx{tVf2+h=HLz$e
z8oah`ejDxr{uRYybrs%Z>A}=9HSzmuKT#GuIdveu)xG}uEbakc;iw$SYuD1i&+fUk
z0!HZ(?hQR0!3MEO+hv36arGjfK*?dkztq-7W*r2b*ZgTl)fVzo--n9Yxkh<J9#cDS
zU4zdOUuOTBHn0hvSFskJDp<WOYHqrnDW_c7vX$ke2}E_MXVo#%vxCoPePpDU2SR|h
zS&2P!tX5yVm@}-%U`D61#j(t+R17&7vWZA<Yqe9R)6&<*hD_VumIx%sc^RD`?g+>V
zzB%`ut0g#C&Vt^;4Yf#j?HhH2nXAW89F>&Bb?G8`s>Q>#Fk-RMgpMt|x4pqARx7T(
zU5=QTH?Y;#RY@9dZbpOswHT*?*H8SmcT$=Cb5|{Xy2;Eb`b2a)H0}64tEuH5ldmy^
z-qr{JtGv-Pokvf|&ULo})Wtk$^{A68FrxA|&F78@cz%tp0uxr#%(};Tk1Ng3SIfNf
zR%yLi3!1YL{lT48r<LW-kGPD$^sMswJUe3hZ_<#pnCYym2QlC$JYYMa_sD)(E8$A>
z(Y*kmtNmudswV?>QPrNKVJ;EH%{C4fdpa1Eb(M);Sk*@MX6W`~wqOt_!Y_hETC>N&
zBs!kwuUv(J1G@TNw^dQXJJ-2jGSf`FdFJYu1SczOTDF=sH65L@BADnEf%fC}<M`vf
zxs`wZT|nUJ?rC(++e_(4t%r*q(ZmNI+K<+v?4~hxBhRyM=cU@uO;*vtWZ-LGd~mc4
z!ac`G)@P8bYUQWGy>TuuQFbh_>e_4z#=G4219C!EE}{Ga?_uW+!s!P~Z%@qa5E^i#
zQ#lc!Or8u1k~1K0DzE}TF4SWDDNC<PmI$Pr7X%K?7c&Dd+m_+E=;;H@7YKw{<ESJ7
z<n}g0=l<GKb2TTob1$eD8JvSCoRT&E_`&Le8m?CQG7`==;@fi2w-9}LxH>jAwl(e4
z)!g)=uX&Eh0#myvm=66QV{{IxCB!a{<j6iJC=Ien-%Cb2-vaWPFWz|-mh}3asV)lc
z#)%GLl7%F_j&ADmqii)-Whw3Oy-$tt;*v9Te$UDq6%;6CQINS3*(x2F{20!JAPRDm
zwbxWaxt}xNDUnWW@z`FMy!GCi6RxPQ!P@WU>%b9l=U-H4_BEaB7zLcRi;J;(M~wuU
zsqm7(8?9acV0u7HShm4FlGv1y@cO8qh}$j-4!Vi^u@P^S3B!F3%jYdq{G9l_nR>#i
zdJFXaf!<kp-KVFr_1U@*D2tF-hrJ1Z*Rg-=%@z#Fx8hwUl9jsZmwpK#X#>>Vi|>8o
zDD!bD2zr0g{^9z;`o!wa+0jXS3V$_aYg(f*M0!wnht|mZ>u9MTaJa#U4Ar^^w9;be
zRs#R7j6`7^<`~kZ%`@NSlfwv^N>tiitOBy#(KQxG<@$i(uTszne?Kc4K#MPz5D0)+
z8*<U{h|s~#eR2S&rIuWR*19mzBnoV)W{J{)X?TzhT;C(`YANR;dF~1bv54ucA~kQ*
z)VzR;%N9LM&!}pGk}p!kD$Wy5DGJu9wB!{~Yt-E-?W>kKnrb0+%{7a^DzA<Be0FU=
zONJ=RJPdntUqB4?iAmjvUd?_@oXDrew+xQ@{zb-egyZ6j{atkG_)R<ld$ita(tT$*
zGEBMiWOw(3x}(|Aef)Rs)y_hc_{(_Ga5z8;T8Qv=y)5|?4%_hX)iW%gR82q!i-}&p
z6F1pKE9q|T+^WJvyogdt8J=_aQ)bKm$%fVg&hRTp6|U`P91POR+c*p$lyb+!iG_St
zKU|^pydd644Yz&XI{H&~4jpp^+5jEd>!`9L`NkMj;|!pQm$*Aiq&$-HpKfZ{;sBXp
z%get$L*TXIfmM3JNnZyZ@@b4TTzY&JD*f%9A=BP|xOdS`X9M(}KraX>wijU3GO2b?
zn3?KG7K??U_gfaC&;A^)l<@;6nrA6zzlWt#j<-ow@q+BnT|nA+FNTNmNlb7-5cEsB
zfEs&U55Hu;xsim2tIzY^?AnYcB669xbabiBt1D^!6E<8-hHo>7MIYo17pm=OG7AF-
zv3tS^o=6t`m>|uq&db4C8WLcLP1gJc`m11-tfLF`4#4{La{*&Fpdg-Jr10zEzB%gI
zQC3$=O+oj+({*a|R+(p8<UcJSGnvnMf~ylV*D@smGef=&>F)9=h(4B;815q-11Bxr
z+Z09H>F*$xNsHEF=FLKR>KX9!5m<MLS|eqSUAwdmbwn5~dX{tZjq6i6>hldV+y&lX
zYyA}fEhQ_uAT8da%$!!;Ow2h@I3gC2B;@1Kmjw<LbpIoS{A8jd){ny{C^6w>!hh8U
zme4~>3`2DuMv|0J0U@AD1^jteFEmVaMZUzaWC;7?#m!ApXxOc9=D5D^gWuaLJi^Xl
zaj;9oLe3Z$SRXn*=Z8@h9z7wXFkvo2vIq9zLwtBNQg4Jk_NKh;v<vf(?Vm9jqddi%
zF}8k<b~b~1!4Z*N8tNLrztEQXu2-c2T||*VRrRNw&xF=kY%ep5HIZ9})DJq>{W<c=
zHprr_o>2>Sfg$1?a9h$`vSVy3I=H$?*5!R1tJeJi_v&&-BlQ9kb1#tdn2(f!4q)Xj
z4Cjm5!QA5o(&ZG9Q&B_?GBVg!R$exQE6p0Ab%7sKkls2VD@6;Z#dOa*{t4vQy-h#~
zr+t!P6Ht+F*kTaKNIGTSjPNs9&QQ_1f=95c@qj~<JY3JU#2+8KJ7&_9S{D}V&S*qZ
zf#$zIJ7*{H1;FW_zm5CbMK|dH&a~XGU?D7>TF;A;Je|u3$T+CUG=)erhJABU>|xZ&
zkY&F;@g5VkFtg@hz=$=EJzw@Y8JZn4#2%8la0vby|IT69v4pvPtifJ#qOZ?0e-!1u
z#x8gH)*p6*ZH0s$%z_<1FTC7tXKn<Vme%|9?V!M-K8pxZ4?<U~CH4zQ8)9AJ(z3lK
zD&d6GOBMNN%4z`$Eo5AUKv{D`5jaM{YqmoCDu`y+FFHi!k^LCSA(@5G7^fB)wcPAp
z_0)v(U){hF3r^KlO+JYMRpj-P-%CYLrgtlcCT`;2`rDMUE<}Z%omRol{v59^XhzNZ
zMsVelSA6q#?F>IIplSfzAW35~q)Bns46FPj|5$^sTjS1fzh2iFrCY57Xs@D-%)W_X
zw1)Li{v&h!T$cJo50k-)|7trJTn#Q)zGIr5_6C3HRvXl(6XO~>RZPU;eHP(|W_n4A
zxNDd%J=$oU$ZWizG!G9HEJ!2@s6S=IG0;KlGIfY5vDbx#0m%|*1A+J6%lL`|nfhe?
zC;tLcs$;bJ;mKZlR)#_sAJIWzk3<z(eemPQp<(NHQoI)=T*E3WE*LY33XlsgB2#&s
zBpz&hoeZaxn)JJ08TQl|*|(fHUQuL7g#?h_l!V+8j61C5%2_z*ui$}${l$4jddv(c
zWF;W-by-;!uv!-AhjA?JjyyZvb^#|NlaysF^1Mddi!a5c$_fN<w(3SC&Zqv09Ow%z
zpz_L2t6JTkrw-Wb89|IE>?xKwg77!4s1O$v?AGvgHZYv$BM%DKuIEjzE*a{oDpF87
zo`6DYeR&0b54}iU<n&;05ESR-;&OI%mgv<%PILguC?%!Auaj}q>IwPGA97F?orKiv
z52jX3X}95QTSXWNvxOMLYk~K^!|!9gw6u1vsM-%&QtZn-^ATUdou*Goekq!#f7dln
z4~CZ}nlS`xzIg6^)@vVL>s;9D5%g4Yy`Zx6+tPTBN3HjpY8*QqJt+N3$L7{=_s%)R
ztvmt1$OUFXTOD%PRr~+C3PXC<xp_DAjJ=q?G~*-C6OpRZw(o{)OES&Z5m6|JGOi61
zAva;awnbuqTU0G-gx|BGI{$ii7dKfW3wGDP2Fh-~@VqiRIawVRL$G*J<N@m$N>Y&-
zL<o<f@BG*QNvFO_+N@EscS2?;7L60>7y_OomH6Ad+m4t>n}IlAbhDU!++MZRd{C|O
z{<Q4eLC;NkE!(DXD?^x}h~=HVm90Lf)0Ou2VRwg?v4U=ms3`6^W&iStxw$zSa&UlO
z=ytr&3+5F9&OrF=M^0O`9H&6hg_D(bCBOOPDFaG7;nm4+Z0Ec^6a-!O=!ZhJ00ZPt
z%=F;uRy*se`G)+5m@e15tbrE%oN1gF=`u8+Aa!PQ$is=Bp&aztFWB!Df&$N<Kaao%
zaXk<>+c})*iJAK8S#HnV-B$f#(~t-@MP>i^jt|X#E~C%CC+kx6;@#E{XA=(wY<Z~<
zni*V5FrxJ883+IEs+O8Q_|s9{a?L}qhfP!2{$)?h#ppn!DGV=qQJggpreYpQsiMr7
zU^-gGgMh4^OY~QOmEg(ysWzFq@nXovDbjq92>e@XzEmJwxj<69@@Le4t65q5kI%Vf
zCcF;^F|e6^|K4JJjF)+Ksb9og>awlLb)j1xclce?Jo7X6ZWTk}ajb9<ma_j?1jUPf
z{x&`$?=tJdDhk(j>*6x+UFg{Nk7&e^g(JAr+ogefxN(*pxFd_9v>kHtahbNlUcmBA
zRy}R7trvJzC*kiPwiXtY^~!v*4yxqm+({qTQh1avW<utRJ>AR{6~l-UEy>V9+KheC
zzGrps!Z~KT_><<jRIb!_m3C(*)GYq!zx#vy!0`!{q%m8Mz0U#~Ilr#3dHG^|gqP6x
zv!siKWVB)yBPTs8VCUylvqjm%kRskM6qMd#Ud5M9BK4qOM_#n%Tvp>yvZQC$YV;4P
z?KcGd>Qf#GbSX!^ObzE3=Y#Rh8!Q}jQe$u|%Y3E6=+z0&oYJoY`-is0pOLY*H=YvS
zPkEK!gb2hYo9gKdJdI-z9s86%gdQ|(MwD4G!JW*W3UyEcvag7x^_QxBjRBCBMxFFv
z%yL6<m8A$cuuwp^QsY1-PCzDgJL=@0``JHt3i;yl-=nYeZc^YiE--wJDVz}mVIU08
zk(3tBNKUj#u5RaK2RJ+4TdGpJU;q(fGpYLq9PW=~|8gBNsrmFhK8kC$5VN#S*;@dP
z4g*4>kFzdarq>2`ZfSygVXzz#(TSz(JyT<~i!=~^#9qka4N_<DAtlGnNHi8D+x&PT
zL?H@$?3tf+*l*68s&}*oOEn@{{t-4dF3v33p0949Owbsrt>fy_pI2V<kzE*ET64ic
zPm(BLEr#<+BI37=_2{;%{4m{HMXbRVyN{HHg?b<&%fN<*s{=LI+m7Pl!RNsT;@iCw
z#jHQT8~gO5_eEbtN6@i<_ha1U;(`SrGNe`WjN-IlJaOqKhGgJP+CTUcls6W_HFe}z
zX8V9prZPF+SdX!L$}2;5zyGG^`vGt2_yx5A=Z$?!*qkx-ByP>MF?h~!IY_npY<K%{
zrzgK}8{p_P`$6z2lUk3S5OHMSFfwP=W}{%BO*^~vmXW)pI6JcasJWV`_URH&Q>N;)
zs<C@MkN+HLzxifwNA9n-3*JTE@VM8tRba>49HOwNOr@S}xl*nReX{WoF9~j06Jk|}
z{-rlQ7gbC>O{<bVT(bHu;oH*Lv_I{Xe|pk=_8-ufQlJ4i#D8!>_gQN59gz{wTuom%
z%)uGso#FP<2H)3wqc$gc{NXydFD|5Jr;tv?h#yV+i;FljTxRQR!K%5Di%yH_ZZS~{
zhD7R#xN$75>3#r1X8q+OqHP_vPJLTv^D@IsL;#ixP>b^*4+5D9M!1zdS1DbP`J&<-
z;N^yQtr}LV5Tm_UO&RoedT}QGK^g_L@jKji8cp*sp^QAa&Xj3G%~KdIxmU93Q&QpU
zjSUATES-JTI^2!c04iaFaADBWL!t0OU%nOY90kqs7mzUgWCA9@y3f(L7K|OF<3wI>
zw&q$52;1A;-5xm%aBMkQAD&np187k&v+x-DH>^=ME&tos<R-;qAvNiz%Ko1WtTUXP
zrV4<gmF|<Bv*CK;jz$N6RfU(F!R@uF-wVs_H>DwiAb$EEJOe#ZA$joE(0i&+uxErC
z2w&QJyYq&I=BL-pT(nV)^}26jAg~&<2cVxnbFQ8H`ytg>F5>-70WA$HAYw<f_#<>{
z?wd82pjzuzxg8A>&$7K$Z1Q4K6sGW^R&6{h8~vEbkb=_kiLdLnFK7<l5w#5pymu~;
z#g#?pdQdNdr74$71;XYm(A%3Nur_^F0N1`hU(>?T!6x<ptT%M)aW%CpJpZ;pkUs7q
zW}&$=gkJg0MdU@`!2Hh;C}!`TL7JM<*k~f0ZltJH)9F>;!~*Ri+a5V-r?R@&<Ll;p
zF-c$J?6gd|a<1NWLqBI=YLJPKxGFgN49Y|}>{8++2oqM%rGb)$1R|577OKgLmegtv
zZhYCwGH0Ic+bT+_2a1PK!x@uNae^jUyXADHtxzx2%mQi10LN_62hXtdy~!fxl2q2A
z`u(p*ZB1aDyAEr;R$yg+%$ImNACN$AX&+B-l-Nu4NCchvwOs%F)s-NRwBUvJ8Rj<<
zv0*H_)OS;}qj9eu+to}jbk9crTb!UZ3TU1jrIgW3uQ{99=a(V{*uXBJ8HyIyZs9{0
z$eX}_i-3&!-F&olwm?V9BL`}kUArPl0aV9qHLzzt+F&mny)O~Ay(;*GDVyzd3k^yY
zxe?4tr;pm&ZhIQ89eqsRvLCY#<BvcIV~$*uB(H&RVdWh3xKwO&ROj4OFsaP*KG~Mp
zy*YHp0dgv>BlFU{)SMd38xd#$mBqmqeL>QIRggU6a*s4CqZ^yY{cY8M3X5to>kmKl
zu6?1B<K5KMTsq-gX=`lv;gY?jESUb_t9PWfnmzd!wHb|Wvi*wS**`Sy*SrV_<a@W?
z>^$7vK3$ASrRkUd{Uc?qmf87cz(4fOn7*$9GF_NKr^-Lhm?oLVw%M?~MVOPl9j9$z
zdyK!Ert#W(6-CnD=?vzIyYfSDrRRD;PT<1>>9aefU&Cj8!wHECGA`~6g`i76U2l}k
zV7P1zpU|gkEvCI<_c+kPL?yX=d1i>bi?&$i_1IX{>d{O6A__uW&zXt0I)^=J&Joi&
z()7#`=3MuUAk~oCX!$x{gM2Wn=Ad;TQi<_b*L<9Zj{;PN9{h+#dx_hh`^<6B<xLuR
zctwlW&iz$I2a0<V%a%2UTFjh$uC@jTxk~Hh4L>gTIDfkiYC+rS;_fDuFdO5oKAZFQ
zTE*wwBK+X~V6Ni~L#K~dZQ7)~0MpM2l;p;6!7h5e-`dqYTbh>rrnDo;f1C&47ksCo
z&gn{2zXWfXRZgj_K+i)7VmUt71P1r0qRy*|d2yknX$A17+e;xuR2kK@^|c}zyceGn
z4v`H|pb>ppwx0vqiVNGC7-FyndpCy6OD;K@Tu%Q}_Gt|aIu93Yo0*tMJm|!<>^E}?
zn|K}in+GEF6KbosW2R1iyZ&~6E67nTEg@I*O7T9@D67=h&Z(snFKtE;E~VdD-F~Og
z1{@2I%|RbhwpANj_97k0x43GgTK1>9B0En8Sqo187YSK#Mnjcm(GFba>3Pg<+Gz5n
z+_^upTiao-YP0(`E1!@4D&R=|LzZ<A<KS90>L22PK}_H*{ZQ0TC`9YxLmBi4u}+!}
z()#L&R|r&6cKRK|8_#=doTZC4Ss*Rx40EyD*}kOH|Ni{hr%Vr)t$LPIv)cV1VEy^4
zs=Q>a2<7eU{Gd_$#fLjYH0Uu;vdjQ9csW%^Q5*@;!^1oq;he2m%KY@w08n`g6hp^6
z?AZlS$$d_=9sKYp@?>lJPD(+6-2|EogvCBAkBWR%{Fij`XvtJxr%<io%(=>C*-=h^
zqNbq0D=qA3d#mr2sdvQH76<LG$>))d7PfnjZ)()yc&=_gYag#J{=gnc@JFJrRlY;2
zxPbIsIPdG5w78(&bDuK>dd$n&`CDtC24029IMb!ye6Te9i%fUMH4_C|FU1?_0<YNV
zZ@pXLC*e`iHhw17rQvM0qpOFF8Rj++d}-LWZu|ag+xHNGv}_kY%tG~;lT)=*4P|en
z?W2DpE@I!b3Y~tK4T8ojR)fhu!9T-}<<7}56w=QZtb*LO$Gm}wc`M9ZyEN3|&l&7~
z{)TJ{zEqK@)+5o&x_5=omj3L4?_CCv^U;6l)Mrp|CuzE-<$v>J5aEn;cJRCJc+XR`
z#I#C&>`QzFBU{00>*O<#27|do@1j|@cX%KOZ#X*9C_dcrUKFdnhMBV?$-OM{wyvIe
zNB4Q&SfSM11h7zFSQv|1`!YZ7#;)g<zdEtfVZSZ!><#23^C1W8d^dbuyvkz=(_npC
z$nUSzoQCBqZZ@DxuG~brfY>7@!}4+TuH`zgx}E3A-xs)aFJR#bqOiDRQri`XY`#>;
zwnq2@wR7i2Tmd=UmK>HdySWl}e7H*E4+pclPyQYQQr&8p7Hw=vsx;IL#~PUEk1_~A
z#p|->6c_(p^|}e2_{V*IcwK%WFUy@k2xX!D3H$}L$jWLvImQh0)z4Wa8usJ~A&Zyq
zOpRXqh?IKdBo83Qr?1Ai`1)?lfg2k#k-dfkLG4?#BR+xoZ@4eaR@Rqgq=WKMm565h
zEC8IFjkmel2QHr4Td7pdcgp{;DR2BzS(DQ{s&Q(>x_IhJQ}X=W^Mr{n=>N`6W1jNp
zr@jIsGRF_*W=YNTuJ7E-U&MQ3KZ4HV5sV0E$aEpH=ch?&(Y|kI0Xa-8&^z?%yzQn$
zT-s{V;O`V>jq!_DZR8nOJ{Uos)DuJe_)mfTg;V2Vdne!S(?9Dgo|yJF-UZSoQ*R1z
zN*Vjz!uy|!%V1w#1T+SeT0fc6&pxoVav=Y*eK^dqG(P^Ago$dbAzy?oa|>PFEsy+s
zLx%2&bZ(FTJ?n@{^_<0YGyYt8RHF1n2;$dr8sD0X#rcqlm^Fw<Ki2J`D+|ck0|G!N
zasznpKkX=jmD0@rSHPbX--@ScqmdqpXoG1=cvN_FxXtul---MfTd+z&se5U0I$UD7
zeoG=T9O)s64`|<>&N^BDeY7%h5A&46_yV!l(%|aUhvzZ$=X6EZtY5bBe-;X{cn-?w
zr&zp~Y6>kBY|?lP3H0mpy9(Xw5C?`?K>oHKd%IXlTl*`1bykRu<OXQBsZGV!03`&)
z4H5I8L5y+>&8O=V4gZesc)0#YT{v6WnwyiAc&X=0N$%s(!j~e6X#aRn@2p)PRKxc_
zDCiPgbg^saM(OYJmMOb?7YgrL3ag~5dMVL8Ep*RZk^fkXAtl>*`A!NL1HWVGQcwHS
zHK^-hQXc!)u5l~G7{(Nnr%6?B-7|t!^}tXR{>#&Z_t0;$Odq<cNzGEjw@{@7^P<J=
zO`Eq5ukuoxD)msr$OY>DW4R_v;aBTjAG`J!q9bSzTM+$kYymjgI$ZH&O}zD%?NVIt
za*RXKyXRU4{zI<JKYyk#1O0#Y?*+c?O5iS|T`Loo5D-<-*Dk__G&Ar;>rFzfwDHST
zQI2L1c3W{^tuxV(LpxoKqnBPr`kvN1v?8msf6?3ij|^LO{M~><M0;hHlHX3I#o`nE
zulc|Ll{!OF(Fy|(F>#B$KAZ2oH7%F_kEZjCr@H_D_&HX`J_pGj$3ePmva+3IlTr4j
zkdeK=GLI`|TypG@6=fuQhmd0=;|Rw{h?99TkMTdh8~?lK=6QU+pYeW;r}w=@wSE6W
z{Ubz9_K)9A<1qjS)IDXm>0cB=c8)s(c;)zlf&+|%z`E`RB>8=Y4<sP%uD`W0n>vNv
z?7deWnwt*kN|h95a@R)9ws@y@#cgQp`Ten4lhp1itc+$cNA*H1j84!kq!Iqb=_+Bf
z&OsHpJU{cW$@97}lRUc9uN+-*eTcVVJXhwYj`>e6g>xwR6~)jml|JIE1CN8M#Tlk2
z(stvlV^7tm1^q&h{3h53jTR&q1MZ4~-{olKPi&h01HZ3+9s-dnK!O}s?T94xeW$sl
z>*(_J@+junE`fAZwVz4e+&Bx&TCg@(+;7YmveI0gs_Th9-@#+1s{0j87}ToMIp;vU
zIYIue-FRLsFU*O8B*8CRG)*gw$uJ2YL{_hJ)sF=&EU-$+%9(7)EjI(8{K|0E0$m!(
zladH|#*=#6BH;;k(51RCB~p?P#>DqPQqzr0otBM_qkd`LSnYD0%ZH<a1Q86GQZ~G+
zWoW}&-St3>(9jt{paRWzM{#qrV*>Z@j@hXkN40p1-QQ^A$;e1KgHsJU>BhbUspxT0
z->4-}F9tq^7n3<<&jZbjjB$NI@bg*fU-Dhbqon+p9@8U>h&>*0Q4|ZPOa3l~fqAr#
zeg)De88Y**jpp&b>iN@!XvO6*3^fY7(ClsF+trjlgVF)VL@ftc*Jwp}m9xM}gc!W{
z4CxczN$R=0SWE4>{ChI7u;F8pV3csD)N!<lqE~gev}*7(@HRESkeRGge2HLn&xl7G
zj)Z?mDx1T~<t7T4<$S&emjpW%+7x^Ik>f0FX>N&96Qh;T-{ww@<FP++sQnrNk_=QA
zRu8O{3Y6(;x{B<=)#5|04|i8_ZiTn#cWR3V;ih+>8lhDQ7?ot5^zN3ZE^p#&WDO!g
zff$ifhJ*qzD78$h%%t*!)wWx?Cl91ov54aFjOn(Mez5QWp#cPAii^nDz!k_0qw4Yd
z^$3HHCOLOjQ}p9Gl)~OA)kS>h*b`$s*J?QokMDcmu4aD?dI+UiL-?w768CXKn6pDt
zZAj1MF{ToaaTkzGTPcbaxwu|uKO1xt<(XfIlXu<^H$c2r;737Mwu$pRY5=KlYHzta
zinwv^9~ZMQbyxJ2eqBdVBnG0vmKLXojB9#T=|wz52i+#O0o-ycs%A;1wWFyoi5ZjO
zIA7xh?1TiiGOFhl-DkHzX}1tEiOe~wUI7^?zLG&0kH@oQy1vB@xhA_;I!wFYW_)b$
zT^b5fU_ME&Z}TZ`d{VRt5!stu8zxR%;`8;k3FPa2bKKB45VRONp&1~eTgmndrvYxn
z@AarXQF;;%*npm=q<3F;@2$>8@Rkbv?B0{W;@P;tUIn9_sgq+3G&g$x^n#u$7ThYX
zDB*CHntsUw(=AN%d!FRICwG3-UK=698K2qF;I8_U7A<nn1I%Iy(kwGvF=@}i<-bZ&
zQ0}|4lPb>H4BRU4!+qU!OIUTedAbQ{5s=Ob(@lUqEhsxrz%iTO38v&Qo=I-gTUH}e
zf<g|Mbag#TO~pD~ACS71x}(qhadPwm8ln`M580ojht;yvb~7$(%QuDEU9Emj_<9k3
z10uv!!EzI8?J2UW3Z04Mjm8blv!q!eCT>~yl58ZLF76T?ct5jYGcL#0spniK=Tu|;
zeL#9KA3vX&$-5jcd8|{+uynhJJS-jVi^r`;$`_$wauWd}l<;;B_`{rcsNP@R{V2&3
zpHe(q#c)RK@2Ra!ssqrshaVo1Y&$M=W(goovbPXEB&_zn<uAyj!e7mm<Q=jbF-q$a
zY3UH95{4iFb?}{G(Q8KUz3;z<U%Ps21GF*ef$c;1r4O|^yXikhudg-C>`dR}$itN{
z?hAXkgZ@g<vI%?&xhWRly$R>v)!?A3Fj3jOrwRHzSuOvvZLMkW`$xsC{b~AEK{CE`
zdJrMiOqR;+E3pA+y}p#_L1$hQ)UM7C*Mhn~%*rKD;HH0<cC?Qv0&X|2{qaI_fM6PF
z2B$`)1a5qD1ad5CvU^MGGIF@@cfF1g4sqBJR91*dhCuuJ{Q~{D)RWAnzCmXzHoc<4
zJEZSq|H=3)UMX6A$K^j(o-3pHIW9E=0atZP<qNs!<#^&Q6(pZxG))JIC|4fg%3_mb
zhSPJekRhdCHi>Jv!nOzDZg)Nd*#XWKASzJELdMfY=!#FzUOpNVuIjK{I8moIzULcI
z&~dWA3*~snG{^=?XQLPf;kX?gTfwoU1Xs!(Hc|C4{JfLLu%>9t;uv<`TnwZQG64l8
zHQ8(zej?XEO8$JREyXUN<a6Wr;7dy3vfL6I!76^B6SHN~@F1{B$rkYD!L9k*!^l4C
ze;s9E8?~XMbpE?!akmfluqkB6{k4vrcO(>)(fH7Fu|1eR#-nA^JhGNSb=*5HEE!^8
zfU2I4m?HVaw|U|$8)t4lwE0pSLjf!qscF8;(f;r;wLSD6`_jc-163cJQ%lED130O?
zHd#svirm4C<tQ}IG6T3qS27*A@|(tG^lB#?H>-e5%SS;(PP=&JlyvYJz9|PxE*Wbg
z8PDfF2&WiO&n$y^#{0ZwU~9mDZ)mXf6sA!p+@{_{z<+XX_UDCLcx@xuI5lWSUA(tH
z9-3VJ-LAGe^_!z?pq#E%3)mwc?jKW0Jn`<8hk|VTZ}4gtkdpCqxQnARWq<NL)jua8
zH$?jSc<k{?r*rYp)zMMq8jt3s2Pw_%Rrp>_!3`%DuKyG@%{k`7xT!+PD9AWNSVdl@
z{Q0e}BLagl{n+uRaOElyjLQ?#B?D0<ar)(jg0vmwEtPC%TR(Iko5YpWkcGsG75M~o
z`^@*h5F-}}0|4c#;{|lT&;OLlmb%T8mvWRt3wdGr8O`^jBeX91?p!VZo@A`We)E2|
z@U*viQea7eetSDwwIIOg<Yb7XqH}|JdsG@qNl@rmdjE2Fx%gxYmzNouJAPL4shc7H
zQkM?!JYX1Xs3l&4WtgX+axnG9z7kmaeLa!t`KBJAyLSE_oX-@iN>Qj)psMtrh@fuR
z{7$=PjdwgmpbvN8gIEv?vhw6s>C9wy%PPFj*7EDrN#k?7C|jDKp_}n%@1x40mLWBJ
z`=X^9FdI}d?v<iy>8nU;GBQxoZ<^JoyBwcIL9S!D)z#tZX>8xmtZg2vk4(jjO7)(D
z!6+mEb_A6yiMA(EF-$xk_qfw0S7VqE<@d!;`8lfe2juyu5jaT0dFhRRi-h|;hkEGK
z&=>(ywMx-gE`+i*ZV6CW!tGR6A#~@PW|bW3TrB>Dz!--fftbF^zVVPZ?8wMoGWHF!
z8~d{xT7<Z7j(2D1leV?+p`z}qnr*Lr=_L$6Sc8|XjYY~E3vQa*5StR$>ug=RhX)kP
zXFcCBh}Av#yCm|2pCWy>J&8O9RbQsO6I1r5qGL@1+9*a?zr5UOoV{tu_2Y$qViC(|
z{`Y9Lo!~ZCDKmK&XEVFm@@75J(PTxhmzcJ0We4%i3hwytX&#e)j-ifeP`>}bLb`BZ
zgu_5<;Z_TMtP}p;(BQSesFf)bLN?^aZ&sDQ2gKSC$Som)$qjSgsQZu|Z@b#A=A<>;
z!76#$I}L1MMrJ;;L7>Vigcig<-ZK2)#(w(3Ef50|+3VyNu3Crj@6Nv_USHN-dtVcu
zhiv}xvTM4#GiFZxu%qFdPm9B5SbQ^(p~IKk5y829J<Q6zk@Ik9QO^6eoAXc9H?XRf
zo+yZI7b<1gk=c-2`~FJ}I|+nt^H(^Y`4b3K(vNDD<Mq3n;FUn|tZXZ~WtaEEu$|tA
z3M|=uwrWHX<9!cQ(NM&rx=6}Jp1pkEeH9uO9=ZMAnNED>_NMRo8@74QrT84+dL92b
z=&#)LItZemB)p(5k=C+0aLlIeT~k_QTsc&?RlmH6=!rVJxcZyud7pwR2^8ad;$g@7
znBFwj587|}&6p{6Fn#yfEbG@-{?3{57PbZ>b`^%ZJ1Shx*)-Y7KX6uT&iI`N1KDNX
z2TvOc20E;3YMK%$Y+n1Z_CBov$W~2U$^4o4^)y(ZQT1YG9PC9r{28D9x|#_dVHf9z
z7pGTOa#*X+x@l}LvKQeW-pdyktIvnuQGc<$B(I5}j3FJE76izrdPm`hexIIt-8clt
z{@%rPzPC~bNj~J&^qm4BP)-bJSXOZ>H`x#|ThbX+=ews9Kl!%6Keqw+>z??c)}>NF
z6C4}!Xzi~YFT>{A<!b14Z$|QuMa$D}F0&EOSp)s?-O1cqFWyro#D)!WYP#CvK@40s
zts>uoeDpsEd23LO*KLiFHUmdJh~g6=j$IP|K?I#|K`*e+T5@%Lee>feHeUvwl`pBD
z{ok+<J>STv&M!NF)fNRPhz6<CL9Hw+Mq}u1!@tr(J+Mn0-qDx8=NIq21qZfFN9@_<
z+~#Y1AMOg1iHwN8CX&RR%eAcw)fRWD=BI3*%lT7uEwhuZM7iRhkLZ$zFQyqIVG8-r
z<9kPw+YmYbX4-=4L-4GubM0dI_SyNy1oB1KHJ)q~C^G+-$QvMPY5j6q)On{AW*-yY
z*u*KInVlqX$>K}5R`@<Afx?-ijW5Q1vOVaNx@gB^U6D%B1iLY7mtLnnO24~ySwC?8
z<AUXWd-Fz{BdE}jkhwiQ-KJx=?$=q5DG*;#zjCv0_a$+XgQ(!%zj2F!EWBWy0mYK`
zK}noY34(J-m<-6#sClnF@NUTZq^Zy2btU%bsPlt2sGWQ8paABq9NsdCHyMNa8%^P^
z2xw!*lau**!W3p|P2VQ(XzuXLku$Kj#zuAS?e+BWwS--d!WWLT8;ija14S-W6m;%)
z3_3)pU*<f2*ItJ{OH$hn?#PsWXZ3#4a4=JAX4+E0xC6+S7WiqBS6>>zcFank%W)~H
z3PKeh_)?v5Ik`~m*Mw25x<C9!QeWNjaY~U~0?=B!KX}FFootP_#?}@<Rp=yV+rusK
z-pJLJxlQe0A)(G!lGN&`43?&twDv(I2?;f$R0oHp!U;^dXGo$A8Hq>w52_@FFG;-^
z77GIVK$dXRcH~?utY}qPa#64h+Dn$CINNS7;EzIqvy1z@`N+c~fA@-f1|qIDPqSvq
z$Di@750%|c<upzDv$eL~{ibq?BwS(nbgj<|n3nD<9>?u@$^4wd{keOZ6mV%6vP`P*
z`89#g45T$Nv#mLb5H%Z@wEelkW<#?p>4~Z*SBke9+JCf4_PW`Hjt<Hsh(#>{^D8NR
z#&0a1SqiEDO-HyGq0}=muT3`eFAq-7Pf3RPt;TWOUAZnQ3Qftr71w8f0ttS@n#7D)
zkC~^erSu9F4gEA)NZr;<**E$68t}V*pDMbWqYKdU)w%Rpv819wRINe0I;2k)0lVE0
zXp_f81`Bd@e}Y2ZPU%CamyQ5DNesM23M?_G?+_{WOcnW_pXZ0Cg{Dh;j7db?iliV9
zg1=DU!^!I9%0QK?_gG`o(!V-bQYf{Y)K4_7xIQ=i^%}x5y~O=2@*&SI1DRy&9hi?h
zLV6=|MgEF+F{MEMBNHFPJw>YKCWpsSMLTD>LgS;zs+#vXFW6gO(ftN&c({DNHbOla
z=-lp*6vvDGRsJV!ojpAlp^9CFjdv8hOTX20Y%kO@hu(|7o`+wRzf$$1c<$|aY7;@z
zJePpnGC7^<=l}R_{{Bi-w>>(j+dkY|(QQ9_S;DEifiBL}tH4S%`uy+e+6^bCZJ2o6
zmtFAdnzM#EQy1@t9i?qR@zgissj-}d0PNqC#IW~<QORR18!my(2PsI@Q#z8dH}r8!
zSdNYF>ak2%V{?<2Z`XTH=)bO9P>n^)nY9l?19bVf^TNF?BY=pui__!d)14sIiKYBl
zs<gTO-h0Rgk{mL42<|j2@8_66`}X&63whVOB!Y?c16ygow|z*UA`0QPz?$`^)TA^<
zGUS~$g{W&*Jg6TgS>BmftT^}d9s6m`v8yu~+4MkWPH<oH1A%F%_RFGSmcHkX4`cX>
z#lyq%PKZ;!Jr1f-R|wNuj6gi|&;EX`x6=kU-aU%9dK&*!hJpEwT^&Js0@EPA8KODr
z7ILxghp>9z-Bkbu`lQ~Lr<ZuvNI!Ho8rVpLDpEh=$qAC*Q7Gqpe?8G!pDAs?hJ9~T
z{;P|tf*NY$HJ{r&YmT+H@Q!~{)z%)N!2h%k6FJ%R$Dh@if+1-smsBvVU`s|pOKm3J
zX3*i$q6OoV-WlyfxV^=@sT3J6cdMwW6^L)n$nvE@BL`{$COVr_@6u(aA=9^Al2$)?
zrlUMmxrFngV);PXZNHS?h?LVocPGctO?K$@hS6$P)TqllirvZS=2k)7(b$Zbs(XJe
za4drofqHag-TKOJL`VA@-}%&!J&u)|{*)6eOYyfi-7O(%vIR}cgvPDDT;1h^DAG27
zVot0{vyV(5^ZQCBU*FuNLh_5Ex+kemYmR1}3AJRPKOsbs&4cn6p5tx4`6is}&LUyz
zp*o=owB|GY-=rjaZ$k>S$tlUzgP!)chIEWy9+?*{Z7*1De$oERm|b%3Wo95JCzRhA
z>VcFAh`n^WChj>@WT3;Y4laj?3svL)a4e;0=PiJN7TtriJAFdQmxNz$K*66Hzt<c4
zIy<XV<oYY^Rx&2(Mi3sAV1v85i^l8F`Xdq@xLwJeUX{fsGCT9uc*EpYQYg#nw0=2>
z6GjGuQu47ivZAddUUKg_4r%_1CFk#tM~IdJ8!o*Kuh=KQjZKR}$YEv~Qu<q6A)h{~
zH7@}XkJZx#M1-B@TV=Es=<?3<5d4I9zpdnuRPnANAP=jR^`Nt>)8m#NHHQg(8|%Le
zr0!JmHy+#7D+QAYaY@dUy_dl!eh5@Y#Jh8AVs%YU$k9HZzcd6Pv&Xvqjcb~!n`&qe
z-Zu&PQ-NuRli_eqdYXN3vOBDbVN<C31>o>AzxvIR(Q?Qi3dUsS#EvRW3ciO-Fxuc!
z&@xMaD+i|ODl{_UbboENzkg_gcW+OPOo>)aRjuk3Op45Tr|WpW@qVxP9`s{V^XSVe
zVWav5;jbj-rl<S7MMQR@2-D|Ki!}wgWXOY0C_`y{%!}Iw-PCWc%l<JS9^V=Zw*&rh
zs=F6ZgkJv}%+VcR;>@&)^O0tJ`OPoDx5g#3X2!E0WlhQk@6fZ^)KpI8{pwuU3GTp|
zOBM9f{=bHPh&%98=3ylExA;0tYz8oCyT(6l>glmB?C!ZZJ&%lX_;5xQ)~!LOo!(}F
zk+im|b<ad4ORrepZBPC12soKD_pi6ooBrTA@euT}qs9>3RO@y3E;93@3jGE*GDSR1
z!z46>-tgZ}kKdAfy&vYKsdt&4Z0ygqF2%<$n4q&MrDP5It!e$#**zVRl7zE%xg&h_
zi9?ZE<RWm<O8k3Jzf7yKxU*MLRljed(ta<#jLv`y<;7W&$0Fqp0(QYX^&jxrqpwy+
z*t|YM=st?5<GrmOsUzIC>0LfqfK&Pp`6d$w#~_KQRbA;y)8(wGu&6K_iCgbyY`I5!
zWy3=PzJ{soglg4O@x%GSVWDjc!0CQ{Q7{&Bc0>Xb9!+mhPP0+UUE(Br6k>eKN5}kw
z%myzW*Es|ZCVFoI_hVqZ?D7>7pZ@sgWTb`I)nU#qFM^|^q9dZC042NZDW8@Ig(X8l
znT))dM+?Of0zpT!f3H@fb4{6yT-4l}L9}MC!X^n8sUca6rcV`KMy?1_e+`PfR&VTz
z$i5F+ET5qviHC7lY?Gs&c-%Cuqq==FpH%IUmQL1Ra#9Pd3B621=#kpIJuF*BZTabx
z(#4O@!_O9$3$ISs&JGT4(ouet6r|FOkz_kfSo$EZ8KUtvUI*nU)!fipeh!9*Bk|#O
z5drn;2yI$(TECzWMn;pk)m3)tMA>klzD*$lOX-=FqMjW9rw%-sW8Iwn$>AbNN8y)=
zG<$1T(FZ^TqQ1%HP|=4S@wvXD;^$X5-oO$LF9O4_4vvXEqa9i!Il9HQcqkc+7T$20
zZN+Gm-%%a@{_dg(UwPkvk3HILr}@K7IIX<jLK)S9N0>~m0Ejsm!W`CGJ*aTfc)VHl
zPd~lK-v&vzLDS5<1=h`{2JY)DvtigNRj97{4Y2W=89Uc`FLOTzL1*CQg}1H>Ne>JW
z&hRGz+^EskUVBko_aaWiz14Uex=DAE!@uq0@2iqD7M-)n$wQy<&<S*X?-T8dgQh2B
zGn2oF+hRRdaOJEQQ503@ZcpmK&d*w7n~rLsy`$0@>@l%mslYP(S7zj)^1cF=|9PaK
z0momI9P6))I&jFlAisgnb0CQ1k$sI9=+1K^3LOod7safY#jNw=;P&qBKjS?(o79OS
zMxx3~SDHXc-Jd*KKkioeIBnFfEH$QHvam@)Roa7XI-9eQat$w%AfFJ)>L3{#ywadf
z=fgqAe7$5VOFH;sZ{%76AN`~&2+M#4zKCNV{Sz9}xB~H$Kk=K$(7Qv3+^I>h(d*0L
zs`vO$G|Fz-VcFin!APPGvre;-&{ifZ>szw`;a?8l0Lk|fCF)KbaKDqj{M_UMQ4f_0
zcVwKH=^}%=ZS9L3iX6Jtx}qb(Nb7icQ#-$+VcP976v)?Gl}g>7pBW9R32y_$2@TPq
ztw5`qYmT0tNcnDe@Nak0D>Fdl59Ya&vyjsRU#$Zcu&1ifp<n64$_ykEj6}j&$|Jht
z=PS_eb~B4$DRhRO&?+CD5CsKoO;g8E+G_;61U1v($DYuqEzMMBfLhb_nXkk3&h;gJ
z>qw99BqT(!GCo}rO25b#2@K3)tv@%>@|Pq*`-f?sGp^DPse?g|yBO=6Uld`c?-E{B
zm{agc9f8%`L^yN4^~0b!iClFvs)T>lbuy%@zRY)cU*<089Y9be?u?ourrF7h?HwJd
zYr!VHCTp*xG<y><!hYC|&8efUp@q~viJt2g<MLm&+M?+K-+cge0GU<B8Z+I&@V7hX
z=n*a-TxrW!-}1<&2s=NWMlH9)U%x+?W+te~uL*(`n(FC}K6_=EMWQ5HFBA#axq^fo
zSL3CB*)`R6oh;q^uJABD=hX<-+{=qYHSxlw8TfDbA^hv@0hajCf{vx1>|oqlc)_yq
z9%w1zKGNb-w{<F>E`oS|;tLQb)Rydx(BA6g&VY!Kh+E3eA9_tDbtjYe6NICS=%|RO
zJqwGT3w%#BQj^hl)?Y++L_-ik1NZU{WSds;hEssmoBW#G3Jt}Q4ayPe`wD+f%#1>1
zf^rmJ5uAJ9aomu8ao?0RNO)mb9R*=i(VS6<^lqKCe4htb@S^~;X+8-2nL)$cqueEq
zAxR3ARXQW;bK!ekHJFRTza-b&LVLp(ar9*-BQdUH3?DvAv%<>IAAFRm&2q7Af8Fy$
zWI(Jm5)mzJEP^wfDlsq<`pp4rYQif!)M4GMY-Yfj)3qzlmH&je_3=3+_x_$8JGWLD
z$Ip+WG5ya;yz2g>qzGM+FA5_=oqtcv^*gwp37cojxS8$DE|xa!`e@lC8{`WbmuJE=
zKu+EEUUo?akmM0_9ZOlOR#cfO$|KDuZA-3DgItnP{#GI))0AVIhag#5K99j19*3V_
z?ffl5i)oUpf$7oSn&hg0Th!IT;gCa5g#mKazWrp06GVC~X1pvjulP!iL!X?S1m!e^
z@N=beIP$nR2=RSKsWUM-!G!poNtJ#6iZ*){xNzZtVMTtLlA~_>I&9dkdyhN!S$V?i
z&=3VWEnZNr4K`ZwAL;ac)t$)VuNE2<VASo(fsgb?mKsugateU4t_YvC*CvJ?8Zsr0
zwi_vz=N7rUTRpEGK@X0=W7<@s)Ksnx*4BqG((({8$JEt2PI@Ew25kIwHO=ph(G2~K
zOh?BRYSW+6<#Hjf_^8}LTxX;Rc8Sf-%j4Rv)KPkl;Q*!z!<2iA_`KhK7mx+!lDIb;
zMo>EO6p^pyBf#(xCB+VIkxWtwOx^JJQAZ-Z;;iEXN95CYV5L$PCXv2^>SOHY%XwaG
zQ*i1fnD9&V#jY~KRg01_M&z@Ce-%6(@*nD+y255YsJ^NLyEImi++Nix`SLxL<A~yV
zet=O%d-J3mlEM!=t5}&?!JXMuyTLFl+SzXV1W*)1170DN(vAqD_|<>;@=?FS3yO;D
zqc1`|gD6p^rgy^PxyR88MG^rugd;sD?=C|UW9O|-iN>CLwM|j|p(+#ZRG@xu^7<<M
z=!b!X_e$q;AA({~3X8E}dRvljI;sqG8=x!Z6_<_oGW`ojtRK~E+jU_#JlTrvh~36+
zKM~~+bS07;3Y<JpcP*}tDqJ}0?B<<+Ek#M?Wf@?`_&Y9U1f`oHDP>F)DR<ov(2mJ$
zW~X{v7T*wbs*TDdH9fT~-UeyW36&Vcw7_!%az@A6tF>!Fl@?m|NI>1Atp#2QiPX9t
z;Jj1KfRSjYDeM<0+dt{ZPbP$1)mh99!q1Z>q$>^rzmmR<jZ#B!EsuQlQk1Gg_tjhG
zlxyV?bSv+^-Uubj85|yv4=RZ5XsKm&j2Up{!QuLx-h;sci&{E|tZQEk;$fdX=s<j2
z(5UHsBx*rHyl-%!k3OrdTcxPmXe+b_laucJi-rH^QTUD>_?sA%)R`0EC1y~0DgH^4
zp(_=r`U$P&*N9(M79{sT>bj;#gzrIKJ82Hb&vqR9%&*QqyZo<qU~#N5B=y^@;oZ?2
zF{<V9S_6pugin6D8D`RuAmiPvInlO=agQD~gYtL51QS}wC~2I144#X4{A#Qzt``9+
z9^1s6lOCl*%iU^fUE<gv(bT-ZVscEg?)?DxW$^*eX8LKa?i{rL6$K*wb}~o-ME*TX
zwRm?f=rp{KZV}|1&+_2qGnRFp{_i2B0^Wvau*6$1^$q2+kd;1<&RQ3R9?0TN5b7&<
zTEGB3-`FB)oS!6t_EYdeiJ+kDam|iNP0d3?Erd?ntaFL$^ZoUf{vqQ8c`j*~08rg_
zR=>djjwE9fQPF@XBkz6v@A#3EzRoQ^e>SHVV3b>KFfmGm66(zUcH61@?p`6t)Ci%R
zxc!>c!SND%aun!+@pMwqp)+{b<q;Ds(?U24S1PDQzwBUoo68!25!XEr%JEVukO;D4
za`}L#=q{c6q*a0poO^?xZ_jO`1b%LK#jpBY5W@e0S6|1k<>_VUjdV!w=r5uDvrTpL
z<3%tB{gv(?1MimcQIo9G`uN$q<D2^LNo3)nRNme1&*u#l>+_200xaul#=LB6u@!c>
zYe7F?PM{>nuFp7r2<HP|DK>gl;$%r-F7p7IX)vs7I_}o;66IDnbI!*EFfLhB_>BnM
z@ISRuiQLx}m3vZ69$S_n&1AQkV-;^^RNl<$KEHqAW2~0dbNT%IghX{c45SGXntzhR
zq@(acK<h?erT10bZQa|8jpANGBoZ%;me8n3lhoSY<k(Pic7~r>gOk`njQ}8{b0P71
zcC~meI}#M9+1O^)KX%>#JTrTGHz@UmAWtP2l__~E3)%W+FVnOyruoqksPn5wvMyQf
z>n&^=11Kh^$bK1oIklODE71}9IvlI>moz4R)&}=T(a&+c!>qBm%<a!aWs37VlMg?K
zTx+nUxI6dphd<v@;nRM^d}A2;qdnDxPj~B9)`5h*(~^yiO^uc`;JvY7cFF(-#owmA
zm6aKwTlGtMf6~jBhar9NRYqoTM?sk@?K`bKRt>8L9Uc!i9%~wOMpEgJm^eN%x+9Xf
zMg*>gmW}K7s^w+u2E-0SWxgxKCo&iFWuTy&`KGKZ4--2O&Xhjs(y3gX>D3#c)z#`U
zDiDR#_*P#&N?Qy7II|Yv_vBylko{D6@H$MechL38K&FC}7IN=ZeufE&XDD1nRnuJC
zEw2Ags888YQRXY^h)UByc{CUT>5D~q9$f7Fz1%w5TQ0gTIs@3Qc!_K2G69+z2BI``
z9o1)Igsg(6r^#UG{RRYC;HO7ZE9um<Q&8sn>U<B!&-{Q+j9^|>n>Ib0V4{_8^=@_U
ze|+!~HdJoW=F!%Acn(oVY*ge8f?{$u$Zmlc)@F|;5x_A=G%I}w%{jw4I*X1%aqJ0Z
zJ+h!lp1(c_W>+;)##rn1gSS%Iur2Ut-nV=3dExaJJDz9@d-e*@FrC?WNK(hzm0xzS
zS(eDxOu$Spk3tAK-E~YgfQUuzzR>6X;vt`hy}3N0F(A0iyEo=yg%7fy8SU@r!*5Bm
z?g*oBJ`Y|5)y@X$ZMFWxn@Uu6Yp`1Ati|ra2B@<8naP+(y@LoEz59NS97?<14EKk7
z+cKMG%4eM-*{I+I$rDi73!S;|Z%@MumzV3%fPT%*?r4?EAwczV;rinIBJi-5cXxL;
zSTG>8<8@&$h{3#!GVi?$giOk)gN{WX@6#?yE2Et*BK0ywi`Z#So6;-V@XwFQ>bSVP
z1(S0+x^a*p-{jMMANm~cY(`KF5@*02IlvVG`WZT*2zt@4EupS^n@mLDIfW8zMd&dj
z5ODVdmffszEQjwII_&{zAics>?RtPqsU82E@3>2bbpBB~R3G<ZBDFF@QfY;A*X(?S
z>Fbc8%ul&9^R<{aiS>qvk5K;m-)>VzV69bb{q;<5_wr{24d4f44f855jJKDNAG~Tv
z6aaB~vepiVT5Qv7zfVh1FwvtWPlyqCywg<+UX5{58otBBuM4d@@!@GYs;FQ^inj3;
zTh|47ct4kb2Pzn7R*CO~7Z&jp^;}&ZtPd?HhY<2;u>Jhe<3lQ%xVM~|5E+_sL`J+!
zYNLg)|0{k?V>knBgR_pwqyBM{X2c&Z8~O>ldT$}lQ0F-lIS^&*8T8$mUnl?<T21zF
z&i99FTFlJDX(thy!)3659Xp(8Z;k8Hy6wMo$WQGNP#G}XNisL#Z7HA8eSf{+0*N}3
zVRJ`tJ;O5k`f@ka;8EIf${H}{QKbL1s!1DVR0f0gq7ZbuPJAMhJ2J71>@YrNnY6To
zuu$=-eZJ$_@CDgX7N6?(z;Qa`64_GSRkY8h7v9^EH6bT_58tV*+SS=rFo6y-`ESNH
zUJfe!FqF=2YT+!hd)&)=^$yEZZ;6k_p*{L1LoUugb{M>HnkJ62TLsZ?{iC<ezsS@n
zK4jS-UhZNlQ+Z#X+%JP{%H}$*KzRo3{Pj9K!WDSn&RjwhXt$bnpn0hycPPkiMKin1
zV5%{%JMwD!=;P6BI4|ZF2qEaIU6vnUw_uB{tekqThBK@x>nx2&2(T%Uc@J-=lNC2t
zdm7$zESBWY{UJuj`6o+Z#=5RaDL(fuUR^f+4N{=-mY^Hzn-AV}qH;aZdSdS~wMUht
zPO5cBMQu0nR`>(J+>W#`0~P0*xxC+UIJ)bf*e88`!y<+h4E4cy_w5Bob7&+oG~_Pg
zfzxWnrxywBxp$O6=d->qCJouGvx}+|=^e-wsv}ZXi_e8UgFt+*=VDghyKf*ylaVOq
zkD%0Szt~^YEQzK)9hc9Dy(_o-2h%~IPw%`U!Xv-lH+4@3dNQ{xD_pe%e$84l{x7>`
z61!pVbpkTl@>9;WvBrW<o)^3Gb6v_+fF28;XjAN_439!rs|nxXTxrs8UTqz3+Je80
zsX%jNe!+yExi>uu*z5YIQfC}V7vG1R_T2HQo{gw$YFZlcx#P~0Cfw*`$uS3>XEK-x
zkC~#96WY-YS8*UMYCyT01bvXu-{IH|WPap(64r49itaarm5?mU83{~^$WVpB3Hg9$
zQ%!mM?v?^_CGOAoWo?Xhjzd<SBv7DAGxm$ne~ieLfRWDp`4+W8$rx}auv-)o5)%7D
zk@i!jbh4k`Tld|iv9Nse=bSC!p&|)B$oh|r6m+??)|Q`)2J@L|T$udKVk)ZbaI#V|
zGBightINPz-^#N~**bkeKE$qJ9U70vFRZ0*o)9`ejBzcd<mic*>GrwKlO}bT@SIJ2
zLv0gKod=bCo+YKmsw-HYfspzsu8a9G{TONT8`^?oCzijtct39GSY~I$n>s|&nzq(1
zwX$iZp8nJFIfB3@9f%xN#PObpV4Ql?3O1*Rk=^3h@^R9?w<PhHR!8yrY8~SXoE=7o
z6=8ooBCxg3@Nt-YzaXMjz$fjU%0<D7H-QpS-ro$&k|0TbNy^n;M|ubUy{`8aHP;Uw
zCEW>nIl5bOe|p}th9qB9kC|uC9fpehFzAr}lx=N?kF=U?8{0eDIyk%};-}W|#09qz
z56c@3#U>E`RFv9@p}|gc)y>6!p_%6K3gxbooo!WZHuQIwpCqHCYB>)5+`lF#u>y~W
z0gl+k&wObu{Gv?!IM>T-yuJJqxaIYg+`rnDuP#$joQ5x{c8d}arF_m5OdB!kzjtT5
z@~U&G7}&-odngj)QjNU#uFn183`VE>2d4)FlKB775NnNgv%?DJBmKX+LCnWAROr&o
zoOSYkDEj28sXG<=4??_N^);<mIOK6J3bg-d81+dUWmVm<1~|5<=s-F88Qg?=OwJGG
z39UB#uE62P%s$13Ki4DeXRAHkn?*4}-MexNRJ!ANh#2|gW!Ci=DZ_O?1+L==NVWul
zkczJR->r+}rxHcKct|csC6=S1L_!>NJtRjmT!zlfGM6~Ut6CO8_IJ^YrsRr1?fagB
zEG!m7#9PhykH1i!z)unICqujw62AI7akF|5c<1t-K1|*QdlcE-StNlD<5`AlW@e@)
z9j>-spB`MT_7ANk8BeX=&XDSa@anyuYb*qnxR54Hgx43}wNpnM@+#JFxTK`D{sX@`
zbokw=X4!j01z~!x2;S<}N5EsOs^GtUEFA;VZ%j8VYB51pv$IC%5Td8Z!F5;*@pP-(
z3SSXqA>)AcIaZAc5tuCx%G=CIMoJ2Q7fSj-(TNc;PQYX)7)I2Lz~gmDUTiY{PtVfH
zT!=5zyHv)#r8w$r28R4(k*y3L#5Ia@a_;+1=3-~IzZ-TT6nTP&fPQl(r&bTy-^m^V
z{VDPMegS)SHjHL@Ou|Af`w$tM4zM#bk0|x0xtRJY(8v2jwQQ<?{ZeFmy&0T1aqLNk
zRK84HozGt<E-mu;#ZW+DKF$$ktKT_hUGB0O83ma?G2)Y*!?{9Txswc8;d04>i{1V8
zF9rwsP_uu+t(FUo0SV(el7x7Ax)q12KZe*~1<CX&Rtj_uAw9O!Z{N?%CK<rtf#x*>
za54kXq_sE38Xfjn@2iJ*(65K2Alfx~3|?7SS8&s2ZPmCHl{6bLUo8!e@tCi{m}2#G
zdi&s2ndVGTl|?nkM4g21tVM_C<3Wv*G^v>^`hirJ>1@EX{oW#XDsefQ7&*t=4MgNi
zV787(6P)nqp4s`a=Zk*Ja6Z4%L+_T@fQ<gk*F+CAtv;Di;?r$RHD(??kG_xIt|1DI
z;tUcR4b5GU9pG1^%I@77d(rd$&C|KS#VH{YR4UPvQ2Od`h&w;{=f*6zdRKx)Q&V*t
z<6<ZXkbAk?RV9J*Y~j)96O<Y`CXzMBGJb3LR$JC@Ze9~|uLzBaC`i5BHJNR~6i;qa
zC{*t|otr(T4GH|_QASR-T0iL`&i?cp^`LE>b_{ObMpRv3T%kCKTY`s%?`LKuN7&AO
zu!ph4=H})Jh8P4WIDTtrL}RB?E3upC6O1j;p~MII>Z68R=H*pgY<VPs#o!mV{s%&$
zqI*j#_2qxFrm4%lBA#fHDsi&heX%Y9v~_goXY|LK{g>%KZZ;NBgH5QzFW$Jl`?%KO
z$8Q(|u_KAdSVR(V_xJ0U4Xr|<!^)9fp|5U;YnLfeOEoaZKD_-!{2@#7?f%hZ2tZJp
zz~TuQvm7RAZiWtP&OU1|`2gRm=x(NrTtW8`o2)PWC?|QrRROiKID@d<;0}WfW%xAp
zppRgA@76{4Y<2hFTAgguUH4Yq|HxnoU#_b-7Fh`^3=!`X$xc`|G2z{fm~wjwuet)*
z9%F|wV+il-BoW%x><y(a`?~z(00iU$jqwgsT*{<nW==`B$K%LMet>g8E~584*$vaE
z;N*VA5K*7kyOVr9Ggb>>(Uw^-?eDkLhts~Ma!+#KP>5X-4w|M@I=6lU?N`j_r#Mu{
zGesaieHEtMukeN#4Ly!$k=zaM-M4D5^NIlOsYPEN?2`;Mej%;Pg6TNG=Z>3?_per8
zLfMO7wCW+>mKNHlgR1Z}JmoCLBBPo?TuaO^QBU_~n`WA4ts{7??|^>KkJ(uO-ZR%D
zZ+Qbdolv*d#}c5|zgP^C%{DU0>TA6o<7>1A6!w;%l{-~qlv|cI!-6d;WPRc!35^ST
z3<g!$=;(VrUHMCe#=aB49>#V|uK#kUV}TWK9Pkel^>g<e|F7ksWU8l4^9w=tv-4gj
z)-@h~(?4{7Y>?YpZ2kq#mZTPRV$Y$Iw_TNtI-4-OZpNiLd|W-GrqqBW@zMAw3Ufff
z)hnK)Xfg93Vz6vu!xQG!;_d?;7(gXUGE4C(WanC07MId|jvfJcd3WUm@bg)1;sG|S
zV+p1e<|tdROIkp$%5)0fDF0jRXUU4C_Ed(hhw``!`S_G?7DM@uW~lNqpv<4762KJM
z{qt+zch07}RSJ95x(k=<eA3JwHjy4)duoM6_N3D|V2689cw~Ry@8aUDG^uD4kKeky
z44<j(l`YTsFB8ZQzOSmf7<J`$vei#w&*P$dCd$if^+w_?D;m7or<;A-8=H&%?d(5$
z>vcVKCtUO5@4?ijaS`e6=?^GxARw>VlxcLBc9a}lG_#qF1-1S(e##4uVR3gFhbE*8
zX(Q<9im>ZDM`JirG4xBwGLCz2#6D}A2h-k8>=acL1cc0In&fXR?_h8r7kNX5G6+5+
z2{^4L%6NoOFG5sgj1W0<Chfot-^`Z|U-b8q|BwKJr%Tb(p}u5A<Exu%$5K3xTChi#
zH9b9bXv|wX+NsN{T|)ms|MQlKzLGU+$xon_Ie&U>Yi6=Q!pF7_KWDC#grz1cg&Zm&
zC-zZ*fmZziRm?#}*}(;!5t#xK+N-vE@Q&o1`quIq(M$D|k<s6OIkYO2pv550q>T)u
z<&+bWyA{B=l3}G|0!OQBGWx%JWZ{iJXXXbfrtAjsy}fbljwW23?hIY_lg>r~47ww!
zBPnUI-ChF2CYGI-z|4Fwt7@@>X%Hm@bpy7a*RHO9yH~g;?;R6m4U^~0-&^^1o$PL)
z%qwu}1&y<PzMo#Jt!X;4ewvTqjqy+a_?>F(R&oNOq)dIgt7rn<-uxm{YAQxYkw4y}
z?&dSA+4(HD=Q3pMEl!<dQct@@Yi3)}QwyX`eLXO|W7@*Sa(T$=tSETlc`WCVi0?f~
z8+eAJ)vPME{R1Yb&_+JMoU-Y+yz<EJxs+(i0l7n1#b|!cX`<*~&O7^+u~$XM=ingW
z?i%CORzY-tHMuz8gYJp!OhvhQ*EZKibvK9kh-lr(rlETF1soC>IsX5GGjqwk_T}Gu
ztu@iTC^oOw_p?h8zA&i-sQ}-uU!o%XF*%wsX@B&DuspMQk6#AOVy$|vTCa{$H;0DM
zOHuMW&`_%l(w$_Pw`h3++tjqGz*6Bc!mEZ;iv$#|F7c$joV2Ig_!?$c+6D_Q?^Jg5
z2Lf}r<u@Idr0H1(^CYXUyyYVEq<QDB>FM?`?;8|^=HE;`jlTh>(g$y^50<a%I{L4!
zLbLUf_m_y0fb%JeN;pQ^ki2*l74<mWtETqPHLv&eJfh`!;7(syHQ6^6nKIY2AH?fx
zpjLh<idPh`?yrEGqA#vKtXD&X|4MZqko#jY>}1LnM4tR%w5fKTjBn2C^IQJ6lqGNj
z<l~<emzRe=)n*0d1``v;F`4h6FLmOy_Tr$poo;V$*{Z{{sl%3u&1(YYiKA2uh!QE|
zZF9091~@YE4OqXwuTxcS?tY9+agTKX;xi5I87{&oDWLHB8~5jSmX_)+!zYP`SmIoZ
z5QyM$wY7e+brx8Nl<eKg<`drk&M%Ux$|4=rOOAShK#}wH!MWh09~ix8%5`qC4#;Ba
z`5Pc+V#=~ra!?S|eQ1Q65i_uAbB6zzz%tBdjvzaTNxylf0oL=<)04SWe-6Oaf4%g+
zoqsEPdq7?E4eGce<zL6=zkjD{2E^#6GRRo{_tDp-qh+VN?qr8xT%>ya_Xl_U(n(jP
z?H;tn?K>yZf5~u!!>M}q)r&F%31bO@E&F8ohm#0XTgwsCY8J(Uaa}PUm&T<bwp6jf
z>yyXrwB}-H?XT2+A@ITq5g9EYQE7#()Ypbn<=TPCsf;G`>|uEJ7Bwxw)>9;QSDnq>
z^SC)=|67Su{nBv@*2}kl#w_e##+kLDc}1};!Bpmi`v|gW0m!n?!cp|H4W|nQ=<QSS
z`-^58M67`jd=W;L+R4DA36cy6iBHvee*L~`A{Akl0eKOyGH%IML1M?%Lw@K4l*u1C
zsVWbClPs^#`22*2=XcEOr);~4Jii#KRojKCVEr)u79Y?jNf_F+|BvQL*Q@T61@EG|
zo5QiAF7=io$`%@|DNtvnMN9M66@Dj$!h$~?iB=0I-V2^0m2Qe+m83fKCiwsvv!Y55
z@tKYFBmd=)Ih)8ZJKz$yJi;z_M*>}Z_3XC3*MzHR;`t77J$xbJ<in&{ShUjR<(?W^
zxAp|Dc6Ek7I;?d!tJx&|JSMR@HRYi+u`Y}85WjEopv81-+R$u&`FLlzOFfsi2u+!7
ze$tF<K04Ooi02jgivoxX%bu&I3Vnt%59GNOg9Af8#G)+bfLEoE&t(am%!v!i3G7N~
ziW$tG_#iK!uy5YWElbBi%PsbvQQ=pazMmEgZ4}tb9w91koe-~%78ZPg&Z~>F3(WF(
zd0F+zM*%W6AEoKhjWE)?UWDT_+*IFm1;^ubA4FZoDkPo}(garL6G(V9>A3uMufDFi
z4tLzl60SG~0M{3d=Fahw7$Z&?yl<y>q$4cbS%DnFZ}M}_%Z-lT3F*A}Adnjzm2GX1
zVQ>vihjYbj{2+p!mYELoPR{Xq-{MY(fX=L6#z2JBDV#(#U{Z%OWQ;FpoO!-}CjSPe
zmKA+V|9D95r|cPllo|dkwu$H){iW@6Lw{B4`gtIbu;N6shb596nTO#y=<tZJ!cEG!
ze_8(PU1X0*vRY_aFzmbAwF%Wgu}((Za+lw$5$HA7ud+z=-|5wi8L;%cnf^?L*L%JP
zxOKh%=9MT@g~u)MXlk7e5=E+So1G3#SqcaIV&R_=C|;Law=&yi%Z5NJLzCalNx`He
zxmo~yLQ;|N>nd7pJmb^|btFx%!Kssw;?qO7>M2=6s_BkqocPI3bC89K{VH27P0kHQ
z@u8O@20Eb+qrb50Z}V)Vh5uzKxj#R5c!N~5Z<sBzBR<fykC46jwNXg^6WFfneTx&0
z4t9qyaHve`*WU2>DIIjfOfZ1%3B|t<Gfv<gd&BclYBm^ItW?>)ElO3*qHhl!@ov|q
zXZeN{AGoAwLi%ye2oe;Zk{>tfdiSB;4m>j(l9Qe$vgLk*YfLL*8F%U8&eGQ6m^)Ga
zU+VL&SIgb@(U&_I-yWcCS&pK!b2*%Mp}TG&+B~cxthRD0P-(*S>^wTGXUPZqJ`Wfx
zS1v#kq}9d<Q?^D4?w^yLokwiZ0@9L;-fU#Tfl9HI$@P!2Jfg!$j>^M^JtE_goSNxx
zutVX5B9o+y3BB2jJrYID|F7$Cu1xE%$d`PJO1$<eMYcr06y*y6KC&Cqx+PtUvnNrc
zX7Eoe11y+g;>Sd#z-}<8xPNJB3AjERYB4U#>bbs5-VzyyKV4;DQ&gm%qAOEoy0iEq
zErD7@nw?9Nzc;XzpSmFv&QqH!ruf3;_E|9-7uCGMNn(XW<=LJDVcEfbw3XQcCD8e`
zUu<zK+w_plz=Vx^PKLZpM+|P#nPp0&KUA~azN^*r4_NXobb?(X)e^bE0f4hTMiVtc
zm5<YIUiGs^W&UyP61t|PE|JV3DyUZ;@ageG5yR3vLoXH|FfN?zE<7`huM!d}L9tMO
z6a}XVk}+!Dj5J^!{dq+qY}q^XbQKgCqp|<+-*{_d(oH4MTfjHpExaqnXu6TbW6-p!
zTJ`z_?>#y*v3V$MR%Lrf7bqDr|B+EcnRX#D!~0+E<+q#ZI!<4er9)d;vvW|OK7*5u
zXRZ|<q@L_{t@t>X4gM!w=}>ePy!d`LucAI%45*^HW&vGSw`||jv)mlI|NeH%a@v}@
zUfd%8M<;PpJrUdO*`wKW;!>R#UW<R`>5Eh_gsbdD%H^x#Aw(iH2T13wGho!wx~o2~
z$76D6p!C+(ghHJUB*0SAmgRQGY&9kAP(6TcTdV7R)tNN9e1w3IsN9k@EH{5Zb(h|e
z1B8I4q$dMeytka^on;;+1mJC4&jV|bCC&7Bn`LZz=nr_S=zsgNGwis-hk6$uZ_H`!
zyj+dZ9M%2y7h+K!1N2ewoGhqJ$p1jP@KwK%MC(!B-CeH5E?-4nZtgi;Ul8(MUy!+E
zS8;3JgI71@8&|rCz!&F_kAuUAR~yxpN(IIk!qk?saiukD3A+o~KWU0k!L$u)f@>;_
zt+2QAR}eFm0BZ%|?C%+26QOrWY&l>I`YU-$ak!XOhYLvs)YpZ`QySa^^T+ZO{nPt-
z0DQyTdEJv``+1BTb?ZJIovqa!8FV`jxO-*Dlc`@|44bU1Nnt`*@FpJa93B$4&<^%c
zT%@Y8&Tr(f4#G5`#Ew?_Nq?b|w>HxKLE~iqOw@FS$o?~*tvW4HcLk-!8JDzp7RdL{
zs`f$`Tehk|(XXmR2HvMey9_?FM9^@6PubsxF;9?)ekWXhuM}AytVGg949teqwo}PH
zE(y-386DR*w1j*#nT!!rvjC|*@{4@8mhmpU)innOi&NGogS#M3M;8g`GCE=Av`<cE
z6RBA-Wf3WIx%Uh7LJcTPx4Fuw`VTweTzz{rVw8Zs#KTMtWY-CP9Va%wwA}3g;t$C}
zkij(nZk_2`n1$?CRBXG$k(QsqF!M1xl3r<`)zN-~>6X9o>sXx$U;kyB4E3vkYEGuY
zc;!X0X<7KtZnol0Ewktxjp5L6CE8CpK6n}w=&27I+v{zYs2+aDnkfQC|4oK4n^1MR
z0nqeXZfSWBxVSpnIg0v|3pQn3gp{}6X<&fe%Ibsnfu<8f7HYkEb}pChj;UUl40Z5Y
z$OK66cNMQ*?SJN}?n0&HjEW&#^&q`*urEkD^{<1&BmMCft~r&JgsrX4)?m)kowe?x
zE-!h`ja?%OGZACVUukr}8+a_L%?(Zm{CnHL;wfu&pD11Y)~gb8*Yrw^4(QG|>Bs*z
z)ZLfBJ>ZN;B9N>HEF(^I+BZszl5Yoan|n*XWzj~|-uHA9VS62KAgV)3GyEZir-SZ%
zzMmaWiQ@jMGF!VWega$&Flqqtczt1kxbvD(KF&=PLd7(#n>E!bYSwBmfu6Jh_P9mu
zjW##k!j#4;ry8dl%&lmQDa=aTa<T|v_N*diB5s8nZiN-r{at$t$`Y9xeo%ISsN7A5
z=&P{Y4doY?dDzsg-@JX2jiR3m8%xPUvZJyp4~08;(!;7pk#0Y0)@ILlc29^)vt944
zL!k{2KgwKinpMpw9S}WzDr#0+`r;tl0HiK`Cw=kx-c>lxxbmJp)X?m+|2qqmrB2!O
zou;yXZ~_@05J-gPXi(uDbG<&#c!s+wjWXV;I~91l7MmW2W<1*D2{{*7814Op^rSIl
z@J{c=Wn`}0xS|4ohH;ab`b4BaO?h)*?p7)&rjWc0!y9lq(93TnzI%C9MgK@xK+~;|
zol0eaN?eTn&jxedQlvl>yEedOa9C8A0Rd@X%67mOL>><{B>dTp;g9c!_pPWa^L#3g
zF){l+9Pxf(DXE&BS@F}h$xMuT8DRP=TjLC<9uWZDWPk+5#vL0}ZvDx^OZ!_O^0Sh8
z?y$i=zt9<gd%OigEpdbYN7H%6v;F;TJdq%Fj4G`Wp^Bn0TBAnnQALdwrM8w@HESj`
zh*G<D?XA=(F>3pv#Hv{_+nTA-8nNB^Ke!*{dGh+6@5wpWb$zZkQb6dImT9lL*)w9v
zoL|Vr(FM*Lb#>HFJZ;aoq4c}%#7bi;<)`)aQ3yI@V>(dAy49_xMb61WO?>wPQ1<KX
zy^nEMYOyi#Q+@fxayyGme^Hj1RdV$siTsR>FS}PO_atXZ_q~+FCQN3=31e;Vt4nq=
zUpGNjP0|OcKZ~Eeya|2G@tTD%8zX3|&z(FVG42iWXRa=Wa{$>2y6L4ZFDB@hiGVSn
znym@%>=WKnY_hCiy6Hm|fdti+^e8v+hE$^xnskNwZYje_{4g4tfYPRgWkj7J>9u&P
zJ5a5cyW}9?887Pm<nZjGJ)`)|w|LcDm#2f(ET&q7nUK~fE#{20O@-0XA)n`4l0%kn
zL=+^3_4ZBH$W0^8=BWYC)r(Rkrz8znkpn7Rs_I~wiLkc`4w)s$NZyQv^D<iz2m;b0
z95I!uS(WnK!RLQCMhgKun7E_)nq-tHXeWrey^_@6cEd^7z!qEt`wJ+dHKpbMJ0`so
zwH>6OJH^MX$zrSir{}J}VjC_XalbirQ0|q36Vkzf-8RvPt_8Ok^a`>8ztu2miuo~*
zpP%fO;F_PMp9$)Cesnm!beBXVk;rCEfJm-&>eZRYdNk@|SEpuEi!b;4A{ZcFD^H_L
ztII~fK_EU{&sl}yAHY-!!YB9K3@fav^RiXzB^u!8z42CEzdPBazGn?n!&~(PTn|-2
zkWH*!YhMKb7lK-fx4kc3OEzi?8wFGnmi5~ZWc|&y`|74zzok$4hi}02A_x`C0ehBT
zR%$sUh>{gRTTJYF?4mCJ9{S`MtT&tR6<G?8NieQSi?nR!k1L1Pyb-!&j#eljn&HH3
zr~Bc5(=AGti?Q_nOvj_&?!-^=BM{epv^tH3YM%m3WZ;?WC$mt+xF@NyUI4pq{kb$B
zBCxN7Pd;0kxm4}clB-GFmpVeoRk$-G+_NQ3l$k;j)tCgS;P`P4Kg&6BO*V2zZKj%&
z{wClYa(mz)Jc^Y8;I+vZrMA(Ndo+9dO=nNkVJ9E#_i;`J^Q?mDc(!S1&en-EIS-}*
zvny-Jw;Bd;kMUE4CAtXeaiJFT<NT%IZ28f&6Jhj&(ejAm$}`(zNFPN$R3AzI1#+Qt
zuE0D_kYDs~XnWsn@$zoV=fzP&TI!}SU`5eeplbi$?;pss;h8C&U*4}Sh~kY`utt^7
z(o0Z%tv}sJsLPWhBJUE4oc1^jOe4GkDOz;sJ^CaBHxgS_E^Yw=Ao0j}Pn?9NhuZ>a
z+^^;_V<slT@hv+W&jLIax@Ppz>BZx`wKpE^cQWIU0>ULLHxjyd(V~{R=IqrCn#UXq
z(ggU(BL|oRd|_2S&C3-egCZ{<zmV32&oZ95G1;XS(Hsi%Ze?+{sX8^<=;RtLvx$%;
z4Kfq}o&OPAVj*pN>suVXp+BGoZE1aX;x{1ZDK?hA8d}X0MD!)BQ@SLU?ng%?mfe1|
z^bwkF4E_VL-S`XrS(K}&gs)As4+!L_sypJn;_JosUj6sB9FP%|CgcX<VjCrEF#Dwj
zWPZ2~kLADeNy}-gr8F})kPr%`?aK<-y1}d-EdLR17aK}&*U)w3vojhw(US+8-cx9F
zYqb=2ksqBHD!ophTXHh78Jy`94`aSZCzDA22jPB))r%!t788G?C^tX<<UpCHjsYhX
z>^6Fi=jgu++yQ5t`JbN-tSukwvpTM=a0-U+rR~xxg)%s78}JFTC_P)`5*`*1fY~a2
zESbobq9qWj?OkcQ?envL-P9TJD=~Vd`6f~xN(J~$6(^@wN&%dtVx~+WyZ#^?TM}L5
zEAN{E4iTndO`@#)qiBWGmgkF_9-FRBQ%@W&2x%7XAQ9B1@5TOEU|W1ieiS<pzN)hv
zCg-3j=pdj;-`hN0Jo&%hqrWdb9D(5%qDKSi$D`lJ)D4Vp!}%R_8JgTmj*&84AQpB_
z3aNcyFE!?9sXf%kCs{v=t4qt`-DwTNn<+nv3!L5k!<$7G@D?b5T^M74Lo)`$%_vmY
z%+b22LpK7>jkoaqin<{DuE9Fqspa|73dDdqxF)-ggV4UTLv;JN+q!W7?1yjzXNL9Q
zK+uwJ>~Y$VVq=JFtCF(vbwMA5y$>yV+|&Bk(c+Q^*-Z@zY9)ujTdfw`;?pK1r&O_p
zm0^*UjN26zn%a=Wb7Gr6{fltG=nrWT7!}GE!mHe99J}J2{qQ;1&V(=KC(tF8={^f}
zFQQ{<CSSw|po4Wu)OwmT5r9^&Wzje~h%({7q>TO7B(XB~y+2XHS5*5iKW_T1(W1%*
zBYO>#Khr&HORNGM|EKA&?m9jRqs<gq9TdyH>w()P2iujEl@&k6XlFQNycoIFu#T#R
z<4rJ$CuJ}7GJLzp^dE&=ygV;cK2KqmO)~d5P!A-hM7ovzVlVP5NpYAD(R9GPw_S|P
z?kgMVAl%t@iAx(9bC9lK+8WpEqQB4OF5ueSHPDgHP3QSA*XPqN@$fJ(q-|@AZCasS
znjRT5hE-q=72;18pJUL>0|H_EsJ&Y8VbzbXf=*}CdTd+dcjD1`)DiMQVTB_vzUr+7
zC_<)yK%~0!k{X34ZZR`XJ=dY`(Z5R8&iAX!I|vZjnFPeMDWs})r#}i-O68%_M%z}b
z=t9Ad)6-#=%nDrnhTtSmmXgkH>1#huU!H{qv^spCglH<{I%XC@XriI0%2zx)oPD0V
z#KuN$=F;+<)Ci~>y*Iyrjb4Ohd)U$aYT}!=oisvImHkr7ZE8*gQ(k%C(8nWZ^PG{=
zfQpiZD)_Qxn$hjVaC)j9)2Xh!f&s}~^4T*uGk5LZ7yQ17QCdQlcSPzxOb;cuAy2#}
z#l>IiFO63thlLS5ku6!Mz~6=+Ce;+@73Nih?9mcZG2;tt*^K-ZiMu^Vj|kNjVZmHu
z3FKauFK8v7rK}1H4C3S3$;AGT2AuCB+yBBIgRyu$4*P{h-<vqoySVnlH(qu}e<U+b
zW?d^7A6^Mi*bv)fZe>=0<<S8+$VAA^8IoVKD>+WNy?u0XczT$2`5GhzY)_N)0tup?
z41E?GE!)`>iaL3R1bd!uY*Kvdk~kdCXjnyXsn%Dd=K%#2;W7XHQc}wU)g7jH3;lV8
z>X@%y`h0fvBbVHOKUjxf$q94mRxn5wxhCBEGhZ7J^_xFVR20q!7kK>&(B>~F=tQB}
z$rW$7_Lb(UbTGJg;5vPM91xpvQ0$a8;{_52j;2w0lfTO)%L*^A$&thPnDe^B-3Xau
zX!i90WB~kw<p3sJ2m`HdbB>oVVQNkU81cXpSsp)%NRWWxr>G0=uow#~sMu1J{MuUv
zyrzH6+?inrHL6YB$;f`vP|fV1?^*<8lbtfM*sZb>=RZcuNjyG|w#n4(5;}>5sPj>U
zlM$?EW2mnFxn+pLSbox*AOL(#b!XOR-VbHH9k`1-Yu|0<75HF(_@7RwY8gED`SYBS
z+srpUns8@h>5>3ah_do$c|g-XNrTbck-G-DBm1SzBM}_Iv5Yy7>Vqm4T=?ML2&QOl
zq-o#*<JG^8kPEYX5`UL@{X3nN!d7~&5?jG^Xqt!*2!`VI8m3;LZh2bZSbe@<N;vbi
zIx~#QGr_4UZLPU4wzMzJ&3NpQa&t%PkqzBpVQfz@Ae(2nX)WdM#BU5*<S4?;Lc_fK
z40IE-q}LtsBtmHOVn;i<Z|7*648p4ro$$(hBR;^<#x#+idh+)jhei`iaUUv4RbWa5
z7?uc0LphF6<E7lWJ~ExF#pd^SZRB$~hU>Cu1EBq32@0YnJyyf%9Ds>piZ@{U01n=%
zw{Zi1dtgjI&iN=<f7qJR1b@M3l^-XJhNRa3IQZ`IFxn#dp$znjt#rZK_u?ilRn&XA
z&b8O#fMMZ&OmynH+AZZS)c(08EqHH?R7@mM@M0}Y?JyMex%j{lj0`_7OGx>-Gv7Q(
z=0xSGmkd$@O%5B~ujJC)eL?q7huas2Lu8%yY-Wf+$)nbQ)Z=u7xr8<%0V$&po6xQ@
z-~zIqDF3hU<oE)qondw3cy{@Fe$;;{p%S;}Bhxok40LNp9vewg;+P%|GqKV>=3~Qr
z@o>?v15pSkP>hpXDcFD`Fe$oww@r{*KHbqVP)YvMKT&|hF!0}mFUF{pYzI?Yxj$A8
zD-}C!BxZOQ*8odr@}9{_Nd3@RRaT%n`3}$mGlo$#hv{(*fIQ3mZW0=qwi|fxXA<&j
z`_7u8c!#wC^-ea0;3fG_L6)CM`0q8Ajsp0tm#$yg$)fXBZpW1o#$w4A!4*f%*b7T(
z4vYfWt!y5gA8eN9qEIKRjv3Dd19H;8o^c>7lumjgpMjZS(0aMDi1gYA0kHMj6VSsm
z4=ZKE4RRHfn;KJs$Bj#`&mKEUU*3PTq|3jCruc1R4*+a>x~h%06qci(gx@#GTEscc
zWz@3Urj9>qi^|AS+xS=?wmN!ul!J6Or+L3Ku|Cizemc;j1+{y6sh{==rSvFvV=YoY
zz1(#pk1#hi|2ER)*|XLL9uP$jvUPKBal3s77ZlX7n?~FwFT#atrM}>bQR56b4p9Y@
z1}1J7m+Cdl3bMB|PFu(bk_bAoy<zKaM9Zi7y_FA=&&J~f=%+o9zyRrOz{o*c;tG%U
zlzme`o;(62`CY{`0z#%Y<^f;>d)@%_q;T=%Ev9*N7c=7ydpzHR&Mw6;tweS8%VTRT
zDg{lj%a#|UepMG{8J7g~HuO7oR~E{2YFwPgd5~3M(rB*rc`|Vf%P(+`Guj;rh6qfk
zGcpb@kfs&5pr#)(M%P(?)&{?ceb>`nnKE%ggJPa`+nx(uPQA*ft4;a2AjKo#7-_1z
z(v4iv$G{zPUYNeBYM2h(<>|t+^U&_lGH=AKC`D#Tc$Z&`Ua)Ul=pbJga<9&aIOUFN
z(v)KPXXoEt$ODCFAl%ShcDvRqTCWuhfW*SyaUCnL!G4UO(hcg|WCO>e;x}d+XcKRv
z2880o7=TWxMWvZw2s5V&0@uq1T(hA{O|XU!84GY3y8zfAgIWK1E7HYDtoGk7-}53D
zCfe0v1v==wh2dZdU;<@i%bcNil)FGlM6HQKAN;FHCN2E<@Io=<)_s%KI3<glLrKgB
z*ocJ3R#6O{Wt2VLk&SLI)MXyjE%2k@-Qi@qr_<71@(HM;%8fu6dyR2j-Dl<8z6L+|
z7ks*!mb)gRX8hQhS=7j2ecvmSD4<;pgy4%|I!$jG{aDZxEB5h?fdq3}n|}}E?p*ik
z2|MkSta%Bg0lcv3{NFCwJ$ZPMTA|?J#WeF;&(AdEps_mVv2gtxjhoE!P*iv{cg=&a
zLa@oZ=H<EJh6XpQnFr{!?g?qf{u?$GYqVfPb^RgoY!%Va^nFRDqB8)XVe@K^JbOWK
z_ei^N7EZ?Z^&t=`P{0_s)<ajwTy;zN_7?4ZJ9^%Q>8hofi+}CUcM03=<mSA_C9AW5
zq~by6pQ^xWr<{Xd)z$2oD3&`E5~6x&Q8-!~*n>13hP&L@gHA!X02s&g`mqiZ4FSkN
z8=DPPh(c;I5{qQZVatZm_6FbLdh+_g0mn(x>dComWvNm;&BmBBW;IsuY3Xr(<<K`J
zxMIoO6MGe#ItFeo31z)*{sN)eO!E-#v6b1h63?C~+*#;}-^mfGA2*tts~i%#SOM*N
zYs<r4duWvhr*jZ@Jd`x$=6}3fwJp4)+r}VO8g2b(N-wWStd&0V{|)9IHVORI+el$u
z5pzrhEnRf>nuc~~;Wh(vL_h`W&zJE-Ky`4+ca5hQ&d%YuGVhyGquND<9oFt-spMb(
zXO!jA;u%VbWeST;&>=vgJP&FZxY=}MBiE>-b70Uv67FHY#Ihyq_R_fa6Fzb8q|Vnl
zCCBM;q#|{X6YN;VY=N46AQy_Hw^CqlS@y){c`YuNoY+^XmOs+wt0VcxZ7X_g`pQ!+
z<kS`=r{c|B%giy<*UP&<Fe^^tdnRjNec@Sy{Y3qAqEA=2htYW{ZW-aeOWN(}+go%a
z936&+24?C%{J0MP(1f6+l9$xJbtlYynbBP%ohhZXGD7YE>$^XLyF6{QswT8FPPbNs
z3~>C5$Pk+`DJ8qhxz;(|4djl<AYPT;h-Q4l2<vs15MF&odLj4fkeO{ByMb%bg-dN!
zWhe$`K|RN3gUNYJ-0W1drZLX|>2d{pmxgLrYTSywtEbN&W|W8sRNt7Rv!~{rwv`^s
zZ@Xta^&NbNGeUxr74B67{V=FN!^X&0ia<;)DvF-x`5`7YlqSy*=!80cRmbNpQ{}Ck
zW%1OxceqTgqKKX{O_tNf|Lv;}{f^+Wnn2;BR9K%1M8oLHY#Dc$7eZoZXa1gET#&^P
z;;LWUXR>XY&mn(Z@2h2!^F)PI-x&RKKn)Cyd~25<Mt6-F2#^_73;FJYmeN+5XM>yd
zf}%8Jh`-ESDb!&QLlPt=e>y6YTG>cw6WW<i?U^~ti~;-<Qj5m++w(9rlkWMWoNS6<
zHv6tWl|IQF&u%J;0O7w?Z=EksQ*d`+C=5Y>aH$RnG;c9T(o=8={P%jukO_{=#z37P
zoqGkI9AMatIQa1IvBQ5Zj-Ho9>u#z&{Am7r-$42OG`wIsP(Bx{jpeJ|-MK&6*~)Nl
z*fY;_>4^M>^FHXm?0m%A)<BJF6*N{XaWv;}{*w42ZgM1u9C4l&F^Y7)*`vt>5QOGb
zAXKKtE43*O9<^1Sk;tA&``7l6jw4(4@>z#t+3)|UiLEa4WD2hR`7i!cB9i1l$D+&A
z7`JrD;lzzS`9(toK+1aEduPKeuQwFW;s@4a)>I$-*d*D)@`msCFLqtjy}O3@gE>c}
zT3V%mT>?zyvdX`+Iny3<8ju6wnRJD5PEc}rHjMz&J!*Qkq~4R-Je=*yf(`GuG#mi|
zfYhk@jG-S?UyB6i46btT5E}1U;m9t>>ABbCL2zh*Mb3!#$0ide6^Deu2{@$=4J2gU
zEY+eCMnyrH^j128uz+f}o;#eIb~~tUdmsP%;Q;AQ3u45WoIE~R-NZw11ag@pVTz)-
zsd>Atp|*0<@IBJDq{eAn=Rkch1i3jPal_&T;-9TeKGbCRv73(+9r!1}Va>V1Z=;JL
z^moox{y4*d5-glX4TA4Z>Eev-*4#1~CX+`bAYq+gH*p^eREb4IbXYWyRzpSWl@0el
z8#*C@A%!#xc<|J?Pm?AfwZt^V37Q`uZhxR{%<lP4%Tm$4V8P}YiFm8s^mwy4TTo;m
z%6k&j4S@iE)KXDv`%knjh~qpwcCDY^wZ1!pBl#`_8B39L5}NcTOkRm;n8^;C?4-6$
zskI}iX2ss2S^}}_n;e*CaNu_~IIox=*5@7{2Eo1-EXdoXTf{ilGKcn*+TH?Gh?+XE
ze2aeA(g#IDWIKC?feD~a#jFQ5>L^!K{U4SBM4+b<(FyetDF<=@83(*)`lc_w@!Muo
z&JGxr$CPM|L{m#5tJ&*~pmdt(tPG&~?k{p?05wWBtwQC*)}1qR)O%n4IcuyYyPjg<
z_^2O<7wm#Me>`7p`^by!OAfR09WD~K`?{%P<nnMOA-}&~M*XH%_Mz<DKFTp<D8TIf
zX-$Fu+-W{vrsuf<@*eqR3v?G-t-xf5lfE)ji~}RL4!4g8%Ar?>C~{n3_3UCX5C=L*
z3rB-EWf!-b8RbL00G&N!|7I2$f$4@|4Wz7ZMU-}A_;S8wm?hQk-AGvMe4HH*oLf1i
ziE-8xCxPD&Nt-}}?xKPZOj^BCR52#+M2*F|+lk~p*5eRf+?ik7V(8`hIc{_`H)qr*
za|ll_r2i8?@*#r*dQV!D9u3JuQmqjxJZ6qgL#$-nhM(PR%i6SW117F#H4jWk=B8yp
zab(7A<F0)fDA9tnVA1iy{0Ln>Jt56UMllr>-uJkl;HN>4q9j@7zcM6_yT&5vD8$nh
z7VXJxMeJ+TyC=Luu~QOIOK6J3*UF>4uw{<W-y1FQ`E&F0l90dBItgf-5Qr3{Ex}mm
z6kXH9qt^c}l#GV#5v3Ssx*iFk{b^G4l+mXS0R#0GwUNdN96f<=4;j-~u8q2~UQ?yl
zW-bF_;QvrB<>$uk&&}=ZlIN`cZ4(IO(CcAib<k4ODpz&UUtLD<qWy7(MLdY+XS{JE
zC4H2;c+QPNs4cZL(;niUrj5evd-v1<q@Vx<WYgEoF6@|I>BtpT`Ww%+aE*fQq5F@x
zf><m-FQ5@SX%99War$?0l-{L2@*?ALTmz4~emJU5leesMkN?e%VKCR6getfnfcMy1
zv7|#%GsSDyA4teE!Sz4Kg2K-@$hI%;?q`G1QtQjl*81y?Ikz><c5%~Z_8F`0N?-UG
zSCTTA3rQAtSo9~<wVl>>xSU_zj^8N6wYT1sdMJEVZoykyjWa4Wy2DK5C<F$I?kAZn
z{zJ`#8JOO?=D#w21_0#Ii@YJD7>qe)=8lh656QjcGvt`_zau}wGx0THmJZ4vI={C+
zB!Z)UCPhOt&|VK%d9%Ut(kW4GR4}^NaWhG6AWxx#h;oF}2jxjcAlCdymU)JGB}~0H
z7A(jXj=U;}9Sr!YR;DH6(F2u22tvb|QG)psmUPMGndMphkqG|q*5VJD=@|nGdSasK
zWwh6s4`MCVeK*)2^@N3w#WPX@y771f%-r0l)WSuB%PQ)q-JkG`&^R0^__7ry2G?i;
zq%E@;A!V6;H+vZjl%n&ktGCZc+krnUytx;%wh#~4_ElEEk2BUpN&`gTF74Z>5UXc5
zU4TXp(HWbK4P;k|l_tj^21@<3#2krKF(u7v{^n<gAY8$2ij6y49KDjcw9Kk1Sn2+j
za_;AKyiJKHJzuFlMn10g@Gzzs_xtl~&;;_RY!CcM=}2jgBfMsKM<{X?Y$;OyAo4nG
zz70LaZ4AJxXE9qbl4z>~Q&kdYpN))$(Izb`JS=yO<M*5yLwmleB2zi>zNzw&pK+n~
z%vzbcB>XQa?o^Wx|9hPn+uA;-5sr@9pD!(?{f7a)2AZeIM1UkwyCm}KeDaTwJ7iQU
z0=IO@Sm!h^aRe6FRx%mi42mr`vF)Oj6Uo?ea5{LIU94SBd4haW<dcFLQ~0-Q4a~Dg
zy66G=Hf$Fg8OWitg9DS&7IGIKX>o^m5o#UubWO5_Y}*bw^L4JgQ^)^Yfh*+dPkjA)
z`S-M+cj@x)QU|In1hv#yeFLELVV!Bg=khP(@``hShVJCT%C|4tQc|IYxA{FlvVzC`
z=KqdKvcXdk&9oAreEUmaf6lEH!B}j+@>aqa)!T~?LP=^jnWX9=bQ~2fq4=ZZflYCY
z_lLpeafuoCIhx011!pfAI3C(+de)EolEL(Y3Z~_zr`0H{g|5hEc@6-AKmF*jHrIHU
z#4jeAq$m8vlg=fQqCRHnm91GK8uBthVQ(sm@xH3nLO)|^<-f!~N$c)*S<&XVnYkp(
zW2Y@um0JLr$;4B?Q8$`=DfNPR7h#_9xu(x&Ei%4Z1%t(Gv{;EwlOW?nh%qDUzi2!5
zbOETSOepy>Zk|{8((ow<|6l!kO=t+(7(?VZV@EnEO@_X-Jb61}=079RNh9i-?#KPe
zIB~g}XUhh8Lze!C*~yQ$(fLngE`81j2Zsv}a|9VeBPhtl?$YSa@W3qH#pTNYU!>DM
z9lbP-eneP8@nCSvt{?B>$69TvpryuupXC#ZnqjEn#?Ti<p3>gjg-j1CONGf~&aw|{
z%?iqQyR>hn<=4e?PYvJEa8wvQJozHpu<^2OyUpL*c(~l0#$St8Tc<4Q&bYJ$Ygkmo
zgD~dwjB+)O(%<jLG7@N9#J&gZ^3Qi)Q>A--jfEWtdGJpIDglV%T>pBlQxsQ_VGfDt
zy7$xv6@~kcsgj=32ooY(20$ZN(DVSc_%<P>wR<_nm%l0_O@<xqofFjxnY?IWrImb9
zC!W|K1ui~vfxoSLPw-MdfkZ%E&3q=KDU~}E%5S@IMRCpoZ~HPyeWo-kN;T4DUOO1(
z@vy={{_V*oaZ0|W+ooRspzLFtBR3EG@EDJpFj4%}@TB<9Sx(vMxV$DC`#rjSxR=t$
z1RIZ=>InnVrWi!j5Mlf3<mzJd>T*VTsl5Xik}eOXP3xST>iks|!)~*4gW3eZXl>lo
z?>VstHhA$O#MPN8!hcb&AGI<*bG-yE8jguqXzKa5pLi|kr6)-85Q>4<fA=!8=UGFg
zs_Vs!#rlr^c+B;gA(Fn$j)Kf!U|t~uhDzbx>gg4_7|4*%dOloL8&F+BWeb)6QvM7;
zbM^~cNyV{}my9KYn}j{f;urPqO?K1An`%PCPRwv;y?_1kRi}?u&%fY@deews^okSa
zARt|_>@oMti;$2G_p|AK^f)<eg(A33<wG8O)v+OMO3l}F%sjcuzveTN!AbAFDa;t@
znjKhC%Tu6lnx!qPS~js!@Cj%=oj-f|sI?{EEP)eNpv?!b342bNJudMp0ew2)eZ@52
zr~G6if*A<M)9d*j)b6&%RaeM=dd88*|A3uxw<PGPhu<qD3ax~~NgdHuK9s12875uA
znE7a$?|-(C1VZ}nzqs&(zcO%9RbP?Pb#JQz##uIMD74BcOHO8C_pHf7=3Tf_3>2L;
z$6-sgtd1}hj!1v#KIo9-T)yAZ@%d3YrfX7g#pjcv*uKeYH~%Z-$4F{~ghpMKUcAF<
zr$q!u1VZ6t*Kfc^QuRg`s$Lre05h+3Flu7TvuHv55Ej0$J8!tMWWZAOMZNIwAAhy&
zp;T%RN-#&li{G^mu1w=;;*P|X1lS_3l017m^b~nQxOhT}m17^XE!#lZ_?Vq@DA(AW
zZ}#$0sq0#Lwfj=Y!ti-atxKdOGTwUASx)hzLj_vf$m=MX!L`>41hwY@TGAquN<?Al
z!q*p^))eR3;{`gT0>n!$jeU8bfBFC9B)SL>Nva_|S}Qf1YLMuP0ySGkf5A+m^b8Fr
ze_T#R`XjftDi7=Ss~K`s`s#Rn=5TSo^WqIxnp_Nsk0O{Q%`yfg09;B7ivisShNLyc
zfD%v!z6TJ*qgz*zP*dNs&(A#Eb_-jn0rE7o`G%h4LH)_dfSWh6C<-TYFRkYmU3KF4
zzp+ma;}`fPSpi)YJ6m_YliV(^-tPE3efsqJQ$WUm**xASgZiF<kpoW&OwwlsM3a=5
z$lO@QqZlJ9`hk}9rbJ8v!Y{`2seiXaaDG1uOr-!#c+j<hkR42Fc0}5&Pdq7mlEywM
zI`>ir^LmMpX7FN<!L{pnZ)d^cp}uE&S&e9=Ym6uz>%<LDquq*l($QhdXY0=mdkqPd
zZW@c^IgAma(L>rD$^TyPf}qh&li&LgHw%+%V9C^rBp&E(aVXFh(-jeGb`9@VmcfW}
zlJMQG2*Z4tymj%JMGx$qZx%sP>ZBLeP7wN=DlOU*dkM2s4mK3HxChUcm>d?Pe8zd7
zIt-Ayct~>udExMi-aF<E>&EhKub+p*37cy1){~xZRr*@`Ui5TKAZjRmS5Je;nqwRq
zutrry9NJ$k<yBVv8WebS(Ld3;-)Mpwa+%~thOobrn6vraU93e7aGGm$Gf(aB@1H)J
ze%3)6+Md(GLG0eCPh2)7^VrZ2L%ZrePCNdXZrBdmZNrt=rx?Xx->YaxaBA2^dp^{@
zSXogSmb_E7dqE~jo)E512xN)7bmsHZDOa4VeFfuV<G{YIJr3p@i7s^Sz@w-IcqGBI
z-e>1Xj)q$MEO}`*MN5uJy2naK$oB^=6yGV}{fA=j=syivxTJ<{8GUTW%D)}N86=x8
zq}*9UJFyiGuQ$<Dfk;K*Z+Lb)hdIkjr)PJL4<|xsCcb@uz;54Q-Ws1GRl98%;(|Zc
z=$YIG1OCxWWU0Fn8*{c+t5M|R$hKyjfW~yOLU!7|1+U~64{~`^s)kAbMQH2dkwq#Q
zlPp?V^@*-d_OXAH&%({J0&0bh=k3qM!wg-@dn<d-_O~uVLmG@AG?V9zq#$o!h@FjJ
zeg1Td>cK^|JREmj4bpG){kUs*0i+kQba*nCV@)E4{Pt~f;LZdrqHK3o<@FvUHx+J<
zjdwURpp0c@bMy%*YJutUuprNyugSbdMPp(7H!jUFz<o9iLtL8Txg+ikBRL!Eq(X(k
zH%$9dvLS4iDYThuaEe|WzZg{_^JuF;_K5Hr%#YBZ;;A-uo0oew|LP7bT8O@})q(z2
z-7S3<X%<5kLoE;#D-);@Nry1S*-`<D_fEbzRX$eNB*yGD{yyi$1}J_cE5M!qW#0F{
z1Hgl!93vP^(ONfGCz{eG@K>>0c~DtJh8FA>eiD?mU6#e!J;3eAUb6hoed9*wGEK_u
zu73$%snXZx#e_E1<(NY3zo~$aiyraYK#kB*)J=1IHE@)H_1Ay^wNy@;4OJWRScRmK
zxY!W6dsga(s4-}Mp8q6N`WLmJZ0da%M!*VtAz^=grsHac$=B-YaOvuBv&Yy<#-BGt
zUD`HZmxlPFodzBmKp=^V!jy%kBq9w|O8C8cdDp!`fMs@?QyOD2N2{lTjAS3q7vZe|
z>sIPBM8!y$Rlvkejr^fuYl(U5#Qnnw1MAR>74KFiDZC~m?AHKaEd*Hypqzv}ii&j>
zkEF43#zgdIznf4GfW{47FX^Oxk<A7Wc%bXWkdbvWe7(|0a_twxJRRI-@Bo)S0s$od
zr>xNek$W9_Hsn2i!_Hz(gd-|hqmQACOkx6~xIx_dbgU42ZRzOO&*eCT_GKCZzOO<h
zyUjV`05^-WXC3H;sJedzTpV9A10o;9e0mlVIGdwS{N*g?L4JbWCh3#(iIZdWJCA@b
z)l1jdU-PzP$t*wfd~0Xv4_TbvRSt%E&~5a!(@5V4xh-tR7<9AB{3&ma_y|B-?N^{1
z>f={jz7wDe+%RUL0qL@uit5}5Y~3X*7`eF~pK&GZ4V~<I-~tKJtpnEj>9akKwYDs(
zKddb$JtC(Q&KeF7uAV;6qMgo#<;8Py#T`CgFi%oiGXeMjLpeXSeLu2UbcWRBXhlyq
zGs~lkEjRXDX#tOMoSu`z+5V=GfmD>jWZNACl`R{eE#-E$Gd<Vvp8ijQh$`&6?$jb-
zv2ts%YPvBcaB;dStwFNzGu`J(fgJi3ILiPhFz^W3ssHT;Mk$cD+oZ^y9la*XSQIeB
zZmTJv+rZZHVjaDc?xrMJ2ZbeN9*t`~X-QsN=An7u@YDqhec5AqZN7v{gdW8Onz(lD
zdxG*|H8Zr_KJzUiA!9L%QoRll_gYN<{MP(17Y!gIF&oC0{A2ak%?tUi%@o}c<)Jn1
z&SOJ|TT}OF;=U~=y3%ck<w-#);k9#3Ul;dE4eHpSv;|TAFSNa{6<Ki5Vxs&U)g%90
zG;FgwIe(ZsRqFH1O=X&UQ8$UoDj((j%cW{3fZuyX1(4z`BrU$O3Z|W8iHpqUoDor}
zSH5lMziY363OxBYv`-u|jRFZ+^BB#3+Q}@>0I$7d%(NfPZlr+cwVI<<BileFnwMkr
zthRk{AfTu_HbDr2qN~;23ZbfPt928akP@8CS?1Z3uFAG<*J>f~R-oj=zqSE_=jzjq
zxKj9yrS2DZBI*(QNU&7FGFGp~%U`C~tjzG#TJ8Q94rIU#@qAT+LUI@#V04*?HEUuk
zr%MaRGCz+Pd3NssgMYc>LTc&$k*5VuV;vj2ogi3%9!Vu$wnBEiW;fQVh)sXC-Sdrt
z%w6UFI!F7KH{^`oM)7=Aw|@Msl{p0tm7OHGhs+wzoV)S?zP;`9eDChWtPjSU(LMbp
zC;PzA_)XjUyLWL0*5adSZNBz`Fp*9b76Ha0MyqQU^`7j;Af`zGw^Icw#fBDYKpZpz
zzFg8GM%F#^eEKJn|5&MSaf&Ok8#FbZ!1V*BfSih|3=X^}isB{GyfN2*D7XIuVOTb~
zeZGt)h1_17`F!^*RLWT9p_F06O9<e3j6|5BjR$t%!(3Yj{ASqQ2GMIBXJd1VkssRb
z`MQHvp2V1xbQ?xAHe~BJu!(eQG(7<v_(VAhH^`?q$K5~OAHwc+_PbfB($5-ZKctkx
zm|<!WfFy_X+bI5mfG_SE03(PJ(O+um*-W(r8`g;FbH$oYgcAMSnc)<(t+2U?H?>Br
zCvfl&67`T$_gHCx1_*euX!e|$HC+o6%e8-?Bkc{++vAmO^NhbH6;TouC^#BIk9u%M
zL(V}g2rYgG@&OKD-h!_mY|3lB8?#;bHRtz!{m%DMA=FPSYNZ2V|Ay>^!19i|9B;gR
zTUpNdlzU-;3~geB;J>&(wE1&ZtrXfCafx0xW98e?pao}g@|C;n7B)OdYZr}q5i17Z
zd(Esj4yvat3%IuVh;Vqh*-zHq+6C9tW!0Vzt1a{k)T%c<vpLE(=UGdq^eUn7ka{ny
zt6#@vi(<LhoGl;FyLHDZH<W{!nGr!#+fO_Anhq&YDZ&un$&5yKc{3(Z7(oo|6y>=}
zgP#Md-3Oy)s<YnbTh(1s!6m)DweIorY`E0M18q%f#BV)%FNqrgO{y{AZ+DrbL$E4!
zc>^4xi3+Q)%%3E(gq5thV+`*+t?;#KobY>2L>bYLOJ5KlVnkQUZrq?3U9OG42qbur
z+73gX5<q^=uc^w3PaH4yz!l4V6s~dizijV3(QVsp++7UR{FEN4GWZgk4u)Tdd~as2
z$a7eP(hSjBW}T+ZT=b8UNkNvo!=Syd3C4Wp4vHTdl4>+x5+9kKnx)DnPOhfcqJHUK
zq8i^_{tXta2GY+qt`L<5y%>VsK-I19VH!V7LDCe%mc!f>G(F>CpuZ<dd%o*cpV#+W
zs*TbBsy}^v?vd(D{FOhZ5?6ngRPBa%RK157g6U_eA#zv==Gw%KW-n0axF4~Kd|JEz
z=1pu44R6>DblEcQerF_9jLu`Gj_f?4(nLz#=z1}xf-tcNfZkb^Z%9DfpAd6)+weVG
z+lj{VBGf@5Jrt5)<pOT<hC~8Z#rZWw!b(cio;|TmCUi!`|1ljpPS9H&;+O8Z;))N^
z04RN|<TqJR>{cmxbhm_XjrGlMBK3KrNQvFbaNectt#I<_4=g|9Q85$Vm_Hvyp`E#k
z*9G&JZZ0#RW$XDslj}4N35Cr*BHDs7>+b_Q4)1ofyRL8l%2k?fpz7s#D?+D=p|Mxl
zThy$$0UI$eK#yrjNiml@TBY8Dy9c)}x>p3A3!(;+Ri6Naj7+u<iFaVnpIeg}`R&Rn
zp}D!aJ+9^CE|JGBl=>jKJP1nDD}u0p7<EMk_l;UtfA;52r23pH9?Hyqc{Ek}v{+46
z&q9$R@`a9KZIWlMraDBu6aVd1r{3NzqyQXBMGq-JWHpet&*fM_J%1a(khANc;LEwc
zwk~&*E0>}wDP`F+I~ll!p9`wAC(5!q(F-v7803htwoscbSG^CC0izID5wmwtnuIqA
zEldx06E>{+4b#O3JGB#D-^8a87Vs3~iPxGqr6;CO5B&%P4)&CzbR$4{mI_G)TJl3d
z3mGVIVB3@T?2p(tf%_&mI}_}RmmBAdG&kC<n;X2-^O_GOcJ7~Y%tbEvwQes`n|W!j
znPjOC8qL5XXdY1?+n{P1kme#}URMZm=5^}=>iFPvX#YZf0L}3Y8EMS4qkqvN6$z2F
zt#Y)&Qc^(7Y3ozU);e;_E>KEKDV#Bc7M7>f)18jTE$>@C8g4;a{H1zVnqJ6SB4y2J
zc@cJV`D3PUchsu4N11j<$yG%sdfKwi-@p69^7s5pCjwiQPzTfQIO1wx^vo;Gzc-c<
ze<Hgb*c}8{;)hBGX7Zg1TWd@dKHBdFdMwWA$qs)}^rav~EuPMb+%=tsKLP^qAgp?k
zwk;e6^a8bQ|C*-y9%ALf90_4a*|gAx<S3emO!e>e{;a}s20`=q1VbXD!JS7YhHqe{
z9r8c}pYIeX(XsU}K^8hE4wH_GX|Qh&|CPz{JozB~LvsH&CE$&HCzoXTxr}j-(%mfk
zxX}OipOZRqH_={a25)=5rK_V0$@P4J>F<M|pwh2*vgxg=kMeVXx>%ps324ffp~@D-
z@6O!q_^N<9J-z4%7Jca4&JUETH&S7N&=0Ai+TC9X+qL1QO?<7c7bMp}t~>edXV70p
z4cjG`C1oh#i1)bPxRY8lKJttbHou5I^cV)xcXH+<p#MS8odB;lA<dVk6YIH`>tu-G
zbJDY?hn~{?qAOyL+WFv4pBlAh1teY{y$XwdQw03pl|uo0XQ^6jDF0Ba*EU|7Unw*S
z^_%D8lE*Du^fNBDGt7L6#A`-PTez57kHWe_3d}Ug(v89&shzB)!lo^87~6a*$Y@_;
zeEE<()L85pSyzJ4ZHwRFKdn)UjFJE{OHw%G$<U;EdGsjIzmyFm$GeAJJ+?zzfs#6W
z!UI2nRj+s#X*gAU5Ay9_KJ_4{1>Mq78cEMfRw^~<nXiXU(BP&rzU<EJH0(Zm-ZkBD
z&(y7lv_{&ovX=!@NPI9kO|6?eB$EM@KBijewZ3PZiwLgljbIpTh&sAF|93hxbg^1Z
zo8*Shk6pHUn>ja2)5b0We+HG3MHtzN`<AO4TOW=pFKK=jEk;}iiENQ!1B5aZAtO!9
z%#R_8V<)7pKG7yp*g(YN=L~X&`Hl1LmFU;(#wdRexi45K1K#aHm4QKg8#%OfNF+W@
z`aT89VrYP9Nai6C{*7jU!4^PT8sUP<)?F<4hrTDzX<$TK*bNHx$jJSz>ftX-{fKA7
zQJNFzZ{|^s=8gAKEjTlZlGIR{<J;TM--(iI8&%N)EJBua_Rzymx4Fj)<4ZygBy`2>
z#J6%7M2iBFXPdUu+8QL%n^Wtszw<S)0B0*?w2EW|!lmvT8|GWjLULt9D2A7EFBUks
z9ivXuU@&Oa`>96?DFWeR`uzj@#>Q)z^EPnFIfo9|NxuR}Jjp#FnWI*REt>;$^<H*y
zb@w1>fhyp>P7`Q7{<GY?O-Fnh8GLq0o@{GB+c{rZ32l9+U4Bhs;3L1R5UgmRKYod`
z`!JM1I^*~?93n!)Q@(dqa__u~39j+Ri`DxP`5L>y&w<<T>1i$XHD_LK-(`Qjd0}e~
zK@NWdFxd1OT<z~Kp{`Dqm9LJ64(muIB`d(Syl4%)Go7s7gMQ4~#0_Zx0|4M1n1Qhe
z+Z5?7KbYM?vrXKV-(QI1svRh2_#MHp{~usEy>b0#ue8Slvj@g$+ld)q<I*tm65b!l
zJDln{fS6@bD~fLbv4n)-;mOHQm52FO*I+KfOH**Ni@-CBni|jZB`%QMNg?r`%$6?K
z3Jys8EG#N44m3<^bNBk%mq=E==@v~MrBM-g(ry9fYVupN*aZdWNI|=W46hb;9+67g
z?tD-+3K!aJW3-9sZ29&px%w4o$pcs5=3T|+N?S<SuvaI(eznM;h2|G9$b4v<*dj{a
z345ZLc$L1QP-K7zseS3oLf5sPI-E;T>(|B5(^Vp|dYA@F%L08+N440p9SGvZ*S5HJ
zib`iE+!^Y2G{;rxWiId<k5eNFS!{r$uit@T??efWA9<&PS{HZS4O6dqhv&5avro?|
z{*nX4hSAM5xR<UT)MSsuqJ^^rD>U?p1@2Yxi&H3tlk3Ufm@zA|3YnUYf*Fir5}IX?
z2`USq6kWaoN3nQ=>8Px;jhxM?UL>2!&f+A?wOV9}s3Kn>&NM_=6fb%Uz}-Zl(rJhz
zGIGQ?>D7LU&~ed)(&UN14C^UuH7BiDP}B0WI@N_`IdK;8+(<o(oQi3xfhCT_f6ze!
zxKV6SsV7qk_-u)9L`=%J<+0)ULMZKMox=Jn*2@glGU?j~zbf^<rv78sE`#t%j%N#s
zq6~R3@RCSsg?Ujx_5ket7X%=TCcN8NlYGn9dYVwp{9{@$gL)qL=QQw)3}rjs&%N?d
z`q<cFl$Vu=;HR?#rkrhiwO#gGU$ta6qVoCS-$Cnstt5&0w}}R&`z?CrYswu-9V;rL
z8US>H`e4T;xdrHSKiA?Fv4=AwN9k+F0*P?wF0Wu*5+f~Ra<ht9MItnXE?k|?kXhzY
z)_2pU{iYI4<%;|SGNRGYu$_cTMXged9EvpQ&_sA$K_PKnFr!AHIIVA2ghY?xc}>y%
ztr?tc*!m(Uq0Df39Dy%d<i`ngD^98oPpR<VrEuuRivAqts)Gs&*ok6Qd%JtHOJG)3
z3rs!wH`iR0@Zfl)zm&1ta!48V!Agb1<Ww7E$|BVw1Hu>EG#T-bQe=D8tC<OaoK!kk
z^s{CEHn4lmd@>|`w+c7i{Q;GTeYx?{ioiiCsj!U`qRUJFg?Q_UG-UrVa(?`qILF*t
z;e55<v&rA9M)<2VkLQlGmD%W{h0+T&eHtRWy3TistBEb{<N`#cTmxd=r=bF(-LLF5
zQ?jN{-X^O$ERUYP=SiMr!?Gh3>|XjiPzZ{u|5krDB~jyPz7;s9k76PVuC&adHwk~T
zT;COva@4PzGHClK62BLVKg1#Li$^3N;FjL-(khLuw=TBk`b<pm2D|!9@xE5(v|+4Z
zH7o+nM&1{dCh=!uZHLp5#dTDCQ0W@;6lQEE;~KL(SAwQiJn~<GW`wRVfo3M_sH8B<
zSrVFjmjxm}g`q_uH4<o|JO7xye(oY%45^*e#THd$m|TPMyn92PS36y7_=pbzeq2Nx
zK->iWbfD@M0xu_q-LjpSoKq0<jupngl|w8*7<q5NlMWDJH8OzyE+E#aB1NsNMN1Y4
zc&(2?B^$4Dr=TzR^mSurUZCd>82i(|UE5&P|4{${B%q)G)-~IBS-K#ZN8NolQ<xyg
zogC~VyF)pLvGYw?ow)=meG+!m8b#(bm-=2Dm)>56_&caOX?uIGz_>!PiL5hqxJJ_A
zS<~~+DN3IBhvnkmdXCi?1HnSY007?yBmGiVc<9MLjp=uAf}d`NLJTyKe@Bf}z34kX
zw|#PoBJ)2^f-lJHHaGWfTP33mUA&x1tz1`*!14J$Kz{8Lg6<U#|DPw5fOcvc;_pEc
zJw|r=p5Ex)bm$V;nXw}H^r4zSr&@wt+0~5eDE~jhCv?8p$%SH(EzgF*X~x?xljHhR
z@SKlIA3bd4m-3gIIq9JMt>RIl_JH+<e4czPih0eJ2lyCko*qy~1pY|u$GUH~#0MDz
z(&Zarzv0EH7Ks|hPWL_?F7|tRt7~=I&YjNoZ}h++Ivlqm+FclzJO31a-vk=0j}+vu
zXxhG5(dr)VQ(EveQ;dwF;!uUw?QO1{ACu1v%GvtlOwmvacCSwKp*dtc#J3(o-rLFO
z<L%W`ubH%vc?Jq0`zI5b2=hix)u(jS%tFxod_$|jK#RtS#*gF=TY*f;p)x7Ba^^iH
z{cr78Yh3lWzT@Gg`!v{n9}JrwR1V3j3p5&ZCQMKCOtdobc91U>o5J-Sz!%n*>Hr|a
zYkF^w!lk1NuV$-i+@hx7tQ-XuiZSZdQ)B_ff7vF$Fe?dwwrcA&7FBb0i>FObBPt&m
zhDqMxg7SC5TjUE?YqApo-NRPvE+Zi$5gASPyl9;i&Ys-<kX^q;vhH$PBh~;_*DP!e
zc-crtvkB&)Dr?S+V_ws_J(4i&nc%sfhxwYI(1}kn>(N7R%J6QwNDEJ*0O&r>68dT-
zt6oep>EAOZr4=qrjq@!Vpu6#RpBlI^V}wr3h$hY2ohE1jlsTPR9)MZv6hJ3}xM_^@
z<F(n*Kz(DTT<=zhj3+%*fKLE#AV_hzi#h}HjS#PV11)KOz?!fsf8^Oz8V`F;&Ocuf
zQ2`jx{7%pkIkZBHMiE<bMqj`q9d2FP8VFNTV9TZ*%jv){Z@#rmtwg~0z&W;`jk|Y0
zUv~?cw+L!Oooo)JSzG6JSl;~{^h!-jkI6s&s|LuJ{FE65q9eH%bdvafakN{&Jz@oV
zRfQnmKD^#XCOCVfM^4oXqBmU;+!ude)_ng7u-U3kelFMi5XHe`vsSSf^}<2OAmn}#
z_u*d<1=sh6^xHUk?#{k};<L3Pb^z6nPwGec;0>2TCNPqZnhO|GbjNG&HzQx?&3c|{
z((7j23?a0QOlobriw(HbFf9UC|HD`nPY8KxTp#^J6B9woCn=qB<m6|mqkT5?hJiJ@
za*BH=e5qn?s5j$oudf@^h(KE-9>@=P-sH*TqS6%knxkJzJ~;LKn)vl=^XqhRx{H`^
zIURi6iU+NN8)~!8vdUZtd<hLNUA(3ISQC-kl8mH>k%TD29xmt;Pi!A3ej+wj5slKv
zJE8;k-KawZDO&qEwg%8$3N(Gh#t)Of)ZcXb73c8spn$^Y(+#7+6CZhHC9*H{CRp}*
zAL3q((NJAVgE;;xDYRn=b@A_HXn%-Eb`Nrlk;8#bPnc6~KgUq2TD_`LQ7!f54C_Kz
zikXC^@TE)c$ai*&QGjnf0^GjIZByoS|Gw?keHcv7P5q%%r^Ow*r)Bt06&Y|@lU+AG
zcJ)qYlvu-_$!j*=EuJ63c~O}n70fomLt75gK4nNraddP%g}6TsxA=~Y{TsIdR4W_r
za9<1DwC*&h*{jfEr(PgWGs@c|13$oWwsX)l+CO<AF+uq&ZwILVck*v)(owGZ<Lone
z1tXo`JT~9N9+8q94#Ai}2=|B8;QU@8>_of>N&%VM@%7!nuaonI+(pH&${(sYz%*Gw
zkheiHSLC_`C^NQkUq_vzEX9aeCmPcn(8Lvfsr-|#K&Otas}nN2Rk93R-YTAMNFg_7
zE_&4XgI#AXPG;6el|xTYPXaIIj?OrGhC4<7yqAAF@oLSc!VuZgF98_+;!HZYP<o<-
z(w3h>NFbXy&Endb>{eYb*OnK#p?GeE^yUIYDu~|;@Ka4rqC5>@-~r*1U+7gpLyg);
zq$BF7OVZ2(Ct!S~o@-7O8QqUk?kg-i$1rFMO|0k^O@?FFc5ZZr_ga373ZuTofq4mZ
z%6B5&e=bg5JPq#9PV)ffp})U9=ANMlLA-6eI}gCiTRDtPE?Oqn3vLh3I;C)@i3~L=
zDQSqkTUl9bwSIn9YE2bfN-S{M8e<@!3D|{h8g7MWt7jqCMf6sEl=zK=S66(hixIUI
zr(m%kVbg7ot=>L_m5NjJ0Gx`4@W_3;F~C7Wx8+Pr&Q@T>kcEq_&c={pyBNxKoxm8F
zUGTMcwsQfhf=MgR87*xizd9Y39?UBE1~zs%Wu*^}ZO@+r^QP<cjAEQUA`+Xb$;J7N
z7HJZ$yQizGhyA(awBO18b|78<#iAzf{eKTB{Hf`Yb{OC+cc!XO7HcU0;&Y=bTtHOY
zv_RKFocj<0pezDvY$+?!qg)+WSfQ_Pcr|5!njjMUb=qDEC2b^*9|!k*|Mbm+C~5>>
znVIR*_%ijv5Dccb;&tIhvteS0iyQ?2+L=Nt*>BUP<cD9W*1s;c{=&(ORURO~mN~~w
zUEI@4GO93pP~KgW%8`3*B3%@5+x`CMJ6JR<TVW``M^o>^KNG8Blrr?clr`QD_AI-`
z9~nIaxX@VS=rj8{w+QMI-Rl&1d#%8e&!_Rki_^=~gT0k62g#pqZIaBuau&N-p<)FE
z#UF+>W@*K#gZL_(9`Ulo@!W0H;YLQ*QK#CaDx<p-d6YQSk<cOn48YB-50&P&pa1Jw
zKP$#9_^7LE?nIoS@FzoQ!;<nuQs=j4(b={^Z(3HGHF84q@IrdAkVGFHz?-w`mm)Ap
ztv|RkLKg@+89l$qxb)Qx0KZ)mT{E0Uk3aa5Jptos6kv78{8DxBT-(Z)_QAZEPAp{x
zB!d|(Y7|4{>4IwG2jCIK;68}dzQQyxni80Ykd%UD{D<=Nu)2#o3M2;ITzed$QG~@p
zq+$?&OlQG*CR3TwXdm_T=88j!$zVwgjnva0(jH>~JBO9x?{)No4Eu&(_*RV4`MO=p
zd}GeAgz86DrB4(9U5ZC-F$ca=64L8J4bu(N=j3<>deiFb#}OujUGcQaqx?c&PhwBE
zj(SnHep>V=5JKg0&f@I+q`p~dsHK`$DOfgll(+qC?<DzPIvv{eD5%vsq6RS?wf&bo
z)I5szY3}$xn$9wy$?tFDgApTy4G;l|F%SU}Mkw7#m&8w`JEc4C(F19b5mHJBN+?~D
z3d#VLk{BV~F+%#;|HboW@3;G$-QROv*ZEv>K*#0fAco`mV6<GXbJp`2=KLf0DNz+h
zy9yi)tQ?;N-W(9pSc}ISOFOu;-lg;A@1{+otrMk*&Tde6>BlXW)rq6_Qf(gZkBa2E
z$7(W*t1w3<MkdNGG@2CFJp>_v<*1wR*3M54mScqE<$)_7HrW?3&m$N*BOjktYUFE9
z3&VvqxC@Yk4w1rl7xnH!vG@tSG+8@6I#Ix3)hLCBGVO(8>brN!`!RhmN_Aa_)#3u)
zYsHa~1qBqc&?9j4l6?&;!BZ{vWWulL;(N~BUMvIX9~0-YO@5`YI28n(M2w0-us_De
zNxu>$wMN!KZ<k?9*j7&;SG${^{CcEu#))W{@FDSE(Pf@Ky_&dCFCZq6dbC~heqn~H
z!7NHw{mHQ3K~8-QQx4CZ0+b4K{e9vmz8Q?q5<+=ZV}cPrDh!VA9kTsJ+2Xio8pAJl
z+X6bH-R4*5LMC@t1srg?sTjT*u#})#sF+i_yYzU7_jpAzE!Yt9AZe!c#g;*$SOca;
zs8Ajt?ERGh+)T&3=g_e$CcOr1L?n$_tpUdZp1oHM#}RGwgy`Si&C&JMDMuJ7-Sd$(
zlmDJI2>+I&N{2pCPxGpXhKyRA<xO14U%<yq5nW@ur##f9nigBN)l#h5{<EHw!be;9
z$wI=jp1;8p0Q`#t7p|j8X2s4_``e5H1AK)deivd(_h6nknZAM6Nrf@}BZ~Qc9uI(6
zosz0RRuu0;{3l83XdoW8SI||)Ug0SEn1lg+SV_Cbeco4zRqwLm!*;mOIbKCyKH#=n
zUP@lRqXr1oT?PjB70<=FI_`N9i~Jf=AL<^rk@$-|RNRi`ih9v3A`FAiwe7glpao*h
zp>BM^KeS*<N&Ceerv(JwM-2EaE6{836pf~4QZt7}aWWN?54-pn9W*5Uh*s{MjZ4y<
zCeld{029yyNeZM8RC}cC*Tv=7DO~7GVPV7ZR(L1u@bkyiHTsy-U*4Wo{%nk!Yoq5!
zL5&d{#(vvnY@g)hWaMOJW#u~Zbn}=6YkIZo($r!csu~jGcdMReZ*h?00Ft0i&f9Qr
zr@Pg%pBWj_OBq+itKNr#Y_;K4nTWRtX4zNN(yF`&t%uOLlZ(l%ozWV?#ya!qFz=T$
z5El7{^an*LEhwT;G8zzTC=z1!jOTRB&kICyTkn<iJ2)rw@o-skhFh3+a1@-JKS;n^
zoa&x+d4{NtS0IPotcN%NGRMTK|0c`lbL}RXqR*IX<81y=!C(8l&ub_Qq(sooe>gwR
zf2Js*ql4=m|40k1ujjEtD8Jmt&cbsSIf(izBC`|6Nw#S%u^-cMYz8e+@wT3t99+qR
zB4tX(ueQ%lK1<ZRIwM!;<|~bQLE<=*3Qh9MmTu<xZ;w6Od*dH)vB4JPQ!`iAHpRti
z?=dL)!`MIRms*n-_v~<ae1E^`<W3k*YCxOS$+v+RZIz8$k*eG92u$Yb4L%w*_*tn-
zVP;R)yo8a?qiW|$H#W-@YQ4R)f*-6Qj21^rm`O}%+m_7A(oQBhY8i7~?cOs#^q?1?
zmMngMJ3H>tlNi9)Pptg?!HKExXge!(ycgRa*}Vv(q6Veei9-5#vDo!23Q1qHo8g*}
zVOLPJw@5Svk0q6Qa8j%ElFH_#!hQyS`i7&&DvV56U=~c2{T{-F3Kn~`B`X3Y?};aV
zkLJyTX2e72g`ojd#}EY=kF_Gh;>RY)_ug;o?w)o@&gddBFio}$ht@T}J&=ae{oADR
z+}rW~0m9OOfh>-_teUWXb=2IdwPm7u-%Ils*|FR$v0+dy`aruQ-mXC1s$j8o6;R_K
z>rPuf+UsKW7{={dWl<A4pLhHX+JI!L8H|FKPV~q3W3=nRT!bL8S3UT_Ey1R(k<PI5
zjeZWxFJYMX^<Y(SK-&uDpdO4NE12uN_$P}=X#n2`!aF<W1k3ROQuz-CtBz|5t>kVl
zSGF)S5#zr`{oo@<3B`+dZzy=!d|1++qx4|mhpVFaSZL1k>^dbm?BlS<R`|@oy0g?x
zYkIrrl0kM3Lfsf-v~{3$3^8_+b*im9^MY)cgoyPSF(W7EkLibwEE?Juo7}{#prV|L
z%}+mpF07c+P*M-EFnvaBHuLNw5s9Cxp(^S$lLjvq)V)zmP%xe2UrJ606_-K40rOo3
zx%1}vy^Ak-PO>WOi@7}IQ!fy7!Ui!pq^pXw-JtFmUijA!CM!@eKi}HoPqgi)69Ho|
zbIRL&Qmh8mZ<I{E>=~(%4RX`YO&9s-B0(*nGu1y_BA<P7G#HoVZl<}|xYsmTg%fHT
ze9~Vq^?{(|Z23sZzbJoINmy(f5d_Vvqv>Q#KKzDIT*1M?-uUG3qxd*Y;)FFH8YTR)
z95_5H^V)AVx76Jt5S<eY9;$$o{rcvPd^|hERc0HH8^z(34O#pKG1=_w922Hpw*Yur
zcF7Nd0v{}$?^nMjWOa>~zuvn!AL+~?#!E#DYfK!!kV6oce+9>rBJ7y#ZwVh&Cw|I`
zp=15Y8h7@SH<#7U_WAJIx`mp)_FK*Auh{poVq$~1bP^{j(GjU!bjFIvx8}Xmpu7C5
z-h~>o?5>NeZ=CpSNKD^{1{GE2xB<3`F{8PU2l@KO+RBW6*Xe66*cG(Yf#oVUmr_N3
z`txiow{a`bzA@IuUJyHi^c_3tA|an@^=^L2QT=0(B_+dFcaPS*h|2E&Bt!JBg#}t}
z!UJF=XM98lCSIY+hy)GDwAWe=|I6&*-ci=9=efOhKD~meDZTWP%Fl98L=o2C^Sy!f
zMU*0U2kjP@Ey#4CegFFG^`%bg-PzNL)`<o><rraGx1GatN>J48zb9v-lbu)n^T5B0
zkyEX`hj#aAJ~~ckaKdsn#va_KSP6VZEkFEA(kgbtd+7O>V9KZXlUQ?#gi5`4wte4m
zm+rf=_ZT4E+O&l9kPP8vdj%ML)h_ImNi7y$(M#7JoMMk$+iqge_%F&88+_{Rzwk+f
zNWpjq;;5*S>ZNm9fF2OTFS~VX(+GPFekUa@${}=kw!NcQg}x1{y_-w=)84n#!nEwO
zz3(XDTz|0>QRnBbb@)Esv~};_CoQ<zh~GN(<+=%*HAoD?_>Dh>i5+lOT-{a}w(&@(
zc{kDOD4-Zbs9QEn%bQ|)cS`gPto1O)!LKc>BmC@gC2UTvUEb4;uScsYHUZv476+5C
zLDyy;?*CKAR-I06jqBuW92{^en`%B>BY-TV#tr(JJ`D6WQV>r=cINAsJ9U^ZTIbQl
zjL*^HhirCgos}mYIH0<1^S$wxo9m^ml}*%IeEg`3olRhkRuv%>#>6H9iW@9;X%yZ*
zJU>4_v?bfJ9sJBw?qH@xd4wnue67*yHhgAtVhe-XIT!FGEufYcH;ZU81QnFFq*O#g
z_)hh9V^I}!Qmy}?xjTa1OX`YXRkrARIsYP}`aH+PXs}Y#09~qr{c3_<=odD_TinVd
zJrXI08%n6UZ_>-!ggPLu7t8x}!5<1}-m`xm)U#sI2nG=mGuaF0B!~V6|JId5rx7l6
z*JZjgV*OfnLPufV*$CDX-tWC#nf2@j_a(B$k@D9uy(o&Pwde91h0jTcapF_c8#Z$O
z;`By8?)HH{1J6OpUuJgv3CKb`0btM0KAcMH#QC;|rL`$keQdPniILjY`F?Hfvf25&
z#E_RvuFCGu`z&VW9`#we|Gx`5b`;Q&T-MmPH92$Ddb6}Ae|0#zHaK~)UKFBY(n;|;
zrB~X;id>zT{M|261(_5fN%^Xe>j?%^>A_#W1NXZX)NyZmSs&kHVe~~1`|=W_nPN9^
zODAi4^EX?aVVD1Kp^(t<@Q!&wsn~!sU0CV_SA%dLx0vvkH+`T5Sc*t`+2uH0>-VOD
zswTA(LMuVcxcTBW&w~SyPSkT$4#i8TP86;1PfiCPC(hg%=;(+LRgEy=oZ(YF*Qgo=
zSi0gfQzf~4sQVZgh{@<kOqDQDN>Hi7Lw#;gFPnfq8v$~{5d8}tcOn;;dynUhHy?z-
zy7R6F7ZI2JUxU!A^TV^r!(DHOT709XRF&|(kyL0lghaDq@U!;ZeN+pVO58_LyN90z
z-7NU&Drk1tjS#BYgqrPcF4%#;r$Ytcw6C+U{GLYDi|E#totRQkNLWx%$m!*=F|LX5
zm#nKIW7K4-@(1dF^x{3S{=@THnF(Tj;~6>7y474bd7mri6XLP3Y%>df;!`=FrRB3%
zj;qSIU~UeGj9!eVPdO+(d&?G)k`0@Q(>==DmB;kVa`+y!@JwWs-yecd1U>`U_)J#%
zBzh^*!lKCe;6>Z&@ca!n-^>j@wtRYZ?WvHz+%8fgd4V$8&1+8V_%;6P6YUs>sCQ)A
zSiBQi_Yomv%$^pnWY4W$b+~v%(5a*mmY8#VxsX+XPzVS4xwL;~FmZNd%nJE!cFA0>
zfgX9pVMMs7OkA*@qvn%y4?s8M`-|ek@cWKZo*#RY1LoMs%6{#hCpPKHM}(a29tZ(v
znDSF6-(U{{-uU3_ylh9!K;`De7lE12lwt4o`o(2)pRqnE;0*T^vg(|x4rPiPU0FHV
z8Y#RvDZIfSpUH)&W}WY?zsK)o6c&u-on5eVuk;G~MKp}67C08hK>-);1sl=D!1HJN
zBxRVZ?WU9f&&Cj_N<AdM@lXf|xgj84=dVvYZ)QeEHzGRdrfWJ6uTor7@F7HUatheA
z*lqZg4j6*&%C0LWk))k6#_ne2(CV#>Ik8InU*9K!Ml!&%#OC+?4zvx0)T8)B>7Hx9
z79fWfA&0X=5GsB^A^4BXN>1>XcU>|oYZ5Hhf=@t6Q;piCUd?KRt0xQogq)-X$ugf_
zueHhh`^i}BN!EY-D5CZ68g|@?a7JP%h=)5pwgR43VfiKE36mBP*}GJZgi?ldgr_52
zDTF+$G!fk@>}yf!I#~>5^hWe0*?-Teqq!>_*Tj#qhMoiQisGOV@v6K+Hcb*LS;Xhj
z>gg~=!Qqw5f4d{VSyN5T<i(u8TUwmiC@Y6(&4=x#dR_F)@!@5ArfH0eNv4A`qT+Dh
z>w5p7x5A~uZ3|~E>xxQHC-Gt~h~l8lwJ<Sh&Q0fkTC5OP|8xZR$c7wFW}B){ODxSx
zs3m_@BmxA^_jNzLU-U!QbCWYrulNb{>FV0Kigk8TW(lpTV?jEUJYAGLlVYb7{wT#}
z#n13O`Gl$SKob(n<)gHJf@LFp<(;$534RucJu{ipQIo{n49URZP^B35LfxAaOe(=A
zBACmI2c5t_!p)p@dEiUnC|Fja%XVO6%iC!gC?$!E|MCOe1xN8dk>Q){X|kYJyR#6}
z>2~=ll>*I+@L(t3+P^8@8OmCg$Pjpn!_LYgd`O#(j*qYd;n^2A9DZu9A+3)yKt9P{
z;-f2a?}I_IBDAp?jt~t_MSFy3uer|g`g1y0Wb)h_iHNI}LBDJc_B|br^u?>4%ae0}
zEX6hdQ>N)^Z*hfk9ikIc^Blw!RDq>lsKyyxG<w(&b<ZvAoE;vW?OeHU<>jb~G}5c-
z#ZjZ~*P$X${$_g({rkp6On&t;A;AP4K)LhB)s%nYTN}Nw`YbX_g<?LkwEBmk?>&gt
z*Sx%ygM$;Ro8xleYLbIMtRw2KSL6el1@NtoA`cKHk&1L;YOIm>rmu9Q<s;zpMrT*4
z?9PM&Yw1o}=GsvaSoM}5YK(4KR72Z9Kl{~(!n+Bd$h)iwIidg}Ved^E1oq(p0tAIA
zXpcAL`uh{!tZ5cLhictyNXbCPf)lS{ZN#xxmnj3^XqMuwF_PLeBZVezXCQa4q$!Lf
zA?GDr4X#B{iu4is&0_{#J7<GyYeJNbI93d8YN~Jkk9zfd6IVV}b>Fvt2xgE{WK8uW
z3XOy*I8QcK5o<=#lb8E_n^hVz?b|b4=UniAsoWn}PFatnTIAgK5FCo6<ku4!S!l$*
zjx}pSH5{iv95<N=)UXf0bJRsh1n^Kq$FS<FEKU3<t^D;eySVr%VZ3c}G}@fGXgT)q
zV0JI56tI24oW8^!>PfL$z57=e*51#Fl7<q&Zl}BaHeltO#pYw_I4Bfoo3jSI6^OCm
zhrOcoh<<A8$HrW?sx;ak5>-~B!PZ5GL_`C9eVp4<w12B}C}IGt&0J)UC(SaP=kHt_
zs4Hd8obwl2K`g4+ofHKz@&)0A3s(!ae{6Yatuubg{46kPQ_*kN(J4PW%MLZP^wbmD
zTf;Y2%rrAq>xhI<XP|>6{=GK71zq1Ifii0ldUJi-rOi7{Q@zvzIW&LN$S(Z(wo}JI
zGonK~zOcJjGXr>bhMig6;DM$In*IucW%XYW0nFDR0oPnCBZEvuT<Z!24!xYk)(0Bi
z790$leH<Dp3!EQs;5|n$rxTBNEJ7z0jadI+gM1JM^MyjQu#=}BpZ`|p#waa`fl#eE
z$Dk2wVRQ|2=x42gM~fm9C8EJ!)?>I*nst%4veX|BZ~*6v2l4&19G<EDpQ|;A{WJ3N
zM$dlF3*G8*fiEy2eMo&c)BdYacByY#l@l1+&O4(EWM?6@UDl-FL&kC2VIQ4U(ei**
zhNrP_be*|-`bZ$(lNE_Ig2z)H+yiBDa!9>pQoX+nbN%(`SMbZl3ECi>YxlaG^~V$)
zP9>5YK-e>5k#A>tn25L!^yyu|m(EhI?OR>YtQHk`8a+H6T3g9M6Tjpg9C~$leNDXV
zLi5cak~EErtqqpY%%GvRLYUrMD$AF;L+TKNV&kSg;bua4T?F9VgLBqq(C4j*Ni8#=
zKjpgPm)W}}U4KWO0+?2p&TXo`s5>9c{$PZF!V+!N8<kJ?ox!fjoU6hAHB9J)&I@Aj
zXG1SiP)i3#LN_a&AqrLagH`UlTrkwLBkwutc#i8akK?M>Aa7oGwh}=p<+2k~qY<GM
z71PSNrAN9(QC8XW-N4UR^Sgwt($>Zg5@neckTv^qbz>Vqs?hu-&chVB$4ix9-((BG
zY^Emcpl=}`@}e)0?jxhL8pZ0ob~SG)pry`^KShzxhoy~k(1*f<H+_+TAmZkhvil~#
zX(@!qld{A9DK_V=h4}iuSwlpXF6=BiSC-|^W+@f&SrX_=)v#5heJU3M2K`B1up6TQ
zO;`+)Xnvw@dy8VI?VV}${%5hGwtl@wdynaeKqJW(yF6;yPvD?dC###ol}k_HTni)@
z(fpy#B(<@jX>zjHw$zAQ?9s9Lz_q22;XM*<j&F0gzqPEfxbrj2Nay9r7DlMhvJvC>
zl>o;qU-I)c_s|{ekIyRvt{3~|FR{LotdyVC{7T*%O5#{MWjbOXNEu|DLc;dXeg!VY
z7AKJXF|1ZC4m{=lx^_EsXX{ihmyYGC68n;Stjug};jYav63a|unOHKYlfAdBOUE+9
z#tQbN)=Vzz$NwzKX~_v>%hUuheQbWCaH0a-ArB1H>Gn>cs8Av7;>02djWs+D<@IRG
zb@TC<9SY5)uDA^GuS63=riSTjb8vIG0^LD8#>&*wKwa~-L1f7Xr*@6)kza(^;wJh7
zKNSa!mdqR#ecU{2YIQy=#H@o;@6&>xiaZ@#OrBZYNvq|T^5UCerOuwCvcq=R2#`5N
zYuG2E2o>FT%bcVM*_g;9eIM=G#L17~4&WAyH8C-e7K9e~+^l%<mU>v25h$LMRmTnf
z@N5io8a8ucd`#kYpf(m7<@iVX-<KmzB8gnZ`_|z9;9jaPi<CreXY&wq`}-*LgOIaL
zCLl!4sSdh@_*Nj2iiehEgT6iw(1<nj;e!g-*?||NlIaiUy5h=d$YcVXO~-qTcz4de
zg)5`@jHWr>Ti8UWi50jdjHqd;%-#}sao>d21cFMaBjo}qbk(XvX-VR6^TUHGtVH<(
zspaSR5e{A74*4PA!t88ibZ{%U<3)$dZz?Fn5486@3i14)cK$I43UvkY6|+hETNBcL
ze#Z(7=H+m$T6z>2u@@n?18irL62{`CEBP~yhUN+;Pcg?Qe!fz}$)@w$8+dv-VuU(h
zsfg&sUhD{1#d!<Lm{s&}OhgaAakdfOAp@Kp9iIDGPENE{n10s#Ce|>S>xq>7H*LQ<
zO}sxnL1s-^+EX+0b^zFgfa`L3$a$s=C1KwDnZGE^|6qQU_Z}@CEE3+qi;Ip0<rnd7
zy)LFzRXDcL(Kr(SN`KeyYv~Me@NY;%9sRjDD36TH=Q<u+ZZed*?(LmhUyE#GTUMR^
z;2#_$29%U{wpK`qD5^d*STzbEQH-mFDa9y8q)ISmi=BV`6Ufb?+$@f|K~fp(4F~|6
zVm}8-zI3r6csGbNpi#>n+J=$>4mp_i0Wce&PK<t<m}&ZNaq&9neN7Fmj*zSA=Io<_
zLMx(ud^w>$ZiFPx2J%Iqfn3Z`^n;-!8?AiUr2;fQI3g_Q-zji0IymZtwBVm(xUAjJ
z-~R;H^h*{WFCS#=G*qKS<*x;<jO;o7l;KU=ODyV&p+Qyu8WLKkBtoThF}Z}3V+Vi?
z7g<R6>Hs6!-NMYxWsXc?aV``9I@_)OGtQ&!U-J3rZ%QcY9_2B7{nMU+(x$wvq~BDN
z72^0#?DJFMpNgl}2zyGKMkVHs{2g90IJaeYfgiXCcZbusP!`u3oAW!&yVmP>QgG3|
z7;KWpb3-n~m$$;b<Y8WS$%MyxyM6-+{_9c!l#~j@TN-6N3hEV7u<nKvaEjhA5!q79
z?3AJ-__gL*Zj2S@TgCzcBISWd<#B@*^~d3(#U-|&J1L_Z`$v~2M~By!=Z*ebtIRC6
z8<uHd=?{1_t%hsW|L}Q)+7;ZRhT&k<nQ!d}yZ3ck1drhHDdCY>nGh!I(|^F_3|^g!
z*xu&!=bTO`j1gw9Z8$W)G2RphQ4$k+sad~^!Jll5jtUW^%)_%YpSk*(l)==-iyy?(
znA>QpZ~-VZHr6CfqSJ|%=?m%PJtt6-w?AC6WfhK1hfnFML$GiLCE-aQ-&~GePj>lP
z;GfqG;aPi0v3sQ8Z7SS>oe3NR84&x~*4UBwZ~CJ>m)~h)YD(G>49j(G{m&fu?){_0
zd<89$Gu~FbP<1Iv11V^iwfNJhOg|q6LD?zwovnCV^2*hr8ZkMat(WVhNO~_uP-G6)
z0||JwychR+>ZO>pa3v49jaoUdra{$W<Io?>9OJbQDi=+3%=6&@3Qq38^Y;(zJ~yA?
z?JNj)3Bn;~YiMV~D?8%yjMl$k_UL4nC|z*vWo}33myeiJO%ts$!Jm!CFMM;FCqiTd
zjDz^27W#X8B9vx!AIPvT+hpYs^ZkbojGkoFbGqt~3IXXqA6C7t&Wj@=gBJ#IGIoEr
z<0<^bBjRHKha;6Rqk~{pK-vm)5EL3H!?xxU&@?C5WAZXv-?@3p`GNjVg<CID|KlNr
zbKl#hj!j5s_8In5;pF^gL&;$S)dew84&>f{;RIDxkg``oI`a1KkGEaVD`Hv<mgvsB
zxk}B4@N$8F=Pu3fq&n_?9zlfmXIz=mR9La8TI(DBe&8oDpn@cjEAJs#Rq0n2J)qGF
zlv9!R9$MT9mWj~xSod%|sG^?L2^BB(eQD`@V-$bTBow=S94_lQ@lfKY&sPPe$mpKJ
z0m1EYDPg<2@@>ukxh!fkXmFi`=gS1Vp8wl8$>^@VE*E1?w13ugACYOm&hAkrrT4m<
zv(Q%c$I4mY;_I7#4*=qcLDOu9g7Q{|?m3~(%flx5n+vO(b3#c(fmKu0y$?df)$ZGW
z9OKtcc&Q%1O2N-S>U;(~FkyUEpgYcBzXPcD;@I2JKP5`d$Vrx=hY}&<AV!fD`$H}v
z&GNDl&7B>c!Gwjk*iYMZ8aOJ#g&!kS)Lxuj3W6pK->yah61R!;6rArYhcqV(F<B2(
z<n(v{Hx5#_g(7yp0qM0yCx@YH`D1@sqhZ2BPBLWjetQ((Vv!$tMh6})O}ebHCpXz#
zzHa>w>c}?pVBRzB*RT`aOWK9GwpdyVc58+6_~x|99>dZfS!%<qZMNduj}Z}MPsy{P
zk(8pJ;Gfqffw(?_20^^ceYz)Te>%aQQkk%SL7{*i1W!PMH&%5tHM&=`5lAE&F@Za;
z>8W<&I3aV5$5;LiiEX0>QS5yjj*>c23lA@GOGE1o-Gg6&CZ*JMgHQ!$$Ao<2+lUB>
z2}4fed}a^I%AT1C9C1F07oC!duY#sPMnVF`OY&+KH^ZHdoZxQ{4}YFte8%>V4sqQ4
z>tEU+S)9Sd41-?*@xATZgGQSNt6shU{`?Y4`sA&8*!j`rrH`Cv8O`SV81`i-k;g{g
z>r$_enZ7ej`~Hj2E}1z`m2!sP7hhhoE%+xD&eHVK<R>c+-^fB2F4=2F)|zGpQ{}I{
zLN98uzb~f7LT0eH0zSmgFWrXeSrwE?xx8GL(TN`{C156iJw4!J%Tup#et7RgS=_hN
zD<5;u4>b((G1f}3h_EngK8uH+rXkK9o>lHGJ#2CIE)fBbk`+`K5sO9~)XQ;1OelC(
zPrsdH{*R^Jp{u1KT<?&N)ST?xtUSSb;krWsDf(-PJ7>K6cVF!*r{TH$;8)&*V0Gv1
zr(nbmWpyYp8H2MinO$zuBg))5YM-DWPV4&A_91sis#E5+bn1YiVQ&`>J6<z`6BbK)
z3A{->xVLJmTXj4avI#}PosVQ~9CeOugXuX#tC{uFlzC$=s+Xe{$?5(iwZ7$lPxD)z
zSaeP8HjMjFLCwCq;?RgN$4iYg7(H^|TG_d*o|pjK30p&8&`2vvw?ll?l4|roKZ8;y
z+vk$d*kk57L;8_kgqTZ!o}+}7$m1tXEb*`A7oZQ*Q-3=KTQGZ4hx3!aq<>n6g|^qk
z{ZXCcd+LuA70dsXM9n1W?GuHtRaB6yrfDFenn3@)`(~ttMR*6M64T$ynWkCgQ5hZ$
z*?I`n$NNM4-KWK$-BP=6Z|ik8e`h^{O}kiGUxL(Gk&(tvc}F#iGjkEjSgfH(L!#s>
zqSpsqbZelk)hDqLvGW}fyoXqy4&^xwvnopI_w{R12T?yW9Vi-u`bv<kI+Cf@h`hP{
zmCkuPxI&jhM+j<L^0MmwR2LM5)Li|Iurvb(yHz!7Q6P{S{3+#Mik#<VJ_8W3^x*1I
zbIm8G?7w1_N!(szCI?&3ri<K&&DYT`dtl{gP>LNfbEqA63CLRAnmXa75jVn1l%Ouv
z#e4-tfN(BM#0&3}?ab~vR<jAM?QG!jn30kAULW(_4_JY2$=R6xJ;u8)A^3W22S89n
zai1Vh&)*zw&EK5Q;4zc<V4b!;8LnqE##tdv>LlstQZ!P}@M?+U27>Agzdk%XpJ>SI
zoeniLZn~<Y79$QA6?PQFewG#@US?q>X@_Sjoi+{Pa6X~Zz`ujVlf~5_-!uZYos9er
z|La#QKkvenvPHUOCWEV8;w5_k$LdGmz!J1VTnLHZWJyWnFLZ9JM~d*5f`;OQls6a~
zs2?XNKs0JGH!^yI7>%`Zd88rv_JgLk3@Zp+5=h%ltkWCoGs*dL(wD<}b*Md*?g+su
zk?;>q!rJ%j7rm4hNp3VtKx}nZ?A)zxW(_o@sPCLjR$jCVKY~Cs8w>+v@<ZH4sxfP)
zbB}k}Jal6g$`)1acpxg-W&BDitPt0f-qu?bkrdp|TA4`XL<=bAYFSZ~ue2n=GM*=A
zFRbQ{N`I;1uWK)rYxdqf+ePvdL*Fkx$(c~DGG10qbgi4y>Pt#dq3&Y3H-IMOx#72Q
zYhj4L24ya9i7Zw3WI2KO(T(g#gRPaMg_xq5ujgcC3<b*uhEJ#wbm^RjdLXfcw(6g<
zO2(ldV_i!P%zH!3@2H{{pe|{uF{sS739y7Djm!N-nzZGIwvP?&4u^c#-)DNPiOL!{
z8fIZv4<Ym2m7eGbgNu^?+Sk3NG>uV4mG?^gv>xxXy!|Q_-NFbkX;FYwc$Jt1MaE=n
zU;X+v6DK(cc?J27{u5>TL+mx-cZxz8K)M=tw&%(<G#4v0VG-eOVY`^|vG|t9<vQ*$
zTuepuj-{GHk4n1IWu(=aqKsVmdOXTBXYmFpezd8u;DQzKm*XXtg@)0_6T*#mTHuNh
zYB5<Ib*YR+zXKW`-5LOdCT{)R2+|s0YPX!8CnS1-8$v?jOM81~%}iOx*6qf-h-jvt
zqt0Iwv<L+TXblw)&%M^ls}mPD_bnV-BTfN6OAco$;K&!WMwkaQBjJYKN^t()=4W2-
z<-^*q_78SXB7*-d9(aYORgkGd&8MX{v>kq$W(#|w0YsAMWJP9%u_I_<@)iY6r7)5R
zo?Cuko^h29aVED2pj>LMJ<VsNw%B~;DKmh(KY*&OF859HX)%vdobTR`Ny5I+Pd~MI
zp~7;XTd>f<((rqx-c+-E&rrvVzGEbLDydpm5$z)6$1R#$BK>8*LQLPrSEf2qsZuzy
z3M$*g*uOd8JKHjRSibaW{P*<9;blpcU7t+>OJu;$x!(@&vqUBhv>Z4G;E(lf=|M^Y
zNlGTnpMProQ;M*cV&v~qs7J_qo@|-j90!Mo?s?|73e3N5#OFsGfu6t4sGNH=H+O*v
zksQP9je~W`+xe~0eTxGal?ku|Ac2i{u<fbUUC`zwuy=N=85tk5txC7bzoGH3EM?@W
zKF0ddNH?*|#zUBxbhr()U@D8xYZ#*GmJcCGv%F;1a9wS&k-@OQ``R51^K#*j_v})|
z8Qx90<NuYbWe2;-!U<dw&na<&#J_YSiG8&Z@Z|;O_;-lj>t5iTibw@m#m}q|a>fLO
z?0N+=A8~e!Kl|*i)0P{OCEs-vr18QvP3#-Xo1D~sLy$+u)zzdP=J;@{e8g>#RpZ~c
z%%DtN5^h!1kH(Z*xp26`mno8N6zPJ=*-8CuxWpCa<{Y02{0qA9lA2|E8ni_SgFCnN
zt=O*S*tTNp?4du0Tu3ZJJ)JO>li(x(e-NZ-A7wfyqZe~M@k~@1n!71Bf|n_6i_R-;
zk`tx0v9U&avhgM^cxM?##ukG}e%;%581Izp075JaEzNviaPnNX`5%F3KB<kyD;AMg
zc7Ca{(rl3~$fndI-}^9AR#ic&H)ynKtu)$&h?-Uo?KE#Rb*qXce7CzwmXp9E?8%46
z*|o@f1-IFu2~eh>>lF!^dL1rw_CJ;ysmwQ$)c=8yTkgjSFjz(OK4zjqU7Z-DWQ68^
zmbCzu(JHKcDyWu%{NSS3_JTdsq%QwJ6j9{7?QY|g@adD422Xw7b<d>SO5#CEpkQHi
z*yiGv6eFF0&L^WEZTk<%$(e!Q?AV7rY>`k_QGFJs*myE6N<Ck`SgnwF@W-vvh5$(`
zODij*+pyJ9Ma`U5Lc>(=LBoX8zC@=dMo0W+Q(9(MU6+c31X)+p&AgEQSQ3q9gDFU)
z4{aPQ%FWKt&4Iem^}Wz_bAS9Z#hm!4dryv6NYg-k5$6Z-rj#%`2td|yW%=>$>Hn?}
zU0)uZ6M78rn6>h?$;rvn3FD!0++o1udf8pB`Ku$SwM|9CaGiF8zqIGh2yk^VaWNCz
z@hyt>#4_tO?&uXP=Tj}x?i%56LjjMW$A6lLpXpudzu>r8Spo>3R|kYVP^n-;-saW}
zQA@+b^nY4UN*aL%4N4;I25ey9R%z_J{m~-Y$`7XJj(hIm_LFiGh0;SjTRuFN$-MXn
z&ad^%wj<ozwuNp+!?{PoE}}XnnPCetwRQY~r?KX0cH1KRtQ=$YZhMt{4x5{dY02{D
zciP#ZU8pDqCdRf4h$GjX#Yltt9Jx-bY0AnMJnYK_%Sjx<S;}(tbLQ>?Wyqd(MgpMN
zukC2(cjF_FQ67?m_9bRy8Vt9lVt&>9|CVt0szgx&IGeW?A?tZQS$=9&C@AFAA*Qbs
zEj)3vRI|7AYc1(9{1ynTkcyw~Y=48<*tixOF@IBE;ko7D?jHFbq74~{{I<T#Sk~eG
zJtY6WX%-lxJOfS#bjy3iZY#hM;#N(AjSYimJ}n>BjokPK!({S3JUj@J!Ny6IZe3t#
z&$FbLo;2W~Ths`J?wGWPox1bEPwZ^yiX4N3DVRZikN6spcPc>3yAM=6an=Gz7<s;z
z*S@cE&hU3A>UwKl&NyqyTQ$ve^-tlCn<Gr<<%S**92^XU2ZwaL5yA{MdOUn0Hq~J%
zw9}rOpq?EpFq_>BIw)&wJo(~Lvo?;%;Hu}6O=$zO(TnVI%k|(u4+2bBbAqZ@2>y?!
zr(9V3#Q`lx#JQ*Z&C&c#P^)w-v+&FcU8KKdHWlLGAQ4fc`%V?TQhwRYOrGg<h)2U1
zWL;<I*DU7OEWxywY_cjC9`SVFdEb?__-1}|xL2MOXJ6jpn@^nEKN!3NoL!s;oqE+_
zd)%6u@F5YQ6oJ(AF%<%euB{p36!bBJD=#WZB*vx2zdI&Bg=Oe{@K~)hi#aKC>^Swe
z7^$h*TCuH!44~z<-$Ei|-OBg^w|b%rj1x#vxO8;{Z##fAt8GQ-f1ZG(BbC+3NOELC
z;^0bpy5Wr3Obj#B<Xwto?HPWsY$CqbAK#?$EM;$vj*%YldmX!(2*b;$91xgvvfYHV
zWRjv!@*w*kANnYG1KzIHrk$_CbcjMKiF(+C8{?ozo#Wlzlk3GYqc3kpm~1-S40#IH
zN(1v(9e`XOZW%V(p0Eoq;PgPGFa^f>hCs_NyNSF@(W1vht$_bkwo3m>TZ0s&FKQ(I
z$RFj9E9NE13V-W<r{_bY5DAGC{Ha`_vIawu>&(QwrO?fm-i?O^aDGNuJ+-WQO@XIJ
z7Yyf+n?WnU-E3xJqH!){#QdkMf&U%GBvM2|);yRTcr!0}?z8KPqgF$RKPg%6A2m*g
zK%SyUlk;6BA-A$@YxVhL#(jWL{J~N$$Ibpyf1$k0^`+Noe7}euZV0Qe6E-D?xgD9~
zT)<dHw+IcAD_pB!Pa)6pel37kY;KnJed0e~S7f-n$bIr!n7KuoO`STjci%6{GvuXA
z_`8q5Z(T0c*-!?iG;tU$4a;h&iIxVrm4E+{TH;d+)kH~RC4S~^oEq{&Tj(5r=A#HE
zQcjeaZk-ZykNRUHX9*8F#$`?O#)*!=%?`*jo2d`&;3O>p+ti)oKC%Rb><`O%*x0l#
zK3M?T#rEoSkKIEo1LIs!E@F&d4>ueSu0#M1bl%%VPmvNRE=p%odJwSz6~96G=oSVu
zA~dtr5_pf?rvymRyzGdTyAWW7se#DQun)O8dV6<lV@3rAy&n%P9PjuLor*kYd+NQ;
zW<i?YV;(y5A17x<uqBqw%rFpwfi*)tI!+s%kjMVGkS!0ZXhLE^Bb%~Lg~bw#0A3pm
z8a*H=b#`&ZHgf+VGaG`kPj+JFKuj%fNMLBC<;+AyBzX@q-Xb>LXh2c?J6S}g6K!8Q
zaisJu(v2--Xn68}_G-D-O{ktvq)I$2496v}x$I%;CtXa1<d3P}eD%rvVgF=#%8?L#
zvdJlXZld|nKV!nQwI(h;Q|U=rCD3yZ6ly4sc+Ia;ZW{HIi%>k55`RvpUnsPipQzeM
zc}`RoPeKVs<-Stm;oa2<mvc=*xmIspZ>^cCgj&&PJv2~lR#QXXJ&|ZLrg8ZDG$0Q|
zS^t-yzN~tz9s8Vmh2D6Dh6W}Af46VIHISeXx!yFR+;A%7$|^YCTZRTN`GiOXT(5w4
zc1^%jtNq5QdHyVesWmAMWY-2wJed8ep+FMroNF&RTt2g;Cm(*fK`=HKQ{~w_Fqr;D
zLPd*J<4}*rANPG8uDC0*?&I79BhGtH@>GHp@$oc5T=Ckk?z1p7EVC|HM=bhtv&my?
z>|<oG+I$`?5d-WtG|^AlVK%*2qR)PzxddhB;}qtrgrKh!UO`3%bl^P#x!Mb`@2^2`
zS(`K5Nno$(d+kB$Gu<EIkTlG<T76W9e)iqxd##V&cxA6Q6VWN!Lz&7Vc@eM)gR)>t
z8gfkJUxG3{yM)J|ZLGArmkq~#dD(RkN_!c|c+R=-{gjsTlmT`o*I~&*rS)*oPeOz%
z<Cv=1)KfB9<ZZKNkRmSwV_Ep9$ZOM5UE^`Cw}A=-Cp4kfuoC>U2Y=`_FPzpc+=6D{
zCL5`us$-}y%;%w0gV<5-X*enL*dHz+Z+h}YYDa<Ylf5(p`t@wjMr9K?xvD*H>5}zY
zl8RCDgts0qTe6G>$nFV?)yR315UWu>iiq1WKKs*syPBL&@1L&+&r&K#r`|7ZxJaa?
z@D7yIvz&S_GN9cTEvvW;HFN8z7&9#)hA*<v5GVqzWfMH`k{^rLM?~f&+JqCZ%%RnF
zTp>Uxy~oVqUc8r&)JNXv8d~(S{nU}ub@hd-<1S`$ATiD^Y-xuNqnA6ANljH<Ebfzg
zL6B<1F9>?H>+aS>ffZq{5u#li2MQftsF(}tK%iG{6&&}np&F3b7Bop3n0cXK7|mGh
zIwWJeFA)6xfYsm+Q8w>AdLJe6=c0i%fZVrP*L|N{-P7{;)KkKI65n4a$I70n=>4mz
zqv6MGVxK;-M+h2PaJ&M?TLpblPhE^}g|Ay>rNwwp`am45mB*lOdat@wl?YKlUW3dx
zh+-cqF~?(7CByg^c^BQ8qvSS;W|wDb_%@WgY@D`C!->BY|0}LWYBs;+wZX>@pf6hT
zk9uW7ri1X);gv6jkA!V!`%vtW>{;7|DzZ3xq(I4-CU$j#ke=ldIhUA)F&N5rK9e=Z
zP58?eM(Z1mg8B00?e#oIto<BwH|quBP0Q!bCvs27VCz4MZ+Ab18FyhH&j^NhC{!hx
z^Uw&G#UJ}zA6{Nxf2oMdnDOcY)F9_9@RQVzFYkT=5AR8m@6f9#QKS8K6W5Yos&}-s
zNY2jARe9>Qeo>xhQtV<nGU*~78{*E*F%!j^WS96<h`i_f;`uK!N9*<U2P4kabn$2s
zg}zt<Y(OANAl(Cisq~b7X0Q+Nr!wAK&2`ICh;o}#-}dCRG~0wV6D2_aUbN~!Ov_YY
zm%=d917AHW7h6_t=K*)(-Rw{%hqqS0Bl_p(eLSnx06&D}S1@S1UQ|Upo=jY|3;U3K
zo+wl^ApTZCmlsyi)|Xc6H`n`DQqQ;_nVD&^C$(Bl3_5f)#|;qbjnhmo9)hqtJC|f6
zW;2=(OBoc3TIQ}UgPI99EdPzBhGnnMUlYpOfwnhI?#8rST@d1)o<5b`;yCAAxl8E^
z`L!ekIz85Jxz~pk48zHSn1sG5?Qu}+9{$k(zhgA^-Jg(h{8yr3GwGV0vt!)VjZgIT
zT01K%E7ymm@fe96`xfXOzS69Q(7i_g9T$2M<Yh^q8B@<wTVygYph{+c_&3_8hFw>Z
zg9t=O{OS2f==ArG;`-YNOK~=={cCLJM<EPea+YQ^EVQ-4U-Mt{e|!c#y{*+-=SN4}
zJt!u1Gq;%uBceC0cVEJbb1O?Hyi{yTR{h&`Bt5#mPz^15YyYknfAY@cqE=v7A%E8P
zn^*5e92t`|xeYnM+P|G;<ZO)J=R^;^+L_wjzBbsE6S(!Qj}`Qt^+Coq`p+Hm?~C2`
zm-gki(y=6JleI+<M?$#Sg?QDYnyA>uu@s%w%v>zB`Z!RgIou5c0=2`^Tb}K>x79*a
zmUX~TZFPbQG3di<&}(Ng>E97EWm8SZKi8L}Pc$|&Jp%l%%?E!GEXkhC&*40Cv=Vl9
z=@TJYBioa}YD0}|6a5?5@ulV+$2T_+1<f`QQ?zV3%#AQ<*F`sdnVo0-u_&f@+wb_x
zQ<BKxm>+XfimbhPmF|Bi_@CzR3v1jNFJO&(V@;)w#fyw!f*JX~RqBFb1k~x!I97Am
z(;;G1HbB%3ej$+D%W11_mk?-kdq+*sY}aN-EA!`mkXv1aMh47)Bv+aRzU-v@Lr`)S
z*I)*6DKjK`N-7ZRWkvp^<OldvXKe0I(hWJGYP;A2%OD-ro%176Qp?QzZn+ohWBqi=
zMlPN#ktb}jT>?|Z9v)`2+#g5B>O0%Nkey-HoB(P8k(U08iz!(n!hTC<eLOjsW!yVE
zH~F%{P`{&s<NDvHSe8G71SxRKq3x)u?TH|FI#jlk7_Qm~Vq5m2*3Nsij7@K=A-LO!
zZR12F5fLZF?UMZ@od3%|p`%w%#kVVG+rWf_a7WES{`2oQ|6U?HyvN1fwZ+ZG=aiem
zG=9Ov@uyW*xn-V`yT`I?9rkphWuLnOQ8e+_yrfQioD^#63Ab*iFL_HAtcFzf(~9?}
zHXVCJCNOA+(x)r=XM7GFwzhpe7zW3fY=9epHWTSncMBm*wQA|aO5CQ;)1Y9h!np|{
zPt1@K5t~gsDKck%erEELe((>Q&_C0|)1oqz&d~5&h-U}5=gOrTZ%`D(wfE47xhP7A
zHM__7R7Ys;k5>L>vbs%r!KXaA*83#y8wd>(TDKV;m=m>@73x8vcztf-fgN&i$vqb!
z@jWz0<PJ}$7aGQtVbxB+rF<#%;4UHueSgurrLbuYj@}<DUY6rv)ouglMCZ`cjF8Nl
zz1nI<dIcbizy50Fj&`^-@wWQ&t*9(np6j=4mWI#LD-NbK%=3bO-oDmRmActyu8t|I
zNihN+9X;3TOhe@h*IF~9Tpck~AYD9QUZ#Y7HPdP_iyOjyJPwd76gPI16F}mIPp=|I
zZ}D~b{nuJQVS(!mejmTVLX;Ly3bI9L^CgonSl4RTi-Jr2l$hmcvDldI#WKhBoSipK
z_v|YS8bIzi?BqPfJ8}^5&*f5+)lw)GDfgcTI2jFzJu|mWSXUp`z(-e60Rk-AbThF?
zSKJM-Sl!(&43{OLNB=ZB^O7NW5c#`$_4|tiPa<%+)Y`$Cq(3;9M=;Y>R+L+?RG@!p
z=zs*&;SV>NX*~9rll?s=j(rFcT^}2x^P)N3QgSwNUH#S1wyo>*lmt~maX;oZNIk4n
zDdHBV@C{1@!{qmy#K4_C9e}Bc4Z6#$xLPd>K?*=1QSo}h&8@is#tq2Sq_=3X^!bZ*
zH&I$J&Cp}HLfYoUwf<AatUEce_tp}KpU0RGx8-sLKP1hOBmrsH4e7ASk|FM-8>sxg
z^3DFbLaI7o@N!HOF=6<<Q`p3Y_dRKG^HbWrJDwjJ<5Wg{7NmsALv<E5+(!v)fH2FE
zs=FCPXRlMlJ!)={_pi8WfU=*;#J{KRq(*VbK^Th}&G6MY9I?mQu70XvmCb{Xo<rK*
z9SEQFC;>=qK;5HrmBDn(_%=8j=>0@o<a+A$%yU{ssW;EH`8FZ+EE5)HpWdZE-XuQP
zm<$c|T(mT4oD*_yb9SO@sh|b^L7_cH&trxOnKcg_?sD3HS%P>Vrt#Hhyv0Fx1KG4Y
zb*t&KTY--J+E0S$7;Tk`=bpl_B5y^VBE-cXfgZ=#C!Ve89?E-7h%&=&ov_h*^R2?v
zH@z%pf-z@1;Cq13agfv;(0sGF#-W$mWNP<Cdf$A^s7yqo0$ayGTwm?Jqb@U2mnkAN
z=p$u0jBz9YMH`wNaTdL|gjLY!u5Rpf$c8(WsNM?0r|pkwf0!S|fZ7xao!<!Vv)v-n
zMm-{8B1tnBYPRG4TxVswF~_2i6O|{rd#z8a8r_z+(hShEGt!VKr1AXNoijGutNL(`
z#LVa9k?98WoctE!y?NJih(;%ounjSk9;@>((e^$dsa+oC<a|21yNG@B(l2XM&ejq{
ztdSOMnYB`j#hWKWZ`~14Q6WQQ86<qgwLKnx{g&!0wi!~O$x70f7fV(=8Df$$df+IB
z6OUdyJq89Z1_w(mTA7PTC1REekXNr)j>2p99vszECC2GqoFDD(t#_peH+typIWfXN
zOW|g*PnJ<kif%79b9o{ZxKV$ISs6O0hRpdnk95)2s~+oV4|2`-3>Dr7ii$7P^MO@h
z=?DW68{y~rM%l3Md|gx7bnf;_eqt!pCxo&lHy}y&#JYab?mq`I1&XeM1Xt=Q-L*|#
zN$v(ln~GA8s(5MmrzH4-U(^qcc63fKi~PzD35^F+JIT9baiXd!%`KHFc8Ohm^`g_C
ztq!&gA&L9%HKxWE<i0_9yg6sAM<*JGW_@NyqMMHy6;lTO8>`BCuF`tqKhrq(N^hO0
zbOK;j>QVeNdn=m8Ay|TVQEUh=GP%|?iD`0ovv|mNoGH5MT<tR=JWOQ>@d3F_-#))K
zCbQpmGOR=Ru4rVdP$Hm}s?FU@S2!agzNqV*=o{Kj?Y4>*ayl-3XpW(w@mB3rSy@UN
zR(-e%c4t+If)5ncwpr_fFS5|=d{&~)1!&rQL^S-cDcB{%EjV^Dc}qh}xUf)kViU(+
zd2hR=;Dm8#Il3y45~XkqCFAdDaHKlBJSPMTX`lx8x*B>N%#5XN^r6u)ImkP!;qkfL
z;eCtnrM6)1J%M7yzdcxD)+-63BJvVA%eVW*Vp6SfXaCfXv?a6--|^nI>Dl-OcpsOr
z-hUiEfTE<TgB!S6(1#^jWHO;3Jz_Cb7X#bu(o{A#{Nzq{C2y%cd=N>f0dv=(%8Xt{
z*jPn9<WYDyKpSZD*Ffa=e-NU#n$OG5Y6bq+g+zDsLj#RADaouDOGM3FysWPIIs7_|
z(VUWllxAC0vH81Exx@4K`#ydEzYu0;rM;uwU1*JfGl$VEgL-w><=F@K8V)W9`7|J8
zBv2kny32%ING(|*sr(webr?~7Td&;5gobcnQyMVH9Z~%*Gt(+X$LA06b2vgI>Eqq=
z67Rk(o?@3xA|d+jDWa?hMcDH2&FenxPC^(xppP<bzSkO+y|_h`vJQAa`jXNu#X-`-
z!d~-_7EO?m+-sf-SQYgaFD+%mCi7V12I@qPXHQg`pR49@Urk^|<E4pn26_!((P+P+
zTT>V3+#aoxJUF+m6susxvl<FNm2EB5<!DZw6}#J<mt)^D=?5uPW#wkjV`i{ouh5L?
zsz#Z}>y{S}+_ze*b%0w4%1=lx5lCAAzd6k)#a>zPZv3&A$q%En#Vdl`F%|M`AIEYM
zqFHmvgWcmz%ce;3Uf+|2>Wq)w8v6zVLy2t#W74yV(}Sw-viwHhTvgj$S_nPzKLx1>
zLY8T8wr+D{tO9ADP1NUB1Y$Lu>(?MOj>WRG<jow}FfOjwKnjY<ZDq+@iVc&O*SisP
z{;!uXEQb>A)~>Yt?Q04Y9~HTijuFzsh3A&-e2!ClNbdy~MUvVET3wu+UpH27$>_P1
ztv1xV;|TWYH%{W&Hb<Tp%e5IVs+XDaa-{U3NIG+hzFIECnT^YPN>sFMn$+$C50MkB
zGL8+rexC)XFjJ2)CcDdK6Id3rX!Axbmv8UXe0xuDQj&};i9|&SmJ!bO+0&*!c;@1K
zF_U$7NslSjz%+>P_Ue(0L6E@`(RW6!ydspb%8%3csq2t0?SCzfsh0e6yF3>YQnz=n
z3y6yl{KOW6>8bk{!wF4dV09>PHf8<bEn9Owh`|7u)8p_RGcEayj!1T`IM0ime{)~1
zLyPm=@~2KSdpU)LE|-r>VaVpC*0ZhP@Q~Mec9QHV{i6n(bZ~*eQbN!Qzou8{mL*jV
zAs3}CbW=p@FNrXFPmjJfHG#lnL>^%k(n`iEe(JvmrChU9+C4P?05RONVatK<yIB^s
z`2$-F-#ez9GMFAj+8?f-W{rSEzl(o&MXO*VGue1664a{<`N?xzbNG0m^ROrpAsrKK
z?a@beUU$%_KqSKpKl#P%=6L#*<-E)~*>jQO#CN+E;hqBG|8Z3j_pg!fLl6USP2$?>
zvH%whzBNl^h)S;^tAcoG>p~i7m>%CBZ&}(fxboY|vehMqJg!wzrJ0%JWv0X5yt`%>
z0j>2*rP$Y$nomOFP6O3$`S0hyOI|7-afN33vn$82Np`-QjqaX!8_oP;6Bj=Eg`vPL
z$E~3qkN8dAGak(v+*uA3V!V#Q$fpJom%Z_XSna=qrTEbULJ+RMcV{DfMk)xpFVK+5
z&9lO|d9r)bG+})A&LTqH9`>Oz2q56>tfn`MBV%K-2>0>`Gi`(87wx*hz5j7+126~2
zU`OEQrm*qUHG@mz^P|@}k`!4aC9Q7g*|QtKnIph4G*A24|Iu{b|5X3~8$V`bk0je6
zQTART4%uYIk&q~kV{;H*Rz}H6wv4hnNDkREBMyplj+}FN8<|-hviY9R58r>_{BUls
z=XzY%b-yTvA+-KaJZi!Pr$uW2dGd<1I2mtH#OiW^3Fvo)Bz`V7*M%fwh^>+?!lH=P
z8cFa{_=4hWFmP3x*)O?w8V!H0HvnFeHF$42Uet`jjEqW!-FRCZBB79Zr}@bnAS?IF
zN2dwm*V|@a$R+3(2RL6Cs^UML&%$B2_L<9}lkILtv5OBlmc2D$`qspQy}C695eLrG
zZs(ui7;Bia%)gMl!w<{U$%k?lRI$aEIH1ty?O8WK{;hTQyk|5svdykGq(WXBF>l?u
zS8>gD1Zcz%_skOHD^1M6`|%8*b<}e$&^)CMnPq2Ji?QuxN_N9bfHSrh{nrMBti&bZ
zhkM2Iuc@P}OiiT}MTNoCmS3z|5<zbVngsA(Uau-R9j$K86q-xMdpy^nYb^IUY0sD}
zOTQBEbfBtA`5kHH;V;pI<{J-|#2U7OuNCQX-mkzjnH%abGnZ!tKgNauazyL0f#m9g
zNI;V>;e4E;pjzV-{O-K8xvET{Qu)B25iqN~$0vv3Cc(j}DYx3-F4KoSV?TqBLkZ|&
z?T(1^e+Q$u(O~FGquuN38UsbklmN;Dmlb4>hxoMvRex4tg$MAAMGwrF9TH>g$vR^%
zhF|39#iY-%4PC}FxT*qwMEuK?o|}juMYDeJ@#Trq-%4(>Hed}35s?MmoNt_VtCe6B
zrT%X*c39#8ZgfHRy#R}apnz6n)XC9a|I(6Pjy(Mx?o2sVS$d9KtnF%;7(_2L@|<i1
z`I!&?DVypLp9`xk&$b#>bk387ZCDx5>ToBx^51`0ToGJ^-Khswed|{@;hs__jNp$Y
z7x}jbhzXAJ5N}9d^i^hCN8(?-$$4i=w!>b&_qSANzzON!ed*)7gp*o|uV2_BU{ZyH
z*y$#vhp*Kt8*<~6Yun7mBd8MN*Rb=#aff<4D!0Z4JAcuR6(~;iQe@~;QV5QRv3*az
zNiXDPYy0Xj%bNsPQ?6nO)26)nJUlCeTLK|^+g9u&U}i&8un|P`rpHL031;?tuI#NF
z^83k?Z&*pP%hc6x_hi5Od@^74JjUZlIpks;EEgmtDJw9#y_jNq$^Qs_PIAMsr1~^_
z!$so}e+|b<sjl&KQ#<k|dVS?;H~qaO|K@I*o-WmuTT?((x5bjHt;(_TdTqVR6-#T+
zYj=}bb#z-cqP>vKHTsSeuoQCfyLfRDph1-^PJigo<^p$AfPePU4V@o{wD33ZTuHko
zPR-Z9lP9(iPC5U$CiL<y7Y$NFj5@BlXnd=S>Q%C<P_g&YQ(k3-xs!IYJ7F!)d?f|G
zUdi)Dv+DwuN@e<OKvfcG1KT%m5^-6wCEYQ*9Xh`ox)+t~Lytv0uAllHprI+C-~rjT
zQej^D%6sac#4@8(8GJj2!wTjzZ2Dxw&P48EucEOpI!}UuL0AzZpG%iQb+@nsHB`B@
z=XywPrX-sXBJWz@P|9&3w_0AJ*~wg&Yn<tMsz7#NQfgLwUUfhcr7fy?Hb2|yWwnWs
zx<L@NUXYR={IcVf0L38-3*=J*smX^y?$iHEomP5QhCO|yD}DA)!0K(KYQk(#P=m@y
zM*0U<xAE#tjVF}UTzo^@p|fPB^3l<mU%0wJ+UZc@lq;YU?fE|2vpQIbnmYCgP=oG8
z*+KkBqd^WP{s-<(fw*s%&#O_G#yW<tm05x2+sZPdR~dwsyL@e7U+|CS>PB(6&0r{6
z?~ZP_%&MHvgRiw5b3^X`t=M5~RH`qXsDb<FE5NP_l9`a0NMN0o2x%%c@eqT^e&HVV
z7O_+SB~j3UP!*`=P}EHJ=o(N#>#=5zSRS-<Nl5%A)x=c5<Dwz5)WrCMpCaS_Xo_Q%
zwLMyRkLw1^OMpDQ5~0w)51Z?mj?U02yJnbaH8{_|?*M^JGIh#5|4s-LkpJhzRVqrj
zB{rSqsrl#jo+hRRzaP*Q(LJs|UwpB(@_G>sQn{n-O~POHo?rCW6aYuu{XO(_|M4D`
zb;>jWD)<L+(8%{&0e2G5!yi4QzHgjNg10e~gF?Vr4(9l@%ufobUK6V{m1#9(lz&GN
zhFuA`v-uQ$4@Q}u2e<z3rIwmKe&*(S`!s?K`Lji_=`b$xjQ;aV$)k5QAF+{&O>mnl
zjpojOk>(sQxwc-W2zUhPJYQMCTIz+Ef)D}1Vr<~?Oc<W?PtsFno_8$alfjH4Vlnae
zl0eh@?}Z!hag?$CD)HWfu$7j^y|Di9^=_~3m&Thh3Eu`<h#=bNn@Qd)qgLZNo}qWs
zhRy~4g&Cf!GA>OX@YBbX#$fDL+-W|pWLu@aldGxu)ik!*76Rzu(&F;aQ(LvoM4$Wd
z{gXEoG(sdDekNL2XrG_0hZJ1wS89__m5cR<FJ6`lwL-tVvC=P)siduV_$HkuiJPAN
zCIvZ%@U;euef7d|6v`{?l?GOO01nc<<9=YcezKr(DcFvux3HM-u*5=NeYTgcOwWho
zRjw%d9rWORz5%If;*IHkNs(_YwLc~%y5VkeeUFTaP>;MgTVT2THxz|qdn^0qeDXJ+
z|JTuob}0__vUD^kTC{reMPLIqJ8V0EwuB23V(V=K^F+37L}?48C7wrJ{>G6fv}Dkc
zKj5<9-|f6=#a{<I{d?Xlsl@S(s&O3{>p2&&|8FnkGk~oA)}up<po`x5ZhB43gHrUV
zuHy5CH@e`~yyPSbib#$&R)e|zQ=Z$|Z=QSo*JG%56KqEcmMTZRh~i8b*%mlp8pazv
zf6#Y|RRB{eP-IYjn+wG^(k`Wk#RUz<8y8@_B;#{)@4jxWwv&>IlC<=eNq^!|?uxfU
z(b9nl!GIc;Mf6PiibKvfC`ZMB+H1sJ7PrgYUB8($#tL}I^J<HhQF6|mto=_}70jhg
zeEqzrkgNuuxv_YTGNQR0xz>4ql4cOcGiu+>@>-?N4l+%Zpi7PUtZaEJxVe2~dZTQP
zkN@p5G4$-{9?AUqq&u>u8teYU3$h8t*?m`k2dw{GDX3?FaXh3td=bs3M)%;!_v9hs
zWUv~2sYF|)q?)@@!D=h!r)jU-yg<xwfw&^=txyD_$K$_{E2sQ0AukSB`INla)>3l&
zozh5$l3M`SSc2jSY}QK&ai7@=Iy{=mEtCrev+~fIN<F1`?y+1@v)ly|mJNozn^$^A
z4Qw28-Cn7n>l(R5=^U}+mz*>iGwlt(2zOdL-CuX6(6_>sOQN^7!XG=E;116x)0TEm
z&QG_LEpRaAY`H*+=SXBZ<ln+pIFUaDOiAV{GW(c0${`d-`VyBG_SW|T+Luc;wBC1J
zoctxHt8l))WCL<z6O>6rSd&w6J82yFuj36!*wW|fwrK$!OLc}~jhwAGrZuNfjo#n<
z$L{4?<R#1eu3r}c>WOTEVxWQWHBb*iCN-wI-C+1;?xEX!a5DQ&h#L+eQC=9BV?95*
z-t_DqVv8IxCv2r@+;_^p!FEE(6Lt9kMKcnRSOP)~VaTcH^_;vGi=I8y2Nvr_$vHo6
zS`Ry0-7(3YWiJvOTS3y4*|frDS{n=n-=6O(y1Zz~={`Hm8Ew+FHcBydWZ=FcPNPMV
zX1zcQB^Mm{D6q|mv=T))CkanrOsua=mo;wS^$ZjKMAr#;-cI7~^pu0yI>P2^WGl0B
zXNZn)Saok~^Bb-|K7&!fm8rQ~bq&2)jE|^l|6B1xJpGo5unq}${<^Q@b+4fqvqUMo
z>y!O~3U-cANZJ+VctTPk9r<3xneXLk$7BYHf9v<OG#2H7bojXY(Fd_sEI`IeYi@^#
z#B@cTw(A8xA0<4Vz~RVryxQ{Eb^__^cH_&#yU(kt+^Td~y{ZKP498`Og7Sc9(%4#a
z<kP*cU87aH-i$kw+(RCZd9k$RVeGQj@X>>i3ZMU?Wb_b6yP_=9tLJ~u1C9?y@i3=R
zs2L1CYg|wn(m5OS&7BR-%I^Fz;j?qO<j>!PgKFNkEXv_p-Jj)}DRFrw=BVsn`S1SL
znROh#6V6VVv74U*DjwYF!CqTzo)i9$8#F5>htKDc_fx8>@y?`6au_RY?y{lh{Z%wX
zA?bagWf5<`F9OlwBr4JIs9$c~YE^-Q7!4a2{`4vv_z<op;7_j?CB<@a`0Tv^n|QVn
zxL$NdM0PGa-Rd;Fg~G_^khcCF9nBmR8A@W((?6kT(*v&q7D;)+nbpKP9MOlf3_H6O
zKpz7^GE|AY4pA#Vzi3z;7$EE<q>-EIPY>|*m*)qYWa;6!dF*906ut=snewW;gy!`Y
z^7V}L4!;9q^m3!A9VA|&DU#FUt;hAA@EAMb+%HP;M^~f|gB>=Zcqs8lnO*eP{vSls
zf)`YlT)=lz-HYAOh)_+Y)4lyro&+t%rm-GI3yWKj`r!389|BX@+Dh;PIB2rO%Z@`q
z$-op$Ee_$o#oo#ghI%(^Of@9x$(_)@XUKMJ)iSOwS~MH<xg5)*n=qTN)J~}AK0hBV
zCI>5Xy1OOIm*IEApAA>*zdO?c5ky}JS0nsD0De-m3I3@WD)?mgOU&F|5Qy`Ot&Ug5
zC3UJ{4{HmDP;S`Fa6wHiyN+cXXvC%f3g`Lpj`@(k%Qu_TBdQqcTgEwm`}{EaZB>jX
zo*Nu^_OKwGg0d_#VE(XlR##SR%xF=MVi#+=12A(?3RB$#g;0?zV$B}D=~Ve3%6{)5
zpn#l(uK7=;BhpyXii?Z2G}&g=J8f%yY>~w5laRcL9QS-jSm?aW(fQQ;zu;z?;Dz=b
z(V9*{ca8IzdzW*+Z<H$Bn;C6Bryx{@8!jV+s1kUa6rJsgauP)H)}h%l^HC2LrxGDW
zDNA)0Zsw<&jN8ifQ&u>r9TNS0W9k!-Q|s*8Bw17~#BMz!$9FI3guv9Fj&P8tX!CSi
zAOt6#yYk(?G91{Qg=%iZhFc7^7Jo6-B@?0BeV(aYj0=8U|LLY_5Y4vA1KB=e0DFh2
zW<y@m(dAOMMx!yD@50=2L|mJ|4=SRC)xWTUc?N8>J5nET|Duif<;6k!)MFQr5?l1?
zne{-#7p$hUVnkLyc1|?c{E~0MoL?T3=AHNk)ZWHOZ3YxnL~$nsV=9o-10DFSN_S5b
zn^n#iW-8(22CsG}>4TsRNXL3hy>o!^#6+6AQ^$=8zC6~7+eU4r8mwd-7HKF=P%H9m
z_we%KFZqd!8<jFbXgSYy6$5$Y@$-e2wKdIVtAFTbW9=!#B^d{#50R>>x5}~tv{Tx<
zJ}Ept1}4WY;vR;^^)vAFJWFzTyzc}h@MPRvKaoCgh&oxS)IJ*(e6OVi9PJhc+95%Y
zN3Ayg?vWJACP-4rj7z!gbto=Oxe{hpQljdak()N*r_uxC1y1TfBv4+hM1%0yPIM(N
zGoq^Rw9jsCgTl=botY+@-8&&4VaQed8IFwI+nkRII}R6ssE+cV^SV&bsf9{PyE5EN
zjr@@<z6u&hiQizZK^(6CQ`8q+?E1y@(lDM$JX?;dFs~%gr!6?Cl|P?7l-s#)e5P-7
z^Qf*eMeW}3i`b)jU^eOwnR?Yi8~GxdEm!$Xf3{&Ns0tBn#lC%V5vtk-t3^$}8kR8H
zZnx*27=;!n$y(DbKq#-PG$uelAyHE0QsBY|=`;_By!J}g`srD1<jc=~MCiq&U<jE=
zqVDv%t^$P(&&wvkTg~B8tQ^(t9lQkLqv?(33yW!eL1YlsB8Fs9yT+#chpl6C4efmj
z-ZnGP<n4M%d(>&?gEl#^>QrGj8vZi}bwt-Z^`x%Fcy`_97Wg-mR?PQ<&_ry?Qbu<)
z-^<8&LwZHK{aOM<sYd^Ree6%)HU1wN4-Ca`n#N}4w%#K1`h3<4Db{EH@fC0?u7vZ_
z4yfE8Cf)=Vqq`&?Mwu|TUdMcW(>O-7flXP?$tqB!7vQ*13Du5t-nxwp+a5vMN~|ge
ztZee}yF;u(3^%SS!Oi3#MUwyVY)Mc>#~B>y4qw~azMmFo$gBLi^bSY#tfI{2ab>qv
z0?{+{%uil6*u?2dG)h_6NP=qmSXoZN#LMh3k?J+TSC6wZad?xlx+%rBb$B`<s3o@0
zwdoT|8>{6`U%=$zI2uqOJq3U7K~3Nv)c(0jtSj~j%t;J;T%WTZ;<#Ea39Cjul4%Lc
zR}vG|m)MTbi`How@{zgPFU9vkS*rZD(CRj@%yo9z&F7N0Vwd0Cyl$VdX!gg$<r+#z
z5gY-iZWX^582qfQ#UEIl{)oLYwY0hC5#`PM4{;vG<D^E((eK0OGYq7l&bGE}ER-sG
zKt4o|Cp<$D9Vq0;7M)IEXl4Ry3G<PW@j(}4csbqyuMn!GP?#_$%AHtK^jSm^aQs&t
zy{XEa2xw>)Gk-*ywH$=hw#h}fe94C?CINCnA)+aUf-|@IY0?^4m`ftK#1Jq%r9w|y
zQ#Ef(+(J~ez9+kr7ih7E5lvk>kxUuL?=WFjaKosNDe8tlTd5131DdO&^<6hz!yX9G
zc$F(YenR#SOMIiCnIxnx9UU#KFYK??Z8kox%c)tRE^H+Uq|>=SX8822grba942Ywa
zMxJP5Hp_|sLhLzPK_%D(X>+}8;NtLd*AxxA)%;?nQkcpOzG+qA!9-tZ|HiBTXUT1E
z85=)>WPz)l4r%Rhr;_^R7IFL86w4!R`yZqP=$nc@^rIhcR4)!C2s!`Vb9*`IjP=^A
z*(^h?$`U>cF<ccv+>k>hm>`+>fwnA`M4!B$Ti+Gqs3cOd+I*hJ**%hRcY@J81W{76
zK7V@0>!m?ba<zzus0h^24Wsqj=g&X1F$OiQPCi2QmrIDrH$ZY)$4&}uDDfS!iX+q^
zSWMM%1@wH+HrK%#J{x}z!Y&t6l{Y9PZljvS`of8eCK~<q37(Bz9`Mr#gc`)lDDu*;
zUT^DmZjRb_G7nR+@wS~&E{s>*=q*Yjr4J?gdxgBu!OM3$`N-6-k`F1-QmR*maCp9$
ztd7n@=aurE7f-8DSUZYRUUdbgyK3BDdGZgdP5NT99MVfhiyER|+zBqqvE!+FSnGsi
zTAD;SjW^Z>uMN3NOf`)GCQ>&GE5p^CWmZRnS|fol4d*)LZ>Dx;{26S*W$Z1bwI8an
zR(+EH@gg>h{aR(Fp44nd-NHum#zP+tx7Tf+eikH6kSVAqyNcT<KEs6#&8X&^H;ntF
zSF`CRvx8F=!oS?f+ZmLrWU~Fel2Irsq1l}00BWdH;#P|0?B6Q~I88_LUU@z#+f|u(
z-5_ed9olZ@+!e`>aNCeq;GI>>*$n2&AR7r!c6UShPL9w1oxer}*Eu?JS0Jt+EK`10
zQmG%?EuB>i{Nn5!%i=&JvzhYiZShl&abLo(WVciK^aD>IyFzLOaW~>Ptih&6pwG~K
zW#Q_M(P5PiSP~X!F4bjZlN~(lW;1VLP&s3yOn6Gu{ETLV^?_(*hjtNjW#V1~!kY83
zBYd-Fn~7iH)*5}LdK0#RqD(<|Xj-^IO`3t=6Vg7bC=sA1Bzu4C+y@aJdUDL7y0Y>O
zXQCp?4#nbshjf`;Cne~hU6GG#aI{{$_JS(s;vJW~+g|{h#b`ER5mxUtqF`GF@TVll
zP+y0->*4<NA9J4WS}zJvk8Jc*stYr!$-R2YCNU=6uyM68L7K(kq-#g+!5((wBQ~jo
zV=T}5377C*mOH}5WhtUR5FWmFtxu`9gpvV^0oj~R`x|Gn*F3}owBL~beeqQgF<idh
z^a+`l^9?Ay*C7z+y{yyFyFvd0)a!P6ptbtTK$WY-v9J~)?v|y=JlBMwHGR*)!FJDz
z<bhCX%=dY5TX=ES4(?-D&^I{7NGJHj<zu|i>-Y<?*YO>nP-2bu?rc3q$B(Q2v&=AP
zgiq2%Te9!4v=?8V1GWRct7Lcc@i)G%s28V4=MjMVmX!y{8HbWAuS1&fb7!H1K=RjN
ztD||j{BC>9;ObF+ksSAH9s##<%+<yI1m@U(`lV^+?l9uNmf(Rk8gC6$fbbb?C4Me2
z*=~2T)^!~Ld%nk!@W-R|&KG|xyL*7R{9_*3&jwlP>5$k-E~DD3+Z7%Orsmoa^Otcs
z8h|8;(cgT5UwbRsXH53rdjHhI@$vqCMEJ#ihc<se4z7(1NjeVce4ppqB1nE0D=OW|
zA+>r^iNb4@^+_dH?;{Ldm~0A?$tgn}rTchpw!Ptka2WhCm*Jx6pFt<uFIG7B_ZlOr
z0!d#bf8(6HHGwnPHWT&*a}05B-|yrYAYWbZ6Tr0lv2KV4F#ofJbw@tvN)pI!zDlUz
zu@<)IYi|sZ?jGG2zgO(ZbdTI#+&@(7YORgQPrMp)^M1Z_$*V_i_>wFQA%^h<G?Gf=
zVKf6hQw3@iAhm?taH`n(p<j;2ycV0;(}VGdS8#EiUJ6bE@mfaP#e$!vme!j@;KP0q
zO3dRYq6&%eLdxIzo+)G`Ws`JURh=UU@&YW6`WXuSPkj?EKQicTntQSkN&qh6hxm6i
z$~<t{)D-)tcrutDw|Rh^h^H?9)zO5qhv9{^9y{QhUEN*`HM5EbsO0MY$65D7@FnV^
zBsI!fM}+kTyYKj!Jp(TRXqyKxl89`S3dUUmhws1#3aCGEz#WHZ9QI#?M$9pto~MU&
z7UhudFFV0Vlw^xzEBY}H)dA6uFC;*-v%D!>;|2Pa*1u+6f9+gW52MIz0@Y8bS0^z@
z%41u+|6(AF+iIQ>w*l&SJQq|nY3Jt>ApZ;`KjruOt3t*-x*55_Xdo$-v&@juVwnHK
zHIyRT%lIoGioC^Y;XzT(h>k;a29>;=QIDJshT@UZjAiPMXA3V$KHrYD&qIKc$`(QW
z$|i{Vet;r=P${(Xt8d%IjJgr~$l3nc{@!YFNF{OenB0!3NJFjA`M7R&e|=e#UBnZa
z(^h7=aq8|gR*F)t2VQ6EGQ%WN{`-DyGN2;EYFwAE*-}L%@Loo$;S8Z*Llf;)Q2~k$
zR+2T&%2oYYnpf~(JCvK$Px|rbX64Sj%qj_9<ePBSJ5;4Re&J^7)hi;doENu7?ks49
z`dPzhlLiW{xQzPKiv`IyZ(rbSNl;r$OCWImbHj%vzLl*(2w<dos~R;8Z9;iK^Amh>
z9IRG6A3ws_<cYuDS7YgKXl|Ph=yD1)Gm1}kgnNxkES;CuqbQ_*{S13N;{@7ldqq*h
z+)h^Hp3%Vk$+oP?L>6+ImWo60FtJ}%`nn*Zzk!$kfl4PKVXObmH)_}1<nQ*F{FizM
z0vnt{M{fkV0ZV0SYTdt5BR8VCHC5w{L>Ve)0>9p9Q&1I>%`~J_oTtfgw0MQkN57cF
zbT&|Yt$*Ssx6SOsMH9h#zi+$UQ;#j<MiSGn^bAp!rS<jxi~X#`q8K7p9vvw6TY9z7
zuE7(D`eff3798Rd?d4)jcXvd3bs)A~ZdHq-l}9-E90=8XFQ9d)NS1S~xLT2V2{FRA
zp1?Q51E06P&HAe3jTzm59n2EWq#Kjnw1#~ekazv{mf*2%WV3%k4OUQN1IP*vB*Xc=
z<TSsaD_Qa1PJIbMB|9^ZY@K24<D0JQ3E8x!HSqU!IW0L$7wd9@=1bKkiXQw^nUD6u
z->o+TU}QC^e1Bk~e=;v>YuRcwNBtmy*#AS&yo;AdO<qilHWr1(d7fNAak0bwepO~g
zqQ89-jfyDND1#Kf{G<cvDO~&Y$;ZFnQ+#r#*}gmRYr@Ba@vyW7&1g??M+Z?c^_p3p
zOx15@%zhJS=C3tWYLr)Bx5wVeAc}-&+EwZTd~Su47};pQ<*7ERHLQ!quV#@{9?7fb
zh_iDXh%QT&#Yuam-*bKkLD*Z5s6#TI>ln^TiuUUkTAPDQD$4DAs~|>YZ+yI95_Pv7
zcfrx6VMn-Y%C-p_v7U_FQKO6!<<gr_$3SYZH*ZE(V(BeTzbaI-T|U<pqM(5o&k*3x
zeNq@t@H&{a8GR)=j9|9&i?$h2PXD956+rteo>QQOe4=KFo=+Z}M)HJjHwNX!i;M{H
z<$g6GD=i8XUPxs2Bp&``HyzzJxAG0Dhj8^#QwtS1sZh@$Kb(?M1`J(F(1wK$ZFRf^
z&6%`lR8|jXJ2Xz+sa{xtug#|#_(r(#*`#rv1$-mAF-a(F^t^6o=zTn+rJD*5>Ax*L
zn)!k2PglGuc))hA4y<MuSw3WeL<1_1ppMq<EnPo)Up*DOC$!7O6x1QT9lC7Ec;!M>
z?2xxZ=<zGV&v)8yQTmYLkh$Fx-pZI?r^?yQ9yd8X*M@bHRV!Wj$XMyAAz<&QM{R<%
z)hwB_iE8dsE2U#qnJPo0TEE@`U#o{)U3<@pAo|U0xu}Gze0AeoHnQR|5V*3IDfKd%
zHrq`ggwnD=_LYmsCo{)#D{m-NVs;wMry}BebTz{IC%pI`gvh4wl7W|vP!weuwfjN+
zgsHO2DALTVz;T+Ir(2}!9{_)|E~gzFERL=yf4tgwCv5cw|95!+SMM-b7&1h(&N&tS
z(Z$OW8O7O^5_$Yr9?;0a=Zi<G{?VA!>dhhHoPA_8miPB!w4)|J`_^H3vK9lKDm^bQ
z&j0?M^)hL7Tq*yAW>jTz<1q>cDINqntu%r{pAN`_6B(}Z!$%)d06i-O(bsoK2CqN2
zeV{Vf^)?NnyP<DdP?NpLXzmh(7^LzU&PlJ$yA3XreH$%4|I4z%cwj(M60d=2A-X-W
zL-w})N_CL6NjG#73h4A@RHnGmV<2fT?m3gz@3WSu`R7D?TML+6ovrai23NcLfBC#~
zN)Tj8OtG0Qb5{z20CnYdhjdq-*SLsD2JMI}J(`kL%8wJa)>VP?l&0<3+hKk}tCdPD
z<E6p<vMFvB;^j^FEoCztKFA5%DwuE{b$Iq?sDHbG7ZFL8`Chyzn84kPzhchFi?c(&
z%lZ}tTfR(lj?EoW&@)ewFwstBsT@|ZUcs#=$W%Soo2O?0gr;ZHLvg<g<tcKTxxzJC
zWX<lgw|LTgSr5=xU{m%0JzGH8()j*3@qW#l_U1bGtB)CZ)_-S~+VSiupqn4D`@c``
zI8TT#j)@nr`>(aTs(d6U+4|{xi8Aabf5?vmDEn4?%wE!EC>!CNUXq*IYbsU}sPWm|
z6!ae$7R1^_fz-Fj`cSw)tRPsAmhOvqUX2p51l>Cis93amrk@vhiBF24e~tCfk5KU6
z(<{(L6t%SH8Ms?GdjB!F#Qy<NG*imzl*AK8{AP*-m@Os}E6M-6Vhfp~?=y8sA`1tC
zA@a=W3QYX+;&~iY<?-9W*WhLOd58UERg$rtm<c>J(7jJ93N|DR7vP-CE48%Kj&pT7
zJ}I*}<!4H6lRBESSM=*!f~BMyO`Fq{E<zk{ppq{Zcfh7Gj)q`R$p%Dwh5z_-To;DN
z*F-@9UW2x2iM!BJYVUqlR&|J$UYnWDK}pq0n>glFd`hv$p5z>Kv;LgqH)jEc<<mvC
z0ovI&5!^xi84Aim|Mjb;+d9rO13V`&EJh&~?fXH`CD5ctUO$41y1Rejulc@x9%$<{
zx<!^Fv~t|H5YiF|sx5O1R955oru_N_I_rkxSY$7MQQ1Al51-Xll@YeDt*v*mZtBQ`
z=NdThjko7$nVKZ7#>DE&-d!XXKT90|0UpSk2JBqsYv}0w3=|?o!?*a^pCa9FoXrE4
za%67f*O**S-mRC|YpfJ1GlCQfCg!a>ZJ$TuJ~H2UpxOS~Gyj=$M~nmr3R}&2I2nDT
zKH<MN*|U^^*;cRQ?8IoJp<bkSY#I<haa7wou-QkE9#mFKOUo=&vR;BdIo%MB0u-+a
zi^@+yYh;B&=YR7G$8!8wXy8rc7w2O+Wtp%b7hIhD-5hnk{P*qRq`aoK29r>5>!%Io
z&y|BK+5v2Rm5Fk150VGZqbMY%T~%zRR?N*@A9Hqo-ARmEYp*jg3gvD)_tA+Bb~15b
zy+`a(wq)UG##()M(oWK=Wg5%1v;}-g%kjiXMC+;uLxxaHSPn0Hr8EL;U{B;L@pR0&
zmdMs7EnKI0|DU2l?XSi<o!mg4!^}q64^Sz0Cba-u+T85c*35zH@LxZAZ$b@s0pUK_
zi>p@bz6fQ|Yq#(_pWIq`CN4Vhx)3%GG1n0Gj3$X3Av?!cb)W=l)8`EjSOG6sxfhJ<
zX|m<Ick=ZWFE0O2y=<j$;bb|;6H4LqDj-AT+Cb8iS>%SH(TP70t{!z16n^2ST9o9M
z=er5b3F~?<GZnPHg7;J3($Zvh`kn_X?!3YEt(80cR#4T}5-G9xa#sd8>YuRv>u`Yl
zG_yJ#+}pPn5I!rI+VA)0DNy_%qR5vN(b?76`Dt^}aF`fSyE>{VQ?5h<>Y7LMHr!D(
zKBQ55%izX7-YEQzFPdI~#6ld=4w=M3pRD=R@F3q@p&4oZBCzfFt7p563IZ03zK4uz
zmm+0tQCQ{BBe4PPSm&r?lX182OBX_U8r0Pi?<5)Wk`DP^4UoMA0LTb(&&EB=yiNK`
z^DMcUdVPQDOtYcd-f`UC3VH`wi@Uwezm(}<Besk3bj&8_o;!-#_j)qbe!Buy?oKP~
z;aV~^W)J;u1*N`YFX3fXQ&0ER{gFG{@r9qSkEhIMTUZu!rLkQX)W6y5@(1#VQ{ao4
zb?bT+s26RY{<8NGP*n7wpTV7r>2sclxPl10$2DcoR5RvtZ$D!BN{_*AaN4Wu{OA~Q
zL>NH0dsq8Ung|^d309}55UsGf?uBe^qr4p-TBT$=LhlenMfy!ROn(u;I*>|(36e6@
z(cwM*)k~;VQnZITU{TXtoz2UI?WygbR?L*jLC`SRHGWn=-1r)OovIH#H&t4aP0e<8
zTf6j9nEX)VC(_m{kG0eVU>_NH{OnQE(ca8fqGrn7p!WJN=C!q@30NI?SDo+F)+QcL
z#<7K}zfVmhPP$Wx`asc=<hv(*cs0m_+6DD^{Cd!zK-}*!h+Hlf+w@%FuH05zp}3@3
zYP4h~C=cLHAM2=7l7HJ)sGuMW`JFnYD8oRfUvfQ{!|I7E7=c7qXAh{|WF5>BznMIl
zGKo~BeF+zN^8_+BL|EBQzS#ou6`H&E4DT{z&mHSboxr(yeNog9G2F85v{TP^K_MWM
zyLP=4YgeFeq+jq*)Zj$8cAjg2u|S8JLVVJeIvIE{x%QsE&GV9tX>trBSJ^dR%xoc;
zY9V-7Ipo6!b%MdC$WvtBO)~C;(w5eOk?5Z@*tMyU(Lj|mTO#+%M~bZWouA17XE@+A
z?cGshHc4J@D1e)^<au~JwKSAA!IG9($ICr48bS8rbw&Zl@)kMR8Y;yD{3EgqQnLZ<
za7FTZFKS86aTOLg4AXD3xj=j_##hdUc9sQI*Z4ayf2H$(V!e51XU;#RS<>~`8-R&R
z-foyU^nc++m3pfQkt#yUQ_#+?uCt%hg4z+|#7UOEq>S0=4zIiit=Fqj*}-0Rg4&vA
z?PsHiqRZ2Wi;JzFvdJdTdR8=<v)zFke-H0OcB6e;7O`0T!PX|uA220OzLEItGb6}c
zIYe5p9fej__{fR7Dq2!m<En(WEkCBeHAtZ4P^WdgHtBK-o+M^Oe4*x5#9S3;3oyWs
z`4vp@4jAJ<(L<J*V={c&Y!&>TFt&r+u3S&puBtYJaYWw&DOvk<v52W0I0D=~Ka_c%
z=A<PONhH0N&6PGuR8xha=0~kq=mr!z6-^^&rKsejP4?exi}$Gow?6ZreY%_a&Dd%B
zrL2*0rae#o7elxi%n^qn&+V&GdAu(7`8Iv#Hk2CdFol+^pG~SqiR2?8=&`XM%j&H3
zXk-38b7EAEbqm1<Iaaj+c%WSzMK7j|!y!jO8{~G3w)2H@9j8jzopM{Si!!WB(pusR
zMm{pU&SSW&0y1-$*BellHKoeDvVp(8i2Ss=Q$N}vW)2*!R~BYgx3<i_%CA91@tr6i
zVMq1A24&(}V;%3pBirp@ttKpKELm3j2hl#%-=htmMvuA7A)J6>4P6^}xg_2Aa=Z_+
z$qub3o+<Zdkf|b(>usox3U0ngr$K_BO5le2lI7dHnoP?EKNqU+ZZ`XdO?7CL{_#SE
zOG0c+l!Cqz3|$irub}PuvmM?NUlnfa>&0x&C;Jz!HNhM&PM0jC;B98Wcj@ROq7t`J
z)kApGzBIM4z65OV&ulS|J;DY%Ws_HOK<h?hL(Md$N)@(IMR=Gf%%t{O^ox~PLvZ2Q
zF*%xOK{lY`iW!C<eTO><eJBhe4h{9M7XPcP^q+ADHCSc;{dcmpDRX(gg(r`Y2~L<*
zlvCw_a!!rhWx_oDZKkL@Vtdkd&yp-8%a+-;wg-+&fzA83!xOPr%k)1~K*#bgn-i2m
zbDsYiv|`3Q853NbkbM~jb#@+i&~%@~Lj0u2!<KsBcobkC9ZC}=louirts^5I70o_+
z=!N<;717#Mj!!FYOKVpTc9^T1f=*{mCdVXbDY#Aj3)cc7BTs+-j&S0e4L`lu`Ws*>
zafMw4_^l~mK9pG5TnDY)y!t%zsbNtljT5SlH^bf)*@Nk*%Fh=`Z=g%(@VDIGcAi>!
z`1&RA#|*<q16KJ?Y|iz2^+(qilU$g5Q((C#Bs!+_^C|I3YDH30FeO{C5o^OwbY`=W
z&ODnTs{>Q&V8-{CbS~v%Kko&_mSgRV*@Zuf0!H>dtYqL`5+6qhi$z-h*LJjmJg<%e
zgF2mmF}o0J2`j=WC-A<c5j(rl_Z%`AXG?RwLbAv(GhQd_R4e#cnaFqdH`Cd&56Xrn
z*wNN300<A2WSTw;C@`tsh$>jAm#N?MYWaE9Wj^ih=x5s>-!`Epc{O)L)3;mdRxa0G
z0!d_PJ>9iK@u-`eIL)CETUyDciHA$0P#Cn~Pvmq?%N>w=?C(9*pyT7c(3xPh1<iRV
zaszu1xfxJQ@^9g7*D`w27nYwLW}Fw8(^jVtZ-pgy0@UMb^ADU3kC!SNeyV}q?icZ~
zgcj7V*crR(BAD;_2%Km#tGPo04jKW;zUf1LIkixB;UJYF--lA_POEmW8V2B&Q0ptM
zeNDfVY_8EU>D={Lk#n)Ue?8SqR?)f>@4BV(W4PJL<q0*8Ebq-ukFlmqrpFG}Ld+p`
zB`ZpOZF%W4-e2N3opcS@Jcc3c{Z?)|!iH<fnj9Xf+*foxT>OoVtJP}s6v_Za#?!?9
zCDQi(<W2~9q)EC3|NG$d;PkMKFoRDI8<P+Dd2?jj>dwTSxIi#19wC(IW4OQT7aBUl
z&+<W?MgN5HCDWCWNFHmRTn!zAc65#}`L39weYrY1I$v?D5na>QIIGzb=4tZ4w!60E
z{6Z`eK}A66(+Y1rM?sCZ*@14n4GJqbYjXZyO4qwx(>OhtT$N39yQ^y;{MhMjf?P>=
z?ZZh58&`*o<E`fT)!xZc;n@7VP4aY?ursL1n0U@1cd+Ca8qO2I0y7gQ1tEWF{P^hf
z@I>>Lu<FMSbC{<k)7a;oKt>~$_9*RtdbrL<-^&(fzHe>)!)Ca2;{gtmEOTOFV(|O#
z^6TEw#-{!0;p^XCAip>UA`p*xOkeN8V_KiX^zqp=p$Ky}f@{E`nG)!0OsPK5PTaw0
zPUZ^9eU%MFVkZ|Rna5I~)(SA!hZK-mzwD&3fmc#K3_&SHF&b5yr_(Dw`gXxjZ<R?s
z?2T-5;=ALnD+?B-0<Jt;>DoD58fQzs@sdt6^*`x9$|%F{C{G8UltT4&QqNif&lxLK
z0WR19m!<H!ezZ|K5cx71l<sgsz+in^@W%n$3ETUAPeH<*E1x$}zuK_wQv5fmM2$u}
z=2m_ObUOK1pHZd1YHO6FjhAdcHyV6eizK;AU%f%vwEznV0brK)+Qs?$QCu}qQVINK
z_hFE$?zj$P9cgy5ezxfu>pp&}-&314M9&;vL}Vhq9iBNn^;cn1fq$1oz6&mb-8q+C
z9?Se#ZBS+ZhuwNSY>mDoTYVneJl#V@9XV#Y8YbLFjB3GtH|<wV5)DvPqI^~{UCt#T
zA*uU5X(sL~0E?e-Uk@C2fGVZsoUd@qFmL64h&QGsvg#xSJjrH#SxUuIR*K^i@DS1E
z@^ICi_9$0IO7T31_akzDm7`Li;;3?jjKrtCq@Q<R#P{$>)ci_f3t|uk#Q4|`^VDNR
zLHD|I7=PNXVj4JAbTSEBsfme0{rzq#DFdINqe1OJN7tjrU6}R!M~FD`cHSR_;6t>#
zK97DgOep1&z4g5E9&>x8$j5$9Uze73Nl)sB@d=ED4Wrd25(;vy&E$TbP3MafXXow<
zxWgR5ahx7xvw6^FYAvB3+^=%hmEL_n!rQK)<R_kMm%B_90-yClqRu9Z&h)Nyw%F-S
zq>=qEYaXK-%87=zyDeY@KUi7Ri>;%I&>x#*_#;1uNX-XzkbL2Ub5|$dV87gN!s>%Y
zJ^{hKP;j8)#FSZGyXMwLgp;rDkrPdXj3;{Z5xhnnGsxKYmA9?*ZK+A*cp@qMG29B<
z{0aBa+;I8)tA*ND+K_vjRLdsecIdr<tiOP&-vo314_->4O51hyJDqbW(p$<ns2eni
zmxp>M?CQ=lDxt;~Gt@uQ)061YxAS8p*xAzW+t_`%YAi01VVEgeo?XM_-M;Q7`5ICF
zDj)WxnaKC2bR|eHkov#V*xRC|uAjC7zDuwzY7(OrUl?Qz2>+B{QDi({Xjm=0iD+N~
zY&J9}hNs8!t3lCmcKQaC-!t#MrtwJr{Je8KP@){hAB}6I%$2RSb;>ei9_s%ca!fv8
z6(??K^Rs9euvSQqs4|u-DkY-o%CkL8s;Anz%t)j7V{$d)>ZT2*O_NegzxC6*9*yyH
zVL^Rs#}Hh>y|RweTf*!7E`Rq&gD*A#e4FKJA*;yvBjN)q+A*w`Lr%F35><CFQ|YcE
zIode(R-kPl(wzM(pV+qi7GeqSSVIR1a^ymNk8QQcq-Gn-lKWlQ?y6H(x0y)frfM%I
zLz=2>ev6*_7=oFjS|9Tjve&z#)ee(tu+tsNetO`ik8oF`!IZ*fH|4RSnH&H@?I3dF
z^eI|7)O3s;oGY;K_*uOEK%U&^Wl^y17>K*+?aF`1m^;FZk+&p_*g=z6FVC8SYy2wb
z>9mVRna|JYzkcILz%p)s%+R6pS)JoYv2XT>YVNNquAB=wM3=4qj{m8~Vtt_gFpgGK
zBCUO@e__EBp=o4Xhm;;Y*!m5eoWNT<KBGRbPSu^BMVuV%?}ln_CG2f!60?)otZ}bq
zw6<6T)6yoEjD#lO$N&BXsHzLu)f*RPF4vi4CJlF*k?`d9w7;v}XY=mIsgqV^g&GRh
zNZhFA1PmEoj3=yr?`G<Ww00#tvw6^;h+yr3G09pS&o_03`HC84ZvhAug}_7mVM0yY
zvoHmgxEoBn`{&)SqBA_gHr6*bK7ALxk+S>kkMv2uru~7a&PIf#8XdTodZZ+={TTIP
z;<(7yYc9f3-<My2J~fH$l)W3_eWFi!7!cJ0+8>nr116&|E#Cwm;BfEM(l|KoUH?O|
zx%F><YE?VSX6|lO{vaUo`!6l4Ktk5|QuSTwdW(AJj~JzCaiLXH$JfENfzPMhoqADJ
zsI;l!FjWa@gW3l)O$Rk!6#XRJB-Vqpu*bcdluC^MtCYU_W+|5&l6$koaA_r7Odw&Q
znSh1E;k9P92ew@%Ya#bMTHw}@cnX`B5aG2mk5;BN|D5O<1prT&<b0WtXDqR47hJmn
zw`lsqgD$FNWMkzj@etU?2D7ciY){`Gne24RU??}8%4JD(%}<!Cy(?(3<R9*2ci)dZ
zTnQ^Fw7Tut@|q*#Bgo$Qy`LC$L}Vl!O($MS9kba+ZYZ@j&!R4GU<_+@Dt_P9_M$ef
zT0K<yMyAZDIv~^1mX`=#T<Y&vBo|HuaVL6mv+BYkJuNZ=B#t9*=WL!F{9aGfJ_~i+
z-0aZmeq3Bnf)l4BF*;+>zY2wP`V{VKk?YG6OGQDH-{-4tUznuP9Zk7kM8IT>72mU4
zk#`~$78Vt7Z;$MK*!IHmy+nWRivFOU-DV^{n_i+Eh&Q;q$<~wiqbSLfTOlBqSI!rO
zg|<Kit<;yP+k-W*NcplFpI`<*kEgVKk@RaDnoq2(otPencf5<aA;&D2@LwC8O(<*T
z3CV?}kK8-tBf6SpZ?$^I<xeWMVYlnM0s;_|ec&<Pm#m*lF+y3RZ)=V56Y+D+%Ca_&
z|A;mPFlOz4Qz;H#$(Bv9-z~|17{~RTtzAooa{CVO<Z8nEHCApz@swmyg}VZkX)Ki%
zI}?eqe71Nliqyqiua7)TJyhv|5<^Q%r!19;fzl(L@KJbFSk&?Q7i4{9%(&o0C54tI
zWB29$HnCrCq`VsLSZ^Tzp<I#4O<llZXvvcYu-NoMimF=t?x&J_t0be@1ppDDS(40c
zpIWrf_MCyEg?Skn?RR_>Se_qtGWB@U)TX8{f>o)*Q{!Ko*1u)59u)`c>o$(@`KSt)
zx>*}XrE@7q6{m#7uA!8z+T(&A!ILxYf1fZj{3CmJ(C*~yYxA*Yi@$NAo520?mujB@
zXp=g*VBO7JM3|v6OW~*14|mt^5@8fUn6^jt*!+Sxn6Bx<&wR4I|C22jGbC_Wtz437
zc%_NQo7v?x98{W_=xR7bdVsU;q}iC2#3FBgl*oDXQ#DK8R4QKOp=nQFE2AS5<ls3O
zXGZ+?3CWS)hMMSGzYZj}^Q>>*-nk3#?>K}vo!*^96T%}R#K^Xp=FOZM8oA2%-U(Sr
zy40YW{Ob9<K>-13W{}f%a<r+q>6m@va;HRUuxACS#tXva2D8nj=y}wMs=B71UhQf9
zWwxf;N%}e+U!vU2A09|d&1(;}_Z6hQ(v&z|zKLhnTNSb6vQG!CtZL0L&2Zz9@+@$l
zU+!SPYE~hkNiDvB_0=vk(g~N7Gn!LRhLm(l<AZBWY?&B2sSI>f=BN8IK)Y>MdrODI
zxHSxpTh<BD_s<kJYh6E3=g3M^qp2@L!!L6vWT@pdwuATrukHprWjx{jnd|N%sh{xq
zqkp0K>Qp<+vm1Z?_$m|E&xR$xp<n)75OSc5tR;~~Ews;$c4TrUa^N!59)N<YG1*(Y
zKDLuyj#s9nNRF0!lelNk!qU?GDHsX8+vaJ?&028$6UySPkmlc4ST6eB((6Ajxmw#I
zDMk8nNaE7-1I-WX-du2u7%yKXM6_S_z4;JY+)$;zhHZFh*_}^+fKc2;-YQzXNyxex
zsDjeLTC8;_>4iOa+rnK{KmkmcY=6c@yek<yGFDyCYXlnZAbMKnzkQlP9ZkWQ8(*5{
z*!$`5N5^a5e!lR3$4;h#bm^HNB?mYuT5$SsS~fSEzZ-9oiEyNC`z!-}?g@3(&NW!=
z%2kkr(FCVF3>7wzYvP*i%15r1w8WvCJYh9F72{cu!D{M(&IB_6^2k`gn9bug-IJ$l
z=GsVn<5){rUe531-{E;UG$A?~A}XA$QPr0Jg7vdzz0zI?x2U$co^c*<$J8Hm*3o3}
zOPnU;qnHpI#|sE-#U<&Uc&0eVa=aQ#e3KNcyh%1hU-w30zc5uvyW`Zji}lpD`u433
zB_y(Ss8Lc!Gi3@$5|?OoAT|d(i9E?qaOI=aagyR=o4@-$nVw7mj#cPUp7<}Dx*xoO
z6|zCsRvQ?L7lx2u#CzgPpm=0;zJukJ^y5x=O1s9u3vqk)UrQRFVO$^b>TO%F;KCjW
zdUiGju){<5{CmaXJ=+Q~#G~2z1%Y26|6B1vlge)UUE%;k;P>jR1=)B0hH9m}B3v0H
z@|ZE-d?>&$rVGwn4}+OZnc=+9l>hUHk@n7x{4~u>e>B4!eYzfh!;*TgWIQm4QKw^0
zrYprDtI?z}2`0V#gaFAfaf#pY<WFIua;HlJIFn|%^)bszcv&liZc#^e{&Br?)XA}S
zcf`3X0ILTtxE9Gj`r*^E72*U7R!ET^jCxCOv4~MPy2;a=hCd0DSCqy7IH^H5ZNBYV
zIP%yD1?qRvz1RGF>->6nZIE}&SShJIWp)vf+g-x8uV>4YqFHr+<yF+7nn!<cE^Mw&
zEiA|u<ps3p#U3WoTwQfleJ`jbnS<ZlJU9slnEV)?jI6*gmQ)7Mp{rp&HLD8diAE5<
zR@w^GINGr({*|fSa;VmmX&RmzF)^i5wrSe~ns2DQLmZWnCSNN=r&ReX{J9geM%QKU
zo}c~xy}qeDvDxuy6IxJu#h?lvtwY#)vFrr@!h7DAcjG}g-)9Ha`wz`d-oNo&;d(Jz
z3t*dt1K97ah0M|!vSU$7bT6W0LLB%B#cg$|v>C>?_008m-}8NBgV)yDUWX7&tJgiu
zHtkm3`B|<QH-0iP_T(EP1Dg?jV2EM5iE*E+Lje}reRgJR{3^NvZPTvo;y|7kB*~_V
z=&;Z&6b7<~#Jzvxc6TYZd^N%W4#%?tKWq}L3=TKj)@zt03m>R*1gNaZJ;>|}(-Gu7
zwW!T;6cH5w?qJQR&yL+I_zE+sO{m`WIZ&t1HY;1li#sH}q|Vz>5tqxX>V^MG%24g)
ztb^i$TZtQu<D(tkcmfaos_U>C;!#PByaHsO<O%Lwu2~pQkq8|B8U8rfzU3pp?h4|o
zjH`_ycN|~AaB*X1+Z63_g~njl!Lawdq&->wia=v<d1@t2ld*^JY4*J+JCU8@Mb=IB
zK4Qa{XF8NYMJc|hV4SywnM<Sj`<<pGqAWb>`g9u<HwW5xUS6r2gx7a{`An^olxfK>
zv7~CAxE`U^{eo|hf`S}T!UA-QLF$@?6^P0O#dOrx!8P0#6tC`!pY5^yX`isCdbf~e
zDEUT*6f~!5bT~V$y5#9A>Qv?hcy;h&p1F4Pn5bI0aHW5GR=O%vVuU8sYM@pTDHLMr
z^o_oO#67kj{^I!TSQ|K-0&=DLg>0B-^Oc_~y?0*OMET%T-*>lY11|Gzf7mtFh^(Nm
z^?y4X3`jytkD;&$;t2rKxSUV(g$NyW{z~<=_;A5m)*^j&ZZV_+;$aWDd)4w#xK-a9
zon1cXgLku4p{H$esmOE)?nILTuDHZRmVmKLauvh#T&j@bPDU~|;qLONez%qq6UK72
zB;K1ko^!haj!FY!^VWmZf4g#4VEMp>RVhaIGW|RtzNPT?8sR%V9zU9#E4b`FAmL%b
zG#@4T?YYLOm+bc==xP}#L}Ya(BGs>YZB#70@tem42@11S7vz1G_%W_Vr(jJD7N#xP
zR+~XTORAU~zRfOce8op(e>`Q!?F&)AmY5c&^B?n|k)bLeIJg(2T4*t5(?VW1ZMJXb
zBIDXHQkqSrSDrjwkSsKozG*gP6F>`st*o$0C#%v^&ec6cHA|8P?mvB%R59C(tRqJ)
zs=*NB8_9fS#W5H8r`L%5Y0>tu+Q+tVM=Y74Bv~n$nM5-NQj10Re5N{ydFhb(kq<;0
z5_sO2?+1Kv`gL%5$;OI9i485M-^^?Fy9QxhfAey3SJsv_la9G$PEkh!(^IPuJ}w^`
zv<|??C1#GBf_A(*Ed2l8?`4Yqc;D?B?9}t7PXv6^Lm<#tl3En)Xg}rUL3>!ilp9Fh
z#uokWEuox)vHGKx)y-%>8l?u&oA)wYKR$}q=N|(oCqK9q&uzOx*2{)8^i}v6OC4-b
zH9Txd#DX``1BL0Td^YImz7P0$&?Ds3KRxy9Yug`hr4SIJ!={G(8X4}~L$3K1WK;q9
zMOAEVUNg(#fcnP*dDi!bPQVhQe6Yej22oq}*DM-ba^<%2Kp?r_{ThT=FQ({}&TX_c
z2g~Q#zB_HZniS9fU`RqOQ`q`;|9V73sP=mS>pVl_w7hVR_F8OM)cHC|@bYX4(26`C
z%Q;Y?mZkWXBc@8NL{XpfNR^r!Xq_#i5HNgXMLV6aW{9E6cHbJwHr&sBDk+~HsB0>p
zG9sQ+SZ|7j*5aF`*|rU7OQSbRJnGPRFFfNoWO)wC&^N4`BA4}Ximr}(ygi1C>W67d
zuv1QwYy<;?kZ7s&$U{t4QqsqkzZj}F|3}k#hO_y;eLM(aC#9;kP^Go^mKv>H4Wafb
zEo$$1qe_%oMa&vMdsBN?)LyOC8dbGNgT|=+<azPWi@d&%`#A6GI>-0?e0gm4r|o=>
z;*IcV2ceC2*8Kf^fD`-`qF8$R?_~O^TAw>4P9uzuKVj8fL9c|{R^pR`i*u@)rxs<?
zY-7_qD8)1MH+|2i^}_eo`+6bY6{ZDy<@M7wn|yxWpQ|+kH#D}*RJr2f(Q1lu4D$SJ
zTznC*rOG=cJl~ets>;X8cL-f<7MS+1Y%1dpAIz|Viwx`y^w5g#6D&Uke>qlAhhMW)
z;Xy<TMoUE0eC<s;u3X<}9;s=>;kGU|7!~x{vQ`}?r{Dd0&Xb>xZ9;)Qu>)RDskii=
z5!M<1JPPbx%=m9?(bUD2H`{ak9ue|^X5`%#lPbz`l=QwR+geiMyTKf)rzUE|qg-`7
z8|KN7bpGoCz;gvbYx|@L;k7R+f5@8~#z`IEg*py^B0;D|f*K=CLqk#HfhNB_&kMiF
zVw;~uc52oePUWt>Y9X1doi=Y&(0ygpa1{-vM-bVfLlK$*tT&-_h$igyZMni<tF--)
zs!6S28>Gt2YIAd%`JDLX!<wjW^eWa-fzZl~iem>>Y>C~O{fAXnOl>R|6qc)-v)Xl6
z^J^BjRuEx(8e<W+Qf(Bnm)R!v`?qz>Yt63OH6fh6J53grUDsz5H)rdc9Gho<-PxnK
zFba^2rkSPLnYo!PT&U&0d$@z^x;t;K3Kk}AvcBCq)oE*wZkllGUjLWdd);{z-&}D!
zFN+7rqeEHGrQX#4-A{;=UscR2=^#AAh*t*h-QnpEA|`!yK7aOqKK<WXdqwjCes_I$
z^^AmdWsb~HoXU%m(D_;&c)IMiFjrEtdrUBY<dZ=#aM4YyH;p#?{D6n4_j~x?;J%`=
zNJZ{b4yHsl`;4os%FJ)?IosAGzX3-20<4ELSZo;{Nx#bbDFsJpY3gU$`GM{oCPPjc
zq!7q;J*RK8a>HNchZD-0N7{?b)xc_~DvafZY$_JqI6phR*pAW~Yo3M5q%-|?MUp)3
z*gQVrzm8AZl`bbIrSx#nL^QaL%uA_qJhjuhi{A9vo}2wx0)ye&s!W?1fzO1BiNEN8
z6J%o>#lA02ZGZo_S(V_d&x{L|G#9fL*`Vf&^#0=VIRf<G-bOCXhMDZKR<ag9=@8p&
z(QG*mhCox)5)Pdb1%?KPmNK_SQI9mkDE>rN{TXYzn?h|L3B2XO4cQaKtEyLe8(Xf$
z&_ok=KcdHaF)f8WIYOHNRPw<7=kE!tlF?-#_;M+FA7VPQPB2P-I`Ek!jMM<`=XpA>
zGn0l!m!(Lg)G-jy)$oI!ly5RLvM+gznv)G4{J>@vm6tgZdi(Yk*x+H+yU|lClg~)>
z!m`m!NtGF`H4E^V0jvzI>i8QL!4co~MjufZi>(B=MrK@*nHzxk`Cspn0v{)JNBL<P
z?cOJGol!yzD||%PQf*Q^OIL5XW5)g{vQJ0Hf@o2_hbJ4!ULsmFD!<QiuIlX<9!BpE
zo^EqAXzsTueaNRT!8^@KV#ulLYfCd+#!9HA&VprhrV{%jAIygp%)|-&D*of!uB5Eo
z-9bok9`m$Dx&v@VURs?QbU9UioPy3KcAHRwsgw;|Abx#xih#~;oDrCvgvtx!(r)|e
zh$4yzT=Bc<OIvFoP>+;m{Uh|#-)wz9YT5r`=Jq)7rsqIp?`T1qVxyULxqh;H=C9lC
z*@Dr3R?HqH!%J4A#2_4G7lgJ3?_JP(xd}EkTGWj29XF=vpekg@BOMK#Mew`F+sEe;
z@dE<%V%#UyPZN3{jXKKcB;+a?HIpiVDT#o`w-t%??wxj!yK}2Cj?2c1gptFkH1*2X
zJ}N8{sNY6kxMa$K&e2ijmmU?(Z9ltij^fj29kge3tB9g_M7$6VL$Xh?{;cGZ&X4&b
z(e`}%%iYom1XP)QEX474#%L!ww_ns_a#|UiH{ZnxQhV*Y{n>x(D@O<v#PlL+tT!Md
zmb%cm<l4MhLm*r(7#Y8X@PPAj@--mf!BfV1)t3L>u;i&PZyvW%1{VL4NvTT_*b4zs
zP5uy&U!w@JzArwNgZ`mfh&DlLDelYnOqnsi)WYg}YR0_bALTZ;kZur6DpGTFv%du;
zUjG#l^VYr4ym(bVs?U8em^~ID{uxN4cwWp!|9k!HY@P7)Y#pP5uM?UKJ5G{GJOL?&
z8ZxP3IqV!iLs<K{pNaW!{@Ib*Iu-xdG(NYBU;kS)TLGcA-#aQfcr`W`&~AaTFVj7(
zGI~h{I%oH6ei#P<;UpFIO6@rhT!W%Ji#YrmSKXLtL*0h<au1bIXJ1B1feRU%o8IeG
z2*jOy?=CwC8ebi6HWgsv*6fp=9399c-zqs(>d|i5-PLiakb=d&;Thtj=7RVpyx_V!
ze0aFhBw$uG|N3+lCYDraK0Li@>ZR3NUv<1YR}vjPB@%~~0eebT=>B-n&P#na7PBX(
zX2u!wOiGJ0!(d3(%y8}i5TX4~A7U3F(8sy;hSN1}V`HJIx<VsDQM1)gq<8e<GAZM{
zaJ|cv{1#3_!p~M~6stdNK4b1PH;ur|gxQGnn|WK7u=zX}Sfm+(F|n7|y7-1v%{ebf
z$tcI!^V#zeGSK=;hcq6JVYS|QshOSfwa*$3k4>wZ984b`ZmW?1F&f=hZA1e1G<$s(
zq)sRMdv$o!BEOlM&HCib&6_%Fg$2b6>}3JcHI3!I3i?hzUA5JRW@?$e>nzA+kCt4d
z3j@@ggHm3d5J8=tciS>2n?ERD|D|H5otk{<jSK7JFG70Gc|R%{6E)D%`CUqAkUo|w
z{p@h!ozd(gOHJK`xvc^=0fd#UtSr;)Q7F#7U!8DONZ~kJ*m(a5gl-x{CGwsBBs3aJ
zWiZ5lpBE%OfDXB<cA#h+?04;xm3r~B2M)L8+Cb~OsHG$C%)b~4$kyUdz9r^!^dZ9d
zC~c5@BO8on8v^Cr8s=1GfNI^NQ-vkhs)TBw8HJu}QNGD;k+hfvd_H1)s>+q9Td4_E
z$p!?G?8)A*ke0IUucX{@;U0Rblqw1`GuYAi-8Dtja1hYi)fL)lncKUoU&v7B0DHx}
z(6LlKa_nO{7m|N!5dlHScY^m{lDBPcVP0S^raxtkZgrXLOTBUH_}$&;^J$>|CPDfQ
z$x-yTu9gAZ`uA?Y{uCf0y9GYx72<l;SKRj6#o38Wv0y}%B4%i5o?rrkjF;u$c|YYn
zE=yk$1|k*#Q<HJH2M6Db6XtEWhe>|i-dd|NC$q|kN@ugQ)VmpunY4ENJnKG|oz-~{
zP1O?+w_rja4m^JSdh|j;EH*Huk=ebLms6(~Tu#HC9!3vP(cBwKevX>+;a2%1zf5yC
zCP-s^>F?tZ-*qP1cs9d1H@#*Na9<x!WSo8wgi8(FMoPrz4`%@Lf(WQ2d9AU^hes5|
zV|7NkZs(UB=2m9TXhS41=V-uE(-;Uj9*upZ=GrVW1%ebXKGU5uMWJ-DVHuu7@S#y-
zvr3btSs?o__AG7ytN*u?frMD#-rR$guhULb_9k2rY6zqTWg0@_bPh8z+d)0B9V}L+
z9DKIw(`7Ly_mEG_(>5-W@pb8A4ba!1AC0rcv!CX7o0^O$hRM1udKvU&Guq|B^sySl
zql-FYB!089oSsL7^<By5vnLF5TsALmZR{oSX=1k3z{-Gkz%x|s;2^<F1*;zwV3KA}
zV03h_(M8wAxfV)RWB!ahtx>-IcXYh#V#yG+H>lSUZ?#w!k_i1uv^4V{_7e1!=sz#t
zSHd#TUQ~Nwy}a3lMQsI{Tz1Vq^rVh*2*>+=mN_g|1eTUu?8v7&bg!gO0KBp?{EIkh
zi-L>d69mvlChpY9*?iR6fM?vJmwCg;S*O*;ODV+9(<GT@pR@Tk5h+x}1Wpx86Vqa5
zs-=6%QzRA_FuJc45`6j3Lxw&4JA6^`9uF@Mku6n)OGu|uXP1)TZUxT+_4>hiBje?c
zE`?<w;kf|ItW47RYFThkPY!=qEXDrGx6a~r5+eXdrL?I14_8Tf*=<#FV>KWxa0IE{
zJw2e=Wjejnq8xN!ITnPFM*l>mU$H>+LG(x7zLcs~PG+;FGx5>ixGz05Vb<Xqw=Cz5
zJ=+!c7g{eaZxKgaXfW&1(K+~bTF%)Upep7!!$%d~8>pr4?5@|luoQGt&tX@*m3Gq~
z+W@!v1teZPCR3;0NQs_0yY$ayPr$Z2zQVti5q5SyEEr+MF>TE6yt9cy$K2EW-3zKs
zQI90bn;pY>x9REB;w0XT>mb1H(r-bOq4*COfBqYF^@A5EW-y+k8?E-COkfaVdLm@6
zndij9(s)_LpnL3iGgmm5(E2)G`s1%~#Ol4VYGb;wQ02;Rj~YvkpYkwC+wJhClKT|=
zZwZ5V0+s@vKCNjk(kpRDbdBw*HqCiZXtL{IJ?N<MD@xa<!X`YjFPq0FAf|EtK0O!_
zsj=L@ah=A!n3^a<p}F(21#e$>hX7TJ2zSLX>cBe%ahA^Rd3pEs648hqQzGiS6=X%o
zD7Yht`V>((ULFm3k;a3SDNKReE=YON+ax)%k^?ZyrX3r!J~F;AE4tir^R7BUzcsgD
zi=y|`G8q%X%=u&x1&?Ng@0Qopj6BUH&?A(uFWqj=2b8b2OLkp6AiN-2PWPq#!;4@2
z+&!Z+|H(cNQ<st1<KhWp(y3)B^F7V=f!zU<wVbotG&zqJ1S!^$WQ{5`%`@K$hhUOD
zf$=Gck530|nBcwGMk!7&@qPqXVG49uMrU&XAHDx~g_`+^Qr|bDV+W#@Us2Vsh#puw
zvfd|i&FVG^3FzwRB0wd>qiq<5T2ss^WBb^c2<ENN*?%f$vvUlr=F88$0cUVhAXONs
z%m|v4bNtB2#N6&wn`H*8dG=J3YvfQ0U}?3v(BaZpma!a?(zza<5Amx9{k-!v<k{8R
z@(BMM-hcK<r2kTXT?@-!mzYXJ-B60XKW%n|LCbl2e%kvBdfddIyS1wz9>|+sQ28I{
zt8EFiNCBNaq5$M>4R>>~OdFYeBmRQ(1%iSSvTwKVVe1tHT<AA5<3?P3|CGH1(s>>s
zNF-usC3>u^2=jtt|4dmLP+~ImJ$X=RJ_#h%BlvV>)sG-b^jF~cREfFIW70IM>EE<C
zDlzI*Rs48xRX_51x|d#23Q}vlY_`~8{&h{04c%V<fgTG4B2MDBX+59x7f0j}#;3EC
zuE&z63b=RaTI9>?5rc8eF&97?9YjSfLeKnXdFWN(8DY1JV|*axAGQR0$Rm5l#<$tU
z?890gp@#(5uVRi%{=maOH7#F)o4+59;8YwEP+gLeu4tsBe~N8lYjUaDEiTPT$?;;g
zU1Vh-%`kJu&qA;fV~!T{1Oce9&ofJw@vP-?`Qn`xiDdy0KZa;A0w~Wkd(H{B5t0Rs
zR?2^Qo0v#19x|nV`99jwerS|W^W@cQW3`B!oCyfSh7B!yXQvK&B|krl<Sk4k`d^b1
z<dII8p+9SsW;z!`4>1U$$<$lbJ<+#gW9Fi(E2x>AokZ9as*cSN`^tAWtMz=gp~{(k
zVD&JPGHO^ux3SST3verz6iqd6nD1B&y*ymEx=G-;{<9qVZ$kOfyVD#N%tTMLIMqSY
zCbY(mUHY25lOM+TDk#f$fSKNQVEl!Riu)3`O}o^8^82GXrMVBcq#hoJz@CZMa;m_F
zW!JN#^W#=UXt9a-`f8Vjrdjk|&Pfy`!ys9LRc~P6DP8V$Il#(6_C!YDubh=@^O=;H
z8BB$fDXte{H^@(I9J}BW8ro_uWW14r7l|i_a3-ZtJxkaC-_n!S9l%~Ba3pLRu#-U_
zGcph{CWATI^_@GHgb2(nLfKt8Qen1Kfe>>c3$N-uZW7ovnh7MAvUQUu0Y^TP@s;bV
zVf=2{)vb#FPJttP>A-gv(ir@~pOs&ybvrd(-^u$=F$@scmS4%9YgxocTx_I3-(6Xg
zgpaD!Yx8@@zVbPgy0Z5@{T?IW1m*wf{QS3hhuIz~7}2}qY*gV#9ohIPzh!_9q;o`S
z8`^n{E(-N+xVR2O`~QFDA-P$jk*qz$|9+~})fuR;G40<Dmg@*Q#a#{TFLwrI*HdHB
z0?dfAyux!6%<gWJ3#`;(^cF5I#f!JkhHn!u3g=m$+YN3X_uZGsihe@^CB-R?qvPAR
ze`JPA^Lp>?C4MRwqRQyB7gAf4589Dudjq``d=qj<oJ7>uy-L>7vgYQ*4e;+=`k32&
z4Xi3(9WUQp`O(p-%e-%$HgaozC`#CGOe4&VIKTR*!+%FBU&Sbh;r0JsDU$Py%F?w?
zyO?3r8#^tnJ`fDo{eEIUPZNo*dk)@1=I6UQ?XDj;2~xwpexMMHK+3+-M_FYFqcF1@
zy3I~b2o(e|s5Uz5G!30<Nfg;U%bD#<`V11j?S^sJ1-}HY&eaJX2I=(TWCTGGa2{}d
z3fx=@-~8nuNdGrCOLk?}2BNVxT2)oMV!{<m7d>GObE{(oBbc0ns_RHaN}5}%`rmAr
z-xiYRsFuj@#eRr(0v*-4f=kt(PVTm8gSQ$+JeX;xBy^)aSd$@Yl?#Lm)D7R^YBsLP
z1mlKj19y=bQLC#*xT#e?Kv6MNe{H4d>t^m=$wZY)%yUlrWS!?)<M7B9sgRIPMMVoC
z%z}cO?5fD11bd<gXP}B~3prejmN1XR9bOGQz4vk`<GBV{mWawG0x=^sff?DwjI0v|
z+a7MR_qLJ2(JAkwDyzm8nj~D<Y1vlyR%i*_0C04t+Wp1$T6e2Ep^bf~^Vi1bkNZG3
z8~M}H!bjcbq>v@P+gDykkG+gP1a;gKdvxCjM4|uqZkV`!uU;Fi`)P4k)t^8!(4ttg
zQ9F(g&hs8SxyV=IIA}NY_L=V^Bwg_RD0i4FYeGWftV<T58Z!Yb*2oYT#h@_KO0y84
zwL>d;b=xp10%XDSfj*aZioiNN-T%opL5KC$(oB!!DG_7WOTeytcRL;*eynEeP)wT*
z+i+xxG!WwcHdox`BAH1FCeF+VlJMe5;RK7|^&$@-Z7=m(S(qY4=u@*x^7jJjgJ20E
zA<Ob*zZ!DMh!&|r#)D<zr@Rqbx+}mg-p2&P5&AD_s2;%x4npg58Hv%E&8~0OSUEYX
zC)<SuB1E+HB#I?yuqz^5_@*kKh4c8zl<=fQ3F6H1GA>ULYDSGZ=b7n3M`g492Za=|
zjNejjyBk$f(swlb7N_oEP;=#Dur`ZXIB}KZ5d*aYC77K2l>?Nk%=tiI5vUCkzg+k$
z-n|WVxW#3|8D??!&XpS-h=Bqj-A%$SdTuFCEXverGmS1xkJL(rndnbX8oM;UeaNo$
zI~%E0hORAxd%+*rLm-5TY$2>?bhmzVq^faPjc}KUu#ss>7b;z5{1u5?^-yT2a$RVv
zG@Ek3<=pNlbJ5P)9HCQ|?6_RHxx75XW2#&SrVgY5{L(4mJUI2YF0Qfax;!H|r`-H&
ztiBz@d-QP@Lc|c}r)hm7k0E~s<_DHmv1h>wI;yvD$|IwJ$Mn@?XvnI!0zovfvbxW%
z9DIIowa?ytQ?YWC_gSQ};;cz<!P7*z20AP-B|&?S`<H^S@B~32r)1?8>jX+9pcNnR
z)Uyq<wx)=sVYmM+?RN#8USG+?_vr`>67y)p@qu5vRuMLQmzpXTlIag^05yTgJY6#s
z_(#*9Ca(wBr2J@5)FY-AV*lmf{aYhe{y88EgRkwiZi<J^q^H;2%kM7#^}kgLP5S#q
zQ6f3dinP0<z$*1B<o0?VI3=9Qok;Hy^$OGXsa~Cs3YRV+FphULWQLNCe%o;<d+PZ&
z8ES*5m~>w%BaSI`Q2mSScpw^&#r`=qb)JfUgJzFxlGY0Y#f;gR%#N)4NaMulpXq51
zAFM24OX&95IS3h-CcB0jPmy`QBXh$2aeK?z`SOfL7ot}>Jc^G5S;Hb_bOV^^M(Qox
zytIaF+^k_A)YF4deq^|S23X`*^^=hETPqF-C%zQ5Vn@Qt|7@)lka{Xz5L>zUM)aQ&
z_7V8*+L8a|(b5GS$8zkN&(z=5sod)$LO14kfWwA$N~Bgli<itBe1x9gnZnbPrliCq
z-7-_7e#1#C%NDhwz($01nlXJTDLug-qakTF+eqbzE|s!yUVWC%^m#eLR76HJHIhbk
zCNrb;lW4uv`-OsCpgb!uk24X<(ysp8;05s$l@s=UrqbCTUsK+iKgv3Of?H5OHDlG8
zzO61FE$ORxR{eG!g`z#p`h90bfan?Yu_S%}9xI}+hYU2%q|+5fufo_{2K4~Ra%wZG
zFg`>1&5cV?IuqP*nO)1M4TkN8gPu;l0rh8d&<>o^B{VPm0f@ie`_S&xtjZ%D6qrdE
z&=P>cB`y`KhDyQymnFcp@G@|7y|R*M5J({AU%HR-hn!tDEN*<KL58GIemCa1{3196
z)_ldPt;c^t?dWsL*4UPS34U7T2Z@!hg`h8>?}-p<p8V0+m8EvVO?4ml2X1hxhzETZ
zO-xK$KR5ucr>w3p-B;~^OgBM160DrVA)6rBQ>ZZOlO?=xT=5e$O;J!Gg(6hA-N7ND
zp<M#by95k)Hg4Y-BQK-&VEgjl!M5LW$iJiQ%Y#HuN*)C64W}9x1K9hM5ymHk5a!#%
z)V(pMa-byyO>=p({ZgiL9yh6&K1Z(4-7eb>K({0Z3QKY{p6?M}6M<~?<=@y?3|IxE
zT=%Y;Jq+A(!)KxCm$%5fjf?}OkNdkA{{!90BQ}f><0Rn>BeMpnx0p2OZU)?=49f<`
z;P@-5iwSld-5m@oV)0LS#%178ILYzr$ga2Ol)u}-P0MClvL#0beV$onQ&qE#MyE`G
z*3dmFaZt6{hXn$a{4gB}*QDG33EW(rawvzS7e9+k;?qb}Lu$$Xx3+O`7?_0#FuUWn
z@GbKLeGa^BjZ{^TPJyl1ye-3uwLtPy@gnI9oD)0fQPAJqGo_&|A84D|bLaPu5*?YN
z15Er$r*LqdmUaPRR@@=dh782Nid_X(+^%s1*iZs+v$7iGV43!3)o=gaP=^3!Zh;W`
ztuJ&cIh0uYRcVR&?-DsJk!eoJd`uWnNO3yH&k0wdAkZ{Xd)xYa3rk^wHX9!-mjiXr
zz?}I~MP^gM2y*SpU45<LdsOQi`}h~1CkSdg*XqO<A{ALKy4a=}cxIpr@a#6V%w%A{
zI~@1B-I5w-DXRuYy0Cwa6NuY#oy=@=?sID#K6UPjwghphS%>{tTKwiiZ=t|KG)a;|
zLSe%6pR*-fezOA`f9(U1eQ{Y~c70a0w*5Qo^QJ=BYg#o5icR||qp?g8KkLF)nq&x4
zg*Wz9asE6tab!!O;qyr>U^dle;+%hVw=O}6h;wv|VRC|QYPB7|e7&@PeMSKIVX=K`
z^GELgu2-Y|_BocXMs7~JZ#IFORnHv15K<iIH3-LiVfk9jfImkgPIso^Ez8e!zo2q|
ztMw!SxwLV3-otxvl`;Pk)52^#AzRFi$L~^kXA>ibe(2VidNBiDcZXMwt}nXJc5luT
zZVo1Hc174*;y(Wf8?2Pco+y!0ZrCle9B?`8DtcAkHnX|8i6JaVx)ra4hpTmvn4a8|
zgu{7?uxI}WA3^T;rdS+kdf**V=gKf!dtsI9u8WD#uIG=YtRM6HJqP;74obp*d@MH_
zne#}?PGhs4p?d*y8XY2P5GQL=pmRY&NT3%u^y_f<-B&g>IlxTg%}9cq<(A*^s8CV#
z(lOx1@%gHZtlbq2CFLi-okyG<4wk#<<u)JFQGJj2GWR1=g$Y6mN>57k9AA}4K|*)_
zoDjMS<Z8dkZ5UM!+XoACsci4kk!yM?u$N%bk;>PH<N!)htSlqLFI3<$v;<UIK2rZa
z+D=QhE=BC9I^_4#udXNaAYyIie*6U;Ky_2-GDT4MH;oj{v<ZG9=YcWz3(-n}B4t=<
z`y4-F2-krRO+pj{dli4dlP8e=k3G)qZy$h&6y*9a5lMV8DcPvGkJ1q>r!LDpMx7#e
z2RuV>0Kyw5IJahXUtF~(Bn<Yb(77vQmk@xSApxO+F^!Gxryp%2*LI+klW%ARMeo3P
znZ06W&2`PFylj*TktWnxnQxgXFxj)IiLaRMgWbJc=ADsn-emjgxwDhM!4`8L{hscW
ze$LvSHAOk^Y?SITD%Mrw4c##d+<|U?%5blN7}v&6udEfPswe2$(~qU)DkUYq7>!6!
z#E_dFnG^M(+k_xJ8aXAgk(~>_n5YS$@mFMimTm+<Ju6jlO4PU6fFDMUa%{W^fqNbV
zgT{1|sxi~IeJod24V^!dS)XJUoe7saWwj4bf*B(+Do4v1w=?di$0h~bJlSI0-Z2#e
zG$RlKUFjL+*CXboq|d!D8W|1Ra5Yey*_P+ESJx*9^5a|`M_YC5sFD4wwp7&vwG`VO
z0}SzhEcxWcLZm3vLY^IUib}9o&~4SW*3L4ADVq);$jeLF$HH6uiB>jJ$89!CxZ6@O
zBw%#Jsi2D{%yNvQCd-L>(Wg$k?!6A6Go#DWnO8<tHQjy|6W(I;a4b8+F3v)WK?Y2~
z5^lA7?6ouK?#NmF4JHD4O%!FCSz=6g8<cgTcXyi!gOKy>la*-KHcE@=Y-*C0{{pO;
z4H~etKRfEgIC9yY$HP?w*MsFI@DFSUS=-08M9Ye1A0XAaZP59<7ue(DKUct?<FPRb
zJ+8-Up21bw2-nUnGTx$5%<9xC;X_;2E2$YdH+OawjB;>|DtB*9u^%cbfAG&ilKTBT
zq$A+hSI8r)ujCrPre#ny&aXd>1LCOAtT5u&n4A5;_uYQ^Zp@;uU$TJwJ1t+YRPn}g
zyZrj-P`nnIk`1?h&B1z0gftA;q2!fZ{i6N932^z^@N*)k7uykJuQ>~$Dj4K)(X>aE
z{m0y~(*1!hj%bhNgD3H`yf0kFm;Vwnz^FkdF{|-OhoyKFnAP&Hsjlh^@tT5k?Q#1!
zQ6Dx~a|0w?JW#cc%UWE^Km+TIE;Ac1S@)FdR!~${7Lwickp%pxM4?%fWu?q5Cl-{V
zl8+%BvsOr#*G%pMQLgIxlJ`Y;0^h@1GQ(vuVP4j_c0=`!@~$>B7TO~gAqI-*I^g02
z_Y}B3TpEa;#3#R`Olg~b&p+}eTq#7hbubit;i$nCMpBv;`>>x0lD$Zwz<KKs`XRCu
zUT&oQ9=$gS1E@qxO{yZH#BX~xwg?jvuD)u&LSE<6^ZK>bk$|qX2iAIImVNKeA2b<`
zHNGQwmSi=uXSq>Py1u}jq)9{I5`(%6lr*1B&4&)M7EU?-wv>f`2;ih5qRG#5@sZy^
zzRN0t=0MXaS|z*h${?-dMtN)<c#Fqg#j3HkR7e^!0-|N6Uxc-N3Jt4vZK<O`#YG!?
zygK@`#18gD20PrOk4IY>rqsVc!uftZe&>dM&ijrPDdEmDSwKp}jrZOvSvuOt3Hl{b
z9J<plN)+}$A3-~)VUdf(sKjZ2zYkl~p-iPT8HphLMc<1URP>+u5E(3L?NDUCvO5V3
zzW=1&7~s3Z)emiBAONerHLyhI@3%7$YYxSxbX1S!wpGTq&X_JL(yY!bE%@bz2L4+~
zm^-CgJns1%vrzbPq3Dg{p4`&@##QR97zwBCkp0I^)1sAQm}nmaO=kbF&Xc3$a07>1
z!L}1ZeP!ptaaP=kocRm2$DDp5GxoAMdI`e9K(`=Xdwer_Gx<|HcjDdj{~+e$yKM}P
zq}-2bsTri!v!o}v`NC14AC~yJ4|rdH*Cr~kQ&YQf&gJpT>x%dbX+?m1xA@B8#<kN?
zy43~VK2{;MPAUQ-&4Vx`_HNwqx!<FtPFeo^MMh(}*xmfHvT_ZoxX9{Vq)yxmqY4dv
z0N*z2IKN+fB@*{N@a^SkC1pAZ5j|4pVxb}-ZgVOl=C3^k*YtzdwN;oQ3RMB&<QzWq
zkK@;E&vPhBGygluY41`Rx41Dzh#)ciRXbfkr+K{a(*&!Xw#+6|5i?^k176Q5V;8Cn
zzAx7NX-F%In)eS@87PRBdJK<zI6G(nyG6t4N2af8cjbHz?$uNUed^4Q!kPV!ko8t=
zR@M<Er3_n?xL*hgf$S?S>e)40QfB$-SQI|EHz$AsWs^iR&^J4okRJ3Xc;}SQnnJC|
zD&~f9M5U{x>h@#Bw=QTGwT1B6?OiWvxqNZ@<rQfGL->yB)?3jU=C?QQ)u^WdrgH_(
zjc}oznkS4BUdC>@Y8N6+Ssb<(z6-7OU&mu$IpzqpuTp)}%3JvwqMkdeljOX~I}hKu
zKJk_=D;A<f=Zz9YG%35&B*n=3SO}pkSr&)#WjaRjnV8K!gE{g(h8D<9QCkhaSX;!y
zHnsjPs$b8wKQzBjL~E#R_?Q#|0=$k)#Om%ZT??b!xD{F3LbtwC$4>}97}gk_rsI#l
z@^Ac)fkAN&sf{wQbr*q+6|KQhuf;W8lexI+=nL)=6+_RfV?{X|aIe%4G^SMyKtHDi
z#NE8_lZ0(_N9$r}!woP<jA@^oNK%l=%cT>NDrDBxGKjDL>Ub%OE>{V-I-1y54xP7H
z0ws~S=}Ukg=QTmX!%1%Wm7PASN--bKqyn`C-5zQ^A^>PqjfY*EccZJdTDv;0w+V=T
zLV~7E62hHC(wKA~u{uXeOa^K`|NHlNdwXtV9mW~`OQ#2I`xEEUxYIU0vYmWqW@k39
zFCGVG2*%fRjNeIJJOFM~`4*zM4gzhT1kF4<IF`6_`EpaeDPI9zK~2v0hyUR?9k$>G
zeFq)Ie_s>ieC$v-$;67DJUw($?rN0{t1c;7W(UkPj6vS2b4CUlk<?+WYfQxCWK;Fk
zt$d%Xgc1g(j%EI|v&XQgNXlena3;2A5sA8^S8Z-^(Fx(bcwn=pWU6d!bL8wradGn8
z7ad)L-K>N%Dp8+RZ>5iD+c_GBz_s|Wd{TXi<{QN@M?e3IYjVMgl;h1VdF2qnd+ny<
zcYEFs>cmDE#ECvyZZ#47BADd{vvu=)BgMxl1yTe>k_jyByj)zOnV*cZh45!^X>g<4
zWc5eV+Es6xeFDtWJTo>`f6|maQSqxl+CeU`Ez9rg4NdB+e;A@fp#xC6I|lDY8AwDu
z7G9%8M5O5(T3R(8@1Wu2k7-ADJ7!H*6z;$r4K0zj;5?(}rWNKwyCVUzIhg9@XXr5f
zkerIq*6N}Tx!CYbJtvlS)<tPfS35Q7F#78*)dCWt#z)t*f?rT~6ICyRD!%{=*rao=
z_Xf)Gim+IasIw2hWh`q&<-OR2$1lM5;eOB$l6Oe-y<Ge$K|TTnO)pf;K5vp@bZx;b
z0Osr~(dbohE1Crq1x0Fc@rpe6g2Coyw{M+Sldl!Ki+NVvW|K(_V;obY)=7doSIfQ>
z-!;J49#L+Nta~d>jFivs-TGx#B@Ut&CoNX$>xslIQx-PamlpjpJM47T_X6p=RB{8y
zx3kLO{_pblFM7bmI}yah#q-(yRWB5PBAd)=SEZ|?O@iaM_vwKW4fn0)IK|iGPrT@x
z=i7od?VsKD6aTt;C^4b!_175YbuXgp>6GgxZdHcbZX#wGIQPG~<~kXA-5%r?O@nys
zs&-qX#wTY&xclk^vx%`9+nDCEZOCkR?U}=zf}FcoGh+IwLVv70EjwFAh72SDyhOa-
zrF{Kn?srWJdQROTXo*7Vg$|3=6?@@ZH=()i<Fl@}*Jk%CyOAi+e3k9Ol_~DiQlQk^
z*p?#gd6WON56_)W4>nG&uof9l^@U6B8>x#M6Tv<>6(a^mAED;5mvTrpJJL)Io{XG(
z%7y&+L#vkEMiCDAF(1}E)V^gM59s9Y*6i<pjR}{H%)}qAm&N8w6@6uK(q0Pb8nsw1
z>8c6H5iU6k{yhmR3DXH1T;|n*pvyR_UH89M35fr9MIEZ7v}q4ZW!&%;S1CYn6)_Qk
zoCKZ6mq+VdtE(Tpt6F57TzqEsXY}#uGMuVv<$YL;tu6Pfj~)i(AxpNfX+Ysl>*$ER
zrBjtW`;WSEEk>1{w(3|-e~U+-W*?&yMBhTs^Aq^f8}&ux5x4W3HsHkJWi>^HB&#3_
zbYe`SQ&p<1JNfO3@<yrH`>}1MGv&^+jsA?a<<+fd3r~}B0n}W9tpi!)U{Xx&E7Waz
z)hBT8Cn%SS?sV5mdN7Z5AHblGQqvlHC@&^9rTjqbEu<`=eaXFLUU&6Ze0diOYl?H-
zl%okIz~aucm)@SlAY0ETV1rM{orekjQ!OlaBPLmz${62ATHLYg*0Z}q(l)DP3)lB}
z1`Q8L67o%okVTu8`3w<uC6~WwOww_SX*c@T+@s|k*5UG|8GnNiEe6Qo??MLag$TNh
zLQ1rcprkQ8&JHbV7`pVD#LN1Q*ITy&a(L5_J7st#_5M4YZo`i*5^}#+jBgn8rST3a
z65>u*pU{0NZ4z-<3F=6kX=l@Rd#qHyV&5RPp}P&ZIKUE?bB-q5l!NE3mLsPnRYk4e
z6x}UtF{-*Jj1O)#2ZS~|Haa%!KPHsm+XMt<jGR%o5u6bWt^3g)%q3V?zfJ(ZvUGhy
znV$J}Nt!z->$5nVKJU2&22U`5+r&)ZFIiO{Z4p1CPEu7z85`79IaHyIL`3P;^+d_u
zJ&SW9X+$DO){IXJ0<vZ=cq6*Ig0du*!Mc~eX*cZ-b-)CmvYa^cuWi1$IUU-4#s(`r
zh9uLw7)yfT=p@&g5<j|T!Wh!u%kTK;OX$U2NM{?0$o*HNXWlgFe6$cvkr2Uh0^hOx
z$7!hGn-B8uw)eGM*Pg+S-By@w)>A2|8{e<5L~V<0=IJAaPF6+4!ZZVD8`e;2EL!z-
z(rQj7UN3-mae2qV!7GU>duVPr7px(K2^5wnay&)I+)kCRUmzPQ1(hL1A}$UcTt3->
zlsQDOq|6QEoVksliwU_OY-yvsmZsKJUQ7mx0M%Bpw2)SDl99pbq0bCN+>~1bJk-|}
zCh(aw9s6>PnXzd|64LkOJb1aydZsNI9c&K>V6xNU=d`Q(u>`f~D<8$N?D7=xz*`NU
z@o#{cR1mh}zY<TPD+CHQM-II+@n6FEd3+=|{gA4g?8F~8f=$!3$0;lIW;!yRS1IW|
z(WBFQwez)7IP1v=C)1TKxw^PsbU1c;AYiN_4Vl)*FCu#VA7dbhV6e0k2mpTKy8i)&
zUkiv-^9<{+hZ6?u6mqk^(w80jm{HK%wF{&nRpQj9w@~H0a0oObyLRpNQk$eNb_*S1
z>#cqtJ;*c@U~+6ehmfFwsg5?v`J-By2OPT%y!AARj7rN0B=?5gVgkcS9TgpFzJcn~
zu{RNi>$c<ELmIW3M(WVEjKl&MV9#mM&teZvTt|DceU97l?#P=D91pnI2?q^k)8D6!
zgUpWqo&B2}iLUve7Ig1pLq>i^SwoWh)T6Q-k&|fb%nWMTXKP30HuI87x<(<f3gqw{
zhr=cSgrL8<S}Mt{-seCncWH5q#qPQ8m=<kx2x0Ks-odgQ2_t90fwBO5Zm#gA*N{p1
zw5eL3PDWW8YG=~Qf}^_HZTaU?h<}h5aOa}SCYCw;rMQ_5f*JzGAqg$+v%^D7b^=_y
z#mLnZ9?kT;{3T{X$lV+303|(DdMiQG?>LTRu_T6|KXd=vUEcmq7NZ+fpv!dAsY+>;
za|Z%!R$S*$DMqD@1HN-}nw)SEWSCKGn<oDIWl8*E%4Sn6&Tp0AJn(O26}xc3LTu6I
z|9!trD7B-M;=vl+-@^eN4t?*hJzH*qIsaI;D@Q)Id+2{DcGM0<?wo|h*W52Iru!`=
z3y*jF?m-=sv^Z^XA)52F#K&S07Q-g8>EtB4xjN#(C+TeglsO>~Fg{-d%xr&5I-Q!+
zW2^sa>?(`yQ_5|aZQLg2WtsZSD2+gv(G!XLA}(4XpK?jk9IEdaNUhBZ?Jl4wETx^{
zeY&4ds*>24d*fm=_|QbMYQ@!bQ=u&ZLHL91l|L63Hq1-m{9s4{5`#p;Cd1oo-nEPL
z)d>Vx)NcU95Vj)dEzw&9cV{d@2NNL4kN#d%t=fcqR<#B%UlNhwgwvPc@xd9SVe0SX
z#QvG0a#&xD9UmSN7V09oDjS91VXqNw76#G|RzjqU!uW5@+UX?e6^{IOE&uDaoBJa4
zZYJdwldTpaJH5Y0-wRK7{hwN(T1%%=M{9R`cRQdSc`Q~DII;52GdNgINtuJFfg{LD
zn}t4>f{DvTkZtn$jmtL>_*bM^38=xY#`Rvd`fEYkV&v@GPP)bWQeN-FdnWM6;Y>_M
zVzDFIqI-4+f2&;$mZ6GaN)p0*D&&lp&~p(@c2q)`RG6c__RzrWtWWsZjFOOze0Rz8
zVt{s2^Ea}eubs{h6F3H{Z+y;=@xoT!p^7;P4-M&lmDza`li&F=Mnd-9=xGhf;l}6>
zumf0C0$=`MKHI-%RJ5pi*VbAEWI*h7)Wb*nYjBkRZS}t2ob*0$xqZcPeRQ(Q;cMK?
zb-dtstLsWPs@efLyB-{@#FkrBnUn`@aRjOw@5({K2Mu`C;<zj=mKVDNE>4!BD=rqW
zOJmpP4+^V^%ycSK8y}Il+~N+)m^IbgELlAqQ2@ryp8qZS9TiZD(&Bf`O8q%jDCI@O
ziCm7Z$lBe#h!)xFR?6v8ViWetS|o7KM2Jb;cQsu)xBq~z9-uP2<=C2eEVT!?gMt10
zYkA2)%1I^47t1HgYHL#e@Wz(L&fNY!`M2O=Uu8~CPn&Z?3A2J#>`bUaw*nnM+sTDv
z=HIpy6dPhjtIp?hhi`Jb&0Mpqr~@q*)ZiEAEJMyzM~C;727_$m2}_YnW=l>yj4gwy
zoym^eDIjNa<e8-SqvZjqAw!?AFMuhcc^Vo)GU_bkWKvmbWwF~Z@UfIAx|!-WC)98C
z+WpkpRF@RQ8mz>@Ra0DSN<=hP9&Si7Bh+uR<yAJ4S)qYtiahCr1?(mtZxk<g+*?Ro
zfc4DG%n&7JPYz{IhoxWeFD>o;3~d1}4#%(q(Yx&x-_=P#FQcGbZRwRN!1F}e{^=4|
z^qY4V$uJK)@s=WJIOirSIW<=iZ6ag|Nqt*`&n&fL!?c{Jx)t=cC8Z{z#Qff|CPM{t
z`ltyjiN4IoxtfWRff9@>6F9GJ*Y_Fcx$lUdaHG^2qs>KHqmej@UgSwnPw@Bm?N~_O
zAS2wDE#oZ{1#Dc&*~jM$sGj@h20%UL-?uyNT~rG^lozob02`^-XX@uos!KWR0n0}I
zZnP(Vl-t@H>v-Wv>sQZsT|O}Ks1BnX$5KB#H6aL*4ONwbM7EZrwP%W^6OnrtjqeZy
zmsF=U2VBvHB*D_m1jS8WhV=Y{Ts~D@+|N3z8|${!vgVwb3<mbd=l=;ts?}*<U|rEm
z$AmKRqIx$+Id{YuM)FG|3G{4Chsi^m$jxnUf2~STd~w__e#dBRCL9t|y8F$Q`CKT!
zeT~%|N@TGFHc#S={7x8DpKsun1_<f=R2H2w&JDLJVuP;9&-p|7y0~$zUc#<#b86(j
zjhV)@WbN@r07CyoD4ZO{8wpierfc?d`>D_Y;IFZ`lR%lhsqf-bk3YSVU8GUaaFMu=
zoI4Iy2$jw0syy2=0yX3}Ep@zmy)0fZi>%*he3xF{Zl`0@&9S_=yjSgd(Zx;Nhe4{L
zmNF3|zdE?k3qRPairz_o>AP&)UOg^s<zHWZ^5|bR@Rt4ktp>KNUyaSRG!}0RKCIkO
zp=dnlThS_U(FCi%3&J?51Mm-$=1KFxwYf^`Mq`ib1s<e!!~|8_x<s>Hi1ND10aOzR
zyVg7s4MB{(?Au8{88LCujISHLuFrh1X+7Ni-%qoVEVrpcq>m`VraKHhdQ9K9q^^KL
z3Tz3$M9JAMep{kuUp+|`EsA)BB?F20!6Ye{a62D0onWbwHSf2#KB~u--->S3C@Url
z(NfAUw-VCkM~%%$*N5hfl^s;kF<I8!uPM)d6P_hQuu6oMk`yahEL+_C8#pB_Dh@=`
zxzYMX?+`Eg9}O~HM3#S18_B%5>Jzn%OR)w1JEXCA8d*?Z`DqoQBV6)<7<cmhG$bym
zhkMF)qpB}Pb!g=9@84hr`EL)dM>cB)535<eysARgkEzSR2mCE}0tCIvBEoGY3eCY#
zwCp{y-r09%p1z+_ZV`(-i~DLgAFjWk%U85evAwcV&G9>Vf=nx>aJHv%C!nf$onz-e
zAVHMaw(8hZjuaoMh6`EVKWV2n9H%NL@}kaPnPYo|a<0;t7TKNkw0hORb<rA8UY0=w
z5-0kLEH71y5K?)Pq)O>n*fTkvf{2Xg;;#o$T37WV9zBXhz@7~SWDwjb))xawgfJNk
zu`H_%S=4!{;{5)<AeCH?3nz;#S^Jg%62FiFL>z-}sj+diBdcGK*3}51;v&sCpR6+$
z@%fPE<p}SP?=n5C&B!21grJ{KS4eFNNYxy;GFR8o^4HEPU`Pj2h`FEhKjd_e%h$dj
z9KNR<N|%Qd1gV#I3-H7d^zQoN!SV|y>9UbvB}D=>;S4jOe4{?Hhtys3mIEvZV5Xw|
z_O4jbM2+`J(OtR4S4&yZ7twUh`^$dq7F%siDH-LB*^SNo0xl$E(q+2}-PwBG7D5Sd
z?YO>lHMMd>80ybA5W=>8++JxjK&VLTwqInQg7;x9%g>Hhy|UEFA0vkymE~h*VUS4I
z=`^X=wI7Tzy4toku9dGb=YUJ<jS|hK{GZg6>@LX#SNUeUk=O%Xw@W2DS6)-z-yTeE
zZm!B*fS;GomVuhFib!fU6|pEOk5nonuLOx2Ef<Z&#EMx~M}}Dy3Kdf>pMe~25<1>F
z56$!yB~$U_@xjjJd*DOlZuzjCD*9+HhQHb}pnlYsCP`u~R6tgY0wg4hwmm;`Bw$v}
z>-}NfIdgTgq0vO3MIAM@(Hzz8<>gK*?VLGt)C>bCZA9J4<4rl)plIRr6WY(83B3XL
zT;tmYUZT4O#47Z24r5s3r14H7S*fWgd0Ea2s2wDK7dyc5^LnKF@*i-`0q802(N3+L
z>?F@I<~m~5Ky04@KHlG24aX;IrnVT|{(e+fx4F>71R+Bm_OUMx_~<o+7hC@MQ<6iT
zdlZA0o3OXALq}Z(xA8%vUknl^DUCZ*hruQz$<B;pwua#a6GetTsz{KV-6J^ScT~`#
zrM+8G3;f&cl(}b_x*=;gw;$J$YDadi5D9yf{L5a3wz}J!(2al<sNpfBR%rdga<%)+
zBf12G`RN4~1RjQTm7Nq>dvMp***%W0wBFT1<hlAVe+y-%jaKxtK`CFI@D!KrlFF-j
z-zj?-Hk(;th(L?ds&er4^*mVjtSd^xx9L->Y5Xdj_1HS^<C3)%l@}hFS~W3D=O?8l
zJpNVgIYHuI{BQ2&-|=Ozjk_Vq+#l-#v*cllN{X>8fQ+J#l~_L65GKh(O8^PA%|9i4
zBUo44BLWfI0;8QYIQpvkO6COHw<PnkqciZI_vtm=)Ujf}4Zm0-%)|hlsXnr#SF%co
z_I2T#!)^}X!d_N+KdNE(!bS&gugSmS%1o%6Eie&TP#C?uOcXRaKi9X8BkV`8^W%WG
z!T?;~E0ik?Q$LO$IlDX-Q#Q`90~v^+BhB;WK*XA<>&aG@8yJk&r@gFj5>C8Ke7ocP
zd#vl8StB)^>$>2<55wjKs5A+ioZK_<6yhz1E|M^sr%%CaW`QSW_{rt-;7yf@rHa<|
zMir)Ck_u6JlMz=nwPQ`=Ueu0+uv=OL7MFkG`bDGiP@l8(4Mr<7oEN;mQma+2wNa49
zZ+tt@aZ%WG)hM7G5g3?KV12sLufuOC>Bafl8wkE9nk3KlWinx@A>k_T2P}Wr7D8qn
z$%OVyih2Oojw~!6Nv=>nwdz7@;OlWES1ax<ot-pXAQz4ZB9^y@?tT@#C?|ek-WOzE
zUT#Db)&m`+kPtbs=;)e{b-L{%T*Jr)8GXl@EM2rQ^QbZDu@p}<H&bRY6F2G0P=Bur
zg64++)|@)~i+)igAbCez*!(=jHV&5b8O5V*qF!i-FvqZl_kSn-+3-TGw5YOrqz#0)
zy#not;iJA1d8s(DD`(|9)kO?E5Fxu&;yHwfi)<_<sn8UvPvQNlUHu!~_c<$oq4QYB
zWp>E;ZL`G%Ds>F{)q*ITSRO>*&(N;a7CRHi8<T`4{w+hz1%{c-H5$Phy9Z1!r4_pb
ze764%mfiO}{{5RNn-Du&;a%*ysXczH=}0*%A-K~o7t0x;)jz$k#4%uIagwV@qmM<q
zIOho(n%z@@;P1*~D2<D^F21)}u;IbPfNCJ1Ndm@Le~(nE#piALNFQ5JPPwY|RPXjX
zYO$K4pFUFm@_A9WqTI<zLRdwuq~|N_j<9S^KunTg<e;eM#Apbn+867%_1dx$?Xt#F
z`zM`T?bXjV4XJE|*DlaMSoA2cC!=+L%2oQO`F=^hoj0+a$YTT8=^r^Z<nn&a`Su3x
zs^sDnP+07wTN>vqMLEL+@KVbkH)S^o6%v7eeV%)By?^5y8t?+cWl8dSK7=wqcG}O8
z)nNHPui=cT{dfeIH{p;cf5!%DF<NH*mjysu|3q0jw`N-j0hrC~i3hP}j#;c<Pe|1E
zK0AN?;c22hj#3-Tm*go<_ibO2h__qbkK#JPHY_w_*E<B@GgUV3TiyN-*-8lp+?q|3
zUw2jL7OwGQZ}4mrem?#1x6{23YY)AFoBsqWHGd-XzMqSuZP8aEixz}xd(m?LG(;<o
z(+p>&KA*M>{O)>hZr3`z9lV_@L$()4_SZDF)M@u6r`k8u!s-9c@B9vzrNuTZvq6Ru
zlT7JN+cz6J?#o6J^4zQof-+B8%TjoiU@|M@;oR>r)tKUgO3|uI;EJ?{v<MIwISIl8
zBa^3d#&eCcOQF)#b~{0m(gOub)W*`Rs*uJs<nS2~{5YPrBC{W=!byfvU~%W{YMnPu
z3NX>>L-nM7ZA(`;XmgkmWFKzOJf<0nikmDo6@ta7^rW0j)n-Li@9Vvk&#@fFV@x}Z
ze4>?awo7c72HDhLgK?I6cla_q{giU=iuftM(Y7G@d|hicTO?4VrLM+zc6RoIx;=F6
z>6{6$(lKj49+i}M;b`?G4gdGZ^Y<hd;Q;@f{;q7ZQRA!AN1ECq`^F^$sKK4g<O-I6
zABR4XFCE~-xKRcJ%GbSGk)m(E7fp*kk(y!pF}Bn3Pox;?+wSOh7b7`W4j7}?-27P^
zlBvR<LAZpqt8S5b#h2w)2jN1`RKmg!H=LG%Q&st1JN~HBCVS9g6H?OH@q3L>qe8h6
z4AAu?Y{a8A;9bw_Ol>)N-34Vaa7r%U5W}j8fz`PZJJ^zL0zcx8<JK7=_dAWgxL8=Q
zx3r6!YS>i)%^)9g`l!zBKM^+*#&ifgE{tsCd<!kT?<%-R=&gL3Z3`qC!M7@0p8%er
zp(mLn=?E$^6N5XfP5Jk%W2IHG`QUyAS@r5<SPDeuK*rnnVtb?ZX`Fp0lrN&c$6lM-
z0L_0i<=@f{oMH!*FY)b9m1!;(%sb0PN!^WM&ng<(6G|q!LqY>XgF^{_=jUg5H`6ri
zcTsZjENw=9(O;8v7I)f(jzXA$=unf{%|KJ<&nZz?;RXqPP4AU$v|ocJcWDW77cpgJ
zz+~h8DXIBq9jB`h``9{F@bcfp|H%R1FElNU>1@byuOu7BxsI-CvuLp*J{HonS2*|Y
z@~Gp@)$OzO7J-T~segox9=ne&esFS$=3ji)^=y}kl|=O`J|e(h#r;b^#sp|S_xJt`
zCoNt2Bnm~aR+4QGROU;1n~*!x`Jofwe{*oMavjR-7fzplOY)II6R~DrvE@72MuYLT
z*(t}5#a$fLZ9)y-Ttun3PU=>nqm^18F~}XftDj{2luTTRauXZ^0^8=1@i=*rXc1c9
zU0hcyRNpbVbHG%<$0%1;xw2kR9F*9h&Z!_-HSbE|i-+<aT^YyXp?_EQIA)-v#Aa(*
zq2p>3IJ+ZyThic81Vj761$nA%@kpD*T5KZ`gE&lUd<T*p71}ZXw1@%Bhkt8MZ2H2z
zhe0-oDy~74ohYK!2xaX22>gFEeS0+1{~!N`B)QC`TDfkdrovouzvNOra?7nEw~0dL
zI-2mN#K^VW3CUff<hr>Pq0A+73k$hhCe0=H-~0Rfo!^dsony!A^?pB|&&N%U^^6}&
z?V@br(XRGUkOSbSjQRaqA`9wo^uY|p7xAK0&=Sq3XfQ3~Lu$T$oc*$r#AlTR!3<d_
zhf6x;a{Ax-_4Sai3;TJk8lqrYWQ$?%XtWxSgug~!+TtR0cJ>7v!p?aj=7k5!q5KpY
zC-Bj>g22v$LL(c=u{EFHEo5{C8n8biRM=K<1|ZN15THD!;jGRQX=oe$1##Zr-lIi0
zF}7hu8bc6t+3(Z&7u59E?e}S`Kd!IM7F{Z9FRci6h~LIhv*`zNTvY5M+GxG9vH|WZ
z{_gEtw5V0Aw6({rTafNr@%&gF|2G}*j7*MRPZ@KAm=g&7>r&QNYFBcW=R^Lye(PwX
z-Ur5}GJ*5kC#rbc6={(f1YKgVnj?l_zXA_t^RI!q2$KeiG-SJ6!*>79d(?YoD#rS1
z#~sMMGuYM23@Weg^9ybaIS2{HYvkr8q^G;s$5zywtX0c6`>1`G{o-V0ZhXK<$yk?8
z99!p7Xuh-ZJ;&8bUdvd5gNd{f@;(QfX~zelPN2Pa^I$Lff2&lg_LOFb;en~()yBLJ
zLw;G{1idcb54NMXhVF2~tS<Ry6u&-i1ns?_`Ef4w6G^aBW<ziGb^&?O_S6!sON>`}
z8?JWz>1vx1PmOgW5-D$~Q~0tUWXMKa_veA$^cxFmj=7!=lkezzoujS=f|eWzxo3I+
z%EG|FSa>)*Ip$eud93#~<hN>bbMs!vlkJJ&St|2CeY*@EWwyEEtIB67=Co6W%9<zm
zAdg(msE15l-cmoaEBK7H+}R7(F#5dZ)Zf+OuvYpu72KnC`WmB-C4{b&_>#1QvEG%@
zEf0!2pF8mgv-sz`)e$}@DK|p-Z9d=u)FM-!OT_6yEkf*CuRamVDl*i(U?G|p_PQ91
znF@WvtP_(en2sbsW3sDSKm5@(oUJjNdznsU_RKCi78;##5&s49VY=e?<gMD_;poSP
z?Q-#!=K|SFyIfwlh<{r5ds?jG|C{`2mnRB)E-HI3s3-;RKzU=Ga0j(eEIkfj#35`4
zV_o!iQ-jm;Nm&*vqy(6M8S#@|DBYP&2O@tP&E4iTzF3KU`m|*6jM9^^mL@z9%5)=<
zmi4A#m~`8dB15(}9gOuWE;EIVUD5-|myq(eUf<T`AX-8BY(LdJv9=%myAh&cU%i`-
zcQ8<rk6m|=GXPZkK*q0H5bzH$zj2Lo#k}s{QQGFITYTs5yGa&E7kHg%A4!`jO)w@H
z8ZJaZ;)a6kxvp9FGr#mEmkM=a;o&<<?+k0dl?Q)yz5C*#_@RhoV(YT+A6(@PhkI5%
zes^W8Q(wFYK6UDQTY$>%tvwOskSzpJ%vSD+qO9`sPzo{S1xN@yqZbgO-+|`O1S>lw
zlFMSCa`Lyzz*gAOtkViASBCjG68uq|*7+&JaaBCj;o)8t25owI8K1lOX3aqVDO)G3
zQEU$@tfJrl#lW*qE8e60{LMZ=^|^=Th8*GPH^cC@tzlXR`(lpZca`sVP%yMK?QKE-
z%PJv%nWoiSDcs`J0J!aqzC%f-aSa!Kt_IyE`>sg5&9oXVNDyU#SsEj_+YeV3R>{g@
zM}HVpGjHv~H)6cuh(%J0g>@V5_2MiO)Y&@v_4Q<pD)GAX3tJdb6>Ya5EK&^JQhTu|
znRvaG-AFc%mUlmdyJ$1C67Z4)30f#L)TVx_U-o=a3sV77nwpw44tINi{Y}{;Qe_uS
z(E3G1!5--yfnzCjrJFDQuXG7@K7^D)`mi@qbwwri5!)?&@j`gJb1S*Iv9YmbXMR)`
z`HNBhFE}n}BYpwH{xjT+=s7by_ta@Bpm>ec5oY$_@-Js?Xntu`(QEln!YnR%zrvI!
zryQC?{r3RI%;5n!=yo)#g2GjOC|?v@&xmiLGtp2u!-WHsLLfW_Z8=r4pnNgGeST_H
z2+*;faJ>V5r?67>m6g7D@z8VgZ|nayNBa*phwVq>(cB1}4xxj<)oLkfSM3bvPj|)d
z@BY(1>^jO>EV}8g9q2NZ$SRSXCGqurUf++?pMZ>CiV3raoG!XZg(r9inUbE_9%fY@
zu)4RtyRtB|N^Zwqh}QhvvQ`W}q*l!@Bg1{df>UrdW$|N15E)h9$u)L`RSC~4_MEdS
zYy}qOf@RiLw#d4ee1d-Q2+Jw9QV&8ZL1wWBkIHM!1>3&nv^sonnuQ>C(tlX2y66hz
zniy2%B+9<o_;7QXaW(L@P3BFgKkF-7e;NE0K|%%}tM6s&yj(NKJW!_C(Ud1;TA|BP
zDBFe4y=fg+=HF#YOp|8k^lBRa?L-u%)prOKG%B^JMH9laV>}doIXiLchZ;p6RKB=n
z?eW#aNO0FVj%H4K!}&}Ars`3~#5V(=hQ@lnwUltws>Se8v;SMIb&Tyct2N|xPxYMb
zd3nvI=84r}UttG_KIY3|NKKk_(S67RU8M(7$|)WMxXa4!PuteG8PDl}3K{p4W_Fqe
z_U|&v46kp%x)(ye{%Cu_0ePNbbXxyh)bi>?P;|T1NKaMk{yCT5<VvU<L2n8up<xXV
zAPLV^V8?k-=Ow>cCoJBLk8L2;I#xP{h3)T7{M}^EGad8nIWym9*6(N8RWSINZB+v4
zX?7lOr9Nw|V}&tB!^V(jtj`;=i`|zj#IL-#dG(Y(OHf=&l&{J?dzz-k{-3Ty;2)JN
zcG%EL5@^Q}7kffL(u9e>y9o3wZ|v`H{M$?>?1fFPZj)(-)2`P_pS^tPsvFOSE-It%
z$`r$Oxe7=&s=YV0?P*#~!1lUjRwd!BIyUUIq&~Y<xZD1DYfaHSU~g3x?80t6+@%8<
zi*W32<!r{}?+|yI?3G=!+OgTu*~NaxJV#T^@3Ez(A>+{Aw2$Xs08mz6!}W!gUB*J7
z^6C&5%(J)m$6#@)7ymY=`1?*?&sQ_?5s=-in%#s&_0zj;`$zjzN24K`W{b=n4Y7Xa
zeD!7exr`w4`s%^v#NLGRwsKJY=(h5nyD-Ney9+1VF1<eqg@jNW;W1gMb_U5DnKau~
z!l`xKDJ3Lh^7&bA;8EGfG_LTDfQIqsa2e*kZ~Ge?6RihbM@F<1(mF{@nZfc0cr*Wu
zHCEy*s`?g%rE8W-N{imB*yi>A-)qfRdO=OLy!J(zSXUU(Ol~onb$Go}^UBu5leqaf
z1dk84_})SXTF9-<jm_h0L9>6BdGmPoK;hxc0WZ+8Kmq)765rw3^{?D5bjlTZt6JQx
zl7Kv64~-#2Rrc7UiVR<2-yAT!NqLRWp!`|N-EvQ#2^oH0Ecr4`cfG5%?{lyKN_4sN
z%pR2C8gq(L1TPkX9b~GVG96$|2pj6yM$<dISL2Yar@gFdr@k;2%g8m`q4!wK6FLlU
zMf6vc0=<zMdFf53%g)8eJXMdewLYbA?2pZ;2}Sr^V$)EpG(!g2t7Jye=OTD<+!*7y
zXD{Z$m4hr%OeIy)SKNAR6DY~hqB~zJ{TdowWA)2YnDZ7zhwZ;p8T9If>%P#UZunU8
zka$!;MklR?S9lSxnG5Eo;%;2<gJ2+q*U1t<aikC(-3iw<uf^Gq;*^wh9x|oJ#@V}V
z_b3$vS!sm1zR#^oD?c^%#;!x`wTrQ^&9<0e5D7JwFK3-euH`tjN@}@@!hPVKY?#tp
zd2;`B>4GP91p0e@eP`-ubrCoiok~+!_|b8h?-=JrnFgGkSLg%C`yyqLCZ$T0!X|(Q
zqp&b+Nae+$;~1WZ0c$@cYV}hgeqUo4W~qy|xjIa1&jxMUhyTn#+M;VrJ1o?81Y{h6
zz1@F1J&UvJ89n1Fx8gKb18<KiD=XLQrQhp0Cq7nitvlSF#TKLSyI7DlysA+v4}7wE
z2v$zn(^`*IR;w=Zu=l>?owzFsWL%bmX1p-ELQj`%-3M<@Sm?gNBW|t*P9fsMp6j5v
zGWOX7hy!JR%mls3P<a%62g@_sAsoYkP#*{zZ+^lA%hvnNWXa&GaI~cB&D!JvXz6BG
zx?>Y@z1IB)>2zixvAtm3u2uEwVJCQjf-$>mM}N&^!Bb2H4Ht)h0WkUUHZW-XP{h4U
zE>J^K>h|cbbK!82$NoxC2*g%B=J+B^k1(0=wYs;|0f9Aiv#D@cz}Z)vHKh)1mz^4r
zrQ2G++w71lNzwmH$^~|Irq;a0u-Ko%mDrfNCefOgTF1xQzKtQPa)JrV4MVd#Jr>mx
zE{0U>xz5+%Og#q(Mmv+n6Zb@l(6?8~*mLZuG}2{ohdvL=W=yyIR7HWkokf`<!S<dW
zYILGlC5Y7n9&fjc1(HROq31~&dR{-D@RfhLC6a;0a3a_y{#u@9*+!+wU`KjQ$nM9m
z4PRJ=d)tz-ylV_B(~4sIZ}xdun+WN2b&S+DC|No!0nu-kU1Rj_V@2a^K~<<lU0m>$
zRMlb`oJ;G(>*$$oM5c;(1=Y2WE**C#b_iK}an$&o@u#deg_+@R4tlO0bEto(QCllE
zX?@C$kdS%KkMI3|!|f^=R>xn&cvtc+o^+JynWci#D+?XJV<?AXpE(dQ<_=dLcrh1&
zfseh;RseTgs_NCE7g5d)WU?75=CMqzl~-8A^@o#DcNahqq|n(W=z`%P-0|d&n7IEL
zIzCqSq3-O&;&pvV#<1<g)dpxcC_BEP5#-r5m_r|}GRx;5tGJ`yQ-z1#b~9GZT3(+Q
zoIScZk-cg7^8Vf6pkxE<UJA{n_g&@6bE@!D2=M9XNP?a1U)wd_YXO@9)(`IM|8vWh
zLHeE1BH($%E<uV}L=D-H@*%fQol@P-Su3m_A_{VTPip(XM2%MOoS<w>j4nEc;qjp;
zoW!g8>7<AilMiU}!I>@j(@TOZkZ_$8OW|k@?q{r}jnFTGdrO~c8!}$~S7F#8jB<;^
zj{#Fh>tcsgazD@j9O2cP1fF?Mi5(qG9fEgEc9h;!S;BOuT|F)q<y=3t25wQYL1Sy%
z<O7xaHCbQpq<`zRtJnOT@k=GaBlvZC$s%~Bn@%_G{Yhq~l(0VMQZE1GT_K5(PiB4l
zkXE-E5VF6u3J%kb>6?OZKOUYItmih6m?^F5RAP41tG9BLm{M8yGct?^(IJRgQDx^z
zR@v~<VlkcFpq7{+<z;(tjiUV*c!oSeRl#3v?xU|aJu=89ZC-q*o{vBt=bF5R+E>TK
z@~u<%_BvqQ#T47oO`|jKq>vTw&PZ?T-9;L5FWKHSLKPu8x%j59D@hbMr~~ILO%d#p
zg4WAD6z?Kkj-LYmjZ}`#A0GYzh3V1JnrQKv9lyIHMP;RM2oGNIIgVHba?@0%MxMu&
z7?$r7ocIa+G%*N9(EB!vot!$W`8r{dXoItw19nSL^S<eRp4_T)zG}i-?)CnC9!k#@
z=dKlbk!f|5o1_uqJ6BDhl5R;va#Lg$XSdgm{sn<Gz>EE2Yg54C#<G8ziGkgkc<Jz)
zE&Hms2!y31Pt=&iZ9X)ZQI0e*61Wxgaq2Oy{>rIvKKOvpH=Yy!4Y|HDDZHyR3P;H)
zSO}bNOK9go%?Rn)*m;dF&>PdQmpM}dJ<sVNIU}BM|H~K@AE5x>xzcJI)FsHT2&vL;
zp9L|75jzX18+}3~>s4Hdmp71reo7{!RO6D|tHzcR!01;=aZ{75(oUnG?LF91h?cgd
zCM)KfukmE4e&-^rtK9sBQRtJ_{jl5p6^-Q5|B+V%a$G^%+pd@FP#!;-Ge6NznmmIy
zRtyo5-3WFb00GJ0QjI%%m>4Xfz<TDGH47Ia3eNg*`>O8R5(mEoOCTh9{B51_o$MSL
zaMYNYpLaj^S#bfLERkYs#fgmMH!b;CM!sh1GTnx-x^?{FiebZL=8m65tv8rbm78Xx
z=jkCG$FV?5U7BW>Y0f*5q;{G3ikFfh_eF&pugr}H42#Ob^&Dq8+N^7#ncGz#A9xPJ
zJ887N{-ce&qtQLyJk>b6)Y+Z&QP3(ab~N9rp{12R9_1deBp+skD1>cM!<0?*z9r+k
zRn3K;Cf~a~cFx9J43pCt;nI}bHzk?~?EGDQocX%_kSZp3zkQ5rr?=`adyFL84bG)|
zUfCf2tM_gV**ol6|98=XCmOXl#~9ri&uY+1DD#h~`D9UtN6|<0KM8XO-xd!%W66bX
zxw1W0aZ?zIUfiEK*rO?Ky-j*(SHCIWU&36bk9s?{YNN%>(p`W2yeh}T0nfCU%ehkv
zoiy?~+_L+3qxcSkIUk@ZB3mU(CxM})e>VZ`!(c6u-nh(#o{x34mziDLV)eLH*3N?&
zIg>r-3=9QboPlJQ1eF5!rQ8xw{Lxx1K5Na7IHwIC@+|fkw}yEVOAq${_;d+JIYT~!
zq4YOjwfQPrObeDeR8{qxfl|n$<u-j|n(3NyzK%9sTqP=lcN=6*X7fC}IzqHw1Bbh-
zJJmzAJBxW*tw2y~E1=D#cW%pf+)^|?sx-yGIxCh7J)aSk;`uFKio(tJ$@4A<aOc%P
z$=!Sv$h+qqZfpa)WUveK--iE#=fJ_Gn(_RrtjCm<)9X}e_l<8I`)-O6j(+5(;$30w
zyEIs$FlkO>cgxrJ$)1IYH@*TcvAh>-$(g9Hk_BUX_*K9=-d0?>KpgJTXcczk9=dm;
zS;-x_kRMgFOu`mriB)+fr&B{Ta{p&J2vWQdbVFZHZd<A)hK#@O!jN>hP3HTjcK3_%
zOs}<kLc2-=M4XtKRU5R16L-~-viVN;bH^#TN7hv`wf7IEK#6r05J(S{F0w(V8ZvAr
z*7v!SMH*uR@+rKNo!SA24tQ#kaPq)p(#82DMml|S>hM68puz>}CDxSr9&!uY3!iW?
z+B*<Iq*_eY<P9e^)Y^DvrWc(I(6lw^fJ{%1+!%Cw809^3nxQ6H(EQz$vg`2_k<VIA
zMt};74AyWisvNKy<!i*OZM?)^oKm<_H!Mv&j;BOqS{C?rm*ebQ?d&~u;P<{1X4bUH
zNXra<u+#iVfV}a462szr`~zD7x}-hqn`+rzg)_>>5nkgRE1mawN+G6kITz#G4-a{d
z{+I##T|n6OI&bgvM17a`(N1e?i-u<C(IL1L`@)5xGy@J2a)~IgTTBSl;noY4Ab5?@
z^XOx8Hl<`jk%9Rzq9|n$Tx&WW%}zKT&i2gGZ*eh1f5z{Tor5?~+sDp?hqWsQ5aQIs
zU(~4eFM3Y_s=2vF_AHjM=%}%;Unq#KnyEE^VL0CYn@yTUz@_nygW}EYZOttOEHB4K
zXHF&Hzo}tkB>b374r0x_hZ5atZ>MEQaO;8E=<Tl2^x1heZ$r2qJX(dv9d2RtVY_hy
zSo3a82AtjmZG#=}RipHs$N$^0?{CfT?~ZvugA5)`#8nbc6XW%kp0@@C1?{!d!T@*|
zO~r(&sKx`P>*U^Q`{UM(z&sFllMhq|*mRJ$D~R?B&(D_7!k2hf7{+rE216nlnF10?
zit3_ymWOoE`#D)}#`m4|^U?80HV!b}IC=(byR$xD_-8&{wrg~Bt?I&2Pde2?%Q0R1
zaAwggFACL>P_JK{^QxjDCcAQ0F`lYuD*cUMQh1DU>LYUV-)=}_PMTNqJ(&_<bTn~o
zjr#t{(fT5lPWNk0dtY1K5StCHzRGt)DZroQK4-+m4!L+e?;GpIBg6$`Hks)`>UH}s
zP(AbCU4Mi=Yd$O@1>Zdn`_hKK{>v$x)4a%wUS8>@8%f2MT6lUHB07X~VmZ!c&f>15
z<1<HzCqH?J7oipS_&&lfinG@MQaZ+nJN-!&8?~aKLT?g#th~Opiz4VUtwWsPqfkr=
zodf)pYPD{2M~(&D?a7D^;uQn)-Wy}bNlItj?a(-VH=edg#QuQD(X%t1g15fM>D|AD
z+hB-$)L!LVfOJHSJ*maSF-~JNCTS*AQ|lsYe?mkj;>BIh%HWE!@G0PB;_I)YiDW_t
zlu+A#n-bG_J99%QGfB@-Ux$mb#9;y!Byfjko<Qk~Gs141@~@g6c(CVI-p4*L#8tS6
z(=&{CmpXnm2{T^)AI^(sx**U$v2l}eTVE>H_%-teMEG;ktm0e_KfjU^t_o=p>+Wjq
z_eQ5y%tto?UE&LDGl1C=P{Oj!Xs-vwZL>S0Rf)hMGcUcz-R?}#O%!jt92YAx3f|=_
z%#>SjUg6krrnD#hO?s-%T{B^j!oTGAda*z@oqEIsn+NC99UT)MB)F^n{u#QzaY$XH
z&B$~F2zVVHx0$Z1#wmQSzGvE1LS6rF<s94+Y-#trfRfS$&PC50GPKO!8R++rYUxeb
zqF9p4`)tbKSIWrLuY=1%nCSt8yS)R1vwq}+Tu`;)=@Z3>pKHg7mJ>Fw&JAGVsMsd;
z-~X!}1J8F94BAbJ+3{{o1n-CPM_I=(cOt0BZQb7kfZ%<bkxNqrpJjR-96hdNE3F28
zg;TMufVwb)A@*~18@wvf!9inRv2cIFeJKJA`5dmxA<P0%yCCOZRsZNpa$~(zQ5|(z
zm#rNJ!6KryCJb~IAka+uvR~lt=HWpXji!|-;cIpl-6BH$!(AtcdAxn7$;UI>hJd$i
zdruwB(zi=ycfbT=8zI<pMV<^lVb!@D7T__%ZrH1SVkGqL?I4%Wwx0Bq935A?oA!%%
znun%{V{7QaX5JB1JM{36)*`w%hDvZw#-$cJ@k^9HL{%%Dk5zSTe+Y@o@PC=TU94Ek
za9!ReY~X$h4{+xUl-7%em%jQh&^3wc9EVzEY;u`0Gv+Rp9b(v#tQ%*CGCALs0z^Z9
z46FNJHh(W|o3(LxTgj718#=+!tczTD*n)A2{g66%zCr2~L4{q?DoEj^aIjj<f3r!J
zT>5>VO$uwZgygImH$SOGnTExKk~z|M8V$@L|KTxAtQ)sQ;V#7qv+y@&e`xnksX3aG
zW#d3YB0yPWQdx99mk!P93dF`8=&4J!9S)8A&&0)uitX1QEYE~ow2VtGB9M{6qt$JO
zPqsOc$VgZ?6tdR|3NVK?be1$z+Tx%pPH{VF#*#im(!sAsd#XD-e2o?^VubDpq)UNj
z6OI%6SH=?W80ZLMEhFHt`xUOqq(`<_2V!NU$13P#8D4J}Ln$80zw5wX5x+^3b`|c9
zcRVIEVg38|XMQ*tYRpaZm>7#h)yygu`)aR>4ayPi4q6ZCtszfBg2V72v8g{Q*iNI_
z8|j(qiC!0vLzq$(<7dId9~BcY71Z)vX9`AK;be;ZkB8EuoBuZUcQ^L`${wvcCUyxt
z|C6YFFf+Rd8ua#=UFZ6>SGSHYu(gF^x23jlyAh2qJn*5RAtC#?^oew~TY#p9Fc>xh
z0QLV9Qn!+|g&Cw?#6I?9GkD={R&0i7fiE+hotux#g3yBt`J{BO(BNIr-}8j30amCl
z9|nL&8;loW95QBT2Qyf0(`TiMWr7cLI^DB8zW3J2{L1!CthMzvmej@k3Lo?2YigCv
zzR2&(LE$n)_sqfg0mn2dPHSzA;eQSrvYqYW|F)G6<J708Ho2OCXhQ6??C*M!`>cEU
z>c;!hcc>-(Vp{YW%`eW{l2V3F)!M)9o_N+{tqb{_qpMidnoNM3cO7%>Cemd0XjtKq
z9V93%LzodD<M8j@Fk`!j>4i^}>@w2z;uLD7!q2+FjWi&h914W*-s2GIg$`YW!350h
z7cS|#9qws^@Rx7d_N1b3cz1nKkMZ|>B|M&HOTY5C89}uaO|tX~BvD3PD+=Jc_oWzP
z1stZucPdmY79u-V%x&pGmC&CfS9Fn9{<uxIL0jVRCrm!#xtyeQZEG{*Z$w6&7j2~0
zo&LxT;92(BelJ5`5BOnZU)Kw@i2YJ?sjk%|xcy4}I&08AuRTwG@yfz9W~E-fT}Zp1
zBH~=Jxieoh@^oykFeEqKYtL_NjbqXpBW|hbC_^Wm(=q&>mYTs#DvJ_Yd_Q{ac^jhq
z;nZZnt&D*(v90_UdmRx_?yoU?wBASdJv}{q7p*?6sr@zj;(6+n!B7gulbEI!|Af-f
zQRKc>DQb~6+*h)>Ji(lQoRV&q#ij`C9_%r9q!*@IwEKE&Y`MA6b14>mdf)WoVlwo&
z+U39p<^TJ*D}!0s&gU>j$+V&xF9%RUxdZ;y^GEwtM}JrMm>N}GOr^Cz@Ptx<-TtHf
z_1P})#Ye?*qJQdp9j(r$aYB-$K#}6%&eY)=aKHpC?=pC(Dw1a~SR%{gSoD`MM{jzQ
z%uReqNNBK@s9E~iY&H>KFG0Qi|5Nw1F*P<fQ9bhgbg=W?mI#8&`NtvP2MosR^Y^Tt
z5*wRn*Lua9ERLRopaPZFjx*-OUSfZP$dj1n+A`wBTCdanFI#8)N-fx{>VG~V<VCBe
z*_Bq)u7e)N%s-zapRuV`y+Wd(q;Mw~A7xuVUM`Vdmd$n=5X(Sd@}0G>aT&fDa=qkV
zk$v~c#PKH=3?Sm+2o}B)K`6T$M?@M<)3FcSlJ&T1*}LJYtQY7Zs*autO4=G67Cy}G
zB>$X17Ze$14<3_mBPLaZ6b$3s^UMYz_#-;9_FjgGP*%8FSLC#eG4^*-LD7J2tRBO<
zN5*6_G6v9%RlN|@9vezT9G8mvnLKgyokvezoK4c_itD<<&icG_j%c8&jOV5)L{vaQ
z;8&9j`Z-(=1;S}PFAq;ImhdL8ReA3cW^F7B|M9DO5Q<K@i0T9E^b%4EBZ7hLvYlD)
z?c23`J(q2n(`Mg$YQVx6JN`@(9hKt80o@g20|NZTkvt3UD<@oY>#NH>@@Bp`jN|P5
zGW8_gcHev$;&s@vIqqC+NHDD&Klo1h#AkqWI1PDFJkqgR_r~1a5Y8fS?UmGo2joB?
z0rk|M0}=SS7T;jqA`ekawy3yv+h})Ln@v}@%GI4HHScf$VUwvW77q;8^z<sWO#|hs
z01r9R(D1igC?vx&A!myHS*#)CXNPGCecQVq{Pbv$ZCKtj{u~jt=hCEfuix*~5zLgZ
zTq@A#NDwdIbyN0LHiKS~*ELSoTY7M_gad9w4Q|}6Mu>3xOvGki?}SIeF=7jd%{Glr
z5VTB+Z3S&dA2@3#;NicLRP=rBjsoDIyANI&a9LF1ZPtHuDE2?la%u*yeY8(K+Hc)w
z<{cg;f?!})u_y-{p5+kfbLr*Eopb0*^x#7<?CLT8=Smh!Bn(3_yokhdGLXIcu&)x8
z6IW&FO^=GMD%@}bz0Oin36u}y=or;bm<u5$`O`i7nzXvgi4pBIJka7$HOeGa5|wU0
z(QIjI0)Ier9d}yLa2Ct&62)S<!VK0MQaKt$`+D{3iS8R<M{zBdSQ(mnflaDJsEu;s
zKM{SfuH`KFl0=PVsX+WGdSRNPYhpY>yg}r4K^F283OZ4hJ0@*7xVaD>A%B+IJ45zz
zZL#1|1Rn*GoVp$!VRRX+R5ya>L<U8twvJR=mpQDg|4d!JrBN9gk%@_e=-%)6_s0he
zMkABS2c$IokA%gdN4S<gdi7PN)NWb&UM`yxdzW+9&L7`&=RCHv0O;c0e5~nMS)fCK
z{ATFG{TxPD78rdU?D@Tgo2Q)@HGG^fZoF=8e{5Jqbm!V4DIF2Q#?}x4#0JkJxnqC8
zh|{l>fou9toSHM!Mk<-}TPMW%!5$e&YGpo5p;cr&tiFOS!nqJN^6cxlt^YPm0<v-4
z2<Zo@;^xC}%q9Ub;b&Pt8VV#6Q+(@r(NK)3p7L7(qyp<Tu81T`lk%~;5zIGXc!VxT
zoPL036q@+Kf&wM8%_Oj6<-9#FS(o37+BkGh6kFpBVtw^O$ue)N9XnnI(ecsIHA2g;
zhuG)kIffnVg5R-5YHon}0Wxja?Eh3P<OEvo`sa=u7XS{nn0CTNn&{rbmiPNq8^Nh~
zs7nDFGkhXB`7ONLrsJfze;hbX%Yj5g$KiNF<fqBu%i+)AoxWpYTc9N)$B&CL$My@^
z6B35s-vT?r|2Lk_k9Ku+rBkVN_e-fK#C~e;FRvXj@{V?Z;3EcNo`Rr>4?>i=E>B_{
z03MHltVc}0!>xmxsV}2c*kVz>9QG^n3dtilsID$Yt-BY|EK@i<X7Xz?hk&mnCxzxK
z;KGIALzl-`zmjG)IaGlz)jlQ&CLH<}>6;ct=v$D-VLM?OEtJtq8>;7>?vBT3u%E+V
z^aZ|;)YQ_#rWT>V?54|)b=#Y=^GLu)*NFw$s!HbyuIJwruQyb-iGC>jv>ciOb_-af
z;a;cDAJS-w@d$ZHWH+h{c9w5wwb!i-GGHQHTkHC^rZdzNbpx}~H7fJ)UX{OfQCquQ
zM+V266j!?ksVn+pLfLydNu}F^h+BkD8l@u)-t5kq60^K-Benu)SS{VQ1<=q^dWQeG
zfPY<#`a&>VnHI!|GF3WKgop0Yif$kAZAqV4x_Up56D|Gmxi}Hrzoyn*8LvY`(K5Q6
zjSzJs#uHdTJm3|HVr^3F9P&75%ELZ=RjvQdxvST7VaIU4qm){c-*}^;a!Qv5TeL&_
zpO}*0lk+f~#v(kic=*wu^zZ^UJ?LS2M*$o1xN_f(qnv4C`f35&b*NCf;#)QeAf%5q
zW^CCHU6>Jsc2h_yUg#9mS>4=YZp%)AVXfU#&=2)%yC%=*jqWZiY%k(m64-KG#^OPR
z!PIK8Ge|u<_K$)TSDuBbZ+98?sWB>!@}|tIh={|*<|q~-=9~><KCj~Eu!H)#3Hl#O
zC_k@Ge2e&-QF71`ajz&9j|J$%H6yE!Lk<pCN5Q=nJ-uow4|F0orP`iw79K1~6axWf
zkjn2)RCE%-<>-ydnAhUyhi~}{zkb@<*RR^HJbgH;I7wS&u`qjEZb(c8^^KuiC%8OP
zy{;ewXLSa$e-sOR7ZFKf<(+*$VTLyjZy7%I;3blYQ*FsLU1V<h0rm?!0U8c6i@BTu
zmI|7Aj`dzp>cCfzaO(iO1+asH(J9gZ2BckP+SEml9L|_&zi-_^JUL%uYh85QCpr8C
zrVt(j>%G?DsbIMrFIo9>(mGYo${&LH-txjq!N1fb3Xv^@jBG<h%KY2Q#@F<$FvO7)
z3;y}8nHVlZ@LvxIO}9mcW1R+yBvXU6nz`_kW*{WiOup&Q^JeM}!tXXmF!xjbptu#q
zcyN#R$OJO|xaI11FkSO@Xt~Lk5N#7O+DuQT7TUEuqjyE<=@*^3-f_1A2R_<8?~7<<
zJ<c|+j!GMFVk=7ZvRkwn(Qg3%b=oGgZP4V4UEPtDUVuFdOJw&RjX*AoJ74lAxMiB1
zFWkthcaA%-|G;^t_cHrKrykcZd~oI#syL5TKs=UH!ra){*+Hg<q;#ubE@~7X7NXYL
z<NyX}7AYyQaq62+)@zkZ`8`YELmX%Om1|=u;ji#fnuUO3|GmJ%M!$KA7W>R4j>0i*
zhYIMq383?*{-p6hN(&ewW2n%F#|#<eZsPnZkc|WVHPrdvt%qTT3T#NW-`Qt%^aprK
z5yu7bCt;iaUN&Yw?+$gNzIac*$r<x2evk{g{C+BRaD-@8wzYM}#MsK<3k|m1>$qm(
zovfRDJ+6CB1nJi7k`dF*5j@qTm`sF?EyEdWK(V^89zmGjPx_Z9tuSV2!EJ7=E-)<2
z-t_hlY1ZSUE3L4z;l`mK4D69EvCf9h!t$he)&D#0v<h|cWv1a!MDbtPbG&;r71JB3
z3mns<MfcKcd7oS?PC3u5rpAv(XA;V<DoaYb@%(rO@l9_bxH!c{WSp2Bp%?X~pS%Vn
zVjj6Z0cEI%2kU#Y+oMcZhnoV<1vEu)t~#1I02S8!o7ZpiApAJ&H8!%p6uUebx+Yv$
zS35FX^Tp5&3X8RpOY8uXz1aRZEMg!Nb-g_#?1Wt@pRozI?QONMtj@;1IZ^J)Uxe*E
zyGCc%H*4;VGM9rjaw`|<)t8BBUckg|7t!WkMa_8hLP&**pOhe&e^ON?>v0vt0{EYG
z)x>$=XlmbE%=CZ|!1~knlY=p%lYshLa}(|7#3t$9QrT;jy)pf&B4G4&mIRW_3W28Z
z+&ME`_X=h>yf^(c)7xOm`=Tma5M2`hVTMM(YDp4{!Ae}9E(IaS0y6<r5zO*=KUMT|
z^qR#vzt1DVtf7wnx_KAT$^K592LVkJ@q~@*1J8S!QjkS44ySJg<sYa_u63HN@$^)~
z2f2mK)ASQ80?2j5@XU&b;0Q93RqETF!(YP21NMS3C%@GZJ!v*)8W(CRH)p%HQC|?n
zDHcy5Sl5Zh<s_=y9Dq1KW!4UZUWRcviBFaeS0)M+2~K?$SYc>tir3nkrKfj24pQd3
zOW}JSeD3T}y|kx$c^?D0x;ekGN(cLQ3<Tf5?beG!s94-mQsN_8XP3rGK!kh<-o|S3
z?`@Sq#RvqkMZ$8@a(-hef-0bv{JxK!9o_eh@JH4HQr2-`;qW?eb7-5!BL02#gCd=A
z?iSQvmott{gm9qFpnD%=7#a?}z3crj_RxG%-c7mT?O9#7UvLRX?F5o|!26#-KsSmJ
zM23}T{rXkz_oO|OOlo%C<=g#!hsb-~+d=bJTX)>phvbI|7$<gX;|^mQ#0odDQvI$S
zZkdbe>jv$=)s=Qv3_PILRW_BI`@w0oFLh&uGmw@>i1`TBa+POk)-HeC<7Shh5$LwC
zEG^*BCn%^Z;wG^@siW9A@*u1g)c^ac<yKjHR{Y)f(*lW6*DsY5329m7&oZM0?e*=H
z-1XtQhWP1>iyae5>0O4bZ!$-`Uc1`e^H^h-HUa7^TD3Kso4`5D-yO3+(OCF5bI>jp
zb_g~PhxRq98)d3?n0{R~(9*0bc4L3Y!2GPkIiz`3*eQd<*|)j~^+I7xTbtAMfS#2r
zAkgipt`w<aBn;9NIE7gZ^gd_kiTaBQpsQRbjP5ROhLCa-!9-{h#j~Q8_oPX!{~H6q
z!_$sSIHOHzBkEF+l+O(u*`W`a#gy*{cR#KUGyoFBeD?YSa?+QXk1B3=-u*M9CHdQ|
z8g-d@8g90TyD~`R<8i8#9*NeD9Fitv5o&DgJn@EvC3X?62z`iCv)YgF6EWTI^!Vn2
z?-tXr?@`Fi>@q`IREEk0Rv<{l2QZ?)7cpb*1nh5bRqfGe<RECc$+4Ud*$RBS_{_2!
zsg`|s;TyURynCB8b_mtA<sK;1q~u$LhBe=->LGDKX(o`Ss`*;MGEcjNO}~3pA2*+V
zu&DRCr1@)^O;Du>#v+~(8^EgXZZTmqZn1*%B)2nWqj>>%(Bw%qn6(EW$rVwIDuD%3
zVVMafRNN20LZ!cT!{;LOa=~OTH;AmOT|LnT!Dh8xPRHZ%AwO66yEx*6Eh%_7zhCYa
zud981QI1*9`euNp<7g55c{ubetGsUjq^i6CY$PSS1nN`nrsyse&rReGHRIbk84(C^
z++=DD9BFBY_Ay!pq=L78Ex=s}CSI2wSX(FCmhT9K+SO+BsJ>T^*bX!GZ}qAos7@bc
zI3HqfNBb|W<~AKjVYG)E$Y%wH-3Ilojc-8o=cEoM9o1g5f1y00WU8|HD|y+Mm6>?c
z6Q8wv{SAHrQ`?A5v`aKiyy9<+0^RdLck5Tc#fsR;H+lpQkU)z@cTxn6X9BWNBN-n{
zQ*6Ig=a&<0WpG-u7UyB?pSx8gwSJe7+nwR;d{exNj&JYAMEDg5<FymkL`A`pa%Xr{
z@wlxCic$6;nd2hiJJU5ZB>3*ad~o9<JHA@jTa|Kh>F~H|SNi*RY=QuTha5o{uFNO(
zR$pPuwG7tmF9Exn=K~rkblY->PMkA*FnXI;`)FgmBu{H<YAxE?>6hp#hNt=TfR#b;
zy>_efa%$Zq``o;i(4(#S&G}>wj!*jcHq#7m=(72n-w?1*1J7=CWs%iK2M@Ltw?H8#
z5_ue~7e*m(c@ni5z~X4nPC#RB)dZ<_oKhq9$9SUdGK9U7&D1F}d^)siBKix(P-mi!
z@X*t%!u8U;U(1H{x7PLD56=&y60h+Z2SqPayMjPI@<O1F6DHtZ7j#nJTK(6z_aBu{
z&}dfSF};&^p2S#1cdyBK)cF;O10i~BaGdBT%R-$<LpPaTppKaN&(2-4!474oZOVi4
zxWX`sL_PQNDxvnw&W^S%A2Qy0tlcTW_Ope{jNzWT#i63^d|*)%;8Kn9(N35R`U7?f
zwG59(8mGdL_*mzx>t0`0q$Zx)el4%JHYzLH^!PPc7^7rxjO7<h?`S5bJT_qE8R3Pk
zLc=2*ikoj6uv~ez3HRsJ*GK<RDReiK8h%%H6(U7YmG`57CMFTgw3K{*xQvtXQs#?P
zS8`6}^)yXA9h>-2i5Uh`X4+Kq5}`xqK2HI|<cpo6zSIp}&W`Jh4jn1Ch2j;g(?R4i
zNB3ie;!+k;BSL@==FAZiG$t0Pu7HCmNUTV(p1B1q40QRt8z5~{CZRFkgJ<j11-9Qs
z#RTy4HYFP^*>LWwuWUr<)Oa6z@{MGc$p`j)w4a<NUXqSvRI(~~4jQ2*>X3+N7)xWj
zP1aIormbfWDTP=w65LZ>#)^qUMYde!BV~IiFm;y3;$>;~=^P@pzedn;eZugwG=4ow
zGN3ZNut>k&V`C6?5Wu0#9+mk9gOYcwueP?-g9M!~6vTg9m-}*nB_Nt6qVZ=Jc09&m
z;_O<#qxQ(l@AC-Vu}~e3|2|pGgft7PDm`!A|GOE|STnv-SGHmF<XPV>IaYD&bgAji
zr3ejqBc~qEt^yy9CH0=diCTF+74W(53w%`SSnFS;Hl@1Rl_6sZAWNTs7anYHpE*4A
z0b17*`iJ7<>vd20D7SKFsbLLH<&Vto%~qKuPA$+?7v<*kA#_qwX0}3L02*1)kZu8j
z6o(P73TgH>5nt!PON%=CoPw4Fgm^@`Oplbwh^xAo>}XHZ70(K5&~eW%-T;)94moIC
zn~oFBWKjU+8`2jWw{2yWBCc;P1Vkyv?0GlG153qg;U}f~V;GDeF|El~0C<qUFyfns
zuFiZk;a1L+{z>7UOg=%;A(GgSd#xl4>}7a(%_`2P^kTI*krp~2;p_IBa>>VX(^mgW
zTborfU&auzj-R7AXQl2j;L@imhs2lBsnk*5F`9_zB!n<i-#L+hgz9r43gP_GP7CD)
zQWTX{&|)xUtND7-mH7G@_%wY957Y~k9Q7bJy5IkneCZj9WR>bBh4t_#F0EwzlI{#W
z3=`luNV6ZRJMGwmcc^6kWym#?!Ro|@yePp86K)8YAm>Np28y`oez@80ZbaNHHY;$p
zL4k_9<WxIk?s;jL#9E8>YOKu3lxIykW<V`=7$sizTpL^6Y(mT@8E0L}FDKWP_aVdu
z#Gki?hc^7W@8rQF(iA_257rN3MiSJ^IeiyXp10Lxcaq8N|J$CXj{i2H{xCWwxd3yq
z!m$lV+5bpy{J@q2Jm&0aA{frSpNa=@^THUG7&e$iw6piv+xoW=oLS|mo-2!tg;b6!
z-D&+nB>9Vw{K1sS##eeeipVno*G_VwbQ2!tBVDf$I2vR;y-Z%`CXC`eM|eZnGTWnh
zdCBfh#qJc%tQ5;{WiIRrmO5=(KUbksEqONr+&BGH!w!H=ZPStmwbfJcE~eA$gHOq^
z<=YnO>^!!$DS^#0^y*saF!HG-o56UF1VgxPmtb=Cs0BHr<zD>`DFx?Q6jRfpC9c-c
zh-!#Xf1}Wz^cb#d1Z^84RhW1(O$B@{`L4O0I+Z3yiU(>weBx_X{<GStXu|hC8pQ>%
z68X9ny$*mPj|^}nvbsqc!g(Uk`Ok%mCzO_td-&%o^Sw<W#?5vJTkQtQ!m2S-kf&#1
zFIma%Wh5-lzJ59<2tBVKkc3+`k+I+6b1y14#DR!rb8~b3kbx+%H)2)We);JTU|8n$
zWx8-G-_<>K!h#b$-?~T6(|sJ3&TJCI2{Hk^=cV1IT58kvzo_UvT4gruT1T%z&DBsL
zUbS#EKl-_rjjkkQ+1$v_@_oHPRNYp9hW+ABK+9$3?)xd)+6~RPXxiL#1Rg%9SnS>%
z`U;&eJL3b7icozwQOuHzHa{+uO>TS>n_gRlyqhDy&(Bu`tC|Nh99TZWkSb?=2!Nib
z&aKyD;pGoKSFA-ed0hx7`Qc)n*f+A|fhR%9=9Hu-u5U+4jd#xUX8yPwsY9<T${BXP
z<dys2K56|&Q*5wlqrasw&i*#W6csHKyYD7uz!rp5{OCpdw~)FS#_~ctO=KtHB(xKr
z8E1ae5@&uxDb-Z@WHOXsNQu(ekOj(83d%)(3_U1iYxbDin)iuQrrBGWoREuxsc_x}
zQVTAM7eu&M^;3DG-iDhRMi9!Z%d9IdLZL1<F6hY8j`2Tt6}Zq1re$<=qot%Uk>L@D
zN3Io~71;&Q&9^Mh9Kv9g&RjRzN#|s;W%uuB!?COmaMT2?mmCp2BWc7IAg8Z24A8V6
zHQ89+1eyC0E9)oGnEqe#n;Krg`x7u0NMojZf17yA{ILwbgy@3}9{_m#7`_$cpGt2c
zo{|}*R}UAzN0mYg5RZ-48Q9%rW)}=^@0M)inJlF*!jq#HN4W338vRfiady%o798gi
zr-0D?!{^U<_3n4tC<Es498m+tX<oA_df$XuSjugGcvNV@O~<UCe3Ys*4WHaD6S*~z
zH6%qGas8QwCweLjGj7mNYVygkSOz8EbrO#L=L3wbMdv(wDDd58Gx&;}x3x6Oc5Zc;
zU-(p_!wE=a8gg}ehuXDw;LP#|_gd9F5BJD5Rf?vrZn8ros{_Bzwiz5V_J(>WLRim0
z6b%$sLfNIyxyUo?y^(Lx;pDftx0c(lGVH5vhJe%{?5`H@tRnZx&sX~$#aemWVbKv5
z;~tlYF<(i&={-I4^>v!&7Z|jSq?))y<mOY+iBsxe>(&=``KGVe*DeRDfn&O+L2;6l
z)WY=7wz_c^Ah_(&!f#I4ug4iVuV1eJrn}Xsp0=Pj{$x;>w)`%zU07>(Uz-#Fa|Qj+
zgzE<UsH=XbS4KX-lGKgKzVPTr(al4o%h0wica7g&DKFI<o=f=(TbLr`6bcaM0^X+7
zm=p!WRm=^APk>Dcx&p}_#dX;7D8poUgx&@JnDWvr;F7ryCSl0iP4}st)X%c*AO)pc
z))BCkg#(1L!Y!p!Zo&A_*XgM-29%tETl4HXOOU;sA8$&={5(x;nUgL)L#q~JJF3@z
z3LQVype`_ZR{StbJ-7o^Qrh`WzaXKvNs;e+i}J=qAXm(fNh4xtF0aoed#H|r_n;GN
z)9`<|GXVb-Ea|a-W<Io_3l7Pzm40e(%8=#IV~Oq#9h;>!d5_$kzKEgQoE0(F(6>Cz
z#K_}Y68Ct6)`G|}D0ana%rJd&Ww55{3pp5*BgejS`YXCC$HhtBz~owVVX}YUp1UzR
zcD(6PRSA8T395Vpa_bAievFS7#vDYwL=<CzA5Fx+1g<tVLwFf$2<(I&7otRQ&Y42R
zj?=_0nhhJq-b?fHa=T*$7HV;;_q;d}l{3|<zQ4~biQaow{_s+HuT^h+h6@Gl+A%;L
z$FlP-?WrrgvISYX;IJK=o|hK`pcHHKc)pt95$SyY`^c+Gy6j(YDCG;#b)s!y85~f<
z$T~T3h8GspKaIxOjS?$|_4hvfR}_A-fR=xG097c*0}i1x)RLfRGOAcyx_NPabLX#d
ziS#&IVMXzvbbk4P!yq>c>{HS#%^6&I%Pfu+rnw3gqZi}!C+bp(9ZC9E*n%IAZKpeU
z5z96cbRgH{ba6LWSDt$m$!?$|0KXfO{K*21&h^PO&KP0gX|<~E@Zas&F_DKkfrBd~
znMK#D;ZQVW$Tqw5NXjxVw`4{xL0!kMVY$ycLH|J^EE*o&hVIN^!d$vliv%sc9psOC
zY*OFZ3hweRI*l~3rYZ4BM#8&l%tWuhkSILXp8EN>#n(3zdL8*^1M;sV0gVFUpR<#w
zC@UxP4%3^cBCOr5WU}b#vwjKA%JCLKnkTin+|OCW-5RD~1cca^El%(c_6i-dC3+3o
ze3f7$WD2kpDH@Z>uN>-$X(n~nw*E?1$2Wc+8}pSB|1~7o{y$B(lB%ENk1W#I0t13X
zzN*Fb$4wp`C4wou-a)qkW$)y#lsk|Ha^|u;??QKZ3$Mg$-pD@@DBbJq5Pt=wVgD8h
z!S=T|xTX%IJkH5=!t^&kI8LjK>4e64*m@y@$=;}WxCO^25rcQNKeNT<Z%O<@i1w>y
zOE-{FBE{yS-JN<+O$&M(zDDJqge6aF`U*SLR8_earYNt>lr9IPbS3tje{Y5vKNCr4
zc(6Tk4w)r{R0}V|EosR8dN!(gh+qj-{F_8TM5#p81m8&#Iqou%laqmBi$XLB@1KSH
zT9r$4^)20h7d6LY|N3#JYOz3cfyZ3_@C0#nwAq{XMQqDeoJEQ=;%376t{h8KG)!98
zCC9>t7oVj|C?Eq4pluQ&4*5_2nw8Bq*xOPM+Gw}vquZm2dtPg_53w;3joRbmkGb@c
zqmE(P;#seWcXVS^N+VI}7GNe!8a^vsRmT6u&kEZqtR0OMD<B?!CATe8?m`$|?O+EK
zZ!onF&VsGWPyWM+#qx5wh>9>F|9ws-DPK?56CbGa0-l8K{XIOa^)$G}=(~S07KyR3
zQ+ltMn(|R-nE1+~BGir?;4jIx_1(cnS?si%=&Mtf6WilYD)w54cAh5Wr%&$d28<>=
zRD4f@5MXDG63iBtnTtJ({ps^y3)M$oesi7E$rGVU0t&a~>RI3X(Yv&BLpnGz=dr^#
z#WS@ntL%W&L`y<LDNK)EJ!+FJ4!cD;-|C}N*gW!PJn+nI)*vadcvJKlk!m?hg#~nv
zF4C{?6YVw=2#F;xwQS=&sWyw_!_W+>M(cj0s=hA2l7i4<v+Ii9+p8|#K_Ak6q3m2d
zaf*Hr$fPc%k>Z-=FN0b`eSyQ<7~7W+8uq5rKtXw;r;x^HNAtG2a&Ar70?hAnQ;KTe
zy*fpc<<k=ftk!UIk`~KiSG!j6F&!N?Y%qLtz+RZNJN;jT-dD$r@b5=|mUCDZ1m7K3
z;EqI@$6fm|blD)mxu!a??d3Y;iSm!&=HV3nT(sZm5?4GzEp+l|)&;clt>xd3MUu8S
zfs0X>k9(bjb^W*4-I2`wuD&ulVW7TFiT>d_pu(L@;B(>F@G&ieet}uWmO;Wxk>CFf
zMsK~Z=i#ZX`$=Y<v*?WZq!{@(UmBuCv0yXgN=6xcix_B<v^Cm)9T=>yVp3G$`aLbc
ztJgm0y`fGV;UYVuvbaG0viS|6{Xfow8Ne9@W;oe^2PIWO7JZ_UnH{6%9*2&6tk2$%
zy`tl-DR<-F!zqce>6;<ju(-=-7GZwU(r!`rmlhgzr)D;58NHTIjRT+9O#yoi%?`LT
zY?7Q>yB{K`)4Y*-56(&)|HbE(&cgj|_vdaicIqegXiE%iC!*7<szz(S$;eD6(t+N8
zxS972+#7t{Cw|{KU{#HUjwLl3<F(9sz@zP&i;1O$5Q!d2Ng!FhmCuh<d6S-!E7OyA
z_-}`~nC||q*?QaO>3O*ls|*a+`A&Y!S*C?bp>uvd4+oaGI^}S*wRMQ2Rn~gZ{(JZ4
zwAJ!nSSwAXWZADJWOwpQJ$Q)0rT72p4{q5hi%Cez3(3jc5HFPbg`qU9TAOT#R_u*<
zC@A@`T7E-cM^=AOEDS=%QJ5)O%{khkYN5fZ>HIO!ftW7d`FX3f^NStCiYzh$M!387
zch7p`v0CpURo+Y>AIIl^HW*d#S}0Q$#try#?gyyGTP_BZp}P)ykAXVpoSlGw1(Uz?
z-v>g1^5s}`JB25BeMH4!zaZ7*^3BR-cTWpDWy!b`ugb|>J>4y+$A+jc1BE|Fz(+(C
zUM+nmW;P)F#f@v4q6j0D2{j~lf$k6#w2E{a_mIFKgTq=TN4+YDUumM##t^@@x4qc>
zUm|fa5&ZWJ35e=m;8;<FA#H0Q@=JC@X+`6w@Xod+LO8UPd+Y^46_SD2aBK<7z@Ndl
zHiv{|nW=WbqsP?Z7TA;DUoRHC*p=cRTlU1#UsC7zX!&-%koX*a#;@Get;Ks>dSt~;
ziOmmp3gUE2Kp7`_A0j7o_KTZ`@f<~j6p@IbI^U1yL~!o|&!Qj<w=l?JqQfzX!Hmev
z6~i(S)(}fCuYX_3W!wNK-`?YBY^75%Aw0r`E3Vr~)rT{M72T=&{AK&vadJ!?p<CFk
zuk4tV;&R0f78#9)@p!%x5@qSuF?LJ%KQx_tIMe_C|Hs6LnL|a*VMM0F5at|nNC<N%
zlCyWn`IO_FV&v^m$}wja8Y+jFp(t`@p*J?<7&+uLXW!T7cU`~h+P}MYZF@amugBqj
zzul|t@lJnIbTP|?#S?mbuXwn=DVcETKUpU#a`Auq+s?ohXo4G@8h&W)FBc$hp;gKk
z&-xrfeWg90)psNlmE%iYTZ<^Evq;E5?0SZgJ;#k2Qc|KtMVhV@Wx|&#`oxwP%vm$O
z<?*YYvlI!eWVE9Jwmb6a;Vxk2qtR&NN7eN<!9(xOVPAav5D0eQ3o^U9I)B5!_|4|c
zWaDQLj*@w}dkE>Qt@S#^)nZ%QC#5BV->D_^m>E>ucF5s&0S@^4n94oL|HfV|UxvEm
zl)E7*aO~|QbxE`j{d0jL#*8wtE;R04Pi46E`sr2wRHq~uqrdTcLzhSM&3odL&24Sd
z>y2x*F<(U6!fG>xsK{XN`a2u8)*khZE27qO!gp}VV8=dt4qL1N0UHvjY&$?vErZqW
z+TQm24px^Cz=!8=_ijH}bEgqWq#~8sjo19=Aq2|*SnF<<Rm6?rc#Qj)byo)=G^!7z
zmT4X}y7SxZcEF;w&9ui+<hrN`?DB!r)f92)lX-bwh#tqxbT)Juw5a}6Y2=*+wb6AM
z$L!o`AI}7j$qrmwiO=d{U)-JZ$d}$e;ORp8iwNUn8bN6iacP7PeL#q5qbDg>KvJK@
z<rnPNLCcdc+wuOcb+AmOtDd}!|D{0Re`FnXlQK{`#V@$;gC;jTsP3K-#1G@GiYOo9
z4yQ(zAU@J;QTgSk(k4egbCxHLx~#`>8m-5vZQ)X#>Asx<C`RrsO_T|n>VkachjsTc
z-TRE6vJJ>q!UnwpTV}{z+UnSbV#O$loe;w>obV$|j#46}l1GGqq?9jnawavX@FhEP
z%bV~iD=f<Vk#JuJ#ThC&3J5`kv9}^rfgRUHPTl#<VIAkoTM9z*C0)#HNE4hR6yh@9
zBdfCmNB-S-+stoyg*h1s<M{NXvAn;p8N|h-)^ky0k?&U#oRa=*9qPrr+?`0ZJv2Sa
zMMIA-u!t)Nc?R>}Jmu5V;;A1qXJSsf*H2FE-~?r{`h@p<=1uPgWq?2sSYC5`L}uNs
z^BKThufA=1a?J2wBn$;B!Fgy(AP2;5Vx>^?-N4y;M^0r6y)>Ca-dL+?Ydz_bo^EZ4
zoboz)q{Gyy8&xrvhTz(IZ1P#S-a#`d`fid~0ykQ00J-3zP@3Y50N-oBBlC9UF`dKc
zrRe2I&ihwFHL?ghI{}9~zmZ=bi?2@j8-xGAvH9gL4~&~`z-8^9;xw1q^wP2!ezZ=H
zrB1F9zcc@dD;1F>i!y83L$R}FN@7<p4mZ;SiL1@y;!e|{Pj<TS^r!lGL*j$Lk5U$;
zBsK(0IR&Sg;4t-&;Rcgx|H1*_6c6ahPP=Vr?P;pr#{e(AJLxf##a$rX+dv;9)5TpO
zeEVJQxk9VMOK@Wx0iZ0IP44+xpb}9D46pqlh_uP_2_1KDqY2rYz?ld^%VCozo7X_C
zq+l%QoUnLI6kdXDZrBoaEp+)+m))we?$a5cov(Glo<=JZ>u&U!6lM@A6Zw@J>ra`~
zyHs((0wEUO)>^jZl2d^0(??>puI{UFJkQelzODMk0N3O-smvAvYMEZAhl;X^{m<Xi
zf_u38=qqo^M*&^S>FD(U`LH^Hv<xj_RcVUQz&~jdW1eJcP2=w5WY_A^h4$~}ed+BW
zDk_bW{@Wzut@edwangUbe`w}#sU^yiSm}#${l4~P#0yASJ36ju-etvOR2<|UHP51`
z@0_`cJ0d>|&Xp$5g%YP*sSQR`|7Edr@o53GUDK!CUt!wwC7!aEhf0PRg(4E><oQ-)
zW&G#bC4?8FAjWe?jQ)&02sYuYbB+ROe5$<NAWi+nauMx%t+-gp4#ehAT%q0;_xrds
z?QX<X@c`DxkTxqNCC2i&BlYCeZV=X4>E<dff`#YhMSoV58H7B1o0TIvYU&;7ibPg6
z-wVpD*6jnY5<aG0n67H*1~vZe;J?`fw;Th9UxBtB)czZBFcB5a@kK8d{9n}Ix%TSH
z)@&8gVmUCEUh<u(Q0n`j{ZA{+O>h~jK#+y*Ki@Q!%AV#J+<9M`<)m-u2>GUJM7KG?
z{an|;kx!3mYBbLFPJL_Y_tL+}!e)~3aeaKE0y~#?P*EJ6SYu#J-Ag%;&s?aR4~frc
zrH+pud1*4=e)CXhly*D&nd#5ZbEnA=)DU3*?WJJZi^ix^cui(zL1bA-_F(T;2Bk9(
z{`fH*1Oc(|2XE}2?Cq_sjnj`j#&Ik5Wcpfb^V>yLBdk7hp4%*_)U$>@8A<Y-v_5LL
zRaNZ;lN5CS!*s_%?6tS6Z0kv^F+8UBR~d!M3MD@wp4v-_o$ed}l4e5Qo%(uO^OzLz
zDOx4#sC8qbriO69)Q)`SlD$`XM6<pGXueJ!WjSiXo1~B|w`9l1RD?Tme0s-hg^e^l
zhY#b2r+=$OAQ}FuZ^T`bkb+Kgv?+$6;Wv23WO8KRd)>UnY1=h*w$Bb}etL=)kL!ta
zPm^jD`M>LhSfFw2RRufXzYFbLSKKh8H+$`hPaMk>!1iN1*`hj;nQjjTj+%DKW5h;D
z$)i@gn(?tipf^F39pert{+Zq#sq_f+m*^*(PHxZ<z8J<FO1aM^jk1FmON8F-?PjTM
z*;hae+hjuxanD#AbP>XT=X<{oB<<VU&<`RjZEt67ZfE<_$s7;Gt%o&*`oo_+=bhLC
zgXVATGlnI%oqgEP4UHZ)hkc6n-5xUM;Fm}fY=nS87XpPkzTli5`}o}EZO11>F+kna
z{2{mdLs!=omd}B$1kEt&q0pH^$C#1C&u>5$qY3lBxcZXyT`L*OE6ast-d==6c_p~X
zZMh6Ag;@I`f(3Uoxt_9|z5o0JU?nKMF<}4c&eGUywU*;_(6K}?>j|VH#WgF*vDsPN
zTv`7n@-(qv&f&JYia}9DHwE`|i%bQ6ec@w_s!B`_y2hwJmFJpp!{f0ur<zE*(2A#t
z;-z4X)2Aldnu#XEIinHSCs6S^PxaU9LrVzZr}j6{d5hsWBYAT#<}#e1f5b(sNl&P4
zE!4B^UVsy^w%fb9+b}9#8F{?dC+|q>`j4AYL}|8YHXG-!{_3~<V1Zq}dPHb`Ijh#6
zs79$=Q-S*te6~67MQ0FZd<9=hIE`kI@jylLd;6}zY^Gf#;!7UoLthC>zmp<A*jr&J
z=gTNuD{AW;a1P@<IZjUnT%~{q;jDR__+5%TIyOZ%qyToSHzrh4E@L2yB2Kj(=J3qZ
z#(B%(;1g-@|B`b5({!q+{B%ABrRBU-6sqQ|pcGY({z@^+&o>?%T;lsZw`b{bMewoz
zPY=0|5#IHI<p#V0g&8e3K8%hvXI1Q+YM(#ok~}oymdpA5)&Ro<*~Uv__<ZH(G%P5d
zlQ5ZhO(qn|><&6HC|$JT2lIciX=za~c)*Z44nB)Q3faACdKo)6EN}9D?6AQL964FU
z&n6336KF&*BIfh9x}RV+Y4oGRRxhFLWBE+&B^3oZCQNbK$o=c9L2Klo^gvs0`s7LE
zhVA<5NkH)uVY_QL(r$I`X#x&b7io9DWW+BR7HeZgyDUOz%3GAVhXq)aIW>*pq%9mr
zBYBsh!--ARxmCjbByBu8hBVRKShLaC%?B}?L3Sa9dBhuefBHW$a!*Gla3jxg2fpgc
zj$LN9CQ&Uxl^K;WT%h#pIXNk>=iZvlm_#xhc8o6+3#R>DZv2-~@|BzH$2Se0(=1BL
z@q9{9b%vy6P*WwNy|AkYll(0n0S+8LlrJms=fBQb=OUaPg?`CdIsE!E`*6fXGSkJ{
zZ<sdhS$NUHVxInmS5(s=!|^?`eymR=J%PK|{IyPRVBjyQq~Hx=3bS-2>GE4{F41SG
zhgW*{Yaevq_W^-Yyf%MscAbxZcXvTEIE8s=SXM>%Ba6_iJ~Ie@pHm~l83OUbmDETt
ziW8G^D>ZpcR3aJKtJDn9c}vIPImZ6^{#pKcj(EfR-HF4l3EJ3RTV&YTpq(A=bn7s3
z(qnuUE~+ao*fCiI7LJL2==5cFIa2*i{@~K~@0~3G7IVymsJ0y^k~jP|tP{w*@tRM<
ze=(&#)p?qw(FJqVgq9=e4QIH$$r?03W>z)&%E0KK9P+@p*KZ)il?^wTrUn0U?G_a%
z4R8@{ktC`{jy$>naUPN3mSapH*ONyl%C7jy1rcq8pxvM0TWuTqo1Z+{X#|0yX4jdd
zLPODq<E#;wE!b`Q9`qvb60+dlHihy#AO|F{@zzzpn-}CWsQ{q)Z9x@@%;M!rgQh>m
zfvitU(u(l4vqn3Ov62FU^O$#Iq>VR`KFs}?%%J`ymWuZ2aZ~E*Bi>cCxWI{!+IE8w
z(q|6euiIRDU-U@WzVL7TkY?X!T!c`9^Md`*gG&`*=c9Wwlse>J0fy$bX~7P%xb4G}
z;{(;w{}$hl&>pqnN<YoFJRBH|<AQnu#8H@kcUgv;qpRFx!x~_d9#7@8P%Dt4PZAsV
zx-?Eg9+Xc;Tq1Twp4w>uSguM|KJtT)^@+cwsD>N^+h_vi0JH3$h~(mFnT;^46mZkO
z!cbdvZg{N1L+DuumP}K`CK@L5C92;{<huJIkFiWb#{;*zgx0Wo?J8yY71O1NQl-es
zZ#VnikFJV}L1yig&+F=KbHRr)5~5IPNM$8dyiXjXx}lW$7FXR!sMTnZy;CvYmk#($
z$q0+yzY5I`9f|9%;HO+^S`f{QcD<JufmNr7ZP++13Jcw->&rymiw0-8vG|mJ{={;0
zyW&6G8EyeYuDQ2F2+47$N@q4D1}~Ade>P_YTKN3?8JsgGd|0ef!7sufz;rE8$RK)5
z#~j^>U<$av5-jU!TgEAYj;dO2t~N9jbrNteIn!ca7iI+e7rA4iwUbd5RV0R=wd?kv
zfAq#QbEabW0?zp}F^M#4i(nnYl@|)~jivAP;#>yqLhmM`WkGKbBRuS;H*jG9sdZs@
zX`1#(W8H#TIfJ=Ku1x;%W6_nQ_C$EJV<ECDv-f~PSvc^^&!p7lp%!gk)ikb6dXo3l
zH&d13A9kJ0;8No{wdKIkf5UUmww5%Wte))dP69G*`3oY+Xj2G9zQCBUy`U;hcvy}D
zJ19WEvPAO^jZRv6hmq)yx{B_!qn~B)5S~)Uj+%)_k4UiiD3oI}nd&Xn=s7ttOe*^O
zSZEgpHF-wr9l9*6TqK8(V^SZIj@?x$ty-d}8ihRQp)3RdkHkClHCkEK19ns=9<=Hy
z5b}#6|D2C|$;`1E6yCNWG;Z@!2={WTJ=HPEW>P=|#h{5ti268d{#<RPS!pJtq<Z1M
z<dfFz75+?!p5}oEu%>dQzKxR&^&FpNivQ(R7J<J=;nGz-{e_@ic_#!4(Btxys7OqF
z#>Adt->b^yWZ*l!*9_1{OiFJ<)(fjAPfqO4cF&k+2-|nRU#ytfw@a|+z}fSan@M6+
zWimjueZ3kd6gzeG>PetN5i*n+44JgClDXm}y`P{<q1#;%NRfZ*RU1@NpWW1lEEWN*
z7ncfHe;Dx<1zCVmd?DN4i=H#D6OTZ}DabS0VPoo11|>suAIaQ<X5T%|2r-_OFU#yz
z{X*y;Cqm1lVcbfeBm<cKY3#n5DLf3MpV5QbsG#?|<v{`8IenGVllIf)dgio$+d>4Z
zCwy1CoVb69v~wC<Mew|WS75A0X~x*ryq=8vleBjZ>>@YIm2Qe<PKm{x=tabwQ7?()
zWq#=M&E9r1UZt(uki5xbDKxC6;y*)}$Oj)TgP2xnx10Y}`;*q2s`hS=OL$zn8y79g
z7$9f0>Lt*++6zkmX1gM>n+@zW)66iXEM1Wj8wo+VMz$7I_WtxC2#{#V_*Bz5kXz^p
zWiHz9pBUbQ-|`UvMQ<lOG-ul_3Hyt;o9do9NmCCBv7MBzX{-@ijqq3;h;uL$WCbT=
zXd5TIv(nKO#Sh<QTWUJq32)BBExH-#*9CI@_h@|YeSKJ?cy&A!Ja>B1yDjQ8!_W=_
zVc6LTWf^dH>NIWId{-KG(7PK--BY7^l^H@|PB$od{(@hbK!O>jaIJ<&GCA^KZE1X5
zdSs7}k#QjdbGREMu<$(KKM(A1`wgSYx6w6?fdV_v5TYuP6?5ebASc{?+0L&e|9n+t
zXG)O?3K5`|@KW56K=E;VF}wUn*{%(_2FM)7TrGutnHBW!lrp2+mL|Xq?w~Jl872Cs
z*p~LD_LoKW@lTe4(b`c(Z|>bU@FK3))5bWs{KsG+l5#8!<dnmcAxzHO0H}QbxkTMf
znwn>gcjH)K9VCq}9S0NOZ2Ic^L}0>4wGta(RS%u0Y?#m%e4&m2%z?(k7PkF;q26)Y
zb9&+vt9wCVZwT8306Cz*4lkYpPw7lS%wA<fl*x1bmB`V8^tv7F{Kn#EfppxUHn6=h
zf?#j4=KJi*7E#YT@CIdLfwdBQD@wA`z5H&RR=$&xG*%dQ$<0s+cDZ;RkGytOA$B5^
z==r$s$Y)Gts&Q}aL_ITuF_U*O&+kW3hD72Z_VQAta<amIZ&y<9jj(s}-2kB+5)fvB
zUvhl&QW9blGP;eH?b}p9!U#FmqKd@s22VYB!pc)|A<Ifr*%YBm-%P0_8E$4>jH{sP
zhbT<cDV+<aARKBhfiy(RMc>gFbjcpc&_BBP-K_J4e@{hUqNp#vo3c?y5&hL0A|<wf
z5iO%In3Ed>yYDhtSuvkCcl5QFv);%(uW1jK9q)J;F9R^c_4~l*K$jAV6a#(O;tXdn
z^Bwc*LqvT{GCUj2?p(LyaPNv=EhLJZSk=tWCyC=>5LU@4wAOTHF!#~H$^Ye2WwJI1
zZ|W;37%V8r#4d_G%gf>;En6QQO+=EsfZ7@u_C-$7t9$ygC1lEAx)S$+K?vS{B+Q$m
zFAF7mLyVlaG<ffBKAXwg^&{H)$aG_Cq6L^=klQDL;jZ#j(Am>}h0c1I(0}E{Tpx~K
z!GZDy2=}B5LN{U;rHb*q*a763w#<!`C-I+e{WrPW>kU}Hw$ECoTeA&P7`PbySJ&p|
zI)+lTpUE&m5!IfnwmDxK){ZR0HeA;uLAu9Bd{9o8#k5#a5gzt8|IOXuJ7rEr-RKPn
zQC8C^VEHEpVm+zjLbWvgo6v&KlF$VZ<C4T+(>q?B@|&~BXCXbsZ1rbe9rrhfPX&`O
zV7+!T1J5v`ZL`X!Wt#QLF)P*(uOusya<{x9+t&~-AuYDFb$oTzvud}da&@M?L`9@y
z?%OS<ptIV^*zCVPC6LPdM8CFC`S({UyJ@*Qu&0lc+$0x5$zCnCQki|(?S>>e?&R=v
za{6?y2K&e$2W7~4GWK&d5nY;rXD$Q*=uFwA7aCvh;dgw=Ip}C3v7V<k)BpV4Ppy9A
zRR8bAO3_M6wN$0+n4GGMVRmnxzt-rtIMk0}cZN>cLF5JJ*VRDpxEj<}6Gp7}u)+f7
zc3cG62;C3Ny(H0?Z<^-Z6?!=PFYUifZ0zV078ZXt<T!6JIk7k@MoY~Le(q-f!O@+b
z&4=gbO!;EJ#D1pEm0lSbY_&n=ls*l<6;o-jDy7biZU<#^4!(8Y5D>aiYZ{E=5`}so
zgnz#G|CQ0Ppt``zdi=o<yQt56S+v7^i?YolTpdWN3HI+r9vdGTFzw@Z0t7h&UdDsa
zI-yBU3T&?SaRbYU&Nw~`-!|%JF~U`5M<<1Ai=vfT*giw#RunhGW&GQw`0mbV5q>4v
zt>`f|VIHVL<+ri2u`!S5D1UbOKUS4Sm-iq0!Gt-lJd#Ou|D4AD*+-x9Z8>)M?)<iy
zy4oAy1<YEJjv{=bo4AR!26|)t>G9-|9qs_olUnOQ+zjN|o`>WEU`)RpLIuppV>j~T
z?NPM|Q$C~G8v<}|>~--?s@jjpOkEjg&W%!<;0je0Y}OV(OFc?HqOZ5GEF!yjyA9Vp
z&H%sS&hbu$S!l$Oc+ik|mi7Qo!VYJDfv*$OaP!*uIK2&bvRiYu(RQ+W=Ksj-4&s|q
z<-fD;`2e6Az3|ZH5*O1C@)dWFk=!~d@_A51L108%y$T3kk*yi0dOvsWW>#)hRVPM_
z08D8P4vx11fLS(Bt4ypH!XN(;>Qs%4$tu)@m|EC)gjLx~N!o^Hq%gS{5Vphd@W}aH
z?a!ixlDIBGRb~nM$p&gQJ;EC)jd>zYr0$VX&q8uIy9?1gJlj6K{81IA!psI(K-3G5
zm^{SW2MD<qZD0pB><;%k_6r$cMr6eY<I0}MK~#o`$K;;pG(C1g;AQSD0G)85-QPg^
z-X*TU$!S1=6LzM}b`BwPsngVMXgFxD9S+9nK^fA`->rw?-y1F*9E41-wl;n*Quu9a
z2)#Jy<k9ZXyf^pxN~3<TESR^`h#{xxN>m<*#~SHS{muZ1g84&$DT}pVzLtI%@L@p5
zzZ)U^mp*2pv%41Bn=Ue!p>+qr^howw`*}xwF6}3dK()6@`-&l{Reu4G7($0{;v{NI
zK$qFGIigfz7=;r+h8E}I-*y-}2`fJw();cxRs3eN$VnUiJJhy|Rj)I>gcMAtsQ_+4
z?PU4OKCUJ8Kt6UHj2{eQ5*NO-Mtb=*FYAV^5xNiAkIf_ei5j0=i;TFA^}UuZgcUv>
zs%jKhFf`VM>(N+aqJxJnvr72m3X^&JD`LyK%ah~AMvcr#d`esio$zyg7Oboh?b6WQ
z9{ZEnFr_g8IeTv-i`yyJi{)3143m}S&CIeHS*)&%SD4AdI4c~4_3@_2+;bS2T0MT*
z<ch?n*<0Ig%CF>?WUl3laD5_e*wV%idnXUa?WRx1#~a5}ZFn3y)0#SQJ`WZNy$1`w
zw<to9)~~Y@)q(J|<@;KRI%On%$1J(7FB}2hLBze`O`4k(ch?;B_;Eyd!qD+0bZ-&3
z7KRj)6iW$h%J5S!D0kz2<)FaKR1DKHs9<M|1_N>mmVng=(sbm}g7?|s-@U3`jsX4R
z_w{388zXnWqp!JY`MktQkww$0vY!H8z;rKH%|3lmTPH|fIG(vnh0h1ZDCG3ta1~<b
z;YiE;wo4@N<Y)i?vH@Vz|7Ovxr_V{@&tg(MIMT|XNHZ$PZey-An@}ROz>NP9DvGhV
zj#jsQjS_DcYWjM5fAyPQS}sIg!nVF)Oc_*QJcxWfhNF*<kC768W-u0Us&o%?+iP8p
z0Wm=NP)Y>_m%w@x?gD=BXC+sHFrBkbJp9Jy{u)v)Gmh3a)AUE~l1ZqG!6+;!;r5J+
z*(lALO4rQrA3$dHt^xbkQ<okB-;!AtQk_X2uWtxfV!!g7l(-WyeQB{G0|+Fg2rEym
z=U&FW&Dmqy9I}p(HvDFNPD$dt&wYcw_f4cvvD|3pPNoc*O9~wEs%CZBAqMh;zXO`z
zPK4$xb)ui~;b9G9fN*E;e=`%+vJ5A_HZS2GLx!QK2Bht4zeih^Hfxel_Y#5l!j$#T
z2sA|4CyinHHhdAPD3;v$`Lji{flykxZzE1j3&NfvDSP#ja{S*F{SN0U`>4)jIKJN&
zB`hoaFP3wPWhE2hWFKH3$NLy<6vbHWb#YiE1S$#wWg_2{Kh2@YE=`mL=W|PAEk})T
zUH`U@T8DW^M&-$(P*Esr;nY9LZ!w0+yqM1zaH_d*pcP1GM;n?La6@Ku_L<z86OJg$
z_4oG7d^?H>8tbd;C28~*R2f@&*j;!J{S|`;f4Cshl@|rUd`_muJTUtQtO^>`emNHh
z%P!}Yz8k4*L_ps}8>ag^T2OPtpBzbh)6b5#PQB0m-?IH^42CLG%KS9H|NG$hfYN)f
zK%=b%cqE0@Od5gpQ?kHeOu!~;$DB{Z@<;ArSbc7s?R#8}7H4ArvaiNfO;OMR%pqCw
zJN8$|0D-~_&U~>LGSCFl@?INuDQz$LDBuQMsC-J@pUcKQkDi|GZ|$xQ5+7mPd9y*7
z!v;_7#DcpDl!$3i;h3vC#<bI>-jhhIOwv%k>tm9bB7S1pE3`I|fFlR(AhJ<}o!<v_
zd4vP=*V|@h!n*0z;(9}>DJ}efJfdy`kiA-8iSnN@>uKrp+=x<Q3zzdB3XudtN!v;s
zlCoKTq=8?*NZ*WXW7stBS-OYIFQ>HD95uu~54bRFW}wfS<<(eK!ObFq2~>R6-pPl*
zC;gE}csR&=($c{j2NGAFuwh8q=^Ng11qI2_p|K3C!s0eWm%Go~0YI0N)YM*m<Dx4@
z{?$m>rF83HkQMeS)5nN5+{3nkZI_$@lyj0O4h(uMfAQ(31}=-bu?Ejcd86KzyunK>
zQ+p?qlPV(5?^lt7W3@+BFCx{{QQ_URZ#F*KGGJ|=bCw*qlxSfL-lee8L|~Lkbkn|v
z7)CCtE@kQNyv}vg(~4@aAl2+z?en2PD3io+Ng$QL(KQ#i*g~y)8tTbCQXLDkT^)sj
z>^#LNU;HcIsZxBqMHms1<|u)!2%H`dtEyiUfVUh+^y4K^_x#-|M|4QC=fI=Y#b2O6
z=f~}cn~^c<#ebUe5DV7T?=wW}=xV;wZI{mc7<Q>b#50@n&@3rK)EJ$251W!ELg+^S
z&3Kf5?`386D)R~5o%$)vGq9>Y_m$h}X^Y(LJBImm+fYT3WMn{nGi&o9gkr8;%L`_>
zTr0}XoKI@{t~t>B9sgCZ0<j6n)e%+{6V~#=kvRfc;1jjJbL^Y=YK4#HY=jxmV@63~
zMiJ-uMOn??%QNvzaT%g&Hx~|6HQto>KGZcWV60*GAhv~*PIoiM0jle2+wqcgn3dKI
zBoK)EsOQBTArrXP^UwAvcOh3jr>i$S$)VGt%EG|rNKY^Ji`so1Qmoym;yHe$v-i0D
zV^aHeoBqqcn|-Nk(|G&*Ey}{qad<Yt%m67S=TBW-3-YLMexQ$x<I{#OsDe1WnL%0K
zA9)K2wUJv-*XB;nYqjMqZM(S<iH&QKZP{12ntty_POqwq_&Ro@ztX~|=Yyy=tLrJ+
zHy1Y}#DficxdaP{oCP;htn(!Z`?%wP!?x2djm#!{O_26Zb;PvI2DL8_cDpaae4pHy
zn!4Jm$fb?(PemHLguJo})?-5buCJfSFHj%c{?U9Kc|pz35WxeWE};0FB!bf5_V&^s
zuIXWZ8ND5712ZE?t*y8|+pULM1|CF7HPoP3?oO+xtJ$L4BiGAR>(UHd*eB+zas5=C
z(e{93&*=6y0wRt+@NN)>*XP1V%gJWXgNs_7A$OwF&>~=EiR&WeSy}v`l}F>&0y5$)
zKu@j_d3}MGf`Kz$Q2H>T#AJ*FG8K_WGBChpX-sg*{|8zWh@?2LkXJ+P`f~$<<up%8
z<6!7}IZ(JALu#`i2It@j_1IWkit05t=6N4VB=_KNnRrf~%mtz~zGg@-HPu&+P;Ysu
zpb{TnB75$7+fDmj_0wrQn@b%;euc-V)KyRBZ4eGx8^b(5>?vs7oB1Td(IArDyrJn3
zu;&D^SG9kOQ6l8|`+l^%4ywGY5Ft_UCc8@&Y5{|gkBXRY<3#-E4lyE*y1f1=2;Cs;
z%XN1qW?{o<E=IpQaBwuX5X2xW8e8~)t)etdgkjmT(Q*TeG|YmKo;v)hI36lKi@uP0
z-+)VQRjYHK?ksdIF5}Ct6M6c&Fv}wR7p0=k8uY~5T}Hxis#DKNnYp1|QVZm6Jdc1{
zfWN|(nAqoW>l5`JawtWxzlf`daCn~<Uc2^lFTaxJW_dA(eVk7sh7yy{3g+IInMOv*
z*|*;pd9$=rXL#@K1HbNcNX7MsjeDz;Yj&sG<G7=RHDGo-_6mk&_rJ5`68gUaVe2>m
zs7zjHZG=(1S9jk$EbqkVGob<DC^V@4w3Tk-fup~n3wB9u%<M<}3VQ32Vz7MV<kx;n
zGut=k7f|Y-?PO{If<oM<hXbe&do-J{OqwqCyib6qqXCLJ^hp=K94MDJRvYQ7-Wx)8
zHq%`>(||R3!dk2oCL$+?{VrD4iC|1)`(aBbdxsIp&pC!*B#wVI+EGor9VU#f>hZWK
zV@zK++>fodA}?P&*}fD4S;pYi2fTDaTnn0g(LD6A#t577c)l!CBieM0D|Ar)p^jsM
z0>FL4f=YFIHWZW31#))${WcX;QBSWfd#E09ND(^S1UwO6sduyS%b9k&b92|@nbXxa
zvkbUC*<ga1wJ$Jg49sOX4y@J-;;b*Jl-DZAD*HysL*HE!Vfa?E>p3$)dR#^J3Y3oK
zuPx<eL_l}O3bRF_1p}s`xsws(h1Etk<$^(Jw2^js(32h1Vjkgzzrn~A9?tKZ^6p8n
z#IT89s+klw<u~GTKU<zhZcK+~T16_BhqtK-Z5TpTUIm)A8s0^s3Ex}lZ~#<IR&chJ
zS~Gk9Ipsk0c~Y}4yK5`aVzha=E8^^+_sS=(eYfn_6^4ms^T>qlj0Q@84Sh9Y(qq~|
z=0|F;IO7Lx`3*@3ln>OMY6;HK_qNrE|B~dS+P=(kYoh-~JS3`Iw-W)c_PLC3?m!T!
zhRd6W$-miPT<l*|`98(q&Wp%|IUBNr?UG<wdBa?{WSN*nK>q2ijkaY=e4?l<@x@M!
zoPup|jjWO5;(ZqPOEY+BFQdmYA${L?gtAn(KKZ}xpsAplucB_>3YGZk+r2T>W$lk~
zOo~RQi8e$ptU9iUa?fXNr}{Z8-%EcY@7wnc?#&caqLMAv$ez!ND;7nkFsaGu1h8fB
zEx`xwHs$q%4EQIaYSB|3M_WNht2F@F(|bBHgd8z-U4(g&drtOMZ%8B{T6%rpXYH4?
z{=u8R8$>+H5h>$yVng}qCpvk#e^_hme^gT+6u=S)+C)tVz1{56)iBrW;0=@4fh9w*
z&|bID$ma0F<1O3)K<jvv52l3LjUNGKW=>Wno-!qT4>GP>J}H#RXS88_wQIMLo+>>)
z>9Ik|{RACJ<}yq2t$Uo9h!;S_m9FIOS+6#Xl7mpYqH53Jd{JEAZ5b{k#O*4Tu@5K8
zV2n|v@N(oE>FwWci<tIUwgnl-=rqCXcA1ePs?iD!m&^UltyS316JR{96YJGmBf0#s
zGEP?ohOG~}@I(#y<Xz&}j36k(bFv|jx;LQ_5qUT#bOs#B!zH&?*x&7Kg19Gfa&p%;
zowt2T@7Kf(R#bNJ350=x`kE4=PVsw94-ZVBv21)de^R<5NTJ`3ys{j#5vKO3s<H{6
zorSC7@Fxl}IbN=PXmusyrdCdmycXJhvZgUL<j``J_=?9gxhUDeDcKEBq2+&hJemM=
zGC(Hqfo_+sO$N0^=+_{JwkaV`Hdn`XNkSWK@0`Mw$^~?{b^H4+)G{Uwxqaupc%z9)
z>A~<ApyQgJY-j`zbg3Ev>w?8<FU*gehd7F4v1r$6@1#X(>~G+qEky{COiVQRRTkH4
zR(&_5?OUpf@N(_GFiNhgU3KN059)#s&tq_r@neuwgw00h=P4!3=qc<@T2oUD&(4I+
zs*6;>d_!8gBx6)C&r7d~sgR*JRLs{ziPCP(KS_m^|H8AQzcbnxBiOSD=RYYw7RFtZ
z_u(h5jD-<zCCFrQr#nLa79O2Kr8^&%BmTgAwV0iUft~J2^EbY9rB|#VG)9y6)-C_2
zVSl0VkWp9n#d#^HemC+TO{jtL)sLXQDEkTGnw|h_6fYx;gXvp|lO}#1DjBnmh%<SC
zsG|)he-_n3WWm`O!3Tv|q?A9uar*LN>i2&CHXq<zWRO!F2HF|i*`)W`wnpU1L1SYo
zF*ufYu6_Q^;%`6e`~qX(uF~@i(y)7a3XtvWj6x;0<;y~(C=Zk8Y%8p)w3|Ox1UZ(^
z@+n^y7Ud66{h_CEYRZac7u7Rx0JbwOO?gW$Qe~wtu<Q=181z4IcAj!v9u|qtgk#}R
zO2Gj>_12TX?0CG9KE1xGmpSG^T)(%3O$LzK8N;G~XJGj3#`2@Fv7^@N`i8}s?qtm<
z1{nks3uYOl?%2ryfAZ-Oib+b$NGuab<4{)XCv5Zj^D*37Hg3nOzH>!%R`6a}BLYKF
z5~7px8`#ydI5typG%qVnp9>uDuKxXQH%*&yl?428TW;^^nVrVF5-X-_?5L;q$w)_4
z+ei72)C&d%QASaGj{AVVy&(Rw-zO@K1e}@-vm63kTPxr<JFS#j1B{jEz@{x^HtXwG
zeywu1?|_5rN`rOSr9c^`;==BrsT`1Ce(vMNz|6*R4;+!cCL3|Ma}~Xi<&@Onf{}sP
zE{mdlBvF~=%<VR0a_bp0Y&UL63?$nQ`FNcof3x?UWL$=sA0QK6jqoJXHt60pH5dRH
zn=R_kbiBMZyVdTYCJ)uaMdCvRrl|2$7w0j1%V()qxC1Uhq$z|DDXief{xRh&s<)L&
zCm*h2exI^*zPRQ{J#z%eGw;)JCZ3#mY}n?`e@c-zYI}1R^Qe*y#A?Pn>Pgx@!}a|k
z@K3H;R_PdY5=|_!`F*9<)rE|&f(+C?#=bB5uln88V_iOzSN#0^v%V=utAj_qCv&)i
zaSfjPY#MNagouBV$H%$66w-0PBdNQ#L!&qj_VAxJawD(Kd4f(1E^8It;QtXu<Yp<j
z_h;q10#ljxKWOeml{jIMKFmY~=_OB$Cf}2rAlJ*E9GU3ianW&6F=>#iMd+?M^^Ffv
zHa5M#1AzVgQf)3Ui2TM`18(}GO<{xJo%Z*xe|5FvDnt5aU2A<e8XZcn-|;GKCtt>;
z`Cu&M(p)NlIk)QrgSGM~VzvsmT)RT73^#w_b4VJPpVdAX%bA>85w=mw<*6W19HGz_
z;K*1!MhX--#l^=iO?-U_5|3g<A-w7yV->~ZFD8E|+b4B|f9&l-8X@Ch9^;&;y+XLq
zqpkwl>G8UQTYP6C60#(Wg>K*2#@)ca&g&%CWcs1i-V_MdH}Z<br<Lc?RSbTQ%yf*L
z9@a)G)o(v*mt!|p<BJ9(-nU!cIpekBiUq@D%(&5l1kz+j@1-u`hp&4{ifx4e-^42g
zkBH;NvPWkDKB$w(9M?tttDwi_L}?H7%~c>K!n?qjv+}YNTjyhD=COZq>ix}6I^>pD
zDs?6ocWe@3sL5h5md=?05EInirpT2&IqEtOo4<iWTW)Yy2#w%&iQeUw^OE#NNT}E4
zJEGDl;WXfZ{#A^Br=|OYu@4tZpDGEid)ZCXTFtRc?^eIoePUqI)e^CMvK|j&Tdan(
zxfG`&lA|x(@-NIH<o@2oZRL(DHcC6@<3&W2x7VoGg*lNeVMmz+!jYMDoFL;8+$Dpr
z?q!hGTD>2nJK5F3hVJ^-4H{<*fjBeUcKLmS^WJV`N&=}jynj+n<bSH&p4VCDQ3A!!
zxV`~p^TV>Tb<$jVu(XZA%(8|@tB0GL`4aC-^Ma#T8Pb&E<@<4@5Zb8q=i3--&(y>F
zV`aORJ<SgXro*>G+f*2P9L`?vOE3AGE;pEd_K~RFloZv*Y^KkWJsPVq!Z%#QTcZG{
zJ8~)SnG=9*U3~AJ+!6BO;lPq=bg9xpn%unyi)e+lb(2GV`k{~H+1~afvS-bqYnrxZ
zrjJ^GvAz`WF&CE~)ayd@aC5C*UtWpVH;ns!TR@-)#49H-w{|1NwQP^prK!LU9k0Kt
zgHcz#(C1z;^NkNL!p%t0G<Z@gTFM}K+3Rk{Mh?VfuB;G}$}jH_5bHkX{yWAs2Jd?M
zUD5LjuYj&N>P4zik2f>P*UsRQIE#7g#G&;er72^P+JvirZ>PiZFxU3=>3&^hj<SU2
zYKMmZPb@4QRK$_i&kUQGy215^HDc~`)r3jd%LoeQnI*E`c50fi<w)xT{g~qT(R~wi
zH4ZPO{j#OVDtupA?DGk4iof#uh_8V^6b0ed$d}e2A8giWv>p6jJL#f~L?<dimLy1d
zP*Pp46K;Sz4p!J`**oEm(7NN6<Co`Z{PJ1LW!(J<>D1l+u2Ac`pa7naUSk5uC}EHY
zIHH*4h_Jv30x9!|&2~t&o$f6FrT_YRN}Sik#)dS$HkV#vzRZj)%lL@@6SOTqh!S)B
z`#yF!v{f}vT1c>fCcVl!yc7aWw)T)-SmBt#*_g5qb2RUbS-b7-1fbQVR%PLc&(Z1U
z(4V$7Cq=VjV^W(LKzy^hwb@H<>V^rq)YG7RI`SS0QXeSh<Hv(x^j;(fi(GAVZ5V5Q
zZJpa1{&ywU8@Io{tFcD!vOcf$uJ<7P(8Pn&F<Kk{3*x1#^z(=Fejs~XjVV=_2Is`-
zzJ=BYYUa=rgd*vKHEn0q*~OsM97e`3@@o%#@(FXZK-NVrrw{<ZYEk^iV0+uiUaOe*
zg+8A0BYh0mj(ONrRv)RdvbYMr=udk5b37~Nmb2zOLnTQ2>el*Vp};f8j2k!N>?fWF
zQU>}GWfAZ7dmacV;qmNSx480+g|u;5<-cn;TzK)HO|<~vO3NhCWwcu4!;QPhg>dD+
zd_avwBA?7AXuwl~6%{oqDAL~Q=C*yKk9jlIIZQltEQ15&Zh4d4yyKFTxSsV^deo2I
zToOm|$HT4-<SQ$sWfU_9+H&#3MBdjfC6jF7u`CVd7<Z0LOWKr1Kbj^$$N3xo{favG
z-E<EP4?b|!upRr5G1gxRl}%_ya4~nf<&Gq*mov*;RAuV^5emL~dE!7i3kxpn2K)S1
z-^BG(W+L6eQcZI3#>`vWUTfPxP`P=c&yy7vT9DjJzirb_cHdl%>iQ~hq9ISR*Lk@v
z_)-Qzwcc#fdTj5?2QR6++5lgt%62d_wE4$&nfxt_Cv0Wra-4{us@@THK70bfYxU9S
z(ET^*S!hO)JqGVJXgMw2#B;6%)V^%Y;3bn*<m|`vrSnFVjB08>d!Im2u)PV$Ne*@X
zak?;j`G2DqLDXk5GME_jvlCuA_UA&Hw*4=!)AhNzw&Q~u;D$g7#-;X77&+RPOSJSJ
zJk3h!YMJih;{%1RxjYE-_>#rR!ixCThaE{X7SwojC@nNOxY2upbrawUd3cF<>1u$u
zA6$G^b6y>@9BT(;XfPS35q7q|t_ejRpKJiv;G!#i2e59Rora$|lM4t~11#iyqo8A6
zIUGA7l!`R!<BG!dJv6eh%QvP2kzIm<w4C}7#eS@79FWUX#@RqwF5}@AEv~$o+517N
z^6+m*S64#4<w`$`BkTd_cO|oc7F#{S)6ZEK3fImmZ<!nfG80lLr(QUvzBn&p^PUoH
z3B~xK;{B70%_&0+&9Nm@<m2@&p)(47`ph3+@meRY0B1~J4Sn=l@3khP6E?(~7$>1?
zN*}o#-mkE?%6;<uQg9i)UV(`wwEn2&=?QgJ$lKelT%KVlFk1F)RY-mm#0kK&e5hU`
zq^6SBg7`nb8p20G`lOehKL<T;SQ)z`Y!tI7qVEVzNhVNqJU)6YwzqOz6IMzCgOy_)
zy~&DCa9_vo-c3*Ztf>uiyliq`Z`rHY`tsA@V0uWtSzrC*2lwvX0}`9;o@K?wTtU2#
zQ|dp8@E}`=H0MbZ2!sW{?8T6_X>v<gNSJ@)v`|XdT(9@a^G2aX>bhCo1A%Ay(}H1>
z-8UGXdOOJxD~+x{wXpKC;1D=AtKDfeS81D?cE%;;i*2W1AxqZdW9~IEcnIqoKLbTY
z<@|S*4GID+1C@J=r@z}y{}N8}oYK5EpkMN0|GfC)pH<@g`PpZV!&em?@tDq;&o`qI
zql?4Fy=Iszheo3fznP#Pij}(d2dI9}3enHB<>-sAy4+sdChp3pO;|N&2R-p3$4QjP
z7|?L<s)FRYIvk`?gDfyDJq>BK%LY5;QeO}z_W9*kG1TLcP}9|@w&5gNKfu(E<b4;_
zok#i@^9}%Wc-9B5LEaXc^!gz)>k~07E3Wk&BR^J>xbgPvOfmd1Bt;6X?#_(Er||&)
z&DwjOcj&{+inYV(yH%>sswK{07=8Ty`G{kE!kF42>z;g_#EEV1%wh^oOg##pD(lFD
z`VXM_N9+Jxex}CRmbXUR(Nn2@&;tlcc0mF-PwIlHz<~j?5X&Rn0vSB|^0#le1r7XD
zvY>To)NWncd%b{$Q_Y+R-ws-F?SIE}?w3zlqa4i6MJe|D<C!StJ|Oa;%DeRhcajRE
zGIVVa;ITg5yT>~}Y;Y2(5Fn5q=BSTs-e9p*?PTw+TpY?7E6bB!om<%Qd0?z9UQ&r7
zz_bmys9{yHEt7|ii`*igP%p9{Bv_I!0vH>uU)O)Ry(?;5DbI}ti<a@x6Wk`Axf%Pu
zSSfp8M%iAd%fGjjkb3d~zI1?-*<e7<XI$;h$hCGm`TZAjwNbhi4PQUq96bBIOV?OW
z4UWqj;VnKsAZWZeF5{QbWMNq)xG;?JG9gqxodjh$g3^S`IMt%2sIAARBq1+vJEM>;
z_>>|d0SwrZHTRI5+HT?d`$Gn51sd_Nw3fYod%q~Ia@DWY@CO%}IB{AC<^b&g;a|0!
z=!#&y>lO*=Gct^rJ(MXPv(D60{#M{!$;ka01umNSqwE~_CIFfS7nLJm(D%?r%N`pG
zI{`SfMrYbp6nFA}6L|yxx@~I#$nrnDHY2!YQg(iQRXjQ|O8!s4qMt!K`No%bwGQlh
zUO<}ka(KrBsDBI%_rz!7ee+8+gP*D1gXV<Hqq<8|zCTgEA0T5#WsR~o4wF?4l>S_k
z`=Q?b!@@%4V|2mYv)-cwjgtkcSC4;+e+KgF2`?vh;+~tv*9qlfi(S`}e+o-HZ!QnZ
zD8I#81?tJQNCH1ctN9lWBaY8+KeMw?{<22ow!u5u>(-TdBXKh9ix39hp?=c@0Ys><
zgjbY6Q=I&@d9Sw$sy{)>?<@v0lQhx3^X>2-e_CFr-dn0>>&QJ)gan`M`_TBb4dEZ0
zBm}A)u@6a&<*wrWpDSKf>sHTzjHr1(3UgV#<wnR?7$K$`W7gi}Y%oiPzKve$v#kF1
zsh^37iQgY~@!__zKG5Xxo2(L?^8;QCYQi#4{BFrz`{`Oz9_9aH4(>(YUUO(w={rI{
z?*`g|RUzDArbb)Hw$G!{==*r^EN_RyK$kXcgWzUh_|78Cn?|y^5X~q0%v?=X%EwYE
zFwFh5H^J}he*Qp}o1Tma`0-2<)X=`z%WB)I(-W(H5x3icW7Ujv!iV%wRPSwjDH0rN
zBO;a6R8?=Rmem{g$qk_5fq6+w!Qz{^Iv3}<vGLxmr)Z732H0)m@3vk2lXX`Zd7@v$
zR(VHSDws#Ekh23n43H1HxAX(l7UoXZs{3uZ(sklOXbT<v8P}j_lR}HTqa&fW0*Qgi
z<YF)i2R6f}JDQ8wvWfl)@>LY__OfdYJvdn%|6j#{G6pS6-Prs!3e3MY3~b)K%`JK`
z4y}xaQLU?%wpm-tpKg!#{{$8#H@gE=r-1h<FQ{Q#s)xS`q%y}1M15IC{@K?DQsj@U
zdpbJ0#(rpKeh`VOe6aCFbz?D?L<$yKe};o(JMk*s@`!#M{Btd}_Vs!0A+vFd{(h2(
zot}$>2_sAKi$2JWtjbBvNxR-=EsqMqwvLvP>>$auN-dsOt8Shwl?gy%H>TNx-$*z>
znEi945U&G?#8*V^In0ag^`|YlOKLA3zTWLP=>qaSWc0-@DuP4=6ktAYHupS>TySKY
zjdF5W@Dpgi?bKJ4LK;2QmR$V|n&>6c(p|UIjxf02Ywba-ZIHhPeId8I4)!fuA=_B3
z;@wYt-EfDiN7M|?qk}H*4YQP^*D0ickAs<kD=VE5Zb_cP8ys8-!oo4v5<b6`+2r`9
zD9KgcZkJ(69p#1n><T(pBmZ&O0LOGIICzEDl6v!Jln+;Wh@zEMpoJe;e6T}__4SB!
zS#ZL_hcC=G3?6?n-=?fU*SY*lUF#dV2)n1VrRxZnv*x6QvnNXJA0A)X#zDC`&<f`Q
zO<=pNKcba9*%_JjM@Yd}Ca#D$3k%hjnC@RTvskOIYIN04VPBt*d;tT2cySf~V-P&n
z)bB><4_T73d6b{SN|e8}2c*6!ism2sY47<(>Kf=aAH8!79xQYr3)v&}*CC}bAv`SW
z5q$Ad0ySf9VwT!PIoq{6-QVz5&(Gy&{m4|_{i$H9-y!(OktVPx{BXRwENp}6&3{?g
z{GfA`69WykAEW!lPu(MND>#DAxU0QD5m!)@?hJdfac;~&UjlA*%<pG&PVOw^yln8t
ziO`L2o!jz4P{qtSKeV*#l9^4#txg~S1-0PT6T*E_L`&DdDiEy(D7p0-)fbYT>tENu
zyGI~As>)?fpY2)KR&X0K@a-Bayixh4{i}I&qZBLq)j^n!N8=bQ5BJOrf_z3rt$Dw|
z%xD@HX40&Q^-OW_FQ<z^vobW(7!!NvpsER@tF4hI9sT{ilO!u?yYNg3LZ6wJ56!zj
zxzTDjQL*FlAphH?o;b)=Omb;`p5S`hO3qm`#+#X}!Z=N9QT^kM&4V$0t8b}rsG+mW
z$QW1!u5GR<wp}m7wM1-}(?0+h+1~J<$o;v65swP{(S9Id#Q<fDRaaA!>V2jiU=9uU
zS6KqAeTTCrP7lac(Ic<^l^JQ`!eVV-hbrGZaxy@RB_hjUU={~1Qx|bX6+0>K=~keY
z&ka`WvSIY21l~LYQ~|%<FkWkF(!q!)B9~M%&Bg-PY;i2k54HGkLcgveqOB_;loCi_
zE)`*6bm452DJS8_n9UmTttG|P-6>u5EtA58KU3AOCIs=50vRuT&aOXz{mw{fD8LoG
zx%<8lLJ+<kbmRu(NI+=~e^wq-6uNh?)QEi6k+BX9(YROEaJS;Fn|q46uP10_i0rL(
z#hLfZGHz(ft;+IPGT!SaH8Coi!30lru-Rl3E<`B500*2~TXCPbrO$hZ7|bF7SC?y~
zksCWZ(O_oqczOLz$~vgqf>`w-rwF5`t;>jXw|wm)rg>33aeKYIZtPFv2W}?K>>%r^
z%n6S$`rc{4qukJ(#kAsV+p$OPV9ndtG&igWxTkW(?-c{Vb+hN7lh)nQAinhe!Yb`m
zw)5vji%Mre6TGnakeXQ&sxV{G?4E2Lml$XY##`GI9JP7SI}wjzHGSM?5oZhaPr5z-
zVef27TIh894A7HrNLMj_fcV1li{XB%f404Z2)0S6ossF21zYZ(bnS8&n85mFy<Ze}
z;`f2`ynMT3-CL`dUO!13JAXER21I#AcMuyZi3m1EhRrdWP#0@_GG@ve=p;IR2dum6
zBg#X1TypmMhS@{MX@_qOp24s=j(sB0iLjO%1VR@*RcLxFl$8N%K(+4a{t|auTixq&
z2eoPcUMV>QPh-E`Qi9BWp)y>aHH}-o?zQezKE*qnjS~BBqrAg+LTIby66-vKSOw%A
zd+?T4*VldPYcIy@*Ns(s;FNFb&n_;01vYHBKS7FUAwQa&*w5G1GbvCXC0Zo?Fo8DS
z=zZ3F;~wZ_aZy~DnJK8iKQb_kZe#RKS+jz8%Vl6Xg5Ke!-fY{LkzDZc824$w{^gtG
zheA%x=+=EIT_ZC5>7;x8psw>z>Ok)oP9Yml1VoOjI5*IO8*Jq`As|~!w{+^f{Q)Ic
zM}(Xv<`u2iubLeS1_2O+8{CPEB2Ey72K7SH5rsne@NOa_ouHa3HE^(V8Zoi4sLhtH
zgMXmKj*xxcR8^q@+s-P-CHY=V?Bn7ucN+*196_;wuQC;{q2}!sRNsf*B)+DtcK1MM
z^w*!XFS2(>{puEYdQ1LVAQmL-m*&(B%lJ{-Xz6YTv&#0~L+WUJs_@Rsug`$8)yhX~
z|IGuU-3}LwGA=ceJ{62O7?6<p;|-SA`{Ujm@)-zCGz{9+bu_%j-PRKF(qU$22d#G$
z^JpNo<LB7VLm6HnVwxp)ye8^1jj?dr$Bs)69pyJ{siV(K1V|}&WHmdt&2k?QzfR%5
z{>KZ4D1oX^{Hn=P(a*B)Fv0F*p-)$S9Zg@>iEs;*@_ew`LD_oR{G+<*K~Hw}cqLJV
zo1ZTy)d3#7p>1>zcLQQ%6jo#OOcbY(rv;J?HphN3<+*r~H3lRw9N00$ak6?%Ur<z;
z3-LaU$xKde86t4!@5|+YBSFuPBJ#nSpmI+*vF`Nm@8iXexw4=@KJyWvmHxf%it{Js
za-!o*#v4@MjBld_#aW}B;?md_>*E@$A|!lJ>zaV<PlW@g^KaG_4scSs@aNA7P1^q#
zSq`T03ANzxiw8p9e2F()4B$Zh5lc)trAZ{YK}=K5lZ1%KeZV!$kT}MaVwMnz83|4M
zo4@|_%P)U=e0&oGLWt&exm?t`yOvT~Yh8PA5g|qf5ad2Iud2|Rl~sEm1F#6RLGp*L
z%+<k-5kiodXuqG1H-G!%%^VXEn?xR`J;o4)19Oy6Td!Tia6(qIuBbzCNXNt7x39i`
zaraG(JCV$k5U<PS>nuTv`R3-X^&i0P{d+$>-KQx|F$BVLIt$Z&o_{k0g8RA_?LJS_
zci;Vx^IWTnD3M=<g#M`!e8rKCiUri#Q`-I0KmDKb6SZ2_^?!`!Of*e-<cZViWa<DA
zq^opy+mL2pL@>>%tQUZqGom9S)B*#W@hMLYsfe1_ZjKz|wA)Pv-ARd2VhAzDm}>1y
zDa%q?tt-$;!sx>|SnKP<<NLR7_tU&P9B-IeB(MY`amOL>lMqtQDNP|};z%5r=;rpt
z&Fx*vlRqt^ZGSh4)@Z{W(X990YOAHzwfCylM-$Y60Rngc6vW69Ndhn{awh;V>xLDa
zm^ddkjvma{wbrGUlS_)ze%>8sLLy`^7~o_#H-yvDet!4x=QsB+j?;@nPC-(TAS`2B
z`f1>D!}I_Q1uK9$=r$PzBmiw#8`GBc{hFq(J6Zs6aDU>Eat5L?d!h!apxss7orr?~
zt5<U)Ay9Mx8m1YJr5U&mw&m%v!s@DpSYjgTpl*Olq!q0!r_0mF{WQCKt1Zpb&F$^Y
z?fey&@z;cdKU~)H<#IY-mbEQayJI(4*4orV46Su{5GK2dBZXniJOcpGHr#4vwHAoP
zEZcbWcz*-imW<?EUs9NdWoj#RJo%Wp_u(Gp+xNyyL=sp6dov|eL_;)k!|^i&=xW~7
zN^5~(<hik}7^{p=d@k6CEfAQd6mO3E+q+|&W2^1-@Ob|C0Hz^u%!$*~IiD?mJi+^d
zXTwgYzyrF^EW%MF^@`x#jll=gR{^c}MpSBBkbfj>HUx+Wn?Ma55Y##l9dB-LZ}0Z|
zqeyh~5OPdYjG36b_Aw+#%pv3$vq&UjH(!^_zyIl9-~93?v*qQ>8#gX%fB)fqZxs>D
zd+Ut|Ly!=7EQ+FnE2HT;cvfp2)9u&1<!EE(7(fltLcn>Jc^8kj``f$YypQIs)e6YM
zB7Y+07GsP-BuFjP-Bpzk$*~y-M&>w8hi||6{=3(|pYnkyA_f=}&@fodHh6AGlJoB6
zi*K~mT9(I$4*(z{De!uF0)!C5f9IP0kK=1r^bVKPvYT)3ZeN*VZ-%Id#^O8*<yZ9F
zC%!QFKuEHgF~#HIuDn`JJD`>NRJsxpvw!riy=stv4!yNPh(I}KVHRc~Vxj$RX26?+
zUe5je)Xq<R2o?G$2|J8ZLpSJMPp5?$YFlIGG=&@!kpP0JUM`nf8#?Ug9U}Ky&reSP
zsOrMl+Vb%D?!)~}OzE&cvIOQpBqNhDa|j{leD$&scTxpFCW$D)0Ci)eZfZ&&(|@6@
zYAda;z1P}%AIv&rM*#>15it-)kt7m`1&EALfdK5J4%LC^g-p9hQFv#dT~GI?^W%v`
zjxTNq$hxoRDzt35V62bKg0PG!dG~JKx|P~W?X~v0xsh(A<AIOsfDN&_M_~~Gb87_@
znF9_8tB>M6F`;$^z!-x%#VJY-Wq&Q!)w}ChNg5CbAhyrUCok&?!(A4U5dg6DJ|xKo
z;0E2blx3bLl$24<OL=@cliPkyDR9V#kSBB?Zs4$-tJ~9WRik%!_i!q;Ue<QGtO(38
z39`8lLzmVXA%+l%324AlwfA0Y8$unVF$_6o!ogg-LU#plVghtx5+(&#Re$VAgyy;t
zSi#LViP$)vF}?uI+cmb8ahO8%G1A~VM)L{5%xmvpiq<0gK8G<#q2{|Oe{*+yeRuft
zFYk0|Zf3|sd_Q<dSBI14rZI3#kyD0{7tJ4AcvEOG=rkC7Lw{k!%o13#w&A7?2r_j3
zh~}uMRa#{(gIN9v<~d{wK7U}4EYofm(!>%S29PG^Jm+Z&AzV#aflx$(gcu|+^Xc^P
z^UweO^z`1Wgh0&Hdt1x8oX@RSW}4=F*zbiX&k5YLt2!R`({abW`kP<wAD>!j+HG_B
zasxyJB2WM`bYr$4lrrx2ao)v{NUfD}S(ft+Q;bnWT5sUCESJZ}hkwiYTw6OFZ`cg2
z_o69EPSdxqe)r9*-yQaM+5iMtt{}Fn6np#6ktyW3+s*r&cfzr(jle8NWOBq-%U4~=
zUrDc|aC7tG)yr?00&J^k2QvUeWEd=qE71kPhnk#_92}4kWIrGKi`Rf~|L}{S@0)fO
zVG?GUa>gO&=mvm6qJIbn8SOh4mK38>2SWl9mSxeks5gT_fZlrJ(AC#c<H8~&ff1Nc
z+|f*X*VY>%PAL;f@77y8*EULUs(`f{(DCWTix)2e4<aHVN{B2`B!r+b#*|aeDW()8
zA_V{tNyNcNPr_h~M~}*U*QUK_TU%ROE49_yx|%yOf&d7h0Dm%)umlN#MUco5Z7YdG
zb-$={^Bv1f#-t$zX0qPx>TafHz3UZ?IBtJqf+_MhcQ?n~?CxvrYi(<3Pv_;dl%@7k
zwc9r1I+Qc&q>O`B0EomSjFnKFQjFbA{i^YE2S9flOUo#Mqrf($IoxK(P2+eiRycr~
z^<DuuiU^Bb+kaeFXFNxDa8$RZSejF5hL`1XX{EUKa@g&5dCnnb3d{tgU0ZiwzRoE5
ze?Hv(X(_F1X|{F~LImtKh)G~wd+z|qrHBM31jKIEw2v7sa%Q9uM2E#Ffuna*b3;a9
zaaTkzH#IdNANP)$n-SACTjjn%&S6_N3DM994ph*nd4Dyt>K<Y2yv9_7sd;sihP7Mm
zR$5<cudT;~hdIBxJKXMbX|{AjqNX|o3o#3jGm{823wgh8{iqvO!&R{Z7?6Kn(*Y6D
z5WqIQ8rTY$nH$=Ge>CC3EJ*A~JQR#Xqm*S1%q$W^oJ10$APSK)hm>;8Q%Z>kgp9YF
zMI^9@yMH}Cef;_7Kd$8jpzcu0#a3Hu0r$W{jCENd#<-jEl(nl{QLxs`ySL`0^iq54
z+7<1}!v_FzG&cf}NYhU8L@CoW$2^4)*{yAr9e{DGo0^$c_fksNPC_ia;r}S^=HrXI
zZ(hE9J>@-us5*_}6%l@tY+T!$BYh%rPP^TFn182zsi)eMxo!OGue_4K${ZYXo^OtK
zcegKv1HhOq8lz@5^857wHtV`=KnjtFOjXT7$cMunF{$^`>t$V-Rgz3;+D-E&gsKH#
z_!$LeAtdX=I}O~G8FJ<zfvCEB>we8k8+bt7d)JXo37JU<+094A*xZxE-88F$^`c#C
zEq|>K&;lYxbG)pl)>Z&#ksu+2!BYup8sew`#qkVB3WO|@n1VZ+feivUz%bV8T2+ho
z(t2&J)!v%wRtDjK7yuczITB$BNIYoAj%Z_H4%Xd@Vt3kbWVBhD5i=30YFW#&tV^k*
zQDyF6DuE9<eE;h1=CJd&thFwsEu}u4E`R4sTe_EKHb^=|fM|e<oBaPecIU&N8jw;-
zR#tVp)@%^QmyW<fY05y@HWa}X-3IPHtfO$qpSt$iDq;i~tU&-%1GkaPhG!FqBO-@x
za9QfIwsl>4Ev`z02Ci;tnvn%b3}Mvkeyhn?fYPk1HFb3yKBuj9X2uPkWp2Lsntzgu
z2qJXTtLg>-NeDv-UCrFahGu+DVG2a5U|`y{yOKM&H#2lVL<tOo!KLcz+v_2)FaeIn
zrfN5)!GWl1;@Z#y!KcdH)*%8i2Q_cqhTW*Srx<PzyPMs7e=3(+)x32bmq$!2kq$Te
zkYWqsgL<k8W_D%CUzIZB5x(U{B7Y(#6hb3cau7f@W7zP7AiYU%%oUkI5s8eMd@w;b
z3lWxxB-cc5!ZL*OF{K#e*yKwHEW@9L)%xl5@b>**-O3=mpD(AOofMW31kvBWf7i9A
z97M(nwYlNtye#L|8=THdslBVKfnCjl(1C#5l?ig<n`2HBB5BSNqlmEf&VPj4N_6Y%
z4r^KKsuCon7z6j_?wCTpy?OEK)$8N&1yN{Skx=})n%iP5+kESixtn>2X`XiTwC{a&
zYfXLAK76I57{6-I+gce-xG#ENpN{nJI|J8;lz;y7_7`8oCf3<B!JmHCujGJFuK;5$
z>7QQ(0RH2{->Mhew9=Oc_J4Vj{)=D!)r(yFjD$q6y~PI9yuLDCL581SbacJ8+woWD
zVB=Q8e|)s~#hJrTjg7zkm+jKKKKDN*(ed%d)7IfSXX9V|3X1R`dyJCo@PT#z`xWr^
zR{rUPe=>x)e(+zN{a>B!Prv>*U-h*F{G}S}*Wc_HKk`3}@w#{|OMm@c8!G4N|LrCI
z7vp@n{OXFlUX6eAi9R3EXXY+k&&_qn;D7Wv|Mp)$^-sR|-xsI)%S#EjANb<7{MGj~
zj)BG^VVnGiu|V6*K0m+d_4Fd()&z#JlNc9-Ut3~-`!=8M^)GJKzdw>sH~QyF#jj7v
z0EH8ei}v$tZY}>I4}be2#;VGFv)TIFxBKF3;q~VIlwq!S|EK%uZ-4dIC-8d3;&pyu
z%gSHgsh>anr`zZAN837MpYz7`qWKT6iq9w1Kk4#*{ZAp}|Lzh0{YUtXikr{z9MAC_
z&+-3}_&NmIb3Dg$JjZkV|1O>r{c}9Wb3DgCcswWi=Xj3icxR4(@OV!2&+#13@f`o)
z@to+N<2jz=IsU=pInh7Ib3Dg${Da4HqJNI(c#h}z2ao4O{~XWp9MAC&9?yyXIiBM=
op5q@po)i6ZJjZi9$3J-dH*{<dJX?lfn*aa+07*qoM6N<$g2G35i2wiq

delta 320177
zcmW(+c|26#|Gt(OQD*QFvW~G8vK7h}BRgfwz8fKX)}cEz7?Ca6my)GuRQ7G`Wg8?J
zS!QfmN`~wae!jo+=e@7jJ?Fg7^M1~IdF}+BBkLtcswRkj=dNe1_;04d3tzXdT#k+?
z^T?h*XQ+mR!xG^1Nm*Qm4s5u?L+)a>Q&=?8^+#DT$Z`*N9>=~YlFM~hR@_VB0mJb>
ztqgGl&nJ`3BRca)Zg*dDlr(z7!eV6U$hCm#`JqW+vGshR<Ja=g4Ks_%3evX9u3e=p
ze2&FK3GC(`-i&2F94YMJ9`=uu&!7MJDd(m7*;D_xYySC6hi6Y%JPv3XXQTGXr;|Yk
z?-S9~rj^)hk@K_vhGu61dihKu7uZLN)K7nVs2|Di1@5n?pZ*m%`rg}iSlNCUcj4sl
z{pn#<3zet$kJ)vlZf>>NKio}o5s{lu1X_>xz-ic@wKnS7$+6$a!P2PKNsZX1BadI?
z;^qClJ%{68_h%#4@S$gudk8Pl2*N4H)X<_^p;WW$joZVwHrkG;@)fGxmt@q0FGcPt
zEoSMY$WIKF_m~wOemkPJYzUnG5pdkjo%lD{{%?>RI(yu<PuNi12xo<#9Hg8cC`NXJ
zivPB*oyq|-XWI)U2Tw#0PIBM>&22xPZa;@9S6%l6Hj_hJ7niPqe=q2qOn}D|EoZY5
zHlj8JPUZ#R3yBUp#kZ)9i^;dB2a`p|llKX~EG|jg9V}csS!h2<YzHv7upw`U?Z#eJ
z-`ukIhZXPtXO&8Ktol)nI{KFxSagULQ<<-OA0-8EJ6`u=#5A<+;!VN_%=XUqf6#Tw
zp)K<7DN(v$zMaEYSEl{+s2zN!OF2E+7hcGl{fBCV{TFp7275VevWr7~lTLl}orkKo
zpWlXU+mfvpzNqJ%B4T*<m^>S`^}OicL9hDBeQ@`_`cU!k!IC-<5N85LpV`R|1W%hr
zqNr!D|LmWc2-uWY-<Agt%fa&i8jW^5S#o^5LD+9QB5yEYHV%?E4w{ONnx_5Qj}6*>
z7{C`^3ml6Bt#jwDg&oYt`zdW{g6e<R)&DSpn~Z>eFJI`tP1e)R^@iDtfa9P49RCd5
zZ({@6PS)Dh;1uxha42m@19;Rh8#JQ2wtINC4qv3**`^mhC7MKheg61&@8jQHvCP2c
zTKmZw^DA&$5!_bYpQ&kYKQ)@&F*+67|HX2u?XMN{moR%oh}DHJL~Wo2fYzTze$7Jv
zGft~ej+9}5zyIdhyn!DhW)DdX4kPCjfA7w-f>t21ZO5Q}XZJmL{3-Dg&D<Gi+UEaf
zozX>N@U}$2N<k>ws8Vvn?Bk<&^?&ng;O@UOeDiIDF#`XFH~w7&X28?qOV=DCCt4`u
zHguR4YBMunSvfA%FtatOvU+rOVBl)}<JHa0%@y!u1)RV1cs^$&a=C@FTXnW+5TwI+
zs4Q}CJl=0<UrlUV)e<;_!p@Ky1^Ay?Ty3+1uKj=OOV4V><~SZrP0)4}vmf`R{OjWj
zo%!>>Fa8gyQK!fI*NFacr)rHtoik9ylHzkkae{ujD1-M|Z9Apvf1j)W%>w^sjf4*e
zscogDM6QHB7PDK5xx5zv9!<|yRrrMu+7|A&u3YH;(6=_i;PG!udCu?)C>=H&R9F)k
zcCuY`x^SuepkO0%rRZOOQAx)CJX3wYX@oF2%D>;(OQSK&4NzORhnH$9Xf-633ePxV
zpMu_KUXw=JO`VRZAF~50$3IONF&=8$IfGe=w7V}H|Gt=rI|lo7)V9tUHhMG)oxxkN
z=fc|uo#jG(m(wS==9x^wXaUIqqevI-<qPQJZS_-m;LKJd2m9;?J7>;5LgS%!IC75I
z8u>rmoq-M9jArCNz|S7y%T<;n{KC2~SN**EN8n!IzpL$ASO2#lG-bOmCF-~9#DYy?
z!O<BiK~`Whnl;kt?^gY^@72;bUda6X{K$Wc0wAfTaCcSAKW^mOOsW%STIkZie?Zt~
zCu7z%KV9u2n4$rS`O?gXi3M$I{^Nb{biZgUP@VhCa&P@bR|Wo~oS{r@-P7b~RN#1&
z?3)$DALk5$MYuB0!!Pk_JDseLH}55%{z(BrUVuv2r~L1n1ddmA&>O&}`svE4WNGXC
z1;2y0MTc({2|tYt2K|mne)zh_8~cwpZc+XuUpuP4hTf|wTBoLL{q#6!?Y_XYoL~|O
z2WHOP=slXe=J4m~eMzbRUuuzRj{W}w{@LuoGhp-S{nmwNXAHaMczUP<wuhWq(&U-^
z?bA2^bC9i;_M?{n_c3(a3VJ#Gaqlh4$A*S@-1~^}O0~U@DTH+&KXk@o`@(<k!w$!a
zP7=>puHvxH@u=>MKxh9)HD@ep-~Q)v`zRV5w54nnUm`s4{2Ozrbx!IGW-YVF)U4JY
zA|{Wwo}Zp^FsM|z^JXG6Zbbgf%7rI8<?X*L-Z!6~9D(Z-kJmq4d;I6?R-E|`_n*n<
zle>2Q-3}vXN9eHMJYnAyxKw^dQSIx~ZR;OxPq^EGe0Opr{{o%BLC2uOUUkFSO0%vx
zp0sTosR3cLN0j{{)wlAIh4byK{~hmJ|19scmf4d9dyZyLPqGNV-HUdX&KTfyHRWV=
zl;!AKxtayi;bbcdoCuAYID0YsZ&?v^Zn-qFX*Jg__+P=FL08748t<p-ejg_vo;f_Y
zCpo<N_t~kM`Od6I=_$WQ%*omH+Rfe85x`!Rhf`8(O~T=Hkysq^evkzWDo6`t&eG8W
zphK1`*P!1V`^XDGqXV;MQ*bd36DAeo??YSO^l?zJQH7p{cE0&N?9U_YQl5@zG!6#a
z?4!#HFkac~Q!%+EJ(MY1RW$zf*HmM9RF(sb=A+?-ste)k+rlHa3O4ILPoQ>n%OM!|
ze%t|*Su$W3vfup44lS<s^>`}oOI^Sscz6#0x>TX>Z?rUz`yUHV)UO;iY;HhdVZofx
z@>@-<u8sA}-ir(XJ^UOd2dxoir{37o($u@#1Pu!=JYc#Ce4m{?+VNa0x(s&-3S|3e
zxDGQ#AQ5(NK%x5B<*F(wtm~Q5@$VFW#>ZX!H1Mu26XxEYpUJ9Q@la18)mE@DnSVYu
z>_@D9URT}8{qoQ$ruWC#u?hd17XX~M@F!tm`>1k*5pT10s+3xnR@i_`QQy*p6nB+y
zWo7MdU(b8BMRxj5Vd6o*9rJ`u=k8$p$>_xO_Qz899mjrVr933o`6S?C*WOR#*YYB}
zYTmagwC=U8iHv-es;b@=OUMOHzW=Qp>@Zts_APo38@6~l;Qp^<GLmNodj|}RJlJ&%
zJUTh~MY6r<+p_y_aieu{;S=+iLP@$36Le^sMWd2H2wL&u?qy0?nQG3}`8I_p9rtqg
znokCL+)(yXS@*`5oBmp$T;|6kRPmyr=+(~ni>$1@o-@8T@FZtms32%awlP+?Sj}@z
zlS!3{T7Q{-CU38N+>0wNW6Zaq5?@DInNlis=4xmOzrRA_VUwyp2f1H4u3!W}`H@N&
zT{j4%X7lFUN#6$cWoNE3PcH(RUatzDckBF$^ISgdD^hp^d3c(9H_$jin@H?Zp*2zF
zHXD%af@?J9ga%cPfeKC5?+N2~FbN2n26EMUz0C??x6$Um{$dQn9B!OQ2lXrr^(nOV
zAzT%@Ai}Q6DWOM<@Uy7X(?4Qd*<&3R_Fy&-+SiDeS=(lTP5tU-RTJ-8J0?u~wi{X5
zjCH?Wbf=(3hm`scoz6X2wns~q%CG|^8AL*ZF%CsSUrPsLvuS^!@=T?>=e&A|dIK^*
z4dVkv@1VTUZI$=KWLflI@U;E(xI;_qzO<i7@(ryk3Fk0s|5DDl^0aMZmI`*z&_+WH
zj2C6zmH(OP%+0azQOx6p_aB`soK8;Lf|(irCS6-5Bacpx+mn-%uQ{qk9(<keN4>6)
z`18qj#1U-u_n&Seq`3~XDqdLIy_uDSB-IUXht#M}k)B(ppzUr_)FQ(oe-5O*#mCM6
z>+JNpPNR80ct3xeLZK|4_S&P5(1^&T%@0kN@A<=*)5BE1_^fPw3i?_9B*Os5^)d_5
z#R3ris_d_aR2j9zG0<b0-g+A6W1sga7@Kj&%YlL##^K^xS0t9`f!<!7x*>jPWk%%2
z`oRx3Tkrel05hGQ41<1=Abq4(CXvV@CKlfkZicR`;K!wZ9;bNufj0P*<Yezk>Tu6|
z^8C@5mu@;l7EUzusfx0#%L^J7XQStdGfYRkC_zC}wb__y;%qDH8tTopZ}$ZLfTcsg
ztPE)Xx8?8~p_*5kWK#3?!sF*Pdyw<VxJ8+#x<D-{VuQ*6V6%~*ojF|KnjPyQ=)l0h
z$jBs!a=*ee!)&DfuyQ%@pdva|UoTuqdD3Uavc(|x+sj=b!+w3*9v>vUkYT_bx45`C
z^@|cQGh3wpin5aKekJ_+MZ<5)f49KWK|-Csa2NBX;gm5PRIFF5C7e3t&hyo4?&CKb
zMKQ6w=g<8jMlYN4s~hOIF+D3xrAB;R+-N+m?<GH8>x@tVeNUah>s9*v`d6%xUOrTz
z^z!izB7V&-6feORYOxIn+87&zIzDn+sN%UTaVf3DK6E?sQSIaqp4X3vCxJYDb#Hm-
zs~O?+f;9R-_H8p0!~7DC+lbpX&%BWe7MKV4SX*NkFREo|%gqV@5ecb{ThKUX4rdmb
zAPgXQKUm(ZAXJ_f*;p^yjiD*K(!>99d2sYHf7vqhA`B%`+S<$Y;ls1s@GyWL)56$6
zX}cS?aRCBPf=2iOm?Y0Wa23iQ<j470d1C_~Pn3Yk%e|!GS*Y^)rs^)lnSuS4vt-_j
z)gB8?(nY4rvB|PA<l3xHeX}xU6G~r2V0&Z?wIr72oKENFW<nH4e1fVcs}Y>e3xc9T
zfI`sqb^A2k7^bxBH1Vt?-IAxc%JG}THQ(f(r!-St`9c`~LJv1E0CRK1IxOEZ|FeA^
zBkOx!e!tM5eT*}&r8`SvclY9Yh^@DMZGQgOTAtpIf=GPJkX}hX6Fcn%E$+`hWxp)b
z002!aot|pu{h#wx4SMDNyQVNn?E$410laaV$U4jGbZ`XnjvNl67%H!5YsjfdPY*{F
z=a>HIF-$D$g_nT85*e1Am+{@cR54qw^TyDH<nQAdz)W%Qd;G>SWn4-Sqa>1^@dSG>
zTN{Z~l9hFK0v?&+>nm$=_=|Xld~WP(Z$q$xEM@*~9V!Tf!?)YlgXwxSH0KeeZjYR*
zZN^<b{)ufWQF$gTR8Pu*h(iQLU?@!lL{M|t@T-$A9=xh#mYVt!fI}P6SjQU8DsFCM
z4>9q-Bp?wPhnczw7fQ>3f8-*s!T|B~91Wacm!ZAStqo48@7+08rZx|ui1R?anLH0v
zq|})QhJjqF?47aQ;7`WUK(O=kC|n0pPll-+;ciu1w@fp29Z=z=#eL_i;w`A9^vafE
zmBIf?eqa)~aW>wB>kyCSrW-CUsOyRTE^hSR_`P2MSE|S?htO=&OOJ@Jr@JLB)Yiq}
zTfTV=usGANC!NhKB3Y8~3a?+(R0CO^4I_*33>t4d)CpzH2Gd6RNI`Qx(WuJAl#TNz
zD(;52wN6sA`-Dsf($jTP5eNoE5Hq7F$bi(5(LvF^1f;_aC6%)*y=#&Qo@+fBRyX%a
zXvK9rTYz-8Oj4r#4ZNtlcU`B8S(tc5OMLv>I5al?R#>BTp0@-YyuRu4+xK*2Vnp6_
zK2qP<ovYbv&B(&%ZI|GqQ2ATCFOZG5kTM!FITCZ4MKnqLqgr&rFKR<iG-S9yx~%`;
zH`zLehg4`e|9dlc_q^tV%DLtFia2UgpQNs2AM{N{3t6SX@-q`Op^piZpnRdylP7N6
z%l@-xP2LJA1aDk^IC$4Kbe}S0XFEZfT9|is4!i`{OTcC0a>~cQ*-Vdfv!OLu8U2BB
z|8-+1&G!fcxmRwA;KAk(#|?Gxrpna#B-H<Uvd9t{6&e;678xivQUBXY`e)`gqoLA9
zeI7YZgVFItI#)PP?{UykO6)A^HnhTao#3e$iqXJIs&H_}WpIlm{4k9R;9?Zr((vNT
z*3Fjclk6J$#_4(mfM#<uVK6ymo_6^1d^Y&OLX##f>j#`a<0{v5fd`mi+Cfvpt>I3K
zm3cN*8{$pol}Sy-aNWfva6L)IC17z(OcwY${$QBHtEuE0IzjkfpW)(?5=*a?grG1?
z2~w9(nLD*`6aE8nE+I&oCBG*vmq(DP1}c_G?4QI4iQEQaoDlmse-tfJtQh_yd5(wX
zhXw}(dlo-1l=HWO?sS>$WHKxuz=mzXw19sZAF83Rypo^5MfWcFJ{@4Jhr<eX(RC5M
z@QH2O$Zi<~%-NxE0f15m<zlz2lm>q?t;Yd#{M2PpNjMu#73sB>5<Bfn831;db~xIN
zzwU+wBcNebTRhZXABva4m&tK#aDL1GnOpfFKyEdp0c75^Gaak)MfM@dwZpfDa}wfA
zeV;D-X?_v@dK%Vck7j&sm|w2qXVSXQycjOQ!bRp`3xq!s6!M48%em<#D61cQ(ITuh
z9Y5(wHyu#!?e7n1Ag_m<#bt}n$^EEX#+9#<7G++jJZ#7-Q+-jrnfl{LV#y;iufT5A
zSgM{C8)%@(NWPA*Ggfx}tc+SG8xBb>_4F{~1py&8D7$;+xDs1UQ+yMtj35RJwy!A8
zQ7+F2^4<Q;ua};;QQrb_Y>g7riSKUI@9rRSs`=Hks{WgcsEZ*4!{LCYW=~HSJfmlZ
zmNvnZ4e54Av=(w>?d30$%`7!UIV(fX{f-cb@N|}pX>H8mrA=)tDdzO?r4j7uu9yiN
zl$sA&+A1Y4H`bpo)6BNcHcdAaclNbkvr}xbSA37JmJy;=qGMUiHVxcg?=Kq;3M<JN
zkomp7u~F2rH1Oi8PW=7kloZ@c&uQOKJ^f)rUD(|G@@Jf$l9=_3Sw;^Gib=gxb^!?@
zT%3_zT)^h4+mRnD|LlH8zgE`r4Kx>Gc~@zi92xm2^6$Wlp^%9dS^&->QluX0|8t(v
zB$>5g+NeJ#HHZFQpmFzIHMGZ)kr3@N>BhEi&9w9sFMB3&F||x+CBG{|@qUnoi*c9X
z48VjNyDL|16|5+48*?FX#4=qsodrw_nwp)yrKH5w+X=Vvu2@DT4n(tHRypfFp^}|N
z1y`rY^&;}}=(%@O=6q?{3EVl+G<|8DBSG|Gn9E@`k0S%Cl_zT{6KJ%%IC7NC(ZzJ;
zlegdG<Q^4MLe}T+*`@$4(#n@H&G*e{Gt38MpavViE=nfQXx5F?1RM3NfdO#MakY(l
zf@SDc5k?n5eQrT{*N`!dRg!n5$WlH%z?}f?7@C1%aNY&zHDaV!Bb;q?;eAM`8$Hqm
z<wEa29?Oekg7U`XD1A>dDFt3?eGd#PsqLAToLSs~=#_G4vb?J!g;$Lm??wSE8gSS>
z`E{~a;u20faE(xTGYI65{;G5#2){ie3{5~Pt(fb}HrQOJ#aWHV6Y*kqib&o*__~`T
z+aZ+TwbF7*1U6AdKAw>VA<EV?X)}?>w&U;b>t~ZCCkpLbGu0oie3hXI5G-4H>>v6}
zT({Yf>`8!}Aq_iG$;3pi3+0M4OsZ=L5GTrIysE1p+pOb3Z)sO<Il+SF8k+2!oIa(j
zG@MnVCu7pMSJ9WQygF;u++Pa}3Koi&d!th$)~A|Vghw^?rs{_O1Bm7bm{VF$7xoKH
zg!df|8V=cuuKJ$hPjc!kL${^75*}Q?NaGt^j>qrqeeG<6cvg}QTggkNPfIe=bA(_k
z8G*UpWNHW)^ljHG_w#7&K1KLlUVO!_FJS@_%C5JlbfGD%W8Rm!!n(&?=x$)Y@{%bK
zP;~(k=gei`>EmE^OOTe<T9?auK#9qt#-G($Lb>u6D#)d5Oo3KOJIVC3t+e6Vu1u0{
zMqp0a^@Jb!SP5C#;s*Yt)imYUCztQ0_q=5Z7muX_BfRmJ+O(n0Hx%KV&Y>f3Gcvjh
zpybxvkvak2L#_MuI$}OmC<c0Sq@~s*FQ0$_pJQ8~xwoytOhz^nI?tjdr#U=R?uU@r
zZ3GJVa{11}2_2>*!5YVu1E^GgIZ@z!S5DYf$S5x0f?X0a-~cK}<A+<69H?{Q%BQF#
zU4D>aM0Yz~W~?@sbc7!B((WR7E8d61e`?XfVrk|_$3trb_->hHCOl^`?7$k9prYq!
zB<ZpI%f1Xa4PrI@Jl%}lg?C=w&QE1+S+!`*A?M^3<pezd7>YJfW`o?_MFSm>x!A4r
zjE)upgJ}V`A%9Z!;Q@>&XM{Ils4iFpEF;Ow=k7L*lPX(oJr@=nE%b*om9_mOTt6GP
zSu`0qg`8pV-Z7aH9JY7-!L`Pg3OQMJlf_2lmF_5}Lg`m(t+_1F0G@5f;(THk@k1DM
zV0u1rW(dxkih;zA^C>_zhQ8SkMc;#_=?1#Y<7Bo!-4R~+w0c%!;ex^qTfm`l;xh<M
zLbj<f9k!`~WN0)V+QNYGWdh{1=mQ3FzR(bZ1A?J7j5Zy1mznWp7TgAcx#XGH3yLDE
z0lWqh3G~+^c)osU&|jQ`C8l&4b<@$P-W~GF>kTN_)<86nBZOuq#+vJvsaV;nvd+Po
zD?vPQ&bcl%IcRb%ikfKv4!f_~4`mL)IZ-H>gu9jbJ*0=Rd-hDOr_5!xJIMQ55V%do
zAVhKegGdRZrW<VPsDCsK4Rk~hq?OZjHGnjl2i`P*F4tZC!q1-9@x<VUMM_>Ce1M|h
z8yYlm@j<Kvt9#$i&8;QMt?Vbt<#wi1<2kVg5<Pa_W)=!H>7yVYag<KAPc!v=`S1|g
z$Rc|%wKt%)Os6;AhhpUZAV?e28AWJt9O&qobDj_Ub>|!{6s60AjdPYQGnIa=TuaL1
zfiLG1EEU)AmUlwCAy>G#Bn=JEa<>xG&wvK}eAcHONBB6VYr%`KhwS3uRndh3zkJnR
z--JPeyrIx&a1Id9h-2Y*qTnlgQKYNAyITQdBI&;wA(*JB=!5OZ$g^~>`lpvc@$F&L
zY8hb+9X-2;p0z3JjwbzDRGUO`J5w%WpLN;C-|uz4ki6Yq9a|_{nV>7jWmpQ?gruts
z;O7Bm8Ak;tQKsTgW4<sN2){E`7hxSZyr^(tQCtMf%Zu0n&auE*5LRvqNDTKkRPW`2
z&$aDW#*_pVkUFB*jFXm)^=Mt@wf6mA3eszKe31EYG5X@LH537DxlBth)oIJBVBx+j
z3GW%ci%zln;gDMvz;l053H!GT`Ej+8lLgnP<RCs%Nusn=IN$Yo!SyX-oMbN?BM437
z(&GYiR9L#mEJS))q4R=_=bhkeX}G)FRRfY%xHS08kiE~x(O6v05Fyxm84mB6PeH8C
z1=my9q=ZHFXn}O3sY@EL>kqMO?X<7kr7mMvy}X*UzYg{f!gXsK-C+J70k~innJA0-
zLTEq*Q`<_602(Y#vOlP61|1p-$UOsX_+DKVe)S?<DM*M!f!+>`fs9ko-u61&mX#7&
zZu9v4-hp7c@^DSx<)G;*w^&2F0XhM2?#Jo_RhI1#cHc3D^>sdAVQFw%>XNT=DAjIw
z5r0X#)@EhBmHuIlCl^ET^BkKET!&%r{g7>6>MwZl8Mb=s>PZ!7ma)_zelV`u;l0_w
zyxPcg5aZc6;~=Hhz_<HcX8eIMAHo}Q;_|UfT4>h&PmWJufK%Dvt@cMh*?{E5MjnPM
z*4bV@^E_qSzRbFCiG-Iel}!~gA}~l%==GJz4-We_Gyz<3ES(ko?{JcB64#k6*aF7$
z#}tyXq#Y@~ioT-oyv0Ru{ZD|gUN-C=&T<h+OmO!d?_v*d<`0xxcJhM8zHN@>hM3I3
z94d!fnn=x^i__OApJ6OY7ql`A=_@^nP0;i7>-F1u1>W~H;q%U1pI;%r`%;wYVIsyu
zzr%t|Z7;z@2lV0skR{}hENE{=ncSl>-wOWV!n(@JKgq9-08=o|`GtP&7eN*}niO=y
zMHFhjCqp~Ve5RY1w<NdlTVwU8{e|m@Ahw>(brkwhVD;of+qDlpv$Wa|>I$VPkJnyw
z8w{R8J8&Fnkb8pfd=%GZuQGP_1kTSnX||?&q61l|Xjf~KF|WESr7|^8k>z}5w)k7J
z<@xaewDin|>6YN4%#YdvTjW7?1tOg5v*%p;?e9=j-6VldNtWIvUzCp#bB(dAGo;*}
zrs>_jB}ow#gn5ylp4u#xtFQQlT1F*lAuzP>yd8ftvE_b_7FiD&@9OyHRiWUFA;tYV
zr?isPbDJe8-&8_N-x^}CDl6vVZ0<i%H8k@mflH|AoB}8x#MmWRQ9(gZcsa*i#3IUv
zAECBlb%68CdIJ(m8C?D2;uu=4R3!Xx)VH@sBrAX`!f&eil_otq%q4r5v`l`-Uz*{W
z-4QjZA>$#03TW_;hbSgi-XC!o<cJp;PlEt98+p&!gSr7B2UGq8oHnQ9Xu5hu1zZm*
zd>2UsRcl|z+A0SB_`px)<VV5gh}DvIraZm3Ti!|tX=3-<M&nkyqSNE-;_POUxDE7C
z&C|Z#7C@$XT!2Wz5Z}*b<i3$Zw8k6i%Qo!yZx0nm?8j-F=}L*%w=DMK#QSS$IWt4=
znry=q{a76%I=+>ACRtVf><^i4Da#0wUhGH$Cuz<2QxvyHNAI+prI36fZ@f_TT;5dK
zjb9-@Ew>c##fYst98+XEr6oCpsEYE<9UP+C0YCF@pP%1A*GF6#fXpAP_L0jx6~;)m
z_@L~MFQL-?(6Zx$Re!E{z@!U*$Iqu_Qu=l5<lX@emmp%J7`ic_%<dEhNe@+3z6nwu
z8bh7@>hzVHa=hY!ARUbV1j*V(chX{|EUHQ@UdP!oE`X4m{e{dj)5*6u@aF|WWZS=b
zGW(YTr+=h^h%EpJ0(!{X<{n*dGXNcWIB(qTp@!`w$su#{IANcMs$Bl+$ESq=j9`ys
z{PfW1G7{HmXGN;qZF;gTa%qhZoMHNykq#fH6n`$0sv68mJ*={}P!E;)LC+<LFf>$I
zlJgFg$9?A)So3W=?(zY!;u&SA=C@yNh6FbNq2m-N@R3yYOPTehv76AN=~(8K5kA4M
zDib!-spo}7<A{`Z`_zalY^RISkh2iN)P-k{k;02oR8>-HT>94(#JB>MkJCfBqG1MY
z67Wnp`j<bxM%-r=ML;w?Sll>-+#sL-I7O{}vZd=IHV`I6w61n#wMuBH4=J?gS_p96
zo-XhW<w;HBc>;HC@P%P+)BmE_3Ku)`1`QiV+8D6CB4w4>If+7Bd2Z-=I%y=E1>tqo
ztk&`62~d+{lW!MrXZ24843CnFAT&Gr+AI3&w}K;_u%Y{2D6T$3y}Q{o;`48>oqL;+
zj*xv`Uk|;f;NI0Erwe~6@>+BT?M>JR?$%hvFf1fN5C|t4Y%Rpshg@r>0qd7!apIr<
zg{lj8>zAmr93l>NbT`)n8iK`={X>~V+*0f{^~~>YnzZPJ^gw-yODr3N5P4^Zk{duS
z4<KoIXE=$WuEk3}_5*~q!5>nWd?!YuuY-e1g}{7$QOczm+2>LNJ6wc?V6q-a{yQ$i
zr6r!%Gv)MLm_4E-uOF5!KgoXXhLB6CD!`fzGYCpBvnLR@dHK7Q9<5K4^+pW=!v0Oa
zkZE>{*j1nmL`|>P?-rUx2k0Bhm1$h*UiA$wGnISj+E7_xZ($^*$20f=vH4KZ7VcIy
zo*ipC!Aaxwit4qTRt+(T5oy+Zh@(0{hYb7E$~I-T*F%`0SPTbYf^-E5xyK1}hD`XD
zJ%ywzm7B+s?a*QfHvngOS!gL=Ysg&6<~GN?j4OyxEDCM8yMoYWgErj~tk%Wm<yJJe
zG_K^UdW3FN5J=p5N*9#n7eX40l$+Q^s1+2L#Ps;DT#^mvK+m!UR)UG9IW|{b2>P+4
ztJpWFVCX*~)6AvXJ1oKv@Ow=$O~L>DXd>aPj4C0Iu4@i7(JRSRT(<Z{2(26YX3D&i
z9S5e%zYtnU&z4EdZfRyR+gQeXt^seal8nV}sZ$=v+kEFDM=0*#1nVGfX>P9)X@l2d
zZAq2mqoaR9Zsp~2=qKd})#UB_dU2+tCGrFDD(?&i=FqYAu7P9Yl#`|>Pj({+4)29$
zjO>r#Ca?f5hW@7P^^lqhrh9O4O*6gxK~N^sSu(tDEeL&y+W3b5ENQ>d()2c0Ip6X}
zK?K1#hg#bOp?jdX)~`1`{F(VvTpi`xbwK7jvBgF@Uq8nZ{ek0Eldbv%UO>)Mr$gi%
zE+Nhwk$Ry^IPqQC_}}B8N>5ElZMLn2;@w1lK~C6Rd6`gFnP7`GYJL8l<nQE`CXiiD
z76vfLvddwwvDx9cVSMH-!RNM-8HiFuQf*1I_JPYt_H-adY~Cu;p@IXVYmlCj*W#Uh
zPXpC2l8Dfhxi53B!9R$O?n_lUJWmNe!@@Kw46pT;)aQV+ogclcuN5dyyYp$)n&T7$
z@ej4i?~{-^{Z@saD3i*HhGsT`=fki%2X7e`NN8p=Jqc}We2(V<o9Q2gr2rm8z{KRT
z+QrEsVa#MFO|;n+ct2l5%Z)Bfl5V8gqsX8b%1{u0If0G$qg#-y$(X*<&@>+qp<y=t
zFqB@2W%17+n9y(h{Y9Ay()!*5-t;TE^^1#C{;6Qa4W41jp%;NouTLC&c!YN6^}D+G
z>x7ZN?KC<*G-%DpC|F#?z@-b}Ng2YATds9dt}yS`^Tj7Q;U`~o3y%-4uns81p&zMU
z_)Ejm4Gs8ZDpMC;C5FNe2Nm0yX2%|-i`WNhme_c2@WG5On+}K=spZ-X>$%Zl94`9<
zm=>)t$%}29fTOI`TM3X$_JXkB4yi$G`Ya&2B>j&3@O7SjLCymt@8E|*%TksgJgS~P
z7$+gi>aHtsZR%ROpeQZ(K8ci@%U8t9s;8e`pJ$DO*{)wy$V>|q(vu9&F6(~X9pBX;
z)=RZfVRV7s(HM_2v@<0SWqhZr;&o)gHPz)a2;GS>H`P$O{S{Qhi|X0sk*4x`_3*o|
z7~>I(sTWCOUf0XgJ0yj0B`S!n|L7kykmkb}S^4%DrG2FNhKz3Wb+8y+#y)Qm78KLN
zLFQq`Ktq1Zr_s}&MW^mHljpytZfA$-TCpj;RNl#BF(h+fx0RCDh?Pmf#Q7V^XYE0P
zP!@toFke@w3yf6u)?e1-z;xdo`n|Pepk;hCmTIQoZBAMn@9gZ9Iyzcmy(pc}+%a`j
zYLKP+PeGOio=Pj5nUV9!D&dpQK;7Z|pMP^&I+df^%~NH2_85cVrzOinTB*@&fwYhA
zIAIxTnm)<`))NkO!E`D4fi@c=xBE48&s0q-7cZVktiT*yv_M=ZF~U2zN;7>xiREpb
zC3=JK`en=wpG0%{&_tOI06FImFtZZH5exiPocybRTcTW{l^nD1QwUINW4ng>Y<N~|
zPLPojLD00U50C$<5_K@WK|>tPr>!m#d4bBmL;o8;gpbLS(d!3ReUo%0kr`Ia74a+i
zX8I8DCH`-qE<4Ow|E#?H$sry`Ib^$0L&>5=-P{d^vBl^hzunweHd-GSqV2oDog*VC
zse;RfbO-y6tD0URve4k`WS&5PD_WH4Sp(&$o!40r)8vTxNN;&|PemL39{W<!aZ0Uy
z<L*BhPE@Oe-;xm|A|>qLl>wsRhb#>?_AMBG@Z!_zpmNLG<|bQx>udnN!T}?M%HvyX
z)><1V?}J(*|EBBeOQbYGYC^^+CdtXxZuw<r(yurCVtMv!*Ts}T-{#nhAuX1n_&V$0
zx%B4<QO>hS9R49>eMp~%-kh}nJ)Gbk@J$DIN4m<ZERjvJOL>L#0@vOB&Q(b`1I)+v
z)0wep3Tm$7hp#U^M2%|NlZ8dNdplJPpW&1|00@wmNB1K?L7BLg?%io9iC5GB?%?s2
zS9T2^<s=AZP6?dHEDrUjJAKF%4Kk~_abIv}lXMGUGaG2yqf&8YPUktRc#G$LFGD%z
zIDrgAxn1fxK3cLx>;qLJ`l>0guChd>q)U(Y-9DvugaE@pv+~`=RsVG!sb8dutQ;Q`
z^4ZyX7ETZulEXiec^V5=77McY3VzogZ8tU)X3Iind|Esq*bZu7^^^GI9Y=w}_q;=d
z-i2DG8q2GVCX0{!{ZCF9r5i{@bF5`0A&3!84Ou!^U0v;=jY}Vt5d`VdR<J!e+NLKN
z4RR_G4_9uwakH%ie3G0mG^KH}2;I&<hwCzzg2sph=;MCSRH^-llzDc^Z}PIdw|^l?
zDJdfnnpd%>_Rg|fAtS*5Ix`(B3uMzwh_!s)N=b@MGxW{SH36#18;%tJ3rNn?r1Tv4
zxoQqKYdOOOuW&f+`5@5v6CK*j`%aJRfVL^WDCJh4^0jJ{aZ~|0DqA$Da=VrMu-c8G
z!t<gAd?DlJgN6t5Jy=95N3^7w1MSNyBoU%Xugmkr=u*mcNM?S5_>3)nTjj`86iwfi
z4*rB1q1T?JK1HxNA^llDjLmJNAACh;%fb`Fd~iSoZ#0r<0Qx3rBlVhCp6K2fJ1aun
zL_Y9#j>8ySfC$AfVuVMn;^6&;-F8_$zU%P(*jUeiciZDfnpT{2-<U>e-{&}$-_Oo>
zX#eo7__jVh6swSop0BcBX}XhLX=;457QDuk24vdI-pq{Ar8Z9fu5THdS$5`y({}h>
zkq(v*C5PEXfU8?F%K@tXpMAZk!J)tEpw2<Izmf*;LSF5jwQ%Q?QB%uPb@FEVM5Lxl
zrd+aFe|S*!v~R<dib^Y&5TdNd%6o|E;@)Sw&+GMteZGqKk)dM{u5!=o&(;XRenFMD
zH1;%{jhh~@GXD00bpkAsdKyTyU)>!pxKs|<)tdVmr1aIsN-&pa(1^SI`QObi!FX-S
zr%l`!2bVj`$RrVxI5+-u8_7CdD3*)rTc6==d%pwF+0o{-C9h}B5lu1rIJ_mGQEYmi
zZI@zfe8acQm$0Al)il=3*5W#*3iXS#YKZ3gOt6GvG$nz}40Fy!@|6_+1z@?@$I>!X
z1~z2oBqH2yQi%4p8w791cjm+WP#Zs;Y4)`!N$scp%b2@s{0xt54%jc2jdqeFY935A
z@1vr4*_i1x%(R@6zM<PAf3w>CWOXPssjuLfzi5Dygnul_NSGv=--;>E*qf+4%h_*E
zL@aH+2(Q8-e%9X2#$k1{arb`6>c5W1fe-8(f{a+qP(4_dIMbn5^&(66v$K09VRub!
zynxqNk!O7fujbv%ZS>l8|FnSddni`Ahjr`qEzUMe`SG4_S_&#Vf6DS0DD8RS!F<v!
zx@K4JfJjpjk(Yv^KY9|BFQq5Jm>>yRzqM{I**JdP%26ssifhL_;R=^r6m;i`15LM-
zjJ!xT1&Vp3E3yUZp~=?SOfG3%lv;LXDBB;l4LPJ+Muw^J{KuJK4l0UF%)X1ma;ro(
zab2YauwlG9n%^iZ;g%LDPA~*gP{YnN-V0wPiZstGk+b{tC2PRt{(9ZMjciOdbaV#R
zN&Z=T5d7BBAE^(wPREJI<S2vI6ICc<1iZ^S71A%9iQ$H^y8vtsguu^yPqM@A4I9?c
zNFQ?7l-Ol1DC~>WsT&_#PuRb!;g~#chQw{2oZN)l;0^8;H{@^|7<cJHKL7rZkx)Ph
zsSBGm$B!ptU?FA!Z}VEd{T?Xo9;Y-;3EV%@^2YBn=qiihzRSj5W(Qw5i)#gpFD||&
z*VBGgN=}Sd1(t8Je>sULN+pd&5E^`qx}6%P^)mAf*qZ)rif{%1kinSjCJvhWNs<0C
zJ`Fw%<a*S~FrQuHF3I*9B_!<c*FU5GVF7xVn@WwP`%*D{`~?^^oF|AsUJ|MMd!Dcl
ziN{q<*}A6p#8mkdg4<SsFvLR(TeEco`mluhCpcZ>ZIx?<MOEaXvD82b<U`I%45dY<
zmXao0+9Q$ElEW6QgN<;8xx}BvvqB5L`i2I3FY;h8m{H}6qwikc1ZO}fGv91@^K@)a
z+9(jAS6`?$G@}Y<PdU?~<-AITw<4C(|DB#b2~^|={DcPM<;KP>&I!s3pOxu$I%C6&
zn|LL!lyCw_w*UHe?wXmYuG~sZ()~#uQWb`G2y{;YG~&dhZ^TU6c~7#10q`<5#dx5O
zWMjp4@nxyGv!`q8<G<%sq{uXe7zpwGeteD@Oal_<%4=IySTDeepwF*4lK^#mv2dMh
zS|_LYx`J=*DxW?*2hZ@2MDw+`fA{kXY4H&FnFdYfrq`<|Qh#hWu(h@Ib&3YsWo{uo
zz4dQr$MJO6u|4XD0ap6vDS|fPOw;rA-%f`BX@rh`ZapCBv7~kx?|TR&e8aaJ>STKf
zbw?7@SIo59oxI3Jpqp#9SM#99WL$S^QcUb^n+BEk_6`q2jyH*h!T6qx?6daT%2?bi
z`gWMul(W+JzSonZx>uztY%0w&izROsG|Y#jt7osY$%(3CA#7fF1M8~6bL=m|B<kKg
zt$0JlN1>yR_ttIWCTtQpAw{?=X<Zc;)1bI~h$`{xJ@JH+Z{_9X{zZOc@Al*9BAoBR
zgq~~ONCDN0Nqe#vmnGn0z=d1uJKM@`FaE1H)eozE9VJ$F%;>1D@OtEEdi^Bu7Nu~a
z4kD+ohwW!@D!$SD>^8(e>iLFZoHl}O+}*nVS;Ci@WLe_V)fpwWvG^2|Y4_k1S)wpZ
z7A_Zr4?cv|4_aa;DmZhT)jEhn(-VH}r+Z8Hok6y)W=V*&W}VmPH^E2${w|&#Ii5}r
z9{b!2f84BM-%yw7)542ARs!UR<j6@9LX+M!^9thke4o`swf*kn7IaUqNf9vG6@M8T
zu}rx*;^Z#j-gtE@q{(n-gG*IjP^@mK%snq#)&y(USlXfkEjFj<zc7XVS*4?#3A|YD
zg61CHmR5pt-P+nX2KXIwi@O73SBkdlXQf<N5W%=*!s?qk%d}C}Te%IwLd;vci{(X+
z5BuTm1VY=l_hc!6WU6g$kmJSZa`JOAzIUBr-ez~TrNujqzV8E<YlPlRbb_92Y6?O4
zE}Smla>#6ovsET*&+fD<7`}7FPwwCYJLu$@?$w&7t?e?YIJO;sP39BwhB^OUKiJ)9
zpK)+F8vXhtHL<UFK*^h*RN*~3x7GX1SH-z`E}nbT3GOV>llaT(iGg;y&pm`8#`)vM
z@q^w`9zm4pK(ie;ByCr=xOY`p(n<YrdTmHbf7Hje3Bu3iA@P^7U6uyQHZLg)j)_mB
zdE>LrmzvVv)@&T{K-IeL+=0CM6<J&g!KJB2J!1d$v`eY|O7iRQ5)aXIk7`Z{cV3^J
z0Cbd@D9&!;LU!u899TevurBV&a}34$oLDtcg(z!=dtsF-^HOy$O8p_1tmCz;Qt2C?
zTSJJ8(M(UT2u-zSpycw!MMR(e0j2-4rYISiLzSJ~a&l@o*=a=KoD0|_e*|K|<~;2V
zr8}}(ko#VCSNVhT%P`?|r!Xr4AN7b0z*@g^tRYV6(tD~P?dI^c;xX*?nfPAaO$_oj
z0}EqHpPA07T1uC5G!8gko5=bpps>xPS@6zKQv(on`>*Hu??l|f*H|AG5TzDdsJ_K@
zYNd19sG8g>YtyR%gKMVXc9OE6?n?Z1a;AM_Lqj=S{YZDoqUxQ9v;S>P{X}Ov8|k%T
z3j?X;whf&%VIdvwsPEO%X1^+uj2~n`n#4a}>3R8-di-(2K<cG-P8#Ymw&y8=Uq(f$
zgN<LuRTPGTqiz@bo9M#sffl+nnk+Wbl~wNG3=JK|>n*?Ua<L1QpHkkhEWeKs6})L7
z(0BDr-VxKAA+}u`2++#8q7WKLW7XRmruu~|z)Nv?rT~n|Co53wxNh+QUdR~<jrmqA
zDY}5NU{0j{YObb%+E5KbmbeS1L$y(1g3bG*Q&jH3X34ccf*n{?HPJuc2>$yQ(#}UX
zcP7PsS9*a?165_f7bQR@HvdKNFbbBn*d`j%o8LM6399$eBHT1n-_TcCa;TrR;lV6r
zYTcQj#YLEY>QXUdj)kfU&p>Wxr^8@fKJDuK`)G9@H{&l|rsfZdU8-d~VF$uvY9`|E
zOmO7iG}gGhT+p3ZSoa403e6kUN~Z4jKHf<t)^u`cYU+<1$?l<+Q!zIdzN2qVE&V;*
z`DZ>c#UH<zE$l#4p~0*yfsZTbqUD{Feyc|41aTT5851YlT)DKIXt4$E7w+m}pn%Xj
zR}UxmSj;9yEiSM0mJ=K>gz(?);i3yV&+|!^j$KF;EXctPm?|&bvTN9{3L;-WZOb-4
z_mZB7lYO_ZW@z;;_t|JN=ltus9YMGVXHOq;#9xPXtB1ZIkRi<D@16Xe%PL~=n!NMy
zV(pda=Q)TZ9Mb|f1OtJoU5XA;7efmEpc95a#3sm_xvV`;zh34jkzHnDl{y!dBtHwl
zv^6!s+*^V<<;wJeF9prAwF^Taz@1!&neAsDs14tcBWgqX6?hVW3p(V_Q4aq4m$mx3
zr>L78JNwD1>YmQcL+o+=v}{KZrA~5)2w{Q?G}z*osvj1qJ$_eLqVuro%KmY{%tjwa
zPTPUaujYs6{x0obi{ia;^QtI4AOgre4UCxu`5PK08U~jtD(a-CtAD3VFaCkva}R5l
z9zjIY>P`$!^?s1$CrD{$m+aq0;GdoSg9YxF435S7NToDnHB9pfWtj?Ma1k52t@fPw
zb8P8f&<{rhiqxVGw$|SFw&~DAGi_*&rl%{LLbp+_-}{Kec3!(gQhYV-m#EqJ>4!Za
zvgDho*MDjK3u}bM9}6NqGTyGX4Vw>-NB8)IuT88Kf9hS0Qt=H{G$9@xpPXICVJg5}
z1OXCtx0NC2sY|SBWo86rQiy$SdW~7+!vh{&iLB=6@W%C&-jC|O`?gqhRr`tHt9A9h
zHUrz+@@^-`8|{ye4@<3Er1QR&MjeZSM8ogX&Nm|bj$AW}kzBs-7MxubT%>OvUrR5H
zvv$8)DqR=htzkD9ThrS5?AokDTU#xuzaYRee1G${p|VHaQF6|7{kyfV=(X1-+y!h6
zBr%Q@rVP(}60}_i!{<HB{}`XGwyXbT%GqAb!o<mbX_@vHM%}Z|jD^^4T=UzX0H=>X
z3g~QbrUANwN$ljXX`Px62thrK?Z_7Ldx0YLsFQz#vtU%K>gPqJ(idY=*DH@obP9Qc
z2VN3u(_DY7c08b<-Pm6hn_5_1(!k-=;*K7&i`l5swC)#P)aGOF>K^jc4UiW^g`{jE
z^1m?-_w;?h!Km4~hzk8Hnpw?8;A_8)p*X%2zFMcUXU?t~qU@rKcMZ1$=D*pQ-r1q7
zNV)A4@>i68IMWCC;^TF4qZyHcG|WanW_RenaNH_<R&Ur{`|Af^>9l!d9H)EEv}E2P
zw_=0rOhJJ2f2cZ&4{>;Armvg2j}|Da(2j-1;$ZamvJ;HaFIt4zE8*p6FbQI4=!S5z
zMC>BK)+P^x=vMxmw*)Z>!~Twx;<}TW1CJos>02)!3A}DHnXeXh`|>x}bsSsb`5ji+
z_U!avP3L9=hQzB!3bxJqk6YC$!PZ-ig)7-#l=nDrW`nR>ugL0AHm~-OrQzJ4RiE17
z1c^7ha7{ffC{E^RR>2iaGbvJ2QMDHqjpucAbT}J(QH~SA+0O@hN7QKo5Y6S!T*2S2
z*_&oY0V7M*G?E%}Z!59ZSw`I2iM`*5a%?q_tG3xstK8{sZ~oqYq6d?55qx^3`Di`l
zGBv@<+j(9ByYIgityjC6KLQk27&mdr%ClWF5O7sbmKlMFN`iILrP+*Z)Z^7m&j1zv
zpF&qiad{Vht!PUZd-aRRY39V<wN5_%pw{({u~GYNCmb_Jy>Z7|O8AS%)iNbjd#Gvc
z=XB^=y_%}JfrMjwq*UoQ_qV}+w~N3whoWTH{WGTtYbCR%zi@qIBCKgD;XeHAhLe8Z
z;P&?MV*x@PsSvDkPdE2ypz*~vR^2bz-KC<B_I88)o`~)UjeO!^*K)C;7X~z?md46W
z2>aMp@`iQ>W~QC-Ov<NQlX@0qm!3r$UdMDgish{5XVDXr7VtYR*iP@}=?}w5^6&LF
z_2k!ttnb!UV$n)zlCqfSFfY-=&54GimDlP~VQp&aAR!7J-rDB=d8bPg#Wr+kT;8>1
z@5A(^CFM?P7dA;vGCNIPM$??;W4z8C4R|1}jw!+|&U!<+Iphul2G{XuaFrjYWE=P&
z=RKF(pOv4N$>0QIlok4t42in(d2fL6Vo4|shc0(HFY!r1wt)oGDnI-t;kL3YLm|6_
z(zE*nP*G6~$pFW~;x<BJBv3jX`;(%Yg{HeUYt`RO2R<YxKz_Ksoc{a%$@3uPQu!Nq
z%nO3jl_m9c-Q$uZ=0q{2mJ8cT%2y*}IJJxW`Ju(`_9}A^RFkVJXz2@8@9I{P5`~-+
zNiVFC+RR3_w`qmmikmwv$THNtv2^hzfKh+DK&*nF<xk`rNi2I{Ar;1L_Ao;b2bRc_
zp75jFRe3(XF-3UdQs#x(G<4|`0RE$ZcJlu5%eO8HiAm0jtBgiI4qesnOua2WK1PbV
z+BB96i<)x6q~w}oJs1p=Ha`kL9yoJ+oua!d`mIPd2WrvvdOxk7+SyfM@d5_->5Gd2
zG4#EQeKLijSew#}kXpAAxEpVpT1;x9w4s~?eC)H_>zCO%%PRC+QQOOy2UMS@v#^}-
zizExAIqZYL&~mx(ZCrx2JYA<;^AUpOkn2h{e$6vR6WL?qg_L=oCsitLba83Z7pB&(
zj8*WA#4CzAWag&H(#8fZm@wV7{_qZ@hX_idNO4hr_bf{7q_fs`BbpqpO)rJCtoO>s
zT@5^|^W8<}M2|*YDKlSlfoZ(K;ZsOQT;v(7_}I#YEF7Jc+c`?|UI_U8aTr2p&iVwJ
zAOo<g$ncHIM?bL68XPmb+&z=n#}(3|D#<$x(-*`*UUX?|WHgeReb$=)1%PYc`%~iN
zRUW{Z05%m_T7h0(+Ypz)l-_T{5n(@CwwYK00%xLwsa0%(@<ZE?XJ>u)*Lus>#?I37
z)_*7G7-D+$v{Qf{q#;(`^mo+NwL%nka*~b0xrlpg&YYb{R{s%yTnSUjMGTDQaUq=l
zRWZ_ynuO`kzNjWK;h<;z0>maPvs~zmo9_FN<d4B-5Z~sDc1SS1W(uL_fSVZaC)~Jz
zaMsi7&Yu1)VaOpRc=JLf3gwDbUUq9?rV*EyH(}#7PF9tbDNuAR&hKzWxiBC{n3Uo$
zGV$E}GGcGos{`47ue{+i#pVQ@EX`k>ztSZ+U7b;ZTlG~)InYJj0pUnvq@_^ie76eG
z?9q9{k0q&=h}+eD%QQAfB+Rp`12rO7WX}1lFZ^M$ESrj4*`|26*wdT`Y(HD4AK+oG
zQX;e~#?*NJMFFJX%g5nO$J5u^k4H|YPaARKcWHT#Tm=O?>LpFGd`1HYf8)u14JqKE
z;g#N1+;^RZJSq1aP@VG=LP&ZlSw|Gw_4s#=9+GC|hq}BfJCIdcA*Ycn3v>>(Jw8@t
z&)VI}YX28(e_sC5TR%U)l-@S!U1PNM(+aP*Npr*j<~ZwHdr_g6wU-w60tgR_BX`tP
zce<0bmobm*@53X1@AJdb2CJRQ`{XYtLiy&#dWEB&Zk_e-!T)dIj}Q+V)gx~GEY^PX
z6{)jf=Y4P7RIbMRkF&5z%`R^dDfsOm8zf2Q(~Y%i6?OH>UxU+mIsjZ&_MUKvK%FJ*
zC;pjQzY|S1gaeiQeTaIrc=)gOn?ACa0>$V>vKWfG@xC00;ea_lJ`F_wzV`mTc%aV(
zYtp(Ya%vQO;7<PwPoT`QZCCri-PEX|S{>ox0_3s+O0>#HG!-w8%9Z_7|8TmQOV{Nm
zA9%XgNRGcnU}Oba8~QO^Slk>$wYo7;*FI}KP}xQc)7u9AJq*sKF|yNwmdWA0?O*&K
zP3IoZ^!xttIixvnln)UzhazWk$eB4N<x~!lbI#}U>Af^e<eYPe9E)-slS2r#QNqZK
z<dha8=ksshKYsu0uRV6W-}n2zUf1jOymE}l*tbMC;@SH#R`})sI4<wY%Mo>F+JfDr
z{WeV?<2p!E?i`3hXD2h$_eP#MkNU1EE2a~A;-c-P$km#$SPgr(T8+`5R2~a&Fo<z}
zAqA;?KJ}AEBdgKRs<s>~BLq=To5QlQ%Y#8I`E=Pbbj@PFN}&A6pBboL9h`aJcw&JF
zV+zyV97RsBZjy-r*C89rn^sz6x}dFVB0xl#4zt|7eli=auRp*Stu;D-jiYYZ{7s$C
z%KBbd&wM`=Reh<+*vy?)q1mY-ylC3!Wh;;4mr(ad3FwjnMLN0|@rSdLl5|?!Hlw>D
zA27@=A-|0DM8Y>UU^-~8`bvVE4Ixx}c&NLF&grXa0(#h-^P3R=@z!*w%CJw{JwS>$
znU$uPRv{2gCjDtL6(cQfarX~$ckP;o)VAg_feHfoqFqaqnNbi8TUA#%#Y|D5k0X8<
zyIQQoWyVYzz$a+?sr-?<{TqjO-YCwMrU_gdbY*IXda7Ow{7*-esX1~ha+Y#kxNnHt
z&!~^=CcmGTN5m0vZ4!B*t&7)zw^q2tGuR5&lI!#4<~=j0=r8zkwI(pv7P{l)l-<zk
zJ4N0#w~VPpFeKL*|CXWSL=U~UwM@wkwX3UYy&CkRfkzZi>YL-|**eafA9<OO5+u;I
z!+^A{17BPU>v&6y>LIE9Q%-ZbC<iy==-?6J{JtEtR^QXrts_eOP}2|mUR+*YZnBrl
zhUVGwj8##njq3({OM}TzcVxTon_(8AkBh%!S)6CmW8*)p4owbjI)UF)TS$hegk%w#
zXI}ZL1rt%E`|{rPMXXBf;l}q6prG_f0mrf3pbic7+hPWku2xz72#((ET~DH-V+OS(
zSy?f$KqyR`%4yTW7y+Y6Cd=I7Mq$ix-1MB*!LPUvr2YB0=;*&CWt2(ruO4?$$|7HW
zGLl^Od%+iB%ez<`B}Z+pQWZ&%fohaLzE>wXU?;FeBk|PifDS~hl&THoXMN6oL|<td
zddnw&!WDVTcGY`>d-MfQ$@7!_zUKu{PocC5O7wBgdkN~z2Iy*CSovA4#{>#8$5a*b
zi@HELdezfhwEX(s=D2t2m3<@Q>e*eyG%mac*BCdCv0kr~Ae(zREp)JIeKC~xhui9X
zvc{>jd^SpCY91pE{_S!Knya=XJr%!f<+kOtLL+}KFA{5TjqBVp%gNmpsq&ghu#eB{
zvegZ@I^I=;KM+(-`<?VScr?}cVXI8AyS0ww%@w&)F*&s+c6rK~!xa+U-JV+$MrXBd
zC9)EioSF|fo$wF2xSSL5W4unLwHml&RTvy*s_&?{-z~5|6c{4(-HFGAg*?;53aF(W
zy9o3rb8{MCDq4xB8!P;N{ig075%pgroFr4<_hKLR0R|kG?Nh`Ndm3^&%*{Ua6x$G@
zxRz<c7<*Tohkj}QMMpWI_{ZGB%-ht1p2Yez<JibvY`Te)l(?;>vG(O@fh5)>pNf6L
zBfod?9$k_C`N@~9LI`Dx-12)!@Kk7lr;}kvgO%RkdyS*9wByB(;~wAGoEToj`lBG=
z*y%t%z=v-<va9}X-gETTX*zVP-HrynpEUUUc#(gtJfy=18rh<2Iq5qmcU#OvdGOn!
z2U<}g7JmD1@96%U+T7abg;UnOsNwQI?)9xaTiGBB(XUdiT)_hF8yZ^-`1zvV)BuFA
zC^7&8F8))N-Iymo%$rb&9=%NZ0G1zf9ZA;!@(mFydVyy@;qsZ>VqbqylK}7KZ#2_x
z#VJr?|0WgbR=xKad)*Quu(E>qD|^H}y@&XRNZAB)i1nF!PY*k>b=fTWiEm9)1<ljF
z_fGTC-%<Nw*3PE?364H#6@kMzC6~=LCU8e&>7LMLC&g0@R7XyeRAo&K(@1Y=F#jHK
zksAAtyCO4Bl3$JD9OVjSIziH+opO6s?lF|yP&+uA4==n!E1c-&f4Vh2#9f&~-X1>-
z+g3!<R?AR2%lPkt*h<Rz(8C&db(;K+_E2^@=KTO>&_x)7D>A`D;|mg|hcg%b);GJ7
zo&c6(&D3kTbeXF>ywuV?;`Q?oW+fn8s2qNoQ<Nhs6AXr*H`*fQFCb7=$@GwXaK50N
z%V@T{YR*C?56aq-!}OxIsos5Bgz4RP&n{)c;|*6H|GOk85^rGQfJv}(MIiKDMT>RT
zo#<dM5|Z;0Z1TZE=z;<n2^j;&>v<;N;k9WO{odCdpIsP1=`wfAxHo+B?*hBv+KVhH
z8dblhtE~l~$ts*t6sibBfYU)OJ?%rqjKM^4@d@GpR;02(Y53AYd2y!SReEOXUMOhR
z@ldhZi9k9MaLcC!u?GtjJQX7us0K|L&`#LCo4TO|*QVH`twQZ{*-JGheOdLfWb$I*
z{66utd!BlwoSOgVa_bgw-^`h#-tE(($8p?_KB7aSPOqwIQi9QjS^BoGJ@i#;zdkTK
zuTS|Zpus%OElE2<74=A18Ts)o5ZNBhJr)J1VR=7vU8UAu+xRi8&7)lNXJvA<yK_Px
zBya%_r1RE)L1o-F%jKKo)S%FW?$|@D#L!pZ@HS=LBdN1e91w2jslgt*%wKC{1U=*8
zg1+!2r7iyMN+;OZ<T%;#^rwe`;)~(mw%b0--(-_>O4KSt)5170bFvnQ?@zh^tAAIY
z6prvPD7p~9aX*uu8hxbJ*vcHWod!$XD<!^~d&eW`irL-Lck_%^RF4ArMrL25AFur8
zPm|#KfxQ4q0lezVz3{&r)cbpj2WjRVX7Z#vRhBJTO3uL%)v|q>OmS5<!U(3*owaIT
zyWSznqwe*tt6h=GjvRff=~G_?O4@9nap^2|<i0e<Y5u*LbXn#0H%$#2M#wfq`8LO3
zDpxGKs(SPs3b+(?ix_zz6+eoa8(i!bFyMSaUCDWm06@1~?Q&>D)78xh4O6pIHD9%l
zv&WlGHZL-{@xIn$@l;D6TE`Mm>eca<jHhSUsTLkwc*^j;cfVa@*frjqd7)DgXn5E!
zb9H=Nl2qVEY?EC)QUgQaozFklpisk(`dn64++F*lYS)&1TtFa>QJGI*3N_lV_7+X+
z=xPe+pFWfvdH!GXv*HI+XmBvEqLMnAdv)LzC+P9@XRF12oNz>bL6Msnxd}hO-&+7+
zz$v?F-<(lFvhte?Tis7N&8(caP>TBGis#T^Z`xk-Ot6vKbvYVvhIP9XUouZ}$anUk
z-x=EUbxrll{oCW_H{beuTfLIfrZ@aXtiK8v=t$u>#L?({lY9SlIK0z5U6z->wiGuz
zKQ|cIA`q1d<T$7pe+(4^{g~NmC$5L}XsBnSt2NW25pQu#lHOKgM<)x1Hn}rCsKZ80
z{$pLLrks!olatf2(z_``z0E(DMJ+FGbL!9{yy<Av#`Tj7O*ker1{dRi$jBdUw*k>v
zzjXSC1gadIZqTXe*K_g@r+GTAIb{p^c~>)5s*XwkqKdd_F7qF^drmda_kh#V^ZoVi
zcjTkP?drRwrJ6uo_rnL1ZsKvi`34+WbtvkSJ*UOYJ^GXTFJ)qgzJ%X-Q&V%%wUw><
zNWG9bHGRr_cWhC&KGepzK6DM}DzfpW$^<Q=c=E1tT7{CkLvBJoY0&ceU8=v3?n#mS
zc34NMMwPy;tKS!`+ccuVPr=R&mEB!kFF8%LoZahjSXpy6?_+KB+pK*b1KyQ%Y7wAl
zn<|TGtyfCn)=1pviJcqrs3>AbLd0YRP8V=(lHSh)5H+nq-?ug^v;j(7bn3B7#OD4N
z^y3Etk1d{6yBE`BttjWm9pUa$9VPSKC*A7)dDXj3lkewTw3MeEhkfn6Yi%v{>T*Nv
zSEM$QN|w$KgxFPrqQ3u&y{fAD`|+XXeg#D>o6_^NwuKt_)Xq_{e%bs-*=-#Ht$2Rg
zxKF^~n%>^FpW5<pVgtgsnu^2e4&n@pDSbk=>6qj03E{F$HWg%$WK$hSR@0?1z5l&=
zJ-JEBGW0ctK=gnl)3dOcP!@&rnabtD(bTvBC@?o#94}s8DV|ax2op@O0qM}g8RZ2F
z>|AHKEkZ{PH6G>{C}(iYug0N4GJ_Fn{O*HO)kg6x?g+rL#fRvYy0CzfH<Diw^+}99
zQlY)@)zMv2uS%z+ww2_Ae;J17n?_|H<K!ZBOz<_Se2}_~oYd4*kAg*(Sl3@5neF6{
z)9G<Lfv1}f$vSwg{;Tpl5;WQ=Z;7D{H{uRY$va=a@*U#tTPlquZ{TsFM03=Cq9MHM
zBS}Se6o3>1;jH>9Ln;V}PR6ypU@<9K0UY&nHBB5CLE@WR{kyBBnp6?jmPk?5_y5k$
z#h_oy1d;D=xMWsanS;nTzt1a#*D8P|90vVBf$>Q;%{14!O3-g5#5XE1XaBB{v0h@b
zdH?#aCw4qua<jwWa*_TRjES`P?poG?Ie$5zw=_GuCwt31jqr9~n%a}Z&ByT^AMEB)
zp;NoWvHF{!>l@b}wyksU(#p_^6vxw0xOghk=WfwhO6u(#j}$RhOE%U(UcGNogJtC>
z{s>?;#F$D1c{MlLUA*VC@PbZl)n(#%$GRYzauuBPWA^rAQF+dP`Sp-*vrIJYU0=Td
z*Hm{En(!<0F29uJve6d;&3zf6`LDR>x%=ljh|s_<Go7sZsV2Zc$4t3V*@gX0#9c#n
zQEHl)rJ&fU%SIS3=PVkjMGlN6rzds9`p?q|&Lw+TSkrT>bOyTW3^wJgrr#CNRDUpe
ze9D$_H*fu&=J}d!Xirz`X}8r6uMQu8IKG`^i!<rEe<<rfPVbf*{7d{QW>)f;FiX}h
zId&_4GmhHt2L6b0kAC8zfs~LAgv0NsvL+g69-W3gn!-wvn;N3K(komTmQ<M+1_VF!
zi1dnm<KsFWqo`iW#0xGNw-Y{ZkJq(&)!(|Sce(QQ0KGHfVwC@zugNp1wJ&*qV*GWu
zIA<A?M&!P=<%%Ld6s7W5BE7T!^?~VQ8<d_7Edw2#mX4Kog^|Sqta(AQOUQLkwknK(
zpCui=cM=y-NqU~V$4tpg2ARyJ(Qz1yIq_%oD&rw$C?|(P)4!ew6IWe$`7;kQsGhTk
zHeAOwE7CR_Eayru2$ebPumg;=*k$%B*qOm)h&E6Gv?~?)GtE%m;EJ8Up}YkQL?`QR
z3#HO95MWt#eJhD)x5onZ-7BtYWXF>-kg)#xt0mj!2^q3@wm=h^Qu_bEhl@<=a1C`|
z&GVDP((^-U>TloEWkyZN?+4z4Fsa*@Cdl}JIphg&p1H&>l7BtWx(U<LHgUH+JXoMZ
z^>lrh6G*53!W=vnZ=Di#G)8R{Bk1dLaF;VHj^ER;K?lER0w=e{-2B&<pL@T+jZ>Rq
zf-?Hn&JEtr(Q`Ansdu+)`)2Kqm;HNA^3E4~KEy^p{b}A7dWsbG4=3+cwK_E9h1O@k
zmdnhOjp@-``K|^?wtJXeP)^pb-*5h)6BiMemP)uKWNM`&AS7p&k0jvsc1}_myFz(&
zE#Hx=^ayQgdqZP0W4yA`ZlU|9(@8{Bki0yNlXHH0g|{TiBNqR(IbHq+4~GxSm;5%q
zaDlEnhyIlMoGfk6{uvqxoe#cm3W`+47T4D6Vv&dMcjy5*X3aefeH||-hpEm*lPnt_
zbY%^CnA?40{fv+ZP4;U2*!3}R3Foc4hY}v92P4bm9S&cv<!M2|NE8y;hjTkVe5d)2
z+%!+=Ir~TUSX>J_Y>_xl!mShk?ZNYWW4ryPaJzSbqgsMov$dSWyx0w3_{LzZgQoXs
z^ovq)07F6P;G6k5NA1*pEtQr${$aDY-8r&7pynQ5K>a)7P7Y#7B?RBD7vEiz4AI#M
zGD`IGHoE~jqu<o+1csg%T%a|0o)BX9#Wt_pt=>-3N^dpxwT%+9L4b*HPOxZ%_{mCG
zIQDXPba0mGfVHW$o>}qPOYmjE{5#sp?J#Y9Ait=ffQFW1C|X60Cu3<qSQw=|nBSr9
z!UBhX(ug{0xenrVA3Tj?r`DQXBfZw5O&XT3YZ8xBOGCFa^8XbDU@G<(Wq6yv$!9$y
zxYgq7Y%>hKp#;t=P(Sx_Ce+FxD)sMZ1|)fnhNnL`0`6_Y{9<RAuv)7(-xHeYIVS*M
zFe>Un0$|as+ai+XU-Tkpypob2_ZWl_?Zn^pKV-D-U6$;NZ$^TUb?@cmTiTi(uk6X<
z)%zZA90Di1bO|NNXWQpTnz2XYNq$3^sduDV*v5L}H@T;__Z{3}^Rl+M#kBD6O+~R(
zcM6fxe<xp4P2Fk|{9hGukt$`6JOo-|?&-nxT~QU3kwRV2LV$mT;62x%O!lv(ei+Og
zJFI<uAAdjv=#j%Z>R{!-$$_xAxY|*`{=cmbf_W?Kr!;W9|LANFGY=fsYbFk7*_=ce
z+T>ZOyJ`MfCuqd#zxXVKABxi)GzHljn{~x~_@*lJS;DM^aBG(K71-_o_+7Ey`TmlX
zyAkNm-vZ+9zpd@nRd#{6^L30<Vj`tqU#D=5F+QV?3H|QfyZ*NlR)`fsJ*I?WzuyG8
zwtJc%)bDHW7WYg}NKV4^j%k&QKe-S(w8-z*o66(VFHgIv90u2~+HJx+p@wAjc?oL8
zFL*waVB;jUV#Gb+$SMoKkzthm<@<y83Mpgh$86UtT?f9i-AdzA@23F?(!n6`e>o_+
z-eTN%^ouP+5gTc;=IO4q|2u!+oG5)9n>6hEL&F|(vbhuC@F;H8*XHbGyt<lfWo60y
zfUoNlazG&ZH?Z^~V_Rx>^o1J0-|UNLY`f4~`(98B9}c+)(-sBj-D4=*)sdP%LniKK
z*%WHd@sHK#26g!OQ!gq0g^U1(f|9GJ9n6XQdW#eHo?L@?TUp=md1K16;p(ZfgsM7D
zUC5_g3F0>k4pT6Yf@gRC$FE{vJ04jG9gJoBz$5dO*Y^8EVbIcqFwIKK)BJ?P&($v#
zo5AjT7KR|1WqnxyGsw@c_d?Hy`ztgx7k+^wZBbeO;y{zcwxe>IJ|k+{fE~R3eqq4X
z33KdgpBu-wPvqr@%J1YX2fz8!Btcv*udCnRmbv0GlVPj+ep^krL~5!o!DeIMz+aLh
zZ@mxxLOFxe`pL;lx?5|9D|5E{AYrTzGh927871T_@gfBXs>L=i7mMjC_=b12jzzam
znOA8!+mNx62<50!R)pBI)XZ+m{($fbGMJNNUj6cU;}rX)DCeWERJi2P&LM3Mm!H#<
zk5-1C1=GU7<K=RiSob-Sqjh>JNQaBTy`he5Ud@pUWvspARPI!<pSuBi0`A(EWq=G3
zz2)ulZc-VrpT%mP1>9LaUcVss?&|~iCU?ESIBlz+$hcljZXKK_a#rBoJ8I|IVFPoK
zLGJF^rynWMK+0K|Kh)$cp3RXHy0Wry%#tmBYc0#&FEi~)w$owzJdLbcz|lxS%9rk^
zMWOpQl|Y6_vA$0xTw!3c3w6Iu>=-#@Nf}N({R@y1B`eM2(`6r+;IqXUQx)W>LY|%1
zWH4{RhEsomho}m^p_<DLE_g}QNmw7zH~d-NCZa&jZNJ1(qUCL?p`b8Sa3zdy{WnjV
z5$cK)Kckkr%ysUUnYreu>k=GH<Ynb{C8EGe*cWqmgHlN5C5VNBGP8Qg<x8{+|3VIT
z4uQqxrqh4^=V#=3fIQNhO|5}Szl26Rgj&39)c_8LC7ftPQts>jW$ZdNTlfhA%}2hc
z1Q$^j8#a+^0y@PZ(gQs7>PZKT+!<`3@l^HWe_HtMsSq|KA2l-h&ObeUiv8=|`R*88
zbN$P_GL?-2GvVw+x$`*s*Wgr`&i%nUz&eFX((rF)(sP>6e!=flpVp7hM)CINqtR90
z9~oH=j2KuC7^eI#*_}0V?M3($M~R`q27~feH#yY!2G!EQ$zof=v#F8(f&;=#;o#vz
zi35c1ui1Ba%5@0Ss~R*|_?u%0yc|2h7>n&8iL&<fC8HVTwX8DlDhDO%x=&^SpI6uz
z#l_#EVx%yfN52*EL-|+ylrU&C9!x(+xXb|pX9&?Iw@Mky(Q<$@j^C@?xS1j=y5)or
zmHm2?2L8C$gP%)*PU5<U*oMXvRd)uE-eiArK&KB_Ma-%0v0;kLHT)NTi}jrPu@l{L
z`K>?fe|{(b);yIyrz>w1X54QBl)OXVz<>2e39v^v*5|p=q$wMi6c~nfl+{!Vn#gj=
z3Bnq1|3YHLTbsNnLpY3omlE>B!#0h<u`7`I^a`~->el~SqVHeWY;r>G74XOiRpvH?
zsCN!w6FZaGt1gXI<BSo<RPD-t=<Ij>^93DaggC-b=Xy@OOjOUj00|JfD{A&W=UEP@
zt2I`0_JrC5ajr0kdR<h++PsOGnlfh+ltr-z>(r#A%SKULX6OA7OB)+;wmXo=a_gsY
zMYi*`lA!WXqK^hR)>0}MiR**Q#ageKA$(H_>Bo!EDRNsw<$ZdV<~cjF7v*oCfy85M
zrc6rlA(!39JcU@00D7Poow)`bMSZyLERwaBU7H?s%_uqr0e;>4dw~a`-1|`|(O%kH
z@&X*?t#bjJs_}Mye+XPUW=r$aqJsk>`{@Z%Ik3o3I|K1>`7S`+fA5x#FD?_wxGAOm
z!%(y^8S7{D4CXC{MLpKUzJD7kX50UF`dsUYTS-NAEn6WlzBO*Y#u*WfWkP2<$IAqn
zhOI~kjZ$P!TRLb(MPRV58y@htM9jPSw)!(1a6Vs8eufz4qfwDorzD)(r)%7lqaS~|
zLa&_=0d0T}aDap=h+4eLHdwfhxuv?Z+$S$F+yAtjUG<&IYL`AD!Ql#&#E4v;ZAM-?
zT4Be5VnBza<BD8ROicbQR)NQx6qRTw2k)c7zfOm8kmx=Zv^*&qn=C(;i8nD^In=`A
zKiEs7C-z7qu`IM8dNX}^g6o1sVU9NF_81vE`uu@4<Q3#`{PV~M$zd&Fe-}1Zc#I6w
zPd$8Hk-Qtke1XTmbbh|AoUr;66tWG{QcRKseE9Ra(w%al1)2G%&&4RDiLA4Ab4%*`
zL8}3{d-*8$RcHbu%(^4zN5%pRSE#^tk)oSk<{Bg0H~e=Ap=VzPZZD);9tjkV^)K66
z-Y{>IjXPbZJaTw+b~06}p>gg>Y<Tvs#n(#M_wUNEO$&Xph{ATgmt@nWI~c}cuB`O0
zfB<y?f5wl*%nPI+kBsEDHp#dl4CIAO6MNyD6}MoHA)tFJ>qjh5l28!Hx-T=MtF`C+
z2t!`3y=3%00D*TCqrO19!#ni!OwFf+>*|*(9G9lmYu2|gJ*Ur?SE**vc8A-ow~o36
z(#`W2llNFCBp(w}qt|nK*8g5?g5?1va=$)ij1^Q`T55DaJ;NR;^HW`0u~9l8UKdk#
z-xaFnlwS-|)VIlP^?Bfo=kw(|UVtXNd8Y5~T~9-8!`#5Rk(ii;7`MP*!QoTpmbY#G
zG-|?v?dynOP_S+x+|2#5070KppTLTekRRfZgWJ5XA(&ga*%3-LPJkAm%O+jeAh$A7
z32J5~zb(R*Rp5I6fs)MWIdI!WCcmJ%R_8v~=jj+(d&H8|Hf#kS`IZ~`EP^e+GbMR$
zxtOBHcieb2wEl7wJ4GLoI<iwaReO>EBE}gON`8+!Xxp=uboS7L!yt*atS5@W?XUhi
zxn86V6x$b=KRv-ohlc}Hd~Z7N@R3qD;u1`)@bqYnGd3o?;Cs+{x7~FahO0v(%HbWw
z=%pA9VCvQR`F5(!$@IWrz*KbmP;UY!NmS)8FzhG8_Y%SwPdhZ3*6(zDKYQEvVBu+a
z%cFNwbHufl57Flvhd2!JJ|}1n@nUNzDNqIhG9PLftNVP~ga?*eUQeyhm^?iYy)STE
zB;5Qt(c_anD<`-a)aj*a``&<F8C+(Akbgt!o4iEZ{}I3Ij&fbd$;gbrznFD)OiJP!
zo^y!Pb(Z<EJdhQXHg@xN;rGAX?lWGy?UJh_Z_`>0kSvqCc3=J;bQCZa7kEhNp@cpZ
z!3Q=oZT(#oepdm><H9Njin^3~y0VBX4D<I@Q9>|5A*RHFnMRCn^cPRnN2V%91z$dx
zAmfF_?)(`k+-tDk-ODaaKd5snp`nvm5u3#J_`Q*7b@3ls$clP^FchmS*5hVGCxfl^
zgWMrW2JCbo&8T&1HoqVPjPjd{8;|1Kt(l3&i!2YUeFgBTlH<M|<qOr@wMjZ!{?CyX
z#;dpjkjD95$luP3OC}04HrtJ*#sluRrtPl8+L~5J1~J|gK_zvn29H=sjkiojO%xhe
z3By`2bn1|RXP>dyXm_62mawkkyd3LETcXv?#|3%C%my{lkp4q8;Najaujlk+XLnii
zY<((>517ikwvcMFBv&|&U0a<8iXN-H{PIw0m}Nl1b3hIhUzDiY*m7UjUTzd;luvUq
zy|38xZML9&iLU~#t3_|6o~E8>GQc0ld?76($FJNQj)av{b7ned+hfBbwbHMkvZ|H2
z=-_Z!Xl^FEW{pV~0s7z(;=3UtC7}7qhj{Ko!0)L8Ozc0ckZ>}oB}tq%b;j%Wj!<8J
z#;=%?<-X4W^J%+Sjdu@HyYKTZ|NTZuzfml<{J}b>uH8@icr11kGk5~`tI4HVb$9=N
z7(lykJ1xn7+Z_&(!tI<b4#~Y(tfA^khgg1hq>RaKH~;!LX0=sV@SBsUZr>&V|2~o#
zQ4*A;3AKpP{WbUJx00r=GU^kO^<z77s52E*44Vt#eFCIL2mIW>SosO^ga$3_S0y}t
zKRs5ntt;nbGXjW>voeozsbh^>DT~*lhd5<ijLe|1lQD;nBJf$L>-lr&^osL)%t*if
z6`Ypi0*N~^Rr;ALpUXH=Gy(YL<YSP4Hdidy@Ol!l-Ht8t;>Dz^G8+s8(5Bhc3-h=z
zaCfsiDRC#sFC$FQW%d`FPZ`dhKiSY2V9C%=TWO@sv`*dv4M(InJrB6p{ECH@MLuCu
zShrp2TegBkX#cos-_V`iJC|;QeA4Pawe790`XZU~w@y2g(luia7Nh~qvg<**eywf3
z)Nd0uxuxXam2{D_asZEDQArVVw)UT_PrrPDRT^hkF_bFz6BnX-$?zv8lk4DlAGnyl
zAoUw`qTYKxX_$E0f>&xk+E<%@Qz4+>t5wA=FfW}qj`a)GjG7!*yBOc0-lfrbtFer?
zQ>85FNy1)@lkz|Rv<e_BrOfk|Y_M^>^6BQ*(*znP$VXgrp=s0WtV~a$(t`@S79IIk
zOsb>nuW`>SNpA^s5CMJxV`G%lT)(d2m&hnedi~i*p+ji{)!p%LXE&m~GYS5i1G4Z<
zx3uTuvYO`X%Z*&dZ*;PGo<DbK$_%t%(6#O<Le}A7kkZ}CseVTUm6{ZdDaI`a$)c;G
zT5OHA*CBJ!k2|PBNJeIc$Xt%&(=Z$y@%ztETc-g>!O9x8VJI-w_A^QQvZEF}Pz(CU
zjOGa<NNOOS-<G$XG%)*g^$r}~Vd7T!$IO0mH!=I3#}(lI+d3;=u#vHgiddFKVJ~x8
zeKgT~KSN32FQ|FeauIlVCc`k%XZ}4%<#cD9-D_`W_mFCukM;MG;ked!ht#9i6GwLq
zAuK2?`eh*a^A&CQlZdLd&a|z4b(cz}^AF3ie0@KC1yht{J|3*nqaY#8ilnx@#jhIQ
zT*2CqWGB`d7^$N6{`y?B>YuiT>^%?jOPz7}hChK5r{Hj>#8;}J`!t+!<Jk{A;eY^f
z#p{IO2@CI<*A+h+ABbN~J+y3=CCMd2l4VnLUSV4UQLK<r*1Iu*edi6tX>`YQQ_YV(
z-Q9nQM>uI<?N4Oks{5TT-jK=lnyJuMjDr-upd&zs9%`J-qstxqEB2j<aJFNrvqM#f
z^|ZAl%dhKNptQ7gcd<QC4q_M#F!gD3gGnuIQGbuy>Kk&c5yF)caDb4i(FdIFK0qu9
zkIUUQ-BxI&8Y1e4qGDq=x^jyRmx98%K6}NH_8ORaSKazH$>cY!w)>09Y`IxDD!+DN
z&Z-EUj^gHsjEP`*y`gwA?p8QC8#Rydwd5&Nlq0EB&PCVL8@?lpX91%E(xq{ylCv8W
zDfllWSBu1W5_|e_=<(*(#?i?K4S%PXW7%5uBpN7(3Lp7wXPMaVrt)&N7LO>as-M4?
z(`)AKZJ$mk^ooomVW>`4T%<~=6j}y_MZ)lG+Rv)T@z0Cuk~(gIOjJaSP!SVR6EjxT
zRJatp<Jciw-IfI0fEj-I%mNX4%rUA_;N`u$d~{_+WqIz3P{9v$_6VG2+D<b{h@tm4
zfv%SpQjwz)$YJblS05%LEu6(HtB0_-87|6`CMoRIrB4<!gjlSoY8dtHS`%W4f5uEk
zuJ*N1>-n}R-_zZFa^Mbe;Z*3%F9IPe<1xTF_S*#|TMYzhsK-T|Q^%IfQy;devo?U^
z<6Ue44V>;9w|}}VkI?U27Xy?fG<`08&i6x%0oPvXcB(Op1T3ZuAkE$(@%_>#^W3<N
zk}Xd^|H|j-;o}D{E;`Bbkop%89@*Xr!39oCg+~9F!2fCUpSoFzN9S#eU5F$rRMF2w
zx654-jpsb_rKRV3#ZMZ0VixqJxP!Z$O|d=9_lnnIcVf}6$4J41jQaU(K_6Z~*=w?)
zi*TPBMtmJgL?<U?ySnoK!E5myC(8PD^L*_~Uo^}V+u}|vD5#Tblo9-O=MxjO%H3hV
z>*__$mw5?L+45FhyOaP?l<-e_u;TY3+O(R!K?yh*v@#dPoAv8HdDBw<8oS>gaw`x$
z(-ULUqrTqCpi>~%t>J&ZcYXrIk}v{sM_c_v>?@g6NZ%v$J2KYyIM7n5$XLu~P<8Ik
zzAt56hHOJcfCRZ;dCO;NGttT$mH#B325OsMR1|H>Siq{2Ri?GA++uTJ7_%{xv=t}p
zMbqo6xpS@7&-v0sU2JPaq?m<ZCLu(#EtV+gY~v31srxTe?Q+Y{hA?2U5f+tN^JK>M
zf$DrIt|kBQ=){MBhAn3XhT#^kUcJWnICETq$Ef<xpJKOSoLfn?=NFN@tjv3pMyULe
zr@{#p7bGfQ=k&euemvOOIX-{3H-#kt)Baj!cAswekfq!Tzw+??iBpR`JYpXl+TYJR
zzFmL|xbC4|m4z7S-?jvhKS-U0VfC%7t{>Mzj5~@99_Xec4aEXY&4bos@7*^D%|`!I
zcx4E2@-CK72EAGvk&z=b8%~cPGwvB)$E#B@kfQ`ixPJT6^f7H;%`1k~I*gwan^qs%
zB(;i==0*C0C69f4DSwTNiW~DHOrn2tUAp)a8B1>r{S4+7s|6<dbUKFT{+N*+-gqHU
z+CWr8sfH%Tk8KHqN1$jb`Im9ul%Ge^I?0&-Iu1`yPp9myN5n-%Ms~MV{WOtjBI;jN
zXPeJY87g|jf_1{25hgcWDErsc@{=b!u^nqQ1|x6=e$af?Pz7w8*v<2F?#rRln@e(O
zt=EE0g9>q3K&2x}X-n#r$4raTdZ}?k^ySPjlFkc3j$oLwG5oI=>21-PB>0p1J@vw<
z=HFNBbvK`5e`@jnhCM+I=%W76KvBX2aHElLG|o%IbhOXarQ_KjUT-f5H5bk?6J<}H
zI=DmMd$^%v=<@28j;?`>E(|6p!}=*aBLCtwo*8<j2Xt25X4xwKynu9TB2O7hlWen$
z5!D15y(6=|sRdGK7EIvlq|>5xe2p4iIi&`G--QyReI1HCl>+?NmH$MM9Owj>TZ-=r
zvDLTw)m7}k5_8&3Jb&6B$F*SJ?14GqRaq7sT+TX)AYED-%cnC*JJ}OH!-rO^;K`T4
zO~?vG-hx1?7#qC}MBo-UyjuNE+>H;VHRqdCsmn^CHT2DtFZF=BJ%F#r_+6z_o9**{
zu&cFFtUSp8O;x_$41ypL#n(E37$xaz>!;+^uM4p%ckur~6MN^Vgf=fKbvH<SoD}rz
z?~8G4L;o5dlzo4H6=&0g&?0xOQE4CLGf^dvEqqssg8=a#6K$auQxE?=0?u%|Zqc!k
z>jw51YB$?g-85X`B}8q;q4e!kEOBpLmqO9(7RZbsMxks9<dX3?bC4yg`rB}=y87lF
z&nKXu_YDk7!on&-7V+2p{%H3NoE#lJ>#MyK$-A7sJz_XhSSzSi$Rv9)J}sq$k<oYQ
zJ@{*NHW0qvN}5((=z4>_@Vv%FxVNIHgY7%3ynLgg`U%y2dsV2>|1DSyVgPml39;7a
zzeaSx#;uj~U{GOGp7Iux)B*0F<Zh*1@H@NMI8x~&rRQpfA81@C#!5H-j%)_Iv?~h<
zuZRH$3yY_xr<!Mrnmw_x4F=)7t`4v5UaDRC11uhV^iAX80{!+N*!j0^NKfT$YF1Vm
zAXTbK<n?laH-myr?b>`v=CeYBTLxvaO-E0i2Qh`Y!m_J5Rbum!;{!i^y<7PW-WaI1
zonog}9K55uKIE9UDC&>)-@n?6R0<1tpw!y<O_a&trax(K2J^HzAkcLXgtq&}$iXZm
zt0D`^ud8J<bafW&Z43%+<;IoM(Cc#dsf05J>m~?Z*rEc^c_maXY0Q&ER<Hucf$syv
zJ~@8-w8)hW4&n!{sYsz4UuTKPx^Nlcxf}h9XoZdN62<1GNLQg!Bi3Fqm=;aIj^pA`
zR!(B+6_pNM^4UMEf3z=2U28I$4C%teQV$-xhe1GG%zjzb*sCIHR|tWyHm2fG<)o__
z1>1I~_b2Pw^ZM>x5`w`Zrtf@KPu*cy{6wd{qPs`u1-LKVQI+vVtiMcxU7<pV3VuPP
zWreiV+1f=iVulLNOL)}#6iU6=p?6e*3WRT?MsdZrzxzek`=&aBGAeWaE#n>pcXsw{
zHxN4MUVCcTZ>%ndbO5Qy*!hyu9=DI3IJsq%MA8pTHPyev2s6T(;1&=6oo&ovyCXit
zoMJZs)#jxftIh<*%3bV6^i6r!G-QJ<6_c<_P+mRm#Hn3&<zKkAy{0CaLZkMM_g2nB
zunKMc4Qs-TT`Sc-xwC<!XrAP6OQ?B<x>F%ytC==7qT*uFQ&V}>jXXyuyG0$i8si|i
zC{;&s=O`%F&ouQc`9>o&`(d|x$~K}T_%i_S3U$oN8LbWmDXfRu6>wbBN3aaRB2w5B
zk*lk#GS{kR{C+B3)LIDmFSj`Oi`a`)!-cUotgE+F(%vf5hU|KU{*IKc-u@s_+8s03
zR-Y5(U|N~;<nQm9wy4zFgaQtvVPREd3F`%kf0WYwsa=pOn5oR{QkvN(O&b{ulpzhc
zx2ll0-G7>{c@}W%<@x^kE}#}09Y#lgMaulZ<sy^P6YNDu{7}so5XgXb*kM=s{&_GY
z7kv<*@}#xRA0wy+>$Z17(9_5c5UDLU2z@F@P2=XbfLz8Dv}>;0m%Q<R0V;ZSQ+fV(
zsloARQc1(mxb368sC?>Y?hH>4&jaL13Truf5Hss9YOCfT4-)3yRnl`#uc=T~2M0If
zLGA9PVgrwPEzM`Cu6FM7WQ9L_PuRmkw1O$8zyF9aHiz$S4Fv~nf_WjCb5~OGZQJJ~
zoDdsxEp)W23hDWs!b-+3?O-gA?tf0GZsbyY&<qaH7v=ao?lT0nPgZ_WTn)q{$9P3M
zR7+cKT6_{?&SY_GgU}`#j8-nm-OWy7-yK)DOQlgOuG<Fjzy(}6pr+rwl$wjHbC}KR
zu8>1dk9MfBcuS0s31hKrJQUfYOCh!R6T|onQ0y5c!ivox*n(Th-Rj7aXPy=$w<LuT
z6NMeIjkv!TMfm;*yTXM70GZ~o7j1qUrY4?WR%SXCeY<Ss)v;BzOXl?E;~RCf0|QlD
zj@&dVvG0cesz&Y@$kS<U!RTCJT1FynAOV?6OKt_X3!nz~xR|#UzTONo@+^Fe$g=#d
zFyBWrTU84|6}DE{1ev-O=5F;tdPP<{mqWzI{W@W$V&PVBah2*~K;O`s5;alhUTHEA
z4^O5F`hhkpj@S+2?li{sjX2eJYX0m(p^ZZ2=0nR?%CUYbwE&m?TJ*_3x;b3{$X|_o
zgz+>D-(OvB(#kggl`rvF)aW5QR6F(QIY9E&EMOA0@B$5wOKzrrv%rvt;vEtk6!<%*
zV7TulUVYEOVic%dI?lSPGot@0_DujZfm_7e+q7F^J4gdFmma27ammxLBI|WzA!H11
zFFEG>&mO_RottcMYsO*!A(LQhMd6HL+;@WD&zjH-_ttqnS?+sT0cH7@Kl4vW)biJH
z!$2X-dTB~UQ0LihN=13n$KHI?>YlTc2jGK$hq&>nY8T**6J10o^I5lE2#iLqjP8hd
zGPLkq1hT-zRkdkPhdclGRF|cyp3pA=V##FjPdt`=d8ITGvJLf1gbCjr=r9>?5jNF5
zzlO3fy=_!b=F&bCYTpxcy0=E$CAa#GumsGjMj$YeUzCd3KnNKsUPAL^Q|NJEs|?+M
z8MtUB5NIJ|OPi##Zo{vXGG?d*>gebol6p0158)^5K?1Ty=93>kbw?(F`bPSE0zhz(
z<Lq}OFSTkK-Nh$W2o9*LvwJAJd^~dD)SVw_LvokLgjxkKyhdw;L~Lm3ZahWiuZjsR
z6}O5|o}IO2aZ@5OGOJs5cMg@)2-qn4d%o*{2WiBH7V4iFPHomOX8LT3(goOIq)U5Z
zj!wsSmu=p}CnAHbwIJi}gw9)RGcvRYV+obztUZ@>WX|@oUx6{d(y@Yws<i(<3%SQS
z4Ei^(xw}>g*JB)t>gJps$FMw!*M_y|t}T#Sb_v*9PlQ|poW(s$EKIdtRduDC{oEI6
zdp|!^@o1`|mJU+QqP3y?vkQ(ucZDGwR+<fE_oIBolHd2<$?l4f0bEqQeCuu@$6Wb6
zDwlGCle1PozwP<Z?RQLQQPqJYfUEuv=n7oKm<8<2efAt6*L=cM=Y4%d6_)phl7{Ad
z9fB-|aS31=k|08DHK)3{llGQbs;;+_p{rl(&Fz2yik6~W#-Z|`_a%iqFpk>~e7m5p
zw7C)Ctx*#6IBG+d?|amK6e>(hOW4`*wb{>>NCx!y+JKlnCiL{f^K5jaC~$SRrqwf@
z81?CQVk*-U$AN*`*T1hs8+66;&<sZ08U)e#!}c}ng6Uy8f(8E$H~wKtfw<U?vqj2z
zhksINirw`!fjTWvFNi-2uO(3C7(@Q;-}Qz#5_@z_4SjI1=~z`C+PO9VayD`L`z&NI
z>ILL#-_T<NEp1!lj?RZ08-M_6&iOYa>mCOO?c+?#i9pR69C??xeV>)KGW@YH+fowl
z+ZDEwi-Iqjq^h?Qr08-LQ9@5)V|G5oMy~cEF;fu<35CkOKDr}CeW}G$1X`&-Po-Kd
zCPvpHgqRYMU!+?@hGnMO51>y{Ks*yab3kV1SdlYXD&iu_|L1uOlU^iZiVZ}kS?izO
z8YKmIFhi3T)Do34(1?s_b<3{Zvz^_wZAxiR>{-y;2Q9rI8U*>(2wjUeJsua+QEXam
zL$7!JF(Y`p%6&k#$pl{*orf4VFDVJr8MQIYSB5sXF(I}2@V-;N^CKbRoXxQ3R`+Xo
zeoETVBwh(uNb0(9@UY6QgmSl<{%tEikpZC5&DgWQc6cm))IgZb{N10lf2Xr>%&t#0
z1tQ9limc!dJm`Kz+1Q}_iMd-U^K*KCDsO5(okH0z4sM)dp@Ct=bqq%(ny4EQ15zib
z<g0Bs(|bZrr9kJdqPcajSa^8&o!vtsbrY!uWfkXb*S=Bj>GmP+E#B39Im*nb4rppl
z+LF6FudL?Wwk;()wY51+GRm$7h2MI_FeXuHCU>{mZRdqs3F2Rz0z5MAN690?`N{6~
zV&_n#*a}vo?fCe3`GesXRGOn#XF`F&`b?DlzKe<^bi$6c8T8ORDk9>)g_+A`Nl2T&
z29jPfPnYOT7MtdxCnq~uCl&D94*_r1n}Yme_^#H-0vcI#+Wm9_t?VibXehAQlvxlv
zWB8gW7*FHm1aSu`ixz2Q3C!C0x^vOcN|-R#_!+RLY<Za$<V9}kPguPh%BnuzereOy
zK*ViF{Uuyh^Eg-%^DFFoh3Fvdp+L&NBJbvHAMaDB%EcLu>XD8+qt#Ug&clGNUD{29
zhele|i3v2{pBq8WU)KOr$>$9hq4Vk*b5SfNQ=z1Ex+o=qKifoG>b#NI)KIicbx7fA
zI)z^|?qGqmsA};vAU^Q?r=ipx^^3nw7jc-W`VT%$ggW=Cq@fqrWlL8I812`U*(@#&
z5Z1@5r8fc{Mh!^#9PV}Vk-YnFHVod}L5W;zNFq7&8Vt9>Xr2`4Wh8Vk98637tXGex
z(E}r8-=zew)6vH<D>QBLV`M(onnV`Hq?XVm8Kx$~`@B6U_4^G>;DMarD^uBxu`q!3
zDc=(+u{>;Y)Qd(UnO;PMA>*Mw`E6Czt@}-$;&e_d3!$1NrBl*9K(|Iu+bMePW9&WA
z50s%c-%!jZ?vy$W^OQV)Mx@>d4i;i;5Fo82(I#@E0C~+4-*NS(4v9LnVCP6~<n}E|
zA^P~t7<%bW&+UqRFLk_Ow9WuZ49GIDMyR(D*n)8?g4tL>PDBgAm0QC?WBGG17h+P7
z>7ZL`?>J<qPEq=H$*N$~&0yz$f{Np9Cc&nWy*f1SO)Ru_2;SZzdE?6)8)r@>T$vM5
zM8b_kqtU!$RfNR&CA612zflwTRY0p>z3)`sd%uQ(2}`Kqop)1Y2WqJg*ntvj{m6vL
z#73X_{Y=zkG;zwevmv-7JEPOr;6gcdKu_D`r4Ycj>Y?#xmGl-Fc%~&&GK{k^KYEHw
zgZ7Oiysg#Lf(EtfPh8sxxmCI$K-rMmZd#gq<xTHAPaaD<RW_uDwAeK}c+XF%zsNFo
zUKt~E21|XKb!eDcRQ?B)6{%|q{j~ftz5U#t0|awq&lp!(m7pG@(_wvnUl1<$<h7pq
z5o~Wx-e=b_F`5l&cXAH+^EBJgd!&=`nh}-#{2t>~=lCQp_>?7qeGcKtQi+%{p&dqk
zhKE*oO}q^0?Md3CpYOXl_&k!C^%^8(U?`lr8mX43UCg`E?zh0ZqS~kPQGbgpcDTra
zouBD?lQ6=YRgCdw<LST7I!(92)s^p1f7)M2;bk(l>{wkThqgM9CMSj@l{uL5{c4ik
z3IsMNRAb#1X>pNsIN=rADQSq5flc^Og%ey*IB@Q#y0^ha4)t~q<J&z7Iu#AF<>KPU
z`BEKlafc^>B&|kVjDcZRFvl8AqL9w5S*R?{KS&GJ=mNHI4}$ZsDaeDbAvk!nuP;h`
z+P$*jvKB5+(qqAJ&FizGkveZ3pIV3-G5nw%WEz~fJnJ+4C>R^-{w57fGiFOn)6KG0
zhThyTQPT=gJGp~7Xy(1tH0j=z^jwgg;Ui>7h`AcjGNr55R=Z^{+QGw-Br|MTU!X#N
zPx<DyL;5<_{t>nCA&+D7uvh0sY%eOI9O2D`<1yQ===PJ{Re>|i`Jraq*>W{i&T=z_
zY%HKH<vLT6S{=+9TW|dv?2Ih=PwEqhPpD7f;aESuWqpX?H90qU?y{hpp>HDGqE0yu
z(*yc$cbJ&tS<x5e1o886mDk}8#=37w233o|>0x)uTz{UM%KW_j)Mc7tIGk<03xiaa
zUNsk`hlL<BgzMX8W~P6A#jPzYP?-eewt8YWbilL3G!GYW?$gQ^EDL=GQArHQ-yl{I
zR74zWqTM)xHTM>4?u-}-gX-;k)j=eHN^4&PZ{|cuw5NA*mFT~o*u{4AF9+>~*TWvI
zT+o;7|ILp;WEpZ?6V$nx(s;8z>UNz~eKd7k#3s3A`Cp;&*+KvDz5sREOV?diR^D<T
zk72LuR3RNnw!Ze!pe0Z2Ea~Ow>edDQ*PbP5VSkBXe~*K1LC0|9HxTWE0CAvY#>BU<
zdcfJeaiBNX-sOTk*Q>c;F_l#Z14ToYX1L(qoVl^5PZT5f4=e{K`e{L1RXt;{k;*Cy
zL}Or$D^-V^=64lz!6#B&Oa!XK{60XJj_~u(Y%zBHPT8XDpMB!+Zu$*QzfL{zwXH3u
zf>q!g1DyX`yR8}ZTkEX%+k0RZnTKJfr%ROPcR?MJd>k<3pLe@|KS{3gjM5s7bT&m!
zOm;@^@Y8@0!bZlQYnLpI3vYK9-=$tc)IERprKuhVTjMQ|Y%g+6JJr#`Xe~yAdMog+
zpYk)XX#LheZj%?Jm8#YgzL0;@w_c;?1B0XKT8uc(ysG#|g+UE_Fd(8bEID_bufP9;
zxAH3?6*-}N>>G}Mu<Nzr%-1qyqiI2+j(wR4GK!-oxpxcAnZSR%D(LqxN?h>}_QtXf
zAu1bspuyIAlF;SqNROPmsB<|{*~)q)k|#SYlgRW$mfl1t-Z{U>8b6ZEpu<OFWDXz9
z%!MqL<Y_64_a=h)fj|7Mq4HV`9%Eiz-5=skPnLHNLpwL>Z8Rq)qb#PF7Q~W04N$d9
z>;eM7A@T0{zxlJHjm6GowUm2PD=Q#wutbiIt!kn+3k{a~hX-kw&lC+w^1SVgitt@0
zjw@Mw0uO2-nAzEX<pgnEk|%eiUpR-P(ES{n>9yONNsE1?EiCe3D|U_jRUdFjUGF9N
z7Fk*5x5d<22)^@Ep*Ney*AO8100On(S4Xg=*eAEQh1oyIJUQ5GFMcK*7j=Yvmz|0F
z53t9{ss6&`t&V@dM-u!9p-U$!$crGIFMa$5H+}7?bn>UJ1$X7;E-U?>r1mR%{N{Gl
zq#2W~^_%KyC+BP=g_f5G;ArT~pn5!{ZoP&ykr>du6rTe*t+DezCFV)viuw5|lzGyl
z@8zdx!^)h6U(E7kTko5}nG(e-o{HnZVhxk1%Kmzky;Ar-Ih4iHVsu>Ln$c7qx$$bH
zNpOkrDB;+bzEqz7B7(*ANAY*_FH{SVt$JE4Z(b)Fk`+pt<OU1Jza|t0hc%LMWSB>l
zCx@^ozwWe#1Fsm37TG@I$vqHfAiu{)l#>~I@Ie1&sK7>A5&d_ak1Pc=Do9|9)rSm0
zD35F>m1@k{OU=vxC#M4M?!I{=&%w}lr*H+5C=1zsdUzRgc@-oMA|Dt0e}9{PPG&zX
zE=BybTld`lF6AXU&r{sZnag=YXxZj6DC@QP%6JgAvsh|%{E({dZvRi>#XSsyvy5Z>
zpZ;-jQS{i@RrVqQ{gT(Zi}Q|8Kdi6U#A;&e_UACC`Bg%B=DCbN@MEu98>T{!Psjh8
zY~DOQT^2U<>E@rj+WB$pA*(hL5^>@%OQIRegsVLq8KDzugk;K0El&IR?sA7g<b_tu
zjmfNj%-~zLLRH&`#W$x2d$AYyu@q|Xenk|2yTA|cJRk_jU77sDD=(Kg%?-s%%pm=B
zjk_gQpIj5LrLiKmk*RHMRfg9&LB=^oog7FJL$$JK;hcQVKCPZ7Q6au{4B9H!ZqBwX
zuC0OD-31$nSdG}@qczMtIm%i|r*VVucG4l8)YTjSDjwZlKBEq~KlRr<|Cfg&sr)C{
zr^UWG^9KmzW0MJLOS2t{`Hc6jm!7%|LPC<`9XB6MDu8ZT%Ui9}XC~al7g30&18#*-
zrxH<}Z*xtrJ90b=XXzk>(DhdPk<Y~=*Nhfc8jzW|(jlyNah04_h>dG(EU95?u5;fM
zQQ#)UDb%e2xHvgclbNQ)EIQ`mm56~rzKynZfcKghj%C4|32sd%e8&i8WL#c9N%(RA
zpgtakgc}$UPGLVmd@u*<)U0i&_o8zA5*-q9RXa078J=3=;Lt{ni?!*AI}3|-Al~PJ
zphps2wg?=9!%zN9b_waX(>W{aD9bNP>~T4Nj`3E!z{p2@MX)1>*{3hZtiNK@zqAbq
z<FfJ&`^+;Mh~3K%^CK@kJy7KEt?#_p49hMm_>qz3N%3PtNEF{L2{IeN`o>y+g`)4>
zu`pFxE|!O~8KgoOw`)G>2cm=-bgR;2E?ufKiE-e#T@Laene`Bs7yeucfi{M$|5+h@
zC~Cf7zD=F)=R|3||0tb$Vg>64ymA73FI<^r<Kn*2AK)w%d`R`An1D)w^O(0qZ49?e
zniYh<LUof)+q{|D?=1P#uxng8>PO!KlOboTq+PK}?NBxl6~S#kMaJ4=Vt32T|Bt3~
z@n`z||Nk6f%4tZ*aYiXQ7cz$==R=_bBUBV}$k`lTNSTtul5<Wu<S?h&hUF06shnH3
zGMglp)*N#_{a&Bn?e_f>%yqr4>v}vNkNe~E9(}fVM<_#RBX?Eng`-w(Q59C=y}SF-
z9_T~id%>#kcE7s*O{S04=5)w40Yo%foK9iivn!lOOGg*ykdCy3g|9X>eG__qD)kh=
z)s!v5xfy)GDPM+AXn!%G_;AvY=d~i!&02Fq_0!9irJ^q;lcNIU8)t9PiH#0ZHyf#P
zz;2X!q>9p<88nuB>$S^ksSE)*ZBdJ4)YlGzYd!|T_5G8bJJ;a(JF>BS$|3!XaBdC%
zMLfm$bsLV*4pv6IKI&>fSPKUi57pG2^;>@YjH^E$|Ek6>6qBt~Qq?cCC_80btX2(E
zOR~j&gIn^OL1C%L?jXC_QlqHkx>}(Daj#>tPtxSy1`q$#eSY+b93W?*{fFDATYY9{
z#l!f+l%!JLmnf@T$7IPM<@2RkbxcM&xPNFmFKTS{9lUy*F<@6(2)A_q71jgyMwO=(
zVu5<%dJq+6k_pDNmGEk<fQ+^Fnv4IdSQhV^^Sis-RO^MxY?`GaF*0uxf0|?C)#)3w
z7YQgfm9@RLN~Bf;xvYWkI6C@j>zH%i$5a<?7xN!-aMQXJO-%(+MR+So%3R}u21DzU
zr}G!Cn#3OzXMeCRAO?+o;g$q1FBw+)i5+lW`9i93#Yzj}2pq>k!4iA!rV0h3;0D9>
z(<2uGtDT*aZ?~#biaYrw<JI^d9MfB{ugV?qs1Ne{VlvQ#u<~v4;<#lgBCo3Rk@w_n
z&;R;$#Z#T}ZBpFubTJRvpfH4d^7|p;QlMbiR|@*8nmTp6ef&c3GiY#cu!7jFWMNsi
z9d0wh%XZ$@)nxECvQG{IwS&Fq6!DH%ULvx#A2AnVNqoMX`w~Nwr=q1_=RVNyP_UR}
zJG3)l3V>Dnr)TnaRG`cNASybJK{zeKm~uA)pQzN5_}Es`V5pZ5^aMY*#`vi01n=Bq
zm<;@J%>8eT&UOZ^eAB+4;{kIi!m0R<1#lVQ*?Q8ZW{$&L9d$mqNQguJ;)1$r>KBQw
zhc01{D0KIbVu_;)35h&DT7fL~^f{S|chhY^Z?N)8u&$FfTw1IEhULQOGanpeWH0&_
z<hu;rFQ%V?%_7c93&XsoGA^RDNei@aln0fy_rN9Rk<%%%&X=R5M6WW^y#AEw_(8W<
z`z5Q?EAp>(7kew8nF!m%{5oWOf~EJ+bQTyE2kEMrJ8@HX;hr3+8r>TK2J3hk3qZmw
zb1L(`QG5x#zBZiwRG1mNvDxHiUqo)==f0qEI)Rf%6UNmeY%a8*qg4qW$`ctKkuoLk
zdpO>w@C$g>3kxSmI9WT{zY={}<y*|8ZHzhNpG6+!60^!Be^-5s89;VF>$du^`)8+k
zk@S1W4sX%76?>~sWflaAx%fYP3~aJlcSGG98*TFt`QO*pCJN?|F6T}NwSVX3x;3={
zy_oDwjv{pd2Sz7SgIRVcJ|Z%gOh1$b;_19uXJ+ehBU?I5m&*%_3supFG1b+ebPu8V
zPtms9JKni%ysC}Hb&N;+4Ucor2ndS7;k844y5U5zmba`#9Zw<EwY7oAtmX!Os|v8D
zDeR3a>blIS+X%A{iEZMNd=A9-uYBJ!HZD4yD_sJa3i$6=)<W_*G^+Ws{QT_<U?7XZ
zM~$4^1JQ7Bx^?rjTgU?V_+2vwcI>W(T7e0B5d~z4_#zm*2sc1ap)U22LsnL2zY&5<
zs#Uu3`9D<L%+%<43mh7ojkQ;q>b?xCJq5?%UyW<tV5(|0=SN)hvDSZpO&s39Fm@v7
zgg+>UuJ}%)j{0+0efp>XUH?-yIXF@9jnEUlY1a($%pN#A$5E;q9Y#F5)+XL<*seqs
zlL!;1S|<W*50n=rh`VxTQ5an*5{bmxJi7049^I;W+*2MHta4k~J0yJbrbjyZEoFb3
zEY!8TrfwX!o0-g~0zuU^P|z2tO|rFCB8qy+h45R$zCW%%(r-?dY?X|HwGWbmzwng0
zU!)UUi8HMR9@Z*k9d{}Y<OsE1LfbRd9aLs#-)bx`&9+})c|Eor6h<P3;9xW;0!0)N
zxLV@U1t>k4Z28UyLvkkV=La3vd@F+ah@IquJV!`PHXY)*vO=lJTW<%p*4g2-+6e2^
zH${u;f|pUZts-+tCV#8MH9tpraY+2olKFYZ)F;Qc98~5cYa&sNsN*lsn8M#K#nGr#
z^;e>d4GKzkJVIL0ix7E8TEkoR-lye)LC48@fNjxubW#9U2cI^PRDm$HUXE0g46MI*
z*R6?|PuFh8(%1JiHfqFtWJlq5;8nhA)#NWDprHxmi`}#1{UduJ`Dl4zfr@5jc1}&v
zTw8?>v!_r0MVpwJ7P*0vE){5H7GfWo&m`Zp*bIa6JSsb`2^S!*MIxsn$Fx^LlN#_~
zkjyWunAJgCSME6*ZukeLe%XH@y7Ok^0*8syqB(MX<Ko5gDmZfFc?Hbu(<*X<ans|b
zD;=k64;nY!LKf#NeHPsLJ}LMGv6|O!R!4tsd&?Lqziv!fXOmZf{XHsp`hs^!boa^P
zx3YL*m7U#7i~%oqRhZ~xi{<AUV55Ql;jnDhU~?_yRRFEIf%@LX+v}O-bP>eFJ+A&g
zn6TAbA)2(njxS_goP{s=uX7<XcXod|<mRybSp90kQ4?;Fr1q48Czkf}3n9csT7-L{
zZ?V>KDkJ`2Z-crURy@}~@DdY6trYW0gqnQTVsKR7m;>_)lxrD#mT+4D+4tL46|VBh
z^6aZZ83hmOi~K4_PRJuYD-qe~vfjC0vVu1<40}SY-Q(yJRMzspw+*P{{1S^Y=!e0d
zqVta>K{v({QfeQ2G+x&hAcMrQ#jd^e?SE;1gsNv3!qBaxU)iqzI}UbZ=^1{yA7(vK
zbZic1@4OHs{UCAwofUxHujr?zC}V~5;hIw(_HxGWj%$V-tgL#(Z*NYwsQO7=w9IUQ
zy9)VSW8-AR(vzVY4&EPay6HNkk+kN;*uNzFhd}95Kb|ikU(~kC^k2zP_vjj4eK)wO
zQ+<$2*tnOW`e|4vL+ecpUB(16Sbr)a^qb+xQvUuy*R4K3v135?f48psm~(R`JUu=5
z7vqx0OUg~V#0XFm;b_$`VD2y?znk(p2*4X@TkS~XLXo4q4%_x}9mU-rp93#D^tk3i
zK-HX|Vz0KWk8(vrRfolm>E(0pj_vb=`1NrsrAEh+4D#FB3<fMecDFvsFHEE|0%;+r
z+>WbUdKE5J{1_l*D|6Kh>!?QmS}+xJ5ySQ9AxB`uMka^1nmFZu&kY<y&4wkIXzQy_
z)bQVFup8>(M9CP4HpVoDFS=Wh4Q7)wyB-`QVl?hLJ#Q9}yCaKvF2nI2HsSNc(+uf=
zylXj0r(9DQC~_<a5f)HeHvD?K>RkNMnOj2Bfqo5A3xJB%DJyGV3N?-zu~EouM&3?e
zz9Pdwow()n|DAckZ;D!Yl_L?l$35O*ZeN~HzTA@SyN&z(0xEA6NTm_j`jx#pml6o#
zFxx)cKCwCcI0cQ{i^4RZyA0d>7a8U@Js^`-MQMbb{K~`PF9(fKqJEeCg0I@KU6N4s
zCmJXoItzi1PSR0zB;UG#lDXRf5A=gdj641k*(@d#*qP}Hzz2Xn(7QR){$!)5r-9<O
zMTg8pj-)^Rm3jOt;o(`1ys9M2t9xjFA`K)>_~UKc<i|>3!hDKRx<XXJht-Db4T;d7
z?pZ2O`PWye1j=bdw$S3rgLoaS`v!@TK)_7E?Te9evM@8}Dqp$Cb!>*viL00Th^sWE
z@?zO1A0HHOC_cqtE~c?vR1V3>eJHv8V~Fltm8AMuEYP;d@xxBVH{s0u8xrz}%H(f$
z7H@#je$Fv-uoj-vO+fx+D76*Go(<?agTE?tMB}0pRd&HvCwD%gBarn4Wz&;b0R*x-
z*RB!0Yie^>QFKCzmf<W?`_5>J?+S99eVL{qn6?B0%g}dXLqm8OY^4LmWBbJI(hSJi
z7qC(>X`9{|uWb<NjSyMrI+*XTP7zh?3p|`CjFg7hi*fv*pi(4yOR`rJv_(wqlQc4f
z54Em6lQK&>ndX%9@P!ZM>V_*YH5>%aI+vr)+oq@<I}G-NV8ZKiSs^Hj){E>)D$g{{
z(baJ6oOJddPmMeBb1<VTlEtow&x4p`+#{Oa9%iWvW%S{T1cnoq$>luWE|_yfr(U1g
zyl|aLK~WbVcp{GCLs!tNbZf~r1^48UBUzcyRw1$|x89+&pniirK)}4o{B_*|lj_|V
zN1JH>TPv7uN_Xz4%hwWta|rGVnsgK8Tb>PDOSC`GOTR>eSeeT=!?P5@_Lg~&AH-Re
zer3+3iZA(sULNv3ZNu|dk!7oJ+r`GAK_t)T_jFu#8rCzpd(z^Cp0-?dZ&=X=rMqLi
zgQlzI@x$_XVF4cKSiogVat^*551hEK^KmlqzrSj?dzY^!O^fp-hyv7qb2~S5vDlQL
z9l|2YzNVsLg6Uf@r!pcb`O1{#_eVAgX_W2g__=>3E7Olg4A9oL3eg)B%Bappbo=9=
zgGCI+FqIYUqU$A(o!Ie+-v^rzjigH$9W2Irdk|GX6%bj`$-UCj7y}qA9_*?ETgITd
z<A~H(@2;XLq^aQ`SJsE`X7~%-oP<CXB|VL5q`hj}9h27qQyX?YjyY|%q_J>Z*E_cy
z?@+S7Ri^xrRQ{{{my!u%UB_TOORswc#RJ<HFW9=6i+56erdc>j#}S7%;ihDKL&w<w
z2JspRr~~zbtZc3_X&>t}?%h-2w3Z6i?Bx1OK9Ax5Y?*HD$D5@EQ?eb0E5ScA5JzcW
zjFHjK@6{2-))xFp$oj{CWu^Mzs4S)Nv}BdLkl%#_#%0Y5g3>)?&1oT1ub)L<mV-w}
zm^4RqD%gg1B;-UZlW_)~1swnb%%J^-0{B^n+{G(vzF}>xudS`I+4m;^^ufuFpYe5}
z6At}InK{RF^4GI|EXDXg4IGgxvvJ5(f#l$Rg}xwCTDWeC$p=*ziAPz<!1m_;{Ib|W
zf3yfZK%kdb$8#5B`>S?ll6QJ0wtfP1OgKLi4zY&?PN1xvh8beW8}~kbowRNQ@Wl3p
zfwc&Iqrp|LQ=xN@{Ky(Ke*=#|a1S75g;p=n@Eh?%HP^H+YIC#1pGc&C9iJ4-zrhW2
zsjzjjwp!<g;V7^dUNEy-oZu*wLnGl*9cau$2bozW!|eI7`4Xr`Vt5UX^FgFG^o~8;
zw;)6%x&TpGr1k9nJZ2dq4`B0&_r!F(t*x+kl*{BtzrA7)uvx6Oy2v#~tVS&=&zCpS
z6d_n)vZC3v0EP2A4G+8Tyq*?~=Xz?+y?w8~<~yUX?l7#m!Zbsl8Qr5|yuY+R-)r!Y
zBrE`-#Q9Usyt5SzT}fD7^Q)~COMq?!Huzb6Y05kj%4<?F7;<tJn2!2f(~hG1zix}6
zZ}edE`FKoyX@u>~g<hnbP5!ZWgoX}Xp_2*GJ_#4T5LHfiA+i~D1vP*i*EGpoH_u6B
zHW+zBL`_Nb2XbWYCJJ?PBU>9DySwyY3T3?&j8s<9pGAD4qk?FJyzg0p`Y*fBfYl(s
zJ|Ue0=%5nX{|NvuhzvBhte~IpxGPb1x-vUC<Rh6x?(H0Bsy80&jvQ>xuWsd*c-$-r
zDjW4jQSEJI%jO{G7U&#PUx2>EuV<RV=dz#1B)jr<lgpY6IP3B5*Y?mZ9>ci-bYW9_
zvXc=~mg##cbrc^NT;|FD*fP!X-duK{4E@!^IuvdQFy%`Hy)LBj5@3I2HdfzA#vvo6
zw^_b&E4YskNn{Md^-_Pi`L6|w!^>;3v&tn6L!bVx;!#i{Hb`Ho;*&=Q)k-bY*8W+I
zwZx+9{tdmGyb4RmoQ&&Rs*aWldy>8SjgRO`myxV-5{@#9e7Y0O+1(ERIyYDUp(FaF
zqX{P<pC%kp5K=HX1MdCVE!Myq`U+dcYFMS3Q#Rmx!@&01S}Y1&wC;hlt?=l`UHXpb
zc7M^A88H9AZ<0%MjY+uzJ_zTh8?qkBLsK5azk&8IW*j|8DY?x4ct7o;kuGLcC}m#a
zd44+IQ3Z{%^OKEtYbOc`I5zl8T8(OU12}=-hfsYGzL^^*X}{$ve|gP?nuZG6OjK(e
z0fFmS5s+ojoULP5dF|{f43G3^aLCAzGuBYh4Ir}6qyQTEZ?%feI%OjuK-L7Ia>ZKJ
zb}*$^p&G0(tHHbaE5S`#pybZ+3G@k{sdKA4XP$S4qYYIjN$cNk*MXnFop|8Up}a57
zhq+U2$%FHP4vV8MrjB#j5{c&(VFFLg)vUCVUMD?D$=4;hud||2>Lf)0?`6H##E1Dw
zV3h_xnFB6xJdu|_KUB^lAJd#E@q6|+<$Tzi-?f6mmgoB95~1=chSCnH|B)70j6X$M
ziWh_CWNX4QOkUSE2hhHaJy`-g--7c2LT~JO**RaS{%DNlpjZ~>;ul0J5@cgJo>KAl
zGS|lZ#urcBtlo;!GqPQcMH)hq*Pl#%y_NjY_G6y)a8I&lT_lm#zIzRchM0Jo`q;yN
z3MM?zlOYCpY_Qt{E`=9V&CF|7nj_%)xjhKk+b(fJ(hzQudqI#yeZc1Py)p_?TKKVO
zmG9L0v&>MHKYmyJ!7|FIBGEU{)ZoSS-QD%bgDK<K{n)rT_mRfa3Z!V_a<0Htxh_?N
zL`yW(^C#rbABgf7C!8fr!R?N<b{=6OH!iwuA&A<6p9}2yn3mF04Nvu~IPv(R;27*B
z{IL!{Od710&a1up24Jv8uX8%#>^#(1%(!^HcLz)VHp5~!rkAIeyR8y3v8AI(xsXp)
zbbMPOCQZT?bgnUjbBki5?xOv7a-52VL^XY>0kk}&$)R`IT9^>q*X~|TX0cI%(lVb?
zGtD1xVa(%=CYO!}2K{q<Uj&s*)I_(U!umTdu*(j6$|tA-s+NJ1<Bw)T+(9E~XXi^)
zop7KJ^dw>eti_CvGw5;;jG^1@#J_oc+U2Ewaj`6)3V4p!*EW1e#ydIbSVVSM==7fl
z5Pt5rRh@tDh&xTKwfaU{V9tq*LpU-;4mTfas}V4k?Ta%z-cW@jwzC5ot3D?uK&LX*
z2l>rROp(AncCuMF-=!-8ijL(MF@>lS+gITMJNxVF=U#AffcC<9!qu}hlv~39i#9vI
z+zbZr&>q9g+3kfU<JqDyMc1GqX+$Da<BCX@v!FIdy2-C^C$Z49nJdZY*U@>A#{+^L
zRC0@wB{h29zaM{#3hIbXZg<g3(6$m;hTh3hjRbtpl;<mmAUvlB!R|J5j#Q_)TSF-S
zj<$$5fsDT^A?^(*8hX^E<UK^egu9hfFRBkJa#7fdzy56dqww)6DcI!Lmy>j*(jBdS
zIA?;AzF=tvN#B&4Ta+{bGG$hAwvw<fh-s*;*O>cIhweCoqcl>cLr{0z@8Q3&laKO4
z0DIxvNyj|xWu-Otn?GWI#kWcBTXxZElb}k~^HKauab3(};ValQW3$itOz!DsoS)yB
z&N}0jIJtlLKb;%lxEC?jIu5h*)0&tFr`Iuy=J!_z$ziQKSb`yP@bB`n0yFmfOIMmw
zmdj$GAH7#l<#|)9Z3>@$#f;vIh^tUbz@-W+bAQS`MmHe6HE4g;d+#4>z!)6z94svE
z`?}nB`4{p%BSru21-K00wDf*Kg1dg_0e^J8VSXzfy-c_ozY&GZ=3OH7{b7XzZETxW
z6}uE{RkA}$h-%T5IFBznR@NLzfx^Ob69&AZM-pzEy!K<^@a|`Aa*&Eb1z1TSOS03*
z!=v-vEc{^LVB14ZPIk&rR;d5tDmdIy${>>mIvWPy@Qz5Y^M*x~*~I)RFdj=M&PM#V
zRGrcvxj6MgWrM+B+|1|~g^_kT7b~;1M!1EW>;CjkkH$=G#Qh53X^RQ_xEFE!3C8h>
zh85Pg_$#i)l)pgmuoL@lb^=f}P)n`MH`TlIb0#~v=GwS-ObjU=v?P<OM>>e_6Q$xz
z;env8HM=(Z?mn%_DSv}fQ+JjNo2rDW6N$IvOo_IR_7A{?`{OM(f7@)Cw72KeUkT{>
z@OO=iE)3!Qd7(r%ENpG~83u9?gGGxCId_#!frZ=nvx|=*=9bgI!efZw`)u!u{zHWY
zI)}J$6U_z+ir#&n${j`pBF1Ti2IAVWoweZDk8Ma~@O7uFLak?5)gvRd5y(MI*~Co`
z^+pGBN`Fy$w=Cp+&*3Xq3sFCNl(I1EoorN|K~!<eFH%Pi1aViOF-8DlZ~9RE19#en
zRz<JG&{U36E@uLuC+P9;%lu#Ajz}vNW;SBY$V}zKpws6RU!$CSP8G{KFf?am+;xT;
zfZDirT+$?Ran6u`TPDND9+qbE8#ByRd$iZS;x<Mb{ha;^{Be4aHd;rZwk``dpqL_2
z`Hd<(fni57=N~SxNNJ$c`!{xacT(mrx7^{>Bu?XG9syoU_a^ufVbC^e`~uU+ZYp*@
z0Xj`t$JO}-B-!daeO47obRTkUbmv%Jp0CM9I!@Mzb(@%LAKzNQhJGcC8_pv1H?H|F
z%HI5?YeP5dYiZ~M4+!q4fS4|vc;aBjBBN_(vpJmY?$0LcC_eC)Vq~3EQm6{31DS*t
zH|$%0z^72TIj+!<rwsuvxf6w?L%|8vj_HspSL@;`YztD_;rC_7l|WZI|Nd^Z!@;ft
z;BuW-7s+eyo9F6P8enlR;iSCzeoz>T-j7SZq1PR=Xoii7n!QAd=*O*3jH?Ro^gci%
z?HBF6hgFV*7ASVR26QmKzfnz2PMp4Y?_4eLc9>HXc3b1#jqryH^z*s`5Fana-QCrv
zwT7J~VCMv=#Ov8>3sc&`8`xX_;pl>@_C>$F2vRFbWp+Bu4^N1rUK4V9u(UdhTVtRu
zCQ~9JHQu<g>Qj2BO`tPDZ;9rZN->ChIrQ7c`J$^T%nJj|AUN^baS{7ABxooI1dfdl
zBgomg!yY#qi1_h2KZTDe(dhFsAY-4!+9K!Xde<;nU$&Q=Er%cRN&31|Mm&nu2ih}W
zT4wgU3)BC=wbuo{fTwnuUE6!WHY-xjNF!wZq&xEq>C&5fRjf{rT-NN$L|sOGuW*7V
z1orG4IkMw-*9_l1T=?0@MH~>MGTW!Mq(1nN`dio?m%(igxgyn|tE(HDp>ag&h@OeE
zC+Az9dr&oRBWa&Ynrelw&bkw8Cx+-;sY=K63Q&D5XQ-_IxT>}gSPKnhc|U*VK;_!G
z-05Z}bzpWaL~lw^()ZTITMAf2QMKdYH&J~pYPGc!?sjy}^7Zq>rbU3`2Q4jg#qf(g
zK0LxbW)eO>I!NE?-)+nW^_e>fLKhsR6yCK5n7RpCaOw#0oA64-It4ISTZ(ey=7Sok
z1GE-gxSOFUtoC8L|HFvu+M?yQeTo$0!%k3G*~JCuukPDgR;mepi5)u(Kka*hLe=`3
z@(FeAU|}7-(c<5-u_RC&nM{dc+u8ZksR7{2>)%osKQE%YI~mS7bCwy{+m8GR@fmB&
zdbZc1=)?E!%rW*v`(f?MG!;pc>v*U+945kdy!^g&{ROcLAPw**t|(Oa96O4r50xiU
z3J9CHK1B!$*Hct&nk;S<e~gTjsha<{H0tV+vliQV_^!_aFyyMMJfc+SLQO1QsKu?J
z&%dk^nI2sSt?jeuBi!Q3%0h7aw#-}p_%lQnINuXmExG}gfULC#FbB4COG-eckQ%U7
zJDrF<B>b2&iI<(D0bPt$HJg^4RsKIOK!(DQripl80w%a8GiH$w?isY#5=g|2SBYmL
zfU3Ugi~2Ik%4txMKz!g1p@HR9>|1bB399k5$>gU2_DB&Q&yzw;VIBCO+cUeYe2hN-
z6u4Bqqhk$r8&sZ3&*gKw(b0GNvOl|7H<H!^WrVXtMKgHxI5`eS5{UyYLmH6LfNmSu
zhnMZXC%R$#wL3;Hv>cqCY%cDG!L337J?YpmXSn%7IrZE`q-~$&1=XUH6?HEgciMdy
z6&;3p{w^>F*2qj|39$F&s2|Jj*-1wgAt9lv4mNm0ii?ZJ)pBSkPo$_E{OjGBI#}8V
zo8O`ire~I?y$XbMv~F+&GTLj}c*3aEP#(H^`{{07a|HK)-%$JvN@NH2;342(0$L~6
z8VHML_=GNuVJ?emALcc=eyBo|!q0Vm{ATzleJDmxkF$5Qg}<Jchf|MPq%@=*l-_nK
z5bsXySgbNf%-BQ~Hx<_Sfk3aApoC<u<C3KdF?F@=K}FbX?9Hk4pxhL1|BO&ItB`^a
zar;z9@p7o{vv|4V5}cXA!oZe{Oj=(y=J6rnlv35oX;p!$?%@8N@{bQ4W$#AckyU?}
zD1Qi(RVMy(kWaM6mHScIwZ)-fbx=OYr7~9|a61z;Mhy7@C&ui~<AvWaiW@E1?;)a7
z@!aC62%-L_pxq&ey!jwJ$FZ)XaPcZJaGWdO;f)v%SUaC+?$%kk36$4mU~Ll=4=t3|
zg7JR;9R}6gxOlGF!?UdU!R!N7sk5u3I9rA8NzU7&_A=$;etxo<$^DC4_F5_OUY{gA
zN3l^l&NsNNGM|&8x!dZF!*go#3~G7gA8^)h(=H6b%kkp+q1I$CbLH<*?zu*#K7sm5
zr^!7rD!e{{Lxcc1;9-3@O7Z7j$_9~DNpQCdZP)b*%4Kb#JJg6X^B<9lSOW<E=3ykT
zHSTKnkLVyIG#{KN#2XeL_x)fu2o#jSi-EWMwQy6h;U1XmLq2<M%f!-$`x1J)WA^77
z_+ewNLwq(ly<R@$6RuJ3-nj#NyDiRaLAbBA)(hkdw-A7E^E!LZm$vThlEV_na#S;^
zKkq!&tuM5ibx8h}xTMmbAWmnK?kjIS5xU<9KGy^6=wkwTEAtQFbL7~N>$g<(@z;pq
zhO&Yxm6#hFFJMlexD)#D8*%QbD$ROPBA;qBs%b5BN;nXw)ZII+;XqysiAKj3F)yAX
zH=#h;0?;Y3cW`i^wb!%CtJ?0EPg4FsaN~?qu&QjE;t>=|m$v?Td?+c)`jg-K)&P6N
z7y$G0B}PE>3oVO;e31{x5zwGJvzn_;Tw~9!;no9TW;a?iW|lSfHkZM)RBY77zXa}D
zKOgU5eJgtvo0~a!ylg^raY_1{g{KrdR3u=bETMQ^T+*{<5j_S_Z@b5^@Eg!?9DG-V
z58{j-rt*@ie(AhWO`dp*rx1qFy2o*mXOk>-mS1P_ww8&Oy*hg9PPXGpn4WD+^}Bz#
zb7#He;Uf0$SC<H0+}PUM1}>#@_e?k>U^stP9Brf5*T|z}y~;jCr;^MqR4ti5-U0Yh
z{WSkN!Z@;cuNs<?+NWs6W+%<`E}y|5q-E?D{h146lm5Dhg1LFM4T;Or9}?*j{jc>M
zTn=;1K>Q7&m^ca;9?0#AMQ^P{M3dMyx5bn%++C#ONbSiFlI4>kxs{rr(iH~$;Kz`6
zqq&x&dk1)A`LXCZSSYWiOJb7%u0TI*MMZBmw}LH+$5J_?d7%R0{FE{7&Lu6(<KpGM
zzDf%(sNT|$$so^%Ma7;kmd!9#O?}S)LRv9loq=LcgeBaYL%<<r4VaUC-O}9WZk(A;
z5dIil%<P^1EttSniyX2*m}Yg7@?FiwjsCE3XRnHW!iEs|x<v1<8|nk-KJXH4w~DlT
zgqf?cRpLuRe0J%<vv?q&l4Uf@+3n!q_W4(!YHNUQ=tx`a<`Rd-W5pVFI4O{8Nz4bD
z=QQjzUpv62xE6+3Bb$E^>RoB)>65z=EMR|Y;D5<{@+$Se3zm<wmF3M<55pHW?r;Sk
z;WqgRk)Am&aDqz-kWbF8>hDhxS}d+Af-Dj63wc!!%~l&AKfGv!o1-QMNh!2(Dx;vr
z@$-erp~eAo?m=t_-!Dy9c@v*uIP&tN>8c4C8SGMD8n=Y3*d6<TH&pfOZ{)3w_xJmI
zef^GB6=`d}E~&Z*|E!duTiup{j<o)MRl(Hyicp0-46Mop#>JCs#N@Tc1EqEz!|m2$
zNiVRXF!?;%{L;ULy&Xp_PrWGmMngvdE<86o>-)H0qu_BInm$kXz4fl1h-)z5FfS&$
z8plQN@9v`8a8f^Tinm1KW05aq?GrXbjDWZp(*4s~BUV)^#?s74p_C1WemP_0kke!6
zRTg-869sw*pd%qnZqO2TUPj0`cK3BH>~&Dk3~KTt$Aho6!09HJKEfH?yxyDYY*<ZU
z+15#!{}`ONM*q1-Ee)6&m?y0L9&#Ng=mv{^NOgpr(kV6Z`W<lnU*OwCBeC-zMUQ!l
z`3PD#MVcV&qt9!?v!w3h>4}$q2qz2k=pFVEY@z~$z?~L$!lGmTiWz+}=Bm@3JK`ed
zeO0b_HeT@Gx65C&HMtqQKW@*l10VViNehFtjqW57j?u3RUr*#*5W|nPo=fST34;&P
z1a}&8S!mR$0sP`#L>uV%Y*CMC{3DaMI4OQ1jFaase_3Ws!9m0QcdVy8rBXA6Nq;+^
zUs?ie|1z#_q`66mJl8mP<Z1^!$HlZ1(T9D*9TvIq{+o%=<qV%|{@T}l?!R;*x}G`n
zf^3&3n6eUl9@$~Aw;*Kfbs5W<*5ep5Y&BR<c}K;q*OqbN<zSAVH^YJgczrw-Oy!o1
z`%J<TuH>2gLfX~*Dhkb4{?x)o`Hv$9%@s-jNcUu(9@DgE>d}SGPND5A`4grviiY_Q
zubjxb@Bm!%$N6QQSeCvL{7(m-P}r8Qh)#YQ{7F)P64!Av2<vEum@YG|G*J*2x#<0)
zH>|=JUK-pO6SIZh-<@3_IanF!+Fj1265Y=}egcu_kvRmD5f+vi?V)o$Sc*_Im!$y~
zfn182!BK6ZusX1}yMx4otytz@!#Y^)x4DH)^C)08Mw?8kwvfM)L8(K)<g4$E4Sh&4
zE#=ZrQ_4trM3ebwsZw9q8yu!l8txHRhpU^WQrW=;*ggkiWG8T5W5zihK8Y7#bMi*q
z(v|zE%meAW^6`wy1M7rebKKdNfE679mQR$ela<PQJs|}+>_iqqRa)4)IP+7F;^d@G
zkE`}sxmTC1C(0t+bmQXMCe?_Hf0ySd+q3P+u6G^9jRV(6`se{{NPheVNzTz8q>icw
z1rg&yG0_F~{aed=#xDgI<v?H9+)U9JZv50>2a(}=TPje6|D%eg63<;Ax$RemoKg3Z
zn~k3xal#MBAsnc9a}9;$C}kI;lsb=};(h5|G(?JsR5ZOE;=S~%<I%aWL+r7-@eb;l
z$?Sx3D3mMlxI$J8<m^5COy^F>_dsHY%6Vn6Y;4ct!B!6N@HLBdAdTsDO)Uq)Aiax+
zrkmMpG)kR_|Gv3@cLTU^IUKW`#9_q;nS8fy2!WoI$5tRLgB|dBI}%eH9A<K!hjlG}
zxV3!L2oz1V#Ae_BaU1DOFL)%vL77ksV$HU{RA;p~bBEwN7g^yPM-H3G&52p?^H>B2
z9|}t@z3|A~w0q~s{=T`;@%)K05CCv?qtsD?Lq)m_QD9SGlMYnPb=LcPId}?%S}Q3y
z(n%m)!W|zTKrcXF)z&`u_m>^++Qo+6$aAw{`q=}t@$>)YaNoh+WUFu58w<6R7`f6O
zsB`GqR|zV|FbRwf-FK{4j}{2(J<C2iPlAqyArL-hB35QYik?O6*sujC#0hHx*&DL|
z2P>O!_Q$O(0h>)h10fv?JNC98C8lcwP8}-WpiHXFa&fkOZ&#apHPY_Uz*ODcpP$bZ
z!Vw0xcK<EJ<ttm6VkjsJP_9}YTl2IdBXg5&6$U5v269)8_j}b3mNq~!KbC&|PzqL)
z&onuLt6bjn7Vkd^zw4!tjjX}rn`a*eLlGj8V1686^mf!O`VBBB9X6kfgeO0It495H
zguQ=@8(^c`?7Yz%`?pJ9-}XOc@=HAYJ@l~Lw!h$I-Z!$s-3gFn;Xq2=<4YPLvWEYl
zL3UJJabvhY!1-ZnZ((n0pFPzF<d9Wl1hq@Sg=(LItp<a;H^$L*MJuCfe_6HX=JK$P
z3V_h>msZxp6X?3}c6+kU3#?f^ak+P}|KjEQJEnsUiwwK|=m1m{x5yIVjLpB|N3wDy
zpTp9^lKfZ^G4VzoBSb`W{P}v?`gTNgTnB>^2cnI$H0~A)Z3!LBZ>X!QAMExUcNUK<
z7!8qom;X)APw(|k>o9Lm4|f09xyhmN6lksSB|lbTW+yMOI1tIu0hPGe10&{q6gp~S
zKSg~+-J`gRQfnzzdhyD?<A$)_uN?ZJpChZTrbVbJghe`Ck1V_V;3{kAZFywVT!$;I
zBXXS5lMm@Y36xn5Vs*aOu2LS*cF~<LM`B*J@xqTVQz2QnwFJidRHbLb=WHhY0NVO_
zZ0C%dkjDD)CN?|7eM*fQ-Wap^`~$+QM}qP;oIQvXfOk%?3Fp|DyionY1WOIdM?X7K
zibe9X|KwpB9Nd~reZ}%x_D*rgCnXzeyzbM>855;}qmdptd#u%2rZEs7txpa37kvRL
zs*rfsM~dkf8p#y~J+$a^bkqglDV61u=%wjWu5hRS(h(8%g8YUQ-LpuGE)Fd&^n*p~
z*$%;*KoQozBS~r0G$kAJsHcO{zJNPR^P|&7oo_?Jk7!tgb3TmmRKHfvijIzHqtHik
z*Y;Smz~e6E$B=P?(4||&C#sty4qf-|n^`_Nh;YL{xBTp4{Xpoft~MZ|jNpLGJ$NML
ziQf3;Ul&j`mGqr%SZzB9`N%1#;aGgI%-nHK8kjZh>2afaN1aIhol+8?jQ~>x>NJof
zu(se+>=YfFl(ml1-JPaq%x!OPgQ}l4u6<2u(!P3zOrBYeLd+2d@?mo67N5M^KqYN4
z{B(bGu@Ah}J*N@y80p&IUatoB{tc+BPbg|uCts51Q;4emEh`1%;L&(;1j*gV5gA3M
zUr$SwgxOf9M2AG4;PH`{t@t@HP8XKGQebiN+qVG{&(i!n=!>A2PZ5I2l)01>B2A1#
zub_4=7>P`2`2-LstfcAkcFx^yd@$>V(%oSdO%bJYHHU%J*LC|_4q%WuerpHBEwZQf
zhk$*S+)6U_Qq7$wql~<~ycHe1E2*|t3N9@9g%Pu=kpb1S#3FFlZplV%L%yMKG;=+Z
z;jy<dkFr<2+nP1BbaDPT*3A_61o;~+58ZOsS-Grg2am=eUFfN3>zX29Z)N^}<L5@&
zwSF2<aVI(!*}1dS${tu9Sj}x5!4snkGWvVx^&C>Ui*T=`o<QgN$yNxt^pv(=ZP_!^
zmp3}r<MWthxgjZppduNiN<{z66-WJISui;hx4k<%%O0oKpN!QVUNrtEZ}K_yQi43Q
z$VS=fAwkc?!0SwX#@-gs)3qV|V^nhCd@CR$FngaxSDZ+XdD5eC4c6^`VetD|;UZyI
zI(30(6m2-vk5I{VM7^QZOEnL#{@dT%VX+Bn15JV%4qH1yY2B(Oll;R2>k&g~ZJP2?
z$u>VJ6D^?AQgz;F#>{Na?QnAx_pXrn%7tf6x~+=`S^3;!TzPs*URIcIwE_N%b3pA_
zfrr}m+Sc~$>K1#JK##y#cR%z{HZzsy?&*d~z$S+HC!sG#`JTa|z6zpG{A&It^&e|t
z__oODTa*q|)YrgQEsqd_+1ju5<+F>b>RA-?<?_O_0=zt8W<KQ%T8nB&3{@k8A|}l7
zoxX7UiNun-NrGn+hH`v4{QDYf3;5g<xYsWyXX7LOczQA9y2hP5VhTLBtgWD)l?{Hv
zkD6?GKZTKQlKjlRP9Se_E4E$>l|YnZ2VV0EGB&>zb3trm6U!FSN<fM9JF0S)DV|MR
z)^k`bs@7+!2xCG$*&XgV&)}1W0k*$jnWD3U9I?n^xao$>kZ>8lIa5al5DYD@f{DGH
zljOL*aoM{E>!czHr{Lqa|80Y<g8OT^xvc2_y5Tx}%1=~|h(^qK2Q^F~>$p<Gsnhf}
zn&&=yPs1!xwMTZ8er6P+YT7T-C!;V);!>6!w6oxzF0!60T3-nCn;`v3FN8kyg&{1J
zQb`8o9MNgqrcD-lt(O40>-CLNErGB@5P1$IguF<PdA;4#&71*mLl44;I!L0NN!6?3
zg&pH{;4(0;_Y)j~F=tXUk!~Emzl1OeAMhV8-BD9oxIY!cBbb_eNH~BSww^hCbZ=cl
zUjMO003!x)nW$C|DghJt9qv_H)QPV6xV@F#m6edrGlgwj<$%UWN(fTM+UkEhFg`vo
zTwSGP4y><64$j{AqA+H5SweXl6n(ceUesi3IgpE19~47-u7XwfoZV;vOtWbUrU+3T
zf|RauGy68_+Gs}$x|(*Wi21$OqS8odt`8Df-WWb&s!;y}>P6ATdr6ulj9tC<hJpiu
zSmAg(&ly0^KDH=S-m!FQkykNEGLW)P2&Z5BT@CKAV8_rmOG{tRKFRItk9Mt^^t)tc
zCbc%qQC6{2nbSBBe{;plG}n{gRM<Lw^leJNnW{chtL&r*)6tw_{8tbo`;faF@|5E>
z-;o}jP6!6V2eT}sOl*K7(Vc~XS`Uvv@VmD?0H}LRbr$O{>5Mq^lMyutFX?lZiot0K
zV^V+US7El?PDy=cVmewOzY`=9ZvQ+l7HcLpT`LZ<^3JqxF@`7|(PH6{Fq1dUZzRY<
z{)U8!oQcoQT4TgUO}3x77>BqfJajH`6ZdRy(IW=)8_L0?5xy!DQTbRG^38+oMqmS2
z3O0EUN+^Pm#O~DmFANmfr|J*I?)_Izr~Kz>p7LN9l&kKBr|H#;&ZbhV)Zo@2`pfWl
zbCW{aiM;X5qw70AU;MZriy2$rn+GSh%c+hGKf0^UY+fPhd7A(ooT9o={nciJ_3#Av
zC*+8JC7~hQg{SI@qvKULvN*l#0O*x-4tL3!jqneD+SKbdr}8wo5lsp#iI3ZwkpuR&
z*C-_ADwQ!rUbn*~M-hCzcueEO*Q~`bmc%<vYb^a;%%di#)Gu|@f6%n1h9T3Tg0+dl
zosPp!g@lHOo0b^UBo*GL94<tL^H6(lh>Hjrk$mY1hS%&m)C{{Y{p0rqKyZ4Lgd3oo
zz41lon3>Sx#4}6GK+}-HidVr@AQALZ?*|?9AM9luY%>q$yY|O{ouGr+fA*i<97ABk
z`rcMMH%a#(ojaSFlj^zJ8VX0a()X^h+=r||F54f1f<h>q2O|%+SLxx^ADBc&535v$
z<=r*BM(Zs&J{tW!h~5b>mdLBL|9O$9t7OqYREOFCHJ7*=KJNpO@A3XzO@2_^;(I_Y
z*$<01xchS>Wrj4f{GeDcG(K-d<qD@kb{KL{sDEbppT^CjJ#H<w`gT^fc*?m5Hx!*a
z0W#j3c*9XGDcQ<WA~)QHKx;oUwV|~yWDE_~lkTU<^;o=p0X!NqRpc~OO0pc(Lsq(n
zPt~o9|0rwl8+sQWVCSz(Kb~|~@J)|u8-J=g|LfjT_N$nnaQAEDow$MJ;Jm8a0q#Q|
zVjWST0sZ>m`+i^xc}?wMGA7I-ozu~?y<v|)*;YF;18L!ei70A{Jh1jIxX?=CxHYUa
zi4T`ilZ8NR0hQ)Aq>ST<?E;I08aUP<f1>k$J+innHpU)*s8nza3H}opH*cyntS>|r
z$qvkT8QJS=PwzdMnK*4C&#l+G&Twsg-)HrOo;#zc!3)v2=I&F)c(nmr;=A685;u{l
z+olk{z7>0(&Dn(dI>FO3K}WqgG!r|E71hc=erpMkNf%M9XxPa17UsMHfqdy+i;0w@
zB?T;AIo@Cb>_$&f85_pC)BR*nJIhvYRKH1|Je!@G_!~`P(`qLMviq%HMHJ2{+O(6R
zVOD01x=u2OMRh}iF%lQSK%M457?HA0asA1ixM`?>-n@zShJS{uRAX1?r&IODt8!@+
zV9;L3fki0{*tupS{-EB@Mm)z1GflwltP!-bgkSiI_#hmKt84H6`i~nd#<iXCDCshd
zvVIrn?{C0)hPdZ(05(5?kBu#meaFkmr*IqEgV0e@sf5^selu_*UeY>LzVljAOFqMw
z=P5!&!Luy7_>$xG#(+qKKr4>cuGRrim+#7hrG${xr9j)i!ebpSC(O*#61v4)-s%Rp
z*MaclBQVWSv%j=!sF^;pmwogEkwCOoeUxDKLCZUvm8{SAR9;x`u|~}t%Q<;~#gf{T
z)MTR$>Wts*Y(D7pEO~-N%JTgR)^mx1M68ZImsY)nyNY>QL7<LJ=)B3sA^_Rh8$QKQ
zlt1pQ`5jo=4Yk}v#Lm*b;UCZh)~KG_?HIX<?u_4FS()9MWyf}?73K%T(EaN=xV&0N
zQs(>RjD$D0*e%rXHb(O1Qt<BP=4O;?f2xqE9iK?=ASf||6$bB<txis&>Bsn6<yOK_
z4X<R4N@Nko4xOrLf*!E|qKa4CT1cF)qx+|P!niQiDJj9>tiYEpH$=2f^Ry=ld)cex
zKZ?N77^EOPHSVD5bwv2@#4}BPf3MQdXf|0EBae%h#mG{J-mVcAV-5~{W8S#yr{AGb
z&Tn8o`kcXHt#^Lx9z?o(QX5WL-M{W*J`+ffTdMd2Io5iT(}M2;pq|W)J+0N#dw*CH
zZSWQ!829)2q#<XIXpcaRu3!I^Jp0wRU;kKLtcLb}PRxYJGSIpuU(1fGie6EezFQfu
z|1!_TPqLHt#KA<9`-0g>v#i}!_s=!i<+YixJpFfn13S)s`QQv~(>08sUEkS`K45PU
zf&KL@kg^qxTwjVh07CYbLbjGdTn{_HO|mRX^wg*inBkVmcbH982~)-@7&utU|ALi%
zw0wHc&ri}4A@VSghtDi#BYe?<1#c-T&*8%$fBvpAnoxk-R*L_sO{6Dbs(p(=iobv4
zI0pLkA8cCkZotQ}QOi=y+o(Thv_-+XJ&r4sq=UVw1H|J<K(0aRyq<&w4eviqAqCRV
zZ!boi!as2~i@J|ehF{^Qs|{?0{i(<K8W)U-{<!Z>8%w{f?Oh&5WfT@1K4wY?T*^IK
z)s$0X?LHT6RPv`Ks;8NrQ6=RvZV0nXs2FC}T#iD`xmwTHCL_=POuv&L^Pu6<6G%mS
z%v^he8Gk01m2{UX7;avmjHkyPZ0=COLNgE+x2LCLTKw1z8q0INMZ@|<TTV@m<DIWJ
z__28$DkA1NequhC?iE<$L~SnI$#|RQ6A#qm!k?GCmj>>n`a-3@qFEg8yZK({*s<@)
zAgxtGU%+i=UfxdQ;Y(Ij(9jggO`?)lLfT?G=%7lAiwvSYKoSU@C3p2aPTjFzb@r1?
z&&^8I*0wq89nc}{{ouNViRousgze|5+N#Xxe{&1An=9K5O?S-XGd2UWT>o&NJW+|t
z-J`894*=|vE?|3Rh8gD%`3>9N+kBX8sZ&ax>WsfhMFsIHyquc_xp&3(%uDv#>Ti}b
zpv{n6M8umb5BQ0}k(SV>&;$e`Qf@?}d#uLOM_I~5`)-zZ1K1el6$I8<g~GOg9qlwV
zNht~m{=7*Ham(_vXEJv>_v$3^l`cM^p;;I3j{`dQcwNg+_^(Wlp4E`aukgzkXcexW
zg>=;UvvB^4BgW|T@rN_y|9;h$*-nteYA)Fiyw#;PmzgA#1t>+KFNw(LWqB9aT#zM8
zV9zUPht{Y{r!V!r;BdT!&`va_N6DrqJDtG@Pk(<7HFUzqu#(}(DbUD=3#tS5+JU(3
zu$`HWAdjv|_|JEPi;dMn>eVdAQ<oHutL`oazh`O|K=ViW274yI>nE0=gYlZv5=USg
zfB(ej+FAb8=;b#Tn_ifg&+(BWtT`s+i(W`EKkfCkn(H#x*Is0@i&+H}W}e(mGdXzy
zK9=xm|M(-!)QS6<8taF$cFKr5MgY)xu-@oErIKS$IbV^m5zIQ(E&6?WPO97Oun1>)
zTlkyk0$zm-*pbkt$d3_!^W`1=a1j0B-MP-i7=Ka$rzuhGWZ?cmD19HK0*N&y*Vx0|
z{Pvbs)w@@|r8fjE+*{b#@e_xyhRo$gHV!y{q0Q=$)`C1b?a67D)w6eGcL2g7E#BzN
z0P;nT<DX3zh8%I3p+B+~W;8VLcbbNtTx752(wOQS{jU^0LT}6K?MB~uV(lVXE<AKq
zt@=C1SuI%Vf1;C$yziyoE2e%%s;l-m-r(lM`TMsUMTFm;?%BQLek3}<EANsy4|i}u
z0ZCNk3B>CT^Itypgi0lpTnfykYu$EN7QeO{Nb8idGkE0if@xJDp7IrgX(q_HI*(xf
z)wjGO*>=4HzOV1i<bnjTsW9X1h~H!JduxZZ^Ucv>hsz2d3ElK>N4*w`%TByyG}6uo
zVmN%~V+>sVGc{hUENri!<32$Hp5=f&g_QPl17xeuyDQ|a{Rc6Br+|2{lo0&3?QU+J
z<`U|dTrH^<lFZlF>g<T4%L*y)&eKQKb6MUu=i0KZ-BO?4sZc)r;`7;7^AEus_ku<$
zF|~!7tMhM#kK_>>eAM#$HshR&9(sI)f<}jf?b*9>sjs?zJcKaGDi_`Pw2Dul5rgsx
zuBWhOj!wFE-HMz;KY)9k3x?*xBKesD|6S3ErZUE*Gb>(9SXrz5&F{VK)i9+}q0;As
z8_;=Sb^o<{k80EDwZK>8u$$s<=#xu)UR(EI(El?HB=<EJOMaKM6w5xZUi~?r!U5U`
zf`9vYfk34xl~it%{7SKb$N7lYReaF}C%m+DZBAf=BY}aqAlbg`5+g&|NH%Z$PCLC5
z$GZI#)0VTL&McX7zd1s7%p^-DK@{aV4Zm<HB*OAui`k!Wk_dp8CJ8T$wYT@(xRf5#
zQ6RVH67k5=T)4;au)O0-rRKT?I%z|p?Cg=_6KJ@@?;LtL>IPzY)!`Xezv05#h>ZAL
zq{?3F0Fc3ZW;D|Nz;a*jZ^~3qf4@Ue7lUwc^fD&d>hSX)zn9w#j#y=R?Ml+_nMs|_
zgdc)wsh0RW(wkVA!w<s;FWE;~L&qsssR64?=Z|*Zt^ASF@Jr71L(*f}7=y_swQAN8
z`4?J4p_t>aV<Mt&UUQv$mRji?z+SB$>Dt{L7$5<Wi$J@yMNlpSMWqZ4x{u@zv<zN~
zq7mA_Mg~TDfH2Z8-%lwyu7k)PLm!!2aX+6WoR}0|%e52`ZL;Z$b!@Dgm>!iOJwidu
zKK#D^BZ_!-BOqw7$NSu3{}_R72~qYzJig0LOBJ~ltn<OuVo0%DbcpLqVkV!4H1v`L
z1o(QN<_^_-MQIsyT774HEj(D;3wdTh>pJ4{b%9atQqJPnhx03gVWyn@Mch+eMV)l}
zjBr{;X9GLlc-?R!tD{$*%Vn`PrBKTh#uI6M5-xZW?o^YfUzB8q#0-+(<)sw3P?v4G
z_E?eRz(A<Hn_tc9+o(ICOy7NU%<d`?$mlv)u8g#}1Nm1LjW)_Q_j{3k8`u#y4Qnf^
z`%0xoGJW%{r>&(XWcvdKYKEC)mtiWIL<JWI)hW(JSs|)bpnz2KOn?^e39tQ+ao$L_
z;3V0NrBmEfNr9UQS7w(mx|%fhN?q8-`|_ennOv6;A|h!!3h*&4JJsihx`DEotG~X;
z5a&><y`>w8@qMmL=(iYI9+H>tVDG>Lv(2vB+x&A{)hmgTe2y%Q%?$5`a8u54f?UeE
z<BM7DIOs)FFYymo!lQ~0vNEG~^PKne!lu~wZi?SRr+1bdG3UNhUOJ=ajL&hs^kJ2g
zPvt>Qk7Is5%&<o>`Sg_sk(NLJ&R8~JYCT{r4j(Wc<Pp>#?>x7f0`lFX4?>LWR2{;f
z2`IfPth=E5)eIWEl$v@npSN2P)gNwLEAOLqu{FACP*j<-{B9r)7x!A^k;o*u_^)Z$
zDXSl4vJkq+Af0~Oq>T3^2&z-cQtc^jsb_vT3y0j}4_I-zoG`unV;Z<9@kLl7$X)_g
z+qtlP?y#_5Zp+tHE>q`p1}aFSo~84d=S8dVVSz#eFV19Ne-fz>UB$%>7k-8L>e~~Y
zbE+S7g}YkMrnwG84aiT+wWkBBH}&tGr9Wo`rW~@m&;v)_zh!EM@Ek=6eIPp@&h;*6
zWIn}YJ2tATTOLO|J^^qiU^Lttnb8eqKc)wCMpk#f$+FrD`Ww4CNgZ)JEA#8Kt8pk2
zEiVLF7lY|G>p285<Uy<7nGBX}zH9a`A?j58K|lk>_Nhs-k08#NaGQ?uH&pbEx)*%!
z3kIXnS!h#on81|ye>9zYJk$OA$2W$N*pQoYn3>~E=1@5w=af^Sl2gt(<dE}ul*2@Z
zgdvBVqMXVh<Ph$NAvwj&AxtbAIm{`3`~LC!*I%>G9-sI1zOL8xd`@y9pT=WuWjqUj
zeodI}NV+<qbjO+hxwJo8b@M%}_kU~$?wwmvO1?$APvYS5LiYnF_1MAuURMt(?Z3D5
zHAGw&(+PywPJ9Ub!qO10cPNZx{me=5uM{)tXLWW&etKDA@>4@`Ivm!hdg3?b_NJH0
zrOXhzE(OuMA9)cFBJqLZyivV3u<n0;{CBuJ_9VC|U6esS?-SZ$(J);t3HguI`Yyjh
z>ZI+%t1sDDD(RpHuCMmwjkP~#efyDJjyW?{FRiIjo0?tDWu&|@?z?Hb_sG&OUNVVJ
z5Mf?9(rIZr{OfI_Ru$d!D?x@{ho=X{_*C8FqXYwnC-X_lBbfVZ%NOMlenc9Hm=uZd
z2@tQ*37T|cVntt=^=%i<&4uY8TC^=pd(e!BK+ztVK00W3QUM4R{j9ED2z_=-#O-o-
z@oDVyKZDC}45$m&I@)s(qw=a>B}T6)_7S6~oy`q{wuvbf>?n6l|A=v4p3=Bm^Tr)y
zpzqVIOj?Ivfj@TTv0aG&%K-WXK(1;yqD`jxJ4C`z8;x*+yO*=%4U9W9bMxVw;SB-R
zKZ*DlLYSOAzX^2fSWs3){QGLA_X1E72T^%3|9H<`M7D=vB?@po+TJ{GsqpJgldEli
znEL4MdFs=_19feVzRy9HdaOH2xhslzP7Cb_@z9a_MS2X9;@`TpMoeK+$6I54qaNAa
zuz*XU6&0YY0QV==B2Lc6x?x0`!yU6?4mmi~S(~p@UE7UUD#XzoYHnKV!;*Not9pdh
z)KV(5h|)ZF`5)**6!@H<uhPNf)tBgFWmWLhZZB$B{hXe+$xlJK4qoQ(r1dJij3^>2
z%vd57pMmA0`u|t%0Juy#JAKc0wYt%C*=7q>V<qb1affYcQnsLH&($xIPctQ;ii>zl
z4Z{L)mdC7s9>vL5`W(A`Ehp{!(|7>@pVv3ZVu_O-$XJ?$E=$1b7*H?^2rfmA$yo8<
zIDX{N>5I3oK~BsaFAx<(t$uC5EnN2P3u0p)b<A4ex#c7kLbaje9dNxTx4w|k`#eUN
zkG@e_bFKcZ!|n`2?`^B!7tTJiBS|M@+UFrNq1@Fsp;CT60GRfS<_^4qgSzB&a<$oV
zH5BccHyKFZD;FYkwN{dhlb1WUx?||&%o;s*q(Lm^$f)Y_haAO2ylZoJkBDRwAFiAX
zhQWBg1h(oFX3Msmqxbd?>d|sk^HfvamzFclwXL2VlhXEC-~JrMz9Ttk#2bVL<%@qn
z-k{lw8^Cng$rYpP?q_d*z6AJH&P7o(7zdN|?%&`Th&vPrG!dzm)PH&)gzaPK5s0FZ
z*o^o~bU4&kld+`0MJ<85>lF|+Png8_u;_~@ZE6l!h)%c`GG#IbpuaN0v-lH?`ItVR
zX?VZ{`a2}L#irhe&14no^s40O0{SQ_;`%jyk*HpEvkG&NyY|;e|GBA21=h^Wa@vo3
zi^2<G-xzDh&y=TIeLLz4cPQ$`7Q6d_`Bm_=>StaQoecS3ohg8g4SKYFr6)kA#F%Yd
z$t<VDfzDccSU41j&i~GGbx{4YXa+b_w3eVeAo`g`iy8#}x;`Agex=?J1I~uhot#h0
zsor2vhLO^AsD;Ef%*$<+7SxAf0wTb$@$eE55dL0N=e1h&QTy4>4q;$yaiLSII<q5m
zqLuLXARJ-*$|A%A?@WG;vG$pM$!(Gu;9I+OGHrKn*A!j&&}Dhhv$J^`CnbV&cQj|5
zj}$OG47WIrA3hw&Y3x@o7SH^N>#X*=@b7eb?C<4FxTV{&SB<-LbX(!^xKOJ<`t*zt
z-Nt;2;bN~e;P=47+=A-ckqK+aF1fW8F9Q!LLTflEY#g-29;sWZXb2`=fFzSMW+mbb
zH=UH8TcKO@;n_EXaDb99O5~f63l9x=xhhRtF{E^%k^D9+vxYpkB{X8;S0-d2wKrTh
zjbA*idA(#%0RT0F2Q9Zf%VyBSXjj-ZL_+G5eMj<D7`Lqf)zv;9I%mtgXiho0N|%aE
zoVOe2#uB!Twr?I%KH(>nE}&2MLe%C~wn96K(E4s|gZ|ybtN*lh@n~8MN7La|qQWn-
zhmSr9cZNho&5M>jo0LW<jqEB-NeDT*y}71gmE*;WcwI-zOj&}+{_&lnOQ6Dk-zM!^
zXmz6JDRg5d?r9(<djrRcq5t8WvTys=%~V4pjZ2b?@mH|QhtZnrV;6E~t+D&DpIL9A
z<1nWgzcYt8GB|hv|8dWacSIs!yEbFI&g7WiB?l{zC-4si1^%_o!X2CED;5mD2bL>w
zNV6cRUQ-=@{oc+nrY!BqCoi}Ke?aVdrv<?yrT4#0h+h!e0B}H3Iv7f0?CfwkOyM)?
zkxKie@<3Fwr#J6%HKUT5OdbzM4appxc$+W5QrR7%R>U$E&&LZ^8VSwwu3@gIBG|k`
zGziNAWf`(E$^|t(gD;r<%*-3Ku0d9`PJ)q77nU10w-$t2{aCaIl;3FxwbWnwz2E-H
zkxoqggXio2_?er7uNr8;)zD+~;bQ)yceSFc_TkxT5HBQf#C2&7Sh}y7H$_5kBR0cP
z(rKR}(q0G;{#x%4UETgSjpae_t#jL%9|tkF{u2xW(CxArth;>kKD;S=BHRKO-;3Ob
z58fuZz!BHSDAnHB<nx`KE#dRsoyb?3pR`^w@TRA{tU`=?bk@^z-xa)4)BB!t^PlOK
z6Au>`;Tqp-@T-n2{(jqEh=MF=iTyU51xa^s@gKMIjrD=T#)@fGvkIP>(qAFei@kh6
z{?rB|V0mutu!2S6?#_Pd+vn3yLOU1z)`u@a*CX0#9?Jqa<WEa`((Fizm!mm=4kcLh
zTkKj8rpZ3~q{pc1fdGL##DN?)#Ujf!LSC-ar;9c47FX3kH&vIwM)Wj!(YF!qkWU>h
z4B`+MDR^Jb^{&;Y=}%Gppv=Wr(H53W08Qv63ksLw1rsiV@YXnrd4tzGViURF4eu?)
z+HoAYN8$%FK^4CVM)M37^IsJQA@-cURAl4o6qjcR$8?uFK@Vl)2+WeKWZOuS$}>d(
zmW1G3fV1-Uw)?r3XtK4>tM)6pmRX_-T*l0J#3{&8u?(XxaGB=Ny@1`?u5+$)eVUD|
zOl7QDlWD^5Cex^-wu6>}haeH-SET=8;KZdpwKsl!9T>m+txUJi_(QP7H;vlcm$5b}
z5n)>F`TAKU)?YB-nYw`@zm<^Ft?4LoC*fPm-;Sw2=)W(}UX>f^xYJ1vWgrsjmZ#<!
zw&XWGGDTE)fiRXCmz~lR+}@ZN8He2LA2jXZJ@oaRcn7lz3gpSMtuf<={nEiGQCvky
zf;dPDq-Kf2-w&!-%CfxOLeeO*r+^S{pzn3qlR30)6W2%8|90<SbVoXvU}9MjEu`gV
zUpL0;e9@N@Rs+8Ibq<U_THN1JreIw=sV8%D3nvTW+g_Qvr<8X4u?g!4)Aq?%26j@4
zr`5TJL#0bMKcsydtd#<~C2ZS#udFeI03fAKNe{_$vNGfG#ZQ6hRVqj>QPx1~B|{oJ
zkTsrJ8<?znkv`|duH-%AAt9Vl05_?AgH`x1G;4~*zkcFQ%;~kPPA+f*=e(Ozc%#Y5
z0nRDe#f53~>B__jO`<e9#^K|(f}ANGf}sE*I|(rAxEXxhgaBUl4&E|Ji4ebz%7I&i
zmKxr3CG{dK6)z((VA(49+~eV2F9BAv?vk9SPZ*DIf6y&dk!TpnG&&-pt>~Ybm5-_D
z?DoQ|y+28vVNw;-UXGQ%Wf-?iRkPs@YqA@#OkE8Ax4~DVE2gtErn8pz_kQiRytlc4
zg{=<E;58~3yYed}-y4_51jw*t8!_2%P9}Yu<rKFBAq?2Xtx29_=o6k&d9gDZfHMOF
zFnAySXgygG!jc1qsVCjdz}&3WnMB+9*HxAl`d~Y&57PB(8frXy7JTW5n6b%u?L9sA
zLucLBJW^?>rMnXen@0){)+2FMziRynPgk4lpG{g*qBenZ-=$#<`LipHh2|A(tectB
z9d>ff>33q2^brgS^!FqNc-`n2B_qc8S+=9K_P>Q7wMPWPakv{mvs(?hik9Y=&)uk^
z)z{CMlrJvW`s94B_!De<mA*fwv^Zlp<9Qh>`~HK0T|<!n*JP-E@-LNXu=ZSxMtbe+
zxUZ_iP#Oj-vT$-1-eAQ1X0F|jvob`BO@zelu~+=ytJWTCn5l%3sY$9f1L_GFfc@og
z(NwnGD{KY~Gmz?eUBe48LL=erbVn+q9XMl<ZLeh`+EOLnd71U0%({&F!0H>0z6t>i
zm+hloDef%x?=tTh9#0QdJTuK5YrSN9=n2tE_I}g3=pXre#u4#VzyX}c3QI2&9=XK^
ziP9$El{=qKL*OETKqIB+A<e!sAwIN$E?-|e>{esFf0)lJNL^JVy2w1}^RE3{q^>}T
z5c9H%crE`jhkTMrU6Diae;hQ=Oi%_tchC(g$?%@rA5-|7gbn(3gdv~;$(5RI`CWFT
zw|n9*<eTW1@uD;JD3NsJ6;bA_Dq*vAZ!1<FgWl4|?zc0+0?H8pfPi1``Y7VDdIg{K
zuzTGd5I_>B0C8?b|Cl})pguS|Tnkj0+V!$EAGS93QEzcPoXvEuj2|g61qDns)8Nrv
zbj#FFc*Pa?MdsY7NH+vq%AnGe)I9%3bA>T056|;P${_Z+-!shH_6-HOpxSJt(zI*M
z1Xk46qUZdG8ml2&pKIl;BDWs9SJ_!`k?S0mGGnWM7@UuOu-F+D5h3>a{On{A>ssGD
zI`K|csC8c$eZD@hs8x%dy40(vF>(=Us0O{y)!2r0w?JfD(m%^kGCvHTy0>Tf_<qgP
zhnn3!{r&i(P97+fuDW$y(?dos0svzd<p5k2*kF0}7etUQD$dWq2Im!Q+N+VF)OLlE
zjBcaGpckdG?WJ91^~>Mi!XY)j2lG76eqWUhhZX_!yOQgAdtOMTt{z!mm|W?Kj<#E|
z!4e1^!;8}9B|G6{D_lL@#T9m80%KTRF&U^@{3|diVVbC6z5f{fziqR=WX=T7TCW>}
zRW}46L)PSoqZXn}fvP{Di+Tk@QgnX0SgnHwK<abX(C0I98!?ONq3lOauRHQ%&W<b2
z$0^t;4gRC|*%6h-d}W3-j*mWUs^xF*!|c8}H*Ix>3@7}SO#x`<;N<nxL=qmVwd1hu
zUP+1otu{{dSC&4cz;~iyd6+3<Qb+ovX>o~V14geoSpq)emBR&vdeuyKcm>h$AoObh
z1B!7@P&hwzd1FtNRSsA6&laz1a}#yzEIj(J#|AGnBhxs$5J3^*ux{k3nK-5NS=hW)
zKb{uNQvcK(S|lmhIk2+TvfWl_k|4o|Xg4gaE*d38EFKKBH|wR8V<_{G4AxXIB7bfC
zk=9<LkUwod7X64E`cCbQg{fQ>`lv5Nb4=Y>LA?CADF?KnB=id5R@eX@fFJZ*C^(ot
z<@YWH{OKuLXm74FyY!U1TKQvC`&)?&m^>uaiS-?8e$tsKS_}r?&R}`Iw#ml`_26<6
zPb<*WDaVEF!<jO^yyf};>DZ_2oSJdm1as8#h%p3u$Ofv&%B|lA6Cd%R_2YU(6kQy!
zi|F&fPScs(wSA&VHIsJVrCqP3lx<h7`FGQ*FJIRZ{cngywmHhq%S2KwrJXAEYn7(f
znm!>vV363pT6s~CHxbmWWUd@7<#rJUHh?62BYmt>QpDw_WSf|RKe>!b-r*BU@*~cT
z%g8`AV@_Rp_!hd)8w%09Qq9+5qN0XP!>&9(rN;>Tuz%@U(4Do38CPGxAB{`b9m|r`
zxRM;W@)F%I3zq9z+8|+>ajBNF<&KlzdYvT=GCkF?J4YTWU^&sNY8>G0`Om$@C8=6+
zQrdKnWmc6Tw_AAyEQ(6;1CD$Mo0|OlD^DdiP-SF&!VCA0j1@ZavXh(M@{^<WHI<J)
zSdIDXVi5UxPrk8PioH%er^(GsSkCa)46F>#Os+(~QY;;k{~D^AvTfRHdaK+tL?O%x
zDy1{6(MpI}42>(NDrDntO-bwL5d~AYX9ZgOgbFAlQyO1x^As6Y>DY+l*FM?Htxu?H
zP{=KQe{ZGURv1PT{ubO{q&rtHTdBxi7(rkXODwr;A_YA1c$aJH;`8V25B_2XtEaye
znM&fl_;{yFjSSjx)%kYfNcrc&Q=Ik#S_e90NX{bTEJo6vi{I$hrd3IPb)`ddK{~_1
zM({IgZe|M|cHQ#j<Bf8|?||wo@#ZCtK!>ZlO3(=xO>s0=(YDRKtMko_C^dL=u-li9
zFno`zO$7>6KNFObi1o~ojNT*3i-5%iebg!gxpXYJQ0903S>9o@)1AfeD3--j#)~Ju
zLQi(A$6~0okPaf1cG9`UlGj-@R^@J6smglK;B&NU1-|KqiD~xs=T{6D9wK7D_j1E<
z8Qg%(Zggvo!M(aLBDSdia*rj5K~bT!0hWu>{eVUr4lJDE-dtInpoWY;99fAH*X&^d
zU;FS@urt>ZE~40;tj{s~>+l9-wMRzX%+DCWG~^|{9%1V(f}y~`V}k$28wweo|2aB4
zA|6#0JEG6pIIRdv*PKxzy^Z&-T`8>;Y!vM1K+~qTs$1oAR6TlDYzwNq=Iw#eHIF>J
zS#C3dmxYIgPnUifumL%qZvT?$h^h@SO?VvVo_E9dWIWxF=XF|Borf&c=x}Rzrt_Lh
zn|<9%_eUbdV{E4qlkkw-NqvmA)B3x!an^F+2hY<!3VyaUcs-&rX#mo26X)YD;<hxD
zh~{R175UllwipEfFw_*maOQ@<J*2an-2i$1W5S<BMuBLGTk%%kxJ!ku4PRmge`01z
z)3s}@ag~iBYLnnU*5rMAtxxu&!<OD5*|%oE6OIWFmwP4ggX~I|#Z*qq@J*&qAOM4u
z>;~}p@$xBU|9ETsY=4LDl#O`scJ>5}?m@I3?$6M2gP@sk8xvVeTRTQTV)alYdY&2<
zCzGhC6U9i%G=jau|6Y#!74P%Gzm@*B6Hac|I+rdQzYB;%fJMSP9p%=!-2_e^|Bj+w
zHX6)^Vh2J61|AF4XwJCvRu=N>NwVpQ!~uhDY}K5Uj+C%?VGgdEHfNVU6b(RwFOu25
z+QVU{W(E1~@UdY}{Zse6pCAnd*>^Xl65P1*Jz&P>ZleMz0Dp%mCB>BL>HL1zzCM(u
z8Q#sq>{{nZ-e0I_?rOeeBYr>ll|^crFpIWFWYT07xKpDMIcImQOlLW~sg#mQMP}YF
zu3WPOdANmy`TM!v=7{%g{fizC<6<^3cxc%yWAp+!lds@vHt6>_SS`KurovVoWpe{K
zFh1=Y^^O)}CvvJjHDO;@<$l!_c-efpu55RT0M#DMbmmt{+8FG|XCS%o&DwE3y3pzh
zYux}o6MIGKM!u0tq9u&U*eFj72=oPu1*VY=x-=BIJP~Bfm$1_@^hq=q0>RCskE%Hy
zw?^s16*-!hPezviyIg(60Oi8$;bsaf<QucPUt|4+9R_~FH<(iDes_f6OAmUT?UT2*
ze3?K}GQDR_9QzS;A65kz`-%G_;b`8@YO9EGCJ9e36>46Djs-gpDTRg1*edH{U75Dy
zo6p~R>;Vn~RqD}%uC0^5^e1`}D^T0~zwN}pHc%XP(5IY@d(ybS*~dpn)uMUS%E80s
zKCKT9`K8=$=SJ7Q(*%)FdOCY!i`@D@z&DMsskw>i3xw#%56NfBvk!r!Os!-s6Penc
znSSRC!`}WDW-%2Yd<hzPYz!Fi5AA12Fai&JK8<)0yZjH@H0U{qP8=Fx{#W_W+S(tA
zQYSw8vQUn|l_Q8zyd-|udNjx#gjlr&_jVLW>B6BOLp21Pf4<mv%cQ$aV(L!T+(H+z
z2OAUrqzXJz*g80V|H+;n4^zZCn;1z_oH-)8so3Vut_~w-?WQ{8qU8!6I3@t_%b_5G
zcGS4(x`h@!@YyuVlIKOJd}K=8peH+qU5TyJSPzFgm6p%4jpu`G+Du1gduz7}W>UCV
zpJ4LxJa4<q9ENri?Sc*A&hWVJlXSJlz}toKgCDd`FN7emsTIUijmMPr?dAUWOXtRi
z*rt^(KcV|5I#St76B-w`hq(5m9oz0xI}ykbf)4-fe-aBDSCF%+-3|2X^dE>PCEOzS
zyZ^LYcW<L7o&PMalTV6sB3$|(s(-8XDz1v`Ec{!=iMPJ6MVPkZ+&66iUqkjiUKnzH
znxbPvvh~*IWcSCB$8VRpLBkf)jT;nLMa4ZsYCq2tpRyy`s`qObsRUOo3SlSo0k*CW
z9i^#^bs!)P1B=cNme+_8nihyWkIl`rOhzw(9996d?g;=y+NkVLLuucx$uRm&6jjvC
z8={F^7lCDp&()a;wv@4r^aeQ}7$N(qkuRx>4?zGv!aO^}m&Y8UmkB?pr7FGhwK2@Q
zlhGQsQcnPzrBL1@^g9xgG43{I{R#TR62b=HVGyY~okSz<@;3--;FVJMA4eYro~&Rk
zyy*f>4A^KRfF6VCIf}L=ZTtueV;mel`aS%*rs#IcD1123Hpe}`GKP@n>F5+AaMB{Q
zC8<9d>0{`|hLt>ckW<D&Yw@SG7U;7B=o?+`FrUb=$3sG-x7(FTtP={|=&mSz2JJL<
z#e%AheWHct_l6oc23*18A#M462aAq5TR%TNSUewF6s8e1;`$Z;c6>UV^&9D5@0;g-
zk@qp6I`Kl3Q>%KuknFS>P0y_FT%ebou>SF}D4+y<X?aMglK<$p;pgr}QW3gL`9OUo
zoQXf})`hV<-cZmlx>Fg8t{i!Uw8njz6@H5u^c1fwa3vwE%|9UTI=_BM9#v?sxvypR
zk~I<b>0vm0uTIJa0F?YL;#{{Doqyu`_h6qeelmRaXSq+!fd6xflKS;EUEb@D=6Sn>
ziPvf+63FAN1XtXq`b$%P4OoYSZJ<@S9iiQp%AdX>Z#)QOb!}!s-%cD2lI?|#-)ZgK
zs=X2NR69P!t!NDLW+ZACAAE1BAMyFwhL9(KzwO$2%Vw@&x3FLQo#?ttDb)p&85jyL
zAxQ?K1O$E!mlds4{pwY1^g9!sDIfsTYn@I%Y3U@=FmEb&Zhn!$#d}DN1W<uC-mlj0
z(<qsAFACO`Z)g-<^#WsGSAR0jh=gCwHNlY-MrLxuw9dEug=5c-g4?W}huC!!()c{O
z1jN2Z)Ks~55xa<fSXaUX944!yG4<ptkf;30O}*dNPA;YGi5s07$FZiLWY<Ef*WIPl
z5>n*jzJLLyX+|||l)o*45B$l#8VggE^uP}Q1~5kHSfi)^ZZ!&Sk|<D+NVaZ-svK2r
zIC(f)S`xA`+mUod+s&p@1hjjYB?7H92-`)*au^vHGI%e?!HbkBRGg{urId)+#{u5h
zO2ZpQ9{sv*Brr^OG0Maka4Hciurzig$o2e4wKvT2V$Y!~LrF4IoW8-mR1`zlM5Z(R
z8T5t_TK*HjxG*pG<_my6Qrr#IW8?nZD9i(Z6cKk(6&Mks|EDMa{Y0soF=fKPPtG-p
z0=VcG2!W_@_KV9-YRoY8wpTdkjLOZ{)-K_J$4zuEOtF3ge?)$6wYgUC$T?r7O%y?x
z-tH~4EYG$C0aFHvwrUwF0DQDH=1x@YelS{t{#AG9Z+7{&61IL|ch}acIO$O4gBHqL
zTkcY?EO(p_{Y2CFeHMED_rd?#?E|5EAe|@3bH(4_Lmu1oULM9t%bT);rWNAOR)m7u
zsKrpbdm2f_fHS&|L5utcqcoUls(0fTP6(`=<+$U2$%~;!QB3hiD7eSes&-}=-3y|`
zHl6*L8@JZL$E-#a$uW(I55MSso_5DZT!<>StSbJ5G~L`5nqQI5oqh$V?Ly{_7m_r{
z+0aQm3UmymYjzujP8L*QA_2^{G*`0Kn&s@o?TkG2`3dN5c#|dHXPWQxUG`j@B>X!g
zm*i3K-s$O4BXyG*3gK7=SjeDJT%b~m-t~$81d?mA2eB6(sVM!i-=4f`#p{JX{(Wq~
zrR>~ZaGAL@G^#_mc;NsWrlAD+`qZWb^t^))_a89j)%}~eat~*o<v2G|L{<b>Ru-BV
zBjgkpP9+>GM;hc4ES;~)U;uuj4m-$ITN}50iBDSAC$YaBI&i|ZQRx1TFzQ0gVZa^-
zz&o<7uxd?dg#HB1|4fW&(V5(-+i%X1>cXdSU(?{~1_AIVAd)jFUjwEwA%Oq>H3fo~
zApm2umm*(B*itoSUi$g7^vTb}I?S!vCrUE*2KHb1?o+G(%-e6o&+g<;OROE1FzshY
z)-*S|&1C*kE+|5SJ_8#Y4v}V3bhy(QcISi97&5_WD3NW&DE0;Jp{i!pzZ@BiTJ{N<
zH(9qcOH)yBgLm-tiTU`*rG>kqN){__ocuAvP&NG}9q7&N+d(nrzl%AaP{R%zYeyb7
zzT(Fl4o)S-<sp0LH54*2T#Ab7theIWzRM`E-c9E-W?wqgLbo@K!FRQ_MGeZYFNRGO
zVd(jZkxPKyzDd)NRHH%P#pW=O8&rhha^5C>Fn+sVfG}OwOv&^pr2{_`EZOK&dwP!g
zR8QwvSP!;W)^<hyRecracP*`K%F=v&*IF6^TvjhVDv8cR7R%-C?thQ+uQ9aB$u}mu
zUWwcAjd_D^o|zfm*xOaNfb9KfrDq^c+8?fd&{&qg`IzyZytKdGod<92bB`lCM?>BV
zwC(&zKilf6wIj2sVuBgju!ERII{Tcz?}}X>b57)&YqAhjbOMAsDH@mwN6HIwap3}_
z=~JKmz@wve8hPa)Tz$G0XV|l=;q-_H6^@8z_=Yyrd78qM>XgJWKDFgrOW!FJgF{}x
zObK4nL9%{0Af<8Zxn$nc*#SxaqP)E0h~sU_F`-JXSPLEbj!Z}<B!^KQAJjE}Q_ppm
zM#ZgE;N&NOdb9%yad>z?I0nrSO!e@}++ZzA5Svc^OD^vAFA#;espoQA7~KFPp8KDn
zty3DKf<so}m-;3r+h-r|^=|#Ra}C1ZBjqa1Es{qb&&q%tH6Cbn)zOd2^E2OvITO}7
z3RW6BN;7TD%hQuBzHyDYIE#m{9LsWDu2Q6)r0CbZd)HjK0~Mo!W0ZU}d+&zlRg<&s
zW`32q)E7R}0Tgw1d$ZX^P2tO!yCQ#%wjNqMBX_q;Gc>tAAklYMym0gtQui82m;ZpZ
zTb+I$>Ptf2wsvZo5-klSrI`C9@w2kxK>l~xS7j*e@VJ8pL4Xe{7#6p)?U^LR@5Nlz
zHkv3E4@vxJEv5fOq_402z^~JXT3IA7o)H(<zia)Vex1(iXu_jBD{H0;_QB}Xw86*Q
za!<H=QZhs@STq}PiSC9QK5oV2J*<TN*+XGZ7ySN95aA6T<XcD3ZlzZ9O6fmqVBx-$
zxwnTO-+R6uydP19?#k4Xv+vL3=D#S}%je;S!tzxg#J9K$+@To74K<mrDb3K431T80
zhQhL6sUy!HwgKbM3SYY=g!-U6yg32R)wNQJmonWQ@i7Q63Iu?^e=BUpa(|VM*%7(S
z9Y*ygKSsHLJ~Be4EXmOf(%y|*==+=EBEnb}u2*vF)zd4U5FNmsLnRBKBA1HWy;qet
z$7*j(ds*K3u$?JoAf#h+A+@1w3x%9BSZ-XH(TY7;S@9R9wkFRGE~8xlD2~8+h@$!{
zQ-HQn8XhI{WH9+SwCSXMK`81Ti_?W|z*zWM&*MCdA^;xPvwP?z`I(RZp)MWCa2Wv;
z7Z>4e4G7TH8H7jYR}HTPtI9T?b~km}VFiq{Dx_}QxwdNp!sXU&F3j>f#h;yPc}D${
z>S%iJuGm*hKv>6{8npd^&d7a@yxgBI-?g_BaZcaN9Pa3^7E;R#B7ElNZzsgrUu0FR
zlB$xD5LarQJ_+xD_e&~R7lNQgLDE*E_r1wt54rLa5;>~75^tn-s+4>Wa<c=N@P1Z)
zc)X@6XHw^T6yCj0q6!bIo9z~kS*a8ck@utXhv+t6F=v-Az37P7S&b!jRHA(l^hVHy
zlBRhY&7XX1)6{syyEXIPU;;QVNHv$WDMv<onNoYwvJZ3-@nF0D)ZMo(wWU=1g08wt
z=6yh-n?lCI@8N-i8QRuxr>x+G4CT9ttRGb;2X2ZYWux;UUir5faAG67rltT6KYU)I
zRhIAv*0U8KP4v>jU&pN$JAWMz$L%y#loK*B&IFK1o1PXLadDWQw$sjwKs9V11r}AY
z0wCH6ES3g&3E!Ic+b`&S4J;b=Dl&9Q%cqlm7kqr|wp!^Ac7>kua4h(b+nV(kz$nhu
z0hoh$IGSo0a-p@5t8x-pLi@EuBEAmkC?X9PT2*GD1EdkUfOr|UAENnQLH^4lMEe0J
z@yu+*Zgy2+kc&sn2(zm5Z`5Y#b!D?(@}p4Q1R1Fe9b)pcM6ci<8+>nJ$_c`U)!+tK
z5c620XI@8sbKN4g2ay%@r;?cMA|O?EmowZYqj;{MluOU$01Przg(<{EpZ;ScItS`u
zWn@L)Z{WdIT0bjC^+La}t8W5$(nW*Gp0WbwkITWg_CtvNtrgc8mqO6U&tCa;>lZpQ
z;gE_#J2JVwKp#}8KK?BY2{j1+9p%2PP#lN4=zYWga-0#a;?2qezVq*Ga02Edk?LMF
zWDm}SDg@!gQo+!Q{$Ap|H78sR1i>Axpc@&om|$NUltn}|;-Bt%b=lLqW3BY8Ipx9G
z!TuKcq~*6CRC}kKaA*`QOmhRM8oUkfsR0`Q`SY<eMuz5I-DwDr_lh&H=?QSa26y(k
z^Yv|~jlFre-IH0mVIfQTc>NjJ%6q{XrUkx9XG<Ri=j(r*Zl9U6W;90d|KdAB=Vpyn
z-A{ncE7f3&74sCGk<W+k0)8F^W@Tr9p_N(O^eLO+MCP^EQjdthOgt)MoV8Q}{<`<j
zYgBlB!nL;b<Y2+yU-$wSJ^2walxuagjy^uB%Ki71MOAiH_RTCADLq}LJ3h@(o}{0|
z5dgQA9}P?E2)m#+Wz*W(vPnUUNdVO?{!9#i(((hO3y`Msr#izJ_4vB<!|Kgysmsn3
zOb^HugNo;o{A|VfiMmz4NO?fnzgx<v7{EDqQJm4IMV8Y8qT|f*#~yhr17(RLS&cF{
z&xQ3uyQI<cG6&O3cb%(CnG=@zUN|!<AYg-Oi5k@rGrnG}wYB7*9%%UMS`lZd2{PFS
zS2>;E0d*No;m2ts-`$}mI=4HT%gvD0%^c|X>sn&YT5Pu*eSQ!bB_k}{8e&&Kxt^r}
zW&G?C&D!gL^BkI8pU8E;0>1T?)#~m~B#UC0KZ!UGG|o#@d>)Jfr6mOGZttXigHHPX
zXzD8ikBGgxDeX0nRwFu$OcDitFih$(f-&(1{RdzV81oX+`hs>ZkWYRB3<JP0an50W
z*8^Oa1MCAfA|~G&l7|v{`=u%h#aWtRled*YvHruc#Z^MIV;~3Y^2OKG=dX0c1?U6l
zzVQqZ6t7!}SC&^+HBvDR+W6s3n|Ia93*qj7=vpiz(>L@KdD^eleetA0iSSX+89=wm
zW{GCBwUVHh<Fc~cjEw_x9DRtbAp}RQ&QbRqr4YvXj{hh^{j6~vkss$2FYOCQ#|D(4
z)k)j@dZ_p1CODJPs9LJOe{QaiwEH&=sxUUhp(Vt`>g$Ahu{qtv+vrLv#P9z-$gz}m
zKR!LGLXT-VoS)K{XX%Tk=rbz1K8kEc{OPo)xh|+@v?gRE-%jk6!Rw;>NeffTpHEy*
zBBPoX24n^WCS8MRhG$}2wpf5f+U+2A23B2=KI}Q05*VtJV3=x<VDMTLplHS0RY<a^
zujlLXCFBmo)K4*34#h7$erM(3qI-oNPtZe)l_T&<<@!R)z1e|RC#>xr&{REQtju5B
z5=c+mUZS+zl@Q64zu2rH^J%Oq#U|k>vZ$VX7HT&wB-->kUyghb4$hl#_m#^vMee)|
zdR7_rF6JTpTOFy(z}JJAnN{0rn|oD>Tc1qG^<Qj{D5^AGQrKBHIa&)j!=8tYQ^dro
zPFy<g({6=nY_33~hnL-)nMZk006x-!%=N_i-jOb1%lRSU1<PJ6nywz)S~+m^5xZG=
zIc~W9zY_0w?)Oa79(NG{c-nnQiQ$wM)%FWt@D)qJf82dG$7^ks_B&fnR(2w#itB<V
zCam)|8sbNT3@Y_H=GEFI*jvMK6IE6bDrPA?%{P!!nxJ32PU)LKWtV5hf&i7ZC*ZPu
zHg8Hg5RdQmo431x*5dIMnx4WlKn!?WF{?kn{{Arv9so7hI95iC_FCpgZVi=}@e1>;
zGA7Hn98%NsM8$CoPfq#EAo0KS9ONl0^yA%FQbGzBRH-iC&Y;T`FW~2Brx%I});9dZ
zrzP@V5Eleo+34>Z@u+r|o)G6wXm2L0EK~XmgW%<btAXgYrt4fj8Wq8+BkQ{xXFDzR
z!(Cf+wimB45x{yM_79jvTp>vD;_IE<vz0OIMc20nvEBYB*Ucu{Ox4G%D-o<<oabVz
zYCw3c&qRoW96VvLe}$H{E7=vOs`~>xwtl4BLw4<;r(676PX<>0Zgu(l+8|k|>BTP{
zQ7|YMH~O*5SCwM(s!TC@KWAXNCb2Kv*HefRkI4dg<rQ_&CYZYc^H_F<!=he{#RZim
z8+ByePZr$eL@?a8cFvA|cKedz@q^#*02}e#2_gymY<@{`C=bS*zIuf!L_(=YQV-Bb
zosa4fk4wk{C_fZioSBntcg^Sz+mJ9wXq`q^Nd<`;DVT+-j#TYNbl<SD3Y+YKeF^U-
z_NdvKtP=ic<|@5<=6!=Uy!O*isHh4$#lRd)IBuDZX%=)3L+^z~#^3^Pjz3^4M9fD@
zvBH|{e5O0vF6xyteUU84b5k9|0G!$Lzg3~!^W%fUY7A(EKQsO`!{p0+2~U0*X#}8s
z?oL68f@~L5|Emkiyzo9-#iGiRR5+|YmbRc)JFF0HsMVBqz9Ye)dEMSr+M*v+11?ej
z*W%TCd3jw{b+Hco(SOiC2(R~XAyj(t=OFtsP<6yOA!BsiiZ^u2wI;_(b!7L5j{mJJ
z6oa?Sb;U%T{2eCHz0;=$xr>^9{<Y|-!b)&mbHV+zenrmL0U*5}Q(52avqbYYyK3*!
zme`@*=Jdeczg@y59M5!AV%3FOlcUuu2SVElhwTxp#y>x2;wl4G?Cx=s7F2<B{MFyJ
z-Tn2Mx3rOCN4q;y)4~3{2;OpzEXA}(m97V0%l<t}xC;lQ^Qdq(uLB;rbrsz3nut#U
zG4i=0p+aFwy!_E(cZ6e`dX>#8LhkG`TBa{PFJNJ05dA$%)Q-!9{END&G|C6+KR<nO
z?Qk4OeZg`>Vd_>BJ-_$j_{f;~!TG_-V%G=bV=)GZf49Gm&$ogvK1FM5YrE$MFJe#O
zOpE8_Zu<`vr&0{Sun%Kr%0Q+w>iJpjOLkz=%OW5a___5e5Lf~P?YDMAa=$YG#ere1
zM+`tgAgH+e{6Xtw5KtEg{G02*1gvH6@UbHOf4?c0jTuNrA7K|47X9XA=AmAz_AIYO
zdS6`NG!o(yT2SM%b7DE@+j^w@ZcuxLyGjR=Fdoj|E&p6smVuKWAx`4Uys8NZG%dUB
zH$G*|CDOK!N<JJI_ys7|lPH<4t>zZf#(>GMtc;(earpU-)OHjE;BNAp1eks^6dOsS
zD_@+=u3>(l;ojsZavX|pb2>`#o^uw0OmWJo)RoCHxyg_F9FGL+f=zO~h9>M0>+!J{
zTBn<Qe2Fv<+DT}K@9=`aYzFyb2y?JUo<(?9ztT*DLx=bI(f*nRXrkp!+ZEWFb|AYU
zCp~6Vs!$w=gWyzBR<kA|G|k#pvzvWjKt<$O6?%8QIQDF!_#Br(Y2}>hs>j}ym9hCA
z-F13Ou6NJRG$NXedU$toe@Nz(fA9TvSoFcz6TanfyBRI-RJ5k1_n)#y+6@4<&CS+|
zt0RtVC5$eJv=2_A7p$llc`l3q93#~TQ>4GNM#^eh+i7=&v4&LN`*da^&9}o*i_TU+
zlcMJe7)le2T&^YdWGjH$<~{&fH8=G{E&l{C`LN5H&d$$U@|CixevH=%|ELNmDO2P6
ze5)<|%~4*W!MC1QL1cJg9)wlJ%YS%bQfqX#keeH2wM*YD-~s|lp;uKbqn(@{8Ez^h
zh%zgJ(wr|IE?>KRYIDW#%|TJ%Xj`@dy4BCb_s+<Ob?0@_*D{t_pzj|Wa7-__0ixg+
zqPjnnL-Pg=SzujbBqzWZPI{y5?G^idZ5LSiPEC@@3wV$2i03R29KW8Gke8%!FFO+O
z%ZUByiuXF{Y0o9Vna#BSgN3e`n5e(eSJ#;M@fvXE2xmN&D3HnAt$1-U|9*W4THm{w
z@dNM}$rS-pJ{)h*5g3irN(*7s?xaoIO1V3=(;)*wqLWu}d{)x#)Q%2O7pfp2(0KZ|
z%MVL<no?Q4idZ&Jv>c+BVYcM<cf&DTc7}|qh`l&WdY{Cakm!L((u2rgfA$7WL-HF)
z3Tzvx3*c(R{kOXtek`62RYQxM#en<`!+;r?hy0=uY2I{a9vH{+{n4tOj=;o~pevU0
zE(zcZ<S!^)BWJ(Mb#WrPcvH;Slglj5eZZ^+aLO%#-6#gwHNm(G@ShmJ7$<f>PZGDB
zn3ITcgeL#e0-@dCK}{P0HD=A6{M2d=fi(Jt5XtTIQ-eZ_xxcx9zEg<{^19Bx_og>c
zb>iqdroATlLa)2@Jp->O+HD{A=OQWqzAE0=`7<@9zMA(z4clwMSEVkRb~7m(%#8Nu
z4j&btpFRxWW|%pEUY>sCsrlflT>Zob1Gtvv?qG6Lsv<(VBF#qecjvXA&T5zOg_y?%
zC&FQ6TB>bhVjv6BD>VP_JGtJ3-|V1D&%56w0u}u^c#v*F2_RD`2`-%~``2%){{?&`
z@I|U<U!7PxrWf>MT7Ulp-P|3B@$GE8awWtUlJRHqEpBd^LJPnwsEM+PH~-=ze)fDr
zu<zs78NV3K)0g~DfU=2p(xE!X_xBSzhfIq8qCK8Ny$OZA@AlCxQ_Ws@v6l$%cq5ex
z8Fpy02qnPeGv+Q+Nnhu%QrB3z(DK|#{HL9#Lun8{w$HuDmk9|YX^>ur<QA5JY@bK|
zJlck8kf&h&I3PY9qzzvM_woJZIXzj68I1q(AhH2EFmtxidfBunz_0{>b@veh^%x`E
znQ!QNxPjQTY3M%QsKJAImbic5n+p`$z}8A7(Q+uu3$i|vG?M}_Qhrsbg0Hl^Y0kH_
z2_5(L?PMtxH6CL0#fmy4O5&k$I|GO2T9akTyKC?3-kxawZU-zox1J<{|95?>hi93j
z3|zeLIJVp3i@9aFkTWvH6~z1a!@%F1z&1xYm0Vv3bfVYQ-9i$)&6u-n<swM=H78s=
z5KzPaGviAJV<^^JbAAjQ`f0GwN`HVe#@c_yGr19~@B+y;fw}WkVo$ONuc#PSlFbcA
zCJcU4JJopA^G?m_MtzMQ@VZA2WjP5XWAwz42ve80s0Lfrb~puaF)l=1scIW^>Q7Mv
zIl;Km;y(dRb$0`K#kYBN;%@nfxh(f*#e<YfvwaDRnshnTnOKI@%WMVsmf1|_ehX*N
z{>gwZk>D6Id%W;T&2<R)J<|FPdsweQAZ{$Y{ap<u2Vye%hIlcN4sVi#ZQY?!v{#oP
zP9%SSDqZE1>qEzhMW{aa<Cf=M%WmITI6=3+gX@3at3n?)ZxzszoI3qFxB7SKHD6#%
z#H?M17KKWsKIo49yHPcUroX1?+os)!FLWHo+Q579Li&&WKHrH@jNxZ+(3X}`-tqWD
zo<sTmFvhx{et^o4zU`)WJ;r11`3l@dW`83_#R1&-`*hvRpI(nkh+{*DW(dJB=#9vj
zqx$o$^TVTDtqOa`2U?;-gp;iun<%QA5k3xSR#l)mdCGS#Vt}fD&PYP|IRkhBID^(%
zyq!%`Qe8q%+S1dv^0uYsm>#+f@WU+(Qyd2VDJ19~yx$U&sI1dE*P(YO=#@4mHQ?%f
z@VDmxz*j(Td`5;OgGQW-_ERbJJoKI68}G0I?PMSq-`X|EMWHA0*z5XB8NE1c+fE6?
znA8q1|GyPKwfXBT(k!Ab^I!8sT|mhk!l#WY{WRB;9&z>NJq+si5(^v%NqL*-HzMFz
zx6bg}!=il_D!2@do-{DL#R5yiJ;-aV#kDt>iYdP6&}tcBho~4ww0%PVQt~BdS<(wh
zneW~+b(GI4z|6?<{ysj{y4IzIvh#DBKCS-t_)bdJYOp5b>zW!{a!kI-Rek2m(MrB1
zDv{3pq%Pl+;r-*lE(%$-*g|{K_TPyF45bKVPv@O<$8^kuByT^Cy!ibCCnM9ivaV=)
ziM#G(-16HmV7B!>VgkAk5wf=qD3V^_Ydv!AzNvd(n~yW0SDx$w=da{Y^SFU=2Isgn
zo%S${;C)d!0)I8dRdbU6Rk*nDuDZp$hL;z7NW?u?j-D1EIrP0=N5HczH;MRh0a<3&
zONkK;>5Xn+33%pdY)8lk5A*caT69eG{Tbi;GqoE^|Ip8kMs!?fi!Y(Mb;pblN_bn_
zT;KebJF8)}GzZL?47j-#xTe4~QSxZ>k4@|ewita>{i<@+BW~pJ+uAm)$$v&a5Ir&e
z#^T1+l#|w0|C6rC&a>lPx`Xo=j}f{~<AoBYg~)b3#N$!WT=VMb@o|3D1|7vO*MI7N
zen2-H;byDQ7R6bE-$;&*bu;AQDvNqQUjiZFYRtb*DR^7E){xoj=ZsN%1E$nlU_rB!
zZNSu<5nHUX7)vM%_@6jaQ9f2+GE>K4<~<3<`h>yp$a@`&{s`6-6RDZpd&U)76SP0-
zuQDR3QSLh+h;lhG*mKOF6bhg$<O_Lo@ue1GR(sI3Y&2x(714|8rYweaJ>ZSF$HBO^
zi>xS6LiK?yE?~(0o8Ab#*T?GGKL+Y!^%KTDeUT`^68?T<m^B%1NVyDCRAI<_enl3*
zb{iCa=;;Y_=28*A>bIg8%Gt_SlN=YX#fW}w(Y5#xYPbf{dmr!m?b*{BwZfK_?>CGj
z=mwUQvhU}VqR-(!+rA7}@C^WLv_FKudkI%ykpBRFF^;L(Ro*<&8{kH3J?&V~^b@9c
z5l%#NEdWcD;XlZy%V@1*APEYSa;THT+@c-tJSI<@0W)U{rNZCnCPJO)W1QH-xbx+8
z7Wx0?CbEa^dl|icXWp$;C^nxCK|P(c^+Ed*TEcuMlqw1#Wv07M$!mpE4kHZAthT_0
zx|7!K7(3x%&;7K(z$V)y9dPCs{Ospswp3{PB?i7y%qTu??A?3$xaZW%^P)%u9DJMa
z1;mP-<8*n25e9eBEzfL-{SgY|d%~KmbyDJG3;XGuCm7Ms2W1v@lsA~X66$j28v4#X
z!GBbAvigO-3DzXx6!b<@@>2MZk=cg#s6_n0FJ*{!9ymkavM8L`wC}M@oSV}WO07v!
z%gS(bdmK0COPqU8(HwH@uWPwO$iKU*;|Wmz@)pXADKjxw*>{$<X4+c4Tu1SsTyEMA
zf8cG%_F9y800Pq`;^((w(cQ#XXuj45p`>Wl2V=Xz(i8j^@k<9S^TPDYcBgz&dqqEr
zu0S1LQ?NX+r9xs=N1$~zemPmIg}|?-1@{xLXNs|k5L(w_JG!-|5BshkpI3879;&J9
z%6f0vZTU9XW%OO@V?v6mC~_$@1Spg!_ikh$nDx5%&%)J{(n_+u8g4bATL#(<01CMg
z=I(IkUB5Z<f<XljYKJEp5N8-XbvqEQfh~8ILurGwViy1?RX>>(p)Jg5dAIf<@}`<d
z8X=haD&^*oDLd<Y8O^WKL5+Jgd*rd3U}G8y4T`+yYp8mJyu_~P1ahKIw*!~5RRHsx
z*S9iBDx3bT0emHbSJ1v~RgvQ>M`;?zhu|(@+h2OXAk4$`oUO3`)Z>IVAN8;Qcczl}
zS9oTubWqto<Mx!M(3K>I)E<$<&6TsgzlO&a=ZfkMv~h33bco&cli=lk<9_0u=eLp?
z)Qa}mgYqpN8pG+3bmC^bU#yTf5U95sQ@(N}{G}ZTU{M~i8bw#nd_lS5)klJ)+_!3K
zYMZA|a^)UqYA&?KoX?>D(%WPfD2S=q6zox+4@y1yZ4|Y}!C#oRcy{7Hrz%7a`u&Or
zRi=D-h7Y_4j*6Xu12n)r4w-S1&lfsRBsbIRflECSE1qQRt}?4AyOi!ykUDzXKs0~T
zNnXJKZnP+dej5*T^7a!^5fykmE=m;a>l^87iVs2R|06W>`tH9OANkr9%`ub_zu|PJ
zD&q{54E%T;zRXAIdFOuJb5<Dh-6<nb3ZwDRWPH?8d&_`p%t?6hI=#8&m0w{Bw6<G3
zXs0%ZF(!|<zavgh(<<ZpO|?rFLJ|a?%2&t)7MZ_9|JN%h^DY-rf4=tMY6`ShFu(8n
z-d2~X+zjANXMq}^?O#S8Dwbg8ySQ0X8A3KnbUU7?*LroeDwmv+SUVf}z;CXn!%32X
zQ?)&U>JWIgmRoFBKm6wh?c)4;R?4lcH~apJUA%w#g`kqm-fS`V=_TIE*^l$V=R#*)
zx6?i^pc`MrF+PgwkR!y5en2*cFV6EY<TW7WU&GvGm|w?}M7^LX@AfIsldk4EVBT;-
zf6ZgGOzm5ifT0R}uhg17xvMsHlLKhFYP$SOQ~F1#U2vA`*URk9PoUiJI2CS&(xk66
zv<dr*v88JY96XuHZ8#Cd=N@kkrXlm|h>f;?(HB!HZ#296(}NNF>X)32j~WVUC{fh8
zFTuIDFg6UpLB@`wyB+)>Bsa%Ica5M3FUPW<!^rrEL9x=KA3_5pR+66|h(~GuPcruM
ziWFAh`P|O?pNA|HR=0$;QSq)_XY<9xQ@~1^CRaLOUK<pK#FAp4zU&2A=yQNB$O~_w
z3zQ?H4h2p%D+u_k`I6>^4)dzO<=wTwSoOPz3`8O5P6l5_W}=n3@q8zqo3+ER=VPW$
zLIJ-nc&$(o@B<1vrLO|#@T)yA_nw=%D$3i}C%Uk(02yyzpwj9z)?GYyH};O!i`RGW
z6A3NuP_plX8F~*E|EQACqK|q@&scXyA8mD?uhE;LoneYM7;{W}<H|XpaXJuX-AeT*
z_c3|#^A3H?H<zX2wfOqO15RTy;+*Xz0C`CHb%gPt8Fu=@PTo8`OL!(<-mC(Z5pnCc
zp2i+-Ec)ZSbh!@BsgC`tV0;3kmfY1^I%nEuG^DzHJm|C4NP|AIRN!6*qZJIb-)uG?
zIff7m^11ZCib;qArW}U~hfopT!o$9^%w_X3mGqmkk+h6ToGp2{eR_ICFHa!-)<k8^
zQ+s98B+d5<A(CEkFGQ~gU%UpmAO>cY@FiI@M($EDS;HG36sfKTfRIVqJ+PU3Kd;sG
z+=<tC<FQ@W+1WLd&)?|ksmQibz8vmXq{2is5nZISc`tjv0t0r)y0Ut*J!H#j)U?T;
zu(iV2bZJ@YY01N8D`VY8TGv4fF6^+(<^}+8_B%Pcdlu`=p1}03IY!-S=uhNLftaXw
zwG{2_$`?Gkt6)(MUBG~d-Q`Ux<2$Cz_vO}sIG+i+6v=i(C$P&NT|W`5+aGd&5)4H?
zL4Gk^h%Zw38Ni~q&KK<DlH(+VX3;@vL#{0SC|&uMFWA?){Dwi`e8~CDdnY#p?6O<x
zLIF~-D*h?}VEt^t=YyR8jT=!a7P(SPAj<}D)%?3^?=98v(0&ZW8#>5qeJ5r%krj$J
zox=0XOa1`x%}zF<Lj>)$ZuGz|FIxhVlmC9RQU(Y&aS2r99EtcF0^mYB?}a`m|3J^L
zZ{J!~ACTMW_v~MM@aKbXG62v_aP>(GkY)ui7sK?m<KkogS^C)n2Yl4`Z?6xrYl)fN
zT&R<qsb0;^y;teR0c_l&qsnr#Zi;x^xn^}k<CXa#Pvl1yGywU!I=*AStNv|U<2GmQ
zPvN$4QOCO6lfNyro2Eia=G_Z(kB^_vWq9RFc+jtOMjvNl2-9C4bxwtt8CNPaf)#Oz
zZ)1+g)Wz8I_WA-r<qi?&hlWZMJMl>qB~dH?2N^-;zQG(~iZRX!4<kDG!|{B_M>F+l
zej8p91Yjf)e$MW{mv?_SU_(P3>EQ%G1WX_VJgNaZd~abvKyWD2j+!yT-ONlyc&r_>
z=;lqG37DA)U0pr2kI{xhz*G`(;grmyc7<+R)xFwYb*+^Mrj&BdNr(W6g(PcG-}|z)
zb?Y8MNJL!9R7y!H6R~z3Dh@SABxd%3jqKXHMZjUZ0S|Zg?yi5~PLu*+Nbx#{f>|hN
zx7u3WHi7^`qh=JunApS33>=7vkqOAUwz>i+F_vjcQtI9wxBFZEg4j=|lsMRZK>Mb7
zs{6*wDJ2mJB$6~bZ__-@r+KeiU2UG{QgUnU@wjuSoJz{+L~vp#431F2A5E8L`yZ;X
z{4T8gPjx;3jF5liYPMFmty@l5C^TT*cZfJ4Od>)YNWstn)ZJ7qg82aGkEQwD%rx$o
zghZL=^ZE7Dr<}8?wYongxj+bvfP#*MKqKe!`9pngO%K~09`bJY(avy>gYr1WRyf>%
z!^%YXRPnjvxH1jEIzaT^OD-uPkqnH?&{?V3veuy)EBSx0t@%Oxj%J3fkK6|UANO}a
z2BfaO*UnG7VniSUF(IJ)NLSL{217+h7{zyK{P@?#P$nbN;8GGWfO!md?D$NE5y$a~
zL&Af9@Pw3%wl2_Np8CFHe1F-8eh&eVKt1-R;{utOkJ;^Di$4#|h^PJrV08LbRlD}y
zd(+;Om`i^uK7*PQLR0ndoO8*USkz2gGwp7AbooY72aymPsrBCNyX5W1E;m2sj6Gj3
zO>6DDnuZ4hrm>m<i0FJikI|63BN3<cSnhk>t+fW>z4unXe*5~<r_cZNPk)-vC+((t
zfB*9KxZMNs2+JDkjsbc$PbsC8%}_xR{y*B@v`K%G<hT-is%qvQM`j&pkYEGkU}koA
zWd8r3Sy}9G1|;t4Ix;g3cQ?~xKd5<Dk^6<UvK`$5k#1yHMufYosXl!N3Lq>p7&X~g
zIdKUh^Kf$n4V~1ke13I0P7z(|C`~nq_RN$yR-Q{K4@6`rM5PegosTkTD#*qFfj~MM
z1#^GyZFj#Mh6e|+gDuIg2y&=lZMtmR+Pk?Ik$Il-({=OhzN$xS-qo9WH$P6(beQWj
zWu*O7=e67Adi(rxeY)J6dJzG!NSUT-p5{^{5X-tO%d&qu0C4xNy7k^oLFw<7c$fi3
zp?QltV1@iRY2{84?rmB1<z~y>DK;sS936k6!QBQGf<aX(gSB<QQxy=w!ey!l@6p!n
zdcADh#rkc%e)*?g|N3-(^XA>#)8~g}Syio;$|9v!nPv!E#PM`Iolk9B$^G5iH>bnF
ztuOccMD}*FGe5jNOm8RQ63wxJF5NC`-+pf7W;F4@DDs2c=YWHn$G!FDm};$ulYW2s
ztlKJ#X_}`}SqcQ<;ihd{El{ROL_8AsgsE!Rl{>N2QjW*tbT~e~efz@?KLD}Z?}fq0
z@FwmIa4=DjDBOcJ{MD)@5hWfC2mwLvp&FW!C6ByLxz8M)_D;iI7dx??%w2*+*qs3M
z-pvgJVJW4I14%l}HGFA0ul+FTTp5?a9sv}8FUvwiQ<1|o)gn`sxyX=EbzQf%ZB579
z29TX{HYt%#wC-+1lIFt{!a?SlBO9X=Psw`#LJ?Wrv-)$c4|Z<E&gp-pVP<)W+=!%c
zin})pY+VRGq2v{GAkyU`m$^Cr1rOr!{*E1{cDvkr(?DVp5EeI!B87RHrXu3z%l&?T
zy<V2v6~JLWJU%`ikH=C=Eg}eYbE2t8x9_4>0^i5FasTUI|9rpQ)^!!)sZ0!R>-P5T
z+vDkkh^@8T?fUuiXJ$@6%dKth%D}s~Zz%kJy%bK)VzhPr{ORfU|NMs!fBST~EZV(y
z^AMJ7Rl>tZXp6ZyMS#BgN0ORpn8I^^2#b9uGHjYhAOC7;79w1T3kwC(l!M@<Fx4`)
zwMd>sNZw2_`XBdP!b8=&!W|+)0BHCK05DT?Pa2e@kqAhG5)Sw9*80}9sdbB0Eq{yl
zzAbB^Acg_mV%w}+sE4^1HSOBFrWnM{K0jZ-yxeo~SW2C0NjJP{FC40ETUYIWod!t`
z*0pz2H4G2c(aaNED2NfY0_0qEE>`H)x2>ZwtaWw>l_ImNc4rQ#_13z(hdVjoU{8G_
z*IJ0C>{g3R;UrbU{eHbXKYzSmJ{O9A{LMd=>0o4s^PBo|!+kN;QcJD1Om*o^BaX+z
zfBEfi@t^(^BYynV-&6Qs|MF*lihh=Ocj6x&=C_Ah$Q$AwbT_$ev2;>=-x&K{uUB?M
zGAk*UF|XteB8|9h?cVi#et7rhSMGLy{;X~7%UzhKQq3AHDOPD!r0IlvI-T8o+nVJH
zlxr>0#j^Uh<z{Zn{oc35+G>3ZYX~J6BX_70C~1x{fW)z+3i!aRnW>b25=1E!NiCv>
zTMQzK;C27>bqh6kMSGK*TiM<lPyEVmsHF@c38<;xw?;I{D4s%OV^Ag<B2zO4%wk>E
z)>~IgF}8ya5djGhJj@em49z;4gdlSkPWuuXJ6^JU>K#FmoOt08OoZLDLK3YZM`UUd
zkZ!NeGE5ABolT$7v-`7uvrA|&s!0SKWT56iH&siCWniQx?!KHmjS1uy>$<I5YrVIv
z-TA(?O(dO6**$vOdhY>f+q4#{B`d=q9hqtC8!;kKtM_iIaC<yIsGIuFhS5jFmrtL~
z9YC!j#0Z!h5lyv<NHV38I%#J2%asVs$u}WpDYd48qMJq6j~_lf{m=jLFHcXGt!X!p
z5toxRN>6|12z0d!ii+VL2zOVv?kSi7d&Ewcr>}F^*ZB;|uubVovJ2b^?oJ-NmzN>|
zIDc&`3}QI5fl{2F<dZBiHZU<?n!eoE%d)Op&k{e?DM^{NRuFpMwrx}G5uVB{^-wca
z^<D9slRPED6)uDz4q|6EhH(UkBf`N>0-_+dFe3p4(YCeLnwzEaCWq;quc}umAfhTW
zK@fC29uanZxx73-9p<uaJ-`-FbG6(QNPkx?$9bxfKoSV^G(DWp$9bNsJf2V5x3;}!
zTTay9*Lav@stiygUY$4K-emELQscW@od4*HXiR#8u$vt{zLIe~(9~~hJ8k{pG(Wt3
zR~VNs9{~?-=qpIPb6=aAo4R_;(>zVnVLoolrrq3p+}q4FxC5cRwQXH4FWdd1+kfJH
z>)Yae3$x7Y5{3v{c9k?rGI=0DX?F@v0wpn+6FDB39qpy?>(iZoebe5)PD$>rs=fD8
zYW7=1RHQg4%t4&6Mq#c+fZPceE<wUzcz9R0aPly<ZA%+@*9hw%(rlhHxFmx5Foqm{
z-22x)f<C`4JQFGMa88qrgz3pGhkqG(gm6;$b}8}Bq!gjzk(sgkVtO!f_KVJWFl3>R
zWPX4&&>a{`ALPmdNc9zVGooy#!aZ8kb=}r=Gjmmq@F-EaGE=yvGSu8ndv9B~np$Qj
z5+TgpEZpHc^n_u6l#1c+x{bfD(BCfC=`bHp=QpRb5SyB*<%K^S4o&+!&wr1PkEe%+
zruzKhr{nP;B7|6$l^Ka#b@QdQ(D?Av=l}K3|GF$qEzH1D6T%z9Vt|+}(A`!u4+3)`
zc6ST!uIg^#6r4n(7}MF{RAKOKF}O()M8T513YduKm5!NLpJLeGLl~K3=4N?nIMB_T
zIl^mYAy^u*=T0)=aF#@t(tows0!?k}-OSZg%?;22Bl7U*Z@n`IgGxc2>Qv!~Ze17V
zB2ob|$ENyxz27(00GE=)p=lb0ewrq8U01u`?@22fc5LRR=H@AEmm-mR4>Q`o;UG9j
z5iQ&ib*>~L4#}K1^R_ldMgT@zx2^Y`C(H<D=2Ao|2Wmiy)Tx|$Gk-IGeEa5heY(6n
zN5nMG)A3|M%eq}Jx96uP(^dtG_gZ+W{Cr=-qtseNrc&mqPL)4@{%E}&<~oHvGQBHN
zQYSIh=GYt?%VN_8{fUTD001BWNkl<Zk@@(u)%n3j%mq$1&4bUC4KLsb4ZPeJ<`D9~
zd-w3BN?1qVC`{KIbAO4Bwyr>R_ufo<pUPCE4lN9k>|;I<!c>JRye;?T^0HiB+|+g3
z?pIYs1cGa=c>?WRNkmvUBW;Q;%Y9Dx&Z(FeArE*e?<v=@G3ev3?DBu}%I7Ik<<xbW
zjP^55gewzCdZ(eOt*v{62s3k;#`_~!1XPX9VXj@ZtC^=z#(yFcchn;ST&sXcN1><J
z{b)YZj1bPIF(+@{x5tB&BTV722;oq3^WfdhGW^xT5`?tvAm!0frKwix;Bu59CV<_Y
zGJQM@C>f}t9z;Q5X@D>Lw?A;9{qzhE3z+({Z1?*D4<Qy73KWteqV998LbPpL;(FG}
z&3bRzJ2N|bSbvoCNGJs&M%)9-vaI*zw%mU<9(^uGOPOmu9FFkNe!pI?rIhpeTx(71
z)zj(3;II(lwl)*0wVcmKVQy`krup>na5_Ew@%MlF^wZP*wpx$0x2vUsB)RGY4@WoG
zpdMi!9!@dLN|V7Iz_2CuA#Y0{otfPNl)@;H1R}HJF@Iz5xGH%~hHt$R(a0^x@q<9g
zVTy2%)uKCyAi|(9S2G<^Pwq_2OqEE8L6mXEIXRkX3=SMQ;KX&BCT3wdPIa0mF1+^5
zLy2fQ)(8;u<Kx3TRS+^w;=cC#+6^pa0yJ}9cYOf4+qSLiy7u0)$`g^QsjAxE&J!mo
zfdhp&kbmZd2EY)_jcTd$oca(qx2?(6ig~S54I;y`ZR$IBI|i?LQ<dZN_;4aNk0?BG
z#=y<<um0ih@3({5`o};1OKaM@{q)nP`Eb17?mvC_o4_Zb$A|NLn1j%Jb7GFtd%xZ;
ztu2Qt+FM~d&E>Hk-corJ{Ww)Q%&pE}o^C(Atbd=M*9B9PpeX+@w)_82F9K)2UBqjl
zj|fly-Lbt7kGL-z!>iE4`S3VTZ{EMhws_ynoXBUUV<l4)iYk=aB#|;tv#O2-CnDUK
zq6nA5OpyTtQ(+Boc$ivmp3&Oj%o35juOZ-qo4AoD11Auu6cf3b3Q(kkdR}|Vw9Guc
zvVTejrGdM9MpHQxl|r<qcbRGLO}Johs(8N@BA^z_)>~^pu#_y~ph%@_bge{%rBo^4
z5-g$W76A+Nfx>7vFq7p%fO4rE9O)91yX4W-gD`dxS*Ay<9uYwh=~bT2-U%i#3x>&!
zW#=tLiidgF-XMA|O_GV{qhLo6C>XF@3V(wvz{rPa4E2IM_&udU3__%tk6Y*a<@Wsa
z)O#a3l~RCU%9XcIb(*JIgzHoRZ0nW~LainHGuk~Fh)ASRpl_{r?LDI~zT@?MJo-9K
z0JL}QjX?RHO5T+by{oAL<R03st=lwD;n>#I9?o?thf+$BAi3SvfBj#7`s<%Rwtr?}
zgdk2k*5t#a18{_T^i1uF2xMxMTZDz%=t-H06QKu%2p3^yG6JQrK}aYe2pv#9MK=00
zgyIOdBobuZ%%(m+UCLe6eAAS$q_#1J9bf>Spx%hH<Nc0ltufPF>sZ_~2guFUhQ^V=
zTq>6uOp4ac(KV{NQ2<<~Zt9DQaDO*nyLAtcl){y%mLgwu$mqS_@5{Drz#!EGV9CuI
z-%M^D4B?z3GUa{=<J1I|Qpz-qzSRTf6kwvtl|4Lc>lyM%yE=i1Ig-s+>)~_`529Mj
zR7yFYjz9d}FV7DL^G-rP{P4p(AGg+-#fQKhwMdyxQtIwXh)7-vBN$YfB7X#4pbwQF
zq#P_z>V|Sre0W)Y`m$bc)@ZT<ko<hRSY`@;rEz^FUt`B^lWf`D?_1P6U6#{aPv`R~
z3T%zOZ6L3NLy2bIlYi%$f)}Z^)GEF2l|HJZ5u`Qky=|r)rrx{v=Bl1R4-o+(>DU1#
z@<^oG;YMzpPGKGr!RonG6@MvC$-tqZFDDoW4^WeJ5GmY^jNO?Bw}b~$uWD3fhkb?+
zDS7Lma0&_wH(~{+rC1=G;Os%nA_WAfXB+?Ok>DWp@Yd6BC)q!-|LzP+!F~ZS&bE;o
zjtF)daw%j_6%0>p4896ri87xXL_;ymA|jUVs0iAfi@)8v(<}3aD1X-;L>}Y>+usZ!
zF9LiY_VUgkLT0A8x^8Of$jpf;A77cL>gKB5wWA6Usx4_erAVYE0%)xXbFH-=rwD+H
z1&oN~`%3rs@#t%<ZQCweBad1uGrxWN=JD}dYu&bOS?<enS9Rvvw|=`^AI`rL;pdm9
z*4puWI-O7Me!s0>K7V}v-T(PN{_>~4g^?Kmkt!o&mk^vPIC66h*GR<_G04meVf*rx
zB815RBuXw+D-(fz+@OWiw`q40jM&?NabZU@L;+^m`OR4?^QKbsX5K9`Oi-8*26qP1
zOREqeWxL<pj=i^1M2?4=%0uGK!nO6L9xhR>l(cCJW@G89-ha)OEje4QlWon-O5q^Z
z;9#L3A*!|T>oCh45v^_a`>nOc>A8_^JU-fuT||(AV#i+S2?&^F78gm99fO&IWH`P_
zl_KUaGxM%$0ETpph!|41)?7#&j%R?}`Qdb~Q#~E$$G4A%;{>K(|I<HBwaoKWiiq%G
zp5DB9AlE8V3V%}(56tsnZQa8?T$m11Js;*ZRJ^e*GaY7Trqa1w)IY54FQ2YYmu^0F
zo;>MqKif+R!Pvq=Q<1oncE4V3AmD^<c3sxz>-9AA!|BW@@Z#1Zlo1pA%;@9_Ak^K0
zKrAehNpc`CQJo4AdfT>j9dVV*vTf_OEUw)xRetDb9)E>Jz=dgCPeQ;I#$=V#zZf3P
z*n+@9B{&Jl5#}D0YzJXfCMgAEy{C@E#eH{x!j9U<FpwSW#Nj*`5oVztg$O}zaPR2`
zZxPHWOod69jW|<w%sf{$jy>-Q(M`$x8^S4wncaoi=#@N|>$v1$Mt1t7%t7fko+Rkp
zOyt16*MDscsrzn|z=NpG_WF$A^>YqUak7yHg9#20gAuT?IB<%Pu>~XygO;(&9^}X(
zRjN=yz}<;(Iv!I?bvhireED*}UYGlwD5QwHmO6phjil5f1;Dm!iE18?XBMVXofuro
z#5+v=A9{V4*G*h67j4Z=+p@fU^Y(mxC@c)p-hXe`Yi~_hq)P8D%<ta3efRd=JWr?d
z5lpRh(CY3V|N8Owzx(~m^Hsa6xf^BKa8NhxZXod>b##R9(E&^(DQ^DCB#U%jWeP4_
zt4vc7=5S+SW(Z|iaYUriP2Q<#8>tdJ2#7Fd`L<NV7@E-4ntD$y*0^&VzPEK5Yl7fB
zJXt`@aJSZbgbSx4WW3?E>DDwnqzDnI`O>>Fv6oGuV0FK~+=MfCW6SU=bB2&7H8_F@
zg*o<xd+*(>FU#VdHaNx8lh~2L=rP>&fW0&Ih)j!%h#Ub8K;2cB0x$t3e;+;|+*IGb
zeRG&AGi~d#E_c^X5k*R!N|`FLggJ?^xDXCgEgaUD`^&wOpC@{Is0Rik%t9ymd|m$e
z4<9bq-j$VLfPdHh{{QK<H<DQl=<8CJ5=+PypVGTx*|tqv$f+JGMb)-$Ycq`kJ{GnF
zV03kzRSzRdwAM6G1xN@Hf06l{+qN$EWqEm7m;27AitOLr0~tCDPoQz|L)^Tz4v@PX
zk1XWs-i*7GBh!=+uzkTv<PtPR5FpG^iVnZxy}i$*cEHp6BaDbkErAHL*NtfJP9+H?
z3I1h-fD%)wQkf%64Pn+>Z%JVAFhCqET!kPU0W)7)<K00v_tO&We|d~Wgj*JJPdbD<
zJB*mgv;U3&4|xlAKnLk-5&^g2Pette`t)lD|IG#(uLE12>w{VoPT4q8B$R_Zn!wpc
zF}`Z@Uy)&Wz@vy94~I8z9<{aJ+Vyhj+ZLwUTU)m})xz9+v#!1CVLsMUS-7o=&S@{|
zsVh#U!uKIJ-v!+Se@we-5BDmiYv0z@r2t@I?b>^TN8v&uK2<xPGgtdCAC_(Ht%KOr
zpFh9+<qv<|mX2UTNkKw_!`-}_5hHWu&Eb(|m5G@n$nt7u!A<xr8|_-C5Hg8iq?P1V
z043W!92g$dZsvLHG7Z56?W?c|m?xs+I>?Rzga~D1uZ?{Gf0K_}Zulzaf-78;KgUdP
zL`0AVRLwmhi^5D6v8lCa;BFDA`)yk<5-u=!ky1q9WDX|>Q{ilXkgGIiHq+Ky>n+Pg
zI&UOEWM*0P_gj`IdEXH}f)suHNd9T1NGXb-a1N))89$X(=IPJ{B0xY>EyQ9m)MQLl
zr+Gdch~#`ee_bxmUp{|sZJTPDr+Pl0)-boeZ5ttkAVl+Fp63b70TS|3D*&cARH<Yo
zyl@=Yk4y(h^xi~3<=*4z-oCu7U5f=P!Rg;Tnm^AgzWE#bV3>tJU70ZKB=aJwZQ9iJ
zIE!G`ZDRwwviU4z1zN&;hj)UPddNSsw>Hd{iQHW^e_r`@%d)J?qOExZz4nhul@F#!
zK21_*fk=*vYwx97)KbE|IW*7lwBQQcnIue+w_i!Heo&2LpJ=~9tuWG`2GwNjqMg2F
zV-k@FoRdG&)dSJhwTOxd5eH1oJct<<DgH#2u{BL^HPvK&rbQJ;$g8$1*_1)U>W9Js
zHD~0bf4sl;6}pYPdsOq-Em8oaOl&aj+;?M@Z?1X(2;W_f6Q-hYMv_Q5Mrc(elC<m7
zC^8ER`@vyu-TKxjP-_upWAgCUHbP9b9_B+~VWI$sTV&J`Q%aGf2$<A;SynS|y(gpD
z{rgmc-v{&%oCqTF@Nf=~<$l|UueYm+%+qYWe^CUND0OC<!)&?Vx3--g&s`P7rAAx&
z?dAUb@kM(O7b&2f5~gX36#)-*SI-3|gP5cwN|^mc`r08alLQPBfiUILb=)#R9?lpr
z1yK-VCnq=ps?JQzT!bja75mK{u!u18tnxWVysmaV!2)3q2RY-FGLey$5pJc<BAn>I
ze+G3AcV|GDOA$`#U?w6*fY~g}JwZevFf(%@E>6_CS$M5OBS(9cXJ)mos%G~3Z=h!8
zy_=eiGC#nmeUX?0W;$?=+yx?*X`bp~>P(clPuCQ!xvKkc1mb{z3(H}yAl|x)pcJZ=
zOW{QM-~aH7bz7caF8AB*bUeI$d{|ERe{fxwdu!V~)${RCkH^F5bUYr`ty=^W&o!#B
zzz>zgV`e-Y=E;`>loe|ynGW?dKi%5Pt#{`kNy`h3S8tP_<uy9xA$QD1IFhE$yQ^J9
zr0-q?b@i=l>y1z6a+vLW3Z9(S?kf`=f~`t72e~ljc`CKCFabLe763$8L=a)Be_gwF
zH8rFo*&vfKlL$?D-+CCb8WNFt^wt;mZc!^lT(gxh3}I^qsU><z>7y5JiCr^eKjOnZ
zWOp9e4IpEv3W3C>X{T@I=|j1^A<jFxtGV}iu8qNnDWjeMW+8Hf>#p)or3|N?teK4?
zC)chqkfPW@6L;*naM|KAJY}T$e=G*=VdN<%+wYSdeDmGQmV`JMl>E9eo4q#k`AE{f
zbRGz8dItrvP7Ew>43>GO5CFr@hwocku6OGSp!a5`S<KR}vz7`bi!g0ORO(bq#c&s?
z^WgwM)o!<&duUf-5#iQ;wvIl4bzLbi9S-Nyskgo?>#{B)d^{dc=i~FIfBSm7H|E29
zlsX;fS|;JCp3V>VbuFd7eSH7BfBoI(pPtm<Lyk?L3}6h~g=T44GvwVwAPGvoFagxv
zy7ow&o=7bs>6|ni@{@9#f^L{<HXyunMT(5n&Ru9id(27RKs$$>QsbV@FbM;0+K1V3
z^6tSbpu)Q_J0jgG#$@2$e-*vA?yOXZFx4VLfZINUs=Eb<%eHk>J01?1NWda_cDcjh
zK$Ma;4Z%VzC7<nV+jopH@8Jv#GF7+C<1IpznFv`AcM!`WEOIM^(@cdeES$q4RHG@n
zLM>$^%p@Qou3V;4CSi3C>(g=07z)>33%z^)_U+p@bt=<5{pAmTf5^>Yt#zIb$HVmW
z^d)I<$J1$=XOaRo4hYeqRtn|l5?X>M@*1teQsQhur5p|q@1OtrFJHb~t>waT-^mf-
zXY;xtpa^?-lGlSNV)r*6@u*)rUIg6H)wiwP*G1nP=cyha&!@-v-0XT?UY=hJ>*R5C
zHUWh)0nDY;!|~W!f9t)OIWdXIG)+<@fN^4zfQU$lIhnzw)OkJxpw?AQf<;(JrXo;s
z>#=nT;yR9c9}yXD?iwTVn2B>ImbZ`Z6hj0lqA&@gl=RBZw$t5=m~((FrG#4g%tVgZ
zENLyDpoi~ntdhB{9I`|3B2+LwU{|dfG6XM0%skwft2>BNe*-!6lHZIo!-dE^D58UF
zi6qXAmFu|Ivxgr9iu^x@iG8B*?lEK%!)5bz1rx7-?QX{Dq&Z*&ijZK3Fc*e!L}YO%
z7EWg_=Fr}^<$ig&+^)CferExI<e^f4geExz1FBHXs|=D_>om<$%6vFvYrFQgbAf2b
zlfL8iUC>P|f4Aj$JWO?(=Q+QeZCjY7)WX7t<1xCf_xoX<MRGkND8<9laUB+)KYV%m
z^la*;k%3}1`u9wE%HNX;x3CD3*W4G+;G9>}n{*p4l9xZiY(KuUFV8V;hqAs#evf%=
zhn=(h4PDZ>J3SI|9dts<X)P%#qjQS1DUNaDW|xtbe-{{IqlI;Kjqr4?NUolZAIOf$
zB3&Z}k0tNol+S@+VVTy)dqKg#u)@RLatitiQ}471k95KjuC*ZCH8Cg}@;?s_DkLZs
zby6m#5ClhrIccyuC2_}+`kxpsmBO_O;0O~6hEo`YhV{&ga91HZ9u5x=XV-q1C+ph6
z+&eQRe_S*j<{)ZayIK(`g%3qSjVvmn1SE{K0Ujhl!eus;=j-}%S=O}&D3LoLr{!0l
zil60GMAV$fQpXhetFy52t8ypXeFDhTTeo%HT5pHSR2Y!Qhj+jD@$`q^|K2`*w%BGw
zhXhbXoEXRx#Ld(*123y8_I8;|Qo&gYv*enuf0QC+>e}29V3tyfRFSN7t{NZ-lz<Z(
z@o;DL-O~}^)UDtZr^b%agd^C=JuN?wP-M2y1N^3(r`ZDR!6R2YK$+OhB$+tn?vg81
zjx;RUKNmp~YB$@$f<5KjOt2j*qhVK&Z<u!8Ga<igfC(C*9t<Y}OT0o@L}_gsV{#p!
zf77cnD#DRi77-X6gv37GLE~}59zl}Is$~2GFm&b#8Kl(#Q+QbJZzH8A$rChhTf4qo
z(o>+$g+^7OyxhzJFht5cxBIfK+trqb^8*n}DZHEdtC@vH8l<P^?swhB|M4FUU|rVZ
z;W$mx015)@x-xUEMYYd|IZT)PUFNACf9D<!7Du#gJJczhpFcf+{M)COr;DklgSrQG
zwLTb*!zDCtq_kZu49fMX>ByUjKo%Y5$}9<%xQAIO=|I3)z(>K02!yE;Q9431k_iK3
zZD(LugT&604vVab0D%}|3>bcaG?ofx6o}kRzLGIxh~+Gk-aPX}$70_-;FfEFfBl!i
z!;^Z!B8ho(5zc48HA$kh3j)VVc|fK%v~%f;ZJzGl?nxb%TBQ~-Q`P91;SiK0BOzfC
zF4aXeT;P=;9^?TJ&Q8qCtS7q~IkXalqN}=@FkHK})utL#)wXWSb-9*p>)ZWyILuX!
zQ!VSdx_7{M2wXx+Rr76Yy>~gxe}!c(q-qrXP=ZZEx4z!M%sA9TDbsQ3{_*37%XR6N
z$yOK|mB8!n>t}k2h_IzmBIo5At3C{W)Saq@(O^Q=x2{`jTNPn8?~e};zxnO|c3Zbi
zz1d3*^C-bQ6$wrwywD<|lAuIn=ASz}7%WnV!UIw!770)(RZ0n<tE-2Ue_EzG)jEmf
ze#KdWNQi{sY0<a4`UY4$l6F#w%V?AL+D7|$I3C6fZ_Xxm6oNo7INg-QS&1+<M@afI
z3ri6Pm28AFMS=!3<@Kz{ZGh!n3$JD8ZVo`qkx2dc&j!N!`aX7>v+#&+V2<MBPahv8
zpK-H&tupoqnNF4ZVT^ofe?~I>9DohPbGNo1uDB6Ixmg&P=)i<tPoja3(zqvy^OyAA
zZ<p)+dhcuV2g_1JpshDGV-YTeq||xFeOcD^@^Ud9eQ!;-lUmHu=u=ZzS4#P=oBsRg
zl|>4HOl?~>1c<4%Mno*~;isS0+uge!4~KG?>*+{{Zcb|KUTgjMVBz_9|JU!oe0te7
zZQZoPEG%N{>V(nE0uf~HNl)Ju%#kACEZ#scb4j*iWQ1l!1UyJ~MtgF|Mn^x^L;%cf
z+14sjL`ITKfJf8d2^U5FE6X;(nW+kwA_5>n&Gy)`ua{;%0Y400<!MBV0B)wL&1_Yd
zu|5GMf15DQ=Tl#oRiu_uMbfd=S}XG;L_LB$h^R2nr_<?lc2AkNuIpMEG|Mcta1i@Z
z$h6;=>+`3NQm2PE@9O#e?Y@0}S^oUt(_O>T<b?M6P82^IZj&zt73M19=7vF?{$_B_
zdx2j6nTEDt3L*{DZe>0?Ve9?z{V#8;UN?BZe}yu!^xo^l^E8=iVVO#$$kZWiTU(c<
zEh{{-nFRor0Bx;<S$nkB%$<ejd7kHKp6Vo$(%T~B5gf*z^)~~2Ai{+yPzKygRCY~N
zqytAbx#6B?%h=iuUQc!=1m0`ffGv@-BPM3Hecee+RG7d>!*QhF>NrX~zD^GmPyiN8
ze=+{^2s6t^2{iIl@@vR9GM278huwqBp&o38*9Gd=i!|(sGL|HAjUEF-$a&iVXG#Ws
zlFri3EF#iR2ZRxbGGNhs3Jv=%VE=*wlDZl)a~IZbx9e?NHvp&Ok(k=HK0iM%>uT<0
znmO6c)1l7u!^4BAfpEXyw{1HfkEhe|e{eW>c$)rx9l(AjdL@V+9v<gucJt?#mvFx?
z_tyJ-n2+c4uYUaT(}xfD%PoqmX5?0<$@*sMY_Tov(}yp=|DS)jyxdx|*4z!|ut5bG
zBUXSS*kA9g{OWRT=y@_b6BVIJ3J<)|^Cwx-wqP8-c{K)r4FLOoNjuyKWvXRAf02iy
zUn-11B!E#cT*!-*sg$Wmn5ns{#fZ%rhtRh``H-`Vq?1%`mcLHBIYfA1EY&i=kb@He
zC6YZYPH1d~(rtjLltfB%wJ@lq38$cldYE}v>$!x*SU4mytM1K0M~~#`DJ~wA&riTM
zlMXV9-lHqlEw&Zf$(45b*timEf3DG5Po6?46C(<PDah4(69Sm0Dyt0AU#`NLAS}nj
z>3ll7$FeNzvce6fq>YSbc4m*ZZMT=(<>hj{z5T_3>#QV~r9I#7mt|8f3fZ@2{FTG@
zb7D8S=uVb46Kpuy1;(yz=UR^+&TXWylwi&@jgqb!BK3T(4{zjnjZ1e6e*uLV5W-YM
zTnb7s&`Gs7-|t@=t8+49n1e{&(OU9zSftb<A|ygABy5|RX@rJ*q$L(G+WOZEvLoHQ
z&*IJ*vlL-CMUaL`#y{<X#T_30cVG9IO7kRP-r;`V*Z*K4j@{pJP#(V4-0_I=$Wq#A
zGDy-e_A}kToduHa3B>L`f0pDI^HB3qbia9*DJzNn*g}A4xb4&KT*QC~0J5ZQAcKI?
z+$&3e_LJd}7GDTAb4#iR3Ge#s{UZ^B*aK?5EbaOE<@w9=uYUYxN+q&B0mRIi4B9Pf
zsq^9Z_AQ8LU=nVvwX|knPAyeB*Zph>Qv{qJ9;%cue|mY!chGx3e?LCHdH3$efBX#-
zedLc@Z};AO+s?JN-gRx%<K^@1FMs~<;cuT>v({ZT%y-iw<`Dus$WUY7kxxxT?orHC
z-E#)zDqKZUUKnn~!X%{>;#BsBxvTF>mK{Q+9q(`ioad=9!%cT&o`#`LZf@bZ`J4*Z
zDn+E5nOW~jQe=OAe_q@9Q4R(I!%~U|;JUY-`%OpeZb%d`Eq!S`Ai2v5@BrY_bT51;
zwxwXuhW4&uciIC%y_;$8s;ZXHb=tbd0Fl(wQ#atj5fKUVp;&}^H<jkJxo(@S3$`X*
z*_>neJtY<%xf4(|)kLVKQiVtWVF(4niZGXADm9nX;prd9f4MxYb1l=>`*L5_<$gZQ
z<UKm5d!-I<%k}wwdD+$lET_kJhlh79=(@J++7+Y@qi`CuhC!*=;eem%Mf=7#XPUh^
zM%IFHkb`1qDF(hklv(O9gpZoaG*vkq<aFkF<1ixG8GBJhqLhTAQtNM~DdSh`y}Jd4
z)F8|3r*2Bbe<PKs);&rj%)Q%?e;}Mf_R{7;D9J$_GiiE(g)lf>&p<@k9jD|wJoB7v
zcYw)ecP!GzS1|r5!o!Iuou?uq8FVq8BPLHWBz$jr21q+j$Yh#vzPHz<*WP_o?8`P!
zO%=v!G#p7dj&a_=b|;@#aRC4iN*Y>L1Ok{Xt9+W>e}kNW!n?t9l7PUf<axyG?M_w@
z3(fI5cXC85K5X#;jG2>JLI78b*7eKh=bwK1_>cErpZVxXF%kiDou;-doAyohd^}I3
z9*@TzJIe^R*tU&`N-d?7pfH%4e%C(sJEH{6s8k1CF4tO00BT<5S;K$#yZ^hbtGe+t
z(Nq<2f4SXt>-U%I(}&N0`opKc{ORMm_TF7JlTn9>fsG`{aSA&?EJTu1s%K&8TMEBx
znQCQW9+G1u@RA7}k|>!2>Y+IR(AUWzuU?L;(^P)-t6x&s^?F&BEh1a>9a_V9vWIYE
zY)zBx9x^Uw_APE^dzuGc%e^FTpHNB>PKwbWe>t*9gg4cu%B51R<CA!}sZ0}RzOs9!
zl)NfO@>xR8p|4Ng*Y#QC=z~CVGxALh8-$%V9}e;W7a|Hb>xgiwlWo?YF1Oo`b=8Xi
z03ZNKL_t(FwpQGWjOCjT01Qadn~tq{mOt&itxH?)#U>ET)bID}bT|^+w2=s*2m>TV
zf8=~Vms(YQTUKq2!U~7?)p(QBxv!UYd%j;U7I=L3{x|Rc?&19}jp|QN*XR3s+nST?
z>bafcf`0?Z{#>t=4yn2JFpV&L(@f=Y5BK50MS(r`(#%z%X7_EqF1OSD+JpIUJpbZ{
zU(oUXa_{$*5kk%qWfFsH7>IQ1>$co)f0yNcTer1aSI>eb4wnJ+<z1F;srx89i1MR}
zO6nGKm^p>heq0gA!x08V+J1{<vTKGyyh8FxOvtT|0~k&fxoX-y0rv5mk&D<!66k2r
zUJru2d4khsTrdD`+VSl1ONEJvUYVx89~<NA9-Vt02Cp9nyziJToP2O%0C19tf74Jt
zU^YNpq8KsbY4=8hE>8fHr*lpO1(6#gLM`?_Jq9p<Ftb4Lh!_bCCnsVmJeZ+r4457+
z?zmp=pFce>>q><!T+M@tg;PR2JQD3nfs|1k4~KW}-ql(ev<JE&y%YvR{Ci&C>2*`9
zh)i|*^^d=HcXc~FoZtQOhdRxd9m_>nWS(W7w%%`-t9EUd<@u+lzy9HGpFUikpYE6C
zM*%l~<6-*Y{hJ^E;g|pAKmXJHcKOpEbzN8Y>=Xx6kH@lKM#yg}w$(jA%#<?*_9gIG
z0qj;e$q~yLIPYK{jNcqfMOeG5>X5BQCXZ82rMz3*Wv$QgU`Tksy|fX7Y?Jo-=AKxU
zB}5-A;|oOu1q75fOdh`TEttR#fOwh?-rd$UyWQ@UXcDd@2{`7Tz(Gu6t~nPBbf3x1
zmizsQ`FK7bS$IC&mwNz3<S<W3=F4xU)@eSRq?gc00TzG!=KbIQcv<!7a{K&pyKU(Y
zP2cXjUTZ!5oSRqxrd@mMZBy;;if;;i($)uvgRzX<(O^zCItvH9n{K*!Imr2NF7x5N
ztWVe5$Iq9~A5ELV46LRat?9O`xBKmRy<N4f;R*x<M8s7j9TS9WCZ!Ga21-d=Cm<CM
zl4m`HNoap(xC27~w1d23YNi2z(ZBP0CeFsocNB?8(piIpeawiXl^!(Yq<*?$9DKv^
zJ823YVHSBu?|4X#q#58d7@Qs<V<Yx&n)=-R#`uQ42osSoJKW*0)GQA^+kX4=SO<JV
zl%RCK+-anU-TN5hEny_lMu3<hGQw-I|NK`dHc@{g<QW6b87(tM#^l5))HRQ-_1kTI
z`f{C)$2w13i<<>;?$xq+>#l`)p6A|rTHGAx0}=J!w{`V!cj<11WN3fKZTx*0{WMKQ
z%3(hI`o|xyx7)I=sB$<R&F%JbndW&q9Z~CL`Et45w%hXj;mgOreEQoTKQ1>->7RMn
zzyW_>PwB5``fgb#j&Mslqr#mj*A)q@1Cjn(sn+BGc|=#u6^aFQ4@>8;d>}n<bqM3#
zo5%nB+u!`R|Kqp6`uiV#`su^V^Ow(Gp41Xcq>O4~rWhtkNrH5>42MYB|0^vaUhkek
zLmJP2xY=j|fCwH-d`cXj0%7K0L4Z7pNML^i%SjIY!x*0O$)-eOf!A8U&p;e7Gt-=V
zcBw0bfnnhko{p6?vrvRaWc6WT<Y~ZVu2bvQw<fJ0j)&tk7v{|COg5&FOm)_-1k0sV
z0$knNx~!%a#5^CWS#PbDDolstA)hX${o>V$SX|X@6S68O)DM-<6AtW+qaNnd!=rz4
zrTJV>51)Vf<6nPzx~#f(wSE1Ri#%kl2ax)*@3@J-`_+4@kDFPKw1C5qbnF+zU#mPu
z*ul=U%6x)aq%a5&;yTTzd9IKBe4D__<!QUE+BXMxx7ORT-fq{Y+vU04FWxr}AGv>$
zTIt->5DR#S1R+9Gk&)u;E`?1w?FfG)n`c5YLr2w&*ZGtF9hgnK!Fm9>;UB1U7V6g<
zALhH#Wsuw=lJ`lJ9@vo{@kE?V>HI@2yaFlV#%s6-WkSg?Vd9Zh%da%H)E3b|%CaW9
zWCz5O3qH@Fg+YYJS0jP}<=T$g@J^1(ZI<r^2>Is6V(HZ)KRKYhe<(O2|Ac=S*#6$4
zIpBa3`REs8?4vX6$U?g>_wD0PPj7#Ce|R{Td5+%ALiP1?A^@r7;o))H)~VL1)M=g&
zNL{445tH`b_o3}Sej9%uMlYohNh$SwI&WLM*v-PqR7K=)Iz7I7SLXTNHjz@ebnvI2
zK7aVj$NROftC~gx`Jjz?ByfMrV|_(L=8@#n1R0WZ6=r~^Dzz5jF(;du4N5E}4G|WR
zx8{&??qhYX&zUe6dORI}``aJ?`+xth|LuSO?fw2zE4_XD@b2B4>%FOB7%&MH7IRPK
zRM=~p40gzZgr(HdT7RwbU(q2NFig^R2YbvlJ=4I{nsrr=fcqW@S;l{^DLjpWNr<Iz
zgj&yPqN+)gpgiUPF2%<VHj^o}cU4D_P()!N=G~((3K10&sTMT`)i=y~DO0Vr<ke+@
z$f-_;I;k5=>E3SlWu6YxVLBX+tu<2*n7gK<xH|}L?%mYV;Gve}{<^2|SRzQIa;X-u
zzP0UY?N*?Z@I$R9@d<xg(Tu}|d7h@}tUNtk?;k!t|McZqLwvxtxyIPh`7E_LBz-5R
z_&@#%g!WB)b5pn@47(`io4<LqcmYCcSvA$_Qg|x$?VGpn-oJl%cvv7?P;+0`ySw^a
z<#34Q-j>BZ)UCH|dHJ$DecoQ4^nOEY>@Gwo1*LEm-j9>mJGy^T%2dldm9)Us?n)!a
zF4reKns-8410|UqiS4~+2<K>wJXHME$ZT)eBf`yl?_-1=SVuB82^njXUl<LY0RcSL
z)Wpe^3P<kG%wjlQXL}s*Acni;#_IJD`WCF;FG(s$%!DOG<dwvudo#`Znw)keoyWkb
z2qSLD)m>H1J<ET4B2~<^^YF*HQ)C<ruvA^;RLmjC`{6XX$?dX5s3WHZCluz&lDuel
zZ<}6TZh!vMUmt(*rW~g-&ESI8)@92?*@&p6rqub6CL!}wO3Bn&W|kuJJaZ9h+uFD9
zx|6>Tx|wPfDYcgHs4|`Aa|=zvg?pUN4~K`dUoNM$-=BZ3Pd`09f4*E_*0zQ=OM~SU
zNnxKEM=$`VA}C#o)RQ?Aj<I;7Z}b|Gt-ia_aAS7|8DJO!zp-1z&_D}9h)$EdeRKT#
zU%vm%uYdWA_vcc<?9;^OhvN@Fyfc4(e!fu<Q6?~tLgN0~pdllMhvbSWXUe=ocxZC-
zm%mPsqoIEud*(z8Q@9Ocmf_q7!oC^9STY511~x=;G(E|bqu)o2-EUFb#!;V{W*EC9
z+M$&K=pLO+0E1=s*z*X$OXgQvCo_Q`YME=9=V@kXreNs?cc^-R>NJBW+%mN}-1e~*
z9#M*%PRFTMDMCz4L%Abmnx^A)I8M`af4OhBXV!n^bSQ6TK2;QGSZtd?FdYxacMtbY
z&zJjOKD|6$Z<{93mR?=a139RV#EpTQ{CqHdylU3Rjs`It<%WCX4i4ny3rc0Nwcf!q
z6#=1JXQnp~kCSusST}p#HUX9MX+A$p%Wd5j+t$9V+x@zJ`n-MqqR-E?uH023iZF?c
zM2~;rqX%YG;dv_aJWZtzvavhCDU{%asmQ>ev$@S14+GuZccA<K><`Vb=fmKT^1iHu
zhi7Ewt1H;P!%ac#Ncr(RN?!ZqaZn5d{&gD_gE9?7r2pV}f}HjwrPQCp%(B*#S;m#;
zN_8Kqhm@}h59!l>5*kqViop~qf-rY=v&4Uq!UJIehGXC0j<vyfm@P*D5puAHr=MKr
zLK~-iE7v2LkcZ@5axMS>86g+B=fk%3r!Oy`K7KjAJ->PXCI&MpM5Icn8Z2Dh)YEV%
z_pz2jm<VC1Qza*LV{PAcBYz)sQ%VtLBDB^KG?!UW1Kz!xrx&V}QVVfgw!i-AZ<l|k
zWm`>Kj7Y5UNTwkezM;e>%P?h_K5mRij|(F3zIht-oZSw7h*=Hn;ItkYFx+5oS>>oq
zha$gtGylV{-v8r2{rKar-k(m@eFd~uK0h4Z>$}T!dHVdsBpElK@Ll+zP7*-GP7Xvq
zKLK>rK$==-H{fZ-31mQuyjo`OMpA#VQ)&{+7;w*iv(oDWmIp+PeHtSn&RirhVKmDu
zL}r%RRfCYqLW2X6yAB9ab2*q{xqiqtE%#Ew#HlDTrSK{&Od?W`C+Zq32EJeKt@o{W
zmQsmSdnHlry=@u8o(4>Hnog(FG}RHfo&W>$G|ltzG#?Ix2-|$S9oXK?)0=-o5i)j1
zM6-T491oA@`SJMl>Gt{Z@`u0vZP{MEONU-&Jmj7%i*YdSG>M<<rP}=eq3u0-Wjm5}
zu`rprOu222Bimi^3eQa5K>Gi{12orE0<P)XhUWNfs+P&jh|nN%)#e2RsE+^^2YtvU
z_pVx%CgO|F=yt@n6voZO78!rwI+$&5Rn4l(S}Nc`%z(Bmb1kKhDN$#^R$ZU10CDI;
z9Qu$tCYD<3yws<2eR#BoCpu55Ry9)uaSb3bwQ>gtAw*6w_9=BKG83qoY1L{{Oe&fX
zLEvV+WLFa!06>Dx>~1@WH!u?5YFlQok>57Z+BuYtGJto=j)XWwwUK{Nx5<Tj4C}Aq
zO|SZqT$5*w1EbnF#ReP6?0Vk36!!je7@L9#p#2aLb$~rET~BMZMldq~uWCX-Bv!@C
zyw+Npq@=l`H2Evh_OUl9zKAqoG6EAQp`iQ1s?x?Z^$q1bCNmGm05l;a1*PU0*t&wM
zm-GDia6Ud>x^X}-sfvGqT}+H&wblJK+#{u7vSG6u$2Nc7aKitDx&42>zRdzbrIuC7
zngfREdCf(}<Ddj+$bo<U`KR#t^Ux2U-+%n&??1o)@%?;J6_VDey?`6uZaZRAst~wf
z3gOD9H-I)2^%@!j5g}x)s^UQ*<{D=AOK9^XAfyloq@o&`rHFsKF5c|N@8913=}*7=
z<3If&j;SAGNG#_{0eAZ$SL?bE7+MiTSgSa}#$y7R$pDxTjhNAQN|owl1#b2TDE^F`
z4y5D~RFUf-%aGiElGtr>e9R>hLi-jxPQ>R)M8J%O+qLCzxq@unu34_l0bz56LTp?D
z0(5Qy7&3a$gsOkowga8MU;w72njoM`_2{)CnIw@J5D`ZXip(XK$EPQcfM1riOC5(;
zYb9g?Bf>7G6g*-q0Fh6XyDs%z@5UVjuujh;>%PN7>W^dU0xa`7bgA!Ch_<^q^@n{`
zy)5&2o^lqEmN~JlZSUBwH6j2Q0x?mH{F@#1=60=3%yEB(ZtTiOr}mCa(WjT@qI{KR
zM1fMvI?u~A9dD1J>nYSC;j*lGS+k4a5v5irHB{-cXi*Xn4nzP3#XOV<195UaMZj_B
zQi_B?-s8F<V69TJlx!l5=yMeHcvEctmeeTZjh+kka#LDkk<}*urYlC@%4dMhyvz)d
zjK~__(FlK$m;y7RuXz#oKfTu7h(rNk>%v?{`0}Iuy#SaA)GIXSMG=5=5jg~|joRzZ
zdtzpxX_L{#L<sKbO<*8ZOqH3sE~c&*#d%qM-r&|z?K^IyPxSO@hZxM0L{YsEZL)Ae
z&vX%010g^%1Z<xa!PJ^>wlx_SxKW&^WtrhJ=dyoRDN=Lw6Fjj0<v!9DIM7xkmXc#&
z1nj%cMGlwCWp#{J?YF|{0YFt8h5Yj60Wrp|3n@NNPfML+3aJ~If@XR8^7#1qDKF|V
zf!m^TMBa#q7^}EPz&888mJA}$9<^KV`p^Xt@ue$7A9`Gup!#S80T>~IA_|iRvb&r8
z{q2A8_5IDwX}r5Xj)#8A%hPkdELf!?QHx!Tv5V=&s~EO%3$%)d2m=M@eE1gL*j;X^
zfeHrZJ|67-v3n^wx2R2pbo;w)0`{N5eQX2g6hjEob{JGuFEGH<Ul9O+iDKN$d0HeC
zINjL|6x>l6?CN3pGIsMPrq&X30E{UZgByRyR25dYxv>@mpP~WOQl@$KLL;d9b|CUD
z!L({ErR0+Oo)MwVUt>&>kh_$aFl$W=p-)3JB%m1Xk8k(8u}ciL>`r@(;o&?#U#4l!
zqQ0}xv^}p4EZ)xgR&KT--f#C(tzZV5w$I;)pPh*ED->jt%Y%6@SVg5u1v8W9r-y$J
z?|<3t?uXq0`o5g#Gd_Ksre!+M=kt1=WX@0|0pIq!VJMT>rIb}_EAJ3FaEfW@yVOOF
z*!PJEt!Ar=pd2`0$ON^tA!LKzJ&ks&jkPKM!9N0Q=}2wX;ej(ux>f0BCT66ff)=ce
zY1<<9rq_p2NKD(f9O#9A_A>MJhq-^v7TW2r?P-~+nS&LLHr>VtKI(1X1Oj+VdUZ(H
zWT_7A0y=C7SND4Fuw$gVyVK!#GN4i_e0{pit29Ll0=RRInJNJ)ZCezPnOh5^>Oc|`
ziHp<SZq3w;=*qmp1}VCAg7du5!>`;1NKLBEmu0!EkyEa#nFbCfYNEl9Bc^{q&b2Oi
zZKtDiLGrpT%d*z2wiz-1!>@0oSB5}HwdljwX@Bgxu_vZ^nWpt}OsApi5cuKg>FcM5
z$1l%XN!<<iIy!S8hLD&*tJ`}u3Zrg*+2*k`euHnLb@#)nRlyaJ&3yo_6pW3h7|;SJ
zqox46F21`ze)sOxo45DlZa9D4-wgY{&bpSR<eFkcuv#rJr5L&xO*YiG?Ob2$0APfa
z0uk6+d`4`<o^cP@I9uQtRJ*i5wNWhXddO-!yYV`1Xd;pA#YE05VrHrWZmcTc3Pv?W
zb?nbcZ`jm_K)Qu6_?^U!t!$!AlhRsgZx-5p<5}Cy9?pOiOx(`}O`3nFp8{Hl!Lip`
zs;e2Qh_7)KLo{G)nl`D}{6U-N789qKQVeK9rW`nhE(VK{#^e4EfBGjzQ(3fx(4|%C
z)A{mzndhaNF&KjZm^jnTU_<{53|pm-0L?r2w|lv!vX%4gtatFj3(p3Dw=M1q)1=C#
zb}*^+a(@2!;r-jUf9ijSn_V1U?}Ldh&rf~e6`>j=AfOl%^e37Xg6LWwRkcbLB4p$k
zLKjlsbwk1k6d6pk8WF@0V@yQME3fO~i_IHmFDJbv#ceWATP%3P4BM5706<J^tl%tN
zfCkNju&S!#D}d(Jm~FnZZt3CW!$yI6DPP)CY`e8hcCY2AZSH@;szB%_MlU&5h;Dj`
zwsC%3RpRVsFf$`RbFr08Uh>&y`C!B}4)Ol(bpQGlQOvohS*bc_xt3sU=3pl1et*X8
zxv`mnI)nyD6o`d3?vk2$CeS98bDD!1G_QJWkUhd?&4z$tm-Bpio+3w)3TPpOA}fLh
z#x77mG%Yk;^+S{I*B&DTBsN0nHb9BdNH@}rbR!+o2t0tq=nxrQ(lA1#HcCKHY3Z22
z2LWM(h~xy`eSdiWh37f<eeQFeD=dh!A%LRu35tFk`%Lg8%S<7l`aSUPl)r&_Xr;hX
z$;Pb9suYV`jI&r64G4WE&naR)MScX_{5?rO!Us-dfZIg#j^=kz(JHhz>*7AHoIT4a
zo8i4<oK~!h1ZCM4?q?&{4s1=uf5CD!DY4PnilkV-Uq5>BD(b6%+dbdcK*%WuSlJw}
zWyc+jtm%#Zz4#pk+NnNyCM#^=L>_JB^-H3=WHNe{ZHXbZNm)@{-Qpu@)Yta<7iI7A
zHQ=c$*p8f;2TZIByuTyYZ9?;?jJjm%OwYZQp`S=5z+&Cg|MEnm!<99=zA8gW%gU5~
zIMBkv{=mHF-p1FPLt>yzigz8A8R5Lc__iPz<yH1ap+^PTmSbf)`LX_8gBoJmihu)<
zjl4y#U3c6b%SZkj?Th$#DbKa^QCAOfIWstOndJ|gz8^k3t`gD65r4>jHE<j6gSEc}
zy4xB1JVn-#&)xf$ZAb<dhlRP66s)x)(q}{wABrc_@n(8#@dcaXktxTg_9gq{y$!Zk
z8(Lwvd4|?G`p%-Sg{-QQWu8x4C`JQ>s#Qaf=Unq3It6!<TBlbG3D-ht*QF!@+jw8$
zZW4EYCU7f53_OYnO`efKR+(hH0udq%ey>8|QeNk+l(?9ZNI&-i`Fuf@uZFkgF;f(j
zw!Ds)=|wW1NE<eL;Z9w1AXz)><APM;kWDauU?}$|#~Cb&_l^8Z_KxpORbK(!&jIyR
z3u#6x{kjl%hXif~J$Q|Q1->De<(ugixNJxg#CvEH({q6=v2@I@zc${N?lg~MN7sSj
z@b3)vX)*=>nJ(?MW9aCg66sAVKa(#m>63fo!12-;W+BXFs?7Uf$JzSxQ>qAw0DRVo
zg{RS01>Gz2rmzkmoD#59^#Mq<1Uz#5BzzU<?H|?(>uHhRH@BC=kGdyo?0v;r-o|?B
zX5zg$`MiuYGjj<v6g{-yzPDtFvZc+=DHmdX&I4J|xOhc+A`Yrw-n%N%9Ku1-hU;b>
zIQej3jr|JKB9CGBInurMR(VLMs{b%o(8%T`7a9n}{R_GIeOJrP1phqiu;kW-CniGX
zh;>qCeeVj*@e#`g%`1o#Y|Ib1o($Pb<Z?Q6-ir_v)0HkKuz4zB(tba$tPCp3M8gN;
zw)1;oYDr?stb-CuFkluJd-%qi9!?-SYW}c0@`Vds^%Gz0dnG6F2GTnH!uy}f4Bz9?
zgi@0o`<fO7{;upcO<+TzMJr|OM?zIUk)etQF$ZiGGEz{0J6gl^zBbaSpRY4<GWggw
z`(UBXwbo8e*Y=Oj&*oy0d?xCQjLT=I7nId4sXHy7)VvlXE(UM!toYuOcnuOm8orv4
zYL*Vy$lKW&84;{5kQkZ4M&*J?28`G*JJ)(LS41@$d1ySIn*x(Fj^Q^nQ(%R?J1k*X
z2Qtu8dS1@TFTBF_mG&BgQ>oG>Pq?lnW#oz9hZU`Yt=^w57Z5JF{0%8m_wGRW6Mvbs
z;7|jLuHgN&VMHLK9VVtE;B5WOc?jP7xR8PbUn@RBvYukS_S_E{zMSd#p;6Zf;%E}r
z#sU&M_y2^#0lU9O>;$r6jhPQgp{e~tM9`~V>7ujX;Ogs0hWN$v-MQ;y?8408;EZU~
z44?U5cYn-p0uy71$2iI?TrLj;QXm~WjM+%6@XCscRdHo|xhMw{my_dihx!q)8b4MQ
z*ttaX-M;egU=L!gp?P1D_~K3VMd0QlwX6ste<xHNn3)Z<UW<clhc;F&b`Wbvh^SG|
zd}#DIz!50;+kixyx?HBgh`^hx%j4?4zLH6E;`}AE72d0Q%yI@L=yz8cv+`?arre$<
z$C+30Bl%xW9fZ_Qeo`yI_rI}Iz+;ymt)|<vT#OeK1g~L~)p0YDZ1IbJ+9}am{GfY;
z@(T@sq4H0^g>-c9kx_93%F)ut#hrXRzek4vuHv*lpI=8#Ugf<)37R-L!(!UZNWI_v
zr!T?*>Anc^)UPbLC}@Lcf!iTwwI<U|_XDbY?RdZ|@qbKMcwAYs<BJA{Kb#r8l&<#0
zGOs2KAJKsw9vVJie-9BhLX<OkQvGpoKO6=Q+ci@c-0>146L*17^Mx*p&rzI2;@YqZ
zI*C#Fz(`|qnL_XM^QPeH0<Jk^y24$dyQSd7QVnj`jU3{ZQzeopS54QH^`)f4h`;<q
zu|6SUR#WE5HjSPXF9JGN#`%c*M}DQU8+!0pHjE}k9R`)t7Z>;0RP9)pxi}e#+rI^1
zhWLoN&8T?Ss?m?deQE3?k^vE=u-K0)TKD6ge{xr_vwN5Z<02vcX*b5eMOYE}YXlZl
zG`APR+5V|^hNCQ6D|N{w1S!^@yzX2Cx(H!}j*~X;BHJxNCMI-r`Jrjld_9eb94Fo=
z!SZS>_o+nH3jN`q!~3w^_+$c*S03O)w2=IbA_FQlWvbGSTacI=<`)MO4*2b-RbTJA
z$fxiI7Cu`(!r{<^gV)F5y}3H5Y<Ii(WukZz@=Crk57XtGB_5OUPb7;S)q|BvIxA`L
zpNkD8%yKe9!Ls)rsh5`Jl;a2DnYBnZl>B~hPojf@!$WSafvmYPw;v{0lyT!ID$0Bz
zm_IdkX})sSM?~?W>J3|FObxQFS%~>(clgb}t;<~&q2i#9ogKF#=P(;BQ~JkkVTD2;
zAAGN&Z8q>Iv?YxdNUtlmR_7$)3Hq*7;M$J(<VZjdyk^oHxfx$UqdS8A76LaFpw)ro
z%FkG-LaotXA}E}T8Jp>f1YUwx*n=jiX>()Bv-gxGOpYTOz+gU06&Kw0bS{W6IVJ|`
z;EM}a;(Enyv^r74f3@vA?&!$a1zx5wkPv?)Qm;W$FZHID#=&WZDOYKaz^c{Q|7%wr
zgE=j*t4PW+C_cY<Mxv0M{~?t`-7-7cQ9t@kO(J6y%#C6y_&H^6-3jTumF74bl!~-p
zK>?~P=3;Ul^7g|7VrOTU_W&iSBJ`hJmE7L5xAY%TIZBzKtDpT6tna{rSf98ydZc{e
zj30tgDX3}>F;Q%d=_mZUE`G$L^gxu0qgkOC(6}im`um?cSG3=gtAA)82$)e4cpqn-
zhobmi`{+X>n4`|F261^QEIWa9W?@iJaJSw;S#@SxGbMOCQH?U<V6T)8{Z@}Id}lY^
zIA7=#$@PDMKJMf&asJ8(HxjhoQ(CCDUr3j(H}NoV-H-+it9)@?Ry#x5df9s>?K#l{
zt6}mc1{^f>dSU^Ft5r2o+26aLy2QsuO9^dMZF?vin$#*0GzE?YetP3LZ<E*gaftiM
zLV}2EVAjs-{^QQQn&U$#97a+H=CWurFZ0BZl_!4&@lHj`v!tW1?KN2aK`t7J-n!_c
zAFB<B*5`8;+gx)~@0Dhmtb>$Aqa{25J-{c@nTzAalnZ!EYhYP*i}&Xq8e{Mpfwet2
zva=q&lmV_T@W?)T#!a1-lSa6jLG1nu1o`iae`BL8=zu%{sh8(yUsId)Quk5Om5vVV
zz2*Pk?-%}E3z)JrptNTB%D_AUa^cXgpP0GY7(|@+;bWl+@k9syN&sFHEC;$VUwp4g
zz+&j(0veJzRFl{cdBepVIj#FOKH!sD0Qp#L9EzYki-0ggaqVQ>@BHfYU}F)}i<0W?
zMOK;b{kx!3df{Pyk_3;?<|J}J_4F_fV|LQ1C(~(Z=Pb0KVhtu^YsGo_np#jlSleq9
zPJLio)`xUXrLK+(Fj$P4hl{_}LIijj(k39rX9bBna~hDp;-6S3j_2rkh!Iux^H!pq
zA#g?xl81i09nsi?ye{ZaX-mpD-Rzbdk0F#0oMY&a!b!3y%;JtT)G}Wno*au#JV$>@
zg^nM;K!g);2pcvJ5O8qW^XJ?*uMP|=uIBvA0U@Oim<t3r<5p&N`)+-1(JK0191~XK
zxGrb#?YKHYN|5Dk;!}^JYO8vwj-zi7;YE1J`lE(VYNPOIN7AuA)}BA#zMtcTrsOQc
z^J{NxKO+B#5Sl<*3@hk3g~c9Ctc%JQRgZ+qj?8j9>>KW75#6&?_26wa9=qhq;r-<M
z%dg{Y0rL{T{&KqOXQ5fWOS$RZjB}QAn;|Z;TTu>Vu*V%I+EB&}d%CACoEt4pMfQDO
zy(51~e*8?&vz5W$N6^d%QVz{*V_~8-<)UCAWmE*&3{*E>jKsh0)Aw!&EI7Eo^w2bT
zb2CV^RsGmAiC@H@_7%|p>ArPEr%MXFleKj_KPw)%E5$#Ok?W9zZ??*syfj90sBBfD
zVOURuzh%ynasS^989~)#cjBmf{aBdFVw$@vJrlTwSNdHxuaMTJa~-d^{Ef?oLMJ0?
zDZU98uhL>@SG<tWq<6FiB(bs|o{^#8Ab6<tyhT?sOf3z%MAqO%br}7M>%W5=ACqZj
z1;DgSf%OrDx2uMfAa%3%-l%zRc-Y0?W9&18Va2RP(h0_rj{^1Bb_hn$$Ix@4$(yz*
zP-X01l+)IEtthhio9L4lmGT#jXdw<MA6iASr=Y*u)z;^GbGx{;K3SH80*Q`l+{(@<
z&uF=rpvS{v?mM7Cdy37)K({`57Cni;cfhsx!ny26wG5UhqOc{J32XqyaB<&zxlEf#
zP4hpG87(M=obd?LD8QMfclou<oh84uN37Vao+e@vVnJd28H(F`Bnk7#bRpX@ZvSFY
zz#Ou<;(PZ;^!#fk&J0>kdyi>oc+xiKYzY5IVY;<&?l1#)LhoHr5wK|Snc_1@2beyY
z><2R}kYo8~oI}H2Bs5ky(HYd=17}+INn_S-kNa-!@BZ$V`Eq$ztx*iP`eSb)KtiHn
z(%|&9vX;lw5jOCy`$NQixzFkY{n@L-#}Zp+?Sw?_p7!@omP}T**DM_yIY6a#&x!k5
zBvrf-V8k=MlNskorg-K1xZMAm*MLtS?kNiMRpvO`<UHb7DvFnul^tmgdi)6zO~>s}
zD>dm+Jin7r?VNA?cY8asLmsTED|-h!{5=&Hn7@!uY0<oOij<=6j4lJoUAwV2jx$fQ
z$}t@om=IGM&_S|GNgfjQh84K4tS^=pZ#!(m+Xz?E{FcsqEfBkA<X+vn%fQeZayhRA
zuVk;gi%3c)PR6q9dnF~_@?kg>gdh(sa2Vf<b%4{M;VBMD+MA8(D$d0fbRyg_BPYUJ
zn(j%Z>-OBGE5`4SyJsP(v$9h%C3h`H4k1m-f0*OnEPCd(T$7$S@e(L}=+4>M8O5EK
zNIIx44VK-huj<-%*W|5=1V~|>?=tF4-Ug1RC3J3@l5P~_FZvArs;5^XC5jFy`X)SC
zcZLaGqYc7Wbg>}z^7upJ=dxU6ILIhaN8by1kBj)u&Zns81p8HXs>)fz1+ks~e)aNw
zEORnxVOa8#<!be(Eb}u?)p=btlX2)nebftFkFei_;I3R`4wXu>0m)NBu$-)H+Mw?<
zjEt;IcA-Z~b%BxQb9d)$jCb&}B16r(IgW<5dr)88YnO*-ndyX_0@oz|IgQXP+LwfH
z2?as?9n2slQul|IO3V|YqUPi7NZ1gB%T?<K+5Y9d-rA@{!xR_OiWtqLh?H7!;tsvL
zFRTp=i9cB_ppHAkfqK4N8ZdDZFQWoI_vE=K^Oanz?*kzS$fshSjxg74e2yo!lNsm7
zQd~TF(w+Tuf>*um3yZe7J5hfiZ@Ty+z6k0~f|$EvQhxjC)i*acTE#HcgEvmGr1{vB
z6YeZZH0RJ`unUXwogCFyNSc32qW5{M9+e*aAjeZc`YA~+056-17_v*bUM`<);uhQ=
zR9FrDko$PxJ*fcPpuhu=p+VQv58I<Bc<FoCJBOc9_%;YHDmJyAADJvE5wI1Q5urZD
zocsKS+#GBsj5M(a6cacaRs_CFsJ;rYnZPF(TM%V5kb~oLl~^FPYrCEDTh7z@2KOM!
z&Dh0wGMNd$N`h=(&F9mTuV1g9K=hHQOFMe6Dk67XZ|eOjNq$*m+Zx}`ymlSj_wFCd
z`ZFbro%hQmD8H;cJ_&mvK~%*{oXJE<39)npr*`assa)y(nd7W&hgEn#+iUO&iyWT4
z=d80xN=zj`x%mjZyk|r)=31h&msCYMSL$BS_K{2t*bL>pd`w8p^q%b9>cUcfG5z?L
z>wL_{yLOVQxwF()<z{bM1r^9Y@h!87nI}X+7Av+ELm!pewat!HR|k&6)D-qn#Zv1X
z_+B40^+t);qwxVgk20_G=)b9>s@i{B@;7IHL(ZhfO#U1C_vbWUEU|`9++PFL)76FU
zDOm`#0p@ph9*O-n>K(Uc?FvHJwy!@AEK5{xIgqhymirUA?{k#4c72Rtzqw5B!wKFF
zMHdbuuXr4IhXg2L0d_+y3xVU<1&p7z^}?y_$n)`}*$l1*LruSMvgzEeP&vukx#aTH
zl+{JFwqeCj;if$0?nw7?q>sUa#ewkgs4sC7fX*F{1KTuh-XVru(#H0qw4icEe>f*i
zoJCubsUH0!;nu+MHe=uBthbYHP`@W|i}WAE4^rz~LF)&PXzLlHuN2i(tk>3l{Fu+z
zny9=7s)R@ezGw7*nmp!VbCT5tv8Rv`q5?UQpJ?hKgynSI_R~kaJJMkXXZTTj;eWvY
z$SAED11*a(i-U$|lMN;!ZtLI2;>r81y>t;FArn&E3ci{FZ@>D`MywsXxcsc+AytEy
zNx5G?6da;#>v3uH&h;nU=(bjJf7e^U-hzFiq#dzFt4@pbm(xS+=&=K#f#=>Dvi=ki
zcgChpO*QRDU&ML5fZzXS+&g-gSsQEu6o`4TE0_FdR!Fa{ljLfxH6HVY&XBZLVzs2I
z`zcm*m#;Kttm2ysQ9&y1EOg!ZPnUrMi-Q=E!b8GV7a3xFA!s_24XIFcM#n@B-z5dd
z&-c?pp|*ol%Y9h0Z{ky>sR^s$G`^c=(OoYqb|Rk9tJ|s#=?18Z_|`|wKPChLr3M9v
zGi6j4rZ3FDfTq@@_Q7dH2X+L*M*l5KN?%j4O<=UU@$KxN5OZP+fs4~SXv)ntKYwr0
z)~*SYul@c0dlKPLPyhZ0rB#nkMn0atuQOG(o7;I=8bznbz^5U&DM<kviZWLHXiCKy
zWNqti-fGHkoXE^S(Km>PIFWaNp6B#mKSKydgaRu6`|wkbz3<=O%}K=0^NLg(f~bt}
z2!+?~Rj3E0!o=G+G@3o~V&h2uTD0%hH_wHKukDgj?C;ua3smX1jA4fhUS;b;BR#Mg
z86E+gV2*5n+rcB-+l_;e*IzsY>e^;QZ9Hmc{aZIShUo6+Ap%L_8?nlOh68yl3j?Dy
zQ(YCwjcE_DstC8rj!_umFpK>D6&F0O&J~`fZceKyv~h3FC3b%lG_8D?2<bNy;<EQ;
zwouYPJ&u3&-kc8PpnY?YW2jYIsd-wVFlR^(7ncxmEz~FWPnTQ>1A#mn9xRh+?bDGf
ziP9-4=r=D`V(bA_PbOfI<1AM)`B_^#^XcEj$2_KlbVw733K>5arOMZo{&E982ag;p
z-&h88<m%_^;`qqW8)4zp2o(0by12|CX*@!I);%+nAMtgUNCU1T;TBYBc7Y?|5t_E(
zBPKI58(90f)k)xedPp+u_@2nRrTxk3_AN9zT|nm({auyqu^7O)n5{Rmw~#$})1EX@
zNrj4mzJhs46cp4fqZaw0N82YoVPbuwTYx@ET7~`0pQ491I)68mo?lhd>qEk9Al>k$
zjIF><Lc)W|-n)#M<Ey5w%(MjEbV!gHPu+Fg&#8aO3||ubD7D4><>bVEyc=!6j1)*T
zhoMRO>$%)P&&q*2Jfd08LWRB!8+K0AG)fSNYqK|Izw9Z#@aB#yjv$4*#;7rXDV2jI
z4~JVE+wpzNkOGYnIwaArhYj_&HJpU!2@DD!xlc&#tTu`%`O%`)hTVL7<ayOqpzy~+
z98sPP?Zd1*DiWjWqJGlEZ^unnxGmjJpKm2O^=Ab|FagqsZz&lJli*m<P~>VlH|O&)
zQVs2~Pr!P+%<c4o#P=ul!a@nLE(zDL*qh7HO9XH|+IKSw$b?7a@8I2I*A6ZCBmo;@
zXf8)A#4!m*Gkq|QIRb8R^0%i=Cm$ps)+PsGmtJfk+6=i?0;q$h)lT*fOjj3I2iNSk
z7uW2-?dAD75Qaue<G<sBoqYVQHMq$Qk9I%?N*tb{uVr*>!ixmoGHIWnCLRRs2E2Zy
zwAnr=;mF`QsJ=6OA2~z^Ax-HlyWn3oi-ylB)aO`%`pb!Rhn)IVi3svXOB0|Md!{dU
zLkuBxNwwL9va$6^Ng5pch6x&#-##=-EdX3xAPpY(Hh0inc=Vtts7wlN!k6EYPDa`3
zmR4J2E@<*@N8ZkaH3*G|D6V(=Rj??B6k9fSL>2-6kicJFB8xU6dzuE-dOBjwvcBh-
zGOVk_FeWj<T3jKDAqMucUUA+!#fJ*Fw#H@~Xs>bg&$u`8Qgr+!4@)p^vIH>-IRB^+
zUs%~QYb0!brviFth>WtQc*+YJ^063R`|D0GD>=LW*X#C8Noo?DguEdBh;ItPXHnTD
z=Wab;n`5P#bNL5vCAPRmdEM(wBIK|x$F`XfG;)I#tPrmD@5CICl;T-T?)CT#EXSs*
zYUOnK@+g<QY2yMva>$5AEju-EeV7q%|Li@ZiVF#;-Ri&Sab7}S*{vP}_b40b*=099
zHJvMTsu4zt&B~{Ku1CK8ZPSK=cT7;+%{Ve_RrvR_ncqfipI~1OvL$|7U>^A&XnM&i
zNYhhH6>YUM7HeTK?j#C~dCKwuo8FHf_`^3PgRCX~S|ey;I34#H_$hFJw@kpG!0L|_
zi*KUi{_$;Gj820Na^|k#`l}zaEl=&oQ8WZY;?=5sP&egrZJle!<#yq*;Hu8U6}xZo
zr-$Qp4zQ|R!-aFlg)V08g}_)I>vWUgL06a24^Kmph8+|8>=6;8Li78(8}`=Q7T=}<
z+xn-T7@S|uEU@3+%!K3n{{w{8oOrQhFWnzI1)^x^CluXz!i=aO-ayIz+wIO;<ki-$
z4E;3!sEf#mpR!lE0$r#`A|70e>L)<co#Bpe=EkXSPfLK?>Z7!#G$y<JYK87bEZ!ia
z8{!9#CPWyPluXJGPCh{Nx3CY}{<xJHKWUkdf=)^LN(5Z^3IL%dyf%B@ojO>3XsN1|
zSd}(&cc(~-7sops(S=^r{GusTaa*Ip1qvG~{mEuPQVQp%fkI|!m-=sw90N{t@3mNc
zL;XZuQh3mtIQMJN<tbCxcuR@Cfoo7agcFTJ#}rbP6<n0RNF^s(jwCY1HXkKmj%QP)
z)1DjifrJ7b0UsKC?35V7)4U!@L*W=TH#m}Pns)8G$QqABuLdHdl3(QI?CzT925m*|
ze?L03?O9F@c4RW2oMe{+{z>ZGH+oRP*WTc4p>-B64yq2ckj64kttN$uad;}-<!$>p
z(6=F!bBxMefi~$nbtz5!aC_`Io^|z{W3Hgdfftky@JcpODC9VQ3l$S{e{M%Y`X$=1
zV%b2Ofh(TjE!X-d$aYFsc8v9X2MV+`n25{px{yC=_VU#P_Is*ylJzf?Hv`%)p^(m0
zjM)3&j&KXKRI7cZYcH-_kkKPSDyGb&S`F9zZmhxddD%O^ANjOVb;gPuJacF41M_Kb
zU<3r5tR(4~1L9cs5NQH<61IPdkhvX73htyi3>%bKnL{Xs5YPfE1)D1!O*qc0+Y&=3
zH4|8~#oh<)c=5v#oJ4T+i_Zcx{{jRM=FqVo`8VY;5_i`JTT{s%n`)(er3Uwd=JeqA
zogD`2e~0N`Bz+p^?H{$RS3M!0OyAiOPPgu34j!2~?pzT>xD^pwUo?IBve`+S88|*P
zHYHPdd%V!mW!4CsbdR;d1InJ5E53U7?{su<bMOQ;Yv>Uufe~+a7hKec(<RoimN*PT
zIA8z19L4F`7FXAXu*}5wZMf7h=W?~)sj*%P>J2kox7U^kPB^;xeK^kkWB-u4b?cv1
zuIs`nA77>BoIMM!F`aWL*Z606o;EQfW=Kz43t7ljO5h4mq_`dqsnJ<b6wDN;<iRt_
zN8C9)Z5K^P+CGPRtw%vHFSO?x2|RUq{W-0Wr3G%(-cRLT`tnH@=qP^nS^#~qep_Et
zy%>T`6(Dn=ihgs4?UtNc<q?VJnyryAM>#sF^hrBC=n3f6GON$oh*95)<No|^xH+Pd
zE~y1;s8$2mC7YHxD1L(167Ig?-5S$IV=k`ny9)d$DIOI%kUebQv37u(**FXdiIC}$
z#S|Ite}!x}mB{056xZf29ZU&I>8$*%TeIujXQ{3buYS~W@mqG7#0JU!>ViblR$mm@
z1Z*u`ZvT~NzLuL&Q@A4vQs2Wg6RT8YcMEu=r~#kgHCl;v1TT<X*(NFsZ{E=AZY%yV
z*S>G4$7wFA_G|W@h$+;BNGJL#vgE`XiE@dsV9fm{vlYsRj=0a206__y5{~H3yKO>p
zTOC=2M8$-ixRzFTdzj*a*rcD#s3XVaN5_yI`9Cuuj4%#5F;(aTgH4^O>234g*)LEp
z1K_QA-PxN+%YOxp^j|5(J1+O<wcoUfK2g>Wmr{skAaFJMA>Lu>u$K_ox5k>DP*C8w
zpf2{c3zF$bQaK-Jy)&v>r~h+s96GC6JL61}t8JF4$);V#v3-0vZ)f1~0$S7TCbFT<
z^RTs^EUlK;+eOiGWVTrIeYz5Nc5N<QZZUuuovGa3Q%}6peNjRNCk6YilSZ*fDwIB)
zz{NZN^O~XrS7d|ZW;VoqC9LP=Z!dk)ZHwD1YMG}iFyftzZG2opoFXuPP3r~f+N1iV
z%}@(}HQgvrjumB?M5F!CmqM2Omxd;^4~C*LxG0rn3Ww1h9Y;5R8>K1roUd>=cA)HA
zFZ1sklQnAVJ<`jwvvziphlgnI;_W8eUOaM;BSpA8j^P&(ainNzEVt;Gt^~uSSvg7*
zm=^dLxH~3dSnPZ`L833c!NP)tbCY2{q`BdyGkn2DYAMmVgJ2!36Q4~>QeED?cjY-+
z?+g{4G-|cnp3wE5+=hz~fsMK600!_|5pL3c^V4=O+ux%0M5HW*?;h06<-bo@8>5Fi
z;LA2>WF;~fG<5Rs5`Q{nY^kapwTnU)14f$HpAlA0DT0Qg7NLhhoAnY+x4Q|qTjM2>
zAx{I(PdSHuPp;0+f}4W-Ucbg+&Lz%$)b`5Bz7&rBnMwcOL+e%x?ob7k>x<>Mo5*w}
zHKa1pHQkZzEdI2-wY<nyZs;+N6u<Veq4538N!Jm5M@dW8>*5XpIR|0F6Nj=YDSRYk
z%+o;zM{*6Kn)3Bjc(#%s&rr@@T6|Qrl4iu-aK;sv!^2!~<35#m<xfopR7fi`7W*1@
zb7B0D%=F}$?R+SCt+Ei{vb{JmYo>dnO>9BRRZhc1tCC*pRzE?})UNf5*Wp#$eCM~A
zQuFVGASlOsSfVj>YRXiqz0+^`BV{ZJuc(=QpkH?q0i%>&3xUMhlbWAQN1K#ww+@D?
zGiPiPX39i=1G>cVZZ(-B*N48(_S4Y}jw#TL#K#EXNv+yIo19PpzX4T?Wl-v4H?H{k
z!79gs|D|<KQsgGoL_}C()+;X-87)%Z7V-=a`*S9M)^oF7<9=-ytk>LbPdf|7-`cF_
z{U^Lt-86}}B+-n>ve5c!s->dTrjiK<XWUIhWrGwKYi0W=)$~;GzLtFgqbU63yfv7B
zCxqSnMfCl<*ac1&9w4~Hl53d2?V$TG|AOrPPm{Tj<WD;zJAylx8^@tYFbgk{32b^E
z;3j$bjaF4jeQ>&;|KnHB{Y~$+(}crA%=PX8f9kQH6e9oA{^54e!l^eGb=2W9ox|$E
z86F+ZTF7rjN%zXi;F%*2pB?B~a0stqirM;PWaVw9fPR&6XYz51xl2J!4fhXa1+MYr
zIUdOGO4RwFp+T;2VaHgNukYoi!7XMQ(TUNN&*_blJ~^|t&FR>EE437Kbb0JvxtFh{
zL6&$up}xlAYi)!aZ)p3KQ=;3)m)X7R`cK@2L(QRV@jRkK{?O+TH>ksMrdn5HU3Ga<
zt%ujD0J@u<X!Jq4BU1v+^YqtXylxi#CgRG8-94pKo&tXhpP%854x^83t--XED$3f4
zugfQdayj=t+T8p*x?QMFmrwrLn{F*%6IWd=tcQ;fBb*ZV@%Tj-=3Gn8=Os?=TFUmx
z#3Qs84iwzC282+9C~t3u0yI$+0byvWR<8{Z$Q(y#Xd_}rZ-0(7X|-r|^(u`;<$VN`
zmS@?oGbq)mY0Vj`Fmckj)8jL<p6$>}K~6B^nOQO~3saC|&Y?LiJL~3x=vhiBo_8^>
z+7O0WVl$4$Gdmtd*M-`=q$s|ShRZ;4a!>(wDNx#&H=2u<@s65X&a|b62Pkdvb`Kz6
ziY6YY`?eq9D}wc334;1PoFw@+s5b7of8X;M=>=)-=^%l7cD`j~60VY~68M}QkBRae
z&8l|e!}DAnAAy&ZwOMI~A*k%DDb+N)r<zh&VWht)Jtq-~G7$-#$IpT%`u27eDK(a3
zEDm<>=BIpgNEAHsEFy=s9(eaQeF6|c<Im(L_1mahlDj8dxVpAgicE_<V010029fY5
zGSr|o`1Ih8AT?2U<LMv{ADH<U@7ll~dAYTDDTClMv0Tize0wpdem;F@NUF_}p)}!)
zU+PpE6FBG+Br?#YuF8R;xC~rXWixpBBu7Bg4*w0>%dDnsK+c=eFypzkfLhSZiFe0J
zAj`H}F@jM_e|@$xk|n|K;?L=*b6y5-1w4i?&}QCZ{>#Qk=}dxLHJ5Y=zbYN&X>m1q
z(OYuYNQ$cd#r}sT2C7a7WTLyBQhBQNHugtoFau%PNoKy_+ph1Er-Nq9k`T_BiOU?g
z2{P$kX5C)->zD68((C=A14K`52AWtB%q+Fg`~m@<GtM?nU*ZWmna?Cfa0}IDJ_&=A
z!Pa??44!!86$@5w2Xt{Px{4iMW#?YuNxA8@PaL;U4SLW07Ap%12FG_nbmtLuS!Ume
zAJC}8k}9~CUtaz@UAz5<7uZLT5rDEY#(b?$KK#tO3-YOc+#j_18u<8DihvH3&(FbC
zu8CZEe#_j`-s<GYy=<D+hW!j>F8NCIAIL7MuEwM;#*b9R-Jz0>ux4*++(NT4%Y;6Y
z$!~~Gn<0hQ_ZOWj{fp!nb7+l27hf5DCT~+MR#^kxG6IKLez9nA&%-{08QATivQYv<
z#-mD}F#Qov&n%=9pePoQM;^02oa-uCzo-@&O2A%7w#nA8G6jZ8>bqvglI+~>2oAuW
zbMF|J)<r`>8QwuGr9q!1OX;6drmnGd=C{>wHII|5GU>oAL${dUQD!p)<678X^E4P%
z&X0=foa|SJlRuA1tbY!Y{N2fp#2W1>{d$!XYEfPN+&QxfDEY80@SmIjXi7tkDpqmk
zx^>i<k%{@sZp;e95YCAH&iH!LqNRt9N1L<2WcuW&bEXklqDozMU1`e^sAc|{&gar3
zlV#!j-rZBmI}ylJ;AZl6^AsD*OLcpE1Vmh1g+%sRZA&fuDLFO<sn-zF_-og)-%&AD
zFtDuG^)UAb$Y8?KN`zx7ED<$2py=rSFS7L}1VpNd97I2kP&1ef-@XL`^d?BQWc1%P
z#w`USB1=N$u5U6)O>Mo_7k2Fe$Q)9=rrOv<Okz>1tE+aU`C2w+qN0ja-D?L6Cl>zZ
zRkb-_Q05!GukO5t9y3y&J8emL249d+bKT0=k_tedq=BSyNRM?OZwp<j1;a?HvcX_8
z0>GPUQMy?`kihbw^vLmNe-bHQvE`AB)tl`M!e%c9CFJ+^JjR}%UeEj!?OKzzrW&+B
zElhmEe)4SIJ~ln*(9_1X1CVi`Apz38&4HIMQ2g!A2<Bj88&jYAA0)KzDD1B4@a_lf
zHjsZc6<8PpBglx5lAp+Z68xU`{mYHrZf#q3xtpz~+xgF7k`)$zWCW$~UgbtMCrv4I
zWnSBAn`f)VI<geoat1E0Rgd)ntzX-5ul|7Jy?B!D$t!3?E2*#AYm|P5L_O6)R<$HY
zC`DN~!Cd){4Ce!ja-5m2xZ|kuW(mFA0|BreoQ&5ruCk^U`T12G*DwLJEW0jSx}rB3
zo!hBw<xHbL9@sSmnc`5<wI+x0!CETQf>__ie^?iO&f8`z@&?7*axLJMC=%bvXpTOX
zFX!a&8Lm#Nxd?*8O6#974MN0gP?X=JQ>2=4MYT03@2LV2^?#Z0juxRS=&Ha85cIm{
z{&;h!a;>AJw!+wpiab33D?y=pWwV*RZmvl$9Z%X3)|z9uqLYdp;vtH|`<lAupwi;Y
zm`666OLk8Gh+6l`Alh;)w&JXkqBxMBeIuy50Pzj+j9}`A>F?-$tbDn*>jeKxRZ=0U
z80M$@fBJyu`aY*|YZev=FlXzPk&TRC7#7_}?Fv6fTK@_n)aT&NN8HUE9=bW{4P*og
z*sn1dbZ8Df!Xy0V>f=P0vDeZpSKGH?C4#ez_mB**uvuGZQGE?~zD}bsY?9921<v0~
z2x|Wo1{PI4%AST~PdSipMs@;tzua>l>Lxw0rHyefmhaA}pMaQUrp2gpac<#lp5#O2
za2a{C8OWP*PRBW=l<F175>#VyiClQMOc6e8Hh<^eg^aA6-id__S@+T|L5M1Qj7a>)
z9247TX9r0>*Y3&TuOKYMq^Z29X{TQ_Xmoqum~V2Gr-5tKloD0SxYD``r;6G`1M3qD
z4-?wOhXw`oOh9R74LAeGpO=)s5@W8MMoJpZEdsqm#Y>(Aj#7dem2ciO_jO@Pyl0v>
zGb&u@dt_uwN{T;blZrzQ71Uyj_JsI7Nke;NBITNPxM!$=kn?AUwnc~XvXLl_gM<h;
zrh<N0k}VxKqqCP)qwy85qgk=NJ!=|s&_{IT7$$H6iR)W3TW#V$?FvS3>r6;M&D(ar
z3dqL-UVAR={nrt8d1kCY?A}MQl+nMFNf*T+ML^79YC<H&L*;5WBg1+oq2j4^Fw&be
zp~X9~UMzsKCM$D%s35LD>zj}vN+EH6ZtgBNrbR;Y^P2&r^5@p^v6Rr!Ij)!8up$Fo
z&5*1Mm>{-z0~3`NPOTRzv$J9$i-r`N5NV3<28WNELca<(#U#6$)>?weN<;TG=#5AZ
zB`DY9TAA|vdva$J|NeT&e$yi=C?emOXI^1`M}pzyd%hScjKY|LIH6FjcZ|+Wl*Ic!
zWWY3{TiE~#^%e1D7a+WF<yT`0j@aWOrh8ir>|6?4yH9jToO7xOQfbznbo)}A2y;8i
z|E<*;D?HoWJ;l_IoNR4$|3qnycR?<2ghX~j3^BYlc<9K)iKJg`z|fBBdg^GHbH~a9
zb`5mUv~QR1>qsz_GG$wp6bs;sXG;OM>0LZIw#q8a{?ea<`^a&qk-9uJ?ncMsVg*=z
zLb-`(qGtaA6jv{ukcyLiT(F{2cBCj&(&lxJ<1*FETd-2K4C12Y9v)BmM#<fMc_Jn4
zm0Nn7N9OSJ+Fk}zboc{<>F&Z{SStq65>2SGl4>_Oy~=EieczZCQmm3LkM~s$>StSi
z-Cp5Keg3uz1}R?+{8Yc?tg0;b7;w;fr<aa6F&x8GGZ%adl{4FU7*%fu{pehsDxj#*
zuuX2lF2~K7OJ2j(j*LlV3QA?*Q6lF=y^piys~L`BQF7pU*yJu=J~A(d8)6`X(wD+(
zcJF02E5f66FR3ZC`|+4K^l|X45v+VFb!78!+eMPaQG(+If1sRgj$2kBKvtR{P&j;e
z(=1;^6(P?cl55VZ|HM7CZRPgxw0;4%l7`zs%|0_)q?SNmx4hx3EWmT4Sza_WT>jY|
zon)84hW~8N$WxjKH1{;u6yUr15H=UZq*`|TVLFc_mHT@?-6PrBoA^6D-a4^*iE~dR
z?G#9o6s9MGIY=tRbJW2?fP-+e*h>TFDh7wOQwNe;YA>I8C-o23y9Zw}-2Z7yIXs@-
z+M0Ees7w&gbIQuQxh-+^NF2ZSbyTS=YD{KFC6@ZNk8B8wl^hK<y{Tyq@e`2SBW{`1
z1skm8WBC5Orp+sJWtJB-9Az-yH{OLQ_QB66PsApY`8e+LKojB17JvtSrLOVO76DO8
zg7JJ>4fhgz>2sGuSc%Dj$-xrbO2bKu|6%2i>e_O){l(>)f^6#F{b5}giLi8canXR!
z7u@&$lg5bRV+_s((-08GyvIE4j_s+@T&3*x8g;T&T(1(s<;1rkT?b|ZH^15e$e<YQ
zHMym6fN*Wj_~`Nq(0OoZs>@=ydhm#>M|hy1HZM>;J8xlmJ`mZKN0?WDHtDfyeA6yk
zp)u%N$R;P#h!B(Dce&i!uN%Me0u^=njP=lTJkheQ+?A#j)523ml_{{oJ9FMbWAefI
zv@^k}4Yy=<Us#iT`4>)vZC|Cp1jUH3pjq!Gk^%nTN2{E`?FBb@BK`7LU$52e0ivno
z_CQOM@cl~3+EZqlQo)}vjTBggsO>z~V%|{ta~4_0ENuu~+JI7N<$LH=X2@nCCsURP
z0j)}zpLY7uA-lXxNio~-3|){L10Km6+&y&C!3+6uKISVk=pEpC2N*|Ea7Fwq(MsFY
zKfxP3PXY*kI<^ray7>={p|G~=V<BC1zK7RhlzgAfiQ%wrAt(5b5pnQp|KKdIbcane
z|Jug323^LB82tH;`pi%carIb+JQX?(_gqV_FUZfAhQ?($Xxp*sEUXq1kivwt)c4V7
z!eqNSo0gar-0;<HSykIykjTMVE9q*L7w_MB0NJY$E#>YZQ1jU2Q|5y-^vrdDWF<yd
z$@RfbXz2A7l0P}>c}2qJ#qsXp&iN$%*%kY$YiyiSk??C|uuKuvdG)4!pBbXzF?IE|
z-Vxi{ZS26XYYGeTmszay%~jaV=`(=)2glWbvcSQsS}F8Af8?=siCF)S2uwBP5c6jN
zSbGbKQdmPIulX_5uWbI6C7P9GkD)6pe-Ao~>(vX)hLcfzg1o+;DlnP^QE_@g$f>Bw
zWxRrFJQ3RePY9CK8m6xx<kL?V?8rhh^4KIXL@TH@w>CCop<Mur#abr$F~s|2xLC~9
zKXVmVAk=+6?B=TF)O5J)UYR|d4Zb4*h-EZr*b2<`j#q|yVGh;$f083NpRoERUu;rt
zyDvIMV0g6`k0f}iZ0`SHB#k0E+;f+GmNX^7Q1CD41mQy<Ahz<?{bc+@uWA<>G5u@O
zbo)r8wYWIurX&_ZsAu$S|546i%N;Tq-M8XTbeQP|ycPfCa={)nfB&}i>R|R=6%Zdy
ztrsjW<~r`)hRyflKV*eBoAVOBX^9ACc;?x~d8eO5hY&<HxL$>vM+$`(0Fqq7$=;MH
zR6jyjIle77Y|kf|6ArJFC(XOpi;2)c7wRgRpt1cOBO7-b*<bt`KVZK-F0t+F$tfR*
z-*>6giP)qJGg~oA8Sso+Xe&m@06Xs6V>FK1$_{F3b}G98(h|SOYF8fLA%IBhP$>8N
zF{LH!?(QD@+{~=?-u^9%QC3Ky8LnyT4ZjZSv6;(`TbtOi1#n*0^M=^tkmZ9QSfvGi
zJ9BbMqsqp1xF#!KGP#4tW&qsP>E%XV;vb7f#OaA()9pVWTLG=<x)ccTIyKFWx>GpQ
zE6*GpmHO^Tf$ge%RE27>T{kTIai$bG0rx#4G4<+$nq*`x0g5}bc?jZ=v1p+zV!wMY
zRMdHoP0p*X&5gIMZ957!+`B}>1u~ndn%7#vDWdCpx_9k{Nr~cnI$6kKg<|6!;`he<
zSb`Glmq#QaS?>I!fpS1!Vc(5QD27;$EeQDc_b?sbFt2`7Mdv)&iIIMuX~%(A0t|X*
zLxwl)Z!f%2uanT5AEp0Do}Q`j5YgPRM4ZT9r{6|KUJIvuU_P4uH0u%Fu~4XGjt@>C
zfFw$Y`FkW|b1n}b_wYJLk?!j@MubpbNSAwihyUp%Q+&mtP5@Bo5(t6fbVC%CK*R@)
z(KjuGFDADeNOg$Dv*QvU3@Km~!1I)cD;05HUN5Rh4oMzX*)#2o&Q0ef3UM*OwBQ~S
zTjDN-LFk@+BGJr}s>f(+0x3cKfRbm}rDoVM|4KlPcw2}fku=j0X2UN)nd4I5!*Alj
z8_6x@^(gOV6A)Ti=KkpMvyMD|Y{jJ-jb9=VOvU3k9fA+wSa0FZu1(f<({nkz!NL01
zp@EOio*n6cbVWY%DQ+W&+*f(fGWsDG?Pm^0&6!-HvY*}A*lqz34~<IUSI1JJVLujS
z(mK?3pgE$uaa~^viLA@3Pd~>|1OH}0lc8}<O+rmT+p^c4w4T=^V;0roSKA6=Dq{6W
zq(@`r5Xa=CUaW7ImB7N2PUgpdrPx5>O_PMd=ff6HrjrEMepFqD$gsM*)nzPLX+d37
zc-vw*;9sNN>qoNyjh@R))}G6fo4@s@+zOo!y07n1?HNefr@`ju=MCq#J-oK39630L
z=?G~7QUdM}k5Y2)*-XnQk}&D?^NyB6wjg#nyp7An;qJ!9=41&~@W?2N&SQchmQ>rr
z5+Oy@QY9!Th!hL9T?k|tZq(lnoSQckR6J!H^zfKKU;|l(VW}j)gSu``N+K^ufGdRa
z`1g^g*FTH{O?a;4ge4?YW|72Dw0UG}P!}L^Af4R%uJ>!crPEjGhE{Br#w%tXvetlG
zioZ{T*=6n3u_1j$9|Jr42+#87T;#DU8FeJ<W1Xii;8H?v(l7;Tq7+@yr(>M<4B0W_
zLc|K>ZuxloY)k|TNw7y~Pj_TbZ)99fx)DoI@TB@pa{7^+th@|{)z)UBDmCsq<P3P8
zW)Or+bPyu5&p%5~4`v7clZ2}GBfA)&tajE2gpcjbU%}e|IRI~iyGao3-cCGcZFxHX
zZaFQQ2qfCyi&)d3^qukWUQyf===Vy!lzkAEaASQsOC%+n0V1hbXMimF>^vH<Y4Bc8
z^0sqD&RPB8gWsbwwrqNlLOwQ_0emn&(a~C&?@#;vFempvNu5E-Grd<L`$0+-LkwAY
zG|#EMIKt>&Weo8i#_v1mPxqt@KBH`Zylfhn7&|DmA0(~bb#Kk=Sp$j=N!5+V8M5vw
z?6Cj+{XIv2&%VN2YUQa?yL0Wctw~(Z_<lzRm^Lri!{SJD&0TR9O`TdB%m~ynNd^$!
zRO$TqL)shB#qVBlv4pw1t-(y{KMMZd{>k^G&dXXD#{GeG;LKe$(T_o0MC|vnyCVsa
z7xH7ihhG$3yst*}4mx4&Q1b~=Y{sD{k-X-vq`-GHQCqT5#P9VV#{S3@Nvhotnf+ii
z{wfq4btnj(`-pKtNmx$1tN<Ps(21SPFmsI1=gdw#ucRQ8;m$+*NBfOAXK!9@QEN)r
zPT~(2Wcrh_rI)v*<l9_3<@PTtfl`8@2cq{fIXH<D2b35{^9uVlpHxv2kTSMscAp1n
z-lHkAzb_vic7tC?9Nq3U-M;Eu;kKWPcO}In#IJvEJIeizYUm5c<p6kYdHeiFH~zp%
zv8D$ka@x3)Ew4NTe)}Tsr9zir&ns0$_g!e`pCEAcPyS}7k0{GcgO|AKN3z96m4L-s
zEhCo!_Q(A!dBl4Ky-%L|M_yx7>5?kCvy<<uOg-rRo^w5=`tmfkq0J5ZM#I<3j`9n7
zaK-{_6ZhlZ;o<ULl@XBN;Zo^l^0r@lb)|=_FcBQeqX&Mj&ye?+%8})qp;~L+WNuYz
zNK_X+czn1Kt>oI4ja}JVX~JGQOZ<)u4wm@u1oJy!nRPhYWC!bPE|RQKVK2dw_WGeX
zPj>&{;~5?_>*%k3mv#3JLX@?DPl^2P?kFBrS-uQmGSvLFUk427P<r^QgOxykycLwd
zds0P|kGJ|es;^z>F~*?3-j=r*0q(9#-CsDRbe<@sFGU&2q>364f-X~a54&wj%eL+i
zUDl?=I^>a)mfJe#mom22+8;UEtL}9fFNmP=ZcNpQo%eZZ3HTVel4Jdr%U-7_zF4)k
z)KXO9cn>jz(*e<hs4W=fX7l_znT*$ZUn}iHLP8!>E{C#`tR1Y>GA5;9!;108P;goN
z*Hv)VanO$p0d(+ng<!a~DkqWQT=Ax;i4zUE8tX5ex<+K)63LEwDb*Q-q(W^LzS_aV
zkFmPE#{B5JIAlHAW+uL%`o1<sIhFX$Vr5&Dd!22a9w6gX%TXAZGq#QviWhCd-n-so
z#1`&23sgNF6*zFZe^`7ZxPV19v>BVAKm8)RK&xxK&N~$s1^ZOh$m+(!s;WouA)vtZ
zpSXreHmf%1anRlKqsRgwX7Lw0Vok)+dLz;fBBQ^k?8YRAV~2W92cXX*_Ze`GgNqDr
z;Q_Db0C9S(I4xIVt+=&T!O2RYwA8!22k;6=0z+(uf=}SvIS*nH2O7>+HjL;!ZNd`W
z8C?2g$<6PZYW%|I-`|a!UHoSF{Om03GN{*PQuPUhQYifLv^AeKJveRo)z=dl3M0Xy
z>)q<~L80E>-oxQ(lO{xRMl3_F<PKQFjiKKQ_<H$j+_txSzN#8Mh#MKvJ1S0h{P(Zv
zR;I5PPa_R`^fX$8V|pC*15ndmUwm#*5ZEs*b{}71=5`NN#O`v9dk`upJ#^xvIkYJ*
zZ%b!wGu`8SCe*sUjdU|<((y3KAl0yzKn418iqPJ@<HG5(-x^7h&dE|0vKq(fU`JyQ
zB)+A1U1|JRlL)T+6n%Z;E5p$_*CwzK9ueKnq81i*UKn09q?_Q!JNgGZ7CUM|U;y?p
znS#|k(_}bHxZbrdq*7`@RJ`W|EBjc%dTea^BCkL9&_uVLUHuUsnodP7JwsgPqq#?$
z&6ZC;4e2DS`>yH!=_r~SPv~B{iF|q&xOtB#TfLF{^$Q=P>cB(7^Xg@lAB%2+wOtMN
z^Z;vxAa2Nu5fgtfQ}xZ$`c}=wA!h#1=@n`W@#=+fddiCL9@<bvo43kMf>lo)>2<a=
zH<qYc<+q@;?*q43`S6tYn|Fs6Oe(4C!#oJSd=&daRe_OHGDb>tF3>t($)mCjps?eN
zA3ScBE@I37QF3TwYqK-DuhZt$wL}^CtsxcDgE;jALfVw7{{=`@BCt`kiYfc<ReL=t
z<W_J>kSErDc--N~8e?beVal4<zbcZd>$2a>K&1b{S_%!ERApOk;Ta;QU|j0H&mL;!
zfU0`)Cd604-^bl!yf%kSMrRYy|I<CTFsf=YZLP!*{@3EeyoWV~rz8~)jX^b7&~^8d
zoUFpzK@x?>saU&hB(4EExgZta0U1E$JZ>4`(K;Wxq)a)L9b811ae<z8^>laRbqz8U
z;a!pbBEimT+!Fdnk}uD4FA}_Q>>tbqyjj4CJo~?$kIoEKcnH?5R)Bnu1LFOsoI@Nm
z${I9=b|i%UilA1Pq(<u_D1iz&43glN6da6BFDbtLjVO`3`Df37_XB4z<Wll&)urZr
za)<3dSi)33C1;M+d?*1S5d~)oC)G?W2#eZJcs`B@-4)4O&gmHKYS$bBRZuNBU!Qi|
zC%m|rLF0kAzEBx9pmhEO>w`T-iD?il<4y0PMGHJK;W&n>+Orlpj@&2UQvZ&xf8Q^;
zEGv3~4j-zF@x;<y4c4rzd|z#Va+aY@I-!99%%nn4!ZLA8P^<UU9W(7i?<&gGHj)9@
zXB=)5oiMoRrZO6<tI_`)CO|2@nL^~ufEVp=bUym%EVlz-pUgY6!zc$g{pg6}Nl5)C
zkU4b!W!<GQ61_)|V}dkc2B)Jpy#Y-(@Q9@>HKc_*e-|}@7+ndMr43DTV{!cY;G06!
zNubSk<xdA@f^`zjed$YN7vwhl)=LzVdI@KK1CE@GV=Edsm^}KA=DbnV!KsXM7$aJw
zi^I(jSOd1b7G2FR*KL%@hlV;*h&Z!^M(AdQXvm@(v0_dB^;?rvLyg`XoG&!|I@VbA
zMgncL1U@I_arH~4)s_8}v2D7JML0;ZP;Cl95-Jsu^hx$lC?Jlr_5m?=52zRtV6k>C
z*R|%`s&mSuZiW9x({=w-`M>{TZw@IV;t*x;J+qRL%yMjUijz&Y8(AT%?2*hITaLXI
z%6=P1wv2;gc0!!+y+1#E|AXT^j@NZv&*`YA>kBAe=EQjXUhwYBm_g#WfhbWs_x!k(
zv%euuH*@6{5m@rL)tH+6vU%F9`Wg4whvB0MfK-#4`<7u%*SU*3NBd)HDdJs6TQ%*K
zf1>K0EX(!TwaEMg3kos=*jI5{r}=&VetBs$YP8zCx;4EN0p~*^5i@31no&c2eW^ST
zUb+tG6_@mj+mtt%$$1I^7K9Kw#=3>Dyx_VYbR+q)DE?Ky?{OZk2t!SSC8wpr3p(e9
zpL=pFw%D>3*<t;#^NWVC0QQ)r6mo<=-Xxw^QKBK-tvH{nEg5+Au$${f`+u|eWrI!v
z5pAoWXyq8vPb;auy~~mEfjpGvG$LCn{0xvHhR_UzyBN4o!j6EY%eBivZ1+`e_unIY
z?B(TJh^kaVJGR3WNy=@o%nrR%zQk(crZa-UmsTn0au-4Su6(QJy~n_3EeqY~_npT$
zw}uZakr85g-2PU3u}<bAM)L7Gh|FDO0)LgJPfzezX}bH1vWKF6#g%@vSMLVDj?7e{
zDt^w2l1{`H0qRRP`za^fp3Tv@A`bKn42Eb0^xSL>y)@%qK73A;WyQN!(r?&#4{_^w
ztf)kUlX5HR@#<(!wo*hPgFwOKLH|RLU0hpRx)inRcBw=>VHT_0%0nY{?yoz5gnPLQ
z<Yi#sFeX#8@HmUVH&p$}<h&^Uq(PWc+(5$4qB7kH(4?pLwTwp%<|h)?r8$WoRxXuJ
z2J5iJP0w%smbF`fTWR+6JIovQpP&3!Ea_33>1+!}9-OM|FHF^Khb91c*^^_NOxIP5
zUw^TK8)qHz%cSO{JzWMcV<`K;(Z9b({{pT(0p-2&juF2=x=6B!R953S&2jmY+>ZiL
z;w6p1f}#2(txszW4m~@<B+GGXVd;Fps*zRZ0?2oCcG--*N;nyJ)qQlkdEQ%K9@YMM
z3KAz!nxzA#KBX-*s1FloX!I;CM?J#JAIsDAi}M5t1h(4F1{%lZfxG;abTn(yATO`v
z9L?Nb(yC##!e`6MQfvpotl{us_ICqh<y~Nsk(oJM={}5H;vvp3W7{=mxpi<sWRd7j
zqoLy~{Snlu+>)Ek%B=4khj8xH7d7Mnr9q7C=+=BqDZ=NT-jw`4`K<?KD7rR(imE^>
zWUDY9u0M)yL>+XFRiN0_DIGwTCA(6HmgBXH;^UOjX|$!=H;)3>2cD6K(f8}n0MM><
zzXuMV8t$kfxNw->UDXhBZ!*~)v%m0nYjZD)M>hr9+IDq@5iTT?K*9-<n_DydOP3?<
zE;(-S@>Emj3zrbZy{FQCNJDg!D`Z`~!nE?W4N6f@_bW$CO$H5bVMVTqJ+U8hzV}Z=
z94}+^eM0EP(ee2b0qd84$OU>RRJ)WG_c3jv-8F%h3-S&9_;ph{B=zyyfvpvC9$w+@
zUY+h|$aLCn<K~J$3N#Aq3&cT9{_bxgtr9AKihC!L{yEPazx$7L{D42rg#8K!R@J_V
z0x>zYq#XQJOYqs%`M;p6(|yBU!<qbxp=rfrj(1HHx3gML%WgM}8S?|4+jpom%qy#T
zOzbV(RMws5dw8Wt!ck4|V3W9syU=GoQ>=Ug_PW!ug6@?)U!&;BM%Wce%UD0(iy#y_
z0$Kd2{q7xVcNof@*mW|*A1k?)W>{o2)trY|-lj|YrJ7h?KKMaH<coQE(_ZAtexK_1
z`>c!b^=5SF)d~JYF%}9qaVXq<KyIM_`XOH#HxsB|{Tn`1jdH}&c*maID1I}JUuY<p
z`5|4#N-WRx-=-w?XH90QZw*@MkN@-L>h?&vIpSO4&62Iv9@{IUn@yKjhqn^QAd9yy
z4Bx9e^3cQfKzRecf8}!D4)}MZtmV#kZNcJ=wtx%ia`uz04t)haJ*}2{t4qts2;$)A
zTZ@KgrBxD!29cq=3k<^Kog(+j?!q9BZhphFyfwI>&M}3jq_v~i(#EZDVOZm9QJi%~
zrG+LpqXWD7kdT)aHI>54;rSdFZBHEv@frA!-NGnk+O>eWz){-A+xnt42M1QXBTXcJ
zex~1SGq*V3wCw}CXuOpH;}TRI{n|9FdBEQa#t4f5E8kD<a1ujynFzAQZI><8YQe94
zBu`I<W~SqcNj_W21gqFxpJduG_>*XhPs@Kj$*Y`=9gNLNBXy}B!_R=Ra-kKajWHvM
z_9$efR@}Cq%|&f;Z0DAa4hf%}I2x{xtX{J2BIULr2ee6X;Rtd6yAt6=HE`{_apFXh
zU+P>Q4)~c~SfpcQ9R2S_I$L&!h6bLWUi=LR36fWsE<;n;Qc~rd$d|a4ps%-2oc`z$
z`-}S_Ytqu9Sf|WTH$CoV-G@LV^zG?|wN|NTN2ZMn&tA9CO?fJFYJw{9y)&=?sN?$R
z&%ZyOKx^0jDVVKs>6^9HVT-MVTFq>>3HLi1Mr-MoIiRW~y~l>zs$A!%v{qA)+EUD3
zQPAg!pLiNAe)y?8CdemA>=kf*)UR}p2~U)#J^mZ}J_@R(Vb2ey^>^GU?-!+aakpGr
zO!S;`fXLp^sAb<aX^nh;qbN6b9ieid?C19o@O|CgYq8Y(_EDBqb+xn`nj(vmj>y0Y
zAc3U?%kh5F@Hm=OrQn`(7h&W<+{-zgYRoNjdEP8B&=11joM11OT(pD%KIHsjztr^r
zM;Z-P_%6Sn-2OppXh7YVidyG$u_x1ii#3?xVH|GL+xP_fR-Te4P1oQqi5~5vq}YDo
z_{wW;pKc;(+cM>mB3nmOgK#G9UsDYR{?yZsrB(~q5UFq5`2lW)yv;bVG*j&`{v*Kd
z>{!;PQ-SrG_HGyY3*oS}^5No<%tX7+XQiQ}*h_-{IH+m^{^7d+f!`|;{PJRfg*dt4
zdyeZw4rX4j;ypS4BZ<?NWK{EPjw>L#OOFfs9=N>!m@-oLU?aR)_UXMMum1&FhP2v_
zvrNsFYoXC5l?vi%Ib!-_tr`DmOy3r}74?btDMT|d^43opJ<>D;Qp(3)N8A^`wG1JS
zcq7NJYdipR5wT&6npdyR=@VI`PC}H#$iav`i@X0|WLqMIP9bJoOTV^?=YTenc26&w
zzb9;Br}ILKPbe<)HRut`pPcQRd+4~KFq(2$Sqk~RfKIa+yY8-#4%I~*5hd0yJG0eA
zi6hyCHIfpBhG#k-C{>Sr2iJm1Tmy#@1-*Bdt$(a$j?OKIH;^KSiFwy*1+n1{c>NtJ
zZuN&B!XOXi4malL3pD(N02Cg+s6zcl!7F9r`y#3deRv|mt6sk3^4+g%41?<!X*D;U
zGQp?4U~-@mc2s%uQ!4X%4fjww=*8Fn0{>r303ohDm@#Bb8tiXW%X)7sB}B)4Aa^Fx
zKPvaBkgkeNkUD1Q{xNu}mK$-EyC~ja@b`~h5?xy2qK{nyHXH5@xUe%2-n3?-_sS|Z
zO)^5lLP@r@OgW}lHJ+!X4NH#CsqCR!g9T<99cXzpDCs_BidN&R_=+vrW7Mwx#bSof
zLptM*Qhvz~^sj|$<&WI4jCu_pUTZX(^6bnA!<>%js(GA3!ket3&ogHC9x9S)I4&~#
zP(1XFb^Q?qs1VIrkz~Kj#=vtbJ}H?V=KXhtWlUo;`{KI%YF*U@N%|NvqM}GF;K^R@
zAQ$>~r$4Nl_p5Wkq)L={nK-={`>%eN*loM6&aHXo6cca8M*{o)Q(?NAti_9mCdJkc
zd_u&j5HXNQI>BLT?R;v$O<%msL2w3YsS)1~+MWREvca>%!-9gld=%0?PA@cQBbhlI
zY_QGEwHO?73}ewCGXtlr+q(5!a?l!S>~`>YJAReiqSxwP?P7`RUPfMZ5_^+s_n|@k
zA!**qPFh`R>?lWy)^pbXrlzLsu8zg7*3_;qr}`zRYVbEltkD?q(?^e`gve&ec2#-o
zDJ6hRbpOOC8l8i5DuZX{z3|@`9TmNK;nx3f1PuH?K7VAT(%t2@?=<fq(^(s{D3Pf+
zj#C!<N*o8%P5ksgiWj0mFL02Sg4H(|N&V6Ghg#p^Ra5nkJ}7b1f0T)Anj0S?(NU|q
zT4;@G`hnYbvw9R$D%u^JUaZLWBOi+K9+3m;M}{A_Yld!fQLpkRocoVIFCSnpbWNuC
zWwEzu|2Y49n@lkowPqeYVJG01L!VvP<;7`G>li*~CL#)4iKqYj=ze5-b<Y(&qbN3<
zl17OkM$b4~*5FV6c(OQF)z-%LuN|!2$)1PnPWC<V5#hu7SyZE3M+pOOnmH+%A{2om
z9a}z`SL6qZwV?c=i1tCG?jaL7IE<9|h!%-g5X*3(=#0R8vEj`Q$Shho^IQWwGgviE
z`5;%l1sh!TPl-R2jgH1?a3nD#6Z((6ZM?lw-HBTbDgVVB8$17DrD_D-s2{KY;@guC
zKdz{$sj9S8QtI~vs=i;ns<Zk{229LcE(|B+a-GHYU)5RHI$OSp%KK80v~GJBQ-qd*
zY}1mZtq!Hpc)QQn;FU`nNAOsj9}ClTOn9wWEv~^k#qq%4K?%A~eh)ORJRc3T#4<nU
zC|^4M^Uw48_X6hnJo);gi;pR(x9P)%QChNtI+<%?NnYLv%h7B|R<Mvfz{ALGU;a2_
zceckc%h6zDF}Pfj`xBQL+ozMqHqU(;MW|-hT!ysV=|)&#(9^x+#ucu^L#~34%Rqn_
ztZF#T<OkmR_rLvS4nMiI<J@ZOFj`i`b#yX9|I3o<L2_)rJ200_fbD(7@S~|n4$7x<
zKp3i^)@RI|PD>;4O;jDgx!sffC^8CmkVI&?{UMw*HCvlo$DU{v4CA#1;hd1mi$I%N
z9_f;4f$FK9Rnb;auz}m1t}VyV-f%J}H$u*L>G|Jf(TJhj8nhlgL34MmR!mzTe4X+#
z(mi<M%DV-l8|+^byZ!}w!B~KR>%UOa#+uS9^aflHiB6E3rv_kc!J^LD80WYB%%h8q
zFRR~_Sr6yn*HYkCm=lv9>>4}{a}8@<J1_`Ha37~qjeB6zFu0R>*Wh;R3_oWClLNnl
zyj{)CX7n3y_REjpc57BeVbKn4?d@Qhum1BJm5nDfe>;Sw#i~SCA^h7B65@)2QgKkm
zXm0T`*5~s;pbIG2!<5btTZK?T)Cn1bom21R;KcG5`!pJNce6IP&|(P&8*>$BWt@?9
zm-JFnMepJp_#DuKp4YGH8x9h<wbqWsm(<ku4h~!ljxm3{!)&|*JFUx*s4@<X6#Fe)
z+iCq+7P<P5(&rRf)|t))na(y=NW3!1FX?x<wme|vWe)f`@?v%aLgZgxCtI4D4p%?i
z83|5@`I42DlfF`_N|CZcn$uNMCq&(8XDKPlms~rjb~Cp3f2gimadF{&wLd(3>?vPR
z#v1$G(2!9*x5<Kv?b5&TsfGnkCtE&ow@x@P`>7k<V$VE+7`F2}-;&JZE;RD1#G={~
zZ~^oW62L$ff1~7N1_d#Q&c^P~o|<8_tJD{n=8h_m;V54U$+xONbk+PcBdP~M(k@Yb
zqb%{`6Uo?(r2E2}de0;KH1(>de{CNaa5%wzWSgVb;YgfRDCW}r-iWJArUDk1mDD%k
z)h!XT<j$N*J&!`1Fl#`=moG_ot!7IBhMg*wym-$PZl2El;l-Ky92ZlJ4k^>!vN&C|
zg`B3;Zq@$&m5iLvq9DD&?W{Cj1(X62<pV8)h(XY2-tN_{w#0dGkR%b`o{iH{YO=_%
zPD8#E-?NPT8sWreYVw$4N`3&Uj#FP(6-PVU8?37rmRWeiN?51F@0+}HNwEbClk69S
z$M|BVmRuC~`cqQgYN?Yx%OXoPbhD|&gNx8pp4+ie#e3xrV9d(Ado!R^i*Yvg)HEqU
zh2Ousm;bZV*h7}NFQZuNYeALonN`rP(&2Cawyg8(e=FC`vyC5`s(#WrQpR3hfy{1a
za1e*7uTUwvqiVQeWLSAzs0FxDzcwpDy*hnzk7Bvam=hAI`*AyGM#j#kz%+*^)_vgU
zHhH&^l5v3I-kkz2fbbg`a)Vz_AX|~<Y&w<F&jO<;mUIcCkGJlYxXiDpDXVq^VL|f@
zuxAySna!A|Nh`WMm1Mpe2w2&jzN7mk4;+~>;j8abh)Qp;8GgxiT>@UFHIw}qb+3Qj
zcxd`a3bOmm#;4Zxr5>$;fk7Dl{%KuQu7_5(;YdUQYA490+{t&=N$s8#Y*3u@v^=(*
zP~LvI#93-j%pLyKyxMK1#r{Bn#g{q5ZQ>2<vnV!3r^I|mKAvjx8AU^0#y{molupbR
z$;stQ@_9)V6q>yl(4v~7^lF;!qJ!;-{~NdBYfUB%vZrKzL)F>2P->VL<z!Hkz0wxs
zZ*}PBih}V>JK)ld_5J$PF&N&ltk11wHsxn&?uNr&FxaWJ6w7ID^+du%kp;Cl#aern
zUQ?pFvyVuhsWK5>0n@B9@D2j2GUQb`?((GhlQT0$tE;!q13y5D(F`qHS-R;gL!uKV
z!p2GIWr=~6X8lSVcD5BA-dlBI&kD$9ky>Sr{h<DgF-0NfN$C@F7?T#qKhU^G%`FgA
zt$>AFQ7*d4RA<0%P4N&PxY^7HjY3|}9S*|MTW=v-@4t_G6;3BpiBQ?ym!rOYLpzl<
zvJP6WAlqC7z)>iHEV80I+$7<%^9ZijSh}k9PO0YL=^zYwxMU>cv=kPvRmAKp_m0r-
zzIa&X64y7cH)ZkL)phmV_qxIJ#sTPUT@G_{n!>HVe(A*bkN{iVScqeFVuV7{{0|)N
zJfz@l00+O;+tO19ZV3+q3F3!wnn5991*ZfkFVzh|>d1oijk>FwjVLb2*d~^_$Y7-I
zHia*#mBJAGw{t;i3rSlpG0m9)RDGQ5sptMHp`Pmb8Va}Lrc8>x?paLw$|lp+HR1X~
zZB}L+-zMC;5feToucES?Pan*W+O0-|9<jhKTP*lw!w>W4Na=W}B?h2Ql$@=!t2}-G
z(E#Oh9a|xz=kxQhucWyP&`}R}6Rmrz%tpzechO7L&Y+X|Ce^{RD+$EWE8GM`^dC<f
zjM`}ZlIJ`;JXq-t4mn#}TRRu0o1RsOa=6ydN`l$spo*E8BqDQNHJf`jBpP9-4v&y@
zi%=DEmj{jYxu=hA`ufpXk3JthTb?Hh0jvrfc()&z`jC?QKJPVri@TMVWaHQJiP2u)
z*{xQ4WkOWWqlidRYS2YqHSeI{z=)xIrzKN@O+oT?WHlSNWAl?XMMX!gr=Bl$Ih>fO
zdcKa}m9fIKuUyL%OI{*s=Jm);bqa@Ez!tbD>_haA2!*h)^Viu}$8SCE74>6VJHRwO
z`OccyQT*ybYm-SH;{c4`fhvN^+zRQbuX#E?lKa!o76gCZk?<CiD%?r#XMJmI9_?hS
z|9wA}d6;Wq@=Y8u+{f-WGcU=>e{^!V4i?DE>L4Losm7TnQ(pi6>|8GOyEJnxi`LeY
zs`LJaZu|Oov@6-vV(q&Bfh+OTw?G9$M_E5(DtK#zbo~hrs5~i8<{G$cY?>c2ADyOB
zpw^*^aMtY)P#!_EMuO`PP0rL6G`ie@!vyZ?-uYF2NOygMHImj#orGUcf~T!gbkhXW
zVJj8Ml}xyFs6vv(_KfY3&(ip`eYrF8_R|;E&rX$EtQ9PAis~d;7!xaP0OHF6yDMNG
z-E0^#81oexH(_Rev(A*ES)7cKTAkW~)n8(On<sAKdz^dY%y9L3%e@k|-p0b2p@OXF
zM|7wNA`%L+v=6`PlGmeFH$-5;1%A=%(zFph^TW92=4O1R*ov!9YSu4t`Wuqv^5O}a
znA{IhuVXLQ31@vHOmUh3fQ3o8ll{ExnW9W*>*=k^AIHKgTf4gou+A>1*^i9eRNo)b
zQi2tq_aRT~epz(tXD>8`N|}Y(9Z>biN)8|Tz!vu@a?=Xv=L=>0AEt`aN{~O0_BTU=
zKh>JyhAP~=>%RH1A%RGU-evCaG<K=C@H{5ur-|Ibs)dz}o+TU*T4wvnd9kq&+#94S
z5aj|E%!Jh-0|}==$_JV$&X#Zy1U+}>HsZU2;f;h62j!Au&o(ucPBoR#xfL~#Phuu9
zdxTz80iI=)W|+W5@MssBB4-;RrifDcy@6>NJlxyuB&f4(ea&-kHgLOG^Y<NZKT;WU
z<c4tm@w_vO79iz`vCWwsq3dYPaiZ<5KP-9hEUIeF-=6UCdM7Km{augZ`oMQab;mi*
z*ZxNS<+1Ar@})K#FwbA>m!Hs%wwZxmbsaU;O_$OV;v_RB6+Q?ejtGGi&$tI#df#hY
z(UvZ_HdGUh0~E&n#Gi*gZD7aa>Y$ds{BrH!==w_S8Yt-w5%OF5*FBq12l4Qw=T?V7
z$bDt%=5w01$XnY2sfA&><Hi-I&GWK&pQ`x`8Gvw#I54-8QL(dyykZCa&0QNH{d&yx
ziTmG#_$>*oD0hnR*|smTD~nH}I`>h|6HJsz<xB2z9oK*V4R-(C2>{s3C^ta;3{En$
zA$I`ay)g?KX`s4aM|`Tw?j*a|V$MU{TYXJ)azzwk11-WGShLxBqsge~32#o^>F!Do
zDRg(*j*5KqJo{t)n@57S{n!(5oiAGoQ@lf&1+fyDlGaN>TBbNNKPY6q3$tp1?S!c<
zt#HEK?F;HHJ$~vueUqemz@rnJY`iiiCB+C3Kl_5Pt#mbYBj4VJ-=0CKuubN;B3C*6
z!V$A@Hxy6ei27W)G$nJCW8##2$%m*4eWK}9fr=a1BH|E8D}1;xx;H8{ZkkAb8i9P%
zq9mzJOVaRc_%qpuT~|K!HZZ*NgI_lG#5;Q&i>wGzUN&sMT=VQ0%PdH8Ducc$IRgG&
zErCkW6cdKClgwMhjMT99Y_!R6gY`JJIWsV<!q3Z8!_{10#vh*fpB_PqP=dYr1k#hu
zL0ZQ{t)X)tvb7)>el1G7!t2uA96(YN{YZn^pf^vJ6vpdz^ZD$wYc*dcd>8Sc_jZMD
z=PUws8PKJo4Owm3q^pmWR7X$e09HbB?~~o!y&!wS97Xbp<?V!%h7YA5tZy~X+DLTD
zErsI!h{)fzr<@M1FI`S8P2qhNOI@g8JQNxn;0AK`(QL*H>ufuVt{S$mr($)YybJrD
z;yAW8(EL-pTZQ*$<*woA_f<s2;lkgf%WZWIbIsT4L#10LNYjTDexB1-K=~a7ad$?+
z{q#HG^vlfF>D?H*KOX96rL;I0{o2~jK{9X>ODHFZ)5omRwu{rYKg)2o#O77Y@cm+b
zY2d_6gl<6iXnzvFOXO^dF!?g7_k6jyKlSI^ixt#-m8*Ku2T+UUl*XFJjoDR)tZX62
zEQx3%WCw0SUOQ4VQA-R0h$P*cbzsE^k+jz85D!lmTHO9=WN3TNZdyXOjXdRB$bH6*
zz2R3F+{N$xnXr49cf9fmnn0Phl~CzM-rFcPvLTMe5x8^Rl61($&haH-5G<J>{N1^%
zyR8`?9;VP;AuDbuQAh{@F+4_3));GZjESBbs&v|Kv__Aam<b?n8}pfDA9QJ6Fl^rm
zYUNJlK0IxA{%-GX{-9HI6W8nki}~v*xBnV_OUA>!-<|mDQkWP=7AxJ>uoXLxWHq`@
zmv++CnO>rl$x2@}X(wQQJX2x3RR%x%k>|<w7Au9Pq?t#AaUAg%aVc1&+=&S~qfH>O
z^KpqK)^mHGQymB$Rev&rGF&2UjlI0kx<iz$`LW4=U6mO0<WhpC^skh=lu)9I8|-h0
zb0CybUOg#&EiX*q{|>}0iWeKv12467>}MAn8z}3gPck#X*=sVNGb}H|qsl?AZ>v{X
zIs<O{^n<ZegCE0!0z#hi_#EY)5HdWq%tZq8Ydx@dpc#W%$v7{~m>DrK0Da{#tpCgG
zXtsFk-fyq*{Gs=YL-RCF%0yDz;JJ?ew{qj9D<N{L-Rl~MVPV0BE*voeUy~HD3$C3%
zk?j;^ALi%%ACR;kPA!3eA2!XW^4q6Oc?CxQVi6`u2FeLnQD+gEv$=COiUpb70wKq@
zE`TzyN$?A8AH!7U&ZdpBbE9O<XRqk8zYK{d#`f<7v?*~Kd9e$S{^}|!GKg)z=XnOl
zd8~AGbal>e8e-7XTj8swh9WwdVQM~NZ#4X^GMxM0`tJ$9dcH%yxX2$L$Fpdbn~h3U
z;+{qMAsP1J(-~){s#X-tj%$xKc5YiyG$R0h6@Q9wdRN;@#@meC|H+rd4z?Sq0UWR2
z6paV^A%!J%nWWE(-Yv#yGN)cjYtzrdEi-oGBdqI+oRD$PkP>oxRIk;=xg7Txwn91^
z;b>U~#$L@7Xfdya8``zJA@uy;uddIt%L}u~0`}DQUq1oTufm5QpSMsl%`v1rUo`-p
z4z%g^=#O8s4{6X9qipBI!pTPd8hM!)3m$gMwqw|4Jj0Cmzc}CZx~47qKHAO{4_nts
z!+`B4f&Pp+1Ah<JrvOl7_GhZxpTX#?{j#PB4j*lWK-Oi;AU@U9W!sN+1sW=Niz*6|
zVnKP!FxHO5hw?wy%io*&=!?I9q=7^MIwkdyYMI9BDpDxXV?sOYJ&GK>vK<z*x)#mG
zRlqw%@$!bSQj5loZ%8t6bkj!Z2aVCfRuR9NrzSV}NhviRJTv$Cj_8_U>Dj;M(lmJn
z)kxvi$R&)^sdG+Op@=6eO6+wwz`ke8Qd#X-Xe4Q?@ns!udg6_lA-bX$h-q=$YHpU3
zQsH>r4|SsMmrv5vjmTay8<iG6PI>aEC+q&cQTZUl*`aEFU%<sri*MuIp@e^%azyW#
zn2evZI&uuuXG9{f!zW-pq_;5@iQF@ccdNCJEkS%gsFV2DSArKi*Bvf~!QxgD_vqW<
zv`J-U%b@bu`|9_mQEjaq&?rKw6SaD+=5{!F*D~$ZP6eu7KaX^@YJ{gYSX>K|8UcB>
zdVL)e7)ca!>ihI5Uoz#r%&crC;XR!bVLOo@rCP4<1(?{vmaO-;6|lVS?A2?a{Mt`&
zcvn2zQ>03L8Kx+vquzAxV`pwJE{)vz>E>{zm*UR$P3@@g=}a=nKXN3=4HloMFMGVh
zi6pNKSRlq)r{UVsDbRd2%7o&r`%%u{aXIPH=%&e8idyO-jU=rhW1|3ihmZe_5hlWg
zMz!Nb#O&$c=K1%B)tcYMbb$P}x7U3z%?0dyh-6IS^H(J2%3DDsTcpL`R-#~py><e*
zNjd-Z<FGkkco;C#=l2fZ$d{Xe)1)rEx~qk5vbFvfTj^2xf<nuL#5Zyp#e3VD<Ov-+
z6?3>X#MjjzMprA`TIfcA&z)Yzz{3u~FOgK-GD><vG;R&n#eI4tiSii#ovp39+k|(Y
zAM2Eqj@@z8Q|0d-;kJ%rzr~gqZHXMipX?H{96@{(Q3@b#3Js&sis8(TEH|=0<@m66
zSWQ0>Ifr)m$MN%wXvrt`6S!6eKU_>lAJI(8eL)HkLk>>(3u!5>ZVNyJKgs`IM)&>b
zYnY8+E%x6SeUPC3v#EE5hXsymcwDsQl-MhVzxcD?YH1%fu70f~6J+tyG6o(${iDcL
zu&<FZ1_qQIc=~8QufYla8dmQADKwe$bW0E=LDycdw#N`S%4Q+%W~d?}RxtUaA>{I}
zKy)K+khwTp23C)LHwzcV;+4n2q6e7?FV^L*K%802j}GCJhK(P6OqzF-g~7mz7hH==
zmz<SV3LhW8s1`=Mis+`qsM-sn^+tbwE?#XUa{_v_R}N#>FC{|%5>uK#OQhAM%VF7*
zqYFjejmgMo=9Mn&gXd7b(;fSyAjv^Eu2YluwvNu>{G@$sztF?@>P4Gwr3J*dU?}^6
zKq;BSh;UiU3OK)P;vX2!c5vsi+ppV%vlZ!wa?rZ~&E40c`yNTF<31KFLiiIizkEwq
zE`VLWQ%vEb-DB|27;TlTWKUrTzgw0r#H5Hd8yyr+>_6V(FN@N3oOR8D)Wf(ZC(N)r
z|Ni``_#mKd;Fb5n=7o^S%rvu>bgfV(PrBs;RH3on*h7!8N2wy|I`orH>s2mz;{-Or
zS6{358$IaXsnIGsBD$|G94w{2+_5BtB?1+ASfe=)cP}+h$CjUs&)BB<Vh_Sdid&+>
z-1R17^@u{I%!`o?%{W-r16%y}0#QuSt_Fw6ZoYrZqm$a~dbes^TZDReW{85y_*Cn3
zU|XBD9#VA^{xr(jaw<h>3oew9#Rh8uAitWcY_?<8Jil~za&<2mlw%u2<F-4Ifar%?
z4-;Ut{Q>eHeiTvT;LofYzW0RFnpZn3=9IF^5{hB!xeQR~cR!nMgs8Sx|7g>6gkUOq
z&Ij4UrxkI!-JIFu@s3NK-n0l>!L`2j+CMlr@%uQQjKj&d`Z3E}cuh(+7DpSNWbM0r
z`@WB*ln**yRt)3!jWs;ZayQ2qC}Cin_pUH=e(*(fkK@Vj`olI>6Ka7Nvz9HtQ3a*L
zB)M4rN;%wyw6iGBt3F56%W6`yi6IzSETrpE@QAtLte|}iE2ZZ;`_Ea&`K(qR_)kHr
zo0?}E%IPw8V(;TyR$4PJLMc=()DeYnZ`q7$)bzZ)+XfgJ=AX=<Z=1CZ1m@H_tBT=G
zgfQ`xsxdWOgG;KVS0O=?>CBxpV@eQlZmpRH(G^v|OLsGXaCvr!hi~n~Kd(qom6ymZ
ztNdIbhP|a4Z==;02p<!hI+rw`P3fz<9rDL9B)>-W@A^``<d@H=bZ4?#q?!~ODI9gW
z-%BN1AREeozb9k2H8Z_|wd1<t)k;KYb>91!QVLRW;y?R@dD$OItOE8(2KT8gp%)YH
zF2(6P8>DkC`6nCS?xQJZ85z^&${ZBE5l@6LaH8qp*f@!A4%=t#jH6qnk1^joVLT6q
z_-;7rUe<Llfi!^o%HF|pqTHFmq)j<$z%M$ev!)#0K=)ngu)_}Ye}X%nUAHRHlPUY5
zPp}4E(HSuCcH`e=|K)n_;6l67VQ8f^X>h4QA+N7}kF{tfw$@H9PS^&k!TR9mvuG=9
z8ReI{&slG&zPu&Pd`*D83nseu&emg`%!u=fi(iHN|3LD*#`8;?+Ab_#*Tk(J2373F
zGH}hxNF~JGNtOhPQZ<vcYCe@EX8VjG-Kh-&-!*s3va!~^bQUnz+x*d2^y60uB`jVL
z(^fT(3?rv(U)tXv%|8?ICbN2Hn;;m;`}+g&=MylskiPx$87`Ib@|QC~II?cT#I-^M
zZm!Sh-n;v5<3H8GaAp&^z2w*h!m%f9#SL?5b*@F+H|8F|BtGVARjFRJ$zQb0OwKAf
zswusGQZ{xTHyffk24gag7x>n3;pp~$nLbEna%FL~iLN`2`y%(%%&r7{l#lW6YMbgF
z19+MoZy{Dndt@+rNs()}MobtbrDx&KW;htN{3!iK`CfRBP0z}Rp<lQ6mYX4+5zW>5
z6VuOHzsm#iMZDPJ1v({--(yOJpjX9KZ)_uU_gDYA%6(~ee27g=CMeF&uyo69c|}qj
z>He-fH^>=HOj9cu#j3(o=o2PH9M~yR^trD7M;g-MQ(55K0Iee+tCNZGjPNFd2Z!SG
z;CsdY_A|pqr%!)aOY5Wd%e#_NYDP2)owJfRA5Q`dq0KZ7y5FJWQ;iupMYx!&)0T0W
zkD@Gy2UF^Q0sw3t&Lk7Fd43+)W}7tVFc&?|XRKWW{nvhW=&0Ae1f~qU*4=pgQ8o1D
z=4@O8+f75RgG0?f5>d_t9Slx#Ar}`jFb?`)?@!(pPbG?(dBSbPGNV1g=<wL)_WsY*
zRr0_xXTSUqUf30Kp@xo+q*U@%y)cY5fojTBmYtm73lVX<LuWRB3ZmE+o~p3A<vX|4
zA53x>TlL-NPV+eQ7yV}8;C#3w?>m-U%ha{$2bwjpc!f-+j@z*xFn-Nyces|j2*=~e
zhd|KZgUbLsG)BdE&e+WSzsPg2H8Wh>7qEQ##0u+NkO5cl>rBoN8pk(my(+}GP){Dx
z69$_-t}m~z4!OF6X`;qrG1_oECQvIj+*Q&Do0%y#KK?P!&W2<%>V7AC;Rt3R^=vGd
zI_axY{2hUl!)4V)l`g5@-63F-$&PaYem19QY$PHz^-I~CIgg2##IMrExLA{9fG-T)
zVOeE-SQE=<b3Kv%s`iy1pI4M_fdvA{YHOiNZ`1E~M6q*y78n;}C-p=Dws4wd=#cT6
z<ml8CUgqoC@o^`rkxa(jui#}%gksJb<U1Rz64DnYhti|$rPcamA&|~Ep|IQjSi9x<
z7LV{_D<QgG+$Hx{={<Tm|75~C;I>C@EQNSkxcmQN0>-Zz8%O9Xxs!an0Lv*>Na3MY
zIO*pG*!eU}%iHSky@25TUuFFseO5je6};3iaSui+nH$FZUUesz5o|KXiJwp2R1h(-
zbu#%;=D(?-iS0i+^7N5UTz8*)7mD%M;2sh1i=*!~5dK@obyhkHAe=~ev>`=a-A|)X
z@v6Q{>b<#w=c^FYIDxMlp$i^z#ce*S(35(nTVo;V!+vcRsjC_er<3mp^*@Os{UHbV
za*ytRm%I=j1mb6-@C5%~o+7exD5M9{`s`+xC|{snZz(jf*GxA3%CE@*Y!`8{<TO4~
z4Sdl>r2>pd-hu$iJh{X#<0fNQ4N{gwe}x%<(x3;zqWQCMS|S(0j*LfL=21orXM===
zg`(HlWw}d7p1>aA)1k)Em2a3pY21)Q_t^%=DvOl*pQH>wi_Pi5B?lG_`x?-O1<2F)
zPfUcA{@w|J_3S8C?27xUl44cJh>0bG0$TLJPA=_d+Q1v1=Ge}PfQRYGG5wqGR5=Ix
z-PJxoTb`4v8or?Y&;}--Yp_E)-8^YRI$bo@8Jj1{DrB3gSnC{&Z)i$PRQsYs5x1=#
zmwE@khf|p?sLMS0#2XvV(`JD{1C4)p0uC7nOKWv%cTO)Sl(4S0`T~fpP>9sSI7mIt
zU(Vlu1-Lviy50hU0|;SrD_@s-tw}WxojHQq+K>OKrmBY0INRj=pn??a^qlXSZ2h1`
zv@d1Jm$06-=d&`bAFYEJ;nrHs?H~(AvKul*Wktl&WHbinL7~~$8+hRyDo*$lbA-#x
zL@x#Zh>4Yfx~xMl$NbW${9dv^^e3a00=uzH;O*^K(G!X4J2&smbjXhNBz8o2{JuO4
zO1$0_lN#bP!c_V^CUUJ#|EZq*db^E&!gUmcJ@B(Xow)yt6gTLK`0+t|K~4Ja%g;rA
zmKoMy<rC?6pQ|H35+sYT!<!p<XR3|yVs}IMl9fnZixq-?$Y65XndxI~?6m1J(AG6R
zKv#zQlbf4xW>1^Uz=HPrfSV?64xd)Wg1Gvw0(@FJZ@&(fLjEcCbGsb>+$XA~SzG%F
z;!Uq*=U|nqt-bGe@KRIWwbCUhJ^SX9d4;69bYu2#o;uf?T2g0iLCZmTO?E2bKWCrM
zsCtm^|CnTO>E?2zrhN8=of-f2rQ2`l%r@{(oYwwx-MOkE8D-n2?#|%PeMnJb8Q<^G
zP|cLS&+`f%t_my_=tCu~yv$O`_7v!&XvPKncH4Gn=9cs6Snwf1bZME>iB{q?w%?0&
zz{Ad{eCi}9-?c&^6E3K3)y;@fBt5htg?vF-w0{0vL55{qs+_bJ+AVt;ke!p`a05_f
zKif&3+J5(;lS+{z#*0!lTwRrmqFZe*b|Lno(PqcCVl}xAX+sqz5RS>#7ZEhCHk-}C
zE@;Mw1+?s&BociT^AY_?m?&knelj_Heix<n94y}Zt$LClZ-D>z`!|!LUcdcpUoism
zS=Pu@Ah4S&)=RxeF#TPxJx3dqmj^f)w6XoN;m!K6rc_s^ZY$Gsi&1=FRV6)=&f4Z#
z9qO|b9x=vz`;f9aTX1xGq8D*3DPu}>pIW;LB_kH!#3{_A=;rB3dIKEK>d3#WTd4o4
zp3%vDZfzijDm)~82fG8poh{wn?VT$0p}&(_EX-j=pAIG2!xgaYVam(hoj~x}dM?+M
zZ}9cDh6zh0EKc#E4XuM5d8o8#nYU_Z0JiaWhcHTS<C|V}dfaI&Z&oRo;L0hqvJBdh
zcL?s7H=5GYrY_@)7JnwUyRCnCcdFK&S7qRW$mm|?*ERZxUwlqdM^{<LizA(Hq9N=Y
ziJ>7aN(cVyVQNWSZP&-mfL&RjfQay$8gz{NYT?s}<!M?guyF_Zn_x2@@+M)PN>2d&
z0gk+_!w{QK<8BFhnnbuD)l~?Z-wA9ow>Z9tSF)9TNT~wl^c5L!q@m42`tPX>Ups6?
z!G2Lw4Bh&+-6ICdG1vvagPSVy)y<<Vj(z`Pb~B5fKpUjh5HTz8^wt;Wg7NaHgMg1w
zy`=1cS2KLtW1y5~bg6&zQd5rhTo7TBK+eL!l%?36gYPrYk{HNx6)~S_vi?9^naYNy
zKXq$$bi<V(3SaKCAJ6MQcN>l`q^6BWsBhO#W<q3x{(9Vm+S~8nTDoEx`lgx`>)^lm
zjTusStEOn{k%|C7#KdlSBw99+@~jBnvf+2qtK_}zU?<d`f|OF^9+-Rc)C><Xp=Pl_
zuygW^;Hyvt9gS%k;yl{i-l`YRY-Gd{5BkEkP4)Ucx7&JS16o>x!xUAz^U83bxHCk-
zMxE54r$-Ai1d9$YBzKICwxoXMOw)LCkK|RVeJnKZeK_E%7dM&eQM}0&%R~yfuTb>P
zDaZ$C)}Pe3H5y)!V*vNMDzBg)VCKzaFa^Xs5K*DtwNQ!%S!1!xsc#Xp9Xdvny>qP#
zm&-}9An$@#u8q^3wielbv%vX!O8IF`>xbdZ5CuDL0i*G8ZJ82x)`2V&Ju0oT%B~JD
zN@w{Duqo#^w^$QI;Y(y4oU_X;y}eKZfBXEZOV#eQIY>h&?huV-e!;W8UtwsEB~aY-
z?*wDqKVmDg&f&)2<~8(3M@noM-cqVFsuO8_`1Y(?klT&+-F`T6LT}(<6N;YXB{wJO
z4U${pBs?Tgsiv~jH~22D7!%1X@OepWzdX1E2xle<)9L9mt#KP~K8oHGzdZovvw>^V
zvT~P?%M+HAjxJeI*e8X%4`oQXObnLXo3VL5pfKsFbZ|K1n*3{tuL$b4ZXo2koC|dC
zj<xQuzoJWj8CSC->uT~_*~V_`-0Zjc^VU4O5-!T^5CszZI2Vpn@Zg#Pwaj=E@0)-G
zGo#krp-wE6M5Qb6?C;-zIj*h1^FQag@h?r^jP~TjO=i5G4vU{nB`X80YuPZau6+%{
zNDxW2gsWI}?!-tNX6CU-5QWF|^J5=@eH(f{F6OD$dSjbcSMrIE66y9rKz6>cK1Ahw
z)pFOb&AJx5?zMlxS)}QD5`U+c1W=S>ju!mk>c!gT03BPGPw^Q}6B@p<ui4KG;V#Pa
zX|=7K9PRh-Pil@QCQuBM<?Hu*5{|Ym8Pr0r*2-(UuaxtR!nsZKiAdj^yKY1~vtA&$
zT5W4k)T<?H<<BBgdZIRG(vRSZm{U_a(hf%ECY|*aXeC`kRa8L}L?kO4(CcSMf1Dio
z3VFb(nAx{>8gRCCbvo)s*_B?H0HMKDX>3I%NTl7IIDG{T#6;hLni0d|1r#tRp-*ya
zrW7+-H9k7(!5}q`dvu*Ny}1v@enL9mk5rNW=%`|Y8kaQGd2hnAP2;sTB9m>7Dw9J)
znr|mJ4-)VvA*}-+Nj{4Jm{#wb)j!C$C(qUIXs(YsRo@($t*T+@4a7$4Gcwlxnl$!>
z=}2grzSxC+-SaOkT|H}hnj}D*nxq*d|HtW1@8_8I_5IP3KjYUY9v`1}P+N=&*dIYP
zilFcu;5+(O&2s8&d%r`o61Bc<Mj9_qAQzmrj*i|&uk_a4qcIX<Afb+0fyI$oM*KZI
zAJ132FO&p{EMCZUI-;MTtjDEAie^gDymb{tSMhP5RNN+X2~vn&hGDI_vHq9tmKnGc
z9ziXMQqic#6O56lkNn_&8iHH%Px!myp9K!;Qc%WNdsp)hbl>WhESnKF>anaHxHjev
zir{bKeOsyS3}L2#vT^!PYMKOSD$z7NcmF@Rr(e!s1DipD!%=YxnSzdjd@r~njV_ng
z{SU+bP7%1Sr<0we4x21dXb5#=xI{u+mcyU5p#MENW+K=;Oh*lk6v@zL2-pF#|GItY
z>S$}Jy<N^Dg`MiPzJFnaCE*O#1UI!6!+?T_N51>*alqj1htr=2b?*w7XX1y4cf4C`
zhEZUos;k6+7whYFQTfBS?x)3^9e9V)nN_vpT4gb$((fMMFOzVk?dyA?g@H*|Sx?7)
zD)9e?+&7FecCOGbU>FP?Yo&hAi&ar*=5*WU2>d$RNGsS-bMD<KI^0z+ZwmH_kT~AJ
zHAdWR6$OTU6}uM~4b~6%OXzJ9rUGBKz_rD>y(Rnzm-Uaan|vf-YJCH9Udc=9@FaLM
z?Jy_`5$#`lTKrenft6Ds_BxL326#0dZ(SYl6Ap19)wsM|nQyYJq9(;*;k+yz$r4$p
z6v~A7IM-rm&+KWlO>+~vN+28d-|W5iP5vf1pgNuHRRZNpKca3CRySNrfZwDn2C?1N
z-@ml9FxWmAtR;->%ByV^&!x(dKYFQqpHHZ=p$%%CxM5vU?ah*1YMv4IQs5>thoaFk
zQFc7<yh&VdLIne3O5T|;*jAR|T)_W(Iq*=E`;a~3jbW!xuToyEtK^>Ip(zF&!X2YP
zbLh&u{=b{+SP#38!_NG7NyyB#PTn=sGKY{p(KzZo*X<?ZMoCseVQezd%nt3!R6f=J
z+V8Bd$W8j+8Qe&AYFjzW*L+d3GaC}5Dgs>ovz{{6e9@DhonjI{_C9XIPe422t^5IH
zN=W8KPG(@v7VK@En`E@6x;TE!i>$^Rpq<UruhyMD3FRFQ=4xU#6y?`|o9-!3US7Hx
zL{Mrbj&TUx`1DzLIH=23xg{kCn}1mJCj3{v5A3Vx$zVStr0g8-;|11ibqb=OQZ}_G
zo6ZU!GX<fP5IQQkS9>_V_vUoQGpI)T6P+HCaB|K!tDz+I(4vMo$*iWIKW&Ww0XL_l
zmp7}65ISQD4^cTLdu`Ka0#T4MFgP|)Fz^H>j59%!y@w|lE!8Qhb`aJ^30z>#P{fce
zjmI$;+L)gQ9C->fU%c~crp`koX7cbkEHz|MXhjA`hZ;xHna8qnM_{|!@-v#igr5j8
zzJDTHF8px3qD`$4{<ot=JD^I2)iQggkr<`5oq^s!f3VJC14~vuf6cmZpL1t|Otz*d
zVk0Ah0~PFQ@M%J5{FJXyiAQegJDKJ{;p;-5<&i9!FPhJ_Ff~U-j`l?(ec?TQ+i@M5
zMNcn@m&n82t7t1nhTF#Qd1*#Ao*Qr#6={~#NM1p6e4D|P-)KS{pz~X+=0P62bcc%R
z*m)NhxJCypjhDYL_M+}JX33A&<CB$2Z+e`S=`8y6Y`As<Jd;Y@J(o3=;upqaSslBT
zmQ_1K{X?%5^FNkaxV7+VIOs@9QEYz>4OO@|CbKPMPZ)GlnzqoN9osw|MqVb`Z#4<S
z7@KS}5fjs=Wi(lU=BxNA+&_GqqX$3S<yR-Zet|QYu#g+pyq<&UuyDTRP9ti*=1T4=
z41ZS1YO<Y{T<`62s3Y(VMEe;hK)p=(5Ol`uE39?Zgym)rt?BOkV4!VA+ZG6GPE9kz
z4mEd`H0{%Kz9~6+72VQU+<&N5PF*2|aFb-o62&QsR04vitn6^62u&WAA!X_C93$p5
zToG?FU#&@e4wH0%&Vs4*`uq9AkX4}E*YqjnR(<zE^4K$0w)XCBYar0_-&)2m`cUaj
z<@{Dj3{(SxHT%Nq$|#|ry!aRFlZ^>^TtfH+1cg1eX>08?l-TTejCs_O(J_V($p@)$
zW(_?XAhXDf-d9HptHKCc4@$@Gb#()a`)7np(XzR-rGPmQ{~9_DP#(htDdu;FT?GYq
z9FKj(d^0&3tC$bzO5&w;S%tSsN{Qh)R=O1SlCVCal?4UZ>*Gv2wXloV`LhGkGQ`E?
zHzhQ*h>HhWC5<boLCvwOo8Lm|z_$OodN|P6heqilpFR{{qJ3_as_uN#m^4nJ?1KiS
zF6a)pbnu@$?Cb|&AEWL_apKP}ZD)Td1UOcxe#?~udD!dY0DlI<3L;6~NKATUeEf9Y
z#{$r`p+AZ{-#ia$-RhV%F`T<`p~|SCcN`$!6^K6t_Xg!IU}b5~W)DIFY5o=m2+pY;
z)s$J_!8oIjVwiIHi$lH(>DrmpgwQ|Byij7l2N`dT#H6@+8($mI!=SBGJ0bEMW~H;g
zLbjd$9k(}g0e{!dH-GP+C+dEFUe-8iMH|lUR+o+<xiYTFJ!PtKT@BUPhy<HSr<hIR
z1TI#y0)zV*&C<2iKS1NGh3Wy|k_v|P&M*719QbEhJC2X#hI$X{dtC?7k<#zAq;!MZ
zw)r^_i0<g+0;pbcnG<*-FvCl0^RP~D^Opf?I>*5a{UZmrqev5`cX@Ss;MJ_=&U}6=
zTwGietBE;uKe*#@z*Q?|WKI3l-Cc7jO@Gx1QUd)a)E=0yD*MsPr^f^Mces+#Feg4_
z63wZIwKi}VMZY7zV5c;NNZcShZ0b&kLzETGm6&KQjaZ0ywb0a87Kve}q-_$+;#FJ1
z#~xX_inJn<DT_w@3Np=N-8p<LH7O~N_7_YuLE9SF#e+|QA=-KJvGy_asVSphnZKWe
z{s0p_+YH{n6%<OX95DgrJUJOdDC6xHKi{Q<fO_XM0^Hd~zXI)PybfJDXj$N*$0*=B
z8o*G$6m`#cEfWp*VKH;x!VEqo9|W~d3qoGb&NF{&JaGvxhV~r>#Q@j`<7DZ&Kjoj!
zjCAmXwAHLwyl&?;CcgXPIiD^X&RUU^eZS1z*9L27A{W=QDG6{A3b$%7*72Xyp<7;y
zG<d~v8J3Dy`$N3M;ln%9ut>J-lS`=MXC8sS+f>|3iyUCmQe^%Z7N`7fxTFxZ+i@6s
z1sW_E&)P%P5>=z7d<BN9cq}%r4n`N`Rzkt9M{flm5hu)zmO%L9alGBt?;W*EK<ye2
z`~`o%%CMin@oQ{8cf<YL5%F$29wZqecRsnf>x9^}vu81xP>K!|B7O*4e;PF{L$xf~
zeR`)g#*B(b1s!(y+kmJ<iDL6Nck_thwi=%5NOO)NBIS<rgIKj|oZ7Y6^#M332A0;)
zQ1XQr<YoqfnH{22$~L$*9V7Sck6j_gGpnI2yjeg_1BzC&djUjSb7SR_o0F5}l9RKD
z<s;pQgPoiH6Y5<yFe}Gtx9knLXA?`+BAeb#MMklPv};&-idAQ%r{ARPd8;k}+VA~r
z^aVy<QEvzYU7sEwmmiLoXX;mg)N$KbyO&uy&hIzOVm44fxX$Hy_o2cSr`RixrW|ob
z1;Bh=5C^`MX`-gPIl)C0V*}RNm9viUi!nx*;x)%&+F%a^(iVpEFr*@li{ea^Y_J~{
z@>$kH7Owqw4SU(4L-iygetWZRI{c-quRy?xzp=TSfM0uD7Eut_&7+XktAA&9TA)M~
z6kR<MKr=PYw}`<GJPtj$mTirUoNcxM7gQ;?I%TRggsV5Y@5hKw7NjuRWOXL`QC#&3
zc}_~q{(GW{6cg)(S=GNQN3CG<*j{>{MXxRdYWHR=x1PcZwW$8?neMlAM>KJo+VWxy
z>sa}FSt`Z}E$^{(eW5_Qq>B$GiIWuO2abPKk5W0$p~^(@R8kljB%NmF1?(Ky7Q%q3
zKn=QyI|1?;X_>9gDf#6VZVVd}gR7xWAP5~&>`>rgCj}|v198)ez|K4|d`I1ftbhBL
z5!stl7&yWyTI7wU$Qvifb9%Gfl!Og&twdwXpR~^j4W<uaB`GGfJN_ys-A2V+iru(h
zP?gVSTNOt%c>Is|Onz_JBiI18bLeD`Iv#E%6-p~`&in)tTG3E-zP4pM*=>z5Hl3Ei
zR2OkJd7PZ~dZMactLAeWQ#+{9Yi4TD;@ir$t$GLt2Tiz(sZA4P1rLmzWxjTTqj%yg
zAecqtkA&k<OOW<0ve=FMd)Tr-TTC?MqDHEI3V$+;&AdLfySm`I20mW@#&m}rX~_*9
zU$(b@IwSXPwS8=qxznbXx<xy8MKb598$QpYjIiHQIiE%JiwJ#<nvImL{#%i%ZS7Qb
zee0&g%fSjC<_?-=1*}1QTpe7bRMxRXK!H|5HSD_I=z4c0<a(-Gad|nm#v`~|RtJtN
zHnFG}_EqUr#&iS_125gOdE(^hJAEk2gg5I}z;M`Uy7e%*J^!V!+PEUr$mr7N+rJo^
zB>s=4^A4x_fB*O~lX-+tWQ0$Yy+^i^RUHmFHraa~gg0f+$ety~IJWE~Bcb9L*<@uL
zvK{*vzxVf#-*uhK-|x$N+^_q-pU>wb(b+r$+`aW5JZkf*+^j=8?Wt(e3uMwmvzF1Q
z)ZFs=ZvmvLPIXVH(kPp1=x@m>Bf-&D35S-kb8>4{f7@WxrJj1#WL&g4Zq+A$tsxy<
z2pCvWLI8u7=_`cj`!X?n-D_l?4JG$u*;^w<Z(#&fo#fD0arkBBl?wrO!v$YM{6Kk+
z>(r9GN4xUq>`(&BoNPVJ^1GUws?BKu;?zm0y@cye{h{N(1<SBr@P%|by+FMs%BE{C
zVL1=$X@9H~NAv|UyyUSb%A#ybB<b^Mf&D}p2@rA!{TmTkcfI;#54$i>vb?bGEV=kl
zp7W&i#>ulh=&+QV<_F2y&n1-9I+knEe;U&+1$W9P1O(tO<DoPrCz}tEH=Fh?s9)8J
ztyDbHWF#o2benn!jgvwK-f&FCd6m~ceWysi`xV@v`<Q8!d8_}rOPHV+KqMn#BbEEE
z0I2?059OpO=&o73YZ%)iKq2QNB!kNV(^OxIXgF8a(WntQeQSUdJS*`MqS_RSZ13(r
z)eJC0)g1`%(MTnP*K~P~7&LoT(%z)L84NAsMsv>kteUMyRdK-1muUmP?Q-lUVXpj?
zc#^auwAU<z18#*669q7lFd6bPo9<Eo`#y@eTq429#hVZptnJVC*q*D-UqeGBpOV07
zb9P1#!`3k&z>HKm*Bb(Ij>X3Uy+L8`+%^_`3Bn(ID|=qH+jauRP?4W5jXuTw5lKvP
zpMRiQbtb0UU$>Tvn%Izo>+daXBNy0@`xFERKN?;UggbOb!zuWC*Mp4vTY(jekgJQc
z(Wd@0v;Zb&pHOh`damPg|MJBB?MX+->8S-+E3S`lbfb=jOm4}mf;_)f-Dx4lQ}JKl
z3Gwa*m%)w0r&H-6LC!M7a0WxK!4^n#52=*V%~Tlu$9L<O?Yo#U)S1GhubsicJ$emw
zlux@n8|^M|T5_UvdA5TEFjF*|J83#}vofTymZVZ<!WQj8Lj420xZ`nz6F=`Cw#ZNT
z{P6za6RuN9d!p{epKoSaiYb&M6Us@(UORNx4KB2YoF8uB`p>+<s(5r4M-1tf$wX`F
ziZcBwp4>pR2fxWKHXci+()=WNp&nu4dE@t{ZWv^}S?RpsH+COLXS4R^alPqDICED5
zKpXW2IC}!=s{N~8dDqR!=BSkW2L|PtbzV$RGD(Tuyy#w*t#OMf?gYpa+8*l0F0xEn
z9LN!=N+QhVx0sgnE11OuO+|p&QpAgOLIv%QJAzO2tNI<_*bj22;PlQpk%aSh-sHJY
zt132NBPw+s=m=(ugivUg#jcA|Jn&y2C??+c*l7)tF$_nJX&%Ra#^;wqAYpjN=w;FE
z#7`V^KBzW>GJ!|0h+gn#vdm1(T2;rh&E8^iYNnB%Nm1M;KRq`gs@SG;8Mi+U{Oztw
z$7jd4#%^6s<UV0cN+zkQ_Coo#iJ{#SkzF%YMpc{|@M>bfdj6~%oJU_G4AT&6s7C=h
z!LzN!ioRu=1h@58U*moR#Sa^b@UZt;$sSD$EsJK<3t8YA7eP5G0AufRWBY}8g%%Qt
zzaN)7gfsn;1Rj}DRuo<GO%4AZiU&Bo1pCvB$nK^0*3!?t3sNSuAg_fp%4(wdQ;^Y>
zH#L9=Rtirtj|bmk!E}sx?C?vQ2N6{#qC}U26k{D1Nw@x<EIcMvjRL6&{igYA2d-rn
z_F0vF?{0{%{^ET@`g}bPKVK`emNFUtkFe4eK#oM5c3fR>1yCl=&$qJ%4*8z-TLAaY
zWoXSLO<%1w7>LI6%T^aEs(a~a^k?UNhCT*lz0`vjaFR`#&nZ4_g3qd>nNuaY?ryJQ
z(bi<<0iq>4&6iirlH3EYV3zSrrN;%UA>eJMCww+(p$I;AM?USu-#X4Gl`aad`@nmB
znRgNnxef$*Z41E3(k{pKGH?ojryY~x+rqV#c(NCkA8br1%nr;ZxvcrqoK^IJsOo2~
z&hH%@`zspLZS9zXzt$JhS+Vq8Ue(NQw|(}Xa^`4Q!`6RPXE~AXlime_@R!e#?(@DF
z61fiB78AdF*dOV#7K6`+EI=<dSjf&#Yw09>%=Y%Z^8GOE-$bY{mqd=u)Ft$>5hb=n
zxZr<G=ASg5Ohjs{8mRk_hEo9ja>3vkr2~Z7rUl&#-2E9fpVOrvJitUm9MP=pu(J%}
zsRaeEKUo*63Q2KpyJNc{Q*N5k%CrQ<0HoR=bX<qA$d_a#%EqJdF04S6V@}!~m{M`a
zUd2bxf65YJHiRBU`|Xq3{YDPTrcJW*xHQaFF%8~jVToIT&D>^&0t5A|1Mf|pwe%&K
zRi%&DK7Z2dU+uj|qppkSwh{ku-zDt4H^#1E({^z4-&gQ3Ub|^M85O{!#=F4fAH5b8
z^5>H_{Ln}4ae*GUS6<0(`y1r7$AZAm?+<!Bk!?!z9hV0iYPnUQ^1_bZt-TKI%v9<^
zvDm2I6yt;`DRmt&1m0-azgRYRx9u^p`i~PFuP{JeJAO|dE;ar(QiCwe2C9UH24SZ3
z%Pb+2j%OEV8(16xL%6I_Q;roarG$LX^uy#mAf&0wq%d%Npc|;0NI}JV{LDjEammG)
zf`TH(HjH^y%L;x~b+F-|DNccpG%cmaK%ZK2wpFO#bao9;My>fuz8IaCk57cEXuOIu
ztmF~66SOC^D3v=J(v&1bj>C*a?ImAd1(;{S5xX*wY;>kFnV&-}$kzyaQWh*pTnI6|
zJ<eqQP~|G3dci1cZ|y4G+dO@>ZpB^yN-k{_jccih4V~u_z_$*J$WnB1(Wjiy4u9}o
zVLSV{UI+Y@jmuBxIyd8}e)z@1yc&$PQ`kDkU~*daQUeg$ps>5xO?fng&3Zu|hme-i
z;IKZ-L88Z6ZRKEEg(;Jg)#V;vSx=_A!f9hdV;Q#9sMpC=NBRC=OVIhB6{u%{sp&+1
zhc$R$rJmzAR#B;ceD^&tgwcWy0=@$X0FX`26@#@4N4pDi_vmmqTw7IuAlhV{Q)RZc
zF_DU*>%<I~M<Tu?FhEN*HzD);-`CgW$y23pn)d0xb)oC;cF)_j`2&+aDn9Xk@e!rN
z6+T7RJ(sgPf3^2KX*Yx|3Hc=C;f+q+LAbbn*4k!N635kf#`VE!xA3M<EK(ZSrqvDa
z9!4#mEBf(MtNnw?{coYk<txN{sM2^L454`O?n4uno=6$Hxq5N2x>*fIu2UQ4Vps|D
zgpcS^G12n-6^~4NsEZv1Bd`=BF`1vv)^qWd5+m7<Tyu&d2BD(<l#29HU&5X(Kcti&
zW^~XTy#K*2Wi@F-f&`-MpaD2(sylHP4ut+9pOt(%4$~U=ZFlB3^HpFE%U8tx4LdOt
z6yv1HsH%>m7WI}SiEZzrC;8ls_~InV0C8mIKNbAoYH1QFbhNzV)t;ZRn(=%AT-<E*
z?R!k2Y?&U#l6Ga?HI?%r+&wg7!lDL7!P=JLm=!AG!~dQ3J=?9`VPMxS5$S#`O}FV|
zTpafe9#JU<sm*+=mH|5yjhs-to>4fSVRo&~3Yz`)a{H@<q!na=O)X!6A@1Mt5m^V>
zSyP>y`Zs9sp^&d98v*)OAWv_zZEREF`h010$xde8#~s|3kyhrBQ?pr7t1z_$>eDEO
z=Tp?gir{~tJj!_c02cO?`#3X-dFcbbTcuk3WYOwHB283rVz{<KW6FFsVni7zjCiLO
z5Y_e4ODKTZ<YpQjU+*Ig0m(r3El8Q6$^$bkQIux~rlsTSz*+xji_-+RVDOZ6w?HA^
z`f!`G=o_9W&EX%U>tpXVT4LRf0<M2g?#i$49xpSw<Y(X!K)ZRmFoQ*=*U~SsoTFOt
z@wP|MLK2Gn&Ve@>=Tj$U-pP}u#f2=ENS5iYQQH8}&MzxRqt_o~?(KMS3{sC*9nBJ(
z9NGNT4*EQ&kRv%+HDScC$)qxER4O<gX9hDw-+1kKDOw2Qa>d_n!zhIyn{8DN>1`l6
z8HouoNC!%AB_*SAkJ$3`l@0oB_veqyBr5yVfD;e=>CPIVV5c{D{iF;rVf-CUC9`uL
z)kgp+&iT*v&Bff_38nn!!DGn@P1%G=!1ak<K3Aq;`}3Zc)<UpvyXO6qS8fT8Jg*@u
zrqo28sSiKZ)oIc5^A4?C)L>-i7aq4?UtxfAJaCCz=nIh+ltnSl<c__+$;}Ne?Jt@-
zF1M8~@Jf@ut*mdn%+<979oR@{g10cyoW;Iui54Bf4<9XDo|d@%DNh#QhO$89iXfJ~
zzl)gQ@&yH81mx+_X4K{pFR2Y{XrLPO3(E)OJw~sp%>&ao5L6*mSsO;A7EfbOZv&eJ
zMC6^&mSZetSbV(GFH>!P=fO>dA>oB~y-@{W{K9n~aJ{$GedKkvjDSKR$~43D5481X
zl}+O)AfW=KuPWiM?>f>TfV+i+r#l^CrCwDiMhFRy*WlUvtp3(Wp$L{J77b<`bXXlN
zAn#>jZP~wI{7$h{<zMUWd`Edh$R%)bw%jyo0b+Bvl9G1YcfH}+R9`@PP+!RPR048s
z^cM__Wl~Toc%JbgiR*ao%3~D8)l=jodeoOgh-%ucDgp1Gu=NWB>)nSbRAsvaM(c`-
z!);cR_G+@X*EoWKcfWhgy;;p^SJVHz2LpD0#(Xo>Zw(NzXn3OTo&S3{0qEP`0r!2g
zQDaA&h>)wz67W*a_ec8|u8zem6JD}kZXAb*Fi3!zN#!EtWzmoDZ!PC#Z-X+7=9Ram
zaDi7GYO!5NGq07jah(_l%~x+_n&mRyYDs&3aQ+BY-4hBIi6{9|!{{|d_w(t$)~lL@
zBmT(6Ja6f!9A80UVO#?+2F=5l-Su-8nv|UN{nvVr-a-%e=r4>qc#@xJ&DIz~n#`E&
zOh&HkF_q=59KPOY)+WH&XvtUeg9G<j`b+<rk6AzJN6S>%%=coUQb8D1PUR!rO4|88
zXfTDKC`IA`eJ_jYg9?q;*?1#4WS2(WAZD<y{iLru-&xVjE4mR_L!ft%O_+V<_4c4(
zL<#?HWcP4F@PFUH6R@#1d+$qwtb&2~_YeHV{R7W?O<P$w+;&ecr(WsV3i3_#C{q@Z
zJ5A!La6PO+gaa~aHd&jST?_2U>yzuFj@ud(Rx*Ak?um{R5pi5H&!gY>*RAwG*BS=3
zc{PZMl_v}>)CPeTmSD;**|A@xTC+bEwq$Zxd#^kb>%V-@yas0Gm!H+~Jj3f-uh1ju
z(ksyA80Ww#YrER_k5%>-NZ#V_)u?47_bUdd|1Bl4zJ#+4e6~+A;WpvR;tq&bizK|U
zb3!ki5Y(xoW1`Qi?w%(?CjwIH{leJ7w4OOqnc-}1C`JNv<Nim-f7kNjY~ZHeYkYb4
z#UdQD?}6d<96KO#$KwssqTITFl++|$wpu)lv@iJM++yB+?m$IFqmj9#zi}4Vfye-j
z!JNixPA>7#WLaxo4e<vk1$HncL`hL`?pSU8&2F2#uD&2FE)~`aB~+2va6heRaj2Y7
zrvgiWHbDP{SowR|7^MeoEoB#%Ra{l60Jmg@t4Yc7zM~x2L>8sq<KURTKE3KD?51z{
zG~T|ZrXo_altuoR-7=P`L!<NYEy`D;h%ckMq4DwaF`aTK0}+Gu=L!BXMBG^6lf?HU
zcXPtD9?vpEa_KO0!zXz*a%%SsW<vk|DN|}XBLzCGnHPeOt0uC&+5!KIoo-j@t=|5A
zp=K*&?S0b3_Ks-6LoYS7ouq6b^zDh2aXNS{BEtPWtK;@yzlJ}j>2{k0CB6G=Un~^<
z1h-)>J3oC9<Y$rYxp`lc%V)l&3e3M}5dSf)sHtwRe>HuJO8B@W<E_Ow<YqE^i!xuQ
z9Jx!Vp#2tViK*4!H~4IIA^2>XdG|QuO%eXhQk!%kE&%?;)|`W*fP*7(-Z{QklM<#S
z>hM7~f3h+|V3&voXiUrX(h4xq<`0*tcBXK$Q#@xWMCP+(bz3Q!o0<EFuB2o*{)U-&
zy)^hSo*rPVZ?6ASwnT8cUUS?r+dDm=e@(R`X)8!C?%7Kcm0R@6k5H;2WDnu{s7gmj
ze$0EPUob7~(SMKeO36=!VT^`7B}F;4fqcgUf5bkikiBQ~2ADC>?iQ4DDfvub)fQ9D
zORgqUc)i%imxqsf*K=PRH)6+O<-}(~JJ4K%bqo=`7M#a!m&v7nx;lr-b1GqUJ5^h8
z3t1U1$S;cEOk9~-%=J-Ygnq_XMYLr>Gi~^+#Y9NN2H+#c3sG~z=AY@RlDbzl^S}Gd
zO(0^!YSnlEXk~vojK5Gz)3V48ephO;^cjC<SH&m7pkR_zSkLv+z7o~?NWG`<!29T!
ztUk}?wyCGP+K8sMS6sUk%H?B<Tgx@wBBpKUTmQD1G38CYwd=Cp&zzWQQ-Ak8FKEAB
zB9ttT)(u)qCCZool&xrZ^-Q28&TT7*Bba$A>b=(`u)i#b>^|C*u3l*W>_!*VYp-@!
zIGZcm213dokO#U+pP$fi;Lc>9u${MW68Z=w^}I~@L_!tlZsB#-@->^3Ts1}Cxy3~h
zEf53YOc_fjrq+rV^t(ZCU{Jpz=<1k<8{IhDSh~78J8ORQuyk%d_~JOHTCc>OVP9Bd
z==vOp7^V}&WoyIJ`0m<2dGTQ)S$9k#H1q-796kEQ9i;Zr6kTxZU6vOWT{4P<-aq{F
z`eCf{M>Gtsj%u{ye5pOo2XUxSP>(ZfZg2Sox@1Z?l!9aiH=C3~wm^ov(wob*wLrjA
zd9r1vvs;C0&6Wh-$7Z6_GP8Wy4gO?yfX0AY#1i)Ib7r%{sB)u>$?BX7KGz&A+wYx;
z6!knd-tmMvnd%t*s<G1%%>j?}iSD@O-`ORJH1#hLOe3P=vb(88k}G%0X+x>0!&Kv?
ztE<~Du)@I6Dswwt8-X^kvm0_A+y&a?^VWQ4^Y4KvF-d0imtE4b`f1r-cHHB@?GK$#
zD#GKp!0FnJ{~0JYXUs-*dX(YOHA%IxE3U3YNVoy&ZFG!;T?tV-f++bbm8cA#hyzhU
z@FYDE6GS}u*B09sB5=Q*n~rmQrqO7iH#+v0Qk^kgThv#v?!Us2SYH$8O6-}7GIuOI
z{^i{xoNeOLuujDqOg|j~Gy?Y?D^JqIjjJ9skN>lojDY#BTPyEvA^61T>6SYaaRFlL
zBo(M98v3Ph1M1{9y2P(pMX8m<4)Ejvl~IAW{3xg@M__M_9>4m@^XplqHP<RzFc)q9
zL*{XJXV=N{&Sb}xm+|d@D#GQU{F70x)Ag@U%@a$we0l~9`aDI%p94bvQ4mWnnJpV3
z6y=8xXxTIgcLCiyE?j@F>VT7;Szc7%vZq;7N}Y}0@^D}N+{Nid1>c}JKUx|7(8)!U
zoUiu>X_4Tok2eiVTSOOKsMVc%ccdD{T?>WXFbd*kY{%6guXO25RL~?_-#2keRP6qn
zNgAuRAVN>ZZu_a10`P||<B_$f&w#0O)^jk&aM=jVNWnHOgtK52%}vdb^Qf~28>x-s
zpAScNqh$PQLEnn?n*RLN0SK1gP`b`&D!BFnni(nKdH*crbp+WzO-l+^f}G*GC~#A$
z@Mj`B*Lx6i>2~c8NW6nyV8_AC2zMbr=1qiRP;(wGDgZFozs(axDw$y#|D&t_^i)I1
zRr5(@I9XaY&K4|Ej}`tbEG#H0D&*>w_!=u89q%mn(e1uF8Nn=GMLF(V|Bi@`x|G3c
zZ@vv<<qkNgyvtRHvYUcX$NVvVGdeS{VqUy7Q3vVjOg`C)G1fJR9lw<qH!n%k#+5Fx
zDK~_7Gy~RPx03BuonnO|rTj%1NVwiY|6J8P<Si(cu6Ey53^_eq+Nu4);azE_6m)!s
ztD3c%%HjATlERsS6bj_Xv;S6KRWiCUq3DW-H70bj5t(n<NSn6@zQM13DTyMqResbe
zb{$K_7juX292|J$t_93#x!UmA*bRSKY}S1}^Z?M3qvDz6p^VS`5%8a-HLR;MHQQ=<
z3@UqCzS-Yj-d(d9YiAwTl>k1A1<(6ekKE=ObHQXd@J`7bhDt!f=(&DNJlCN{6FhGu
zhVtXaE@HZ6&KK3FyWU7tG&Gu$icoE-?Pa<uR^k;V1%ybSsP{otzjqF71#3NnXAQYO
z0N(nFlJ8<GUUPk)lljg<<o~d>a$|qvbm@3$Nd_Kau)4FfBq7Kxmxx)ER~25t21M0#
zOxgc#a}+=@p|e;gU|4A8QBUX^#Gu}QpWm6^FGT#>e)4(QI{l?cx@$g~*T;DzSNi8P
zL-;y8l}*w_b6h}!5s^>LRbx-*B<r3JWYM##428P<K;;(|x$@d$1L&s$4kF#MClI5p
z>l`5`CGFk>GkfxuF&Izypy#W}ZDHW5UjYkwq0RNB-lZf8H#gg)wxx4;deJs6=Uy&K
z%qh#m?Z@*rya=u(WVZgd{lAqte-XHyU6je3+wLx<Q}oxHQ+%Ag16^d2(Q4N~()Gnj
zw@k?S((&2x{$ZPr_h63YE&DMIDWaF6{Iv8pDMqqY*skSzU=iMf5?@AB$|-II>>@Es
z*z$fBQ^bD;Dj!%{4D^}w4@Cu9_Wz!PSZrKVLGTC-fg9@(9m<JTx;njT0zDw364kR$
z2HAy#Y^Xn?s#e&MjKX~oUp0U~nE8ZY79TR(_`Kj=2$(2}_>GW=K{<8mJ`3>sb@x@j
zrz9+mn;TWs*Eee`+q2$h1?k2utAAe>@e^&lE5r$k2tbFY(%fSO!<W|T-(az<{<yrP
z@3b>fHcp(O+z78;#Ok5K<HzhAJD`vE-p;O@ySow=pW0ZN4)dCs09ukpBr_L=q+&Hf
z<Kn)CrTzX5gAB0<aY~4YhTUet1=+~`?bnzcJPzj3XRJXaMBn_W6Z7mlnvtAlFbZlD
zzhd@MKg(+-v9DyZTDsW5ekG+e5NA4SO2R_#KNKC(#6cHz=?2QD-(=#BmIBUwuKkZT
zcK-c6_?vx0T8{h1I`B9Os&%ZR4v8Q8Q`?f7#1R5sYw-2B5rEHGxVTihm~+26?hlzV
zGwqnSm|Jm816iWRm6ihVANR#?^qKADA$e`1)^4(~%94LUXy+^~8p}^h7-0W%7I2ET
z<RNa@XX{~u+y@hIiB@$Y1q>Ml#NMm<K;_y)8VDHRv_}T`zA7P&h41Obesq_R9h4Ql
zMN($?LhK1YlL1i$iljl?C%(&@sv=+eSLejZ-0;(6Eed{|l)p6hn75O-eAA=;Mpp!x
z*9N94E)4NvIABXJoPx4t&Gy-By58u)>1hzbUuQgaCL07IcgwCm`W|jq!N<OI_4gZH
zRRw-uCW08qspEWDD~f?4xQY;!j1N#_2OPH&3+h4;n|#lLb2DafvSl&ZvDpTyoFbz8
zT&>V#v$ApODNpBiuFP0NGX5S0ajwi~mBobuq=*W)2g?80o^nv<F?=uNi-30S?O*XH
zB}3OF>BA^aWn^&bTmC_V{P)+zuy-wWjkS1r_=EuX8W><VUmw|y90<+ys%l8hg$&Db
zgsH7tZ_z*K<3<m*xWzXfq)a-R-{37rOsWph=2B*tQSFRla@u$2+giaR)@T&n35Vii
z)T};}+Q-ghMrw79_%hGhE@}#dXNmB?dX^ySdxMNrynMR>>&w_96!KLgQ-cio!F*>d
zClJ69b?MRE8iGHo2Jm!7)$?)-m1QfY6ZNQiP3%JO$+po#kSk^FuCS1haCJ3iv4JPm
zpxHWt<!22y2$?in$eF%e-q_q#YP$ws9oTLC`y-J<?t>YA|8=FW|2AJRZ5-VYbrmlJ
ztu)AYtU~?&rpR_N(3l4$srtbnsU#4P6Y6)ei;fQ$=>72{><9f<&O(ZDo{V}oJmscb
z31$|9nI(XKakyptzJH*i+bqH>9nt7Y^nryqTZHJ^g*#APVJiyO-|A+fHMPFU8DB(7
zT3lQVb9SUvI^~d(ksrld{PSo?_5SsfnNm^YgIJ7##}>=YRQgU&W;MaHLm<dBqsNn(
z?dt<Q6I1#;$*TB)*H03HrhSvyjK&?co9Lcn_D5w7WAp#IO*RbEC7L}7@E&Z)im^Z5
zB6UWjpPl&=mM$9yA!6}g?Ljiw=H%`{<63>X>v&+uQWvQT`{;1QCpuYIm%l&R!ib(d
zX1Ih5{h4^;O<WTx52x24R|c3$$BP66YSa!^9=|IIne;x~%3_Cy@hw!q%ocNuqJYyV
z;VN3pnT+GJ$qd4xanY>+9?4tczFTS@pLfH`G94-@|3zb<*345Fa^d~?H#jGul(?*j
z#dk<vdsY0y4AV9|I<~lsv8*9B-PB3Q(AjhAX4>>17b69EL#*b}AmDD2Uo44ga}7ZJ
z2FtOwAXfe6Yn`4K<y^6>j4$@;IllD_F0m*eGCY}nElMR*ma1>HacN3c9#MZsdEdi6
zPhq?cPcX;s6y$@YdgHRBkLEt7EHP&dBF;c^uM4RbwKdWAxs*Lp5m#79k>2#j|7@Lg
zT-^j`s@~(-J(Cp!kgDC+5Y=ok@C+xIP2Om79c{HMEZel+daD#y^1}eC@!jc}VR^b1
z`3RW=bR@OE&P7`{!pegHl_yom*TcTQ6TBWZr|I%Upb;gZ8;5Vd+Z(sJ`J%T*haZ>X
zH0L$<ON8_Wm*8JH(%vbIkx{|*Idj3|>jUP6AiNE@<bhoRf%#uPi-hOfk~yXXGRiXl
zp;RtTxF{v({%b1jBf6^oG5!ho0a}Q-lyR7L0?o-Sw9{~L+r7(XOwiH!7mMKI{)INe
z-~K5M*aR32>3ivOvN5~cA$t$(1aP*-w1n;}kLG~OwevMNN89+A?6jWY?a?;Qm+>Ji
z$nRID3YWm@><MsjemUn!8T?=9kNMES`OuD?gUI2r5j454S-!i{SLsR{r7h%R@eG9a
z;IhbfDBFiJ@L(qBEvM`4R{8TdUP_1g#^2$?p7|&%Hd7C@n6Q2s%1%~M@p|tvi9;!*
z4f9o)x12(i!FCeiB{(fg&cS1(>b+w>`;Wf2b^(B8%K-|K4<Dk5MP-%$l2;mja2hyH
zVHi!%V^zZe=#32tjAqs68mg|Re6xBbw}rBsXy8Ahrp|m!pO-8v<P~x>pV<H8gU~~~
zJ~TGgTfqfUrG~g%`>5!aF+z!bpQhy&MrJE&j+cx!Uo!HR+TLI-KJnd&D5=Alu&nOr
z1`x&3e*bB0T*v9fmruJ3UkTj8c)7gBfZxTp86a>8-tbSmerg%%SqAUAJ70}NLq5&H
z!#Wum%@xD!a$#>`MqgB+QqlL4+ctYe;?4dXWmF5beu`6LZhfUcQUPP406T#5wZC2+
zgo`6xW2sWqSlz0tE~_RrwT7yX=w8T{1u)n4WDRWp^<bQV9UPHXY|&?qfwy0Z$8oI)
zP2}pPWm!`%Pu<VHZ%SkxT^wm=qSaql46CQi5))^Ag+EQR=B6lh0@WG#KghYp)zNxa
zChuf}*Sfxgs5^z;pj+?`2~&|4lZKl9%u~wln%Rg%AFUdLnF00;-=DlWg)ywk0E3T+
zC^;u5SzUpy?`Je{b=6%$SZ;N@(+pM>gp?o|y{dFk8xxb)J>{FVMPsGpf|N`U`0Iyc
zI`B<d>4;}n+1a)Zc7-TL?Ll9#L&#SxTDLVR8O7JTUS5T2KM9AnX{8~GTy#WymONzM
ztC&yXO8yhA1Vz91|4z}d{22z0lnu&TrndcTI4~HD$!lgzh#ZzX^81`fv{Q1d;Na{T
zq=xNzN<S!qkItjN*(H;>5)BpZWj7dUyDYe#1bdZ*%Q@^OK){1f{p0*xboND>PNDE4
zQn|@50>yTV>~EetMU&Ewe{NZKuM#phvj~iYsn<_r8;bZ;-3gxy0os~(=rRf}|GCd!
z{+(2$9+gmHFTr34?j6@xv+f<B9BmhZm7P54o=UP=xM64Dt0S72T*WuVmlZG1w`H?M
zX9PCf=Xi_u%YRYN`y^YcFd>&~2WxPposU27Sja74v2NU<<D9~?8+}O&*PHIw@+yj3
z=j(-GtR&<jU+P_f9CTTa2zB?A??X%_D>&%;k@e{g`oSkTRc6ykB-azNH;Q=&#~y)-
zbHRbl9d8sA*-J)W7Z0rG%}$_J!Yy?|KMG`NepNeRh|#^qQxVx5g454sk->gqlR>Y>
z4(=O$aOFE&wRR$IkiuRvnIv9%3lEmc%~#`N0ZP>L_1gVF8;ZuD)rx0Xpn;O*K_*)g
zUDRCK?8`EzV=q>To{Z635HO0!RzzXz1T*K6oi0#j5TGc?;l3**q_}m=ynJP?@NpTc
zKsF>A-oD!^<R>}-yOSP@c=Y+_TnXLvd37r6{HPiT@b>lg#jl+n`Uka5!Wdy`l%b?F
zuWza?{9FZyP)~}vB2%>mwGz_ZEW=4F_N<f0PRoCGbCD%kE2{t5ZJm^6u6Co1Dxa(2
z=<S~y$7)FXptU+V6GAM**V*q{9)EO!-7Ox{iGLi%RZ4DcAn15spV(SFY3#yrBudmz
z^VGmuhzU#l<^Df{BIun5A=EF`{>ty(=?sg>qf!QBS2Z-)_1E9V<>B+eUX{K!OR&GO
zf*(cs!fV3R)lB;rcO>P5Nb4WNfxlHW;M!#ZPaytU4MDj=%h;Vd@Y8~!eC8mIV^_p(
zI`$WicPcosS^Ka>OH@}(tDp*QhFRJz1UGj)t|~!spE?P^RM1Yg+-t>6QY5zA;_qQP
zE8w@!Z8f3(-vmU*Ao}OtfIOM<NV-rl%8ABOJyT3~G}{yli|_I`M{Ucf$0)<K=F_{Y
zlENl*g-_%q7=s_du?Kao8Ae8)>m%N5Pl<F|BB$svg{>7Pds<s10eoJ;(xzr+Ol6TT
z)FC4`lF-5NE8lu{REZwGBGaBlI0?!M0uH5iMNx5&Ljm7l&!)}S=8v?AnDmt^3O`ap
zwwlvk3;I$dw|!QO{Im=GBs?<ZyX^1D(#Q2SX8d(Z4m%rBSBn)8aP)6`JZA{3Lp?hl
zX<nDx9oNSgcQ9I0E~FLn71IF(25z1EH#e`Vu@OSn&(|LliABkL?^LbxAK)_v)Zt6}
z^_OQDj_b?k7S|qxW_dPqFrjD%KO|IL6+|I?e#MOMXH|k@V^w5z>7JJmY;*bs>BVKE
z!n0v$)NOC|L0KNDFML-1VMzto>D)8XQkt5JZB%BQ#V}BY<0F@U4-^qA&4c8%)(Tl(
z()+#t9-xX~Z1>;NKgdo{iw2VOx(oc0E*-V=a8~QbgqJ-<ZVmNkzhsvu2kIOCeji?s
zxI8%6Xjs1YX(R#dP=-eT7)T6#Fo`vgA?2=UsA@^)i5Gi5B<qg#2p5*|dO>2~qLLk|
zJG=gMb{9Eml+j<e%FW&Cvaz&r2!2_ukSx8yf(>_(17X8K9<@}Te9a!_>ceSNbg};$
zEAc_+R~UCS3a(SdtGc^LBk5|6UAGe0uSRsk>lN;sP}4m(wx{L1w~j_~!j^TSnd-}u
z;)fYmFSeyg??EBl$6W~+w<%-+aP<$n8P{Jv*Swh#OGf<gR*(Bus=)n#zd!P`<Zw9O
zxqn+3ePqPEfUxr~S*B+_Uf595mC1UyGDFeVy9!~Sh*;kX7!o<ks%?elYMD{$7^if8
z;wZF|{MG5oT!DrYbrIv;*85nvzqqW3Jc&|Q8hTP~2S$qT;=BZ)4*H~OzR`(!HrCwC
zkIx{TOd7G#>t_R^;`9&^xRW=sFjc=YEi~7ipC)q#=+M#o|3!aXsQ{+|u}`nHV|IaF
zU#CBdpN&$u(BQM1FPXHaBNtgkct^s1^OY>?bKL05pDM`}R~BdiqgJe{Ddt2dKc1vb
z>EZ(w6z?VVDf*#)b&rCKj$=%Bxsi0CT#2i>5pyn5&*keM4swnxP5OuzH?L3olopf#
zg+sF-ZeN{w5c36QSlcSz<K>kiO0l-K3pLOZ73ZebH98^I;CkCBL3?Qhi12VR(_pOF
z2!qLv(@xU{u~h75p<H!w%#+pgAy?%{i=cq!AYSqUH}~s>YuxqKByct<h=Se+;}7yp
zIsUq_fTK<8crK_YW^sA2Ca_JG8nlido*o9hIoShuBf!}uNGcsi!jFuAlauQW&>25o
zZWPrY_r-Tr>9{Pi!H7h(6ur9P-+@~7XrZH(^}io43hI_@MaSpd*dn_;GjZGd8!|)M
z3`&`A-^Pyl>MC~ykXrNXwUeDYXgSU{_JJWum!r$e-ZL<MV`s@14_bk&u@rj77)bbV
z+j7okT-8bFf%rC_Jl#t*JY-$ySNrRqW#9?KJTo;qJkr7PVDj+8ZN;~`H>-v1eANrr
zl-}n$SB|p@eBumsC>tM0`R6!S{|eXYo2QR(!rH}K9&&75`<-c!@^Ru!<2^f?jH=zz
zC9!xp;WEaoiNoc-Srz~zbc~m<Fk9H7&ES4Q$`uJE_P)PH!Yzjjif%~v#rETuPZofu
zI`bpg@ujt+*}mD?wt<Ivh;JEAm8iG7S!R<*k{|BxbrE2h8v|1EK%o6iTi0h<>J+;7
zNkk8hFh&+ZXJ<+v_}LtEipQ|Tza06_n4qpx_lfhSaYNl|FTkt@$H_!4em%C+TxTnL
zRyHl#3+gWK6N_t*4!I19a}nJN{ZGy(7%|~!rtI1T3Ruu%f+|d0>bSO~si|^T>diZ}
zm1A~fKLlPx<Q=VtI?Ta-QKc{}KWeTpnWP&$aSkzs{T$ETm5b7_(rww`FEV7h|9AMP
zbKan9w9x0*Ip8E6-YcqMK3RiGEsuU{bBEhvyVs1FM1PJ)<{zN|<av?n|8}u|pJ1c@
zf^mPHogq4u;zfiv+{Sw55L(=*$G{?E$0xzd%SbOuq`^Zy5&?zr(|A#mn<P8k)YYYW
zmqrR6zHJJJ6yL~Ge?^uSl8!S#Zm4zev|Rcy6Z_`p1BoB*u+d`THggCiC4C!UfY^CL
zt~3v~f}+yg+#K5BnPYx&y{&3)F(@Z~G<&`ml<V~Gd{7)qmn600z%nf=^4WX9vtu?5
zR+J89<9hZ9IJ>?&U5c80zN=JK@<Ge(&?+M1g|50EE#&A?T*E|=!-rRXwC_zpd5akb
zs(2#+NOX~P<tC;jesjPczSG}v*$4iv|Lj&PhO~lBEl77gGU^~~cRc3cP}^#=!9Bj!
z17(wcVe@w}INEITgv;NKm_y}V>|}1o`Ek_sm3xPLesR)CQu`)wbY^dHg#|Y{Z!oh7
z3Em0UR^MHy86=^hXhZF1GY4;Ugxt-MW<cshvf3YZTEp^_`LQ1i5W{3gW0=tPS0lKn
zL)~8EU6GL%G#YM)ZBgLB**-(9_xGV>|0AWD58xwJgL2RAo;SHyb)5b++sbotCH+)j
zvx=aOVMD5QEEtsujC+mTemm9ZP7ZU;)CkS6(4Ae1p75#1&g}-Tw}wqI*P_2E0zM>i
z;%XIWr?s{lVthi9m7uaidto~UAto#!@xC(Ood#eE$&~F|9=5PB<JO%`NB^jI0aFUW
zabJ_~<HsCPpGt-?!wjFP5?RKJB9^k8C{%cE^T$dJmF#w0HWqJpTyb2l{uA-*rk&Py
zZl1Py40*er>s?A*jG`cx;|>`C^rh6piC9uH>-A%ykocE3Q+3%8)LRO~Xa_PLOXeU>
z&Pqyn*)A%bTSFtuOerxhiS|dGPzlO$C1}rXyU0k6kdRmE<Shh+T2<QSWN2k(tb|2E
zvu+wV|1{BRbyH^ydeAVItF33FRD+Z>B2vexDy?uknoI`VZB5k7)D8w@(oLHTs@><q
z-K3lMt<Lu@Wa6;Ty|X_QIl(K@OtrZW3uWy`-f*i~<JwX{Pu!V7MxP|$_5+<L>K}#R
z{qI>0F|;{!ZoFy%nXKHCuoxW_Uf}roMkvl8f=%{60at^k)`Q9aRlyqrxOcmU-{5Z6
zymqYhwtV45F79Ct$&3LMnj%A^Rex#ziS7N8M_Oa>69r)jlNAx16&nhNMzPDeFE-lt
z$LEK2>1kW5dkMkGL@bfHMwCkj<D1L)m#~fas>nEQ5n+a8-O<!WU4u4!@T7&A#S@N9
zPFHQA=@(z^?MRt^8mp%N(@1Wk$_K-b5^S`}QkiZxHYf%KUX21A?H!jMMvY|Jkpeh>
zWfDiLVnM%uE?F<(N8w?PTk@&{C`cV0*sKnud}yTNgm6B*)zXY<ZV4nzmp66n&F%Rs
zTyAz;5(=(P`(`;h3QBNK##qniqoUQ-7MfJs<7zk0BfnC2i8`f${o%&`)&9nQIoQ_x
z!<;z(;&;W&0ef`Urzaif2N}}ULZBg3hC4`9Mj*6tsn$t@QD3yu@>{$a^&m#TS5DVq
zLRR{)yRosR9xt8gWjxuiGQ;Bs(X{ViMBMEW<4P%YGl7lxPA5Ju*{Dx7RUTJ&`Y8}G
zXjS{fE_^MEmT-mL>@P3>N@Xf}9aMSuwNqN)`nRhZKs+K9t>c`w=5FJHbZ~jKl%bx$
z+@IS1!|h|2&X~a?93{1;ljhcAqA}CPlrI#4yX^9Op$I3mq(0R#=ecwHuPi_6gfU|7
z-!EDRd}Ow$_gFo%Ghjz&^+BgTMzB@|9gGzON4P}oBAu*17Q(=}fXLop1av%{{n@y0
zzTyS&s-tD<xlsnbt^ThR2`cwTYF@c4$ve6Va>E`S_P5BDEDGi8^UaP2_Br0vkins}
z^ftl1K9GUr<~(<DpCm>~UhqKXu<M4>VX2DN^nqRc_buyn>m<rD7u<!4TQ99O-?$bj
z+-FWo)cJM^dGtbKf;x#Ps|*j3#LG0NXu%91^ZhizcNRA2pghP``%177j`)nqs%^}5
z4WrDyU6bql3${P~v`lNrdCm(9PjogeZ?GJJ@n#4(>5zx2|Djb1?waJ($F>>BS4Aku
zsJg3ud`%1+4rSq<W@)CYu-5cM%VE!O)^jYKsl10_c@KFRSV%r{5q@a46!TUy${YiK
zdP-?0Kmjf_+X3s=>yO-9W{aT%4t5Q0{7r|i?c698l`9(35%31@L@;VSmY7~gd&YuA
zUPkZXJJP5;q_qew*o-CM@3FQf<2GcO1Y>(@q7fq?w_~=E5cQNBF8JNRNu|*9Pw;-6
zg-mA(DA%=|VfU+|D$Xuu9qiI+01UbC?VC|nTh?2JCq(J892_O??N^6am)$<OpQZmJ
zFKoycAI}ahHakCL^Qiv#Yas7rdANMKzT(OJo1wu^;BZFbDd$$&i=P-75E~4xUkN}$
z_fB^S287Pc0R;E!?Wo;$E^N2|`nhPN8w*mA)b3tn$X1jmTa*<`Wcx0Ffxbg(8XkaQ
zMVEWsO%}l?SKUSn*QYXq4Th>yyC)rIv%*T(7g!lu?mG(NpR%}rWqtDPjW<#kEr)p7
zHSLg>+0Lc+_5DX))ZJsn*5nFTSDS$uH#Wj5f33{!v4@#d-$vjfDImY7NbR!HL7|uH
zv`!~|=~#-9ou%q9B^uCNW24Pe+9JfojoeU|I?vjIK_H9ohTxDf4Qa9aixR>n0xe^=
z8^$wlNBcd2H@+N?yDrD_QW!94I1L7vNgbv}{3Jn3owsLJZZ;OVPJxwgT2eA9TFZun
z*o#Uza9VZNB?r#63U-mLLY89G?6=Bh>fFeiUCwrvUOMwt0)ss8Y1j^}EXu&ykRRP+
zU?SnVmtT64h*+Cx9RN-*cl$o!!#V${H9;y(_&&`22rx+y_}Jf48ZEs)BEPn1+E3Xi
ze$YE8mBjqb3;#ZICMX+i3_o36XbZla=JMflYC!P+a<`xZYjv^#im~Y>xoF?u4KKYf
z2JqK>#MUrC{P~v6dxW}kWp@AXkDmo`m17mMYGvRoSv9OegE1j9^k-)-9{h?~O@Yg*
z-#dLyr4W>q&kY&e*QiA;Y(S+3C8QJ`(|NHW`cwZlDXn#8Y`&B(v6c&|bweztYQH`y
zk2r%Q|0=bLgu&m}UP_gohwPud_DM&qIA1hq4vPY0s>BZk$o^cWbN~t)L-ivhkbtBr
zjjs>sFQ1YV-Ix1=v|;P+`Glm^*!x>pZOm%M+)Bqgq%XuB{lh^W)amglv%E)-rNY)*
z?PToMq}t(7ha~|`%F;p;{-i#fE50z6RI~l=vD!k^qE`cM`LlgnduwxZ!17}tP-cF9
zp%UQ#Kt-po>-eVmdXe4KY-gdZHQ;309&2x|9>WTqXlwbjNs!r!lo0L<xLPak0RO89
zH@7>5Jo)Kp1z~Uf{>#<!U2yEQbk?OMiVNBcB}YRtw<BgJb?~RZEj8XLM%E@RPtjX3
z{e-8{vyuF0%w>zV$&{v7;~nbsoHFYO1PEu$9Ux_vu<Z_9_wBZWN=c8Wq@sRYjpGdB
z+_yfQg#4i~n#{Nyj#_N}A$m*o=cZRs^ChTW0ZuRr!31za9Ka6;b$=Oz^8PqK7EuQ7
zI^@=A1+|28;)`(vN~}y1(KHIuAnxtBYy9dYQmruS(=_wYWI5NPsWayH%?BnoB<|PC
zyik7)7qo_thY~@kp{&Wy>4GTR{O3bpyDn)2BYE;gwUkRn<e_Z9Q?BMqp^&ScwY9ZH
zh00mg_=Il!hy0a-dW17m(DKNub<=nudm&Wg#UF1W_b(@^aVoVF5k>eHjk=$l2YFqM
zUwDZ7tg)MI^@gw{I||<Xl@7q1G8lZD(*(y0DQGx;{}y|XP^vE+J5C0}^L#5C#Pqkw
z{S;BR3e+kMH6|-_{WsVG$Vtl4n~jFab$X?&&Z?q3jfK&RozWux|EafV>G8Y9jo)FJ
zMkjq8qlO{$g<WUcUXxJH_X(A<iYI_koszm5hXa$3jq>e^pS&(EgaF@Qr~ya9&V&X>
z;Pv_3aL__94!Jy~L5I*KQ+bEWt&5@eleH(8V>@p9ugrID!qsdnGq^ybwd!u4Xj+?F
zc@g%Fy8uGdnIA#<9;x@eXsmM0fJy{q`o4%!u*XT#t8&L40$G6uzH~?Y=Hq>6^@5=c
z=6uv8%lsCZqM@GR00^NBv>OJ?49OOu3WaoR+CN(&++#J%VB^$3`QQJri<M;Ej1JtB
z@&EUDJ5p~+>^tqi`J%92p3Oi*KJ0P&x#urr&IM1~=<}KNk%%9I4c@HSJ9`W!FCY#J
zzRO@(-yRq73uXxq3bw3oUV(bEt*$rOKU-?uWWWF6q`k4G=nGK3r{Lk#WfNHhbFSRJ
z5&f7`*m<T=*Xd3na;3bpZq0U@r^Hp3>^Xe<d~N1oQQO?(ynFfi&z<M5{i7@qlifmz
zHbaANIAGb}!xHkRdszP4R3F$1*CvS~1oB(r%;HD-D8BVO@{hLW*pIZL4)>Qqe+Pze
z>Y4H<#){#6Ebumys4%;XGV`D%noQMUopMOR$CE8kzDc8IoIX*vzc-9*@vOetZ40?v
zBM)gQS;li*?e#5O?;!stk=Z{hGE_V4B2;a*(G6=+iTJTuZ$*W$L_yg<7(esQ)}ARI
z9hC;o@RJKcgzmoCsO9=qO7boY_^NDfJ}l!cF+~A^9GR>)O(fo<*Gv)rf+|}$RJHo)
zl!A~!tww2aV2ocPFR$jb{FCniS%;>LpIJ4>Q_A_3yUXrRpi!j`@2bT<h&SYV>4^yA
z6?FS0CSke)rGB9NbGMA_OI9q>jjgpjZcaqq?h&tRz>#PoX1_$CSIL$B71rce0QEig
zvq9j<#&p&LTM=6pdxtn*(%xf!YKHatEx%(xqL&_Px4=MnIBG1_%lr2SZ?0a%-J8Y(
z2ic8+?i#$1J8}-DJm39_p|U*xS;OFCZ%+nWs&+9|3?riw1$xdx+2uOg)};EZt?T8T
zjpK;lVi^HiCC}99qo+r-?Mq`0*$Y3$WTF8>5;gCYp!JGdAGzLYO;3DlclKd^dH*{!
zXfIccX~gahZ&1r}z2a>T`6HTFq`y<{ap30W#%Sv04}&?_`!3y-4m}kg8&_67d;z<q
zI_c}E?I6?a&w{ixa5kTapd#vO6*{W0u*Q8Z6aB2lBPm%g(2_dX3UmI|vv-lOVvwu?
zgvF4(&zc_2Ua>XluS@rR*l(!iWC(?=!r;2Oh2CitlcpJuoF}X^d%Q+WP4ov}b~{Sm
zjBMQJ&Fxq)%t?h66~g`Yww40A^WNo^2$Anz%Ea4!i|%>}&-$WJwd(hwPxtm}L4xl)
zs5PmSq+_<n)0Ld^q3<t(7u%DThv_iv02hk)Yisj9TRJbt@f?1oDVMrlv+`Mj=y4uC
zMEe;w7q9+$_}k3ZqdH`2QM!QZ*7rf8mrgUi^;2%tY-_TUZE3k$dLPH-vnO+z&TE+~
zpW~00LE~i<LGk<(U6g2|+S3&paAaRnvb=Ftzp$NY2}&jJ_Fhh%JpRIf>$M63PLAh#
zR(sq!h6+bv`pK!c-skI;C|z7_KIbTQ6qBf(P`d5t&To0Qk1{jx$<swBYptzgU~xND
z@IqyafEcvc72)jZqUICrLOxozeX*!IxqECPWD$IXEdU$xt9?gfqo<mRayT~7p{bs$
zElDb6S6r23ws5xC#pC{^cmrI0=N5wlscw#tYXXO2kiSADoEq}A+iamD0RL|d)OEYe
z8XiC~OQ<kDLAHie7c>l&s&LpgP*Qz$P-2_ax(+s5$oc=cH>CR{s{8Q@XI9jokm@?r
zkuY7<&lfZ1A2rky*tr<Q<;z+JoGWd!i+!`S^|1Wzh}J22JB{B6z(C*w&DRbY>Wqvx
zkh$Lo0Uf;KpWQys6aD<Nf*NS(-4uRTLfktl+N{XnCRY@rq=xphp{)CV-a^K<s^Jw}
z60bkAmI_ccJmMv$%@00oDKh;fF{JJQ>3TXEu8ihGG}c!+{zUmWk&k`rpW#qdc6m-E
z<-gK=_ua|-4eJuSAz(-!q}dver5nI`N_@OqICw-<^qjLxB=RPfy`O1}X9m6Ludzj+
z|KO7I8~gxusuY0f#IVVVSK5;&X+5+C*#I)6b~rY9J~RXp-bq2LvZzoXX{knR&27J;
zvna7TJ8Qq-xPGy;_!e)*%ns_B)_!+RIA%r2?)3VaG-#rL20HWdBG=hQCJOX~ui2XG
zvj%fBGjUJqJmgC9E0v$ZCHxtM<SnVDT1ttrubc!mRK6=uvC2n5U`|4ku4n$a1tq?7
zT4&deZ$8|6mg1tlODk52*4??6Y@7wJMrT9rk`k`3KiDw@xj?Z328O+hOFL5QoZQIu
zh|E6cl$;CzG762G3XjTaL9eq9p6PkRl_|fgP?xJR9HB#L@<4@SHRYfRO(2Yq><R6U
z6~4$te`(qy=fbGQ!>Sfds(7YT!x`RhTz?&-`*M$EAHRS){zjC!3T^wQReKZKereQN
zhc1ZKUbud~%OfIAypS+DmBSW})>Tnk@zexUE(?I`i@sDZ<0jnErqKSj*VV!`9w>wQ
z5s1{LoJhW9S{FziDw8Yo_-&}N!S^sMpWS()6uz^h+h)1rTdTlDIE(%~QSXe4xz?<d
z`ocD?0K6y<rz$-8?p%ZhJ}ru}@;Zdsa{2q8pPn8j^M{;Yl#j_J+Qe5(tThOdy?0@;
z2JVq>U);A7HPPCZPp?b`i%P`($3QUQeD$v!d2fKR7;7`r#u#hp8+CFK7&Vs{r#kaC
zdD~@3EPw2oVNUYhTg_#R5=j^xM%8%Gh+`p}ptm%dfm76_7Uk-+QtD-SFE1e>`gBwe
z`f{A_x3!+*kDcSyQDK<ju%Sr~ZgqAO*jxb4j*h&Hon3Wps7tzcl9aAbN1rPx<|dB1
zY4X#Hk%)_N`EAHIhaKt7!v5k{w?{$jSIA|MT*9aX2(aG%TPVBd+AcJUF4m6^2NOn~
z_y}lV-)DCHZ2K|?39T(4fWZPYPW-ctLJf=}8X~Ii@&#;n1ngoVTp--GdlVS$mMNZ@
zn8`L0)V2OFgO3U+$uNF>55{?zYxAq#b+&nk`^gXT;)$>6!D*U%ZtP*mC;YeOa4mn=
zI@Q&otlm2mK0MzMPUYAE>-gQz+RUSTm0Ap6@<p!(3n4dDS2}A?o`9l839&EPk;-a%
zQQbaY|7ht~uF+}?46xzm40}bs6EPXI!57x#gd1Y-W~xk<g-aB9-a$qwBn<pPl+W_H
zdOId7++fqUOzd9#&W2adShz>b4Fa#SXafKin;)RbsBElh5*N)vG2%HlA(ANDd+H%3
zS<T&$>7-``WKOtZuuTXm7w1aJp48D$$b^kGTi|=7V`cq>I=jORdq!nFAG&6seCiJV
zjy|8YJ{6bC?D5grIAii(ku}@OhB7T@YmEvCM{cu8n?dreljxi3oIAGxE2Q|Ypizk`
z1O~>MmzBmMT&)FQjx1xI`meGpvDvcp3H+#Btxq`!$IX7z_D4sX+zW<zOk~%Bzqc0f
z=Lt8yKFkg!{2xc>{>bF}|M5AGG;^qygdw#`EaZHMk>rqLnaDXK=fiT`M2sBYi8)k~
zL#ENeHq5CU3mqJ$FsCuGOmZrU@BR7y2m4|7eO>qK^?W`a30^E^(wvDfws+~}w1K`1
zZfB`_ViSF<#7kZ)vR_BR+*dW*-`oO;vLidMxOs}=|25j%c|~g-lEe*5DWrWqP`(ZJ
z1MZeu?LpSHcU>}nK#QqEFQ}<wmBd9#X?-mW!m4|C@cp{lf<B}(7%-IDiyfNypo|9P
zLrMAlo6@GYoyKAgy&+|kLp)*T>Lp_bZr^bBEd}2NUewm<u^-c#O5V~9D%P{7j(Pp;
z(}%<qD;uM)h;veHaZ}c&7cUmVylHmwmo8oCgd{Ed74^yR_8R|_f`hJxX-1l10ex}t
z`{v1(hhwVy{51m`u}zOc`Z++OiCl>Q9W>ScQB~c`VzI*7r?XhO4%zU{8E^g{+$Y&L
z%=xv<?!p&UZ|<+{{L1t{e);*hqjZU+$peuT`ai>kiot#=PNih9pESNb1jbJt3AX@a
zo61O|_5<X6^5~%B^ZT8v5*vQ_Da?f~owY<QLw%X-UgrAxBE>PnHYPNUv#?WgZfANU
z{3kF!4+spF4%at)4C3<wJ)xlk$OdOcCtp>E%x+)J!v3(PmDXH&;~e$XOab*1dMdSJ
zAXu3qK9q*57UQ}HNNR#Y#z*);+{>mv9~Cm3w36tT)cK7ahwU?pBxhNjYb%@Q>nktV
z32|N&`vwBF{Na3pwvbJxz<?{pn$-Bb2LAIM#)fsCTLn|Z)0``vF(=A%bO3y{Ot>=p
zusyIp_qQK?VcHREPc8ZrC!#jz<`FXWO1uZ#i94&Y<{K1WKr5VJbtY=ra7dI9G#XaU
zXIXn%g6ZI4bJ;AT-ukpI8}9AFck=@yaGtTe5*_H`O#n2n0WQvJ9PhOd)sV3Xg@ASR
z$FT+7`mRPNvh{;Cd(l^Ce#bM%qw2blx3ZC(DzgbR+97Z};KAuR_O^h!HXb>xa|neS
zX`LpFTq>Ntrir&mGVxH?)IjlyB6@Nq_(_nGi&0V$c27mf6bdyWFXT#s`*`3hS4_#>
zoz+x?ASjYgGc@9Yfadm)h*|MaC=hB$!_Z2icmHwLY1F#iQe(~T-|L@n@h;R@mq-M=
z!wSJ4bA*usOFud_8_kv_Kqnntabzo*j@?!iDv54zfAWT_rY1ZC{@)jSAGg3DC#Qrn
zwS*^RN1pQhUi2{`>7G{b2#KvNN*^D0rcRufEOt4NIW{pmFczf<1|+hISG?At!>APh
z{vE#${+P__(Ic2{A9aYqpYB)}pHvGAi&aq`UgIj5sSPlssZ{ZOmTQU=MO0c8e7$j^
zO0<izM7@yRgQ<WzSWN<KuCKkXAjYGO&Cg=%>3ONpqsy^+uC0yCwuH-8oob1Dzcv@n
zPczV6AT{$u?c=t+%(a(S5UJeTS;!Qawwh3AB(m`UC8>jX@Hlbp?^bwgOO-_cnGqZx
z=~P3$kf)KAR(%p%%)vFE<8T1{XY1z4b*x|qGa*XrQ(4K8n$u*87g{`%x4R>A_1kes
zN^bZ7a$cqi@g#e<T1+gY@YckDIF|mX=HvQby*jVA-e8FH2WLxm=~gcvsKm?d#HX+W
zPj_*x$)n<N5>Vss^S0_sW^GM0WmimG-;BpNC=UTaG?O%Va2;uBmkt#bQZz5(8J3um
zWTd*u_iq!SA+X$BfYE|da9+V$Au8D#KxJTfqx9UE?~1i@BmHpRva)^!G2PpWr9BeG
z?1;1dy4XKg^44;;AQdi#=(%ItTqiJSX~X|BbaD|Uoab%q<rD%Yi6lOG-P1u@X4u_N
z_-79B&tAqE|9gE^-O;hY+;^Sh+uFjEwY?OYuMFyfBRy-g2#RS^CLb>I29k%Q6)N+S
z5L78IwH&_yNXX(9%P3fM;!F^k)<5VNPB5PBUj&}eIpYw(V*5Hc3P&uyBUxr#`8+9=
z`0_*ZdwT<ZtqVscMhdEEKpQ;zQzsJJA*<!FjH|o&h?qwN#U1v%cG*)5g%mR8JNXX!
zE$QuhresR4gGD;8X8xQ~$K2=Q=M9z1zJc|%Ew{t#E*!Y+MaJB+{JX0<Hw`<H{nn1B
z=Ki0_N1sjK5lzM<$I$KZd+~Zt+<eAY(lq2LxnuI|m%$uCL5IhN`i9&9gPS3X6IoMl
zQde<Gc5kL4D1co6<ugNe)>^HI4!!n@Yj6oQSH|sk_I0sT@%zAI2eJy70Oca^-hs=f
z7db0Wp0o|nu0XW2-@i!Wta*+(TDtHK53`1~t)3X+R_-_02TEVMBsFXd^Rhs8$4}M)
zp@zbR3vm-w7IpTH6wD=^qKlo+ga9v>JPvdjb<xq-E8P6MLb*l0L!XW$3r(%^;_9Tr
z7omTsj8c2cUFcAh%XHnkue+%<OYL61STBq4>=`90Xu_T*@YK{BR-~&4KbYQX(G%g~
zC7O4%;ZQdR4S3B%0wt5DyyvC%7Ho~~85=O~xD|wdwH-zcbn)bkOb))Pm1n-iJ?wB*
z`NWyn+=?wx4Otw!+NsU%RC_tTva&KkTU?6#yT(fEtrnxjw7QL<CswEhLx1BatVFRR
zODCrO9ra(Wj_13E%1Oy&s24$#gmq>-jSiZMX7;jziLaB>m!0pExZAs!;%L@}o&pp+
zNFJq*T0=?SUJj3sywwT+fLO$R4OscCsnsoe^>!v~huZ13ddQ%$xu)nSq1g#zKUg8Q
z<&fd`^Vf=NW*)&yG@yB4EacM@Ifzkdwp{x6e_vY;=<ulZNO}Lj>LYt|*)g6XyQZ)-
z8SjqOgP=0KjBn@#{d!>);vy3{(2a@qO4cKng@B9wB@rjOU|d0{lcX@BpAleMk;Y0a
z{Bx4`6z=T$*2;tyztX(aGd{RIcrDV3J{$&1&41JMYBzO+<cJvqDlb_`ZuD@=SwNWA
zzi>}`HM(~g$OQVjMR;J<DU~92Ww~Pk)~!d+mM{dJE?NdIFIy;3-cy5VmEt8TW{=yC
z=1Oba1EPJw-?!sPdiai~)Qm$sc_u$V`VmKPPFkE|T&QX#e-)AU!Ii*-7fLhA3i%?&
zQNMD%WPd@0<JYKn;@lV5Ps2diw7IIX_LEgGXpth+1Pt@O<|Qs|=P?g7nd^^4nruo*
zXgk~S5ITS7Mn?ed+st4(a{o80cQK@%tDX6b_>d70I54Tsasm)tqs8`J*^*184!90I
zg%8ci1t2#^al66V?~JZV7_Y`H!L%8_)31Z9M}=Q$B+zf9m*z-9=?5ep%gR}E7H(4t
zeBJ1COy<FcXO9EOuFOfP$<(z*_rE*QuEE#H$6~yBThBbwZ8#1S|B|i-6{wQAGk|F?
zsbWew#Ot*OO)TSkB^SF;GC;^i2)DBpoe*R2BqYMgHu=P>++xBF9fH40A(f2QG7+;v
z0HF#bT=AQYL@*7uj$)mgpN%dlB&&?pT_Mk9gMwMPPh`bxZBIHB-KvQ?Q`x`(Apmh=
zbN%j~wTTx9g*qMH)-qWaVolZViUa?5x;ag1yZTx0cT~p>*9q$DI;RR8?be=w(VD@P
zoy0DBx-`*Q>!Jp~m?(U!`!QS>p3=w4%>iOMMr<js`5;$n(78&&^XD2EooX)opMg81
z3}hg8zA~k&s`+7%+XVUoa@Inpdi>TbNm*nbpRjAQI^ZR!#?MiZ(R7G17D*a(LYFKS
zAsU~ROxKM8PS#9CxETY@_Di|B_p`Io;4t`Ui2AN8NADZd^h1n{(5P}bdkXF31nSdS
zsajjuyo<dj#QR_J{fQPJj1HB>?k>F*OHFLcxLI2`bDH-EE>!n#e_EQ%p9?n&b^I0i
zr&r8Tr**^?UU;}!u7=uH{ur&pl5NMHr<atUNF^9Yok3LQMttN`h>E~iCe`pEN3G2)
zhZ&COVc$n%LF!uUGC^02>Y^+8$Xvf#qN$dk-b(MkhWqfIFjZ?B8(j<2O^ta#v8XBk
zT&jB{4bb%9Vp|8-CF~=+NOxsG$$ZYpL+YXCJ_-u<E5CSzHX#Y>+u41_?MCuEAfmXj
zwFzOss*f~k_B3aehiN5F`g21-y9=)9;qqf>hP<<^1i~LKLYRB(wyJf9-W)-R%%h+R
zLhRfd$3Uc^P3kB~xN|u8NIKg$$_w6;EBJA^aUKUlj3*iylx8ok`_f7tr>@1*-|Ex#
zkR?SHoa>fCa@B~GuQ@{EOCB$5!iCfY`LRyEe#ZO?0#cW-BoT2P!NaxWhxVLYN%ay|
zkV+cM(0W%}={Eipt0BVLcY;d3QBU)Ap__d+`=zE}92nj-P&q0!Czd1kDFh%s{6RC$
z;3BDNQ?ts(&;7t1sQJTS2N7}4zS);@a&oq5tR&u>A-Axm<?vpN4B^h1N==u-U!QD0
zO}l^MID5&C*bSHlt*<DTN0IICV-&}X3wW~-RZ?CVh)a(ZiinOU3Ucwo>k}*Su3BsC
zA9()M8L{flCwm*ip6!7_lHY>K1dT%ixC*F!X0o?2SFZ1EL#<-2qc!_aOVBmrZ%?)+
zsLd|5S;SPPhU|6-M*&QsydHf`VT{Zke;6@sH;Ghpv*1)MGnzWVmj9YyYHG#Ojq=*^
zJ$=AOr|t@13(5XbM*9OCnFc<G>`(!<?X`q9k}l}{Vh8jtd=y>i>%tghYU8|Qy=t4B
z9{4HTz1=D*utAp};wxi%gLPTXhwxq>EaZb~D|^6TpJfVFYB*O2brn<TgddABEDWlG
zBL(@x$5wBiRIijugMW&sr91@YI@YKHUjrGN;owXpVJSf-(TMYkpsD3M?wzsQ%cuO-
z@gRw`YwpiY=|cS=#P)J=)7S3LRWWAb$&iD;RdMf!6QMyYwMJ5jY|bvs>)UB!4Mx)L
zdxetv$z90R<<Vp6jrffF0vYK-q5TN!Vd5{a$524fp!hqgTYce;XU`{tY3SmNnzzm5
z{^p{icHifizBToWF)DKwh?IXsSG}4vpwXNO1V>`*l7p?k2Dstwp9Qk?rROinr_wmF
zhSO}{_nmkdei>oB3{U&w2?^|dFx}BfQ&U6U=hC+F_Gb|PHA-;#R`5(SO&$C35ye@+
z6l9|1ltO+V<4%%21cpvq#xcj8T0j|kM0STw!hwyrnqajQeJehp$74?2Q0@;AOk$gq
zLS~m$y3V=&zqsx8e?hVFbDk2_F~B?f+^*Z+@3p1bS<f@I$(w(h?C)E6n+S1wVll)B
z8zh89v=2|4SYfEZ*;5rF#z&A(`6Q2a0k+u85^!VA@W-wZeB3z+7qDTbL!k|s(pp-N
zy!Ufh%s}fmux%5tLNfypDUT)8SSzm^hN6ez-9KPBljm+jz}xxLr<*|b$(|#@2ba`-
z`>1x-Kh+}>@2^wUsnM9vBpM~&`TKhfvEXQCQ_WJ#mSJ@+%~n52gkXLz{J{Y|gZRX_
zxHyB5%4plBfdfG&y2_S^97z(q&)`RyPP;Z>#7zEh=i1NTvz^?fw8b$_|5%|@n~H0o
zAOz<#gMz{o0RHNq#r=);JylM4ZTzB-n$T!l-yqzF&)myc>rmqrv>myIMmc9Atq!Oz
z^xALJPQV{Qc8Th0-JU22&DT^Ek&OuE_Qx6IM2|g);yBhX(!Wugj*0Vg)!?C=ca?f)
z@1A;Mv&-_doev37-R;cwR{IHM5oW&QCIjF5ww-j?htyl6bHp@9L126(^)<OlWpVyC
z|Lx+yJaqT*fzrO*=`dMP-;bWU@_-73(-bGWi?CFEH28ED0ceVEy4(r$g_$cWU47}?
z5BOBR;ygs%2>*<(=Du=$RRFNfKaFvG+XjOludCV6q3<-ZC@SSUd)g?WA6dAI$ltMU
z5n!p*+240UO&bIbj(ir!50ou}wg1~RZrp8@D<B^RPTKx3t-ot&)$pOcPzyvrEgd71
z_BNW4ro*9nfg$HencG_Uk@o^vLh@yVdigtOD%|Cxsg*F8X?#Nq6Zq|8;kxRR#PmEJ
zsCcuYf?{Z(H9D}o(;vzHhueEQ+uy=i_dVR{Y2l}4Z4%*fKd3*NKE<Hmy=AM3LQ2tK
zh(eY9=(wMAU4QchCqQlno-4D4KIGdrMi}=72S!~RWfa#SB_wpq666lt>AkaqZTArs
z^{^hPXi6^{vDdEJ%S65l3RG=NNEobh8rYwx|MX;vN;+lGXQ8BkgS`uRbLC%8@>H!q
z@Qls*v&jg>q&UgeDF(b3o@LS5vb@sVHpO@t9IbK%v))o;k@0(D!@^?!?*?bH859O|
z7Z1KKN+ywb<f}@(GIc6T)fH+^i%$vZ`Av@ooW9FODqrKXzm<x8dZk9-I9O8qUaq}m
z*T|Z=g>%5})~_BMp_&yeJ#1b;G=)`YhA!1UV=aMgkX4V%)#kh4Bjm(Co1FEr#fgOs
zh-e=kVr-Ev%=a+r;<i-iKxgm=I#RttOiT>-d#$sx^NUQpODeqDGczW%Lv`9cIB>Up
zf9(^E)=zCcqXhb_t0+{OD`OHe;x-G(Sk_mIt{|oj591`OpPv3|KOM}8L6{jQf#|Qg
zkA_(Q*ZH*JFfo4xO>;=tHA`V+7=fjhnhVgk^%NrifHzJlPM~AxtqC!EQMD!hba&MW
zDk;!nX^kb8{X4uBRR%0AN{#@Z-Bh~r=a#r7uDI*e2^1Ebv3={1WcbY+*v&RXd39G>
z=j<A8Y4*<7)#zlB{N3Wa<%-2FfFM-RsmQgYa0PpM`FKZq0NGxIXe65**0;VDpfGb-
z(p%m<vNwucnL$`KNcyr`GqY-1u=`|{nbkxRgdi*W-ZG1bxw|f>$rnG4oI#8^j|G*u
zs(hTJtglig6LJv>oMynsGO@|#rp$9IZ{xUz*?+=apE6HwV&7FgQ%veqE}Kb$npzD9
z-^#k!>J&1La*Z$(y42F&kw+Bm=qM=y20yz14wMn}FLd0f6w;{{I|J`!68T<Tu8pcC
zM@Ky>a^ksmP3XsCRS?(MW|KsnWC!UzGL)q$a~32_Ec0R+nWW+xm-}HMT+VAEj>k?E
zuEJrUicB-Vy?Su_u8yXuko{snqwvg)j{f*Zfi>e&d9}d}7>g!LS4;Xpts<X*ah(?%
zWbw#qaWz@B<-l1P8I+R~vz5ZQ|FI5sp?&Nx#Nc$Z(csUv=;ilcf-S(SzZD_`4kgbK
zG82P#mN=aAs!l8Y{hR0;Y9x9suF;~sB~da{Hj(q)e8lY0N}R}DJ?Y{rZ=%S-FYCff
zN^bYB#6GGe7giSE4CYW7@HA#xS{Sjw_s_3|+UGhFKPQKG2`_7Ow8h!^_M*rhSTcdi
z&{Y<iLTniu+&n0XkE50}QLEUiTUX4|q#l*=H9iXv;LXT9M}A-FL=JMrL2PZs##1}b
zi6N@=Nw?W(SH0tMJ)UD3>GTpcZ;>0;z^8I+OUoz)1xeDD=Xpm|a8Q~%+SLFN7u(f8
z7}FP4xZW!!CTFDoqbiIXrIK%NVk0#~D;!&-0l4Dz>BK(@3fsPqr`;-!`*ssD3E~PS
z*kNQ8J^<V)6(Iu=115`d48M>oo3SzUx<H4dx>TVbj@oRD)&4J4qnyrAgC}EA%#|uY
zKcv%BI;1A5$BxIoo{T!*`aTj_GWKWlFEC#^Q2yMqiRq}>qx}P$mX=0l{Q5OiR_qic
zo4?38|0&`@Rn>7$l7B)I??3PlSkx+TbxX9(jaGLEUHwSU!2v(r;8FB2l$U-0xtT0W
z7&sd)(=)rWLK$JqZ`)r}FwXhz8gN40%M!r{l9UMlC~PGxj_+}+a|)z6Zbd5*r)MLv
z{je$Z6HLcSry8VM6gwqD;=JPaVp-7zOmOPg2gj-Q1@_KXd4E0%#cBC5#dd;RX+c2t
zU>hLM9m~hrGvSqek!i~9IOWhEN;gC0uU*~7qYyc}2@15=l%RVt4-I46Bu8>%BSnpk
zW+r(i+v4LMqAP<Dl`M@4qL-*?wkda#<vYQkXkgWc_f832QeM7ga2LDyhBz4v3_*Si
z#iFUM7|qJ~$%83wEhKJg3fx@i45i65vkUKvz?G|IH(ihnWFC@FWlPzci&%Ri>`9mh
z3sk(0qCn4UP3-J*UMeHnwN??12!m3%(HDf&E-9FsD4V3l5i)9@18n#>F*+MEtFh$p
zUiRKc8O2Me$Q?)r$LIRHsk!jRICJAJ=9erx!-TlF)==>{7oR8wh`mfIo8$Hk!1r)f
zb8K2xaQO1$xaPrr?My_avq+=ldT_hY8{o}}OQaV|ffUUh#f9t$NLeuegt-5?Sj9EU
zc<%Ep97uV*m_ZqV4Qbn$W+dgMo0Qqpd>@6-Tzi#j0=tZ@z&t$6QtgUTp0HZV$awp6
zm;w*+`Nab=;0H)?J_$x;8OHbyu1#e<%COy?m&L;Dr^x}=^Ioy6ndrEZ!tTr&#G>cW
z__A{)gSNTwD;z=VuI2BA3o2XXrIrGa4mHtdq}J9{wBN78eEdWB3-d2aekE#E<QI!5
zRA@NfTi@KO-{0Kd`4vqspfbj5?*$99A=|4aDiz=kFzc%|mNWye^u*z`_2civ$7?^b
z38U@qtVFqpymXGJ{(MgkOz<GcD#}6V?SKE|P=bLa*ito0!WW{G!_w{sW{S2CW3{4Q
zH#z0`wzfWQZ*{v_J5u`fIC>N_c@UV~;{I&4yx=T^Bb=;~;M}=j7lyBkQ-i5T?9vyz
z1gu+3+L!{=(?`JsNBj6)GxT`G*VWYF&fsxt<Vhc5p{t7PgXMft=m7pI>fB;~we3wR
znP)Fp+yR@Ibf5Y<NVRZ%YS0QGNQlB09jSwkrsXWwN6?~gn5L6Z9)agutc*SgBmttk
zsQlX6F;~*D^T_DJk6iu_{IWzj6q@JjUXsBhD3Lcb9>BoubahO7#(-wH3e7>j6B$KW
ztlyntM*I+345Bg;nFt=R3fMBTiduWT1Mw#io`XB!qAU%i=dL9o9#gk#Py7G03FZj%
zVjp1l%?eJB`QEoOO<YUs?DSNsfN7h_rfDC<NdH$vdL0)jWOD76&HKs0hm$;3Z<$f<
z!S#z{{VO<Ou!+UUCB5%fj{SaRI@PrPT2jq7jg_rkZ&_@6VvGxQToe%GkToVqfG#Pd
zpV8$F3DB-L8-)prjyuD2x>4{*UTJ?Pyq|IDsM|=zSp^YLQgV}`$;@Nj?5%(PD^(^S
zKjkiC8G?1g<&z(=st4`BGgS&Gz+}dg*K*kbhmlVW;1ND0au+U*cho$j<*8<JqubAH
zmLSL#QTERuB(2c~7I_i_&7MRkE2-Ne*O|{e!AvK`@h+z?lNH5h<h$T}>YyrEcG_@n
zj*+2}kt!!z*T8MD#6ta=`iiiioZ+5-0L7VB(2FUqsv1fYgP%WIL!aW48f5CE8pClZ
z{#-I;h_VHPBDRRCKYdkP5mLE|3gtD`xqZh6@pfHPQ^CH>;6>Ln$CBogfGr^e_KUIc
zx<#e`waS%=3yxpU`L=yzXiQcglr7rwHq2bR0}Hn_G8CO?zCB@$j&ksmwugoH)l9_O
zbtV4W0GtC&4IztvnhPj3-{L;6>fz?Hr(P+~K)y3AiWJUVRi8G}=i10su@tjd%brca
zSJnf-T)~9I(f++S?e4XgyT4~&<p;U289ua<W}(63;j&ej=|@@22C->hEj+=$&icN;
z8Nb%ac~2Pg#qH4%(;$kbiM}iZu!B7scXC3_?=?Zc#JY*Or88X2AKe`~gNt!E6w0}@
zp9TyC3?E1L0`wKW8rNDgM}iCLOR4DqMr|xgMd9&K!-T=qAZj1iE@K>E_7i$H^i(CN
ztl@&SgKW5b3kB^e9QL*Qt?1O$e~mG%T0c?IpW5C*RL#{jki)W1mdA_L)a^v`@~j*K
z_>9#bC@*s}1JL8t;Puz;IPXbh^{!q7Ho_@r2pu$(0`2wGmd!?<fXwz6s*-rO<G&>)
z&F*uCH)}s~?|aEZN_#N0Xa789s{Y0r-dvm*50HFds2L$|+`B!ZcD48Pg^(2vP3!7G
z{;jup1Yy35qWKEl#-<RNQ`Hx<@b*@w!_<ebg90T590dF+1GI@1TzJCY-^AYAu_63L
zjT2<h(*g!+5@CME`cYRp5+(0k5i{ZbX}yJ^Tqj7ed~&E5Jww1#jy50SNdgJ_1uHBo
zgA&#&mC5OZ^AUY-cRpV}G->^XyH5OwK!s{;LwOYdzYGI?=+HH@vix^p?WX2@5gVnZ
znf$>+#nib$-t&AoQZoH0p7{`Sbuh$<r|gu52&jYkCcaQP!UK|rtDg}^4d|p76|qax
z-%rZqf}gssI=ZFsi~=MDM;>n*<hJ&)058@0TV9Dfzrz503h}zNX~r&VEiK3B-1^_0
zm0A*i6}&o}8Jp|+8Dz`_leeY{2woNll*m&dr{_9eCvia==fx-8iW`)T)jYz5?5D*{
zjT@>6Pe<*k3$ZQ~XY=FgxW9l1WBgvJXlDzB6V%2$H=!G+^2Nb*Jlxef3uqgP7lrk~
z&u0oh$>v*goCtp03w)Ra2~0*iJWkCwf%0j(UfEq31F)4%uu0MdzNnv&%NLE6AyUN8
zEuZ*W3kM8&MR3`+ww-I~zuI`nPSn>+j}Dn-kL<>_&?a)dv>!ooa;ocZwinog<h#0L
zpa_(7PgrTycJf}^3%b$i&x4ea%8~9DmAOQDrK#0zRaI4r<94Fax(0w@conpAx^T!5
z<#+>i9J<71V^{m%uQqP!HGCsK*%*qjaF|c4W&wvDZjRVhOpmewOCV2NJ2BURthyeE
zaRPTH?r*Si_D$LyeiG;!Jb0*=)jzAR++!C)u_5FJ=LlJcjMp@#XDEosp$1@|O?f8F
z5V9>if-^EFt)Xrn$^%bxt`X4W&vIYz1M^mF>cc3P5DGb9@CHv?T3->#5{Ng6U;O?%
zNgpR_ioI1k9_RmlpJm>LYosR+@(j#AUJ8<85x0eOxnjeO{nPGi3-NBwQCi=fj!TqH
zy$_nf2fRGY(~A5ZY?Y924o-@*4Dwg$`i!j?*AYL-ZVaASxX|9XQF-E7lfN5nVqxd!
z9V?_p?jBn^1@??cFWE?4;#QMg1%%Aaqj2^!l;uBEXQUACPxLWy$p=y3b5}9k0IOhV
z+r+NZC{!&B`?I*IX7l=KzpHN;(i<wZ$;s*tGkYP-0X6gOb5#Ke=3?9=U~t&ULRNU3
za|$F_%#-ZTI9}B>Q{IScYe_n0yq%xkIV1{t_}nyHCKfG@yt+SKgqsl9pEw=shYJ`-
zok;NrX3N@(qB6mVLSM{8l>)v4t|o9v{#}@~WzGKYjsFd3JlRHnH-^_DcQ*TPv2(+}
zGybsF8iF{lCmqqEZ>{RbI1>|RpuIBn>*qyvs*x1ues+MjtO<DLx~pq}*+>s_>JZT+
z)6SJi4Vv&9?a6e!$1B2<yxNOP*TFw<6z98SqoY_tx(CG$2loM!%VWO|r@|KoPK4ec
zZmb;(SLtug=kC-{AT@Nm($8PaqOQQj9CS0qc;baKLAJoU%kzl$i;}Z}2?jUo`=i~i
zXi9mQzM;!@@6`X4-C1iZmr^$o&4kaa&E{i;SfH4CnK2S1f%#rAp%pi(h5v_NHcHu|
zGnXO<boMzwjSRH6u`yB+K9fHOP~}&*JJl-k3BPx{#Kif8kcg<gzng`$v0F6@cfR&^
z3~O_@cB5P-S6)ur(vK@JoB>cF2&P+X5?Hm|WhKS++QCX08~N@uc~?#MY7c0LE``c?
zD+`2eDr>4OXe{L8D&9NyFGXg^bd=?nmse#&b9SqZ_BM?E*&Iq}UuGTx>B*Jj3eacp
z;+{`v@^=GFe+^?Y2KYy)-2=x(^LRvcw~}Hn@g+%m^Cug6Oh)DbMu|tPkq5e|o@KD(
zVN=wY$06&QVhKe8k6wmdYAvQRs8K-_uN)du01q|6%IV=?VMl0@zdr8ML?a`F)YOF1
zDwpy-j4FEn0a*uA*%D&(af7Ktc6k462zwb)#<tFsVJNJl3&@01$VHiI_-Zc?4*{_V
zpB~$5DcFh5LnAH)XQ4Eq;`sZ)v_C5g-Nr+?@YIg{R35(k*XKBM*#^GHPCXQ%kgZ!B
z-58sHaGSrP+k%{A2s&Y(8R!rP{B_Y{{~$p>o}}m9p_iIA)}*&22Vd0pPPP_EcsahY
z56l!V;wj7V<@)6YE2w)bD2T!%&+CXE)R7fq0jFqo^>K-o8#f0BjRU_qz9}KQ_k9_)
zu3YUM%__pb)N!P6RipVMqU;2f%DbMjP-TIDtAklTUDb^6J>(D($nLZ1l7bza9C0cV
zcKGjOnOP=6<B-VY_JV>)9ip;-WeYIa)9mAPKoA;dl9a7;<qZC#6{q=vxOz_B(HNOj
zr~-(1EX-KY=dhhPf>Kl51W=coE?gUSb|sT~WK5KNQrj+jN(nq<Sh~=-NhX&vF)&<J
ztIiWuPS7{^Pe(G>euLx;6K@ohkYYQ~l-pj6byjSP*m+MM*>LGJ*(Mn;`OvQi(=YPK
z_kI?Lx!K;j_fM{UZ$<72n|x?3;2dZ>OAPBHyxFRc%;QsEkP<`6VLQW&j@$0ehTev?
z4c@oAYOJ_idtT_5dsq9(jd!i(+tA;*&fZHVzk!GP_1WVsl~vVg!%G`@FrqvEd}95!
z4CwP<ph`LLXs_@8xw(J8Kg-TRl>a55(>Y7f&Nj6~3I>pa`X^m)r@XA2pk4^tsvA#z
zjNAx`fJerj<ZC=2F4zO}`iP;)$v%$86rQppD0$T-MAY;<pIf&==;b;!xz``zG^fvf
zGTH~I8UJ!{JHxo&t*2wVh}gnlGhQ=XlxujgxWmjW{mx~lN$zV6{O*fPu!AK)*&!eY
z*Z=c6VRG$a#%k4xJD+M}Ls&9QI~mJ26O8j|ZXY`vcYpF&l4Xd~-}OJ?2_wQ*=Kg~e
z?E0KJ{tBjkaiXwZH4U-an-ZK9HX#1R+mfYSm4E!@1(mYz(Y6b}^sWmJI91%nE&Sbb
zCU+rohBY9jKm*H#*8Ji`3JPkqbm5ch7qVqdkV=$$e@nqogTw3haP6k;b^v40b5@{n
z-7$Uet<&mhClhtf<*d-O>vtTYQtBR~-V!I>ft~*Uuk=xZ#&c6&8})cM0_Y4&rhKo}
z+oSTmY7jWTQG2}Z%HEyWc-&XZ85x~l&FB2wEnV58c&^ap{d7MVh!8s`tL80QcW|KQ
zets@*UNIGoc1e7|J4^HJDR7n%2aBG)CuM4qJY`)JAaqmh=_$CUw1nlGRf|+v&O-E~
zMFsmS!g+%?4puPG$B2TdFk%tof0KdnRfi828uLGFm^Nlzs23gzI5IF_CA#8d=|#E(
zO-@fB2$ICBD4F)P_9;a}5<|&2C25{5tHDNg)i@g6bQXFvrs8)DTWs<HDImglq33m@
zwI4ryg_P*fUFu+o?1u6tnP#LMCJ+b=`<o#PtH8-Z!^Q7xuG4&nt3HqD{d*zC^s4Ba
z4;~nKGvpw;i@^bU<JfrZQdyhb;a6UFm;o-N&2&|--l<5@kN=MC?bVu`e3$=j?PV>c
zrL|48^XiA_R`=0-VsF@3?Z_U#K;r>VLZ3K_^Sbn=yWziIIs4I*-K<1W87Inp%6hJS
zCMjql-fgUZ^L&<`S~Q@npKvxKtM?6ZvHT#=WNSR?JI;@t;Q>^iN_)bHzxK3$$W|?>
zRlx6-jF<q)4tu~zQ5YjDCS>f0p+%{<Ucqb^w*UQv`@_Kj`-;VB;Uqks#U0VxkYVLU
zcmxlj$qAo|ycEh|clAztgZqBKJ916Kq&UJwi6O6`KivhzwSg&Uv7XQeQ)Ls5_i4Xu
zo?M)+Z9mX_d|OAmA}Uk`&#YNB85|gt5fwKD6F&3YLY(k|tSyF(C;ZzXr%$taXTYg8
z+qQz+K?B(0yL!!WeAh|Jb@cop!yNk3(ptr7zi~$yXmPHt!sfTOqm55_PjxuYuQbQU
zCa#Rn{ewY02}z%UrXbdP=JVLYlx(-bX<S%=-*!-zKRJl$?Ac9qQV-LDp&t%8c8T9`
z8l%L=HBA+akB}^Cf@-a0Uf6bn(%~Td16`sWt(HgP9@w0JisJguEaf})sJ$5P%JhKV
z^ruPd3Y|Ly@XyDOW^GXX&&LTsq}vZ0`cu2YGgHU&%$qCL{5n@<BOdpJ^;I@-a8c$a
z*Z3-6u+o8ssKFER_wd}V%p~LElVI@R`=n;vpd<NTBI(%orNc0nx4jrEWMPbHbhx1R
zRmBZxvR30*C~82;+Yy__SnqbEjF1#8zG9s_)vSiUqrJg8W^eRlykV&r8ehjIl|U-G
zTr2bJSqYJTR9E6ILZVwLu%-@|n$2DIFm`(<+E?0}_Zjf^t<SLsWK5;QW$bgMbyAe?
zv=;pDAK|b;j3|3K8SEgt_IvhMIM2o0eVd7rD4qY5Tw-6>kW@hN>4eFErRB%77(>zh
zebrgG*!!i&b)**qeT1I^__tO^#AlO>Z$%ls@`eoCv~;n#ppg-j85j5O+%kh8P*fK8
z%E{?a+3FZHB^lhPe)m8(QFLj2c;S-~;DqW0#5##hXHT5DAmH@#_@W8Y66#}uI#O;d
zMW{x$(%W}caXYL|mN%@|X=S0=ojRVo)kf0dG-SEu{|#KM9NUt8HZY9;UfBO<VPT>D
zA5bG*__I)Nw5PfFA1(r8T|_sIG#cdtf$pP}uwOqe0^7PoDmB1qO#WR}aYSNF%-+^S
zh}8tZZVl+KWH7vBzGz>Y2v?~jIdirCJLZAvph>o?!@tNr`Dk*{2R65~N32qlH7R|!
zwkJPO|LIfxr<=TM|FakdQAQc0B3%WPoqRR4w^MF!qx~=LC*4Rqr+w5s=Z<b`1QyGw
zW0FUL*CMnIbHn5mE{@*wu~=euGNJuC^M#AAeSywAdfq^>A;iCG(2Du?s7zBFTl^k^
zAo*NR;&~Cx*%H}_{_yATe{&d%UQdnicy%kDaymltp8UFo?Qo>Ftb--ICk;yrGbFO2
zzMU&9IgM-4Ilb>K*8B9NJTIQ-NM|79(_)~HHcFHioA!#*Sv<D))=5nPAXVO87CfHA
z*L3v7LbYmr-)%f@XKxLP5Hx=&<(v7cBy*&U)SHV*%ZfZN*w{qURQG}Bv7H<2YVes5
z3nYJgY2iBO*WZ@-y>PcTy5bUCkE8NXW4N(`G7cnf%0Yw=4u}hDU#)(-e9ntZKC+pj
z#J@X8>8vmTC)kcLEZ<vG<Uag9*nAJv2PO98v%?&nXq8mn8F=b2rde)STa<j;bf!Zp
zSq1_qCbsljx}=sF9R`(3+!xE1=HKxWtVG2Je6ZMFl?@eM8G~%k?3*-pdtKCU!#y|-
z(n)xSL}C5O=7%BTy71{tyrVUf6nIjb^o?2<aM(-1Z)H%md>q|$I27n>NQD9^OFEsk
zX#0c^BXg<1qY8Y7SgnywUfKz%r8mWX%h(KgX0C8{=-YXzQ<jY?B>uBJ&0(WtJ-&!h
zvIZ!#uWP;!;sY7fKhp11=*)saZ~)J!jou#Te~90#bZnj4a%26qfUsK^$oQgmCCv1@
zEa6c?Qt)`H)Uh{kRTm?}oBwmF8yUK}7sdzr9XHumet0VtmU0x91TO!XKh(MQ`wnh<
zkd6h)H@W%Fo2rYOF>b1^h~<#lu#m>B#M!A=0;>|2Ljf=5`tQGgQ$X#w{Z0Ta!9M2M
zcCOI3Daw;4Pwjw&4(QGi4ey}A7cc(AH|eYNAGqGy9*4#11_Unl$F!=zEmpOw^9Sbd
zG#~3pzR=iGM{^Ax$K&sU9z2j2>`1|X_oGT|hPQb?E)_oUQ>~&i%&@H_XM1=lF*}JO
zWe$MDh0g==UZ!Maj3p9RV9(IZSzn{CPXGJ)&uEfqWJ$1Sf7l|CH`Tg((K>LDKWnmd
zCtsC@ri=0<fkS(x_~cEXr9D><$OoKr(`&>scJ<M%)pt*(8+$3qBh?N-+4eGcVlRe!
zZ%a+Is$+ocG|I4UcAFZjRBg4^-2u34OI!`>b#2iw3yHfrlekAVSS;4owjSP6e1T)o
z;O4x~;H=}ue)9cr-Rj<qX>w7#Xl0k}5UXuNaSF;Hu)FwECP!Q;LnBO55n=|t5S6~P
zu(qd`qCt?qBO!b-w@-%9@JYWUASfW@!7(AVWSbSOG(TxC3E?pyR?UAGVOI6I_{pdK
zsBe~0A{R~Mad{o0sMN3VVsFOSl^W{&Wrlap)BZb_xBZ322g4WZ=;}t-*=pq@Bfy?8
zi8n-ysON<9=9Jcv!%H{emKI0}V-a}wXEfqO$T9vr_`zZQPY%+laFJ_49t_Ka$c&$#
z*HD|zfqZB1qM*$0)M26R>c(%-a*(+azx-YH!9H(M97{j`Vb?DX7wTZP-Jt|8GCecu
zc=@hcO=xYaJ>O~?Qu7DR45!6I5QbX2A|BfRs-hIYx-tjzh0}`&wWsAXQmRO`=Z7>P
zX`S#wlN5?(kTW$1d*XT4FuV#yKy*)KR%h0A<!vKCCbscsC=^4U2!D3v>Bdp}tjR-k
zBUJ~+1<0ie@z6$cAgaIC{bq1~zx|kmlKhq<&wV5W1G5($$UKEgO8-*AAr6_xs_ewO
z85uspCF-`H{o~w#UZub#1w0Q+r9#19^hmcY*SKo9{leL+yhm|2OKNc;4ZJ_@9L4%&
zCV9^k78U~gtJU`X*--t0|HRTdo9hx%H`vi#G*uMAhY{YSG6qB$F%##K!es2BK&pTf
zgdFbFKcKE|%kt-TYJW<j18+VZTJTq7Xnvg@gur9mA^s`~E86JdAau(d01;?>^e!HX
zFNI*&wX-<5aF<Vu@vXAG2urnLjI@xU&zqn##kHaF@!Nx9)BBxr`#*11{Q>Je|3@Qz
zmy>I!pIaO&RINLhcO8mUz}HBW9V4Q1PM$PYcI5Yz78**O8I6^rJJfBtcG5ZaCdJB6
zL4B4cUAG~wQ^lkdwde<05msLgPFhe%&TrUtf(`1pIS>?)($O`nVLg2I;!ZN9U<l2#
z4b@)Bw@&E1$@}B;U+w@dt`!gRhm6KY>D3oKvJp3#kIBHs{P^n=FeNI@jV78bgPoqr
zeI5&_r|Fj#7II>xJ2ftQWL`{xezy+(OasRq8(r0FH7!(--iqvmAGDgq#0=70eaa<k
z?jmqNDaysk!(2lFI&1N6v|%J5``_Aw2($105q1%QWQyfHsEzM6@gimVL2J?Qv*gC@
z2{W%uyu89$$3UkEYvBQu1+pim_@0``fd{9jg5e?OD>SlM+BQ%31)f5kDF5A8jMBdb
zP9h$3b@Cm5a7H&;<{K_3Kma1{0!qMnXWqm0A|7S0UsL&0FDishbJo0N;vPY+96B}x
zPw}Yg@L)Fif<&w-RC5$dTjD-nq)023K@^vnv_8m_+?D^Xpns>oTIC3zaY*1fw5uop
zx}0QeHlGDQrm57sI2D0q8c1rFgxQ9JijDRCyk9G;&4=8MSiB=KNL$Roq=3=HOx#ar
z$zF*^&o<Vse*3wEaZ~?~c<0O_XCdnSm8}4bRUa>?N~KZ*f|0iWv3R~p2{mJLs`F}#
zN+Bobi966N`_+8ADO&q5uh+?6IS4gHNiYvy{5rI`q*rVk*a-J(5a<hx#~(SWg`5AW
z)N(#<%k#$u{@H6)-8a7<4mxHJHUU0t(1GCDJGxuAv1^!Bar%DqwGE`KKwgzo5Pj;g
z>e!+^bxCcp)D<z-FEF`?(Ur|<<u|plCSTYEly-l^-3lu6{|i*1=&V66RrsHq0oCpP
zaH%aSLZ{7$AwI6_%PEVN`MbWS1U*UInOIo5ufj3fTeCf%>8i3VH|1cph+BxJFN6_s
z`@11V@jx_bS8l3(7lPQ_*sMqQ?Eg6+E(e@De^@yb7bX=RE;5i})rERDB(Q2_^~fCa
z^pdo|VG#13O5w*~exbNPhUvA(udRmRO8gg0SL&p$j~TXcrO)Q6eG*bT)x<0S(9G!L
zgU$YwLY$R?NnV<#s012ErcK-hSLRPNyVv!{x<YSTAVZv{2lV&XR5~Gp->|MIzT3tn
z%j35*G*Et-CwgNto1DF-UpMQV&wvW+-o3WX_A6sIEpHYWwcg+^MjBO{TEf+$R5f^c
z8owEG%7bASK%s#i@~7ryPRV9tq^s=5n_tj42PfCjJdGOjYP`5>TIv<$*@90aT#6US
z()NMw5&f>hJUY-fpjh03^})}{bTvskq2D~bB9s79Du^`RD}nzw086IBy)apKsy8~$
zrAIESZy8I>&8ihIc|NYfWOUB>SJs@7&n<bV(teIM5gSM%bi9*D)&$KxIyBJ@=Xxv$
zgH!u#{6mfzAaU6a`j6Py9PCZAi@%$*5;Gqu)~bX!A7tu1Hzr7x62BIKz^8!64LUkr
z;0yQjvR|)L7Z+CW8gIb#pGf(9?<~dt+(gh|n`-36e8%7EoF@QUa`A?x%BUmXn^2I@
zPX@aGgyPT&mAFg|GQgFMDx9+OlJ@Fd=Ec{DOY$}t;98FLE~Vh3a9;n_;`SngrL)b?
zBZIUOTG4j#B_qizEiS{U!44USKB%v(<g-8rr~b-B9YAsQXQB7zW5f0vj2kk|*A98-
z!aAW!-S}XxC0Id*cyxA8t)=}*FH*tbLTih=;mu#1@z;4KVzBeryKYaIIOtbXgZQy{
zo@Hn+4thPM{ofXV`qsm?GYg{J0-3b&bMHD69!$~=20w(yB^cn^ZnWS2gQgI_aX31c
ziPc5K!It>|u7uJ(egh}dh+Dt7-;ODvct^6B+28wYf&&6bWvP9p(vV(H!zxdSdpiqJ
zPsMe3(Ivf@J1H`C*`Fupw1X>0RNCDOT1K}|0T-IJZJh4ospIUpauCyC2Ds#l5#X8I
z)!P7DJ@IdU#9(dK5<R5w;Wb_b9;3p>YrJTlTZNL>d4KYoW41I4g<N<Qd3p=axbVh+
ztN-_2c#MlFeBr`-ffqMjIPb!HhHqi1uuuZci`H+m4f8un_gT^k82{VME}CPm&HiK2
zo*3>eM7J)K@{|@`zfL^*9`szVut8+4K1mt_gUB3jtkH~!Fk3Brj^;hX(^5Es<~@yD
zooTC((VZH4N~{m4-I=$U?=n;}Y&f-yAn5WUbLWS=zWTg97VH@PsH=bjT(#rf-B;Nx
z=zpOgFMdza$%Hp8*o)WucJ}aGs>%d)zIZfion7fEDpW!CZ(XpbF)}nw=R4+&?GI(3
zCoDDPm3`p+{-838iczNnAOT|uW!#jj`@f1~ifZ?t#U-;dtdd-aR+M&A0+jc_+Ci%<
zFHyf1#ng@$n8157B2T~l{C!8x+k?&VeNTBY5nC!0aH5RwRmH2*XRkB`=@eFpKukpO
zBM+LiwLLiL3L~aV#8dnx8V->&eVw^I2mq-o+IfvUU$0>mQ({tFJuh+G>>!S2FFr5`
zKC7k>`l@Po_Q410XqAM?>Y0t)gDK<1#Q|grBUWG56x_T|9UY=lpps!8hcz9xiGoTd
zAABdk!bh9?hGb-FIuV5v6pXrhm)1qzl^8OAhlKo{2huE?E(pAc!hId3<ueEC`10^n
zrX?{D9??>lIMH&!!A#<?9WLm*GJhK4cr{!Nl2i?|_c(F=TI0Eib#g*{^WTl7&6XF|
z)<p0{Fkgg-2I{f7&VZ@=Q{!W)`3eV<An9^kHY;pi<HpB{6JrrQgU@+Zw>_V<a9_v#
zY;|cDn%p3j3REMQMw9=zK%@?+=-+9<`D@-hxA@<cLfk!QDiClA!98{`Q7Svv{2BV?
z>E6a#NFnD;<MFDa50aO|-Ck%OE@C6Hp7Gswbw5`a<VI`xxA_T2XNty$dn@t!M1Ufd
z<U_xw9>a1^Ym|p&uOm|{FiNJPMFCUPdy8=Q5G}`&Uh(VF_)*<(kbKY{cOt~FWj|5Y
z7WA~=DfU9oHS%nID-P=a*k)C>Jt6*Y&$x5{uciJ%l_ToyrK--Z^|onnrxRH3a`Tz*
zow>v`VuMgeGo8DsUVIe#ivO@o@7f)G=lZirMV2bAKtsR_LoT{FMKzua=ETz7_XA}2
zHpKR=ptsYce|$VHf4Zrqt&JTdIU7@a{WmvqXKz=os&nbj0`9KV%}<=y4DxkO{g3H(
z%Py_j7u;l7eClT~y5n6>8R7I@@Baw=T{8BB2a5Hsz%J(T&5!k&h20TL?=zxZ*EvxM
zZId?M7D@j>;rwja&gt)jk*S2a#21bf=W!~7V?S3}R>pU<t31;E1BX-O20aRWM1&^s
zUQny8{EiMr;wr<WPlT#>ly`ZMs+<GuLIXq_6c<XUz)60-0_Ke>FBp@0z%O3VLQXlX
zpw7yDw-h=skr9kY%}r%2zrGP?+D=P7ShlHgR0`G1WDJe#d<Z(OxFcVh+G$~`c|(it
z?pk*!5A<HI>YI43snk2t*Bgh@^Wq+8LE%lwW3U<{oW;Nk`sDul*tOdmsNaCGIEMu_
zX@2NuAtEIZh#>n9pZ364d1NyNIL>iXZ_hcvC)bRpcJCiXRbjl?_LdIEQ+@Q{DFZ=F
zYWaAtto&PMx^_>6bznb-nhv{Sfrxq|D*iN@4(9q8Ir%W`kI?U$)}pUS7SNg$#d9@X
zsjj%*gY0!_ykhtAN<oOo8T^M>){R_3icHyv)ePbn#zqXAMrXn9rFPgs-$tCH4ZUXR
z&<cIy+?*ps-(?bcZ4raRk^7-o{efYP3Z?$dmKz`6{XoBu5>etgKQd7n-^3Uj3@*q>
zUJ=$vNWS&=p~Yrk|GVw0heHoaWcMzWgtP&<o(UjvDpt*U${>JhU$E4wV1ehDHQL$9
z1Q68tmUSWdMqU4KxSnJui)AV*bC1TM1b-7p2%;Q=KG{Cu$u(8N%kvhnmmIISU;A-<
zJppoe*VMvw6g}yf!s@7|cdjMMh#|8qBrJozvq^})RY0}T-<vbm?~5pkExmdUM~<zX
zF>l{zK56^I-Yq?*u8t+}Mhf<YM8kQ6Y}W7W{N(Ny7gjIsgg2EZLpQHqn+@h(iVRQF
zhJ?XBBT_FO+lqOwjnb^8oUsB4GY2Q{W)g0coc90N-BH&+w~Rmur(Yh3r!y4}@QOPG
zp{bj(tsC&9ChWe_l~XZvG7b1>i5P{>>C5F*vv$Sgrf|Rg3@5uI8_6p-_crTMqV0DT
zv7-lDf_C(tmG2UppEo!f)ZfdKpE;$Ujd*!DijQBTAzYsc^wL)4xQ2taPlg<PS|+Ig
zP}O}jxXGOnJ`qqf!gKHF7exmp5mbwV^LDV^6Fc?0HT8iBh9MonshYU4vBIFHC3N28
zm*N^tHF=X%lJhA2gXAcRBBr?w2hF-eSCX<@n?PSLnNt*W@5?((bA>MxG{FygVOQSw
zAA-GS#y$o*Dir^!Ye7@7y9XR(*ckmMg~)?gtk{z$?zORjZ9(MbTem5RM;Ar=7T23!
zG}WX+bD{oX_VQBV7jRL=LggAt9=FH;X2l8!;^%E>6g8FhX6XKdv9VXdiFa_ti$yx|
zv4+RO`Opz$;~pNtZ{hCi&S+1ilm6E9?BiH+eeDGW<#(Su51SB>B$vR!fR*mXBAzXk
ze(14W_<?&3CnxNxua=YY^c1;jsk9{n^=c{RMb|(|+r6`mB;4{b%M3#EV(pbXwa!1p
zdB;op-e)yn9^4MFbnU6Knhg2q94MmE^*gj!Zgp`uOop!#Zv}=lF;0vHQf|7}-E^;a
zts|&)<)-9#Hv>siIoY;<+>%axVHQ<oPG=4UM)Mp>>-yMsk&dQcaFb@!b^=1WR-X!m
zqy_z0|Ix<O!QFlHS*@#6KikyDWOx}cjC+M{hZXmPeb7mhhg)Rko!AhkCuss7DgFI_
z*8l8?d`DCQJ3*<|3s;}PtmtoDBBmzc#HM4s_-8+j!6fo_H5BX5Dmib-d?;)>zm(hd
zXD<wPx%^Eo|Hbl7{iwgm{57P0s&t1G{~*J6GXwDvC$_yRCp#sE84o`I{~?vT|M%Cw
zC#|sSryxtAxb3~z1T>A)9|$o~GC3O}V7e;I`&3d<TwMfKgviTrSY}Xxdc)w3;6%eh
zX*K351&#h9N+AX!PI$~fd4;rIIMXJwWUiZFOEp#MRc%>6qwn<YrFtVaslhsn9q{hi
zJ1vFV-n@r|>AwbV3%Pks7!@A!;Z@@KuTa;A*N*3LVWtl+l4qi@&4>5A#Z6Y>4k<%+
ztuyS-_TI&zGL>VNnlE45JgCqZNbCG$1nfutR0SQx_=te6maNFf&=}|%6$^sG|8e%+
zZ%uU3)*-Zjlu(o+H583XiS#Bd6oE(w=~a;4r3nLy5Q-F~gn*zZ0fK-^?;u^8NR!@s
zRX_v;?!5PT?)L|L`7M)~$;>(X>~qfEYp<pJ6Z1R@zI5bUwA%c{3Uca}A^iyI1+0NM
z;HM$_Vs>1wL3r#h|G%rCv}W_!Zea7-JvFoRUf^=2OsFh25{@HvcniUN)EkbSq7ZVB
zfKncEYHvj-l(hWXducp3+B^`o#7f`gk`vP)WYhGNxI$s8{4DbYya3$Z(dWA_Nz)*G
zG@c@e_4UR=87oH{CkI<6W~08q0IIv~d<LJFMPRkb&Gjov%;_9)?LSd(sr*@IRPy+x
z4w6Edf`?pGOgzAg+DRvs@qu3hAph?oX)dj5dd#m#g5VEg#ic)PmAIU(^cFOSot_-6
zNYH=?_>>6Gj6wHb7rXV<LLy85mly<2yMSvCg565k%4L)?J_A6S;y$owPq)k7Canu<
zH3V8TrPuL>(AXsuDZ0$OROO-3te2pH^v?|4e|#?i5vkfOAms*`Uk#oTX5oltW#T&x
zu;2(y)9I#(E}h>ruT1r5EV;wDgy!f!I>^Y~+vPT34}a61HIZJ;l`lsd6MRo$8YHuS
z3dqjHNZ)Y_2f`H;o3s5qW~$Y@9~>U%{PTRd6B;I^RwSv>Lh)hspyi3C4|@`%>5*vu
zq-!~Wwvy~Hn5$*i(pX!+JRI?uXdyf}J|&SmHW{K(z|{GIhdifKnsy_7$|PW|_-ON8
zB|g5?jIG+fcWbB0y|#$=I5!hF<2pR{k>|^0NasvXFc4}=PtMRv#dcpsX_YwHeE-H4
zu82JXL2?ur(V{F)L8Y&PtxWym5j@rD!*6eD<rUAD$@ZE@c^*cGq;<)@r>HR6E3n!i
zM>;(&RBhQ%<egmk1tw|;L;QHtO2*3nam$IYq~vNnI^9@wg%^><hC6z$>SF#SpPBA`
zyJ-06UErdQ>)ej9v(s8po@3abCw4Bp=JM~ai~tK#!Z2^qObjbgRZNp)+;W6OV1#fx
zsN`jf9Prk#a9SZ_k)YBS%6tWPod<3!R4Eo+Q3+}gR1VW=@<;J7#%fk(hO)fmU64-P
zn{5mY4&IpXT{U-YEy*2kcx|mEMEu4ffaY<X1wPi27)en|868TJ!C*O|MG;~CP|vTT
zf-*WA7os_<zofhF%foSP=97=yY=r;Y3E6-F;k3*cW&$rq0T}WQ0VjO`Pc^v{(%2yX
zdc0*I3igCG+D^x8-hF4KL1Ata@8Cv-nd9&Lq+vcZCvmtRowQND!s9PHkaT>UL=S-3
z@3Cyf*;0Wu-j`OZMs7RNJrAU5>>zK)`aWk?6FM5I>fUYDcg5U04-W~yCt_N16)q}T
zUtxkNDN46wBgrPc5ux7B>o1m3d!02BdJ8AZM00-@kft-_SbI1*RxyGfEE8!H<!Bqy
zsjM<^GY!*;amjIp29C1+I6I!ZSOAVM_hS_QF-72H$|2iKPbYnjn~f8BAcfULnT)CQ
z41=mkBL2uW8~4g(i{VS91V89=9pHBSP<39Ws#4zx-^AEr1GhQ{E1gmE!C}}P>`moG
zUPj-fUv3^Zdff8|54=zhl!h9PT09`42V)guCR)ao@Wl=Px@JH+d$(YNmh}hhX7BG)
zkWNqWawUV?6?(vO%lOy$(R%aQ>Sv?=0Bxut|NR8kT*b-^q#N`Z%-0OK@ICpJkx}jj
zmEx|qxg6pi9ZiY)dY5M5{Oq{-@*HG&KaT9S$!WH{*PKD+4mBZ>%CL2ZUKiZs+Eq+w
zYw#H5te5?>6IIUQ-K?d40W?o<;YG^UbO%M68ydo!GTymE6G_T4L_MHpG&W+rEfFOt
zN1Dy&u^DL|P#JmPSy&S~L1#5Wb61Ho#C@!I#`D?PsKAw7Vj`rmrc4ctiVcg1TpY8e
zUhZubONM~n=5#4Z@xuT7BqM{`6KX{EZ%vUj!`>5mkNRHo#nDm*2)FReAbLV2pf;)*
z3!YFWU`r8{qWQZ--|KrKb(oRi>=x2A`#XKDZR27g{9a$+@iB_PJe>GPfZxtd@Av7(
zQWEOxhCk?K`=%=h%wQ}R)NtQg`MuB<<D&B{4~FgNWB5Q6>8<k2FXD|YI@)C+sg=Q@
zYO!JyA)yMh9X?ubPT5p(1rz-u8Cxq~{F4Eri8#_^IArX{s?X`4DaEHjmtk=SzKtYz
z%fDZ^w1oc)?LGYGAuKhRv9*>9oK%An5#u5@;pXY4=ua)?TP7B;FI`5*XSiBFp@;1w
zGjGTyIKR8~Lmr&5ey2i4OVhPIWu=cvfqV9vYkRzlB}<98)9}c9q-xkZOVV%P(Q`V0
zilRWFE??vc3suZX(B<Etz)Ml?2fT^xt<x!F$$g(CrEV-*_1C!F4&1h!TgJFmPh73s
zoN4VvIm?AIJqzqwKz(r01j*j;R7w`NgJ(RoVJHf=u7#lNcvrDG-=dOHf^OM(!P&`|
zoUy!YO)LRbr=)hz^HGBGdOx$$nerS^sodXR@r;=-P^jc%a6YSM4I-ei7M3+Jb+*~K
z^el=PdZo!ZTKqA!p!o|@f1!^%E7;=nv5LwZ=N=u#4`X?lN|SXq1BeSNnha!G4)uOZ
ztZEvpK3AHMjdv;Qgfk3F=hYQej=^HhZj#ZFsL_0Ga~`cMkSmt7Q{H$dOo0WKqytW#
z*BSz!VjmCZDu$nKcqtZ#kr(|fUQe@j<&8d(o{6!`apn(EXjVVD_%b~(-mgcCHB-Tp
z)QTHZ!dH2?6#@HWNQEa@>C%^cGhcHF%y7W$vI*y?VnRDb$${{!`wu+}Ll543>$;Va
z4{#b}*~BQ!d#$r-bycB*1fWvoc4Nr;Q?C~IQqcgb@8(f*mj}im3W{_Hy7Ic|Y3#il
zJ0B3YOHhfilh<)!UrhbV{0N3XxE&qroR$G+D|28Fd1J+L6nj(j233p;h|(??ek$F^
ztLe6Kv~m(6Q*9pIg4&nQKkFLqUk)D1&GF0~2&mPk)-V(;cO6XzvNh4CJEtZ!7q`+r
zS&uWm*8ICV<<)$CaaMmuDR{bnG4-nV<j4|SZ<UU_X`<Q-3H1Y4>l>L!4yzh+Y!s{O
zs6JOsnx7q>4ExsSB*p|-k?=%;cod}TbN&tK+4KE3?#1C3ho^@XO+KL{5v>f&$?zGr
zF#Ol0`d|LJG@Eun4m4i$F2$iE$Tyw*0k?X>1D3XV1{Y7dRt8H?D?UA|pF1C(`K21i
z#M8<85PF4_fsu%OMkJaY%&CVucyfb&3{3F%=3f2=6oC^9Sdi?%RGn*)g&5g<+nqj#
zxVGfFQW{G)WTnx&c?$D>TH}(cKY_P>%ZknHVPAD%XrM)9b=GLyL-khUfiQ2vu=s$X
zSYwHW+XM70-#Eh)hl1ys<xzIYF)>|WQfYWYoQ6-^yTG6q@8%^VvBjvn`%?}owN=Zv
z>Qsu+c6sC`QEf6p%*qP9i5gQ!UpyUNqH$oRk9AO=o~}r3xRW8C&~3-@Lp;f?Xv^f8
zD6dK7BVZ=Sx-2(Ggx-&lo&LFK9**i6yt5k_JN>CH(DRL4jZ8)7a+H(OgPY29^2QY8
zdQ;DR*I$k}P!KUC8x{@T+;ojdwRe@SPMv~?u%b_}R<1!e;^gE?U5)BM{wRUCsU8)P
zxNA|ASL%B1=TbRl&Y7HimYf;B^QrE5pJXyp1$Y2US@L8Kb$~H4@YL7W)v>Wd-ZDXx
zPv*))La(a!t$SiBCEaK%3{{)$1puogYm5W?F1I6xv)$O(COp>3ej0D`+`zcxB~80K
z;Tdb*2cJ6!=+#)arV?{8o(^VJEY^Jz{Y*5Am=$T}=wmW|@41p;NxrLx6PM<6>-5JW
z;GM~_`4>qwe(y7hiOAaA$mAEei5a(1lPuT0QuWyT*gnn~!?=j$?<JPPbjqO*%l2AC
z1JSAk9y_i@JFOVhOMQcN`p|FvIRn*-mxm`_xjBR5X~|JCaWtq0OWv|z(znY;a~>q;
z0CxgY9t$}*cGYZk_SG+wZl2BZUW2AUe;U@n&BCR2!U-M$rR3i(LO-sTx3Bh(;A4lQ
zwAYUtNb~I)nt_k`b&E$|T;-&Mrs`IZOolL2R7FN71hT)eqzhbdUxw9NIY`$`>^uQi
zKQg&CTC+MH99=%FtHIRDck9NR^q$wdX$Z^V>vx~3Bi27}GYfDg`pSO@jsv0@(cdFe
zBFG&~heMEQ5H_Ud+srI8G0NqW{lEMBGv{rAhO!sm(l-Uf{0o>1GmFyAd%Z?#eaeBe
z-phl7E+91cB7b>mV_%d;-!JiNf~9uVSe<<_Wp>te>H18`Kr%P$IEFZ)OL?`Zejr3;
zdxi7h;^|N2SSR!N=Tb)ou7F30gI?jFM%Jjg)<C84-j$}oo6`;n2jU9ypzu=Qus_y^
zC)V}t>{v;uMZfXcjmml!&hMOZ&%!QGeEZ#c+-{py7Me<UIH_?)YBQ1k^%ovaj%H=$
zi{@=-BuXBbYxT$^GQZau{^WVWD1e25)Yp}b)}FaJ)^pr3ZSX_~M&O;TkSMhSb)&7l
zN9g?I?x3+iz&5QbQxx64q_@;z=~t-(&+F4RNm*I4&jNWL1oDyYF!>p(qSy8BqobxN
zzhmh<A^9WZ<DAmepV|T&NdNlMJP2Eo@~Rx0JfKvrjOib0HjLX{eEa4Q@+p+&%z0Yu
zF)p*GwXrrJ!9tfwSsaksA@VNY3*TD3A|CjmiR2GeKX7rXrGr`&AM9gv00|yS*n9><
zUi0u*nkm@Wbh$9vjjL!DMttkC<9XY1veC=(gfC@DoKd?(2u7;n)^AZh`CXot>dv!z
zzQ7k}RURcsfBL&TGPo?s%1Qh+W`xQpm)ncwWCI#yO-&5v+XCRmnx&=CfJ-(Vs`Hih
zTC?uKx{k52Te(5yRrlT9{&(8HTWN<xqgtf-K@|$FVfaHFqzISmHT7SjpjkdFq{->X
zj+LC3;yy?kE6)K!Rx%?utE>h@GR95Gt`kz!zDn;)3vuQ8o%oIpmOV0?i8NgP@!Py~
zRY)41nZ<xe0UnBoz6A@b8v*ES5ozu{Aoy>PB4*f5>tMmi&BAM?H-kk?bX;gH`(VRA
z;|JV9WwA+k<{T7q=Dr-iJZa8-(6-^}yAMtgPa1{CJ1X8h03V^AxlNQ{ZG30k!a5ya
zd9B7$YBc93#8rI;P*ZvwBlG6Lk~l;B7=9@kqGSwE*<X$!z>6!#cwhGV7Gny6H|dFX
zxV_!x+WJdj+CW~quVl6=Wb>|LmvRwVQlrzGf3CZD^=;c+bX|TH79mcFcMktbwhsyp
z!51L4g2G%p$3nhg_LwNX!m)b&eggt-$n+mSiJ+}pRF7_7^AWV)yc)YpX^IZ?^O_th
z0o*M-OHh8pRi0bDTg7rz9gk|$zhoBqe-qD>OK1y{&)MDETWhmp?Sv!rq5k_I+gI(1
zMR>d?B&heb??_XxX>o0OWx55@t)$k)`0)d3r=J|6%$2VDzXZB)(W!}dcoW*TSCjpG
zV$83^@SqYMKx8!<8o1{scD&_rwz6>qw4MK3xPYKl?zpSm)X|sjfJa&6E}nTlCn5Lh
ze|K=_j5`@|C`<uY77>3e@-{LX#}7NAIKEf)2yr*ic!v_X>7?y6eWDk-FN2U#5}Wef
zIn`7t&37&Kq!_qot3?W_4HhysPGS9S?lux>K61`SAf9Vy-P$dvw$9~6h>B?gqF>>p
z@StB5_1}BUxj6&bZZa4QwtSr(pyb(i9h)8r?|sPy0TUu6V^4y^WSX1)HE%b%eDp9g
zR+Podthqk%^EB))$-Aw;ugj0P^~{Qf`+G8sf&7_zGY+oe{yoqB`K2?&usu4@Kshq`
zn5OWrXkqJ3lZ_jnly&|g7_jz8;O6M8#c__)UgH?vPkgU!dTrX`(MEB6>bLW~x!PlK
zw=$>F+D>-AfuVn$q8}q=idW=yf<M5dW}lwdUN6<KO>z0@j%Ki=qo?OQ8Sy<C9lZEv
zcgw6Hcd!#_)8OyIK^_UEaP+13v=-wv(9zpUs`nxkX)+wqslr+@HyC;r0nc`36)3dB
zvaOA8SUf>3-{nX89X*VeatjV$m6cB*IhLn$V#l}h!f;v7n4i3yr{eY0rmiWBu;8I2
z?ffbA*-K@v(~QXgMknfcUH}Bom*>{+{TP}@$I4Jp!;s#up_IIgWJ0fXmK+hGy^mni
z0akLkpo#Ol(;d#9(eD9`dnp=plnkU1cD^af{$&2I-tr4B4kfOYkae9HtVh`ze~asG
zo(;S#8V<ENNgG`L&R<ZKb$tF*GOOAu(A!^hC0jcZpQd7=Uf{yc6*PZE#w)$oZ0-iS
zd&%2Bh}A(pz_knmwh_TjENjp{RMMMgdBwi=OaJM<JrWnstPglk3(7v@V|^#4JiPx=
zH*8R8u@ISGx`ii@AIq;Ip!1bboj{^C(g%jEc$hP}<n&qI@6UFBd`Cc7vj!R#{gzJe
zq_u{kEYmgfgX5@grTss$g?T53Dev6$NFO|*Kje{9?egaNxNH;=S{H|Tz54FMNY&0q
zM^2EcRrY1LH&7!hJBekNwzo?54j|Uc6`o2j%U?x5u-cl(KBhA6>yL%Et2uJ0`)$|h
z$KlN0#NE}`>UXbtY+$*)b9C@Xhtn5YnDgZ7=kxQn_zXHzjptZWM(*o>tUAz`xb*wq
zk`MHL6H@gnaanw|uTAkf-sIkS?-{scezc#Qvh8(<y8&;Fm7}ur1+x(EEuM(LzC}gK
zcy|*GQ`(L`0TnXrgrdhZMjOY5^IBI&33@vGUT%rc`EU*-Dlnx9eOh#1OEv_m&qF3o
zc~zOri8v%dCoC8c&7$%cFDUUcvqUP(C7ZY~JCk`o|J5@`evnvT#8;?OtD(Qwj(O-+
zd>mOK3J^!tWnLX}`L@e-JrVUVpUI!Ho%MapTLYtj`jMY>P#6Tu3N2t2-REC~^u6;?
zhFwwNYHxuTUAtIiHE_LM9AHKBOpZ62)!VHSl(THAiz;etG^C`ADYjeL-RgH0`xtYC
zx9?lc*J=OWC0V6F&a*G(oDrOEG{SkUJc^?}7?6q)O;~v>231n_fc8y6z-p;mSBUBC
zdO?<soJx)A8ZRQ0G!vF#>yKS`qx%BVm{hRHd}hOWbKlx!=oDM6mkqU((#IFH+Q`bo
zt`GJEK`3H79l~g+REQ4Tr6yqEGEb0wXHUim!Y>H-+vTr?E!!T6&}W0_toY*l(vlni
z4t`o?)i*UE-=PWf%oA|S5k%IysLUOCG=KkD5|t#_EUsfyBYLkv(X(@r2vgyAdR@=R
zV?ZVD(!<V(kR#^@Wg?>?@G%;OZp3N}hv6ocNm>Fk*MF`2<V$Wpok<MhmRcVq8H<zV
zSrAV=R@^`FvFu=mgf&{%ZaI}J=K-Y~byYrTKJ0&z=&lu{|NK;>p@otx+Tft^zLf`2
ziXtJICUwaA>N_OF;xR=*8EIa{Y?ZG=$NPm@KLt(-z3Uf;>cnd1xkk^phY!~}?+TUf
zof&eNM`$g@%+tj|zF@p=2eWs%u={4_$OCuk$Oz^`{v1+EqRc7T<!7I%MgfoFbz@&)
z@^)&9ZLKPw<WX1V>tk8G&LtcvXjW3wr3RUVw8Ao&-(!50OBTB?DK5ZbeITn!zq)D)
zXEzm)mjCJ#(N-xh)qX0#YP{Nm+dgk|;Fm|@Z`Vp0SxXN|&@ox|tQu85?h&>Z(9ZPb
z@||6lk4wkn7L@2(v-@iVfi1D<p1<B7ybH~B<Pyr2c%wmLGzqm7oDTVr!-uL^hx9xm
z-DX|pH@>qN7W!k8HtTV+hL7tqYKc?6HH0dG!$PV<P1aa2T}*YFUOemiPKw{2si;qk
zxsY8AP0T4xz5dRH9eqcQm^b%erx-D@I>W$lAgP~0^%+~p_A1o?K#%mhSvwENjMvKN
znHZL<_S1iGh{&RDXH|vi@qLH5-DNXY*=SKp>(+tuqA3t+g)7dg?Kmj7sqIL1yyc|c
zM|7o@_<kVmyMD_%AnxLDL#+77iB;0zn{?vQz~}TSWWGoJ_dltV{i`>ly=RV`@j;EL
zzCqqWr`un%nS<`Ay}gwX;$MOHy-(Bsg%UPz`Ni?rp03-zGU&AtQIF=Gi+T0Z$76>l
z8I!`s0*fG{3W~sro2trv$||W!8~u~9jtBBt^G$|CB=io@M9E!y`7d9CZv6F+hAWYo
zI@iJiUA?L-JFD)m<8BBUB8U0!Rw5-ZasAtcQ5qC-&BAM@EC1{!9C2~auSPt32)SDu
zS#B@fA|T$!5P{+m<@v<q>EHv0=ves<H5wM^zG*6T3FTMm#MOmM=8B{roFL?Xf2uE@
ztkUxkc&%~=cq9ufwD#8+n7;WoH1rmM9;vdzVwe~QGa~lqJ7Of;A|3=Y(nQ`ISk-<@
zML(7YwU}dq8!!c<yu9kbL10(uQz^R)!C-gm%(!b?XN!hC6ZN*l`2BQ|2zz}_KQ{7H
zHTl3LF?@~AXFvX*!Va)L$fMq&B3aDRrwN)iC%i4;#{EG)$oui0YTN^vB$?vJ*GS}r
zt?jHmOeG#Nb96sy0bPE=%Kl@81c7)C7TAlHpOQr$StI;`54|Wx?Oo%8n=S9$+S1k&
zYau$I^F8q2CDr=h3}!c8jBJQmNfk$CG(LlJ03IE*3__27+zmnEZy*f^mQrI76Rtb7
zj{6WC$1^$TUAwQ}Z%V%ZeK9{Q+LIvbxtb;bH{w@y;+H&D8maocH)TFN%7J#j6>{;T
z^(4Oc`Ic_3wSNjy)L4jHg%)3F9@TJfakH2IZF{Q>wEX?Mh~RuPADXB>$5tI8dbH(6
zA&^#)IP(c&ymOZaml-h@lE@paHBJkOSf+R=HcxU}ArV){4v8wsEi1%h3tjWqhyq5{
zk8)Lis)x$zl_ZV_3L+3`v5mes50~{BDrH~E(Fj`nJM#bodl<;k7r=~wq@hTIgo+oV
z-j2p!d75TjRIGPMJpI*t+Aiw(TO2A4@Lo+hSR0h`=*5sa1#@&X@MQ<wSKDpgJCK?Q
z^7Q&FU0GwtdJxiGjN$sGDO@hdpF`g!P2-S&>Ku5(9kQ?N?Un^&ZXXroEO6J%HW&|R
z6Ldy$L{g1>)D+?y;J~R^vB@yipDqkB$P#Ph1VL?@QUtZ8#N+qNE*z5O9UM3y?l$&r
z5W*GaEJ%BkZ-j>va)m4<Xttx{<C`N4iwrpp8TKXx3%v<&mN{ts)8?uM9Z8?G-uMUB
zLXV)&+t=dBub0tF23uxfA{$-g-mQ8&9Oo#?ch{iq-H|tEw$&#s7MD+e-;#@oj9W#P
zrF3&bA<D9ZoM~UP5DAyU+i!7zuG($XqxqkK<tx6m(-xALVj{Mkm%~ux6`q-mWar04
z$NTs3sJvnVt8T6=t&%{mlLTFx>noQD5h;zWUylzL8?Xl3^YgZswgDBgvZZEQWN^Mf
ztL(w4iH{p+h3Z)gjHu{`!4F7zrBH9>I)ZkmWq=k!wpY!<>Dg}m<-!#pgY0kT^<1Cb
z=mHUzaa1Li=IEsc1sXzvfZ~6r7TEp#IKFk{TFD;0(33Tk6$_0F&GAJ-A*c>kRcsnD
zaWmoFa?f|Z*T|LaHt!m)Zy^dLQu(Vfrkz*m-tFNVSDw~NS@hjLHyJSDaF-~JOJ~Ky
z^GH1<w~m}DCq8z`8@~dSj33^)<~(2-IE>NF?7s$<Z?(BL-jkJ;HRsG{wmRZ(<Ay*!
zr8by@3teHlsz@AKqKmlfQXzkzOF&mmOW4ea1Qts|YEPz{I1~wu!QlxTsyZ#o@Vo74
ze^uS$<CAalJWKsZo%emfaH^_t-ASm3(MXnASzp1j=VD>iTLIANfcnu`_Aa3XRVa<q
z5Q~)|-zCqc`r_G{ZDB(1rAk5Jvmg_G@%&A0>H&*6%JNrZKib=*#|k9Vgs!lP#jx6A
z=^*IVR!p&PprokEWG8wwKaN-n@j>4FzpSA$GS9bf2&&ZXF*Rn!FUCMg{j~2Op<UUy
zz~`elV40SKCm|IRyPHB11+Q=%M@(3e>!@HB$tf9b6Bla>uTr6TKcU<rzkMJ$xrP{r
z|2KeE!!N$_K_0A6jcUN&ZM7@ML-NuWA%#01_ddNG3K+{b%f+XD3USZ8{gOLU!+r$9
zt^&hcXIRQtvK=5vNmjzdXR7cZh<B*d+&Y1wezap8%G=gn2Wjw;nLIt7ZVM$kK5AAH
zU_7C=dAACiif~fbVSRopNO+zLCf6P<Ds6JtaVow3d3U#MY|%?zlz5~=f@VW=bJN~}
z#>G+Vm96{9Dm0M{j0bspYvqK!EP~6&hg`iCT&)Ey*X-wimh#Kjj9s~Iaof4Y-H{o<
zGTHij`=Om7{vxYjj^kPI$1NY8QhCR2O9f56SRc0|333T<z7O3H=C@mQ6W~s0t$gAs
zUY)<w+~)=voqJAv0%c5w$=3V+PYXSKVoVu%9!yyy!YuLlSDh*JNpxyjaWCUMC6`*k
z&q7Ex!C{u#Hc!ks-6{)@bt1o60eJXrZoByL61Uum60+b9-bCXzgW3lde6-!2ydC%8
zgZ^syqt&v<KWG}&A0m}#Z<Sc4!C*4FZBFGAGy-b|-^8t~-Ywstqr55Lpo3nw<zQul
z|NEKr%|vyfPYshV8}bzY&B;*GflWR0i^-2~uI>)2g85Y#tEmdBHU<U+8Ib$U=;K34
zKRq$On*znm#y!Ee^5~ci<#I&^b80URi^@@gSFcCc5zKVk1a%uVDat+Cp?S4Q!~Q|u
zqKU8Z9(;*zi@$%xnSrQh?H)w_WMm2(FN1m)1WrP#4@2e?USbSi|GnA(q4&%XRGdK>
zpy_z>Q?uhq<j8a|?o&S{fzPq;9;=i2vBq1=TPpnPz7W!_^%;}UfoB@L74)U9{F928
zF<4Qy^Pa~We%M4~IUKyzMAiJE%kbML__)XUbtNveI-QoivPo)*y(R9LJCV59(UQ)V
zkyi=Csw4WJs7ND;7vlyCQx91J#hg~n**=%b*D%@zV~WY5@CRvrK#^ahxu*c!M3x8b
z+|Hbde0KfyxH<VZJyK-PjF!Fz!;I`OEc5f}4oQQ0n)TJvF}C_jrxz_0(Y$v`>pjcO
zwZWXJ=i_VBXRJ=;i4S&;!dOVNZs%2E9@m$iYFl2xN|F?6(~Xchtu|p`uj@%*K|Z%p
zcPRu_ddO#DmH3%5fvjg&^yif{XT?#3WzROy`#khTeepQ`(o=Pi9NW#%=`w=Q<agRm
zjJPF$6EjpvT94v!tqTZKa*QK;YcL^$$9gas=(PO0IABo>oNLP%b7*Zxf#v2ZEN!z5
zHoF1m4J&YZk<#BzQHU^IUWx}VeM5cTo+Y|R7$uW+Q^~XgIP=(A)>^Inq6&6RaxLNA
zL}qv<mL$8C3c(;7NqRj(hNvuj3#!TM&Gel%vZt<e%lUkE_V@m$=m9?%Z#%1s&{|`y
z_rgEvmwnr%90&&8Veu{J_>RTXtum2d-hA9d4cC>l6L(>?7(2fTbK16>YHnk+V2#U~
zXM?ahzB2$wxgP~?o7yl^2_Ulcx*-`QKNLa98tqfrG)E;9O$p*HalYe&Uh>By7C9H&
zS)Na2m1XH-e@3%-w9WS!KVt9d|1{w~R{5=}+t@MDl5@Ji1(AnlV8{}bc{nhf-up(c
zBQ6z02eKe<vH5n|@fD9sJN7J^E39Lcy0HsCWr4wjK30Sjl3n7-<KVu$SKra`V1W|C
zDC;wdH<~%Cu~7^pni@a$6U7sJIHZPadX|z8#9un}Oki#|T2)%<EP-oZ>D8fWY%D1+
zNzEJYAum$K;^3b%5is0MUIEAxNB&iu%II&D{m)m+H(Qpy`NMC*Rj|A%xHWkiLrIra
zz-rTQxU&>vrPY?o07)77^<!c4B8HpI7Zsz~9u->~J57MMNoro@U?BEf<@^LmzFsJM
z+qUri$5@ikWxZnfnv2mW;_ElagFfD8@9r~*U7eBDghy=Y`1Ud^PmKf@%XI7AuWV#D
zXlLMso3b-XnO3u3+<zF4RnF%tX%CCg%cx9%G63RCi>yIg)FTK-w(hjg*SH+++}c%&
zNRhcazW6>s$bNVCTi3Pz8`xs~!`(K~u`f0GjsgAVoj&+Mu&do4dPzs6{;-}C*nd0b
zpVAs*R}KWhx_PTmJ_%Ui;Z>~69cK@%nUHMhc3FygCk8_v!8wb=X@7^IJJ+CuR7(eB
zCcu!rF;b+~kB>BmwrL?BP16XlI0-Jdyu0th=<TgyB+B!FwbMo1(D&QK&nppX?U2q}
zIxS=pUZv4ulBkNv)97m+(v4D6S%J=f(tDe9Q#2|=FIOT%FOQTvqBvqzX{IT6@vKae
zRyr7!`S>*v<(`6c&MwWOQBgU?-!G_4wE#9fEG-No9ff1R{{|U*)i~dSVPmsTpN#yY
ze~po9WzQ%Tw5re1DSuu3+d^vyD}mk8Mu=+@%~9WEBb-r-EqsBqcePK@3=gWg?C!Wy
z6niDR8uL(i?a{Z=*~%SvNy>H&VQWaaJf`@4W!%78`Bs@+nD-}H**MH_sg&WH&J2}A
zs1TqSFcY8~ylI##v8P*Wkgj5ydUe8_U*|xai*tv2+s3?vF2;@zk!3gO-g%?Pe+jZA
zo@(Hl>^~)lX6<S<fTDSxcuQmvbh-)Q{^FivvJXt!<LogZMWq-GdJcr4Bh<9qo+}c|
z*9FI?I^I%?uQMo-ju?onE5&#$S=*fa0Y+vgf}6>l8H&Cs=o$_TOD6{7Ke%gPH@}EO
zPz?QjMur-$JM!KzodE__YW*SRTSE8q_v)V>2hgfaZL08&-2Bb1wFrBL8~RgD2`i{+
zdn_J1<samN=lj!Z=ISuqDHz?JO?v$;!tv<&k@v-7Yf<oc-=wBdv5O>c1*$5=51?g%
zc<(EbVed!w5g$2!lgJ!5zFKq*VoUegcd*E{=LPH5B9+@UPG0sR861;AfU`_JE*38X
zC)bclbEPOLaxibVip_*l(xs)cO_RH>M=SO383{POef=-k38AFQqf&@}qfe@G-60FX
z=%qk~5|xf*dV6b?ici*L*wvN%;Wdu<)ATX+gzvG*nec1~>HetRjAdG021zoMU9fMK
zvH$z1k)%Lb*RS>U$C5T91xzqJI*i3$6A|M7i5`tv#w~7SD1+ycjAmFSL$3qHckMB)
zs)x4UYos239(dA6WD^#ZKpn=eSxsk2^_tbM56?ElrzA-ZuMOm0W>AV)TfS^bi6zyy
zpUfH@$j-^VC%5O^NGzkL>pUW8k32zEakhsa?k>b0U%^1M9F-(8Ib&toe^wAZ%?|{i
zN=tg@t#rbg<lg?z!P$rvaueWTXDF&+mJ&l!gSn8B%1^H>{B$r<St4!;u2U5#bwuhb
zdOBe%>C5j!3(7supA>mzxk7tt-J@R;MM)5@`uedcQ-Yn{ym}Jn5%073lwlmCC}r5s
z3I?Qj*j*z@;(d4Qy$2IUY{(C(b?JtAE$^x0#ChMyhPAyMSKR5`$^j&|9-t)Y($oe%
z|30ayHS1YgSh#4TR|FRhU<-Peo5h<CyegN!AIC*vPeUZET8$F~i&FcYXd%4(ci~_J
zuXt3QoX~0#^N2(FYD6Jpv=@^N94;>!;9-8}wCW1#bA63i1j7qKf&2NZ<Tb@|dh=$(
zqA7*3HLC7YD@7|4K$Fe0*Vxr;zZ~D3-{LDS*}gAx`Voj*33r^-3M>UFA4S8fI_{9j
z!&R}4B_l2+MWf%n{FGE#q@euXlXt1!IF5~4x85>;9Pzr<(Vf=^o>=>BR6JogSrukA
z-skiPN3i#_ZZT6fb*KenZ&EoEauR#0^WCGxab(e+DSD7i!2UhObM@cp4^wU^i5350
zae~IYy3+N@3W<+=?kRQ^PBKc9X-HG}ObsJs3!y~1Zc4~(NpJL3^+`L8Hd@n4qv6R2
zl;_0d(2e3uJt8zBsyixi$__~$mokKf<9I~xO23!lxE3}gAvBWxpdE~DT^m&LlwA>b
ztLapb`}F|_T#d71FxHn+4FpxMCMl$yRD<(sWbupT1%vkv=Si1DecJ^qcwfMS@8t$y
zZp(P};aMLDtu8se&R@<RU#j7XcKKE=U`)8LvN!pudYi8xhN}+5($K%T=SOt2XPPW8
zV6ojSjZ>cPH0b-{_T+Y&#D~%)LuyQxt&x)6YnNnmK&al}$%q6UBYs0#hc3tdL$a$<
z&MhNt0aoRHB)RPY#alH;=qGcxA8aVNL}V6g9rKxR0A@Q&+KmsfQ)ZIebZo;tX@h}&
zrj3Hb7?*r!a}~-1PoouhWNZBYcB>}gmc_FI!ZbW`i>`mDO%2X3jG|b60D(9K2*M4P
z4fLo1NQ9Uyk6ww&8X^B?kVz#44+A;f&9ALLl}C+ftLz9IhD2o+3DGp^sGk*KiYyF6
z1Q!PwR(fS)eSQB&M?8y&{9Vje)yw7C00kzmZ%%kZsasvCog=txr6y6eH7pvEh$8C}
ze!+@^^-&|*nWYi%_SDJmnheM8uQ|9liyi}UC`ruRph`k5Gc|9!zw6%i#IRKMy?{2~
zaI4qXQm!Amo49OjCO>kz`!cKT`ExJ^!OdwcD?hC$Yc6B^C?w>|)6bx#y<8%LoPYGn
z5~q<&P{kJcIL9DlX@V;!KPLT_4$AonQXV-x`tqBd%7E0K)Nov~x8vj1``3l>nKk6V
zezM9)l|+oot$Y|y|Jlh?NnO;-rHTp1W9i(T{<Tn#sk1Q;Gp!yA^YNuk_aoC3D4Hnv
zN=@(pYAdl{_i7%6@@=2Db<)qfM%8L*)vi2q1@cS`a#n?o<2>GfPe`GAn@GLRDmwCz
zQYGqDuam;zL6gT+xZt@Hd%TlApYJh%+EQf^<}4W`vA30_D`=7JvJ)(R?JVZ4+OGN2
zK+j&a?~_CYAKLWx@bW_sVWrXTy|;^4x?N&AUXZY?vdRV_B^NR$QBu0`tt~O8K`Z-{
zRjsJK(j-MAkLYWxQpr7!&W4w+-lC*e5|F-3Y@ASDCJLw^Hp+vh*K9mc2vP_T<=R^!
zMb&Zwd-{C`!Vs&EBh1|21Tf!>X8P%<XGE1ilAw1Kzu?+YjHgb!Ykcuxlf8JWc*lRG
zJuI>UL)zK9(GrxY$l1`T!Np{=<Rf3h?d)PbL3GuaRyXI=W?x*AItyBgz2kW~{pzsw
ziSpldx(JV;(f4#Rvj%IZ8{sX0`6mMP>rCtPi=9VROjOffhK$%>?4&a_xi>$3*{i`l
zx0`lMd-1dJfv^$4A?0!X3C%p38JBr7R2`!7W%0(&Gy(ayO0R}n5vy^ev-2Dn=If$D
zSYVz^UyIl|gq8gSLMhWoKTxY-TIVFs;siNOUtS!We7IoEK72(g-ancKe1E#weImW{
z2d#Bmin7ILA;W_w_wrJ$Vc|aAY8vy08vsZ1ovx{cI6UbTXHr|#uem>+zY6^Su~u*V
zt-5Lyx_y1=#-#|LKxpim6VBb&&q}(K4X@QTFb@m;e)W9G6eb0Qd$`QMHPI+G56gm_
z{q5g>ZIWx*_~$PeQh+4Q0Z}&yYx{A<E$PseA>Glkr|8Q+k%}hazd+g3<)0z3x6dxW
z$tjFQWJtZS3>!PvosUiwIbQC_-}%%=C%3(?NTTa$x5D}4r(F!^#nflRL%D76hjHmA
zap5uG(FOLd2Sa=Rooc!C6M6c-=ecpnsWtz1vG+@df^{J9ze`JfwwcetL!a6t65r&_
zfk&RybpC(b?efOgKI_vT85fHp7mG|2Cx*GF%!sBHA+K8BrLv37vRU6=q286>^-Wv#
z7Yp1M3q3BtDxKWZp8ed6lfb5fH-lljei^+hfxvO#T;N=o?gO!AOYku+nHF}Br~L-!
zE_UX^ePS+GY%W%k&VMG&_zz<ge=#Zk`jrIyGcF4m&8P|n8%3ZCFj&3(n{@fN3;5U7
ze6iJhF=pX)-mP)(ceGT)g7#E{NFeyWk7~|z&w&x$|Lzht2MkMv4bT1SpM$h;|7+#u
zS){zTPXhT1Uh9T#SW1+G%nx2{59S<XugRZp=mNu8vqph`!COAVmIe2_4ZM*H+?}rd
z?;WaVo=i=YJLjHnI|KigxXUtv=FVo1M{UkW!TsKf2yl|GodKD5;67tTTr7hqp5-ex
zf4_Fv|7x5&YaA9=IIbNekDw3Ve-83_ZE+vpg#109yF7PR*fP@%+cX50q{5d_Qoo0i
z!kHoAJLBOyC0=1$x2=E0>4yJ)%_401S7hTWXH%bdfA>G<dk2|ir^K?5@%q`LRjK9^
z7v0cF#q(jsbC#p|j4NCgDchH}z|(bAZfFGe-$?G7t<w6{gOlp;t?Ifn5HM^9`CBq~
zDinCxe&Zeq#0xl^60xfDZ=;o(279r2|4Os>pY4gdHtzEwZjzQ-zqRl`&cL6}>t19L
zUVo##f|jn&!5|{xzv5>a<1YSt!us7hw_3~a!Q-b%^``j^)z!zHQcb`wHIWboNW)&(
z-Wdp+rH6=wPV{mGFQ;@VY;s@z;#P&(97x#so$XZ5dAHF*_Wyg2W%b?0<JqL}!}s~j
zfxu|7b)}`S+H!NgcHm#`L^t)-+q?C9P1TJ%9+3V2Y8;yDZeKbt-HLl|a}acKkP8ed
z96#+aecDl=c<zD#PU-uef~Iy1uDSlr9yc9VmW2t4HE&?NWKY&<y@J{X|Md-WaBwti
zS?z^@GWO@sSQ_T(FZb!WgXT#fg~Q3YXSKQi^?mN;N$%zD-o>sey@F9v$jp}Yv3B#J
zHgIGL90|?+7CN2{-J`#lr-!z<OUn7J_APN8MbbZ20?@k+2Ll-m%K~87(R7P!@ad{p
z?n$3BwB>B>@&Yu8xw91R69%V?AIIZ#HWxKcj!la>p}WCb7IU&${W<>`+5NMPUM@K`
zfDJP|9WZY7-^&>0{7wa$(Grjr#mfUl(w2jqV|hqW>lJJ76%DBwe<sL>RIt7w%d4~*
z02%g1WO_lXlFHeBy4QF-W|MPqSU}p6Dds#M$NueVQs~Iqy<?EG1#G%Yuj_j6i2s>T
zSQlqt!)`!XXqVjVuXb76n9C(^p&oh2zkRT6{_AX(=D!_t$6<4U3twN%{97fHyMF%Y
z-)R{rTOT-g{y8aZ`#-BMGAsZ#J8zwzot;6vFaDj>eJ<rtJjfh`v@iv7TLI@sm|>&a
z-&`w0_FE~z(^)Pk&rCOf>GzwHwbrL#Q}1#4z0GX=omK`mq1@b2X<*};!XA~KRQBHf
zU$CVv66JRP{_(o}bA?4X`(D`TR@3g*pW|a)^CnVo_@@PdcO?PlL1PL3ueVMHfu>X_
zWH6{x3^HF@zZ=qPG3&ckQaz?%19~(`%s~}sQW?NW23SU3t=M|d4E}F6VM7Br)KL7}
zO8>uuG2Z$RAr?BV5L>u+rVEns<(`fda2*%!{TtXLi?Hx&JTeVmG4%us$GbQ#ZUDb;
za9;rXF}X({ovj*zet|Sl|CN*{B_=i%3QvHpTY~REYVPTQ&A$$thSSlqrk$#(dx!7n
z71m@1Wk8E#85I~8d=htY*R%0x41Dq6z2L!6^s-#jlH7j-;6H~j7&?34S-(>i!xgeF
z1sq5L>oMVlyC<my@+;NAS+zDY5SR^KbF)6c-w3+MXxx4j2!2&Ko<5iyZ~9|47&4Z7
zv60)!${nbSZUFziY5#>t!{zxN*ag<VUu86aj(uWo4*1UoHcgvPQ8B=_12<_z^W|Cd
z<y@Dvy!=1=&_^+DY~u-iz~wP;X)$*?HW>b=931SS%{x;~C8w`}KxvYDow8MJOszK*
zPkO-NI`f|s{`VWCf?ESF*5*zI0>c-Xa{nfce*j(kqnB^`05b=v_m}qjpeMc&wSPL^
z45nX)n-Aog{XnDK<`0i?yDpb``K8lS{WAA!NMlaF)8$MFILlqiZaZG|GW@^UZ|N3v
zZyGpnFVH|ok%3Ir5s;(n2q=`1mzU>F`^srgQk~oRd=S>>*yGvmO8mlA8wj>WLE`Om
zufdq3cF)z-Cr^bT`#Tq=ITVQ^dd-~G)j9Rybsn4?xeIZMy5?KEUHyul<epiJ-YXr#
za}t&^uNJQYi)o7g9ObM(XW4uXoAOJl`va`ZhHU5T<9?mhWM+iAOesu-F>E~vZ2rBy
z8}92QAKY%5-KX2cHJYbk1Lr>Kxr#Z|IRCTmY`ePaR27Ct|NW`tC0|Q|Eee~~{#URN
z%>JN#LkrpX1-0wxN_28UYognYLN`B@CC$C+^`$h$p2t=^Q(Tq&l+(5lqzh33HC)IJ
zcZ&U9c*|vXU!-C3dlwthmL@~7Qx0X!#l0dQZ9Lo41`kkQ|9Cmla~&jFPLP^%{V9Bz
zbZpJ@^;gz%lFaUMS6bZSPyvuilnxH83bWp$)vY6$|103TuTg#U`TD=O<2AG-SLo3a
z%i6cnBHhgod}ZrKPY>^T|LDA3e}Xn+gVtmnmrnrMr``gkDnn%^#VjC_l2mED3UpMv
zo!85}CW98fwhcs~?1sIyFib3AL8rBbQa0mrL9$g;Dx#DqDPytEtST9J$w~us9PuGC
z#Ji#?D5xd_kp~q3T2$%peF0qFXrQkR+2J1SyulxZQGeDg!hEJu?v=)5_19!-4G=e#
zQ2AAEXJ(37NlO*hUdc$&_o|tMm8VtPq0k#*{-%{auk3DZA-e?i6O?#76IruZ4ZN#e
z@J=&#USM|x^YTY*C8UaVGm;CNW+4!zK5UdGnXRiZ^(GA9f=@A{<5CT@QV_(=d9Zu~
z%>IPQ)tJ$w#I}$y^y?Utv>0h4=c6HHbe=lQ4m2p9S0s?PxFQl?s3R$!S^9}4(s^c1
zahiq+@&ecA-Z+u4^iT^`UWEzEI&I1v#Fs8~KQ{QFBcr?3KdFO5<{Lc{>0mX4Mbh4B
zhZ~Tg0i}<}o)D*W<9@6D(Wk6kA-wAR{JwSjE%`CrHT(OdET?9Hf1-z_6py0AhxGXa
zNFn?*I+)j3__WRMFPFDF|HNK7U_$pSabGqomRkz19mw!-aFv)8V+I5|Qw^rXX?gEg
z9S!XtgZt3jduJyi|9VNmj`_1uJ^k<FWkB!ZqrPWMEJChONPDV*BLN;X7KqWB_ug2%
z2x(+U@f<5ig^CqYDoK7v^u4=}nW|cn<}5LZp}g%!&-<i&*JUtBeQHp6m-}PPhPTA_
zUYYMNpF2y*XHB_B<)v;(bjR9A-tU=W1%EPbet^mYwSxTo?s$R={tl6j0*3C6>QkD3
z*rg^R1}R&rt}mW=(ylrWmyUu%?-rjKeW{g^hT=1?N7To=Ao2NLc(bX#x@Z!}y8*+v
z$%l>#iS~4`J`grHRfPp3kkYcUxag}j9W7x~PcXUm+S$=Q?1XH!6l~$99+QwmPQ=aj
zWET|<sp~x*Xsb$(Nf7LY6H&#^2i&rCa<T<W7rXkU6V;6S;vz}4hDL`No@;b6yNpR$
zf$G`Rj*vV^GJ)av_~0b0w&F)yYkW;nBr;M$Cv13I=-QH#-^OBoAh-t9om14mJ~^8I
zBQ--OA1Xk?o8g-e^#+2hyj28Au<K3)g93BcbnRbDg?7(EhL5=U@`_gO7MX<gC#<iV
zpj)WT!<evzTa&VOQetE`W$H;0q6l8mvCpI<A9V}|I&>JaGf8<zr|bj~ft3EMExgYP
zW4UxMrK2HA`Q1)p4-C^09C=wq=14Uo5_y4~BbR%7tL0T*h5$9zv)E4I<@?mH5lD%m
z*vN@K^b0uE5J?=f&y7H^#w>ZVo0qX%o55HtP)UmXSV>278LXlFkP!TnX8zumMJ9Eh
zoVq&NnA}g}H6k+h<bFI-$<fJMM4;F;Q_z$a?J->8!KC8fWydOPIo%wz@wjqebvaR~
z9o6}e(!Q_57#KCKgc!zw+Q_P~(#}3C4-u?8<ONBH^`~|EBdeJL=p*Z^j8Vp_i||G1
z1Xm}7EhE)x*n4%%D<v}H>vdw}um@2@$wW3I%nd>3F9@yL_U;jWv^|{;Gfp9hb7W>{
z8lyIf{fTZ>@P#v(l1m}gjAJMINc~<LX!W$_cc%h3yrm4&?CDxaQs6B{(MK>c-Y8~1
zv8*gSKC?FKEEfNua*E__X?kA-grs$7#EO=)n<0Wk2Mp5-yk^eCW_L4`KNdEd;UIy;
zfvob=xKG%JX)=CHqn2cMNts0X)ah1jJ!>_E*(|Fa66EgQSPTz#m9qks-zz-E0~Ms>
zumJPW>5565X8J4bp}HTWdCAG?RxDJ8%aSijwJ}WC7KXk?>@zW`WxpIPwtV*wg;_Fn
z&$D8?diUl2FYo;`>h0vz6PLiN+CPhp4)Zj(>2BHIf_-+wL*Htucave-Pwogi`Ha1&
z5^!Yfa@R6nD<>m>+v!L+$#*bx13YF5jAiF4xQ`|v<FPm>NGkW^<_}!tW=Q2N8NQ%#
z$?_&=w3Gv(%4MpBZe26t1JNLE+2DwRF)j{Q<0T;jXA#pMyc<imniv(etJBR4l?h5`
z{LUHCXKbvdb2XNvih_<Gt)k>DMtP&XL}%<X=X4)2??zr7R_p-GOu=rNB3_kaqa7?V
zvYy3e-ipkORCV;6lJqqKy|Wflz*)*!g_@)<IXhXWAN@PE%Lh{3(g#CFg8Z+2Iy##V
z&>T1iB|ovNV!j&c5avqcmS{roEkY-aNleDgB5;hkWjrW?$^cLQec!&6hx5FE%;%>X
z#9dR~XCD5zHCaj;fe_9+izz;DI|}LE{16dgp~lT8_PV%+NMwiBYB;I*OCZUoiK92P
zsnQvtzDlBjN_RCxO(hmuJmw%N%o2f%qA%ZLr6JD>G*K0FcG*!>1bqe_wM|y=F60bT
z*qT*V9<uJdYZ2F=5=bBHys?|T6|zn)FOHp;7;2osK?H$G#!cU8Ig}o`xvBgx=%dY(
zjLe+KHd5Q-$9=-z8Tr^bnZ<?}(T8v*mBjC^23+lpnCCH6M{dXE+!Teg8R+-i8dX}_
zRf{W3&Ca~9{X~oc>g_M3-;viIiQrRT&Aa7_c%|)Qs054r{)UsCE;PQ8;DXguBT3G5
z&;b?0Ud3Ypo#78)f%_|MWS)~}Vg!gZK>o3E)QX)FBnUFTufx@67QCJRn8N2!CCcs*
z7axbOeP0|;jZN5D-|RCjs4MgI3eXcX&LqTAomFLONe~e5yTR16)yxm!WTL(_R=C_)
zI|X;0m&qSe@AvB6<qgp{d?P>*l*GuXu0$4~NZ$dpIew+L^jpgxqf`$pm)G|-`9++p
zY+Vf)8LslR2v&B`#2OCAX(83@6EPo@5=(76jL_IjqE8=5pH5>cbe^L-b(&@~>g0D;
zqGk)p6dh`vU(j=1g;kouk&MibJWvJwsqUF-J6vVB$~GE{Cfhs;phk~YIn;{U4p$va
z&;Whk#CxR4KsY;JjT}M}Rar#am(MI<04Y|bi^yU`)KSs}K)%1q#3n<%E;~RGVGh<<
z7$h^|W<1Ss1z4ZX_wG3h8)wZ;O??Qzy$aDu1%VrME;pO1B(Wrw4dQhs^A<zUaeU_$
zt@fz5?3)Qg-z`3eb-&BJ;n$m=>T^d15c3ODz#>JRyPYUpZKv`@_)xDk=r~$zryY@!
zFJGh>=wRGw`0uXCzjPqEOIE=2dh2zBpa7-mRWfj9^2CQrZe<bP_?EiqdEd-?t%))$
zGZEt_Fh6HYi-mK{u%Sf3ij)L_HJ6WK>?Ga{MNec^TpeJ%>@lwj@-msPKR|T=@HX*S
zEMDeq{dJ!F79<(JWLFFJ`-A%h_YcHHW14dN*AC!?smP@3lpkf)h?#F;Z~{)A;<g&$
zV?K{k9IE}H7Fe@Wfz=s)Ow?Jo&3p8%cDaW$G7}}rgR5t!^fcf{BX4<PhOz6?7!%*P
z(DSF!>upYa0&c?03{+GyAQ2iVxrrExeAE>uVJKxR1u=YHQV{(;v!%-VN-?BRwdj4;
zduS{h597PyIA-L;NK0faEV4BV2dA0oE*7}ofAywxP|7Pcj<*}K-WsB4je~C97t&i`
z`kv%-Dk5*y7D8tWQw#VO%&LLvH^}(fqREPlWwC0`?oThzVa-foA_eLS-PJV?aW@I;
z6t`FznX#{b$+q!d*Z1dB3_qs-LqAE_Ih!!9{ZRUi?F!Mn#ZM680P4KGcViF>>=d-U
z{_e^HVNY^VBT}2G3N^GItR;cQ*1*LfN{Q~C%dh)!MUN%*Y}X<<l3%Jgu6A=62Lx0s
zUM5uqy7=-<REZrAD?owF`(FkQv}V3Yo?p(5;Hx&|iJDH(h267JQp&-`yTkK8bYA$F
z36PM<|FPb(xQ$&6!U=UpJnIuLM|Cy`4>~d`p*t6OAdga>{|`;)9?tat|NmKNL)o07
z12d9_9F|kg8kQ)^oO3p5&MP@(Xf$W59EOl{b3Pk6B$*Gxn~Io25|WVfDg5^P<GbtH
zpS!Nt^}3$V=N^y8{r<e)EHXfOKnNQ`=Isu3OJ1Xp1O`CBv25?SWa=Qow5u{Owg@2Z
zt&x{HCtLO8ITQ;iqWFsMDxpRJ8Q_Lx2Ry_f;N8v+Wb03;C3C@qLLy>>0*Iu<ks&+K
ztXUoVbP~dmyOs#aKq$h)OmBAf{jkQ5+lvf`;1FH1UA339<bvx+3QFf(OzalMGP>Us
z0XnsA6|yc=1t+R61|VG?#U1R=YyQ;So*(^q&IPJV^d>(b`pOej1$|8I_)&hl<&u(o
z`tDBGfK8J+5q(ZlQc@`R@3ut65lv+OD>e*MsJ4{ccow}NC?X~{m)p^HvfgoQU;ub0
zDUP4*MgSoTGoMHtEM+9KaUqkH)1MSBN;k*2I^YJ;z151MuJ-slchh_5<4Kk9Z=uT-
zdIm*0VZ-f-1aY~n&GX4P5<>C5p*X6DCgEZL;$w5+*XJxIc5$KfIb~sm>K-T<Uwr*&
zRwGc?N%}I8YgZ7(V-2>6KuJ<lJ1rPrhc39n$e~JbiFYdEcfNWQC`XX$(;ENyCnog%
zRXz<twN!Cny2?WxQKY~ATBy_v2o*6&#T6hpe-WssUkPQyFDpru0?J5x0_zN-8`N!s
zM1QjK+=2KOhPXP6%ty$$RP~x*$x9qhLZ2<^o2Ep^|8;El)YEEPNQm>>j6*n!!}VV;
z-j!401E>SoL`1AHx1Pjr91igbproN}88%<1B&rmV@fnA~z&wMbhDzm<D(^m)ms+8Y
zKoYHZ)-vUgy!p^JB_cXO2+Gm^v^1eVEk&9|DB|gbK6cy|Zj@ExOGy#l54H=~%56y?
z_~h@B7J2Q_2B^za={WI}*@uspA1{}_2ANZwIf&7^(d^BhY2^XC_Y8`u+0Hta%u?e{
zxFM9vXKwC^;uB3y*3i}Ca9jC+K*|wZE7a);CeH5W?8bOEDSfvOSF!*@{A)A8xj>X8
z>}SuvY@rM`MQ#K)H^tg!|AovTPggln#6i&9XEikv@TCQ=_!h>+$L~E{tOP-kvK|4X
z3=Ndvy4Xk<Uq1XEyW71~uTD5#{sNI?1{KBN<WR6~DPXGE(!#7L$$B135Ya;$EiaB^
z1f7{3%@{179vQyTwkHIDkzoOB+UNe8HaHJ)U$}S<xm&!rV)mh+zrSLLB61dkyT&s2
z2|Y3$tZR}2Ecpn@j5MLo=<B7VE@+Qs;<%syBCX=n<-RU-3Pn`#f8|EFt}aQ`JyK3e
z^z4(4QQsMzG!Zxz#7*GW->YpXjR+U}&Iy1-h$tY$+g=K%#1a>6u|n;nq7m;Iu$N_;
z%8k2PXU7W`gSUncn~7|@oQ*S8S>v|M4GH**t-tQvP+)dsOXIXfhT>fCe>Vl$UBvG@
z8**9z^ow=9xfP<RVAMw*eQf@QHNXVG;IZg}@xXO6x|~E)2v@i4B_e{6XEB$%F6*M&
z(|Z#@60lNzc6TANRPAtUHF}rvUvX$$AEU1n!D$AM>s6=SgFMdwKqCMK0?68XcBl0v
zOVm|IM=ahV;5&|_Nv@((c_dCZ=ePf@2WVVC3amtjR0WbLdXjgiI;b1qu0I&OT_<>+
z`I*WNbaNr(JG4Q_))vp?dSUF2@r{+QYb+h`?Hx09x_0s|l1jTvSP_TRRFpKl;vKbd
z+K0X$B3i;S@Q+SEZ41ykIb=PU?tIS|-tE3IOs>O3)oGn;kDt4AyGRM-_u1o*znne1
zfBKn0vqJnv@J0GC=CR356xJ;={kitk*2-woAd2vd3?)NtW}!Ai56U3GoZR-soQ6me
z0@;MS`7Kj{zYc;Fz<z6dPBb&fj8n4eK5jz=8Ahu^Px6YXJMHNpc3`KZsQ^63Lw~fr
z09NbOCj8$VP5O^-b}aNAb%$da6WMDt_Pt(x(Y?`0Y*JE}CxaRN#6%Upn~>q^1U$t>
zFn6+LSKt>A)Mdk3%HesR@}7<;r+Nzm6rcdNAXKv{%59;z_DMwCw-#*+SRq%<ul|+y
zQ-jmY>^e9^&(;HqvL)m81DZf-fuZJKv#vDXT<*?Cy-r2<@&J0+jc~aREbTcf$D^ux
zK1XD%IpbHL28ApF{p@P#ZedGv%_k^;IrQB)UzcQ65mEWne+Nf)V##xwGt<*VbjHVv
zWx{V`Nz!nSSyu)<8}>3sD0o1<a%pujI4mTz?adpu_VA<aCGzo|zB52redV`i4-5cc
z;y8C+<(&CDeGH%F0atGK7mYV1*{7{CD?CS`dIlD$5DAw73KWA$b=8cc5lEUNI1T}B
z2`3k2y;Z8LN{@Cnmy%4Q!^X~1L;h%OUz9QHeEui_fsm4<DjEQT-DtZS(5;MU^y^GB
zsnVB9gUgCz@ZR@tjC}R)&d4J{9OR7BcR$EbP5=Yci@F4$prWUqId<gR>dkg0Aff=!
zn$w7<x1Z+6Tf<)Mw$TWHS_D6m^6Q~@Ijx6>w-lJ#a}Nqwx-f}P{!5dB2?{{{&jGu8
z3iC}eQZW{vNMrTNZFNA!TEO{O-LJ)-D*w*JVjwm|8|zL~7cqc@5)g1=bH@P1Co;?~
zqv3I=v26UEcrTlB@6Sf&9H`PipE=%LU3%Nr5WW?TEB}##C)i`aBwaPHfoe%<-Qqw}
z_R+(9QZJ9`?8Bhtt>K4*2soc-YM+-(r6R$&@*gd>JrDI$Eon}Z%P-Ro>kR%J{5QRc
zDWB`8-FA^^PR5<Lu$1~D<jfDK212;~5^XU3-u>U38~9PM<H&eUIN!0a^bF%VuQ!6q
zf|Wr%7bXeBth@k4A$`<@E&G2&G?5&wJiZ(B!<{W6e`HH7dGrQ+fJTjD!w{imzt@9M
zhL%z;;`csphRwTm)+?t(izJUGjf%X3k4Pv|5o|;on=T24=&I3jbJbc&15W##1Oy+T
zCn)Yce;!i1^AkvMh&U`ALSxihxR5Mox2nD#fn5S#S89oER*$R#1K|278z<~8UAX_{
zuTA+5z(7m6*lAlVhk!@S-CYy?FNVIp9{d)O##v*sR+kf=?e=|};R|pUx+i`UBju*I
zG_2UyhRv|LQRsaDilA6vGRCqqt}>P)%@WL@T>%;_c+oVVQ<J_G{!m9lshRe|h`3J(
zu$%S5VN$6%i{2HKL%{JR2Tb)S%2`ONs9{p;V@%0&r!{(THsx=B`rXCxq??1Adn^fR
z|D$|v#(E25KsEjr1KoUNB?*HG(k>Zk@IrR_*2WsT_?4=?OtO~e=H@n+SsH|O;6r}a
zTn6QWWguo|Uzo}m?#+d(BC-E7Calo+4>(KBg~>RZm<k22msf2f(3pog_-J{$arZ6M
zWR^-rTrFerPOuLih>#GHO!4!r^{i!lThI@Ji}OjM43+rtR%9)8cS#F9%cv3$mVFPV
z#O<~ke6${9VmDg}M?+Ybky-ZeE@P=jaMXJOc(z?Hx3aDFT9f+C@QZ{?owD6jIn7!w
z@`S$<l<g1rSplEfEkIuaF^bU*$~aL()Hvsn*saGvPf5*fll4yw1r``o((*bWuhi|w
zs0N3C8nOm*PY_!upOyN76l=iXfQUHQ02sBgBL!l88};-N@x<ZRb^cfNAY8RCAI~OG
z_Ct+TN#AT+rv^9Q69h*#{u=B4H<&v;=%_uV-Eq-qjx^b;!@48~@R@m^2g}`y9ZJQB
zI%wD$NOm&c$V;1`x;@bBk)#M}UP_M17yB|X(}rAmc~QkhAQpm%>g|<&>WEha6if4o
zm(?fqW?Qa`*z__`rCmnL3?3W7Pbi}`3~e-s;p_u8Ky(r&^-#iaFrZA}U2OZs5{Omt
z3ed7Uai0&0bk!r$q9wE{hC+N+XU~FoB1|*@A7y0|63mpFrr$(n+5XQ@wVY=S@X3VA
zV<xhi!;qDQXF?lCM@?eg3l4<*6Ze}J8hC89C6F?=p-4{H0`{*2vaIkO^}70pbA?Qh
z9WGYkB_RuQS@&J7v;>=(AlC9rj}aAp!5LeLV?-DeJ^Q^3v@bPMxpra3yiJ&zH@Cg=
zjeP0DQ(Ex6_fOT&Uj9)2w|??4YpCH@K^%q6+@e79&*HU%+z@wmIPjk3K_Is{_q~iF
zF!wuFoap|3|8nj2&dvr)Z(-prRiW1++;PG03GMnW;;twT6qfXN*lz7;XUCSvn`PBJ
z@a#$Da!A7A%~<it<WptUFq?&p@_f=gN@~hqw>Iulg)VQKfFP83y;@4-Ey_1918aA%
zs}zPz0;Gi+I*Xs`vL3atXkOxc`ZnNm>Zk53jwtv<6a<fQw{(vQ?Y^H_Di>^}SVx5r
z6us^VJ#!Rm$Gh9@Vedd0M?xb#2|N+^zMjDhs_%=OIi+vL=L{NKf~1;!v1^_3nOhg`
z4;viRM_cDr#P{|G*ZB>V1BE6+7LMjxhZuglc?elEEL_M^e^9#0TJ)~y?A0y(m9~kg
zsj+O03ModYVjWF|8kYS)<Z4}KtXU&LkAz8CwM;#P_;!<iHNx@u2oO4A2+GUPjd%k)
zZ*}e$5J3XlOLteilRvq8r6{K=_r|aF{n?zw;8t=qCMf3mjpUsdI`5zDKc2i{B`nU@
ze@uOgn5=fHDzod1Ix2_^o7OzJdHyd=4Xgc06bpBQ0;vf2b-N4JT7v~*ZJq?3q&i52
z(%Z=aiFBJF3pdw5RJw)BFsHS-UP`u^Mt}_m+1enwx<F%FV4nL?QH6r;pUFK>jnt}n
z{Ad6itE(3$m+s-EXCI8u?}5fmq9Pv=;_c?~b&qHEtlgm8o1%kd2(Yysau?BW0*?_}
za^~$2<%N#+$O5a2_V`><2~zpVIJl*6OT(Y(WOiGecI6TVoZ&FjK0{FrScHl|U1cL&
z$mw<chU4Vv4u{PamWA52?x!7+Q_?emJtCTitX*98$kL|J&vvhV-<n_eR#3E@-O{u*
zLvF+=>pKRYP|>+lTyeFI*g78I2omQpHRk}pKTF7VO?usVPnaZNnEpA@DWJ0*V<=ug
zHy;L<;G9FUE2xU<-q<=69J?sFmELi@KF?aYfrtGaN9!w_@jJz39F}emRM_|~_;azr
zfJhas`t8ka_h`Gz+P5Crt*MAv$<B?AWwL*@hQG|yR1$~WA|m<NEM8}TtkFr+U0M=s
zeO_%H1c{e(VFWGTDl_t|2<9uIj4MQE_p(d{-q%`PY_}!>Lbb&JHED+Wk0xG80Au|N
zl)yt@uv%c_Z>PAR@cU^JT6%kLo=hjV$D%LP)QClZo+?SOvL|)R`3m8vF5o2RMu(_p
z8*do=_efqR^1T1KRYZKRjJ%y_jc=Shk-i@gx@1#L@Um%c?CFAP(?2e_H_Gah&eu`n
zR8l!(7qlm<8j?2uR7mhES&9qRDZu}n@bY24<P0k1{_+wB{64nAuC~5_btMhLz7MKq
zj5j_csEIYiU+%}3vqnQuGF2eTt|0yS>KjquFHkJ$OMNhM|9>w$jwBPV)*^j5low4(
zJ8xJMl5RQ<g<lspvAbXY79@TY%KJh4gA9M*Pmo210f#s|C6;n>DXniK=r=p#Q#NSQ
zZmmuA>vBlBX?#4Wf9FQbz~Sy8PZOy;fE%lR7>NQa3>c$Qx8ApvZawC8Y<%#h@stz5
znUxtvvH*$VCQ?DSB0#dNT*H<Ub+=0us@~5CI4I6LlZw=HHylv!dLC<bL+G`~{ffTF
zi{3Z?4(@x=i{q7agEMF^E`UAU#cM%G{KtYO`jG=hzN(5N)*||wt?RHc-@R0+i4ajt
zjcGs7dNrq-QK!q)ivtsTW#5nn!m@xOmH?FQH-tCMs(IW`Rk|1CZ0U?eAv*iEjtFho
zVhS?=+Hc%ns=?EQ8>DAS!XHSyBHDc(o71#z?p%=6g(r{(1h8y;mh(sJ&6P$(<ob%2
zWBdECgU6wotB9bSl@-?N+!<-r?jiNfo~lMKr7+iwrc$co0^tsaSR1hCY#Q`(RAHjd
zv*07b@!?5BaYN7I?adAPdu%yDP>9lb*#u!+YHdKPkh+)3u*N-bEmxkh4El4#8BKQj
z$USjh>5A^p!K0#9q^obmlYws=O6|`=yeldi(&UHAxxP+kU7zZeaJR@{y?S9hROgmV
zxwNWU9Ww47N?#vnxB^94N5_2_7I)Bo1}|d$Z{uIWfA&7Ql(b14uptuGK3qp2xuDTD
zD5!!fvyUhMHXz;QkLLiZLMYbGOQ{+Tw;8Flo}lv<UtX)@WfF1!HOxDB|DNtDheJB?
z7n%-#*ptUm3II5St4Ws^Z6ez@VY}RQz8(5meW4GSY=$Y-1p?H3uJE_Dvp&x@NXdtX
zu7g@H#KpBvJpc?5f)x0MbEZikR$Dstwl7UF1(ePaq$|W{A>@?*&=7bU5OZ4mefyUD
zM=`6g6KiKX)S$j(+USLU42u)~(?^@D-3$iiyQq2WhF9p`a7vWUg!;%-{E}yemEqU3
z)XP9vUP_~(K9-Y&7NE426>uqiC>RYW=DU_(auk|#1$m9&Tz(^&1-H=wL7xiGAOjOZ
z<ey=xH-$MOYGcn@c##yw`g$u=Zu;=SC*OmfEPlvC+kkV=$si?)4wFE&f%+&EGIY8#
z@W1)slA0~uLg}ukIELvPcOWY!MUkH`;J&xfm5G|RTAtkuY);K52q(<NJn5U)8gvPB
zD`Dlmo)CfE`(X~n&iBFsHvgM*YZuVXyZnI-7mbURqt{q00Br>s2uCX-KYLfmbde*t
z=6})sFB;wK@mA6-Fvcv@?3Mw0jk&Yhdu+#SKPC__Usykm`=xfk^jCSQTxl&jTa)@s
z)=3sPWc`tD3o4@KYPbDcy6bqnkuE>B4I!te&k;03*AS+?|6`=ldfQ!(As&PmV*#P+
zLxI_!G^dK~qU9Y^#xYxJXP3GIrU(DRsNs+F%W1M&?*JeA#VIK^!gVmMu7UE1u%u)~
z?Nu?R=FDdJeLd%?U@QP-T-sah-63HK{4Y5(Rm<V#MWg#rWSdm7@_;^wLWYLimpVK2
zf^dj+*1QLzG(U|_aIs|N?+pd7<N#fS%!NLf7ASDIY52-|rP<U`QT4Q&Z>b^J=+Fgx
zZFs&$lzy+IWM1?9fIiU#K2ZnMb#rk8>bnH*^|A0p_adori1G<7F=DCgg?#W+yZGoV
zbYr-I<Yir(gL*+Q?%GqMh(}z2jH@JqDKSsNg5NVg@dEzT&r3^8zoF(z+Coq3>vxc%
zrQwN{Kv%M-%0j)LJ{AM)?tMUdMX}lMc+7qzGiishnXOrnJx2oEML2brhZw6K%9>7|
z_48%L+rGy>(}aiFYy6LOue439<DAh+nATUJu>4;fQ1$c<Lc-Ydq2jig6a<_wngNd@
zRlC3Li(v24XV?E;u__HTTulZ;V)GziF7dp4_$ht>KOAgE@}UPeN};H=L-c1$)##fk
z2>ly}x6li-J~jrZX!KxjwzG$CQ-MrumWL!<Z{dzA4C(VZ1*QAcl`;5J>;4^b!k3nv
z(be6<qn63G?P60!5M*j<ln^D&##b0H<LKa~@RG`>t1l=Bi}PEq5Mu?`am4x&-4m2^
z^}ItXD=8n~EVvcPq?8&mISrFJlcRl4XnE9uEk#L232kBLAqivi>EttCb;c*w^kmEq
z+wqH=dm^CFvm%4+jEVX;tR|{$*D{5J96_5*AM#`_!}*3AJ7x&>8ec?3L<P^%kah`X
z6zW01<ejg<4_uQxUx~aDPT|A%^vOC|c(cf2tYGrxrkQqD9M$$gpv0#L;^Qi7&m4zI
zrOn{cqq9}igPr4}pkU7jRy~j7N$$9WGNqwujN@Rq=xO4bvkV*VTT73${v!rs^L&kF
zZ1wq5ulF}N6e0)@-L)6q57l(tYA%C$9o)<$$@PP{hq@S<jf2y6zVeGUsiBuAPWQ*S
zp=_RPCZ#IwaEkmfWk-zMHz50k>W4~6@Cg<3b**k@Ci|Z>Bk!-RXolj>Ko{+)rm+!_
zc!YT&g@vy0$33gq_dg*S4^4W*QT3@IdT=cFInTCjcfxAx79fs<4k}Qs0e`%5iUJWK
zC!2ij>8hqZWPdvIyF*3!sbm)rD6mj`emGk}#v>y+-xhbR_fZ$*3Ll@vIjANZFAKIS
z)ss&O7+MmwK}}lBtgYeYP-;n|m<yZCeK?h*R_U)*`6ei3^dR_@<Wp5U8q(Cc;mti1
zwC5R#AmP5^Cus5W&eQuqpftCWIJ5?RUosAe?Lp`Vg{tRwN74U1DW5e@mt*@)Uk^3i
z!c?bCRk0e|<N43bWv1y?c1_pvoGU~qt@Tnz*hqmQG!`Nwp}VRg(W^D&6TeMy+>qWZ
z9tlZ~!ZGOQmjs1E6Ww(=?_n?Vp@6^IF!O}wO2yTM)g{-#CBurZR318&y2Vf|_ISvG
z$t418P!`<c1wF2Ci%fasun<T8$B|bM5xF)cXdSnQWr&+$u!166_F_eYAqmgJvOWy&
zOtLbyU+xBiIxo!7ACT+Gz2D4*C)KW6=-Uu!^2pp(=Y>OW13mFhCA_M>rY%2);Jl7T
z!C$TfLHg@5*j|}i8^C)qa5kuh=M&?hI$RAY65?IEr5|Lw+UKI+<y|Hz=ER1@q2dq?
z-?pnrD}yMV&C}>1t63#6d`OXs5XSbfY%$6Eie8@5sY$N{|AyQCr<oBDl5Q5O2-Sjq
zrjpPgWhnsP-e<>lfD7i?@djZcQvd}7u&GM9=`(s2PM!@01K2Iwl5v*4g0}xB@gGZQ
zukL#0q{y-YGwtp_Q@`VFoIvbka2WLF%zkbUKZ)yF3~p}^*FMnWFfK$vpc$<zT}}HS
zRJZ+C*2Wv&0!t;)pFAj_THE^SchOhO1Ys9(%@?4TBS-*3ewHhzkbc*Ze@9RC-;m#3
zMIml!FG7QYDXm7O)d-2v70r58@y@>LugMoyo1R|;+bB*c5cTqFsnro^TMZRLG>Q_O
zDD%`ldS~5^$uEivmXhGsfO>q*0jhXdVHVC3BhV(>g(4s-2oj6Q$S8Pe195u3mAIvO
z=WB^h3xQw+)k9S@nBIi+-dTIC4h-5$QkM2-zVk}$-I9)|C3$Llxm&V9KKHK_H7)o6
z<`2<&W?z~8-0+5J4Aa^WULl70%UVJbX~DLI5<;+=&ij#9EJn4eQllVxQLOqV%Xp&T
z=kebZ5Et?C)M&HjEt})57egK47se7=s~Rlf1ev3x^UvgV>2n9i$H(jIA=62l0&ezL
z{NAoHk+DUc+qBcT!LNPh08uG7JPz#YBlc=pn56-pj0`?&Dfv1{{B4cxLFrGT9^<1h
zg61aB<Dkj^zX1m$HdYc!rC7P>ucM0=Yokr_L$a@+;(ncNNwaI_*Z)I%HCY$)!d-wM
z4$QpU?V_(go>(iAnrhw4K#bUyF)CcV??Iqom=*$eT4v{^>EZIB)&G3F@15a8N~Dwm
zR2sfh@N##D63T(VsHu)k3C?S+!yE<^*DChK4J->+9&tI3S9nL;SC^e4Tia^Hb-zzz
zR(}7sof(=w(H22W8G7H0F<^tgRuCuXrx^ulJ-;O^(yEm~QRtExJ_E8=fCG#(iab$4
zQcB(@`^U}Pu%K+J^MX$*t)B$!l#o3YONGH+4D^)0IsW^iX$tYVHAnu8NWqV}15oK)
zpaK}?d=%byhQ0Igg~srs17#(fpYz+JO#dxIRAX9|LGx|il-QkHZxJ;g^mz&5+Vz(z
zK4lK+g839yGj%v05;KQeM2*&{X4Ru1-EB&^Me-xuRd$ehy9XOK?U!^^geaH|La9mC
z^eV?J`Jl@uc0Q?&`SHRL=+YRFkLdZ!F9CK&&)+ONIGd9oOZY_3N-9w|-rzbLC7J}k
z1Xwm|64QrrpZYnf2#|%>5i;+&*`oBte~dQRurfXPa~%F$`)exw?LSdLg+KPH?;c`o
z%BiT%(eu!O23CPHE1F8aWzwxrOan)l5lvDvC<{H-)*!NMmP>Ka!T$tO%~Js~VJDRs
zYvFUL1^^&L%EH3EsCsLftvdA2e|Z!n5VdD3nv>FPeK+<|$AceY1gMz4e#BqtkI()y
zx(u`wcQ=$eE<=aV>uMnl%3r0r*v+sG8PX?j4GwhH560cf9bLl?caDcPHHYQ`>(iNF
z^N$Ou<b4H4QJIaM`HI}r$om@SA%Y*Q@q2|8Ge$~eC2mOB^+&?Z+M`Fy^rotrfq_Zt
z@4#%+?T63i<~1}}glAWWlYiT}tTp>tNWuTUk8vy*!!DvQYI1idNZoO0^H;^q!no^}
zE?`|5bW^!8`_nqv_^l1&LK8m$DY#z5gEX%7^7GRTKw{QO;iZG`m#TpopF2+#bh-}R
z?@ibi`*Q~F9O;3>|G57(m@7<lX&;4qTK*g&lb@$zmTg0RLdY>i<$jN1)+&$M<0!Pa
z#?h=mKw2Q0?Y4)d3)@TDLUYg2oOb%k)LSZuEL`{SoNUsvA|11oV<VcFr3l362JpL4
z88w!ByxkxRB$pE{by1m9*OS$|)%HmMmRL@Jmg)5UjJa@v)}2WKB>y|h(F9@K<sZTI
zmu`!xV0;N}4aFNf27wP!5J3lr+k4EJ6A#5Bo8w8x!D+$dlcGz<CqK)++rE0u{d;a5
z3O!QuX>zqQ>yN3hff6NkGM_Mq4~LIeI{s}R&9$dFCe!<acM1qBLvs`1-&9vqlK4n8
zJENvv9K^=|*XOA|yO4WiIIO1B>np^$>2hk8Ab_C2Qt6x`*UB9%n|G%jwhX)0#CAzQ
zv``mZrKOTyJ5^KuSEA@4?g0ffk7GEy?xi(0EsHL#-j6;Uc5oQ%eOM8%0v-RRsSCZo
z{OrC*XrcV1d!U&)C89gJxd<mT{rf!$_Sq*ckr6O6dJ~gsH1Bf`DrCMQ%O^t9<PLmG
zCj_^y3(vJM<h0-g=KOHUOk6M!5Nu%S#CckQO`yltEDC_FnXBrJvzLaG#xn6|fkCu4
zLraO^*usX=5C-?m`kxnZ)L#FK9KXm*Nj4B0Xi6-Ihc#9-k+lo><Z*{Y$(<9?16!uK
z!(bk%r>xkVkAA(q@>0LHQ&BF==%or_lozyhc-fT=z?kz-bR-<In(2T40uMi?tEwtC
z396s@&Xq?iFro6%JiaYn`L&v*<kmSjV*M^pbL@KE`m!(U7dtD*&5PL0T#LkqIFp0~
zA#E|CdDleTnJQ2@dQ4whDx&gCvkS!t^7-~!6Z0*DCd9_())|w@m`pQeauBYYnL{o>
zLwC%sXjN>AJrJR+LcfEm7s3t@*jAf#?!K_3nx0a$;mhgWr|)ZMTO^>kB~-9tZ@2y?
zhGK(4#XQp3ax0G7Ql9eKZSpZ&P-AV)ClYs6K7RBXw{0A?RVk_P1n|$oU0lZ7PgXec
zgMy)D20&e$T)-|bv-ui;pq8XVRp1^}?@9#u$XCs>+2Bo4C~*O}3oPz$P1?8%eE2Mi
zKK^t&aooPY_9Dc=!RN(ql?jzdMz5(xO$hLm5+59lj7t$xI<1R4Iyyc&Iy$5`?Xs?(
z95Yw`jU=stF}fsK-COePC-WW0%<Ipi4v$=GuBXjh6|n=5{W=JSmVfJHlG5Z(Rz;j$
zRmhbqqn9P7=vwz;9<5DW0LIXTc^z?!U96)IH#k*HPi1c<6$`IFOw5BxXhoF1V}XAN
z%G~j#)U}xzv9VLIP+ykm4S^yz<~#US=358v@Sp#dSx>*~NS)+eqOWYPSQ>P1Err<i
zgWka;G@rlnBP29yE?g8;tXkha*>9?H?5Lm}jQcjtX)=d7P<eiLn4E;=w-w$M|7foa
z_7X^)eGel1AM$%Ix%b0e7*~IWuCrB69W>!Z@riy_q{LUJ{5Oz5;eKQ-Jef%w;3)OD
zRbJ5$tnhT^Y=xKKrASELq=ZDy!LK-Yxn#G=V4ticfzPMh_Mb_L_~zNy3G_EtsD16p
zd@k_G#h`@c<tE>ZCtGu$mrI6n+ervF=h4bG6Y~5#Wi^TWX_dL}<A=WRoqq-%LVpr`
zRxYA|TDc>rPUe~N2j|RC$ui-gthwcre>*F#cTROrPjvj3e)4bG<J5gKO#h#wxnmXy
zAvw9}FTwGl>K2xA^A7rNdV2HJ_Z3WCDOa^qqO+@}Mku%Yxmfp>#q+p3tvv@TthtGO
z|8rp?DcbYJ%V@yg%+Hf)>Q@EbV3K+$TTV3~pT(OMCl|qLS(I1XQ*WM>1rwkkc^((s
zQo>soq+%zbw1OwcLd3g7%t`L&6kC|KjXVGd!AZ$V(qi0pcCg!72&#f@=7(|1Y%UBV
zU1qL8JewZK+<tvn7g4ZdZjI6X7J5z)zNJt_Ayd|6*91qWE=nowj<=^(&0}Y09cCM_
zNs9Dp#WYHM@AC3WGU52Me^&lV^Hhr3iPk#^3Y|du#;fJ~Mg&=M)#Ei4ghvresBpq7
z4>L=>Yjw`Jfm_51u2`jI`u#jJ4AQeCia9+XvHTu<RbXTKb%tNZT3@M`iSObWx^QFQ
zJXF^+bmJd7LgS@lN@M@@djI0xG5ay)Md7+Q^6y>sC1_-Wdx+cioHRQsE^6TXzI&L`
z6CrQttY|W>Gkb(JyB}!ixtL7<=thV&^S$BC=KRV5;6P3jZJ2?(SsKu?F2FpOeSNEj
zKTHmcYJ}mlB47xTnwL9Uu(ed|B;+0Jp%oGV|CZ@$&vIlGTr$EvfJPHbD*Ok;c)IvK
z&Nk8M^qG^LUDnAO{_JR*xx35Tq~<s{lsgl?zad2k(lC#uA)fxH^A<oRP;`kRl;~S|
z3UEYRMy#-4H2ZnJmji;UY{%K&<eelLIO?HcE4TCRb%-Kch#_>aVQ21fpplsN1rJxN
zJ=&oyE(uDJea~&ybo4wmRS(^19RY4oskZ?@qb2o+Y&Z2eja}f6Op~X*5_<viucwf=
zG?24cPlp`3F!X~g-Jbw=GW@TyuF?TOTe!~LElCha%MwK-NRtki5Q146Ga=WWS1YYo
zmWB3^xhL_mC)><_i(|`T9f^l@E_rciu|h8X=u7H9{OO}cZmQGS!{hLigB51enj^{c
zoFGaUz}~qlGb5$8V^Y~!pOM_}6$fHdBJ*LpH!@tG5jt4?f7PL~)oV;XM5e47q{Ezs
z!jF~@$9!4U$iL0>p~e0W6?)~9mS7k6{f4d71OKgQSE~S$S0ruH9LzNtHt4wo_7u^J
zCQjr^U0LyO@AHO6q<~&p71Pg;*)|ph28(V%7DUV;G&#K)UOpEg&k47WniT73aK2=>
zF&t%MHbiLdj_C)+TCk+AaqHH`mp&pN(B<9TXm+$swU}qX1FPne(*g=tq1Zk8;Q?JI
z{E#^leni(f9u7TRc0Ar9=!~vx7aJY2E?5Lj`t<%;ZNy?P8sSV)dn=#&mzmuB09k+?
zj<cI{F7C(trXAby&dOK-Z!dkjdBGkBER6ObEM#-XCe6K{^2ya9#2D-U$tPN*qQ3Z4
z?d!V+2f66sfJ@FeIc_Db1J)q~bS~fD3)1~s($q9YI9whMvC!_>+?wGn)s^z0nwI@%
z8TO6d9Md|M>~ZTjH+(j@J!1fEH2u4G^A?1WkZ7k7qLDl|DgLeD(Ln@#TNo*`m_0^l
zuH9B-&b8EjR~t+pYf>Mhvu4~e)fXp(ogXB}mHItRjvAc{4CTjQUAl@VEN+#G$<+o*
z__*Kbyxnq>uai_%I>`<hd*41algt$FlVwlI|B__q_I{H-7mitd&#OD1xw$;%dop)&
zBt|$nISxEIhWOa<bmR3nkxR?W3uFSd9++HT-J4Ba8Gmq+?dq3XDSF2SK<rG&pRmOV
zsxLm?{ZV-4di)>1CDER9=E5wWE{jKi%Te@V`w^}}8b)JuOVN%vO;O4BqaiF1iKZyv
zztD41f-YD?xa3>6VFW1o=wdEa63|pzS_kSw=oR`|0cqfo70o=C1R10t;k7}NoJE4*
z!&JE2{ESX}nab01KCOn{I^kv2lT$0Uv{C!H%-7%ad<x&71S`7Us$Atj>skW>>ZcC|
zy(T6NcJ{*^JHlI!cb4atcb^b`#w>^6tb@ds!wLVs_-*3<_l+5B0R+aHvD>F>Rle&p
zw-KS)MWCnGj)$t|0_`DkgzQ1i2T46V)~(z5h^l49l4p}utO_t0XUaQb*lIz9g1GuW
zY7V?aA{8RZ=9VD?v~dOKvuoQN?0(hZ;YVBaA;(dBR!-x!EE!ouJY4Sr6n9^?k&8!{
z$4wIZb=0D%Ld<zy1k-KCNce_nQPDTdFIZtlxqKKuIJwJgz5YLfzvOu4!Szj2uSvwG
z7R=L-;}e~%CAu6RyOKoL6k|(~d)sZOwua@R?d2d+{`#q^YJ?0v?W+p-b9ix%{YLE<
zje?k~fX?p1)O2wq(rz;YZ=x2w_la6hl(nD9o(n(GMeE%$H00wmsrTf+Quak}Xw>+w
z+5nX==WuqWLx(_kpG>RXwgR0TInEH?cleGSZ;#27_Bx32O?FD#l@{6~(^XB&V~*1q
zie7P}_iB&#){nMVa5LBP)w`%~`n^^*_m1Zdvn4))BQZ*MKgEy_#;R8E*!qQ8JX>M4
zMIXFv%eMZp^l7cxfY3jOVmj>v*5@Qhr^o9-QCVE2y5lZwL3mFj+o2!vuDdJtp(y=Z
zN$#K6j$bz=CsIE1f4==98vU1X&(hyUb9Yu0YjE(p7oDFmns;FPKjf|}<R1Kus{I<7
z!=#%h7eB5xnvIp3Fi=r7DkPfhGPk>;l+#njfk-$D7qmHZGDA37Kjh7=@dgpQ^-YF8
z>WFTxXUNNsX2L$Hr_YZA<=E`f`6qxXY-{Pp6F>z5<djPOZ7XwX*t5jUqjLNIDVIRX
zc1XSK_~x}8I(=_!caHvSMPf1n(E6)km$h%#kJp)NGso0uQK*B@m^!08VE<PR-5N^H
zz^)R(QS7WYu4DK)G^FWj)wz4I6M<u*0HuK+wB%%Z_zc~8H6kiDyA$-<rd>jpTOyYd
zK}eVwa=xt$h4_zE_RbDQByP_KgtoQ6pLF0%{v8@n=Z@_sHdnvb%>|t#L}lg_MPLpt
zDH8&E>p?OLVw<C_)mNN8760l!e&}2A?w&G-el2L&!_!3)CM|<DB$m1@WFRLOg>iC%
za;6F}{z#bUP-~opBI`0ws=l*=qQrhJft*JsL4-U@QeJdY>MgZ%0}17etN2Kbm``d8
zdpjQzMTZX4s~mmh$Cg(nGD-J}ZrZ-8h?k}@YekmOAfj?Ti2aImQGZ<h(sK3u>{qeJ
zk=$io%${!MsC-+~XtcV-4&62Jl#)G8fbn_>A|%pJNioRWu*K7!X|C?l>2rj2wI0%Y
zh&Y_#TCV;v49D@Vgmm%aj=J%GH{sM4&HXZ4BS_6|yBFo$0wfYNa!XC|)kVNSm~Kd6
zFbvE#8yh5fS?KGQuH@#gHnKF|sT(7)uL>F0<tr160JV*#huHqnY;qEa@wm<Yr5~+-
zWb{TEA3r#F&zfHX8A*oc#}b4D{3u9uzWs|}C!b_`D<$4<4K6!d6Qh;8yEb#Ocf68(
z6s+)2QWpS2<!cO5k{9h|kg+dGun}*HJkpl<u-{m6Y2<ZL-0p(Cs~pK5Z)xrl<YEap
zAJJK{hc<uAj=k1$<wvaGO(|z_!>56GZ`Og^CNJNdl&&M5FUV^{$VjC<*8W=#QWNJx
zavAj1D*I!D?zy?ZvsdI+At|#er@`XRH-Ge}l5E%Lne94bitvw9n%VmaY?rX@J>PiU
zX%3Cd?YPBcl;uK#f?xbIxe!h8dZ1@|g>SkPBK1|@54D@3hp&{XCL|PJYVIp;B)2{b
z%^lJjHSsgUy7LhUb2Jua?)b22jWDI3>?<!Xw$CF*BK1<Pmz)APU9&U4$`hw}bfk#X
z4jWL;?6@`N)JWur{uig#6SBJL&oKWc>C6X6i1^Fi<sw(KAH}ha_D#Q&Z=O6V%*Z#9
zcE&6H*T20pzr7MdlX9A^tQ;#FP)|0WNilNb1X%R?Q7#jTATNweW22=RuE7~M*<pi#
z-w#lY)91$3jbhQ=j6%4pl5d+{P|G_6%&<`Y1sa7=#!}dRC}rVWY&e{aZ3aQy62-bL
zGi$7)jNg3WHVcfAx*^9rR7%c(fFo4Sp_I5`26?BGM3NUZu_n*m_a@Kvz!UsUbkd5B
zc39~VgoE9VBi8ppV<Ko^g{Bjq)jFDi9}I4m!*ipi8c`P;(=I@Rdk|adK(K+6b&Ckh
zp>btC<P)(!(9<*;tmHkexlps_a~p#uHZ09Y36?qpj)%-#vm2w&DK6BA9cm}J{y7xZ
z^=w~#deNxpK|QW&_ILxeVRFg>rr#4jYf11fAdr_izX#BIHqBiWp<qSvG41PV``OqW
zDET|MuutSxN<_Eh7vT>KB><_xJAWnnksI9QK1*){1G@9Osm%UmyJcq08hvxMrm+HN
zWqB(f@aNg#LI0SuC8vYD$v2utqyj%wTxj7D4VwB_P(+Qz+gn4_wzbE?oMR3NnRK-z
z*O_EjVu^|im2<!EwjV${c|MZV#m?|dQyJCdlvMCXDZGu-ZTRWt-rO~n;hCYe+cqJQ
z??6P)Dvs4Z9x~Hv`<z;KDuQPDp^8X(E#X%2HT12%{MWy#J?#1Ld8mF2(wzS?APb8Z
zG-CraK+7O_H@FB)T!M{1K7{~+xWz<|*UzDE%)Rt&O7Gtu2{}1B6g%9cxL($4=uNHY
zWn*1lQl!`L>*n3&_s)cXBZ-BK1ie}CuK&fkX77WL*0AmL{-RUoERVh3To}4y^K<34
zT2P8!Xu@K0$w^Q~g~_QTH$7mryV&mGuGrd4$H^Y|xq~Xl=|?F)0ukxn-+`A1E5BYa
z1LDhzp+w+6TL4gz-;FQC+T<>g%h~t_IeUeTZ1Q;N(qYl9P+FV)wRF<nUUQb5%fz+x
zV4?NAz{~DxyLJIf<yIAs!}T`3P2aI!z3pkTWF9#0vIB9I{ASBhc*sy$%#CZI)F`ks
zXbXzsx&}k*4dd-={i8N85%)&~`z}UT3e2>j;d8pL%&g|F-;#DW@r!K>z23b)=}0Z-
za3zr68<l-cv{{TB5N!i$4T$*?^2(<%tb#Ubtpd2ghb#nXCohIil1Ji`!0PFVwBv+M
z6Fb?96xT%5)?>b6fk=V3qqfVY6var;m~ad?St-H*5L7Y6sBbWoUA0R1Vxf7<&I9Ad
zroUP3F8Tp+p`f%@S{&yW1iuF~%1_^j=jjzqi8M&7=*r7A1|NSZkuuXB3s0~dL<vd>
zE$nP;1f2*El~5YIm#KR7)Svvdy2&OMPt%uYSudBKT%!Zp9NFfNO?4fdQ%h0=m|LiL
z_q`PDzw0imDa)s(fHXLkHC0AR1|4mTShT93lR**xT^m~tY<0-mSyY>Rsk0=PsH*@A
zljGLqbAOv)vx!nwwf~=gn?n{t{0r}Con<DM-d;ef!?kmGN7o5!6Z;#-_A;YkqNib{
z4{-$V<u)Oa-M^0?KgxssyP{i+a4oy;mm~p5W7gKi7whjedR0CNoQCp+ygW^@ldXGc
z97Zd-U(ol?Y4rWWgt?UBiU*afXQ6GiH%$AcGu3)(Rz9A)8K^I$=e7xB=UW})CQ1|(
zMb2BhPVj*89#!zyS%1#OUyR=gHo4$i)znT<j&SRJ^yU2{qnC5)$sA&#|JK{X3Ati-
zvtoQAMKQ}NSZwDZeQy8a@vBGkLuXv=aPm#uZrY-}z1BEU`mm~Z#RKt@);I3goMR}u
zBHx8#I9G(egc9Z1J)(Yuc=9JWNe$m<2zG-5_#hIuOkMkPevrS%oF%hqd!QwJ(gQsS
zRA%D?U)86CRXY7>)yH%`uuU(D)F`AYz+o~_XI*t${+ECAiRvO?sc<z#lIZ@mNcYJI
zMBM5u?EMcy1Q@icGB!!i%U>;b_tA%UQ9Pf7=$8IIm4#eqpMw>|g+=|%dj}~tIFeet
zCa6P`f-GSV@k1mk`IIDz^PZ#PSSA5qc;TkBhuOUR?U(t*-pm#iXTt~ppwR1qY*27Q
z@R!S#H^l8Moo;ipViaBjWXijKDc24TDJPI9MX+#sM)WKGm5Iic-;~+*yaPKpng*@<
zyKw=gSOdT9Jkg+hS<*!}O4132X5WgS9Wy=UCL+KT0QpXg&JpI!eUcsSF5$<PwW|}m
z1WV4`DrRB6R}OX25$)dIOxdW!`1=+MuSpi3L<ZW#-u4n7{7YN9po4@<>ZtA5Xqn|Z
z<2R(g0)lMm<HD{9Y|x2`B47EY3lg;PRsT_HK=8|=`m>g|(7Zz?JOZ;IF}%kOxx=>Q
zWoCa>$MMRT<H^Eu)01x(2}Yl<p`wy%MmmL7=;4g7p++X7kR}uB{>k_6l%NV<bB^wM
z3&Fr|v+u9=puW6L8C7ah`(S3^dL@BzLu1s+vmdBAtg*wc*5j0>7D{5PRk$GHGh2yx
zB4fERnI@hmaM1ejBiScS#3YPnZ!fDw!<KCS&a0cV!dy#djRMbT6UsU>StWuBJ?zij
z=T2nU7GRU*TS}M(unP%!3N6dR5cV(^7tSKrvrQMb&3Xn}so9=X9u!-Dvg_pTzvI7z
zBVodOvE-=%am`*W*g3bLZ*g<h3$dSU55)-KwLy~aRil}o7W|&>ZnL!cHhb3Wl<Cgu
z0@{l`P<n}chTDW6ZqXWhX(X_|ZXDdCYveJ7)dyxw`rAlFls;TWM0t3=a+lh>G7}Je
z_qG%8PlaP+e<01iYD{cxZFiAXey#bG#WGy4V^9KLF{(y4z+g6sred;l!M)nT`Hc@e
zC|;QUKMP*y`kz!u!%qo}1a>Ok%Aj-1)-}oY-S5WLHo^n>^txlNjgVG<en41Qm;ldd
zFvY-bP65BE{G3c&x>yZ|yA_sr051<U<QyJ{X`<N4O)FjHx;)bKu#1Pzc;cCaBH^>n
zB0{U86I&}QE6Zb5zP|FzVgMAXn`f-re$B3gJrvl?9lnMQyvIt|yDYHqXq)Hb!DsV&
zDFDEaP`5rQmVBko!Et)J{Wu>nF`<?t>M$XTd}S=F8}aI8foc6*hAg7Ad~sn)4ob_2
zSO9y*IFTdjx4EyL7P`7h{BiQd``>*rLGlE&+?`^GCZ$ieKD4UnMj{_vv+J-=Qv9tT
z3c#5n@Ag(aum%4|j4t}!mNl-%iJh?bY2W51Jd*Gxi!?H%pxMtgBJyK0_p{45?~F}5
zRyE02`DQ32lU>z0fz(Iq?o3Nh>(X<9hJ5A>kR!yub*d%M#j}uQ!aJVJC7iI<bnZ~{
zqVMj^&fyzh8W?{*s5`*Uk5xVTcf4<ZQ2&G#q*6D(nzkOax;mU0V;6HJTP~BB0VO=K
zu2Tfx%0f-~{P;hu9e&R(N*pW^cfD=e2L!PU^Orw!g|0njg2pB>p3N9E1Xr0_!M})>
z?KoKEJnVfm%RqlM{M!F9VbSE8vGj7AL3SFXTK1zV$lrJ<yv?C%jJZ8OzaI2w&Ke63
zXtQ7bEvITJB1*RfeCYN#B3yr+pekN>V?t@F;5$!7gydCNO_;mpLWD=iM58hsD2bN*
z4<sT*2>Z4sw(O|WlYID%SLUu`vmK{<Ww-J(n+gLRu;_8_$uJ}AWc{WAo9-as>BXVN
znzfyP(AKtVyYsoOg8(-Y>ba&;$6|W|V{7^L+M5<&;gw>@bvcw6fZ*&c*|6LGplRh^
zmQAm9@IuT%V^M*e<Mpbe5azW$F0haGKt!qIHDAYeZSCo|T_pk}NJQjM8_7f^vq}#&
z0I{_K<8WS&p_%o1nUnhMobI*Iv{?s>Da04!ex4-P@7Xx$@Je9639rY3OXT*nKZRBx
z=k&hqy}fUf_p;aw!9-Jx7X{N7w-*v7iv<4uDZ^OIp3gJ#LrGybb@ThT&yzut6Sf_a
zf3^Rybd94q`o$q@en#&p*TJLxZ(=^1%Pf>>)5#&>WRq}GmEN)hgh6j4`i^lc$~9^H
z^yjH-SPQL4UtR)g5oFutSc$X$%7FKwC+lk`L+0RE#>Ak^eYP#Oh-f=s%6O<1xxtb+
z<j<hx(32vV^KeF6(<8uHL*5NioRaErAPrJmYYF|DEB72H2hfGWHrP!BR!pr}H5`sP
zUD+f*b}mb1u?uja?vJ$T1#T<ZL9P4@76623mjB=(E?@mht96#WZ+cVH@_fDtL`6}h
zr?dX?<G+)0$H)GJj)9K0{~6%e+}SPV_<X}!cpyLse8`rC9eL7|{{1&l06TZQetbui
z8nTo{O8Kt6V`B%27`3JOGulkppLAX6?hm|rJI#n%9p_dUhbor)yjl7JZXdR2;h7Ip
zqgU7WmAXhN`q-VDRbD*VA3AB#$z6S0(bm@f;2KlD<KM(#<J#e-n)c3ay6rMhh7i=V
zYZOM<OP-={{Vmnd8MVg-v?*E^mf4T%Jf)v7aN?V+XKr<{E916yglFfZYKpauAi189
zQ&t5f6467X<;FHvFWb5T;CyXK`i@37k_4}!d-d~6bX4!SNJ07XcJ91svn!B-H(EBt
zi@1?9m&HhVz_$ue;9T$z89Bn`0j0X`(LS57v{Tt!t^*GY1~;J+G$UOn7elzyhfpko
zp_X(CZt-;hq4`)vszgGZyR4Dl|9%X$#nSo{fd!5C0oYW*fnF0NNxsR8ir~l%$;WY2
zxuNp1yI94YIscBe<QD@XI9>M33haBe%BWx4^Q;tiM~4}QgHdY`jR1OA^zw~Ox4uiR
z$JNxy00JMd+j&R*XV6ks{lb#q-t+(xR<|LaElj?Z8r^5C#ukro`mVT`;TnYf-&f^X
zYuspdTZ7z)R@Wp1O-gwOKL<A<Zweo}io4=MLTO2g=~-Dr?)U`7#k*HHPF^%=cKT~B
zT06Lj5~TvB8Y|<7LL9$K7iJGqvwll-Eo8dx*~4vzZHjNCi>v@u2t)dhTyn6ka&|j5
zlQACh4#k(Hp^B}hofh-*jt(8+&kpx?w09$gXK{(_E*9dm$~{4=biA*A1vH+Oy`XSE
z(Lo|V;Wz0PG2hr^``;MvnZg9%{(()wsYShVTVu0(D73JQs$iFgD&)EkRB@ruRsLD=
zpS#5F-kjri<w}Ul1Q?t9UI|A`gqdDn(o#Pd^8Jq<qSto+DOnmAR_?r`nUt9vD#hoU
zejD}W!_IM#<%V$7P+S#oLkY?U5iEtDD->}X?Un#4sDq*@27JbeTipV~ivNxi#MXA&
z>2tlV#fRX3pG?Hlu9;@pBq)wjwexc|J{>cXM8Ai3l=*!o+-Kt__ZMwfGAz#c`R8X0
zp<O*+gM}@LR|uhrO;*r0H;a?X3gwcDUexgXQbf)#ut8AeNi=P&F<_vfq5ny%3p^<1
z)8;x#U*N6!;rw|~?Ach9{gH;WNAKvQ)14-pH}f`L%ckI={Jj&=U*ccV*z*imBqhn8
z2Tm0X`uAEBvY#O@;1%wzX^|c=l*gztmS<avxzP9o^&XbuGntII2-;n0U+4v)I!!R{
z-l*5QEIXuSOm!p2mGL)2_g}Rp*NE;O`V%Uj5h=_r>H1m;PH~58M=L7_JAq6UcS9Hi
z1Iq{b*lCG%n<eMwO5S$v@^f^xI7rE8D$e@l-bw(tT-&7mv1WAT%brQ;b40_HGq$%r
z3y1@IglfcIO?|y261oQW!YRt|HH}KRR0jV!?V=YV2*`NHN*G@c(q=_}$TQO@?M!Q&
zW-eVAe~i%x0k9N`5)XDDSHERrkup7dfg3<lfGgdcpENHNxh##h!bpL{=yhYNS^HNA
zVy%xPZJDf|!xlH_PncFst!P+YzFl+Is$1zP-We+7!wj&J9PIYXMNYmJ3AvK*&AgUA
z*tM#d<j?S2y|>=RUgdAzEi}%?_o1r{@slO>r#rU)+uJyL5n7ncy0VoW$|60v>WHrG
zPG{_Q<u|l^rD(7%6Wv8AH4hpq695D(Zh+>sUyBMz0wkc>LA$Oq&|zdBVxVH;T_0Z?
zS76|XBXrSLJLxyO{P!zgpW-=#Qc`23`7?%gjIipaN+u1TuT8qM@R?B+NH$~3(~o(R
z{~t})9Z&W5|KIjC;$Gx)WnGtqC^IuIF3HX)TV_T^Mz(v&xJFju8X1*YX10tFa<7C?
zX2LbDJ>wd``~Bnhc-;T*I`^K}>pAO}x8m`Fl*nM#-w({Irk9_$e3bQTW@VIq3`nrm
zm%Oc<dh|X(IlpC_v#J}8`b$V);6Y{pQllJVjD}F&lqV50o234!8{CW=_-twy)(76_
zt)A5KWH2`Dy8?x}N5qNh#aW*J{!B{K<f8{}Hg~O~?dEOm?Ch$ubm4*yKi7TtR$=Rk
zLdL1iTLKZ6M^7uPwo{>L1Ty@{3mOfb6k>8?7A8neI(4Ci-Dyyi&V;%K)q=Gj=8_{r
z{qhMZ4EVDd>7u=fc6!jtx8zrXNj|-bvzE<|O3j?_!n%jpC<W-t?=#*yiWK(~GJZh$
zQT(}+L>fa|T#Q9?#a;pEVKpz$&1@_43Z0nf6>=xt`S~0A+wFEC|GiHFIi{eaMH@ZC
zO*|d5e@EcI3Rc#lSfwR_93`MO6-W->I$r5NCVvC7OuF4%p}$0}Mh~Y@Cx^t_<a+TV
z-<!<O1*!k;1SJK}vxpw4yeUerEot*=-=EMV4;+WKah}e!ED1mL<vep6HB9RvdwDLY
z*l(5bZP{k`-?s&q)OXladrsT-^yN4v8@FQ0j&0vfQv0I<1NWExR3JOPEe8`Bt#$Pi
zqx!iI=hHC4`#l@GyASHM{@tW9!HbDjmQTZ}t%wnw{cWMNFrS8qHq`<aX^+Z)@yf$s
zCO8t*6ma=(Va+eO+1kviCB~E<Ff%i~+x@YF8|V#ZW@&r+A5wxPu^xnE=T|d`=3esS
zA-$p}?~4}UdX<NsZS6B+f1jZ?zJ8%Z639Mlmo5F`2Bz&vMi9$!?F4oVugO(s-mB6F
zu;-)(wLs($2s`5L?r!y^;)+z$q=)Ahn0idS1Z(L@o;3e-BY(or<ouo?{x={qq$cYF
z)3X!hqZP?}cU7XHX?{bpx`|eV%v9?#c^5<yDoG*B^n>eC&WAtwNHI}rr?Iz$Ut1va
zb8WCiF|J_zr|Ca%m5aLI!{xWun#QIIwC;iui864>{o|D50K7jPy7JJjBdq1*b%sEA
z)j~&w6tbSCu<a8p`+BmoiC}Z@8JnrOqicqzMigmFwIk&?<#1*Z+RW6G_=*#BAJg7|
zB4c37A>o%77hx&|DWS19+0JGZFJJZia(<p9IT;e%!k|NuW3@f-8BZyJS-*0!686zH
zm!{>#KuZ5p)P{ZlF_L|&meL5SSstK=d>&Zf7+X2T@HM`dRm&hw@`<7ttNP$TL&?>Q
zGJF!oNFqob0poX~7GCYtpf>$#5cIVgwSLUVVwIToZ>;yKi-))z8;9v>uPgXLiH7V~
zm?N8+yYow7;d`3=9{rE6r#%d`IDB2r$w`SvioV6Kc*H^>jdSt8#PKV>b{xD$or_P1
z_jBT0HI{bHcF)hJy-{JL&FKhKpRXQhSkjX>6rn}C#@TJ^i3|UDUkFU8H`6MH`Vb)m
zV0;okzEbDugCKG{K6l_3F(4skxm*+M4-G&m(ftMoV7_Il|7tR~?;DE8kMcpAwbeFS
zqK|x;c{mX8Rd9^aOE%GKm68v{5-~YwIf4*eUC3H`ME#KLNbC6guWn_ZlW;UCevo~<
zA}wB5aZ@w3YeKQlexb5y&>fX5J9&O9%=#92*u;lb;_-h#s4c^NXAl0jR<O}LTJ5(l
z?}G<sj~(T#9GM?R_SPLX-|J&2bJG3_O_eZH7+c{6ENttXb&<2-^C`6iED<Zc;MD)o
z+3(uuzrAIBQ&wSEioyhR@z+Wofw?!{iN0Sd$;m;-y`49&zBV81g<=#=tMUGi)jvfx
z*%hxxAMiH=-jGzw;WA{EO%s_~b|_S7Rw^#NJSdUtM+r!4L|duH*y=tKRE{?S#Ei??
z^ahkAr63ls%+&?aW2*Gu!a}@ijBlI8gCnnBH&bd+s(nl$V-^Ed?xJMXhCZwmf1XrJ
zIFNO$6s@~QD0yi{iE<Q6-0^)Ck~ex4d|e$*lS*ZtXDKN_pIFCXg+5gDe0g$qPKFwn
z@>aiWVV|6Y0Z!SFb;N@*OF7a<9@0%4IYLKDEnh!zI#IZ3Zo07uvs~K5T8>Ojm<(4D
zqX6Qe-}-m>RGGET_KF)G{)_rP7!!pM6#Uem)5s(pUAG5A(W`^0sXybxc~Z?c{^ILF
zu%F06u_hb&8Y5%+C(wo0gYTK5KaSgIGWrf_gBLvbbuH?~9A7H*q4j?M8+O^R0K`PG
zxq)j_ra83lv#&nKVpt_b644OO)PI~`KBR|=z@AI$l`n-ofX{^R`SJE+3Gvv5fR(WW
zd)$zN#laFPnw#`*({4V3R87W;FT(Yk0`{cwzI=^gG||zPlwhiTseT1NW6`w6=W7?`
z4__|IGyW$~5<}Lyd_I^nt;%BLpR#VyDA-3Z<x&acXq(Wu;2dsI6+N9y#w>K;v-Hal
zb)v8h(t{9a=6A<@dI-msV^!9)hoc)_3Ef7rOh#cjD~zpZIjR!n`+t$ShUbo5{slM0
zwr`KF?)Q`8U~Y%-GN~+kN%)7ShTtD}`<}tn#o>NT)fWQP=sfb2G?awVSEgE#8UOx%
zyxWnW3A6E#`j7cdDyPbe5t`zmx|hxsJ}yDOR9=|zIB;<EE$#-1`3=Kg8#zeIIDWhe
znMM_RXC;IBmB~j`e#6{84ak2?FYPPVPko*J{S+kbgu(94@i~3;TQa!h(Rr{lbDZ|2
zl0MOO3@)Pi7b<`!LlU-kJag#LW%?FhXTGJWeo%$0#{5<e&pvbz5Usllhl6EZ(4QKL
zg)dW=oy*>b@K+8xF-#d+h374P@oCE8K*bg$d~Tz9IPG;9D9tX!Oz}MvgO;qI?;;F1
zgCj9y!dDby{?NdR7Ms>G^P8h4L)px`LhxBv3to42+|)~(@&~4RjQ5xf{qu38i*Wv1
z+1sM0&yIiAZHHT-q6rm0A(pDBHvM??s>_^^o!i)WC}_UMA@7b4b1&u$no9nTqATh9
zgr=8riMW&m7D9R&?@u=exK(6luYq}0__-l4hI&x^l8>k}r^VXfShu+E!y&~wLkIS&
z?l#zLkx^+133>{Un(zzqA&xFj9gA9Wxq4eq4W2-)Wz2!b4=9xOE9{i(h%70-RFhms
z#;|)3(>&gAZtVpLJnWrSNwh`9Sl$f!e*e=jvLay`$Qw&`q)u<_tMF8sb<Bv5#qjG@
z$Vu}po@`E&wmUC(&w^YKmN}ZyxxXbJ#v6f_(6Ot_s-K*Pt=FcM64MJqdC6zFY2Ng`
z7<#J%zXH9ZW!CYNL%7GREpB$(ap*Nfr<a<dl<O1WL1vMd-YHZ7mW^71D#7X`U{9xG
zFe~yvI?!`BY)FB3_z9LxM+=Mw%M!@&S9P_wUOgWv@0~%v<M)%zT+{i~-#I;%Mq{eM
zi<c~_^ezVKSDjD&PJT=%jo_9aj?t88EIqxA-O~P%)95)?SzC8SwGZlud|w0i5rp|-
z+Z5oG_Gx(^UTQ{bpyZ_n1^vuok|d552;A6~2Cbkvo3!6rmhzLmWdnWafxLW<#)rSL
ztl$FDm;<CJcZj^d>FB)WJ;A1=oCk6H(aIo2WAudiQx*|gP7X!Kd@M`zX@`85XZJkB
zt$~c<27R7%X)8gBD#_RrsBS}pe(LpM+KmdIzY%MZ5<?ZJ385GD=1UDXPB#j)kQ@^w
zqB<D+j=qZ179FL`7KKMr542AA8cK$LTY9&@S<4<P#73LP4S@WteD1V|PU#p3dD*;l
zoewOmd%a{WJu|`m*Z?D{OrXv^Jk3?9q_$5=QtnF@j5POrGcekC^U>U`xD-d#OghF{
z){&ImYVTuj)co->`~7V8f}MraIq6#A33QiU6EWa3_c2KT*8L(v3jXgEV|bsG(!@-E
z(D^bBH+g%qDT_J}M+kn(K#f-&dwRHi=AFj`V3+g1A74d0Lb4j)lS^qmSYY~QY-MX%
zs>Hz$COaG|t?%(CM8%XT4>tnN1_%g8<>-zHF2M(d#vllO7R&0<tj`3fYeWwZKhZ?T
z8@<o-+LEnNhchUu%Cf#(*s3ZDYD5>Yx!k{xpSlLm2_?8&vahiVIgd=0nth}!%$d_U
z9`E3P)@fbbupFVtM%njSb!!oNOSNzD<ZSrzUkX`6@hgQhaxE<zz4L^Av!<mx?!>Rk
zzJnEoRRErT4^Muc9X^fuC3duk^>9Y&mk#em<L!G~1Iz?#YQi^0IngWJsm>D=R?>;8
z*R7uyi;W3G=*j-LjYVx>e9Lx_itGnzF|gfJ?5kBRNG%*)|GMSUx8dT*?LrJzW>iJ!
zxkM<Z;Eghca?;FvPXFTvr|wL4UM_w-cS1FevI8?F)NJx>v!5E@f1Rs}<z%(>QZX@5
zqBLbQq370l`xfOw1d1H=K}O<%a3<jFn|TVGFqL30=G-?PgZ2T`0rJAi1}@TefeRJB
z9ufc4Y?#(UPzUTSJu2-x@=OIBnc7rY=r*!&<e)1yJ4=tZS;XRoShM@z1@+lbYc0d5
z^P{_JO_g!ECo%QpH3s~!t}X#|QxFlWIaffEx=k5msxJ-^c~^7G4-n@1OqVDz9*_%B
z)BE~SRS7P}U}z`3vj4$HZhx!>-S9?0M-V+)Gxn}Nw(g(|goyQ#_oH|Pt{Ay-cl{4i
zZ<KoR{U^<w7H4_NAVu>NF*rYps^UuhnMub?s(fVTy@niDV$5rr{QLIFB<)@wtCa8v
zk|Ro4+?Zj-7AH7yJK6MMp&(Q5fKytBXH(XPxHg4_;o>i-L0kG|HoL}T*%lmbz9a0M
zRGqTYF}EEd>UuwwBB%pA+@@5dvkK#v1RgEzw*Z{09|dus{m8ZGAFoll!qPX;!Idv$
zXK0S8{XkD(5V!W@%16vsVC5APloZ8}74UHXI$rS?;JSxSVlRa%HhAR$g)ZIV*s)4Y
zuZQ$s2mK0$2n96uWHSvp0(@rhNWYf}lG+8I3U$_nF8KZi1IpNJ3i?6`I)DP2Z_zEN
zJ9ewr@?X&OaS|wQ2X)q5t=cimm$CpMQxKQC1(&z9se1=$8Tq4uLK~0m=iDjJ^aid5
zizyM@<-SsX`ql5(a!=^f$7M(Sr(O7!LD6Jw3_18LR$GrF0uKRwTLr^)FyLqSNLZzf
zb|YJ=?;wqKck$aPDC+yuL4gH=qbgji=k7#af5e71Hr12{#l$%-X_b$2WMb`@^}-uB
zCn!8>6f=r2alTM!-$Pxfy(c$j6fDa3MPMsi@B$b?sxCkA>fL{JWiy$tBazW~UdI%g
zAudRKy^-sb6;VG|n~j4y&!ah%5+fy<P{Sde`)js%lpd0A@0HE3AL@}s^i)ljCHS(T
zsh@_km5rf3!XF4?L-mnjaH%d2xd*xo?==*o-)U4S6daQs9>5<Wj<IqBZzbuf<0ANd
zs`-EeL?6`2hnY9y0`&a;G}_D?6jatUGQLARga>RnQj#Gzs97@xkI<KOXb~`RNj^`V
zKne==<q@jJRdu7$rR+g^0Q>6+Qpa}VLnm{nvy&n#TY-GE`wMO1GJj-YO{0(~kc#}q
zG^o8&w2HUV6mo(4ENZRW>W>s>(LL5$5|s)Tf~Weh3YPTHk2%#j#2Yd?q#1}MVd!C%
zPYRzTYq~t!*e8*~F0uW1KqY2N`O!)AR2DM@6$i><_3|?2iTdc0Z`;{<0cxI?1vZn?
zqjl;#puBquT%XIcYTske5McDFR6c@r4~m>v+S8@U@-N{OAXm!5b*Tjf)r<jJslkcq
z!{_bRa-QqKcW`4Ok%iT-)3)Y9S0=FZ2&tYP`gAEn`N41Q0Hsly-j3PVs5F4LmM^Y_
zA2o9Ld~Cv9-VRk<QA_v7vW2xsq5{E<hN!_uGcfO)t{CO!)lwdtZ;j{NjAlwJ>`fRD
z#5@tEtI^*lIW~V)+_>01J0cygIHn*2Z@68at#rQb1W8Rwb~90!zcI7iu7(UtKDYDs
z_NM=Lago{=ZUT#P0_nw}cbq#ou4lZyjuH@kC#a5%UwI&gJlModP7`lJ9@_I$7IL1b
zt6xo<vyg#oACoL`AR=Xs0=y+V?Tsh+a7#uyH?8C8cLl;VvU>Et*ThZ5Y}_MD$%BFS
z)Ns}Zz(#I-SwfkP);(%-Vl^?zISOyyIDt*OFP4Rq-A6*RAC;&Hd<NgDo=kK_y~E`7
z%o6H|%cE@-N0r{~c8w7bh_<=MP=7!4o(ow87t-mKbqk)h%vF?>=f~6U$KCc@a9}bk
zE23q+@ot~;;3Q5+aIoT|)wI_nmt^iq2VbuvK87Y+%gKPY#0k`=4N6%v<a6XZ{vV9e
zmh#*DVABlqVK2|%FU_k7?guEt6^2*AAz}dZ@^m+dd%kmizT>lXRlE0<2WH4du4p$R
zS92Rhe%$LE-Y1l`WOWHYSc^K#6lr>=$oWIG1ybe{lz_v3OeQlHb9#c>*W0yP9|Ut-
zey}Z66ulazu4xhUJu+1^8a94Y6C>6uHW0QaCSdRMJ|>$qDAS85`G}f$2?~tU8$|Ly
zW-3IAu-x4b)<QE({X%jeT`Nm#Z}asH$KBo9-IZLS){bRWt$V1WjaGmE6BW9Yzzv~<
zf<C@&ZzJ2TN`DR(MCMS0wHy*RPfwSZms_0)lFU|qZBOkkj+|AKNhv89o`=RGZoptv
z$b~w6sQ?`ny2rj!8O8kQYl-jg3g<7;4r8$cUlCPuWlCT{0soQ8LtV1gp(LwM9}PTo
zh@4NPH}>9@q4ps>rhQ-*OD*juo%rW(jgY!Sh-qmy)x;kG>61^Hdg~nj^}Xeg`p<Ak
z8mU=J;q<jZz{HGSYGMzRSfhRFf07FRiG6lUA`YN_<fd-D;(&rbSi`X1ciH(TPZk3<
zd82+EI2MrAqZ=eIT-qv=D_GY9Z=JRk%_I$TP!psD+$+4IWb*q{yj976<J0f`;Pz6q
zR4`g82?mCEu)qrmxwOiLR~l}(nTOD&sgyJb$$F}_f*g%xzu?GA+rr=A=~0<&p<dzs
zCb44Rpl0P83X5k?n^Tg$PHV1dwtnCO`P0a1Tp#jS+enr*esN|=0)FqGb-n^;BF8Yw
zMky#3QUY=YCAt&a4;#qHW50a--EO9h*nRVUOZaHxX4XVzc4f=}{^KCh8iRG9`R|9~
z_tTkA&ZS<8y-dR`Pb*Kt=7xmcOaDDTIT%<bo8hBTqXJ{n&2{A*UX3=lj7BwCssAM`
zD)eD-V&w2>Ot)i_!clE<uyV@EV$fV5*Uh19Xc`;QpGlL^wW3>0bfrps?5XDa(AfS9
z<@^*f<uW<F?7Qap4~N_IT@!sHA|-B6`EF0r%PSte3^P@Duz+d!aEQMqly%H5y&~zc
zI?jL!({frEk1`E3P0JwU6@(i;k(9mqB!<QUTzAbUu`v%{a3fHRPP|sk&#$Qg&NK>%
z0wF<0*-+N8^wd6if^{u71w<J9E_3CYleVl=4C$Jo1_wM+P#XJ+HzjE1{A5S+`^e}?
z)9nPqS>J+W&5!<3zaJyLJ|8lcXma;M396_7u1oIrJ-*J*pFf8OpPXKf9JvgobKlR*
z_j1|deU>%z;lqb-oEMj!8VN}<abq+x3^cH;HqGX??6q#XL&j9^R{~MsFO0@g%Uaef
zva;iT9s4dVZC`hYxaBpnP<M7ZcYkfDP0lfCE{A1^Z})J8nGQLyz71dy05hUw#0s_X
zcZE&z`-=0ekI^sa#UD2}R%UAekre#wk-3WPE$uBW3d5zpY1qSm4gB_H-jhn3OP{`K
zukrfVZ=C6im}GWJ!!LBS;M8}YVBp-}(5ZfTvZE8;U91hk$A@&wjvj1&FBUWph|YG4
z7LY{Po~RGSi88Ld8iJF98$R6uq0uO7amu2?<u@GAa-=rj#a`XryDE3@Dw7{CtEMON
z3+%e7M5f=GdY3YrBY3bj5AI^XsZo>=Mt&)`$-|SJjhD+V*;(cJLQJ$obgFq-538Qh
z`wX~AB~I$<g3r7hwYmu~jtkIF+Y>32(seo8c$Dwj@c$Hi>o(UBYJozyTMTgZ3hJ%-
zrF|oGIxeTiDFV!xTX8xdezyhu&FGO$k9BIMVsSpD!05sT$Grrb)L=!v&j~f)ti6@-
z6G_Wxdg+HbSz|K8$f#fVr!ZQ$^tffV8`Jt!t0?%^{@Qa3{8@-isCTqA&Z)$46&efd
zE00^w4E0EX#x(*^fb{)x{Ldb(?dgtf&lCw@duzJGZgR6E#q!2u!N-pn=~#~Kd&m_y
z?t|9je?n%DD;qpf5n&y+lb#;#)$zXo$E3!lj!n{p7MEfkr7`6yHS^9?j+S|98}NB9
zO8U{DS-6Z@0AO8BYHeJWDKOY?d@1MMpWAdC7^o|f2h&Bxp)oyq>l`3^pyav`2Q3iy
z^Xiu${#Qa_aOof{C*}Jd<?Dv)$@lE&utN<NA#z<WB=uWtM`^z#X}iR$o1C4${>wXP
zbWDli4Nsa2Wf}9ip{@oOiXZqYN5RLy$e0LvfsoK1zFl);IWvt4_Ss{b#|U%tu_u?O
zmG!Mw+saK8ozx$e@V^hFSQ5G=x03OX;ua!u)=ixi^}~~b-iIftGMeZ`lY4@~np;q9
zGwqRzhv@KGGV<Sq_#C=Vm?bq$H*OLYE#FN_Gb|qSUaanNrmsCeYvDZFnsz5Slv;}N
z^Z#1SH)zZlgtNcB|L=LWYDXK&FEmu;wy7@NbfpJM7$<qj=m-BOhMmu~FA2^83$z4l
zNyWU#vEW<%BDmk1COtIK7|z$8WgIKQs`JK|hrRkk3PPJMZs(b%tLJ<}Zd>Ys;N5Z;
zsfv3NkITo9j6i+kW+>f$cM>La+g;@+J(O}b3fxqASGnVi+)A0Xf(x-Z7HF()nt;5B
zSvN&t`BS-H7}Z^^F`c7mb?lPNk&%<;sUmgQRg>@dtl;dDvCny#iiXA7a7p@S70doP
zhAV6Zobqulb2-U#Rg>coDGNDnbXp6U2%y!H;Sm^SFgtYZDRDFKdoI-yjRBrAjdys4
zioe1vMGVizF~86FZE+{Ql0{vtK@XSCc7pQs=G9IrOBO$VEF7S-^UZnMqHID{oLl*E
zyR97sw^$2nYC5SXI2(@6_NW9@wD_Yut*E+fAa5~Z`{^~#X^MT%?5s26v+;+mHdfP&
zV~ySf_hNwZSsGkBna(CE{U|&*II^q!d#~nf{5N%abAaNlsoF5Ni-%{Y*C|f56ZJGp
zfBN6i2@!{L_O4{X<-cMM`13=>zs!pL7Bz<4n_!%Ppiulf2T&pc4l=F2&yNJURJ5~v
zLUfHZG>znLwN*_vR<$J`%QcyZl;wo#a^0cih8RCt+iGcj;sXD6wUFnp8orOQTQ-k1
zz)}|xEy{F{ACdHamQ^Q>#;_`u0yQSH_T!Eg6U=@W!7wKI6o}PAs0V{NEp)O=vdK9X
zQ{#Ogo;88{ns%c+*zyd;*YVhjjAFg%xt!lS60&snXkS2=qWkn>gh6t$EIv+&zwX#o
zN7z(0NetbRmyV;P@ca5&JR<Y#fO7eICc}Jtn^FM>*W&W>*RG~U)a0wPa1_tE-R0hb
z|8ZmfcHoOr&*`JxGym781I8*5r~fX_m2?Y6=HA^5;WePop#pXuc~YwZ^3<USzFMzE
z-=br0Z=LI3CVB*zC`ycB2<(cnH(_jZEesW!UqT73u=ye;mix5n^GD84_y9@}7sns3
zg_53}w?WY?|FPa;H)j7m?uTOe{Vwigj!S9?2k+Ln3J`+TAzhzgT@Z@&X9s$U^|{Yg
zp>thMUhV`)R9>%IU`>@vPIuU*1_qH}f-O!oypdM(>-1tLAUwjhAiv5mC2OwF^_?7#
zjhXDWrUG{GhcRGe7#xapZ%s1}RZPxUD=Nu;xC$n=78kdCf0dDLiwZ+O?hO_V3?$e5
zHOU1w;eu7>#hRKJi?Xly>Hn;Vp`(zJ{uKMt^I)zq+-%%m=tG<)vP(w)`kQ@w7hW$w
zP2rSPkd~!sp@PHSf<mNQk7D(xBop3E*-L^}+=087<5|&_rHgWyY-h@{%FW+Xs7Z}W
zo?&Ign<to`f69~R#IDu4PkH?XY0|+Ff}lG1!gY9pM6@O~1i!D>Yq2^sV<Yf*)rTBP
z+C+F!m8HN;#Uqq!{JwDJ))OD|YO<^)B0T6b8A~ldoqC4<Ry$wYt!=Ow|H-4kl@-QJ
zC4&kLY;Cb26a$g5EM{FWt(5v0bMJP8cY0>X<n3Q-f&qR_KGf)>C}6v)A=8o|_2g&9
zsan3elNOS(MzV^y?ovWqGJxr((W%%Tzbb)FGsi$VC55^1mEJ>rh~sXzH`Ia$(Sc90
z7$TXPnx0_)Q+i2s4!3B^^7PE6{DWq$LTS*<&&<NF{*fOp?fT}>>h<d=qTO|Vmfg-?
zHQ&uP4V)RaMCpq}6@Jbxac&LuJKpM1vJVLf0)bre)(f<%k-aQ@^M<9N!Z>p!r1X{P
zzWuS{@B3}X+uK{)%75!@AE6$a)yOjyqa*3A2_y%;4jIvg{AwiPu9CU{wPduopf$uH
zTa@jg$VzaF()?Wh@V*RHmk&=>Lhx)Nl9P25BKo2~IBS%JN=Wgylv2>vbJhR=2=xw~
zo?f+%28(}wnZL;ertt0PDQc}GhP13h!IG2oJa3HevJ3qDV9~VMs;X*hn<YgyFrlW~
zrrStu`<p{27h%VQvxW0nnFk_*n&=<hLuc6GsKCP!Qe)M0V-A;OmLh*kYq=E-%mMzR
zC>U~w3zF$3`mKh?u|D&~r1gDAY6+b3a6_igd}C#KD@~#v4lh`7$A9U#%ITEOq_Vg{
zeD*bJkg?VPw}3Sn8yVux-meP&hgw8p(|Mi1-&>IUMw<YrYu_}U0XC_q*$aiHzB>Jm
z7kj$y$OL&7D7RX9k8d)dVE|yG1#^^-QByRKpPsqO{SSY1Oba-E$Mt!8Bj3fv{;(=4
zJi@E<V(-U<0rT{8S!u+Me3CYtx?lJz3`o<f!I(bgNa_Nd?tfR3ioA{r7L`}_BuK`V
zs9%9-rdmkMULOEU#GZaV?Wmk2HqLzx4jhhQGVQ^tQTBwduD;6+%|5LcTgl~e`95+B
zP7xg}*rN=(OLzO-gN1T-m)!~c-P!B(|KWtX`vz7bvG~dl4%_jaEG(aptM4mgC4|L3
zmd718#E4L2z>(}A-YPE?Fo1N)OiXFXl#(<pz-IpluTS=hldw8B`o<b%q%izx@uyaB
z?Z{EbrNIq);CN%zPWFz4ut2T?65WM(+l?Dq>nUZCYg%daoDPs3H(t8fI<ratyM6PZ
z>3ADWbZOr5#8OECO*RDuRb&I}IdT5tbYU?L7joXfvdM$r{edF&d}zKX`&aF;jG$&&
zUB8`mUeFnKr8C&?k=3InwWWp(=Zex`+mhv>+1V3+X7mKf$OI&)<uiTH<7MiWx7Sq#
zd=oJftUxIy$V9=+PJxM9szPfI7&B1+lAw}5QlK9w)5NxDBY`52z^9fgCW$`#URQwB
zomTQ7W_7$tKoMeN8dwnQA`J%Bp>LB^YJ65JQzqzZjRT^0elWM^GPm`vTFqCHp3l~K
zR&;NPTxR{W>yy9<i)mmIV#>hRZydN@+3_j*BoPV(r_vqRYxEn*K9YC!`z@)avki*^
z)>l>(!?QC_2p0!OJ>)Cn^rD9}ol-D&b~xK_9NgTF`Z%e7+}?EO`2r;`MUfZbVf1^Q
z%2xxXrRlzVPTHHK4o{Ei>A9&!x!-~N%`NTv7p6o3g~Z~F8AP7J4ftwhIWtgZa(8}4
zZ!8I+@mgs)|BOI(4yz&UG6P-{mzND0v6Ql47f8-ZuvJr|4ONOYeF~UEL@Gqr4iCLc
z_d6U_4|2>bzt#Ji9{$*ZV41iMefM%iIV|Ui$i^M;;|h~BRswN(N^3s!38YFrngVG`
zSkCY&53bT<-)hI3G4Ny8y*tBCjt-Yg`B3Mb7sHnqxZ5eDt?7n10k)U+fEvb~)>S5n
z9wGpMex&m&v(<KO-|98i5qEKX`m!5alaXc1PDQW1kk&h90j1Y2%fP?;BBO10+sY$+
z`ksyGBcXUVIFnYSzv^cDe*p*;aSH0+T#|mUvXl;4K)9id7qUZ9R03ZvG>@T%jy{vm
zs%EFMLG8SvY*~R&1I+f2S3ic%H|;JDZ-;xc*pXl8S+OU{EyqhGqEJ=$9!rf|(p=+~
z6_EoY4Efn+XWKa38J?TDe6cx8j>4Ef3gD{rcnZ+2m(PULJU7ZAU7~iH(M7w<3z`hz
zH{Cp67?C2%`m#M5Q>z{7M{MzRVn}u{M-?q<wL}NaT;}-GSUPZE&yoVotIZD*pwA6n
zzyY7rq=H$~oNn-6bx`l^&86;z-hK9~Wt1Y?SWuZ?rel)}-l7CdT<eXIloi5c_~sbD
zH+JgU_{MYO`T6sEcad)U4abdHvrp}m2z3yHi&0{i1jfEZW9Vt|Lu83tdXAe&#-ugs
z5Hi4v(SvIavBfCKSWenA1kZk<5_zMb9ZcumKH}xfGn}=#d$M=%?`(H*Gvr)Ge>qJ0
z9XIiS$x=AIP}<c-xa?=nJpB0AaY>Jve||-7=~BsIxcAU(ttrygGf-(u6z^Q&SNWR0
zM>Y&#mA>U?c4)kQ;-79z1%$eUDzKs|M5W_*Rztnp*8jcAtA*5(Sz?L*%E!`+(pq{j
zy_6L0|6EBKc-hS0YzljC((-9~0SRh3C@2-Lmz)OrR+Q~22P$$pc=UKaR0&>+rx-<0
zXa89Mbvep#-E5wnpB3M)`<V@R{+m1PX~Ll{c5t1SO?(j-C)0~#6KyG5Ae1@MSBowc
zVAuf+GoLfqpuyTyKh^I_rBKkD#{CBPKo17>#Pk<i55qm)1@iC!I#+Myi7LS#)4~mB
zEFxCukbEW2MM7VT4yX+K`bp2u7ZBF;Ok-6f{n7+gpDo$G0?iz?AC5qZS`X)DlfaxJ
z%9s#tR<B-kw>1(5FusSQc;n9`k}2MsrH}4iH>Qs7W;v9gLDPl!7@&8$;_1|bB#hr`
z?xj&VDA9`dCXHEduhA+HJX*GhcVgq%Oq%JWp7H01J>DkaiW@Q~-&%hN4ndv$EBhm>
ze6ucu4s!Jt?qFdoj_mMc*u8Dn5pi-B(6(jolan<a+`Y|z74=^`ij}vTWHdiM<#UhX
zuT$t2DJU!`KQlh@?9ltt_wr)1vFeduA}#M@HK2Md*Ae}yQ;-GG3)Eq7Q+x%J<$smc
z?6nf$#)50`n#0DI3%_4UvBB$N?k(NBDmewttXHF=lce9ZhbMRW_`s->B&6dGn0Ue0
zK(Jm*2FVrnd@L5#oB_wXJm)s!t(xi|P`O6Ghr8QKO9{CrUvbrpKTW=u9*pQM{<&1g
zud}_lT&#MxD!eIaxwh(#%=P!t*cXZzgl||wUim!}Vr8x(B+E^#^h9RTM1MPL8j4^Y
z$WF|hyZ#bN=y7n3m}aZ?MN)Uw`WFKCcq()4&Ua(-^KDW21>f5ng0nC7CXbtr8GYF8
zd*id$P5tJs7fVEE(5HR+#xrGg5X-(I&D%#(-cD_qC}-X8kRym1!(DR-mZfmKOnKjG
z8>2v!wxvoM2=m940E5qnTS^CtYaV&>RJQys&_bXCh@kj;`Tk5z64I;J?AN-w04^#I
z2nTC60ri!?wM8Mh<j&{7Hzl?Be6<0HPINC*Y=sWSIw94S&(kCRlo=_S-PWB9rZT0X
z44M3~pjcwLqNu-59XUa2J(`ao8?E|>FL#gacncamalL!;__EHM^vqY?v2?um+c-T2
zI8#d9*9m_hRTGkVU?`CK^N+K4%5CCdtJ^jc>H)(GANNNRV|hRL`?jU|0fXUHqucJr
zy68sD8OI#LSj&Wt=7eZNh7R4D0KAL;r{{ETHbLEc{tgD~e&u+it)i8V&9DLfKH@dK
zu(T*e*NmsDQ@{VrFDeLDK~}+?mLds{W%FZl_=`V48s>=!PBXA+65+}`FRTZUnC?WB
z*nl}h*6Z~SIi7NTfPlF3)w7AHSmm*u^Ma@<Rzo+(35-^}u2PRes27Fgc+bs|``%S%
z=Vm?<j}D1FV?DUlzDScdUf$l9VtTdr!h)Nj(Nv1s7Z(&@$;FN-@^!Opta83QmMiCd
zkNECz7C#h$>asPwMGZvzOFK|g7or3pJB`G}{iC&>mJCIGpTceIc+bqwIL-GlFMwhe
zN<>N&JH?cYyzN(ju+)21xND}nn-T=)>EqP>8G0e(Qz#phBbwD+T}2dBV)X;qAuf0B
zz<FI9lScYK>zNas+(T_8Zb(xh%uMbG#Q<Y^#Q_U%OBBQGCbM)ygKZ~0P@f==<N3I&
zbp9adIu#(opk8*m%s2%McX`X<M1W`Mow*q?|G*q6(|lpjy=DvfSPg72UkL{@wP1N-
z=<DUBg@6NxREtcFI8k{0#PpoMFl4+X-<tsUQJJl&i8>xBu(O>Y(xh4l{avSkY9s9(
z=kII^^a&e--)XnE6fOIs)YF}g4kS6U32q7n$ON$cKQIN*Y+V(n{`1REOi1%&@Gg3y
zHF+VQ%V5qvW+%x8PyJSP`Rb20jyt)vdN`h4!Li1wT<?WQir0*i^b2eL`I$7>=^sTJ
zJZU<Gva${N-`Cx4Ax#<P=4MkQZ)L4a+~J`bx?boRKK~p+F2WVLC76EB#N%%kFm6-`
z^Z8XhO|E&&9&9}%OVA_!Ell#E&NgvbM3tGkNCzfhh$`y?Hx&%!;mwkDv=cUa#Jd$(
z>XzQE@K>Zk8=~&N?^H#gF#LLD98C!p=jIWUWJL$58thwRE3~h_pQB5pi;g=UX$s#C
zhw)Rj#Ff*Jo%oY0V(Xs;cd(b&X)mOoJ{GEK$XRDr(=HM)O-tP^EK(cvtGg~WCYdfi
z;AW=Y`{*Z(e!Csj!o_fv!{4KwEHh3u6vID*ort;JysFd1?K%MdkP@C`ug&y6l$_)-
zO@6etP$_go900*tfPWGibU@us1yH`gFCT5@{%>wj2|xRFe33$?I!|_|a$^N3S?NoE
zt)os4XR&`B2Mnse{fY~R&J=nst}UGKsh)@Vg+`FFzVT40p@#3P0^MM6j<CqsJA81N
z5HB6lDaYkjy2t}akz;*;zEp9BI-Jw0ZM!XJu4h(e7#HSH&O87Q_XlYgdx{3XDiJcq
z83%f25tr8%d@V>FA4Hn9%Ho2tYUa0|0B>N6AwcREN&yC(u}K;_lxQ_wO<L8MWe=PA
zE=4av!O((g{cS!OK07thnw=e&O^2pCD>D2t(EQG)CTQr(P~v}GDl@lv7HHO1j``+i
zXWm>MdT*DwG3+#MzH8X@4G(U`uUIj8gjbtXe!`D&ySrKbZmj&04BeNKeDL@wiCKG%
zme*0@FhmryRYHo;Jdm7f^%C?|d^3*<3T(G-V3D&y-M+o`C+!P0?;kiT!DTT_WoTdQ
z(9m%frZ*;7EcIU|eVR?U)gzaDCx&E8-8UX@N~8>#o3}Q7XL;;x4W9juZ5U|)^%(t{
z0IG`<*fC$I$7E*ny!$%hoSEr=T1q+PVXd~4gpt7tONe5%1}Rz-Q~Rc7+vAA|J5=!L
z@#WDNibXZz;<SeOMGrQ%JWBasNR7R}P{=#r9yfQCrq35{`*&@JwT4tZnjnaJoac1J
z@59jz=gEn<I5sVaCBZi^UsoiCepda<D=8{942HcZ)h7BK3k@CEr)Om2#4VtqZF-q-
zRFC&lKXG(LU@ZX&oQxrQK(BklCFZ0)&sra4CZNIXVMDMmqet|Y6>|r%3i-oJ_D?=K
zCQbOYxm1MD3mEJ4{0pVu0a(<*ir3nH3Q0%BR(uG&4lwvMl-x_@N5blzBBMEhs}Kl(
zI^xlb4&s<ID$Hv#9Q&m4oBfqg*)OyJFQYjo@0Wvp9&RX?L<;o&F1HdeUC?nioE#|7
zyxqeCk%yiB$O<6cy1yH!F=VX0Bf6GCl@%7Pd9v($eo|_8fg?BTiQ~?;dPpO!6?gpV
zqrG@gH|A#}`Q>T<K?evDv|DY&B#>tFD1`d#UI%Ls@~@XTjQ{>g95RLm!ZW}ePs@t<
zto4Qh{lcS>{STeSouQ3|t!UuH6t&r2n#v(iANa`8K51vE@q<_UCdt<QdYYoXaVg@b
zUiY51orBbG$=gAKlnzQs=gWKbm7#_sZ*$v3v5rtjuwoG#t$o>Y?*M-)zzE@RVN&a~
zrstojJdSA5a&7W!R~p~9W+;+|BRxoso9$d(kO1(rnymDDw4l6B#e}66RPfwqDE|-T
z9MXQVg7W!!oSpalOnWHGPACyNeK^$z^=zi0jwx!f*X;Z0nkvEN*W4u8U5QCe&E5_r
zr+}vi+k)x4Fx=h!zm7K@yfgx~Ez=@{EDgLy%45i9-Ce6U?Zh3T-7+;?|5CGLP01o+
zes_>Z1iliP?OOtphi#xR@MXdluSKJZrCrk<bdBDs8i#B0YkBag5&8-Y`O6=x(QH^;
z7srA`R7A6p%W?9)i`8v$Wf{9WLnKn{S5?Myrhog4c@dZQmZuN<`BW}Rm*->Emtn2S
z+tuHaTO_{!2Zw-Z3)`8aDOeIXv|x(;+q`3r%ip47IJ-&EEK~rhwNdwz4Hfu<k+bsf
zvNCR{x4Pr{@Oc<gg2JS`^9?<~>Rh~~)`cmXwFxx3J>Wh6`3*7q=C)(Xz<YhwjlN=P
zw1i=U0~*Zwqp^CD*y=@+(Z^fs7U4Oz+Y`ICwS-HI2Tj*IsvNz&7}R1V;4@c%l7wUe
znj1=3K<jm*^zb%uokDpxgm}cwp**K^XQ@eXH$KWy3b5L!i*;-M$|!o-{B&`x*hX$s
zKzSA6c27Ml#kgP|dDi}qan~BRxi%CR7<hafeXu#bI62>OSJlSvT@r|QBDOo={1cN2
zuoLQhNma&{-=qsLq%<B@+jY#&y-|$_xg`H{hfiyvw%Y@svDbp{4MrL+$*la?u)okP
z840nj4w3$^)46+)2|*#;)&EeZVOpX-m6}d4vRgpz27V)Q>J-{jcr^|A5nD2#MCODd
z(TT7k698c#)4=k90U!m)nnfR=MrHKHMrV&2?HU8NNziv|zoxzy1EVL;nR~<AnTLM4
zgB?(`lz+)Z@)peu8H^j@lZ`J&WYDaXj_>@U!5|vO>$eKxyT!}7ELI_aqlD%g-17<-
zM%`MFNoZI{^uwfL27Qd#pfOLkSzONj?BASUjMX25_0mTxKY!wVO+huwUz`^S1v9%h
zZ|<mGEKW*|%WZq999cm+ZEr)m1iiQBSSL}i3K@N8b@YfO)(oQX1_sAll0OmCpAwkI
znfoVaYm22LB<Py~|AUdH_Dv1_yOf+QIb(}sy5i0Xv+oOA8)zfZ{mYM1i*-ak=Xq%l
z76`qg1H_O?tL}z|x2KL9ZOFDDtbF-$&;6lD==JW_Rys>^j80ZjvT1X7JwC~6>JcVh
z*e0jdGko?4-i>t5!7_g?Xe*sq-P+9k*arDAF&p6Oy>jvy|4PEvwu;yg9@>1-&r)!?
zh&nqhxI8&Iozb&_GE#9rQ2MRPn_}oNv{UohSkYo<HOuZg6&Gp_zdSVLZ&Ka`dj!`a
zpkY1<k6~Ic%6l-t7m;Z1?vfk>fep;Eo3?5Aq_r5xg7w?fn<TCbcTfXSMH$8x-Tu=@
zrCC|z*!%v{6}O5&xu%8hz=DdU)wolIlv&^n6(Fyrdt@#<8`uy|mh1xsETyX75rb1}
zV!&?o5RR6tlj~FxsLty9KqT2q<<=2;Oz5#w(H{#7?l?OluJl(ASGox?0SJW5xJrki
z@mYAxh47#@C_|PMIB`vDOV&{){NJf*B7Us2>6q-;+w`;!q|sDj#sulR4UKL7>tgkN
z+I(|Mr!Ae<5bxX_FY&sjBw;C<0vcn1;Z&H7g)F$RBPdY1<pyC(9dPdQNJUolm;Ip#
zB;3F<lc!K=oC*C7(;ZfE)%FqNByM?O@2t-DO^x26V13^3zMDvK9SE$yKNI(ysL|j<
zuMU^&nVagc$>+1Z#gcV+;y&xptl8ZjQua{jTbLd>oXuwv@Y?})l+UxHIH`V6Y=8Fy
zq>fh_+)&#UjaxR*Xps{XNYXGM`ySGS%Z2w3$3k526_(1#%e36r6aZGGeW@L;j@Ulh
z@i||Jvon<2N1asjseV>1$TVgc*V$(ZZw`U&2IuQmg8n$d#^Kz&F)bx>bZE?yvOca3
zpO{MtjkcX^Z@6os&>Fj3@kX?8W7gDBbXjH`wLd#M6UV$Mu7BqNk`Syio0pYqb>-Jo
zLCL$gfoC6cTdlA;?l4!uUWruDk<f6K)X@5OF_ZGXWNm0yFK<BpjfF031sJ&z`cB|T
zuFr_-G17q{x@fVriZ(K7CD$sDuYH>|cYL+S(C@DJzOug8(cb6tv;8L2#lHD8doci@
zV8DH(n(6dB>TC~p+i*Gn{8&VgNu7P6X9mtalEyu<@Z*J$h{%PxC!Kwcs-wNa8^ZK{
zotQLRO40Cb@=8N)#dP){ppDn9Ph856<@Vju%Lh$`i;eEEn@H>Te}#=saz)yEd9BY<
z)PEwgDcXJh?OWt%vHgI9T+)8M(+mol6FfQMkIfd$&|qk0`E)0iO1p9;_tSb%L@@2A
zF&VSJXXYOdM=a$W(=0uHDux8g2^nZek6a$pG5bQ*ylBzpI@5uE-qXxU#hK*>yj2t&
zRAR6wZZvGh=0yVVLMc0WqU|suXQdopxeyTKQQHL7%9rNtpI7>)t@PAsj2o2^^BP3$
zob5e7cS-C#&(&`EY>hf!-p7qiJr7XoXbQdoQoSG*pprYQf2>6Fumo=LQ&S=aAF8q%
zLMO9`W2Ro--}j|v=jUyziSO=Wl@O^<QlEBx*b_<V1%bMKD0^(|H*1evV|BH{eMZ6J
z-SS{G=GDEB!hSvR8GX99Y^XvoReI4&NdTP<f$8}T24Jr2C8=ZG^?Tf?1*Z$^khGHV
z*xwR7Hn+(m^vRmD&G2#Hn6mEA<*=O0>~tNidNEvM)6?~MYzu7gu+Gl*sI(u~G|VzZ
zdo$-6Hl{rJ&c_#ARCElnMn7sr{ZupVZ<So55RRWL^|5o`Mci`_M{R%zbFp@EU_!7(
zCMWQx9#2@H$O;IQQ{7SqgIda{x+b>>&5Cfi8Vu>Og)CIUo-`f~7ZfbdqolarbzTz5
zK@4-Z#*8hNTDRo!6w$7t-FWII5DecgoAhT9nL>7+r?GC4IvlQ5hj_FHi~}K>&goR;
z@16@tfGk}s3{WO|<UkT^F+q!Gl-*7kSTlUO=2wuv;_%$gIQ;<|hF#k|qQy4RQp=*$
zgdL5D2HnMC5{<-Pfj5Y)o}{`w@;V|Sdu{7`{>y~Mk!N-|OJpzpINYxKk%_KzUG|pM
zhSlX}Q$OnL;H3ZZY>((YeA_V8Tu=RVp#-}S%f>Z*lqa=flt!iq`h{*f8*b8WsOVJ4
zCC^=<z|aH&%uF|XZ902;aA{D*3w3wPnB|w6{-Ok0r=jnTZ^6x|@7IY&(>Xk@LK|m4
z<?Z^T^&md+%2bm|hw{hv*^hE$9N|`JHx1-#`^u89Lz=Pi!xX%!d5L1&r<xeMs@qe~
zo7+~FQOa5C_oXC1jr8PE$indCQpKx~`>r)gf|u96P}0j#V!-ea4+8T-lqQBTzV)KY
z@5cM?JyCR^&aR9AB!ti2KoFk<uJoSf4W%1W-}c$CSd`FepD<CqCW?F<En3=^+q69&
ze)eyHI2q?{$H&{D;_2b{QkB}IS_;4T<g_3{DcrkgnBhK<S8uN@dWV(!)iG$>3Kbl7
zV!eF(a_?F8ieZ6@*JNdb0k|r;dVWwoBjR~d>H}Ya=77$pSq5TM;uLqY!o;YcsD{nS
z&DN&&!}H8}Jnre%zSLxS>xslw!ij$-(b805r#?@}LVaPsx-;Tpb79$S+@4!Fj&FS6
z%2RX-UQt16{_uD)1tUS(depdGVd?&cz#0E4NM+oY`4^#e(=!ozjZ4@sx7rH$ZgRKa
zyC4mg^rJmDF4Vkg_|*y%I->9uiV_Y@1<|JpK#Qa(3Tc>_u9#PEmKJnK#O21W&0ie3
z<&>7Yl(~^Fi{mn9+sr`kIRG);6gQGqt4C)yHZ$%xA2>Z>?37RB7x{+0?eLg8Q?MEI
z<K5RZpFfHPUHWlCl%n9+a!=BZgs}CEx7%+J5tFzcZn95V(B7=!cxwUG9i>h2Bh=*6
zXZt9NW(saKD4|MG^w9#t+&w<eY47h%`@WQa4_iH_C#<|Zo#f!gz~lMjyobaebCx18
zAo6G$V$(ocb|x)st6u(F^j1AzBsR8inHHH84GKg3#3^c6_FOBT)vOAXu0#XXiN|3R
z$%8%G<oeUYRCdyo))>miNmM%M9<>k*-%69y=k-Z4#3OY2c&u)x?)}`Z@WD>&rLX=R
z6}pH@k>ah=%c9)1hQk%yVTgt5Ij)TqD7XLF)isbtmLttL64b3LZoQ?AOgL!YBnC`3
z5j?Jo^agik-PfUpP^3w`P+}X}&#Sb3UO|Wi^7ilhn28tjHlmyY)s=5Eg83zzZ@A6P
zD!fPnwK22kcC$0SE@%DTj4Ug96<4v}+TTi=tNX#%`g)R;p?duy=$sD|WdywuOj?+Y
zGvp(0ewV|jOOkhISjXq5cPOdoFTi>Wm+4QIl(K4rT)kNmzIbG7b3aoD^LLFOH{6W~
z2X~XtcSkshz_aHYgC>q$7kjB&-Bu8W1F1n%Unn47G`uX;B^_mn-=?xc4aw-6RPxQW
z6HB?cbhL=hohqIzc6PT6NfA7pQMgs7!7XZA;o2+cdoB(L^_K$5qrX@=qgw(!-P^9E
z3C<LKuxXkIec|IkZd9KrY~aKp95@(`C59F7Mcwk&_4X|cbb~RqN-$aq7=k}PhG3j-
zAUV+Xg3@*Jwq3sT8|ijIakPA&_B`2=;zMjt8G$^TKhDBACncjB3|!{=S6zO`P(7Z@
zx+y_*4YgcX6)MdYozYh3qVkt3OJuhbHGzARYM$<dOMOOB?DRkjDIG{$<(1TnOXh`G
zgc%DyjCL0>6Z~n2Q%B<Jeg@Mz@el|gsEf9F^mn-@gOTEoM<=q4kZ^}GTf4^oeq8nC
z*%9tG%Y27TqkfMzm@(4Gw9mpR-=i;TyQgUcr7v!qg$nYqm+G9Oh>DLE(fmpl6(qE=
zAZ6#{?2c;h&^U2?JAJ!e`4Z>sh~IDy<ndy#_(~D0#IuYZm2>6gNT#Ntr^t>{AF^(1
zVWvW6j_9|=W7vbOy#Mz@2<EikSI#nqz0DP3eI1_w(wn=J1oGRy)0%9bASQxQd1HWb
z1G4};U)^bdamhXjGZ0J=mH?1v63b}KWbjLk+tm5LZuyW|#tTZ$_(HaIBRNX;!YDkP
z8qD1u++<E?YxR_>?Nv$Xt%XbwdbJwBs(Ot;O^fqO#?EI;JHv8x&UfR;@BhTUePe<p
zhPu6C#IH4EL!om;<YH??pIZlp>Yw}_cFUh~)s;xmrEb;P3+^*szPQ-xi3{B7sI&_5
z0oWk0=s0@=Uv9)$xtl8}bTZ3>v=`)TDq)fE#aQ>M*=w=U?zL?srf0V&gIq28ccGei
ztkJyfh3?YqVv0v1b;^GeYLMM}7#W!_G#Dq2(CMfF_)qc@9?Z~Z$%vAskpP;f(~yg?
zfL!U=pH`crtSsW<WLDhp?co#;tGJY%1@N^p);=ic0hD?GnB{><&H!q06~c9lw2Vgk
zjV!+u>%fdKmi_hsGe$HMz)-kQ!*a97BLId0Y2oeJUrD}Qd}cq)BJ?DpKSbd86DTIQ
zd|BBn1?nzt<H_5DTHLD>ppG&$aW3h4tl}UHc1$X&fFxro99Jd$kk{N#v;gfvwen|N
ziZTFUxz~BGK_O^vIz2cZq^F0eP}G{7vm~zb-WO9mO>k{rFhWGBR*yab$(r<C3qAVU
zKeRJ$*Ryou!8lBplzl(i98)gAK^x`j*sQ?TFB$Uok$5%=NCyp@D7-20w_S?<I)E3~
zRZk0EJvyj&e%)!?#wh^^fM_&84$gVeXecJZU;X8e|D)+F<C=WDHoh@pAT>Zl=~3bj
zaCG<Rlu!hTNhly)k~iRhp$LM=Xha31bBvHikX9*Wq)5mFq+`Uh|BL6v_F|vCxVO(8
z=bY<0-%EP}EZS<U0M)wKLmS^j>NK&#VCo9I^OHHY_e8oCpcYO%;rCMVEKj7f%kM7j
zh0?lQd8`eLwK4HA$sg%&+jYw`gC6v}W_myRg(KdZEa4pp1wee}b4}$RWufC#KjBnQ
zexR9`hKE(QU2&39x^{CDvbE-Mie5?0SpvwSiBZoRi?qy8F`!=GTCn~-q%d6zsqa#K
zTTi2pUfjUDk&PCl&nUlGy8}w<MAa}f{q_49$y-QWT1gQSD^RfI+s-r7)0+`vd=MB;
z+F1q?_WW-|p6!(Y?JAB9|0%TR4~H5l@$62s{!-5Wbs2X((!b3D@0b29`u$gdcK;g_
zljkdcil+QSLP$H~B9v#F*o7Z=Kq9bMh<}k=Mt5(qAJU+ZvFXYA-t?pA*PWlUx-s@O
zZkLQ-|Buy@E(glz&{GIY)hTwjm@HggLQP8o%;41cWVWZpN~$8Si`Ybi95M=LY{!^U
zNV4ER;L=*@W#_5b^B>OK^}02BM?cx@3SFaDPxLL<d{egBkjq^H`2hxmAMZZl?u?GU
zE_QUBlySmBNtVAQ*}A{Y%EsOYeQ61@wmyjFll*}7Bx-|8N2TkKu?G+e=0u84kQUHp
zN6*AyJmy#Fn(slJX`u`E|BUGH!zeyMKgA8gL7FIs;y<f0S?=4QrNSj}@<fj6iN}=A
z%}(z9t&z=<5%>HOxK08&*O(|<U<l*?75IdTFLiiN=Jwtv<2Z5^UAx(uH8(qdRTMT7
zpY^QfL;2vyTMY%=)L~=*H)U7%7SO+5nr@NY8tbvWAYXHqe{btrR#V3J@GFxOyJ5W?
z&pfVbG5?8tssO@3jY{_@1jnToX7i~uNsSKq1-u1^jgS`t%I1~-l&R`7{FO$AEJ|*<
zZMYdfDI<`nhnTo##Jx?wQ1^v)2Z`JEG|=;ohPL~!1@UP`rJSIEqkotO`oNUbxT@6c
zy-VP$6*h3rNLX50Zef@{T4E2h_i57SjetqG5x)?5SVe95F@G;l#>G~Ki-U(lkn)R5
zv_TrCj(Ys-;pfO_PCrgZYj-Z80U7C(v?Fui7M+xT-`a`X-Ejjd6W~cZN0?T)9oDXT
z^Utww#3@M7<14O2P)H>8FQD%+HAkU=*;C#929EfBNEHo%!F)M#9VCc#52~=2hXA=(
z`U@y;@#IESmNyo+W!>WUxW_)Tc}P@A`FfXy78wxYvrOXs`L0JrdLxrm%ur-(%mq{G
ze59X91?u637rSFE?v4lHn&f*VU%9`yK>I#q2k`}Rz@}FrRmHwQx$pf=nWLe1p3<#G
z_^uH?P@lZ|Y$B+}i>A}+aC|(WA2?Y9{3r3-)ng0!q%w|YMzVt^tx-bphw<{LRe($g
z;%;LIQUp;Xj}FLvz4F^jp;ndJ+>W@89sfhX2L9L%NOJ^EBBG-7aBzxe+Mv!OQiH^c
za%xdc1q@1T9T1__T-IkIi5l(xW}e^pZbyH=B#9>>`;doG(wJ(v_wSY%eeJ=az(eEj
zjZbBMn*Oq{oSL%LjuqEaxZ>1sH^%p-6j(8lxrmPrqxK_1^ktG+3iE(d)BMazLj-B}
zbayi%s(@IrD5#&LVeAc?n&dP$=krp%`c@-oo0C6{LIap<$SDg&-?yoGeY5F)0}gTm
z(hNNEo}U!s17&_>D#W{kSb7l&Lr^30S;m;<NMjDk=;<SJ>fYmTriv!_!=!sOrfbJx
zg&<9ha=)+wri7uL9p8wEhXcJ&yv5FJl0{u*J&T1N9X&Rg#xX_Evh1n8bN85R5YR{!
z)VCZh0|ToDXz*;ylx#ZR$^&795IR?ymEi~kD^52WEsKXeO))2v!F9y;#MhEALL&6c
ztAcd~5R8;GOXxH&)ipSy5jQgg$(|u9%?Ej|ze|Cf!Qpjt9MxOoj*=gP=EM)?!|w^x
zK{v~I?)TC!IF6FR0Ov!zW7xgf?9V^mp(wFHJVF?S;yJDNW$qIdVOaOmQ-kR&`u4#n
zq(xTKTy9zTgFELd?QEQgjTL0D&p11yFZN})6P)6oUs{RPVJP;<vL1s=(R~Wu_%`L8
z1V!;A?`waOObV}ZV`ss*|BeM^>ky|#C~E)C3l`<pu?5R|z62-GUhA9;yzdq=qzGWs
zyW2dK?QaSE>ghps{3kRygFTV6b%4DR6iD-Y_zueJO%LM0G!%vejNgf%<b$oDH*WL<
z=AhG_a1}sZsBu*{VDEW6=cegO6!RCR@I!Tt;PqgR{t#Wv&ib5QwYIGG7ex~$(pX1C
zSXjiwltaE_K}TC7?tY4R0aK$U4KUlf@=W}+VCLeoh$vEG1KEuxc&8jY4@XCL8OOQH
z3!x;X8Lf{Flz=0j);WnHdXFiOz}J#1=CZQvqvH}K!x)7L;lJxV@ok?~v{!#T^5BL)
zK2v`dkz6B5aXhaQ>s$ZvYNilh#eaQFIo;|H?mc}3BsC%{gDO3zO))&w9O9n#suc%v
z?|gm#1_&s#vd-&eQt}}Up4)2Z+dJoBr?*Rd$n-Hsi6NXvNPIYUQ3=8U=E0sQLBs%o
zF1H3AIU7mTyxf(CT6#`N-<}tJ<Km~&sBdy0k_el}PtVU@s3|7?CZh+86}b;v?d9Wa
zOXhy_Am`{?W|&H^!hiM!#4LFl(K5EG8kO45Q3%AR-B9hEz@ZeRSC(o+&`_ic9HZ~7
zvV1Xpy%-dqUp^U^X<$>Fi)!i}kpvpqW>FK&D4IZMeezOe_f;;hehv;kLBlFo8?Nhm
z9Ph32A2r6&>jDOSxy&&)vD}5j)2HVI0$$zGvA~QtYwtYnsli7nKyTXDXDZK>Al7hO
zH1qCDH_4f3=Ud7Z2xL0i$iQ--nN83FBJa83NyLYiR+jyA6171FzE~eb0#G!cu{KK#
z+z5{o<!8P_;}#sRkM6!{Od(VD*KhsbPNFs#bMqzhYt5eldz@9(3!|bd=NG<DXBikR
ze`&J(QtS@noo6!5HJq9<wj)+z0yj-^Y&O~?h`E-@CV6Km8k(DPU47S9mbUmn&&pYr
z+;rj2qcl-#mtAu|La+S?27cANc6PEOc}?y;x&0_yadHUPSmj3TwkRj{c01I!NoZW_
zqaq@`G~)DZrzfE#cO^`%gB&#0me+7g!R!6k%GM1+73))V^yMu4qV%PjFQuQC$8mT#
zUE_~9<RA0n@p1mzk2BVbDhu92N()tOjk!`!$)K9dYO$WT6Sj{sQh)-r$m3&8o6Iou
zkW+1Vw6kZsSmT6^-Ne>oiFKKMwTovXL;7g=d(^y#rQ^fB$`g(}FBJh8{Km8GEynpt
zFNf7ByQXt}o=UIH8N2NbYBc;h4U$>@?tubPQW7FESnR+#HNW6ky>J<EdALT1$ZAuh
zV&JPyhzY?YJIT0B0kD>xJ-N?0dV!Ng2W8%0JgVC8xa206MPyu}F-d=_QC#e&UZE}+
zwPpSiYV$>m7920&WEd*18Bp`hT>4-5=0Gnc<;O%(=}k0iitb1LBzE;wQS@@>JMyj6
zet%zW`>4bbXpK5OGx#u@;lpf-F*7quw2<b0YYTHLguNR96pXEcTStaeZV$CQ7)56e
zZIPi(`gM0B0d;<~Gtx*vd(zhRU2iUCgn{3m<%fzC(R;?vFy!u~qaRAtH51`2pJzA_
zu4v54*ErQ6>s47=g)5tZj#&f*9r<c~b-<pKdH!4x>xFVLw#h>D-uX3ALfKi0ANXel
zk9#M;HC%uhax%{#5>K``=U$xct*o2`wpA2^W6+BN!Ec;?bZ>K>XDDQ_Y{m_+9KGE#
z+FMfi`InUv+&L_?R%$Hj8tt3n-CnWTC^qeS7pLLEqb=l{Qr!4i+~6U5I;HVkVf98a
zBI>Q>L*{?>ze-(DJjFUe&%D}M8Vd7A8s77h$?$;Ke}7-JcQ~Xr{blUn*kX)*RnK<?
z^t2j>ziVh9%3w+hPnt`4&2hXEyn~st^%0lbz6I)x!Wa1SINtcUnoH`V#odv}NF#eF
z&<XV#)~fW}5J*z&S9ck61js5uE-4&Om|ZYEIB9El40L#cyo|oyQm=1tR11c|sZ(`j
zqydp1yoVj<heKAoQez7pQR*E98=u}S%{^{t=Jw5DV=9F|N&C@Z>OkuhV)P#aX<t~F
z8u?bVWI5{8H?u+B=W3W2o^TQv+Uj5#HMbd@;7+r(q?Y48O+Hk^-bzp6V0Xf~+AM|{
z6MYlZ^%mDO8fO3Z7^KV&zs5zTo(iI_1LxAbkPNS3OuToon<fV)GcD=9;dWcj%k$cU
z`BsWpK~qKA@xHw4Gj2a776fUt^mNAR_@(#+jllXt6xBXM<QyU3@_Z%u{>=AxA&5ty
z8y}4=?@?_hNpPvK{Ri4QYE2N(oSoC2oSwPoHS(!EfhP5}CSO)4-aKLTt=vCZ1t5wY
zmaOX}Z$EVvw!Q?mWMN^4nb&?9$k9d5VVM}gWl&cM9Vd~e`niZ%ldEtLIefvUhLh^J
zI8$KU#`}PTKPq8bC{R<0JUb)vzXaaa|HES+;UwX@8WnBR4mpj`A1mDO#;J@@$%9_n
zV3J@KjOxyaajOT@I6?=`QSAY6^$Ww69<$dPxN#$x;|AmlcO3)c+!vzW->>!*`Wy6U
zYp*VM-mC|Uurh{!BABeAU^l$CDud{&akgF@cRb0tX>DNV@`J!IY;H*FoHstlFiL?Z
zxqlf29Pf?Y#tsS53P@=+%#8U=uT_(gLZ4<wU&t=-jfG&kZc5=jt^>ON4NWGoNZz4-
zP_BEFI=R7<W5_6Hl$dS3sYu5cjK}?{n+mEdj>kwfmUMl3hWSP%1*O}rDl|xyXW*u$
z%(S)>?y~<i%*Dewo%70<bAzRxFFxMg#uLnVQ=1_P>ZZ0ZRUBn^uroB2kU~n~4JlCL
zISl8=B%JriTxI7e(*r^hUZuw*Lj1f00$;=X=sQ(AAsIp3ubU+wTz3{{kKvc2dfX;H
zM2G1$ZbLn3ycO+d!y`EJwTZKfSyIJd)oO#}-%?S%O+4_xWtTs@o~gU-Eupww!tDc{
z;zJ4w@-f|JA~EsiAC7MYw~?{5`3t6A4jM1-IE^nb?7e^--vtI|#qWi<4g&|L29KWn
z2zH?4*h%$!bSgC@ffc0<ojz!MFp51IA`Fp+2q!0juBNSXTYrubZ-(`~@7j4i+j==M
zX*b?q((i6)pz!iM{U?Q(vFxhOZFyBS8#VpHC9&NYtf1iZ&N<}U;TDP=?2Eno9|%Oj
zc<PLlF0Tx+3?a*D8~Fhga5ha{I{Jk9*5n*9t+bm(G2+FjIjMj+#5%I&2Z3;~4FDI*
zDOj@T(rM0o#quE^CG=U0<N8RLK;Sh&Lkt?q?)=}2@G@S$cX|y~c$_@V739MP4IJoR
z$k9KLEC)}1s+0cvBbuBS^&$3e-3Fj&MUga(kQ6UJz+1dZlDbC&q-P)_G~;?J3GtoA
z>$UsWSNE8-rXDl1#5(&#wTj&yE5UAWcbxBiQvQ&eMY++!$j@BM2Y->0b+FjmTwaE^
z?a8CCxMlozHR%47_mFKR-WP&5@u<nmkbkd8fk?$5QWMHOk^;;;fxH%}*;l%^uLyKo
z*;U#H0rsm?w|0f)oacKvmuUql+wgbiDpDGhEGhN7I~y7stKNiNGhkwZi|RRf=J}T5
z4(I!8>l3Q+hvx)h0T2-$8D+ONInl$nB{>`eR*_?hPjzrdu`g+2zGw|6#&mWCUJ2)_
zWB8HI#r~FYu4m4Zyo61h3A+b%a=Si2uhA6^a41<+_mzS_<=|lz12TO^h+T@8ySgQF
z-tTUJ6n;590a0d#RZq!+-YAGreBk?_VgAC^+lc&hnwxhB!8-C*_OFd4oagIEQ7L5#
z`lS^9y<v$*1l;rWndy(Bqvycz*td#1*crh&e|Bo4@24U8qO&T4xSa*(Z#pAPi+aTX
zF>08%Q~QqAtN6!K(Vhu}htqcDS3fD+`cgLKgf~4h5PB2ZelXQ7E`%~*{Lx^L(O$Ji
zUf04;j}r<0Z84!J2~ma-JEEM{7ks{AC-=XLapL(S`BPNwu_TQ~_s;r2)$emFmOFXX
zWJ(VnJ6%2I1GMKCnBn(kG2>PlEfoMSZ;GX;%IY2a*6ijg>uS8K!xWQQidjm6n5BVe
z(Zt*|<J=R6&A`%xNJUe<VZp!aUN`gM8YBHko-LUTt2BN1JF>4PJYeC8Jz@WQMR~R|
z->QYhfFKvT3mxrLhzmaA_?{}bETq}3aCnVrb85<IW6hGpdwICBaIw&lHxGDI?jHYp
z6=29g3;EW`(5?3*vhwPKR%VtksAN(F^AYpBFFhLwwAx-_eJ7WdKhE~YV?!MfizP3J
zJnK!RRpkPwx_{ZR-IT6|q6I}?+g7v@rmCB(*;1l8JP!HN;c;1CDqCAg3f?qS^s@6D
z0+dghA}B(<q7>)yJ34^NQ{a@)5phdsP3mDHS86Z`0<ROqfj-O!y*(PKz;CS5E38n&
zwBcq>KTQ?K8`roq#l*?S=iY+R!6-2Q$WF<JZl1k$6`uCfGZ#vsWOo4giby*Dc3IG7
z+tGY#U*hxOCsysnwszzHB!BbPpSL2DSqyafi}`v9vj;q5&~+X>Cm}*D5B%xqIi}1l
z632rNAnlUH)5{d~c}Fv>j2`46aLZKGHEHuNgBFVh%DjARZ_~fMts^hodeZUYz)>Au
zm>}=j%ywhtEei`(Do##`ej{HF77Kc7+cYCJR-Az>wn)=|%5_K*75S9YGMw8ND(PhP
z5|Y&k7k$z4+RKFxNtwh`xWUsd_rSAa(gen(5DTI&6j1Urh)Z-VSjbobsF%jA1YP-_
zkblvXYgA-Z`vVCt`=1Z7Ls5eX!=|U4|G(0G-Zi%x)b^Ol_euh)^;A!QE-JYC^74^r
zulV~2!6+#+QB~vdLSswe^&2o8jzBJm@visnCc)_wRf~Qqe?wfm`A~_S>r*4AwyF}u
z8Hs7)xZmZ;-ndgjiJpPzXNAH3Py2f~d!|O#Tf<p(rNN$C-)uwhg_TI*(X<i^LzqD{
zV!<0%r7es`y515L%NBcpv=Hr;(U<=c$UF8HoSBewMVrM?lp!c3G;`cy^B^b`P>wv`
z+vmlu>qFGROxT+S5H;GM$dilH{)P4rSRVt(RbcPrlSKWY1X{oXGTdW4MTxut)7OHF
zbqgYmKUCURUn9;n*lnXZ6h)By4fE54O}VkTziSl&>bS2kUd9B}Iz%IYI8%L1N0?>4
zB#%j^`4)x`1Y1d0XzZf>?<P4ia7|Q{3hBA&-v;{SJ+o40)VyCTP}FB{O?X_*@8lB+
zG*7HmmyofQ!1+bXFheZphx0K}_t|SqD7yZA8AXd?{qHnIAr6kGvFT05r6cwNn3(Gz
zz4iq8koCKycWs{A0s+sccXQG1tjml?r@~5Hm*~yJuU&=#B2emUed#W@Tz{WZ#2$Kk
zgEKRuDes4s+xNtCdDaLZ;4%}SygPJ2BHNOwHHk1}y_rI=ZSB8^4)xE_2;pI=D~zS-
zOn2gHwo27U8}xGX<p!97eoV}HZm*6a=d2!Nezt}nR&xDE6-5l}36DA96iuGwj?8eJ
zanSZ&0F4W~gS~fsM}7vevX$aGf<>g;SBKdY>EAkXdyHJ8sN-8GkfNe*RE)bB*E}Cc
zOgB%R1Pfr8m=bzu=C7!+&!}^eueaW(t;25ApdK$sO{r0oMc`h&(&>+iplFQ3Y;36E
zAH<u4Qe%TU=0&UG^9JQ~|17MD5kv?`(bI5wTNas5SgFfr^6Jl|cqUuXTuQUAso~ly
zXpos%a-MD>MW&H!7!v3qcPzAiDn$q&4eMk!?Y!Bd`Ke!<3wMUYMw*RR-R!O3-&L{y
zJ64VFINv-OQ+IIeXgfdrvv+||-3+14>-_eOUhi2PG?O__z|V84((?+N?kWfHnpaC+
zT7)lf!M06Qa^vH#F38-&y{7q8MD%-M-bGrM8v?8g6G3!a^DVZ~V3yaRfgx8CDyb=x
ze3|$M?2@iY!*nT2$leEEjr}`4to+`~O;ApsEAlq0rx*UGQbo(l5|4DBr&^*hctT|e
zl?Gk|39{O(#3AwyTZg+9_SJym=h{v2l(49ybovi4p;XEC^)>;8cv^l|+wGZF_k{5J
zgWD}E0rKPXs#<pPlL%hWX*vrd?6oiZXt$7o=>Xq^YO~$L`rliMCFd82H}7Ufw8i-l
zYwmagjnlThtlTss#cNQsTtPK)2!Eu0adEJ6;?-Q~96DR%Q=3{N#|bdnP`*4;Na>w6
z?vCg17UWAr8c=qEIbJpHgM<33dZt<37O9z+KPcF`yuW$Gk<qTq?e004g7}01Nv-j>
zL9UrD2XXhD=qy%&G>jSF@}OEKireBsf2l+?qkHaa8*8G!a5PXUSKuN^hf!HrqQ&Kj
zV?or(!M`2LC58m!UI63atya;j4JrWX8k#XvFnu%K{m^FpgXKFFhD+DR3G^_lcrMD(
z5RA5o{}`6(Gtw#&oz6;|QrIFu(>L>L^HYA6*L`t%n#34doIJx<@73VuS<{m7&9I<*
zT2|u^y?+WBbchF3mJH#lj2<Qh<#mRzg$LhE$1t|~+u7Zd#seC&5ib$5F9G`$HNM$Y
z-v~<s3xm*m#gPLf3r5RZiLVi^ic!ZCVi{HOY=WB8LqYCyEqyEgUbv~C^W&W!Z6(uI
zQJUbrAumQE4j1~8xH;8OPEx1YVhR^*s4_ZE{MSS<q2!t<m6V}>5$&-GjZLFfG#G6D
z8w{%*uD9hy0>8TKUVuzPe`_$RmnSf(C4OEt4$^pKWug0ypd$Ukcd&F~s@s^;Vf64U
z@`S`&0$d(dSC5sLKYACLAsT~bRgenQ2$@+`$E2aVO^9`O<#|$y5H8)h7M#)-JaZ9G
zUWn4l3vo`dEGX@UJnOyu^_ZLS`O5(s#7$&Qj_hp3ZD2?)?ylt#hs`g)8`1i-0Vo5s
z-HEBF5dC9MYo^ZD+>A`Yw|_FD82*4jnJ0fX>DLR?)HfM*qzrV~k?;0SB7gYneCNPQ
z4Oj`sK8;ZLElB^NWodVNe;;=748bwdi`L%%H50CpR3XgxioY(1w@@SPyP4Q>!Kdy|
z_{K)t1aiASJW}$Hn*s#A!!98vx|U`!5W4t&dR~PK{*qHjbTe4lV1V5`VO~78&Lvgl
z`=Ck#n}iUPUW(m=>66{4^SzYn%N=z<y)Dmr+pPr?{(HymgCCSZL_<>$#y_f9B-Z(!
zKE$AxZWiv!qxe+7dIa3vV{O0yf<DY0k5h!bv;{8I#G3Vjz1dYMd;OBR_6(z64vb<%
z{T{|+<~bFG3Ng$uQws)LFZ)RQ40987Xtuvw8jQ$*&U~7=$;^zVf=&Kf85^Ho!!LBS
zpQ-Nw@@MLY{Q{3x(Xp8}Y&9X)d*Q3%%HRI6LinD$7*MXeRTg&%u(B8P3qOz6c$d4k
z1RTFIlCunWc>`fcUq5d(XDutDiPVy0mhBiO?vYtqN2#`%mbVO`?<$EcjeD-ma~VfR
zdk4V~Ula@fU*RY5W%g#%gp;tgS+`9S=o5S7y$$I<uZgyY{i0@;683F}t<`?HWT63@
zQ1Ip+IW$zNCw;GHX{kw_t9UNm)lg~$NdGPzLrWP{-3FFS<Vy`-w9xaL8CebbyeEWd
z`rW_$K2*%JZKOOq^y$5)%J%*5&M9jNn>TAlC$mjOxkFZ#{R&Eq<veqy52jROe}Dp>
zYB|L=S9+^KId~Y2wfoyf3Zd5(uyu_s6G}-;J>+@JDo?ZwGp(I1+k(X2NkppyIXvRu
z%FrL+qJOXBJmvQ7F_G=lrKKg*J*fmdVcnf(QA5YJrdcVTk9oOaNTar4t9Ymm;mnrB
zQvXGvVay2E2_1&Frg2v)DKs@R6peK*(to97c}cxJ_3CH*2?(S=F*4DmH(X1@mwc29
zVODsuPmzv>zmj6nd&>&;_CG)X_X2oh+&Nu*>Mz2RI;fy<+wq}4hX+GP?%1Jp9-a^k
zJ<3gkZEI1^c+pRROZF-eHzoJRKSH56cd}wC*v?x9?FCPagNnxLivDPw+E_}nOw$50
z)>a5Oh!nfesH^~Yj)WiDdu|3*1~s#G&Yg#~R^oVqThAK^X03BFp-q6tb`S%UhJIZ0
z?he!-R=+J|?_;IBEROHt02O#LtkTHyML4Xm2gPC-<c=RYJKJ0t+Fo$z=okx-C#N&0
za{cO$m=%~1Z{bHH-kTf34HWkD?SkOb9PoX?U(BfXom;8Y;XyYW`bTRg{|$o(Zgnvt
z@eHmI^XMF=m~5s55#S+9ex@*4qkvmkSBlp$o50b=HzcR$Cr2n|4Oi_)i@s>cl(*7I
zvhuM|zrTM|iSCz`V2KO<<TUu4wBTqdk+_0iAWt`3sNBaCE<8+04-8E!Wwip~FQY-a
zf+$}mh(Y330W9JR29%W<gBBa^3CN14sl`tbic!VBmi9OxCE_2uZ~Mu~cM0-`onmy-
z2lLaRCx>UJ7g5a*rZr9O2Eb-Ca*<1tf#hw)p%uAmTwJ;NTU=YO+S0OwDo;wHD_*%G
z=!#~{w+|twF@q<iM3J#itH?W<yhp8$<h3E+2r;;hLdiEuNWwqFR+{CE$@usVGi(_V
z+AQHYi31MiCsn+O6!*FXU9*el_Q|2<_8qu++im9)is%pMVN7~Jm1hp|^ejSYt0Y16
zO?pWBD;G*_y|)baRV_dhcdaYkY3SL6k<zURsAqGmCALJ|-30S`H*QRPP1<Ah;kys?
zuo*A=jg^L6Q{u|(<uPGs8z5b*Vb>f;s9Pxsz??+DFQ*zw(;&!EWysG~_)}rAq}=bw
zMUW@nySSY}m#Rw-9db|GpN(BYk<;d5dFeaITc2}wj=A*0xRXMUAOB>d-?#L0FqT~+
zD;Yt(Fiu3o<Zv!pF%{|uXJ3S|n!Zwr={Da8!Sj0Ya&Wo+P+$Q$u?KmLEjR?Err-Y!
z6buH6S=JT8{5>lZAn(#GZfXWp<HA8DfeJh$;+3A!Bn9(&=Hw_&%V;tzTYw@&*ToIv
zI~nQ@W7N+Y><c7$R)N-6c`(D*ccH9u^|s~BNAqti1s}nwwNP*mT(G<~v2A+B*2h);
zsx;FV))|Bt!(WgWZLx=Hr#po6b+AF3y*K$-+JW#$K>4=O=8dhuYj=H)LR$|&QT-;c
zLX7@sS{#ca@~J{!ihRJ6aa`*^HyZ)DLboUUGIpk$hvAEQag><%Y5#b!CDsrnpzl-e
zOaY+c41zqpoK@fE;h}^Cu)B%6vMi+|M`qh{_7qBg$$20Av!I(xbaLi}a>!h3sBz0n
zQWRn96I?6?nOL{BxIc7rW7~k$pZ>8}FxxwksWuYg6WlWKnI;ej<Yd&_YeSzZM!05^
zvJ6En4Z2-su(+zBu)H-wT@?=NbeC*?IznD2mA*NvAGLA%U~kHs(hppao&L*~nRMfi
z|HF5M?{P>4B8L@Kd5%PfP#x;_tK*^Vvl#@7HS!;|MjWBA80|RUB|Dt<hfc!sr+?nv
z<7k=WyJ7zQjBq=gVk;;#QgPG&mezf>uMlA;kdu}ZU3W~^jrQMUQPRZjGNl4wwBBIP
znzMWH5jSb;IX!QcQ!04<_l`wfR<Ns^B(7!Zq@^MtXYAmFa6p{Kule|_9TC&VYZzTL
z3z@ZeV8B*R=FGY6e@$2UZ<dlXwFcgW23Q8*Z+NtivA~=k$o^g$PGe;fuX=Tfplp7o
z$>*bOBdFCfbb2}6=KTq>#eEKVvAs0qB=N%Jv)LJD)vp-nYW6|}hruyl=PJYHtz{Xw
z5V@!*T6!AZyqWp)%>yJ??!Q0FJIiZK{R!{0t8fxSEmcMeiNDi^m_KGe3*+D`>3t<v
zA22HmM0cyKB*CU`#CZ;x$SXJL1=hP}&CzFxM-1aGq$U_RKvGyi<y9RNR=PekKf!-`
zPVcv~Xk5U%-l3Jc3hT+?C-*Z55d^PBhO}IH1CS3(hxj1aIeT!)SK;pwGs1BVyIt)V
zb$NPv?$s`VRk4A{QQyweR7woTKDB`$83AB?inY)Qt#-cv`fmw4Xai9MyvsWZGRP5A
zC!6m)6xTHJ*A7lrHi_huq<Y(E^VyUk6#i*p+XyCaH+|sWoPT?oY%lPE@f!~IYoGYe
zl6TOL_m`G;dP@HIvJG%JpqYg9p4gg2f32M+jfLt(dA!$i`)M-1v$WJRzVqj)!e67d
zEO)zX`-t4v@whf7>j&jPxn}{34(hr6L-NVEcMHGLDCX%|Lj2ioV|WIr{Z)$!dotDH
zZa+CZjmycS`RCo+(%`+PBAp6dMb7+w^(Hz*ai!EZtFFeq+WJ14nbz0-9-m8wxP{Xb
zK(Y?s-z=`X9Q5LtBRL*A9WyBem65Jr7q{JOv9)h0#R~TKY&``s4dL|iFR4`}y<9R=
z5tQCrZJtDwhvl!{-|RdnH&1zuC*9A4Q0$z{a##5p*MD5?N#IFNrEVJGfAU^=xn^L|
zKK5D2=(n#qoRo~OK=5w+Cn+)yY<t2gsCx!VlQ)KG(bTOu^syD>uWI!Q*P=-w9CBZO
za8Q2Bo~t5~$aTZP$>#VP6R&zCGi-_nMQlMjC4wTD4S1HO%WVV6W1?3ENOW@98yjRa
zEuU?<*L|YYwxOtEWo5ETS<u`*0P@XQ6C~6BeN@h<uI+pcGqtl81|dvZgYIwG($d=4
z2);=(q<B}-FS2y9c`(=Fac^@_;U=3@@l0LJv0u^+OTdsRAz|6?l2~29MCzGV?--e0
z*$)nljB3sG!376!$<N}U(Akd>gv!O---x2+sF{`SwCLCT*{5|Ib%#ZVuHTBni0!li
z@|OHwtdktyyfp-*Q?j<dw(>USisM$vlyX0{=x^AC?r*xM%Jr{6egQ<rxAWa)Gq#hn
zG~Oy8AV=qDegf-_vrFMyO5<=Pg@Wz6Xdd5xeD6EzR@qdmy~NPLgEePBAw6hQ>hZoJ
zIQdpY=EcNhH#c|SjX%*E<bo*Ei26ogtmO1tm=pex{mNI2G~;9Y{uc=p7=@KmT?ge)
zj)Xlj^<*VNb>T1pLM?6-9;qA&M1_Z{I<CzClVDB#U=kUxNIsEkaba!t3-e}pf>))i
z6SF-=G(VX%R=@+0_V<Y;Ddb?~A@%cdbyXD=!2wiPsez&5TIp)@6rmd*v;|>1lAa!W
zvJWK-5#!5bC#`zq)3BNCVlfq{3s1Z3r#{Fv0gF5|$4}a>o9Pd+H=iTZQL9Ka5K6x9
zZmt|O8`EYnCfHXppek`W&GPDw-slf*i9hEuTz3!;f0!5?e=+;RXPI9-C-GV^1sW7v
zYiA(p+mbtNJBgjYN^3bdM=$SA_H))817+0s&nitlJRZrNZJ$?K9I7_MD)Gb_&quxF
z^L>20Ct(>isHITcY$_RH@dX4>baGHS**PT3uXOiPI)9iB!-K4XHH14kj9}ZktRG^&
zfVgf_yl8zNF7stOXev0LH^UX3x=+O{<+AIn7uXL@<~!EpmVK7oepjg1m+=uaM&)Zu
zFVDb}T?6nZch1=HovZv*Nv%AHh>R`b%;+|j91Ex(si3B&%Ec36;DQHo472Pw6}Jc~
z?tCW{U%4BO8YP14H(Fx#L_b^7)thrC9QDHCGp#{hxJumAd_o2ts3kby-eiH@fqMNq
z4Yv=+FP+c;{$9YJuDDk;zz}RRO{RiQ@juKx+HU-gm~_pz+ul=Yy=8d=ubCM-vl&8D
zJmJ~#$&ow^-M(FNga!7b?$wX@0+(BjpHF*cpCh$$?~}z`QH^(C+seb>94zi|rtPlh
zjYz?ApNa|-CA@ejNZ44*z(!ZNhx?v_vEzmFc!$r?vL1N~tfk4^&5wx<A5EmzioVrk
z>lJy7mQ0i^ZEm*hovF;S14&9ggbyR>p^U|D?Ck3wwQ34>f&P+&kuNQcu2n;`89D)l
z>7rNR_c}R#bNMpK3P1xe>p_U5QvB9?Z69*((Io@FK;Wcgi8ZdPSg^PEi7^TUUadK7
z3JzDqsdw17M}?EGTJKkucW~q$YkHeJ`r2M2J_t|PNt9Ai<yHqS)|ap+hrvn(KJuG%
z#+J7f4F&-Bd%u^*d)Ank$daR)n%dtZas@nC^uZY8daJeA$ct2L<Vh<1a$Vw{J<0Lz
zH`ZSzDFtf4DY?va<VoqzrR;(|(*qm4v!{7t3=JR)(wMy9;Bm%R-D1Pn-yy#DV$d;5
zdo7<j+FA+Z5LtLPdE!=Ehb?O9udCWpyKQ8e2+&#wZH{z>kam5=Qu^m=e}BsHM;)1a
zf%y|@5wu2zdnyZo?W6Ju<eZ5lo}<|r22cKIeHUX0<Y8^mLQ7etr}TT<^K&oNj-$QM
zHiv7QwwXLfOa%qWyvePR?2O!I(OFcW{tQpV_iX03IVmG^ajeTuV*6Di-2*ZR+99W4
zzy_ow1xk_^db#$<m!*?G7Ifj4qKDJziEa-P1j}iJKYjI@)Mxa$mDk@l6#;)|_oGI#
zwmCB#gdNt3Wof2l345iDEAz0ME`Wl+paQEQFx49QR)<v|APR_3ia37ia<a1YXX%e9
zz6wv2(!H|kw$zLKgfp`y)rk#7<WvLl|DHyuv}Bdu|3%idO)TMZlO9uc+im0oi8t=@
z<|382f24}#>L%vQ&lQLBG;bfcR5WPr={)2&xGDN7jNDM*s=WAIhh_LIOrfY5Pb#|%
zT%+s#*(8lVFAkpo`s`|>@`Lk({fn@$aMir|>Yz!EsUyF;7iO)S3SBQx)8&;j#bE|D
zVx9$__ERfzW0g6IWwEyV$~>-tPjovd9&(;QE!{?qBlOr$xvw5;YhD%{o`Qo!+N;f$
znnbV?D2N1r!U8A99I&C82VVB12+zG)i+C%<Rr(*#A;MZAfgR%ues<^4Fv%(3dzIvY
zKl~w`^Xh!qz<moXkaQe~|K}UyD%A@4YIb6UPkq!$6p6rld!fDUaEm-T>>QFJcN)6~
zgrO7itcx>Q8<kyrwnJVnO<3M0r?`G<Z_k?#2Rz9oRo#ClD0Nk?V)7ksr%>f@Nl#UJ
zE~fEmYuB%QOz|(FlH#cmPi4MLybkubROgbFQ<deMyxR|`avi4h_gERCkLD9jSbOwj
z6u{~3H$bkse_OxXc?P+LfSI2tqP{bKL%)L2eamX{Xw8)<w%rvh<w{|SmNDmNm!xL|
zDuc8F@r1ew2ce;alDRdegoUURkQg=d+vvx|LGDp)$?88KP|39?urz<etgJ{rUA<UG
z^oP`7C0{)@W(`ay4;yOQt}T~v@&R8yif@;&1{I(0@9g{woaKEErB?9b{tf;V$dRtM
zO(zi93Bkk&Vjz6oa4z_8MNm1Nr~$yv-iyjWX}iFRbDM~7XEZ<>nO*`Rvtq0{J}bN_
z;`A}}TDov%rgdk%cFX8^Syz4ONNqp3M$$FAdkTNw{_fTg%fd{RB|Uel{j-jL=?1}i
zUtM)F>1)|vxr-$zEo3n97N|v>Ei!6hgS@n>@QDQ7@0`7hpK?AZc3V|mhX4k*QkYW4
z)31LM8o?9ZHr0HvA53%UP^b6s+iq)|3hywJyf|{~v488~{FwazZkhHLrKev1J_!!r
z_PX!kv)qve6>xo<%3G9^m5-1#6D<*3XGEil-L^d;Oq(~T_ktn;Rs4*7<V~}ks<1Q3
zV4jS9nPk>P9YKl-0q8>Y5O5M~=egNN1A4d4@w|zNYEwN-3bvHY+_Vz(ikTw1xmDl{
z!*&F)9W4KMUDN>G`F?3Mlf&>=t_`|^Dd5As1Mhj&uxUh5jZEyAipp0o_cz2WgX$M<
zt2TM=54!YvXvrW*COT1}h|5i+r4on7k(UeuM>W`8c=A45{7R#`4<NKK?g7qsexY;F
zsPJIjxyg}@7KtHppS~STnv1&kytL87+ik;A)N)X6)4>cM6o>0Onm##FcLXkX$-M)*
z)6KX2qJPCdGAdL#uiDd7?QlIUnTg~GoTop}^;sil;*d@P@6@O5p|W`n$Cpu;HMbK)
zH${G$iPC!bWYhs!Eq|ELrh|U^-+ooxn=r?iX*J#Q0DE$3c>`VWa~(#$>*unT^uoN$
z=5yS6z4Su*#W0z1a|&t9Y$gw~5VOTBZtEJW`liFNe3V*9jp-nJTl+Qg+u+I4vCO=i
zbC~y%6QTkBKHRUha_lzwx@#vJb&gl|_=^fR+Qu*-pd}|*w%&cFDE+3;ODEWXyZPeW
z7D@*M(HA$)b*b=xL~H2M5!<?3kanm%ue_YfAndxI`uXV+3Y!S)equF49#DF7-kl%m
zVe7fB0ga(fR(;gR%%UsUhtnJ9`Px#dliIb<1T*LcKUVYL1-&d}+YOgFOsRX(s81Qa
zWBUS#h6-y?qKq|R-}=j`wkWRl`U%o7Y0wCMt-;LLb9<&l?kuBjbuY@D3U&|w)<tm|
zEhStym(z9iut}^DNW_!<u<YQI)2F$`+|)t_!ZFTnARf0M`>K1i(@UxeP^|{cR5qeD
zw44EKXe#>X&tNn73Yf)t@IF%aPw4IO3nEZ{TvC1K3w#N&36DMrZq-3!W}PdoUxxeO
zaLOV#sWT~e^a!7+Tn9tHI)nc8fBSBQ%gMpbjFk+L2qzION#<3p0Ti^%sT96~XNcEN
zWKOL7MT<DhJ#FtP&t(w5TLp#3Ij@_4nGN-Q-==`3g*eD~?_h+2m;yM$_W8A5pa5r1
z0J^yPq&xNdmR2t7wZb%)53<c!L60{~+JZgZFt4(%4Ed!+idu^14qoaJTE&}R3fj#?
zJqKyH6;I4fN|QE#%Y+-GN84(XnijXR1i?r&?`n5fMf($bHX~untUs>|C>WWJaP5=a
z#&3otb742(#$IpTzVu0oI3c0XKHyb%0RQQk6>W#(R$+_Jc@%GNG$lGpZRmmPEmknI
zls}`?>q(?2?uGe7xQnh<@rN94Um2#&$dIFytUJHwW8KxTrfPj8ud}m*-LZm>v$G*A
z$xFL)=wOQU<B+`c|9((<BtSK<csw*?aC>^_;5;}S2&r#QFWK{VSS<;S0<wonzltf`
zF*eR3`Xj01TN$($(WZJee6XoODYE)!R^Z@p7<pMZzp6*4x;V6_s7K*OU4|-~c>$H=
zN2b!>51_{8`oPe3Ydz&_uRVE_QxgD6<f4YKXakl&Ms!E|D2EY7G5@bg8)+KX(zoz>
ze$-wg>(okrH-XL5YBEIdwBhWa+RWI@k6rUhNavRU_e?jwl~nU3Mzol*e(Jy?Gab0N
zhr!@g5!;UM<dXqfF6{xD&x?q!ZY}aMk%}WgWY*=WlyhI+M=h=;6CUp+@4oc3ey7;(
z_0Ih3d^#Smm|wVb3mYZx^Qj=-GJD!+SlrXmr=h)YDPS7fY3wXV-@V+JQ1Oqqas9oE
z2vPRLQ6CyNFxwASdP#GKE{0DtV@HAUw_4bb_K*d)*KlK5Nm`u``WRQ*XffDR=CVcS
zIw$rc;^E=x#p&keSwzdHc2<AB7OE9z1jQ`vyUp-l^SgYm-haBRIJaJJG_^1EJvfh$
zxGnu=un0(F3p0Kr{?isTaA|QaS^ATy4bCDemZe9{G)*DR3e(8^GoSl%SY~r7&?*Gu
z>pQISJ4t}$my|VS19VT|5iNwdbD}Ybw0m$sARZCDUeI(r_a`wZGJcTx!p*|D3NK3U
zF7mebsdPNLXm3<wvv|pV%n{wGdE?8E1!$P@Lnx4X{bc9Mezc4g<U{3eu+Z2Cfk7%A
z5G4gd3zqG9zgX((C{&b5CZSMmqBT4FcEJ5I6|-IcY}0Wo!yp&)aZoW>gQANg>-f=0
zFn_VQoU^q8zFnu-Y|W6(gHY3c5Xqg06^|W~qBXP3`zG#F)XN(ltvb=p!FaPbm(nD~
z+!~;}NuI2C%<V<IHj$xzcLN$wYGuBBv3Y*Jw%~AiFf>gL-YXVR@8odvNcftJ@ep4d
zcfbD}if?Rdv!8G8Xx=C_T^rv$3sf2#FU`<-+?Ux&=fa4j{m?w6^3$0zUIcMFbPnBx
ziTe|w7I|@|m`K(iUC*?+RflyLr9M)z&;Y66zr;Bcu%(U$=50TT&KKd=34UYKd?PN5
z%jrIagbD<5FJo}Hs7f#T_Hk`???3&$Ab1Khc;YOW&>qqH>CFH1Psuf;e8CtiixNzV
zC2``x<LTJppEV_>Ik429Ft2^zAYPmcol1_~DUtcb^<+k`0}ercs@TAq6fEs!ARtj-
zwri!Gz$7CQLL*O%1pQW&w>Z7oDGg6ss_`~m<Nz;W%rk2<K=knURXipnb7fO1x3wZ+
zF|^Ic$DJA5x1xFfQ&5}mGq6&9P&bUhXyyxnQzcm<+_p~aEaG3&e)(Bbx@Kpj@m*Dy
zbL5?=kcu#cXb_{tGt|jX%{uu-K0t#Tr5!rR_imm<TG?Rt+V)wv9sNq#h$RNff4YD6
zp{!In#L_ZI8V0XejE0u2R)IAwiVN-MAc<F+Mf5a3tcf&liyQC(xf}6l5q_qSE3z-Y
z)GE7~e*T*7=*BFSpFi6Dq?ph`0S9GsQpdkPwW*e;tOyE=3N5wRrR+iCfb`O-Z+!=0
zmj~EtpX%vaPilQtUxD<EvM)-=2R3vspb&G;d9QFslGpZ8mHXjRg3|o?$Ly?d+CNkb
zM+r=a=9@Y$!!F+Z{FYiEo}`<?4?%`<x_5bmE8s&as(EH+YoBoh4lmxL%5D4BpNNnU
zc&1^fZ85evx10@z92RE<Q07QfnO42|RW}r$u@8L_{Dq3WE`@naMzckFq<GjLwV5Bv
zxsMREMdpmo2qlhlV`LWbt;aK15d!WfZt<T{m#?79L(N}MPBJnsq}-BLqFYc*<FEc!
zZSZ+;^X0N5*%Y*gzr8@tt5;T=rO&A4<Ip!3xo^5xXS{S8)tm&(hiyCU_$#U39vXL7
z%PTI*C>peQ3eThyANn69q)z{Tb3O9fm(1jit;mJd&_epyXJ0t}jH_Jkuc@1j=Ze%^
z{yT71I!!>(iyPI5Y_4~5yA`aIo()v#x#ZuHzLWKo*hG%vC7rk9gSN}qdC+1M4`SRv
zE_aDyEE?K90{ZW8^ni*mE9g~{p<u{tYezeIt1eFnQ(Wq#IB^RpCTQXD#`gQ0EttVw
z7B~z4s*>5&TS6#Vt%PiznXJbC;PHbhLgm730trj&4dDVSV}-rRO$|JuQGe0f!0;ag
zcow~gVFZifRdu1IDZ|kE3La!)tI0qy1V$aLv;GxNblX@PG`fm(VTV;Sismu)_aJg=
z-L5;+gJMWijve+6j<?5EV6WM9RdnY-{K#>YR%IvXQVj;7z)ViK73G^?Q`(*1Xgn(P
z=tGWHC|Sw!F_!{tif-l?de)(oZ@a-UroLbE|H$Oc`P=J~v);$0zG+hlz!jJiZ3IpB
znV8B692m-?-yff6LOng_tO(?dhW5*|F~|1v(-sGL%btV>a_Z&!c@}4+U)RYy>~pr;
ztDbl-8~J`?&o3)&yr7_nn&R;t0OPk)JEPDb(?3b#&j*a^Wgkr``uBFRn>0mSbyz(O
zbQmph&E8N7<7~lyo1CzhX1u0_p|d<$oOI>&g23qnOyTI)Uj>Itommux!@L(Ix^Y&N
z0?}GlnW(;wC~}w&H%Q6pX3_?Pu(8bM$I?$(?;(nzw7AK&S2I#HWicoNe*Ng~Q|jm0
z&DWVJSO5dC;GQopW@)|W0Q)~b_6hOw?0@v8sE!5sdbZmqch)U84S!m{+jvZB@j3c;
z6TT;vw{{YmHELw<mUYD}PNX=?J%3PTY)D3YNNVf_X2PP_cGaB%s)*6XU>4n!e3{dX
z$JxI86R6MOl{KaUQ==4$YG%HFeDnQ9QZix$Fnoce?=2D=da(M+wKH<NGvuJ1yrnI+
zBAOJ<Cdd<QRcMG3<jL+kuZc|2D^>c?qYqEQL>{kHm3Zb{Q!MZ-&8^pUqekkw!PrG{
zZ^Mag`1oiE3k%W%oy?JuVH0t#p01Z8zHLWGL+$gY2mZiiR3s3oY({3D>^cI9W`6;C
zYPs&oO5$TdDlk-)LR?Fgj#_|UAe7C;@CwT-nvdC<(1D#!En}FW7C}`DLRpie%`527
zlK&x_TTTRh{ul?(jDyDBk=)~ps$K>!Mk{8X{F9BE?N+qXQ=$K7&{eA~pT$?wROLm_
z9q_H_t}u;L8k{5g75`SkyL^A~oJG}FAP<v{4ts!fK>VXpFd9&*rQdAv)|qSA-70Wg
z`#lhW+A%*9U<pM|*LvG^SET6tlEfKz8Nse^@%-z@Ciy}50YU-on>dW8=!y8X4BPVX
z3Swg=;mBv9<81ffG?++$zR>#)Qd)IjQy&aK#sYI({a+*3XJntDAkn-_lvxh{eBXgJ
zB<7ZUp^?YqJS0k{r^hus3Y2KB1!-xDoRVAF^@C$%5XGu%6%LA|$H_I6Uw;b;L-HXk
zM$n1N3(CO6j<+rvE^`N^^h#6l4oz`jkUBkYmkl3VH!UWHy-;1<47uN!Oss0e-?6!}
zsdF?*x^B1xl1FH~2}EeDEAX^fd@$B|&63$z`zbpv^cL40kDx2#`(sb<ecHUJUYNf)
zAs!WwU33>w+a}=kLYa}*{=pi0^uF8So}_i>{8enZKSab<-Pok!Cfq`NCB<S61VZb*
z)oUV?OYn_bq{qrXZqk@^ch_lNp-b_G>5}UP)B5_G<^iAN7p|J<@ujhi`!al4WYS=l
zC=$CeZ7UFw<elvWTN9RImP*~)Rl|w}K;!}9L1Z}4I=A4nsCG@!496n*w25HO-6Qte
zDHSAwN~tL(wZA~5Q-HLUVJ5i8-G*u0*WYe|oWRZP4#flD!z$+M*B|gdP(7R?HjZ7H
zfRp1$t_a*Q%WhZZN0dN+z`DOK^Y<F4heu>O3-s^qBWzta)h`5kF};_cpOu2L!m+;s
zAY5$t={4+LiP_lNo&nE;-O`B%Q*Dk|mO#i&fcHZSEhy+E6iy}zz$A4$yzua%I%Y}H
zwFqJE*s>cu1aeN)HQBtiKP%%`;?2g-{sgL-)a$*%WXI)vQMvtsQ3om9_tyN&pW~j`
z{cyje1#zQwkOz<CJ}SbC+CA-R?C`mA6yOrz8&RAnwfB#nRr7c*go?L=rtYwWnU$8;
z&MZ0IjfLeEqQG?Aee`<Vrq{G&Bjr-9`RL9|!AHuLrLqDl+T2hG$RE<5TWA>r@_2@!
z&=2=)i?O2{pz6UjGqBWkmuZkAQBogi@I>0*_IZyFl$-VVukK<6%N7K=1?&c$Z<0=<
zculmK#3_)2uR9?DUy_W!=qGVt0xFG4OU6%R@>bomuh8n5bATWm?^rpnxu(7+=;(e)
zzu`+T-ih&6pcWKMqI6q@V3?qrWN6v9Zzf9I{*=7#q8mNk!>VW=e`a60M`H$yAbRRo
ze8JlC7!8+OS02(rvmSwO0h4&Ig~rGvxp>wr%qNvR>6}03=X82cl@_0^$n^a<JSV5+
z5hE5h8-Et$S>&Xi6eWE+vhT+L;m!SxDPGfSYir$TMz<tWFyh9R=WNw^7-2zzNFZz;
zkYD}>Y7VADR#*6wtEg{ODK(Ur-Qed(?p7boOXkv2s5~0~7|j8o`6D8Jno7o(_(_Dq
zFn8j_p3{81KmS9Yi@E14$}yo#A1ugUro&?&YPtqy`E1DcbDY=W<*__saJd_;ZN#hB
zH&M&?Ic3(C$uN|rP}znI6IGpv@Z;abAnKRD4>=%=Sm3H0EnAYd4;H-12Bo&BsHiZd
zYpGHLQe(MB9^@1=oy=0Gz}}s2nNq~q`x`Q$hH5HbZ}OJYc9qI3NnQlIlv$&JzZ@j(
z>Um|ByTTHb(Tvo}$fmgW&Wy8L_?)tX`9?2icI!<gmU6|BfTq-%6>0%MN|}8zL+y`E
zHIYOlk%+B~Ca95R?H*7A^yL@Ur35LdKkw$8C~q)>qyP$D;&-u$=EjngNJXyV^5s(~
zG|c>C{WXP+h`u40{OM=r+{_?k`k|^ID1=z~M}OZym>}4p1fI2F++vj0Kodd#^t+rz
z1(SaEu6g}IMN@JV->9P0WR6Ru?Y`R8<43Q<M=OG&6En6<txRd$#^R0c-v2+E&ij$7
z$N%H^x)+y>dnJ)|U9vK+88WVGR`$rwmXVRY@5QxBM#;TKB3qG?y)W<VkX5)wHW``O
z-~0LD`!}5HIIrjH^?W?Yj+K3j;``52Qz4kojVlYe_H@UsULMo1XIUF=7Lbu|$h&BY
zZzes<*Im6ABRD_poGr_GZT_s=?p3840J)aW`j$jkW|Zu^<(9hzeSpYK7LDC6gvWIF
zm<AwA1-}feu7@kUTPbN!B-6^?jNM*ZvCc12T<}q)PVcTrLt)PEl0{E@rz2=aH4;Ww
z$+z^uMn0Dp*ET#|qMeZW-<lV#UOa5P=h*(N*wbtrrOe>kqwySKiQ;bD(KQ9433~hI
z>nDGvrW=c^ezyx!sJrkdefFAed^uf?x0(x8l$Qt#%p3@RDr*{Fgkb_!9nNMm>6%h-
zwl`bh)>~g5do)?s9&er=1)g1=UgB$mLDE2F{AMtSpU$K|MRcfidsi_2BmvJE0^-zF
zP|E{-IV2DT|9G@L_S%`7JHDz>QF&XK?wr}F1RBSM<no&e8zuawmIh$x)#9))S0^JQ
zeop{GcXV_Hz)4tnuKa<J>z4(6;axq{`1laeU0@x)Ed<1miG04Ba#uJ^GIH&!-xoZQ
zJsk4=g<_s!+5<5sv??<iGCBLTo3%WCMjIf+n5LarT8!>tg|2LXMs6gQG3VwXfuqlY
zeXw$I^^kcTK}ebT!($!250;Y<Ndeg(0NNaPp^D=PF+40hFNYv_u8!0#vP%y~gWp3z
z9puZNh&7LSM^7Ix*P6loW+ed3k0<D<f>)ERSy<cgbi~{-iSbq6=hNSODm*4>`r_jH
ztFIw1^~xFax+w2s*;vQMM2E4$j+rH^`m6X7{ne9E>s2(JvL^LV@h_!uuNr8YJRx44
zFHcx0G#+!J5{VxQCA)9DeFxb|)Qx8Eoh#2*_3JjtMDlzF)iSTx=J7#Nb3n`N)TH<B
zq>9SBCsFRPAdt{U2z7g{S4B!Bj`uZt^2*MfcM>{kN?Jt6ajyG9xw7tJ?|)|sr_0lI
z{mZ`PUv&E2emB4!50rJ{EmyklI500fNWXP;pEk{eni8yuu*PJ_dn1SHpZP2AGD^g8
zQ5VGfrqc-DhXYX@)abn`1NtBtM+bloo^~FVv}_-4!l;1gW{gpE+K-3mIK`XZ-W_4$
zD*COad~8pou)~c7%i(}8QPRkRKVrK)<8pdu)G(OeE^p>NDlqMok_c3YXT7+bb0#8d
zLvjfAUF23J|8X`o!}!^S<-_%U2o`)|$RqS0sll&|60(m<)OmA_qIAQPbSZ}y2nWev
z&ZZ9!`!`iCSKs05t|`X3M-e3ZP-EfVQ2@amSo=BF5Y!muG8wR>wf}cDDpj&cdiQfj
zks9L*p$iASdU}b`nA~)Sw*`exv|ij9S={{0@6mz~x4E$#Cu*<}obo+2+)#)jN7aSJ
zU9S|IyoS4elgCL@?-wHw0iyZM<5q$Zl<zGk`u=R{A)1oHc6imJ<x_8C1s2No0;<We
zny%7&vZKZ(RR|h2gW_jc0?L^*z4V-pKeA+8xnSkozm(=>agp?MLfFxAwR&?-D9z68
zWER;s^l1~uM}7v2lCz3SayVTEscGG`ljC6INKh5@J7fTM5l6{fkI{FRnDbG6@%{tJ
z-u<p6Qa5&F?>o2b3?cy#MHo~iyA4RaAo-L*Zu6<DwL?&>GGcpL_5g5FydZvjy0Wb{
z{j;|QY6%oXG00K@QW?5snJeyP(YoGtRjXN#@F;%E<^gkIBaMX-8su@a<a@Hq^Sz{J
zZJ#`yKpgTPw2oaXj*Bs8m6LaDX3zHB9MtdL^(X2$t5K6CDp$CqO{S-&Dh^qptW;CF
z@yA}r0d1t|!&ci_Z*P46*3q`(K*_1%o6MLFqMGU1aDMTMyB;fpI^CZH93mDfZ^vB>
zTj4K)I`FlyM?)Vh*i-p@L~*OJwgWT)g&eh9+fgP;aoN=6APg@rC)g|jiNpeE0nv0S
z)By7I^}uGYot~9rVczR6>lYBm%=KXA`|3+6SNA06B@hP2cr}8k8O4HcIvpYAOJ~rT
z>WFmpVQp(fll2@~k<<21U@{-3rR^@sM1n)0+Qo?+99c~}6(MqPvR32WkU*E?$u8@S
z_*gTclKtx##K-BWnv3?^S8C#2M$dzWf>YDk9tp)>1I6BhJx>qGH6?G-EjL@jz*1Jh
zg7tUEM8UW-R<CaZMUVIbQw{^KPB$kp7Z>ly!Byj|*>PcOaT+yfWS@^J0oAbZwY%@O
z<Q5YEj*J%co!>XWI^DanSMIG{SkvHEe(yLbte%U37s%C0g9#R|a^0EW>Q3N}875;e
zA%@Ky9hoSipXQ@e1~h;Rin9zIipYIS_kLGM><1dCC;W$abuKnglmLZ*Q7o(*?rMXa
zjcEjyZ$@csFLsUGgcIg>COu9cJS`36jQA>>ahrFIY>q5wWMIb{UW<#R_avvZUV%<c
zS{QX4<<?5wV5{v>U4Nw2bZuS!BIH-!nSkWf9rd;4lY>!xOx6cFU5z4*Ds82=L;Y@*
zui*_;a3nsOZ-GJK=Y0SzE=6f-!}Ey{wpU#KaYHIQ!x(i&ltzp;Kn?^Vb8=H|>$KQ<
zm-VUfANz22JDH7Zd2ep8y^(xj_7#>7!tfA&0-LVOJwO!^u2qt(t)?deT0p^MEsSyC
zM_{pgB?_Ri^i@ZHwVJ_n+_n7IYbt;Nm`rf-*s0w*3jFmhL$dejQ#z0u4G=(j_O&Z6
zZ>v0es<P6K_uKA>C+ReTBtCg~(MsB$-m3LFBh|bUuqe;r7h|4xU%9Wcz2TmZ5&DdP
zb`rro2#N$|B%R5)$opa-qv6YTAj&*W{@R6=sd_Ya<{{8!xZ6>lh75C9xZgqZ9yOpv
zw7*fOmXcOyl{=JD9lEhkz%snmH2?1E8}OXFu`%Zl?+?e*SGFu*{^{Wk*#01&BH}sI
zhs^OwAZK!f8|_`|h$ixR2i_ag7I}7lax%KQT0sj_WF_~Dt>!3&zo1r(9*G-|OE4YI
zPUY@Zj7=l`vi^;m41ymh=)gM98Krd$je5yPyyn_V(&A_eE>G_S!2#p9P_3|L4wO1d
z$i`oT0G#=gpin2VpQNrp9-Gfj`|-BqXy9zRW{21CWdbDg>q{Sq7KU+pSTzmj=o*>2
zMnloTwa6luOJw7vYU`I^?44Th>J{~nUGKr!(-h(uh4_J-EOt(cdHisBIW=UwBP_E&
zqAP!;x89}qO+@dOyBrf|u>u_#M_KkpYDv`ig79BlmPW8ony@w7h}xi{4ri0*rac2f
z%<CJr*I12!TOpca)&g5F+_G`UUemRoj`v4L{Vg{XZ<ck%8X||&RD~^Q{6RrJcZlJ6
ziZ-GH`C1>d+gcAN>$dh<D?B_}K1{_Y@o*D<`RWFoo|BQ8%R<)Evqvq*Fm-D}zxAf#
zv;2wUg{5}v@0F<_(l$?=b(-8MY5Mt8steJiwn&lYbl4takF}|(r^IIR-ulfbbD%Od
zqGmn&sV3d69R0gIzXch%YtN|(BI{11Oht98UKz3)BY6<i*9^-BUZWPEzO*3YEm9|u
z(VbfA<+fT2BM2gN4`H%W^yqD4cGPL`ZUfP9JfNN-D5l?8$y8DQtIp$|5t+eoR_SLN
zuOMHrw^lIoTqC6J1oD^X+$USxxyRyl$7i8))2z?){W`k(Z)eI@C^*s$&Qk)eQ9z^x
zS#H%Ez+eEMeZ?^WU23_QrA}R*XbnXlbN_1P<p0R4y=Xu{N)@YnL}|z_Z|o}uAV(;m
zB#~1U0#BAqPY70A`I{PM>-U68la@kuh=02?GzOFMcD8^CjutZ^GJ1&)|FseX$N=DS
z`|O~9h6-@dlfC=Uw$VDBfTlUjN%zo@cnMx#Uz;ZuMx06bIZ#>Wk;yMy#3&lzk~V}n
z9d#8&xL1_yCnn%nna;Mo)aaY14MypxX^%bU$OW3uet#IP4@RuQsyrme9W66r(=^#b
zFe}<y&06V3?iC1ULaoHp1AraHLK*FK6V8_Be5kpx=|*x(^mMC#bNcvLc>r-En~qzn
zF_?_@pHUm_%4zwRCd0Ir`&1R(+d1*Kgsbx-NCGp{0RTX>Dm7su=y<miLt-*kh?t<=
zaNSJPj6xh2ZjCNmq3nwj?+nol3gXAW2}?XTR^xM|7A1uj*U5L=e)eM{bm#}pLQzZk
zE7&OXD>)UG16h$~_CLH?nqVrRo!uJTiT1sbpkC~daf9_jtEbgyfQL7}WJr9KJl5k&
zme&<06CXOWs8f|GiW;S+V~2y1oy%%y9pDCVd2Qo*($-l~OIuw4dy9wnB8K;bY47B@
zSkk_UGKL(KD@YN^dZEqPz#5d0msgm#=O1?k6J7&?&_@e@CZ#lAZi7a@q__#Tq;OVq
zL9nTx2|H<kB1#UB8Io1b8R`?m-LQgj^#`>uCfE_76xB#)AF6_?GzHUYvfzUb4M(bG
zUwfNT_Q>AB&H;-y93H{kGLKjMNkL1V5sNY)wgJ+dl`w3P+l?YNb=TZ}R;3^5lsUg)
zW<vzmyFVu`E|}(o&)17`_=S7o{l-8<2n6`+GZnRV&}a)utj^neQv4JC6aKhvvvIqn
zEc1*{+q&4Tye0!7lH(#A)IYnyV+#lUVbQQr>e~0GBU)1^xnXG2>R9fnH)cW6tk?Wz
zU#HmJHllX5@kG^U1_l63GlV4*m%47-I6_chj`@WZKo4*}caRhLUZu9pB9O+ARY`6q
zpn6U}F9-M~s;v22ypzC9f8VMDj?Y-N;rJeLcsqy<ZuY`LD1nl<eQ*Hq{`l|F#oqK#
zn^Id&#=f+!z6ZiV`Y6I`@~m}-MRld~DM9MmYreqxnm~oOWV_Z%BKd4+E6{Nv=0A>M
zt5Id&$5(25G2Ee)?OTDmrBsIK;3Jy{uxH2Ckv>$6FQepF|5gF&GJH+ZdL3bpAyI1p
zwZn_8qkWaglhwJ%OT1D%2qBld)xe<8>I-P^9Ns&dWdITkz#1sIfbw|lxaj$O>nLlg
zrFU`gq?u#~E(E7adOgEW(>jTP{SY!a9VCuvJ~V0aO-JKQo{JcZ5?eQ0>mvR|aMzt3
zT}GbmudcR*M_wL;7mr%IxS*afJqfHF=)#ufOE>`30hE%+xknLU-k;_s14bvBYet$f
zl_%JtA+4D&$ir($Ce5>HvSZ;)IZgpu7BE+F7?$D8NN`uuz3yA_Mp(gp-v#7m)`c;6
zam}jW|M$B-LpK_!YPbf!LB7K0qk9*Xj#i(!+DGnYU40%f=LB_7s<L&GSsUDt<OQwd
zO0t)9a8NV`s{a~H1DS|N#tdJRkA2Swx%WrH|KnQx*|5X_0(#+NpO6RF(+FKg30wn7
zqI9OhxLaOIe7tME<0EAF7c(Gh$kT;oj=F^sM5GiK(0dkCY(=)%iv5qx&}v(`Qx`xw
zQx;kEx&6}27rx8YIW!i^p$*04d=a&i^Y{BZ#u^K`mluMKFO#5H?C|ohR;QBWQDp^4
zeXr3L>a-SJlvBL@WM!=qYHse^AIR*i4z}y7)+prS#rn#=9wLL8>G*w>`;&Y>9>RT@
z(q&FjlyTf0e(}E-hm&ClIV!tIBgR!rfI@JvdR9@&NC@;<s51w2rn}0M)EW^I5vqJ~
z(1uZXix0_VwQrfT+(~iHXgV2=lfyl9(M+24sL+<vz0Q-F+2mJP@mBGD%IxX3*MLaT
zxnYg6gN=8QUeO~AVnRX_bYQ-6yR3Qn81$~(U3i;Va0~>X*i3fG283c{WzApm+j5S~
z09Uy>U*6iPt+O65=8L@c<#kX09bOuS|88IBkpwOW@+-^H_`3=+hS8%~esg@f9Y#HG
zuy)PZen1#g8b$%F{__k`$x8$!Ca}@y2!f+7b=ltR2ghR!QoLs4cwFps$j~VrbV&mX
zzp;$*T`f9$O)!~IBO@V|7DBKzM;hJ`L#(kiBy0?eX2pRh^SL=`0l3Ued(@XNg?cq(
z&*d5!wjy?F>uNo->2-$#RV~$@`vAo{ZWm&G5F0-d*>l60)Y_WKtlYZe{)w#3-C44C
zE}y_iixEiZlQxqdG=>@Od+uybO&MHEwWJu(sjE0tHhHVP8*Yh{tCt+4pr9>p?3P7%
ze@0>Ld0jU@2bgukGfqqrptR^h`3c>v-B%awvFDEE*%E<}h20x;;q`&l3&mLM|4OYa
zpN~Huk9>pTJOi~PL<HuI?vo^%4Iq?`P&u9l;8gabl|K^Jj1MVbzMtD=b-;$H;8lwU
z&VC$GO>tB>D-TZ+`98IJOd`DHko|?+50l0xT<_Ma3d_GnaH|EZZ8bjjnw}1bXnTy0
zcma9_e2m|D0V3ZeoZ)TT!cR`>oXd=Fz{F|>_LD?LzW_$6#^cLgDt$94ud<8fc%u^_
zDHSUH-KcOZIvjq-NTE-~mFr~47MSOVa@PTTTeF|7rB}XMUtcfN3&<3JZMTpn$*aJl
z$^mEp7TcYEH=g!|=jsE$7Z8f%jJLV@^7M4`)7<4>6%|ZdM4OkD89lbJZ$I8k;VOiB
zclT>E;`YY&u9Ib9CYws>>oh!&<NEocNJuz${HI2w8IKFxe*iQ8vzl)P2mAZ)0|SjG
z^5;$>LjLPtq|}QlY<~Gl_fwU`oNct6Sy>J{r{7lbTWG9sW6A{fGqVf$F1t-P5?nf9
z{Km3qhkRJh`IPV>1_+RfcH@rif^l+4U8cg_WQ-oXbrJ4F$Hz7_HkBC_`c>adD6#Cb
z7Fc?Y_+FZ)cIFx&B=V6ZS(+knYtDAg*0wIdlyg2#!Dw!~L6pZoPCMO7^gmup_TvEC
z<Kk^?lZJFOP=bo!?k&t@-)E0jHXNXM4US&^CXP{2yynkLzd=hTgOpL-gaUn@`*1qW
zQS=PNPcgo{gBfPq+*Y8Hi;ATmtt@a8W||pS<D}CD#djzZ2evz9Uy^oTcZ{-u)CPIq
zBOm4Wb>TYEgA*0@)Ypu4U$V$|4L^;HisJ7mLqflpWa=N2wmX}oSlnDDJ+lRJ6-bWE
zKQT`?gEVQiq$VX!tG<S28L=RqTe{`;@9#4ueSO9zeD>3j)NEw(ZQuHxe!tO+<caCo
zu)BB1bF$v}1QlY_aiR50y4R&(!RGbXS)(#D6>!fU%tGGMWadpET8@>8qSicLp=Lbx
z&t6pV+Ja#qAHuWmR=@$@J6V~r2CUY88bO#~tMjTp1GvV|_!wFMgcsNYKpy;j@!Rq7
zqWrRovb*-UwQL59y!0u&1mN3v%a_0yU*FvcV@hM<T9GC5>zgCmE`~5y#^pW>t9XIK
z!k3Fk(7pM>ZTpoPz>6`m(TteW`fIUllSP$_OG{GC#V7w_zIiYHG2scrU2SLPeHCd_
zMvn?pp|e(j^2o2ruvm>jVsk={z5w6PX(||>r0xx#frY)mv$m^MZ_L#WrX`}SHiDo4
zrp339jkhnB;jL20MXbm2-HgEphp}al7w`pM#_g)}aHs#cZp~nIl)(8tAHQ-;5aJtd
z!$uL(jwHW8=5EIr9<@(|b|T4NUG6%|N?8IDR$R`+LzE<ilqNu#upi1)S#b=W8!RZ|
zrQGjwpr+g!-*2y$AX<)%g1g4_HONlKH!yWO<%uojXU#UHnADmGc@@!p$-m1Hmc>Te
zo;Y9;i=OM<t4NzqJJYR)5_bM`#}|QC0fOlC91%+BPq&N^CKe7YVeisTWPFu_<WNmI
z6HCNoR5D<mYT|Z1^rl!W3er1r<5<~pw9-qE#fd6G&?o`&G|vq9#D74JFHd=I<_d)W
zn{p%6UL6a;NP&zwBmSIK7JMjY*kqwv>q*w=Xj&&Tg-KZ{?XV@m0H%i_kcTi99+v?V
zTuVUDa2X0kN2lA|TmW}Ee+hYE@V0OL#2Q};lKl?|R{!wJ<-6|jnez7Z%lCOh2=IKd
ztR?{{2h%Zqz_aoyTQbwha-(^BvWDh0k}$=WlU#!-an2R4+}h?Y9o+@+7i$N;UQheT
zVg3PH0F4v4A~CS(XlnBYv$z4UaN>C&5}`vfVP#b%L-4IT7JoML5HfGyW`PYPTiqge
z$<c!FcB~?A3hCrHRLAFQ&|jW>!Sf~@@1^I)-?hcvwL2H~N$&yFF?o<g?R^nsiS{&{
z0V4q?ADQl#U2g~4Q$_r|ATRbm&FSg)udnkNQ>wpuRsHU%9WNJ_bssWs1b8Lw4D8Tp
zSd*37899pJJ9*pdCvm7X9)jvhsS499CvUUqfGt~<w-$&L!1Ct|vGOySu9c(K0!~SC
zp)L-xHUG<-PA=!J$Y9Zn$;rdnnf42cdQSX)8i<FY<vO0_6^IY`4u9_zNER4^_k9IQ
z#S|!nf(ga1+TiM@oEw`HR?}x|+d!;dG+L0EgQ|e7#(p;LTXDPKp}yp26JedFT_R}W
zPf?-`iN^SEELXQ(I|$ftJ&7TvKqu3R3LK~`8tWe5raJikWi2Afle{T@+i|msW|3?^
z7_3>Mich5Nm+qJBm6nhg1PUBho;KIqAwSBtIgf=f9y6=w;gQO1IjdzE(lgG_hm@({
zOwY=SUfRoQ3fQaiGtM00w_k(UFw)~fCuZ_GV847q?6Gn?osf2_;HoH%NN3U|+yUcb
zz;|*B_MLTOB_pnaE;|MVco-PL5QdZbA`{o35zgwrJ2w!>?QkNEF4}z5*;~P-uIEd4
zjS)l*rR$)s5l)Vmr=TFPf33vVk;@L@VW$UsCnaq;%5U*ymVpJ^fDJ870c!!hrkHi8
zKftco*`nAjG&0i5)4S!cVs0Xgn}kRZ^36KxHVwr}NUMSSeX~m3GFNMN$XD}O$%YR^
z5wcV~H(Zz#g@~eB8)WFAdxnKu#Exy`<zC$tStoOGF}ZnWBa>p4BVo#DY%IBG7n}rn
zsa;2@!7xNZ#7+BD^I?L{KGRTZpZD&#_i5c@^v5;5GC2Km15Sv#pImdTy_@>xLP8I7
zY`!G&QW+l<zsM1#rEME(3Hz~Bnb0Hh#t=%>;i;y|*+to9G0^i1`1t$rp^1DAOu!{g
zUbIr<>1+)>=SBub-<-evItAcFqWEs2`HHjh_;m5)bYta`Him3>=o<}D7&;VEJ6lm0
za?Vl@csd<wdrW?}xC3-biUzKKkuwac4<NCZzNc=s@!qMkf~Q*w{}uUrq8K)}ZFVO%
z6*V~7;ypOdlMwZ5FwSz5cH~$$Kz;Tx9S?``{oR=CCThZI>*h51xbFB&cJG;-Gj|rK
znR=!_fDp+445KxgUvM=nbFQ{9AFLo6&-tTs4EQT%Iw63&;!vjuDd5Md3i@fltV$9K
z`{!I2e66x%)IpUBH&8PXLdA@<o2Sr*Rn;l^vMn;G#^upgfE<z0Rk&ZRkKF>rVXU_l
z3C%h!8QY=U1s}XBjm;+sn<gHa9+?96zwO&uT!F<K7})_p9<%bh6-9UEw2@Q3W}SvB
zYKL6?lG!F|pU=gEBwBzJ5F97P!&BCPq(<5S<$eYyiUu&6%Irg&6eSPalNU>~*@!*&
z`b2a(Zjgp$upNE|xfK2J%QEfK$6l&J0(}u{@eddf?To*qDX4fTgM+ajl;|EU_5OE{
zYc|yrqsidUgM`a)unFONBtf9AB&z-~0GYnt!p->?)a@f8;m!b9U@4fM-aO>K+V^|A
zdUbkrxw)(_9`JZ3;s7>+ixplrSJ~HvN(@Z5K5+z|XqF{QoFknS5L)NYa>xg<bv^m-
zkmE)xAuhjnS=CIbrOd#HKaO-YtT9F8;yvbe1U1Z+Oa7p_l|(MnU2G10@?>`*Q1gJ3
zwHOjL0Iw(o^E6$VBpK>iSJocziYO*m$Z;BT;kr9-GQ`~AdS(Pj-4N{c=$V;;DqcN$
z@Lza1-ElO8AbBCmAW6}l|64_;5z_GN^4~(p$Snti=Z&P!Uc@5R7zhfE=DNnIhr`g|
zB$lKT{{}DF<#EiFalR45r^3DycOg}(eJ^#tVIAO#@BTM$QenxL(%GU~zI?x;8#5ci
zz}T6d^2+UX@=x&UYS`b*%<Sf*4uD1-mz^oLpSJ?nSg{w@h!YEkzjlH{?O$w+RMQ2q
zSk#^R)G6gmoQ?X<X4GWFmLbb9Y>2XawwWx7;RL-w(ciNEcfH#u=c)dy--b3<%tib|
z&9V%?jJWP9eX3{F#c?ESSzW2DXp;j>jr+FzLm|XlZ4eMm<DOsh92(tJ&KNav7&`r~
ztu->FvDG@;a*V<X2}}4^RXVCqR<w0G5{|uYRq3y-Fn(CmFS(J%<3q=e=}c<`1;tqH
zljnX(l`L%EFsmxnA1I5Daor~C$`|WdYJ7l%h_z$W9iC#nj@V5ml;hgy{(kG62@l+A
z&)~SSofT(@!`=AmS1jQp-N~Nl{Q(lCeaAG+JM2?=Se@mx(FWNl|9GM=sa34G@>Q@T
z?u9n|33%-n2zc+d2lN!P76CC%YqS7lRm-KW*jVA%(geqi)^gv18Xap-dV0`UZTBfH
zZi{ujte7yqKdCllJm@@nKU1G<-%_LxosOA_w%X-blP0P#*^PmSatZZjf%X#RzSYS6
z#rxd#;{!Np`bOh)T<|I02b<s$a>4CNBs&POsFwD|WRg#g9}}kaMKrGt`yIuc(B-~(
zxcpotBBKav0425DQ_K1QA14~X6~Er$m2}$q<XMM&_~GNku|(Y2*YIKG*G02B5tj+8
zP$pjcI9n+)ONO=O`0?g%<Gc(X>7{}CM>JU%^01~b?#1=}{Y7zo@u8cX45{HE+V<6r
z9>p$9gaUcjWQ>yE6m0@_+XYC=cEmuqb)qJvgJXLE2BgwTq2I&LMZ1K{j@i17zSZXI
z3=s^%qQlxpzxqa=aIJe`N;e>TM-;nmLk<s~86@ir{!FLn93T#WU;_!!ypkP#Bofpy
zAH9yD|26CYSho4JkMD-`NP|}HpJ*W*7K{qTtZ5M#&EOjf-s69yXNoccthl~UliV7n
z`;5#JGNGwR09$=tT*|_VeL;3z<zj&M-D_{A`W^Zc^&ftGSaBGx{CEEL>GKD(FY}nA
z+Vf`vH#_4RSDsu0YI8})dH9n-KM1>Tmh}(5cl?UVBK8L;D%+~4v?{l3SFlR~4S@`=
z;LqRzn6>iZtk`P8t>8SWHKg%H0oWKDY-yH3i#k)byqivnZgRNG(NQOHRJoGC%vG8k
zx70(m04!IwUfzFUe`2EF-tTMKmpK#{g)Z`1y+_@=bx)!#JYx3vLuR8zR#dr$8erEl
zi3Udzrlwvw=5;~&3ld)DT=ws;nnxJ>qoav2Oand6L)0jfR_(^rgs;N5Q7Po+f+cWx
zlry7fY{PCvVQNi(h@23RBNrM8JGEzGHwK^%ux~_w_AC3`q=c1`#v4Fwo_`<b2E)B=
zn!XvpA1F1KX9?UmMY`^&nQGz4R+vFfW(gvtU^VDM^oHQVnZE!SkQ3q<qM>8Q1(U@1
z4vY$n4A3JxF!KjQrAGtISzH^LipLV?TYm)FeINk^!4Ii%fZ>>d%&x7%O%5S+d_aoF
zv`ns;wn{+)?`7;2F=>GZfsV#bQRCg{7%GbutONXmc3Q&Z;K<QN(WuWmO{f31SWZE?
zj=H$|n?g`m7hGX^OUi_$&FL4BT#Q650N15~)L<N7$y+fkyz7adUgd66jJV7qTwkEG
zkzowre%-~i;1%`D&n1N5$kLNZ+dL>;OkOA(;qD)lIF=x!-hA|rIJH+-muu(i;<_W)
zEec5cST3zdp#`9z8j?rRYV#;29<KT1?1fO)Hwtk-Zt{|i3+z2*Pv{vK#Bh!$1?PSE
zPnim5+R>OUEz>unz-YiIgM0@zwnxCJdZQ7JRqA``pPVf;9)Pp5nqcS(>#~V}UwYFa
z6%TJ?e_C~#h<(8oxVs{&u-(c>{Q+&WM-ykdo{T?>{$neww0d*c2`Y_dzg8tbHC)?d
zC3ILj-5R#Z_SxYUl~4hX1BNN#;Dh$6Woh)Q5H&uLJ3gKpDdJ}-wvRC(%`;il^C3JX
z=r28?!XCHnUk*=Xmmho<d4?$KmvFrL_U7F4#>K^njx~t2%CmDaB7K?Z-w!J5Kz7yV
zUp`Q|g0mX3?tz)&3P9AqM?L5JzgyW3ky}(nVnAHi`kMQ;bJ=#Jf+z+*X?E8$f_x(A
zsg6b1wJW1~RFe`DUaAkhVc)H;_3$Q>BWgQH5eGW|AY%Y^%C=gdEY8&;xG6ngLAcpO
z-qQ-s|4|a?CIq&d`c$#G-@iGFcZTspxLIU9vmcylarXkE0KbYfvG?NbAO_o#4uwpv
z?h0R>Z^$;#TmS2MjHyXt>@{gsL`$6y{rzbMo*!*cK#!AnIh>{r92#vTqO@8g+r2+H
z2(!%6ZKN0@AjKziOj*B7PTLh_n1r!~7T-UrBYm?NW&sIO?2@R6Tojdj*29u$CtttW
z(mD^!`C-UC?{}?NCX=_Ij&$WB5Ey?FWk(o$nLAcaMUgBO;T{}NFU%klT@&i}4t-n7
zH~way^XSuB0Q0mBYw(}MvHwo@zSMET9*ScB9UL&bJgi_mJ~q1JbE^lafqZqx`kk1{
z`LX^xOk{KOjKBJ=X3NHZ8s`6|&E7qJ!?1vKJA|<2-ahPlK#aHr<ADOD1U><!I`DwB
z1QaV0o<Mwxnb>n1ik8YklA?ejC7*?|^W3`Q%UX0QBe^SC=FMFx?)pfFWr9yMf9nt=
zQWb}lLTYu^xa^`PWxYtyc6hm+5SkD5TXjQOq($w05T}}bqacV^^_wTM?XkBDOG5I;
z_QAZ&+6giwWA<BB-q-GQ3T-k{pD`LaxFS#SgEt?;$WBFa@V?LgIe!We$9#vfZahO0
zeSATzc-sjctVnrQJ<>fa`=@a_fIOVMl9DW$;xo!<WD(MFRcNP<si5pLin56D0yQRY
z9v*eB;%#%i@Tyc@+EkJ`p*}$y__q@3sWcv<n`jXzs>g^>q%L!Hp}uvse^NYPJ~Q#A
zUI9}@CDu9VIktr}(i*PF=3(eo>Rg~8b{PG`!bTZl#cuht`?_Ho!!XK9Ip3e4iT7V;
zRDx(?RdV$<4}B-CnjVvPA(O!kPRyTx44qzrIxea%Op$F*pNF%#%Qjwn_}{4fQDxmq
z#UO4}XWCexDq^iBrDD&>S@Z`ppj{EGt&Vxl%um4tzdO9COl~ccCW=_>6NXWlm9e<_
zUus205p?<@v(E{ise!i~emo&)1D@(r#!~y9ez?A?E+E=Y5uw*q@+Mth(`Let=9SK)
zq^w1I)BCSQrSu;Iuy?5jcwYBmq-6f=BWS{#fP!^b(sDUT_rY=-+H#fgzj~e=#)n{l
zIvpl@2a3@WP(}OOnhu3Ar}0x;hbMdg`u9gyKh0g77+XBp=cFNIzM*}EvNLDd3$S_!
z+ia;RF5hf~zQt+OBdfEAoOV^7vSh2j88~QoEVkb6*FqZj<ByruD{;06^<c0Ae22tF
z32FksRH>c)`^%H>TF(#2ab!gJuTL18@y-#BYFwT4Nn8Tq^Awl;>#nX}pD6i%r%`EB
zhtXmMeQDmH*F`vC@k&3E{{7>873Lq=_67&^dS>wFx^MQkE_v42#IJPp`#Jjt(I)qO
zYiMSGJi1(zhVcK!q*gYv<&rY>PQ(pGi!z98eOW}oL*@4=Yn`X8T;IUzJ&;sP)GVKM
zUj_X>GH+pxR2gx%x}|3NkhX^}wh;b;`fFuGuz(YId?Y?O-=;;zv#Hje<$h>p+$R<Z
zb!)Jmz|5m3KUzJS##_BZD6J+ZOls)ovgaCuUM@f(T_#5BR((d3WRvQNzoomxmaKy7
z&Akeyo3es<uBo%p{eJPGy47rquy!plPBJ=(-re0%;J~sp;i>Ate7kmR!&niMZp66C
zRXBUgmvo{vU1=$q-v6|~>&fF<yI(RK4)w>q-!-GziWSS66Kz2@n)4oER{{&mUr8-t
z{rMU=S39q1w}{~Sor<<r>(<ur+d?)<Rlf{g81R8;!G2F(Q09HS<`dG2nXQDbOc3!~
z;MIKbn?Ko0c~DNSa-X)Rbk02Ks>j1!rRfPd0p=ObR*q$F((Vh(7qXFNy^4B_HXFsj
z+K4D(4U)=mDD($R>C~0{omw!4Sb4wws~w$nH}U1>Wn-NzmDHpWOPb<W2n=3vO;AF+
z7^<l&n;DG^sSoq58BW3M765W3Pz@1jg84`H8(t6sf~mAAd6uFS*<p-242p|<^Y;e*
zcmM<*4!Go%)Rppw=FB^H^%H4C32MQtnmq7?c2)wdL&k|rJo}%zzvs)#-ECLre;2#=
z<C~@r5s0EkUGSjf`8;eYf-HfM#`SMC?$k+q%RVz<QZ!n)(#3fEv*K2NzPrS~lBM{z
zc0>#5wc6Aa#JKqd(e61X`n%@1$(aWrBLI@d!c;DX$nht+c5%Kaj-PC~D?`tSq}CO2
z(6ef{pOWX7em9W^bR{ztNBj9Wcmcf56muPc)P)4)k<6XIr34K<Pq_F~f;bhR`>WDT
zf#Q(qj;YK>%+r(sK7s=MfaFC_uKy5x4{Q#8vd*MU<r;;1>Q!9qVpGAmRSqzGcw+t{
z9Z?i&XQmZ1Q4uuPL^@I;YWS3yCj6L7(bfjvPra>3=QhcD7C9X>ZtuR(`gWV8(mDg+
z+8MX)Wl12o&Uv?_$bJlHk|AXuRhrP=r+Hs^tbi58P;BxaGd5Pzhqpa`Q9D>;R3((;
zRgeO!=&lL5%K=PIRdf@J)!6uwqemw4D2M;sz1;r#a`r58JP>3?*FRgnwRO0BIx22!
z8|bNrGD*e$Oq1E2@|?W$&n|h~8RU|`fU~w>A;X_MoqznfBs^CJOwASoQkUK^Kunx|
z)oz7j>d5|~K);wT9|_kF4&HEg&WIEp(xK?A>nj89iIy}i`@4X3en}gO`dZ!$@Q665
zT<L8i=p@7Gx)(;Ig{6A^T@xkHFQ*tbxbjSBVWJ)lm@4I&8LeZTPyAHxA->1O8Knu7
zlFsWiY%LWV5AE$;gn(=9Hy{*ojW&Vp=F*Os=gy)!M#j<FfB@BN99*QqM#f161J+FB
zYx)`Ke)!>V)y&<GkGo<i6a6DU>QGRu<zWe*%UA4IlI}kiDAb{f_gdkKH3-gFmbcB`
z9b!fo$>($1I+L5Jbttq3BrS18+WsVF56&zhqd33d0-RT;ax07lLUf?axsta402NGQ
zwExSVTG?L3c?LVOPEY0XsPo<B!NrOH)W(A-FHK^SNCKsS*Cg3Bt#Wxfx%`!$6PCs$
zH@R7nvcQ!TuJCr!xli)y<jpxfqk|9y0LN!^xh#%3xzwMlo0w>ll2i&0#<Xlr_7D08
z7wg<~^D33Kob=e4e0I`!b@=IOe-y*m*4kQI`yu&bA5RT$uMQXo@sGFOl2^DraMD!5
z-$_a~zLC&??8Pk$wMa*rH9Xl(xDM6(etRA_r8FI*9!rrDXU~7gMGoAz^I6<08E%EZ
z;MX;sVBRaKu}B`6qbf6kg?~kwa(!bXk~MG&X|`|A9}I5H&U6Rus43+!V$5FVS8i*R
zPJ9{bQ-|1h6^U638@LU9`cwx5BuvH6PKga5*iKpyFB-|(Kz)qY^8at84<{_^C7EFK
zoE0AIixVa?12AVGvmb{GlmXI*n9khc4!61IU?#JLX}lK^UGH&kHZHrN6ZwaVWDbmu
z0^<XI)dt-2EE>ZE*i3rfdq*y6xY>}u!u)NX=8QI%@0>+8gtr{6PF`pfbo0qlzNe&1
zVV&a4#L`@Y$)`C4(1}=QH0IWhl1{$9LF4AW$)53N|6Xem7|}2VG1&|?`j_?p5KVpg
zkcTe`GMUrk=8<-qXK9`>N+Ml!HjR=ZY|Jvy^WY9Bf(`@!&)K_zQD$G+-@B#8zU-qa
z1Wf;w#TksKiV49nP#2Dj@TNAG8s;1I1riU-Th`ZHvgo>&UVx+ZWb!EtU`}WR^4_By
zyyXKSv`7sTqRgG&loA(@m<70g3=lFghES+7uPaCnXkz6uqmZ<~U?;o#Kc#DA_4_#-
zY4`0IFgkamJLBUL_>^?uf9Ttv`2gN$3NJ&;w_$n!RL=E<xh0bj>f3ghgx5@qhive>
zEu9iBzLbqI#<hX~03Wy=7xI($J8@R`b=6<0We`PblYql(^9TkuX^ev2E7`)RQPJQ1
zliXLAqi!}T5mx|?h2*3jIyWyd4qQQsU&Op6=4j$*HjEqJANzwkIYemub_gq`M>Pdd
z#0-EJ8uhLFqr1AhwJh#De>yGXyeEQIPp)SqHHA(-^PDW{?;l-0>0d74j*JKm-LeU&
zw+J4fc|J&U#MiO<j(S+)Nvwr=le1V{7+m#BGcK_N?*#AqTxlN~*@64VukB~BJZgIj
zg8*_xGpk*+xrA}0#ecCZRV&;UGc}C=U|Jg}r<(1rSnTxORl(De{O}vH8@^wO!Wl!Q
zIHxj)hk@hvNl3<hNCKLiIlQ)E>`7MC^!LC=kh84zvCot|H-*R>aFHy@5B3$u%5mAe
z3KR^Z6Wz+1+$^fvL_whgQ$GN;i#X<uUQn<?5DPv1UD|us=B|6m9<9f9Zb&Ffw?m^-
zWA%ep&zhR{PqzK1CIW0GG;nxli#~M1_J?eZf2}(kwx5RZtJ%ssi!M+UTXoh1uJ_9z
zw9!&70Tl(IkKHhu82$3LcI)fg+9GI+M8#FZs#S}Zh#g$ZQfieTn&!eH{}!W<$|>P3
zw2STmo*K}m8tF~f=5BM%M*KS<1u-%*W?s<Co&82oKwJ|X31RFPjXR${k>5o%b!jw#
z$=ImQg&#LkOHmR6rYJizI>4)8uUMQxM(o55FVh8w>n3>vwA5XYs)Wj9S9y^qd{`g9
z$%ZVycYgN#+drx8n4G`umSCPuH@&S}UTB{6d?*S|8(H(>%GyGs6@YO-_sU{NT*!|^
zDJLloz8OV99&tWKWHsLe^19vWpjP88U0NAH-m^HNm=F3lB^H)i&L?8@<*0d{TGE(f
zr_-f5KlY@n+5dQ7Z5Vw*D~1zP2d_ji0P^LLG<HiUEuzU9?<-0>EM|19b9F|HNY4QR
z--pobzfMGvA*hjL%it+S@AwW6JNTwVf=xW4r#4dV-4)rVFv!a>A3A;3+Sd9Q;A1nn
zf1=Zs`msypk5CZ`02^-<35WtXiFH};3o@ZB^|R5|#}_BdrdHgO6OX-8RFsr2PxlUj
zLn3}zu`4)))`u-&L~jwI^%;ZU_oe@=>Aj+3>)st_3UTD)VL_{ES0-gsN9%-jN`K6i
z0qQ)7Szk-hih09A4KNORd(hWwL)-;OnH?%C@~VYwRP$_=rf@^&Q!D>dY;ax!$ZCp_
z?h4cZkUcK9zcAiLnXyN}0krb-kTqVLv>1UP2ym8g=+l6CIcXKvu#NB6H61;RCe#^u
zR3Ok@+ry{zvw<=)*{;s&=E2;dn*}KgY>oH9+VS7VH~=5gM>*f^IQ;q%cay{tB=IJX
z;<HtKK@45pLVmBkT_JIE-CtpjL?Wro%)R^WVe`Cf!U*38_nln^vie7~96qd`beZJ3
zg_|!6>ea`j!v;#W`JkX4r3b3>Ms=g1t<!ozWQ+P{BeTna_xsuKY1Ma$?b{M@3X*n6
z{pi1qp0^L>V7XN`WvfkLDB3*b#&lohx8IM0cDS~;cfQ9;=+J*0`y%38aZPHcWzMV4
zmK^3NE44LSe_NBM9Q*jh<%Vj=Zw+(!Sk1&OK7UIL9>z-7IgY-e%@!@RSP2rDL7)tN
zfPKD<0OQ<U9IqF9fhdS$nqQ?sA8Wq&#P$u8X9-2C9Zr06QU&Ks7+ur9q4hU^JMAq+
zXldE|TfmGjAN4fV;%WI7DG2`r=Y501HY>F4(`NpHpLO->xcI4K!@P>0st*+0(c~+C
zy?ZW!3Fk$p5Efpf+1+yqV3D#xlD}l;ne{onh1{<0Gym7{sgw7O`Zh3Q=7+!d{YUz!
z()eLRi|;odBt^?zloCM)>wh)Ww})0!=GvOVFv3<65GL7UDAlpo<*0ZY90B!QeVu?b
z5GtB*wqWtw_mKx*V7k=QUpom!N5QJbbvr<q_VI@m9zvTZ;SDzSJCL7ZXQQ9=Z6nF(
zk^beqnyosCxa=s4pX>otC6fakCiz(Z6h}a=+vxWCS;WWiRR3~&2ds}&6-NYjSKqJm
z!HyO(QtDnX(oe%)gj#b6&=v3NU?gszI}4TbZ%hmU@Dzr>b)$X4s3Tr|P^^m9TNuC$
znn=ZF3wj2(kFv0*8WeZF(JrHe4U%NAl!*<m7(t8=kdP6I0ra4kr()Y<Tknd17{K?w
z7v{+Vn9<p(*|W6^pE(|Y*|C4S>Am-jP$jo-vIT$YO)P2;FL~Sz=@^e^FAj?^m(J~4
zHaAU;<Gp@TS!?}a(OaAUB|PWVWMGdGF4<>Gdmt24;wKuLXgV(7+RN~D`{l+hYYO=Q
z7%LU>j*Sad*3B5Iym;o7#)(6dNn;!GV}FRY-Tv!oe{U{uc#(Pe_BX2SI5HJfgq#U)
zMOb^EZ7$z}by+{OPussxvzX}LyapJv%(zb;0xOVAlK~YaqmMQv<F;(L-`Qd!&(~MQ
zZCkCyCw*qTJrC@8+xUEf+5vo;tXNXMkb0f`Gg>UF1SX<Qj5fJlUby_jFa=<Mx@Ay!
zIojE-$u9wfmGH!8%VS?arNm$GbvsUlrKIWF8h3jQD8T_9U*ckm(hZJLbQGforhBD)
zh*-{_6r(%6AOL~iN7@a~fN{z43<~Ar)_JD*;H||r3;<<W_@%i-At2x<*H`v&(Z$JN
z!e04hFNNQ4m;J_Q2!z;Moul=n%S1?j-1HP=sN;5ND9`^_d3=3WlL$<J95l-mlO6@z
zmHYFwD+U&>UX@?f&xq}aGX`<%Jyhwnd=;6Zky>z=oM6;P>WrbM#JmWwa4#E}3GM&7
z?QQvO?ntLNt-*~^aY=k>-sM|Sy0b-X#RsO;ps<k8i}TS%ah33fmZRG2<b<A{7N2+W
zn3FFSnolmc<U;PGNj+uMzpGf+Q=O8?7Mlp;c_l^z@Fo|BDb-gJE&?(4f&tMAY17jS
z*>&`&C9o{4_T*JmCMGw-?^ai5hEuf8>)JQ7*UIkb*vB{BPW%kN&Lc&Us^R{O#LG&N
z$YxPkmEO*3?~>cC0RIdJFq6CNeATSm6`PwiE!JnVaUmx5Fd+njIDnwTx@7hKQ5zrh
zLvX@eD%~UKWZu!)69-IX2ic{utQNs#Uqxc>cPcV_YS%NnAQO=6?xf$czZ*S_C(CXy
zqI^;!b`TLZ0d#}%4=MsGWH97w47<>q-X;5_&Zf!97kKGvkdo@@l;zpp@{0y3GFF#7
zmdU*v;#Q9Qw?|G}bQ0!^$!EG~bJk5#Acp*Ae1ZyXX;F-bxH!Rgb6@@)T^(whsl18*
zQVlW$*5JWkK{1%k{==Qi#s1~x&i>Al&8GeMv%|>OHC`nF6%TKWOu$hOEGA_JD%L|D
zX6pPlk71DBObI3q@(v(Hkjz*Z>>46E8wNp0rd1i(wbiZ;4W+24L_{<vamf-zRG!bF
zvcty<^=@ymrV{v=^a<R>@k1jKQ?+<v$fiI~?q`9-OpK(Tu-0`-?b^KVkrAF?^lu0y
zUy0<tfj;w99vMLhM-lYvn%p8&{9HJ$n`I$96oxOB2432`&;r7fpnFw%<AiKV98eR8
zLc<u@T{_?PW>O#t7GxBI%MXTqD|Wo><Kq#T`zJhznh9TUz_K3O)bYjCnRYG(M`4tX
zd-e!Ewgz+yO)7Z_`w{%4tLWEe<@Y~0Wtq(-hKiR(<qbP6B+D|J^Cfck!k{0y>7qb)
zzt;2Y;ICs7sl8@r?^fvO_KJuzN$(iHPy<pBvuwKiB+c{Q_~}NuF*`R^G4v5@UBIu2
zB4#)DA+k_4{ABN7`ETeiGGU`~@X!H&zZPTyRKVxff?_e1ofk!ik!0TLoYzhp=IXME
zv6Uk6<Eo#?hYwS%Ztr5X6A*mb^<%2I^~!VI&C7QYP<K(7oqL_t^*E!V$^dgSPKIIy
zugQMa)5G;m{i}cLJ69)jZI`=b;o^i<?w*7w|Dq?HHgylD76g_^eB+(|EpaMd3h_ew
zQG3#B-Rl_G{?oT6gfW9)T%$*{1vU8Y=SYa(iv`vW8lt*WDa7b;-SzbQ|4kV7tF;C)
zU8MA?Ufy^mvG-SV??piy-r}<9-fshMHJR!lIbi0ROX14=29o0$>k0gb2ve1{&Llvf
z-lRYoYV{1XSVB#@;&Ks$NAutCY?wZYLy8DAOvc!k39748CGm<uI-p6zpjB!4@_&|^
z%N_alUTT^GTi9alrYJHBKr@d5jGArF{q3_y{IlJ%rLRAAx?F2Z6JjxOwogMmTdS{|
zXk@r{xbGHGuVU5qQdl8yG}(M-XPcXgYnyvkw`7nGKWSfV(ZfspGGY0j+VO<oz+p;^
z7lxwQHjOs5zM6iL%i`Me9;0jf*^kDQ022JEj~>7b(~@%U$8qv9rxme4ywA2amrqY7
zZQpTmlM~?2DJDrt5;~mNJZLZ!4%cU*U<K0=C*Jfb$#8N2RR0_0`zfdm!M&_@D_FY4
zZJZ15iK69tI5UW*`B}E4bA5)?%{sYx0|S_J=6Us1^~rNfr0Nl_@YoBo?Osm%ltsc?
ziW>{1i6BU3>8_eW^^A>=ePlL4V9|<RLeDTlQR;4j%RjC5XYTM~2N~d5%uOi|r=WO~
z%K7}C*+06q7r!Kx8I1FVwYG?S-*mbJ8u&aYgR@8dI^qTMa;`8`CXP2zyI3n+Vj@GS
zb7wx(t$Q6os57Vcl6*}$9S=rCIcVzk*S9ZQCj#!h>n*VH{uJiugyw(mZ=N|p3vGMY
zcXoB6+U{(@W0$veS3u0@8cp5AQDg!7p(q4-bacLaGRaLQ?*08s{t<d9=${1NTL!{U
zirPG~cZ2r(b`_l#2WjpV+v~y>FdUqNLhj8ij|Qa0snrjW`_Jrrg1aa{fTf>w6cV&j
zQjN~&?=ur^`yX<vZ?)Uy4A?~{PEr_g6>NW(OH)vRHWxl{YRt?I!Ku<+J4ONcg^3V5
zF5SV<6uPdc-F$#BqK_Z9#6wi4!;#QR6b3jNbEt;grYN~AD-k+9&B}{G2o7RF{6pQZ
z+Gu}qJJr4Z=V|mP=hyw_1T(hSsQmb{5650%qa#^&uBXf+H<D@B2;ZpOSpS2rZK$2y
zu;rm@LECI@g??~W@(d~&CAnG2!uZ@PY{bImvJbyOYmls;`lF#m6K73l0XE}Lxs~n`
zYbzg$PS=V&9OT>5Y1kdXvp0e0faZ9?CeR)F$Di=;Fem4uDJqKO!m(WwND~y=vVt#b
z0^zB+X4}pVIPsdzAOVagIo`F^U)}OoFw5}Rv~5$woSgjKTnuPiTXR!LeRA40y!-MB
z{F*_YzDK0JBtbc3Kl22gFd#Bu!NcvsuAKcTq_64MxmG|`;%C(#&MAy!uNThE#>vJV
z<{s7`#$Vzo3wja_NWKGnkqUL*Vv%~uR~!+*1Y`gV511U9yJGBIZsogZ+D8#+deH-P
z+qc*_T=$g_eDNMavOa}!D7pk7jnuPPC@vNz)buZjo|Zde#;Y!3Zm^mQk>aEKdi1F*
zERj`+I6$$V7jK8PLo=uTHvQ-exz#BDbNDCqY^cY*4TxAQtpg{6keMEjsGQ_ADuN}q
z=hw|GO?3l&5oYdN<Lq0#<IVMEul`!mBA@AF^1o!#<4>PHDI5(vw!u5w;73oq?!7fR
z++Q84n?58TkJXTq2-5L8WzrnU^Tm>dcR43>MTZd-=*XgfwY0ChF#q@TG~9BssPbXD
zJI^iGLf)g|PjwSvYF(aIpTa=zEj`K*8VTe-2(Jpt84Q3T@qW3;=9jm1qxZGKLc`C8
zuKu-MZK_<k&<<?UedDv3tUgJA<ghSuup}her^<8<XA6I`WP0F(@?%|5D{%EqN_3Wg
zR}jRhDv>Dcc2lZ=<;So{kQP&p=z!v;E8;;$KkH)c+;7iPLU#oJv#`?GY`@$h9LzO;
zdm*aar4o}!3DfhK5-EWi?_?R>A3_U7rG%_}XZTlIo}$u+n#?SGz4Hp7<nu^EV&@tG
zh|0Wp07i5Xl`pi;MCOeTm|2lcRQhdSXdJk@_rOUj?x-Ax%yXNSr~QL)D}9PAs(NiU
z<lEIgD?{4XUm1l)vP^RNAL4CWGxj>y{Vg;8kEXK@Yx4d7_DDx>fOL$}e&~>p7>INU
z7?gBNcisbr3`I(0Agz*0sC1W7LmHIPDGg4#o_(L=`2DkgckI};d)IZn-mmi{BLa@E
z+*^4+5Fp~I|5QFFqlsLj@m&`iPyWg&d->ojw=Jf1si8m66}8hCW)S2vWLiAG{hB~b
zx$a&72u#~~si5T!_hM02`qV7LR=+gkD7pTF6^>l|qfD;poVDx{h^)6w-`kYZbD2Au
zIkQ6x>+`1S(GYAv{bc;bcuDd;fVty*{_j-g#gBB=2U4Og;iP%r7EJieDas``=6WZi
zdk08Ck7{`{$)L(zvUA~Z_&D}W6`8a`Plu==Gb7SD38cT1WXPh1UhedAWkS3edK=`H
zHIO!BuphZVkV8z4)r$6i;Bwz=Tj>>x=1?9lls@?25QroifjwpUByS%A6p&?oBT<NN
ze4v<UvHk7~=~6U8U0==1*&%WM77Lw7EhJ-f_h_5oW?eaJi3|ber%X^73n~UR6b&`6
zPZz{acx~@GFVk{g9P=t$YW(WT`bAB;#m|lLvw-rVB+19A&7KUTHPD%Mw|*d7>e#8B
z{yxFz;=OwBM^e<o-SB4MS<vLePZ1|`C79_~BzgOLh9$4#=32#SDKA(Y0{TEKV)4bl
ze^gInFqjP&>8H$=8L|qZI17^kW}?6{{V{9*pLE}tUm;oU<7MnfZ}HJRxfPfaKtyjb
z&675&*9iEDlj^@68Ld_4+x&ii$T!VSz#^Z87Rn6ei~3OPYdbXxWQ35yrHvwH@VlAr
zFjhF^hi|$OGjWdWJdeu&5tze{Hy+bxc@%&+O0kYQAJZNn0TW{5SN+dt-*BY_{Z&I@
zBUJK>%)Z$cPj9!$5mG+kF1zG3C2D=0ZJxlG*iaJuijh2ZCM+cH=*@N3qs7_CHx;>D
z>!!idYDFz-N-!XsY~)X~wau}(s%q=a!P(wk1Q5&wzw@~;yg5!oS=34+u%<FNm`bm$
zNVUAQ9$k2MBIChq?XH1A6=vJjU9Tmgbzba(upY<7RU<>M!q_>@lEK?@g=4-Ug1g%w
zokd%jFmiP7A^%3n+uO<y@7$BBl0V~P0heX_XLc8BbOZK=8n(0#K+q2ot<4M9|K=`$
zGXnUBd*S}C!NI}Z^Kw&H%s%6xtFj6EeSLL?l|db95Z96rgWG_OXkm85_V2K8jVA9d
z{ij2>_q@N-(~a<Dl4WL9gGc#JAtan80>Mo08)+o3F25eIeW-5}+}QH;;QLc_*XS%$
z8s3Z0R|9x#VQK~Z0ThKGW6BBQ2gDMyoy3!B$`kh`S!nEC_Z>>j?YmepF$kjL=B2yz
z66ykCgsFmIn<&x!y(i@Nygz*$XEXhK364%c6r?wa46a5JXWmnH>{FDUf^UDFDuSnO
zjqIuM+oObkaWg5=Lv2nTXM~Y2kQyd!*^*Wj*a4A=nqaz<;9$D_ytcN|iy*}+Tmm|N
z>+vfv`}?Fknn$4W2izIG7NBx>+%3@6*_;;aF_kcV!Fu=XmI<`sB$ZuoIA2DZ+a3M%
zKZaC}wum?P$k07j%nZIgiNEd!yk7{oX=Gf#&U`g2`z1q3*-cZsp?OQY^N!#TRzu_}
zgA$-kMC7DJys5vR`9a@k^G^=Keqf40P-i^#cgu=hm$E88Q}xMVy$f7~fkW2TK2KE|
zT+D6gb%a!6g!8Jozoq-%ws0PL*ff!^s;Z*Wy5-67XMf`1I)yDceFnE#$f|1+jOqs6
zcX2dyz_PBKC476cSz(j^Tin1&Zsumo7l2l7CN!8OQ?q%K>Y3`z>DgTKMf>I2Fx~#}
z`rnD;>A_Fp*LzEm4<0{SIC1)&@o14fQnhV!v}IZ7FZE`4`1Scw*zazMoM-t27X0O!
z_syBAO}^I$7ss-jq$I@~v9nE`)~xn@H4~K`^^GU@JXhV0-N31u#9Vh7qDla=2ev@i
z@8PIz*AwSYx70LVjkx^>Twh<^T-)57x8F>+t0?7oVW<kP&i`BRDqWYaVqRP!>7h37
zqa9n2K_bD43hcTm_!R5!J6>652WEzmKfai50Q4ZJHq=fBPh-S9EOQjYmnp@Y`rWMG
z4@5+{l2|6Fu5qG(95T(z+PFuFH3g!QJ`k;&kalEx4w_NMhxdRaNS`rR>ZNEP^8U(D
zzlaMP>%%_xb}J%gLFVPiXub#A^MHtK)#=SMDT%YTwx_GW4#FL_|27%Eq{B^ChxFew
zE_Hl0gFcjVuGAwyh#0%zhN#R`5e5G#J3`z4F&Ttf{YK&s`Qm1IApYC^1i-d1Tq9`a
zQMP#B_r{$4g{w{RoAaYJKV0^Me}&;5?^Ybv1K(b~o1*Y-ZBbDcC5%_FF**HC(Jr5z
z%T<jy_i?w_nhBG8lFVicrzf&{QL@+@>&(o=RO$Rt*Gwpd$8*9z*x(ltI{FvyL~IA`
z05Pb8uM|5c=j$^;QTlZ}I}GTXGj)Y(q@OjaKv@KY_%~xyvt0Ok)K;-XPU_*p+OHj+
zD$J!veNg`4-9oRvgwBDqDEn)5I$8;Y_jOEi`m1CFk3amm21wI^!6-=KMf`aC_1gb(
zqTq$dhMRrv>*EF0^I;pWxqk>c+G-ZR^RQXgP+_&j!m7U@2{23-fEo?d9(Xc$kCMqR
z?X%siIuOHOqO*HD74=IyWi*4APuDhj*J`3pPv`&fN?8&1MdDS9rFU8M%AAyQn-l~7
z`>S>Kjtm)%8Nh<}6!imZTI0w_!{etnaO0TiK@avJeO^{v!Uw_o<Q;F!I}p7~nLb08
z-+74zuU;i2JVnz2l)E|Z7I&uB>lq|<W+mb^C>ElNDhEvc`N5XoFD`ahdgFnc6V)3+
zsb|yXdW|H;E#AZjHx)VT_*g@mB$m(hr#!J<%|-Cq<s~92cz|0q?ER5Q+S=u)V~5(H
zMhJT_(Z>wf37@2nnw?=~lp)_b2Pl>`%k(!{PfyuBv_w77^9qE}07*VtaKAfobZW)(
zX>~O@=ANzJo`}-!Qcjk})+VY}$fQ^_maUY+mZuEqr86R<s_orKmXP%drUlb82WJ_>
zjT<IRI(E+v2k{?bYA^7|n=8GwVotw5iiy-R@_v+kxMQLxT8L2ho<<VdLx?{k^Fc)X
zopG_IiVlERp|kRGvceyp#iH+Si{HzLJ+bN`lo9nAdkC_(%;rxBA9kQC@@dC>FM?D$
zkeXnFo)qaI6a$~V6d5$}@J$}y+`CY{`mMaTzW`kBIc#edwLPnhr1l?l<$dG?br?i*
zoXETDNdyNn2ftk8r6rn;2BS;SBCj%*wU_AV4}ix13q+mw95YU}L17&@r~<P#GPJ-c
z#94hclL%Yvu7T}~c6*ds8sP)dH&cNTv~BVz<!v=38R_U01?(;3(`GV-goKa@F_2Gz
zoOg<=28^w9qK+6rTCRXR+cEMtJpQrmqxXmDOZr#OLAOsE{*gD=0XIJC`k}3Wa%8Jj
zRYe)Dg=PMw<G)NkDBkBl?y)SQpqT;T$R`}*b9}gWaO|br#W@#wM;Q1-v>~(ir^qMF
zE3A~9uDzw^e>b32H>dma7n>~syPo@TxZR*n--#Yl2FoJ){^IGvW@W|wq{k)oA2mGR
zZj=p=Y>c{I+Z?PKuGiZ)=p6wh1L$uH-jxnE)Y!VfcaqrU$II%mC9E2|RuUK@a^Ikf
ztRm!x>>ARtpVLjfAfyHQX)lWoOhP}cU0of|1#@5j+t^o)x;Y#VrrYrU9n2ry)OU3?
zcXU^*o_0wF>O-B`NUQ+S{xnf0qp0hjpW%Gmd`(Zs#YSeos!bs*QOpN89Lt`4G7=!s
zIG(}g7`pNilriA`*J{K%p_9@7@q^NNxD`LJIP<SU-6ds4Ltkx9(VgC%j4_Yw;NZ+w
zM}egr=K01=XZmwVNYO;gh~3LmU6)_%NS@~&2?&@V0?uRmNHy1R1Alz}H>5Fw?61r%
zhq4ahzSN{Sw1(Boj~@U(m=yl&L^Hl$vmQ=RE?9K6?$!%b+d3`3Y$NwV?#%p%+P>g^
zI%=cX8WhH9_QZMl_7~FGCtnaEA>A`wT->U9<{*Ux`~E2W1l#@3+vV<&nMtEw_1j02
z&tD@6x#Plh)y*1FZRN(rJvVn+#Ox1=Zy<ID4H*IFb)P|@?0z6q9SxS|b&%1bm7<fD
z(MzKdCz~eADDUbw3GmnLm$z5n-W(t`Rayhta;;VdH2`FYh6gXaEF`#vs1h7&?V)z&
zQx1^#;7{4H7c!p1lXN<T+CnSx4;Wy&qu7rwwqJAT%B7%dcL#x<gkTVHLu0Yv(=j><
z$;rgbh*O;gOJF(-0G>_5V~Q5LH=mam$IjNN+X%Hivyhtx`)7XbWlaTx|1cN@tHaPe
zK1v+3aK0}uQ0sFKMYTdgWeS>mE$2P?h(UP_%{c}?C0-#w0mDt#+bc)jz?Bv#>hfP@
zrTxlH;qK78Rrxd-wN~uxN*}}G!^!{deZxy!+yEn$#uLW@)_g>(tb1O?y(*=lpw_Q+
z8amRbk0~&z)GWKo;`7a54crWFd;8F~$c#@$_Q<O5`k}ZfKdC<VkNLUf5P6E`qd{@P
zQGa(@+(xDKIffHUfLLyf2ly+_=Fo9-dqrKJ#dF{MlOA3R5$WR^O?_{;7U54w5t%Cg
zXzR@L9&~PKExcxaA;4y9L?B$x&%I&Hf3j0swQ%98>-z7Llz5tPwpVVfH%g&f!bLtr
zI898>uI<e|9oOf<KGZFeLbC6;hVREZuRG5N8s8zZtzPyc{>=bbbIi`klJzPGe5Ah(
zf(ymK<&n`W#9fo0Kb7H(lZoK=FtSwMi844P%<-O4hjUPPaJUM0Ui;-(*c)||@^nf*
zYvmOKdmyRUr0lVl)g$xM5T%}#RtdY)q0QA9*24Np_P>m`at!|nzW?Ck?=kgkdM%vo
zd|%|>fBN2chS?7R{j4jA2OI)md-AB)e`nM%UajUx#nK@5_i%N4tz{E*9DnT!@sjX~
zDo1kBM1=>z<@mR$wylnq6mLh)CtVaMD{J_jlR+9)KqwWoiMN`7lAK2<ej#szb4KB%
zDf(o^2_T(ZjuA-^^@%7GQhPm;!M+mF&54cakJUUU-36RYmNKj;HLyFsKn;c~s1~{E
zrLtI}WX^YwYb9>$hPU}HKx`@B9^Z0*@(afIdc$HFKS~b`CZWPnOp!1V+b%!v1cM+F
zBB_feKZ8%|@pkYeq~U6b|JB(B-A3bl{swUQ;qve1+R;}rOYe98h)D@BzwnF#?Z-+x
z^)w(-6ky>3aqj%$1?LNpk{<l)ypC?funWDtB}Cnv%Od9hKUzvEf5{j5t$yR^%BfX(
zA;-x7!<1b3`6VhD2{zF!!(y+smhd3xj3O$I-<7HsJ03xxMA~3B#Pgp$vmuz@!{i?6
z9rtGzA6wvQ1a8@6W34Pf&u}v!{X(<TKubp8H>aKL<MGuFzVJJOaKkt$#b~OF!oO$o
zks%Qw5rlJWL)OOT_0nEQ-eV7s;NJNf9<^NeT#LV<d~rNzNvs(|dOZ({f-ZUIa{@~O
z5iHI0LHK>i(x-NnXRODUx-IWHH#e)i%9Hd{9==Lfc*~`h!t)#ju}fLtO(GL`t@7V7
zkoyC&N3*DkRi9*#8q8Ou`6m@E?h_Bz4MWhQV2$kKW=ZjWnfFRe2K#cb<609#;ZOWu
z<@%(9Zl@XxJLRFX8aZ@`(GOpyTbwQ}T?c%ws-iPsmK6=$V;df>0-X{-Ow$+_;%*62
zRStJL*J9_#C4MTL!0Bl;ED3*kdT;;`CRQ*6){AXeX^81Tcl18Wi0t8ekCrkOQOip;
zPGwL~-ojhe_G~qL0ix}{7-9xRP?M&d&B$$H5NEXH*g0)xD*t#GBxWd86oPB2n;L7O
zL3XvCTwk}fh<_>{>K)(oDG#*3@XyxtoAJy12JxWO;~*2epNjH+3%!1BMhygvC})DR
z)H5rLHN9Q#y(9uBJgF9(1G%0&(4KTLs`s)hma*Xw$&qp3a;4yrb}*TmK%=W%u?Ixl
zvoF-Bk0h!C>^Yl~a$|^7I>OS*C?F=~h8jK<SICt&{>aV{vEZPT--x2IMuDX#=+d~2
z(0A)E6Wt?sUqw%5sYJAp>;qlCD%|;b3wh7x^KBODyyled-xryqN)z`Zi~Bps%0zMR
zyQeqy8;a-sy{@+=>CY8d+JvX$xK>BLh>xVv6U8b}<^$neZW>6D9WH|LT)Yb&ljv4{
zcB}P#3ZjNy?^fftI9dA|dH(Mo;nCL?5%o+gI|%1{HX3AOj=1<rd<c9>CL?}d4WXjn
zIN2fUu3v0grAA6t(((yTbD$LH`EV2z(X1$Sp>+$dpZM;pp_TXxI_^DgRV9w$l9eGh
z6)tzn=0Lf)PMgBdPv2%CLmY>~-^uVDr9$3tlb!AC9X>MGrD*GNlI%E*e(+$V?ID{G
z9|-cs%`2BAH7G+GgaYD;8E2Sv_tu5!HR$b%MOe;0({>(QOBblIviL1ujtW>NNL0}N
zFp+bkqIkP^>l;(ve+IHBzaLt%J1^bkA2U1=1jG8&vJ&KjdMI=T%;NZoDay1Jut)31
z5lV)t*Ow)VMqXjzCrIr<r-$E8rZIt`Rj69WEn-j0;k%dwj|%CMl|o9R2awAErRx0A
zx(mu=t<X%TV#Lea%X5ph-^U-^lqPY3%xIZ;foX$o5J}2#1Pl{4M)DufEP@PaYQ2|h
zgqLZy^nN8V1s7ms^SV1j$tYt?g8T4RhX|}gaqqsJz2l1<Y0ih*Xy42(Efy35=mqiD
zbqX9S3YtQj1xaKbkc>F0ya^t0s+?%hh?><Feu8fq&TWuWo&NqIt9j>J-~1QH_Kf>m
zPvW606Y2jbIccS!`enXA0yutDBea3N2bQ891k1wF>%7OnH5~0M7nBfpI1Qyc?&q!^
z`d<CpJo|^ks2`uVJ-fTj#Npa#4@>M&TQYKFK6cN|{PH^A+Uiq{kHE{1(2F)huI2%6
zU!2FQy8;2+>Xo3=OB@4KSD21Y&NR1VU@^r=l6URzW%I$n!p%zn8E8)R4U0FUdr*C{
zhOxOmoW41|ShzVJ9K>(e)jA$M=%{_oiQ74t<#Mv)e;=?&2@-xVzz273JF2X(q-;AG
zzP-1i%(<nq_)zLwstZAJ@B<QkN8)yyf)U-xMlan-{fQi5rKw(?E*)QcDPsGSj?S+B
zMm4rjA&=5a8)1OqgANAB3|S+W*9sLmi=X?tZ}lTtP#Oyg-F#@nC-S!Saz0eBR=(F(
zLm$tY+(|bV9^=bH{i$EZZ}2p5rcrCflLaWMyp1yNJsvn=cdWr@^(>P#oOmtW$J}B`
zx#kgDtvxU+`?{7Y-COyKruj8GUu}X&K%<~bUj)5#RtxY*>Q%&t<}oC|dlV^{mrS?_
zzvtF=K?2tm?TVY4x5V9@rX_Gk__~^p2NL3fx*??Axby2Ksqb@V#xYzy+|CL{F^i&?
zb1OrCYC8v)fv1I90cFZCnQuK}<&;p>oUKMk^iOIzvPVmv+g{Oz#PSq8?-gMDO#MAQ
zsP81xY_(;_^H`-7rcl?bE1yiGjcnDMvth!BMOQsWagyW>(7O_mNo!hGlD#r?HMEVF
zW@b!)A8ryFl?0n>mSX9L|H<P>8;Tg7WORKo`RVD<F&%dwgb!(l=9*~lUx)=ErzNBO
z?or<hxJ#>WbAmt)hPSBL%zI(Dfy<@i<0T2beP+{bPieVd6<pWoCmKqTq3^anioVCo
zaAJzC7H)uP!jd>y$Wv*zo_}IM^O^txZkT^;BOC?pM8OH>)((><3;SB=NZ5h}&M)pB
zb^qAqjyqd4T12C(OogS7hC~9S#zC1+*n){A)5)DA%LUWP#S_A@!}aUKV;>dYR3zxm
z-eiUM7am%`j6<sE#=DV)Uxkk1wX!?nqjhH>u>Y?59`&p2%VS}rUvA#Cq8f-^J}CIn
zH){e?%zgtJ>h8|ckllD5*nIEB3*W<JQ^~d-tHH9`c}pWsVsg^gZGza(Wnvj}(61Zs
zh{zOB>J9t4Vk}Xx=%Ld8%44=*SG*GbQD#szI9-4SQ_dZuT7A<Byp`Ug@j5YYOmJ--
zKrqElb}vaxrWKg(57kU}1t?~GRA(sWPz{z|pTgB41JyNcQQbXS!h%Tl8izuJj2L@I
zdC-@TeZS4M=A_G$z<;zdEDK!wpLpt0gLTa~vD5}GGQ1?3hB-L}=%*0lq)+!fJ7Q$t
z|0BofRch*olR}p^L#RKRe9CrTfOUldRJ(SZ>ekuELyx*gDY95*HC^sC+1xVy17S2I
zvPg*|nKN!qbpkudqaL`9y&;Jn$-y#6YMff^l87*>WnzELe$z_;Hy!8oD=#pYt{igA
ziPD3*-!e%F$@Xw0MK@+ekA9*S1OgnjA3Q4`*BYlnv25CJ>)e$fg(apn)}W_>DZlDe
zdJ!0?1FD@K6Qb^%QN|%nY~Yx;ErT%`ZPf}`f^=G}tp%FP*{}x~AgMSdcOz|GAsxQ@
zFbD5M6eEb5B9X^8^`!=>m-Eb{@QiAf4!VMSH1|OqwZ6|iy0Nt@e!ufnR0+Xdgc6rw
z{KXw<Js(=RrNj8|Kj2+ToW+0?a~ACM?CdI^;1S3NZg$TGx7WTFTifTQ{uh5ouho(-
zM!UyEBeSIJ`fL~=?^sX)LPI0BXNK%{Tg%FE*XHDGM9rHl(#+1kbD4?!XlRL_L39f*
zaNTorOUFUi2P@sX-9ww}LjD!;Y#dglW`^zvZ^Bsh04d&Tqp_HC9RO!HPEpUM!iib8
zpd;3A(#SP=-M5+)ew>C)z4&p@@7W8rkvPz=A@B<ia@t}lm=sVn$+w2Cyz~9g2SlAR
zuY5o$L&H2^9gnaXjznh3;rDb#UimEYE%Va*Kx`fu{po<Z(z^e;-DJTOCP(VR$8!gZ
z4UISVqAe~%8w>l%1K1LLWBu_?lT!Oho!h>TzW}KmjU1b_8mvp^5fr|RF!tw|^3(?F
z!y3oF!w#|ztqxwSfz2|l2FMmezYQ~A{4ExLefQl-{915pTk^BU?ou$K;KeJcie8IP
zeXkiOom(+zR?Y+#q@->NlvUeD?M~D1nMO~M{<sE7OX7SupiaAi7bYp1T8mjk72xb6
zNlYn?{L0GU<j{mZ$eAGMU61vXKU{_7%2VpKCc>3=%9T{S2p_bJHm^Ejz<--4ha<k`
zjn6IZzf6+sWi&3WjzS^S?EDBl$1oQfIzHuCg%i&0XDTUKv92M40GVhPPsg|&hikRF
zebvi;tsX0YDzS(Z1ZOOyW@g>u`ql5c;@O?!gwjDdYCKmHb$T}-a~rCLAe0$#LB9Lr
zd_@OlMd0=}*|vX9vPX86LvE(*OJMRpDD$#0JsM#jzF}JW7qGV$FRQc5=ucYGdFQ$v
zH1b@|klt4f<RQB?6<32B8!HPyMJ)#%{!D37y$2Ybc}KnqFm96{Ml|s|Qa{5S4+Bx>
zHaFMPCF0`y``d&sB`j|DMIVg(pNxg?GJ#G(*TV;|uMf7KjJl|)tvlMI#d*OXdn6F*
z;bcQ*38>gTE0&ZpTOP=5x5@~mDQ=bdsMC<(a3JJ#=`bWby^LHFBI21FP%Pytip`V*
zG@w7Sw#{_=??{6VwqKCmnq<iYL59Xd#QLwbNo!)o?#209{kkxudLef7NseoEkHHb~
zbb4Gtv(ZYdvtZvgL|xPQZfW0z*j@N*271h}R8^VdK*{r2?xpp4PoZxevAr?8wj=RA
zq-*5y<XH~muF9gL_a}S5k$oVQro|LW0O~(a(=U(8GVq%d?rxspp;2Q~lFRZ&uMziL
zy*H)I^s+q42s#Ew2)oc13J?#ICeLST^FREAyGeV++jlw;HE>n$<88~EqtM9keec{R
zj<)|iH||~iA^LiU|52+)OY4bvpu*Jo=}P4O);uJ(lYf>6ts%+rL5q}^DY{sv0@F52
zB6I8O9b0tIVGrhPmc$8A`q|svgR;x*$TCK4nax^@iab~R85{s8ac`UHiRxQz$8FLF
zxhaRyLM0^LX@AUsqr1L66DH^HTZ6JRKBVqI<Yc<$(c$NEC61<|F8=*)NmiaWPJC8B
zw35$dw_b6CJYC~%cUrgvu7_`Ya|FkCzdj#qs+l)kJFvM{2c-gMf9sS~BhPj*!vqfr
zPP{#;O$0Yp^Qp|`j6^>f(^E{Y<M6pN>nuR=*5VXS6}YPTJXn>lQYGrox5RF`{7?~^
z!ovFHv9O3+3=(wa%s@)Qgfv~l;g|Nh&wz^s!r+Gyw>S6C-ev|$?^f}__nrw(l6ZYd
zW$ZUc$jDHrHA%QPGsbqFBoP?gtbs|XRXF7dssr1fvUtD-f$ug#^Ty^nhsG@PS>viD
zBr@x186O<T3^sN|7lKGf=}*6IE#%LuDpwg*RVdU!7_Pe&!$9%BK5TZwMo1=Y7N*3D
zW^;e)$CZ3Ine-f=nusasHe0t<AMb$2%-j#R$*7pJ@UUAS-+YlW9SG2TyDFy<5YbO?
ziMyo%Cne7V6lm!@$H8!cwjZBp`CSOS4G=cWQrTETV+(d!6nKX|4TOZ;hWEeaKLrhs
z2|iCBOy$M8&y0&@_i{}=cezXi3$cY;{xY|JzVjrFVz;>cpw_6-zGWqMa08YfcG{E_
zN2+Hk^7gOT<v^#}3!vtf|0^FY7-=tcI=SXjVP{&`xRvfw+WZn|yQyX>WAN|!`TKIT
zM1C?s4KzG1!);Vms}kG07HN5R?()<AK?ARXXauAK+eYq!)S6j^!GknFOB#1edgCvA
zfbeh<5bqf6{R02VAsaF$F3JqXNenZB_~YqjFSom}hBk}^SUuMI;x)Tc1zc_5$F~WT
z(_x#u{5%^1uh#O3?33U@-0zX&hUAEo@%zZQqAA5;smZC`ueI&hSNWFv-k2jcDwSX1
z=8uEMjWiv9baf@3y-L3lL_!R)B^KOX$;sUNz-4{%%`R|#(#y^5@iwe<&dg>ZsgD5O
zuCurUudM(KOu)9~PHoMW7^b8yqga*FO83p|*7~DV>x-wro=o=V!K@J3_t>g}M=G9Q
zE^2qbwqF{Zq$=Tesg9qyXdoH7eiA|*W<So?Uj{q;BaFu@TR#=2mwmEC+`aa(J_tjA
zNGMHgpYH7OX1=yU-O?R(KEG_Mxo!4aU#U;IpVfg&BtF8mwQ(lJzz!8Enz+=uS@+M=
z4aWv5H^?;Vjm5lJB7vDjuFJnqAjuS}hAVrYpOl=~<Y!=BiP5Bw*d`ve-yB6Mo#%{_
ze<ZR6{qZ%wz+eXL!kScvrI`giw{T6LD>z|yh#83e3M3U9<J`P0zdqYYppECnkS~*@
z0zWraWNss1a5<ZLC)|vmd4|)2_ZE5zg#XL4_4VTr#a!=&Et&9o!HDlqBm7x9JZvFE
zlme*dXMyKG|6`+i1Ckva)awHke9|e4Yt$x#1>5k1ZIe4hC&>Rn#m45d4+I##0R6MM
zIV>Yo`J`heq<XK-t}*SiD}=HbZg6lL7#!dJ-<w+2XTaa>5QdL}o(*a6VU?J?^sSH~
zrk0lBHuckk?$CbdOe$CJTyzozYcM_ck{9y!J&`1RrD;EvpUL%dgs1=2X$hU`Ree5i
zz7MD@<aQ@!dYZ$lf8JV;(T?wi64iO`RxqgN7UOW)7BgXPwG-e!21BE>2AGT$k}vlg
zYi#nwhP^#^OFJ@_au&vacDIwxIhHi~=Mul`9dz0SE;i0XPJH|`<EFg`OYZM6{T_<|
z*i?bEY*TlhPal%8W{-KfN`S<PS5<9)&AdY_^`5R-NZoeno!h`%(?NmDBYPs-V7@dk
zr8t)9b0s+av#dHF7LrD!CXk%RgC2QfnolM`{{3z@kxz<n6`wV!GeVdDKf#av^6Td^
zfod(3R#$9~snMJ2S$$3)qr>5jyhN@$bfB!icYKLo=KfI-`Pde!VT#NH+>t(7i4{Dg
z`lhYT&8<L#@LS9v8Sm#{&w(C%#2@RDGYJC*`aB~oySB)N_KODWhN%YF#RnC;2`p=>
z?=g^*j6~Fwo%QgUnh=2|s3#2!I2R=h*h1vFc;0(bElqYX6BBc$2;~nK@=+SsdTx7H
z4c_1}H`SG@SC@wo;Yu_)KkI`}{`E4RgKuyPn5s1r4!~-jDY7x3*6RgtYBt>D@5*Mm
zHYrM|XY<F;8V?k?p5)U~#yt1N_SUo0!zBXd`EX^Jn+u3)xELpRG4`CD3xdoE7Hw!|
z!un>18JvSy95bjixw6?ORhiZu!6>%()9T^eiO3(uQ2B&NCw<L5!&~(C7bTQ(7qjQ#
zMhilpfSPeM^n{CyeAS>QHugoR@RZ3|=aV_ft5kb4OsahnGDTO@=LCw~+RLk_bb8IL
zsRmxm82@biLL!akHd|6QimZ_yQxUkI8z$cM=;`sDeb5U1t6vV`%3svrko7xf0{%RF
zKDxu)q&~f?SIzP4uUGlsZItOxDupeW1?H7&16djz_gopNX&2GN7M-L><V4KNiOtLN
zHG*O6IBe6pKbm9H<^w(8T{x^lrPTvP6502ZJFkyls3-d5t@*q&kPeBoWbzUJ_Lcp1
z{QT&n5n*g?z2|Luc-BU2ql3p5?UEERb63U5{4sTLTcxG<+^$D8D-REasa=M4w+G@Q
zVB((T?$mMUhLw=P_ENt1%0xHbmOH`@;>Emy5kExdm`H-x5iIM~pCr;K75)hP?nwT-
z6D@B7^C>BWMRGb<RgJ#f(Pr3TGM2AH^l8Mt&#Unxi2wSzSXS^IFQUIaO)|;)j8?*7
zMbJVb<1$owWkrT#7J4U>2Pr0{eP)k<22xw^Ey0f_aeu7zAYna9!|~%^uJHt#@N|8k
zN0Y;2<R(va7pP!;HI?FJxM_Ye8*jn8Pfa)HjZCBSbiMaQO*9*{v?SP%<(nVTIxRm8
z$_7!^EQ_~au6tnu8^aq2i@de<UT?e4G0^)TabBLd1m9cd1!Dcivkp6s<nD+E4?&|K
z(GjJ07RyTAsR1XW>@{vZqn;2c0g=2T%3xo5h#Jp3XfXxn|K*ti3OVce-OZAGmFs^c
zgaBkCKdC+*1N<R@0pd94CBB`zA#b4FQL&3*LmK5N1nwUh2GxB(9+P^%xSxUMrG+1@
zk3{={Vp<aS`V%R2F6UxKIT}^RWQq7rp=5OR(skA+*B&S)&-;GC{uDD>r$7hgvMK<v
zV=@ppK(Fe89@1L%-;F^JB-lxTKJMB+hLHB@IXo5WcqW8p{~SEWW+<;8Fbhh6Nr9=y
z5sax|58-ho?j$XY!X^nPcx(laU)x;!ig~t4co>Pjjw{OO4p5ThO*<m9s;JM=WPNDG
z-vVQ#q#AE?%hIhdDJkEAj^2Zz6F+MF5}qV-T?!3t*k{q6&$QB@q>tU0iZg#XQD(>x
z$BUATJNYOLvy%$-$pn@RS$}QXUIc5GQ3PP+a6BR@B)n4TW4_h?n(#YT+YN&0Nt$_l
zNZ<?nb(PB%j+wUHumeQ@?LPL`n=jMYrxPOol)T6aa(my5%|t$UfoiyyFIe^5jMY(K
zxib&uN0`~0H**&?AxiO~iJw3JT3J?00WR}vQ3XKE!sCY_^0}wpHyy*n`<S9&Irc3h
z7$dfQ+-%zCR=}I&H!~bylauQp9%3HF8h}>&EJN+@!VhB64>gr<NaRT~d5P|Zghk1o
z1bPftd3)~<b~_!ZI<vAv2ggvL(e{k<hcOhXDC5jm+8~7#qP@*TS7`YZmX~}uEuL+d
zCyo=DU?RmM81P2c|76PnTS3nKbNuZ3^5F1#rTyyQbUc1#MOcVwZpQMH6%R~HA28)z
zK9KMM|Kp(~mBp{uSn1|knm*bsXG+qz8*%DqnfHoA^W6O{MK$j8nxrNePHu(`kQPk)
zOem{o^vMm{2^teeh3`1T!z3%BBn%WyM{|?!Mocd4VJp(@clo)noth>FA4Ogz_=%n5
zmM1`01jzq^|5ZEQ@NI1h#M<!`0gvtP))zzFxl%`amy!xHx6WF(%C-pj_ceOd`*&R0
zo0mM{Vd3N-q{q!lx|g1APgpKkjL*Y}trcqE@^{`-X8v}@jZKreJFQCj+WuDjJh(Q9
zzqq(4rquQ7CgSG#<AMCOA0DYycw?dKM>``(!16;j$8L+WGpC06`}?;v0<D${c>$5}
zZIUra$v<UCveIlF8R>fGrY{mYYbH*{rV$GVB+{5WwNJ{_vAo{o6i76cUaE;XsX2V^
z8jeQkISKdeY%!J356Mo8M3YXpr2bk^Z>gK$^&w$_8be;EQ22k_M@HvJK3>mj8|#tR
z9!bHG&Ptw03~01}yh>^Z+Slf-lCS@0_`SRZ`sL<=&ZMRBN<?N^(zcNtr|;0RD@8A5
z#`dgpWqi~!77VP^cCV|;P+|$RA(Ymbry*IziX<i;?PSPCf*lxX>tR_|luocDOgHCq
z|8K_uoTnx;TwT}zXB*ywPPoTM62yrxn25gqDwjLDc(c3x4RG%N6U=wmWA}MeJYRQd
z_K#ndcFvF)O~HkVxQ)xB<UXaB6~Rp+v_8LQoIHoC_V;d7<^BLegPom{UlpGjpvMg}
zsC-jR+|8R$yey~l?y7;hjl$~VcsTgU$I;In5dz{WLMK~G>3Apzp|6hT>%|9zDJ?{{
zgkNrU5A|9q0SA~Pr$(!;UHKsb@WY}6mF)2r&ad=&yQz|Ztb|jeqY39~PlMYXWzAo}
z^yOt1?lfAAn(ohfap^1MJIx~;T)lh-;ifV#Yv~4;sVDkHovGzoyj_uB{2@UdzF04v
zxW}?|+%dnDe&_A}+!J5h>mIY!#}cJrzrg#G(2>4C3I~_|52K(NNWZ>M01eek+hPlX
z(ZP0aoejd~{P-m}Asbrm;N{;Ni>t3A1uectV_V(5ab0josvHeMg6Cmb5eIEATvmzD
zK%VBuCi%ZLJ``|9Q5MWkMTJ_<FNp19x;Guo&a`pG;uyle>Dt=YW3ZqfO0r|;Cwk%N
z9B505VyPnHldryIOMRT9`Pn*uUYvIReCc}qY0^e$dBLZX*KlxvfZ6d++GUn`GtX{A
z<AedT1f8bR<5h`OQcHwA;e6YI{iZGXIFK0TsFsG<SzmmKc4q?hP=mQ_9o#r*q4(MJ
zTBts`s`~x*qw}zNk9yH-nQdT@mY4z`0cJ-*JgV#_LH$d6m2<?}(z%rAJGR283q>LL
zBd!MT_s2h+FIADYiPV7(#lM!QewsbTn$IlH$j7ZzN58~=;gtN6F1+=J$hn_3&>t5D
zv&$_mzFIqy?Q2(3b+AqCC)l9G_X^#uG+wS&H;P$4rtuHAVKyx(2s2f^JaeFu1*$Rs
zg9JHQo=?renbWk<c{W54UP~mcpFDZ-kdEJa-`sR%g$MNb&Vx)OG6h9x67plO>A_Cx
z=5;B1(4)OMBEm^_e){z1=hcN9xW2cQ?9iV|o)f$Evd0<$Jwum~+j`^wM0;D?Hd-+H
z_CL&+PH9J$M!1U>Mfmxaz`|en3*ml>oQ4~XdE0~qxw+!mdapx+vbZ%tyHHW*x3Y`G
zmx1mGJ`DW<NiAUc+4{V$a&T<2bRw*wFzhMgv-#<g`lCVIS975P-koZY0RAHU;vHbe
z+Pp$$qj$|U!B}1P`@7|SMKGd4bvRIkQE0v|oazyI>EbO~{iX7Y;#YFVCM99KCQm=Q
z>=Y;OQ4qO{j>}A%#62*7u}rT35$(V#F(!cgE;>POI`07Z9gSjaTt63wJ?V>8W^YMQ
zgLfG_#UCPs6z)JVMe$b$Oes*ParoA)@jb-4?K-Mkn&@O|RZ)yOCs+HCZ;e>LEvxF|
zS>`&)Qn!_1bU?CG+U)7-E%H^T48f>bj?98`2IsbSSsxU5d8j`9#;v`dnit13a_Oqc
zesdz~aLWS*&CsEKmi6D~=4OX1ya|~^nH2sIG`<_LurD{4GN$KJR8+gSbv<-nS$`Ug
zR1yI}80>W2ln9Jsr}w$6LPZP%2cfpk0@So}viFCdZ?uT==)El`bKQ1syH;aS20@Sl
z;@_pj<DLq^I-hN>>vV-KSePb*1e`O{{3(gm!2liAkC9(GbJg2Y!7evD+dP>FRPjLX
z_~G>-VXnEDgj;o%W6v-`=qW2pr*B?y0#%X4Fm-~SaQyh@dH^|cUJ>%y6Bj18&*iqo
zzRrYX4RCj-kdMCz<dG4P#s68M4J!F2M2r|iF?3Oa_}WZ(+Fzt(0yL^Gw^xDh%|Xio
zKq8J7e3A(BpXMTEVIn2%>@H!kvNmWgx5(7ee5moRZ$I*1ZB{%{yfb_L@GYzF_?2*~
z!B^O>G`vQ%h`LC$^-yX{MhupN7*zawy*P=BB=wcHFnse`0lWN{nAfrFfjd8RyKHrV
z<+B9iM2zulY5d7|TzLi0Xn?$5F5O1B!Q!Upc*)>-gt>SHdm7Q%+ax)!sv8->tGwxU
zlRrE}n>(+xI_bDcUm3ru2%X;ZBC3xF{#;TvRegJ^-nn@38MIMbjfwPW^KZ7u_uUDY
zWIRbGuu?``;C|xs#Ct|by*t~z;dWVk_tMe|?<>+5<!!Bpb)KkS;Z!GGUys^qJV$@>
z`I!=C!gye)vaeukF^uP2Q}bEvcAqb43WcYj(?;`htOrdEjwPVBl5G!bmih94PoxK>
z-ueNa>7?QE^U}-99rT>m?KZwDcXh#cN}3f)bIeKSE4&WX-tsZWIbpv2BjE?Qw>Bf1
zZhUUKhgI{qy=#9&1g|!hmgzgCnPj;Lle${tW>Q1)T!fueIA}M<{mkEOI$2K7zY7ee
zk?Vo*z=>~3du)&4n4I$lgzbNuCs#|mgNcHoRYv8l7w@qVXM#M436r6X&wP5d9bMXc
z##&{OQe<*2vsZ;&iR8g@gT8>Ek_ZPF2{TBaYyDI=Z~5fNPOM#8jwa`?{!Wn_L9*??
z28FUxD83Wc-A>h);Ck@{Y}=oEU;TaYTz8fN^QOpTWKRo-G!2H2kmpDuq>Xye_e(Kx
z=srY`$(=6^;nT%UCIN++taLFO^jbuR&V_1!YPsl2HUp&d_p>|k7r^=2)tZx=o*YRH
zDBfq5p}75PPTT%bWL#d5+USUzP5%59C+9E8A5*)VS6##!k4v(%8_^t4Uf0(K*5f^N
zXi|tJGWLCNVU`|0YMug$ou%RQd+=sEk`jF6wN9RM*9l4R3v1{KhzL-<q?Cp$kjb-~
zzJINcP9ld$Ja*Jh0Qh<fz=?Mwu;o~g5YaE`hp%~Xn^v6s@gJw)BDE70M?SasGN7b6
zhQ<mB%#zYZD<xK}f^PiHt)p`tK0nZ4xhIpp`q0eGAWh5T|MDU!)u3Fm!r5%Iy)0dP
zVvg+3XJ+cMER!xmA@%d{KIre4O_Nfuw5Qv=a8u!bDU=H)0LA;-^5E=dyOA#(T!sjF
z1AIj0HXgS<h^wou%t#L#VD*=I(zwC05hqNlfRx>(jJ&x+B6a+&%sjR6FP}b~3<`e4
zJ8PSG?NyFsRJ$4IyqOap_j2Eh?g2AutWTItDn73(E92#pPV@KZiEA`gr$jDRDn6f4
za>b18-xeMN&?YROv@85_N-p9raPb%aON<V^+LSfF<C0IteV<{zH@WXKemOEEf|PP~
zAs#eLJ?F`{v6)|3AW|<UZ3uc=_o$YEoJ>+XlfvI_KDbn8OpAk7sx4dV?@xL*J~6CF
z$7nJGC=5aQ-Y$Y(_rnK|%;aQ^VbnK1)t4qvfgCS@<B8FLul;yDb!oK(6`_?Y+gbAK
zK>`sPrhmVoIvKn7<s0O<%FpMn=AD5>0l_XX8e25=`HsSKs9LV>E93F$lM^3i0Wp4F
z&o~IIJiS|nn=Zav31dP3Z$Jl)e)=)JY~m#nck!ySYDDa91d|e@3I)e+ml&TnWrp$H
z(UjbPq2Zd-?`LATLEN#Q`7{Clj-Zv;L0#j&Ts~uOCrqJTl$=DIkeDte5s_SLTNoBt
zSk)0WITnWzV!hqIiuy&6uj%XG9!d?@RVHcZauQp!wt{&lB5=jaFI-*gwCH%%5P!3=
z#t;-bQ)i&w)J)<7cya($z~q%s33EOTLjt@_?8K4|AE#+L4E3pJgAL@F^kMyH?DEo3
z6S3K`kqOZ!c8^43mwO*Up<Wpe1$5eF@z|v#guvfwvmO}S+W-D>n@U2qSCY;NTlLD5
zXU5ka#GKFzde}|C`VOY|om>v9WN2qOZ9#dtVxGYd#K?h@<)Bi(S+Nwxv&hqW8mKK(
z)8Jup)*!c=HjEqeRa}{q8NO=ZN{*Mh?HB??T4q!PlPy%(Rh9f3@b0=yB2LKaV2Qt;
zCmA&}m_vASNP>7^>*_3StZWGzO@%zKPZ#zTe=HoCm<H1!R$~x*BcgTd21A|1!mr|1
zt$&(iqBO&k7Hyekx+9Sn(X5hEfXOBp{pGPrZZUq-$KRhIQ|AUwLry~I9;_ONR>~jl
z&J6`*zUmvql<cV(jMsf$dN@eR(MPQx{cCQzt+lPaVq0<8^j%q~sHbIp;BK*RBIu+t
zi-KL@Ghg<93O}5SX&FfqNuLl)?x3}!p0bq_FBY(TOuBfPov4gA4(<h=0|}a2O!Rhd
zJ2ka_<qF${uxz9{4_qiSUo;N}CPfqR5~UiW9BtxvEPwgVAGbxx^T%JDGW__E&NL6N
zP4l5GM603T%isM~a~2@@6azk`ZV@cpJV3q{Dtyh>t}=9s?`xfTGon5HS5jAe*SAv?
z4<wP1!zs^IQU38tnE4R~5GO=!+ZQwS8Bu)UB+41o3&+xjt&vwJ>T=VH-}Z&$@7q5+
z?cdhpwe{%@#7_^N|GOhs;-{ErxN*NM9%QE=VvwJgrwW`~eVb5tAfIV>YDJ}vV#D;K
z$MY&Dy42B&sbzNLF?>aaaWB-#eEyvnvv@37Jje{A+wLQcw!sqva}|~)+IvGPZ+f0m
z@V$|1ua6+l)M?Xse4o#r_Tvsh$3Vk9*F<4tVn|;y`ZNiW9_rR*Z~v~6HMSW++7er6
z`{9D3w?{cgNO0j}&h99t><LHU{#Lj{;}<r!+wv-^NG*sO`9peSplofCC?=2)YtkuA
zS`<?zR_*V1=s)1H<>IET5FH$|dDi5Hy7i(-L8F0{^i_IykDOfH@or1axXHRMSrJ^h
z`=a^73&qo)qdyKuTx7umipe}(NjVD-_d0DoMmwqR9__WcsWRpffbz91c=*_@uE|)A
z(Dss>OpMSaiNwRg%l}<P|9==D9idOx-J9Ht!(Y?|9>5mO-Gz1Z2Sg`!=XSj-KHU}T
zdXUlwf9!d9eO%Wf?s+uQp{^cJU?GPQkX+7Dna=z3mNw!TPfXb*i8BB&LG6|jbv2~@
zyU;>!2(`=4)^m{_G*-ez<gs+`ytpM0>~JOpuB0qM({0Blw|0$i@zEsyJ;Nd?6Bc)W
ze%|#vYclwN*+j8dkzr96+v53kd+rcKBTwUB>q}Lgg_Ufl<^9M=47W04Av#@UFR_g=
z&FS0CZW<Fsj9H}l6>t`)z+J>76k(`D1HRt9Ipq%CXFfZ*Il8>`BHVeibX0#aD72U;
zAIUO~Q@2Ex)%gsau#6u!K$2i7v*Tn^MztL47E}(Ib;X|e`EKg2RQdW`c-n8V)CYpm
z+J)&gy9YKq2E>RU0k1rcG(ISCX886BPX|0<>Vub;*B{MBfL}JIFu<NCK<k!zN@^49
z6P`jBUOL7;^zVl>8TgS;)GrI^c-*;&xu+i)QvT*>d`)7gjUXbbsP#0jcM7~`TG$n*
zvk}_X+Nh9YL{#bpt=>HTd(DO#YzmOOgMFw(!o%QDq3(rW3OP9~XeFGR<T?hV433VE
zYcBsC;=S{u2y+wl)AfV|vm*R`dB!M$V(sH(Y`AMIkI?N`pXJfUS*0Gu4WYp&Lt!UV
zjAM5*`n4k9D21`f<x{#49uBXcS@MOHO?S)%3tw_U`uH9+Xcy;7#n}4DN0YF8?=u;$
zD3(ZpAhf}7jQY5UoLp~ne0Fd0iR$&g({92rHrHMWV50M@BQ=Kysgv1J;aw*}#iQ7T
z5jV<L1s=<4C~_n}LS;4|Pg!p7er=JotCf2Dk{anF&r!D`DMHvyHa|DAd8QH>8X9i%
z$p0ryU3c0>E>_7?K8?qZH;vc~{HHi_HQv@rJUa2h)^NzQfe3ovWKE0Mj+$!uaHB}&
zZr3AIz-`CftK2S6zN9(?Sr*LRXHLD6!Z-J0($Pme%;v2`_lYwI8~ctGg6xC45#<kr
zLd)Zdz%U0n>49)v@+e)%XEv|B;y>$uo^D&6`Y2A6F{a+3NA~cUXlUr#Wmq%@ZqFo<
zd8FQtYlOUDuvg}AkCy5`To)>FpbKmK3|#C2YtzHtm<kU?d&iGfpq()CUrWBH<=1Cz
z)4}R;yq4?~OIu25@&Oic>*Ad(D<d5B>s6TJ%P@tX5lV~!V6b2I@P=Q@ZxJgM>m9x4
z`$V>0?>Rr9A&AZcF68=5t);Z*;sA?q>ELm~sRs?;d$`h*tT7MVq=Ek&%4hE9YE3TS
zEK8L7pRu-o;xB0flOs||u}8B&vdew;XtGW<t-z_sTj^*!3zf5j{vfkpF8ti0?%v;>
zrSBx+Dg7)Nkg-R+S|6|CUxEpd(D}3Ft6RJG4&8H3S&l<O+gEEyaZ?NPVpUc9a&+~#
z|AQMjDRL$Cj{B-`{k^qgi`g9;Gwib>2m`WuL|f|D?PAu1zIjNP-eyua{ubl|mFN0%
zEoR2eMYzu-M5k5x|7+s^^$W!{DgN1;&5+VVpnRky5YEe1KY7Sapw7iVbtq-NRa#G4
zY<J1Lg){Jy$MC*r{IJPAJ<7XycAbYF@@;(nQJb9Xsr+xNOg;Uc4xfK8K6N92WJgVb
zo|4i2bFDsJLm$}@)4LK79(;Ye(%l_@(cFu};RqA4b3<q3L}G(Zt)E-HlaPp!S$&eO
zFfG(UNNbB%Mcnr$&%k$osXPg4N?e_~hH&i^{nB8QS2eMEuFLM!<O1OftGhlXXyv|!
z@;yPdd2UE13+?Qn{<9)qS_0-~pBrBV(_ebY=3mF!G%8ld#(xGgrQe%b<(FoFH-vXd
zkq`&o0B+%L1R7sIw+s&6qlGnL`Gm`KOTL8P+;E4*hX4HUzel`Zbesb~v%z@sqyqQj
zm%Uz{u52Sh#gfE6JI)o}L<nvqD5bo%_-b<5Qp}-2p}~PvIfdOa`Zu5~vSf{`E+eXw
z{+gd*9iW9^Y<2n}I#Fgh#B<2Ry1hP{O%Br}-r}r?kI5$CjsEu-rlZSxvqX>^!DE(K
z25g1#9idqHbP;u(5N&-|;qic9vdd@l^Ij+78-vr^R=c?I_PL1G5dV&$mo&J`W5A~!
zsNX6E66ZQ7z!3IRK)qeG=Eknhfwtni(#6_ix{ITRN^o>K{RZwzkt=WE=5S$ZvFcT4
ztvi(r$&>nbCQE#5<Xc;c)0Wozl~hfX2<Kyi=;@>W+}hXt{l;U)Um^CHS@a0*9^t`b
z_Z?}5uAH>&P&&3fJS|C2Wn=QSCDBxP(=#?{a#(46{C^f0fRmvk)(WkILVsH=&W_<s
zRq6;^iyhrmOi~tzMya7e*xgi~uWAfogqPpv3=dFAL{N9bt7FprcW*^0YAp5baWuBK
zssrHJy<aErYbtGdB`Z;M7fQu;Y`tzG_j!1H0uG7;pFKv59i;o&uztvDT)>1{PfKOk
zvCT!c5$4gd_JZ!gIeM3|98aHq{dl6<a`_KWSC6|g0?zTk&2)Yg{>jrjCYhhS)9=hg
zC{#Fv4N;gQ(W$V9^ay9KfuYZ3Q()`unRJLJA~i<{N?thKq-nHyzIX2?zZpkiB)&ge
zSXuGd<#1Ka+P%(mfp~;RL~%aozaXWKlsHuLFFAmM#vv?BF(d7%5Z&3?+P{_L{hI@B
zfFmd_6XnMCBh5d#+j<@6*~rU7c}7jJ6bwiZgfC)vD?f7Bu11iwnR`Wfodj;bWaY4W
zVV^)-J@jjv2vpvqBh&>?wyi8X0w<Ggya&gm9s-EFSIPG@l!&R`w_8VexPNhg-3taI
z0qy~P1_5VcK1cNc<6oLQ%8gkGXd)s9h9<~yOT4#7Y1?xRxp<J~fH3q+=x2|L^5jdQ
z9f4-82|1XzVBrb9N?jflnn{5sF5Q1!FY4jKA?2O&Q>9SqXeuWT!r<?C==1+*I`4lr
z-?xp2*fkPbwP&o>-h1y^loqv1jmF;VM$J;IK4KLes1XzqVznq*qZFYcRPC*1?I+I<
z-@hQq^~!bM=Xspx@jjZ|tu8W2H*vQ<s5UkF;3NY{NuZ*9r^jGQ;Ps1IVCYYEPp<s4
zRLOmL8iIbDeAV~6A~3ds8!t*oWZfA$3veu)UHZ2zXMJOQUfhCXKELpOO?91i|9oBJ
z?o2nfEB)0#L)ennaZ1GJI*7)t=TwYOzg&wKMA!*Si;Ek~n+(wHiR>ll+F?W5(JjK*
zVuBn|jjB71=w6RX3wE{yC<9Z%NCwSgA$%3ZVY9EwmthG$Oxcagxg*gf&)`;guJf)=
zavtAzW2*9BlH0+u;82o(Y7Jx&a314~=!)NqOxh59;qu!}V-0N~KJ|w(@DX0(w%0!~
z9m1*Lcvc?bx30-m!8Ra_($|W6;lk<^b3LfPawdFnU*9Kg23jTnGs_I@j+g6xsVU;I
z!_iY=ouWs_*S+1~mme%F7-@j^lIdyNBxj~%PZr1Sm`?oa#c`q15W&XxansiOhiZp?
zeS5E|HV8Dw!}Y|ZMz73>QI(DEhu9AUjxTfNU(`8aE&WGnS657oQa3(N$1ZT2a?xIc
zA&!aq3yuP!_<aQ6iuYt#<Q7V`*s@HVjVJ$B@EWo<$$Gulxp%X{$}x&4vYx6&7StpO
zk%9Jm)0TcSI;$+xD<^ph@|pj^``ptq5LziVGVSH6oG2k|1W>yHb32*s>UMd|(4;3%
z=o=n;@6tl&Z8;1_NmI|WSSGhRY7g1m^`983vSlvGOEv)oP#1Ti3h`3krmB&{j#?vo
zZ&POQbmZ0L+2v_>%-M=Br`YM8NtlErT*OIDy=qB{lX!7~Ca2YAPnKbkH%IARGI2!M
zf-N{mhyUP3I=j7o5?ieE_{2lVr_+il`=Tl@4~ErTdIr4&S$pp-VUGomFVI~!I`JLX
zKUA#7-?)KV*E>C6j*Qx_Wqj_NIg}2*BkfJL@mzU=ZbGUAeY-*FXI@?4x4K}lU`i}Q
zXncI3)$!^&4^MGIB;&o->Y50%MFY1`CWjKby#=+s5QezU=<CQ>vUSO=t+VrpE_+|3
zY)Y0M$Fx?<KAdjWYBKtxSi_gb{|$xJ$ABq|g&Kit(`V1`zUGp2{dbamJaAl;UY+)8
zkyV=ifl8r^<wq8S&v;|R){e2;va^`0LI;p7z;Y0C7SVQj@^=S?42Jt4t30R9>ZWTJ
z>RW|fsO<lweV93|$<;9&b<=kf77FI7r!+7KFV;QHDz>d@s+m4FDf+j5zx{FJKnu`+
zBD;ekes0h0v&Bxl3szr%x!YN4(K^$Pq}$UINu))NVG%5EbJW7vKKF1etfr3Ly`oUz
zdda~;vZZ*lBcSt92n5b~E+4u7DjVlZI<Rw<rlI|YaRUD*Q#-G^WsVW(jEd-7w$Il1
zpg&=~MmUj@($b*^+UFKDl;5oGZ~<#SOCh0pFu88RKN%0xu8wGbf>D2mA{Ea1YG~@&
zt+uToAwk#vw>MKQ>U!tU7q~duSR&vZ?N$xdXM^nT#}>JOLd2xP9)L(l<rB@#jv@S9
z8nTA?)mUx)9-%b}iC3t)EkaZ&{F41qAEp)#$Fxu!Iowz8?v?1lOKTDV>4}oBA1JF3
z>_KKrF^Vds{r>A@w(d?z2n2&mb4#v81wMa3vOcYVI99Xix#;d{dcHQrgA^(O`f!SH
z6Yq&?NIvY<5@D02O0TOK*|eKuYH1hci45k}vJ5^R+PR+Y1}+c9IDxARL^-wtsmDpJ
zaOl7Xu!4N(zlv7TQ2?NYF<3Law#An7{P#0(rye=zK`D|r=j2j=LYp`#Mu{jsGKjUU
zFp-%k$P2>NR!}I1;;?7lpIIDmH*j9{wcIkvrMD};5&8XL`g1d!I=k)0u_%j3QJZhY
znF-@{>q~*Szy>Fk`bUh)H3vV5CYd3Kjkd13X4WfuZ$vOv8xSAH7gmL+_^=%Ao^6~B
zgo^DsQbc!l%lIyiJn=P(Y#&}53MjR@!+DVy7Ilv68bz0hUal%_OwFG*zUkU4yTPs~
zIE!nb5%3FenDKFzAcs2Hw$E7gr&?LDmkez<TSR6lAoFM9LR+W8Q|^OMWi-bCMd|d#
zPptJL@6F&>z{PWf_Y>b{;jbo;)GC|=u|`VLlYal<cHsh0uZfL3B4CEI!jmdina;?*
z2#Z4LDUL1`yV3Kk>qN|ige<B31IyV*aArcCG6Q$CcdzNRXAb<7Qz@Yk@!(}*uY6XP
z!+ue{p81zYlB8@B`H7J6w`Rxf)1U6@b`d$Y%q9>41<WpKsbNeFEx*_oR~2B|X(fU0
z7ToPdq6>mw{R3_;PqRRjqszaCFmt=7&TIn&D<)DzVlU+4qS3=r^PU1uo<4`7Xk#b9
zv(geuiTY8hx|%!R8CkbQznLHKgHR(ISo8;Q{adW$REr9`=}kTbZg2x;-O*hRJ_g^<
zqQZfT2=~0%tJMVTu(8~mK(71{7t*o@aY(Xq%O6NF0V~-nzmtR+4O_X<kh(B)z_ZSE
zma+e%j?SCD5Up7>TsHXr;AhrKrP}p=GH}s5lpMOZ&^@G3BopfqTm?VdUsCGJU$diU
zt`tJNtM0@E$#`#m`^;lhnDjTVF4bbwR0G&=u%3EC!;>G*G~n6I8amNo-p+lV@J7gh
zOjBm(bA1=Dy`9p+YWL;UgyZ%37Q$@JJfTT>teagJWKTPP7txSNe;PfWCwLdsm!7~#
z<W6_GaUy-J3Ug^<rOKvIdf?a}wl*B3xK?48tcuo>i(l_Y_&l53U7@+GsN4>W7X|h<
z`T(h`&W^dd8J43RB3-9hX<9oT1>!OlkfA{<b?;xlba6PcC23H|fWy#{>BrFQ<efPh
zj31ANzI@v|t7+7RMKE+RJzhA4@$K&sJhg_!y^E{s75pOpVzHr}KqB3n_NuY-M%wQV
zowYpK!&t8~R3&$?@{}D`iHx<U0o`{i>=-5-Rc;zzaya*Le34`CZ1;8Sd;e|yhwk&Y
zM1Md2XZ612qX?NmcET{U8iSGh_Ku_meGw(&r@WoX8AG8>C)a6%j3^DC8qdmpbNG@L
zn#KwuCZ#l|A7<;J0ZU|=>kIN3$LHjSbE&GT_-+)l|NOM0AS>iix&>#-ngQ6j=@9y$
z{)#7ODhkWy50_-WA5h47d)bHg4wI-ajyF%U_$=BxP4Klt7ELoP!Ty$XP;RZe*Yd<q
zx%IeZta?Q^pT;%Jmq>z09@9Cl2j~F*#xE3}zTAvn|4!l*Q;N7bpKuiH-IZ4DR#RwA
zth4|%$lS@vl~gbC4Rv=ns0Lcp*d8y3b$ZnV2kVy9qqUnC_JF7xZc4iHxqWI9t?xhK
zvauNKN=Wyd?_sp=XT0+{?^P>~<`nnjK>i0IG^QJm4X+xqwX2yPf_MGE3c#j5E}Uh6
z*zxnBk)7gmJraes$PthN;wIZ}|K^4&c16|O$`T(`tAAy}w)>j~K=KGeRo|S5iJ47E
zvpHsj>+5r;S~`2oN+8sQ#D$IVf?*DMZZ^!}`DD&rj`hyYR#taF^g|E(rD=IKRDZDl
zdI?IPR$sNY@;$6|>`L(d_`0Bu1{x^3YBV8RZ#2-(^*W-s{;;0Yaqm#kWNA^2cF&Yt
za_R!!i7$jxbI#NPt60`PRW4fCcD8DRl_J@AuWBoJrAb_;)#uhQZOjJ~d@DP3Cb$;C
z-)XcHEAOWls=*ZSE~9PN=7-m&X^!6u$CK8SI;Zu<LXoh*`vBBwiRe1!wuHIbmQ`>~
z?l}=@+2Q~p1oY-0C?caF*hXyHNQ#bl^oa`yrdcpe<sA$>P`}-?oW-SWH+&k>xenh;
z{+e|kWNS%H$#S@V2wd+Cei(oGZ`ZT5pE})Sx_o}>4rTmIaiNQq2N~|x9b&m5ce(pz
zdV8mwl_bHqcD2!pz1FU#cWX4L*`k+zbiM1WGs$L)pGfwkt=G?ouu=AK`&ybgno!}v
z_?|&NUVba!3E{H@g*a*5+A4AXc_x7*^9BEt-1MNCHnN|_a#6>Eh)8mybiB(F^7E-A
zcooL0zq@OFce@vPJo_Ap#QXYessxO2_EE>Gs-7jg2YH)!7381qMc;Q(d^Q9Ibpm&G
z>)gHWQS;=aQ1;Cs>Eba|<j`}%NCxC2RrfC>AqXf?BM8c2V(*9wvhF>K=x}U_IVvhX
zaQ-+`-2j)Ty!x0-!87VwQc}@=9`4{XbQ#hX9j*fW`>R7wR+b_?b#X;YOTolA)Ir@F
zON~TT&_KRFo<nYZ`@%gu?yAp6dCS6GBR(gsK|KN6GMZ*hnzo9}l{8&?U^v=X2WP8r
zs{&3hB*l=}U1ikJS%iY)E&?|#=DRjDwHM9r?GrbcGWMPx{nopz2xZ%D!zH_m?f(7p
z@orj?PTz~i;Ki{ijHbCEbCPHN!4Co~>EPOJmd0(kxM5>9$RH!9VYxOH6m7E>qw5>O
zmx-5NX?|PX^Ikx}V3UKe)(D!xVwq)Pfyv}0BV@jXxD~oj8^53|){`Tz^Lr<RfL@wt
zF0%MGyuE(f=1||1a=AZ2w^v${xg=h?&=nJ8zt)0E#2;}$h?1hum}50QP04rP_m|ZX
zlUnKHlZdx-dK4K-?9O=uC%4uH`4G5X+9{?1C!o+It_tn^9FWW5JTAy!MN1dxTC{KX
zfV18^;wC=%d$pG^vC~7N_9E84kr>w$Pk_YcSk9N4CQlG_g=AK$dP2!1qA4LB7$3)F
z6I=a`*hO5^`5wh~+z>IkXrV5qo3)V#62`Uaj8;<Vyu2hYZq$8sY|*c+ITX=8H;TUa
zss+^tV;NC8mf{Pa&s#V276BZvC}GF7E6($jB%Y@8!d~v+l>O+ZB@`0Jc}Jy6O@E2b
zN(2Ow5BVL$6{cQ;5#EXMUDDTwDFm3hesj%pW+F3wsD;9h{73Q9X|$NWiNfH6a61^R
zw8mOeG|OkbUPnpB@VEg<>m>W3N65g?V|;jXXLT?)c`0U=`w{A*5vY!s<Wqm!&7t&5
zFxKhC6vI2P=F!zOTH%Lm&)=B#<G^Dgw{1QGsNB=i>1A<#3Pn!%N43ep)s~&Agp{JX
zipBX!Cs+3n3Qm^+eM$|+jq@VmpDS+F_Zl_sUzv{tz42erI+o6Xzt9CG<b(k;SC^$X
zHQX_(j@JiU^&I;fhX5|Qer>`NAqnoo)Dd3UQ#n-flPDgL<fG}s731N!=)bf)Rm(kt
z!8VBnJc1weoqM<ueSbC&uN)Bn5f4aKM`jlXQif`?39$B84a?BERq6@~+rUN+BVv?*
z%N@cx{5k10*(!&3a;lKIrY8MHLQ6@r7i4jDckuQR{`5)>2-o>6`D5&dP}J{Auy`?Z
z1$QT)&=DCL5gr<G+S`7$k!9q7v#@)!hC7aMPr}C(hP=16s7D;ym=Jvnl94%Df4Df$
zPTv||fYIlP2T9W%y@HN&hNB-$h1+l$i2KVUsCQAwA?3T@<k?eNMUT-7s$CcRF)y=V
zJzs@+_K$*q^Rr5|%Yh$2v)h?I;XC7hmO(frk3j20wV%v2WILlAL;sMG(+ZGq5BF3d
z^tB;abm&q3Qe=WczTPA5hUJpgX>%QCQzC6a7!RyxlRa<tM>x1IqcW6^7(})0vPl>!
zFg^2K@da$rA7uHX9z1U+zD@j@i>C)tMQxV6^cOgDQ)F`!th=`>d0To4`-C;{oiJri
zv*CXqHJrpXxMvx%iyp095KsT|2}NuVk0y9BWe!|rs{QsKxX}C%GX<k?a$M-@=!(HO
zEF>B0KbTt6ncl3%admY!w!nM)Pj+#dWhL!jCP_{4GUuHQ*uBYdUn&n09=$&!1q;(X
zw3$FD2c68&3paMJ7UAQFqR_}uSMI#vH@@gY@!iaU(02cB&IzCW!=Iw3E8cc?HA6er
zzJU+j%HF0A_zMh7*cp1SE7;1Wd+y^CF2)nZjaV9&iL$V>vpeqI!I|pJrvjr(`DDQa
z8peJ^c++u4mnO(EiyWNQ9*ou}VNeJnQUXpSxLmk`&0!&CgW&g5-kv<0^c*0Pr<-Op
z68vvR2QFF^sMPuSH~gr@Z^eplJ^%RG*AvZp*eHkOO4uAHC3A;TOj?dZiG{|ZS*L%$
z_|vs<et|W}(O*S`igCtVdxp+-mi!ouE^&|jO&~c>O*)u<VUWl{g?QelDtCU2{R)5|
z20PrX`HvWD%gmk5TDE4)P+pwaiqq`vRUy%#%5Kp6_$#V(1V0A?;WmHg3QG8!B@_4-
zPr`~1CfeU88<;wFJB+U36c^ZNX8t<IY$PFK9+JpYzTfZGcc9GtqSe}_$VDG$XajrY
z8gtrlL!g;1ao+q}JUh!z?&*n(3k2kR?=$c%7Jop^+R6}K*<SYOAz`7zXosHRk$k~4
zYK6?DRxdkSm4&*tasmu#c2Zg1ElT+aDfBx><;_|mFG&NkM!xJ<YmsUP@a*^O9+;dy
z;^rE^xZl?|aJ49IGQ?Mbm*1^WZ#EG_J74&@N@&OpAAO%0o@&=RhHg$0>Tg!H2CZKZ
z2!&7!>>0=<Wt7{P$435GMUNm6Yk`P)Q-k7G<-V+7U;oiEKNMNtG@k(}TYrj{;2>I@
zzxB_QZ6jpCg!N_0u=@63Mmj9K*i>xqv=X7b%XZm<y-<E6C#ta49KSG)-Fzt`ZR-aj
z*!AQzne;;x1Q_Q`6zC`c!C}fOYQp|=#QERKjbu8<O&)acu(;&1_H#@f8Z@F~;bJw(
z82k#QGqE3*^88$4;hRqbQpI0u2_BCrCV@^3&z`a$FAQPANn9YFoF!+gmg~5$1Dta~
zWeW{$?w@2kN4K1X-d8|h7mS{6t?Uf>-y9SZmS-N)oBXP1{vZx?IIP6jWn&+@RDu7t
z`wRP&zI-x!S&k<kI0#CwwA|J$SMTrZQ*YRe>RjmVYC-WF$a56hW_@@Pk}u2@E2=IR
zT5kxNk+D=MlId9R!`@k^vwN<r`K0~B@t5$u-f$4Q?zvmG>bQiI6^wFIPT@1JTzjt-
zSP=5yF%NZ5$|V%|#1I{PuP-`%B3Oo+YwpmK7xdh_;YAb<ddgCJw;1{_(-F+|;E!^i
z^Y`S5n3=uS{f#Ie+W}`7%yKWsnR`@704k22wH0JB&zaY)IkWZ7|Hjxs@q2yzYKyGs
z9xw41Bpp-C59%tE$kyyba2!Q2Z_a@{Z2~b7Qf8{!8vGmp2w`T075;1i^i;kgg@sdh
zY_d4nYDbAAc^A|hZyN72#X~JlTirN9Uc;FCCAUvK*;=0p6(mIFsE~;J8wlOHB}HWG
zJ?k}%gS+gD7SXhiq}8}2h<lq7We;eK5F#G1knC(GN;cy+u>A5|$t1mJYQ!zJN)__M
zcV@35b|it`Pr2n*nNEE<PrEsFBtdRa4XTk|AEt`91h%nrpXMJg(1EvI7#y=A^V@Ci
zezI^BD;K$LK+wl(Lnphftq%3Ja7vG}hhY>XZ#(05>(j_N+U_SQUALO;%7tkFLOjRz
zzj&d3MV!ySUb?zBRgHGRT|K?6F+4P2W<lthv?(yL7!+|P&R!}{n4;qIRsuR@k;M!z
zL7y;I6XG<f`r7>CDb(UT+`HeyW_Fz9beMS??k*cxun3gs3VuTvs`Cy>WGWGP+)L;A
zm|$YUI<<MK<1K^oqnk_bz(0Z%<@j7sM@+IyScW?DYHE*#ey8htq}RPTDOXirRFrgD
z_!&4aU3a`_=}%Wxqq+5Q^R(Q44O0_-ml?b%382huDvCn}@J8Qy^F16Zdb3Is#U>u8
z#T2L!dN-5YdCn8k=lc79FE^%f2=;*U{XWiloU})NjV4k8eH>FtkZYsDrz(UHZVsE(
zyPrLae`-sD%Is>lB&SKaB6Nj$141>NfimB-d!Ub>D8v3r_#UZR(@}rsP{s);CZzND
zc)6p<v>@^-#eZX256Xw8=BT0O`3CCgoX-X2cwJl=^LE~_;C3Y>icq7NUl}%j(pe=K
z_1h;iOwWlWV_!W^FP>;E?Z?ZfhtJUfb65Z7=Vb^=A*y_eZ?4jri6E1(Nm@(+UpF_I
zQs%t8p(<?ZBcJUbuaE1U1dd>;#=&d_^hrgwbkW<HZc;VXC=_iX4>go|z`u}8v&zbs
zFd`wy(e*=$bW5!Zd-I)XOU|J)Chc!KI`Q-KtB1iGJukSDUU_?S==T5VmOHqa7SkBM
zUBE5SpX-)iQEB-@PHJ`ff~FfNdz+~RWmNYt$qsuIU|Im0d)PiN+EG*yNsqPzpI)rI
z+~wIG27T$Nk$o{XfBrnJXq{Ee@3sMhRORjHOwL84XCzGARkMit`Y4t6`r_}O8}|7N
zF8LU@ePkvnyAZGNm!qEGQXqP(+gb?W{>i@)@iFVI0IIMknfFV}$|EjkfSP+w8_{?5
z)CCG>vESEP9t7cq4SiO<Sex4pPrC~$dyucMrfTnxMHV)-Ye6hPWbwgpDwUvl=q~nt
z<gAOU-1&k#`MT465IKqApiY19?om)!cxc#}_%1(PJt5{U<R(JZyQbG!V1hzn;&IlO
z!52(|eCZxF`7dqU=QAgOp+lvaojjNnJ((mIRKulw%%i>7R4Ybq#g#Y}bFQ2hWY(xX
zd{|w$`n3)njGS#UZplR5Y_&*wF<c!3Y2v|in)F1VCa!bUy+K0ZJZlTsVcLrdtBT;{
zSFT<sVLy)EsmiuHmDvi{h}MXt)upkp@?ud}jn{ITEJKS&M?kVLufZS5pFV$jJ+=~x
z&EsO}CiLf{qLnT^*>qK5zdE;n4ry2xuC=vzZ7BlLhv@tByDlv5Em*=}_Bg6;NO4=!
zLX>yMl+Z(AMz;dcGLx^EQ9q)&iC;v#n0eLD2>A$gpEom!e2@qFmKxS;`h2MRTK0js
z%xz-u-j|~kIshmkrw6yz@axhunlf#wZ!VwHEREvUew~`yf1QM<+JB{vW&cXkXN>cx
z_qHd3`n|Y<iZ?CiCR<eG=oyeQoAc<U*cGll*8D;PvOm6Dxl`Ml{z#9ayxeo3xIFiq
z={^Hky~f1VtuLN}2!-0W)zP6MszY^kcFIh3byUD$fF_TLofQK&EC6j?B+k|M;NTU6
zJ7wK(#$<_>IwSe&B2-N6`d{zBLf7R=J?E7eXHjB7xaFL3wvYSnLY3OTk~Bz6SfS)u
zmdEVg-7G`Wp_w`9_(>-R>i4?a1I}!QLPC_8uD|wlVUMR#SW>7kTNL<M{B3nb?vS`)
ztSo6Lz%Si|i=yk1q~)WFddf`?+ZVLSmjCaGc5y=@d2C_b(90!~a#-JgYb7cp!>e7j
z_aUpr6`dc1hdt)K?AzJes4YLv&u6%;C)KWt#=Wwe8sN{&PH8)}vMggDN;-9oxL<n_
zfKt22^tlDq$uCGy(AWN1@g%-$R+1X*2XoR)242OFJ@l5c+IyN>_wyIMnDTJvz^9N5
z3ivI9Z7mo-gm~jvEWnJ^|DEcN?5z*5-e$E<wuKwNh|_C7Md+gB0v0857^UL4dumkR
zyR^|9d83?{D~0AxR<!PIQ`ChjN?+~oUi|?6q17%I*Z9w#pIAR)3%FO_ByMObcG2t4
z24w1-F!N#0WPeO=;c$pxdt%BqVRT5>@S%kj4d|c0obTtlI`atO!2(UD*M504RkuOL
z`32ourq^9Bx=^#pyFTuns8mypgPq#s0qYjmPzg{34}Ywdw|~3Hp~?lrhaAdIR787=
zBXGsHCbZ%SWDS#Q*c!?VX^Q?~F4q?V3QX*ne;y8PRG#4neg0sIKUFjnG9^7sc&L_g
zY>amyHHD%>QEgA=N8J<~o|=ZwaoXA=&SF$b?@FPpcqm?67*dL}r4WM!W#4*~&PKT^
zkpy|CB<cml5s_r6-6o#t<e(na=+nTvws_u?W52_dpmeKIJnXTbxb3NC+iB7apaMxR
z$d>&re)88CAif|lM7h;Kvhlv!^<ffY;FN%G7fZV<_=#)zxUp&)2?}vO|JN;9)k%p&
zEJ1&>stcAAgj?dy0yxM9!K4%}&KiMs_vs;Hx-qo-DXTAOy=#Ur*ow~Syy8HiCyn7B
zZ4B8x?o`u&As-R<ezyH-<#S30l({Y7Eq1fJ7~{2ea;)1M-$y|M6x-JWGl~e{Yhns#
zhMU@|KTFxR>PU!VFB{UEW_WGuJfu51A~I5ey~K>L%-_Ij6r%2eTHQ?*Lco|%M6C?5
z=nw@*algUx<zek;ad@@03EtGWJ%=IK$dZdI4W`ad20Qie-YPKQTLXrTDLMlNE}|V6
zA)9dyzcutz6H|F=s|u(36!;+hxda69TV{JS{fy9aR?pj!eNm-2W#vz|=fAL~(2f0)
zo$+T(8Wkhf+uXYQ2c}Oqh{kE{`Knxk-*r1hbW<QbE%*5Dz#Xzxf9^G&!=a-Ml*^D<
zMa)Ys2`WVuNe#IjHQ*f>lT5lQO?@%3xN`M(rT6;iNrgFXws2l-`r^1{e(yMjTq;SQ
zdZpE9;z5!zjO{Z@NbvO@_Wb&4=*P{;0C1Ce(@F1$jc{-m6iyaWD-9t?@t2Fe?Uid<
zLcCzTH@?0I^mM(W>gZ{j*DVmLbz~>P(M#$gS!aQ)D9ktwK$EVyGg8ItkuaHuMY#rE
zABG-;vK<k<MaD3g6jb|au>!@%Dz}Ln#CWqg{9B@@UTpexj-F~!4!liI*~Hp5Pm6U|
ze?7e53}j9=XZlnr>nU9$#X`q}L}BKq2>XGRH@U!t*WtqGT79Y5O9Rc9;K+UHh&;Dp
z$NHmV*={`G)mn$%!d040T<3p(T#*~Ve45<5aTXPZ!nCy8hH>VKE9SqjdtA@W@!*4~
zzz#EuG1JVi>R4Lb-8g96?IOMzjS-fzqb#Xe_at2L{M-6wyj7XrE=2xUg<WtmSHI7r
zzLe3#x8z=46SF~|8DiU5Nz}+*7o8(nKab0?JpDER^zP!@D;w_z+T~eW(WDMYAC={`
zSX*s1aOrNEm@xM%Q?tbJ6kC1+OG$BQ-Da$Ze}eSDgbZBFkDRhAyiBHpkM&-g+gZ|)
z3-?9iN9;Pu>PD0)NTEiCd<a$%6S#xd8po)K*ia)z%hauVs7e@Jgk>J-v1f15`O{td
zK`#N|G+o&^tJ<b>yw`jjGi8f{5;e8=H_my#>72Tt+bdLR%x8ttYPs(h3$93;yifjm
zl9+%!zB~>tnC{ykkO6Wn0}d(Rwj(1?l0Dgi(vgT{?R4$Ve|~hxo%0ZQUE7DsH37q_
z)8p7hlKxoz<=Elh61kGgnknwKTk#NM4JM$@3e`qfUp;Y_^AT4$QcWeNrn^XHppY$h
zP1Xx}<NxS&TS4}NXN#Z7at9p7dB-tc`6mawbD!BsmCi?KM(n4GbS&k_oiZpQA8QW3
zv-udV4hdyClPz5XY2HrI&SGxif6sYi=fhj0bxYbwbYk~kK-gJ>k_rMJS3ofA4Fv$v
zk#?ew>=<QVTQ8E0$Rtx~h$c&~3g^sucfP2fP?_&kx;`R=w%lC&xbw8`xJ5L}+_e2*
zkuX>?yO<b~-jxJ}A&sI5)B26+?&#}FZ+k?Fcv|?>R&?hG%BB?PRin7a9wWy8m)rW;
zdg<fL9KEAthPC#(y{dKa!~5?}bw_|&jQ3?%Qt&_i@q<_8b5?25*JC%da`}5(TZq6s
zeP&}Tl4(pw*Y(Me*o}@br)rj;fx;eb<;i6S*~XmpPwVN6sJ3;BlL@+&-a5Qd9(pSp
zxvI<5r@oX`U~c00W6iAIw>w6`5dpIIbyRcgVzd0to>~-8LR+~skyclOVgy=<F)(J!
zY!9azR4^|+Yb{M{_(Lf<Iud$DA}XHU+noc`&1X~RJ>7dIBVf`9pN9#)5{r{fx3uga
zqzpztmQo9tjqMFD2V-E6)(0fRx;cV`-5O)_Q<idFj)eZGHc5T0<a;8;Z`7zoS!g^;
z1}T#YvB-f5|E8c&*-as40P-pkYyPwNRptWrvbsCqgfM+d2w}i%B&2ZH#ULk3-=eE{
zLDYBcEaZoUb`Ff>oicURGVY3zNyztiC512KZF126mI+YIkN<hC<`A^nKr8XgA$T3i
zrf)|TI6uJD#njjmB|Eb`&Y11X^YE@Yp6}s`%Ux(Myr}{1@({|KNao74o}0rpM>SEu
zQ0%@i*LEPh+Pn4j2Rpdqd>ts4GH-3^@j?6AlJld0p71VDZ$<)H%ZMXDbvMM4IS0mH
zym#u>RXN=D!Qv(FBqrwiU?ZZl@Z!OD!hmeu8Uu35c{}(w7Jat4h_dES|1CN=vXoUR
zQJJpqYz`oF@{`^|EY)Gq9t!kX2+Heps}@CN)J;=eO-jK~F;eI3uD%XpjCP+Vr)Fg8
ziDsbQbf;6Qy%A;k)pE-?9a2GH-gxb1{7%T2^z6{Rt$mwpbv)v+90Ps3Z=PSI{@}-C
zX36X$=PIc(2rOMYn^NkOiCoj&iPBKl04$>Z1`vF;biJ{E=@S~k5~7s7jLpOC9GoAY
zAA8&RuqnEiG#cv|=CE-i|8_`{2)OJ=6Dj0Mj89A~Zg_?&8#5|fGfO@UVdJCoH1vEb
z8k-(dBvz_+b&OEEnMQ~Koj)9jhN<(W)@;^v5MJ(oA7~6LnGkO6{urY|kY{qss*sL8
za5LfKj*Omm*i?FUWZV9Xh4|q`*-sEhQsM~e4xI_h3DK2SXFQP;+0f5XU-Xq~W(Ltq
ziq&1`?&!<Ih9ZI(aw66zZM?L&;x-Jf6?r>+>OExMm-C-Rva$fT{D+SD>FO~yWk03>
z&K!=A-WZmCi>@b|F7zLdPeLN^!{oP5fIel@CBm<HvlrZk+m6*WCg_FCFf#AJR$Xlp
zF{;zAXDy$fn4k)MY+~uwZhryupUyl4tINwr?D?WEI>Upc`Sm%Uv?j|uI*MTo9Zqj6
zZ1+ajfOz2YM%wN!T03Y`vXeI&B#&p%RU^GGpEC|>Xxr6k_FaWgKF_86lA6T^5XToJ
zJI|Q4tHa!H^Q6R(AbilCRu(>9!EF)&_av9ICD@RMT5M4c({A>bqr#sE+3UkdD`H}G
zcjf@kYBX*qN&)eyxFmy`E1^gEzb}*1C=qXwT(&Hj$YDegV(2)e4SyjP`%om^&7bS(
zuN^Z<KF8=cVu3P)`2zryhzJUV&)L)5#MrCik5>&+Ym>WOC>R3BTquejeQhnA{mjLL
zMpRD3z=T~*F{987sKHZ~q^sukOgT$e#wko);Dlnj>!`57WNq@xV|$nH3YlP{=Gj$T
zwb52stSM(8I@}?}550gyN=(SV;z=dO7jxwc{r5qIFHuL39;zp>s|axC32Tiiq{(Jp
zH!egG<KE)e1nv1=EK+c3D@YjKO>_n+zH^n{;cyqwWOR4dls=xFEIiS@{kGkQt*0Qa
zAdXVFL+@J<=Pibq*Qvf>?+*?mCrn@VMVEghxc=d`gz6KIZJg$47~XPV_og9-eRB&{
zChWSDlnlv3n`~-MaVr2J7<FhZq`RU4FIZ_GKD(DO;Ukwp^--IH2pLZvSE8%1?U0d0
zTiN+yVfUE<Ljt#;q+d^trGb!?M4~4lg>eVbeR*1G2H5AJDMl2CVr|+T5Eq?Tua-sD
z>rYvLcGPq02GjCWk>8)vav&xqgh71F^qWh2^5~B(b!&RDa3H8@g*{T;>>gn3-Q(~5
zE(X}?s|fqF1uxlC;@53fzXf_KU#b>!9gt5ukb9UV@GT%Lrhh3_d5m|#HvqqKQ8zMu
ze3Im7!gZ2^?HC@}8`(7}&|nI6Wll@yqfy`--3mW$|FqHQ;HbcDPCcQz(yYdTxPZ6W
zKZ8T=KrO#d0E1xi$NdjT)J<!~gJmN+73_~>(b)55h@3Qq>cQ=<H#D(>P&|K1MuN8S
z`urICTvkf#J3Y55IdM&cRKf$f3Mp&3zf-a2Y7-@BHQYNH+^!&xDU@PUfy8p&surc(
z>7UK9G);x9UC;~$JiTo&r(-)>-0v^Xc2eB5OpH?p%0~{<jW_Fjh(SbHYGz77BeI%#
z38VC>+29gyq|Pn%W(;wwc!8W<x<2hd?@Gwo(b+_FiJ8A5M@TxCG2b`o3{-Xa=UMD*
z^|=Ku+J0g85P?3dFYPL>KXrFFAjzj*X_So@EKb%^eaUAZGF_zguH)h^_XhYk$Hw&5
zB?1iuzojpRL>9N{lzpJix3{xLl*@V}kvsAq+~h=u8izxba{IJQfBGvJKat!Coxs_7
zQD_I*F&_5EhwJB%K<8dbtP~H)^Oy_uwica#_NiJMT5%{M1S`q$XUDca5hKz_oyM#d
zlWR*oI`p7g)K5s?ngu=3E=%)BEfz^#?&AY0o%MUOo<zSGII_{CtI!LZWXAK*59dD)
zpoen1gSML)Xeg{GcoX~ub8=urAWOK_i0pkupL6c56afR5E8ou|B@NyqzcpAkD8B5>
zMV48gD~eNQYAKq8s~Iwu%^wA53Vm7X)F~=@L2U4S;{U~&h+phxITraXZf#xBZCwBX
z*6h2jNE0>tALqgw)ege#Fl+aZP+IUc%7ivGBz;udrsK@`u~+e0y(IYT@?^!2TK)NW
zx;)>va+rbk?YE}{(0yk2<r(6}5>Zc4NWb!A4qt08cBovB{N?I%Sa$zL6_ja`_&4AD
zn58R{oah~jcC~6jrsJ&Z<-DRYJsSY!ZQ+e6nmH;W;aItJCt82uI-@NQA}eDV`hyti
zDIQ2oExUCRdmIvZ^$%BSbnk~5XNA96TNIdrPRnrM?^wwyfw<Y!j7eZ2T`u;7f2uv^
zm<@7Z^xIM2xsndxpncgi?(&cb-nMDa(wh4zH%NYW>uNbTiG|F1${3p1?++ML^9<+M
zeee&~XQ1o#EAk~+QChGSQ|Cepd_hHAO}G{zo*q!O<~vBxwR;=Qx<4XzMhZ7Kb~`b<
zz))z&HW;}IH)`Hy8F^Jh4TUh4xpVAdgQ2T&yOWQ;jdj{*-y*Oo^F!O63zJl92y4kp
zF__^rUimML!paX!h5l>)1$<4_w*U$I)`tm@bUYD7G83)ja{1ZhEyP^pGw^8$32T-3
zPESvZ?apsHRnwLjB`qIz%&pnkk7z;vZUhzfJEKcOmc(hmnjT2s9V-h_zU~hJU80Wj
z4mStW-3xgUZDNjM)AiFAaGR;<arNQYU&UPNJyLqA1WKNCE{VE2aHeGS^=A7H5umgH
zw4>zsSm-E)=4kLm$;qtQPEPsnWlsGL%JA~k6F#%fVRne%k%U|e&G#k86u*bpP5Y$N
znH7d#<xkssBfXK8xgiQ)-0>N1$n=lZ-uB>P=7KmsC1Q+XX8z>EyD!ijzHUys>00UB
zt+n1H6nrg@Rc2lQ;yrQYtlt#)@+Qxyn2O%Z2zQho4C6~FjdKf{{9K^!>W#5p=2H)4
z<|v*p@vY?2b9+iz22a#=xPC_Rq_4MY(|uf>DkHV+tN4)?JMgVEKmM(NA_*Bil`GFT
zW3!8mtBw7|mDc-%<|riew_rw<ZuK}wnXr!2xS-bAp;Vj!w82hW2$Ol-G&-~USAYU_
zP!ym{y}v6>=wS~JHy#FfdTq_tWDk1?ne^tLT=d2O`Tf0(kMC1vScZr4xHqmQ4j>R$
z8wQQ?>m75~mx!Cg>F#U7rEvWQbwWas$5A$fWcl6GW1@FSzOxy~x=fL9hW(=2s$10}
znj^NB;=m(#HL~#fzMh<Y(x0xc>F~-kxj40>D~abeHJ=o6mN=ym(2tB;o2QYimCd*{
z|K%t73^h76r(&fyB@_CCJwhw#&F~<b`~X7-kUeU2x(2nn`nwQf+1gxDjgs;HFA+S>
z7GNqg1Zq?iA*%t;Jio90aoL&a=XLA!$^+jx3?QB62B-A=&WDhuEvL35z{u_}H28`3
z+JenhW@oW(ffC+Z2^Bs&O@8P2g~d8uOxyPh<AKOL4tNY*kQADOvouu!R0uyW_tF@y
z5-8&4dj}zkt#;p}tN4^m9tT!OMDlPkP%%j535T=3bbV8_5p6E>F6mRty<w13-MsB2
z@RDb6vbo9zJ~cVICPk9bo1f#+g!P)TY1*cn@CWWKl|e3+#vis<l}S+-e9E2lvP1>H
z2^1<%5#J_g+q8TFa&X#3{sf+mhmLs!hs)*J(DY$++vrv>GSG^K`ZicsUCwc;CmXbn
zq4-ia=+C62d*&$BMwIz4Y&y;P{n3rV_{nN#HtZK_4L>}LcjYcHvO-Np<dg}#ij`Zq
zsi7__5ZHJXQC0Unky`hkO9<kuSNZX=du)NJ^Q;Dh(p{4I$WN|2WKW?RjM$JS86i0h
zi#SDKWJn~Gf)q0pNdIB#^d+c0hb6`HczPfIlsBzB+}OCXNm6DT*l<=<lQi-Vd>bUM
zpQpaCd3kwo$Qg5qMhuxD_+?4lfmHEwY3kmfRe`4)Ew9wod7O-woRzV?y+{#n`x@&^
z9~K6(KKHI>EEqTJS}gudC{-YrrPX%W%j(%J6HTG<MR{A5A0CQDI(DdUWCKFbZ{q2d
z6aG?EKSvt8NDkz~nf0I&rvzV9N7s{<GwVmskp+bjL0^PA&Qye{@%tRb05jBQ9WRx(
z%fa~N>FVUjZy4z5{I0a%h(V)1@n;a3anINQ(g}OvG5;0qOsjHnDOy2CqHkcq=t6L7
zZ@rU!ks@&~{J6hww*^~mzmrH}98ElV`VwUoNE(~_;Tcr{YF620xLkZlJY#RZyMD`k
zwpqrC#`2jV&(cHUKMuY80FNjiT{%UQ4vB<hP-Q6=ZnKK(7#=#f4Cox*a`zz|NKRHC
z`$PK?m6XRN(}iJ5)H$4$+~vc7kg-1~%GUP%^IT#j3x=$eRF!u!f#Q8!w+(VB1A2<d
zHD~is7Pd_7Ja5x`7qf`Q@JP>R4OO0@?57O#g;?yIs|_O;N53baotryX>x0A8K_yIj
z#c7Gk>KHRp%rc+35ck01e^WP%NVW2PX-L7yP*@#$RcF&vRlX77f1><d9^~v@?anw!
z^fbnob!nIOF9FY~ee%zAyYQ>Ws`nHUMIqp47_Z$IuaC7Va32$(YPc|VlIuG0Zq<`j
zpQGevZysUl6#EYJ6bIaMlQ(u_voKxnPtjS9QdBOfnE1qIQRa*5-<qBM88m^iZ36vK
zUj7@UtftCl#$egJGn-db|LD`P^1jB?T=O#8cr6}Hk6>@FL*+oj-mHqZY(^Z0h6;<~
z>1&4@?R%q~T|>Q;Lgf!B+A`VW<srl9>8&+F14k&L)Sd%7RxssS%cT3WJ1zJ+Z=_Gs
z_}3ZV9M)|iKeBG^-V=X2HM1l#cO_d#R6$7ePdT=QZT7uA0$RV7cz5IuZP<(I;Brb$
zC$1gR>$)F!xes3EcFklEO+lrJNfb7uElMHkY$GU4-uPL?+)mc-E9bbLaHfp4<kO**
zlk<qqE(ZWS7yr2N%vNSH7Fvi6FDcqQOP_EibPaWJp2)yMMv=EXukR?7WHJTE>FPQ;
zc^FN-SCTMW0u|x&Ot$FjguY3e5+td9?v(&VV@I{<n6q=YqP1WLY%p?U49q(ztIoB1
zqH<8z^6OR1;&8g4O2&+E@%tM7=pX&kfoYv9R)9ASqL9iJqy#5{QEtt$B&c-XZLg)>
zHM4)D_d=iVKkiDNFf-bpDJ@e|)wg(jDkg)PDuQbc8#9u@d^T9Va_NjE-<ZP}4HQuN
z<(wS9$Jh!KmS9EEH-L^c7kEvR32#|U-+}h~ldKbQ7YCa5_>49A*FJmFa8xW-gM{1h
z0EOQld|!{Tm$RRWOO&E|rnyU|!f$9HPq*&sdJ}#e;GNL7<oGg}fe45w5F&r}bn~Zr
zxgR6ZhZf|>ZEizG7jBK!3mU_8n3tcOtP9LkJh6rVeDXnKY7ZsFaI$D!!f^+Tvfuq!
zJxmR~iYn9>Z}V!Ea{XT1)Lin29V=1=XfuHZHhz2kCeUg$P^7x37Ev0kOl6LKzJJMD
zf0VLu@<wrve$~>33>{cpuBJsGRU57SnZqqL!B1{`n>h<)FR)|>$IIX>8S-miuPN<>
z$r%?dkfQ3aC^^s-d0X54AY9Gfb73OLJ6NvvclnG7XT{PVW|_V6T%R$*vn+r{v1xMh
z#8D87DcA<*;OLL9Lm>}akTleMuiTh2CwkBPiu1qV-UX*((x0E@QyEt1|6Yg8=h?26
zFLvUFjdLtn0$9imC{ri=1_vhmRM#dHGWGf<t%^ZPnO}_S(TLKRb9^Rn8QSE=l<-}{
zZh{Qie>fD?iG}0F4^L0Xfo(g4y$g9KD>3NQs_QP+gX8I(k@rrU-3^`7+Req^R}>`O
za=5=co|BxKN7%&fo6Ub@>RLT@dPJM5Fmwv?SK}lO9$CA||K&I1@I%R(I`r8}z}dn6
zzC(tM1tDdNCh+ra6T(^9v@)3>;UVfxv}7g~1i4v>xD%D0+wHt3L@_JAF%4%*tddEl
z5V+@Rkp(Me9=G$r)p-q4HP}e+AgU6lJiP53?1_Yo$nz4VzQ{AGaQ19!ex^u&U|5<F
zUx@Jx<DCnY!RzU3Z@RlbH*5~1?HRrur=LRfF_JdqkojGeT;ofU^q+A!icYDei<#%4
z<6fsrT#(im{`B;tK(fCwJM9{S)D&{$5aVO@K6IhXC-@c}d_9iIM*dc1j*-EH^nb-f
zp9|X*5{!6E<73fVRs;@!@?U5H&%3G?c+Hlrw|6x~wuLTF^7^x~?$sEb&;ed9LaX-3
z3YQmbCnq%|lm>Gll2XrblJeV*jd60$g9(wx#^Il+)$`8*9*;t<ZUB<LVB|Lmg{KA;
zB2P%6+7YhY(#pnn5@7M*k8(5*3}yB=(@UDDIqc6DY>~ke*z@EmvN5gX&SxDkgXT5g
z$M$#q-tVXArxpwp)88U8IAp{&7Z!Z!^aQW6#rcqwUBN;(NVM)%<vyh3dZIxiY+&%<
zDU;A$Cv);90CcoIQ;CjbGRLg{H65oR*YaSrLRRaNxSYE4a!CllXvwXlQ2XmAa-c_B
z%~x5Yq!^_QO5Rg;N=XKb0n?|<{r>qC8$rq{${at|a$i@UXj>z%z~cqtO7Y|mQ6`-V
zWK!m%;cRA=f5VuPa(~(1*L7==7|)gGeee;Dse{`~x?8Pu)e-YJ>rMI_W;`9t{qCp8
z$vOY3a^N9<&QI5h`M%xi;1WmRdgGLUcMWyyzU)KPsgj5j{X#6LlE?sJ(_2H*Yy*<V
zy4H_L9&?5nexLXoC9h(5_7FY}qJxPFfxWbXK5P;VHJ&^*f<+ihGBOV;aPu?wKg~57
zeZ=pLoPygSC_-T`9;gRDkeXhFqu^xEcmjNpq|Q&d`<V=XdXhY%Uc;-W#d%Z^UXn)I
zOkIhjkemmI<r*(TZEfJE-PvZGZcan!<|oAny_#@ssk?{v8DpmpAn_gv9Fvhzk(&ZM
zF7hP0F!kKr0e^Xp$_c1Mp-oc@%8*BakCc!#ZKsoxAgJvr*e37`krUwJO!Ay#lW~z<
za2SSD+<WakBSJiDi{G@C3)Rzp$D#Z3O>PF{_LmeNY?>wg#vYJiJ>_l{Z1)KHDur*Q
z*zi_xE=Bbe1Wc^xl$;+LYcR&RA6!EncYW7(jfl7fi2*Y`2!T)GH9xS%r(`kV$4^H@
z!pan72S&=O?tJ;AuPKoXuK~HMT(6cejlFq?ZJwHaag@*|rB~F3nw>M2Czt?<KRnE#
zkn(G?z4NffXC81>9OUJ`ldW{Oq%nmgU(3Z{xD5ssx4hI@>4{_O%Xl8QI`10t$3&G#
zgXv(kdh?L$10ht`%4CEnJ1ZjVVUERj{lVQblim)aft1g~QxEY8sprxP5xr!IkW@=%
z?Sw4xcu0WZk~2$2GT~2FLQ8P`jpMav5`voG4)-4^WdV1k<$@O`t6y-F3w(1eq#a0}
z2+L36reRj;`AYIzdOtU@2i4Qj5>Bw_$%Pj6YdN-!_r`&KWfzk2-RljLS7ZNSwiLuK
zI)l8%!$=^`!8lPdM*_KZA^Q4gN!gKOv{!sNhn#z&1%tz!V<IN-KW=sRvCpfWxHLEh
zU7P`yfbE|WgV58lld~N(+l|on0`5mxyDbsMfcX3YX^8!q*!`Pp{Z^@YaK!hEorrd2
zglEoMV-?ERI?C4Qh<253hc4xkg<fb3TwSI~yz;ZNvN9p`^r|KLj1aNgi84{5qJ`)T
z_6D|z9K!8eD4eqW+2&PLFkVfdRc)ezoc$JH)W(BNPw7px$+lI1Ux7TGFtevtI?p;8
z%_L5jL|=%~D{z~(4CX+;SkgmA$d5HJnYGnZh|oKeI1UL%=cc8^eyl$JIiy?}kY8qv
zD=dse8&;iRS%}FQwB}5_KBt^^%&UG~kO;WaJuXw{9u=lcAylY1_w^-DV<GV04pHE0
zcJ@%3r^g;`wT0!8ugv8uzs*6yWjs8#6rJ>TTRW--^%YZxfiLoFYkebRHd0T_JUk|x
zL$|Q;fTVHnXF??{RL<4bi<z{YQX!_0AhNdN{gay<kEND7M#}P!5}eSaIpAuR?Zwid
zstot}Ti4y6p%B~qg|8JbMM5M`0meD|*lbH4Q6YoMAq8!kr~>rA353w$DFFi=cEhew
zgSjL$OB?cGpav(yWz$`UqN-VXsKkFcNs@WQATM1Kj(iK2O76AvzD?@@L%ztcriU^z
zaP(PYt#`gzxZskXBAcSuXUZ%FC~uC8)M>CCjPM*Aa2^czd++wk2i*dQRv>R({Y6^$
zb_Q+H3`Ra|Eg~mPh<kLbIu#ckl9YNDmY*@e01a7G=B8?*ICUA~P07knt-2rEPLZEr
z`F%;cA~$?3S_??o6@~Z_XU5vf!8ft;)WM{rFZ&Z%p2ku=9W<@gQ24KRBTBK9j$jY|
zng2kC^+9uCHtAe5-e($ETU#Tfq{DG3Fa*T3uct4rip4ybJ2yLdgYXp0m@p7L4&$Vl
zvowM9%qkPC#aBm{2jRFnCYI=UXV91GR`@5My7|)DNgBQOG`V@}v-U-~`IqjP!v!vJ
zWoo6X6O;HekpZ;ukuT~J7jfM(wT{r8s|q-M6bFh`>QMyo$S@IIboI{JrNQCTTLIA`
ze3n!9Mv0tfCs~FsBd*RPt`su{gc&4xPwLDvdo+eU<&85r)gGe1Ym-nB0g3k_NQw=2
z!|8tOrU(}YGCG6kv>9`XExxIQw@zW0YhP;lJ-i7hC{T0ruR&wxyKeqz8+$jEZ&qMI
zTJ-q}TusZxmCni;FA8ONZ)^5R=7#Rg7YtzFxJI3X=9k8+^DnZMwBCcsiyJhynIWKp
za%xx8a2u?fy{-4KMUt~3z*g{mXVwReJ4BovH^OUbG)d@*$3`E(NQjpsZ~fw)8~;Kx
z_HOSm!k$#c(vR)?V?)Lh)mYN?=XvZ02`UY*3pPyqwz)0(goVmWPT7BNh`f?2cNT2>
z(D&u`DzvVSbvAF%;$1UZl9D1W!0$_+ro4sdgukGci~Lj7Or5d<Ct!#~cp{7P?*XAF
zJ~#+yn_DH#4*1q99*io~&rvqK2P-x#FwLYh<4pRipKHvn-;SX<zoPC!9v+`ZwMEq(
z-y_yyR=N^Wwml4<_nCIH@tTOPnJOTP?Mj~8q2eWujijMd@?xgwz|A}RZVcG({JTp-
z-i>~G@*}LY`qE`J9*8b(0Gw|ONl043C)9Kx3cW`I$c>I#G)~d<o0dxYo4G>I$yWP%
zM+XO0nw9;1E(1vlEsA(NuZ@0EKvYz#$dGl7Ebb~S!5AilkUVb17UwiP_r5w<+3M}}
zWa@WjBW^34s(AjNJwsCZSECU2cOqGRf~<U5!&S@0BGuku08_HLV1k4OqYLCn10tsz
zcnG(fwt{^r`7o!%<h&YvJ!l;LGr$oErVND@DoOCIItzN$dDUUOy~>8XK8rIgi73`j
zWGD(V?j|i$P}5n>$hLQ3w@`4_V$OcaT*+L-keiLQlu-_y8;>)*$Sc_DzzHVBi?p_r
zLjLzJ=v^G3{b3;n56a7dEn?tBkshwj{PaAUo=A}`%&0dUDZ}?!MUU}hq`PR^)=G&(
zP{tQs{%+apF%4#>?giWVZVM|z`uWoT#<D+^-=S0tr7u7!C{(b$$vW;ECvZ(lxUuU}
zilOR%axH|~yfI}HtNB8F{Qi?IEmpj&q?1lk7ZHeo|D)-=|Ec`{KYolv8OO-VItSG$
zdy{c8$_f$Q_EyN=^Flbs$VzrZaX3O`o?~`I;-KttjO=w}W%E6sAHILUx!umW&g*(T
z*W-DAXj0k&gN0S1B>e*VKG^=RbsB`yaFrhZ!XY%~x@+V$_AQpscHBa2`BDZmb7Y=F
zR~s$#xZpwC!z&E_XSk;=BV!HNSG`OIs#|{3y>>}auY^U5CP(2^tMfsK&Ju;M9(h-?
zW27F9Q!3;JbxAs+h+6AEh>C#vO*SFH*hhLZxgwi@!eAf9#F#hEGR~Xpt}i8@mOO(4
zkIqjY#X{bK*}}rv5(z3>H*IdMnLvd|R%Oxu^=ijA$Nws@z%)LaQRa5bcs7z=|M%nm
zHf!NoeB?uVqMc`xQAGJu!O3Oi!>v1#7|$^*e*Sj#v8Ov>t7hwJA3qPeD5?`B)`ZIc
z^8_X@Y$Bs5AGOBoP;=Ili&&6C(W8{LJp)yHpT>XZ=gk=o1J|_{rQZ7vbiw!{L9}GA
zJFj!`&MD=rD-MPW?s>%Mxp4K&o?b`#yZ6;f@Jl`&wiAl|)G|;pe#=jzVvSJ~`pt<+
znVw4#ne@c@SxT)51^pOhj8-<h-j0&l%N{5y5-?+`*uZ$GzYJK#LioKIdYdg~%Lc2s
zaGf=;H%{^EtZ|Xq*)ge%t_+YEB)pLibWJ}dx`Q$@1{C(BJjXv)`OydYv5ozwg@sw0
zZf<mG+S;HP`ReHg4*1U#FFE9oZB}Pa(4!G3dfmJNjdo!dGZHZq$T@LMl&+Z01&ku(
z3Q3VgyDiHe9$B(vE=WSoyJLhNsHakE-q7k~Latn%`b_3#Ey%sszjMe?2mtRr$Lj}f
zBqZgt^A}yajb>zmA6N%e)|VfC{)efoG>)$Av=Onpfy(bEa7V{$2sk-=l&(ZQ+~fkw
zhh+FhatgVfagIam9GO<O((Y+x73)x{0Ec_UXxp^3t5aD-5$%4hP+zS`V(>#AdIt`I
zrkkbc^lG6EaK5{lCE6JLY^#R<;jTi2H|pObRrTwBrYZFM6zbQL1}2MI=i$_*+vz`v
zueJj_t2o!Z+S=46dhQEg^Q6q(KqBAN|I^MbEut07BJFX3V-OEs-()Yo8zayU1>jnb
z&^;S5!S{0#9W!<3Y0V?M?yu+GYx<NrG4@}ke*v4Sc5{Y3`1#NOe%^Dj?NC-s%lIj_
zBCZmC{`YtR=cPzqqtO(qt*NU?-o<J%O$t7zr>FjAHiFM!=^A4VnA4TR<&%Km=7IgG
z5uu>~JOo5V)ZMkx4Qk9HN)x!h06sQ<7p%VpzA&D@x=^^dQfRX)dJ<{Xj_~}ir>VW;
zV)FHo>nsK4(LvR>tvPrkdG-g?S0i@PK(2NS9iW)Ovyi2e0J^E3{ufh~HN71?H$Yg;
z0)uCs{Pqe768pM&h!M#{n~y3pIIg(lL%%)$ADnF(eBso@Lyy`w>yf$vycKgUf=Ei-
zvll}t<`1m)EFJ`L6_gg?+5WVAuKb=T;@jkq?N;$1jstopQy~5(MTa3PBNZuSzz(k~
zh<9R!X%!^N7&{0o<oAy8`pyk05?De!@}Hym8vj^CtC}qs1bU`&R5O=85g2S0e)H*t
z%(#A`h|0ym-Y#N#H)xG(y}t)n*WYMCF#L(m734-O)(VXWue16&2Vdc`PM%WA05<Jy
zvzZF-b01-ZPcg3k=v3#*yeJ3?5_MZ9wleM-6O(WH$g<#=YoRU}#0U+)op)xD>}9I@
zM?j#waA6?pyY&YNaT+R2vxQjvK|@n$WoJ}$f!re~dAZv)F~`rG8QL5PpUA{!DAE)v
zc%Z3f=_E>+VFJ>9y8mg625EZ&P!F0!BOt0;^64Ht=#cj>Sb~GXEIkvApZ<V3Nn<TZ
z@&T&9;xDMGD%#paRIdCkNRGHD;9{r!YLORcn`K4VlFx(4$1ix=89Wrp`z@7nUtjw#
z<~5E-l^oegiA{42yfS0+eq<Y2vBT($?H0m}Ge7*&+wV|O31LKf!C=6u;?u<wKdv|k
zRMFCnLPFMomPJ*YhSXv8i%@L865$>A;Ju_K<ZkTA3a0m1OJMy#p4shfKi_cTHk!5C
zx4HQg&ltr1I4xhNx8#fFOjK09T*~qS0k^*Et;mD%RnknQN#_$(uleGSwW=y4g<Vqc
z9Y6fJH2){R!Jl8Z6gW6lm|JnLc#1zKe%umfpw(!4d+l!<l9;$5`utYcnQW(MqEJUQ
z<q<HuTj)xFw-_H%ha$?`L1C|J=Y(55AHJamId&vdr1H_;peSpoK12=keM2f|kj0aC
zM8&t%2b+ngF$|<erIHEW<7-K2?TN>eV1wZV*Cx@%$$C@)V3$rwMmHp6J`4iFt*k}H
z7hP7QKyYn|j4jZIIC{S4-Mi}-h#k_B)K!qOvze$(n&YJT-GxaX$zidI`dm&3X$ipI
zZb1H>p*Bn&WuBMo^0jp;PU*{S+(UOhzJA)7_pe=^{KQ_*CMy>6cTjMTq`&Aqk@1fu
z%F%$E+ysaKN|}$Ekw(EgxZ-rVZ~3VGYTZk@v};F&C;4t~Gxb-j`*g0rT|H8X9pyS|
zn~^V|l~Xz^??0|gLrK1_<;=0s&}>#|oE^m3u>Hl|%}Q2d`_oqbFnDq0!@VCf=6Q{e
z*;!xdZT@bvuG%+bTdrSdIjO`rpp$j>@#LvK938?y8M-<48kA`zouZ>3V`NE9BQHKI
zYt=Oulx>pPpOo&H(jA~7Ffn<O1h?Z^ng2E27sJlOiIF#8hxj2qE8@Uil$x<UcG)o;
zs$@bS5;Yw%b9$!xnWtfNY-A)|pqC@w+-(z+4P$<jUXxSd`E(kyLOK}wWwvEWBhM||
z8i5mMb7rGnjS76<b`P)3<=NFP0=O7gIVY+_kiImva#B2q=rT)Y#paW;KzylnXaK)i
z*ZLIe@P?+w;YF<&q=}gJ#VYUD`@j{wk-=MjdGAi!kYybIfs3e@^L++z2HoE52vTul
zHs0WF;nqb9Ub`RD57Xi#>F5@F5Gep)oXqf=pNlp4y=dcfSYBl%H3Uri*e*tVi6yS$
z`?icqyQN6h1K*z{$^->vs^-twvM0XeG^qU2#RUR_p=Te)KtaualXOXHk%Jk^>PZf$
z)iyv|OPF<*X(Pro%5jR8+#+_<-E6E9eMtYRYc_z2n)<cZmLJka_)BXTkOxE(fwex5
z1C}FCk=2E1T<4C%s7-hcL(+~Y3=AFG(VHX~UQ;5l(U>76eQJ4>5bQtdCpWhP6mI4E
z=q(YGBGpwj9f|ezCH45vQEX6hmQDEzO{E+kF4Ruj6%QzTO|t7eWRp?nV(aD61sUTm
zrhawRZNMf~AdA#yMLi<`V<xz|q~Q~iydWgxXyFWn>Vfr%CH*KS%ebM9R$2E6Co~9?
zk0zOYd!Obr5F<ys9H#s@=F@oyDOgVu-#hPnd4Y;M2t6FXyk(UvhIhJSlHYUtn2b6>
zzBcy9B12#3hoc<8(d|xOcI=(`y6LN=d-^sHZLSVgtUVi7S4DxVV}0Nv5I7>>)EM;~
zMCwPUx3&9IQc?u7(ywbu@suKualBWWJUPMlNH9JMrj>kz<wiqgU-6W6b+y)~jNcD;
z7Hl^nvwh^3n97F?Wa|(2LZ(^ckZZj$tdGjMq{`6M_$;)J+kUBqv@{=tj|?<{!-wj0
zX;*{3S-64qsId29Iu1m$?mee`LE;3O5x7WmT1hI>Yz?@Sx|(jkI1s(s+qv51RZ#+3
zlhq<fr0vp?DU8T|?NloRt!AjwtPmz_l4fp0f8yg1pE$iBcj+_DkRf?L46aQsxQq(U
zyZJcM{D88x3acKehvh<%Rx5}7*n1%H@GcCgz&w;0+-z9Rk6{#NssS<}?_A`R_4Ijp
z+s`)FcUz$hjc+1OCnGLNTX>8&Kk25^==7S!fX7R#s~f9H!M4x6pfTT@-I0?)lO3#7
z#5F^~$bleT+#z8{3l6<zK6hA|9_e+cX7d@;<M)^MPEUMJ!t!ipEcCQ6EOU^OVmdds
z{RZKuc?h*0xzHvxz=o&OPV^Vek%-XDc&v@^ZfnEUyEaDgj`rYthxnTp)e#_7S#6Cg
zuGU{Oqf0!V@w!9tx8UR%&#2~UGrqj7qL^td_jJR~vVt1>D@UsyCQxfFib=`-bm83l
z&&MvqDs_ZnE*@g3PT#ZWQ0YHSqQhB@*|0T}4LJ?+*ph9U05&4k=}SQh?~3@~Y`sfe
zDBJ;YcY!cGyn8ODi^_=6()78%xNtyXonM&je14@a;OxB7!C_Yr_egG80vhhJF>UhE
z49@o<r}>HG6IB{jkmQa!b;^fOs(09^N7RzfdV@KC7Tm?M?M{pzwZ6Tfcjf}_XdPPZ
zM=Z)cz5Z#01!z;W$?`wJ@kTh-3Z-7_<bgg%zU#+j`kx#u`?^<Z`p_CMP&@a=cW0E^
z$#LCroNcK0u$W7GCn1)e_ZV&@Ev-&MJ?A2eB(>f0vJM+wr6UaQ7|pD>6Iw=BR-p0O
zp_rxG4fA?=)y6(^^$*f-3Zgr8qvq#9cv`4{KV3mAa67;=4zP$=k>&oT3l*TH{PJM2
z_=-pt2_;)w2_)gX=G>m}QGJw6;!I<CYoH|oy_vvyimq(%`mUyxZ=?;o#VBL!sHa{+
zOA_I```-p<2U}$ZH`klwtoh0Dl;Zf-<3<oFMm%xeiphwE|22vI-KP3mzedZ}ySwa%
zo#146tuGg^&i`li<mvygAa3sRA9klbG$ce(Y?uUNyE<4LRJk}NxnhUjIVc#TCS!>V
z)Uv%&%4##EgaPZj3t`GUr4X8<tG@n_L*^<fB=w@7V*A<gC5|CwN2TrR64A4B<tnH2
zQjB)fwavY+!E&tcL5xJ7)El*4YPms@Gx+|h=-~KtF(>HL%IwD*@vjgPPee+Rmc0}r
zj&OncJF{;VHT%1wh2YU$NY##&v8jftad#|6Se}vhG0>$-JymC{k?KC}@en1!#Subn
z%N`T;Ao2)*BvNx8J5DnseXI56*%&im;k5&tRsKlvv|L+cgW1fe@zb0d|LuElYCJh-
z`ix#1&d~&7n`h|)Cr2p>_m4DAt!%2!dwiQQt^6yNR*=~*|3>f)PRuDA>C$L>C~if`
zr_;S21jwl<&RAF1%WTUb*B-qLU8JN=olwklYPGZuTojRHT+aAJyHm)c+y;b%$>|RZ
z1oJVbk#-hPE{X+mS;1y*70zP~*FK1+9Z=SLftdbCE!!yy^#!!zVS<R(C=Fd!JEAdx
zqiOZmWAJ0P$)k0hm^`xbA_JzJ(xL_v3tWUTFgu47sr6y#o<F0#&ghsrytunaY90YW
zchkq8_NCZd9rp~9Xs9Ze)Qj11QL$4};ve`wJQ*=*R|ZNV(<Q_W?Gur9;4TOXCzg-m
zxBaiueDm|lio76WxpN;T07Az&TI%Mmc!Qb$*uvl@)!owcFLq_-l9GWkkH<>AE&Qwi
ze<KDKN+x@UMryYN-FeL)vp9B&QH6)OwW$wYe!ct=j$ONCy3Xvc@oT{Djg@|zTTy`V
zJKRO-v4)%uoF(QNK1TfEb+GssvPem)x1#egsj*)7)w$sGKhk4UHJ*<W>LuTNn=+rq
z#97?OW$b0|%`zYGMT)N?aJvND;&uQQfcy0>RZVS!d+}`%?@3VZJ(8fry<+kCEQw==
z0xm`8fGgs{LaNc}=?URvA=~Kme0Y3WrpF_7<Ky#=Ht2$=ht?fU)y4mOv};~yR$zC^
zM#Ec{LS0+c{+o%?fX^Ua(t@DrI`UlfB-SGX!(HMw)aXcQAJi$Hnwm!HBncB2Z7#Cf
zPo@F6_A{@LFxmIK@+ra{pGamsd7{&YiL;;)n&Ju>WB)cQ61pOb-hz@9*hP^3p!1-J
zAWiveOngU=z{TJj<QOy3?i0zQVARivYN0EXJJRDroW<yLtp{FuD2<$>EM*WwYa>2e
zI;@98{-}H4KF{XEPU-h&=&G)2v070ZSlO+-rF77E^0X`7^We2j#e>$o5C@yYXOTAz
z$nHPlNF@WqDa0b7GeT29>3$2-HMm~1`ca0l>9CVofsJ*77e^02(+e&}>a?g38Pz#e
zb?p;B<!Mi$l6<36dHZI3<+{nHc||B~$x?bO2WY&q*ZofEdXO$G?GpkONP5N>h?E_a
z3yYn=4JLuK48wtY&GKPQvnx)sW8bG6>Q<K2QAm;CPJ=!QeFOl@@>pR*yu^Qo1(M}p
zbFagP3eI1&=}Uxg#}HY^%pJgvB$=Ewt954`mYRkrk_bYCwmi=!ga17K8+5!qce%oQ
zx%<B(-qYV#8-7fDFOZY9yccptDva)PFmIk6A$C(=!?j7_xAD<$spQo~afXuIh||61
z`Ws~5BuaGwU75PRO(#<Ov_K!lOfIr7G3ODL>I?bfZ<guLt*zDUSTu8PRUmYHvTQPr
zzm-nw?EkgWWun1c-Kbg^9Zh!Ke2`tQ%PMR<v6Jusp{b6>^Jk~UG7{*_jRM~s>s0Pt
z<n9!)!KoS9Z3AQ*{Rrx8gO5Q6pk=r!Yxi2N=@VeB((K8N&o|&XF85)mt<p`yoIhfj
z!Y-6~6wg$@1y9x)f4$Ss&F<pyS|B<jwHKP}QYP~n1QCJVVm<gfR_8Hh{t|e+^-&*0
z?JIfhU>|pI?01>hez8zUvYJ#<)Ybj;lmW4$(gK7W9g~un>8nfZ)$ZA{!p`Z98L|n`
zj5P4Cb;7aKP$7*y5AAYOwxw_9-}BwJpZsQiLhm);rrPQ;xx0Gxn(`!OyPq=m>Nwk)
z+_g{XU-PW%SEN=XGH~mJqxOr<_NyM?c=gJkox!qu{uSKOglk~cD35pQ@OkxOL(rU|
zlrdpre}BJk->E=$_-_XsI!h*emd?q@m=6422s=N<9g&>KY575iHTH1NCZ8S=mx}2S
zi_AWZWy7{}!@q}OiqBhf26aN;X9ZVMy?nHB@yD5^Sj?2d2t+|@?|z!iU2`a$<%+VY
z+w-I{pKx2X@!!5WU(91WwNm=3vu+o7F*p6h3qW^P-G-_^`uXTu8R!LH)Ii%`YEwXR
z*nr<)*I%XxOr=T&au)cMzb{8u!V(jL4L><>feS={ogpmkiMprFe3`pmsRBKV-U>;Y
z$BR2xAJ9+kNK5Vyq<p9T!vZq)H*47S(uLj4G>n_A*KfH!bEJ72ZmWe0ia5IToG+|%
zHy_9J2Q=52H#+@`6i~%YeD9=+1Op@!3_1LMZGRu8YT;Mvw~x{q2W!-w6U;&GjRx2F
zBA+0R4aJxQ)6&0gOzl4gAGK}gb)nhrEdDpJ!*hAkej(a^F+Hak9;8mi8WOp<peXHz
z-{{+S)B0Q)i#=lD7cfN_MDqB(jBiGY{mRy<TXCH-&%U`2dH2(PVnk>}vk6$LtBO&_
z$w4CX+=i{+GKr@N`L+%;X02{w{<VY}KB1e{Z5%_J8_7QLuU(7xe~*gB&hY#Sb<_8M
zq2pKaUV~zjBp-^H`2HCv!kHTxy{gxJn;P^%WAGLzscbVElr$2I(VW22C3nKRkVt2*
z)x$9no}L<8ZI^rQ@3Y*ED}Z4;@-9QBKz+ujoYASz#+og^iZ!QO^l<-aYxU3iHB7IJ
z>BiqfDUxm)@qDqEaz}B=eQt$rxY_TXKn5)mMFq7IA*avV78-Kubx$xhl+S5EemP}x
z7^R{Vc5&cjVPcvX4)S9IW_ZzBO&QO9Hu#xkZ#L?Qt`TB+HVB7Z0DPwA9kH_?R+27B
zllWIGdC{QM<(_<VgMXX*J{?HWz5<VH^mnNwS$5%0c6YJurwg8!m!v~4jJOWFCjan;
z`UiyAQ~8q<YF?F)_o6oKb?(gRCYy(c3WF7Be&74w-Td9_>r1Cbn2H2oIIoB>QdUk_
z944t{G_{c|K`B5!oY=fZUsXTQ*f%b!(yDlMetH^gR%kIDl28=d^pRu6gLJT_`f2U6
zIJHqS<U5=}eyU#{o`Pti6;K;(-2v<8d1ZU<<Tt)}Ube4GT#m)I9U<+i-9hthtrMh;
zLgD1P*P)o~bV);c?a1;%IRRUyNLH#UXO`{>kh#6EJrK<uBLEd#En2c{s#h2sBudc^
zi|H=<Z5)4u?=aUYukJ)-;_TA9Rutx{hF%b&p57w2Zm4h8vOjw1+BE83VWi#60wsDd
zM%yMO+pBemWZdapT*&t^tE%qXK0Mt!JUQvBb)UwB#ON^D%L*DvBEA$kYBfu$&hr3T
zm0#`n0P=^JTb=!{JJQxXG^l;A)A`+Xw&T(EK>z9|Zj2>UgyUkN3vzUG;YXLemz%)J
zLUnsqRF?alB1@l~z0$dd6XW{TMIoHS>tZC959lTX9WEn=?_M+?Wg4<4!#KS>H@HF=
z;sb}6*oz&FQ9p`NpP=9;PyJ8e9V9ix0w<ISIAL2XSEt8V6n{X_A@%=_W;t89p??V<
z5|w&Z+#hwm?<Z96(LaizZ(dJPee&1C)F;g7>{27y%a?i&?8E36lhD?)9rUttp+N8&
zM2y)-Kf;$<z)rXS(+081!|G+Cx{#&$+ir9fKDXQ5B`PB?D?MF~ylJ$-ec>mif91S2
zkhPtPQ2&OUHmR7l--v+=G`;=v5Skg!q@zb4MQKtAwY68yCnY$4cM;SJipvWIr7(&}
zj#R#z5{4W;A|xpiw1bD3p!B=8(UCDS)kjb81*5q1&1<qVY8x~m!`&CwZAN@2+&?vt
zTA!A>v<wFji&DurgIa1A0&C?oCLkQxs1@OvHGhLt)MMCbKl|U|KAXEFEFyNAX*-y@
z!VL$*hdZU-*`pT2LcW;10I7c_(X^VSw}Zk-0!z@ZA|k|uao8?|L=8G4y&C7oq(tTi
zZuT*jfsG@L@gDiiD>G{woA#EL^Ok<m>43^<%SK@+_y@P5!7Yz*e=GDqAd(RUqDpVw
z$3I+SOzOBDNz6KF^_fjv6~V$kh+wfty*pPc<KZ(Zo5%WJJWk(VBp#IREd7h$Qm&qQ
zC>lvF0H@JRbsSH844zMAj^{A3a&~jp)mvL}E}f#|H`MNhNv4Ooq3Z`Wr|1#GyI!0<
zG@(@n=YNdJ82D^ewNDNJ;zBh?b+zf8DUU>CU8w~@giTCR(hVIMS(eo&xEv;z&^&I^
zh<U?pIEMK7BYaku8v3K3H{!U0#9$<8D%!(bH#f!{I>W4b3EB5%h^@ko#ra*nr|mC8
ztDSVxC3SY#h&=X*{6<-?q{QkVg^_GhZCC6zw%xD92y3b(3idz+((B0P@3%#P+(1os
zMpLJEE}1!@J>E5!^=-!5;gA*?bdr$(r5}aSudJxQ3Q68Z&#t=^byjVTbgs745%K@x
zI5E}nH&=$}2p;Y6Fayw1sXXM`np)7=Wq@QrU&)u?o)HbSp%32hT8F%srnq-SrZI@2
zUv0eHwelKSW(yEhxy2KykZk@u^VR*DosQNeiKoq*p(|kfRr(T)YtD+|)4ucv-pDRv
zyWD+4$0~mq=J|s^#g{&gk0b*R|7q1RIJBNX8ax}{Ke)Y}&;G0TnmUWShVI2%o&*+t
zlrnv|EwsE%Y%N4R@U82{pGfMTTzoIju91vHvyH}=4VO!R&i7miIJ?|TMU}g=e&KN4
zsKgJ}=(BX_#|KTtI*wadP;$Lg@HBNiQr*#f#q8&+^f-tCW#GzV*JGLAxS~x&GJgIz
z%;4E$`_c$a@2#E+;gSz)&6(!tT&CBr<6nvhvX$FcX!*ROrHOwJ##4*M4h32zkrnwr
zc@X`im^GIi_*?3n6a4WZ2T1Bl=`+TUIv<mP8G(*P#n1FftYdgzzMI;*qJoPPt9Gng
zuXcM#H53t-GF@?-&u9KtLR<&+2rER)3;svsP`T_~x{jF-89cBWDxVd>9_PKLO_EYv
z3>m}58%b*g3**3rDB$|RQ^4ng%pT>-a}2x#G!K|CAP^{&epVkpoW3(BlJR>BVbOKe
zsVZw&a1cFO`6!;7Ky&YS!zMUK^nUZg6~dr_UTuE4EBRmX$N4^&)9HjOtHD4Mns138
z>XRv^ZHTeh$Dl9dYuq)x7VxBg%_4C*aM8}<s7ILp*Xve#`Zhf6{hw}vFhy<^FacGn
z4#3ifHj3i4i&>^+#-wOg&*FN>%oc_x{<f;LTuu&ODmITy+zYuFar6(t5cO?;m1J)C
zd~bg_Ov1>hx9?SHY55_(XCyWo<L#?lfV3Z{V+mPGjx4%nKr;^TbU0&y3*vR*#Ej*m
zwudT88R=HStF-fv9ZNMjVcppuc4j_j0)H~`<J2{>&5v%gfNw|gfK(|ms}3cJL@M`}
zYT_{&{oqfG>IAMbr#?bag0Pv0Z?o>OvL|AMzX)qwE*?XOm~nrSQ_*jh^$Lw<M>-&D
z5@8z72~pWi?lS7^*?BbPGnNAU{_1HI{MT2U%r`!H6(k^~?u)|2$qgO|RT_%`S<7Ap
zE1TbI{(s=DzuLUn$O7eiZc}>Ss7ePWI$-~y^zGD7k?TjdkkjnH=;mX9X~?grNOJ&@
z`3Ft=19Hhf_NRSkLGvQD`>9H@5@RAr+|3_moyb)oB$zm@?MQhS8VG(X$VOq{>mGCi
z<S$e#^Fr041I6KsiZ2Ir^5cG&z#W@41#G1%gSU*beq{8BJW9nKtbtY{+AO>1Kx)*m
z5->BnP%{$gw&_O`K2G6dB*)P_mO#&?2~S+%p(@oxN{{D%%{eUB5f6#oCrPFYQNhpL
z18Dx%k8woC=a3um+S{rgd0sQzlo&DmTX}44S|$%FlNn2EKL!$>8Z#<>A36%RB{%Ya
zL`G)huD56+QSkfAq$T0}c(c9jYJD6lyYpKN>fJYi3-OYY^{%MI2Ijsn_LgN<;vsQw
zW$yyF#7&`VXLXYvgm5l|-In|=mSm3E9?E>?*f85Vj-%q^mnB&zkCP}JxGd3jC0^y6
zr{!v0FQr|~fMo^gdnF>i)+;5>rbW5N!t(OoUbLW>1CbMBxBk0c{oUpsqiACPDTev;
zk8GK?Q;SsAwr0z#+0`@FQ_uHbE>(@mUN$E9AGtLT25g$$68V!U;_~8+$oI{P?L5=Q
zo!Oap>y(E+A@;^nzPG^Q<<2)^`njI__f8Vzy#{p3rTqaZQPPp#48Y#~GvW?LrdHAg
z<+UWzul!)aT%KO?yl2l=E0gDC(6^$khrO9Ku-G6DJ8hgQABN-m{N$gmG9!}Zt~rvc
zU&`32!gVja{6kYboC?^t1OM*t$`5`ms}pO?V`RE^1NNO7WZ-7zbLpJy@9S^;?A3BZ
z<Vy0C>kGgHbH!=)lFpM-mxcNEqy3~jg|$A`#1AXYKb&$!hb`R*dq0wmjcdV%q;RSG
zt&cpnUPd)62^KTN5U$(#IW~Hd3{71x&BjQr8xAYsWWTNnkJ0^X>{iM0`eE6B#+CJE
zie@7kl_plTBHzD88vEllw$l47ovJV_w@6G1^<+RdMxz3q&91;y^bqZ8aq->H={+AG
z8GmUc8fxfjJ&~4{b7__aQ`Zci@jzUR21wEq%QmJvA2qW+e81vUkm<zg|M>1*5Z|3Z
zbusA&C2fC^rf;xm5b9RDJl9{E6aM1=`kc4y;p0<JbSdwP+<=PVe}5M-6ubWHLH&R{
zs%;0rveTT}sXLdC5$n4;)pKAmRh_}$d)*^&^%@4V%w`eechAXcL`qh)W4vj_<9>?f
zx!clj*;oC2dI0Z|;n0H^yj`&R)fz9e88|;rkGCI!9yre?r8_^a1X$Cu4YT6BCuk{(
zuqM^+r^Mnn-VEY<UnZ>x?GgL9YL)ZTo^fCYn6a9R8Ie{Kw|!m0QPyNqDL6Rn*K8@W
z+OpKL6r7uk+Lss<mNRiPz6IxtW@+9*cuRWxo8CSoj-EAQrytn5vIf|^*>a~EMCwM`
z`{xYm4;B_O6r$GUW1@$y6|+MT3FxX)dpEo}W2LSAorHXQT9%LOqg1XXkyJlKiGaqw
zOy!X=B@ldM;?hQwd%P`l#elcJN{elSRO~v{dwPmPBK99b4K^7C?wVJV_E*9D1Z&UP
zw&sh4i#*dnQ+=E5z8(15XyuJa3ek5IB_cy2elGX?3rbxIN(liLI;`)i-lprL_HjWI
zB_lmUWZz{9U%k3KEdFKRrHgZZ0)R0SDYze?JInGz`D~2t_j<c$!R1c9X)uZCvHjf+
zNbCgT6LnSS%QG<$2x7M(YZ_QRcX3dm^{ck-oG;^BUG8q&i1!A}k4jzLHCM;VC$^-{
ziF-!F?|sELg$~#4AM<b!S;v^iJZoJvJ3CSruKu*M<ziqo@xlF7m)y)(z<l_4`KV=G
z2Isw(&joWbVOf*#9y>`|+DP&4404IFTms!Mjn6-q+G_O-wMo6VTkq`kA|aFlG8FVK
zjqN|7A#EfLHq3@ATh;Ihh0k4SKO$t_<tHk*ZRvBz#&di{vrLs31o1-`PhUajwn$G^
zX|Io;`Zu{GlEg0e>Qqh`2>xaGGhgo7oo--El_7F*!pdrgxZ@#{_YD2=S0u#XzM0DI
z;x7B(s`G)mYZ$@gTi1k0Vo3&=pHVvG9Xg(xT>Z{nq2Nc29`OV*1^YsFt6wxwiN_>O
zhtjwp*gXyjSuy8sL0i6AbZ#{E)6*hjEi313$isVlJvrX}h^D^)<n$AfIX_z*)u&!z
zFAhhclOjFP1T*R{mbDe$bN)5_H@XzIWX1zcQH$3`Gkpg(O&KWREBSKXw(4I3G8;3q
z1k}TF1g388e!|n@bxYpF<k4$y$J9*kR997<R;ih?p@-I3zo|oSfmM;?fq^?DyVo9-
z%kf5)%aHJ^JBN!v;=!+$AH>osW7mPcF!@5g9^6S#{^08HLQq&lTmGQni*iFZMn*}e
zQghaFHNd~M9n%&b9(Z(lh`+GfI*Su_r7ReI6-y>gT0GD(6Ukcb=}D32)wttTIa8-j
zFaG8Epa|Y-ek~5qcY_#y!$p!E1=Dd4qT%DCN|_f0N*#dP0y5k2CU)nZ+=j=0THwdm
zZ4P%V*Tn1m8nA7;<cv{i<aOpF#<@K!D*j8Qij7u&joiTH?gazyY)e}^WT-jujL@)Y
z*!wm2gQ`e0+lx4M>R-rl$I4v)@|_$K;$WzSE2XDet(^{1;&FZL*vH=Wl&|V7|DWtQ
z;ht3X-lqVeo|n|xGS8gkPMaX}tt!p*Rpv=nD=*sAHl`p?fY(qYOoVv0bvlCL)ntht
zM4Q&aPeMqrgCcK139GTB>=%$Z_pkhpI@lpTg8skUW^3KaR156qYE!|W$<Og%(`(b?
zj{{y(ONuxb^HD8<b1=%U0&w`vJXw{t;0t2J1dv4RU?Hb`R_+VyqF#d&zQj+9b=C9-
z=X_lur{*pJGl?@{aszO1-7FVz7;>VdO0B{0S>+NGO%RjggEFJ`_x*AxBQ+8fB#sS@
z-*dg}J>1>>8zyH*tyB4xn(uaM)%eJ}$t)eRqduv|Zl)E`imSPHTvplN3ICczduc&V
zATFB;e796wY-HyA^_5T>52Hu*Q8n>@?*OA+vo3npSB@t+MZaftl|S;I!;X%NgS|Kz
z|EKyA3eL^;r<{?Xe6MYm$bhZU`k1dsVU7OhKPA2%MV)mySBZm<gL8Gv_7oX^J4<#|
z*|;9MBwH0D`-^mvG_zck3-nfp<5NMv%Um`}S_mEP<$N4cTr4z$zCyFyJcE`3)nz4I
z)tPS_Sbwsf#o#awOog4E?=CDX_$iARC`F=@dug~VSxgz&rOcIDX}j_yU);)Or2W-j
zfqv=xhfK0315=e4MMI`b)BlU&qg*6EiMWVqo3oy4Ia?nORJmF>L=-lfHbQ{IO&6xN
z3$3^+XT!{surOur!V;mh58x)3%6bk3>Zl(^0_sT+Ypc1Lc9QOD`>f(@n&jt3JiY&^
zip)ay^Sk;-6z)v?kvU`LdaEQGt4%4Tr7)orL+|g3Blk;K8b`HvoDK7uLE!Qo+$jF=
z@FGuai>E-3LBFSO<W3_}4+JE2fM!ArEj&XWK5Ex|w<4<7BOO2fdzX4HwCqanO#yjb
z3i;&t1p{L<-6d&&H9AFYZCMtB)RjuUxVYF(QE9o_#pP8Hf|L~*VB3LMyz|?vW}yub
z+9W|(a;4<$4Bd2_f1<cAbyPO_^_tSP{L+)5L;hz=;3@r#2A4k}3czDONBb@NB<`|h
zr(BUIqy3cOywd_kqh!u*WCTMsYI-r@<}#<6;t2oBJnnz$baE}$WuU00+kt0T+0xPH
zppH(|KI5LfEBY3A=rgCLw86GLSDbQ|f@pJsRMMlDbGOqPMa(=sPDd?j>6^L9EUm1#
zLA^BZ`?cth&2yiydBFDO(NVQZOUV8L{*Yk2RZN=ws3E~tx~vePA6yO_MzURxzK0JW
zn`${`43Y3u-(-roRKx3<F&?nE8v+@X+SWe8qo6FyNY(s4+#<|WE1c;&n(B@bYY&Je
zd3AZ=gwzmUi+cn@W2@M%6?-;@9T~d4^<*6LG;@<Sk|y&7C~5o8wwnkkpoCWt`|&h0
zWEyy3>#d1t)=tQqtIO+*veVvJj<dE0``_r3Cd8wR(>~u}x#yRL&mEo{zcripI{c!L
z=;8i39?8OPo}o4U=~{29kFmLr)0bRfFgZyXlS)T!S9Sl7p}&aqCzi8_a%gG~xQMek
zC$v1cmsCdqCZj(dp))M8BS#nDK&*4btNrS1CU#9}Dc-XvlZcy;XIsBL+PUomTL~U#
zZJQyhgR6Ud_8=}86CXN<hQA6nfB5F&cCZ_swsHUcM~tzrKc6ae15-z!4bK{~@}U<&
zMv~|pRhnSctvEX%-1q1o?yT@~dGxF>K=#Oc!WKY^7QH{b@v5|bi6r~#BHjDX_Wzjc
z^a`$;$Hd!bb-cdMtV&(9gP6veR?oJC9uhD9{R^4c`8^>yto`++{LC#yO?RoghFQyD
ziV;fh+eA**Oh9!-4TP42&vTN+V}`XWB?;MRtb?vy<82$l+^cHRuiz)|QiBzvJ)VAW
z@&c?GyS87Yj2V8fu(!YbeX#5E+w$Gs&PTqman+Q^G*ih>|Ngd^h^y)>@BMofkB)0f
zuvCY1Uw9l{60juGs@C9eS%FL^lE4)^xpg$YuUPSX?+&p!p`P1kt0%i@{ajd);>~h2
zJzu@$$_b0=+`^CkF2~0eH?bO{bQa_(><+-3ueSShu>-=;U6A3v;de4O=|3iM%%H!1
zipqth2ol`hd^R4606Rnu#nFLmiBMM-%=W0#+Y#t}9c#|vUrLFr#)9-ZX@*bfPBX$@
zqa&L8OP))c+Vq}AnV`NkD%%D%7qpc5&{KK*f#X5&5A@lme&K-^!=gJ8Cpd)c4q&sr
z^KbN8)9zZD_&*xs#Q03XBP?AHQrs}%O}DI{5@W*87TshDac;f4*Jg-ccFx%Oqn?yb
z_FKd=BhptFm;Z(-KKu>6+})?}Y1w=%ZtnW-l@Vj|5AlN45bK}M(Y7Dh>0oSRl=gb@
zQD5r+WabluN@75|EDlod0{BJ(n6-iRf3Ev0cq^4I`9iu3CauaSv(bISr+q4;Pk0?D
zyUaJ2{cfz##Lt|y%@G?o%JN@4r&wWE#Z^CyMvs}NmD!C|F1HQ6)cEIn&Dcni0_04-
z)cpKeJf&4kG8Dn`TQjQYJuHfvTg}jsjHSaoHYz%rH40jszOQICkk!|GNM#9#FgLQv
zcH;vaK$ZDgofJ=O>BOP2B8PwYh&)Z3%-8HcebH#eOnG*DV?U1Y4)poK)<zJ?t<mLq
zb@@8X3xwV82?%^YZKOK9tJu0;{zXl2Y-8TSscRv*8{R0#TNfr4{jL^7-ouWm!?v{n
zN0%3(SEnmIDqA~qhMbYbSAhFFPl@`xkzj?w0^YWW%j1)yxjJ2uQFW5qZdqc~{IqQo
z%T5tL!XCVKhIbA;+Q47r9g*rv|7rI%OhZNvx<0?Y5hhegK_y$>C&jIY63y!S_=AgI
z{0?K<ePeZFtQ06UTgWJ9AUIdM-0(X$M4Z8D#uH&$U1j7d=ocUd6s&602j2uo4ZWHA
z9Y+_i_ovoHYpb2(!E{N4)pEN&?nEzCP!Q#bZ`d5zp4fx7{Jah1U6*ARX5N)cYLso{
zoo#Jxo;rdRKY(PDT<S>IJX4<bP3BxOC?YkkYV~jB)7k#YF*I6z^t~?Zg~dvUe|Y=k
zD+k(`CC!aEQME;2<;+iDLxC8&k={h{YY4?2jl}Cvby%v?=~ldS)gE3}RN@isDR<n3
zMm!XD@I9xe05#<xk=YLs5slc;#ta4FW;7ZG>d#6X`9oz)m+GKdp28LVT7UoZyt459
z(iQtL6lKgEQu`Yr3Vk0+;q$qaF3<=557G@*9Y-*b29uY=q)lFx%Ol{a(yODqi|tWC
zQDNm_kzO)|=x4fJ?_O?i_ttlb5a#CS!hvSXf<n~rddB8A-C?QabW|{=wbe~I;BmId
zs#EUI+Ac8N$%)@HKKkC&qw#T_QPko|ZHD}~Q116Ka*aam@{PDppxN$d^~f%|qo%f>
zv)bMopTsIr%}a)fWpd<Ch^lLdAbAxlsu_mBff$&d!yrU=xF1rs!DyS)chuSq94*r8
zBeLFxPznP#!`sULmDqaof2nbJoc!6gPZdsDua*2naGux+oQc)AI9S%wH5$JefLs$X
zWCK-7o0dm{9Y~SU02yWcbJ^Zkc-5(t9huspRk=8H7_a8jWgk$mit71VCT4Vli}dxe
zDhGLZXU_!SqK}m0#dbcUvmAdNU!`^b?w31K)c~ge*q3@s(@<k1RW3&CVBNK`88fxh
zc>A%<ig~6}8%unD(#lkdgUYL>+dX;Xrk)Wkp-1~Gr#y>dY#cw3pJD3w;$;ui@b1wZ
z$wgT7YIC=*=j6PF`**_*Li}l#d`ZkSeyGH53&S!KgUp`j6enr^E}U<lk!9n@oio^u
z0VuijDjRtajCAA&RuO#%mQD#|y1jf%KCv4hY60UF9c8Xk+`^2|ne#m8hhp+#z}LxS
zT)>t^JuMUx@%-H@^rky9FN&V~eHn)~nV<9=Sw{luF8Q9c(phEj+D>?zZ_z68`^=sp
zg!V5JkJ{arotJDA&K9fkadp$%L=O*u=;_(G$gVboo_|^fa~|ujcgMV>f|~eP#gaeq
zbqOk!De#pQcorL+zUac4eAm~Zl%+aX9&Cb9X-m!i@jlglt;geQ;>Cwa#-dGoYAV|p
zcBD^Q+<mA{%qR4Ieuc`B4ez~AF4h5ggoyU`)>b+QsFF9D>~^d-5Lj`?*cAp$1%UD^
z&?aP&9pE1{QeywCTH2=KU2pk|in4N3_b2eBo0hD6e^j6T*hB!q&z@V#&VTJi#Qw%V
z1n=C{`DV&$AAkBgiE+#R7O!=q1o6k}RHkV{PK;@WWZpTxl$;K1L8Ix6MVYlt1AVk0
z;}|v2-D0r=T*jfIaG&3#N5Cz>AWA${?wt;my})M;?*E?pq%}*$Tl?I4B0xXe|8cGI
zg=@WD+r5_UscQ7AI3)1~HOmB9wpi)J;V-Mt&Qe-Q>MY<MJ8B5>N3BqOLfOUt8kn2A
z!BxIlPxO@1D%jY1=DTWmA2S?yNg0#RB|B)ZhczS1q8fvZu}O_LpTK1RQ(rYi29&W{
zC?|jiqUuFnO<r=GWcWcsAvz~BKQXt*&fxj45Cv|Opf3`k70>nP*Ji+QAnaAcZOSc-
z$IL4x1nVCogJ<4ifx~EYE+;+_{0rvg*EOj#pi|)BcoX82G$ZLdS5mR<#i@|z{J{Rf
z3Dqm56*9)$Mv?SecZX&GjJS_WDvy^l@vM%JEIW?H4~Z*8iL-w!N1lber!U_LG`(0^
zArexKb|NO{T5l4@_{xmk7=`v_<_bz-tINB)n>JUcBo+P6Re-W$Z6COr%w$_Rc1U9b
zE0tWn!_16!t(0);1Y@>k4k*6UgDfkm>CZ=1kTU^l)%BJG2%v9Se)a3mEt9D`PwFnI
zp#SFU*BqOL1Xu4DuMBZA8R<@M2S_G6%y}Xp|8UM&H$<OHfd<+FH$MdInJ5@YW})AH
zF8Y`i%QJSdns<HmVZqKulE#hlSX0~O^~>$7u`hRmDdg=o2l*06pZ5w34-NDUZJj=Z
ziAR5K25z}n01=62GCpsX@1yohfIsYCHaUK{2^XU)3!}hVZfS-o8#eLRSLA5(2m$(<
zaY-Y&LsLqqWu;z7m&eTPjQ;B4$+JBl4uptUK)i)ya-K+uP-5}(gm792rCLXDLA$ad
zPjXh#PIWbj8!}A@VDBC|a`i~o2xqVIkc$Jm3YTO6`zJ|U+^o`XATe-Skzk2=c$jZ}
z1$0<34_OiU;G@5I@Vrwm+D$RT<=w{ZbUKcrXZ4m)_Si_+whS&nl9T``Ot1Uh8?<pd
zG#}-@JY;C6$H9qw32<jWIjawjKM5yc&K)y{v?BgAl-P>^>lKNt`?na2mR46^NN%0F
zw>JX=H&#mqjAW8xW;_FoUnvjjBbX8N%{Pk}x}WPRiE1R6?7xZ>7OvPXh|b0gWmsDE
z_9ct*D@rmmdc7#kHne3nfhNJaL`st7Ys<TTc$lCj5#q652p$bt-qt>qrwp;0m&rYk
z9yEj5W5Iih<M-{vR<lp>{)2>`YD#u?&nh4wT&0d*G1DID>PFS!E^O@kF79H(u<xVm
zt=%kv>xd2{QsB?mTNQ#-_T=N~zOZ7j{rHp7(=bJ4MefA4fc<Q8zqGgl(k5?6V>{`B
zd5;q8Zkl|B1;gt{ZfFvQ4za)bGy4PwzcRw?gf$_Q8}Sx5WuveopX)`eJNEoUQUDuz
z(zB?RkEE%yO_lcVwBt4>ZwQi<;?lO&O*2}$&+#Zek|ISduzq@GD=@xBh>cuhh`s4^
zyxk4)*a%5dbw79ccVShf?P7P_E-<gcRJs&IybpD)A6wi#IO$tOt&-|})G7)wSlIqD
z=g6UltN+z~vrk=-YASwKTbQ9Ez=exsXjZW4Qhv>dGayt?iD_ElGDf<F3@-Zc1A1Wp
z;QZXbpkU^O*EPxVL(c~mN!&~jojpttHS_V3XpuiXsi4?)R-w(S+UzcxBOX0Op)bQo
z?7HOrmTcG8`XG?h%4Eo0AGq($Wb!+koeaI4e7CKs_~-TqfET;ej0({}Wqx_-h!kVa
zklF_p7!_@lQBwIPIiiTSI^ZnO)4IL6eJ*LbBi<lBmygFHO2FPo^n3mk?1kCJx|F+2
zxQeB})ytA;dsfvghaO0_**isi$;j&VKUXf<MX`Su6A64EFfxiWwG2;p&mVVr8h3!G
zwzUCl^j2LY{T7F`ZczcYF<`MoW9D~U<l8cNa>O`-^5N@;JcC)oCws>SQnVKpvej3|
zD3$o~vK$Qf(*$9*{Ta+zLDW~R?Z4Y_apt0>&&eGwYdsZ&(u`=36DCy$o=8id4SnGm
zl_t~84;LFpZ@3C0LC*84dLRCsAMOrbEdtB(f&AUG^2HotD1hZGgNd;TsV#ns`H>MN
zksLMc=4absC<#qk)Qmb|iWCdz?e~G%@bV63y&AaM-Q5jR92nMcZJQnUX3tqXky249
z<l$wo2eIE15D-C^1eKd~jXdplt%P#Qed<Ck5%2%L@io_hbNUv?7f6+B<^7@9u6|J2
zoj=Z?qD{lO9)Jar(eGb%;-g5aY%7ou!=&AZ6bo~^Kl5rV3_5WV?+1D;-AqSd$VXKh
zrKIcGj$LND#@}Xcc7|nCel^%k-JiPwTmHOULxYTOJUa3|IX;%j!PXfnRox7~C3q(W
z%#riFegth)iVNPcc+xHukiXF)w5mg7w%$SDSMGEH>MX9K!zWB?TZTPD{MjF4*%yam
zMV4KZ2Zo2ot4*if`#wEO!rE;i>|no-dRvU%Ii(LT?ENJzI4BkrBxm7j2EH3l#hP>X
z(~d2;Qm|B+4n9`cG}>|?dDS}{C=Y$%uBvVrKYN;hw0?SD8tO<*sn=C=yZZuvc!8DW
zjX2o_##Lrpg9}e9?|t@BaVRITwNbcTiKqJdgEnmuWR#;nxn%Mx1+~fWjs>+#v7H!}
zZjl4EMbl#Js`;(8ZA9Te*&}v^>(PfD+VpiM%}}U>5&Tm+>9gk~i?Pd9zhK9z;pOI)
zNj#>_GtBCaws<U<Tv9FrYW%#b>lqz;jvRnQqmM%vItqq2r?G>Lfi_h<BnM#V)HlB-
zD4%J4bFXdOMim^hp@jt5KEO^aoH^IbHItHiE(1EZ*Bz}w2bHOw*!95ah(?Yh?6zDo
ztfZqx|H663a_6W~c3f>v(r8gt>^gmvyn&=e*F<jK42Y&;LMm%Nm?w_@%iF0`AmxK|
z@mORO{f4zKEJeRihLk9gqSQWyS4O>)hLM6v)&_7`h)wOri=^ok=no=#26;=g*&O)(
zKuXwE`SK;7<Jcp`gEVnpIAmQWI6v^D2q#BT_Vm%@{^_sTq1t@Nj=_t~%(M%W_=~UT
zD}B1O#bu3uu~Lxnt@S8d?=%AUCEiAjWRPZJG1)rqbUYhw4E!MS)aC>ikjJocpg%JU
z={6bJIyu=RX}s_i_|d1l<Onc!#C+|;nEBo-sJT-?sVyhAj$qB07QaykR()A=137e~
zlu)rzn}J$gTu&cu*BgeS)1D;-M#D}Q<hb>QokUjFJ|su8j1+y|1+Jtw9~48ifAnjr
z8>>%?#8nF8o^j`Z(^C8C$k@wwFznIj0#sri|1;Snf*5FD=f5=9cN&N3eJw1{;OEyh
zd4Tc7`M&2VirBKu6A2sieKxkk`}d2ZT-xV|0T$|I^&nMUn&(-D+0%kt*;?P9YiNMP
z#PVdvYRKWZ!>SG*Akc`J!q&0a(CVq%GihOGuiOfLFc->@1r3KKa@cIknwrWkHu=6r
zp-4lPxHdLM8YXtNrHC!8_dHmMjjdAt%=x*Iu$fQk1mua4hAODVa&N&Um}IzqA!;*-
zFv_`8%|S9dcAjGo!w(be{}SEMIrpy3(pqW%BE#q<HQfbfi4CUH-osBGOO{t@a4S*r
zm1e7Q#t}yGn~l}V=jZ==Y;1tzX5OPn$f`n?p>c<QPU+(QK5o34^!E4mr6X!y(8Egb
z&UIU4l)5vq23TcUD^A47Aq~w)n02ju^7O?tFEBMX^Gp*zKW0K@gQ^xDQ=K>13zJif
zI6cm42QL4uI>~#;KC5rKj}hIEMn=w}lDU4}6ZKKB8bIlkKftzMOhkYDp$*ocR80`y
z3Y?zK%gfu$ODSwiKw>NY^oh^@GR;lo2%=D~uZm@RdowzGIHzrs=}hT;4f2dY1=0wI
zZD-J|fZ5Mj0;d-)|7{$89}?aCe~mx|Fx}`$oNEmat)G*WbWmdf+abk2vDCPqBfFyH
z<G>UM_ToF0C6OOFz-eWxLa2SFZ{-<(UaBWWX>6oRYH3i~Yg<v5aHTL5Dj$uT23pB>
zOT~<kL74a@&fn}!q|@P2oBH<`af`s!&Xq&%ibx4z;!U^;eUiFsvH;G*cR&WX+07!m
zC`YUkyCWd+6}KNIZ^gqva&HDU&UQ~XR!INC*|y2QhcQ-j?!>vUmk4`CvY7qhk3w&?
z-MdW4y?WypJsxJe@S3@;8<Mt=!0!x8G%JgLGu1q$)DV!UXD#D{8>@X*_mj67@I5*A
z!U*Z4oq=(qhpC3TFmO@MN<OgaB5|wIN0~=;-688Y!F@}iQsN!?3e}zZ70W%tdth;X
zsz3jsm6Zy43I>OvT`kkE<b-cPQhAa|%qcXx(Q0zNxv()U%T!SPWT&|jNx*bh-PbMB
zpPrG-lzG&(3)jJdJk9urb!YpC9!Vd~xOaeK#};AKZc*Lk^BFNj?E^l|8G;o*0Yk<s
zAbe)Q53>{;e1p1*3Jp?t7d%j}?HICu>VYLBEqcl2qvR{dUJ4s)_kspuTnBVL?u~-Q
zyGIUN>mxXDSsm-<R=-&b8bT5<|A`x=C9OeHpPr<DjbdX2-4K5#gp{Jq8?Pp1D0#J(
z%HiiM4qMnNGBKnYFo)U>(7XU8$I5}P)58P>ei|pQYQ0Hl!qoFtrL?Okhvo1p+bCZh
zPL`z7<OTOtH`j0Qh%=l^c)K*0$hbncm_2(SrO)w{VsZw^X}q(fA1pO<QZO$;KI-9P
zZJgXzbro^8<zyj?fi$VuW@z84B6fM1ZJbDOlP0--1zEtlkAjklijuM-llNT8G{FPQ
zP1Dj(?QdXMt6P@FS6a7@X-kB~fnS%280!tscvgALwS=DiA5G`|mgM`laU8jFYek6z
zwL!x~O~sx2<EmUG?nREkm68)p6Erg|3tU;2JG4w(xXM}1a;D<UiDE@g9MApz;n6<;
zj^n<$uJ?7G=k+>gJFDtsQyH~;4r#)tZ?a|mA=+*}h&uecP}6+43A{ZA?OL6IgILMV
zU#KN>dW`RPTxq+MbZVZiu9mcw$J*F(!;nRE@5@i;9^=?Jf=vG`uI@1w!>B(?Ps@yC
zWq&HLyfh)tIHu?JM0(_0>m?4he`kF+CHt|L|HbVd9X22KuK;tw5;I)mRZ+M1E8<y2
z;0RA=(MPvyDi{F_OvOe(Say(1OF-^r(A5?NPS9_=B@cN&d&~(VS+6jEiky8L(P)TJ
z8*&Pzg8NWsg;WBG4j%eRc+^_hzy=2gi8xmGkp=Bs=zOSd`Q)*gBg1ZPZgCSOYp=JY
zW2v1*-mm~h1+Z&}MuupoWP6kkzDH6?dqHd1(FuB)3S+1QQ*B)axIR)(@vVc;lee6Y
zShR!n-t}QjjU*m`UIaBu1ON|(HGAq_A5N=tr7P>a-q156zQwQbEYIYW`3uhmPNWfh
zmg^^9SSPEjoE+~zEM;YCf@R)>$Megn*T#uQ$7BgeDKPgB@$}mj6#<x?C=GKmbK!Eq
zJ3VN@dRQks3~D?FJ!=Yk4J?x8-322^dhl2Hxw*iQknN~%LioCGB1!4_?0Khd;j($c
zHTuJ0*4NI69m9%JN-v`LX2|8)B!r4oc*zO7NoT_>|5v0Wu<(R%>Iae9dYulix{;Ny
z&1&W4$JxE+2qhD)43D$aZ-<L7VWI`reMIvYdpADa2G-JN679z_WIf$XtmDByVw5pz
z)Rv&Em-D+(nJ2OLWw~Xz^!%zcaUN#1l~SH5P)<8Xpx^qW4M&1%#X|}@_C2M)fv}S=
z%5KDYijvRh{qz*gIk4%u%5_U9rx@Rh@-O$R|K;EX(>DTYG_1{>*ErM|+i81Jna4_Z
ze^iBnBBP=JDUbzpE*&c6!Ec;HC0)3w7xbAD@4K$a5qqt1WriGNnOTB+3Pc28#xlVn
zR?#wx!aUArZ!2FpUJDiJ<b)>1Ou9_F&>F)xw_5Rl)xj91f?1vTqHtK};4a`10aY?j
zJ3ST5=MCY1QPjrzrn<U1$@*W&m!Qum0e!AD(E-u<80vp;(9SbDb0xU`vhnQR`tSi&
zD4TYo{qt*wzwX-IIF9GX&VfR1m3RX{tNEfEM}L0r+`7lhzsPrq?9*$ka{gU%{KV{o
z%*D0eYHvvD@9}pqf}?GaBzCCg0OnIPYjyL6^s79<iKTgF>)`wVmJy>NDH4CC=ds?~
z@v3{yk1c+TyMNyyPeUYdOH9e*+Hl}9^g%aZ$wx%!<o3HZ<8-vOcAlg3>86^AALpU)
zy84r&ptj|fvNB!}Xkh$5Ep<Ft|5upB%TVJ@_8<eLPsmNBE8W(;fuDZ<i}nNViB#I2
z@Yja-z|N)vbM@2aBi6w>utDBtl{w~LtKcTT>u0n7pj6#;sYuF0qi^%Up$2iS3;Y8`
zJ>5aJ5w`gS%>V8o6bY=OnP>SpmWBe`R(EUl5AJ3F;WNa~*<h`h-Ggm~0`b<_-~SF<
zS#<irho8Wul%@5s$EHGt#Jk14kyrI8>ZKgY#um^7H~*M#`ZGH1KiY@u>-^NnEdrwt
zfV~rJ&3sbr^G)maAKyL)a;Y!JGHI+*yavU`e}Gd|X}lan=;@8W<?r&>>08#ViUOt1
zp#sbrKs{1~Amqiky+Z3P!W|Ms%+rzYPA+66y@1_O&QK9ktZM#Q+CtBIQB=JZPc8L0
z{vB>(punzL5&`CZzUl`5)4O>XHQVhS6Q$u)h&U#HJ8jNRCPnxe&q90fh$`;zyfz(T
zJdA6(lj&(w?*mSxQdt?#2us%H!;7bU5;{Uj-i&3aq9?xv{M=$JfifaoYk3ax`UzUX
zKl@`e#?Wxig-k)mOT@m%FNL**)P!B`t<%xciZgSo@L0dx{Ohz+CVauCu(K7`kx^9n
z%K~~vR9#hPA?K8CY$h=I{GJGSdgE!`Wt`VvQHat_=Mxg2Hk&`Z_^9_5fM9UuE407a
z=xD-%uBp8yc}Qs2qE3Y<l^jbV$+Zi5js^gO*}to571PeNCMm3%2T<P(ESbz^U1=cV
zw5!yQp9ij4ac{6<KG87c9qXu16PTLp^#QC;ft+V9-|dRtEgpm9R>k6#U8QJTw7M1~
zHRR<1O&jZ}#<+0*gS9{t`B_4jp<qhBx5LX;Ch<mI4D0XawyzFnV)=1FovR_tr<oTK
zZf|dxmkjqUon0!dF4xz^3w!XvDG(04DP$Zj&DQB2#aw+sa4KBU%FTPqEof^V9u<RJ
z3~uV_@AUnO93~sN)7S(=5N@~`F0)yjY#~_0mF99bQBO7Oz4xpB`(Hd{mAa$yqlTAY
z4>MxR&lY7^uX?V@u)nue2E>iG@K9J=Y(oA;bmp%8#e|tF>0X;rugc5AuvoRxQAZEG
zFY_xZy*75Y;Gi`sz`Cl_(vC(zq!fq-FvKmH2nXosbMErScs+=<8tz`=`qVOgcDFVo
zKQ(#3`>N+^qUEX|x$BKY2Mv?0DzRimewVgae!HX_^~r(=%k6%)q@~->9JMyQNXkv^
z&(l5HKWxR1R%}#YT$>{{HHfA8lu+S}xtrr9qu<lgV4~ax_tMJ<0K_da{bF5mG4-dK
z3oYUuyCOSB#Py$095ku^awWm`(&8DpV<_X&e9p1JG3ePS_b$IHcJzg)dj+@r?oi4t
zIZO?qfn{vc)~oHMh^)^lcd>W9u%aKG+AK)={4m7|m;VA9-cbB15NJnMNQ%g$k^E9Y
zoP}e%ameMzBNy3%-dh`8J2~!(Hh4Y%XJ?bnAycRzn3Vmr3%3KFyD4sF54+0h*4^+8
zzlJ>(8rua3Rje3X=nG=QulI{rq!|JnBXct7cd6p~1c=9sFu<kDZY9l5uRiJFcM`Yd
zRZIYT@b^rO^EAGZGfm+cV>2{+Jg($k60Dwkv?n4Z-JBW?#w!^A;;Hj`k-`JTx6b;k
zVg?GJfiuFm<YJ(CSw5$*`_KKG6yU17bh8yD0Xb|77U{{p1n|wAkh)?r*X|(Yq)Bl1
z<e)Wb?-(}vo82+U_!$#WdFhF+h67E<>T?3_V`~AfM}Iqy)>j$V0JY$6)c#*q0PX;H
zIDupRU4RnDpT-+J|6{`kee^-HCOc@&=2RxD{5YV$xXCY_ewlo1hz0I^vjhfyJw;dh
zj}Av~^cOu28EcBbzfn#ZPIKMAcs0#bKk!{_U;E4Mg?}r90eLz{0QkvBek06t^ek4_
zwU3glc^`gFA<drj7r4^WV0%l^B!1)sp|@S!AmwVecJ->LiQZa0gRuo<FV`>1BbU5&
zoN)^H`Puo-8`8Qs?--q&Lf>#ii@J=PmsvWst@MQ370NElb#^1uqt4X|KL}(X{;31+
z-c#Jcjx-J-^{O`|Q=y2P9P&zGyJovvtA=eptVW^}J^d6O1}`MZa>--1BD~I_QemQa
z)uzCJ#(R7-rapq5gO6g4Z2j#cBcmH}cLk%TIo!#^y#;17wWlux^4azQa$!aD^kmaG
zT*T(nAd64j2P?_hDNLmeZ&HPHGpqq-p!)v8V5{%Nkj>Emm+2`d^#8&DM>yZeigV`A
z?r8=bKP3_bBtaRkG(*^g4gNmrI66|>(HlGVB#ayUeb~9=DbFm!@%QH&s=smS-W`Hv
z@9~wQ4w)iS4F8HvvdY?gRgu;ifm^W0r79;*dwhcnz(zB!TaqMXAC6WY?sbpSqjq<f
znvee6!0in-AFjB@?9C1m^E58)k2V{W3HT}B1m-N?>_N*8aXaC4`HrA2R3ft)c>=Ex
zt<fqljCSKP<E%jlvOUwRlf{9|XS?g7S(|5-o*1xqLZ5Pt4GlL?A~KhVHMcpg-PRAV
z%pxF)-F@rJ9ZMDgqE%P7Z%m|SMZwEbhL05^m5}Jhu#6JJGP^n^|7$1@llvW7nVk8B
zTIL182SM2Mke9CcZ%onkzQtO+&mUd*v5NY^gLYmz`dx9PVM+BJ$B`#WGLY@IvUUFG
zfTaM!T9@5B4)3OU)x+}R(2nASUxb9kL_?Sfw%MuTXuSew(B=--iP^7B((|8Lo-+m8
zjBh+mur&)BjIlSE8$pu2U0x~-Z2FuK;fRfQl>RUK&FLmb$69(CVarMrPA}t4j1??q
zB&tXXdJ2L7yk!+mq65}Dx>#;XvVK+yTwSV?`UU4kStWHq%SO;I(4k~D_rMd+Q9szJ
z#kHoLn9Uwp_JqFmvwR7XZK8yZ96=Ur=^d)qX%do9<ttVu@f`7*T^$~ruOi5z>h9x@
z{&gPhED7Zu4e&QbNA7n6Kp1-W;9zU(^wE}@G6B9FfqT-TRj4I6kI50F(l9xRT8ix%
zo#HKb-Y{fep1f{{&#*ST@$@O~sI#W|V4glt&jv#_bW>RKA2bs(SPzfuHP1?D7<Ndu
z7$I|)m^JvkX{#Fr3URP@u)*3dkHs|4@r`YSU`m66_a$>!{R#A3%-|l2CB-^j!_67t
zX*F7hi;%i>R#1x_99xP%5Eo+29;z<_MKEE5JHL1K{)W}{Nr-j`cR*7O&Ln17uSF{y
zOtY2(er?=YrX8%z&PFu8nx1-<mnYR6t=+6sjcjoinNWS}UuNV#FB~zu@#n#$bSmq>
z9fF-#u^e1hH{_9wFlK9|ZYC0P<49k}Xu=>|k`uW)Eb|m9CyXOsvzV%!&VBM0)(vHq
zqW5EMeI2pTJRQB$6U`DA(U#f2AtSE`l3&!~ca7P<Zl}-ZiaXrf+S~gb@fY~~D+_q~
z-%j2Ae+%7;kH-GI?8*oUKA{#g7@AWUl-2Q{(dmy#RV|Gk_V1b+bzBvSyd8mj!^$RZ
zli5jCS<{I*T=}<G?^JR29{F<l94hv`C7@5zazL5-GJJ50`GJXB4JDyv2(h^1wg0Ma
z-41FI7H7ft69a{iCFL&myoT_=lyr6QYj#`TiTcvAOySuoiP9mYgf$PGyGT~wc=3UA
zx3|k;C|&=<8B^aL3&Xa{TzmvJGA~ZphG#_i*TrZux4w*0jbq&EdOuvb9jtzn9>%F0
zBIXOGIa!kX?Kq&S&Lc{Tv#ke3p=itU{Ax_E(YU>ZWO^otqC3xYXeVyz5!}dDWnS39
zR<co(fFxTcAV43XTP|eqTk0dQQ0KhLx~(1Z8XoKesU%SrOKY<DcCPH=#xkQ0Y9E_I
zV((~1wdar`U;Njc&WlJe<ZNq6NgWs(K@NjRnQ8i*a}2C4e49@&x;mh6^&ZHwS1-av
zEkV2sS%T((6_T@TM;Ac~ernq!$Bh^xul%kd1dg)>%8#%z=z~FT+1PAflr13HbDcfK
zXUk$c{|g*)X1h};51<+B6wTvrOiSgtO!YPI|4$Oz_pSqAT0u1#e--&#6=ggKL6A;3
z06Q#9O_K(CC}Bdd+Ce?4vmq=_VKiJ}FBoyqd~{fohdW$YSnq*b;%WGPSDlz$cdU9M
z36tM@)WSDuUyV73eA&ISzVdGiV3VZ41T7{h&&9={PjG-K*>GIkU*vlwW?iGgu_T`X
zUOV^czS5`7ng_bwiD32nrbyzc*e~Qz_MY6!&x{`W60of6^r7nEcU)EA`6`L0_=6Z-
zMxk$3hp;Rt69GSBZvO$40@T(fy`M~BuKWWuiS44!0Pn`XS1zvoSlcWZsdN~`5L4jm
zThvzE>;vqPd$TTZR*%qH^HNJe7%GVw@}%nn7N#H@C|8VBH|kjZvTM@uxi#Q4LLa*G
zfNIhV^44W#LSkckHt!=e4}!P<Fd4P|^QMzF#wi<}DpB|=kH9)n`=$x4>VZ65zkjdK
zN&{ACV{Wr{Gls%;%XB^#)|n1Gm)NByoE<HUilz}mL)Ff`r&L>g<+ZJ*428ZLC65eF
z&}&?AyG-2S4c(|c5)(Yw=G*q6tKQ5;F-gR(O8jS*XnRhjy9&s-2l2Hx(k2VJUHLE4
za8DyZaGA@>XUI`VZ|JK$Rmqe~fK&2zd)H9*Zrgib4v_quDYS!}QTWXBTWc(CZ8aBh
z^1_`HS1cor%CaaUm{raM7pO>xiYr{Peavh{)af~&A%sJ;l$v==P+M~RT~HT{OZM*U
zxgW?p0dVQ$o|~k>`R0z<2nX8og9hE)tB&OQVSC3*1(-+a270ns<*=P6&$f<Ok3di^
z8O2Y+97>zQ8EnPbrN>$>qJLK7tjp@3!b^NDe8<-_O_A5cK;FslO7WoEh8xI}VBZaw
z>xK!!?1~__>79Q&Rax9Tjd)QK9K@XhpMI9anTj;3U`9BMkkqZdLX7Tl_|N^|EKTNt
z#x--ZM9O#wU(gqYIUe4lT(%}!c5@-W8y$-bjUHW?jpqN_7I%93za|4*D%BTU6h|6<
zQ<5D|cu$v=;7`IsCf_8L0E=)=hYF+UAo>i@=Ww)*+pPhPO#N3YfNp~2EDv?5r!WD8
zKsY!~F;U%e5NLT&x0+XZ0<eF4!8XZpJ(q8O^Lkt2G+I-%G+ML-oR2Dw4!6Scg%9VS
z;&%31iPE^e6{e6bn6bsZ$k`H&*;d#J-HWSx>=zy#9Tm0jO?>u>^A?UaEv*y1_W(O7
zy^)ITq_)nE0W~*R$K?ZhzpEn|DDRk^9Ytq;ev%oGlN-D0QoXsd;>c67VaZfn3oD`y
zhSy9>3*zT#e-2w|v$fJi*KM<G`BZ@KX-Qc|IiC8{Tib~wtDO^t!h`hIC_GcsL`(4`
zbxsu5JCb)WPtsLf_lic|b|*C_QDXX_aC~w(AOTEbrU`aRDkos&r$=3z(4%#wh%fdN
zyGj7=s<lY^-|zIf57MGy@`^vDx5$k(jx`s55I(ftW|N#MYA$$xwV#ac9zDRQ3DToP
zfLvMHc|#C7#m9@MkgONTbX)ATb(XXF<~w>iz|l9Ny8}mS($R}nNJw1-`HAxoc#9R7
zC!oqlbm``IX@vXaV_E8Yzb1Zo7y1ljH8fzzn<pEl(pF$XT_viT&WrR)yIcHFB{A&1
zTb$=@-rGT7A^IMn7s#8Mqo4r#NvKF^U44YIFL?wwtPiB=gGFDAPrGzR$rPd;YeLcE
z8rhi!I1R|gsCTrcbG0=Q9TJ=20bfsE4|qNmXzm%+DS6~f<R~twN7Rz`$UVP1`jtaW
z#{8mV!b>PJTRO13CGUwQsVz-*><e(0Z9fsfGH?l7S9zaZ)bc8w?Le%B-h>7cE{?0I
zIg6jWzlzshp3@0((=!P6FvKc5jsn1AuF~@}ov97he}J}jk!uBG$g%>fc`0RhkncFP
z44>gmVfsezZAFDgJ|}X&;Z(tG&d%;eonQtCJ!MwDu;(z~M-%{6fq#WPWfz5875P~Q
zhm8OIVa6?iQHPXBvi8El(AlHiyTLKQx87|&>~4;ZeDG>%b3Lk;N_z^xnevWmba99K
zyN9g@RoVI}YPW#1Ol<)>=-pS1{tDb~z)^qm(L!_S<<3?W!K70p$1>0>4-5F#=!8Z^
zx3!eCYaTG3SP{R!;gs~EBvsj)!@89KKG}#hl1!v;0yRy5oxHodpwUnM8iR8S*k)Pi
z7(Q%E<9IhV+^|vmyxF||BOZ^p?^bf9SI0TZc<o!ttiMD$k7m}1nl!afQ;J@?7S$Yd
zGY;1e8HXQq8AO+S&a=1x64c|hyY6*<(>FujmqN<29U`Z>BhMRwWMZN0B1PhCTqvHn
zm!aCOeK)QK5WA!5dsbb_dPaAHn?_K6F}0`O%`B!vm(<rgti$N__a^FJkh5JxVC|FF
z_>_ax6vsx4{OT05RaMwRQ+Yo-lP1TYM&@rGiQ^~BUOH#h@`~XeiHBG*5>K#@27(s+
zjdC?dZzU7G^mt5Xkl*I+{;eUhg6gLiYTplQvOTcqu^=$(+r&Ni3&ce@_$ei13NPqy
zHKn^Dw{P?!pDARFpG}dYB<xpCUEJHpBtj_R`m&f)8&;DZ5c+JxDu^9>?Y5pdyv|qX
zSsU2;8<caoSp+xG8%Izlwbo;Z;ZxsAOQRD{Mf#4Szj_O%lUAr5*3;oXBPA0uWuY2%
ze%mI>+$Q3~;SQjjOr%)NM52BH)SbtI(=zK`8QZ@b>V-Ug(1bDG(ZyN&-EW`)Lo+g$
zGi;^kNSKGW@nr!eD<g7~L%6m!>*3hS=6gz}pF>A02B*FzDhvIXy~*Zgg$Rz3=D#S*
z_CQQCw_Io%m{jwK;bCM>H`?rxBd+U;WkO0<O56n{lfnU`Bw@&P+f5o~4)cl~>6Ji|
z#Siqqv=ULwDBKL*^fri4Zb4tdoH^|Y_8>%)@IxF-uqUK52sb9ZAqdp5L70kd&>+H0
zWjT&mQ?uln2iz@+h}qlvw?B&82L@?-k<>r<yqJC0qix-z71yH&T2h_9W(XSSS|iV_
zz5MV;6%Tu$&-(Yj&Go+x!`F`SK-);^Hw`DnCyla_WS*Hg)u3HQ!hf2fQK>zk<wOe$
zNTw%NNGv1;$e)VI#n;vQTtPe6ka3rpxTr>#iltCzNB{Dzy{*50_kdq=Lyy{>DLa3@
z4|zXGU(!4TA(?@HGyY;}9n(eMo0}R&>M8YGcCUm<IT(=gdHGObG^zsydTmkQ=06AO
zfeY<!uCFr=0~>T1f;K#B40xy3qKZ5cV1qpK;QZ!gp@!e6ez9u?EMhih=@;~iOLocB
z+DsWdfap--7#?388AsVqiwH2fb{JEfoTy|tJ5RdE{K3JW`A%RgjdAX7xEHWga4)o}
zk6x+DJhxXsY%TzZ+a2zmyT8NB*k=%7`hi}2KhALL*Y`dKHP||LTW1kV1B&SMiOs0G
zWOg{m2dWqV#dI#qrfP~w#Hv0YaZJc%0>IQvZBpjvNP+u2=^SE8V%))sc5D_21DYQ`
zi?;xPxwFa!+#>67n<ylb+DDu?A8QIBq^Da@AL{q)$N1n_wDNN71(F0(=$yZQFU=qY
zG#~*>>SAk9H6%b&49>Bm_<Lk64Xdp?XgY121cU%B#fs3w&k5($S7J_yk0Bq!Lfkm{
zpo*g3$)`K2I?UkQYoZC7XKFt;kjb_s!KZ(8vb1BGG$GvQPWcz=tmA1_Hf*sou?ZO-
zUUR$kI33;Q{l9d;UI`dUfJEq=1VKu$RyXlip<ErNBvJ^!ffb@1{5BqY@WeB@2XS7_
z-6Co4!8w_a8c?NRagLwK57@j?;R#?4;}pGwG1rN7MG`gZ*azWO_5Z!ExUnZTsd1S(
zdQM)M{a2q($L#J_-~iH6On6jO6hgc46)+iP{yv;P$UEG0J=|Y9TA0VPGL-T6<4rjx
z6+Z+%4o4JU_pj{j-koS2y?9n2h)3LJPPl%^8zoDovnM1714Z-5JS7B3tnRs-$hkRF
zaT0a@TnAWR*27i|NS-c+H@YnC2Inr(clW~(c@?APuy#YXncaoi^>yzimpA<|99#Df
zt6%T|;PuWTR2P!hw^%wtJLfAlx>HAy5wKHnV(enVyS)x~2dFQKx$BJTKWZ4x3)>33
zem_zDqVhex;WAblaA^~}K*F*`5Cbp6n>E<%0(L#MprMhMPkx-)CKo`GN~B$>uY(%<
z>|U@7wE`Peg7r&uu3f2^d%!&$gX4PslsVoI%Ef$*r0U)~gO_fOf;sv-@xizL-#@o(
z(;F)H-`(38;$I90YQ3@w#pPs6;%x}wJ@djg!8b!fi7p2wV`cKMC(?ex+KPKlKCHS}
zkYPVw?K6=G4)c`4!yb(xmC@9LxB<&mrQ6ps6M<RDxh&MV8cNsl^c~~my)HiVQDS7@
zII*dnvt{VLY2<^5u1G8D2WmD^p@h?tZI2jL!YRv*?K3&bnQ0^zlyDleX%g2;IOTBr
zY22b82Nyg{;Ts-3C>hJ%C$QLYdMl7#2)HVAweml`wRLLD;7kN&oCNwTJ5Q2}TLOp8
z2Li!wU?{~CM@N~P)@HP#3>!#rX|{|FpInHKpQEq8bFS=p0bibQuyuZ5q*PtDN_v{y
ztu@%E?cVC(*IJ9dx<N9xeZ5~z;3BS(Ln8r|ZD*^yU-#&MULn*UP{eF_f#h|E`~VP)
zz-Sq$zSaT5rx84l3f0HZOQRpZy<VJ~LOR#1BUc~#%&S12Qe%e(i!WOHpG}})qz%~P
zk=~Av2{d)#f0*hO!}})Ozghs7dCNvh>~%nBd*-{YDCa2`z+2vpsKY5Rqaq@yPJP-Z
zmJ3l4I?b2~_QE&Mhe2h_)jdK&&5RDoS^(*PnQ33m8aZsuYx2Y~owv+<D5YjXGqP-A
zZKiNhXjVL76jk^`;y<Uxy0VM-;zdwfe)Opl)Z)!-egjoCL!LMT9_QcNSf&qgn0DQ;
zx_3#hTd5tKCOUw|skQ+3_Gm4y&HuZ&85ME*Z*86+zSTsQxCqg2{;2<}T=?G^&a({g
zu<se%&TNQ3J1AhuUySA#5;9ygZrigxSS7(!`DHB;n8VN=u7uv<H=KHpKDT<}d+%nV
zXo&JVP><f4A8=T6LH?``#dupLOaOJpgDQd$;iTXci{)|L$XOXZLc{cv+U#$4g83?z
zj$?$`iibulrrftDrC(0>c*Js5J1r86h`xCqdoV?|{9;A+_BQdwh06DiVrR`T4|10}
zi<sT?(4VXJNE?IKdcWGD>2LjH)t12`!9Z`F@3m`6YRYA@u_TWcjDp^WCfo01b@h&}
z6j0L({h9`xILDqqzzQh#;_6gf?hm|u?5}b^E?J<BbNa@=SJqz+nHzW&0~!66rc5?N
z&h^+c!FuW2pj%I`W>OsplXq1cd8pKPX&0t0^a`g5vYcUPn%N7#4ctJRp`GodK>F0d
z=G4EFFmc~Dg2fcnV?;AClaTUyyVqRoCtqCd)E62qNMjB#57W|YL|p0TEZ?cB$#qs!
zm`r+b{Wts5>I3NAM_nMV)yG#>MNixC8}QAieMk4~w$lDE80x8VN>}LC$m?x16{#(1
zP|lQXmRzhWG1z%TXYCEi`lZV9vT0kJmu%_yWLg>t;U&c0=Da7H8VBd{I2Y(xLDX<n
z;KXSse|)jNvI`uN{s8}*QJJGZ3`ss}0(m-9(R(}YYEKLszrQdbK=e-u9iTq0j7fzG
zbC;mWq1VvxXAYnQ*%Li-aFGHqTf=A_xTPA+tSq0{lJeuUv56mfVf9xJYy&BBAGd^e
zj*L&xCno5=b>&Ngz*c=<+m-0;OSy-e7}3R0=8y%$Ix0(!?di&INH0U9qoZ$q`V-OE
z$E<#VzH+p5G<&okxElFR;`a#OmPfFfu|;~3b22~~k@=4qiY^>~$DQX?KxGU;M1D$w
zcwXOcsV}b1P0%Zy;y%?Ab!jig<SVq+abx#WC~Yyk5AjI+g7w`hF%Q8wkyszxH{Q(K
zSBTw7u#|A(k2JCB5s771me`(QQc@0w&2rH^4P*EY`%uAYu|OPV`LB{t&#~S)i&2Z&
z4ZQQRJq<GxTPYu$kv~B}AM0^4a`dNEjTk;LSj<pA8#q=Snz>|OjdrG2*LPhj3MJ?1
z;w)f-@PZ6c&cI6(eU|u~L?})Owi|OMk;g4g55zHROb(0Axlphp_39>eV}-3`J)*ur
zt7AS$wvr^rx2d939AG!{!#0;@G*0hcnW*URzk2FcORB68JaNX<Xn4r5${wA6F%@*L
z1=^&%($(Tedw)%hBVmkQ54#^yNrl_YfDL%kKcms)&PTyLf`nB+DIAF<aiD!`8hbp}
zu%`3Jd#b<Tu^wKSqUe2@t&Q4>=1oso=&h%y#<#FNgP?8_Z6+1PlbPh10vf_RuG*TG
zw;X^L)lRSL7yBhYF8j7FaF?;VmAYcGY~o4L0RyQ30*9QkN2hT<B027~EnGx7l;bAA
zG3%}r^t1?1M#vT6DBtQa6!DT{wvEWT5hmSX#EcIX?3Qc+dz8VwSUks%w14us2%c&n
zmDmPi`R>i)Z*<P5a}beuXp&_s59=VNd0JrpUs#hpnf#5}y?ejOiFksF(M0-{3ZkT}
z?~9;^--!U1FBalF^MwNpT%_a#5>UvR(3j|-l17PQ72Hp=Xn?~V4xC^63X(FFnWXum
zr`x!lm2SKc7O}**;JqlUdA8=%?!ShttkbXS9*0kiggcazF|_v+$V(GkjI>|^8%O-<
z?^nud?oBjY2FR@q^_phb{tJOT;sU&$g5f>k7HthL(U-h)N0}=TF;pjoQEwMVa`YXa
z(%j=$aMq3!9y2YR<!{uS>XhR&OD8siPkzrKhPzagFEdvUC2O-c)>2N)re<?n!X*U<
zi*8_{?zm@`?wK_$4>T(L(=ks1K_PY@b@rj}?5__QzN3sIyAAXW)t23O<k>;=;jdg4
zriLiVWSP-!rS-6QOOdaW7DlV8ZEc9tI{*z4XM@DAudC$U2*9C_nx-Xnqj*5~F3zOT
za*_u6yDhX!0iyf7y6E4Cu?lHb5J`c#P~XG!;3`;U(ib9}N4$BKG3+zZS1JljohSa<
zUk#v#R=X(LuU4ZAOtf-i*^9*WL7X<}nAi)JeiLSnW{ns$b_}RRTM?R1jl9mCZ-#r4
z03A?S5dVTXBRdF)htAay-zN=dit$WZk*{vb=@#>!Bj_?a9=74e)d&|Csl1x{HaNRn
zr+A1|VnPleJUNV_VM~L5w<79aRnV<d^jl(6*fP~MVsG1h2l`2Aa8238%VyB~Hh<AK
zR;L==k8QUIr)e<08S<yY-htH%aQDTuBZERx;BAG~FFuu)nqC>+&ut=)TjC*Oxh;{p
zMRn8Fx7#4UaXpH-nzB~JzpxuqF(M!H{3dBBNq)+{`4<c2K(fKlV;EO+t>QgUs*634
zhjm%i<wHY-HJhEEXH~-+jQ~rby_0NG)cF!oD9*0@kow*+6cAOOJ{o}H+DO4Ytzb_S
zUtyp|ueg7YgRR*WZ-<-1_E(4+d8bN5c{tp?AfnP+08R~@c)e|#qj%KD-s5%71(T_#
zz_`oGi`h}(H?TpI3&ppGKACcgw=M5h>gY6HW(GGpRg(!w!Y#<K^N=69sOEWT)&-Ih
z(#-;%5>7ybB%6xk*%VXpw3bl8+c#;bpC{S-VU4er1{++6!Szmyp;zocG98NiD}$ZD
zr|j>~Kh>S2trN>Wah}0xN3r(y{+e<oxXG4Bn7@;QK13A1O;Rw=T=I4pwJvqIdW}6)
znffGLB%K5385U13ri@5$1FuX}y;J`vnubK9#)TdLBGO3^$8GlWiu!uLo95SBJy7zm
z(T;{yl<MKi7r_YEh!YFC;nCCjxYo|s+=cV*0AUyq1$9I<aWaZ<jq-UACcjiN$b0N(
zb^-%)v84^gAXDs(vZ_=W9T}pV>4#53jkJxQ&cLi?^yQU_sEHAa{YDkhNpc2R`aE8o
zauR#?OU9S#>cJJ@iqdGy3S-BCf_-mObC+H3)SAY1<qmO!_jlNGc@rT9z{S2lY*9d;
zPyj_0YI7EsVEW2~RCvqrWBu3a*Y=gG=oi+0zKvP6D!v|;x8h>e+XkA$^GjuO_hjbs
zNB9*Soy`o|4l?%Kv^#RxMQ-tH3HqLS_TM%CQh``rkJJ?4V7?xP2iGR`D8D{FCJ?dc
z;r&ygs4U&PUpURqbK10)N;qC?)6XXh7tV|?j<orMJ1LO<H9d{}W4b0hEh)3Eq+a5V
zuDb@i(o_npG!}#DGUPM_vr7xHS}}JE1>I1F(OxsM-ynP7)Dy_<SDBhJaAd$_*xk8T
z!;0xvb8%Q}UYc1P&Lmeyn6O25yWy}6Q)ZXp2z$WRfHitFKnrz4St4QH6ZZ1Ig!N^x
zV`expZDkvOv;(p7#RxLM;aaF(3jEQ)=sA6?4H_lIBT4;JkbO_`&0G6Pjl}Pn9CB|R
zJqQTVqVfGGn$OSDt`B*FvmGsp%C?t{7U}lYettYN<p~+GHxN>VwbKId1R<8<)mAN6
zO9}S27CbEw3s{&*g!f=jLYlb6iM~<ubexUN&P0FntE)tNN1Eyj7vF#Z1Ro0YZuJM}
zBv?Dm#F+P9`1&1jLAL0-nfIyo=h)8=)?dL?C+_@?(&=^dt$uMapHMq(IZx_@0_rh?
zA&AAY7&{iUu?LrI*|5xs*Cq4zwrqaRQc}3^h<#PpegqzRp8d|~qn^o}k8dA~;6&OR
zGyXT9RAL<M;2;L!_$NmMx2*MRAn3uyYa8mI>RYKWvAYh`Z%<TKBaO1MW!WBPC!4Le
zi*fcXr^6J*<Z8<vK8vi_v=;NR4Y0cVfLMF<b7F^z*%PdF`<t`;poaBtg)!gxHZIGt
zGrFS-Rf*0n=_`AntQKf;PLTsgSv>=(D$&&6@GPY~8O*RzrS4uJn6d#TFX^LlMT_OI
zzsLy=4jUt5H|v^lGZ|8F>BQU<#8M04S3&uW8_lV$j0d$?4w=O1GXa{(oWk&gG!IkA
z!_2beEXDb%>@9q>|0WQ8rGyDKl_r~cj0yZqcPgVS7N)~Y%K2AUt#Ni)nYAS)nfdkg
zkITT@yDUqYe~w|dCXi8Amcr}$y78_5W-&**hnO?gZmQQn1n9f2P_+!D_ua!!<mB**
zfS)k_WWLX0lpH&3Z2$-jbitu|%0hiG>Knh_NpZ{U2<Mv0_l=%4qs+t%10~<`*iy!L
zsRlmQB37K8V{10~?DM!1#Qoo@9!o1CyS-wvjN*%Jxi%-+2Qg=!cDDXcKLtR7&P0z|
z&NAG(_)B0J>T++E*ro>ce54@{-d%I7sUO-}|Fg2G<T|<))<n3#Uh6mNc7Uo`?F(J*
z6Vd1SWhKb=o!~TZ?26($9z=dzat1|4cBcM4<yvS&1hvocQ_PH97eptze%g{l+0v8o
z{auc=m83E$lbwC*4@PKb>p&~tyo}gAQGsjTpT8ere~Cp?U*yEdT&XQ9j<`zkoOx|$
z?mJ4aRB$Yt0~}S%SsG!6kN~}JAyOB9h4Of%J2bn(+@Z!k%al&%$dAv?ekf5ZrlAZN
zFVJ`5APX{eyh?RLMMUj{h?o5>_ucY|ok$4@$oEK&X^h37I9nv?3rKG`{T>dCJNk|9
zhb;w6yHuDv*PP3Tn+WpE0T)`xs@xH$;ATI+say42F9_IXhhEwekRh=_5g)Q&meTD<
zLyz}`pR%zs<aQXH`t`$GI^DiB*%n*p1Q136qh%YhZhdB;j>x~&Tz4Z{<8CT!(u?<m
zyeX_L<7mM(hJW_7OCQsE=vEJIiA-K*j#x3zd@X`7!l)R1P7$?J-^c$&Yb}F1Oxwiw
z?(n9A-WkBYT7I6o0UjR*rT~rpBBpodDC>ZM(Oq@dT06%r)B`lK!z~RM*(pH~Av~be
z_qfwkPY_b#G{(jzdj;$o-d{QKbuK3@jd0=7gDXVhk{Icc@aM0w1~|xHQ1WUUm#vYb
zO~vYRIM~1<y+rkxHGlNaM^n_t2_A`)mn0fTdfV%L`g%}MPGrrMh4+GvzDzB-!6*HT
zc0(cGa<e^3>Gh%^W{CT@eyOo8$kvLz$wbpT)GFR@i{068TI<3(@hEn5c_kpTO5-(3
zOYYY@5cu^BJyd_aIl>XmV5Q^fmeGl317*gTBgxc1h1AETFt!e1b`vPQ!dm>8p^D=A
zX88E$K=A=ym@u4jk>Ypatc=^*GUsvuRWJX;y*bWs9(W=c(a~m@BCM#GK)x4>4z*Ev
z9b+rQALq%BcmzRk@%uo7`7#fLc^#jFTWLWx*@?on_s~CA0efn{yOq9DkIBDLQPhpB
zE-W3_6puVfELrkp#aqS-gSvVxBDuK!XkTr~2+>+AVo|BB2L~T?`|Qe@)%LYc72WpP
zUkYDy5fW!$?ejr<fAIa#zx6$`<Plj*f(8W0zKVREnh^Sq(O0T15O;NTm68Sox}9&I
z@lYktG6`I<F{gpgx$f$qZ%}VRHa0u*J<6~Y|M|7jMhCIm=k*w@q0|dLmVVFe=}H0J
z4I_X;zWsLd(JLNsTnV-7vSO-1L$l`Ewn5+ZswAJXh{ltonu{mmB5ye8B3d>~DmX#<
z=IL9k{RJlNP?2q?z7JB|uTfoT_MDNMTev^lnvQ$G6?ltn&@uLmKugOpkmdK3h`hMu
z4pPtl!Q(C;{uzc3ZF%TG9yLlAb$x(UOK9kCRkv(?_@S3s@l>c)vCaK;M)kOvzegno
zsB`(dx3x2y>H*yIqUD(Un@Z$?$aI|XVE2+ug?r~im)yZczN>AXzeg#iVqzhI0d9_H
zHe92vYDm~Io**rYKC7<}wnCN1A;(KLodULf??={&J$$}BTZLut*2X8-9oK`_-*a>r
zc{4<dQIixJ#6UWr6Yf_j3g)2SL1%<m@Pc20m^c4H@EJBdJt;l}F4~2&Xqyr#E<tfo
z6Lxv8cLRp+vP#I=rBG{m<C%f4PWcy07jb;zIV8Ew?98;8mr2;w1Db`Ut({CleK-zy
z?zWbXGF@C;m%P0_M3m3px2<Cn)@aeW8@3Fqo<u$$(b@(>LaE|mF{$a>9oLUB`Z%#%
zgy!i~si1GLQuw9xVo&9(mWEXu0teNS1|F(4*8X`*9Fu0ZK?6P6Dr2C4P;1&FT$>uF
z{FKu<pV(zM7s!X4>_&XfI{qcC1sW{9^+0VSJM^S_@>so{F}Ik|_l))ilEtfyxx>Lf
z^B6UFd#thj`-oge=d7384{qN`5$n*S$s!g3UD=|r7`B|I65JQA`vhiuE1S5!c|STz
zN4qJfk7+-2)zL9q<>T~CrM96fxa=A`sdP3Hh*vH&R!G*J0R5GbsFe~97<^QYg_1M>
zrFU17d-=^v$R#E2cf;YI)0mw?fD5u0jKkhMfi|LKmez@cq#z1Yz!A46dS}?8ad%PH
zWettm+RagA-v5c$gBC9qckqn+`JWg-D2ZgBfyU3(XJkK*-Lb#&!<HW>`SFyZ;xRqz
z?~b2a>&d{v>X&~o_RhNH*mum$lgr}PVEpethrbtqJu@g7Mljn5507rr8Z1dxWm_pN
zsp2d4YbM^02N^2<dZ1+?tqTcBwe)kSc0Uu+XIh1J{AtGj-;n-m7AqRJGCY!n+~DC9
z=9II=s=Rh7w%^mwQ<BAD4P`|T=@MZQKgnrbpV`Wp$^2b9Yl*bD4&U?Sa(b`}iUTan
zp+}}9?FA@pu7sZvoUi4O$7eJ48K(D*`GN>RJP^+w1fVehg*VWGNuda`y>4%*(VMG5
zf~a0iHwUrn`h*}0)-U%eb$z6w6Rbz!1*BfvkU#4>_c(hdPLoZx;W#2!{}`9iS``ox
zu0C-BvGt)VErg%yn^8}9*l{-<O!G)7MV-tZQ#2)-rKMq}E)4jgpI4hnLG^lagfA4h
zk4xT&pj9JPf7<)@4&Rg<h&<)iSr<#NvEkQ;H~6>xcAPS_<mdoDEN-eeHRgKn^_0mq
zimgavpT8Lbmng0uXd+TZp<E<b*e2CxcZT#8@-Z-GH8(F{z2<w_XpZ!jI&(;QIt)ME
z9U2c}<&+_L&SaK_i`a?1nK~F4jcB%vm&AnjGbr|kDr$!JK{oyawVrL771Sdc_UKc8
z(+JSSF??0TV;Rk`j%ow~Zo|V*d+>8td4OJQYlj2peE)h!;HzYo#vjZaX$!(-C&zuo
zr6QDi4~<qhlyc>gb;}oi&=u4LU&^rIPe{Ibx>J18K&n$XzxV|jNMy4%|NIZE74fRV
z(LdG5g1_g$d8A`PQM|-NMq^T4m{ULHa$>F-2?!zq(F&?>PMFQ4r9k*4KXhZcH~OL#
zE_R6cfc@^5oSjzMVUe`%(I3|s_8RMI)pwPlo!p5!PgCpS<!49gM@@({nVX}4_7<s?
z`zrgS_`;$^V*UkLRKZCO^ZB&ku(y?})};kx^XI74gqEoV?@*?c`=X8moPxeY9<D^w
zD@TwNSgpx9LoAwLELnCumM7i4>kk8`7_)bO|KCdO#0|*T_szT)WFB{gy7aX5oak9k
zytz}Zxf>bwj|cImN}|K}rav78s>YaCJm!NBopUTH;A0nQ*ZL&{5%#@QK&_^=54Yrg
zr4GOr^JynhejSr{9ZU^Vhw4~<hq}7$miH&dD@TH^*!Zr3dVI5VqB<b5(4;I}W@y8C
zxvbG}?SRmetO22x*w1n45kjC?M*>1<P|*q@1PcJ7h~SpRU40&GXJc~lD(j=bRJZHR
zmXZ7u+QN&%!H@@P7D03Fz;U7cW8Ym{cRk-d7xiiBEuAYMcG<u|$D2a5_BSQiSK{jY
zyXm~(gYHfPK9bFclRYZJg?czRm$%I)aYUkm_cWjR9>-d*7kU9{ZT6t0%ONL>DrCrF
z?muNJ<y}?2keN?VAcK;j-A}5I5e_Vuy4unO-VYb5Nx!KGC69O{9llaAlr>I)4nzi<
z30wfO2^Jh6!bQ=|fQW6W>hYF08RN*1SS|j8C!dF&{C#~9G}UU&X^I1c4x?{z15u_`
z5_20>Kjgl|K(df0$^Q|Nv`~)6oIp{8#|*>o;ze!t(C4265EiRU4-_M?rVHBc<vloY
z#%d9jZ*|6#A3AlBW73lHwq^=$vZF99cW23b|0Q(US)mX5OQo1}`q`W8C3bZD=xC+g
z4PK?6n`ne~Zbii@8jh1K=^<8IHE?x5E|BTa(G^#Fd!j%ux3Xzil?db;_C|4lmhc-0
zNIE3nn`1wWX;{fqn3TmxMjO)DCpHdKA?=`hUBH5^kwR@%Uk@7u+uU0;FEM%?XA2Li
z!d`?v6x&PtK?~ZR7y6%9qnKJZJSVmF;dl>Bzyh8K?Z83X-t?Q#q69dOh3W;Cg+_)S
zY_9)ncu%w*U6i-K#SMfZRb4A`1|MiabYuMH=ndrV`kWdfF{Gro;b~ge`_@`WT$+Bx
zV>q+^#5Gzq<(~hQ;h`1*3s@&OscVl6C5yOwz3TY)rxi$u9{@XtJE0A<+;(tvffh5i
z9&F`=%Py^xV2>kF&i!xW+-9wO0M4B}N7CqkAHPhWdHMc5_qdc@QF;50QW+|7@95}l
zp;YO&)wPswzk5%DN-rK)?!P{y^=VIWD2C@I<V%ulk3eiO&v9ntw1IzP_Vyd1?0DU?
zOpforPnpaD+>;!Q(FuWp?H?VV1m=5+T2C#YaNLh5Y_%L;d3a8T+|z9dkJTe^bTLXt
zvZ_ldTh`f7cLLas#`=)5mi^E?1#Y^&P~D`{9Bny>=of+%ayV)6t^VP;l@gS-$W(FN
zuB7_YX%_w^h8BiTpAuR1=z@`$8f@S>5Mt&#N^O84kdhQ9D^DPZjxDA%>Th^Y6olaz
zE>vvjH8t=67%VBG2%@k>uFxQ$9^vr~*Xf>y&BKVuy&c88X9>cIV@TtW_~K7og%>jP
zEntab4W$Ut*YW<}S=|D{KN<pcPMlYsRwExX|3e0|*VTuCX39rsL+NR3E%&^{vrFLV
z%4$AbI?0y!^lRh1zBAIcoRhu#AY7kxBcJg*IIr?>y0s?PP9c*1`?;Ep#G$zC)*nHd
z(`O42wnqlS#$F%fA&;I3dy?M%lE*>T&MC^ue(K5bIB~9O%AQG!(Cvv2ypuioTUd_|
zo{6Y`2u=N{c}C<-eAuw+m*O0PVnST9o<~Rpd25Or<8(*(7WZb4WN#!HlZZ<#1gb~8
zNFN9T&JOH&pQ%K)p_FdS)e4DrvU(fmNAv^}O~pj}FJFKNf2Hwg7b)Hd?}G`vPQD2V
zLH)k)-f-1cBPu4E^$#dm+1wl?5{aXmP5!<kko%WUa3WF>@(G{I+zj1O#;6p|I4D;i
z;tj#Pg@nkJzXBd)RNeN+IcG=Iq^Cy6W1mhPf5@HA4BrD7U=dLbJcHF!9oI!Y0`Xdw
zgH^wL1>thXq+dX4Su_0-vL(JH+coqW>kYv0Qh!$02{+>TlH<{3^NJYb?6^Kr4>b5P
zYSuqI{1O?n?RK>pTQu4Da@^6saRiCHqLY73@L8L>Ob7qSz_kwC1C51FJU}+<wGsb3
z9`@)}@R{GAr((O75Zc<WhRYXwv#1jFVu$XTA7cYGESH!KIVg(t#ME$_X|Esp!H%s^
z+UeIto<_ny|KHArD1XS((Yj3Nb239mmQ43Y8*=K);iQw^hH|S0pW#b*%k_K1g8M!w
zL|QAyg?ti6tMOar@@jKS<h<;GvdpfB7Q*AhV^~>>-BE`v?q(3)uIWrikg2?^@3(HP
ztMrJ8YU<B75AWYRcanFBS-I}*tR_trYAMc={NB>jScb)(lS94J<MovJ27kyq7USA&
zZ-(|?iS&(J=MwaKu(s7A=&7bPWXrK}%rC%ky}lZ^RsC|4Sah_t^fXglF?9@KtT!hk
zo{;}LV=jkTU*;g`&O3t|?`{O$a$=;t)c;7auMTV8v~Jj(wN>QsN+R1SYp+(e{!9kV
zIGfDR=H{;pa=j&So_O}g)AOaf$*mA>8LuxQ^c@CB$!>7_Jg0HX+-Cwrr#{xgU}`ZQ
z^{)vRr!St|^Dxb%{IZSSjX2q(xYBHQihGOjBcB%UY&$Y)L_-+<J@Ti18$=%R4KC@h
z%$PyNrb2$wA~pSR%AIk|9_}8@?_nEn+oivl`s5?awgfr4iWa9jX1PH<oa#E>aNrg@
z=qw1t&RH}v+q7M4CuT;gVNn{V==PRR{=*T#gzt93!Tceq-5J4Y;_AF6?r5~_Kx7Sh
zobGb8x3YjcI9z8kC)TSs3bqAl#<3;yX;6{BJ~uzhP~ONKn}rxyJ>%%?f^}5&Bo3un
zi2mk;XeO7YhD{HsAo6FoV<~U8<HM7Mgo=cur9600(aw%$L9w|87K2rnyI~8DaA(=R
zY2H7K<jg#+=gUGzQ)w=K)3Lc}RoHbh6bQa_XCUsTiX=sjR@Q`Gq1O|~JDWC}zDvtZ
z%|oF|6u<Uf<kd#CQ7fK=)OqOvcd*)Rkl5$J9BFB*r$R*;Jg4h7k$yLM|7&?u<FYLa
z@%Y8`{<&qJU1Lu@&CYjp2hH@vY42tHx}PO2eYF4+aC?2kz*MQlT2NLWlkG=utU^XQ
z7cSiYcH_of6ra0ejbsaU8^!k;)Mh9{AUL~gjgR*xO>KUguLzZt(+X7I^BB(>8Sh4J
z1Yv_rIAYWOiW>-itY29tZ@Tc6>Zj~PL?`VsRdg$ROyRc-<#)IR9+Tv8Md7ck5Uw?7
z$wykCsetXsI+6dZR*Ih9s0g^-NW}1U{uGoe+t9;mWHr`ErezrN^mqSb?(#Pdg*pao
z)6oZ8f4BA{js+^MFPgf=GA{gTjn)u9yLwYYg--q1IML{bt<AY1eJTlI6-SEue)i^u
z_@n1tIfF_RiqAU8FZLHVPOW&m(baXs&IR36l}M*VvBD;*8GRjIgccb%Pproh^prtH
zi^+tS%4?H1$8`1tJAD<0yxkT%zKp%e=8#o?vT?kJ_rib6v2N$uG-TipE~K%=ruA4x
zN|&CHh1i}xIs1n9y3j9#+sXfA|C3jGgS~FB^Y+VU5k=$4ZTBccE)tHt)z@J|^G=P6
z!8cL%q$)G+v$hGblke*!E|475<cnG#Jv^X3@esajWULk8-<w$6|0J`>H!B!57wT;O
zv)95<{sV^!nGqHXo?fa0joA*Xh1@l~kCK^-g@)DGhg}w!DzC>(36aV)VlTc`ek*JR
z&ne9D?1I0ONiUIj2)}_FJg}dtW(Bxb95Dx(0j@67jDieR>pNWRIU?z4rwv}myTt2o
zebCn`*m<C%t(|it#Kp=y>n`EXM09l12A7x2>aRVQb;%3|@FU5F={ylAkG8h9mbMmW
zx^-BH+(RA3-==VYCGPSPFMfV;S<he|qz6*(s*Su<Q9}&4n&*Ps=(l_vPZ_CZtvX*T
z@XeAp{?EZ9_#OJ{_CyTObd4Nuac+ow$sLAHsZ2`a6=_Layq)bALg%qE{9?+fV|Lvx
z&>+zL8)E<8pFgd5e1<7y#6J7wi2qhpnBsQyv}<184JL1_XNGs?RzAzcpWc;Yz5iCR
zJyt!~*+cZ2(&#wuj7);C@jl<{(Gg?!N+O`3ls3xUG@HGjMS`^jraW;B<<>L7%8G&`
z4@jO*)O&HY<I&;|TV)nfF{L-noi`?4JE;}Zvg55_3W%#r{shtd=D~`|hRsT1=%t$S
zVId}OT=VlgY-#F+0In}wceI*VNkP_cMNe0Z`jgj-@^CS!@aX;DV|z8TovbFvte)!5
zM-^^g{_AYLzBaX3-S*vQi}>#O`Ffy?aN)l5=tOk|rOMa<bS@#~-7l0nr|Eo{&#JxV
z)Yi`W`oD?02Mcj4Dg`{9m{l0Txa{LqgiE}l`k;^SX@~*aeM&V$K~1q-!EQM1|7be%
zXei(R|BrphG9!9h!=Q-l+aNp18lvoLvSjT0auYEmTe4><OVOY*_CYAh62?--*bR~`
z%aHARe}3ou4u3nhGjm_p>v~<!*W(eFu-~l90r9vGkfp^{6KU!G)1gQWgL2GgOv*k(
z7htDnd`2$O*4nvfs6{Z(mlBl^^%%FP-&U&${kQ#j5bahu{rTPmn=)fH4P`08d73dQ
zA0J+vll|@0?X!o}vt7VrsRKdpt0q-E)ZU)+R)?>RE&A1gsD7m5)GICO@-eQwU<(B0
zCux81{I7YBA8=%YNHu(+0%YZ?we2``Aa)q?=$1}bsAAPGfdQ`R0qu1C<)jGi@u9Xh
zK0x)YK&j4KkBRRcqE>Vp4L)i)TO+Fi{LNA%p(^C7N~Nm$a;b`p1Y*Rtjm=yE5Yuw8
zxi<iuo*nP)9e|B@%d?}+e9zwY^iR6f@FHMmWM@$H)!(3k)pS!9J~dSnJT)uVJmz5N
zE#rH&REeRV@Ew?Fs_6ob#du!2|5){Ge^*ow9B7|Qma1qL-}7K(-IW{K3;2#V;-LT`
zod0p!$IYHJf|0|y4mFTW^DkJ^b~Dl!JN|nsyg)7HXf5G4$h~?SEQ`Ulh_#kI2Ljjl
zd=j;}WfWpUWw|}XEjPD#KXVAw@7={+mEq>mz_ayuegk)pzP>={en_Nf1v5(n@=ry<
zuZ}5mZVLeh=8Qm_oQ{_^wu<=KX_lMId^Kx<X@ee>Z*HvPsGPZWpF#!G+)t~#T79@F
zY2gTHW@doV2k(tD{To=#Lw+G4fS;O<w}It$SV?wz-S?SqpLadDJeq6>jVP)ZLMc~>
zn^J@_N}F1%U$ipw7kVDENj?ZB&XSIuhm(dwAG<@*I;=|%#5O{FFb%U)!8jtA00&_z
zj@wn2Z>bf{)sjzaX46BDu4`_PpUU638JdLh^ZstDyR5W|7vGbkfy04<x~8DEzf4WI
zMv5L+)~hm(f4nv`rsAHoO}9>gQtK`eYO^o{XI8iFwJT3=J++Da7HyD4rCU~8dv-Rc
z8ANq+2qgw#PtpAhjqPVQHoiUJe^1m;q!V&RoLaHB&)8jSdQN$R<SFy%Rin=woszOX
zj5*#k7PkRusirK2$-)83f`VXPCeABt?ud{#YJ*SIWInL$=6gR{91IxVm01~(xhd-H
zy{2&YS@rY@-&0Y&SPs#|#}@Cg1&WAVp1;|-x_iOFi<9izvtFD`4);Rk?+PRIvZU>}
z2af8k_dm=-AUwqR=pH^vj*k+XbghAMv1vKKqzi5xL2t`I=DQ+*qcS0F4L|PaGZYG!
zvsF1=7IIf5hLW-7meHNzzz9|2D&)`|5?dvI+`DlruSnaC0h<djPBy1@>*DbE_@FVj
zvInV`YNVRjq)n$mP1W)>p=b0gO!-zZXl+<u-vvWc6hiNfr!XB`XR{5Ak)n)oKaYCp
zXP4L)h^PO969EBI#_uN1269bBD+X3njek)jAQKwIs~#e&<Yd38%ap^CcJ3IHTXDn^
z$GSNbxoO&4y;0zDCRmiOH%zroncMg#+C8@)m`tP!kV-n`Q(<kNQ=r|(r`@%e_0zpu
z39sEqP(y^HHS)|EL=(deR`I7he#)<|Y4oX7bF#|N0}vE}M8X7XRNFsI799*+`wiZ>
z28$U7XS}#bcT}HzQ&QzIiE0Y5^~GXGS>IwPJCf+yKKU8>EA<Ux4&1A!?S;hmJfT&Y
z^Zy#RY&VFB5pl6`U#GOFr0FgWqIt4j#%n&pkU!ps4_ho#<Fi31-&sr~-`2!zYfXuI
zagrQB;K5GiuKth~7rmLgC+HljZL30}=oJ@TRW%-b>`rKm20{P-4lAoy@hdTHSR3H@
zz;r-{xD`1S@MrbnK#(%FtRQx@P~O&Ckr*j!5x*B@vKyoZ|5v1)+Dk1N#`nVOHrJRQ
z<gwQnR>5mp&)LDUjy%*rcTJB`hcBlejRb5`w50dnS2XSwU6nMH-?nNE=AG-^F(Q}c
zPd$G~Bf-mN^f{u&02L(1_nE>t9Ln+3X!LWk*)o(SO0ngx$?XvXA7jg;C(o0g^tU0q
zZdb6RaXk*|NxV!PqTSKT(4j^njy%RFHDK#FPw(j)enwg>15x>7LgG%9OkJ6@OILwE
zbQSIwtCDmcKG90*gZO*jL1;NPGH{l=f3nruf@X?e!Xqty?b%kjJDNHORx}<hLgTez
z|5BJypwv6m?>c_Ep@~A6&Tq4|R_rpq|G{mJadrAdW?WB~-CM5A`UDtw2g@DHw86}v
zr1Fu$Gw20sgl!NGL(ecL8!;~|WCJk~O1Ex_=OS<3$>A|~g4{hgu>wzT7w4_#2Ydgk
zM~cCMge6FXN$XLn|Hz2mID6b>BqdFm=+ml!<DQ<C-GJg;d_$3Rx$U~6FQJJL;o~#U
z=g-)ZcJle+Dt&U2J8i;lRKdk*RY}vq`RVEA=B7`93_OwTj#~i<zaG>96s@jg{|(|F
z{91No!cP-)(E5_#u#I`?@4@Pq!fLV1QA>z$Yzm$;+(<xhP<|y0oE?G%BIYWNn=Ad$
zjAa30*&NAdL#1;)=$aet0@az!gHG+%2(=-6c8;*Rok?zz@JiU+7DiG25~~n3^#k9H
z{>vqmEuq?r*?}3<UE~wsJOnp!?wZAYWM<@#Y18Jv2Fc*3kE5dMb~rrfpmc#cTf5Wa
zMOwc+{)xSPio7&zVI^Dc-$ZD{GJlqO{hUo1!IPB+*Hb-T6qx0H&j$(Wo5i}MV{H|a
zyXK*dgO{bhavb_d+JgIrE*H8vz7<8D>O5WPq}u5@7QKEc+aEZs)2^qLVziKUM_lDF
zfIJ|rIG(Tk@b#VY0ZSCcYk8{;n^@7j=O6w%wRp3BQumzsHrGAb2Zf_HL+wALOA$+Q
zU#^$Xk7H!DJgaGmPRm2LlO`>jCrDIi<;v4YpSjsBixdln`6~YGEXIS=?69<ef<^qp
zs|$EAS&=yQ9~1z^9qa|4Y@L4V=rJ==tIozgp<>#WO7q-Tdg)n2k+kR}!q$L|=p;fu
zG**WMePi)g$~@e<JuH&WQh_zu*{bGkic!5LFsrhoBAk_x9KetrU*lC%C2DUHp7K7p
z&PrN;TfSH7CURxGwg3E$+UE=We~6;o_Ce=-yl2{m!!}@x^TlK*&b}@!l=Tqm6u-j+
z=`B!P8~S=Rq<^LFs?{_!!y9o&iC;rD5KEezZ|h)$dCpX%nU^GU2<xX(WX>WFwZqjL
z2#<eP|91CCOVC5FE!%g|WKO(0aVosZlC5&Nw~M7}eQ9aEst?~&WMYwrxGfMWhM7D3
z*GOK)o&)D#d$Do0$!p6<#PzSfh`98fl%)6fBl({5>-{hr4m@Aq4MGm#*)l!6apcW#
zjIS+Dv4%J?;_Nij353VEF}72IE++WXg@)y#^6ayf16|9D(^YlHI`H%B+05Bnh6p66
z4vAKn8A0YXIll<Fc~24cRzdBWI<OVYf`9_qE`M<LzEfr$`;%mIY>gCo+VDP=qwj|D
z+9hqhkW=$MV=BnkdN=g5+dopmIiNhwfBq`B-j?pgXMsG?_(56Y4+qnzm5$k_=nb49
zTV~HK5*l1@w{eQ_Z^+8L$SDqF#YlwjSs%$<os{Z>qd>-YjSSx&L{8Wonxi%n0kr5l
z>f*nH-Ho|soqS-NV4C}qaiHF7czb9%VQ&DVz<ZNYNKZ(oPaEaWW1FPFlm45HXo-oN
ze2ta6qdhj2EP^iQu$!0|e2+Cyl8bS(xBcaS`mq;uj{kA-cQ@H%u`56*gpvMwrhckF
zV$n=4xj(>0E$WD0v$(_&MOJ|Tl_H`k>{zncOolIxY}F@o=1^R!%uXd@Z#fi1pYhEi
zV^4-YeETImoz|Ng6&E9T@=)oxl>sl;o<h+9GZ_;TiTi=nTi|(UB2B|<kj!(e`o+Jt
zy2s<i^Xl^>p`lXlczW1n1rEfDz<bWV^)RI{{p1;C-anyq5Fxip6Av!{NA@A4DKSs%
z8g<yd)BC0rQ<VCq6+5gil8PQw4RZql_2=nu1^!m4Wf2PlBPZOe)pjHQlyaX-ZqiOB
zqLq28{c#T{|L)Um?xT-_iUZR`$Mn~OAeNF?`f%p?-`}0IkZyDeZw@-b>0k={fwu-9
zLYKsLLX%rEAYq{#5f%X5bpbcZ>j=JU<T#Wr!ezN@P$pv(#}ql85l;hojBTqfj4P9Y
zLq5I<zIRz?E0&>lZQD4C`oA*aq^g0_-QCk&f26R(>+=5IpRKhOxY~n(;FGhJ!}a5T
zA#;T{WUnKfQxJk|;uRbg8uha_HSB|V(o*6`wO6a@EQSR#8EgPKx{f^S&;;)Jw9#Su
zUA(m7>wbi@J^7eZvCHNCc(#0hdtZp3a*3L1v<HT%AaXlJU!al=uTImm*ze9^_X6jj
z4jgZSq|~#R6ZdHgm=26I{Tn1amhmPrltG^*V8NNv>Uy7(j+c`K`E7i$@C&uX9_@VO
z@;thqTKf@jcrL^UP$lvY$k<;sNl#z>oPX_!4Y@X&I9W@aThZ;+5y-Q}JJ@bazu}f<
ziV91WF@0B~<q7<hx)RmUnS>baf=jL1{dO?e7~*Z;aPU4pqHsl{m1GcUhL!K~{n6st
zIYs6%QEMAMVk>pk8}nUnR1iHU&%L34&OMvxWS4*iz4f<5gLN|K@MtSL>}Z-Ugt8b*
z8zaFt_<7U7N|q8l&vUO4)aIU^Z3FpZSxc!hxQ(W-4yJ>wu0mX~l=`AK2Z{pz!PNN2
z2nSvyYK4`NKEgk>XMI4(CQpq|!(CppS0M+rUhX$PYj@`<{Sei?hcV(PPqn(lPYO?y
z1_T3rTYXNzec_}v*StVMUcM8xg5UEo@E`WicQIk*cq($+oTV$Fn^-p9xK;7as|grX
z9fzgzqKV24CR-G#gBrwGp8Vvvyxn;b#*pEi?aT0YHi4lBlhZqkz<t5DUUfrl={n3(
z8l)cC>S>#v&?`URw&kJWNmv-{&rc-aZxDbq6{V%@6vt;PWyB?RNv-Vf96`|DzqU4r
zGv?Xq`i{vEzN^DaakV&vMoQe(?!4zuI62B>TlLXsO?EPYx4nv^O|wb5aDyaRo#(1=
zB)t>MO2M^XOS%!4{FIVYN#<|d<SqWo%nH?Crz1qMFS)c9`Q`rHug2cr8C?eWHLfbC
zO0#&m=qEZUm3zAGe_liGgn#3^Y6AYI=N_|3v8KYqJVOYC0UHov^>1pfGdiXtXO`YG
zMULTFc(KIlKtMs2=w(;TN=&raNiFVj7+VA=AqQd01B+~vTVz}(d#BY;<)GUtTIqe-
zq5aDj;U)eTv#&y(i_OtmfcsfTqx#vujiny#)ZOD1+opJh&*GxS;=3gU>Pugt_5+QM
zC5|9VsPi18)#P6IJGi8=YH>__^^-S64h`^F+`t~p!tavZ-(nA~B7JaJW{tT=_H(|Q
z%#hn&*I)uHlccaqB+~C`?ee0<mH+2tj%Zd@D{2I1Btq|A;Q~(Rq|-}AFs^i&rPbI;
z+!d+qX<OeiMWKER)mN+9%R7Q~d<<79CSv}4c8#TM^dW3U%#pe9TD|VLNkYf{Yxz~j
zv5W<>($XwVMHG2l^wM{Wx_j`8YlKNX>Nh7+m!eAoVlKw|_`cFH#d`e490uRSU%~ST
z8HxrInxf}eyns+N$_hJ+x(_jxR=Z^valh#dyha$<_+cjVMCwtmnd8W4h^sSmcbqQl
zPTR}qT-e`i5b#9C{fT_VW3Y({-k^K78tvWX+<Lh5nY&3pREFvgx{&f-pkizcWvT~q
z!0y3I7dL+Q%)GyHH${yR{*nhFZJny>oy<b#4uxL#L5;|)d1;rD?l&`?usa+H|LG)C
z2#=!(R;Z2H)?lB*81}0f?g3EL>#j835vTsgkQq1|#F{!UC4}QOvn-QVP;+T|)|d8B
zpM-5?h2G5Csh}c19Ui6T$Ut8VnA6kLDjkiL<;E~+purXr4J3iSAG|02rj=rJC9hH_
zFm;a&Ac-R9!Dn)G{F_aDN9MEOKOL!=nD)slN>xg2zf4W#TPh5@>r$4$Hux^?OIKxT
z!sIxnrV%x-Bc6#2lDaP}v`^8gG%gl@z_JzaUd6r3s-{{?y%vqLsUfz)4ZkBvcOKaJ
z*i{9XXsDJaYQf;~<{%tHs-7(4#$h42D+5&Y;e)>Wo5<WQri3@QZDTrXSID^sbtMkx
zdm96L1kB+uI!lSJMSQPk<#gZgczv}b=b`)Sri~3=F86N&-oCSuZh4oJ2HAt^LT(;f
z?aq4K|DD#ydxSgK<<;7ATozAa?h?A2njOe-5HNfkrDlvD$Fh6^1%b)>snRTzaRAlT
zE1Yye-ac(}9jE8y&UReX$HqA3Wc5iKTW?GgnCZyOhs{K6U<5L1tjQnARuR7Lw7scv
zWAcez-;zB1+T$tc1a_ozcyU}>rpRGKWk*t9RT3WbUG}QF_n>M$l;%T`k60-iP9pfO
z52^B=i70B6Pu~!zwlP^)_?RjUP@MPQJnXYkRC_WL(HClPUlcn@+f{{kWo&6YJ8o%o
z#}U|`jwJCed(cY>FllPgU11a%UhVtRmDy;X$a8yPEs+=Qf7Fr8c)hxXFIQ#<W#dI4
z=NJyA2)L0^l@R9mu7UDu%;d*yEf7Ow2HM3A|Ni?IZcqgy73GVzU-AH8nRyQ$BtHK_
zjPO0_kQ}raR=?QCgVBktLaw(3jg&jJ*nbXrsaOWTyI1Dxv-4bo$y>|@%OYsFh|BlK
zHxgCN8}h?b<bLc1wZ5H>C|K<mHGm-!FpbkeVw)QqH>FIaEBTkE*kAVw#nLty?|xj~
z%W5<?#BY@<Oglq=uSo#Rqh(Z-hI~7`z0O;H7Tp@+%(XaAU#c;=twtSa_xxKN%eSX#
z&h5%$Qon9+@JkC``p_cAfS?*vI_QcnjY*b*zM&G-DO-`F2=NR46nmxrMJfbhR~c5&
z?U^=EWUztRWed;Gsz&{zL+xX;$Gj$(`=#NRsW{%yhdVve5q1Gwgw`1*EyKzw-?dSS
zyygqgFL#ZLW05TK<o14<jAo&{eu)mA^I1tyano`aQXOCYufH^1G*KT-^-(xZJ4J)q
z{&Om^SG@WY#nsr=vTy!cP$LV*T;W?qm3W?9A&Yr7YfJL<T;mxz_he43^Slp;IoprI
z`Le@Z83UOO8Tf#!Kls%D*VJ3W<F!-Dupf!{N~>b;{G^IyebYAxt1q1$cYH={{gbj8
z$sVEqSTFs=o>Go7g9ZNeZI-HD@oJE=v8K|h(vLs`PM8g5+g^|L>Z?r+?fK=NE=}fH
z+h2`MQ%xYF^W17cp5bd3^++V4(MM$0!o<CGn;VjCZVWsZ+26wcY|OE_*VY-DGhfQe
zYy9s`j`}oq`*3<R$?YrKQc@p2pqOOgFl?Bnm5Niu#!wF}^%wdx8yTHAOa_ZpP%2$o
zEn0n88%fn`HeZ+@4;~S*vnPBWc`hUhdAH9+msw}G6(Pj|=SK^s_@HTW@Lb5wY~f)o
z@=-E84?s25ykortf8AR6L`s@rcil6DQfPWBn85Idy{k3tFsCcC7-=)h*u_kmb&}$z
zXg$DWw|}YVG3`9fl9`J3!M1Y3jgd{05!)4iAhnhHP<N2}*B0|?bwKV_ybyyRKn3xT
zyAsyi2*-MZYkniBc@0OFNHRBXo;2m&*VzJyUY;0Eo@&wu{q?iUJ)yq{VV6))cPD6V
zSM{O@H7oyeT(ws4z0js=_FD$!7W$qh0%>bE74eA)kYKS8$ARl%+|m$D5jwSppyep<
zXJ!8Y8(&VlPq1Y4cF)Q}Ur(t-B78LOftE9mqA0>|!is^imp}Y_#h;Uix{7gJTs*J}
zyw9I~nOB9p5B7u<9<owRRNAYkK<TK3be!`mAL`g`Rb;3L>bO_3Q3;kF8g!ZFeu~_{
z?K|Wmhml`3MCY@YqYBJpIh1HkGOLdE!wT*N+beni_cp!~4R6|t4vCxYkyl^Q2e|!h
z*v82VcEc*(y~Mn!NtZH~r+%gh9RXr_ouzhWhA=G4?``=${1lX@p6zjm_Z!orGrViG
zi4;i+k7tx<X5Z8diIWD+cp#YgkTu?j>#}m&RM~0FD18lm(UG1kz8`x%o~?u7YzJmx
z@(&D;&j<%~o$40^kY~8*+qrE_FWu)?fCoXs41C2B>d|4RD^=Iwzk}y$fx2K^1c6GK
zAFV$>?~&1x%-*1KWEpL-MzyMbr3Cmd?EF7ZnXAReAI`YpgBB6ZMOXPG_p$?rk=NZN
zB3p>yY@Ff?4qdGbsqf(su31Ki)J9|r4i_CBbACeDc$5%=1pAQY-uAwPvlv-LL;!x{
ztRpvCXOa4jXrfj0+`mnrb!vKsH_2vj;lyYwFGsK)XjF^(sN@y?nyqK4XZ->MloYI{
zK7k~z8526Dz!DxQpruGQTWGSHo2@-@h*Z>(Bg1;Kb63y*Y@tUmV8-)ZIt$-wcs1t2
zpZ!(<Ity*B(;fTPlgvR1k2yWs1*<YwU*iYGCBR(M#@+@2n29L()Z3i%8mCJ_Uf;m@
z6k7k-larKvj=E`XWL*0sp}r;ZzPuE#t1Z9Zqoi1oG%C(mG_PHo_03>nW6))KB}p@A
zK9x*q{6~ZfV`e@@>7R!;zffrR62PPQrr~fE?79p5yPq5O1GJALzy!G(%>nviQ>*YY
zKV+YcPwvS9`4Qf&v-kI=3)?vPFM2j8l5|l=1OGov@RJkLQ5Q>Yz1rGs!#g=S_~Xey
zw&~>$5u%uVxOb^d(+OcOAO<RJKOBhBjSQTs-|QW4YpiX;oStnMGU(K<9Rg>6Ib8J-
zTXzdlfvqK^6Nl}JL+k`Rg0hd@v!4Cicr-wlHTsJM;iT&HR!6BX+dSlx4FjZ_%PRW5
znu=_CIFHwD{(Vvs7j<$!F^N6}&H14)Fa!3}7^8S!P%3s<D3!k-833caE<h32+dncY
zZZK%(!Ve*Sw;G844jL()H;VJazN<H^0Zv~8Nkwh>Sd-qPbT_{12^=e6*JGO|19*zO
z6Q;CcBNMx6$wd1!*_%9?$J~AGJ??UB!R3~OGOyB%1l_y+M64&?i#WqLn4Fg!yAypy
zOWY+sSJp3f-m=Xq7T!G(DK)f~d^e*j>DYv#h=N9&IjDH4N9LAz_?;|DxVc3$3alwN
z=gtq;)nkC;{Y{%fRu@K@)i8%j;D5!!a5~HCp67Qn8^S=0ZC%}Rbc#`ePKK7g=%&+s
z@4CpgiHNyyU$F7PoHlnImDEalo6w<tjn0lLo3lzZ>gec55;#9F?L3caDbjZG!A$&{
zw&gXx*)@u#mrms0@8A_Iu&FzUaqPr7bhdXyV|@6$V+fPx2AHQXT}nV!*1`n&wVxHD
zuaa#~#`#2wVuHC#1wy#FfLPanjifT_!6?R-SLnP<YxC@tY*UuYv{L5O)IkUxzDJ6J
z8r$Es2D>wDeFz;GZ(TO`dm!xMzHHKMuG-hABn5;%)C}Ueew@zz^-1f1U%T`aE_ciZ
zUnI$=cKUDc2hd?x49whGO-9FDl>7QRn%$rdS&)8-@N0_QVoaQ-KUGS1Wfa9s1$fy1
z$?Kl(U+bk9Q>rgUQ*)$EHpO{++k1OH6pOs#-o^6<D!{xyS(-1%TrzAHQAmi!RcthQ
zDwq5)6*bYW40*gSP@Y{`WS)w%ymmEf!E>CA6>U4ytqKY?0q`zB_Gf$W*=*;rMBORM
z=r|@)ZPr@-&%MdkzYlE!OhipcEji|18%uB$;p|4AP0{dbi}!s+W)~8{<{l<Ex@{Xg
zk5MR6s^7%RtcaR3stO}b{Z)*cf;(rMa1Fe)+Gov=5&SCDdzHMODQIF(J>*LdX1(yU
zN!=;hN?mIJo&5BWxIRR~LeM3%huuUl;jvOVHa}RoUCp(XQinQTiiNgzo)?~XOG+;>
zPEr?k?NN{p6JxgqwbN`QVWSpT0i%@8(<Ak(7nJN#)u|d22R|^5OLV~<?Yyf#FH3Wq
zwo8T2`>)<j!pR4)m7{nk*6c;ov}0D)CJO{vEntAV!0q=`15>T<()BZ3ht<D7EmPP1
zINB@C9oP<srfs=hA+m-cplQoWtows(qjyAPhVKj;b`<>$Hwk3c{K>Ob5fjsLc7SiX
zI41jcs)PMj8u(LWQoYe}KrxsC$$i*Sl*1MuIoEU&(UueaJkA9wU;q`D4$!y6_@4ND
zS_QmCnTVRakf_u{x*Mm_Z}@JMI;Y$!KtS~pq`w6eT;-GCQv+oO>pe+Tx1DXaHzGRc
zw#fo<POmv=&5$Y7m1g6Du{D!8ySZ90I-ut0`vbFyA_ix;r)Ic7-{=||>HL8r%qA3l
zDzbgVB0Y)Y>HBN}P7pnzP?~!@tdmRGfQ-&{Eqb2#ULic&kKE&ZsX(E5&8GnzA2js0
z^Sh+R%7~a;oTV@$K1y@hd}?;`<CG2I=wGCwB8{xx7##C5CiUk(?Vbm72JlddyWsJ3
zGMHd^va`F8>v(=J<#-WXAgObDbX0cOnzOm`uUenMX-m-GXXhQ9(?W(?Q=@4G2rB;3
zP;}zm+N$jxPj;L3Dc`?}Cu_!VuRYn0ywzM&9lq^OwDZaHKI4bsV^><IOCQ6>va8_V
z&3?z=a6|YebCi4uS}Q{}@wq^+XF+9n)@b#A^a5Y=grNNU4rK6HGaDghwzj*Vs&EwS
z{STDTIZjG<s8sUm?dc5-veyHULz61uU~8!=aleo6u@1iAF6UE6SCk$O{;h&Rdm{Sn
zWXyB#ZRLPJ>Ya9xzE1z}N&geb1-QND3I7&g{Wthq-PBG99X0P2J|TA!|KJ-3$aW>~
zjPiKg-Rp9YZ6Dk1UKf<OCKmThsDH-pGrf%`wo&zwa)52WK08HMArQPV)X=(Ds?rv`
zX`;hQMfE}KNwrmg8|}5rjcmHKgLsu@<+mA^xx)A`zo%_Pg_8%7bmjc4%&pseJ6cBJ
zT-|B2uQDiFyxZzlV(hh{`aFBovEmf4wt@m=*IG;ZM!T21d|<#fvwpv{u*MZ(UJ%OQ
zt|8@PMShb;t-(t18i?x(+vb2FTH+7Wwh#y;mzEFH5Qi$i+mmksf>IeQDJ>iQ#mqVu
z%X@5mCda44s!pWogN=<%;QVZ$^P<kQGdet+U?YnbU^JTbES9KR*<A<%Ljc*yPPW^1
zKjs=wCL&&_npzaCst)~9nt5sSL8e@#&`Z#rE3NDeg-n(v(5D@}zceQ%kmS6KHv8Y5
zP0m2L1F0x*P?R=v7gmw&CuNjpo`M7kE@)@S;>_sh34?c2tGT9<b&gvbwj0=&<}-ZY
zHm{ZHFY@O0;%5PvY@H<;Q%jDv2ifMIGdbfRG*SpO6+#ZeC%p(^R<NrR=|=X6R;l)d
zCs!pMgWo@`Eco5{E$Y<%9<OVxl|@+Znul}bNk)Mplia?(lH~P}$QFBw&)6D24#g$G
zDheO_ZE=cYKr#31qBG{f?K-H5h^8HfHY0~gZdFoMdexpQ`ZB*k2B-+Q@OhiH91`i%
zI%VUU>R)W0fnLK&hix9j;M6)hD)FL2my*T-oJzb<K@Rm=LWA?o3T!QNqJ~h)<30@i
zJTdU{VEdo5m1CRl(QQUxGGd?2LSG<pK?cE*AhDmxqqG9Lt*PcQGsTXb5b9eC;cyLF
z4(AZ?d;2ioQbphk2Qhl?dxWF682RLkuqO+5lz2qy;(^TB0DI?YQxQ{2JZHxOzky)C
zQ0|c2OzZeN?6WBSk@k31V?Gkn)N@e5+s7%S(|!}?+FR6Z$|s5L*g7c+v<Qwx%|<#J
zo}3+Tc3zxZAnMu94`g)V`4%&U_M{-|zXu%{TjIzB6Ju1bt?Zb}rD@1N6;J5Xk8XFj
zu?9R*Y9Gi(jRGRP$^#!Vo3)_`pxdbYRjUe80GjZ5lG3R%PE60wSBpa*DgsT`U@Zu2
zFZ!EBGc)agYC0+EQB8U{EIm@s*TEg|E2)wl21`0Vr2FEozH`Gkxkq)k^*-HiB6+MW
zlalPE(G9m8(rCn-9}=p3#IS{wjqY3egODUT=f`fb5(o`uH=$Uf@hvIfYS|X9N3#K0
ziat!=F}!|_-wS6um3i?<DXi9hkj9BJ4|aJ`^O-`hg_i4Qk-jSzLB)JI;0<oJ3-m+C
z*<s_^if_rq-=-hndq6A@@ONS8M)qg-s1B$OWD8&S-#)RKBpk#<<7dsYIkXA_)NW>v
zP0tf1&(B+dHtdPPJSRX~bv3msF5|)cS~&pVo^kNR#%5;qugHM$SM``?>V=OON42vw
z)ShY^5|IcqH<G4y8Ag_vfEl!W!AwYLeq#ri$=(EMNF{Ym_bs1G)40e{QElkW;Utp6
zY3=Q?9x7Qe8+YpqKTt2ma?EhXc2WHC7-#H(Hp`SrRk4`>1MQ%UJG)Z2SOPD1J+!S8
z_cC}upxQMl#$J(?ew63KOPfaA9~#kONl=IjD(uv&MJXqv<!_?8;xaSP?1gOtu}@mW
zy7^j0`(TL+x<fo7q;+Ry;PiMZsgFh4R0j@!b-cTNou5TC>iD!|Y9sKSSXojR@;2v9
z+RbvJWh8*%p_F-JJu!*n^eUh#yG%)$uQ_EloEMK+`w)kSP4^Zr^!BEmg-KnP-kg3*
zg=n_EuHjBg?d}eD%m15gT6NMoG3iYN+w-i#IoW&Qrv*uN)yKCF?wNbj(7`>ixOLlK
zO_R2wCylef`SBD!fFgTuU3M3k-PSE5XwvimWnJB&oScX!>AbUTAsYtN-7a6r?Dq=u
zO_8hOJn=@i<kBr&dA_}Ph@0D^&RWdY9#(yrzxuPGO?%z74HOj|r_ZUGnqB>I>Edrg
zd4`K!RxxS6-e~HfEx9$<o%B1c7KQHB$>SLxg)?YK<DGpU`OAr6$!d(hq?GKoZ;B#-
zznSlxtPE)sziO~$!t!X%7h`z^nC_11+j50Z>4?xt(&&bTjU<dRRa@6oTq>YP6daRh
z8hS*OpBT9a@#SW#Glnd6yp%9veoKoz!0m6~vPr}itiPi!-Ap0I7HHe5u4HBy`Pcu0
zXO~S!NW+85EfA8BCTFPgNK3u>LYWPqK=)<zbm<&%XpM;6Z>yom;*2e5i{omciA{iw
zKJ>_9*-uHgs}cemN)JtPh3zZlR$=k{VEXWkQ004ZPN)K_znB&u-aK@?U)smzAcrxT
zz9M0$fQBz72REUWkToW^qBK-=?NYenc_Vgi?srvg1fd}$=<+AjByXxTKL9~cMOV?C
zkpI#_Ag_|Mz}A1~`QP0s$Kv8*i3HSiNy&Yk2PC`)d#k2_=!!UEK*n^B#{zMMPmu#d
zJ~-%ziEN7@9&(=po6+kpc>6LP9W3pM+XbAl)S@ZVOpzBC>R->7nQSM$bPamt`&3#t
z`i+3MF;sWQYX4)F?RGZ;Xc4hpgBw$~Eki6x+ZrX6qEcR>H&<&EW9xLKbre0Oy)Z2~
z_}zerg4`ArttV81mje7&C4ADIm>Fg_XdzAsHBW}*p5sw7yb6Ru&uM1QRL!>%m*QR&
z^0S7?gFR@07Ryf+*}}Qz2NT4Nx{mX$LdA=E;+EmJ`T9YSJ?jdH-bB+d3#14(b2?8|
zUh9#>YaFqTGc7uI%4!9RU@68Lh|%OiAZl;ZHzcVL>&*a#E;T>=?8!xLBle~Hs>htJ
z9|ifk+XvA+kE2gcltL#cr7?MaL*}-Qx2%0Me&gpvi?GI^Tce+gKG3Dr+0tTnLhH=V
z@HCX&P0dPp3ImL68{Mb#!b^4Z#bwINnR#?{TK~$!i%nPrQkR+ujWd^e=S2jbFNI7$
z&T03sfak9wSEJNkHRBtJzP_lAldaIUv(1H_c;&PGL)Xz?cF!aH*Yh>7SFDMnL*wm@
z)fm<rAUIFrnZy^bPubPIpQRF%0{)i8QQaB#x|>6jNdXA3UWznm{?%)uTHnHNj!~LA
zJv>*c2y2@&`5&cPKtJW9v2*90q|GqvzeV1%V$&~?<4nm?{Apr_d<BJE8jd?KehVM=
zjJK|<>iDAHlqoYb5`|WZlpA^~V?))EEBQ1$SWkzPnvghAOT?eTe-gfJ1$SD)C0!Dm
z1L6%me*+qbB`A*2Jt+OfQ{R=#TO#<I;Ta7UB@w6#2^BN`0isT__k{1uTfS#N{wK>K
zpdhwq$Hakdt3vrWjE_8cm6?a6L^MjwPV*Sy1j?qX_b4Ga6`JOhxFDeyf-vSmN+@yf
z=U~?i+=hylMoQr_*rKJ^nv!dqrYvoP(I)BxQcTV&UpF}vM9X1$&H0@`*gue0_*YWB
z1Gtcr91@~+Pw6fM?Q8y>rUakVIEvDjIja>DH&Q*DUcg+8RGGOFU-Tszo~)Oh9|_Fy
z1^S@-qwU6XGDq_JKNKcVWk%AKg`R2ADe^l#kz=mgLdB5Jd_gLtVp0z&EULZp!Vl<B
zxA9)xm1A^;eweu8!qW%Jgxcm<dir@N`UfW(8jr(TJHZL8qod*XCWZc7mj&%WW#=UJ
zRfOcL3=X)ZPlp}pq18lEI*Vz#q3C#ZXtJgoMY$MP1C{8wa>KVwPNW~BGY{Tf<$0n1
zOge$EiBg2OPApEl<SM1ZGmM`<1h6B)3A7_TJS&G2igUn)>0Hc7x&X8A4@8tIQF0)>
zxa0xlQ<%1fg%Im&4qaj(W=ELtQ_^rhG_<J&TXBzib_=z=H5;`@IsuJIJyMh-!Es}j
z_wjO+@UI+69DhF+flr1(DbRm@If9Ob4oa>*dj@$nxj3os#6dHpo6!sK@Nk+|J<IVL
z8jOn;zRs_dK}Dq;?^1)YeE%!gkQZ?|J}!~JcU;$*x2$Ezuy%fkNUZYYak*i=7odDq
z{VK{KNeN1$sO3WMU(`sjvAy{}PoGKOQdCPSzJZ8xAa=AOXHZRTXE7JQYXZnyOAWv5
zK3a+4VCx&+z7>P4u|U(bBEH%N8~b1Iip}`QXGzXL@4ydxHJbuv`$EZzstJA)58q)M
z5NNbisRdVJkLF3C<wQ=kech+GIGp<V9$(B)6H!|5w*UH|Xuo@g4}(Rvkdkb(W}BB+
zyZ~G}%<qzuOX=Ed17~yG4Y^?c7bZ87#1h8e7aS^b%7(6etO4SzLoKP+4ySsJoYsOK
zet~rTo-{;{U3#-TQQ##TIJ4Qvif^D22-@G9d0nPc_Zz`ONkj7INqC#@=r!QQwDX_M
zXxp)=I+6T@9hM&$Xq8bx3KzwWS29tsvl$di9)G--xu!AnUM{c3#`-QLJV=W_PoOzc
z-p8_rdMh!~@;*Sv5{WWwB`Y+?9i+W?To4U<X|5pzxgM&({aJ74=gl|D_P=n2l+3B|
z^X~I!3Y&^OD6S9-15c@P^-kdE3>?JH<(pG43Y6ElqLpeSYQWVU!c9JDm@|%xzk*ko
zl8Lu;H}%YN4lsH2=2+{$iIOcz-|ckpI=khbBkom!wLIXh6Q^W1LEuE!Nfp&z@|cgM
z9aKtN`rlH$NuImdI`Qpei}$NpiBfHxzPfk0+p~@Elp|4~)wk_^7@9uAa;VJeW}up5
zPSwE5u1txtk)EY1mlmTxdTrKg{d*gB$X!C<3AbQ=zWm!Sh6?i3{6ZZDst5A>HpC!V
zKc{kV{%{k;qv%|esARBf=W`dox6)8#y4sf%k`JO23maqE#-#K!q%E;_gxz%=A;nO~
z&5)+flRh9~Z<9<oyD7lZ!}0obJxM*|w1rDVgNn|(Dx7vg0IC(>ce3J^LZR_wd2$yL
z5V`l8;M>vG5`L?aDbUX`j)yt6Ux(sX-#AkR07g_lkOI`y6JQzb4HHfuFA1i4rWUTt
z{D59f&|+LmT+`@v<{m4Z<+=Iarrr}G6!uHy4J>(vKHJ!g`+;^vhXV%DF5{42?bTst
z;LRf_%R*HlSLv+dXfsf?bhdKJ(Zf$AsJS{>DMwP3+^Q}JECzK`GHHc!siC2cK21pg
z#|xF}0;)s~>jZ1(m(qV)o;y{H9^ZKMYlz+kTZ>Qo@@kMi<CF6D8P)b_X4}t9Nlcen
zu6(v>Yba>bEy=u1-?r(IINd4}Y&P-r@%-w{`RT%LIuO&ziO*hSd&sjGlG^6~Gp#2i
zwL(Eh?Ds^(4~<BI&Bz<)pu1I-zth~qbb&Nvb97q5MjqD_hxm2o$}@)O6~~Y-j18n|
zc+3bB2};2>GGx}qD-glokBVhNCOLCu6qebLov1wO*={*9P*%hV(veEtSB2Giei*a2
zDv~8(Tjx7r4)Te4N<w`Z+*iI-H_O#o4jEO)TMQXLxBE;{-HLptDSO!gIfmh+1fXW#
zR^Bz30xG&}eec`pPxFSlW_~6oCx5e3VQTsG{yrl@_0o5$mQS`CEKo)<yP-jPE7UlL
zAD5f@)56W5Plc#Ze24+!uhJB7-ow&)fhRPjuj=xmK^JD4^q<OxInZUu(-)VF_3)ms
z4HQGbT+5}{SWIP$*iDAgCV+oH-ge2~-S#{MYC)O*g^RkVul){@2XyTeSE9}4&T=6-
zpn~G7oC#jZM*%;=)cB(hzqGW}Zs+CGtGdAEst+Rj3I5ai9FwW3R)m235{cD5n@0%!
zp84A7zg9kpT{@-9^S^6KUIu5EMAW!HfA3#2`$?76-Z#9Ya|qiih5$aIrA!MOR8l>c
zGbm|n<o+ANvRh1<D{r#%qB%3PQrX_Ht`~7Se@ZU&1e(JS{!KLb&PFv6=DsR{8)$tO
zSW|76%&xX7naOH@s${xHCD1xyGrW2^>dr3dS>XrdYLRz_e`j0x+3%r@Mx)(Uy$#u`
zOwd;pPe+sL7j8wL3<17|!|CZP*y_09a7~Eir^$ryX`>_pfy}r46q@hKrYK(%=kbZ~
zm|VYniFL#-M|Wtb5aIal<A^0?n~rck{DG8ZlY|d&bA#b#OVRLu^v*w2eH5?_9Eq^k
zOtaskkZP(kots<AW?FQNmCo%6dASwi1yjt=FW)gJJUb65Xaw{F@^jP4ET(-Hiu%G=
z6d{OB1(@H%u#?u2kCX4^@ssUHtEAr&TZ<fG{Yh1o4mn0`am?^7Rv)Avnl-0A+nhl>
zIFFV&ixo?<Uo@Do{Zu&#o0!JiT^!}I6iRP4;>nJ4%%Z4``Z>lVpH^{Ki|eRB%lGzw
zXJH?xC=o~Vz=-?}#whlsjh9f1`5y)~TV4#%KhbSZ?#q9gHK3Lc4fS%s9o&5I-vR=$
z=U!X;yLy?{sJPXib+E-gz?*sRVVq^Xhel6JUdfeEzw|tWrC5a;Z|+Ioi`FE8<6KMe
zi#H39z4zF?3mze^fcFf)OZk8MNZ2nn$Ur&x?|9F?cLjXW8FB5;cn`YoCslii8RM+9
z1Kj*>Q)y{2Yd)DA8!Q^JPT-h;w;F}cadPER9578X{0X3hC8%AWBsE9Yb)}jqP;FO+
zeKJ&R+eIl7Hx*n{4CBcxRTux}&Ufm3`GzM_6RasM+-B0&>j^V6#)U+-gn=<RMc?P8
z(%gF)UjgH<3o&JUR|j08%v3Eq)v@^ZhVu1^cj>t`g#Bpss@OFjk%*naKFIg>Llb@p
zH2H<r?=bSNiy%m$t>+<V4Eyv8nOk58-Za3pc*c|}<rpr(1>Us0lqpFg@2z~pJQo$P
z77hOfs?i=-G<(9<YyZqTzo3<W6#Ab1@B3+}9bm*Bwwa}oyS!fN;bSHC<kJYOTPk5F
z=I2$~<gU0P4M_}mQ^5=sP1sy;CyjKPJ-O@T79uunB-9bUM_kzXO*%ZNY4aVczmx@`
ziD0HV|5xa{8u8tyc-0jsH9~7>_3GRFUEe)__VqMZI6~HA2!2G`@oK_xE~5EI`u?UQ
z0=Qw8E>w2y@9FuCx|kP%G2}xZZ}54GRy~`{cB%#{UfOk~6t;oyf3H&7dQeos>ZPQ$
zw>(FFvK4F+tABTUEF9Y3fh8ArSyiB5Wwr8;m7s#A6KDM%et+;CprUhzn?1aysNyKG
zJ&P;MZO>OoI%A?}S^u~UoaN3%zUczuc~G#-JBu0S1p&hY9~&y|-R!j1F8CHjoFP~{
zn^5LZIz)(uMu0|-gdXvgxpz&)#Csc~ns*1f0r!RLXNv_Nu>D~7{9N%{H){AO;&n8K
z8;7N&7k08m^&#7WfPThM$e)x%B8PzYl=5T~OV$2Tn=`b{znNIEDMvDwY#9VV_lEiB
z3d*<IdySMn8D`Ko+QC&=o83=q&Ca3MDcxx`1VWO6KVOQCO<dDw9${;ESL2E2H*<T7
zL3&Ir$Od{3+4ANo33oZo+hE7TXZ}E1-pX5A$!4#4^y~9m5B*h@r0(w)@}K4jz)Zx5
zifKZ!{ohoCNnpmw+gqlq71aRC5?i8&dv3*wImO)(`Vqo}=xplSZDS$?6B*ajdL@7#
zqUR>-@%wIJ_h7fah_)XD!vH0=Pu0Ic!Qw{zZZTp~SRp7V!vi9>#?gMgRhEbVL$HJH
z&vGdyX|!F=q5AphZs#eP?>qpoX3Nk;$M0}F%v}Vnhy*26B8T<>uz6N8$D$<~pwH$q
z@0?H{exjrP>dz7*yA=1(?=R(?jqdgljggO)yIQ*zroBdO;52cNFrRJm9<{a}q81x{
zw6hzetSAqTsdyTN^E3^PGTPv+Z0ii3!jha*rIV?zY3-<%q=ACftFtcy=qZw%J*<)v
z#EsU{K_dl8fOZS85odnIGFmNwR+JOA2)J$o->dnTQW&|KU5HSYWsZNWxP($ig?dh>
zPzNMT-)n_=W7x{Yg{Qs`cD+XHH)K~Dsua7I`zg(wZG~QVenOTv7<vxKb+rc3ga*YD
z&q;-fbJ)iANOEqt8f8Pz0t`k)*=8z=>u5Qv$os&YVF+L>^YMyvQBC=WsHvO6OI6}?
zeVp&>Uleb+@_+)gxu}z~o~1UMlxP`54=JShQB2g^ibVcF9W-k4-W^7}WY15A=vl91
zcw5%<{4$L0&Z8B^U`gZEnZ|m~ME@ckgqemLqd?a*T~H|KN7`%mu++PXA8!N4uaa1L
zLS#1+0ho0{o^US=@d-rRw^B)mDp;nz(05DgmrAoRut-Wna_|elFm(cSpSF@G3~Gg&
ztRk#JAh!b^Q7Jsqpk=#_>Uvg&d{;;b$&CM_{iL)!5i@Bwo!Z<V^pjb8|5f3?YW%0=
z84U_jHMkV6_mO=acwf&DA1{XRge=u_E``4X3_sWz|9ls09?{nr9WlbZVM9xgYh&k>
zl7cjcJ~N5ZOAPAcyz&QfLz_C-(qpK>bKv>cpSVYDlFhvyvjwd+m&%kCHjQL7diU(A
ztwn#$puhZL<AEwcBsK#iFoSVV^Y2i@?&KlQAMczwK7X%v|JJIq<jSs+`YVYF&uWG8
zUSQC0bebobk#R_cvyW(d8-MwdmL-1Dz;dC#gs}=1JIFhcT|brZRQCzLhIWws!!_me
z18bpl_$zeA1AWSO`J8eIiRDL(%JwC1Fjfg9W@uN#*l2(cOWz(L-86S>ysRcmZ^4)g
z#X{&Aa-Xh%X<7e{Qk7f{oO#Yl*#~v*2l8@wW^W!!_k3xlu&kop8i&ifpvIX*BA!sw
zZIhUh*{MXytI+TRVTTnLNZUByhh71u9ub*lO`fmWdi3y2+yADdv==#&1g_C%=4FmF
z;LZkE)X(<&s;Zha`4c|owy#4VOzU@aH1Bz!atCxVE#AW|lhX|}|G{0{Aety%V6U11
zYva&>HHp<NwG`@H`fkTN!y!POOy$$GbY4&7t5kx0xU(EXaAh=8lDbV_(Ksd-blpwf
zJ7pT4)G$YkIr~RwBh<OfpoJ<ac3bT_0_YR-n3D+89aK#$e5XO}d|#Em!k~NW;zk6|
ze_w=PGr7Xvd!9U<$E3FGO7nYwPVy;((Eh64j%C{W`{Je93XFxoQPPFD`rVu1*25md
zI!k#G#Y(SCDw!dg1qOjlmSc>m(665T12m=O32XgOVeQ<Zuwe_{8?Q+f`GS2)D1|qg
zf#xn?yE|2yxuOBN3M$~UY1V|w$j&_P8&`JMd)h+5it*1Oo%<Df@i;(#<^$EKP=Y{(
zGl<^n%~X;aT_A1w1*(*frU&{G4qnUKgx;3(#!Q$SN<8$$yv5E^2Rp?=u0m*}Kk8&b
z22*CBkii0HitE=*sJL&lkOs`Z+x`9a5Dhl0{#?^SxJj*be?y_5RW0X7=ewr9+uMgr
zL<uI%M1IW&{~Gu6{eYJ5`;Z`Nu~_A`@2QlmvD296o<sj)j4h^kZ$gxm{5i}iUk~3h
zqp}<!Rd1Q}YQ|1HYsrL(C_{qrjD2=Yc5RI->mR!otMeZ$YK&_TAushW*%X`RF7<>I
zfy@hKqooBB$xk-=v($N(-%j5^K!=^l>OlK=i$~yN+CB5?2moC+Nn-AHyYE*n6*Mdh
zfnb)cWW|KFh1{*zN6#)jwwKAOi@xK$mGe$M_nhQPnrw~Ct^;S#(L(jqhh7G6(TN_*
ziD#kEk%{)|6&ot;AA!-fqxU&~{IjdS(Wv3Xl@#_9?9-9`Q?TRsCH(OM+(>&qkM`=<
zuj`N37pVDUfMO=@GRn}=Xm@xTLR$<vJHyfL{qII8)+#yRoorUZL?Q(4O!GF5>WK~2
z>of450Uh(iXF7^30$;%O9Uh3Eyz``&M=M1-SqBla_psx1gYROMP;%~}&>0czx76Oi
z!l&L~1N+=QVU;XJO&@piTI5EvOo4>`4=>SI#F!z#Hf<8w2k}B*p)`46RM_14Yj$R<
z%15pa%cgPV3S7F0!<7-CX;4Y3txnKSo6wMw>GPk7KAUKoYd_x13(<i~Ydq4>p7Ls}
zT}>1c^>H9ZS75VM)o;a6L*Sg0l2wBI2^{CorBBF9hF1H`x%l>c6w243(4{8vE&A)?
z%q3vu{(8%!Xy+XXfp6|#*~o2#(+Fvw*qi}rMPVU#Ph;CrehOY2Og=2m+@M>0P^(vq
zFw`5H^vYp)GwS`S>xRbDNyBbt-1rE%@8Lb&6U}q;!Z@^3s4^i>-Z@C*n%_`<T*ZWF
zG9T^}gjBamO)9Ia;V!}ebWc5D%bGHfAIK`claEZuFmH|cg;se-$^Swp+bOoytGw33
zVEfOA4R{XCnKbak1P>{@-KK!j!lhxiGc|=^mtdS_loUVj@TO3WzDh0BHw$wuH8v$z
z?ctMk4nyfEKavJophMmgYa4=}_{CodZ_f^Tf3owEz+_ulDUM--W>%KCoOH#v0J8@=
zT{B*^x};e9go()A;=H?sluoIduNFCkpSHVs{~p5D2{ECmV1(;K*MAI9678>K!XfpI
zlLFsp=V^!JSHlbRvIWYDR!@UeUl2mL61xO8wYWy(nxFJUNuc^1tM#^b&Udp7Tf`3n
ztYLXe{Qaa^CVMYT0*98+wNi;3pc;cU_oDmFA1vdxnh@Wj@RY<b-%GvLD`dn_Tm3Tw
zIuvZ_M5E{^;BpqF98HloCEb|gpQp8N;_0_J9;GF<2hOs;p6Kr|F1su`9wl2@ha&r;
zLP0ScZg^sdu_b=H?dsUXLV2qvBTrJl_LB^EV;1@LAvtcj-&|3=fHS!Nus+C|+^S@P
zj`^RXNpIikdN~i<e>k-s{M2)${h8*+)Iwtpy_~fX8c!Dk3$8v`_e3^3R98qJtw||c
zVv??YepJ`DME8k@Qrf~h_}f5~+LTqEXVDm{js3M8(|o1zgntt+nH%+G?mXxs=zWXV
zr->DiR|H3EPz;Vd|MdhI$?*I19Dgk_Zrp1aEK>OLTv8F{wlq!Q9jdo(ISj9wZ=e7t
z*@nrF%w}U!;A*evh&vxbMb@xu#ID@v4SDKP2k_);r4$Y|7<&rx?3vRKn(JJ<{azRt
z)j!udFkndPA#U7vl%;uaAJ6O#N$elNir1z{Kxapn6P7_y55Q0GoZFg(be}Dm#&pz8
zKS;Qf_t1!5rRWVOgf<z$Q{J>~_aKquB^L*4M*ZL?32Piz5X~A^k=Z!<oA9u>7Ar>_
z{TJS$26+qyv8V)6a)xX$tLBmPChYZcRMULhTU!nkTRJ*g*)Z$Z`cO^RRBbMEu0<*8
zWPJ*V4(;MRfV=KIvObEFnQQPGu*n5~0^e1PSCZ0F`5}}dv8h7hnp#e9p#U9Q!^OKZ
zlE3Yudt@@DJ<)Ob@8#QC3*T9LN4(VggP+TDN<E#}bneWbi5B${4gPizR%{wrY&f7>
zh~b$^D>A$O@Q)WwtiDnA1Lzk|^oN#rj?yt(kTnz#?a3+Z$M(Rwfap_c%GKxnp_^j-
zJ(jY|Oy|M!|J8OaOi3mRRAMXKHY3c}sw?=K^|ATZw9-V)M{7oe6e75;V#t+zB&cOy
zwuGeV-g49ojn)iT4X~@^4NNQ3vQ*qi35-fp$h2BA#TRhdA92r|Gjo2znK?7(C{&`5
zcfq3HZiq1b_=eno?xETP!*bz_lGG84{?-ST;tJ2gvDenq_YWF0+1hvY6e&K7?8<<=
zyyf9q5|G6M!PTCyD^MugSa1g8r-xz?ELu5gL-Yx*ih1$j6rYs>P%|_4y;fIVWp9^0
z<>eBiMt(^SG(|mz&|0n&7FtRDdy#Jn^Ai~@g*jDaH;<8PC`oZ1!hR4PT~sU<%enY0
zDl-xz1Y4%U;2EylNtRz|%2kw~C$l^i><u?`a>0&`W@mXt0qqvvenFi(^{DfH9q`*u
zInmy*#ltnhfOkjCcR4$3pc_%F`rz*(;>dQ)5;4R^P!;9e&JypncQ~`I*>)fl^ylvB
z)3^I;E<f1Y4*j*xiseHc@ORDnP-Af`aX}?48u_U<ksTN%{xAtP3BGCzifPJmA~zxa
z+STvRgvxpRYWuUHWAVd3m!BHBOLjt75@rlPQn@-M@DVE@+_RBdh97At(v(3s2wT|_
zDS^YPXBzWb{JX$fbjwpbRyXHmH%Lq~t8|Mm)qVUp>@jIFdN-;eleD*&n2;jmCz&2*
zMW(f8S|Sx6sTX4J%_Q!8HgwDP$*zm+99+pE7x2lj8}~gKX}nX-C1T%*I6G&k1zj*K
zC<MqBKOEAIcV?iGBp8ymK=29FhEI{|k37JQb8k*7M57xV_vdwcHsh>JGA_Xp41P20
zh5sW0s7z;HxdU=6E#=!!$2W~z+y{wnCrTTy3TG5?JYgpeBlCr9eq{2b{YwnP1)hW5
zO^ck{lGm=Pv0zKxqxZMnwDB?jqgCYsNEUq6&uuX(GcK&4_F<Rc#5Qq-tk5leB!6V4
zKveIQPrcT54Er3hArgE9KWV@!8t6dgzVDe2zBE#0SGOX8_YhD2i>8n{_o(b@DwzIx
z9QATCAwH>PqtCgMyl@jVxa-=GD8roP<XB3|t}|wj^tz-$ejH;HpW3$z0c^Q!jTFdq
zhlh@y#q(M)`T6sq&B8#^0A|m#SErZKt?`l2Kk5jmAd?uRHS&0-YCOna7joVW+H$Uh
zELFvQdP^7{1#NfnKDl)#221?C^!~$dUfQx-hp&jtB2nhDvL7JVJn|@|kSBI8x&!BE
zH&mgIvO6a10;M`PVjOA#P;_qleOG&=!v4-LtN<2$42|*bXscPbaFY@zDO6|2R3ySb
z*JU&4h#%8<@79DNSKo#E{pfhfr-~jwYAl#t90vpbVz-}sd)kg^{p*|Q0}xLo=lqyr
z_UrPeKH{rfY(W_Dp#5$)PXwe9SpK_*Y<n0w>+#|Pvvy$*htdXQhe+;ra`~HR0b7up
zH{@bnW55neWb5s*=*!ko3}JvP9@|JEb|pQ-p{wD>f33HPO8e4(Q_biCVz2!3XBIN6
zj4>q}d-Z*>Tt%rJQVvDfadZ4CCdMthaAVl{J}YFIrhKVesCvy<yQN(`%>^cG{@c^E
zF71^WmOT@J_K`BvGy3FS_sx48K&!atVfGtM^!i!BSa~(?>&k`V)aSs`B*%VQk>Tuo
zZqXw#e0c7X#yxF$_WHd!;lalo<!_iSm|icR%AlKZY;6HuRsQ?rH*1pEP9g!A2*DQy
z>gCag7~=m#dkxYYj%s_8K?358Bxqd@ohq6P1H@}#;t7d^p)I=jClF9Mu)0#Eo5$*v
z?;rP5j(s?xIOJAg5J#*#46Nj?$XVT$1HGcggp9ATc~*R_D$aLV%U=)xtFOXV<^>Y!
zqB_D<MqK_EPCp;Kn0LJ==4hQ^)arPfT;1$-?=|5YJp2AA9ag+=S?IDPBQETbg&Fq`
zfm5y$a~=BTl)9^F%XxEc_}c6h%Cp|SzR>cX)?8xN9sTNDs5q&hp<8033R&Z*r9|MX
zI_U3^RLxW3(T*=_zeVX-SOO=Z_6}_zbf#kN`il6*lyvK22L81K*nwOPk}RL_nzVAK
z_B=Y!{SOUJnsnH4FAdB~bH*q`a<}tpZNUX2iG}|!yY6*`qFt)!sZg325gT<dl8;O|
F_dmUcqF?|3

diff --git a/docs/dev/articles/understanding_masking_and_padding.html b/docs/dev/articles/understanding_masking_and_padding.html
index e78f48338..e4a399333 100644
--- a/docs/dev/articles/understanding_masking_and_padding.html
+++ b/docs/dev/articles/understanding_masking_and_padding.html
@@ -31,7 +31,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -493,7 +493,7 @@ <h2 id="summary">Summary<a class="anchor" aria-label="anchor" href="#summary"></
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer>
diff --git a/docs/dev/articles/writing_a_custom_training_loop_in_tensorflow.html b/docs/dev/articles/writing_a_custom_training_loop_in_tensorflow.html
index 43d948655..888349828 100644
--- a/docs/dev/articles/writing_a_custom_training_loop_in_tensorflow.html
+++ b/docs/dev/articles/writing_a_custom_training_loop_in_tensorflow.html
@@ -31,7 +31,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -229,21 +229,21 @@ <h2 id="using-the-gradienttape-a-first-end-to-end-example">Using the <code>Gradi
 <span><span class="co">## Seen so far: 12800 samples</span></span>
 <span><span class="co">## Training loss (for one batch) at step 400: 1.6450</span></span>
 <span><span class="co">## Seen so far: 25600 samples</span></span>
-<span><span class="co">## Training loss (for one batch) at step 600: 0.8552</span></span>
+<span><span class="co">## Training loss (for one batch) at step 600: 0.7421</span></span>
 <span><span class="co">## Seen so far: 38400 samples</span></span>
 <span><span class="co">## Start of epoch  2</span></span>
-<span><span class="co">## Training loss (for one batch) at step 200: 0.6606</span></span>
+<span><span class="co">## Training loss (for one batch) at step 200: 0.7254</span></span>
 <span><span class="co">## Seen so far: 12800 samples</span></span>
-<span><span class="co">## Training loss (for one batch) at step 400: 0.2213</span></span>
+<span><span class="co">## Training loss (for one batch) at step 400: 0.2476</span></span>
 <span><span class="co">## Seen so far: 25600 samples</span></span>
-<span><span class="co">## Training loss (for one batch) at step 600: 0.3709</span></span>
+<span><span class="co">## Training loss (for one batch) at step 600: 0.3578</span></span>
 <span><span class="co">## Seen so far: 38400 samples</span></span>
 <span><span class="co">## Start of epoch  3</span></span>
-<span><span class="co">## Training loss (for one batch) at step 200: 0.2968</span></span>
+<span><span class="co">## Training loss (for one batch) at step 200: 0.2974</span></span>
 <span><span class="co">## Seen so far: 12800 samples</span></span>
-<span><span class="co">## Training loss (for one batch) at step 400: 0.4781</span></span>
+<span><span class="co">## Training loss (for one batch) at step 400: 0.4441</span></span>
 <span><span class="co">## Seen so far: 25600 samples</span></span>
-<span><span class="co">## Training loss (for one batch) at step 600: 0.2455</span></span>
+<span><span class="co">## Training loss (for one batch) at step 600: 0.1461</span></span>
 <span><span class="co">## Seen so far: 38400 samples</span></span></code></pre>
 </div>
 <div class="section level2">
@@ -341,7 +341,7 @@ <h2 id="low-level-handling-of-metrics">Low-level handling of metrics<a class="an
 <span><span class="co">## Validation acc: 0.9049</span></span></code></pre>
 <div class="sourceCode" id="cb9"><pre class="downlit sourceCode r">
 <code class="sourceCode R"><span><span class="fu"><a href="https://rdrr.io/r/base/Sys.time.html" class="external-link">Sys.time</a></span><span class="op">(</span><span class="op">)</span> <span class="op">-</span> <span class="va">time</span></span></code></pre></div>
-<pre><code><span><span class="co">## Time difference of 51.62403 secs</span></span></code></pre>
+<pre><code><span><span class="co">## Time difference of 39.32149 secs</span></span></code></pre>
 </div>
 <div class="section level2">
 <h2 id="speeding-up-your-training-step-with-tf_function">Speeding-up your training step with <code>tf_function()</code><a class="anchor" aria-label="anchor" href="#speeding-up-your-training-step-with-tf_function"></a>
@@ -421,7 +421,7 @@ <h2 id="speeding-up-your-training-step-with-tf_function">Speeding-up your traini
 <span><span class="co">## Validation acc: 0.9049</span></span></code></pre>
 <div class="sourceCode" id="cb15"><pre class="downlit sourceCode r">
 <code class="sourceCode R"><span><span class="fu"><a href="https://rdrr.io/r/base/Sys.time.html" class="external-link">Sys.time</a></span><span class="op">(</span><span class="op">)</span> <span class="op">-</span> <span class="va">time</span></span></code></pre></div>
-<pre><code><span><span class="co">## Time difference of 0.5661066 secs</span></span></code></pre>
+<pre><code><span><span class="co">## Time difference of 0.4121809 secs</span></span></code></pre>
 <p>Much faster, isn’t it?</p>
 </div>
 <div class="section level2">
@@ -710,7 +710,7 @@ <h2 id="end-to-end-example-a-gan-training-loop-from-scratch">End-to-end example:
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer>
diff --git a/docs/dev/articles/writing_your_own_callbacks.html b/docs/dev/articles/writing_your_own_callbacks.html
index 43a826d70..904b68d62 100644
--- a/docs/dev/articles/writing_your_own_callbacks.html
+++ b/docs/dev/articles/writing_your_own_callbacks.html
@@ -31,7 +31,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -1088,7 +1088,7 @@ <h3 id="built-in-keras-callbacks">Built-in Keras callbacks<a class="anchor" aria
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer>
diff --git a/docs/dev/authors.html b/docs/dev/authors.html
index aea853f5e..74fffeb86 100644
--- a/docs/dev/authors.html
+++ b/docs/dev/authors.html
@@ -8,7 +8,7 @@
 
     <a class="navbar-brand me-2" href="index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -102,13 +102,13 @@ <h2 id="citation">Citation</h2>
 
       <p>Kalinowski T, Allaire J, Chollet F (2025).
 <em>keras3: R Interface to 'Keras'</em>.
-R package version 1.3.0.9001, https://github.com/rstudio/keras3, <a href="https://keras3.posit.co/">https://keras3.posit.co/</a>.
+R package version 1.4.0.9000, <a href="https://keras3.posit.co/">https://keras3.posit.co/</a>.
 </p>
       <pre>@Manual{,
   title = {keras3: R Interface to 'Keras'},
   author = {Tomasz Kalinowski and JJ Allaire and François Chollet},
   year = {2025},
-  note = {R package version 1.3.0.9001, https://github.com/rstudio/keras3},
+  note = {R package version 1.4.0.9000},
   url = {https://keras3.posit.co/},
 }</pre>
     </div>
@@ -122,7 +122,7 @@ <h2 id="citation">Citation</h2>
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/index.html b/docs/dev/index.html
index ed877943c..4a1a2ff1e 100644
--- a/docs/dev/index.html
+++ b/docs/dev/index.html
@@ -31,7 +31,7 @@
 
     <a class="navbar-brand me-2" href="index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -139,7 +139,7 @@ <h2 data-toc-skip>Developers</h2>
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer>
diff --git a/docs/dev/news/index.html b/docs/dev/news/index.html
index ab2bcf17e..ab5c2c40c 100644
--- a/docs/dev/news/index.html
+++ b/docs/dev/news/index.html
@@ -8,7 +8,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -54,6 +54,27 @@ <h1>Changelog</h1>
 
     <div class="section level2">
 <h2 class="pkg-version" data-toc-text="development version" id="keras3-development-version">keras3 (development version)<a class="anchor" aria-label="anchor" href="#keras3-development-version"></a></h2>
+<ul><li><p>Added S3 methods for JAX array: <code>str</code>, <code>as.array</code>, <code>as.double</code>, <code>as.integer</code>, <code>as.numeric</code>.</p></li>
+<li><p>Added <code>str</code> S3 method for Keras Variables.</p></li>
+<li><p><code><a href="../reference/layer_reshape.html">layer_reshape()</a></code> can now accept <code>-1</code> as a sentinel for an automatically calculated axis size.</p></li>
+<li><p>Updated dependencies declared by <code>use_backend("jax", gpu=TRUE)</code> for compatability with <code>keras-hub</code>.</p></li>
+<li><p>Added training loop configuration helpers: <code><a href="../reference/config_max_epochs.html">config_max_epochs()</a></code>, <code><a href="../reference/config_max_epochs.html">config_set_max_epochs()</a></code>, <code><a href="../reference/config_max_epochs.html">config_max_steps_per_epoch()</a></code>, and <code><a href="../reference/config_max_epochs.html">config_set_max_steps_per_epoch()</a></code>. The caps can also be set via the <code>KERAS_MAX_EPOCHS</code> and <code>KERAS_MAX_STEPS_PER_EPOCH</code> environment variables. Added <code><a href="../reference/config_is_nnx_enabled.html">config_is_nnx_enabled()</a></code> to check whether JAX NNX features are enabled.</p></li>
+<li><p>LoRA-enabled layers (<code><a href="../reference/layer_dense.html">layer_dense()</a></code>, <code><a href="../reference/layer_embedding.html">layer_embedding()</a></code>, <code><a href="../reference/layer_einsum_dense.html">layer_einsum_dense()</a></code>) gain a <code>lora_alpha</code> argument to scale the adaptation delta independently of the chosen rank.</p></li>
+<li><p><code><a href="../reference/keras_variable.html">keras_variable()</a></code> now accepts a <code>synchronization</code> argument for distributed strategies.</p></li>
+<li><p><code>Layer$add_weight()</code> gains an <code>overwrite_with_gradient</code> option and layers now provide a <code>symbolic_call()</code> method.</p></li>
+<li><p>Transposed convolution utilities now follow the latest Keras API: <code><a href="../reference/op_conv_transpose.html">op_conv_transpose()</a></code> defaults <code>strides = 1</code> and the <code>layer_conv_*_transpose()</code> layers expose <code>output_padding</code> for precise shape control.</p></li>
+<li><p><code><a href="../reference/layer_torch_module_wrapper.html">layer_torch_module_wrapper()</a></code> gains an <code>output_shape</code> argument to help Keras infer shapes when wrapping PyTorch modules.</p></li>
+<li><p><code><a href="../reference/save_model_weights.html">save_model_weights()</a></code> adds a <code>max_shard_size</code> argument to split large weight files into manageable shards.</p></li>
+<li><p>Added elastic deformation utilities for images: <code><a href="../reference/layer_random_elastic_transform.html">layer_random_elastic_transform()</a></code> and the lower-level <code><a href="../reference/op_image_elastic_transform.html">op_image_elastic_transform()</a></code>.</p></li>
+<li><p>Added <code><a href="../reference/loss_categorical_generalized_cross_entropy.html">loss_categorical_generalized_cross_entropy()</a></code> for training with noisy labels.</p></li>
+<li><p>Added the Muon optimizer via <code><a href="../reference/optimizer_muon.html">optimizer_muon()</a></code>.</p></li>
+<li><p>Added complex-valued helpers: S3 <code><a href="https://rdrr.io/r/base/complex.html" class="external-link">Arg()</a></code> methods for tensors, <code><a href="../reference/op_angle.html">op_angle()</a></code>, and conversions <code><a href="../reference/op_view_as_real.html">op_view_as_real()</a></code> / <code><a href="../reference/op_view_as_complex.html">op_view_as_complex()</a></code>.</p></li>
+<li><p>Added signal window operations: <code><a href="../reference/op_bartlett.html">op_bartlett()</a></code>, <code><a href="../reference/op_blackman.html">op_blackman()</a></code>, <code><a href="../reference/op_hamming.html">op_hamming()</a></code>, <code><a href="../reference/op_hanning.html">op_hanning()</a></code>, and <code><a href="../reference/op_kaiser.html">op_kaiser()</a></code>.</p></li>
+<li><p>Expanded numeric operations with <code><a href="../reference/op_layer_normalization.html">op_layer_normalization()</a></code>, <code><a href="../reference/op_cbrt.html">op_cbrt()</a></code>, <code><a href="../reference/op_corrcoef.html">op_corrcoef()</a></code>, <code><a href="../reference/op_deg2rad.html">op_deg2rad()</a></code>, <code><a href="../reference/op_heaviside.html">op_heaviside()</a></code>, the new <code><a href="../reference/op_sparse_sigmoid.html">op_sparse_sigmoid()</a></code> plus matching <code><a href="../reference/activation_sparse_sigmoid.html">activation_sparse_sigmoid()</a></code>, and an <code>attn_logits_soft_cap</code> argument for <code><a href="../reference/op_dot_product_attention.html">op_dot_product_attention()</a></code>.</p></li>
+<li><p><code><a href="../reference/layer_layer_normalization.html">layer_layer_normalization()</a></code> removes the <code>rms_scaling</code> argument.</p></li>
+</ul></div>
+    <div class="section level2">
+<h2 class="pkg-version" data-toc-text="1.4.0" id="keras3-140">keras3 1.4.0<a class="anchor" aria-label="anchor" href="#keras3-140"></a></h2><p class="text-muted">CRAN release: 2025-05-04</p>
 <ul><li><p>New <code><a href="../reference/op_subset.html">op_subset()</a></code> and <code>x@r[...]</code> methods enable tensor subsetting using R’s <code>[</code> semantics and idioms.</p></li>
 <li><p>New subset assignment methods implemented for tensors: <code>op_subset(x, ...) &lt;- value</code> and <code>x@r[...] &lt;- value</code></p></li>
 <li><p>Breaking changes: All operations prefixed with <code>op_</code> now return 1-based indices by default. The following functions that return or consume indices have changed: <code><a href="../reference/op_argmax.html">op_argmax()</a></code>, <code><a href="../reference/op_argmin.html">op_argmin()</a></code>, <code><a href="../reference/op_top_k.html">op_top_k()</a></code>, <code><a href="../reference/op_argpartition.html">op_argpartition()</a></code>, <code><a href="../reference/op_searchsorted.html">op_searchsorted()</a></code>, <code><a href="../reference/op_argsort.html">op_argsort()</a></code>, <code><a href="../reference/op_digitize.html">op_digitize()</a></code>, <code><a href="../reference/op_nonzero.html">op_nonzero()</a></code>, <code><a href="../reference/op_split.html">op_split()</a></code>, <code><a href="../reference/op_trace.html">op_trace()</a></code>, <code><a href="../reference/op_swapaxes.html">op_swapaxes()</a></code>, <code><a href="../reference/op_ctc_decode.html">op_ctc_decode()</a></code>, <code><a href="../reference/op_ctc_loss.html">op_ctc_loss()</a></code>, <code><a href="../reference/op_one_hot.html">op_one_hot()</a></code>, <code><a href="../reference/op_arange.html">op_arange()</a></code></p></li>
@@ -421,7 +442,7 @@ <h2 class="pkg-version" data-toc-text="0.1.0" id="keras3-010">keras3 0.1.0<a cla
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/pkgdown.yml b/docs/dev/pkgdown.yml
index 2b9292eb7..8643bbbd9 100644
--- a/docs/dev/pkgdown.yml
+++ b/docs/dev/pkgdown.yml
@@ -1,5 +1,5 @@
 pandoc: 3.1.11
-pkgdown: 2.1.2
+pkgdown: 2.1.3
 pkgdown_sha: ~
 articles:
   examples/vision/autoencoder: examples/vision/autoencoder.html
@@ -28,7 +28,7 @@ articles:
   understanding_masking_and_padding: understanding_masking_and_padding.html
   writing_a_custom_training_loop_in_tensorflow: writing_a_custom_training_loop_in_tensorflow.html
   writing_your_own_callbacks: writing_your_own_callbacks.html
-last_built: 2025-05-02T17:19Z
+last_built: 2025-09-27T14:52Z
 urls:
   reference: https://keras3.posit.co/reference
   article: https://keras3.posit.co/articles
diff --git a/docs/dev/reference/Callback.html b/docs/dev/reference/Callback.html
index 16ce07c7c..69ed8a220 100644
--- a/docs/dev/reference/Callback.html
+++ b/docs/dev/reference/Callback.html
@@ -16,7 +16,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -358,7 +358,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/Constraint.html b/docs/dev/reference/Constraint.html
index 95d0e2272..889afbe7b 100644
--- a/docs/dev/reference/Constraint.html
+++ b/docs/dev/reference/Constraint.html
@@ -58,7 +58,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -220,7 +220,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/Layer.html b/docs/dev/reference/Layer.html
index 31cbd6274..416e2f964 100644
--- a/docs/dev/reference/Layer.html
+++ b/docs/dev/reference/Layer.html
@@ -36,7 +36,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -351,6 +351,7 @@ <h2 id="methods-available">Methods available<a class="anchor" aria-label="anchor
 <span>           regularizer <span class="op">=</span> <span class="cn">NULL</span>,</span>
 <span>           constraint <span class="op">=</span> <span class="cn">NULL</span>,</span>
 <span>           aggregation <span class="op">=</span> <span class="st">'none'</span>,</span>
+<span>           overwrite_with_gradient <span class="op">=</span> <span class="cn">FALSE</span>,</span>
 <span>           name <span class="op">=</span> <span class="cn">NULL</span><span class="op">)</span></span></code></pre><p></p></div>
 <p>Add a weight variable to the layer.</p>
 <p>Args:</p><ul><li><p><code>shape</code>: shape for the variable (as defined by <code><a href="shape.html">shape()</a></code>)
@@ -386,6 +387,8 @@ <h2 id="methods-available">Methods available<a class="anchor" aria-label="anchor
 the type of multi-replica aggregation to be used for this
 variable when writing custom data parallel training loops.
 Defaults to <code>"none"</code>.</p></li>
+<li><p><code>overwrite_with_gradient</code>: Boolean, whether to overwrite the variable with
+the computed gradient. Useful for float8 training. Defaults to <code>FALSE</code>.</p></li>
 <li><p><code>name</code>: String name of the variable. Useful for debugging purposes.</p></li>
 </ul><p>Returns:</p>
 <p>A backend tensor, wrapped in a <code>KerasVariable</code> class.
@@ -421,6 +424,8 @@ <h2 id="methods-available">Methods available<a class="anchor" aria-label="anchor
 <li><p></p>
 <p></p><div class="sourceCode r"><pre><code><span><span class="fu">compute_output_spec</span><span class="op">(</span><span class="va">...</span><span class="op">)</span></span></code></pre><p></p></div></li>
 <li><p></p>
+<p></p><div class="sourceCode r"><pre><code><span><span class="fu">symbolic_call</span><span class="op">(</span><span class="va">...</span><span class="op">)</span></span></code></pre><p></p></div></li>
+<li><p></p>
 <p></p><div class="sourceCode r"><pre><code><span><span class="fu"><a href="../reference/count_params.html">count_params</a></span><span class="op">(</span><span class="op">)</span></span></code></pre><p></p></div>
 <p>Count the total number of scalars composing the weights.</p>
 <p>Returns:
@@ -613,7 +618,7 @@ <h2 id="data-descriptors-attributes-">Data descriptors (Attributes):<a class="an
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
     <div class="dont-index">
 <ul><li><p><a href="https://keras.io/api/layers/base_layer#layer-class" class="external-link">https://keras.io/api/layers/base_layer#layer-class</a></p></li>
-</ul><p>Other layers: <br><code><a href="layer_activation.html">layer_activation</a>()</code> <br><code><a href="layer_activation_elu.html">layer_activation_elu</a>()</code> <br><code><a href="layer_activation_leaky_relu.html">layer_activation_leaky_relu</a>()</code> <br><code><a href="layer_activation_parametric_relu.html">layer_activation_parametric_relu</a>()</code> <br><code><a href="layer_activation_relu.html">layer_activation_relu</a>()</code> <br><code><a href="layer_activation_softmax.html">layer_activation_softmax</a>()</code> <br><code><a href="layer_activity_regularization.html">layer_activity_regularization</a>()</code> <br><code><a href="layer_add.html">layer_add</a>()</code> <br><code><a href="layer_additive_attention.html">layer_additive_attention</a>()</code> <br><code><a href="layer_alpha_dropout.html">layer_alpha_dropout</a>()</code> <br><code><a href="layer_attention.html">layer_attention</a>()</code> <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_average.html">layer_average</a>()</code> <br><code><a href="layer_average_pooling_1d.html">layer_average_pooling_1d</a>()</code> <br><code><a href="layer_average_pooling_2d.html">layer_average_pooling_2d</a>()</code> <br><code><a href="layer_average_pooling_3d.html">layer_average_pooling_3d</a>()</code> <br><code><a href="layer_batch_normalization.html">layer_batch_normalization</a>()</code> <br><code><a href="layer_bidirectional.html">layer_bidirectional</a>()</code> <br><code><a href="layer_category_encoding.html">layer_category_encoding</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_concatenate.html">layer_concatenate</a>()</code> <br><code><a href="layer_conv_1d.html">layer_conv_1d</a>()</code> <br><code><a href="layer_conv_1d_transpose.html">layer_conv_1d_transpose</a>()</code> <br><code><a href="layer_conv_2d.html">layer_conv_2d</a>()</code> <br><code><a href="layer_conv_2d_transpose.html">layer_conv_2d_transpose</a>()</code> <br><code><a href="layer_conv_3d.html">layer_conv_3d</a>()</code> <br><code><a href="layer_conv_3d_transpose.html">layer_conv_3d_transpose</a>()</code> <br><code><a href="layer_conv_lstm_1d.html">layer_conv_lstm_1d</a>()</code> <br><code><a href="layer_conv_lstm_2d.html">layer_conv_lstm_2d</a>()</code> <br><code><a href="layer_conv_lstm_3d.html">layer_conv_lstm_3d</a>()</code> <br><code><a href="layer_cropping_1d.html">layer_cropping_1d</a>()</code> <br><code><a href="layer_cropping_2d.html">layer_cropping_2d</a>()</code> <br><code><a href="layer_cropping_3d.html">layer_cropping_3d</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_dense.html">layer_dense</a>()</code> <br><code><a href="layer_depthwise_conv_1d.html">layer_depthwise_conv_1d</a>()</code> <br><code><a href="layer_depthwise_conv_2d.html">layer_depthwise_conv_2d</a>()</code> <br><code><a href="layer_discretization.html">layer_discretization</a>()</code> <br><code><a href="layer_dot.html">layer_dot</a>()</code> <br><code><a href="layer_dropout.html">layer_dropout</a>()</code> <br><code><a href="layer_einsum_dense.html">layer_einsum_dense</a>()</code> <br><code><a href="layer_embedding.html">layer_embedding</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_feature_space.html">layer_feature_space</a>()</code> <br><code><a href="layer_flatten.html">layer_flatten</a>()</code> <br><code><a href="layer_flax_module_wrapper.html">layer_flax_module_wrapper</a>()</code> <br><code><a href="layer_gaussian_dropout.html">layer_gaussian_dropout</a>()</code> <br><code><a href="layer_gaussian_noise.html">layer_gaussian_noise</a>()</code> <br><code><a href="layer_global_average_pooling_1d.html">layer_global_average_pooling_1d</a>()</code> <br><code><a href="layer_global_average_pooling_2d.html">layer_global_average_pooling_2d</a>()</code> <br><code><a href="layer_global_average_pooling_3d.html">layer_global_average_pooling_3d</a>()</code> <br><code><a href="layer_global_max_pooling_1d.html">layer_global_max_pooling_1d</a>()</code> <br><code><a href="layer_global_max_pooling_2d.html">layer_global_max_pooling_2d</a>()</code> <br><code><a href="layer_global_max_pooling_3d.html">layer_global_max_pooling_3d</a>()</code> <br><code><a href="layer_group_normalization.html">layer_group_normalization</a>()</code> <br><code><a href="layer_group_query_attention.html">layer_group_query_attention</a>()</code> <br><code><a href="layer_gru.html">layer_gru</a>()</code> <br><code><a href="layer_hashed_crossing.html">layer_hashed_crossing</a>()</code> <br><code><a href="layer_hashing.html">layer_hashing</a>()</code> <br><code><a href="layer_identity.html">layer_identity</a>()</code> <br><code><a href="layer_integer_lookup.html">layer_integer_lookup</a>()</code> <br><code><a href="layer_jax_model_wrapper.html">layer_jax_model_wrapper</a>()</code> <br><code><a href="layer_lambda.html">layer_lambda</a>()</code> <br><code><a href="layer_layer_normalization.html">layer_layer_normalization</a>()</code> <br><code><a href="layer_lstm.html">layer_lstm</a>()</code> <br><code><a href="layer_masking.html">layer_masking</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_max_pooling_1d.html">layer_max_pooling_1d</a>()</code> <br><code><a href="layer_max_pooling_2d.html">layer_max_pooling_2d</a>()</code> <br><code><a href="layer_max_pooling_3d.html">layer_max_pooling_3d</a>()</code> <br><code><a href="layer_maximum.html">layer_maximum</a>()</code> <br><code><a href="layer_mel_spectrogram.html">layer_mel_spectrogram</a>()</code> <br><code><a href="layer_minimum.html">layer_minimum</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_multi_head_attention.html">layer_multi_head_attention</a>()</code> <br><code><a href="layer_multiply.html">layer_multiply</a>()</code> <br><code><a href="layer_normalization.html">layer_normalization</a>()</code> <br><code><a href="layer_permute.html">layer_permute</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_brightness.html">layer_random_brightness</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_contrast.html">layer_random_contrast</a>()</code> <br><code><a href="layer_random_crop.html">layer_random_crop</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_flip.html">layer_random_flip</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_rotation.html">layer_random_rotation</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_random_translation.html">layer_random_translation</a>()</code> <br><code><a href="layer_random_zoom.html">layer_random_zoom</a>()</code> <br><code><a href="layer_repeat_vector.html">layer_repeat_vector</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_reshape.html">layer_reshape</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_rms_normalization.html">layer_rms_normalization</a>()</code> <br><code><a href="layer_rnn.html">layer_rnn</a>()</code> <br><code><a href="layer_separable_conv_1d.html">layer_separable_conv_1d</a>()</code> <br><code><a href="layer_separable_conv_2d.html">layer_separable_conv_2d</a>()</code> <br><code><a href="layer_simple_rnn.html">layer_simple_rnn</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br><code><a href="layer_spatial_dropout_1d.html">layer_spatial_dropout_1d</a>()</code> <br><code><a href="layer_spatial_dropout_2d.html">layer_spatial_dropout_2d</a>()</code> <br><code><a href="layer_spatial_dropout_3d.html">layer_spatial_dropout_3d</a>()</code> <br><code><a href="layer_spectral_normalization.html">layer_spectral_normalization</a>()</code> <br><code><a href="layer_stft_spectrogram.html">layer_stft_spectrogram</a>()</code> <br><code><a href="layer_string_lookup.html">layer_string_lookup</a>()</code> <br><code><a href="layer_subtract.html">layer_subtract</a>()</code> <br><code><a href="layer_text_vectorization.html">layer_text_vectorization</a>()</code> <br><code><a href="layer_tfsm.html">layer_tfsm</a>()</code> <br><code><a href="layer_time_distributed.html">layer_time_distributed</a>()</code> <br><code><a href="layer_torch_module_wrapper.html">layer_torch_module_wrapper</a>()</code> <br><code><a href="layer_unit_normalization.html">layer_unit_normalization</a>()</code> <br><code><a href="layer_upsampling_1d.html">layer_upsampling_1d</a>()</code> <br><code><a href="layer_upsampling_2d.html">layer_upsampling_2d</a>()</code> <br><code><a href="layer_upsampling_3d.html">layer_upsampling_3d</a>()</code> <br><code><a href="layer_zero_padding_1d.html">layer_zero_padding_1d</a>()</code> <br><code><a href="layer_zero_padding_2d.html">layer_zero_padding_2d</a>()</code> <br><code><a href="layer_zero_padding_3d.html">layer_zero_padding_3d</a>()</code> <br><code><a href="rnn_cell_gru.html">rnn_cell_gru</a>()</code> <br><code><a href="rnn_cell_lstm.html">rnn_cell_lstm</a>()</code> <br><code><a href="rnn_cell_simple.html">rnn_cell_simple</a>()</code> <br><code><a href="rnn_cells_stack.html">rnn_cells_stack</a>()</code> <br></p></div>
+</ul><p>Other layers: <br><code><a href="layer_activation.html">layer_activation</a>()</code> <br><code><a href="layer_activation_elu.html">layer_activation_elu</a>()</code> <br><code><a href="layer_activation_leaky_relu.html">layer_activation_leaky_relu</a>()</code> <br><code><a href="layer_activation_parametric_relu.html">layer_activation_parametric_relu</a>()</code> <br><code><a href="layer_activation_relu.html">layer_activation_relu</a>()</code> <br><code><a href="layer_activation_softmax.html">layer_activation_softmax</a>()</code> <br><code><a href="layer_activity_regularization.html">layer_activity_regularization</a>()</code> <br><code><a href="layer_add.html">layer_add</a>()</code> <br><code><a href="layer_additive_attention.html">layer_additive_attention</a>()</code> <br><code><a href="layer_alpha_dropout.html">layer_alpha_dropout</a>()</code> <br><code><a href="layer_attention.html">layer_attention</a>()</code> <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_average.html">layer_average</a>()</code> <br><code><a href="layer_average_pooling_1d.html">layer_average_pooling_1d</a>()</code> <br><code><a href="layer_average_pooling_2d.html">layer_average_pooling_2d</a>()</code> <br><code><a href="layer_average_pooling_3d.html">layer_average_pooling_3d</a>()</code> <br><code><a href="layer_batch_normalization.html">layer_batch_normalization</a>()</code> <br><code><a href="layer_bidirectional.html">layer_bidirectional</a>()</code> <br><code><a href="layer_category_encoding.html">layer_category_encoding</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_concatenate.html">layer_concatenate</a>()</code> <br><code><a href="layer_conv_1d.html">layer_conv_1d</a>()</code> <br><code><a href="layer_conv_1d_transpose.html">layer_conv_1d_transpose</a>()</code> <br><code><a href="layer_conv_2d.html">layer_conv_2d</a>()</code> <br><code><a href="layer_conv_2d_transpose.html">layer_conv_2d_transpose</a>()</code> <br><code><a href="layer_conv_3d.html">layer_conv_3d</a>()</code> <br><code><a href="layer_conv_3d_transpose.html">layer_conv_3d_transpose</a>()</code> <br><code><a href="layer_conv_lstm_1d.html">layer_conv_lstm_1d</a>()</code> <br><code><a href="layer_conv_lstm_2d.html">layer_conv_lstm_2d</a>()</code> <br><code><a href="layer_conv_lstm_3d.html">layer_conv_lstm_3d</a>()</code> <br><code><a href="layer_cropping_1d.html">layer_cropping_1d</a>()</code> <br><code><a href="layer_cropping_2d.html">layer_cropping_2d</a>()</code> <br><code><a href="layer_cropping_3d.html">layer_cropping_3d</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_dense.html">layer_dense</a>()</code> <br><code><a href="layer_depthwise_conv_1d.html">layer_depthwise_conv_1d</a>()</code> <br><code><a href="layer_depthwise_conv_2d.html">layer_depthwise_conv_2d</a>()</code> <br><code><a href="layer_discretization.html">layer_discretization</a>()</code> <br><code><a href="layer_dot.html">layer_dot</a>()</code> <br><code><a href="layer_dropout.html">layer_dropout</a>()</code> <br><code><a href="layer_einsum_dense.html">layer_einsum_dense</a>()</code> <br><code><a href="layer_embedding.html">layer_embedding</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_feature_space.html">layer_feature_space</a>()</code> <br><code><a href="layer_flatten.html">layer_flatten</a>()</code> <br><code><a href="layer_flax_module_wrapper.html">layer_flax_module_wrapper</a>()</code> <br><code><a href="layer_gaussian_dropout.html">layer_gaussian_dropout</a>()</code> <br><code><a href="layer_gaussian_noise.html">layer_gaussian_noise</a>()</code> <br><code><a href="layer_global_average_pooling_1d.html">layer_global_average_pooling_1d</a>()</code> <br><code><a href="layer_global_average_pooling_2d.html">layer_global_average_pooling_2d</a>()</code> <br><code><a href="layer_global_average_pooling_3d.html">layer_global_average_pooling_3d</a>()</code> <br><code><a href="layer_global_max_pooling_1d.html">layer_global_max_pooling_1d</a>()</code> <br><code><a href="layer_global_max_pooling_2d.html">layer_global_max_pooling_2d</a>()</code> <br><code><a href="layer_global_max_pooling_3d.html">layer_global_max_pooling_3d</a>()</code> <br><code><a href="layer_group_normalization.html">layer_group_normalization</a>()</code> <br><code><a href="layer_group_query_attention.html">layer_group_query_attention</a>()</code> <br><code><a href="layer_gru.html">layer_gru</a>()</code> <br><code><a href="layer_hashed_crossing.html">layer_hashed_crossing</a>()</code> <br><code><a href="layer_hashing.html">layer_hashing</a>()</code> <br><code><a href="layer_identity.html">layer_identity</a>()</code> <br><code><a href="layer_integer_lookup.html">layer_integer_lookup</a>()</code> <br><code><a href="layer_jax_model_wrapper.html">layer_jax_model_wrapper</a>()</code> <br><code><a href="layer_lambda.html">layer_lambda</a>()</code> <br><code><a href="layer_layer_normalization.html">layer_layer_normalization</a>()</code> <br><code><a href="layer_lstm.html">layer_lstm</a>()</code> <br><code><a href="layer_masking.html">layer_masking</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_max_pooling_1d.html">layer_max_pooling_1d</a>()</code> <br><code><a href="layer_max_pooling_2d.html">layer_max_pooling_2d</a>()</code> <br><code><a href="layer_max_pooling_3d.html">layer_max_pooling_3d</a>()</code> <br><code><a href="layer_maximum.html">layer_maximum</a>()</code> <br><code><a href="layer_mel_spectrogram.html">layer_mel_spectrogram</a>()</code> <br><code><a href="layer_minimum.html">layer_minimum</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_multi_head_attention.html">layer_multi_head_attention</a>()</code> <br><code><a href="layer_multiply.html">layer_multiply</a>()</code> <br><code><a href="layer_normalization.html">layer_normalization</a>()</code> <br><code><a href="layer_permute.html">layer_permute</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_brightness.html">layer_random_brightness</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_contrast.html">layer_random_contrast</a>()</code> <br><code><a href="layer_random_crop.html">layer_random_crop</a>()</code> <br><code><a href="layer_random_elastic_transform.html">layer_random_elastic_transform</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_flip.html">layer_random_flip</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_rotation.html">layer_random_rotation</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_random_translation.html">layer_random_translation</a>()</code> <br><code><a href="layer_random_zoom.html">layer_random_zoom</a>()</code> <br><code><a href="layer_repeat_vector.html">layer_repeat_vector</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_reshape.html">layer_reshape</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_rms_normalization.html">layer_rms_normalization</a>()</code> <br><code><a href="layer_rnn.html">layer_rnn</a>()</code> <br><code><a href="layer_separable_conv_1d.html">layer_separable_conv_1d</a>()</code> <br><code><a href="layer_separable_conv_2d.html">layer_separable_conv_2d</a>()</code> <br><code><a href="layer_simple_rnn.html">layer_simple_rnn</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br><code><a href="layer_spatial_dropout_1d.html">layer_spatial_dropout_1d</a>()</code> <br><code><a href="layer_spatial_dropout_2d.html">layer_spatial_dropout_2d</a>()</code> <br><code><a href="layer_spatial_dropout_3d.html">layer_spatial_dropout_3d</a>()</code> <br><code><a href="layer_spectral_normalization.html">layer_spectral_normalization</a>()</code> <br><code><a href="layer_stft_spectrogram.html">layer_stft_spectrogram</a>()</code> <br><code><a href="layer_string_lookup.html">layer_string_lookup</a>()</code> <br><code><a href="layer_subtract.html">layer_subtract</a>()</code> <br><code><a href="layer_text_vectorization.html">layer_text_vectorization</a>()</code> <br><code><a href="layer_tfsm.html">layer_tfsm</a>()</code> <br><code><a href="layer_time_distributed.html">layer_time_distributed</a>()</code> <br><code><a href="layer_torch_module_wrapper.html">layer_torch_module_wrapper</a>()</code> <br><code><a href="layer_unit_normalization.html">layer_unit_normalization</a>()</code> <br><code><a href="layer_upsampling_1d.html">layer_upsampling_1d</a>()</code> <br><code><a href="layer_upsampling_2d.html">layer_upsampling_2d</a>()</code> <br><code><a href="layer_upsampling_3d.html">layer_upsampling_3d</a>()</code> <br><code><a href="layer_zero_padding_1d.html">layer_zero_padding_1d</a>()</code> <br><code><a href="layer_zero_padding_2d.html">layer_zero_padding_2d</a>()</code> <br><code><a href="layer_zero_padding_3d.html">layer_zero_padding_3d</a>()</code> <br><code><a href="rnn_cell_gru.html">rnn_cell_gru</a>()</code> <br><code><a href="rnn_cell_lstm.html">rnn_cell_lstm</a>()</code> <br><code><a href="rnn_cell_simple.html">rnn_cell_simple</a>()</code> <br><code><a href="rnn_cells_stack.html">rnn_cells_stack</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -625,7 +630,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/LearningRateSchedule.html b/docs/dev/reference/LearningRateSchedule.html
index 74c504fbd..1ea5a1f6a 100644
--- a/docs/dev/reference/LearningRateSchedule.html
+++ b/docs/dev/reference/LearningRateSchedule.html
@@ -50,7 +50,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -240,7 +240,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/Loss.html b/docs/dev/reference/Loss.html
index 1612e9f97..3cad4e8d2 100644
--- a/docs/dev/reference/Loss.html
+++ b/docs/dev/reference/Loss.html
@@ -14,7 +14,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -204,7 +204,7 @@ <h2 id="symbols-in-scope">Symbols in scope<a class="anchor" aria-label="anchor"
 </ul></div>
     <div class="section level2">
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
-    <div class="dont-index"><p>Other losses: <br><code><a href="loss_binary_crossentropy.html">loss_binary_crossentropy</a>()</code> <br><code><a href="loss_binary_focal_crossentropy.html">loss_binary_focal_crossentropy</a>()</code> <br><code><a href="loss_categorical_crossentropy.html">loss_categorical_crossentropy</a>()</code> <br><code><a href="loss_categorical_focal_crossentropy.html">loss_categorical_focal_crossentropy</a>()</code> <br><code><a href="loss_categorical_hinge.html">loss_categorical_hinge</a>()</code> <br><code><a href="loss_circle.html">loss_circle</a>()</code> <br><code><a href="loss_cosine_similarity.html">loss_cosine_similarity</a>()</code> <br><code><a href="loss_ctc.html">loss_ctc</a>()</code> <br><code><a href="loss_dice.html">loss_dice</a>()</code> <br><code><a href="loss_hinge.html">loss_hinge</a>()</code> <br><code><a href="loss_huber.html">loss_huber</a>()</code> <br><code><a href="loss_kl_divergence.html">loss_kl_divergence</a>()</code> <br><code><a href="loss_log_cosh.html">loss_log_cosh</a>()</code> <br><code><a href="loss_mean_absolute_error.html">loss_mean_absolute_error</a>()</code> <br><code><a href="loss_mean_absolute_percentage_error.html">loss_mean_absolute_percentage_error</a>()</code> <br><code><a href="loss_mean_squared_error.html">loss_mean_squared_error</a>()</code> <br><code><a href="loss_mean_squared_logarithmic_error.html">loss_mean_squared_logarithmic_error</a>()</code> <br><code><a href="loss_poisson.html">loss_poisson</a>()</code> <br><code><a href="loss_sparse_categorical_crossentropy.html">loss_sparse_categorical_crossentropy</a>()</code> <br><code><a href="loss_squared_hinge.html">loss_squared_hinge</a>()</code> <br><code><a href="loss_tversky.html">loss_tversky</a>()</code> <br><code><a href="metric_binary_crossentropy.html">metric_binary_crossentropy</a>()</code> <br><code><a href="metric_binary_focal_crossentropy.html">metric_binary_focal_crossentropy</a>()</code> <br><code><a href="metric_categorical_crossentropy.html">metric_categorical_crossentropy</a>()</code> <br><code><a href="metric_categorical_focal_crossentropy.html">metric_categorical_focal_crossentropy</a>()</code> <br><code><a href="metric_categorical_hinge.html">metric_categorical_hinge</a>()</code> <br><code><a href="metric_hinge.html">metric_hinge</a>()</code> <br><code><a href="metric_huber.html">metric_huber</a>()</code> <br><code><a href="metric_kl_divergence.html">metric_kl_divergence</a>()</code> <br><code><a href="metric_log_cosh.html">metric_log_cosh</a>()</code> <br><code><a href="metric_mean_absolute_error.html">metric_mean_absolute_error</a>()</code> <br><code><a href="metric_mean_absolute_percentage_error.html">metric_mean_absolute_percentage_error</a>()</code> <br><code><a href="metric_mean_squared_error.html">metric_mean_squared_error</a>()</code> <br><code><a href="metric_mean_squared_logarithmic_error.html">metric_mean_squared_logarithmic_error</a>()</code> <br><code><a href="metric_poisson.html">metric_poisson</a>()</code> <br><code><a href="metric_sparse_categorical_crossentropy.html">metric_sparse_categorical_crossentropy</a>()</code> <br><code><a href="metric_squared_hinge.html">metric_squared_hinge</a>()</code> <br></p></div>
+    <div class="dont-index"><p>Other losses: <br><code><a href="loss_binary_crossentropy.html">loss_binary_crossentropy</a>()</code> <br><code><a href="loss_binary_focal_crossentropy.html">loss_binary_focal_crossentropy</a>()</code> <br><code><a href="loss_categorical_crossentropy.html">loss_categorical_crossentropy</a>()</code> <br><code><a href="loss_categorical_focal_crossentropy.html">loss_categorical_focal_crossentropy</a>()</code> <br><code><a href="loss_categorical_generalized_cross_entropy.html">loss_categorical_generalized_cross_entropy</a>()</code> <br><code><a href="loss_categorical_hinge.html">loss_categorical_hinge</a>()</code> <br><code><a href="loss_circle.html">loss_circle</a>()</code> <br><code><a href="loss_cosine_similarity.html">loss_cosine_similarity</a>()</code> <br><code><a href="loss_ctc.html">loss_ctc</a>()</code> <br><code><a href="loss_dice.html">loss_dice</a>()</code> <br><code><a href="loss_hinge.html">loss_hinge</a>()</code> <br><code><a href="loss_huber.html">loss_huber</a>()</code> <br><code><a href="loss_kl_divergence.html">loss_kl_divergence</a>()</code> <br><code><a href="loss_log_cosh.html">loss_log_cosh</a>()</code> <br><code><a href="loss_mean_absolute_error.html">loss_mean_absolute_error</a>()</code> <br><code><a href="loss_mean_absolute_percentage_error.html">loss_mean_absolute_percentage_error</a>()</code> <br><code><a href="loss_mean_squared_error.html">loss_mean_squared_error</a>()</code> <br><code><a href="loss_mean_squared_logarithmic_error.html">loss_mean_squared_logarithmic_error</a>()</code> <br><code><a href="loss_poisson.html">loss_poisson</a>()</code> <br><code><a href="loss_sparse_categorical_crossentropy.html">loss_sparse_categorical_crossentropy</a>()</code> <br><code><a href="loss_squared_hinge.html">loss_squared_hinge</a>()</code> <br><code><a href="loss_tversky.html">loss_tversky</a>()</code> <br><code><a href="metric_binary_crossentropy.html">metric_binary_crossentropy</a>()</code> <br><code><a href="metric_binary_focal_crossentropy.html">metric_binary_focal_crossentropy</a>()</code> <br><code><a href="metric_categorical_crossentropy.html">metric_categorical_crossentropy</a>()</code> <br><code><a href="metric_categorical_focal_crossentropy.html">metric_categorical_focal_crossentropy</a>()</code> <br><code><a href="metric_categorical_hinge.html">metric_categorical_hinge</a>()</code> <br><code><a href="metric_hinge.html">metric_hinge</a>()</code> <br><code><a href="metric_huber.html">metric_huber</a>()</code> <br><code><a href="metric_kl_divergence.html">metric_kl_divergence</a>()</code> <br><code><a href="metric_log_cosh.html">metric_log_cosh</a>()</code> <br><code><a href="metric_mean_absolute_error.html">metric_mean_absolute_error</a>()</code> <br><code><a href="metric_mean_absolute_percentage_error.html">metric_mean_absolute_percentage_error</a>()</code> <br><code><a href="metric_mean_squared_error.html">metric_mean_squared_error</a>()</code> <br><code><a href="metric_mean_squared_logarithmic_error.html">metric_mean_squared_logarithmic_error</a>()</code> <br><code><a href="metric_poisson.html">metric_poisson</a>()</code> <br><code><a href="metric_sparse_categorical_crossentropy.html">metric_sparse_categorical_crossentropy</a>()</code> <br><code><a href="metric_squared_hinge.html">metric_squared_hinge</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -216,7 +216,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/Metric.html b/docs/dev/reference/Metric.html
index 3b6772872..afcf35c8c 100644
--- a/docs/dev/reference/Metric.html
+++ b/docs/dev/reference/Metric.html
@@ -14,7 +14,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -289,7 +289,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/Model.html b/docs/dev/reference/Model.html
index cff35a9dd..5a7fc3383 100644
--- a/docs/dev/reference/Model.html
+++ b/docs/dev/reference/Model.html
@@ -20,7 +20,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -156,7 +156,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/activation_celu.html b/docs/dev/reference/activation_celu.html
index e95e081f6..fac6dd01a 100644
--- a/docs/dev/reference/activation_celu.html
+++ b/docs/dev/reference/activation_celu.html
@@ -12,7 +12,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -91,7 +91,7 @@ <h2 id="reference">Reference<a class="anchor" aria-label="anchor" href="#referen
 </ul></div>
     <div class="section level2">
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
-    <div class="dont-index"><p>Other activations: <br><code><a href="activation_elu.html">activation_elu</a>()</code> <br><code><a href="activation_exponential.html">activation_exponential</a>()</code> <br><code><a href="activation_gelu.html">activation_gelu</a>()</code> <br><code><a href="activation_glu.html">activation_glu</a>()</code> <br><code><a href="activation_hard_shrink.html">activation_hard_shrink</a>()</code> <br><code><a href="activation_hard_sigmoid.html">activation_hard_sigmoid</a>()</code> <br><code><a href="activation_hard_tanh.html">activation_hard_tanh</a>()</code> <br><code><a href="activation_leaky_relu.html">activation_leaky_relu</a>()</code> <br><code><a href="activation_linear.html">activation_linear</a>()</code> <br><code><a href="activation_log_sigmoid.html">activation_log_sigmoid</a>()</code> <br><code><a href="activation_log_softmax.html">activation_log_softmax</a>()</code> <br><code><a href="activation_mish.html">activation_mish</a>()</code> <br><code><a href="activation_relu.html">activation_relu</a>()</code> <br><code><a href="activation_relu6.html">activation_relu6</a>()</code> <br><code><a href="activation_selu.html">activation_selu</a>()</code> <br><code><a href="activation_sigmoid.html">activation_sigmoid</a>()</code> <br><code><a href="activation_silu.html">activation_silu</a>()</code> <br><code><a href="activation_soft_shrink.html">activation_soft_shrink</a>()</code> <br><code><a href="activation_softmax.html">activation_softmax</a>()</code> <br><code><a href="activation_softplus.html">activation_softplus</a>()</code> <br><code><a href="activation_softsign.html">activation_softsign</a>()</code> <br><code><a href="activation_sparse_plus.html">activation_sparse_plus</a>()</code> <br><code><a href="activation_sparsemax.html">activation_sparsemax</a>()</code> <br><code><a href="activation_squareplus.html">activation_squareplus</a>()</code> <br><code><a href="activation_tanh.html">activation_tanh</a>()</code> <br><code><a href="activation_tanh_shrink.html">activation_tanh_shrink</a>()</code> <br><code><a href="activation_threshold.html">activation_threshold</a>()</code> <br></p></div>
+    <div class="dont-index"><p>Other activations: <br><code><a href="activation_elu.html">activation_elu</a>()</code> <br><code><a href="activation_exponential.html">activation_exponential</a>()</code> <br><code><a href="activation_gelu.html">activation_gelu</a>()</code> <br><code><a href="activation_glu.html">activation_glu</a>()</code> <br><code><a href="activation_hard_shrink.html">activation_hard_shrink</a>()</code> <br><code><a href="activation_hard_sigmoid.html">activation_hard_sigmoid</a>()</code> <br><code><a href="activation_hard_tanh.html">activation_hard_tanh</a>()</code> <br><code><a href="activation_leaky_relu.html">activation_leaky_relu</a>()</code> <br><code><a href="activation_linear.html">activation_linear</a>()</code> <br><code><a href="activation_log_sigmoid.html">activation_log_sigmoid</a>()</code> <br><code><a href="activation_log_softmax.html">activation_log_softmax</a>()</code> <br><code><a href="activation_mish.html">activation_mish</a>()</code> <br><code><a href="activation_relu.html">activation_relu</a>()</code> <br><code><a href="activation_relu6.html">activation_relu6</a>()</code> <br><code><a href="activation_selu.html">activation_selu</a>()</code> <br><code><a href="activation_sigmoid.html">activation_sigmoid</a>()</code> <br><code><a href="activation_silu.html">activation_silu</a>()</code> <br><code><a href="activation_soft_shrink.html">activation_soft_shrink</a>()</code> <br><code><a href="activation_softmax.html">activation_softmax</a>()</code> <br><code><a href="activation_softplus.html">activation_softplus</a>()</code> <br><code><a href="activation_softsign.html">activation_softsign</a>()</code> <br><code><a href="activation_sparse_plus.html">activation_sparse_plus</a>()</code> <br><code><a href="activation_sparse_sigmoid.html">activation_sparse_sigmoid</a>()</code> <br><code><a href="activation_sparsemax.html">activation_sparsemax</a>()</code> <br><code><a href="activation_squareplus.html">activation_squareplus</a>()</code> <br><code><a href="activation_tanh.html">activation_tanh</a>()</code> <br><code><a href="activation_tanh_shrink.html">activation_tanh_shrink</a>()</code> <br><code><a href="activation_threshold.html">activation_threshold</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -103,7 +103,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/activation_elu.html b/docs/dev/reference/activation_elu.html
index 4b74d758f..aff647e5f 100644
--- a/docs/dev/reference/activation_elu.html
+++ b/docs/dev/reference/activation_elu.html
@@ -30,7 +30,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -101,7 +101,7 @@ <h2 id="arguments">Arguments<a class="anchor" aria-label="anchor" href="#argumen
 
 
 <dt id="arg-alpha">alpha<a class="anchor" aria-label="anchor" href="#arg-alpha"></a></dt>
-<dd><p>Numeric. See description for details.</p></dd>
+<dd><p>A scalar, slope of positive section. Defaults to <code>1.0</code>.</p></dd>
 
 </dl></div>
     <div class="section level2">
@@ -117,7 +117,7 @@ <h2 id="reference">Reference<a class="anchor" aria-label="anchor" href="#referen
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
     <div class="dont-index">
 <ul><li><p><a href="https://keras.io/api/layers/activations#elu-function" class="external-link">https://keras.io/api/layers/activations#elu-function</a></p></li>
-</ul><p>Other activations: <br><code><a href="activation_celu.html">activation_celu</a>()</code> <br><code><a href="activation_exponential.html">activation_exponential</a>()</code> <br><code><a href="activation_gelu.html">activation_gelu</a>()</code> <br><code><a href="activation_glu.html">activation_glu</a>()</code> <br><code><a href="activation_hard_shrink.html">activation_hard_shrink</a>()</code> <br><code><a href="activation_hard_sigmoid.html">activation_hard_sigmoid</a>()</code> <br><code><a href="activation_hard_tanh.html">activation_hard_tanh</a>()</code> <br><code><a href="activation_leaky_relu.html">activation_leaky_relu</a>()</code> <br><code><a href="activation_linear.html">activation_linear</a>()</code> <br><code><a href="activation_log_sigmoid.html">activation_log_sigmoid</a>()</code> <br><code><a href="activation_log_softmax.html">activation_log_softmax</a>()</code> <br><code><a href="activation_mish.html">activation_mish</a>()</code> <br><code><a href="activation_relu.html">activation_relu</a>()</code> <br><code><a href="activation_relu6.html">activation_relu6</a>()</code> <br><code><a href="activation_selu.html">activation_selu</a>()</code> <br><code><a href="activation_sigmoid.html">activation_sigmoid</a>()</code> <br><code><a href="activation_silu.html">activation_silu</a>()</code> <br><code><a href="activation_soft_shrink.html">activation_soft_shrink</a>()</code> <br><code><a href="activation_softmax.html">activation_softmax</a>()</code> <br><code><a href="activation_softplus.html">activation_softplus</a>()</code> <br><code><a href="activation_softsign.html">activation_softsign</a>()</code> <br><code><a href="activation_sparse_plus.html">activation_sparse_plus</a>()</code> <br><code><a href="activation_sparsemax.html">activation_sparsemax</a>()</code> <br><code><a href="activation_squareplus.html">activation_squareplus</a>()</code> <br><code><a href="activation_tanh.html">activation_tanh</a>()</code> <br><code><a href="activation_tanh_shrink.html">activation_tanh_shrink</a>()</code> <br><code><a href="activation_threshold.html">activation_threshold</a>()</code> <br></p></div>
+</ul><p>Other activations: <br><code><a href="activation_celu.html">activation_celu</a>()</code> <br><code><a href="activation_exponential.html">activation_exponential</a>()</code> <br><code><a href="activation_gelu.html">activation_gelu</a>()</code> <br><code><a href="activation_glu.html">activation_glu</a>()</code> <br><code><a href="activation_hard_shrink.html">activation_hard_shrink</a>()</code> <br><code><a href="activation_hard_sigmoid.html">activation_hard_sigmoid</a>()</code> <br><code><a href="activation_hard_tanh.html">activation_hard_tanh</a>()</code> <br><code><a href="activation_leaky_relu.html">activation_leaky_relu</a>()</code> <br><code><a href="activation_linear.html">activation_linear</a>()</code> <br><code><a href="activation_log_sigmoid.html">activation_log_sigmoid</a>()</code> <br><code><a href="activation_log_softmax.html">activation_log_softmax</a>()</code> <br><code><a href="activation_mish.html">activation_mish</a>()</code> <br><code><a href="activation_relu.html">activation_relu</a>()</code> <br><code><a href="activation_relu6.html">activation_relu6</a>()</code> <br><code><a href="activation_selu.html">activation_selu</a>()</code> <br><code><a href="activation_sigmoid.html">activation_sigmoid</a>()</code> <br><code><a href="activation_silu.html">activation_silu</a>()</code> <br><code><a href="activation_soft_shrink.html">activation_soft_shrink</a>()</code> <br><code><a href="activation_softmax.html">activation_softmax</a>()</code> <br><code><a href="activation_softplus.html">activation_softplus</a>()</code> <br><code><a href="activation_softsign.html">activation_softsign</a>()</code> <br><code><a href="activation_sparse_plus.html">activation_sparse_plus</a>()</code> <br><code><a href="activation_sparse_sigmoid.html">activation_sparse_sigmoid</a>()</code> <br><code><a href="activation_sparsemax.html">activation_sparsemax</a>()</code> <br><code><a href="activation_squareplus.html">activation_squareplus</a>()</code> <br><code><a href="activation_tanh.html">activation_tanh</a>()</code> <br><code><a href="activation_tanh_shrink.html">activation_tanh_shrink</a>()</code> <br><code><a href="activation_threshold.html">activation_threshold</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -129,7 +129,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/activation_exponential.html b/docs/dev/reference/activation_exponential.html
index e63fbfbec..06a3ef742 100644
--- a/docs/dev/reference/activation_exponential.html
+++ b/docs/dev/reference/activation_exponential.html
@@ -8,7 +8,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -78,7 +78,7 @@ <h2 id="value">Value<a class="anchor" aria-label="anchor" href="#value"></a></h2
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
     <div class="dont-index">
 <ul><li><p><a href="https://keras.io/api/layers/activations#exponential-function" class="external-link">https://keras.io/api/layers/activations#exponential-function</a></p></li>
-</ul><p>Other activations: <br><code><a href="activation_celu.html">activation_celu</a>()</code> <br><code><a href="activation_elu.html">activation_elu</a>()</code> <br><code><a href="activation_gelu.html">activation_gelu</a>()</code> <br><code><a href="activation_glu.html">activation_glu</a>()</code> <br><code><a href="activation_hard_shrink.html">activation_hard_shrink</a>()</code> <br><code><a href="activation_hard_sigmoid.html">activation_hard_sigmoid</a>()</code> <br><code><a href="activation_hard_tanh.html">activation_hard_tanh</a>()</code> <br><code><a href="activation_leaky_relu.html">activation_leaky_relu</a>()</code> <br><code><a href="activation_linear.html">activation_linear</a>()</code> <br><code><a href="activation_log_sigmoid.html">activation_log_sigmoid</a>()</code> <br><code><a href="activation_log_softmax.html">activation_log_softmax</a>()</code> <br><code><a href="activation_mish.html">activation_mish</a>()</code> <br><code><a href="activation_relu.html">activation_relu</a>()</code> <br><code><a href="activation_relu6.html">activation_relu6</a>()</code> <br><code><a href="activation_selu.html">activation_selu</a>()</code> <br><code><a href="activation_sigmoid.html">activation_sigmoid</a>()</code> <br><code><a href="activation_silu.html">activation_silu</a>()</code> <br><code><a href="activation_soft_shrink.html">activation_soft_shrink</a>()</code> <br><code><a href="activation_softmax.html">activation_softmax</a>()</code> <br><code><a href="activation_softplus.html">activation_softplus</a>()</code> <br><code><a href="activation_softsign.html">activation_softsign</a>()</code> <br><code><a href="activation_sparse_plus.html">activation_sparse_plus</a>()</code> <br><code><a href="activation_sparsemax.html">activation_sparsemax</a>()</code> <br><code><a href="activation_squareplus.html">activation_squareplus</a>()</code> <br><code><a href="activation_tanh.html">activation_tanh</a>()</code> <br><code><a href="activation_tanh_shrink.html">activation_tanh_shrink</a>()</code> <br><code><a href="activation_threshold.html">activation_threshold</a>()</code> <br></p></div>
+</ul><p>Other activations: <br><code><a href="activation_celu.html">activation_celu</a>()</code> <br><code><a href="activation_elu.html">activation_elu</a>()</code> <br><code><a href="activation_gelu.html">activation_gelu</a>()</code> <br><code><a href="activation_glu.html">activation_glu</a>()</code> <br><code><a href="activation_hard_shrink.html">activation_hard_shrink</a>()</code> <br><code><a href="activation_hard_sigmoid.html">activation_hard_sigmoid</a>()</code> <br><code><a href="activation_hard_tanh.html">activation_hard_tanh</a>()</code> <br><code><a href="activation_leaky_relu.html">activation_leaky_relu</a>()</code> <br><code><a href="activation_linear.html">activation_linear</a>()</code> <br><code><a href="activation_log_sigmoid.html">activation_log_sigmoid</a>()</code> <br><code><a href="activation_log_softmax.html">activation_log_softmax</a>()</code> <br><code><a href="activation_mish.html">activation_mish</a>()</code> <br><code><a href="activation_relu.html">activation_relu</a>()</code> <br><code><a href="activation_relu6.html">activation_relu6</a>()</code> <br><code><a href="activation_selu.html">activation_selu</a>()</code> <br><code><a href="activation_sigmoid.html">activation_sigmoid</a>()</code> <br><code><a href="activation_silu.html">activation_silu</a>()</code> <br><code><a href="activation_soft_shrink.html">activation_soft_shrink</a>()</code> <br><code><a href="activation_softmax.html">activation_softmax</a>()</code> <br><code><a href="activation_softplus.html">activation_softplus</a>()</code> <br><code><a href="activation_softsign.html">activation_softsign</a>()</code> <br><code><a href="activation_sparse_plus.html">activation_sparse_plus</a>()</code> <br><code><a href="activation_sparse_sigmoid.html">activation_sparse_sigmoid</a>()</code> <br><code><a href="activation_sparsemax.html">activation_sparsemax</a>()</code> <br><code><a href="activation_squareplus.html">activation_squareplus</a>()</code> <br><code><a href="activation_tanh.html">activation_tanh</a>()</code> <br><code><a href="activation_tanh_shrink.html">activation_tanh_shrink</a>()</code> <br><code><a href="activation_threshold.html">activation_threshold</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -90,7 +90,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/activation_gelu.html b/docs/dev/reference/activation_gelu.html
index 5ad163003..ebfc2618c 100644
--- a/docs/dev/reference/activation_gelu.html
+++ b/docs/dev/reference/activation_gelu.html
@@ -16,7 +16,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -99,7 +99,7 @@ <h2 id="reference">Reference<a class="anchor" aria-label="anchor" href="#referen
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
     <div class="dont-index">
 <ul><li><p><a href="https://keras.io/api/layers/activations#gelu-function" class="external-link">https://keras.io/api/layers/activations#gelu-function</a></p></li>
-</ul><p>Other activations: <br><code><a href="activation_celu.html">activation_celu</a>()</code> <br><code><a href="activation_elu.html">activation_elu</a>()</code> <br><code><a href="activation_exponential.html">activation_exponential</a>()</code> <br><code><a href="activation_glu.html">activation_glu</a>()</code> <br><code><a href="activation_hard_shrink.html">activation_hard_shrink</a>()</code> <br><code><a href="activation_hard_sigmoid.html">activation_hard_sigmoid</a>()</code> <br><code><a href="activation_hard_tanh.html">activation_hard_tanh</a>()</code> <br><code><a href="activation_leaky_relu.html">activation_leaky_relu</a>()</code> <br><code><a href="activation_linear.html">activation_linear</a>()</code> <br><code><a href="activation_log_sigmoid.html">activation_log_sigmoid</a>()</code> <br><code><a href="activation_log_softmax.html">activation_log_softmax</a>()</code> <br><code><a href="activation_mish.html">activation_mish</a>()</code> <br><code><a href="activation_relu.html">activation_relu</a>()</code> <br><code><a href="activation_relu6.html">activation_relu6</a>()</code> <br><code><a href="activation_selu.html">activation_selu</a>()</code> <br><code><a href="activation_sigmoid.html">activation_sigmoid</a>()</code> <br><code><a href="activation_silu.html">activation_silu</a>()</code> <br><code><a href="activation_soft_shrink.html">activation_soft_shrink</a>()</code> <br><code><a href="activation_softmax.html">activation_softmax</a>()</code> <br><code><a href="activation_softplus.html">activation_softplus</a>()</code> <br><code><a href="activation_softsign.html">activation_softsign</a>()</code> <br><code><a href="activation_sparse_plus.html">activation_sparse_plus</a>()</code> <br><code><a href="activation_sparsemax.html">activation_sparsemax</a>()</code> <br><code><a href="activation_squareplus.html">activation_squareplus</a>()</code> <br><code><a href="activation_tanh.html">activation_tanh</a>()</code> <br><code><a href="activation_tanh_shrink.html">activation_tanh_shrink</a>()</code> <br><code><a href="activation_threshold.html">activation_threshold</a>()</code> <br></p></div>
+</ul><p>Other activations: <br><code><a href="activation_celu.html">activation_celu</a>()</code> <br><code><a href="activation_elu.html">activation_elu</a>()</code> <br><code><a href="activation_exponential.html">activation_exponential</a>()</code> <br><code><a href="activation_glu.html">activation_glu</a>()</code> <br><code><a href="activation_hard_shrink.html">activation_hard_shrink</a>()</code> <br><code><a href="activation_hard_sigmoid.html">activation_hard_sigmoid</a>()</code> <br><code><a href="activation_hard_tanh.html">activation_hard_tanh</a>()</code> <br><code><a href="activation_leaky_relu.html">activation_leaky_relu</a>()</code> <br><code><a href="activation_linear.html">activation_linear</a>()</code> <br><code><a href="activation_log_sigmoid.html">activation_log_sigmoid</a>()</code> <br><code><a href="activation_log_softmax.html">activation_log_softmax</a>()</code> <br><code><a href="activation_mish.html">activation_mish</a>()</code> <br><code><a href="activation_relu.html">activation_relu</a>()</code> <br><code><a href="activation_relu6.html">activation_relu6</a>()</code> <br><code><a href="activation_selu.html">activation_selu</a>()</code> <br><code><a href="activation_sigmoid.html">activation_sigmoid</a>()</code> <br><code><a href="activation_silu.html">activation_silu</a>()</code> <br><code><a href="activation_soft_shrink.html">activation_soft_shrink</a>()</code> <br><code><a href="activation_softmax.html">activation_softmax</a>()</code> <br><code><a href="activation_softplus.html">activation_softplus</a>()</code> <br><code><a href="activation_softsign.html">activation_softsign</a>()</code> <br><code><a href="activation_sparse_plus.html">activation_sparse_plus</a>()</code> <br><code><a href="activation_sparse_sigmoid.html">activation_sparse_sigmoid</a>()</code> <br><code><a href="activation_sparsemax.html">activation_sparsemax</a>()</code> <br><code><a href="activation_squareplus.html">activation_squareplus</a>()</code> <br><code><a href="activation_tanh.html">activation_tanh</a>()</code> <br><code><a href="activation_tanh_shrink.html">activation_tanh_shrink</a>()</code> <br><code><a href="activation_threshold.html">activation_threshold</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -111,7 +111,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/activation_glu.html b/docs/dev/reference/activation_glu.html
index a474d76ad..5e059e330 100644
--- a/docs/dev/reference/activation_glu.html
+++ b/docs/dev/reference/activation_glu.html
@@ -12,7 +12,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -91,7 +91,7 @@ <h2 id="reference">Reference<a class="anchor" aria-label="anchor" href="#referen
 </ul></div>
     <div class="section level2">
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
-    <div class="dont-index"><p>Other activations: <br><code><a href="activation_celu.html">activation_celu</a>()</code> <br><code><a href="activation_elu.html">activation_elu</a>()</code> <br><code><a href="activation_exponential.html">activation_exponential</a>()</code> <br><code><a href="activation_gelu.html">activation_gelu</a>()</code> <br><code><a href="activation_hard_shrink.html">activation_hard_shrink</a>()</code> <br><code><a href="activation_hard_sigmoid.html">activation_hard_sigmoid</a>()</code> <br><code><a href="activation_hard_tanh.html">activation_hard_tanh</a>()</code> <br><code><a href="activation_leaky_relu.html">activation_leaky_relu</a>()</code> <br><code><a href="activation_linear.html">activation_linear</a>()</code> <br><code><a href="activation_log_sigmoid.html">activation_log_sigmoid</a>()</code> <br><code><a href="activation_log_softmax.html">activation_log_softmax</a>()</code> <br><code><a href="activation_mish.html">activation_mish</a>()</code> <br><code><a href="activation_relu.html">activation_relu</a>()</code> <br><code><a href="activation_relu6.html">activation_relu6</a>()</code> <br><code><a href="activation_selu.html">activation_selu</a>()</code> <br><code><a href="activation_sigmoid.html">activation_sigmoid</a>()</code> <br><code><a href="activation_silu.html">activation_silu</a>()</code> <br><code><a href="activation_soft_shrink.html">activation_soft_shrink</a>()</code> <br><code><a href="activation_softmax.html">activation_softmax</a>()</code> <br><code><a href="activation_softplus.html">activation_softplus</a>()</code> <br><code><a href="activation_softsign.html">activation_softsign</a>()</code> <br><code><a href="activation_sparse_plus.html">activation_sparse_plus</a>()</code> <br><code><a href="activation_sparsemax.html">activation_sparsemax</a>()</code> <br><code><a href="activation_squareplus.html">activation_squareplus</a>()</code> <br><code><a href="activation_tanh.html">activation_tanh</a>()</code> <br><code><a href="activation_tanh_shrink.html">activation_tanh_shrink</a>()</code> <br><code><a href="activation_threshold.html">activation_threshold</a>()</code> <br></p></div>
+    <div class="dont-index"><p>Other activations: <br><code><a href="activation_celu.html">activation_celu</a>()</code> <br><code><a href="activation_elu.html">activation_elu</a>()</code> <br><code><a href="activation_exponential.html">activation_exponential</a>()</code> <br><code><a href="activation_gelu.html">activation_gelu</a>()</code> <br><code><a href="activation_hard_shrink.html">activation_hard_shrink</a>()</code> <br><code><a href="activation_hard_sigmoid.html">activation_hard_sigmoid</a>()</code> <br><code><a href="activation_hard_tanh.html">activation_hard_tanh</a>()</code> <br><code><a href="activation_leaky_relu.html">activation_leaky_relu</a>()</code> <br><code><a href="activation_linear.html">activation_linear</a>()</code> <br><code><a href="activation_log_sigmoid.html">activation_log_sigmoid</a>()</code> <br><code><a href="activation_log_softmax.html">activation_log_softmax</a>()</code> <br><code><a href="activation_mish.html">activation_mish</a>()</code> <br><code><a href="activation_relu.html">activation_relu</a>()</code> <br><code><a href="activation_relu6.html">activation_relu6</a>()</code> <br><code><a href="activation_selu.html">activation_selu</a>()</code> <br><code><a href="activation_sigmoid.html">activation_sigmoid</a>()</code> <br><code><a href="activation_silu.html">activation_silu</a>()</code> <br><code><a href="activation_soft_shrink.html">activation_soft_shrink</a>()</code> <br><code><a href="activation_softmax.html">activation_softmax</a>()</code> <br><code><a href="activation_softplus.html">activation_softplus</a>()</code> <br><code><a href="activation_softsign.html">activation_softsign</a>()</code> <br><code><a href="activation_sparse_plus.html">activation_sparse_plus</a>()</code> <br><code><a href="activation_sparse_sigmoid.html">activation_sparse_sigmoid</a>()</code> <br><code><a href="activation_sparsemax.html">activation_sparsemax</a>()</code> <br><code><a href="activation_squareplus.html">activation_squareplus</a>()</code> <br><code><a href="activation_tanh.html">activation_tanh</a>()</code> <br><code><a href="activation_tanh_shrink.html">activation_tanh_shrink</a>()</code> <br><code><a href="activation_threshold.html">activation_threshold</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -103,7 +103,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/activation_hard_shrink.html b/docs/dev/reference/activation_hard_shrink.html
index 0948e7401..b52e249a7 100644
--- a/docs/dev/reference/activation_hard_shrink.html
+++ b/docs/dev/reference/activation_hard_shrink.html
@@ -12,7 +12,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -86,7 +86,7 @@ <h2 id="value">Value<a class="anchor" aria-label="anchor" href="#value"></a></h2
     </div>
     <div class="section level2">
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
-    <div class="dont-index"><p>Other activations: <br><code><a href="activation_celu.html">activation_celu</a>()</code> <br><code><a href="activation_elu.html">activation_elu</a>()</code> <br><code><a href="activation_exponential.html">activation_exponential</a>()</code> <br><code><a href="activation_gelu.html">activation_gelu</a>()</code> <br><code><a href="activation_glu.html">activation_glu</a>()</code> <br><code><a href="activation_hard_sigmoid.html">activation_hard_sigmoid</a>()</code> <br><code><a href="activation_hard_tanh.html">activation_hard_tanh</a>()</code> <br><code><a href="activation_leaky_relu.html">activation_leaky_relu</a>()</code> <br><code><a href="activation_linear.html">activation_linear</a>()</code> <br><code><a href="activation_log_sigmoid.html">activation_log_sigmoid</a>()</code> <br><code><a href="activation_log_softmax.html">activation_log_softmax</a>()</code> <br><code><a href="activation_mish.html">activation_mish</a>()</code> <br><code><a href="activation_relu.html">activation_relu</a>()</code> <br><code><a href="activation_relu6.html">activation_relu6</a>()</code> <br><code><a href="activation_selu.html">activation_selu</a>()</code> <br><code><a href="activation_sigmoid.html">activation_sigmoid</a>()</code> <br><code><a href="activation_silu.html">activation_silu</a>()</code> <br><code><a href="activation_soft_shrink.html">activation_soft_shrink</a>()</code> <br><code><a href="activation_softmax.html">activation_softmax</a>()</code> <br><code><a href="activation_softplus.html">activation_softplus</a>()</code> <br><code><a href="activation_softsign.html">activation_softsign</a>()</code> <br><code><a href="activation_sparse_plus.html">activation_sparse_plus</a>()</code> <br><code><a href="activation_sparsemax.html">activation_sparsemax</a>()</code> <br><code><a href="activation_squareplus.html">activation_squareplus</a>()</code> <br><code><a href="activation_tanh.html">activation_tanh</a>()</code> <br><code><a href="activation_tanh_shrink.html">activation_tanh_shrink</a>()</code> <br><code><a href="activation_threshold.html">activation_threshold</a>()</code> <br></p></div>
+    <div class="dont-index"><p>Other activations: <br><code><a href="activation_celu.html">activation_celu</a>()</code> <br><code><a href="activation_elu.html">activation_elu</a>()</code> <br><code><a href="activation_exponential.html">activation_exponential</a>()</code> <br><code><a href="activation_gelu.html">activation_gelu</a>()</code> <br><code><a href="activation_glu.html">activation_glu</a>()</code> <br><code><a href="activation_hard_sigmoid.html">activation_hard_sigmoid</a>()</code> <br><code><a href="activation_hard_tanh.html">activation_hard_tanh</a>()</code> <br><code><a href="activation_leaky_relu.html">activation_leaky_relu</a>()</code> <br><code><a href="activation_linear.html">activation_linear</a>()</code> <br><code><a href="activation_log_sigmoid.html">activation_log_sigmoid</a>()</code> <br><code><a href="activation_log_softmax.html">activation_log_softmax</a>()</code> <br><code><a href="activation_mish.html">activation_mish</a>()</code> <br><code><a href="activation_relu.html">activation_relu</a>()</code> <br><code><a href="activation_relu6.html">activation_relu6</a>()</code> <br><code><a href="activation_selu.html">activation_selu</a>()</code> <br><code><a href="activation_sigmoid.html">activation_sigmoid</a>()</code> <br><code><a href="activation_silu.html">activation_silu</a>()</code> <br><code><a href="activation_soft_shrink.html">activation_soft_shrink</a>()</code> <br><code><a href="activation_softmax.html">activation_softmax</a>()</code> <br><code><a href="activation_softplus.html">activation_softplus</a>()</code> <br><code><a href="activation_softsign.html">activation_softsign</a>()</code> <br><code><a href="activation_sparse_plus.html">activation_sparse_plus</a>()</code> <br><code><a href="activation_sparse_sigmoid.html">activation_sparse_sigmoid</a>()</code> <br><code><a href="activation_sparsemax.html">activation_sparsemax</a>()</code> <br><code><a href="activation_squareplus.html">activation_squareplus</a>()</code> <br><code><a href="activation_tanh.html">activation_tanh</a>()</code> <br><code><a href="activation_tanh_shrink.html">activation_tanh_shrink</a>()</code> <br><code><a href="activation_threshold.html">activation_threshold</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -98,7 +98,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/activation_hard_sigmoid.html b/docs/dev/reference/activation_hard_sigmoid.html
index fad0bf6f9..b428484ab 100644
--- a/docs/dev/reference/activation_hard_sigmoid.html
+++ b/docs/dev/reference/activation_hard_sigmoid.html
@@ -22,7 +22,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -101,7 +101,7 @@ <h2 id="reference">Reference<a class="anchor" aria-label="anchor" href="#referen
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
     <div class="dont-index">
 <ul><li><p><a href="https://keras.io/api/layers/activations#hardsigmoid-function" class="external-link">https://keras.io/api/layers/activations#hardsigmoid-function</a></p></li>
-</ul><p>Other activations: <br><code><a href="activation_celu.html">activation_celu</a>()</code> <br><code><a href="activation_elu.html">activation_elu</a>()</code> <br><code><a href="activation_exponential.html">activation_exponential</a>()</code> <br><code><a href="activation_gelu.html">activation_gelu</a>()</code> <br><code><a href="activation_glu.html">activation_glu</a>()</code> <br><code><a href="activation_hard_shrink.html">activation_hard_shrink</a>()</code> <br><code><a href="activation_hard_tanh.html">activation_hard_tanh</a>()</code> <br><code><a href="activation_leaky_relu.html">activation_leaky_relu</a>()</code> <br><code><a href="activation_linear.html">activation_linear</a>()</code> <br><code><a href="activation_log_sigmoid.html">activation_log_sigmoid</a>()</code> <br><code><a href="activation_log_softmax.html">activation_log_softmax</a>()</code> <br><code><a href="activation_mish.html">activation_mish</a>()</code> <br><code><a href="activation_relu.html">activation_relu</a>()</code> <br><code><a href="activation_relu6.html">activation_relu6</a>()</code> <br><code><a href="activation_selu.html">activation_selu</a>()</code> <br><code><a href="activation_sigmoid.html">activation_sigmoid</a>()</code> <br><code><a href="activation_silu.html">activation_silu</a>()</code> <br><code><a href="activation_soft_shrink.html">activation_soft_shrink</a>()</code> <br><code><a href="activation_softmax.html">activation_softmax</a>()</code> <br><code><a href="activation_softplus.html">activation_softplus</a>()</code> <br><code><a href="activation_softsign.html">activation_softsign</a>()</code> <br><code><a href="activation_sparse_plus.html">activation_sparse_plus</a>()</code> <br><code><a href="activation_sparsemax.html">activation_sparsemax</a>()</code> <br><code><a href="activation_squareplus.html">activation_squareplus</a>()</code> <br><code><a href="activation_tanh.html">activation_tanh</a>()</code> <br><code><a href="activation_tanh_shrink.html">activation_tanh_shrink</a>()</code> <br><code><a href="activation_threshold.html">activation_threshold</a>()</code> <br></p></div>
+</ul><p>Other activations: <br><code><a href="activation_celu.html">activation_celu</a>()</code> <br><code><a href="activation_elu.html">activation_elu</a>()</code> <br><code><a href="activation_exponential.html">activation_exponential</a>()</code> <br><code><a href="activation_gelu.html">activation_gelu</a>()</code> <br><code><a href="activation_glu.html">activation_glu</a>()</code> <br><code><a href="activation_hard_shrink.html">activation_hard_shrink</a>()</code> <br><code><a href="activation_hard_tanh.html">activation_hard_tanh</a>()</code> <br><code><a href="activation_leaky_relu.html">activation_leaky_relu</a>()</code> <br><code><a href="activation_linear.html">activation_linear</a>()</code> <br><code><a href="activation_log_sigmoid.html">activation_log_sigmoid</a>()</code> <br><code><a href="activation_log_softmax.html">activation_log_softmax</a>()</code> <br><code><a href="activation_mish.html">activation_mish</a>()</code> <br><code><a href="activation_relu.html">activation_relu</a>()</code> <br><code><a href="activation_relu6.html">activation_relu6</a>()</code> <br><code><a href="activation_selu.html">activation_selu</a>()</code> <br><code><a href="activation_sigmoid.html">activation_sigmoid</a>()</code> <br><code><a href="activation_silu.html">activation_silu</a>()</code> <br><code><a href="activation_soft_shrink.html">activation_soft_shrink</a>()</code> <br><code><a href="activation_softmax.html">activation_softmax</a>()</code> <br><code><a href="activation_softplus.html">activation_softplus</a>()</code> <br><code><a href="activation_softsign.html">activation_softsign</a>()</code> <br><code><a href="activation_sparse_plus.html">activation_sparse_plus</a>()</code> <br><code><a href="activation_sparse_sigmoid.html">activation_sparse_sigmoid</a>()</code> <br><code><a href="activation_sparsemax.html">activation_sparsemax</a>()</code> <br><code><a href="activation_squareplus.html">activation_squareplus</a>()</code> <br><code><a href="activation_tanh.html">activation_tanh</a>()</code> <br><code><a href="activation_tanh_shrink.html">activation_tanh_shrink</a>()</code> <br><code><a href="activation_threshold.html">activation_threshold</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -113,7 +113,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/activation_hard_silu.html b/docs/dev/reference/activation_hard_silu.html
index d0909684e..9e29652f0 100644
--- a/docs/dev/reference/activation_hard_silu.html
+++ b/docs/dev/reference/activation_hard_silu.html
@@ -20,7 +20,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -106,7 +106,7 @@ <h2 id="reference">Reference<a class="anchor" aria-label="anchor" href="#referen
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/activation_hard_tanh.html b/docs/dev/reference/activation_hard_tanh.html
index e04bd1db3..49cb575f0 100644
--- a/docs/dev/reference/activation_hard_tanh.html
+++ b/docs/dev/reference/activation_hard_tanh.html
@@ -14,7 +14,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -85,7 +85,7 @@ <h2 id="value">Value<a class="anchor" aria-label="anchor" href="#value"></a></h2
     </div>
     <div class="section level2">
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
-    <div class="dont-index"><p>Other activations: <br><code><a href="activation_celu.html">activation_celu</a>()</code> <br><code><a href="activation_elu.html">activation_elu</a>()</code> <br><code><a href="activation_exponential.html">activation_exponential</a>()</code> <br><code><a href="activation_gelu.html">activation_gelu</a>()</code> <br><code><a href="activation_glu.html">activation_glu</a>()</code> <br><code><a href="activation_hard_shrink.html">activation_hard_shrink</a>()</code> <br><code><a href="activation_hard_sigmoid.html">activation_hard_sigmoid</a>()</code> <br><code><a href="activation_leaky_relu.html">activation_leaky_relu</a>()</code> <br><code><a href="activation_linear.html">activation_linear</a>()</code> <br><code><a href="activation_log_sigmoid.html">activation_log_sigmoid</a>()</code> <br><code><a href="activation_log_softmax.html">activation_log_softmax</a>()</code> <br><code><a href="activation_mish.html">activation_mish</a>()</code> <br><code><a href="activation_relu.html">activation_relu</a>()</code> <br><code><a href="activation_relu6.html">activation_relu6</a>()</code> <br><code><a href="activation_selu.html">activation_selu</a>()</code> <br><code><a href="activation_sigmoid.html">activation_sigmoid</a>()</code> <br><code><a href="activation_silu.html">activation_silu</a>()</code> <br><code><a href="activation_soft_shrink.html">activation_soft_shrink</a>()</code> <br><code><a href="activation_softmax.html">activation_softmax</a>()</code> <br><code><a href="activation_softplus.html">activation_softplus</a>()</code> <br><code><a href="activation_softsign.html">activation_softsign</a>()</code> <br><code><a href="activation_sparse_plus.html">activation_sparse_plus</a>()</code> <br><code><a href="activation_sparsemax.html">activation_sparsemax</a>()</code> <br><code><a href="activation_squareplus.html">activation_squareplus</a>()</code> <br><code><a href="activation_tanh.html">activation_tanh</a>()</code> <br><code><a href="activation_tanh_shrink.html">activation_tanh_shrink</a>()</code> <br><code><a href="activation_threshold.html">activation_threshold</a>()</code> <br></p></div>
+    <div class="dont-index"><p>Other activations: <br><code><a href="activation_celu.html">activation_celu</a>()</code> <br><code><a href="activation_elu.html">activation_elu</a>()</code> <br><code><a href="activation_exponential.html">activation_exponential</a>()</code> <br><code><a href="activation_gelu.html">activation_gelu</a>()</code> <br><code><a href="activation_glu.html">activation_glu</a>()</code> <br><code><a href="activation_hard_shrink.html">activation_hard_shrink</a>()</code> <br><code><a href="activation_hard_sigmoid.html">activation_hard_sigmoid</a>()</code> <br><code><a href="activation_leaky_relu.html">activation_leaky_relu</a>()</code> <br><code><a href="activation_linear.html">activation_linear</a>()</code> <br><code><a href="activation_log_sigmoid.html">activation_log_sigmoid</a>()</code> <br><code><a href="activation_log_softmax.html">activation_log_softmax</a>()</code> <br><code><a href="activation_mish.html">activation_mish</a>()</code> <br><code><a href="activation_relu.html">activation_relu</a>()</code> <br><code><a href="activation_relu6.html">activation_relu6</a>()</code> <br><code><a href="activation_selu.html">activation_selu</a>()</code> <br><code><a href="activation_sigmoid.html">activation_sigmoid</a>()</code> <br><code><a href="activation_silu.html">activation_silu</a>()</code> <br><code><a href="activation_soft_shrink.html">activation_soft_shrink</a>()</code> <br><code><a href="activation_softmax.html">activation_softmax</a>()</code> <br><code><a href="activation_softplus.html">activation_softplus</a>()</code> <br><code><a href="activation_softsign.html">activation_softsign</a>()</code> <br><code><a href="activation_sparse_plus.html">activation_sparse_plus</a>()</code> <br><code><a href="activation_sparse_sigmoid.html">activation_sparse_sigmoid</a>()</code> <br><code><a href="activation_sparsemax.html">activation_sparsemax</a>()</code> <br><code><a href="activation_squareplus.html">activation_squareplus</a>()</code> <br><code><a href="activation_tanh.html">activation_tanh</a>()</code> <br><code><a href="activation_tanh_shrink.html">activation_tanh_shrink</a>()</code> <br><code><a href="activation_threshold.html">activation_threshold</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -97,7 +97,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/activation_leaky_relu.html b/docs/dev/reference/activation_leaky_relu.html
index b0ac0f732..eb3a2fc8c 100644
--- a/docs/dev/reference/activation_leaky_relu.html
+++ b/docs/dev/reference/activation_leaky_relu.html
@@ -8,7 +8,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -83,7 +83,7 @@ <h2 id="value">Value<a class="anchor" aria-label="anchor" href="#value"></a></h2
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
     <div class="dont-index">
 <ul><li><p><a href="https://keras.io/api/layers/activations#leakyrelu-function" class="external-link">https://keras.io/api/layers/activations#leakyrelu-function</a></p></li>
-</ul><p>Other activations: <br><code><a href="activation_celu.html">activation_celu</a>()</code> <br><code><a href="activation_elu.html">activation_elu</a>()</code> <br><code><a href="activation_exponential.html">activation_exponential</a>()</code> <br><code><a href="activation_gelu.html">activation_gelu</a>()</code> <br><code><a href="activation_glu.html">activation_glu</a>()</code> <br><code><a href="activation_hard_shrink.html">activation_hard_shrink</a>()</code> <br><code><a href="activation_hard_sigmoid.html">activation_hard_sigmoid</a>()</code> <br><code><a href="activation_hard_tanh.html">activation_hard_tanh</a>()</code> <br><code><a href="activation_linear.html">activation_linear</a>()</code> <br><code><a href="activation_log_sigmoid.html">activation_log_sigmoid</a>()</code> <br><code><a href="activation_log_softmax.html">activation_log_softmax</a>()</code> <br><code><a href="activation_mish.html">activation_mish</a>()</code> <br><code><a href="activation_relu.html">activation_relu</a>()</code> <br><code><a href="activation_relu6.html">activation_relu6</a>()</code> <br><code><a href="activation_selu.html">activation_selu</a>()</code> <br><code><a href="activation_sigmoid.html">activation_sigmoid</a>()</code> <br><code><a href="activation_silu.html">activation_silu</a>()</code> <br><code><a href="activation_soft_shrink.html">activation_soft_shrink</a>()</code> <br><code><a href="activation_softmax.html">activation_softmax</a>()</code> <br><code><a href="activation_softplus.html">activation_softplus</a>()</code> <br><code><a href="activation_softsign.html">activation_softsign</a>()</code> <br><code><a href="activation_sparse_plus.html">activation_sparse_plus</a>()</code> <br><code><a href="activation_sparsemax.html">activation_sparsemax</a>()</code> <br><code><a href="activation_squareplus.html">activation_squareplus</a>()</code> <br><code><a href="activation_tanh.html">activation_tanh</a>()</code> <br><code><a href="activation_tanh_shrink.html">activation_tanh_shrink</a>()</code> <br><code><a href="activation_threshold.html">activation_threshold</a>()</code> <br></p></div>
+</ul><p>Other activations: <br><code><a href="activation_celu.html">activation_celu</a>()</code> <br><code><a href="activation_elu.html">activation_elu</a>()</code> <br><code><a href="activation_exponential.html">activation_exponential</a>()</code> <br><code><a href="activation_gelu.html">activation_gelu</a>()</code> <br><code><a href="activation_glu.html">activation_glu</a>()</code> <br><code><a href="activation_hard_shrink.html">activation_hard_shrink</a>()</code> <br><code><a href="activation_hard_sigmoid.html">activation_hard_sigmoid</a>()</code> <br><code><a href="activation_hard_tanh.html">activation_hard_tanh</a>()</code> <br><code><a href="activation_linear.html">activation_linear</a>()</code> <br><code><a href="activation_log_sigmoid.html">activation_log_sigmoid</a>()</code> <br><code><a href="activation_log_softmax.html">activation_log_softmax</a>()</code> <br><code><a href="activation_mish.html">activation_mish</a>()</code> <br><code><a href="activation_relu.html">activation_relu</a>()</code> <br><code><a href="activation_relu6.html">activation_relu6</a>()</code> <br><code><a href="activation_selu.html">activation_selu</a>()</code> <br><code><a href="activation_sigmoid.html">activation_sigmoid</a>()</code> <br><code><a href="activation_silu.html">activation_silu</a>()</code> <br><code><a href="activation_soft_shrink.html">activation_soft_shrink</a>()</code> <br><code><a href="activation_softmax.html">activation_softmax</a>()</code> <br><code><a href="activation_softplus.html">activation_softplus</a>()</code> <br><code><a href="activation_softsign.html">activation_softsign</a>()</code> <br><code><a href="activation_sparse_plus.html">activation_sparse_plus</a>()</code> <br><code><a href="activation_sparse_sigmoid.html">activation_sparse_sigmoid</a>()</code> <br><code><a href="activation_sparsemax.html">activation_sparsemax</a>()</code> <br><code><a href="activation_squareplus.html">activation_squareplus</a>()</code> <br><code><a href="activation_tanh.html">activation_tanh</a>()</code> <br><code><a href="activation_tanh_shrink.html">activation_tanh_shrink</a>()</code> <br><code><a href="activation_threshold.html">activation_threshold</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -95,7 +95,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/activation_linear.html b/docs/dev/reference/activation_linear.html
index 13aa22c0e..bff2a9ed1 100644
--- a/docs/dev/reference/activation_linear.html
+++ b/docs/dev/reference/activation_linear.html
@@ -10,7 +10,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -81,7 +81,7 @@ <h2 id="value">Value<a class="anchor" aria-label="anchor" href="#value"></a></h2
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
     <div class="dont-index">
 <ul><li><p><a href="https://keras.io/api/layers/activations#linear-function" class="external-link">https://keras.io/api/layers/activations#linear-function</a></p></li>
-</ul><p>Other activations: <br><code><a href="activation_celu.html">activation_celu</a>()</code> <br><code><a href="activation_elu.html">activation_elu</a>()</code> <br><code><a href="activation_exponential.html">activation_exponential</a>()</code> <br><code><a href="activation_gelu.html">activation_gelu</a>()</code> <br><code><a href="activation_glu.html">activation_glu</a>()</code> <br><code><a href="activation_hard_shrink.html">activation_hard_shrink</a>()</code> <br><code><a href="activation_hard_sigmoid.html">activation_hard_sigmoid</a>()</code> <br><code><a href="activation_hard_tanh.html">activation_hard_tanh</a>()</code> <br><code><a href="activation_leaky_relu.html">activation_leaky_relu</a>()</code> <br><code><a href="activation_log_sigmoid.html">activation_log_sigmoid</a>()</code> <br><code><a href="activation_log_softmax.html">activation_log_softmax</a>()</code> <br><code><a href="activation_mish.html">activation_mish</a>()</code> <br><code><a href="activation_relu.html">activation_relu</a>()</code> <br><code><a href="activation_relu6.html">activation_relu6</a>()</code> <br><code><a href="activation_selu.html">activation_selu</a>()</code> <br><code><a href="activation_sigmoid.html">activation_sigmoid</a>()</code> <br><code><a href="activation_silu.html">activation_silu</a>()</code> <br><code><a href="activation_soft_shrink.html">activation_soft_shrink</a>()</code> <br><code><a href="activation_softmax.html">activation_softmax</a>()</code> <br><code><a href="activation_softplus.html">activation_softplus</a>()</code> <br><code><a href="activation_softsign.html">activation_softsign</a>()</code> <br><code><a href="activation_sparse_plus.html">activation_sparse_plus</a>()</code> <br><code><a href="activation_sparsemax.html">activation_sparsemax</a>()</code> <br><code><a href="activation_squareplus.html">activation_squareplus</a>()</code> <br><code><a href="activation_tanh.html">activation_tanh</a>()</code> <br><code><a href="activation_tanh_shrink.html">activation_tanh_shrink</a>()</code> <br><code><a href="activation_threshold.html">activation_threshold</a>()</code> <br></p></div>
+</ul><p>Other activations: <br><code><a href="activation_celu.html">activation_celu</a>()</code> <br><code><a href="activation_elu.html">activation_elu</a>()</code> <br><code><a href="activation_exponential.html">activation_exponential</a>()</code> <br><code><a href="activation_gelu.html">activation_gelu</a>()</code> <br><code><a href="activation_glu.html">activation_glu</a>()</code> <br><code><a href="activation_hard_shrink.html">activation_hard_shrink</a>()</code> <br><code><a href="activation_hard_sigmoid.html">activation_hard_sigmoid</a>()</code> <br><code><a href="activation_hard_tanh.html">activation_hard_tanh</a>()</code> <br><code><a href="activation_leaky_relu.html">activation_leaky_relu</a>()</code> <br><code><a href="activation_log_sigmoid.html">activation_log_sigmoid</a>()</code> <br><code><a href="activation_log_softmax.html">activation_log_softmax</a>()</code> <br><code><a href="activation_mish.html">activation_mish</a>()</code> <br><code><a href="activation_relu.html">activation_relu</a>()</code> <br><code><a href="activation_relu6.html">activation_relu6</a>()</code> <br><code><a href="activation_selu.html">activation_selu</a>()</code> <br><code><a href="activation_sigmoid.html">activation_sigmoid</a>()</code> <br><code><a href="activation_silu.html">activation_silu</a>()</code> <br><code><a href="activation_soft_shrink.html">activation_soft_shrink</a>()</code> <br><code><a href="activation_softmax.html">activation_softmax</a>()</code> <br><code><a href="activation_softplus.html">activation_softplus</a>()</code> <br><code><a href="activation_softsign.html">activation_softsign</a>()</code> <br><code><a href="activation_sparse_plus.html">activation_sparse_plus</a>()</code> <br><code><a href="activation_sparse_sigmoid.html">activation_sparse_sigmoid</a>()</code> <br><code><a href="activation_sparsemax.html">activation_sparsemax</a>()</code> <br><code><a href="activation_squareplus.html">activation_squareplus</a>()</code> <br><code><a href="activation_tanh.html">activation_tanh</a>()</code> <br><code><a href="activation_tanh_shrink.html">activation_tanh_shrink</a>()</code> <br><code><a href="activation_threshold.html">activation_threshold</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -93,7 +93,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/activation_log_sigmoid.html b/docs/dev/reference/activation_log_sigmoid.html
index 776595fd3..664ef29d3 100644
--- a/docs/dev/reference/activation_log_sigmoid.html
+++ b/docs/dev/reference/activation_log_sigmoid.html
@@ -8,7 +8,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -76,7 +76,7 @@ <h2 id="value">Value<a class="anchor" aria-label="anchor" href="#value"></a></h2
     </div>
     <div class="section level2">
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
-    <div class="dont-index"><p>Other activations: <br><code><a href="activation_celu.html">activation_celu</a>()</code> <br><code><a href="activation_elu.html">activation_elu</a>()</code> <br><code><a href="activation_exponential.html">activation_exponential</a>()</code> <br><code><a href="activation_gelu.html">activation_gelu</a>()</code> <br><code><a href="activation_glu.html">activation_glu</a>()</code> <br><code><a href="activation_hard_shrink.html">activation_hard_shrink</a>()</code> <br><code><a href="activation_hard_sigmoid.html">activation_hard_sigmoid</a>()</code> <br><code><a href="activation_hard_tanh.html">activation_hard_tanh</a>()</code> <br><code><a href="activation_leaky_relu.html">activation_leaky_relu</a>()</code> <br><code><a href="activation_linear.html">activation_linear</a>()</code> <br><code><a href="activation_log_softmax.html">activation_log_softmax</a>()</code> <br><code><a href="activation_mish.html">activation_mish</a>()</code> <br><code><a href="activation_relu.html">activation_relu</a>()</code> <br><code><a href="activation_relu6.html">activation_relu6</a>()</code> <br><code><a href="activation_selu.html">activation_selu</a>()</code> <br><code><a href="activation_sigmoid.html">activation_sigmoid</a>()</code> <br><code><a href="activation_silu.html">activation_silu</a>()</code> <br><code><a href="activation_soft_shrink.html">activation_soft_shrink</a>()</code> <br><code><a href="activation_softmax.html">activation_softmax</a>()</code> <br><code><a href="activation_softplus.html">activation_softplus</a>()</code> <br><code><a href="activation_softsign.html">activation_softsign</a>()</code> <br><code><a href="activation_sparse_plus.html">activation_sparse_plus</a>()</code> <br><code><a href="activation_sparsemax.html">activation_sparsemax</a>()</code> <br><code><a href="activation_squareplus.html">activation_squareplus</a>()</code> <br><code><a href="activation_tanh.html">activation_tanh</a>()</code> <br><code><a href="activation_tanh_shrink.html">activation_tanh_shrink</a>()</code> <br><code><a href="activation_threshold.html">activation_threshold</a>()</code> <br></p></div>
+    <div class="dont-index"><p>Other activations: <br><code><a href="activation_celu.html">activation_celu</a>()</code> <br><code><a href="activation_elu.html">activation_elu</a>()</code> <br><code><a href="activation_exponential.html">activation_exponential</a>()</code> <br><code><a href="activation_gelu.html">activation_gelu</a>()</code> <br><code><a href="activation_glu.html">activation_glu</a>()</code> <br><code><a href="activation_hard_shrink.html">activation_hard_shrink</a>()</code> <br><code><a href="activation_hard_sigmoid.html">activation_hard_sigmoid</a>()</code> <br><code><a href="activation_hard_tanh.html">activation_hard_tanh</a>()</code> <br><code><a href="activation_leaky_relu.html">activation_leaky_relu</a>()</code> <br><code><a href="activation_linear.html">activation_linear</a>()</code> <br><code><a href="activation_log_softmax.html">activation_log_softmax</a>()</code> <br><code><a href="activation_mish.html">activation_mish</a>()</code> <br><code><a href="activation_relu.html">activation_relu</a>()</code> <br><code><a href="activation_relu6.html">activation_relu6</a>()</code> <br><code><a href="activation_selu.html">activation_selu</a>()</code> <br><code><a href="activation_sigmoid.html">activation_sigmoid</a>()</code> <br><code><a href="activation_silu.html">activation_silu</a>()</code> <br><code><a href="activation_soft_shrink.html">activation_soft_shrink</a>()</code> <br><code><a href="activation_softmax.html">activation_softmax</a>()</code> <br><code><a href="activation_softplus.html">activation_softplus</a>()</code> <br><code><a href="activation_softsign.html">activation_softsign</a>()</code> <br><code><a href="activation_sparse_plus.html">activation_sparse_plus</a>()</code> <br><code><a href="activation_sparse_sigmoid.html">activation_sparse_sigmoid</a>()</code> <br><code><a href="activation_sparsemax.html">activation_sparsemax</a>()</code> <br><code><a href="activation_squareplus.html">activation_squareplus</a>()</code> <br><code><a href="activation_tanh.html">activation_tanh</a>()</code> <br><code><a href="activation_tanh_shrink.html">activation_tanh_shrink</a>()</code> <br><code><a href="activation_threshold.html">activation_threshold</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -88,7 +88,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/activation_log_softmax.html b/docs/dev/reference/activation_log_softmax.html
index 5cd30b3ba..c04236e58 100644
--- a/docs/dev/reference/activation_log_softmax.html
+++ b/docs/dev/reference/activation_log_softmax.html
@@ -12,7 +12,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -88,7 +88,7 @@ <h2 id="value">Value<a class="anchor" aria-label="anchor" href="#value"></a></h2
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
     <div class="dont-index">
 <ul><li><p><a href="https://keras.io/api/layers/activations#logsoftmax-function" class="external-link">https://keras.io/api/layers/activations#logsoftmax-function</a></p></li>
-</ul><p>Other activations: <br><code><a href="activation_celu.html">activation_celu</a>()</code> <br><code><a href="activation_elu.html">activation_elu</a>()</code> <br><code><a href="activation_exponential.html">activation_exponential</a>()</code> <br><code><a href="activation_gelu.html">activation_gelu</a>()</code> <br><code><a href="activation_glu.html">activation_glu</a>()</code> <br><code><a href="activation_hard_shrink.html">activation_hard_shrink</a>()</code> <br><code><a href="activation_hard_sigmoid.html">activation_hard_sigmoid</a>()</code> <br><code><a href="activation_hard_tanh.html">activation_hard_tanh</a>()</code> <br><code><a href="activation_leaky_relu.html">activation_leaky_relu</a>()</code> <br><code><a href="activation_linear.html">activation_linear</a>()</code> <br><code><a href="activation_log_sigmoid.html">activation_log_sigmoid</a>()</code> <br><code><a href="activation_mish.html">activation_mish</a>()</code> <br><code><a href="activation_relu.html">activation_relu</a>()</code> <br><code><a href="activation_relu6.html">activation_relu6</a>()</code> <br><code><a href="activation_selu.html">activation_selu</a>()</code> <br><code><a href="activation_sigmoid.html">activation_sigmoid</a>()</code> <br><code><a href="activation_silu.html">activation_silu</a>()</code> <br><code><a href="activation_soft_shrink.html">activation_soft_shrink</a>()</code> <br><code><a href="activation_softmax.html">activation_softmax</a>()</code> <br><code><a href="activation_softplus.html">activation_softplus</a>()</code> <br><code><a href="activation_softsign.html">activation_softsign</a>()</code> <br><code><a href="activation_sparse_plus.html">activation_sparse_plus</a>()</code> <br><code><a href="activation_sparsemax.html">activation_sparsemax</a>()</code> <br><code><a href="activation_squareplus.html">activation_squareplus</a>()</code> <br><code><a href="activation_tanh.html">activation_tanh</a>()</code> <br><code><a href="activation_tanh_shrink.html">activation_tanh_shrink</a>()</code> <br><code><a href="activation_threshold.html">activation_threshold</a>()</code> <br></p></div>
+</ul><p>Other activations: <br><code><a href="activation_celu.html">activation_celu</a>()</code> <br><code><a href="activation_elu.html">activation_elu</a>()</code> <br><code><a href="activation_exponential.html">activation_exponential</a>()</code> <br><code><a href="activation_gelu.html">activation_gelu</a>()</code> <br><code><a href="activation_glu.html">activation_glu</a>()</code> <br><code><a href="activation_hard_shrink.html">activation_hard_shrink</a>()</code> <br><code><a href="activation_hard_sigmoid.html">activation_hard_sigmoid</a>()</code> <br><code><a href="activation_hard_tanh.html">activation_hard_tanh</a>()</code> <br><code><a href="activation_leaky_relu.html">activation_leaky_relu</a>()</code> <br><code><a href="activation_linear.html">activation_linear</a>()</code> <br><code><a href="activation_log_sigmoid.html">activation_log_sigmoid</a>()</code> <br><code><a href="activation_mish.html">activation_mish</a>()</code> <br><code><a href="activation_relu.html">activation_relu</a>()</code> <br><code><a href="activation_relu6.html">activation_relu6</a>()</code> <br><code><a href="activation_selu.html">activation_selu</a>()</code> <br><code><a href="activation_sigmoid.html">activation_sigmoid</a>()</code> <br><code><a href="activation_silu.html">activation_silu</a>()</code> <br><code><a href="activation_soft_shrink.html">activation_soft_shrink</a>()</code> <br><code><a href="activation_softmax.html">activation_softmax</a>()</code> <br><code><a href="activation_softplus.html">activation_softplus</a>()</code> <br><code><a href="activation_softsign.html">activation_softsign</a>()</code> <br><code><a href="activation_sparse_plus.html">activation_sparse_plus</a>()</code> <br><code><a href="activation_sparse_sigmoid.html">activation_sparse_sigmoid</a>()</code> <br><code><a href="activation_sparsemax.html">activation_sparsemax</a>()</code> <br><code><a href="activation_squareplus.html">activation_squareplus</a>()</code> <br><code><a href="activation_tanh.html">activation_tanh</a>()</code> <br><code><a href="activation_tanh_shrink.html">activation_tanh_shrink</a>()</code> <br><code><a href="activation_threshold.html">activation_threshold</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -100,7 +100,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/activation_mish.html b/docs/dev/reference/activation_mish.html
index e64bcfb31..74d8e9875 100644
--- a/docs/dev/reference/activation_mish.html
+++ b/docs/dev/reference/activation_mish.html
@@ -14,7 +14,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -92,7 +92,7 @@ <h2 id="reference">Reference<a class="anchor" aria-label="anchor" href="#referen
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
     <div class="dont-index">
 <ul><li><p><a href="https://keras.io/api/layers/activations#mish-function" class="external-link">https://keras.io/api/layers/activations#mish-function</a></p></li>
-</ul><p>Other activations: <br><code><a href="activation_celu.html">activation_celu</a>()</code> <br><code><a href="activation_elu.html">activation_elu</a>()</code> <br><code><a href="activation_exponential.html">activation_exponential</a>()</code> <br><code><a href="activation_gelu.html">activation_gelu</a>()</code> <br><code><a href="activation_glu.html">activation_glu</a>()</code> <br><code><a href="activation_hard_shrink.html">activation_hard_shrink</a>()</code> <br><code><a href="activation_hard_sigmoid.html">activation_hard_sigmoid</a>()</code> <br><code><a href="activation_hard_tanh.html">activation_hard_tanh</a>()</code> <br><code><a href="activation_leaky_relu.html">activation_leaky_relu</a>()</code> <br><code><a href="activation_linear.html">activation_linear</a>()</code> <br><code><a href="activation_log_sigmoid.html">activation_log_sigmoid</a>()</code> <br><code><a href="activation_log_softmax.html">activation_log_softmax</a>()</code> <br><code><a href="activation_relu.html">activation_relu</a>()</code> <br><code><a href="activation_relu6.html">activation_relu6</a>()</code> <br><code><a href="activation_selu.html">activation_selu</a>()</code> <br><code><a href="activation_sigmoid.html">activation_sigmoid</a>()</code> <br><code><a href="activation_silu.html">activation_silu</a>()</code> <br><code><a href="activation_soft_shrink.html">activation_soft_shrink</a>()</code> <br><code><a href="activation_softmax.html">activation_softmax</a>()</code> <br><code><a href="activation_softplus.html">activation_softplus</a>()</code> <br><code><a href="activation_softsign.html">activation_softsign</a>()</code> <br><code><a href="activation_sparse_plus.html">activation_sparse_plus</a>()</code> <br><code><a href="activation_sparsemax.html">activation_sparsemax</a>()</code> <br><code><a href="activation_squareplus.html">activation_squareplus</a>()</code> <br><code><a href="activation_tanh.html">activation_tanh</a>()</code> <br><code><a href="activation_tanh_shrink.html">activation_tanh_shrink</a>()</code> <br><code><a href="activation_threshold.html">activation_threshold</a>()</code> <br></p></div>
+</ul><p>Other activations: <br><code><a href="activation_celu.html">activation_celu</a>()</code> <br><code><a href="activation_elu.html">activation_elu</a>()</code> <br><code><a href="activation_exponential.html">activation_exponential</a>()</code> <br><code><a href="activation_gelu.html">activation_gelu</a>()</code> <br><code><a href="activation_glu.html">activation_glu</a>()</code> <br><code><a href="activation_hard_shrink.html">activation_hard_shrink</a>()</code> <br><code><a href="activation_hard_sigmoid.html">activation_hard_sigmoid</a>()</code> <br><code><a href="activation_hard_tanh.html">activation_hard_tanh</a>()</code> <br><code><a href="activation_leaky_relu.html">activation_leaky_relu</a>()</code> <br><code><a href="activation_linear.html">activation_linear</a>()</code> <br><code><a href="activation_log_sigmoid.html">activation_log_sigmoid</a>()</code> <br><code><a href="activation_log_softmax.html">activation_log_softmax</a>()</code> <br><code><a href="activation_relu.html">activation_relu</a>()</code> <br><code><a href="activation_relu6.html">activation_relu6</a>()</code> <br><code><a href="activation_selu.html">activation_selu</a>()</code> <br><code><a href="activation_sigmoid.html">activation_sigmoid</a>()</code> <br><code><a href="activation_silu.html">activation_silu</a>()</code> <br><code><a href="activation_soft_shrink.html">activation_soft_shrink</a>()</code> <br><code><a href="activation_softmax.html">activation_softmax</a>()</code> <br><code><a href="activation_softplus.html">activation_softplus</a>()</code> <br><code><a href="activation_softsign.html">activation_softsign</a>()</code> <br><code><a href="activation_sparse_plus.html">activation_sparse_plus</a>()</code> <br><code><a href="activation_sparse_sigmoid.html">activation_sparse_sigmoid</a>()</code> <br><code><a href="activation_sparsemax.html">activation_sparsemax</a>()</code> <br><code><a href="activation_squareplus.html">activation_squareplus</a>()</code> <br><code><a href="activation_tanh.html">activation_tanh</a>()</code> <br><code><a href="activation_tanh_shrink.html">activation_tanh_shrink</a>()</code> <br><code><a href="activation_threshold.html">activation_threshold</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -104,7 +104,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/activation_relu.html b/docs/dev/reference/activation_relu.html
index 90bd4aa71..def1c7add 100644
--- a/docs/dev/reference/activation_relu.html
+++ b/docs/dev/reference/activation_relu.html
@@ -16,7 +16,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -121,7 +121,7 @@ <h2 id="examples">Examples<a class="anchor" aria-label="anchor" href="#examples"
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
     <div class="dont-index">
 <ul><li><p><a href="https://keras.io/api/layers/activations#relu-function" class="external-link">https://keras.io/api/layers/activations#relu-function</a></p></li>
-</ul><p>Other activations: <br><code><a href="activation_celu.html">activation_celu</a>()</code> <br><code><a href="activation_elu.html">activation_elu</a>()</code> <br><code><a href="activation_exponential.html">activation_exponential</a>()</code> <br><code><a href="activation_gelu.html">activation_gelu</a>()</code> <br><code><a href="activation_glu.html">activation_glu</a>()</code> <br><code><a href="activation_hard_shrink.html">activation_hard_shrink</a>()</code> <br><code><a href="activation_hard_sigmoid.html">activation_hard_sigmoid</a>()</code> <br><code><a href="activation_hard_tanh.html">activation_hard_tanh</a>()</code> <br><code><a href="activation_leaky_relu.html">activation_leaky_relu</a>()</code> <br><code><a href="activation_linear.html">activation_linear</a>()</code> <br><code><a href="activation_log_sigmoid.html">activation_log_sigmoid</a>()</code> <br><code><a href="activation_log_softmax.html">activation_log_softmax</a>()</code> <br><code><a href="activation_mish.html">activation_mish</a>()</code> <br><code><a href="activation_relu6.html">activation_relu6</a>()</code> <br><code><a href="activation_selu.html">activation_selu</a>()</code> <br><code><a href="activation_sigmoid.html">activation_sigmoid</a>()</code> <br><code><a href="activation_silu.html">activation_silu</a>()</code> <br><code><a href="activation_soft_shrink.html">activation_soft_shrink</a>()</code> <br><code><a href="activation_softmax.html">activation_softmax</a>()</code> <br><code><a href="activation_softplus.html">activation_softplus</a>()</code> <br><code><a href="activation_softsign.html">activation_softsign</a>()</code> <br><code><a href="activation_sparse_plus.html">activation_sparse_plus</a>()</code> <br><code><a href="activation_sparsemax.html">activation_sparsemax</a>()</code> <br><code><a href="activation_squareplus.html">activation_squareplus</a>()</code> <br><code><a href="activation_tanh.html">activation_tanh</a>()</code> <br><code><a href="activation_tanh_shrink.html">activation_tanh_shrink</a>()</code> <br><code><a href="activation_threshold.html">activation_threshold</a>()</code> <br></p></div>
+</ul><p>Other activations: <br><code><a href="activation_celu.html">activation_celu</a>()</code> <br><code><a href="activation_elu.html">activation_elu</a>()</code> <br><code><a href="activation_exponential.html">activation_exponential</a>()</code> <br><code><a href="activation_gelu.html">activation_gelu</a>()</code> <br><code><a href="activation_glu.html">activation_glu</a>()</code> <br><code><a href="activation_hard_shrink.html">activation_hard_shrink</a>()</code> <br><code><a href="activation_hard_sigmoid.html">activation_hard_sigmoid</a>()</code> <br><code><a href="activation_hard_tanh.html">activation_hard_tanh</a>()</code> <br><code><a href="activation_leaky_relu.html">activation_leaky_relu</a>()</code> <br><code><a href="activation_linear.html">activation_linear</a>()</code> <br><code><a href="activation_log_sigmoid.html">activation_log_sigmoid</a>()</code> <br><code><a href="activation_log_softmax.html">activation_log_softmax</a>()</code> <br><code><a href="activation_mish.html">activation_mish</a>()</code> <br><code><a href="activation_relu6.html">activation_relu6</a>()</code> <br><code><a href="activation_selu.html">activation_selu</a>()</code> <br><code><a href="activation_sigmoid.html">activation_sigmoid</a>()</code> <br><code><a href="activation_silu.html">activation_silu</a>()</code> <br><code><a href="activation_soft_shrink.html">activation_soft_shrink</a>()</code> <br><code><a href="activation_softmax.html">activation_softmax</a>()</code> <br><code><a href="activation_softplus.html">activation_softplus</a>()</code> <br><code><a href="activation_softsign.html">activation_softsign</a>()</code> <br><code><a href="activation_sparse_plus.html">activation_sparse_plus</a>()</code> <br><code><a href="activation_sparse_sigmoid.html">activation_sparse_sigmoid</a>()</code> <br><code><a href="activation_sparsemax.html">activation_sparsemax</a>()</code> <br><code><a href="activation_squareplus.html">activation_squareplus</a>()</code> <br><code><a href="activation_tanh.html">activation_tanh</a>()</code> <br><code><a href="activation_tanh_shrink.html">activation_tanh_shrink</a>()</code> <br><code><a href="activation_threshold.html">activation_threshold</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -133,7 +133,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/activation_relu6.html b/docs/dev/reference/activation_relu6.html
index d3ba37311..decb4e910 100644
--- a/docs/dev/reference/activation_relu6.html
+++ b/docs/dev/reference/activation_relu6.html
@@ -8,7 +8,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -78,7 +78,7 @@ <h2 id="value">Value<a class="anchor" aria-label="anchor" href="#value"></a></h2
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
     <div class="dont-index">
 <ul><li><p><a href="https://keras.io/api/layers/activations#relu6-function" class="external-link">https://keras.io/api/layers/activations#relu6-function</a></p></li>
-</ul><p>Other activations: <br><code><a href="activation_celu.html">activation_celu</a>()</code> <br><code><a href="activation_elu.html">activation_elu</a>()</code> <br><code><a href="activation_exponential.html">activation_exponential</a>()</code> <br><code><a href="activation_gelu.html">activation_gelu</a>()</code> <br><code><a href="activation_glu.html">activation_glu</a>()</code> <br><code><a href="activation_hard_shrink.html">activation_hard_shrink</a>()</code> <br><code><a href="activation_hard_sigmoid.html">activation_hard_sigmoid</a>()</code> <br><code><a href="activation_hard_tanh.html">activation_hard_tanh</a>()</code> <br><code><a href="activation_leaky_relu.html">activation_leaky_relu</a>()</code> <br><code><a href="activation_linear.html">activation_linear</a>()</code> <br><code><a href="activation_log_sigmoid.html">activation_log_sigmoid</a>()</code> <br><code><a href="activation_log_softmax.html">activation_log_softmax</a>()</code> <br><code><a href="activation_mish.html">activation_mish</a>()</code> <br><code><a href="activation_relu.html">activation_relu</a>()</code> <br><code><a href="activation_selu.html">activation_selu</a>()</code> <br><code><a href="activation_sigmoid.html">activation_sigmoid</a>()</code> <br><code><a href="activation_silu.html">activation_silu</a>()</code> <br><code><a href="activation_soft_shrink.html">activation_soft_shrink</a>()</code> <br><code><a href="activation_softmax.html">activation_softmax</a>()</code> <br><code><a href="activation_softplus.html">activation_softplus</a>()</code> <br><code><a href="activation_softsign.html">activation_softsign</a>()</code> <br><code><a href="activation_sparse_plus.html">activation_sparse_plus</a>()</code> <br><code><a href="activation_sparsemax.html">activation_sparsemax</a>()</code> <br><code><a href="activation_squareplus.html">activation_squareplus</a>()</code> <br><code><a href="activation_tanh.html">activation_tanh</a>()</code> <br><code><a href="activation_tanh_shrink.html">activation_tanh_shrink</a>()</code> <br><code><a href="activation_threshold.html">activation_threshold</a>()</code> <br></p></div>
+</ul><p>Other activations: <br><code><a href="activation_celu.html">activation_celu</a>()</code> <br><code><a href="activation_elu.html">activation_elu</a>()</code> <br><code><a href="activation_exponential.html">activation_exponential</a>()</code> <br><code><a href="activation_gelu.html">activation_gelu</a>()</code> <br><code><a href="activation_glu.html">activation_glu</a>()</code> <br><code><a href="activation_hard_shrink.html">activation_hard_shrink</a>()</code> <br><code><a href="activation_hard_sigmoid.html">activation_hard_sigmoid</a>()</code> <br><code><a href="activation_hard_tanh.html">activation_hard_tanh</a>()</code> <br><code><a href="activation_leaky_relu.html">activation_leaky_relu</a>()</code> <br><code><a href="activation_linear.html">activation_linear</a>()</code> <br><code><a href="activation_log_sigmoid.html">activation_log_sigmoid</a>()</code> <br><code><a href="activation_log_softmax.html">activation_log_softmax</a>()</code> <br><code><a href="activation_mish.html">activation_mish</a>()</code> <br><code><a href="activation_relu.html">activation_relu</a>()</code> <br><code><a href="activation_selu.html">activation_selu</a>()</code> <br><code><a href="activation_sigmoid.html">activation_sigmoid</a>()</code> <br><code><a href="activation_silu.html">activation_silu</a>()</code> <br><code><a href="activation_soft_shrink.html">activation_soft_shrink</a>()</code> <br><code><a href="activation_softmax.html">activation_softmax</a>()</code> <br><code><a href="activation_softplus.html">activation_softplus</a>()</code> <br><code><a href="activation_softsign.html">activation_softsign</a>()</code> <br><code><a href="activation_sparse_plus.html">activation_sparse_plus</a>()</code> <br><code><a href="activation_sparse_sigmoid.html">activation_sparse_sigmoid</a>()</code> <br><code><a href="activation_sparsemax.html">activation_sparsemax</a>()</code> <br><code><a href="activation_squareplus.html">activation_squareplus</a>()</code> <br><code><a href="activation_tanh.html">activation_tanh</a>()</code> <br><code><a href="activation_tanh_shrink.html">activation_tanh_shrink</a>()</code> <br><code><a href="activation_threshold.html">activation_threshold</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -90,7 +90,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/activation_selu.html b/docs/dev/reference/activation_selu.html
index 7afb0326c..3770599dd 100644
--- a/docs/dev/reference/activation_selu.html
+++ b/docs/dev/reference/activation_selu.html
@@ -38,7 +38,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -133,7 +133,7 @@ <h2 id="reference">Reference<a class="anchor" aria-label="anchor" href="#referen
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
     <div class="dont-index">
 <ul><li><p><a href="https://keras.io/api/layers/activations#selu-function" class="external-link">https://keras.io/api/layers/activations#selu-function</a></p></li>
-</ul><p>Other activations: <br><code><a href="activation_celu.html">activation_celu</a>()</code> <br><code><a href="activation_elu.html">activation_elu</a>()</code> <br><code><a href="activation_exponential.html">activation_exponential</a>()</code> <br><code><a href="activation_gelu.html">activation_gelu</a>()</code> <br><code><a href="activation_glu.html">activation_glu</a>()</code> <br><code><a href="activation_hard_shrink.html">activation_hard_shrink</a>()</code> <br><code><a href="activation_hard_sigmoid.html">activation_hard_sigmoid</a>()</code> <br><code><a href="activation_hard_tanh.html">activation_hard_tanh</a>()</code> <br><code><a href="activation_leaky_relu.html">activation_leaky_relu</a>()</code> <br><code><a href="activation_linear.html">activation_linear</a>()</code> <br><code><a href="activation_log_sigmoid.html">activation_log_sigmoid</a>()</code> <br><code><a href="activation_log_softmax.html">activation_log_softmax</a>()</code> <br><code><a href="activation_mish.html">activation_mish</a>()</code> <br><code><a href="activation_relu.html">activation_relu</a>()</code> <br><code><a href="activation_relu6.html">activation_relu6</a>()</code> <br><code><a href="activation_sigmoid.html">activation_sigmoid</a>()</code> <br><code><a href="activation_silu.html">activation_silu</a>()</code> <br><code><a href="activation_soft_shrink.html">activation_soft_shrink</a>()</code> <br><code><a href="activation_softmax.html">activation_softmax</a>()</code> <br><code><a href="activation_softplus.html">activation_softplus</a>()</code> <br><code><a href="activation_softsign.html">activation_softsign</a>()</code> <br><code><a href="activation_sparse_plus.html">activation_sparse_plus</a>()</code> <br><code><a href="activation_sparsemax.html">activation_sparsemax</a>()</code> <br><code><a href="activation_squareplus.html">activation_squareplus</a>()</code> <br><code><a href="activation_tanh.html">activation_tanh</a>()</code> <br><code><a href="activation_tanh_shrink.html">activation_tanh_shrink</a>()</code> <br><code><a href="activation_threshold.html">activation_threshold</a>()</code> <br></p></div>
+</ul><p>Other activations: <br><code><a href="activation_celu.html">activation_celu</a>()</code> <br><code><a href="activation_elu.html">activation_elu</a>()</code> <br><code><a href="activation_exponential.html">activation_exponential</a>()</code> <br><code><a href="activation_gelu.html">activation_gelu</a>()</code> <br><code><a href="activation_glu.html">activation_glu</a>()</code> <br><code><a href="activation_hard_shrink.html">activation_hard_shrink</a>()</code> <br><code><a href="activation_hard_sigmoid.html">activation_hard_sigmoid</a>()</code> <br><code><a href="activation_hard_tanh.html">activation_hard_tanh</a>()</code> <br><code><a href="activation_leaky_relu.html">activation_leaky_relu</a>()</code> <br><code><a href="activation_linear.html">activation_linear</a>()</code> <br><code><a href="activation_log_sigmoid.html">activation_log_sigmoid</a>()</code> <br><code><a href="activation_log_softmax.html">activation_log_softmax</a>()</code> <br><code><a href="activation_mish.html">activation_mish</a>()</code> <br><code><a href="activation_relu.html">activation_relu</a>()</code> <br><code><a href="activation_relu6.html">activation_relu6</a>()</code> <br><code><a href="activation_sigmoid.html">activation_sigmoid</a>()</code> <br><code><a href="activation_silu.html">activation_silu</a>()</code> <br><code><a href="activation_soft_shrink.html">activation_soft_shrink</a>()</code> <br><code><a href="activation_softmax.html">activation_softmax</a>()</code> <br><code><a href="activation_softplus.html">activation_softplus</a>()</code> <br><code><a href="activation_softsign.html">activation_softsign</a>()</code> <br><code><a href="activation_sparse_plus.html">activation_sparse_plus</a>()</code> <br><code><a href="activation_sparse_sigmoid.html">activation_sparse_sigmoid</a>()</code> <br><code><a href="activation_sparsemax.html">activation_sparsemax</a>()</code> <br><code><a href="activation_squareplus.html">activation_squareplus</a>()</code> <br><code><a href="activation_tanh.html">activation_tanh</a>()</code> <br><code><a href="activation_tanh_shrink.html">activation_tanh_shrink</a>()</code> <br><code><a href="activation_threshold.html">activation_threshold</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -145,7 +145,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/activation_sigmoid.html b/docs/dev/reference/activation_sigmoid.html
index dc64980d2..1a6c60557 100644
--- a/docs/dev/reference/activation_sigmoid.html
+++ b/docs/dev/reference/activation_sigmoid.html
@@ -20,7 +20,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -96,7 +96,7 @@ <h2 id="value">Value<a class="anchor" aria-label="anchor" href="#value"></a></h2
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
     <div class="dont-index">
 <ul><li><p><a href="https://keras.io/api/layers/activations#sigmoid-function" class="external-link">https://keras.io/api/layers/activations#sigmoid-function</a></p></li>
-</ul><p>Other activations: <br><code><a href="activation_celu.html">activation_celu</a>()</code> <br><code><a href="activation_elu.html">activation_elu</a>()</code> <br><code><a href="activation_exponential.html">activation_exponential</a>()</code> <br><code><a href="activation_gelu.html">activation_gelu</a>()</code> <br><code><a href="activation_glu.html">activation_glu</a>()</code> <br><code><a href="activation_hard_shrink.html">activation_hard_shrink</a>()</code> <br><code><a href="activation_hard_sigmoid.html">activation_hard_sigmoid</a>()</code> <br><code><a href="activation_hard_tanh.html">activation_hard_tanh</a>()</code> <br><code><a href="activation_leaky_relu.html">activation_leaky_relu</a>()</code> <br><code><a href="activation_linear.html">activation_linear</a>()</code> <br><code><a href="activation_log_sigmoid.html">activation_log_sigmoid</a>()</code> <br><code><a href="activation_log_softmax.html">activation_log_softmax</a>()</code> <br><code><a href="activation_mish.html">activation_mish</a>()</code> <br><code><a href="activation_relu.html">activation_relu</a>()</code> <br><code><a href="activation_relu6.html">activation_relu6</a>()</code> <br><code><a href="activation_selu.html">activation_selu</a>()</code> <br><code><a href="activation_silu.html">activation_silu</a>()</code> <br><code><a href="activation_soft_shrink.html">activation_soft_shrink</a>()</code> <br><code><a href="activation_softmax.html">activation_softmax</a>()</code> <br><code><a href="activation_softplus.html">activation_softplus</a>()</code> <br><code><a href="activation_softsign.html">activation_softsign</a>()</code> <br><code><a href="activation_sparse_plus.html">activation_sparse_plus</a>()</code> <br><code><a href="activation_sparsemax.html">activation_sparsemax</a>()</code> <br><code><a href="activation_squareplus.html">activation_squareplus</a>()</code> <br><code><a href="activation_tanh.html">activation_tanh</a>()</code> <br><code><a href="activation_tanh_shrink.html">activation_tanh_shrink</a>()</code> <br><code><a href="activation_threshold.html">activation_threshold</a>()</code> <br></p></div>
+</ul><p>Other activations: <br><code><a href="activation_celu.html">activation_celu</a>()</code> <br><code><a href="activation_elu.html">activation_elu</a>()</code> <br><code><a href="activation_exponential.html">activation_exponential</a>()</code> <br><code><a href="activation_gelu.html">activation_gelu</a>()</code> <br><code><a href="activation_glu.html">activation_glu</a>()</code> <br><code><a href="activation_hard_shrink.html">activation_hard_shrink</a>()</code> <br><code><a href="activation_hard_sigmoid.html">activation_hard_sigmoid</a>()</code> <br><code><a href="activation_hard_tanh.html">activation_hard_tanh</a>()</code> <br><code><a href="activation_leaky_relu.html">activation_leaky_relu</a>()</code> <br><code><a href="activation_linear.html">activation_linear</a>()</code> <br><code><a href="activation_log_sigmoid.html">activation_log_sigmoid</a>()</code> <br><code><a href="activation_log_softmax.html">activation_log_softmax</a>()</code> <br><code><a href="activation_mish.html">activation_mish</a>()</code> <br><code><a href="activation_relu.html">activation_relu</a>()</code> <br><code><a href="activation_relu6.html">activation_relu6</a>()</code> <br><code><a href="activation_selu.html">activation_selu</a>()</code> <br><code><a href="activation_silu.html">activation_silu</a>()</code> <br><code><a href="activation_soft_shrink.html">activation_soft_shrink</a>()</code> <br><code><a href="activation_softmax.html">activation_softmax</a>()</code> <br><code><a href="activation_softplus.html">activation_softplus</a>()</code> <br><code><a href="activation_softsign.html">activation_softsign</a>()</code> <br><code><a href="activation_sparse_plus.html">activation_sparse_plus</a>()</code> <br><code><a href="activation_sparse_sigmoid.html">activation_sparse_sigmoid</a>()</code> <br><code><a href="activation_sparsemax.html">activation_sparsemax</a>()</code> <br><code><a href="activation_squareplus.html">activation_squareplus</a>()</code> <br><code><a href="activation_tanh.html">activation_tanh</a>()</code> <br><code><a href="activation_tanh_shrink.html">activation_tanh_shrink</a>()</code> <br><code><a href="activation_threshold.html">activation_threshold</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -108,7 +108,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/activation_silu.html b/docs/dev/reference/activation_silu.html
index 377f5c6c3..f39c692c0 100644
--- a/docs/dev/reference/activation_silu.html
+++ b/docs/dev/reference/activation_silu.html
@@ -14,7 +14,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -92,7 +92,7 @@ <h2 id="reference">Reference<a class="anchor" aria-label="anchor" href="#referen
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
     <div class="dont-index">
 <ul><li><p><a href="https://keras.io/api/layers/activations#silu-function" class="external-link">https://keras.io/api/layers/activations#silu-function</a></p></li>
-</ul><p>Other activations: <br><code><a href="activation_celu.html">activation_celu</a>()</code> <br><code><a href="activation_elu.html">activation_elu</a>()</code> <br><code><a href="activation_exponential.html">activation_exponential</a>()</code> <br><code><a href="activation_gelu.html">activation_gelu</a>()</code> <br><code><a href="activation_glu.html">activation_glu</a>()</code> <br><code><a href="activation_hard_shrink.html">activation_hard_shrink</a>()</code> <br><code><a href="activation_hard_sigmoid.html">activation_hard_sigmoid</a>()</code> <br><code><a href="activation_hard_tanh.html">activation_hard_tanh</a>()</code> <br><code><a href="activation_leaky_relu.html">activation_leaky_relu</a>()</code> <br><code><a href="activation_linear.html">activation_linear</a>()</code> <br><code><a href="activation_log_sigmoid.html">activation_log_sigmoid</a>()</code> <br><code><a href="activation_log_softmax.html">activation_log_softmax</a>()</code> <br><code><a href="activation_mish.html">activation_mish</a>()</code> <br><code><a href="activation_relu.html">activation_relu</a>()</code> <br><code><a href="activation_relu6.html">activation_relu6</a>()</code> <br><code><a href="activation_selu.html">activation_selu</a>()</code> <br><code><a href="activation_sigmoid.html">activation_sigmoid</a>()</code> <br><code><a href="activation_soft_shrink.html">activation_soft_shrink</a>()</code> <br><code><a href="activation_softmax.html">activation_softmax</a>()</code> <br><code><a href="activation_softplus.html">activation_softplus</a>()</code> <br><code><a href="activation_softsign.html">activation_softsign</a>()</code> <br><code><a href="activation_sparse_plus.html">activation_sparse_plus</a>()</code> <br><code><a href="activation_sparsemax.html">activation_sparsemax</a>()</code> <br><code><a href="activation_squareplus.html">activation_squareplus</a>()</code> <br><code><a href="activation_tanh.html">activation_tanh</a>()</code> <br><code><a href="activation_tanh_shrink.html">activation_tanh_shrink</a>()</code> <br><code><a href="activation_threshold.html">activation_threshold</a>()</code> <br></p></div>
+</ul><p>Other activations: <br><code><a href="activation_celu.html">activation_celu</a>()</code> <br><code><a href="activation_elu.html">activation_elu</a>()</code> <br><code><a href="activation_exponential.html">activation_exponential</a>()</code> <br><code><a href="activation_gelu.html">activation_gelu</a>()</code> <br><code><a href="activation_glu.html">activation_glu</a>()</code> <br><code><a href="activation_hard_shrink.html">activation_hard_shrink</a>()</code> <br><code><a href="activation_hard_sigmoid.html">activation_hard_sigmoid</a>()</code> <br><code><a href="activation_hard_tanh.html">activation_hard_tanh</a>()</code> <br><code><a href="activation_leaky_relu.html">activation_leaky_relu</a>()</code> <br><code><a href="activation_linear.html">activation_linear</a>()</code> <br><code><a href="activation_log_sigmoid.html">activation_log_sigmoid</a>()</code> <br><code><a href="activation_log_softmax.html">activation_log_softmax</a>()</code> <br><code><a href="activation_mish.html">activation_mish</a>()</code> <br><code><a href="activation_relu.html">activation_relu</a>()</code> <br><code><a href="activation_relu6.html">activation_relu6</a>()</code> <br><code><a href="activation_selu.html">activation_selu</a>()</code> <br><code><a href="activation_sigmoid.html">activation_sigmoid</a>()</code> <br><code><a href="activation_soft_shrink.html">activation_soft_shrink</a>()</code> <br><code><a href="activation_softmax.html">activation_softmax</a>()</code> <br><code><a href="activation_softplus.html">activation_softplus</a>()</code> <br><code><a href="activation_softsign.html">activation_softsign</a>()</code> <br><code><a href="activation_sparse_plus.html">activation_sparse_plus</a>()</code> <br><code><a href="activation_sparse_sigmoid.html">activation_sparse_sigmoid</a>()</code> <br><code><a href="activation_sparsemax.html">activation_sparsemax</a>()</code> <br><code><a href="activation_squareplus.html">activation_squareplus</a>()</code> <br><code><a href="activation_tanh.html">activation_tanh</a>()</code> <br><code><a href="activation_tanh_shrink.html">activation_tanh_shrink</a>()</code> <br><code><a href="activation_threshold.html">activation_threshold</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -104,7 +104,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/activation_soft_shrink.html b/docs/dev/reference/activation_soft_shrink.html
index 7fcea74a8..e5e6fb785 100644
--- a/docs/dev/reference/activation_soft_shrink.html
+++ b/docs/dev/reference/activation_soft_shrink.html
@@ -14,7 +14,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -89,7 +89,7 @@ <h2 id="value">Value<a class="anchor" aria-label="anchor" href="#value"></a></h2
     </div>
     <div class="section level2">
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
-    <div class="dont-index"><p>Other activations: <br><code><a href="activation_celu.html">activation_celu</a>()</code> <br><code><a href="activation_elu.html">activation_elu</a>()</code> <br><code><a href="activation_exponential.html">activation_exponential</a>()</code> <br><code><a href="activation_gelu.html">activation_gelu</a>()</code> <br><code><a href="activation_glu.html">activation_glu</a>()</code> <br><code><a href="activation_hard_shrink.html">activation_hard_shrink</a>()</code> <br><code><a href="activation_hard_sigmoid.html">activation_hard_sigmoid</a>()</code> <br><code><a href="activation_hard_tanh.html">activation_hard_tanh</a>()</code> <br><code><a href="activation_leaky_relu.html">activation_leaky_relu</a>()</code> <br><code><a href="activation_linear.html">activation_linear</a>()</code> <br><code><a href="activation_log_sigmoid.html">activation_log_sigmoid</a>()</code> <br><code><a href="activation_log_softmax.html">activation_log_softmax</a>()</code> <br><code><a href="activation_mish.html">activation_mish</a>()</code> <br><code><a href="activation_relu.html">activation_relu</a>()</code> <br><code><a href="activation_relu6.html">activation_relu6</a>()</code> <br><code><a href="activation_selu.html">activation_selu</a>()</code> <br><code><a href="activation_sigmoid.html">activation_sigmoid</a>()</code> <br><code><a href="activation_silu.html">activation_silu</a>()</code> <br><code><a href="activation_softmax.html">activation_softmax</a>()</code> <br><code><a href="activation_softplus.html">activation_softplus</a>()</code> <br><code><a href="activation_softsign.html">activation_softsign</a>()</code> <br><code><a href="activation_sparse_plus.html">activation_sparse_plus</a>()</code> <br><code><a href="activation_sparsemax.html">activation_sparsemax</a>()</code> <br><code><a href="activation_squareplus.html">activation_squareplus</a>()</code> <br><code><a href="activation_tanh.html">activation_tanh</a>()</code> <br><code><a href="activation_tanh_shrink.html">activation_tanh_shrink</a>()</code> <br><code><a href="activation_threshold.html">activation_threshold</a>()</code> <br></p></div>
+    <div class="dont-index"><p>Other activations: <br><code><a href="activation_celu.html">activation_celu</a>()</code> <br><code><a href="activation_elu.html">activation_elu</a>()</code> <br><code><a href="activation_exponential.html">activation_exponential</a>()</code> <br><code><a href="activation_gelu.html">activation_gelu</a>()</code> <br><code><a href="activation_glu.html">activation_glu</a>()</code> <br><code><a href="activation_hard_shrink.html">activation_hard_shrink</a>()</code> <br><code><a href="activation_hard_sigmoid.html">activation_hard_sigmoid</a>()</code> <br><code><a href="activation_hard_tanh.html">activation_hard_tanh</a>()</code> <br><code><a href="activation_leaky_relu.html">activation_leaky_relu</a>()</code> <br><code><a href="activation_linear.html">activation_linear</a>()</code> <br><code><a href="activation_log_sigmoid.html">activation_log_sigmoid</a>()</code> <br><code><a href="activation_log_softmax.html">activation_log_softmax</a>()</code> <br><code><a href="activation_mish.html">activation_mish</a>()</code> <br><code><a href="activation_relu.html">activation_relu</a>()</code> <br><code><a href="activation_relu6.html">activation_relu6</a>()</code> <br><code><a href="activation_selu.html">activation_selu</a>()</code> <br><code><a href="activation_sigmoid.html">activation_sigmoid</a>()</code> <br><code><a href="activation_silu.html">activation_silu</a>()</code> <br><code><a href="activation_softmax.html">activation_softmax</a>()</code> <br><code><a href="activation_softplus.html">activation_softplus</a>()</code> <br><code><a href="activation_softsign.html">activation_softsign</a>()</code> <br><code><a href="activation_sparse_plus.html">activation_sparse_plus</a>()</code> <br><code><a href="activation_sparse_sigmoid.html">activation_sparse_sigmoid</a>()</code> <br><code><a href="activation_sparsemax.html">activation_sparsemax</a>()</code> <br><code><a href="activation_squareplus.html">activation_squareplus</a>()</code> <br><code><a href="activation_tanh.html">activation_tanh</a>()</code> <br><code><a href="activation_tanh_shrink.html">activation_tanh_shrink</a>()</code> <br><code><a href="activation_threshold.html">activation_threshold</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -101,7 +101,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/activation_softmax.html b/docs/dev/reference/activation_softmax.html
index 0b40447bd..a03230ed8 100644
--- a/docs/dev/reference/activation_softmax.html
+++ b/docs/dev/reference/activation_softmax.html
@@ -26,7 +26,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -109,7 +109,7 @@ <h2 id="value">Value<a class="anchor" aria-label="anchor" href="#value"></a></h2
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
     <div class="dont-index">
 <ul><li><p><a href="https://keras.io/api/layers/activations#softmax-function" class="external-link">https://keras.io/api/layers/activations#softmax-function</a></p></li>
-</ul><p>Other activations: <br><code><a href="activation_celu.html">activation_celu</a>()</code> <br><code><a href="activation_elu.html">activation_elu</a>()</code> <br><code><a href="activation_exponential.html">activation_exponential</a>()</code> <br><code><a href="activation_gelu.html">activation_gelu</a>()</code> <br><code><a href="activation_glu.html">activation_glu</a>()</code> <br><code><a href="activation_hard_shrink.html">activation_hard_shrink</a>()</code> <br><code><a href="activation_hard_sigmoid.html">activation_hard_sigmoid</a>()</code> <br><code><a href="activation_hard_tanh.html">activation_hard_tanh</a>()</code> <br><code><a href="activation_leaky_relu.html">activation_leaky_relu</a>()</code> <br><code><a href="activation_linear.html">activation_linear</a>()</code> <br><code><a href="activation_log_sigmoid.html">activation_log_sigmoid</a>()</code> <br><code><a href="activation_log_softmax.html">activation_log_softmax</a>()</code> <br><code><a href="activation_mish.html">activation_mish</a>()</code> <br><code><a href="activation_relu.html">activation_relu</a>()</code> <br><code><a href="activation_relu6.html">activation_relu6</a>()</code> <br><code><a href="activation_selu.html">activation_selu</a>()</code> <br><code><a href="activation_sigmoid.html">activation_sigmoid</a>()</code> <br><code><a href="activation_silu.html">activation_silu</a>()</code> <br><code><a href="activation_soft_shrink.html">activation_soft_shrink</a>()</code> <br><code><a href="activation_softplus.html">activation_softplus</a>()</code> <br><code><a href="activation_softsign.html">activation_softsign</a>()</code> <br><code><a href="activation_sparse_plus.html">activation_sparse_plus</a>()</code> <br><code><a href="activation_sparsemax.html">activation_sparsemax</a>()</code> <br><code><a href="activation_squareplus.html">activation_squareplus</a>()</code> <br><code><a href="activation_tanh.html">activation_tanh</a>()</code> <br><code><a href="activation_tanh_shrink.html">activation_tanh_shrink</a>()</code> <br><code><a href="activation_threshold.html">activation_threshold</a>()</code> <br></p></div>
+</ul><p>Other activations: <br><code><a href="activation_celu.html">activation_celu</a>()</code> <br><code><a href="activation_elu.html">activation_elu</a>()</code> <br><code><a href="activation_exponential.html">activation_exponential</a>()</code> <br><code><a href="activation_gelu.html">activation_gelu</a>()</code> <br><code><a href="activation_glu.html">activation_glu</a>()</code> <br><code><a href="activation_hard_shrink.html">activation_hard_shrink</a>()</code> <br><code><a href="activation_hard_sigmoid.html">activation_hard_sigmoid</a>()</code> <br><code><a href="activation_hard_tanh.html">activation_hard_tanh</a>()</code> <br><code><a href="activation_leaky_relu.html">activation_leaky_relu</a>()</code> <br><code><a href="activation_linear.html">activation_linear</a>()</code> <br><code><a href="activation_log_sigmoid.html">activation_log_sigmoid</a>()</code> <br><code><a href="activation_log_softmax.html">activation_log_softmax</a>()</code> <br><code><a href="activation_mish.html">activation_mish</a>()</code> <br><code><a href="activation_relu.html">activation_relu</a>()</code> <br><code><a href="activation_relu6.html">activation_relu6</a>()</code> <br><code><a href="activation_selu.html">activation_selu</a>()</code> <br><code><a href="activation_sigmoid.html">activation_sigmoid</a>()</code> <br><code><a href="activation_silu.html">activation_silu</a>()</code> <br><code><a href="activation_soft_shrink.html">activation_soft_shrink</a>()</code> <br><code><a href="activation_softplus.html">activation_softplus</a>()</code> <br><code><a href="activation_softsign.html">activation_softsign</a>()</code> <br><code><a href="activation_sparse_plus.html">activation_sparse_plus</a>()</code> <br><code><a href="activation_sparse_sigmoid.html">activation_sparse_sigmoid</a>()</code> <br><code><a href="activation_sparsemax.html">activation_sparsemax</a>()</code> <br><code><a href="activation_squareplus.html">activation_squareplus</a>()</code> <br><code><a href="activation_tanh.html">activation_tanh</a>()</code> <br><code><a href="activation_tanh_shrink.html">activation_tanh_shrink</a>()</code> <br><code><a href="activation_threshold.html">activation_threshold</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -121,7 +121,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/activation_softplus.html b/docs/dev/reference/activation_softplus.html
index 58276412e..03cac33ae 100644
--- a/docs/dev/reference/activation_softplus.html
+++ b/docs/dev/reference/activation_softplus.html
@@ -8,7 +8,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -78,7 +78,7 @@ <h2 id="value">Value<a class="anchor" aria-label="anchor" href="#value"></a></h2
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
     <div class="dont-index">
 <ul><li><p><a href="https://keras.io/api/layers/activations#softplus-function" class="external-link">https://keras.io/api/layers/activations#softplus-function</a></p></li>
-</ul><p>Other activations: <br><code><a href="activation_celu.html">activation_celu</a>()</code> <br><code><a href="activation_elu.html">activation_elu</a>()</code> <br><code><a href="activation_exponential.html">activation_exponential</a>()</code> <br><code><a href="activation_gelu.html">activation_gelu</a>()</code> <br><code><a href="activation_glu.html">activation_glu</a>()</code> <br><code><a href="activation_hard_shrink.html">activation_hard_shrink</a>()</code> <br><code><a href="activation_hard_sigmoid.html">activation_hard_sigmoid</a>()</code> <br><code><a href="activation_hard_tanh.html">activation_hard_tanh</a>()</code> <br><code><a href="activation_leaky_relu.html">activation_leaky_relu</a>()</code> <br><code><a href="activation_linear.html">activation_linear</a>()</code> <br><code><a href="activation_log_sigmoid.html">activation_log_sigmoid</a>()</code> <br><code><a href="activation_log_softmax.html">activation_log_softmax</a>()</code> <br><code><a href="activation_mish.html">activation_mish</a>()</code> <br><code><a href="activation_relu.html">activation_relu</a>()</code> <br><code><a href="activation_relu6.html">activation_relu6</a>()</code> <br><code><a href="activation_selu.html">activation_selu</a>()</code> <br><code><a href="activation_sigmoid.html">activation_sigmoid</a>()</code> <br><code><a href="activation_silu.html">activation_silu</a>()</code> <br><code><a href="activation_soft_shrink.html">activation_soft_shrink</a>()</code> <br><code><a href="activation_softmax.html">activation_softmax</a>()</code> <br><code><a href="activation_softsign.html">activation_softsign</a>()</code> <br><code><a href="activation_sparse_plus.html">activation_sparse_plus</a>()</code> <br><code><a href="activation_sparsemax.html">activation_sparsemax</a>()</code> <br><code><a href="activation_squareplus.html">activation_squareplus</a>()</code> <br><code><a href="activation_tanh.html">activation_tanh</a>()</code> <br><code><a href="activation_tanh_shrink.html">activation_tanh_shrink</a>()</code> <br><code><a href="activation_threshold.html">activation_threshold</a>()</code> <br></p></div>
+</ul><p>Other activations: <br><code><a href="activation_celu.html">activation_celu</a>()</code> <br><code><a href="activation_elu.html">activation_elu</a>()</code> <br><code><a href="activation_exponential.html">activation_exponential</a>()</code> <br><code><a href="activation_gelu.html">activation_gelu</a>()</code> <br><code><a href="activation_glu.html">activation_glu</a>()</code> <br><code><a href="activation_hard_shrink.html">activation_hard_shrink</a>()</code> <br><code><a href="activation_hard_sigmoid.html">activation_hard_sigmoid</a>()</code> <br><code><a href="activation_hard_tanh.html">activation_hard_tanh</a>()</code> <br><code><a href="activation_leaky_relu.html">activation_leaky_relu</a>()</code> <br><code><a href="activation_linear.html">activation_linear</a>()</code> <br><code><a href="activation_log_sigmoid.html">activation_log_sigmoid</a>()</code> <br><code><a href="activation_log_softmax.html">activation_log_softmax</a>()</code> <br><code><a href="activation_mish.html">activation_mish</a>()</code> <br><code><a href="activation_relu.html">activation_relu</a>()</code> <br><code><a href="activation_relu6.html">activation_relu6</a>()</code> <br><code><a href="activation_selu.html">activation_selu</a>()</code> <br><code><a href="activation_sigmoid.html">activation_sigmoid</a>()</code> <br><code><a href="activation_silu.html">activation_silu</a>()</code> <br><code><a href="activation_soft_shrink.html">activation_soft_shrink</a>()</code> <br><code><a href="activation_softmax.html">activation_softmax</a>()</code> <br><code><a href="activation_softsign.html">activation_softsign</a>()</code> <br><code><a href="activation_sparse_plus.html">activation_sparse_plus</a>()</code> <br><code><a href="activation_sparse_sigmoid.html">activation_sparse_sigmoid</a>()</code> <br><code><a href="activation_sparsemax.html">activation_sparsemax</a>()</code> <br><code><a href="activation_squareplus.html">activation_squareplus</a>()</code> <br><code><a href="activation_tanh.html">activation_tanh</a>()</code> <br><code><a href="activation_tanh_shrink.html">activation_tanh_shrink</a>()</code> <br><code><a href="activation_threshold.html">activation_threshold</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -90,7 +90,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/activation_softsign.html b/docs/dev/reference/activation_softsign.html
index 92125c056..50e78e167 100644
--- a/docs/dev/reference/activation_softsign.html
+++ b/docs/dev/reference/activation_softsign.html
@@ -8,7 +8,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -78,7 +78,7 @@ <h2 id="value">Value<a class="anchor" aria-label="anchor" href="#value"></a></h2
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
     <div class="dont-index">
 <ul><li><p><a href="https://keras.io/api/layers/activations#softsign-function" class="external-link">https://keras.io/api/layers/activations#softsign-function</a></p></li>
-</ul><p>Other activations: <br><code><a href="activation_celu.html">activation_celu</a>()</code> <br><code><a href="activation_elu.html">activation_elu</a>()</code> <br><code><a href="activation_exponential.html">activation_exponential</a>()</code> <br><code><a href="activation_gelu.html">activation_gelu</a>()</code> <br><code><a href="activation_glu.html">activation_glu</a>()</code> <br><code><a href="activation_hard_shrink.html">activation_hard_shrink</a>()</code> <br><code><a href="activation_hard_sigmoid.html">activation_hard_sigmoid</a>()</code> <br><code><a href="activation_hard_tanh.html">activation_hard_tanh</a>()</code> <br><code><a href="activation_leaky_relu.html">activation_leaky_relu</a>()</code> <br><code><a href="activation_linear.html">activation_linear</a>()</code> <br><code><a href="activation_log_sigmoid.html">activation_log_sigmoid</a>()</code> <br><code><a href="activation_log_softmax.html">activation_log_softmax</a>()</code> <br><code><a href="activation_mish.html">activation_mish</a>()</code> <br><code><a href="activation_relu.html">activation_relu</a>()</code> <br><code><a href="activation_relu6.html">activation_relu6</a>()</code> <br><code><a href="activation_selu.html">activation_selu</a>()</code> <br><code><a href="activation_sigmoid.html">activation_sigmoid</a>()</code> <br><code><a href="activation_silu.html">activation_silu</a>()</code> <br><code><a href="activation_soft_shrink.html">activation_soft_shrink</a>()</code> <br><code><a href="activation_softmax.html">activation_softmax</a>()</code> <br><code><a href="activation_softplus.html">activation_softplus</a>()</code> <br><code><a href="activation_sparse_plus.html">activation_sparse_plus</a>()</code> <br><code><a href="activation_sparsemax.html">activation_sparsemax</a>()</code> <br><code><a href="activation_squareplus.html">activation_squareplus</a>()</code> <br><code><a href="activation_tanh.html">activation_tanh</a>()</code> <br><code><a href="activation_tanh_shrink.html">activation_tanh_shrink</a>()</code> <br><code><a href="activation_threshold.html">activation_threshold</a>()</code> <br></p></div>
+</ul><p>Other activations: <br><code><a href="activation_celu.html">activation_celu</a>()</code> <br><code><a href="activation_elu.html">activation_elu</a>()</code> <br><code><a href="activation_exponential.html">activation_exponential</a>()</code> <br><code><a href="activation_gelu.html">activation_gelu</a>()</code> <br><code><a href="activation_glu.html">activation_glu</a>()</code> <br><code><a href="activation_hard_shrink.html">activation_hard_shrink</a>()</code> <br><code><a href="activation_hard_sigmoid.html">activation_hard_sigmoid</a>()</code> <br><code><a href="activation_hard_tanh.html">activation_hard_tanh</a>()</code> <br><code><a href="activation_leaky_relu.html">activation_leaky_relu</a>()</code> <br><code><a href="activation_linear.html">activation_linear</a>()</code> <br><code><a href="activation_log_sigmoid.html">activation_log_sigmoid</a>()</code> <br><code><a href="activation_log_softmax.html">activation_log_softmax</a>()</code> <br><code><a href="activation_mish.html">activation_mish</a>()</code> <br><code><a href="activation_relu.html">activation_relu</a>()</code> <br><code><a href="activation_relu6.html">activation_relu6</a>()</code> <br><code><a href="activation_selu.html">activation_selu</a>()</code> <br><code><a href="activation_sigmoid.html">activation_sigmoid</a>()</code> <br><code><a href="activation_silu.html">activation_silu</a>()</code> <br><code><a href="activation_soft_shrink.html">activation_soft_shrink</a>()</code> <br><code><a href="activation_softmax.html">activation_softmax</a>()</code> <br><code><a href="activation_softplus.html">activation_softplus</a>()</code> <br><code><a href="activation_sparse_plus.html">activation_sparse_plus</a>()</code> <br><code><a href="activation_sparse_sigmoid.html">activation_sparse_sigmoid</a>()</code> <br><code><a href="activation_sparsemax.html">activation_sparsemax</a>()</code> <br><code><a href="activation_squareplus.html">activation_squareplus</a>()</code> <br><code><a href="activation_tanh.html">activation_tanh</a>()</code> <br><code><a href="activation_tanh_shrink.html">activation_tanh_shrink</a>()</code> <br><code><a href="activation_threshold.html">activation_threshold</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -90,7 +90,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/activation_sparse_plus.html b/docs/dev/reference/activation_sparse_plus.html
index 412c587ad..564bf6a2c 100644
--- a/docs/dev/reference/activation_sparse_plus.html
+++ b/docs/dev/reference/activation_sparse_plus.html
@@ -14,7 +14,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -85,7 +85,7 @@ <h2 id="value">Value<a class="anchor" aria-label="anchor" href="#value"></a></h2
     </div>
     <div class="section level2">
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
-    <div class="dont-index"><p>Other activations: <br><code><a href="activation_celu.html">activation_celu</a>()</code> <br><code><a href="activation_elu.html">activation_elu</a>()</code> <br><code><a href="activation_exponential.html">activation_exponential</a>()</code> <br><code><a href="activation_gelu.html">activation_gelu</a>()</code> <br><code><a href="activation_glu.html">activation_glu</a>()</code> <br><code><a href="activation_hard_shrink.html">activation_hard_shrink</a>()</code> <br><code><a href="activation_hard_sigmoid.html">activation_hard_sigmoid</a>()</code> <br><code><a href="activation_hard_tanh.html">activation_hard_tanh</a>()</code> <br><code><a href="activation_leaky_relu.html">activation_leaky_relu</a>()</code> <br><code><a href="activation_linear.html">activation_linear</a>()</code> <br><code><a href="activation_log_sigmoid.html">activation_log_sigmoid</a>()</code> <br><code><a href="activation_log_softmax.html">activation_log_softmax</a>()</code> <br><code><a href="activation_mish.html">activation_mish</a>()</code> <br><code><a href="activation_relu.html">activation_relu</a>()</code> <br><code><a href="activation_relu6.html">activation_relu6</a>()</code> <br><code><a href="activation_selu.html">activation_selu</a>()</code> <br><code><a href="activation_sigmoid.html">activation_sigmoid</a>()</code> <br><code><a href="activation_silu.html">activation_silu</a>()</code> <br><code><a href="activation_soft_shrink.html">activation_soft_shrink</a>()</code> <br><code><a href="activation_softmax.html">activation_softmax</a>()</code> <br><code><a href="activation_softplus.html">activation_softplus</a>()</code> <br><code><a href="activation_softsign.html">activation_softsign</a>()</code> <br><code><a href="activation_sparsemax.html">activation_sparsemax</a>()</code> <br><code><a href="activation_squareplus.html">activation_squareplus</a>()</code> <br><code><a href="activation_tanh.html">activation_tanh</a>()</code> <br><code><a href="activation_tanh_shrink.html">activation_tanh_shrink</a>()</code> <br><code><a href="activation_threshold.html">activation_threshold</a>()</code> <br></p></div>
+    <div class="dont-index"><p>Other activations: <br><code><a href="activation_celu.html">activation_celu</a>()</code> <br><code><a href="activation_elu.html">activation_elu</a>()</code> <br><code><a href="activation_exponential.html">activation_exponential</a>()</code> <br><code><a href="activation_gelu.html">activation_gelu</a>()</code> <br><code><a href="activation_glu.html">activation_glu</a>()</code> <br><code><a href="activation_hard_shrink.html">activation_hard_shrink</a>()</code> <br><code><a href="activation_hard_sigmoid.html">activation_hard_sigmoid</a>()</code> <br><code><a href="activation_hard_tanh.html">activation_hard_tanh</a>()</code> <br><code><a href="activation_leaky_relu.html">activation_leaky_relu</a>()</code> <br><code><a href="activation_linear.html">activation_linear</a>()</code> <br><code><a href="activation_log_sigmoid.html">activation_log_sigmoid</a>()</code> <br><code><a href="activation_log_softmax.html">activation_log_softmax</a>()</code> <br><code><a href="activation_mish.html">activation_mish</a>()</code> <br><code><a href="activation_relu.html">activation_relu</a>()</code> <br><code><a href="activation_relu6.html">activation_relu6</a>()</code> <br><code><a href="activation_selu.html">activation_selu</a>()</code> <br><code><a href="activation_sigmoid.html">activation_sigmoid</a>()</code> <br><code><a href="activation_silu.html">activation_silu</a>()</code> <br><code><a href="activation_soft_shrink.html">activation_soft_shrink</a>()</code> <br><code><a href="activation_softmax.html">activation_softmax</a>()</code> <br><code><a href="activation_softplus.html">activation_softplus</a>()</code> <br><code><a href="activation_softsign.html">activation_softsign</a>()</code> <br><code><a href="activation_sparse_sigmoid.html">activation_sparse_sigmoid</a>()</code> <br><code><a href="activation_sparsemax.html">activation_sparsemax</a>()</code> <br><code><a href="activation_squareplus.html">activation_squareplus</a>()</code> <br><code><a href="activation_tanh.html">activation_tanh</a>()</code> <br><code><a href="activation_tanh_shrink.html">activation_tanh_shrink</a>()</code> <br><code><a href="activation_threshold.html">activation_threshold</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -97,7 +97,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/activation_sparse_sigmoid.html b/docs/dev/reference/activation_sparse_sigmoid.html
new file mode 100644
index 000000000..8e028181c
--- /dev/null
+++ b/docs/dev/reference/activation_sparse_sigmoid.html
@@ -0,0 +1,115 @@
+<!DOCTYPE html>
+<!-- Generated by pkgdown: do not edit by hand --><html lang="en"><head><meta http-equiv="Content-Type" content="text/html; charset=UTF-8"><meta charset="utf-8"><meta http-equiv="X-UA-Compatible" content="IE=edge"><meta name="viewport" content="width=device-width, initial-scale=1, shrink-to-fit=no"><title>Sparse sigmoid activation function. — activation_sparse_sigmoid • keras3</title><!-- favicons --><link rel="icon" type="image/png" sizes="96x96" href="../favicon-96x96.png"><link rel="icon" type="”image/svg+xml”" href="../favicon.svg"><link rel="apple-touch-icon" sizes="180x180" href="../apple-touch-icon.png"><link rel="icon" sizes="any" href="../favicon.ico"><link rel="manifest" href="../site.webmanifest"><script src="../deps/jquery-3.6.0/jquery-3.6.0.min.js"></script><meta name="viewport" content="width=device-width, initial-scale=1, shrink-to-fit=no"><link href="../deps/bootstrap-5.3.1/bootstrap.min.css" rel="stylesheet"><script src="../deps/bootstrap-5.3.1/bootstrap.bundle.min.js"></script><link href="../deps/font-awesome-6.5.2/css/all.min.css" rel="stylesheet"><link href="../deps/font-awesome-6.5.2/css/v4-shims.min.css" rel="stylesheet"><script src="../deps/headroom-0.11.0/headroom.min.js"></script><script src="../deps/headroom-0.11.0/jQuery.headroom.min.js"></script><script src="../deps/bootstrap-toc-1.0.1/bootstrap-toc.min.js"></script><script src="../deps/clipboard.js-2.0.11/clipboard.min.js"></script><script src="../deps/search-1.0.0/autocomplete.jquery.min.js"></script><script src="../deps/search-1.0.0/fuse.min.js"></script><script src="../deps/search-1.0.0/mark.min.js"></script><!-- pkgdown --><script src="../pkgdown.js"></script><link href="../extra.css" rel="stylesheet"><meta property="og:title" content="Sparse sigmoid activation function. — activation_sparse_sigmoid"><meta name="description" content="It is defined as
+f(x) = 0 for x &amp;lt;= -1,
+f(x) = 0.5 * (x + 1) for -1 &amp;lt; x &amp;lt; 1,
+f(x) = 1 for x &amp;gt;= 1."><meta property="og:description" content="It is defined as
+f(x) = 0 for x &amp;lt;= -1,
+f(x) = 0.5 * (x + 1) for -1 &amp;lt; x &amp;lt; 1,
+f(x) = 1 for x &amp;gt;= 1."><meta name="robots" content="noindex"><!-- Google Tag Manager --><script>(function(w,d,s,l,i){w[l]=w[l]||[];w[l].push({'gtm.start': new Date().getTime(),event:'gtm.js'});var f=d.getElementsByTagName(s)[0],j=d.createElement(s),dl=l!='dataLayer'?'&l='+l:'';j.async=true;j.src='https://www.googletagmanager.com/gtm.js?id='+i+dl;f.parentNode.insertBefore(j,f);})(window,document,'script','dataLayer','GTM-KHBDBW7');</script><!-- End Google Tag Manager --></head><body>
+    <a href="#main" class="visually-hidden-focusable">Skip to contents</a>
+    <!-- Google Tag Manager (noscript) --> <noscript><iframe src="https://www.googletagmanager.com/ns.html?id=GTM-KHBDBW7" height="0" width="0" style="display:none;visibility:hidden"></iframe></noscript> <!-- End Google Tag Manager (noscript) -->
+
+
+    <nav class="navbar navbar-expand-lg fixed-top bg-primary" data-bs-theme="inverse" aria-label="Site navigation"><div class="container">
+
+    <a class="navbar-brand me-2" href="../index.html">keras3</a>
+
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
+
+
+    <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
+      <span class="navbar-toggler-icon"></span>
+    </button>
+
+    <div id="navbar" class="collapse navbar-collapse ms-3">
+      <ul class="navbar-nav me-auto"><li class="nav-item"><a class="nav-link" href="../articles/getting_started.html">Getting Started</a></li>
+<li class="nav-item dropdown">
+  <button class="nav-link dropdown-toggle" type="button" id="dropdown-guides" data-bs-toggle="dropdown" aria-expanded="false" aria-haspopup="true">Guides</button>
+  <ul class="dropdown-menu" aria-labelledby="dropdown-guides"><li><h6 class="dropdown-header" data-toc-skip>Model definition</h6></li>
+    <li><a class="dropdown-item" href="../articles/sequential_model.html">Sequential Model</a></li>
+    <li><a class="dropdown-item" href="../articles/functional_api.html">Functional API</a></li>
+    <li><h6 class="dropdown-header" data-toc-skip>Extending and customizing</h6></li>
+    <li><a class="dropdown-item" href="../articles/training_with_built_in_methods.html">Training &amp; evaluation with the built-in methods</a></li>
+    <li><a class="dropdown-item" href="../articles/custom_train_step_in_tensorflow.html">Customizing `fit()` with Tensorflow</a></li>
+    <li><a class="dropdown-item" href="../articles/writing_your_own_callbacks.html">Writing your own callbacks</a></li>
+    <li><a class="dropdown-item" href="../articles/making_new_layers_and_models_via_subclassing.html">Making new layers and models via subclassing</a></li>
+    <li><a class="dropdown-item" href="../articles/writing_a_custom_training_loop_in_tensorflow.html">Writing a training loop from scratch in TensorFlow</a></li>
+    <li><a class="dropdown-item" href="../articles/serialization_and_saving.html">Serialization and Saving</a></li>
+    <li><h6 class="dropdown-header" data-toc-skip>Other topics</h6></li>
+    <li><a class="dropdown-item" href="../articles/transfer_learning.html">Transfer learning and fine tuning</a></li>
+    <li><a class="dropdown-item" href="../articles/distributed_training_with_tensorflow.html">Distributed training with TensorFlow</a></li>
+    <li><a class="dropdown-item" href="../articles/distribution.html">Distributed training with Jax</a></li>
+  </ul></li>
+<li class="nav-item"><a class="nav-link" href="../articles/examples/index.html">Examples</a></li>
+<li class="active nav-item"><a class="nav-link" href="../reference/index.html">Reference</a></li>
+<li class="nav-item"><a class="nav-link" href="../news/index.html">News</a></li>
+      </ul><ul class="navbar-nav"><li class="nav-item"><form class="form-inline" role="search">
+ <input class="form-control" type="search" name="search-input" id="search-input" autocomplete="off" aria-label="Search site" placeholder="Search for" data-search-index="../search.json"></form></li>
+<li class="nav-item"><a class="external-link nav-link" href="https://github.com/rstudio/keras3/" aria-label="GitHub"><span class="fa fab fa-github fa-lg"></span></a></li>
+      </ul></div>
+
+
+  </div>
+</nav><div class="container template-reference-topic">
+<div class="row">
+  <main id="main" class="col-md-9"><div class="page-header">
+
+      <h1>Sparse sigmoid activation function.</h1>
+      <small class="dont-index">Source: <a href="https://github.com/rstudio/keras3/blob/HEAD/R/activations.R" class="external-link"><code>R/activations.R</code></a></small>
+      <div class="d-none name"><code>activation_sparse_sigmoid.Rd</code></div>
+    </div>
+
+    <div class="ref-description section level2">
+    <p>It is defined as</p>
+<p><code>f(x) = 0</code> for <code>x &lt;= -1</code>,
+<code>f(x) = 0.5 * (x + 1)</code> for <code>-1 &lt; x &lt; 1</code>,
+<code>f(x) = 1</code> for <code>x &gt;= 1</code>.</p>
+    </div>
+
+    <div class="section level2">
+    <h2 id="ref-usage">Usage<a class="anchor" aria-label="anchor" href="#ref-usage"></a></h2>
+    <div class="sourceCode"><pre class="sourceCode r"><code><span><span class="fu">activation_sparse_sigmoid</span><span class="op">(</span><span class="va">x</span><span class="op">)</span></span></code></pre></div>
+    </div>
+
+    <div class="section level2">
+    <h2 id="arguments">Arguments<a class="anchor" aria-label="anchor" href="#arguments"></a></h2>
+
+
+<dl><dt id="arg-x">x<a class="anchor" aria-label="anchor" href="#arg-x"></a></dt>
+<dd><p>Input tensor.</p></dd>
+
+</dl></div>
+    <div class="section level2">
+    <h2 id="value">Value<a class="anchor" aria-label="anchor" href="#value"></a></h2>
+    <p>A tensor, the result from applying the activation to the input tensor <code>x</code>.</p>
+    </div>
+    <div class="section level2">
+    <h2 id="reference">Reference<a class="anchor" aria-label="anchor" href="#reference"></a></h2>
+
+<ul><li><p><a href="https://arxiv.org/pdf/1901.02324" class="external-link">M. Blondel, A. F. T. Martins, V. Niculae, 2019</a></p></li>
+</ul></div>
+    <div class="section level2">
+    <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
+    <div class="dont-index"><p>Other activations: <br><code><a href="activation_celu.html">activation_celu</a>()</code> <br><code><a href="activation_elu.html">activation_elu</a>()</code> <br><code><a href="activation_exponential.html">activation_exponential</a>()</code> <br><code><a href="activation_gelu.html">activation_gelu</a>()</code> <br><code><a href="activation_glu.html">activation_glu</a>()</code> <br><code><a href="activation_hard_shrink.html">activation_hard_shrink</a>()</code> <br><code><a href="activation_hard_sigmoid.html">activation_hard_sigmoid</a>()</code> <br><code><a href="activation_hard_tanh.html">activation_hard_tanh</a>()</code> <br><code><a href="activation_leaky_relu.html">activation_leaky_relu</a>()</code> <br><code><a href="activation_linear.html">activation_linear</a>()</code> <br><code><a href="activation_log_sigmoid.html">activation_log_sigmoid</a>()</code> <br><code><a href="activation_log_softmax.html">activation_log_softmax</a>()</code> <br><code><a href="activation_mish.html">activation_mish</a>()</code> <br><code><a href="activation_relu.html">activation_relu</a>()</code> <br><code><a href="activation_relu6.html">activation_relu6</a>()</code> <br><code><a href="activation_selu.html">activation_selu</a>()</code> <br><code><a href="activation_sigmoid.html">activation_sigmoid</a>()</code> <br><code><a href="activation_silu.html">activation_silu</a>()</code> <br><code><a href="activation_soft_shrink.html">activation_soft_shrink</a>()</code> <br><code><a href="activation_softmax.html">activation_softmax</a>()</code> <br><code><a href="activation_softplus.html">activation_softplus</a>()</code> <br><code><a href="activation_softsign.html">activation_softsign</a>()</code> <br><code><a href="activation_sparse_plus.html">activation_sparse_plus</a>()</code> <br><code><a href="activation_sparsemax.html">activation_sparsemax</a>()</code> <br><code><a href="activation_squareplus.html">activation_squareplus</a>()</code> <br><code><a href="activation_tanh.html">activation_tanh</a>()</code> <br><code><a href="activation_tanh_shrink.html">activation_tanh_shrink</a>()</code> <br><code><a href="activation_threshold.html">activation_threshold</a>()</code> <br></p></div>
+    </div>
+
+  </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
+    </nav></aside></div>
+
+
+    <footer><div class="pkgdown-footer-left">
+  <p>Developed by Tomasz Kalinowski, JJ Allaire, François Chollet, Posit Software, PBC, Google.</p>
+</div>
+
+<div class="pkgdown-footer-right">
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
+</div>
+
+    </footer></div>
+
+
+
+
+
+  </body></html>
+
diff --git a/docs/dev/reference/activation_sparsemax.html b/docs/dev/reference/activation_sparsemax.html
index 4b5f1d80e..3caa13d32 100644
--- a/docs/dev/reference/activation_sparsemax.html
+++ b/docs/dev/reference/activation_sparsemax.html
@@ -12,7 +12,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -92,7 +92,7 @@ <h2 id="reference">Reference<a class="anchor" aria-label="anchor" href="#referen
 </ul></div>
     <div class="section level2">
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
-    <div class="dont-index"><p>Other activations: <br><code><a href="activation_celu.html">activation_celu</a>()</code> <br><code><a href="activation_elu.html">activation_elu</a>()</code> <br><code><a href="activation_exponential.html">activation_exponential</a>()</code> <br><code><a href="activation_gelu.html">activation_gelu</a>()</code> <br><code><a href="activation_glu.html">activation_glu</a>()</code> <br><code><a href="activation_hard_shrink.html">activation_hard_shrink</a>()</code> <br><code><a href="activation_hard_sigmoid.html">activation_hard_sigmoid</a>()</code> <br><code><a href="activation_hard_tanh.html">activation_hard_tanh</a>()</code> <br><code><a href="activation_leaky_relu.html">activation_leaky_relu</a>()</code> <br><code><a href="activation_linear.html">activation_linear</a>()</code> <br><code><a href="activation_log_sigmoid.html">activation_log_sigmoid</a>()</code> <br><code><a href="activation_log_softmax.html">activation_log_softmax</a>()</code> <br><code><a href="activation_mish.html">activation_mish</a>()</code> <br><code><a href="activation_relu.html">activation_relu</a>()</code> <br><code><a href="activation_relu6.html">activation_relu6</a>()</code> <br><code><a href="activation_selu.html">activation_selu</a>()</code> <br><code><a href="activation_sigmoid.html">activation_sigmoid</a>()</code> <br><code><a href="activation_silu.html">activation_silu</a>()</code> <br><code><a href="activation_soft_shrink.html">activation_soft_shrink</a>()</code> <br><code><a href="activation_softmax.html">activation_softmax</a>()</code> <br><code><a href="activation_softplus.html">activation_softplus</a>()</code> <br><code><a href="activation_softsign.html">activation_softsign</a>()</code> <br><code><a href="activation_sparse_plus.html">activation_sparse_plus</a>()</code> <br><code><a href="activation_squareplus.html">activation_squareplus</a>()</code> <br><code><a href="activation_tanh.html">activation_tanh</a>()</code> <br><code><a href="activation_tanh_shrink.html">activation_tanh_shrink</a>()</code> <br><code><a href="activation_threshold.html">activation_threshold</a>()</code> <br></p></div>
+    <div class="dont-index"><p>Other activations: <br><code><a href="activation_celu.html">activation_celu</a>()</code> <br><code><a href="activation_elu.html">activation_elu</a>()</code> <br><code><a href="activation_exponential.html">activation_exponential</a>()</code> <br><code><a href="activation_gelu.html">activation_gelu</a>()</code> <br><code><a href="activation_glu.html">activation_glu</a>()</code> <br><code><a href="activation_hard_shrink.html">activation_hard_shrink</a>()</code> <br><code><a href="activation_hard_sigmoid.html">activation_hard_sigmoid</a>()</code> <br><code><a href="activation_hard_tanh.html">activation_hard_tanh</a>()</code> <br><code><a href="activation_leaky_relu.html">activation_leaky_relu</a>()</code> <br><code><a href="activation_linear.html">activation_linear</a>()</code> <br><code><a href="activation_log_sigmoid.html">activation_log_sigmoid</a>()</code> <br><code><a href="activation_log_softmax.html">activation_log_softmax</a>()</code> <br><code><a href="activation_mish.html">activation_mish</a>()</code> <br><code><a href="activation_relu.html">activation_relu</a>()</code> <br><code><a href="activation_relu6.html">activation_relu6</a>()</code> <br><code><a href="activation_selu.html">activation_selu</a>()</code> <br><code><a href="activation_sigmoid.html">activation_sigmoid</a>()</code> <br><code><a href="activation_silu.html">activation_silu</a>()</code> <br><code><a href="activation_soft_shrink.html">activation_soft_shrink</a>()</code> <br><code><a href="activation_softmax.html">activation_softmax</a>()</code> <br><code><a href="activation_softplus.html">activation_softplus</a>()</code> <br><code><a href="activation_softsign.html">activation_softsign</a>()</code> <br><code><a href="activation_sparse_plus.html">activation_sparse_plus</a>()</code> <br><code><a href="activation_sparse_sigmoid.html">activation_sparse_sigmoid</a>()</code> <br><code><a href="activation_squareplus.html">activation_squareplus</a>()</code> <br><code><a href="activation_tanh.html">activation_tanh</a>()</code> <br><code><a href="activation_tanh_shrink.html">activation_tanh_shrink</a>()</code> <br><code><a href="activation_threshold.html">activation_threshold</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -104,7 +104,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/activation_squareplus.html b/docs/dev/reference/activation_squareplus.html
index e97e8b6f5..84a4c45a1 100644
--- a/docs/dev/reference/activation_squareplus.html
+++ b/docs/dev/reference/activation_squareplus.html
@@ -12,7 +12,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -91,7 +91,7 @@ <h2 id="reference">Reference<a class="anchor" aria-label="anchor" href="#referen
 </ul></div>
     <div class="section level2">
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
-    <div class="dont-index"><p>Other activations: <br><code><a href="activation_celu.html">activation_celu</a>()</code> <br><code><a href="activation_elu.html">activation_elu</a>()</code> <br><code><a href="activation_exponential.html">activation_exponential</a>()</code> <br><code><a href="activation_gelu.html">activation_gelu</a>()</code> <br><code><a href="activation_glu.html">activation_glu</a>()</code> <br><code><a href="activation_hard_shrink.html">activation_hard_shrink</a>()</code> <br><code><a href="activation_hard_sigmoid.html">activation_hard_sigmoid</a>()</code> <br><code><a href="activation_hard_tanh.html">activation_hard_tanh</a>()</code> <br><code><a href="activation_leaky_relu.html">activation_leaky_relu</a>()</code> <br><code><a href="activation_linear.html">activation_linear</a>()</code> <br><code><a href="activation_log_sigmoid.html">activation_log_sigmoid</a>()</code> <br><code><a href="activation_log_softmax.html">activation_log_softmax</a>()</code> <br><code><a href="activation_mish.html">activation_mish</a>()</code> <br><code><a href="activation_relu.html">activation_relu</a>()</code> <br><code><a href="activation_relu6.html">activation_relu6</a>()</code> <br><code><a href="activation_selu.html">activation_selu</a>()</code> <br><code><a href="activation_sigmoid.html">activation_sigmoid</a>()</code> <br><code><a href="activation_silu.html">activation_silu</a>()</code> <br><code><a href="activation_soft_shrink.html">activation_soft_shrink</a>()</code> <br><code><a href="activation_softmax.html">activation_softmax</a>()</code> <br><code><a href="activation_softplus.html">activation_softplus</a>()</code> <br><code><a href="activation_softsign.html">activation_softsign</a>()</code> <br><code><a href="activation_sparse_plus.html">activation_sparse_plus</a>()</code> <br><code><a href="activation_sparsemax.html">activation_sparsemax</a>()</code> <br><code><a href="activation_tanh.html">activation_tanh</a>()</code> <br><code><a href="activation_tanh_shrink.html">activation_tanh_shrink</a>()</code> <br><code><a href="activation_threshold.html">activation_threshold</a>()</code> <br></p></div>
+    <div class="dont-index"><p>Other activations: <br><code><a href="activation_celu.html">activation_celu</a>()</code> <br><code><a href="activation_elu.html">activation_elu</a>()</code> <br><code><a href="activation_exponential.html">activation_exponential</a>()</code> <br><code><a href="activation_gelu.html">activation_gelu</a>()</code> <br><code><a href="activation_glu.html">activation_glu</a>()</code> <br><code><a href="activation_hard_shrink.html">activation_hard_shrink</a>()</code> <br><code><a href="activation_hard_sigmoid.html">activation_hard_sigmoid</a>()</code> <br><code><a href="activation_hard_tanh.html">activation_hard_tanh</a>()</code> <br><code><a href="activation_leaky_relu.html">activation_leaky_relu</a>()</code> <br><code><a href="activation_linear.html">activation_linear</a>()</code> <br><code><a href="activation_log_sigmoid.html">activation_log_sigmoid</a>()</code> <br><code><a href="activation_log_softmax.html">activation_log_softmax</a>()</code> <br><code><a href="activation_mish.html">activation_mish</a>()</code> <br><code><a href="activation_relu.html">activation_relu</a>()</code> <br><code><a href="activation_relu6.html">activation_relu6</a>()</code> <br><code><a href="activation_selu.html">activation_selu</a>()</code> <br><code><a href="activation_sigmoid.html">activation_sigmoid</a>()</code> <br><code><a href="activation_silu.html">activation_silu</a>()</code> <br><code><a href="activation_soft_shrink.html">activation_soft_shrink</a>()</code> <br><code><a href="activation_softmax.html">activation_softmax</a>()</code> <br><code><a href="activation_softplus.html">activation_softplus</a>()</code> <br><code><a href="activation_softsign.html">activation_softsign</a>()</code> <br><code><a href="activation_sparse_plus.html">activation_sparse_plus</a>()</code> <br><code><a href="activation_sparse_sigmoid.html">activation_sparse_sigmoid</a>()</code> <br><code><a href="activation_sparsemax.html">activation_sparsemax</a>()</code> <br><code><a href="activation_tanh.html">activation_tanh</a>()</code> <br><code><a href="activation_tanh_shrink.html">activation_tanh_shrink</a>()</code> <br><code><a href="activation_threshold.html">activation_threshold</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -103,7 +103,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/activation_tanh.html b/docs/dev/reference/activation_tanh.html
index 2ae67b9d1..ed427f7c4 100644
--- a/docs/dev/reference/activation_tanh.html
+++ b/docs/dev/reference/activation_tanh.html
@@ -12,7 +12,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -84,7 +84,7 @@ <h2 id="value">Value<a class="anchor" aria-label="anchor" href="#value"></a></h2
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
     <div class="dont-index">
 <ul><li><p><a href="https://keras.io/api/layers/activations#tanh-function" class="external-link">https://keras.io/api/layers/activations#tanh-function</a></p></li>
-</ul><p>Other activations: <br><code><a href="activation_celu.html">activation_celu</a>()</code> <br><code><a href="activation_elu.html">activation_elu</a>()</code> <br><code><a href="activation_exponential.html">activation_exponential</a>()</code> <br><code><a href="activation_gelu.html">activation_gelu</a>()</code> <br><code><a href="activation_glu.html">activation_glu</a>()</code> <br><code><a href="activation_hard_shrink.html">activation_hard_shrink</a>()</code> <br><code><a href="activation_hard_sigmoid.html">activation_hard_sigmoid</a>()</code> <br><code><a href="activation_hard_tanh.html">activation_hard_tanh</a>()</code> <br><code><a href="activation_leaky_relu.html">activation_leaky_relu</a>()</code> <br><code><a href="activation_linear.html">activation_linear</a>()</code> <br><code><a href="activation_log_sigmoid.html">activation_log_sigmoid</a>()</code> <br><code><a href="activation_log_softmax.html">activation_log_softmax</a>()</code> <br><code><a href="activation_mish.html">activation_mish</a>()</code> <br><code><a href="activation_relu.html">activation_relu</a>()</code> <br><code><a href="activation_relu6.html">activation_relu6</a>()</code> <br><code><a href="activation_selu.html">activation_selu</a>()</code> <br><code><a href="activation_sigmoid.html">activation_sigmoid</a>()</code> <br><code><a href="activation_silu.html">activation_silu</a>()</code> <br><code><a href="activation_soft_shrink.html">activation_soft_shrink</a>()</code> <br><code><a href="activation_softmax.html">activation_softmax</a>()</code> <br><code><a href="activation_softplus.html">activation_softplus</a>()</code> <br><code><a href="activation_softsign.html">activation_softsign</a>()</code> <br><code><a href="activation_sparse_plus.html">activation_sparse_plus</a>()</code> <br><code><a href="activation_sparsemax.html">activation_sparsemax</a>()</code> <br><code><a href="activation_squareplus.html">activation_squareplus</a>()</code> <br><code><a href="activation_tanh_shrink.html">activation_tanh_shrink</a>()</code> <br><code><a href="activation_threshold.html">activation_threshold</a>()</code> <br></p></div>
+</ul><p>Other activations: <br><code><a href="activation_celu.html">activation_celu</a>()</code> <br><code><a href="activation_elu.html">activation_elu</a>()</code> <br><code><a href="activation_exponential.html">activation_exponential</a>()</code> <br><code><a href="activation_gelu.html">activation_gelu</a>()</code> <br><code><a href="activation_glu.html">activation_glu</a>()</code> <br><code><a href="activation_hard_shrink.html">activation_hard_shrink</a>()</code> <br><code><a href="activation_hard_sigmoid.html">activation_hard_sigmoid</a>()</code> <br><code><a href="activation_hard_tanh.html">activation_hard_tanh</a>()</code> <br><code><a href="activation_leaky_relu.html">activation_leaky_relu</a>()</code> <br><code><a href="activation_linear.html">activation_linear</a>()</code> <br><code><a href="activation_log_sigmoid.html">activation_log_sigmoid</a>()</code> <br><code><a href="activation_log_softmax.html">activation_log_softmax</a>()</code> <br><code><a href="activation_mish.html">activation_mish</a>()</code> <br><code><a href="activation_relu.html">activation_relu</a>()</code> <br><code><a href="activation_relu6.html">activation_relu6</a>()</code> <br><code><a href="activation_selu.html">activation_selu</a>()</code> <br><code><a href="activation_sigmoid.html">activation_sigmoid</a>()</code> <br><code><a href="activation_silu.html">activation_silu</a>()</code> <br><code><a href="activation_soft_shrink.html">activation_soft_shrink</a>()</code> <br><code><a href="activation_softmax.html">activation_softmax</a>()</code> <br><code><a href="activation_softplus.html">activation_softplus</a>()</code> <br><code><a href="activation_softsign.html">activation_softsign</a>()</code> <br><code><a href="activation_sparse_plus.html">activation_sparse_plus</a>()</code> <br><code><a href="activation_sparse_sigmoid.html">activation_sparse_sigmoid</a>()</code> <br><code><a href="activation_sparsemax.html">activation_sparsemax</a>()</code> <br><code><a href="activation_squareplus.html">activation_squareplus</a>()</code> <br><code><a href="activation_tanh_shrink.html">activation_tanh_shrink</a>()</code> <br><code><a href="activation_threshold.html">activation_threshold</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -96,7 +96,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/activation_tanh_shrink.html b/docs/dev/reference/activation_tanh_shrink.html
index d789d192e..1631c366c 100644
--- a/docs/dev/reference/activation_tanh_shrink.html
+++ b/docs/dev/reference/activation_tanh_shrink.html
@@ -10,7 +10,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -79,7 +79,7 @@ <h2 id="value">Value<a class="anchor" aria-label="anchor" href="#value"></a></h2
     </div>
     <div class="section level2">
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
-    <div class="dont-index"><p>Other activations: <br><code><a href="activation_celu.html">activation_celu</a>()</code> <br><code><a href="activation_elu.html">activation_elu</a>()</code> <br><code><a href="activation_exponential.html">activation_exponential</a>()</code> <br><code><a href="activation_gelu.html">activation_gelu</a>()</code> <br><code><a href="activation_glu.html">activation_glu</a>()</code> <br><code><a href="activation_hard_shrink.html">activation_hard_shrink</a>()</code> <br><code><a href="activation_hard_sigmoid.html">activation_hard_sigmoid</a>()</code> <br><code><a href="activation_hard_tanh.html">activation_hard_tanh</a>()</code> <br><code><a href="activation_leaky_relu.html">activation_leaky_relu</a>()</code> <br><code><a href="activation_linear.html">activation_linear</a>()</code> <br><code><a href="activation_log_sigmoid.html">activation_log_sigmoid</a>()</code> <br><code><a href="activation_log_softmax.html">activation_log_softmax</a>()</code> <br><code><a href="activation_mish.html">activation_mish</a>()</code> <br><code><a href="activation_relu.html">activation_relu</a>()</code> <br><code><a href="activation_relu6.html">activation_relu6</a>()</code> <br><code><a href="activation_selu.html">activation_selu</a>()</code> <br><code><a href="activation_sigmoid.html">activation_sigmoid</a>()</code> <br><code><a href="activation_silu.html">activation_silu</a>()</code> <br><code><a href="activation_soft_shrink.html">activation_soft_shrink</a>()</code> <br><code><a href="activation_softmax.html">activation_softmax</a>()</code> <br><code><a href="activation_softplus.html">activation_softplus</a>()</code> <br><code><a href="activation_softsign.html">activation_softsign</a>()</code> <br><code><a href="activation_sparse_plus.html">activation_sparse_plus</a>()</code> <br><code><a href="activation_sparsemax.html">activation_sparsemax</a>()</code> <br><code><a href="activation_squareplus.html">activation_squareplus</a>()</code> <br><code><a href="activation_tanh.html">activation_tanh</a>()</code> <br><code><a href="activation_threshold.html">activation_threshold</a>()</code> <br></p></div>
+    <div class="dont-index"><p>Other activations: <br><code><a href="activation_celu.html">activation_celu</a>()</code> <br><code><a href="activation_elu.html">activation_elu</a>()</code> <br><code><a href="activation_exponential.html">activation_exponential</a>()</code> <br><code><a href="activation_gelu.html">activation_gelu</a>()</code> <br><code><a href="activation_glu.html">activation_glu</a>()</code> <br><code><a href="activation_hard_shrink.html">activation_hard_shrink</a>()</code> <br><code><a href="activation_hard_sigmoid.html">activation_hard_sigmoid</a>()</code> <br><code><a href="activation_hard_tanh.html">activation_hard_tanh</a>()</code> <br><code><a href="activation_leaky_relu.html">activation_leaky_relu</a>()</code> <br><code><a href="activation_linear.html">activation_linear</a>()</code> <br><code><a href="activation_log_sigmoid.html">activation_log_sigmoid</a>()</code> <br><code><a href="activation_log_softmax.html">activation_log_softmax</a>()</code> <br><code><a href="activation_mish.html">activation_mish</a>()</code> <br><code><a href="activation_relu.html">activation_relu</a>()</code> <br><code><a href="activation_relu6.html">activation_relu6</a>()</code> <br><code><a href="activation_selu.html">activation_selu</a>()</code> <br><code><a href="activation_sigmoid.html">activation_sigmoid</a>()</code> <br><code><a href="activation_silu.html">activation_silu</a>()</code> <br><code><a href="activation_soft_shrink.html">activation_soft_shrink</a>()</code> <br><code><a href="activation_softmax.html">activation_softmax</a>()</code> <br><code><a href="activation_softplus.html">activation_softplus</a>()</code> <br><code><a href="activation_softsign.html">activation_softsign</a>()</code> <br><code><a href="activation_sparse_plus.html">activation_sparse_plus</a>()</code> <br><code><a href="activation_sparse_sigmoid.html">activation_sparse_sigmoid</a>()</code> <br><code><a href="activation_sparsemax.html">activation_sparsemax</a>()</code> <br><code><a href="activation_squareplus.html">activation_squareplus</a>()</code> <br><code><a href="activation_tanh.html">activation_tanh</a>()</code> <br><code><a href="activation_threshold.html">activation_threshold</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -91,7 +91,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/activation_threshold.html b/docs/dev/reference/activation_threshold.html
index 6710fc61d..f877f7827 100644
--- a/docs/dev/reference/activation_threshold.html
+++ b/docs/dev/reference/activation_threshold.html
@@ -12,7 +12,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -90,7 +90,7 @@ <h2 id="value">Value<a class="anchor" aria-label="anchor" href="#value"></a></h2
     </div>
     <div class="section level2">
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
-    <div class="dont-index"><p>Other activations: <br><code><a href="activation_celu.html">activation_celu</a>()</code> <br><code><a href="activation_elu.html">activation_elu</a>()</code> <br><code><a href="activation_exponential.html">activation_exponential</a>()</code> <br><code><a href="activation_gelu.html">activation_gelu</a>()</code> <br><code><a href="activation_glu.html">activation_glu</a>()</code> <br><code><a href="activation_hard_shrink.html">activation_hard_shrink</a>()</code> <br><code><a href="activation_hard_sigmoid.html">activation_hard_sigmoid</a>()</code> <br><code><a href="activation_hard_tanh.html">activation_hard_tanh</a>()</code> <br><code><a href="activation_leaky_relu.html">activation_leaky_relu</a>()</code> <br><code><a href="activation_linear.html">activation_linear</a>()</code> <br><code><a href="activation_log_sigmoid.html">activation_log_sigmoid</a>()</code> <br><code><a href="activation_log_softmax.html">activation_log_softmax</a>()</code> <br><code><a href="activation_mish.html">activation_mish</a>()</code> <br><code><a href="activation_relu.html">activation_relu</a>()</code> <br><code><a href="activation_relu6.html">activation_relu6</a>()</code> <br><code><a href="activation_selu.html">activation_selu</a>()</code> <br><code><a href="activation_sigmoid.html">activation_sigmoid</a>()</code> <br><code><a href="activation_silu.html">activation_silu</a>()</code> <br><code><a href="activation_soft_shrink.html">activation_soft_shrink</a>()</code> <br><code><a href="activation_softmax.html">activation_softmax</a>()</code> <br><code><a href="activation_softplus.html">activation_softplus</a>()</code> <br><code><a href="activation_softsign.html">activation_softsign</a>()</code> <br><code><a href="activation_sparse_plus.html">activation_sparse_plus</a>()</code> <br><code><a href="activation_sparsemax.html">activation_sparsemax</a>()</code> <br><code><a href="activation_squareplus.html">activation_squareplus</a>()</code> <br><code><a href="activation_tanh.html">activation_tanh</a>()</code> <br><code><a href="activation_tanh_shrink.html">activation_tanh_shrink</a>()</code> <br></p></div>
+    <div class="dont-index"><p>Other activations: <br><code><a href="activation_celu.html">activation_celu</a>()</code> <br><code><a href="activation_elu.html">activation_elu</a>()</code> <br><code><a href="activation_exponential.html">activation_exponential</a>()</code> <br><code><a href="activation_gelu.html">activation_gelu</a>()</code> <br><code><a href="activation_glu.html">activation_glu</a>()</code> <br><code><a href="activation_hard_shrink.html">activation_hard_shrink</a>()</code> <br><code><a href="activation_hard_sigmoid.html">activation_hard_sigmoid</a>()</code> <br><code><a href="activation_hard_tanh.html">activation_hard_tanh</a>()</code> <br><code><a href="activation_leaky_relu.html">activation_leaky_relu</a>()</code> <br><code><a href="activation_linear.html">activation_linear</a>()</code> <br><code><a href="activation_log_sigmoid.html">activation_log_sigmoid</a>()</code> <br><code><a href="activation_log_softmax.html">activation_log_softmax</a>()</code> <br><code><a href="activation_mish.html">activation_mish</a>()</code> <br><code><a href="activation_relu.html">activation_relu</a>()</code> <br><code><a href="activation_relu6.html">activation_relu6</a>()</code> <br><code><a href="activation_selu.html">activation_selu</a>()</code> <br><code><a href="activation_sigmoid.html">activation_sigmoid</a>()</code> <br><code><a href="activation_silu.html">activation_silu</a>()</code> <br><code><a href="activation_soft_shrink.html">activation_soft_shrink</a>()</code> <br><code><a href="activation_softmax.html">activation_softmax</a>()</code> <br><code><a href="activation_softplus.html">activation_softplus</a>()</code> <br><code><a href="activation_softsign.html">activation_softsign</a>()</code> <br><code><a href="activation_sparse_plus.html">activation_sparse_plus</a>()</code> <br><code><a href="activation_sparse_sigmoid.html">activation_sparse_sigmoid</a>()</code> <br><code><a href="activation_sparsemax.html">activation_sparsemax</a>()</code> <br><code><a href="activation_squareplus.html">activation_squareplus</a>()</code> <br><code><a href="activation_tanh.html">activation_tanh</a>()</code> <br><code><a href="activation_tanh_shrink.html">activation_tanh_shrink</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -102,7 +102,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/active_property.html b/docs/dev/reference/active_property.html
index ccc35716b..18a3630bc 100644
--- a/docs/dev/reference/active_property.html
+++ b/docs/dev/reference/active_property.html
@@ -8,7 +8,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -94,7 +94,7 @@ <h2 id="example">Example<a class="anchor" aria-label="anchor" href="#example"></
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/adapt.html b/docs/dev/reference/adapt.html
index e0b36a215..3f3ec391b 100644
--- a/docs/dev/reference/adapt.html
+++ b/docs/dev/reference/adapt.html
@@ -8,7 +8,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -164,7 +164,7 @@ <h2 id="details">Details<a class="anchor" aria-label="anchor" href="#details"></
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/application_convnext_base.html b/docs/dev/reference/application_convnext_base.html
index c4e1702d4..021ae42cd 100644
--- a/docs/dev/reference/application_convnext_base.html
+++ b/docs/dev/reference/application_convnext_base.html
@@ -8,7 +8,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -183,7 +183,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/application_convnext_large.html b/docs/dev/reference/application_convnext_large.html
index ab660d870..9d805eb09 100644
--- a/docs/dev/reference/application_convnext_large.html
+++ b/docs/dev/reference/application_convnext_large.html
@@ -8,7 +8,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -183,7 +183,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/application_convnext_small.html b/docs/dev/reference/application_convnext_small.html
index 62e730298..a325c367c 100644
--- a/docs/dev/reference/application_convnext_small.html
+++ b/docs/dev/reference/application_convnext_small.html
@@ -8,7 +8,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -183,7 +183,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/application_convnext_tiny.html b/docs/dev/reference/application_convnext_tiny.html
index 4c8a77f4b..fe7280d1b 100644
--- a/docs/dev/reference/application_convnext_tiny.html
+++ b/docs/dev/reference/application_convnext_tiny.html
@@ -8,7 +8,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -183,7 +183,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/application_convnext_xlarge.html b/docs/dev/reference/application_convnext_xlarge.html
index 2456643ba..cd1c90a67 100644
--- a/docs/dev/reference/application_convnext_xlarge.html
+++ b/docs/dev/reference/application_convnext_xlarge.html
@@ -8,7 +8,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -183,7 +183,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/application_densenet121.html b/docs/dev/reference/application_densenet121.html
index 0df47b837..c01d147a2 100644
--- a/docs/dev/reference/application_densenet121.html
+++ b/docs/dev/reference/application_densenet121.html
@@ -8,7 +8,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -169,7 +169,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/application_densenet169.html b/docs/dev/reference/application_densenet169.html
index b0a60fce6..55ca2cb2a 100644
--- a/docs/dev/reference/application_densenet169.html
+++ b/docs/dev/reference/application_densenet169.html
@@ -8,7 +8,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -169,7 +169,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/application_densenet201.html b/docs/dev/reference/application_densenet201.html
index b031e2f5f..3bb02529b 100644
--- a/docs/dev/reference/application_densenet201.html
+++ b/docs/dev/reference/application_densenet201.html
@@ -8,7 +8,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -169,7 +169,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/application_efficientnet_b0.html b/docs/dev/reference/application_efficientnet_b0.html
index 82d0127c3..add6f16f8 100644
--- a/docs/dev/reference/application_efficientnet_b0.html
+++ b/docs/dev/reference/application_efficientnet_b0.html
@@ -8,7 +8,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -178,7 +178,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/application_efficientnet_b1.html b/docs/dev/reference/application_efficientnet_b1.html
index 8de9a359f..d95b46bc4 100644
--- a/docs/dev/reference/application_efficientnet_b1.html
+++ b/docs/dev/reference/application_efficientnet_b1.html
@@ -8,7 +8,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -178,7 +178,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/application_efficientnet_b2.html b/docs/dev/reference/application_efficientnet_b2.html
index 80ea371c1..fb030d422 100644
--- a/docs/dev/reference/application_efficientnet_b2.html
+++ b/docs/dev/reference/application_efficientnet_b2.html
@@ -8,7 +8,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -178,7 +178,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/application_efficientnet_b3.html b/docs/dev/reference/application_efficientnet_b3.html
index 3ec59b84b..352aa2c74 100644
--- a/docs/dev/reference/application_efficientnet_b3.html
+++ b/docs/dev/reference/application_efficientnet_b3.html
@@ -8,7 +8,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -178,7 +178,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/application_efficientnet_b4.html b/docs/dev/reference/application_efficientnet_b4.html
index 686fe291e..70e402c1a 100644
--- a/docs/dev/reference/application_efficientnet_b4.html
+++ b/docs/dev/reference/application_efficientnet_b4.html
@@ -8,7 +8,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -178,7 +178,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/application_efficientnet_b5.html b/docs/dev/reference/application_efficientnet_b5.html
index 0e99b2f5f..3501fbe77 100644
--- a/docs/dev/reference/application_efficientnet_b5.html
+++ b/docs/dev/reference/application_efficientnet_b5.html
@@ -8,7 +8,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -178,7 +178,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/application_efficientnet_b6.html b/docs/dev/reference/application_efficientnet_b6.html
index a5fe62381..ff6fc49dd 100644
--- a/docs/dev/reference/application_efficientnet_b6.html
+++ b/docs/dev/reference/application_efficientnet_b6.html
@@ -8,7 +8,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -178,7 +178,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/application_efficientnet_b7.html b/docs/dev/reference/application_efficientnet_b7.html
index cfc7a41db..c2bf98e70 100644
--- a/docs/dev/reference/application_efficientnet_b7.html
+++ b/docs/dev/reference/application_efficientnet_b7.html
@@ -8,7 +8,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -178,7 +178,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/application_efficientnet_v2b0.html b/docs/dev/reference/application_efficientnet_v2b0.html
index fcadd1a5f..b9f2486a3 100644
--- a/docs/dev/reference/application_efficientnet_v2b0.html
+++ b/docs/dev/reference/application_efficientnet_v2b0.html
@@ -8,7 +8,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -181,7 +181,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/application_efficientnet_v2b1.html b/docs/dev/reference/application_efficientnet_v2b1.html
index 9c1ea2dfd..c8d4733d5 100644
--- a/docs/dev/reference/application_efficientnet_v2b1.html
+++ b/docs/dev/reference/application_efficientnet_v2b1.html
@@ -8,7 +8,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -181,7 +181,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/application_efficientnet_v2b2.html b/docs/dev/reference/application_efficientnet_v2b2.html
index 783dc5450..4df49325e 100644
--- a/docs/dev/reference/application_efficientnet_v2b2.html
+++ b/docs/dev/reference/application_efficientnet_v2b2.html
@@ -8,7 +8,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -181,7 +181,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/application_efficientnet_v2b3.html b/docs/dev/reference/application_efficientnet_v2b3.html
index 1a9578ad2..3f50a11c9 100644
--- a/docs/dev/reference/application_efficientnet_v2b3.html
+++ b/docs/dev/reference/application_efficientnet_v2b3.html
@@ -8,7 +8,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -181,7 +181,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/application_efficientnet_v2l.html b/docs/dev/reference/application_efficientnet_v2l.html
index 13db4437e..554ec82cd 100644
--- a/docs/dev/reference/application_efficientnet_v2l.html
+++ b/docs/dev/reference/application_efficientnet_v2l.html
@@ -8,7 +8,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -181,7 +181,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/application_efficientnet_v2m.html b/docs/dev/reference/application_efficientnet_v2m.html
index 9bffbf073..670c107d8 100644
--- a/docs/dev/reference/application_efficientnet_v2m.html
+++ b/docs/dev/reference/application_efficientnet_v2m.html
@@ -8,7 +8,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -181,7 +181,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/application_efficientnet_v2s.html b/docs/dev/reference/application_efficientnet_v2s.html
index 35d272587..65250c856 100644
--- a/docs/dev/reference/application_efficientnet_v2s.html
+++ b/docs/dev/reference/application_efficientnet_v2s.html
@@ -8,7 +8,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -181,7 +181,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/application_inception_resnet_v2.html b/docs/dev/reference/application_inception_resnet_v2.html
index 9f5a1ae66..f0a7a7978 100644
--- a/docs/dev/reference/application_inception_resnet_v2.html
+++ b/docs/dev/reference/application_inception_resnet_v2.html
@@ -8,7 +8,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -175,7 +175,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/application_inception_v3.html b/docs/dev/reference/application_inception_v3.html
index a919db56a..93c6b40d2 100644
--- a/docs/dev/reference/application_inception_v3.html
+++ b/docs/dev/reference/application_inception_v3.html
@@ -8,7 +8,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -175,7 +175,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/application_mobilenet.html b/docs/dev/reference/application_mobilenet.html
index a28f5ebc0..15273fcf1 100644
--- a/docs/dev/reference/application_mobilenet.html
+++ b/docs/dev/reference/application_mobilenet.html
@@ -8,7 +8,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -196,7 +196,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/application_mobilenet_v2.html b/docs/dev/reference/application_mobilenet_v2.html
index 7cb199549..c262d3a01 100644
--- a/docs/dev/reference/application_mobilenet_v2.html
+++ b/docs/dev/reference/application_mobilenet_v2.html
@@ -20,7 +20,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -203,7 +203,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/application_mobilenet_v3_large.html b/docs/dev/reference/application_mobilenet_v3_large.html
index 0a46bcf44..f9f38d0b0 100644
--- a/docs/dev/reference/application_mobilenet_v3_large.html
+++ b/docs/dev/reference/application_mobilenet_v3_large.html
@@ -8,7 +8,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -229,7 +229,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/application_mobilenet_v3_small.html b/docs/dev/reference/application_mobilenet_v3_small.html
index 2bc023f4b..0c59613a8 100644
--- a/docs/dev/reference/application_mobilenet_v3_small.html
+++ b/docs/dev/reference/application_mobilenet_v3_small.html
@@ -8,7 +8,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -229,7 +229,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/application_nasnet_large.html b/docs/dev/reference/application_nasnet_large.html
index 26d15f6c8..c0e9980a8 100644
--- a/docs/dev/reference/application_nasnet_large.html
+++ b/docs/dev/reference/application_nasnet_large.html
@@ -8,7 +8,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -168,7 +168,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/application_nasnet_mobile.html b/docs/dev/reference/application_nasnet_mobile.html
index 49ac6119d..66f9e80bd 100644
--- a/docs/dev/reference/application_nasnet_mobile.html
+++ b/docs/dev/reference/application_nasnet_mobile.html
@@ -8,7 +8,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -168,7 +168,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/application_resnet101.html b/docs/dev/reference/application_resnet101.html
index b3f7df3fb..44ff443ef 100644
--- a/docs/dev/reference/application_resnet101.html
+++ b/docs/dev/reference/application_resnet101.html
@@ -8,7 +8,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -165,7 +165,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/application_resnet101_v2.html b/docs/dev/reference/application_resnet101_v2.html
index a19b46853..fe566b7de 100644
--- a/docs/dev/reference/application_resnet101_v2.html
+++ b/docs/dev/reference/application_resnet101_v2.html
@@ -8,7 +8,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -164,7 +164,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/application_resnet152.html b/docs/dev/reference/application_resnet152.html
index b9e93f3db..4f82c2eea 100644
--- a/docs/dev/reference/application_resnet152.html
+++ b/docs/dev/reference/application_resnet152.html
@@ -8,7 +8,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -165,7 +165,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/application_resnet152_v2.html b/docs/dev/reference/application_resnet152_v2.html
index ecd9f360d..d50ebab6b 100644
--- a/docs/dev/reference/application_resnet152_v2.html
+++ b/docs/dev/reference/application_resnet152_v2.html
@@ -8,7 +8,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -164,7 +164,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/application_resnet50.html b/docs/dev/reference/application_resnet50.html
index ad4b3c0e4..01a03e573 100644
--- a/docs/dev/reference/application_resnet50.html
+++ b/docs/dev/reference/application_resnet50.html
@@ -8,7 +8,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -165,7 +165,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/application_resnet50_v2.html b/docs/dev/reference/application_resnet50_v2.html
index 3d3ff46d8..4ba36919c 100644
--- a/docs/dev/reference/application_resnet50_v2.html
+++ b/docs/dev/reference/application_resnet50_v2.html
@@ -8,7 +8,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -164,7 +164,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/application_vgg16.html b/docs/dev/reference/application_vgg16.html
index dd666c298..91d83edcf 100644
--- a/docs/dev/reference/application_vgg16.html
+++ b/docs/dev/reference/application_vgg16.html
@@ -8,7 +8,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -174,7 +174,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/application_vgg19.html b/docs/dev/reference/application_vgg19.html
index f4b35d4ba..47409fb2e 100644
--- a/docs/dev/reference/application_vgg19.html
+++ b/docs/dev/reference/application_vgg19.html
@@ -8,7 +8,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -174,7 +174,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/application_xception.html b/docs/dev/reference/application_xception.html
index c3fd07c47..ddffec522 100644
--- a/docs/dev/reference/application_xception.html
+++ b/docs/dev/reference/application_xception.html
@@ -8,7 +8,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -170,7 +170,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/audio_dataset_from_directory.html b/docs/dev/reference/audio_dataset_from_directory.html
index 624d694a6..f366f6218 100644
--- a/docs/dev/reference/audio_dataset_from_directory.html
+++ b/docs/dev/reference/audio_dataset_from_directory.html
@@ -34,7 +34,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -239,7 +239,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/back-compat.html b/docs/dev/reference/back-compat.html
index a334f42de..bc92c6cbe 100644
--- a/docs/dev/reference/back-compat.html
+++ b/docs/dev/reference/back-compat.html
@@ -10,7 +10,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -85,7 +85,7 @@ <h2 id="arguments">Arguments<a class="anchor" aria-label="anchor" href="#argumen
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/bidirectional.html b/docs/dev/reference/bidirectional.html
index 2f586f57c..72f4892ab 100644
--- a/docs/dev/reference/bidirectional.html
+++ b/docs/dev/reference/bidirectional.html
@@ -10,7 +10,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -147,7 +147,7 @@ <h2 id="value">Value<a class="anchor" aria-label="anchor" href="#value"></a></h2
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/callback_backup_and_restore.html b/docs/dev/reference/callback_backup_and_restore.html
index 1e1b4109b..5bf62d767 100644
--- a/docs/dev/reference/callback_backup_and_restore.html
+++ b/docs/dev/reference/callback_backup_and_restore.html
@@ -38,7 +38,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -216,7 +216,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/callback_csv_logger.html b/docs/dev/reference/callback_csv_logger.html
index 405120f1c..6ad3d6823 100644
--- a/docs/dev/reference/callback_csv_logger.html
+++ b/docs/dev/reference/callback_csv_logger.html
@@ -10,7 +10,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -107,7 +107,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/callback_early_stopping.html b/docs/dev/reference/callback_early_stopping.html
index 70f77700c..d8cc9db65 100644
--- a/docs/dev/reference/callback_early_stopping.html
+++ b/docs/dev/reference/callback_early_stopping.html
@@ -22,7 +22,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -185,7 +185,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/callback_lambda.html b/docs/dev/reference/callback_lambda.html
index c88d398b5..117cbc07b 100644
--- a/docs/dev/reference/callback_lambda.html
+++ b/docs/dev/reference/callback_lambda.html
@@ -6,8 +6,8 @@
 epoch, logs
 on_train_begin and on_train_end expect one positional argument:
 logs
-on_train_batch_begin and on_train_batch_end expect two positional
-arguments: batch, logs
+on_train_batch_begin and on_train_batch_end expect a positional
+argument batch and a named argument logs
 See Callback class definition for the full list of functions and their
 expected arguments.
 
@@ -18,8 +18,8 @@
 epoch, logs
 on_train_begin and on_train_end expect one positional argument:
 logs
-on_train_batch_begin and on_train_batch_end expect two positional
-arguments: batch, logs
+on_train_batch_begin and on_train_batch_end expect a positional
+argument batch and a named argument logs
 See Callback class definition for the full list of functions and their
 expected arguments.
 
@@ -32,7 +32,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -84,8 +84,8 @@ <h1>Callback for creating simple, custom callbacks on-the-fly.</h1>
 <code>epoch</code>, <code>logs</code></p></li>
 <li><p><code>on_train_begin</code> and <code>on_train_end</code> expect one positional argument:
 <code>logs</code></p></li>
-<li><p><code>on_train_batch_begin</code> and <code>on_train_batch_end</code> expect two positional
-arguments: <code>batch</code>, <code>logs</code></p></li>
+<li><p><code>on_train_batch_begin</code> and <code>on_train_batch_end</code> expect a positional
+argument <code>batch</code> and a named argument <code>logs</code></p></li>
 <li><p>See <code>Callback</code> class definition for the full list of functions and their
 expected arguments.</p></li>
 </ul></div>
@@ -135,7 +135,7 @@ <h2 id="arguments">Arguments<a class="anchor" aria-label="anchor" href="#argumen
 <dd><p>Any function in <code><a href="Callback.html">Callback()</a></code> that you want to override by
 passing <code>function_name = function</code>. For example,
 <code>callback_lambda(.., on_train_end = train_end_fn)</code>. The custom function
-needs to have same arguments as the ones defined in <code><a href="Callback.html">Callback()</a></code>.</p></dd>
+needs to have the same arguments as the ones defined in <code><a href="Callback.html">Callback()</a></code>.</p></dd>
 
 </dl></div>
     <div class="section level2">
@@ -204,7 +204,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/callback_learning_rate_scheduler.html b/docs/dev/reference/callback_learning_rate_scheduler.html
index 415d7efda..6ddde712d 100644
--- a/docs/dev/reference/callback_learning_rate_scheduler.html
+++ b/docs/dev/reference/callback_learning_rate_scheduler.html
@@ -14,7 +14,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -129,7 +129,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/callback_model_checkpoint.html b/docs/dev/reference/callback_model_checkpoint.html
index 5f663bdf8..5ade68e01 100644
--- a/docs/dev/reference/callback_model_checkpoint.html
+++ b/docs/dev/reference/callback_model_checkpoint.html
@@ -36,7 +36,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -169,9 +169,8 @@ <h2 id="arguments">Arguments<a class="anchor" aria-label="anchor" href="#argumen
 decision to overwrite the current save file is made based on either
 the maximization or the minimization of the monitored quantity.
 For <code>val_acc</code>, this should be <code>"max"</code>, for <code>val_loss</code> this should be
-<code>"min"</code>, etc. In <code>"auto"</code> mode, the mode is set to <code>"max"</code> if the
-quantities monitored are <code>"acc"</code> or start with <code>"fmeasure"</code> and are
-set to <code>"min"</code> for the rest of the quantities.</p></dd>
+<code>"min"</code>, etc. In <code>"auto"</code> mode, the direction is automatically inferred from
+the name of the monitored quantity.</p></dd>
 
 
 <dt id="arg-save-freq">save_freq<a class="anchor" aria-label="anchor" href="#arg-save-freq"></a></dt>
@@ -266,7 +265,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/callback_reduce_lr_on_plateau.html b/docs/dev/reference/callback_reduce_lr_on_plateau.html
index c580e5d8e..851377380 100644
--- a/docs/dev/reference/callback_reduce_lr_on_plateau.html
+++ b/docs/dev/reference/callback_reduce_lr_on_plateau.html
@@ -14,7 +14,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -156,7 +156,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/callback_remote_monitor.html b/docs/dev/reference/callback_remote_monitor.html
index 4f427ec2c..2ba74be6b 100644
--- a/docs/dev/reference/callback_remote_monitor.html
+++ b/docs/dev/reference/callback_remote_monitor.html
@@ -20,7 +20,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -133,7 +133,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/callback_swap_ema_weights.html b/docs/dev/reference/callback_swap_ema_weights.html
index 747bc22c9..21f2cdbec 100644
--- a/docs/dev/reference/callback_swap_ema_weights.html
+++ b/docs/dev/reference/callback_swap_ema_weights.html
@@ -24,7 +24,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -135,7 +135,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/callback_tensorboard.html b/docs/dev/reference/callback_tensorboard.html
index af5f755b5..23fe638d9 100644
--- a/docs/dev/reference/callback_tensorboard.html
+++ b/docs/dev/reference/callback_tensorboard.html
@@ -48,7 +48,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -276,7 +276,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/callback_terminate_on_nan.html b/docs/dev/reference/callback_terminate_on_nan.html
index 306ed89e6..4d3fe34b6 100644
--- a/docs/dev/reference/callback_terminate_on_nan.html
+++ b/docs/dev/reference/callback_terminate_on_nan.html
@@ -8,7 +8,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -82,7 +82,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/clear_session.html b/docs/dev/reference/clear_session.html
index c11a76950..6f1becc78 100644
--- a/docs/dev/reference/clear_session.html
+++ b/docs/dev/reference/clear_session.html
@@ -92,7 +92,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -217,7 +217,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/clone_model.html b/docs/dev/reference/clone_model.html
index c4f6e10c7..3c78fb155 100644
--- a/docs/dev/reference/clone_model.html
+++ b/docs/dev/reference/clone_model.html
@@ -20,7 +20,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -233,7 +233,7 @@ <h2 id="examples">Examples<a class="anchor" aria-label="anchor" href="#examples"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/compile.keras.src.models.model.Model.html b/docs/dev/reference/compile.keras.src.models.model.Model.html
index 20749f609..2fa48cf7d 100644
--- a/docs/dev/reference/compile.keras.src.models.model.Model.html
+++ b/docs/dev/reference/compile.keras.src.models.model.Model.html
@@ -8,7 +8,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -225,7 +225,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/config_backend.html b/docs/dev/reference/config_backend.html
index 2e4b79639..e0f3d520a 100644
--- a/docs/dev/reference/config_backend.html
+++ b/docs/dev/reference/config_backend.html
@@ -8,7 +8,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -78,7 +78,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
     <div class="dont-index"><p><code><a href="use_backend.html">use_backend()</a></code></p>
 <p>Other config backend: <br><code><a href="config_epsilon.html">config_epsilon</a>()</code> <br><code><a href="config_floatx.html">config_floatx</a>()</code> <br><code><a href="config_image_data_format.html">config_image_data_format</a>()</code> <br><code><a href="config_set_epsilon.html">config_set_epsilon</a>()</code> <br><code><a href="config_set_floatx.html">config_set_floatx</a>()</code> <br><code><a href="config_set_image_data_format.html">config_set_image_data_format</a>()</code> <br></p>
 <p>Other backend: <br><code><a href="clear_session.html">clear_session</a>()</code> <br><code><a href="config_epsilon.html">config_epsilon</a>()</code> <br><code><a href="config_floatx.html">config_floatx</a>()</code> <br><code><a href="config_image_data_format.html">config_image_data_format</a>()</code> <br><code><a href="config_set_epsilon.html">config_set_epsilon</a>()</code> <br><code><a href="config_set_floatx.html">config_set_floatx</a>()</code> <br><code><a href="config_set_image_data_format.html">config_set_image_data_format</a>()</code> <br></p>
-<p>Other config: <br><code><a href="config_disable_flash_attention.html">config_disable_flash_attention</a>()</code> <br><code><a href="config_disable_interactive_logging.html">config_disable_interactive_logging</a>()</code> <br><code><a href="config_disable_traceback_filtering.html">config_disable_traceback_filtering</a>()</code> <br><code><a href="config_dtype_policy.html">config_dtype_policy</a>()</code> <br><code><a href="config_enable_flash_attention.html">config_enable_flash_attention</a>()</code> <br><code><a href="config_enable_interactive_logging.html">config_enable_interactive_logging</a>()</code> <br><code><a href="config_enable_traceback_filtering.html">config_enable_traceback_filtering</a>()</code> <br><code><a href="config_enable_unsafe_deserialization.html">config_enable_unsafe_deserialization</a>()</code> <br><code><a href="config_epsilon.html">config_epsilon</a>()</code> <br><code><a href="config_floatx.html">config_floatx</a>()</code> <br><code><a href="config_image_data_format.html">config_image_data_format</a>()</code> <br><code><a href="config_is_interactive_logging_enabled.html">config_is_interactive_logging_enabled</a>()</code> <br><code><a href="config_is_traceback_filtering_enabled.html">config_is_traceback_filtering_enabled</a>()</code> <br><code><a href="config_set_backend.html">config_set_backend</a>()</code> <br><code><a href="config_set_dtype_policy.html">config_set_dtype_policy</a>()</code> <br><code><a href="config_set_epsilon.html">config_set_epsilon</a>()</code> <br><code><a href="config_set_floatx.html">config_set_floatx</a>()</code> <br><code><a href="config_set_image_data_format.html">config_set_image_data_format</a>()</code> <br></p></div>
+<p>Other config: <br><code><a href="config_disable_flash_attention.html">config_disable_flash_attention</a>()</code> <br><code><a href="config_disable_interactive_logging.html">config_disable_interactive_logging</a>()</code> <br><code><a href="config_disable_traceback_filtering.html">config_disable_traceback_filtering</a>()</code> <br><code><a href="config_dtype_policy.html">config_dtype_policy</a>()</code> <br><code><a href="config_enable_flash_attention.html">config_enable_flash_attention</a>()</code> <br><code><a href="config_enable_interactive_logging.html">config_enable_interactive_logging</a>()</code> <br><code><a href="config_enable_traceback_filtering.html">config_enable_traceback_filtering</a>()</code> <br><code><a href="config_enable_unsafe_deserialization.html">config_enable_unsafe_deserialization</a>()</code> <br><code><a href="config_epsilon.html">config_epsilon</a>()</code> <br><code><a href="config_floatx.html">config_floatx</a>()</code> <br><code><a href="config_image_data_format.html">config_image_data_format</a>()</code> <br><code><a href="config_is_interactive_logging_enabled.html">config_is_interactive_logging_enabled</a>()</code> <br><code><a href="config_is_nnx_enabled.html">config_is_nnx_enabled</a>()</code> <br><code><a href="config_is_traceback_filtering_enabled.html">config_is_traceback_filtering_enabled</a>()</code> <br><code><a href="config_max_epochs.html">config_max_epochs</a>()</code> <br><code><a href="config_set_backend.html">config_set_backend</a>()</code> <br><code><a href="config_set_dtype_policy.html">config_set_dtype_policy</a>()</code> <br><code><a href="config_set_epsilon.html">config_set_epsilon</a>()</code> <br><code><a href="config_set_floatx.html">config_set_floatx</a>()</code> <br><code><a href="config_set_image_data_format.html">config_set_image_data_format</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -90,7 +90,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/config_disable_flash_attention.html b/docs/dev/reference/config_disable_flash_attention.html
index d08845814..7028f1094 100644
--- a/docs/dev/reference/config_disable_flash_attention.html
+++ b/docs/dev/reference/config_disable_flash_attention.html
@@ -16,7 +16,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -77,7 +77,7 @@ <h2 id="ref-usage">Usage<a class="anchor" aria-label="anchor" href="#ref-usage">
     <div class="section level2">
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
     <div class="dont-index"><p><code><a href="config_is_flash_attention_enabled.html">config_is_flash_attention_enabled()</a></code> <code><a href="config_enable_flash_attention.html">config_enable_flash_attention()</a></code></p>
-<p>Other config: <br><code><a href="config_backend.html">config_backend</a>()</code> <br><code><a href="config_disable_interactive_logging.html">config_disable_interactive_logging</a>()</code> <br><code><a href="config_disable_traceback_filtering.html">config_disable_traceback_filtering</a>()</code> <br><code><a href="config_dtype_policy.html">config_dtype_policy</a>()</code> <br><code><a href="config_enable_flash_attention.html">config_enable_flash_attention</a>()</code> <br><code><a href="config_enable_interactive_logging.html">config_enable_interactive_logging</a>()</code> <br><code><a href="config_enable_traceback_filtering.html">config_enable_traceback_filtering</a>()</code> <br><code><a href="config_enable_unsafe_deserialization.html">config_enable_unsafe_deserialization</a>()</code> <br><code><a href="config_epsilon.html">config_epsilon</a>()</code> <br><code><a href="config_floatx.html">config_floatx</a>()</code> <br><code><a href="config_image_data_format.html">config_image_data_format</a>()</code> <br><code><a href="config_is_interactive_logging_enabled.html">config_is_interactive_logging_enabled</a>()</code> <br><code><a href="config_is_traceback_filtering_enabled.html">config_is_traceback_filtering_enabled</a>()</code> <br><code><a href="config_set_backend.html">config_set_backend</a>()</code> <br><code><a href="config_set_dtype_policy.html">config_set_dtype_policy</a>()</code> <br><code><a href="config_set_epsilon.html">config_set_epsilon</a>()</code> <br><code><a href="config_set_floatx.html">config_set_floatx</a>()</code> <br><code><a href="config_set_image_data_format.html">config_set_image_data_format</a>()</code> <br></p></div>
+<p>Other config: <br><code><a href="config_backend.html">config_backend</a>()</code> <br><code><a href="config_disable_interactive_logging.html">config_disable_interactive_logging</a>()</code> <br><code><a href="config_disable_traceback_filtering.html">config_disable_traceback_filtering</a>()</code> <br><code><a href="config_dtype_policy.html">config_dtype_policy</a>()</code> <br><code><a href="config_enable_flash_attention.html">config_enable_flash_attention</a>()</code> <br><code><a href="config_enable_interactive_logging.html">config_enable_interactive_logging</a>()</code> <br><code><a href="config_enable_traceback_filtering.html">config_enable_traceback_filtering</a>()</code> <br><code><a href="config_enable_unsafe_deserialization.html">config_enable_unsafe_deserialization</a>()</code> <br><code><a href="config_epsilon.html">config_epsilon</a>()</code> <br><code><a href="config_floatx.html">config_floatx</a>()</code> <br><code><a href="config_image_data_format.html">config_image_data_format</a>()</code> <br><code><a href="config_is_interactive_logging_enabled.html">config_is_interactive_logging_enabled</a>()</code> <br><code><a href="config_is_nnx_enabled.html">config_is_nnx_enabled</a>()</code> <br><code><a href="config_is_traceback_filtering_enabled.html">config_is_traceback_filtering_enabled</a>()</code> <br><code><a href="config_max_epochs.html">config_max_epochs</a>()</code> <br><code><a href="config_set_backend.html">config_set_backend</a>()</code> <br><code><a href="config_set_dtype_policy.html">config_set_dtype_policy</a>()</code> <br><code><a href="config_set_epsilon.html">config_set_epsilon</a>()</code> <br><code><a href="config_set_floatx.html">config_set_floatx</a>()</code> <br><code><a href="config_set_image_data_format.html">config_set_image_data_format</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -89,7 +89,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/config_disable_interactive_logging.html b/docs/dev/reference/config_disable_interactive_logging.html
index 85340fa55..879d9f40e 100644
--- a/docs/dev/reference/config_disable_interactive_logging.html
+++ b/docs/dev/reference/config_disable_interactive_logging.html
@@ -12,7 +12,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -76,7 +76,7 @@ <h2 id="value">Value<a class="anchor" aria-label="anchor" href="#value"></a></h2
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
     <div class="dont-index"><p>Other io utils: <br><code><a href="config_enable_interactive_logging.html">config_enable_interactive_logging</a>()</code> <br><code><a href="config_is_interactive_logging_enabled.html">config_is_interactive_logging_enabled</a>()</code> <br></p>
 <p>Other utils: <br><code><a href="audio_dataset_from_directory.html">audio_dataset_from_directory</a>()</code> <br><code><a href="clear_session.html">clear_session</a>()</code> <br><code><a href="config_disable_traceback_filtering.html">config_disable_traceback_filtering</a>()</code> <br><code><a href="config_enable_interactive_logging.html">config_enable_interactive_logging</a>()</code> <br><code><a href="config_enable_traceback_filtering.html">config_enable_traceback_filtering</a>()</code> <br><code><a href="config_is_interactive_logging_enabled.html">config_is_interactive_logging_enabled</a>()</code> <br><code><a href="config_is_traceback_filtering_enabled.html">config_is_traceback_filtering_enabled</a>()</code> <br><code><a href="get_file.html">get_file</a>()</code> <br><code><a href="get_source_inputs.html">get_source_inputs</a>()</code> <br><code><a href="image_array_save.html">image_array_save</a>()</code> <br><code><a href="image_dataset_from_directory.html">image_dataset_from_directory</a>()</code> <br><code><a href="image_from_array.html">image_from_array</a>()</code> <br><code><a href="image_load.html">image_load</a>()</code> <br><code><a href="image_smart_resize.html">image_smart_resize</a>()</code> <br><code><a href="image_to_array.html">image_to_array</a>()</code> <br><code><a href="layer_feature_space.html">layer_feature_space</a>()</code> <br><code><a href="normalize.html">normalize</a>()</code> <br><code><a href="pad_sequences.html">pad_sequences</a>()</code> <br><code><a href="set_random_seed.html">set_random_seed</a>()</code> <br><code><a href="split_dataset.html">split_dataset</a>()</code> <br><code><a href="text_dataset_from_directory.html">text_dataset_from_directory</a>()</code> <br><code><a href="timeseries_dataset_from_array.html">timeseries_dataset_from_array</a>()</code> <br><code><a href="to_categorical.html">to_categorical</a>()</code> <br><code><a href="zip_lists.html">zip_lists</a>()</code> <br></p>
-<p>Other config: <br><code><a href="config_backend.html">config_backend</a>()</code> <br><code><a href="config_disable_flash_attention.html">config_disable_flash_attention</a>()</code> <br><code><a href="config_disable_traceback_filtering.html">config_disable_traceback_filtering</a>()</code> <br><code><a href="config_dtype_policy.html">config_dtype_policy</a>()</code> <br><code><a href="config_enable_flash_attention.html">config_enable_flash_attention</a>()</code> <br><code><a href="config_enable_interactive_logging.html">config_enable_interactive_logging</a>()</code> <br><code><a href="config_enable_traceback_filtering.html">config_enable_traceback_filtering</a>()</code> <br><code><a href="config_enable_unsafe_deserialization.html">config_enable_unsafe_deserialization</a>()</code> <br><code><a href="config_epsilon.html">config_epsilon</a>()</code> <br><code><a href="config_floatx.html">config_floatx</a>()</code> <br><code><a href="config_image_data_format.html">config_image_data_format</a>()</code> <br><code><a href="config_is_interactive_logging_enabled.html">config_is_interactive_logging_enabled</a>()</code> <br><code><a href="config_is_traceback_filtering_enabled.html">config_is_traceback_filtering_enabled</a>()</code> <br><code><a href="config_set_backend.html">config_set_backend</a>()</code> <br><code><a href="config_set_dtype_policy.html">config_set_dtype_policy</a>()</code> <br><code><a href="config_set_epsilon.html">config_set_epsilon</a>()</code> <br><code><a href="config_set_floatx.html">config_set_floatx</a>()</code> <br><code><a href="config_set_image_data_format.html">config_set_image_data_format</a>()</code> <br></p></div>
+<p>Other config: <br><code><a href="config_backend.html">config_backend</a>()</code> <br><code><a href="config_disable_flash_attention.html">config_disable_flash_attention</a>()</code> <br><code><a href="config_disable_traceback_filtering.html">config_disable_traceback_filtering</a>()</code> <br><code><a href="config_dtype_policy.html">config_dtype_policy</a>()</code> <br><code><a href="config_enable_flash_attention.html">config_enable_flash_attention</a>()</code> <br><code><a href="config_enable_interactive_logging.html">config_enable_interactive_logging</a>()</code> <br><code><a href="config_enable_traceback_filtering.html">config_enable_traceback_filtering</a>()</code> <br><code><a href="config_enable_unsafe_deserialization.html">config_enable_unsafe_deserialization</a>()</code> <br><code><a href="config_epsilon.html">config_epsilon</a>()</code> <br><code><a href="config_floatx.html">config_floatx</a>()</code> <br><code><a href="config_image_data_format.html">config_image_data_format</a>()</code> <br><code><a href="config_is_interactive_logging_enabled.html">config_is_interactive_logging_enabled</a>()</code> <br><code><a href="config_is_nnx_enabled.html">config_is_nnx_enabled</a>()</code> <br><code><a href="config_is_traceback_filtering_enabled.html">config_is_traceback_filtering_enabled</a>()</code> <br><code><a href="config_max_epochs.html">config_max_epochs</a>()</code> <br><code><a href="config_set_backend.html">config_set_backend</a>()</code> <br><code><a href="config_set_dtype_policy.html">config_set_dtype_policy</a>()</code> <br><code><a href="config_set_epsilon.html">config_set_epsilon</a>()</code> <br><code><a href="config_set_floatx.html">config_set_floatx</a>()</code> <br><code><a href="config_set_image_data_format.html">config_set_image_data_format</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -88,7 +88,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/config_disable_traceback_filtering.html b/docs/dev/reference/config_disable_traceback_filtering.html
index 83e789247..64404f251 100644
--- a/docs/dev/reference/config_disable_traceback_filtering.html
+++ b/docs/dev/reference/config_disable_traceback_filtering.html
@@ -28,7 +28,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -100,7 +100,7 @@ <h2 id="value">Value<a class="anchor" aria-label="anchor" href="#value"></a></h2
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
     <div class="dont-index"><p>Other traceback utils: <br><code><a href="config_enable_traceback_filtering.html">config_enable_traceback_filtering</a>()</code> <br><code><a href="config_is_traceback_filtering_enabled.html">config_is_traceback_filtering_enabled</a>()</code> <br></p>
 <p>Other utils: <br><code><a href="audio_dataset_from_directory.html">audio_dataset_from_directory</a>()</code> <br><code><a href="clear_session.html">clear_session</a>()</code> <br><code><a href="config_disable_interactive_logging.html">config_disable_interactive_logging</a>()</code> <br><code><a href="config_enable_interactive_logging.html">config_enable_interactive_logging</a>()</code> <br><code><a href="config_enable_traceback_filtering.html">config_enable_traceback_filtering</a>()</code> <br><code><a href="config_is_interactive_logging_enabled.html">config_is_interactive_logging_enabled</a>()</code> <br><code><a href="config_is_traceback_filtering_enabled.html">config_is_traceback_filtering_enabled</a>()</code> <br><code><a href="get_file.html">get_file</a>()</code> <br><code><a href="get_source_inputs.html">get_source_inputs</a>()</code> <br><code><a href="image_array_save.html">image_array_save</a>()</code> <br><code><a href="image_dataset_from_directory.html">image_dataset_from_directory</a>()</code> <br><code><a href="image_from_array.html">image_from_array</a>()</code> <br><code><a href="image_load.html">image_load</a>()</code> <br><code><a href="image_smart_resize.html">image_smart_resize</a>()</code> <br><code><a href="image_to_array.html">image_to_array</a>()</code> <br><code><a href="layer_feature_space.html">layer_feature_space</a>()</code> <br><code><a href="normalize.html">normalize</a>()</code> <br><code><a href="pad_sequences.html">pad_sequences</a>()</code> <br><code><a href="set_random_seed.html">set_random_seed</a>()</code> <br><code><a href="split_dataset.html">split_dataset</a>()</code> <br><code><a href="text_dataset_from_directory.html">text_dataset_from_directory</a>()</code> <br><code><a href="timeseries_dataset_from_array.html">timeseries_dataset_from_array</a>()</code> <br><code><a href="to_categorical.html">to_categorical</a>()</code> <br><code><a href="zip_lists.html">zip_lists</a>()</code> <br></p>
-<p>Other config: <br><code><a href="config_backend.html">config_backend</a>()</code> <br><code><a href="config_disable_flash_attention.html">config_disable_flash_attention</a>()</code> <br><code><a href="config_disable_interactive_logging.html">config_disable_interactive_logging</a>()</code> <br><code><a href="config_dtype_policy.html">config_dtype_policy</a>()</code> <br><code><a href="config_enable_flash_attention.html">config_enable_flash_attention</a>()</code> <br><code><a href="config_enable_interactive_logging.html">config_enable_interactive_logging</a>()</code> <br><code><a href="config_enable_traceback_filtering.html">config_enable_traceback_filtering</a>()</code> <br><code><a href="config_enable_unsafe_deserialization.html">config_enable_unsafe_deserialization</a>()</code> <br><code><a href="config_epsilon.html">config_epsilon</a>()</code> <br><code><a href="config_floatx.html">config_floatx</a>()</code> <br><code><a href="config_image_data_format.html">config_image_data_format</a>()</code> <br><code><a href="config_is_interactive_logging_enabled.html">config_is_interactive_logging_enabled</a>()</code> <br><code><a href="config_is_traceback_filtering_enabled.html">config_is_traceback_filtering_enabled</a>()</code> <br><code><a href="config_set_backend.html">config_set_backend</a>()</code> <br><code><a href="config_set_dtype_policy.html">config_set_dtype_policy</a>()</code> <br><code><a href="config_set_epsilon.html">config_set_epsilon</a>()</code> <br><code><a href="config_set_floatx.html">config_set_floatx</a>()</code> <br><code><a href="config_set_image_data_format.html">config_set_image_data_format</a>()</code> <br></p></div>
+<p>Other config: <br><code><a href="config_backend.html">config_backend</a>()</code> <br><code><a href="config_disable_flash_attention.html">config_disable_flash_attention</a>()</code> <br><code><a href="config_disable_interactive_logging.html">config_disable_interactive_logging</a>()</code> <br><code><a href="config_dtype_policy.html">config_dtype_policy</a>()</code> <br><code><a href="config_enable_flash_attention.html">config_enable_flash_attention</a>()</code> <br><code><a href="config_enable_interactive_logging.html">config_enable_interactive_logging</a>()</code> <br><code><a href="config_enable_traceback_filtering.html">config_enable_traceback_filtering</a>()</code> <br><code><a href="config_enable_unsafe_deserialization.html">config_enable_unsafe_deserialization</a>()</code> <br><code><a href="config_epsilon.html">config_epsilon</a>()</code> <br><code><a href="config_floatx.html">config_floatx</a>()</code> <br><code><a href="config_image_data_format.html">config_image_data_format</a>()</code> <br><code><a href="config_is_interactive_logging_enabled.html">config_is_interactive_logging_enabled</a>()</code> <br><code><a href="config_is_nnx_enabled.html">config_is_nnx_enabled</a>()</code> <br><code><a href="config_is_traceback_filtering_enabled.html">config_is_traceback_filtering_enabled</a>()</code> <br><code><a href="config_max_epochs.html">config_max_epochs</a>()</code> <br><code><a href="config_set_backend.html">config_set_backend</a>()</code> <br><code><a href="config_set_dtype_policy.html">config_set_dtype_policy</a>()</code> <br><code><a href="config_set_epsilon.html">config_set_epsilon</a>()</code> <br><code><a href="config_set_floatx.html">config_set_floatx</a>()</code> <br><code><a href="config_set_image_data_format.html">config_set_image_data_format</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -112,7 +112,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/config_dtype_policy.html b/docs/dev/reference/config_dtype_policy.html
index e6c2e58d9..19498c5d7 100644
--- a/docs/dev/reference/config_dtype_policy.html
+++ b/docs/dev/reference/config_dtype_policy.html
@@ -8,7 +8,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -68,7 +68,7 @@ <h2 id="value">Value<a class="anchor" aria-label="anchor" href="#value"></a></h2
     </div>
     <div class="section level2">
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
-    <div class="dont-index"><p>Other config: <br><code><a href="config_backend.html">config_backend</a>()</code> <br><code><a href="config_disable_flash_attention.html">config_disable_flash_attention</a>()</code> <br><code><a href="config_disable_interactive_logging.html">config_disable_interactive_logging</a>()</code> <br><code><a href="config_disable_traceback_filtering.html">config_disable_traceback_filtering</a>()</code> <br><code><a href="config_enable_flash_attention.html">config_enable_flash_attention</a>()</code> <br><code><a href="config_enable_interactive_logging.html">config_enable_interactive_logging</a>()</code> <br><code><a href="config_enable_traceback_filtering.html">config_enable_traceback_filtering</a>()</code> <br><code><a href="config_enable_unsafe_deserialization.html">config_enable_unsafe_deserialization</a>()</code> <br><code><a href="config_epsilon.html">config_epsilon</a>()</code> <br><code><a href="config_floatx.html">config_floatx</a>()</code> <br><code><a href="config_image_data_format.html">config_image_data_format</a>()</code> <br><code><a href="config_is_interactive_logging_enabled.html">config_is_interactive_logging_enabled</a>()</code> <br><code><a href="config_is_traceback_filtering_enabled.html">config_is_traceback_filtering_enabled</a>()</code> <br><code><a href="config_set_backend.html">config_set_backend</a>()</code> <br><code><a href="config_set_dtype_policy.html">config_set_dtype_policy</a>()</code> <br><code><a href="config_set_epsilon.html">config_set_epsilon</a>()</code> <br><code><a href="config_set_floatx.html">config_set_floatx</a>()</code> <br><code><a href="config_set_image_data_format.html">config_set_image_data_format</a>()</code> <br></p></div>
+    <div class="dont-index"><p>Other config: <br><code><a href="config_backend.html">config_backend</a>()</code> <br><code><a href="config_disable_flash_attention.html">config_disable_flash_attention</a>()</code> <br><code><a href="config_disable_interactive_logging.html">config_disable_interactive_logging</a>()</code> <br><code><a href="config_disable_traceback_filtering.html">config_disable_traceback_filtering</a>()</code> <br><code><a href="config_enable_flash_attention.html">config_enable_flash_attention</a>()</code> <br><code><a href="config_enable_interactive_logging.html">config_enable_interactive_logging</a>()</code> <br><code><a href="config_enable_traceback_filtering.html">config_enable_traceback_filtering</a>()</code> <br><code><a href="config_enable_unsafe_deserialization.html">config_enable_unsafe_deserialization</a>()</code> <br><code><a href="config_epsilon.html">config_epsilon</a>()</code> <br><code><a href="config_floatx.html">config_floatx</a>()</code> <br><code><a href="config_image_data_format.html">config_image_data_format</a>()</code> <br><code><a href="config_is_interactive_logging_enabled.html">config_is_interactive_logging_enabled</a>()</code> <br><code><a href="config_is_nnx_enabled.html">config_is_nnx_enabled</a>()</code> <br><code><a href="config_is_traceback_filtering_enabled.html">config_is_traceback_filtering_enabled</a>()</code> <br><code><a href="config_max_epochs.html">config_max_epochs</a>()</code> <br><code><a href="config_set_backend.html">config_set_backend</a>()</code> <br><code><a href="config_set_dtype_policy.html">config_set_dtype_policy</a>()</code> <br><code><a href="config_set_epsilon.html">config_set_epsilon</a>()</code> <br><code><a href="config_set_floatx.html">config_set_floatx</a>()</code> <br><code><a href="config_set_image_data_format.html">config_set_image_data_format</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -80,7 +80,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/config_enable_flash_attention.html b/docs/dev/reference/config_enable_flash_attention.html
index 9491f1317..5c875c93b 100644
--- a/docs/dev/reference/config_enable_flash_attention.html
+++ b/docs/dev/reference/config_enable_flash_attention.html
@@ -24,7 +24,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -89,7 +89,7 @@ <h2 id="ref-usage">Usage<a class="anchor" aria-label="anchor" href="#ref-usage">
     <div class="section level2">
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
     <div class="dont-index"><p><code><a href="config_disable_flash_attention.html">config_disable_flash_attention()</a></code> <code><a href="config_is_flash_attention_enabled.html">config_is_flash_attention_enabled()</a></code></p>
-<p>Other config: <br><code><a href="config_backend.html">config_backend</a>()</code> <br><code><a href="config_disable_flash_attention.html">config_disable_flash_attention</a>()</code> <br><code><a href="config_disable_interactive_logging.html">config_disable_interactive_logging</a>()</code> <br><code><a href="config_disable_traceback_filtering.html">config_disable_traceback_filtering</a>()</code> <br><code><a href="config_dtype_policy.html">config_dtype_policy</a>()</code> <br><code><a href="config_enable_interactive_logging.html">config_enable_interactive_logging</a>()</code> <br><code><a href="config_enable_traceback_filtering.html">config_enable_traceback_filtering</a>()</code> <br><code><a href="config_enable_unsafe_deserialization.html">config_enable_unsafe_deserialization</a>()</code> <br><code><a href="config_epsilon.html">config_epsilon</a>()</code> <br><code><a href="config_floatx.html">config_floatx</a>()</code> <br><code><a href="config_image_data_format.html">config_image_data_format</a>()</code> <br><code><a href="config_is_interactive_logging_enabled.html">config_is_interactive_logging_enabled</a>()</code> <br><code><a href="config_is_traceback_filtering_enabled.html">config_is_traceback_filtering_enabled</a>()</code> <br><code><a href="config_set_backend.html">config_set_backend</a>()</code> <br><code><a href="config_set_dtype_policy.html">config_set_dtype_policy</a>()</code> <br><code><a href="config_set_epsilon.html">config_set_epsilon</a>()</code> <br><code><a href="config_set_floatx.html">config_set_floatx</a>()</code> <br><code><a href="config_set_image_data_format.html">config_set_image_data_format</a>()</code> <br></p></div>
+<p>Other config: <br><code><a href="config_backend.html">config_backend</a>()</code> <br><code><a href="config_disable_flash_attention.html">config_disable_flash_attention</a>()</code> <br><code><a href="config_disable_interactive_logging.html">config_disable_interactive_logging</a>()</code> <br><code><a href="config_disable_traceback_filtering.html">config_disable_traceback_filtering</a>()</code> <br><code><a href="config_dtype_policy.html">config_dtype_policy</a>()</code> <br><code><a href="config_enable_interactive_logging.html">config_enable_interactive_logging</a>()</code> <br><code><a href="config_enable_traceback_filtering.html">config_enable_traceback_filtering</a>()</code> <br><code><a href="config_enable_unsafe_deserialization.html">config_enable_unsafe_deserialization</a>()</code> <br><code><a href="config_epsilon.html">config_epsilon</a>()</code> <br><code><a href="config_floatx.html">config_floatx</a>()</code> <br><code><a href="config_image_data_format.html">config_image_data_format</a>()</code> <br><code><a href="config_is_interactive_logging_enabled.html">config_is_interactive_logging_enabled</a>()</code> <br><code><a href="config_is_nnx_enabled.html">config_is_nnx_enabled</a>()</code> <br><code><a href="config_is_traceback_filtering_enabled.html">config_is_traceback_filtering_enabled</a>()</code> <br><code><a href="config_max_epochs.html">config_max_epochs</a>()</code> <br><code><a href="config_set_backend.html">config_set_backend</a>()</code> <br><code><a href="config_set_dtype_policy.html">config_set_dtype_policy</a>()</code> <br><code><a href="config_set_epsilon.html">config_set_epsilon</a>()</code> <br><code><a href="config_set_floatx.html">config_set_floatx</a>()</code> <br><code><a href="config_set_image_data_format.html">config_set_image_data_format</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -101,7 +101,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/config_enable_interactive_logging.html b/docs/dev/reference/config_enable_interactive_logging.html
index ece0faa2c..a05a3ab28 100644
--- a/docs/dev/reference/config_enable_interactive_logging.html
+++ b/docs/dev/reference/config_enable_interactive_logging.html
@@ -12,7 +12,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -76,7 +76,7 @@ <h2 id="value">Value<a class="anchor" aria-label="anchor" href="#value"></a></h2
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
     <div class="dont-index"><p>Other io utils: <br><code><a href="config_disable_interactive_logging.html">config_disable_interactive_logging</a>()</code> <br><code><a href="config_is_interactive_logging_enabled.html">config_is_interactive_logging_enabled</a>()</code> <br></p>
 <p>Other utils: <br><code><a href="audio_dataset_from_directory.html">audio_dataset_from_directory</a>()</code> <br><code><a href="clear_session.html">clear_session</a>()</code> <br><code><a href="config_disable_interactive_logging.html">config_disable_interactive_logging</a>()</code> <br><code><a href="config_disable_traceback_filtering.html">config_disable_traceback_filtering</a>()</code> <br><code><a href="config_enable_traceback_filtering.html">config_enable_traceback_filtering</a>()</code> <br><code><a href="config_is_interactive_logging_enabled.html">config_is_interactive_logging_enabled</a>()</code> <br><code><a href="config_is_traceback_filtering_enabled.html">config_is_traceback_filtering_enabled</a>()</code> <br><code><a href="get_file.html">get_file</a>()</code> <br><code><a href="get_source_inputs.html">get_source_inputs</a>()</code> <br><code><a href="image_array_save.html">image_array_save</a>()</code> <br><code><a href="image_dataset_from_directory.html">image_dataset_from_directory</a>()</code> <br><code><a href="image_from_array.html">image_from_array</a>()</code> <br><code><a href="image_load.html">image_load</a>()</code> <br><code><a href="image_smart_resize.html">image_smart_resize</a>()</code> <br><code><a href="image_to_array.html">image_to_array</a>()</code> <br><code><a href="layer_feature_space.html">layer_feature_space</a>()</code> <br><code><a href="normalize.html">normalize</a>()</code> <br><code><a href="pad_sequences.html">pad_sequences</a>()</code> <br><code><a href="set_random_seed.html">set_random_seed</a>()</code> <br><code><a href="split_dataset.html">split_dataset</a>()</code> <br><code><a href="text_dataset_from_directory.html">text_dataset_from_directory</a>()</code> <br><code><a href="timeseries_dataset_from_array.html">timeseries_dataset_from_array</a>()</code> <br><code><a href="to_categorical.html">to_categorical</a>()</code> <br><code><a href="zip_lists.html">zip_lists</a>()</code> <br></p>
-<p>Other config: <br><code><a href="config_backend.html">config_backend</a>()</code> <br><code><a href="config_disable_flash_attention.html">config_disable_flash_attention</a>()</code> <br><code><a href="config_disable_interactive_logging.html">config_disable_interactive_logging</a>()</code> <br><code><a href="config_disable_traceback_filtering.html">config_disable_traceback_filtering</a>()</code> <br><code><a href="config_dtype_policy.html">config_dtype_policy</a>()</code> <br><code><a href="config_enable_flash_attention.html">config_enable_flash_attention</a>()</code> <br><code><a href="config_enable_traceback_filtering.html">config_enable_traceback_filtering</a>()</code> <br><code><a href="config_enable_unsafe_deserialization.html">config_enable_unsafe_deserialization</a>()</code> <br><code><a href="config_epsilon.html">config_epsilon</a>()</code> <br><code><a href="config_floatx.html">config_floatx</a>()</code> <br><code><a href="config_image_data_format.html">config_image_data_format</a>()</code> <br><code><a href="config_is_interactive_logging_enabled.html">config_is_interactive_logging_enabled</a>()</code> <br><code><a href="config_is_traceback_filtering_enabled.html">config_is_traceback_filtering_enabled</a>()</code> <br><code><a href="config_set_backend.html">config_set_backend</a>()</code> <br><code><a href="config_set_dtype_policy.html">config_set_dtype_policy</a>()</code> <br><code><a href="config_set_epsilon.html">config_set_epsilon</a>()</code> <br><code><a href="config_set_floatx.html">config_set_floatx</a>()</code> <br><code><a href="config_set_image_data_format.html">config_set_image_data_format</a>()</code> <br></p></div>
+<p>Other config: <br><code><a href="config_backend.html">config_backend</a>()</code> <br><code><a href="config_disable_flash_attention.html">config_disable_flash_attention</a>()</code> <br><code><a href="config_disable_interactive_logging.html">config_disable_interactive_logging</a>()</code> <br><code><a href="config_disable_traceback_filtering.html">config_disable_traceback_filtering</a>()</code> <br><code><a href="config_dtype_policy.html">config_dtype_policy</a>()</code> <br><code><a href="config_enable_flash_attention.html">config_enable_flash_attention</a>()</code> <br><code><a href="config_enable_traceback_filtering.html">config_enable_traceback_filtering</a>()</code> <br><code><a href="config_enable_unsafe_deserialization.html">config_enable_unsafe_deserialization</a>()</code> <br><code><a href="config_epsilon.html">config_epsilon</a>()</code> <br><code><a href="config_floatx.html">config_floatx</a>()</code> <br><code><a href="config_image_data_format.html">config_image_data_format</a>()</code> <br><code><a href="config_is_interactive_logging_enabled.html">config_is_interactive_logging_enabled</a>()</code> <br><code><a href="config_is_nnx_enabled.html">config_is_nnx_enabled</a>()</code> <br><code><a href="config_is_traceback_filtering_enabled.html">config_is_traceback_filtering_enabled</a>()</code> <br><code><a href="config_max_epochs.html">config_max_epochs</a>()</code> <br><code><a href="config_set_backend.html">config_set_backend</a>()</code> <br><code><a href="config_set_dtype_policy.html">config_set_dtype_policy</a>()</code> <br><code><a href="config_set_epsilon.html">config_set_epsilon</a>()</code> <br><code><a href="config_set_floatx.html">config_set_floatx</a>()</code> <br><code><a href="config_set_image_data_format.html">config_set_image_data_format</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -88,7 +88,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/config_enable_traceback_filtering.html b/docs/dev/reference/config_enable_traceback_filtering.html
index fa1c4d02d..e63cacca6 100644
--- a/docs/dev/reference/config_enable_traceback_filtering.html
+++ b/docs/dev/reference/config_enable_traceback_filtering.html
@@ -28,7 +28,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -100,7 +100,7 @@ <h2 id="value">Value<a class="anchor" aria-label="anchor" href="#value"></a></h2
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
     <div class="dont-index"><p>Other traceback utils: <br><code><a href="config_disable_traceback_filtering.html">config_disable_traceback_filtering</a>()</code> <br><code><a href="config_is_traceback_filtering_enabled.html">config_is_traceback_filtering_enabled</a>()</code> <br></p>
 <p>Other utils: <br><code><a href="audio_dataset_from_directory.html">audio_dataset_from_directory</a>()</code> <br><code><a href="clear_session.html">clear_session</a>()</code> <br><code><a href="config_disable_interactive_logging.html">config_disable_interactive_logging</a>()</code> <br><code><a href="config_disable_traceback_filtering.html">config_disable_traceback_filtering</a>()</code> <br><code><a href="config_enable_interactive_logging.html">config_enable_interactive_logging</a>()</code> <br><code><a href="config_is_interactive_logging_enabled.html">config_is_interactive_logging_enabled</a>()</code> <br><code><a href="config_is_traceback_filtering_enabled.html">config_is_traceback_filtering_enabled</a>()</code> <br><code><a href="get_file.html">get_file</a>()</code> <br><code><a href="get_source_inputs.html">get_source_inputs</a>()</code> <br><code><a href="image_array_save.html">image_array_save</a>()</code> <br><code><a href="image_dataset_from_directory.html">image_dataset_from_directory</a>()</code> <br><code><a href="image_from_array.html">image_from_array</a>()</code> <br><code><a href="image_load.html">image_load</a>()</code> <br><code><a href="image_smart_resize.html">image_smart_resize</a>()</code> <br><code><a href="image_to_array.html">image_to_array</a>()</code> <br><code><a href="layer_feature_space.html">layer_feature_space</a>()</code> <br><code><a href="normalize.html">normalize</a>()</code> <br><code><a href="pad_sequences.html">pad_sequences</a>()</code> <br><code><a href="set_random_seed.html">set_random_seed</a>()</code> <br><code><a href="split_dataset.html">split_dataset</a>()</code> <br><code><a href="text_dataset_from_directory.html">text_dataset_from_directory</a>()</code> <br><code><a href="timeseries_dataset_from_array.html">timeseries_dataset_from_array</a>()</code> <br><code><a href="to_categorical.html">to_categorical</a>()</code> <br><code><a href="zip_lists.html">zip_lists</a>()</code> <br></p>
-<p>Other config: <br><code><a href="config_backend.html">config_backend</a>()</code> <br><code><a href="config_disable_flash_attention.html">config_disable_flash_attention</a>()</code> <br><code><a href="config_disable_interactive_logging.html">config_disable_interactive_logging</a>()</code> <br><code><a href="config_disable_traceback_filtering.html">config_disable_traceback_filtering</a>()</code> <br><code><a href="config_dtype_policy.html">config_dtype_policy</a>()</code> <br><code><a href="config_enable_flash_attention.html">config_enable_flash_attention</a>()</code> <br><code><a href="config_enable_interactive_logging.html">config_enable_interactive_logging</a>()</code> <br><code><a href="config_enable_unsafe_deserialization.html">config_enable_unsafe_deserialization</a>()</code> <br><code><a href="config_epsilon.html">config_epsilon</a>()</code> <br><code><a href="config_floatx.html">config_floatx</a>()</code> <br><code><a href="config_image_data_format.html">config_image_data_format</a>()</code> <br><code><a href="config_is_interactive_logging_enabled.html">config_is_interactive_logging_enabled</a>()</code> <br><code><a href="config_is_traceback_filtering_enabled.html">config_is_traceback_filtering_enabled</a>()</code> <br><code><a href="config_set_backend.html">config_set_backend</a>()</code> <br><code><a href="config_set_dtype_policy.html">config_set_dtype_policy</a>()</code> <br><code><a href="config_set_epsilon.html">config_set_epsilon</a>()</code> <br><code><a href="config_set_floatx.html">config_set_floatx</a>()</code> <br><code><a href="config_set_image_data_format.html">config_set_image_data_format</a>()</code> <br></p></div>
+<p>Other config: <br><code><a href="config_backend.html">config_backend</a>()</code> <br><code><a href="config_disable_flash_attention.html">config_disable_flash_attention</a>()</code> <br><code><a href="config_disable_interactive_logging.html">config_disable_interactive_logging</a>()</code> <br><code><a href="config_disable_traceback_filtering.html">config_disable_traceback_filtering</a>()</code> <br><code><a href="config_dtype_policy.html">config_dtype_policy</a>()</code> <br><code><a href="config_enable_flash_attention.html">config_enable_flash_attention</a>()</code> <br><code><a href="config_enable_interactive_logging.html">config_enable_interactive_logging</a>()</code> <br><code><a href="config_enable_unsafe_deserialization.html">config_enable_unsafe_deserialization</a>()</code> <br><code><a href="config_epsilon.html">config_epsilon</a>()</code> <br><code><a href="config_floatx.html">config_floatx</a>()</code> <br><code><a href="config_image_data_format.html">config_image_data_format</a>()</code> <br><code><a href="config_is_interactive_logging_enabled.html">config_is_interactive_logging_enabled</a>()</code> <br><code><a href="config_is_nnx_enabled.html">config_is_nnx_enabled</a>()</code> <br><code><a href="config_is_traceback_filtering_enabled.html">config_is_traceback_filtering_enabled</a>()</code> <br><code><a href="config_max_epochs.html">config_max_epochs</a>()</code> <br><code><a href="config_set_backend.html">config_set_backend</a>()</code> <br><code><a href="config_set_dtype_policy.html">config_set_dtype_policy</a>()</code> <br><code><a href="config_set_epsilon.html">config_set_epsilon</a>()</code> <br><code><a href="config_set_floatx.html">config_set_floatx</a>()</code> <br><code><a href="config_set_image_data_format.html">config_set_image_data_format</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -112,7 +112,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/config_enable_unsafe_deserialization.html b/docs/dev/reference/config_enable_unsafe_deserialization.html
index 51561fec3..bfc4c4242 100644
--- a/docs/dev/reference/config_enable_unsafe_deserialization.html
+++ b/docs/dev/reference/config_enable_unsafe_deserialization.html
@@ -8,7 +8,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -68,7 +68,7 @@ <h2 id="value">Value<a class="anchor" aria-label="anchor" href="#value"></a></h2
     </div>
     <div class="section level2">
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
-    <div class="dont-index"><p>Other config: <br><code><a href="config_backend.html">config_backend</a>()</code> <br><code><a href="config_disable_flash_attention.html">config_disable_flash_attention</a>()</code> <br><code><a href="config_disable_interactive_logging.html">config_disable_interactive_logging</a>()</code> <br><code><a href="config_disable_traceback_filtering.html">config_disable_traceback_filtering</a>()</code> <br><code><a href="config_dtype_policy.html">config_dtype_policy</a>()</code> <br><code><a href="config_enable_flash_attention.html">config_enable_flash_attention</a>()</code> <br><code><a href="config_enable_interactive_logging.html">config_enable_interactive_logging</a>()</code> <br><code><a href="config_enable_traceback_filtering.html">config_enable_traceback_filtering</a>()</code> <br><code><a href="config_epsilon.html">config_epsilon</a>()</code> <br><code><a href="config_floatx.html">config_floatx</a>()</code> <br><code><a href="config_image_data_format.html">config_image_data_format</a>()</code> <br><code><a href="config_is_interactive_logging_enabled.html">config_is_interactive_logging_enabled</a>()</code> <br><code><a href="config_is_traceback_filtering_enabled.html">config_is_traceback_filtering_enabled</a>()</code> <br><code><a href="config_set_backend.html">config_set_backend</a>()</code> <br><code><a href="config_set_dtype_policy.html">config_set_dtype_policy</a>()</code> <br><code><a href="config_set_epsilon.html">config_set_epsilon</a>()</code> <br><code><a href="config_set_floatx.html">config_set_floatx</a>()</code> <br><code><a href="config_set_image_data_format.html">config_set_image_data_format</a>()</code> <br></p></div>
+    <div class="dont-index"><p>Other config: <br><code><a href="config_backend.html">config_backend</a>()</code> <br><code><a href="config_disable_flash_attention.html">config_disable_flash_attention</a>()</code> <br><code><a href="config_disable_interactive_logging.html">config_disable_interactive_logging</a>()</code> <br><code><a href="config_disable_traceback_filtering.html">config_disable_traceback_filtering</a>()</code> <br><code><a href="config_dtype_policy.html">config_dtype_policy</a>()</code> <br><code><a href="config_enable_flash_attention.html">config_enable_flash_attention</a>()</code> <br><code><a href="config_enable_interactive_logging.html">config_enable_interactive_logging</a>()</code> <br><code><a href="config_enable_traceback_filtering.html">config_enable_traceback_filtering</a>()</code> <br><code><a href="config_epsilon.html">config_epsilon</a>()</code> <br><code><a href="config_floatx.html">config_floatx</a>()</code> <br><code><a href="config_image_data_format.html">config_image_data_format</a>()</code> <br><code><a href="config_is_interactive_logging_enabled.html">config_is_interactive_logging_enabled</a>()</code> <br><code><a href="config_is_nnx_enabled.html">config_is_nnx_enabled</a>()</code> <br><code><a href="config_is_traceback_filtering_enabled.html">config_is_traceback_filtering_enabled</a>()</code> <br><code><a href="config_max_epochs.html">config_max_epochs</a>()</code> <br><code><a href="config_set_backend.html">config_set_backend</a>()</code> <br><code><a href="config_set_dtype_policy.html">config_set_dtype_policy</a>()</code> <br><code><a href="config_set_epsilon.html">config_set_epsilon</a>()</code> <br><code><a href="config_set_floatx.html">config_set_floatx</a>()</code> <br><code><a href="config_set_image_data_format.html">config_set_image_data_format</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -80,7 +80,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/config_epsilon.html b/docs/dev/reference/config_epsilon.html
index f449f7369..e3f3c45b4 100644
--- a/docs/dev/reference/config_epsilon.html
+++ b/docs/dev/reference/config_epsilon.html
@@ -8,7 +8,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -78,7 +78,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 <ul><li><p><a href="https://keras.io/api/utils/config_utils#epsilon-function" class="external-link">https://keras.io/api/utils/config_utils#epsilon-function</a></p></li>
 </ul><p>Other config backend: <br><code><a href="config_backend.html">config_backend</a>()</code> <br><code><a href="config_floatx.html">config_floatx</a>()</code> <br><code><a href="config_image_data_format.html">config_image_data_format</a>()</code> <br><code><a href="config_set_epsilon.html">config_set_epsilon</a>()</code> <br><code><a href="config_set_floatx.html">config_set_floatx</a>()</code> <br><code><a href="config_set_image_data_format.html">config_set_image_data_format</a>()</code> <br></p>
 <p>Other backend: <br><code><a href="clear_session.html">clear_session</a>()</code> <br><code><a href="config_backend.html">config_backend</a>()</code> <br><code><a href="config_floatx.html">config_floatx</a>()</code> <br><code><a href="config_image_data_format.html">config_image_data_format</a>()</code> <br><code><a href="config_set_epsilon.html">config_set_epsilon</a>()</code> <br><code><a href="config_set_floatx.html">config_set_floatx</a>()</code> <br><code><a href="config_set_image_data_format.html">config_set_image_data_format</a>()</code> <br></p>
-<p>Other config: <br><code><a href="config_backend.html">config_backend</a>()</code> <br><code><a href="config_disable_flash_attention.html">config_disable_flash_attention</a>()</code> <br><code><a href="config_disable_interactive_logging.html">config_disable_interactive_logging</a>()</code> <br><code><a href="config_disable_traceback_filtering.html">config_disable_traceback_filtering</a>()</code> <br><code><a href="config_dtype_policy.html">config_dtype_policy</a>()</code> <br><code><a href="config_enable_flash_attention.html">config_enable_flash_attention</a>()</code> <br><code><a href="config_enable_interactive_logging.html">config_enable_interactive_logging</a>()</code> <br><code><a href="config_enable_traceback_filtering.html">config_enable_traceback_filtering</a>()</code> <br><code><a href="config_enable_unsafe_deserialization.html">config_enable_unsafe_deserialization</a>()</code> <br><code><a href="config_floatx.html">config_floatx</a>()</code> <br><code><a href="config_image_data_format.html">config_image_data_format</a>()</code> <br><code><a href="config_is_interactive_logging_enabled.html">config_is_interactive_logging_enabled</a>()</code> <br><code><a href="config_is_traceback_filtering_enabled.html">config_is_traceback_filtering_enabled</a>()</code> <br><code><a href="config_set_backend.html">config_set_backend</a>()</code> <br><code><a href="config_set_dtype_policy.html">config_set_dtype_policy</a>()</code> <br><code><a href="config_set_epsilon.html">config_set_epsilon</a>()</code> <br><code><a href="config_set_floatx.html">config_set_floatx</a>()</code> <br><code><a href="config_set_image_data_format.html">config_set_image_data_format</a>()</code> <br></p></div>
+<p>Other config: <br><code><a href="config_backend.html">config_backend</a>()</code> <br><code><a href="config_disable_flash_attention.html">config_disable_flash_attention</a>()</code> <br><code><a href="config_disable_interactive_logging.html">config_disable_interactive_logging</a>()</code> <br><code><a href="config_disable_traceback_filtering.html">config_disable_traceback_filtering</a>()</code> <br><code><a href="config_dtype_policy.html">config_dtype_policy</a>()</code> <br><code><a href="config_enable_flash_attention.html">config_enable_flash_attention</a>()</code> <br><code><a href="config_enable_interactive_logging.html">config_enable_interactive_logging</a>()</code> <br><code><a href="config_enable_traceback_filtering.html">config_enable_traceback_filtering</a>()</code> <br><code><a href="config_enable_unsafe_deserialization.html">config_enable_unsafe_deserialization</a>()</code> <br><code><a href="config_floatx.html">config_floatx</a>()</code> <br><code><a href="config_image_data_format.html">config_image_data_format</a>()</code> <br><code><a href="config_is_interactive_logging_enabled.html">config_is_interactive_logging_enabled</a>()</code> <br><code><a href="config_is_nnx_enabled.html">config_is_nnx_enabled</a>()</code> <br><code><a href="config_is_traceback_filtering_enabled.html">config_is_traceback_filtering_enabled</a>()</code> <br><code><a href="config_max_epochs.html">config_max_epochs</a>()</code> <br><code><a href="config_set_backend.html">config_set_backend</a>()</code> <br><code><a href="config_set_dtype_policy.html">config_set_dtype_policy</a>()</code> <br><code><a href="config_set_epsilon.html">config_set_epsilon</a>()</code> <br><code><a href="config_set_floatx.html">config_set_floatx</a>()</code> <br><code><a href="config_set_image_data_format.html">config_set_image_data_format</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -90,7 +90,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/config_floatx.html b/docs/dev/reference/config_floatx.html
index 10f52d46b..d2312172e 100644
--- a/docs/dev/reference/config_floatx.html
+++ b/docs/dev/reference/config_floatx.html
@@ -8,7 +8,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -78,7 +78,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 <ul><li><p><a href="https://keras.io/api/utils/config_utils#floatx-function" class="external-link">https://keras.io/api/utils/config_utils#floatx-function</a></p></li>
 </ul><p>Other config backend: <br><code><a href="config_backend.html">config_backend</a>()</code> <br><code><a href="config_epsilon.html">config_epsilon</a>()</code> <br><code><a href="config_image_data_format.html">config_image_data_format</a>()</code> <br><code><a href="config_set_epsilon.html">config_set_epsilon</a>()</code> <br><code><a href="config_set_floatx.html">config_set_floatx</a>()</code> <br><code><a href="config_set_image_data_format.html">config_set_image_data_format</a>()</code> <br></p>
 <p>Other backend: <br><code><a href="clear_session.html">clear_session</a>()</code> <br><code><a href="config_backend.html">config_backend</a>()</code> <br><code><a href="config_epsilon.html">config_epsilon</a>()</code> <br><code><a href="config_image_data_format.html">config_image_data_format</a>()</code> <br><code><a href="config_set_epsilon.html">config_set_epsilon</a>()</code> <br><code><a href="config_set_floatx.html">config_set_floatx</a>()</code> <br><code><a href="config_set_image_data_format.html">config_set_image_data_format</a>()</code> <br></p>
-<p>Other config: <br><code><a href="config_backend.html">config_backend</a>()</code> <br><code><a href="config_disable_flash_attention.html">config_disable_flash_attention</a>()</code> <br><code><a href="config_disable_interactive_logging.html">config_disable_interactive_logging</a>()</code> <br><code><a href="config_disable_traceback_filtering.html">config_disable_traceback_filtering</a>()</code> <br><code><a href="config_dtype_policy.html">config_dtype_policy</a>()</code> <br><code><a href="config_enable_flash_attention.html">config_enable_flash_attention</a>()</code> <br><code><a href="config_enable_interactive_logging.html">config_enable_interactive_logging</a>()</code> <br><code><a href="config_enable_traceback_filtering.html">config_enable_traceback_filtering</a>()</code> <br><code><a href="config_enable_unsafe_deserialization.html">config_enable_unsafe_deserialization</a>()</code> <br><code><a href="config_epsilon.html">config_epsilon</a>()</code> <br><code><a href="config_image_data_format.html">config_image_data_format</a>()</code> <br><code><a href="config_is_interactive_logging_enabled.html">config_is_interactive_logging_enabled</a>()</code> <br><code><a href="config_is_traceback_filtering_enabled.html">config_is_traceback_filtering_enabled</a>()</code> <br><code><a href="config_set_backend.html">config_set_backend</a>()</code> <br><code><a href="config_set_dtype_policy.html">config_set_dtype_policy</a>()</code> <br><code><a href="config_set_epsilon.html">config_set_epsilon</a>()</code> <br><code><a href="config_set_floatx.html">config_set_floatx</a>()</code> <br><code><a href="config_set_image_data_format.html">config_set_image_data_format</a>()</code> <br></p></div>
+<p>Other config: <br><code><a href="config_backend.html">config_backend</a>()</code> <br><code><a href="config_disable_flash_attention.html">config_disable_flash_attention</a>()</code> <br><code><a href="config_disable_interactive_logging.html">config_disable_interactive_logging</a>()</code> <br><code><a href="config_disable_traceback_filtering.html">config_disable_traceback_filtering</a>()</code> <br><code><a href="config_dtype_policy.html">config_dtype_policy</a>()</code> <br><code><a href="config_enable_flash_attention.html">config_enable_flash_attention</a>()</code> <br><code><a href="config_enable_interactive_logging.html">config_enable_interactive_logging</a>()</code> <br><code><a href="config_enable_traceback_filtering.html">config_enable_traceback_filtering</a>()</code> <br><code><a href="config_enable_unsafe_deserialization.html">config_enable_unsafe_deserialization</a>()</code> <br><code><a href="config_epsilon.html">config_epsilon</a>()</code> <br><code><a href="config_image_data_format.html">config_image_data_format</a>()</code> <br><code><a href="config_is_interactive_logging_enabled.html">config_is_interactive_logging_enabled</a>()</code> <br><code><a href="config_is_nnx_enabled.html">config_is_nnx_enabled</a>()</code> <br><code><a href="config_is_traceback_filtering_enabled.html">config_is_traceback_filtering_enabled</a>()</code> <br><code><a href="config_max_epochs.html">config_max_epochs</a>()</code> <br><code><a href="config_set_backend.html">config_set_backend</a>()</code> <br><code><a href="config_set_dtype_policy.html">config_set_dtype_policy</a>()</code> <br><code><a href="config_set_epsilon.html">config_set_epsilon</a>()</code> <br><code><a href="config_set_floatx.html">config_set_floatx</a>()</code> <br><code><a href="config_set_image_data_format.html">config_set_image_data_format</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -90,7 +90,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/config_image_data_format.html b/docs/dev/reference/config_image_data_format.html
index 267b5577b..d25742827 100644
--- a/docs/dev/reference/config_image_data_format.html
+++ b/docs/dev/reference/config_image_data_format.html
@@ -8,7 +8,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -78,7 +78,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 <ul><li><p><a href="https://keras.io/api/utils/config_utils#imagedataformat-function" class="external-link">https://keras.io/api/utils/config_utils#imagedataformat-function</a></p></li>
 </ul><p>Other config backend: <br><code><a href="config_backend.html">config_backend</a>()</code> <br><code><a href="config_epsilon.html">config_epsilon</a>()</code> <br><code><a href="config_floatx.html">config_floatx</a>()</code> <br><code><a href="config_set_epsilon.html">config_set_epsilon</a>()</code> <br><code><a href="config_set_floatx.html">config_set_floatx</a>()</code> <br><code><a href="config_set_image_data_format.html">config_set_image_data_format</a>()</code> <br></p>
 <p>Other backend: <br><code><a href="clear_session.html">clear_session</a>()</code> <br><code><a href="config_backend.html">config_backend</a>()</code> <br><code><a href="config_epsilon.html">config_epsilon</a>()</code> <br><code><a href="config_floatx.html">config_floatx</a>()</code> <br><code><a href="config_set_epsilon.html">config_set_epsilon</a>()</code> <br><code><a href="config_set_floatx.html">config_set_floatx</a>()</code> <br><code><a href="config_set_image_data_format.html">config_set_image_data_format</a>()</code> <br></p>
-<p>Other config: <br><code><a href="config_backend.html">config_backend</a>()</code> <br><code><a href="config_disable_flash_attention.html">config_disable_flash_attention</a>()</code> <br><code><a href="config_disable_interactive_logging.html">config_disable_interactive_logging</a>()</code> <br><code><a href="config_disable_traceback_filtering.html">config_disable_traceback_filtering</a>()</code> <br><code><a href="config_dtype_policy.html">config_dtype_policy</a>()</code> <br><code><a href="config_enable_flash_attention.html">config_enable_flash_attention</a>()</code> <br><code><a href="config_enable_interactive_logging.html">config_enable_interactive_logging</a>()</code> <br><code><a href="config_enable_traceback_filtering.html">config_enable_traceback_filtering</a>()</code> <br><code><a href="config_enable_unsafe_deserialization.html">config_enable_unsafe_deserialization</a>()</code> <br><code><a href="config_epsilon.html">config_epsilon</a>()</code> <br><code><a href="config_floatx.html">config_floatx</a>()</code> <br><code><a href="config_is_interactive_logging_enabled.html">config_is_interactive_logging_enabled</a>()</code> <br><code><a href="config_is_traceback_filtering_enabled.html">config_is_traceback_filtering_enabled</a>()</code> <br><code><a href="config_set_backend.html">config_set_backend</a>()</code> <br><code><a href="config_set_dtype_policy.html">config_set_dtype_policy</a>()</code> <br><code><a href="config_set_epsilon.html">config_set_epsilon</a>()</code> <br><code><a href="config_set_floatx.html">config_set_floatx</a>()</code> <br><code><a href="config_set_image_data_format.html">config_set_image_data_format</a>()</code> <br></p></div>
+<p>Other config: <br><code><a href="config_backend.html">config_backend</a>()</code> <br><code><a href="config_disable_flash_attention.html">config_disable_flash_attention</a>()</code> <br><code><a href="config_disable_interactive_logging.html">config_disable_interactive_logging</a>()</code> <br><code><a href="config_disable_traceback_filtering.html">config_disable_traceback_filtering</a>()</code> <br><code><a href="config_dtype_policy.html">config_dtype_policy</a>()</code> <br><code><a href="config_enable_flash_attention.html">config_enable_flash_attention</a>()</code> <br><code><a href="config_enable_interactive_logging.html">config_enable_interactive_logging</a>()</code> <br><code><a href="config_enable_traceback_filtering.html">config_enable_traceback_filtering</a>()</code> <br><code><a href="config_enable_unsafe_deserialization.html">config_enable_unsafe_deserialization</a>()</code> <br><code><a href="config_epsilon.html">config_epsilon</a>()</code> <br><code><a href="config_floatx.html">config_floatx</a>()</code> <br><code><a href="config_is_interactive_logging_enabled.html">config_is_interactive_logging_enabled</a>()</code> <br><code><a href="config_is_nnx_enabled.html">config_is_nnx_enabled</a>()</code> <br><code><a href="config_is_traceback_filtering_enabled.html">config_is_traceback_filtering_enabled</a>()</code> <br><code><a href="config_max_epochs.html">config_max_epochs</a>()</code> <br><code><a href="config_set_backend.html">config_set_backend</a>()</code> <br><code><a href="config_set_dtype_policy.html">config_set_dtype_policy</a>()</code> <br><code><a href="config_set_epsilon.html">config_set_epsilon</a>()</code> <br><code><a href="config_set_floatx.html">config_set_floatx</a>()</code> <br><code><a href="config_set_image_data_format.html">config_set_image_data_format</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -90,7 +90,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/config_is_flash_attention_enabled.html b/docs/dev/reference/config_is_flash_attention_enabled.html
index 364011505..584c3f7f3 100644
--- a/docs/dev/reference/config_is_flash_attention_enabled.html
+++ b/docs/dev/reference/config_is_flash_attention_enabled.html
@@ -22,7 +22,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -101,7 +101,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/config_is_interactive_logging_enabled.html b/docs/dev/reference/config_is_interactive_logging_enabled.html
index 0d535bc1e..44c6df373 100644
--- a/docs/dev/reference/config_is_interactive_logging_enabled.html
+++ b/docs/dev/reference/config_is_interactive_logging_enabled.html
@@ -12,7 +12,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -77,7 +77,7 @@ <h2 id="value">Value<a class="anchor" aria-label="anchor" href="#value"></a></h2
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
     <div class="dont-index"><p>Other io utils: <br><code><a href="config_disable_interactive_logging.html">config_disable_interactive_logging</a>()</code> <br><code><a href="config_enable_interactive_logging.html">config_enable_interactive_logging</a>()</code> <br></p>
 <p>Other utils: <br><code><a href="audio_dataset_from_directory.html">audio_dataset_from_directory</a>()</code> <br><code><a href="clear_session.html">clear_session</a>()</code> <br><code><a href="config_disable_interactive_logging.html">config_disable_interactive_logging</a>()</code> <br><code><a href="config_disable_traceback_filtering.html">config_disable_traceback_filtering</a>()</code> <br><code><a href="config_enable_interactive_logging.html">config_enable_interactive_logging</a>()</code> <br><code><a href="config_enable_traceback_filtering.html">config_enable_traceback_filtering</a>()</code> <br><code><a href="config_is_traceback_filtering_enabled.html">config_is_traceback_filtering_enabled</a>()</code> <br><code><a href="get_file.html">get_file</a>()</code> <br><code><a href="get_source_inputs.html">get_source_inputs</a>()</code> <br><code><a href="image_array_save.html">image_array_save</a>()</code> <br><code><a href="image_dataset_from_directory.html">image_dataset_from_directory</a>()</code> <br><code><a href="image_from_array.html">image_from_array</a>()</code> <br><code><a href="image_load.html">image_load</a>()</code> <br><code><a href="image_smart_resize.html">image_smart_resize</a>()</code> <br><code><a href="image_to_array.html">image_to_array</a>()</code> <br><code><a href="layer_feature_space.html">layer_feature_space</a>()</code> <br><code><a href="normalize.html">normalize</a>()</code> <br><code><a href="pad_sequences.html">pad_sequences</a>()</code> <br><code><a href="set_random_seed.html">set_random_seed</a>()</code> <br><code><a href="split_dataset.html">split_dataset</a>()</code> <br><code><a href="text_dataset_from_directory.html">text_dataset_from_directory</a>()</code> <br><code><a href="timeseries_dataset_from_array.html">timeseries_dataset_from_array</a>()</code> <br><code><a href="to_categorical.html">to_categorical</a>()</code> <br><code><a href="zip_lists.html">zip_lists</a>()</code> <br></p>
-<p>Other config: <br><code><a href="config_backend.html">config_backend</a>()</code> <br><code><a href="config_disable_flash_attention.html">config_disable_flash_attention</a>()</code> <br><code><a href="config_disable_interactive_logging.html">config_disable_interactive_logging</a>()</code> <br><code><a href="config_disable_traceback_filtering.html">config_disable_traceback_filtering</a>()</code> <br><code><a href="config_dtype_policy.html">config_dtype_policy</a>()</code> <br><code><a href="config_enable_flash_attention.html">config_enable_flash_attention</a>()</code> <br><code><a href="config_enable_interactive_logging.html">config_enable_interactive_logging</a>()</code> <br><code><a href="config_enable_traceback_filtering.html">config_enable_traceback_filtering</a>()</code> <br><code><a href="config_enable_unsafe_deserialization.html">config_enable_unsafe_deserialization</a>()</code> <br><code><a href="config_epsilon.html">config_epsilon</a>()</code> <br><code><a href="config_floatx.html">config_floatx</a>()</code> <br><code><a href="config_image_data_format.html">config_image_data_format</a>()</code> <br><code><a href="config_is_traceback_filtering_enabled.html">config_is_traceback_filtering_enabled</a>()</code> <br><code><a href="config_set_backend.html">config_set_backend</a>()</code> <br><code><a href="config_set_dtype_policy.html">config_set_dtype_policy</a>()</code> <br><code><a href="config_set_epsilon.html">config_set_epsilon</a>()</code> <br><code><a href="config_set_floatx.html">config_set_floatx</a>()</code> <br><code><a href="config_set_image_data_format.html">config_set_image_data_format</a>()</code> <br></p></div>
+<p>Other config: <br><code><a href="config_backend.html">config_backend</a>()</code> <br><code><a href="config_disable_flash_attention.html">config_disable_flash_attention</a>()</code> <br><code><a href="config_disable_interactive_logging.html">config_disable_interactive_logging</a>()</code> <br><code><a href="config_disable_traceback_filtering.html">config_disable_traceback_filtering</a>()</code> <br><code><a href="config_dtype_policy.html">config_dtype_policy</a>()</code> <br><code><a href="config_enable_flash_attention.html">config_enable_flash_attention</a>()</code> <br><code><a href="config_enable_interactive_logging.html">config_enable_interactive_logging</a>()</code> <br><code><a href="config_enable_traceback_filtering.html">config_enable_traceback_filtering</a>()</code> <br><code><a href="config_enable_unsafe_deserialization.html">config_enable_unsafe_deserialization</a>()</code> <br><code><a href="config_epsilon.html">config_epsilon</a>()</code> <br><code><a href="config_floatx.html">config_floatx</a>()</code> <br><code><a href="config_image_data_format.html">config_image_data_format</a>()</code> <br><code><a href="config_is_nnx_enabled.html">config_is_nnx_enabled</a>()</code> <br><code><a href="config_is_traceback_filtering_enabled.html">config_is_traceback_filtering_enabled</a>()</code> <br><code><a href="config_max_epochs.html">config_max_epochs</a>()</code> <br><code><a href="config_set_backend.html">config_set_backend</a>()</code> <br><code><a href="config_set_dtype_policy.html">config_set_dtype_policy</a>()</code> <br><code><a href="config_set_epsilon.html">config_set_epsilon</a>()</code> <br><code><a href="config_set_floatx.html">config_set_floatx</a>()</code> <br><code><a href="config_set_image_data_format.html">config_set_image_data_format</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -89,7 +89,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/config_is_nnx_enabled.html b/docs/dev/reference/config_is_nnx_enabled.html
new file mode 100644
index 000000000..1b0e93dfb
--- /dev/null
+++ b/docs/dev/reference/config_is_nnx_enabled.html
@@ -0,0 +1,93 @@
+<!DOCTYPE html>
+<!-- Generated by pkgdown: do not edit by hand --><html lang="en"><head><meta http-equiv="Content-Type" content="text/html; charset=UTF-8"><meta charset="utf-8"><meta http-equiv="X-UA-Compatible" content="IE=edge"><meta name="viewport" content="width=device-width, initial-scale=1, shrink-to-fit=no"><title>Check whether NNX-specific features are enabled on the JAX backend. — config_is_nnx_enabled • keras3</title><!-- favicons --><link rel="icon" type="image/png" sizes="96x96" href="../favicon-96x96.png"><link rel="icon" type="”image/svg+xml”" href="../favicon.svg"><link rel="apple-touch-icon" sizes="180x180" href="../apple-touch-icon.png"><link rel="icon" sizes="any" href="../favicon.ico"><link rel="manifest" href="../site.webmanifest"><script src="../deps/jquery-3.6.0/jquery-3.6.0.min.js"></script><meta name="viewport" content="width=device-width, initial-scale=1, shrink-to-fit=no"><link href="../deps/bootstrap-5.3.1/bootstrap.min.css" rel="stylesheet"><script src="../deps/bootstrap-5.3.1/bootstrap.bundle.min.js"></script><link href="../deps/font-awesome-6.5.2/css/all.min.css" rel="stylesheet"><link href="../deps/font-awesome-6.5.2/css/v4-shims.min.css" rel="stylesheet"><script src="../deps/headroom-0.11.0/headroom.min.js"></script><script src="../deps/headroom-0.11.0/jQuery.headroom.min.js"></script><script src="../deps/bootstrap-toc-1.0.1/bootstrap-toc.min.js"></script><script src="../deps/clipboard.js-2.0.11/clipboard.min.js"></script><script src="../deps/search-1.0.0/autocomplete.jquery.min.js"></script><script src="../deps/search-1.0.0/fuse.min.js"></script><script src="../deps/search-1.0.0/mark.min.js"></script><!-- pkgdown --><script src="../pkgdown.js"></script><link href="../extra.css" rel="stylesheet"><meta property="og:title" content="Check whether NNX-specific features are enabled on the JAX backend. — config_is_nnx_enabled"><meta name="description" content="Check whether NNX-specific features are enabled on the JAX backend."><meta property="og:description" content="Check whether NNX-specific features are enabled on the JAX backend."><meta name="robots" content="noindex"><!-- Google Tag Manager --><script>(function(w,d,s,l,i){w[l]=w[l]||[];w[l].push({'gtm.start': new Date().getTime(),event:'gtm.js'});var f=d.getElementsByTagName(s)[0],j=d.createElement(s),dl=l!='dataLayer'?'&l='+l:'';j.async=true;j.src='https://www.googletagmanager.com/gtm.js?id='+i+dl;f.parentNode.insertBefore(j,f);})(window,document,'script','dataLayer','GTM-KHBDBW7');</script><!-- End Google Tag Manager --></head><body>
+    <a href="#main" class="visually-hidden-focusable">Skip to contents</a>
+    <!-- Google Tag Manager (noscript) --> <noscript><iframe src="https://www.googletagmanager.com/ns.html?id=GTM-KHBDBW7" height="0" width="0" style="display:none;visibility:hidden"></iframe></noscript> <!-- End Google Tag Manager (noscript) -->
+
+
+    <nav class="navbar navbar-expand-lg fixed-top bg-primary" data-bs-theme="inverse" aria-label="Site navigation"><div class="container">
+
+    <a class="navbar-brand me-2" href="../index.html">keras3</a>
+
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
+
+
+    <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
+      <span class="navbar-toggler-icon"></span>
+    </button>
+
+    <div id="navbar" class="collapse navbar-collapse ms-3">
+      <ul class="navbar-nav me-auto"><li class="nav-item"><a class="nav-link" href="../articles/getting_started.html">Getting Started</a></li>
+<li class="nav-item dropdown">
+  <button class="nav-link dropdown-toggle" type="button" id="dropdown-guides" data-bs-toggle="dropdown" aria-expanded="false" aria-haspopup="true">Guides</button>
+  <ul class="dropdown-menu" aria-labelledby="dropdown-guides"><li><h6 class="dropdown-header" data-toc-skip>Model definition</h6></li>
+    <li><a class="dropdown-item" href="../articles/sequential_model.html">Sequential Model</a></li>
+    <li><a class="dropdown-item" href="../articles/functional_api.html">Functional API</a></li>
+    <li><h6 class="dropdown-header" data-toc-skip>Extending and customizing</h6></li>
+    <li><a class="dropdown-item" href="../articles/training_with_built_in_methods.html">Training &amp; evaluation with the built-in methods</a></li>
+    <li><a class="dropdown-item" href="../articles/custom_train_step_in_tensorflow.html">Customizing `fit()` with Tensorflow</a></li>
+    <li><a class="dropdown-item" href="../articles/writing_your_own_callbacks.html">Writing your own callbacks</a></li>
+    <li><a class="dropdown-item" href="../articles/making_new_layers_and_models_via_subclassing.html">Making new layers and models via subclassing</a></li>
+    <li><a class="dropdown-item" href="../articles/writing_a_custom_training_loop_in_tensorflow.html">Writing a training loop from scratch in TensorFlow</a></li>
+    <li><a class="dropdown-item" href="../articles/serialization_and_saving.html">Serialization and Saving</a></li>
+    <li><h6 class="dropdown-header" data-toc-skip>Other topics</h6></li>
+    <li><a class="dropdown-item" href="../articles/transfer_learning.html">Transfer learning and fine tuning</a></li>
+    <li><a class="dropdown-item" href="../articles/distributed_training_with_tensorflow.html">Distributed training with TensorFlow</a></li>
+    <li><a class="dropdown-item" href="../articles/distribution.html">Distributed training with Jax</a></li>
+  </ul></li>
+<li class="nav-item"><a class="nav-link" href="../articles/examples/index.html">Examples</a></li>
+<li class="active nav-item"><a class="nav-link" href="../reference/index.html">Reference</a></li>
+<li class="nav-item"><a class="nav-link" href="../news/index.html">News</a></li>
+      </ul><ul class="navbar-nav"><li class="nav-item"><form class="form-inline" role="search">
+ <input class="form-control" type="search" name="search-input" id="search-input" autocomplete="off" aria-label="Search site" placeholder="Search for" data-search-index="../search.json"></form></li>
+<li class="nav-item"><a class="external-link nav-link" href="https://github.com/rstudio/keras3/" aria-label="GitHub"><span class="fa fab fa-github fa-lg"></span></a></li>
+      </ul></div>
+
+
+  </div>
+</nav><div class="container template-reference-topic">
+<div class="row">
+  <main id="main" class="col-md-9"><div class="page-header">
+
+      <h1>Check whether NNX-specific features are enabled on the JAX backend.</h1>
+      <small class="dont-index">Source: <a href="https://github.com/rstudio/keras3/blob/HEAD/R/config.R" class="external-link"><code>R/config.R</code></a></small>
+      <div class="d-none name"><code>config_is_nnx_enabled.Rd</code></div>
+    </div>
+
+    <div class="ref-description section level2">
+    <p>Check whether NNX-specific features are enabled on the JAX backend.</p>
+    </div>
+
+    <div class="section level2">
+    <h2 id="ref-usage">Usage<a class="anchor" aria-label="anchor" href="#ref-usage"></a></h2>
+    <div class="sourceCode"><pre class="sourceCode r"><code><span><span class="fu">config_is_nnx_enabled</span><span class="op">(</span><span class="op">)</span></span></code></pre></div>
+    </div>
+
+    <div class="section level2">
+    <h2 id="value">Value<a class="anchor" aria-label="anchor" href="#value"></a></h2>
+    <p>Logical flag; <code>TRUE</code> if NNX backend features are enabled, <code>FALSE</code> otherwise.</p>
+    </div>
+    <div class="section level2">
+    <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
+    <div class="dont-index"><p>Other config: <br><code><a href="config_backend.html">config_backend</a>()</code> <br><code><a href="config_disable_flash_attention.html">config_disable_flash_attention</a>()</code> <br><code><a href="config_disable_interactive_logging.html">config_disable_interactive_logging</a>()</code> <br><code><a href="config_disable_traceback_filtering.html">config_disable_traceback_filtering</a>()</code> <br><code><a href="config_dtype_policy.html">config_dtype_policy</a>()</code> <br><code><a href="config_enable_flash_attention.html">config_enable_flash_attention</a>()</code> <br><code><a href="config_enable_interactive_logging.html">config_enable_interactive_logging</a>()</code> <br><code><a href="config_enable_traceback_filtering.html">config_enable_traceback_filtering</a>()</code> <br><code><a href="config_enable_unsafe_deserialization.html">config_enable_unsafe_deserialization</a>()</code> <br><code><a href="config_epsilon.html">config_epsilon</a>()</code> <br><code><a href="config_floatx.html">config_floatx</a>()</code> <br><code><a href="config_image_data_format.html">config_image_data_format</a>()</code> <br><code><a href="config_is_interactive_logging_enabled.html">config_is_interactive_logging_enabled</a>()</code> <br><code><a href="config_is_traceback_filtering_enabled.html">config_is_traceback_filtering_enabled</a>()</code> <br><code><a href="config_max_epochs.html">config_max_epochs</a>()</code> <br><code><a href="config_set_backend.html">config_set_backend</a>()</code> <br><code><a href="config_set_dtype_policy.html">config_set_dtype_policy</a>()</code> <br><code><a href="config_set_epsilon.html">config_set_epsilon</a>()</code> <br><code><a href="config_set_floatx.html">config_set_floatx</a>()</code> <br><code><a href="config_set_image_data_format.html">config_set_image_data_format</a>()</code> <br></p></div>
+    </div>
+
+  </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
+    </nav></aside></div>
+
+
+    <footer><div class="pkgdown-footer-left">
+  <p>Developed by Tomasz Kalinowski, JJ Allaire, François Chollet, Posit Software, PBC, Google.</p>
+</div>
+
+<div class="pkgdown-footer-right">
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
+</div>
+
+    </footer></div>
+
+
+
+
+
+  </body></html>
+
diff --git a/docs/dev/reference/config_is_traceback_filtering_enabled.html b/docs/dev/reference/config_is_traceback_filtering_enabled.html
index f965d2bcf..46a696412 100644
--- a/docs/dev/reference/config_is_traceback_filtering_enabled.html
+++ b/docs/dev/reference/config_is_traceback_filtering_enabled.html
@@ -28,7 +28,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -101,7 +101,7 @@ <h2 id="value">Value<a class="anchor" aria-label="anchor" href="#value"></a></h2
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
     <div class="dont-index"><p>Other traceback utils: <br><code><a href="config_disable_traceback_filtering.html">config_disable_traceback_filtering</a>()</code> <br><code><a href="config_enable_traceback_filtering.html">config_enable_traceback_filtering</a>()</code> <br></p>
 <p>Other utils: <br><code><a href="audio_dataset_from_directory.html">audio_dataset_from_directory</a>()</code> <br><code><a href="clear_session.html">clear_session</a>()</code> <br><code><a href="config_disable_interactive_logging.html">config_disable_interactive_logging</a>()</code> <br><code><a href="config_disable_traceback_filtering.html">config_disable_traceback_filtering</a>()</code> <br><code><a href="config_enable_interactive_logging.html">config_enable_interactive_logging</a>()</code> <br><code><a href="config_enable_traceback_filtering.html">config_enable_traceback_filtering</a>()</code> <br><code><a href="config_is_interactive_logging_enabled.html">config_is_interactive_logging_enabled</a>()</code> <br><code><a href="get_file.html">get_file</a>()</code> <br><code><a href="get_source_inputs.html">get_source_inputs</a>()</code> <br><code><a href="image_array_save.html">image_array_save</a>()</code> <br><code><a href="image_dataset_from_directory.html">image_dataset_from_directory</a>()</code> <br><code><a href="image_from_array.html">image_from_array</a>()</code> <br><code><a href="image_load.html">image_load</a>()</code> <br><code><a href="image_smart_resize.html">image_smart_resize</a>()</code> <br><code><a href="image_to_array.html">image_to_array</a>()</code> <br><code><a href="layer_feature_space.html">layer_feature_space</a>()</code> <br><code><a href="normalize.html">normalize</a>()</code> <br><code><a href="pad_sequences.html">pad_sequences</a>()</code> <br><code><a href="set_random_seed.html">set_random_seed</a>()</code> <br><code><a href="split_dataset.html">split_dataset</a>()</code> <br><code><a href="text_dataset_from_directory.html">text_dataset_from_directory</a>()</code> <br><code><a href="timeseries_dataset_from_array.html">timeseries_dataset_from_array</a>()</code> <br><code><a href="to_categorical.html">to_categorical</a>()</code> <br><code><a href="zip_lists.html">zip_lists</a>()</code> <br></p>
-<p>Other config: <br><code><a href="config_backend.html">config_backend</a>()</code> <br><code><a href="config_disable_flash_attention.html">config_disable_flash_attention</a>()</code> <br><code><a href="config_disable_interactive_logging.html">config_disable_interactive_logging</a>()</code> <br><code><a href="config_disable_traceback_filtering.html">config_disable_traceback_filtering</a>()</code> <br><code><a href="config_dtype_policy.html">config_dtype_policy</a>()</code> <br><code><a href="config_enable_flash_attention.html">config_enable_flash_attention</a>()</code> <br><code><a href="config_enable_interactive_logging.html">config_enable_interactive_logging</a>()</code> <br><code><a href="config_enable_traceback_filtering.html">config_enable_traceback_filtering</a>()</code> <br><code><a href="config_enable_unsafe_deserialization.html">config_enable_unsafe_deserialization</a>()</code> <br><code><a href="config_epsilon.html">config_epsilon</a>()</code> <br><code><a href="config_floatx.html">config_floatx</a>()</code> <br><code><a href="config_image_data_format.html">config_image_data_format</a>()</code> <br><code><a href="config_is_interactive_logging_enabled.html">config_is_interactive_logging_enabled</a>()</code> <br><code><a href="config_set_backend.html">config_set_backend</a>()</code> <br><code><a href="config_set_dtype_policy.html">config_set_dtype_policy</a>()</code> <br><code><a href="config_set_epsilon.html">config_set_epsilon</a>()</code> <br><code><a href="config_set_floatx.html">config_set_floatx</a>()</code> <br><code><a href="config_set_image_data_format.html">config_set_image_data_format</a>()</code> <br></p></div>
+<p>Other config: <br><code><a href="config_backend.html">config_backend</a>()</code> <br><code><a href="config_disable_flash_attention.html">config_disable_flash_attention</a>()</code> <br><code><a href="config_disable_interactive_logging.html">config_disable_interactive_logging</a>()</code> <br><code><a href="config_disable_traceback_filtering.html">config_disable_traceback_filtering</a>()</code> <br><code><a href="config_dtype_policy.html">config_dtype_policy</a>()</code> <br><code><a href="config_enable_flash_attention.html">config_enable_flash_attention</a>()</code> <br><code><a href="config_enable_interactive_logging.html">config_enable_interactive_logging</a>()</code> <br><code><a href="config_enable_traceback_filtering.html">config_enable_traceback_filtering</a>()</code> <br><code><a href="config_enable_unsafe_deserialization.html">config_enable_unsafe_deserialization</a>()</code> <br><code><a href="config_epsilon.html">config_epsilon</a>()</code> <br><code><a href="config_floatx.html">config_floatx</a>()</code> <br><code><a href="config_image_data_format.html">config_image_data_format</a>()</code> <br><code><a href="config_is_interactive_logging_enabled.html">config_is_interactive_logging_enabled</a>()</code> <br><code><a href="config_is_nnx_enabled.html">config_is_nnx_enabled</a>()</code> <br><code><a href="config_max_epochs.html">config_max_epochs</a>()</code> <br><code><a href="config_set_backend.html">config_set_backend</a>()</code> <br><code><a href="config_set_dtype_policy.html">config_set_dtype_policy</a>()</code> <br><code><a href="config_set_epsilon.html">config_set_epsilon</a>()</code> <br><code><a href="config_set_floatx.html">config_set_floatx</a>()</code> <br><code><a href="config_set_image_data_format.html">config_set_image_data_format</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -113,7 +113,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/config_max_epochs.html b/docs/dev/reference/config_max_epochs.html
new file mode 100644
index 000000000..ee0efa900
--- /dev/null
+++ b/docs/dev/reference/config_max_epochs.html
@@ -0,0 +1,124 @@
+<!DOCTYPE html>
+<!-- Generated by pkgdown: do not edit by hand --><html lang="en"><head><meta http-equiv="Content-Type" content="text/html; charset=UTF-8"><meta charset="utf-8"><meta http-equiv="X-UA-Compatible" content="IE=edge"><meta name="viewport" content="width=device-width, initial-scale=1, shrink-to-fit=no"><title>Configure the default training loop limits. — config_max_epochs • keras3</title><!-- favicons --><link rel="icon" type="image/png" sizes="96x96" href="../favicon-96x96.png"><link rel="icon" type="”image/svg+xml”" href="../favicon.svg"><link rel="apple-touch-icon" sizes="180x180" href="../apple-touch-icon.png"><link rel="icon" sizes="any" href="../favicon.ico"><link rel="manifest" href="../site.webmanifest"><script src="../deps/jquery-3.6.0/jquery-3.6.0.min.js"></script><meta name="viewport" content="width=device-width, initial-scale=1, shrink-to-fit=no"><link href="../deps/bootstrap-5.3.1/bootstrap.min.css" rel="stylesheet"><script src="../deps/bootstrap-5.3.1/bootstrap.bundle.min.js"></script><link href="../deps/font-awesome-6.5.2/css/all.min.css" rel="stylesheet"><link href="../deps/font-awesome-6.5.2/css/v4-shims.min.css" rel="stylesheet"><script src="../deps/headroom-0.11.0/headroom.min.js"></script><script src="../deps/headroom-0.11.0/jQuery.headroom.min.js"></script><script src="../deps/bootstrap-toc-1.0.1/bootstrap-toc.min.js"></script><script src="../deps/clipboard.js-2.0.11/clipboard.min.js"></script><script src="../deps/search-1.0.0/autocomplete.jquery.min.js"></script><script src="../deps/search-1.0.0/fuse.min.js"></script><script src="../deps/search-1.0.0/mark.min.js"></script><!-- pkgdown --><script src="../pkgdown.js"></script><link href="../extra.css" rel="stylesheet"><meta property="og:title" content="Configure the default training loop limits. — config_max_epochs"><meta name="description" content="These helpers control the caps that built-in training loops obey when running
+fit(), evaluate(), or predict(). The values can also be provided via the
+KERAS_MAX_EPOCHS or KERAS_MAX_STEPS_PER_EPOCH environment variables to
+quickly constrain a run without modifying source code."><meta property="og:description" content="These helpers control the caps that built-in training loops obey when running
+fit(), evaluate(), or predict(). The values can also be provided via the
+KERAS_MAX_EPOCHS or KERAS_MAX_STEPS_PER_EPOCH environment variables to
+quickly constrain a run without modifying source code."><meta name="robots" content="noindex"><!-- Google Tag Manager --><script>(function(w,d,s,l,i){w[l]=w[l]||[];w[l].push({'gtm.start': new Date().getTime(),event:'gtm.js'});var f=d.getElementsByTagName(s)[0],j=d.createElement(s),dl=l!='dataLayer'?'&l='+l:'';j.async=true;j.src='https://www.googletagmanager.com/gtm.js?id='+i+dl;f.parentNode.insertBefore(j,f);})(window,document,'script','dataLayer','GTM-KHBDBW7');</script><!-- End Google Tag Manager --></head><body>
+    <a href="#main" class="visually-hidden-focusable">Skip to contents</a>
+    <!-- Google Tag Manager (noscript) --> <noscript><iframe src="https://www.googletagmanager.com/ns.html?id=GTM-KHBDBW7" height="0" width="0" style="display:none;visibility:hidden"></iframe></noscript> <!-- End Google Tag Manager (noscript) -->
+
+
+    <nav class="navbar navbar-expand-lg fixed-top bg-primary" data-bs-theme="inverse" aria-label="Site navigation"><div class="container">
+
+    <a class="navbar-brand me-2" href="../index.html">keras3</a>
+
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
+
+
+    <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
+      <span class="navbar-toggler-icon"></span>
+    </button>
+
+    <div id="navbar" class="collapse navbar-collapse ms-3">
+      <ul class="navbar-nav me-auto"><li class="nav-item"><a class="nav-link" href="../articles/getting_started.html">Getting Started</a></li>
+<li class="nav-item dropdown">
+  <button class="nav-link dropdown-toggle" type="button" id="dropdown-guides" data-bs-toggle="dropdown" aria-expanded="false" aria-haspopup="true">Guides</button>
+  <ul class="dropdown-menu" aria-labelledby="dropdown-guides"><li><h6 class="dropdown-header" data-toc-skip>Model definition</h6></li>
+    <li><a class="dropdown-item" href="../articles/sequential_model.html">Sequential Model</a></li>
+    <li><a class="dropdown-item" href="../articles/functional_api.html">Functional API</a></li>
+    <li><h6 class="dropdown-header" data-toc-skip>Extending and customizing</h6></li>
+    <li><a class="dropdown-item" href="../articles/training_with_built_in_methods.html">Training &amp; evaluation with the built-in methods</a></li>
+    <li><a class="dropdown-item" href="../articles/custom_train_step_in_tensorflow.html">Customizing `fit()` with Tensorflow</a></li>
+    <li><a class="dropdown-item" href="../articles/writing_your_own_callbacks.html">Writing your own callbacks</a></li>
+    <li><a class="dropdown-item" href="../articles/making_new_layers_and_models_via_subclassing.html">Making new layers and models via subclassing</a></li>
+    <li><a class="dropdown-item" href="../articles/writing_a_custom_training_loop_in_tensorflow.html">Writing a training loop from scratch in TensorFlow</a></li>
+    <li><a class="dropdown-item" href="../articles/serialization_and_saving.html">Serialization and Saving</a></li>
+    <li><h6 class="dropdown-header" data-toc-skip>Other topics</h6></li>
+    <li><a class="dropdown-item" href="../articles/transfer_learning.html">Transfer learning and fine tuning</a></li>
+    <li><a class="dropdown-item" href="../articles/distributed_training_with_tensorflow.html">Distributed training with TensorFlow</a></li>
+    <li><a class="dropdown-item" href="../articles/distribution.html">Distributed training with Jax</a></li>
+  </ul></li>
+<li class="nav-item"><a class="nav-link" href="../articles/examples/index.html">Examples</a></li>
+<li class="active nav-item"><a class="nav-link" href="../reference/index.html">Reference</a></li>
+<li class="nav-item"><a class="nav-link" href="../news/index.html">News</a></li>
+      </ul><ul class="navbar-nav"><li class="nav-item"><form class="form-inline" role="search">
+ <input class="form-control" type="search" name="search-input" id="search-input" autocomplete="off" aria-label="Search site" placeholder="Search for" data-search-index="../search.json"></form></li>
+<li class="nav-item"><a class="external-link nav-link" href="https://github.com/rstudio/keras3/" aria-label="GitHub"><span class="fa fab fa-github fa-lg"></span></a></li>
+      </ul></div>
+
+
+  </div>
+</nav><div class="container template-reference-topic">
+<div class="row">
+  <main id="main" class="col-md-9"><div class="page-header">
+
+      <h1>Configure the default training loop limits.</h1>
+      <small class="dont-index">Source: <a href="https://github.com/rstudio/keras3/blob/HEAD/R/config.R" class="external-link"><code>R/config.R</code></a></small>
+      <div class="d-none name"><code>config_max_epochs.Rd</code></div>
+    </div>
+
+    <div class="ref-description section level2">
+    <p>These helpers control the caps that built-in training loops obey when running
+<code><a href="https://generics.r-lib.org/reference/fit.html" class="external-link">fit()</a></code>, <code><a href="https://rdrr.io/pkg/tensorflow/man/evaluate.html" class="external-link">evaluate()</a></code>, or <code><a href="https://rdrr.io/r/stats/predict.html" class="external-link">predict()</a></code>. The values can also be provided via the
+<code>KERAS_MAX_EPOCHS</code> or <code>KERAS_MAX_STEPS_PER_EPOCH</code> environment variables to
+quickly constrain a run without modifying source code.</p>
+    </div>
+
+    <div class="section level2">
+    <h2 id="ref-usage">Usage<a class="anchor" aria-label="anchor" href="#ref-usage"></a></h2>
+    <div class="sourceCode"><pre class="sourceCode r"><code><span><span class="fu">config_max_epochs</span><span class="op">(</span><span class="op">)</span></span>
+<span></span>
+<span><span class="fu">config_set_max_epochs</span><span class="op">(</span><span class="va">max_epochs</span><span class="op">)</span></span>
+<span></span>
+<span><span class="fu">config_max_steps_per_epoch</span><span class="op">(</span><span class="op">)</span></span>
+<span></span>
+<span><span class="fu">config_set_max_steps_per_epoch</span><span class="op">(</span><span class="va">max_steps_per_epoch</span><span class="op">)</span></span></code></pre></div>
+    </div>
+
+    <div class="section level2">
+    <h2 id="arguments">Arguments<a class="anchor" aria-label="anchor" href="#arguments"></a></h2>
+
+
+<dl><dt id="arg-max-epochs">max_epochs<a class="anchor" aria-label="anchor" href="#arg-max-epochs"></a></dt>
+<dd><p>Integer upper bound for epochs processed by built-in training loops. Use
+<code>NULL</code> to remove the cap.</p></dd>
+
+
+<dt id="arg-max-steps-per-epoch">max_steps_per_epoch<a class="anchor" aria-label="anchor" href="#arg-max-steps-per-epoch"></a></dt>
+<dd><p>Integer upper bound for steps processed per epoch by built-in training
+loops. Use <code>NULL</code> to remove the cap.</p></dd>
+
+</dl></div>
+    <div class="section level2">
+    <h2 id="value">Value<a class="anchor" aria-label="anchor" href="#value"></a></h2>
+    <p><code>config_max_epochs()</code> and <code>config_max_steps_per_epoch()</code> return the current
+integer limits (or <code>NULL</code> if the cap is unset). The setter variants return
+<code>NULL</code> invisibly and are called for side effects.</p>
+    </div>
+    <div class="section level2">
+    <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
+    <div class="dont-index"><p>Other config: <br><code><a href="config_backend.html">config_backend</a>()</code> <br><code><a href="config_disable_flash_attention.html">config_disable_flash_attention</a>()</code> <br><code><a href="config_disable_interactive_logging.html">config_disable_interactive_logging</a>()</code> <br><code><a href="config_disable_traceback_filtering.html">config_disable_traceback_filtering</a>()</code> <br><code><a href="config_dtype_policy.html">config_dtype_policy</a>()</code> <br><code><a href="config_enable_flash_attention.html">config_enable_flash_attention</a>()</code> <br><code><a href="config_enable_interactive_logging.html">config_enable_interactive_logging</a>()</code> <br><code><a href="config_enable_traceback_filtering.html">config_enable_traceback_filtering</a>()</code> <br><code><a href="config_enable_unsafe_deserialization.html">config_enable_unsafe_deserialization</a>()</code> <br><code><a href="config_epsilon.html">config_epsilon</a>()</code> <br><code><a href="config_floatx.html">config_floatx</a>()</code> <br><code><a href="config_image_data_format.html">config_image_data_format</a>()</code> <br><code><a href="config_is_interactive_logging_enabled.html">config_is_interactive_logging_enabled</a>()</code> <br><code><a href="config_is_nnx_enabled.html">config_is_nnx_enabled</a>()</code> <br><code><a href="config_is_traceback_filtering_enabled.html">config_is_traceback_filtering_enabled</a>()</code> <br><code><a href="config_set_backend.html">config_set_backend</a>()</code> <br><code><a href="config_set_dtype_policy.html">config_set_dtype_policy</a>()</code> <br><code><a href="config_set_epsilon.html">config_set_epsilon</a>()</code> <br><code><a href="config_set_floatx.html">config_set_floatx</a>()</code> <br><code><a href="config_set_image_data_format.html">config_set_image_data_format</a>()</code> <br></p></div>
+    </div>
+
+  </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
+    </nav></aside></div>
+
+
+    <footer><div class="pkgdown-footer-left">
+  <p>Developed by Tomasz Kalinowski, JJ Allaire, François Chollet, Posit Software, PBC, Google.</p>
+</div>
+
+<div class="pkgdown-footer-right">
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
+</div>
+
+    </footer></div>
+
+
+
+
+
+  </body></html>
+
diff --git a/docs/dev/reference/config_max_steps_per_epoch.html b/docs/dev/reference/config_max_steps_per_epoch.html
new file mode 100644
index 000000000..a1c838315
--- /dev/null
+++ b/docs/dev/reference/config_max_steps_per_epoch.html
@@ -0,0 +1,8 @@
+<html>
+  <head>
+    <meta http-equiv="refresh" content="0;URL=https://keras3.posit.co/dev/reference/config_max_epochs.html" />
+    <meta name="robots" content="noindex">
+    <link rel="canonical" href="https://keras3.posit.co/dev/reference/config_max_epochs.html">
+  </head>
+</html>
+
diff --git a/docs/dev/reference/config_set_backend.html b/docs/dev/reference/config_set_backend.html
index e83993394..1f8336727 100644
--- a/docs/dev/reference/config_set_backend.html
+++ b/docs/dev/reference/config_set_backend.html
@@ -8,7 +8,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -93,7 +93,7 @@ <h2 id="warning">WARNING<a class="anchor" aria-label="anchor" href="#warning"></
     </div>
     <div class="section level2">
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
-    <div class="dont-index"><p>Other config: <br><code><a href="config_backend.html">config_backend</a>()</code> <br><code><a href="config_disable_flash_attention.html">config_disable_flash_attention</a>()</code> <br><code><a href="config_disable_interactive_logging.html">config_disable_interactive_logging</a>()</code> <br><code><a href="config_disable_traceback_filtering.html">config_disable_traceback_filtering</a>()</code> <br><code><a href="config_dtype_policy.html">config_dtype_policy</a>()</code> <br><code><a href="config_enable_flash_attention.html">config_enable_flash_attention</a>()</code> <br><code><a href="config_enable_interactive_logging.html">config_enable_interactive_logging</a>()</code> <br><code><a href="config_enable_traceback_filtering.html">config_enable_traceback_filtering</a>()</code> <br><code><a href="config_enable_unsafe_deserialization.html">config_enable_unsafe_deserialization</a>()</code> <br><code><a href="config_epsilon.html">config_epsilon</a>()</code> <br><code><a href="config_floatx.html">config_floatx</a>()</code> <br><code><a href="config_image_data_format.html">config_image_data_format</a>()</code> <br><code><a href="config_is_interactive_logging_enabled.html">config_is_interactive_logging_enabled</a>()</code> <br><code><a href="config_is_traceback_filtering_enabled.html">config_is_traceback_filtering_enabled</a>()</code> <br><code><a href="config_set_dtype_policy.html">config_set_dtype_policy</a>()</code> <br><code><a href="config_set_epsilon.html">config_set_epsilon</a>()</code> <br><code><a href="config_set_floatx.html">config_set_floatx</a>()</code> <br><code><a href="config_set_image_data_format.html">config_set_image_data_format</a>()</code> <br></p></div>
+    <div class="dont-index"><p>Other config: <br><code><a href="config_backend.html">config_backend</a>()</code> <br><code><a href="config_disable_flash_attention.html">config_disable_flash_attention</a>()</code> <br><code><a href="config_disable_interactive_logging.html">config_disable_interactive_logging</a>()</code> <br><code><a href="config_disable_traceback_filtering.html">config_disable_traceback_filtering</a>()</code> <br><code><a href="config_dtype_policy.html">config_dtype_policy</a>()</code> <br><code><a href="config_enable_flash_attention.html">config_enable_flash_attention</a>()</code> <br><code><a href="config_enable_interactive_logging.html">config_enable_interactive_logging</a>()</code> <br><code><a href="config_enable_traceback_filtering.html">config_enable_traceback_filtering</a>()</code> <br><code><a href="config_enable_unsafe_deserialization.html">config_enable_unsafe_deserialization</a>()</code> <br><code><a href="config_epsilon.html">config_epsilon</a>()</code> <br><code><a href="config_floatx.html">config_floatx</a>()</code> <br><code><a href="config_image_data_format.html">config_image_data_format</a>()</code> <br><code><a href="config_is_interactive_logging_enabled.html">config_is_interactive_logging_enabled</a>()</code> <br><code><a href="config_is_nnx_enabled.html">config_is_nnx_enabled</a>()</code> <br><code><a href="config_is_traceback_filtering_enabled.html">config_is_traceback_filtering_enabled</a>()</code> <br><code><a href="config_max_epochs.html">config_max_epochs</a>()</code> <br><code><a href="config_set_dtype_policy.html">config_set_dtype_policy</a>()</code> <br><code><a href="config_set_epsilon.html">config_set_epsilon</a>()</code> <br><code><a href="config_set_floatx.html">config_set_floatx</a>()</code> <br><code><a href="config_set_image_data_format.html">config_set_image_data_format</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -105,7 +105,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/config_set_dtype_policy.html b/docs/dev/reference/config_set_dtype_policy.html
index 85137351a..ba8f95c15 100644
--- a/docs/dev/reference/config_set_dtype_policy.html
+++ b/docs/dev/reference/config_set_dtype_policy.html
@@ -8,7 +8,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -80,7 +80,7 @@ <h2 id="examples">Examples<a class="anchor" aria-label="anchor" href="#examples"
     </div>
     <div class="section level2">
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
-    <div class="dont-index"><p>Other config: <br><code><a href="config_backend.html">config_backend</a>()</code> <br><code><a href="config_disable_flash_attention.html">config_disable_flash_attention</a>()</code> <br><code><a href="config_disable_interactive_logging.html">config_disable_interactive_logging</a>()</code> <br><code><a href="config_disable_traceback_filtering.html">config_disable_traceback_filtering</a>()</code> <br><code><a href="config_dtype_policy.html">config_dtype_policy</a>()</code> <br><code><a href="config_enable_flash_attention.html">config_enable_flash_attention</a>()</code> <br><code><a href="config_enable_interactive_logging.html">config_enable_interactive_logging</a>()</code> <br><code><a href="config_enable_traceback_filtering.html">config_enable_traceback_filtering</a>()</code> <br><code><a href="config_enable_unsafe_deserialization.html">config_enable_unsafe_deserialization</a>()</code> <br><code><a href="config_epsilon.html">config_epsilon</a>()</code> <br><code><a href="config_floatx.html">config_floatx</a>()</code> <br><code><a href="config_image_data_format.html">config_image_data_format</a>()</code> <br><code><a href="config_is_interactive_logging_enabled.html">config_is_interactive_logging_enabled</a>()</code> <br><code><a href="config_is_traceback_filtering_enabled.html">config_is_traceback_filtering_enabled</a>()</code> <br><code><a href="config_set_backend.html">config_set_backend</a>()</code> <br><code><a href="config_set_epsilon.html">config_set_epsilon</a>()</code> <br><code><a href="config_set_floatx.html">config_set_floatx</a>()</code> <br><code><a href="config_set_image_data_format.html">config_set_image_data_format</a>()</code> <br></p></div>
+    <div class="dont-index"><p>Other config: <br><code><a href="config_backend.html">config_backend</a>()</code> <br><code><a href="config_disable_flash_attention.html">config_disable_flash_attention</a>()</code> <br><code><a href="config_disable_interactive_logging.html">config_disable_interactive_logging</a>()</code> <br><code><a href="config_disable_traceback_filtering.html">config_disable_traceback_filtering</a>()</code> <br><code><a href="config_dtype_policy.html">config_dtype_policy</a>()</code> <br><code><a href="config_enable_flash_attention.html">config_enable_flash_attention</a>()</code> <br><code><a href="config_enable_interactive_logging.html">config_enable_interactive_logging</a>()</code> <br><code><a href="config_enable_traceback_filtering.html">config_enable_traceback_filtering</a>()</code> <br><code><a href="config_enable_unsafe_deserialization.html">config_enable_unsafe_deserialization</a>()</code> <br><code><a href="config_epsilon.html">config_epsilon</a>()</code> <br><code><a href="config_floatx.html">config_floatx</a>()</code> <br><code><a href="config_image_data_format.html">config_image_data_format</a>()</code> <br><code><a href="config_is_interactive_logging_enabled.html">config_is_interactive_logging_enabled</a>()</code> <br><code><a href="config_is_nnx_enabled.html">config_is_nnx_enabled</a>()</code> <br><code><a href="config_is_traceback_filtering_enabled.html">config_is_traceback_filtering_enabled</a>()</code> <br><code><a href="config_max_epochs.html">config_max_epochs</a>()</code> <br><code><a href="config_set_backend.html">config_set_backend</a>()</code> <br><code><a href="config_set_epsilon.html">config_set_epsilon</a>()</code> <br><code><a href="config_set_floatx.html">config_set_floatx</a>()</code> <br><code><a href="config_set_image_data_format.html">config_set_image_data_format</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -92,7 +92,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/config_set_epsilon.html b/docs/dev/reference/config_set_epsilon.html
index 61ab73068..2a81cbd69 100644
--- a/docs/dev/reference/config_set_epsilon.html
+++ b/docs/dev/reference/config_set_epsilon.html
@@ -8,7 +8,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -92,7 +92,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 <ul><li><p><a href="https://keras.io/api/utils/config_utils#setepsilon-function" class="external-link">https://keras.io/api/utils/config_utils#setepsilon-function</a></p></li>
 </ul><p>Other config backend: <br><code><a href="config_backend.html">config_backend</a>()</code> <br><code><a href="config_epsilon.html">config_epsilon</a>()</code> <br><code><a href="config_floatx.html">config_floatx</a>()</code> <br><code><a href="config_image_data_format.html">config_image_data_format</a>()</code> <br><code><a href="config_set_floatx.html">config_set_floatx</a>()</code> <br><code><a href="config_set_image_data_format.html">config_set_image_data_format</a>()</code> <br></p>
 <p>Other backend: <br><code><a href="clear_session.html">clear_session</a>()</code> <br><code><a href="config_backend.html">config_backend</a>()</code> <br><code><a href="config_epsilon.html">config_epsilon</a>()</code> <br><code><a href="config_floatx.html">config_floatx</a>()</code> <br><code><a href="config_image_data_format.html">config_image_data_format</a>()</code> <br><code><a href="config_set_floatx.html">config_set_floatx</a>()</code> <br><code><a href="config_set_image_data_format.html">config_set_image_data_format</a>()</code> <br></p>
-<p>Other config: <br><code><a href="config_backend.html">config_backend</a>()</code> <br><code><a href="config_disable_flash_attention.html">config_disable_flash_attention</a>()</code> <br><code><a href="config_disable_interactive_logging.html">config_disable_interactive_logging</a>()</code> <br><code><a href="config_disable_traceback_filtering.html">config_disable_traceback_filtering</a>()</code> <br><code><a href="config_dtype_policy.html">config_dtype_policy</a>()</code> <br><code><a href="config_enable_flash_attention.html">config_enable_flash_attention</a>()</code> <br><code><a href="config_enable_interactive_logging.html">config_enable_interactive_logging</a>()</code> <br><code><a href="config_enable_traceback_filtering.html">config_enable_traceback_filtering</a>()</code> <br><code><a href="config_enable_unsafe_deserialization.html">config_enable_unsafe_deserialization</a>()</code> <br><code><a href="config_epsilon.html">config_epsilon</a>()</code> <br><code><a href="config_floatx.html">config_floatx</a>()</code> <br><code><a href="config_image_data_format.html">config_image_data_format</a>()</code> <br><code><a href="config_is_interactive_logging_enabled.html">config_is_interactive_logging_enabled</a>()</code> <br><code><a href="config_is_traceback_filtering_enabled.html">config_is_traceback_filtering_enabled</a>()</code> <br><code><a href="config_set_backend.html">config_set_backend</a>()</code> <br><code><a href="config_set_dtype_policy.html">config_set_dtype_policy</a>()</code> <br><code><a href="config_set_floatx.html">config_set_floatx</a>()</code> <br><code><a href="config_set_image_data_format.html">config_set_image_data_format</a>()</code> <br></p></div>
+<p>Other config: <br><code><a href="config_backend.html">config_backend</a>()</code> <br><code><a href="config_disable_flash_attention.html">config_disable_flash_attention</a>()</code> <br><code><a href="config_disable_interactive_logging.html">config_disable_interactive_logging</a>()</code> <br><code><a href="config_disable_traceback_filtering.html">config_disable_traceback_filtering</a>()</code> <br><code><a href="config_dtype_policy.html">config_dtype_policy</a>()</code> <br><code><a href="config_enable_flash_attention.html">config_enable_flash_attention</a>()</code> <br><code><a href="config_enable_interactive_logging.html">config_enable_interactive_logging</a>()</code> <br><code><a href="config_enable_traceback_filtering.html">config_enable_traceback_filtering</a>()</code> <br><code><a href="config_enable_unsafe_deserialization.html">config_enable_unsafe_deserialization</a>()</code> <br><code><a href="config_epsilon.html">config_epsilon</a>()</code> <br><code><a href="config_floatx.html">config_floatx</a>()</code> <br><code><a href="config_image_data_format.html">config_image_data_format</a>()</code> <br><code><a href="config_is_interactive_logging_enabled.html">config_is_interactive_logging_enabled</a>()</code> <br><code><a href="config_is_nnx_enabled.html">config_is_nnx_enabled</a>()</code> <br><code><a href="config_is_traceback_filtering_enabled.html">config_is_traceback_filtering_enabled</a>()</code> <br><code><a href="config_max_epochs.html">config_max_epochs</a>()</code> <br><code><a href="config_set_backend.html">config_set_backend</a>()</code> <br><code><a href="config_set_dtype_policy.html">config_set_dtype_policy</a>()</code> <br><code><a href="config_set_floatx.html">config_set_floatx</a>()</code> <br><code><a href="config_set_image_data_format.html">config_set_image_data_format</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -104,7 +104,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/config_set_floatx.html b/docs/dev/reference/config_set_floatx.html
index 2bde60b85..81edfc483 100644
--- a/docs/dev/reference/config_set_floatx.html
+++ b/docs/dev/reference/config_set_floatx.html
@@ -8,7 +8,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -104,7 +104,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 <ul><li><p><a href="https://keras.io/api/utils/config_utils#setfloatx-function" class="external-link">https://keras.io/api/utils/config_utils#setfloatx-function</a></p></li>
 </ul><p>Other config backend: <br><code><a href="config_backend.html">config_backend</a>()</code> <br><code><a href="config_epsilon.html">config_epsilon</a>()</code> <br><code><a href="config_floatx.html">config_floatx</a>()</code> <br><code><a href="config_image_data_format.html">config_image_data_format</a>()</code> <br><code><a href="config_set_epsilon.html">config_set_epsilon</a>()</code> <br><code><a href="config_set_image_data_format.html">config_set_image_data_format</a>()</code> <br></p>
 <p>Other backend: <br><code><a href="clear_session.html">clear_session</a>()</code> <br><code><a href="config_backend.html">config_backend</a>()</code> <br><code><a href="config_epsilon.html">config_epsilon</a>()</code> <br><code><a href="config_floatx.html">config_floatx</a>()</code> <br><code><a href="config_image_data_format.html">config_image_data_format</a>()</code> <br><code><a href="config_set_epsilon.html">config_set_epsilon</a>()</code> <br><code><a href="config_set_image_data_format.html">config_set_image_data_format</a>()</code> <br></p>
-<p>Other config: <br><code><a href="config_backend.html">config_backend</a>()</code> <br><code><a href="config_disable_flash_attention.html">config_disable_flash_attention</a>()</code> <br><code><a href="config_disable_interactive_logging.html">config_disable_interactive_logging</a>()</code> <br><code><a href="config_disable_traceback_filtering.html">config_disable_traceback_filtering</a>()</code> <br><code><a href="config_dtype_policy.html">config_dtype_policy</a>()</code> <br><code><a href="config_enable_flash_attention.html">config_enable_flash_attention</a>()</code> <br><code><a href="config_enable_interactive_logging.html">config_enable_interactive_logging</a>()</code> <br><code><a href="config_enable_traceback_filtering.html">config_enable_traceback_filtering</a>()</code> <br><code><a href="config_enable_unsafe_deserialization.html">config_enable_unsafe_deserialization</a>()</code> <br><code><a href="config_epsilon.html">config_epsilon</a>()</code> <br><code><a href="config_floatx.html">config_floatx</a>()</code> <br><code><a href="config_image_data_format.html">config_image_data_format</a>()</code> <br><code><a href="config_is_interactive_logging_enabled.html">config_is_interactive_logging_enabled</a>()</code> <br><code><a href="config_is_traceback_filtering_enabled.html">config_is_traceback_filtering_enabled</a>()</code> <br><code><a href="config_set_backend.html">config_set_backend</a>()</code> <br><code><a href="config_set_dtype_policy.html">config_set_dtype_policy</a>()</code> <br><code><a href="config_set_epsilon.html">config_set_epsilon</a>()</code> <br><code><a href="config_set_image_data_format.html">config_set_image_data_format</a>()</code> <br></p></div>
+<p>Other config: <br><code><a href="config_backend.html">config_backend</a>()</code> <br><code><a href="config_disable_flash_attention.html">config_disable_flash_attention</a>()</code> <br><code><a href="config_disable_interactive_logging.html">config_disable_interactive_logging</a>()</code> <br><code><a href="config_disable_traceback_filtering.html">config_disable_traceback_filtering</a>()</code> <br><code><a href="config_dtype_policy.html">config_dtype_policy</a>()</code> <br><code><a href="config_enable_flash_attention.html">config_enable_flash_attention</a>()</code> <br><code><a href="config_enable_interactive_logging.html">config_enable_interactive_logging</a>()</code> <br><code><a href="config_enable_traceback_filtering.html">config_enable_traceback_filtering</a>()</code> <br><code><a href="config_enable_unsafe_deserialization.html">config_enable_unsafe_deserialization</a>()</code> <br><code><a href="config_epsilon.html">config_epsilon</a>()</code> <br><code><a href="config_floatx.html">config_floatx</a>()</code> <br><code><a href="config_image_data_format.html">config_image_data_format</a>()</code> <br><code><a href="config_is_interactive_logging_enabled.html">config_is_interactive_logging_enabled</a>()</code> <br><code><a href="config_is_nnx_enabled.html">config_is_nnx_enabled</a>()</code> <br><code><a href="config_is_traceback_filtering_enabled.html">config_is_traceback_filtering_enabled</a>()</code> <br><code><a href="config_max_epochs.html">config_max_epochs</a>()</code> <br><code><a href="config_set_backend.html">config_set_backend</a>()</code> <br><code><a href="config_set_dtype_policy.html">config_set_dtype_policy</a>()</code> <br><code><a href="config_set_epsilon.html">config_set_epsilon</a>()</code> <br><code><a href="config_set_image_data_format.html">config_set_image_data_format</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -116,7 +116,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/config_set_image_data_format.html b/docs/dev/reference/config_set_image_data_format.html
index 1624f2c67..c99954ace 100644
--- a/docs/dev/reference/config_set_image_data_format.html
+++ b/docs/dev/reference/config_set_image_data_format.html
@@ -8,7 +8,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -93,7 +93,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 <ul><li><p><a href="https://keras.io/api/utils/config_utils#setimagedataformat-function" class="external-link">https://keras.io/api/utils/config_utils#setimagedataformat-function</a></p></li>
 </ul><p>Other config backend: <br><code><a href="config_backend.html">config_backend</a>()</code> <br><code><a href="config_epsilon.html">config_epsilon</a>()</code> <br><code><a href="config_floatx.html">config_floatx</a>()</code> <br><code><a href="config_image_data_format.html">config_image_data_format</a>()</code> <br><code><a href="config_set_epsilon.html">config_set_epsilon</a>()</code> <br><code><a href="config_set_floatx.html">config_set_floatx</a>()</code> <br></p>
 <p>Other backend: <br><code><a href="clear_session.html">clear_session</a>()</code> <br><code><a href="config_backend.html">config_backend</a>()</code> <br><code><a href="config_epsilon.html">config_epsilon</a>()</code> <br><code><a href="config_floatx.html">config_floatx</a>()</code> <br><code><a href="config_image_data_format.html">config_image_data_format</a>()</code> <br><code><a href="config_set_epsilon.html">config_set_epsilon</a>()</code> <br><code><a href="config_set_floatx.html">config_set_floatx</a>()</code> <br></p>
-<p>Other config: <br><code><a href="config_backend.html">config_backend</a>()</code> <br><code><a href="config_disable_flash_attention.html">config_disable_flash_attention</a>()</code> <br><code><a href="config_disable_interactive_logging.html">config_disable_interactive_logging</a>()</code> <br><code><a href="config_disable_traceback_filtering.html">config_disable_traceback_filtering</a>()</code> <br><code><a href="config_dtype_policy.html">config_dtype_policy</a>()</code> <br><code><a href="config_enable_flash_attention.html">config_enable_flash_attention</a>()</code> <br><code><a href="config_enable_interactive_logging.html">config_enable_interactive_logging</a>()</code> <br><code><a href="config_enable_traceback_filtering.html">config_enable_traceback_filtering</a>()</code> <br><code><a href="config_enable_unsafe_deserialization.html">config_enable_unsafe_deserialization</a>()</code> <br><code><a href="config_epsilon.html">config_epsilon</a>()</code> <br><code><a href="config_floatx.html">config_floatx</a>()</code> <br><code><a href="config_image_data_format.html">config_image_data_format</a>()</code> <br><code><a href="config_is_interactive_logging_enabled.html">config_is_interactive_logging_enabled</a>()</code> <br><code><a href="config_is_traceback_filtering_enabled.html">config_is_traceback_filtering_enabled</a>()</code> <br><code><a href="config_set_backend.html">config_set_backend</a>()</code> <br><code><a href="config_set_dtype_policy.html">config_set_dtype_policy</a>()</code> <br><code><a href="config_set_epsilon.html">config_set_epsilon</a>()</code> <br><code><a href="config_set_floatx.html">config_set_floatx</a>()</code> <br></p></div>
+<p>Other config: <br><code><a href="config_backend.html">config_backend</a>()</code> <br><code><a href="config_disable_flash_attention.html">config_disable_flash_attention</a>()</code> <br><code><a href="config_disable_interactive_logging.html">config_disable_interactive_logging</a>()</code> <br><code><a href="config_disable_traceback_filtering.html">config_disable_traceback_filtering</a>()</code> <br><code><a href="config_dtype_policy.html">config_dtype_policy</a>()</code> <br><code><a href="config_enable_flash_attention.html">config_enable_flash_attention</a>()</code> <br><code><a href="config_enable_interactive_logging.html">config_enable_interactive_logging</a>()</code> <br><code><a href="config_enable_traceback_filtering.html">config_enable_traceback_filtering</a>()</code> <br><code><a href="config_enable_unsafe_deserialization.html">config_enable_unsafe_deserialization</a>()</code> <br><code><a href="config_epsilon.html">config_epsilon</a>()</code> <br><code><a href="config_floatx.html">config_floatx</a>()</code> <br><code><a href="config_image_data_format.html">config_image_data_format</a>()</code> <br><code><a href="config_is_interactive_logging_enabled.html">config_is_interactive_logging_enabled</a>()</code> <br><code><a href="config_is_nnx_enabled.html">config_is_nnx_enabled</a>()</code> <br><code><a href="config_is_traceback_filtering_enabled.html">config_is_traceback_filtering_enabled</a>()</code> <br><code><a href="config_max_epochs.html">config_max_epochs</a>()</code> <br><code><a href="config_set_backend.html">config_set_backend</a>()</code> <br><code><a href="config_set_dtype_policy.html">config_set_dtype_policy</a>()</code> <br><code><a href="config_set_epsilon.html">config_set_epsilon</a>()</code> <br><code><a href="config_set_floatx.html">config_set_floatx</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -105,7 +105,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/config_set_max_epochs.html b/docs/dev/reference/config_set_max_epochs.html
new file mode 100644
index 000000000..a1c838315
--- /dev/null
+++ b/docs/dev/reference/config_set_max_epochs.html
@@ -0,0 +1,8 @@
+<html>
+  <head>
+    <meta http-equiv="refresh" content="0;URL=https://keras3.posit.co/dev/reference/config_max_epochs.html" />
+    <meta name="robots" content="noindex">
+    <link rel="canonical" href="https://keras3.posit.co/dev/reference/config_max_epochs.html">
+  </head>
+</html>
+
diff --git a/docs/dev/reference/config_set_max_steps_per_epoch.html b/docs/dev/reference/config_set_max_steps_per_epoch.html
new file mode 100644
index 000000000..a1c838315
--- /dev/null
+++ b/docs/dev/reference/config_set_max_steps_per_epoch.html
@@ -0,0 +1,8 @@
+<html>
+  <head>
+    <meta http-equiv="refresh" content="0;URL=https://keras3.posit.co/dev/reference/config_max_epochs.html" />
+    <meta name="robots" content="noindex">
+    <link rel="canonical" href="https://keras3.posit.co/dev/reference/config_max_epochs.html">
+  </head>
+</html>
+
diff --git a/docs/dev/reference/constraint_maxnorm.html b/docs/dev/reference/constraint_maxnorm.html
index ca6b7226e..32f16c86a 100644
--- a/docs/dev/reference/constraint_maxnorm.html
+++ b/docs/dev/reference/constraint_maxnorm.html
@@ -10,7 +10,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -108,7 +108,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/constraint_minmaxnorm.html b/docs/dev/reference/constraint_minmaxnorm.html
index dd5fd25a6..2b35bfdd5 100644
--- a/docs/dev/reference/constraint_minmaxnorm.html
+++ b/docs/dev/reference/constraint_minmaxnorm.html
@@ -10,7 +10,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -123,7 +123,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/constraint_nonneg.html b/docs/dev/reference/constraint_nonneg.html
index 94c0cb235..d3ece9cfe 100644
--- a/docs/dev/reference/constraint_nonneg.html
+++ b/docs/dev/reference/constraint_nonneg.html
@@ -8,7 +8,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -83,7 +83,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/constraint_unitnorm.html b/docs/dev/reference/constraint_unitnorm.html
index 499f8e03a..445649efa 100644
--- a/docs/dev/reference/constraint_unitnorm.html
+++ b/docs/dev/reference/constraint_unitnorm.html
@@ -8,7 +8,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -101,7 +101,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/count_params.html b/docs/dev/reference/count_params.html
index eb4edfa53..66d5cb257 100644
--- a/docs/dev/reference/count_params.html
+++ b/docs/dev/reference/count_params.html
@@ -8,7 +8,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -88,7 +88,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/custom_metric.html b/docs/dev/reference/custom_metric.html
index 2c2579d2a..8048dace8 100644
--- a/docs/dev/reference/custom_metric.html
+++ b/docs/dev/reference/custom_metric.html
@@ -8,7 +8,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -112,7 +112,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/dataset_boston_housing.html b/docs/dev/reference/dataset_boston_housing.html
index a7cf93fdf..746af3148 100644
--- a/docs/dev/reference/dataset_boston_housing.html
+++ b/docs/dev/reference/dataset_boston_housing.html
@@ -10,7 +10,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -108,7 +108,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/dataset_california_housing.html b/docs/dev/reference/dataset_california_housing.html
index eeafcb400..0fe1fa710 100644
--- a/docs/dev/reference/dataset_california_housing.html
+++ b/docs/dev/reference/dataset_california_housing.html
@@ -56,7 +56,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -187,7 +187,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/dataset_cifar10.html b/docs/dev/reference/dataset_cifar10.html
index 49f84d0f7..b8c0e63b3 100644
--- a/docs/dev/reference/dataset_cifar10.html
+++ b/docs/dev/reference/dataset_cifar10.html
@@ -10,7 +10,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -87,7 +87,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/dataset_cifar100.html b/docs/dev/reference/dataset_cifar100.html
index a562b4e73..56c37b596 100644
--- a/docs/dev/reference/dataset_cifar100.html
+++ b/docs/dev/reference/dataset_cifar100.html
@@ -10,7 +10,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -93,7 +93,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/dataset_fashion_mnist.html b/docs/dev/reference/dataset_fashion_mnist.html
index 590b7b40b..e4289b6c8 100644
--- a/docs/dev/reference/dataset_fashion_mnist.html
+++ b/docs/dev/reference/dataset_fashion_mnist.html
@@ -14,7 +14,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -106,7 +106,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/dataset_imdb.html b/docs/dev/reference/dataset_imdb.html
index 16619b6c0..e296e28d6 100644
--- a/docs/dev/reference/dataset_imdb.html
+++ b/docs/dev/reference/dataset_imdb.html
@@ -20,7 +20,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -162,7 +162,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/dataset_mnist.html b/docs/dev/reference/dataset_mnist.html
index 270401ced..45d5d5895 100644
--- a/docs/dev/reference/dataset_mnist.html
+++ b/docs/dev/reference/dataset_mnist.html
@@ -8,7 +8,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -90,7 +90,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/dataset_reuters.html b/docs/dev/reference/dataset_reuters.html
index 3297c72e1..40b6c65c2 100644
--- a/docs/dev/reference/dataset_reuters.html
+++ b/docs/dev/reference/dataset_reuters.html
@@ -12,7 +12,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -146,7 +146,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/deserialize_keras_object.html b/docs/dev/reference/deserialize_keras_object.html
index 8f5f3df0d..cb035b8a9 100644
--- a/docs/dev/reference/deserialize_keras_object.html
+++ b/docs/dev/reference/deserialize_keras_object.html
@@ -220,7 +220,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -409,7 +409,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/evaluate.keras.src.models.model.Model.html b/docs/dev/reference/evaluate.keras.src.models.model.Model.html
index 4a22e1926..368bb15ba 100644
--- a/docs/dev/reference/evaluate.keras.src.models.model.Model.html
+++ b/docs/dev/reference/evaluate.keras.src.models.model.Model.html
@@ -12,7 +12,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -189,7 +189,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/export_savedmodel.keras.src.models.model.Model.html b/docs/dev/reference/export_savedmodel.keras.src.models.model.Model.html
index 706f9b760..ec055e5e0 100644
--- a/docs/dev/reference/export_savedmodel.keras.src.models.model.Model.html
+++ b/docs/dev/reference/export_savedmodel.keras.src.models.model.Model.html
@@ -36,7 +36,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -200,7 +200,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/figures/op-gelu-plot-1.png b/docs/dev/reference/figures/op-gelu-plot-1.png
index d34697042776579594b2859afa21cfe37cfd6331..45e2d00b7b68d6939555ecc9f0231d5f5f14a0e5 100644
GIT binary patch
literal 16638
zcmeIaXH-+$7d?u4)vKbYfC?z6fb^m$0ck2tI#Q*J^cGs^O+^rtB1jKKq=^V92uKhR
zkxs<WJ18|&=^fq*LGk|nZ;bcpy-)9q;kcTU?6c3_tIRdm+}B@GUh?Q+>cb=?BuAxg
ziz$(i>{%xK`FAgT!Xj<T4F5S~b6dlngyhIc!k;~jJ)Y|%B<D$_#BQoM$4>T`Xsfif
z9{42`a9-}_iQ_NszALzkYkyT4fuIhk6bn>0D7&i^_G}N!Me#lU?H2Wc9F@IUmF?H4
z1369@C~!YIN|}6YA8CF4+C%LmrQ`|wZ}qwjTcV!(zVKV5YfR0(YoH3pnHs|LIFF<p
zfw`FJf8R&)?=f#$k{dQ>=1E8%mA^bla{TN8A(99E+@pI){&{kfjO08YsVfOd+5i7l
zRVB=W94UEuVaJK6MpY6L8)JR_A9%c(nHft(8P@VUoscVPX{x8E$VnY@oP;F4HB+k%
z<1oIuFfL<j!$UJ4@Sal#<33x%B&~jjJy62Hz#vI2Bb-xA6OYFqJ9bQ3{R01CtFeXE
zi3v+LH@B1{A!_pS#|-K_J2mmQo;pl*^W%>N1;q1Ys%5l|hiBcgi;j+#jy!*m@Ece7
z5Jef<ghySElgG1C7<jU|l8^lS`3iN&&Uv9lJ)HCAF^bL)IXSO>Mdq8gypfK4$?$fR
zzsd9}h1xtk{?9mp5=k&lT}=(kZEMpnFGommfw!4I_>cPT6+^_{`FPm)>XoAsNpfNc
zihJS>RIg}46ivx-NB5tZkG+67ama`FM33TwQD2_J>SX~Esb!|yzk~;gqUcW^wz`N(
zyfJ&z`lH(*I&efeFyL~8{o0dXUd1h08ur1x^rB81EQ3QA{|p*R=IYV?&99>B#nAwx
z1k&YF+qE}Gk7^Z~PYt%V?OXN85vpNcz#GA9-ZwHzfoUf_H{+(ZYV=1kry*v=5i6zB
z{EAj%WBbWhoyNO8-Zd2HE$|i3*htLic`g6?ama`L?z{fMnisCc!U)7tnfJj<%NMU*
zTiD!Kdm4FR;aPLDiZZk4*y)*>nN&sJWG>Wo9dq+HKaFm0{QMRtpqZ)m?S=W(m=u}F
zh|Qv8xrl(1P3JZ~=%$T|%m%I?USSqB&tz|{&2>Dv#Czt0aNe@DtdvxtV)d^d<p+;a
z8(yV8dO+0CG5=7B^YSl!2?^irai@BU;8D$7KL-Zn5=o`>28V}*-Io<-t|WVg#Ky+P
z$8(NI8Qi}=j76)jPE&)qX=}&cp=M3G)49wr{WZ+j+FB8T?0km|mrTpLs^hv?M5g^h
zo@)+G5x!m#y%pq^C)-BR+uQ5!?~mrS3VowZ^5C$HloS=8O+PM4QfABDg^P>J%gf8&
zL-zd;`iM2-!Jdur?9$EEakZtc4%LN#Aw2=-X4mfF7K2*eA?#jSWp>!rv6-m*hyH<U
z|9+`?&Wl6mtAt(iGVOa#eRI%4xsQ7#;Hh$w(ny)q9;s#I<>Z|7{BB}wd{|@;RUq$z
zYq44erRVq4mW3fE-}Q=JukrI2>Xmw4iOmSJDW@=@;N|9~?z(d(kQYCJ37ORP^z@t?
z#=e+DrCiW;e8t_(%}98@Z8?Q{SH6`+?<bcei9ka>b-eyxwkgTa&wu&y<%7r0KA&eF
zdT_IW$|vKV1bx%JtDRz~@;kD!^~*B@a=*T@gMt6H;a{_`O`=bZ1=0#>Hs*$r9S|~c
zZY4NEdQ~oxZ)R@p$dMzt@3<=xpZ<@nKhyKtP|a>DT^UA)#;0`4&d|~2`i@F4P{6a4
zZ6%UySLgA$CJhmz%al)GCOf7sS7FuuE>R|u(_ySRgmCM>qozK}$;pY})NviGKH(qK
zuwQN0*;Et_4F|(mlr62SQuz3|{&GW=uW<rS-(vZBP@{@dlEX)EM@=ZQ9#I9#dgH)j
z<;0hlm&GI`-l(v$9Q$_SuMxw#8br1x`uqET{3u&`r}TL@1f;spyB3$0dO17uRT)Sh
z{GS0=+~=@}>5{c9G@7QbT>2k2*Kfyr{2t#z`yp#DaBx^HjI}7I%KmV;k*@#U{?C#e
zOH54Mvu95@oBA;E&hY=~%+4tY1hV_1?SOAu8dGFa%|7pV9x=ql#P640p>t*z76OBU
zhV7U;?)^3FrM1P$46OoWT?7Z4Co{i(+g~sLLLQ|&<2^VugpIjo(GV|0DH2xy?$w__
zmpTtCy^52h1GBMZk&^#wNyC(slo}&Av@Xd;ea>LuzkvA>@I{q7DTEMoE@D_xsCbO)
zWBHpJ8$)G;u9QFBon_-k6kf^B&aTvRgD--CiF9{qq|4jd5Ld2*8VFqxMV0?;nR7}o
zf2*ccxpLp5{H^pn@T05Lfp9_K!5lYBNlO#p<73wG83dzj8_&(not&H`<dXe`fq8ij
z{3W*F4FsEi$-HtN^W%3IPc;Nr7^n)4oZ>JUdPphsADcKd)mQ4(|M?-@pW~E7-o@Gb
z>TIvGV^7%UzCN$TiH=jJj76BRC3}>;e;aiykIA<)q`t@=jFd`f?C~BOi65*{xazCm
zvnP5eYD}-LuB~Nj<VwWY4f?|K{E%X|5+QwvYnFO?1k(zQHK)_-NZI|BX;QfZC2}o0
za?F}ir8NJ~5l>N3;U_1pOiY+`{=3MhyGuOQ>Z5t+JeL!j8EX#xWhJMVS63y(#nV+-
z(`7=*U(tMH*51!U9ep4P2PR~4PQbCgtgNiOTy)IU@b5V);1^cG5q|x$agh!shZyjC
z@(P)nd9_VVO<1g4eCXig|MAuvnwp)E_Jhs|TvGV&6!(I)H>D|LYZvBN{%3|XmF<Bv
z1S1DH!D1q1ck92lP=01^+sq>}&#e3v+OT9I>x$XB#bA=P+M%)Bztil=VC3xVtZ_s9
zxwB{4z5Zu32HGBWjU2<TekbJR<*&5S^Zd!qN<qlb(9nGAK4JYY|A_zn#<e&(MMXs$
z8=DmAp9g7vIx+9&cPU<se_)`Ty!=#GLH5OkYSKS`Z<vIBl%~R}8eIQpVvPaFd}L2&
zmQIoHQ5x2}|FgM=BoI5Hnv7QE9c}cVH+^vr#p5*9eTe*07#_PHd+yIpnB%Y;YfEIb
zg3B{ibZj~yZ~tR{>EZ}j>;xgVe4qW9*ElngMh~`)mVvS-mfyb8^s3`{s~p~LF!l3G
z1jE_8`3rVg>r4^*w_`E{i5F<}`1G`h+oENP#hnN9xWLQfQ$rbKIe`iyAzR~!6KFId
zD>HLUG24Osv`h`#bILzyC7HbEtxi$CNkcru^E;SR`>VM}={ER^w+8p|b1vvPDDqUE
zX%d!}77}uqHy%ekBJ&XF7H4!hWY=hVm6w;-JP+_*a(I0FRDb_V0;>6*s+-;L?OT39
z0pLnAMkXdMjhsP9TGFRqk>ZnRjw~rS>a+_LztHNriCep$tyj9a789)N#x^#wFdnHE
zd3C73y016wIoNr3k&{Ap+sNSH*QzRX|FeLAfX9#jsi=FtDh>#QB@95%*RNs#dKyn(
zkK?4L7bO6X6KBr~nl&=yX<A)epzyq4_8|2Nbs*Ok>#_M@&lrKSwbE;8S+N^qPlKLa
z*KNPn_<lliW^qwoR<`?t8Tr=UJvS8L>5H?o*WH)jNgHAIJc^8rge0=Dx{$*6(NZKp
zwa9U)%f(PXL)4&mRygWXg9xYe=PUxQR93jx7MPuFefRENWY%k~8@lqX9`kR^Cz&!u
zdhPpC6Yd_=;#Xj&&s4K8GGcU>NQjPRad*a=62f7gX69$AjrRssXV0C}%C}5!!oSVT
z4Cj=jn(gB@Y8#BRl`|B{uw=HpckgO`5=8H(#p4p$Z3hn?wC>39Y2HV6xcsQ{4j8Bi
zLaJqI4gp3U9*%hN;)QZO+FO}`B9Kuh0q-vS?Bx!Rhyb(3uqda#``LFk_FutqU!I5A
zj}pW$5I_b6ALh`JBWPg{On5jY+a|Wm|MTzEkop>j1;5YIE1kmQHGW-qMbm$18*j6i
zqXx#uY0jP7m=_g8e17p-y^QPSHn=@Z)iVK}29M*=%6DI-`iVQb3r<c?X&_u&Tyl-E
zod7a&6WFV*s;819A((6%weqJ?<C&io^b!F!c6P3dFZ_@Xm}asbJrg6&lN}=^E!~-K
z)ukY^G5=-6UlQ>r1Ek(7LK@dcUA7;Vm8m=3H?p?~e!885Q*Yz2Eg9;tFl!6r>dTSG
zdq@;6>g<nvaEpNcS(H!%BO?gw-cJ|34&jdedc5s`6qhg^`Bqwh<Y=y$s|;pe#O#SD
zVukx+h{D3cxse94fB$8kIzkxii}wD=0ODX0+hM?VpfcR1Y2WTz-9XLtmw9_1BnzSy
z!aUBtiP|Sf#Buxmkn0;8<>loZ?9wsDgty(3^+?8hdoZ&p92SsmOhUreFxom|-H33T
z_(mF%2#F+IU!Taui$7EJ{6`Ns-`cKTVjNJp`sKcW{oB%&M~HW>jA!&W63dPlYgAH;
zc_f#f=db><M)%8qCTLcIlT%S?PD}7KH!)F<;ga|>LF`qG)a~1R5G4R^%B(C9=D<^Y
zn3L@=AvdK)OCSL7hJ=K$hQA5Z3*McG0m^=?nMF&h>)5&LVVhcnWg@1g+-FGA1EPwG
z-`{HF*3Rn?#<Tx)nDZcUybf_hpXZttpb)44Ra2*L?M{RZQ&V5BUF^EhoUUBA8o0X{
zF5-wEKYj=c3%9nm8f{1r&5e_jQ%y~6yI$R%D3v&(04&mdWfnG~Zl+i@LoFq?_YWYu
znlx(M`N8bYojXHox?s+bfx*3-v|pi<B!gRFs_^}xHAfiRvB$=kQ@xCmXpvW+je=J`
zdHmS2ucWwJ@?~rn_uJ7zVksFTj>?7S-Sj@dFCY-P_V4bD@sf!2^mH0t%jNkd1vb0m
z1pmS7?YF<VZRJ-tMlUfiFfcQ}euVDWT{!Y!BzWQGNP^embYJPCs$C<J6U5AHC*Q1P
zTf+A*c~aIN@gbn-rm!S*vD*@+sHngc_v+`9i?luP%BeV0OH0dU^Snl%+iXHYJ!NG#
z@vIDziE_od32|{&*4Ejm3tY8zb!Z{C#ol7K!n2v@3s|ah$-ffHW}*SsNFbh_=Az}d
zOMdZ!JTmD5xQFZ_YZMo87tJIPQc_X^juU*I>wMY!AlVbpYtc5YYU*|%KTEi@(@z$J
zGdun<4DtYEmTvJXBn<#bc7m6{IBWjNew1>9xF}_#l?@Hxr7rHlf7!bxD%XDG+a<Y}
ztLDvq)A}d=jP~Lw5{hA2Sy@HJ=M<7)&)rd-s_(>@4XFbp(3dECpsv2x{Q6E%P*8Vp
zz^>WKqvl3&Nw8!=Gzj^p-t>RmK9#)Zp92TJMO_wsRa8rOOsYL4I}7o3{lyUP-@k|O
z>#;fywbAZA!;O-Xk`pIRG@hgT6I!nNj2c5JXa#?uqC%zJ027b?qWX@MIJc*gi0g64
z&Gi+i4_?f(?+*48CkKUK`+I{F&$Czmcy78SY7RUTFh(#HkHi=`VJC=zs#XH=@Zm#!
zef?9XLbLgHS7v__8nzB-0jT8bY&icIyE1Ww0YT5k9>E2Yejwx2`4}<W{KkIHT#BhJ
zi~o4GYi85S6I*Rxbs6jML&QL-!iwp@I9L=pPEubrJ>P)$lSDL<xek=a4fLHYP}gbG
zcriQXCPX{El2r3K$Z;#K)qbd<OI`x?{>3K}mmk<Nae&(5y=S%4nmF#hyD`QQ-tYzO
zEhFPKUKG{jQLt${S<vT&*0NelY>PbG{?2Bz%`F(4UTiz`<7C#E1}GiVo*y~Q$-u-!
zLwX=i-Xw<W3J(v^66tz|hK5nMsGRT2oQ7;OyOk2??@z^VXQ-<iAIpTrVz2r1{ragp
znk*Z^pogQO6>?T8P_1KSO|_dHZAr55<PoSmqpWKE{j;M?PHVO0Gk<GckA)V;$<E?m
zHDGT;Lqq8*Z=}OnHMF$Q{YTW*)q&e`rcx?}qy@->K+O<7+(s!-H%6>*S-HBd28(Pu
z4_1eCcWZaZkByF+nwXf>dYLf>?op9Qx`RpLsSHryC9^m;Iooo0x+J?uJzM*il=pBj
zHo@haoeqsaT9T1Q(D3ABTvXIYv!+Wd(nEWRiP#G@Ff??{vi%+SYMrBm@iAQB-vDM1
z7g&q;_Ob9yW?!N%X66mPTp4xQVZ-fO*FljLXCJ*MSL9rBA8c1qsr5z3!|_Tib73a!
z4)hx=?B<KJec$W0@Kvl;WF=l#yL;v9-SXaz6Uc%vmo~Y3_xd00KVm4*qx{vWhxvW5
zt`Wyrn8`<dX?D0j=bJv+RRD44R%4pVse=gFJv$ko;}uPuJR#vectAq+O?9vXY!g9D
z4jr%1K1WUmQD2izr$g>K->euSm`(&<uK-@GnThq-AM<BXK=Bw<2K4ve+h1^Q_Xk(R
z5N%mHGSbpx07Q~aSg2-3Q-8f|kinqtjl_F1ee1jJ9BX!`CC`yOcxvOov9Yc%hqjo*
z_7d@u(MSEUI<C_7v5f3w0Z+52&7;bpu1M!J-F4n}FQd0sT}WE2Tku%lYvfUb#z~RQ
z<$>X;mk;$_&Ivy|OohvN<4hJiD4o%wsiAR5WR!=8$0y+d#GL)OK$N%VbZ_xdi2JQ?
zqs6MYQXX|5@Q%M+nLqh{)p;Ym;#-4fj!t9&l|jIe(DfqymY`>Hk_8oCN*IWXBb9_+
zUtj09AD-yUlX*1|Njoz-Tu4_LM7F8!mNx%Y(uK?YPOF}+zj<3wk~jB|-eF?Em2&D~
z2Y^}q`~p-{C(o76zZ)NRSuyd(pw1Q#&JVa(o-xqynu?u~&!D~YekCW;IDTh?k!^D3
z=3_wQF)75eSCrMkc=09+-*OBhj-<*>d`O&2+p@c!b8j*!t0bdAe#|&U%)$6yqDktY
z$f>Evrl+mV%uWV#ilOFS)n+fOT)c0rH|ZPQbXxcIQbkTw1Ew(gD_yw8jcw>6P0`a0
zB`G{^-vQEo&%xQ7s5!y%4bjB-Ot)_}3GRc7<!lz1xYdsyOAT$6sZwjruD0j5`CP`Q
z$W%G>r7jw;$OEv?HWFTw+V}gc8xnM{C5j#`!fYW7mV7pPRr9k&ZY~wPw$#?ve*5-o
zLm@HVDv2Sgg3j4HIwHIBIr_@g!~l1&q#+uN=405Ciggk`Q+mX2B<m!Vo!B~8f5yO<
zT`nlo!`6D^V8CVI_D&tPvMbfEy~q97*B5HzkhVRA=RM1t?b%irUWmk=$y=!xnmocE
zc$3^~w#X)%EX}ge;CU{eJu!9;()IwdNVV=OK_Zc7gEefwCZR{v=3lhB+dGWpM1@%O
z2J^I{teamV9SDZD^(jszy+**GG%El4Ybi?h5k0`T1gfG02b8$R`}-9$163~Po}aEQ
zJnClamQtIza&OD};Mz#ay35Q@tbi1?DrMw#xvqRWr?2^FB_d*+4Me_5N-8ccwz9G+
zJR3+4SqA5X(Nwc;PgE~IcBSp&KZ4utKDhiAy|y&HwLaGX3<0GGg_CnDc5XHs8Ptv%
z@_BOpZvL55+#mR9y%Y}<tu+W22sPB*yHV0cPxo-azI<PlR%NshSlfF4{(B{zUaslD
zwYWT%&5r1cM2>V8qd)eu;@S1MEQ_`*V7w^&`nXk1!?Tcr->a<qc@qBJMBSOrTH`bh
ze5<gk_qB0n{!fv6_w75)q4_n1f#p2L<&XpRz!YyOciF+yJ28b5hlOfB1_I8@oBKE!
zr5-w?=5hjhqt-?z@_QECA6X7&)>-Pe5vWs*8684#0w@rY@4#J3CkSg4-$q*^!ZWw@
z2ip6KN4?QZE#*aik;jOkf(`aM$EZ#s;LM$3R|j22sXYi;A%z~pcX}^RhW|^9&l+W;
zmpC{k*QQHFTxJ=nLX#hi4%1n96dUNdJF$TCMPy1!x_2M$Vw2gk?L36<3knLJKmU22
zZctJX<AQVsG_O@q?G*ZE(DKP-c#8LSur?NV>>=<;wc%_)n006TJXVC<ocWff9QH=^
z%=R8&pNb(iH#aXWd|x?vJQ(?OmC9l<q3D+6!i;I>a-5GkXXa7A_H&X;!roV)m;|=c
zxX?OkbtqDgor9yRz`CKBf#rCq{u>CKe9O~LsqtAtcEi716|Rt8qbbsz9~Uw5y56h*
z21N`P>`Pf_V4^f~jcv0fFA^4xdLE;fzuC?{zxL~nM~rI0y;IkEqh=G{@m4v;vCTgD
zp!jdEzBFVKc0E+C!?+q5S@3m-vD<e-Df~*yRepHPtA)2vnMH50_Xoy&KzZ-a$3>pp
zM&-F@adNu4z3VG;8|&)~tWt-;ZEk5@;t+5PY@qoy^(*X9mQzz>kabLf>C7id(E^84
zULmyy9k-d@p>~H{6uvQ7c(wYIrwK&D{oz_`lOoNA9XWwj&DU}qt@0)WUD^`8@~X&>
zOCQ?yFKmtk0vc$*bzUo5dd^BU)a^^Tp~w^N(X`}#x^#ri?efLX#O2elj(Y0n7sYE8
zladl$j6aKva@9>l(B4UJ4!6uGayr93%HbryFT+DjgxD+Mh(m`C0d&OSa7yYG=LkuP
z%pjAkXH6gX6I;5zz`&-Cwp*KMHOe@qTU1iDHMwceiR+0|F^B97U4;l1%pp=z(nE(r
za?Yk7?{iQ#6}PVc<c>4$^jGH5-4cjsC_Kr3?~OzIL#J13S;BME`>kU-DAC<Se3%l9
zyl-FtEuN8YaYnD73V`P)QWkG|ei?{a&|4cRC|}K6l`r7jFw<zt3%%5I-okeHUxchb
z(B&zm`mE>ApJ!*69dl*4weFjKi_JA7i_XOpzt9^N)*z%+dOvQizp<Cmo^w%aI9F)(
z;K<e*Y28X>**0mC8-~Mz-BKfvmbKFtRIfm|V!Qu6<LoLsU;a#&3tQ_qT_;3EfBU;N
zKP;1^n_%z8JzOm=&!(@rju`M|$a_W?R;#P4p?y%R240$kewDBOMn@nxp~+?4KEeLn
zmBrbZJQK+Pyq~y-|BiiPv(R}KZO6`Vzmy21&QXQs)<L9r@ti0hka>?UXJi*AB|1%{
z1VS^@5wd_}Fl~4yxIi%Sd4-Ly$+rD)d9-4|@R;OdO$`isovO&aS_$OF&VGCvzr@e)
zvbDKEFBue;%fNE4?Khvm&W|}nMn*<T`hk0^s%@7|U=%|D_f!IeL6>zYhrN^%p&B4k
z12TTdy?ggwym-+tOY)*0Qd&x<K%rE-3~|(s$Pu!0VhU}3Jo51o2k<?}B?WO;*HCuN
zp=HFnR5#<PDkFL<Piry$-MaRh5Wss*y=K0*w~UB;Um=cwM!>&hWRaz`{tz|eXMRXz
z@5vn~@|`Z8J-^bMA{0_g=W)%Z+xLmH{jbKG+Zp;4RnPh@@&fJOT7cz&U|@zEot)H@
zw?l9I)I@M8%KsDf)$bvIF$W&t$8)&yyWMn`vYpuKbmORq5&KjSfdBxu0XX3>@Lrhq
zJu4lS%RkCQw&wA)Uq$TF=U5w8em^t_$!ppEy~xQL0CkFV=w5=8puG`mr6p7Yt`0?4
z%x$W!=XVN>IA(RB=N$X9Ew<)pl-oQk46|L&tT2}45;7$Abi6}S6CNJu_^bb>5roPH
zc;_Fsp+hvH2e>0nF1Z~xsK*#>T?ptAbo0Gz9NjiXPr4dP+y^~BB(z(un>W8^Rei&J
z7s8Vy!smcTiX(YMX3R@5miwO#o<bFaHK9|EjgBrfs5)lv<*5<kCzAT{HBBmNDTs6R
z{_TcEj+i{tHjnQnM}oSO*2>4OlyCdOT!#+Aez<-G(kchXe)m)SV3a-wJ%({+#gdlQ
zQ-YW6hQ?HJ2}{*|f;meck|;OUkkvKS^&V3^D;?3Df-}5UKeuZrY#I&=JrRWj;f=C`
zbj&&-41T5j&WSg(1&VDI%Ke1&w%kk;IL`?zP+u`x@uH{MdWT1?Y;iPi6q+ej#T;vi
ziycQ%7tgl^Jo=F(u<k<MV13dDJx?@i;T{L_lP6CeKmJlFJuK_?NkVMb#bv0-CTvY4
zUpYhicHL{avWg~cLtrseZbpO3%W_41SbbWdlXD)cPV}iqqr;1nT|kiIdx{Dghi>o3
zVdvgn4?YviZSXmYzi-24;_G`lUa{cGX)lX@$+$YS#9$xhWSe<?b38IU)538bHMzRP
zwQ`W|He1E(z7h|&gj<di?Y2u(IzU+f13^b;{fO(0TIOnh***eS?1M(5x4Gi=b@%Nm
z%sdYRgc5Z3AYfkgab<7~2@Sn(Vv^=e#rTQ&ht3{8?o$zw&US?j?}Xzz^O0TY4T{r7
zOtgcPGpZ@c!O!qx!gf{k9%WB_KhW!4wWlk=XDwFQ=DBhdSu=OxQsv{Wqej2as?!(O
z*Vn@%Q!<`wYih<cGFCslg<e@%0g9iN*D{Jgw+Ij}Ej+9A6@ZH$Kl1q>(`{C+Wo?<b
zZi{U5ci+75-iD5{-D~03scf#b^Y?~c1$Uq8eY!j~cVXHEW%Kr}?Ww^s=Y}ua`35+5
zvSY_|KRLgLk&hr`rDRnHCdEs8&$!JmPg1LLoH3}+!Zr(?cAM_Y8UK*z6E~O^NkarK
z3USwmrlxw6rN}|UacYPIA)8fj@`Ih##^U$ZhS)sUXM4!CNS_aN{{{xm&dakybp%BR
z|4049d|$wL|Lq$FTm@vTR;K1P809ROt~1jKLQ7R_g7@fwa)^^o6YW0+2a{18q4dB!
zLB|`O?%!_ZxzytX?7BS)#Xbe*8w%8{H%;n;)4(mA_$U`AUjVcjw0O1k*3sV2{@Gzp
z_7)V{4b`5Ym!J>~iu$6bxkFs|bsDRwwJ!B}ZPhR(5Xfz9?j6qLL>?B3N%>g5_x+bM
z{E*Byh#jj-sMHzu3th4@GF?SZ)Bfa4hlSbKhx>2sP>uf2tt2ER*Or%6-pnsuB>N*A
z@prbm#K;&Y?BNRBse*|2=pVVsBV%eG|2AV-q9fCskKps2c2!xY?CskhJvUszU=ubN
zUeO%h=`xjdl0tL-&>@!}z;FMV*8S8w3?ZGLT;_&fFezYA_}MQb1f>eGA$d^=ec{3d
z=wq_4yAXse#76R$4=yXSMoGAKsDHkcLX+>e0_MIEHRJJAEZ{LeKfi5Wdq06F2WtEN
z2baxt7fHz~_{pgi!m{kN&%c`T@bNiU22fROu)<V#+Ej<a%)(?J@G8U$YGmsQIXNx;
zEFpm<*%@r_2P$f6>N98VemQ*}lSo2rXI@_7U7GA_d!w46uIrLQAiG8P?DR7)TnTLC
z9_I~|80apvhhE~{*~km<J7PcX>)S5))Fe$lPFqJuW)n*I->*5`m0ETW__~L|!3{E{
zC*k>iNWLT+%KCc2CT5{ID|?{nfwCj8hs4`E(=+Y_57C3A1jblqrNR&Sct_9>WriMa
z%U+rXu?6+1zlP&)VkWn@<l*9)21$u9s1jI8X?NdtxP7>w&r<+C77!>>@)X+L&|C{1
z1i}gA0}1Z4QuxySe+|jg#C$N<!hWbW9Ejxn_wQLY3J6QJ^U9w(>uYK{KA1HDtyeb>
zWccshsL#MwJUi{?;-aCU5soi9{ztM@Ps$TGdpU{s8a1J+@m$5H<#^J=Om^e}^{=tw
z&rSgyny!>`g{z0~JEDdoN(K!CrQ6~Jm9V=rf63{;rpVg#`d@o9d!QUacLRa)Y~Qn8
zBhEGDf#Rh)gkIE_l%=nLes`On@g^lDdAPf?v)8U{T_G-py{#?F_$IHtA!oG2{%B8h
zO-&6U_VL-=D-6EDgh~9vxTp3RtEhx!XJ<>Nur8hWYnskzQ2w0e(sdt+b7Cbh;}ig*
z2oiv3enV>QC}%VjXN3NA|IJY`#EwLaK({L+=Rh3=q|Z3q-5*SFPojq5YgiY-+ek^@
z63Fr2%CGuCo};?5`Pj#gAG<A0Edb~H^T{nRRHDknza}3Hdu(K+YDjT+Ph@kk5V3&_
zL5;Tt*5Upd>@m;L61e;QyCNsveGg&IL~&2lJv7Xy$vp4ebyUB@ja`er)NX8QYHDM{
zU9}GcXW&uWZ<uaO-5srohW?CNMho=se>%(lHI^nC6aawQ`rWoaB}RZtEyEDwfK89#
zwW?iuw;N83U88-)WZz()KYtEiuoGe5wGr7gta6%s@oX*I!g$+=r{V6jQkyPiGj&n;
z?go-e5oWim#5x7R=m#nTp)lvDDBF!n_b*yWNJz}O=E&*)wM>avkQX|Qe*Y1!uDeZ#
zGd~x`;Pmta99lOGMNf<%FFW43kzii&scUGubjx$NKKit<hE2I&**)$vq#zgsonh+Y
zE7yfF_<K03>Xi^k*LO11G9j^U=k~a`xE!yJH!A5T$;oqF7tE#VUh5XSg6@rF-B)Jm
zR(2bxPNHV@IzN39oN8K1&}ZvO+)71$?;G!#GZ5*x>i}I|i7^wep>TE9(_8VC&@}?E
z&$cd=guDAot=c@V^B+A!BO_2^{O&pfJ3^4YT|^+%JFetgcFeYFSwl-l#_6zk+4Jyl
zql=|K8jHV=FF;=~A|m3FlERPE`?tdVYE7?7JI$wE$tNA3pZnS_F2v5}c-Go_`s>%P
za%kQe7MZ<3YC57&AbC`Rra3HHgP5lx3yV=#^Y{Y%tFOR1Az9>hGad!i=a(%(UV3`(
z=*3{e;!mHTXPDA{v(B)~K{-XQJ*F*=#d7AvrdQNnic~>m4ECBu>!+n&H>HM46Ch_y
z?Z~h*9sF^=A)VrP=Vzg=v^jQLpX0a1^yVHKW)2H{-F5zTGgIUELPtl(;^LxG!flQH
zTFC9Tv=~*pLpc@H&|+o0ksMmQuJfi^T3YD}k980fino;w#X(WS5@%K|kxqzAK;kLH
z@?9t|B=`s<Fkplkr1PlA$>%Cg>s8OudvtS)?;%PgUfm`Hi6_U-Q3F5{OcUKr7Gemc
zrvU-6(3JM^k$uL!>j~9;!_Z0fT$?Na86*(iyY@R<Km|bc<jIS8xlpbfe-G98(PhpU
zv{9ga9TH^O^#wyD&-wHBzy$5$Lm(CHA*Qg9Tr4!Fp{JhbFy5LI_nriPu#*rkU_x%(
zzFiHZixJS?1gj@@%0n`-TboOz9GV}Vk7D=A5L}2Df5hTJLg;})&ClbTjQG?;Oe9hC
zf;&Q0KjcTv0;CQUjB&BCAPU^}qMh*)u@e&$(EfF;4P#MCc+I!#V1}kV*RQ+I4%S=|
z4Enh9d3{BN`0~;TLr9D+^P{*BdP(4<F5oq!CH{`ON(EFD6cnJi>6)v1N`MPHQ*7h~
zMMivRXd{$ac=<cjglA<DB}ifjLHCuRNT^Xlrj}F0m@s!R|1j>zUT(JyL(GMs(DTW~
z-akmsb$5;o<~$-IUeGv&{*%(#PoR4F9eGJ{auw`4nE8#RzU(_mz7z+!*mk$}HqKaD
zx)wyq5M}E(s)u&NxY$!(??;3SxT!+ALihg+=%NE{;UDEstV=g|fx`V~cYi+?;73MI
zQ&LteMN#g)@MCK<LAeIg12HBaZSK$ijMib6*9kEblx7KSdt3<cWk-8h#p~_u4fw>|
z+PY!<!KYow^P^=L1O~8%!iLqyIVA@`$@m+h9PbJ9!I{p_%xHYH(XSw5+`s#>HfHt!
zf+01Sf^u;8R`j+6RVkxI`t-%j`g&QAd=To1iY^f6YLvLIfDG0p{HxAQl))nr!TyZ%
z1@H6oL2hRRn2?}ryMSTI16`WW<Hzmo?KbI>P+xCDOrXwG-eC=@f~W#3JvUZo-fZ6?
zfIlF)ltOqxj{<xx3U#&0H5s}=6rji>NC!nx%7-E8{90UebaWiP8%(j|(ookaP4PA)
zh|rxrjezV>vT%D3yfsmZD7DgAkHu~*m2MS-s&I;UYX`Hzf|vrz)Ya7;p4B332@x%a
z+I?^Vp&e|C$NEf_QZxPbLJ^akkrfXc+c>mjwDT>a#r|hBih*ZbB?QVXBO_zy-VXdT
zGPeiqBx|@iLF~}-_kbzZbI?oz)dI*kB0taqZ~Qw4Q@uiu42_IHFV(mvlu`AKN(hEw
z>aSO><=wY$-xE@*MkssI$3ymlr4d1WXyb>KfnX76+{;FB-}Hqw+)bi>$juRGwt`53
zpmt0nK97h1lLvW07bzJ?op-<DuzkRA0s({&BcswSG-y2P>MR<Qr63MqVi|2gbo(3O
z`ytO9AtSSHecSPb&Xd_h2#BNKb}D6SHP!xpV5}trWUC;b`)eh|^Id*?@nMvSfWCu7
zic*c<Nv9q2ASlE5Foi%wfm9c2YyH~kC7nwIhbG1&mqaX7OVAZZg4lC-p*;uKk>7~l
zB_ErdoD7`mDB8LNwlK?%a^|n8&c8y3g@s)Z^@X1MOEAsah_B~h-3GOBG;_H;O-<MM
zZ)H23=<!Q_i3Mqwf<gn-EYNCY(K+D-&wp<4ePlDwsrB`goc6@q$v4-ptG&-l*DQ)l
zTk_&erZ(a$B=DIG&}S1C-t;CDzWd49Hj+~(j8pO?;n$$#KDRU}G8F0hqJ-atjZvWQ
zWt$nTZU@Z*a<f46N#=6h`4#-ouB4a#WRIb-anPB|KSj1y6hML=mL<-LdPP%<4;-ZO
z=$&>WYg?JMspc*yT{@d#mC@!|+_1bNf9Ra3r*z!tluEts;3Tvh@4o-(qtJ_SsO#y`
z0WsUqkdXWGue>&0wv~$SiK4kVIRO#|NMu%CULGme>>S}r*pPK~b%oDTQc`fH^$iVP
zi{el;aImvK6?h475RU?5Xl-5e(R%~ZOn^XT_o{1ZY;A4tK&LUYwWUQhQ_aN4s3l9M
zivXhzKyw)mXh5YkHHCze2QB8%bu_iGK!=B`g1AO$llIig7r>L1m8so*$WvU$Xa&zu
zP<*iJ;zb||mY1C?#zRKMA^%;amIK+He=|Mq*+DYe=C>M6U%q@18)I3&)i}-?j$^iZ
zaR2t*#)74WB(kgx_2SwFcg?xA?wg_m?p#@;209T*RvkHvz==Y(ip>}tfd&;QE*d$;
zfF8(Q`6!W8Q&(pYfJb58zGbhBQp<qOJje38{?h#kQgU)Fpfbz^&_mC`!GS<TURJZI
zQvf{pMm>9KaBy&J`<j^0&QOE92zB+gpHqi0PSf%2&sL+NqM(MB5EHv<(JG7AOipRS
zH#Rmt)@wE~G~{mD40-#X{y3ZKv+D-0M1;6MS`}w^qAc&U6+TV5%p`Pq>SrO}6Menu
zQPory>r=^_bd_Jeywc1wR}gVubSN~t8UtC1fss)NL@q2Wc<{F;WOSe^bYEY7ubZu&
z(E?pPV4Sm*(}IeM+`JDQH3#_oE3C~t512ZF1QO)8?Cf9rykv=4wdP%MW7cG|5JCjA
zKBq8}-6z#JsBu?)eCFX&BSL6Q<;0BF#oZ%0$)pn?r=FY~hvIfHn>w%(U?m9V1)Qd0
zMZ7#glnt!~z;V~in@u5r0C-0eP@#+c<xR<_fq@$nxedSoEcUuhU#@IyY62N<7^k!*
z95WFR=<4m&1$C$DyP?WX%};qo>D3>COLAoDbm=}!FgDasfH7sheOqtwJyJGB37|6|
zM@?YjJy&a>l<Mf}>H`j=Gmjt%IZQ^@P+$KR?!$LLUInv41FmPeYtKK5!a%ENW3kH`
zm>ZAD&b%YS0QaTzzgy&K{y6AS7C}Sm+V2esXss^`4GRS!X$H@vtgOtXgQ1|IDS~wc
zGo_(HgBZ}&nQ(nt?6jz*T(g6A#@x&d6`VXVH66_ri8#)e=z5r#nxYL*%E!cwsbSLN
zkVZk<eOS1@wie?&^K)fxq`<1napc=M6O$1{uX7S)N?KZ4NP2ea+5u>UHJnob+rJMB
z=~H$66^;IR5ZZiwhJ30LBU}FM8=+^ZQ{>38%ng<R3U9E4pn!n-Z{Ls|IYs~(dYuOX
z4;-bwt!I4#@ToC2hEQ^T|Ni}Fehoe5!~xJV_`|!hsi*#!m{^}Gf<p>ZMFE$Kb-f@4
zKtcQD$&)^aCN(wbFfHp(GZN6BOey)=I=%pZSp#*T3Tt?9umULDq?bXU_=V6#xD$kw
zR>)K<APDI))LvYYyB`3$T+kXXg3Rx!zkf}2^=mlC0wIndsDuhgcgt%Gg#pI_<g*25
zjsTo!<rqqtd0Ill0Xm8V@!^mnASQs_OCUk(FDD$k0ZOE+)XPgnMWxIiiqOtvsTa^+
zX@PTPSZrokX)%z#g+S*G)kQ)wM#sc-W@;6HBu-phJbY8(ATE%bpjac+_g|xZiwM`+
z^9!Pfg=G~KN+B5mtb#`+Jb!Mpwdql;o2`@%<w%BV#^RUIts0nWyr6yn6{&BVu3M&2
zT?7nGE8u9kvZ;N%=ihzXyrHka!GljuFWR1hXbCX|F6_5&#sNU9fOCMcw!OaN-F7Y$
zxS@wc6&Q_$g$2YDxbX7&hpnWK5>4kn=gRz+KT?VRIG0LjzJDsGAfTSV{<Q;iOt+<G
zyBU;l800rvARaW^sr!E)c=<!C5l-U#MgiL=5q^WQ-{&}f^GDl9SD3kTDT}?mMa9Iv
zg6=b%Gx@H1mNwK>KNb`@UVyJcW1qm=QgmKDLkcJFmY;lvgC$V#33+rmk;7p?9U2-M
z|6^I8TRM3zCY7sX`TqLy3@CS1pe9|I=$Hd)ig@f#x5V8AZ0Xr44nnz_sgdh@nNSD%
zj4S%}A5kD7`TT!-Fq5ZC*KC!Yo&9m2R?V3V75}e&XAfJQT%loGFLH|@?P4Pw@B=Sn
z4VNY;%A=z-Qlu#X@1FGmn$SQ^?NrE80G!BzCj!8QgP`*3GQ1=tKmVUjev)rhilhvt
zG7fcKSEpRq447C>aGc;@yP=vf+2!qe?VU2_?B?tX&dEOeizP})y73+{*Pmj}H!{X|
z6_0<lic@~p>)ngwv_$U@=tx(XS2b@}KI?jatA@4m`9}xY3{L&ivtgWK)R=aDkLy7?
z9@pr{v#k=GJO}zDoAwutfBR+A6?JUiaIDlPC+2tR>Iq3yj~3KF|7!E<wH9$M>tO2e
zfAXl;T)<W9)lGaTLl|AbeZ{6dGFvCNJcOSa{xday$l_p2ey&x&y>uFiCA3;d5T7^f
zX<{EBk#u>hA^SSNYnI!6MGSNNZ1y*EJ^S`n3UA!&(t;elz!J9UBQ1IlKehL(<yCDy
z*N5XczLFNbnZ1m|5xV$S$)RqH!yKn+x`OT?=E6C}PXJPmq+41TOyNujGCMyZ8a8W%
zKDIis(Hc%=Agfc3RSkc^(VX(^^2U0!NY~;d{`LL(FPZ8I5|Q#-smd7ULwB1eH!4c6
zckon3y)nne<QI1E-_kAQj+oP$Pb<3a>ebJ%O#k{Ru&&WlV_&2!s3j%m(=Sc+v2@8N
zaK(=5SDdNy`epE|+dfXjy-On*c~pUxN!`)$y*jiH#g9f(cLcw<`ZMseG)Gna0$xL|
zBI?4>d~Aovr`65R61O<zucnm?i!@rl_j1z;*^G!dT9F|0%gEi%cDl@h>Oi<rG2+Pw
z4?&Yg$KaJG81;=gzeXQhqh@W94<iK(`;^;U4HO$z!>M>ub+ji%SWNX={AqXtGC5iw
zm6;kg2zxeH2OQ9E8h?w{5@k|Y<8=G=6x*cvFW<2Myrg$(yqg!cL@iu_Rp;5Gud8$_
zhbm*4oPH9nP1j_X50{;9*0T9}+l!W~$2-ptMq1OW#mSyDV2O%!oK^$*QFrD+D%VOo
zYj=US<)2%^MY_Dqyllu4Bjy4w9~p_cW*7P*PonY+4+B%mU4bH%6K^F!CbB9ZaLI9s
zGFs5ke!_nxK5zfoIoYj|+RsNP!kaTg5_4O1ZkIY4N_}-IHDb{TUl&`?4Ueo~D;9~v
z74_#`S{$G<m>Q)rIH1LLgfjPi!TXZ~2}-JY@&?}5H#Qz_4y)__z~BL!Wm>13s-C=~
zx-i<!+&?OXt%_F-kYWnjz^(Q;?nS=1b}4JiatwQ7&2!PCM-8vZU~ngMJekBH^x@^y
zg7+%vnU<4Q?BeT^YmC1HUCjoo4iwO(>NvU}O=ct#g^|dD*1p!*tM6v|>K;o69#Ehg
zxX`EzTG2^5rx_XAEANYp8{Qo0X+3**T2u7{{+@fzVh<*9^|Gke)&^6WXI@`e%@>{c
z4N)gUrSg}UC!UMsb4(`~l|g*Xtm7S}a!Pf~74@2{+x=dE*_GUhbLTkNHiE<Lo8+^L
z{O7r1YD+pE==Ypeo-%2pI(xM)Bh&kV*HhUHoTj?NyUmec-1~=;U7dTkJ}af<ylx#m
znyAHoN{w*boSgpV0g?yz|35og&de|_JO6IR;wPNN?v{Bb8KmpNp@0u1VI}dVeR*6U
zxJ+lDytLa!WV#_-_}N^aOLlrnBjX;G2f#M6>g;=f$LR!3@Gq<MxO$oU()IEWUwiBK
zITmJ5A?G&VWIP>9Yd^JK-^RwGbL584y~MqW6b6-C0>L!q43oXo=p+;e3FYIPWQzYx
za^<GQxto;FRJ-=Auos79qxeZEAHkj5Qx%21MO~8rph#TxuE7EQD8m02!0OkXL5h|y
zW){<D!Z;7T-1{82?^AV@?fCjp1B2iL$5<;-{-p_S+T}^AnYGf9`xSN15Y|={weSC}
eqx&DX4%YCT;3jiwJp^+kkrI~|%er;%(f<Wjxn_F+

literal 6996
zcmd5>c{tSHzfZQJvXm`bCCQ#WWD6nNV8)1yB7`g<%UC9RA<;tFWg8zm$!?OZ>|s>4
znMg9WNyab?_cQu^f6smHbD#Uqy?@+4=6TNNytmhTIp@63na2<_gX1ixSq>dKblk{L
z&+^bAx-Hu82t7EHvdLz8=+I#@#007je$vs=9X@=Ro}Qk8fq{{ck%@_knVI>>kt0Ws
z9%W%+VP$1KcI?>k<Ht{&IKjro#?H>p!NI}F$$9eR$y29JadB~Rb93|X@bL2T^6~MV
zK7E>>pI<;gKu}OnNJvOnSorMOv**s8JAeMXsHo_L3m3%1#4cXEC@wB8At50pB_%B_
zEh8f%D=T~H(j_@LIeB?`1qB5~MMWhgC1qu06&01smoKZTs;a4}sjI7BxpGBALqk(j
zQ%g%rTU%R4N9XF*tGc?ndU|^L`uf+dT{AE+Ff=qYGBUb;{kpNSv5AR^si~=%nVGq{
zIRpZ+u&}VSw6wCaf<mD%7|hz*8V-ls*x1~-al_Wu*3QoE=FOY-_Vx$_!ok7e)~#E&
zZ{K!wbaZlZx^w4_v$L~{i;JtP>)pF|-Q3*V-Q7JrJnr4Q=jrL`<>lq=?d{{^bN~K*
zUteE8KR<te{|65q1Ox;G1_lNN1qBBOBaz6EkdV;O(6F$u@bGXH3ia^e!-$B8$jHd3
zsHo`Z=$M$8*x1;(xVT4;9zA~iI6gi;At50#G4aWhCrL?3PoF+bPEJlqNqP3{+4JYm
zU%Ytn^5x6a)YP=Jv{$cQy?*^VJv}`mBO^02Gb<}AJ3BikCnq;IH!m;m&6_v*`S}F}
z1%-u$MMXu$#l>hex}>B80D#ic(z3F$^78VEii*n0%BrfW>gwv6nwr|$+P81t*45R$
zd-txszP_QMp|P>Csi~>Cxw)mKrM0!St*s4%!MuO}zP-Ku!-o$Y9UYyWon2jB-QC?i
zJv|>ke*E<5Q*Uo?UteE;fB(S1z~JEE(9qEE@bKr)pGQVUMn^}-#>TK%?D+WjmoHzw
ze*OCG+qa2{iOI>ysi~>y>FJr785|CW$K$_$|2{iAJ2yA?<HwKr`T3tee=aO6{QC84
zadDAAAS^8{EiW&xtgNiAuCA@E5sAd#zkjc<umAb;XJcc7L?Uf&Zj#C5t*x!??QIH$
zva_?ZySq!JQup@u_V@S0#Cn6l(m(2Bc*F0|Ay!V>k8Wf-eD}~HUV0-v9cWPIN>*?v
zw|;uhYr#Ngfo?2cpg)}1b!?jBT04iYuc0=Bg`VQWpvW@{371$NeYwcOBJ_y)oJ<^<
zLtBqd{~9kH2bYD2@5S#d5kBX2t{yR(X`%L9@HRB0g|ev>7v4l5Q#Y8O5mpz<TQDuS
zjhz5ET!H?sQEip_p$~NJ9xRbxPBB!9A3<2Lw-lWTGSwlbN8*(j|6dZi42U~^hi?D0
zX5!kNq%WdxD)2rkwN--Hlz0-1(u<Aa^e~d<mV9Yb)-u(Hh<e0`ssFGeBLBm<Otj9X
z3-P!sobss;F&4>e6lqI@v|fT}&_jpfY>8)#Agn?F4-??}kgyF7&mUE7=Mi+7%tdNS
zp!e2&o%*^4!ozRm1v5g23>dL0>i7Kls-4Kc+J16DG~DhmpsvH7w1iI%*Z3Poj)hm}
zGk!4sEh8IpKh-g~(9xjYXxT=;x-)6zqF?QeML$!!ia&y(*`Jk%E-7bE4qLV;>+F{%
z1~n2bgFf~lGDRH_;v#MKjy)1Aq5;eA&ApUEe0@h#Gj^G9-^;j*H!fN4ZSwtzPx+R#
z6Io#K;<nnp%y)EK=-fiL>z~&y-tBW}1TWlxq^qT<#6Z@Zu_Imvc)ABHsFczHY0&eC
zVZYkOm;>)+pjxiSM)jnFKnYiDT5(kd=sLsut=ISX>+pHV+X665v6IrI;Q@mI7Ds%n
zFH(n(@E=FGVy_=eGGlvc-`Ocy(Sp~lL#oug_02>&=*;U5ap%;PvB)?=Of{U^>TO&9
zm&2N*-qcwR2d#W8T;s{)`paQO?bT^f`a8A=lX6n=<^rvlAYOIa>mOKYBN$Hz&#(UT
zVIYHF1IK5cKAHBkwxpSkEu|nDl|w8J#>Eu<j+M<f+*gwIoe6Z!*aOAi(R(GXo@8(q
zDw|kGAQw*Hnx5GbwK@BT`A+^d4W+95r;o6DVSxiBR|oCQ><+4<`zpy=*VW{|j16zW
zi@N?%avO%CvMjWs8QHL`9q2uUzC48s;791mwL7Z*-3$j&?FUfXIx2s`b{9}aHo3`Z
z&UxHdh<zH~jT4CnnK6Z7NSuTJXmtaSs|RD|J)3SpHK)I|`vm5#@Q=>ZrQs7!UNrZJ
zx<(SHUx`QE+jU7F6fIk)|52h^ft>sonp^7smV;u50TF-kT2&(B{xV!q*j7y`FQGZ%
zXF}Q;Db=Gz0xH`thh@8Q?kx|P9fP)2k#`yg{t%!ztlK{p&qH$*tbJtN+v&|;)bnKB
zcy-`T9e@f-{ROT&upmk-S~LEdX}L7~@4$)61}gb?M;GAXwWbga;YL8~mpiD_sJ;GH
zMlR*f?~%)F>f0y6_5RqVD|ve%=&*lM016cqKoeTS_r(@0aC2uvZN3i6t4QkfZP;up
ze=j^U1g<9M;ug#5^(UO_0R`HQJ+T8DHE`5E;4tid7Ah>{6*u@Q`L{~AIXQ0x#HPR4
z+#7uQK5K!%R96oFN99>`_S4RMigQ#}F-Lvm#7@we*qy)7u?cc3atm&q(qP`8vU*1D
zF@F?iQ^CqO@>_-cPmP?F^9MH+2!ixDPXW>z@O4PdiH#O9vH-stJyzjj1x0Krc2Jv?
z{#|h?QG*s7dn>J_W6+`8w}dZWhCrSr{h@)jfMU?b+A~!r0w$8ccgmXj`E9I&9vUHb
zv;SWT;ivpBYaZBwlT<=|cRkPu;qenh`-3`Rowes@%ZecIL1ENVHpRUYz?YaBMcIFW
z4E(wGB_os-tAH+{tQ663Ubusj$!DSA6tkYb=l&2BgoZsRK5WGbtz_IMqchcLNLN%}
zbUzXVy@QbQ7vxMKtxWegUOng*qLtmdjlEdUC<5OLD{4>u4aKXZ-jjV(V&&X4o&4EZ
z_pmIW^F@@|Q+;C006c4w=epE2aGhuO#rBm=<{_Etps4M@@fAqxM2oC^1X_V+PZe53
zyucm<g0ZANw+i;iMZV<7)E?ub*&Cng&e1*vVlf~sdD57;x)<!(rE<Zrd3UJ?PLxH1
zVisZg$j1Y=XvB>RrfCk(!LrmazFoBP{4j{+4Mj)zTSY8^YksLs2U-Zq4Goi+cSsP8
zC{7^9FRg<K()tWqU_`Q}r(rR@BH!W0iq6)d!>+mc>#CoCw8l@Z4?Sd|ZF1fpnmn@V
zkHSdY@T1zK(3_=gOHT(UqYEe>bh3(24_1d1>NDWlq`gmk{5Ed5cJlJ`(QVAkgxs$0
z9~Q9OjIqrjl044{=Hhax4mNWcIK{W2%5pl_(g^X?4}GgwX=h5-`gIumUdq?%_2~-b
zkUQxa1M6(a#ST(<)c*9f{9jHTA2o{eL{V;$nys_gLYq)2J9)_{$?)emd~r4+<iW=c
zEi?H_@SJO~qUIJ>#K$~<V%h?(&WU}stHpy5O7oHRuJc5?B*zV(%aRLhv=Ec}1?O5!
z4~JS0u{&d%wIrFbl3h-+40btS9K!@dM(wpYAR4;SFs*1U>p_$EfEH){w|*A%3^UGw
zYeS@grEPZG*mjrDIqtOi2pUV+z6`YWkXI8Jiz#X98f}z^7Mm*`(>*LUjKIQN0flUE
z<P@tJ$&Viq<Is3n<0cz)?KN<0#?$=A_f_xRoyRB#4&WH}&wU7W)JiX&F&K*5B-nS?
z^Q}Zn=#oO0VGfzr;+9e$^C^T9SJxa>yX6KUfZY3t{G5131~=IHxyE?urQ^#yq~<$-
zCoA!cAvUmoYjEqwE`JP7{vHgsBU$aJ1B6?b&r1q%gQh~9zLgsV^QyF?rid#~`|I9@
zf{lSPP`TP*S7qY^<cvHRFS;(0jcHiB6hTe{_b-0zQ05t1DQ#?85{Bqtw>Mj6w^FY}
zhYiWwn63G7C3*6>A{XNu6{l#zHW7x}=l{^<W1D&w3E;3dHqPIzG9d9iI0B0$X!x|^
z<3wlf^X#T8d1z*2shXhW!G5Kf@&J)t`yS>HUITGPXS11;(vD_Q@0}qgNC#$%Liqq5
zmp$|kjno~?Z#N9DG&r%~R8HyL)JMcKmQY}IKjJ~%+cu@)y$ad!-<v3ptVYa!%C?^L
zio{e2<@JL2jj_1j74?qoC3_jjgP@i{F?AU|j_tr+@-I~pG$MIf_Bq98HBy#9k}cO<
zk@7Jce@IxVf=~$&hy*Yrr~!1OHH;j<MN3pQBrNr7AHbml>}|aI+<O@bfUptKna$64
zAFf93kM^eHwKN<KT(bys2FUAR?6)z2>MP@;T@{oV!uBNqMYyrnbyx$yY5i$;Q||)v
zv?|WQfAuWAd(PBi3FY(J)<y0D!cs*?gMZ{lU2x%<063Y=@K1Wl0H$&vA;IP56|XHw
zL@%V(qhd#K`{|<07LMo3u3uH%*>yJ!=umM90kgVRYLg6THE$w9f?d01xzQkW@aQ%F
zGA$rxh{c3=I|y&%_A`~P|FB|0&xj^mi~Wt3c}n;rB)t8LNd(^60ny0m2A(=Eli^Z)
zU$B#=PF_sF!l(^NL5w1mnkebxbrMwB{6=ct@ZwWr>7W&;)}AP)>scNJAN9qQw4)#(
zcFO-S@bp|gihYFMiWQ-N4QU&$09O`$7?y0egt4gta82&UQ<s!kz(tg`?W=v4ZAE5l
zz|*`H|CUwxC@rgFrEMYE(O~szQu8E4l+T*d!n*K^z>_dl`pOt|uQ6Z*|JCg(SpH&w
z`l(nlel-$<4zO(u4EZZxviCBs;1sJR39jN1Xxi+Oc0yq5^$u7RIne}3%M(W8V4NLF
z#fba@?wvmtIzs*Hy%%Pg^|WlkS{?)TAh89K9#K$g+n$*6e!U0zuTo%dre2hbru6u<
zhXC(*4>IFfC#IkEFi`JY8K*w+snec_DG3v_-iyO>H!kciv#p)(w)n{siq1BFLWpb3
z5Sd~Gt7mGWdU9#7cH*4pM*F;L*)#4I82~q7Wm|m)mg7gry&r}0+%{PJkv{i=5mk2Y
z)b3OngY>B5{A86+Dxy?(RT~v(K+@8bBOF~??BCz87|{BLiXrGfDqj_=v@ukXL+ti>
z>0dKyPRK>lNz~{K`l+#xuMDo^uH5v)t1QU)sV5P(rT94m*e=;Bzi`az8uXcr+QBM7
zVANf8!D0@cLxJxSS+IhUIcnn^?(Ivtlr*u+W+F0&p+nkMBrfr|%pwjRFuzBJwi-UC
zTtPEdwh`Ehp_ffXWSBq}OVUY%Z~C>H>}XBS7X{Wa(=iNVk#Gbla7al26l4Tk&!8z`
zf+tV<>5`h{LXEy^TxP1nbHS;bWt}MSsG0>Z&ahc6G(thyo-csybS`(x1h7_Lv5EqZ
zDmU!HxRxlS6|phH9>A8QD;@L_w%=t+a?l|ilPKNWqRQqXvuxy~mW32!ZHX+UZG9E7
zvfx?#0EYRMH0Pj8qOh+C%pUg!Pt}$hASci~-#Zt{1cNzb-5C3&jz;j$@bUz(fxR)p
z0^2i^(IN%P>cg}`ET4PlAx+UH8Ug`4{UFs8i@8*|NCx3P1GFY)6u+}WE1AvBS(}-q
zKxZvT^Cd1%Xpn%t>eoNH(Fo@2qj!&`yMz4*7nB+ee|-V;MFT5ravPPFhddsnJ#G)o
zUzHA;hdMYa#a#qjr%b|Hw)f{FSjnNir0+AS*{pSVm=!fEq0AOU)CI-taQ3;N5f8lf
zFD?raL1%0M*^f0|aRk0GOX~3aLDz9FP33o@KbW2u1aCz94LYznDxe5g;>#llFJ15p
zx5@9B+}kA!C~K+;ey1U=j4UO46W$G_xyYDLE5ZS6_wu+~)PVZtQ4SDJA_3F#*+mpu
zDWkEH^K}>`NdOMc>S^aeT0Eh7rOJ*KTiR>T(Xko<UPIt|q^!Vi71lZ=#*C!fh`d??
z&iUuKS3C=VYU(=bdj9U+?Eqw?%rsCMPr3=n*GJDV7tb@{0Eda*u_{$Dogty$Ku&a8
zlqh0zR!VgJdROUY18$4NETD{5yIAeC^EBw&(luN=+}?Y-qLN@5s8*xRi1qtSMu&IL
zfo?H^2=M5=yF2p5rqIj4!kcn}m^yBY0e~luY|5ho>7#6kpW)P~N_B6rf7L^fc1ERR
z^)n&}*tsM!=z%(#QTPFt$KO{$>NMc`sEOcBEzT&<MPk2>sV@sPfR|1Vcvw!QJr?X^
z-;`67R8+lbT_x|dr)<MH#yY%6IfXI*<~WU7MzNvLrPbvk3ty28ziO~Q1ZAzMDW`fy
zt4Z3VW8gTdvY~<zE2($8D4|xtKbJeyAEp&lvSSPOA$MSy2-H3+SU!Tt3btluM+mDX
z0655-6mUa_*e$Sd=T)8(Mr`Sa7vh7H!(f;E85(AREQCm-v(Kgj!kZB`(rCo-aR;sd
zw$426(DN{yU+pa#F&-G=leT3A;1M<cuPbqPbfmYA0DFU~A3&h7ze>Wx*nZnW!Zsc|
z^iFMxbAncV!Mk2K8;y!CX}dW8C=Mk5Xx#3o3c-Urgn<dE@8*dtIk3X?j7iQb<Y0gi
zYmliv!_v&04H_<h?lT5DsxpJ(7fbf;#(rUbNDvwLmG)sZf(;DyS!^X|V{Vf;=&?T(
zb-8ZXPa3WHVM<<hS!fG%D#d~t5*0c|i{=~3?wGEoyAH{sP-3aHsu_P0G_FT-LJ(oB
zV4`9PJUegspu(mB6bijLJxJ3b8K{LV)zZR6hwWi*S*hoS3hQ@20fQ+I_SIq7q2}=c
z!NnEuTJf=TuWt}hQK3kOHAs5mHe)`J3>xo*VH7+w62WZHdt%r6zMd%r6-dXcl^t(p
z9ZBZgJzF|=lLW=6hrKMY1z%V`M*lIp7x1BRj#po@nBuv(bO&pxLsB6lU^ihcY8j0(
zI!ah?!-StzHYeJmw8~dKJ<FVR`@roTv&Jcj8^7E5rcZQn(72})gYuewXX<F=Y4dXA
zo#`}B8fj&8u1JS!FcUl}GAEqANq8&T$3w01@Eh$>dw0rC33-#dep;f^RzIaB>E*)9
z;BTcE9DcIje7C-=Jh*GpN|0Rl-iU<lQd;isn`{VJi7!m>MP%%abvDPcH5GZpoe)$<
zgm>UB3?dr5%sV*Z^srj@=gI5w;&~J$j7OG611`o@4qgz%I>S~tiMhEkQjqd3OWj}S
zQozKs774ePeN0orB^jTa-|t(5-CImFk`3<{N(^B|3|fS>;73(@aG^%-m7f)?Ly#JZ
z^Gk`_zPNK}ALx)MFP1gMp~hwWRi8X^TE)BcsRS39T$W{K_S24m3#4xY3l=TAB})p;
zPqOMqM`v<e`-v#EB%!fLwq^Rv4oNYx`I^T4?c3Fy88tf#wOza5MTlXCZm^BKx{|t`
z21ohZ-P*w;jtfs{lVtMpDMg)a!F#bAk{3dkxBR<inrC~tKK-(|2-dh!4tC99i!8*a
zVs@k56f-NFKe6BXNrw))#u*QIU!TbgLEb@VB7gSo7GXTqxiu%6=fZN*!x{6CX%JZJ
z<0uc88TBf^xTLgAe#<^=mfC{`uUTgKsJx?fcwKdAyS9+IiMqz%kj9*09S}(+g&40x
zIwOT<G~Z?s2nP3pgJ(S22Jt_>=N-I<K&+^7Ja>ZeiB{p#Pk++h3IlqLI1{LZE3)Pr
zC>{o`7M0kA>r&%Jj?yJvzH!D<$|3JM9Z6@lwByePF)_#<i*(f{X6q0mPU#Yl9YOFa
zMG99V9IK@s*A;NvX<c=!h9gpJ%CZZX9VGlR)5k(g`lPtiw&~1b`pp}cjF7~KO#kg8
z>kIZs+2`bv@mW?gvwbtOlA+1$fQ0B0aYFLfO55|APRFC!cawkCNX+m18ZG|%Q}`mD
zx|$^dH%qp>?vCO77}A~ed>7ACV7wzYv+TmljIutLKQ=NRTTQsa_wrZ?70{nm2*@m8
z#uURU-U`q8q}-O;k9uLLP;)$gOHaJ8V6!iICs9!BTYFEltkjEbC#lPG_TmpN99K>R
zT-FCiKN8Y*U6*-39SiSPUHEZDWi0uR(-^nHiOn5hBfsk{Un_rdp(4u^;?^n7yd&Aw
zMdx?8w5s$Io~PH@NEh2U?_i+4C3kl7pWH6fgj>ozras#RjYa1pgE1IC88urFNwq>+
zS51$RxM-cjQD{&XzdS0jdcZwJ;joKmMr!@XCcKUd#@*t4kT%`_!87&JE9@t<SM#N1
z#CZ+|2ZJkfPfFl%(~s{&;-CJu@2d@rnT`X;ON-A01K6IiL~3=5l@IMIFKeDZ337LB
yOS-y7A8;=PZCw+YS)H#6NzwaXyqmA@KlU||?0qF%4*o}Q$VlHzuj1;h$o~MBAmG3N

diff --git a/docs/dev/reference/figures/op-hard-sigmoid-plot-1.png b/docs/dev/reference/figures/op-hard-sigmoid-plot-1.png
index fb272c26ed07ae70add47bb97a29a375cb7271b1..8ca5306bdae80255f6a7e4c6523a874342f74c5f 100644
GIT binary patch
literal 17921
zcmeIabyU>d7e0!CAc%knNJ}XtNDhr6AyU#Qh|)21mqAL2bPFRO-8Gc7(jy&1h``X@
zb<dz*^}TD|_50`E|L&~yzOKcGIiGXR-e>P;Kl?cZK2?y$!y(7Pz`($hc`TuffpK9U
z{U7#4@Dt`IrcB_Ut2U3d>@hHKZ=nCX@Ez%OjDc|*Lq_7Es!QSu()j_A2@TddE`=e>
z?MhPUc;#NX_MWAcex8+)2Bt;XyBwC!_bLgbsnjf529>fdN$BL!4G1nhASAxW^vL+)
zRhiWj`aHF_@78r!H-#sA^iAYelT&8e>?eht@6Na;U&6p}iwh6|vmhWL2mj3WqAp-C
z`AgwrJP_b<!@y`yMO?yodOd_1<CUH1HU>uIBN!ION6P>ISO47sMNI>PR4Evt`^`5U
z9UVmO*nVHc<H$oY3JaeuD`DqXqCc9;N&#0wlJEP9H;BVHn3xFu_ghk_PXk178nRSb
zqMk7BGD*<Sxauu)Vqm;_bsu*38UjyVlZgm{hue#~`pDrC5lOp*L(|2#5|DQ3OlPZ6
zN1hESbK}>_#<QqZM8?q9u#9UKbGM0(hcPg6iRus*s}gt%R2h`QM!I1nj=C$IWs+43
z-@IE_$8F2Dq9+@V1eFGtaB*?-^77>6<y)u&3O6@5DJdzNsePw3nN(nio5aK;qoeOc
z2n-hNC|Frpi?HafbEo2etUE;d=}tCfnm}AnpVEsZejh9F>_;5e5iuQEe|pNq#>V#9
zdK(iHv)}naXb6`h0&gcoFJ|l`0$0))YLKDK)Xzcnj)Pg<&e(X{Q*vr?k!6?QTB=kC
z`3w980!^mA#>O|k-le;{yD#t`%DsWO^<_jDyExE{ZyXMXiwG}O4##b^xGsDno!O=C
zPAP;mDxjJuDAOq73!mOeqaeYaKnX;|6V=rZ3ma=}7m0-?^{}yV7TIObAr{}>n{Rc<
zGGN|E;9dW-w-VLQ-wj1Zu4j<k^n+kve5X4Ou5*Y_NKG0|iM2bbR4mhAs}}G&-k+#-
z*LUNDKq^dnl2g16v$XTNySwwX%d-3VHn+BjaeNaJ5+qvbd;QAF%0fayP)!7G^dcg)
z#$EAg_>X$=OiWCwboEqK4ZptiIy>2`fFhMyq6P;CwetosC?q8%)z#Ix;yT}B?X0h3
zU%$>iLnHg};p>u;k_UJirDUE;qxwTz&vf{lc)eRF?Ck86m6i3}IN8~|J3HC<`D5o@
z)0NGPj2O#8?kXk^gU`r_jG^#FD?=kAwF13o+XmRR;qmd>)lOCje_E7ZV4#eGLg?GK
zZyT<|o!2H_wuK(-??*;PYHDh(3tPVmIX?E_=H{lLpm;>x|Lt3r*U6#L@k^yF)#zyP
z^bskif&>%_m7B|^udiQ(brEAWJSj=U@ay~GVND^S!{`@}N$KnwZ3@r^$#}iyIJ34!
zLr!kK&mQKE#E_{RLi=>z`#rdT%lgNlAlWc#k!$vq^PFA<6w~#=YWi9NHqDTsplu?$
zp#0Q8#F;&8i5i<t*S)dGmXWYcD(*`4a~9x>JUl#L^%!m6{c$sVM8!L%n(mH^qIP``
zzQ5cW74sI>VuX;w;lki4s!<&MrS{25F*Y{2T+E}&=TBTn99!kc_q)R5@hEuR2FNMQ
zr`|B5=WD<oJ6d8AR$cABw~E9RX}17Bf13-R=4I3-vhRDRFjcg8<qHaIYwgi$wXLmf
zzrHuCmfs<ly0P{KyY<b&MZY_2k!G*dAYNWxbbfq-nvRzk3<l6kRzayjs-4&Jm0*4W
zsUg?Z!Cq)OqL1`VhXHwGV`D=@Lrcrt2S*%2ts2PZQi`|EJv!qfA|xzKQByM(coq7g
z1mwYuUr%fhUEN7SO9MGWRrV&dPN(GrV8{TdDOm1BNyx&&!qQNInAb7?hDZ#48pZ9a
z=MJy(5q0PG?dJ!Zv&z_34u8KPE&&;<u-;o=M}1Gn9YLyJ{_~9p7l(-G=yJ2ZI}1L1
zf&tK13=$9?gH*X6h>(H;YZsTwekk87)Ro8L=kbYw0%?hamuZz4>*?uza(ayOcbw<v
z<G60k?eDwLxzpn!@ZS5N(Ay<*AHK1>8yXPsUfUG*_U}-{NXv<uDzzN-8`u7wszh3X
zKRyj~vhG<t*vCKPm;6BB^=!eT_((`dK1oS}fbe_C;6Yc0e8OnC)mZ*BA>~ImF8+P=
z2GR8v3*TvVoVlz6woMDa@PPo?>*LGCdIPyGyf#07ccWJd>(4~B!clivSh78jcCE(B
z8HE^VZ~ys12NiO>-uuj-#_PS;zl&hgcWB!3c<|s$Jg-G(SC{WCV%tAk*tsbQ2@ema
z5^}Ocr1=v3XJV9yu9Ks^z2y=5+5b#zk`InBFH^}>cvb;TD*D$TofRR_!91Oz5)(!6
znf7D2{Kr-1bR;3bosTx!XhMwAySDzC-6#wi%u|YZ=x~XF7M^+)*C<*9u2t*im?G*K
z53y$7e(&?wv}`XOB0hY$K}2M~Jj{9Nnhf^eg%nFbKo}bvb05f7%gD&^Bl~xi;2|L)
z;mP5S#b9on!GGpO0ZRa0*Vf$J+|oj3`B?nHpLx`YqSpxyk;PCx@5H}GI5<SjC0N&y
z@t8jC%aF$t_}2>hL`VikMk)b220p&QFaK`nmKq{NJqHgPTeIG)w#fe<NAA>+gy?&p
z?0kIqmYmPhbf5O>g}(+jcd%a?%!5gXeJ316{k8TEjG|0?WoCwmm{?N!Uvu||LDh3L
zIqiNv#>dAm`p;Tr5nWvgkeQhoGn{|Dxknn(BThj?G%9%Mk6UgXpp3x@1+Rtnetcc|
zpLJ+C*w3_v^kpj2h<a!)6JEXeXR=vA(0=o>f`YZtG7BTFxMBJme-E(=DU$6O++n=Z
zc4ujb)SX}auf?@BHHATTTOKKq4y7b?XS*o=#|QwEh>6uZE)62c|6P6k;bP8olfMe(
z-ya@w73fykO*dj{Ym5rx|DKXVOqP01rhEbk85wg-M+<e(6an&9eEI%D1SO;O{v`@9
zz4+c|b7_)66W(Vhivv0J8`D!ap9{k|AHe4A*)6Z#8q7=5n9Lf>H@U;Ms)wx2(3y2x
z^V*UY^+iq$3=A|jiUWWZdMNkE{|6U`ex1j3qc5{cMhzInp|GODM4*<i2C?NTZ5)c<
zv&^}Yb1>7%cg<|0A&zh(PVxYv&TqXNRGC3?xPHYS#G%sYEaj|=7cZuzr5SRKEYVz7
zz>|P9DxWzX%boNzW`>dT|BNUwagHe4U++`j&4@1Z_vCtf+IB#>IJXU=uDC^95XrrG
zVOP66?XpUU2xTI90Q9oxD+}Iv_jwLO3ju}GAsX*lq->o=qb+5MJtX~eu|4CW`YIhb
zcTq-qaEFEO>9P@b6$ejl0Kj)yU(alGYDN-v7`F9!2BoU2X|cGe*LLTcTW`U19}yPy
z$=VbKMJm3vjjyjSKoLA5@tLZwuCCXwUw;uMZnGT)5ap$d7oFB7PL?;BLZnmx@UXC8
zUH#&d(P5RFo7+qs!0)3k&0)V>w!FNO<&E^PR4Q=n%g(5~SMEuXFyUr9?*Ax3Y2QEp
zA3PW7_(d6^G4}=&SM=a#1c+#&?vkX)r;{khD!h^qz1FR?=ASjfv$~Ca<jB#1Qjdf%
zNoi>lBcp@ELmXaq47X=BF1Cm?Nv!LX1n#$=o4$B<KV2S4!Q(X%zN~Q;Ia2LxEhJK6
z(o^TK@Iea3%+9VLBNIN=3VeNSqL%gWzR$$hZ{NZj!M~OE^PxBW9vaefaIEzyiajea
z?vkxfSI$Zn^<?oS#DRw=Cgwh6lAvj|;5$7%vYsZ-*W^C5`*0d8N>AT}K!hKzy`&Lw
zlb9uq;z-Tv1j&lmdgAykzNpKBW7LzVgap}>y`qDI1Ii|d`2C;3$r_}@KWB+aRBHDL
za)71f;{pQ%E3796*ET8Dk%bKeA$oB)wg&t~$g9<hvkli!rY0|~^mA5v7bf*SS`&((
zF^ITy+>>;>)?hLa&+hFw^Hx58ccBsDc2JOlf6A7c-!s?eoI;XSMy@tJQ~yGIQq7tS
zBFZAkx<8GFM=FQ#3FVsaPFh-8(y4XZQTEn&i<@!l_Wox3haAo}iyq;+=QT%S9ZvFA
zEJNmqwTN{Vd_lJTdu`Qv#PIyM7t?I9Fs42i5uA3p$B!!ku6akys<(5+DUG7Y|Mr~*
zlkJRw#mdPO-qX6xmKa>?#K)?tIS(hJihxtUST-UdCdS0Vl9rOHn-d9uYTn)tp7^x;
zbJnX(sywSY&C!3(QFpoX;785&Pg!pUq2S5N0(?&m4m6s&<sPT+bzt=1!K;LXF^dUU
z1E$Kg`@Ne_j9SYky@kBFc}ohLo7=PErE~I;wdP~BCksB?F&Y;iONfg%f+_B<jNRRR
z@WA7WTja;|9Fr8zWY25%_m!sgAUvECT*Nq%PoMUlo*Yf?E8wj<Hem=qd4E3_o*K9j
zEnw1}Xx5jZmZ!ZFveukUK{7Qx<x^Dn>NV_zz!R-oW^n+&d+x7K4;Q`c*fh?C8(c8n
z_CaH`;37F+@r~9fzMdxVt&a#IKM)pLHRp1HgLK9!MW1IOTmmFPYopeElhU}>F)lb)
zPoh8sfWL0rsABKS;Nbo)+k$iCs&qx|=Q;xd0xAw~g1Y2?<EhQfr+8D}yt!T8XD<O^
zd*JZs7OqJ{BB3$)>`MuwWMNk>4i4-7&q~%u55eOv!O8al-r|hrnzcJf6tHh>YU<d~
zLVxd9c^I7DIW;vE$87?ytc;u~0Tz0N^1Dyc{YC)R&co8k)GloM!$+5$M!?SSKr|HP
zhZ*~8S>HKS*kk;Cd}>ynB+S!-Js=k`00yT!;WFurCA@i4waRYRZVhC@KkH?PlWLKv
z=>)~>j~_n>>7`T+@1oyE_Un7ITw7aPkHp2>E56UUqJRDiz&@_(B;z*jjOWdvJI87y
zx8bzGz9r8UHsh`^f4H}`zcC}+gHIz`^P^fERJNVJ%3QPPxR@A-^XhmMtHzBHiocup
zoz4Dmdtqz7$Jx$~t{C#q_W5w*6*FkgfwUwqCwB#pT*BP{pV^ghZEbFnaOl;%Xu)}N
zgs!1^+!_JWlQ1{(+8r^u{D^|;pRpLY@n|1scgDuvl9laveF^IW@*bGNZ%F<)vvYiW
zyrAxog^ewEKba6&hym()V4`q^xYw^=<C3yFI5;rWnfx<eGZ)_U<4>KB{I9EJtD1dz
zbwPZY@~@|SdTI=XLNzsqH)lIAN5%dwR+*`<r-y@qp}D(T;o}KCuu&RCqb5J@97BW|
zw){4i_vwL&v~(Ld1GvuEzdRzhu&{S@bhN);b@??mD+fnUL&HOmA{EBDKp`1(Pltqp
zf)%C2LoTnR6b`EXkJ5xm>=i}D#W&n<|3Et;NXP*?#?PNmfmGy+i-swM7~F2ZUi#kb
zmS2DowXjR!lPDGthHe}__-p3?HDge}nwpyG>Ly+Y^#y(w0L4Z#7DesbdwckJc)W63
zlD`c662zlN4)mW;C{s{NQ;WFWnzQ*k(5<>oke$}oZJN%p?YTFj{of<5uC9(%J@>xh
z*nh_fMpioyffEYis-G~^-`(?L0~IVc7nd^X96uX0u#(quu^zOpV3A?o*;TXPwN$Qu
zmu?_!)O-^Mq)^PAkh+f_U-?Q|f8{zkK8|L8W^QF=726~N3xWbN13rEwL@yK7@NI1{
zf24F?H9OG##LqGDpwqo#O>}N?H~v0vulI`za?^e5Mi3kTo+y1*?KF`8tXii6>L)M`
zAa~)EuT=fc%tle@$wUfG$=ANF!}<cVvUjDdwqjnS!l(YJtm^ecuQsd9F#!}0G-&rY
z-rs;bF8!F9@fCfV^=kkzhULs}YTHeeL<#13{+Q*|b4gKJK5By2;MCvoW9I%<+WFz1
zn{LXtbw#-)SS3kYh~v<h_H0?wQA+&NWU&izn-+gH|JQ4x;`B3%i{=Ifw7X1~!JH#O
zDbk;#(es7XD!EO;p~=n-26+U@QG?kar^+9RUq{|kHetv;t5jLq!yo3eel;s#^!BbC
zLITCr*VhM2OCw>^yp2-Q!&i)b@mdcC$>J719tuLdr0!aHTA&^=Jt^dA`(v@F(Ak2%
zgqcG3z9dm<$2IW5$0ubiMS%3c#+2@DQQR$EFlm$aY0F=wjaN<#ew4|bwS0Qp!vY<i
zOyfYkVBP!i;nD7juaJb+ZMeZ}Jv!LCE$g+dsgo#MQh)bh8}c(TzMjpciRu1TTv@?{
zAkGIb->N})2=hBjOG`s2An(dRjM8*$x>ze^J!t8+Y!r5-9B%zgKBqJVE?}V9h$)bk
zP~4DB5s$5r0cvjv37?yXpk4<CJgv1K2(NA#AU_?OJu{jc%zEh=@#a&C#+i+zRWE{!
z7I-S*=gxaN<t}b+ZmzCZi?%AzXMv*=byNx$aR2%t_fW*$mDdV|D14aI?lD&@1cKZt
zE9YrN{~{3y2SsoT<lK_tk#W({yDJrwPiWD_S<@R*@(TAbvauhT6GtnVB3SqZlN<8p
z#;Kj%Wlj;hB3C?0IX~&#<C4A9j~S=vDoyHcFBtQa9(WOfk(;}_EW`OA7s-0Iti3dR
zKhsbpuhlm|Clphv+j>%#Pj&)%Lgdt|kUxnN>f?sD?|HjO{DE!Usu%`V6Ey<7yp905
zy}-XbyxAChis)K$9sNFG^y6uXE#8rxe8WMoNXkJHw9ZUJrG<7)5uZ1p5^k+|DJZx3
zYgpUa+EVoVoSQP=jwFZ(#YUnY`!>LS->?4mQmJ~k&1QbaHfHYC@h2K;c6mz3hY=f7
z26diW7q|y$_S0mmqBXMD^y&sIzx#>u-h-}GKC^G!O~qr0vq`kTnv83rYwDg~PwsY*
z<VPd%Lb>F5PfyRLrl$1WXCfdZ)D$inA5_W4uv6;~Bg#6`_TA&cRG^F^UtL^`Wpq?x
zT0`28?&WGVIVC&eugbFXWewyTovzhqEDo;R)C(1XW$X1t=d<?!jPMLQy2^H@6;KJ~
z7DJ)^*JXg4kf8MMM-{I5Wa=9hEW%SfjM<Yi4l1DJNgv5ip;b^RjvOwxmf}h2UCYWM
zqfG6=`Aeg&r@eWB1C!o*%J|yug%D3MJwiY`sK((_3lCS>XZDMq=c8VW?TmbA)Qw##
z84BUFc;+_S#5*S_#d@8pqsV&OxIxOgIbHm-zKweQ>dDp|3FFzPWzZnvTEPM5k40qY
zvLc+zHF<AqD+n+vb6M12BT)KeW6;sfhnMd1SrlhGx51-+?hz^KmAkD~!>m}#UrqF9
zDfJ*i+)no{awy91pO#8hIPo&On-T+rLYbCOK}bwIkow3UEMMJ2L;U$&IH5cCi|_R%
zuLH-X2Ij3(^l0wq+9&K@%FGvWoNpX;M#@ZHc$ff(m>$>Aj}+*QRROZm$;}wxuZ8c7
zpeO;^;=>2JhLB9f6u_=(oYRyM5%j)*uCxV!7iV>T6<IwAio;6VnNLp@vM5-h78m+5
z0iTKr`T5h(SDg0nzFx`3-lR9tYH^RN0adMX>U%Bi&Gr{NvL_nG#zrrheea7OE}=jo
zproQocm1Hy3#w|(*nSPPllWh!0%eynpiADs=w#ECXXoaW#j<*OdggB>57oc_Dwj3V
zZ`-zcd?r6~_T4iuPiom^8@c|@kF5s&`Dy+Kt<h0!nQ4E&`=8J7VNK#O^DMe(7OyRo
zDt&xbKmuavl8RXCtn0C*j1h=l=8R<KWF^TRcHQ={QxEr>H=J|g6?PBZ#JhU%If=-z
zgS7~7hs4ZVU?bdjm&4}(E;9pk;)Q~*6G1fkNB8F1P&D{SH6Qd;i`*H$dY^c#n_PAD
z<V&eLDt(+c@EM2ZjF{*!->su^Ly0+E&)=ed-G@0ovRW=reQ9N-7K#+Kx6cx5eFy?J
z=L2s}ycBoi1yoV0w?OH7u9eO)YB6d#qh$A#Co|~f!JSD)jhCTA$*vv^E_>dh*5)cz
z>Oo#9J{qpQ1`XB%6Fb?xyjdpad3@^*NW3c}C7I2?FjQ{>!UCF`RNf-6y?H`m@*|W|
z$aCd1DoU+FLEjlx6QSB??l``7O75+H@T0>QqQJnCJ@8pVBW8Zx#s(f)&<V<ulkGmm
zhK7dbHu}qW)YrQ)YRPH^1r}t6_Sy!|AnnDuLT)eiIGpe1R2k2{OOiKN)LV77G)#cA
zYCD_iwY797a&*fo)@@4{H|>WU9rlvOPJ>z`5(f4`$Z=5>23SuXbn$gaD(A}MC{|`A
zt(+1@F=9j6MvR;&+ZoEXdeh<EpqkFtcb&1;d1Y_1ds43Jyn(PNV0;K))hk;u^$@20
zbs#iAJ?1!B=a~SwvSt^+mHA;FRy;8MQ88-Ws9B(G5E4I7sw7+!{%JFwbY<w%3vN|e
zjULgL!bhs6G_4mvExB24l=0~k;J+w^oa8EL3H{X|Y`rO;JC=nmpV^rxfA<Zl;EsxB
zr?wj9b;>wiq(Re@Kl9rA^GdbK%%f~iRsbw~$Q*c$Ls5VzCiX+Y57*ER`QupyIuq^8
zd3O=>zSvTNwsSy|!Gk8@j`pkp31>4y2sls@6tA0AJ3Q5j)W~sh44yq(!pJ$N*VPI`
z#9ReHFtWDu)2C0dv9Xnv{HgvDfKLTqIQ&q5S_v_niK1#+X}%S@0U%xkS9I2Z`N_%2
zQ2sNEVDB}6<vv1WA=Zeh1(=vMmn{-TaPEmkodxx$ZXk8bC-4!w38SNjfj})H0g=$}
zwkI<0Bto%lfvkz|su#P$=GG~6zoOcf4VQ!rJ(G0wOu)HqZEYnte%uVomDgSl_JWqH
z2TmpQ9)eY;F^B|pD>>I0x+{R>p3uzr!}M*OhJYP<?FIAR)a)Cw9`t#=-G=|>`7hTk
z0Odv9R<{ry1s_kRi_aGY2l`!t=1(yQoxNE8b+`Srw=p!I#9Z&tD{77UKcrc8d=xPh
z;|lmd&|i7=3gc~vG&bmF;BFglXE&$d-_>V3t(^3m*DEF8wc4CZjTAwu-2&-oGoEX}
zyeurNr!|Bk<YbK;Ab&lV$dB7HXG(9r1f<KxjE;<u9XU|nQfilcs+~9Is-4Gq!azU;
z7G9Ix2|O#A-?kCNq%VJrW)$qa(#|($+!Grg7q6uO?QQbjqjE~=QFd}^yea?Wo(k!x
zgrsC#Y_ClbfAam7i?!ZaB*?aqP1Bukb1D@nKfi5puO8P8Soiiyib)>jG+ly-d|X3#
ze>y)X+Qz1)C{P+x@R&Z|-@tyI1U!<X24S(ynCW~n?&8!O38l{8<#S@kgUe}8Tb-?}
zb~<Nld%8_O3B2`^hXpJ+O@D46M*{?HXBrwBrB-^tL$iTYc4xqR_{<eG4D;4;lhGbe
z6=T!Q{z~&HzOwSWO*RV~fKIMia%I|7w{BH}U=<gfZ1S8AnC_G)LmSg7mdjQ=Xw*J|
zFAu7g=(MG-v@~TEMD0CNFa`6}tHf<gD6}EzEFwp;X%!nb`dmemoEGOUhE7l{vkmmu
zsovl-dOGYamVNS9EDNR9ON!Sh`>CqueZA<Dm|H&o_>$k9o*38J_9*-LZYn^nj)TMK
zeh%6TRGAE;qic#ayL8~Z(D}jG8U9d}j#+YURtjBqwL2h~-05jo%^3h}^jMvzkUS`K
z*<)Wc;D&fT>B<Ndik?5SZa<BwD=j21nn5@>Ar?DcfeJXZ7G=Egqh+mjPY)1=%~W9o
zze@z}BhfZn-8oGqzY9#HR{zrih+}~bLPWU&S{RhS9GZm>mZz}+_lx^PM~c4Q)8VK{
zh+M(*Y-)VZKjfHydzikg0n0cS-Hrl%Se7W3+qZACu)M1S6h;m>pF^mFHX*U)mHK(7
z)H_@LWGTy#6_Jlmatqd+%_>#1Eo~ZZ%A5CpE^^uUAro;&iEI>ANQ5j*oB&A2aih5f
znu@rP7o>RVk)ih7bgJLCuNo=rj%j~b3@|*8fal?M3aCvH(lmi01Q&o)p#x3aTbP=Z
zwz^Dfsz@gJHPp<VKGCX}?6PLkA|ucQB7c(3#l&RV<cC)ZdS>VdQV}{)({%TB^rca_
z|IM+wYR_jYLP||X9Nlp)h4d$8m&w=?G3_lSAfVpf-r331D#;kWL`Wip2pN7v5OIpB
zVl40R;)9gP<39MHV-7>d0<XzgI*FC}uuaYJ29cE-L@53)<nr?YJSuiTi*L<!0g4CA
zwgEsq+$XiJA%SHU$U|t99KXU`|L2zp<b0HEwv6+Q(Je*piz?~?P^ORP^y*-))+6V*
z^9hT={BaCGT!D$e<HWpAGgG{tIEVCr`s4ysTIQBI@>Vs7@}L;>+@JQR9550E70E9S
zCq#*ro&7ov&b#2?{*w=bTq&RsLs!q}IvECmU%h%2G{^GQ^Ws-yN}~^i!D0Hv3AHYb
zSJ{K|A&%enM~TK~wk^g#RWyCNDvQ|J@fJ7+VrO)oMvIToB}7_+G^AiY%1x1i$Ui>B
zV*B9En<lzL$oZr>`(n9GLc_ws09Cie@m(WB6ubw(mx5$rPK9r_?ZHA7-_@dYEmWO*
z4TW7|ttUOc1n*$gp<aaYof%g?Z@<OJz%UNFn@a`tUJH)x?QF#7M*z{aX&=M>Ot`KP
zW)UN~vByP|IsU}${l<Fdhw=>e5Q=G?vEqjgFCA{4vw5=M<j5P<ftDc&%4j>=$Mdf6
z$s2dIp_8B7_6s~+^*akESC>x-YpKRFhEW~|voR^2yIP>*wBsF=!KLq;gz}r+mMTCb
zz~O37hMgH!CS|gI<K=24YDogH<E^ux&5@Or1v)6KJI-0mTfE?eV7dhcYgL<dVBsYW
z=YGqcpjuM@FsUf)I<i>17FsUppu*?EUrDIMbQ6G*qrEln0SXk;)(*fcBwt;Uy{9ol
z=;pIV6J_Ov9XAxT>#}5F3ZCwK&U~b1E*wFjq22B6n>#zi?zbPj{qp4t6bTO@<Ca=}
zje#v|ou;>G@ibuv$#EuZrn)8i;@a0?gMpADe)~B&b@joeYGF*HIzmu)KHz{=EJ94Q
z20Zn1!|Ia{8CZ{K6md76S(1TYS9;2HliQ>_I3U2~WGuM0v-k_TB9Vl&zl~Uu6I}lu
zL!#bPft(bSyBJ0!;c!V}|L6#G745fC2vE#@A=pl%=rGXfvoC+gc~kH<JiHEcAm)Dl
zj7;yKaEwDAO-!T8p^&-VeAfi_9A6q-VglsjL|yh2ZW9F;7Z<Gw%NnC-kg7;f1@Zl&
z!uMZ(PdtIxtDYGOGWuddUvF&TfuFTBmfcO{)R*A_bSNMzxlVb>tmBN@x-nJI0uMKC
zx1E&mtsfNrt*3|blkG_Hp5ueD@~me^hjQlrj@w3n?EbXe97qHRd6-lP%gj+oX=lY7
z+zfOW<2A$I8of^5`*~XE5_#qXNA9T27v1{yv9udcnr|f$iaXG2*;NTMFc|JE4j7#U
za7G`1E_vaDGZO5wCABI7`{mb-aG8h?lM3TQ>ymb^US72NqnQu!R~_lK0BwnlgOd%~
zS|egFf{p6#0m3}Gj{?mZG&(N`RlKRG_~fdPXh=SYfQ4sH>c4Gu{Lo%jr0kE9iU*uc
z)RWl&Li&kC6)6p40W?kFwGI8jXL?ZJ%pW>a`HFrB>seJ!?qE$^HE-1;a~jH^L)d5}
zXYvir7)5IE`iEe(fIXiDgy!V#+g3H>gDE`-8`Fu9dS3BujeLOf?6XzEW^O9t%Cn9|
zbB9gM;ukzqWIg0KR6472T}U<#F=t+;S}ZqTeln7E5lt``8IeDH_)y<tqeaJ;PY^vK
zi0J_$mpK<8hiSMS{xuYKQcw$%gXKVF^`<)c3E@%?Zky9@;T%_vs9Bxz^qU?+y3~Zf
zbqXvAnT4sfy|27Sq`Tk-1qb~@JP?p%4J?AW#>|vRsTx=P4*CNN^5IX;nwL3r)amRm
zN4TX>S3dQezDqklc5nFp84@1Nc=seBck#W;)Fo*ytoI}mkPjbXVk(|Dys-o9U#qgQ
zBzof}Vv4xzak0Wk@UA+#QU#5^jiD4cPH(D!xw(7qX^0Nsx)d%-DH{XY`o@hPKYyy$
zg;B`H5jULkS6e`B<q@BhH2L!cpR0a1Fu2<3jPqNyM!fn8r-k=%cC>ATyJA>Zdd1Mo
zfe;7Kh5lw({P69SD0h4*C1Xr<HqZ-zW|Ucatfz>*Yu@v6UCqEE{)S4Sw>O_JnHkda
z_udb&uvS4N$pDH!@M}j0?$xWz&ZaQ>6E0SuFv8V~MOb^qYQg;z{IxeX+rn_MJDXW~
zy-eEsoyq5tb%ng`e^|Uo1Rt>-0C7?IKY;veZ%fOZf}NdRJ5L>55G%o?@R+a7@KOML
zfkuO6WsZ`&Gt0|t;dVY_g@uJSwzh<xx5TZ1c4v5OEJ!qX$dW1QiNbxOO7~f@C*8jH
zg6N&Da_S#5EG!~(hsHIhd&JJhKi?G<73JksgCRbvW`pYci48pj0ud6@`CEe5n7;&V
zHy}&_`8;dNi(1T!&DZWS@@bZ8RaMn_1s)uH?RN!Ei#p-krWW3OlD)scy(O*@a&w34
z?t+*L!@un|h7k@HK*L|TcFSfPr6k?<j(z!qUEC)Wxya6B+uZ!T{k4J&NWVOD*=DKE
z5H#JGgvm%fbd@^!-lAH@@{=s@r)M-$((Nl8fa7d<gGs!$nONGv;0y36I*ME|-Cp$P
zv=_q(kW+<2KChUY<lk$TSFF;|v~Nzkx0F#{4qLKg*1gYI4rqM}iUdIG-xx_S>|l6&
z-b-_V&8IA%vQ6pETU)no=by6eAV|hf&E|$>WVQ9NS00QsQ|CY>A`=oQ$jRM)ekD{E
ziZqThJg?7<&bZ1G`@h9l^PYUE*Vvga@g%DDqd$?-Mph<EBj31zchY%#p8?&=$jC_e
zXev(!L(s2OwGKkbC;O#@CDPTZ_?^jjRCTDe5{j5rM7h<>V;ws%F$sHAFRi=#(S=J_
z2*xuncQ7cRXZzI6L1p{?^w)5o1TvOtbLu>Iyk+f*55Xm<E&0oPyPjI=t`g1Vk*~~7
z%oWttoScie=YzJAg_*NQS7sQijmy&3F8BsOC4g$H7BroKYOA<9N;?jSFPbi%%tnp&
zqrwh7-TKRqJ}T^322nFm3xxNXG;OTRSZG*o7N{9cD9ZdQh_lrT^c;`ujJCe(>s?KA
z0Z{giV)A%_z8DaN-IjDyS^Qw1Gk1cxdw2*xRWk$#>bQ?6#c@w7>HgvpJMrOYe{*ZP
zo_&lP@rbVy$XZQ8F6+;uSk!z_n;%$IZl4phW^UhL{l|PK)oKTjPvL5-<Ml~={y)av
zhx_`!rO-TYIiQM-I(&w!9NKS&UctvQd8G%E<+{R=Phl0->dke^whiBJV*YIPgWeu3
z8C%x?Uv~-j+BMyB%eV(b$_tUql%N3!$ZCzrDqDS9Mh^*Cd%2o!&afaaWjWWzY8`WO
znfYMu$n(WiI?gL*a)pbn@2^t}qkjDO4s<rJcR9kjFn!TAdk5+syj#@HKzrd~NGBYp
z!QZFvY7X?86zM1kDp#GcnIoBb&~G|Dl*(xz$hL0@D`3=n?51qF@B@+m64%*cF}fPp
z58W9fkMm1O(FcOBk5o@^jBrAdP*TgZ0W$T7ttyk~P-=?9m`Auhqvb7O`Fh`=f);Cv
z_V8Y3tBBiRYWKYWxCA;1z_LB-g$3wkTmT3ccbmL+u65K*C-D)-<k`nMVn8EqPV7Y-
z$?HRer#p-z_t?3BF5`i*@m#iAPAH8S`}S)B<kPnsD5<lK;h&%GKAWV`Ihz#2?|{SZ
zI`z~<<?^k62_^=J_7%{Yv`4YBs?i-{AAP?KPrag!tohmr%R0)k68doZOjF0feP~cW
zhi~WOnlWZ0Fmb9UP(GGf46S}o4+3o>HAw%L>;%#{z73rfh(WGKUZGT%`6v_h#A9XG
zuAp5Dao3v<ucc8uG%_~M*RN0B7qQaAMBwq>3WMrxFg;XL{|`M9EqC*^JMuoXGnNyL
z3jrI^LXG{3(^Zo0<@ESx`!gv&DjX`1Dfj{Bo~u)lPg^alj(mD4=UmWsQASZwA1H$^
zsGrMM(60cJ+m{9xF5PXli+uI!)oe$Mjt<54s3qtzU+IvA@?R1*c*m>hNbj>UR^jgA
za?2R#ssNZIM+^RF;k4}c?pXnpxv{bFtNZix5)j2Sqx;6ga`gA_-xn6vjXX;WfPQ}c
zOYb%aq+1lc=KXb^?oE`d0LuUG!BcJR(YdY!$#b&&?<k+;n#RV2=;*|o&l3Osa(l(3
z*I2QUEC&ZiD*iSQp!{!?W<a9W8g-Bmp0jlQXq{CNsF<%NngN*`pa;zdSP*AAfHS^w
zE<%!T0i1soAZk`d%ZOG9&P%jkYA@o*_;~osw(FNB`{vYuoBVHCcQrE_wrYL}0wt2a
zgRTd7J6bA8Tx8I6g@dC2aK<=ATXvw!2xK)t!kA#^zCP6em|Kp+^A7a+n*-4Y;Fa9l
zc~_u23JAcqx3@2L&j-!T^BL2^nNz-g{TfOo1k{Lr%TQKq;0(XCZ0*x%+H`L<>g#_8
z3pLG19Wbx0tpzQ@m&SLhyhHx%$yPa9EeDgv2lTJ95CeLwUshLPy2-|K_~y+;x+8=y
zAHdcO#gHG5qnM50;3R@(YP#~mcM3k4Rl|{C^d1tn^Ejl+8ybH4^xJoS&j{KT&s*!h
z2c4*KZE<EK_`3!e5ZVF>VOK2Y+pjK{-~8Sw(6X<>F(Lnc-nu*ARv%Obn|hx1K!JWe
z&=7}{hdubc-#)0h5Fmf_1i2ybT*-EBWwg#C=K*@Tpno|CxVW37_@D*!%b;CLo?Bgz
zOIX(}M@yMUHn4$G;rGtz(K9nId<~}u=z?C9`Zf9#0~bIWB40PoPd#U6Z%@C(I7I2Z
z9sApzwpds-@+m<l_?H0j_fJ)_L3s;0h-|lSKlu3f73f0UI?l|rY!uW#pFcCvpUdBV
zuTW|dC_S#=Q!h_X(~ke$3<(HFmFt!<Xb%7jfb#m!0$J@dQq^-na|dEIYWU(sqzcZT
zeRb6X0O+$4jGV{x_V)jV^2w~Ucd>fh+sUXCI?;(Z-HM+Z#K#>>`8JPd>+wk3VXG5p
z^3YV3n3&k5VCmc7RShl<^d2w`{WG`cr1h7`)aSg{5)dUp`f^tQk)K)R?a`;Gv5@e%
z5)KE-Ika}amb33^&H%tuK;Ibt&CrkR1E6xAHcd<<gK~6Ts-UWD&Hp$I<{hYA%(K~>
zMOCYab(!jG8kG8sMgk(3nwkpG^KXeGunD%Tj^<O<^!Nk;5T`q<&Wd>-oh5S5W7p<U
zBk<a6H45}Z?B{|(rQXm=|I4Z9+3b5xk!uC9aHhlUuHYA;B}-NO=8u2$7wMn6tYqzW
zrSORc0BN8j=<5S%V2B{P8;CxI7jybe4qh|L7kJE274ph-*>WANrnZ&u$R5lk&{y#&
zED<6-b_O(v?#O-CudkzRJAWTqp7%x2EiP}qIqK+czQ^vHES;1QG2Pfa^{>3@QeyM&
zapI`}mfnY3oJ^_%2|G}kpDt<S>&nQ<3481pPWv<XXhg{nJ#%;o^k};eeA55u(VDfT
z3u-G?B53uGH@62O4uO8zy5Amj!MC>^lII7^9f4;3GK57VpDm_ClE$x>0?>>@XL&%5
z0s=JftPm;sL+n6H)ofK6oF)Q?rt$F_yV-U$RK*LCdTC;E7u_tPMUjfX0DP1uEcp?<
zKiyi2Bc`KFo2A~<<zZG_>jy?zsW9qiIqLVC!lZ8lRX)I$Hsm2v>@h*W(ovC-g_Ovz
zE)%gIHvdt;Z$2R-BMS`;?f0fRLLp!=b)RcI4mN>U2h?w*ICXD5rRwDl0nQ#D*Ild+
z>q57e>ImN$tUAKJRogE<*8LIE-+3{;vmJqi$@e@w8H4MP|D5uZJn_Dj6f>IgBpnAE
zd$`PkRQ_2q&?UXQKla1XCk(nju0kt9?z}VUaj2Kt**ej!9N{T)$I>!~WX(H9mk!);
z2<=jX{E{Scp*hK4$<8Oe_Gl%>J;CzSqUm|ey!YqN%6y@2CR)p})>t~YA{U!p*bH<5
zQX0@=r7jyUZ2cK7qch%24%EkbdxT72K&Slk=J%|(l6iy4gZ?G&T)N}4?LFtxDAJY5
z0CNlCUHf_SvLE;fP+umLQGzB{KfjpOx({{oAE{VSk&gIYf6_q$y<-<}7NA*)KP2<i
z(@TbCSKCZC0ug8LC_ze;4ClFUQQ{;RTpzhF1S1uNRhZXLeO<-RRrEco8Ozc-5p<*|
zU*t8|4UL@Br$Qb)#5;ug_w?67+4X8%K>h{nWuORUDyW_NpC9W%H1)2|eo|#Q0ve(w
ze6e1^B&?5DJXOrYi%J&n0Fq#_Kg+?{_4j(w`JVwU;JDZiGQ8OS)Fp8H1+diUg*7xa
zps~Rtp#KMCd6?KK)9!=I(t8K*&?<p*!At-Y<ZOu9cLrx?XYf7#-%(152AL@2%q}3{
z3i==~@SD+r0-Zr6_OuF>vyP9qdSEaZa(uj*ItXaXe?{U|lyP^W08qaGl@OaDIq1W@
zLkp3aLZLTq+yL!Sh}W@8Xo%d~moPcB2neWn@=O|VeCYt6dUkpO*Dro=W@8x{2$&oo
z{gSe901554JWg#-mV^rr+|c<gVq$?M0QDuhJHI?!)T3uW+c!0Qe)|YW-lQhA3dK{w
z#S||RKnDM3NPUN`7E*n=6wwncQ_u#@Xed&;0~jHocazd`*c!902k<SFjC(G9Nb4&P
zKi~NkCJBfS!hr;wW3(1;ed^)K<cmLV0}8s}4Ak6Kk5~%lOTVC?pi}FjKQA=h_E922
zCjL^%0TL_ylpQzj3b+=6CO?7lO|bjTFQJ^*>A}24?Q7rvZ48<;6--O`C6<c@dgi~2
zK#xg)Y;J2KB_Ro=6xifl`_IyimrWY%syt4Q4+;PX)BLjrAXekvMHvG@8sJqxE6M9*
z?C(RMFAElg2Zn`lFf)^JzG$YH{Pt&OfCjG9rRNP0%z+-aJ5VAP*Tns2o--j(Kxc4T
zPYB<=oB8iUAo}Zk3`+p5F!Qoe=Ly%$;5Cy!`v$CJ9Bk46M8LQtZ2UliJ|fom=bb>k
z7RxCG0b1G=F>fz200aMHdZ2Pb2P+koKEQs|4rKma2(T2{kd}>}-o%fVAaGabu$;#5
z&ntncIxEu>Kw`(lCE*totXskQYh&~lFu{RT76{D9`E8yHh<5y$4tfhTyWaqO_vq21
z4j0!*(ZACHg0o;o_*HD|SctV=)!9zTzs4t%bU@d{Iyz&ZCXRH+N1q4^BhV;9i{n0d
zlO+DHs2zT{?DL;ZU66+S(&v?yl;nI>_`5<dPX=lv`tT_zD1Z)W!IanEhhX_K2W<R8
zLhd8QM#SJA%HK!8{t<Q#;C55J<w!9O=v(}q3lLL&2!Z}B6zSYw{@F8NraN6FQ$Xd`
z+uI9ztI9(E*-yE6L`Z$TH>m&e^6q6({XM<FOTk5%z|c@9d;6t{TA>Edzt2FsAYxR@
z2GqkYF8mF>QvUy0d6p#lHj?50-R3fqzwaYm8Tt1cAcDj?biV=OIeL2fKwaweu>TCU
zAOQhT$H2e<h|a~A6aPE|t|>={=mLtj#lgHWa3ksUZqL8wmcub=0C$qWeV4W{8q!sQ
zzncZrw=%S7dlRsiBL1GdYg{l5QNdb^zI<39M41~ml8aT^IgGDX%8SQHlN&3J_PX^`
z9N$Lp0dmLO3m$t&XFPHq)M#1GM`>mzCL0?Yf^7n2a1|4d^4q_tn}wL=1N-v=UxE_o
zuDg;xPuDg@zRe6iniTryY9u>V>B`W(dfz?AwA`T}cap^9<T`^UKX79y6~Fn<a=;bC
z0gd{x_e<{k+sk))Xv}~5tGFGNg$c$_I$t!}N(UDcEG#UFjJJWqO;1l>i7o#*JF6r%
zFgyJFR;7{{Y-pCRvm1rkgEE>UF=p`<Kl$F_Ayf2$n;W*qdXZ2AX*pKj)zkz7g~Ann
z65I|Y5IE6VW^7Ry#7fZ2PkOI`svwMNq^Tq{C`mv+F(%oA6S_8ieAS&ETlnIm7tMmW
z3bL|erDhQqKtBF+xE`S8+jBl^9DHFT#GcDxFXy>&W#w`cvt3IMpLbMoJ~P+E@p5=e
z#0m$MYhXaY{oy#~Kk@bXdo{=%h2%#kBK~#j<OR%1S!E>szL2zBI5-<YnVFd}F)<CH
zUmNXif}8@>ESPA4e-+ti)=409Mc9$5drNV`Ok3C*vU)F_R7F`ip3f>9VG4k$lamu@
z<LK7|-SH>otWxbf@W1l%49A(nouz806_z6!*%rt+5*hU*-5`|l2e`f%NPDK*G74<`
zSXP$E^VS>aO%f7NE2*`gnd0K$fFOWrUdfd_k2$)tM4W!&C^A`Q^ipF>qlWhG-Mfs8
zv+1O2ENV%Dj$lmJWDS-m;1~sym2NxP=>B^h5M%(a!k_@&)!Z!M?C_SV6!dNQ`1q`k
z`)*gJs<N!Ctl-(R5kiVfN|KV202KX5F*i2{ejOk@x*0-KdNHxd@92**-3($9FsA4C
zCewJ`w2cf6L6|%*90&}6Xgt7|gGGx;<(Ks6%+u1#OSElWqxg#1Hrq2YT#VbizkHPz
z&ic~5)Md4OI7ukqDPyBu)0QKa>lR5g44v8m|MVrQ$R`loplo_>Ab<n1A(}jEpg8f?
zXsxqhGA{^D7hUR46|EWs_x5ywB0!f+_Sf}7bfTVW*F<Zf!L2tm4(MP7H^8O!AG32Y
z3mdd!Xu53Q0OJ8+0s0Pd#Mh<<-`2(xcLCsyR`0aH<uGtF{I{f>?SCL+cfx%wJtIR~
zM<)ne9cE-?1iZ#$P&tSCas|D6$Hm2E4(>nz<R29kwKMHcqmUx{Z25J-qujBZDjg#u
zuDi=4$;rv!plSk@@w=cPB5<+#WojqT8Ci{%nt?_R`q5aq6}V~r0^C*lf4Zs;uyUgu
z`hz;#Ff(p#uBRkvH~gv3wzEg(A(9Kb3$)UO%IQzI$c*s2-LycVl0@YRiSr`EHQE#g
zH33<bJL3+%y%+KMJ$Qt7`xT5Y{Qsvr@u>QBMYi^Fx%7}Ak-Q<3o5(Y($<$^Ztunq#
z(%WWFWtOMkXLi@vZTDs!8RH#YTN!E&AG{jQauW6yO982(W;Ws6{KdFGI;)9v*3e=(
z(MrqG&Ec%;om4^{7Qz!B0|PSw{dWPJ?K;?l*dqhL=6+`P-&tBm3@#mPM$%jobWXIT
z3oFi9Ee-Qd9QS%WT(vE2tG~NI?m(APeCSZhJQ->Z;iBWLW)guy*Z3m5<c8jfFkHxr
zfJyO?6|?)E&U-|^!}`I4Umo^Z?OVpfUb3j+M}$6`m5N=)B{zaxK@sMx?qElG_^e!g
z`C<5{x>GfwiVgP^b634B(<Y4hQ;O4!Y|*3-9kll>Jhx><d$wwJzm+?)Iv*V7OD#XV
zDtRz8RQoK?O3>#{RtxnzCIBB4l4=mnNjZ~0uk=qy$XN9m`kZ}Mj|N<EM)#_kQH!K%
z5`5itz6WX6S04ilo{<|r+O}|+U6za*etYAUOj6xzBB2pG%C>LI$Z^X^$rjb*DjVI6
zmRT<>K{*-~-%XvTwd5cfXD{V=l=EuWm8KBYo@H1HV&-n{7EkdjBsnn{&e9~cU3W;$
z(zbGvicSy8$qSz0?`|F$sA8fod0$6jIw#<2^pxW+S({l(GM>(+Y=}}5xw3!&vv4)j
zZLcr!BP*7shZ=`Yn#X!?*hG(%<(5M{@;Ymvdk8D8Mh~w|oUkPW@4I@-uL)a7CxV3>
zI8{30)R*|vG(2TPXY4twl2*9)3TyZBpV17YSu}px8sDm*nHA8JwQ{?sI9lBcaWO%B
z{VdYZ#cYY?F%ekRcY3RA#YVQH??srTgu@9I@g@Dlq<C}(DRJ}D`nT2h0ht43MXaRk
z6E=%7PL0v&2?{3PIf>c6c{)G@1OyJC+I4<2krm~^)?$`9>1PO5=|^tLyh-z5$t_(~
z{H(GK_LOaAC0r@mRh6Sk>hx`Z^N>N#BddCfa}n<BDL6FRL5+MpW~-L_cBC32CTqpM
z`kv0TYf8tX=|UXsE!0udh<#@Z%{v1}hh0lKW9&7@^{x>rykJl(HaWw@vb-`(DzPU<
z3Q3~zV+$$zrXSQjYBYCRz7f$3Ku8@`FNTk}J3{&An-?0-@XBVm6_aX^i^%eu>=Cnl
z&ijh8ZfEOjDHVm_p5hx$+r>rhHXfA@+r-vUU1M{DmljqEGx#fzFDEG9)H=%V&>FhU
z(f01`GI9CMc4%IC<-s+J?H-<n-080zcia)B(aGM&eP{5_cJdV@Y`r%F7g^qNZ`E8;
z$9bxxqF%=&)p)Z~{p`3aZt&K69EW43$(d5ikxOSD(cTf8V11yANFPAG0U}ofzsW5x
zx|9_tKJ%X9&o6ucT|${ccR6rl?j&hf_H7f%8;?1B-G_!Mp0Q_(*sW^NRM>u8w+>gi
z$&$NFm3w+*;b1v6kU`R-ui0*RM*&Yr`PmOvgY&<^A$Y?O1H*;!{}guqU;C>aZa?Rv
z!_EwKiv4EzXQOy&<VI~^63es~R58d!UL`-#qpzOO7q%aoWOU}coTpUJT@zR}{OICm
z44^E35q9B$o9a;X4!LWF0#eAAD0D4Fmx!D0zA^a2A?YfK>lTOIBj3CeQzjebmWwb{
zvG1^8q)%@V?Ve>jb_-YT*%b}Hyzu!pkJ~F{rW?_hQ@l)u3OW{7uW38W-rKdoV7VMZ
z{UB`fd;0>`*uA8_vIh4K(TF^piR5fz_s1B-uY#$kj66<}TNV3KdU%wEl$lTJ52gJ|
vG&;<qU*O|@togM?=jrPIn?Gc7hBfFfJS<lGst^1%6$}|k1&Ms|=WqTm==sv%

literal 7293
zcmdUUc{r5s+qW(2$i7#IY>D|=qU^gC3}!6JmJljymJws$BN7!^!?-CEO3D~nl59nT
zk{MB{n~bH~G8&%S@An+P=l$pT=Y9Wr|G1CizOL&yuk&-B_jN9x>&`^kTJdsAax*b8
z@t(0ZcVJ>--r9dSS>c)d4L%ztCYEK?d9(#QWM*b&VPRooV>@u*06RN72L}fyCnpyd
z7dJOI4-XFlf#BukJ$UdSA0MB9fPkQ&ppcM|u&}U*h={1DsF;|TxVX54goKoo)S*L%
z4j(>z<j4_eX=xc58ChA`qeqX*%gY}-c1%G*K}ktTSy@>{MMYIr^~8x2YHDig>gq@&
z@}GbH(a_M))YR0{($dz}K6&z_j*gD5uI{N*r%s<ft*57_udi=lU|?uyXk=t$Y;0^|
zVq$7)YG!6;Zf<U2VPR=$X=P<)ZEbz#%$c)i&z?JX?)>@lHa0f4wzhV5b|@6e-rnB9
z!NJkd5sgM;Fc>E%Cue787Z(>-SJw*{F1We5UA%bF-QC^8!^6|l)62`t+uQrnrAt0O
zK9?_F_VxAk^Yiof_rG%GN<ctBU|?WSP|($@SFc^WcK!PG;NakpkdV;O(6F$u@bK`6
zh=?0EZbU{#Mny$MM@PrR#9*=5*x1;cH*dzp#l^?R<8ZiygoMPz#H6I8<mBX(l$6xe
z)U>p;^z`(MjEv08%&e@e?Ck8EoSa*?Zsq3Y=H=zxzI{7CKfj=$;Le>pg@uK8@7^sc
zDk?56zIX55{rmSHJa|x2Qc_x4`taeyva+)B^74v`ipt8$s;a8$>gq?29@W&;)YjJ4
z)z#tg`1<;K000^q8X6lLA3uKl<jIq!rX~V`@bu}^=H}*>mX_Am)@RS2J%9eZt*!0F
zix=(f?Jr-xeD&&8M@I*dNbKzFB#}s6U0ttVzwYkt?&<02?d|RB>wEL&O@Dv?+qZ8A
z1_s`}dp9^ZI5adgJUl!yGBP?k`u_d<4<9~^jg5_ukAM95kxV8}OiWBpPEJitefsq2
z^XJdg)6-wRe3_Y<p-?DPDs^^t_UqTLb8~a^^Yh=nef$3X`;Q+#78VwM{`~pt*DnwR
z7Z(?omX?11{=K}sOrz0OR#sM5SJ&3o*4Nkn{P_bx(8k8b=H}+s*4Fm+Hl0r2+1dH~
z_b-FN*xlXT+uIxFpMMLVc&-p@*Kj5#9>M*Gd2A;BFB6l5_8D^%bad(OhcP#WF)!2R
zk!F_d9+n+POM1F4zP-zqdYjGUWGKIhi6FmKBIi-0?m;CLH3?q9*KrbxM`ctFR<KlM
zSQ=|(DrH}7eyhSi@t*6S*r=DguYMhKD+iWO*O36ng;Lr!IWGbLJK_p-7^q(n&V5%h
z3VK({n7TPsJ?sh57O&?-4rDtE|9>`kO5E}^Nbv4UaINqz^JMN(QDN!<3B;`yOoF!}
zQ;TsAFv61N-oYvE!~uBNqoAUcfhxM7`qbenu>R{K`r6RQu+M`V?-EoIUv0n5_4AJ2
zXm{VP7xG%>$Jp^IlG{`^aVxFq?Ph2L?NYQfX>8BiEa|{zZirmiYRPn8O>NzcpjW&<
z4%$N>S55<)-&Ehe^WY<k0cDb%cC*5dTV^Q&*YBwrsz2ZL8!<PkrmUbvQR+|bs8%CH
z#V!`sttrj+0vBr|2Ol6eC~pVvt=2}COl$qsN3GrpxdyE-=?}Y^5KEQ4C432`3E5*E
z{Y;8ZH{J|gAAA!$K$VPfB@KMMp8g=sxm!&7@~6cmpZp8lLg%(-+JE=W+s7FMc$V<Y
zN1#0kM%F*spU<u0J99DP=hH$@1GsjJmWg;%oeHpe?&i^NH2TMTs5o0^2874@oW~p9
zxbq3=VK@7NHs+ZJ=qNOUd+Q0mkaa{B4vh4HQj0t*NHo;~%}^SzilStnu!r85J21vL
zl)F*TScqCp>89|$pDj>~4E+a4v_WD8fQwSVhyxyiw#P!n3;;?MfdEM7pA$<16!{&X
zY=8T-O0cFv=2I32QX7vw*WHbepW<uv#$#ESF{BlH5Q!$T3Co`{{e#D5cOx)Hg%{wr
z!zqpow>-_1N-%XHI?IO=o&&VWpZA0)ZqEtuBo~__gH(Q*+l<UJ@|2P*u&)+rK#onn
z>dRwB7UHa->u8Agpdi!mKWTv!$14bcQh;=Tr6ZeRG+b;BjG$3udL?+^$CY5K%p;g!
zdlx(y#Qysd-@bG;-T8;PKpG%cBsX%pYBLd5-B=GVFXt*CFm^H-pl7MnzQw46o9V5G
z`1g&HZ0PixE&B;OiOm$hyS-3_fT0Z}N*z|hqKOvXyg-x0isQeHMjO$#@h~=)W>UGF
zMIO+m)MyK7v+|is!yXIi0Lx>UWe^zkLp!A^|B_`UzuMVFz;F~pUVsZ7EP^nw93mqq
z+U~VDnt06H5MDo&!UY4@y#U>BVo8O4HdpsL#QD;5rv3FY&zZP`Ch#oRp+<U+P8W=k
z<>|VDRrbHk_T|3!(;hml>jbqdu)@Ac^-gce`x#dOp5P4MG&tJh+shCxXh1=h`nyGl
zVgVz)WqSH(|6c8S=X+i}mNmZP(`uED<KOrB(g%rQ*U!93*q@FA9OD0py5`yv;=uUc
z_jz?n<#DU;Pj)oXFpqd``EzqE9Gtd6tx-N30n>VE*Nw@G%?x#x49fY{<2WiEn4Hde
zRe+bkG+Ftsg${r7?H%r3#CL{-*Khp3<H^5h4`GKcEbTb6lR1<xh`nUTH1$09?(p{#
z9PeA`o1k8#%-Hou*!6jX-@9dt?Afz<)ZRI+Y8qRIZqDyYVZ@jeBMCXaAJ0_Gok2nJ
zd?&bW7Fs~Zup9s$n>5@LGuow<pRGQ+^hLkcwGJ_tpJfGAq7}&7n6zqoXivENb?106
z1vZk7^6PMpI>%0K<G#pA9yg3&1q@UEPDhU>B)#qHZ~B9qy3PNR8AB~xdo&KU`Jxy<
zB17wc1<ZWD&NGt$1izyiU3qaL8Pv8LNorhzzOjm6sBD$yc}-o+JjWMr)PC^Sz*G3q
zaXz)`L4WTH13D}iYDUL&%xHoj+Q8by@k2Fu<F+H@=&Z#!glS@LuZCA7(FVpY0Zs2Y
zbb&HsOEPuO`==zD$bI$2olS(eG1=;{(}U<-mvlh!wjHz?Xa2;DJUB{!1taD{#}#_l
z42VTcRDoQV9T~lO5d7x5Fi2V}W0u3y@!e_)ARH!5$rpigf&<O~@VFlKX1$DA6Hiy_
z&We4BcwPYuIxt>5vvfm28tW3PM}!iM$(rdRijx15(+5a9;q%w<*ceyM{7pm^3uaW@
z>0^u1kQ1CjQ~^@9uJ(B!6x6&n5ZB?)e+-YUB~*PYAWYjsY+?9#M9fB;LIwE!GQ@ya
z;VhCy=<Z#qvLY|aB!m4|Gg{&FZ|P(+Z@|igf<AgL2+74Bf&;Y^bR^|=@Bto%MK%it
zto$gb(A&>E512HGAX8T#-zUNfpZ+nSXfm;2>_j5fYQCV0tO4~pVp?t%K$@2dcdf#A
zg16#N>^8)FF{_hXnqfPD*`pv}fgaDldT|Ti#R7fxs8agbbLr;{$mPqvwtu?uyFGM`
z=ZCe7&<5l@tsCYvGhuR2(BD0^N=cv;EY|OhWlvzswi6F^n7%)j9EcADjP71`g5^BM
z13hTBKl=&Hgc1V*>22b+syDjn#5JnX$e%qgW?@IhoNI?a(^<lnq4KL|Htx%{l0Xs2
zQ*b5acn&B=SM`Ln$f##%+#;WrYZ5*_j#0;vcV8MP1gY^kn}~!}vAwab)SU)xTIMr3
zMX?9cKmqv6Ts;@W#nny+Pw#d9CZ^{^hX3klL?0vb)d&)iz@|OMYsCz^G&0|hbzduy
z%A{qR0=UIc*nY*ea^F~kubnr3)I0L=a?Moaf_ZiC1JClq;Slm&8ynl&IQFETp%Hi?
z^*YLOKW)GF<Eo<1&G~bD7j2YSQzsf+_W*8fTaO#kL|>XnkV)I}Gzk}D6m<Azpvqn~
z<%r_^OTNrAKO;p}Q)*NIBg+$5*c$=YHYRpkNDr6QD7QYj20sZGS`@J-M!j{StD3;8
z>aFo%tB%eje!u>sa|r${{O&E>T3O<;kB7;@r5cwmOhsjU`(HCe%QD?enL~UGP-C)P
z@{$<D-}#Z{Z_GuG4gS<ua4Z9C+Sz@{WEp$+C!@xLi%i-;z<5(gTf<^)?C+<m^;^Q%
z@A*B{$gl#qtspC`v!*XsINEcuwOf*5@aDfekKRl-5mel;jKT+=YOIZ(&dZo3nq~f^
zT|*6K_O4&qMx~T6iZK`-J=&#_)JHw}tI<i<byM=^YGO9>sx{|~3<=Ep(xre+MB`HJ
zy2#Gp@2BnV7Xz^u+MZ9F>R30?A0)nVZAr0@Q>2#5K98E1>ATY+b3E2y+xV^rr0%y`
zmYB+x?MVqUwWLcI+4J6YhthXi>@ecwAdZF6dwZHQr~!<pW(Z$w5fYeum%CjdsnQOX
zm-olA7YVAykJS8rQlA=#{{m2PK3YN|mXNcizP;=-xG??|wt+=x^|+cLg7stiLKz%w
z*<7M~7Sq2*KbmTz{m3+$*o;=HeO%KcE5Av2<`Q`lq0jFQO9c<Cm>qixws?FqU>)QV
zr3Q=qKGk&DiP5Q4_L55`2b}l0u-7gMRE`0Znc}s2RY7kwNs)<+3c!?abJHZIn94pC
za86HCXOAfh!<AduNXK{VOX?#aYO_-Jdx-$tZA9MO@aK6ldhcZBCHx}*cVR^Rc%j!S
z?vJ~`D_vVTii{ZhDGX^{fldv!hJD9D4PE+rlCr3rcD5cHCEat!>^+Qv8f@S@^07f|
zBs%31z7W8HHWW=l=F4h&&-#(*$foDnZ+bL<_<k?W4?K^sqVj0sE1EhI4nxB^H+)1s
z3bMPrUY2;xJf7#S>{|V9e~GubdI89H1v4miN70BgJpyPt%e)_QldYeMgB)METSbLl
z{e?D)Wri)Zp3w`)ooM}n`-;P-y{n}McbTi-2Ah;Vf3!co6%14s)OW7Y=R>?<Qyu<l
zhuiWUveZUvkd2x%HW62kxpesVS<+TxBH|0qLDT7GyCm5haB+*H4^@x$ynmBrZI9O*
z?b#nEY*ypI=%KBYB@Rz{bZBBoG2%;2oZa;l`rP4qde%3VDNg|ubP2ic4qJX09e0At
z%Yj%=hP#pq@OW%_8V`?5DY$8mxN8aVseLzKJ!sx(nkw6jH{3hfMUa6#u_U5`$Tx?4
zi6dR9WEexw#ma8}Gw8i899)5T6rP~GQw{;64Mq>^k^r6XRBl0Kw5K}O0wcZ6Eh6uL
zkN+G-(VR7}*m}1dq29JGmkfqmv8HL8ij$2_k3P}7q<a?LT2k5d40)5-=v(~6{gngn
zU1P}6X{C%=C#jFdX{?RDNe&ig6`x@9MoZ870F~Vhomx*bZ{I~Tb~9lX-quztsnP)|
z;@-xxAA(P!&)P?+;zu2J+UN*!6G;fE3d>QenhEJnh42XuAr~v8my(>%kfWxmpPgvz
z%nHQM0_E0G=5kW6=R9)_cJfl<S@CIEkLgXz><Br2OX%|lmCZ5do}0h6zo}GWaQXW7
zuL^)XYo4&X?*Ti{fmrrZE>V(UdysRo|8i|VJtOlL?~F_{I61>AUYiWZ*Y-!->hMoR
zHs>ifehk(4kEC4ZcY*pc)m|VC;DehUveURsufGgPlZrMGFN9R%CmOzfiNLPN6XqUP
z(A6`K`&bFoX3GGy&(F>c1@|y+jO=`N39BIhlOn_dfZC$Cwhd-akf{*|Z1bUgb|O0$
zHtPR(!HeM#P_@OLLzOO{3_+J`ei%&%D0{~O_T4D(Eqpz9N#dEjJC0%Bm5FKZs?xtp
zDtwxA+78pf5?raW?8+wQmm&iMcU5c&62<(uuxQ-mLWln>=TR}wvQO<X=UZ$gdgU-f
z)pYH&$xEmed?)k6{}8p_CSy~eQ>mRzZAe@KO$EN}y@va&<l6u#-2Gczp-kB7YNC<<
zzcOPLXmlSTF3Jw4M)^(DZ-LU0oq3(M>~l7DixZJ<KiQcvp7*XO!KVo)Q~CYb<0mss
zud)^WO`JBcYl$iqpcM}Gey*&`%{8N$2a1}Rf%s^^Q(FY7stVWhFly|A{$qR^cO88_
z(|Zx+0L>Ty41Y2TK0)>Ie=I4QA^flBta|^vsJ0ri@#&BEkH=pDJRzMf0w<i2HbGs#
zc)S;oCj+Z*3ZR&N4pQQ&rQ4UlY2KbK+vwj_;9<3-N)c$ZrRU{ho?_3qc4s6o^-eG!
zDVWo=e)r-f<%=$t;c1c1(?HWBNtHs->IYoddN@U?0q1@jP3t-GikyLHB`^M@6#)N}
z;YF;W1&4QEyNBQm_AGhbe`xRib-Xd$Tb!Ut=pLMIQW!RaXd`S0%~T}dZz4`EKdT{&
z#b8Dgkbrce)gO6oeEh0v7xE6=++B5!GbtFLNdXjjxVh-vgukKnBWFgp6?4ydI0+D3
zp<0;B=-lE5r0J~i?!p5YlGjNu@dGf=bHm`qI$)Mj8`=pH+pc~!IM9Fe8Q4<Jgs6ra
zFRyA4_siynCmwr8*BzqBy2aSNer?@I#~B<Dx8wrKx=xa%7u|WqEO~%3DT-|G&dYpD
z>jwH{89Sny9TvX>=KZPcj8uS;**e7ySGaf7&4$Qh7{~7?cq8_a+cN|{%h5x;RV%=D
zl;#y3nzHT@Wz62##V(`*%-2rhcUO23a5Gu*rXHqg;>#U7)=XTOz3%V??;KcO)tQkf
zC4p;2K?<R@`^By|mfKY3A-FKif$-S}@D$wK^x1uFN-pWm1gt)kSZI*_iJao1`=+16
zkn-GNK0B^!2Xj1SyZLIQLw)@6xdd<fMA&O6D0XMOELZ2eOGE{Dt#@WqhNjG!Vh-Sb
zV7drO`DRdJ&{ILGSbBXt&5O8Lp^;{hjR2_R?*+LHT8|5+#$DkuXYyg3zkTb`{XQwT
zg(Q*SyI^@Py`LW@9)m%oH1eJsFQ&r+ph`T3KS1T*a@x_jddJeo*r;-|jyf6f{A|zI
zPZc~D?DLj$ofvn5S<;x{min2HbYb8HC@bA;QqLOFtJ<ga2~}x}!L;qR@>u`a88s_a
z5n^6wdGdS_ve>Rx#ZX<ec5BYGOT9@c#AjoWKj1)uP`DVYQ}pvvUt#pBe6BqsPlq)F
zW)N1Hl&A@1D+r@ZS}99PFjwpRKRPGJTu?ez3W?y9$OZp%TVtxDYYt8Cu^h#s<CGq=
zh()Zn^3BX5$j-IKqO}L}v-3QE?Ga>UBa$`IdvA|+A(5B=hieHc=7*(#ZM>|AWHWYh
zP+92#*H4QB9Dop9`f{rL%~qUe^;_AH2(`1^WT`nOGEU<V5`hU8+y$DfPxr-gcOPv9
z{V(_UR)RJ??hst@44TczB`IzD74L^0bESX_#23#L=#M74>NdcY&PA^p9PW1BFLEm#
zv0nw8pz-1WLZl_RB((utzrM-H2K+5i<m#W*7AUfv+wxulJU&mxJrv!_fZwdi#T%#l
z7oXIaSd*!wqtUFG(VPr`8sa~50&Sql0(5E#SV$YN0K1)eU~m<9U%(FXymjrxEm%|+
zE=QiTp>1Qbr67&*;)5a{>u2uBaNg9~MZmjw0i=6BoA+#M%#mcft;T^mc)8bsu`vH2
zJ{Oi&B8zRBe=w6R1p6QJx+cOi@?_kDBQUof%`pblmt8N@4xhbc4D2l~^s2y6=J4^u
z=3$3{G6}eMD7@NN+@uC~A4{hn-rZMg&rsa__1J=}DVG8A2+5T`5UDihC+BUOMrp$m
z|DZ|OtX(X;iz)ty1z#RNU^jh*zzl2Rgmdoj6*$Z#n9(>xfwiy6Z2ZPheF;LR61)%R
zj$#`fN*dt*W)+Syx8!+@(L0wvFaZsD2^JfQUq24Zr8n$eJlhUMRmmT8W@K~_O6;K_
zT{nm}%t<;7E2zfcRJbgHMV@;G6j(xXowo?Au+Vgq)|pCrRL1e*^dxxqjpH^C7SIJE
zM!&3?;H3Z$NRE{EVjtE1=mL~|BopS8{~i^w2or0j;%m;d03?V*NDs#O4<q)bz-k)%
zIn3UMkKcwVS<~m`R>u^*<dhHtu~kynT=7BUJB7>Vb<v}0o1V~0_YqbRD(9q=(RQ6y
zutl|R|HO&y{Nms+o8OU@RRJ${9j;!I9v_T(f#<g4n(fUx?qZ2C6>i!YIxN*yED&|&
z9s<z0`QxC#eWkK?AJz?xxdhi}v~<!}&8Vvt`<I~D9{*+j{LH!gdM9COyn<!v|M~A%
z-QOa*8E>aN^x^BwI^t9TFMJ7#$Fd#VzcPYvLOu+z&R*fTcR;e}3m39+nn(1HSM2I|
zBy$>cV2Fix+BaH|k6rWx3!;HFyG<UEt=KhJ|9Xy#5bZ~ZKG#i%W>GDZJe5`6$$`Ae
z<RyA#stoCWwaRZmu#0ed3CF^*{5|#oJGR#GEmHlPkS@ll1LQ4l#{4hW)2U{wz9FN5
zkBJ*=-)4~ZRK-?FzDOs7u9$76zq#a|%B{?&`J4arXAX95-S88;tz`EqZ!q0A$vkmn
zknc-WBlAwS=R?GX?@grjc$PYI`-^DX(%SElY1U)L3*T8+^d=I1_mwZd<XIY2B31-Q
z=_Q$;uqJCchu>Cu9}_^JK4o80qv&moYu@&J15Pv;hc%CB4<BaTjFQC1yvFq>O}YL}
zGGYX52i?)Sr`{u0D^kZeC3f`uc=LFm^y+ZMZWyPIx((}2u6EzD)!!-66Kb~t^7(|E
zX4ZED@5%peh_)G|u?e&5s?X9~SP$ibU*qGQBbpiy$iEx+OIFlomFJG|h`$`&8w9Iw
z6qiQQ5<W>%<MluC9FbZ0ZdNcSpqS8ioLjU>$)%&pYc(tB|8^kl<?Z=c;j1I1g&?46
zn4shPYf&!RuG^NdCG%=JN_QakU}RF?<Oid>FsC?>&DX5_WZ0GxP7qQxG{zJMi`A@@
wr9AXnWiIP<Ti=LI-&DQI_W$&|VH+2$CH}eOMs_m%SB2?}g{}D$Q?I1|0zf8D6#xJL

diff --git a/docs/dev/reference/figures/op-leaky-relu-plot-1.png b/docs/dev/reference/figures/op-leaky-relu-plot-1.png
index e66d27a632d252c8feeb3c572a887146c7cb34fb..385f87f8a1c58ad4b3ed3f0ccf68e52e5ef698ae 100644
GIT binary patch
literal 17394
zcmd_SbySpH_%4jC2x5RVhyn^Ipu~WPC@C!<B`F;;Al;})C?HDX5DL;I4MV92NK5B{
zDBT@W=bjmZ_xF8koxjfc=geBJcfCB%>}NlF-*Mg7eeHRES5AVMkcyCifPh%?j;JC5
z!M=6uzoYx%FDz0fm*79gtnR4W5D*+cjs3T;7301|KyZ#gQuOvc$EX>MldH_yuOstw
z1{YpeofYm-+*H|AY*%y<)wm~}wJL5Zd7eSfua_*lUGa+|xw4cU(-{`Y_QhP4Q=6x6
zn+Ttg4yrXkMJ{dXqjec`T$}m12FI3mewNGCMlIbal5g4=c4=z46ZD0E;FhMn897X5
z@$=gw1Se?@3lcp3&N;r1;GoZK5`uGlM9u^RpB{WWKydVgCmq2pD~e?Tf~S@L&);fu
zx8!l0q^#`rN|#PY=6-_5Tmk|mh(Xu3mKKGV(S6--2P-^wGqp<4wi9vo3<Lx>EP8Sh
zoR&uO&AY<%$9Q?iDfMXu?6-!4<=ZpQ>s*g+>gnkj7|@;(9u5r)8yy;YXHS+B|1Gh_
zq@k|P+Rje7<fN0Ku`yS*R5*KvJ;TOzr^WZK0utiQk=j`ZssJ-ph7x~&qI~JQ`@MJr
z#N)NIR7(hZMn<upR4p&Fhjab|b;_ytBxg>C+|9VpgTFC4D(cc!yq$xCYRN&KA^#Yz
zXgP@?|C?8?$ii*6-pR(dB)3hpXA0ZH+(wtZnTed6PGWqY38Q?Gmu}fsQhKH7ieG5c
z<?a==71`G`&M#tn;WVMedGtF1&loOI#QAsMnB%AkPhzaZhgc(8zH(Cp@Seileay9I
zm{)G!P_ti<-8+33o?orbu@rBjT@_Cg59nK7*IjMy@+#F7{WIvT)!UZ-yQVvJ2U3`;
zb56+52<XWduZ^Z}T0i$#s#kLHSQL4|xHd$lnxe>b@#4k&v{$rVGM+6a!E&}D{ACL4
z=&MnF-5SXL#F6-4WdrZ(KfVg^E1|>C1;^}m)qMHlpYy?WW3keQggT%)Dk|#zdq%c?
z$K~-BdXKGl&Xm@cH=LKhQq{e3`PrGF(RtT9ow~TVIFL?Iv5cq~ar(@e73Z>`BeL@H
zI{9YGLqbfJIgGWJ9XxJjbe+}Wy5r6EAR3)jWVkiel^Z9~MLE9`za_LjKUAr7*82qQ
zb={8yl0h=eN-8R0WXS#!2M$3&t(9xT14TAtFL^>3)~Yi!3e(>asUxhdt!FVMN!aPU
zPfScqNMLdL2A`FwF7h3v^78V9g@vFS+(C?%u0aka4>33Url%0;8f`N>ua6U}#&%yX
zvaV~HuXTTra=XSMFj?ij`HS!0zYh-&gC%}dW+%AyJeuFes40qHXnR>{Na&jN@YiqO
z?rONnWb2QSArzA^l6T-!)G}feR`U94WrEwLbl|#u+kBj?6Y6mdsUKK|_>IJfLRh|!
zA8(~yr&rdaEi3agZe)((LFKvpoYR@$ovJ|+(x?$Tmkrd`O0M6Ief^qY>!1O5FE>?y
zbD8$~ZA)$I_=dN!{5GQ(V_o%>g=)#?Lhc)?*<me9uY*|UqFhvw^+=r$y?gS<Y(f$+
zw0=$|xJLD;lC&%S8GL+v>fPSPeSSMegphfTol-w%RkJ@Omv|tZ7Hm#2$-R3N%TWyI
zQ^&l(9OtdK))%I_vPh_TWjf<YNe&Ad`yxe!-^q$1G+||A-f){XzTsJ<jbr5eXW&D6
zlkF+)?(W*!+W+GUOR;{<OH8Sg&G@%>*Jdh)_k4+kp-h|p^l82PcL9`pp2D;|krexq
zR-gdh%6dAAe2<e!{TOeFfByXWT-h4kn&6Vx?Pkh2>eqKo&a}v8Pptepc7|DPXhtDy
z?=uQmloAgfIilfYqVFB<Y4gwVCPWdg4vxcLU$?infAT!CgvpSx68|^e)$x^I&(B<B
z*C~(cIx*tC=Sf$wO8{%cYs>U(+?$K~R|&^4!pOt%EzEJKxL*+c=mnjZn*`?Od*AKf
zJK}2tBO|%!>vlakhD?K@L3<||1W%HS;QTb2`2>Q4k<IS@7e2y?pG@+ek0;Kr&JD)K
z#N0Ib_BLov>H%N;7*R{WolDqDudZ=%88<{MRFUh<vr0<t*+kOT^@X1z!oqdo9L;Hr
zObqBD|If;t@vluesURG&_;+`AgH7%`aCkmdV(*ZnUaf6yVC-!P(&39&{xf9h-8*-_
zhOj6-<j%V)^50me^r6gUjf%*-aL^rPDM(mJ`Ojny-I10?@tB9n`tTa<ou4i&AeT|Y
zCDqI*8YY%wEyTv8nNKMLu;B?d(>*Reo`*_yu-t7V`M>ikZf$KfYKX{oT+rK)QaruK
zY?^tH{d`x2sUglrDHIVJabF`Aj|Za-JAdH<EiG-;QHzE>yV1x61C2ICmzcDiFP8rA
zKu3v)qIk?zV4p_VB_(~4eDPLf4JHOu{^AhlW8Qa|X=!TyTAsMz@ZVS6b=+Eiq^1_Z
z$z~Dr?4P|f3RQ!zx&5+5AP|Ny<v8l5OPYtcsPGm;o$Kp+)WKnGpv;9$DeyBGMcagJ
zQ&ZFG>Z%#}`K-J=ZW0m_T;zr<rVsHjEiJ9r>uY^fUkJO_ui}c$@1Hz3md57#3gvs^
zLOj79|D>5}HC^5BQT#T~i107ICpYDqcV!Ziu`W;dDAhWX{<ZC3N88{xZ>Y|mJxfcQ
zA+At&8+IAzUU1<IDQTYWr~N|}&DgX^sQSrY9BZGF{OF#1^YEl3<E~6CG7bsTzk6IO
zju6Ff20pOB|L@zk?>IR->5f1YCFI6`i^?63f_PiH+$tR{;GmL5=k?&5jrX`XBWW)q
z1?K(x_vvcc=b4$sXZF6|r}65+L66!XCd0ZAg+fNA|IX?PdSeu2Y1W?HlC3XcH>E)F
z*G?2$WW~j+-|(0pbWfCZzvqjTDO~gRaJVV4ae!?<&aWg{(F<#9Iv;eO<mTq~j$D%6
z;c{kKYTEFVRBsik&|P<{+_I$Z%F1+DOYSJW%_Pyoc9-<;+P*)7d5A&ht7O~<M^`L4
zOZL@6oC6d1rZr37hwi}5c^>LwPTrA*@SXDV3*^-|gBG_nA%*#SefjbwJ^eB+=NTC#
zk|7OnpJZ+Ce(FGap#pEned|ljLK)bcM|4<3T-<xo@N3zlk-E^1G}Y!P{x+$}?EY6g
z$6k4S>nmHVZBDY6b;^)x+~>C%RR&4CxN)vxOT7DeefUpa;*9sTYc>dEzvJv%#BMPK
z(cx%0V=rm|TsOuu-qS_$Ub}V;a%(XHA_4As4ZCUB{OV6H<ldRC{_^EZ6~(|2?@zj~
zEQ!n+S$2cXV*J-g*eS_M`^@{%P92>En{k+wSfNYR8l9e=p25Mv4%6MADsB;+Y**EB
zvx6xgI({};>&|WQmo$`ARQF5lE%@y-yGitC*c>9al%`)8wNwrs9avL~v}8B?HZ<~m
z|GRIA^BhOyrKIjxxNi^F2J5U_TRmYxmnte{<_~rqLC^F&dX;adu=tU%b!tUUPVO^D
zR>Rk?C*3g&@hyo?9RVNf**u><4ZEUYWNaLcLZJrJ4tgbRUH^5BxKx7s?pE@PcOM=<
zX;FLm*(8r;H5Da`;_b-Nsj%oPFe8t<2Dviw1VZK*<jwtqoS`8hc@HYRySvp^<H9|U
zlQg4;ekWD9$m6^`K2jf!@^SgZafF->Z2ZhR>$)T<E)XfiqjtCFJRo(Nw<QK_2pxCi
z=R5S*TfF>{JL^9K1E>*iM6eO;;~AXeDV2_k2fc5Jm&j7f{!yF#a{RF4=|$G=(>NFh
z7mX^tiLYoBfXS@{SNOtJVH`FjJd0suW_AZ89%dORY<u7a-UY>8AQ>eW!L|z?Mg|AB
zaaDL>$D3y56T(mNbsWlWsx#lTEm1j5B@B1<_jqS8{pD_(D9dsu>x9POlNbVdJS4ap
zM=cq&fB*h44xQQY^(|FO2K2s|XT^l<6!;nXk2Bx6QR=q23X^P@JO1yB`{Gb2gr)l$
z#vu$FE&tBE*V$J8S|k9IGV2kkBR)^%G)}?8_G=y@Ka3l(tur147Vj|8ngG*0<N@*N
z`!Sr`nzpqB@aNT9{%3xD0mz1TvW||9?!TXsGm-ZFbNnl88tg^d<^sY%3^>M11~YqY
z=>2FRhApeC|0D)rhuV$bQCRqMpwJGV`_{U!yPLf|??||({-4N{S2dn^2XG5Kzg9f!
zsox7+x;FJ|X?#J(=Da%HQ}pQPXODv|#PiQ^ku0xxe6%;;{M4yaUL+O`{}!U}1rl)X
zty>>cmD66JpHjjX3~W?CRf#ZQ%ZR*;aU*zDw_nSq?a2xXLjSh7_pYtY%4A1!a`K~}
z)nq?&eUXnZ;v#=yhA4tL*t8+yDw}HN<pGN^>=kM>Av+$UU@uFDU7qO5(!uA--<H+}
z;03-gjVVEJa(?J`x%Jm-GKA1qBV}&iz70v^N^S0b?1u0@BGa_S8_iT8@+^mTKR689
zZYs>+;=hT6N+7~HbRb@^O9J-#yT$iyZ1p)Qwzjt+W0)JOwn}2HFlRAMSNjq!#dgwQ
zE^8K9eS3S`ZnB+aBlF){r3D}Xj%|7Du6aNz(Olg#rZH~Jtu2*I&6F0CqUn>^b<9`&
z<f?!xvNE{P$BrL&0dp&U#%uI%ie%!5v8HGN>tWy6So#{b^I*OBdam;<hJwiZ?c2BD
z2@8by4O4L>yL<!(^b+&M5V3BnJ>TAmz@jud6#g4gSq$;$(Idd*>rluR&0+WRw*kbA
zSL<>D63x;md})baket*s-Hp5W8D2z33OLMe&Q<7?Iot8uV_hBZeD@`7ZNQ!e-8U`1
z_@25pL;r6`?{Q`T6`Y(_GrjrN&v<uf_O}pQ=`*Q*vc{YEeGywO<EE&gN}qbt-UzI3
z;@>V+0s()9ilPB3E<-`WzfP2bw1s#Lxt_623Uv|tg?kl*DtOmZ!i*ciJT8d)-~HbF
z@4fC_Me~Dw0$y}=b0gzB8%JG5X|p85N~1!Fj}iHC)HWu$W*w}CCnRp56?DHzlXz&5
zbI%9JVFpmeW0TLD%hF2t1xl1n5qfQ2G8-hBMN*-x{QUeWT3*w5lQ0uQ09K^nF$>Z2
z_&hs18`4I-am3BJ{ry@P#(cMyaW`!2qP>&3c>vj?6q3xjU-K=-^j9h_4Ww}3qiE>(
z=$Gebz}_~$dy<d}oR6b+qRgbZuFm0D^@YLm#EHn2S!v}#nvM!2e_*PG&@m^rpX<*h
z6wB%Ri*1a}kM|W?RzTqi%tKR?+?_igJ_jr>FKZMcQ?TiIVqyY{-+pRINy$G&F#Ji1
zZh3K}{?n&V^<h_(7T-F|_Frvw?%4gf)c1<#R@)uLo!qb^KR5ViyeCv+kD9B#naKP?
z!;?SN^HsL;y~H4~#X{xV;{u0E&Kg?67`3&v0`}9d(liNP^}u^T;M|Gb-IACQPOty^
z`htjvNU>0CkIX8U^JX>u!P3RXz8SZt<;Ao8ZzZzCk#DA#24!eFU8d&*g~qkTXDiuG
z;N#^zUNpb^mZPg{>7vqr>-DVp+ZjGvQ{yvc#Jf8WH${%H3zY2GNjy^sf9J$Ap49A<
z$Hp-yc-rh6F2(+ar@*l6fu}&SIM`ijO|ZK<g0B<7w+LybC`XwPU>+(v-#Tx-rcRu$
zM@;j67!ul5AI(=-(pLIWWcHC*38<08pfqHnT=eAJ+}y}WLc>sus8cO*C9zUJK6{b*
zA)zR=Um9&>5b?fbby@mLJT;VtR+S-844K{=_GZCVlQ~9>RK+AA^gj?XL+Xg%Rc+{V
z?8w)V{Y7=5Y)}kuo%`zxgrW!uNH;cPO(k>XY*Pw^|DsAAEe47}C@QfPw1e;Ge=%|7
zB{ESa@-z)iy58Q<R!4=Qx#7sk#AG=C{Sy?s^S%;@ji{Ef_tiRR`^{z|fwj7hDV}t#
zcOf~-J)HP5z0RTXo{^D}iOD!{V?}et6jqh(!8yoXv|&dbM}=UBiE4s<*&DKi5L6ek
zUq2sBOiRXX)IHU~Ij?gNw~;J5wp^yI?-zd7xHvo4WiQ2@r^DRf8Ak-r<YYZ;-fd`a
zGMyaRvP{-*y;vpYQDOM`6|E)v&X4WuPh-=<Udbid*xR4o!Es#cN?K4c#kWL+hE{F+
z?{w*S$VyAQj72-F&lYqy6hn5T)Ne2zBB^9k^X9i_SJo&cYyXkalER|?by<f1KN$(z
z;K)dRh}@VG2b+H5W2XX<F;x9AWHf5C@08mdv^%#qr}ZjhYd#-1`(^an=)IOUk<z4B
zkG%2BgtaK*46{rG-=nxth|`Xh?QJxvs%;swU3JFdUji<}_d5MuGpan??JX3KOmQAR
zl7?d@X48;*`ucX)F(02C;GfMM{xCUeYR@yMZT~iHPW7ixzKdaqLF%@``i{+ksgfTu
z<*)Ez&@~>_(9jUcWqe;5yNjdh)$N1!nd*f?S(QHV18Lz~^{ezR8C$o-)eP{Vth`0m
z%E}5DpN+*4r<qseq6qo=0J^r-?O4H!60rkiRz~Yj&g@P_a%R|+$On$5C`1$EW4N_H
zc7jWbi-(CQJ{nrOGN7GC$*GRa31}2H>8J;|c@@O-(zr}@r^s9`WTC-Va9QKbf`V?n
zd8RI_(?g164!+2pd^@rBfa%ng$s5661suyd%L}xsJ1{!azFe2?|9RBn3uHBv3=on5
z+=RNSTAu@RuoE9F+ZeeU>y(mTVsGA-tc`KDGYbn>L}~+~?)5s8edLS|)pdv4LYK>(
z_TlZ)y7<9lxQ;ud{(*sk($pV-fw$A7!WU5b5||t7m4zj%=^M_bdN1UDs-%cl7HBzp
z<hfT*bwqctwuYU<QAyHtnD+?@0C?G0SzqWY_k$^1Zu-V!XlA!<CDlcel&layFrSX4
zZ*qsR3p)>6k-?jHH9o5E!x(!Occg#wwaAdH2Jr_SSiVF3vohlK>(>x@08fsn1=09s
zp|dHL`GwroKCM`?CqRu)6V29stY2geb6Lz2Z%|B5mG|!7?}a!{&Y|5<$jE=vqcWuG
zEeDU)RD62Hh=E`^9ZyEEs%N6zq2{RKPpP=sFt*6vz56X)J-06GN<H>FI(AD>M;1=Z
z=5NZNS~ucWcj=@Lc`!jj#WH4ov>228-qr-^wTU(#7fI_|woB<r*PK0f0xPc?)0lVK
z*0doB58=E+y>xsO{Pe+t2fU|xBe3hCf*IbLk9{FhzPlz6Tb=!I8$mujJ)CT!UT_>|
z+KoTN5Ko>wX-kqXe8>64TPoBOJ9!fq@d0}tsZ$4#5vQ>;cvN{$RP<R|n)%#7*+|+8
zfqu>v4maof12*xF0s52<<o%=?6h#lDprF1yNvuRdcOyc`V+R1eK?d&LtjE82Jltmk
z*`zAUHaZgv=Z)rtZ{Z4wvpncygoJMdT?(LVa%HOB*V0(Q%OuqMUWFm<_Z=+go-!AQ
zV(Ss-l}RQCWeC<zCw~51|0&^KVVFKxXdbF5AkU$C<9*QwywiCFBI#~8egG~3bsUSU
zZiqOar0%sG_x8!US6k;&_&WU2mp%o-LQ(IPWZ#uguO~S8eA(3&5I_V}j@_Cj`r<dx
z*?h_fZKBvko3?;(cKV17^ckCnJT|*8Pq-vy7GK(FwYrH<4C-I06YXbuM<ynKr%>Hs
z3p#|1aDKxu<ZSI%j$!NBwcDS?(w8aW*S>zR^wE2(-(QNLPE=G<s{H(Uv^h3QM$-D6
zJJxC(9`+{gWbbAKN+X0mDApBcyDfbR*S7Gj3ia|d#<c;wo+E+m$+a5#98`h4EA}PU
zPXB7fjG0@G(LnJ#LG#Ob#@Bs{s?+AOyT)c|0({<(omRnpk3-Wg*JezPgCh&jD4RN&
zgPPqEcf7Vv=5DHz=gpGS4*V=_?}Vd`WYbEH-OP9VjJMP8%7~Pd6hIh$P?Hz2&SJx!
z(-GnOnG>$|g_f1}q134guLCN7-#pHI?%c;ts4nY@iCyHxb<kt&heAD~a+p{3er&fN
z@o-9x-&FEH|9kAm$_RixSt=QCGJv`JqKZ+wS!Uf_`do;nX=iADEp}(q&+(w}JG=`l
z`5_e*6+c&2S}d<#k_L`UWirE+?%6rRffCEq=@IYEL_$0WZs0&KEG=miS-k^pWf5yn
zvJ4}%Iz{KGjy3t=qed(>=3!q!1~4eJ*KO}6F!KLP)nP5NK$bu?CnMwkLiv%;KZq=j
zb&~U5d!YR&S4$&Q{J=`Lr6^q_lMXuY7vydj*ak;Ml_D`{DC6_}evteH)UMaC=6)Z_
z<5z9i3Iadrt&8X$;IA=gyS@U8G<&fe@bEmEnQlbqm=g&N-v^IyZ!((QZQ`U^>2R*g
zaUQ4ebr?TdQeJ*{v7W;V2si)&XN{NTqH7wQ#ztI9l2%eS72{*d+d4&~<eI`L^!-g=
z@Q>Ps7s$IO*`)Cb3K3zJ`$U$?)dpVol!<$UpC>D^X7j=~6h_CZo0}~m1MP3V{Uea2
zfMS_y4UG{Z(%SgBpy3g<eLcBsbp7d0*1b&3MA@qZfk5U-zpLSPo6O57+k$M5?$SV1
zzh-~d?$g{J|EtV~{lyfqZ&Vjgq!Zz=a+?Wvrh0B76xSBLc?u<GDIrQN){fgKvRChX
zu;^R0tO!I;uA3DOU{L%zYP*9fjY0!j?FT3_Y4}vQ3@{B0f-YqOrfI9WIp)>rrJ504
z%2n7-(n)G+m)#vVz!p$ngoGG){l(3H@UzNJU_d>#ufq#8uHKiUY$+0n0WNE^(Bn(f
z7T-s`=uRIL^LlNQ+_Sgil165y`GJ~-zIEyTvh)g%n&mO?a*hDU)hQjux#@<m45O&p
z?I|JRu1c4VaT<H^SX!Pqv$c64nGJU+czlvB6M=1ge7rA+F;XTGc(;eT2dJ6)=NA_E
zp@}$Ys9V|9g}8ZBJIF^oDCVWUf7j}ajBAR-Sar<q;l{+}JQ)NDLUQ-}lXp*zPuZ|n
zCL~zeYK<&6Nzl9UO;)bPQ2m^-^j?XuuaKxop9>PYK|Z6Jf2r03kb|}LB0$&|DxOd0
zy;)u#9US#0riM=0McFruAJ2>(H`=z0pyIbSFflQKf^MX{Sj<*mO0rWqS$koQB^G7t
zxKcy==GRQ4VM9~GSEZP%qu(y^Eq;5kGb^cj#*Vo<!YyE};%1SPRb=TkVr>?Vfotxi
zG@6Whg&Q^W9zYS<wZ50+aOE!|@U5Wf*K0M?4B|a8?K0&d)MOf7Zrl@6D@_vWwAq$*
zwnF`>Pm6g!)iOPG=k9Y)m6_?n_qjR?YbI6L@OTiZgrlT9b~eT`Gc!NOHgPQ?WYmeA
zea<Sm_e^jvmv!3Zw8;jDHw2VjOBZwMS5Y<r_;lV11x+Q(4&XUdGW<E5`HLvhh0w1%
zY{ZMI<7L0fZ=SWyR4c2OaG!fnM@|g3=-WFw(^pssHV5sXoC@{sX01P30Luu)lJh+b
z=#8#<))Ds|W3RvvROHU>;Mo09wDYW=&SO&(Lth9Z%eiy9z~Tc1F)ap++~GKsM@G4`
z`G$?<^H#pFPazywsC6o*5n9j%P{_l~Mt7le`$@@jplaJ`H!+~Apz(UQ<7iD|mB#rC
z2X0J>GyceTn(EAei3x4UQ0w1>QHd6eF={6)+oEG@Hs!V_7UTOXVx{8i=Lug#r&v8G
zcmaKvGu&+-7}ZcgjE4>%&eSgJrNI{C?>_#xnBhHGnN-7jw9i4;G;lG?hjwH8`=~;+
z^ZG_X^@QutcXgOvdLs~Cg_eWm;FIz@aSKlHOp+ML5Bjaqc`9x$U0Gw8lG{J7XpK0R
zB^?SGEM1Lo45SY}0+~hhJ@1**r^|PDHfw_~3N1~k*@--<dRv&aAmydhTUNNzL`*$O
z+wW0-$<WX!F=^_VOUwtE`3Sw&N81u>4Hh9#7h5NIR_1n%+Qw1WO$0y&bnNgbUkiQJ
zLLL0(#8SRacR6x9aUc1w^0hZVmu?cDg%mFu!K<2azXKfE7)Sq;s)^6pYrB!t7Bx}m
zU*??D8!V#cS4MGX{z$idP(TPM=e#&nv;$j1KJ<EedNG?NFw0L%9@S>9G#*=D=L-E-
zHtjQ3bD}RMx(;FcI{OW&4xM=`8(hl`{Qms>{6*E?8_(rz<L!@`<<Km*dUk(v7eU-`
zO^{L<C<!R_zcBD}vY6>wx`bFMM*%0!fK8VCILe`sDDks8U}<xIKd0D}kyMo5kdiwR
zz0f*xfJ?_g?`4M3>|o1-xpp4%s=(-tkB~&Wol(b*9lM{w%gFftaTZ`j!pmJP7d7RP
zErl*jn-y~3c?|njOkPvP-i;{S-3QBGmDtDa!^@_Uerf$z5U%GAc=e@hU%Sw1#mb`z
z=4uLk#ioE66iNBAnMgmUG9R|7i$dCf$wEJk{aoHv$lZjOM+>zw<U1P#5P8M{UaK-=
zC%YV11Xq9cN5@YQ#_-#CDkIKkV|cT)%cv+RDQRi9KWBY`uf^EJX{Qc&ltU?)%l0Td
z_!=#1yJ=@08ueYt?;`YX8mhShKIgXZ$}+b_T-Aqks~mCW#PybdUZ&)Ul-pVI^7WgW
zC%Rjesq>QSDD;`N?!T$d_Afd98TA>r7A7d^aCUG5%Xsks!lf$go+-5nymfDkjYa7V
zRbBAHHgF2MM?=|E<Kp60XZlpXW=Y^&%Ux+PUUtJhA;f8}JS8WbI;@S{7knhQM<zb`
zVJ7S=DYTjZvVD#<JBNPvcAn!|qd>XUZElQRvX>Y3U5k0iuiS6P6OLzhF2JMr4Qw@1
z0v3L5k?#aW&-nQG$B!S+yAdK}6duQZu*UX$dvp^OZt%7ze$t%>yRL9Ai`Hss8aVk^
zeDTG*SF_Y{0SZVz998y5v5_6h%b`<FLrqP6^5k9rW~`J0->Zz2HKbyCaK_?|2#T1R
zmy`$F6#c9DK$&n|zC4X_7|c*gl3Td>a@_mrSZs?ihW{*?-SlE?F;aP|V7xO)DdQT|
zXuI2lwYS(Z$5ZM=QN3*oZP3SnZ9=6}H30ZN?&G5WKuA$rp;OzwrM$6VrCnb?wRP>K
zQTo{;3q5~R&QTqrK$f%5LkmrDq^c?qpe&zu5zyTRmduaqSLsi-#MtTbr5}^pWZaPO
z1^a~xatFz44!aGUnmfXA5F6A~21ZMrmb{556-(@Af~F5PV}q0b%{io=b3F)W);Ec6
zE0v36btXy#CKy+ojKS6>dW{y<RaLY??ykU0Gxd|GVD9Q(9hWX#Lb^O0%8Si1zT<yY
zlDxvDI&P`V`@1c!=Kf;V$4x;5RC1bX7M>LL-e0eJOuNEXn!fPit?k83GFD})E2?uD
zOc#+`d~v2LP=OXVCHr3Hd6CD2^^P<Po7w(i=z6$}gsMuFv!%mK&&NAVl2X~3^)Go+
z1=t)i?qSP$guQp_As0LQEcA<VjT(RCAh5SuI~h(aD(mR=tTn~**{!(SvoxbK`YP^M
zQBGM9`eCi`hm0o&+e0o64$J<cM>=jR%C&%)@orIT{Y^SBu)$*t^WivM7H`JT8@Dyh
zeoyh>ip>XRH8~WkUQe**`I&bN3fYI<A0|7}mU<bm>+-sVnj?d%Fk7#?M<e{DsKkt*
z>yAo)P#x;2s?KR4`_Ti~^|Y`6XC7S}^w=Gb59FIYlZL$XVe7%et&Yv;oh9T<Ma4pB
z>PpjxCKlBn#DwHVX&)wG<2sCxNF?-mfqQoZ`1?0bC-)`Sy=&Cb+1}J}?aa48ow$d*
zM=meKMw5;{^;|ihA2@|m7sN}q7VBLd{FD)-YDCVP-urY88Z=udl_62=mbRu!h0asz
zk?w5!sc=yS+_|%D6v&dM7{RHjsd~3&&^&#(q0m)Sh&t(nS({90>r%9qa_Y3`LEjcz
zh(C(NcDdQ=xrg<cEJlZi4<0`3zByCq`gb{5VnwS{^tMlwO|ogcpuoLN>nIf6H=z+0
zuI()OGFrcb_>dzVPD`LiY!P$Q?H90j*e04`uMo(n=DAdGo%FO??%_md6fnkMSy-ZU
z?j%g@BZv^NR?yKG1z!Y@GTe5DP)Vwb@8C=pIPI0WK&1KO*JK7%b$7EH)6>&my?QlQ
zv8z3M0FpI<F-{@#{Dp-aiq`_VE1(^Fy?G1!Akfu7b>D8b)SayliHzpDZSwvaVF(7l
zJN=>vYv@ZdFfa)1tjAyT*%Ky|d$*R`<ZE8iPT>eoh(_mjDa%+kzOMQnM~{`}*6k#8
z4X&9E2Q`lc1qX9zm##N(Ff%h#5gnEcVw9`{9o(Zw*#!X&4Kkjez7GvOU|RW_2Z3W~
zbTsjgKnsJ>s`aGfw`OEucmSOw=3vQ4&+B}AAVo3brZQhY5N>%vP_XP?iX!w)aedwN
zi@uAoE#@@Sw#J1r^jtej(VHYUMs)k>`I|kA8T9=zMUO623qz)cJ~gA{Vb9x1ZR$hf
zQ2#?|c8^13HUI$LeG%vv^9cyBIei0y4YVyyD1>~i=k@E?!`M`tq;@`z^u1!NZETcX
zxklK{Nnzl6wp2W5F*d(HVn?klX(}NlT*cn*a?i`Bxc0GOnZk!>wvL?vw@8sT)7{z7
zS*HR9K~KT}{?<XU7cgI=EYk`t*T1-&d3)%{e(d%$_1&-@z9S{|;<HlBgAQV1!VGkF
zT94&DS;G?v*9ebOP^e($6T7A^%0zrH?_xJRvBn5-;_+w6f;e0F`pK%^s2fh!ckcMI
z9v}yY{Jqy9rc010Na%060y#>`7se!tApP#mg0n_Z#1<^Bx3_odX&lP?V>tUQ<f8Z<
zeSUU~_2qrI{kIuf7Z!W$?)dxr-}XFQ>m2eFYiYl=Fmcp$czI8@%NlaygaXBHqsC%s
za}=1<&ze9;O{K0rmx%rixR_6Q=1hFo)R7Z3m)drYLv4HLw}9m&CPc>B`B$E48@7v6
z)8t18*M6&s49sj`X2BQbRRr;Q46m0sFM-O=&USKgI&hfid->Rp+k1_=%1{~C|0Mja
zs<E=<;^T7!T?}O4pCx@kx8P>{FrNz>lant=29=hTVcA)1(gmFgU=X4f*_H&L2x9~V
zfFkTV;e>s_Z`tHai<i_0<Y+^r3K@PK@e0bJ4kH7D_nw|2pbaXX4&Z<{KZh4LiP$QU
znwpvdk-t3>t^>INn1erVdn}w66aYqzZ<Grj_JE4&#6Q248c0-0v$MCKZA&5vcOHhB
z?f)$x@<kGk;n_NsqSjH+0pVf_EWhGs-U9~?B+19lj*eER(czwhpB;-N3Oc+Hk^#Ss
zss+;Qebq+48;<k3J6ju@t1~|}@YnFd&-dLHhYufyxY}2TYyST+jw*ATwOI@(3g$Y}
z$k+bI+!ohB`Ge)dM6ukp_x`9BoMxuV#0G-XZ-x<TO}{5*hJw_-xV${wn&2n1@^#-{
zySO25W@e_Qre<rK=Ze4d+bmwRj{-J4%dR;-ImxOc2-N1EEmd6St=L}eV^vPIo$Xie
z{2z0R348Mf?Yd!&XZZf;Ti?I3IDGJ6H?Cviy&tdYd(UZs94s%ppKn$`f#2po){m-4
zc>f+6FMZ!Y(Dx6g4SQMq+Ss_FS@f-^M?Dh%+}}fe=Pg8He$}yM*^Ym^o&7(iL?JCF
zr=4pg3)E52f8LSMgbNy|EpcMJw0rHWHmfnK{8uWFM9_1(SA^f>-*cN8XJ%oUsOP94
z1&IDHoCl2+Ow7$yJnYE<GMG&Vzr%PPm|8(GRBzP*)P<&P;-UYU+P!!bNSpvdeyOfb
zRo-hB!;!q$&J?WU1AHa*J7&li6oUk1v=Q|Oss5f;**NIg1OT&y>i)+uBcKwlEp4p^
zdFjlHb?kicZmEY{f_Hcb6**M29qbZk!NTueL%$|cM>lB7vZ!{ot`gFzsAF|`3FD&x
z;i2V8Hi{$oF5Lxqc=7gTkB=7BJys>Oze`UJlD%oRL!*pxt3oM|@z9A(Zm6vx$3D7s
z>v1xEKEw3Okw=(Us9n;-2E8j0bVigi)tZ>H<GmktyaVQl)ls@Z+w)SOnGWJH@5}@J
z4HV=cg1Z1d{`KbgQj@8^nM^;Y`JJR+%Kg$k=F>D<b*diuV>R_ECu9Szt9_K*hWFRd
z=musRgk=8W`u6tr!op9<kVPdWgJmwNSOsyiV!{hh5VuuNv?Ucj?78MR{~gK&5P+m;
z2c}a)a{=1Rso@N3;@}dI*9`A*#6KohufUiwN)8ZI*}5oa(d~|v%%wfZml9m4N83mr
zE+rS5Xy}a*LvJwh@$vEVJ3%^+N*k6)YU39W=<D;_R5}WE9t0*jIyy7ePwK+L!lI&I
zI4ZV+qN6boLqIdxS6?q(${YKDn<_Ua=izeWs8@J&bR$LXC<?PzBgMW68Ywd(9nudD
zy7l+{H1f<9+u9lx>g1$f3fWv6(Es|2zC>d#*7?!IiZjT)hmN>`OpM|TCjzksLWTPo
zs{n<GsknjW+L$ew+lh2`m>VFAZ`@B2AYP`;vGo=)#@l&g+%Vx|D(c#m8Zk|`-K|~P
ztr=^*7wpZMw#AvAo~jN=rApZ^`O&0jIXU}u?1F=Wa*Uf}zJC3>xi(j^^J_}P=Y$s8
zHYFO9yL;ZEtPxu(5|ZB6R!>!IoApH5tB&fX3ft;8HpL2Mb(rbv>1k?eLQ)Xg-CDu|
z{l{5_Exdl9nK|Tma6gES(`D`LS3#P4;rw~I*qfO5lELoVzYZNb6bh1F&e;5Syne)R
zFKDQ+v_Qct)BXw%owSbGp4^K(Jo#X(p#KI*OS=iTSnu<dLUrrEK_3m`hci$tQ&Ue3
z4GqoAfY_)GV9Qw-OO-xmiNCik=H%pn_?G_EsYg(BK=xRf8x&$>{L<4?JnT}C%4_%(
z;w}UqEYO|$${iFG)G#D-|Neal+TgfY&2zcW2XEdHY<R}SsIO8%?*ZQno<IKs@Iy>Y
z47Aq3_tG?rb7TM)UVuD~mE-UOfT;fR1-eRp<Q$N&asVPg9eG7LwcSrh{hBZG*%P3?
z<3N-eQpXMa;e`tq)~0(P>ANhCpCTn~&ofmz^m3C#6!E)b&B72|Yr8mH>+9<aYJL|d
zr;YXXhrM|zdQ~s1X{NkRVvZibk===o%g$~OAEv9Mm01l52Q$mMy1GKs<=rhlCC=Z2
zVOM(VS^y9>;_wj@g98<SCa|b1fQsiwjxqb+!A8CO#eYKjgw1<N!>^ceKYwR?(*q=F
zkk7_67|4IWg|KaKY+ha-wDiQlEoy2UKqjtJ;STDWNKa4Y-Q#fB0k<lkuv7_=xp4F5
z&GYB=pv~o<tzG6U8+)@1BD8f_HLThLe4~J~n&@!l>kATj_h*3NG4Dw2fMX{R!$YoU
zsMHb`12u>nZ)AKFm|)<S1AKkqRDqn&3EBbIrN$2*I8i(Ce{O<@uG%I(0-Y5s9Za#O
zuWxOpZ!Nd^Cbs;%mmUD23Ae_lo+F`O^z+9L&>G7}@hKD{wdm<%vBg8_E(i2T4ny-n
za(bMZhlhuiRVuAx0Bx5BCgOR7L_5de`AC`yw#@|q@9{;3tfzqUlRM1f<Kv}6Sx3M)
zO<Llpc+8TuLQcctij>wjp8xcAOeSKS`QpVfzyRRb?I}vI{EG_<l+Yv-%DSI`6S)08
zcA^o8+}T=c1}BA2NJt3C4BY2G9RXVEgKGyXPE~gX??yf2%2t}1Q7UNx0ReV4HcP9k
z+U2fJpHdF}=cc>1!O_v%Q&|<_qM}jI8HB_Jy8<#5c6N5JjmvxQRrkPlS{|M-rUONM
zHMONb4S$Z9-`TJsnNN6O58l-pu4g@ElDrNQX#;~{u#)<4j=7l`S~@!I1gTKHxmUlH
z@Hi(*XGL#}S<r#9gsZ{~%V=xLnf+%hsyI}M!`z$7NC4F_f-b){HxcHY{&Ng_Z<5JK
z(*>7+<`C>F#4M=gQo!h_c^|&YqWCT4rzQj~wh|w@wxOZF&{79xR^><D2k05h337)3
zKYy*8=6~*GMi<*nslbi!SZf;_tTWQk0E?in9=J}l_s-xs)^3AqK%Ho&b<o{(Ei&s!
z-QHXq%rTVqA|x}ouKAzE@prKee*HSzVNN?k#Rb{|kcT(d*BbyVwzRzN+VT5yP5`$$
zj$#NrMajxqXTrTY*<oH6!U8e>yCMTQUap6)6jXxJbmLp#FpmNZ3P&KozSh^*!Tu3g
z@UBM3@t>Rhab4?CD`p0Ucd%Np+aSWRvHCDsLG|~w(w*OQ$9<MSfXE6qMmfG^d7{l2
z&^stW6y6H6?_K@$?`gW8pRC}t4;W;;L||u%QW8WsP`~9>DG1|qkJ#f=08*L*u)X4W
zKq(LlVOQ-Z+fAWu10wRgps^uRlD)PjNQu!bcjZ5S{`~dpy)7**eSI#Uo*n`qaj)do
z&Hp?e0xEcWt6Ba<`Iv{_-aZKo46LZAaNOmH+v7HY;3<MGtMFVvXW%k$D9IM;XFwEY
zT}<SEUz>AWdU$JN2}ogPW}w5W;HhvIm@TVDLHhncG9~goa}*1Uj&}S0@d+3m&?wxd
ztq|{wA#{4VcBFFdnIcDed;2{bJ}meLKtlpg28<S#bw589Ayd0&zkiYQxvWC826tTn
zA{_6oaO?4PXLHBOko|jx^mqw;1{j%nes57x5u`Fm0dO)1YIO3^+&w2Ru<3MHN=gb=
z2Ph&*Y3ZqfvQ0R5)EXz|7dML+96g70HtDI2OJ!PM!F*P4g}n$Nl|c(-WpR;p^~9+^
zXX4IXA762Hv$e%Kw%2iT=&bgD`@jJ$MOY%&=Knl+k;GY~Kj`LxFztq*%kDKPO5M++
zcUPS4F)#cf$2Df%@0ePFFan1&Aa}70is0MDaD49E?pp^F-UrR@StQ!AKlnW6mJ8(R
z<dv%Hm3CMxuSLqm+@OREzCQnegM$O2GH9%wK`3R*hnLv^kwjW7F_<!{J9sm+u0k_w
zM`==d>5?uDWZ?f#-*Wjgv5w9l^ze{<1v4G0Iv{eI1DO;CtQh2e<cmM6x?v;k$jXN5
zH>>0;BuJ5&UlN%m77!4C|8^3)xVV5^uA}~yi%3!%pcru+|Dj{^RZP*l_LVs&*~A$j
z6>b<l0JyO<)*Q}QYjDlb(6G$vCwRTrt{ywKYZ@6DNl1ThBcLXbUIWXsfPJcoQ05>(
zue!W!iya)BcGcC@Q^OTLdS_)>g2IoHB!E%fd(8UUd!{54R^>+=+IiVfD1(`Kcz84^
zFD#A4dMG+}Hp2U`Y`qN<Mq!Wg(F>j38Ow$zHlRHiK_d%C-@bu_xjx$_fj-LwU1HiT
z(KpMs*(fF%%Bp<Yy^f2jUI7j?PNbu=M_D)!4`Sof&ce|GZmJGAFbM$IEd1?3IF=S5
z9xayyXF2Pen!*qV3z1SJ&H*+Vs}|-*w}LIpwB;XsGd3_t)_QsxRDIl3aEh>3`71jD
z(N#L{&75Ev(R$C_eBhy1V9KR`hfpq}72v!k9CG?|GU_+p^2MLbJohV|dgCr27w+Fj
zf=Iu^8J54>zvm#<U-ISmL8QL~3HG>^_kL=D__g_YtawIVX!|`JIz<CcvbQgYj{ZJ)
z8)oX~=LeI0DU4zcW<elS*T>dIM@E!W6ca!jV@4j71tE@fv8ui5?Yc;#C^;$V@<I*0
zjYsVB{E5ztzO0I!RNN8QQ8>?20n98%`MNNie$97UGOCWq%F3#<OBS6!BTqo^@&EU%
zF|~%a>Gt&W^fMieS_&Jha|;|k!YI+5s@^{GxtWBg9~Y&X8aPm0CfwL_MIhV+<uLwS
z5tC$HLxU~;L~UfO$E_ir*k&)>N#_NQ9s&a2|J~7Mxh?A5_Mjy`rkbiYt@cf>93<`5
z)gPikiDvaT51_kP;_dmjC>4LWKj*Wpf2X~MaN)hoHhD||WCdD6^yo;9canko+^CNC
zB}Rs<j6Z3~h-0gIcX4$v-Atx-kl3>GQB%r|{Nt}BW_!E@``q#-^qh@8uRWpd4IGSR
z?$&ZhbQvVEoXVYdT*QpampBmi#Imb-7$rX2B{mbZ+IW6My|$%oz+__6Ek|-D$iTo?
zN8s(d=E;K#gG(Elr>%@jw>utB8fs2Hqp4LcAyps#vR>BjxcK}b5UiCeRav2KHuVRB
zd&L$Eiq_T5vls>ixg~=ZX3+y=E*YC5$vX{;O-&{$+wos*;R|isjobAOTU*`=GjYQ~
zHbYAYR{oN_<0WQyOGuT$<OD>voOe4y_j`ApT4j(r9PWLfn#fH1#ntL*HkFk0+NRT!
ztzuG}GwUm1W$4fjP9xO>muBZj67plIS|LQ!W~)c3R^PCJRHVPGRMR>q$E(!(!R5{6
zxW}=5he$HdtzSr9^ts)vK3hfRj+)|lGngAyw411@eO&V@I^c|c!z)rA{o2|SbE^xa
zDH?5_LQ2w3D%KsG)e%}%Qe?S1GA+*$`)S{t&^g_j5#@cgtup@jn3xLnvWg&syqvAX
z*um;mZOYzWvG*PcVj>G;r(GpUnoQQTq!6Yw=63E$BeBP~-A!^Yqy08l)bfx2lxo#_
z`I0YMPj6MZk7Qh>gjB!vCl^Y1J8|X_eVIj{L`T*22d@}chkr^iF-Ja=&Em>(NA!7o
z?o5ATUJ?CG^y8y*2{W<0v%_y2UEYLVPq)9qdU|U2t=X#_ddBLSr_8go1Ox~CvA+vY
zQ&V%XzY8;XQ-7l}TxK(@A*N*Z+xC$f-lp9}s}~eWPn`ni27IwO{&fh_;PO~qHG{=q
z?>(~yjz<*@<E-ryH}o5t+S2sw%~G}X{I4pR)m65}g!UhtdVRGz;_5<83oV-Cj@wvY
zu1{E*y4J_5&Hi0oZs>rr<?)h3xk7!`tTV{z)C~=J<5aECMbWJi<uHu2;QqulPRpsL
z9fx1@q8IcVhJP3&J!r~olGAf#>j-g~i8syKUZJl}^OI<5j9#4D2{aQ(y&o#WJ!^5j
z9fjWgFk4}XmSpj9FY7GzVA~g7D^;r#Na?U5<g#Pgn#OL{b)LG!I(jRyHz|TbU))$C
zpe)!gEjQkGFi(e|IX>KHXxhJ1WO-;T-1ej;Db>&L(gD}_@Ij50-RrVe$q1{udEWCG
z3E{Kb^Mfygl?Ev>V&n@WDP0JO=9yvkUfWl6t{oMl4}Z?Dz1dws80%!Lsd)^sH}8Z}
zPpLZGZyeu6T^^M`PiQ1$-g!CS{A6%-Q-%?$Am0l`;(t|3Jkh)-=j90xo5hFgy=ls#
zzg{LjvKZzXUdbWZ{hE5!s@M2}hV#+&!mrEM-Ao+<TX%PO!ZO+t4eg$I=%%!Eho*ID
zkDp!g;==yK!t>YX4iMZb|G)Oj3u<3)l$(65LH!Ky&yFezrC-l12}!b-Bw%`cn?(1W
z4C+ZVvN5`7Ca`pBN?3)ebPze7jJ^8yHKXJV6+ycE?8YI?VY(FcDKyfgLB3L_n(|cl
zH6rK7V{LOJCc|#F>$QVM<&$siuMjXjJwx_*>~(A4w(ZAzR}H%yjB`6O550W`uNoub
z@o?Z-<v`zw%gOmX1}kKqy(0u6M-K}=_KM<?e(I1PBb2Wcy+-ceksdLx90xItK#G7;
z@MG@#W$Fs8*OP@QKi@sLjM(U}6=EweRml(OHQYx^j{6A<I7st<>L)8^clXI&)oI@9
UZvPCwn?WEcCMTLHto!tT0seX!X8-^I

literal 7242
zcmd6Mc|4T;*S~C$T}0NqB#JDlWErySRtyHoGIWzQ5m|;A`@SVbvW2o-V;8c_D6&K-
z8bg-RsAOyn#xNP4Yuw-8@2}^-=lSFLV_vW8x<2P~&igE%^Eu~CGSu9J=a|?rCMG7H
z>!yZQOiV|17(Wgc@JrS<-%TbaW-9as%m{ova^wgzGcyYd3o9!t8yg!tJNwb2M>#k+
zI5|0w9XrOw#l_9d&BMdP%gf8h$H&jlFCZWwC@6UR_;Dd2Az@+R6DLlHh=_=aii(Me
ziHnO%NJvOZN=ivdNlQ!r_19mgPoF+>=8UYY?Afzt<>ch#<>k+vJ9qy4c?AUpMMcF6
z7cM9%DJd%}tEi}`s;XYRcu`GF?b4-7moHzwa^;G;y86|tS2Z*=G&MD~w6wIfwIL9Q
zj*gD5uI{yK*Yx!C^!4=(3=9kn4ULS9jE#*=OiWBoO|M_S{`cR1o0*y2xN+m=&70=t
z<`xzfP$<;W($dPx%G%l*27}qy*x1_I+S%FNx^>Im-v0LO+YSy6j*gB_PEO9w&Tu%~
z#l^+d)%DJuJOBLikDHsDySw|{yLUZ2JUl%;y}Z1<y}f;We0+U<@7=rS=jV6-{(XOc
z|A2si2M-<u1_lNN1wDNDFgQ3kBqSs>G&C$M41qv|hlfW*L?Dq!6bcm?85tE76&)QN
z6B82~8ygoF7at#=kdTm=n3$B5l$@NLl9G~|nwplDmY$xTk&%&^nfd6^qpYl~$B!Rp
zXJ<co^5p5$r#U$}xw*N|o;}OU%X|L(d47I=K|w)bVPR2GQE_qcix)3SN=iygOUug2
zUcP+!>eZ|A^74v`ipt8$s;Vjg0MKZ3b#--3O-*fWZCzbmeSLjHLqlU@V^dSp>({TF
zo10r&T3TCM-@JL#*4EbE-rmvC(b?IF!C<ghY*$y;+qZAuy?fW)-QCmE^Zxz&-rnB6
zzP|qc{(*sk4<9}Z4h{|t4SoFhad>!mWMpJ?boA4wPoF=39vd4QA0Pkn<qHmno0you
z<MES|lT%YuU%!5xo}QkWnVFrPotv8@5D4@0^WVOGBNB-V3k%=BfB*60$IqWX7Z(?Q
z{ra`Ev_vA2mY0`TR#sM5SJ&3o*4Nj6|Nc!TlQ%XtHa9o7wzen~%J%j)l}g>&+1cIQ
zrO{}A{`}e7+oRLz`}_L`2L~jgw>=p8oc^ZvflN$Xf{fphv02m}6O-ufbwfQ^Xu)b>
zScH&GOuw4e+Ms;<)gJD&{*_9to&?Kx@s8K78X0QD8xonAj~KG^@f(`BK}}BF6%r6O
z;^+JRn<FVI!64CJJ~2sNK0)i$HB<e8!XUhK{@M(k^svj00+4Jl1gf>Psk058_A2s2
zZfS>=@T()T$qx|BZ%;xyOJ9T<>5)@nh{9~W3dd6HgoNtk<()16Z!X*riu99x^o*)d
zeIC^d;fn5jf8b_M=1N<6*f6*wBPhi2QBchGp{2_$uc|DJ@paq1yax-Jrv6qgIiKn`
z)R=pe990@~eS2B*AJv3b7Kv9DF##1a4&=9I-4vvNMRq_ginKIft6k5j+FPuAy8i`2
z(;12+SglqZq|<g+HA)oz{Q}FHK#q}D&czzoU87jZl<u5(irb*wq?;nVA5Fdl8Wq=M
z*Pmoz<jsoSR054<%yPCxa{hPvQ|+&GsB|q2sEqe|c&}c(ygJWAkqPUFJ;V2D@70W%
zc?q~>=z4(V^Mhw^lb<ef6x>>4%NdpmuUQ)4XzC5YmWQS4e4QY4inbHql?YJTS{hm-
z2cn|NXk)Q(Z!MkE3F*R_WeC4p8KfbiLFUbxKargc!~obR&yXek%c~NZMq$2c7Z2S#
zrY*!0gx_g2_L5GjXg0$V`2#@l#ubgT4&-avTKbv4SU@AdMsJ?M<v!G`Q9JKIhC<Y{
zU<bc1Vl1apj+Rd8Tno8%$fO&My_yqBxRm7EYe1Qav!lljQTp8#q>|jHuMHlXN=YQO
zaaJIn|N6nO@7V&9T<y!6teVij9;dmH&L1w9FgW;~_|G${`#jAV#QDBqw@5_+WCScG
zl+#L|`8$KVK#x5q^><18DKkJ!UQ%4~KHT=%4r#f!TOa;`0XiK^vexD9Vt85AS3=5_
z<zEtOdVIz=BV{;469gcAb3ew61C(P}?`rt|=Ari!U`V|i&hIh*d7&ehPp`{@v7ud@
zerNJulj+-xo<{5*DrH0GVE;U1r>!#dRrN;xq2!y^IFX=_9$(E4E%O}MpQHFPFsbzS
z|8r9P$)TP5fYD54x0!!Q#L%co_6df-XRgDcKW`qYTdqfGQ>`w|tz)=_9VZvrhh^xp
z@DlR2q2bWVU#j*Uxh;VIJnMvUjWR5j_IO*QcKBZw#*~QvFc8D>?Z(_^*sCve{nUu$
zp>k(m(3(^=sYkd{4|%Y{bg#ySFldZJsRC<GhcxA{w|w<Esm?Je1an?JN)Rg=_1ktI
z!zxTDkcc6z5?9Z$2VZpBdwKUr>GR8I=?~_77k?{t)g^_WQ@qqD;{HKI<Oj7-;A+F5
ze)MrIQEN9B<#B4Ef_XZjI6KhZyyul#t2xgDpAdK$FZFSuyPNXh1AWIQb$j(-;NRsc
zm|PW4TX}lNnSw6E>r+O^KQ9?{B=4y22dy`Br}SGw<m>}>M`nz*qYhpPJ>WM1;Ys26
zXk#VQN!HP%RTpwltkxhS7>sNk&y8c;uOF&J9I2aJyI^+OfgF`+WFb0d!|)TYc`utH
zAMSXlBV`}k!5pba0mxso&j#l(+?BUb=)J-LF5af>YX##Ym~iM`1&h3|PK=$#84rO<
zItL*4yPiZ3S*iNN57;H=-p%7a%q__>PUn|(6G)ueXau~D&QpAiLi$ZqE0<wF@t8NI
z3m};)i{6$2R0^$yTmi+L@BCd5#m@*<na5LYr5_fdkZ3uZ_2sY^OH8mj_Jcj$Q>vhw
zaVGPLwI!qcFpGIT+#f`Qh$brz{kEVjCWgV7Z>YrY`z8yB*qowSnn<#7+1~6fJzVM0
zll*Kz(G0yo!4~O~cnm1ci)|vT35>|K`0-~=XEp$YV-3{Z(5~)c2B=poMn%Nr7zaOo
zJ^ar(pBD&bzA9`Zm(dm!)0)?O&Hn`>cG@EhgdwoHSPvL{O7_N0C?+J24E#1_Xz_4T
z#2BRIm9j*-;}^c!f&zbhz+d2a$}sHw@^nwK6o7ws98k94P(W19p|$rDWE&U?Cf2uF
zgdRe@Msep9kQM=`r!y}!$TJcS?d-~uNSL^Z4g5MIC~w)oPmKPq<gL>2fDyYP>U*8z
z4WERem@p~iP^0H~FG1Dc9Zyiw)oHl@>sVzGZy70M+hc0*6s~mu-!kL4nh(ciR(ndn
zv~jfYI!k>_$z~*A&HziBR|KwtTh~vF%k)XPPp$Mv$OM{RaoB4<+Ps;6>D~c_aSt@z
zK=pU~kL!}TZnLd9VSi^fF)o9kXRxl=>sup|IO$tX^!^#XqyP!rxT^-Byd3oW0=M#f
zXlf#%>5?-;m)k9TE<8Ox6jmi$WPq>v%bTPq66XC&;S;)14BP6p@VN2Ck~kgF2wu<}
z&Uc7^4@v7YHhSw=5ILJ}QDn15n2xal6~~<^MhzZe-HI^e?$GRWc!1Zm$w~O%z!5-w
zkdF2Z<<BAJo*kJgLJLQ3ny-6F5FDY9!g_$9ou>Rz6W0ZY_eqiNFJ^=2yQ{GB31(Xp
zQn*Noz3lZM{>nMgg#l_ubyUqxaICo(;Y`r{-wC9pI7@n;(42}wCJD)fQr{xUM)Xsl
z5Xp8@Dsgl(!JfuE(zx8-Iy;!ljT3kMn0N)k??84E#Nx1?Y6S`>feHf~dY{0at1W;E
zc)Rf<LR|&q`3V+z>TISii0D3Wu%!~Ptu0wuRBTy1NriIvc6|yYmmC|8d@wstZd>^P
z%1R;OGcO6=Eg>oEAUC_<)Ks6oIKmay%AyP`iAz(AAA`F;l;c5m&gKIa-M?#!OLN~J
z(iA91)0)fD<L$#>#ZVTzi!cil3ksX$Cita_R;U%Sc5BQ_8~i-lUyPro`AY!Gn8ysf
zWo5KX>+}SBRO63ipEr+75SlMZn+rUgT2W5{zC9LcYU(*h5ES1zabMK67NQ#S%Ay&l
z5<dV}8wnL1LCg84lLG%amY!%|LCarem~g6X(Ry0?@v9x=Lw{KVN{~LR3lWp2;$s7U
zHk{9m8}|Yb)}^i6mHT^z2PZH!3jN`YVz5#3u@gt7p49Ag$axsq>C9y_u#Fp1yp9hu
z6tK0q{n|EIn4wlh18KBZv?DJ9VhikBIG|Gr6A2)($3MSBGPD&P$PW8Wc0;N)IzJiS
zyktdBxqi1t5d!E^`>g8<<{8*8RnEQsb-@N{A6|&IHem)lT$9vFJ?P2BI*HWQg%X<o
z7c^?B=Vwnf3`zW9*PNV@amWk2oP70@cNw9v3SNE}*zX0dl<dwGhR4Z(Sgp~tkh7^A
zO$+zzi$c;tGhPNpGmig-2v-WZizJ^{P1>Yp#)Ap^SQTmsrs@i&z%L+y8cdWEwywk;
zx%8`N$hPw$0N;34siI(W08LQdq3U7od80NHL!H6e?1x9AVe;H<oUXtu!m)jSneu0P
zZ-wgh+}5}tlM9!)P+KQa#DpgCjo$Rsh-e-O@tS!ObD>5pPPi_K1V=eKK5I6i7@_<3
zX?&a)P6#+S+S7Hjago)Yun4fSYe}|ox1TYa;FJls%sEah1qM<4R~w4`Ag7g9J+WDs
z1XbaKh*1T^$k3P9m@e6topBFbBzNJb%AL+xy~jsA_pKD$%?Nc8#XTJ5L=zM=klEi|
ztf9AkD7+BAq4heEG!;g#D<8#r#ytfHCwKmVhs@qI$J)L05=JNGmiWUg7GSxlX{`bt
z-GgN`G~RoP>ZBU0aMFY#w1*7Qg0Q7G@30(PCZ=TBM}V;yZ|6}zXI#KzmPxYM=Nok(
z>%BlA!7Iz9y;V_dq6!;%*{RfTc{DU+;r0uJ(_YBHUx*gsIEfLP55sovi!I!HZ{M(g
z&mODbZ9qXt)SQ4!-+Q0euu-`W^|rDadpiLJJcf54TOjhJfxCsMl@|G8*@1KY0T|4h
zA0st)a(!TgPQ}*PoJ<9W=dw*^!wv<tmh^RAPyBb=*F;bXX9UJg0OCX5d|nv#(y3L7
zae&@+cHvR;VmKd@A%)!G#Xn;uG-yUpZ6$r+GSIqCG?D7VZ$^z<d&c>S=x)@Lii@2N
zN_=4V3B8$}Xxhg$9>RF7B?bwuDPpBACOczOrNe9~fs2=D@BiHo_CO`I>0M<`Wr#mY
zsPC2$5Nsgg*r|cwsRx+=YDD*W1W6D#9X9pgKCti7T22^MvcweTUZ9=<?@Uby8?z%z
ze?om%7yre1L7aLNmyGB)7j^hRk{OJ^q?G0oHE6hN2$q;qFnW`+paB@YRM-KM3I((G
zdFa%S`f=+|MlD201<jr-h~4QWP3ENT?m7J15+>ZyowZ=hGOELvqWijoRGy<Q$*@p!
z)UO`hu90`P5qpI5&433H%se#jB7{-M<Yl1iqQ7RSjajW*rSteUWuog{ENJlkSd#A8
z+IPm@!I3<3i|Atx<k(Wpnb}jMmM2dn!NiH#{_gWc81#hJL;nI#F5I}``e~fcbKTqs
z`Z?`EH1;(4&A2rOiv!ssTlwC$w86*&S%o^9+yGi{;u}~@iiR5xN$NgQuz!eK$03)>
z+3}V8cF~&WQNvHG-;jnDG%U=UPqq;6PukG?=fv9{3)|7RdR=YXjVTIsVwVHAl<oXw
z(KR#-vU@gv{4#*&9WB((-P9I*1>WS5gG=WOnHF&lf>&0D>(QW(WGbs^5P@2@B`KWv
zHCJ42FBd(hiFGQ48b_+=lyqeQD;6$*AV4ndf?`C@B`;u|AAy}q=TWdzX-vw$MqypE
zqi-L{YK<mgOafNCDk|Q_`Sx}=Q}~oTCI`5HQB9Z`-82~SC^}>pX#_T4;2}P2z)%Fs
z^5K??NSwO{vs`fb+W}`^Y9Uuo=g-{NyjHr654|5bbo?0%ggGenMJ(?~Oy&rU0~VPR
z(sovqNc#^@ez4(uo>EDf=ulX3*2kRqGQ@pL37JNy57@)lq4%S0((H?%5FQGw>xa~Y
zV&$BfP@C!2hVc3wTxMjk1{WabUq@1ZRqvrknJbas?;N+D2d{l6G_66}-%XFg>Su%0
z{Nb*Me1WL*FDdI_M4XZhT@<Tz&j8dI2Hva@_E3!a8~*IYEul9InZQ2Q;%XCE!l?tO
zS<^XCB$rT7P%+q*&BX)K4H;StnZ+c0#?R3h(yWDEe?&!v3&&JSNFnK<bMD#+u-LqU
zi`1SlJpo$}-dNx&EQ--{Q<!rHET&xK8GCa{_%!t#He5&kjA(0$GnVM|iuT=t=Pm=S
z6`jOs3<_6YxC=H+ovE<mt+sE9pz9!jMZTvt7Z9Snc*k)6%hycMH^vl$E|n}c-0?R?
zlp*O$*&s{d)>H<=QrfAzhJ3i?d0w1y_@pC<#C+uJldBN6anR!mFkR354-BR}KxBBz
z6sTdkTYnpzVO3@v-=In%$a0{Lre2Z9(9p!TvqV~D2`to?YzXTLN9@6)x&pwpoeV*?
zR~m_(r@$^R*O3yE!q2E~XMjk0-JQl4h>(EOdf!`LHGvX80#xd(EJOK6_7pCUq`{38
z5*J-Fw#b^^U_K{+i6WX6wUNgv&HSES*FNWLjog>)h<6XuqsaX6_-^y_p0wuqpm;p&
zVC6=DJvd_+dQO!`*P^BEKkJyH-vDDb>J-Csc-L-|_Aaq1eYl(2^`mSao<AL5`#XcA
z2LDyKGjl5Iz%NvfqE&ZKD=27e7{9{l^o<9XTVatRx%<vY)s?483W~9PiuK&~$sdr0
z@k!u{k(+BHnd@5`5FWEfs|VzbO4LEqEDal>552|V421}*qp?9fag(ZEMt8b{e#eo#
z-q=z!QrUxx8qC%&pHn{NKsHCC)~ny6xSCkwEu%@C1u*1?uMG_gIwgf#JUBJtMV!Zl
zmE5b#J{2CW(v8CFNhEql(Teu3bO}{gMesOoqp|Y$f>z>q-0yUYMw1GE?hGM2ARUAy
z`>WBuz0;KlLdH_znlr-uhNv{?5+^{}Ia@5Z{cPiWUiTZ2{@cKw4&-j+;+A3om`E?!
zA#M7;#JvpRnPC<MJPeI-BGsMh+gZa;?{wc_!cBkp7C9F@t&0jmoM`;ZfqbG0yRrEp
zpeHB4-BJ`bnxqZrb}4Fn`y81^&xj$N%rc}T{@hxDD334M4<!2ba@o+^()oIM;O@VU
z#;^h&@4axv;XbB`OVhoDDn~(OD@lCXc+99-zeGXQ78nssxLl;BYVa46=;J?E+S1<~
z0nb|#pxYD*U6ktrAg&`h(@Ek-0FfbVMaLWk_|$>{+i!7O<Wf7ip~@#EKMibXz(e(n
z)C6BAxc`(Qrg?(JtZ(n;0s+!A6oqeE4O3YM^(Z*BO5g$F`7&7XGzyQF(X#92@^(IA
z{l!J|q9t9S7}p56&gZb-f?1=*cfd(bHYrs{LRgL5G~jnQ-->X<Nh7N)G8cJO;tz%D
zXk+hX%5sNk*+tYN25cilaU!*1Qhz>`AInfV#4rvk)~E{^)i|orb11XU1lJ{%-sNFm
zRC$#jEO<F_i^>+s`Fm{*YNcRR*C<SYQFK49KAhf6<gqw%IAi01`Jv8i&pp+T^weW%
z6o#W|Ga-)!%iW(r#E%{l9Kl8EcJ!}qW{O3AM;uBa_;XsioU;;++#Y@xFwW+zF9Z@=
zkT2)*KV2Biu}##YG(V~B(hdGdA`XZ%Tbe@m+Fph9o6p`3?=Bd(Yk4il4dAi7ZE^D8
zROk;3-Wcq-@}~K<SrW5ZK=sbNtRYpf2b=~eSz#jaX-X_2g%b{JxWI)+q$XaRV0(;h
zM8H?kat5`k)81j~7i`yW-AT&jk0-SS{7qT<Oj|HyfPskX@BoW_Rb{lna0B)+>)6*f
ztIrR4zvj*ia|yeiEuhG&)bn0?=Eo}U#bx_p_*YMjxh?BKt6LvmpPaV(OX=^;;^S8v
z0B}yCAb+_L&vB}^?#oGshNy#Zc0fHUTY28GPVH`WsLS!j!|7f@#1W6I5mA^$zZ!sN
z0dmC-=UZ8;%?Pl;h4B?IIz)sY7<-3~`+t+pDOlIHaw&GtSz<2H^Q?-Q=dIKs)&!@3
zvMf1TS5tS3M``FkH-y5<@P^U$`aRngLZU5h$#3g5p`Eg>@<mL&fw4qRXs7G{e_co?
z$~8DuKS`HjZi-8ge0D}<rKXiN0p}V{7D=AI@$3Z@9J#7=4T$&;8es>W5?6a!G_L)(
zQ|7DCh=w<I%@eah?t86wO0mB94IS59FTtyIOw7d%2nl(e=Df<6*}0L3fWO60S=!)H
zx5m~_p0d<m6gHeun~0+K)@6unyLaNv#*d+Lluw_>)Reo=$e+G`T=S3vyRY(o&=$%S
zZu~jy=JZDRDa+IZ@7r^)^h@JL%#Qy_yX+D|-k<XcG(??5J5+Ff$b;M-KVSA9V<F@1
zOe}7=y`qTvOu^ShcvYYMIm8j?#dbY%fM1~8t4`MD;63~_sn8%`bL^7q=q>A&{bUQT
zxxL29pVPf@HbI@mi`#MECPP!p?{(=xpS=y^MD>aJ&sRFnP}hNNcN|AD(2fYiead4X
zQAiR&5ie)K`(9Tv-9fS3rCz|#%8}xZW^Epu&mu;1|1S*b;XOpdVZ2|>cvHQA5Ik1b
zbt*>B+@jkY8gYT(Ozp5XAAJWtOU@>0LRcL1$juR-w@LR4ied|2?1&9c`#L?I-#W9x
t+KX55dGm9<{>zg6B**{Y--3PVD%5=6DXtwp@ShQ;>qh2=wfZhG{{?iPi4y<-

diff --git a/docs/dev/reference/figures/op-relu-plot-1.png b/docs/dev/reference/figures/op-relu-plot-1.png
index d246197c26637e35b034f238a7ffca0eb6d46ec5..d867ae184c9e2e76ef838daca506d88af54a8589 100644
GIT binary patch
literal 18241
zcmeIabx@V<+cvsTL==z^=~56>N<u&yr5lt^=|)1jB&0z?5Ge%|B&9)0x<LelB@NQu
z4f|aF-uInvW`DZpn?3u#y=K(s(Z#y&`?}6Jj^jM86|5*PfpeAYDguGPk&+ZuMj+4*
zP`@xQz~9`JHoF6VTzM|3;ebG35ukpd{TTB)K_F-lQlcVHU6NME-1(m#o?qVZi&B!|
ztE3zJ^j4YRhEANOl4MApw(1D|gmFmp_nN`MIE!uODlL4yJj(@Qt;*QZM5K;&^xI&P
zcpNdwma(q&$i1rLv9;BfgonDlLT(HEU0s%@t(}?s9&;3Tqa;P)HeN`@P{JQa#357&
zpJ~M%1cERy?K0v$51{}8(fnZn4RPn17%oEC?v5J*5%Re2B0`bk|9@-v3S#|JVIdm_
zN6ZQ#0<oS@QnFKI+-YQNd_2>0J9mMDojp1_TJZ4q^&&q6qFO=dWW%qvw)WE}6YtSV
z{~H1#c;e#XHa0d!nFxeSs@LIc{8^Pl*WVW#!(u30QL&S8T7{c`H9F0kC*(7qbwiLB
zk_BR{m((Y1J+HEQ#CT?3$T|E6S!Om3<BHkmoXPU=Q%CP%=iJ4s%BR!_gudePej2er
z7_wwDGxBj|yqrYm>Hhl-^Thz>y`HmvTdI`s00D$>-O@g(=lbbL_1zq9ri0?yE55H)
z-&TwoKCH<&2+pdbrbhUn4-O8xy1GVE3(@moieL-(_0xOvC%aor9P)*ac~e|}!LNVW
z+f$dvp(XI`C<0L$BzlV|bar($IWBH)auSb_P}C{yjCxEm{9tEd>IQiYQMk^CfI)1C
zZ*JChdt)Fa?Y!qAb*Y@*R?jyCVnY)jAAhDLoX=sdtGMnL63MLoMSp4FGo3h@&!;!4
z33a#k;uPE;8;iYotQE<b{6X+pgIP0o_gsH}|IExxDxdn(rx|&9V|r3(h_k|^q$El{
zr<2X0H;ig?v$IOI3#n05+3e4<JN(P*>bnv+NU>WSF4rB-C1xt5)(+{z=HkS68W$t?
z_V&Jh{aWR`{`1EV3oEOqN=i4m-yHPFlVG>NPo?tl*#B<(`juCC3-()APA)t#aU_aX
zLOzM7(4?Ep)O6nOGHy##lXj(jz^MMI-cG|;a<=%jvKk}z@U8_{`JYc-?YUn-_`Jj;
zCN4MbB(k)$ynXwlkB`r%PoLm^MI|I~sRfh%{G>AHV%dn0#|``A#>1Z+D(BAexYkSl
zH)K9xH@A=@-uq8a8@v>LpPkL?yf)g@)D#vLM$T;$MlF=GqJ2?m=bnScM9Y=+#MqTE
zZ1(!rbc9bdm(8fUj*gC&*2wqocqAkwl$4%R4VU=%_=+6xn}m}GWZed)<!Xs$o?|>f
zyw3;=!!|cx+}YV_k6{=Z929WfGWwaTfsN7F)a1WnhR&vM+ahhr^voQ)>oFPvyCjj@
zR#Q`x03Dx%L|sMY#p<up>gsA(i|FKKXLX~Ch!fh|w{Oq)rtWWTv2$?9$jVZ1+a$ev
z_fFk&TEe4T?E)gGE-sFUo4Y(H2$LwZ)V%-GmoHynBih>9{Qc33(oe52x*>e(SQ!~R
zdwOg-V(${O>KYXznVFe0<&#6g!zmwF5F-1|Am<R@CY+al=0!&nh>3~0yB{|O;tmZA
zto$mqTp2Djnm@wIh5S4Fl&4j`b?70OI)Ox*+t}#p=}{997(9C>;*>v+s)lC~$NTGQ
zs;c6mqAR17kmt%$Qs%G#V`F1Es(H8ger=;_sJDcaRPcunA0+8C3k}3nRewS#l$4ZQ
zqZX8P%8&mRMunJ6-gK#k{mstK&dA7!6goTj{^2p~0xXu%yxT*lzTc`AS65%Wco7~F
zvbDXP$Z7tgKT`qT;mWToDvFNm)4i5Xj_`4AoVABY&HwNLpNa}1-@mM^Y-MEyi<rgC
z!Xlb-bry9y>-4OwtPBjM$9t>VEc?a1f_!$<P5xJ|H(AV_INbk=hH$xj`5Hx5PEJf*
z92Y0&&eqniLPH4=k(VSkpHKAEq3BoV_4M?>rmxqX3-a>v!h4W$iLYK^*xK3_rPo|j
zg^SqcKzZ9)=)<Mr|Nir*yuAF{(o#`L31Khun{zB^Aseb4ot<$1GSbo>o}P(`iH(ho
z0gtA>JZo(0*)@Uo;!{37HHA+=5S^IlzWC#|gM)*XmKIs~-k9@5RPu5+>OpLO{`^S~
zTW!RiAUn5$EH2QkIr`lmv$A5PqM`yfLOLFH3E@)w`}gmYlam{as-mKzvI*?Q+jHF(
z78V43Tj+mol#r0HY1W>YN&U<7sRqdCwTU`CF|p=v-+WfIZU5XV#^&axhK9!K>Z+HQ
z*QGzNl$4aSvoqKmEs`-vad~HFX9)=jCnqiuyw*tS-mWfAHa7SvQg$OuWZyUu^vKCK
z7M7O8%$ftk!!Czg#xgRUurYIgUQ>;MF_f#6`i>V78`6S;f(!~Ngu$XzR8*>}s<6<8
zEn)cQ3X%esQOhR1ySwY==7uhk0qbjQY;Ive#pfier<b~-y^etjf%wqyaAqbZ)b;oF
z)|QvGzB=AbmQTI^>;kfKoD}uMtoYZj$E2jJt*`5;tD6}al~@ekBPS=napMN00xTN`
zD%@m}_?*|Kn?tv@wh9XiA$S7<0${5lxaj$YCSLa8AiizGeZVsr8W>1ROH)u%#>B)V
zCnc?{uFn4c-4?Ou{2rAcX`es0cXi1sDMhu;td3Rfw?)yyI$CkLL&-LhI8OYNA5-h=
z>+|#R(b3A<+PUxF<55sp85k7VP6+ZQ%R80SUEx3_b%n#cys$7LT)fJDR@&VBv#+mj
zVc}*&09Jd??*22>+GJ*BWu>LvqNSy^v|MgzXb8T};N;{4!BOOJN{b4qiI16?nHd?{
zwH^ubnN?L)#p=1XwTGWRe0cV&#2oHM0(Ccap`oE8#io%jUy3|`{v0AjK>;_iH8eOl
zDmof6TF93au#gCzsG6FZo!w^ngqI~Oq5W(}41)s3<;#qB?nr*uxTkjs3K8NF4W2qI
zv85$97gt<-JgZ)<HVs~qkoQ@Pg5XWQU+zADt!N%SeypRd4ZkPInn+7uP3`5re}7|h
zQ%OyY;lJPF!k+?VPXy0>e@#w7VdCqH<#!J(b8~YcRlM}YaS?txj*gD@_FLVF+~%C`
zAeoAb9gK}XMnqhLJtv1srmtA#dF=G)(Id!W=y{=`GLDWV<>lOo-MSDS5TWF-=hznJ
z=1}BpurX+vne)oa<LmXGs;bi8zTMu|_L_3FrhWkffl&aR4?0N3$B%YA6u7vNFP49@
zCCFY2e4MV7O<tIG*+(QhGt(4VWDJnz(Fa96y@{Vc<9g>WV_^Im9)?d5NFT6ehUAWa
z_YRi!tMv$%fItltUo7+sPoDI@e@_E70b<Y;%FH(!6-~{13=C>|dbZ}~32)xKiH$Wf
zFfg#TmUGJA$AtwZ37ejt?n&ljprwsTOw7s6^_c5Q$jxQ)^z<xp*u_FU6E74X%qv$a
zUo6K}Rq-Xu6Jw%talthOq@<*bB$kOGeZRfm?pF{(4iA4fE18_szkK;JeGGD7AT}w9
zsdDJGn*eJ4o|{`*LOX(9^6=q9axSa4Lf*AszaG`N?`M_Rojru$A`$g}^dTMfqD}2J
zEmW6?r1Cu4F;`cAo0-Xg>^s7T8i0XM70GSV^|n{&d~|sDAg7=f*1+D*?g|!Gddp&v
z-X(})Kg`J1>zL?BBobB-pw6QY&->EEWa5}^@~NFSBo-p>tDhXqKy<#OxW>TD+}+ic
zTU&d!G??>rO!=_x3u*^GR8<{A|42wkfD!~?voVa6qpPcHU~rICyF%Eh#48Z`ANJM#
zwJ{D(&W!%W9JPXourN*zj`aRTwE|t0r%&nlH0x04ez&x=G+EFS693o0095nzj*bo!
z69`jjWo2a}335zS_1i5}X9BqJ^QS5;ZTivCkxrG9imGb9UL6W!pu;x3uYjO|0OsSX
zDmLvcD=oF-NrJvzYB}U}w3rDWLqkVc^QCz4nM1tUBsB>F<KvGXKVbcm1>E;GXIcR&
zLtw%^5D^g>&6Dw5RNP6MOROiio+Bk;F$y5&etxZsTEHUrSI`xi$nzf|WtV-<UYs7i
zfNu0s_&XB0`1^NkR8&-Is*u}`S&;)ymRD+DXiA2a9k!K&kjry}{@dQm<~T7P1>8mo
zHnkNtg{^hO_9vgFbS@wkgrL9IIW8)zsJwwi&5(+L?19F2>(;Hdo(J!bYE0bMKTUie
zjV*L-zH_%h^sPJF=npwPrB{k2b~-EjFT9G)-gPjJ??gmK;$mXDtWDb1opQZc8aV60
zMhtu(9K6lSS^z~WCnqr>!S(#?w5X^kBZDqG<>QvhQO&|fmBBZMhuqXv%qnXY$DLyg
z9?V|2Y;vjt+PZyg24`Q?Y^{rmQ7`wk$+_;t<)x(~ZZ?GCBMFJtIA$##iXPufSmNPi
zFQ%KWKtjVZ*8kjULJ-!3{fs|Wd&_tKfuNCNnP(Gk5zT6Gvq~sEmoeR08GXBfnYR!>
z0$X*Y+%}cNG$JKM*U4!QS^*^`<>0`;&wL#jCl-R8bARon#!G?2A9Jflk8WQ7q(FLi
zMv$3*$6{>f;!v&nCyoWLf(ze#Djnu~pblgD>a!=<+1Wvpfd&VC&1n8s{kQ5H(#MLA
zz4Kd&3naRa?f0wET%RIHN)9}{R8$7btVg=KyCJ@1Wo6@7^#q;Qw2K^~F1qU?d@KM0
zL5@S8%>DF<h?=_Bq}bZp8Zas;HU_e<?Io1eY-G8uo`(m(%M&P<WMpIn1O&rYPS2m;
zM^;Muq26F_Z4I4BMNu&*BxGrAZM@pG7*P6fp<#QE7+KOX1|lmRYGAX0UF9eaITr~z
zxdRV{h6WN40@TSU%5Ygo`u7ortr2pmf>2LRp^N9|=R^GjumiBG@f8j&A6WtlPb~pF
zt+1aB$<4Kb60EJA1Z?p0=g&DgjBqcnAQu%y0bXcmB&4U$K)eIz_#~Hzjg1YJLd1z>
z3xKB2)e7zW(9ljubI490869sJL;oPbOu0mHC(f;BfE}I!lY&=9L4o(y@Ai$2jgXM5
z5MQtX?HwJ_l;Iwz8}^0NdiClRfE0IV-)?Rxf}Zc{_07!8UMI^ip%6&4934TSdhV}Z
zzh1w_yL>qiO0q8vBy+^#2{p7w8YIMiOE~%dcCQe0wVp(7R(AFu9UamN3X$zb(Ufx9
ze?Ej6@MUWR<<+ZKhw`BwtZI6C)?{Zdt&LT+_uLkTDuuYxGHZ{tDyylfDK6$n?JX}U
zN#cEx{ikI$VL`2=Q~4j-BpyccU$hB&-rKh%SXfv^>A}QdHV9$m=H_N=WKr|4Uz$=<
zcWG$efBtL^*X`@;gPy{`_s&J6@1jq}Jr))sOmqO(uoiN%vR$wpb~7ypn=^1TPN*FU
zE`k4FTC(6uq`Yxsc5?Fj(2(;~LxAh{^V6^K!tfElQ9Up&PF_JFmEYA~M`sMWy-D}G
zX8rpRoox~H5vWgTci5hjlaZ0}rMYnNVor88JUMiI=Z#5`B8T9+&_;wCVq;^W|BrWe
zc6N0!v9Z~WR@gxcfW<bNCqO}KA37*?b`B1qK|%V3>WN&|SFx~ia&kh$!ek{SY53k9
zG@zD7N>w#3Ie9Gh?h_W>>Nb;NElo}NWZoO(<a`zbcVKB)QP+wd{rVy&BSTb4sc(Gz
zY-6hN!Gi~|6J%Y)XI|u}OEnFDeZiR8JHLIHoSY0m6p-c3n>P<AR>&dZgxN>yyz7dJ
z*l*o3nD0r3`Uyq9*rXdWMYq)A2C}b-5_Kn!|ITb!uJ`~_l0;BN664`1LII79-PvrV
zrsj2odIzf%x}qJ81`TgIzq}l%<AZzmI(mC;e`H80qzVGjncZ}O?q?(sgAF~#r((vw
z5(v-G_wSgvxBzA!0U3a*6o5s{`QSl&1U)G#G}i#E+S}Wg*-rG#+6x`86klhMmwkI5
zjeLHBN}L-(^Ya=7k0QV-W@c+MvxV8&Qb=B3-wRfsJs42yaj>+qBCDWqes;VLUHtLm
z$Entp=4NIadwcYJ@965`ahufS<hr2?QuDjK&&va_J32UMY;GRR{AT?sJc&Vebv67=
zX+}oV&`^10<wJgc?K_0m$hpGcQBmmKprf&omX3~&n|uGqM;Tahs1uxIt5xf@w0!Zu
zQK8BMfI8;Q8)E>i^bUI~NNC5NTfe28oXWDZ?;-o>a8Z=T4xmX=YHFQ9b4XB7(A_7W
ztj4P~<m7Oe&-e9FS7EqwC#Z9l1#}UJVdwSnG1$qFkdT)z(U6sNH&D+Z@8RJAMcCcF
zDm7J41oe6e$hfXf2=xZkQt5sLw8YB`?d6IukJpKdiOHwn;NZx}-q$xDL%4rNEw}i;
z35}dDswf#WwYJU!DZ3Tb1`H420}|=l^_K0xe?Z6|ssNj^o?by_rUafSqya?rr%!W*
zhOIlhyO5bkR8HC9lap%!hBh#mUR@o7Wrb$Un+((zy~sg|9~I|JAZuh~W*#3O6B7}U
zk&@<VmUgtXhy#}a(y?MjfQwp^snJRY&;nxP<0Hevad2>eAr18PJ@h={L{?@YfH%?T
zDJUpF6M=MwkWNoGw6`w;LIdR~DTxBvmxX{&p|iBG09oS6)2F{TT>y&<*#DMvcjrs&
zPT_q4n{(#^>OI^0`}^D5GEj|JSX31iF|S?YVq=45M4c?pz(*#Ag6niuf2T5Hln{!D
zgtW9Y*fCIRa`W<_+%d=}4I`25M(og=?m`)-k%Q7LE!{EO8NcX=+N#jzqq#&|Xmijx
zpdUy>EFsJR976~h>gxlc0u|*~e}Ci5CID!V7}_Gl@4kiiJOuG1Bs8?{bWaPw8>nC@
z{I2XgJnAAMXr`Y%)=`C;02BQJ+yZpVmX;RCFt{cby6^UUulH6vgONl!4Z!TzG@P6R
z&CRztIb++6K4oNV9C}>5c#)c#8WOoET?r3jPV(E#CWs!GzQa|{wr*}mKl8MPay1N%
zjm4Z;EE+%q!O;9WiMfM?Li>v4wiEJ_l6UUhc^wnuU}qN`5dpx`q!=j^&vKiO%oxg+
zzVOG7A1f;>*+|0RV&mhw7=B97SYu=9XlWliadwa$^vSLFzoPF8ycbe5%z$-&0cg-f
zg&i=S@Wv?@7nfYE@`!dLdmaih9K+vUoCwCF@rh_><S&Y|QZ@C&y*JG+&G?2b_ZYLh
zs|AZP($A?W5KZC=3caBB!fpWt0GO4R$DG=Wi;Z1USa_3<ta;bHFO({I;&&9^i<#`Z
ztMN`1IzM>5GC!uvC?+p#Oy)e8Al%*EZ@lO;EGZ`kMS2X7&)({YLK4qSJWN?RxuHC*
zOsJSnoFU8m-YmOu6LX6^schtY{=`bWd&Lf&C9kBe`!8ATGjmN`{>>fg_(qSd0aEpU
z0OzYGPcKSl{rI8$@L}cAksFj=5Fa2DJ&*S^w6tQ9LyrJ!nqS9p>`XsjuZ>xjz;MeG
z(HIoG-rA3uwz@Y+K9(Nuv9Tml18oAjj>Bbc+i`6*wV}FTwD;6a&CSVt&hp~o^xWKP
z8XAY7egcvK763?{5QSl5RjCAG%SVk2zKoAX{@&&$TWIM?l=a@Q6j`6Pjghs}K=Y}e
ztoOSNt*6{(%--2Kvw!iz%Wo|$pMW>ry&Jq@=GTCz?q}xs$+eKF!1Y_VmoYo(*)Xjb
zwV?e;Njlb?#kE$Gdp`A3nW<hypj7mk)z{Z^+l;n#chA~GBvcM7KfI4&CKH2g7~S3>
zMSpQyA3-Az9hczxb)=P(ynN3q9P+u|)Pnqcuk~?(k`tf$aTUaZ#G?=20Y&unMfJ{u
zOpq=SDG0@aiK(l<KL&0I6C(NuKeCma!=!6I)!W9(>SRz&4>qr|qQXBQV9K%S;^Gal
zi9wgTjFP3kC+5vj)N)C=pXs>2d`9@(0zNNH)7I0&#m){D9PesCC=s(}k<k@oB}V*w
zNNxJP?Lb?<M?LdTWg>qhWqn|>n2e7bx;{hf(XutVaahfE#{=PF2_hk+wmx7Hz^tby
zCq1b`-p6~^R-Y+cyzd^Jdn~%_Hj3r!X^$7=(4Wtnngpzt1`H8+ZM+E6aoWj<AsQD#
z^Wk1zTN89#Py|lr;c<d1baLOY1)GC|<Hd_z&<tNwT7N&3>C?Kbt)8<peP6;Y>&TJ3
zcadwEjGXs*nGC{(@qeX|T-@Bf@M>?bblbhhaw4#CraXF2nlp`WexzW!!a7ap+MeLF
z$GY8u8@C9yk}7G$`D4U}36xyWlp-P`xE?$ZaNEfO1_ZnnWLeD=Le|{x=G5Wl$2RS<
zmUS|-lr@n&nYBrLo+{IqpOk(PxQrMl3vFg+W7DX1DTG2TOfw9W5)|B_0zCz%e37Mq
zv$&sMaJty-1m{l=SFSKcDfZ7Ehw*hX)XXJuf-DQTX#B?y15?v-D`XMi+-reP6czFC
z@t>gt-V%|e+Z_9~EhiOEOxfK8&wOX^7qUE<YY1QJd5MdGh|Y!~PhYx(X>W?>$w6U3
z0SMdg0KICah_VuT?w8^eClRQW=~V3+eoI*|_PVC?w9x%*MQ4SI{Rf&)My1#3*Yfgm
zVVZ@N6%cMZfFShtQVY02H^L|-7U$+n?pCwqDD_GmY`fEKkUTs5+~(?+F$~0b!`bOE
ztIpS_Dk`(EbLKzj7#J8pKmj=%h^+<ed)xK94m!jN18!&T19Lq$MHSXh4^n;Y)*{?o
zw1^P8`A|r7b#;}LKwjeJ;Mj(bGd4B`EyR&mUP}H`T}fGjpmSiUclxLH3ypWLCtNS(
z6b~XhfcJcK_O4CQ5jZNo7#R0{7Hx7C2zi+mvSI2u5tI6Ti-9ZvTW03w03tukO6|u_
zQ|$&jAG{zEI&L`3`27UGHHZ9H2I13Ixs=`D0GTfn)ses3uU<hA=z|IklbpW(eo*Xn
zVIu*0L92pYZ11_m+OzQSRXF#dK~+2PNV_DElVz0JBcDWr>DzA9(%cMW=-}W$T|*-(
z8B5n`Px`|b1-hf<n|JS$npcx!Aqtx2w*iMkJv9Md20;$0qQ~JDaMI{xEX?WKBaBsN
z>sTLg&n)lgsK(JV7G#vNRVIcxB51Av!-Qh@KMd1*^4m-3$vr*t7=9v7oHRl4xeK##
z5?s!rXZyA_d9#bH+BG&y>~=lU2$ujTtj(>hfbk2o%0Y^U%J#C`1Z1U{WGt`!S1+;U
zl=|)W_YRxmw>3oz&gFseYHDe@ySv{YB!p|Q=+&-FPCh@_m;xjO_vy7Coko21_-tV0
zjE1{LSoQH^eA6m4pZVRry&`ocmiVe?jTnZ8hSvgV85seT=rHHa3q@X}bF&loem400
zDD}6H5$`86oBjK7w`nO6LFNAsjP+!(ne;T|ZKeR5kth(On5Zbb*^V1PeEt3XD=K(k
zKJ(<s6VN`39P0H>J}!)CrS4Qa60+XrleN1eNQi|v2N@1tT~In9s>jAW`#+(Y8q`S|
zK4SiL)^zenDKb{>_8DnnjVoxv6n2%`#KiG~IchmsS@%i80D=$!1)HB&NaPA%G25uF
z)1|1+KNK>$1UEzQcY-^OBT%@#<tnS!d75f-qKV`-FKo}UF~3oYyHTi)I0*ea?Zwia
zZpI)MsI46{n@%dde9dOU`E!VKhsE@$^<wxiK)T*P^GX#hEph^a>Ho|t@tSs*huD&d
zwIJ8^Hup!azi5=M&~Tr_xoKp&)*f`qV?F<QS`d}$2f;x>B+!ojBl+>zj#sy}Nis4r
zf?_})Bjv>DQb`<ATb`C{b7a7wk}x4eb9;AcC7_>tU2W%@BRc|5;qgJk*?)ThKzqD)
z?b?4xbs~skpzC4d-~hsxb7FD%NF1W8I`o`?fwSU^iX<MoNYK{~!-irBg~YJ#Hwa-o
zZI<|MlaY}TkV&==JpclHev-Mgu<#l(XFeb;o@(8xL+><DY=)Ej6X9jPg|NgIK^vz|
zl=2oB2-rW!ABhMGQZ|Doke-v0lZ%Ur07)`2F||ct6l^y<swt>RF5DSOWW2U@wX*cW
zyXd%-HHoWct_Yf-ii!$mtul2jEylZdFQKCYVOt$8q~+yxgAg~8@Edadu;5<gVUqOD
z`&eSWy;wq?o)#Ywl=F|t<pu_8n9V|{6BXH^_lSZHYrfa;!GW`!b;TOyH=6RlQ29Bi
z5#uY(#cK4&ww13FzgT`OOyPR6?95C4<oQ)}9}y-%h?<(b0s=s|+5xvJr0^#^uq5)U
zzt6@7Fs>RIYFbd@+8m50>U$K}5M^<dhgIO)<Uc5Xy_V~-1NqJg;qwJDdLD$If|OMI
z;r4t@bv1~wfIeEo$unV~o0o^$J*{CXTtq?tj~~->b42*~^#+Ap*1rgeh*Z_oh)GEk
zUcVNGSkdCqB86eozuABN2udT+od<vo$qFn2FmfO+b^hni$WlvUmJ?<Tlt`@h>wi!F
zbDMxVeCtF}#($9i$uSu2OwNKN0P!bwt1sjRBQ)*^iic?*KfXBFcm~CR&;2m0HvLsh
zbi<V3lB%Mqyq>-}zt>wrwx*SXx+_*`mq{pG_;rPag<)2wto);u+S_B{`-l8|78usT
zk56n)hv=$P&gT{!e?Fg0``U6jApQcy&Td^$9svRR&4728NeJEC-C<ggo|Xn68OR+I
zGxO@w(pRt3BkPf}C`z2O2x0|`z5d_=FSby|vmR}!5E{%!1F?nXXGbwfHzr7?qU#WT
z7vnoY;QspRmAH$`0ql<LXhjqh{`mMn<~IaO0XIB*Z3pu>y5{emI`6VPm^O5{_{;%|
zg$(Py+3(p|7<5@p)FyY%dhH=ip)I|IDVwIQeT~ar2U4||2d(OQXG#dBFJ9@`dEdHZ
zR3&ZPHo~vz<;BaeD=<L;9ty$0&dz>zy6$Z#^HJ3OnPb0}sxng7WwBq1A&{M9r)O?%
zPY^jpd;am5Rs_#pb%Hoq<DSAJSp*3l+~wlpBFqW+sd`67j*)TXM4`L8LwYgr<h8^h
zUO}4?v5C5@)B}?3j2~vvvOp^R-wb4%;-sbar}w0fFn9YzO+EX1xz`0&VxK%C>q_8~
zraHDpWEK2f)Ywp{QQt|6BiAb^5%O?8;2g7~c06;++}Mn@-XwM+^9(Aw{GJPjM6X}J
z9vvM8mX-ht>#tuw%u3d4k0VyJ;pbiKHll578e(xL^0o7%Kqp?9mMG9EIpovoVWHH%
zMd9-mC5{;zgB}voo&+@yIt55@(aFhq+1b)goCle@!!P>X&&MCnsO<G>rqEzkZZZm`
zPJFAH;QLmO&=<bVz%V~QU!YzHx}^a)0>FZx`sB&u&L@q)Dajl<vXVA}+)ec4$)xen
z`;^FbQ*)y@JnRGw5lc{Kr|m1T0NzwQpBk%j-d~?6NJ}%&*B6F-71CX>O=zcL5D+Zp
zKDnbG-81}Q^}+4^uQ+uf+i$}vaS$&60+<>bmzI|wA09$yhnEY`-s59H0ppRx#7oK{
z#hwMR4=1Wzb4otQ$Ok-Z6xMRZywZ-)*LQ?*vZJGssVO-b8I_P1z#WjgsHszLKHi8?
zz%SsD+r*odqE2Kj)v!2Y)1>n0__q>+Gfo~hj)w67J^^IQ5Ybz>xVX@Qz=pzR*fM!t
znGp1Fl7Xe=<=vefOkW_O-5ni21_t<@R^Hau3L+~fJ#S52ebSI!vOW{~731tB+GK%&
zp&@7jpruyV)Eplj#j)r}XZAD5zC|rf?~%6{E5QKajUlHK*S(~1y<RU*UDF>Dt09f8
z%a%FQC2tWKLBYXN3JM^Vu)?UCon5ok;!SuscpE}sev9mz6d?CW^L|_88SJi^nLYP*
zs&~(<OvuX2q>3*!vv;$KXfI)jQ9kec=%}`m(&lJAl(#(qp0@fx1OFlPOr9>Er^TmC
z#<@7`2&&b2_|=&1F;TJ(#Z*4q3PuVs<ezPnmLEe3{V@u;0G0wCbcTg>UlK+>13!O)
zV+ZJM>mOQZ*BQqQg%#vRXkq)d!sB6->^8yPAtHNhKd<&P>YARkRu6^~v1mi=G|XSK
z8LzI)&tE%bKqrv}KZ}*MHB3UlyJ2CG{qki!FwkMEJ*W%NOivBIjfcG<J!~%v9a31(
z(X#w{c#lOEQ}~v*x3`c`9SF$K7%yE4fJr5c@d0Y~E$(nnkkEbiu_=f(Ajyqv(|k>l
zDAz^C`x+M;aXtfp84V2$R7MyXfHnd_4Kfbv*+sHDhLih!xcy~E@q0c(_YnNoZ`=Sl
zZDeGWk(M^Sz7DF~(0@rzmrjHHJ$Deoaj*<<{myvS{zX)XLcQtcp5Wx<l>LL%zDI&8
z=qIVIJ(?om4&w`7KR<wqFlK-{FDn}sOPEh1Va#*s!Z-cC!9f*m?UU8=36x%^T?q|0
z23R}X8;{KGS(xSDzw>X=$<GUHi6n%78J*WN;9b5*DDdt5KO`p-Opw16_5Ln7>HC0|
zDs1;}$q5?|#INRmN=_s+pkFI$fW;`BoEuOm4@E>^U@Od|-pbMkHh$>aC>9DrT}cBF
zA;c7ThzPC*fPMnfu!o?~{rlfrT3)S~-G9HKMd?$d{N#z8oLsr}2#6!j&}Ux?>swfm
z!OZ86K18Aj+M%K<cq5XMv|-@#@uM+_yvoYaAl;UhK0sFLzlXIImV-GQNcw>6+q=7A
zQY+{%Ckv3`PpbQ%gSvRt4MrA{l5Jpq8XX%0fl-*|OJN~28!2h&TYO}%QF`fBnB1mX
zSy`EJYCeDd#n10zOuLnZg|xhU%YzII2GC`*0N60Iu|;*8%tMzP{_^bX<Zx+e2^53I
zmO~Nn)d=7Hr^F&m221;K!=)N(l*ifOJVdRY=dQ|=Co$k&P**3Bm*9U3g5RBh)|o5l
zzTo)*{Ss*K2gM4rzO>siae!;v{J9O$vSt8UKs5e-&I6+r;DAslKvi#!ll~S$<#P+f
zQ(Q{k*n|X9YU=YwTp`e=8)4+AqjL(*5+jL=(Wn)}|6g^E|J$gX>@_yL=hEMljOf@{
z@SOR0puzwxw7ICX6yEuOojtptV4~Dg&DePMXm=UNt^pcooWVo?tw}h8bX!)&buI8Z
z5mA}b>MsSM)7TX=8Y0Lsf`b38_7D*ea0yu{tPRkNZUp!VNFtXmh?tsY#>6}UFM*fW
z8H_Ga7eWU=O?#R5`5AO%Kwp(rRSNGo+uPfrBD3DQ<-4LyBaXUC?%%WmNYWDk;4q?-
zkO%~!F|bkNi~dWvii1OOT%0PH<TR;L{B8*NWc;f2JWhJ(0CfW<2>{Vn78bHTeS*i)
zEOK}`3`kQl8*D~NJa)b`nDjBE#KdY!N>ej4|51hU_#hU8|3%@Oz*Hf;^?QCET9AAS
zKg>Xw1qD-p<^1}EGX9WY7p1@Cc?4HU`3EI$B1C;iNm&_`^Luyhh>D55aCQcshV1UP
z{^JF_z!tFj1`(a{uW2w&6IiB&qbc~Dtd4ddQx?IX=?EkOY-=z`C@rPrvLZqD346du
zghmPwU%qzbEk?#f7_kx)gCohv!UA$b+KC0v5>i+)48Sy?Z<yoHv_^7}gvljxk&=)=
zX!^n4tij$yzwDg-T2Ya!Tcar^7ErIBt6f=?m32o*$or3x>7rd=AC8X^CKS}GSAPr*
z34tvWv~756eEbloAAqOYBQU1H%1uKmEhF<El(B-k?L}ActYG+o-wP(suMrTlL3~k?
zO$LR4Fdjd8L_t9Twku5E3a?Y%Pum-Da4XX+fU97PDRh3~U}!i~?Ya#Hti0S@*mJ0G
z@XYKm5T;p!M(MON{Nerk=V1D@9LgUY8UkfzZM-HC9*GSY37+NO2`b$cSX>V<A5>IU
zQa-ddY6`-){F!U#<kZvA5fU6+_~i=?ADJWpgk+N%v>z}(z}N)ljTbIlfMyHV#iyhc
z@Hlu5;XnsL7gX>MZb|bD#!h%oz~hvgpAWMIkjj8_gWG_ZMf)wcEfr+g7GzkHlDPP*
zk`hONbkMxOdVw08SpH<f@Ox%p@W_etH9u<Y(*JIF#iAnsGoAr#{_m5+ZP4VPG(nvP
zXEQX-elRxj5lh~KI@$zkeQ08$7Pve^n_+jKfTIjp5A^9rk1oLSDno&y(S>jUJK-nU
z1pCctF|Ne1A3yBDU^hHGtduSeREX{dTx)vvZ<G+{9A$+xFakp?cwKwvot>P(MK&}v
z6zK0i>F{L(7Yc1t^Xc(EphntTx182SNl8gRXJl}~F08GsspP1lu0al)qz(9hgp4di
zG!6Wg5N5DjqB1heFnu<fzYxq#BQ~KMcaoRdOLYF_sZH_5zM*IJa80I7NeIUDCR!79
zWGfaXCJ)$AA&kmw#_lmQj|~n&dxaU%d*3tGqw#55rf_}&>ON!YlhpFysMcADTfQ_w
zwH-UJv7CP$olc?A3ESEh<m6aBeH#1u^CGxb;HP1^KoYJ4uUSz#+Krv#{6xX=^X4)y
z(*5nBSN8sGBTng8IvMPGa8vKqZN36W0q8X@Yx*!a2N%^b!(D`Vh4$LcV*dk9&W6Uu
zUeMWLmteW8tNE=y`|qbssO>bAWl~BnSrnm*;3;>^1^ULFeswUFrWNNxc!STiv%MWz
z>&X}2dAd^E3np+j7+;p~SF@hhro|6FNYy)J{pmGGd5!*6jLHq!Ye{7(<sDvX-=^uO
zEN!#W)3>LM9V@hL;3>{OULtY%-+FV=%APj351exxJu{IYIL#!a(ktH-+IPG1NZRgf
zs$@g#Mr&kdiM|G|pRTE?Da7plgf}(#dC16)tUBc$KlTGJAoM6+jLDcnW$Fblua)Wf
z)-xWObdxgLy}5{SClP9$>~m^r;V?+TFe-r~7}x+WgPs9*6&DvbtbApiI5<=IJr3oo
z>-WFCSJl-8zptjU^7QiZ$ml2q6%`BN)j%9_ao9()r_2a*etv$i!q_`FfS7Fr+7h_y
z()((U*T7|tHVEm}*7r|qFWTTFAtvItBusC5d-cFnEh^dsZU`>habUTx;sVcnz*Tbv
zN)>Dzn4kcL1DYR$00xl}JneuDvaUeS_66}6w936bCqUNlZZHGD6c~Uu+{FVZ)%U;u
za}16#03&P?MexYM{02?|RC^p20p^0%17+XAVH=_z%4}>vAI>+Qt<BAcu3H(BbXyx6
zUy6$Cot&sROt1xSUOENm!2S1sCtp4tP>)D>z=R79TzYzXzz0x_pmTrhU%XB)3!<tb
z1tjl9I`}sMfw75+x-Vb8x3!rKW~+eF@!q|AK&k-8pbqO`zXtz{PPL1YuC5^b8z*PU
zmoJJE67vw+key473mEz||HE`3k$HG|=@}UK9zH~lj6euOj%UFz{rfu9R8X)8Q3Pgu
zyv99};^B|+@g9?6@Yn+*Cn1EcaN095G11%03da^IDqxy?lZR)xsmW-f*3*$U8Q|<4
zn1g>`UmLBwj)y0l-UsbaSyMCn{d;{_#phFxKmgKzc|$<>9Bv*S0UBmeSs4ox6Cec`
z*D%Di)A12w-@4=G;}r`|?C5BFa8&aqLn-PpD6ECk8sKn(vjK3!82E$&!a^|N{)?lx
zVK(~NB<f!D{xt~qM+29Za2*~dI#>`P?t!X;_yar$+@x??;rw@u0x&nfYavwc_55GG
zg3b)TvnO-`@)Jz8)z_b#nMu#jmxr}+M6Hc)>kL3qeDp$4ZehIvX6I`Zw?Hdf(Y{ae
zOdNOPOiV%o20Ihgt}){2Fv5Ju#|MWD{v+06po~Q0#bC~3prZp(p$!IZ6%}wK1O^Dj
z$d!*W@zL<ucY*%6Lt=u*q(%Qe&<@bbNU_CXjstW64s9U&^j|kDU`%>i*xJ_F&4_Di
zk_9$yZfvNR+fc@|16JlF^ZMz-31Fm}``^SQ>)9K$jd5_JLsZy|lz~5a9pWA)xB~;6
z;DuIGi-+ybf+n>w9vc_e(cTW7AN)$-T?chC;q6<{g+UR);6puCBuK?5yINb@@42~q
zjEt!d9l+l&^T*wefZg7o`sbh+1y`zZg-&;3z~d1vj&<!C`2E0l2$TsbRmCg9po?&o
zv(G@Snwmtp5|d+Mik~&U0@V*JRufguxa9L9vQT`^@INRn{CTy_Z)-69gfU;LpeGp-
z(Hu}9Cr+OUXoqKixAHbF28nq5`53B520TnazTshEpg#FOno1K3jbqYy<iz45G}W!{
zQ@8#%Zi2@ROMfB<=STvtQFe57g+)f5PX$tgBk&nmabQ#pXL6vKom_&{_wEN9GU%|2
zjz{1R1ou05CKnu!Kqv!^0SeFtF)UqOdS2ctBsz!}pgDuN7Kk4gcv&DEP_}lmZ`)92
z4*1lW;_~vWEiH4RqLe&6Pr<<k+wf<e%Hk6bglRGv1TFCDfg1(fxn+5I;F^GWcb-nw
z9(bt1gz-%Vo_t&b4h6x;yeCEA_+ZlzOyDIYoX~Di#ytnLYB(}DIi_9bH8wJ$MT6()
z<iz{%p|Oz>;P^-?fp;s~hys67fp1mFI`4A`XRz%73uTad2j@7z=R=H-575*{4Pbq>
zAIQaU@C>XzFa?K_1ZLg%H*X-WrkLK)^uPnTXu(y%a7uB_wBGj;7&F083TJAdfSS4D
zIiR9DZ*`=+H-SR}s1F~X`{r~rjH1Dc1PJvuAMrPG)WiHl861*$MDWmK+QULap@o1w
zzp$w2)_=bNJo^RN*}uxH6Cqb&dL10x4lWs>*kF|P;fBInt>NWWTTrkLND=zy^vnzx
zA7PoHlKIsGa{ZPF0W}UvvM2bp!C?b1%+kzEQc_aC7zwfn)Mgh(*g|g0zdKehZK!^!
zA((stX|BOU6yG`a<(VIxNDOR@dh-UNBOlz?$UX$>Aj;&62ueOKu%^QcbTcS>U!lN2
zjsNxQk5kPIVT+Yc?lS~Q_&3igVjK;^t^f`Pz-w+{!R6$@4#Tg|yr2H!ML$vw6KJo0
z1_Zhh|7IyJ+^B@rRz$sCgNYr^kby^x@!maf?V{$qkou5q&W~r)z?=auDqcr3@LMIz
zuP-e{zj?E|ybLFlh=N7YMI^zqC3@@b-OnImz^HO<%?7+KaCF9$+jUdH)%6hm^XB7M
z@Wf#82ZN5#+7yx)iXt%N-n9_<Ob}bEm-1groNSNQxR-zUFa<7L=qvC7Z-U8v0JLNo
zwOMFq5omXh@+a)mAEhaKvwXI3`E^_D{5hdRi|;JUc~KwfxWbSU+zx8{tgK)f0WL%E
zcJz9E(K{&<VwpW@Rj5_G<METp+vBvg`(sAN+5R%|`UcYT%x+s~?Hq}7UcvY~Du^Fk
z$^Q|liwg^pU{nPvUw3u{Z$%S6+4b(*f0JDH0B1pSi#70pDp~f`qK1gLug7(|r*3vG
zs_IUYkHozlJpm8FPtwc1GkW%{etsJ~;u7IxAfBY>=ALX!J%;m5VAm%Lr~d=6!Nm$m
zF)%Oy&nbAckPc>A(15{k2!&HHRS1refv*w=Cmcc(#2DCi81!sGC8iY*GqbXy;IdK&
zkwKD<47Nv0D-pmA5E$^J04ZvTq}GFre*O%eAaFcuC@HadogBdDL<%0kQJo8}O+t}5
zIj~?_rrqyuz?>^OdJ8<F90^@sE~OWCs=T~8%JzG3O*Kbj3R1?qkE{{`kQe&JR6FZ7
zb|Sk@daEr>m9_Ld8+Udbe`0}2`3y3QThL5iKEluMv5ieW{0_1Q*p=7EYmS?V^`5G!
zX=!L&LB9Z=34B^{kAtZ|klOO{^5A^M=j?3FViP$q4TA9~u^R;CZh-CJ>nax7-r544
z+xv7^4UQfHF^7W;P(iy>h4f$o_V}^k$-ySNNLQ-RIdlq85nv$G-MtEqe86f@p1en;
z;8qF>SOXhHsDj(JJ6Y@e0|WQr^FeF+{{1;1)AdEu3;OSOk3)2&oiOvNYHIc?0||B=
zo$xt-S$g$$B|Y43<`563<Wq%nbm6#1R)^5kQ<$fz_+9QBwY`RW=`$;tn=^@$8~}6z
z#|j97;ZICoGaH*veP$p<P_P@d!O?@g`peX?eQ@ybBvJ@w64^O9fMZ~vfT_U=FrZ3s
zb_j3;=qE6q0_7Hb2@t1b<m7?aq~O$qbAx(1I(hHkZ?q=gxpxn;-)p1(avLqaD_FGq
zpy9jkSw%^Le!P`m;DKPfwPfo#*msSLc2{t$Vtt~1KFuJF_S|7<fkkaQIDLNQnULoY
zAD{QxiSvZ#ev2(6zGlU9pwDTgrMn*<2R?oJlwUvq;&vBy4vwsjLtKLG3|s}n!|D<e
zs6$2&u#np6;;7J6R#gQsx*vp3ySF~ktA8KNzV{89FolGK0EEH-I*j6Bbh800;!ijl
z3N~XfIDs?;2ZlJ{R9=%2JEUD}7-_NBsT&+V08Do8?%h>nabUeZ1;5Mddi@oc0D<lR
za+N4-;&pOz$Z-!iWdgDz;H<jcrOPf_=SM?zaC~q=!TD^<IcuEbnM2=fRz-<k#UPHg
zzz@e)yid-rkEl3_8;)0VdCO~SYa1G-!?6<bx|qf(baeEM!xQMu@Z?U8jsWXGh60-&
z#1q(qFasX>(qy~|MBG3iFh)?{3-IwHh+bx#zluGHVG;<>5BCpe4&bP;I~OGwPhCKV
zg42{KBW;^2`LLT;e4!R`lI6kathU!Td!79FM;gFim@;;POr`eNcqvRHOeNJ0FjdgF
zBNpat{E2&romAGk6MOK}1GYQ0o4;7JEdK68m6hLG(My*sp{`^;Zm_TmQxhPBnWkb2
z|IfZGfZ-W26{(=d!H}M}C$!!26R&H5+guDH<RvLOMt&=V(^$Q|%TtH00v{C3^z}F5
zCbXYCp=D*AP>{m-t<C)#6h{#g6ME#`CgPOTp~mfPHy1Bm1&<VJ5P|;oB?WhMcR1hq
zuToR$!+UZ6>*F_Fz3$Z3Y#|>^waMSC-8iYH`mfKM-ab*Rh9itV(;#;v;1Iw+zw1C=
zTd-pHY**m3^=s0PdoKsPaWM}b#w**FXLl88@|t3^h(Ch)q8;Cni?YUgaou-%&!uG^
z9EGlRH}b1bSZAO9pnu1##d-LcDd_BKL6X|_&bbduSn4};$mQ#)Ib-Li?nBJy0S-gS
ztlm#OoCa4E{MPrQNaeM6RJ5;Q_nfSnk8n^59ITA4`|n<{G}X9#@^pWsRL7)6kTlQ5
z<m>``BEirczLLCc^K1dP2d}k%*Zo$k<T`5iEpK^R)3v&OW+)lO+;Rh>vHfBuqHKMx
z#;azDK}1D{L8;@@ei#-Jn{tn$vewHvh&tv>sW97n=u|g4TuJS0sIKONNh^Di<@r`g
z|DglZVCU>8nN5+RRvZoHp~jr}YTe1xfCn`jN@8#LAHOCgAaUW1(CHU*qAI-U7JdE%
ziM@S^S{mPf=T5QMbu480*9ni}`ilig@dWd8AN2cVB9qK(Jes2yY0pn=mn~feGB}kF
zk!bHi?#*mAV%*d?Y!cT~xm36G1L?n&*&R@ygH23vP?@|UIW2N<pp>2zvQ&ICr=%#`
zzBT<77dFG0zgmw>u*0_EbhrAmRfY3-mGRrrNBUc7#vD&Gbhnsq`M!=mnz4+M<l*tQ
z5wjk(jJJ7!d4+WMVd^{CBQ>6p&hL}7uD%OpDqPx*E77Lajn9iXLeQcXdeyK*@NPY#
zYx+9r{T<fD(%=Vw>>W3yV_GoSt==+!)xnQ1Qmwwje$PU>JCZEGKMq@DN5IC6luaqY
z_dL!OOMtoV%|yW>tyxp7;9a@o!+Y?>3TcKsa=bI;;y2kmo(r;_%V>(pRdSRU>WH>U
z;WxCrqW>C_-Ow3_>-Q%1s_sKu39EO31Y)e|rB<O6MMEAJtvf~&m;;L$kLF~f7r(w~
zC*?Kmc1b`BWf0FGanNz<_1t?gaY~c(rW)trgvMP0QClSC+Uwv5ivx|@n2)6UHfDnP
zIP6D$GCje2@Y^VEi8cA5|B9kP#=6&c<!bx!kYP^T+aYJ&ghcMP{6#C<tnQSpME(r&
zn=%g?aPzy+M^vuQ7v>&o=d}v<L{zgc<#f(9wq3hTWmGa`!|`CtX;}40YUJZ<37e^i
z^zVk=4UF@v!7|&5dkwVKb>d!n=99&bj^Q|hm_lc9fi|<1N^-H$;!59wE6fSf);GtN
z%^y0GFUZCitGadgZRtHM_`dyxW>(s!G<UDTIJ6|t@cA}HwY{zwe_~(}IJgkKlX@Bd
zXXWYt-PfLMT&P!&pwMYt-^WMqQz0q&bto%^i;O~?p#d+)dPT~@ozMT}_L7JKk~H<b
zF5^ghk&(oOgUcDBxA2jqh$aQ+z7Ll=ZOy`1EsvQ_r+*cy<G-Xs@d$1ASFMDMd%yIy
z_GQzfk}pkC*FU;6X)KzT6sqxP_EpBgNV1VB+Cd<w46J=UoBQ3RAsxJ&f<NZ@bXk#h
zh%-<_zF~d68euKZe&Ci_7}x(nPMN`@R|Ts*0(}D)bdU{xB}-R91<rOw%L+mf$Dy&q
z*{GrqDPJj&g@`j2n8+cTQKvfozxo20bKKl3H<(E=c(&nNXb@6j@}l{V^nL#?Enj~4

literal 8250
zcmdUUXH*nHvo=AJEFg&FU?eJ8a#RtKsASM339Be^Sy*yjkSHRcL;)pd$r)BaL0K|L
z9+14Si{!X>FudP)&i!}quY1nSIaAYJ-8Eg^Pt{ZNQde7z<}BM;A|fIh^?P?75)l#a
zo!*qB;G5VTIxQk1l4D&>eN|8-CMG5!At5CtJ#*#^85tQlIXMLd1tldV6&2OlvuCNP
zsUZ*u4Gj$~EiD}#-MMq;=;`Sh7#Pl<KhMa>c;UhYCMG6kW@Z)^7FJePHa0ePc6JU9
z4o*%^E-o%^Ztja0FJ8KI>GI{vSFT*SdiCnHYu9*qczAhv`S|$w`S}F|1Ox>IuV24@
z<HijkAt7O5VG$7#QBhGbF)?v*aR~{Dn>TMtN=iyeNl8mf%gD&c%F4>g$;r#hD<~)^
zDk>@|DJd%}tEi~lx^?UJ?b~<m+)-6ky?ghrnwr|Zd-v4U)$iZGuc4u#si~=@rKPQ{
zt)ru(tE>Cq!GnhnAL{Ap>Feto7#J8D8X6fH85<j$n3$NFnwpuJJ$m%$@#Du(DAe5C
z+`_`b($dn(%F5c>`pJ_gHa0f4wze=B?CH~|&z?QAv$K2t{P~L)FYN8@9UL4S9UYyV
zoSdDVU0ht?aJZ|hD*}OVb8~ZdcSj<T9v&W^o}OM_Uf$l`K0ZFazP^5be*XUc0RaJl
zfq_9mLBYYnAt50s6e=_{G%PIa<;$1h;o%Vx5s{IRQBhIR(a|w6F|S^|ij9qZ{rYuW
zT-=*CZ{p+Q6A}^<6BFOQef#d+yQHL~<mBX(l$6xe)U>p;^z`)i@84%+WMpP$W@Tl4
z`0(N5$B)_B**Q5mxw*M{d3pKy`2__9g@uJhMMcHM#U&*rrKP21Wo6~%<)1!%s;H=_
ztgNi6s;aK8uBoZ1t*x!AtE;cCZ)j*}Y;0_5YHDt7ZfR+0ZEZ!P(VstmZf|ez=;-L|
z?ELcO%h#`8ySlo%ySu-A`}Y0&_nw{}3<lHN+uPUI*WceiFfcGUI5;#kG(0>!GBPqc
zI{M?skFl|_@$vDAiHXU{$*HNS>FMd2nHc~8W@l&T=H`C>{5d~Azp${dxVX5qw6wgu
zyt1;gy1I(RV%OHz*4Ni@I2<02-`Lpr_3PK>=H}Mc*7o-H&d$#6?(W{+-v0jn!NI}d
z;o;HI(ed%|$;k<UKq#%_)CLa#mD4>FI1v#w!|6@jKk0i!M8q+ten&+gnX)!XZTtq&
z)Z*-}v#1Y~OW^Nch!W9)G(;NNqF4h|g8R-fs(ywvoD<0ADYuWPmT+A144G4i7l9z?
zBeeaK!m_JMuUakL>JER($|M!Qx~Bm9e3j)b&}hCuIK+P1k-*6jc6Yy3TV%iUnAY|W
z8&t;<zmgV_&-g!a83z+KSYDU@^lWP5CBxnA|5(><w)glOE&?e<jgF&2pFjKW3Xj8I
zh+3Nn1Fl6O_L3@()7Q83n+*^P>>>>x%4oS|MGGc$(-aU1`{WSeXN73@f@cR2W!0w=
zKS9>7d1(3*t<u($0rMWdZ&DUWs%I&3%fe~nW>amlM<1a=dN2x_<^8DB?w3M`lTO|q
z*+-a-Diug&<Hk#Q2lmEkmf^`YC^0trjST)x-i?K7<Bp#Q@BJa30Uq%h;n{WsZ~M38
zQt`ZYt)pX#6ji-<)6CBKcT)4^vSbSmYTNA#(TP)g@u%anmhrz@Ta>8Fod;h!dadUr
zCcSlR!M~=jTe+;fuwqzUMtG3EK5??f0u^9FdJ4a_N^!3mio&AVZ7clbP>EsRLQ0>$
zuAV&}uAYsPORfgs{<z&9iTVrZ@j7o+znHjJbv;;Hh~per?yn(!U*|44+iGhhySt0v
z5laz*U$mJMJzn!xog)AJhqd!g5XbS%Ta3<K<!^%}!KKuyXQlkwMb(%&?$C;tykTQ~
zWste1`V$#?yqv^_&ftxxyS7}iNCS;{ZUn6GB`w0fL=;tR=@HxeN_x8Q=$iL=h$DKI
zT+N&`t9a6$Wy-B6nkA{QJ92Vx{Vv4u6PMllmd?$PC6rsW6*AdCdQDP!jfL7ww4tc(
zV<q)3A;BTD(MZ5P-W@o3nXH!Qa^pn@UNYZBzNPX8>AJ~SG+?j)+NTA$Te`*0g&xme
zqgOdhD4M7ooTX)%lB~U{FT}O!t(p9ZNdk?R%yZa#cb}*^{RS=ouz!%(Vfk|9QEf+;
z5X%%3+57_Eh;;6!LTn&lAG`e3yPY4y@x0yw$?jSa;kH5AOR%!%hB$gLpl93SeB7ID
z5WJw&@O&m0)BT{*(7g>WnKNbMr`XajIK112uUg=hGSHtv+0EZxt+zy)eY)D%CRK<F
z+oE+t<E!q%i1yY~b>zlq(Bp5<9L63zqO9JM^cIIWn)T<H_jyYSNWO2}d|3^e(w%av
zhCc7D{r=-xUeFbJ211&4{*x~`aW4UT^?1qad$*efV)~=q+woOT#xEHIl22P;jBcQG
zb(Tc=ap1m3^jHXB|E1ICZVgxPP2N0g1YoZg?N{qrGGb+J-C1viG%Hu6$)C&FW9M4~
z*ZJ=Dncex_i*uxQUe-vn2e?|OS6k)1bTxZ-h@+r(QHaFa#Zdk5da#Y2-HcrK#qp-$
z4RDcSe~Nu}OxTz)%s;e7jysvb-kxX8<{ocdbnp!V>~H^~^)Eyezk#+%fjcNe?6W9d
zkT3J-)L|&{V|s&*y#eCI+-Yz}1s*x^(LQif(waMU3J0AG%%7Z2)*aNks5-f;pk*?~
zAiLX&|79d~09a0D2FMnezm^CE=7ohBW*qV}(R9OAfq+ek3)E;gs)J3~6Ldv6v-kC@
zzN%Y;=eO2FfcfGvW;>~<%EQ2gN$InG&1+gLP!<{VZbD$ZIIC%k>``MceV-ETu@8Y{
zlZFL)j&)sIuHs!{&OjsgD*ikVuTH6%d0GB<#5s|MixSO0^I$(>Dv>>=8pffxBk?!#
zdK3nalN@Cv3tu+!PMVzashlE0k<iS$TPK&%P77a%b`|h%mP>MM@tY&l6>r{r`%H;N
ze+&0j_UJ3jynK4-%AiByDIWNTsO@O#;3=on2x-{#q3LFQvPGjK^W<^S1|m(J1<YM(
z-TmSe8E;Num}<%OZX&!e%*hBMY@K+s$*j$2abKiX7?mh`+!$_AQapG6?F2Kr`>h$j
z@?--o20kbR5jI2BP97Qjh)NQ;yW4>;uM|==+K+|zGTEL=^9FS8Rd%}~4N|DYf?J{f
z-|2wK0thY|uvw{-<vH?K)X~G>PgtPYmt8E9vvbr$2SA^{`*rk7QTd2B5*Gv5<T)4X
zy<c>D`{)aj3?i%%I&4@|!m)gN!&_>t6<_|O%U09NOQ!xpVJrT%>!iBqCJ$EAs_YCZ
zF@ph-zaF_#*wyvXS}qu{;gcjQKTf&JTxLRw?$%w(k05$iU{VwS%)<h5b_4l4JYXg_
zKzH-PFI`XiS%gfV6kvh6NqW=bkEh;UH)_SZ#FZ@JoEaYSKaOfS3lSEl)8AiVY?hwq
zOjxx*zLTlF;2SwyCvxC#qrn2Twc_E-4Sda7I09}jh6|ZtdI{}UOQDqL?ym33_YAd{
zE`0aB1`!tIXJ^l7YDo=>7imag<nzkSjI4!A8=pmYhp9Z4x7KpaE!eglv_P_cjR;14
z-j9A^e!Azy7^I{R*jUC>?8E3G!rTQ{P&}(*)>6*pEU3h&P&PjNhWDkqfVd%%h8K+5
z^)$K~hngrR3uJB0UkHR(aG(;Yac#%vrD7h6%F*01V}bGrbdo+A7xUj?y?z6nA@&Kk
zQ%&qiZTO8g{I61l%TK4v<LeV#sxG1uDXLLj?bc~Ad2=sRAi~rLG)A9fuWWxMk`*FM
z@sTa`T0VT9jW+Q!-sRkz<g7<??|WvQ)WN1~-pva+^6;8R4CwBbKB2L8I%S>Jz(CGc
ze19Sf6LT82KDpy&8-EMFfAKM?nh!KOf8O;0MA&0n=x~!b0SdSoK!g!WwRxE&Qu>0+
z=8j(kKv|<-&>MXaeQyzI0!p%^E$%q)uI+Lb=qv+PZ3TWQrJx!}wnRSDNcUCU{E<6K
z<jEYGLxk2C{{w{IC+l5wERgC2e}QnKBog>oc%Pl=D8oEnjel3e8p%KIMrl|>!`AZi
z{%g?IhfL1I?}Y-^ek0*E?4072M|5pTao&Ydz(>K(=0}Yq&&x_;W7<U<T#IyqR`&Q$
z$e~P&Jl-Cv<KrUo8bjwN5~qV8Ohy@{D8wEvC(#>sP(NKy822Vc!E&bHjXS~5V-;i4
z=S@Q0A3Gqgtn{SX*f)I(1%Aj>GWAykaXj=eO+)MnA!c-`dV7oZ6Rgomqh94)(ND|D
zY>3(?e5dlsh5;XjOMBAFW9K$$_8jE})rD>!*|O3(%s=-T1{~`l1MUwYWrVTQ8Rx2c
zM0mwCl@jJKlSioIPX`^Jyd;ls%=YGR3#8<0ccLdn@cDT#=z`dw9)Br1z+8piu^A9;
zkhr-NlPlJdPAlCxeMzK&aX$36b6EM|GcGz68E=|NI)VjqAv!cgTsep$J@h?elTgoG
zEx}{V#)LIBUR2(if29fctgAl(?*S12=%RkS@!`%V#EPuzNHkT40`KzKkMX4reaf6U
zARen-^LE*c2zPXNXwHfjk7uTGk}_!Br+99bu_Je~;U}V~`P>GA(VcH7UT2TG2|eNp
zF-5k27{}9X-!m1oMxp{^LOz#-rFjNUXX#Q;mlj|KxhR&8v90)T`iHmHvdi4e@!l-q
z!ix-EUUexs)nX_Z4{UGh;S3%;G`iF!27c7$G#}h$^dOjfsQLj~`-S;sZ|+djJ<=H8
zu93dObLUX7JDVD5o(adsJv}DRsp9ePzWaG?daZzy^Vl^KL&(^srlRccUOO@@A(dLm
z*qBb&iLTM{0gYDve$-=~eQcq-;dk}E(u&{+WFIqy*-5#*T5iOW7Vr<YI|;G5P3n3^
zrs{A|$$$Cl(GN-r#U?=*?Q}x?>rA5r8=}2LY9~HBU?`@~lodVTW-?cQ%^Q&xbFMR!
z>7@Uo1~XlqYVT=FcC!&Zb_nt4CD^i&bz%uSU81px`&v5L3(mf2S#Eul=KXQQT`Xxl
z=n03DeC%DV_={>;CvwpJ7mP)q>$Fy9;qlxu6!!EBPY2sz3b!`Ly}-^To;9Bg>eb$7
z0o^ire6EzEN9U#*1$u$pVD4XMPW5h#42c31TOLWC0nwmG*UwiNcHk@S{A`ib(uuLk
zh*sNH^J(X-1FbQYZ*HJFM|f$`nZ<OHL)GlwG^0JIVA6uGkmy02-I}0{qb8HPT!|0R
z&`71h4IEDJqH=)Vv9r7JSA3AN(^_7+A=OfW;)8A(m13xz);FzhcXnJXW1fC50Vq<X
zLYhThoqk^3s&@E0DNERE|IiOiC)iEv`tM~q?p$0D9g~-`2JUwehV1|tHtuYg9LF3Q
zbqsh9P@?sil1YB}YG*|R%kzB1SE5-`9@B#_e`nQpDlzJ7(|byenFJa|h|Um+4vtJO
za<k{Gr>-5)d6dzu3=;Cw#SRx^Zp&1k3f~P;5U9lOrN8i~_?=((A0{z)ZB3<mC3b-3
zCH={>ySj%c3*R%J*t3C}_y4Bb_s4J{rNMxz+n;=!{wHa|4NGKfyQZuz*){PqN<t`R
zBmJ<;+9phK7xw~08*&}UZ<V{;DtQXTL;|Yu<z@6M)Dx=GVncXUlWI2Nvrh{n0Xf6>
z*o<(og&umKcd*SbXGyGPe1`qhN1S4R6Y-DR+wf@cK$zMS&6!>Ued-<h>vF~==ccS)
zVIUbiTKj$V@D2w<_nV=B9OLf@g~;ScHr=!A)bcyBCppznGgO0U4cl}$FwD;kw&|=$
zh&ess#l3nJF=zzNgv<>f;l?m$vfq)Bt;AF5%$Sta=$)qUH*heq3jdYv+Kxk4LhEPz
z@MQlOnMY!LRjqE8a&8F0a%=!|H1dY~X;CClv?(;9<h;ojFOsA3G{O`~z<W6<UG@ik
zK9%3KK{QeYnK_0eB!Ko-_H3y&a)M?x7^v?MhqCW08I`FHw(Zti9n9g5pPi2NgNwex
z@kiHV4OMo3(>?3Qwna(Dmm7sWEs)PA*aVnkJQTCFD~S0N3z}K$B@z~-WJDV-{0WFu
z`$MeaNC8ICH+CEexjpgE?{E;KBQEhtsnvF<q+Xk)WPw8dM#Rq?0a^GT_Kt0MCeF3Y
z$45k44N%?ED;yzy#b7=;_4;#Z4gSgWq&)OW3dt}q?XeK6jgcSl*7!{gKXLWk^%!Y+
zke--jr^*?K*juU1Z>V=C9Ij6;nDsA8I&au2q}b=X|464R-3qVKbHZmSw(m=v<mRmZ
z<rLQlBkbkj5z`wsb8VvXHvb|jvj;>wZQ|xS<eAqMlw*1i5q?rGJzRmjDVo_h@c5pO
zP;kWdDajBn+*&r0NRJAqnc1DiGg=@mzw-yP$A<ryZ4}%+gJRY%a<P0!iq&THq6Pa3
z*c;<ReGRxiTG0gpA(d<A3Z2*`l*0to-Oiw?`G1p)-bEQZW)ySpKa$a&HV0-oDI5fs
zZD>h*)7>%(53n!HhfTjzX9MO0wcxTqw(9FRnfua-ooP55a+3whx1F|H$@?f`OQK}_
z3o}Icjo$N1T5?|kriQrof`AZ)Q?7A+eX+>21y7>x!a6I}jdUs)h06y3A<lovhTyN?
zWFsWDf%M|@)Pk$O$wpu<#2IpQHvH-)NH(amRQ&G5i=Mc)$dpm+)1a7d`ARc}^3|s^
zc||jcG@Sbr{z6BsD}n*CSDEXFY1!B9Uyi~-asNMr<3aP)XSC?eKZN7f)^Eb$VI)&h
zaCp}`DiQ3A0zCf1IIJ*zY^X)EO5s5Vw&%Mcqwu$_c<*l3rfSAJ@^KktENQTR{31dj
zR0e_(|1RvYVzOhcDhDgN#*$|Kn?6K60fP}Zs=+_P0FKIpS{!P-;+4&dJcB#$6<sAF
zU-E|++ym-e7DVLJ!=WY##53kXI~;2*c=@`*jY)<kS&SN1W-JKs`0y{LfaX%5H@_4y
zzIRAl`t8NVC`$(^KLfX{?E6?1kUDyN`o|5*7N2sX7Nd+5CVu2kUfE#*^V23q*)uPE
zDRIyXUbNIA@;b!@)@^i4n-berbXLTuMfE$hF9x=@_?WzCHNe)tQDW5EI2iEglOEyk
zh0AGXdlV@WTOy$xre;8s%ZHddJ?;Zs5DfJI+72WMwP5D}VBhHvN%-I{3bnLEa=ppb
zL%{T52A83-RkvBvDE<~t?qOv>Mj=o67(ss#UI}W30JYUE1759+a}A%YFqRiki;os=
z8B2UM+h`hv8;1fif9oWSUXipG$l`gy1Fw`|ef3ADWC2q~k?)NNH&6K$JsRXngoa~(
z8t#dju3r91XIvEw$S{d7D_?l??Hs6TiF8m6;hTM9BT5B&<`M*xmmKh_Fr`<}6e}Z(
z0%ZQ;cRN=Q32mFw(}Pdn*%_{)gmrgxRaxPSVG0yvh3G|^qtv-+BD-;{x&<<*4%v~k
zq-(SqlP6J1TM~J&kJT`_fi@qX{0t{q1%C1fGq5o%%H2N()4X9o&M3+ElgyAcGOnhz
zZ=@ynycguYaVYoJ$K{y3%?&k$mX)xKuXW!>wX85IRIB)Of+bc5Z}C$ELeM28Q%n=S
zBNe7Vc@1Qlgu|y7l#g@lp>+(*O-^I>196YOrhe@u@#e(Do0-;M#Fam69EIx!1GTe{
zYy4&x9tb>3BhR=2!NhiLWRHlOk>6VfVcG&2ef0+ghr^qx-&AB`^dBI%rv-^YFf{+T
z$ayc?E)jX&6;yiO8GPI?BiJ_!q|iU;{hSs=4#9-HpWv8hqd+8#!h8LJS_@AxbLTE`
zHLU1T3!X!IOtpTGc<CBKr8U|Dx%0O!qp`#WfQYv9uP_6CGc^iU^tUntN^~gL>R%ZB
zxK%9m<N^xzT-`Io(RiAE8Eig}f@%CI&VcR`f?yv00nYAu`#oEhwA6pV`D(Zw0C^Ag
z+aKWk<zci=1;O0^f53SlBF}CxRysp>oG*XBohw)og6XKbUD{^fOT-UVFTcW)78OQK
zgrIt;gn3XeB2Ue|%L;g&8%_Z+mq$b%E^{)!(g=<4iL#<=JA;Bb{DDr;Q52wN*$UNq
zH%<i=Tyo2j1_jIi$3b+kDj>XQrkYYe8&jdpv%<83iLQWT7394!G`-r67%l$hUzMh8
z(c&yx9R3eEKY?6=94#LA2RX0*rmT_rC~0zEGS%{9ni$~SvtI58Hjzr4PTJrkXuRgn
zzw7}V2prd0{$~m?oBU_4AQ~`54zI&`C6!lP0}UWY<&-#Vq0x1!q=@o^*B@jcI_A);
zEtC2oSViOUv#6Y+f0Y_=APV#V)4x&;s7s5=DfnBfxrLQ6Zo|9IL#EBnipl;4&TNQ#
z4(Hgn?g4OH-GJ!CEhNfsUq37>m-!8xX{L@uw*&50;W{MI_=>@AIC<MKI`a5Ycxfo`
z%mn!avvnqB#s{2)3zgHp_2lG5a1ci`n4O~s9&B}$Lx>12Ws5UJV3u0m;jXc_BF_zq
zB&Zw}zmm(N1*<g-f5g_`xQXpKEve|d`~PrLIso7|V@@wP`Xci*VifK~iq>2BQ&b64
zAXgKSHwd96PUdc_Hv)|Z9h$ki@)l2eAg|Rp8su~at;ZNkIxeB05DvBqrUaKf#lb^u
z0RrMAdDE!8si(<xWA<Z2qR^wBI;%QhgHUYyxCxKgX@4<*D$!yumUMj8f8t<1MG-~_
zUYQf{C`6AGt`S^wat?~@;14l|0Q(}j^!PZU`g#BNEyB9nj@!<px!5tL=`f3?*`a*Z
zcC2@Ds_9wKhaJCl1ZGX7_SFvQb%R25sP;DjsW&j4RGX#@gT33&rEn*4&s!+D+gaMb
z^xalqY`Vfzi>hh-NHf%smn6FqWeL4KgbuBx*=Cb;)_=94rIovLqk8%Em=ZAG8>=X$
z0H*z5&CGCMi`g3YC+}e$SrIw!x1)(|p4)DYq~w$maz%byl{@4{XgbSJN)2u76&$7x
z)A@GV|9VpGPVUH3N*J1p&hydP)pC9_ctYLmy}33I;j+lH;`Cjb8#qjTATj;WxAH3m
zyBYQZf`4n6Jvqgi9K)^n)kkv%PZngIB1JPM>6&BJIn!KM$XemLlw3ULtrklAlC5HR
zz9h(<&y99c^4`P4lA|XhJy-A2l}VJ*`sF2d5B^ZO?dfaXbm^H<MDxDkZuQ10mg!{U
zyHQ7d?qEqXfFI^&MjUoW6x4g;XMYtx*Z~?z%m--L7ls_~W5U+E>x!4kv%9;P3D@(E
zl^($IpSUiM(w6xdI%-Z=Tk6=3>)oz6OZUS!$a4p`ZHH^};o#)pINVwrDlsCYm25k$
zh`tBQg2$fQH|>%0@rpiP&13cEn3HUFiY=s;)nUwAlRLPKc*zrJXEkliRq}s7OaAlM
z<UbGL&3JRjaOJO1=KXC*vEvYR5xl|n`0~r^ZOL<oKQGlNCazi|Z;f=~gNtJTs>j^u
z^L2_SURNJl^lFXS*^xyb@X5%jnz0yjc))>Da|r~fLjK#;Zn9?|Gqp=F?YmKkD!B0d
znJi*OBCVu-pObShE$t)=^8k6f11?1r*{D6ZS2S<iUgdf6DL=e=l>#p(i<eV~Q%ldn
zi>IWEamr4TT`W+N60F+5%cL<mw5=|+Y;I|C`Skb>Hb0TwYRV&;*MV1Nz6)eVHNHRO
z)rr^y2T)K5QcEAev)d;7VqIsKzlW9hI1UMOv|`5OV`#l~Ff)BHZr|yG60XDTUavq+
z!Z(F>^fePd1@fbU0i6lqso?59vl&NE@)*S&LI(^&_GP_uDBM;kjiP3g+dI`HX{0Xz
zx6CSkkb;uhk+x2$3kN?j1q<yheetOCP!tW?xVr`6^3g`4hcX(`V=0ueg5%IN#bqJ$
zzvdmQaTB^Z|JQ%!j;tv<l!Im~M)fEu(MrR)+7ma`v6q5_0l#f~oMLr*n#w(v6Dmxe
zmCi*z2%(_lk>}BEEGyW~UBHe!d2jl=FZZDZk3_Wm)g4BXHhl2^>^kskJm<z^3h8tM
Q|A7*zt7_jVzGV^kUyRXSLI3~&

diff --git a/docs/dev/reference/figures/op-relu6-plot-1.png b/docs/dev/reference/figures/op-relu6-plot-1.png
index cec4334e6edfcf5d477133aec34f7c2d68436fa1..0672bfbc98ae08a8e2b343f89151b102157e72bf 100644
GIT binary patch
literal 17089
zcmeIabySsI^e(yq1q2C6MH&fdL<DI`1u2n~5TvBLOGOZnR=OJ$kPaz9q`MoWySwh(
zp7XmW?zm^1GsZdhue-;m-}kZie&6-3HP@Wa^UP=AuP85#gGGjgKp=3QJ(YNYK%g8W
z|6roRPwvZ@Fu}iWSU=UULm;pTkbhA6COyv)2x`PLi6<|e<2NRawIYU5(YGT)ec13H
z83v#+zI-WRolK2Igz-sT(#DD<pcpqkFCfiyGEb8<zlxZ&Me#eI;sh!xZA)m4c=#Dj
zO-=ye4aRE0E4*`C+t#?O%AFnOsV0KJrfUd9bfBC744{Q1fD++7tGJIq-1JMuK)mK7
z5<(yvQ<hN>w{a!$5cSqft_Xy$crO})ll=exsGp;EK}19ZAtj}}z;%Qu+neIi#Kc4`
zb~Tc_1;&z8BqWZz^R1SXRlT^V2*tR|(>0E}ckjx|%4(QbClU+M{QmVzZ+)b&L>Yk?
z$c*E)YO|bEPy3&qbhRt_XjE@kHb%y!)(A#sX6B5H46Cb`s;cZ997{XSsPDWRXE8sD
zkp$9{1ZuEG&ZES2=I7@AwXOX6^{b4GjG>|7ATRAnPI~%FC8b{rwjaf4;yOtJn|$zB
zhJ3_Q4Cn#^0u~k)sK%e>&A;tU{a9A^Xd|RrKA`pi{r-`MB&8jCY)#b<BfahE!U73b
z3+-mMXo2jn{NHe=s1Wl_R&^JSiybj-ZEeQgiT)C_kK)(}{L`<W9=>@dP$&JOpf8D)
zBoLq0CB<I3FX*LL-uzMPm8|YA!3w&Wkiy<#=VNmD=~|Dh=P@tZljhXIJ$gxX6Gq+s
za@T2#Un>k0l9Jfhsxo+mo$)!F4cqNZ?ZVC=`xZ$Eg6z}r@v+%(erJ-%)y%IC%9+nc
zOWrEa{Rtu@B<$?$gujD=f?&lC4i0o{T$`5m>}+jcs;IQLw>Rk*1_a31*c1i_%UfFJ
z`};qWlJfodv8g?ZgN=<1zJmNaJe-x4m4J{?PEHQ~`z8*KZmoMxT3SR*%xJb!T0?^v
zu3w{3F$*(uNlD4>-X6})n?=RN@ZtXce)zibix;eHY`WFXN`Wq!@(E;QWUEU{<$8_i
zjEq@Bd9Reu=M8m=RYJrX<%}-5%Hte&zgjo#ym~KvV6(pd(E2d(z~gVQ_Wk2yY!2O8
z!Ti{Rwc+V12ic(*6j74Yw6vO<8ag^U4|jK3Iy&R7`1>b!C@6$HPWQ*ltwJaS4o~)$
z&@ga2+S*pfN?8J%9`N#(|M)@eW4|%R!@|<u+>DH&H_gE{4r@dEEB$F{X&HC@0|Kya
z-u&$EAM?Pl?E?m$FV5Y_*x1#@Mb#HC^erqpm-cvgcr47#v9PfxCnw?QeDSC#si;I8
zRtMIG3oy{phg=gsefsp|1G=PygtMEQw6t{p#k`?#n(^A56(fdwUkSt$wr;lN*Rj)C
z3}DjyeLoz7O243>71lpAq$(@BR>L^_T|@ch%X{tB2x^vam%@ExTiYUEUn%&zs>;>W
zG`Fb8R!Ryzs=c$dRj+u|#l=NMMWv~!$^6kPYJ4IRk|Oo*@$ux8AxEbNW`>3f0|PG|
z92{1p$aqY@#K(Wn%sgBjWT6a+`ueqL)_Qk$_qV={o}Qkh<VST@c6N3KpZSFa8@~9C
zj_2^1$jHphOoqv)Y>O^uBC2X?QM=L{kvvLI2F~Rdhjh3-wRyiq?KlO$At^YsSonO&
z7G%C8N#eQjCsQu&kx37=#_wwg4Tv&3YwHTzg@ClQH&35_ii{lU%aA>Bzijx3U0PZS
zi<VPRAYwVeM@Sg)4h`dx)zrl3=;+YDS5s5Q+x|>pVc|c2{tz-e@9FQ4<1vdYD?3b;
z41{2dWP6pFkwHvE^m}P<sWUF?+cyD+)tBPpje3o~8tUrK&dw!o`~Up<^{cnHQ01E{
zB&o&4#mmb}Utiz;Ou7Ev-WU-t;iPWeB7<iei#KlEu(7wVhH-xQAnxgTX)F@2lqN-g
z?_SGt#ML&X@0e;LL9NqpVaxG+ygrKTGx6zEi(yZ3kDqZYUK_rssHhJfJP;ETGYBI<
z7%)G4xPN}KS6Em`OibK2Xy@Q?keW)XqB3}dYd>pS3F~G4C=Q>Nz@LviSd@BZ$2mOB
z@Sb1AmoHzQ>ad^WWM)FFrKYC7P*u&$&Q{mZxIg;o{rmS%@6r1F7&T8(Y?-yb*WCIg
zDvE`L1=c7xH<yHzbl8=WjxG(Rgk7`bY0=o;+cfErL;)wunp(=5@R<H+>MSmYT^$5<
z$e*CQSvfhQbzWX(W@`xoPE$2*QJ3$fKD?V-MiE`yocyV#qEc$RAS3qlXmhf&qvQRD
z5A@>MXnYm75C)q&bIlY2j$d#$*4HV8TqU>^#6Lj6j4>N4d8?%rZ|@9WfyIB((chDd
zyqdX%1vWNz6o;-Sya5C1isfP9<P3GpM!#mZkDz`fDJjXKmfz9R@)RN?FmP=kr+>8A
z_{3d!LkkDf2tFZB8`d&BT(14oF^XNA-(}ap&@dw?2uHIv7^<_VFQtg5xv44t>w3iX
z>(|lHK8Rc%&mB#BWl`k*deV#Lebdm$i0satXMXO&b;&b?!xgA8qR1bQj*R$x{0PhN
z%r_+^Wn-)~?QRZALzg-tltwsd3tp?pS8QV9<?+rO47(3^a0dB2ssscCki(cs0->B0
z=++4vcaaMS2&_zsBzGa-XSk<kFi~ljH9wJ^_o^Ji%yD(V-!b186Zx=&&z?PleepFB
zMZj{R0v@?4jEwa^fQBdo8Lwr0Xz2F-igK&<+ECu)Pp7Qd?AZn~Aw-LWp5E-ITXg3a
z7lVU?u%95g)tL@5kEM_?bu6PKA4*C|RsH;#m6i3gqJoH&bZe~C44#P`*C^6{#ty@K
zjMr&%!fvVSt%(V~m@h5`+3nlT=f^hib#~;}BMl7<qGMv7X(r<Leb*>HKX-?LAra8P
z1L35kgakGUtdC-a>d(3G;1QD}?=v%NYG~wX6lX$&CMUmc`*QySI!-D@?vq9s>p;<~
za?70TY^@6G*{!;(#k~_xPfysuCE`3L(Y<J*7W~#T^#Q~zrw41KL`3OXS-b`<_$7i4
z|K1y`cCk}eza7#{=R?kKpA9v%wzd|&X^(t253Z`HtQ=7yNaNEymy?^T`toIu+;u=^
z**s9R>FMbK352K27s5`PeCeHW7m+PjY}C0rU3c~6zVdJV9f*6lqMX33e~<A=B9-k`
z8M%Pt_wV1$8+~!)<>g<MnHQFnq$gkl5LA>^&C^uT(sJMVD~an@48?VM*`%a8z8?9E
z!96`a8M09_`C5<ke}7Ud&;e}Gi+kM%dEv4j>fg(^FutXvC;?FA!g~sFn(Vl|vNBoc
zrMt_=&AkT|%+=M^p6=bXA>>=E+Ew=REn$%F05<SiPK;F8&}~R;bHLKfyn+kk+<jyY
zcuGo23epMey_?wBXP0MNZ<pSC=}F-srcGxXK2lOrhMSG0XJp_|3XY7A$2l?`<H7FP
z&I1%iUt=3nSZLE8$wo*(uq^ey90oM=st`tL#r8r;>0LWz3o$IV`!~n~1zMFa`VPu1
zCuI~Ac)7XLD6lR+@VFw58BpIPLjv_W-|<UGNQjM%efRDcB#O6w@10)57<i&0BKl#Q
z2smw4R93F6t(_eolk-{Ex3qk6WOv4Zkto_ddGZ861%xIGjR8DvjqAbg;o(qkZ%P98
z)Vn({sS|L4D*F{$mky)ipH5qzTT?<+Vafc+I3!Q%_PRJNwU3UDHqqA?<HB>t2NZ5<
zdAh#>E9bVqtfZqeb+|DuD<c!jV<sg)RE-B4PEkB3Cnrlj;l+y=u(#kETcgEGwH{|C
ze(&+>Kg0KFX=wDeXMWYV9Zf-<2obq#F)9Y^MO(wh_isuDenXV<nve9<xH)li7vCUc
zfNe5a?Ggdg%ZVJ91OQ4vZ5iS$EZ^qm4P0$zl#!uB2y1!k)~!B02??LlQl2Kk?HeO^
zun|05c8eWUp2xG0LrF+Tmbwy(fBg7SRFq1ACHy@&P$&9qf2`Ck#+=UVOuqB@_!WAx
zC%@O_Ic#PP)}b&u>G*^MM)`OXGc#EMBKI#HQI^-jxHSa|6!34+vVX!_ol5QhK|k(z
zO<}e7kKUikVOotk0mMKR0Bp;t>XYT(G#HW}E=9oS&%aBV7vk3$2O9$VUc6I(E^+@S
z>H2`C<5^quA3?DItBy5cTZhkkWPTwOL0`kdcD835j}o8D$$fRuK^&Ke`{A<bRI7`N
zf56wbwA@@;a$4!j`1Jv!zw0`ciqkj7hK>^QxY`TWdrkR9XCDa7Mk7!QKd`Sg)+I$p
zJD=<>{JT`sK{#TWnVnq~_A@Tzw7`G>rp0&Nxq+#qhSS1srvVAU9|ymE3)7&h-XHb(
zB3x}-(^svK76N<!cQ5j_Q^Sb%ui_fqB}E9gj*hwv=WExU?_$$PFcDGmS-<X#<tY*W
zmDN(cOw8&Pz3RYhsk=S>?3q7DU!qlFuaixC_R7%3Rhf=g#@)?F=!#nvRaH7w_Bqla
zl&Y$#M_bcWBA$ZzR&+}SI{W6z!z*=FmtOY>Ffsi?Nz%kNIV_BGPj^lP5iJ>CS=1oS
zm%%<ODte);yw~=Xr9$ZY`L{Vb@=+~z8lT%G4Pokt0|j+;b$xw(XJ<Yd309J`lM}#e
zfO2}ft|z_YeuFY2`|;z)le05f%3}5JvyFbbm9`&EX^&|kxgEbSH8r)OdabOko}HgB
z53x^!+^zRHI0|4#0-d=f*4N*U%p2uax??Fbd)CCL@FfU+i?I?LJ3D1HH8o{rY+T$&
zJUkf?X?VbAlNd)2OkcmAnV*N0anLPt1!?eR`JEh&ANFX7BvLXmQ+<5`a`K6Ct7*wV
zl8fy|JlK$#6j-%Hu<eAmE-%g|s~jS5e>$u=!Y&PPnfV5pZ)UHfvlFI`l|tBczont!
zy*}kJGY~uPZh*zQTT?X$`}?rJ2}nsHzf|>SDfHuB*Tq7P1`=h7<Ax?+)&HE4q1tCD
zBth*cvHKO((1(L?NAk8h&x^4#3%BK-lxfd1HUWY0;bBrX-8$s0mGgCKV4n=b$i991
zHa<26|L+P#6hdGaa6o-lwd=v^#zj7CySH!O677lN{QK5exs}&!!}S%;hYyi)jr{p*
zV^+9Z+#?I%EEsr{>+9?1Knx%(AoG5Peb0|vl5F_@&P1V~+1c3@717OeL4d6n+P<=q
zSl)Trjl6QBjI3;PFa^w-4#1xjMLt-I{eAlqcMl(AI{h&+qHb(#+?uF<{rX3G*+_C>
z_h?6^>>J31RflO9AI_S$tpqBoPnpg2?}u=)?j#XiMMdg6mSZKYot;6J_p<MNhez|f
z&%|`|&VzJhX^Va6a&hVao7@Ol=ZFTbA(v|ga5g0WwZ4qNprFn`QjTB>fy}%-<on1h
zI;EznT4vB14#^qNB%xSWl1QETXb})12u>d4aGj@t1(B1J-(yq2T$%jIYsvkh09ivk
z+=Z_?<9Pe@G&EwM=0chag6y|l3w0Kj>U4dy*e5hRJlvm<F_`>uI6U8Dc$Dfb$kSzI
zWdO<ndL9D**xtU)XHC7KsAvaM>)Z?g769i@#|<RdgXuABk92W!gIJlqCOwI$_hM#Y
zIocQ}B_?h%D!xGz92`t1^_kCUlYB#KTaFW=?J7Mzy~K1-rf=5=-^1ggsOUje*vvR`
zyduD56cPn*1vZ82LFO8d`uWZk8Glz0fBP^BAy<2I^Yw-HDBzT&tm>kWWm%9jT<0S;
z3q-ZCp#jM4hxhOK9oH*7&JG#mVtcx-Ph)Yzf(`@d!^XmrkLR<2I3Quu%+AeCU5G(}
ztNY;75@A_QRX?_x5(JC_9ANO9T9U9wEKCRmJTWzLk_d2ca8^Q_^lkXa!=6fT5e$&P
zRN;#~O_K`3!omVJP2ULh1@^#-O{8Lh9>nPR+OV$I25c<26if*W4v;V2n18b4Oo17K
z8qhG{E|IJA;_&c3c%<=CGbb3?VrLxW=~6TpGc^hd3ZUgT@O?}gNS*-Ov|-txIBI^l
zBP1%$ND>J2pB~oS#@d?c{(VR|S#emVw;*KxDP=emflWm}euPdK`QQ^!P~^enrl(8r
z5)aVAoISKheOEs|?gnY%n`++Q9cO-iekeqcLQDw(4d9GHv_aXvML{uHY^)IWln4;%
z_RJG)ZSAqj<VOH&dXlfBqg##^Wk4MIgg%R4sH{M~W(4p1{<=y);ApC<My4;^pn=>7
zIpev!d}b`BDOCVv@{>1(>ZtET`GHc@SK;izW+W>CHW+l5mt91p4wy5jBGVp62~}16
z5OVEIL^T!ED_s%hdXsfrh-wyCAz=5QOgL>#0bK-!P<M4H0G~W#BJvnaC>6F73tYVn
zLYZN+v9W;}c64+EU;~j1A&rZPSx=#V=h>n3B<5VyxBzPi>8FhDPr1w=1#W(A-mSVi
zJf2Xei|n97h&sqdvOcz+@h&XnqDhBrx4F3q)j&Pj_}(aKjaXd0_!*XTSJjyLVdjZt
zrXX#_qw@A(M^Z<;ftTj#dpnUL?#J7?D&MBu9<Z^=x%waw?t#_Sg3sDz)nC10ymv2o
z`A!adaGF}lww6nd_()sfoM$y*#f6#`8}W5X93c;nPKDaGX5?$>|1P7X#L2Tg!@~+s
zpN38mKY|yqG|PLTVO9Kk0dGd;mLj6GJ(K2#FMPh5#*E|#PeIkyZw;r1DFC&px4$1K
ziEekI5R$nmQWHA37_r7p+qAv2&hltiwLLK!nL`upjo~(aWHI*Js2B>s>dXwMUc)s3
zf#-F%zvXkhtAAm&QEpqHOHs#ys9u1S<#xFK*!|cVb^xe6r0iN12CW2RDWbmC4Md_J
z_Lh>ofI}4mLHEIzmTtzS;D;UP=g1Ta#ZmMFvQ}nf&{LS2nJq0YzUV+J@=e7su_mXW
z2qAw=>%xkUDJCv1V81NNOVC2Zu6(m+=6t_j0k{T`K?Vi}<gVg($PEt0g=Z0Ga`pB#
z-1*xCBo<1j2Y^jTPF;yY-!d}+X(FZLWE9X2?z6E41q9IgeEIs-#o77z^fYz(4ibgp
zw15W1C>MLr1&GG_+#Dg$QC@-p6c{I3Dv*06B^nBV9Di-U;-DZeb%)P-=5JFVy>?P!
zqUZ7UEWm6?eh>V3T)ojHG&D4Tn*)^tb_%TjF0bY90}C=#<TFzfVSx$;Fl}RF<2vfQ
zfPgmO;jo+e-)G$QM%RP2fm{hCaqsBpYOh-aM3aWbMgZ{q?_p~2EvIWYR#%xRs;a7{
zsvH8L?#RM(qyHKi;k$qT8UJ$`8P~JJ7xm^)&rR@wht^bAKjh&VFSEEyOzZ%o+*|5~
zQ94|c#H)7)C<BG>(SrxELT+WCTb&(lK-vYuCL=&PaEo;k!Knp6%+F66G??MMSERRZ
z130>U`}XmnVZgf`$l<Rm8vXF_DEJ3}$>0NOdZDFNZ8cpBg`}_Rx&sCzFJD=C`Ld#-
zaNNA_-@_v!R+><-p!D1{wy^LxSXEisULE`fA-w+=ab3L%<#h#ABMN?dlcDb#uv2SY
z_7)=;71GP2-lL<!wHSVSU0r&kpcNGrH8eD^>r^k#&)ct$a6t-%WF)VW)QerTrOoGg
zs&B9!@#Q<MBb|rxH{40Bv%{@H&AY{3-mdfl0&ahQ`=1?c0cT=iV!937H8K)J<<tc1
z)iHsE1gE3!k9OrvIVEd-DPQ+EkBk=i|E`n5lDN6rdQQ{7503D{A?wS$ReKL{@VhCH
z)DMr!(#S{(v$v-Q6AO#qdHXfYArQW*L-}4Rq0YOtM`Kcx<w+$D%W{}B5}_yEmlR&p
z7PHg1gl^uaDNq1mU-YGbSg53Qt4UC_UBGPr4GFGAfp%!vednzxnQ(#!MEe`(OKA%z
zt6>1fDEO>JnTTY3f!|3N=+rDrp)%)ge9k*HuKkd3mAR&&;)5?HlSrjIV>Ed&T#?Hk
zf{vK}GchqSH8llVeqZ{tuC6XE_Tl#SKu4zJ<guFxOJ$Gae+pYyUtD%Y_?qf!)G_wk
z4CQ+<*#GjNM_K(!$SD6E0NbBGQnIovyu3#vg>OI>E)hqepi&P|dT0|<rLVM$xRSK{
z&3uThP!KZ_7Lvw^vKsOE^JhrpzJ7kPn-?`kow0yRf~iCj9odu9G(QCDXlt&&ohvCa
z{ri=)fNFiy*24MBQu*-rLg3qUG;cD0fB%)?f+S?s0LV}QNObb4sL}z?sWdOQ>%|>6
zkLHSC{@g_mlK5(3vSd$lD+Br<dxRq07R7(T-aQBMy`)i}Zgej!<wa2JU%l#6vUv2m
zDTgO9z~_@pfGk%bMm7e&)2937`3YdeTVg;EVVny~OVR>FihEsT7gPSiN1D!Oco)CA
z)E74PQYD&NhvnBN?PxRhiY(~ztal%JSG!%FyM#~*{rP++Ejc+kJDbnu?{A2eKHTdA
z)yn&Xj$7`|?t5f)TQvl~(Szc6xRwHobp^dHY2)53JhC;N$v;tD<>Md`ow_((E7;lD
zK|{aUGdL&&x=V_<AE@_l{ZM14gjRhsdQIymL@+vt=Y{Lsj^3I&a>;6Uf0q<k`xC1r
zbAp)pW?*0-AtePGqWpvPo)mFK#hx;Y@#Rd^Dq;6bmZ<5HH{F>OYkb%U_hJZGem=gZ
zegKl=1)SKM1gY`h4w>^ig}OtDUTn1V3L#xb8<$WED71Y!L?T{F-KRjf)7#nYHHT2i
z`n9&Uo}Qi(?Rg72XDx>WIV5?hX(&`KH|z^@Tk=rUm@XYIG<K$D?Z=X6dN*1iRzCpt
zVq#+Yu35?qz=WUQ6*3TT>l6wsCO6}i#fgJp*}se<%sO$Dy$13l;VDwAV*E~)-L4~>
zEe=n%ge$O6z1_d%=8hG<kp^WgM>&g@mR7Uevc_zf9dy01O79{kqkSn&4#F4LC4SYb
zizv9_$j-(MYnU4vl3SutAN>iT($&$a1yM-I?a;b%0=7`f!n<nTw500vyg=z;>DsAB
z7S00Io-I^!QxfOr6G@ED5#6J6Kkgz<4grG!tMK+lh<(b;Tm(I$x~67wV!~tz^>iwx
zcg@Ybma?io_~87IMQ$zg`)@D(zkAPpx9VJklY?mUyb;@mKy3^yE!in@b8=Kb0t-a5
zov&$R(J<|tVBm0yd(L<hF`xyigcV3X_?Q(yLhlI+Pa&lkrcms`;Isl)XA{Ym4@a9^
zO%@vhR*EbW+E}8&pz6sd@DD;qqrH6%P(7#<*f(y#jLwUdep4&rlgTx`%IaX3tF3Zt
z{3ZQW1_TZ2W?Ng^^Ye2cznUPngBSynnRcaZ)5u7)BRky7Xa%eDtSfr*D36guer#sl
zdDGHA$uRB6ixe^c8w3ep@Kp|LQkbKacE<Mh_R#qfIb$}V7WP_Wy8eM7=<AJNJS-^l
zDtdY;>a3AK{SEe)djUG6-n}v3wluWdL>zNPF7%e`O|9+ZBC0oss)`EzC<^xWlfmI-
z;)(#7E2SaxKC8hdaY9z>VReKH^u2(LB?-FF-qESBPJzY(XmDCNwNh3*uMfK)h9g9A
zZd`kRUpezL`QyI&`sC_rEp2T8`Wq$U@5o)UmXA8HJ`uVvmsdQxWbu79yuU4Db9pi6
zl~56p)o*e9sf@!j?xa12lr^GDeDQL9xa=10BFgb2g(P7LU=~2i_#IY>2nmY~+ex4u
z3Q(rDT6wC<XmL@6>YPn`kc`)Mp@ulW=5XzB%Yx+YOrf+!_B@}5Z`A5uBa7g<#ufPi
z_wQKz`toSNP$1ROFa54{-qDBT+xIgV@e~q+WrSz$0dKae`+0>}{Z52kEXzxsxL%`s
zX2ZE$36(zke?_cn=&D<`q=Q1Lw|5uXrGv>`wpyd3Z^y4=P^&<b0a}%6!XQcEV|@hO
zQIL}T5%$$dkL~l(J;HC1m!1rdRGiKWbsyrz50XEwu_DZftreHlxUTS_nbyZT?toJ0
zg~@#v*+(JGxdfkn`?$|fV*x1E)YRM((tvIjzPq)<jbs{V#BOQ#fOL#1JyY+NOX@M(
zv|`yGj?ShrFkjy%7WOXTQrqh&iO9-wBTp&qJQwf#e|P}~@5<lpo*Mmd&Y&Vk&NVS?
zxM(i+iSz7N##|g};={X-E(pg5`ud7WOFg!p0r7)c5j&4~O-m9uSoo%SW@cukKZ{Td
zs3o+Be(UQ^lUApVVYwJ~gzd4BQmONIWY8zcjqoRUp1x^4nq{;cawf}L|M9o)*R1PX
zn{L~y8d(*k{h@_s7L1Ehr)qkVWt3V;8JPfRn?jcfL`D$jDz!F?4BJioP%D)!q`Lp!
zXC_)C_4gmEQ}`6Phb^&g_0}L?DT*ub=Q;1tdfolpD_(Ov1SZYB2M=s}QzZcl6&Dtc
z78%L{34@NJJV47dQ_1cPLD$dqbAG=n=k|3MmvnJ7e(KGQbv%(4cr%Ra9j{Sj&^qmP
z#WwnN5Y*Fr?VqP7CxvxyRna?MUI?U^a0~4By%?X!Uc%yUWegU4814MGuTQ=L71g`?
z7S_i;G!i~?dEi5Z7}&zuW1njU9mn6FQK)r{jaZQc2_m*LM$8?Rl)`{2(Ma4Q0BUsS
z_U#_r>$~vT#BcN!bq-GUG(nt*ZDM4J+t>i;HRyGDVm$Ig55yD|#d<@dS+ra#9gUpf
zA?LlBt)Gr7Lo&Kw=^j51JiEwNp|h~|yN{@}Sslnhc9gz+VSQEB@#l{bMD*d|A?&&M
zT%9RFp1sYvMMtlr@|y$Jui3pPFmK#=4AcP1;lSV^^px(BkoZRWDP(&dv@uXs;)|)A
zZ`NdpOO4FN3n0dT5{GabcQG+D6NpJcv>&dIuCJT_W0|7z9w%Yyso&jf3M3wUMn&iX
zV+YNE0A%)<m>4KU@r^X)h=Y)r7!7mt@AEbq8lxNI<-lR(firNgzrLRB<#};13=PY%
z@$tQra{#~`>OT;q4?ZY>wwx?{%U=S?zN4a|Vq-Ny2ZpXEw3^TMSHuNKTX;a*UyXoX
z5P)KxpN>KJ;gOMxYiogP<MqZseif6Th+0DV4{wIPC$tADowg_^pHg4dx=R(j-u$ci
zu74Ae?YgD8r(YPYN{}rr%$Y{IG(E+Y;q;2mfl@lgYV428L}Z+cz8Def<b<Yv28K7Y
zlhY9h53-bX?kBq-l~oY)xvQ2glGF+FjlFScN}ujBw=?;!8c9&>)jLp|Hn2r`*VH@W
z4T=f0<dm(gtwDeSZbxGP48n>^5f74VK<#eQ9v5z%VEnXN1U?kA=J>j{&9cXYuO9DX
zxF;Z%K`}Vm+2P{h*#PhWEEyy~uhW$ejD={`1*d&M;RK9y^ffQ|^Pekp4RCY+SrtU6
zAo<D84wx7EKi}c*0<aF+G6)}IDaoGMo_;xz9<Ic}tD_6QexNuSF=yoPhFm#hRv&ej
zAD!FyF-0G3uf2WOhLJo23RDDxT)3c7p}M7|CA3YT?fX*v$%SIxvZTi7g!~kx&pg9K
zH@y#2sm&3?{E7%KCXLp+gAkBonD-A3z_)Oh$MoKvTGxYI&>!sSo_HwbEk7=sd$e_1
znhp!$-cVsPm-#%V=uI;&G_65Wn}VJV6BAU}Y${><Y6C;fks+115M8fujr{EKf!eG2
z$AoTVbBi;GeIg<v2?+@h@y||AbF;HYMn|DK5Dk~V<9O*|Y_6o9ycc?5n9zJxWHB79
zqFkj$v1^BWlO;722U5j6)Mc<XZ~?R8wVDE>$Kb#~B!@0L^egI(RYa<HPk;5xsht;l
z_)rGoi><G_+4rvx-=BB2##4K~iefSX*(v_9ouH7=4H}E_G7rEA(8Pio2h74X?eueE
z6Kw<Y>??&&60{8stuGKI_wL;TK|$6Js1$IOv}lag1XFHToh2GHRCJfmEjYJFdS7Pj
z8?N^{m=T^)u6|w-M9_m8Q*ATn3wy}mHX-4MHG)R9oLX+xyzw>`X(#ln%i-O|QR09q
zIe{R4Tf^t1JXoU$Z+tNZ`S|6{%|z%T!5Fu<x1pu>`0-<CsU5WX1y{}9<_pGiB-dTi
zKqSY43IvKNG}-DdPrpH)feF#5cILtk%-QC#7IH<Le2tGk?^h7X)2+J%8i;{QUg34+
z36#BOce@lw;_g~Y7#;KlAMo(d`>1Pb$|nk*K$Euw!UH<lp}JtF7#kbwODQXJ1d!Dv
zsHgzi;k+8)4ruPDH-?6WA{lug+5vF@rKl&!QZQYxx2|K8>~Btv1B2}C4TCuFvbGXM
zJa&Ng5{T7#AYg&M1I=!?qfKacfDXS(4>@4I7qT`krQmI1;!+^EK<9}-*KBSED*f;B
zl-jlKF26tB3SUz$8eIqe3s^iDE?$6qj;BgWg+;q{0<O>pone?!d;`(zG61?ctZj|U
zo-aQ1y^5fX&%wa~`==J!-FA8T@}+^56?C_Ap_PDn0~6C^IG+O4EnVa@cD=uj4SKEX
z;rhz@y8G4T1xR17fT_a`n~^tM0CgRZI?&vxn^!<BAb){K*C%)VfD9U$^X-rywY9X;
z#nse?K}=NYK)dM;6UqT8j6<h76m|;Ig>-)r1sz)OEm1Iw(o$03At*^so&ih@OA^be
zhYEwK(+i>$4F-P_^sr@hb-kd)2SS5cZ)y`{eURlLTu*U;xVzAh1cF4brK5A%&QZsz
zTU%XS?cU&X!^Gv-5jhvPsHnP^_MkSK8W{~iq{F;i$H0KOeU2O#DLe}3E`54%V`$vK
zxc(r=6$t1GNRNzPSXdZ{_9@XAx6uUNpv;4)@e$<nA9{_jf0UJ#L!i^~_xFvl6z?0W
znaeCX({&_gFA6i#5;FeOZ85b4{QQzUnpga|YaN~1$H2&_D~6l#@#D(u?5EWD&Mq#n
zu6NYF`)JafW!$wnJn@##)6V>^B|J3MVcL_Q6)sz#Iyw5#cJ@T8uw$A^^wM7<l}6n6
zGljq-K|w(V1}R<Ly6kLBHoQ{FMQ&H6w#%=KnwFL2PRS@gj}Hq(_9@DC&$6#Ss_w4k
zq@$aIcG8xRzlU=DQ|KU+gKMU|oY4nJ3v{la4IO-&>%AlUYs?tZ8qNztRqQOwQ-D>m
z8{b7pCG3+5n;7>Q1G{TDxwv)!F>VRP#Kz{PrP0NuBL=`|Aj>Yz`fG0I8Y)VYU_Ivp
z4p;ezGPx~_Y3o6cln9!TA8b+<rzh$CA;oc28;3C}cR{IJ<*mlU#B6D43Dr<X@B`_h
zfDD?HRFISN4D+$={3ioPJVh30y0O;$`t>c<Yv1BLZuES4P2`mwYx8+2huN~?8k>en
z!g%}XB+4pCG#`PzfuU*bY^$!jtqnpRY8&Z>Vm-No^#D%6N+?C4u$hjr%3B*u8lQ}Q
z2Tzl~$Bb7Ko^O5|w_aYL%)ipongDlzW|{dY<n*QPB-n;Euv<ZkNE;Zm*`JU?pRVB^
zDzvrFOFKdj)zvR_hF&}r!~Wr6MjszPzXHutQzS*OLrI(IP)Ec<D8)T|OZ(PE>=R@&
zKm{vOsEg2o>PZJvkCqlIML-i@={2NO?ag3eVG;J!9>~7fCq<^4SXj1BPFWOK(dd$R
zqRJo~fG7`naASFS1T=CvxlRDE$l!%6mAnYe6-XUJ!^1~MN5;jYEUMozF)?2k?_?2}
z-pSXwA=(KR3wG_Q^wd-_OdBUBs)r_u2gScoYAF_R5%%*-OKU4DNWVvFULG4spd{_B
z$(Qv{VbAbCeE1M_u=GwHpab}rKP=ITysd~46s#oRl`{s@CJ1@{At6)`3|b&(f{<W@
z^ruiTDkMVEmvVJIoo|hJRbp}rAHN4U9c+#mNE}4w(EVU%w}h4~Ij-?wt_t`EOk6@K
zKnYaDCm?`~PXZP`A)y)o=}=4m#;b=?fTF2Q=q~}DQUDI8H<P+KSXt9=--yPM#8Vs|
zF@NOO7&axwC?X=l%c};hXiMhToX~bq++HiF!zJhI0pYN57P@N4hI0uaN(7|0)i0vd
z>OWrpf-d2R`60C6{c!F&!VVTrQV<~JyaUp!Jezja00^GU%mMh|b}WW280ctFgpi)#
zY-J@SrP<lp@MZv~&~<=x4qB<bGD<PB8zZO1UJi1kBNoy*ESg|CS#<ztk3Ix)9)O9u
zv(29hb=PL@>aCzB%Lm=&(%+n@gcf68PY<XVBOsz~NYuN7-pc<%U440N%?yMWGqbz5
zZgF{@JO8Kj{RDE;Ko}Gmq)?ropYIde*Y|>yl+?+|32-?m7AHRdH%kAVn?rVNV5rcv
zfsP8w<kKEe=z*+$N=?;=Hl2NRLPF49da~uepVie}{q`pW#p4DIxbPq!7XJ8=Oo2sq
z4W@G7BN)ygp+Z3bbpyJ|P@p`4p@V<bk=+CzRNBx=$R$8Tzh!4XkLFAPQ5u-Y1(aNy
zN||hKaO|{9KoAZU=vo;YOJmyD+U_14$US{((wBZ?L#sy-d?VC$04vO3!`Yh{8y_wA
z(gnCAr@&rm(E?I3H#e8jh}4nIhCb_4nwUUqn%WLTo;whGfMT9Me-7QYl<aI1Lqkac
zqKDKFy`xn~Wqj}(Xxkw5_UEdwKq&%&1%xvk>_|(50TURJI61AG0*GOTAg~~bLmA=?
z&h_Dk20>^F0M2}k;y=*4Zqh#hfdaN4`~%gRjGRaa%w&d!cgV<O{QiR@h&KAbw`knl
z+*T~4FlnIlZVu!SHVO7{LQHtyl#Axn>q&kBO>|2bj<a(`V4y5uF3{Z3F(mP*S>WNO
zVC|E-LxO@XHh*q8u8-(};sy^R0S^=NDIlN<2)=>Ay{LA!`}gl~zODxxjr=wiv=tuM
zfZhTk5_^;^rjm*ZfXMxi*rcHMqeDxRnvhSOavlQ!3IkLt$RjNHn3yyWC}7Z8hX>wJ
z>=8v4Sq6It2XO0x90jNs%rj7ihbwH}09<${1|d)j83$hDqobuJy<(t+AUo=WPd>n0
zWgC)m=nw@q(a_L@&sMwafp)Y2DTSB#p#iiAvJJtH;QVt45~CY5qyt<KkOoSh>*|UK
z32E~`mzCw_<V=pnK)_~hhZqCPCpj3zF)LyDKoD~G@IXE#4m@Qk5TsS;zXR}1dSnp`
zA^G*|psv@&$2&qj@JBQcXe=l+Y^<#DP`4rZd=#4n2Zqf{dPOx1?$e_0=}|7IHV0Sl
zGE85;{hmU7b`9Y$2i+BbDpxC6Ntj3+tfHg@w7VT1!er^aul%7T$8&6Lev{KFx03+P
zsvu{pD-OYC-K)of6Ek#-0aTDwAca9C1V&foV0MT8`mq$Wf57t7mr`1K0Eq>lL};i-
zFUv!UvNb6)Lbt8zQUzYeCC{BT_kS;F+fh?fgICiL4Rs(A`rSI!&NlVlDB-gpZgX;=
zxG3@pydfTUy<8?5#3Xe=HCpRUgNh?S!mb6aiG8r+K<ZC?Z0GC9<Xy$T>f0~89dhWf
zr;4x#mjlub2F65r*dNcH**H4ZTF<_R=KO{hf^QW)`4M!dz#<0jza0qd8tYjx9UUFm
zY`weGjL0RTW?=X;I$8;^+6P}asm^g@3`i{4oxHsugPJ}8hd4+&_SxY5j*FA!!mA%g
zekoZwOCC@b0oKRxSug@YSj+R=s`s+Kam`!vHFVThR#xWc=a-kGzkK;Uw*&il7flim
ztZ5S!w{G7CKUaf(VWO}{6^vc$5HXJh8?_dklC8i1A*zrF`qL*&=*WUi8*=eBE|E|@
zuYX`5SlIvWorKRq?*NS9&v^;z-C!`&U=37JRh99B&Mi5vs>G98$aJ;-urM&Ua2FX&
zh!|{Ac5tVG<p8l&KY%t92l(u8YuXD5wEfSa69oM&C~g0IYdoX`1P_clZbMU8K|vu!
z5xB%92#3gl2m^pO&4#=R_!O*FTI^Bye`ja8K%4-_IXDCU(+8Ge4?*uK6y(cv;F1*@
zs4a|8B5*v8CY}DlVz53r3r;b5Qh7tqUlivpcyia)FfcHnaVhL|>E04bqn?a$0zM3H
zGH^uPqnU@&4_=SbA3vhGjpae6g`plJ;Hz?=pRjUs+kwadM#w-CHgHymwttv_o$Ou;
z%|du;0JcTytYR&&N}xp0)iCmTSN|dts#k08=y=G@jkH!u0N}Tz^j3hgF77-KqrZRu
z20wi3HkcNdT|<qbopx->N&+5U@M8B*PuG={*h78+=8{f<HIoJ_pqm9K9NMK^TwHLD
z0{9LEBrYh1)3ti6aJN33`r#O14_83lNRQ6k9~>|;RaEem2)2Mdq+Si*Uji`39WH$z
z13uFMq|wH|Nxv!wAq<s&n;0G=owPh`=ASu6mC~EzT`I;gmLR3Vvu7)SbHm{e9!Qrq
z{&z#|K%w4tfWsKvMjg@63MCMO+d=VXAM-~DBLgr4fDU;3Gt<(bQhWne17QTt^B9Qs
zZsgPFczb(8FAlQbEipiK7Z(>w`-&W|gD7WYGqSQ|0X^x17az7lM1(j*Ulg)JE5hu7
z<79&fkp!6G_yx4~wM$LE&S#%M^uC3U=jK8&+IE2)c}#n9ass*pSo-)}_d$I!1&#_*
zx{Lr(J?XE{lrww4cHn&md`!^Tp=|>Sa+Ogh8F=}}DiO6vb955Y*9-1N&{zq?PQkPW
zo?|Mn^IO>ain>r<w_PeLE5joheUKnyYz(3)-twL|2XbKGM9Bc3@5(kTjFFKM(H_Db
zJfQVUU}|!ksecEG33SUK8(Z{e2KoA;!%opf7M&y;r~q(7td9-IUOsrv;B?8quAg@S
zaH``k%gf6^&do=D{Oal12A&Ol`|ZO+uzk#|y<S0|Q3f<nXWUH*Mk5@YARxbB))H|$
z1YebR0l?bhFB215XlQ6FJq-=C&`#H<^d^H@b(aAT2^{aB50Ru*0`UaMAb{#)P2?ng
zfd(1S(xdg!ez+q@RbWN#15@e0JHiww2{^7V4;N4cHi4DW9bBT|Nfw6vJeGo}C4(uX
zLLOBb85!B$w#N0_1BwG%Lm`^;9gHXzbiA2Hur3k6`GQ$%Od4<oh@5XX!cKrKhcs)|
z`$4MNbrxI+{g-sWXMiN(-gtI)cH<Q`Y4bKpO1(hZp&tn8NwkyGB}f?Z2oH~nq+~`(
z$;)TYa#B;@w=aOZ8qNt-KEn@0o=RgSk!2#Xt(<6UlaZ8c+TFEnmCgNs9&cJ9xrYbp
ztcuE410+=dWI{DgwA<p92xSnQ?V|vIV`3l;0Hh^q9v2l-VU*MjBPsQ2zU=X$%rGKV
z`SIl;PF$HC8K>>bX;m`T(Q4?ifSKN1UcL*`3pkD>X~X9691nEg6kY~VUqo;a_}}ZK
z^<edQ9N6S~W^;feQ+D5@(*7)*REGDX-~m1I^b&NV;1~e7j9*-oA`U*T2kIQVo7KT;
zPe|jCZ5rigbrFAEFupfYTantm`OQn}yWLQ@<nY$y&*6@a@6eS8Tni}?&dz`z4^+ww
z5Z56%ksc^)?8oG|aF8oIyQi<Ot-G6-le49(3+xzj3JO4Jq11tK8)?;sTK0e9%u}fJ
z<=*oIex8R92fDkhAmQUt3Ytt-mENq)ML0ZJG}~7>x*CfV*_mj{+iS)-wBJ}6Xl$~m
z|B=&v^l@v|_U$W^*=AiN8L=7Es<h1j#iQ5?&LuoH>4ElcHXIdd`twIV=D`9)A6#5k
zPR_y3j^+OS!m=_jK>}R@1rGEFa0x-X6H;7ANC+5IRv{8$w*Ytg-+b`tNup~QBQ32y
zfLADN(BuchZRtJ+syAv(_p*i-WmZv1$!bkG0d^Y;rr(}_oiUv9C<QLy*7y6jgp$%u
zu@!SnE%+>-G1vge0QVQzR>*M~IXDzpNk~aZK(G4L^!M*yeX*xHI+T8mtsNcnJI?TV
zpjALZzzNWo1pXG_9AJpQbLWm0J1--nU#HR1!U9xvVp7uLQS*1Ait6eu`i0Pegz5<f
zQ^@+D${@A1yKe`wmEf#YY%JOCK%ii_RoHx1e*WgL=N`)ZC-otp91QfBwO2<;Ma~z8
zLT7lCBjsl8t4YoKria>DlEZ<uj&C5-o?aYHi=3H3_Z+MZkfK48JYI;>1-y*ZlT1wb
z%ttt(nTUrM0vnDGUi|!Jrh{!z6TucSJw4siV+~sYTyLNNEbX0u-K)2+PqWJYKAbj!
z-Xok~g2DvH@tWaGg1-I(=*<K6hLa><U4>UwMa4slu>mk>jf^;hq6In))XTo1A)OMF
zFu<JvSAly08a{Tv1l$^$r8EEEonG<`s=YwyLIG8RHy(@~`yD9`K2Xv*kGb&Um~fAe
z)n<Lo^7al->B^8*izvoicX#Ks9W>G7<KrP=c^>k<(brea*E#{H3@MqI_$w4O$VQNp
z*rQ&ls6YTy6Zm7&z=<=+Rea>FHl$GNVYMLRK#2oybNHa%e|6?*KGZy*(}++^g7%hQ
z<1Nf<;c0nrE{fjI`oB5w^d8bkNLZK>m{m#HH1BC_poAtsjp$AgP;)i=JEp=YJ^7<_
z%vs)YPJ`5n-q6a*3rsis9<m5Tr4L~y;(zI=ZhqpVvxK<#8=!*VK)7m{4)OOt2@g!>
zQ4fzZ?~$je+dgq818D#>2qaf%(Dj6T1c2dl_&{r^<ub5YEIhod1ToLqy_1LT!b!f}
zOa1dpdjN0#U(XFU1|~M<7&nIuasB7d!!2UZ-Gy(8RKKuQ;2XO?PYO3|T-tkk>@F<l
zDNN-{3Wtz6-B-{NNc!|YJeeF3MObc{^n+UH)vq4Wpo1$$86Am4+#r^Ljpaje#y1n6
z!NAk@7x%l+Nwv&_nTK0c%i|<e9d)@yTOSjHpD#WHIjB#JkTBJ0Hg3<xc2w~B1!^F6
zm)zQ;IsJvJW@4vEf$<QLFVP3qOZY!8#|eENHnw8x^~E@hH}=EHo3?9Uz|~D1mX@~P
zS9hVK+vw{SqA>O${u|+fQTv#RzMzxNC_Xoxn}y-Xs$EU-KuGm+%kH|@Rkqd7dCv?u
zb{KS=6SJTYM*NVj&40I;u6FGwZ<6E96XgSI4L7uDx=PwY_2{xrm8{`1HQKS5c`Udn
zZFgs5{T(6MrsSah5M_+AQ!9<!Rhtfqpv^8-EwN9w%{sUlyjJP2<vH49WR!-OZ%h|j
zP7XB=>#Fv2(eeJiUdWqRzP6CwO6y}W(w{L}v*yGsa#iGuHXV-_zO<<Ky0`eiLI~Zt
z(yQQV+AEGzfJN<<uJrCTX*Gp4Iis>qs=}2t&!4%5*}5+E{!XoS8~3xf);v|XzwvbU
z$vy*%w33=aQA^uUZ~y)FHugV%e|v3ro>043QcA6VCeD~XJ~P$-7@aZH&{UkjzJ^vu
z%iz-#E(}5(&R%z3#qmx?zf+N<#IJ1StIba7jauP6aW`c{G3K)sd+Qs*LI*9>IAeDf
zyw%UA%Rl1NGU(r0ylyo+6SvnipYC#7Y5#BEchVa-GW5$LElIg(1_O1T42ueIut-b9
zGU3y@M<jGxWj5A(wY#G{@zm6BkfoUojTDPC{FYyJCERT`n8GH&$|f!GkZciS?@?L)
zJ9(Fv@;T8=MW2`$1=OCYDO`y6`}qcB1X^`Pg~cuENdV8?iLdP0Bdn```_?M%%1k2s
zhqB$r0D)O9<^<nYOamH%j~n`ms2)%2?5aoq=<Pgfoo|_*B)j$E;QaE94!17axdy$@
z4>E`433DCwmpj%94mOSk$J}vL<}M}Tt+(_V{!qc>(b|qLHTWVBsIQQ}7ocD8Gt+6e
z@)oy88*_I`+BbK(%}wE6TfJ<w%Ngc}o4n~7Q8ttfT$Bt}w1){iTGp9S%o~ckPkx#H
zUZ=$E$LRDN$~)c<{$BXg1E=@R)i2(q_Ey<F&ci=+P31=x@|Gfv?W<Q_UIBpwbX_{t
zYT3iB7Z^XcdiG!!`B+hERpOsezsh;*q`+&IfPPXCAA)Z%CFJ#p`K#H&`ig$WyN_{4
zuo_&uqk^B=;P(cE*4I7#|MC|JB*%u|{pQ}hE{{(j#=L2<zc+pOR8)YVo)@t6%d{~)
z?>D=BF=x6r60PD23)}K^vDr7CBBs6Jy!=~ZM5xHuojN|83M0#%YUk~RIlAYZbnj1Z
z0$<Y0(2LNl(x$TEc2s(WyTOW>zD7vwjdzRm4|yGPU#qaoz5IAb5^pQe#E!lDq8HvN
zS)LVK>K9u5*;2?gX+}9_0Bmf^yortxZ(gUL_$GqN>%{-y)lA~cTgdw(fM0A8k}~`e
zzQw!MBk@7<so|fCI1PS^??FUwn^CjxgCic%2ny=|v0v11iSafDlVG*55q|vy0`W{z
LULs%Y_51$|wI&=p

literal 8073
zcmdUUcQoA3_csxuTb&?w5h7Yd&+5I45TdQ#!$w=HN3=zdC_(ftiIV7K@xh87(QAm+
z%VO2lpUvm@{PUdedCv2k^ZV;}&dfRY&g;JK%)K*r=FYrhpXzH+k>4dJARwU9(o}s$
zKtPE5TS<xVE$>fg^au!u2%hR1tKnHfLc&|OZV?d?-M)RBn3$M^gyhbhJEWwfWMpLI
z<m40-6qJ;dR8&;d)YLRIGynjAmX?-|j*gz5o`HdZk&%&!iHVt+nT3Ugm6i4G-MegT
zZ0zjp92^|??%m_$<OBkNAP|U)i;J6^n}>(z{{8#Byu5sTd=DNx;OFOm`0$~CfPkQ&
zppcM|u&}U*h{&Tyk3>a9A3uI9CMG5>E-oP<At@;-B_$;-EiEG>^W@1BFc>T=D=Q}_
zCoeCrprD|rsHmi*q^zv0qN1Xzs;Z`@rmn88p`oFvsi~!<rLC>4qobp%tE;D{r?0PX
zU|{g{=~F{P!)MQ)85tQF8yi1={@ldG#MIRE#fukaW@hH*<`xzfmX?-QR#p%Q#M;{0
z#>U3h*4EC>&febM!NI}N(b37t$=TW2#l^+d)z!_-&E4Jo<;#~I9v+^ao=_<C)vH%t
zUS8hb-abA)zP`SGets|*%-`Q14u=N>1Ox^K1_cEL2M33QgoK8MhJ}T_e*HQ;Jp9d@
zHxUsLk&%&6QBl#+(J?VG2m~TFHa0FUE<Qf~?c28r2?>dbiAhOG@7}$8|Neb)a&k&a
zN@{9qT3XtN4<FLg(=#$MGBYzje*Bn~m6e^Hos*N3o12@LmzSTPUr<m`SXfw8R8(AC
zTvAd}T3T9GR#skK{^`@F&!0bk`SPWrqN1|0vZ|`8y1M%7*RS8ceXFUdsjaQ8tE;Q8
zuWx8*Xl!h3YHDh3Zf<F5`TqSo5{YbWZEb67Yj1D==bwK%IyyQ#JG;8NP$*P)clVDU
zKYDt4dV72O`uh6&`v(RF1_uX+hK7Fr{5d>4{Oi}Rk&%(n(b2K7vGMWoiHV8H$;qjy
zsp;wInVFf{+1a_dx%v6|-@kt^EG#T8F8=xRXK85(jYcmoFR!eutgf!Et*x!EuWxK@
zY;JCDZEbCDZ}05vU@(~7-QB&tz5V@tEEaokaBz5dcyx4he0+Rza&mflio@Z~&d$!y
z&o3@6E-x>yuCA`HuWxQ{&|j@W@G(I4O4H1TfPjMTZzUZ49ePPXz%HVts$>k$#^jJ$
zGMMJ}WM=O0)~4hnsDsR~qz~5sv;g9ktP3VA0daSEd4}lSB*T4j^`akugjlTPJuhCY
z<kkiYA4C+GZ}-@(#tkVVZpgrxR=a2-K}z-8<@pr$b4a^L-sR!9&j$;`=h3@QAQL{o
z*IMi@cL;o6!_~w8CzSWV>mPynwvrzn-Ve3PyRZ0t<L3k3&~VI+(9>)sz%Vah_`!dc
z14O&uls79%XH|WTcFX`WSwM8B)~CkD+lTs|bZVMuJx+6NsUl*gS!Nz>>4{z1>OOuM
zR%^1H+-<Gd;@ffR-%itF=}(N(Ss8nJlg@7K?d$F9w$poMJlgwBwDF(5I`l}V`n<xt
zooDxQ0}bifAkwc79m1F1UAnej=2!Lo7P`c%V%;C=L8^JOSEjSUu+6wuu=mGtpY(8R
zlHYJ@SV6|-HK?yK#y{0L-&un*$!6GPPkON}n(!Ff&cAwrTXq`KVOq?OtZUNHR4dxL
z|2;ZmU$5b~ni9@b6%;lMp0n`unHhIGbLVVH5=Zs0skb;ITaXjH=^jFjjq_VIvwjO~
zdx3l!>b(A%VfJm?chf6{kD&^7r*-x5gCElc<=rGUYh&L*XH7+#)0_+ak(FVN|J1@d
zF8%g~PP5mqW9N=w_HyTE>gpFfjRqc4T^Rb)fZRj5wT7$-$=@L_Vw&>*$>oUtUe=g1
z;aR0|HD|CiC*}^R!3lUmayVq?z#J9d?Tp`Sv<p01ZtzSB^@wn#;%m>?Ix-o{^-yj_
zcAB4h-`mPKK?hNlN@a|`vRkGVISso1L$551yP^Aax`sqx<I)A*87A!-rY@7nqoTm&
zwi4EO{|{z@F>u#lvCgU5K6gEc2YPG1#dW|*D69QI`|!h%xr94rr(Y<zUW?6ns<g7H
zTUf}vNgO&;=VxTPE}1m9@cl;Sa4ij-6a&|y)S7FYKb2aYje&a{?1fv-<dj<80aWWv
z3a<6HvaCLN4fjYsWv&WVFmI*?LW+(uue(U{zVz4AU_Y+x7Uw5oYbh9Z2+{SLYDc99
z?%%$r?c74w_ZnPET_)^m!w9kF4#O9gpGZw=C=gRz#Nm#Smp&EAR*Fd8GcIX+ho-a<
zp#wtf-4$QsVuLC6qQVeG<jJG+8|B`@(^SlDbbU{zlBFmK{~Q_0od)5|VE>PnGL!!E
z271KQy(nv3B&)gwwa0r##1#95A7}A}JS3h{5xKT6z_Vg3RT2~quYA9lRUNMFPlPou
zl1>#^CHzZ%w8!nBI}0e%c8SsTKbGnqwP*MlEa-Cqszuwjazpax?yg>h!xIVwK*g%n
z!3uzCp*U{bUBe2v243Rml#KuJgGCyY``fTc=!M2K?60t%p{PPH-wsiC^%Naqibaes
z?L8yW#0D8!LXALS;UPMlTuupD`pBwq@A%8V<c44ByX|YYqubbz=+s?|Xbb1~Uujf?
zR99eNwSZxX2x1CgKSk?pW?i6d4X6g`#=uncKaDb@+!4TQhpHRm=jU(WRPRN{G^t}g
zjp3VPMlruzBLW3*=kz{6h}itE_M3pgLOv)T4WODe8pBu2wGyq}s)TfU#4)R9bkr^7
zq=<Av46OGP`01|!Au8;oS7(#Sxo*cKSo0-x6HCaBrC-%wYbeXEPOM%`>&;B%We{QS
z+P=7Fs=oGlk`v_~nGs}ZR%cRw4TNy(qBSOYbDvf!Af`y8qRrUu=s&Gi2tYA2m<dO~
zMR!P#_oPJ)x~cESCk1H!go7_ZTY<kB0U~cAM59)akjjC2+2n38!~1vdnGcQJoiYa&
z9Sb#R`j)1bsx_Kw9O~ob`a#!GkCpp5zN9-Oij9Dtc}v3{QGGXSRqei460e{mJ51+Y
z==YluSI*pVj-B7re``Lq8Nd^N>K}R?<&wOKGyAtYFu#(DUzlmF25aILXRfrSx1@z&
zG<+$TMcPtI>C72D)DeIrnNCqgh-$Us#{{S*<mLF&-za8`p+`&nGnivJ<KmTC`~cq`
zN_L%QN!3*DmWO9xNk`;wWhy-L00tcPqZvX*g~n`wM#Ly1=N@>Kpz+~n4jPovdKa|m
z;BfBYDLHoB3>+iC=JaV^0A*Ary+06dag~}aGkX_ckoDcD(D!%QCkzML)rju$__yqp
z>%>JAy#Laj+|>I=`GN?_$f^5bm8EHzwUVr3;_*^HbM5mzX0&TcZj)erZObV-0GKPR
z&~qDpJXgKI?K7b$`+Rr%=lg@Dhq_Akgearc@25@PU%*khpxlS0wo<2d4~{6Y#{A>V
zbdQa&GsKkySmWl#UICDeV3a)=QP!`26{*xA^~skC%MkGA^z}QAqVgj=EtA^MXb(th
z1Lbl#<IYNV#{!xd(EN697xd%KtII~EP(~}S0`s9Lk|!Z5vT{tkJG14c0n#B70D}a&
zc5>w^4Z|%C^e3zixiu00F*SDlfg}qVKWW)G7&XRi|EIu8DKqs+4BR|*IBA1HPVH)`
z{<)&8&@*RQ!MjB#hhh?nviJLmn(T}nuWi2I>0Q);#EYCWRs87dNO1}84s$;F(O~O(
zu~{Fgz*?0-h%yyhe2(*aXp5$-ngqT14jJ*TLOPBRm`k2kB*Mixs)INCHQa=?Dd*iF
zjt^SK0x*ag<*>ajh;292CBTX|73bj!Lir|00R|C}9<t_j70lmB9s5W8-0qLrM_OPm
ziBgGWY>STAE!3Ee;{hgln8YXnn1Xvbpa!rOVJ?MjGKgwmrkJMGd9lM5!f$ZrxRI}i
zy&K~FAf9;6Gxnq|pWz8BLhSL#)nn5_*IX1mD{7Im5DdFI53{PywDZkLCo2Q=E#*Gb
z?Kz^a2nK(SrQ|%enDgw>Xl0?p(S9iS+}_$C=^fl-9Un}AR(sXI{>AFb`5856pd-gC
z)OsKaU`;c}bxFBAk>l`_Tsp}#D+lYZS-MG1{TvAK2((((z8kun$(_vUlKw3MMu{-Z
z+6?0>7T+MjumGwhPoMB5oGxmDiBVsM7(zIclJ-XQ39)Xqyb15_beOvKyoT3r_84#=
zGkyg1Zxdj*k`E|M<>yqvBzXBot?rf2xd(oM_YkJ(gP>f(5rt8GB_PDAKj^J_rtDi7
zIX$47ci`+SLSv_dBjOHvVg-}ed@!8k((@-AE^%hVX1Z2c`Y*t<D0TP~rb3Hh!N(?-
zzdYhFUb|fl98HEWP2%c`*0J8+{gr4)hWgSIci-}C?$)g6!*KX<1~SihNTrAn2$7T2
zTkY4d9{E>BL#=vzh`uF#EYUF521|p|k0HhSNzY3Ba^GkacS@^^24e41MC9G<IsYE5
zigjg0yB}ul-r|2-AVHi!tc<EA&rS<TnO8V>T!GWBa{;WgMAE(QLn$#VNdT@;`{Sum
zTH3vd7~HJjQq1q{yUO4n+>s$YP_c>jSU0>bpBjqOixl-RxCdh=D(Ht5NaL78t)3EF
z&wrgiHU~;^DJxgcz{L>{(SNL9@K+?q)=%%8!W`^C7zPPa*&K`Cf+)R}U@ZVtKlBwD
zsFoRTQ^X@I;rX(|&Gyw_MQbn5-wF`Rx%8APW#AYu%YysF4TV&YGK)X|^q*bV*^dKN
zFLSpPa@&ANM}$<F=8YKtR&y5Y?cGSkjrhy%zGrqwBzNeB-qPQX6!cHTB184SQLJy|
zxAyd)H=IM)8`%dq5By|14er!><wE5#3~QvVT1MW_cj$8g)++J}X7DPfs#z>)8`xiG
zN5CJ{7Hu`wSG1{F_XdDvP8PP>zxurwAsqkL8`3$bG@sZy`wVVX<lain^!VkaxYrn$
zGLrS`@lU6_*Z7`glu=R-KtfGWT%3BLc{I+gp9wamRu=n2+=Ly3E;31x&mui5Z7u%*
zYWt|<<C(X4ZZOO5#;tnEC{_KFIv0~e9f$9U(<>!U`W$i_!%|P{YYd%l6t~CzeC#Ep
za><>T_{SZHZ|`_uvC4lUaDU|m=d7q3w=S}c4)3^N3HL6h;a{;}RugtxH2&J@V{$k8
zuf{J;BT`?D)`b|HDI4PuN#`+POcRe$rX$B-WqD2=u$q-BqJ2g1vm>H<g*r^Px?Tk=
zx`+)8V>oT%M43JV^q7Jf^SXU>Ee8PBYSU!Ou+m@phRm7z>lN9{D#7<lZuT+vCUjJm
z-34g5+OD`OJiJ1l{z>Mpys`Ppl1hx_)!))cIvMF?quR#LXFI2@NBmE@$vanjv+;YV
zar}nD%QZ~zEgT0U*NF@rU#x`A|DtH@liw}+^D}-z2Sz2jRQfG`%=ZD=y1Vr60EbGD
zZm?4NVxt_u!4KgvMqg}01L`q<{=XCCV15j<J%Ke*-GOm1RVV6{J*eJerudP&h2Sf6
z-lf%nx%%G|SnB<gFNIw8>==d*r1X2n`m@Ps5)-L2R{>Kei?(nMbbzpfYS5h_+5mea
zs>oPqdS9%<ehYrua_z(Zd<gTqcmpGMC#7%S3+rSnk@ld+JJTm}o~wAU@ePbpfK>K#
z`?ui!LLIbuH2j%giR30ZhQ$QH(>6U2Ey9mjCvDTwjH*^&cx6Ii+m5Aoj#c0chbQJN
zU3jO6AJC1>lG1bH%1EjkJ?2b|2LJ}WlPp%f882t~E4Xwo?S*G1DNrdAAr#u_d?Wg{
z9A5tGC}as0TI+$r!si}no&5{7Y+O6Sg%vH3v>C%$#?kc<Pz}7Ip*{GgF%?GnMG^VN
za;)s<(V`MK$`1$;eaL&R2D1#t2UXx}cy|Uk;@v<?3_kLRPy*_=!*mto)A0!4;kD%?
zD>sNwwQdlJ(}Y|~UoeIB*M<jipQaR=J3%gyu923b?wFBJ8l?2qLG5o-hxhSLIYiU&
z!8KGbRl#Z@r1Uq=*SnT7kSMM<F;>w-iI=IcxN}<Gg?LW0`jYta53I~L9gAe5FnI&8
zdMn-q>%YD^Ys4VhLVO|{2b*v&LeF!Fw_SV6nQhVS!M7oHf~gGGlrHGdS<x-Ejm>Oq
z>2$dL+xTG$IqNp=7g_ksDl+z_^0apN%2~wglzE298t0{4mXDpy5S@S6iSiHIHud%?
zZ#rYlU<w=a9~v?E?&9T(;f{<Admnc4rS4n#hfF4ggs^^|5I+2FiQk?kF9$Oen&R9t
zU9neZMK)uy;^t}@xMM03J&yTTIe*3TzI_m|Kb7?Y2jW+Nf?Uwfjy0-&unre=eRx`W
zI=LWLBk^bGgZ94t>FdO<2KN~-N}wa)CVu6ul9AD`7X(=H{SjDI#n=jw_=F<zmhK;@
z>kgBYLeVUq*<tTX>)={PKW9alzK-)RTujJN0(UQm!_Ulq^Y;Z6OLAs^CPX`~uM3wH
z+vR@{S=d)Z?hLM$y&IxG4Cbc-LSRgX>FzYT#T?6Ic*K&jMJvuO=~p85HXaK7xWUNM
zlm?6ABgA?idgiSz7uKDlEU5q@Vj36TjuV2jqPe7~fx)=vmd#aH?O;6FXw>AbxXtGS
z5CNBTu#wYj)k*l&NdO^`{trizk!?jGdjJuUAGl`bwy7IV49l(8a5I@Fkpc=d#T+UG
z01*L;^;ac=_~l5bh^)xo{*fkE<k|xuKn)C}N=XG*v}=K*2pj;`XS`DJpBT&-!a@}D
zP0=KkZ6~En$=#emSN_5lr2Gxj$x2v8&`h!K3i+()ZIBt#+HI*Vy=Jg`eedyz%6wP7
z0TJJs{W<#8zvyM&o>xyzb=t?+VRq+2mEx6GC(*9c38DxXrIIGo;;uy%4`r8zoW?g!
ze`KxxwmtOWd%@72yiST<Od<p@obylh@x4(Y!wY=&TFg0N4KY&H+$4!N6Q7~KDGJq8
z{^dH$fBVfLeCE+s6@86V6pS)l5&6<`V|YrjpFi>zw#G{=Q9vExtS`IS7LjT&dF}FD
z8wimU4yG%n$S7Jh&CVpkdViDiJ$~3lh}VQ?mi+TBKGGEAC_e&-&^Mleu{_fd`ArZ+
z_-$zRAsrEh<xM!e8Rz>e)*fp;q7Pz0q{j-yUW2+Y>(kU|M?0nd3F{u03pvyJaCpF@
z)I+tgy_yOt)PS<uBkS+xl(OT(01=v{vw7B>EA}xJq~&G)g+uOA&@8_y0oMEJ6Mg8u
zwTU7)ngs}<9$lwR=D;6s5Ad|E)RNlIQ8iXXI#CL)SfI`xL5N<p!7(ZSobG|2YbYYw
z*~w@UC>Uj&X&ix&C;o*bdi!&l;3!Qz!wsYJUL2OsEh;BLm)M_s8_4N`qowh{dzz-l
ze*`fs=-2Sg3XWHKFXO)zfqi5FBAK0Pa^TyE^+jwxyg<mq^bivahb>Jt_%jts$I(UD
zlx}AgG~<RxK$p{5G9U9Y^6^%srE)eOtdY%%qHbZI$!yoIYM$HR%_73m$t0JHT{_n0
z<P6Xv()+*V`}~k2gu$9W+(DO!qVH4<;-X=cSVg2bUA33vDl^`;zkt7}X<#;f-QtIn
zFuKcMTc}Ww(@6q^&<TVT&KWdogQJ8Lkqm8nR@!l$7@IAS4iG~ADW<6=ln}$xK!AOw
z83EUM0mw`14I7`Ena;w`I(eCTfa=|y{Vl3+xYk?!jInj0S<uVdXpeR4qNc|kySpgU
zTi6CMKM_-JA;>tCQxS;~yiOKi!(eMT!r}b$`(V|?jjEAETQ&ex%uyiuijp4|Aglpg
zfb|zC$hFMwG2wgWGP|5ec|LB1t&3k2%@nY3`ju!aklZ~Sa-Es3Yb;{6?%3HE3R&`?
z&+|b4n`}+T9VMMxc~pvo&4w-emw7oRv<|A;h5KS1q2|`tb@5WX#d|r-uaaPt79AZq
z36`uZzX}!?ckF(=!XGLgvX9ugh^=Z|dcJZX8WZG<n!N+@iAn<zw&9zte1JGjLI~a@
zFBxR^yQsX;L|ZTb%6&*hPq!dgl+y_VF7SH$#$L<q96xR)Lo`Nlwck=+3sD6}GXNKu
zYIwa$c(P%X{EA2+cJ0;vkc&X25Dq+`P06QqmXTIxuStmZSQ=%wj_A<uv3yR5mHf(U
zx2fB%Q)i?}gzh%lvC!_tF5t!RjF0baYa-X4cO<CzbtmSk`TFe?8AM}jM=g2iW=#Z)
z+*J_?Z=j^6)NLqY?W(+mEzFd>JR0%Law&U9j*2hQh4QQv<Z#X^A>ma#SVr{rsw*S(
zh@W@GoCm}5EF3=alG3_W_ml|3fM?#zvUC38-#W~~cTIv;d+bXL8i>Y6DhGmDBgz%l
z0lWxO2(bb}*weZ^J9S*v-%+CCc~1D0L$~m2$`=UXnQ6N^kD5Mfh2KLkcURMDMSyn;
zbCmIqTnr)JWX&w9;3#%Q<O5w(x1fc54i)^8WZ=b)sMXo`5}@MkyrbX0+)?>ggvGNO
zaZG)=Z8a;pLh$$A<=+=Q?O12jON5HgztdRT??>a(^Ntw79KQL>6J<Cn>OhEA0|aCV
zEJfj!-9p9NSZ@@FE%*G=2hjjTvb%^PZY4`i-*A}G14J?gztTtL^^EC*nDI<ga1~Ep
zA%_Y+SRdhU?t6O$|3!DTbK_rkT$_}-h_G*(T{CKm51-BQuiQdM)fn$BZgx@N2Sb9Q
zaM00aZ7%+<0*-#Eh*V6j@?KTB$kv#=3lIT}ea(`AJ;^I)1VVrUeDvN^H?A_J=~|ao
z?&+f2i1v&Cp2Js@9Xs>c+4I@+L@_ngJ`@kb4r-`5q%(3-hAO3dD|^2{3Q91C3rAYN
zFpMo0^+^s`8mA`t?S0+tPfWuf#_6E!qeYcQ_B}s%?|R)xOTyfCR&aEu%PYOr<>nJj
zw2hr3eSLMx7ISxQ8!)%}OT2Z!_WiE$O#9uh&tmMaCWQUs6pVC0LTUCM3@Z*N%HQkA
z9a->d3!x&qm3(!vN%y2XFe+dJ3}=<?M7H>8j~^czP_x)*j*g#XJ!?x+5Dq+bW9~z<
zJ*U>>hnloHOabUU*Y>Dcd`m5Cd)=l2pvdhUcGk>!&{e`&mAhp_%S&kUkfYI;)#{K9
z`B4BTlw~s)yLDiaW!X?azu^5PzO;e!ThQ#Rnk2@~gUbxr<22TO5@_-_O}6`Zwmom;
zlQN9@>6W!CU!h|8)Ml8v>-X5)PdBM4`?hu2Q-k@^Mz-ydVX4Qne9a+l-&sBLnuZUv
z_e2|WN*nTO{=Pu}m)>AcqsGXNSWHgRxYak$@EOj;7|+<u88~X-3KHuzwz-J#5dE`i
zn6Ti13;lmQE*F+ISbte83VQNm!I<l+ye)XD=50w(ztD#%I)j?;4L&fB+0SMJYyZbn
zu^z<zz>VH?eNxZDTL2PJ5-{LY`LP}bTV-mSEl&8+aJn0&KVLN=*`L!h#Y7`cf)XSD
zzmzy{w0}#?B-(%~ZB?f3<^`m@vS{de32I7h)AXV!L}o4quKpi2-7vU3Sg#<YdoJA&
zIl332dg0w5QUtd#P?RnD&(g~qDjc?WlkZ#<9@o0IxY*GFQvdWLEkQq3UnF#IfT!){
z=^-B(KlJ=KqDpMVb%bucR!xm2&CPOVLv`k;VJ9RL)jo5=TD5KE8R^j6KUPRglV@6P
zOCP7sM_$(GxHoWOctbaJTytC5Q%=A%-cK6)W{RS~nekr*n@5s6A!gmqlxy-Zc7_&K
zuSe&)o1{?3Srh3ehEH0lf~;F`j+N$>)mwqCT&FklV&sGSPbcm)Z#)HtE%P$Odq#V2
zdQ4x`T>scR0rGsWw-lNL@uj@GRCpk9R#rF?JNhpgCjzZ$l(qegM{R4DW8><2`Jm2h
z_yKB{HQMeCJj`7Q5NC72a;Y|F%=L@zwmeG^V9R-s1G!x2Ndy14|8(KBO>QTX4`b?#
z%1KeCcH`9$fCf17qhc^)PnJZxGC1O+;w85-oeSSnMntt;yG2S-ome>;{)&4LxTY<K
x0l3}5rs-}QyyY+*(J$co&;M2bpDi~uQj(gdCGxi=@&6MEwAA!ftCVft{4b{-xugI9

diff --git a/docs/dev/reference/figures/op-softplus-plot-1.png b/docs/dev/reference/figures/op-softplus-plot-1.png
index 2a0b841f4bd224a1ca15f6d94dcd08b6dcfb90dc..69d87ea65f24a01339363ef5d711f576278ae13c 100644
GIT binary patch
literal 21578
zcmeIaWmJ`27%jT}L@)>g0TED9K~X}bK~PXYL_j2@QBp!mIz`2x6hu%;8U#UUq(w@)
zJ5;)*n>)AXo_l}XbH=!1+`o5^&;2Rv{l;3)dY(DwGuQi4URIJ~Fa2Hufk1IX>Y4(9
zuw#YzXZL^j$ysTmGx*;g3n^7=0%6}l;-4K&y{;Ps!g0clYgZNRLnnJ3G-_UMlg-L4
zpEZ6fCHR)%^wXzj#$Hoj&&`w?eaI-ls}gxe^2|#qJ1OZ`(q|sLoZgoh^Qhmhc4N!P
z%}Kw=ODW?0*y2XC<nTQME%x;f*~=rBBJM>t8j%nPT%5{jI|+p1VzI`Z1TF#U%LGDz
z`2YXc|99h{%PLfyotT*Du(5KX^DKej%B8HV?Be1=N$LC8-h%Yp+S8{`U%YtnY#*&Z
zBMD(Za&CS;gx9QV<?kP|-z7(qNB4LH^O^T$nRQpn?jX#b++3ST-MKu_@xLGb+!DbF
z&n~s_h(K+%-kz%Om+xmJe^-1ExkEscnn2hx*#D|<M7O1+RyjyPNU~b$1^J(kS$vAr
zE=C8mEf>13&ySVfB^AMJzvdmdE|)}}_SNyivawbR^M#b{qBmU{_D<Pf6XzFHSOYI~
z>?DvU?C5V?&c9h7kzb;)>GzIxw}(v!iTFP8>7G!Dt}CLVh+65JHv|0rjZI9dD8=?a
zkzbe$7?!P{7e45(+Pt`Kma!yPU*Pv#KG49sc$kzByw6AC5+7f7cDA3756^`Qe`jZp
z3pEy}KD_%>IkZ4aPe1AG)}5dRqhI$JGJMN_{#?5KNj+@mlB{n2IW7WWLk?s0_Vw!=
z=T)nIPL<DUu_-AWoSbs^W3C#hDgWA)_e6WUM_E%tQLVSB_9r{PgqoVnK)>z0rKRP)
zd-vS(E}T1O^83fjoq^PZ0a2&5g|R}Hb=U2opMmFa2f>A->tE*f-mKnt(`}RS?c28`
zyLN5cMu&u`TU*alQIHa5O9zLBF1u~@badRkdiB>p$&-s0FY@s4eE<G^FfOXU_vOcr
zAN~EC`X4gz7%N`CzU!<5<|EthnCQdqtlHMr#hzSiOg+9~V`a5>@7`~2+akFoq=lo2
zhpVo>>15aDFYvr2WAnLBuWyGORwbR7xVYU+-{<&v$w+w(4UO^f@ujJ5S~@y@=M@X9
zClt<&>0i>bV%9gUPP_d0@ABz4q7g%-9Gqquk1le`=I<cHy$%Ry?&@M=W{!%AvRa<*
zMI;Um>YY7%HqDwsqH$kRu=eiQ!aLQyGc%E%*^)a5chb_**x1?GIXP2PQbeqWWps1~
zrn<8+%-3x>Jm`|WK794nx}Sbi{XZ4?orE~dL-Eg_^t7~m51L|Ly?Rw%Uf!8*xNq;?
zPl<_d7=4bDPU)Oa8dR{ZDsf;VxR)%Ax1M8UWb`;*7b<W^L1AvRF*+vZ6e}xhn)N2B
zI4_~t%|9gM=)r^cO-zy^BQ2MwHTUe<bNlw~Z;lmIo8ek}2v@f0>FKdhZ0Cn>SzB8l
zJsM)to}#3vxU#<9|INm0(#Wm&8Wmx7^WdRFxMUuOxxv}l*^&APK5lM%M@LdJGC!uv
zAO5dTX%)GKg@l|teq6`Y)Ol{O3e&{IZ*fIKBXm;3?eR+{!hq0JXZlF#bE;z^&ZDEF
z4NXmJ8>{ny=TtH>Gpp(r?anw7+-Lqy_vZcn{X2xue7e7=zrR01EV3s2a$ZuBN$)os
zJ;`+;B$49f%G%m69upcw{>6*ZGBS^BYy#iB5pbA0)tOsJ5W;cH-$g{s{Qj+=pm6ZO
zfeUuiJ%?CjAKBXetgL(;Hc>6VKp^zT#m86F)SN$lytuLPwybPQZf+<3DJr@>TpNND
zT}*yRAQaxx(&}3nYd(GYbgFg{2RnOjQxk42pgY?fw--KXr0p3+Mp)s!apQ(ugourW
zMP+rhdb&YvuJuSvOw8!`kJX}*aZ<v-Q3fH~<fteQA0M;fn)kN0w(Cn%H*emY$ZMeJ
z{Z1U?OhG}x-+%saYJQ`pqqCpt;$aIkHZr=au5My(E@zvuz3UKeyB)E_c-irhs_OgD
zP>n~A9vK?O`})?sJj7aAx5!DnDea-n_2up?v+CN~-gHAbGqb71iS`HF^rE7oow-|#
zuP_NockkX^Sy>TwS>wKVQ9)hZsqEF^$;nAtPOVcy?M1{YwUPJk9StQV<&l3cef=TK
z7%wj`3k!=cJ<qF2qisqu-5rE2%}0+Gs)IOvyuCLzH(BMvO$`nCj9bol8aez&yx5|-
z*;$On9Tk=KhK9<zIuWbE7hk^2Cab37Vl#;syUl$5{P_zPz7`bpHa7mng_`%}fBpK^
z($<zmsJ;C=aopU-#>b5@0x~i&jYbb2KE$9a-?@{Xojo!*cuJ_NomgKtqA~OjjE&n`
zTCPb-PK=J8Vqs}(C%zD9p_4}agruVccef10_WVfwA(F4wBeIN)VG*ut4<<XlE;ux<
zZtu8`Dex0<+e%JIklKGuMy4i!RZdS&Pc}@D$E=Gx&D!PYzi2pr=FFLsC&e{2d+X{_
zkVWL={Jgwws;PaL)aWNC-ZU*TTv<hhkdTm}o?gPIPbuoz4_D`gn3<XX`|m#ll>srv
zIdd%kNZ8x+Q&GuTSbXvGlar9}%+Jq{k7q^nk}2~M2yX4Itx8&2OmuX?0Ricbi{n@>
znI`S@M~_-tTE1bNa{4%m35ZHbdHDBt8IHVb*DgsvCL2r32WDosWM$t41+fZs+5H=w
zXX)wb(b0dNkWpD#Sz+E&G;*Wj;%pa2Z(}&qqFq=giHBmobg8+l>{(dY=GNxAd4J*I
z{rlyWlr;10FW{1nvEZUbkrWK|^=q*($C_ft#>SA^kbLm2H?k&2oX#Q#98^X|Mp2O(
z<>=Y7GUV3Yo}N!hNr$PaD}9$Dh!^d8h=#^?v?22SB|}Fir-~mx9@*KAjEs=&+V%Cz
zm)DF_q9wje1oz{+qoSivva(i`mg-LT<R&S97IfRn&(2={a<7zblr3<Z1cOaz?CSay
z9eoM8^4qs>O-)UQ&nahQWqCe*dR`;v^`wy~0paeRgVDo+c5!ybd@`Rrc~e^2pe{5Q
z!Eiyh9492Zf)Lujy1M$O!Y{tj2%#=3CDoB`=$Du{=70JoVpUg1l8PX%dF9F#tT;6_
zOhU-}_nmU0n~EA5H90d+rl@c+<PR_;+2;MO^TQOhwC(sU*0GVHVUWLn6avLI;}|Jo
zOkB;>bP@>{QKEbQeu5&~Gcgp8T&HD@qeqW+HoDRw%5+9i_;Ho5e0(sKSs5AJe0)J~
z-XO!=Gc^44@CUMmxFQx+ntnCS`MXzQ6$0PBeaET2wYBMFXJ@x_=gzd~O&yuVPrC@S
zi}!VPdAPZ4XZkMNjJF`9v$C@}wenZKKilW+TJmIzt8r)Xex<v2E6d81Zr_gl{CN`~
zD$i~P<%V7J+uKPEQ6@}1-77IMv5k$5j*bq0e}9y`{DOjGqOQXK76!#`<YRJy2oaaL
zzkhGsyxE;=%}PtFqOKlNSm<hQZhq|=NoOM|31-4G6e)&|j_yl(x|5TWsN2>CYMX<D
z18yuixhA((0vFTJ)YHSj#1xM61P2>Ec#s$u$HK;@CNE!{%r1xU(UFQ&NK8^mMX7PI
zwERm9u-@K~kPx6lGc&U)UtV_vBG*%ZikDtqsjBG)efbV#R7}aKseG3%<=IYkdF<L(
z`9D8JO--GVk%44U7s6LxU+*U!Yiw)`NZ=o4qH|$Nb}wPT9P1Xfb^dpmS56K;1%*$2
zglI*1xom_;*rdizO5(Kd(bUvL?nBt;+f2MmO`Y*Ra;c@YwXCeny!YF6+l(F0qsZJ(
zM1K6J)0V6X$T89Oxw)mqd2w8+f2f9T^GQ(=DHeS3#)ps)V(zD=j!8+G{{34lQog07
z<qY*t`Tvnk$#)G84@*c$eE$5|dblR>^JkQrGiT3se$^%G%q87}S<<n>?|giG@F5Wq
z5v<Irsw%#RJzSSAsTVp6U=H@|xh^-Akd!<VL$P&eXYp%A)B#mh2V|$8Kc$_V)`NqC
z&)$hMGB(Ej&ytHZHjisNs2@qF3(f4<aFn(A`E}R*=(6k2VuEtFyPjr|xj$Y6-q6gm
z&BU@?STOC$xxm4}!Na2@BjY`(@x-{;ZXu(bEKeokLFd>YcO*xXs=0AyPs++TYruGE
zHs@FVJpM9EHs6CFNdmT$9VlDtymLT*6YZ%Hyk;LFB97t4THV`C<!m+P>_bR|q8raY
zrC(HCu?aJ4Ch6F{V`utl-wN%v=D5AvrB`RZeBq*{4JzurtE|kz$tl|KeuujN>iyim
z*U#!|Ru&eT!^GG1Yu7A@In+#^!oD&6mG_a=UqeY<X4Mi|hT9){SFR3=^IZshqP$(0
zplP(DUo@gC!?^X6Qj+rKl!TB_9zfaSTjT_pNYt#pK9SR>^^t%91`X@O1O5HSr=}Q<
z+2t^J=~bqejbrqg&k2ZqiPjIzO>&(4aQgIzL*oah-<iE?7-k%eCv%T9p6M&tM{_<{
zz}i4e%mWiY-kO9wUVxRIX8lHIfN${p$rd?bPVV};^_sJnUFObYh22TjcDZ|Dvm@GP
z?M@EaM&}Nlgr7fuB0UOU-H+h|C<j(?TAprgY&;=EyN7f{{>=LBst0WTF8zMWy^iuW
z^5KUGp2^EI{j{{SGhLZLA$Xbm`u@z*(-TM18(}7^bJ#&J2>9^fo863-gTv}q-3p0o
z*MMivYZvCHr{9s6M|K?~=FN0W9m0t2U%tPN#VjQyWo-N@JDaaFSLz9nkwG)EMq}dz
zt3fGI(W2?;hX}OK>e;@GLhmLuVoEX}65Jh70n9toVu1}tBHY47-CTj$goK<>6qA$B
zcQ#f&#c_r$$6FEwtcQP;mAR}fJg9!h4g|(`;lhQB7k>{iHjP+CkM3}Hz*_$B;RD;5
zGvEdBpFiKacI}$FdUtub=cL9dBCrWS`95?0yhg57Y*Nw$YRP$a_IQ8?YLSCW+L1WI
zk-_2N5AWZbb*Aap1Yf|so>NJwsjQT`c@sB%T!?<vqK$&UwUdg<Zyzm(Oqk#zG6S;F
z=IpO01qGWvNBQ{q`73=TM357?o`A&6%~=6&@B8<Xlao7f;sg+QnswtY<azh6w-Z}1
z79h98bXMv~fgn1)o0)?{_Qs8ulST{wp<sB|*4N7^D=(FK(SjDG8`e{jlee_Bg$r26
zd3%$0HjeHhW)EUsMlP(Vuiu{iRf-J#?VI4ed%s-Pm)?YV`RZW&NZ*Bpg}r@CjAfgN
zwgQ)RTf{9Oap*<;y`8zE`;edMlpZ`7s|yoCj^jLYCdaHh3l}HixNwG>TM_9T7qkmF
zN=Gp=G7=@Ct*MFr!u=mr0jwq_Ccs$?)YPYrAHQPDz2<?_Nd$r@wX|^G|FK(1sa42s
z`V?NQtbRxNd?z&?kr9t0C1g99kdR=%I;U@MzXHC9q986VPRpSoWy|46ob&QSw6p*K
zhYlTz{`gT|QL&||>7s7AcVo09=iM*vxSrj_;|K$n1NoiW{k^p{69FS4vW1I!PRYnD
zL>oqg*5oXgE?wg1?<g;qxOI!2oBI(!pNEIXAr_hALbRL2Pk1BeJ{+kFD=8^a&ocd*
zm9@0I%*n|~ML|J9Ma3#a|35wi)Xxe+hVSKISs=~cxzkL<ti9iOfO^xS8#!eb2*pxb
zh0gE8!fMOPG7V}!oRsu^`|jP=Ts0>y9LrI&lz2g<<>hg}P5JquVPTq*lJWWZJ<ZK`
zBqg6uYAjF^Cq44ZmoJ|_jUyWrZOnQ3`T2ePH~`#?;6f1*x8-(3Q6ZGSj)+K2Na$#8
z*45E@9UAKV>-#PNf!t_GKmRbV9YoAReq2as{m08g`uh4<DM(#Fe2a^V{IPTAh{4`x
zgN4^$xY_nu-PhMwx59U7d>mAoosDfTIr%9edfj3m@KMc&52tJ1Un=|Y!`a!H`tV^P
zK|zH(cTO@hpE`ZI%D0>j1^3Df2;!gLzriA!nwv3da$$m}fosm6m$BvWeB4SwXvhTq
z1o)VroBQ<X6Ifw!SC=3=`z_$-Neyozm}^K~T3Q0N%*)LMnC;E6jQ;%Dto8#B>N^&Q
zo}})nU|g~*u#Pu_z+7*h9q`_P0|yk@l*GmVAn5f*?sEZKf4qZl8yFbizAn3N*jrkD
zMIrH#ILd2E*O^O4g)kp|8yg#2RJ4t?<niQ*jg1WjIeAWQZb3moRb81q@oxOo)YR~k
z)5MrP^)@gNch=S24bCbRDPQSZZg0UwLl~v>k;udP@}lJg_e|BuWoBnD(5;}vC1q?Z
zmtdA3Rn*pgj*1eKlQRc|7#bP^4vCRJGe2B=QiwL12+ez6dV2#mOpcCP*xJfUOM{yN
zch<i-DOp)pmhk1<4#HYfUtb>(5hEjt&Gg{l+u-0Fos$2MPHb(-<NEg?RvsZOUjrk`
z&0R%~IH#Qa!pG+`!r9pP;Qsx{SUXG4ogy^6L7mpjwR$EI{r>&?#H1tuSR^arykgu;
ziQ{$>=_P>W7BeybaG-*)vqN7|x_b37=7$37>?5s8ibhps<=q^MUjVQeA{utJYx|$R
zeM^mjCBnpy`w9ySRaI4|r>DWvw6aV)yE09J&Z{z?JzIq&X3OnKjM{Je85o>_M{$-D
zCn8bo0sia5gq9%$;8IBM_7VuIU0q!$L>vIB6RAbx6BGS~F5N$VT+`N$c=F`Qq)`n4
z_kAiUIT;9v;gW&Oty?}pK|#^cx_ALL=jRJ|HU<+Br_}J|BvFWfScqH<0OjsZKy;$+
z;p}PA7xsYqxa;NS=7O;Q9v&{VnNUWxTb&;X3=Bjl%Gq)Z6J?9aM<6`O$;sEfXdW0D
z3A=4Mqc*g*w*I3&_t`Ld$8RYrDmpqkiiu%vyzp%wpW0ug&#_d#gl&H%#?*(|Sqn0<
z=KxrM{soSUue`jHkY|WW!^nMJqBbdj%7r8PG6<mT%r(o23J3@|I65kvp;lHV|0qhg
z_2fUo9&isttdf#aUS8fFkGol>s<&?4l9H+fpvPK#3OrBPgJp_~MPSv1U!E8l0r~!7
z@e5&Ce>++NBGMk>(kFTL>=|Yo%sM-h_$qK(et7k22@xIFm4zbo+)He6(<>|XDk>`A
z$_UGYByZlm^Soq6N{NvrRRUB7pHx<+0VoO&*8=EEOniWFfqID1Zxc;^C}`3*MDtFO
zCw;FsJ>R`1`@MZH$O)nn&d&LXiD#*(>X8w?mzGxe9BoI0A3TT|60_xwTg;9!=X98~
zEBPZ`*{smk_=i=>H%7_rjPv&U*9zLY56yD5Y$mDBY=jSw3wG_kw}Wyc@;`C`+yV+~
z_QT$9)+2TGqT8mXrozrEvye7F4w1EJEAthqs|fYXXxX3n^zqXt{&Pe3^ai78J5RSN
zIoK^HW%PAsb}`Rt&SZ;Tms4GMJVeUc?3gh0waDn^p4~o&j|rI8^e}GaPKl1k69}XY
zZEaci%U)3&<t0DL3vf``(y{<n3>?9G@glPjeZYG4wbT)-byn$Zeadq)!DFOt%@kSf
zirH@SvJuLM@9X}R*^(|E%&MH~X4bDW=)3UxSMrU544Xw3zA{>}!yQ}V0Q1M8gFsv9
z&v?-Mbk8AQrps6p-;7(?w6(Q0It@N<#^yXbJ9(|Wd0)}f;miitqOb96L;htiS2#Qd
z7|sfBgls#s&Wl`jdzJXvlYb;WRO~Mg@7hJ@WRmAE=IVM$Z*gk1HE;d(MJd0uzUQC3
zxY|ZexIzWA3<7lV(j|z1Ov3gSD2luG(Ll#QNm*SCY3Qu*G0V+SyitE)&c#Vi>D7pG
z;Mm0UTy?ec%%0`RLk4X5emV0gPi%T*%Hm>vG%<OA$e-T4XB8C8zb#s1L3i|M(YPV9
zI_ly^-5J6aCJ3Vd0l<hzq|hR5ZEUVfN}fA+E+EWAYP7vS*}uB})Tj4-U$j%to5@D|
z_B)?F(JVjhMm<-g6PYY*!(1hA=8+xslRrF2`O;wzC00u}X{YA&L9rjq)7$TWX}*Kc
zV?Fr?1OV`AD=07@KmHL)tmDGyz<^F#G{qimx)vuBEl)=++0pR4M(4@-kPBC{Y3noE
z->qy(8ywP}{(C=iD19X+`S+3ZZ>l-Z{ro>TN^9zGSR{|^@bm=sgFuz`<x42PB{MX(
z)2Cw-5=K!_^+xE~)(*S0pRDc_l5PugU2x1j&)I0QnpCglwd<$c6D@)IO{!x})huV?
zvlL{fL`imGO{uDmjf@0^g}GqZ0l$tMJ$i`bjlaK`ZOVxl?Z$MuuTH@n=jOXpJbf;9
z_pp{7A5uEkC$^8!4<*WQsw)!(M%Z>T7<Chn9J%oB-43EM=*#<bBerO>{<3}l1p|+o
ze%`ssC^qZzbY9vXp<|(qmlDPxCF>g+R##UiB_%z7_6&t{1tKj`rp4j>{SS6FdcH1_
zS)gy8TN)E+_g1+R!Q<dDNpj@I3AM%N&^%}i;>$(P9#*%W{Gp&Rb&ybaRav=heR*c*
zfB(H>S4ZA$Yi>^0t2!8JdZ}_tp(mQQOWTicdv?%P>RQIgt>vb1ns@h@4+a=;w)9O-
zB?$UHPkxPNN{UtiJrfgTjdh?|<RA!7g=kVh4#MDCt=$?u@2p^al(B_Z@bw#XK_r}<
z@<6{Kaxg(%o}3i6{C&;FW)YlWWpx#FuX45NLR;JS;G<v7onD6je(_|2oIE?+o4wSS
z2`cZsdy%-&qer_T`7m8})YI2L4Z=uI|FhO-xt=`B^}5{pWRc|?eW`ml&27FEsPcC+
zoBb>e6WaWls28)7P_76j4fV^-%?))F_1WV%;1QrmRUOZ@MfKiCMY8sv9WPcJJ4%ke
z9DLAbo?QDZ<_<OUMMqx0U#CtUCvbfP*oNTk){|oisRe=ykSKa!0u}>=&RoxgEupnv
zCv%d1ntV(07Z3kwbc|L%(se^#dgwc=AE8*q*wj={SNG2C+mO{{jf_SyeG!+PXc!qk
zfBbmGHsi#n{&7$7&YBMs8BL>mx|_JC+HT}8sud30-5j^E=BnFS+zuT;UrQ_e$&*sZ
zD<~cqeYTS)pS)wey*>ZClXdC^8yDNz_*YUd=f1Wi%}}{rvMIl-bueuQ;m@AkyLVGj
z(H=agATQ5v`TGZ;Jk&H$3n|3y)IM|fD1`^1Z!P^vu1-wF?y0O%IJKKCP^XM%KVhIB
z1&-y^DMe*v%xXI1c+>!G?QM`MOmWj%tKCSL5B1h|MRjdSNnMHfo3F2?nNKnjxZc*)
z$)X$h>eXI|K*+!#7&RYws=K>GA|fItjl|W8qDC$*q?&KLg-ubIA7cpR6uf<vCWxBU
zeH)5Wc|!vifN}BTohV|MBSp5%=;$Z=Sgl`AY13Ss{O^wqrQrF>m(s>7BUGC^3CDM%
z#1Z9*wKZiWC9{q%B^@31(99_*?Sn3_7Uo3H#BTPDXVlf#|L(JpN;$AHT_5}(B%kDD
zQ>afQS7k!^FEBHE9Df0Uu|8BFU{YiD8RO=ZO=<3S@g*_a#V3C$Y38>}<9!K)W?=pL
z+S=Ck_T|~x*RNl%td2xLNn*YA;hHVS@fRQ%8WLA)d#~zU+d&|;LH`MW=Fp)4@be}k
zZnV@OfTAXOgD^}QUD@w;ZY?XWe3~=mw<`Y%iM{1nN5V?f%^;560;ixSPyRsaLr%o$
zKo1I>S1)$v4)D1>FyALomB*;PrrA@SO1n-!P$@uI5RX*=$s<aBCT+>$pjLv9Dp8MW
zYGkfl0S}uEDViL0IX7OE{2x0NLA#-&BOx}nwXKcj*s&r=ZpFpLb#-;fln|ueFj8#I
z4!OvzSG*;oQkeVn&C};iSW3zz?N?MIJ%lT6lD-U`1y0!z!(YCvLKsjn3PQ`vG^}Tm
zmDOPK@?8-13s<8z7QA)unYE(w_1buEDcXPs&Ag;R16B)!25o2tadC0M!FR>Pz5`Jj
z)`dPWG|WcUNV6`{UMOPP^s+A6_;lcd_3g?%w{!ld7r^Vg*nX0$NjJZ!e);44>P@$B
z=)FAzw}a3j0|R3}ePTO%c5HAk!~XAYh)fzq-js_Wqj}ZRAqGZ?3R`X}rOezaCM=mu
z=H9#9MZLVd(5{<XT#S`J1L?mP8UwJwIW{)=g+~ef*Xlnf3pqtpm*m}{JX%Sq8=0B!
zB|QEi)@Js6xdUP4kel1~op{+BQc|;j|3Z8Ji*^P2tdAc*K7XD+O41GmrM|o8cb0eM
zVXF~`UAuCH#wWTR1w8$_qAdK5@8;;esTvgY_x1imgwY$2f<fHCPf^LCasny~3k(1F
z@xvledaWgOv{OHbiLzJl=7;TN6@4`qSL0`2=JfXN+efvFor<d^Gb?L+Y|MvIC<|a3
zY9d->6&0+^%-_+D!+h(zbvCLPh&ml~D;*nHe49P9_1e1MrM>q4Jv4JJi1p|Vkq{hz
zjg37pFtD9yW2g6d2Z=T?kjG{15mF-rk+kTNpiTWKIVwTRwX*n7xz9miH@AaFRXcX^
zb)5MVa~EsQ<9J$nI<PJX1*jPc)8^*p1MaR@htE~k?b;sYyq-G6@VR$>?ZdiZnC3U#
z*~Z6&202LdK)>jlV8!BFZ{FMnN1yCWufkWVwZ}Jxq8I4J^ybKK<R73RtT{pY1d+UR
z=MGL{W@^fzkz;|*L`#bzI<~mP?YW?rTXVeX2Ur6~DnwEQGbQHgId_;w2WA+&xL;02
z*M)p-{@{UtK}|4<9psDo*;ze9!+tC;ABqSb>Gmy!g(b=OV-&6nDW6VHTK)MVO0^g|
z&&UlzAg*@vW;LcaLewq%pVB=xFmRlcQ(Io%y}GNRsEB%VkiyWTU60gJiM=bAE|Zc_
zgvM4wb8~ZBn=D!^^70L6Z|&KW`)JHRFHfkmaiD)u)X#NX?WV`*N@(EPr;lk*Zr-&W
z4o!@9_>xXW*90XSO@fCeCPMeVKfPpFhnPT@WEP7gXJCi;sN4JdYT|Q$1It;{pUWw|
z9+){(U6T-&&o{ZW<#1Adi6E|wJ}78*_0ON)|6ag|oKOQFjVWMhsf?*C6y(yHruqoK
z*UG;q%y)JFwUOt=5!_Q89C89&yPm5PJY!H-`Y<%lpR)=GsLIRF&J9&V`frZA1wQQj
z*)?P(+-_IkeESi<?d|1as~ev+iarW%Yt|np+e!E~i>5F{r$X0FHDzU6)X2-1FO!mz
zHa0e5x$8+jNo^&DG^?Mh<EFvYZu6ptoaD2&7<g1^j!{r@ogx|x8X8^)2DU=xFD|~K
zruHp1b_f^4z>pKeUJ9IP8gRZ&J^4a7Pu=y0ulUWoD|$RqREwUf$$al4)fr9v^l9&&
zJs`0~koZA0va+&XzeaoP@VRrc)w-uP$!utL3~4U+f1t^C)V$kUb^5%J?(mN{{I_Uy
zM#rY7wZQknFFS>XhB7cP^z`<2cXgo&XEoVz0YR3%+Ei_2a!*E{J}Is3;PvrlFQ@D&
z;Unsv6qk!%&#x^`pouXyI=Z0wA}1{knuir6V`*t=+_QOe>ga=rO+Jn^^0TI6pDDBW
z_8F?j+;YgVj@K3tVPw#G0lIF#z2$QEZZNb}@cn}a52C?9^G{)0G1wGk*s%F-(i)(u
z+$<y}cS<hZRo9H;R9MvhH;w5I<TU&QaZdnCbj>hL=s3s48Gva5Uk(4G#w$#lufp|L
zc29qr`!M}!@OAvfu8Qlb3CBd=D&*!=?j|F|7#SJuM$g304?^1F+?;<<kjkAquxqGY
zxk5m|IxU=Yq}0D;!eJKM=*zfK{ZQkr{Y0k5#NExc7XoxZ*WO4zs7=6<tw}1Js%fX7
z<DESD=O4QEm}%KgKDu;6C~B{_^v^_Qg(&Cqi{tC1{W4(}-UWTkbWMrzJ;>;J?gW7=
z6RaBI7F3_um>8K5-l>+vJH%FK@D<OcvgNB^=Uzzp%&aTRKlIh<x9jE9kmU{+HhgiK
zl5nRx-(f!SPCODP2BD*~6Hr%6P3_ekwj=a?i)hlWY^+{-CU#p<abal*&4t>U8gR~E
z=xQP7R{E0mQ$eS2v!5Hhfd(;}c@V~d=zvf_&xs;M?to~?X;Z=g9}^Q3nuoxA;9ijE
zAT27g*}x{CH{!9q7snvEx}Te#ou7XNhlVus?%g{iH((xaUS1KqX$FjC5m8UVqSMr5
z<m0n5Fo?^~-vDHR0P*g;>hH0!mlHu3c5EE@k8oQ<!_>$qI4H;t9Vo!#o}QkpEM6#V
z8N${pbJzN}E)11Cd7J+!<({rye9}Zs&~M!n{5(DN2l-zzxgY5S@^NyarJ-@Ow@=p0
zYlA2wB{d2lfh#>0y<N?d$=&kk!}q--3CXWM$Q^mZ!O|ROM$Qp=kfg-p1DTE(kXCH$
zby?Ym(AR@FwGGg$(AQu5`}aMcITP-pV$|(+I_-$-s?tgBktmvH&rk-GgLKk$Dpsja
zW;?71A0fnvM>i5ve35H`QDYRqJp;dm4!jgN609Cr702S-8iN9PYwyNOuWpZpT{lW;
zlNk>q?Lr!HFR?(E127*&2ty8m^3tVrs6BGw!tcWv4UM_{T$7GI)bH&)%Rx3%Xcl*X
z<f@5-U40|ZO%ooJ=QwI6rs?r<$CcT8T3TAijzyqT<Fd7}3U>(c_7k<?>Z)abSk0d<
zCi^0SS?m{WsE&o;;$|58-alwt=+~U=uG+naurms*8`Fc#3xx+x5C|8Lj=*k)zdx&b
zf3wIf%(AY3c_45qh?eKPFm?F$*4q~UyQ$h0rmC^-Mbd;*Z0Lv<xn6!IhM*C!{QWd7
zZ7%Ls2#PlpON5>1-EAI*>;}3EQGcwJ$CEsMQgYnud~Y4+JG@FY(%qmYYB2ORd;4!w
zW5BSru->bl9Nx#j6q;FH>9CSqc?D5ME6)~^k`V|c0IR6ek||a_>@0UJTp63kv==fw
z2mhS&ku#u*_;kFa|Mt80w}l+cJEk`0C7ef&iF}@WT-p@&m67>R1*PFhRPiphq=|?b
zZi-*we^y4j=-;^6_pGh0fs&(Dr>ff7-M#K|l`v3`4jm+^v(!H!U&GG_c?-f8gfvuc
zQLUn=nc4y!C$qt-h2Io^a`$)~7ZfGgig{pX>G1RMS=0VS^5mxO>B56T_lH-khJRH*
ze|N*uMl>t*k*vVtwC92CdDm2}g7NK_yq%q$IQ~`&17Y?kTj2P>!0WKE0?R+;ps2`e
z=zV0z#YxD@`c4{g9VddVJddBN6W;4;;Q&{OffKY0yB*uboedigk(8YM+Ok?w!x~@f
zL(`jGFe$v0G#!K0otU_qeROJa5_43@M@9gww0CfT!XjcntJ9vM2~!Q$8fXVHd9;KN
z)FSd1u7<r|s%4or7uIS$vie<nwde*y_>I#h-7d$!Gi&)?@ayjg^|8&2%IOnzKgo_a
zVup}E&|4V#^C#f#+aEC2n09<YMSmT(k5sy7%OFJ0vhTjn-(*V;t2;j&x<!U*9@Va$
zZn@uN=Qr>6otfXdqW0S2=m6^v)4pS~_JrA5j5j<YD#>?YlhQzc4$}^?`z~@xh0goC
z+gHP`UvX&P%#PcYszs}8XSJugE?B_w)CiTcMB&SC>MUEu6+7KimF4BXK(oMqfk8oX
z$ns!FSmM53Uft=2RGp1Fgw0Y3PwSW-4R(g09Xc2BI}vp90M4NPjtmbkjWx$ZQd?PD
zo16RH-%&#zlun=%BTB4#dKf~$GrS)5<XnwzWcf#j=<cp4?f}io&TjAM$U*lFMFJXJ
zR%Rx8Xe=xS=#(HG5Zmu`2f?V3P7fVBrmdvJV>MU_=7QV@F;?dN#b=o80Vtky2QepE
z=KXbLWrvR(fxoH-4Ws=(N}iA*2A{wnRJ;LC%1nRJHm>==NeR>gcI`r8xPgf6Pz^3A
z97D$nEl{Ku*y4VVjEFj~e1R(yWlu`##pB02FbJ>VVF-qCD+9p}@n1(r$KKu^<^_3q
zdH8{j9Xm$+jE>FD;u;9-u!BYalde7|C)Y)YZleK(nYnX@8n9siEr#L@*j^SE7h79e
z(5`?W4TWA+P0bNWF7{@SKO+U{VaW1vhtWTajU{rZJ$qh47&t_70X{3l`|qCBNK&+3
znx*CBQd3jqZr$SH<#k0y8>$Y1RU`jluOP6U4q8^lq0o}yP^m62Z)<G)H8@C3L$d^J
zzApH}lSz#$|DoM^$Z>mX6Oj`T5Ws%^yr7T}k`X#15Z6mfORIcIPobIQz6Uk<fr-f>
z5=(R+!Ha+<Z67_lcKte74yd7&JBmyv2wxc5n^95BTwE3qp|NI=@iDN;$#j#RXEsW9
zas`6_0}McL!%SUvS^}8m2dGq6pXtdB0QIFuCr;+{$&<Rc8H5n(wg~($kQm_+Me2!@
z4U@LzK2Gt9$=v~^yAPICfI$o(G#5)NtMSQ6ZZ581UNid6Mt2#=`@<mElqeTFc07hk
z!zAk3^Zh$Be6Lmhr}aku^S>uaReTDb9k4t!VMoTsa^1FF*#f_T`@*~iF*7ZCM-`-b
zu9wiG(d>rA_Ta$-sBD-u2UNn8loT9h?p_p`4k^eH$TBeAIZjS^z-sI2;5lTxWI*1T
z>&}5ju8uu7eQjr_07w}#^KZaIh{W2tR;Qnd?V+S(5u$I{g}EqAN=Vp`=KlD&8af*&
zG|+;|tE!G1I|jCY%a;4&KD1f5vLK(J06u+60+OU({SJ3df8@yRGt?Qv8$@f1ZhA(B
zF#7grqQh;34g)gsE)tTO+FC4#M<iGf#ozMs!0f=X6JuhiA>Z!b50GuMxweQ_Q~)Ez
z=x&Vg?c0fpP<g@s+0`<n&<#eH6}UlMT(>dmx}M}tV!v2N5n}-R3Lrg7A<jcXL-Ri@
zT{+UHH;D6e8~sONj96x-4F(5=6O;{Z8k`zk)doh~v^X2ggd7}bGPbw0fIBl^-4DYe
z1chHCBk18aFm4>8a{o*0-T*EDE1*#Tw~WoBN64lq#_2C!?8bC65obXdQ-C7P$jFGk
zVX&~n9I!J)L9~HRr&;d^1J>dS4i689aR5P%9vUlV0}hD7LUij~VS&b72|h#D^CWya
ztqlzjBXMaxUv-~DpTr_Y)U&`bgn7_LPm`_+ogzO50eB%z4GgkyS<Ufsk`fXhf`exP
znlXyBSj7_l@81g`{Xx)$6$A0^P0wSD#esoB?0ka>Rfimr(bmQb|0<L%bXh^qdJx$i
zUv+6=6T{hv{@QYjp`V9u-)@9^0{%!R2Zu1+9ugXc;$K6YzCj30F}hOZ34C8BQXWo?
zVNe(q6<E-I&CJyQ>(YF*BVKqhoXjwJFDxv)<Iw#1{ks(=9wx;U)pSzYu;;0dqsTtm
zLGyup0lWljN)U(U8U!skm61pxao@D%)*(kEHsnE9k(89Ye!UFL>Q)FZ`de7bWF#a5
zX}q(6V9Sp_gTNpYBPj)6)c=XUp_U%qzrUfc&sX=$>>^U{M{QWpy3+K$!0bW=tTCy)
zrX90m&8(1QoilbjA3}(1g2i>BCkb4EmSA8Y_*)VN3N2BX>@fLnka@4fBlCi7!gv7o
zWr9zjb^7f2bF?Xs($ix)=*Tf0jxCLik@4|-Ku1{0ASo^`n*b4TU*p=b-2Yr{+}RL~
zlOUDmefyU7^(#8s;sEyORiYu??XW^gym3Z)AMl3O=H@?RW7zHR7hO;I%sh_EhVoy<
z;7DPVN9(@XOc0S$XQzX`J&e+l-gJ%4%}-(}ocVA|E;nRlX$~C7b=?%)x9=u^KbB2o
z<S&2(v_+A2y}=)@Xy3noAD9EC2M$nns9}g#ocYL+s*)1oc@)uqY<Pq36gtd%qJ5l|
z1&{3|Vc~qF>w*HtpoFK}@S<Gdgo6MLA|z1M)@HO7(Qbplv>bK|&0QT+;thGBX<%hx
z@$~s~?L1pEv|gcJ;vi^dQSU{eE$%=XY}jRPKJ)tZevjj5f<i5a>0@Yc5c$*crS!@S
zW_!@Y)Rdf@Tt-UjZuUb)9|-`6dNfMa)zwoq^CWO8Vwd>!%-=s1aK8Tf^$XEOz4tU5
z+oz8okDocSyPetd7}0=S^{~Hi>enxrM4=gE=~ek3<$HKu^IOKJPtrt7x>4~dQvE1%
zr=u<<e@A-j+V9VeCl>X#g&cG4xRm>nm#&`bN;Vv-SmcaHk0yQm*wETKG(2qZ@L?hH
zI%>{qMvCByVp<nD-;g?CuRNt5hKT8oq}ucspQ1n<7HBFobHh3>4#-hmY1hiPFG2dH
zr+3^~vE2L23I!U=2pPj+ei&shxWYa4q2v4v37;9ot;-@_ULQL-vlT9FuTC=*9X`3e
z=!<lRb@Xqqg%_Bcj{kq;S#ay2gtxV|p=I`2Epr?!1Of|e&k=$)Y>4GV@|A0HW)B|5
zdmk7AjnX-iX!BiM)NFf;jOT7AVSj}$W1ib~VOEwcOqTKS@nK<b*H0vW{*2sSV^GZM
zNV)6tt}^u+zgLr&Zin#P&<pOmz<IgVrkqUUUCZt55yp5xC;gvqi+aH&Rc_zDVDVrF
z=`9#9tgX|7g4%%t;e#2QoJ5YE{QcV_N%1=s->i_M2t|8dUBIQDZ1q)|h5aO6#U24)
z+j+ctdXI7IBtMU26A|fyu{tl$>HE_?$OX`WU0q#K!Hpc~c0aLjJiV{z`&#p_3At_8
zogoS3g##+<j$LX_xt6gkbpFs+AaW6v6=)t9T!bbnty%B4z)6kbP$^-*@US2%!@s5S
zZ{NK6i=2yi2jBi#RYgP_FsmYaK^#9VM9;;Ve~Cs%9-)iv3X<~j7Jtfp-n@AOwuPh*
z{vc6ZT3jMS!IglULyR%f(%LPIHo^>|diO5Ss%dB1Gn}yop-vu-&gR0xqmw72LA3D+
z$c?a>rl@6NS?<IiE-5&YK~4k#g)w~)cE082G78yxV+^5KAxH<YAfEiFuAYDy2t^X<
z_4MwjZ{Hk|`;l-aCp$iVFaF-o1#XKL3ihHP5#1~G%+1Le92~T;u)xKXb#w&dKpQyF
zPLAWp2awuvDy{5?VL3S~o)kxM1)Cci$l~CR;{Uy3(kVu+Ds5{6QEBVxaW3~hf@{Y}
z!6TF-c8r{o`f@P=M#tWK2Wwf`I;>6nS~*pV=rG2oaPi4^zj{nQ;!*_GC4CJ540Bpq
zniXVLEISkfe}4rOAc*@&PvX1lfV|Jt*482=VYdMsL`a}G42+s^%EIh^faqdjBtLP+
z(S1iFfIixYd!I!mC3QD918UA=$I$7U?{f0;z#@Wh(hQ_<uH$H489#V%_S`wAf#MzT
z9YQ38Uk~m;L4JO$TTceugJ00=*8|hj94sv56&2agF@c59@x%rfm<P!WyNU<k7~Iey
z8i~NMeSB&w-<-r&2eig;ztC#Hz|bO;uoa|xg^?BQz;-lD{u&2&e@jbDWaK?Fv&&F}
z&@I6V_dLVky=&zSY_5NSngK%NVq*cfD36H<!`m$u*@MkCfrq#Ab_mS!Q4KtISeiud
zw;Lb|&X@IS&LS*!ZWz~WZP9f}`!b+==`lA_;07I<R{#JcSD`w~ec--<p;J`k77#$K
zP`!Tr8HOVUW)D|)1PZ%))?GEVw$|3pt}YlH1q20=Dqz!Ut*P<G=UV=K4glH8(lRq6
z1KTKW!UW&hIn|p7gUBLaEierXpBsiSQU9#3SCN<Bf9TM}<RttLIk3$DjiKWLKX8>S
znfsHkgzZ}ZvW|{x0M01<;3$akHS}Z9A4fTj9C|E^;3(|J3Gq91H8l-QO=wkUQa~P^
z>CJ<EJkMd?Agy`F25Hh!_nDUNZft`&hK(^;XxI{l@kmHGgHQ(#;N#;Xehvg}_A($S
z;3@2^z>+xL<kZwH8JTb(98_|U1>)y+y<!@03V#0{1LBP)80req>2u>oE-aCThK6vb
z5P{$rVG9*+*35|e1KiDsaS$3D_82<Dy1Iv$m=HBf^Yx<9Lyswc>6|al#+H@3+FE$%
z(G0!&MdxvJbo7T0YN+@9hHe32lmq(^BK^k&1Xi13iPv`h`gIgOK%hT^gTjsr36CF_
zV2Ckxv-*hh)WfJ0FuB5xiIj0GOb`|~UD!jx8RKQctnBPqg&2!PfkK@F0dp{(P>^uA
zzw`6!tE*_*v~_hkVoM%U7Q$4IjATU;2XzYy4qm|}L;Diq=ZAG1FJuXB9v)Ce*hxJn
z9}$REewGtW%XI8dLH7vCAmm6`PEdeu+^9nO!Tu5O^Z>^F1KOCttssou400sAnLv{b
z4HifiSjdpMac@HC1*82{Nx6G=#)9c`u_!h)^kzH=$jAUMv_Zo?J$)W<0firJ?_-RN
z?B~wiw7ue<3O=xshE5fb@U393UvOPw_ZF0XNL?f_-Pz5cNxf5ziEg(tA3+-{_~HHg
zV$qEtaT8!$dU_f79+*6+R;ZwjF^s~*5g-MU+L{LThD8E<_|T$5et3`HLk7S-;NPZZ
zX8$1R6?gL|FPR3YpuocR4Xy%00`!jESpZS6wxRKWLV)j7bZb980JgAFi%W7{8zcIN
zfpcz1OCz7u2Vc;In+o^@Q;oz4IY`ftWQ7R`9F1T!P<Tw*WP-U$dV32J5-I@(&|N7l
zJ;~3{&%?6_F3P{Q*LxozSEz`KAokP2`wWZ;-4`$bSrQimO&WR#2>;dUo#LgwOr9C=
z+`%dbu!*{fIwEoXI$E%5s;UShVhG*6i$fol)PCF;IeKsOm9K9&pShN#WX0p1WGI;j
zX=vWTK?-UG4>HOa`j6KoBp?h0FdiBx!cHtt>=}wfZimk;Au$pAJfP44vSV30B2T+S
zOcwxPp`S(kd<8ZQir2wXhMWZya3fM46#yG_AzTveS<t1iy9HK`rt0c_>{K;ilH!#R
zHqcmcTN`7AB8r@br3iZ<!XM-3vNC)1#tpa@tJ~XE#KoUX#t=jgzGCuxiUEP%bSpxn
zAT3P+DrJ4W`HvR|Aj=~2T3`=Xie_Gd;-^=yq@lb+PUyV1!<CErGFL8f8l9FZLBm*C
zQDT66Oa?38z-n7fbbpAt)>K!+oZL}UQv;Za@VB*P$o!85mgKnBrl$Ky>@di{VGsK(
z00uUiba!+RoBZ&sqa{zhS0C@l@d$>0W>3!ID_{eMAhKv@nZEb%5JU4pS2r4<D>-=*
z^a=(}Z2kxf4Mn6G)Q9)Bv}B;@0L#Klns#z0`EgvbA7ThrbmTHf@WB$%JbZkq>e;fW
zqu9C#GK+Q~JjGE__aRAO98o5qm*5Mhgm{YGQO8QrJj`+1-a`8nY75SVc?2{;dW(ws
zjr2xzx}H7_5{O-(kZ(|c%l(c;1O^UGPuG=}ird<{m$-||6(@l3Lt^<eF#+1&ja@F)
z)p_t1vITw!2snB6Y+^zJwqHm@H!3PAT^1Hb0~bpY(l;F3#2sLDlw1k&3&m16h<1VF
zVFrc`6pRla=ppdP%F42{f5sB}l#pOyWrh7~#DW0?2X2P_E)KH;S3zLR%wVX+ZmJOw
zyvnUT1S$6@vTqWx3aFAHA3m&s#v*NiyW@ql096B}0(l-t2p5E??kRAx18#x^bZcu1
zcm2@RG~aR2*yX30&u#+Y&>-=-04|$QIxs+7Wu@|9)oU1Bf3$3X{IKH^qo$=*S5Z-c
z{Zg8mX6EK@(6PXR4Gj!XW4k&!P{>U%=>V;$h`<SeJFvRJ_6uZzzDo!4A^c28<W*bm
z2~s%zK_pI8SPV}-;5f)12q#c4SR7hBGkn-91<!m+bhJ2ZG40xJus_3)3>czO;AnI0
z+SBmxKEOHLKlXbv?%!WrR@REzkL^-*b$v}u%#h$<O2Y7C_okJb3_<X5)bnHR(y{ob
zLShT3FFe;c6~YDPQ^@jQcW}5Pks-d8&00;2jp3a_yAq-D^w~2Q#Nf^cjQA8Ezq~L8
zr}eY6w2P<^cmXTIrG<!vX#xj@rXCv~@AcvZG#la$#=N|BBnJo^7%(K3a%5u6LVL36
zNkFJ2H!8x~_nPb4VBTtK=md1PwUuN4YG)_-X=8Uc^b+alMsOI^j7B3QV+;*S3^rZ?
zfJ3Blu>%PlTknGUvK$x^k{2EQ1gFIMnz_2h^lRB1?AZxegy|K2I<o^?Gf~d)zFs&o
zFfckjZJ?_=aZ@Z-VUNeC6N$(1wTFNW`Z$$BB2I;Lk38#Qhr>5Fi|@bI&xk(%x@$*a
z9ssj^VoP727F%G~G6l#toU;81)tG^vo*KM2G{7@~pFv2tqpc0$t+%7268F>81g9&}
z*WKNX71Y|-CmY?Urmnudfh`m1Gt=<rL+!-iAPRsK5JN6<Ez5rp2zcE|RNVVaK}jk0
zpDdjIpg9h0Bh22~`ufo$NAwL0UcGn$Jt{sY#~uU^vw>s^IT{7Z*4i37^Wx!?hBq7r
zF_=n`qF^TmVuUWB{8<e<KCz)d^!|?*&;?N#pyXgj-<w4~3imn+=|;PEm-poy+#m2k
z{eM0pVUpaYv{s!wj@<A_Fe`TFLQ+bFv-kY@WNc(IsVTJD?lzxDQu)1ovnbVJU3z1A
zW_@~VWX8>Wsp#^+Op%X7^rDhOzlyEx5*C-J+Z24>6I=CEtttyHsbnsx6fT^XFcPq0
zK)uIExb)Be{kyrPfPB)5O#$tnx#@kLJc&+E4~U2;N=$qy7Ku%fu-^o|eT&SF5+eKl
z;%Olvb#PN4`x&e%XIIxP^ePcf60jLHHeTKT1C9phvoPm3H#GqY9TT?yg3yI49b$ua
zjzvjd9~9(yFd^)u>wp&*i_Nem_)SQNka0@_R!(7iOQFhD2v?|Zkb0o_H$WQ3W)lDh
zOfI^9s0Ek-@N{T9&=i5l5VS6LHbB9!lCfJ8x5mi8pqza70F)q=X<#{YQPcsmv2o*T
z#Eun}o!Scp<^}IKt;RIImFWef#8h%o8as)X<z+mt2-1+tUmC2rU-Ec#4WS^7B3M-g
zh20H(XaFZ-vkP|&=7PI~c8gUGP><Av@CSE*poMCHRgN)*c#T4Y2Twq*L3zgCq4uCE
z;2lX7N{yrsJ$59EAU+CT=3vw%qW{j#VRtIFW2AvS!x9hr0fzww6Yqd4c!y<-`p}_s
z{}gN)ZGRjKjT3BNTI$W7`JRhe1geFzEL#2yN)O1pi}u9C#5$7(dPz7WvL)65kPZZ4
zkb4Xi=;74Vl#BRfpja|8vOsAFN0aLpSY?8t&hP+l0uewV1<e6ULS)m?)6>w>9;Ktp
zcUe~!7mvoefIN%qM9AapSln0yxL{OXz$B;_D5p?|=a-j7fy#mWaWF6vlmqP4p<>`4
z7#wWj^T&oG5wJ1fJDdewyr@Pa<RaoqO%=x#0*YgbG02!}upn%*%D4FC;ptf(DgGS&
zC0F=9P?|vGU%q?^t_94uu$hZC_7!Xpla}5)L^*qlbR+ER<@c^ZM{OlU7X*z8CA8JW
zUEai{{E{#++^#76<i}CLy*bZ_jVU$jFa7+?O-(o2Ss@`L!ZnAw2965c1z9rif3!E7
zniQ~O9gH4e3aAQm02{<-`1Mf?;U|DDR`c^GR2JzQH&%<5rY|QXp7Yzel3VrcHC@i0
zUq@;_NwvOv(8_i=)969#;`=g{>+GMujbA*RxLY<}GrL<g`<tzrpwdKg!X-fy8=7xr
zpKKmjx)4a8N$+EmcPP3RK-eLcv%j)eG^H#h_H9aQOvTIy6-!TU&j!EsTy2#A>kZUc
zBr3dPq2V210Av!6g=kCF*8V5m9So?VK5*bfpfu5wj6G#yl~{}35`Vy^%zJXaLWd+`
zol~di$UPCPm@T}6TSEWGI0$fXq@b{}un<L_rO8fggo7QV)ri|T(-BT0-VihV5R8Pg
zU_k+WD5QIoZ)fKX^w5Y8WPoCdlz<$B?&OPr6F^s(R2na6UuQ4AKabhNYOk#|!#@$A
zsQo5JM)rTpXi+S+T~~*YO0dVBnwAz@gwmig3ko7@sHL`t)DB~_W1n%R=d)*fJoX<v
z7^RTt3_`oPsrrs$@7}cTDZwz`5r;5Y!%VeIH8V-c=hCsT<XPI;(Z(0ueRtTfkDX@U
z33Qxqga+Z!BGaRz#78Zhp@!|YVeAoxi3u4CI=Q%*7<Ov`9ig;?6`Ar9TeNsw0|lu#
z05aU!zIc`jH$BkShc|D&qM!ir#t0YJFBVHKc!vmeo24)ISgUx1%`l%ADxR-F*=R9d
zyi8^z7$(EAUL~}+w8R@yYn<6-%1d?RB4iIRnqDLBwDa9SAG^h{ZIEB65hZT-gNF|d
zRF^<<FTPJMEYwO=R8~>3W@uk-hUpw~Ge-~~FYl}Bmj_?|xc{|d%GOe=`(buAh|H;j
z>7A3)(|z!i%Affm3cfqYp8~WWtAOSGf;Om14@8f;B8&2Q>HV)S4*jO_D*I&-EW`3n
zAIWXd$^j;=3QMFk)LTS4y-yr=OXYu7WJ5j#LItLbQ5Zu{&pw=X`)c?I;OgCX;GqH4
zL%?XbEgdcrQc|rPixX$gi~?Q?h7q?+seN8TQNR(=BvoV!)OJ}6yvDq`)NSq%Zb(;d
z%6l`m3GH}LWjyoa{HarV_2HKRP{1fa3d><m``kW3+^K>+Bsg=5c9AQHC(^+!?D~B5
zN)TBLDH{onho1OMmi~icXvCLt0SHUSrp3hQb!9w&sHB-`+=|W)LId6BYaVjLgq72a
zml`+rS+_8%Uto`XwBh9yCo0g>-h697{b=>Ri_Nmx>~hu4PIL9^!$Yt*rgv_PC#6z4
zV~<ulGBuvKLdn3-1q}~q8I>$PHnuiY;1cQtH+Kds)78~p_kxjKu{jML3qUvnJ-zel
zS(7L&;5kIz1FzKjOi>XeZK#lz%8uCOfiT3_{eohE9<bM|SC0h4%I-rP|3?sfj*0f^
zaMgtHabCXc0;-MO3*MkonI<a8n^smtON+?HY$>282;*`q#J)rH^lWhgrCc;#r{RTy
z9twbiVY&J4{DFR&w-6a2e1g~gsHniU=5xTfm~j@CVUP^R0ueNWn__NYR*5CY*!VtO
zIrb#rStAgXPzixd>uYK>zUW*9P=quIIDnoO%H=n>jDYlDG=jK?PADBYsb!p2QtO9W
zgItZi-rgw-{uqU`_460s%V_Kw#j^vXPW`4&)!@LB4@US?GBcH86<+E(hmQ#O21v)!
z`$$B}WB->WLwu_-G}jC@vnyBJ!7kwXc6oG`u=lfO-jWYrADw{J;FFw{CsdANFQSLW
z6aDTKRh~=aQJp$SmAUlQvN>9rBTR51zG8TpPOssRn3c6P6cSamR(pDqu}WT#Pfr8y
z9a6bkrg`r7PuA+R-={9t&;K?!YMFT_v=*E7C)HP}?Pgv$ts|3UXTC7`@_X%5PDweK
zo`%TO=%_qfpmLH@Pt{3|*rnF_O|Si8YaFLjek^H9_v}9PaO=Q1YCo?$#~)rVk}Akc
zUx*&@JRsU}Q-fOR4E67>^Y!!S#~{ZtRQwfARY)}DJz8Ojq`*+oIc4knhk^3B28W%s
zbwHIKSU8Xo29715YRKxD*olBDe&3T%vEaH)voMt})!W@%vkC5bOhyf&Z#DT2E(WaS
zur5;lNm&CAUd1AzTI4vUB;`P$eO@I`ApAUm$BUeK{*jFEk--BG{;=BZ_`mz%=FdX=
zr+;$5+`tm3<~C5fOPqIiovb!UKTEyWQ03RR29cfO>*O`REI{lnU$deWUo|vGrdTDu
zntXHw`!<E`rbE+{EHsTG9!I*6dM+Iap2(nz!vll<*B2n^(2zcvYPvltj*rrH8S3DB
zxw7l>g#~GNkN{l0ok}?z8TG~g=V@)ap3mVv2tVU-M8Z`n!W97$Jnv*^Kc2GkR=i;+
zL7wve{-MT@k9tvlUt0*v108q&)#)peN^Mo%=3UMa@AHm3pDFZyx?fDNNZ`YUs;b}0
zJ|V|@3JZ6Pf`uGqo*iLL|I~S#bJS2I<-7Z+Tja<Fe~=85n$VX9N5|%cADnqb>jg;z
zVMhXVw^i@uyXp+ZJUcs7P<OsgX+b%B?f8&}OP*58g9@jSX^WmUy~RuOc3Q^C^&5Zo
z)w6_~iC1q>ErcE$`c!pNK0bb~Whv*GoWsrM*S5Witdkd;4{sgXQfOh{?J@P;BwC{I
zUF%TIK~0kF)$x!0o29=eFKWouFBZv?zmPSt-7RLmS=hf;)73s;VS9^X*X4N8lu%)6
z-_ax2iv!-A`yJ%j7xr%aY!&-Jc*uovpZ&~k25pM!eX;c3V~piL>2tsR>Di(ZS#kOl
znNr=km_r`Pv#rqCoX)I7e!{<)*1(sIoBGlALd0dk2J5w<pH<@b8g6c|)@Wzg_t|gf
zZL6nNbtL~?K0x)Q&~-eno8@`>xb=|&o<JoYRw`$O#y*AMw2SAhy7fcSZ(33KvP7#6
zPq50SNjq;i3;M8}V_}^Zs~-zG$4YhB>#J`_kjQ22#?xFsD-6dYmh{&It{;DTZ|&he
zk3WshJ$u{yYXWw$ki9W3r4`{@TYgR+@||T%LRQM|tz1XVvTS)g`=3zF=zaU2<`q2K
z&%D3nv1Q5fmr;Qy!^~CU*Bkcn>MA+p-g=$ja*{;Mm2N?ypK!~oaq{>H1u@4E8Jd~f
zBGiTNj*#TV7DnZ#1brENsb!!y%H1zi(JNjW?jQP@PiXhD5pT`%4d<QCE2inUj$icT
zuZc6D(iW}X@}%-(E2q6|XraO4kg(hF!3)=F^|c+cIuXIbC3D&P0t<dc$7=2xu{!)K
z)%&$b#+=6P=<s}l3BA9z=Bv}WH>LliiF}y2E+waW`{jDVC_z|x&k1jpj<>hX!)vn&
zqn4><R7W0qt3wX$yq^2IZ?1P=gEeWHzW?)AAGaTbTl1Pbt86!Qg&Xy@Klc^=t6Z)q
zY!O>>`<&$UjeD*$-&ViH?UNLt+GtKx6%W;XYx3Gxb#rb04j-q}X8EIus?GAn=TSS~
z=-Ne&kcTIoqRP}3zF%OaXTJF3N11M+(-w8tIS%J7{d*zWEBU(uJYLUZxj2TTY>s#(
zo3Qv*gg4hSPRvUc{#-q<)UCa}K0m3Jec{r=r3#DSoLn-;IzP+W!AM?Yo-0pPV*VGa
z^MAQy|G)oGw83($chu(jPp3U6G@s?^Do>ru63RQ}zV7~!Y~}HZdtTw6`?7}qkUDi1
z{^-4I9nEBCZF_^T{P+M#Nhc2>Cww|dh?!QCK_sf?+lI`VF{8w_SrgBXWTO#Nlb)=@
z^vWYc+pA~WIx`79wB(9}wUZe`aS?Va!6G-;J?(7S3Cnw~QZ-PiPYd1Xv=$X9(*5z~
zho7`ls3v8x%hiUR8&}%fXHGlwTCRuGdnR}MrJujo^h@n|`py4xsDh)(a)KI^-)OUW
zQ||jkC-&{A*POS=J>ATUhC#b#x5fS?w>x+Q*8d~V{eK@c!rOJuTpIZ@x8P!J<W<6r
M>$2BU#B@CV4?T{N#{d8T

literal 9302
zcmd6NS3I0=@UL$5)mzl436T&jdXL_TwxWgT-Rf--5hQvSM6~F=vw|SVDp{S?yH!_r
zHov?7=W{;i=A4Uj@m_pq=9zb9-g)Mk_cKqdj+P1uAp;=>1_p`hYehW_49v@a4*(lY
z`Eo&~iGhKQp`&4-goc=ym{?d?*x1-OI5@btxOjMY`1tq$0DypifRK=oh=_=on3#lw
zgp`z&jEoEj1d@}JQ&3P)Qc_Y;QBhM<)6me+($dn=(b3b>GcYhPGBQ4T^oWUxiJ6(1
zg@uKcm6eT+jh&tS@#DuF92}gSoLpR7+}zweJUqO-yicAydHVDzA0Hn-KmW65&jbVn
zo<D#7;>8O=K|vuQp_eaT3JVL1h=_=aii(MeiHnQBdi6>|LPAnfQc6lnT3T90Mn+av
zR!&Y%US3{7K|xVbQAtTjSy@>{MdkJD*Q%<jYHDig>gpOA8k(A#T3TA#+S)ogI=Z^L
zdU|^L`uYY228M=)Mn*=)#>Q{nyfHB`F*P+cGcz+cH@C2`0D(Z3mX>edzO}NlvbMIi
zv9YnWwY9Ued-v|0y}iAIgTwpx?;RZ-ot&JUot<4=TwGmU-Q3*V-Q7JrJUl%;y}Z1<
zy}f;We0+U<{rvpEV6eZxe?UM$U|?WSP*8Aia7ai<XlQ6ySXg*?ctk`*WMm`+0*Q)>
zijIzsiHZ5};X`a}?8lEE<Kp5zefkt1AD@trkeHbG`Sa&5U%n(IB_$^(r=+B$rlzK)
zrKP8*XJlk#W@ct(Wo2h)=j7z%=H}+*<>lw+fBpKkprD|zu&}78=-an%#l^+nzke?&
zDJd;2Eh{T4FE6jCsQB^YM`dMYRaI4Wb#+ZmO>J#$U0q#$eSJeiLt|rOQ&UrOb8|~e
zOKWRuTU%RudwWMmM`veeS65eecXv-uPj7E86bkL@>-+ihXMcbHz`(%Z;NY)czlMf}
zhKGkoMn-=B{yjQ6IyN>oK0f~E&mR~JHZd_VIXO8sH8njw{rB%*I2=ARGc!9oJ2y8s
zKR>^)u&}texU{siyu7@!vVuS$R##Wo*4Eb7*EcpcHa9o7wzjsnw|91Sc6WF8_V)Jo
z_YV#Z4i68Hj*gH><ni(G$;rv->FL?o+4=eT#l-~*g}S`Fyt=x&zP`S>xw*Z)y}P@+
zzrTNYcv!8m2uAM#0{7P@UKkie6#pK~iN%mx3=CEbRYf@iaL&OZzPY}AcaO(WV4&O0
zQZqFSV>pfT!;k0nebE@C(P0sQ!>9^ZOwlJgD*c8PhPtp-#Yg@8?2e13%|1<}3m@uE
z%zJbR2y&9Ww6u~=AFj)~xVXo!LN7btvJNHF0G{g+wc%r&g&~P?5%h#mAu5PH4`5se
zm{-6msUQcAj`;tQ!4ACN;rcpo@}!KgyLM3UYZq9QmngsT)&+CpS{D57e;BO??M@qR
z1wt!?p@<RtxuC_>r+a18b;h#bMrE-eYk1+7;dorp%`XrpVwwSg{Bd$EQ?6-O9fow=
zID}2$g>pYyvU9p)!JqlJM40`6<;AQYB0U{QYH7U+TofIcy=>U027=;$<eA)<t@jH&
zV+*P7YIE<;*|@2AX#H9fDy3DZ_ZaKRl>L|mdtj{fa1hJGEh?b)i`oK>&MKLNFO;Og
zVU3nn7loZe4jW#k`Mk?4vwVbc#Co(rn@&#pfZSF2&)P4oOw@k0r)_r|c91@S%ZrMp
zbK8GrKK9u4fLv%hxzisDoJZxZTaqHC`*oIk3KlEyS)AS3j**7PCjv1@N9#~Q26FOW
zc~eNe7O#@?&3*apF>M0Dba}9gU#81iJ@rP(h&eH$ae-*b?*8IgTx;l%IVl?cz1rRp
z`_d;hJHpnHrXKsoB>GUe3meUu&k(z?K3wJ%kFN?tnjR=q4X;IKesT&TfxsQ-x72hu
zsNKZjvS1hAD5sx{`oUHCITfrOG!?54^fET)>?$A|Kz08!Vg=#7m^Z&dh%CvVPPP47
zYtITCEnDGo-~(Y7;EnMEHa)#(H3Vt8mkY2Uxo{EgQa)@1f}AYkS`+^S)<!t7bu_i{
zPQ;O1?pc*aRG<-I$FUcjPkTum*H{6}fByYK|89b;DUr&84AD5H9Eohy<In?3LEu_M
z`2DvppSi$Lgk)jJIaM8f5w#!5a}Mx`an=r_a`*WMyn<%})GU~YMi_R%mBUi+7L<!L
z9686cxgdV{(kn!lG6wA&hw`>h*YfOlG{@ATvkLsvi|VAj%!ak&!erV0IBc|cBx@-H
z3+j~L<JrJA<No`Gi!1_ZS^~A|=Y(|0OrPljLGzN7$N}$sivfT5nH3s-dJ@F1+Ti`v
zdGk4-+U2$!*d{Gm-P*_6k!Iqw!X_Svm*aLQ2B@|O+4EO>vyw^w%uVJPkF~>h$B)T^
zY2dKpQnytb?SlUxQ&Ry0zuEHb7a%A*CHDF(htg>CwB|~ItwY45hN1YF$;Hn~j0)<M
zET^PrkSWACQUC<St!MbV8x_*e*E}LZG!D+Xj^mEpDMJiILwO)@u6n<sX7!?WL#tPG
zP^b9jf~91Zg5b$oCK5!WQq;IP%&0m$iD-s2203RvsL`OZZ`(z=h5&-hY~S5S{mJjF
zWk#osVpWkk{M&BoX)QDQlk{PCmxm}MbA)Tkzz&IkAc^(64}WTBA6?Do{lZ2xLih4$
z$njzPkKE4j5sh;3wo=v4omHCqX56q4jlIQ~+pdz!Uoi$XSvy2#7Ea%CmLK<eQ39%Q
zn~wRp3GSDK8OB{$J2oqj0r_hZ3xB(ZXjO+JkXc92=MZMv7Ll8ubLhYg1WJA8%(J)?
z!C0<Dqqxzy@w|)RDWjlfv}lz^#FIL#^N>D&p(JR!NEZNFF+dM5M(v8q^v1Yf4#pm?
z<+a_8Y<~P#1vo=acUoSXznP#sIsn9w#r-Cl%)SdK&;O5yQ!Q{3$cQ675GVmfm~|<l
zy2@nj+VTZv*G<P|@6D_+^g79YC{M$qGFam8KDmSUf*P9osl5l}F9<I`8Hb;Q^Y_0<
zEC*k@sfykH4Q&?wC89PplGcd@t>~#ae;Tm<!Nz#CoAWsEyqME&+hJ}=Ib8uvUD_!W
z>hStzQEh>uD-muhI;sK>8Sn|L`l-p*f%lX=YK&2GBS$fFULH(BbF%K!;g-E}GrT|p
zb$Wg;#asTUeET3e3lAY|WS0~|g(@{b8kLzTfa8U|u5b^>FLX03<zYYw{LOvpHY{=J
zX-bDM1a4qF`fA)iPC3`;WD^Kdu%OXkY<YE=Q>r&KuK*?yMoOA;ygY6#+CO0Hh(4LU
zqbuldRa<6;I(^HMzJOt#>Rgf(8;2no-cRp~b|o)jzvJ!3g1}!l&N>p}mynwk@B%^n
z7W*p`YfnsDVtXinAnZlz{d4i35Alh6!uo=&9eBet;acAHYfGGUco4X3r+xx!LnuRm
zFggwB>F8bx76xD>90;tVUs2-D!7$1B$Ahre5dGOR>xl|(x>shlUOxj=zaMLV6DLx5
zu!EBxj%3)RKVk^=YWB}j!GWXUyAiB2$!R!2-#m#h3DhZrz)Ib+puH1Njg1gi_e|O&
z9P2x#TY6JP4+IsN$H~oia76R|G>$+r*aZeqm$x2hLe+qvWb-pY8cyR}W?85n5EN-}
z!pu>ea5Qw43~5CBATIhyRWz|CKz_g!&HUZW;(=!<Jm>Nn?Y)-<!$y}l7)8u!Da45l
zb^7G#=3QT3YRGWW%<`TP0;e-Bbe;OjupmMI1Op-5Z`dVQh`I~3;^4<e2uCgFHnPz>
zA95z8MIsqiHF@OWwDUu=FW-m*K|-%Xt#AdZmDwj~Ajm8>8-t@_{Q-Lo4i`B<H9chF
zM77YUzWlWr5b6Z+OwSz$?rN+T3!&r9)zVj=#kX(@`Ei8UlLc!6uLkW7IVhJ@Dbmq;
zzFz2e28I5?6+{Dux2;t*%al}S(Z7Y+I<C#coeis*LQJq#EvTRs-C-z)z!s14I>ziF
ze8lw9e1Zu#kl{|EwL%{3gZi*1%JyP|llI|;M3?|t(Qc)(q^xM#!p!M!0;oO=T4p<Q
zU!bW=k3u>=Gs|D|BIdN<2($oHFP+?2th)NB@5x<5fuK&#MR`B&<LIPQoDl&zaO1^r
zCp-z|aIAb{9F5%F=X5jJr+6q~FPyjhg>D8+-aX42tlfL<Y$J5V+VN`7QYYch-&q4b
z62$bIrsCIPWSUE={c}K2aOxQkDwi<>$XRUd@f$I6oW~5M6h0L-Xsxv*=3K0es(!P7
zdf;a%@#)SCW;Jdp+C>bdG}zq}MB>dUcylY@)<7)z$9SG8xs7k(#Jk<T9G8$fT14E!
zyE_;=>-izvo0TiHR%}O~mVF%41}UVguO}_FMMF**wHyt%UWDEWizr?Y0<?dp@fkc?
zclL)AZq(i}=3748F!-x%8?l~>0;X%EE$d*7dEo-8j|HcGt8S6EX?j;%N7Y2xt4PXj
zT``{|06>WW&dLtW_50LFdhK5n31$zj)NyHgjnYj>;n{aE<%;#C&Y+bQhKYy+XLoc(
z0%-p{0PbAQ;JD)WS7#=LAJ4a*fA1nC%)@4vY6%*fkc!8LAQhd5mA8l0FnzdtXh1)^
zKal!q^fU@WxL0&bTaT2zi&=c^duT%J7LBhy?z9+1BZn(a#8+~comypv#(&&Vq#<u)
z9?tO+L<P_+iR@2mDg?ypJd(I+$kRJxP^IF;O*=b2VkQ<F`qWMH_1M!oUqJ(m$@FVc
zp`3@ov*m%iENLS=i1FM<;Cxh0HltpdB>rFg_fMEtgNK+Q#zx}R0ki@zj$Niy?q0fk
zhVq#+qOK!ewvM_nPWXV#n$m%4O;r8@1&NO@<0mIOK=n5jM5R*XNIt%PAdNF({??=@
z@H|txDwVAxNj^?R<DTvB1F+E^t*YOO=u8J6J`4nj3$(MocTIXxhEAz){#y*%Q<Kme
zzg~0<jI|0zTawSS^L2>KE_W@-)Q;>{+AYZtsY#TG)Jc~`U(e?5&)9j+W$(b$DEUa!
zb9SF@M*o_qt<bQ4wtPaZN;EEyTqSo9MLe2;rl+JlvgD+L{9{2%D2%jW(}IL733oVy
z>DPYz9+kw1RG@LnQ8yIU9d<2*0ECXy{WFiC(Qnhu_Enx<Z)VK{<M<tl&ittY`1WEa
zhX02mo81#J>L#R;kNf>=Oz2vZu;)w;Zmfj0fyc8;U#<Ad?#di?+jBM#FVFrOh~GU9
zS|M{a!^PMjfTZX;rjg|>LFggnKYcl%>$!w83sjzdK~lZdonC1BHrJT3i##|-`Ov>d
zy`I9^aCEjNs(awCkF~f(KRP&rLnU~IR0E&8k_)hApr_vr3wATj|EUUU+V%)sP`-Q6
zf$h{pxtXATR5%(X$Doh&7AoZfVu>CzO65@zK)e;vQCsUpv&-euUn=NYk~==rr-pC$
z=(eaPD$xN!z5!!@{_IffB9%>&IrUKgg1y1Lepu_nF{gpJk7H<z@q|v1#zN)83@EjW
zP&Av3sR^088xlxHHlf31W1-otLmOzgSgbMwRqt{<Hq5R1m%3sc=E~bcWF}7}A2*GI
z?UpK;p*HC@`@N$I`o3kzls8ZrrQr@Iame7l?YcsB{OZ)o?j{*0Ng9%Ce1CWsVM-6R
zN%I9rsgsc!kCN@kKq;gRw=9?c_Wg^-ZhZGcie5^7*+jDSjVLHH8pvGaS7?Hn-HM8=
zE<OA4h9x&SZ8g}B7?7xYD48<o*YMRg(XqcKDwN{pT=Zf7jsTsclFIL87F4z+yVBVy
zWB7xbF%xxnzPkSls<V-pztUOd5U6T((mJTod68*MP3j**!xi%PXW~8+gZ!Zxo+z0a
zc(2VqAiwrs`VQKjt3gOLDU%=Wtw+&&IN3G7EZA%<kdhCO*l!aXWlPrf++c%y09<{q
zyot&b6V-}A(isFCZ8Gh&ab`esiK)Xz6CDf%X%=_-Uwi}4Q~JL>eQe4fH*-eQN%Rfz
zyXJ~Kmj>VOHat)Uf0?Uy{h#!iym`Ic)Usg2wF{6T5=}6WBrQl+k$f*M>Kvk&f75#n
zUH>?EDV_f-;%K_A$m0hwB6}^gN;^Qg(Cr03_aH6vIF~uFi#~MlYnI`$GVPAI1f)c>
z%Ez;o@L%D8PTE-|ZT<alF1|eJ9RSd#<Za$;;;1nURTC|C!-ABkmW|IoInz2%91g}p
z>>QWWEb82E*h@E%gt@5@OOB(<o#sxEPt3!r7L<JsFO2DhgRcHfSx8{)V<k881l;`(
za`-!SMoW~#pULcw9?eV6kVlddsc2`_J`*uJA8)_H|0i*jRnF33!s{*h1Vw<xk+e*k
zxeBGIeDAR^{-81LkG?-YpUPrGjMr&mvdQoIgsRFf#FOH_rRxm&OL0PsIWC56mhZpk
z8FbmXp8~oej+df7ra`BpTk3faAOVr24&KJNQYzJl*OD4XK!~OMd6Uc$v`FOvwGG=+
z581OEXA47%sE%fcQ1rf_x+uL;C{u*0mp1v$5m@flj^z31z3E{V&e?qN3rsM*EQif1
z`L4k2&p83}`-Y`@1&&ULS{qN*aTE|#^>j0&Okht$rhvW!qSjbN_>N?8H8>ZQb||+}
zscPV~PVh3{{;;fEs+>9!S-z1TOkQAV1d~En>&jgokTuku$}9Fzj<v(F@^=7>;GCJp
zpwot}9N3V-lUy}1fk_=j=fm1TNSz@;*j%Wzx1Co+3Ng-=iEwo%!x11aU1d_t#zh!%
z3$ZNHZ*m&EQ*k(v1E=~^@-aUEnbe8`6UbOQ$_>Ed1al*;jQ&Z7a?)YQ@;WJ|4NtKF
zaqsbCa)@zinT`6lZF#r0!Al%y#V^S!R`IjNk5OnQ0mL{yI~TTssn8L*6Qt~e4Fm~`
zynF0zEgjQDJ1ByMFkB@Q5L)8fDaJ%glL4v;kNn}m&DuGeUU`Rd_m~Jn#80s^&!j|{
zN!!*X7Q{Gwx3)=j@K4~YrXPmrzYE!J6XyPc_N}p!zyx%cvYQ%RIB6YwNO#_<p2*tq
zgbL3!AQlw2&0NF=&~6Q?-3U7DSFiORNB7|+SyOul*I}(pEvRkgZ)kR7jIO|+P9IWl
z7u!+*q-4f2NyI{)rL)9o#59UkvLC9g$!rFk3dUaa`W1nkDnC=1SYTF0`v^k|8;r_Z
z98Gzy7a1)>;gQG`wfqGMH@ns9;GMjp7XWSEFK44~b@#$Amqt}MStV1eGUIt<OK)u2
zwmxG+^|v#c!TO%VkJ-?gu@Sr+{sESfGgEuS|G_f-jfP{_(f%2I?{PL-NQp+-AKx}x
zrTDOm1K0HkWD0!%lES`HKV1pUW`UG^LK>YXb7GF{Fe_jJw8?ke@94BJ)u5=5XXR+s
zb$oXYEB?J5fHuDx101M6M<G?NM$J*BD_LJC9-{1v{R*de)VsstKvJ~FoFAdh1B>J|
zr&Gw7GZs0p8nH~w-&p6xDwtG^H8775?kt(?T9lKI&L;w39v0apQB3}Z6rfEB#i|*G
zr1~`%{q6+!Ky*?{MW0pjYUx9SUcS+~RzY9_mmGMalQN#(yoZ;mE&ot13kRW^E<r}J
z$GrSra?$H02{c)gfwul~bhP*w9bRI9cGC>VqoSoaZtDEs&${H$$7uC66)lQrg9)kE
zLTJo|GB*wZr^#Bdq}&iglbw<{oY{_4<p|Nrh@i>4MlCM`=2kJP0`^sxsQ}vLoUrWW
z&G<&cW&924NTjVP7md9Ki}}6+>JgnRn4*(v`lqcc(Ms8IQEk2)n8Jl*B(Sx-Iy39Q
z_c05ETRZQUi}coY@|gE{H~|5|a_Zq;@4|W@s`?@zU#xj%zWT$V+;g@^_ws`ebu-a_
zZj@C7p9Ot!A6=F~J=ZH@D$kY|ll1#N66q4xMI8FxL~8jtwC{b&BKr;XZBla=S}8dK
z$?w!{zvyaAfM~TV(i8A(b1exZ{RD{dyRE!v-e&vspclHtcDg&4;j6Nphy4_a@Vnz)
zp$|YHhLOmdAZI@umnwsiQGs>#km~#UXhF!6P$up`@p@>Sr5$O@K1F@>q4@L2qcv5J
zG3S}X+pbNCojhy_fOa$#V@p4>#MlF^6d$0C*KM5L_tApg_>UAmDKHNw@J+~M+@OTZ
zw?I-!R!I+u!rP;&_|jDam=u5+m?t*0SKrYH4YNcqrUhuv1TlHMfeBj1U(1srVn^tu
zSAON+(n_8$jA@~_T1sfI&k-zS%nU8e4bax`5$HRX9!Wsy?WHS@U?XBpI#?wv7%n6p
zs23wwa1jLlqgtjSdh34?bZCr+hy@Rwj*pP?X!Cvz1m3brlAADwkOk#P8*7S|x>7*6
z-*$hA@q1^FpNQ{GR)_<r?$`_QGVUU5%gNG4u?~kJht-)|EF@+6M9uG(3`ns7)eYew
zN<C~NPG94jbm#~!BGt$C*Uqo<{H=4I?}5M#Sum%Ck2{d$Y`47RFmRCv3z2%7)y#Ki
zu~6kxhJFPfpgPAppRf0NG9;&an_1BW2;wbLh~5hLMASSzbBIRGLE|4r1h{@reX}=4
z8=xlUdPKjr@ugHkO{hD$0v~Z)v6+)7+a9xjActaAlm#z<yhSPe6Zm$1w#uP6{t?)X
zQ7L1I%r=z?M}&z?kU`b9Dc0Zb1t{W8)leKiWx+Y$x&4Bre(O&;5-dgnaS+G6j_|?b
zCIfe|@I$!|=$6lUf4H3<-c=&EE!4)%MhIQ==;atfd@*pBS46S?lmkB^y{@lDiKzK`
z$|#^U%7ORCG9a@{3et@Gph+nL0`%!Mc)Os^T^B-lHm8i@unj}LIe4*>L0dF4YXslP
zgHWLxt;k-i_n?t|x+**wb`&QdQ8O(>XMWb<R*WR5HYEtioa|k&(jUAClUjUvhwqEt
zUBr&;xtoMi#TC)>PTTs98)3-N_Fgx)#<JmXfm#b~6wi7%a`crwe8aYCi6sWF&0+BF
zii|bLOE9O%pV^R7L&AG1?Z)SvHE8lMYR4oo&`KQnmd6r%NrMz0u>1{+^5H6m&ll8m
zM0Xc)Xz~jek&P^m@e9h@DUNlyF(WAoq^Xkw_cyhhs5EsSu<@OG&Z?mJzC|KGE)Y2!
z*m;c#U&pHAUx>HGplfUTk!H7umJwVkU~}(GaJN#oozeyeu~GJDvZ)9`Qoa4M@9ttO
z>liNrEO$AuL@c!PXS1&4TXS$GwuEwhzH~JZlI|lim9uyfmi}BB`Bg!s-g1XI<_WMX
z+;8OIhr*mK!Q~5!1T*m(<+MpDpdIV2%+kSrxlNYs)8icm+UEYO#kRd=Z*4_3z=N)x
z>(^Mho3VmCIm*?wG#^jbHLvi)FFd#iCqo+U`5pQ!RPEukv7mso?JIa&Bg&&KrjTy3
z4{mO;2p?9o^s}8$LDpQYetHd;%N_N3<%F^iWWtN-uDAEi=&Y)>8{rM$wZNLRsF%9I
zX}!dM(?ov^IfNO6`6KdWX1ar{Gb+mR@<0;WBRK6VtZ=IY%EctGxs`dKlhe4*w6}IC
z#T-l5rvDC036U4ACd-t<Pm#u{7ZY}~JKlfQ8}@exP-~D_Xk<IVt=Iy1<h2cxRy3EZ
zVkOnT;X^aQXN}8t6>SzNL+`?t^~>X=HO@gTn%wfv71ZMr$Ix6!Jw1b<AD3@nwS}vX
z2gz7u%snqJ3)W&1m|g{J|0+<xCu5ombzi=R1_u)Q9W1s=$b{Zp@S8Q=W~H}>9PAoC
z4FR1@cJzB~Y7b>RcUgz@_T8Kv&{Hx$;#g@HGn;3e-Og+loO<zRdeWv*Sj7AJ_Ao<b
z{<}=R!@TX>-G<UW!giK}5{WZPPZdT|`^Kgfeu2Jg-AOD+U;22%Ra?LDU;U@H)eb<=
zoJ>b@cwQGs%IF~UIsh*K@7l#Fbo$=pra94gr~d4G?(e;$@k0gwNSI-06lMj}$Ok*9
zpp%W!P5%3X2f+zZXQ#J6;9c?YRYeJA>SKPlX;@4kMluBzMn*sVSUL0TFKzE#-Z6a3
z;F`a1C-vS62x!4?IrGVIL?xfoj93`qHtjKOeWXa*Puo7Lia&gQOd-?-4gH@p92o2E
zZl=k|fERQza@M*jCOwYKY?ZwCfcG_!`(wayk(5OHRjt2k6&7u1^B4#gA?6XG-NBQ#
zJT3G`^VbE^Hf|wO%;M+bolT^5WZ|tu95$GpOxsvAp`>SC^ySxf4y_4^n&uYTl6I93
z?H~5YS0@JjJJ!qxol8=4|9Zc%ssFadw`Hq~;yKnye;IS8rN4+DgGW|7<|tfPoExiE
z(&H~mN7G4)o-&ciO#cwM3sSXu2q2Xs(V9FjJ_^WYIpA*R#>3@oE1nUkn|)imXkB90
z)&H&{|KObj39Yg+VYI|(?cJ<noKlxvct2^JdA5?zAU-~4Ysrkg*;doITlYv|Pf+&8
zlX>D2a%kxx^h-i!lM|nB)`76V%XynT`_6_ns%Gh=&m1U&oBg@-W>c5i(MF5u7b8|b
zytJDV+uQNSm8==<8AetcV4{xGy*9swC*!;HUf>{*&&>87#Cq<7hEA9FH1luzb#2rr
z+<eEjQ&wO8<d^45@Mc_1S^bP?>9HffQQ7h*6?)X|r@DfJ8+h(Zzb4jqjK77u&oX=Z
z$X-Ijn_+uSlaJ7kY5OJY#iDqh@Q74$?qSNRWD3(7Vr!3lh<QbD&ZiKrxP)_$S}4dn
zW-ee367|y-sQMs)I7iZLi~W5v?NENNXk>ZU({yn_d~U)gg{{7YT({#yy*#<YeXSrC
z6dSAHy-a7nHd$xv{QG`YfcCB_>iZsN^WL=dRQO=?k#pXRD~^-r9(~)S(*q}SkxUOq
zd1QXJQ<#t68s@dl{I?w~Ct)Q&ET=c~mlp*W=6S&x{z3BvE(1+j-;xW3v_vygZD&TI
z^RQSq-uH})uL|r|n`0khA4GN~RG&(Z!2V49{AJL60m4vUmW3lF){Znx&FJQQC~#3s
zxsARL8zn6=igU4-o!7fiYOk~-x7%vlNMvbCTv2U&Y-(9?W>jlyRAzB(V{E(!{$q*9
zW^1Hef4>{OOWQo)|II)8L(@gWS#w|GRmM!_T-Izub5l)2LtRrtV?+J618k@+IW;oA
zs8U}=1)tl#sUo`se|#i{mWS{(4zM#@>6a)Mtdf|MA@bA<rtK8MKjXJM6}d;v&}~AZ
zRC*)bWd;8F?*BCX)ujzdOt!>+c&}OM_V$je(J`Z_PoW$-LXq4|^+WsW8a+JTQvgR-
zks;{Yq#^d({}GbumJ&OP-v$rExnD?5yRBpGCAUh)BEq@C|F;w7FEk-?i);7XGe0<n
zFFsN=Il$Oc%KOWhm~^G2?h~a|5fvA~fz$<#$^07@?~>}RZ3dc$hnI94#TttqJC+!4
z7HR8UUTRrR=e7O8EMRB!G;&^FY^LHjPn>{}_fI}Og>fIK2hY#xSAS66zJl9U@~o*n
zg&T!5^O)w$!N{j4pB}@wg>Q4Z8ZCWIb<W=YBp4bUw)J!T`yK49cz&i9bQT+9X=T^A
zvsRWqICUR-_|)I5=!+MrcQdZxJgbh#HEn4&C^k{LEdTh%9QdS(%SnlTFp>QX9;)A{
z&s1FWxV*%8wMg6X(y<788m2#FsykqG_f*&>GD}8FJS8EaH1EIF2<~0M6Z8cE<iEQB
zu%l<+Ra0C9ntIe^R>=+H^d(*u1_a!)1YZs@{l6I1KlG}-D7QNL7Dz}Pfn<4owO9Xm
zn^``U^;U&ebCmgyROA!6OYNYdojloMG%bed)7a#HG_ohJm@WEC0s)5TWg6PDdWq-*
hvq1j;;tIiaXotk<b&%2-FWPquRV6LOD*3k&{|nMRX~+No

diff --git a/docs/dev/reference/figures/op-softsign-plot-1.png b/docs/dev/reference/figures/op-softsign-plot-1.png
index eac6f369030bbd4ecbf9bf44e9154b895c28c725..2822105546a2fc30dc7cff7358060152923b2727 100644
GIT binary patch
literal 20855
zcmeIahd<W;|2BLoNkbH6k5Vc!LiUQNBr2rLvRC%rQ7JQoWUELSA$uiR$x4Z2WoJkB
zxQ_GlxvuMXU-x}~ANTJMxX*`-&hvba*Xucs=kYwA=c6mi=P7nG>?RNh6c;a?yG9_8
zEEE5cZ^QqbR5U$-AG>TW+_WbU_8cJoBl*(fwN4<g5iXuPtL73n*<+-wW>!r$D^Rny
z=P=#dhj)GMJq&nw%kM~vK2^YMrw&(EiN$xGfE~N{+(<e?_Uhro1Dv-r4jc(~P#pcb
zc6W46$LP}cf%i5W(yqo~_D!dBD`VGwj!Mx;C8`hzKC>y|%(wv3J@S+UwxKHv1cL0o
z|Nqzj-a3RRJNfXcs;mE9o-@c*B@jFe)HuJ@);_4ZMgA*x>1~Sl!dTl>cVSagQ^US(
zgjsH#BAc^%dU~p=s#JYt{s$$`Ql+M*UW(>9CccC4zVOMDCr68{Lqy#Ogv0(RI|)>F
zC;sbK9lO&7nJ1=mTZY#jS?}Rw{Bhz`jFjesLr#~pP4-2kI@FAl5ZumG(Mk$8c6a=-
ziOruVlfV93F?_G?r8{PejXQUy?l`wr!xVcWFbhXMahF!|_{xW_ACXJHe)?~2syQ`h
zMZHmQ>&s3olF@A4MIg8d-)(7Ar8}~)IAYvv!aLuUd51JU<Ps@?;L9<)U-Gj1oI<8;
z$MY|fQ=c8b9Xv+A-cj;pQX{Nt={GGYYc$OQ@p2@qaELW-a*vnxqkr-rzIZqq8EnaU
z>AL!*z6RdgS!Q~v?_2f}S|=PEUNgSV$Z9gziB-`ZwC|c?N+9R3hD`=#lv@>z`zwTc
z3_V-1x&J%yRH^Io>}?yHSrgvW7cchG^PlS<ClKzD-;`Hrx7qbDA@jxZXdqW`P>`XK
zk+l0<w4RrlsVSAEAb~I-`(rvX<+T{Q+Sczqcowgba=&`>X7RgsYCbtQ#=g{g62d*E
z02YEw{VuQH59UWk)AJ<0XfDTZ2G)1VJ^Fqq{T3x||Lg&ewswE(ilqxR4VH|9G9R16
zDy3QFwTF3mYPS*I3%4DkHIV;0%9HSWf3J0Gf7LeES#APhy_iSSu{!Q|yhuzy)K0>f
zn@)aNxU4~I$(R~}FzbH9ar^)2d>6X6kq~4=Y($uNqGMxYyT5$7cH_p4D_5>)X~jlF
zRO{y~{QVn!>gJInM`W#JX(S2n6MxOGWM*d8)YKRm8R7d(UuAYi#^>hdtT%7CPM$nW
zo|2n;O-U)EqC!_z*0*ukPJ~Ha{hP0^Z$rfK=%}dQ6B8|QqVvu1;(`n@jaOm*t#52a
z;!;vlo;`aO7Z;b2k&&C5J2^S|`t|GIzke4L6m)c4X9+!@*1qB)bze3tENo?Vz%M6f
zE>_qYFXz!Id1pJQU~N6~?TtoXxx4dFec1i`_x1Jl$;il5Q<WMT8&%cRn9i7g_1j6A
zdO23%m;1T4?LM<_aQO7}iz+HDpFf{EdGg}r%Nw%;ayM_%$17*2r-wa$d@HX@f+gF!
zuL4Ko;NXy#mxl*g_m;J_w{vlE8RXp1=80|`{$eV4=1l&lPoL-ojL)4vpRHZYf8vDc
z+Tv6~LW29!^we_ok9-T<FZH2ALMF{|b)l>i<KspqCJ}3kR~HCuM_nXzy}Z1xU%&3;
zv{>pq^XH$3rk0k4g$2&2rKOdw_WIc~O)IOc%*;Fb`YENQrT+ePQQTTs&)Q%{>=+j(
zr}?d0k7HtrZ3aHn*I#gVUwixZZMtfjl*e*wef^2!$IokOKC8Ml>mqUT<VnuRPq+9^
z&?we9^KfxdQ&YRoj|7K@UsF?i5)hzhZvNZ3&y$vxR@i<-$xfOcyS>1NMR@(1^URqu
z$BrEn7Z(>5Eyh+A6hu`IP?4TZzH;u|xf`br#Vc#5s&eu23JVG<UA}xqNQj4rCn_q6
zh2QXkvGK_7Z*Tm3eVt}{J%4{qXZ2pU9c_+hVPTo*$P{<{{q?o_o6ny=TlJK@Oi%CX
z=-^;yZ|~@+y~T$OIwvo`XZP+g2e)h24g}Vec&<71Rd{-OcmxFphq6eTTUe|$@>YI+
zcBT=J4<C;3@>>7;_z0IeSQmPHiCV60yUbF}u${e~9hNmbB7*wBf!%wE5Bj|da&p`K
zs;*qS#>vBzAaqAlLE%Asd=E@yF9pT?pFbZzex%#0U~exF(eNZXIy*D-TU*;nUS3vF
zJ44uXdb*jp`RnlT1F=oUe0178g(W4LYHEU_qBqsm&qzpUYiX^mugpI`A1dYg=S5`X
zw-^BvCMKr!jg7R+u})i?o)i?p>+2q5er=~cw>CH2ec7ge7PIs4cujP0A3PX|;0#Z>
zq^S7p>C+%4Q9Im{p?3i3{g#<BZ=0q3+tm-geEGtqk@c9Flu#iFr+xZ#dG+ta*RNk`
zd3C?SS*~AKR8%}bdR0$P+UV1h+1c4duhsFfw&w`{!2tmbe7AgFy)so+ZoapR+Skud
zBUAG)R<WqGlvT>Z<=boZ-rnB%(H6>GyH1}u<GHzB{Q9-Ik<qb8#h2I8ErnuRMgs!_
zqa!2f>FBUXSo?`zHGx7xLWd6@e(>M{{3GnilOY?C%a<?XDv^$M@7~Q!y@2&+U|_&y
zZLI!97E`}^^(wZMn)>;b&2PqKR#sNi)6;POzOlS@QiTPA%)Wj8^RSwG_wM1I%3SB+
zda23DcWd->Ze+l&at!LCA|tsq-;MQ_yTi{ZC@2t%8l$*N;RA>)DhU$B)_qdW&P%X_
zY`scdJ-w|z!;Qaw{Zit{!)7Tea!N|-V6();+**^)BBtQ@r%#{89aL^C#ugR1T)s>Z
z+caWV>>;%^-BWt%5ZBS8FH{nxL`6mIM?PP@a>d5ZZgskJ#?a7^jFLs^dD~MZj;_Rx
zO5c>7K0NTIg@y6<&d$!Ep`rEl^$?4vYMi*`Fxlk241%w+vO*dR?TWR<;&rvPGoC-E
zW@WWDHjWAp*S~#xuD@!#nVA_&==aXf3u*1D*RJ6VR=s8EnVC}q1E<ZtyoeWf&bpDK
z4?nmRbxPEEI_~b>vznR=epR>l=!C6%pM-_E{{5|vr@wvu`tIGkuiw7GAJ+a(q@|=B
z;p9|YG-1E-w1k9kcdY29AnsU5XmPwf4g2)>?_V`e$JzcWdV2bhkPsXyPSETNLeiMS
zYSLN1@bK_dl|+_9hdONi!(P5Lvb4+y2~ma@#Kmdr=rBhr4%-#)C3~Kgr7S1M&dq&U
zOH2Il;j>73FJ9bh8FgOgOnUL+C=U;!SSI{BElt>A><Sjs(a|wiI<)d`IwN6>o`I2(
zXo54n<v2SM=S4+D{+Q1&oii34F0QVgTN_Ks$;q%FMnv;GlNPV#ft`rolJ1L^xZl6O
zzvblQ;Ms{HYIRxJUqC}hANa5K)XSxwYm2i3)rr#H&4&D1nwp)KPIW&PVGg^fIj|)c
zG&D3+R8(M;2m_NXN#e3?+uv7S)6{fBh(ek<L__l=DoXLf1yc(P6GKBL371*8)1gC$
zVC6W0R-sk7%N#=U8mw{jbBz5=ucTOfvwn`(+LSOicP6|ErbBdfMj^}2e2eJ(d`B0T
zqI>s9BNd5n%+0kkGJ1*d^Z2mb!otD>O4i<-p!mtcUW24DlK#fom38E~*49==VJn)0
z2OX6%80qQr-@QAf`R)V{Pe|jiCN9+7y}Pp$38>}iX)J%H!C6CXZM^FzHWxd)G}C8o
zWaK#gQ&di_`USS+9_jx5VH;~p7G`Go1qBE{?eE{q$jQm+=xpfmn38<P{Yzg{S3k(a
z<bf4MN={AXz&q}4-&s&t7#I-n$3?=IZP}yZ-8%;ZgXf44uCB$tzOuJ(C&PDCRjFYQ
z^D7c8p$&!|9UWfFv&q=^!IU=v0iXJ-{IF_u!Hj8HS-lpcx_sv;2`_e0Qc~{T?e=Hr
zBEoV+!(<WV32yE?^^XtZZIO|YgSElf!GgRzdl{McFD^z{^;KMc-iGAR{^g5t)3d)5
z*_90R^fh&L&OhHOCSQ3#!DxKvj-=PR3lbc1UT&XWwhtV$<=~-1y+yWzpCY-K3q7x$
zc!Ic$9s3UCl9ZHGQsPz<aBy>T6B{n&w(trdq^!(CQSsBqN7NhZ>jjowxp%sSBqhs#
z{P+=ZTm@E~+cgPqd=?vvReTf}7_l+p#*UP^uA!@|iwlJTVx{0dhy{6hd4<+}EcEm@
zuU*Tv9jtx-{=MTw2eY_SMtQl;*|P(<qzI17VmAFg&!0aJ4Gqo6I2GO4lV_q*m$`)}
z7$n^U18a_~-608JN&3!p>J%^;#ooP$vvziNuCA`QYCsNyye?u0v>Oh<K@t=7939Wx
zm!13jml^h2Y(I)+SwvbIZAqMWDSQ0*@k!NGgs7)6F@LAJqbG$)6Ok_$t|64-71G|D
z9!^dLPoH)IWP~z_!d1ebJ^PIi<h8zBR3yR@T6c>N$pe@mQNs1ZfDK&jOM83wk01EX
z&CNYEHFaF+sn2b9VU!nRIz`&kFgB!YO>J$Z<SU$9T%MkuOVd3tWHWtzJNV14UAxT8
z#+R2}I3gNu@dX41A|wG=_Sv?lhlDhDWNMkTBwS8=F6p*+J$u_2H;xd>EPjZJitXe{
zQT>k(0T*#+$B!Lzb#p_O!RAzWu3f)$DX{7mveaZ}jsa|ld2O!3b>4V>q^Y^N8Lu*|
z+_lVr!e?L!K?8ds;rizsT&XQZ9{$36>J*cR%`Id;|LPbK+h5&<)&LGjmicaf&GDO1
zY&mxDVg!e%xVTE3$bJC<N#+0+8B;)6<mNarM^1KjIa%2c4GkxyrKLGI6y)WNa2+iP
zk_hT>t#V{4@6Gu{>f_3=Ecvr%k=$0+)?#B~9zA^ccw=9Gw@ll1^5h2($N+8k(DJBh
zXaEv~ut>VS_aRwXSpmSsEnT{J(e=;JqeqV#h8v^TU}zm3PVn+emm1aITn}azkD8pE
zok|`fxu<yN&eS97<5@bT9i5%8Qd5Dij6Xj`68?fev$OSX+;ExuRWnfSUznedqa$Yk
z;v`6Ux&g7_RE8!y&z{BgJgTj&ZBJ9tD{)9qPCl!lvD8!Qytuf?OsRd0JjGy7LsZlu
zSJ#z~AJ6Uc&jX^1<kHx^At3BSycH6mSkLh<#}Kf8|29`uZO2{%sxB`tr>P`f*U*4X
z?q^qkPoF$-;$1<(r^d#u)ro9lV`Ia#2U&hFqOJo64!{mH)YJ@(jSn$0PW1P`zr~k@
z<gTOBG+;yYPlxd@aGYJc9>KJUi;9#xS>O5%C5EKBqvKxy6&Q9_c6MT7VkNF_b#+z1
zq}itUXFxyzR#+)c#1`<!+}u1CfbgB^{w~+Mf-dLw$h1@s+c{7Dc$t~0qo#H)`ATz~
zsDNo}5`16r;>Cjp4{qQ>@dH8Oy1Kfg^R%{|9lvHpG|`>-kRAXkG_th!)6#|p2BIF<
zDzJ>n%CbU9lq}~DPz1a@JUrYIFAgiTU}R*>o!_=XykZ3li;2alZjFo^RfBWaR8>(9
zFf%co;^szaAZ*#$+S@B-_T?x;uoH6j(2${;gvULs5<Mp;=V!Q1Z~0m^rL+f1*zSTm
zfM>v5%F4=u3{ADQsfcvLpCXax_wC!~QqQ^0f}pZf1R)RE_0_AfwWS$Ub@6d=$oV5J
zi7ILBqfN1J*3!AbI%Ge~&bK77in1mr+<dI2s{JXka)?Hej~|oo-i>JV191)u?JAKF
z5uxhlR^sdX0javItjul7Ab-FbwpfylV1XUN#Z}eSorfu)eCp{bFD`cNDRC5J2(Hne
z>dHGVBy=4W=;VvmOyYaeiQtv)v~P|6vE#@6@86G?@+_&S=&i1<uBy7rMlO}e%D_<K
zv0|5NAWP{pF5xphHrbVjYKxWMus(!Yyhu9>u=W1^nlSeB5)u-?6a_E<<0b}%U<D_o
zF?uX`PjmROODGQz{qUF>8w04e5Ed>3tZW;1pr?;P`VC?@6EGRn8gc74SrTJ<cDA#v
z?W3TejLb}6<BEz3tUa7#bj}4fWKbIfAa+obrGf^_TiS1j2;7WS>I9?$ybooS-b5+{
zbl6X7JN)S&PShC9Yj9V>V~0`?!C?<f0;m|(XXdL{_4V~Gv;B9G7Eqrf%j)R#Aafxp
z*Ef?(9t2z-8k}>%UXF~6OnlE)QBr~d#7lcu!0=G2AtCGQAJJslqzD=#xo}-ar`%)Z
z4`2omFZKa33=4`90<lT%>eV?^t4Mr?CifC3*xd->#YII#x2W>nfxy+?*4Eb10e1(A
z8=iA1wCa&<tgb|WX9HG7B6XIPl?7qa6f10DZf^Cn2*z~d`gI}OUmsBk6BA_o{(DGZ
zK7z=wBXcgQY04Col)?f60)m3t1(rAA>%b4u(X@v;-xmT%vz<ph3MVra1i;Y*z^JVB
z28Oz<q(nrbG&CRebL#hzf7niIKytwb_GwTTa*~S+0aX6{d6j2^W*;B!A0Hogp6R6^
zCub99CTAqxzs*={av&{l?u`t1UvH(i-u3GpO-*<3S`;uac*C^qxBd1@%0y!eH7h(;
z5cZ>a^?Exxh=>BA7N&ZqyFjPHBWBWpEshT!cMYYBlao_PNy**2cS%S{a1dY~EIeGl
z&vS{nYg|TzXu$x+)o>RC@e!<$`{Lw3C14J=7zaE}w>P)_ZYyx7Ll`Osc!y1~Hfq6F
zuU;W49zAv}GAhao_!rK06t^NkBPm1Mm1E!!n<y{$1OmXi!}QJvGkmM`_C|WjeSY^b
z0+7Rf>?R^DD=RCk21f?=fN3#Hd33h4gie~Wt;6!SMMp#kSah7)vEx2ij51(lT&JzA
zE#f0BkM@<TR}bTq(lnAj@0rCM>btwU+uQ8{xUqkrOsY|)q10&o(|QJn^8MD^oACHC
z6+L}dZ?D(L=NK4d^phtK0|FAn90MjzTebQhoFtIE&dx5($q9>!O1*NTs;UY_*4K1Z
z<Ylc257EU_emvL#9|BT2($yj?@_V+PFh4(10t`pD(rW{hOG5}*IXUvJ50WWLc7!+f
zBS$(d3$S1%C95scTS`tmV>*0}+u2UQF#*3FZEV=c^$`rWR$9Ds^YZ2vG{)TMkeALW
z5g(rsq+BF=L9BRfu8g8!J}D{L^Z9cSE_nhdob(r}N)!W(e7D|bY8CdDxmMre;};M>
zhJU5D{@gE%V>VG#_x2O()_Df0vsC9SdV9)4q7;5q2OSris)+n{YRrE7<54UtY78J!
zSdsMR^5B=3mzq9*err$vShnAO;pb%RPUU@kg~CyHLc67!&Uk&9mNaoZ#yMDCpmOum
z<uVbi{g*z=EN5uuzkc^_p|H=(eQl9DqCo<IJ2bQf;ibNLN4eT}ouJ{P{_^!%l%f@u
z<iTtYPuG4HWZv@sn@4-W`Q|nsp8YH=AZNUgAprS#d9wlSP)@UmJ1NP@`QdgiIrcal
zmRHbK)>c~Z-tNK~adFUgMaRj_CoBBPldn1Nj0Zm4hjG9w3UYIwiP#?LDo$D=GL@!f
z-~H1<`0{!WwLe|YbH8vhv_Dc>Jl{jL^VY=NwQbUpG8z$2o)A+K9(A{ooWUy<hb2W*
zFMp*lh%-_C7V_aI@xe4-cIYI}S({!dmydSkzh}PCPEg9!ySTX_pp6U-NqDY0plF8|
zFD@*A8{ohNXiNl%x;=ZEQWAo%N5fmXIm_d<E<CYKgZpoolgYw?hEWM2ej&4+4`C{H
zU$O#7zjEa<s^YG;Bn|e7<{MV4$!+Yc#l}+F$*ad3tyrWy=s?<Z1<uXQEpyYAWFA*Z
z_#7kP&qfY#1Qdg!lu62?NUze%aMZUqBqJrwtcdhn%k|a3nn(o&1^8WJd1q_ud*i$v
zj1>JA4Gj%2QcsDD*RR1uv7e@t+4*I$WWV}1`nhO1`<pj!va>&L(N(*CorjzI;H^_W
z+24T8<KiZVhe7!d0%kRS`ZV2{Q{CPkF{w?^$t%mvJkdN;H@v7?7tGJkM|b*%{)2gr
zP%SgZ8(O<${i+8(CrE060TL1_0CV{H>1h=8$n2~_mULZhgpyn5TAKb8<V*ZCo)WVT
zi!2F0>i7Mc*0YU2Rd>xu+PqHJaeVsKe1mEma5?xDq^@J+M+5~yadpA}KYqN2Yangg
zA>VSb!luA!xLjl@xl%eZKJU%&>#*LXiQ~lgl3}fpVOS~82nqsW)dy~$-E~)`=gR+`
z>92`oV}c8_qDq<LlaTNLfqQg>WLffV=$P4$Kw<OXJeEjfYm=-uStlC$3p*l>D+Bgc
zXjHN-7rjVQp=5JGtuZh#(ACui0ts0Ve}eI;upK1pDkf}Rnw2v*uiSAr`^eG_#puc)
z@(MZ|soCpF3bCmXda?^D&ySIMIT5Pt9UN%(?|&2!Ackxe%m^A&@I!TV&Bu@MsaZXp
z+_G@aHx_xC-&47iceK{<^YeeKseuP5Q{BtY|BE0fjC5z&#h+_WvM#L@zI~kwXbqRc
zN{P_u<=sLf<BS4yS;N86(ZIl9Wqqsn(4V0O>N|Jt*s+s%Ja{K=KynWh%NImJlcr~|
ztPFT1c59ITl!GvWlt(!jQX~byW0b?l?YiJsz~-Q8z(yp#f5_6$HMimc5P|Z}S%ST_
zrDX;M0l0I+(QV62fCzdgj~_QdLCu;-T58x34$K|SxT~KMiK*hiKjaN7rL=D7jxj3)
zaXnPw?83rC>SAyf)h{)N5}WXW_pBe7B;RYKTseVi6GR5825W0;;WKA?3aweYk_nrH
z0G56Wq%Bs~#J6wx4h81h4S(v&Gl_^N?^hw7rxxt{lgP+yU19}Yf~5EC83WFeD@U?^
z8ia#K4v@u@Cp6B^&M>+iJ9eP%0VIlG-u>^}Z>XzBf<LH}6JxHVmn(dL_0&h5hujB(
zgV`Ts5>mmC!nWl-bdLxg!KgP8^?Lw!U%otc<j9A<zQoCM1n=GQl(T0}{Zl|EO1jq?
z@)ufmeE^V9CY?<q-uh%sAT5vp&2FQ`$u79DVI9db<J0Fmm-&(1ckW|)o0AhG<++OL
zGE2Ky*6D~(<=y>~@1u#VAoyO8;pDzWZhwD+S|vhfrxQMvzIdJ%#8JfK$780hzHE36
zj0cr+{pB6A--+k+X)qK*Oc54#W^!7OX)p|r-QAyk`!`vCqa^@wN5)Oq!@&Fs;?Mf#
z+Drp3RhalT<Jus4IsbhqLl2#$LNtTD86MxZ%m`r4D`eCE9w;Lsep`PcK!wL$<Z|=|
zgkr&pEcaPEZ*Pqs&_DHFIx;+*|IUwhKr48`jy;EdR(cbcX^R{tv-NeU&)XZc8gBLg
z=|F!!ALvK)6~eJ2Gq=xC&Z>Zfa(E|kmI}s-LIzv_aruwq+O}vCTVNbpE;qn%zIt^*
zpT)<U94oI<4470@G&DRM869oiU4ZNtRCVj|KL-q9=}$!caP8{VmGQJhn26<_J19KP
zV~xE@@!lOAmjUX&97XrjgB}=}KrRZh#fOM;)+Jd&iG3e{o12=NIDwj)+GCjAo?AHR
z470RXS#hzbnAkJLmmuUnc6HGg*2q!<VJ^D^9ilz=;#xZC*&t@|>lZJkqj&O8NZ+=6
z9_V3q2@RU_jm}im)N9L=`+cToR01k8IoAqJuI}F-I;c0QHh9{swR?PG!r4+rh3XvQ
zI5<&M!Np;O&El3+E6rZ%3Y*_ATE2B|GiPl&`+N8Lr?Ss+{Ib>hRk`_Zn44CAJu++M
z0#(%V?c2AO78Om+Rah?Eb8~~hniRYCVr-*lMEY^e(ClEmWC>@=)|SC%3m#W$*NqP&
zD;ky|2{%L@cMKmUxxj&(j;anufGoMWu@=F+v%k^f+Cn>}e^I@2g~X+Us%M@ScibBJ
z^ZL`FApNAJa72>GNc4l>)2_WJIygLjj&}$1^#9TVxKF0XfB(X`!m2l<_A&9MBj4+I
zwsSNzFHh<uP!R~5k;?22Zd(<>X_>*PGrJqj2Gl;f7Y7U36dg+zpqk)AM>bV?<Fb08
z<RM9~xs~7$jy~?edBNy7*C&?3H@3065`0*GQ;z&BR&U96jraV>GFWEWoAhPv>|vEX
znx#^7GR|}0^fb|jym3QBTpS%K;_U=FGPD(*&tfV5RbJf^mTj5;k$n31(Y_gtV1Y*z
zZaLn2S#k*FDz|R+BM1Tw!qIfgTn5I*=;-M$rWtTjyM-%DXD{4lWGU-lSnexkc(z_|
zR52$os3&;4m|{T@4o<Ao_wPSU4n}D|90EuU$&VO9w<Rtt9c!75yZ-I$cInu#v^PO4
zFE8GyCzrGxrbxTgPVyK;CF)jB2{dmkEiI3prhKolY<N!ktaOj(>LzE?+@abZe(%UA
z)S?G9M$oD-O65?fp?SsK>E6mvnIk)+4@!fMZftT=iRzqz!5~5j>QhC858nBUwo9k`
z=&15f@p26Po^scvmE6tAcrsjZ`-2d{#N45uJb9ZbhWAO?nhNf8gNg?t1>qPG9-iyH
z<%K|s@ahqR-3lnGbs4=g7+)9`MwV2HyDWB@(+B;Is0wbEr&?BJp-wt~z2L#KRHx0l
zZiACb7e#l4@$A{r+Eh<$uQd{536~?!YE-mz%hMcz3U$qZmAPl9EFQF!+PCL89wW$5
zesyK`{8_A;$wkfSw3qfcJ0(%45Fb@%&&aU+%uUnO6>@_%vvV!a-kCSV%tR;ZTjhSa
zx_@iNJN2#MlNYD$w-Z{a9TQ3(EiPSdHofe$w{K&MH`jRc*_4G}DmvMB#STig*A~A0
z5*>7c&~w?;bPR>GD3da3k)^3_;al$sxzr>c!O}N$8wPE+yl)>nd*y^0=fP4Mx$JrH
z^#F{(?W(e}M}T75+UBbKcA`4ArgBSArZwoEKE=v*yf(@E-1GLa6ua$B8as{hBf_YB
zm{0<N8!@f(eCy`s29C#TagrYCu0kO#g_d5}x*%%P+rvPhapKu;uH1y>>}SS#$A^BH
zH|OL_&A1vGlCsHpdX@vP*U5oGC@B#;Fi-HvSC(%HDP<D-xKZ>bCMx{N(-*&F+H1~7
zU)o?$GaDA0s}2;l<mcb^xa!N7Kj2cBy^Zs_i1aa7V8oRQ<+PNv*Nc&E^oOo4R4ST8
zpL=lZ(TPvO<-AFpoVPXH#?D;NxDgf?CxM(q!6*b5%Fr&(MKUz>-?q#U&nl5pIiB%m
za9cTv_`CdXH!_cn&uH!y+Btc0?e?|maK>7fdYT0WFh$?L7lO7aD_ehmcYD#^J(3!u
z>cTW;5ly%0FWH5w^6!?OHGCIze&U3{Bd5XW_L$C&y>84z(>5|fh+gMPr18|U7r5Ur
zRG4qKtQ2{eUF`uH##C`Iw)kTYx?e*0@j@k_H$So9a}E37KyB*+uT}Moo3w#EohiQ&
zK$27fHm^Ani>a5AC$AC+GE|$U^p{mfHs<%=)b@?=cK5GV5-*h7FTXV%(|<?a)=QeF
zFj36vN!Z=0hi!)h;)Gp)D!v)q+UO%C_;|ljZJTf85!K47c-B>)S(ZB|Eo>I0I>efx
z?V3FF#yawsv1@mTQ-xZu^M{nC(omOmJDbsG#q2k-_bCZ<&%Pzofnv!t{(8?_JxMns
zrxQlOXD9OgRQ`V7EO3hX={#3b*k=><%IjMSqw=5odHGy3HN})08|jmAI@}t;3l)0O
zws#j2H`NJ*Wiqkgv6El;GXlano#Z>eb>3Cwcy+~P*mZ}*h=$(7mwe~Orka`(TQ+9@
zRlzX6g*GL#{b9oGKZ*IDPbzpi<tu&`@phAmDx4ZBKNC`T-gAn<?b+h5V9mjd^Dn!)
zV<uj6De^C8I=<}6KH@j?jh9@CXN7lgJGs(X;-jiHS#Iuz&Dc0xDf`(SmtVtLnv%eK
z@^RM;ayDw&_X}m-S9$2!qg(UUYtJ^l@t?&W+(%CSay4R){S>bL?ThoDOHVC+(C!gR
z7KaK1@GLtki$%;~OhtM_5atHtN=)Vxp4S$J(`{!IS!Z;%dgq?xk>w|y@M9zQHND(f
zDsiW)bdTQFy_5e>9Zu33g3>P^lI_h2lHIl<U{b+kFf%jTsgii)!yFu2s}iXL+5=*n
zr+FVQi+H8A8pU_ScnnTa{vXnv^^{pa^8Wt*7TuuGPz`nUje@HLDQZy0)6<)CHlT0O
zA3obA*Yfvbn@X3r_3qKg#PeNWo36k1Wj1dY)ZUh2!gX?U#o;mf!WkKpCcHv|f^WMz
zx#&qRDJy5aeyymgnm70sKT*Q|yg=BhJubf`@hAVm{qTarm$YX?_tE^Ps*d1m<K)!)
z_ICCCJ+#Cq9T&$$L*uYCtwW5`<+#JiFMk!+6r{VgM3%<3{p~{q<Qc@l_}b$yUFCn4
zWYy-GCNq9%YXjJB2{_0FSp={I+UCn^=P#2ih$PZ%)Jl@MXPq<F5F|^?j9B;F85q<f
zG++^!5IcOw!nLUBhhy&%F2`dR$+BM-2(AtLZExR}vY{Vx`HB2@IblJ!g)z{SVF>@-
z%P+?A<3n_AK4u!zxm|4c>w9RQ^nDaBg!*qQyHwrwXi%TCxT*LnFzC<q)Rc~gSKU0%
zReE)9Y&q9@*>V1)-$Q7X7n8mnptrX0o@q@k{fk@4_M@7C;eBcr-mFZyJ!&HppIDrs
zZF&rUweL&NoJq48bF7Hq2xKEq;pS~M`tqQ?U`~4PbI)(|HYTSCgmjqkaq;IK8GfGU
z8>vaFzCIlknf6bz+&32~ud+TD54Bb68L`XZ?fLK4K9XsFHr<@Dy)<I9WomRMv)Nt(
z&r&=5w*Jwoe*8N3Tj6PMm)W;o`c3z(1%3oO>uQ}@c;i2h!1KC!yJNtUhLO^nfdaZ^
zpE4{BIL+xp=u&2bh0#g($(ldj5=7y4@l1@w9T9L6OP4e@CroO$Eij>%R=NJCG1sV3
zIZmYQ=g;*eZ<3^L>U9AF61R}rS}^<;Xak_!{8V3m<K|6tVu{$g@4zzg1&@)hcy8tz
zf_o)ByO%|xC^`ACh)8Ec!y}Y@TLJGA>G!ny4ULR!diA+7GsjO(S`t$Wb$r0Tu_7^e
zb+6{C-4`-{lAFB+gNq%XotJ+TuZm4fObp3j!(|y2sE&@#zwgrMY4y)4ia5RIxogI|
zC8&HQHsQA3Y2QAz>&>dy!#-L?u$3L+(MB25-P(G~)buGH;o*P&F|aG%MQ`ZT)h^1Z
z%Eu{E(R5Wk-@6vDpY-fo>V2$1mNP#b2s&uP-MfE({P%B=&LK>q{6~-8Q><xrZyg+X
zpy8u(CH*V->~I*E)(6ua5XGTd{9pRG|I4o%bFX8gJ}l49E-wT)T~z1Z9++-$=HtnG
z^`xPLq=(x4X3edwiG51|aGd7Nn>XlFc0D3g1UIw0=~m^}td`_0_FX)g$>~My<dRTP
zF(0{vCiaO}u2h<qLhf)!OQG&>-_#++LeC3ZM@*MOV%FwS{lx-iAM)R|M+~1t0$ZmH
z?FiPgJ>s><n;MiAy0cYYR!$Dh&bM~MioR?h*3t3JuG}o~+HeCM;ITbkVPP`G*R8t2
z_|Fz&x$01@5=+&a^kHXPJ|CXQ_;^VWVxj(E2BAYE!bah*>0>uJuCotP&J;y{dw-nV
zx9=%)$;5?ELo0O6el}?EDPOp-?DF*S<Jq-4r%2r1DXR*{WfZP7S${C5kUGr4)o7lX
zZZ^XIB2vduZkO*Z&G?I=;&M;!Jtb@dg*`PoTIf8ZD<F^uP8DsGo^to-$t#4uf56NH
z%clMYtIF3a=Z4$<u*JPnGrs`IgQ=;KQczS>2g(xqz3U8nWK-t<C`QG-W@yul3SU0*
zyeu;P)fxRKx~x|-3P}dezq|%k0}sehXM-Ji+`OjRx7$xrzEqqT-}l&*ThU83bHmf7
z-bt1Ii6lWr1$SF)t&+K<DzYHfcQHbqinJ|?c8uSw+N^t-^+8*S6blrMTFAUMxdNZ+
z#5$dc^`=s=XviREpOJlu2pu3MBWynFWzTV;etm^`?@%p%-nH{w78V1$j<J=2(}gY&
ztqO<>8dg``atDc!G$299;oBqiO!we53;I~CXN(qqS?5&Scekj^ownXC8_+V!d*dw|
z3UcHM)TCg#p;|d2PV>I-t80>S7=M?Cu!U`HphisO9VL10>iu#w=Pq0zYdwR8DJY2_
zb5yhFGkt1kh!?TdN8^N8K^SiuDkz-0J^Qr5uT*F^;Uvgo*qAg6gbQf<Mm*iwA6xCU
z(ArHw&p!kXf|Ziks?5y1jMCY9-}f8Ig^U8mpTS{zua2hy_Ex$tHMg{&NgYJoz5^)N
zKfb&ajYix|PiftN4b++t&=9EtZdB?rS5DAadJ$yzqNVeX;Fk@Zs!mIxu9F1P9jd~u
zZ@|<QmzF{jd0kBnlCwWZpEMGT^!N5DEVQ~KncV5C*Z=|kJxeDrBqS;#Vt#RvgSZ<0
zDgM2Xe471uX9iA;+5e43c7^i{3qUF<i9BU_^uvclIbxlhP;79T;@vB<ZE`bE4z1(x
zWo7U^$O6FdqaAV{m5wzftwHh<G-2~|b2kgEbf5%@Zj2YUMtAbwfdCe_d^BuO+9FKI
zyhUaHkBGJP5_p8tq9S12AT-p_@&jK(#YQ8!pdi80QdM<({L4#p+CqYYlvGseSQgkm
zURh{m1VTk~8?E$bY~*Bqw#%~v|ETG9%UmHo3-Kbf0bUZ|1FLj}|9!{!*x1Lqy6<SN
zGm-7w=^Qjoa_>NthOG5RLw<-O_&rw_pvqHMQ+s>k)aN!oD!Dc}UL$jJ25RaEP`zAS
zZ$N=KF%d{Ue5F_rW7;<gnhzp9RH-e^%?gcvt<?`%+~U!&r(hOafD{3Cfe!TRH*d7w
z-XcM%K}yc<Hl_oH1zNg}Y&~8Mj#R*s(b3U~iAzo=2vx*o1#QxX5+T$dI}hE>M+t%s
z0Xq=*;DG`<S@!n!bpvQ+ojy+~OKHE*T8faXQ|bh^qN}s>>-X>H<>mirX<Z@s-rp~|
zzym4=i-Zf*xO(-pfB<AHzxyhsiI)QHe))xS=k7g!&VFAuxqTeG6Z-N*bMxOuC8E?L
z!<!4G#>0o?C_R1IjGN>Brz#Zyha5a)s}=-`oF`8@65SZ2umlU`C95Q1nKn7=!`$3D
zH*YqK&WrEsg3hoh_1c^Y;U25kr{uz<D<?o@Vw2F=Y(lDhXBrk8%ZSS<*)}qktkFN`
zf*udR8u$zlz{qcC7K8PX00CUG4W@G+G#LoC`wt#mOaF>CA$BThIC%+Y&89*n%*=c_
z`3kvjXNlvR`}cQ5H@<8$g~yKV-j?`$^q|5DnPm0=HR;)RBRxG6re-fBKf9!~Pix(5
z@|6-MoTVZ~cOC*&a72&8!?DyyP9KobIN;J|pJ&Y?La$T9{xGU&s$F}gp}E_>V{|lf
zb-X+66h->bp49~c>DiapKonzl#P@;T#|}fA1zNDxsKn~`g9xTWhfAe~mU9ma`m)v1
z3A8(`vv#!Tvh^74N-Eq9y^NKW5WJ3$kMHzpbXOszr6Def^7n(1|Fmz^7Q8DZRb!MG
zLId|pl50Ydgz6PM_1Uy`@AbJl=tcc0S#xikLY8u1uF8I%*5;B;slt5QxSTun^oq&S
z<7+HcT;5LRE2(d*ItJ$&1;~E@wxS?}B=P9cqh2e+v`|1B=QZrx(fX29D#avd{&V_|
zt54CSj*T@)_<p437<qH|&d?iK8XG#Rzr47q1f>#NZt+R65G9Auh+n%F{_z4qCxa{1
z%2Yq#w-~-SqPE;?ex)%E7F!C1p6@4fpyGpJLoEe>08x;zkWgw`8d5u0;iOAp-|~Jg
z$UPrUjpMF5DoL({LK_x<S{uBjH1vfam*IQm?o0aKjAM7|OKu0G4jiD8)Vdh<XGw4O
ztRs{%#ZUo&SAW|zIW{)-cjCLW+rovLH)DXXbb`Uk%d0yoyxROmE8(%sCnB<(M>{w;
zsII>CC!1tlMNu&zE$#SOs=Z_eUtU~9%NIKF6K-SoOJ0n12mIn$)2>KdE=bBv4*n26
z{@lp)cHu|I5x((Cx>ksu8a{pUTGU5`i&A<oeeQW1n&-k$l!O#N`}9c>Og6mbvCORe
z|7VM4^X7M=-O$r}E)rFQW@hP0>~bn~!*<XqL2!E7pmr~sy2#vE3e-WEP<YHt;X}HO
z*X=*lsr#L4KNOCRkBeG%@_~Q1_@0FZJk(OnU0o~aR#x^+lO-)tx%}6-M%u4Dt!{EL
z<~OV|RA7<CcxF4%AnNW$HwWVL3lOI69(zfuyC67K!<b)tGgL9JP~f||#uLL?tqH|E
zJ$bUeH+obaOwO6_g^{-4KghTSP(D~%f(cVqjd=Q$4qkkLeOc7)pEi!ew06azFq*lt
z+_1|=4X@3|#aE7fJ~P!wv)q*&ji{cRlXF!=19C#>IV(ZLL(`#eVBkF2dDcmg@WtVO
zkAE=wq8wG#(7br{H~anLQa)^GFY;&=e1`bBuTPqdjSU)qQPCCX<vMHjy5(gh934Hl
zBF<B+9m1%8mJd9Vfw{RN)xE&LOXwXVGaWj0!>2~5wQ<1w1bHaJkEG`+BAwIv@*E|L
z1RMD;H1ImR$dcX!CLDQv+I)zxeu$#)Cpz}kA3i`{MYp#Wo#(}wK5rx=AhS?L;gCsF
z!qJI8DH;jI-i47&*9EEz4?*H$*3NPHaufCiVgv|lE1~6pLeFKJ2*J#~hu|RxDHs~+
zP})I+yguO1x{pP|1q#kTf9$mVcCbNgDa$4;?Tt2gT?jLyplK_R>gr&qG{^DdYpbiu
zG5q&&+MaC))PsnYJbD!#=-~qZpt6Kw13Drw(<~x3Y{VAg4y0L2L|rKPkuWjufPw|&
z*iALHuTWis5Jiu^RR>ZTx3?HpKnMoT!D5JTaEzc*#T!AmA|i2!Gsp=0ruef=VzD8Z
z(9zQBez>2VL+L3bPSd}BrGQyI7D=e|PiGuU#|!~83ky-g1^heI5CM5XU0oeiaA;yY
zB3}9)g!>mkK|zo{g8(1`DM<FQsA<aaZV->8$OqQrs-}|0l1DKaqV>+S4MC8JnHk@$
z3=GiJ*DZH*9ID@c>C&a#b`l5ir_cRaQX#Q`C=gwk7*Ts;H8q<1vbK(nXnXbf?$`s$
zwLy7jD>Go0<gU;<ahrY_^m)Mn!<IosV{@R|IzjUZ3$5rR=3})Rx%?t}I#2=fL!A$C
z19Zg_kT65dhX?`zi60ro5piZ2;rm#VL6DTlP=IStli-LzVD7jLXkf2rX{Td;1&5s8
z0|+Mofi&I@Relho@HFBH(j5A(P&P0@vVg8lJuWhnc#$&D{GmVKPr(QY{TZ}_&WEz3
zzj}4*X6{EQ_Pe`dmFrw~^k-v*O84yDi$uZd=->dAt$bR0aB#47Psv}r8x4!C_4yX4
zWyVMVnlfQ5Sz3i5RksXHOhV~S|H1}4h?QdkNJ;4wKmUQyU8KamP^;(4>gt1g_YiBf
z%Uo<b-`++$2wcp-oC}Oz8p9IAl8uDKjf_Up;k-7_l0U0gXPK)#h%4MXgoQZm%NH*Y
zXYeP+DRS=6znAuZbE9?2MkEgK7lK^4q7?24T}$Hlh4*F|B&;wp<pc?$V9`mzCxH|j
zpbKyXl`3Si&>|!KQZkFZbDH`Afm_|jk2Ewi%PT7bkB$)Zh?}K=(_xd13=RM7NkO_S
z4Cy4qR#>hNA7n9Q_mghN>_OZfH>BFA6`-F)w;wVqJ6H`=;=}=jJmbr#_EBy_I1lXG
zb(+NOD2?Rf?DF!Sf8@vS8ydzT-{iWUF*`R$FX^`M^t1s=iPgm^_rY3<cM|V4(C2D^
zzzTf<&Lc;z0l7g91p*FQ-9B7MRn=h>b%2GCNYd=SKlzj(Nrx#hGgwYsd_4LZP}O>P
zcnD!Y1N9%4k|<vgcbWtg$4g*Va*f;9FP(5>t3wvVK4VcZXHy%(T#5*Tv=8wOuXYjN
zkt2{bwN+OeKyRhQ5i$9KAPIb4qUq(e36%n%wv~y=aW1Y>SRbfoNGrGR+?SzSt_jr{
zx(uA0pW52O@Fv-9@{~GhIXSk_{6IV9A(dEl3*vQL9Q4tDpeQ=79QQCZRIk)26T?pq
z4(~8)({-EV-cForn}0PF8JJBe$5_a{d*@KpB6b4J0EEJkF>6Iso?=J@(G}X{Ysk}+
zWdzwB@|3=q1;Qyr?S|?BejDoR`C>k!4Nbu$!pF}Kg{1%+TR-CTcCx*#-@ZYf93LOg
z+~rDgk1BwL^nUe#gW5XSH8gn8SVFHQH8ax_TjR_2;ll?Gl>{i{53#U7>OO!X=vw-+
z87UR{db_VRW{0MbInhl*&}J4irAlnkE3}G-)Dr!3tg?s4I-F(PVHFZ^#N^l~PXuj$
zoy%q4PA>41Z$}@h=}d^cpo&XuNyMIED~wwb5F?3`Pq==xULoR!Zg*m0f|5z36@tjt
z*2k0egb$SLZp%UtI!5v8NkKjcr7M)1$uD2ZBRr|7%wS+~KQ*<5si_EJkE3ILLIO)*
zjTkiZ4i18vEJ+qgV*`C9(0yQAp)?1@0ll?CT07<##0={nW2ne$V--r}$)Ck}FJ657
z`7=)8Ib!fQJQO_eVevCEyol-9Pr}20jf|iLKRhymEh&KQ;q<(`yny2n5rHRwP}sCB
z1#5^^nEyw@0AV9;^wy>AK0FUZ-In=*`H_`TRba#fmJYlHHA{YeK13f7**sSWHotSn
z6NwhO38Yf!hkkT-f6us?ce|Plp))rRqV^MR%jcoFf>@Y^j&6CnN8HwS4zmm}Y7-+P
zRNUtvcwlF*{`m1lMTL}zhzo!r(>@bOr0^Z8FXZ!=?Q;fXG92rB?p0pX(P0K_yScvd
z>zBSS8#EomLqph<`0#Mq<SW|R+6Wjf5(y6;)Ix&scVPj^9Z~+p%a@d_QkEtr{x=zX
zrcs-Dvq?%y!s6ht2ts%dZmHC13On}Y>sO5CJPHjhEGP(h_>idmL>C6B4tn0!A3s({
zniwG#LVMppltKndN&hW#Q_~{HiR-s+-Ga#XAKCv+$mF2+#^xu)#JFPK4yn-p-aQoe
zx#i`Oe0*1AW%VJVa-RNaW@*U;AOLX&=0yPQ>%CYBv!XPf*|ne%*x3=hHZeMdn1wY%
zgB{zqu{;<`1SyXm#fUpkLk@?0i@q}y!gYrHh;mTc`tM_2#ow5Jf-K(~!bM;Z_O(n5
zkIJ?!i9zp&9D@7_D(l3F6I2Hes;jEL$;xtqoHj5J$cyo`{<(j}S)q0AD}P>M5Cg(F
z0G|@aiJv_^aGRT_pbqYaaE&<9`Mj;(OO8NPMU-O%<+ir=PoP%xe&+_Ncar%X#@JC(
z5+4vPCRkvEB-_boXldKW9ky@x8yho){uo#C$E8d@@DO;8o*zHNFnfYshaMFpRyhR)
zwVD@wSV-l0<G`##G;#=x4WJ5?<`563wFB$|Tre;*6Xl`!&wc*DuNagn(68AIc^%9e
z$lDXDHy|lg%TSAAR0Hrr^~#l3>FIN`vu|^AG1g{SDL`1(KjC&5ntY5}5Qj#f`-OfA
z&<!OJaKQ27$7g=zPk=l?$V8!o$`O)e3GdBym=I`)KRwQU$U8`oH#9X9z&v5iAvV5#
z<;ty`8-Qmh=eJtCHw8SFZF2jV(4jlA%90NG63PP<>zD;YyBlg6SRzI~DFPq15D5GC
z6F(QAxENwiqFca>kct8JARq+F1d0>~^#PGOB9Ba3dyLWujs+y;HQRsI-Q68?4@fkq
z+=2b*85tpQ|6W^5MmfKoiUFrif+hpEglgyT;Q?@TxE)|d4Am+;2OI%@Q;HRug31w+
zOBuAptt&T-Ei4$RsQ7S7NTXn9d0mqW3l67G=lJ*#An!-Kr(_g5b@*^{d3iaqR9_!a
zFASmuuoaxZUNXPQ!aamOx_}Dz0lQ*|hLAe9a0b-S0DTZ@si~{i81ly`8Do5+K8zhb
zZ|G?w6B1xOGLUpySlHX=A^d|(f;1mZQ5>Jx{$gb0I(F<JiaPxq;xtijFT`1E80*~U
zud1h)otkQFWOU}v4?%c4N5sX&H8KK;Y+D4B%3Qd6h@Wx&Nl9$HyrGIOJHCGxi1`fH
z0J~?^m1_i{6{ZDt?cU8VBxGe~hW)X%vnv5m0ZAk&A%S7Cew1n^%}n0j-p7u?OUf`>
zVPrIn`HP)9{Smx@5Md~YwK(0<czYd2w!v*<lj`i(2yT=#<?n5ae<I?*fgv!%cX$GZ
z-2|~>LPA&x)a%1`h)V+e{D44MuS#ggmY0{IfP$z316MOpB=+@rBXPmfFf}!XRP5m3
z0Gxs#456{O6LXQnc=($)6Nm=Lr#J-f$&(kKx8YhJ>|-`JG!%r_U<&Wn*`qSq_fxiv
zp}?Q`XO;+~5~yvFmhmSb3RK`)T8#0^(T^XST3K~``GWo<#&toF8u8HqK(IZ;<SPg_
zNX;Py1gma?A`R9`oEFW=0Y_yiREm8>_g_#@?fm&s03TQ;WY~c6DApD*<EVJ)64=$_
zkX>VD=pfg1Fc~;Ero_WG!U!^jGUF$nCFoZ_pm_Q6B>>go;Yq9vXfli`wKO$xpE~vV
z(<e@(HcrkX<SD7C3fkIb<>m2>!~5v@k3w;P3L0?~`zh+am=PbZ<Ka;T0bH;G=C`rE
zA%TIh60UZJhA%J%fjRGjf<+iP&ZLFAW@H2*2i5~FBAz-~xi(#j0ZBRrhT57M6fx1!
z(e}mK$fgJr00NA3bWd?$I=Tyris6592=6OLfmNrcv(nQ~KC}+mzp(|=8?-BqflwYZ
zNk&Ex$>T-<T9IfW2myQ8IDCM#&31nOec7w(>fD&>78VW-3DGUGxsB_)e_tM^7+3>d
z+zs#z-jCVt#pxb*Kp$)fhE=<O4^hn^ej@+E*q{WX-M`<=+8W+6Vj}{N2gG`nk)d|w
z%2<(i{@`DryWRK0<Kj>?WdIf7hpFi+^j8;mk`W~3vQN0dm9P#-0q|zfMhK($d;+x2
zPHX6bGQ7oyasU$F*k{ki97-^dib*oS=LAW2V{2>7g_xU|EMcw|;T}ygs4x~6o$!Gq
z&)dlMF+EL40NDWISU_C76ef&@+tANqdq5a``U8fq6bua5|KT}{i;DpuP*<bRx{I9r
zshHzy%;5vGVN|U0R+TT1I3*ol64Io!v>uF1N~!`fGR!UDqXW>~`HoLjfR27mmDDFe
z(KO8pL6*or!wj^vE+t=CCVsL(^0jp4SQ|ueB#WxfP7(0x=>5UcY@M9;?b;Os^S~f~
zK)`OlDlFdD_IBjiSRf^2b_@yG!m(kqagy#GRaFBJ&BO37Uc3lS5fMOw1$5PGOnqZ^
z3?UwO0*H@ji)Zl~q!WAw0n$UI&46!3g_pTGXRrc3-~c={90P+(!hC$K0NCHZLxum_
zMJlmh`VwJGfzNTi27&YMe2X-4An?XBwud2W4=_{#7snvK0u^auiv}j_Po2UD$uF#e
zj<z;NH+wLoi$IA`<At6KXzx2ei{{V-K$f^;ZH>`#jFJrsy#y1+5fKv+A(z%J?&OC6
zHaNe+v@gvd83TU1BNgwMn$~~%v|711EqUVge9+*sdsL&{E}pkO^)(NV93i774N{}_
zJ4^BK+~xLL2PW<-hzDOb<%<wHW2?z^n)U^AUC4ps*RC2fSX$Oz670QA!%Izl=c^Ja
z_4WtGv<Gw=J{dWyY`s@|e!;S7TC~DL+MC-XN|dstY;CY->E~WD;?P5ru|04b)(%((
zi7^&N%)szTg4t(hJ^cdKoY=la#OCIXKW(6J`}P282UO8SRt^yl_ym*-#IukXp9nZW
zN7vcWF)}!aMGsOKM0@8lM?0pL$oA4<`XUTkE=;pw#1ROk2}Ec^_%Uo2FOH05h5w<9
z&AdJVpn;)EBw1P-nxWBATJ(f5ih@}T=yAcKQBxy7vSw;+gYr*c?2eF@jSVCWWPVkq
zf@GwmuU`MQE2fdp>9XACkB|I7_9Z`j>QEq6N=eBNOCjsdw}(%jOowp;cc!Ee)p3}_
zKv%`m$_mxC3#et3IH0n?YX|^mmG(%X&cZ0hSwFI4==DB$;G~nbo8ZQ7(h6qFYU)Qm
zjF8AY0pvhlMf|W4cs0IXzYb0UU^z>CrYtWx8P|q!WAS)pzpA7YpO6=~Z`+1WIzSw#
zFsK9RFm=bsC=RL@2vBj4Pj^_{S<G(fsj0oQ>Jf`=LbNdEj{#o!Y;2D}hl%JK{f4F{
z3QP#XcOE``2(~OZm}1{Pp%^7i+I<9RDN#RNa^g@V9`HbLIet}CbabyA<q1g?W_a#4
z8WYBO9qI(1+M@f12^X5#1y&bppZ&GY>T?<!qU{Kr3edButYYU+YdY!T^7P`|$7zxz
z4SVlaah2>i{^@Uv7k7j<eWs_{KQvh4P?0|Ew-lJyxKxp!Ey#5*V<(}|&b?JUBwNRE
zQ$x$bOL*b8&gC`Ns-m>>SwD>%pEbwu5c;oM={?u*@F*9bbgDZO`>naaX<(}9@HT>t
zaAC_>dAPBp@bc!x=gRArKRSm<2F^;Eci&Gfc`p9C;76*)MhxSt?LMW&-od?8;^lwF
zbYj0(mT$~1vZJZ&r+49Nzq@gn<7`{oX^991!p~-QlO8(R@cEJBqn_5Wzb7Yciac(X
zkr0I2IZfizqKxk;YyTadauDz!6mgq58f9qwocvQBdhVXR-7!K@jB%mNY-kBrW}SZm
zs;NADmdmqXozPAvgPR04=8fJ74t#9NM6$9*U~upm2GM(a2wU4X$UM<-#?aRpMpUzC
zBAiLzNw89#rgW>z5oUQ-GWqf41ZC-J$W6DgshF@+?c06Ci>1fPXLR2>Iu8Y{c}=L6
zZM+QAm{HjkG<53}p(yrzl$FO$39*S}qnR3xz4X5ZA8n8jvSwQu%IA_J{W<6^xDsY1
zRCM@Wx~#aIlqDZ1YW-?x?`E1!AcX4CPnc`Z55@h}vpsu?;Ig~$8401mbydQxM%wD@
z1y*UwnS%tk)1f;(`~0KVtE$&XeRP9gG7=s$ouwp%x6hPYl^F5I5IYuDKV2ebKmrVN
zm`c9wo&vw8sgbO=kY<7}o@g$NN^|u6Y;EnInVCsdq*0GtQL6(K!)F?p>fHBgOHBCp
zYdlT*R1d*th}`Xeo`QXhM8=L(l5j!!Ptww2MN>2DU%u6Lfq%a?7S8`^=Yj%=j)vxi
z<H1#VZ#RX1zETr6FR<Wqw;H0Ypew_>Ly~q98usIZXheNk2uIJiZvXcbak@M|SLZ}|
zJ<R_8rgioI)APVu`c3b<DR!;p`nZPY`Yv>jsARQQU)gn)rZMYCyHLiIuz8CF7k|Dj
zS#U@7?`ZEX%8_7JhR^rL*62iqQyYiE90_k|3l4>FyHyUBCU%GNG<;ef?wjB~eqGvj
z=4-B4T2#&W{rS%$UvedbX|+nVe-HgAZq(lD<#RWjezP-{!FheN>a=|69fuKj9Pq{W
zp{p+kMe>SWHop!e>)u`2GA%8e`XC;<ba`WNZlAjP)LO!2x&=$e(x_v<&vi$2wNYxs
zeJ|O&VH>xss%hx;?K_XUqfpi}I-%R&M58;-jPGVo6L(W)*=4ezC}U_hGT0&^Hu)*{
z$)MWQ)%NW-A4$oQZ-r4SbW^fQG{*HBiQ0Ea4|PjgM@40MtjWlgcKjWC;5JZS)y~dg
zNX8{gw<;+jHPCtYIkWV-MfpzlG%ezx{d!x4&p7{JkGQs~yV4zZz47}<#cdtNvTAl&
z&5GQTcdu21<4jsGbl?7);xBEb==H^OZX>_soH8$c*{KouvWt$TEx+EJJdnK9FsC3_
z#CFHO$33~PwxmL0xw&`MNxj%$Y`JXzfTFPV)%#bTY*k!W*O*!V`8qSr$x_VRaaZ=M
ztE#MY*RFUa@kwS@R#Gvme)v*XnKK^qx=u;+`>XA*-HZ_R)sjn|(rVYlhS|3}giBl_
zEjKUpxUONML2-CsDP2~ZbwPI^Y{dFh6VLvcWB2lE+@u5#D}9*Z7fzn{F!0pzQ*{YE
zn#OP1xiJu3xVG;St5s)D=j4IZ-ma%_=s&Pm|1f_x^_q|K!#B%~AG{JrUX_l^S--p|
zU}qI#6@ES-X(^z^eRH^srMEx+Ntm10c!faoRk^Ysq0{LUbY}#b8x4lXA07xlZE{=T
zyM;;5pSVOX{z^A_W!A$?X`zXKF4u<+<#FD-vif4{=s-=+nM{FsUp0%r2`V!&alhX<
zUKRV5p(Pa)X0PfHfg9VP-1Et6`g``U<jS7&UX%XATS{)O-KJ7Sy4%wv!fg7=v`1Bo
zbGX+!?u;*e8QFDLllCR0?xQ?CI^osgI^AUl%ZXHBZu`dhwJ-w>K0B)f-SG22c80$5
zdOMw+n08Pye7R>!<*)um1@{bVo_x?V!IfT9J6-a__wt4Zb-Le37pzivvGI;hztvsg
z9wig2T%9fFKBKpM)Oa{pMKH(4o<gBHcuc_c1%(?yhPEPX+y9&I{|E2=CkFlR0P){=
zkxjfcWkuAMnrpl&d7GEzO3=+W%1-7w2W<9`yY-9R9dEDos90Y4J1VA<&8azEAH)Ck
z{6x5qRuV9}BZbeV!}>(zXS21PALFg!E9?g+Zy28|ir>E=vou+nsQgKkZ*4YAzPda=
zYe&&BB{iSvR|P+U-F0qvIKPoHZ!bC2e+VxXp3J{==<85M-<smjl9YpqBJX&179As%
zw25D&>|1UsHLKwbdE~twX%bm^YPp-1C-c%aD-w%^_xWmC8(}J650B4r$Uh889rWkE
zJ??y-KA&gLviCn>Ey{}UZ=C#ZEW>{chyS&pTO>bIzRw6!*(u_K3<($ImCt3#-tzq)
D@h%gt

literal 9079
zcmdUVXH-+o*Dp#jQUjs)A|OQ&8!a>e5kZP{sfu)@NDEyd5D)~BE*KO<niQ!b5Huhl
zEz&z7^qx>cLd_ffKkr@7d)Hm}%l&lMnzPoPJu`dnnKOINZ_hsQMg}?zwA{256ch})
zceNf+P@E(Fc>t7TPRaq(eF}>6M@IUl+T`l_^XDlkDXFNasHv%GXlO26xBvhEE?&Gy
zM@L6bPtU->z{tqR#KZ&y0-2eaSy)&=AP_4nD;padJ3BiE2L~r7Cl?nNH#avA4-YRd
zFCQNtKR>^KfPkQ&ppcM|u&}U*h{&Z&mqbNH#l*zK#l<gQzAPajAt@;-B_$;-Eq&$6
z6&V>BSy|buSFg&+$z8j4?fUiWH*VaxdGqG2Tesxp<rNeZ6crVfl$4Z}mBC=Jii(P=
zs;Zirn!38WhK9!N+qX3}HMO*~w6(SG+_|HpqjUG}U0q$>d-v|?>FMd~>)*eB-@w4Y
z(9qDx$jI2(_`!n*CMG7Prlw|QX6EMR78VwkmX=mlRu3OOw6?akv9YnWwY9Uev$wZ*
zaBz6^=+WcHj~yKypFDZu<mBY+?Cj#=;_B+^=H}+^?(X5?;pyr5^yyPCFE4LzZyz5Y
zUtiy6&z|}D`T6_%2LuEJ1_lNN1wkN?;NakpkdV;O(6F$u@bK`6h=|C@$mh?Wzj*N?
zDk>^EIvNUv#>B+L#>T>6u(-Im`1tsjFJC4kBqSy#zIyd4DJdyAIr;VL*C{C}Z{EB~
zO-)TpOG{5r&&bHg%*@Qn%F52pe*5-qPEJm4ZtlBx@8EEFUS8h&_wPS^_>iBUUr<o+
z@#DwB!op9VJ{1)e6&Dwml$4Z~mX?*3eg6Epyu7@kqN1|0vZ|`;%a<?J)zvjMHMO<1
zb#-<1_4N%64PU>0ZES4(_U&6!Q&V$ub4yDL0)c34ZT<fJ`;Q+#kVs@(TU&d3`_G?0
zJ32Z#J3G6&x_<ro)!p6Q)6>)2+uPUI*WcfdLZN>D{yi`-FgQ3kG&D3kJUlWoGCDds
zHa0dsK0YxqF*!LoH8nLoJv}osGdnwrMx*EE<}et{{QUgF!ouR>;?mO6^78V^%F62M
z>e|}c`uh6D#>VF6=GNBM_VzXwi^bt^J3BkOySsaPd;9zQcs%~#;DA6N5Q)UY!^5Ma
zqvPY_larIv)6=uFGZKlUo+|an87}(Vweq8&pkw*-oEw~pIH8~rq|nvUFooo9%>>(k
zT$*sc{R^R!dk;r}^;KO6>f4r@EJ#E=ZMVCIHtTv4Te<Q!TW0B3GmpCZ$<m*bqlNXw
z<pFB8Pd(}?%FC_m=Ima44^tI~y5FSA`qjG#?+p(p#i?dP{8UaP#Zt55xz1HRi4CE6
zP6O>^{r{r8!pS$BV=QxfB6Tu)=0Oqk3su7&QHY6Tud1LzE?h)9(jpySrr?6C_opvn
z6ki1rVSX)n*Eg1L)@TG#LRW4~ViK-1jolpK>ysLoq4?7S@<Ve=)}3R90SF&*g)vQh
zI9+;f`cL0v<%GT1MLjL_s@HQ|?j4d<%eO$=k;&8bK!PU3z;Qhw?9^^1R1jPw$9C4W
zXBc5^T`DSdD@^jc68pjQNO-x~>!;&d-!G3(Ia$M9h<VhXqj6dJd}rO+k^d2>d{VtV
zvsvdauMRONAQd2P8IexOW~{=zB?lwb$mT+x!6X35|0<}U`^gb!%Wdmcls)EZewLS@
zD+L*_6~?C}rCo`8J73tdl_w|St~`5VQKTCFn+m#;{7DauQ1kW=fx_-OT&ZOXp&R_%
zIo`{BpOOFVWYG6z#hcd;z<39M_pTTG)D<)S(Ma6FPscjPVdF+8FV^49|B2-joAK{+
zZs*Fv>Vvlri2jQl=~HESM<&2o)0_6=O^=@#Mh1C^^H|3pQtVkgbvbTLn|3ks*TxOc
zS=`nG9W9sKiRq;c>#J09g?l1%*?4-?BG;2OAtpm%j-<g9ImE&F7v-ri!!Y%FZ)eEk
zh~dTeW@pwdLBV13XDKT*XbVamDi^4>cK2<QQyPWjYOdYe-aF$j3}q=X_hXMTTj-8r
zr<AQHX)M{t8TPs^Cg{E{{CUY)N(X9ABZCiZb1}n#msS%I+Jl+z#3vc5GZnuAsL#Ks
ztMOBR(yHO-k!Qt#%%Q-iq_-hkq~E=}^32}SQ_Xg^^-^HqZ*D9DqKDF1?~c}!?(|Yu
ztH%y0J)9dFvc=Nx_<G+x4LeEQ6~9uV7_x&vfOv$!oj2cEnzJr?x?!EiP{D~7UJv!P
zare;Nl{0qA;qvhu+VqC^y6`fAr~|;W`UW;j<(KEvelCph5f6S8$XLYUZ1Be!7}^wd
zS6Rk;fK(lc{6S$FnBnVQb_;6rSf(|Qd}-(Tp3Ekx@-3|luWdQ%p)rAG;_L2RAUt%~
z1<kdNb-wbNuG<!N`LUj;Npz8(lrW9X$L>Yr?)R$jmYByp9y(G2ZW!o&w9U261a*j4
zV}qkBG(92nkaa^c_)KA9^7)qPO@v33So_D7kfdj8PFTrWmTypY)6v}WB0`KHILO8}
zebI3{%k@jC&RRnE;ftJbkCRcDLEowwrg||Zj85D|FRvIgVm~Z@efOszSiCmZS~{q%
zRxgN4G`@$H8w57HCc0R-`t^LWG1An-TKvTgzk#!aqdYhd09g>^XVzR(*CKqo&{Q&{
zly=_FDgIipf12dDm8&C1ZqugnTeT5<6fSN{VqM5*x9_0m9$gN5?PZ@&wN)<PO=0vr
zFA7X{7g5_)2LVFh+nJen6crR0{UoqF7JSuK)xus5jT<kw2MpLFB=lOFP}^g%xMG`J
zeosdWVaSzo)L>lSql7mhuYz(NmSkAO^p$W8%m^@4XMWe;P`>2w;Pp8Z;Zo5d;RU<>
zewGP3m0k4dj3r1Nve487?j7_^@o`QH^R)b6Q<6xRK52c(IPR3#x@xEn5i!9&2oA9t
zr7GHgsed>54M!70?R7%$^x%_CMZsp{Q^j9Dnp%ex`Znt4yO56w_zVNDH!<=?`sXpP
zR-54LG9b}=BNP+05%JzR1Yh>W%^T||LXkSe(Je%L@WG91ni)#sEsygzct5-gokawW
zy|~E_@a7HTpi+VqwWOsqy}w*HXyS%_6A9Q-2hMV?yU={S1Jf6Hpb+r>>(%}VC8uM`
zVG#|;!Ay?8>RU{szp1gYxjy8xMY48UQkKI-sEhT&k%KY3XR7VlyrUFvbhAmTmWmM%
z09EbCoD|F`z;TeFGis?EoJ<1Y!O0pB^O^1{J!by<diQBn-8{r2=<o`LFL2Ti3_<rc
zWiPYro|jb~r;db%<hrOs4tVKz*L`Jg_*&vbEmqXMoM5=k!_YU84$+a=okuRG;2C+j
zka1R!AXv{!i|T10ud>_`?5;h=<4z>qCl4T;k1p4gT>v3WQtd#^i{4_)E?_{`U@Fgh
z=aq@H8T7Q*cbgQ$2AGgFbLN)aKm32Ok)CK1-G^)ZsgT#PeDX1xKSc??eyK}iMyNp{
z0P5>lxCxVhKiU?wRrT>y=Y}s0)a9Y?e4>U|!%F=0rB0*dK`sF5vvRc7qncskmx?K;
ziiE*aE+E`^eKCb6$qL(80SKeSU$%XY)Bgq+dUJ_+iZ9wGsDQsv3JBNeyGX}z6S5f2
zBi)n~k|v`8S)T3=Ox$%30cOE1VIgT1WX^s_xQ$-M*kfbKo)zV5=aJV7ePf?S9ouW|
z6E&tIlt@efRNBB$0{4Y)8MM2`?}RlNpe|j>+t&+Q8!n@Hp&`IGg5c|cqFs#bhwQXR
zv)U>bb%AiMmu0?V+)wO3=EI|J08r55n9-JkLXb*<%zHTW55riXcstLGk3~M89|{Jb
zo;O8EnHPp-WL0zF1zFW0M>4;fJ&x{YgM29Pf~MrU-HF?JcgA?QTk*<!D&$aLUwH56
z4(-^bZGPuMCTP({m}xtrn<YQlEdYkAWjH*u-2C#Hxy=}*!3-@b5%R(O$m`TL&uUk`
zPTm;%)5gCzUh!%y2A(}#6a*NeK5tp%-}<<cq^WYj2?&Sh>b1YCI~9A+AFTv1?BPio
z7!Uw~5b7!y0)X%&S=z`~lguU)3E9c<a+1zSQsK(w>%2&dFm0Q>6eEY**gmj^a##nt
z{m`d|^s!F)Jtqtoy+Tq4T3dYJZv)o0<;;k)beseRgw=1;&?A`)>A0@$2hM8``*B9a
z<1+43ePCbrnu{c=EX4P{<bP3TtM(>(71pYLR~FdrCOgA$!lbs~*VktZ3X>Rn8A#+b
z4(jore_i!3h?3MM3xAd^Sbs%ZCH4^~#GFJ}1GRtQ+uB|ZCl*Q?d#)KPy2w1Z$6!*q
zut8X#Z6n~+7J-q82Y0_EhA>k83rrZug-M!JxtA?3PXmV?IbX)81O?c8u+@gX=#;fK
zclS!lUNupUMWROt&Nonz!{{8(^ze0eq+#*ORn{#W(@Kb%%V?j6a}#O=6GZ&9_*Ctn
zI!-MqgfV)6(LOi$YemVnt6lsyO$1DB|K;-X;#IPaq(F$q)P>K*w*K-s<&%t2J@eVE
zm=OB50Y-tr1KP7c*r>HFuWxYL^WCez?8s@WT=`4(!bf+4C@b6Kr9n-Gp`z4>usE)y
zvzm~At#74u$5Zyrsbd&Fb5f(!T!4vk^?{oYRgzGDtsP;Wz&q^t`y7VhOC?@_kfHFO
zim!<7p2=Acb8+W;N+iz{-@JxpN$-4k6eGa<x&50Vk%hpld7%ta6mGcgvMft`j)`%y
z26^voPyZgZPK|KCfMM``boJ3XYE7tF=;7l?;kZISqdz_5zZdpGeoFD4?BA6X1Us#d
ztc80RF=_BXi#&UqL*@0po~2y3jl!AUZ1?bb`<N`016s6xk2GoxKno#a#h(NL=EJ;)
zO@32jB$VIVDCkA7<5~SRAnWUaBX2u8ll^IsCiszdUHby~i*rDDexH?2@)HCsg!V`f
zti4OqC#6KaF0mSo8wL-;Y2WSw@VvDeke94XfO!N9o^_v0s~XH6`VPvFRb0T7C^h(e
z_+|UWlIafwH0*o#mY}~`a9N20BXKAcvi5e|(8%xy6~It;68m;#|Jtz0)pJOb{BZet
z65FujRSGig-ZG!iU29nAN{NvOZX*5MsLUdkus#(8-^^p^3Pq(ECPbe93z^BNnHFjC
z+omv)$Db0<+WQ>GDqZ0z#<;UM*$ae65l54`q^^z7rK`~)P2SIVTLjYJITHo>&V)uR
zD#7&jh-krzU-xlC)ObNQ#3QpX0tXyq_<le35>p!*wP99#!cUed2#RprL>4r}T@aTq
zG-%e0aN>^@kNbu;GW^bZ4dVf~soc-;TeMhqf}9BwyiW|awx-g|W67{vrnwd^(5|aZ
zRhTXWE=r+8HxJ%Kxlz`w;K^9|E+RK{OjIK*ZnFD?u2tq%{VLP2=G7P*Ncz0a{Ckr6
zJ*hpJljUlf`7%pT>-Y$05n1VdDuGttidKMdAXZfR?jJT3cRB#FamRTlD8JS>_Sr!g
zkTt2=@R$Y9NiPJ7a2q_j{z|I{8(IyU5pKPPMWUwf0tt%*qvRjJ2WsoszLXQSOcFCg
zp?^Oo7AenqAd_&$i)YmrIu^U%##Txy*^>clOe?iz76sz!rR%$5(3JpBZ8s+X%54#;
z)2|wECwj1(Jd>`NhMRM|=l`-fqzld@X^?*0jksquq=OwizOeQhIUab1iFDU=b3(+B
zF3Z)wMkeW1L}&V4+Cm|8a!tWnDm0E=F&3_cy@k)V?BFp15M96P9s=ufoSHHx!>{X_
z#$r&sW_KPr!DCRRL(`_-I)4XX2aZTb^;xftEVErOnX)2r9ndoJiqJRWD$2)}TIYsY
z4E_wU$lRC7+*^?a|H9>L>GblfRUsa;#H0ta!v!ny7Evb3gQ#^`Y3WG3%lw6aBBls-
zEOL*oIk$~98}~>YIcfBei#(ldTqibc=2L;r%m{&>kvyR{YU`0g-1RMm1YGL>7OK5b
z_UVmJc~gZs;y3=+S@m~)rfl5nmG&}?&>~b%D}9G$zVVQQuoQax$lve2KK_7Z7E-Ql
z<$b;WH)@j!vx=K)_RtOASDF7>qu_lPp|$^Nn+6_3+5*P-(dtt);+W8+=A~U^>%OX%
zZBvE&*4aT*AcPH3Sc@Kpx|97X-UhO@Q<tpu4qO#cxvkPn+v;1fEOx9zcC$Er;=Y&j
z%tNQZ-R~kOwZo@ROER*=EzS=QM&#6gzN11D-F5m3tL^mbTYZZP^#l#mPnCsa4=;01
z?LSS>R$c#{=2%kbc7%TOz1%eP-vQ8|vmPuW1EJ$ky|V2pF!DX280Nr|jT^dhR@Ovj
ze}^j%X>Fs42HeO$)WIn+!<x?q$#yWEE%ppUl7?&_lKM|g&>}t)ex34*NSE{(K`(}l
zGJ=rhX5)gu8}GqdRTuBt|K3t9h{5Twjr;%XNbefSY{mAslSxwKmA;fWl4ouDWmUTF
zkLwMnE>~!V(MARtuT@_lo5a55xUJM)fFq<~;S<IA&q}Wn{&oS=wz@|zJ5$%8+~$-f
zJEsHUbTCp6?T5qU_{hgbet54f48nR=Uidag5nI3?ZM=vqtLvOLZrWu_M#_Ie&8ny*
z2;|8VIO|fyI*l!}bujVEOl5&wJ|${01?*>{v{HK#U?vTS{=Z8KL*x!FxNjqSh}C@A
zkQiH6kh!K(JDrw7M@kr$v;-!uU0}N*Rt`R_pO+zf#GT#hrJA<@{Io0ifDjAo3+wA^
z(j$BH)n}MKYxGMCy36IgIk-Kwf$3cC*bw^I6gq*)kSi!AiFZoB@^Atc?q0Jt@qhVf
zl(euoQo67T$9b3@3_>XSjjrg;9|myJOyF|(+jeW5a)Nrtr{h8Zp9WN8N;OX|qV#GF
zKdb}EyZ@-jD1eIW7P+@$BotywrTVNpEYKP%RC{x*r82dP{mNf-13}~^i#93#=2L9r
zSL4D>w*5k$(Ljv6HLBoiN^}T)hXzrGm#e&S!i;PWxl*xT3EutG%!p@AjK<lzO~^Sh
zU>Vz<<Hn<KSy4FK#ve`c%e8`GD~Pm};*Pz|TE!~>bjj4qkSIt1+q#@31x6WfwIOaY
z-J`BTLnjzkgeV@K9(MY`A59HJm@qAJT*P->As-Ltpx9|t%j0w@Bc>lT|6Ea^vp1q2
zn^hL@NAUp>puUSb@f~s-#>qcu0qFSa<hQN!i#(7Tiftb&&kXX2*HodA5)2dP*B%44
zkN(rt`zYAWSru)N=}C!EE<B5w40ilPi46Me7X0wQYgn_2noRruir$b#B!tjrs1q4y
z?vG60XeaL?Q6NHd1y(!Tk^djM&zSsKWhM20a3f2q*QxvZV25X2!ty*u+284Oz3-Tt
z{PAQ5pi^3Wd1j4xbyTQ51jBUlvCQpx@x?_B3{b3kxbwzSaI5NzRS6)1^2#oIl&dLO
z_Dvu{gwFn(&nE{FvN%c%-oq}Hm*wW7->_rV1>}Z&&~8D8q@+NNi+k2ZQYKvLLRKI#
zYZN5vI6|I0(-w{^6139Rca|Dx8x*Yz^BzUW59Pqd8bT8TAvMQZ$Sib45^8kDVGt7!
z=K#!et0mWav}r+l#G;HEj(vht8SOzXT=K?eeB^hfrFKGnyYz6%cTz7XS^|bPdJu_S
zqL1MBBrr{49JOYy48?0A-Dem+@v2tfcg5Ypj^|2TuHrtQVTZ)y9F-?$o_1sEbnNL7
z&qL^M$>Eg{e-1Zt=cjw&^YFxm<2CEQQ1lxi%ZsS&b|}2UvZOlfS4Dd;T_+zRJ3!2i
zY~ll*Any{JP&Q1T>rM4?9}`Vo-Ru#`Bw9S@Ci$idV}o7k{TAGJLf|a?R#`_oxiu)q
z7GHFe^o?mAZ=y_t<Z$7y@F-P&Ymwjd!!&ZZtS1UwaDeTdTQFnDyRF<j_u+t}E@7L-
zgrW88+wr#!a_hI&6)o>5huxi$B2qpePmNpk^u@CCY!P{JdA4w!?|uer?uzDn`^#s!
z62&HvB+)ntm=Ws7ts~U>NZm&1=(abhzs@bTRlpvo2AP%+tt1RJJYEhsxurtLb|yAj
z$K&M~bvcYtKa`x2q3eyP8*M9R9~vjRP;|B`lt{G&zM`umMeQFci&q>3u4Bg58d0om
z0{tqoQ*77OUUha(s?Z#-=iS0k?Hp@4F=F#3n1N?pouSY|KC#)ArIFGA4PvD2%hM{-
z2sBk<#fa%KJ@CxFryl6tEo>X~0p505@C20C7~3-nJpS|^-liaU;@J9A;Ibz#abhmI
z_<|fGfHeMDC{NmqUE#`Y6<CToF|s#Kx=jD;&LggM#YAdkgJP<m;#vQNA(~AcmAEhJ
zkn)5SJGlbW_|c<4ja?Abd*uDJ>A;PntpyRXb}-PcpmDJBM#w_tv!AlX;&S|eO1UuC
zL&INA;kHU7I~*NUJkC(%lkSFhQ3X6X1n3xrYcH+coc#S<xU5=iIF%CwRl}GvPy35w
z_9Dpk$!xZH8{Ai9k(^kSXo7~YEFe`*bfWC|85kb*MDN=)B#d2UB6U0;pK56#=$U0T
z-`tb^+*T%T-A)@JgcKvpkubyGd>GNDHk{hhHM3lx=*}-3sP76VSDcbZkQVU>HZO&(
zufg<a+_6nGU*)^%EHSzj!c9Y^=N`licip>sDcFLNg;VbjaO#WtZSwo9vOZTrS%m}Q
zRii`LZ{X<d7q&fWma1Gg@&{+ZtvGF%4}E08-BrKUk>^8Z#`Qx2ANXn5=-XPc!ldF$
za}8OqiQj{aP*IG57}=7=C9e6ru%3FSm}zwt`rBDCt@1LL#LRcK_iIs6n>xEhN5HpO
zGvcQ@jqUFDGb-Zc)jNBWNg>>|S7|HH1lv_Ryo1oY1B?ZQjm?PLau>&;-bd>0ne){}
z_PX}TwvX)UiaRwJQ*p_&7lBK3geM30_gdE$9@r&^@b6av>|5W%rYv&^CZNG+>{rE-
zN!vF}I-Jl}OELeldTUOO5&8L`lrrpb@d}1!7KSdV_8H&IE_x{d+<VtGQ^!4KTl<a3
zcLrlIRMrS6zI_v7*?Y#nZLr5}Y}n2PW+XX$X@CvOO37UDb=O|CK2ynEmiIq}Rs9&l
zjb-$F*7|Z9Rc*JeScY_QuYzsd``n+o$|Bo_e#pN(_j*%K=t`*Wox{6%x$0OPH?iT=
z>V%H8N98SUs8NTma*Nk{FhuJZZPP+GM{T2gLK{igGL`knnW(QxcvB{PrvuuVu4o0d
zXxrviA%YNYs5u8e7`UwPWjp3W;j{GW-y&^8&h1VM29K^0nGsHgR#oYaUtJST(^<1q
z__K4jTRSFyiMWZ=Scu$uy#{M-Zv5WUF)Q+bSvJ*0Cr$58dx^OfV%0rm=Uj8h22>YT
ziMEyDP*Rv;@aN+$%VMcjTCE#DmyQ&@bsn|9rC{^oEu#rT<kHaEHE+7m5jr~CPIfWr
zRId~SmvZy;(~|VegsC{}#7Q^!Q3!^lx@Yd2^{Q*2df46AwSP5k<OP<dB8@@vAlFl1
z&LQuFrvJjAvBN!#z)6+cliGyecQ_Z#IA8OF?06N7A8=h$5>{+e>m0LP1<Ea0k8>Di
z=t3>V!j)PT#)k&`f7Tc2R4B+8j9QGUAp})Ll*ArpNXJ*zFl+ap*n2Xi$OAgt@=l*v
zMaHDo#9aM+tgzAe;^(2<^^>uqiXD4iOBp>GpFC&rfLfe^b@)Ys$ZGL|GX$SbS($u)
z$FIJ+0n|Qj;LpjEZpME47LtILY31IbJ;dd|7R@NQqE&I;>5v~X;0*MNX+2q}Un?9q
zSASq$ud>OCh<{XFTpXA9o_SvJT+iv~kqkh#?_!=<?h}4zZwyFrCsmC;k1^B+U>_NR
zE!Caq=2ww^=UHCR#MF?r02ULLQg%;!a{F%M=SyGg4b5Vu?brrU*T;l@k4{b~yUhM*
zie;jCC`1>zl}za^d%2%>=L^T=IDqdR()ML8ER;=bpI5B%P4g1r!%8Q7s@Z}{AJCVW
zzQ2QA0j)ZCxL~J3kNTxzlc~;}dU5yV4Ywj)23F5#<5P2Dy1GX@!6RXDI8j2x4@s*u
zu<B={-FtH$2%hHpU@z8IF<F$pz!sv*{j2utcZFSY)jo%rWAp+vtJlnl;S&6}&hxzZ
z{X^*chlW#2m<1SCwo7Z$p<_l}v0uiRB_3>Loc1=CFP&pR%3u`j7v=CVII*t0rqF);
z_hG0vvd6E3OX5Vn#5;+5X+$SKF%FPSC&6mSXo+bVaWqUeR8c#K81vTsJ>I4IWpMHR
z$EwREIZfs2u3K!m@)1K4MCE>Nkr)Aj*;!qC`Wan&st8RHUh-v#)iqli2|gZ+ERb@?
zhS?AL1RJ_YYn}@{@@VLr(ygK@)#MR{MAnGa#%3NBd14<Ew?rQ~yPHU_heHMSaa^)K
zqW!GR&R;RW1<KeyS^9*rXE1v55r@4$W%K3N)A}u~4I7@4@~+W`4jXeq23u;5TjH%q
zlpkv%esO#0Mq^`5AxzJ8^ZTw7mhT8(kD;R!<J?-BF2`c$17ElrERX|}=M*9s?7Fh`
zIWoZg_g$3yFLJE*KkuK&kw$GB9lv{|annDEv9K=2E!hhn|HgF#jQ@Sb<Q`-v{0BDw
zuWxx|1q-?U$*yw@JJNE2tjGkD46Y1iA~VPRU=r-eu?+Wk6NVfBCp8^r12Oux<$G29
zbK~EPkxRX5{_>u-ZbZ}P0kv(i8?$|OdVjNiM{4UHlz21~IX4(V{3Lw(H`9k*;w6i8
zMtTqV3H&9<(13}oj+`_73j+So;KsIpa;Tvg0*$!p0?E2&BiORlBm=dKTRZ6%zJc6$
z>EtD0nTN*<fj@0%LiERAxU32$jGV0i((#*WGzOGyO*`rm9!}F%6VCj=GVd3ji=IjU
z;{P!->xjv$LtrCsOLCsPE@2D!NlwH&Rn;jp6@WYq4Nsd;M#N9yGbkW*@-P|}zI?6b
zacDjr0b6sRi`+~ys8VhbLWjp9Ij94N!qy(jY49F6JO2zk+wLDkKKx8>2mfjBh!~#o
zK<s}5M-_)Ak#*$O)7R-Ye;=IxBSEF*96J0?n+3zb$+(>3T3;M!ZG}e%R#S7U=Z^{A
z--GfZhyif;)Y;*n$KFj+U%*gMp@%bqW+!LDzE`&5=j5v4BkvGO#Bt);D>TP#piWEH
zL*vdbnpD1cHP3MN^JV`nW&r12M7iFatzTa32*dgBQH{#f{#;+Bni}H4&A8kE?*zp4
zIE<>E1ZNjjp#$pKU4s0hw$@4_36D^Gp9I1GZzU`T|3$0DpJQDj5rAxPGj49G7FZWb
zyoe+%=x~gw8KE5XfoET!;9(Fa8<f<k@nuMN_&OOCc@ND+x<;SJ13RcNQ@#9mYqdPJ
zRAQL_9kTx4Xp;p6^8W!89Dn}>_#c5TdDN+ZL$GnIYg+#BpFX<U23i%j9isjVc|{Iu

diff --git a/docs/dev/reference/fit.keras.src.models.model.Model.html b/docs/dev/reference/fit.keras.src.models.model.Model.html
index 1409a14c9..d9c241faf 100644
--- a/docs/dev/reference/fit.keras.src.models.model.Model.html
+++ b/docs/dev/reference/fit.keras.src.models.model.Model.html
@@ -8,7 +8,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -324,7 +324,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/freeze_weights.html b/docs/dev/reference/freeze_weights.html
index 264ddfd5d..6c04d47c6 100644
--- a/docs/dev/reference/freeze_weights.html
+++ b/docs/dev/reference/freeze_weights.html
@@ -8,7 +8,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -473,10 +473,11 @@ <h2 id="examples">Examples<a class="anchor" aria-label="anchor" href="#examples"
 <span id="cb1-46"><a href="#cb1-46" tabindex="-1"></a></span></code></pre><p></p></div>
 <p></p><div class="sourceCode r"><pre><code><span></span>
 <span><span class="co"># Freeze only layers of a certain type, e.g, BatchNorm layers</span></span>
-<span><span class="va">batch_norm_layer_class_name</span> <span class="op">&lt;-</span> <span class="fu"><a href="https://rdrr.io/r/base/class.html" class="external-link">class</a></span><span class="op">(</span><span class="fu"><a href="../reference/layer_batch_normalization.html">layer_batch_normalization</a></span><span class="op">(</span><span class="op">)</span><span class="op">)</span><span class="op">[</span><span class="fl">1</span><span class="op">]</span></span>
-<span><span class="va">is_batch_norm_layer</span> <span class="op">&lt;-</span> <span class="kw">function</span><span class="op">(</span><span class="va">x</span><span class="op">)</span> <span class="fu"><a href="https://rdrr.io/r/base/class.html" class="external-link">inherits</a></span><span class="op">(</span><span class="va">x</span>, <span class="va">batch_norm_layer_class_name</span><span class="op">)</span></span>
+<span><span class="co"># batch_norm_layer_class_name &lt;- class(layer_batch_normalization())[1]</span></span>
+<span><span class="co"># is_batch_norm_layer &lt;- function(x) inherits(x, batch_norm_layer_class_name)</span></span>
+<span><span class="va">is_batch_norm_layer</span> <span class="op">&lt;-</span> <span class="kw">function</span><span class="op">(</span><span class="va">x</span><span class="op">)</span> <span class="fu"><a href="https://rdrr.io/r/base/class.html" class="external-link">inherits</a></span><span class="op">(</span><span class="va">x</span>, <span class="va">keras</span><span class="op">$</span><span class="va">layers</span><span class="op">$</span><span class="va">BatchNormalization</span><span class="op">)</span></span>
 <span></span>
-<span><span class="va">model</span> <span class="op">&lt;-</span> <span class="fu"><a href="../reference/application_efficientnet_b0.html">application_efficientnet_b0</a></span><span class="op">(</span><span class="op">)</span></span>
+<span><span class="va">model</span> <span class="op">&lt;-</span> <span class="fu"><a href="../reference/application_efficientnet_v2b0.html">application_efficientnet_v2b0</a></span><span class="op">(</span><span class="op">)</span></span>
 <span><span class="fu"><a href="../reference/freeze_weights.html">freeze_weights</a></span><span class="op">(</span><span class="va">model</span>, which <span class="op">=</span> <span class="va">is_batch_norm_layer</span><span class="op">)</span></span>
 <span><span class="co"># print(model)</span></span>
 <span></span>
@@ -498,7 +499,7 @@ <h2 id="examples">Examples<a class="anchor" aria-label="anchor" href="#examples"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/generate.html b/docs/dev/reference/generate.html
new file mode 100644
index 000000000..e780e0d58
--- /dev/null
+++ b/docs/dev/reference/generate.html
@@ -0,0 +1,8 @@
+<html>
+  <head>
+    <meta http-equiv="refresh" content="0;URL=https://keras3.posit.co/dev/reference/reexports.html" />
+    <meta name="robots" content="noindex">
+    <link rel="canonical" href="https://keras3.posit.co/dev/reference/reexports.html">
+  </head>
+</html>
+
diff --git a/docs/dev/reference/get_config.html b/docs/dev/reference/get_config.html
index 8e87dbb5a..4766d0bcc 100644
--- a/docs/dev/reference/get_config.html
+++ b/docs/dev/reference/get_config.html
@@ -16,7 +16,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -121,7 +121,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/get_custom_objects.html b/docs/dev/reference/get_custom_objects.html
index d57c6554b..e20e67c80 100644
--- a/docs/dev/reference/get_custom_objects.html
+++ b/docs/dev/reference/get_custom_objects.html
@@ -12,7 +12,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -123,7 +123,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/get_file.html b/docs/dev/reference/get_file.html
index b8ad8da05..0c5ce6192 100644
--- a/docs/dev/reference/get_file.html
+++ b/docs/dev/reference/get_file.html
@@ -22,7 +22,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -189,7 +189,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/get_layer.html b/docs/dev/reference/get_layer.html
index b221fd23b..0a26bf3bd 100644
--- a/docs/dev/reference/get_layer.html
+++ b/docs/dev/reference/get_layer.html
@@ -12,7 +12,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -103,7 +103,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/get_registered_name.html b/docs/dev/reference/get_registered_name.html
index b8bc95106..f21c3f21e 100644
--- a/docs/dev/reference/get_registered_name.html
+++ b/docs/dev/reference/get_registered_name.html
@@ -12,7 +12,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -95,7 +95,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/get_registered_object.html b/docs/dev/reference/get_registered_object.html
index dff0def3a..cb8f6c15c 100644
--- a/docs/dev/reference/get_registered_object.html
+++ b/docs/dev/reference/get_registered_object.html
@@ -12,7 +12,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -117,7 +117,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/get_source_inputs.html b/docs/dev/reference/get_source_inputs.html
index c540511df..2ce73a694 100644
--- a/docs/dev/reference/get_source_inputs.html
+++ b/docs/dev/reference/get_source_inputs.html
@@ -10,7 +10,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -100,7 +100,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/get_state_tree.html b/docs/dev/reference/get_state_tree.html
index f9dc819df..ad58d623b 100644
--- a/docs/dev/reference/get_state_tree.html
+++ b/docs/dev/reference/get_state_tree.html
@@ -14,7 +14,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -173,7 +173,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/get_weights.html b/docs/dev/reference/get_weights.html
index b5119ab35..5e9b81d5c 100644
--- a/docs/dev/reference/get_weights.html
+++ b/docs/dev/reference/get_weights.html
@@ -8,7 +8,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -106,7 +106,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/grapes-py_class-grapes.html b/docs/dev/reference/grapes-py_class-grapes.html
index e561b2a51..97c19ed5c 100644
--- a/docs/dev/reference/grapes-py_class-grapes.html
+++ b/docs/dev/reference/grapes-py_class-grapes.html
@@ -8,7 +8,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -173,7 +173,7 @@ <h2 id="ref-examples">Examples<a class="anchor" aria-label="anchor" href="#ref-e
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/grapes-set-active-grapes.html b/docs/dev/reference/grapes-set-active-grapes.html
index 5ba217059..df391ba6c 100644
--- a/docs/dev/reference/grapes-set-active-grapes.html
+++ b/docs/dev/reference/grapes-set-active-grapes.html
@@ -8,7 +8,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -125,7 +125,7 @@ <h2 id="ref-examples">Examples<a class="anchor" aria-label="anchor" href="#ref-e
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/image_array_save.html b/docs/dev/reference/image_array_save.html
index f6df2eacc..0e388a214 100644
--- a/docs/dev/reference/image_array_save.html
+++ b/docs/dev/reference/image_array_save.html
@@ -8,7 +8,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -122,7 +122,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/image_dataset_from_directory.html b/docs/dev/reference/image_dataset_from_directory.html
index 8925726e3..1b626c380 100644
--- a/docs/dev/reference/image_dataset_from_directory.html
+++ b/docs/dev/reference/image_dataset_from_directory.html
@@ -34,7 +34,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -283,7 +283,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/image_from_array.html b/docs/dev/reference/image_from_array.html
index 068b5be3c..cff9677a2 100644
--- a/docs/dev/reference/image_from_array.html
+++ b/docs/dev/reference/image_from_array.html
@@ -8,7 +8,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -115,7 +115,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/image_load.html b/docs/dev/reference/image_load.html
index 7cf65629c..26e10ff6f 100644
--- a/docs/dev/reference/image_load.html
+++ b/docs/dev/reference/image_load.html
@@ -8,7 +8,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -136,7 +136,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/image_smart_resize.html b/docs/dev/reference/image_smart_resize.html
index db8ac6922..713e9ec1a 100644
--- a/docs/dev/reference/image_smart_resize.html
+++ b/docs/dev/reference/image_smart_resize.html
@@ -68,7 +68,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -206,7 +206,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/image_to_array.html b/docs/dev/reference/image_to_array.html
index b5b691fb4..c62cecc9c 100644
--- a/docs/dev/reference/image_to_array.html
+++ b/docs/dev/reference/image_to_array.html
@@ -8,7 +8,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -115,7 +115,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/imagenet_decode_predictions.html b/docs/dev/reference/imagenet_decode_predictions.html
index fa81abfbb..9ff9ceaf7 100644
--- a/docs/dev/reference/imagenet_decode_predictions.html
+++ b/docs/dev/reference/imagenet_decode_predictions.html
@@ -8,7 +8,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -89,7 +89,7 @@ <h2 id="value">Value<a class="anchor" aria-label="anchor" href="#value"></a></h2
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/imagenet_preprocess_input.html b/docs/dev/reference/imagenet_preprocess_input.html
index e92511bdc..c08925e65 100644
--- a/docs/dev/reference/imagenet_preprocess_input.html
+++ b/docs/dev/reference/imagenet_preprocess_input.html
@@ -8,7 +8,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -100,7 +100,7 @@ <h2 id="value">Value<a class="anchor" aria-label="anchor" href="#value"></a></h2
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/index.html b/docs/dev/reference/index.html
index 0ce4bd9e4..79c6eea62 100644
--- a/docs/dev/reference/index.html
+++ b/docs/dev/reference/index.html
@@ -8,7 +8,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -262,13 +262,13 @@ <h3 id="save-and-load-models">Save and Load Models<a class="anchor" aria-label="
           <code><a href="save_model_weights.html">save_model_weights()</a></code>
 
         </dt>
-        <dd>Saves all layer weights to a <code>.weights.h5</code> file.</dd>
+        <dd>Saves all weights to a single file or sharded files.</dd>
       </dl><dl><dt>
 
           <code><a href="load_model_weights.html">load_model_weights()</a></code>
 
         </dt>
-        <dd>Load weights from a file saved via <code><a href="../reference/save_model_weights.html">save_model_weights()</a></code>.</dd>
+        <dd>Load the weights from a single file or sharded files.</dd>
       </dl><dl><dt>
 
           <code><a href="save_model_config.html">save_model_config()</a></code> <code><a href="save_model_config.html">load_model_config()</a></code>
@@ -1070,6 +1070,12 @@ <h3 id="preprocessing-layers">Preprocessing Layers<a class="anchor" aria-label="
         <dd>A preprocessing layer which randomly crops images during training.</dd>
       </dl><dl><dt>
 
+          <code><a href="layer_random_elastic_transform.html">layer_random_elastic_transform()</a></code>
+
+        </dt>
+        <dd>A preprocessing layer that applies random elastic transformations.</dd>
+      </dl><dl><dt>
+
           <code><a href="layer_random_erasing.html">layer_random_erasing()</a></code>
 
         </dt>
@@ -1488,7 +1494,7 @@ <h3 id="core-operations">Core Operations<a class="anchor" aria-label="anchor" hr
           <code><a href="op_searchsorted.html">op_searchsorted()</a></code>
 
         </dt>
-        <dd>Perform a binary search</dd>
+        <dd>Perform a binary search.</dd>
       </dl><dl><dt>
 
           <code><a href="op_shape.html">op_shape()</a></code>
@@ -1693,6 +1699,12 @@ <h3 id="general-tensor-operations">General Tensor Operations<a class="anchor" ar
         <dd>Test whether all array elements along a given axis evaluate to <code>TRUE</code>.</dd>
       </dl><dl><dt>
 
+          <code><a href="op_angle.html">op_angle()</a></code>
+
+        </dt>
+        <dd>Element-wise angle of a complex tensor.</dd>
+      </dl><dl><dt>
+
           <code><a href="op_any.html">op_any()</a></code>
 
         </dt>
@@ -1789,6 +1801,12 @@ <h3 id="general-tensor-operations">General Tensor Operations<a class="anchor" ar
         <dd>Compute the weighted average along the specified axis.</dd>
       </dl><dl><dt>
 
+          <code><a href="op_bartlett.html">op_bartlett()</a></code>
+
+        </dt>
+        <dd>Bartlett window function.</dd>
+      </dl><dl><dt>
+
           <code><a href="op_bincount.html">op_bincount()</a></code>
 
         </dt>
@@ -1837,12 +1855,24 @@ <h3 id="general-tensor-operations">General Tensor Operations<a class="anchor" ar
         <dd>Compute the bit-wise XOR of two arrays element-wise.</dd>
       </dl><dl><dt>
 
+          <code><a href="op_blackman.html">op_blackman()</a></code>
+
+        </dt>
+        <dd>Blackman window function.</dd>
+      </dl><dl><dt>
+
           <code><a href="op_broadcast_to.html">op_broadcast_to()</a></code>
 
         </dt>
         <dd>Broadcast a tensor to a new shape.</dd>
       </dl><dl><dt>
 
+          <code><a href="op_cbrt.html">op_cbrt()</a></code>
+
+        </dt>
+        <dd>Computes the cube root of the input tensor, element-wise.</dd>
+      </dl><dl><dt>
+
           <code><a href="op_ceil.html">op_ceil()</a></code>
 
         </dt>
@@ -1873,6 +1903,12 @@ <h3 id="general-tensor-operations">General Tensor Operations<a class="anchor" ar
         <dd>Returns a copy of <code>x</code>.</dd>
       </dl><dl><dt>
 
+          <code><a href="op_corrcoef.html">op_corrcoef()</a></code>
+
+        </dt>
+        <dd>Compute the Pearson correlation coefficient matrix.</dd>
+      </dl><dl><dt>
+
           <code><a href="op_correlate.html">op_correlate()</a></code>
 
         </dt>
@@ -1921,6 +1957,12 @@ <h3 id="general-tensor-operations">General Tensor Operations<a class="anchor" ar
         <dd>Returns the cumulative sum of elements along a given axis.</dd>
       </dl><dl><dt>
 
+          <code><a href="op_deg2rad.html">op_deg2rad()</a></code>
+
+        </dt>
+        <dd>Convert angles from degrees to radians.</dd>
+      </dl><dl><dt>
+
           <code><a href="op_diag.html">op_diag()</a></code>
 
         </dt>
@@ -2065,6 +2107,24 @@ <h3 id="general-tensor-operations">General Tensor Operations<a class="anchor" ar
         <dd>Return the truth value of <code>x1 &gt;= x2</code> element-wise.</dd>
       </dl><dl><dt>
 
+          <code><a href="op_hamming.html">op_hamming()</a></code>
+
+        </dt>
+        <dd>Hamming window function.</dd>
+      </dl><dl><dt>
+
+          <code><a href="op_hanning.html">op_hanning()</a></code>
+
+        </dt>
+        <dd>Hanning window function.</dd>
+      </dl><dl><dt>
+
+          <code><a href="op_heaviside.html">op_heaviside()</a></code>
+
+        </dt>
+        <dd>Heaviside step function.</dd>
+      </dl><dl><dt>
+
           <code><a href="op_histogram.html">op_histogram()</a></code>
 
         </dt>
@@ -2119,6 +2179,12 @@ <h3 id="general-tensor-operations">General Tensor Operations<a class="anchor" ar
         <dd>Test element-wise for NaN and return result as a boolean tensor.</dd>
       </dl><dl><dt>
 
+          <code><a href="op_kaiser.html">op_kaiser()</a></code>
+
+        </dt>
+        <dd>Kaiser window function.</dd>
+      </dl><dl><dt>
+
           <code><a href="op_left_shift.html">op_left_shift()</a></code>
 
         </dt>
@@ -2599,6 +2665,18 @@ <h3 id="general-tensor-operations">General Tensor Operations<a class="anchor" ar
         <dd>Turn a function into a vectorized function.</dd>
       </dl><dl><dt>
 
+          <code><a href="op_view_as_complex.html">op_view_as_complex()</a></code>
+
+        </dt>
+        <dd>Convert a real tensor with two channels into a complex tensor.</dd>
+      </dl><dl><dt>
+
+          <code><a href="op_view_as_real.html">op_view_as_real()</a></code>
+
+        </dt>
+        <dd>Convert a complex tensor into a stacked real representation.</dd>
+      </dl><dl><dt>
+
           <code><a href="op_vstack.html">op_vstack()</a></code>
 
         </dt>
@@ -2736,6 +2814,12 @@ <h3 id="neural-network-operations">Neural Network Operations<a class="anchor" ar
         <dd>Applies the HardTanh function element-wise.</dd>
       </dl><dl><dt>
 
+          <code><a href="op_layer_normalization.html">op_layer_normalization()</a></code>
+
+        </dt>
+        <dd>Layer normalization (Ba et al., 2016).</dd>
+      </dl><dl><dt>
+
           <code><a href="op_leaky_relu.html">op_leaky_relu()</a></code>
 
         </dt>
@@ -2874,6 +2958,12 @@ <h3 id="neural-network-operations">Neural Network Operations<a class="anchor" ar
         <dd>SparsePlus activation function.</dd>
       </dl><dl><dt>
 
+          <code><a href="op_sparse_sigmoid.html">op_sparse_sigmoid()</a></code>
+
+        </dt>
+        <dd>Sparse sigmoid activation function.</dd>
+      </dl><dl><dt>
+
           <code><a href="op_sparsemax.html">op_sparsemax()</a></code>
 
         </dt>
@@ -3093,6 +3183,12 @@ <h2 id="losses">Losses<a class="anchor" aria-label="anchor" href="#losses"></a><
         <dd>Computes the alpha balanced focal crossentropy loss.</dd>
       </dl><dl><dt>
 
+          <code><a href="loss_categorical_generalized_cross_entropy.html">loss_categorical_generalized_cross_entropy()</a></code>
+
+        </dt>
+        <dd>Computes the generalized cross entropy loss.</dd>
+      </dl><dl><dt>
+
           <code><a href="loss_categorical_hinge.html">loss_categorical_hinge()</a></code>
 
         </dt>
@@ -3768,6 +3864,12 @@ <h2 id="image-preprocessing">Image Preprocessing<a class="anchor" aria-label="an
         <dd>Crop <code>images</code> to a specified <code>height</code> and <code>width</code>.</dd>
       </dl><dl><dt>
 
+          <code><a href="op_image_elastic_transform.html">op_image_elastic_transform()</a></code>
+
+        </dt>
+        <dd>Applies elastic deformation to the image(s).</dd>
+      </dl><dl><dt>
+
           <code><a href="op_image_extract_patches.html">op_image_extract_patches()</a></code>
 
         </dt>
@@ -3963,6 +4065,12 @@ <h2 id="optimizers">Optimizers<a class="anchor" aria-label="anchor" href="#optim
         <dd>An optimizer that dynamically scales the loss to prevent underflow.</dd>
       </dl><dl><dt>
 
+          <code><a href="optimizer_muon.html">optimizer_muon()</a></code>
+
+        </dt>
+        <dd>Optimizer that implements the Muon algorithm.</dd>
+      </dl><dl><dt>
+
           <code><a href="optimizer_nadam.html">optimizer_nadam()</a></code>
 
         </dt>
@@ -4372,6 +4480,12 @@ <h2 id="activations">Activations<a class="anchor" aria-label="anchor" href="#act
         <dd>SparsePlus activation function.</dd>
       </dl><dl><dt>
 
+          <code><a href="activation_sparse_sigmoid.html">activation_sparse_sigmoid()</a></code>
+
+        </dt>
+        <dd>Sparse sigmoid activation function.</dd>
+      </dl><dl><dt>
+
           <code><a href="activation_sparsemax.html">activation_sparsemax()</a></code>
 
         </dt>
@@ -4633,12 +4747,24 @@ <h2 id="configuration">Configuration<a class="anchor" aria-label="anchor" href="
         <dd>Check if interactive logging is enabled.</dd>
       </dl><dl><dt>
 
+          <code><a href="config_is_nnx_enabled.html">config_is_nnx_enabled()</a></code>
+
+        </dt>
+        <dd>Check whether NNX-specific features are enabled on the JAX backend.</dd>
+      </dl><dl><dt>
+
           <code><a href="config_is_traceback_filtering_enabled.html">config_is_traceback_filtering_enabled()</a></code>
 
         </dt>
         <dd>Check if traceback filtering is enabled.</dd>
       </dl><dl><dt>
 
+          <code><a href="config_max_epochs.html">config_max_epochs()</a></code> <code><a href="config_max_epochs.html">config_set_max_epochs()</a></code> <code><a href="config_max_epochs.html">config_max_steps_per_epoch()</a></code> <code><a href="config_max_epochs.html">config_set_max_steps_per_epoch()</a></code>
+
+        </dt>
+        <dd>Configure the default training loop limits.</dd>
+      </dl><dl><dt>
+
           <code><a href="config_set_backend.html">config_set_backend()</a></code>
 
         </dt>
@@ -5268,7 +5394,7 @@ <h3 id="xception-applications">Xception Applications<a class="anchor" aria-label
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/initializer_constant.html b/docs/dev/reference/initializer_constant.html
index d2a8eb4ab..dd9854041 100644
--- a/docs/dev/reference/initializer_constant.html
+++ b/docs/dev/reference/initializer_constant.html
@@ -12,7 +12,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -107,7 +107,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/initializer_glorot_normal.html b/docs/dev/reference/initializer_glorot_normal.html
index 37b6719e2..d341f6ecf 100644
--- a/docs/dev/reference/initializer_glorot_normal.html
+++ b/docs/dev/reference/initializer_glorot_normal.html
@@ -14,7 +14,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -122,7 +122,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/initializer_glorot_uniform.html b/docs/dev/reference/initializer_glorot_uniform.html
index 26276fd79..4f58d89a3 100644
--- a/docs/dev/reference/initializer_glorot_uniform.html
+++ b/docs/dev/reference/initializer_glorot_uniform.html
@@ -12,7 +12,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -119,7 +119,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/initializer_he_normal.html b/docs/dev/reference/initializer_he_normal.html
index 45d20fecb..3d4521e65 100644
--- a/docs/dev/reference/initializer_he_normal.html
+++ b/docs/dev/reference/initializer_he_normal.html
@@ -12,7 +12,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -119,7 +119,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/initializer_he_uniform.html b/docs/dev/reference/initializer_he_uniform.html
index 70772a7c0..118ff5e7d 100644
--- a/docs/dev/reference/initializer_he_uniform.html
+++ b/docs/dev/reference/initializer_he_uniform.html
@@ -12,7 +12,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -119,7 +119,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/initializer_identity.html b/docs/dev/reference/initializer_identity.html
index de08d9059..e7cd14441 100644
--- a/docs/dev/reference/initializer_identity.html
+++ b/docs/dev/reference/initializer_identity.html
@@ -8,7 +8,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -99,7 +99,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/initializer_lecun_normal.html b/docs/dev/reference/initializer_lecun_normal.html
index b99ac82d7..917ebe51c 100644
--- a/docs/dev/reference/initializer_lecun_normal.html
+++ b/docs/dev/reference/initializer_lecun_normal.html
@@ -18,7 +18,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -126,7 +126,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/initializer_lecun_uniform.html b/docs/dev/reference/initializer_lecun_uniform.html
index 6ec758c44..699909245 100644
--- a/docs/dev/reference/initializer_lecun_uniform.html
+++ b/docs/dev/reference/initializer_lecun_uniform.html
@@ -12,7 +12,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -117,7 +117,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/initializer_ones.html b/docs/dev/reference/initializer_ones.html
index ad54ddea2..bb4f47980 100644
--- a/docs/dev/reference/initializer_ones.html
+++ b/docs/dev/reference/initializer_ones.html
@@ -8,7 +8,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -93,7 +93,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/initializer_orthogonal.html b/docs/dev/reference/initializer_orthogonal.html
index 04824d3d9..2f1fdf1c6 100644
--- a/docs/dev/reference/initializer_orthogonal.html
+++ b/docs/dev/reference/initializer_orthogonal.html
@@ -24,7 +24,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -133,7 +133,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/initializer_random_normal.html b/docs/dev/reference/initializer_random_normal.html
index feb526f3a..0e9c75f88 100644
--- a/docs/dev/reference/initializer_random_normal.html
+++ b/docs/dev/reference/initializer_random_normal.html
@@ -8,7 +8,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -118,7 +118,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/initializer_random_uniform.html b/docs/dev/reference/initializer_random_uniform.html
index 2a69782cd..4b33abca8 100644
--- a/docs/dev/reference/initializer_random_uniform.html
+++ b/docs/dev/reference/initializer_random_uniform.html
@@ -8,7 +8,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -118,7 +118,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/initializer_stft.html b/docs/dev/reference/initializer_stft.html
index 7c27df001..12c710bae 100644
--- a/docs/dev/reference/initializer_stft.html
+++ b/docs/dev/reference/initializer_stft.html
@@ -22,7 +22,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -141,7 +141,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/initializer_truncated_normal.html b/docs/dev/reference/initializer_truncated_normal.html
index 1b124c6eb..0ed4f766f 100644
--- a/docs/dev/reference/initializer_truncated_normal.html
+++ b/docs/dev/reference/initializer_truncated_normal.html
@@ -14,7 +14,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -127,7 +127,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/initializer_variance_scaling.html b/docs/dev/reference/initializer_variance_scaling.html
index 9ccebcb46..061c3b4fd 100644
--- a/docs/dev/reference/initializer_variance_scaling.html
+++ b/docs/dev/reference/initializer_variance_scaling.html
@@ -26,7 +26,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -152,7 +152,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/initializer_zeros.html b/docs/dev/reference/initializer_zeros.html
index 08bb732e1..6490c20ee 100644
--- a/docs/dev/reference/initializer_zeros.html
+++ b/docs/dev/reference/initializer_zeros.html
@@ -8,7 +8,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -93,7 +93,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/install_keras.html b/docs/dev/reference/install_keras.html
index 00de58c02..f42c35157 100644
--- a/docs/dev/reference/install_keras.html
+++ b/docs/dev/reference/install_keras.html
@@ -8,7 +8,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -121,7 +121,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/keras.html b/docs/dev/reference/keras.html
index aeecdde08..73437ba49 100644
--- a/docs/dev/reference/keras.html
+++ b/docs/dev/reference/keras.html
@@ -10,7 +10,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -79,7 +79,7 @@ <h2 id="value">Value<a class="anchor" aria-label="anchor" href="#value"></a></h2
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/keras3-package.html b/docs/dev/reference/keras3-package.html
index 1ac68abd4..c963be932 100644
--- a/docs/dev/reference/keras3-package.html
+++ b/docs/dev/reference/keras3-package.html
@@ -12,7 +12,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -114,7 +114,7 @@ <h2 id="author">Author<a class="anchor" aria-label="anchor" href="#author"></a><
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/keras_input.html b/docs/dev/reference/keras_input.html
index 5f05743d3..9e0607490 100644
--- a/docs/dev/reference/keras_input.html
+++ b/docs/dev/reference/keras_input.html
@@ -18,7 +18,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -173,7 +173,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/keras_model.html b/docs/dev/reference/keras_model.html
index 780207f84..6547735c1 100644
--- a/docs/dev/reference/keras_model.html
+++ b/docs/dev/reference/keras_model.html
@@ -8,7 +8,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -118,7 +118,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/keras_model_sequential.html b/docs/dev/reference/keras_model_sequential.html
index 62d74bbbc..22e7a3129 100644
--- a/docs/dev/reference/keras_model_sequential.html
+++ b/docs/dev/reference/keras_model_sequential.html
@@ -8,7 +8,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -209,7 +209,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/keras_variable.html b/docs/dev/reference/keras_variable.html
index 763b2ff9c..b62385795 100644
--- a/docs/dev/reference/keras_variable.html
+++ b/docs/dev/reference/keras_variable.html
@@ -14,7 +14,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -75,7 +75,9 @@ <h2 id="ref-usage">Usage<a class="anchor" aria-label="anchor" href="#ref-usage">
 <span>  trainable <span class="op">=</span> <span class="cn">TRUE</span>,</span>
 <span>  autocast <span class="op">=</span> <span class="cn">TRUE</span>,</span>
 <span>  aggregation <span class="op">=</span> <span class="st">"none"</span>,</span>
-<span>  name <span class="op">=</span> <span class="cn">NULL</span></span>
+<span>  synchronization <span class="op">=</span> <span class="st">"auto"</span>,</span>
+<span>  name <span class="op">=</span> <span class="cn">NULL</span>,</span>
+<span>  <span class="va">...</span></span>
 <span><span class="op">)</span></span></code></pre></div>
     </div>
 
@@ -118,10 +120,19 @@ <h2 id="arguments">Arguments<a class="anchor" aria-label="anchor" href="#argumen
 to <code>"none"</code>.</p></dd>
 
 
+<dt id="arg-synchronization">synchronization<a class="anchor" aria-label="anchor" href="#arg-synchronization"></a></dt>
+<dd><p>Optional string specifying how distributed values should be synchronized.
+Defaults to <code>"auto"</code>.</p></dd>
+
+
 <dt id="arg-name">name<a class="anchor" aria-label="anchor" href="#arg-name"></a></dt>
 <dd><p>Optional. A unique name for the variable. Automatically generated
 if not set.</p></dd>
 
+
+<dt id="arg--">...<a class="anchor" aria-label="anchor" href="#arg--"></a></dt>
+<dd><p>Additional backend-specific keyword arguments forwarded to <code>keras$Variable()</code>.</p></dd>
+
 </dl></div>
     <div class="section level2">
     <h2 id="attributes">Attributes<a class="anchor" aria-label="anchor" href="#attributes"></a></h2>
@@ -132,6 +143,7 @@ <h2 id="attributes">Attributes<a class="anchor" aria-label="anchor" href="#attri
 <li><p><code>trainable</code>: Whether the variable is trainable (boolean).</p></li>
 <li><p><code>autocast</code>: Whether the variable supports autocasting (boolean).</p></li>
 <li><p><code>aggregation</code>: How a distributed variable will be aggregated (string).</p></li>
+<li><p><code>synchronization</code>: Strategy for synchronizing the variable across devices (string).</p></li>
 <li><p><code>value</code>: The current value of the variable (NumPy array or tensor).</p></li>
 <li><p><code>name</code>: The name of the variable (string).</p></li>
 <li><p><code>path</code>: The path of the variable within the Keras model or layer (string).</p></li>
@@ -180,7 +192,7 @@ <h2 id="examples">Examples<a class="anchor" aria-label="anchor" href="#examples"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/layer_activation.html b/docs/dev/reference/layer_activation.html
index f452f0db2..f542efc56 100644
--- a/docs/dev/reference/layer_activation.html
+++ b/docs/dev/reference/layer_activation.html
@@ -8,7 +8,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -109,7 +109,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
     <div class="dont-index">
 <ul><li><p><a href="https://keras.io/api/layers/core_layers/activation#activation-class" class="external-link">https://keras.io/api/layers/core_layers/activation#activation-class</a></p></li>
 </ul><p>Other activation layers: <br><code><a href="layer_activation_elu.html">layer_activation_elu</a>()</code> <br><code><a href="layer_activation_leaky_relu.html">layer_activation_leaky_relu</a>()</code> <br><code><a href="layer_activation_parametric_relu.html">layer_activation_parametric_relu</a>()</code> <br><code><a href="layer_activation_relu.html">layer_activation_relu</a>()</code> <br><code><a href="layer_activation_softmax.html">layer_activation_softmax</a>()</code> <br></p>
-<p>Other layers: <br><code><a href="Layer.html">Layer</a>()</code> <br><code><a href="layer_activation_elu.html">layer_activation_elu</a>()</code> <br><code><a href="layer_activation_leaky_relu.html">layer_activation_leaky_relu</a>()</code> <br><code><a href="layer_activation_parametric_relu.html">layer_activation_parametric_relu</a>()</code> <br><code><a href="layer_activation_relu.html">layer_activation_relu</a>()</code> <br><code><a href="layer_activation_softmax.html">layer_activation_softmax</a>()</code> <br><code><a href="layer_activity_regularization.html">layer_activity_regularization</a>()</code> <br><code><a href="layer_add.html">layer_add</a>()</code> <br><code><a href="layer_additive_attention.html">layer_additive_attention</a>()</code> <br><code><a href="layer_alpha_dropout.html">layer_alpha_dropout</a>()</code> <br><code><a href="layer_attention.html">layer_attention</a>()</code> <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_average.html">layer_average</a>()</code> <br><code><a href="layer_average_pooling_1d.html">layer_average_pooling_1d</a>()</code> <br><code><a href="layer_average_pooling_2d.html">layer_average_pooling_2d</a>()</code> <br><code><a href="layer_average_pooling_3d.html">layer_average_pooling_3d</a>()</code> <br><code><a href="layer_batch_normalization.html">layer_batch_normalization</a>()</code> <br><code><a href="layer_bidirectional.html">layer_bidirectional</a>()</code> <br><code><a href="layer_category_encoding.html">layer_category_encoding</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_concatenate.html">layer_concatenate</a>()</code> <br><code><a href="layer_conv_1d.html">layer_conv_1d</a>()</code> <br><code><a href="layer_conv_1d_transpose.html">layer_conv_1d_transpose</a>()</code> <br><code><a href="layer_conv_2d.html">layer_conv_2d</a>()</code> <br><code><a href="layer_conv_2d_transpose.html">layer_conv_2d_transpose</a>()</code> <br><code><a href="layer_conv_3d.html">layer_conv_3d</a>()</code> <br><code><a href="layer_conv_3d_transpose.html">layer_conv_3d_transpose</a>()</code> <br><code><a href="layer_conv_lstm_1d.html">layer_conv_lstm_1d</a>()</code> <br><code><a href="layer_conv_lstm_2d.html">layer_conv_lstm_2d</a>()</code> <br><code><a href="layer_conv_lstm_3d.html">layer_conv_lstm_3d</a>()</code> <br><code><a href="layer_cropping_1d.html">layer_cropping_1d</a>()</code> <br><code><a href="layer_cropping_2d.html">layer_cropping_2d</a>()</code> <br><code><a href="layer_cropping_3d.html">layer_cropping_3d</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_dense.html">layer_dense</a>()</code> <br><code><a href="layer_depthwise_conv_1d.html">layer_depthwise_conv_1d</a>()</code> <br><code><a href="layer_depthwise_conv_2d.html">layer_depthwise_conv_2d</a>()</code> <br><code><a href="layer_discretization.html">layer_discretization</a>()</code> <br><code><a href="layer_dot.html">layer_dot</a>()</code> <br><code><a href="layer_dropout.html">layer_dropout</a>()</code> <br><code><a href="layer_einsum_dense.html">layer_einsum_dense</a>()</code> <br><code><a href="layer_embedding.html">layer_embedding</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_feature_space.html">layer_feature_space</a>()</code> <br><code><a href="layer_flatten.html">layer_flatten</a>()</code> <br><code><a href="layer_flax_module_wrapper.html">layer_flax_module_wrapper</a>()</code> <br><code><a href="layer_gaussian_dropout.html">layer_gaussian_dropout</a>()</code> <br><code><a href="layer_gaussian_noise.html">layer_gaussian_noise</a>()</code> <br><code><a href="layer_global_average_pooling_1d.html">layer_global_average_pooling_1d</a>()</code> <br><code><a href="layer_global_average_pooling_2d.html">layer_global_average_pooling_2d</a>()</code> <br><code><a href="layer_global_average_pooling_3d.html">layer_global_average_pooling_3d</a>()</code> <br><code><a href="layer_global_max_pooling_1d.html">layer_global_max_pooling_1d</a>()</code> <br><code><a href="layer_global_max_pooling_2d.html">layer_global_max_pooling_2d</a>()</code> <br><code><a href="layer_global_max_pooling_3d.html">layer_global_max_pooling_3d</a>()</code> <br><code><a href="layer_group_normalization.html">layer_group_normalization</a>()</code> <br><code><a href="layer_group_query_attention.html">layer_group_query_attention</a>()</code> <br><code><a href="layer_gru.html">layer_gru</a>()</code> <br><code><a href="layer_hashed_crossing.html">layer_hashed_crossing</a>()</code> <br><code><a href="layer_hashing.html">layer_hashing</a>()</code> <br><code><a href="layer_identity.html">layer_identity</a>()</code> <br><code><a href="layer_integer_lookup.html">layer_integer_lookup</a>()</code> <br><code><a href="layer_jax_model_wrapper.html">layer_jax_model_wrapper</a>()</code> <br><code><a href="layer_lambda.html">layer_lambda</a>()</code> <br><code><a href="layer_layer_normalization.html">layer_layer_normalization</a>()</code> <br><code><a href="layer_lstm.html">layer_lstm</a>()</code> <br><code><a href="layer_masking.html">layer_masking</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_max_pooling_1d.html">layer_max_pooling_1d</a>()</code> <br><code><a href="layer_max_pooling_2d.html">layer_max_pooling_2d</a>()</code> <br><code><a href="layer_max_pooling_3d.html">layer_max_pooling_3d</a>()</code> <br><code><a href="layer_maximum.html">layer_maximum</a>()</code> <br><code><a href="layer_mel_spectrogram.html">layer_mel_spectrogram</a>()</code> <br><code><a href="layer_minimum.html">layer_minimum</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_multi_head_attention.html">layer_multi_head_attention</a>()</code> <br><code><a href="layer_multiply.html">layer_multiply</a>()</code> <br><code><a href="layer_normalization.html">layer_normalization</a>()</code> <br><code><a href="layer_permute.html">layer_permute</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_brightness.html">layer_random_brightness</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_contrast.html">layer_random_contrast</a>()</code> <br><code><a href="layer_random_crop.html">layer_random_crop</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_flip.html">layer_random_flip</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_rotation.html">layer_random_rotation</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_random_translation.html">layer_random_translation</a>()</code> <br><code><a href="layer_random_zoom.html">layer_random_zoom</a>()</code> <br><code><a href="layer_repeat_vector.html">layer_repeat_vector</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_reshape.html">layer_reshape</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_rms_normalization.html">layer_rms_normalization</a>()</code> <br><code><a href="layer_rnn.html">layer_rnn</a>()</code> <br><code><a href="layer_separable_conv_1d.html">layer_separable_conv_1d</a>()</code> <br><code><a href="layer_separable_conv_2d.html">layer_separable_conv_2d</a>()</code> <br><code><a href="layer_simple_rnn.html">layer_simple_rnn</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br><code><a href="layer_spatial_dropout_1d.html">layer_spatial_dropout_1d</a>()</code> <br><code><a href="layer_spatial_dropout_2d.html">layer_spatial_dropout_2d</a>()</code> <br><code><a href="layer_spatial_dropout_3d.html">layer_spatial_dropout_3d</a>()</code> <br><code><a href="layer_spectral_normalization.html">layer_spectral_normalization</a>()</code> <br><code><a href="layer_stft_spectrogram.html">layer_stft_spectrogram</a>()</code> <br><code><a href="layer_string_lookup.html">layer_string_lookup</a>()</code> <br><code><a href="layer_subtract.html">layer_subtract</a>()</code> <br><code><a href="layer_text_vectorization.html">layer_text_vectorization</a>()</code> <br><code><a href="layer_tfsm.html">layer_tfsm</a>()</code> <br><code><a href="layer_time_distributed.html">layer_time_distributed</a>()</code> <br><code><a href="layer_torch_module_wrapper.html">layer_torch_module_wrapper</a>()</code> <br><code><a href="layer_unit_normalization.html">layer_unit_normalization</a>()</code> <br><code><a href="layer_upsampling_1d.html">layer_upsampling_1d</a>()</code> <br><code><a href="layer_upsampling_2d.html">layer_upsampling_2d</a>()</code> <br><code><a href="layer_upsampling_3d.html">layer_upsampling_3d</a>()</code> <br><code><a href="layer_zero_padding_1d.html">layer_zero_padding_1d</a>()</code> <br><code><a href="layer_zero_padding_2d.html">layer_zero_padding_2d</a>()</code> <br><code><a href="layer_zero_padding_3d.html">layer_zero_padding_3d</a>()</code> <br><code><a href="rnn_cell_gru.html">rnn_cell_gru</a>()</code> <br><code><a href="rnn_cell_lstm.html">rnn_cell_lstm</a>()</code> <br><code><a href="rnn_cell_simple.html">rnn_cell_simple</a>()</code> <br><code><a href="rnn_cells_stack.html">rnn_cells_stack</a>()</code> <br></p></div>
+<p>Other layers: <br><code><a href="Layer.html">Layer</a>()</code> <br><code><a href="layer_activation_elu.html">layer_activation_elu</a>()</code> <br><code><a href="layer_activation_leaky_relu.html">layer_activation_leaky_relu</a>()</code> <br><code><a href="layer_activation_parametric_relu.html">layer_activation_parametric_relu</a>()</code> <br><code><a href="layer_activation_relu.html">layer_activation_relu</a>()</code> <br><code><a href="layer_activation_softmax.html">layer_activation_softmax</a>()</code> <br><code><a href="layer_activity_regularization.html">layer_activity_regularization</a>()</code> <br><code><a href="layer_add.html">layer_add</a>()</code> <br><code><a href="layer_additive_attention.html">layer_additive_attention</a>()</code> <br><code><a href="layer_alpha_dropout.html">layer_alpha_dropout</a>()</code> <br><code><a href="layer_attention.html">layer_attention</a>()</code> <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_average.html">layer_average</a>()</code> <br><code><a href="layer_average_pooling_1d.html">layer_average_pooling_1d</a>()</code> <br><code><a href="layer_average_pooling_2d.html">layer_average_pooling_2d</a>()</code> <br><code><a href="layer_average_pooling_3d.html">layer_average_pooling_3d</a>()</code> <br><code><a href="layer_batch_normalization.html">layer_batch_normalization</a>()</code> <br><code><a href="layer_bidirectional.html">layer_bidirectional</a>()</code> <br><code><a href="layer_category_encoding.html">layer_category_encoding</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_concatenate.html">layer_concatenate</a>()</code> <br><code><a href="layer_conv_1d.html">layer_conv_1d</a>()</code> <br><code><a href="layer_conv_1d_transpose.html">layer_conv_1d_transpose</a>()</code> <br><code><a href="layer_conv_2d.html">layer_conv_2d</a>()</code> <br><code><a href="layer_conv_2d_transpose.html">layer_conv_2d_transpose</a>()</code> <br><code><a href="layer_conv_3d.html">layer_conv_3d</a>()</code> <br><code><a href="layer_conv_3d_transpose.html">layer_conv_3d_transpose</a>()</code> <br><code><a href="layer_conv_lstm_1d.html">layer_conv_lstm_1d</a>()</code> <br><code><a href="layer_conv_lstm_2d.html">layer_conv_lstm_2d</a>()</code> <br><code><a href="layer_conv_lstm_3d.html">layer_conv_lstm_3d</a>()</code> <br><code><a href="layer_cropping_1d.html">layer_cropping_1d</a>()</code> <br><code><a href="layer_cropping_2d.html">layer_cropping_2d</a>()</code> <br><code><a href="layer_cropping_3d.html">layer_cropping_3d</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_dense.html">layer_dense</a>()</code> <br><code><a href="layer_depthwise_conv_1d.html">layer_depthwise_conv_1d</a>()</code> <br><code><a href="layer_depthwise_conv_2d.html">layer_depthwise_conv_2d</a>()</code> <br><code><a href="layer_discretization.html">layer_discretization</a>()</code> <br><code><a href="layer_dot.html">layer_dot</a>()</code> <br><code><a href="layer_dropout.html">layer_dropout</a>()</code> <br><code><a href="layer_einsum_dense.html">layer_einsum_dense</a>()</code> <br><code><a href="layer_embedding.html">layer_embedding</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_feature_space.html">layer_feature_space</a>()</code> <br><code><a href="layer_flatten.html">layer_flatten</a>()</code> <br><code><a href="layer_flax_module_wrapper.html">layer_flax_module_wrapper</a>()</code> <br><code><a href="layer_gaussian_dropout.html">layer_gaussian_dropout</a>()</code> <br><code><a href="layer_gaussian_noise.html">layer_gaussian_noise</a>()</code> <br><code><a href="layer_global_average_pooling_1d.html">layer_global_average_pooling_1d</a>()</code> <br><code><a href="layer_global_average_pooling_2d.html">layer_global_average_pooling_2d</a>()</code> <br><code><a href="layer_global_average_pooling_3d.html">layer_global_average_pooling_3d</a>()</code> <br><code><a href="layer_global_max_pooling_1d.html">layer_global_max_pooling_1d</a>()</code> <br><code><a href="layer_global_max_pooling_2d.html">layer_global_max_pooling_2d</a>()</code> <br><code><a href="layer_global_max_pooling_3d.html">layer_global_max_pooling_3d</a>()</code> <br><code><a href="layer_group_normalization.html">layer_group_normalization</a>()</code> <br><code><a href="layer_group_query_attention.html">layer_group_query_attention</a>()</code> <br><code><a href="layer_gru.html">layer_gru</a>()</code> <br><code><a href="layer_hashed_crossing.html">layer_hashed_crossing</a>()</code> <br><code><a href="layer_hashing.html">layer_hashing</a>()</code> <br><code><a href="layer_identity.html">layer_identity</a>()</code> <br><code><a href="layer_integer_lookup.html">layer_integer_lookup</a>()</code> <br><code><a href="layer_jax_model_wrapper.html">layer_jax_model_wrapper</a>()</code> <br><code><a href="layer_lambda.html">layer_lambda</a>()</code> <br><code><a href="layer_layer_normalization.html">layer_layer_normalization</a>()</code> <br><code><a href="layer_lstm.html">layer_lstm</a>()</code> <br><code><a href="layer_masking.html">layer_masking</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_max_pooling_1d.html">layer_max_pooling_1d</a>()</code> <br><code><a href="layer_max_pooling_2d.html">layer_max_pooling_2d</a>()</code> <br><code><a href="layer_max_pooling_3d.html">layer_max_pooling_3d</a>()</code> <br><code><a href="layer_maximum.html">layer_maximum</a>()</code> <br><code><a href="layer_mel_spectrogram.html">layer_mel_spectrogram</a>()</code> <br><code><a href="layer_minimum.html">layer_minimum</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_multi_head_attention.html">layer_multi_head_attention</a>()</code> <br><code><a href="layer_multiply.html">layer_multiply</a>()</code> <br><code><a href="layer_normalization.html">layer_normalization</a>()</code> <br><code><a href="layer_permute.html">layer_permute</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_brightness.html">layer_random_brightness</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_contrast.html">layer_random_contrast</a>()</code> <br><code><a href="layer_random_crop.html">layer_random_crop</a>()</code> <br><code><a href="layer_random_elastic_transform.html">layer_random_elastic_transform</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_flip.html">layer_random_flip</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_rotation.html">layer_random_rotation</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_random_translation.html">layer_random_translation</a>()</code> <br><code><a href="layer_random_zoom.html">layer_random_zoom</a>()</code> <br><code><a href="layer_repeat_vector.html">layer_repeat_vector</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_reshape.html">layer_reshape</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_rms_normalization.html">layer_rms_normalization</a>()</code> <br><code><a href="layer_rnn.html">layer_rnn</a>()</code> <br><code><a href="layer_separable_conv_1d.html">layer_separable_conv_1d</a>()</code> <br><code><a href="layer_separable_conv_2d.html">layer_separable_conv_2d</a>()</code> <br><code><a href="layer_simple_rnn.html">layer_simple_rnn</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br><code><a href="layer_spatial_dropout_1d.html">layer_spatial_dropout_1d</a>()</code> <br><code><a href="layer_spatial_dropout_2d.html">layer_spatial_dropout_2d</a>()</code> <br><code><a href="layer_spatial_dropout_3d.html">layer_spatial_dropout_3d</a>()</code> <br><code><a href="layer_spectral_normalization.html">layer_spectral_normalization</a>()</code> <br><code><a href="layer_stft_spectrogram.html">layer_stft_spectrogram</a>()</code> <br><code><a href="layer_string_lookup.html">layer_string_lookup</a>()</code> <br><code><a href="layer_subtract.html">layer_subtract</a>()</code> <br><code><a href="layer_text_vectorization.html">layer_text_vectorization</a>()</code> <br><code><a href="layer_tfsm.html">layer_tfsm</a>()</code> <br><code><a href="layer_time_distributed.html">layer_time_distributed</a>()</code> <br><code><a href="layer_torch_module_wrapper.html">layer_torch_module_wrapper</a>()</code> <br><code><a href="layer_unit_normalization.html">layer_unit_normalization</a>()</code> <br><code><a href="layer_upsampling_1d.html">layer_upsampling_1d</a>()</code> <br><code><a href="layer_upsampling_2d.html">layer_upsampling_2d</a>()</code> <br><code><a href="layer_upsampling_3d.html">layer_upsampling_3d</a>()</code> <br><code><a href="layer_zero_padding_1d.html">layer_zero_padding_1d</a>()</code> <br><code><a href="layer_zero_padding_2d.html">layer_zero_padding_2d</a>()</code> <br><code><a href="layer_zero_padding_3d.html">layer_zero_padding_3d</a>()</code> <br><code><a href="rnn_cell_gru.html">rnn_cell_gru</a>()</code> <br><code><a href="rnn_cell_lstm.html">rnn_cell_lstm</a>()</code> <br><code><a href="rnn_cell_simple.html">rnn_cell_simple</a>()</code> <br><code><a href="rnn_cells_stack.html">rnn_cells_stack</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -121,7 +121,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/layer_activation_elu.html b/docs/dev/reference/layer_activation_elu.html
index aa9272dd8..68dde5a85 100644
--- a/docs/dev/reference/layer_activation_elu.html
+++ b/docs/dev/reference/layer_activation_elu.html
@@ -14,7 +14,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -100,7 +100,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
     <div class="dont-index">
 <ul><li><p><a href="https://keras.io/api/layers/activation_layers/elu#elu-class" class="external-link">https://keras.io/api/layers/activation_layers/elu#elu-class</a></p></li>
 </ul><p>Other activation layers: <br><code><a href="layer_activation.html">layer_activation</a>()</code> <br><code><a href="layer_activation_leaky_relu.html">layer_activation_leaky_relu</a>()</code> <br><code><a href="layer_activation_parametric_relu.html">layer_activation_parametric_relu</a>()</code> <br><code><a href="layer_activation_relu.html">layer_activation_relu</a>()</code> <br><code><a href="layer_activation_softmax.html">layer_activation_softmax</a>()</code> <br></p>
-<p>Other layers: <br><code><a href="Layer.html">Layer</a>()</code> <br><code><a href="layer_activation.html">layer_activation</a>()</code> <br><code><a href="layer_activation_leaky_relu.html">layer_activation_leaky_relu</a>()</code> <br><code><a href="layer_activation_parametric_relu.html">layer_activation_parametric_relu</a>()</code> <br><code><a href="layer_activation_relu.html">layer_activation_relu</a>()</code> <br><code><a href="layer_activation_softmax.html">layer_activation_softmax</a>()</code> <br><code><a href="layer_activity_regularization.html">layer_activity_regularization</a>()</code> <br><code><a href="layer_add.html">layer_add</a>()</code> <br><code><a href="layer_additive_attention.html">layer_additive_attention</a>()</code> <br><code><a href="layer_alpha_dropout.html">layer_alpha_dropout</a>()</code> <br><code><a href="layer_attention.html">layer_attention</a>()</code> <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_average.html">layer_average</a>()</code> <br><code><a href="layer_average_pooling_1d.html">layer_average_pooling_1d</a>()</code> <br><code><a href="layer_average_pooling_2d.html">layer_average_pooling_2d</a>()</code> <br><code><a href="layer_average_pooling_3d.html">layer_average_pooling_3d</a>()</code> <br><code><a href="layer_batch_normalization.html">layer_batch_normalization</a>()</code> <br><code><a href="layer_bidirectional.html">layer_bidirectional</a>()</code> <br><code><a href="layer_category_encoding.html">layer_category_encoding</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_concatenate.html">layer_concatenate</a>()</code> <br><code><a href="layer_conv_1d.html">layer_conv_1d</a>()</code> <br><code><a href="layer_conv_1d_transpose.html">layer_conv_1d_transpose</a>()</code> <br><code><a href="layer_conv_2d.html">layer_conv_2d</a>()</code> <br><code><a href="layer_conv_2d_transpose.html">layer_conv_2d_transpose</a>()</code> <br><code><a href="layer_conv_3d.html">layer_conv_3d</a>()</code> <br><code><a href="layer_conv_3d_transpose.html">layer_conv_3d_transpose</a>()</code> <br><code><a href="layer_conv_lstm_1d.html">layer_conv_lstm_1d</a>()</code> <br><code><a href="layer_conv_lstm_2d.html">layer_conv_lstm_2d</a>()</code> <br><code><a href="layer_conv_lstm_3d.html">layer_conv_lstm_3d</a>()</code> <br><code><a href="layer_cropping_1d.html">layer_cropping_1d</a>()</code> <br><code><a href="layer_cropping_2d.html">layer_cropping_2d</a>()</code> <br><code><a href="layer_cropping_3d.html">layer_cropping_3d</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_dense.html">layer_dense</a>()</code> <br><code><a href="layer_depthwise_conv_1d.html">layer_depthwise_conv_1d</a>()</code> <br><code><a href="layer_depthwise_conv_2d.html">layer_depthwise_conv_2d</a>()</code> <br><code><a href="layer_discretization.html">layer_discretization</a>()</code> <br><code><a href="layer_dot.html">layer_dot</a>()</code> <br><code><a href="layer_dropout.html">layer_dropout</a>()</code> <br><code><a href="layer_einsum_dense.html">layer_einsum_dense</a>()</code> <br><code><a href="layer_embedding.html">layer_embedding</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_feature_space.html">layer_feature_space</a>()</code> <br><code><a href="layer_flatten.html">layer_flatten</a>()</code> <br><code><a href="layer_flax_module_wrapper.html">layer_flax_module_wrapper</a>()</code> <br><code><a href="layer_gaussian_dropout.html">layer_gaussian_dropout</a>()</code> <br><code><a href="layer_gaussian_noise.html">layer_gaussian_noise</a>()</code> <br><code><a href="layer_global_average_pooling_1d.html">layer_global_average_pooling_1d</a>()</code> <br><code><a href="layer_global_average_pooling_2d.html">layer_global_average_pooling_2d</a>()</code> <br><code><a href="layer_global_average_pooling_3d.html">layer_global_average_pooling_3d</a>()</code> <br><code><a href="layer_global_max_pooling_1d.html">layer_global_max_pooling_1d</a>()</code> <br><code><a href="layer_global_max_pooling_2d.html">layer_global_max_pooling_2d</a>()</code> <br><code><a href="layer_global_max_pooling_3d.html">layer_global_max_pooling_3d</a>()</code> <br><code><a href="layer_group_normalization.html">layer_group_normalization</a>()</code> <br><code><a href="layer_group_query_attention.html">layer_group_query_attention</a>()</code> <br><code><a href="layer_gru.html">layer_gru</a>()</code> <br><code><a href="layer_hashed_crossing.html">layer_hashed_crossing</a>()</code> <br><code><a href="layer_hashing.html">layer_hashing</a>()</code> <br><code><a href="layer_identity.html">layer_identity</a>()</code> <br><code><a href="layer_integer_lookup.html">layer_integer_lookup</a>()</code> <br><code><a href="layer_jax_model_wrapper.html">layer_jax_model_wrapper</a>()</code> <br><code><a href="layer_lambda.html">layer_lambda</a>()</code> <br><code><a href="layer_layer_normalization.html">layer_layer_normalization</a>()</code> <br><code><a href="layer_lstm.html">layer_lstm</a>()</code> <br><code><a href="layer_masking.html">layer_masking</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_max_pooling_1d.html">layer_max_pooling_1d</a>()</code> <br><code><a href="layer_max_pooling_2d.html">layer_max_pooling_2d</a>()</code> <br><code><a href="layer_max_pooling_3d.html">layer_max_pooling_3d</a>()</code> <br><code><a href="layer_maximum.html">layer_maximum</a>()</code> <br><code><a href="layer_mel_spectrogram.html">layer_mel_spectrogram</a>()</code> <br><code><a href="layer_minimum.html">layer_minimum</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_multi_head_attention.html">layer_multi_head_attention</a>()</code> <br><code><a href="layer_multiply.html">layer_multiply</a>()</code> <br><code><a href="layer_normalization.html">layer_normalization</a>()</code> <br><code><a href="layer_permute.html">layer_permute</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_brightness.html">layer_random_brightness</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_contrast.html">layer_random_contrast</a>()</code> <br><code><a href="layer_random_crop.html">layer_random_crop</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_flip.html">layer_random_flip</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_rotation.html">layer_random_rotation</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_random_translation.html">layer_random_translation</a>()</code> <br><code><a href="layer_random_zoom.html">layer_random_zoom</a>()</code> <br><code><a href="layer_repeat_vector.html">layer_repeat_vector</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_reshape.html">layer_reshape</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_rms_normalization.html">layer_rms_normalization</a>()</code> <br><code><a href="layer_rnn.html">layer_rnn</a>()</code> <br><code><a href="layer_separable_conv_1d.html">layer_separable_conv_1d</a>()</code> <br><code><a href="layer_separable_conv_2d.html">layer_separable_conv_2d</a>()</code> <br><code><a href="layer_simple_rnn.html">layer_simple_rnn</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br><code><a href="layer_spatial_dropout_1d.html">layer_spatial_dropout_1d</a>()</code> <br><code><a href="layer_spatial_dropout_2d.html">layer_spatial_dropout_2d</a>()</code> <br><code><a href="layer_spatial_dropout_3d.html">layer_spatial_dropout_3d</a>()</code> <br><code><a href="layer_spectral_normalization.html">layer_spectral_normalization</a>()</code> <br><code><a href="layer_stft_spectrogram.html">layer_stft_spectrogram</a>()</code> <br><code><a href="layer_string_lookup.html">layer_string_lookup</a>()</code> <br><code><a href="layer_subtract.html">layer_subtract</a>()</code> <br><code><a href="layer_text_vectorization.html">layer_text_vectorization</a>()</code> <br><code><a href="layer_tfsm.html">layer_tfsm</a>()</code> <br><code><a href="layer_time_distributed.html">layer_time_distributed</a>()</code> <br><code><a href="layer_torch_module_wrapper.html">layer_torch_module_wrapper</a>()</code> <br><code><a href="layer_unit_normalization.html">layer_unit_normalization</a>()</code> <br><code><a href="layer_upsampling_1d.html">layer_upsampling_1d</a>()</code> <br><code><a href="layer_upsampling_2d.html">layer_upsampling_2d</a>()</code> <br><code><a href="layer_upsampling_3d.html">layer_upsampling_3d</a>()</code> <br><code><a href="layer_zero_padding_1d.html">layer_zero_padding_1d</a>()</code> <br><code><a href="layer_zero_padding_2d.html">layer_zero_padding_2d</a>()</code> <br><code><a href="layer_zero_padding_3d.html">layer_zero_padding_3d</a>()</code> <br><code><a href="rnn_cell_gru.html">rnn_cell_gru</a>()</code> <br><code><a href="rnn_cell_lstm.html">rnn_cell_lstm</a>()</code> <br><code><a href="rnn_cell_simple.html">rnn_cell_simple</a>()</code> <br><code><a href="rnn_cells_stack.html">rnn_cells_stack</a>()</code> <br></p></div>
+<p>Other layers: <br><code><a href="Layer.html">Layer</a>()</code> <br><code><a href="layer_activation.html">layer_activation</a>()</code> <br><code><a href="layer_activation_leaky_relu.html">layer_activation_leaky_relu</a>()</code> <br><code><a href="layer_activation_parametric_relu.html">layer_activation_parametric_relu</a>()</code> <br><code><a href="layer_activation_relu.html">layer_activation_relu</a>()</code> <br><code><a href="layer_activation_softmax.html">layer_activation_softmax</a>()</code> <br><code><a href="layer_activity_regularization.html">layer_activity_regularization</a>()</code> <br><code><a href="layer_add.html">layer_add</a>()</code> <br><code><a href="layer_additive_attention.html">layer_additive_attention</a>()</code> <br><code><a href="layer_alpha_dropout.html">layer_alpha_dropout</a>()</code> <br><code><a href="layer_attention.html">layer_attention</a>()</code> <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_average.html">layer_average</a>()</code> <br><code><a href="layer_average_pooling_1d.html">layer_average_pooling_1d</a>()</code> <br><code><a href="layer_average_pooling_2d.html">layer_average_pooling_2d</a>()</code> <br><code><a href="layer_average_pooling_3d.html">layer_average_pooling_3d</a>()</code> <br><code><a href="layer_batch_normalization.html">layer_batch_normalization</a>()</code> <br><code><a href="layer_bidirectional.html">layer_bidirectional</a>()</code> <br><code><a href="layer_category_encoding.html">layer_category_encoding</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_concatenate.html">layer_concatenate</a>()</code> <br><code><a href="layer_conv_1d.html">layer_conv_1d</a>()</code> <br><code><a href="layer_conv_1d_transpose.html">layer_conv_1d_transpose</a>()</code> <br><code><a href="layer_conv_2d.html">layer_conv_2d</a>()</code> <br><code><a href="layer_conv_2d_transpose.html">layer_conv_2d_transpose</a>()</code> <br><code><a href="layer_conv_3d.html">layer_conv_3d</a>()</code> <br><code><a href="layer_conv_3d_transpose.html">layer_conv_3d_transpose</a>()</code> <br><code><a href="layer_conv_lstm_1d.html">layer_conv_lstm_1d</a>()</code> <br><code><a href="layer_conv_lstm_2d.html">layer_conv_lstm_2d</a>()</code> <br><code><a href="layer_conv_lstm_3d.html">layer_conv_lstm_3d</a>()</code> <br><code><a href="layer_cropping_1d.html">layer_cropping_1d</a>()</code> <br><code><a href="layer_cropping_2d.html">layer_cropping_2d</a>()</code> <br><code><a href="layer_cropping_3d.html">layer_cropping_3d</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_dense.html">layer_dense</a>()</code> <br><code><a href="layer_depthwise_conv_1d.html">layer_depthwise_conv_1d</a>()</code> <br><code><a href="layer_depthwise_conv_2d.html">layer_depthwise_conv_2d</a>()</code> <br><code><a href="layer_discretization.html">layer_discretization</a>()</code> <br><code><a href="layer_dot.html">layer_dot</a>()</code> <br><code><a href="layer_dropout.html">layer_dropout</a>()</code> <br><code><a href="layer_einsum_dense.html">layer_einsum_dense</a>()</code> <br><code><a href="layer_embedding.html">layer_embedding</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_feature_space.html">layer_feature_space</a>()</code> <br><code><a href="layer_flatten.html">layer_flatten</a>()</code> <br><code><a href="layer_flax_module_wrapper.html">layer_flax_module_wrapper</a>()</code> <br><code><a href="layer_gaussian_dropout.html">layer_gaussian_dropout</a>()</code> <br><code><a href="layer_gaussian_noise.html">layer_gaussian_noise</a>()</code> <br><code><a href="layer_global_average_pooling_1d.html">layer_global_average_pooling_1d</a>()</code> <br><code><a href="layer_global_average_pooling_2d.html">layer_global_average_pooling_2d</a>()</code> <br><code><a href="layer_global_average_pooling_3d.html">layer_global_average_pooling_3d</a>()</code> <br><code><a href="layer_global_max_pooling_1d.html">layer_global_max_pooling_1d</a>()</code> <br><code><a href="layer_global_max_pooling_2d.html">layer_global_max_pooling_2d</a>()</code> <br><code><a href="layer_global_max_pooling_3d.html">layer_global_max_pooling_3d</a>()</code> <br><code><a href="layer_group_normalization.html">layer_group_normalization</a>()</code> <br><code><a href="layer_group_query_attention.html">layer_group_query_attention</a>()</code> <br><code><a href="layer_gru.html">layer_gru</a>()</code> <br><code><a href="layer_hashed_crossing.html">layer_hashed_crossing</a>()</code> <br><code><a href="layer_hashing.html">layer_hashing</a>()</code> <br><code><a href="layer_identity.html">layer_identity</a>()</code> <br><code><a href="layer_integer_lookup.html">layer_integer_lookup</a>()</code> <br><code><a href="layer_jax_model_wrapper.html">layer_jax_model_wrapper</a>()</code> <br><code><a href="layer_lambda.html">layer_lambda</a>()</code> <br><code><a href="layer_layer_normalization.html">layer_layer_normalization</a>()</code> <br><code><a href="layer_lstm.html">layer_lstm</a>()</code> <br><code><a href="layer_masking.html">layer_masking</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_max_pooling_1d.html">layer_max_pooling_1d</a>()</code> <br><code><a href="layer_max_pooling_2d.html">layer_max_pooling_2d</a>()</code> <br><code><a href="layer_max_pooling_3d.html">layer_max_pooling_3d</a>()</code> <br><code><a href="layer_maximum.html">layer_maximum</a>()</code> <br><code><a href="layer_mel_spectrogram.html">layer_mel_spectrogram</a>()</code> <br><code><a href="layer_minimum.html">layer_minimum</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_multi_head_attention.html">layer_multi_head_attention</a>()</code> <br><code><a href="layer_multiply.html">layer_multiply</a>()</code> <br><code><a href="layer_normalization.html">layer_normalization</a>()</code> <br><code><a href="layer_permute.html">layer_permute</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_brightness.html">layer_random_brightness</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_contrast.html">layer_random_contrast</a>()</code> <br><code><a href="layer_random_crop.html">layer_random_crop</a>()</code> <br><code><a href="layer_random_elastic_transform.html">layer_random_elastic_transform</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_flip.html">layer_random_flip</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_rotation.html">layer_random_rotation</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_random_translation.html">layer_random_translation</a>()</code> <br><code><a href="layer_random_zoom.html">layer_random_zoom</a>()</code> <br><code><a href="layer_repeat_vector.html">layer_repeat_vector</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_reshape.html">layer_reshape</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_rms_normalization.html">layer_rms_normalization</a>()</code> <br><code><a href="layer_rnn.html">layer_rnn</a>()</code> <br><code><a href="layer_separable_conv_1d.html">layer_separable_conv_1d</a>()</code> <br><code><a href="layer_separable_conv_2d.html">layer_separable_conv_2d</a>()</code> <br><code><a href="layer_simple_rnn.html">layer_simple_rnn</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br><code><a href="layer_spatial_dropout_1d.html">layer_spatial_dropout_1d</a>()</code> <br><code><a href="layer_spatial_dropout_2d.html">layer_spatial_dropout_2d</a>()</code> <br><code><a href="layer_spatial_dropout_3d.html">layer_spatial_dropout_3d</a>()</code> <br><code><a href="layer_spectral_normalization.html">layer_spectral_normalization</a>()</code> <br><code><a href="layer_stft_spectrogram.html">layer_stft_spectrogram</a>()</code> <br><code><a href="layer_string_lookup.html">layer_string_lookup</a>()</code> <br><code><a href="layer_subtract.html">layer_subtract</a>()</code> <br><code><a href="layer_text_vectorization.html">layer_text_vectorization</a>()</code> <br><code><a href="layer_tfsm.html">layer_tfsm</a>()</code> <br><code><a href="layer_time_distributed.html">layer_time_distributed</a>()</code> <br><code><a href="layer_torch_module_wrapper.html">layer_torch_module_wrapper</a>()</code> <br><code><a href="layer_unit_normalization.html">layer_unit_normalization</a>()</code> <br><code><a href="layer_upsampling_1d.html">layer_upsampling_1d</a>()</code> <br><code><a href="layer_upsampling_2d.html">layer_upsampling_2d</a>()</code> <br><code><a href="layer_upsampling_3d.html">layer_upsampling_3d</a>()</code> <br><code><a href="layer_zero_padding_1d.html">layer_zero_padding_1d</a>()</code> <br><code><a href="layer_zero_padding_2d.html">layer_zero_padding_2d</a>()</code> <br><code><a href="layer_zero_padding_3d.html">layer_zero_padding_3d</a>()</code> <br><code><a href="rnn_cell_gru.html">rnn_cell_gru</a>()</code> <br><code><a href="rnn_cell_lstm.html">rnn_cell_lstm</a>()</code> <br><code><a href="rnn_cell_simple.html">rnn_cell_simple</a>()</code> <br><code><a href="rnn_cells_stack.html">rnn_cells_stack</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -112,7 +112,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/layer_activation_leaky_relu.html b/docs/dev/reference/layer_activation_leaky_relu.html
index a948e63e0..6bef7dbe3 100644
--- a/docs/dev/reference/layer_activation_leaky_relu.html
+++ b/docs/dev/reference/layer_activation_leaky_relu.html
@@ -14,7 +14,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -111,7 +111,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
     <div class="dont-index">
 <ul><li><p><a href="https://keras.io/api/layers/activation_layers/leaky_relu#leakyrelu-class" class="external-link">https://keras.io/api/layers/activation_layers/leaky_relu#leakyrelu-class</a></p></li>
 </ul><p>Other activation layers: <br><code><a href="layer_activation.html">layer_activation</a>()</code> <br><code><a href="layer_activation_elu.html">layer_activation_elu</a>()</code> <br><code><a href="layer_activation_parametric_relu.html">layer_activation_parametric_relu</a>()</code> <br><code><a href="layer_activation_relu.html">layer_activation_relu</a>()</code> <br><code><a href="layer_activation_softmax.html">layer_activation_softmax</a>()</code> <br></p>
-<p>Other layers: <br><code><a href="Layer.html">Layer</a>()</code> <br><code><a href="layer_activation.html">layer_activation</a>()</code> <br><code><a href="layer_activation_elu.html">layer_activation_elu</a>()</code> <br><code><a href="layer_activation_parametric_relu.html">layer_activation_parametric_relu</a>()</code> <br><code><a href="layer_activation_relu.html">layer_activation_relu</a>()</code> <br><code><a href="layer_activation_softmax.html">layer_activation_softmax</a>()</code> <br><code><a href="layer_activity_regularization.html">layer_activity_regularization</a>()</code> <br><code><a href="layer_add.html">layer_add</a>()</code> <br><code><a href="layer_additive_attention.html">layer_additive_attention</a>()</code> <br><code><a href="layer_alpha_dropout.html">layer_alpha_dropout</a>()</code> <br><code><a href="layer_attention.html">layer_attention</a>()</code> <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_average.html">layer_average</a>()</code> <br><code><a href="layer_average_pooling_1d.html">layer_average_pooling_1d</a>()</code> <br><code><a href="layer_average_pooling_2d.html">layer_average_pooling_2d</a>()</code> <br><code><a href="layer_average_pooling_3d.html">layer_average_pooling_3d</a>()</code> <br><code><a href="layer_batch_normalization.html">layer_batch_normalization</a>()</code> <br><code><a href="layer_bidirectional.html">layer_bidirectional</a>()</code> <br><code><a href="layer_category_encoding.html">layer_category_encoding</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_concatenate.html">layer_concatenate</a>()</code> <br><code><a href="layer_conv_1d.html">layer_conv_1d</a>()</code> <br><code><a href="layer_conv_1d_transpose.html">layer_conv_1d_transpose</a>()</code> <br><code><a href="layer_conv_2d.html">layer_conv_2d</a>()</code> <br><code><a href="layer_conv_2d_transpose.html">layer_conv_2d_transpose</a>()</code> <br><code><a href="layer_conv_3d.html">layer_conv_3d</a>()</code> <br><code><a href="layer_conv_3d_transpose.html">layer_conv_3d_transpose</a>()</code> <br><code><a href="layer_conv_lstm_1d.html">layer_conv_lstm_1d</a>()</code> <br><code><a href="layer_conv_lstm_2d.html">layer_conv_lstm_2d</a>()</code> <br><code><a href="layer_conv_lstm_3d.html">layer_conv_lstm_3d</a>()</code> <br><code><a href="layer_cropping_1d.html">layer_cropping_1d</a>()</code> <br><code><a href="layer_cropping_2d.html">layer_cropping_2d</a>()</code> <br><code><a href="layer_cropping_3d.html">layer_cropping_3d</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_dense.html">layer_dense</a>()</code> <br><code><a href="layer_depthwise_conv_1d.html">layer_depthwise_conv_1d</a>()</code> <br><code><a href="layer_depthwise_conv_2d.html">layer_depthwise_conv_2d</a>()</code> <br><code><a href="layer_discretization.html">layer_discretization</a>()</code> <br><code><a href="layer_dot.html">layer_dot</a>()</code> <br><code><a href="layer_dropout.html">layer_dropout</a>()</code> <br><code><a href="layer_einsum_dense.html">layer_einsum_dense</a>()</code> <br><code><a href="layer_embedding.html">layer_embedding</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_feature_space.html">layer_feature_space</a>()</code> <br><code><a href="layer_flatten.html">layer_flatten</a>()</code> <br><code><a href="layer_flax_module_wrapper.html">layer_flax_module_wrapper</a>()</code> <br><code><a href="layer_gaussian_dropout.html">layer_gaussian_dropout</a>()</code> <br><code><a href="layer_gaussian_noise.html">layer_gaussian_noise</a>()</code> <br><code><a href="layer_global_average_pooling_1d.html">layer_global_average_pooling_1d</a>()</code> <br><code><a href="layer_global_average_pooling_2d.html">layer_global_average_pooling_2d</a>()</code> <br><code><a href="layer_global_average_pooling_3d.html">layer_global_average_pooling_3d</a>()</code> <br><code><a href="layer_global_max_pooling_1d.html">layer_global_max_pooling_1d</a>()</code> <br><code><a href="layer_global_max_pooling_2d.html">layer_global_max_pooling_2d</a>()</code> <br><code><a href="layer_global_max_pooling_3d.html">layer_global_max_pooling_3d</a>()</code> <br><code><a href="layer_group_normalization.html">layer_group_normalization</a>()</code> <br><code><a href="layer_group_query_attention.html">layer_group_query_attention</a>()</code> <br><code><a href="layer_gru.html">layer_gru</a>()</code> <br><code><a href="layer_hashed_crossing.html">layer_hashed_crossing</a>()</code> <br><code><a href="layer_hashing.html">layer_hashing</a>()</code> <br><code><a href="layer_identity.html">layer_identity</a>()</code> <br><code><a href="layer_integer_lookup.html">layer_integer_lookup</a>()</code> <br><code><a href="layer_jax_model_wrapper.html">layer_jax_model_wrapper</a>()</code> <br><code><a href="layer_lambda.html">layer_lambda</a>()</code> <br><code><a href="layer_layer_normalization.html">layer_layer_normalization</a>()</code> <br><code><a href="layer_lstm.html">layer_lstm</a>()</code> <br><code><a href="layer_masking.html">layer_masking</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_max_pooling_1d.html">layer_max_pooling_1d</a>()</code> <br><code><a href="layer_max_pooling_2d.html">layer_max_pooling_2d</a>()</code> <br><code><a href="layer_max_pooling_3d.html">layer_max_pooling_3d</a>()</code> <br><code><a href="layer_maximum.html">layer_maximum</a>()</code> <br><code><a href="layer_mel_spectrogram.html">layer_mel_spectrogram</a>()</code> <br><code><a href="layer_minimum.html">layer_minimum</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_multi_head_attention.html">layer_multi_head_attention</a>()</code> <br><code><a href="layer_multiply.html">layer_multiply</a>()</code> <br><code><a href="layer_normalization.html">layer_normalization</a>()</code> <br><code><a href="layer_permute.html">layer_permute</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_brightness.html">layer_random_brightness</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_contrast.html">layer_random_contrast</a>()</code> <br><code><a href="layer_random_crop.html">layer_random_crop</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_flip.html">layer_random_flip</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_rotation.html">layer_random_rotation</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_random_translation.html">layer_random_translation</a>()</code> <br><code><a href="layer_random_zoom.html">layer_random_zoom</a>()</code> <br><code><a href="layer_repeat_vector.html">layer_repeat_vector</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_reshape.html">layer_reshape</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_rms_normalization.html">layer_rms_normalization</a>()</code> <br><code><a href="layer_rnn.html">layer_rnn</a>()</code> <br><code><a href="layer_separable_conv_1d.html">layer_separable_conv_1d</a>()</code> <br><code><a href="layer_separable_conv_2d.html">layer_separable_conv_2d</a>()</code> <br><code><a href="layer_simple_rnn.html">layer_simple_rnn</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br><code><a href="layer_spatial_dropout_1d.html">layer_spatial_dropout_1d</a>()</code> <br><code><a href="layer_spatial_dropout_2d.html">layer_spatial_dropout_2d</a>()</code> <br><code><a href="layer_spatial_dropout_3d.html">layer_spatial_dropout_3d</a>()</code> <br><code><a href="layer_spectral_normalization.html">layer_spectral_normalization</a>()</code> <br><code><a href="layer_stft_spectrogram.html">layer_stft_spectrogram</a>()</code> <br><code><a href="layer_string_lookup.html">layer_string_lookup</a>()</code> <br><code><a href="layer_subtract.html">layer_subtract</a>()</code> <br><code><a href="layer_text_vectorization.html">layer_text_vectorization</a>()</code> <br><code><a href="layer_tfsm.html">layer_tfsm</a>()</code> <br><code><a href="layer_time_distributed.html">layer_time_distributed</a>()</code> <br><code><a href="layer_torch_module_wrapper.html">layer_torch_module_wrapper</a>()</code> <br><code><a href="layer_unit_normalization.html">layer_unit_normalization</a>()</code> <br><code><a href="layer_upsampling_1d.html">layer_upsampling_1d</a>()</code> <br><code><a href="layer_upsampling_2d.html">layer_upsampling_2d</a>()</code> <br><code><a href="layer_upsampling_3d.html">layer_upsampling_3d</a>()</code> <br><code><a href="layer_zero_padding_1d.html">layer_zero_padding_1d</a>()</code> <br><code><a href="layer_zero_padding_2d.html">layer_zero_padding_2d</a>()</code> <br><code><a href="layer_zero_padding_3d.html">layer_zero_padding_3d</a>()</code> <br><code><a href="rnn_cell_gru.html">rnn_cell_gru</a>()</code> <br><code><a href="rnn_cell_lstm.html">rnn_cell_lstm</a>()</code> <br><code><a href="rnn_cell_simple.html">rnn_cell_simple</a>()</code> <br><code><a href="rnn_cells_stack.html">rnn_cells_stack</a>()</code> <br></p></div>
+<p>Other layers: <br><code><a href="Layer.html">Layer</a>()</code> <br><code><a href="layer_activation.html">layer_activation</a>()</code> <br><code><a href="layer_activation_elu.html">layer_activation_elu</a>()</code> <br><code><a href="layer_activation_parametric_relu.html">layer_activation_parametric_relu</a>()</code> <br><code><a href="layer_activation_relu.html">layer_activation_relu</a>()</code> <br><code><a href="layer_activation_softmax.html">layer_activation_softmax</a>()</code> <br><code><a href="layer_activity_regularization.html">layer_activity_regularization</a>()</code> <br><code><a href="layer_add.html">layer_add</a>()</code> <br><code><a href="layer_additive_attention.html">layer_additive_attention</a>()</code> <br><code><a href="layer_alpha_dropout.html">layer_alpha_dropout</a>()</code> <br><code><a href="layer_attention.html">layer_attention</a>()</code> <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_average.html">layer_average</a>()</code> <br><code><a href="layer_average_pooling_1d.html">layer_average_pooling_1d</a>()</code> <br><code><a href="layer_average_pooling_2d.html">layer_average_pooling_2d</a>()</code> <br><code><a href="layer_average_pooling_3d.html">layer_average_pooling_3d</a>()</code> <br><code><a href="layer_batch_normalization.html">layer_batch_normalization</a>()</code> <br><code><a href="layer_bidirectional.html">layer_bidirectional</a>()</code> <br><code><a href="layer_category_encoding.html">layer_category_encoding</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_concatenate.html">layer_concatenate</a>()</code> <br><code><a href="layer_conv_1d.html">layer_conv_1d</a>()</code> <br><code><a href="layer_conv_1d_transpose.html">layer_conv_1d_transpose</a>()</code> <br><code><a href="layer_conv_2d.html">layer_conv_2d</a>()</code> <br><code><a href="layer_conv_2d_transpose.html">layer_conv_2d_transpose</a>()</code> <br><code><a href="layer_conv_3d.html">layer_conv_3d</a>()</code> <br><code><a href="layer_conv_3d_transpose.html">layer_conv_3d_transpose</a>()</code> <br><code><a href="layer_conv_lstm_1d.html">layer_conv_lstm_1d</a>()</code> <br><code><a href="layer_conv_lstm_2d.html">layer_conv_lstm_2d</a>()</code> <br><code><a href="layer_conv_lstm_3d.html">layer_conv_lstm_3d</a>()</code> <br><code><a href="layer_cropping_1d.html">layer_cropping_1d</a>()</code> <br><code><a href="layer_cropping_2d.html">layer_cropping_2d</a>()</code> <br><code><a href="layer_cropping_3d.html">layer_cropping_3d</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_dense.html">layer_dense</a>()</code> <br><code><a href="layer_depthwise_conv_1d.html">layer_depthwise_conv_1d</a>()</code> <br><code><a href="layer_depthwise_conv_2d.html">layer_depthwise_conv_2d</a>()</code> <br><code><a href="layer_discretization.html">layer_discretization</a>()</code> <br><code><a href="layer_dot.html">layer_dot</a>()</code> <br><code><a href="layer_dropout.html">layer_dropout</a>()</code> <br><code><a href="layer_einsum_dense.html">layer_einsum_dense</a>()</code> <br><code><a href="layer_embedding.html">layer_embedding</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_feature_space.html">layer_feature_space</a>()</code> <br><code><a href="layer_flatten.html">layer_flatten</a>()</code> <br><code><a href="layer_flax_module_wrapper.html">layer_flax_module_wrapper</a>()</code> <br><code><a href="layer_gaussian_dropout.html">layer_gaussian_dropout</a>()</code> <br><code><a href="layer_gaussian_noise.html">layer_gaussian_noise</a>()</code> <br><code><a href="layer_global_average_pooling_1d.html">layer_global_average_pooling_1d</a>()</code> <br><code><a href="layer_global_average_pooling_2d.html">layer_global_average_pooling_2d</a>()</code> <br><code><a href="layer_global_average_pooling_3d.html">layer_global_average_pooling_3d</a>()</code> <br><code><a href="layer_global_max_pooling_1d.html">layer_global_max_pooling_1d</a>()</code> <br><code><a href="layer_global_max_pooling_2d.html">layer_global_max_pooling_2d</a>()</code> <br><code><a href="layer_global_max_pooling_3d.html">layer_global_max_pooling_3d</a>()</code> <br><code><a href="layer_group_normalization.html">layer_group_normalization</a>()</code> <br><code><a href="layer_group_query_attention.html">layer_group_query_attention</a>()</code> <br><code><a href="layer_gru.html">layer_gru</a>()</code> <br><code><a href="layer_hashed_crossing.html">layer_hashed_crossing</a>()</code> <br><code><a href="layer_hashing.html">layer_hashing</a>()</code> <br><code><a href="layer_identity.html">layer_identity</a>()</code> <br><code><a href="layer_integer_lookup.html">layer_integer_lookup</a>()</code> <br><code><a href="layer_jax_model_wrapper.html">layer_jax_model_wrapper</a>()</code> <br><code><a href="layer_lambda.html">layer_lambda</a>()</code> <br><code><a href="layer_layer_normalization.html">layer_layer_normalization</a>()</code> <br><code><a href="layer_lstm.html">layer_lstm</a>()</code> <br><code><a href="layer_masking.html">layer_masking</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_max_pooling_1d.html">layer_max_pooling_1d</a>()</code> <br><code><a href="layer_max_pooling_2d.html">layer_max_pooling_2d</a>()</code> <br><code><a href="layer_max_pooling_3d.html">layer_max_pooling_3d</a>()</code> <br><code><a href="layer_maximum.html">layer_maximum</a>()</code> <br><code><a href="layer_mel_spectrogram.html">layer_mel_spectrogram</a>()</code> <br><code><a href="layer_minimum.html">layer_minimum</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_multi_head_attention.html">layer_multi_head_attention</a>()</code> <br><code><a href="layer_multiply.html">layer_multiply</a>()</code> <br><code><a href="layer_normalization.html">layer_normalization</a>()</code> <br><code><a href="layer_permute.html">layer_permute</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_brightness.html">layer_random_brightness</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_contrast.html">layer_random_contrast</a>()</code> <br><code><a href="layer_random_crop.html">layer_random_crop</a>()</code> <br><code><a href="layer_random_elastic_transform.html">layer_random_elastic_transform</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_flip.html">layer_random_flip</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_rotation.html">layer_random_rotation</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_random_translation.html">layer_random_translation</a>()</code> <br><code><a href="layer_random_zoom.html">layer_random_zoom</a>()</code> <br><code><a href="layer_repeat_vector.html">layer_repeat_vector</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_reshape.html">layer_reshape</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_rms_normalization.html">layer_rms_normalization</a>()</code> <br><code><a href="layer_rnn.html">layer_rnn</a>()</code> <br><code><a href="layer_separable_conv_1d.html">layer_separable_conv_1d</a>()</code> <br><code><a href="layer_separable_conv_2d.html">layer_separable_conv_2d</a>()</code> <br><code><a href="layer_simple_rnn.html">layer_simple_rnn</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br><code><a href="layer_spatial_dropout_1d.html">layer_spatial_dropout_1d</a>()</code> <br><code><a href="layer_spatial_dropout_2d.html">layer_spatial_dropout_2d</a>()</code> <br><code><a href="layer_spatial_dropout_3d.html">layer_spatial_dropout_3d</a>()</code> <br><code><a href="layer_spectral_normalization.html">layer_spectral_normalization</a>()</code> <br><code><a href="layer_stft_spectrogram.html">layer_stft_spectrogram</a>()</code> <br><code><a href="layer_string_lookup.html">layer_string_lookup</a>()</code> <br><code><a href="layer_subtract.html">layer_subtract</a>()</code> <br><code><a href="layer_text_vectorization.html">layer_text_vectorization</a>()</code> <br><code><a href="layer_tfsm.html">layer_tfsm</a>()</code> <br><code><a href="layer_time_distributed.html">layer_time_distributed</a>()</code> <br><code><a href="layer_torch_module_wrapper.html">layer_torch_module_wrapper</a>()</code> <br><code><a href="layer_unit_normalization.html">layer_unit_normalization</a>()</code> <br><code><a href="layer_upsampling_1d.html">layer_upsampling_1d</a>()</code> <br><code><a href="layer_upsampling_2d.html">layer_upsampling_2d</a>()</code> <br><code><a href="layer_upsampling_3d.html">layer_upsampling_3d</a>()</code> <br><code><a href="layer_zero_padding_1d.html">layer_zero_padding_1d</a>()</code> <br><code><a href="layer_zero_padding_2d.html">layer_zero_padding_2d</a>()</code> <br><code><a href="layer_zero_padding_3d.html">layer_zero_padding_3d</a>()</code> <br><code><a href="rnn_cell_gru.html">rnn_cell_gru</a>()</code> <br><code><a href="rnn_cell_lstm.html">rnn_cell_lstm</a>()</code> <br><code><a href="rnn_cell_simple.html">rnn_cell_simple</a>()</code> <br><code><a href="rnn_cells_stack.html">rnn_cells_stack</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -123,7 +123,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/layer_activation_parametric_relu.html b/docs/dev/reference/layer_activation_parametric_relu.html
index 9a881982f..cbcb6b13d 100644
--- a/docs/dev/reference/layer_activation_parametric_relu.html
+++ b/docs/dev/reference/layer_activation_parametric_relu.html
@@ -14,7 +14,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -124,7 +124,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
     <div class="dont-index">
 <ul><li><p><a href="https://keras.io/api/layers/activation_layers/prelu#prelu-class" class="external-link">https://keras.io/api/layers/activation_layers/prelu#prelu-class</a></p></li>
 </ul><p>Other activation layers: <br><code><a href="layer_activation.html">layer_activation</a>()</code> <br><code><a href="layer_activation_elu.html">layer_activation_elu</a>()</code> <br><code><a href="layer_activation_leaky_relu.html">layer_activation_leaky_relu</a>()</code> <br><code><a href="layer_activation_relu.html">layer_activation_relu</a>()</code> <br><code><a href="layer_activation_softmax.html">layer_activation_softmax</a>()</code> <br></p>
-<p>Other layers: <br><code><a href="Layer.html">Layer</a>()</code> <br><code><a href="layer_activation.html">layer_activation</a>()</code> <br><code><a href="layer_activation_elu.html">layer_activation_elu</a>()</code> <br><code><a href="layer_activation_leaky_relu.html">layer_activation_leaky_relu</a>()</code> <br><code><a href="layer_activation_relu.html">layer_activation_relu</a>()</code> <br><code><a href="layer_activation_softmax.html">layer_activation_softmax</a>()</code> <br><code><a href="layer_activity_regularization.html">layer_activity_regularization</a>()</code> <br><code><a href="layer_add.html">layer_add</a>()</code> <br><code><a href="layer_additive_attention.html">layer_additive_attention</a>()</code> <br><code><a href="layer_alpha_dropout.html">layer_alpha_dropout</a>()</code> <br><code><a href="layer_attention.html">layer_attention</a>()</code> <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_average.html">layer_average</a>()</code> <br><code><a href="layer_average_pooling_1d.html">layer_average_pooling_1d</a>()</code> <br><code><a href="layer_average_pooling_2d.html">layer_average_pooling_2d</a>()</code> <br><code><a href="layer_average_pooling_3d.html">layer_average_pooling_3d</a>()</code> <br><code><a href="layer_batch_normalization.html">layer_batch_normalization</a>()</code> <br><code><a href="layer_bidirectional.html">layer_bidirectional</a>()</code> <br><code><a href="layer_category_encoding.html">layer_category_encoding</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_concatenate.html">layer_concatenate</a>()</code> <br><code><a href="layer_conv_1d.html">layer_conv_1d</a>()</code> <br><code><a href="layer_conv_1d_transpose.html">layer_conv_1d_transpose</a>()</code> <br><code><a href="layer_conv_2d.html">layer_conv_2d</a>()</code> <br><code><a href="layer_conv_2d_transpose.html">layer_conv_2d_transpose</a>()</code> <br><code><a href="layer_conv_3d.html">layer_conv_3d</a>()</code> <br><code><a href="layer_conv_3d_transpose.html">layer_conv_3d_transpose</a>()</code> <br><code><a href="layer_conv_lstm_1d.html">layer_conv_lstm_1d</a>()</code> <br><code><a href="layer_conv_lstm_2d.html">layer_conv_lstm_2d</a>()</code> <br><code><a href="layer_conv_lstm_3d.html">layer_conv_lstm_3d</a>()</code> <br><code><a href="layer_cropping_1d.html">layer_cropping_1d</a>()</code> <br><code><a href="layer_cropping_2d.html">layer_cropping_2d</a>()</code> <br><code><a href="layer_cropping_3d.html">layer_cropping_3d</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_dense.html">layer_dense</a>()</code> <br><code><a href="layer_depthwise_conv_1d.html">layer_depthwise_conv_1d</a>()</code> <br><code><a href="layer_depthwise_conv_2d.html">layer_depthwise_conv_2d</a>()</code> <br><code><a href="layer_discretization.html">layer_discretization</a>()</code> <br><code><a href="layer_dot.html">layer_dot</a>()</code> <br><code><a href="layer_dropout.html">layer_dropout</a>()</code> <br><code><a href="layer_einsum_dense.html">layer_einsum_dense</a>()</code> <br><code><a href="layer_embedding.html">layer_embedding</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_feature_space.html">layer_feature_space</a>()</code> <br><code><a href="layer_flatten.html">layer_flatten</a>()</code> <br><code><a href="layer_flax_module_wrapper.html">layer_flax_module_wrapper</a>()</code> <br><code><a href="layer_gaussian_dropout.html">layer_gaussian_dropout</a>()</code> <br><code><a href="layer_gaussian_noise.html">layer_gaussian_noise</a>()</code> <br><code><a href="layer_global_average_pooling_1d.html">layer_global_average_pooling_1d</a>()</code> <br><code><a href="layer_global_average_pooling_2d.html">layer_global_average_pooling_2d</a>()</code> <br><code><a href="layer_global_average_pooling_3d.html">layer_global_average_pooling_3d</a>()</code> <br><code><a href="layer_global_max_pooling_1d.html">layer_global_max_pooling_1d</a>()</code> <br><code><a href="layer_global_max_pooling_2d.html">layer_global_max_pooling_2d</a>()</code> <br><code><a href="layer_global_max_pooling_3d.html">layer_global_max_pooling_3d</a>()</code> <br><code><a href="layer_group_normalization.html">layer_group_normalization</a>()</code> <br><code><a href="layer_group_query_attention.html">layer_group_query_attention</a>()</code> <br><code><a href="layer_gru.html">layer_gru</a>()</code> <br><code><a href="layer_hashed_crossing.html">layer_hashed_crossing</a>()</code> <br><code><a href="layer_hashing.html">layer_hashing</a>()</code> <br><code><a href="layer_identity.html">layer_identity</a>()</code> <br><code><a href="layer_integer_lookup.html">layer_integer_lookup</a>()</code> <br><code><a href="layer_jax_model_wrapper.html">layer_jax_model_wrapper</a>()</code> <br><code><a href="layer_lambda.html">layer_lambda</a>()</code> <br><code><a href="layer_layer_normalization.html">layer_layer_normalization</a>()</code> <br><code><a href="layer_lstm.html">layer_lstm</a>()</code> <br><code><a href="layer_masking.html">layer_masking</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_max_pooling_1d.html">layer_max_pooling_1d</a>()</code> <br><code><a href="layer_max_pooling_2d.html">layer_max_pooling_2d</a>()</code> <br><code><a href="layer_max_pooling_3d.html">layer_max_pooling_3d</a>()</code> <br><code><a href="layer_maximum.html">layer_maximum</a>()</code> <br><code><a href="layer_mel_spectrogram.html">layer_mel_spectrogram</a>()</code> <br><code><a href="layer_minimum.html">layer_minimum</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_multi_head_attention.html">layer_multi_head_attention</a>()</code> <br><code><a href="layer_multiply.html">layer_multiply</a>()</code> <br><code><a href="layer_normalization.html">layer_normalization</a>()</code> <br><code><a href="layer_permute.html">layer_permute</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_brightness.html">layer_random_brightness</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_contrast.html">layer_random_contrast</a>()</code> <br><code><a href="layer_random_crop.html">layer_random_crop</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_flip.html">layer_random_flip</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_rotation.html">layer_random_rotation</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_random_translation.html">layer_random_translation</a>()</code> <br><code><a href="layer_random_zoom.html">layer_random_zoom</a>()</code> <br><code><a href="layer_repeat_vector.html">layer_repeat_vector</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_reshape.html">layer_reshape</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_rms_normalization.html">layer_rms_normalization</a>()</code> <br><code><a href="layer_rnn.html">layer_rnn</a>()</code> <br><code><a href="layer_separable_conv_1d.html">layer_separable_conv_1d</a>()</code> <br><code><a href="layer_separable_conv_2d.html">layer_separable_conv_2d</a>()</code> <br><code><a href="layer_simple_rnn.html">layer_simple_rnn</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br><code><a href="layer_spatial_dropout_1d.html">layer_spatial_dropout_1d</a>()</code> <br><code><a href="layer_spatial_dropout_2d.html">layer_spatial_dropout_2d</a>()</code> <br><code><a href="layer_spatial_dropout_3d.html">layer_spatial_dropout_3d</a>()</code> <br><code><a href="layer_spectral_normalization.html">layer_spectral_normalization</a>()</code> <br><code><a href="layer_stft_spectrogram.html">layer_stft_spectrogram</a>()</code> <br><code><a href="layer_string_lookup.html">layer_string_lookup</a>()</code> <br><code><a href="layer_subtract.html">layer_subtract</a>()</code> <br><code><a href="layer_text_vectorization.html">layer_text_vectorization</a>()</code> <br><code><a href="layer_tfsm.html">layer_tfsm</a>()</code> <br><code><a href="layer_time_distributed.html">layer_time_distributed</a>()</code> <br><code><a href="layer_torch_module_wrapper.html">layer_torch_module_wrapper</a>()</code> <br><code><a href="layer_unit_normalization.html">layer_unit_normalization</a>()</code> <br><code><a href="layer_upsampling_1d.html">layer_upsampling_1d</a>()</code> <br><code><a href="layer_upsampling_2d.html">layer_upsampling_2d</a>()</code> <br><code><a href="layer_upsampling_3d.html">layer_upsampling_3d</a>()</code> <br><code><a href="layer_zero_padding_1d.html">layer_zero_padding_1d</a>()</code> <br><code><a href="layer_zero_padding_2d.html">layer_zero_padding_2d</a>()</code> <br><code><a href="layer_zero_padding_3d.html">layer_zero_padding_3d</a>()</code> <br><code><a href="rnn_cell_gru.html">rnn_cell_gru</a>()</code> <br><code><a href="rnn_cell_lstm.html">rnn_cell_lstm</a>()</code> <br><code><a href="rnn_cell_simple.html">rnn_cell_simple</a>()</code> <br><code><a href="rnn_cells_stack.html">rnn_cells_stack</a>()</code> <br></p></div>
+<p>Other layers: <br><code><a href="Layer.html">Layer</a>()</code> <br><code><a href="layer_activation.html">layer_activation</a>()</code> <br><code><a href="layer_activation_elu.html">layer_activation_elu</a>()</code> <br><code><a href="layer_activation_leaky_relu.html">layer_activation_leaky_relu</a>()</code> <br><code><a href="layer_activation_relu.html">layer_activation_relu</a>()</code> <br><code><a href="layer_activation_softmax.html">layer_activation_softmax</a>()</code> <br><code><a href="layer_activity_regularization.html">layer_activity_regularization</a>()</code> <br><code><a href="layer_add.html">layer_add</a>()</code> <br><code><a href="layer_additive_attention.html">layer_additive_attention</a>()</code> <br><code><a href="layer_alpha_dropout.html">layer_alpha_dropout</a>()</code> <br><code><a href="layer_attention.html">layer_attention</a>()</code> <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_average.html">layer_average</a>()</code> <br><code><a href="layer_average_pooling_1d.html">layer_average_pooling_1d</a>()</code> <br><code><a href="layer_average_pooling_2d.html">layer_average_pooling_2d</a>()</code> <br><code><a href="layer_average_pooling_3d.html">layer_average_pooling_3d</a>()</code> <br><code><a href="layer_batch_normalization.html">layer_batch_normalization</a>()</code> <br><code><a href="layer_bidirectional.html">layer_bidirectional</a>()</code> <br><code><a href="layer_category_encoding.html">layer_category_encoding</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_concatenate.html">layer_concatenate</a>()</code> <br><code><a href="layer_conv_1d.html">layer_conv_1d</a>()</code> <br><code><a href="layer_conv_1d_transpose.html">layer_conv_1d_transpose</a>()</code> <br><code><a href="layer_conv_2d.html">layer_conv_2d</a>()</code> <br><code><a href="layer_conv_2d_transpose.html">layer_conv_2d_transpose</a>()</code> <br><code><a href="layer_conv_3d.html">layer_conv_3d</a>()</code> <br><code><a href="layer_conv_3d_transpose.html">layer_conv_3d_transpose</a>()</code> <br><code><a href="layer_conv_lstm_1d.html">layer_conv_lstm_1d</a>()</code> <br><code><a href="layer_conv_lstm_2d.html">layer_conv_lstm_2d</a>()</code> <br><code><a href="layer_conv_lstm_3d.html">layer_conv_lstm_3d</a>()</code> <br><code><a href="layer_cropping_1d.html">layer_cropping_1d</a>()</code> <br><code><a href="layer_cropping_2d.html">layer_cropping_2d</a>()</code> <br><code><a href="layer_cropping_3d.html">layer_cropping_3d</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_dense.html">layer_dense</a>()</code> <br><code><a href="layer_depthwise_conv_1d.html">layer_depthwise_conv_1d</a>()</code> <br><code><a href="layer_depthwise_conv_2d.html">layer_depthwise_conv_2d</a>()</code> <br><code><a href="layer_discretization.html">layer_discretization</a>()</code> <br><code><a href="layer_dot.html">layer_dot</a>()</code> <br><code><a href="layer_dropout.html">layer_dropout</a>()</code> <br><code><a href="layer_einsum_dense.html">layer_einsum_dense</a>()</code> <br><code><a href="layer_embedding.html">layer_embedding</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_feature_space.html">layer_feature_space</a>()</code> <br><code><a href="layer_flatten.html">layer_flatten</a>()</code> <br><code><a href="layer_flax_module_wrapper.html">layer_flax_module_wrapper</a>()</code> <br><code><a href="layer_gaussian_dropout.html">layer_gaussian_dropout</a>()</code> <br><code><a href="layer_gaussian_noise.html">layer_gaussian_noise</a>()</code> <br><code><a href="layer_global_average_pooling_1d.html">layer_global_average_pooling_1d</a>()</code> <br><code><a href="layer_global_average_pooling_2d.html">layer_global_average_pooling_2d</a>()</code> <br><code><a href="layer_global_average_pooling_3d.html">layer_global_average_pooling_3d</a>()</code> <br><code><a href="layer_global_max_pooling_1d.html">layer_global_max_pooling_1d</a>()</code> <br><code><a href="layer_global_max_pooling_2d.html">layer_global_max_pooling_2d</a>()</code> <br><code><a href="layer_global_max_pooling_3d.html">layer_global_max_pooling_3d</a>()</code> <br><code><a href="layer_group_normalization.html">layer_group_normalization</a>()</code> <br><code><a href="layer_group_query_attention.html">layer_group_query_attention</a>()</code> <br><code><a href="layer_gru.html">layer_gru</a>()</code> <br><code><a href="layer_hashed_crossing.html">layer_hashed_crossing</a>()</code> <br><code><a href="layer_hashing.html">layer_hashing</a>()</code> <br><code><a href="layer_identity.html">layer_identity</a>()</code> <br><code><a href="layer_integer_lookup.html">layer_integer_lookup</a>()</code> <br><code><a href="layer_jax_model_wrapper.html">layer_jax_model_wrapper</a>()</code> <br><code><a href="layer_lambda.html">layer_lambda</a>()</code> <br><code><a href="layer_layer_normalization.html">layer_layer_normalization</a>()</code> <br><code><a href="layer_lstm.html">layer_lstm</a>()</code> <br><code><a href="layer_masking.html">layer_masking</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_max_pooling_1d.html">layer_max_pooling_1d</a>()</code> <br><code><a href="layer_max_pooling_2d.html">layer_max_pooling_2d</a>()</code> <br><code><a href="layer_max_pooling_3d.html">layer_max_pooling_3d</a>()</code> <br><code><a href="layer_maximum.html">layer_maximum</a>()</code> <br><code><a href="layer_mel_spectrogram.html">layer_mel_spectrogram</a>()</code> <br><code><a href="layer_minimum.html">layer_minimum</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_multi_head_attention.html">layer_multi_head_attention</a>()</code> <br><code><a href="layer_multiply.html">layer_multiply</a>()</code> <br><code><a href="layer_normalization.html">layer_normalization</a>()</code> <br><code><a href="layer_permute.html">layer_permute</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_brightness.html">layer_random_brightness</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_contrast.html">layer_random_contrast</a>()</code> <br><code><a href="layer_random_crop.html">layer_random_crop</a>()</code> <br><code><a href="layer_random_elastic_transform.html">layer_random_elastic_transform</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_flip.html">layer_random_flip</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_rotation.html">layer_random_rotation</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_random_translation.html">layer_random_translation</a>()</code> <br><code><a href="layer_random_zoom.html">layer_random_zoom</a>()</code> <br><code><a href="layer_repeat_vector.html">layer_repeat_vector</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_reshape.html">layer_reshape</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_rms_normalization.html">layer_rms_normalization</a>()</code> <br><code><a href="layer_rnn.html">layer_rnn</a>()</code> <br><code><a href="layer_separable_conv_1d.html">layer_separable_conv_1d</a>()</code> <br><code><a href="layer_separable_conv_2d.html">layer_separable_conv_2d</a>()</code> <br><code><a href="layer_simple_rnn.html">layer_simple_rnn</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br><code><a href="layer_spatial_dropout_1d.html">layer_spatial_dropout_1d</a>()</code> <br><code><a href="layer_spatial_dropout_2d.html">layer_spatial_dropout_2d</a>()</code> <br><code><a href="layer_spatial_dropout_3d.html">layer_spatial_dropout_3d</a>()</code> <br><code><a href="layer_spectral_normalization.html">layer_spectral_normalization</a>()</code> <br><code><a href="layer_stft_spectrogram.html">layer_stft_spectrogram</a>()</code> <br><code><a href="layer_string_lookup.html">layer_string_lookup</a>()</code> <br><code><a href="layer_subtract.html">layer_subtract</a>()</code> <br><code><a href="layer_text_vectorization.html">layer_text_vectorization</a>()</code> <br><code><a href="layer_tfsm.html">layer_tfsm</a>()</code> <br><code><a href="layer_time_distributed.html">layer_time_distributed</a>()</code> <br><code><a href="layer_torch_module_wrapper.html">layer_torch_module_wrapper</a>()</code> <br><code><a href="layer_unit_normalization.html">layer_unit_normalization</a>()</code> <br><code><a href="layer_upsampling_1d.html">layer_upsampling_1d</a>()</code> <br><code><a href="layer_upsampling_2d.html">layer_upsampling_2d</a>()</code> <br><code><a href="layer_upsampling_3d.html">layer_upsampling_3d</a>()</code> <br><code><a href="layer_zero_padding_1d.html">layer_zero_padding_1d</a>()</code> <br><code><a href="layer_zero_padding_2d.html">layer_zero_padding_2d</a>()</code> <br><code><a href="layer_zero_padding_3d.html">layer_zero_padding_3d</a>()</code> <br><code><a href="rnn_cell_gru.html">rnn_cell_gru</a>()</code> <br><code><a href="rnn_cell_lstm.html">rnn_cell_lstm</a>()</code> <br><code><a href="rnn_cell_simple.html">rnn_cell_simple</a>()</code> <br><code><a href="rnn_cells_stack.html">rnn_cells_stack</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -136,7 +136,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/layer_activation_relu.html b/docs/dev/reference/layer_activation_relu.html
index d2f62ca6d..48b32ab69 100644
--- a/docs/dev/reference/layer_activation_relu.html
+++ b/docs/dev/reference/layer_activation_relu.html
@@ -28,7 +28,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -149,7 +149,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
     <div class="dont-index">
 <ul><li><p><a href="https://keras.io/api/layers/activation_layers/relu#relu-class" class="external-link">https://keras.io/api/layers/activation_layers/relu#relu-class</a></p></li>
 </ul><p>Other activation layers: <br><code><a href="layer_activation.html">layer_activation</a>()</code> <br><code><a href="layer_activation_elu.html">layer_activation_elu</a>()</code> <br><code><a href="layer_activation_leaky_relu.html">layer_activation_leaky_relu</a>()</code> <br><code><a href="layer_activation_parametric_relu.html">layer_activation_parametric_relu</a>()</code> <br><code><a href="layer_activation_softmax.html">layer_activation_softmax</a>()</code> <br></p>
-<p>Other layers: <br><code><a href="Layer.html">Layer</a>()</code> <br><code><a href="layer_activation.html">layer_activation</a>()</code> <br><code><a href="layer_activation_elu.html">layer_activation_elu</a>()</code> <br><code><a href="layer_activation_leaky_relu.html">layer_activation_leaky_relu</a>()</code> <br><code><a href="layer_activation_parametric_relu.html">layer_activation_parametric_relu</a>()</code> <br><code><a href="layer_activation_softmax.html">layer_activation_softmax</a>()</code> <br><code><a href="layer_activity_regularization.html">layer_activity_regularization</a>()</code> <br><code><a href="layer_add.html">layer_add</a>()</code> <br><code><a href="layer_additive_attention.html">layer_additive_attention</a>()</code> <br><code><a href="layer_alpha_dropout.html">layer_alpha_dropout</a>()</code> <br><code><a href="layer_attention.html">layer_attention</a>()</code> <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_average.html">layer_average</a>()</code> <br><code><a href="layer_average_pooling_1d.html">layer_average_pooling_1d</a>()</code> <br><code><a href="layer_average_pooling_2d.html">layer_average_pooling_2d</a>()</code> <br><code><a href="layer_average_pooling_3d.html">layer_average_pooling_3d</a>()</code> <br><code><a href="layer_batch_normalization.html">layer_batch_normalization</a>()</code> <br><code><a href="layer_bidirectional.html">layer_bidirectional</a>()</code> <br><code><a href="layer_category_encoding.html">layer_category_encoding</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_concatenate.html">layer_concatenate</a>()</code> <br><code><a href="layer_conv_1d.html">layer_conv_1d</a>()</code> <br><code><a href="layer_conv_1d_transpose.html">layer_conv_1d_transpose</a>()</code> <br><code><a href="layer_conv_2d.html">layer_conv_2d</a>()</code> <br><code><a href="layer_conv_2d_transpose.html">layer_conv_2d_transpose</a>()</code> <br><code><a href="layer_conv_3d.html">layer_conv_3d</a>()</code> <br><code><a href="layer_conv_3d_transpose.html">layer_conv_3d_transpose</a>()</code> <br><code><a href="layer_conv_lstm_1d.html">layer_conv_lstm_1d</a>()</code> <br><code><a href="layer_conv_lstm_2d.html">layer_conv_lstm_2d</a>()</code> <br><code><a href="layer_conv_lstm_3d.html">layer_conv_lstm_3d</a>()</code> <br><code><a href="layer_cropping_1d.html">layer_cropping_1d</a>()</code> <br><code><a href="layer_cropping_2d.html">layer_cropping_2d</a>()</code> <br><code><a href="layer_cropping_3d.html">layer_cropping_3d</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_dense.html">layer_dense</a>()</code> <br><code><a href="layer_depthwise_conv_1d.html">layer_depthwise_conv_1d</a>()</code> <br><code><a href="layer_depthwise_conv_2d.html">layer_depthwise_conv_2d</a>()</code> <br><code><a href="layer_discretization.html">layer_discretization</a>()</code> <br><code><a href="layer_dot.html">layer_dot</a>()</code> <br><code><a href="layer_dropout.html">layer_dropout</a>()</code> <br><code><a href="layer_einsum_dense.html">layer_einsum_dense</a>()</code> <br><code><a href="layer_embedding.html">layer_embedding</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_feature_space.html">layer_feature_space</a>()</code> <br><code><a href="layer_flatten.html">layer_flatten</a>()</code> <br><code><a href="layer_flax_module_wrapper.html">layer_flax_module_wrapper</a>()</code> <br><code><a href="layer_gaussian_dropout.html">layer_gaussian_dropout</a>()</code> <br><code><a href="layer_gaussian_noise.html">layer_gaussian_noise</a>()</code> <br><code><a href="layer_global_average_pooling_1d.html">layer_global_average_pooling_1d</a>()</code> <br><code><a href="layer_global_average_pooling_2d.html">layer_global_average_pooling_2d</a>()</code> <br><code><a href="layer_global_average_pooling_3d.html">layer_global_average_pooling_3d</a>()</code> <br><code><a href="layer_global_max_pooling_1d.html">layer_global_max_pooling_1d</a>()</code> <br><code><a href="layer_global_max_pooling_2d.html">layer_global_max_pooling_2d</a>()</code> <br><code><a href="layer_global_max_pooling_3d.html">layer_global_max_pooling_3d</a>()</code> <br><code><a href="layer_group_normalization.html">layer_group_normalization</a>()</code> <br><code><a href="layer_group_query_attention.html">layer_group_query_attention</a>()</code> <br><code><a href="layer_gru.html">layer_gru</a>()</code> <br><code><a href="layer_hashed_crossing.html">layer_hashed_crossing</a>()</code> <br><code><a href="layer_hashing.html">layer_hashing</a>()</code> <br><code><a href="layer_identity.html">layer_identity</a>()</code> <br><code><a href="layer_integer_lookup.html">layer_integer_lookup</a>()</code> <br><code><a href="layer_jax_model_wrapper.html">layer_jax_model_wrapper</a>()</code> <br><code><a href="layer_lambda.html">layer_lambda</a>()</code> <br><code><a href="layer_layer_normalization.html">layer_layer_normalization</a>()</code> <br><code><a href="layer_lstm.html">layer_lstm</a>()</code> <br><code><a href="layer_masking.html">layer_masking</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_max_pooling_1d.html">layer_max_pooling_1d</a>()</code> <br><code><a href="layer_max_pooling_2d.html">layer_max_pooling_2d</a>()</code> <br><code><a href="layer_max_pooling_3d.html">layer_max_pooling_3d</a>()</code> <br><code><a href="layer_maximum.html">layer_maximum</a>()</code> <br><code><a href="layer_mel_spectrogram.html">layer_mel_spectrogram</a>()</code> <br><code><a href="layer_minimum.html">layer_minimum</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_multi_head_attention.html">layer_multi_head_attention</a>()</code> <br><code><a href="layer_multiply.html">layer_multiply</a>()</code> <br><code><a href="layer_normalization.html">layer_normalization</a>()</code> <br><code><a href="layer_permute.html">layer_permute</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_brightness.html">layer_random_brightness</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_contrast.html">layer_random_contrast</a>()</code> <br><code><a href="layer_random_crop.html">layer_random_crop</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_flip.html">layer_random_flip</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_rotation.html">layer_random_rotation</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_random_translation.html">layer_random_translation</a>()</code> <br><code><a href="layer_random_zoom.html">layer_random_zoom</a>()</code> <br><code><a href="layer_repeat_vector.html">layer_repeat_vector</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_reshape.html">layer_reshape</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_rms_normalization.html">layer_rms_normalization</a>()</code> <br><code><a href="layer_rnn.html">layer_rnn</a>()</code> <br><code><a href="layer_separable_conv_1d.html">layer_separable_conv_1d</a>()</code> <br><code><a href="layer_separable_conv_2d.html">layer_separable_conv_2d</a>()</code> <br><code><a href="layer_simple_rnn.html">layer_simple_rnn</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br><code><a href="layer_spatial_dropout_1d.html">layer_spatial_dropout_1d</a>()</code> <br><code><a href="layer_spatial_dropout_2d.html">layer_spatial_dropout_2d</a>()</code> <br><code><a href="layer_spatial_dropout_3d.html">layer_spatial_dropout_3d</a>()</code> <br><code><a href="layer_spectral_normalization.html">layer_spectral_normalization</a>()</code> <br><code><a href="layer_stft_spectrogram.html">layer_stft_spectrogram</a>()</code> <br><code><a href="layer_string_lookup.html">layer_string_lookup</a>()</code> <br><code><a href="layer_subtract.html">layer_subtract</a>()</code> <br><code><a href="layer_text_vectorization.html">layer_text_vectorization</a>()</code> <br><code><a href="layer_tfsm.html">layer_tfsm</a>()</code> <br><code><a href="layer_time_distributed.html">layer_time_distributed</a>()</code> <br><code><a href="layer_torch_module_wrapper.html">layer_torch_module_wrapper</a>()</code> <br><code><a href="layer_unit_normalization.html">layer_unit_normalization</a>()</code> <br><code><a href="layer_upsampling_1d.html">layer_upsampling_1d</a>()</code> <br><code><a href="layer_upsampling_2d.html">layer_upsampling_2d</a>()</code> <br><code><a href="layer_upsampling_3d.html">layer_upsampling_3d</a>()</code> <br><code><a href="layer_zero_padding_1d.html">layer_zero_padding_1d</a>()</code> <br><code><a href="layer_zero_padding_2d.html">layer_zero_padding_2d</a>()</code> <br><code><a href="layer_zero_padding_3d.html">layer_zero_padding_3d</a>()</code> <br><code><a href="rnn_cell_gru.html">rnn_cell_gru</a>()</code> <br><code><a href="rnn_cell_lstm.html">rnn_cell_lstm</a>()</code> <br><code><a href="rnn_cell_simple.html">rnn_cell_simple</a>()</code> <br><code><a href="rnn_cells_stack.html">rnn_cells_stack</a>()</code> <br></p></div>
+<p>Other layers: <br><code><a href="Layer.html">Layer</a>()</code> <br><code><a href="layer_activation.html">layer_activation</a>()</code> <br><code><a href="layer_activation_elu.html">layer_activation_elu</a>()</code> <br><code><a href="layer_activation_leaky_relu.html">layer_activation_leaky_relu</a>()</code> <br><code><a href="layer_activation_parametric_relu.html">layer_activation_parametric_relu</a>()</code> <br><code><a href="layer_activation_softmax.html">layer_activation_softmax</a>()</code> <br><code><a href="layer_activity_regularization.html">layer_activity_regularization</a>()</code> <br><code><a href="layer_add.html">layer_add</a>()</code> <br><code><a href="layer_additive_attention.html">layer_additive_attention</a>()</code> <br><code><a href="layer_alpha_dropout.html">layer_alpha_dropout</a>()</code> <br><code><a href="layer_attention.html">layer_attention</a>()</code> <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_average.html">layer_average</a>()</code> <br><code><a href="layer_average_pooling_1d.html">layer_average_pooling_1d</a>()</code> <br><code><a href="layer_average_pooling_2d.html">layer_average_pooling_2d</a>()</code> <br><code><a href="layer_average_pooling_3d.html">layer_average_pooling_3d</a>()</code> <br><code><a href="layer_batch_normalization.html">layer_batch_normalization</a>()</code> <br><code><a href="layer_bidirectional.html">layer_bidirectional</a>()</code> <br><code><a href="layer_category_encoding.html">layer_category_encoding</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_concatenate.html">layer_concatenate</a>()</code> <br><code><a href="layer_conv_1d.html">layer_conv_1d</a>()</code> <br><code><a href="layer_conv_1d_transpose.html">layer_conv_1d_transpose</a>()</code> <br><code><a href="layer_conv_2d.html">layer_conv_2d</a>()</code> <br><code><a href="layer_conv_2d_transpose.html">layer_conv_2d_transpose</a>()</code> <br><code><a href="layer_conv_3d.html">layer_conv_3d</a>()</code> <br><code><a href="layer_conv_3d_transpose.html">layer_conv_3d_transpose</a>()</code> <br><code><a href="layer_conv_lstm_1d.html">layer_conv_lstm_1d</a>()</code> <br><code><a href="layer_conv_lstm_2d.html">layer_conv_lstm_2d</a>()</code> <br><code><a href="layer_conv_lstm_3d.html">layer_conv_lstm_3d</a>()</code> <br><code><a href="layer_cropping_1d.html">layer_cropping_1d</a>()</code> <br><code><a href="layer_cropping_2d.html">layer_cropping_2d</a>()</code> <br><code><a href="layer_cropping_3d.html">layer_cropping_3d</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_dense.html">layer_dense</a>()</code> <br><code><a href="layer_depthwise_conv_1d.html">layer_depthwise_conv_1d</a>()</code> <br><code><a href="layer_depthwise_conv_2d.html">layer_depthwise_conv_2d</a>()</code> <br><code><a href="layer_discretization.html">layer_discretization</a>()</code> <br><code><a href="layer_dot.html">layer_dot</a>()</code> <br><code><a href="layer_dropout.html">layer_dropout</a>()</code> <br><code><a href="layer_einsum_dense.html">layer_einsum_dense</a>()</code> <br><code><a href="layer_embedding.html">layer_embedding</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_feature_space.html">layer_feature_space</a>()</code> <br><code><a href="layer_flatten.html">layer_flatten</a>()</code> <br><code><a href="layer_flax_module_wrapper.html">layer_flax_module_wrapper</a>()</code> <br><code><a href="layer_gaussian_dropout.html">layer_gaussian_dropout</a>()</code> <br><code><a href="layer_gaussian_noise.html">layer_gaussian_noise</a>()</code> <br><code><a href="layer_global_average_pooling_1d.html">layer_global_average_pooling_1d</a>()</code> <br><code><a href="layer_global_average_pooling_2d.html">layer_global_average_pooling_2d</a>()</code> <br><code><a href="layer_global_average_pooling_3d.html">layer_global_average_pooling_3d</a>()</code> <br><code><a href="layer_global_max_pooling_1d.html">layer_global_max_pooling_1d</a>()</code> <br><code><a href="layer_global_max_pooling_2d.html">layer_global_max_pooling_2d</a>()</code> <br><code><a href="layer_global_max_pooling_3d.html">layer_global_max_pooling_3d</a>()</code> <br><code><a href="layer_group_normalization.html">layer_group_normalization</a>()</code> <br><code><a href="layer_group_query_attention.html">layer_group_query_attention</a>()</code> <br><code><a href="layer_gru.html">layer_gru</a>()</code> <br><code><a href="layer_hashed_crossing.html">layer_hashed_crossing</a>()</code> <br><code><a href="layer_hashing.html">layer_hashing</a>()</code> <br><code><a href="layer_identity.html">layer_identity</a>()</code> <br><code><a href="layer_integer_lookup.html">layer_integer_lookup</a>()</code> <br><code><a href="layer_jax_model_wrapper.html">layer_jax_model_wrapper</a>()</code> <br><code><a href="layer_lambda.html">layer_lambda</a>()</code> <br><code><a href="layer_layer_normalization.html">layer_layer_normalization</a>()</code> <br><code><a href="layer_lstm.html">layer_lstm</a>()</code> <br><code><a href="layer_masking.html">layer_masking</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_max_pooling_1d.html">layer_max_pooling_1d</a>()</code> <br><code><a href="layer_max_pooling_2d.html">layer_max_pooling_2d</a>()</code> <br><code><a href="layer_max_pooling_3d.html">layer_max_pooling_3d</a>()</code> <br><code><a href="layer_maximum.html">layer_maximum</a>()</code> <br><code><a href="layer_mel_spectrogram.html">layer_mel_spectrogram</a>()</code> <br><code><a href="layer_minimum.html">layer_minimum</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_multi_head_attention.html">layer_multi_head_attention</a>()</code> <br><code><a href="layer_multiply.html">layer_multiply</a>()</code> <br><code><a href="layer_normalization.html">layer_normalization</a>()</code> <br><code><a href="layer_permute.html">layer_permute</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_brightness.html">layer_random_brightness</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_contrast.html">layer_random_contrast</a>()</code> <br><code><a href="layer_random_crop.html">layer_random_crop</a>()</code> <br><code><a href="layer_random_elastic_transform.html">layer_random_elastic_transform</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_flip.html">layer_random_flip</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_rotation.html">layer_random_rotation</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_random_translation.html">layer_random_translation</a>()</code> <br><code><a href="layer_random_zoom.html">layer_random_zoom</a>()</code> <br><code><a href="layer_repeat_vector.html">layer_repeat_vector</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_reshape.html">layer_reshape</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_rms_normalization.html">layer_rms_normalization</a>()</code> <br><code><a href="layer_rnn.html">layer_rnn</a>()</code> <br><code><a href="layer_separable_conv_1d.html">layer_separable_conv_1d</a>()</code> <br><code><a href="layer_separable_conv_2d.html">layer_separable_conv_2d</a>()</code> <br><code><a href="layer_simple_rnn.html">layer_simple_rnn</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br><code><a href="layer_spatial_dropout_1d.html">layer_spatial_dropout_1d</a>()</code> <br><code><a href="layer_spatial_dropout_2d.html">layer_spatial_dropout_2d</a>()</code> <br><code><a href="layer_spatial_dropout_3d.html">layer_spatial_dropout_3d</a>()</code> <br><code><a href="layer_spectral_normalization.html">layer_spectral_normalization</a>()</code> <br><code><a href="layer_stft_spectrogram.html">layer_stft_spectrogram</a>()</code> <br><code><a href="layer_string_lookup.html">layer_string_lookup</a>()</code> <br><code><a href="layer_subtract.html">layer_subtract</a>()</code> <br><code><a href="layer_text_vectorization.html">layer_text_vectorization</a>()</code> <br><code><a href="layer_tfsm.html">layer_tfsm</a>()</code> <br><code><a href="layer_time_distributed.html">layer_time_distributed</a>()</code> <br><code><a href="layer_torch_module_wrapper.html">layer_torch_module_wrapper</a>()</code> <br><code><a href="layer_unit_normalization.html">layer_unit_normalization</a>()</code> <br><code><a href="layer_upsampling_1d.html">layer_upsampling_1d</a>()</code> <br><code><a href="layer_upsampling_2d.html">layer_upsampling_2d</a>()</code> <br><code><a href="layer_upsampling_3d.html">layer_upsampling_3d</a>()</code> <br><code><a href="layer_zero_padding_1d.html">layer_zero_padding_1d</a>()</code> <br><code><a href="layer_zero_padding_2d.html">layer_zero_padding_2d</a>()</code> <br><code><a href="layer_zero_padding_3d.html">layer_zero_padding_3d</a>()</code> <br><code><a href="rnn_cell_gru.html">rnn_cell_gru</a>()</code> <br><code><a href="rnn_cell_lstm.html">rnn_cell_lstm</a>()</code> <br><code><a href="rnn_cell_simple.html">rnn_cell_simple</a>()</code> <br><code><a href="rnn_cells_stack.html">rnn_cells_stack</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -161,7 +161,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/layer_activation_softmax.html b/docs/dev/reference/layer_activation_softmax.html
index c8661247d..065973332 100644
--- a/docs/dev/reference/layer_activation_softmax.html
+++ b/docs/dev/reference/layer_activation_softmax.html
@@ -14,7 +14,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -111,7 +111,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
     <div class="dont-index">
 <ul><li><p><a href="https://keras.io/api/layers/activation_layers/softmax#softmax-class" class="external-link">https://keras.io/api/layers/activation_layers/softmax#softmax-class</a></p></li>
 </ul><p>Other activation layers: <br><code><a href="layer_activation.html">layer_activation</a>()</code> <br><code><a href="layer_activation_elu.html">layer_activation_elu</a>()</code> <br><code><a href="layer_activation_leaky_relu.html">layer_activation_leaky_relu</a>()</code> <br><code><a href="layer_activation_parametric_relu.html">layer_activation_parametric_relu</a>()</code> <br><code><a href="layer_activation_relu.html">layer_activation_relu</a>()</code> <br></p>
-<p>Other layers: <br><code><a href="Layer.html">Layer</a>()</code> <br><code><a href="layer_activation.html">layer_activation</a>()</code> <br><code><a href="layer_activation_elu.html">layer_activation_elu</a>()</code> <br><code><a href="layer_activation_leaky_relu.html">layer_activation_leaky_relu</a>()</code> <br><code><a href="layer_activation_parametric_relu.html">layer_activation_parametric_relu</a>()</code> <br><code><a href="layer_activation_relu.html">layer_activation_relu</a>()</code> <br><code><a href="layer_activity_regularization.html">layer_activity_regularization</a>()</code> <br><code><a href="layer_add.html">layer_add</a>()</code> <br><code><a href="layer_additive_attention.html">layer_additive_attention</a>()</code> <br><code><a href="layer_alpha_dropout.html">layer_alpha_dropout</a>()</code> <br><code><a href="layer_attention.html">layer_attention</a>()</code> <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_average.html">layer_average</a>()</code> <br><code><a href="layer_average_pooling_1d.html">layer_average_pooling_1d</a>()</code> <br><code><a href="layer_average_pooling_2d.html">layer_average_pooling_2d</a>()</code> <br><code><a href="layer_average_pooling_3d.html">layer_average_pooling_3d</a>()</code> <br><code><a href="layer_batch_normalization.html">layer_batch_normalization</a>()</code> <br><code><a href="layer_bidirectional.html">layer_bidirectional</a>()</code> <br><code><a href="layer_category_encoding.html">layer_category_encoding</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_concatenate.html">layer_concatenate</a>()</code> <br><code><a href="layer_conv_1d.html">layer_conv_1d</a>()</code> <br><code><a href="layer_conv_1d_transpose.html">layer_conv_1d_transpose</a>()</code> <br><code><a href="layer_conv_2d.html">layer_conv_2d</a>()</code> <br><code><a href="layer_conv_2d_transpose.html">layer_conv_2d_transpose</a>()</code> <br><code><a href="layer_conv_3d.html">layer_conv_3d</a>()</code> <br><code><a href="layer_conv_3d_transpose.html">layer_conv_3d_transpose</a>()</code> <br><code><a href="layer_conv_lstm_1d.html">layer_conv_lstm_1d</a>()</code> <br><code><a href="layer_conv_lstm_2d.html">layer_conv_lstm_2d</a>()</code> <br><code><a href="layer_conv_lstm_3d.html">layer_conv_lstm_3d</a>()</code> <br><code><a href="layer_cropping_1d.html">layer_cropping_1d</a>()</code> <br><code><a href="layer_cropping_2d.html">layer_cropping_2d</a>()</code> <br><code><a href="layer_cropping_3d.html">layer_cropping_3d</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_dense.html">layer_dense</a>()</code> <br><code><a href="layer_depthwise_conv_1d.html">layer_depthwise_conv_1d</a>()</code> <br><code><a href="layer_depthwise_conv_2d.html">layer_depthwise_conv_2d</a>()</code> <br><code><a href="layer_discretization.html">layer_discretization</a>()</code> <br><code><a href="layer_dot.html">layer_dot</a>()</code> <br><code><a href="layer_dropout.html">layer_dropout</a>()</code> <br><code><a href="layer_einsum_dense.html">layer_einsum_dense</a>()</code> <br><code><a href="layer_embedding.html">layer_embedding</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_feature_space.html">layer_feature_space</a>()</code> <br><code><a href="layer_flatten.html">layer_flatten</a>()</code> <br><code><a href="layer_flax_module_wrapper.html">layer_flax_module_wrapper</a>()</code> <br><code><a href="layer_gaussian_dropout.html">layer_gaussian_dropout</a>()</code> <br><code><a href="layer_gaussian_noise.html">layer_gaussian_noise</a>()</code> <br><code><a href="layer_global_average_pooling_1d.html">layer_global_average_pooling_1d</a>()</code> <br><code><a href="layer_global_average_pooling_2d.html">layer_global_average_pooling_2d</a>()</code> <br><code><a href="layer_global_average_pooling_3d.html">layer_global_average_pooling_3d</a>()</code> <br><code><a href="layer_global_max_pooling_1d.html">layer_global_max_pooling_1d</a>()</code> <br><code><a href="layer_global_max_pooling_2d.html">layer_global_max_pooling_2d</a>()</code> <br><code><a href="layer_global_max_pooling_3d.html">layer_global_max_pooling_3d</a>()</code> <br><code><a href="layer_group_normalization.html">layer_group_normalization</a>()</code> <br><code><a href="layer_group_query_attention.html">layer_group_query_attention</a>()</code> <br><code><a href="layer_gru.html">layer_gru</a>()</code> <br><code><a href="layer_hashed_crossing.html">layer_hashed_crossing</a>()</code> <br><code><a href="layer_hashing.html">layer_hashing</a>()</code> <br><code><a href="layer_identity.html">layer_identity</a>()</code> <br><code><a href="layer_integer_lookup.html">layer_integer_lookup</a>()</code> <br><code><a href="layer_jax_model_wrapper.html">layer_jax_model_wrapper</a>()</code> <br><code><a href="layer_lambda.html">layer_lambda</a>()</code> <br><code><a href="layer_layer_normalization.html">layer_layer_normalization</a>()</code> <br><code><a href="layer_lstm.html">layer_lstm</a>()</code> <br><code><a href="layer_masking.html">layer_masking</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_max_pooling_1d.html">layer_max_pooling_1d</a>()</code> <br><code><a href="layer_max_pooling_2d.html">layer_max_pooling_2d</a>()</code> <br><code><a href="layer_max_pooling_3d.html">layer_max_pooling_3d</a>()</code> <br><code><a href="layer_maximum.html">layer_maximum</a>()</code> <br><code><a href="layer_mel_spectrogram.html">layer_mel_spectrogram</a>()</code> <br><code><a href="layer_minimum.html">layer_minimum</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_multi_head_attention.html">layer_multi_head_attention</a>()</code> <br><code><a href="layer_multiply.html">layer_multiply</a>()</code> <br><code><a href="layer_normalization.html">layer_normalization</a>()</code> <br><code><a href="layer_permute.html">layer_permute</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_brightness.html">layer_random_brightness</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_contrast.html">layer_random_contrast</a>()</code> <br><code><a href="layer_random_crop.html">layer_random_crop</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_flip.html">layer_random_flip</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_rotation.html">layer_random_rotation</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_random_translation.html">layer_random_translation</a>()</code> <br><code><a href="layer_random_zoom.html">layer_random_zoom</a>()</code> <br><code><a href="layer_repeat_vector.html">layer_repeat_vector</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_reshape.html">layer_reshape</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_rms_normalization.html">layer_rms_normalization</a>()</code> <br><code><a href="layer_rnn.html">layer_rnn</a>()</code> <br><code><a href="layer_separable_conv_1d.html">layer_separable_conv_1d</a>()</code> <br><code><a href="layer_separable_conv_2d.html">layer_separable_conv_2d</a>()</code> <br><code><a href="layer_simple_rnn.html">layer_simple_rnn</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br><code><a href="layer_spatial_dropout_1d.html">layer_spatial_dropout_1d</a>()</code> <br><code><a href="layer_spatial_dropout_2d.html">layer_spatial_dropout_2d</a>()</code> <br><code><a href="layer_spatial_dropout_3d.html">layer_spatial_dropout_3d</a>()</code> <br><code><a href="layer_spectral_normalization.html">layer_spectral_normalization</a>()</code> <br><code><a href="layer_stft_spectrogram.html">layer_stft_spectrogram</a>()</code> <br><code><a href="layer_string_lookup.html">layer_string_lookup</a>()</code> <br><code><a href="layer_subtract.html">layer_subtract</a>()</code> <br><code><a href="layer_text_vectorization.html">layer_text_vectorization</a>()</code> <br><code><a href="layer_tfsm.html">layer_tfsm</a>()</code> <br><code><a href="layer_time_distributed.html">layer_time_distributed</a>()</code> <br><code><a href="layer_torch_module_wrapper.html">layer_torch_module_wrapper</a>()</code> <br><code><a href="layer_unit_normalization.html">layer_unit_normalization</a>()</code> <br><code><a href="layer_upsampling_1d.html">layer_upsampling_1d</a>()</code> <br><code><a href="layer_upsampling_2d.html">layer_upsampling_2d</a>()</code> <br><code><a href="layer_upsampling_3d.html">layer_upsampling_3d</a>()</code> <br><code><a href="layer_zero_padding_1d.html">layer_zero_padding_1d</a>()</code> <br><code><a href="layer_zero_padding_2d.html">layer_zero_padding_2d</a>()</code> <br><code><a href="layer_zero_padding_3d.html">layer_zero_padding_3d</a>()</code> <br><code><a href="rnn_cell_gru.html">rnn_cell_gru</a>()</code> <br><code><a href="rnn_cell_lstm.html">rnn_cell_lstm</a>()</code> <br><code><a href="rnn_cell_simple.html">rnn_cell_simple</a>()</code> <br><code><a href="rnn_cells_stack.html">rnn_cells_stack</a>()</code> <br></p></div>
+<p>Other layers: <br><code><a href="Layer.html">Layer</a>()</code> <br><code><a href="layer_activation.html">layer_activation</a>()</code> <br><code><a href="layer_activation_elu.html">layer_activation_elu</a>()</code> <br><code><a href="layer_activation_leaky_relu.html">layer_activation_leaky_relu</a>()</code> <br><code><a href="layer_activation_parametric_relu.html">layer_activation_parametric_relu</a>()</code> <br><code><a href="layer_activation_relu.html">layer_activation_relu</a>()</code> <br><code><a href="layer_activity_regularization.html">layer_activity_regularization</a>()</code> <br><code><a href="layer_add.html">layer_add</a>()</code> <br><code><a href="layer_additive_attention.html">layer_additive_attention</a>()</code> <br><code><a href="layer_alpha_dropout.html">layer_alpha_dropout</a>()</code> <br><code><a href="layer_attention.html">layer_attention</a>()</code> <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_average.html">layer_average</a>()</code> <br><code><a href="layer_average_pooling_1d.html">layer_average_pooling_1d</a>()</code> <br><code><a href="layer_average_pooling_2d.html">layer_average_pooling_2d</a>()</code> <br><code><a href="layer_average_pooling_3d.html">layer_average_pooling_3d</a>()</code> <br><code><a href="layer_batch_normalization.html">layer_batch_normalization</a>()</code> <br><code><a href="layer_bidirectional.html">layer_bidirectional</a>()</code> <br><code><a href="layer_category_encoding.html">layer_category_encoding</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_concatenate.html">layer_concatenate</a>()</code> <br><code><a href="layer_conv_1d.html">layer_conv_1d</a>()</code> <br><code><a href="layer_conv_1d_transpose.html">layer_conv_1d_transpose</a>()</code> <br><code><a href="layer_conv_2d.html">layer_conv_2d</a>()</code> <br><code><a href="layer_conv_2d_transpose.html">layer_conv_2d_transpose</a>()</code> <br><code><a href="layer_conv_3d.html">layer_conv_3d</a>()</code> <br><code><a href="layer_conv_3d_transpose.html">layer_conv_3d_transpose</a>()</code> <br><code><a href="layer_conv_lstm_1d.html">layer_conv_lstm_1d</a>()</code> <br><code><a href="layer_conv_lstm_2d.html">layer_conv_lstm_2d</a>()</code> <br><code><a href="layer_conv_lstm_3d.html">layer_conv_lstm_3d</a>()</code> <br><code><a href="layer_cropping_1d.html">layer_cropping_1d</a>()</code> <br><code><a href="layer_cropping_2d.html">layer_cropping_2d</a>()</code> <br><code><a href="layer_cropping_3d.html">layer_cropping_3d</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_dense.html">layer_dense</a>()</code> <br><code><a href="layer_depthwise_conv_1d.html">layer_depthwise_conv_1d</a>()</code> <br><code><a href="layer_depthwise_conv_2d.html">layer_depthwise_conv_2d</a>()</code> <br><code><a href="layer_discretization.html">layer_discretization</a>()</code> <br><code><a href="layer_dot.html">layer_dot</a>()</code> <br><code><a href="layer_dropout.html">layer_dropout</a>()</code> <br><code><a href="layer_einsum_dense.html">layer_einsum_dense</a>()</code> <br><code><a href="layer_embedding.html">layer_embedding</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_feature_space.html">layer_feature_space</a>()</code> <br><code><a href="layer_flatten.html">layer_flatten</a>()</code> <br><code><a href="layer_flax_module_wrapper.html">layer_flax_module_wrapper</a>()</code> <br><code><a href="layer_gaussian_dropout.html">layer_gaussian_dropout</a>()</code> <br><code><a href="layer_gaussian_noise.html">layer_gaussian_noise</a>()</code> <br><code><a href="layer_global_average_pooling_1d.html">layer_global_average_pooling_1d</a>()</code> <br><code><a href="layer_global_average_pooling_2d.html">layer_global_average_pooling_2d</a>()</code> <br><code><a href="layer_global_average_pooling_3d.html">layer_global_average_pooling_3d</a>()</code> <br><code><a href="layer_global_max_pooling_1d.html">layer_global_max_pooling_1d</a>()</code> <br><code><a href="layer_global_max_pooling_2d.html">layer_global_max_pooling_2d</a>()</code> <br><code><a href="layer_global_max_pooling_3d.html">layer_global_max_pooling_3d</a>()</code> <br><code><a href="layer_group_normalization.html">layer_group_normalization</a>()</code> <br><code><a href="layer_group_query_attention.html">layer_group_query_attention</a>()</code> <br><code><a href="layer_gru.html">layer_gru</a>()</code> <br><code><a href="layer_hashed_crossing.html">layer_hashed_crossing</a>()</code> <br><code><a href="layer_hashing.html">layer_hashing</a>()</code> <br><code><a href="layer_identity.html">layer_identity</a>()</code> <br><code><a href="layer_integer_lookup.html">layer_integer_lookup</a>()</code> <br><code><a href="layer_jax_model_wrapper.html">layer_jax_model_wrapper</a>()</code> <br><code><a href="layer_lambda.html">layer_lambda</a>()</code> <br><code><a href="layer_layer_normalization.html">layer_layer_normalization</a>()</code> <br><code><a href="layer_lstm.html">layer_lstm</a>()</code> <br><code><a href="layer_masking.html">layer_masking</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_max_pooling_1d.html">layer_max_pooling_1d</a>()</code> <br><code><a href="layer_max_pooling_2d.html">layer_max_pooling_2d</a>()</code> <br><code><a href="layer_max_pooling_3d.html">layer_max_pooling_3d</a>()</code> <br><code><a href="layer_maximum.html">layer_maximum</a>()</code> <br><code><a href="layer_mel_spectrogram.html">layer_mel_spectrogram</a>()</code> <br><code><a href="layer_minimum.html">layer_minimum</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_multi_head_attention.html">layer_multi_head_attention</a>()</code> <br><code><a href="layer_multiply.html">layer_multiply</a>()</code> <br><code><a href="layer_normalization.html">layer_normalization</a>()</code> <br><code><a href="layer_permute.html">layer_permute</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_brightness.html">layer_random_brightness</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_contrast.html">layer_random_contrast</a>()</code> <br><code><a href="layer_random_crop.html">layer_random_crop</a>()</code> <br><code><a href="layer_random_elastic_transform.html">layer_random_elastic_transform</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_flip.html">layer_random_flip</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_rotation.html">layer_random_rotation</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_random_translation.html">layer_random_translation</a>()</code> <br><code><a href="layer_random_zoom.html">layer_random_zoom</a>()</code> <br><code><a href="layer_repeat_vector.html">layer_repeat_vector</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_reshape.html">layer_reshape</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_rms_normalization.html">layer_rms_normalization</a>()</code> <br><code><a href="layer_rnn.html">layer_rnn</a>()</code> <br><code><a href="layer_separable_conv_1d.html">layer_separable_conv_1d</a>()</code> <br><code><a href="layer_separable_conv_2d.html">layer_separable_conv_2d</a>()</code> <br><code><a href="layer_simple_rnn.html">layer_simple_rnn</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br><code><a href="layer_spatial_dropout_1d.html">layer_spatial_dropout_1d</a>()</code> <br><code><a href="layer_spatial_dropout_2d.html">layer_spatial_dropout_2d</a>()</code> <br><code><a href="layer_spatial_dropout_3d.html">layer_spatial_dropout_3d</a>()</code> <br><code><a href="layer_spectral_normalization.html">layer_spectral_normalization</a>()</code> <br><code><a href="layer_stft_spectrogram.html">layer_stft_spectrogram</a>()</code> <br><code><a href="layer_string_lookup.html">layer_string_lookup</a>()</code> <br><code><a href="layer_subtract.html">layer_subtract</a>()</code> <br><code><a href="layer_text_vectorization.html">layer_text_vectorization</a>()</code> <br><code><a href="layer_tfsm.html">layer_tfsm</a>()</code> <br><code><a href="layer_time_distributed.html">layer_time_distributed</a>()</code> <br><code><a href="layer_torch_module_wrapper.html">layer_torch_module_wrapper</a>()</code> <br><code><a href="layer_unit_normalization.html">layer_unit_normalization</a>()</code> <br><code><a href="layer_upsampling_1d.html">layer_upsampling_1d</a>()</code> <br><code><a href="layer_upsampling_2d.html">layer_upsampling_2d</a>()</code> <br><code><a href="layer_upsampling_3d.html">layer_upsampling_3d</a>()</code> <br><code><a href="layer_zero_padding_1d.html">layer_zero_padding_1d</a>()</code> <br><code><a href="layer_zero_padding_2d.html">layer_zero_padding_2d</a>()</code> <br><code><a href="layer_zero_padding_3d.html">layer_zero_padding_3d</a>()</code> <br><code><a href="rnn_cell_gru.html">rnn_cell_gru</a>()</code> <br><code><a href="rnn_cell_lstm.html">rnn_cell_lstm</a>()</code> <br><code><a href="rnn_cell_simple.html">rnn_cell_simple</a>()</code> <br><code><a href="rnn_cells_stack.html">rnn_cells_stack</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -123,7 +123,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/layer_activity_regularization.html b/docs/dev/reference/layer_activity_regularization.html
index 9d8ccd78e..bbb4c3524 100644
--- a/docs/dev/reference/layer_activity_regularization.html
+++ b/docs/dev/reference/layer_activity_regularization.html
@@ -8,7 +8,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -106,7 +106,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
     <div class="dont-index">
 <ul><li><p><a href="https://keras.io/api/layers/regularization_layers/activity_regularization#activityregularization-class" class="external-link">https://keras.io/api/layers/regularization_layers/activity_regularization#activityregularization-class</a></p></li>
 </ul><p>Other regularization layers: <br><code><a href="layer_alpha_dropout.html">layer_alpha_dropout</a>()</code> <br><code><a href="layer_dropout.html">layer_dropout</a>()</code> <br><code><a href="layer_gaussian_dropout.html">layer_gaussian_dropout</a>()</code> <br><code><a href="layer_gaussian_noise.html">layer_gaussian_noise</a>()</code> <br><code><a href="layer_spatial_dropout_1d.html">layer_spatial_dropout_1d</a>()</code> <br><code><a href="layer_spatial_dropout_2d.html">layer_spatial_dropout_2d</a>()</code> <br><code><a href="layer_spatial_dropout_3d.html">layer_spatial_dropout_3d</a>()</code> <br></p>
-<p>Other layers: <br><code><a href="Layer.html">Layer</a>()</code> <br><code><a href="layer_activation.html">layer_activation</a>()</code> <br><code><a href="layer_activation_elu.html">layer_activation_elu</a>()</code> <br><code><a href="layer_activation_leaky_relu.html">layer_activation_leaky_relu</a>()</code> <br><code><a href="layer_activation_parametric_relu.html">layer_activation_parametric_relu</a>()</code> <br><code><a href="layer_activation_relu.html">layer_activation_relu</a>()</code> <br><code><a href="layer_activation_softmax.html">layer_activation_softmax</a>()</code> <br><code><a href="layer_add.html">layer_add</a>()</code> <br><code><a href="layer_additive_attention.html">layer_additive_attention</a>()</code> <br><code><a href="layer_alpha_dropout.html">layer_alpha_dropout</a>()</code> <br><code><a href="layer_attention.html">layer_attention</a>()</code> <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_average.html">layer_average</a>()</code> <br><code><a href="layer_average_pooling_1d.html">layer_average_pooling_1d</a>()</code> <br><code><a href="layer_average_pooling_2d.html">layer_average_pooling_2d</a>()</code> <br><code><a href="layer_average_pooling_3d.html">layer_average_pooling_3d</a>()</code> <br><code><a href="layer_batch_normalization.html">layer_batch_normalization</a>()</code> <br><code><a href="layer_bidirectional.html">layer_bidirectional</a>()</code> <br><code><a href="layer_category_encoding.html">layer_category_encoding</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_concatenate.html">layer_concatenate</a>()</code> <br><code><a href="layer_conv_1d.html">layer_conv_1d</a>()</code> <br><code><a href="layer_conv_1d_transpose.html">layer_conv_1d_transpose</a>()</code> <br><code><a href="layer_conv_2d.html">layer_conv_2d</a>()</code> <br><code><a href="layer_conv_2d_transpose.html">layer_conv_2d_transpose</a>()</code> <br><code><a href="layer_conv_3d.html">layer_conv_3d</a>()</code> <br><code><a href="layer_conv_3d_transpose.html">layer_conv_3d_transpose</a>()</code> <br><code><a href="layer_conv_lstm_1d.html">layer_conv_lstm_1d</a>()</code> <br><code><a href="layer_conv_lstm_2d.html">layer_conv_lstm_2d</a>()</code> <br><code><a href="layer_conv_lstm_3d.html">layer_conv_lstm_3d</a>()</code> <br><code><a href="layer_cropping_1d.html">layer_cropping_1d</a>()</code> <br><code><a href="layer_cropping_2d.html">layer_cropping_2d</a>()</code> <br><code><a href="layer_cropping_3d.html">layer_cropping_3d</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_dense.html">layer_dense</a>()</code> <br><code><a href="layer_depthwise_conv_1d.html">layer_depthwise_conv_1d</a>()</code> <br><code><a href="layer_depthwise_conv_2d.html">layer_depthwise_conv_2d</a>()</code> <br><code><a href="layer_discretization.html">layer_discretization</a>()</code> <br><code><a href="layer_dot.html">layer_dot</a>()</code> <br><code><a href="layer_dropout.html">layer_dropout</a>()</code> <br><code><a href="layer_einsum_dense.html">layer_einsum_dense</a>()</code> <br><code><a href="layer_embedding.html">layer_embedding</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_feature_space.html">layer_feature_space</a>()</code> <br><code><a href="layer_flatten.html">layer_flatten</a>()</code> <br><code><a href="layer_flax_module_wrapper.html">layer_flax_module_wrapper</a>()</code> <br><code><a href="layer_gaussian_dropout.html">layer_gaussian_dropout</a>()</code> <br><code><a href="layer_gaussian_noise.html">layer_gaussian_noise</a>()</code> <br><code><a href="layer_global_average_pooling_1d.html">layer_global_average_pooling_1d</a>()</code> <br><code><a href="layer_global_average_pooling_2d.html">layer_global_average_pooling_2d</a>()</code> <br><code><a href="layer_global_average_pooling_3d.html">layer_global_average_pooling_3d</a>()</code> <br><code><a href="layer_global_max_pooling_1d.html">layer_global_max_pooling_1d</a>()</code> <br><code><a href="layer_global_max_pooling_2d.html">layer_global_max_pooling_2d</a>()</code> <br><code><a href="layer_global_max_pooling_3d.html">layer_global_max_pooling_3d</a>()</code> <br><code><a href="layer_group_normalization.html">layer_group_normalization</a>()</code> <br><code><a href="layer_group_query_attention.html">layer_group_query_attention</a>()</code> <br><code><a href="layer_gru.html">layer_gru</a>()</code> <br><code><a href="layer_hashed_crossing.html">layer_hashed_crossing</a>()</code> <br><code><a href="layer_hashing.html">layer_hashing</a>()</code> <br><code><a href="layer_identity.html">layer_identity</a>()</code> <br><code><a href="layer_integer_lookup.html">layer_integer_lookup</a>()</code> <br><code><a href="layer_jax_model_wrapper.html">layer_jax_model_wrapper</a>()</code> <br><code><a href="layer_lambda.html">layer_lambda</a>()</code> <br><code><a href="layer_layer_normalization.html">layer_layer_normalization</a>()</code> <br><code><a href="layer_lstm.html">layer_lstm</a>()</code> <br><code><a href="layer_masking.html">layer_masking</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_max_pooling_1d.html">layer_max_pooling_1d</a>()</code> <br><code><a href="layer_max_pooling_2d.html">layer_max_pooling_2d</a>()</code> <br><code><a href="layer_max_pooling_3d.html">layer_max_pooling_3d</a>()</code> <br><code><a href="layer_maximum.html">layer_maximum</a>()</code> <br><code><a href="layer_mel_spectrogram.html">layer_mel_spectrogram</a>()</code> <br><code><a href="layer_minimum.html">layer_minimum</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_multi_head_attention.html">layer_multi_head_attention</a>()</code> <br><code><a href="layer_multiply.html">layer_multiply</a>()</code> <br><code><a href="layer_normalization.html">layer_normalization</a>()</code> <br><code><a href="layer_permute.html">layer_permute</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_brightness.html">layer_random_brightness</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_contrast.html">layer_random_contrast</a>()</code> <br><code><a href="layer_random_crop.html">layer_random_crop</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_flip.html">layer_random_flip</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_rotation.html">layer_random_rotation</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_random_translation.html">layer_random_translation</a>()</code> <br><code><a href="layer_random_zoom.html">layer_random_zoom</a>()</code> <br><code><a href="layer_repeat_vector.html">layer_repeat_vector</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_reshape.html">layer_reshape</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_rms_normalization.html">layer_rms_normalization</a>()</code> <br><code><a href="layer_rnn.html">layer_rnn</a>()</code> <br><code><a href="layer_separable_conv_1d.html">layer_separable_conv_1d</a>()</code> <br><code><a href="layer_separable_conv_2d.html">layer_separable_conv_2d</a>()</code> <br><code><a href="layer_simple_rnn.html">layer_simple_rnn</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br><code><a href="layer_spatial_dropout_1d.html">layer_spatial_dropout_1d</a>()</code> <br><code><a href="layer_spatial_dropout_2d.html">layer_spatial_dropout_2d</a>()</code> <br><code><a href="layer_spatial_dropout_3d.html">layer_spatial_dropout_3d</a>()</code> <br><code><a href="layer_spectral_normalization.html">layer_spectral_normalization</a>()</code> <br><code><a href="layer_stft_spectrogram.html">layer_stft_spectrogram</a>()</code> <br><code><a href="layer_string_lookup.html">layer_string_lookup</a>()</code> <br><code><a href="layer_subtract.html">layer_subtract</a>()</code> <br><code><a href="layer_text_vectorization.html">layer_text_vectorization</a>()</code> <br><code><a href="layer_tfsm.html">layer_tfsm</a>()</code> <br><code><a href="layer_time_distributed.html">layer_time_distributed</a>()</code> <br><code><a href="layer_torch_module_wrapper.html">layer_torch_module_wrapper</a>()</code> <br><code><a href="layer_unit_normalization.html">layer_unit_normalization</a>()</code> <br><code><a href="layer_upsampling_1d.html">layer_upsampling_1d</a>()</code> <br><code><a href="layer_upsampling_2d.html">layer_upsampling_2d</a>()</code> <br><code><a href="layer_upsampling_3d.html">layer_upsampling_3d</a>()</code> <br><code><a href="layer_zero_padding_1d.html">layer_zero_padding_1d</a>()</code> <br><code><a href="layer_zero_padding_2d.html">layer_zero_padding_2d</a>()</code> <br><code><a href="layer_zero_padding_3d.html">layer_zero_padding_3d</a>()</code> <br><code><a href="rnn_cell_gru.html">rnn_cell_gru</a>()</code> <br><code><a href="rnn_cell_lstm.html">rnn_cell_lstm</a>()</code> <br><code><a href="rnn_cell_simple.html">rnn_cell_simple</a>()</code> <br><code><a href="rnn_cells_stack.html">rnn_cells_stack</a>()</code> <br></p></div>
+<p>Other layers: <br><code><a href="Layer.html">Layer</a>()</code> <br><code><a href="layer_activation.html">layer_activation</a>()</code> <br><code><a href="layer_activation_elu.html">layer_activation_elu</a>()</code> <br><code><a href="layer_activation_leaky_relu.html">layer_activation_leaky_relu</a>()</code> <br><code><a href="layer_activation_parametric_relu.html">layer_activation_parametric_relu</a>()</code> <br><code><a href="layer_activation_relu.html">layer_activation_relu</a>()</code> <br><code><a href="layer_activation_softmax.html">layer_activation_softmax</a>()</code> <br><code><a href="layer_add.html">layer_add</a>()</code> <br><code><a href="layer_additive_attention.html">layer_additive_attention</a>()</code> <br><code><a href="layer_alpha_dropout.html">layer_alpha_dropout</a>()</code> <br><code><a href="layer_attention.html">layer_attention</a>()</code> <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_average.html">layer_average</a>()</code> <br><code><a href="layer_average_pooling_1d.html">layer_average_pooling_1d</a>()</code> <br><code><a href="layer_average_pooling_2d.html">layer_average_pooling_2d</a>()</code> <br><code><a href="layer_average_pooling_3d.html">layer_average_pooling_3d</a>()</code> <br><code><a href="layer_batch_normalization.html">layer_batch_normalization</a>()</code> <br><code><a href="layer_bidirectional.html">layer_bidirectional</a>()</code> <br><code><a href="layer_category_encoding.html">layer_category_encoding</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_concatenate.html">layer_concatenate</a>()</code> <br><code><a href="layer_conv_1d.html">layer_conv_1d</a>()</code> <br><code><a href="layer_conv_1d_transpose.html">layer_conv_1d_transpose</a>()</code> <br><code><a href="layer_conv_2d.html">layer_conv_2d</a>()</code> <br><code><a href="layer_conv_2d_transpose.html">layer_conv_2d_transpose</a>()</code> <br><code><a href="layer_conv_3d.html">layer_conv_3d</a>()</code> <br><code><a href="layer_conv_3d_transpose.html">layer_conv_3d_transpose</a>()</code> <br><code><a href="layer_conv_lstm_1d.html">layer_conv_lstm_1d</a>()</code> <br><code><a href="layer_conv_lstm_2d.html">layer_conv_lstm_2d</a>()</code> <br><code><a href="layer_conv_lstm_3d.html">layer_conv_lstm_3d</a>()</code> <br><code><a href="layer_cropping_1d.html">layer_cropping_1d</a>()</code> <br><code><a href="layer_cropping_2d.html">layer_cropping_2d</a>()</code> <br><code><a href="layer_cropping_3d.html">layer_cropping_3d</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_dense.html">layer_dense</a>()</code> <br><code><a href="layer_depthwise_conv_1d.html">layer_depthwise_conv_1d</a>()</code> <br><code><a href="layer_depthwise_conv_2d.html">layer_depthwise_conv_2d</a>()</code> <br><code><a href="layer_discretization.html">layer_discretization</a>()</code> <br><code><a href="layer_dot.html">layer_dot</a>()</code> <br><code><a href="layer_dropout.html">layer_dropout</a>()</code> <br><code><a href="layer_einsum_dense.html">layer_einsum_dense</a>()</code> <br><code><a href="layer_embedding.html">layer_embedding</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_feature_space.html">layer_feature_space</a>()</code> <br><code><a href="layer_flatten.html">layer_flatten</a>()</code> <br><code><a href="layer_flax_module_wrapper.html">layer_flax_module_wrapper</a>()</code> <br><code><a href="layer_gaussian_dropout.html">layer_gaussian_dropout</a>()</code> <br><code><a href="layer_gaussian_noise.html">layer_gaussian_noise</a>()</code> <br><code><a href="layer_global_average_pooling_1d.html">layer_global_average_pooling_1d</a>()</code> <br><code><a href="layer_global_average_pooling_2d.html">layer_global_average_pooling_2d</a>()</code> <br><code><a href="layer_global_average_pooling_3d.html">layer_global_average_pooling_3d</a>()</code> <br><code><a href="layer_global_max_pooling_1d.html">layer_global_max_pooling_1d</a>()</code> <br><code><a href="layer_global_max_pooling_2d.html">layer_global_max_pooling_2d</a>()</code> <br><code><a href="layer_global_max_pooling_3d.html">layer_global_max_pooling_3d</a>()</code> <br><code><a href="layer_group_normalization.html">layer_group_normalization</a>()</code> <br><code><a href="layer_group_query_attention.html">layer_group_query_attention</a>()</code> <br><code><a href="layer_gru.html">layer_gru</a>()</code> <br><code><a href="layer_hashed_crossing.html">layer_hashed_crossing</a>()</code> <br><code><a href="layer_hashing.html">layer_hashing</a>()</code> <br><code><a href="layer_identity.html">layer_identity</a>()</code> <br><code><a href="layer_integer_lookup.html">layer_integer_lookup</a>()</code> <br><code><a href="layer_jax_model_wrapper.html">layer_jax_model_wrapper</a>()</code> <br><code><a href="layer_lambda.html">layer_lambda</a>()</code> <br><code><a href="layer_layer_normalization.html">layer_layer_normalization</a>()</code> <br><code><a href="layer_lstm.html">layer_lstm</a>()</code> <br><code><a href="layer_masking.html">layer_masking</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_max_pooling_1d.html">layer_max_pooling_1d</a>()</code> <br><code><a href="layer_max_pooling_2d.html">layer_max_pooling_2d</a>()</code> <br><code><a href="layer_max_pooling_3d.html">layer_max_pooling_3d</a>()</code> <br><code><a href="layer_maximum.html">layer_maximum</a>()</code> <br><code><a href="layer_mel_spectrogram.html">layer_mel_spectrogram</a>()</code> <br><code><a href="layer_minimum.html">layer_minimum</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_multi_head_attention.html">layer_multi_head_attention</a>()</code> <br><code><a href="layer_multiply.html">layer_multiply</a>()</code> <br><code><a href="layer_normalization.html">layer_normalization</a>()</code> <br><code><a href="layer_permute.html">layer_permute</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_brightness.html">layer_random_brightness</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_contrast.html">layer_random_contrast</a>()</code> <br><code><a href="layer_random_crop.html">layer_random_crop</a>()</code> <br><code><a href="layer_random_elastic_transform.html">layer_random_elastic_transform</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_flip.html">layer_random_flip</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_rotation.html">layer_random_rotation</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_random_translation.html">layer_random_translation</a>()</code> <br><code><a href="layer_random_zoom.html">layer_random_zoom</a>()</code> <br><code><a href="layer_repeat_vector.html">layer_repeat_vector</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_reshape.html">layer_reshape</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_rms_normalization.html">layer_rms_normalization</a>()</code> <br><code><a href="layer_rnn.html">layer_rnn</a>()</code> <br><code><a href="layer_separable_conv_1d.html">layer_separable_conv_1d</a>()</code> <br><code><a href="layer_separable_conv_2d.html">layer_separable_conv_2d</a>()</code> <br><code><a href="layer_simple_rnn.html">layer_simple_rnn</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br><code><a href="layer_spatial_dropout_1d.html">layer_spatial_dropout_1d</a>()</code> <br><code><a href="layer_spatial_dropout_2d.html">layer_spatial_dropout_2d</a>()</code> <br><code><a href="layer_spatial_dropout_3d.html">layer_spatial_dropout_3d</a>()</code> <br><code><a href="layer_spectral_normalization.html">layer_spectral_normalization</a>()</code> <br><code><a href="layer_stft_spectrogram.html">layer_stft_spectrogram</a>()</code> <br><code><a href="layer_string_lookup.html">layer_string_lookup</a>()</code> <br><code><a href="layer_subtract.html">layer_subtract</a>()</code> <br><code><a href="layer_text_vectorization.html">layer_text_vectorization</a>()</code> <br><code><a href="layer_tfsm.html">layer_tfsm</a>()</code> <br><code><a href="layer_time_distributed.html">layer_time_distributed</a>()</code> <br><code><a href="layer_torch_module_wrapper.html">layer_torch_module_wrapper</a>()</code> <br><code><a href="layer_unit_normalization.html">layer_unit_normalization</a>()</code> <br><code><a href="layer_upsampling_1d.html">layer_upsampling_1d</a>()</code> <br><code><a href="layer_upsampling_2d.html">layer_upsampling_2d</a>()</code> <br><code><a href="layer_upsampling_3d.html">layer_upsampling_3d</a>()</code> <br><code><a href="layer_zero_padding_1d.html">layer_zero_padding_1d</a>()</code> <br><code><a href="layer_zero_padding_2d.html">layer_zero_padding_2d</a>()</code> <br><code><a href="layer_zero_padding_3d.html">layer_zero_padding_3d</a>()</code> <br><code><a href="rnn_cell_gru.html">rnn_cell_gru</a>()</code> <br><code><a href="rnn_cell_lstm.html">rnn_cell_lstm</a>()</code> <br><code><a href="rnn_cell_simple.html">rnn_cell_simple</a>()</code> <br><code><a href="rnn_cells_stack.html">rnn_cells_stack</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -118,7 +118,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/layer_add.html b/docs/dev/reference/layer_add.html
index 9184718f6..e1b75ee92 100644
--- a/docs/dev/reference/layer_add.html
+++ b/docs/dev/reference/layer_add.html
@@ -10,7 +10,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -114,7 +114,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
     <div class="dont-index">
 <ul><li><p><a href="https://keras.io/api/layers/merging_layers/add#add-class" class="external-link">https://keras.io/api/layers/merging_layers/add#add-class</a></p></li>
 </ul><p>Other merging layers: <br><code><a href="layer_average.html">layer_average</a>()</code> <br><code><a href="layer_concatenate.html">layer_concatenate</a>()</code> <br><code><a href="layer_dot.html">layer_dot</a>()</code> <br><code><a href="layer_maximum.html">layer_maximum</a>()</code> <br><code><a href="layer_minimum.html">layer_minimum</a>()</code> <br><code><a href="layer_multiply.html">layer_multiply</a>()</code> <br><code><a href="layer_subtract.html">layer_subtract</a>()</code> <br></p>
-<p>Other layers: <br><code><a href="Layer.html">Layer</a>()</code> <br><code><a href="layer_activation.html">layer_activation</a>()</code> <br><code><a href="layer_activation_elu.html">layer_activation_elu</a>()</code> <br><code><a href="layer_activation_leaky_relu.html">layer_activation_leaky_relu</a>()</code> <br><code><a href="layer_activation_parametric_relu.html">layer_activation_parametric_relu</a>()</code> <br><code><a href="layer_activation_relu.html">layer_activation_relu</a>()</code> <br><code><a href="layer_activation_softmax.html">layer_activation_softmax</a>()</code> <br><code><a href="layer_activity_regularization.html">layer_activity_regularization</a>()</code> <br><code><a href="layer_additive_attention.html">layer_additive_attention</a>()</code> <br><code><a href="layer_alpha_dropout.html">layer_alpha_dropout</a>()</code> <br><code><a href="layer_attention.html">layer_attention</a>()</code> <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_average.html">layer_average</a>()</code> <br><code><a href="layer_average_pooling_1d.html">layer_average_pooling_1d</a>()</code> <br><code><a href="layer_average_pooling_2d.html">layer_average_pooling_2d</a>()</code> <br><code><a href="layer_average_pooling_3d.html">layer_average_pooling_3d</a>()</code> <br><code><a href="layer_batch_normalization.html">layer_batch_normalization</a>()</code> <br><code><a href="layer_bidirectional.html">layer_bidirectional</a>()</code> <br><code><a href="layer_category_encoding.html">layer_category_encoding</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_concatenate.html">layer_concatenate</a>()</code> <br><code><a href="layer_conv_1d.html">layer_conv_1d</a>()</code> <br><code><a href="layer_conv_1d_transpose.html">layer_conv_1d_transpose</a>()</code> <br><code><a href="layer_conv_2d.html">layer_conv_2d</a>()</code> <br><code><a href="layer_conv_2d_transpose.html">layer_conv_2d_transpose</a>()</code> <br><code><a href="layer_conv_3d.html">layer_conv_3d</a>()</code> <br><code><a href="layer_conv_3d_transpose.html">layer_conv_3d_transpose</a>()</code> <br><code><a href="layer_conv_lstm_1d.html">layer_conv_lstm_1d</a>()</code> <br><code><a href="layer_conv_lstm_2d.html">layer_conv_lstm_2d</a>()</code> <br><code><a href="layer_conv_lstm_3d.html">layer_conv_lstm_3d</a>()</code> <br><code><a href="layer_cropping_1d.html">layer_cropping_1d</a>()</code> <br><code><a href="layer_cropping_2d.html">layer_cropping_2d</a>()</code> <br><code><a href="layer_cropping_3d.html">layer_cropping_3d</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_dense.html">layer_dense</a>()</code> <br><code><a href="layer_depthwise_conv_1d.html">layer_depthwise_conv_1d</a>()</code> <br><code><a href="layer_depthwise_conv_2d.html">layer_depthwise_conv_2d</a>()</code> <br><code><a href="layer_discretization.html">layer_discretization</a>()</code> <br><code><a href="layer_dot.html">layer_dot</a>()</code> <br><code><a href="layer_dropout.html">layer_dropout</a>()</code> <br><code><a href="layer_einsum_dense.html">layer_einsum_dense</a>()</code> <br><code><a href="layer_embedding.html">layer_embedding</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_feature_space.html">layer_feature_space</a>()</code> <br><code><a href="layer_flatten.html">layer_flatten</a>()</code> <br><code><a href="layer_flax_module_wrapper.html">layer_flax_module_wrapper</a>()</code> <br><code><a href="layer_gaussian_dropout.html">layer_gaussian_dropout</a>()</code> <br><code><a href="layer_gaussian_noise.html">layer_gaussian_noise</a>()</code> <br><code><a href="layer_global_average_pooling_1d.html">layer_global_average_pooling_1d</a>()</code> <br><code><a href="layer_global_average_pooling_2d.html">layer_global_average_pooling_2d</a>()</code> <br><code><a href="layer_global_average_pooling_3d.html">layer_global_average_pooling_3d</a>()</code> <br><code><a href="layer_global_max_pooling_1d.html">layer_global_max_pooling_1d</a>()</code> <br><code><a href="layer_global_max_pooling_2d.html">layer_global_max_pooling_2d</a>()</code> <br><code><a href="layer_global_max_pooling_3d.html">layer_global_max_pooling_3d</a>()</code> <br><code><a href="layer_group_normalization.html">layer_group_normalization</a>()</code> <br><code><a href="layer_group_query_attention.html">layer_group_query_attention</a>()</code> <br><code><a href="layer_gru.html">layer_gru</a>()</code> <br><code><a href="layer_hashed_crossing.html">layer_hashed_crossing</a>()</code> <br><code><a href="layer_hashing.html">layer_hashing</a>()</code> <br><code><a href="layer_identity.html">layer_identity</a>()</code> <br><code><a href="layer_integer_lookup.html">layer_integer_lookup</a>()</code> <br><code><a href="layer_jax_model_wrapper.html">layer_jax_model_wrapper</a>()</code> <br><code><a href="layer_lambda.html">layer_lambda</a>()</code> <br><code><a href="layer_layer_normalization.html">layer_layer_normalization</a>()</code> <br><code><a href="layer_lstm.html">layer_lstm</a>()</code> <br><code><a href="layer_masking.html">layer_masking</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_max_pooling_1d.html">layer_max_pooling_1d</a>()</code> <br><code><a href="layer_max_pooling_2d.html">layer_max_pooling_2d</a>()</code> <br><code><a href="layer_max_pooling_3d.html">layer_max_pooling_3d</a>()</code> <br><code><a href="layer_maximum.html">layer_maximum</a>()</code> <br><code><a href="layer_mel_spectrogram.html">layer_mel_spectrogram</a>()</code> <br><code><a href="layer_minimum.html">layer_minimum</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_multi_head_attention.html">layer_multi_head_attention</a>()</code> <br><code><a href="layer_multiply.html">layer_multiply</a>()</code> <br><code><a href="layer_normalization.html">layer_normalization</a>()</code> <br><code><a href="layer_permute.html">layer_permute</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_brightness.html">layer_random_brightness</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_contrast.html">layer_random_contrast</a>()</code> <br><code><a href="layer_random_crop.html">layer_random_crop</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_flip.html">layer_random_flip</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_rotation.html">layer_random_rotation</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_random_translation.html">layer_random_translation</a>()</code> <br><code><a href="layer_random_zoom.html">layer_random_zoom</a>()</code> <br><code><a href="layer_repeat_vector.html">layer_repeat_vector</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_reshape.html">layer_reshape</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_rms_normalization.html">layer_rms_normalization</a>()</code> <br><code><a href="layer_rnn.html">layer_rnn</a>()</code> <br><code><a href="layer_separable_conv_1d.html">layer_separable_conv_1d</a>()</code> <br><code><a href="layer_separable_conv_2d.html">layer_separable_conv_2d</a>()</code> <br><code><a href="layer_simple_rnn.html">layer_simple_rnn</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br><code><a href="layer_spatial_dropout_1d.html">layer_spatial_dropout_1d</a>()</code> <br><code><a href="layer_spatial_dropout_2d.html">layer_spatial_dropout_2d</a>()</code> <br><code><a href="layer_spatial_dropout_3d.html">layer_spatial_dropout_3d</a>()</code> <br><code><a href="layer_spectral_normalization.html">layer_spectral_normalization</a>()</code> <br><code><a href="layer_stft_spectrogram.html">layer_stft_spectrogram</a>()</code> <br><code><a href="layer_string_lookup.html">layer_string_lookup</a>()</code> <br><code><a href="layer_subtract.html">layer_subtract</a>()</code> <br><code><a href="layer_text_vectorization.html">layer_text_vectorization</a>()</code> <br><code><a href="layer_tfsm.html">layer_tfsm</a>()</code> <br><code><a href="layer_time_distributed.html">layer_time_distributed</a>()</code> <br><code><a href="layer_torch_module_wrapper.html">layer_torch_module_wrapper</a>()</code> <br><code><a href="layer_unit_normalization.html">layer_unit_normalization</a>()</code> <br><code><a href="layer_upsampling_1d.html">layer_upsampling_1d</a>()</code> <br><code><a href="layer_upsampling_2d.html">layer_upsampling_2d</a>()</code> <br><code><a href="layer_upsampling_3d.html">layer_upsampling_3d</a>()</code> <br><code><a href="layer_zero_padding_1d.html">layer_zero_padding_1d</a>()</code> <br><code><a href="layer_zero_padding_2d.html">layer_zero_padding_2d</a>()</code> <br><code><a href="layer_zero_padding_3d.html">layer_zero_padding_3d</a>()</code> <br><code><a href="rnn_cell_gru.html">rnn_cell_gru</a>()</code> <br><code><a href="rnn_cell_lstm.html">rnn_cell_lstm</a>()</code> <br><code><a href="rnn_cell_simple.html">rnn_cell_simple</a>()</code> <br><code><a href="rnn_cells_stack.html">rnn_cells_stack</a>()</code> <br></p></div>
+<p>Other layers: <br><code><a href="Layer.html">Layer</a>()</code> <br><code><a href="layer_activation.html">layer_activation</a>()</code> <br><code><a href="layer_activation_elu.html">layer_activation_elu</a>()</code> <br><code><a href="layer_activation_leaky_relu.html">layer_activation_leaky_relu</a>()</code> <br><code><a href="layer_activation_parametric_relu.html">layer_activation_parametric_relu</a>()</code> <br><code><a href="layer_activation_relu.html">layer_activation_relu</a>()</code> <br><code><a href="layer_activation_softmax.html">layer_activation_softmax</a>()</code> <br><code><a href="layer_activity_regularization.html">layer_activity_regularization</a>()</code> <br><code><a href="layer_additive_attention.html">layer_additive_attention</a>()</code> <br><code><a href="layer_alpha_dropout.html">layer_alpha_dropout</a>()</code> <br><code><a href="layer_attention.html">layer_attention</a>()</code> <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_average.html">layer_average</a>()</code> <br><code><a href="layer_average_pooling_1d.html">layer_average_pooling_1d</a>()</code> <br><code><a href="layer_average_pooling_2d.html">layer_average_pooling_2d</a>()</code> <br><code><a href="layer_average_pooling_3d.html">layer_average_pooling_3d</a>()</code> <br><code><a href="layer_batch_normalization.html">layer_batch_normalization</a>()</code> <br><code><a href="layer_bidirectional.html">layer_bidirectional</a>()</code> <br><code><a href="layer_category_encoding.html">layer_category_encoding</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_concatenate.html">layer_concatenate</a>()</code> <br><code><a href="layer_conv_1d.html">layer_conv_1d</a>()</code> <br><code><a href="layer_conv_1d_transpose.html">layer_conv_1d_transpose</a>()</code> <br><code><a href="layer_conv_2d.html">layer_conv_2d</a>()</code> <br><code><a href="layer_conv_2d_transpose.html">layer_conv_2d_transpose</a>()</code> <br><code><a href="layer_conv_3d.html">layer_conv_3d</a>()</code> <br><code><a href="layer_conv_3d_transpose.html">layer_conv_3d_transpose</a>()</code> <br><code><a href="layer_conv_lstm_1d.html">layer_conv_lstm_1d</a>()</code> <br><code><a href="layer_conv_lstm_2d.html">layer_conv_lstm_2d</a>()</code> <br><code><a href="layer_conv_lstm_3d.html">layer_conv_lstm_3d</a>()</code> <br><code><a href="layer_cropping_1d.html">layer_cropping_1d</a>()</code> <br><code><a href="layer_cropping_2d.html">layer_cropping_2d</a>()</code> <br><code><a href="layer_cropping_3d.html">layer_cropping_3d</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_dense.html">layer_dense</a>()</code> <br><code><a href="layer_depthwise_conv_1d.html">layer_depthwise_conv_1d</a>()</code> <br><code><a href="layer_depthwise_conv_2d.html">layer_depthwise_conv_2d</a>()</code> <br><code><a href="layer_discretization.html">layer_discretization</a>()</code> <br><code><a href="layer_dot.html">layer_dot</a>()</code> <br><code><a href="layer_dropout.html">layer_dropout</a>()</code> <br><code><a href="layer_einsum_dense.html">layer_einsum_dense</a>()</code> <br><code><a href="layer_embedding.html">layer_embedding</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_feature_space.html">layer_feature_space</a>()</code> <br><code><a href="layer_flatten.html">layer_flatten</a>()</code> <br><code><a href="layer_flax_module_wrapper.html">layer_flax_module_wrapper</a>()</code> <br><code><a href="layer_gaussian_dropout.html">layer_gaussian_dropout</a>()</code> <br><code><a href="layer_gaussian_noise.html">layer_gaussian_noise</a>()</code> <br><code><a href="layer_global_average_pooling_1d.html">layer_global_average_pooling_1d</a>()</code> <br><code><a href="layer_global_average_pooling_2d.html">layer_global_average_pooling_2d</a>()</code> <br><code><a href="layer_global_average_pooling_3d.html">layer_global_average_pooling_3d</a>()</code> <br><code><a href="layer_global_max_pooling_1d.html">layer_global_max_pooling_1d</a>()</code> <br><code><a href="layer_global_max_pooling_2d.html">layer_global_max_pooling_2d</a>()</code> <br><code><a href="layer_global_max_pooling_3d.html">layer_global_max_pooling_3d</a>()</code> <br><code><a href="layer_group_normalization.html">layer_group_normalization</a>()</code> <br><code><a href="layer_group_query_attention.html">layer_group_query_attention</a>()</code> <br><code><a href="layer_gru.html">layer_gru</a>()</code> <br><code><a href="layer_hashed_crossing.html">layer_hashed_crossing</a>()</code> <br><code><a href="layer_hashing.html">layer_hashing</a>()</code> <br><code><a href="layer_identity.html">layer_identity</a>()</code> <br><code><a href="layer_integer_lookup.html">layer_integer_lookup</a>()</code> <br><code><a href="layer_jax_model_wrapper.html">layer_jax_model_wrapper</a>()</code> <br><code><a href="layer_lambda.html">layer_lambda</a>()</code> <br><code><a href="layer_layer_normalization.html">layer_layer_normalization</a>()</code> <br><code><a href="layer_lstm.html">layer_lstm</a>()</code> <br><code><a href="layer_masking.html">layer_masking</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_max_pooling_1d.html">layer_max_pooling_1d</a>()</code> <br><code><a href="layer_max_pooling_2d.html">layer_max_pooling_2d</a>()</code> <br><code><a href="layer_max_pooling_3d.html">layer_max_pooling_3d</a>()</code> <br><code><a href="layer_maximum.html">layer_maximum</a>()</code> <br><code><a href="layer_mel_spectrogram.html">layer_mel_spectrogram</a>()</code> <br><code><a href="layer_minimum.html">layer_minimum</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_multi_head_attention.html">layer_multi_head_attention</a>()</code> <br><code><a href="layer_multiply.html">layer_multiply</a>()</code> <br><code><a href="layer_normalization.html">layer_normalization</a>()</code> <br><code><a href="layer_permute.html">layer_permute</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_brightness.html">layer_random_brightness</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_contrast.html">layer_random_contrast</a>()</code> <br><code><a href="layer_random_crop.html">layer_random_crop</a>()</code> <br><code><a href="layer_random_elastic_transform.html">layer_random_elastic_transform</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_flip.html">layer_random_flip</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_rotation.html">layer_random_rotation</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_random_translation.html">layer_random_translation</a>()</code> <br><code><a href="layer_random_zoom.html">layer_random_zoom</a>()</code> <br><code><a href="layer_repeat_vector.html">layer_repeat_vector</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_reshape.html">layer_reshape</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_rms_normalization.html">layer_rms_normalization</a>()</code> <br><code><a href="layer_rnn.html">layer_rnn</a>()</code> <br><code><a href="layer_separable_conv_1d.html">layer_separable_conv_1d</a>()</code> <br><code><a href="layer_separable_conv_2d.html">layer_separable_conv_2d</a>()</code> <br><code><a href="layer_simple_rnn.html">layer_simple_rnn</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br><code><a href="layer_spatial_dropout_1d.html">layer_spatial_dropout_1d</a>()</code> <br><code><a href="layer_spatial_dropout_2d.html">layer_spatial_dropout_2d</a>()</code> <br><code><a href="layer_spatial_dropout_3d.html">layer_spatial_dropout_3d</a>()</code> <br><code><a href="layer_spectral_normalization.html">layer_spectral_normalization</a>()</code> <br><code><a href="layer_stft_spectrogram.html">layer_stft_spectrogram</a>()</code> <br><code><a href="layer_string_lookup.html">layer_string_lookup</a>()</code> <br><code><a href="layer_subtract.html">layer_subtract</a>()</code> <br><code><a href="layer_text_vectorization.html">layer_text_vectorization</a>()</code> <br><code><a href="layer_tfsm.html">layer_tfsm</a>()</code> <br><code><a href="layer_time_distributed.html">layer_time_distributed</a>()</code> <br><code><a href="layer_torch_module_wrapper.html">layer_torch_module_wrapper</a>()</code> <br><code><a href="layer_unit_normalization.html">layer_unit_normalization</a>()</code> <br><code><a href="layer_upsampling_1d.html">layer_upsampling_1d</a>()</code> <br><code><a href="layer_upsampling_2d.html">layer_upsampling_2d</a>()</code> <br><code><a href="layer_upsampling_3d.html">layer_upsampling_3d</a>()</code> <br><code><a href="layer_zero_padding_1d.html">layer_zero_padding_1d</a>()</code> <br><code><a href="layer_zero_padding_2d.html">layer_zero_padding_2d</a>()</code> <br><code><a href="layer_zero_padding_3d.html">layer_zero_padding_3d</a>()</code> <br><code><a href="rnn_cell_gru.html">rnn_cell_gru</a>()</code> <br><code><a href="rnn_cell_lstm.html">rnn_cell_lstm</a>()</code> <br><code><a href="rnn_cell_simple.html">rnn_cell_simple</a>()</code> <br><code><a href="rnn_cells_stack.html">rnn_cells_stack</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -126,7 +126,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/layer_additive_attention.html b/docs/dev/reference/layer_additive_attention.html
index a210c3608..2095dd79f 100644
--- a/docs/dev/reference/layer_additive_attention.html
+++ b/docs/dev/reference/layer_additive_attention.html
@@ -40,7 +40,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -171,7 +171,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
     <div class="dont-index">
 <ul><li><p><a href="https://keras.io/api/layers/attention_layers/additive_attention#additiveattention-class" class="external-link">https://keras.io/api/layers/attention_layers/additive_attention#additiveattention-class</a></p></li>
 </ul><p>Other attention layers: <br><code><a href="layer_attention.html">layer_attention</a>()</code> <br><code><a href="layer_group_query_attention.html">layer_group_query_attention</a>()</code> <br><code><a href="layer_multi_head_attention.html">layer_multi_head_attention</a>()</code> <br></p>
-<p>Other layers: <br><code><a href="Layer.html">Layer</a>()</code> <br><code><a href="layer_activation.html">layer_activation</a>()</code> <br><code><a href="layer_activation_elu.html">layer_activation_elu</a>()</code> <br><code><a href="layer_activation_leaky_relu.html">layer_activation_leaky_relu</a>()</code> <br><code><a href="layer_activation_parametric_relu.html">layer_activation_parametric_relu</a>()</code> <br><code><a href="layer_activation_relu.html">layer_activation_relu</a>()</code> <br><code><a href="layer_activation_softmax.html">layer_activation_softmax</a>()</code> <br><code><a href="layer_activity_regularization.html">layer_activity_regularization</a>()</code> <br><code><a href="layer_add.html">layer_add</a>()</code> <br><code><a href="layer_alpha_dropout.html">layer_alpha_dropout</a>()</code> <br><code><a href="layer_attention.html">layer_attention</a>()</code> <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_average.html">layer_average</a>()</code> <br><code><a href="layer_average_pooling_1d.html">layer_average_pooling_1d</a>()</code> <br><code><a href="layer_average_pooling_2d.html">layer_average_pooling_2d</a>()</code> <br><code><a href="layer_average_pooling_3d.html">layer_average_pooling_3d</a>()</code> <br><code><a href="layer_batch_normalization.html">layer_batch_normalization</a>()</code> <br><code><a href="layer_bidirectional.html">layer_bidirectional</a>()</code> <br><code><a href="layer_category_encoding.html">layer_category_encoding</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_concatenate.html">layer_concatenate</a>()</code> <br><code><a href="layer_conv_1d.html">layer_conv_1d</a>()</code> <br><code><a href="layer_conv_1d_transpose.html">layer_conv_1d_transpose</a>()</code> <br><code><a href="layer_conv_2d.html">layer_conv_2d</a>()</code> <br><code><a href="layer_conv_2d_transpose.html">layer_conv_2d_transpose</a>()</code> <br><code><a href="layer_conv_3d.html">layer_conv_3d</a>()</code> <br><code><a href="layer_conv_3d_transpose.html">layer_conv_3d_transpose</a>()</code> <br><code><a href="layer_conv_lstm_1d.html">layer_conv_lstm_1d</a>()</code> <br><code><a href="layer_conv_lstm_2d.html">layer_conv_lstm_2d</a>()</code> <br><code><a href="layer_conv_lstm_3d.html">layer_conv_lstm_3d</a>()</code> <br><code><a href="layer_cropping_1d.html">layer_cropping_1d</a>()</code> <br><code><a href="layer_cropping_2d.html">layer_cropping_2d</a>()</code> <br><code><a href="layer_cropping_3d.html">layer_cropping_3d</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_dense.html">layer_dense</a>()</code> <br><code><a href="layer_depthwise_conv_1d.html">layer_depthwise_conv_1d</a>()</code> <br><code><a href="layer_depthwise_conv_2d.html">layer_depthwise_conv_2d</a>()</code> <br><code><a href="layer_discretization.html">layer_discretization</a>()</code> <br><code><a href="layer_dot.html">layer_dot</a>()</code> <br><code><a href="layer_dropout.html">layer_dropout</a>()</code> <br><code><a href="layer_einsum_dense.html">layer_einsum_dense</a>()</code> <br><code><a href="layer_embedding.html">layer_embedding</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_feature_space.html">layer_feature_space</a>()</code> <br><code><a href="layer_flatten.html">layer_flatten</a>()</code> <br><code><a href="layer_flax_module_wrapper.html">layer_flax_module_wrapper</a>()</code> <br><code><a href="layer_gaussian_dropout.html">layer_gaussian_dropout</a>()</code> <br><code><a href="layer_gaussian_noise.html">layer_gaussian_noise</a>()</code> <br><code><a href="layer_global_average_pooling_1d.html">layer_global_average_pooling_1d</a>()</code> <br><code><a href="layer_global_average_pooling_2d.html">layer_global_average_pooling_2d</a>()</code> <br><code><a href="layer_global_average_pooling_3d.html">layer_global_average_pooling_3d</a>()</code> <br><code><a href="layer_global_max_pooling_1d.html">layer_global_max_pooling_1d</a>()</code> <br><code><a href="layer_global_max_pooling_2d.html">layer_global_max_pooling_2d</a>()</code> <br><code><a href="layer_global_max_pooling_3d.html">layer_global_max_pooling_3d</a>()</code> <br><code><a href="layer_group_normalization.html">layer_group_normalization</a>()</code> <br><code><a href="layer_group_query_attention.html">layer_group_query_attention</a>()</code> <br><code><a href="layer_gru.html">layer_gru</a>()</code> <br><code><a href="layer_hashed_crossing.html">layer_hashed_crossing</a>()</code> <br><code><a href="layer_hashing.html">layer_hashing</a>()</code> <br><code><a href="layer_identity.html">layer_identity</a>()</code> <br><code><a href="layer_integer_lookup.html">layer_integer_lookup</a>()</code> <br><code><a href="layer_jax_model_wrapper.html">layer_jax_model_wrapper</a>()</code> <br><code><a href="layer_lambda.html">layer_lambda</a>()</code> <br><code><a href="layer_layer_normalization.html">layer_layer_normalization</a>()</code> <br><code><a href="layer_lstm.html">layer_lstm</a>()</code> <br><code><a href="layer_masking.html">layer_masking</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_max_pooling_1d.html">layer_max_pooling_1d</a>()</code> <br><code><a href="layer_max_pooling_2d.html">layer_max_pooling_2d</a>()</code> <br><code><a href="layer_max_pooling_3d.html">layer_max_pooling_3d</a>()</code> <br><code><a href="layer_maximum.html">layer_maximum</a>()</code> <br><code><a href="layer_mel_spectrogram.html">layer_mel_spectrogram</a>()</code> <br><code><a href="layer_minimum.html">layer_minimum</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_multi_head_attention.html">layer_multi_head_attention</a>()</code> <br><code><a href="layer_multiply.html">layer_multiply</a>()</code> <br><code><a href="layer_normalization.html">layer_normalization</a>()</code> <br><code><a href="layer_permute.html">layer_permute</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_brightness.html">layer_random_brightness</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_contrast.html">layer_random_contrast</a>()</code> <br><code><a href="layer_random_crop.html">layer_random_crop</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_flip.html">layer_random_flip</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_rotation.html">layer_random_rotation</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_random_translation.html">layer_random_translation</a>()</code> <br><code><a href="layer_random_zoom.html">layer_random_zoom</a>()</code> <br><code><a href="layer_repeat_vector.html">layer_repeat_vector</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_reshape.html">layer_reshape</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_rms_normalization.html">layer_rms_normalization</a>()</code> <br><code><a href="layer_rnn.html">layer_rnn</a>()</code> <br><code><a href="layer_separable_conv_1d.html">layer_separable_conv_1d</a>()</code> <br><code><a href="layer_separable_conv_2d.html">layer_separable_conv_2d</a>()</code> <br><code><a href="layer_simple_rnn.html">layer_simple_rnn</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br><code><a href="layer_spatial_dropout_1d.html">layer_spatial_dropout_1d</a>()</code> <br><code><a href="layer_spatial_dropout_2d.html">layer_spatial_dropout_2d</a>()</code> <br><code><a href="layer_spatial_dropout_3d.html">layer_spatial_dropout_3d</a>()</code> <br><code><a href="layer_spectral_normalization.html">layer_spectral_normalization</a>()</code> <br><code><a href="layer_stft_spectrogram.html">layer_stft_spectrogram</a>()</code> <br><code><a href="layer_string_lookup.html">layer_string_lookup</a>()</code> <br><code><a href="layer_subtract.html">layer_subtract</a>()</code> <br><code><a href="layer_text_vectorization.html">layer_text_vectorization</a>()</code> <br><code><a href="layer_tfsm.html">layer_tfsm</a>()</code> <br><code><a href="layer_time_distributed.html">layer_time_distributed</a>()</code> <br><code><a href="layer_torch_module_wrapper.html">layer_torch_module_wrapper</a>()</code> <br><code><a href="layer_unit_normalization.html">layer_unit_normalization</a>()</code> <br><code><a href="layer_upsampling_1d.html">layer_upsampling_1d</a>()</code> <br><code><a href="layer_upsampling_2d.html">layer_upsampling_2d</a>()</code> <br><code><a href="layer_upsampling_3d.html">layer_upsampling_3d</a>()</code> <br><code><a href="layer_zero_padding_1d.html">layer_zero_padding_1d</a>()</code> <br><code><a href="layer_zero_padding_2d.html">layer_zero_padding_2d</a>()</code> <br><code><a href="layer_zero_padding_3d.html">layer_zero_padding_3d</a>()</code> <br><code><a href="rnn_cell_gru.html">rnn_cell_gru</a>()</code> <br><code><a href="rnn_cell_lstm.html">rnn_cell_lstm</a>()</code> <br><code><a href="rnn_cell_simple.html">rnn_cell_simple</a>()</code> <br><code><a href="rnn_cells_stack.html">rnn_cells_stack</a>()</code> <br></p></div>
+<p>Other layers: <br><code><a href="Layer.html">Layer</a>()</code> <br><code><a href="layer_activation.html">layer_activation</a>()</code> <br><code><a href="layer_activation_elu.html">layer_activation_elu</a>()</code> <br><code><a href="layer_activation_leaky_relu.html">layer_activation_leaky_relu</a>()</code> <br><code><a href="layer_activation_parametric_relu.html">layer_activation_parametric_relu</a>()</code> <br><code><a href="layer_activation_relu.html">layer_activation_relu</a>()</code> <br><code><a href="layer_activation_softmax.html">layer_activation_softmax</a>()</code> <br><code><a href="layer_activity_regularization.html">layer_activity_regularization</a>()</code> <br><code><a href="layer_add.html">layer_add</a>()</code> <br><code><a href="layer_alpha_dropout.html">layer_alpha_dropout</a>()</code> <br><code><a href="layer_attention.html">layer_attention</a>()</code> <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_average.html">layer_average</a>()</code> <br><code><a href="layer_average_pooling_1d.html">layer_average_pooling_1d</a>()</code> <br><code><a href="layer_average_pooling_2d.html">layer_average_pooling_2d</a>()</code> <br><code><a href="layer_average_pooling_3d.html">layer_average_pooling_3d</a>()</code> <br><code><a href="layer_batch_normalization.html">layer_batch_normalization</a>()</code> <br><code><a href="layer_bidirectional.html">layer_bidirectional</a>()</code> <br><code><a href="layer_category_encoding.html">layer_category_encoding</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_concatenate.html">layer_concatenate</a>()</code> <br><code><a href="layer_conv_1d.html">layer_conv_1d</a>()</code> <br><code><a href="layer_conv_1d_transpose.html">layer_conv_1d_transpose</a>()</code> <br><code><a href="layer_conv_2d.html">layer_conv_2d</a>()</code> <br><code><a href="layer_conv_2d_transpose.html">layer_conv_2d_transpose</a>()</code> <br><code><a href="layer_conv_3d.html">layer_conv_3d</a>()</code> <br><code><a href="layer_conv_3d_transpose.html">layer_conv_3d_transpose</a>()</code> <br><code><a href="layer_conv_lstm_1d.html">layer_conv_lstm_1d</a>()</code> <br><code><a href="layer_conv_lstm_2d.html">layer_conv_lstm_2d</a>()</code> <br><code><a href="layer_conv_lstm_3d.html">layer_conv_lstm_3d</a>()</code> <br><code><a href="layer_cropping_1d.html">layer_cropping_1d</a>()</code> <br><code><a href="layer_cropping_2d.html">layer_cropping_2d</a>()</code> <br><code><a href="layer_cropping_3d.html">layer_cropping_3d</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_dense.html">layer_dense</a>()</code> <br><code><a href="layer_depthwise_conv_1d.html">layer_depthwise_conv_1d</a>()</code> <br><code><a href="layer_depthwise_conv_2d.html">layer_depthwise_conv_2d</a>()</code> <br><code><a href="layer_discretization.html">layer_discretization</a>()</code> <br><code><a href="layer_dot.html">layer_dot</a>()</code> <br><code><a href="layer_dropout.html">layer_dropout</a>()</code> <br><code><a href="layer_einsum_dense.html">layer_einsum_dense</a>()</code> <br><code><a href="layer_embedding.html">layer_embedding</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_feature_space.html">layer_feature_space</a>()</code> <br><code><a href="layer_flatten.html">layer_flatten</a>()</code> <br><code><a href="layer_flax_module_wrapper.html">layer_flax_module_wrapper</a>()</code> <br><code><a href="layer_gaussian_dropout.html">layer_gaussian_dropout</a>()</code> <br><code><a href="layer_gaussian_noise.html">layer_gaussian_noise</a>()</code> <br><code><a href="layer_global_average_pooling_1d.html">layer_global_average_pooling_1d</a>()</code> <br><code><a href="layer_global_average_pooling_2d.html">layer_global_average_pooling_2d</a>()</code> <br><code><a href="layer_global_average_pooling_3d.html">layer_global_average_pooling_3d</a>()</code> <br><code><a href="layer_global_max_pooling_1d.html">layer_global_max_pooling_1d</a>()</code> <br><code><a href="layer_global_max_pooling_2d.html">layer_global_max_pooling_2d</a>()</code> <br><code><a href="layer_global_max_pooling_3d.html">layer_global_max_pooling_3d</a>()</code> <br><code><a href="layer_group_normalization.html">layer_group_normalization</a>()</code> <br><code><a href="layer_group_query_attention.html">layer_group_query_attention</a>()</code> <br><code><a href="layer_gru.html">layer_gru</a>()</code> <br><code><a href="layer_hashed_crossing.html">layer_hashed_crossing</a>()</code> <br><code><a href="layer_hashing.html">layer_hashing</a>()</code> <br><code><a href="layer_identity.html">layer_identity</a>()</code> <br><code><a href="layer_integer_lookup.html">layer_integer_lookup</a>()</code> <br><code><a href="layer_jax_model_wrapper.html">layer_jax_model_wrapper</a>()</code> <br><code><a href="layer_lambda.html">layer_lambda</a>()</code> <br><code><a href="layer_layer_normalization.html">layer_layer_normalization</a>()</code> <br><code><a href="layer_lstm.html">layer_lstm</a>()</code> <br><code><a href="layer_masking.html">layer_masking</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_max_pooling_1d.html">layer_max_pooling_1d</a>()</code> <br><code><a href="layer_max_pooling_2d.html">layer_max_pooling_2d</a>()</code> <br><code><a href="layer_max_pooling_3d.html">layer_max_pooling_3d</a>()</code> <br><code><a href="layer_maximum.html">layer_maximum</a>()</code> <br><code><a href="layer_mel_spectrogram.html">layer_mel_spectrogram</a>()</code> <br><code><a href="layer_minimum.html">layer_minimum</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_multi_head_attention.html">layer_multi_head_attention</a>()</code> <br><code><a href="layer_multiply.html">layer_multiply</a>()</code> <br><code><a href="layer_normalization.html">layer_normalization</a>()</code> <br><code><a href="layer_permute.html">layer_permute</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_brightness.html">layer_random_brightness</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_contrast.html">layer_random_contrast</a>()</code> <br><code><a href="layer_random_crop.html">layer_random_crop</a>()</code> <br><code><a href="layer_random_elastic_transform.html">layer_random_elastic_transform</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_flip.html">layer_random_flip</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_rotation.html">layer_random_rotation</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_random_translation.html">layer_random_translation</a>()</code> <br><code><a href="layer_random_zoom.html">layer_random_zoom</a>()</code> <br><code><a href="layer_repeat_vector.html">layer_repeat_vector</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_reshape.html">layer_reshape</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_rms_normalization.html">layer_rms_normalization</a>()</code> <br><code><a href="layer_rnn.html">layer_rnn</a>()</code> <br><code><a href="layer_separable_conv_1d.html">layer_separable_conv_1d</a>()</code> <br><code><a href="layer_separable_conv_2d.html">layer_separable_conv_2d</a>()</code> <br><code><a href="layer_simple_rnn.html">layer_simple_rnn</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br><code><a href="layer_spatial_dropout_1d.html">layer_spatial_dropout_1d</a>()</code> <br><code><a href="layer_spatial_dropout_2d.html">layer_spatial_dropout_2d</a>()</code> <br><code><a href="layer_spatial_dropout_3d.html">layer_spatial_dropout_3d</a>()</code> <br><code><a href="layer_spectral_normalization.html">layer_spectral_normalization</a>()</code> <br><code><a href="layer_stft_spectrogram.html">layer_stft_spectrogram</a>()</code> <br><code><a href="layer_string_lookup.html">layer_string_lookup</a>()</code> <br><code><a href="layer_subtract.html">layer_subtract</a>()</code> <br><code><a href="layer_text_vectorization.html">layer_text_vectorization</a>()</code> <br><code><a href="layer_tfsm.html">layer_tfsm</a>()</code> <br><code><a href="layer_time_distributed.html">layer_time_distributed</a>()</code> <br><code><a href="layer_torch_module_wrapper.html">layer_torch_module_wrapper</a>()</code> <br><code><a href="layer_unit_normalization.html">layer_unit_normalization</a>()</code> <br><code><a href="layer_upsampling_1d.html">layer_upsampling_1d</a>()</code> <br><code><a href="layer_upsampling_2d.html">layer_upsampling_2d</a>()</code> <br><code><a href="layer_upsampling_3d.html">layer_upsampling_3d</a>()</code> <br><code><a href="layer_zero_padding_1d.html">layer_zero_padding_1d</a>()</code> <br><code><a href="layer_zero_padding_2d.html">layer_zero_padding_2d</a>()</code> <br><code><a href="layer_zero_padding_3d.html">layer_zero_padding_3d</a>()</code> <br><code><a href="rnn_cell_gru.html">rnn_cell_gru</a>()</code> <br><code><a href="rnn_cell_lstm.html">rnn_cell_lstm</a>()</code> <br><code><a href="rnn_cell_simple.html">rnn_cell_simple</a>()</code> <br><code><a href="rnn_cells_stack.html">rnn_cells_stack</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -183,7 +183,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/layer_alpha_dropout.html b/docs/dev/reference/layer_alpha_dropout.html
index e717c0741..88590f4af 100644
--- a/docs/dev/reference/layer_alpha_dropout.html
+++ b/docs/dev/reference/layer_alpha_dropout.html
@@ -16,7 +16,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -126,7 +126,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
     <div class="dont-index">
 <ul><li><p><a href="https://www.tensorflow.org/api_docs/python/tf/keras/layers/AlphaDropout" class="external-link">https://www.tensorflow.org/api_docs/python/tf/keras/layers/AlphaDropout</a></p></li>
 </ul><p>Other regularization layers: <br><code><a href="layer_activity_regularization.html">layer_activity_regularization</a>()</code> <br><code><a href="layer_dropout.html">layer_dropout</a>()</code> <br><code><a href="layer_gaussian_dropout.html">layer_gaussian_dropout</a>()</code> <br><code><a href="layer_gaussian_noise.html">layer_gaussian_noise</a>()</code> <br><code><a href="layer_spatial_dropout_1d.html">layer_spatial_dropout_1d</a>()</code> <br><code><a href="layer_spatial_dropout_2d.html">layer_spatial_dropout_2d</a>()</code> <br><code><a href="layer_spatial_dropout_3d.html">layer_spatial_dropout_3d</a>()</code> <br></p>
-<p>Other layers: <br><code><a href="Layer.html">Layer</a>()</code> <br><code><a href="layer_activation.html">layer_activation</a>()</code> <br><code><a href="layer_activation_elu.html">layer_activation_elu</a>()</code> <br><code><a href="layer_activation_leaky_relu.html">layer_activation_leaky_relu</a>()</code> <br><code><a href="layer_activation_parametric_relu.html">layer_activation_parametric_relu</a>()</code> <br><code><a href="layer_activation_relu.html">layer_activation_relu</a>()</code> <br><code><a href="layer_activation_softmax.html">layer_activation_softmax</a>()</code> <br><code><a href="layer_activity_regularization.html">layer_activity_regularization</a>()</code> <br><code><a href="layer_add.html">layer_add</a>()</code> <br><code><a href="layer_additive_attention.html">layer_additive_attention</a>()</code> <br><code><a href="layer_attention.html">layer_attention</a>()</code> <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_average.html">layer_average</a>()</code> <br><code><a href="layer_average_pooling_1d.html">layer_average_pooling_1d</a>()</code> <br><code><a href="layer_average_pooling_2d.html">layer_average_pooling_2d</a>()</code> <br><code><a href="layer_average_pooling_3d.html">layer_average_pooling_3d</a>()</code> <br><code><a href="layer_batch_normalization.html">layer_batch_normalization</a>()</code> <br><code><a href="layer_bidirectional.html">layer_bidirectional</a>()</code> <br><code><a href="layer_category_encoding.html">layer_category_encoding</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_concatenate.html">layer_concatenate</a>()</code> <br><code><a href="layer_conv_1d.html">layer_conv_1d</a>()</code> <br><code><a href="layer_conv_1d_transpose.html">layer_conv_1d_transpose</a>()</code> <br><code><a href="layer_conv_2d.html">layer_conv_2d</a>()</code> <br><code><a href="layer_conv_2d_transpose.html">layer_conv_2d_transpose</a>()</code> <br><code><a href="layer_conv_3d.html">layer_conv_3d</a>()</code> <br><code><a href="layer_conv_3d_transpose.html">layer_conv_3d_transpose</a>()</code> <br><code><a href="layer_conv_lstm_1d.html">layer_conv_lstm_1d</a>()</code> <br><code><a href="layer_conv_lstm_2d.html">layer_conv_lstm_2d</a>()</code> <br><code><a href="layer_conv_lstm_3d.html">layer_conv_lstm_3d</a>()</code> <br><code><a href="layer_cropping_1d.html">layer_cropping_1d</a>()</code> <br><code><a href="layer_cropping_2d.html">layer_cropping_2d</a>()</code> <br><code><a href="layer_cropping_3d.html">layer_cropping_3d</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_dense.html">layer_dense</a>()</code> <br><code><a href="layer_depthwise_conv_1d.html">layer_depthwise_conv_1d</a>()</code> <br><code><a href="layer_depthwise_conv_2d.html">layer_depthwise_conv_2d</a>()</code> <br><code><a href="layer_discretization.html">layer_discretization</a>()</code> <br><code><a href="layer_dot.html">layer_dot</a>()</code> <br><code><a href="layer_dropout.html">layer_dropout</a>()</code> <br><code><a href="layer_einsum_dense.html">layer_einsum_dense</a>()</code> <br><code><a href="layer_embedding.html">layer_embedding</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_feature_space.html">layer_feature_space</a>()</code> <br><code><a href="layer_flatten.html">layer_flatten</a>()</code> <br><code><a href="layer_flax_module_wrapper.html">layer_flax_module_wrapper</a>()</code> <br><code><a href="layer_gaussian_dropout.html">layer_gaussian_dropout</a>()</code> <br><code><a href="layer_gaussian_noise.html">layer_gaussian_noise</a>()</code> <br><code><a href="layer_global_average_pooling_1d.html">layer_global_average_pooling_1d</a>()</code> <br><code><a href="layer_global_average_pooling_2d.html">layer_global_average_pooling_2d</a>()</code> <br><code><a href="layer_global_average_pooling_3d.html">layer_global_average_pooling_3d</a>()</code> <br><code><a href="layer_global_max_pooling_1d.html">layer_global_max_pooling_1d</a>()</code> <br><code><a href="layer_global_max_pooling_2d.html">layer_global_max_pooling_2d</a>()</code> <br><code><a href="layer_global_max_pooling_3d.html">layer_global_max_pooling_3d</a>()</code> <br><code><a href="layer_group_normalization.html">layer_group_normalization</a>()</code> <br><code><a href="layer_group_query_attention.html">layer_group_query_attention</a>()</code> <br><code><a href="layer_gru.html">layer_gru</a>()</code> <br><code><a href="layer_hashed_crossing.html">layer_hashed_crossing</a>()</code> <br><code><a href="layer_hashing.html">layer_hashing</a>()</code> <br><code><a href="layer_identity.html">layer_identity</a>()</code> <br><code><a href="layer_integer_lookup.html">layer_integer_lookup</a>()</code> <br><code><a href="layer_jax_model_wrapper.html">layer_jax_model_wrapper</a>()</code> <br><code><a href="layer_lambda.html">layer_lambda</a>()</code> <br><code><a href="layer_layer_normalization.html">layer_layer_normalization</a>()</code> <br><code><a href="layer_lstm.html">layer_lstm</a>()</code> <br><code><a href="layer_masking.html">layer_masking</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_max_pooling_1d.html">layer_max_pooling_1d</a>()</code> <br><code><a href="layer_max_pooling_2d.html">layer_max_pooling_2d</a>()</code> <br><code><a href="layer_max_pooling_3d.html">layer_max_pooling_3d</a>()</code> <br><code><a href="layer_maximum.html">layer_maximum</a>()</code> <br><code><a href="layer_mel_spectrogram.html">layer_mel_spectrogram</a>()</code> <br><code><a href="layer_minimum.html">layer_minimum</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_multi_head_attention.html">layer_multi_head_attention</a>()</code> <br><code><a href="layer_multiply.html">layer_multiply</a>()</code> <br><code><a href="layer_normalization.html">layer_normalization</a>()</code> <br><code><a href="layer_permute.html">layer_permute</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_brightness.html">layer_random_brightness</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_contrast.html">layer_random_contrast</a>()</code> <br><code><a href="layer_random_crop.html">layer_random_crop</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_flip.html">layer_random_flip</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_rotation.html">layer_random_rotation</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_random_translation.html">layer_random_translation</a>()</code> <br><code><a href="layer_random_zoom.html">layer_random_zoom</a>()</code> <br><code><a href="layer_repeat_vector.html">layer_repeat_vector</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_reshape.html">layer_reshape</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_rms_normalization.html">layer_rms_normalization</a>()</code> <br><code><a href="layer_rnn.html">layer_rnn</a>()</code> <br><code><a href="layer_separable_conv_1d.html">layer_separable_conv_1d</a>()</code> <br><code><a href="layer_separable_conv_2d.html">layer_separable_conv_2d</a>()</code> <br><code><a href="layer_simple_rnn.html">layer_simple_rnn</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br><code><a href="layer_spatial_dropout_1d.html">layer_spatial_dropout_1d</a>()</code> <br><code><a href="layer_spatial_dropout_2d.html">layer_spatial_dropout_2d</a>()</code> <br><code><a href="layer_spatial_dropout_3d.html">layer_spatial_dropout_3d</a>()</code> <br><code><a href="layer_spectral_normalization.html">layer_spectral_normalization</a>()</code> <br><code><a href="layer_stft_spectrogram.html">layer_stft_spectrogram</a>()</code> <br><code><a href="layer_string_lookup.html">layer_string_lookup</a>()</code> <br><code><a href="layer_subtract.html">layer_subtract</a>()</code> <br><code><a href="layer_text_vectorization.html">layer_text_vectorization</a>()</code> <br><code><a href="layer_tfsm.html">layer_tfsm</a>()</code> <br><code><a href="layer_time_distributed.html">layer_time_distributed</a>()</code> <br><code><a href="layer_torch_module_wrapper.html">layer_torch_module_wrapper</a>()</code> <br><code><a href="layer_unit_normalization.html">layer_unit_normalization</a>()</code> <br><code><a href="layer_upsampling_1d.html">layer_upsampling_1d</a>()</code> <br><code><a href="layer_upsampling_2d.html">layer_upsampling_2d</a>()</code> <br><code><a href="layer_upsampling_3d.html">layer_upsampling_3d</a>()</code> <br><code><a href="layer_zero_padding_1d.html">layer_zero_padding_1d</a>()</code> <br><code><a href="layer_zero_padding_2d.html">layer_zero_padding_2d</a>()</code> <br><code><a href="layer_zero_padding_3d.html">layer_zero_padding_3d</a>()</code> <br><code><a href="rnn_cell_gru.html">rnn_cell_gru</a>()</code> <br><code><a href="rnn_cell_lstm.html">rnn_cell_lstm</a>()</code> <br><code><a href="rnn_cell_simple.html">rnn_cell_simple</a>()</code> <br><code><a href="rnn_cells_stack.html">rnn_cells_stack</a>()</code> <br></p></div>
+<p>Other layers: <br><code><a href="Layer.html">Layer</a>()</code> <br><code><a href="layer_activation.html">layer_activation</a>()</code> <br><code><a href="layer_activation_elu.html">layer_activation_elu</a>()</code> <br><code><a href="layer_activation_leaky_relu.html">layer_activation_leaky_relu</a>()</code> <br><code><a href="layer_activation_parametric_relu.html">layer_activation_parametric_relu</a>()</code> <br><code><a href="layer_activation_relu.html">layer_activation_relu</a>()</code> <br><code><a href="layer_activation_softmax.html">layer_activation_softmax</a>()</code> <br><code><a href="layer_activity_regularization.html">layer_activity_regularization</a>()</code> <br><code><a href="layer_add.html">layer_add</a>()</code> <br><code><a href="layer_additive_attention.html">layer_additive_attention</a>()</code> <br><code><a href="layer_attention.html">layer_attention</a>()</code> <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_average.html">layer_average</a>()</code> <br><code><a href="layer_average_pooling_1d.html">layer_average_pooling_1d</a>()</code> <br><code><a href="layer_average_pooling_2d.html">layer_average_pooling_2d</a>()</code> <br><code><a href="layer_average_pooling_3d.html">layer_average_pooling_3d</a>()</code> <br><code><a href="layer_batch_normalization.html">layer_batch_normalization</a>()</code> <br><code><a href="layer_bidirectional.html">layer_bidirectional</a>()</code> <br><code><a href="layer_category_encoding.html">layer_category_encoding</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_concatenate.html">layer_concatenate</a>()</code> <br><code><a href="layer_conv_1d.html">layer_conv_1d</a>()</code> <br><code><a href="layer_conv_1d_transpose.html">layer_conv_1d_transpose</a>()</code> <br><code><a href="layer_conv_2d.html">layer_conv_2d</a>()</code> <br><code><a href="layer_conv_2d_transpose.html">layer_conv_2d_transpose</a>()</code> <br><code><a href="layer_conv_3d.html">layer_conv_3d</a>()</code> <br><code><a href="layer_conv_3d_transpose.html">layer_conv_3d_transpose</a>()</code> <br><code><a href="layer_conv_lstm_1d.html">layer_conv_lstm_1d</a>()</code> <br><code><a href="layer_conv_lstm_2d.html">layer_conv_lstm_2d</a>()</code> <br><code><a href="layer_conv_lstm_3d.html">layer_conv_lstm_3d</a>()</code> <br><code><a href="layer_cropping_1d.html">layer_cropping_1d</a>()</code> <br><code><a href="layer_cropping_2d.html">layer_cropping_2d</a>()</code> <br><code><a href="layer_cropping_3d.html">layer_cropping_3d</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_dense.html">layer_dense</a>()</code> <br><code><a href="layer_depthwise_conv_1d.html">layer_depthwise_conv_1d</a>()</code> <br><code><a href="layer_depthwise_conv_2d.html">layer_depthwise_conv_2d</a>()</code> <br><code><a href="layer_discretization.html">layer_discretization</a>()</code> <br><code><a href="layer_dot.html">layer_dot</a>()</code> <br><code><a href="layer_dropout.html">layer_dropout</a>()</code> <br><code><a href="layer_einsum_dense.html">layer_einsum_dense</a>()</code> <br><code><a href="layer_embedding.html">layer_embedding</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_feature_space.html">layer_feature_space</a>()</code> <br><code><a href="layer_flatten.html">layer_flatten</a>()</code> <br><code><a href="layer_flax_module_wrapper.html">layer_flax_module_wrapper</a>()</code> <br><code><a href="layer_gaussian_dropout.html">layer_gaussian_dropout</a>()</code> <br><code><a href="layer_gaussian_noise.html">layer_gaussian_noise</a>()</code> <br><code><a href="layer_global_average_pooling_1d.html">layer_global_average_pooling_1d</a>()</code> <br><code><a href="layer_global_average_pooling_2d.html">layer_global_average_pooling_2d</a>()</code> <br><code><a href="layer_global_average_pooling_3d.html">layer_global_average_pooling_3d</a>()</code> <br><code><a href="layer_global_max_pooling_1d.html">layer_global_max_pooling_1d</a>()</code> <br><code><a href="layer_global_max_pooling_2d.html">layer_global_max_pooling_2d</a>()</code> <br><code><a href="layer_global_max_pooling_3d.html">layer_global_max_pooling_3d</a>()</code> <br><code><a href="layer_group_normalization.html">layer_group_normalization</a>()</code> <br><code><a href="layer_group_query_attention.html">layer_group_query_attention</a>()</code> <br><code><a href="layer_gru.html">layer_gru</a>()</code> <br><code><a href="layer_hashed_crossing.html">layer_hashed_crossing</a>()</code> <br><code><a href="layer_hashing.html">layer_hashing</a>()</code> <br><code><a href="layer_identity.html">layer_identity</a>()</code> <br><code><a href="layer_integer_lookup.html">layer_integer_lookup</a>()</code> <br><code><a href="layer_jax_model_wrapper.html">layer_jax_model_wrapper</a>()</code> <br><code><a href="layer_lambda.html">layer_lambda</a>()</code> <br><code><a href="layer_layer_normalization.html">layer_layer_normalization</a>()</code> <br><code><a href="layer_lstm.html">layer_lstm</a>()</code> <br><code><a href="layer_masking.html">layer_masking</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_max_pooling_1d.html">layer_max_pooling_1d</a>()</code> <br><code><a href="layer_max_pooling_2d.html">layer_max_pooling_2d</a>()</code> <br><code><a href="layer_max_pooling_3d.html">layer_max_pooling_3d</a>()</code> <br><code><a href="layer_maximum.html">layer_maximum</a>()</code> <br><code><a href="layer_mel_spectrogram.html">layer_mel_spectrogram</a>()</code> <br><code><a href="layer_minimum.html">layer_minimum</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_multi_head_attention.html">layer_multi_head_attention</a>()</code> <br><code><a href="layer_multiply.html">layer_multiply</a>()</code> <br><code><a href="layer_normalization.html">layer_normalization</a>()</code> <br><code><a href="layer_permute.html">layer_permute</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_brightness.html">layer_random_brightness</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_contrast.html">layer_random_contrast</a>()</code> <br><code><a href="layer_random_crop.html">layer_random_crop</a>()</code> <br><code><a href="layer_random_elastic_transform.html">layer_random_elastic_transform</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_flip.html">layer_random_flip</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_rotation.html">layer_random_rotation</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_random_translation.html">layer_random_translation</a>()</code> <br><code><a href="layer_random_zoom.html">layer_random_zoom</a>()</code> <br><code><a href="layer_repeat_vector.html">layer_repeat_vector</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_reshape.html">layer_reshape</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_rms_normalization.html">layer_rms_normalization</a>()</code> <br><code><a href="layer_rnn.html">layer_rnn</a>()</code> <br><code><a href="layer_separable_conv_1d.html">layer_separable_conv_1d</a>()</code> <br><code><a href="layer_separable_conv_2d.html">layer_separable_conv_2d</a>()</code> <br><code><a href="layer_simple_rnn.html">layer_simple_rnn</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br><code><a href="layer_spatial_dropout_1d.html">layer_spatial_dropout_1d</a>()</code> <br><code><a href="layer_spatial_dropout_2d.html">layer_spatial_dropout_2d</a>()</code> <br><code><a href="layer_spatial_dropout_3d.html">layer_spatial_dropout_3d</a>()</code> <br><code><a href="layer_spectral_normalization.html">layer_spectral_normalization</a>()</code> <br><code><a href="layer_stft_spectrogram.html">layer_stft_spectrogram</a>()</code> <br><code><a href="layer_string_lookup.html">layer_string_lookup</a>()</code> <br><code><a href="layer_subtract.html">layer_subtract</a>()</code> <br><code><a href="layer_text_vectorization.html">layer_text_vectorization</a>()</code> <br><code><a href="layer_tfsm.html">layer_tfsm</a>()</code> <br><code><a href="layer_time_distributed.html">layer_time_distributed</a>()</code> <br><code><a href="layer_torch_module_wrapper.html">layer_torch_module_wrapper</a>()</code> <br><code><a href="layer_unit_normalization.html">layer_unit_normalization</a>()</code> <br><code><a href="layer_upsampling_1d.html">layer_upsampling_1d</a>()</code> <br><code><a href="layer_upsampling_2d.html">layer_upsampling_2d</a>()</code> <br><code><a href="layer_upsampling_3d.html">layer_upsampling_3d</a>()</code> <br><code><a href="layer_zero_padding_1d.html">layer_zero_padding_1d</a>()</code> <br><code><a href="layer_zero_padding_2d.html">layer_zero_padding_2d</a>()</code> <br><code><a href="layer_zero_padding_3d.html">layer_zero_padding_3d</a>()</code> <br><code><a href="rnn_cell_gru.html">rnn_cell_gru</a>()</code> <br><code><a href="rnn_cell_lstm.html">rnn_cell_lstm</a>()</code> <br><code><a href="rnn_cell_simple.html">rnn_cell_simple</a>()</code> <br><code><a href="rnn_cells_stack.html">rnn_cells_stack</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -138,7 +138,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/layer_attention.html b/docs/dev/reference/layer_attention.html
index bf45321d8..c141ba2b9 100644
--- a/docs/dev/reference/layer_attention.html
+++ b/docs/dev/reference/layer_attention.html
@@ -38,7 +38,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -186,7 +186,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
     <div class="dont-index">
 <ul><li><p><a href="https://keras.io/api/layers/attention_layers/attention#attention-class" class="external-link">https://keras.io/api/layers/attention_layers/attention#attention-class</a></p></li>
 </ul><p>Other attention layers: <br><code><a href="layer_additive_attention.html">layer_additive_attention</a>()</code> <br><code><a href="layer_group_query_attention.html">layer_group_query_attention</a>()</code> <br><code><a href="layer_multi_head_attention.html">layer_multi_head_attention</a>()</code> <br></p>
-<p>Other layers: <br><code><a href="Layer.html">Layer</a>()</code> <br><code><a href="layer_activation.html">layer_activation</a>()</code> <br><code><a href="layer_activation_elu.html">layer_activation_elu</a>()</code> <br><code><a href="layer_activation_leaky_relu.html">layer_activation_leaky_relu</a>()</code> <br><code><a href="layer_activation_parametric_relu.html">layer_activation_parametric_relu</a>()</code> <br><code><a href="layer_activation_relu.html">layer_activation_relu</a>()</code> <br><code><a href="layer_activation_softmax.html">layer_activation_softmax</a>()</code> <br><code><a href="layer_activity_regularization.html">layer_activity_regularization</a>()</code> <br><code><a href="layer_add.html">layer_add</a>()</code> <br><code><a href="layer_additive_attention.html">layer_additive_attention</a>()</code> <br><code><a href="layer_alpha_dropout.html">layer_alpha_dropout</a>()</code> <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_average.html">layer_average</a>()</code> <br><code><a href="layer_average_pooling_1d.html">layer_average_pooling_1d</a>()</code> <br><code><a href="layer_average_pooling_2d.html">layer_average_pooling_2d</a>()</code> <br><code><a href="layer_average_pooling_3d.html">layer_average_pooling_3d</a>()</code> <br><code><a href="layer_batch_normalization.html">layer_batch_normalization</a>()</code> <br><code><a href="layer_bidirectional.html">layer_bidirectional</a>()</code> <br><code><a href="layer_category_encoding.html">layer_category_encoding</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_concatenate.html">layer_concatenate</a>()</code> <br><code><a href="layer_conv_1d.html">layer_conv_1d</a>()</code> <br><code><a href="layer_conv_1d_transpose.html">layer_conv_1d_transpose</a>()</code> <br><code><a href="layer_conv_2d.html">layer_conv_2d</a>()</code> <br><code><a href="layer_conv_2d_transpose.html">layer_conv_2d_transpose</a>()</code> <br><code><a href="layer_conv_3d.html">layer_conv_3d</a>()</code> <br><code><a href="layer_conv_3d_transpose.html">layer_conv_3d_transpose</a>()</code> <br><code><a href="layer_conv_lstm_1d.html">layer_conv_lstm_1d</a>()</code> <br><code><a href="layer_conv_lstm_2d.html">layer_conv_lstm_2d</a>()</code> <br><code><a href="layer_conv_lstm_3d.html">layer_conv_lstm_3d</a>()</code> <br><code><a href="layer_cropping_1d.html">layer_cropping_1d</a>()</code> <br><code><a href="layer_cropping_2d.html">layer_cropping_2d</a>()</code> <br><code><a href="layer_cropping_3d.html">layer_cropping_3d</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_dense.html">layer_dense</a>()</code> <br><code><a href="layer_depthwise_conv_1d.html">layer_depthwise_conv_1d</a>()</code> <br><code><a href="layer_depthwise_conv_2d.html">layer_depthwise_conv_2d</a>()</code> <br><code><a href="layer_discretization.html">layer_discretization</a>()</code> <br><code><a href="layer_dot.html">layer_dot</a>()</code> <br><code><a href="layer_dropout.html">layer_dropout</a>()</code> <br><code><a href="layer_einsum_dense.html">layer_einsum_dense</a>()</code> <br><code><a href="layer_embedding.html">layer_embedding</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_feature_space.html">layer_feature_space</a>()</code> <br><code><a href="layer_flatten.html">layer_flatten</a>()</code> <br><code><a href="layer_flax_module_wrapper.html">layer_flax_module_wrapper</a>()</code> <br><code><a href="layer_gaussian_dropout.html">layer_gaussian_dropout</a>()</code> <br><code><a href="layer_gaussian_noise.html">layer_gaussian_noise</a>()</code> <br><code><a href="layer_global_average_pooling_1d.html">layer_global_average_pooling_1d</a>()</code> <br><code><a href="layer_global_average_pooling_2d.html">layer_global_average_pooling_2d</a>()</code> <br><code><a href="layer_global_average_pooling_3d.html">layer_global_average_pooling_3d</a>()</code> <br><code><a href="layer_global_max_pooling_1d.html">layer_global_max_pooling_1d</a>()</code> <br><code><a href="layer_global_max_pooling_2d.html">layer_global_max_pooling_2d</a>()</code> <br><code><a href="layer_global_max_pooling_3d.html">layer_global_max_pooling_3d</a>()</code> <br><code><a href="layer_group_normalization.html">layer_group_normalization</a>()</code> <br><code><a href="layer_group_query_attention.html">layer_group_query_attention</a>()</code> <br><code><a href="layer_gru.html">layer_gru</a>()</code> <br><code><a href="layer_hashed_crossing.html">layer_hashed_crossing</a>()</code> <br><code><a href="layer_hashing.html">layer_hashing</a>()</code> <br><code><a href="layer_identity.html">layer_identity</a>()</code> <br><code><a href="layer_integer_lookup.html">layer_integer_lookup</a>()</code> <br><code><a href="layer_jax_model_wrapper.html">layer_jax_model_wrapper</a>()</code> <br><code><a href="layer_lambda.html">layer_lambda</a>()</code> <br><code><a href="layer_layer_normalization.html">layer_layer_normalization</a>()</code> <br><code><a href="layer_lstm.html">layer_lstm</a>()</code> <br><code><a href="layer_masking.html">layer_masking</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_max_pooling_1d.html">layer_max_pooling_1d</a>()</code> <br><code><a href="layer_max_pooling_2d.html">layer_max_pooling_2d</a>()</code> <br><code><a href="layer_max_pooling_3d.html">layer_max_pooling_3d</a>()</code> <br><code><a href="layer_maximum.html">layer_maximum</a>()</code> <br><code><a href="layer_mel_spectrogram.html">layer_mel_spectrogram</a>()</code> <br><code><a href="layer_minimum.html">layer_minimum</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_multi_head_attention.html">layer_multi_head_attention</a>()</code> <br><code><a href="layer_multiply.html">layer_multiply</a>()</code> <br><code><a href="layer_normalization.html">layer_normalization</a>()</code> <br><code><a href="layer_permute.html">layer_permute</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_brightness.html">layer_random_brightness</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_contrast.html">layer_random_contrast</a>()</code> <br><code><a href="layer_random_crop.html">layer_random_crop</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_flip.html">layer_random_flip</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_rotation.html">layer_random_rotation</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_random_translation.html">layer_random_translation</a>()</code> <br><code><a href="layer_random_zoom.html">layer_random_zoom</a>()</code> <br><code><a href="layer_repeat_vector.html">layer_repeat_vector</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_reshape.html">layer_reshape</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_rms_normalization.html">layer_rms_normalization</a>()</code> <br><code><a href="layer_rnn.html">layer_rnn</a>()</code> <br><code><a href="layer_separable_conv_1d.html">layer_separable_conv_1d</a>()</code> <br><code><a href="layer_separable_conv_2d.html">layer_separable_conv_2d</a>()</code> <br><code><a href="layer_simple_rnn.html">layer_simple_rnn</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br><code><a href="layer_spatial_dropout_1d.html">layer_spatial_dropout_1d</a>()</code> <br><code><a href="layer_spatial_dropout_2d.html">layer_spatial_dropout_2d</a>()</code> <br><code><a href="layer_spatial_dropout_3d.html">layer_spatial_dropout_3d</a>()</code> <br><code><a href="layer_spectral_normalization.html">layer_spectral_normalization</a>()</code> <br><code><a href="layer_stft_spectrogram.html">layer_stft_spectrogram</a>()</code> <br><code><a href="layer_string_lookup.html">layer_string_lookup</a>()</code> <br><code><a href="layer_subtract.html">layer_subtract</a>()</code> <br><code><a href="layer_text_vectorization.html">layer_text_vectorization</a>()</code> <br><code><a href="layer_tfsm.html">layer_tfsm</a>()</code> <br><code><a href="layer_time_distributed.html">layer_time_distributed</a>()</code> <br><code><a href="layer_torch_module_wrapper.html">layer_torch_module_wrapper</a>()</code> <br><code><a href="layer_unit_normalization.html">layer_unit_normalization</a>()</code> <br><code><a href="layer_upsampling_1d.html">layer_upsampling_1d</a>()</code> <br><code><a href="layer_upsampling_2d.html">layer_upsampling_2d</a>()</code> <br><code><a href="layer_upsampling_3d.html">layer_upsampling_3d</a>()</code> <br><code><a href="layer_zero_padding_1d.html">layer_zero_padding_1d</a>()</code> <br><code><a href="layer_zero_padding_2d.html">layer_zero_padding_2d</a>()</code> <br><code><a href="layer_zero_padding_3d.html">layer_zero_padding_3d</a>()</code> <br><code><a href="rnn_cell_gru.html">rnn_cell_gru</a>()</code> <br><code><a href="rnn_cell_lstm.html">rnn_cell_lstm</a>()</code> <br><code><a href="rnn_cell_simple.html">rnn_cell_simple</a>()</code> <br><code><a href="rnn_cells_stack.html">rnn_cells_stack</a>()</code> <br></p></div>
+<p>Other layers: <br><code><a href="Layer.html">Layer</a>()</code> <br><code><a href="layer_activation.html">layer_activation</a>()</code> <br><code><a href="layer_activation_elu.html">layer_activation_elu</a>()</code> <br><code><a href="layer_activation_leaky_relu.html">layer_activation_leaky_relu</a>()</code> <br><code><a href="layer_activation_parametric_relu.html">layer_activation_parametric_relu</a>()</code> <br><code><a href="layer_activation_relu.html">layer_activation_relu</a>()</code> <br><code><a href="layer_activation_softmax.html">layer_activation_softmax</a>()</code> <br><code><a href="layer_activity_regularization.html">layer_activity_regularization</a>()</code> <br><code><a href="layer_add.html">layer_add</a>()</code> <br><code><a href="layer_additive_attention.html">layer_additive_attention</a>()</code> <br><code><a href="layer_alpha_dropout.html">layer_alpha_dropout</a>()</code> <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_average.html">layer_average</a>()</code> <br><code><a href="layer_average_pooling_1d.html">layer_average_pooling_1d</a>()</code> <br><code><a href="layer_average_pooling_2d.html">layer_average_pooling_2d</a>()</code> <br><code><a href="layer_average_pooling_3d.html">layer_average_pooling_3d</a>()</code> <br><code><a href="layer_batch_normalization.html">layer_batch_normalization</a>()</code> <br><code><a href="layer_bidirectional.html">layer_bidirectional</a>()</code> <br><code><a href="layer_category_encoding.html">layer_category_encoding</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_concatenate.html">layer_concatenate</a>()</code> <br><code><a href="layer_conv_1d.html">layer_conv_1d</a>()</code> <br><code><a href="layer_conv_1d_transpose.html">layer_conv_1d_transpose</a>()</code> <br><code><a href="layer_conv_2d.html">layer_conv_2d</a>()</code> <br><code><a href="layer_conv_2d_transpose.html">layer_conv_2d_transpose</a>()</code> <br><code><a href="layer_conv_3d.html">layer_conv_3d</a>()</code> <br><code><a href="layer_conv_3d_transpose.html">layer_conv_3d_transpose</a>()</code> <br><code><a href="layer_conv_lstm_1d.html">layer_conv_lstm_1d</a>()</code> <br><code><a href="layer_conv_lstm_2d.html">layer_conv_lstm_2d</a>()</code> <br><code><a href="layer_conv_lstm_3d.html">layer_conv_lstm_3d</a>()</code> <br><code><a href="layer_cropping_1d.html">layer_cropping_1d</a>()</code> <br><code><a href="layer_cropping_2d.html">layer_cropping_2d</a>()</code> <br><code><a href="layer_cropping_3d.html">layer_cropping_3d</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_dense.html">layer_dense</a>()</code> <br><code><a href="layer_depthwise_conv_1d.html">layer_depthwise_conv_1d</a>()</code> <br><code><a href="layer_depthwise_conv_2d.html">layer_depthwise_conv_2d</a>()</code> <br><code><a href="layer_discretization.html">layer_discretization</a>()</code> <br><code><a href="layer_dot.html">layer_dot</a>()</code> <br><code><a href="layer_dropout.html">layer_dropout</a>()</code> <br><code><a href="layer_einsum_dense.html">layer_einsum_dense</a>()</code> <br><code><a href="layer_embedding.html">layer_embedding</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_feature_space.html">layer_feature_space</a>()</code> <br><code><a href="layer_flatten.html">layer_flatten</a>()</code> <br><code><a href="layer_flax_module_wrapper.html">layer_flax_module_wrapper</a>()</code> <br><code><a href="layer_gaussian_dropout.html">layer_gaussian_dropout</a>()</code> <br><code><a href="layer_gaussian_noise.html">layer_gaussian_noise</a>()</code> <br><code><a href="layer_global_average_pooling_1d.html">layer_global_average_pooling_1d</a>()</code> <br><code><a href="layer_global_average_pooling_2d.html">layer_global_average_pooling_2d</a>()</code> <br><code><a href="layer_global_average_pooling_3d.html">layer_global_average_pooling_3d</a>()</code> <br><code><a href="layer_global_max_pooling_1d.html">layer_global_max_pooling_1d</a>()</code> <br><code><a href="layer_global_max_pooling_2d.html">layer_global_max_pooling_2d</a>()</code> <br><code><a href="layer_global_max_pooling_3d.html">layer_global_max_pooling_3d</a>()</code> <br><code><a href="layer_group_normalization.html">layer_group_normalization</a>()</code> <br><code><a href="layer_group_query_attention.html">layer_group_query_attention</a>()</code> <br><code><a href="layer_gru.html">layer_gru</a>()</code> <br><code><a href="layer_hashed_crossing.html">layer_hashed_crossing</a>()</code> <br><code><a href="layer_hashing.html">layer_hashing</a>()</code> <br><code><a href="layer_identity.html">layer_identity</a>()</code> <br><code><a href="layer_integer_lookup.html">layer_integer_lookup</a>()</code> <br><code><a href="layer_jax_model_wrapper.html">layer_jax_model_wrapper</a>()</code> <br><code><a href="layer_lambda.html">layer_lambda</a>()</code> <br><code><a href="layer_layer_normalization.html">layer_layer_normalization</a>()</code> <br><code><a href="layer_lstm.html">layer_lstm</a>()</code> <br><code><a href="layer_masking.html">layer_masking</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_max_pooling_1d.html">layer_max_pooling_1d</a>()</code> <br><code><a href="layer_max_pooling_2d.html">layer_max_pooling_2d</a>()</code> <br><code><a href="layer_max_pooling_3d.html">layer_max_pooling_3d</a>()</code> <br><code><a href="layer_maximum.html">layer_maximum</a>()</code> <br><code><a href="layer_mel_spectrogram.html">layer_mel_spectrogram</a>()</code> <br><code><a href="layer_minimum.html">layer_minimum</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_multi_head_attention.html">layer_multi_head_attention</a>()</code> <br><code><a href="layer_multiply.html">layer_multiply</a>()</code> <br><code><a href="layer_normalization.html">layer_normalization</a>()</code> <br><code><a href="layer_permute.html">layer_permute</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_brightness.html">layer_random_brightness</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_contrast.html">layer_random_contrast</a>()</code> <br><code><a href="layer_random_crop.html">layer_random_crop</a>()</code> <br><code><a href="layer_random_elastic_transform.html">layer_random_elastic_transform</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_flip.html">layer_random_flip</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_rotation.html">layer_random_rotation</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_random_translation.html">layer_random_translation</a>()</code> <br><code><a href="layer_random_zoom.html">layer_random_zoom</a>()</code> <br><code><a href="layer_repeat_vector.html">layer_repeat_vector</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_reshape.html">layer_reshape</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_rms_normalization.html">layer_rms_normalization</a>()</code> <br><code><a href="layer_rnn.html">layer_rnn</a>()</code> <br><code><a href="layer_separable_conv_1d.html">layer_separable_conv_1d</a>()</code> <br><code><a href="layer_separable_conv_2d.html">layer_separable_conv_2d</a>()</code> <br><code><a href="layer_simple_rnn.html">layer_simple_rnn</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br><code><a href="layer_spatial_dropout_1d.html">layer_spatial_dropout_1d</a>()</code> <br><code><a href="layer_spatial_dropout_2d.html">layer_spatial_dropout_2d</a>()</code> <br><code><a href="layer_spatial_dropout_3d.html">layer_spatial_dropout_3d</a>()</code> <br><code><a href="layer_spectral_normalization.html">layer_spectral_normalization</a>()</code> <br><code><a href="layer_stft_spectrogram.html">layer_stft_spectrogram</a>()</code> <br><code><a href="layer_string_lookup.html">layer_string_lookup</a>()</code> <br><code><a href="layer_subtract.html">layer_subtract</a>()</code> <br><code><a href="layer_text_vectorization.html">layer_text_vectorization</a>()</code> <br><code><a href="layer_tfsm.html">layer_tfsm</a>()</code> <br><code><a href="layer_time_distributed.html">layer_time_distributed</a>()</code> <br><code><a href="layer_torch_module_wrapper.html">layer_torch_module_wrapper</a>()</code> <br><code><a href="layer_unit_normalization.html">layer_unit_normalization</a>()</code> <br><code><a href="layer_upsampling_1d.html">layer_upsampling_1d</a>()</code> <br><code><a href="layer_upsampling_2d.html">layer_upsampling_2d</a>()</code> <br><code><a href="layer_upsampling_3d.html">layer_upsampling_3d</a>()</code> <br><code><a href="layer_zero_padding_1d.html">layer_zero_padding_1d</a>()</code> <br><code><a href="layer_zero_padding_2d.html">layer_zero_padding_2d</a>()</code> <br><code><a href="layer_zero_padding_3d.html">layer_zero_padding_3d</a>()</code> <br><code><a href="rnn_cell_gru.html">rnn_cell_gru</a>()</code> <br><code><a href="rnn_cell_lstm.html">rnn_cell_lstm</a>()</code> <br><code><a href="rnn_cell_simple.html">rnn_cell_simple</a>()</code> <br><code><a href="rnn_cells_stack.html">rnn_cells_stack</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -198,7 +198,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/layer_aug_mix.html b/docs/dev/reference/layer_aug_mix.html
index d066dcc76..9724badb2 100644
--- a/docs/dev/reference/layer_aug_mix.html
+++ b/docs/dev/reference/layer_aug_mix.html
@@ -16,7 +16,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -161,9 +161,9 @@ <h2 id="references">References<a class="anchor" aria-label="anchor" href="#refer
 </ul></div>
     <div class="section level2">
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
-    <div class="dont-index"><p>Other image preprocessing layers: <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br></p>
-<p>Other preprocessing layers: <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_category_encoding.html">layer_category_encoding</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_discretization.html">layer_discretization</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_feature_space.html">layer_feature_space</a>()</code> <br><code><a href="layer_hashed_crossing.html">layer_hashed_crossing</a>()</code> <br><code><a href="layer_hashing.html">layer_hashing</a>()</code> <br><code><a href="layer_integer_lookup.html">layer_integer_lookup</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_mel_spectrogram.html">layer_mel_spectrogram</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_normalization.html">layer_normalization</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_brightness.html">layer_random_brightness</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_contrast.html">layer_random_contrast</a>()</code> <br><code><a href="layer_random_crop.html">layer_random_crop</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_flip.html">layer_random_flip</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_rotation.html">layer_random_rotation</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_random_translation.html">layer_random_translation</a>()</code> <br><code><a href="layer_random_zoom.html">layer_random_zoom</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br><code><a href="layer_stft_spectrogram.html">layer_stft_spectrogram</a>()</code> <br><code><a href="layer_string_lookup.html">layer_string_lookup</a>()</code> <br><code><a href="layer_text_vectorization.html">layer_text_vectorization</a>()</code> <br></p>
-<p>Other layers: <br><code><a href="Layer.html">Layer</a>()</code> <br><code><a href="layer_activation.html">layer_activation</a>()</code> <br><code><a href="layer_activation_elu.html">layer_activation_elu</a>()</code> <br><code><a href="layer_activation_leaky_relu.html">layer_activation_leaky_relu</a>()</code> <br><code><a href="layer_activation_parametric_relu.html">layer_activation_parametric_relu</a>()</code> <br><code><a href="layer_activation_relu.html">layer_activation_relu</a>()</code> <br><code><a href="layer_activation_softmax.html">layer_activation_softmax</a>()</code> <br><code><a href="layer_activity_regularization.html">layer_activity_regularization</a>()</code> <br><code><a href="layer_add.html">layer_add</a>()</code> <br><code><a href="layer_additive_attention.html">layer_additive_attention</a>()</code> <br><code><a href="layer_alpha_dropout.html">layer_alpha_dropout</a>()</code> <br><code><a href="layer_attention.html">layer_attention</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_average.html">layer_average</a>()</code> <br><code><a href="layer_average_pooling_1d.html">layer_average_pooling_1d</a>()</code> <br><code><a href="layer_average_pooling_2d.html">layer_average_pooling_2d</a>()</code> <br><code><a href="layer_average_pooling_3d.html">layer_average_pooling_3d</a>()</code> <br><code><a href="layer_batch_normalization.html">layer_batch_normalization</a>()</code> <br><code><a href="layer_bidirectional.html">layer_bidirectional</a>()</code> <br><code><a href="layer_category_encoding.html">layer_category_encoding</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_concatenate.html">layer_concatenate</a>()</code> <br><code><a href="layer_conv_1d.html">layer_conv_1d</a>()</code> <br><code><a href="layer_conv_1d_transpose.html">layer_conv_1d_transpose</a>()</code> <br><code><a href="layer_conv_2d.html">layer_conv_2d</a>()</code> <br><code><a href="layer_conv_2d_transpose.html">layer_conv_2d_transpose</a>()</code> <br><code><a href="layer_conv_3d.html">layer_conv_3d</a>()</code> <br><code><a href="layer_conv_3d_transpose.html">layer_conv_3d_transpose</a>()</code> <br><code><a href="layer_conv_lstm_1d.html">layer_conv_lstm_1d</a>()</code> <br><code><a href="layer_conv_lstm_2d.html">layer_conv_lstm_2d</a>()</code> <br><code><a href="layer_conv_lstm_3d.html">layer_conv_lstm_3d</a>()</code> <br><code><a href="layer_cropping_1d.html">layer_cropping_1d</a>()</code> <br><code><a href="layer_cropping_2d.html">layer_cropping_2d</a>()</code> <br><code><a href="layer_cropping_3d.html">layer_cropping_3d</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_dense.html">layer_dense</a>()</code> <br><code><a href="layer_depthwise_conv_1d.html">layer_depthwise_conv_1d</a>()</code> <br><code><a href="layer_depthwise_conv_2d.html">layer_depthwise_conv_2d</a>()</code> <br><code><a href="layer_discretization.html">layer_discretization</a>()</code> <br><code><a href="layer_dot.html">layer_dot</a>()</code> <br><code><a href="layer_dropout.html">layer_dropout</a>()</code> <br><code><a href="layer_einsum_dense.html">layer_einsum_dense</a>()</code> <br><code><a href="layer_embedding.html">layer_embedding</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_feature_space.html">layer_feature_space</a>()</code> <br><code><a href="layer_flatten.html">layer_flatten</a>()</code> <br><code><a href="layer_flax_module_wrapper.html">layer_flax_module_wrapper</a>()</code> <br><code><a href="layer_gaussian_dropout.html">layer_gaussian_dropout</a>()</code> <br><code><a href="layer_gaussian_noise.html">layer_gaussian_noise</a>()</code> <br><code><a href="layer_global_average_pooling_1d.html">layer_global_average_pooling_1d</a>()</code> <br><code><a href="layer_global_average_pooling_2d.html">layer_global_average_pooling_2d</a>()</code> <br><code><a href="layer_global_average_pooling_3d.html">layer_global_average_pooling_3d</a>()</code> <br><code><a href="layer_global_max_pooling_1d.html">layer_global_max_pooling_1d</a>()</code> <br><code><a href="layer_global_max_pooling_2d.html">layer_global_max_pooling_2d</a>()</code> <br><code><a href="layer_global_max_pooling_3d.html">layer_global_max_pooling_3d</a>()</code> <br><code><a href="layer_group_normalization.html">layer_group_normalization</a>()</code> <br><code><a href="layer_group_query_attention.html">layer_group_query_attention</a>()</code> <br><code><a href="layer_gru.html">layer_gru</a>()</code> <br><code><a href="layer_hashed_crossing.html">layer_hashed_crossing</a>()</code> <br><code><a href="layer_hashing.html">layer_hashing</a>()</code> <br><code><a href="layer_identity.html">layer_identity</a>()</code> <br><code><a href="layer_integer_lookup.html">layer_integer_lookup</a>()</code> <br><code><a href="layer_jax_model_wrapper.html">layer_jax_model_wrapper</a>()</code> <br><code><a href="layer_lambda.html">layer_lambda</a>()</code> <br><code><a href="layer_layer_normalization.html">layer_layer_normalization</a>()</code> <br><code><a href="layer_lstm.html">layer_lstm</a>()</code> <br><code><a href="layer_masking.html">layer_masking</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_max_pooling_1d.html">layer_max_pooling_1d</a>()</code> <br><code><a href="layer_max_pooling_2d.html">layer_max_pooling_2d</a>()</code> <br><code><a href="layer_max_pooling_3d.html">layer_max_pooling_3d</a>()</code> <br><code><a href="layer_maximum.html">layer_maximum</a>()</code> <br><code><a href="layer_mel_spectrogram.html">layer_mel_spectrogram</a>()</code> <br><code><a href="layer_minimum.html">layer_minimum</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_multi_head_attention.html">layer_multi_head_attention</a>()</code> <br><code><a href="layer_multiply.html">layer_multiply</a>()</code> <br><code><a href="layer_normalization.html">layer_normalization</a>()</code> <br><code><a href="layer_permute.html">layer_permute</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_brightness.html">layer_random_brightness</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_contrast.html">layer_random_contrast</a>()</code> <br><code><a href="layer_random_crop.html">layer_random_crop</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_flip.html">layer_random_flip</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_rotation.html">layer_random_rotation</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_random_translation.html">layer_random_translation</a>()</code> <br><code><a href="layer_random_zoom.html">layer_random_zoom</a>()</code> <br><code><a href="layer_repeat_vector.html">layer_repeat_vector</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_reshape.html">layer_reshape</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_rms_normalization.html">layer_rms_normalization</a>()</code> <br><code><a href="layer_rnn.html">layer_rnn</a>()</code> <br><code><a href="layer_separable_conv_1d.html">layer_separable_conv_1d</a>()</code> <br><code><a href="layer_separable_conv_2d.html">layer_separable_conv_2d</a>()</code> <br><code><a href="layer_simple_rnn.html">layer_simple_rnn</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br><code><a href="layer_spatial_dropout_1d.html">layer_spatial_dropout_1d</a>()</code> <br><code><a href="layer_spatial_dropout_2d.html">layer_spatial_dropout_2d</a>()</code> <br><code><a href="layer_spatial_dropout_3d.html">layer_spatial_dropout_3d</a>()</code> <br><code><a href="layer_spectral_normalization.html">layer_spectral_normalization</a>()</code> <br><code><a href="layer_stft_spectrogram.html">layer_stft_spectrogram</a>()</code> <br><code><a href="layer_string_lookup.html">layer_string_lookup</a>()</code> <br><code><a href="layer_subtract.html">layer_subtract</a>()</code> <br><code><a href="layer_text_vectorization.html">layer_text_vectorization</a>()</code> <br><code><a href="layer_tfsm.html">layer_tfsm</a>()</code> <br><code><a href="layer_time_distributed.html">layer_time_distributed</a>()</code> <br><code><a href="layer_torch_module_wrapper.html">layer_torch_module_wrapper</a>()</code> <br><code><a href="layer_unit_normalization.html">layer_unit_normalization</a>()</code> <br><code><a href="layer_upsampling_1d.html">layer_upsampling_1d</a>()</code> <br><code><a href="layer_upsampling_2d.html">layer_upsampling_2d</a>()</code> <br><code><a href="layer_upsampling_3d.html">layer_upsampling_3d</a>()</code> <br><code><a href="layer_zero_padding_1d.html">layer_zero_padding_1d</a>()</code> <br><code><a href="layer_zero_padding_2d.html">layer_zero_padding_2d</a>()</code> <br><code><a href="layer_zero_padding_3d.html">layer_zero_padding_3d</a>()</code> <br><code><a href="rnn_cell_gru.html">rnn_cell_gru</a>()</code> <br><code><a href="rnn_cell_lstm.html">rnn_cell_lstm</a>()</code> <br><code><a href="rnn_cell_simple.html">rnn_cell_simple</a>()</code> <br><code><a href="rnn_cells_stack.html">rnn_cells_stack</a>()</code> <br></p></div>
+    <div class="dont-index"><p>Other image preprocessing layers: <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_elastic_transform.html">layer_random_elastic_transform</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br></p>
+<p>Other preprocessing layers: <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_category_encoding.html">layer_category_encoding</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_discretization.html">layer_discretization</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_feature_space.html">layer_feature_space</a>()</code> <br><code><a href="layer_hashed_crossing.html">layer_hashed_crossing</a>()</code> <br><code><a href="layer_hashing.html">layer_hashing</a>()</code> <br><code><a href="layer_integer_lookup.html">layer_integer_lookup</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_mel_spectrogram.html">layer_mel_spectrogram</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_normalization.html">layer_normalization</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_brightness.html">layer_random_brightness</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_contrast.html">layer_random_contrast</a>()</code> <br><code><a href="layer_random_crop.html">layer_random_crop</a>()</code> <br><code><a href="layer_random_elastic_transform.html">layer_random_elastic_transform</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_flip.html">layer_random_flip</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_rotation.html">layer_random_rotation</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_random_translation.html">layer_random_translation</a>()</code> <br><code><a href="layer_random_zoom.html">layer_random_zoom</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br><code><a href="layer_stft_spectrogram.html">layer_stft_spectrogram</a>()</code> <br><code><a href="layer_string_lookup.html">layer_string_lookup</a>()</code> <br><code><a href="layer_text_vectorization.html">layer_text_vectorization</a>()</code> <br></p>
+<p>Other layers: <br><code><a href="Layer.html">Layer</a>()</code> <br><code><a href="layer_activation.html">layer_activation</a>()</code> <br><code><a href="layer_activation_elu.html">layer_activation_elu</a>()</code> <br><code><a href="layer_activation_leaky_relu.html">layer_activation_leaky_relu</a>()</code> <br><code><a href="layer_activation_parametric_relu.html">layer_activation_parametric_relu</a>()</code> <br><code><a href="layer_activation_relu.html">layer_activation_relu</a>()</code> <br><code><a href="layer_activation_softmax.html">layer_activation_softmax</a>()</code> <br><code><a href="layer_activity_regularization.html">layer_activity_regularization</a>()</code> <br><code><a href="layer_add.html">layer_add</a>()</code> <br><code><a href="layer_additive_attention.html">layer_additive_attention</a>()</code> <br><code><a href="layer_alpha_dropout.html">layer_alpha_dropout</a>()</code> <br><code><a href="layer_attention.html">layer_attention</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_average.html">layer_average</a>()</code> <br><code><a href="layer_average_pooling_1d.html">layer_average_pooling_1d</a>()</code> <br><code><a href="layer_average_pooling_2d.html">layer_average_pooling_2d</a>()</code> <br><code><a href="layer_average_pooling_3d.html">layer_average_pooling_3d</a>()</code> <br><code><a href="layer_batch_normalization.html">layer_batch_normalization</a>()</code> <br><code><a href="layer_bidirectional.html">layer_bidirectional</a>()</code> <br><code><a href="layer_category_encoding.html">layer_category_encoding</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_concatenate.html">layer_concatenate</a>()</code> <br><code><a href="layer_conv_1d.html">layer_conv_1d</a>()</code> <br><code><a href="layer_conv_1d_transpose.html">layer_conv_1d_transpose</a>()</code> <br><code><a href="layer_conv_2d.html">layer_conv_2d</a>()</code> <br><code><a href="layer_conv_2d_transpose.html">layer_conv_2d_transpose</a>()</code> <br><code><a href="layer_conv_3d.html">layer_conv_3d</a>()</code> <br><code><a href="layer_conv_3d_transpose.html">layer_conv_3d_transpose</a>()</code> <br><code><a href="layer_conv_lstm_1d.html">layer_conv_lstm_1d</a>()</code> <br><code><a href="layer_conv_lstm_2d.html">layer_conv_lstm_2d</a>()</code> <br><code><a href="layer_conv_lstm_3d.html">layer_conv_lstm_3d</a>()</code> <br><code><a href="layer_cropping_1d.html">layer_cropping_1d</a>()</code> <br><code><a href="layer_cropping_2d.html">layer_cropping_2d</a>()</code> <br><code><a href="layer_cropping_3d.html">layer_cropping_3d</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_dense.html">layer_dense</a>()</code> <br><code><a href="layer_depthwise_conv_1d.html">layer_depthwise_conv_1d</a>()</code> <br><code><a href="layer_depthwise_conv_2d.html">layer_depthwise_conv_2d</a>()</code> <br><code><a href="layer_discretization.html">layer_discretization</a>()</code> <br><code><a href="layer_dot.html">layer_dot</a>()</code> <br><code><a href="layer_dropout.html">layer_dropout</a>()</code> <br><code><a href="layer_einsum_dense.html">layer_einsum_dense</a>()</code> <br><code><a href="layer_embedding.html">layer_embedding</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_feature_space.html">layer_feature_space</a>()</code> <br><code><a href="layer_flatten.html">layer_flatten</a>()</code> <br><code><a href="layer_flax_module_wrapper.html">layer_flax_module_wrapper</a>()</code> <br><code><a href="layer_gaussian_dropout.html">layer_gaussian_dropout</a>()</code> <br><code><a href="layer_gaussian_noise.html">layer_gaussian_noise</a>()</code> <br><code><a href="layer_global_average_pooling_1d.html">layer_global_average_pooling_1d</a>()</code> <br><code><a href="layer_global_average_pooling_2d.html">layer_global_average_pooling_2d</a>()</code> <br><code><a href="layer_global_average_pooling_3d.html">layer_global_average_pooling_3d</a>()</code> <br><code><a href="layer_global_max_pooling_1d.html">layer_global_max_pooling_1d</a>()</code> <br><code><a href="layer_global_max_pooling_2d.html">layer_global_max_pooling_2d</a>()</code> <br><code><a href="layer_global_max_pooling_3d.html">layer_global_max_pooling_3d</a>()</code> <br><code><a href="layer_group_normalization.html">layer_group_normalization</a>()</code> <br><code><a href="layer_group_query_attention.html">layer_group_query_attention</a>()</code> <br><code><a href="layer_gru.html">layer_gru</a>()</code> <br><code><a href="layer_hashed_crossing.html">layer_hashed_crossing</a>()</code> <br><code><a href="layer_hashing.html">layer_hashing</a>()</code> <br><code><a href="layer_identity.html">layer_identity</a>()</code> <br><code><a href="layer_integer_lookup.html">layer_integer_lookup</a>()</code> <br><code><a href="layer_jax_model_wrapper.html">layer_jax_model_wrapper</a>()</code> <br><code><a href="layer_lambda.html">layer_lambda</a>()</code> <br><code><a href="layer_layer_normalization.html">layer_layer_normalization</a>()</code> <br><code><a href="layer_lstm.html">layer_lstm</a>()</code> <br><code><a href="layer_masking.html">layer_masking</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_max_pooling_1d.html">layer_max_pooling_1d</a>()</code> <br><code><a href="layer_max_pooling_2d.html">layer_max_pooling_2d</a>()</code> <br><code><a href="layer_max_pooling_3d.html">layer_max_pooling_3d</a>()</code> <br><code><a href="layer_maximum.html">layer_maximum</a>()</code> <br><code><a href="layer_mel_spectrogram.html">layer_mel_spectrogram</a>()</code> <br><code><a href="layer_minimum.html">layer_minimum</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_multi_head_attention.html">layer_multi_head_attention</a>()</code> <br><code><a href="layer_multiply.html">layer_multiply</a>()</code> <br><code><a href="layer_normalization.html">layer_normalization</a>()</code> <br><code><a href="layer_permute.html">layer_permute</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_brightness.html">layer_random_brightness</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_contrast.html">layer_random_contrast</a>()</code> <br><code><a href="layer_random_crop.html">layer_random_crop</a>()</code> <br><code><a href="layer_random_elastic_transform.html">layer_random_elastic_transform</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_flip.html">layer_random_flip</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_rotation.html">layer_random_rotation</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_random_translation.html">layer_random_translation</a>()</code> <br><code><a href="layer_random_zoom.html">layer_random_zoom</a>()</code> <br><code><a href="layer_repeat_vector.html">layer_repeat_vector</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_reshape.html">layer_reshape</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_rms_normalization.html">layer_rms_normalization</a>()</code> <br><code><a href="layer_rnn.html">layer_rnn</a>()</code> <br><code><a href="layer_separable_conv_1d.html">layer_separable_conv_1d</a>()</code> <br><code><a href="layer_separable_conv_2d.html">layer_separable_conv_2d</a>()</code> <br><code><a href="layer_simple_rnn.html">layer_simple_rnn</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br><code><a href="layer_spatial_dropout_1d.html">layer_spatial_dropout_1d</a>()</code> <br><code><a href="layer_spatial_dropout_2d.html">layer_spatial_dropout_2d</a>()</code> <br><code><a href="layer_spatial_dropout_3d.html">layer_spatial_dropout_3d</a>()</code> <br><code><a href="layer_spectral_normalization.html">layer_spectral_normalization</a>()</code> <br><code><a href="layer_stft_spectrogram.html">layer_stft_spectrogram</a>()</code> <br><code><a href="layer_string_lookup.html">layer_string_lookup</a>()</code> <br><code><a href="layer_subtract.html">layer_subtract</a>()</code> <br><code><a href="layer_text_vectorization.html">layer_text_vectorization</a>()</code> <br><code><a href="layer_tfsm.html">layer_tfsm</a>()</code> <br><code><a href="layer_time_distributed.html">layer_time_distributed</a>()</code> <br><code><a href="layer_torch_module_wrapper.html">layer_torch_module_wrapper</a>()</code> <br><code><a href="layer_unit_normalization.html">layer_unit_normalization</a>()</code> <br><code><a href="layer_upsampling_1d.html">layer_upsampling_1d</a>()</code> <br><code><a href="layer_upsampling_2d.html">layer_upsampling_2d</a>()</code> <br><code><a href="layer_upsampling_3d.html">layer_upsampling_3d</a>()</code> <br><code><a href="layer_zero_padding_1d.html">layer_zero_padding_1d</a>()</code> <br><code><a href="layer_zero_padding_2d.html">layer_zero_padding_2d</a>()</code> <br><code><a href="layer_zero_padding_3d.html">layer_zero_padding_3d</a>()</code> <br><code><a href="rnn_cell_gru.html">rnn_cell_gru</a>()</code> <br><code><a href="rnn_cell_lstm.html">rnn_cell_lstm</a>()</code> <br><code><a href="rnn_cell_simple.html">rnn_cell_simple</a>()</code> <br><code><a href="rnn_cells_stack.html">rnn_cells_stack</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -175,7 +175,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/layer_auto_contrast.html b/docs/dev/reference/layer_auto_contrast.html
index 4562e7d4c..fc8a72fdb 100644
--- a/docs/dev/reference/layer_auto_contrast.html
+++ b/docs/dev/reference/layer_auto_contrast.html
@@ -16,7 +16,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -96,9 +96,9 @@ <h2 id="arguments">Arguments<a class="anchor" aria-label="anchor" href="#argumen
 </dl></div>
     <div class="section level2">
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
-    <div class="dont-index"><p>Other image preprocessing layers: <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br></p>
-<p>Other preprocessing layers: <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_category_encoding.html">layer_category_encoding</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_discretization.html">layer_discretization</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_feature_space.html">layer_feature_space</a>()</code> <br><code><a href="layer_hashed_crossing.html">layer_hashed_crossing</a>()</code> <br><code><a href="layer_hashing.html">layer_hashing</a>()</code> <br><code><a href="layer_integer_lookup.html">layer_integer_lookup</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_mel_spectrogram.html">layer_mel_spectrogram</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_normalization.html">layer_normalization</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_brightness.html">layer_random_brightness</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_contrast.html">layer_random_contrast</a>()</code> <br><code><a href="layer_random_crop.html">layer_random_crop</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_flip.html">layer_random_flip</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_rotation.html">layer_random_rotation</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_random_translation.html">layer_random_translation</a>()</code> <br><code><a href="layer_random_zoom.html">layer_random_zoom</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br><code><a href="layer_stft_spectrogram.html">layer_stft_spectrogram</a>()</code> <br><code><a href="layer_string_lookup.html">layer_string_lookup</a>()</code> <br><code><a href="layer_text_vectorization.html">layer_text_vectorization</a>()</code> <br></p>
-<p>Other layers: <br><code><a href="Layer.html">Layer</a>()</code> <br><code><a href="layer_activation.html">layer_activation</a>()</code> <br><code><a href="layer_activation_elu.html">layer_activation_elu</a>()</code> <br><code><a href="layer_activation_leaky_relu.html">layer_activation_leaky_relu</a>()</code> <br><code><a href="layer_activation_parametric_relu.html">layer_activation_parametric_relu</a>()</code> <br><code><a href="layer_activation_relu.html">layer_activation_relu</a>()</code> <br><code><a href="layer_activation_softmax.html">layer_activation_softmax</a>()</code> <br><code><a href="layer_activity_regularization.html">layer_activity_regularization</a>()</code> <br><code><a href="layer_add.html">layer_add</a>()</code> <br><code><a href="layer_additive_attention.html">layer_additive_attention</a>()</code> <br><code><a href="layer_alpha_dropout.html">layer_alpha_dropout</a>()</code> <br><code><a href="layer_attention.html">layer_attention</a>()</code> <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_average.html">layer_average</a>()</code> <br><code><a href="layer_average_pooling_1d.html">layer_average_pooling_1d</a>()</code> <br><code><a href="layer_average_pooling_2d.html">layer_average_pooling_2d</a>()</code> <br><code><a href="layer_average_pooling_3d.html">layer_average_pooling_3d</a>()</code> <br><code><a href="layer_batch_normalization.html">layer_batch_normalization</a>()</code> <br><code><a href="layer_bidirectional.html">layer_bidirectional</a>()</code> <br><code><a href="layer_category_encoding.html">layer_category_encoding</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_concatenate.html">layer_concatenate</a>()</code> <br><code><a href="layer_conv_1d.html">layer_conv_1d</a>()</code> <br><code><a href="layer_conv_1d_transpose.html">layer_conv_1d_transpose</a>()</code> <br><code><a href="layer_conv_2d.html">layer_conv_2d</a>()</code> <br><code><a href="layer_conv_2d_transpose.html">layer_conv_2d_transpose</a>()</code> <br><code><a href="layer_conv_3d.html">layer_conv_3d</a>()</code> <br><code><a href="layer_conv_3d_transpose.html">layer_conv_3d_transpose</a>()</code> <br><code><a href="layer_conv_lstm_1d.html">layer_conv_lstm_1d</a>()</code> <br><code><a href="layer_conv_lstm_2d.html">layer_conv_lstm_2d</a>()</code> <br><code><a href="layer_conv_lstm_3d.html">layer_conv_lstm_3d</a>()</code> <br><code><a href="layer_cropping_1d.html">layer_cropping_1d</a>()</code> <br><code><a href="layer_cropping_2d.html">layer_cropping_2d</a>()</code> <br><code><a href="layer_cropping_3d.html">layer_cropping_3d</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_dense.html">layer_dense</a>()</code> <br><code><a href="layer_depthwise_conv_1d.html">layer_depthwise_conv_1d</a>()</code> <br><code><a href="layer_depthwise_conv_2d.html">layer_depthwise_conv_2d</a>()</code> <br><code><a href="layer_discretization.html">layer_discretization</a>()</code> <br><code><a href="layer_dot.html">layer_dot</a>()</code> <br><code><a href="layer_dropout.html">layer_dropout</a>()</code> <br><code><a href="layer_einsum_dense.html">layer_einsum_dense</a>()</code> <br><code><a href="layer_embedding.html">layer_embedding</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_feature_space.html">layer_feature_space</a>()</code> <br><code><a href="layer_flatten.html">layer_flatten</a>()</code> <br><code><a href="layer_flax_module_wrapper.html">layer_flax_module_wrapper</a>()</code> <br><code><a href="layer_gaussian_dropout.html">layer_gaussian_dropout</a>()</code> <br><code><a href="layer_gaussian_noise.html">layer_gaussian_noise</a>()</code> <br><code><a href="layer_global_average_pooling_1d.html">layer_global_average_pooling_1d</a>()</code> <br><code><a href="layer_global_average_pooling_2d.html">layer_global_average_pooling_2d</a>()</code> <br><code><a href="layer_global_average_pooling_3d.html">layer_global_average_pooling_3d</a>()</code> <br><code><a href="layer_global_max_pooling_1d.html">layer_global_max_pooling_1d</a>()</code> <br><code><a href="layer_global_max_pooling_2d.html">layer_global_max_pooling_2d</a>()</code> <br><code><a href="layer_global_max_pooling_3d.html">layer_global_max_pooling_3d</a>()</code> <br><code><a href="layer_group_normalization.html">layer_group_normalization</a>()</code> <br><code><a href="layer_group_query_attention.html">layer_group_query_attention</a>()</code> <br><code><a href="layer_gru.html">layer_gru</a>()</code> <br><code><a href="layer_hashed_crossing.html">layer_hashed_crossing</a>()</code> <br><code><a href="layer_hashing.html">layer_hashing</a>()</code> <br><code><a href="layer_identity.html">layer_identity</a>()</code> <br><code><a href="layer_integer_lookup.html">layer_integer_lookup</a>()</code> <br><code><a href="layer_jax_model_wrapper.html">layer_jax_model_wrapper</a>()</code> <br><code><a href="layer_lambda.html">layer_lambda</a>()</code> <br><code><a href="layer_layer_normalization.html">layer_layer_normalization</a>()</code> <br><code><a href="layer_lstm.html">layer_lstm</a>()</code> <br><code><a href="layer_masking.html">layer_masking</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_max_pooling_1d.html">layer_max_pooling_1d</a>()</code> <br><code><a href="layer_max_pooling_2d.html">layer_max_pooling_2d</a>()</code> <br><code><a href="layer_max_pooling_3d.html">layer_max_pooling_3d</a>()</code> <br><code><a href="layer_maximum.html">layer_maximum</a>()</code> <br><code><a href="layer_mel_spectrogram.html">layer_mel_spectrogram</a>()</code> <br><code><a href="layer_minimum.html">layer_minimum</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_multi_head_attention.html">layer_multi_head_attention</a>()</code> <br><code><a href="layer_multiply.html">layer_multiply</a>()</code> <br><code><a href="layer_normalization.html">layer_normalization</a>()</code> <br><code><a href="layer_permute.html">layer_permute</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_brightness.html">layer_random_brightness</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_contrast.html">layer_random_contrast</a>()</code> <br><code><a href="layer_random_crop.html">layer_random_crop</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_flip.html">layer_random_flip</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_rotation.html">layer_random_rotation</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_random_translation.html">layer_random_translation</a>()</code> <br><code><a href="layer_random_zoom.html">layer_random_zoom</a>()</code> <br><code><a href="layer_repeat_vector.html">layer_repeat_vector</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_reshape.html">layer_reshape</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_rms_normalization.html">layer_rms_normalization</a>()</code> <br><code><a href="layer_rnn.html">layer_rnn</a>()</code> <br><code><a href="layer_separable_conv_1d.html">layer_separable_conv_1d</a>()</code> <br><code><a href="layer_separable_conv_2d.html">layer_separable_conv_2d</a>()</code> <br><code><a href="layer_simple_rnn.html">layer_simple_rnn</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br><code><a href="layer_spatial_dropout_1d.html">layer_spatial_dropout_1d</a>()</code> <br><code><a href="layer_spatial_dropout_2d.html">layer_spatial_dropout_2d</a>()</code> <br><code><a href="layer_spatial_dropout_3d.html">layer_spatial_dropout_3d</a>()</code> <br><code><a href="layer_spectral_normalization.html">layer_spectral_normalization</a>()</code> <br><code><a href="layer_stft_spectrogram.html">layer_stft_spectrogram</a>()</code> <br><code><a href="layer_string_lookup.html">layer_string_lookup</a>()</code> <br><code><a href="layer_subtract.html">layer_subtract</a>()</code> <br><code><a href="layer_text_vectorization.html">layer_text_vectorization</a>()</code> <br><code><a href="layer_tfsm.html">layer_tfsm</a>()</code> <br><code><a href="layer_time_distributed.html">layer_time_distributed</a>()</code> <br><code><a href="layer_torch_module_wrapper.html">layer_torch_module_wrapper</a>()</code> <br><code><a href="layer_unit_normalization.html">layer_unit_normalization</a>()</code> <br><code><a href="layer_upsampling_1d.html">layer_upsampling_1d</a>()</code> <br><code><a href="layer_upsampling_2d.html">layer_upsampling_2d</a>()</code> <br><code><a href="layer_upsampling_3d.html">layer_upsampling_3d</a>()</code> <br><code><a href="layer_zero_padding_1d.html">layer_zero_padding_1d</a>()</code> <br><code><a href="layer_zero_padding_2d.html">layer_zero_padding_2d</a>()</code> <br><code><a href="layer_zero_padding_3d.html">layer_zero_padding_3d</a>()</code> <br><code><a href="rnn_cell_gru.html">rnn_cell_gru</a>()</code> <br><code><a href="rnn_cell_lstm.html">rnn_cell_lstm</a>()</code> <br><code><a href="rnn_cell_simple.html">rnn_cell_simple</a>()</code> <br><code><a href="rnn_cells_stack.html">rnn_cells_stack</a>()</code> <br></p></div>
+    <div class="dont-index"><p>Other image preprocessing layers: <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_elastic_transform.html">layer_random_elastic_transform</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br></p>
+<p>Other preprocessing layers: <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_category_encoding.html">layer_category_encoding</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_discretization.html">layer_discretization</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_feature_space.html">layer_feature_space</a>()</code> <br><code><a href="layer_hashed_crossing.html">layer_hashed_crossing</a>()</code> <br><code><a href="layer_hashing.html">layer_hashing</a>()</code> <br><code><a href="layer_integer_lookup.html">layer_integer_lookup</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_mel_spectrogram.html">layer_mel_spectrogram</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_normalization.html">layer_normalization</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_brightness.html">layer_random_brightness</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_contrast.html">layer_random_contrast</a>()</code> <br><code><a href="layer_random_crop.html">layer_random_crop</a>()</code> <br><code><a href="layer_random_elastic_transform.html">layer_random_elastic_transform</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_flip.html">layer_random_flip</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_rotation.html">layer_random_rotation</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_random_translation.html">layer_random_translation</a>()</code> <br><code><a href="layer_random_zoom.html">layer_random_zoom</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br><code><a href="layer_stft_spectrogram.html">layer_stft_spectrogram</a>()</code> <br><code><a href="layer_string_lookup.html">layer_string_lookup</a>()</code> <br><code><a href="layer_text_vectorization.html">layer_text_vectorization</a>()</code> <br></p>
+<p>Other layers: <br><code><a href="Layer.html">Layer</a>()</code> <br><code><a href="layer_activation.html">layer_activation</a>()</code> <br><code><a href="layer_activation_elu.html">layer_activation_elu</a>()</code> <br><code><a href="layer_activation_leaky_relu.html">layer_activation_leaky_relu</a>()</code> <br><code><a href="layer_activation_parametric_relu.html">layer_activation_parametric_relu</a>()</code> <br><code><a href="layer_activation_relu.html">layer_activation_relu</a>()</code> <br><code><a href="layer_activation_softmax.html">layer_activation_softmax</a>()</code> <br><code><a href="layer_activity_regularization.html">layer_activity_regularization</a>()</code> <br><code><a href="layer_add.html">layer_add</a>()</code> <br><code><a href="layer_additive_attention.html">layer_additive_attention</a>()</code> <br><code><a href="layer_alpha_dropout.html">layer_alpha_dropout</a>()</code> <br><code><a href="layer_attention.html">layer_attention</a>()</code> <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_average.html">layer_average</a>()</code> <br><code><a href="layer_average_pooling_1d.html">layer_average_pooling_1d</a>()</code> <br><code><a href="layer_average_pooling_2d.html">layer_average_pooling_2d</a>()</code> <br><code><a href="layer_average_pooling_3d.html">layer_average_pooling_3d</a>()</code> <br><code><a href="layer_batch_normalization.html">layer_batch_normalization</a>()</code> <br><code><a href="layer_bidirectional.html">layer_bidirectional</a>()</code> <br><code><a href="layer_category_encoding.html">layer_category_encoding</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_concatenate.html">layer_concatenate</a>()</code> <br><code><a href="layer_conv_1d.html">layer_conv_1d</a>()</code> <br><code><a href="layer_conv_1d_transpose.html">layer_conv_1d_transpose</a>()</code> <br><code><a href="layer_conv_2d.html">layer_conv_2d</a>()</code> <br><code><a href="layer_conv_2d_transpose.html">layer_conv_2d_transpose</a>()</code> <br><code><a href="layer_conv_3d.html">layer_conv_3d</a>()</code> <br><code><a href="layer_conv_3d_transpose.html">layer_conv_3d_transpose</a>()</code> <br><code><a href="layer_conv_lstm_1d.html">layer_conv_lstm_1d</a>()</code> <br><code><a href="layer_conv_lstm_2d.html">layer_conv_lstm_2d</a>()</code> <br><code><a href="layer_conv_lstm_3d.html">layer_conv_lstm_3d</a>()</code> <br><code><a href="layer_cropping_1d.html">layer_cropping_1d</a>()</code> <br><code><a href="layer_cropping_2d.html">layer_cropping_2d</a>()</code> <br><code><a href="layer_cropping_3d.html">layer_cropping_3d</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_dense.html">layer_dense</a>()</code> <br><code><a href="layer_depthwise_conv_1d.html">layer_depthwise_conv_1d</a>()</code> <br><code><a href="layer_depthwise_conv_2d.html">layer_depthwise_conv_2d</a>()</code> <br><code><a href="layer_discretization.html">layer_discretization</a>()</code> <br><code><a href="layer_dot.html">layer_dot</a>()</code> <br><code><a href="layer_dropout.html">layer_dropout</a>()</code> <br><code><a href="layer_einsum_dense.html">layer_einsum_dense</a>()</code> <br><code><a href="layer_embedding.html">layer_embedding</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_feature_space.html">layer_feature_space</a>()</code> <br><code><a href="layer_flatten.html">layer_flatten</a>()</code> <br><code><a href="layer_flax_module_wrapper.html">layer_flax_module_wrapper</a>()</code> <br><code><a href="layer_gaussian_dropout.html">layer_gaussian_dropout</a>()</code> <br><code><a href="layer_gaussian_noise.html">layer_gaussian_noise</a>()</code> <br><code><a href="layer_global_average_pooling_1d.html">layer_global_average_pooling_1d</a>()</code> <br><code><a href="layer_global_average_pooling_2d.html">layer_global_average_pooling_2d</a>()</code> <br><code><a href="layer_global_average_pooling_3d.html">layer_global_average_pooling_3d</a>()</code> <br><code><a href="layer_global_max_pooling_1d.html">layer_global_max_pooling_1d</a>()</code> <br><code><a href="layer_global_max_pooling_2d.html">layer_global_max_pooling_2d</a>()</code> <br><code><a href="layer_global_max_pooling_3d.html">layer_global_max_pooling_3d</a>()</code> <br><code><a href="layer_group_normalization.html">layer_group_normalization</a>()</code> <br><code><a href="layer_group_query_attention.html">layer_group_query_attention</a>()</code> <br><code><a href="layer_gru.html">layer_gru</a>()</code> <br><code><a href="layer_hashed_crossing.html">layer_hashed_crossing</a>()</code> <br><code><a href="layer_hashing.html">layer_hashing</a>()</code> <br><code><a href="layer_identity.html">layer_identity</a>()</code> <br><code><a href="layer_integer_lookup.html">layer_integer_lookup</a>()</code> <br><code><a href="layer_jax_model_wrapper.html">layer_jax_model_wrapper</a>()</code> <br><code><a href="layer_lambda.html">layer_lambda</a>()</code> <br><code><a href="layer_layer_normalization.html">layer_layer_normalization</a>()</code> <br><code><a href="layer_lstm.html">layer_lstm</a>()</code> <br><code><a href="layer_masking.html">layer_masking</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_max_pooling_1d.html">layer_max_pooling_1d</a>()</code> <br><code><a href="layer_max_pooling_2d.html">layer_max_pooling_2d</a>()</code> <br><code><a href="layer_max_pooling_3d.html">layer_max_pooling_3d</a>()</code> <br><code><a href="layer_maximum.html">layer_maximum</a>()</code> <br><code><a href="layer_mel_spectrogram.html">layer_mel_spectrogram</a>()</code> <br><code><a href="layer_minimum.html">layer_minimum</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_multi_head_attention.html">layer_multi_head_attention</a>()</code> <br><code><a href="layer_multiply.html">layer_multiply</a>()</code> <br><code><a href="layer_normalization.html">layer_normalization</a>()</code> <br><code><a href="layer_permute.html">layer_permute</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_brightness.html">layer_random_brightness</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_contrast.html">layer_random_contrast</a>()</code> <br><code><a href="layer_random_crop.html">layer_random_crop</a>()</code> <br><code><a href="layer_random_elastic_transform.html">layer_random_elastic_transform</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_flip.html">layer_random_flip</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_rotation.html">layer_random_rotation</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_random_translation.html">layer_random_translation</a>()</code> <br><code><a href="layer_random_zoom.html">layer_random_zoom</a>()</code> <br><code><a href="layer_repeat_vector.html">layer_repeat_vector</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_reshape.html">layer_reshape</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_rms_normalization.html">layer_rms_normalization</a>()</code> <br><code><a href="layer_rnn.html">layer_rnn</a>()</code> <br><code><a href="layer_separable_conv_1d.html">layer_separable_conv_1d</a>()</code> <br><code><a href="layer_separable_conv_2d.html">layer_separable_conv_2d</a>()</code> <br><code><a href="layer_simple_rnn.html">layer_simple_rnn</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br><code><a href="layer_spatial_dropout_1d.html">layer_spatial_dropout_1d</a>()</code> <br><code><a href="layer_spatial_dropout_2d.html">layer_spatial_dropout_2d</a>()</code> <br><code><a href="layer_spatial_dropout_3d.html">layer_spatial_dropout_3d</a>()</code> <br><code><a href="layer_spectral_normalization.html">layer_spectral_normalization</a>()</code> <br><code><a href="layer_stft_spectrogram.html">layer_stft_spectrogram</a>()</code> <br><code><a href="layer_string_lookup.html">layer_string_lookup</a>()</code> <br><code><a href="layer_subtract.html">layer_subtract</a>()</code> <br><code><a href="layer_text_vectorization.html">layer_text_vectorization</a>()</code> <br><code><a href="layer_tfsm.html">layer_tfsm</a>()</code> <br><code><a href="layer_time_distributed.html">layer_time_distributed</a>()</code> <br><code><a href="layer_torch_module_wrapper.html">layer_torch_module_wrapper</a>()</code> <br><code><a href="layer_unit_normalization.html">layer_unit_normalization</a>()</code> <br><code><a href="layer_upsampling_1d.html">layer_upsampling_1d</a>()</code> <br><code><a href="layer_upsampling_2d.html">layer_upsampling_2d</a>()</code> <br><code><a href="layer_upsampling_3d.html">layer_upsampling_3d</a>()</code> <br><code><a href="layer_zero_padding_1d.html">layer_zero_padding_1d</a>()</code> <br><code><a href="layer_zero_padding_2d.html">layer_zero_padding_2d</a>()</code> <br><code><a href="layer_zero_padding_3d.html">layer_zero_padding_3d</a>()</code> <br><code><a href="rnn_cell_gru.html">rnn_cell_gru</a>()</code> <br><code><a href="rnn_cell_lstm.html">rnn_cell_lstm</a>()</code> <br><code><a href="rnn_cell_simple.html">rnn_cell_simple</a>()</code> <br><code><a href="rnn_cells_stack.html">rnn_cells_stack</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -110,7 +110,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/layer_average.html b/docs/dev/reference/layer_average.html
index 71e07ab4f..7c89857ed 100644
--- a/docs/dev/reference/layer_average.html
+++ b/docs/dev/reference/layer_average.html
@@ -10,7 +10,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -113,7 +113,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
     <div class="dont-index">
 <ul><li><p><a href="https://keras.io/api/layers/merging_layers/average#average-class" class="external-link">https://keras.io/api/layers/merging_layers/average#average-class</a></p></li>
 </ul><p>Other merging layers: <br><code><a href="layer_add.html">layer_add</a>()</code> <br><code><a href="layer_concatenate.html">layer_concatenate</a>()</code> <br><code><a href="layer_dot.html">layer_dot</a>()</code> <br><code><a href="layer_maximum.html">layer_maximum</a>()</code> <br><code><a href="layer_minimum.html">layer_minimum</a>()</code> <br><code><a href="layer_multiply.html">layer_multiply</a>()</code> <br><code><a href="layer_subtract.html">layer_subtract</a>()</code> <br></p>
-<p>Other layers: <br><code><a href="Layer.html">Layer</a>()</code> <br><code><a href="layer_activation.html">layer_activation</a>()</code> <br><code><a href="layer_activation_elu.html">layer_activation_elu</a>()</code> <br><code><a href="layer_activation_leaky_relu.html">layer_activation_leaky_relu</a>()</code> <br><code><a href="layer_activation_parametric_relu.html">layer_activation_parametric_relu</a>()</code> <br><code><a href="layer_activation_relu.html">layer_activation_relu</a>()</code> <br><code><a href="layer_activation_softmax.html">layer_activation_softmax</a>()</code> <br><code><a href="layer_activity_regularization.html">layer_activity_regularization</a>()</code> <br><code><a href="layer_add.html">layer_add</a>()</code> <br><code><a href="layer_additive_attention.html">layer_additive_attention</a>()</code> <br><code><a href="layer_alpha_dropout.html">layer_alpha_dropout</a>()</code> <br><code><a href="layer_attention.html">layer_attention</a>()</code> <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_average_pooling_1d.html">layer_average_pooling_1d</a>()</code> <br><code><a href="layer_average_pooling_2d.html">layer_average_pooling_2d</a>()</code> <br><code><a href="layer_average_pooling_3d.html">layer_average_pooling_3d</a>()</code> <br><code><a href="layer_batch_normalization.html">layer_batch_normalization</a>()</code> <br><code><a href="layer_bidirectional.html">layer_bidirectional</a>()</code> <br><code><a href="layer_category_encoding.html">layer_category_encoding</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_concatenate.html">layer_concatenate</a>()</code> <br><code><a href="layer_conv_1d.html">layer_conv_1d</a>()</code> <br><code><a href="layer_conv_1d_transpose.html">layer_conv_1d_transpose</a>()</code> <br><code><a href="layer_conv_2d.html">layer_conv_2d</a>()</code> <br><code><a href="layer_conv_2d_transpose.html">layer_conv_2d_transpose</a>()</code> <br><code><a href="layer_conv_3d.html">layer_conv_3d</a>()</code> <br><code><a href="layer_conv_3d_transpose.html">layer_conv_3d_transpose</a>()</code> <br><code><a href="layer_conv_lstm_1d.html">layer_conv_lstm_1d</a>()</code> <br><code><a href="layer_conv_lstm_2d.html">layer_conv_lstm_2d</a>()</code> <br><code><a href="layer_conv_lstm_3d.html">layer_conv_lstm_3d</a>()</code> <br><code><a href="layer_cropping_1d.html">layer_cropping_1d</a>()</code> <br><code><a href="layer_cropping_2d.html">layer_cropping_2d</a>()</code> <br><code><a href="layer_cropping_3d.html">layer_cropping_3d</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_dense.html">layer_dense</a>()</code> <br><code><a href="layer_depthwise_conv_1d.html">layer_depthwise_conv_1d</a>()</code> <br><code><a href="layer_depthwise_conv_2d.html">layer_depthwise_conv_2d</a>()</code> <br><code><a href="layer_discretization.html">layer_discretization</a>()</code> <br><code><a href="layer_dot.html">layer_dot</a>()</code> <br><code><a href="layer_dropout.html">layer_dropout</a>()</code> <br><code><a href="layer_einsum_dense.html">layer_einsum_dense</a>()</code> <br><code><a href="layer_embedding.html">layer_embedding</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_feature_space.html">layer_feature_space</a>()</code> <br><code><a href="layer_flatten.html">layer_flatten</a>()</code> <br><code><a href="layer_flax_module_wrapper.html">layer_flax_module_wrapper</a>()</code> <br><code><a href="layer_gaussian_dropout.html">layer_gaussian_dropout</a>()</code> <br><code><a href="layer_gaussian_noise.html">layer_gaussian_noise</a>()</code> <br><code><a href="layer_global_average_pooling_1d.html">layer_global_average_pooling_1d</a>()</code> <br><code><a href="layer_global_average_pooling_2d.html">layer_global_average_pooling_2d</a>()</code> <br><code><a href="layer_global_average_pooling_3d.html">layer_global_average_pooling_3d</a>()</code> <br><code><a href="layer_global_max_pooling_1d.html">layer_global_max_pooling_1d</a>()</code> <br><code><a href="layer_global_max_pooling_2d.html">layer_global_max_pooling_2d</a>()</code> <br><code><a href="layer_global_max_pooling_3d.html">layer_global_max_pooling_3d</a>()</code> <br><code><a href="layer_group_normalization.html">layer_group_normalization</a>()</code> <br><code><a href="layer_group_query_attention.html">layer_group_query_attention</a>()</code> <br><code><a href="layer_gru.html">layer_gru</a>()</code> <br><code><a href="layer_hashed_crossing.html">layer_hashed_crossing</a>()</code> <br><code><a href="layer_hashing.html">layer_hashing</a>()</code> <br><code><a href="layer_identity.html">layer_identity</a>()</code> <br><code><a href="layer_integer_lookup.html">layer_integer_lookup</a>()</code> <br><code><a href="layer_jax_model_wrapper.html">layer_jax_model_wrapper</a>()</code> <br><code><a href="layer_lambda.html">layer_lambda</a>()</code> <br><code><a href="layer_layer_normalization.html">layer_layer_normalization</a>()</code> <br><code><a href="layer_lstm.html">layer_lstm</a>()</code> <br><code><a href="layer_masking.html">layer_masking</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_max_pooling_1d.html">layer_max_pooling_1d</a>()</code> <br><code><a href="layer_max_pooling_2d.html">layer_max_pooling_2d</a>()</code> <br><code><a href="layer_max_pooling_3d.html">layer_max_pooling_3d</a>()</code> <br><code><a href="layer_maximum.html">layer_maximum</a>()</code> <br><code><a href="layer_mel_spectrogram.html">layer_mel_spectrogram</a>()</code> <br><code><a href="layer_minimum.html">layer_minimum</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_multi_head_attention.html">layer_multi_head_attention</a>()</code> <br><code><a href="layer_multiply.html">layer_multiply</a>()</code> <br><code><a href="layer_normalization.html">layer_normalization</a>()</code> <br><code><a href="layer_permute.html">layer_permute</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_brightness.html">layer_random_brightness</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_contrast.html">layer_random_contrast</a>()</code> <br><code><a href="layer_random_crop.html">layer_random_crop</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_flip.html">layer_random_flip</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_rotation.html">layer_random_rotation</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_random_translation.html">layer_random_translation</a>()</code> <br><code><a href="layer_random_zoom.html">layer_random_zoom</a>()</code> <br><code><a href="layer_repeat_vector.html">layer_repeat_vector</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_reshape.html">layer_reshape</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_rms_normalization.html">layer_rms_normalization</a>()</code> <br><code><a href="layer_rnn.html">layer_rnn</a>()</code> <br><code><a href="layer_separable_conv_1d.html">layer_separable_conv_1d</a>()</code> <br><code><a href="layer_separable_conv_2d.html">layer_separable_conv_2d</a>()</code> <br><code><a href="layer_simple_rnn.html">layer_simple_rnn</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br><code><a href="layer_spatial_dropout_1d.html">layer_spatial_dropout_1d</a>()</code> <br><code><a href="layer_spatial_dropout_2d.html">layer_spatial_dropout_2d</a>()</code> <br><code><a href="layer_spatial_dropout_3d.html">layer_spatial_dropout_3d</a>()</code> <br><code><a href="layer_spectral_normalization.html">layer_spectral_normalization</a>()</code> <br><code><a href="layer_stft_spectrogram.html">layer_stft_spectrogram</a>()</code> <br><code><a href="layer_string_lookup.html">layer_string_lookup</a>()</code> <br><code><a href="layer_subtract.html">layer_subtract</a>()</code> <br><code><a href="layer_text_vectorization.html">layer_text_vectorization</a>()</code> <br><code><a href="layer_tfsm.html">layer_tfsm</a>()</code> <br><code><a href="layer_time_distributed.html">layer_time_distributed</a>()</code> <br><code><a href="layer_torch_module_wrapper.html">layer_torch_module_wrapper</a>()</code> <br><code><a href="layer_unit_normalization.html">layer_unit_normalization</a>()</code> <br><code><a href="layer_upsampling_1d.html">layer_upsampling_1d</a>()</code> <br><code><a href="layer_upsampling_2d.html">layer_upsampling_2d</a>()</code> <br><code><a href="layer_upsampling_3d.html">layer_upsampling_3d</a>()</code> <br><code><a href="layer_zero_padding_1d.html">layer_zero_padding_1d</a>()</code> <br><code><a href="layer_zero_padding_2d.html">layer_zero_padding_2d</a>()</code> <br><code><a href="layer_zero_padding_3d.html">layer_zero_padding_3d</a>()</code> <br><code><a href="rnn_cell_gru.html">rnn_cell_gru</a>()</code> <br><code><a href="rnn_cell_lstm.html">rnn_cell_lstm</a>()</code> <br><code><a href="rnn_cell_simple.html">rnn_cell_simple</a>()</code> <br><code><a href="rnn_cells_stack.html">rnn_cells_stack</a>()</code> <br></p></div>
+<p>Other layers: <br><code><a href="Layer.html">Layer</a>()</code> <br><code><a href="layer_activation.html">layer_activation</a>()</code> <br><code><a href="layer_activation_elu.html">layer_activation_elu</a>()</code> <br><code><a href="layer_activation_leaky_relu.html">layer_activation_leaky_relu</a>()</code> <br><code><a href="layer_activation_parametric_relu.html">layer_activation_parametric_relu</a>()</code> <br><code><a href="layer_activation_relu.html">layer_activation_relu</a>()</code> <br><code><a href="layer_activation_softmax.html">layer_activation_softmax</a>()</code> <br><code><a href="layer_activity_regularization.html">layer_activity_regularization</a>()</code> <br><code><a href="layer_add.html">layer_add</a>()</code> <br><code><a href="layer_additive_attention.html">layer_additive_attention</a>()</code> <br><code><a href="layer_alpha_dropout.html">layer_alpha_dropout</a>()</code> <br><code><a href="layer_attention.html">layer_attention</a>()</code> <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_average_pooling_1d.html">layer_average_pooling_1d</a>()</code> <br><code><a href="layer_average_pooling_2d.html">layer_average_pooling_2d</a>()</code> <br><code><a href="layer_average_pooling_3d.html">layer_average_pooling_3d</a>()</code> <br><code><a href="layer_batch_normalization.html">layer_batch_normalization</a>()</code> <br><code><a href="layer_bidirectional.html">layer_bidirectional</a>()</code> <br><code><a href="layer_category_encoding.html">layer_category_encoding</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_concatenate.html">layer_concatenate</a>()</code> <br><code><a href="layer_conv_1d.html">layer_conv_1d</a>()</code> <br><code><a href="layer_conv_1d_transpose.html">layer_conv_1d_transpose</a>()</code> <br><code><a href="layer_conv_2d.html">layer_conv_2d</a>()</code> <br><code><a href="layer_conv_2d_transpose.html">layer_conv_2d_transpose</a>()</code> <br><code><a href="layer_conv_3d.html">layer_conv_3d</a>()</code> <br><code><a href="layer_conv_3d_transpose.html">layer_conv_3d_transpose</a>()</code> <br><code><a href="layer_conv_lstm_1d.html">layer_conv_lstm_1d</a>()</code> <br><code><a href="layer_conv_lstm_2d.html">layer_conv_lstm_2d</a>()</code> <br><code><a href="layer_conv_lstm_3d.html">layer_conv_lstm_3d</a>()</code> <br><code><a href="layer_cropping_1d.html">layer_cropping_1d</a>()</code> <br><code><a href="layer_cropping_2d.html">layer_cropping_2d</a>()</code> <br><code><a href="layer_cropping_3d.html">layer_cropping_3d</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_dense.html">layer_dense</a>()</code> <br><code><a href="layer_depthwise_conv_1d.html">layer_depthwise_conv_1d</a>()</code> <br><code><a href="layer_depthwise_conv_2d.html">layer_depthwise_conv_2d</a>()</code> <br><code><a href="layer_discretization.html">layer_discretization</a>()</code> <br><code><a href="layer_dot.html">layer_dot</a>()</code> <br><code><a href="layer_dropout.html">layer_dropout</a>()</code> <br><code><a href="layer_einsum_dense.html">layer_einsum_dense</a>()</code> <br><code><a href="layer_embedding.html">layer_embedding</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_feature_space.html">layer_feature_space</a>()</code> <br><code><a href="layer_flatten.html">layer_flatten</a>()</code> <br><code><a href="layer_flax_module_wrapper.html">layer_flax_module_wrapper</a>()</code> <br><code><a href="layer_gaussian_dropout.html">layer_gaussian_dropout</a>()</code> <br><code><a href="layer_gaussian_noise.html">layer_gaussian_noise</a>()</code> <br><code><a href="layer_global_average_pooling_1d.html">layer_global_average_pooling_1d</a>()</code> <br><code><a href="layer_global_average_pooling_2d.html">layer_global_average_pooling_2d</a>()</code> <br><code><a href="layer_global_average_pooling_3d.html">layer_global_average_pooling_3d</a>()</code> <br><code><a href="layer_global_max_pooling_1d.html">layer_global_max_pooling_1d</a>()</code> <br><code><a href="layer_global_max_pooling_2d.html">layer_global_max_pooling_2d</a>()</code> <br><code><a href="layer_global_max_pooling_3d.html">layer_global_max_pooling_3d</a>()</code> <br><code><a href="layer_group_normalization.html">layer_group_normalization</a>()</code> <br><code><a href="layer_group_query_attention.html">layer_group_query_attention</a>()</code> <br><code><a href="layer_gru.html">layer_gru</a>()</code> <br><code><a href="layer_hashed_crossing.html">layer_hashed_crossing</a>()</code> <br><code><a href="layer_hashing.html">layer_hashing</a>()</code> <br><code><a href="layer_identity.html">layer_identity</a>()</code> <br><code><a href="layer_integer_lookup.html">layer_integer_lookup</a>()</code> <br><code><a href="layer_jax_model_wrapper.html">layer_jax_model_wrapper</a>()</code> <br><code><a href="layer_lambda.html">layer_lambda</a>()</code> <br><code><a href="layer_layer_normalization.html">layer_layer_normalization</a>()</code> <br><code><a href="layer_lstm.html">layer_lstm</a>()</code> <br><code><a href="layer_masking.html">layer_masking</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_max_pooling_1d.html">layer_max_pooling_1d</a>()</code> <br><code><a href="layer_max_pooling_2d.html">layer_max_pooling_2d</a>()</code> <br><code><a href="layer_max_pooling_3d.html">layer_max_pooling_3d</a>()</code> <br><code><a href="layer_maximum.html">layer_maximum</a>()</code> <br><code><a href="layer_mel_spectrogram.html">layer_mel_spectrogram</a>()</code> <br><code><a href="layer_minimum.html">layer_minimum</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_multi_head_attention.html">layer_multi_head_attention</a>()</code> <br><code><a href="layer_multiply.html">layer_multiply</a>()</code> <br><code><a href="layer_normalization.html">layer_normalization</a>()</code> <br><code><a href="layer_permute.html">layer_permute</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_brightness.html">layer_random_brightness</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_contrast.html">layer_random_contrast</a>()</code> <br><code><a href="layer_random_crop.html">layer_random_crop</a>()</code> <br><code><a href="layer_random_elastic_transform.html">layer_random_elastic_transform</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_flip.html">layer_random_flip</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_rotation.html">layer_random_rotation</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_random_translation.html">layer_random_translation</a>()</code> <br><code><a href="layer_random_zoom.html">layer_random_zoom</a>()</code> <br><code><a href="layer_repeat_vector.html">layer_repeat_vector</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_reshape.html">layer_reshape</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_rms_normalization.html">layer_rms_normalization</a>()</code> <br><code><a href="layer_rnn.html">layer_rnn</a>()</code> <br><code><a href="layer_separable_conv_1d.html">layer_separable_conv_1d</a>()</code> <br><code><a href="layer_separable_conv_2d.html">layer_separable_conv_2d</a>()</code> <br><code><a href="layer_simple_rnn.html">layer_simple_rnn</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br><code><a href="layer_spatial_dropout_1d.html">layer_spatial_dropout_1d</a>()</code> <br><code><a href="layer_spatial_dropout_2d.html">layer_spatial_dropout_2d</a>()</code> <br><code><a href="layer_spatial_dropout_3d.html">layer_spatial_dropout_3d</a>()</code> <br><code><a href="layer_spectral_normalization.html">layer_spectral_normalization</a>()</code> <br><code><a href="layer_stft_spectrogram.html">layer_stft_spectrogram</a>()</code> <br><code><a href="layer_string_lookup.html">layer_string_lookup</a>()</code> <br><code><a href="layer_subtract.html">layer_subtract</a>()</code> <br><code><a href="layer_text_vectorization.html">layer_text_vectorization</a>()</code> <br><code><a href="layer_tfsm.html">layer_tfsm</a>()</code> <br><code><a href="layer_time_distributed.html">layer_time_distributed</a>()</code> <br><code><a href="layer_torch_module_wrapper.html">layer_torch_module_wrapper</a>()</code> <br><code><a href="layer_unit_normalization.html">layer_unit_normalization</a>()</code> <br><code><a href="layer_upsampling_1d.html">layer_upsampling_1d</a>()</code> <br><code><a href="layer_upsampling_2d.html">layer_upsampling_2d</a>()</code> <br><code><a href="layer_upsampling_3d.html">layer_upsampling_3d</a>()</code> <br><code><a href="layer_zero_padding_1d.html">layer_zero_padding_1d</a>()</code> <br><code><a href="layer_zero_padding_2d.html">layer_zero_padding_2d</a>()</code> <br><code><a href="layer_zero_padding_3d.html">layer_zero_padding_3d</a>()</code> <br><code><a href="rnn_cell_gru.html">rnn_cell_gru</a>()</code> <br><code><a href="rnn_cell_lstm.html">rnn_cell_lstm</a>()</code> <br><code><a href="rnn_cell_simple.html">rnn_cell_simple</a>()</code> <br><code><a href="rnn_cells_stack.html">rnn_cells_stack</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -125,7 +125,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/layer_average_pooling_1d.html b/docs/dev/reference/layer_average_pooling_1d.html
index 1b3919805..703614450 100644
--- a/docs/dev/reference/layer_average_pooling_1d.html
+++ b/docs/dev/reference/layer_average_pooling_1d.html
@@ -18,7 +18,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -198,7 +198,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
     <div class="dont-index">
 <ul><li><p><a href="https://keras.io/api/layers/pooling_layers/average_pooling1d#averagepooling1d-class" class="external-link">https://keras.io/api/layers/pooling_layers/average_pooling1d#averagepooling1d-class</a></p></li>
 </ul><p>Other pooling layers: <br><code><a href="layer_average_pooling_2d.html">layer_average_pooling_2d</a>()</code> <br><code><a href="layer_average_pooling_3d.html">layer_average_pooling_3d</a>()</code> <br><code><a href="layer_global_average_pooling_1d.html">layer_global_average_pooling_1d</a>()</code> <br><code><a href="layer_global_average_pooling_2d.html">layer_global_average_pooling_2d</a>()</code> <br><code><a href="layer_global_average_pooling_3d.html">layer_global_average_pooling_3d</a>()</code> <br><code><a href="layer_global_max_pooling_1d.html">layer_global_max_pooling_1d</a>()</code> <br><code><a href="layer_global_max_pooling_2d.html">layer_global_max_pooling_2d</a>()</code> <br><code><a href="layer_global_max_pooling_3d.html">layer_global_max_pooling_3d</a>()</code> <br><code><a href="layer_max_pooling_1d.html">layer_max_pooling_1d</a>()</code> <br><code><a href="layer_max_pooling_2d.html">layer_max_pooling_2d</a>()</code> <br><code><a href="layer_max_pooling_3d.html">layer_max_pooling_3d</a>()</code> <br></p>
-<p>Other layers: <br><code><a href="Layer.html">Layer</a>()</code> <br><code><a href="layer_activation.html">layer_activation</a>()</code> <br><code><a href="layer_activation_elu.html">layer_activation_elu</a>()</code> <br><code><a href="layer_activation_leaky_relu.html">layer_activation_leaky_relu</a>()</code> <br><code><a href="layer_activation_parametric_relu.html">layer_activation_parametric_relu</a>()</code> <br><code><a href="layer_activation_relu.html">layer_activation_relu</a>()</code> <br><code><a href="layer_activation_softmax.html">layer_activation_softmax</a>()</code> <br><code><a href="layer_activity_regularization.html">layer_activity_regularization</a>()</code> <br><code><a href="layer_add.html">layer_add</a>()</code> <br><code><a href="layer_additive_attention.html">layer_additive_attention</a>()</code> <br><code><a href="layer_alpha_dropout.html">layer_alpha_dropout</a>()</code> <br><code><a href="layer_attention.html">layer_attention</a>()</code> <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_average.html">layer_average</a>()</code> <br><code><a href="layer_average_pooling_2d.html">layer_average_pooling_2d</a>()</code> <br><code><a href="layer_average_pooling_3d.html">layer_average_pooling_3d</a>()</code> <br><code><a href="layer_batch_normalization.html">layer_batch_normalization</a>()</code> <br><code><a href="layer_bidirectional.html">layer_bidirectional</a>()</code> <br><code><a href="layer_category_encoding.html">layer_category_encoding</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_concatenate.html">layer_concatenate</a>()</code> <br><code><a href="layer_conv_1d.html">layer_conv_1d</a>()</code> <br><code><a href="layer_conv_1d_transpose.html">layer_conv_1d_transpose</a>()</code> <br><code><a href="layer_conv_2d.html">layer_conv_2d</a>()</code> <br><code><a href="layer_conv_2d_transpose.html">layer_conv_2d_transpose</a>()</code> <br><code><a href="layer_conv_3d.html">layer_conv_3d</a>()</code> <br><code><a href="layer_conv_3d_transpose.html">layer_conv_3d_transpose</a>()</code> <br><code><a href="layer_conv_lstm_1d.html">layer_conv_lstm_1d</a>()</code> <br><code><a href="layer_conv_lstm_2d.html">layer_conv_lstm_2d</a>()</code> <br><code><a href="layer_conv_lstm_3d.html">layer_conv_lstm_3d</a>()</code> <br><code><a href="layer_cropping_1d.html">layer_cropping_1d</a>()</code> <br><code><a href="layer_cropping_2d.html">layer_cropping_2d</a>()</code> <br><code><a href="layer_cropping_3d.html">layer_cropping_3d</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_dense.html">layer_dense</a>()</code> <br><code><a href="layer_depthwise_conv_1d.html">layer_depthwise_conv_1d</a>()</code> <br><code><a href="layer_depthwise_conv_2d.html">layer_depthwise_conv_2d</a>()</code> <br><code><a href="layer_discretization.html">layer_discretization</a>()</code> <br><code><a href="layer_dot.html">layer_dot</a>()</code> <br><code><a href="layer_dropout.html">layer_dropout</a>()</code> <br><code><a href="layer_einsum_dense.html">layer_einsum_dense</a>()</code> <br><code><a href="layer_embedding.html">layer_embedding</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_feature_space.html">layer_feature_space</a>()</code> <br><code><a href="layer_flatten.html">layer_flatten</a>()</code> <br><code><a href="layer_flax_module_wrapper.html">layer_flax_module_wrapper</a>()</code> <br><code><a href="layer_gaussian_dropout.html">layer_gaussian_dropout</a>()</code> <br><code><a href="layer_gaussian_noise.html">layer_gaussian_noise</a>()</code> <br><code><a href="layer_global_average_pooling_1d.html">layer_global_average_pooling_1d</a>()</code> <br><code><a href="layer_global_average_pooling_2d.html">layer_global_average_pooling_2d</a>()</code> <br><code><a href="layer_global_average_pooling_3d.html">layer_global_average_pooling_3d</a>()</code> <br><code><a href="layer_global_max_pooling_1d.html">layer_global_max_pooling_1d</a>()</code> <br><code><a href="layer_global_max_pooling_2d.html">layer_global_max_pooling_2d</a>()</code> <br><code><a href="layer_global_max_pooling_3d.html">layer_global_max_pooling_3d</a>()</code> <br><code><a href="layer_group_normalization.html">layer_group_normalization</a>()</code> <br><code><a href="layer_group_query_attention.html">layer_group_query_attention</a>()</code> <br><code><a href="layer_gru.html">layer_gru</a>()</code> <br><code><a href="layer_hashed_crossing.html">layer_hashed_crossing</a>()</code> <br><code><a href="layer_hashing.html">layer_hashing</a>()</code> <br><code><a href="layer_identity.html">layer_identity</a>()</code> <br><code><a href="layer_integer_lookup.html">layer_integer_lookup</a>()</code> <br><code><a href="layer_jax_model_wrapper.html">layer_jax_model_wrapper</a>()</code> <br><code><a href="layer_lambda.html">layer_lambda</a>()</code> <br><code><a href="layer_layer_normalization.html">layer_layer_normalization</a>()</code> <br><code><a href="layer_lstm.html">layer_lstm</a>()</code> <br><code><a href="layer_masking.html">layer_masking</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_max_pooling_1d.html">layer_max_pooling_1d</a>()</code> <br><code><a href="layer_max_pooling_2d.html">layer_max_pooling_2d</a>()</code> <br><code><a href="layer_max_pooling_3d.html">layer_max_pooling_3d</a>()</code> <br><code><a href="layer_maximum.html">layer_maximum</a>()</code> <br><code><a href="layer_mel_spectrogram.html">layer_mel_spectrogram</a>()</code> <br><code><a href="layer_minimum.html">layer_minimum</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_multi_head_attention.html">layer_multi_head_attention</a>()</code> <br><code><a href="layer_multiply.html">layer_multiply</a>()</code> <br><code><a href="layer_normalization.html">layer_normalization</a>()</code> <br><code><a href="layer_permute.html">layer_permute</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_brightness.html">layer_random_brightness</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_contrast.html">layer_random_contrast</a>()</code> <br><code><a href="layer_random_crop.html">layer_random_crop</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_flip.html">layer_random_flip</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_rotation.html">layer_random_rotation</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_random_translation.html">layer_random_translation</a>()</code> <br><code><a href="layer_random_zoom.html">layer_random_zoom</a>()</code> <br><code><a href="layer_repeat_vector.html">layer_repeat_vector</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_reshape.html">layer_reshape</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_rms_normalization.html">layer_rms_normalization</a>()</code> <br><code><a href="layer_rnn.html">layer_rnn</a>()</code> <br><code><a href="layer_separable_conv_1d.html">layer_separable_conv_1d</a>()</code> <br><code><a href="layer_separable_conv_2d.html">layer_separable_conv_2d</a>()</code> <br><code><a href="layer_simple_rnn.html">layer_simple_rnn</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br><code><a href="layer_spatial_dropout_1d.html">layer_spatial_dropout_1d</a>()</code> <br><code><a href="layer_spatial_dropout_2d.html">layer_spatial_dropout_2d</a>()</code> <br><code><a href="layer_spatial_dropout_3d.html">layer_spatial_dropout_3d</a>()</code> <br><code><a href="layer_spectral_normalization.html">layer_spectral_normalization</a>()</code> <br><code><a href="layer_stft_spectrogram.html">layer_stft_spectrogram</a>()</code> <br><code><a href="layer_string_lookup.html">layer_string_lookup</a>()</code> <br><code><a href="layer_subtract.html">layer_subtract</a>()</code> <br><code><a href="layer_text_vectorization.html">layer_text_vectorization</a>()</code> <br><code><a href="layer_tfsm.html">layer_tfsm</a>()</code> <br><code><a href="layer_time_distributed.html">layer_time_distributed</a>()</code> <br><code><a href="layer_torch_module_wrapper.html">layer_torch_module_wrapper</a>()</code> <br><code><a href="layer_unit_normalization.html">layer_unit_normalization</a>()</code> <br><code><a href="layer_upsampling_1d.html">layer_upsampling_1d</a>()</code> <br><code><a href="layer_upsampling_2d.html">layer_upsampling_2d</a>()</code> <br><code><a href="layer_upsampling_3d.html">layer_upsampling_3d</a>()</code> <br><code><a href="layer_zero_padding_1d.html">layer_zero_padding_1d</a>()</code> <br><code><a href="layer_zero_padding_2d.html">layer_zero_padding_2d</a>()</code> <br><code><a href="layer_zero_padding_3d.html">layer_zero_padding_3d</a>()</code> <br><code><a href="rnn_cell_gru.html">rnn_cell_gru</a>()</code> <br><code><a href="rnn_cell_lstm.html">rnn_cell_lstm</a>()</code> <br><code><a href="rnn_cell_simple.html">rnn_cell_simple</a>()</code> <br><code><a href="rnn_cells_stack.html">rnn_cells_stack</a>()</code> <br></p></div>
+<p>Other layers: <br><code><a href="Layer.html">Layer</a>()</code> <br><code><a href="layer_activation.html">layer_activation</a>()</code> <br><code><a href="layer_activation_elu.html">layer_activation_elu</a>()</code> <br><code><a href="layer_activation_leaky_relu.html">layer_activation_leaky_relu</a>()</code> <br><code><a href="layer_activation_parametric_relu.html">layer_activation_parametric_relu</a>()</code> <br><code><a href="layer_activation_relu.html">layer_activation_relu</a>()</code> <br><code><a href="layer_activation_softmax.html">layer_activation_softmax</a>()</code> <br><code><a href="layer_activity_regularization.html">layer_activity_regularization</a>()</code> <br><code><a href="layer_add.html">layer_add</a>()</code> <br><code><a href="layer_additive_attention.html">layer_additive_attention</a>()</code> <br><code><a href="layer_alpha_dropout.html">layer_alpha_dropout</a>()</code> <br><code><a href="layer_attention.html">layer_attention</a>()</code> <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_average.html">layer_average</a>()</code> <br><code><a href="layer_average_pooling_2d.html">layer_average_pooling_2d</a>()</code> <br><code><a href="layer_average_pooling_3d.html">layer_average_pooling_3d</a>()</code> <br><code><a href="layer_batch_normalization.html">layer_batch_normalization</a>()</code> <br><code><a href="layer_bidirectional.html">layer_bidirectional</a>()</code> <br><code><a href="layer_category_encoding.html">layer_category_encoding</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_concatenate.html">layer_concatenate</a>()</code> <br><code><a href="layer_conv_1d.html">layer_conv_1d</a>()</code> <br><code><a href="layer_conv_1d_transpose.html">layer_conv_1d_transpose</a>()</code> <br><code><a href="layer_conv_2d.html">layer_conv_2d</a>()</code> <br><code><a href="layer_conv_2d_transpose.html">layer_conv_2d_transpose</a>()</code> <br><code><a href="layer_conv_3d.html">layer_conv_3d</a>()</code> <br><code><a href="layer_conv_3d_transpose.html">layer_conv_3d_transpose</a>()</code> <br><code><a href="layer_conv_lstm_1d.html">layer_conv_lstm_1d</a>()</code> <br><code><a href="layer_conv_lstm_2d.html">layer_conv_lstm_2d</a>()</code> <br><code><a href="layer_conv_lstm_3d.html">layer_conv_lstm_3d</a>()</code> <br><code><a href="layer_cropping_1d.html">layer_cropping_1d</a>()</code> <br><code><a href="layer_cropping_2d.html">layer_cropping_2d</a>()</code> <br><code><a href="layer_cropping_3d.html">layer_cropping_3d</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_dense.html">layer_dense</a>()</code> <br><code><a href="layer_depthwise_conv_1d.html">layer_depthwise_conv_1d</a>()</code> <br><code><a href="layer_depthwise_conv_2d.html">layer_depthwise_conv_2d</a>()</code> <br><code><a href="layer_discretization.html">layer_discretization</a>()</code> <br><code><a href="layer_dot.html">layer_dot</a>()</code> <br><code><a href="layer_dropout.html">layer_dropout</a>()</code> <br><code><a href="layer_einsum_dense.html">layer_einsum_dense</a>()</code> <br><code><a href="layer_embedding.html">layer_embedding</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_feature_space.html">layer_feature_space</a>()</code> <br><code><a href="layer_flatten.html">layer_flatten</a>()</code> <br><code><a href="layer_flax_module_wrapper.html">layer_flax_module_wrapper</a>()</code> <br><code><a href="layer_gaussian_dropout.html">layer_gaussian_dropout</a>()</code> <br><code><a href="layer_gaussian_noise.html">layer_gaussian_noise</a>()</code> <br><code><a href="layer_global_average_pooling_1d.html">layer_global_average_pooling_1d</a>()</code> <br><code><a href="layer_global_average_pooling_2d.html">layer_global_average_pooling_2d</a>()</code> <br><code><a href="layer_global_average_pooling_3d.html">layer_global_average_pooling_3d</a>()</code> <br><code><a href="layer_global_max_pooling_1d.html">layer_global_max_pooling_1d</a>()</code> <br><code><a href="layer_global_max_pooling_2d.html">layer_global_max_pooling_2d</a>()</code> <br><code><a href="layer_global_max_pooling_3d.html">layer_global_max_pooling_3d</a>()</code> <br><code><a href="layer_group_normalization.html">layer_group_normalization</a>()</code> <br><code><a href="layer_group_query_attention.html">layer_group_query_attention</a>()</code> <br><code><a href="layer_gru.html">layer_gru</a>()</code> <br><code><a href="layer_hashed_crossing.html">layer_hashed_crossing</a>()</code> <br><code><a href="layer_hashing.html">layer_hashing</a>()</code> <br><code><a href="layer_identity.html">layer_identity</a>()</code> <br><code><a href="layer_integer_lookup.html">layer_integer_lookup</a>()</code> <br><code><a href="layer_jax_model_wrapper.html">layer_jax_model_wrapper</a>()</code> <br><code><a href="layer_lambda.html">layer_lambda</a>()</code> <br><code><a href="layer_layer_normalization.html">layer_layer_normalization</a>()</code> <br><code><a href="layer_lstm.html">layer_lstm</a>()</code> <br><code><a href="layer_masking.html">layer_masking</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_max_pooling_1d.html">layer_max_pooling_1d</a>()</code> <br><code><a href="layer_max_pooling_2d.html">layer_max_pooling_2d</a>()</code> <br><code><a href="layer_max_pooling_3d.html">layer_max_pooling_3d</a>()</code> <br><code><a href="layer_maximum.html">layer_maximum</a>()</code> <br><code><a href="layer_mel_spectrogram.html">layer_mel_spectrogram</a>()</code> <br><code><a href="layer_minimum.html">layer_minimum</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_multi_head_attention.html">layer_multi_head_attention</a>()</code> <br><code><a href="layer_multiply.html">layer_multiply</a>()</code> <br><code><a href="layer_normalization.html">layer_normalization</a>()</code> <br><code><a href="layer_permute.html">layer_permute</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_brightness.html">layer_random_brightness</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_contrast.html">layer_random_contrast</a>()</code> <br><code><a href="layer_random_crop.html">layer_random_crop</a>()</code> <br><code><a href="layer_random_elastic_transform.html">layer_random_elastic_transform</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_flip.html">layer_random_flip</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_rotation.html">layer_random_rotation</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_random_translation.html">layer_random_translation</a>()</code> <br><code><a href="layer_random_zoom.html">layer_random_zoom</a>()</code> <br><code><a href="layer_repeat_vector.html">layer_repeat_vector</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_reshape.html">layer_reshape</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_rms_normalization.html">layer_rms_normalization</a>()</code> <br><code><a href="layer_rnn.html">layer_rnn</a>()</code> <br><code><a href="layer_separable_conv_1d.html">layer_separable_conv_1d</a>()</code> <br><code><a href="layer_separable_conv_2d.html">layer_separable_conv_2d</a>()</code> <br><code><a href="layer_simple_rnn.html">layer_simple_rnn</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br><code><a href="layer_spatial_dropout_1d.html">layer_spatial_dropout_1d</a>()</code> <br><code><a href="layer_spatial_dropout_2d.html">layer_spatial_dropout_2d</a>()</code> <br><code><a href="layer_spatial_dropout_3d.html">layer_spatial_dropout_3d</a>()</code> <br><code><a href="layer_spectral_normalization.html">layer_spectral_normalization</a>()</code> <br><code><a href="layer_stft_spectrogram.html">layer_stft_spectrogram</a>()</code> <br><code><a href="layer_string_lookup.html">layer_string_lookup</a>()</code> <br><code><a href="layer_subtract.html">layer_subtract</a>()</code> <br><code><a href="layer_text_vectorization.html">layer_text_vectorization</a>()</code> <br><code><a href="layer_tfsm.html">layer_tfsm</a>()</code> <br><code><a href="layer_time_distributed.html">layer_time_distributed</a>()</code> <br><code><a href="layer_torch_module_wrapper.html">layer_torch_module_wrapper</a>()</code> <br><code><a href="layer_unit_normalization.html">layer_unit_normalization</a>()</code> <br><code><a href="layer_upsampling_1d.html">layer_upsampling_1d</a>()</code> <br><code><a href="layer_upsampling_2d.html">layer_upsampling_2d</a>()</code> <br><code><a href="layer_upsampling_3d.html">layer_upsampling_3d</a>()</code> <br><code><a href="layer_zero_padding_1d.html">layer_zero_padding_1d</a>()</code> <br><code><a href="layer_zero_padding_2d.html">layer_zero_padding_2d</a>()</code> <br><code><a href="layer_zero_padding_3d.html">layer_zero_padding_3d</a>()</code> <br><code><a href="rnn_cell_gru.html">rnn_cell_gru</a>()</code> <br><code><a href="rnn_cell_lstm.html">rnn_cell_lstm</a>()</code> <br><code><a href="rnn_cell_simple.html">rnn_cell_simple</a>()</code> <br><code><a href="rnn_cells_stack.html">rnn_cells_stack</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -210,7 +210,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/layer_average_pooling_2d.html b/docs/dev/reference/layer_average_pooling_2d.html
index b1a9b5650..7413f3f9c 100644
--- a/docs/dev/reference/layer_average_pooling_2d.html
+++ b/docs/dev/reference/layer_average_pooling_2d.html
@@ -8,7 +8,7 @@
 output_shape = math.floor((input_shape - pool_size) / strides) + 1
 (when input_shape &amp;gt;= pool_size)
 The resulting output shape when using the "same" padding option is:
-output_shape = math.floor((input_shape - 1) / strides) + 1'><meta property="og:description" content='Downsamples the input along its spatial dimensions (height and width)
+output_shape = input_shape'><meta property="og:description" content='Downsamples the input along its spatial dimensions (height and width)
 by taking the average value over an input window
 (of size defined by pool_size) for each channel of the input.
 The window is shifted by strides along each dimension.
@@ -17,7 +17,7 @@
 output_shape = math.floor((input_shape - pool_size) / strides) + 1
 (when input_shape &amp;gt;= pool_size)
 The resulting output shape when using the "same" padding option is:
-output_shape = math.floor((input_shape - 1) / strides) + 1'><meta name="robots" content="noindex"><!-- Google Tag Manager --><script>(function(w,d,s,l,i){w[l]=w[l]||[];w[l].push({'gtm.start': new Date().getTime(),event:'gtm.js'});var f=d.getElementsByTagName(s)[0],j=d.createElement(s),dl=l!='dataLayer'?'&l='+l:'';j.async=true;j.src='https://www.googletagmanager.com/gtm.js?id='+i+dl;f.parentNode.insertBefore(j,f);})(window,document,'script','dataLayer','GTM-KHBDBW7');</script><!-- End Google Tag Manager --></head><body>
+output_shape = input_shape'><meta name="robots" content="noindex"><!-- Google Tag Manager --><script>(function(w,d,s,l,i){w[l]=w[l]||[];w[l].push({'gtm.start': new Date().getTime(),event:'gtm.js'});var f=d.getElementsByTagName(s)[0],j=d.createElement(s),dl=l!='dataLayer'?'&l='+l:'';j.async=true;j.src='https://www.googletagmanager.com/gtm.js?id='+i+dl;f.parentNode.insertBefore(j,f);})(window,document,'script','dataLayer','GTM-KHBDBW7');</script><!-- End Google Tag Manager --></head><body>
     <a href="#main" class="visually-hidden-focusable">Skip to contents</a>
     <!-- Google Tag Manager (noscript) --> <noscript><iframe src="https://www.googletagmanager.com/ns.html?id=GTM-KHBDBW7" height="0" width="0" style="display:none;visibility:hidden"></iframe></noscript> <!-- End Google Tag Manager (noscript) -->
 
@@ -26,7 +26,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -81,7 +81,7 @@ <h1>Average pooling operation for 2D spatial data.</h1>
 <code>output_shape = math.floor((input_shape - pool_size) / strides) + 1</code>
 (when <code>input_shape &gt;= pool_size</code>)</p>
 <p>The resulting output shape when using the <code>"same"</code> padding option is:
-<code>output_shape = math.floor((input_shape - 1) / strides) + 1</code></p>
+<code>output_shape = input_shape</code></p>
     </div>
 
     <div class="section level2">
@@ -223,7 +223,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
     <div class="dont-index">
 <ul><li><p><a href="https://keras.io/api/layers/pooling_layers/average_pooling2d#averagepooling2d-class" class="external-link">https://keras.io/api/layers/pooling_layers/average_pooling2d#averagepooling2d-class</a></p></li>
 </ul><p>Other pooling layers: <br><code><a href="layer_average_pooling_1d.html">layer_average_pooling_1d</a>()</code> <br><code><a href="layer_average_pooling_3d.html">layer_average_pooling_3d</a>()</code> <br><code><a href="layer_global_average_pooling_1d.html">layer_global_average_pooling_1d</a>()</code> <br><code><a href="layer_global_average_pooling_2d.html">layer_global_average_pooling_2d</a>()</code> <br><code><a href="layer_global_average_pooling_3d.html">layer_global_average_pooling_3d</a>()</code> <br><code><a href="layer_global_max_pooling_1d.html">layer_global_max_pooling_1d</a>()</code> <br><code><a href="layer_global_max_pooling_2d.html">layer_global_max_pooling_2d</a>()</code> <br><code><a href="layer_global_max_pooling_3d.html">layer_global_max_pooling_3d</a>()</code> <br><code><a href="layer_max_pooling_1d.html">layer_max_pooling_1d</a>()</code> <br><code><a href="layer_max_pooling_2d.html">layer_max_pooling_2d</a>()</code> <br><code><a href="layer_max_pooling_3d.html">layer_max_pooling_3d</a>()</code> <br></p>
-<p>Other layers: <br><code><a href="Layer.html">Layer</a>()</code> <br><code><a href="layer_activation.html">layer_activation</a>()</code> <br><code><a href="layer_activation_elu.html">layer_activation_elu</a>()</code> <br><code><a href="layer_activation_leaky_relu.html">layer_activation_leaky_relu</a>()</code> <br><code><a href="layer_activation_parametric_relu.html">layer_activation_parametric_relu</a>()</code> <br><code><a href="layer_activation_relu.html">layer_activation_relu</a>()</code> <br><code><a href="layer_activation_softmax.html">layer_activation_softmax</a>()</code> <br><code><a href="layer_activity_regularization.html">layer_activity_regularization</a>()</code> <br><code><a href="layer_add.html">layer_add</a>()</code> <br><code><a href="layer_additive_attention.html">layer_additive_attention</a>()</code> <br><code><a href="layer_alpha_dropout.html">layer_alpha_dropout</a>()</code> <br><code><a href="layer_attention.html">layer_attention</a>()</code> <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_average.html">layer_average</a>()</code> <br><code><a href="layer_average_pooling_1d.html">layer_average_pooling_1d</a>()</code> <br><code><a href="layer_average_pooling_3d.html">layer_average_pooling_3d</a>()</code> <br><code><a href="layer_batch_normalization.html">layer_batch_normalization</a>()</code> <br><code><a href="layer_bidirectional.html">layer_bidirectional</a>()</code> <br><code><a href="layer_category_encoding.html">layer_category_encoding</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_concatenate.html">layer_concatenate</a>()</code> <br><code><a href="layer_conv_1d.html">layer_conv_1d</a>()</code> <br><code><a href="layer_conv_1d_transpose.html">layer_conv_1d_transpose</a>()</code> <br><code><a href="layer_conv_2d.html">layer_conv_2d</a>()</code> <br><code><a href="layer_conv_2d_transpose.html">layer_conv_2d_transpose</a>()</code> <br><code><a href="layer_conv_3d.html">layer_conv_3d</a>()</code> <br><code><a href="layer_conv_3d_transpose.html">layer_conv_3d_transpose</a>()</code> <br><code><a href="layer_conv_lstm_1d.html">layer_conv_lstm_1d</a>()</code> <br><code><a href="layer_conv_lstm_2d.html">layer_conv_lstm_2d</a>()</code> <br><code><a href="layer_conv_lstm_3d.html">layer_conv_lstm_3d</a>()</code> <br><code><a href="layer_cropping_1d.html">layer_cropping_1d</a>()</code> <br><code><a href="layer_cropping_2d.html">layer_cropping_2d</a>()</code> <br><code><a href="layer_cropping_3d.html">layer_cropping_3d</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_dense.html">layer_dense</a>()</code> <br><code><a href="layer_depthwise_conv_1d.html">layer_depthwise_conv_1d</a>()</code> <br><code><a href="layer_depthwise_conv_2d.html">layer_depthwise_conv_2d</a>()</code> <br><code><a href="layer_discretization.html">layer_discretization</a>()</code> <br><code><a href="layer_dot.html">layer_dot</a>()</code> <br><code><a href="layer_dropout.html">layer_dropout</a>()</code> <br><code><a href="layer_einsum_dense.html">layer_einsum_dense</a>()</code> <br><code><a href="layer_embedding.html">layer_embedding</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_feature_space.html">layer_feature_space</a>()</code> <br><code><a href="layer_flatten.html">layer_flatten</a>()</code> <br><code><a href="layer_flax_module_wrapper.html">layer_flax_module_wrapper</a>()</code> <br><code><a href="layer_gaussian_dropout.html">layer_gaussian_dropout</a>()</code> <br><code><a href="layer_gaussian_noise.html">layer_gaussian_noise</a>()</code> <br><code><a href="layer_global_average_pooling_1d.html">layer_global_average_pooling_1d</a>()</code> <br><code><a href="layer_global_average_pooling_2d.html">layer_global_average_pooling_2d</a>()</code> <br><code><a href="layer_global_average_pooling_3d.html">layer_global_average_pooling_3d</a>()</code> <br><code><a href="layer_global_max_pooling_1d.html">layer_global_max_pooling_1d</a>()</code> <br><code><a href="layer_global_max_pooling_2d.html">layer_global_max_pooling_2d</a>()</code> <br><code><a href="layer_global_max_pooling_3d.html">layer_global_max_pooling_3d</a>()</code> <br><code><a href="layer_group_normalization.html">layer_group_normalization</a>()</code> <br><code><a href="layer_group_query_attention.html">layer_group_query_attention</a>()</code> <br><code><a href="layer_gru.html">layer_gru</a>()</code> <br><code><a href="layer_hashed_crossing.html">layer_hashed_crossing</a>()</code> <br><code><a href="layer_hashing.html">layer_hashing</a>()</code> <br><code><a href="layer_identity.html">layer_identity</a>()</code> <br><code><a href="layer_integer_lookup.html">layer_integer_lookup</a>()</code> <br><code><a href="layer_jax_model_wrapper.html">layer_jax_model_wrapper</a>()</code> <br><code><a href="layer_lambda.html">layer_lambda</a>()</code> <br><code><a href="layer_layer_normalization.html">layer_layer_normalization</a>()</code> <br><code><a href="layer_lstm.html">layer_lstm</a>()</code> <br><code><a href="layer_masking.html">layer_masking</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_max_pooling_1d.html">layer_max_pooling_1d</a>()</code> <br><code><a href="layer_max_pooling_2d.html">layer_max_pooling_2d</a>()</code> <br><code><a href="layer_max_pooling_3d.html">layer_max_pooling_3d</a>()</code> <br><code><a href="layer_maximum.html">layer_maximum</a>()</code> <br><code><a href="layer_mel_spectrogram.html">layer_mel_spectrogram</a>()</code> <br><code><a href="layer_minimum.html">layer_minimum</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_multi_head_attention.html">layer_multi_head_attention</a>()</code> <br><code><a href="layer_multiply.html">layer_multiply</a>()</code> <br><code><a href="layer_normalization.html">layer_normalization</a>()</code> <br><code><a href="layer_permute.html">layer_permute</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_brightness.html">layer_random_brightness</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_contrast.html">layer_random_contrast</a>()</code> <br><code><a href="layer_random_crop.html">layer_random_crop</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_flip.html">layer_random_flip</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_rotation.html">layer_random_rotation</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_random_translation.html">layer_random_translation</a>()</code> <br><code><a href="layer_random_zoom.html">layer_random_zoom</a>()</code> <br><code><a href="layer_repeat_vector.html">layer_repeat_vector</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_reshape.html">layer_reshape</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_rms_normalization.html">layer_rms_normalization</a>()</code> <br><code><a href="layer_rnn.html">layer_rnn</a>()</code> <br><code><a href="layer_separable_conv_1d.html">layer_separable_conv_1d</a>()</code> <br><code><a href="layer_separable_conv_2d.html">layer_separable_conv_2d</a>()</code> <br><code><a href="layer_simple_rnn.html">layer_simple_rnn</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br><code><a href="layer_spatial_dropout_1d.html">layer_spatial_dropout_1d</a>()</code> <br><code><a href="layer_spatial_dropout_2d.html">layer_spatial_dropout_2d</a>()</code> <br><code><a href="layer_spatial_dropout_3d.html">layer_spatial_dropout_3d</a>()</code> <br><code><a href="layer_spectral_normalization.html">layer_spectral_normalization</a>()</code> <br><code><a href="layer_stft_spectrogram.html">layer_stft_spectrogram</a>()</code> <br><code><a href="layer_string_lookup.html">layer_string_lookup</a>()</code> <br><code><a href="layer_subtract.html">layer_subtract</a>()</code> <br><code><a href="layer_text_vectorization.html">layer_text_vectorization</a>()</code> <br><code><a href="layer_tfsm.html">layer_tfsm</a>()</code> <br><code><a href="layer_time_distributed.html">layer_time_distributed</a>()</code> <br><code><a href="layer_torch_module_wrapper.html">layer_torch_module_wrapper</a>()</code> <br><code><a href="layer_unit_normalization.html">layer_unit_normalization</a>()</code> <br><code><a href="layer_upsampling_1d.html">layer_upsampling_1d</a>()</code> <br><code><a href="layer_upsampling_2d.html">layer_upsampling_2d</a>()</code> <br><code><a href="layer_upsampling_3d.html">layer_upsampling_3d</a>()</code> <br><code><a href="layer_zero_padding_1d.html">layer_zero_padding_1d</a>()</code> <br><code><a href="layer_zero_padding_2d.html">layer_zero_padding_2d</a>()</code> <br><code><a href="layer_zero_padding_3d.html">layer_zero_padding_3d</a>()</code> <br><code><a href="rnn_cell_gru.html">rnn_cell_gru</a>()</code> <br><code><a href="rnn_cell_lstm.html">rnn_cell_lstm</a>()</code> <br><code><a href="rnn_cell_simple.html">rnn_cell_simple</a>()</code> <br><code><a href="rnn_cells_stack.html">rnn_cells_stack</a>()</code> <br></p></div>
+<p>Other layers: <br><code><a href="Layer.html">Layer</a>()</code> <br><code><a href="layer_activation.html">layer_activation</a>()</code> <br><code><a href="layer_activation_elu.html">layer_activation_elu</a>()</code> <br><code><a href="layer_activation_leaky_relu.html">layer_activation_leaky_relu</a>()</code> <br><code><a href="layer_activation_parametric_relu.html">layer_activation_parametric_relu</a>()</code> <br><code><a href="layer_activation_relu.html">layer_activation_relu</a>()</code> <br><code><a href="layer_activation_softmax.html">layer_activation_softmax</a>()</code> <br><code><a href="layer_activity_regularization.html">layer_activity_regularization</a>()</code> <br><code><a href="layer_add.html">layer_add</a>()</code> <br><code><a href="layer_additive_attention.html">layer_additive_attention</a>()</code> <br><code><a href="layer_alpha_dropout.html">layer_alpha_dropout</a>()</code> <br><code><a href="layer_attention.html">layer_attention</a>()</code> <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_average.html">layer_average</a>()</code> <br><code><a href="layer_average_pooling_1d.html">layer_average_pooling_1d</a>()</code> <br><code><a href="layer_average_pooling_3d.html">layer_average_pooling_3d</a>()</code> <br><code><a href="layer_batch_normalization.html">layer_batch_normalization</a>()</code> <br><code><a href="layer_bidirectional.html">layer_bidirectional</a>()</code> <br><code><a href="layer_category_encoding.html">layer_category_encoding</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_concatenate.html">layer_concatenate</a>()</code> <br><code><a href="layer_conv_1d.html">layer_conv_1d</a>()</code> <br><code><a href="layer_conv_1d_transpose.html">layer_conv_1d_transpose</a>()</code> <br><code><a href="layer_conv_2d.html">layer_conv_2d</a>()</code> <br><code><a href="layer_conv_2d_transpose.html">layer_conv_2d_transpose</a>()</code> <br><code><a href="layer_conv_3d.html">layer_conv_3d</a>()</code> <br><code><a href="layer_conv_3d_transpose.html">layer_conv_3d_transpose</a>()</code> <br><code><a href="layer_conv_lstm_1d.html">layer_conv_lstm_1d</a>()</code> <br><code><a href="layer_conv_lstm_2d.html">layer_conv_lstm_2d</a>()</code> <br><code><a href="layer_conv_lstm_3d.html">layer_conv_lstm_3d</a>()</code> <br><code><a href="layer_cropping_1d.html">layer_cropping_1d</a>()</code> <br><code><a href="layer_cropping_2d.html">layer_cropping_2d</a>()</code> <br><code><a href="layer_cropping_3d.html">layer_cropping_3d</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_dense.html">layer_dense</a>()</code> <br><code><a href="layer_depthwise_conv_1d.html">layer_depthwise_conv_1d</a>()</code> <br><code><a href="layer_depthwise_conv_2d.html">layer_depthwise_conv_2d</a>()</code> <br><code><a href="layer_discretization.html">layer_discretization</a>()</code> <br><code><a href="layer_dot.html">layer_dot</a>()</code> <br><code><a href="layer_dropout.html">layer_dropout</a>()</code> <br><code><a href="layer_einsum_dense.html">layer_einsum_dense</a>()</code> <br><code><a href="layer_embedding.html">layer_embedding</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_feature_space.html">layer_feature_space</a>()</code> <br><code><a href="layer_flatten.html">layer_flatten</a>()</code> <br><code><a href="layer_flax_module_wrapper.html">layer_flax_module_wrapper</a>()</code> <br><code><a href="layer_gaussian_dropout.html">layer_gaussian_dropout</a>()</code> <br><code><a href="layer_gaussian_noise.html">layer_gaussian_noise</a>()</code> <br><code><a href="layer_global_average_pooling_1d.html">layer_global_average_pooling_1d</a>()</code> <br><code><a href="layer_global_average_pooling_2d.html">layer_global_average_pooling_2d</a>()</code> <br><code><a href="layer_global_average_pooling_3d.html">layer_global_average_pooling_3d</a>()</code> <br><code><a href="layer_global_max_pooling_1d.html">layer_global_max_pooling_1d</a>()</code> <br><code><a href="layer_global_max_pooling_2d.html">layer_global_max_pooling_2d</a>()</code> <br><code><a href="layer_global_max_pooling_3d.html">layer_global_max_pooling_3d</a>()</code> <br><code><a href="layer_group_normalization.html">layer_group_normalization</a>()</code> <br><code><a href="layer_group_query_attention.html">layer_group_query_attention</a>()</code> <br><code><a href="layer_gru.html">layer_gru</a>()</code> <br><code><a href="layer_hashed_crossing.html">layer_hashed_crossing</a>()</code> <br><code><a href="layer_hashing.html">layer_hashing</a>()</code> <br><code><a href="layer_identity.html">layer_identity</a>()</code> <br><code><a href="layer_integer_lookup.html">layer_integer_lookup</a>()</code> <br><code><a href="layer_jax_model_wrapper.html">layer_jax_model_wrapper</a>()</code> <br><code><a href="layer_lambda.html">layer_lambda</a>()</code> <br><code><a href="layer_layer_normalization.html">layer_layer_normalization</a>()</code> <br><code><a href="layer_lstm.html">layer_lstm</a>()</code> <br><code><a href="layer_masking.html">layer_masking</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_max_pooling_1d.html">layer_max_pooling_1d</a>()</code> <br><code><a href="layer_max_pooling_2d.html">layer_max_pooling_2d</a>()</code> <br><code><a href="layer_max_pooling_3d.html">layer_max_pooling_3d</a>()</code> <br><code><a href="layer_maximum.html">layer_maximum</a>()</code> <br><code><a href="layer_mel_spectrogram.html">layer_mel_spectrogram</a>()</code> <br><code><a href="layer_minimum.html">layer_minimum</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_multi_head_attention.html">layer_multi_head_attention</a>()</code> <br><code><a href="layer_multiply.html">layer_multiply</a>()</code> <br><code><a href="layer_normalization.html">layer_normalization</a>()</code> <br><code><a href="layer_permute.html">layer_permute</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_brightness.html">layer_random_brightness</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_contrast.html">layer_random_contrast</a>()</code> <br><code><a href="layer_random_crop.html">layer_random_crop</a>()</code> <br><code><a href="layer_random_elastic_transform.html">layer_random_elastic_transform</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_flip.html">layer_random_flip</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_rotation.html">layer_random_rotation</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_random_translation.html">layer_random_translation</a>()</code> <br><code><a href="layer_random_zoom.html">layer_random_zoom</a>()</code> <br><code><a href="layer_repeat_vector.html">layer_repeat_vector</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_reshape.html">layer_reshape</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_rms_normalization.html">layer_rms_normalization</a>()</code> <br><code><a href="layer_rnn.html">layer_rnn</a>()</code> <br><code><a href="layer_separable_conv_1d.html">layer_separable_conv_1d</a>()</code> <br><code><a href="layer_separable_conv_2d.html">layer_separable_conv_2d</a>()</code> <br><code><a href="layer_simple_rnn.html">layer_simple_rnn</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br><code><a href="layer_spatial_dropout_1d.html">layer_spatial_dropout_1d</a>()</code> <br><code><a href="layer_spatial_dropout_2d.html">layer_spatial_dropout_2d</a>()</code> <br><code><a href="layer_spatial_dropout_3d.html">layer_spatial_dropout_3d</a>()</code> <br><code><a href="layer_spectral_normalization.html">layer_spectral_normalization</a>()</code> <br><code><a href="layer_stft_spectrogram.html">layer_stft_spectrogram</a>()</code> <br><code><a href="layer_string_lookup.html">layer_string_lookup</a>()</code> <br><code><a href="layer_subtract.html">layer_subtract</a>()</code> <br><code><a href="layer_text_vectorization.html">layer_text_vectorization</a>()</code> <br><code><a href="layer_tfsm.html">layer_tfsm</a>()</code> <br><code><a href="layer_time_distributed.html">layer_time_distributed</a>()</code> <br><code><a href="layer_torch_module_wrapper.html">layer_torch_module_wrapper</a>()</code> <br><code><a href="layer_unit_normalization.html">layer_unit_normalization</a>()</code> <br><code><a href="layer_upsampling_1d.html">layer_upsampling_1d</a>()</code> <br><code><a href="layer_upsampling_2d.html">layer_upsampling_2d</a>()</code> <br><code><a href="layer_upsampling_3d.html">layer_upsampling_3d</a>()</code> <br><code><a href="layer_zero_padding_1d.html">layer_zero_padding_1d</a>()</code> <br><code><a href="layer_zero_padding_2d.html">layer_zero_padding_2d</a>()</code> <br><code><a href="layer_zero_padding_3d.html">layer_zero_padding_3d</a>()</code> <br><code><a href="rnn_cell_gru.html">rnn_cell_gru</a>()</code> <br><code><a href="rnn_cell_lstm.html">rnn_cell_lstm</a>()</code> <br><code><a href="rnn_cell_simple.html">rnn_cell_simple</a>()</code> <br><code><a href="rnn_cells_stack.html">rnn_cells_stack</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -235,7 +235,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/layer_average_pooling_3d.html b/docs/dev/reference/layer_average_pooling_3d.html
index cb6a40ea4..89b7929b6 100644
--- a/docs/dev/reference/layer_average_pooling_3d.html
+++ b/docs/dev/reference/layer_average_pooling_3d.html
@@ -14,7 +14,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -171,7 +171,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
     <div class="dont-index">
 <ul><li><p><a href="https://keras.io/api/layers/pooling_layers/average_pooling3d#averagepooling3d-class" class="external-link">https://keras.io/api/layers/pooling_layers/average_pooling3d#averagepooling3d-class</a></p></li>
 </ul><p>Other pooling layers: <br><code><a href="layer_average_pooling_1d.html">layer_average_pooling_1d</a>()</code> <br><code><a href="layer_average_pooling_2d.html">layer_average_pooling_2d</a>()</code> <br><code><a href="layer_global_average_pooling_1d.html">layer_global_average_pooling_1d</a>()</code> <br><code><a href="layer_global_average_pooling_2d.html">layer_global_average_pooling_2d</a>()</code> <br><code><a href="layer_global_average_pooling_3d.html">layer_global_average_pooling_3d</a>()</code> <br><code><a href="layer_global_max_pooling_1d.html">layer_global_max_pooling_1d</a>()</code> <br><code><a href="layer_global_max_pooling_2d.html">layer_global_max_pooling_2d</a>()</code> <br><code><a href="layer_global_max_pooling_3d.html">layer_global_max_pooling_3d</a>()</code> <br><code><a href="layer_max_pooling_1d.html">layer_max_pooling_1d</a>()</code> <br><code><a href="layer_max_pooling_2d.html">layer_max_pooling_2d</a>()</code> <br><code><a href="layer_max_pooling_3d.html">layer_max_pooling_3d</a>()</code> <br></p>
-<p>Other layers: <br><code><a href="Layer.html">Layer</a>()</code> <br><code><a href="layer_activation.html">layer_activation</a>()</code> <br><code><a href="layer_activation_elu.html">layer_activation_elu</a>()</code> <br><code><a href="layer_activation_leaky_relu.html">layer_activation_leaky_relu</a>()</code> <br><code><a href="layer_activation_parametric_relu.html">layer_activation_parametric_relu</a>()</code> <br><code><a href="layer_activation_relu.html">layer_activation_relu</a>()</code> <br><code><a href="layer_activation_softmax.html">layer_activation_softmax</a>()</code> <br><code><a href="layer_activity_regularization.html">layer_activity_regularization</a>()</code> <br><code><a href="layer_add.html">layer_add</a>()</code> <br><code><a href="layer_additive_attention.html">layer_additive_attention</a>()</code> <br><code><a href="layer_alpha_dropout.html">layer_alpha_dropout</a>()</code> <br><code><a href="layer_attention.html">layer_attention</a>()</code> <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_average.html">layer_average</a>()</code> <br><code><a href="layer_average_pooling_1d.html">layer_average_pooling_1d</a>()</code> <br><code><a href="layer_average_pooling_2d.html">layer_average_pooling_2d</a>()</code> <br><code><a href="layer_batch_normalization.html">layer_batch_normalization</a>()</code> <br><code><a href="layer_bidirectional.html">layer_bidirectional</a>()</code> <br><code><a href="layer_category_encoding.html">layer_category_encoding</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_concatenate.html">layer_concatenate</a>()</code> <br><code><a href="layer_conv_1d.html">layer_conv_1d</a>()</code> <br><code><a href="layer_conv_1d_transpose.html">layer_conv_1d_transpose</a>()</code> <br><code><a href="layer_conv_2d.html">layer_conv_2d</a>()</code> <br><code><a href="layer_conv_2d_transpose.html">layer_conv_2d_transpose</a>()</code> <br><code><a href="layer_conv_3d.html">layer_conv_3d</a>()</code> <br><code><a href="layer_conv_3d_transpose.html">layer_conv_3d_transpose</a>()</code> <br><code><a href="layer_conv_lstm_1d.html">layer_conv_lstm_1d</a>()</code> <br><code><a href="layer_conv_lstm_2d.html">layer_conv_lstm_2d</a>()</code> <br><code><a href="layer_conv_lstm_3d.html">layer_conv_lstm_3d</a>()</code> <br><code><a href="layer_cropping_1d.html">layer_cropping_1d</a>()</code> <br><code><a href="layer_cropping_2d.html">layer_cropping_2d</a>()</code> <br><code><a href="layer_cropping_3d.html">layer_cropping_3d</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_dense.html">layer_dense</a>()</code> <br><code><a href="layer_depthwise_conv_1d.html">layer_depthwise_conv_1d</a>()</code> <br><code><a href="layer_depthwise_conv_2d.html">layer_depthwise_conv_2d</a>()</code> <br><code><a href="layer_discretization.html">layer_discretization</a>()</code> <br><code><a href="layer_dot.html">layer_dot</a>()</code> <br><code><a href="layer_dropout.html">layer_dropout</a>()</code> <br><code><a href="layer_einsum_dense.html">layer_einsum_dense</a>()</code> <br><code><a href="layer_embedding.html">layer_embedding</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_feature_space.html">layer_feature_space</a>()</code> <br><code><a href="layer_flatten.html">layer_flatten</a>()</code> <br><code><a href="layer_flax_module_wrapper.html">layer_flax_module_wrapper</a>()</code> <br><code><a href="layer_gaussian_dropout.html">layer_gaussian_dropout</a>()</code> <br><code><a href="layer_gaussian_noise.html">layer_gaussian_noise</a>()</code> <br><code><a href="layer_global_average_pooling_1d.html">layer_global_average_pooling_1d</a>()</code> <br><code><a href="layer_global_average_pooling_2d.html">layer_global_average_pooling_2d</a>()</code> <br><code><a href="layer_global_average_pooling_3d.html">layer_global_average_pooling_3d</a>()</code> <br><code><a href="layer_global_max_pooling_1d.html">layer_global_max_pooling_1d</a>()</code> <br><code><a href="layer_global_max_pooling_2d.html">layer_global_max_pooling_2d</a>()</code> <br><code><a href="layer_global_max_pooling_3d.html">layer_global_max_pooling_3d</a>()</code> <br><code><a href="layer_group_normalization.html">layer_group_normalization</a>()</code> <br><code><a href="layer_group_query_attention.html">layer_group_query_attention</a>()</code> <br><code><a href="layer_gru.html">layer_gru</a>()</code> <br><code><a href="layer_hashed_crossing.html">layer_hashed_crossing</a>()</code> <br><code><a href="layer_hashing.html">layer_hashing</a>()</code> <br><code><a href="layer_identity.html">layer_identity</a>()</code> <br><code><a href="layer_integer_lookup.html">layer_integer_lookup</a>()</code> <br><code><a href="layer_jax_model_wrapper.html">layer_jax_model_wrapper</a>()</code> <br><code><a href="layer_lambda.html">layer_lambda</a>()</code> <br><code><a href="layer_layer_normalization.html">layer_layer_normalization</a>()</code> <br><code><a href="layer_lstm.html">layer_lstm</a>()</code> <br><code><a href="layer_masking.html">layer_masking</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_max_pooling_1d.html">layer_max_pooling_1d</a>()</code> <br><code><a href="layer_max_pooling_2d.html">layer_max_pooling_2d</a>()</code> <br><code><a href="layer_max_pooling_3d.html">layer_max_pooling_3d</a>()</code> <br><code><a href="layer_maximum.html">layer_maximum</a>()</code> <br><code><a href="layer_mel_spectrogram.html">layer_mel_spectrogram</a>()</code> <br><code><a href="layer_minimum.html">layer_minimum</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_multi_head_attention.html">layer_multi_head_attention</a>()</code> <br><code><a href="layer_multiply.html">layer_multiply</a>()</code> <br><code><a href="layer_normalization.html">layer_normalization</a>()</code> <br><code><a href="layer_permute.html">layer_permute</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_brightness.html">layer_random_brightness</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_contrast.html">layer_random_contrast</a>()</code> <br><code><a href="layer_random_crop.html">layer_random_crop</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_flip.html">layer_random_flip</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_rotation.html">layer_random_rotation</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_random_translation.html">layer_random_translation</a>()</code> <br><code><a href="layer_random_zoom.html">layer_random_zoom</a>()</code> <br><code><a href="layer_repeat_vector.html">layer_repeat_vector</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_reshape.html">layer_reshape</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_rms_normalization.html">layer_rms_normalization</a>()</code> <br><code><a href="layer_rnn.html">layer_rnn</a>()</code> <br><code><a href="layer_separable_conv_1d.html">layer_separable_conv_1d</a>()</code> <br><code><a href="layer_separable_conv_2d.html">layer_separable_conv_2d</a>()</code> <br><code><a href="layer_simple_rnn.html">layer_simple_rnn</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br><code><a href="layer_spatial_dropout_1d.html">layer_spatial_dropout_1d</a>()</code> <br><code><a href="layer_spatial_dropout_2d.html">layer_spatial_dropout_2d</a>()</code> <br><code><a href="layer_spatial_dropout_3d.html">layer_spatial_dropout_3d</a>()</code> <br><code><a href="layer_spectral_normalization.html">layer_spectral_normalization</a>()</code> <br><code><a href="layer_stft_spectrogram.html">layer_stft_spectrogram</a>()</code> <br><code><a href="layer_string_lookup.html">layer_string_lookup</a>()</code> <br><code><a href="layer_subtract.html">layer_subtract</a>()</code> <br><code><a href="layer_text_vectorization.html">layer_text_vectorization</a>()</code> <br><code><a href="layer_tfsm.html">layer_tfsm</a>()</code> <br><code><a href="layer_time_distributed.html">layer_time_distributed</a>()</code> <br><code><a href="layer_torch_module_wrapper.html">layer_torch_module_wrapper</a>()</code> <br><code><a href="layer_unit_normalization.html">layer_unit_normalization</a>()</code> <br><code><a href="layer_upsampling_1d.html">layer_upsampling_1d</a>()</code> <br><code><a href="layer_upsampling_2d.html">layer_upsampling_2d</a>()</code> <br><code><a href="layer_upsampling_3d.html">layer_upsampling_3d</a>()</code> <br><code><a href="layer_zero_padding_1d.html">layer_zero_padding_1d</a>()</code> <br><code><a href="layer_zero_padding_2d.html">layer_zero_padding_2d</a>()</code> <br><code><a href="layer_zero_padding_3d.html">layer_zero_padding_3d</a>()</code> <br><code><a href="rnn_cell_gru.html">rnn_cell_gru</a>()</code> <br><code><a href="rnn_cell_lstm.html">rnn_cell_lstm</a>()</code> <br><code><a href="rnn_cell_simple.html">rnn_cell_simple</a>()</code> <br><code><a href="rnn_cells_stack.html">rnn_cells_stack</a>()</code> <br></p></div>
+<p>Other layers: <br><code><a href="Layer.html">Layer</a>()</code> <br><code><a href="layer_activation.html">layer_activation</a>()</code> <br><code><a href="layer_activation_elu.html">layer_activation_elu</a>()</code> <br><code><a href="layer_activation_leaky_relu.html">layer_activation_leaky_relu</a>()</code> <br><code><a href="layer_activation_parametric_relu.html">layer_activation_parametric_relu</a>()</code> <br><code><a href="layer_activation_relu.html">layer_activation_relu</a>()</code> <br><code><a href="layer_activation_softmax.html">layer_activation_softmax</a>()</code> <br><code><a href="layer_activity_regularization.html">layer_activity_regularization</a>()</code> <br><code><a href="layer_add.html">layer_add</a>()</code> <br><code><a href="layer_additive_attention.html">layer_additive_attention</a>()</code> <br><code><a href="layer_alpha_dropout.html">layer_alpha_dropout</a>()</code> <br><code><a href="layer_attention.html">layer_attention</a>()</code> <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_average.html">layer_average</a>()</code> <br><code><a href="layer_average_pooling_1d.html">layer_average_pooling_1d</a>()</code> <br><code><a href="layer_average_pooling_2d.html">layer_average_pooling_2d</a>()</code> <br><code><a href="layer_batch_normalization.html">layer_batch_normalization</a>()</code> <br><code><a href="layer_bidirectional.html">layer_bidirectional</a>()</code> <br><code><a href="layer_category_encoding.html">layer_category_encoding</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_concatenate.html">layer_concatenate</a>()</code> <br><code><a href="layer_conv_1d.html">layer_conv_1d</a>()</code> <br><code><a href="layer_conv_1d_transpose.html">layer_conv_1d_transpose</a>()</code> <br><code><a href="layer_conv_2d.html">layer_conv_2d</a>()</code> <br><code><a href="layer_conv_2d_transpose.html">layer_conv_2d_transpose</a>()</code> <br><code><a href="layer_conv_3d.html">layer_conv_3d</a>()</code> <br><code><a href="layer_conv_3d_transpose.html">layer_conv_3d_transpose</a>()</code> <br><code><a href="layer_conv_lstm_1d.html">layer_conv_lstm_1d</a>()</code> <br><code><a href="layer_conv_lstm_2d.html">layer_conv_lstm_2d</a>()</code> <br><code><a href="layer_conv_lstm_3d.html">layer_conv_lstm_3d</a>()</code> <br><code><a href="layer_cropping_1d.html">layer_cropping_1d</a>()</code> <br><code><a href="layer_cropping_2d.html">layer_cropping_2d</a>()</code> <br><code><a href="layer_cropping_3d.html">layer_cropping_3d</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_dense.html">layer_dense</a>()</code> <br><code><a href="layer_depthwise_conv_1d.html">layer_depthwise_conv_1d</a>()</code> <br><code><a href="layer_depthwise_conv_2d.html">layer_depthwise_conv_2d</a>()</code> <br><code><a href="layer_discretization.html">layer_discretization</a>()</code> <br><code><a href="layer_dot.html">layer_dot</a>()</code> <br><code><a href="layer_dropout.html">layer_dropout</a>()</code> <br><code><a href="layer_einsum_dense.html">layer_einsum_dense</a>()</code> <br><code><a href="layer_embedding.html">layer_embedding</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_feature_space.html">layer_feature_space</a>()</code> <br><code><a href="layer_flatten.html">layer_flatten</a>()</code> <br><code><a href="layer_flax_module_wrapper.html">layer_flax_module_wrapper</a>()</code> <br><code><a href="layer_gaussian_dropout.html">layer_gaussian_dropout</a>()</code> <br><code><a href="layer_gaussian_noise.html">layer_gaussian_noise</a>()</code> <br><code><a href="layer_global_average_pooling_1d.html">layer_global_average_pooling_1d</a>()</code> <br><code><a href="layer_global_average_pooling_2d.html">layer_global_average_pooling_2d</a>()</code> <br><code><a href="layer_global_average_pooling_3d.html">layer_global_average_pooling_3d</a>()</code> <br><code><a href="layer_global_max_pooling_1d.html">layer_global_max_pooling_1d</a>()</code> <br><code><a href="layer_global_max_pooling_2d.html">layer_global_max_pooling_2d</a>()</code> <br><code><a href="layer_global_max_pooling_3d.html">layer_global_max_pooling_3d</a>()</code> <br><code><a href="layer_group_normalization.html">layer_group_normalization</a>()</code> <br><code><a href="layer_group_query_attention.html">layer_group_query_attention</a>()</code> <br><code><a href="layer_gru.html">layer_gru</a>()</code> <br><code><a href="layer_hashed_crossing.html">layer_hashed_crossing</a>()</code> <br><code><a href="layer_hashing.html">layer_hashing</a>()</code> <br><code><a href="layer_identity.html">layer_identity</a>()</code> <br><code><a href="layer_integer_lookup.html">layer_integer_lookup</a>()</code> <br><code><a href="layer_jax_model_wrapper.html">layer_jax_model_wrapper</a>()</code> <br><code><a href="layer_lambda.html">layer_lambda</a>()</code> <br><code><a href="layer_layer_normalization.html">layer_layer_normalization</a>()</code> <br><code><a href="layer_lstm.html">layer_lstm</a>()</code> <br><code><a href="layer_masking.html">layer_masking</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_max_pooling_1d.html">layer_max_pooling_1d</a>()</code> <br><code><a href="layer_max_pooling_2d.html">layer_max_pooling_2d</a>()</code> <br><code><a href="layer_max_pooling_3d.html">layer_max_pooling_3d</a>()</code> <br><code><a href="layer_maximum.html">layer_maximum</a>()</code> <br><code><a href="layer_mel_spectrogram.html">layer_mel_spectrogram</a>()</code> <br><code><a href="layer_minimum.html">layer_minimum</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_multi_head_attention.html">layer_multi_head_attention</a>()</code> <br><code><a href="layer_multiply.html">layer_multiply</a>()</code> <br><code><a href="layer_normalization.html">layer_normalization</a>()</code> <br><code><a href="layer_permute.html">layer_permute</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_brightness.html">layer_random_brightness</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_contrast.html">layer_random_contrast</a>()</code> <br><code><a href="layer_random_crop.html">layer_random_crop</a>()</code> <br><code><a href="layer_random_elastic_transform.html">layer_random_elastic_transform</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_flip.html">layer_random_flip</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_rotation.html">layer_random_rotation</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_random_translation.html">layer_random_translation</a>()</code> <br><code><a href="layer_random_zoom.html">layer_random_zoom</a>()</code> <br><code><a href="layer_repeat_vector.html">layer_repeat_vector</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_reshape.html">layer_reshape</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_rms_normalization.html">layer_rms_normalization</a>()</code> <br><code><a href="layer_rnn.html">layer_rnn</a>()</code> <br><code><a href="layer_separable_conv_1d.html">layer_separable_conv_1d</a>()</code> <br><code><a href="layer_separable_conv_2d.html">layer_separable_conv_2d</a>()</code> <br><code><a href="layer_simple_rnn.html">layer_simple_rnn</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br><code><a href="layer_spatial_dropout_1d.html">layer_spatial_dropout_1d</a>()</code> <br><code><a href="layer_spatial_dropout_2d.html">layer_spatial_dropout_2d</a>()</code> <br><code><a href="layer_spatial_dropout_3d.html">layer_spatial_dropout_3d</a>()</code> <br><code><a href="layer_spectral_normalization.html">layer_spectral_normalization</a>()</code> <br><code><a href="layer_stft_spectrogram.html">layer_stft_spectrogram</a>()</code> <br><code><a href="layer_string_lookup.html">layer_string_lookup</a>()</code> <br><code><a href="layer_subtract.html">layer_subtract</a>()</code> <br><code><a href="layer_text_vectorization.html">layer_text_vectorization</a>()</code> <br><code><a href="layer_tfsm.html">layer_tfsm</a>()</code> <br><code><a href="layer_time_distributed.html">layer_time_distributed</a>()</code> <br><code><a href="layer_torch_module_wrapper.html">layer_torch_module_wrapper</a>()</code> <br><code><a href="layer_unit_normalization.html">layer_unit_normalization</a>()</code> <br><code><a href="layer_upsampling_1d.html">layer_upsampling_1d</a>()</code> <br><code><a href="layer_upsampling_2d.html">layer_upsampling_2d</a>()</code> <br><code><a href="layer_upsampling_3d.html">layer_upsampling_3d</a>()</code> <br><code><a href="layer_zero_padding_1d.html">layer_zero_padding_1d</a>()</code> <br><code><a href="layer_zero_padding_2d.html">layer_zero_padding_2d</a>()</code> <br><code><a href="layer_zero_padding_3d.html">layer_zero_padding_3d</a>()</code> <br><code><a href="rnn_cell_gru.html">rnn_cell_gru</a>()</code> <br><code><a href="rnn_cell_lstm.html">rnn_cell_lstm</a>()</code> <br><code><a href="rnn_cell_simple.html">rnn_cell_simple</a>()</code> <br><code><a href="rnn_cells_stack.html">rnn_cells_stack</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -183,7 +183,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/layer_batch_normalization.html b/docs/dev/reference/layer_batch_normalization.html
index 258dc9d8c..df078df87 100644
--- a/docs/dev/reference/layer_batch_normalization.html
+++ b/docs/dev/reference/layer_batch_normalization.html
@@ -114,7 +114,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -343,7 +343,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
     <div class="dont-index">
 <ul><li><p><a href="https://keras.io/api/layers/normalization_layers/batch_normalization#batchnormalization-class" class="external-link">https://keras.io/api/layers/normalization_layers/batch_normalization#batchnormalization-class</a></p></li>
 </ul><p>Other normalization layers: <br><code><a href="layer_group_normalization.html">layer_group_normalization</a>()</code> <br><code><a href="layer_layer_normalization.html">layer_layer_normalization</a>()</code> <br><code><a href="layer_rms_normalization.html">layer_rms_normalization</a>()</code> <br><code><a href="layer_spectral_normalization.html">layer_spectral_normalization</a>()</code> <br><code><a href="layer_unit_normalization.html">layer_unit_normalization</a>()</code> <br></p>
-<p>Other layers: <br><code><a href="Layer.html">Layer</a>()</code> <br><code><a href="layer_activation.html">layer_activation</a>()</code> <br><code><a href="layer_activation_elu.html">layer_activation_elu</a>()</code> <br><code><a href="layer_activation_leaky_relu.html">layer_activation_leaky_relu</a>()</code> <br><code><a href="layer_activation_parametric_relu.html">layer_activation_parametric_relu</a>()</code> <br><code><a href="layer_activation_relu.html">layer_activation_relu</a>()</code> <br><code><a href="layer_activation_softmax.html">layer_activation_softmax</a>()</code> <br><code><a href="layer_activity_regularization.html">layer_activity_regularization</a>()</code> <br><code><a href="layer_add.html">layer_add</a>()</code> <br><code><a href="layer_additive_attention.html">layer_additive_attention</a>()</code> <br><code><a href="layer_alpha_dropout.html">layer_alpha_dropout</a>()</code> <br><code><a href="layer_attention.html">layer_attention</a>()</code> <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_average.html">layer_average</a>()</code> <br><code><a href="layer_average_pooling_1d.html">layer_average_pooling_1d</a>()</code> <br><code><a href="layer_average_pooling_2d.html">layer_average_pooling_2d</a>()</code> <br><code><a href="layer_average_pooling_3d.html">layer_average_pooling_3d</a>()</code> <br><code><a href="layer_bidirectional.html">layer_bidirectional</a>()</code> <br><code><a href="layer_category_encoding.html">layer_category_encoding</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_concatenate.html">layer_concatenate</a>()</code> <br><code><a href="layer_conv_1d.html">layer_conv_1d</a>()</code> <br><code><a href="layer_conv_1d_transpose.html">layer_conv_1d_transpose</a>()</code> <br><code><a href="layer_conv_2d.html">layer_conv_2d</a>()</code> <br><code><a href="layer_conv_2d_transpose.html">layer_conv_2d_transpose</a>()</code> <br><code><a href="layer_conv_3d.html">layer_conv_3d</a>()</code> <br><code><a href="layer_conv_3d_transpose.html">layer_conv_3d_transpose</a>()</code> <br><code><a href="layer_conv_lstm_1d.html">layer_conv_lstm_1d</a>()</code> <br><code><a href="layer_conv_lstm_2d.html">layer_conv_lstm_2d</a>()</code> <br><code><a href="layer_conv_lstm_3d.html">layer_conv_lstm_3d</a>()</code> <br><code><a href="layer_cropping_1d.html">layer_cropping_1d</a>()</code> <br><code><a href="layer_cropping_2d.html">layer_cropping_2d</a>()</code> <br><code><a href="layer_cropping_3d.html">layer_cropping_3d</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_dense.html">layer_dense</a>()</code> <br><code><a href="layer_depthwise_conv_1d.html">layer_depthwise_conv_1d</a>()</code> <br><code><a href="layer_depthwise_conv_2d.html">layer_depthwise_conv_2d</a>()</code> <br><code><a href="layer_discretization.html">layer_discretization</a>()</code> <br><code><a href="layer_dot.html">layer_dot</a>()</code> <br><code><a href="layer_dropout.html">layer_dropout</a>()</code> <br><code><a href="layer_einsum_dense.html">layer_einsum_dense</a>()</code> <br><code><a href="layer_embedding.html">layer_embedding</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_feature_space.html">layer_feature_space</a>()</code> <br><code><a href="layer_flatten.html">layer_flatten</a>()</code> <br><code><a href="layer_flax_module_wrapper.html">layer_flax_module_wrapper</a>()</code> <br><code><a href="layer_gaussian_dropout.html">layer_gaussian_dropout</a>()</code> <br><code><a href="layer_gaussian_noise.html">layer_gaussian_noise</a>()</code> <br><code><a href="layer_global_average_pooling_1d.html">layer_global_average_pooling_1d</a>()</code> <br><code><a href="layer_global_average_pooling_2d.html">layer_global_average_pooling_2d</a>()</code> <br><code><a href="layer_global_average_pooling_3d.html">layer_global_average_pooling_3d</a>()</code> <br><code><a href="layer_global_max_pooling_1d.html">layer_global_max_pooling_1d</a>()</code> <br><code><a href="layer_global_max_pooling_2d.html">layer_global_max_pooling_2d</a>()</code> <br><code><a href="layer_global_max_pooling_3d.html">layer_global_max_pooling_3d</a>()</code> <br><code><a href="layer_group_normalization.html">layer_group_normalization</a>()</code> <br><code><a href="layer_group_query_attention.html">layer_group_query_attention</a>()</code> <br><code><a href="layer_gru.html">layer_gru</a>()</code> <br><code><a href="layer_hashed_crossing.html">layer_hashed_crossing</a>()</code> <br><code><a href="layer_hashing.html">layer_hashing</a>()</code> <br><code><a href="layer_identity.html">layer_identity</a>()</code> <br><code><a href="layer_integer_lookup.html">layer_integer_lookup</a>()</code> <br><code><a href="layer_jax_model_wrapper.html">layer_jax_model_wrapper</a>()</code> <br><code><a href="layer_lambda.html">layer_lambda</a>()</code> <br><code><a href="layer_layer_normalization.html">layer_layer_normalization</a>()</code> <br><code><a href="layer_lstm.html">layer_lstm</a>()</code> <br><code><a href="layer_masking.html">layer_masking</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_max_pooling_1d.html">layer_max_pooling_1d</a>()</code> <br><code><a href="layer_max_pooling_2d.html">layer_max_pooling_2d</a>()</code> <br><code><a href="layer_max_pooling_3d.html">layer_max_pooling_3d</a>()</code> <br><code><a href="layer_maximum.html">layer_maximum</a>()</code> <br><code><a href="layer_mel_spectrogram.html">layer_mel_spectrogram</a>()</code> <br><code><a href="layer_minimum.html">layer_minimum</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_multi_head_attention.html">layer_multi_head_attention</a>()</code> <br><code><a href="layer_multiply.html">layer_multiply</a>()</code> <br><code><a href="layer_normalization.html">layer_normalization</a>()</code> <br><code><a href="layer_permute.html">layer_permute</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_brightness.html">layer_random_brightness</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_contrast.html">layer_random_contrast</a>()</code> <br><code><a href="layer_random_crop.html">layer_random_crop</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_flip.html">layer_random_flip</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_rotation.html">layer_random_rotation</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_random_translation.html">layer_random_translation</a>()</code> <br><code><a href="layer_random_zoom.html">layer_random_zoom</a>()</code> <br><code><a href="layer_repeat_vector.html">layer_repeat_vector</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_reshape.html">layer_reshape</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_rms_normalization.html">layer_rms_normalization</a>()</code> <br><code><a href="layer_rnn.html">layer_rnn</a>()</code> <br><code><a href="layer_separable_conv_1d.html">layer_separable_conv_1d</a>()</code> <br><code><a href="layer_separable_conv_2d.html">layer_separable_conv_2d</a>()</code> <br><code><a href="layer_simple_rnn.html">layer_simple_rnn</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br><code><a href="layer_spatial_dropout_1d.html">layer_spatial_dropout_1d</a>()</code> <br><code><a href="layer_spatial_dropout_2d.html">layer_spatial_dropout_2d</a>()</code> <br><code><a href="layer_spatial_dropout_3d.html">layer_spatial_dropout_3d</a>()</code> <br><code><a href="layer_spectral_normalization.html">layer_spectral_normalization</a>()</code> <br><code><a href="layer_stft_spectrogram.html">layer_stft_spectrogram</a>()</code> <br><code><a href="layer_string_lookup.html">layer_string_lookup</a>()</code> <br><code><a href="layer_subtract.html">layer_subtract</a>()</code> <br><code><a href="layer_text_vectorization.html">layer_text_vectorization</a>()</code> <br><code><a href="layer_tfsm.html">layer_tfsm</a>()</code> <br><code><a href="layer_time_distributed.html">layer_time_distributed</a>()</code> <br><code><a href="layer_torch_module_wrapper.html">layer_torch_module_wrapper</a>()</code> <br><code><a href="layer_unit_normalization.html">layer_unit_normalization</a>()</code> <br><code><a href="layer_upsampling_1d.html">layer_upsampling_1d</a>()</code> <br><code><a href="layer_upsampling_2d.html">layer_upsampling_2d</a>()</code> <br><code><a href="layer_upsampling_3d.html">layer_upsampling_3d</a>()</code> <br><code><a href="layer_zero_padding_1d.html">layer_zero_padding_1d</a>()</code> <br><code><a href="layer_zero_padding_2d.html">layer_zero_padding_2d</a>()</code> <br><code><a href="layer_zero_padding_3d.html">layer_zero_padding_3d</a>()</code> <br><code><a href="rnn_cell_gru.html">rnn_cell_gru</a>()</code> <br><code><a href="rnn_cell_lstm.html">rnn_cell_lstm</a>()</code> <br><code><a href="rnn_cell_simple.html">rnn_cell_simple</a>()</code> <br><code><a href="rnn_cells_stack.html">rnn_cells_stack</a>()</code> <br></p></div>
+<p>Other layers: <br><code><a href="Layer.html">Layer</a>()</code> <br><code><a href="layer_activation.html">layer_activation</a>()</code> <br><code><a href="layer_activation_elu.html">layer_activation_elu</a>()</code> <br><code><a href="layer_activation_leaky_relu.html">layer_activation_leaky_relu</a>()</code> <br><code><a href="layer_activation_parametric_relu.html">layer_activation_parametric_relu</a>()</code> <br><code><a href="layer_activation_relu.html">layer_activation_relu</a>()</code> <br><code><a href="layer_activation_softmax.html">layer_activation_softmax</a>()</code> <br><code><a href="layer_activity_regularization.html">layer_activity_regularization</a>()</code> <br><code><a href="layer_add.html">layer_add</a>()</code> <br><code><a href="layer_additive_attention.html">layer_additive_attention</a>()</code> <br><code><a href="layer_alpha_dropout.html">layer_alpha_dropout</a>()</code> <br><code><a href="layer_attention.html">layer_attention</a>()</code> <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_average.html">layer_average</a>()</code> <br><code><a href="layer_average_pooling_1d.html">layer_average_pooling_1d</a>()</code> <br><code><a href="layer_average_pooling_2d.html">layer_average_pooling_2d</a>()</code> <br><code><a href="layer_average_pooling_3d.html">layer_average_pooling_3d</a>()</code> <br><code><a href="layer_bidirectional.html">layer_bidirectional</a>()</code> <br><code><a href="layer_category_encoding.html">layer_category_encoding</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_concatenate.html">layer_concatenate</a>()</code> <br><code><a href="layer_conv_1d.html">layer_conv_1d</a>()</code> <br><code><a href="layer_conv_1d_transpose.html">layer_conv_1d_transpose</a>()</code> <br><code><a href="layer_conv_2d.html">layer_conv_2d</a>()</code> <br><code><a href="layer_conv_2d_transpose.html">layer_conv_2d_transpose</a>()</code> <br><code><a href="layer_conv_3d.html">layer_conv_3d</a>()</code> <br><code><a href="layer_conv_3d_transpose.html">layer_conv_3d_transpose</a>()</code> <br><code><a href="layer_conv_lstm_1d.html">layer_conv_lstm_1d</a>()</code> <br><code><a href="layer_conv_lstm_2d.html">layer_conv_lstm_2d</a>()</code> <br><code><a href="layer_conv_lstm_3d.html">layer_conv_lstm_3d</a>()</code> <br><code><a href="layer_cropping_1d.html">layer_cropping_1d</a>()</code> <br><code><a href="layer_cropping_2d.html">layer_cropping_2d</a>()</code> <br><code><a href="layer_cropping_3d.html">layer_cropping_3d</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_dense.html">layer_dense</a>()</code> <br><code><a href="layer_depthwise_conv_1d.html">layer_depthwise_conv_1d</a>()</code> <br><code><a href="layer_depthwise_conv_2d.html">layer_depthwise_conv_2d</a>()</code> <br><code><a href="layer_discretization.html">layer_discretization</a>()</code> <br><code><a href="layer_dot.html">layer_dot</a>()</code> <br><code><a href="layer_dropout.html">layer_dropout</a>()</code> <br><code><a href="layer_einsum_dense.html">layer_einsum_dense</a>()</code> <br><code><a href="layer_embedding.html">layer_embedding</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_feature_space.html">layer_feature_space</a>()</code> <br><code><a href="layer_flatten.html">layer_flatten</a>()</code> <br><code><a href="layer_flax_module_wrapper.html">layer_flax_module_wrapper</a>()</code> <br><code><a href="layer_gaussian_dropout.html">layer_gaussian_dropout</a>()</code> <br><code><a href="layer_gaussian_noise.html">layer_gaussian_noise</a>()</code> <br><code><a href="layer_global_average_pooling_1d.html">layer_global_average_pooling_1d</a>()</code> <br><code><a href="layer_global_average_pooling_2d.html">layer_global_average_pooling_2d</a>()</code> <br><code><a href="layer_global_average_pooling_3d.html">layer_global_average_pooling_3d</a>()</code> <br><code><a href="layer_global_max_pooling_1d.html">layer_global_max_pooling_1d</a>()</code> <br><code><a href="layer_global_max_pooling_2d.html">layer_global_max_pooling_2d</a>()</code> <br><code><a href="layer_global_max_pooling_3d.html">layer_global_max_pooling_3d</a>()</code> <br><code><a href="layer_group_normalization.html">layer_group_normalization</a>()</code> <br><code><a href="layer_group_query_attention.html">layer_group_query_attention</a>()</code> <br><code><a href="layer_gru.html">layer_gru</a>()</code> <br><code><a href="layer_hashed_crossing.html">layer_hashed_crossing</a>()</code> <br><code><a href="layer_hashing.html">layer_hashing</a>()</code> <br><code><a href="layer_identity.html">layer_identity</a>()</code> <br><code><a href="layer_integer_lookup.html">layer_integer_lookup</a>()</code> <br><code><a href="layer_jax_model_wrapper.html">layer_jax_model_wrapper</a>()</code> <br><code><a href="layer_lambda.html">layer_lambda</a>()</code> <br><code><a href="layer_layer_normalization.html">layer_layer_normalization</a>()</code> <br><code><a href="layer_lstm.html">layer_lstm</a>()</code> <br><code><a href="layer_masking.html">layer_masking</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_max_pooling_1d.html">layer_max_pooling_1d</a>()</code> <br><code><a href="layer_max_pooling_2d.html">layer_max_pooling_2d</a>()</code> <br><code><a href="layer_max_pooling_3d.html">layer_max_pooling_3d</a>()</code> <br><code><a href="layer_maximum.html">layer_maximum</a>()</code> <br><code><a href="layer_mel_spectrogram.html">layer_mel_spectrogram</a>()</code> <br><code><a href="layer_minimum.html">layer_minimum</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_multi_head_attention.html">layer_multi_head_attention</a>()</code> <br><code><a href="layer_multiply.html">layer_multiply</a>()</code> <br><code><a href="layer_normalization.html">layer_normalization</a>()</code> <br><code><a href="layer_permute.html">layer_permute</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_brightness.html">layer_random_brightness</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_contrast.html">layer_random_contrast</a>()</code> <br><code><a href="layer_random_crop.html">layer_random_crop</a>()</code> <br><code><a href="layer_random_elastic_transform.html">layer_random_elastic_transform</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_flip.html">layer_random_flip</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_rotation.html">layer_random_rotation</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_random_translation.html">layer_random_translation</a>()</code> <br><code><a href="layer_random_zoom.html">layer_random_zoom</a>()</code> <br><code><a href="layer_repeat_vector.html">layer_repeat_vector</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_reshape.html">layer_reshape</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_rms_normalization.html">layer_rms_normalization</a>()</code> <br><code><a href="layer_rnn.html">layer_rnn</a>()</code> <br><code><a href="layer_separable_conv_1d.html">layer_separable_conv_1d</a>()</code> <br><code><a href="layer_separable_conv_2d.html">layer_separable_conv_2d</a>()</code> <br><code><a href="layer_simple_rnn.html">layer_simple_rnn</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br><code><a href="layer_spatial_dropout_1d.html">layer_spatial_dropout_1d</a>()</code> <br><code><a href="layer_spatial_dropout_2d.html">layer_spatial_dropout_2d</a>()</code> <br><code><a href="layer_spatial_dropout_3d.html">layer_spatial_dropout_3d</a>()</code> <br><code><a href="layer_spectral_normalization.html">layer_spectral_normalization</a>()</code> <br><code><a href="layer_stft_spectrogram.html">layer_stft_spectrogram</a>()</code> <br><code><a href="layer_string_lookup.html">layer_string_lookup</a>()</code> <br><code><a href="layer_subtract.html">layer_subtract</a>()</code> <br><code><a href="layer_text_vectorization.html">layer_text_vectorization</a>()</code> <br><code><a href="layer_tfsm.html">layer_tfsm</a>()</code> <br><code><a href="layer_time_distributed.html">layer_time_distributed</a>()</code> <br><code><a href="layer_torch_module_wrapper.html">layer_torch_module_wrapper</a>()</code> <br><code><a href="layer_unit_normalization.html">layer_unit_normalization</a>()</code> <br><code><a href="layer_upsampling_1d.html">layer_upsampling_1d</a>()</code> <br><code><a href="layer_upsampling_2d.html">layer_upsampling_2d</a>()</code> <br><code><a href="layer_upsampling_3d.html">layer_upsampling_3d</a>()</code> <br><code><a href="layer_zero_padding_1d.html">layer_zero_padding_1d</a>()</code> <br><code><a href="layer_zero_padding_2d.html">layer_zero_padding_2d</a>()</code> <br><code><a href="layer_zero_padding_3d.html">layer_zero_padding_3d</a>()</code> <br><code><a href="rnn_cell_gru.html">rnn_cell_gru</a>()</code> <br><code><a href="rnn_cell_lstm.html">rnn_cell_lstm</a>()</code> <br><code><a href="rnn_cell_simple.html">rnn_cell_simple</a>()</code> <br><code><a href="rnn_cells_stack.html">rnn_cells_stack</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -355,7 +355,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/layer_bidirectional.html b/docs/dev/reference/layer_bidirectional.html
index 513501325..0c5d4fdcf 100644
--- a/docs/dev/reference/layer_bidirectional.html
+++ b/docs/dev/reference/layer_bidirectional.html
@@ -8,7 +8,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -181,7 +181,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
     <div class="dont-index">
 <ul><li><p><a href="https://keras.io/api/layers/recurrent_layers/bidirectional#bidirectional-class" class="external-link">https://keras.io/api/layers/recurrent_layers/bidirectional#bidirectional-class</a></p></li>
 </ul><p>Other rnn layers: <br><code><a href="layer_conv_lstm_1d.html">layer_conv_lstm_1d</a>()</code> <br><code><a href="layer_conv_lstm_2d.html">layer_conv_lstm_2d</a>()</code> <br><code><a href="layer_conv_lstm_3d.html">layer_conv_lstm_3d</a>()</code> <br><code><a href="layer_gru.html">layer_gru</a>()</code> <br><code><a href="layer_lstm.html">layer_lstm</a>()</code> <br><code><a href="layer_rnn.html">layer_rnn</a>()</code> <br><code><a href="layer_simple_rnn.html">layer_simple_rnn</a>()</code> <br><code><a href="layer_time_distributed.html">layer_time_distributed</a>()</code> <br><code><a href="rnn_cell_gru.html">rnn_cell_gru</a>()</code> <br><code><a href="rnn_cell_lstm.html">rnn_cell_lstm</a>()</code> <br><code><a href="rnn_cell_simple.html">rnn_cell_simple</a>()</code> <br><code><a href="rnn_cells_stack.html">rnn_cells_stack</a>()</code> <br></p>
-<p>Other layers: <br><code><a href="Layer.html">Layer</a>()</code> <br><code><a href="layer_activation.html">layer_activation</a>()</code> <br><code><a href="layer_activation_elu.html">layer_activation_elu</a>()</code> <br><code><a href="layer_activation_leaky_relu.html">layer_activation_leaky_relu</a>()</code> <br><code><a href="layer_activation_parametric_relu.html">layer_activation_parametric_relu</a>()</code> <br><code><a href="layer_activation_relu.html">layer_activation_relu</a>()</code> <br><code><a href="layer_activation_softmax.html">layer_activation_softmax</a>()</code> <br><code><a href="layer_activity_regularization.html">layer_activity_regularization</a>()</code> <br><code><a href="layer_add.html">layer_add</a>()</code> <br><code><a href="layer_additive_attention.html">layer_additive_attention</a>()</code> <br><code><a href="layer_alpha_dropout.html">layer_alpha_dropout</a>()</code> <br><code><a href="layer_attention.html">layer_attention</a>()</code> <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_average.html">layer_average</a>()</code> <br><code><a href="layer_average_pooling_1d.html">layer_average_pooling_1d</a>()</code> <br><code><a href="layer_average_pooling_2d.html">layer_average_pooling_2d</a>()</code> <br><code><a href="layer_average_pooling_3d.html">layer_average_pooling_3d</a>()</code> <br><code><a href="layer_batch_normalization.html">layer_batch_normalization</a>()</code> <br><code><a href="layer_category_encoding.html">layer_category_encoding</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_concatenate.html">layer_concatenate</a>()</code> <br><code><a href="layer_conv_1d.html">layer_conv_1d</a>()</code> <br><code><a href="layer_conv_1d_transpose.html">layer_conv_1d_transpose</a>()</code> <br><code><a href="layer_conv_2d.html">layer_conv_2d</a>()</code> <br><code><a href="layer_conv_2d_transpose.html">layer_conv_2d_transpose</a>()</code> <br><code><a href="layer_conv_3d.html">layer_conv_3d</a>()</code> <br><code><a href="layer_conv_3d_transpose.html">layer_conv_3d_transpose</a>()</code> <br><code><a href="layer_conv_lstm_1d.html">layer_conv_lstm_1d</a>()</code> <br><code><a href="layer_conv_lstm_2d.html">layer_conv_lstm_2d</a>()</code> <br><code><a href="layer_conv_lstm_3d.html">layer_conv_lstm_3d</a>()</code> <br><code><a href="layer_cropping_1d.html">layer_cropping_1d</a>()</code> <br><code><a href="layer_cropping_2d.html">layer_cropping_2d</a>()</code> <br><code><a href="layer_cropping_3d.html">layer_cropping_3d</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_dense.html">layer_dense</a>()</code> <br><code><a href="layer_depthwise_conv_1d.html">layer_depthwise_conv_1d</a>()</code> <br><code><a href="layer_depthwise_conv_2d.html">layer_depthwise_conv_2d</a>()</code> <br><code><a href="layer_discretization.html">layer_discretization</a>()</code> <br><code><a href="layer_dot.html">layer_dot</a>()</code> <br><code><a href="layer_dropout.html">layer_dropout</a>()</code> <br><code><a href="layer_einsum_dense.html">layer_einsum_dense</a>()</code> <br><code><a href="layer_embedding.html">layer_embedding</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_feature_space.html">layer_feature_space</a>()</code> <br><code><a href="layer_flatten.html">layer_flatten</a>()</code> <br><code><a href="layer_flax_module_wrapper.html">layer_flax_module_wrapper</a>()</code> <br><code><a href="layer_gaussian_dropout.html">layer_gaussian_dropout</a>()</code> <br><code><a href="layer_gaussian_noise.html">layer_gaussian_noise</a>()</code> <br><code><a href="layer_global_average_pooling_1d.html">layer_global_average_pooling_1d</a>()</code> <br><code><a href="layer_global_average_pooling_2d.html">layer_global_average_pooling_2d</a>()</code> <br><code><a href="layer_global_average_pooling_3d.html">layer_global_average_pooling_3d</a>()</code> <br><code><a href="layer_global_max_pooling_1d.html">layer_global_max_pooling_1d</a>()</code> <br><code><a href="layer_global_max_pooling_2d.html">layer_global_max_pooling_2d</a>()</code> <br><code><a href="layer_global_max_pooling_3d.html">layer_global_max_pooling_3d</a>()</code> <br><code><a href="layer_group_normalization.html">layer_group_normalization</a>()</code> <br><code><a href="layer_group_query_attention.html">layer_group_query_attention</a>()</code> <br><code><a href="layer_gru.html">layer_gru</a>()</code> <br><code><a href="layer_hashed_crossing.html">layer_hashed_crossing</a>()</code> <br><code><a href="layer_hashing.html">layer_hashing</a>()</code> <br><code><a href="layer_identity.html">layer_identity</a>()</code> <br><code><a href="layer_integer_lookup.html">layer_integer_lookup</a>()</code> <br><code><a href="layer_jax_model_wrapper.html">layer_jax_model_wrapper</a>()</code> <br><code><a href="layer_lambda.html">layer_lambda</a>()</code> <br><code><a href="layer_layer_normalization.html">layer_layer_normalization</a>()</code> <br><code><a href="layer_lstm.html">layer_lstm</a>()</code> <br><code><a href="layer_masking.html">layer_masking</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_max_pooling_1d.html">layer_max_pooling_1d</a>()</code> <br><code><a href="layer_max_pooling_2d.html">layer_max_pooling_2d</a>()</code> <br><code><a href="layer_max_pooling_3d.html">layer_max_pooling_3d</a>()</code> <br><code><a href="layer_maximum.html">layer_maximum</a>()</code> <br><code><a href="layer_mel_spectrogram.html">layer_mel_spectrogram</a>()</code> <br><code><a href="layer_minimum.html">layer_minimum</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_multi_head_attention.html">layer_multi_head_attention</a>()</code> <br><code><a href="layer_multiply.html">layer_multiply</a>()</code> <br><code><a href="layer_normalization.html">layer_normalization</a>()</code> <br><code><a href="layer_permute.html">layer_permute</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_brightness.html">layer_random_brightness</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_contrast.html">layer_random_contrast</a>()</code> <br><code><a href="layer_random_crop.html">layer_random_crop</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_flip.html">layer_random_flip</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_rotation.html">layer_random_rotation</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_random_translation.html">layer_random_translation</a>()</code> <br><code><a href="layer_random_zoom.html">layer_random_zoom</a>()</code> <br><code><a href="layer_repeat_vector.html">layer_repeat_vector</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_reshape.html">layer_reshape</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_rms_normalization.html">layer_rms_normalization</a>()</code> <br><code><a href="layer_rnn.html">layer_rnn</a>()</code> <br><code><a href="layer_separable_conv_1d.html">layer_separable_conv_1d</a>()</code> <br><code><a href="layer_separable_conv_2d.html">layer_separable_conv_2d</a>()</code> <br><code><a href="layer_simple_rnn.html">layer_simple_rnn</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br><code><a href="layer_spatial_dropout_1d.html">layer_spatial_dropout_1d</a>()</code> <br><code><a href="layer_spatial_dropout_2d.html">layer_spatial_dropout_2d</a>()</code> <br><code><a href="layer_spatial_dropout_3d.html">layer_spatial_dropout_3d</a>()</code> <br><code><a href="layer_spectral_normalization.html">layer_spectral_normalization</a>()</code> <br><code><a href="layer_stft_spectrogram.html">layer_stft_spectrogram</a>()</code> <br><code><a href="layer_string_lookup.html">layer_string_lookup</a>()</code> <br><code><a href="layer_subtract.html">layer_subtract</a>()</code> <br><code><a href="layer_text_vectorization.html">layer_text_vectorization</a>()</code> <br><code><a href="layer_tfsm.html">layer_tfsm</a>()</code> <br><code><a href="layer_time_distributed.html">layer_time_distributed</a>()</code> <br><code><a href="layer_torch_module_wrapper.html">layer_torch_module_wrapper</a>()</code> <br><code><a href="layer_unit_normalization.html">layer_unit_normalization</a>()</code> <br><code><a href="layer_upsampling_1d.html">layer_upsampling_1d</a>()</code> <br><code><a href="layer_upsampling_2d.html">layer_upsampling_2d</a>()</code> <br><code><a href="layer_upsampling_3d.html">layer_upsampling_3d</a>()</code> <br><code><a href="layer_zero_padding_1d.html">layer_zero_padding_1d</a>()</code> <br><code><a href="layer_zero_padding_2d.html">layer_zero_padding_2d</a>()</code> <br><code><a href="layer_zero_padding_3d.html">layer_zero_padding_3d</a>()</code> <br><code><a href="rnn_cell_gru.html">rnn_cell_gru</a>()</code> <br><code><a href="rnn_cell_lstm.html">rnn_cell_lstm</a>()</code> <br><code><a href="rnn_cell_simple.html">rnn_cell_simple</a>()</code> <br><code><a href="rnn_cells_stack.html">rnn_cells_stack</a>()</code> <br></p></div>
+<p>Other layers: <br><code><a href="Layer.html">Layer</a>()</code> <br><code><a href="layer_activation.html">layer_activation</a>()</code> <br><code><a href="layer_activation_elu.html">layer_activation_elu</a>()</code> <br><code><a href="layer_activation_leaky_relu.html">layer_activation_leaky_relu</a>()</code> <br><code><a href="layer_activation_parametric_relu.html">layer_activation_parametric_relu</a>()</code> <br><code><a href="layer_activation_relu.html">layer_activation_relu</a>()</code> <br><code><a href="layer_activation_softmax.html">layer_activation_softmax</a>()</code> <br><code><a href="layer_activity_regularization.html">layer_activity_regularization</a>()</code> <br><code><a href="layer_add.html">layer_add</a>()</code> <br><code><a href="layer_additive_attention.html">layer_additive_attention</a>()</code> <br><code><a href="layer_alpha_dropout.html">layer_alpha_dropout</a>()</code> <br><code><a href="layer_attention.html">layer_attention</a>()</code> <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_average.html">layer_average</a>()</code> <br><code><a href="layer_average_pooling_1d.html">layer_average_pooling_1d</a>()</code> <br><code><a href="layer_average_pooling_2d.html">layer_average_pooling_2d</a>()</code> <br><code><a href="layer_average_pooling_3d.html">layer_average_pooling_3d</a>()</code> <br><code><a href="layer_batch_normalization.html">layer_batch_normalization</a>()</code> <br><code><a href="layer_category_encoding.html">layer_category_encoding</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_concatenate.html">layer_concatenate</a>()</code> <br><code><a href="layer_conv_1d.html">layer_conv_1d</a>()</code> <br><code><a href="layer_conv_1d_transpose.html">layer_conv_1d_transpose</a>()</code> <br><code><a href="layer_conv_2d.html">layer_conv_2d</a>()</code> <br><code><a href="layer_conv_2d_transpose.html">layer_conv_2d_transpose</a>()</code> <br><code><a href="layer_conv_3d.html">layer_conv_3d</a>()</code> <br><code><a href="layer_conv_3d_transpose.html">layer_conv_3d_transpose</a>()</code> <br><code><a href="layer_conv_lstm_1d.html">layer_conv_lstm_1d</a>()</code> <br><code><a href="layer_conv_lstm_2d.html">layer_conv_lstm_2d</a>()</code> <br><code><a href="layer_conv_lstm_3d.html">layer_conv_lstm_3d</a>()</code> <br><code><a href="layer_cropping_1d.html">layer_cropping_1d</a>()</code> <br><code><a href="layer_cropping_2d.html">layer_cropping_2d</a>()</code> <br><code><a href="layer_cropping_3d.html">layer_cropping_3d</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_dense.html">layer_dense</a>()</code> <br><code><a href="layer_depthwise_conv_1d.html">layer_depthwise_conv_1d</a>()</code> <br><code><a href="layer_depthwise_conv_2d.html">layer_depthwise_conv_2d</a>()</code> <br><code><a href="layer_discretization.html">layer_discretization</a>()</code> <br><code><a href="layer_dot.html">layer_dot</a>()</code> <br><code><a href="layer_dropout.html">layer_dropout</a>()</code> <br><code><a href="layer_einsum_dense.html">layer_einsum_dense</a>()</code> <br><code><a href="layer_embedding.html">layer_embedding</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_feature_space.html">layer_feature_space</a>()</code> <br><code><a href="layer_flatten.html">layer_flatten</a>()</code> <br><code><a href="layer_flax_module_wrapper.html">layer_flax_module_wrapper</a>()</code> <br><code><a href="layer_gaussian_dropout.html">layer_gaussian_dropout</a>()</code> <br><code><a href="layer_gaussian_noise.html">layer_gaussian_noise</a>()</code> <br><code><a href="layer_global_average_pooling_1d.html">layer_global_average_pooling_1d</a>()</code> <br><code><a href="layer_global_average_pooling_2d.html">layer_global_average_pooling_2d</a>()</code> <br><code><a href="layer_global_average_pooling_3d.html">layer_global_average_pooling_3d</a>()</code> <br><code><a href="layer_global_max_pooling_1d.html">layer_global_max_pooling_1d</a>()</code> <br><code><a href="layer_global_max_pooling_2d.html">layer_global_max_pooling_2d</a>()</code> <br><code><a href="layer_global_max_pooling_3d.html">layer_global_max_pooling_3d</a>()</code> <br><code><a href="layer_group_normalization.html">layer_group_normalization</a>()</code> <br><code><a href="layer_group_query_attention.html">layer_group_query_attention</a>()</code> <br><code><a href="layer_gru.html">layer_gru</a>()</code> <br><code><a href="layer_hashed_crossing.html">layer_hashed_crossing</a>()</code> <br><code><a href="layer_hashing.html">layer_hashing</a>()</code> <br><code><a href="layer_identity.html">layer_identity</a>()</code> <br><code><a href="layer_integer_lookup.html">layer_integer_lookup</a>()</code> <br><code><a href="layer_jax_model_wrapper.html">layer_jax_model_wrapper</a>()</code> <br><code><a href="layer_lambda.html">layer_lambda</a>()</code> <br><code><a href="layer_layer_normalization.html">layer_layer_normalization</a>()</code> <br><code><a href="layer_lstm.html">layer_lstm</a>()</code> <br><code><a href="layer_masking.html">layer_masking</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_max_pooling_1d.html">layer_max_pooling_1d</a>()</code> <br><code><a href="layer_max_pooling_2d.html">layer_max_pooling_2d</a>()</code> <br><code><a href="layer_max_pooling_3d.html">layer_max_pooling_3d</a>()</code> <br><code><a href="layer_maximum.html">layer_maximum</a>()</code> <br><code><a href="layer_mel_spectrogram.html">layer_mel_spectrogram</a>()</code> <br><code><a href="layer_minimum.html">layer_minimum</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_multi_head_attention.html">layer_multi_head_attention</a>()</code> <br><code><a href="layer_multiply.html">layer_multiply</a>()</code> <br><code><a href="layer_normalization.html">layer_normalization</a>()</code> <br><code><a href="layer_permute.html">layer_permute</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_brightness.html">layer_random_brightness</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_contrast.html">layer_random_contrast</a>()</code> <br><code><a href="layer_random_crop.html">layer_random_crop</a>()</code> <br><code><a href="layer_random_elastic_transform.html">layer_random_elastic_transform</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_flip.html">layer_random_flip</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_rotation.html">layer_random_rotation</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_random_translation.html">layer_random_translation</a>()</code> <br><code><a href="layer_random_zoom.html">layer_random_zoom</a>()</code> <br><code><a href="layer_repeat_vector.html">layer_repeat_vector</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_reshape.html">layer_reshape</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_rms_normalization.html">layer_rms_normalization</a>()</code> <br><code><a href="layer_rnn.html">layer_rnn</a>()</code> <br><code><a href="layer_separable_conv_1d.html">layer_separable_conv_1d</a>()</code> <br><code><a href="layer_separable_conv_2d.html">layer_separable_conv_2d</a>()</code> <br><code><a href="layer_simple_rnn.html">layer_simple_rnn</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br><code><a href="layer_spatial_dropout_1d.html">layer_spatial_dropout_1d</a>()</code> <br><code><a href="layer_spatial_dropout_2d.html">layer_spatial_dropout_2d</a>()</code> <br><code><a href="layer_spatial_dropout_3d.html">layer_spatial_dropout_3d</a>()</code> <br><code><a href="layer_spectral_normalization.html">layer_spectral_normalization</a>()</code> <br><code><a href="layer_stft_spectrogram.html">layer_stft_spectrogram</a>()</code> <br><code><a href="layer_string_lookup.html">layer_string_lookup</a>()</code> <br><code><a href="layer_subtract.html">layer_subtract</a>()</code> <br><code><a href="layer_text_vectorization.html">layer_text_vectorization</a>()</code> <br><code><a href="layer_tfsm.html">layer_tfsm</a>()</code> <br><code><a href="layer_time_distributed.html">layer_time_distributed</a>()</code> <br><code><a href="layer_torch_module_wrapper.html">layer_torch_module_wrapper</a>()</code> <br><code><a href="layer_unit_normalization.html">layer_unit_normalization</a>()</code> <br><code><a href="layer_upsampling_1d.html">layer_upsampling_1d</a>()</code> <br><code><a href="layer_upsampling_2d.html">layer_upsampling_2d</a>()</code> <br><code><a href="layer_upsampling_3d.html">layer_upsampling_3d</a>()</code> <br><code><a href="layer_zero_padding_1d.html">layer_zero_padding_1d</a>()</code> <br><code><a href="layer_zero_padding_2d.html">layer_zero_padding_2d</a>()</code> <br><code><a href="layer_zero_padding_3d.html">layer_zero_padding_3d</a>()</code> <br><code><a href="rnn_cell_gru.html">rnn_cell_gru</a>()</code> <br><code><a href="rnn_cell_lstm.html">rnn_cell_lstm</a>()</code> <br><code><a href="rnn_cell_simple.html">rnn_cell_simple</a>()</code> <br><code><a href="rnn_cells_stack.html">rnn_cells_stack</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -193,7 +193,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/layer_category_encoding.html b/docs/dev/reference/layer_category_encoding.html
index fdef05471..deec25b39 100644
--- a/docs/dev/reference/layer_category_encoding.html
+++ b/docs/dev/reference/layer_category_encoding.html
@@ -20,7 +20,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -194,8 +194,8 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
     <div class="dont-index">
 <ul><li><p><a href="https://keras.io/api/layers/preprocessing_layers/categorical/category_encoding#categoryencoding-class" class="external-link">https://keras.io/api/layers/preprocessing_layers/categorical/category_encoding#categoryencoding-class</a></p></li>
 </ul><p>Other categorical features preprocessing layers: <br><code><a href="layer_hashed_crossing.html">layer_hashed_crossing</a>()</code> <br><code><a href="layer_hashing.html">layer_hashing</a>()</code> <br><code><a href="layer_integer_lookup.html">layer_integer_lookup</a>()</code> <br><code><a href="layer_string_lookup.html">layer_string_lookup</a>()</code> <br></p>
-<p>Other preprocessing layers: <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_discretization.html">layer_discretization</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_feature_space.html">layer_feature_space</a>()</code> <br><code><a href="layer_hashed_crossing.html">layer_hashed_crossing</a>()</code> <br><code><a href="layer_hashing.html">layer_hashing</a>()</code> <br><code><a href="layer_integer_lookup.html">layer_integer_lookup</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_mel_spectrogram.html">layer_mel_spectrogram</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_normalization.html">layer_normalization</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_brightness.html">layer_random_brightness</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_contrast.html">layer_random_contrast</a>()</code> <br><code><a href="layer_random_crop.html">layer_random_crop</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_flip.html">layer_random_flip</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_rotation.html">layer_random_rotation</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_random_translation.html">layer_random_translation</a>()</code> <br><code><a href="layer_random_zoom.html">layer_random_zoom</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br><code><a href="layer_stft_spectrogram.html">layer_stft_spectrogram</a>()</code> <br><code><a href="layer_string_lookup.html">layer_string_lookup</a>()</code> <br><code><a href="layer_text_vectorization.html">layer_text_vectorization</a>()</code> <br></p>
-<p>Other layers: <br><code><a href="Layer.html">Layer</a>()</code> <br><code><a href="layer_activation.html">layer_activation</a>()</code> <br><code><a href="layer_activation_elu.html">layer_activation_elu</a>()</code> <br><code><a href="layer_activation_leaky_relu.html">layer_activation_leaky_relu</a>()</code> <br><code><a href="layer_activation_parametric_relu.html">layer_activation_parametric_relu</a>()</code> <br><code><a href="layer_activation_relu.html">layer_activation_relu</a>()</code> <br><code><a href="layer_activation_softmax.html">layer_activation_softmax</a>()</code> <br><code><a href="layer_activity_regularization.html">layer_activity_regularization</a>()</code> <br><code><a href="layer_add.html">layer_add</a>()</code> <br><code><a href="layer_additive_attention.html">layer_additive_attention</a>()</code> <br><code><a href="layer_alpha_dropout.html">layer_alpha_dropout</a>()</code> <br><code><a href="layer_attention.html">layer_attention</a>()</code> <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_average.html">layer_average</a>()</code> <br><code><a href="layer_average_pooling_1d.html">layer_average_pooling_1d</a>()</code> <br><code><a href="layer_average_pooling_2d.html">layer_average_pooling_2d</a>()</code> <br><code><a href="layer_average_pooling_3d.html">layer_average_pooling_3d</a>()</code> <br><code><a href="layer_batch_normalization.html">layer_batch_normalization</a>()</code> <br><code><a href="layer_bidirectional.html">layer_bidirectional</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_concatenate.html">layer_concatenate</a>()</code> <br><code><a href="layer_conv_1d.html">layer_conv_1d</a>()</code> <br><code><a href="layer_conv_1d_transpose.html">layer_conv_1d_transpose</a>()</code> <br><code><a href="layer_conv_2d.html">layer_conv_2d</a>()</code> <br><code><a href="layer_conv_2d_transpose.html">layer_conv_2d_transpose</a>()</code> <br><code><a href="layer_conv_3d.html">layer_conv_3d</a>()</code> <br><code><a href="layer_conv_3d_transpose.html">layer_conv_3d_transpose</a>()</code> <br><code><a href="layer_conv_lstm_1d.html">layer_conv_lstm_1d</a>()</code> <br><code><a href="layer_conv_lstm_2d.html">layer_conv_lstm_2d</a>()</code> <br><code><a href="layer_conv_lstm_3d.html">layer_conv_lstm_3d</a>()</code> <br><code><a href="layer_cropping_1d.html">layer_cropping_1d</a>()</code> <br><code><a href="layer_cropping_2d.html">layer_cropping_2d</a>()</code> <br><code><a href="layer_cropping_3d.html">layer_cropping_3d</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_dense.html">layer_dense</a>()</code> <br><code><a href="layer_depthwise_conv_1d.html">layer_depthwise_conv_1d</a>()</code> <br><code><a href="layer_depthwise_conv_2d.html">layer_depthwise_conv_2d</a>()</code> <br><code><a href="layer_discretization.html">layer_discretization</a>()</code> <br><code><a href="layer_dot.html">layer_dot</a>()</code> <br><code><a href="layer_dropout.html">layer_dropout</a>()</code> <br><code><a href="layer_einsum_dense.html">layer_einsum_dense</a>()</code> <br><code><a href="layer_embedding.html">layer_embedding</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_feature_space.html">layer_feature_space</a>()</code> <br><code><a href="layer_flatten.html">layer_flatten</a>()</code> <br><code><a href="layer_flax_module_wrapper.html">layer_flax_module_wrapper</a>()</code> <br><code><a href="layer_gaussian_dropout.html">layer_gaussian_dropout</a>()</code> <br><code><a href="layer_gaussian_noise.html">layer_gaussian_noise</a>()</code> <br><code><a href="layer_global_average_pooling_1d.html">layer_global_average_pooling_1d</a>()</code> <br><code><a href="layer_global_average_pooling_2d.html">layer_global_average_pooling_2d</a>()</code> <br><code><a href="layer_global_average_pooling_3d.html">layer_global_average_pooling_3d</a>()</code> <br><code><a href="layer_global_max_pooling_1d.html">layer_global_max_pooling_1d</a>()</code> <br><code><a href="layer_global_max_pooling_2d.html">layer_global_max_pooling_2d</a>()</code> <br><code><a href="layer_global_max_pooling_3d.html">layer_global_max_pooling_3d</a>()</code> <br><code><a href="layer_group_normalization.html">layer_group_normalization</a>()</code> <br><code><a href="layer_group_query_attention.html">layer_group_query_attention</a>()</code> <br><code><a href="layer_gru.html">layer_gru</a>()</code> <br><code><a href="layer_hashed_crossing.html">layer_hashed_crossing</a>()</code> <br><code><a href="layer_hashing.html">layer_hashing</a>()</code> <br><code><a href="layer_identity.html">layer_identity</a>()</code> <br><code><a href="layer_integer_lookup.html">layer_integer_lookup</a>()</code> <br><code><a href="layer_jax_model_wrapper.html">layer_jax_model_wrapper</a>()</code> <br><code><a href="layer_lambda.html">layer_lambda</a>()</code> <br><code><a href="layer_layer_normalization.html">layer_layer_normalization</a>()</code> <br><code><a href="layer_lstm.html">layer_lstm</a>()</code> <br><code><a href="layer_masking.html">layer_masking</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_max_pooling_1d.html">layer_max_pooling_1d</a>()</code> <br><code><a href="layer_max_pooling_2d.html">layer_max_pooling_2d</a>()</code> <br><code><a href="layer_max_pooling_3d.html">layer_max_pooling_3d</a>()</code> <br><code><a href="layer_maximum.html">layer_maximum</a>()</code> <br><code><a href="layer_mel_spectrogram.html">layer_mel_spectrogram</a>()</code> <br><code><a href="layer_minimum.html">layer_minimum</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_multi_head_attention.html">layer_multi_head_attention</a>()</code> <br><code><a href="layer_multiply.html">layer_multiply</a>()</code> <br><code><a href="layer_normalization.html">layer_normalization</a>()</code> <br><code><a href="layer_permute.html">layer_permute</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_brightness.html">layer_random_brightness</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_contrast.html">layer_random_contrast</a>()</code> <br><code><a href="layer_random_crop.html">layer_random_crop</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_flip.html">layer_random_flip</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_rotation.html">layer_random_rotation</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_random_translation.html">layer_random_translation</a>()</code> <br><code><a href="layer_random_zoom.html">layer_random_zoom</a>()</code> <br><code><a href="layer_repeat_vector.html">layer_repeat_vector</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_reshape.html">layer_reshape</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_rms_normalization.html">layer_rms_normalization</a>()</code> <br><code><a href="layer_rnn.html">layer_rnn</a>()</code> <br><code><a href="layer_separable_conv_1d.html">layer_separable_conv_1d</a>()</code> <br><code><a href="layer_separable_conv_2d.html">layer_separable_conv_2d</a>()</code> <br><code><a href="layer_simple_rnn.html">layer_simple_rnn</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br><code><a href="layer_spatial_dropout_1d.html">layer_spatial_dropout_1d</a>()</code> <br><code><a href="layer_spatial_dropout_2d.html">layer_spatial_dropout_2d</a>()</code> <br><code><a href="layer_spatial_dropout_3d.html">layer_spatial_dropout_3d</a>()</code> <br><code><a href="layer_spectral_normalization.html">layer_spectral_normalization</a>()</code> <br><code><a href="layer_stft_spectrogram.html">layer_stft_spectrogram</a>()</code> <br><code><a href="layer_string_lookup.html">layer_string_lookup</a>()</code> <br><code><a href="layer_subtract.html">layer_subtract</a>()</code> <br><code><a href="layer_text_vectorization.html">layer_text_vectorization</a>()</code> <br><code><a href="layer_tfsm.html">layer_tfsm</a>()</code> <br><code><a href="layer_time_distributed.html">layer_time_distributed</a>()</code> <br><code><a href="layer_torch_module_wrapper.html">layer_torch_module_wrapper</a>()</code> <br><code><a href="layer_unit_normalization.html">layer_unit_normalization</a>()</code> <br><code><a href="layer_upsampling_1d.html">layer_upsampling_1d</a>()</code> <br><code><a href="layer_upsampling_2d.html">layer_upsampling_2d</a>()</code> <br><code><a href="layer_upsampling_3d.html">layer_upsampling_3d</a>()</code> <br><code><a href="layer_zero_padding_1d.html">layer_zero_padding_1d</a>()</code> <br><code><a href="layer_zero_padding_2d.html">layer_zero_padding_2d</a>()</code> <br><code><a href="layer_zero_padding_3d.html">layer_zero_padding_3d</a>()</code> <br><code><a href="rnn_cell_gru.html">rnn_cell_gru</a>()</code> <br><code><a href="rnn_cell_lstm.html">rnn_cell_lstm</a>()</code> <br><code><a href="rnn_cell_simple.html">rnn_cell_simple</a>()</code> <br><code><a href="rnn_cells_stack.html">rnn_cells_stack</a>()</code> <br></p></div>
+<p>Other preprocessing layers: <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_discretization.html">layer_discretization</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_feature_space.html">layer_feature_space</a>()</code> <br><code><a href="layer_hashed_crossing.html">layer_hashed_crossing</a>()</code> <br><code><a href="layer_hashing.html">layer_hashing</a>()</code> <br><code><a href="layer_integer_lookup.html">layer_integer_lookup</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_mel_spectrogram.html">layer_mel_spectrogram</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_normalization.html">layer_normalization</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_brightness.html">layer_random_brightness</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_contrast.html">layer_random_contrast</a>()</code> <br><code><a href="layer_random_crop.html">layer_random_crop</a>()</code> <br><code><a href="layer_random_elastic_transform.html">layer_random_elastic_transform</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_flip.html">layer_random_flip</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_rotation.html">layer_random_rotation</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_random_translation.html">layer_random_translation</a>()</code> <br><code><a href="layer_random_zoom.html">layer_random_zoom</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br><code><a href="layer_stft_spectrogram.html">layer_stft_spectrogram</a>()</code> <br><code><a href="layer_string_lookup.html">layer_string_lookup</a>()</code> <br><code><a href="layer_text_vectorization.html">layer_text_vectorization</a>()</code> <br></p>
+<p>Other layers: <br><code><a href="Layer.html">Layer</a>()</code> <br><code><a href="layer_activation.html">layer_activation</a>()</code> <br><code><a href="layer_activation_elu.html">layer_activation_elu</a>()</code> <br><code><a href="layer_activation_leaky_relu.html">layer_activation_leaky_relu</a>()</code> <br><code><a href="layer_activation_parametric_relu.html">layer_activation_parametric_relu</a>()</code> <br><code><a href="layer_activation_relu.html">layer_activation_relu</a>()</code> <br><code><a href="layer_activation_softmax.html">layer_activation_softmax</a>()</code> <br><code><a href="layer_activity_regularization.html">layer_activity_regularization</a>()</code> <br><code><a href="layer_add.html">layer_add</a>()</code> <br><code><a href="layer_additive_attention.html">layer_additive_attention</a>()</code> <br><code><a href="layer_alpha_dropout.html">layer_alpha_dropout</a>()</code> <br><code><a href="layer_attention.html">layer_attention</a>()</code> <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_average.html">layer_average</a>()</code> <br><code><a href="layer_average_pooling_1d.html">layer_average_pooling_1d</a>()</code> <br><code><a href="layer_average_pooling_2d.html">layer_average_pooling_2d</a>()</code> <br><code><a href="layer_average_pooling_3d.html">layer_average_pooling_3d</a>()</code> <br><code><a href="layer_batch_normalization.html">layer_batch_normalization</a>()</code> <br><code><a href="layer_bidirectional.html">layer_bidirectional</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_concatenate.html">layer_concatenate</a>()</code> <br><code><a href="layer_conv_1d.html">layer_conv_1d</a>()</code> <br><code><a href="layer_conv_1d_transpose.html">layer_conv_1d_transpose</a>()</code> <br><code><a href="layer_conv_2d.html">layer_conv_2d</a>()</code> <br><code><a href="layer_conv_2d_transpose.html">layer_conv_2d_transpose</a>()</code> <br><code><a href="layer_conv_3d.html">layer_conv_3d</a>()</code> <br><code><a href="layer_conv_3d_transpose.html">layer_conv_3d_transpose</a>()</code> <br><code><a href="layer_conv_lstm_1d.html">layer_conv_lstm_1d</a>()</code> <br><code><a href="layer_conv_lstm_2d.html">layer_conv_lstm_2d</a>()</code> <br><code><a href="layer_conv_lstm_3d.html">layer_conv_lstm_3d</a>()</code> <br><code><a href="layer_cropping_1d.html">layer_cropping_1d</a>()</code> <br><code><a href="layer_cropping_2d.html">layer_cropping_2d</a>()</code> <br><code><a href="layer_cropping_3d.html">layer_cropping_3d</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_dense.html">layer_dense</a>()</code> <br><code><a href="layer_depthwise_conv_1d.html">layer_depthwise_conv_1d</a>()</code> <br><code><a href="layer_depthwise_conv_2d.html">layer_depthwise_conv_2d</a>()</code> <br><code><a href="layer_discretization.html">layer_discretization</a>()</code> <br><code><a href="layer_dot.html">layer_dot</a>()</code> <br><code><a href="layer_dropout.html">layer_dropout</a>()</code> <br><code><a href="layer_einsum_dense.html">layer_einsum_dense</a>()</code> <br><code><a href="layer_embedding.html">layer_embedding</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_feature_space.html">layer_feature_space</a>()</code> <br><code><a href="layer_flatten.html">layer_flatten</a>()</code> <br><code><a href="layer_flax_module_wrapper.html">layer_flax_module_wrapper</a>()</code> <br><code><a href="layer_gaussian_dropout.html">layer_gaussian_dropout</a>()</code> <br><code><a href="layer_gaussian_noise.html">layer_gaussian_noise</a>()</code> <br><code><a href="layer_global_average_pooling_1d.html">layer_global_average_pooling_1d</a>()</code> <br><code><a href="layer_global_average_pooling_2d.html">layer_global_average_pooling_2d</a>()</code> <br><code><a href="layer_global_average_pooling_3d.html">layer_global_average_pooling_3d</a>()</code> <br><code><a href="layer_global_max_pooling_1d.html">layer_global_max_pooling_1d</a>()</code> <br><code><a href="layer_global_max_pooling_2d.html">layer_global_max_pooling_2d</a>()</code> <br><code><a href="layer_global_max_pooling_3d.html">layer_global_max_pooling_3d</a>()</code> <br><code><a href="layer_group_normalization.html">layer_group_normalization</a>()</code> <br><code><a href="layer_group_query_attention.html">layer_group_query_attention</a>()</code> <br><code><a href="layer_gru.html">layer_gru</a>()</code> <br><code><a href="layer_hashed_crossing.html">layer_hashed_crossing</a>()</code> <br><code><a href="layer_hashing.html">layer_hashing</a>()</code> <br><code><a href="layer_identity.html">layer_identity</a>()</code> <br><code><a href="layer_integer_lookup.html">layer_integer_lookup</a>()</code> <br><code><a href="layer_jax_model_wrapper.html">layer_jax_model_wrapper</a>()</code> <br><code><a href="layer_lambda.html">layer_lambda</a>()</code> <br><code><a href="layer_layer_normalization.html">layer_layer_normalization</a>()</code> <br><code><a href="layer_lstm.html">layer_lstm</a>()</code> <br><code><a href="layer_masking.html">layer_masking</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_max_pooling_1d.html">layer_max_pooling_1d</a>()</code> <br><code><a href="layer_max_pooling_2d.html">layer_max_pooling_2d</a>()</code> <br><code><a href="layer_max_pooling_3d.html">layer_max_pooling_3d</a>()</code> <br><code><a href="layer_maximum.html">layer_maximum</a>()</code> <br><code><a href="layer_mel_spectrogram.html">layer_mel_spectrogram</a>()</code> <br><code><a href="layer_minimum.html">layer_minimum</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_multi_head_attention.html">layer_multi_head_attention</a>()</code> <br><code><a href="layer_multiply.html">layer_multiply</a>()</code> <br><code><a href="layer_normalization.html">layer_normalization</a>()</code> <br><code><a href="layer_permute.html">layer_permute</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_brightness.html">layer_random_brightness</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_contrast.html">layer_random_contrast</a>()</code> <br><code><a href="layer_random_crop.html">layer_random_crop</a>()</code> <br><code><a href="layer_random_elastic_transform.html">layer_random_elastic_transform</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_flip.html">layer_random_flip</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_rotation.html">layer_random_rotation</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_random_translation.html">layer_random_translation</a>()</code> <br><code><a href="layer_random_zoom.html">layer_random_zoom</a>()</code> <br><code><a href="layer_repeat_vector.html">layer_repeat_vector</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_reshape.html">layer_reshape</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_rms_normalization.html">layer_rms_normalization</a>()</code> <br><code><a href="layer_rnn.html">layer_rnn</a>()</code> <br><code><a href="layer_separable_conv_1d.html">layer_separable_conv_1d</a>()</code> <br><code><a href="layer_separable_conv_2d.html">layer_separable_conv_2d</a>()</code> <br><code><a href="layer_simple_rnn.html">layer_simple_rnn</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br><code><a href="layer_spatial_dropout_1d.html">layer_spatial_dropout_1d</a>()</code> <br><code><a href="layer_spatial_dropout_2d.html">layer_spatial_dropout_2d</a>()</code> <br><code><a href="layer_spatial_dropout_3d.html">layer_spatial_dropout_3d</a>()</code> <br><code><a href="layer_spectral_normalization.html">layer_spectral_normalization</a>()</code> <br><code><a href="layer_stft_spectrogram.html">layer_stft_spectrogram</a>()</code> <br><code><a href="layer_string_lookup.html">layer_string_lookup</a>()</code> <br><code><a href="layer_subtract.html">layer_subtract</a>()</code> <br><code><a href="layer_text_vectorization.html">layer_text_vectorization</a>()</code> <br><code><a href="layer_tfsm.html">layer_tfsm</a>()</code> <br><code><a href="layer_time_distributed.html">layer_time_distributed</a>()</code> <br><code><a href="layer_torch_module_wrapper.html">layer_torch_module_wrapper</a>()</code> <br><code><a href="layer_unit_normalization.html">layer_unit_normalization</a>()</code> <br><code><a href="layer_upsampling_1d.html">layer_upsampling_1d</a>()</code> <br><code><a href="layer_upsampling_2d.html">layer_upsampling_2d</a>()</code> <br><code><a href="layer_upsampling_3d.html">layer_upsampling_3d</a>()</code> <br><code><a href="layer_zero_padding_1d.html">layer_zero_padding_1d</a>()</code> <br><code><a href="layer_zero_padding_2d.html">layer_zero_padding_2d</a>()</code> <br><code><a href="layer_zero_padding_3d.html">layer_zero_padding_3d</a>()</code> <br><code><a href="rnn_cell_gru.html">rnn_cell_gru</a>()</code> <br><code><a href="rnn_cell_lstm.html">rnn_cell_lstm</a>()</code> <br><code><a href="rnn_cell_simple.html">rnn_cell_simple</a>()</code> <br><code><a href="rnn_cells_stack.html">rnn_cells_stack</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -207,7 +207,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/layer_center_crop.html b/docs/dev/reference/layer_center_crop.html
index a2b84d4c9..4732f0475 100644
--- a/docs/dev/reference/layer_center_crop.html
+++ b/docs/dev/reference/layer_center_crop.html
@@ -16,7 +16,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -135,9 +135,9 @@ <h2 id="output-shape">Output Shape<a class="anchor" aria-label="anchor" href="#o
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
     <div class="dont-index">
 <ul><li><p><a href="https://keras.io/api/layers/preprocessing_layers/image_preprocessing/center_crop#centercrop-class" class="external-link">https://keras.io/api/layers/preprocessing_layers/image_preprocessing/center_crop#centercrop-class</a></p></li>
-</ul><p>Other image preprocessing layers: <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br></p>
-<p>Other preprocessing layers: <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_category_encoding.html">layer_category_encoding</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_discretization.html">layer_discretization</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_feature_space.html">layer_feature_space</a>()</code> <br><code><a href="layer_hashed_crossing.html">layer_hashed_crossing</a>()</code> <br><code><a href="layer_hashing.html">layer_hashing</a>()</code> <br><code><a href="layer_integer_lookup.html">layer_integer_lookup</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_mel_spectrogram.html">layer_mel_spectrogram</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_normalization.html">layer_normalization</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_brightness.html">layer_random_brightness</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_contrast.html">layer_random_contrast</a>()</code> <br><code><a href="layer_random_crop.html">layer_random_crop</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_flip.html">layer_random_flip</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_rotation.html">layer_random_rotation</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_random_translation.html">layer_random_translation</a>()</code> <br><code><a href="layer_random_zoom.html">layer_random_zoom</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br><code><a href="layer_stft_spectrogram.html">layer_stft_spectrogram</a>()</code> <br><code><a href="layer_string_lookup.html">layer_string_lookup</a>()</code> <br><code><a href="layer_text_vectorization.html">layer_text_vectorization</a>()</code> <br></p>
-<p>Other layers: <br><code><a href="Layer.html">Layer</a>()</code> <br><code><a href="layer_activation.html">layer_activation</a>()</code> <br><code><a href="layer_activation_elu.html">layer_activation_elu</a>()</code> <br><code><a href="layer_activation_leaky_relu.html">layer_activation_leaky_relu</a>()</code> <br><code><a href="layer_activation_parametric_relu.html">layer_activation_parametric_relu</a>()</code> <br><code><a href="layer_activation_relu.html">layer_activation_relu</a>()</code> <br><code><a href="layer_activation_softmax.html">layer_activation_softmax</a>()</code> <br><code><a href="layer_activity_regularization.html">layer_activity_regularization</a>()</code> <br><code><a href="layer_add.html">layer_add</a>()</code> <br><code><a href="layer_additive_attention.html">layer_additive_attention</a>()</code> <br><code><a href="layer_alpha_dropout.html">layer_alpha_dropout</a>()</code> <br><code><a href="layer_attention.html">layer_attention</a>()</code> <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_average.html">layer_average</a>()</code> <br><code><a href="layer_average_pooling_1d.html">layer_average_pooling_1d</a>()</code> <br><code><a href="layer_average_pooling_2d.html">layer_average_pooling_2d</a>()</code> <br><code><a href="layer_average_pooling_3d.html">layer_average_pooling_3d</a>()</code> <br><code><a href="layer_batch_normalization.html">layer_batch_normalization</a>()</code> <br><code><a href="layer_bidirectional.html">layer_bidirectional</a>()</code> <br><code><a href="layer_category_encoding.html">layer_category_encoding</a>()</code> <br><code><a href="layer_concatenate.html">layer_concatenate</a>()</code> <br><code><a href="layer_conv_1d.html">layer_conv_1d</a>()</code> <br><code><a href="layer_conv_1d_transpose.html">layer_conv_1d_transpose</a>()</code> <br><code><a href="layer_conv_2d.html">layer_conv_2d</a>()</code> <br><code><a href="layer_conv_2d_transpose.html">layer_conv_2d_transpose</a>()</code> <br><code><a href="layer_conv_3d.html">layer_conv_3d</a>()</code> <br><code><a href="layer_conv_3d_transpose.html">layer_conv_3d_transpose</a>()</code> <br><code><a href="layer_conv_lstm_1d.html">layer_conv_lstm_1d</a>()</code> <br><code><a href="layer_conv_lstm_2d.html">layer_conv_lstm_2d</a>()</code> <br><code><a href="layer_conv_lstm_3d.html">layer_conv_lstm_3d</a>()</code> <br><code><a href="layer_cropping_1d.html">layer_cropping_1d</a>()</code> <br><code><a href="layer_cropping_2d.html">layer_cropping_2d</a>()</code> <br><code><a href="layer_cropping_3d.html">layer_cropping_3d</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_dense.html">layer_dense</a>()</code> <br><code><a href="layer_depthwise_conv_1d.html">layer_depthwise_conv_1d</a>()</code> <br><code><a href="layer_depthwise_conv_2d.html">layer_depthwise_conv_2d</a>()</code> <br><code><a href="layer_discretization.html">layer_discretization</a>()</code> <br><code><a href="layer_dot.html">layer_dot</a>()</code> <br><code><a href="layer_dropout.html">layer_dropout</a>()</code> <br><code><a href="layer_einsum_dense.html">layer_einsum_dense</a>()</code> <br><code><a href="layer_embedding.html">layer_embedding</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_feature_space.html">layer_feature_space</a>()</code> <br><code><a href="layer_flatten.html">layer_flatten</a>()</code> <br><code><a href="layer_flax_module_wrapper.html">layer_flax_module_wrapper</a>()</code> <br><code><a href="layer_gaussian_dropout.html">layer_gaussian_dropout</a>()</code> <br><code><a href="layer_gaussian_noise.html">layer_gaussian_noise</a>()</code> <br><code><a href="layer_global_average_pooling_1d.html">layer_global_average_pooling_1d</a>()</code> <br><code><a href="layer_global_average_pooling_2d.html">layer_global_average_pooling_2d</a>()</code> <br><code><a href="layer_global_average_pooling_3d.html">layer_global_average_pooling_3d</a>()</code> <br><code><a href="layer_global_max_pooling_1d.html">layer_global_max_pooling_1d</a>()</code> <br><code><a href="layer_global_max_pooling_2d.html">layer_global_max_pooling_2d</a>()</code> <br><code><a href="layer_global_max_pooling_3d.html">layer_global_max_pooling_3d</a>()</code> <br><code><a href="layer_group_normalization.html">layer_group_normalization</a>()</code> <br><code><a href="layer_group_query_attention.html">layer_group_query_attention</a>()</code> <br><code><a href="layer_gru.html">layer_gru</a>()</code> <br><code><a href="layer_hashed_crossing.html">layer_hashed_crossing</a>()</code> <br><code><a href="layer_hashing.html">layer_hashing</a>()</code> <br><code><a href="layer_identity.html">layer_identity</a>()</code> <br><code><a href="layer_integer_lookup.html">layer_integer_lookup</a>()</code> <br><code><a href="layer_jax_model_wrapper.html">layer_jax_model_wrapper</a>()</code> <br><code><a href="layer_lambda.html">layer_lambda</a>()</code> <br><code><a href="layer_layer_normalization.html">layer_layer_normalization</a>()</code> <br><code><a href="layer_lstm.html">layer_lstm</a>()</code> <br><code><a href="layer_masking.html">layer_masking</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_max_pooling_1d.html">layer_max_pooling_1d</a>()</code> <br><code><a href="layer_max_pooling_2d.html">layer_max_pooling_2d</a>()</code> <br><code><a href="layer_max_pooling_3d.html">layer_max_pooling_3d</a>()</code> <br><code><a href="layer_maximum.html">layer_maximum</a>()</code> <br><code><a href="layer_mel_spectrogram.html">layer_mel_spectrogram</a>()</code> <br><code><a href="layer_minimum.html">layer_minimum</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_multi_head_attention.html">layer_multi_head_attention</a>()</code> <br><code><a href="layer_multiply.html">layer_multiply</a>()</code> <br><code><a href="layer_normalization.html">layer_normalization</a>()</code> <br><code><a href="layer_permute.html">layer_permute</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_brightness.html">layer_random_brightness</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_contrast.html">layer_random_contrast</a>()</code> <br><code><a href="layer_random_crop.html">layer_random_crop</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_flip.html">layer_random_flip</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_rotation.html">layer_random_rotation</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_random_translation.html">layer_random_translation</a>()</code> <br><code><a href="layer_random_zoom.html">layer_random_zoom</a>()</code> <br><code><a href="layer_repeat_vector.html">layer_repeat_vector</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_reshape.html">layer_reshape</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_rms_normalization.html">layer_rms_normalization</a>()</code> <br><code><a href="layer_rnn.html">layer_rnn</a>()</code> <br><code><a href="layer_separable_conv_1d.html">layer_separable_conv_1d</a>()</code> <br><code><a href="layer_separable_conv_2d.html">layer_separable_conv_2d</a>()</code> <br><code><a href="layer_simple_rnn.html">layer_simple_rnn</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br><code><a href="layer_spatial_dropout_1d.html">layer_spatial_dropout_1d</a>()</code> <br><code><a href="layer_spatial_dropout_2d.html">layer_spatial_dropout_2d</a>()</code> <br><code><a href="layer_spatial_dropout_3d.html">layer_spatial_dropout_3d</a>()</code> <br><code><a href="layer_spectral_normalization.html">layer_spectral_normalization</a>()</code> <br><code><a href="layer_stft_spectrogram.html">layer_stft_spectrogram</a>()</code> <br><code><a href="layer_string_lookup.html">layer_string_lookup</a>()</code> <br><code><a href="layer_subtract.html">layer_subtract</a>()</code> <br><code><a href="layer_text_vectorization.html">layer_text_vectorization</a>()</code> <br><code><a href="layer_tfsm.html">layer_tfsm</a>()</code> <br><code><a href="layer_time_distributed.html">layer_time_distributed</a>()</code> <br><code><a href="layer_torch_module_wrapper.html">layer_torch_module_wrapper</a>()</code> <br><code><a href="layer_unit_normalization.html">layer_unit_normalization</a>()</code> <br><code><a href="layer_upsampling_1d.html">layer_upsampling_1d</a>()</code> <br><code><a href="layer_upsampling_2d.html">layer_upsampling_2d</a>()</code> <br><code><a href="layer_upsampling_3d.html">layer_upsampling_3d</a>()</code> <br><code><a href="layer_zero_padding_1d.html">layer_zero_padding_1d</a>()</code> <br><code><a href="layer_zero_padding_2d.html">layer_zero_padding_2d</a>()</code> <br><code><a href="layer_zero_padding_3d.html">layer_zero_padding_3d</a>()</code> <br><code><a href="rnn_cell_gru.html">rnn_cell_gru</a>()</code> <br><code><a href="rnn_cell_lstm.html">rnn_cell_lstm</a>()</code> <br><code><a href="rnn_cell_simple.html">rnn_cell_simple</a>()</code> <br><code><a href="rnn_cells_stack.html">rnn_cells_stack</a>()</code> <br></p></div>
+</ul><p>Other image preprocessing layers: <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_elastic_transform.html">layer_random_elastic_transform</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br></p>
+<p>Other preprocessing layers: <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_category_encoding.html">layer_category_encoding</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_discretization.html">layer_discretization</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_feature_space.html">layer_feature_space</a>()</code> <br><code><a href="layer_hashed_crossing.html">layer_hashed_crossing</a>()</code> <br><code><a href="layer_hashing.html">layer_hashing</a>()</code> <br><code><a href="layer_integer_lookup.html">layer_integer_lookup</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_mel_spectrogram.html">layer_mel_spectrogram</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_normalization.html">layer_normalization</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_brightness.html">layer_random_brightness</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_contrast.html">layer_random_contrast</a>()</code> <br><code><a href="layer_random_crop.html">layer_random_crop</a>()</code> <br><code><a href="layer_random_elastic_transform.html">layer_random_elastic_transform</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_flip.html">layer_random_flip</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_rotation.html">layer_random_rotation</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_random_translation.html">layer_random_translation</a>()</code> <br><code><a href="layer_random_zoom.html">layer_random_zoom</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br><code><a href="layer_stft_spectrogram.html">layer_stft_spectrogram</a>()</code> <br><code><a href="layer_string_lookup.html">layer_string_lookup</a>()</code> <br><code><a href="layer_text_vectorization.html">layer_text_vectorization</a>()</code> <br></p>
+<p>Other layers: <br><code><a href="Layer.html">Layer</a>()</code> <br><code><a href="layer_activation.html">layer_activation</a>()</code> <br><code><a href="layer_activation_elu.html">layer_activation_elu</a>()</code> <br><code><a href="layer_activation_leaky_relu.html">layer_activation_leaky_relu</a>()</code> <br><code><a href="layer_activation_parametric_relu.html">layer_activation_parametric_relu</a>()</code> <br><code><a href="layer_activation_relu.html">layer_activation_relu</a>()</code> <br><code><a href="layer_activation_softmax.html">layer_activation_softmax</a>()</code> <br><code><a href="layer_activity_regularization.html">layer_activity_regularization</a>()</code> <br><code><a href="layer_add.html">layer_add</a>()</code> <br><code><a href="layer_additive_attention.html">layer_additive_attention</a>()</code> <br><code><a href="layer_alpha_dropout.html">layer_alpha_dropout</a>()</code> <br><code><a href="layer_attention.html">layer_attention</a>()</code> <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_average.html">layer_average</a>()</code> <br><code><a href="layer_average_pooling_1d.html">layer_average_pooling_1d</a>()</code> <br><code><a href="layer_average_pooling_2d.html">layer_average_pooling_2d</a>()</code> <br><code><a href="layer_average_pooling_3d.html">layer_average_pooling_3d</a>()</code> <br><code><a href="layer_batch_normalization.html">layer_batch_normalization</a>()</code> <br><code><a href="layer_bidirectional.html">layer_bidirectional</a>()</code> <br><code><a href="layer_category_encoding.html">layer_category_encoding</a>()</code> <br><code><a href="layer_concatenate.html">layer_concatenate</a>()</code> <br><code><a href="layer_conv_1d.html">layer_conv_1d</a>()</code> <br><code><a href="layer_conv_1d_transpose.html">layer_conv_1d_transpose</a>()</code> <br><code><a href="layer_conv_2d.html">layer_conv_2d</a>()</code> <br><code><a href="layer_conv_2d_transpose.html">layer_conv_2d_transpose</a>()</code> <br><code><a href="layer_conv_3d.html">layer_conv_3d</a>()</code> <br><code><a href="layer_conv_3d_transpose.html">layer_conv_3d_transpose</a>()</code> <br><code><a href="layer_conv_lstm_1d.html">layer_conv_lstm_1d</a>()</code> <br><code><a href="layer_conv_lstm_2d.html">layer_conv_lstm_2d</a>()</code> <br><code><a href="layer_conv_lstm_3d.html">layer_conv_lstm_3d</a>()</code> <br><code><a href="layer_cropping_1d.html">layer_cropping_1d</a>()</code> <br><code><a href="layer_cropping_2d.html">layer_cropping_2d</a>()</code> <br><code><a href="layer_cropping_3d.html">layer_cropping_3d</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_dense.html">layer_dense</a>()</code> <br><code><a href="layer_depthwise_conv_1d.html">layer_depthwise_conv_1d</a>()</code> <br><code><a href="layer_depthwise_conv_2d.html">layer_depthwise_conv_2d</a>()</code> <br><code><a href="layer_discretization.html">layer_discretization</a>()</code> <br><code><a href="layer_dot.html">layer_dot</a>()</code> <br><code><a href="layer_dropout.html">layer_dropout</a>()</code> <br><code><a href="layer_einsum_dense.html">layer_einsum_dense</a>()</code> <br><code><a href="layer_embedding.html">layer_embedding</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_feature_space.html">layer_feature_space</a>()</code> <br><code><a href="layer_flatten.html">layer_flatten</a>()</code> <br><code><a href="layer_flax_module_wrapper.html">layer_flax_module_wrapper</a>()</code> <br><code><a href="layer_gaussian_dropout.html">layer_gaussian_dropout</a>()</code> <br><code><a href="layer_gaussian_noise.html">layer_gaussian_noise</a>()</code> <br><code><a href="layer_global_average_pooling_1d.html">layer_global_average_pooling_1d</a>()</code> <br><code><a href="layer_global_average_pooling_2d.html">layer_global_average_pooling_2d</a>()</code> <br><code><a href="layer_global_average_pooling_3d.html">layer_global_average_pooling_3d</a>()</code> <br><code><a href="layer_global_max_pooling_1d.html">layer_global_max_pooling_1d</a>()</code> <br><code><a href="layer_global_max_pooling_2d.html">layer_global_max_pooling_2d</a>()</code> <br><code><a href="layer_global_max_pooling_3d.html">layer_global_max_pooling_3d</a>()</code> <br><code><a href="layer_group_normalization.html">layer_group_normalization</a>()</code> <br><code><a href="layer_group_query_attention.html">layer_group_query_attention</a>()</code> <br><code><a href="layer_gru.html">layer_gru</a>()</code> <br><code><a href="layer_hashed_crossing.html">layer_hashed_crossing</a>()</code> <br><code><a href="layer_hashing.html">layer_hashing</a>()</code> <br><code><a href="layer_identity.html">layer_identity</a>()</code> <br><code><a href="layer_integer_lookup.html">layer_integer_lookup</a>()</code> <br><code><a href="layer_jax_model_wrapper.html">layer_jax_model_wrapper</a>()</code> <br><code><a href="layer_lambda.html">layer_lambda</a>()</code> <br><code><a href="layer_layer_normalization.html">layer_layer_normalization</a>()</code> <br><code><a href="layer_lstm.html">layer_lstm</a>()</code> <br><code><a href="layer_masking.html">layer_masking</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_max_pooling_1d.html">layer_max_pooling_1d</a>()</code> <br><code><a href="layer_max_pooling_2d.html">layer_max_pooling_2d</a>()</code> <br><code><a href="layer_max_pooling_3d.html">layer_max_pooling_3d</a>()</code> <br><code><a href="layer_maximum.html">layer_maximum</a>()</code> <br><code><a href="layer_mel_spectrogram.html">layer_mel_spectrogram</a>()</code> <br><code><a href="layer_minimum.html">layer_minimum</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_multi_head_attention.html">layer_multi_head_attention</a>()</code> <br><code><a href="layer_multiply.html">layer_multiply</a>()</code> <br><code><a href="layer_normalization.html">layer_normalization</a>()</code> <br><code><a href="layer_permute.html">layer_permute</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_brightness.html">layer_random_brightness</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_contrast.html">layer_random_contrast</a>()</code> <br><code><a href="layer_random_crop.html">layer_random_crop</a>()</code> <br><code><a href="layer_random_elastic_transform.html">layer_random_elastic_transform</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_flip.html">layer_random_flip</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_rotation.html">layer_random_rotation</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_random_translation.html">layer_random_translation</a>()</code> <br><code><a href="layer_random_zoom.html">layer_random_zoom</a>()</code> <br><code><a href="layer_repeat_vector.html">layer_repeat_vector</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_reshape.html">layer_reshape</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_rms_normalization.html">layer_rms_normalization</a>()</code> <br><code><a href="layer_rnn.html">layer_rnn</a>()</code> <br><code><a href="layer_separable_conv_1d.html">layer_separable_conv_1d</a>()</code> <br><code><a href="layer_separable_conv_2d.html">layer_separable_conv_2d</a>()</code> <br><code><a href="layer_simple_rnn.html">layer_simple_rnn</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br><code><a href="layer_spatial_dropout_1d.html">layer_spatial_dropout_1d</a>()</code> <br><code><a href="layer_spatial_dropout_2d.html">layer_spatial_dropout_2d</a>()</code> <br><code><a href="layer_spatial_dropout_3d.html">layer_spatial_dropout_3d</a>()</code> <br><code><a href="layer_spectral_normalization.html">layer_spectral_normalization</a>()</code> <br><code><a href="layer_stft_spectrogram.html">layer_stft_spectrogram</a>()</code> <br><code><a href="layer_string_lookup.html">layer_string_lookup</a>()</code> <br><code><a href="layer_subtract.html">layer_subtract</a>()</code> <br><code><a href="layer_text_vectorization.html">layer_text_vectorization</a>()</code> <br><code><a href="layer_tfsm.html">layer_tfsm</a>()</code> <br><code><a href="layer_time_distributed.html">layer_time_distributed</a>()</code> <br><code><a href="layer_torch_module_wrapper.html">layer_torch_module_wrapper</a>()</code> <br><code><a href="layer_unit_normalization.html">layer_unit_normalization</a>()</code> <br><code><a href="layer_upsampling_1d.html">layer_upsampling_1d</a>()</code> <br><code><a href="layer_upsampling_2d.html">layer_upsampling_2d</a>()</code> <br><code><a href="layer_upsampling_3d.html">layer_upsampling_3d</a>()</code> <br><code><a href="layer_zero_padding_1d.html">layer_zero_padding_1d</a>()</code> <br><code><a href="layer_zero_padding_2d.html">layer_zero_padding_2d</a>()</code> <br><code><a href="layer_zero_padding_3d.html">layer_zero_padding_3d</a>()</code> <br><code><a href="rnn_cell_gru.html">rnn_cell_gru</a>()</code> <br><code><a href="rnn_cell_lstm.html">rnn_cell_lstm</a>()</code> <br><code><a href="rnn_cell_simple.html">rnn_cell_simple</a>()</code> <br><code><a href="rnn_cells_stack.html">rnn_cells_stack</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -149,7 +149,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/layer_concatenate.html b/docs/dev/reference/layer_concatenate.html
index a222a01f6..f1cf9cf43 100644
--- a/docs/dev/reference/layer_concatenate.html
+++ b/docs/dev/reference/layer_concatenate.html
@@ -12,7 +12,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -114,7 +114,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
     <div class="dont-index">
 <ul><li><p><a href="https://keras.io/api/layers/merging_layers/concatenate#concatenate-class" class="external-link">https://keras.io/api/layers/merging_layers/concatenate#concatenate-class</a></p></li>
 </ul><p>Other merging layers: <br><code><a href="layer_add.html">layer_add</a>()</code> <br><code><a href="layer_average.html">layer_average</a>()</code> <br><code><a href="layer_dot.html">layer_dot</a>()</code> <br><code><a href="layer_maximum.html">layer_maximum</a>()</code> <br><code><a href="layer_minimum.html">layer_minimum</a>()</code> <br><code><a href="layer_multiply.html">layer_multiply</a>()</code> <br><code><a href="layer_subtract.html">layer_subtract</a>()</code> <br></p>
-<p>Other layers: <br><code><a href="Layer.html">Layer</a>()</code> <br><code><a href="layer_activation.html">layer_activation</a>()</code> <br><code><a href="layer_activation_elu.html">layer_activation_elu</a>()</code> <br><code><a href="layer_activation_leaky_relu.html">layer_activation_leaky_relu</a>()</code> <br><code><a href="layer_activation_parametric_relu.html">layer_activation_parametric_relu</a>()</code> <br><code><a href="layer_activation_relu.html">layer_activation_relu</a>()</code> <br><code><a href="layer_activation_softmax.html">layer_activation_softmax</a>()</code> <br><code><a href="layer_activity_regularization.html">layer_activity_regularization</a>()</code> <br><code><a href="layer_add.html">layer_add</a>()</code> <br><code><a href="layer_additive_attention.html">layer_additive_attention</a>()</code> <br><code><a href="layer_alpha_dropout.html">layer_alpha_dropout</a>()</code> <br><code><a href="layer_attention.html">layer_attention</a>()</code> <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_average.html">layer_average</a>()</code> <br><code><a href="layer_average_pooling_1d.html">layer_average_pooling_1d</a>()</code> <br><code><a href="layer_average_pooling_2d.html">layer_average_pooling_2d</a>()</code> <br><code><a href="layer_average_pooling_3d.html">layer_average_pooling_3d</a>()</code> <br><code><a href="layer_batch_normalization.html">layer_batch_normalization</a>()</code> <br><code><a href="layer_bidirectional.html">layer_bidirectional</a>()</code> <br><code><a href="layer_category_encoding.html">layer_category_encoding</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_conv_1d.html">layer_conv_1d</a>()</code> <br><code><a href="layer_conv_1d_transpose.html">layer_conv_1d_transpose</a>()</code> <br><code><a href="layer_conv_2d.html">layer_conv_2d</a>()</code> <br><code><a href="layer_conv_2d_transpose.html">layer_conv_2d_transpose</a>()</code> <br><code><a href="layer_conv_3d.html">layer_conv_3d</a>()</code> <br><code><a href="layer_conv_3d_transpose.html">layer_conv_3d_transpose</a>()</code> <br><code><a href="layer_conv_lstm_1d.html">layer_conv_lstm_1d</a>()</code> <br><code><a href="layer_conv_lstm_2d.html">layer_conv_lstm_2d</a>()</code> <br><code><a href="layer_conv_lstm_3d.html">layer_conv_lstm_3d</a>()</code> <br><code><a href="layer_cropping_1d.html">layer_cropping_1d</a>()</code> <br><code><a href="layer_cropping_2d.html">layer_cropping_2d</a>()</code> <br><code><a href="layer_cropping_3d.html">layer_cropping_3d</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_dense.html">layer_dense</a>()</code> <br><code><a href="layer_depthwise_conv_1d.html">layer_depthwise_conv_1d</a>()</code> <br><code><a href="layer_depthwise_conv_2d.html">layer_depthwise_conv_2d</a>()</code> <br><code><a href="layer_discretization.html">layer_discretization</a>()</code> <br><code><a href="layer_dot.html">layer_dot</a>()</code> <br><code><a href="layer_dropout.html">layer_dropout</a>()</code> <br><code><a href="layer_einsum_dense.html">layer_einsum_dense</a>()</code> <br><code><a href="layer_embedding.html">layer_embedding</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_feature_space.html">layer_feature_space</a>()</code> <br><code><a href="layer_flatten.html">layer_flatten</a>()</code> <br><code><a href="layer_flax_module_wrapper.html">layer_flax_module_wrapper</a>()</code> <br><code><a href="layer_gaussian_dropout.html">layer_gaussian_dropout</a>()</code> <br><code><a href="layer_gaussian_noise.html">layer_gaussian_noise</a>()</code> <br><code><a href="layer_global_average_pooling_1d.html">layer_global_average_pooling_1d</a>()</code> <br><code><a href="layer_global_average_pooling_2d.html">layer_global_average_pooling_2d</a>()</code> <br><code><a href="layer_global_average_pooling_3d.html">layer_global_average_pooling_3d</a>()</code> <br><code><a href="layer_global_max_pooling_1d.html">layer_global_max_pooling_1d</a>()</code> <br><code><a href="layer_global_max_pooling_2d.html">layer_global_max_pooling_2d</a>()</code> <br><code><a href="layer_global_max_pooling_3d.html">layer_global_max_pooling_3d</a>()</code> <br><code><a href="layer_group_normalization.html">layer_group_normalization</a>()</code> <br><code><a href="layer_group_query_attention.html">layer_group_query_attention</a>()</code> <br><code><a href="layer_gru.html">layer_gru</a>()</code> <br><code><a href="layer_hashed_crossing.html">layer_hashed_crossing</a>()</code> <br><code><a href="layer_hashing.html">layer_hashing</a>()</code> <br><code><a href="layer_identity.html">layer_identity</a>()</code> <br><code><a href="layer_integer_lookup.html">layer_integer_lookup</a>()</code> <br><code><a href="layer_jax_model_wrapper.html">layer_jax_model_wrapper</a>()</code> <br><code><a href="layer_lambda.html">layer_lambda</a>()</code> <br><code><a href="layer_layer_normalization.html">layer_layer_normalization</a>()</code> <br><code><a href="layer_lstm.html">layer_lstm</a>()</code> <br><code><a href="layer_masking.html">layer_masking</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_max_pooling_1d.html">layer_max_pooling_1d</a>()</code> <br><code><a href="layer_max_pooling_2d.html">layer_max_pooling_2d</a>()</code> <br><code><a href="layer_max_pooling_3d.html">layer_max_pooling_3d</a>()</code> <br><code><a href="layer_maximum.html">layer_maximum</a>()</code> <br><code><a href="layer_mel_spectrogram.html">layer_mel_spectrogram</a>()</code> <br><code><a href="layer_minimum.html">layer_minimum</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_multi_head_attention.html">layer_multi_head_attention</a>()</code> <br><code><a href="layer_multiply.html">layer_multiply</a>()</code> <br><code><a href="layer_normalization.html">layer_normalization</a>()</code> <br><code><a href="layer_permute.html">layer_permute</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_brightness.html">layer_random_brightness</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_contrast.html">layer_random_contrast</a>()</code> <br><code><a href="layer_random_crop.html">layer_random_crop</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_flip.html">layer_random_flip</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_rotation.html">layer_random_rotation</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_random_translation.html">layer_random_translation</a>()</code> <br><code><a href="layer_random_zoom.html">layer_random_zoom</a>()</code> <br><code><a href="layer_repeat_vector.html">layer_repeat_vector</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_reshape.html">layer_reshape</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_rms_normalization.html">layer_rms_normalization</a>()</code> <br><code><a href="layer_rnn.html">layer_rnn</a>()</code> <br><code><a href="layer_separable_conv_1d.html">layer_separable_conv_1d</a>()</code> <br><code><a href="layer_separable_conv_2d.html">layer_separable_conv_2d</a>()</code> <br><code><a href="layer_simple_rnn.html">layer_simple_rnn</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br><code><a href="layer_spatial_dropout_1d.html">layer_spatial_dropout_1d</a>()</code> <br><code><a href="layer_spatial_dropout_2d.html">layer_spatial_dropout_2d</a>()</code> <br><code><a href="layer_spatial_dropout_3d.html">layer_spatial_dropout_3d</a>()</code> <br><code><a href="layer_spectral_normalization.html">layer_spectral_normalization</a>()</code> <br><code><a href="layer_stft_spectrogram.html">layer_stft_spectrogram</a>()</code> <br><code><a href="layer_string_lookup.html">layer_string_lookup</a>()</code> <br><code><a href="layer_subtract.html">layer_subtract</a>()</code> <br><code><a href="layer_text_vectorization.html">layer_text_vectorization</a>()</code> <br><code><a href="layer_tfsm.html">layer_tfsm</a>()</code> <br><code><a href="layer_time_distributed.html">layer_time_distributed</a>()</code> <br><code><a href="layer_torch_module_wrapper.html">layer_torch_module_wrapper</a>()</code> <br><code><a href="layer_unit_normalization.html">layer_unit_normalization</a>()</code> <br><code><a href="layer_upsampling_1d.html">layer_upsampling_1d</a>()</code> <br><code><a href="layer_upsampling_2d.html">layer_upsampling_2d</a>()</code> <br><code><a href="layer_upsampling_3d.html">layer_upsampling_3d</a>()</code> <br><code><a href="layer_zero_padding_1d.html">layer_zero_padding_1d</a>()</code> <br><code><a href="layer_zero_padding_2d.html">layer_zero_padding_2d</a>()</code> <br><code><a href="layer_zero_padding_3d.html">layer_zero_padding_3d</a>()</code> <br><code><a href="rnn_cell_gru.html">rnn_cell_gru</a>()</code> <br><code><a href="rnn_cell_lstm.html">rnn_cell_lstm</a>()</code> <br><code><a href="rnn_cell_simple.html">rnn_cell_simple</a>()</code> <br><code><a href="rnn_cells_stack.html">rnn_cells_stack</a>()</code> <br></p></div>
+<p>Other layers: <br><code><a href="Layer.html">Layer</a>()</code> <br><code><a href="layer_activation.html">layer_activation</a>()</code> <br><code><a href="layer_activation_elu.html">layer_activation_elu</a>()</code> <br><code><a href="layer_activation_leaky_relu.html">layer_activation_leaky_relu</a>()</code> <br><code><a href="layer_activation_parametric_relu.html">layer_activation_parametric_relu</a>()</code> <br><code><a href="layer_activation_relu.html">layer_activation_relu</a>()</code> <br><code><a href="layer_activation_softmax.html">layer_activation_softmax</a>()</code> <br><code><a href="layer_activity_regularization.html">layer_activity_regularization</a>()</code> <br><code><a href="layer_add.html">layer_add</a>()</code> <br><code><a href="layer_additive_attention.html">layer_additive_attention</a>()</code> <br><code><a href="layer_alpha_dropout.html">layer_alpha_dropout</a>()</code> <br><code><a href="layer_attention.html">layer_attention</a>()</code> <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_average.html">layer_average</a>()</code> <br><code><a href="layer_average_pooling_1d.html">layer_average_pooling_1d</a>()</code> <br><code><a href="layer_average_pooling_2d.html">layer_average_pooling_2d</a>()</code> <br><code><a href="layer_average_pooling_3d.html">layer_average_pooling_3d</a>()</code> <br><code><a href="layer_batch_normalization.html">layer_batch_normalization</a>()</code> <br><code><a href="layer_bidirectional.html">layer_bidirectional</a>()</code> <br><code><a href="layer_category_encoding.html">layer_category_encoding</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_conv_1d.html">layer_conv_1d</a>()</code> <br><code><a href="layer_conv_1d_transpose.html">layer_conv_1d_transpose</a>()</code> <br><code><a href="layer_conv_2d.html">layer_conv_2d</a>()</code> <br><code><a href="layer_conv_2d_transpose.html">layer_conv_2d_transpose</a>()</code> <br><code><a href="layer_conv_3d.html">layer_conv_3d</a>()</code> <br><code><a href="layer_conv_3d_transpose.html">layer_conv_3d_transpose</a>()</code> <br><code><a href="layer_conv_lstm_1d.html">layer_conv_lstm_1d</a>()</code> <br><code><a href="layer_conv_lstm_2d.html">layer_conv_lstm_2d</a>()</code> <br><code><a href="layer_conv_lstm_3d.html">layer_conv_lstm_3d</a>()</code> <br><code><a href="layer_cropping_1d.html">layer_cropping_1d</a>()</code> <br><code><a href="layer_cropping_2d.html">layer_cropping_2d</a>()</code> <br><code><a href="layer_cropping_3d.html">layer_cropping_3d</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_dense.html">layer_dense</a>()</code> <br><code><a href="layer_depthwise_conv_1d.html">layer_depthwise_conv_1d</a>()</code> <br><code><a href="layer_depthwise_conv_2d.html">layer_depthwise_conv_2d</a>()</code> <br><code><a href="layer_discretization.html">layer_discretization</a>()</code> <br><code><a href="layer_dot.html">layer_dot</a>()</code> <br><code><a href="layer_dropout.html">layer_dropout</a>()</code> <br><code><a href="layer_einsum_dense.html">layer_einsum_dense</a>()</code> <br><code><a href="layer_embedding.html">layer_embedding</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_feature_space.html">layer_feature_space</a>()</code> <br><code><a href="layer_flatten.html">layer_flatten</a>()</code> <br><code><a href="layer_flax_module_wrapper.html">layer_flax_module_wrapper</a>()</code> <br><code><a href="layer_gaussian_dropout.html">layer_gaussian_dropout</a>()</code> <br><code><a href="layer_gaussian_noise.html">layer_gaussian_noise</a>()</code> <br><code><a href="layer_global_average_pooling_1d.html">layer_global_average_pooling_1d</a>()</code> <br><code><a href="layer_global_average_pooling_2d.html">layer_global_average_pooling_2d</a>()</code> <br><code><a href="layer_global_average_pooling_3d.html">layer_global_average_pooling_3d</a>()</code> <br><code><a href="layer_global_max_pooling_1d.html">layer_global_max_pooling_1d</a>()</code> <br><code><a href="layer_global_max_pooling_2d.html">layer_global_max_pooling_2d</a>()</code> <br><code><a href="layer_global_max_pooling_3d.html">layer_global_max_pooling_3d</a>()</code> <br><code><a href="layer_group_normalization.html">layer_group_normalization</a>()</code> <br><code><a href="layer_group_query_attention.html">layer_group_query_attention</a>()</code> <br><code><a href="layer_gru.html">layer_gru</a>()</code> <br><code><a href="layer_hashed_crossing.html">layer_hashed_crossing</a>()</code> <br><code><a href="layer_hashing.html">layer_hashing</a>()</code> <br><code><a href="layer_identity.html">layer_identity</a>()</code> <br><code><a href="layer_integer_lookup.html">layer_integer_lookup</a>()</code> <br><code><a href="layer_jax_model_wrapper.html">layer_jax_model_wrapper</a>()</code> <br><code><a href="layer_lambda.html">layer_lambda</a>()</code> <br><code><a href="layer_layer_normalization.html">layer_layer_normalization</a>()</code> <br><code><a href="layer_lstm.html">layer_lstm</a>()</code> <br><code><a href="layer_masking.html">layer_masking</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_max_pooling_1d.html">layer_max_pooling_1d</a>()</code> <br><code><a href="layer_max_pooling_2d.html">layer_max_pooling_2d</a>()</code> <br><code><a href="layer_max_pooling_3d.html">layer_max_pooling_3d</a>()</code> <br><code><a href="layer_maximum.html">layer_maximum</a>()</code> <br><code><a href="layer_mel_spectrogram.html">layer_mel_spectrogram</a>()</code> <br><code><a href="layer_minimum.html">layer_minimum</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_multi_head_attention.html">layer_multi_head_attention</a>()</code> <br><code><a href="layer_multiply.html">layer_multiply</a>()</code> <br><code><a href="layer_normalization.html">layer_normalization</a>()</code> <br><code><a href="layer_permute.html">layer_permute</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_brightness.html">layer_random_brightness</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_contrast.html">layer_random_contrast</a>()</code> <br><code><a href="layer_random_crop.html">layer_random_crop</a>()</code> <br><code><a href="layer_random_elastic_transform.html">layer_random_elastic_transform</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_flip.html">layer_random_flip</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_rotation.html">layer_random_rotation</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_random_translation.html">layer_random_translation</a>()</code> <br><code><a href="layer_random_zoom.html">layer_random_zoom</a>()</code> <br><code><a href="layer_repeat_vector.html">layer_repeat_vector</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_reshape.html">layer_reshape</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_rms_normalization.html">layer_rms_normalization</a>()</code> <br><code><a href="layer_rnn.html">layer_rnn</a>()</code> <br><code><a href="layer_separable_conv_1d.html">layer_separable_conv_1d</a>()</code> <br><code><a href="layer_separable_conv_2d.html">layer_separable_conv_2d</a>()</code> <br><code><a href="layer_simple_rnn.html">layer_simple_rnn</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br><code><a href="layer_spatial_dropout_1d.html">layer_spatial_dropout_1d</a>()</code> <br><code><a href="layer_spatial_dropout_2d.html">layer_spatial_dropout_2d</a>()</code> <br><code><a href="layer_spatial_dropout_3d.html">layer_spatial_dropout_3d</a>()</code> <br><code><a href="layer_spectral_normalization.html">layer_spectral_normalization</a>()</code> <br><code><a href="layer_stft_spectrogram.html">layer_stft_spectrogram</a>()</code> <br><code><a href="layer_string_lookup.html">layer_string_lookup</a>()</code> <br><code><a href="layer_subtract.html">layer_subtract</a>()</code> <br><code><a href="layer_text_vectorization.html">layer_text_vectorization</a>()</code> <br><code><a href="layer_tfsm.html">layer_tfsm</a>()</code> <br><code><a href="layer_time_distributed.html">layer_time_distributed</a>()</code> <br><code><a href="layer_torch_module_wrapper.html">layer_torch_module_wrapper</a>()</code> <br><code><a href="layer_unit_normalization.html">layer_unit_normalization</a>()</code> <br><code><a href="layer_upsampling_1d.html">layer_upsampling_1d</a>()</code> <br><code><a href="layer_upsampling_2d.html">layer_upsampling_2d</a>()</code> <br><code><a href="layer_upsampling_3d.html">layer_upsampling_3d</a>()</code> <br><code><a href="layer_zero_padding_1d.html">layer_zero_padding_1d</a>()</code> <br><code><a href="layer_zero_padding_2d.html">layer_zero_padding_2d</a>()</code> <br><code><a href="layer_zero_padding_3d.html">layer_zero_padding_3d</a>()</code> <br><code><a href="rnn_cell_gru.html">rnn_cell_gru</a>()</code> <br><code><a href="rnn_cell_lstm.html">rnn_cell_lstm</a>()</code> <br><code><a href="rnn_cell_simple.html">rnn_cell_simple</a>()</code> <br><code><a href="rnn_cells_stack.html">rnn_cells_stack</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -126,7 +126,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/layer_conv_1d.html b/docs/dev/reference/layer_conv_1d.html
index cefd41b75..8bbbeee9d 100644
--- a/docs/dev/reference/layer_conv_1d.html
+++ b/docs/dev/reference/layer_conv_1d.html
@@ -16,7 +16,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -238,7 +238,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
     <div class="dont-index">
 <ul><li><p><a href="https://keras.io/api/layers/convolution_layers/convolution1d#conv1d-class" class="external-link">https://keras.io/api/layers/convolution_layers/convolution1d#conv1d-class</a></p></li>
 </ul><p>Other convolutional layers: <br><code><a href="layer_conv_1d_transpose.html">layer_conv_1d_transpose</a>()</code> <br><code><a href="layer_conv_2d.html">layer_conv_2d</a>()</code> <br><code><a href="layer_conv_2d_transpose.html">layer_conv_2d_transpose</a>()</code> <br><code><a href="layer_conv_3d.html">layer_conv_3d</a>()</code> <br><code><a href="layer_conv_3d_transpose.html">layer_conv_3d_transpose</a>()</code> <br><code><a href="layer_depthwise_conv_1d.html">layer_depthwise_conv_1d</a>()</code> <br><code><a href="layer_depthwise_conv_2d.html">layer_depthwise_conv_2d</a>()</code> <br><code><a href="layer_separable_conv_1d.html">layer_separable_conv_1d</a>()</code> <br><code><a href="layer_separable_conv_2d.html">layer_separable_conv_2d</a>()</code> <br></p>
-<p>Other layers: <br><code><a href="Layer.html">Layer</a>()</code> <br><code><a href="layer_activation.html">layer_activation</a>()</code> <br><code><a href="layer_activation_elu.html">layer_activation_elu</a>()</code> <br><code><a href="layer_activation_leaky_relu.html">layer_activation_leaky_relu</a>()</code> <br><code><a href="layer_activation_parametric_relu.html">layer_activation_parametric_relu</a>()</code> <br><code><a href="layer_activation_relu.html">layer_activation_relu</a>()</code> <br><code><a href="layer_activation_softmax.html">layer_activation_softmax</a>()</code> <br><code><a href="layer_activity_regularization.html">layer_activity_regularization</a>()</code> <br><code><a href="layer_add.html">layer_add</a>()</code> <br><code><a href="layer_additive_attention.html">layer_additive_attention</a>()</code> <br><code><a href="layer_alpha_dropout.html">layer_alpha_dropout</a>()</code> <br><code><a href="layer_attention.html">layer_attention</a>()</code> <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_average.html">layer_average</a>()</code> <br><code><a href="layer_average_pooling_1d.html">layer_average_pooling_1d</a>()</code> <br><code><a href="layer_average_pooling_2d.html">layer_average_pooling_2d</a>()</code> <br><code><a href="layer_average_pooling_3d.html">layer_average_pooling_3d</a>()</code> <br><code><a href="layer_batch_normalization.html">layer_batch_normalization</a>()</code> <br><code><a href="layer_bidirectional.html">layer_bidirectional</a>()</code> <br><code><a href="layer_category_encoding.html">layer_category_encoding</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_concatenate.html">layer_concatenate</a>()</code> <br><code><a href="layer_conv_1d_transpose.html">layer_conv_1d_transpose</a>()</code> <br><code><a href="layer_conv_2d.html">layer_conv_2d</a>()</code> <br><code><a href="layer_conv_2d_transpose.html">layer_conv_2d_transpose</a>()</code> <br><code><a href="layer_conv_3d.html">layer_conv_3d</a>()</code> <br><code><a href="layer_conv_3d_transpose.html">layer_conv_3d_transpose</a>()</code> <br><code><a href="layer_conv_lstm_1d.html">layer_conv_lstm_1d</a>()</code> <br><code><a href="layer_conv_lstm_2d.html">layer_conv_lstm_2d</a>()</code> <br><code><a href="layer_conv_lstm_3d.html">layer_conv_lstm_3d</a>()</code> <br><code><a href="layer_cropping_1d.html">layer_cropping_1d</a>()</code> <br><code><a href="layer_cropping_2d.html">layer_cropping_2d</a>()</code> <br><code><a href="layer_cropping_3d.html">layer_cropping_3d</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_dense.html">layer_dense</a>()</code> <br><code><a href="layer_depthwise_conv_1d.html">layer_depthwise_conv_1d</a>()</code> <br><code><a href="layer_depthwise_conv_2d.html">layer_depthwise_conv_2d</a>()</code> <br><code><a href="layer_discretization.html">layer_discretization</a>()</code> <br><code><a href="layer_dot.html">layer_dot</a>()</code> <br><code><a href="layer_dropout.html">layer_dropout</a>()</code> <br><code><a href="layer_einsum_dense.html">layer_einsum_dense</a>()</code> <br><code><a href="layer_embedding.html">layer_embedding</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_feature_space.html">layer_feature_space</a>()</code> <br><code><a href="layer_flatten.html">layer_flatten</a>()</code> <br><code><a href="layer_flax_module_wrapper.html">layer_flax_module_wrapper</a>()</code> <br><code><a href="layer_gaussian_dropout.html">layer_gaussian_dropout</a>()</code> <br><code><a href="layer_gaussian_noise.html">layer_gaussian_noise</a>()</code> <br><code><a href="layer_global_average_pooling_1d.html">layer_global_average_pooling_1d</a>()</code> <br><code><a href="layer_global_average_pooling_2d.html">layer_global_average_pooling_2d</a>()</code> <br><code><a href="layer_global_average_pooling_3d.html">layer_global_average_pooling_3d</a>()</code> <br><code><a href="layer_global_max_pooling_1d.html">layer_global_max_pooling_1d</a>()</code> <br><code><a href="layer_global_max_pooling_2d.html">layer_global_max_pooling_2d</a>()</code> <br><code><a href="layer_global_max_pooling_3d.html">layer_global_max_pooling_3d</a>()</code> <br><code><a href="layer_group_normalization.html">layer_group_normalization</a>()</code> <br><code><a href="layer_group_query_attention.html">layer_group_query_attention</a>()</code> <br><code><a href="layer_gru.html">layer_gru</a>()</code> <br><code><a href="layer_hashed_crossing.html">layer_hashed_crossing</a>()</code> <br><code><a href="layer_hashing.html">layer_hashing</a>()</code> <br><code><a href="layer_identity.html">layer_identity</a>()</code> <br><code><a href="layer_integer_lookup.html">layer_integer_lookup</a>()</code> <br><code><a href="layer_jax_model_wrapper.html">layer_jax_model_wrapper</a>()</code> <br><code><a href="layer_lambda.html">layer_lambda</a>()</code> <br><code><a href="layer_layer_normalization.html">layer_layer_normalization</a>()</code> <br><code><a href="layer_lstm.html">layer_lstm</a>()</code> <br><code><a href="layer_masking.html">layer_masking</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_max_pooling_1d.html">layer_max_pooling_1d</a>()</code> <br><code><a href="layer_max_pooling_2d.html">layer_max_pooling_2d</a>()</code> <br><code><a href="layer_max_pooling_3d.html">layer_max_pooling_3d</a>()</code> <br><code><a href="layer_maximum.html">layer_maximum</a>()</code> <br><code><a href="layer_mel_spectrogram.html">layer_mel_spectrogram</a>()</code> <br><code><a href="layer_minimum.html">layer_minimum</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_multi_head_attention.html">layer_multi_head_attention</a>()</code> <br><code><a href="layer_multiply.html">layer_multiply</a>()</code> <br><code><a href="layer_normalization.html">layer_normalization</a>()</code> <br><code><a href="layer_permute.html">layer_permute</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_brightness.html">layer_random_brightness</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_contrast.html">layer_random_contrast</a>()</code> <br><code><a href="layer_random_crop.html">layer_random_crop</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_flip.html">layer_random_flip</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_rotation.html">layer_random_rotation</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_random_translation.html">layer_random_translation</a>()</code> <br><code><a href="layer_random_zoom.html">layer_random_zoom</a>()</code> <br><code><a href="layer_repeat_vector.html">layer_repeat_vector</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_reshape.html">layer_reshape</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_rms_normalization.html">layer_rms_normalization</a>()</code> <br><code><a href="layer_rnn.html">layer_rnn</a>()</code> <br><code><a href="layer_separable_conv_1d.html">layer_separable_conv_1d</a>()</code> <br><code><a href="layer_separable_conv_2d.html">layer_separable_conv_2d</a>()</code> <br><code><a href="layer_simple_rnn.html">layer_simple_rnn</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br><code><a href="layer_spatial_dropout_1d.html">layer_spatial_dropout_1d</a>()</code> <br><code><a href="layer_spatial_dropout_2d.html">layer_spatial_dropout_2d</a>()</code> <br><code><a href="layer_spatial_dropout_3d.html">layer_spatial_dropout_3d</a>()</code> <br><code><a href="layer_spectral_normalization.html">layer_spectral_normalization</a>()</code> <br><code><a href="layer_stft_spectrogram.html">layer_stft_spectrogram</a>()</code> <br><code><a href="layer_string_lookup.html">layer_string_lookup</a>()</code> <br><code><a href="layer_subtract.html">layer_subtract</a>()</code> <br><code><a href="layer_text_vectorization.html">layer_text_vectorization</a>()</code> <br><code><a href="layer_tfsm.html">layer_tfsm</a>()</code> <br><code><a href="layer_time_distributed.html">layer_time_distributed</a>()</code> <br><code><a href="layer_torch_module_wrapper.html">layer_torch_module_wrapper</a>()</code> <br><code><a href="layer_unit_normalization.html">layer_unit_normalization</a>()</code> <br><code><a href="layer_upsampling_1d.html">layer_upsampling_1d</a>()</code> <br><code><a href="layer_upsampling_2d.html">layer_upsampling_2d</a>()</code> <br><code><a href="layer_upsampling_3d.html">layer_upsampling_3d</a>()</code> <br><code><a href="layer_zero_padding_1d.html">layer_zero_padding_1d</a>()</code> <br><code><a href="layer_zero_padding_2d.html">layer_zero_padding_2d</a>()</code> <br><code><a href="layer_zero_padding_3d.html">layer_zero_padding_3d</a>()</code> <br><code><a href="rnn_cell_gru.html">rnn_cell_gru</a>()</code> <br><code><a href="rnn_cell_lstm.html">rnn_cell_lstm</a>()</code> <br><code><a href="rnn_cell_simple.html">rnn_cell_simple</a>()</code> <br><code><a href="rnn_cells_stack.html">rnn_cells_stack</a>()</code> <br></p></div>
+<p>Other layers: <br><code><a href="Layer.html">Layer</a>()</code> <br><code><a href="layer_activation.html">layer_activation</a>()</code> <br><code><a href="layer_activation_elu.html">layer_activation_elu</a>()</code> <br><code><a href="layer_activation_leaky_relu.html">layer_activation_leaky_relu</a>()</code> <br><code><a href="layer_activation_parametric_relu.html">layer_activation_parametric_relu</a>()</code> <br><code><a href="layer_activation_relu.html">layer_activation_relu</a>()</code> <br><code><a href="layer_activation_softmax.html">layer_activation_softmax</a>()</code> <br><code><a href="layer_activity_regularization.html">layer_activity_regularization</a>()</code> <br><code><a href="layer_add.html">layer_add</a>()</code> <br><code><a href="layer_additive_attention.html">layer_additive_attention</a>()</code> <br><code><a href="layer_alpha_dropout.html">layer_alpha_dropout</a>()</code> <br><code><a href="layer_attention.html">layer_attention</a>()</code> <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_average.html">layer_average</a>()</code> <br><code><a href="layer_average_pooling_1d.html">layer_average_pooling_1d</a>()</code> <br><code><a href="layer_average_pooling_2d.html">layer_average_pooling_2d</a>()</code> <br><code><a href="layer_average_pooling_3d.html">layer_average_pooling_3d</a>()</code> <br><code><a href="layer_batch_normalization.html">layer_batch_normalization</a>()</code> <br><code><a href="layer_bidirectional.html">layer_bidirectional</a>()</code> <br><code><a href="layer_category_encoding.html">layer_category_encoding</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_concatenate.html">layer_concatenate</a>()</code> <br><code><a href="layer_conv_1d_transpose.html">layer_conv_1d_transpose</a>()</code> <br><code><a href="layer_conv_2d.html">layer_conv_2d</a>()</code> <br><code><a href="layer_conv_2d_transpose.html">layer_conv_2d_transpose</a>()</code> <br><code><a href="layer_conv_3d.html">layer_conv_3d</a>()</code> <br><code><a href="layer_conv_3d_transpose.html">layer_conv_3d_transpose</a>()</code> <br><code><a href="layer_conv_lstm_1d.html">layer_conv_lstm_1d</a>()</code> <br><code><a href="layer_conv_lstm_2d.html">layer_conv_lstm_2d</a>()</code> <br><code><a href="layer_conv_lstm_3d.html">layer_conv_lstm_3d</a>()</code> <br><code><a href="layer_cropping_1d.html">layer_cropping_1d</a>()</code> <br><code><a href="layer_cropping_2d.html">layer_cropping_2d</a>()</code> <br><code><a href="layer_cropping_3d.html">layer_cropping_3d</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_dense.html">layer_dense</a>()</code> <br><code><a href="layer_depthwise_conv_1d.html">layer_depthwise_conv_1d</a>()</code> <br><code><a href="layer_depthwise_conv_2d.html">layer_depthwise_conv_2d</a>()</code> <br><code><a href="layer_discretization.html">layer_discretization</a>()</code> <br><code><a href="layer_dot.html">layer_dot</a>()</code> <br><code><a href="layer_dropout.html">layer_dropout</a>()</code> <br><code><a href="layer_einsum_dense.html">layer_einsum_dense</a>()</code> <br><code><a href="layer_embedding.html">layer_embedding</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_feature_space.html">layer_feature_space</a>()</code> <br><code><a href="layer_flatten.html">layer_flatten</a>()</code> <br><code><a href="layer_flax_module_wrapper.html">layer_flax_module_wrapper</a>()</code> <br><code><a href="layer_gaussian_dropout.html">layer_gaussian_dropout</a>()</code> <br><code><a href="layer_gaussian_noise.html">layer_gaussian_noise</a>()</code> <br><code><a href="layer_global_average_pooling_1d.html">layer_global_average_pooling_1d</a>()</code> <br><code><a href="layer_global_average_pooling_2d.html">layer_global_average_pooling_2d</a>()</code> <br><code><a href="layer_global_average_pooling_3d.html">layer_global_average_pooling_3d</a>()</code> <br><code><a href="layer_global_max_pooling_1d.html">layer_global_max_pooling_1d</a>()</code> <br><code><a href="layer_global_max_pooling_2d.html">layer_global_max_pooling_2d</a>()</code> <br><code><a href="layer_global_max_pooling_3d.html">layer_global_max_pooling_3d</a>()</code> <br><code><a href="layer_group_normalization.html">layer_group_normalization</a>()</code> <br><code><a href="layer_group_query_attention.html">layer_group_query_attention</a>()</code> <br><code><a href="layer_gru.html">layer_gru</a>()</code> <br><code><a href="layer_hashed_crossing.html">layer_hashed_crossing</a>()</code> <br><code><a href="layer_hashing.html">layer_hashing</a>()</code> <br><code><a href="layer_identity.html">layer_identity</a>()</code> <br><code><a href="layer_integer_lookup.html">layer_integer_lookup</a>()</code> <br><code><a href="layer_jax_model_wrapper.html">layer_jax_model_wrapper</a>()</code> <br><code><a href="layer_lambda.html">layer_lambda</a>()</code> <br><code><a href="layer_layer_normalization.html">layer_layer_normalization</a>()</code> <br><code><a href="layer_lstm.html">layer_lstm</a>()</code> <br><code><a href="layer_masking.html">layer_masking</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_max_pooling_1d.html">layer_max_pooling_1d</a>()</code> <br><code><a href="layer_max_pooling_2d.html">layer_max_pooling_2d</a>()</code> <br><code><a href="layer_max_pooling_3d.html">layer_max_pooling_3d</a>()</code> <br><code><a href="layer_maximum.html">layer_maximum</a>()</code> <br><code><a href="layer_mel_spectrogram.html">layer_mel_spectrogram</a>()</code> <br><code><a href="layer_minimum.html">layer_minimum</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_multi_head_attention.html">layer_multi_head_attention</a>()</code> <br><code><a href="layer_multiply.html">layer_multiply</a>()</code> <br><code><a href="layer_normalization.html">layer_normalization</a>()</code> <br><code><a href="layer_permute.html">layer_permute</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_brightness.html">layer_random_brightness</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_contrast.html">layer_random_contrast</a>()</code> <br><code><a href="layer_random_crop.html">layer_random_crop</a>()</code> <br><code><a href="layer_random_elastic_transform.html">layer_random_elastic_transform</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_flip.html">layer_random_flip</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_rotation.html">layer_random_rotation</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_random_translation.html">layer_random_translation</a>()</code> <br><code><a href="layer_random_zoom.html">layer_random_zoom</a>()</code> <br><code><a href="layer_repeat_vector.html">layer_repeat_vector</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_reshape.html">layer_reshape</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_rms_normalization.html">layer_rms_normalization</a>()</code> <br><code><a href="layer_rnn.html">layer_rnn</a>()</code> <br><code><a href="layer_separable_conv_1d.html">layer_separable_conv_1d</a>()</code> <br><code><a href="layer_separable_conv_2d.html">layer_separable_conv_2d</a>()</code> <br><code><a href="layer_simple_rnn.html">layer_simple_rnn</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br><code><a href="layer_spatial_dropout_1d.html">layer_spatial_dropout_1d</a>()</code> <br><code><a href="layer_spatial_dropout_2d.html">layer_spatial_dropout_2d</a>()</code> <br><code><a href="layer_spatial_dropout_3d.html">layer_spatial_dropout_3d</a>()</code> <br><code><a href="layer_spectral_normalization.html">layer_spectral_normalization</a>()</code> <br><code><a href="layer_stft_spectrogram.html">layer_stft_spectrogram</a>()</code> <br><code><a href="layer_string_lookup.html">layer_string_lookup</a>()</code> <br><code><a href="layer_subtract.html">layer_subtract</a>()</code> <br><code><a href="layer_text_vectorization.html">layer_text_vectorization</a>()</code> <br><code><a href="layer_tfsm.html">layer_tfsm</a>()</code> <br><code><a href="layer_time_distributed.html">layer_time_distributed</a>()</code> <br><code><a href="layer_torch_module_wrapper.html">layer_torch_module_wrapper</a>()</code> <br><code><a href="layer_unit_normalization.html">layer_unit_normalization</a>()</code> <br><code><a href="layer_upsampling_1d.html">layer_upsampling_1d</a>()</code> <br><code><a href="layer_upsampling_2d.html">layer_upsampling_2d</a>()</code> <br><code><a href="layer_upsampling_3d.html">layer_upsampling_3d</a>()</code> <br><code><a href="layer_zero_padding_1d.html">layer_zero_padding_1d</a>()</code> <br><code><a href="layer_zero_padding_2d.html">layer_zero_padding_2d</a>()</code> <br><code><a href="layer_zero_padding_3d.html">layer_zero_padding_3d</a>()</code> <br><code><a href="rnn_cell_gru.html">rnn_cell_gru</a>()</code> <br><code><a href="rnn_cell_lstm.html">rnn_cell_lstm</a>()</code> <br><code><a href="rnn_cell_simple.html">rnn_cell_simple</a>()</code> <br><code><a href="rnn_cells_stack.html">rnn_cells_stack</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -250,7 +250,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/layer_conv_1d_transpose.html b/docs/dev/reference/layer_conv_1d_transpose.html
index 3d422977d..6f6ffd482 100644
--- a/docs/dev/reference/layer_conv_1d_transpose.html
+++ b/docs/dev/reference/layer_conv_1d_transpose.html
@@ -16,7 +16,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -77,6 +77,7 @@ <h2 id="ref-usage">Usage<a class="anchor" aria-label="anchor" href="#ref-usage">
 <span>  <span class="va">kernel_size</span>,</span>
 <span>  strides <span class="op">=</span> <span class="fl">1L</span>,</span>
 <span>  padding <span class="op">=</span> <span class="st">"valid"</span>,</span>
+<span>  output_padding <span class="op">=</span> <span class="cn">NULL</span>,</span>
 <span>  data_format <span class="op">=</span> <span class="cn">NULL</span>,</span>
 <span>  dilation_rate <span class="op">=</span> <span class="fl">1L</span>,</span>
 <span>  activation <span class="op">=</span> <span class="cn">NULL</span>,</span>
@@ -123,6 +124,11 @@ <h2 id="arguments">Arguments<a class="anchor" aria-label="anchor" href="#argumen
 height/width dimension as the input.</p></dd>
 
 
+<dt id="arg-output-padding">output_padding<a class="anchor" aria-label="anchor" href="#arg-output-padding"></a></dt>
+<dd><p>Scalar integer. Amount of padding to add to the output length. Must be less
+than the stride. When <code>NULL</code> (default) the output size is inferred.</p></dd>
+
+
 <dt id="arg-data-format">data_format<a class="anchor" aria-label="anchor" href="#arg-data-format"></a></dt>
 <dd><p>string, either <code>"channels_last"</code> or <code>"channels_first"</code>.
 The ordering of the dimensions in the inputs. <code>"channels_last"</code>
@@ -134,8 +140,9 @@ <h2 id="arguments">Arguments<a class="anchor" aria-label="anchor" href="#argumen
 
 
 <dt id="arg-dilation-rate">dilation_rate<a class="anchor" aria-label="anchor" href="#arg-dilation-rate"></a></dt>
-<dd><p>int or list of 1 integers, specifying the dilation
-rate to use for dilated transposed convolution.</p></dd>
+<dd><p>Scalar integer. Specifies the dilation rate. Values other
+than 1 currently require <code>strides = 1</code> and rates greater than 1 are not
+supported.</p></dd>
 
 
 <dt id="arg-activation">activation<a class="anchor" aria-label="anchor" href="#arg-activation"></a></dt>
@@ -230,7 +237,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
     <div class="dont-index">
 <ul><li><p><a href="https://keras.io/api/layers/convolution_layers/convolution1d_transpose#conv1dtranspose-class" class="external-link">https://keras.io/api/layers/convolution_layers/convolution1d_transpose#conv1dtranspose-class</a></p></li>
 </ul><p>Other convolutional layers: <br><code><a href="layer_conv_1d.html">layer_conv_1d</a>()</code> <br><code><a href="layer_conv_2d.html">layer_conv_2d</a>()</code> <br><code><a href="layer_conv_2d_transpose.html">layer_conv_2d_transpose</a>()</code> <br><code><a href="layer_conv_3d.html">layer_conv_3d</a>()</code> <br><code><a href="layer_conv_3d_transpose.html">layer_conv_3d_transpose</a>()</code> <br><code><a href="layer_depthwise_conv_1d.html">layer_depthwise_conv_1d</a>()</code> <br><code><a href="layer_depthwise_conv_2d.html">layer_depthwise_conv_2d</a>()</code> <br><code><a href="layer_separable_conv_1d.html">layer_separable_conv_1d</a>()</code> <br><code><a href="layer_separable_conv_2d.html">layer_separable_conv_2d</a>()</code> <br></p>
-<p>Other layers: <br><code><a href="Layer.html">Layer</a>()</code> <br><code><a href="layer_activation.html">layer_activation</a>()</code> <br><code><a href="layer_activation_elu.html">layer_activation_elu</a>()</code> <br><code><a href="layer_activation_leaky_relu.html">layer_activation_leaky_relu</a>()</code> <br><code><a href="layer_activation_parametric_relu.html">layer_activation_parametric_relu</a>()</code> <br><code><a href="layer_activation_relu.html">layer_activation_relu</a>()</code> <br><code><a href="layer_activation_softmax.html">layer_activation_softmax</a>()</code> <br><code><a href="layer_activity_regularization.html">layer_activity_regularization</a>()</code> <br><code><a href="layer_add.html">layer_add</a>()</code> <br><code><a href="layer_additive_attention.html">layer_additive_attention</a>()</code> <br><code><a href="layer_alpha_dropout.html">layer_alpha_dropout</a>()</code> <br><code><a href="layer_attention.html">layer_attention</a>()</code> <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_average.html">layer_average</a>()</code> <br><code><a href="layer_average_pooling_1d.html">layer_average_pooling_1d</a>()</code> <br><code><a href="layer_average_pooling_2d.html">layer_average_pooling_2d</a>()</code> <br><code><a href="layer_average_pooling_3d.html">layer_average_pooling_3d</a>()</code> <br><code><a href="layer_batch_normalization.html">layer_batch_normalization</a>()</code> <br><code><a href="layer_bidirectional.html">layer_bidirectional</a>()</code> <br><code><a href="layer_category_encoding.html">layer_category_encoding</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_concatenate.html">layer_concatenate</a>()</code> <br><code><a href="layer_conv_1d.html">layer_conv_1d</a>()</code> <br><code><a href="layer_conv_2d.html">layer_conv_2d</a>()</code> <br><code><a href="layer_conv_2d_transpose.html">layer_conv_2d_transpose</a>()</code> <br><code><a href="layer_conv_3d.html">layer_conv_3d</a>()</code> <br><code><a href="layer_conv_3d_transpose.html">layer_conv_3d_transpose</a>()</code> <br><code><a href="layer_conv_lstm_1d.html">layer_conv_lstm_1d</a>()</code> <br><code><a href="layer_conv_lstm_2d.html">layer_conv_lstm_2d</a>()</code> <br><code><a href="layer_conv_lstm_3d.html">layer_conv_lstm_3d</a>()</code> <br><code><a href="layer_cropping_1d.html">layer_cropping_1d</a>()</code> <br><code><a href="layer_cropping_2d.html">layer_cropping_2d</a>()</code> <br><code><a href="layer_cropping_3d.html">layer_cropping_3d</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_dense.html">layer_dense</a>()</code> <br><code><a href="layer_depthwise_conv_1d.html">layer_depthwise_conv_1d</a>()</code> <br><code><a href="layer_depthwise_conv_2d.html">layer_depthwise_conv_2d</a>()</code> <br><code><a href="layer_discretization.html">layer_discretization</a>()</code> <br><code><a href="layer_dot.html">layer_dot</a>()</code> <br><code><a href="layer_dropout.html">layer_dropout</a>()</code> <br><code><a href="layer_einsum_dense.html">layer_einsum_dense</a>()</code> <br><code><a href="layer_embedding.html">layer_embedding</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_feature_space.html">layer_feature_space</a>()</code> <br><code><a href="layer_flatten.html">layer_flatten</a>()</code> <br><code><a href="layer_flax_module_wrapper.html">layer_flax_module_wrapper</a>()</code> <br><code><a href="layer_gaussian_dropout.html">layer_gaussian_dropout</a>()</code> <br><code><a href="layer_gaussian_noise.html">layer_gaussian_noise</a>()</code> <br><code><a href="layer_global_average_pooling_1d.html">layer_global_average_pooling_1d</a>()</code> <br><code><a href="layer_global_average_pooling_2d.html">layer_global_average_pooling_2d</a>()</code> <br><code><a href="layer_global_average_pooling_3d.html">layer_global_average_pooling_3d</a>()</code> <br><code><a href="layer_global_max_pooling_1d.html">layer_global_max_pooling_1d</a>()</code> <br><code><a href="layer_global_max_pooling_2d.html">layer_global_max_pooling_2d</a>()</code> <br><code><a href="layer_global_max_pooling_3d.html">layer_global_max_pooling_3d</a>()</code> <br><code><a href="layer_group_normalization.html">layer_group_normalization</a>()</code> <br><code><a href="layer_group_query_attention.html">layer_group_query_attention</a>()</code> <br><code><a href="layer_gru.html">layer_gru</a>()</code> <br><code><a href="layer_hashed_crossing.html">layer_hashed_crossing</a>()</code> <br><code><a href="layer_hashing.html">layer_hashing</a>()</code> <br><code><a href="layer_identity.html">layer_identity</a>()</code> <br><code><a href="layer_integer_lookup.html">layer_integer_lookup</a>()</code> <br><code><a href="layer_jax_model_wrapper.html">layer_jax_model_wrapper</a>()</code> <br><code><a href="layer_lambda.html">layer_lambda</a>()</code> <br><code><a href="layer_layer_normalization.html">layer_layer_normalization</a>()</code> <br><code><a href="layer_lstm.html">layer_lstm</a>()</code> <br><code><a href="layer_masking.html">layer_masking</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_max_pooling_1d.html">layer_max_pooling_1d</a>()</code> <br><code><a href="layer_max_pooling_2d.html">layer_max_pooling_2d</a>()</code> <br><code><a href="layer_max_pooling_3d.html">layer_max_pooling_3d</a>()</code> <br><code><a href="layer_maximum.html">layer_maximum</a>()</code> <br><code><a href="layer_mel_spectrogram.html">layer_mel_spectrogram</a>()</code> <br><code><a href="layer_minimum.html">layer_minimum</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_multi_head_attention.html">layer_multi_head_attention</a>()</code> <br><code><a href="layer_multiply.html">layer_multiply</a>()</code> <br><code><a href="layer_normalization.html">layer_normalization</a>()</code> <br><code><a href="layer_permute.html">layer_permute</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_brightness.html">layer_random_brightness</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_contrast.html">layer_random_contrast</a>()</code> <br><code><a href="layer_random_crop.html">layer_random_crop</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_flip.html">layer_random_flip</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_rotation.html">layer_random_rotation</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_random_translation.html">layer_random_translation</a>()</code> <br><code><a href="layer_random_zoom.html">layer_random_zoom</a>()</code> <br><code><a href="layer_repeat_vector.html">layer_repeat_vector</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_reshape.html">layer_reshape</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_rms_normalization.html">layer_rms_normalization</a>()</code> <br><code><a href="layer_rnn.html">layer_rnn</a>()</code> <br><code><a href="layer_separable_conv_1d.html">layer_separable_conv_1d</a>()</code> <br><code><a href="layer_separable_conv_2d.html">layer_separable_conv_2d</a>()</code> <br><code><a href="layer_simple_rnn.html">layer_simple_rnn</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br><code><a href="layer_spatial_dropout_1d.html">layer_spatial_dropout_1d</a>()</code> <br><code><a href="layer_spatial_dropout_2d.html">layer_spatial_dropout_2d</a>()</code> <br><code><a href="layer_spatial_dropout_3d.html">layer_spatial_dropout_3d</a>()</code> <br><code><a href="layer_spectral_normalization.html">layer_spectral_normalization</a>()</code> <br><code><a href="layer_stft_spectrogram.html">layer_stft_spectrogram</a>()</code> <br><code><a href="layer_string_lookup.html">layer_string_lookup</a>()</code> <br><code><a href="layer_subtract.html">layer_subtract</a>()</code> <br><code><a href="layer_text_vectorization.html">layer_text_vectorization</a>()</code> <br><code><a href="layer_tfsm.html">layer_tfsm</a>()</code> <br><code><a href="layer_time_distributed.html">layer_time_distributed</a>()</code> <br><code><a href="layer_torch_module_wrapper.html">layer_torch_module_wrapper</a>()</code> <br><code><a href="layer_unit_normalization.html">layer_unit_normalization</a>()</code> <br><code><a href="layer_upsampling_1d.html">layer_upsampling_1d</a>()</code> <br><code><a href="layer_upsampling_2d.html">layer_upsampling_2d</a>()</code> <br><code><a href="layer_upsampling_3d.html">layer_upsampling_3d</a>()</code> <br><code><a href="layer_zero_padding_1d.html">layer_zero_padding_1d</a>()</code> <br><code><a href="layer_zero_padding_2d.html">layer_zero_padding_2d</a>()</code> <br><code><a href="layer_zero_padding_3d.html">layer_zero_padding_3d</a>()</code> <br><code><a href="rnn_cell_gru.html">rnn_cell_gru</a>()</code> <br><code><a href="rnn_cell_lstm.html">rnn_cell_lstm</a>()</code> <br><code><a href="rnn_cell_simple.html">rnn_cell_simple</a>()</code> <br><code><a href="rnn_cells_stack.html">rnn_cells_stack</a>()</code> <br></p></div>
+<p>Other layers: <br><code><a href="Layer.html">Layer</a>()</code> <br><code><a href="layer_activation.html">layer_activation</a>()</code> <br><code><a href="layer_activation_elu.html">layer_activation_elu</a>()</code> <br><code><a href="layer_activation_leaky_relu.html">layer_activation_leaky_relu</a>()</code> <br><code><a href="layer_activation_parametric_relu.html">layer_activation_parametric_relu</a>()</code> <br><code><a href="layer_activation_relu.html">layer_activation_relu</a>()</code> <br><code><a href="layer_activation_softmax.html">layer_activation_softmax</a>()</code> <br><code><a href="layer_activity_regularization.html">layer_activity_regularization</a>()</code> <br><code><a href="layer_add.html">layer_add</a>()</code> <br><code><a href="layer_additive_attention.html">layer_additive_attention</a>()</code> <br><code><a href="layer_alpha_dropout.html">layer_alpha_dropout</a>()</code> <br><code><a href="layer_attention.html">layer_attention</a>()</code> <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_average.html">layer_average</a>()</code> <br><code><a href="layer_average_pooling_1d.html">layer_average_pooling_1d</a>()</code> <br><code><a href="layer_average_pooling_2d.html">layer_average_pooling_2d</a>()</code> <br><code><a href="layer_average_pooling_3d.html">layer_average_pooling_3d</a>()</code> <br><code><a href="layer_batch_normalization.html">layer_batch_normalization</a>()</code> <br><code><a href="layer_bidirectional.html">layer_bidirectional</a>()</code> <br><code><a href="layer_category_encoding.html">layer_category_encoding</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_concatenate.html">layer_concatenate</a>()</code> <br><code><a href="layer_conv_1d.html">layer_conv_1d</a>()</code> <br><code><a href="layer_conv_2d.html">layer_conv_2d</a>()</code> <br><code><a href="layer_conv_2d_transpose.html">layer_conv_2d_transpose</a>()</code> <br><code><a href="layer_conv_3d.html">layer_conv_3d</a>()</code> <br><code><a href="layer_conv_3d_transpose.html">layer_conv_3d_transpose</a>()</code> <br><code><a href="layer_conv_lstm_1d.html">layer_conv_lstm_1d</a>()</code> <br><code><a href="layer_conv_lstm_2d.html">layer_conv_lstm_2d</a>()</code> <br><code><a href="layer_conv_lstm_3d.html">layer_conv_lstm_3d</a>()</code> <br><code><a href="layer_cropping_1d.html">layer_cropping_1d</a>()</code> <br><code><a href="layer_cropping_2d.html">layer_cropping_2d</a>()</code> <br><code><a href="layer_cropping_3d.html">layer_cropping_3d</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_dense.html">layer_dense</a>()</code> <br><code><a href="layer_depthwise_conv_1d.html">layer_depthwise_conv_1d</a>()</code> <br><code><a href="layer_depthwise_conv_2d.html">layer_depthwise_conv_2d</a>()</code> <br><code><a href="layer_discretization.html">layer_discretization</a>()</code> <br><code><a href="layer_dot.html">layer_dot</a>()</code> <br><code><a href="layer_dropout.html">layer_dropout</a>()</code> <br><code><a href="layer_einsum_dense.html">layer_einsum_dense</a>()</code> <br><code><a href="layer_embedding.html">layer_embedding</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_feature_space.html">layer_feature_space</a>()</code> <br><code><a href="layer_flatten.html">layer_flatten</a>()</code> <br><code><a href="layer_flax_module_wrapper.html">layer_flax_module_wrapper</a>()</code> <br><code><a href="layer_gaussian_dropout.html">layer_gaussian_dropout</a>()</code> <br><code><a href="layer_gaussian_noise.html">layer_gaussian_noise</a>()</code> <br><code><a href="layer_global_average_pooling_1d.html">layer_global_average_pooling_1d</a>()</code> <br><code><a href="layer_global_average_pooling_2d.html">layer_global_average_pooling_2d</a>()</code> <br><code><a href="layer_global_average_pooling_3d.html">layer_global_average_pooling_3d</a>()</code> <br><code><a href="layer_global_max_pooling_1d.html">layer_global_max_pooling_1d</a>()</code> <br><code><a href="layer_global_max_pooling_2d.html">layer_global_max_pooling_2d</a>()</code> <br><code><a href="layer_global_max_pooling_3d.html">layer_global_max_pooling_3d</a>()</code> <br><code><a href="layer_group_normalization.html">layer_group_normalization</a>()</code> <br><code><a href="layer_group_query_attention.html">layer_group_query_attention</a>()</code> <br><code><a href="layer_gru.html">layer_gru</a>()</code> <br><code><a href="layer_hashed_crossing.html">layer_hashed_crossing</a>()</code> <br><code><a href="layer_hashing.html">layer_hashing</a>()</code> <br><code><a href="layer_identity.html">layer_identity</a>()</code> <br><code><a href="layer_integer_lookup.html">layer_integer_lookup</a>()</code> <br><code><a href="layer_jax_model_wrapper.html">layer_jax_model_wrapper</a>()</code> <br><code><a href="layer_lambda.html">layer_lambda</a>()</code> <br><code><a href="layer_layer_normalization.html">layer_layer_normalization</a>()</code> <br><code><a href="layer_lstm.html">layer_lstm</a>()</code> <br><code><a href="layer_masking.html">layer_masking</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_max_pooling_1d.html">layer_max_pooling_1d</a>()</code> <br><code><a href="layer_max_pooling_2d.html">layer_max_pooling_2d</a>()</code> <br><code><a href="layer_max_pooling_3d.html">layer_max_pooling_3d</a>()</code> <br><code><a href="layer_maximum.html">layer_maximum</a>()</code> <br><code><a href="layer_mel_spectrogram.html">layer_mel_spectrogram</a>()</code> <br><code><a href="layer_minimum.html">layer_minimum</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_multi_head_attention.html">layer_multi_head_attention</a>()</code> <br><code><a href="layer_multiply.html">layer_multiply</a>()</code> <br><code><a href="layer_normalization.html">layer_normalization</a>()</code> <br><code><a href="layer_permute.html">layer_permute</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_brightness.html">layer_random_brightness</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_contrast.html">layer_random_contrast</a>()</code> <br><code><a href="layer_random_crop.html">layer_random_crop</a>()</code> <br><code><a href="layer_random_elastic_transform.html">layer_random_elastic_transform</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_flip.html">layer_random_flip</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_rotation.html">layer_random_rotation</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_random_translation.html">layer_random_translation</a>()</code> <br><code><a href="layer_random_zoom.html">layer_random_zoom</a>()</code> <br><code><a href="layer_repeat_vector.html">layer_repeat_vector</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_reshape.html">layer_reshape</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_rms_normalization.html">layer_rms_normalization</a>()</code> <br><code><a href="layer_rnn.html">layer_rnn</a>()</code> <br><code><a href="layer_separable_conv_1d.html">layer_separable_conv_1d</a>()</code> <br><code><a href="layer_separable_conv_2d.html">layer_separable_conv_2d</a>()</code> <br><code><a href="layer_simple_rnn.html">layer_simple_rnn</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br><code><a href="layer_spatial_dropout_1d.html">layer_spatial_dropout_1d</a>()</code> <br><code><a href="layer_spatial_dropout_2d.html">layer_spatial_dropout_2d</a>()</code> <br><code><a href="layer_spatial_dropout_3d.html">layer_spatial_dropout_3d</a>()</code> <br><code><a href="layer_spectral_normalization.html">layer_spectral_normalization</a>()</code> <br><code><a href="layer_stft_spectrogram.html">layer_stft_spectrogram</a>()</code> <br><code><a href="layer_string_lookup.html">layer_string_lookup</a>()</code> <br><code><a href="layer_subtract.html">layer_subtract</a>()</code> <br><code><a href="layer_text_vectorization.html">layer_text_vectorization</a>()</code> <br><code><a href="layer_tfsm.html">layer_tfsm</a>()</code> <br><code><a href="layer_time_distributed.html">layer_time_distributed</a>()</code> <br><code><a href="layer_torch_module_wrapper.html">layer_torch_module_wrapper</a>()</code> <br><code><a href="layer_unit_normalization.html">layer_unit_normalization</a>()</code> <br><code><a href="layer_upsampling_1d.html">layer_upsampling_1d</a>()</code> <br><code><a href="layer_upsampling_2d.html">layer_upsampling_2d</a>()</code> <br><code><a href="layer_upsampling_3d.html">layer_upsampling_3d</a>()</code> <br><code><a href="layer_zero_padding_1d.html">layer_zero_padding_1d</a>()</code> <br><code><a href="layer_zero_padding_2d.html">layer_zero_padding_2d</a>()</code> <br><code><a href="layer_zero_padding_3d.html">layer_zero_padding_3d</a>()</code> <br><code><a href="rnn_cell_gru.html">rnn_cell_gru</a>()</code> <br><code><a href="rnn_cell_lstm.html">rnn_cell_lstm</a>()</code> <br><code><a href="rnn_cell_simple.html">rnn_cell_simple</a>()</code> <br><code><a href="rnn_cells_stack.html">rnn_cells_stack</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -242,7 +249,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/layer_conv_2d.html b/docs/dev/reference/layer_conv_2d.html
index 8060db64e..ff4140118 100644
--- a/docs/dev/reference/layer_conv_2d.html
+++ b/docs/dev/reference/layer_conv_2d.html
@@ -32,7 +32,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -220,6 +220,12 @@ <h2 id="arguments">Arguments<a class="anchor" aria-label="anchor" href="#argumen
 <dt id="arg--">...<a class="anchor" aria-label="anchor" href="#arg--"></a></dt>
 <dd><p>For forward/backward compatability.</p></dd>
 
+
+<dt id="arg-output-padding">output_padding<a class="anchor" aria-label="anchor" href="#arg-output-padding"></a></dt>
+<dd><p>Scalar integer or vector of two integers. Amount of padding to add to the output
+depth, height, and width. Each element must be smaller than the corresponding
+stride. When <code>NULL</code> (default) the output size is inferred.</p></dd>
+
 </dl></div>
     <div class="section level2">
     <h2 id="value">Value<a class="anchor" aria-label="anchor" href="#value"></a></h2>
@@ -258,7 +264,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
     <div class="dont-index">
 <ul><li><p><a href="https://keras.io/api/layers/convolution_layers/convolution2d#conv2d-class" class="external-link">https://keras.io/api/layers/convolution_layers/convolution2d#conv2d-class</a></p></li>
 </ul><p>Other convolutional layers: <br><code><a href="layer_conv_1d.html">layer_conv_1d</a>()</code> <br><code><a href="layer_conv_1d_transpose.html">layer_conv_1d_transpose</a>()</code> <br><code><a href="layer_conv_2d_transpose.html">layer_conv_2d_transpose</a>()</code> <br><code><a href="layer_conv_3d.html">layer_conv_3d</a>()</code> <br><code><a href="layer_conv_3d_transpose.html">layer_conv_3d_transpose</a>()</code> <br><code><a href="layer_depthwise_conv_1d.html">layer_depthwise_conv_1d</a>()</code> <br><code><a href="layer_depthwise_conv_2d.html">layer_depthwise_conv_2d</a>()</code> <br><code><a href="layer_separable_conv_1d.html">layer_separable_conv_1d</a>()</code> <br><code><a href="layer_separable_conv_2d.html">layer_separable_conv_2d</a>()</code> <br></p>
-<p>Other layers: <br><code><a href="Layer.html">Layer</a>()</code> <br><code><a href="layer_activation.html">layer_activation</a>()</code> <br><code><a href="layer_activation_elu.html">layer_activation_elu</a>()</code> <br><code><a href="layer_activation_leaky_relu.html">layer_activation_leaky_relu</a>()</code> <br><code><a href="layer_activation_parametric_relu.html">layer_activation_parametric_relu</a>()</code> <br><code><a href="layer_activation_relu.html">layer_activation_relu</a>()</code> <br><code><a href="layer_activation_softmax.html">layer_activation_softmax</a>()</code> <br><code><a href="layer_activity_regularization.html">layer_activity_regularization</a>()</code> <br><code><a href="layer_add.html">layer_add</a>()</code> <br><code><a href="layer_additive_attention.html">layer_additive_attention</a>()</code> <br><code><a href="layer_alpha_dropout.html">layer_alpha_dropout</a>()</code> <br><code><a href="layer_attention.html">layer_attention</a>()</code> <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_average.html">layer_average</a>()</code> <br><code><a href="layer_average_pooling_1d.html">layer_average_pooling_1d</a>()</code> <br><code><a href="layer_average_pooling_2d.html">layer_average_pooling_2d</a>()</code> <br><code><a href="layer_average_pooling_3d.html">layer_average_pooling_3d</a>()</code> <br><code><a href="layer_batch_normalization.html">layer_batch_normalization</a>()</code> <br><code><a href="layer_bidirectional.html">layer_bidirectional</a>()</code> <br><code><a href="layer_category_encoding.html">layer_category_encoding</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_concatenate.html">layer_concatenate</a>()</code> <br><code><a href="layer_conv_1d.html">layer_conv_1d</a>()</code> <br><code><a href="layer_conv_1d_transpose.html">layer_conv_1d_transpose</a>()</code> <br><code><a href="layer_conv_2d_transpose.html">layer_conv_2d_transpose</a>()</code> <br><code><a href="layer_conv_3d.html">layer_conv_3d</a>()</code> <br><code><a href="layer_conv_3d_transpose.html">layer_conv_3d_transpose</a>()</code> <br><code><a href="layer_conv_lstm_1d.html">layer_conv_lstm_1d</a>()</code> <br><code><a href="layer_conv_lstm_2d.html">layer_conv_lstm_2d</a>()</code> <br><code><a href="layer_conv_lstm_3d.html">layer_conv_lstm_3d</a>()</code> <br><code><a href="layer_cropping_1d.html">layer_cropping_1d</a>()</code> <br><code><a href="layer_cropping_2d.html">layer_cropping_2d</a>()</code> <br><code><a href="layer_cropping_3d.html">layer_cropping_3d</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_dense.html">layer_dense</a>()</code> <br><code><a href="layer_depthwise_conv_1d.html">layer_depthwise_conv_1d</a>()</code> <br><code><a href="layer_depthwise_conv_2d.html">layer_depthwise_conv_2d</a>()</code> <br><code><a href="layer_discretization.html">layer_discretization</a>()</code> <br><code><a href="layer_dot.html">layer_dot</a>()</code> <br><code><a href="layer_dropout.html">layer_dropout</a>()</code> <br><code><a href="layer_einsum_dense.html">layer_einsum_dense</a>()</code> <br><code><a href="layer_embedding.html">layer_embedding</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_feature_space.html">layer_feature_space</a>()</code> <br><code><a href="layer_flatten.html">layer_flatten</a>()</code> <br><code><a href="layer_flax_module_wrapper.html">layer_flax_module_wrapper</a>()</code> <br><code><a href="layer_gaussian_dropout.html">layer_gaussian_dropout</a>()</code> <br><code><a href="layer_gaussian_noise.html">layer_gaussian_noise</a>()</code> <br><code><a href="layer_global_average_pooling_1d.html">layer_global_average_pooling_1d</a>()</code> <br><code><a href="layer_global_average_pooling_2d.html">layer_global_average_pooling_2d</a>()</code> <br><code><a href="layer_global_average_pooling_3d.html">layer_global_average_pooling_3d</a>()</code> <br><code><a href="layer_global_max_pooling_1d.html">layer_global_max_pooling_1d</a>()</code> <br><code><a href="layer_global_max_pooling_2d.html">layer_global_max_pooling_2d</a>()</code> <br><code><a href="layer_global_max_pooling_3d.html">layer_global_max_pooling_3d</a>()</code> <br><code><a href="layer_group_normalization.html">layer_group_normalization</a>()</code> <br><code><a href="layer_group_query_attention.html">layer_group_query_attention</a>()</code> <br><code><a href="layer_gru.html">layer_gru</a>()</code> <br><code><a href="layer_hashed_crossing.html">layer_hashed_crossing</a>()</code> <br><code><a href="layer_hashing.html">layer_hashing</a>()</code> <br><code><a href="layer_identity.html">layer_identity</a>()</code> <br><code><a href="layer_integer_lookup.html">layer_integer_lookup</a>()</code> <br><code><a href="layer_jax_model_wrapper.html">layer_jax_model_wrapper</a>()</code> <br><code><a href="layer_lambda.html">layer_lambda</a>()</code> <br><code><a href="layer_layer_normalization.html">layer_layer_normalization</a>()</code> <br><code><a href="layer_lstm.html">layer_lstm</a>()</code> <br><code><a href="layer_masking.html">layer_masking</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_max_pooling_1d.html">layer_max_pooling_1d</a>()</code> <br><code><a href="layer_max_pooling_2d.html">layer_max_pooling_2d</a>()</code> <br><code><a href="layer_max_pooling_3d.html">layer_max_pooling_3d</a>()</code> <br><code><a href="layer_maximum.html">layer_maximum</a>()</code> <br><code><a href="layer_mel_spectrogram.html">layer_mel_spectrogram</a>()</code> <br><code><a href="layer_minimum.html">layer_minimum</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_multi_head_attention.html">layer_multi_head_attention</a>()</code> <br><code><a href="layer_multiply.html">layer_multiply</a>()</code> <br><code><a href="layer_normalization.html">layer_normalization</a>()</code> <br><code><a href="layer_permute.html">layer_permute</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_brightness.html">layer_random_brightness</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_contrast.html">layer_random_contrast</a>()</code> <br><code><a href="layer_random_crop.html">layer_random_crop</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_flip.html">layer_random_flip</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_rotation.html">layer_random_rotation</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_random_translation.html">layer_random_translation</a>()</code> <br><code><a href="layer_random_zoom.html">layer_random_zoom</a>()</code> <br><code><a href="layer_repeat_vector.html">layer_repeat_vector</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_reshape.html">layer_reshape</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_rms_normalization.html">layer_rms_normalization</a>()</code> <br><code><a href="layer_rnn.html">layer_rnn</a>()</code> <br><code><a href="layer_separable_conv_1d.html">layer_separable_conv_1d</a>()</code> <br><code><a href="layer_separable_conv_2d.html">layer_separable_conv_2d</a>()</code> <br><code><a href="layer_simple_rnn.html">layer_simple_rnn</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br><code><a href="layer_spatial_dropout_1d.html">layer_spatial_dropout_1d</a>()</code> <br><code><a href="layer_spatial_dropout_2d.html">layer_spatial_dropout_2d</a>()</code> <br><code><a href="layer_spatial_dropout_3d.html">layer_spatial_dropout_3d</a>()</code> <br><code><a href="layer_spectral_normalization.html">layer_spectral_normalization</a>()</code> <br><code><a href="layer_stft_spectrogram.html">layer_stft_spectrogram</a>()</code> <br><code><a href="layer_string_lookup.html">layer_string_lookup</a>()</code> <br><code><a href="layer_subtract.html">layer_subtract</a>()</code> <br><code><a href="layer_text_vectorization.html">layer_text_vectorization</a>()</code> <br><code><a href="layer_tfsm.html">layer_tfsm</a>()</code> <br><code><a href="layer_time_distributed.html">layer_time_distributed</a>()</code> <br><code><a href="layer_torch_module_wrapper.html">layer_torch_module_wrapper</a>()</code> <br><code><a href="layer_unit_normalization.html">layer_unit_normalization</a>()</code> <br><code><a href="layer_upsampling_1d.html">layer_upsampling_1d</a>()</code> <br><code><a href="layer_upsampling_2d.html">layer_upsampling_2d</a>()</code> <br><code><a href="layer_upsampling_3d.html">layer_upsampling_3d</a>()</code> <br><code><a href="layer_zero_padding_1d.html">layer_zero_padding_1d</a>()</code> <br><code><a href="layer_zero_padding_2d.html">layer_zero_padding_2d</a>()</code> <br><code><a href="layer_zero_padding_3d.html">layer_zero_padding_3d</a>()</code> <br><code><a href="rnn_cell_gru.html">rnn_cell_gru</a>()</code> <br><code><a href="rnn_cell_lstm.html">rnn_cell_lstm</a>()</code> <br><code><a href="rnn_cell_simple.html">rnn_cell_simple</a>()</code> <br><code><a href="rnn_cells_stack.html">rnn_cells_stack</a>()</code> <br></p></div>
+<p>Other layers: <br><code><a href="Layer.html">Layer</a>()</code> <br><code><a href="layer_activation.html">layer_activation</a>()</code> <br><code><a href="layer_activation_elu.html">layer_activation_elu</a>()</code> <br><code><a href="layer_activation_leaky_relu.html">layer_activation_leaky_relu</a>()</code> <br><code><a href="layer_activation_parametric_relu.html">layer_activation_parametric_relu</a>()</code> <br><code><a href="layer_activation_relu.html">layer_activation_relu</a>()</code> <br><code><a href="layer_activation_softmax.html">layer_activation_softmax</a>()</code> <br><code><a href="layer_activity_regularization.html">layer_activity_regularization</a>()</code> <br><code><a href="layer_add.html">layer_add</a>()</code> <br><code><a href="layer_additive_attention.html">layer_additive_attention</a>()</code> <br><code><a href="layer_alpha_dropout.html">layer_alpha_dropout</a>()</code> <br><code><a href="layer_attention.html">layer_attention</a>()</code> <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_average.html">layer_average</a>()</code> <br><code><a href="layer_average_pooling_1d.html">layer_average_pooling_1d</a>()</code> <br><code><a href="layer_average_pooling_2d.html">layer_average_pooling_2d</a>()</code> <br><code><a href="layer_average_pooling_3d.html">layer_average_pooling_3d</a>()</code> <br><code><a href="layer_batch_normalization.html">layer_batch_normalization</a>()</code> <br><code><a href="layer_bidirectional.html">layer_bidirectional</a>()</code> <br><code><a href="layer_category_encoding.html">layer_category_encoding</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_concatenate.html">layer_concatenate</a>()</code> <br><code><a href="layer_conv_1d.html">layer_conv_1d</a>()</code> <br><code><a href="layer_conv_1d_transpose.html">layer_conv_1d_transpose</a>()</code> <br><code><a href="layer_conv_2d_transpose.html">layer_conv_2d_transpose</a>()</code> <br><code><a href="layer_conv_3d.html">layer_conv_3d</a>()</code> <br><code><a href="layer_conv_3d_transpose.html">layer_conv_3d_transpose</a>()</code> <br><code><a href="layer_conv_lstm_1d.html">layer_conv_lstm_1d</a>()</code> <br><code><a href="layer_conv_lstm_2d.html">layer_conv_lstm_2d</a>()</code> <br><code><a href="layer_conv_lstm_3d.html">layer_conv_lstm_3d</a>()</code> <br><code><a href="layer_cropping_1d.html">layer_cropping_1d</a>()</code> <br><code><a href="layer_cropping_2d.html">layer_cropping_2d</a>()</code> <br><code><a href="layer_cropping_3d.html">layer_cropping_3d</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_dense.html">layer_dense</a>()</code> <br><code><a href="layer_depthwise_conv_1d.html">layer_depthwise_conv_1d</a>()</code> <br><code><a href="layer_depthwise_conv_2d.html">layer_depthwise_conv_2d</a>()</code> <br><code><a href="layer_discretization.html">layer_discretization</a>()</code> <br><code><a href="layer_dot.html">layer_dot</a>()</code> <br><code><a href="layer_dropout.html">layer_dropout</a>()</code> <br><code><a href="layer_einsum_dense.html">layer_einsum_dense</a>()</code> <br><code><a href="layer_embedding.html">layer_embedding</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_feature_space.html">layer_feature_space</a>()</code> <br><code><a href="layer_flatten.html">layer_flatten</a>()</code> <br><code><a href="layer_flax_module_wrapper.html">layer_flax_module_wrapper</a>()</code> <br><code><a href="layer_gaussian_dropout.html">layer_gaussian_dropout</a>()</code> <br><code><a href="layer_gaussian_noise.html">layer_gaussian_noise</a>()</code> <br><code><a href="layer_global_average_pooling_1d.html">layer_global_average_pooling_1d</a>()</code> <br><code><a href="layer_global_average_pooling_2d.html">layer_global_average_pooling_2d</a>()</code> <br><code><a href="layer_global_average_pooling_3d.html">layer_global_average_pooling_3d</a>()</code> <br><code><a href="layer_global_max_pooling_1d.html">layer_global_max_pooling_1d</a>()</code> <br><code><a href="layer_global_max_pooling_2d.html">layer_global_max_pooling_2d</a>()</code> <br><code><a href="layer_global_max_pooling_3d.html">layer_global_max_pooling_3d</a>()</code> <br><code><a href="layer_group_normalization.html">layer_group_normalization</a>()</code> <br><code><a href="layer_group_query_attention.html">layer_group_query_attention</a>()</code> <br><code><a href="layer_gru.html">layer_gru</a>()</code> <br><code><a href="layer_hashed_crossing.html">layer_hashed_crossing</a>()</code> <br><code><a href="layer_hashing.html">layer_hashing</a>()</code> <br><code><a href="layer_identity.html">layer_identity</a>()</code> <br><code><a href="layer_integer_lookup.html">layer_integer_lookup</a>()</code> <br><code><a href="layer_jax_model_wrapper.html">layer_jax_model_wrapper</a>()</code> <br><code><a href="layer_lambda.html">layer_lambda</a>()</code> <br><code><a href="layer_layer_normalization.html">layer_layer_normalization</a>()</code> <br><code><a href="layer_lstm.html">layer_lstm</a>()</code> <br><code><a href="layer_masking.html">layer_masking</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_max_pooling_1d.html">layer_max_pooling_1d</a>()</code> <br><code><a href="layer_max_pooling_2d.html">layer_max_pooling_2d</a>()</code> <br><code><a href="layer_max_pooling_3d.html">layer_max_pooling_3d</a>()</code> <br><code><a href="layer_maximum.html">layer_maximum</a>()</code> <br><code><a href="layer_mel_spectrogram.html">layer_mel_spectrogram</a>()</code> <br><code><a href="layer_minimum.html">layer_minimum</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_multi_head_attention.html">layer_multi_head_attention</a>()</code> <br><code><a href="layer_multiply.html">layer_multiply</a>()</code> <br><code><a href="layer_normalization.html">layer_normalization</a>()</code> <br><code><a href="layer_permute.html">layer_permute</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_brightness.html">layer_random_brightness</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_contrast.html">layer_random_contrast</a>()</code> <br><code><a href="layer_random_crop.html">layer_random_crop</a>()</code> <br><code><a href="layer_random_elastic_transform.html">layer_random_elastic_transform</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_flip.html">layer_random_flip</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_rotation.html">layer_random_rotation</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_random_translation.html">layer_random_translation</a>()</code> <br><code><a href="layer_random_zoom.html">layer_random_zoom</a>()</code> <br><code><a href="layer_repeat_vector.html">layer_repeat_vector</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_reshape.html">layer_reshape</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_rms_normalization.html">layer_rms_normalization</a>()</code> <br><code><a href="layer_rnn.html">layer_rnn</a>()</code> <br><code><a href="layer_separable_conv_1d.html">layer_separable_conv_1d</a>()</code> <br><code><a href="layer_separable_conv_2d.html">layer_separable_conv_2d</a>()</code> <br><code><a href="layer_simple_rnn.html">layer_simple_rnn</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br><code><a href="layer_spatial_dropout_1d.html">layer_spatial_dropout_1d</a>()</code> <br><code><a href="layer_spatial_dropout_2d.html">layer_spatial_dropout_2d</a>()</code> <br><code><a href="layer_spatial_dropout_3d.html">layer_spatial_dropout_3d</a>()</code> <br><code><a href="layer_spectral_normalization.html">layer_spectral_normalization</a>()</code> <br><code><a href="layer_stft_spectrogram.html">layer_stft_spectrogram</a>()</code> <br><code><a href="layer_string_lookup.html">layer_string_lookup</a>()</code> <br><code><a href="layer_subtract.html">layer_subtract</a>()</code> <br><code><a href="layer_text_vectorization.html">layer_text_vectorization</a>()</code> <br><code><a href="layer_tfsm.html">layer_tfsm</a>()</code> <br><code><a href="layer_time_distributed.html">layer_time_distributed</a>()</code> <br><code><a href="layer_torch_module_wrapper.html">layer_torch_module_wrapper</a>()</code> <br><code><a href="layer_unit_normalization.html">layer_unit_normalization</a>()</code> <br><code><a href="layer_upsampling_1d.html">layer_upsampling_1d</a>()</code> <br><code><a href="layer_upsampling_2d.html">layer_upsampling_2d</a>()</code> <br><code><a href="layer_upsampling_3d.html">layer_upsampling_3d</a>()</code> <br><code><a href="layer_zero_padding_1d.html">layer_zero_padding_1d</a>()</code> <br><code><a href="layer_zero_padding_2d.html">layer_zero_padding_2d</a>()</code> <br><code><a href="layer_zero_padding_3d.html">layer_zero_padding_3d</a>()</code> <br><code><a href="rnn_cell_gru.html">rnn_cell_gru</a>()</code> <br><code><a href="rnn_cell_lstm.html">rnn_cell_lstm</a>()</code> <br><code><a href="rnn_cell_simple.html">rnn_cell_simple</a>()</code> <br><code><a href="rnn_cells_stack.html">rnn_cells_stack</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -270,7 +276,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/layer_conv_2d_transpose.html b/docs/dev/reference/layer_conv_2d_transpose.html
index 4ad5ad88d..e3e1c0dc0 100644
--- a/docs/dev/reference/layer_conv_2d_transpose.html
+++ b/docs/dev/reference/layer_conv_2d_transpose.html
@@ -16,7 +16,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -77,6 +77,7 @@ <h2 id="ref-usage">Usage<a class="anchor" aria-label="anchor" href="#ref-usage">
 <span>  <span class="va">kernel_size</span>,</span>
 <span>  strides <span class="op">=</span> <span class="fu"><a href="https://rdrr.io/r/base/list.html" class="external-link">list</a></span><span class="op">(</span><span class="fl">1L</span>, <span class="fl">1L</span><span class="op">)</span>,</span>
 <span>  padding <span class="op">=</span> <span class="st">"valid"</span>,</span>
+<span>  output_padding <span class="op">=</span> <span class="cn">NULL</span>,</span>
 <span>  data_format <span class="op">=</span> <span class="cn">NULL</span>,</span>
 <span>  dilation_rate <span class="op">=</span> <span class="fu"><a href="https://rdrr.io/r/base/list.html" class="external-link">list</a></span><span class="op">(</span><span class="fl">1L</span>, <span class="fl">1L</span><span class="op">)</span>,</span>
 <span>  activation <span class="op">=</span> <span class="cn">NULL</span>,</span>
@@ -136,8 +137,9 @@ <h2 id="arguments">Arguments<a class="anchor" aria-label="anchor" href="#argumen
 
 
 <dt id="arg-dilation-rate">dilation_rate<a class="anchor" aria-label="anchor" href="#arg-dilation-rate"></a></dt>
-<dd><p>int or list of 1 integers, specifying the dilation
-rate to use for dilated transposed convolution.</p></dd>
+<dd><p>Scalar integer or vector of 2 integers specifying the dilation rate. Values
+other than 1 require <code>strides = 1</code>; different rates per dimension are not
+supported.</p></dd>
 
 
 <dt id="arg-activation">activation<a class="anchor" aria-label="anchor" href="#arg-activation"></a></dt>
@@ -233,7 +235,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
     <div class="dont-index">
 <ul><li><p><a href="https://keras.io/api/layers/convolution_layers/convolution2d_transpose#conv2dtranspose-class" class="external-link">https://keras.io/api/layers/convolution_layers/convolution2d_transpose#conv2dtranspose-class</a></p></li>
 </ul><p>Other convolutional layers: <br><code><a href="layer_conv_1d.html">layer_conv_1d</a>()</code> <br><code><a href="layer_conv_1d_transpose.html">layer_conv_1d_transpose</a>()</code> <br><code><a href="layer_conv_2d.html">layer_conv_2d</a>()</code> <br><code><a href="layer_conv_3d.html">layer_conv_3d</a>()</code> <br><code><a href="layer_conv_3d_transpose.html">layer_conv_3d_transpose</a>()</code> <br><code><a href="layer_depthwise_conv_1d.html">layer_depthwise_conv_1d</a>()</code> <br><code><a href="layer_depthwise_conv_2d.html">layer_depthwise_conv_2d</a>()</code> <br><code><a href="layer_separable_conv_1d.html">layer_separable_conv_1d</a>()</code> <br><code><a href="layer_separable_conv_2d.html">layer_separable_conv_2d</a>()</code> <br></p>
-<p>Other layers: <br><code><a href="Layer.html">Layer</a>()</code> <br><code><a href="layer_activation.html">layer_activation</a>()</code> <br><code><a href="layer_activation_elu.html">layer_activation_elu</a>()</code> <br><code><a href="layer_activation_leaky_relu.html">layer_activation_leaky_relu</a>()</code> <br><code><a href="layer_activation_parametric_relu.html">layer_activation_parametric_relu</a>()</code> <br><code><a href="layer_activation_relu.html">layer_activation_relu</a>()</code> <br><code><a href="layer_activation_softmax.html">layer_activation_softmax</a>()</code> <br><code><a href="layer_activity_regularization.html">layer_activity_regularization</a>()</code> <br><code><a href="layer_add.html">layer_add</a>()</code> <br><code><a href="layer_additive_attention.html">layer_additive_attention</a>()</code> <br><code><a href="layer_alpha_dropout.html">layer_alpha_dropout</a>()</code> <br><code><a href="layer_attention.html">layer_attention</a>()</code> <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_average.html">layer_average</a>()</code> <br><code><a href="layer_average_pooling_1d.html">layer_average_pooling_1d</a>()</code> <br><code><a href="layer_average_pooling_2d.html">layer_average_pooling_2d</a>()</code> <br><code><a href="layer_average_pooling_3d.html">layer_average_pooling_3d</a>()</code> <br><code><a href="layer_batch_normalization.html">layer_batch_normalization</a>()</code> <br><code><a href="layer_bidirectional.html">layer_bidirectional</a>()</code> <br><code><a href="layer_category_encoding.html">layer_category_encoding</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_concatenate.html">layer_concatenate</a>()</code> <br><code><a href="layer_conv_1d.html">layer_conv_1d</a>()</code> <br><code><a href="layer_conv_1d_transpose.html">layer_conv_1d_transpose</a>()</code> <br><code><a href="layer_conv_2d.html">layer_conv_2d</a>()</code> <br><code><a href="layer_conv_3d.html">layer_conv_3d</a>()</code> <br><code><a href="layer_conv_3d_transpose.html">layer_conv_3d_transpose</a>()</code> <br><code><a href="layer_conv_lstm_1d.html">layer_conv_lstm_1d</a>()</code> <br><code><a href="layer_conv_lstm_2d.html">layer_conv_lstm_2d</a>()</code> <br><code><a href="layer_conv_lstm_3d.html">layer_conv_lstm_3d</a>()</code> <br><code><a href="layer_cropping_1d.html">layer_cropping_1d</a>()</code> <br><code><a href="layer_cropping_2d.html">layer_cropping_2d</a>()</code> <br><code><a href="layer_cropping_3d.html">layer_cropping_3d</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_dense.html">layer_dense</a>()</code> <br><code><a href="layer_depthwise_conv_1d.html">layer_depthwise_conv_1d</a>()</code> <br><code><a href="layer_depthwise_conv_2d.html">layer_depthwise_conv_2d</a>()</code> <br><code><a href="layer_discretization.html">layer_discretization</a>()</code> <br><code><a href="layer_dot.html">layer_dot</a>()</code> <br><code><a href="layer_dropout.html">layer_dropout</a>()</code> <br><code><a href="layer_einsum_dense.html">layer_einsum_dense</a>()</code> <br><code><a href="layer_embedding.html">layer_embedding</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_feature_space.html">layer_feature_space</a>()</code> <br><code><a href="layer_flatten.html">layer_flatten</a>()</code> <br><code><a href="layer_flax_module_wrapper.html">layer_flax_module_wrapper</a>()</code> <br><code><a href="layer_gaussian_dropout.html">layer_gaussian_dropout</a>()</code> <br><code><a href="layer_gaussian_noise.html">layer_gaussian_noise</a>()</code> <br><code><a href="layer_global_average_pooling_1d.html">layer_global_average_pooling_1d</a>()</code> <br><code><a href="layer_global_average_pooling_2d.html">layer_global_average_pooling_2d</a>()</code> <br><code><a href="layer_global_average_pooling_3d.html">layer_global_average_pooling_3d</a>()</code> <br><code><a href="layer_global_max_pooling_1d.html">layer_global_max_pooling_1d</a>()</code> <br><code><a href="layer_global_max_pooling_2d.html">layer_global_max_pooling_2d</a>()</code> <br><code><a href="layer_global_max_pooling_3d.html">layer_global_max_pooling_3d</a>()</code> <br><code><a href="layer_group_normalization.html">layer_group_normalization</a>()</code> <br><code><a href="layer_group_query_attention.html">layer_group_query_attention</a>()</code> <br><code><a href="layer_gru.html">layer_gru</a>()</code> <br><code><a href="layer_hashed_crossing.html">layer_hashed_crossing</a>()</code> <br><code><a href="layer_hashing.html">layer_hashing</a>()</code> <br><code><a href="layer_identity.html">layer_identity</a>()</code> <br><code><a href="layer_integer_lookup.html">layer_integer_lookup</a>()</code> <br><code><a href="layer_jax_model_wrapper.html">layer_jax_model_wrapper</a>()</code> <br><code><a href="layer_lambda.html">layer_lambda</a>()</code> <br><code><a href="layer_layer_normalization.html">layer_layer_normalization</a>()</code> <br><code><a href="layer_lstm.html">layer_lstm</a>()</code> <br><code><a href="layer_masking.html">layer_masking</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_max_pooling_1d.html">layer_max_pooling_1d</a>()</code> <br><code><a href="layer_max_pooling_2d.html">layer_max_pooling_2d</a>()</code> <br><code><a href="layer_max_pooling_3d.html">layer_max_pooling_3d</a>()</code> <br><code><a href="layer_maximum.html">layer_maximum</a>()</code> <br><code><a href="layer_mel_spectrogram.html">layer_mel_spectrogram</a>()</code> <br><code><a href="layer_minimum.html">layer_minimum</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_multi_head_attention.html">layer_multi_head_attention</a>()</code> <br><code><a href="layer_multiply.html">layer_multiply</a>()</code> <br><code><a href="layer_normalization.html">layer_normalization</a>()</code> <br><code><a href="layer_permute.html">layer_permute</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_brightness.html">layer_random_brightness</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_contrast.html">layer_random_contrast</a>()</code> <br><code><a href="layer_random_crop.html">layer_random_crop</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_flip.html">layer_random_flip</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_rotation.html">layer_random_rotation</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_random_translation.html">layer_random_translation</a>()</code> <br><code><a href="layer_random_zoom.html">layer_random_zoom</a>()</code> <br><code><a href="layer_repeat_vector.html">layer_repeat_vector</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_reshape.html">layer_reshape</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_rms_normalization.html">layer_rms_normalization</a>()</code> <br><code><a href="layer_rnn.html">layer_rnn</a>()</code> <br><code><a href="layer_separable_conv_1d.html">layer_separable_conv_1d</a>()</code> <br><code><a href="layer_separable_conv_2d.html">layer_separable_conv_2d</a>()</code> <br><code><a href="layer_simple_rnn.html">layer_simple_rnn</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br><code><a href="layer_spatial_dropout_1d.html">layer_spatial_dropout_1d</a>()</code> <br><code><a href="layer_spatial_dropout_2d.html">layer_spatial_dropout_2d</a>()</code> <br><code><a href="layer_spatial_dropout_3d.html">layer_spatial_dropout_3d</a>()</code> <br><code><a href="layer_spectral_normalization.html">layer_spectral_normalization</a>()</code> <br><code><a href="layer_stft_spectrogram.html">layer_stft_spectrogram</a>()</code> <br><code><a href="layer_string_lookup.html">layer_string_lookup</a>()</code> <br><code><a href="layer_subtract.html">layer_subtract</a>()</code> <br><code><a href="layer_text_vectorization.html">layer_text_vectorization</a>()</code> <br><code><a href="layer_tfsm.html">layer_tfsm</a>()</code> <br><code><a href="layer_time_distributed.html">layer_time_distributed</a>()</code> <br><code><a href="layer_torch_module_wrapper.html">layer_torch_module_wrapper</a>()</code> <br><code><a href="layer_unit_normalization.html">layer_unit_normalization</a>()</code> <br><code><a href="layer_upsampling_1d.html">layer_upsampling_1d</a>()</code> <br><code><a href="layer_upsampling_2d.html">layer_upsampling_2d</a>()</code> <br><code><a href="layer_upsampling_3d.html">layer_upsampling_3d</a>()</code> <br><code><a href="layer_zero_padding_1d.html">layer_zero_padding_1d</a>()</code> <br><code><a href="layer_zero_padding_2d.html">layer_zero_padding_2d</a>()</code> <br><code><a href="layer_zero_padding_3d.html">layer_zero_padding_3d</a>()</code> <br><code><a href="rnn_cell_gru.html">rnn_cell_gru</a>()</code> <br><code><a href="rnn_cell_lstm.html">rnn_cell_lstm</a>()</code> <br><code><a href="rnn_cell_simple.html">rnn_cell_simple</a>()</code> <br><code><a href="rnn_cells_stack.html">rnn_cells_stack</a>()</code> <br></p></div>
+<p>Other layers: <br><code><a href="Layer.html">Layer</a>()</code> <br><code><a href="layer_activation.html">layer_activation</a>()</code> <br><code><a href="layer_activation_elu.html">layer_activation_elu</a>()</code> <br><code><a href="layer_activation_leaky_relu.html">layer_activation_leaky_relu</a>()</code> <br><code><a href="layer_activation_parametric_relu.html">layer_activation_parametric_relu</a>()</code> <br><code><a href="layer_activation_relu.html">layer_activation_relu</a>()</code> <br><code><a href="layer_activation_softmax.html">layer_activation_softmax</a>()</code> <br><code><a href="layer_activity_regularization.html">layer_activity_regularization</a>()</code> <br><code><a href="layer_add.html">layer_add</a>()</code> <br><code><a href="layer_additive_attention.html">layer_additive_attention</a>()</code> <br><code><a href="layer_alpha_dropout.html">layer_alpha_dropout</a>()</code> <br><code><a href="layer_attention.html">layer_attention</a>()</code> <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_average.html">layer_average</a>()</code> <br><code><a href="layer_average_pooling_1d.html">layer_average_pooling_1d</a>()</code> <br><code><a href="layer_average_pooling_2d.html">layer_average_pooling_2d</a>()</code> <br><code><a href="layer_average_pooling_3d.html">layer_average_pooling_3d</a>()</code> <br><code><a href="layer_batch_normalization.html">layer_batch_normalization</a>()</code> <br><code><a href="layer_bidirectional.html">layer_bidirectional</a>()</code> <br><code><a href="layer_category_encoding.html">layer_category_encoding</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_concatenate.html">layer_concatenate</a>()</code> <br><code><a href="layer_conv_1d.html">layer_conv_1d</a>()</code> <br><code><a href="layer_conv_1d_transpose.html">layer_conv_1d_transpose</a>()</code> <br><code><a href="layer_conv_2d.html">layer_conv_2d</a>()</code> <br><code><a href="layer_conv_3d.html">layer_conv_3d</a>()</code> <br><code><a href="layer_conv_3d_transpose.html">layer_conv_3d_transpose</a>()</code> <br><code><a href="layer_conv_lstm_1d.html">layer_conv_lstm_1d</a>()</code> <br><code><a href="layer_conv_lstm_2d.html">layer_conv_lstm_2d</a>()</code> <br><code><a href="layer_conv_lstm_3d.html">layer_conv_lstm_3d</a>()</code> <br><code><a href="layer_cropping_1d.html">layer_cropping_1d</a>()</code> <br><code><a href="layer_cropping_2d.html">layer_cropping_2d</a>()</code> <br><code><a href="layer_cropping_3d.html">layer_cropping_3d</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_dense.html">layer_dense</a>()</code> <br><code><a href="layer_depthwise_conv_1d.html">layer_depthwise_conv_1d</a>()</code> <br><code><a href="layer_depthwise_conv_2d.html">layer_depthwise_conv_2d</a>()</code> <br><code><a href="layer_discretization.html">layer_discretization</a>()</code> <br><code><a href="layer_dot.html">layer_dot</a>()</code> <br><code><a href="layer_dropout.html">layer_dropout</a>()</code> <br><code><a href="layer_einsum_dense.html">layer_einsum_dense</a>()</code> <br><code><a href="layer_embedding.html">layer_embedding</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_feature_space.html">layer_feature_space</a>()</code> <br><code><a href="layer_flatten.html">layer_flatten</a>()</code> <br><code><a href="layer_flax_module_wrapper.html">layer_flax_module_wrapper</a>()</code> <br><code><a href="layer_gaussian_dropout.html">layer_gaussian_dropout</a>()</code> <br><code><a href="layer_gaussian_noise.html">layer_gaussian_noise</a>()</code> <br><code><a href="layer_global_average_pooling_1d.html">layer_global_average_pooling_1d</a>()</code> <br><code><a href="layer_global_average_pooling_2d.html">layer_global_average_pooling_2d</a>()</code> <br><code><a href="layer_global_average_pooling_3d.html">layer_global_average_pooling_3d</a>()</code> <br><code><a href="layer_global_max_pooling_1d.html">layer_global_max_pooling_1d</a>()</code> <br><code><a href="layer_global_max_pooling_2d.html">layer_global_max_pooling_2d</a>()</code> <br><code><a href="layer_global_max_pooling_3d.html">layer_global_max_pooling_3d</a>()</code> <br><code><a href="layer_group_normalization.html">layer_group_normalization</a>()</code> <br><code><a href="layer_group_query_attention.html">layer_group_query_attention</a>()</code> <br><code><a href="layer_gru.html">layer_gru</a>()</code> <br><code><a href="layer_hashed_crossing.html">layer_hashed_crossing</a>()</code> <br><code><a href="layer_hashing.html">layer_hashing</a>()</code> <br><code><a href="layer_identity.html">layer_identity</a>()</code> <br><code><a href="layer_integer_lookup.html">layer_integer_lookup</a>()</code> <br><code><a href="layer_jax_model_wrapper.html">layer_jax_model_wrapper</a>()</code> <br><code><a href="layer_lambda.html">layer_lambda</a>()</code> <br><code><a href="layer_layer_normalization.html">layer_layer_normalization</a>()</code> <br><code><a href="layer_lstm.html">layer_lstm</a>()</code> <br><code><a href="layer_masking.html">layer_masking</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_max_pooling_1d.html">layer_max_pooling_1d</a>()</code> <br><code><a href="layer_max_pooling_2d.html">layer_max_pooling_2d</a>()</code> <br><code><a href="layer_max_pooling_3d.html">layer_max_pooling_3d</a>()</code> <br><code><a href="layer_maximum.html">layer_maximum</a>()</code> <br><code><a href="layer_mel_spectrogram.html">layer_mel_spectrogram</a>()</code> <br><code><a href="layer_minimum.html">layer_minimum</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_multi_head_attention.html">layer_multi_head_attention</a>()</code> <br><code><a href="layer_multiply.html">layer_multiply</a>()</code> <br><code><a href="layer_normalization.html">layer_normalization</a>()</code> <br><code><a href="layer_permute.html">layer_permute</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_brightness.html">layer_random_brightness</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_contrast.html">layer_random_contrast</a>()</code> <br><code><a href="layer_random_crop.html">layer_random_crop</a>()</code> <br><code><a href="layer_random_elastic_transform.html">layer_random_elastic_transform</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_flip.html">layer_random_flip</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_rotation.html">layer_random_rotation</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_random_translation.html">layer_random_translation</a>()</code> <br><code><a href="layer_random_zoom.html">layer_random_zoom</a>()</code> <br><code><a href="layer_repeat_vector.html">layer_repeat_vector</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_reshape.html">layer_reshape</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_rms_normalization.html">layer_rms_normalization</a>()</code> <br><code><a href="layer_rnn.html">layer_rnn</a>()</code> <br><code><a href="layer_separable_conv_1d.html">layer_separable_conv_1d</a>()</code> <br><code><a href="layer_separable_conv_2d.html">layer_separable_conv_2d</a>()</code> <br><code><a href="layer_simple_rnn.html">layer_simple_rnn</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br><code><a href="layer_spatial_dropout_1d.html">layer_spatial_dropout_1d</a>()</code> <br><code><a href="layer_spatial_dropout_2d.html">layer_spatial_dropout_2d</a>()</code> <br><code><a href="layer_spatial_dropout_3d.html">layer_spatial_dropout_3d</a>()</code> <br><code><a href="layer_spectral_normalization.html">layer_spectral_normalization</a>()</code> <br><code><a href="layer_stft_spectrogram.html">layer_stft_spectrogram</a>()</code> <br><code><a href="layer_string_lookup.html">layer_string_lookup</a>()</code> <br><code><a href="layer_subtract.html">layer_subtract</a>()</code> <br><code><a href="layer_text_vectorization.html">layer_text_vectorization</a>()</code> <br><code><a href="layer_tfsm.html">layer_tfsm</a>()</code> <br><code><a href="layer_time_distributed.html">layer_time_distributed</a>()</code> <br><code><a href="layer_torch_module_wrapper.html">layer_torch_module_wrapper</a>()</code> <br><code><a href="layer_unit_normalization.html">layer_unit_normalization</a>()</code> <br><code><a href="layer_upsampling_1d.html">layer_upsampling_1d</a>()</code> <br><code><a href="layer_upsampling_2d.html">layer_upsampling_2d</a>()</code> <br><code><a href="layer_upsampling_3d.html">layer_upsampling_3d</a>()</code> <br><code><a href="layer_zero_padding_1d.html">layer_zero_padding_1d</a>()</code> <br><code><a href="layer_zero_padding_2d.html">layer_zero_padding_2d</a>()</code> <br><code><a href="layer_zero_padding_3d.html">layer_zero_padding_3d</a>()</code> <br><code><a href="rnn_cell_gru.html">rnn_cell_gru</a>()</code> <br><code><a href="rnn_cell_lstm.html">rnn_cell_lstm</a>()</code> <br><code><a href="rnn_cell_simple.html">rnn_cell_simple</a>()</code> <br><code><a href="rnn_cells_stack.html">rnn_cells_stack</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -245,7 +247,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/layer_conv_3d.html b/docs/dev/reference/layer_conv_3d.html
index 605e80b79..86a2d32c8 100644
--- a/docs/dev/reference/layer_conv_3d.html
+++ b/docs/dev/reference/layer_conv_3d.html
@@ -16,7 +16,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -137,7 +137,7 @@ <h2 id="arguments">Arguments<a class="anchor" aria-label="anchor" href="#argumen
 
 
 <dt id="arg-dilation-rate">dilation_rate<a class="anchor" aria-label="anchor" href="#arg-dilation-rate"></a></dt>
-<dd><p>int or list of 3 integers, specifying the dilation
+<dd><p>int or vector of 3 ints, specifying the dilation
 rate to use for dilated convolution.</p></dd>
 
 
@@ -238,7 +238,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
     <div class="dont-index">
 <ul><li><p><a href="https://keras.io/api/layers/convolution_layers/convolution3d#conv3d-class" class="external-link">https://keras.io/api/layers/convolution_layers/convolution3d#conv3d-class</a></p></li>
 </ul><p>Other convolutional layers: <br><code><a href="layer_conv_1d.html">layer_conv_1d</a>()</code> <br><code><a href="layer_conv_1d_transpose.html">layer_conv_1d_transpose</a>()</code> <br><code><a href="layer_conv_2d.html">layer_conv_2d</a>()</code> <br><code><a href="layer_conv_2d_transpose.html">layer_conv_2d_transpose</a>()</code> <br><code><a href="layer_conv_3d_transpose.html">layer_conv_3d_transpose</a>()</code> <br><code><a href="layer_depthwise_conv_1d.html">layer_depthwise_conv_1d</a>()</code> <br><code><a href="layer_depthwise_conv_2d.html">layer_depthwise_conv_2d</a>()</code> <br><code><a href="layer_separable_conv_1d.html">layer_separable_conv_1d</a>()</code> <br><code><a href="layer_separable_conv_2d.html">layer_separable_conv_2d</a>()</code> <br></p>
-<p>Other layers: <br><code><a href="Layer.html">Layer</a>()</code> <br><code><a href="layer_activation.html">layer_activation</a>()</code> <br><code><a href="layer_activation_elu.html">layer_activation_elu</a>()</code> <br><code><a href="layer_activation_leaky_relu.html">layer_activation_leaky_relu</a>()</code> <br><code><a href="layer_activation_parametric_relu.html">layer_activation_parametric_relu</a>()</code> <br><code><a href="layer_activation_relu.html">layer_activation_relu</a>()</code> <br><code><a href="layer_activation_softmax.html">layer_activation_softmax</a>()</code> <br><code><a href="layer_activity_regularization.html">layer_activity_regularization</a>()</code> <br><code><a href="layer_add.html">layer_add</a>()</code> <br><code><a href="layer_additive_attention.html">layer_additive_attention</a>()</code> <br><code><a href="layer_alpha_dropout.html">layer_alpha_dropout</a>()</code> <br><code><a href="layer_attention.html">layer_attention</a>()</code> <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_average.html">layer_average</a>()</code> <br><code><a href="layer_average_pooling_1d.html">layer_average_pooling_1d</a>()</code> <br><code><a href="layer_average_pooling_2d.html">layer_average_pooling_2d</a>()</code> <br><code><a href="layer_average_pooling_3d.html">layer_average_pooling_3d</a>()</code> <br><code><a href="layer_batch_normalization.html">layer_batch_normalization</a>()</code> <br><code><a href="layer_bidirectional.html">layer_bidirectional</a>()</code> <br><code><a href="layer_category_encoding.html">layer_category_encoding</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_concatenate.html">layer_concatenate</a>()</code> <br><code><a href="layer_conv_1d.html">layer_conv_1d</a>()</code> <br><code><a href="layer_conv_1d_transpose.html">layer_conv_1d_transpose</a>()</code> <br><code><a href="layer_conv_2d.html">layer_conv_2d</a>()</code> <br><code><a href="layer_conv_2d_transpose.html">layer_conv_2d_transpose</a>()</code> <br><code><a href="layer_conv_3d_transpose.html">layer_conv_3d_transpose</a>()</code> <br><code><a href="layer_conv_lstm_1d.html">layer_conv_lstm_1d</a>()</code> <br><code><a href="layer_conv_lstm_2d.html">layer_conv_lstm_2d</a>()</code> <br><code><a href="layer_conv_lstm_3d.html">layer_conv_lstm_3d</a>()</code> <br><code><a href="layer_cropping_1d.html">layer_cropping_1d</a>()</code> <br><code><a href="layer_cropping_2d.html">layer_cropping_2d</a>()</code> <br><code><a href="layer_cropping_3d.html">layer_cropping_3d</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_dense.html">layer_dense</a>()</code> <br><code><a href="layer_depthwise_conv_1d.html">layer_depthwise_conv_1d</a>()</code> <br><code><a href="layer_depthwise_conv_2d.html">layer_depthwise_conv_2d</a>()</code> <br><code><a href="layer_discretization.html">layer_discretization</a>()</code> <br><code><a href="layer_dot.html">layer_dot</a>()</code> <br><code><a href="layer_dropout.html">layer_dropout</a>()</code> <br><code><a href="layer_einsum_dense.html">layer_einsum_dense</a>()</code> <br><code><a href="layer_embedding.html">layer_embedding</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_feature_space.html">layer_feature_space</a>()</code> <br><code><a href="layer_flatten.html">layer_flatten</a>()</code> <br><code><a href="layer_flax_module_wrapper.html">layer_flax_module_wrapper</a>()</code> <br><code><a href="layer_gaussian_dropout.html">layer_gaussian_dropout</a>()</code> <br><code><a href="layer_gaussian_noise.html">layer_gaussian_noise</a>()</code> <br><code><a href="layer_global_average_pooling_1d.html">layer_global_average_pooling_1d</a>()</code> <br><code><a href="layer_global_average_pooling_2d.html">layer_global_average_pooling_2d</a>()</code> <br><code><a href="layer_global_average_pooling_3d.html">layer_global_average_pooling_3d</a>()</code> <br><code><a href="layer_global_max_pooling_1d.html">layer_global_max_pooling_1d</a>()</code> <br><code><a href="layer_global_max_pooling_2d.html">layer_global_max_pooling_2d</a>()</code> <br><code><a href="layer_global_max_pooling_3d.html">layer_global_max_pooling_3d</a>()</code> <br><code><a href="layer_group_normalization.html">layer_group_normalization</a>()</code> <br><code><a href="layer_group_query_attention.html">layer_group_query_attention</a>()</code> <br><code><a href="layer_gru.html">layer_gru</a>()</code> <br><code><a href="layer_hashed_crossing.html">layer_hashed_crossing</a>()</code> <br><code><a href="layer_hashing.html">layer_hashing</a>()</code> <br><code><a href="layer_identity.html">layer_identity</a>()</code> <br><code><a href="layer_integer_lookup.html">layer_integer_lookup</a>()</code> <br><code><a href="layer_jax_model_wrapper.html">layer_jax_model_wrapper</a>()</code> <br><code><a href="layer_lambda.html">layer_lambda</a>()</code> <br><code><a href="layer_layer_normalization.html">layer_layer_normalization</a>()</code> <br><code><a href="layer_lstm.html">layer_lstm</a>()</code> <br><code><a href="layer_masking.html">layer_masking</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_max_pooling_1d.html">layer_max_pooling_1d</a>()</code> <br><code><a href="layer_max_pooling_2d.html">layer_max_pooling_2d</a>()</code> <br><code><a href="layer_max_pooling_3d.html">layer_max_pooling_3d</a>()</code> <br><code><a href="layer_maximum.html">layer_maximum</a>()</code> <br><code><a href="layer_mel_spectrogram.html">layer_mel_spectrogram</a>()</code> <br><code><a href="layer_minimum.html">layer_minimum</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_multi_head_attention.html">layer_multi_head_attention</a>()</code> <br><code><a href="layer_multiply.html">layer_multiply</a>()</code> <br><code><a href="layer_normalization.html">layer_normalization</a>()</code> <br><code><a href="layer_permute.html">layer_permute</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_brightness.html">layer_random_brightness</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_contrast.html">layer_random_contrast</a>()</code> <br><code><a href="layer_random_crop.html">layer_random_crop</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_flip.html">layer_random_flip</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_rotation.html">layer_random_rotation</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_random_translation.html">layer_random_translation</a>()</code> <br><code><a href="layer_random_zoom.html">layer_random_zoom</a>()</code> <br><code><a href="layer_repeat_vector.html">layer_repeat_vector</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_reshape.html">layer_reshape</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_rms_normalization.html">layer_rms_normalization</a>()</code> <br><code><a href="layer_rnn.html">layer_rnn</a>()</code> <br><code><a href="layer_separable_conv_1d.html">layer_separable_conv_1d</a>()</code> <br><code><a href="layer_separable_conv_2d.html">layer_separable_conv_2d</a>()</code> <br><code><a href="layer_simple_rnn.html">layer_simple_rnn</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br><code><a href="layer_spatial_dropout_1d.html">layer_spatial_dropout_1d</a>()</code> <br><code><a href="layer_spatial_dropout_2d.html">layer_spatial_dropout_2d</a>()</code> <br><code><a href="layer_spatial_dropout_3d.html">layer_spatial_dropout_3d</a>()</code> <br><code><a href="layer_spectral_normalization.html">layer_spectral_normalization</a>()</code> <br><code><a href="layer_stft_spectrogram.html">layer_stft_spectrogram</a>()</code> <br><code><a href="layer_string_lookup.html">layer_string_lookup</a>()</code> <br><code><a href="layer_subtract.html">layer_subtract</a>()</code> <br><code><a href="layer_text_vectorization.html">layer_text_vectorization</a>()</code> <br><code><a href="layer_tfsm.html">layer_tfsm</a>()</code> <br><code><a href="layer_time_distributed.html">layer_time_distributed</a>()</code> <br><code><a href="layer_torch_module_wrapper.html">layer_torch_module_wrapper</a>()</code> <br><code><a href="layer_unit_normalization.html">layer_unit_normalization</a>()</code> <br><code><a href="layer_upsampling_1d.html">layer_upsampling_1d</a>()</code> <br><code><a href="layer_upsampling_2d.html">layer_upsampling_2d</a>()</code> <br><code><a href="layer_upsampling_3d.html">layer_upsampling_3d</a>()</code> <br><code><a href="layer_zero_padding_1d.html">layer_zero_padding_1d</a>()</code> <br><code><a href="layer_zero_padding_2d.html">layer_zero_padding_2d</a>()</code> <br><code><a href="layer_zero_padding_3d.html">layer_zero_padding_3d</a>()</code> <br><code><a href="rnn_cell_gru.html">rnn_cell_gru</a>()</code> <br><code><a href="rnn_cell_lstm.html">rnn_cell_lstm</a>()</code> <br><code><a href="rnn_cell_simple.html">rnn_cell_simple</a>()</code> <br><code><a href="rnn_cells_stack.html">rnn_cells_stack</a>()</code> <br></p></div>
+<p>Other layers: <br><code><a href="Layer.html">Layer</a>()</code> <br><code><a href="layer_activation.html">layer_activation</a>()</code> <br><code><a href="layer_activation_elu.html">layer_activation_elu</a>()</code> <br><code><a href="layer_activation_leaky_relu.html">layer_activation_leaky_relu</a>()</code> <br><code><a href="layer_activation_parametric_relu.html">layer_activation_parametric_relu</a>()</code> <br><code><a href="layer_activation_relu.html">layer_activation_relu</a>()</code> <br><code><a href="layer_activation_softmax.html">layer_activation_softmax</a>()</code> <br><code><a href="layer_activity_regularization.html">layer_activity_regularization</a>()</code> <br><code><a href="layer_add.html">layer_add</a>()</code> <br><code><a href="layer_additive_attention.html">layer_additive_attention</a>()</code> <br><code><a href="layer_alpha_dropout.html">layer_alpha_dropout</a>()</code> <br><code><a href="layer_attention.html">layer_attention</a>()</code> <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_average.html">layer_average</a>()</code> <br><code><a href="layer_average_pooling_1d.html">layer_average_pooling_1d</a>()</code> <br><code><a href="layer_average_pooling_2d.html">layer_average_pooling_2d</a>()</code> <br><code><a href="layer_average_pooling_3d.html">layer_average_pooling_3d</a>()</code> <br><code><a href="layer_batch_normalization.html">layer_batch_normalization</a>()</code> <br><code><a href="layer_bidirectional.html">layer_bidirectional</a>()</code> <br><code><a href="layer_category_encoding.html">layer_category_encoding</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_concatenate.html">layer_concatenate</a>()</code> <br><code><a href="layer_conv_1d.html">layer_conv_1d</a>()</code> <br><code><a href="layer_conv_1d_transpose.html">layer_conv_1d_transpose</a>()</code> <br><code><a href="layer_conv_2d.html">layer_conv_2d</a>()</code> <br><code><a href="layer_conv_2d_transpose.html">layer_conv_2d_transpose</a>()</code> <br><code><a href="layer_conv_3d_transpose.html">layer_conv_3d_transpose</a>()</code> <br><code><a href="layer_conv_lstm_1d.html">layer_conv_lstm_1d</a>()</code> <br><code><a href="layer_conv_lstm_2d.html">layer_conv_lstm_2d</a>()</code> <br><code><a href="layer_conv_lstm_3d.html">layer_conv_lstm_3d</a>()</code> <br><code><a href="layer_cropping_1d.html">layer_cropping_1d</a>()</code> <br><code><a href="layer_cropping_2d.html">layer_cropping_2d</a>()</code> <br><code><a href="layer_cropping_3d.html">layer_cropping_3d</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_dense.html">layer_dense</a>()</code> <br><code><a href="layer_depthwise_conv_1d.html">layer_depthwise_conv_1d</a>()</code> <br><code><a href="layer_depthwise_conv_2d.html">layer_depthwise_conv_2d</a>()</code> <br><code><a href="layer_discretization.html">layer_discretization</a>()</code> <br><code><a href="layer_dot.html">layer_dot</a>()</code> <br><code><a href="layer_dropout.html">layer_dropout</a>()</code> <br><code><a href="layer_einsum_dense.html">layer_einsum_dense</a>()</code> <br><code><a href="layer_embedding.html">layer_embedding</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_feature_space.html">layer_feature_space</a>()</code> <br><code><a href="layer_flatten.html">layer_flatten</a>()</code> <br><code><a href="layer_flax_module_wrapper.html">layer_flax_module_wrapper</a>()</code> <br><code><a href="layer_gaussian_dropout.html">layer_gaussian_dropout</a>()</code> <br><code><a href="layer_gaussian_noise.html">layer_gaussian_noise</a>()</code> <br><code><a href="layer_global_average_pooling_1d.html">layer_global_average_pooling_1d</a>()</code> <br><code><a href="layer_global_average_pooling_2d.html">layer_global_average_pooling_2d</a>()</code> <br><code><a href="layer_global_average_pooling_3d.html">layer_global_average_pooling_3d</a>()</code> <br><code><a href="layer_global_max_pooling_1d.html">layer_global_max_pooling_1d</a>()</code> <br><code><a href="layer_global_max_pooling_2d.html">layer_global_max_pooling_2d</a>()</code> <br><code><a href="layer_global_max_pooling_3d.html">layer_global_max_pooling_3d</a>()</code> <br><code><a href="layer_group_normalization.html">layer_group_normalization</a>()</code> <br><code><a href="layer_group_query_attention.html">layer_group_query_attention</a>()</code> <br><code><a href="layer_gru.html">layer_gru</a>()</code> <br><code><a href="layer_hashed_crossing.html">layer_hashed_crossing</a>()</code> <br><code><a href="layer_hashing.html">layer_hashing</a>()</code> <br><code><a href="layer_identity.html">layer_identity</a>()</code> <br><code><a href="layer_integer_lookup.html">layer_integer_lookup</a>()</code> <br><code><a href="layer_jax_model_wrapper.html">layer_jax_model_wrapper</a>()</code> <br><code><a href="layer_lambda.html">layer_lambda</a>()</code> <br><code><a href="layer_layer_normalization.html">layer_layer_normalization</a>()</code> <br><code><a href="layer_lstm.html">layer_lstm</a>()</code> <br><code><a href="layer_masking.html">layer_masking</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_max_pooling_1d.html">layer_max_pooling_1d</a>()</code> <br><code><a href="layer_max_pooling_2d.html">layer_max_pooling_2d</a>()</code> <br><code><a href="layer_max_pooling_3d.html">layer_max_pooling_3d</a>()</code> <br><code><a href="layer_maximum.html">layer_maximum</a>()</code> <br><code><a href="layer_mel_spectrogram.html">layer_mel_spectrogram</a>()</code> <br><code><a href="layer_minimum.html">layer_minimum</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_multi_head_attention.html">layer_multi_head_attention</a>()</code> <br><code><a href="layer_multiply.html">layer_multiply</a>()</code> <br><code><a href="layer_normalization.html">layer_normalization</a>()</code> <br><code><a href="layer_permute.html">layer_permute</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_brightness.html">layer_random_brightness</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_contrast.html">layer_random_contrast</a>()</code> <br><code><a href="layer_random_crop.html">layer_random_crop</a>()</code> <br><code><a href="layer_random_elastic_transform.html">layer_random_elastic_transform</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_flip.html">layer_random_flip</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_rotation.html">layer_random_rotation</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_random_translation.html">layer_random_translation</a>()</code> <br><code><a href="layer_random_zoom.html">layer_random_zoom</a>()</code> <br><code><a href="layer_repeat_vector.html">layer_repeat_vector</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_reshape.html">layer_reshape</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_rms_normalization.html">layer_rms_normalization</a>()</code> <br><code><a href="layer_rnn.html">layer_rnn</a>()</code> <br><code><a href="layer_separable_conv_1d.html">layer_separable_conv_1d</a>()</code> <br><code><a href="layer_separable_conv_2d.html">layer_separable_conv_2d</a>()</code> <br><code><a href="layer_simple_rnn.html">layer_simple_rnn</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br><code><a href="layer_spatial_dropout_1d.html">layer_spatial_dropout_1d</a>()</code> <br><code><a href="layer_spatial_dropout_2d.html">layer_spatial_dropout_2d</a>()</code> <br><code><a href="layer_spatial_dropout_3d.html">layer_spatial_dropout_3d</a>()</code> <br><code><a href="layer_spectral_normalization.html">layer_spectral_normalization</a>()</code> <br><code><a href="layer_stft_spectrogram.html">layer_stft_spectrogram</a>()</code> <br><code><a href="layer_string_lookup.html">layer_string_lookup</a>()</code> <br><code><a href="layer_subtract.html">layer_subtract</a>()</code> <br><code><a href="layer_text_vectorization.html">layer_text_vectorization</a>()</code> <br><code><a href="layer_tfsm.html">layer_tfsm</a>()</code> <br><code><a href="layer_time_distributed.html">layer_time_distributed</a>()</code> <br><code><a href="layer_torch_module_wrapper.html">layer_torch_module_wrapper</a>()</code> <br><code><a href="layer_unit_normalization.html">layer_unit_normalization</a>()</code> <br><code><a href="layer_upsampling_1d.html">layer_upsampling_1d</a>()</code> <br><code><a href="layer_upsampling_2d.html">layer_upsampling_2d</a>()</code> <br><code><a href="layer_upsampling_3d.html">layer_upsampling_3d</a>()</code> <br><code><a href="layer_zero_padding_1d.html">layer_zero_padding_1d</a>()</code> <br><code><a href="layer_zero_padding_2d.html">layer_zero_padding_2d</a>()</code> <br><code><a href="layer_zero_padding_3d.html">layer_zero_padding_3d</a>()</code> <br><code><a href="rnn_cell_gru.html">rnn_cell_gru</a>()</code> <br><code><a href="rnn_cell_lstm.html">rnn_cell_lstm</a>()</code> <br><code><a href="rnn_cell_simple.html">rnn_cell_simple</a>()</code> <br><code><a href="rnn_cells_stack.html">rnn_cells_stack</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -250,7 +250,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/layer_conv_3d_transpose.html b/docs/dev/reference/layer_conv_3d_transpose.html
index ba7c713cd..2d865af0a 100644
--- a/docs/dev/reference/layer_conv_3d_transpose.html
+++ b/docs/dev/reference/layer_conv_3d_transpose.html
@@ -16,7 +16,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -77,6 +77,7 @@ <h2 id="ref-usage">Usage<a class="anchor" aria-label="anchor" href="#ref-usage">
 <span>  <span class="va">kernel_size</span>,</span>
 <span>  strides <span class="op">=</span> <span class="fu"><a href="https://rdrr.io/r/base/list.html" class="external-link">list</a></span><span class="op">(</span><span class="fl">1L</span>, <span class="fl">1L</span>, <span class="fl">1L</span><span class="op">)</span>,</span>
 <span>  padding <span class="op">=</span> <span class="st">"valid"</span>,</span>
+<span>  output_padding <span class="op">=</span> <span class="cn">NULL</span>,</span>
 <span>  data_format <span class="op">=</span> <span class="cn">NULL</span>,</span>
 <span>  dilation_rate <span class="op">=</span> <span class="fu"><a href="https://rdrr.io/r/base/list.html" class="external-link">list</a></span><span class="op">(</span><span class="fl">1L</span>, <span class="fl">1L</span>, <span class="fl">1L</span><span class="op">)</span>,</span>
 <span>  activation <span class="op">=</span> <span class="cn">NULL</span>,</span>
@@ -136,8 +137,9 @@ <h2 id="arguments">Arguments<a class="anchor" aria-label="anchor" href="#argumen
 
 
 <dt id="arg-dilation-rate">dilation_rate<a class="anchor" aria-label="anchor" href="#arg-dilation-rate"></a></dt>
-<dd><p>int or list of 1 integers, specifying the dilation
-rate to use for dilated transposed convolution.</p></dd>
+<dd><p>Scalar integer or vector of 3 integers specifying the dilation rate. Values
+other than 1 require <code>strides = 1</code>; different rates per dimension are not
+supported.</p></dd>
 
 
 <dt id="arg-activation">activation<a class="anchor" aria-label="anchor" href="#arg-activation"></a></dt>
@@ -235,7 +237,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
     <div class="dont-index">
 <ul><li><p><a href="https://keras.io/api/layers/convolution_layers/convolution3d_transpose#conv3dtranspose-class" class="external-link">https://keras.io/api/layers/convolution_layers/convolution3d_transpose#conv3dtranspose-class</a></p></li>
 </ul><p>Other convolutional layers: <br><code><a href="layer_conv_1d.html">layer_conv_1d</a>()</code> <br><code><a href="layer_conv_1d_transpose.html">layer_conv_1d_transpose</a>()</code> <br><code><a href="layer_conv_2d.html">layer_conv_2d</a>()</code> <br><code><a href="layer_conv_2d_transpose.html">layer_conv_2d_transpose</a>()</code> <br><code><a href="layer_conv_3d.html">layer_conv_3d</a>()</code> <br><code><a href="layer_depthwise_conv_1d.html">layer_depthwise_conv_1d</a>()</code> <br><code><a href="layer_depthwise_conv_2d.html">layer_depthwise_conv_2d</a>()</code> <br><code><a href="layer_separable_conv_1d.html">layer_separable_conv_1d</a>()</code> <br><code><a href="layer_separable_conv_2d.html">layer_separable_conv_2d</a>()</code> <br></p>
-<p>Other layers: <br><code><a href="Layer.html">Layer</a>()</code> <br><code><a href="layer_activation.html">layer_activation</a>()</code> <br><code><a href="layer_activation_elu.html">layer_activation_elu</a>()</code> <br><code><a href="layer_activation_leaky_relu.html">layer_activation_leaky_relu</a>()</code> <br><code><a href="layer_activation_parametric_relu.html">layer_activation_parametric_relu</a>()</code> <br><code><a href="layer_activation_relu.html">layer_activation_relu</a>()</code> <br><code><a href="layer_activation_softmax.html">layer_activation_softmax</a>()</code> <br><code><a href="layer_activity_regularization.html">layer_activity_regularization</a>()</code> <br><code><a href="layer_add.html">layer_add</a>()</code> <br><code><a href="layer_additive_attention.html">layer_additive_attention</a>()</code> <br><code><a href="layer_alpha_dropout.html">layer_alpha_dropout</a>()</code> <br><code><a href="layer_attention.html">layer_attention</a>()</code> <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_average.html">layer_average</a>()</code> <br><code><a href="layer_average_pooling_1d.html">layer_average_pooling_1d</a>()</code> <br><code><a href="layer_average_pooling_2d.html">layer_average_pooling_2d</a>()</code> <br><code><a href="layer_average_pooling_3d.html">layer_average_pooling_3d</a>()</code> <br><code><a href="layer_batch_normalization.html">layer_batch_normalization</a>()</code> <br><code><a href="layer_bidirectional.html">layer_bidirectional</a>()</code> <br><code><a href="layer_category_encoding.html">layer_category_encoding</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_concatenate.html">layer_concatenate</a>()</code> <br><code><a href="layer_conv_1d.html">layer_conv_1d</a>()</code> <br><code><a href="layer_conv_1d_transpose.html">layer_conv_1d_transpose</a>()</code> <br><code><a href="layer_conv_2d.html">layer_conv_2d</a>()</code> <br><code><a href="layer_conv_2d_transpose.html">layer_conv_2d_transpose</a>()</code> <br><code><a href="layer_conv_3d.html">layer_conv_3d</a>()</code> <br><code><a href="layer_conv_lstm_1d.html">layer_conv_lstm_1d</a>()</code> <br><code><a href="layer_conv_lstm_2d.html">layer_conv_lstm_2d</a>()</code> <br><code><a href="layer_conv_lstm_3d.html">layer_conv_lstm_3d</a>()</code> <br><code><a href="layer_cropping_1d.html">layer_cropping_1d</a>()</code> <br><code><a href="layer_cropping_2d.html">layer_cropping_2d</a>()</code> <br><code><a href="layer_cropping_3d.html">layer_cropping_3d</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_dense.html">layer_dense</a>()</code> <br><code><a href="layer_depthwise_conv_1d.html">layer_depthwise_conv_1d</a>()</code> <br><code><a href="layer_depthwise_conv_2d.html">layer_depthwise_conv_2d</a>()</code> <br><code><a href="layer_discretization.html">layer_discretization</a>()</code> <br><code><a href="layer_dot.html">layer_dot</a>()</code> <br><code><a href="layer_dropout.html">layer_dropout</a>()</code> <br><code><a href="layer_einsum_dense.html">layer_einsum_dense</a>()</code> <br><code><a href="layer_embedding.html">layer_embedding</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_feature_space.html">layer_feature_space</a>()</code> <br><code><a href="layer_flatten.html">layer_flatten</a>()</code> <br><code><a href="layer_flax_module_wrapper.html">layer_flax_module_wrapper</a>()</code> <br><code><a href="layer_gaussian_dropout.html">layer_gaussian_dropout</a>()</code> <br><code><a href="layer_gaussian_noise.html">layer_gaussian_noise</a>()</code> <br><code><a href="layer_global_average_pooling_1d.html">layer_global_average_pooling_1d</a>()</code> <br><code><a href="layer_global_average_pooling_2d.html">layer_global_average_pooling_2d</a>()</code> <br><code><a href="layer_global_average_pooling_3d.html">layer_global_average_pooling_3d</a>()</code> <br><code><a href="layer_global_max_pooling_1d.html">layer_global_max_pooling_1d</a>()</code> <br><code><a href="layer_global_max_pooling_2d.html">layer_global_max_pooling_2d</a>()</code> <br><code><a href="layer_global_max_pooling_3d.html">layer_global_max_pooling_3d</a>()</code> <br><code><a href="layer_group_normalization.html">layer_group_normalization</a>()</code> <br><code><a href="layer_group_query_attention.html">layer_group_query_attention</a>()</code> <br><code><a href="layer_gru.html">layer_gru</a>()</code> <br><code><a href="layer_hashed_crossing.html">layer_hashed_crossing</a>()</code> <br><code><a href="layer_hashing.html">layer_hashing</a>()</code> <br><code><a href="layer_identity.html">layer_identity</a>()</code> <br><code><a href="layer_integer_lookup.html">layer_integer_lookup</a>()</code> <br><code><a href="layer_jax_model_wrapper.html">layer_jax_model_wrapper</a>()</code> <br><code><a href="layer_lambda.html">layer_lambda</a>()</code> <br><code><a href="layer_layer_normalization.html">layer_layer_normalization</a>()</code> <br><code><a href="layer_lstm.html">layer_lstm</a>()</code> <br><code><a href="layer_masking.html">layer_masking</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_max_pooling_1d.html">layer_max_pooling_1d</a>()</code> <br><code><a href="layer_max_pooling_2d.html">layer_max_pooling_2d</a>()</code> <br><code><a href="layer_max_pooling_3d.html">layer_max_pooling_3d</a>()</code> <br><code><a href="layer_maximum.html">layer_maximum</a>()</code> <br><code><a href="layer_mel_spectrogram.html">layer_mel_spectrogram</a>()</code> <br><code><a href="layer_minimum.html">layer_minimum</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_multi_head_attention.html">layer_multi_head_attention</a>()</code> <br><code><a href="layer_multiply.html">layer_multiply</a>()</code> <br><code><a href="layer_normalization.html">layer_normalization</a>()</code> <br><code><a href="layer_permute.html">layer_permute</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_brightness.html">layer_random_brightness</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_contrast.html">layer_random_contrast</a>()</code> <br><code><a href="layer_random_crop.html">layer_random_crop</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_flip.html">layer_random_flip</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_rotation.html">layer_random_rotation</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_random_translation.html">layer_random_translation</a>()</code> <br><code><a href="layer_random_zoom.html">layer_random_zoom</a>()</code> <br><code><a href="layer_repeat_vector.html">layer_repeat_vector</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_reshape.html">layer_reshape</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_rms_normalization.html">layer_rms_normalization</a>()</code> <br><code><a href="layer_rnn.html">layer_rnn</a>()</code> <br><code><a href="layer_separable_conv_1d.html">layer_separable_conv_1d</a>()</code> <br><code><a href="layer_separable_conv_2d.html">layer_separable_conv_2d</a>()</code> <br><code><a href="layer_simple_rnn.html">layer_simple_rnn</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br><code><a href="layer_spatial_dropout_1d.html">layer_spatial_dropout_1d</a>()</code> <br><code><a href="layer_spatial_dropout_2d.html">layer_spatial_dropout_2d</a>()</code> <br><code><a href="layer_spatial_dropout_3d.html">layer_spatial_dropout_3d</a>()</code> <br><code><a href="layer_spectral_normalization.html">layer_spectral_normalization</a>()</code> <br><code><a href="layer_stft_spectrogram.html">layer_stft_spectrogram</a>()</code> <br><code><a href="layer_string_lookup.html">layer_string_lookup</a>()</code> <br><code><a href="layer_subtract.html">layer_subtract</a>()</code> <br><code><a href="layer_text_vectorization.html">layer_text_vectorization</a>()</code> <br><code><a href="layer_tfsm.html">layer_tfsm</a>()</code> <br><code><a href="layer_time_distributed.html">layer_time_distributed</a>()</code> <br><code><a href="layer_torch_module_wrapper.html">layer_torch_module_wrapper</a>()</code> <br><code><a href="layer_unit_normalization.html">layer_unit_normalization</a>()</code> <br><code><a href="layer_upsampling_1d.html">layer_upsampling_1d</a>()</code> <br><code><a href="layer_upsampling_2d.html">layer_upsampling_2d</a>()</code> <br><code><a href="layer_upsampling_3d.html">layer_upsampling_3d</a>()</code> <br><code><a href="layer_zero_padding_1d.html">layer_zero_padding_1d</a>()</code> <br><code><a href="layer_zero_padding_2d.html">layer_zero_padding_2d</a>()</code> <br><code><a href="layer_zero_padding_3d.html">layer_zero_padding_3d</a>()</code> <br><code><a href="rnn_cell_gru.html">rnn_cell_gru</a>()</code> <br><code><a href="rnn_cell_lstm.html">rnn_cell_lstm</a>()</code> <br><code><a href="rnn_cell_simple.html">rnn_cell_simple</a>()</code> <br><code><a href="rnn_cells_stack.html">rnn_cells_stack</a>()</code> <br></p></div>
+<p>Other layers: <br><code><a href="Layer.html">Layer</a>()</code> <br><code><a href="layer_activation.html">layer_activation</a>()</code> <br><code><a href="layer_activation_elu.html">layer_activation_elu</a>()</code> <br><code><a href="layer_activation_leaky_relu.html">layer_activation_leaky_relu</a>()</code> <br><code><a href="layer_activation_parametric_relu.html">layer_activation_parametric_relu</a>()</code> <br><code><a href="layer_activation_relu.html">layer_activation_relu</a>()</code> <br><code><a href="layer_activation_softmax.html">layer_activation_softmax</a>()</code> <br><code><a href="layer_activity_regularization.html">layer_activity_regularization</a>()</code> <br><code><a href="layer_add.html">layer_add</a>()</code> <br><code><a href="layer_additive_attention.html">layer_additive_attention</a>()</code> <br><code><a href="layer_alpha_dropout.html">layer_alpha_dropout</a>()</code> <br><code><a href="layer_attention.html">layer_attention</a>()</code> <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_average.html">layer_average</a>()</code> <br><code><a href="layer_average_pooling_1d.html">layer_average_pooling_1d</a>()</code> <br><code><a href="layer_average_pooling_2d.html">layer_average_pooling_2d</a>()</code> <br><code><a href="layer_average_pooling_3d.html">layer_average_pooling_3d</a>()</code> <br><code><a href="layer_batch_normalization.html">layer_batch_normalization</a>()</code> <br><code><a href="layer_bidirectional.html">layer_bidirectional</a>()</code> <br><code><a href="layer_category_encoding.html">layer_category_encoding</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_concatenate.html">layer_concatenate</a>()</code> <br><code><a href="layer_conv_1d.html">layer_conv_1d</a>()</code> <br><code><a href="layer_conv_1d_transpose.html">layer_conv_1d_transpose</a>()</code> <br><code><a href="layer_conv_2d.html">layer_conv_2d</a>()</code> <br><code><a href="layer_conv_2d_transpose.html">layer_conv_2d_transpose</a>()</code> <br><code><a href="layer_conv_3d.html">layer_conv_3d</a>()</code> <br><code><a href="layer_conv_lstm_1d.html">layer_conv_lstm_1d</a>()</code> <br><code><a href="layer_conv_lstm_2d.html">layer_conv_lstm_2d</a>()</code> <br><code><a href="layer_conv_lstm_3d.html">layer_conv_lstm_3d</a>()</code> <br><code><a href="layer_cropping_1d.html">layer_cropping_1d</a>()</code> <br><code><a href="layer_cropping_2d.html">layer_cropping_2d</a>()</code> <br><code><a href="layer_cropping_3d.html">layer_cropping_3d</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_dense.html">layer_dense</a>()</code> <br><code><a href="layer_depthwise_conv_1d.html">layer_depthwise_conv_1d</a>()</code> <br><code><a href="layer_depthwise_conv_2d.html">layer_depthwise_conv_2d</a>()</code> <br><code><a href="layer_discretization.html">layer_discretization</a>()</code> <br><code><a href="layer_dot.html">layer_dot</a>()</code> <br><code><a href="layer_dropout.html">layer_dropout</a>()</code> <br><code><a href="layer_einsum_dense.html">layer_einsum_dense</a>()</code> <br><code><a href="layer_embedding.html">layer_embedding</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_feature_space.html">layer_feature_space</a>()</code> <br><code><a href="layer_flatten.html">layer_flatten</a>()</code> <br><code><a href="layer_flax_module_wrapper.html">layer_flax_module_wrapper</a>()</code> <br><code><a href="layer_gaussian_dropout.html">layer_gaussian_dropout</a>()</code> <br><code><a href="layer_gaussian_noise.html">layer_gaussian_noise</a>()</code> <br><code><a href="layer_global_average_pooling_1d.html">layer_global_average_pooling_1d</a>()</code> <br><code><a href="layer_global_average_pooling_2d.html">layer_global_average_pooling_2d</a>()</code> <br><code><a href="layer_global_average_pooling_3d.html">layer_global_average_pooling_3d</a>()</code> <br><code><a href="layer_global_max_pooling_1d.html">layer_global_max_pooling_1d</a>()</code> <br><code><a href="layer_global_max_pooling_2d.html">layer_global_max_pooling_2d</a>()</code> <br><code><a href="layer_global_max_pooling_3d.html">layer_global_max_pooling_3d</a>()</code> <br><code><a href="layer_group_normalization.html">layer_group_normalization</a>()</code> <br><code><a href="layer_group_query_attention.html">layer_group_query_attention</a>()</code> <br><code><a href="layer_gru.html">layer_gru</a>()</code> <br><code><a href="layer_hashed_crossing.html">layer_hashed_crossing</a>()</code> <br><code><a href="layer_hashing.html">layer_hashing</a>()</code> <br><code><a href="layer_identity.html">layer_identity</a>()</code> <br><code><a href="layer_integer_lookup.html">layer_integer_lookup</a>()</code> <br><code><a href="layer_jax_model_wrapper.html">layer_jax_model_wrapper</a>()</code> <br><code><a href="layer_lambda.html">layer_lambda</a>()</code> <br><code><a href="layer_layer_normalization.html">layer_layer_normalization</a>()</code> <br><code><a href="layer_lstm.html">layer_lstm</a>()</code> <br><code><a href="layer_masking.html">layer_masking</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_max_pooling_1d.html">layer_max_pooling_1d</a>()</code> <br><code><a href="layer_max_pooling_2d.html">layer_max_pooling_2d</a>()</code> <br><code><a href="layer_max_pooling_3d.html">layer_max_pooling_3d</a>()</code> <br><code><a href="layer_maximum.html">layer_maximum</a>()</code> <br><code><a href="layer_mel_spectrogram.html">layer_mel_spectrogram</a>()</code> <br><code><a href="layer_minimum.html">layer_minimum</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_multi_head_attention.html">layer_multi_head_attention</a>()</code> <br><code><a href="layer_multiply.html">layer_multiply</a>()</code> <br><code><a href="layer_normalization.html">layer_normalization</a>()</code> <br><code><a href="layer_permute.html">layer_permute</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_brightness.html">layer_random_brightness</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_contrast.html">layer_random_contrast</a>()</code> <br><code><a href="layer_random_crop.html">layer_random_crop</a>()</code> <br><code><a href="layer_random_elastic_transform.html">layer_random_elastic_transform</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_flip.html">layer_random_flip</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_rotation.html">layer_random_rotation</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_random_translation.html">layer_random_translation</a>()</code> <br><code><a href="layer_random_zoom.html">layer_random_zoom</a>()</code> <br><code><a href="layer_repeat_vector.html">layer_repeat_vector</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_reshape.html">layer_reshape</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_rms_normalization.html">layer_rms_normalization</a>()</code> <br><code><a href="layer_rnn.html">layer_rnn</a>()</code> <br><code><a href="layer_separable_conv_1d.html">layer_separable_conv_1d</a>()</code> <br><code><a href="layer_separable_conv_2d.html">layer_separable_conv_2d</a>()</code> <br><code><a href="layer_simple_rnn.html">layer_simple_rnn</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br><code><a href="layer_spatial_dropout_1d.html">layer_spatial_dropout_1d</a>()</code> <br><code><a href="layer_spatial_dropout_2d.html">layer_spatial_dropout_2d</a>()</code> <br><code><a href="layer_spatial_dropout_3d.html">layer_spatial_dropout_3d</a>()</code> <br><code><a href="layer_spectral_normalization.html">layer_spectral_normalization</a>()</code> <br><code><a href="layer_stft_spectrogram.html">layer_stft_spectrogram</a>()</code> <br><code><a href="layer_string_lookup.html">layer_string_lookup</a>()</code> <br><code><a href="layer_subtract.html">layer_subtract</a>()</code> <br><code><a href="layer_text_vectorization.html">layer_text_vectorization</a>()</code> <br><code><a href="layer_tfsm.html">layer_tfsm</a>()</code> <br><code><a href="layer_time_distributed.html">layer_time_distributed</a>()</code> <br><code><a href="layer_torch_module_wrapper.html">layer_torch_module_wrapper</a>()</code> <br><code><a href="layer_unit_normalization.html">layer_unit_normalization</a>()</code> <br><code><a href="layer_upsampling_1d.html">layer_upsampling_1d</a>()</code> <br><code><a href="layer_upsampling_2d.html">layer_upsampling_2d</a>()</code> <br><code><a href="layer_upsampling_3d.html">layer_upsampling_3d</a>()</code> <br><code><a href="layer_zero_padding_1d.html">layer_zero_padding_1d</a>()</code> <br><code><a href="layer_zero_padding_2d.html">layer_zero_padding_2d</a>()</code> <br><code><a href="layer_zero_padding_3d.html">layer_zero_padding_3d</a>()</code> <br><code><a href="rnn_cell_gru.html">rnn_cell_gru</a>()</code> <br><code><a href="rnn_cell_lstm.html">rnn_cell_lstm</a>()</code> <br><code><a href="rnn_cell_simple.html">rnn_cell_simple</a>()</code> <br><code><a href="rnn_cells_stack.html">rnn_cells_stack</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -247,7 +249,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/layer_conv_lstm_1d.html b/docs/dev/reference/layer_conv_lstm_1d.html
index 4dcc38f27..f401b7c48 100644
--- a/docs/dev/reference/layer_conv_lstm_1d.html
+++ b/docs/dev/reference/layer_conv_lstm_1d.html
@@ -10,7 +10,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -316,7 +316,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
     <div class="dont-index">
 <ul><li><p><a href="https://keras.io/api/layers/recurrent_layers/conv_lstm1d#convlstm1d-class" class="external-link">https://keras.io/api/layers/recurrent_layers/conv_lstm1d#convlstm1d-class</a></p></li>
 </ul><p>Other rnn layers: <br><code><a href="layer_bidirectional.html">layer_bidirectional</a>()</code> <br><code><a href="layer_conv_lstm_2d.html">layer_conv_lstm_2d</a>()</code> <br><code><a href="layer_conv_lstm_3d.html">layer_conv_lstm_3d</a>()</code> <br><code><a href="layer_gru.html">layer_gru</a>()</code> <br><code><a href="layer_lstm.html">layer_lstm</a>()</code> <br><code><a href="layer_rnn.html">layer_rnn</a>()</code> <br><code><a href="layer_simple_rnn.html">layer_simple_rnn</a>()</code> <br><code><a href="layer_time_distributed.html">layer_time_distributed</a>()</code> <br><code><a href="rnn_cell_gru.html">rnn_cell_gru</a>()</code> <br><code><a href="rnn_cell_lstm.html">rnn_cell_lstm</a>()</code> <br><code><a href="rnn_cell_simple.html">rnn_cell_simple</a>()</code> <br><code><a href="rnn_cells_stack.html">rnn_cells_stack</a>()</code> <br></p>
-<p>Other layers: <br><code><a href="Layer.html">Layer</a>()</code> <br><code><a href="layer_activation.html">layer_activation</a>()</code> <br><code><a href="layer_activation_elu.html">layer_activation_elu</a>()</code> <br><code><a href="layer_activation_leaky_relu.html">layer_activation_leaky_relu</a>()</code> <br><code><a href="layer_activation_parametric_relu.html">layer_activation_parametric_relu</a>()</code> <br><code><a href="layer_activation_relu.html">layer_activation_relu</a>()</code> <br><code><a href="layer_activation_softmax.html">layer_activation_softmax</a>()</code> <br><code><a href="layer_activity_regularization.html">layer_activity_regularization</a>()</code> <br><code><a href="layer_add.html">layer_add</a>()</code> <br><code><a href="layer_additive_attention.html">layer_additive_attention</a>()</code> <br><code><a href="layer_alpha_dropout.html">layer_alpha_dropout</a>()</code> <br><code><a href="layer_attention.html">layer_attention</a>()</code> <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_average.html">layer_average</a>()</code> <br><code><a href="layer_average_pooling_1d.html">layer_average_pooling_1d</a>()</code> <br><code><a href="layer_average_pooling_2d.html">layer_average_pooling_2d</a>()</code> <br><code><a href="layer_average_pooling_3d.html">layer_average_pooling_3d</a>()</code> <br><code><a href="layer_batch_normalization.html">layer_batch_normalization</a>()</code> <br><code><a href="layer_bidirectional.html">layer_bidirectional</a>()</code> <br><code><a href="layer_category_encoding.html">layer_category_encoding</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_concatenate.html">layer_concatenate</a>()</code> <br><code><a href="layer_conv_1d.html">layer_conv_1d</a>()</code> <br><code><a href="layer_conv_1d_transpose.html">layer_conv_1d_transpose</a>()</code> <br><code><a href="layer_conv_2d.html">layer_conv_2d</a>()</code> <br><code><a href="layer_conv_2d_transpose.html">layer_conv_2d_transpose</a>()</code> <br><code><a href="layer_conv_3d.html">layer_conv_3d</a>()</code> <br><code><a href="layer_conv_3d_transpose.html">layer_conv_3d_transpose</a>()</code> <br><code><a href="layer_conv_lstm_2d.html">layer_conv_lstm_2d</a>()</code> <br><code><a href="layer_conv_lstm_3d.html">layer_conv_lstm_3d</a>()</code> <br><code><a href="layer_cropping_1d.html">layer_cropping_1d</a>()</code> <br><code><a href="layer_cropping_2d.html">layer_cropping_2d</a>()</code> <br><code><a href="layer_cropping_3d.html">layer_cropping_3d</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_dense.html">layer_dense</a>()</code> <br><code><a href="layer_depthwise_conv_1d.html">layer_depthwise_conv_1d</a>()</code> <br><code><a href="layer_depthwise_conv_2d.html">layer_depthwise_conv_2d</a>()</code> <br><code><a href="layer_discretization.html">layer_discretization</a>()</code> <br><code><a href="layer_dot.html">layer_dot</a>()</code> <br><code><a href="layer_dropout.html">layer_dropout</a>()</code> <br><code><a href="layer_einsum_dense.html">layer_einsum_dense</a>()</code> <br><code><a href="layer_embedding.html">layer_embedding</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_feature_space.html">layer_feature_space</a>()</code> <br><code><a href="layer_flatten.html">layer_flatten</a>()</code> <br><code><a href="layer_flax_module_wrapper.html">layer_flax_module_wrapper</a>()</code> <br><code><a href="layer_gaussian_dropout.html">layer_gaussian_dropout</a>()</code> <br><code><a href="layer_gaussian_noise.html">layer_gaussian_noise</a>()</code> <br><code><a href="layer_global_average_pooling_1d.html">layer_global_average_pooling_1d</a>()</code> <br><code><a href="layer_global_average_pooling_2d.html">layer_global_average_pooling_2d</a>()</code> <br><code><a href="layer_global_average_pooling_3d.html">layer_global_average_pooling_3d</a>()</code> <br><code><a href="layer_global_max_pooling_1d.html">layer_global_max_pooling_1d</a>()</code> <br><code><a href="layer_global_max_pooling_2d.html">layer_global_max_pooling_2d</a>()</code> <br><code><a href="layer_global_max_pooling_3d.html">layer_global_max_pooling_3d</a>()</code> <br><code><a href="layer_group_normalization.html">layer_group_normalization</a>()</code> <br><code><a href="layer_group_query_attention.html">layer_group_query_attention</a>()</code> <br><code><a href="layer_gru.html">layer_gru</a>()</code> <br><code><a href="layer_hashed_crossing.html">layer_hashed_crossing</a>()</code> <br><code><a href="layer_hashing.html">layer_hashing</a>()</code> <br><code><a href="layer_identity.html">layer_identity</a>()</code> <br><code><a href="layer_integer_lookup.html">layer_integer_lookup</a>()</code> <br><code><a href="layer_jax_model_wrapper.html">layer_jax_model_wrapper</a>()</code> <br><code><a href="layer_lambda.html">layer_lambda</a>()</code> <br><code><a href="layer_layer_normalization.html">layer_layer_normalization</a>()</code> <br><code><a href="layer_lstm.html">layer_lstm</a>()</code> <br><code><a href="layer_masking.html">layer_masking</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_max_pooling_1d.html">layer_max_pooling_1d</a>()</code> <br><code><a href="layer_max_pooling_2d.html">layer_max_pooling_2d</a>()</code> <br><code><a href="layer_max_pooling_3d.html">layer_max_pooling_3d</a>()</code> <br><code><a href="layer_maximum.html">layer_maximum</a>()</code> <br><code><a href="layer_mel_spectrogram.html">layer_mel_spectrogram</a>()</code> <br><code><a href="layer_minimum.html">layer_minimum</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_multi_head_attention.html">layer_multi_head_attention</a>()</code> <br><code><a href="layer_multiply.html">layer_multiply</a>()</code> <br><code><a href="layer_normalization.html">layer_normalization</a>()</code> <br><code><a href="layer_permute.html">layer_permute</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_brightness.html">layer_random_brightness</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_contrast.html">layer_random_contrast</a>()</code> <br><code><a href="layer_random_crop.html">layer_random_crop</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_flip.html">layer_random_flip</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_rotation.html">layer_random_rotation</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_random_translation.html">layer_random_translation</a>()</code> <br><code><a href="layer_random_zoom.html">layer_random_zoom</a>()</code> <br><code><a href="layer_repeat_vector.html">layer_repeat_vector</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_reshape.html">layer_reshape</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_rms_normalization.html">layer_rms_normalization</a>()</code> <br><code><a href="layer_rnn.html">layer_rnn</a>()</code> <br><code><a href="layer_separable_conv_1d.html">layer_separable_conv_1d</a>()</code> <br><code><a href="layer_separable_conv_2d.html">layer_separable_conv_2d</a>()</code> <br><code><a href="layer_simple_rnn.html">layer_simple_rnn</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br><code><a href="layer_spatial_dropout_1d.html">layer_spatial_dropout_1d</a>()</code> <br><code><a href="layer_spatial_dropout_2d.html">layer_spatial_dropout_2d</a>()</code> <br><code><a href="layer_spatial_dropout_3d.html">layer_spatial_dropout_3d</a>()</code> <br><code><a href="layer_spectral_normalization.html">layer_spectral_normalization</a>()</code> <br><code><a href="layer_stft_spectrogram.html">layer_stft_spectrogram</a>()</code> <br><code><a href="layer_string_lookup.html">layer_string_lookup</a>()</code> <br><code><a href="layer_subtract.html">layer_subtract</a>()</code> <br><code><a href="layer_text_vectorization.html">layer_text_vectorization</a>()</code> <br><code><a href="layer_tfsm.html">layer_tfsm</a>()</code> <br><code><a href="layer_time_distributed.html">layer_time_distributed</a>()</code> <br><code><a href="layer_torch_module_wrapper.html">layer_torch_module_wrapper</a>()</code> <br><code><a href="layer_unit_normalization.html">layer_unit_normalization</a>()</code> <br><code><a href="layer_upsampling_1d.html">layer_upsampling_1d</a>()</code> <br><code><a href="layer_upsampling_2d.html">layer_upsampling_2d</a>()</code> <br><code><a href="layer_upsampling_3d.html">layer_upsampling_3d</a>()</code> <br><code><a href="layer_zero_padding_1d.html">layer_zero_padding_1d</a>()</code> <br><code><a href="layer_zero_padding_2d.html">layer_zero_padding_2d</a>()</code> <br><code><a href="layer_zero_padding_3d.html">layer_zero_padding_3d</a>()</code> <br><code><a href="rnn_cell_gru.html">rnn_cell_gru</a>()</code> <br><code><a href="rnn_cell_lstm.html">rnn_cell_lstm</a>()</code> <br><code><a href="rnn_cell_simple.html">rnn_cell_simple</a>()</code> <br><code><a href="rnn_cells_stack.html">rnn_cells_stack</a>()</code> <br></p></div>
+<p>Other layers: <br><code><a href="Layer.html">Layer</a>()</code> <br><code><a href="layer_activation.html">layer_activation</a>()</code> <br><code><a href="layer_activation_elu.html">layer_activation_elu</a>()</code> <br><code><a href="layer_activation_leaky_relu.html">layer_activation_leaky_relu</a>()</code> <br><code><a href="layer_activation_parametric_relu.html">layer_activation_parametric_relu</a>()</code> <br><code><a href="layer_activation_relu.html">layer_activation_relu</a>()</code> <br><code><a href="layer_activation_softmax.html">layer_activation_softmax</a>()</code> <br><code><a href="layer_activity_regularization.html">layer_activity_regularization</a>()</code> <br><code><a href="layer_add.html">layer_add</a>()</code> <br><code><a href="layer_additive_attention.html">layer_additive_attention</a>()</code> <br><code><a href="layer_alpha_dropout.html">layer_alpha_dropout</a>()</code> <br><code><a href="layer_attention.html">layer_attention</a>()</code> <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_average.html">layer_average</a>()</code> <br><code><a href="layer_average_pooling_1d.html">layer_average_pooling_1d</a>()</code> <br><code><a href="layer_average_pooling_2d.html">layer_average_pooling_2d</a>()</code> <br><code><a href="layer_average_pooling_3d.html">layer_average_pooling_3d</a>()</code> <br><code><a href="layer_batch_normalization.html">layer_batch_normalization</a>()</code> <br><code><a href="layer_bidirectional.html">layer_bidirectional</a>()</code> <br><code><a href="layer_category_encoding.html">layer_category_encoding</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_concatenate.html">layer_concatenate</a>()</code> <br><code><a href="layer_conv_1d.html">layer_conv_1d</a>()</code> <br><code><a href="layer_conv_1d_transpose.html">layer_conv_1d_transpose</a>()</code> <br><code><a href="layer_conv_2d.html">layer_conv_2d</a>()</code> <br><code><a href="layer_conv_2d_transpose.html">layer_conv_2d_transpose</a>()</code> <br><code><a href="layer_conv_3d.html">layer_conv_3d</a>()</code> <br><code><a href="layer_conv_3d_transpose.html">layer_conv_3d_transpose</a>()</code> <br><code><a href="layer_conv_lstm_2d.html">layer_conv_lstm_2d</a>()</code> <br><code><a href="layer_conv_lstm_3d.html">layer_conv_lstm_3d</a>()</code> <br><code><a href="layer_cropping_1d.html">layer_cropping_1d</a>()</code> <br><code><a href="layer_cropping_2d.html">layer_cropping_2d</a>()</code> <br><code><a href="layer_cropping_3d.html">layer_cropping_3d</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_dense.html">layer_dense</a>()</code> <br><code><a href="layer_depthwise_conv_1d.html">layer_depthwise_conv_1d</a>()</code> <br><code><a href="layer_depthwise_conv_2d.html">layer_depthwise_conv_2d</a>()</code> <br><code><a href="layer_discretization.html">layer_discretization</a>()</code> <br><code><a href="layer_dot.html">layer_dot</a>()</code> <br><code><a href="layer_dropout.html">layer_dropout</a>()</code> <br><code><a href="layer_einsum_dense.html">layer_einsum_dense</a>()</code> <br><code><a href="layer_embedding.html">layer_embedding</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_feature_space.html">layer_feature_space</a>()</code> <br><code><a href="layer_flatten.html">layer_flatten</a>()</code> <br><code><a href="layer_flax_module_wrapper.html">layer_flax_module_wrapper</a>()</code> <br><code><a href="layer_gaussian_dropout.html">layer_gaussian_dropout</a>()</code> <br><code><a href="layer_gaussian_noise.html">layer_gaussian_noise</a>()</code> <br><code><a href="layer_global_average_pooling_1d.html">layer_global_average_pooling_1d</a>()</code> <br><code><a href="layer_global_average_pooling_2d.html">layer_global_average_pooling_2d</a>()</code> <br><code><a href="layer_global_average_pooling_3d.html">layer_global_average_pooling_3d</a>()</code> <br><code><a href="layer_global_max_pooling_1d.html">layer_global_max_pooling_1d</a>()</code> <br><code><a href="layer_global_max_pooling_2d.html">layer_global_max_pooling_2d</a>()</code> <br><code><a href="layer_global_max_pooling_3d.html">layer_global_max_pooling_3d</a>()</code> <br><code><a href="layer_group_normalization.html">layer_group_normalization</a>()</code> <br><code><a href="layer_group_query_attention.html">layer_group_query_attention</a>()</code> <br><code><a href="layer_gru.html">layer_gru</a>()</code> <br><code><a href="layer_hashed_crossing.html">layer_hashed_crossing</a>()</code> <br><code><a href="layer_hashing.html">layer_hashing</a>()</code> <br><code><a href="layer_identity.html">layer_identity</a>()</code> <br><code><a href="layer_integer_lookup.html">layer_integer_lookup</a>()</code> <br><code><a href="layer_jax_model_wrapper.html">layer_jax_model_wrapper</a>()</code> <br><code><a href="layer_lambda.html">layer_lambda</a>()</code> <br><code><a href="layer_layer_normalization.html">layer_layer_normalization</a>()</code> <br><code><a href="layer_lstm.html">layer_lstm</a>()</code> <br><code><a href="layer_masking.html">layer_masking</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_max_pooling_1d.html">layer_max_pooling_1d</a>()</code> <br><code><a href="layer_max_pooling_2d.html">layer_max_pooling_2d</a>()</code> <br><code><a href="layer_max_pooling_3d.html">layer_max_pooling_3d</a>()</code> <br><code><a href="layer_maximum.html">layer_maximum</a>()</code> <br><code><a href="layer_mel_spectrogram.html">layer_mel_spectrogram</a>()</code> <br><code><a href="layer_minimum.html">layer_minimum</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_multi_head_attention.html">layer_multi_head_attention</a>()</code> <br><code><a href="layer_multiply.html">layer_multiply</a>()</code> <br><code><a href="layer_normalization.html">layer_normalization</a>()</code> <br><code><a href="layer_permute.html">layer_permute</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_brightness.html">layer_random_brightness</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_contrast.html">layer_random_contrast</a>()</code> <br><code><a href="layer_random_crop.html">layer_random_crop</a>()</code> <br><code><a href="layer_random_elastic_transform.html">layer_random_elastic_transform</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_flip.html">layer_random_flip</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_rotation.html">layer_random_rotation</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_random_translation.html">layer_random_translation</a>()</code> <br><code><a href="layer_random_zoom.html">layer_random_zoom</a>()</code> <br><code><a href="layer_repeat_vector.html">layer_repeat_vector</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_reshape.html">layer_reshape</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_rms_normalization.html">layer_rms_normalization</a>()</code> <br><code><a href="layer_rnn.html">layer_rnn</a>()</code> <br><code><a href="layer_separable_conv_1d.html">layer_separable_conv_1d</a>()</code> <br><code><a href="layer_separable_conv_2d.html">layer_separable_conv_2d</a>()</code> <br><code><a href="layer_simple_rnn.html">layer_simple_rnn</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br><code><a href="layer_spatial_dropout_1d.html">layer_spatial_dropout_1d</a>()</code> <br><code><a href="layer_spatial_dropout_2d.html">layer_spatial_dropout_2d</a>()</code> <br><code><a href="layer_spatial_dropout_3d.html">layer_spatial_dropout_3d</a>()</code> <br><code><a href="layer_spectral_normalization.html">layer_spectral_normalization</a>()</code> <br><code><a href="layer_stft_spectrogram.html">layer_stft_spectrogram</a>()</code> <br><code><a href="layer_string_lookup.html">layer_string_lookup</a>()</code> <br><code><a href="layer_subtract.html">layer_subtract</a>()</code> <br><code><a href="layer_text_vectorization.html">layer_text_vectorization</a>()</code> <br><code><a href="layer_tfsm.html">layer_tfsm</a>()</code> <br><code><a href="layer_time_distributed.html">layer_time_distributed</a>()</code> <br><code><a href="layer_torch_module_wrapper.html">layer_torch_module_wrapper</a>()</code> <br><code><a href="layer_unit_normalization.html">layer_unit_normalization</a>()</code> <br><code><a href="layer_upsampling_1d.html">layer_upsampling_1d</a>()</code> <br><code><a href="layer_upsampling_2d.html">layer_upsampling_2d</a>()</code> <br><code><a href="layer_upsampling_3d.html">layer_upsampling_3d</a>()</code> <br><code><a href="layer_zero_padding_1d.html">layer_zero_padding_1d</a>()</code> <br><code><a href="layer_zero_padding_2d.html">layer_zero_padding_2d</a>()</code> <br><code><a href="layer_zero_padding_3d.html">layer_zero_padding_3d</a>()</code> <br><code><a href="rnn_cell_gru.html">rnn_cell_gru</a>()</code> <br><code><a href="rnn_cell_lstm.html">rnn_cell_lstm</a>()</code> <br><code><a href="rnn_cell_simple.html">rnn_cell_simple</a>()</code> <br><code><a href="rnn_cells_stack.html">rnn_cells_stack</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -328,7 +328,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/layer_conv_lstm_2d.html b/docs/dev/reference/layer_conv_lstm_2d.html
index b942624cf..8995d09eb 100644
--- a/docs/dev/reference/layer_conv_lstm_2d.html
+++ b/docs/dev/reference/layer_conv_lstm_2d.html
@@ -10,7 +10,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -316,7 +316,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
     <div class="dont-index">
 <ul><li><p><a href="https://keras.io/api/layers/recurrent_layers/conv_lstm2d#convlstm2d-class" class="external-link">https://keras.io/api/layers/recurrent_layers/conv_lstm2d#convlstm2d-class</a></p></li>
 </ul><p>Other rnn layers: <br><code><a href="layer_bidirectional.html">layer_bidirectional</a>()</code> <br><code><a href="layer_conv_lstm_1d.html">layer_conv_lstm_1d</a>()</code> <br><code><a href="layer_conv_lstm_3d.html">layer_conv_lstm_3d</a>()</code> <br><code><a href="layer_gru.html">layer_gru</a>()</code> <br><code><a href="layer_lstm.html">layer_lstm</a>()</code> <br><code><a href="layer_rnn.html">layer_rnn</a>()</code> <br><code><a href="layer_simple_rnn.html">layer_simple_rnn</a>()</code> <br><code><a href="layer_time_distributed.html">layer_time_distributed</a>()</code> <br><code><a href="rnn_cell_gru.html">rnn_cell_gru</a>()</code> <br><code><a href="rnn_cell_lstm.html">rnn_cell_lstm</a>()</code> <br><code><a href="rnn_cell_simple.html">rnn_cell_simple</a>()</code> <br><code><a href="rnn_cells_stack.html">rnn_cells_stack</a>()</code> <br></p>
-<p>Other layers: <br><code><a href="Layer.html">Layer</a>()</code> <br><code><a href="layer_activation.html">layer_activation</a>()</code> <br><code><a href="layer_activation_elu.html">layer_activation_elu</a>()</code> <br><code><a href="layer_activation_leaky_relu.html">layer_activation_leaky_relu</a>()</code> <br><code><a href="layer_activation_parametric_relu.html">layer_activation_parametric_relu</a>()</code> <br><code><a href="layer_activation_relu.html">layer_activation_relu</a>()</code> <br><code><a href="layer_activation_softmax.html">layer_activation_softmax</a>()</code> <br><code><a href="layer_activity_regularization.html">layer_activity_regularization</a>()</code> <br><code><a href="layer_add.html">layer_add</a>()</code> <br><code><a href="layer_additive_attention.html">layer_additive_attention</a>()</code> <br><code><a href="layer_alpha_dropout.html">layer_alpha_dropout</a>()</code> <br><code><a href="layer_attention.html">layer_attention</a>()</code> <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_average.html">layer_average</a>()</code> <br><code><a href="layer_average_pooling_1d.html">layer_average_pooling_1d</a>()</code> <br><code><a href="layer_average_pooling_2d.html">layer_average_pooling_2d</a>()</code> <br><code><a href="layer_average_pooling_3d.html">layer_average_pooling_3d</a>()</code> <br><code><a href="layer_batch_normalization.html">layer_batch_normalization</a>()</code> <br><code><a href="layer_bidirectional.html">layer_bidirectional</a>()</code> <br><code><a href="layer_category_encoding.html">layer_category_encoding</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_concatenate.html">layer_concatenate</a>()</code> <br><code><a href="layer_conv_1d.html">layer_conv_1d</a>()</code> <br><code><a href="layer_conv_1d_transpose.html">layer_conv_1d_transpose</a>()</code> <br><code><a href="layer_conv_2d.html">layer_conv_2d</a>()</code> <br><code><a href="layer_conv_2d_transpose.html">layer_conv_2d_transpose</a>()</code> <br><code><a href="layer_conv_3d.html">layer_conv_3d</a>()</code> <br><code><a href="layer_conv_3d_transpose.html">layer_conv_3d_transpose</a>()</code> <br><code><a href="layer_conv_lstm_1d.html">layer_conv_lstm_1d</a>()</code> <br><code><a href="layer_conv_lstm_3d.html">layer_conv_lstm_3d</a>()</code> <br><code><a href="layer_cropping_1d.html">layer_cropping_1d</a>()</code> <br><code><a href="layer_cropping_2d.html">layer_cropping_2d</a>()</code> <br><code><a href="layer_cropping_3d.html">layer_cropping_3d</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_dense.html">layer_dense</a>()</code> <br><code><a href="layer_depthwise_conv_1d.html">layer_depthwise_conv_1d</a>()</code> <br><code><a href="layer_depthwise_conv_2d.html">layer_depthwise_conv_2d</a>()</code> <br><code><a href="layer_discretization.html">layer_discretization</a>()</code> <br><code><a href="layer_dot.html">layer_dot</a>()</code> <br><code><a href="layer_dropout.html">layer_dropout</a>()</code> <br><code><a href="layer_einsum_dense.html">layer_einsum_dense</a>()</code> <br><code><a href="layer_embedding.html">layer_embedding</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_feature_space.html">layer_feature_space</a>()</code> <br><code><a href="layer_flatten.html">layer_flatten</a>()</code> <br><code><a href="layer_flax_module_wrapper.html">layer_flax_module_wrapper</a>()</code> <br><code><a href="layer_gaussian_dropout.html">layer_gaussian_dropout</a>()</code> <br><code><a href="layer_gaussian_noise.html">layer_gaussian_noise</a>()</code> <br><code><a href="layer_global_average_pooling_1d.html">layer_global_average_pooling_1d</a>()</code> <br><code><a href="layer_global_average_pooling_2d.html">layer_global_average_pooling_2d</a>()</code> <br><code><a href="layer_global_average_pooling_3d.html">layer_global_average_pooling_3d</a>()</code> <br><code><a href="layer_global_max_pooling_1d.html">layer_global_max_pooling_1d</a>()</code> <br><code><a href="layer_global_max_pooling_2d.html">layer_global_max_pooling_2d</a>()</code> <br><code><a href="layer_global_max_pooling_3d.html">layer_global_max_pooling_3d</a>()</code> <br><code><a href="layer_group_normalization.html">layer_group_normalization</a>()</code> <br><code><a href="layer_group_query_attention.html">layer_group_query_attention</a>()</code> <br><code><a href="layer_gru.html">layer_gru</a>()</code> <br><code><a href="layer_hashed_crossing.html">layer_hashed_crossing</a>()</code> <br><code><a href="layer_hashing.html">layer_hashing</a>()</code> <br><code><a href="layer_identity.html">layer_identity</a>()</code> <br><code><a href="layer_integer_lookup.html">layer_integer_lookup</a>()</code> <br><code><a href="layer_jax_model_wrapper.html">layer_jax_model_wrapper</a>()</code> <br><code><a href="layer_lambda.html">layer_lambda</a>()</code> <br><code><a href="layer_layer_normalization.html">layer_layer_normalization</a>()</code> <br><code><a href="layer_lstm.html">layer_lstm</a>()</code> <br><code><a href="layer_masking.html">layer_masking</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_max_pooling_1d.html">layer_max_pooling_1d</a>()</code> <br><code><a href="layer_max_pooling_2d.html">layer_max_pooling_2d</a>()</code> <br><code><a href="layer_max_pooling_3d.html">layer_max_pooling_3d</a>()</code> <br><code><a href="layer_maximum.html">layer_maximum</a>()</code> <br><code><a href="layer_mel_spectrogram.html">layer_mel_spectrogram</a>()</code> <br><code><a href="layer_minimum.html">layer_minimum</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_multi_head_attention.html">layer_multi_head_attention</a>()</code> <br><code><a href="layer_multiply.html">layer_multiply</a>()</code> <br><code><a href="layer_normalization.html">layer_normalization</a>()</code> <br><code><a href="layer_permute.html">layer_permute</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_brightness.html">layer_random_brightness</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_contrast.html">layer_random_contrast</a>()</code> <br><code><a href="layer_random_crop.html">layer_random_crop</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_flip.html">layer_random_flip</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_rotation.html">layer_random_rotation</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_random_translation.html">layer_random_translation</a>()</code> <br><code><a href="layer_random_zoom.html">layer_random_zoom</a>()</code> <br><code><a href="layer_repeat_vector.html">layer_repeat_vector</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_reshape.html">layer_reshape</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_rms_normalization.html">layer_rms_normalization</a>()</code> <br><code><a href="layer_rnn.html">layer_rnn</a>()</code> <br><code><a href="layer_separable_conv_1d.html">layer_separable_conv_1d</a>()</code> <br><code><a href="layer_separable_conv_2d.html">layer_separable_conv_2d</a>()</code> <br><code><a href="layer_simple_rnn.html">layer_simple_rnn</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br><code><a href="layer_spatial_dropout_1d.html">layer_spatial_dropout_1d</a>()</code> <br><code><a href="layer_spatial_dropout_2d.html">layer_spatial_dropout_2d</a>()</code> <br><code><a href="layer_spatial_dropout_3d.html">layer_spatial_dropout_3d</a>()</code> <br><code><a href="layer_spectral_normalization.html">layer_spectral_normalization</a>()</code> <br><code><a href="layer_stft_spectrogram.html">layer_stft_spectrogram</a>()</code> <br><code><a href="layer_string_lookup.html">layer_string_lookup</a>()</code> <br><code><a href="layer_subtract.html">layer_subtract</a>()</code> <br><code><a href="layer_text_vectorization.html">layer_text_vectorization</a>()</code> <br><code><a href="layer_tfsm.html">layer_tfsm</a>()</code> <br><code><a href="layer_time_distributed.html">layer_time_distributed</a>()</code> <br><code><a href="layer_torch_module_wrapper.html">layer_torch_module_wrapper</a>()</code> <br><code><a href="layer_unit_normalization.html">layer_unit_normalization</a>()</code> <br><code><a href="layer_upsampling_1d.html">layer_upsampling_1d</a>()</code> <br><code><a href="layer_upsampling_2d.html">layer_upsampling_2d</a>()</code> <br><code><a href="layer_upsampling_3d.html">layer_upsampling_3d</a>()</code> <br><code><a href="layer_zero_padding_1d.html">layer_zero_padding_1d</a>()</code> <br><code><a href="layer_zero_padding_2d.html">layer_zero_padding_2d</a>()</code> <br><code><a href="layer_zero_padding_3d.html">layer_zero_padding_3d</a>()</code> <br><code><a href="rnn_cell_gru.html">rnn_cell_gru</a>()</code> <br><code><a href="rnn_cell_lstm.html">rnn_cell_lstm</a>()</code> <br><code><a href="rnn_cell_simple.html">rnn_cell_simple</a>()</code> <br><code><a href="rnn_cells_stack.html">rnn_cells_stack</a>()</code> <br></p></div>
+<p>Other layers: <br><code><a href="Layer.html">Layer</a>()</code> <br><code><a href="layer_activation.html">layer_activation</a>()</code> <br><code><a href="layer_activation_elu.html">layer_activation_elu</a>()</code> <br><code><a href="layer_activation_leaky_relu.html">layer_activation_leaky_relu</a>()</code> <br><code><a href="layer_activation_parametric_relu.html">layer_activation_parametric_relu</a>()</code> <br><code><a href="layer_activation_relu.html">layer_activation_relu</a>()</code> <br><code><a href="layer_activation_softmax.html">layer_activation_softmax</a>()</code> <br><code><a href="layer_activity_regularization.html">layer_activity_regularization</a>()</code> <br><code><a href="layer_add.html">layer_add</a>()</code> <br><code><a href="layer_additive_attention.html">layer_additive_attention</a>()</code> <br><code><a href="layer_alpha_dropout.html">layer_alpha_dropout</a>()</code> <br><code><a href="layer_attention.html">layer_attention</a>()</code> <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_average.html">layer_average</a>()</code> <br><code><a href="layer_average_pooling_1d.html">layer_average_pooling_1d</a>()</code> <br><code><a href="layer_average_pooling_2d.html">layer_average_pooling_2d</a>()</code> <br><code><a href="layer_average_pooling_3d.html">layer_average_pooling_3d</a>()</code> <br><code><a href="layer_batch_normalization.html">layer_batch_normalization</a>()</code> <br><code><a href="layer_bidirectional.html">layer_bidirectional</a>()</code> <br><code><a href="layer_category_encoding.html">layer_category_encoding</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_concatenate.html">layer_concatenate</a>()</code> <br><code><a href="layer_conv_1d.html">layer_conv_1d</a>()</code> <br><code><a href="layer_conv_1d_transpose.html">layer_conv_1d_transpose</a>()</code> <br><code><a href="layer_conv_2d.html">layer_conv_2d</a>()</code> <br><code><a href="layer_conv_2d_transpose.html">layer_conv_2d_transpose</a>()</code> <br><code><a href="layer_conv_3d.html">layer_conv_3d</a>()</code> <br><code><a href="layer_conv_3d_transpose.html">layer_conv_3d_transpose</a>()</code> <br><code><a href="layer_conv_lstm_1d.html">layer_conv_lstm_1d</a>()</code> <br><code><a href="layer_conv_lstm_3d.html">layer_conv_lstm_3d</a>()</code> <br><code><a href="layer_cropping_1d.html">layer_cropping_1d</a>()</code> <br><code><a href="layer_cropping_2d.html">layer_cropping_2d</a>()</code> <br><code><a href="layer_cropping_3d.html">layer_cropping_3d</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_dense.html">layer_dense</a>()</code> <br><code><a href="layer_depthwise_conv_1d.html">layer_depthwise_conv_1d</a>()</code> <br><code><a href="layer_depthwise_conv_2d.html">layer_depthwise_conv_2d</a>()</code> <br><code><a href="layer_discretization.html">layer_discretization</a>()</code> <br><code><a href="layer_dot.html">layer_dot</a>()</code> <br><code><a href="layer_dropout.html">layer_dropout</a>()</code> <br><code><a href="layer_einsum_dense.html">layer_einsum_dense</a>()</code> <br><code><a href="layer_embedding.html">layer_embedding</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_feature_space.html">layer_feature_space</a>()</code> <br><code><a href="layer_flatten.html">layer_flatten</a>()</code> <br><code><a href="layer_flax_module_wrapper.html">layer_flax_module_wrapper</a>()</code> <br><code><a href="layer_gaussian_dropout.html">layer_gaussian_dropout</a>()</code> <br><code><a href="layer_gaussian_noise.html">layer_gaussian_noise</a>()</code> <br><code><a href="layer_global_average_pooling_1d.html">layer_global_average_pooling_1d</a>()</code> <br><code><a href="layer_global_average_pooling_2d.html">layer_global_average_pooling_2d</a>()</code> <br><code><a href="layer_global_average_pooling_3d.html">layer_global_average_pooling_3d</a>()</code> <br><code><a href="layer_global_max_pooling_1d.html">layer_global_max_pooling_1d</a>()</code> <br><code><a href="layer_global_max_pooling_2d.html">layer_global_max_pooling_2d</a>()</code> <br><code><a href="layer_global_max_pooling_3d.html">layer_global_max_pooling_3d</a>()</code> <br><code><a href="layer_group_normalization.html">layer_group_normalization</a>()</code> <br><code><a href="layer_group_query_attention.html">layer_group_query_attention</a>()</code> <br><code><a href="layer_gru.html">layer_gru</a>()</code> <br><code><a href="layer_hashed_crossing.html">layer_hashed_crossing</a>()</code> <br><code><a href="layer_hashing.html">layer_hashing</a>()</code> <br><code><a href="layer_identity.html">layer_identity</a>()</code> <br><code><a href="layer_integer_lookup.html">layer_integer_lookup</a>()</code> <br><code><a href="layer_jax_model_wrapper.html">layer_jax_model_wrapper</a>()</code> <br><code><a href="layer_lambda.html">layer_lambda</a>()</code> <br><code><a href="layer_layer_normalization.html">layer_layer_normalization</a>()</code> <br><code><a href="layer_lstm.html">layer_lstm</a>()</code> <br><code><a href="layer_masking.html">layer_masking</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_max_pooling_1d.html">layer_max_pooling_1d</a>()</code> <br><code><a href="layer_max_pooling_2d.html">layer_max_pooling_2d</a>()</code> <br><code><a href="layer_max_pooling_3d.html">layer_max_pooling_3d</a>()</code> <br><code><a href="layer_maximum.html">layer_maximum</a>()</code> <br><code><a href="layer_mel_spectrogram.html">layer_mel_spectrogram</a>()</code> <br><code><a href="layer_minimum.html">layer_minimum</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_multi_head_attention.html">layer_multi_head_attention</a>()</code> <br><code><a href="layer_multiply.html">layer_multiply</a>()</code> <br><code><a href="layer_normalization.html">layer_normalization</a>()</code> <br><code><a href="layer_permute.html">layer_permute</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_brightness.html">layer_random_brightness</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_contrast.html">layer_random_contrast</a>()</code> <br><code><a href="layer_random_crop.html">layer_random_crop</a>()</code> <br><code><a href="layer_random_elastic_transform.html">layer_random_elastic_transform</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_flip.html">layer_random_flip</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_rotation.html">layer_random_rotation</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_random_translation.html">layer_random_translation</a>()</code> <br><code><a href="layer_random_zoom.html">layer_random_zoom</a>()</code> <br><code><a href="layer_repeat_vector.html">layer_repeat_vector</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_reshape.html">layer_reshape</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_rms_normalization.html">layer_rms_normalization</a>()</code> <br><code><a href="layer_rnn.html">layer_rnn</a>()</code> <br><code><a href="layer_separable_conv_1d.html">layer_separable_conv_1d</a>()</code> <br><code><a href="layer_separable_conv_2d.html">layer_separable_conv_2d</a>()</code> <br><code><a href="layer_simple_rnn.html">layer_simple_rnn</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br><code><a href="layer_spatial_dropout_1d.html">layer_spatial_dropout_1d</a>()</code> <br><code><a href="layer_spatial_dropout_2d.html">layer_spatial_dropout_2d</a>()</code> <br><code><a href="layer_spatial_dropout_3d.html">layer_spatial_dropout_3d</a>()</code> <br><code><a href="layer_spectral_normalization.html">layer_spectral_normalization</a>()</code> <br><code><a href="layer_stft_spectrogram.html">layer_stft_spectrogram</a>()</code> <br><code><a href="layer_string_lookup.html">layer_string_lookup</a>()</code> <br><code><a href="layer_subtract.html">layer_subtract</a>()</code> <br><code><a href="layer_text_vectorization.html">layer_text_vectorization</a>()</code> <br><code><a href="layer_tfsm.html">layer_tfsm</a>()</code> <br><code><a href="layer_time_distributed.html">layer_time_distributed</a>()</code> <br><code><a href="layer_torch_module_wrapper.html">layer_torch_module_wrapper</a>()</code> <br><code><a href="layer_unit_normalization.html">layer_unit_normalization</a>()</code> <br><code><a href="layer_upsampling_1d.html">layer_upsampling_1d</a>()</code> <br><code><a href="layer_upsampling_2d.html">layer_upsampling_2d</a>()</code> <br><code><a href="layer_upsampling_3d.html">layer_upsampling_3d</a>()</code> <br><code><a href="layer_zero_padding_1d.html">layer_zero_padding_1d</a>()</code> <br><code><a href="layer_zero_padding_2d.html">layer_zero_padding_2d</a>()</code> <br><code><a href="layer_zero_padding_3d.html">layer_zero_padding_3d</a>()</code> <br><code><a href="rnn_cell_gru.html">rnn_cell_gru</a>()</code> <br><code><a href="rnn_cell_lstm.html">rnn_cell_lstm</a>()</code> <br><code><a href="rnn_cell_simple.html">rnn_cell_simple</a>()</code> <br><code><a href="rnn_cells_stack.html">rnn_cells_stack</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -328,7 +328,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/layer_conv_lstm_3d.html b/docs/dev/reference/layer_conv_lstm_3d.html
index b674d0d4f..ece9ac679 100644
--- a/docs/dev/reference/layer_conv_lstm_3d.html
+++ b/docs/dev/reference/layer_conv_lstm_3d.html
@@ -10,7 +10,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -315,7 +315,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
     <div class="dont-index">
 <ul><li><p><a href="https://keras.io/api/layers/recurrent_layers/conv_lstm3d#convlstm3d-class" class="external-link">https://keras.io/api/layers/recurrent_layers/conv_lstm3d#convlstm3d-class</a></p></li>
 </ul><p>Other rnn layers: <br><code><a href="layer_bidirectional.html">layer_bidirectional</a>()</code> <br><code><a href="layer_conv_lstm_1d.html">layer_conv_lstm_1d</a>()</code> <br><code><a href="layer_conv_lstm_2d.html">layer_conv_lstm_2d</a>()</code> <br><code><a href="layer_gru.html">layer_gru</a>()</code> <br><code><a href="layer_lstm.html">layer_lstm</a>()</code> <br><code><a href="layer_rnn.html">layer_rnn</a>()</code> <br><code><a href="layer_simple_rnn.html">layer_simple_rnn</a>()</code> <br><code><a href="layer_time_distributed.html">layer_time_distributed</a>()</code> <br><code><a href="rnn_cell_gru.html">rnn_cell_gru</a>()</code> <br><code><a href="rnn_cell_lstm.html">rnn_cell_lstm</a>()</code> <br><code><a href="rnn_cell_simple.html">rnn_cell_simple</a>()</code> <br><code><a href="rnn_cells_stack.html">rnn_cells_stack</a>()</code> <br></p>
-<p>Other layers: <br><code><a href="Layer.html">Layer</a>()</code> <br><code><a href="layer_activation.html">layer_activation</a>()</code> <br><code><a href="layer_activation_elu.html">layer_activation_elu</a>()</code> <br><code><a href="layer_activation_leaky_relu.html">layer_activation_leaky_relu</a>()</code> <br><code><a href="layer_activation_parametric_relu.html">layer_activation_parametric_relu</a>()</code> <br><code><a href="layer_activation_relu.html">layer_activation_relu</a>()</code> <br><code><a href="layer_activation_softmax.html">layer_activation_softmax</a>()</code> <br><code><a href="layer_activity_regularization.html">layer_activity_regularization</a>()</code> <br><code><a href="layer_add.html">layer_add</a>()</code> <br><code><a href="layer_additive_attention.html">layer_additive_attention</a>()</code> <br><code><a href="layer_alpha_dropout.html">layer_alpha_dropout</a>()</code> <br><code><a href="layer_attention.html">layer_attention</a>()</code> <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_average.html">layer_average</a>()</code> <br><code><a href="layer_average_pooling_1d.html">layer_average_pooling_1d</a>()</code> <br><code><a href="layer_average_pooling_2d.html">layer_average_pooling_2d</a>()</code> <br><code><a href="layer_average_pooling_3d.html">layer_average_pooling_3d</a>()</code> <br><code><a href="layer_batch_normalization.html">layer_batch_normalization</a>()</code> <br><code><a href="layer_bidirectional.html">layer_bidirectional</a>()</code> <br><code><a href="layer_category_encoding.html">layer_category_encoding</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_concatenate.html">layer_concatenate</a>()</code> <br><code><a href="layer_conv_1d.html">layer_conv_1d</a>()</code> <br><code><a href="layer_conv_1d_transpose.html">layer_conv_1d_transpose</a>()</code> <br><code><a href="layer_conv_2d.html">layer_conv_2d</a>()</code> <br><code><a href="layer_conv_2d_transpose.html">layer_conv_2d_transpose</a>()</code> <br><code><a href="layer_conv_3d.html">layer_conv_3d</a>()</code> <br><code><a href="layer_conv_3d_transpose.html">layer_conv_3d_transpose</a>()</code> <br><code><a href="layer_conv_lstm_1d.html">layer_conv_lstm_1d</a>()</code> <br><code><a href="layer_conv_lstm_2d.html">layer_conv_lstm_2d</a>()</code> <br><code><a href="layer_cropping_1d.html">layer_cropping_1d</a>()</code> <br><code><a href="layer_cropping_2d.html">layer_cropping_2d</a>()</code> <br><code><a href="layer_cropping_3d.html">layer_cropping_3d</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_dense.html">layer_dense</a>()</code> <br><code><a href="layer_depthwise_conv_1d.html">layer_depthwise_conv_1d</a>()</code> <br><code><a href="layer_depthwise_conv_2d.html">layer_depthwise_conv_2d</a>()</code> <br><code><a href="layer_discretization.html">layer_discretization</a>()</code> <br><code><a href="layer_dot.html">layer_dot</a>()</code> <br><code><a href="layer_dropout.html">layer_dropout</a>()</code> <br><code><a href="layer_einsum_dense.html">layer_einsum_dense</a>()</code> <br><code><a href="layer_embedding.html">layer_embedding</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_feature_space.html">layer_feature_space</a>()</code> <br><code><a href="layer_flatten.html">layer_flatten</a>()</code> <br><code><a href="layer_flax_module_wrapper.html">layer_flax_module_wrapper</a>()</code> <br><code><a href="layer_gaussian_dropout.html">layer_gaussian_dropout</a>()</code> <br><code><a href="layer_gaussian_noise.html">layer_gaussian_noise</a>()</code> <br><code><a href="layer_global_average_pooling_1d.html">layer_global_average_pooling_1d</a>()</code> <br><code><a href="layer_global_average_pooling_2d.html">layer_global_average_pooling_2d</a>()</code> <br><code><a href="layer_global_average_pooling_3d.html">layer_global_average_pooling_3d</a>()</code> <br><code><a href="layer_global_max_pooling_1d.html">layer_global_max_pooling_1d</a>()</code> <br><code><a href="layer_global_max_pooling_2d.html">layer_global_max_pooling_2d</a>()</code> <br><code><a href="layer_global_max_pooling_3d.html">layer_global_max_pooling_3d</a>()</code> <br><code><a href="layer_group_normalization.html">layer_group_normalization</a>()</code> <br><code><a href="layer_group_query_attention.html">layer_group_query_attention</a>()</code> <br><code><a href="layer_gru.html">layer_gru</a>()</code> <br><code><a href="layer_hashed_crossing.html">layer_hashed_crossing</a>()</code> <br><code><a href="layer_hashing.html">layer_hashing</a>()</code> <br><code><a href="layer_identity.html">layer_identity</a>()</code> <br><code><a href="layer_integer_lookup.html">layer_integer_lookup</a>()</code> <br><code><a href="layer_jax_model_wrapper.html">layer_jax_model_wrapper</a>()</code> <br><code><a href="layer_lambda.html">layer_lambda</a>()</code> <br><code><a href="layer_layer_normalization.html">layer_layer_normalization</a>()</code> <br><code><a href="layer_lstm.html">layer_lstm</a>()</code> <br><code><a href="layer_masking.html">layer_masking</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_max_pooling_1d.html">layer_max_pooling_1d</a>()</code> <br><code><a href="layer_max_pooling_2d.html">layer_max_pooling_2d</a>()</code> <br><code><a href="layer_max_pooling_3d.html">layer_max_pooling_3d</a>()</code> <br><code><a href="layer_maximum.html">layer_maximum</a>()</code> <br><code><a href="layer_mel_spectrogram.html">layer_mel_spectrogram</a>()</code> <br><code><a href="layer_minimum.html">layer_minimum</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_multi_head_attention.html">layer_multi_head_attention</a>()</code> <br><code><a href="layer_multiply.html">layer_multiply</a>()</code> <br><code><a href="layer_normalization.html">layer_normalization</a>()</code> <br><code><a href="layer_permute.html">layer_permute</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_brightness.html">layer_random_brightness</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_contrast.html">layer_random_contrast</a>()</code> <br><code><a href="layer_random_crop.html">layer_random_crop</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_flip.html">layer_random_flip</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_rotation.html">layer_random_rotation</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_random_translation.html">layer_random_translation</a>()</code> <br><code><a href="layer_random_zoom.html">layer_random_zoom</a>()</code> <br><code><a href="layer_repeat_vector.html">layer_repeat_vector</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_reshape.html">layer_reshape</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_rms_normalization.html">layer_rms_normalization</a>()</code> <br><code><a href="layer_rnn.html">layer_rnn</a>()</code> <br><code><a href="layer_separable_conv_1d.html">layer_separable_conv_1d</a>()</code> <br><code><a href="layer_separable_conv_2d.html">layer_separable_conv_2d</a>()</code> <br><code><a href="layer_simple_rnn.html">layer_simple_rnn</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br><code><a href="layer_spatial_dropout_1d.html">layer_spatial_dropout_1d</a>()</code> <br><code><a href="layer_spatial_dropout_2d.html">layer_spatial_dropout_2d</a>()</code> <br><code><a href="layer_spatial_dropout_3d.html">layer_spatial_dropout_3d</a>()</code> <br><code><a href="layer_spectral_normalization.html">layer_spectral_normalization</a>()</code> <br><code><a href="layer_stft_spectrogram.html">layer_stft_spectrogram</a>()</code> <br><code><a href="layer_string_lookup.html">layer_string_lookup</a>()</code> <br><code><a href="layer_subtract.html">layer_subtract</a>()</code> <br><code><a href="layer_text_vectorization.html">layer_text_vectorization</a>()</code> <br><code><a href="layer_tfsm.html">layer_tfsm</a>()</code> <br><code><a href="layer_time_distributed.html">layer_time_distributed</a>()</code> <br><code><a href="layer_torch_module_wrapper.html">layer_torch_module_wrapper</a>()</code> <br><code><a href="layer_unit_normalization.html">layer_unit_normalization</a>()</code> <br><code><a href="layer_upsampling_1d.html">layer_upsampling_1d</a>()</code> <br><code><a href="layer_upsampling_2d.html">layer_upsampling_2d</a>()</code> <br><code><a href="layer_upsampling_3d.html">layer_upsampling_3d</a>()</code> <br><code><a href="layer_zero_padding_1d.html">layer_zero_padding_1d</a>()</code> <br><code><a href="layer_zero_padding_2d.html">layer_zero_padding_2d</a>()</code> <br><code><a href="layer_zero_padding_3d.html">layer_zero_padding_3d</a>()</code> <br><code><a href="rnn_cell_gru.html">rnn_cell_gru</a>()</code> <br><code><a href="rnn_cell_lstm.html">rnn_cell_lstm</a>()</code> <br><code><a href="rnn_cell_simple.html">rnn_cell_simple</a>()</code> <br><code><a href="rnn_cells_stack.html">rnn_cells_stack</a>()</code> <br></p></div>
+<p>Other layers: <br><code><a href="Layer.html">Layer</a>()</code> <br><code><a href="layer_activation.html">layer_activation</a>()</code> <br><code><a href="layer_activation_elu.html">layer_activation_elu</a>()</code> <br><code><a href="layer_activation_leaky_relu.html">layer_activation_leaky_relu</a>()</code> <br><code><a href="layer_activation_parametric_relu.html">layer_activation_parametric_relu</a>()</code> <br><code><a href="layer_activation_relu.html">layer_activation_relu</a>()</code> <br><code><a href="layer_activation_softmax.html">layer_activation_softmax</a>()</code> <br><code><a href="layer_activity_regularization.html">layer_activity_regularization</a>()</code> <br><code><a href="layer_add.html">layer_add</a>()</code> <br><code><a href="layer_additive_attention.html">layer_additive_attention</a>()</code> <br><code><a href="layer_alpha_dropout.html">layer_alpha_dropout</a>()</code> <br><code><a href="layer_attention.html">layer_attention</a>()</code> <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_average.html">layer_average</a>()</code> <br><code><a href="layer_average_pooling_1d.html">layer_average_pooling_1d</a>()</code> <br><code><a href="layer_average_pooling_2d.html">layer_average_pooling_2d</a>()</code> <br><code><a href="layer_average_pooling_3d.html">layer_average_pooling_3d</a>()</code> <br><code><a href="layer_batch_normalization.html">layer_batch_normalization</a>()</code> <br><code><a href="layer_bidirectional.html">layer_bidirectional</a>()</code> <br><code><a href="layer_category_encoding.html">layer_category_encoding</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_concatenate.html">layer_concatenate</a>()</code> <br><code><a href="layer_conv_1d.html">layer_conv_1d</a>()</code> <br><code><a href="layer_conv_1d_transpose.html">layer_conv_1d_transpose</a>()</code> <br><code><a href="layer_conv_2d.html">layer_conv_2d</a>()</code> <br><code><a href="layer_conv_2d_transpose.html">layer_conv_2d_transpose</a>()</code> <br><code><a href="layer_conv_3d.html">layer_conv_3d</a>()</code> <br><code><a href="layer_conv_3d_transpose.html">layer_conv_3d_transpose</a>()</code> <br><code><a href="layer_conv_lstm_1d.html">layer_conv_lstm_1d</a>()</code> <br><code><a href="layer_conv_lstm_2d.html">layer_conv_lstm_2d</a>()</code> <br><code><a href="layer_cropping_1d.html">layer_cropping_1d</a>()</code> <br><code><a href="layer_cropping_2d.html">layer_cropping_2d</a>()</code> <br><code><a href="layer_cropping_3d.html">layer_cropping_3d</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_dense.html">layer_dense</a>()</code> <br><code><a href="layer_depthwise_conv_1d.html">layer_depthwise_conv_1d</a>()</code> <br><code><a href="layer_depthwise_conv_2d.html">layer_depthwise_conv_2d</a>()</code> <br><code><a href="layer_discretization.html">layer_discretization</a>()</code> <br><code><a href="layer_dot.html">layer_dot</a>()</code> <br><code><a href="layer_dropout.html">layer_dropout</a>()</code> <br><code><a href="layer_einsum_dense.html">layer_einsum_dense</a>()</code> <br><code><a href="layer_embedding.html">layer_embedding</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_feature_space.html">layer_feature_space</a>()</code> <br><code><a href="layer_flatten.html">layer_flatten</a>()</code> <br><code><a href="layer_flax_module_wrapper.html">layer_flax_module_wrapper</a>()</code> <br><code><a href="layer_gaussian_dropout.html">layer_gaussian_dropout</a>()</code> <br><code><a href="layer_gaussian_noise.html">layer_gaussian_noise</a>()</code> <br><code><a href="layer_global_average_pooling_1d.html">layer_global_average_pooling_1d</a>()</code> <br><code><a href="layer_global_average_pooling_2d.html">layer_global_average_pooling_2d</a>()</code> <br><code><a href="layer_global_average_pooling_3d.html">layer_global_average_pooling_3d</a>()</code> <br><code><a href="layer_global_max_pooling_1d.html">layer_global_max_pooling_1d</a>()</code> <br><code><a href="layer_global_max_pooling_2d.html">layer_global_max_pooling_2d</a>()</code> <br><code><a href="layer_global_max_pooling_3d.html">layer_global_max_pooling_3d</a>()</code> <br><code><a href="layer_group_normalization.html">layer_group_normalization</a>()</code> <br><code><a href="layer_group_query_attention.html">layer_group_query_attention</a>()</code> <br><code><a href="layer_gru.html">layer_gru</a>()</code> <br><code><a href="layer_hashed_crossing.html">layer_hashed_crossing</a>()</code> <br><code><a href="layer_hashing.html">layer_hashing</a>()</code> <br><code><a href="layer_identity.html">layer_identity</a>()</code> <br><code><a href="layer_integer_lookup.html">layer_integer_lookup</a>()</code> <br><code><a href="layer_jax_model_wrapper.html">layer_jax_model_wrapper</a>()</code> <br><code><a href="layer_lambda.html">layer_lambda</a>()</code> <br><code><a href="layer_layer_normalization.html">layer_layer_normalization</a>()</code> <br><code><a href="layer_lstm.html">layer_lstm</a>()</code> <br><code><a href="layer_masking.html">layer_masking</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_max_pooling_1d.html">layer_max_pooling_1d</a>()</code> <br><code><a href="layer_max_pooling_2d.html">layer_max_pooling_2d</a>()</code> <br><code><a href="layer_max_pooling_3d.html">layer_max_pooling_3d</a>()</code> <br><code><a href="layer_maximum.html">layer_maximum</a>()</code> <br><code><a href="layer_mel_spectrogram.html">layer_mel_spectrogram</a>()</code> <br><code><a href="layer_minimum.html">layer_minimum</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_multi_head_attention.html">layer_multi_head_attention</a>()</code> <br><code><a href="layer_multiply.html">layer_multiply</a>()</code> <br><code><a href="layer_normalization.html">layer_normalization</a>()</code> <br><code><a href="layer_permute.html">layer_permute</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_brightness.html">layer_random_brightness</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_contrast.html">layer_random_contrast</a>()</code> <br><code><a href="layer_random_crop.html">layer_random_crop</a>()</code> <br><code><a href="layer_random_elastic_transform.html">layer_random_elastic_transform</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_flip.html">layer_random_flip</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_rotation.html">layer_random_rotation</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_random_translation.html">layer_random_translation</a>()</code> <br><code><a href="layer_random_zoom.html">layer_random_zoom</a>()</code> <br><code><a href="layer_repeat_vector.html">layer_repeat_vector</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_reshape.html">layer_reshape</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_rms_normalization.html">layer_rms_normalization</a>()</code> <br><code><a href="layer_rnn.html">layer_rnn</a>()</code> <br><code><a href="layer_separable_conv_1d.html">layer_separable_conv_1d</a>()</code> <br><code><a href="layer_separable_conv_2d.html">layer_separable_conv_2d</a>()</code> <br><code><a href="layer_simple_rnn.html">layer_simple_rnn</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br><code><a href="layer_spatial_dropout_1d.html">layer_spatial_dropout_1d</a>()</code> <br><code><a href="layer_spatial_dropout_2d.html">layer_spatial_dropout_2d</a>()</code> <br><code><a href="layer_spatial_dropout_3d.html">layer_spatial_dropout_3d</a>()</code> <br><code><a href="layer_spectral_normalization.html">layer_spectral_normalization</a>()</code> <br><code><a href="layer_stft_spectrogram.html">layer_stft_spectrogram</a>()</code> <br><code><a href="layer_string_lookup.html">layer_string_lookup</a>()</code> <br><code><a href="layer_subtract.html">layer_subtract</a>()</code> <br><code><a href="layer_text_vectorization.html">layer_text_vectorization</a>()</code> <br><code><a href="layer_tfsm.html">layer_tfsm</a>()</code> <br><code><a href="layer_time_distributed.html">layer_time_distributed</a>()</code> <br><code><a href="layer_torch_module_wrapper.html">layer_torch_module_wrapper</a>()</code> <br><code><a href="layer_unit_normalization.html">layer_unit_normalization</a>()</code> <br><code><a href="layer_upsampling_1d.html">layer_upsampling_1d</a>()</code> <br><code><a href="layer_upsampling_2d.html">layer_upsampling_2d</a>()</code> <br><code><a href="layer_upsampling_3d.html">layer_upsampling_3d</a>()</code> <br><code><a href="layer_zero_padding_1d.html">layer_zero_padding_1d</a>()</code> <br><code><a href="layer_zero_padding_2d.html">layer_zero_padding_2d</a>()</code> <br><code><a href="layer_zero_padding_3d.html">layer_zero_padding_3d</a>()</code> <br><code><a href="rnn_cell_gru.html">rnn_cell_gru</a>()</code> <br><code><a href="rnn_cell_lstm.html">rnn_cell_lstm</a>()</code> <br><code><a href="rnn_cell_simple.html">rnn_cell_simple</a>()</code> <br><code><a href="rnn_cells_stack.html">rnn_cells_stack</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -327,7 +327,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/layer_cropping_1d.html b/docs/dev/reference/layer_cropping_1d.html
index fb0e674a8..8fac711aa 100644
--- a/docs/dev/reference/layer_cropping_1d.html
+++ b/docs/dev/reference/layer_cropping_1d.html
@@ -8,7 +8,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -127,7 +127,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
     <div class="dont-index">
 <ul><li><p><a href="https://keras.io/api/layers/reshaping_layers/cropping1d#cropping1d-class" class="external-link">https://keras.io/api/layers/reshaping_layers/cropping1d#cropping1d-class</a></p></li>
 </ul><p>Other reshaping layers: <br><code><a href="layer_cropping_2d.html">layer_cropping_2d</a>()</code> <br><code><a href="layer_cropping_3d.html">layer_cropping_3d</a>()</code> <br><code><a href="layer_flatten.html">layer_flatten</a>()</code> <br><code><a href="layer_permute.html">layer_permute</a>()</code> <br><code><a href="layer_repeat_vector.html">layer_repeat_vector</a>()</code> <br><code><a href="layer_reshape.html">layer_reshape</a>()</code> <br><code><a href="layer_upsampling_1d.html">layer_upsampling_1d</a>()</code> <br><code><a href="layer_upsampling_2d.html">layer_upsampling_2d</a>()</code> <br><code><a href="layer_upsampling_3d.html">layer_upsampling_3d</a>()</code> <br><code><a href="layer_zero_padding_1d.html">layer_zero_padding_1d</a>()</code> <br><code><a href="layer_zero_padding_2d.html">layer_zero_padding_2d</a>()</code> <br><code><a href="layer_zero_padding_3d.html">layer_zero_padding_3d</a>()</code> <br></p>
-<p>Other layers: <br><code><a href="Layer.html">Layer</a>()</code> <br><code><a href="layer_activation.html">layer_activation</a>()</code> <br><code><a href="layer_activation_elu.html">layer_activation_elu</a>()</code> <br><code><a href="layer_activation_leaky_relu.html">layer_activation_leaky_relu</a>()</code> <br><code><a href="layer_activation_parametric_relu.html">layer_activation_parametric_relu</a>()</code> <br><code><a href="layer_activation_relu.html">layer_activation_relu</a>()</code> <br><code><a href="layer_activation_softmax.html">layer_activation_softmax</a>()</code> <br><code><a href="layer_activity_regularization.html">layer_activity_regularization</a>()</code> <br><code><a href="layer_add.html">layer_add</a>()</code> <br><code><a href="layer_additive_attention.html">layer_additive_attention</a>()</code> <br><code><a href="layer_alpha_dropout.html">layer_alpha_dropout</a>()</code> <br><code><a href="layer_attention.html">layer_attention</a>()</code> <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_average.html">layer_average</a>()</code> <br><code><a href="layer_average_pooling_1d.html">layer_average_pooling_1d</a>()</code> <br><code><a href="layer_average_pooling_2d.html">layer_average_pooling_2d</a>()</code> <br><code><a href="layer_average_pooling_3d.html">layer_average_pooling_3d</a>()</code> <br><code><a href="layer_batch_normalization.html">layer_batch_normalization</a>()</code> <br><code><a href="layer_bidirectional.html">layer_bidirectional</a>()</code> <br><code><a href="layer_category_encoding.html">layer_category_encoding</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_concatenate.html">layer_concatenate</a>()</code> <br><code><a href="layer_conv_1d.html">layer_conv_1d</a>()</code> <br><code><a href="layer_conv_1d_transpose.html">layer_conv_1d_transpose</a>()</code> <br><code><a href="layer_conv_2d.html">layer_conv_2d</a>()</code> <br><code><a href="layer_conv_2d_transpose.html">layer_conv_2d_transpose</a>()</code> <br><code><a href="layer_conv_3d.html">layer_conv_3d</a>()</code> <br><code><a href="layer_conv_3d_transpose.html">layer_conv_3d_transpose</a>()</code> <br><code><a href="layer_conv_lstm_1d.html">layer_conv_lstm_1d</a>()</code> <br><code><a href="layer_conv_lstm_2d.html">layer_conv_lstm_2d</a>()</code> <br><code><a href="layer_conv_lstm_3d.html">layer_conv_lstm_3d</a>()</code> <br><code><a href="layer_cropping_2d.html">layer_cropping_2d</a>()</code> <br><code><a href="layer_cropping_3d.html">layer_cropping_3d</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_dense.html">layer_dense</a>()</code> <br><code><a href="layer_depthwise_conv_1d.html">layer_depthwise_conv_1d</a>()</code> <br><code><a href="layer_depthwise_conv_2d.html">layer_depthwise_conv_2d</a>()</code> <br><code><a href="layer_discretization.html">layer_discretization</a>()</code> <br><code><a href="layer_dot.html">layer_dot</a>()</code> <br><code><a href="layer_dropout.html">layer_dropout</a>()</code> <br><code><a href="layer_einsum_dense.html">layer_einsum_dense</a>()</code> <br><code><a href="layer_embedding.html">layer_embedding</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_feature_space.html">layer_feature_space</a>()</code> <br><code><a href="layer_flatten.html">layer_flatten</a>()</code> <br><code><a href="layer_flax_module_wrapper.html">layer_flax_module_wrapper</a>()</code> <br><code><a href="layer_gaussian_dropout.html">layer_gaussian_dropout</a>()</code> <br><code><a href="layer_gaussian_noise.html">layer_gaussian_noise</a>()</code> <br><code><a href="layer_global_average_pooling_1d.html">layer_global_average_pooling_1d</a>()</code> <br><code><a href="layer_global_average_pooling_2d.html">layer_global_average_pooling_2d</a>()</code> <br><code><a href="layer_global_average_pooling_3d.html">layer_global_average_pooling_3d</a>()</code> <br><code><a href="layer_global_max_pooling_1d.html">layer_global_max_pooling_1d</a>()</code> <br><code><a href="layer_global_max_pooling_2d.html">layer_global_max_pooling_2d</a>()</code> <br><code><a href="layer_global_max_pooling_3d.html">layer_global_max_pooling_3d</a>()</code> <br><code><a href="layer_group_normalization.html">layer_group_normalization</a>()</code> <br><code><a href="layer_group_query_attention.html">layer_group_query_attention</a>()</code> <br><code><a href="layer_gru.html">layer_gru</a>()</code> <br><code><a href="layer_hashed_crossing.html">layer_hashed_crossing</a>()</code> <br><code><a href="layer_hashing.html">layer_hashing</a>()</code> <br><code><a href="layer_identity.html">layer_identity</a>()</code> <br><code><a href="layer_integer_lookup.html">layer_integer_lookup</a>()</code> <br><code><a href="layer_jax_model_wrapper.html">layer_jax_model_wrapper</a>()</code> <br><code><a href="layer_lambda.html">layer_lambda</a>()</code> <br><code><a href="layer_layer_normalization.html">layer_layer_normalization</a>()</code> <br><code><a href="layer_lstm.html">layer_lstm</a>()</code> <br><code><a href="layer_masking.html">layer_masking</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_max_pooling_1d.html">layer_max_pooling_1d</a>()</code> <br><code><a href="layer_max_pooling_2d.html">layer_max_pooling_2d</a>()</code> <br><code><a href="layer_max_pooling_3d.html">layer_max_pooling_3d</a>()</code> <br><code><a href="layer_maximum.html">layer_maximum</a>()</code> <br><code><a href="layer_mel_spectrogram.html">layer_mel_spectrogram</a>()</code> <br><code><a href="layer_minimum.html">layer_minimum</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_multi_head_attention.html">layer_multi_head_attention</a>()</code> <br><code><a href="layer_multiply.html">layer_multiply</a>()</code> <br><code><a href="layer_normalization.html">layer_normalization</a>()</code> <br><code><a href="layer_permute.html">layer_permute</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_brightness.html">layer_random_brightness</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_contrast.html">layer_random_contrast</a>()</code> <br><code><a href="layer_random_crop.html">layer_random_crop</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_flip.html">layer_random_flip</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_rotation.html">layer_random_rotation</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_random_translation.html">layer_random_translation</a>()</code> <br><code><a href="layer_random_zoom.html">layer_random_zoom</a>()</code> <br><code><a href="layer_repeat_vector.html">layer_repeat_vector</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_reshape.html">layer_reshape</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_rms_normalization.html">layer_rms_normalization</a>()</code> <br><code><a href="layer_rnn.html">layer_rnn</a>()</code> <br><code><a href="layer_separable_conv_1d.html">layer_separable_conv_1d</a>()</code> <br><code><a href="layer_separable_conv_2d.html">layer_separable_conv_2d</a>()</code> <br><code><a href="layer_simple_rnn.html">layer_simple_rnn</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br><code><a href="layer_spatial_dropout_1d.html">layer_spatial_dropout_1d</a>()</code> <br><code><a href="layer_spatial_dropout_2d.html">layer_spatial_dropout_2d</a>()</code> <br><code><a href="layer_spatial_dropout_3d.html">layer_spatial_dropout_3d</a>()</code> <br><code><a href="layer_spectral_normalization.html">layer_spectral_normalization</a>()</code> <br><code><a href="layer_stft_spectrogram.html">layer_stft_spectrogram</a>()</code> <br><code><a href="layer_string_lookup.html">layer_string_lookup</a>()</code> <br><code><a href="layer_subtract.html">layer_subtract</a>()</code> <br><code><a href="layer_text_vectorization.html">layer_text_vectorization</a>()</code> <br><code><a href="layer_tfsm.html">layer_tfsm</a>()</code> <br><code><a href="layer_time_distributed.html">layer_time_distributed</a>()</code> <br><code><a href="layer_torch_module_wrapper.html">layer_torch_module_wrapper</a>()</code> <br><code><a href="layer_unit_normalization.html">layer_unit_normalization</a>()</code> <br><code><a href="layer_upsampling_1d.html">layer_upsampling_1d</a>()</code> <br><code><a href="layer_upsampling_2d.html">layer_upsampling_2d</a>()</code> <br><code><a href="layer_upsampling_3d.html">layer_upsampling_3d</a>()</code> <br><code><a href="layer_zero_padding_1d.html">layer_zero_padding_1d</a>()</code> <br><code><a href="layer_zero_padding_2d.html">layer_zero_padding_2d</a>()</code> <br><code><a href="layer_zero_padding_3d.html">layer_zero_padding_3d</a>()</code> <br><code><a href="rnn_cell_gru.html">rnn_cell_gru</a>()</code> <br><code><a href="rnn_cell_lstm.html">rnn_cell_lstm</a>()</code> <br><code><a href="rnn_cell_simple.html">rnn_cell_simple</a>()</code> <br><code><a href="rnn_cells_stack.html">rnn_cells_stack</a>()</code> <br></p></div>
+<p>Other layers: <br><code><a href="Layer.html">Layer</a>()</code> <br><code><a href="layer_activation.html">layer_activation</a>()</code> <br><code><a href="layer_activation_elu.html">layer_activation_elu</a>()</code> <br><code><a href="layer_activation_leaky_relu.html">layer_activation_leaky_relu</a>()</code> <br><code><a href="layer_activation_parametric_relu.html">layer_activation_parametric_relu</a>()</code> <br><code><a href="layer_activation_relu.html">layer_activation_relu</a>()</code> <br><code><a href="layer_activation_softmax.html">layer_activation_softmax</a>()</code> <br><code><a href="layer_activity_regularization.html">layer_activity_regularization</a>()</code> <br><code><a href="layer_add.html">layer_add</a>()</code> <br><code><a href="layer_additive_attention.html">layer_additive_attention</a>()</code> <br><code><a href="layer_alpha_dropout.html">layer_alpha_dropout</a>()</code> <br><code><a href="layer_attention.html">layer_attention</a>()</code> <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_average.html">layer_average</a>()</code> <br><code><a href="layer_average_pooling_1d.html">layer_average_pooling_1d</a>()</code> <br><code><a href="layer_average_pooling_2d.html">layer_average_pooling_2d</a>()</code> <br><code><a href="layer_average_pooling_3d.html">layer_average_pooling_3d</a>()</code> <br><code><a href="layer_batch_normalization.html">layer_batch_normalization</a>()</code> <br><code><a href="layer_bidirectional.html">layer_bidirectional</a>()</code> <br><code><a href="layer_category_encoding.html">layer_category_encoding</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_concatenate.html">layer_concatenate</a>()</code> <br><code><a href="layer_conv_1d.html">layer_conv_1d</a>()</code> <br><code><a href="layer_conv_1d_transpose.html">layer_conv_1d_transpose</a>()</code> <br><code><a href="layer_conv_2d.html">layer_conv_2d</a>()</code> <br><code><a href="layer_conv_2d_transpose.html">layer_conv_2d_transpose</a>()</code> <br><code><a href="layer_conv_3d.html">layer_conv_3d</a>()</code> <br><code><a href="layer_conv_3d_transpose.html">layer_conv_3d_transpose</a>()</code> <br><code><a href="layer_conv_lstm_1d.html">layer_conv_lstm_1d</a>()</code> <br><code><a href="layer_conv_lstm_2d.html">layer_conv_lstm_2d</a>()</code> <br><code><a href="layer_conv_lstm_3d.html">layer_conv_lstm_3d</a>()</code> <br><code><a href="layer_cropping_2d.html">layer_cropping_2d</a>()</code> <br><code><a href="layer_cropping_3d.html">layer_cropping_3d</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_dense.html">layer_dense</a>()</code> <br><code><a href="layer_depthwise_conv_1d.html">layer_depthwise_conv_1d</a>()</code> <br><code><a href="layer_depthwise_conv_2d.html">layer_depthwise_conv_2d</a>()</code> <br><code><a href="layer_discretization.html">layer_discretization</a>()</code> <br><code><a href="layer_dot.html">layer_dot</a>()</code> <br><code><a href="layer_dropout.html">layer_dropout</a>()</code> <br><code><a href="layer_einsum_dense.html">layer_einsum_dense</a>()</code> <br><code><a href="layer_embedding.html">layer_embedding</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_feature_space.html">layer_feature_space</a>()</code> <br><code><a href="layer_flatten.html">layer_flatten</a>()</code> <br><code><a href="layer_flax_module_wrapper.html">layer_flax_module_wrapper</a>()</code> <br><code><a href="layer_gaussian_dropout.html">layer_gaussian_dropout</a>()</code> <br><code><a href="layer_gaussian_noise.html">layer_gaussian_noise</a>()</code> <br><code><a href="layer_global_average_pooling_1d.html">layer_global_average_pooling_1d</a>()</code> <br><code><a href="layer_global_average_pooling_2d.html">layer_global_average_pooling_2d</a>()</code> <br><code><a href="layer_global_average_pooling_3d.html">layer_global_average_pooling_3d</a>()</code> <br><code><a href="layer_global_max_pooling_1d.html">layer_global_max_pooling_1d</a>()</code> <br><code><a href="layer_global_max_pooling_2d.html">layer_global_max_pooling_2d</a>()</code> <br><code><a href="layer_global_max_pooling_3d.html">layer_global_max_pooling_3d</a>()</code> <br><code><a href="layer_group_normalization.html">layer_group_normalization</a>()</code> <br><code><a href="layer_group_query_attention.html">layer_group_query_attention</a>()</code> <br><code><a href="layer_gru.html">layer_gru</a>()</code> <br><code><a href="layer_hashed_crossing.html">layer_hashed_crossing</a>()</code> <br><code><a href="layer_hashing.html">layer_hashing</a>()</code> <br><code><a href="layer_identity.html">layer_identity</a>()</code> <br><code><a href="layer_integer_lookup.html">layer_integer_lookup</a>()</code> <br><code><a href="layer_jax_model_wrapper.html">layer_jax_model_wrapper</a>()</code> <br><code><a href="layer_lambda.html">layer_lambda</a>()</code> <br><code><a href="layer_layer_normalization.html">layer_layer_normalization</a>()</code> <br><code><a href="layer_lstm.html">layer_lstm</a>()</code> <br><code><a href="layer_masking.html">layer_masking</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_max_pooling_1d.html">layer_max_pooling_1d</a>()</code> <br><code><a href="layer_max_pooling_2d.html">layer_max_pooling_2d</a>()</code> <br><code><a href="layer_max_pooling_3d.html">layer_max_pooling_3d</a>()</code> <br><code><a href="layer_maximum.html">layer_maximum</a>()</code> <br><code><a href="layer_mel_spectrogram.html">layer_mel_spectrogram</a>()</code> <br><code><a href="layer_minimum.html">layer_minimum</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_multi_head_attention.html">layer_multi_head_attention</a>()</code> <br><code><a href="layer_multiply.html">layer_multiply</a>()</code> <br><code><a href="layer_normalization.html">layer_normalization</a>()</code> <br><code><a href="layer_permute.html">layer_permute</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_brightness.html">layer_random_brightness</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_contrast.html">layer_random_contrast</a>()</code> <br><code><a href="layer_random_crop.html">layer_random_crop</a>()</code> <br><code><a href="layer_random_elastic_transform.html">layer_random_elastic_transform</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_flip.html">layer_random_flip</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_rotation.html">layer_random_rotation</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_random_translation.html">layer_random_translation</a>()</code> <br><code><a href="layer_random_zoom.html">layer_random_zoom</a>()</code> <br><code><a href="layer_repeat_vector.html">layer_repeat_vector</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_reshape.html">layer_reshape</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_rms_normalization.html">layer_rms_normalization</a>()</code> <br><code><a href="layer_rnn.html">layer_rnn</a>()</code> <br><code><a href="layer_separable_conv_1d.html">layer_separable_conv_1d</a>()</code> <br><code><a href="layer_separable_conv_2d.html">layer_separable_conv_2d</a>()</code> <br><code><a href="layer_simple_rnn.html">layer_simple_rnn</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br><code><a href="layer_spatial_dropout_1d.html">layer_spatial_dropout_1d</a>()</code> <br><code><a href="layer_spatial_dropout_2d.html">layer_spatial_dropout_2d</a>()</code> <br><code><a href="layer_spatial_dropout_3d.html">layer_spatial_dropout_3d</a>()</code> <br><code><a href="layer_spectral_normalization.html">layer_spectral_normalization</a>()</code> <br><code><a href="layer_stft_spectrogram.html">layer_stft_spectrogram</a>()</code> <br><code><a href="layer_string_lookup.html">layer_string_lookup</a>()</code> <br><code><a href="layer_subtract.html">layer_subtract</a>()</code> <br><code><a href="layer_text_vectorization.html">layer_text_vectorization</a>()</code> <br><code><a href="layer_tfsm.html">layer_tfsm</a>()</code> <br><code><a href="layer_time_distributed.html">layer_time_distributed</a>()</code> <br><code><a href="layer_torch_module_wrapper.html">layer_torch_module_wrapper</a>()</code> <br><code><a href="layer_unit_normalization.html">layer_unit_normalization</a>()</code> <br><code><a href="layer_upsampling_1d.html">layer_upsampling_1d</a>()</code> <br><code><a href="layer_upsampling_2d.html">layer_upsampling_2d</a>()</code> <br><code><a href="layer_upsampling_3d.html">layer_upsampling_3d</a>()</code> <br><code><a href="layer_zero_padding_1d.html">layer_zero_padding_1d</a>()</code> <br><code><a href="layer_zero_padding_2d.html">layer_zero_padding_2d</a>()</code> <br><code><a href="layer_zero_padding_3d.html">layer_zero_padding_3d</a>()</code> <br><code><a href="rnn_cell_gru.html">rnn_cell_gru</a>()</code> <br><code><a href="rnn_cell_lstm.html">rnn_cell_lstm</a>()</code> <br><code><a href="rnn_cell_simple.html">rnn_cell_simple</a>()</code> <br><code><a href="rnn_cells_stack.html">rnn_cells_stack</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -139,7 +139,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/layer_cropping_2d.html b/docs/dev/reference/layer_cropping_2d.html
index 93b37cd19..b9d87b6f8 100644
--- a/docs/dev/reference/layer_cropping_2d.html
+++ b/docs/dev/reference/layer_cropping_2d.html
@@ -8,7 +8,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -139,7 +139,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
     <div class="dont-index">
 <ul><li><p><a href="https://keras.io/api/layers/reshaping_layers/cropping2d#cropping2d-class" class="external-link">https://keras.io/api/layers/reshaping_layers/cropping2d#cropping2d-class</a></p></li>
 </ul><p>Other reshaping layers: <br><code><a href="layer_cropping_1d.html">layer_cropping_1d</a>()</code> <br><code><a href="layer_cropping_3d.html">layer_cropping_3d</a>()</code> <br><code><a href="layer_flatten.html">layer_flatten</a>()</code> <br><code><a href="layer_permute.html">layer_permute</a>()</code> <br><code><a href="layer_repeat_vector.html">layer_repeat_vector</a>()</code> <br><code><a href="layer_reshape.html">layer_reshape</a>()</code> <br><code><a href="layer_upsampling_1d.html">layer_upsampling_1d</a>()</code> <br><code><a href="layer_upsampling_2d.html">layer_upsampling_2d</a>()</code> <br><code><a href="layer_upsampling_3d.html">layer_upsampling_3d</a>()</code> <br><code><a href="layer_zero_padding_1d.html">layer_zero_padding_1d</a>()</code> <br><code><a href="layer_zero_padding_2d.html">layer_zero_padding_2d</a>()</code> <br><code><a href="layer_zero_padding_3d.html">layer_zero_padding_3d</a>()</code> <br></p>
-<p>Other layers: <br><code><a href="Layer.html">Layer</a>()</code> <br><code><a href="layer_activation.html">layer_activation</a>()</code> <br><code><a href="layer_activation_elu.html">layer_activation_elu</a>()</code> <br><code><a href="layer_activation_leaky_relu.html">layer_activation_leaky_relu</a>()</code> <br><code><a href="layer_activation_parametric_relu.html">layer_activation_parametric_relu</a>()</code> <br><code><a href="layer_activation_relu.html">layer_activation_relu</a>()</code> <br><code><a href="layer_activation_softmax.html">layer_activation_softmax</a>()</code> <br><code><a href="layer_activity_regularization.html">layer_activity_regularization</a>()</code> <br><code><a href="layer_add.html">layer_add</a>()</code> <br><code><a href="layer_additive_attention.html">layer_additive_attention</a>()</code> <br><code><a href="layer_alpha_dropout.html">layer_alpha_dropout</a>()</code> <br><code><a href="layer_attention.html">layer_attention</a>()</code> <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_average.html">layer_average</a>()</code> <br><code><a href="layer_average_pooling_1d.html">layer_average_pooling_1d</a>()</code> <br><code><a href="layer_average_pooling_2d.html">layer_average_pooling_2d</a>()</code> <br><code><a href="layer_average_pooling_3d.html">layer_average_pooling_3d</a>()</code> <br><code><a href="layer_batch_normalization.html">layer_batch_normalization</a>()</code> <br><code><a href="layer_bidirectional.html">layer_bidirectional</a>()</code> <br><code><a href="layer_category_encoding.html">layer_category_encoding</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_concatenate.html">layer_concatenate</a>()</code> <br><code><a href="layer_conv_1d.html">layer_conv_1d</a>()</code> <br><code><a href="layer_conv_1d_transpose.html">layer_conv_1d_transpose</a>()</code> <br><code><a href="layer_conv_2d.html">layer_conv_2d</a>()</code> <br><code><a href="layer_conv_2d_transpose.html">layer_conv_2d_transpose</a>()</code> <br><code><a href="layer_conv_3d.html">layer_conv_3d</a>()</code> <br><code><a href="layer_conv_3d_transpose.html">layer_conv_3d_transpose</a>()</code> <br><code><a href="layer_conv_lstm_1d.html">layer_conv_lstm_1d</a>()</code> <br><code><a href="layer_conv_lstm_2d.html">layer_conv_lstm_2d</a>()</code> <br><code><a href="layer_conv_lstm_3d.html">layer_conv_lstm_3d</a>()</code> <br><code><a href="layer_cropping_1d.html">layer_cropping_1d</a>()</code> <br><code><a href="layer_cropping_3d.html">layer_cropping_3d</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_dense.html">layer_dense</a>()</code> <br><code><a href="layer_depthwise_conv_1d.html">layer_depthwise_conv_1d</a>()</code> <br><code><a href="layer_depthwise_conv_2d.html">layer_depthwise_conv_2d</a>()</code> <br><code><a href="layer_discretization.html">layer_discretization</a>()</code> <br><code><a href="layer_dot.html">layer_dot</a>()</code> <br><code><a href="layer_dropout.html">layer_dropout</a>()</code> <br><code><a href="layer_einsum_dense.html">layer_einsum_dense</a>()</code> <br><code><a href="layer_embedding.html">layer_embedding</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_feature_space.html">layer_feature_space</a>()</code> <br><code><a href="layer_flatten.html">layer_flatten</a>()</code> <br><code><a href="layer_flax_module_wrapper.html">layer_flax_module_wrapper</a>()</code> <br><code><a href="layer_gaussian_dropout.html">layer_gaussian_dropout</a>()</code> <br><code><a href="layer_gaussian_noise.html">layer_gaussian_noise</a>()</code> <br><code><a href="layer_global_average_pooling_1d.html">layer_global_average_pooling_1d</a>()</code> <br><code><a href="layer_global_average_pooling_2d.html">layer_global_average_pooling_2d</a>()</code> <br><code><a href="layer_global_average_pooling_3d.html">layer_global_average_pooling_3d</a>()</code> <br><code><a href="layer_global_max_pooling_1d.html">layer_global_max_pooling_1d</a>()</code> <br><code><a href="layer_global_max_pooling_2d.html">layer_global_max_pooling_2d</a>()</code> <br><code><a href="layer_global_max_pooling_3d.html">layer_global_max_pooling_3d</a>()</code> <br><code><a href="layer_group_normalization.html">layer_group_normalization</a>()</code> <br><code><a href="layer_group_query_attention.html">layer_group_query_attention</a>()</code> <br><code><a href="layer_gru.html">layer_gru</a>()</code> <br><code><a href="layer_hashed_crossing.html">layer_hashed_crossing</a>()</code> <br><code><a href="layer_hashing.html">layer_hashing</a>()</code> <br><code><a href="layer_identity.html">layer_identity</a>()</code> <br><code><a href="layer_integer_lookup.html">layer_integer_lookup</a>()</code> <br><code><a href="layer_jax_model_wrapper.html">layer_jax_model_wrapper</a>()</code> <br><code><a href="layer_lambda.html">layer_lambda</a>()</code> <br><code><a href="layer_layer_normalization.html">layer_layer_normalization</a>()</code> <br><code><a href="layer_lstm.html">layer_lstm</a>()</code> <br><code><a href="layer_masking.html">layer_masking</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_max_pooling_1d.html">layer_max_pooling_1d</a>()</code> <br><code><a href="layer_max_pooling_2d.html">layer_max_pooling_2d</a>()</code> <br><code><a href="layer_max_pooling_3d.html">layer_max_pooling_3d</a>()</code> <br><code><a href="layer_maximum.html">layer_maximum</a>()</code> <br><code><a href="layer_mel_spectrogram.html">layer_mel_spectrogram</a>()</code> <br><code><a href="layer_minimum.html">layer_minimum</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_multi_head_attention.html">layer_multi_head_attention</a>()</code> <br><code><a href="layer_multiply.html">layer_multiply</a>()</code> <br><code><a href="layer_normalization.html">layer_normalization</a>()</code> <br><code><a href="layer_permute.html">layer_permute</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_brightness.html">layer_random_brightness</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_contrast.html">layer_random_contrast</a>()</code> <br><code><a href="layer_random_crop.html">layer_random_crop</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_flip.html">layer_random_flip</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_rotation.html">layer_random_rotation</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_random_translation.html">layer_random_translation</a>()</code> <br><code><a href="layer_random_zoom.html">layer_random_zoom</a>()</code> <br><code><a href="layer_repeat_vector.html">layer_repeat_vector</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_reshape.html">layer_reshape</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_rms_normalization.html">layer_rms_normalization</a>()</code> <br><code><a href="layer_rnn.html">layer_rnn</a>()</code> <br><code><a href="layer_separable_conv_1d.html">layer_separable_conv_1d</a>()</code> <br><code><a href="layer_separable_conv_2d.html">layer_separable_conv_2d</a>()</code> <br><code><a href="layer_simple_rnn.html">layer_simple_rnn</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br><code><a href="layer_spatial_dropout_1d.html">layer_spatial_dropout_1d</a>()</code> <br><code><a href="layer_spatial_dropout_2d.html">layer_spatial_dropout_2d</a>()</code> <br><code><a href="layer_spatial_dropout_3d.html">layer_spatial_dropout_3d</a>()</code> <br><code><a href="layer_spectral_normalization.html">layer_spectral_normalization</a>()</code> <br><code><a href="layer_stft_spectrogram.html">layer_stft_spectrogram</a>()</code> <br><code><a href="layer_string_lookup.html">layer_string_lookup</a>()</code> <br><code><a href="layer_subtract.html">layer_subtract</a>()</code> <br><code><a href="layer_text_vectorization.html">layer_text_vectorization</a>()</code> <br><code><a href="layer_tfsm.html">layer_tfsm</a>()</code> <br><code><a href="layer_time_distributed.html">layer_time_distributed</a>()</code> <br><code><a href="layer_torch_module_wrapper.html">layer_torch_module_wrapper</a>()</code> <br><code><a href="layer_unit_normalization.html">layer_unit_normalization</a>()</code> <br><code><a href="layer_upsampling_1d.html">layer_upsampling_1d</a>()</code> <br><code><a href="layer_upsampling_2d.html">layer_upsampling_2d</a>()</code> <br><code><a href="layer_upsampling_3d.html">layer_upsampling_3d</a>()</code> <br><code><a href="layer_zero_padding_1d.html">layer_zero_padding_1d</a>()</code> <br><code><a href="layer_zero_padding_2d.html">layer_zero_padding_2d</a>()</code> <br><code><a href="layer_zero_padding_3d.html">layer_zero_padding_3d</a>()</code> <br><code><a href="rnn_cell_gru.html">rnn_cell_gru</a>()</code> <br><code><a href="rnn_cell_lstm.html">rnn_cell_lstm</a>()</code> <br><code><a href="rnn_cell_simple.html">rnn_cell_simple</a>()</code> <br><code><a href="rnn_cells_stack.html">rnn_cells_stack</a>()</code> <br></p></div>
+<p>Other layers: <br><code><a href="Layer.html">Layer</a>()</code> <br><code><a href="layer_activation.html">layer_activation</a>()</code> <br><code><a href="layer_activation_elu.html">layer_activation_elu</a>()</code> <br><code><a href="layer_activation_leaky_relu.html">layer_activation_leaky_relu</a>()</code> <br><code><a href="layer_activation_parametric_relu.html">layer_activation_parametric_relu</a>()</code> <br><code><a href="layer_activation_relu.html">layer_activation_relu</a>()</code> <br><code><a href="layer_activation_softmax.html">layer_activation_softmax</a>()</code> <br><code><a href="layer_activity_regularization.html">layer_activity_regularization</a>()</code> <br><code><a href="layer_add.html">layer_add</a>()</code> <br><code><a href="layer_additive_attention.html">layer_additive_attention</a>()</code> <br><code><a href="layer_alpha_dropout.html">layer_alpha_dropout</a>()</code> <br><code><a href="layer_attention.html">layer_attention</a>()</code> <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_average.html">layer_average</a>()</code> <br><code><a href="layer_average_pooling_1d.html">layer_average_pooling_1d</a>()</code> <br><code><a href="layer_average_pooling_2d.html">layer_average_pooling_2d</a>()</code> <br><code><a href="layer_average_pooling_3d.html">layer_average_pooling_3d</a>()</code> <br><code><a href="layer_batch_normalization.html">layer_batch_normalization</a>()</code> <br><code><a href="layer_bidirectional.html">layer_bidirectional</a>()</code> <br><code><a href="layer_category_encoding.html">layer_category_encoding</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_concatenate.html">layer_concatenate</a>()</code> <br><code><a href="layer_conv_1d.html">layer_conv_1d</a>()</code> <br><code><a href="layer_conv_1d_transpose.html">layer_conv_1d_transpose</a>()</code> <br><code><a href="layer_conv_2d.html">layer_conv_2d</a>()</code> <br><code><a href="layer_conv_2d_transpose.html">layer_conv_2d_transpose</a>()</code> <br><code><a href="layer_conv_3d.html">layer_conv_3d</a>()</code> <br><code><a href="layer_conv_3d_transpose.html">layer_conv_3d_transpose</a>()</code> <br><code><a href="layer_conv_lstm_1d.html">layer_conv_lstm_1d</a>()</code> <br><code><a href="layer_conv_lstm_2d.html">layer_conv_lstm_2d</a>()</code> <br><code><a href="layer_conv_lstm_3d.html">layer_conv_lstm_3d</a>()</code> <br><code><a href="layer_cropping_1d.html">layer_cropping_1d</a>()</code> <br><code><a href="layer_cropping_3d.html">layer_cropping_3d</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_dense.html">layer_dense</a>()</code> <br><code><a href="layer_depthwise_conv_1d.html">layer_depthwise_conv_1d</a>()</code> <br><code><a href="layer_depthwise_conv_2d.html">layer_depthwise_conv_2d</a>()</code> <br><code><a href="layer_discretization.html">layer_discretization</a>()</code> <br><code><a href="layer_dot.html">layer_dot</a>()</code> <br><code><a href="layer_dropout.html">layer_dropout</a>()</code> <br><code><a href="layer_einsum_dense.html">layer_einsum_dense</a>()</code> <br><code><a href="layer_embedding.html">layer_embedding</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_feature_space.html">layer_feature_space</a>()</code> <br><code><a href="layer_flatten.html">layer_flatten</a>()</code> <br><code><a href="layer_flax_module_wrapper.html">layer_flax_module_wrapper</a>()</code> <br><code><a href="layer_gaussian_dropout.html">layer_gaussian_dropout</a>()</code> <br><code><a href="layer_gaussian_noise.html">layer_gaussian_noise</a>()</code> <br><code><a href="layer_global_average_pooling_1d.html">layer_global_average_pooling_1d</a>()</code> <br><code><a href="layer_global_average_pooling_2d.html">layer_global_average_pooling_2d</a>()</code> <br><code><a href="layer_global_average_pooling_3d.html">layer_global_average_pooling_3d</a>()</code> <br><code><a href="layer_global_max_pooling_1d.html">layer_global_max_pooling_1d</a>()</code> <br><code><a href="layer_global_max_pooling_2d.html">layer_global_max_pooling_2d</a>()</code> <br><code><a href="layer_global_max_pooling_3d.html">layer_global_max_pooling_3d</a>()</code> <br><code><a href="layer_group_normalization.html">layer_group_normalization</a>()</code> <br><code><a href="layer_group_query_attention.html">layer_group_query_attention</a>()</code> <br><code><a href="layer_gru.html">layer_gru</a>()</code> <br><code><a href="layer_hashed_crossing.html">layer_hashed_crossing</a>()</code> <br><code><a href="layer_hashing.html">layer_hashing</a>()</code> <br><code><a href="layer_identity.html">layer_identity</a>()</code> <br><code><a href="layer_integer_lookup.html">layer_integer_lookup</a>()</code> <br><code><a href="layer_jax_model_wrapper.html">layer_jax_model_wrapper</a>()</code> <br><code><a href="layer_lambda.html">layer_lambda</a>()</code> <br><code><a href="layer_layer_normalization.html">layer_layer_normalization</a>()</code> <br><code><a href="layer_lstm.html">layer_lstm</a>()</code> <br><code><a href="layer_masking.html">layer_masking</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_max_pooling_1d.html">layer_max_pooling_1d</a>()</code> <br><code><a href="layer_max_pooling_2d.html">layer_max_pooling_2d</a>()</code> <br><code><a href="layer_max_pooling_3d.html">layer_max_pooling_3d</a>()</code> <br><code><a href="layer_maximum.html">layer_maximum</a>()</code> <br><code><a href="layer_mel_spectrogram.html">layer_mel_spectrogram</a>()</code> <br><code><a href="layer_minimum.html">layer_minimum</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_multi_head_attention.html">layer_multi_head_attention</a>()</code> <br><code><a href="layer_multiply.html">layer_multiply</a>()</code> <br><code><a href="layer_normalization.html">layer_normalization</a>()</code> <br><code><a href="layer_permute.html">layer_permute</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_brightness.html">layer_random_brightness</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_contrast.html">layer_random_contrast</a>()</code> <br><code><a href="layer_random_crop.html">layer_random_crop</a>()</code> <br><code><a href="layer_random_elastic_transform.html">layer_random_elastic_transform</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_flip.html">layer_random_flip</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_rotation.html">layer_random_rotation</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_random_translation.html">layer_random_translation</a>()</code> <br><code><a href="layer_random_zoom.html">layer_random_zoom</a>()</code> <br><code><a href="layer_repeat_vector.html">layer_repeat_vector</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_reshape.html">layer_reshape</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_rms_normalization.html">layer_rms_normalization</a>()</code> <br><code><a href="layer_rnn.html">layer_rnn</a>()</code> <br><code><a href="layer_separable_conv_1d.html">layer_separable_conv_1d</a>()</code> <br><code><a href="layer_separable_conv_2d.html">layer_separable_conv_2d</a>()</code> <br><code><a href="layer_simple_rnn.html">layer_simple_rnn</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br><code><a href="layer_spatial_dropout_1d.html">layer_spatial_dropout_1d</a>()</code> <br><code><a href="layer_spatial_dropout_2d.html">layer_spatial_dropout_2d</a>()</code> <br><code><a href="layer_spatial_dropout_3d.html">layer_spatial_dropout_3d</a>()</code> <br><code><a href="layer_spectral_normalization.html">layer_spectral_normalization</a>()</code> <br><code><a href="layer_stft_spectrogram.html">layer_stft_spectrogram</a>()</code> <br><code><a href="layer_string_lookup.html">layer_string_lookup</a>()</code> <br><code><a href="layer_subtract.html">layer_subtract</a>()</code> <br><code><a href="layer_text_vectorization.html">layer_text_vectorization</a>()</code> <br><code><a href="layer_tfsm.html">layer_tfsm</a>()</code> <br><code><a href="layer_time_distributed.html">layer_time_distributed</a>()</code> <br><code><a href="layer_torch_module_wrapper.html">layer_torch_module_wrapper</a>()</code> <br><code><a href="layer_unit_normalization.html">layer_unit_normalization</a>()</code> <br><code><a href="layer_upsampling_1d.html">layer_upsampling_1d</a>()</code> <br><code><a href="layer_upsampling_2d.html">layer_upsampling_2d</a>()</code> <br><code><a href="layer_upsampling_3d.html">layer_upsampling_3d</a>()</code> <br><code><a href="layer_zero_padding_1d.html">layer_zero_padding_1d</a>()</code> <br><code><a href="layer_zero_padding_2d.html">layer_zero_padding_2d</a>()</code> <br><code><a href="layer_zero_padding_3d.html">layer_zero_padding_3d</a>()</code> <br><code><a href="rnn_cell_gru.html">rnn_cell_gru</a>()</code> <br><code><a href="rnn_cell_lstm.html">rnn_cell_lstm</a>()</code> <br><code><a href="rnn_cell_simple.html">rnn_cell_simple</a>()</code> <br><code><a href="rnn_cells_stack.html">rnn_cells_stack</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -151,7 +151,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/layer_cropping_3d.html b/docs/dev/reference/layer_cropping_3d.html
index 53854b3bd..81050504d 100644
--- a/docs/dev/reference/layer_cropping_3d.html
+++ b/docs/dev/reference/layer_cropping_3d.html
@@ -8,7 +8,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -139,7 +139,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
     <div class="dont-index">
 <ul><li><p><a href="https://keras.io/api/layers/reshaping_layers/cropping3d#cropping3d-class" class="external-link">https://keras.io/api/layers/reshaping_layers/cropping3d#cropping3d-class</a></p></li>
 </ul><p>Other reshaping layers: <br><code><a href="layer_cropping_1d.html">layer_cropping_1d</a>()</code> <br><code><a href="layer_cropping_2d.html">layer_cropping_2d</a>()</code> <br><code><a href="layer_flatten.html">layer_flatten</a>()</code> <br><code><a href="layer_permute.html">layer_permute</a>()</code> <br><code><a href="layer_repeat_vector.html">layer_repeat_vector</a>()</code> <br><code><a href="layer_reshape.html">layer_reshape</a>()</code> <br><code><a href="layer_upsampling_1d.html">layer_upsampling_1d</a>()</code> <br><code><a href="layer_upsampling_2d.html">layer_upsampling_2d</a>()</code> <br><code><a href="layer_upsampling_3d.html">layer_upsampling_3d</a>()</code> <br><code><a href="layer_zero_padding_1d.html">layer_zero_padding_1d</a>()</code> <br><code><a href="layer_zero_padding_2d.html">layer_zero_padding_2d</a>()</code> <br><code><a href="layer_zero_padding_3d.html">layer_zero_padding_3d</a>()</code> <br></p>
-<p>Other layers: <br><code><a href="Layer.html">Layer</a>()</code> <br><code><a href="layer_activation.html">layer_activation</a>()</code> <br><code><a href="layer_activation_elu.html">layer_activation_elu</a>()</code> <br><code><a href="layer_activation_leaky_relu.html">layer_activation_leaky_relu</a>()</code> <br><code><a href="layer_activation_parametric_relu.html">layer_activation_parametric_relu</a>()</code> <br><code><a href="layer_activation_relu.html">layer_activation_relu</a>()</code> <br><code><a href="layer_activation_softmax.html">layer_activation_softmax</a>()</code> <br><code><a href="layer_activity_regularization.html">layer_activity_regularization</a>()</code> <br><code><a href="layer_add.html">layer_add</a>()</code> <br><code><a href="layer_additive_attention.html">layer_additive_attention</a>()</code> <br><code><a href="layer_alpha_dropout.html">layer_alpha_dropout</a>()</code> <br><code><a href="layer_attention.html">layer_attention</a>()</code> <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_average.html">layer_average</a>()</code> <br><code><a href="layer_average_pooling_1d.html">layer_average_pooling_1d</a>()</code> <br><code><a href="layer_average_pooling_2d.html">layer_average_pooling_2d</a>()</code> <br><code><a href="layer_average_pooling_3d.html">layer_average_pooling_3d</a>()</code> <br><code><a href="layer_batch_normalization.html">layer_batch_normalization</a>()</code> <br><code><a href="layer_bidirectional.html">layer_bidirectional</a>()</code> <br><code><a href="layer_category_encoding.html">layer_category_encoding</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_concatenate.html">layer_concatenate</a>()</code> <br><code><a href="layer_conv_1d.html">layer_conv_1d</a>()</code> <br><code><a href="layer_conv_1d_transpose.html">layer_conv_1d_transpose</a>()</code> <br><code><a href="layer_conv_2d.html">layer_conv_2d</a>()</code> <br><code><a href="layer_conv_2d_transpose.html">layer_conv_2d_transpose</a>()</code> <br><code><a href="layer_conv_3d.html">layer_conv_3d</a>()</code> <br><code><a href="layer_conv_3d_transpose.html">layer_conv_3d_transpose</a>()</code> <br><code><a href="layer_conv_lstm_1d.html">layer_conv_lstm_1d</a>()</code> <br><code><a href="layer_conv_lstm_2d.html">layer_conv_lstm_2d</a>()</code> <br><code><a href="layer_conv_lstm_3d.html">layer_conv_lstm_3d</a>()</code> <br><code><a href="layer_cropping_1d.html">layer_cropping_1d</a>()</code> <br><code><a href="layer_cropping_2d.html">layer_cropping_2d</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_dense.html">layer_dense</a>()</code> <br><code><a href="layer_depthwise_conv_1d.html">layer_depthwise_conv_1d</a>()</code> <br><code><a href="layer_depthwise_conv_2d.html">layer_depthwise_conv_2d</a>()</code> <br><code><a href="layer_discretization.html">layer_discretization</a>()</code> <br><code><a href="layer_dot.html">layer_dot</a>()</code> <br><code><a href="layer_dropout.html">layer_dropout</a>()</code> <br><code><a href="layer_einsum_dense.html">layer_einsum_dense</a>()</code> <br><code><a href="layer_embedding.html">layer_embedding</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_feature_space.html">layer_feature_space</a>()</code> <br><code><a href="layer_flatten.html">layer_flatten</a>()</code> <br><code><a href="layer_flax_module_wrapper.html">layer_flax_module_wrapper</a>()</code> <br><code><a href="layer_gaussian_dropout.html">layer_gaussian_dropout</a>()</code> <br><code><a href="layer_gaussian_noise.html">layer_gaussian_noise</a>()</code> <br><code><a href="layer_global_average_pooling_1d.html">layer_global_average_pooling_1d</a>()</code> <br><code><a href="layer_global_average_pooling_2d.html">layer_global_average_pooling_2d</a>()</code> <br><code><a href="layer_global_average_pooling_3d.html">layer_global_average_pooling_3d</a>()</code> <br><code><a href="layer_global_max_pooling_1d.html">layer_global_max_pooling_1d</a>()</code> <br><code><a href="layer_global_max_pooling_2d.html">layer_global_max_pooling_2d</a>()</code> <br><code><a href="layer_global_max_pooling_3d.html">layer_global_max_pooling_3d</a>()</code> <br><code><a href="layer_group_normalization.html">layer_group_normalization</a>()</code> <br><code><a href="layer_group_query_attention.html">layer_group_query_attention</a>()</code> <br><code><a href="layer_gru.html">layer_gru</a>()</code> <br><code><a href="layer_hashed_crossing.html">layer_hashed_crossing</a>()</code> <br><code><a href="layer_hashing.html">layer_hashing</a>()</code> <br><code><a href="layer_identity.html">layer_identity</a>()</code> <br><code><a href="layer_integer_lookup.html">layer_integer_lookup</a>()</code> <br><code><a href="layer_jax_model_wrapper.html">layer_jax_model_wrapper</a>()</code> <br><code><a href="layer_lambda.html">layer_lambda</a>()</code> <br><code><a href="layer_layer_normalization.html">layer_layer_normalization</a>()</code> <br><code><a href="layer_lstm.html">layer_lstm</a>()</code> <br><code><a href="layer_masking.html">layer_masking</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_max_pooling_1d.html">layer_max_pooling_1d</a>()</code> <br><code><a href="layer_max_pooling_2d.html">layer_max_pooling_2d</a>()</code> <br><code><a href="layer_max_pooling_3d.html">layer_max_pooling_3d</a>()</code> <br><code><a href="layer_maximum.html">layer_maximum</a>()</code> <br><code><a href="layer_mel_spectrogram.html">layer_mel_spectrogram</a>()</code> <br><code><a href="layer_minimum.html">layer_minimum</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_multi_head_attention.html">layer_multi_head_attention</a>()</code> <br><code><a href="layer_multiply.html">layer_multiply</a>()</code> <br><code><a href="layer_normalization.html">layer_normalization</a>()</code> <br><code><a href="layer_permute.html">layer_permute</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_brightness.html">layer_random_brightness</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_contrast.html">layer_random_contrast</a>()</code> <br><code><a href="layer_random_crop.html">layer_random_crop</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_flip.html">layer_random_flip</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_rotation.html">layer_random_rotation</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_random_translation.html">layer_random_translation</a>()</code> <br><code><a href="layer_random_zoom.html">layer_random_zoom</a>()</code> <br><code><a href="layer_repeat_vector.html">layer_repeat_vector</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_reshape.html">layer_reshape</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_rms_normalization.html">layer_rms_normalization</a>()</code> <br><code><a href="layer_rnn.html">layer_rnn</a>()</code> <br><code><a href="layer_separable_conv_1d.html">layer_separable_conv_1d</a>()</code> <br><code><a href="layer_separable_conv_2d.html">layer_separable_conv_2d</a>()</code> <br><code><a href="layer_simple_rnn.html">layer_simple_rnn</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br><code><a href="layer_spatial_dropout_1d.html">layer_spatial_dropout_1d</a>()</code> <br><code><a href="layer_spatial_dropout_2d.html">layer_spatial_dropout_2d</a>()</code> <br><code><a href="layer_spatial_dropout_3d.html">layer_spatial_dropout_3d</a>()</code> <br><code><a href="layer_spectral_normalization.html">layer_spectral_normalization</a>()</code> <br><code><a href="layer_stft_spectrogram.html">layer_stft_spectrogram</a>()</code> <br><code><a href="layer_string_lookup.html">layer_string_lookup</a>()</code> <br><code><a href="layer_subtract.html">layer_subtract</a>()</code> <br><code><a href="layer_text_vectorization.html">layer_text_vectorization</a>()</code> <br><code><a href="layer_tfsm.html">layer_tfsm</a>()</code> <br><code><a href="layer_time_distributed.html">layer_time_distributed</a>()</code> <br><code><a href="layer_torch_module_wrapper.html">layer_torch_module_wrapper</a>()</code> <br><code><a href="layer_unit_normalization.html">layer_unit_normalization</a>()</code> <br><code><a href="layer_upsampling_1d.html">layer_upsampling_1d</a>()</code> <br><code><a href="layer_upsampling_2d.html">layer_upsampling_2d</a>()</code> <br><code><a href="layer_upsampling_3d.html">layer_upsampling_3d</a>()</code> <br><code><a href="layer_zero_padding_1d.html">layer_zero_padding_1d</a>()</code> <br><code><a href="layer_zero_padding_2d.html">layer_zero_padding_2d</a>()</code> <br><code><a href="layer_zero_padding_3d.html">layer_zero_padding_3d</a>()</code> <br><code><a href="rnn_cell_gru.html">rnn_cell_gru</a>()</code> <br><code><a href="rnn_cell_lstm.html">rnn_cell_lstm</a>()</code> <br><code><a href="rnn_cell_simple.html">rnn_cell_simple</a>()</code> <br><code><a href="rnn_cells_stack.html">rnn_cells_stack</a>()</code> <br></p></div>
+<p>Other layers: <br><code><a href="Layer.html">Layer</a>()</code> <br><code><a href="layer_activation.html">layer_activation</a>()</code> <br><code><a href="layer_activation_elu.html">layer_activation_elu</a>()</code> <br><code><a href="layer_activation_leaky_relu.html">layer_activation_leaky_relu</a>()</code> <br><code><a href="layer_activation_parametric_relu.html">layer_activation_parametric_relu</a>()</code> <br><code><a href="layer_activation_relu.html">layer_activation_relu</a>()</code> <br><code><a href="layer_activation_softmax.html">layer_activation_softmax</a>()</code> <br><code><a href="layer_activity_regularization.html">layer_activity_regularization</a>()</code> <br><code><a href="layer_add.html">layer_add</a>()</code> <br><code><a href="layer_additive_attention.html">layer_additive_attention</a>()</code> <br><code><a href="layer_alpha_dropout.html">layer_alpha_dropout</a>()</code> <br><code><a href="layer_attention.html">layer_attention</a>()</code> <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_average.html">layer_average</a>()</code> <br><code><a href="layer_average_pooling_1d.html">layer_average_pooling_1d</a>()</code> <br><code><a href="layer_average_pooling_2d.html">layer_average_pooling_2d</a>()</code> <br><code><a href="layer_average_pooling_3d.html">layer_average_pooling_3d</a>()</code> <br><code><a href="layer_batch_normalization.html">layer_batch_normalization</a>()</code> <br><code><a href="layer_bidirectional.html">layer_bidirectional</a>()</code> <br><code><a href="layer_category_encoding.html">layer_category_encoding</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_concatenate.html">layer_concatenate</a>()</code> <br><code><a href="layer_conv_1d.html">layer_conv_1d</a>()</code> <br><code><a href="layer_conv_1d_transpose.html">layer_conv_1d_transpose</a>()</code> <br><code><a href="layer_conv_2d.html">layer_conv_2d</a>()</code> <br><code><a href="layer_conv_2d_transpose.html">layer_conv_2d_transpose</a>()</code> <br><code><a href="layer_conv_3d.html">layer_conv_3d</a>()</code> <br><code><a href="layer_conv_3d_transpose.html">layer_conv_3d_transpose</a>()</code> <br><code><a href="layer_conv_lstm_1d.html">layer_conv_lstm_1d</a>()</code> <br><code><a href="layer_conv_lstm_2d.html">layer_conv_lstm_2d</a>()</code> <br><code><a href="layer_conv_lstm_3d.html">layer_conv_lstm_3d</a>()</code> <br><code><a href="layer_cropping_1d.html">layer_cropping_1d</a>()</code> <br><code><a href="layer_cropping_2d.html">layer_cropping_2d</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_dense.html">layer_dense</a>()</code> <br><code><a href="layer_depthwise_conv_1d.html">layer_depthwise_conv_1d</a>()</code> <br><code><a href="layer_depthwise_conv_2d.html">layer_depthwise_conv_2d</a>()</code> <br><code><a href="layer_discretization.html">layer_discretization</a>()</code> <br><code><a href="layer_dot.html">layer_dot</a>()</code> <br><code><a href="layer_dropout.html">layer_dropout</a>()</code> <br><code><a href="layer_einsum_dense.html">layer_einsum_dense</a>()</code> <br><code><a href="layer_embedding.html">layer_embedding</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_feature_space.html">layer_feature_space</a>()</code> <br><code><a href="layer_flatten.html">layer_flatten</a>()</code> <br><code><a href="layer_flax_module_wrapper.html">layer_flax_module_wrapper</a>()</code> <br><code><a href="layer_gaussian_dropout.html">layer_gaussian_dropout</a>()</code> <br><code><a href="layer_gaussian_noise.html">layer_gaussian_noise</a>()</code> <br><code><a href="layer_global_average_pooling_1d.html">layer_global_average_pooling_1d</a>()</code> <br><code><a href="layer_global_average_pooling_2d.html">layer_global_average_pooling_2d</a>()</code> <br><code><a href="layer_global_average_pooling_3d.html">layer_global_average_pooling_3d</a>()</code> <br><code><a href="layer_global_max_pooling_1d.html">layer_global_max_pooling_1d</a>()</code> <br><code><a href="layer_global_max_pooling_2d.html">layer_global_max_pooling_2d</a>()</code> <br><code><a href="layer_global_max_pooling_3d.html">layer_global_max_pooling_3d</a>()</code> <br><code><a href="layer_group_normalization.html">layer_group_normalization</a>()</code> <br><code><a href="layer_group_query_attention.html">layer_group_query_attention</a>()</code> <br><code><a href="layer_gru.html">layer_gru</a>()</code> <br><code><a href="layer_hashed_crossing.html">layer_hashed_crossing</a>()</code> <br><code><a href="layer_hashing.html">layer_hashing</a>()</code> <br><code><a href="layer_identity.html">layer_identity</a>()</code> <br><code><a href="layer_integer_lookup.html">layer_integer_lookup</a>()</code> <br><code><a href="layer_jax_model_wrapper.html">layer_jax_model_wrapper</a>()</code> <br><code><a href="layer_lambda.html">layer_lambda</a>()</code> <br><code><a href="layer_layer_normalization.html">layer_layer_normalization</a>()</code> <br><code><a href="layer_lstm.html">layer_lstm</a>()</code> <br><code><a href="layer_masking.html">layer_masking</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_max_pooling_1d.html">layer_max_pooling_1d</a>()</code> <br><code><a href="layer_max_pooling_2d.html">layer_max_pooling_2d</a>()</code> <br><code><a href="layer_max_pooling_3d.html">layer_max_pooling_3d</a>()</code> <br><code><a href="layer_maximum.html">layer_maximum</a>()</code> <br><code><a href="layer_mel_spectrogram.html">layer_mel_spectrogram</a>()</code> <br><code><a href="layer_minimum.html">layer_minimum</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_multi_head_attention.html">layer_multi_head_attention</a>()</code> <br><code><a href="layer_multiply.html">layer_multiply</a>()</code> <br><code><a href="layer_normalization.html">layer_normalization</a>()</code> <br><code><a href="layer_permute.html">layer_permute</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_brightness.html">layer_random_brightness</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_contrast.html">layer_random_contrast</a>()</code> <br><code><a href="layer_random_crop.html">layer_random_crop</a>()</code> <br><code><a href="layer_random_elastic_transform.html">layer_random_elastic_transform</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_flip.html">layer_random_flip</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_rotation.html">layer_random_rotation</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_random_translation.html">layer_random_translation</a>()</code> <br><code><a href="layer_random_zoom.html">layer_random_zoom</a>()</code> <br><code><a href="layer_repeat_vector.html">layer_repeat_vector</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_reshape.html">layer_reshape</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_rms_normalization.html">layer_rms_normalization</a>()</code> <br><code><a href="layer_rnn.html">layer_rnn</a>()</code> <br><code><a href="layer_separable_conv_1d.html">layer_separable_conv_1d</a>()</code> <br><code><a href="layer_separable_conv_2d.html">layer_separable_conv_2d</a>()</code> <br><code><a href="layer_simple_rnn.html">layer_simple_rnn</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br><code><a href="layer_spatial_dropout_1d.html">layer_spatial_dropout_1d</a>()</code> <br><code><a href="layer_spatial_dropout_2d.html">layer_spatial_dropout_2d</a>()</code> <br><code><a href="layer_spatial_dropout_3d.html">layer_spatial_dropout_3d</a>()</code> <br><code><a href="layer_spectral_normalization.html">layer_spectral_normalization</a>()</code> <br><code><a href="layer_stft_spectrogram.html">layer_stft_spectrogram</a>()</code> <br><code><a href="layer_string_lookup.html">layer_string_lookup</a>()</code> <br><code><a href="layer_subtract.html">layer_subtract</a>()</code> <br><code><a href="layer_text_vectorization.html">layer_text_vectorization</a>()</code> <br><code><a href="layer_tfsm.html">layer_tfsm</a>()</code> <br><code><a href="layer_time_distributed.html">layer_time_distributed</a>()</code> <br><code><a href="layer_torch_module_wrapper.html">layer_torch_module_wrapper</a>()</code> <br><code><a href="layer_unit_normalization.html">layer_unit_normalization</a>()</code> <br><code><a href="layer_upsampling_1d.html">layer_upsampling_1d</a>()</code> <br><code><a href="layer_upsampling_2d.html">layer_upsampling_2d</a>()</code> <br><code><a href="layer_upsampling_3d.html">layer_upsampling_3d</a>()</code> <br><code><a href="layer_zero_padding_1d.html">layer_zero_padding_1d</a>()</code> <br><code><a href="layer_zero_padding_2d.html">layer_zero_padding_2d</a>()</code> <br><code><a href="layer_zero_padding_3d.html">layer_zero_padding_3d</a>()</code> <br><code><a href="rnn_cell_gru.html">rnn_cell_gru</a>()</code> <br><code><a href="rnn_cell_lstm.html">rnn_cell_lstm</a>()</code> <br><code><a href="rnn_cell_simple.html">rnn_cell_simple</a>()</code> <br><code><a href="rnn_cells_stack.html">rnn_cells_stack</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -151,7 +151,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/layer_cut_mix.html b/docs/dev/reference/layer_cut_mix.html
index 3b4736666..7c685bbb4 100644
--- a/docs/dev/reference/layer_cut_mix.html
+++ b/docs/dev/reference/layer_cut_mix.html
@@ -12,7 +12,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -113,9 +113,9 @@ <h2 id="references">References<a class="anchor" aria-label="anchor" href="#refer
 </ul></div>
     <div class="section level2">
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
-    <div class="dont-index"><p>Other image preprocessing layers: <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br></p>
-<p>Other preprocessing layers: <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_category_encoding.html">layer_category_encoding</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_discretization.html">layer_discretization</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_feature_space.html">layer_feature_space</a>()</code> <br><code><a href="layer_hashed_crossing.html">layer_hashed_crossing</a>()</code> <br><code><a href="layer_hashing.html">layer_hashing</a>()</code> <br><code><a href="layer_integer_lookup.html">layer_integer_lookup</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_mel_spectrogram.html">layer_mel_spectrogram</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_normalization.html">layer_normalization</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_brightness.html">layer_random_brightness</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_contrast.html">layer_random_contrast</a>()</code> <br><code><a href="layer_random_crop.html">layer_random_crop</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_flip.html">layer_random_flip</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_rotation.html">layer_random_rotation</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_random_translation.html">layer_random_translation</a>()</code> <br><code><a href="layer_random_zoom.html">layer_random_zoom</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br><code><a href="layer_stft_spectrogram.html">layer_stft_spectrogram</a>()</code> <br><code><a href="layer_string_lookup.html">layer_string_lookup</a>()</code> <br><code><a href="layer_text_vectorization.html">layer_text_vectorization</a>()</code> <br></p>
-<p>Other layers: <br><code><a href="Layer.html">Layer</a>()</code> <br><code><a href="layer_activation.html">layer_activation</a>()</code> <br><code><a href="layer_activation_elu.html">layer_activation_elu</a>()</code> <br><code><a href="layer_activation_leaky_relu.html">layer_activation_leaky_relu</a>()</code> <br><code><a href="layer_activation_parametric_relu.html">layer_activation_parametric_relu</a>()</code> <br><code><a href="layer_activation_relu.html">layer_activation_relu</a>()</code> <br><code><a href="layer_activation_softmax.html">layer_activation_softmax</a>()</code> <br><code><a href="layer_activity_regularization.html">layer_activity_regularization</a>()</code> <br><code><a href="layer_add.html">layer_add</a>()</code> <br><code><a href="layer_additive_attention.html">layer_additive_attention</a>()</code> <br><code><a href="layer_alpha_dropout.html">layer_alpha_dropout</a>()</code> <br><code><a href="layer_attention.html">layer_attention</a>()</code> <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_average.html">layer_average</a>()</code> <br><code><a href="layer_average_pooling_1d.html">layer_average_pooling_1d</a>()</code> <br><code><a href="layer_average_pooling_2d.html">layer_average_pooling_2d</a>()</code> <br><code><a href="layer_average_pooling_3d.html">layer_average_pooling_3d</a>()</code> <br><code><a href="layer_batch_normalization.html">layer_batch_normalization</a>()</code> <br><code><a href="layer_bidirectional.html">layer_bidirectional</a>()</code> <br><code><a href="layer_category_encoding.html">layer_category_encoding</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_concatenate.html">layer_concatenate</a>()</code> <br><code><a href="layer_conv_1d.html">layer_conv_1d</a>()</code> <br><code><a href="layer_conv_1d_transpose.html">layer_conv_1d_transpose</a>()</code> <br><code><a href="layer_conv_2d.html">layer_conv_2d</a>()</code> <br><code><a href="layer_conv_2d_transpose.html">layer_conv_2d_transpose</a>()</code> <br><code><a href="layer_conv_3d.html">layer_conv_3d</a>()</code> <br><code><a href="layer_conv_3d_transpose.html">layer_conv_3d_transpose</a>()</code> <br><code><a href="layer_conv_lstm_1d.html">layer_conv_lstm_1d</a>()</code> <br><code><a href="layer_conv_lstm_2d.html">layer_conv_lstm_2d</a>()</code> <br><code><a href="layer_conv_lstm_3d.html">layer_conv_lstm_3d</a>()</code> <br><code><a href="layer_cropping_1d.html">layer_cropping_1d</a>()</code> <br><code><a href="layer_cropping_2d.html">layer_cropping_2d</a>()</code> <br><code><a href="layer_cropping_3d.html">layer_cropping_3d</a>()</code> <br><code><a href="layer_dense.html">layer_dense</a>()</code> <br><code><a href="layer_depthwise_conv_1d.html">layer_depthwise_conv_1d</a>()</code> <br><code><a href="layer_depthwise_conv_2d.html">layer_depthwise_conv_2d</a>()</code> <br><code><a href="layer_discretization.html">layer_discretization</a>()</code> <br><code><a href="layer_dot.html">layer_dot</a>()</code> <br><code><a href="layer_dropout.html">layer_dropout</a>()</code> <br><code><a href="layer_einsum_dense.html">layer_einsum_dense</a>()</code> <br><code><a href="layer_embedding.html">layer_embedding</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_feature_space.html">layer_feature_space</a>()</code> <br><code><a href="layer_flatten.html">layer_flatten</a>()</code> <br><code><a href="layer_flax_module_wrapper.html">layer_flax_module_wrapper</a>()</code> <br><code><a href="layer_gaussian_dropout.html">layer_gaussian_dropout</a>()</code> <br><code><a href="layer_gaussian_noise.html">layer_gaussian_noise</a>()</code> <br><code><a href="layer_global_average_pooling_1d.html">layer_global_average_pooling_1d</a>()</code> <br><code><a href="layer_global_average_pooling_2d.html">layer_global_average_pooling_2d</a>()</code> <br><code><a href="layer_global_average_pooling_3d.html">layer_global_average_pooling_3d</a>()</code> <br><code><a href="layer_global_max_pooling_1d.html">layer_global_max_pooling_1d</a>()</code> <br><code><a href="layer_global_max_pooling_2d.html">layer_global_max_pooling_2d</a>()</code> <br><code><a href="layer_global_max_pooling_3d.html">layer_global_max_pooling_3d</a>()</code> <br><code><a href="layer_group_normalization.html">layer_group_normalization</a>()</code> <br><code><a href="layer_group_query_attention.html">layer_group_query_attention</a>()</code> <br><code><a href="layer_gru.html">layer_gru</a>()</code> <br><code><a href="layer_hashed_crossing.html">layer_hashed_crossing</a>()</code> <br><code><a href="layer_hashing.html">layer_hashing</a>()</code> <br><code><a href="layer_identity.html">layer_identity</a>()</code> <br><code><a href="layer_integer_lookup.html">layer_integer_lookup</a>()</code> <br><code><a href="layer_jax_model_wrapper.html">layer_jax_model_wrapper</a>()</code> <br><code><a href="layer_lambda.html">layer_lambda</a>()</code> <br><code><a href="layer_layer_normalization.html">layer_layer_normalization</a>()</code> <br><code><a href="layer_lstm.html">layer_lstm</a>()</code> <br><code><a href="layer_masking.html">layer_masking</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_max_pooling_1d.html">layer_max_pooling_1d</a>()</code> <br><code><a href="layer_max_pooling_2d.html">layer_max_pooling_2d</a>()</code> <br><code><a href="layer_max_pooling_3d.html">layer_max_pooling_3d</a>()</code> <br><code><a href="layer_maximum.html">layer_maximum</a>()</code> <br><code><a href="layer_mel_spectrogram.html">layer_mel_spectrogram</a>()</code> <br><code><a href="layer_minimum.html">layer_minimum</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_multi_head_attention.html">layer_multi_head_attention</a>()</code> <br><code><a href="layer_multiply.html">layer_multiply</a>()</code> <br><code><a href="layer_normalization.html">layer_normalization</a>()</code> <br><code><a href="layer_permute.html">layer_permute</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_brightness.html">layer_random_brightness</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_contrast.html">layer_random_contrast</a>()</code> <br><code><a href="layer_random_crop.html">layer_random_crop</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_flip.html">layer_random_flip</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_rotation.html">layer_random_rotation</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_random_translation.html">layer_random_translation</a>()</code> <br><code><a href="layer_random_zoom.html">layer_random_zoom</a>()</code> <br><code><a href="layer_repeat_vector.html">layer_repeat_vector</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_reshape.html">layer_reshape</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_rms_normalization.html">layer_rms_normalization</a>()</code> <br><code><a href="layer_rnn.html">layer_rnn</a>()</code> <br><code><a href="layer_separable_conv_1d.html">layer_separable_conv_1d</a>()</code> <br><code><a href="layer_separable_conv_2d.html">layer_separable_conv_2d</a>()</code> <br><code><a href="layer_simple_rnn.html">layer_simple_rnn</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br><code><a href="layer_spatial_dropout_1d.html">layer_spatial_dropout_1d</a>()</code> <br><code><a href="layer_spatial_dropout_2d.html">layer_spatial_dropout_2d</a>()</code> <br><code><a href="layer_spatial_dropout_3d.html">layer_spatial_dropout_3d</a>()</code> <br><code><a href="layer_spectral_normalization.html">layer_spectral_normalization</a>()</code> <br><code><a href="layer_stft_spectrogram.html">layer_stft_spectrogram</a>()</code> <br><code><a href="layer_string_lookup.html">layer_string_lookup</a>()</code> <br><code><a href="layer_subtract.html">layer_subtract</a>()</code> <br><code><a href="layer_text_vectorization.html">layer_text_vectorization</a>()</code> <br><code><a href="layer_tfsm.html">layer_tfsm</a>()</code> <br><code><a href="layer_time_distributed.html">layer_time_distributed</a>()</code> <br><code><a href="layer_torch_module_wrapper.html">layer_torch_module_wrapper</a>()</code> <br><code><a href="layer_unit_normalization.html">layer_unit_normalization</a>()</code> <br><code><a href="layer_upsampling_1d.html">layer_upsampling_1d</a>()</code> <br><code><a href="layer_upsampling_2d.html">layer_upsampling_2d</a>()</code> <br><code><a href="layer_upsampling_3d.html">layer_upsampling_3d</a>()</code> <br><code><a href="layer_zero_padding_1d.html">layer_zero_padding_1d</a>()</code> <br><code><a href="layer_zero_padding_2d.html">layer_zero_padding_2d</a>()</code> <br><code><a href="layer_zero_padding_3d.html">layer_zero_padding_3d</a>()</code> <br><code><a href="rnn_cell_gru.html">rnn_cell_gru</a>()</code> <br><code><a href="rnn_cell_lstm.html">rnn_cell_lstm</a>()</code> <br><code><a href="rnn_cell_simple.html">rnn_cell_simple</a>()</code> <br><code><a href="rnn_cells_stack.html">rnn_cells_stack</a>()</code> <br></p></div>
+    <div class="dont-index"><p>Other image preprocessing layers: <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_elastic_transform.html">layer_random_elastic_transform</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br></p>
+<p>Other preprocessing layers: <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_category_encoding.html">layer_category_encoding</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_discretization.html">layer_discretization</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_feature_space.html">layer_feature_space</a>()</code> <br><code><a href="layer_hashed_crossing.html">layer_hashed_crossing</a>()</code> <br><code><a href="layer_hashing.html">layer_hashing</a>()</code> <br><code><a href="layer_integer_lookup.html">layer_integer_lookup</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_mel_spectrogram.html">layer_mel_spectrogram</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_normalization.html">layer_normalization</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_brightness.html">layer_random_brightness</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_contrast.html">layer_random_contrast</a>()</code> <br><code><a href="layer_random_crop.html">layer_random_crop</a>()</code> <br><code><a href="layer_random_elastic_transform.html">layer_random_elastic_transform</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_flip.html">layer_random_flip</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_rotation.html">layer_random_rotation</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_random_translation.html">layer_random_translation</a>()</code> <br><code><a href="layer_random_zoom.html">layer_random_zoom</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br><code><a href="layer_stft_spectrogram.html">layer_stft_spectrogram</a>()</code> <br><code><a href="layer_string_lookup.html">layer_string_lookup</a>()</code> <br><code><a href="layer_text_vectorization.html">layer_text_vectorization</a>()</code> <br></p>
+<p>Other layers: <br><code><a href="Layer.html">Layer</a>()</code> <br><code><a href="layer_activation.html">layer_activation</a>()</code> <br><code><a href="layer_activation_elu.html">layer_activation_elu</a>()</code> <br><code><a href="layer_activation_leaky_relu.html">layer_activation_leaky_relu</a>()</code> <br><code><a href="layer_activation_parametric_relu.html">layer_activation_parametric_relu</a>()</code> <br><code><a href="layer_activation_relu.html">layer_activation_relu</a>()</code> <br><code><a href="layer_activation_softmax.html">layer_activation_softmax</a>()</code> <br><code><a href="layer_activity_regularization.html">layer_activity_regularization</a>()</code> <br><code><a href="layer_add.html">layer_add</a>()</code> <br><code><a href="layer_additive_attention.html">layer_additive_attention</a>()</code> <br><code><a href="layer_alpha_dropout.html">layer_alpha_dropout</a>()</code> <br><code><a href="layer_attention.html">layer_attention</a>()</code> <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_average.html">layer_average</a>()</code> <br><code><a href="layer_average_pooling_1d.html">layer_average_pooling_1d</a>()</code> <br><code><a href="layer_average_pooling_2d.html">layer_average_pooling_2d</a>()</code> <br><code><a href="layer_average_pooling_3d.html">layer_average_pooling_3d</a>()</code> <br><code><a href="layer_batch_normalization.html">layer_batch_normalization</a>()</code> <br><code><a href="layer_bidirectional.html">layer_bidirectional</a>()</code> <br><code><a href="layer_category_encoding.html">layer_category_encoding</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_concatenate.html">layer_concatenate</a>()</code> <br><code><a href="layer_conv_1d.html">layer_conv_1d</a>()</code> <br><code><a href="layer_conv_1d_transpose.html">layer_conv_1d_transpose</a>()</code> <br><code><a href="layer_conv_2d.html">layer_conv_2d</a>()</code> <br><code><a href="layer_conv_2d_transpose.html">layer_conv_2d_transpose</a>()</code> <br><code><a href="layer_conv_3d.html">layer_conv_3d</a>()</code> <br><code><a href="layer_conv_3d_transpose.html">layer_conv_3d_transpose</a>()</code> <br><code><a href="layer_conv_lstm_1d.html">layer_conv_lstm_1d</a>()</code> <br><code><a href="layer_conv_lstm_2d.html">layer_conv_lstm_2d</a>()</code> <br><code><a href="layer_conv_lstm_3d.html">layer_conv_lstm_3d</a>()</code> <br><code><a href="layer_cropping_1d.html">layer_cropping_1d</a>()</code> <br><code><a href="layer_cropping_2d.html">layer_cropping_2d</a>()</code> <br><code><a href="layer_cropping_3d.html">layer_cropping_3d</a>()</code> <br><code><a href="layer_dense.html">layer_dense</a>()</code> <br><code><a href="layer_depthwise_conv_1d.html">layer_depthwise_conv_1d</a>()</code> <br><code><a href="layer_depthwise_conv_2d.html">layer_depthwise_conv_2d</a>()</code> <br><code><a href="layer_discretization.html">layer_discretization</a>()</code> <br><code><a href="layer_dot.html">layer_dot</a>()</code> <br><code><a href="layer_dropout.html">layer_dropout</a>()</code> <br><code><a href="layer_einsum_dense.html">layer_einsum_dense</a>()</code> <br><code><a href="layer_embedding.html">layer_embedding</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_feature_space.html">layer_feature_space</a>()</code> <br><code><a href="layer_flatten.html">layer_flatten</a>()</code> <br><code><a href="layer_flax_module_wrapper.html">layer_flax_module_wrapper</a>()</code> <br><code><a href="layer_gaussian_dropout.html">layer_gaussian_dropout</a>()</code> <br><code><a href="layer_gaussian_noise.html">layer_gaussian_noise</a>()</code> <br><code><a href="layer_global_average_pooling_1d.html">layer_global_average_pooling_1d</a>()</code> <br><code><a href="layer_global_average_pooling_2d.html">layer_global_average_pooling_2d</a>()</code> <br><code><a href="layer_global_average_pooling_3d.html">layer_global_average_pooling_3d</a>()</code> <br><code><a href="layer_global_max_pooling_1d.html">layer_global_max_pooling_1d</a>()</code> <br><code><a href="layer_global_max_pooling_2d.html">layer_global_max_pooling_2d</a>()</code> <br><code><a href="layer_global_max_pooling_3d.html">layer_global_max_pooling_3d</a>()</code> <br><code><a href="layer_group_normalization.html">layer_group_normalization</a>()</code> <br><code><a href="layer_group_query_attention.html">layer_group_query_attention</a>()</code> <br><code><a href="layer_gru.html">layer_gru</a>()</code> <br><code><a href="layer_hashed_crossing.html">layer_hashed_crossing</a>()</code> <br><code><a href="layer_hashing.html">layer_hashing</a>()</code> <br><code><a href="layer_identity.html">layer_identity</a>()</code> <br><code><a href="layer_integer_lookup.html">layer_integer_lookup</a>()</code> <br><code><a href="layer_jax_model_wrapper.html">layer_jax_model_wrapper</a>()</code> <br><code><a href="layer_lambda.html">layer_lambda</a>()</code> <br><code><a href="layer_layer_normalization.html">layer_layer_normalization</a>()</code> <br><code><a href="layer_lstm.html">layer_lstm</a>()</code> <br><code><a href="layer_masking.html">layer_masking</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_max_pooling_1d.html">layer_max_pooling_1d</a>()</code> <br><code><a href="layer_max_pooling_2d.html">layer_max_pooling_2d</a>()</code> <br><code><a href="layer_max_pooling_3d.html">layer_max_pooling_3d</a>()</code> <br><code><a href="layer_maximum.html">layer_maximum</a>()</code> <br><code><a href="layer_mel_spectrogram.html">layer_mel_spectrogram</a>()</code> <br><code><a href="layer_minimum.html">layer_minimum</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_multi_head_attention.html">layer_multi_head_attention</a>()</code> <br><code><a href="layer_multiply.html">layer_multiply</a>()</code> <br><code><a href="layer_normalization.html">layer_normalization</a>()</code> <br><code><a href="layer_permute.html">layer_permute</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_brightness.html">layer_random_brightness</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_contrast.html">layer_random_contrast</a>()</code> <br><code><a href="layer_random_crop.html">layer_random_crop</a>()</code> <br><code><a href="layer_random_elastic_transform.html">layer_random_elastic_transform</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_flip.html">layer_random_flip</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_rotation.html">layer_random_rotation</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_random_translation.html">layer_random_translation</a>()</code> <br><code><a href="layer_random_zoom.html">layer_random_zoom</a>()</code> <br><code><a href="layer_repeat_vector.html">layer_repeat_vector</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_reshape.html">layer_reshape</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_rms_normalization.html">layer_rms_normalization</a>()</code> <br><code><a href="layer_rnn.html">layer_rnn</a>()</code> <br><code><a href="layer_separable_conv_1d.html">layer_separable_conv_1d</a>()</code> <br><code><a href="layer_separable_conv_2d.html">layer_separable_conv_2d</a>()</code> <br><code><a href="layer_simple_rnn.html">layer_simple_rnn</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br><code><a href="layer_spatial_dropout_1d.html">layer_spatial_dropout_1d</a>()</code> <br><code><a href="layer_spatial_dropout_2d.html">layer_spatial_dropout_2d</a>()</code> <br><code><a href="layer_spatial_dropout_3d.html">layer_spatial_dropout_3d</a>()</code> <br><code><a href="layer_spectral_normalization.html">layer_spectral_normalization</a>()</code> <br><code><a href="layer_stft_spectrogram.html">layer_stft_spectrogram</a>()</code> <br><code><a href="layer_string_lookup.html">layer_string_lookup</a>()</code> <br><code><a href="layer_subtract.html">layer_subtract</a>()</code> <br><code><a href="layer_text_vectorization.html">layer_text_vectorization</a>()</code> <br><code><a href="layer_tfsm.html">layer_tfsm</a>()</code> <br><code><a href="layer_time_distributed.html">layer_time_distributed</a>()</code> <br><code><a href="layer_torch_module_wrapper.html">layer_torch_module_wrapper</a>()</code> <br><code><a href="layer_unit_normalization.html">layer_unit_normalization</a>()</code> <br><code><a href="layer_upsampling_1d.html">layer_upsampling_1d</a>()</code> <br><code><a href="layer_upsampling_2d.html">layer_upsampling_2d</a>()</code> <br><code><a href="layer_upsampling_3d.html">layer_upsampling_3d</a>()</code> <br><code><a href="layer_zero_padding_1d.html">layer_zero_padding_1d</a>()</code> <br><code><a href="layer_zero_padding_2d.html">layer_zero_padding_2d</a>()</code> <br><code><a href="layer_zero_padding_3d.html">layer_zero_padding_3d</a>()</code> <br><code><a href="rnn_cell_gru.html">rnn_cell_gru</a>()</code> <br><code><a href="rnn_cell_lstm.html">rnn_cell_lstm</a>()</code> <br><code><a href="rnn_cell_simple.html">rnn_cell_simple</a>()</code> <br><code><a href="rnn_cells_stack.html">rnn_cells_stack</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -127,7 +127,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/layer_dense.html b/docs/dev/reference/layer_dense.html
index a6884846e..e6346452f 100644
--- a/docs/dev/reference/layer_dense.html
+++ b/docs/dev/reference/layer_dense.html
@@ -18,7 +18,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -87,6 +87,7 @@ <h2 id="ref-usage">Usage<a class="anchor" aria-label="anchor" href="#ref-usage">
 <span>  kernel_constraint <span class="op">=</span> <span class="cn">NULL</span>,</span>
 <span>  bias_constraint <span class="op">=</span> <span class="cn">NULL</span>,</span>
 <span>  lora_rank <span class="op">=</span> <span class="cn">NULL</span>,</span>
+<span>  lora_alpha <span class="op">=</span> <span class="cn">NULL</span>,</span>
 <span>  <span class="va">...</span></span>
 <span><span class="op">)</span></span></code></pre></div>
     </div>
@@ -156,6 +157,12 @@ <h2 id="arguments">Arguments<a class="anchor" aria-label="anchor" href="#argumen
 <code>Dense</code> layer by calling <code>layer$enable_lora(rank)</code>.</p></dd>
 
 
+<dt id="arg-lora-alpha">lora_alpha<a class="anchor" aria-label="anchor" href="#arg-lora-alpha"></a></dt>
+<dd><p>Optional integer. Scales the low-rank adaptation delta during the forward
+pass. The delta is scaled by <code>lora_alpha / lora_rank</code>, letting you tune the
+LoRA adjustment strength independently of <code>lora_rank</code>.</p></dd>
+
+
 <dt id="arg--">...<a class="anchor" aria-label="anchor" href="#arg--"></a></dt>
 <dd><p>For forward/backward compatability.</p></dd>
 
@@ -198,6 +205,7 @@ <h2 id="methods">Methods<a class="anchor" aria-label="anchor" href="#methods"></
 <ul><li><p></p>
 <p></p><div class="sourceCode r"><pre><code><span><span class="fu">enable_lora</span><span class="op">(</span></span>
 <span>  <span class="va">rank</span>,</span>
+<span>  lora_alpha <span class="op">=</span> <span class="cn">NULL</span>,</span>
 <span>  a_initializer <span class="op">=</span> <span class="st">'he_uniform'</span>,</span>
 <span>  b_initializer <span class="op">=</span> <span class="st">'zeros'</span></span>
 <span><span class="op">)</span></span></code></pre><p></p></div></li>
@@ -214,7 +222,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
     <div class="dont-index">
 <ul><li><p><a href="https://keras.io/api/layers/core_layers/dense#dense-class" class="external-link">https://keras.io/api/layers/core_layers/dense#dense-class</a></p></li>
 </ul><p>Other core layers: <br><code><a href="layer_einsum_dense.html">layer_einsum_dense</a>()</code> <br><code><a href="layer_embedding.html">layer_embedding</a>()</code> <br><code><a href="layer_identity.html">layer_identity</a>()</code> <br><code><a href="layer_lambda.html">layer_lambda</a>()</code> <br><code><a href="layer_masking.html">layer_masking</a>()</code> <br></p>
-<p>Other layers: <br><code><a href="Layer.html">Layer</a>()</code> <br><code><a href="layer_activation.html">layer_activation</a>()</code> <br><code><a href="layer_activation_elu.html">layer_activation_elu</a>()</code> <br><code><a href="layer_activation_leaky_relu.html">layer_activation_leaky_relu</a>()</code> <br><code><a href="layer_activation_parametric_relu.html">layer_activation_parametric_relu</a>()</code> <br><code><a href="layer_activation_relu.html">layer_activation_relu</a>()</code> <br><code><a href="layer_activation_softmax.html">layer_activation_softmax</a>()</code> <br><code><a href="layer_activity_regularization.html">layer_activity_regularization</a>()</code> <br><code><a href="layer_add.html">layer_add</a>()</code> <br><code><a href="layer_additive_attention.html">layer_additive_attention</a>()</code> <br><code><a href="layer_alpha_dropout.html">layer_alpha_dropout</a>()</code> <br><code><a href="layer_attention.html">layer_attention</a>()</code> <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_average.html">layer_average</a>()</code> <br><code><a href="layer_average_pooling_1d.html">layer_average_pooling_1d</a>()</code> <br><code><a href="layer_average_pooling_2d.html">layer_average_pooling_2d</a>()</code> <br><code><a href="layer_average_pooling_3d.html">layer_average_pooling_3d</a>()</code> <br><code><a href="layer_batch_normalization.html">layer_batch_normalization</a>()</code> <br><code><a href="layer_bidirectional.html">layer_bidirectional</a>()</code> <br><code><a href="layer_category_encoding.html">layer_category_encoding</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_concatenate.html">layer_concatenate</a>()</code> <br><code><a href="layer_conv_1d.html">layer_conv_1d</a>()</code> <br><code><a href="layer_conv_1d_transpose.html">layer_conv_1d_transpose</a>()</code> <br><code><a href="layer_conv_2d.html">layer_conv_2d</a>()</code> <br><code><a href="layer_conv_2d_transpose.html">layer_conv_2d_transpose</a>()</code> <br><code><a href="layer_conv_3d.html">layer_conv_3d</a>()</code> <br><code><a href="layer_conv_3d_transpose.html">layer_conv_3d_transpose</a>()</code> <br><code><a href="layer_conv_lstm_1d.html">layer_conv_lstm_1d</a>()</code> <br><code><a href="layer_conv_lstm_2d.html">layer_conv_lstm_2d</a>()</code> <br><code><a href="layer_conv_lstm_3d.html">layer_conv_lstm_3d</a>()</code> <br><code><a href="layer_cropping_1d.html">layer_cropping_1d</a>()</code> <br><code><a href="layer_cropping_2d.html">layer_cropping_2d</a>()</code> <br><code><a href="layer_cropping_3d.html">layer_cropping_3d</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_depthwise_conv_1d.html">layer_depthwise_conv_1d</a>()</code> <br><code><a href="layer_depthwise_conv_2d.html">layer_depthwise_conv_2d</a>()</code> <br><code><a href="layer_discretization.html">layer_discretization</a>()</code> <br><code><a href="layer_dot.html">layer_dot</a>()</code> <br><code><a href="layer_dropout.html">layer_dropout</a>()</code> <br><code><a href="layer_einsum_dense.html">layer_einsum_dense</a>()</code> <br><code><a href="layer_embedding.html">layer_embedding</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_feature_space.html">layer_feature_space</a>()</code> <br><code><a href="layer_flatten.html">layer_flatten</a>()</code> <br><code><a href="layer_flax_module_wrapper.html">layer_flax_module_wrapper</a>()</code> <br><code><a href="layer_gaussian_dropout.html">layer_gaussian_dropout</a>()</code> <br><code><a href="layer_gaussian_noise.html">layer_gaussian_noise</a>()</code> <br><code><a href="layer_global_average_pooling_1d.html">layer_global_average_pooling_1d</a>()</code> <br><code><a href="layer_global_average_pooling_2d.html">layer_global_average_pooling_2d</a>()</code> <br><code><a href="layer_global_average_pooling_3d.html">layer_global_average_pooling_3d</a>()</code> <br><code><a href="layer_global_max_pooling_1d.html">layer_global_max_pooling_1d</a>()</code> <br><code><a href="layer_global_max_pooling_2d.html">layer_global_max_pooling_2d</a>()</code> <br><code><a href="layer_global_max_pooling_3d.html">layer_global_max_pooling_3d</a>()</code> <br><code><a href="layer_group_normalization.html">layer_group_normalization</a>()</code> <br><code><a href="layer_group_query_attention.html">layer_group_query_attention</a>()</code> <br><code><a href="layer_gru.html">layer_gru</a>()</code> <br><code><a href="layer_hashed_crossing.html">layer_hashed_crossing</a>()</code> <br><code><a href="layer_hashing.html">layer_hashing</a>()</code> <br><code><a href="layer_identity.html">layer_identity</a>()</code> <br><code><a href="layer_integer_lookup.html">layer_integer_lookup</a>()</code> <br><code><a href="layer_jax_model_wrapper.html">layer_jax_model_wrapper</a>()</code> <br><code><a href="layer_lambda.html">layer_lambda</a>()</code> <br><code><a href="layer_layer_normalization.html">layer_layer_normalization</a>()</code> <br><code><a href="layer_lstm.html">layer_lstm</a>()</code> <br><code><a href="layer_masking.html">layer_masking</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_max_pooling_1d.html">layer_max_pooling_1d</a>()</code> <br><code><a href="layer_max_pooling_2d.html">layer_max_pooling_2d</a>()</code> <br><code><a href="layer_max_pooling_3d.html">layer_max_pooling_3d</a>()</code> <br><code><a href="layer_maximum.html">layer_maximum</a>()</code> <br><code><a href="layer_mel_spectrogram.html">layer_mel_spectrogram</a>()</code> <br><code><a href="layer_minimum.html">layer_minimum</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_multi_head_attention.html">layer_multi_head_attention</a>()</code> <br><code><a href="layer_multiply.html">layer_multiply</a>()</code> <br><code><a href="layer_normalization.html">layer_normalization</a>()</code> <br><code><a href="layer_permute.html">layer_permute</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_brightness.html">layer_random_brightness</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_contrast.html">layer_random_contrast</a>()</code> <br><code><a href="layer_random_crop.html">layer_random_crop</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_flip.html">layer_random_flip</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_rotation.html">layer_random_rotation</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_random_translation.html">layer_random_translation</a>()</code> <br><code><a href="layer_random_zoom.html">layer_random_zoom</a>()</code> <br><code><a href="layer_repeat_vector.html">layer_repeat_vector</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_reshape.html">layer_reshape</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_rms_normalization.html">layer_rms_normalization</a>()</code> <br><code><a href="layer_rnn.html">layer_rnn</a>()</code> <br><code><a href="layer_separable_conv_1d.html">layer_separable_conv_1d</a>()</code> <br><code><a href="layer_separable_conv_2d.html">layer_separable_conv_2d</a>()</code> <br><code><a href="layer_simple_rnn.html">layer_simple_rnn</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br><code><a href="layer_spatial_dropout_1d.html">layer_spatial_dropout_1d</a>()</code> <br><code><a href="layer_spatial_dropout_2d.html">layer_spatial_dropout_2d</a>()</code> <br><code><a href="layer_spatial_dropout_3d.html">layer_spatial_dropout_3d</a>()</code> <br><code><a href="layer_spectral_normalization.html">layer_spectral_normalization</a>()</code> <br><code><a href="layer_stft_spectrogram.html">layer_stft_spectrogram</a>()</code> <br><code><a href="layer_string_lookup.html">layer_string_lookup</a>()</code> <br><code><a href="layer_subtract.html">layer_subtract</a>()</code> <br><code><a href="layer_text_vectorization.html">layer_text_vectorization</a>()</code> <br><code><a href="layer_tfsm.html">layer_tfsm</a>()</code> <br><code><a href="layer_time_distributed.html">layer_time_distributed</a>()</code> <br><code><a href="layer_torch_module_wrapper.html">layer_torch_module_wrapper</a>()</code> <br><code><a href="layer_unit_normalization.html">layer_unit_normalization</a>()</code> <br><code><a href="layer_upsampling_1d.html">layer_upsampling_1d</a>()</code> <br><code><a href="layer_upsampling_2d.html">layer_upsampling_2d</a>()</code> <br><code><a href="layer_upsampling_3d.html">layer_upsampling_3d</a>()</code> <br><code><a href="layer_zero_padding_1d.html">layer_zero_padding_1d</a>()</code> <br><code><a href="layer_zero_padding_2d.html">layer_zero_padding_2d</a>()</code> <br><code><a href="layer_zero_padding_3d.html">layer_zero_padding_3d</a>()</code> <br><code><a href="rnn_cell_gru.html">rnn_cell_gru</a>()</code> <br><code><a href="rnn_cell_lstm.html">rnn_cell_lstm</a>()</code> <br><code><a href="rnn_cell_simple.html">rnn_cell_simple</a>()</code> <br><code><a href="rnn_cells_stack.html">rnn_cells_stack</a>()</code> <br></p></div>
+<p>Other layers: <br><code><a href="Layer.html">Layer</a>()</code> <br><code><a href="layer_activation.html">layer_activation</a>()</code> <br><code><a href="layer_activation_elu.html">layer_activation_elu</a>()</code> <br><code><a href="layer_activation_leaky_relu.html">layer_activation_leaky_relu</a>()</code> <br><code><a href="layer_activation_parametric_relu.html">layer_activation_parametric_relu</a>()</code> <br><code><a href="layer_activation_relu.html">layer_activation_relu</a>()</code> <br><code><a href="layer_activation_softmax.html">layer_activation_softmax</a>()</code> <br><code><a href="layer_activity_regularization.html">layer_activity_regularization</a>()</code> <br><code><a href="layer_add.html">layer_add</a>()</code> <br><code><a href="layer_additive_attention.html">layer_additive_attention</a>()</code> <br><code><a href="layer_alpha_dropout.html">layer_alpha_dropout</a>()</code> <br><code><a href="layer_attention.html">layer_attention</a>()</code> <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_average.html">layer_average</a>()</code> <br><code><a href="layer_average_pooling_1d.html">layer_average_pooling_1d</a>()</code> <br><code><a href="layer_average_pooling_2d.html">layer_average_pooling_2d</a>()</code> <br><code><a href="layer_average_pooling_3d.html">layer_average_pooling_3d</a>()</code> <br><code><a href="layer_batch_normalization.html">layer_batch_normalization</a>()</code> <br><code><a href="layer_bidirectional.html">layer_bidirectional</a>()</code> <br><code><a href="layer_category_encoding.html">layer_category_encoding</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_concatenate.html">layer_concatenate</a>()</code> <br><code><a href="layer_conv_1d.html">layer_conv_1d</a>()</code> <br><code><a href="layer_conv_1d_transpose.html">layer_conv_1d_transpose</a>()</code> <br><code><a href="layer_conv_2d.html">layer_conv_2d</a>()</code> <br><code><a href="layer_conv_2d_transpose.html">layer_conv_2d_transpose</a>()</code> <br><code><a href="layer_conv_3d.html">layer_conv_3d</a>()</code> <br><code><a href="layer_conv_3d_transpose.html">layer_conv_3d_transpose</a>()</code> <br><code><a href="layer_conv_lstm_1d.html">layer_conv_lstm_1d</a>()</code> <br><code><a href="layer_conv_lstm_2d.html">layer_conv_lstm_2d</a>()</code> <br><code><a href="layer_conv_lstm_3d.html">layer_conv_lstm_3d</a>()</code> <br><code><a href="layer_cropping_1d.html">layer_cropping_1d</a>()</code> <br><code><a href="layer_cropping_2d.html">layer_cropping_2d</a>()</code> <br><code><a href="layer_cropping_3d.html">layer_cropping_3d</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_depthwise_conv_1d.html">layer_depthwise_conv_1d</a>()</code> <br><code><a href="layer_depthwise_conv_2d.html">layer_depthwise_conv_2d</a>()</code> <br><code><a href="layer_discretization.html">layer_discretization</a>()</code> <br><code><a href="layer_dot.html">layer_dot</a>()</code> <br><code><a href="layer_dropout.html">layer_dropout</a>()</code> <br><code><a href="layer_einsum_dense.html">layer_einsum_dense</a>()</code> <br><code><a href="layer_embedding.html">layer_embedding</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_feature_space.html">layer_feature_space</a>()</code> <br><code><a href="layer_flatten.html">layer_flatten</a>()</code> <br><code><a href="layer_flax_module_wrapper.html">layer_flax_module_wrapper</a>()</code> <br><code><a href="layer_gaussian_dropout.html">layer_gaussian_dropout</a>()</code> <br><code><a href="layer_gaussian_noise.html">layer_gaussian_noise</a>()</code> <br><code><a href="layer_global_average_pooling_1d.html">layer_global_average_pooling_1d</a>()</code> <br><code><a href="layer_global_average_pooling_2d.html">layer_global_average_pooling_2d</a>()</code> <br><code><a href="layer_global_average_pooling_3d.html">layer_global_average_pooling_3d</a>()</code> <br><code><a href="layer_global_max_pooling_1d.html">layer_global_max_pooling_1d</a>()</code> <br><code><a href="layer_global_max_pooling_2d.html">layer_global_max_pooling_2d</a>()</code> <br><code><a href="layer_global_max_pooling_3d.html">layer_global_max_pooling_3d</a>()</code> <br><code><a href="layer_group_normalization.html">layer_group_normalization</a>()</code> <br><code><a href="layer_group_query_attention.html">layer_group_query_attention</a>()</code> <br><code><a href="layer_gru.html">layer_gru</a>()</code> <br><code><a href="layer_hashed_crossing.html">layer_hashed_crossing</a>()</code> <br><code><a href="layer_hashing.html">layer_hashing</a>()</code> <br><code><a href="layer_identity.html">layer_identity</a>()</code> <br><code><a href="layer_integer_lookup.html">layer_integer_lookup</a>()</code> <br><code><a href="layer_jax_model_wrapper.html">layer_jax_model_wrapper</a>()</code> <br><code><a href="layer_lambda.html">layer_lambda</a>()</code> <br><code><a href="layer_layer_normalization.html">layer_layer_normalization</a>()</code> <br><code><a href="layer_lstm.html">layer_lstm</a>()</code> <br><code><a href="layer_masking.html">layer_masking</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_max_pooling_1d.html">layer_max_pooling_1d</a>()</code> <br><code><a href="layer_max_pooling_2d.html">layer_max_pooling_2d</a>()</code> <br><code><a href="layer_max_pooling_3d.html">layer_max_pooling_3d</a>()</code> <br><code><a href="layer_maximum.html">layer_maximum</a>()</code> <br><code><a href="layer_mel_spectrogram.html">layer_mel_spectrogram</a>()</code> <br><code><a href="layer_minimum.html">layer_minimum</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_multi_head_attention.html">layer_multi_head_attention</a>()</code> <br><code><a href="layer_multiply.html">layer_multiply</a>()</code> <br><code><a href="layer_normalization.html">layer_normalization</a>()</code> <br><code><a href="layer_permute.html">layer_permute</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_brightness.html">layer_random_brightness</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_contrast.html">layer_random_contrast</a>()</code> <br><code><a href="layer_random_crop.html">layer_random_crop</a>()</code> <br><code><a href="layer_random_elastic_transform.html">layer_random_elastic_transform</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_flip.html">layer_random_flip</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_rotation.html">layer_random_rotation</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_random_translation.html">layer_random_translation</a>()</code> <br><code><a href="layer_random_zoom.html">layer_random_zoom</a>()</code> <br><code><a href="layer_repeat_vector.html">layer_repeat_vector</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_reshape.html">layer_reshape</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_rms_normalization.html">layer_rms_normalization</a>()</code> <br><code><a href="layer_rnn.html">layer_rnn</a>()</code> <br><code><a href="layer_separable_conv_1d.html">layer_separable_conv_1d</a>()</code> <br><code><a href="layer_separable_conv_2d.html">layer_separable_conv_2d</a>()</code> <br><code><a href="layer_simple_rnn.html">layer_simple_rnn</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br><code><a href="layer_spatial_dropout_1d.html">layer_spatial_dropout_1d</a>()</code> <br><code><a href="layer_spatial_dropout_2d.html">layer_spatial_dropout_2d</a>()</code> <br><code><a href="layer_spatial_dropout_3d.html">layer_spatial_dropout_3d</a>()</code> <br><code><a href="layer_spectral_normalization.html">layer_spectral_normalization</a>()</code> <br><code><a href="layer_stft_spectrogram.html">layer_stft_spectrogram</a>()</code> <br><code><a href="layer_string_lookup.html">layer_string_lookup</a>()</code> <br><code><a href="layer_subtract.html">layer_subtract</a>()</code> <br><code><a href="layer_text_vectorization.html">layer_text_vectorization</a>()</code> <br><code><a href="layer_tfsm.html">layer_tfsm</a>()</code> <br><code><a href="layer_time_distributed.html">layer_time_distributed</a>()</code> <br><code><a href="layer_torch_module_wrapper.html">layer_torch_module_wrapper</a>()</code> <br><code><a href="layer_unit_normalization.html">layer_unit_normalization</a>()</code> <br><code><a href="layer_upsampling_1d.html">layer_upsampling_1d</a>()</code> <br><code><a href="layer_upsampling_2d.html">layer_upsampling_2d</a>()</code> <br><code><a href="layer_upsampling_3d.html">layer_upsampling_3d</a>()</code> <br><code><a href="layer_zero_padding_1d.html">layer_zero_padding_1d</a>()</code> <br><code><a href="layer_zero_padding_2d.html">layer_zero_padding_2d</a>()</code> <br><code><a href="layer_zero_padding_3d.html">layer_zero_padding_3d</a>()</code> <br><code><a href="rnn_cell_gru.html">rnn_cell_gru</a>()</code> <br><code><a href="rnn_cell_lstm.html">rnn_cell_lstm</a>()</code> <br><code><a href="rnn_cell_simple.html">rnn_cell_simple</a>()</code> <br><code><a href="rnn_cells_stack.html">rnn_cells_stack</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -226,7 +234,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/layer_depthwise_conv_1d.html b/docs/dev/reference/layer_depthwise_conv_1d.html
index dee93a2ea..ebc7cf0f5 100644
--- a/docs/dev/reference/layer_depthwise_conv_1d.html
+++ b/docs/dev/reference/layer_depthwise_conv_1d.html
@@ -38,7 +38,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -260,7 +260,7 @@ <h2 id="example">Example<a class="anchor" aria-label="anchor" href="#example"></
     <div class="section level2">
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
     <div class="dont-index"><p>Other convolutional layers: <br><code><a href="layer_conv_1d.html">layer_conv_1d</a>()</code> <br><code><a href="layer_conv_1d_transpose.html">layer_conv_1d_transpose</a>()</code> <br><code><a href="layer_conv_2d.html">layer_conv_2d</a>()</code> <br><code><a href="layer_conv_2d_transpose.html">layer_conv_2d_transpose</a>()</code> <br><code><a href="layer_conv_3d.html">layer_conv_3d</a>()</code> <br><code><a href="layer_conv_3d_transpose.html">layer_conv_3d_transpose</a>()</code> <br><code><a href="layer_depthwise_conv_2d.html">layer_depthwise_conv_2d</a>()</code> <br><code><a href="layer_separable_conv_1d.html">layer_separable_conv_1d</a>()</code> <br><code><a href="layer_separable_conv_2d.html">layer_separable_conv_2d</a>()</code> <br></p>
-<p>Other layers: <br><code><a href="Layer.html">Layer</a>()</code> <br><code><a href="layer_activation.html">layer_activation</a>()</code> <br><code><a href="layer_activation_elu.html">layer_activation_elu</a>()</code> <br><code><a href="layer_activation_leaky_relu.html">layer_activation_leaky_relu</a>()</code> <br><code><a href="layer_activation_parametric_relu.html">layer_activation_parametric_relu</a>()</code> <br><code><a href="layer_activation_relu.html">layer_activation_relu</a>()</code> <br><code><a href="layer_activation_softmax.html">layer_activation_softmax</a>()</code> <br><code><a href="layer_activity_regularization.html">layer_activity_regularization</a>()</code> <br><code><a href="layer_add.html">layer_add</a>()</code> <br><code><a href="layer_additive_attention.html">layer_additive_attention</a>()</code> <br><code><a href="layer_alpha_dropout.html">layer_alpha_dropout</a>()</code> <br><code><a href="layer_attention.html">layer_attention</a>()</code> <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_average.html">layer_average</a>()</code> <br><code><a href="layer_average_pooling_1d.html">layer_average_pooling_1d</a>()</code> <br><code><a href="layer_average_pooling_2d.html">layer_average_pooling_2d</a>()</code> <br><code><a href="layer_average_pooling_3d.html">layer_average_pooling_3d</a>()</code> <br><code><a href="layer_batch_normalization.html">layer_batch_normalization</a>()</code> <br><code><a href="layer_bidirectional.html">layer_bidirectional</a>()</code> <br><code><a href="layer_category_encoding.html">layer_category_encoding</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_concatenate.html">layer_concatenate</a>()</code> <br><code><a href="layer_conv_1d.html">layer_conv_1d</a>()</code> <br><code><a href="layer_conv_1d_transpose.html">layer_conv_1d_transpose</a>()</code> <br><code><a href="layer_conv_2d.html">layer_conv_2d</a>()</code> <br><code><a href="layer_conv_2d_transpose.html">layer_conv_2d_transpose</a>()</code> <br><code><a href="layer_conv_3d.html">layer_conv_3d</a>()</code> <br><code><a href="layer_conv_3d_transpose.html">layer_conv_3d_transpose</a>()</code> <br><code><a href="layer_conv_lstm_1d.html">layer_conv_lstm_1d</a>()</code> <br><code><a href="layer_conv_lstm_2d.html">layer_conv_lstm_2d</a>()</code> <br><code><a href="layer_conv_lstm_3d.html">layer_conv_lstm_3d</a>()</code> <br><code><a href="layer_cropping_1d.html">layer_cropping_1d</a>()</code> <br><code><a href="layer_cropping_2d.html">layer_cropping_2d</a>()</code> <br><code><a href="layer_cropping_3d.html">layer_cropping_3d</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_dense.html">layer_dense</a>()</code> <br><code><a href="layer_depthwise_conv_2d.html">layer_depthwise_conv_2d</a>()</code> <br><code><a href="layer_discretization.html">layer_discretization</a>()</code> <br><code><a href="layer_dot.html">layer_dot</a>()</code> <br><code><a href="layer_dropout.html">layer_dropout</a>()</code> <br><code><a href="layer_einsum_dense.html">layer_einsum_dense</a>()</code> <br><code><a href="layer_embedding.html">layer_embedding</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_feature_space.html">layer_feature_space</a>()</code> <br><code><a href="layer_flatten.html">layer_flatten</a>()</code> <br><code><a href="layer_flax_module_wrapper.html">layer_flax_module_wrapper</a>()</code> <br><code><a href="layer_gaussian_dropout.html">layer_gaussian_dropout</a>()</code> <br><code><a href="layer_gaussian_noise.html">layer_gaussian_noise</a>()</code> <br><code><a href="layer_global_average_pooling_1d.html">layer_global_average_pooling_1d</a>()</code> <br><code><a href="layer_global_average_pooling_2d.html">layer_global_average_pooling_2d</a>()</code> <br><code><a href="layer_global_average_pooling_3d.html">layer_global_average_pooling_3d</a>()</code> <br><code><a href="layer_global_max_pooling_1d.html">layer_global_max_pooling_1d</a>()</code> <br><code><a href="layer_global_max_pooling_2d.html">layer_global_max_pooling_2d</a>()</code> <br><code><a href="layer_global_max_pooling_3d.html">layer_global_max_pooling_3d</a>()</code> <br><code><a href="layer_group_normalization.html">layer_group_normalization</a>()</code> <br><code><a href="layer_group_query_attention.html">layer_group_query_attention</a>()</code> <br><code><a href="layer_gru.html">layer_gru</a>()</code> <br><code><a href="layer_hashed_crossing.html">layer_hashed_crossing</a>()</code> <br><code><a href="layer_hashing.html">layer_hashing</a>()</code> <br><code><a href="layer_identity.html">layer_identity</a>()</code> <br><code><a href="layer_integer_lookup.html">layer_integer_lookup</a>()</code> <br><code><a href="layer_jax_model_wrapper.html">layer_jax_model_wrapper</a>()</code> <br><code><a href="layer_lambda.html">layer_lambda</a>()</code> <br><code><a href="layer_layer_normalization.html">layer_layer_normalization</a>()</code> <br><code><a href="layer_lstm.html">layer_lstm</a>()</code> <br><code><a href="layer_masking.html">layer_masking</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_max_pooling_1d.html">layer_max_pooling_1d</a>()</code> <br><code><a href="layer_max_pooling_2d.html">layer_max_pooling_2d</a>()</code> <br><code><a href="layer_max_pooling_3d.html">layer_max_pooling_3d</a>()</code> <br><code><a href="layer_maximum.html">layer_maximum</a>()</code> <br><code><a href="layer_mel_spectrogram.html">layer_mel_spectrogram</a>()</code> <br><code><a href="layer_minimum.html">layer_minimum</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_multi_head_attention.html">layer_multi_head_attention</a>()</code> <br><code><a href="layer_multiply.html">layer_multiply</a>()</code> <br><code><a href="layer_normalization.html">layer_normalization</a>()</code> <br><code><a href="layer_permute.html">layer_permute</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_brightness.html">layer_random_brightness</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_contrast.html">layer_random_contrast</a>()</code> <br><code><a href="layer_random_crop.html">layer_random_crop</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_flip.html">layer_random_flip</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_rotation.html">layer_random_rotation</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_random_translation.html">layer_random_translation</a>()</code> <br><code><a href="layer_random_zoom.html">layer_random_zoom</a>()</code> <br><code><a href="layer_repeat_vector.html">layer_repeat_vector</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_reshape.html">layer_reshape</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_rms_normalization.html">layer_rms_normalization</a>()</code> <br><code><a href="layer_rnn.html">layer_rnn</a>()</code> <br><code><a href="layer_separable_conv_1d.html">layer_separable_conv_1d</a>()</code> <br><code><a href="layer_separable_conv_2d.html">layer_separable_conv_2d</a>()</code> <br><code><a href="layer_simple_rnn.html">layer_simple_rnn</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br><code><a href="layer_spatial_dropout_1d.html">layer_spatial_dropout_1d</a>()</code> <br><code><a href="layer_spatial_dropout_2d.html">layer_spatial_dropout_2d</a>()</code> <br><code><a href="layer_spatial_dropout_3d.html">layer_spatial_dropout_3d</a>()</code> <br><code><a href="layer_spectral_normalization.html">layer_spectral_normalization</a>()</code> <br><code><a href="layer_stft_spectrogram.html">layer_stft_spectrogram</a>()</code> <br><code><a href="layer_string_lookup.html">layer_string_lookup</a>()</code> <br><code><a href="layer_subtract.html">layer_subtract</a>()</code> <br><code><a href="layer_text_vectorization.html">layer_text_vectorization</a>()</code> <br><code><a href="layer_tfsm.html">layer_tfsm</a>()</code> <br><code><a href="layer_time_distributed.html">layer_time_distributed</a>()</code> <br><code><a href="layer_torch_module_wrapper.html">layer_torch_module_wrapper</a>()</code> <br><code><a href="layer_unit_normalization.html">layer_unit_normalization</a>()</code> <br><code><a href="layer_upsampling_1d.html">layer_upsampling_1d</a>()</code> <br><code><a href="layer_upsampling_2d.html">layer_upsampling_2d</a>()</code> <br><code><a href="layer_upsampling_3d.html">layer_upsampling_3d</a>()</code> <br><code><a href="layer_zero_padding_1d.html">layer_zero_padding_1d</a>()</code> <br><code><a href="layer_zero_padding_2d.html">layer_zero_padding_2d</a>()</code> <br><code><a href="layer_zero_padding_3d.html">layer_zero_padding_3d</a>()</code> <br><code><a href="rnn_cell_gru.html">rnn_cell_gru</a>()</code> <br><code><a href="rnn_cell_lstm.html">rnn_cell_lstm</a>()</code> <br><code><a href="rnn_cell_simple.html">rnn_cell_simple</a>()</code> <br><code><a href="rnn_cells_stack.html">rnn_cells_stack</a>()</code> <br></p></div>
+<p>Other layers: <br><code><a href="Layer.html">Layer</a>()</code> <br><code><a href="layer_activation.html">layer_activation</a>()</code> <br><code><a href="layer_activation_elu.html">layer_activation_elu</a>()</code> <br><code><a href="layer_activation_leaky_relu.html">layer_activation_leaky_relu</a>()</code> <br><code><a href="layer_activation_parametric_relu.html">layer_activation_parametric_relu</a>()</code> <br><code><a href="layer_activation_relu.html">layer_activation_relu</a>()</code> <br><code><a href="layer_activation_softmax.html">layer_activation_softmax</a>()</code> <br><code><a href="layer_activity_regularization.html">layer_activity_regularization</a>()</code> <br><code><a href="layer_add.html">layer_add</a>()</code> <br><code><a href="layer_additive_attention.html">layer_additive_attention</a>()</code> <br><code><a href="layer_alpha_dropout.html">layer_alpha_dropout</a>()</code> <br><code><a href="layer_attention.html">layer_attention</a>()</code> <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_average.html">layer_average</a>()</code> <br><code><a href="layer_average_pooling_1d.html">layer_average_pooling_1d</a>()</code> <br><code><a href="layer_average_pooling_2d.html">layer_average_pooling_2d</a>()</code> <br><code><a href="layer_average_pooling_3d.html">layer_average_pooling_3d</a>()</code> <br><code><a href="layer_batch_normalization.html">layer_batch_normalization</a>()</code> <br><code><a href="layer_bidirectional.html">layer_bidirectional</a>()</code> <br><code><a href="layer_category_encoding.html">layer_category_encoding</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_concatenate.html">layer_concatenate</a>()</code> <br><code><a href="layer_conv_1d.html">layer_conv_1d</a>()</code> <br><code><a href="layer_conv_1d_transpose.html">layer_conv_1d_transpose</a>()</code> <br><code><a href="layer_conv_2d.html">layer_conv_2d</a>()</code> <br><code><a href="layer_conv_2d_transpose.html">layer_conv_2d_transpose</a>()</code> <br><code><a href="layer_conv_3d.html">layer_conv_3d</a>()</code> <br><code><a href="layer_conv_3d_transpose.html">layer_conv_3d_transpose</a>()</code> <br><code><a href="layer_conv_lstm_1d.html">layer_conv_lstm_1d</a>()</code> <br><code><a href="layer_conv_lstm_2d.html">layer_conv_lstm_2d</a>()</code> <br><code><a href="layer_conv_lstm_3d.html">layer_conv_lstm_3d</a>()</code> <br><code><a href="layer_cropping_1d.html">layer_cropping_1d</a>()</code> <br><code><a href="layer_cropping_2d.html">layer_cropping_2d</a>()</code> <br><code><a href="layer_cropping_3d.html">layer_cropping_3d</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_dense.html">layer_dense</a>()</code> <br><code><a href="layer_depthwise_conv_2d.html">layer_depthwise_conv_2d</a>()</code> <br><code><a href="layer_discretization.html">layer_discretization</a>()</code> <br><code><a href="layer_dot.html">layer_dot</a>()</code> <br><code><a href="layer_dropout.html">layer_dropout</a>()</code> <br><code><a href="layer_einsum_dense.html">layer_einsum_dense</a>()</code> <br><code><a href="layer_embedding.html">layer_embedding</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_feature_space.html">layer_feature_space</a>()</code> <br><code><a href="layer_flatten.html">layer_flatten</a>()</code> <br><code><a href="layer_flax_module_wrapper.html">layer_flax_module_wrapper</a>()</code> <br><code><a href="layer_gaussian_dropout.html">layer_gaussian_dropout</a>()</code> <br><code><a href="layer_gaussian_noise.html">layer_gaussian_noise</a>()</code> <br><code><a href="layer_global_average_pooling_1d.html">layer_global_average_pooling_1d</a>()</code> <br><code><a href="layer_global_average_pooling_2d.html">layer_global_average_pooling_2d</a>()</code> <br><code><a href="layer_global_average_pooling_3d.html">layer_global_average_pooling_3d</a>()</code> <br><code><a href="layer_global_max_pooling_1d.html">layer_global_max_pooling_1d</a>()</code> <br><code><a href="layer_global_max_pooling_2d.html">layer_global_max_pooling_2d</a>()</code> <br><code><a href="layer_global_max_pooling_3d.html">layer_global_max_pooling_3d</a>()</code> <br><code><a href="layer_group_normalization.html">layer_group_normalization</a>()</code> <br><code><a href="layer_group_query_attention.html">layer_group_query_attention</a>()</code> <br><code><a href="layer_gru.html">layer_gru</a>()</code> <br><code><a href="layer_hashed_crossing.html">layer_hashed_crossing</a>()</code> <br><code><a href="layer_hashing.html">layer_hashing</a>()</code> <br><code><a href="layer_identity.html">layer_identity</a>()</code> <br><code><a href="layer_integer_lookup.html">layer_integer_lookup</a>()</code> <br><code><a href="layer_jax_model_wrapper.html">layer_jax_model_wrapper</a>()</code> <br><code><a href="layer_lambda.html">layer_lambda</a>()</code> <br><code><a href="layer_layer_normalization.html">layer_layer_normalization</a>()</code> <br><code><a href="layer_lstm.html">layer_lstm</a>()</code> <br><code><a href="layer_masking.html">layer_masking</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_max_pooling_1d.html">layer_max_pooling_1d</a>()</code> <br><code><a href="layer_max_pooling_2d.html">layer_max_pooling_2d</a>()</code> <br><code><a href="layer_max_pooling_3d.html">layer_max_pooling_3d</a>()</code> <br><code><a href="layer_maximum.html">layer_maximum</a>()</code> <br><code><a href="layer_mel_spectrogram.html">layer_mel_spectrogram</a>()</code> <br><code><a href="layer_minimum.html">layer_minimum</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_multi_head_attention.html">layer_multi_head_attention</a>()</code> <br><code><a href="layer_multiply.html">layer_multiply</a>()</code> <br><code><a href="layer_normalization.html">layer_normalization</a>()</code> <br><code><a href="layer_permute.html">layer_permute</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_brightness.html">layer_random_brightness</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_contrast.html">layer_random_contrast</a>()</code> <br><code><a href="layer_random_crop.html">layer_random_crop</a>()</code> <br><code><a href="layer_random_elastic_transform.html">layer_random_elastic_transform</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_flip.html">layer_random_flip</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_rotation.html">layer_random_rotation</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_random_translation.html">layer_random_translation</a>()</code> <br><code><a href="layer_random_zoom.html">layer_random_zoom</a>()</code> <br><code><a href="layer_repeat_vector.html">layer_repeat_vector</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_reshape.html">layer_reshape</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_rms_normalization.html">layer_rms_normalization</a>()</code> <br><code><a href="layer_rnn.html">layer_rnn</a>()</code> <br><code><a href="layer_separable_conv_1d.html">layer_separable_conv_1d</a>()</code> <br><code><a href="layer_separable_conv_2d.html">layer_separable_conv_2d</a>()</code> <br><code><a href="layer_simple_rnn.html">layer_simple_rnn</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br><code><a href="layer_spatial_dropout_1d.html">layer_spatial_dropout_1d</a>()</code> <br><code><a href="layer_spatial_dropout_2d.html">layer_spatial_dropout_2d</a>()</code> <br><code><a href="layer_spatial_dropout_3d.html">layer_spatial_dropout_3d</a>()</code> <br><code><a href="layer_spectral_normalization.html">layer_spectral_normalization</a>()</code> <br><code><a href="layer_stft_spectrogram.html">layer_stft_spectrogram</a>()</code> <br><code><a href="layer_string_lookup.html">layer_string_lookup</a>()</code> <br><code><a href="layer_subtract.html">layer_subtract</a>()</code> <br><code><a href="layer_text_vectorization.html">layer_text_vectorization</a>()</code> <br><code><a href="layer_tfsm.html">layer_tfsm</a>()</code> <br><code><a href="layer_time_distributed.html">layer_time_distributed</a>()</code> <br><code><a href="layer_torch_module_wrapper.html">layer_torch_module_wrapper</a>()</code> <br><code><a href="layer_unit_normalization.html">layer_unit_normalization</a>()</code> <br><code><a href="layer_upsampling_1d.html">layer_upsampling_1d</a>()</code> <br><code><a href="layer_upsampling_2d.html">layer_upsampling_2d</a>()</code> <br><code><a href="layer_upsampling_3d.html">layer_upsampling_3d</a>()</code> <br><code><a href="layer_zero_padding_1d.html">layer_zero_padding_1d</a>()</code> <br><code><a href="layer_zero_padding_2d.html">layer_zero_padding_2d</a>()</code> <br><code><a href="layer_zero_padding_3d.html">layer_zero_padding_3d</a>()</code> <br><code><a href="rnn_cell_gru.html">rnn_cell_gru</a>()</code> <br><code><a href="rnn_cell_lstm.html">rnn_cell_lstm</a>()</code> <br><code><a href="rnn_cell_simple.html">rnn_cell_simple</a>()</code> <br><code><a href="rnn_cells_stack.html">rnn_cells_stack</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -272,7 +272,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/layer_depthwise_conv_2d.html b/docs/dev/reference/layer_depthwise_conv_2d.html
index cb553e993..afbf14c7f 100644
--- a/docs/dev/reference/layer_depthwise_conv_2d.html
+++ b/docs/dev/reference/layer_depthwise_conv_2d.html
@@ -38,7 +38,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -259,7 +259,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
     <div class="dont-index">
 <ul><li><p><a href="https://keras.io/api/layers/convolution_layers/depthwise_convolution2d#depthwiseconv2d-class" class="external-link">https://keras.io/api/layers/convolution_layers/depthwise_convolution2d#depthwiseconv2d-class</a></p></li>
 </ul><p>Other convolutional layers: <br><code><a href="layer_conv_1d.html">layer_conv_1d</a>()</code> <br><code><a href="layer_conv_1d_transpose.html">layer_conv_1d_transpose</a>()</code> <br><code><a href="layer_conv_2d.html">layer_conv_2d</a>()</code> <br><code><a href="layer_conv_2d_transpose.html">layer_conv_2d_transpose</a>()</code> <br><code><a href="layer_conv_3d.html">layer_conv_3d</a>()</code> <br><code><a href="layer_conv_3d_transpose.html">layer_conv_3d_transpose</a>()</code> <br><code><a href="layer_depthwise_conv_1d.html">layer_depthwise_conv_1d</a>()</code> <br><code><a href="layer_separable_conv_1d.html">layer_separable_conv_1d</a>()</code> <br><code><a href="layer_separable_conv_2d.html">layer_separable_conv_2d</a>()</code> <br></p>
-<p>Other layers: <br><code><a href="Layer.html">Layer</a>()</code> <br><code><a href="layer_activation.html">layer_activation</a>()</code> <br><code><a href="layer_activation_elu.html">layer_activation_elu</a>()</code> <br><code><a href="layer_activation_leaky_relu.html">layer_activation_leaky_relu</a>()</code> <br><code><a href="layer_activation_parametric_relu.html">layer_activation_parametric_relu</a>()</code> <br><code><a href="layer_activation_relu.html">layer_activation_relu</a>()</code> <br><code><a href="layer_activation_softmax.html">layer_activation_softmax</a>()</code> <br><code><a href="layer_activity_regularization.html">layer_activity_regularization</a>()</code> <br><code><a href="layer_add.html">layer_add</a>()</code> <br><code><a href="layer_additive_attention.html">layer_additive_attention</a>()</code> <br><code><a href="layer_alpha_dropout.html">layer_alpha_dropout</a>()</code> <br><code><a href="layer_attention.html">layer_attention</a>()</code> <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_average.html">layer_average</a>()</code> <br><code><a href="layer_average_pooling_1d.html">layer_average_pooling_1d</a>()</code> <br><code><a href="layer_average_pooling_2d.html">layer_average_pooling_2d</a>()</code> <br><code><a href="layer_average_pooling_3d.html">layer_average_pooling_3d</a>()</code> <br><code><a href="layer_batch_normalization.html">layer_batch_normalization</a>()</code> <br><code><a href="layer_bidirectional.html">layer_bidirectional</a>()</code> <br><code><a href="layer_category_encoding.html">layer_category_encoding</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_concatenate.html">layer_concatenate</a>()</code> <br><code><a href="layer_conv_1d.html">layer_conv_1d</a>()</code> <br><code><a href="layer_conv_1d_transpose.html">layer_conv_1d_transpose</a>()</code> <br><code><a href="layer_conv_2d.html">layer_conv_2d</a>()</code> <br><code><a href="layer_conv_2d_transpose.html">layer_conv_2d_transpose</a>()</code> <br><code><a href="layer_conv_3d.html">layer_conv_3d</a>()</code> <br><code><a href="layer_conv_3d_transpose.html">layer_conv_3d_transpose</a>()</code> <br><code><a href="layer_conv_lstm_1d.html">layer_conv_lstm_1d</a>()</code> <br><code><a href="layer_conv_lstm_2d.html">layer_conv_lstm_2d</a>()</code> <br><code><a href="layer_conv_lstm_3d.html">layer_conv_lstm_3d</a>()</code> <br><code><a href="layer_cropping_1d.html">layer_cropping_1d</a>()</code> <br><code><a href="layer_cropping_2d.html">layer_cropping_2d</a>()</code> <br><code><a href="layer_cropping_3d.html">layer_cropping_3d</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_dense.html">layer_dense</a>()</code> <br><code><a href="layer_depthwise_conv_1d.html">layer_depthwise_conv_1d</a>()</code> <br><code><a href="layer_discretization.html">layer_discretization</a>()</code> <br><code><a href="layer_dot.html">layer_dot</a>()</code> <br><code><a href="layer_dropout.html">layer_dropout</a>()</code> <br><code><a href="layer_einsum_dense.html">layer_einsum_dense</a>()</code> <br><code><a href="layer_embedding.html">layer_embedding</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_feature_space.html">layer_feature_space</a>()</code> <br><code><a href="layer_flatten.html">layer_flatten</a>()</code> <br><code><a href="layer_flax_module_wrapper.html">layer_flax_module_wrapper</a>()</code> <br><code><a href="layer_gaussian_dropout.html">layer_gaussian_dropout</a>()</code> <br><code><a href="layer_gaussian_noise.html">layer_gaussian_noise</a>()</code> <br><code><a href="layer_global_average_pooling_1d.html">layer_global_average_pooling_1d</a>()</code> <br><code><a href="layer_global_average_pooling_2d.html">layer_global_average_pooling_2d</a>()</code> <br><code><a href="layer_global_average_pooling_3d.html">layer_global_average_pooling_3d</a>()</code> <br><code><a href="layer_global_max_pooling_1d.html">layer_global_max_pooling_1d</a>()</code> <br><code><a href="layer_global_max_pooling_2d.html">layer_global_max_pooling_2d</a>()</code> <br><code><a href="layer_global_max_pooling_3d.html">layer_global_max_pooling_3d</a>()</code> <br><code><a href="layer_group_normalization.html">layer_group_normalization</a>()</code> <br><code><a href="layer_group_query_attention.html">layer_group_query_attention</a>()</code> <br><code><a href="layer_gru.html">layer_gru</a>()</code> <br><code><a href="layer_hashed_crossing.html">layer_hashed_crossing</a>()</code> <br><code><a href="layer_hashing.html">layer_hashing</a>()</code> <br><code><a href="layer_identity.html">layer_identity</a>()</code> <br><code><a href="layer_integer_lookup.html">layer_integer_lookup</a>()</code> <br><code><a href="layer_jax_model_wrapper.html">layer_jax_model_wrapper</a>()</code> <br><code><a href="layer_lambda.html">layer_lambda</a>()</code> <br><code><a href="layer_layer_normalization.html">layer_layer_normalization</a>()</code> <br><code><a href="layer_lstm.html">layer_lstm</a>()</code> <br><code><a href="layer_masking.html">layer_masking</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_max_pooling_1d.html">layer_max_pooling_1d</a>()</code> <br><code><a href="layer_max_pooling_2d.html">layer_max_pooling_2d</a>()</code> <br><code><a href="layer_max_pooling_3d.html">layer_max_pooling_3d</a>()</code> <br><code><a href="layer_maximum.html">layer_maximum</a>()</code> <br><code><a href="layer_mel_spectrogram.html">layer_mel_spectrogram</a>()</code> <br><code><a href="layer_minimum.html">layer_minimum</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_multi_head_attention.html">layer_multi_head_attention</a>()</code> <br><code><a href="layer_multiply.html">layer_multiply</a>()</code> <br><code><a href="layer_normalization.html">layer_normalization</a>()</code> <br><code><a href="layer_permute.html">layer_permute</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_brightness.html">layer_random_brightness</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_contrast.html">layer_random_contrast</a>()</code> <br><code><a href="layer_random_crop.html">layer_random_crop</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_flip.html">layer_random_flip</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_rotation.html">layer_random_rotation</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_random_translation.html">layer_random_translation</a>()</code> <br><code><a href="layer_random_zoom.html">layer_random_zoom</a>()</code> <br><code><a href="layer_repeat_vector.html">layer_repeat_vector</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_reshape.html">layer_reshape</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_rms_normalization.html">layer_rms_normalization</a>()</code> <br><code><a href="layer_rnn.html">layer_rnn</a>()</code> <br><code><a href="layer_separable_conv_1d.html">layer_separable_conv_1d</a>()</code> <br><code><a href="layer_separable_conv_2d.html">layer_separable_conv_2d</a>()</code> <br><code><a href="layer_simple_rnn.html">layer_simple_rnn</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br><code><a href="layer_spatial_dropout_1d.html">layer_spatial_dropout_1d</a>()</code> <br><code><a href="layer_spatial_dropout_2d.html">layer_spatial_dropout_2d</a>()</code> <br><code><a href="layer_spatial_dropout_3d.html">layer_spatial_dropout_3d</a>()</code> <br><code><a href="layer_spectral_normalization.html">layer_spectral_normalization</a>()</code> <br><code><a href="layer_stft_spectrogram.html">layer_stft_spectrogram</a>()</code> <br><code><a href="layer_string_lookup.html">layer_string_lookup</a>()</code> <br><code><a href="layer_subtract.html">layer_subtract</a>()</code> <br><code><a href="layer_text_vectorization.html">layer_text_vectorization</a>()</code> <br><code><a href="layer_tfsm.html">layer_tfsm</a>()</code> <br><code><a href="layer_time_distributed.html">layer_time_distributed</a>()</code> <br><code><a href="layer_torch_module_wrapper.html">layer_torch_module_wrapper</a>()</code> <br><code><a href="layer_unit_normalization.html">layer_unit_normalization</a>()</code> <br><code><a href="layer_upsampling_1d.html">layer_upsampling_1d</a>()</code> <br><code><a href="layer_upsampling_2d.html">layer_upsampling_2d</a>()</code> <br><code><a href="layer_upsampling_3d.html">layer_upsampling_3d</a>()</code> <br><code><a href="layer_zero_padding_1d.html">layer_zero_padding_1d</a>()</code> <br><code><a href="layer_zero_padding_2d.html">layer_zero_padding_2d</a>()</code> <br><code><a href="layer_zero_padding_3d.html">layer_zero_padding_3d</a>()</code> <br><code><a href="rnn_cell_gru.html">rnn_cell_gru</a>()</code> <br><code><a href="rnn_cell_lstm.html">rnn_cell_lstm</a>()</code> <br><code><a href="rnn_cell_simple.html">rnn_cell_simple</a>()</code> <br><code><a href="rnn_cells_stack.html">rnn_cells_stack</a>()</code> <br></p></div>
+<p>Other layers: <br><code><a href="Layer.html">Layer</a>()</code> <br><code><a href="layer_activation.html">layer_activation</a>()</code> <br><code><a href="layer_activation_elu.html">layer_activation_elu</a>()</code> <br><code><a href="layer_activation_leaky_relu.html">layer_activation_leaky_relu</a>()</code> <br><code><a href="layer_activation_parametric_relu.html">layer_activation_parametric_relu</a>()</code> <br><code><a href="layer_activation_relu.html">layer_activation_relu</a>()</code> <br><code><a href="layer_activation_softmax.html">layer_activation_softmax</a>()</code> <br><code><a href="layer_activity_regularization.html">layer_activity_regularization</a>()</code> <br><code><a href="layer_add.html">layer_add</a>()</code> <br><code><a href="layer_additive_attention.html">layer_additive_attention</a>()</code> <br><code><a href="layer_alpha_dropout.html">layer_alpha_dropout</a>()</code> <br><code><a href="layer_attention.html">layer_attention</a>()</code> <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_average.html">layer_average</a>()</code> <br><code><a href="layer_average_pooling_1d.html">layer_average_pooling_1d</a>()</code> <br><code><a href="layer_average_pooling_2d.html">layer_average_pooling_2d</a>()</code> <br><code><a href="layer_average_pooling_3d.html">layer_average_pooling_3d</a>()</code> <br><code><a href="layer_batch_normalization.html">layer_batch_normalization</a>()</code> <br><code><a href="layer_bidirectional.html">layer_bidirectional</a>()</code> <br><code><a href="layer_category_encoding.html">layer_category_encoding</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_concatenate.html">layer_concatenate</a>()</code> <br><code><a href="layer_conv_1d.html">layer_conv_1d</a>()</code> <br><code><a href="layer_conv_1d_transpose.html">layer_conv_1d_transpose</a>()</code> <br><code><a href="layer_conv_2d.html">layer_conv_2d</a>()</code> <br><code><a href="layer_conv_2d_transpose.html">layer_conv_2d_transpose</a>()</code> <br><code><a href="layer_conv_3d.html">layer_conv_3d</a>()</code> <br><code><a href="layer_conv_3d_transpose.html">layer_conv_3d_transpose</a>()</code> <br><code><a href="layer_conv_lstm_1d.html">layer_conv_lstm_1d</a>()</code> <br><code><a href="layer_conv_lstm_2d.html">layer_conv_lstm_2d</a>()</code> <br><code><a href="layer_conv_lstm_3d.html">layer_conv_lstm_3d</a>()</code> <br><code><a href="layer_cropping_1d.html">layer_cropping_1d</a>()</code> <br><code><a href="layer_cropping_2d.html">layer_cropping_2d</a>()</code> <br><code><a href="layer_cropping_3d.html">layer_cropping_3d</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_dense.html">layer_dense</a>()</code> <br><code><a href="layer_depthwise_conv_1d.html">layer_depthwise_conv_1d</a>()</code> <br><code><a href="layer_discretization.html">layer_discretization</a>()</code> <br><code><a href="layer_dot.html">layer_dot</a>()</code> <br><code><a href="layer_dropout.html">layer_dropout</a>()</code> <br><code><a href="layer_einsum_dense.html">layer_einsum_dense</a>()</code> <br><code><a href="layer_embedding.html">layer_embedding</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_feature_space.html">layer_feature_space</a>()</code> <br><code><a href="layer_flatten.html">layer_flatten</a>()</code> <br><code><a href="layer_flax_module_wrapper.html">layer_flax_module_wrapper</a>()</code> <br><code><a href="layer_gaussian_dropout.html">layer_gaussian_dropout</a>()</code> <br><code><a href="layer_gaussian_noise.html">layer_gaussian_noise</a>()</code> <br><code><a href="layer_global_average_pooling_1d.html">layer_global_average_pooling_1d</a>()</code> <br><code><a href="layer_global_average_pooling_2d.html">layer_global_average_pooling_2d</a>()</code> <br><code><a href="layer_global_average_pooling_3d.html">layer_global_average_pooling_3d</a>()</code> <br><code><a href="layer_global_max_pooling_1d.html">layer_global_max_pooling_1d</a>()</code> <br><code><a href="layer_global_max_pooling_2d.html">layer_global_max_pooling_2d</a>()</code> <br><code><a href="layer_global_max_pooling_3d.html">layer_global_max_pooling_3d</a>()</code> <br><code><a href="layer_group_normalization.html">layer_group_normalization</a>()</code> <br><code><a href="layer_group_query_attention.html">layer_group_query_attention</a>()</code> <br><code><a href="layer_gru.html">layer_gru</a>()</code> <br><code><a href="layer_hashed_crossing.html">layer_hashed_crossing</a>()</code> <br><code><a href="layer_hashing.html">layer_hashing</a>()</code> <br><code><a href="layer_identity.html">layer_identity</a>()</code> <br><code><a href="layer_integer_lookup.html">layer_integer_lookup</a>()</code> <br><code><a href="layer_jax_model_wrapper.html">layer_jax_model_wrapper</a>()</code> <br><code><a href="layer_lambda.html">layer_lambda</a>()</code> <br><code><a href="layer_layer_normalization.html">layer_layer_normalization</a>()</code> <br><code><a href="layer_lstm.html">layer_lstm</a>()</code> <br><code><a href="layer_masking.html">layer_masking</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_max_pooling_1d.html">layer_max_pooling_1d</a>()</code> <br><code><a href="layer_max_pooling_2d.html">layer_max_pooling_2d</a>()</code> <br><code><a href="layer_max_pooling_3d.html">layer_max_pooling_3d</a>()</code> <br><code><a href="layer_maximum.html">layer_maximum</a>()</code> <br><code><a href="layer_mel_spectrogram.html">layer_mel_spectrogram</a>()</code> <br><code><a href="layer_minimum.html">layer_minimum</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_multi_head_attention.html">layer_multi_head_attention</a>()</code> <br><code><a href="layer_multiply.html">layer_multiply</a>()</code> <br><code><a href="layer_normalization.html">layer_normalization</a>()</code> <br><code><a href="layer_permute.html">layer_permute</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_brightness.html">layer_random_brightness</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_contrast.html">layer_random_contrast</a>()</code> <br><code><a href="layer_random_crop.html">layer_random_crop</a>()</code> <br><code><a href="layer_random_elastic_transform.html">layer_random_elastic_transform</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_flip.html">layer_random_flip</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_rotation.html">layer_random_rotation</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_random_translation.html">layer_random_translation</a>()</code> <br><code><a href="layer_random_zoom.html">layer_random_zoom</a>()</code> <br><code><a href="layer_repeat_vector.html">layer_repeat_vector</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_reshape.html">layer_reshape</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_rms_normalization.html">layer_rms_normalization</a>()</code> <br><code><a href="layer_rnn.html">layer_rnn</a>()</code> <br><code><a href="layer_separable_conv_1d.html">layer_separable_conv_1d</a>()</code> <br><code><a href="layer_separable_conv_2d.html">layer_separable_conv_2d</a>()</code> <br><code><a href="layer_simple_rnn.html">layer_simple_rnn</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br><code><a href="layer_spatial_dropout_1d.html">layer_spatial_dropout_1d</a>()</code> <br><code><a href="layer_spatial_dropout_2d.html">layer_spatial_dropout_2d</a>()</code> <br><code><a href="layer_spatial_dropout_3d.html">layer_spatial_dropout_3d</a>()</code> <br><code><a href="layer_spectral_normalization.html">layer_spectral_normalization</a>()</code> <br><code><a href="layer_stft_spectrogram.html">layer_stft_spectrogram</a>()</code> <br><code><a href="layer_string_lookup.html">layer_string_lookup</a>()</code> <br><code><a href="layer_subtract.html">layer_subtract</a>()</code> <br><code><a href="layer_text_vectorization.html">layer_text_vectorization</a>()</code> <br><code><a href="layer_tfsm.html">layer_tfsm</a>()</code> <br><code><a href="layer_time_distributed.html">layer_time_distributed</a>()</code> <br><code><a href="layer_torch_module_wrapper.html">layer_torch_module_wrapper</a>()</code> <br><code><a href="layer_unit_normalization.html">layer_unit_normalization</a>()</code> <br><code><a href="layer_upsampling_1d.html">layer_upsampling_1d</a>()</code> <br><code><a href="layer_upsampling_2d.html">layer_upsampling_2d</a>()</code> <br><code><a href="layer_upsampling_3d.html">layer_upsampling_3d</a>()</code> <br><code><a href="layer_zero_padding_1d.html">layer_zero_padding_1d</a>()</code> <br><code><a href="layer_zero_padding_2d.html">layer_zero_padding_2d</a>()</code> <br><code><a href="layer_zero_padding_3d.html">layer_zero_padding_3d</a>()</code> <br><code><a href="rnn_cell_gru.html">rnn_cell_gru</a>()</code> <br><code><a href="rnn_cell_lstm.html">rnn_cell_lstm</a>()</code> <br><code><a href="rnn_cell_simple.html">rnn_cell_simple</a>()</code> <br><code><a href="rnn_cells_stack.html">rnn_cells_stack</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -271,7 +271,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/layer_discretization.html b/docs/dev/reference/layer_discretization.html
index 33ada10f8..49174a0a8 100644
--- a/docs/dev/reference/layer_discretization.html
+++ b/docs/dev/reference/layer_discretization.html
@@ -16,7 +16,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -202,8 +202,8 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
     <div class="dont-index">
 <ul><li><p><a href="https://keras.io/api/layers/preprocessing_layers/numerical/discretization#discretization-class" class="external-link">https://keras.io/api/layers/preprocessing_layers/numerical/discretization#discretization-class</a></p></li>
 </ul><p>Other numerical features preprocessing layers: <br><code><a href="layer_normalization.html">layer_normalization</a>()</code> <br></p>
-<p>Other preprocessing layers: <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_category_encoding.html">layer_category_encoding</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_feature_space.html">layer_feature_space</a>()</code> <br><code><a href="layer_hashed_crossing.html">layer_hashed_crossing</a>()</code> <br><code><a href="layer_hashing.html">layer_hashing</a>()</code> <br><code><a href="layer_integer_lookup.html">layer_integer_lookup</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_mel_spectrogram.html">layer_mel_spectrogram</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_normalization.html">layer_normalization</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_brightness.html">layer_random_brightness</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_contrast.html">layer_random_contrast</a>()</code> <br><code><a href="layer_random_crop.html">layer_random_crop</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_flip.html">layer_random_flip</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_rotation.html">layer_random_rotation</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_random_translation.html">layer_random_translation</a>()</code> <br><code><a href="layer_random_zoom.html">layer_random_zoom</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br><code><a href="layer_stft_spectrogram.html">layer_stft_spectrogram</a>()</code> <br><code><a href="layer_string_lookup.html">layer_string_lookup</a>()</code> <br><code><a href="layer_text_vectorization.html">layer_text_vectorization</a>()</code> <br></p>
-<p>Other layers: <br><code><a href="Layer.html">Layer</a>()</code> <br><code><a href="layer_activation.html">layer_activation</a>()</code> <br><code><a href="layer_activation_elu.html">layer_activation_elu</a>()</code> <br><code><a href="layer_activation_leaky_relu.html">layer_activation_leaky_relu</a>()</code> <br><code><a href="layer_activation_parametric_relu.html">layer_activation_parametric_relu</a>()</code> <br><code><a href="layer_activation_relu.html">layer_activation_relu</a>()</code> <br><code><a href="layer_activation_softmax.html">layer_activation_softmax</a>()</code> <br><code><a href="layer_activity_regularization.html">layer_activity_regularization</a>()</code> <br><code><a href="layer_add.html">layer_add</a>()</code> <br><code><a href="layer_additive_attention.html">layer_additive_attention</a>()</code> <br><code><a href="layer_alpha_dropout.html">layer_alpha_dropout</a>()</code> <br><code><a href="layer_attention.html">layer_attention</a>()</code> <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_average.html">layer_average</a>()</code> <br><code><a href="layer_average_pooling_1d.html">layer_average_pooling_1d</a>()</code> <br><code><a href="layer_average_pooling_2d.html">layer_average_pooling_2d</a>()</code> <br><code><a href="layer_average_pooling_3d.html">layer_average_pooling_3d</a>()</code> <br><code><a href="layer_batch_normalization.html">layer_batch_normalization</a>()</code> <br><code><a href="layer_bidirectional.html">layer_bidirectional</a>()</code> <br><code><a href="layer_category_encoding.html">layer_category_encoding</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_concatenate.html">layer_concatenate</a>()</code> <br><code><a href="layer_conv_1d.html">layer_conv_1d</a>()</code> <br><code><a href="layer_conv_1d_transpose.html">layer_conv_1d_transpose</a>()</code> <br><code><a href="layer_conv_2d.html">layer_conv_2d</a>()</code> <br><code><a href="layer_conv_2d_transpose.html">layer_conv_2d_transpose</a>()</code> <br><code><a href="layer_conv_3d.html">layer_conv_3d</a>()</code> <br><code><a href="layer_conv_3d_transpose.html">layer_conv_3d_transpose</a>()</code> <br><code><a href="layer_conv_lstm_1d.html">layer_conv_lstm_1d</a>()</code> <br><code><a href="layer_conv_lstm_2d.html">layer_conv_lstm_2d</a>()</code> <br><code><a href="layer_conv_lstm_3d.html">layer_conv_lstm_3d</a>()</code> <br><code><a href="layer_cropping_1d.html">layer_cropping_1d</a>()</code> <br><code><a href="layer_cropping_2d.html">layer_cropping_2d</a>()</code> <br><code><a href="layer_cropping_3d.html">layer_cropping_3d</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_dense.html">layer_dense</a>()</code> <br><code><a href="layer_depthwise_conv_1d.html">layer_depthwise_conv_1d</a>()</code> <br><code><a href="layer_depthwise_conv_2d.html">layer_depthwise_conv_2d</a>()</code> <br><code><a href="layer_dot.html">layer_dot</a>()</code> <br><code><a href="layer_dropout.html">layer_dropout</a>()</code> <br><code><a href="layer_einsum_dense.html">layer_einsum_dense</a>()</code> <br><code><a href="layer_embedding.html">layer_embedding</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_feature_space.html">layer_feature_space</a>()</code> <br><code><a href="layer_flatten.html">layer_flatten</a>()</code> <br><code><a href="layer_flax_module_wrapper.html">layer_flax_module_wrapper</a>()</code> <br><code><a href="layer_gaussian_dropout.html">layer_gaussian_dropout</a>()</code> <br><code><a href="layer_gaussian_noise.html">layer_gaussian_noise</a>()</code> <br><code><a href="layer_global_average_pooling_1d.html">layer_global_average_pooling_1d</a>()</code> <br><code><a href="layer_global_average_pooling_2d.html">layer_global_average_pooling_2d</a>()</code> <br><code><a href="layer_global_average_pooling_3d.html">layer_global_average_pooling_3d</a>()</code> <br><code><a href="layer_global_max_pooling_1d.html">layer_global_max_pooling_1d</a>()</code> <br><code><a href="layer_global_max_pooling_2d.html">layer_global_max_pooling_2d</a>()</code> <br><code><a href="layer_global_max_pooling_3d.html">layer_global_max_pooling_3d</a>()</code> <br><code><a href="layer_group_normalization.html">layer_group_normalization</a>()</code> <br><code><a href="layer_group_query_attention.html">layer_group_query_attention</a>()</code> <br><code><a href="layer_gru.html">layer_gru</a>()</code> <br><code><a href="layer_hashed_crossing.html">layer_hashed_crossing</a>()</code> <br><code><a href="layer_hashing.html">layer_hashing</a>()</code> <br><code><a href="layer_identity.html">layer_identity</a>()</code> <br><code><a href="layer_integer_lookup.html">layer_integer_lookup</a>()</code> <br><code><a href="layer_jax_model_wrapper.html">layer_jax_model_wrapper</a>()</code> <br><code><a href="layer_lambda.html">layer_lambda</a>()</code> <br><code><a href="layer_layer_normalization.html">layer_layer_normalization</a>()</code> <br><code><a href="layer_lstm.html">layer_lstm</a>()</code> <br><code><a href="layer_masking.html">layer_masking</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_max_pooling_1d.html">layer_max_pooling_1d</a>()</code> <br><code><a href="layer_max_pooling_2d.html">layer_max_pooling_2d</a>()</code> <br><code><a href="layer_max_pooling_3d.html">layer_max_pooling_3d</a>()</code> <br><code><a href="layer_maximum.html">layer_maximum</a>()</code> <br><code><a href="layer_mel_spectrogram.html">layer_mel_spectrogram</a>()</code> <br><code><a href="layer_minimum.html">layer_minimum</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_multi_head_attention.html">layer_multi_head_attention</a>()</code> <br><code><a href="layer_multiply.html">layer_multiply</a>()</code> <br><code><a href="layer_normalization.html">layer_normalization</a>()</code> <br><code><a href="layer_permute.html">layer_permute</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_brightness.html">layer_random_brightness</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_contrast.html">layer_random_contrast</a>()</code> <br><code><a href="layer_random_crop.html">layer_random_crop</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_flip.html">layer_random_flip</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_rotation.html">layer_random_rotation</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_random_translation.html">layer_random_translation</a>()</code> <br><code><a href="layer_random_zoom.html">layer_random_zoom</a>()</code> <br><code><a href="layer_repeat_vector.html">layer_repeat_vector</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_reshape.html">layer_reshape</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_rms_normalization.html">layer_rms_normalization</a>()</code> <br><code><a href="layer_rnn.html">layer_rnn</a>()</code> <br><code><a href="layer_separable_conv_1d.html">layer_separable_conv_1d</a>()</code> <br><code><a href="layer_separable_conv_2d.html">layer_separable_conv_2d</a>()</code> <br><code><a href="layer_simple_rnn.html">layer_simple_rnn</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br><code><a href="layer_spatial_dropout_1d.html">layer_spatial_dropout_1d</a>()</code> <br><code><a href="layer_spatial_dropout_2d.html">layer_spatial_dropout_2d</a>()</code> <br><code><a href="layer_spatial_dropout_3d.html">layer_spatial_dropout_3d</a>()</code> <br><code><a href="layer_spectral_normalization.html">layer_spectral_normalization</a>()</code> <br><code><a href="layer_stft_spectrogram.html">layer_stft_spectrogram</a>()</code> <br><code><a href="layer_string_lookup.html">layer_string_lookup</a>()</code> <br><code><a href="layer_subtract.html">layer_subtract</a>()</code> <br><code><a href="layer_text_vectorization.html">layer_text_vectorization</a>()</code> <br><code><a href="layer_tfsm.html">layer_tfsm</a>()</code> <br><code><a href="layer_time_distributed.html">layer_time_distributed</a>()</code> <br><code><a href="layer_torch_module_wrapper.html">layer_torch_module_wrapper</a>()</code> <br><code><a href="layer_unit_normalization.html">layer_unit_normalization</a>()</code> <br><code><a href="layer_upsampling_1d.html">layer_upsampling_1d</a>()</code> <br><code><a href="layer_upsampling_2d.html">layer_upsampling_2d</a>()</code> <br><code><a href="layer_upsampling_3d.html">layer_upsampling_3d</a>()</code> <br><code><a href="layer_zero_padding_1d.html">layer_zero_padding_1d</a>()</code> <br><code><a href="layer_zero_padding_2d.html">layer_zero_padding_2d</a>()</code> <br><code><a href="layer_zero_padding_3d.html">layer_zero_padding_3d</a>()</code> <br><code><a href="rnn_cell_gru.html">rnn_cell_gru</a>()</code> <br><code><a href="rnn_cell_lstm.html">rnn_cell_lstm</a>()</code> <br><code><a href="rnn_cell_simple.html">rnn_cell_simple</a>()</code> <br><code><a href="rnn_cells_stack.html">rnn_cells_stack</a>()</code> <br></p></div>
+<p>Other preprocessing layers: <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_category_encoding.html">layer_category_encoding</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_feature_space.html">layer_feature_space</a>()</code> <br><code><a href="layer_hashed_crossing.html">layer_hashed_crossing</a>()</code> <br><code><a href="layer_hashing.html">layer_hashing</a>()</code> <br><code><a href="layer_integer_lookup.html">layer_integer_lookup</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_mel_spectrogram.html">layer_mel_spectrogram</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_normalization.html">layer_normalization</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_brightness.html">layer_random_brightness</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_contrast.html">layer_random_contrast</a>()</code> <br><code><a href="layer_random_crop.html">layer_random_crop</a>()</code> <br><code><a href="layer_random_elastic_transform.html">layer_random_elastic_transform</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_flip.html">layer_random_flip</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_rotation.html">layer_random_rotation</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_random_translation.html">layer_random_translation</a>()</code> <br><code><a href="layer_random_zoom.html">layer_random_zoom</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br><code><a href="layer_stft_spectrogram.html">layer_stft_spectrogram</a>()</code> <br><code><a href="layer_string_lookup.html">layer_string_lookup</a>()</code> <br><code><a href="layer_text_vectorization.html">layer_text_vectorization</a>()</code> <br></p>
+<p>Other layers: <br><code><a href="Layer.html">Layer</a>()</code> <br><code><a href="layer_activation.html">layer_activation</a>()</code> <br><code><a href="layer_activation_elu.html">layer_activation_elu</a>()</code> <br><code><a href="layer_activation_leaky_relu.html">layer_activation_leaky_relu</a>()</code> <br><code><a href="layer_activation_parametric_relu.html">layer_activation_parametric_relu</a>()</code> <br><code><a href="layer_activation_relu.html">layer_activation_relu</a>()</code> <br><code><a href="layer_activation_softmax.html">layer_activation_softmax</a>()</code> <br><code><a href="layer_activity_regularization.html">layer_activity_regularization</a>()</code> <br><code><a href="layer_add.html">layer_add</a>()</code> <br><code><a href="layer_additive_attention.html">layer_additive_attention</a>()</code> <br><code><a href="layer_alpha_dropout.html">layer_alpha_dropout</a>()</code> <br><code><a href="layer_attention.html">layer_attention</a>()</code> <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_average.html">layer_average</a>()</code> <br><code><a href="layer_average_pooling_1d.html">layer_average_pooling_1d</a>()</code> <br><code><a href="layer_average_pooling_2d.html">layer_average_pooling_2d</a>()</code> <br><code><a href="layer_average_pooling_3d.html">layer_average_pooling_3d</a>()</code> <br><code><a href="layer_batch_normalization.html">layer_batch_normalization</a>()</code> <br><code><a href="layer_bidirectional.html">layer_bidirectional</a>()</code> <br><code><a href="layer_category_encoding.html">layer_category_encoding</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_concatenate.html">layer_concatenate</a>()</code> <br><code><a href="layer_conv_1d.html">layer_conv_1d</a>()</code> <br><code><a href="layer_conv_1d_transpose.html">layer_conv_1d_transpose</a>()</code> <br><code><a href="layer_conv_2d.html">layer_conv_2d</a>()</code> <br><code><a href="layer_conv_2d_transpose.html">layer_conv_2d_transpose</a>()</code> <br><code><a href="layer_conv_3d.html">layer_conv_3d</a>()</code> <br><code><a href="layer_conv_3d_transpose.html">layer_conv_3d_transpose</a>()</code> <br><code><a href="layer_conv_lstm_1d.html">layer_conv_lstm_1d</a>()</code> <br><code><a href="layer_conv_lstm_2d.html">layer_conv_lstm_2d</a>()</code> <br><code><a href="layer_conv_lstm_3d.html">layer_conv_lstm_3d</a>()</code> <br><code><a href="layer_cropping_1d.html">layer_cropping_1d</a>()</code> <br><code><a href="layer_cropping_2d.html">layer_cropping_2d</a>()</code> <br><code><a href="layer_cropping_3d.html">layer_cropping_3d</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_dense.html">layer_dense</a>()</code> <br><code><a href="layer_depthwise_conv_1d.html">layer_depthwise_conv_1d</a>()</code> <br><code><a href="layer_depthwise_conv_2d.html">layer_depthwise_conv_2d</a>()</code> <br><code><a href="layer_dot.html">layer_dot</a>()</code> <br><code><a href="layer_dropout.html">layer_dropout</a>()</code> <br><code><a href="layer_einsum_dense.html">layer_einsum_dense</a>()</code> <br><code><a href="layer_embedding.html">layer_embedding</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_feature_space.html">layer_feature_space</a>()</code> <br><code><a href="layer_flatten.html">layer_flatten</a>()</code> <br><code><a href="layer_flax_module_wrapper.html">layer_flax_module_wrapper</a>()</code> <br><code><a href="layer_gaussian_dropout.html">layer_gaussian_dropout</a>()</code> <br><code><a href="layer_gaussian_noise.html">layer_gaussian_noise</a>()</code> <br><code><a href="layer_global_average_pooling_1d.html">layer_global_average_pooling_1d</a>()</code> <br><code><a href="layer_global_average_pooling_2d.html">layer_global_average_pooling_2d</a>()</code> <br><code><a href="layer_global_average_pooling_3d.html">layer_global_average_pooling_3d</a>()</code> <br><code><a href="layer_global_max_pooling_1d.html">layer_global_max_pooling_1d</a>()</code> <br><code><a href="layer_global_max_pooling_2d.html">layer_global_max_pooling_2d</a>()</code> <br><code><a href="layer_global_max_pooling_3d.html">layer_global_max_pooling_3d</a>()</code> <br><code><a href="layer_group_normalization.html">layer_group_normalization</a>()</code> <br><code><a href="layer_group_query_attention.html">layer_group_query_attention</a>()</code> <br><code><a href="layer_gru.html">layer_gru</a>()</code> <br><code><a href="layer_hashed_crossing.html">layer_hashed_crossing</a>()</code> <br><code><a href="layer_hashing.html">layer_hashing</a>()</code> <br><code><a href="layer_identity.html">layer_identity</a>()</code> <br><code><a href="layer_integer_lookup.html">layer_integer_lookup</a>()</code> <br><code><a href="layer_jax_model_wrapper.html">layer_jax_model_wrapper</a>()</code> <br><code><a href="layer_lambda.html">layer_lambda</a>()</code> <br><code><a href="layer_layer_normalization.html">layer_layer_normalization</a>()</code> <br><code><a href="layer_lstm.html">layer_lstm</a>()</code> <br><code><a href="layer_masking.html">layer_masking</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_max_pooling_1d.html">layer_max_pooling_1d</a>()</code> <br><code><a href="layer_max_pooling_2d.html">layer_max_pooling_2d</a>()</code> <br><code><a href="layer_max_pooling_3d.html">layer_max_pooling_3d</a>()</code> <br><code><a href="layer_maximum.html">layer_maximum</a>()</code> <br><code><a href="layer_mel_spectrogram.html">layer_mel_spectrogram</a>()</code> <br><code><a href="layer_minimum.html">layer_minimum</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_multi_head_attention.html">layer_multi_head_attention</a>()</code> <br><code><a href="layer_multiply.html">layer_multiply</a>()</code> <br><code><a href="layer_normalization.html">layer_normalization</a>()</code> <br><code><a href="layer_permute.html">layer_permute</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_brightness.html">layer_random_brightness</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_contrast.html">layer_random_contrast</a>()</code> <br><code><a href="layer_random_crop.html">layer_random_crop</a>()</code> <br><code><a href="layer_random_elastic_transform.html">layer_random_elastic_transform</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_flip.html">layer_random_flip</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_rotation.html">layer_random_rotation</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_random_translation.html">layer_random_translation</a>()</code> <br><code><a href="layer_random_zoom.html">layer_random_zoom</a>()</code> <br><code><a href="layer_repeat_vector.html">layer_repeat_vector</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_reshape.html">layer_reshape</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_rms_normalization.html">layer_rms_normalization</a>()</code> <br><code><a href="layer_rnn.html">layer_rnn</a>()</code> <br><code><a href="layer_separable_conv_1d.html">layer_separable_conv_1d</a>()</code> <br><code><a href="layer_separable_conv_2d.html">layer_separable_conv_2d</a>()</code> <br><code><a href="layer_simple_rnn.html">layer_simple_rnn</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br><code><a href="layer_spatial_dropout_1d.html">layer_spatial_dropout_1d</a>()</code> <br><code><a href="layer_spatial_dropout_2d.html">layer_spatial_dropout_2d</a>()</code> <br><code><a href="layer_spatial_dropout_3d.html">layer_spatial_dropout_3d</a>()</code> <br><code><a href="layer_spectral_normalization.html">layer_spectral_normalization</a>()</code> <br><code><a href="layer_stft_spectrogram.html">layer_stft_spectrogram</a>()</code> <br><code><a href="layer_string_lookup.html">layer_string_lookup</a>()</code> <br><code><a href="layer_subtract.html">layer_subtract</a>()</code> <br><code><a href="layer_text_vectorization.html">layer_text_vectorization</a>()</code> <br><code><a href="layer_tfsm.html">layer_tfsm</a>()</code> <br><code><a href="layer_time_distributed.html">layer_time_distributed</a>()</code> <br><code><a href="layer_torch_module_wrapper.html">layer_torch_module_wrapper</a>()</code> <br><code><a href="layer_unit_normalization.html">layer_unit_normalization</a>()</code> <br><code><a href="layer_upsampling_1d.html">layer_upsampling_1d</a>()</code> <br><code><a href="layer_upsampling_2d.html">layer_upsampling_2d</a>()</code> <br><code><a href="layer_upsampling_3d.html">layer_upsampling_3d</a>()</code> <br><code><a href="layer_zero_padding_1d.html">layer_zero_padding_1d</a>()</code> <br><code><a href="layer_zero_padding_2d.html">layer_zero_padding_2d</a>()</code> <br><code><a href="layer_zero_padding_3d.html">layer_zero_padding_3d</a>()</code> <br><code><a href="rnn_cell_gru.html">rnn_cell_gru</a>()</code> <br><code><a href="rnn_cell_lstm.html">rnn_cell_lstm</a>()</code> <br><code><a href="rnn_cell_simple.html">rnn_cell_simple</a>()</code> <br><code><a href="rnn_cells_stack.html">rnn_cells_stack</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -215,7 +215,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/layer_dot.html b/docs/dev/reference/layer_dot.html
index 7d1763cc1..9e3a6b8db 100644
--- a/docs/dev/reference/layer_dot.html
+++ b/docs/dev/reference/layer_dot.html
@@ -24,7 +24,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -145,7 +145,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
     <div class="dont-index">
 <ul><li><p><a href="https://keras.io/api/layers/merging_layers/dot#dot-class" class="external-link">https://keras.io/api/layers/merging_layers/dot#dot-class</a></p></li>
 </ul><p>Other merging layers: <br><code><a href="layer_add.html">layer_add</a>()</code> <br><code><a href="layer_average.html">layer_average</a>()</code> <br><code><a href="layer_concatenate.html">layer_concatenate</a>()</code> <br><code><a href="layer_maximum.html">layer_maximum</a>()</code> <br><code><a href="layer_minimum.html">layer_minimum</a>()</code> <br><code><a href="layer_multiply.html">layer_multiply</a>()</code> <br><code><a href="layer_subtract.html">layer_subtract</a>()</code> <br></p>
-<p>Other layers: <br><code><a href="Layer.html">Layer</a>()</code> <br><code><a href="layer_activation.html">layer_activation</a>()</code> <br><code><a href="layer_activation_elu.html">layer_activation_elu</a>()</code> <br><code><a href="layer_activation_leaky_relu.html">layer_activation_leaky_relu</a>()</code> <br><code><a href="layer_activation_parametric_relu.html">layer_activation_parametric_relu</a>()</code> <br><code><a href="layer_activation_relu.html">layer_activation_relu</a>()</code> <br><code><a href="layer_activation_softmax.html">layer_activation_softmax</a>()</code> <br><code><a href="layer_activity_regularization.html">layer_activity_regularization</a>()</code> <br><code><a href="layer_add.html">layer_add</a>()</code> <br><code><a href="layer_additive_attention.html">layer_additive_attention</a>()</code> <br><code><a href="layer_alpha_dropout.html">layer_alpha_dropout</a>()</code> <br><code><a href="layer_attention.html">layer_attention</a>()</code> <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_average.html">layer_average</a>()</code> <br><code><a href="layer_average_pooling_1d.html">layer_average_pooling_1d</a>()</code> <br><code><a href="layer_average_pooling_2d.html">layer_average_pooling_2d</a>()</code> <br><code><a href="layer_average_pooling_3d.html">layer_average_pooling_3d</a>()</code> <br><code><a href="layer_batch_normalization.html">layer_batch_normalization</a>()</code> <br><code><a href="layer_bidirectional.html">layer_bidirectional</a>()</code> <br><code><a href="layer_category_encoding.html">layer_category_encoding</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_concatenate.html">layer_concatenate</a>()</code> <br><code><a href="layer_conv_1d.html">layer_conv_1d</a>()</code> <br><code><a href="layer_conv_1d_transpose.html">layer_conv_1d_transpose</a>()</code> <br><code><a href="layer_conv_2d.html">layer_conv_2d</a>()</code> <br><code><a href="layer_conv_2d_transpose.html">layer_conv_2d_transpose</a>()</code> <br><code><a href="layer_conv_3d.html">layer_conv_3d</a>()</code> <br><code><a href="layer_conv_3d_transpose.html">layer_conv_3d_transpose</a>()</code> <br><code><a href="layer_conv_lstm_1d.html">layer_conv_lstm_1d</a>()</code> <br><code><a href="layer_conv_lstm_2d.html">layer_conv_lstm_2d</a>()</code> <br><code><a href="layer_conv_lstm_3d.html">layer_conv_lstm_3d</a>()</code> <br><code><a href="layer_cropping_1d.html">layer_cropping_1d</a>()</code> <br><code><a href="layer_cropping_2d.html">layer_cropping_2d</a>()</code> <br><code><a href="layer_cropping_3d.html">layer_cropping_3d</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_dense.html">layer_dense</a>()</code> <br><code><a href="layer_depthwise_conv_1d.html">layer_depthwise_conv_1d</a>()</code> <br><code><a href="layer_depthwise_conv_2d.html">layer_depthwise_conv_2d</a>()</code> <br><code><a href="layer_discretization.html">layer_discretization</a>()</code> <br><code><a href="layer_dropout.html">layer_dropout</a>()</code> <br><code><a href="layer_einsum_dense.html">layer_einsum_dense</a>()</code> <br><code><a href="layer_embedding.html">layer_embedding</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_feature_space.html">layer_feature_space</a>()</code> <br><code><a href="layer_flatten.html">layer_flatten</a>()</code> <br><code><a href="layer_flax_module_wrapper.html">layer_flax_module_wrapper</a>()</code> <br><code><a href="layer_gaussian_dropout.html">layer_gaussian_dropout</a>()</code> <br><code><a href="layer_gaussian_noise.html">layer_gaussian_noise</a>()</code> <br><code><a href="layer_global_average_pooling_1d.html">layer_global_average_pooling_1d</a>()</code> <br><code><a href="layer_global_average_pooling_2d.html">layer_global_average_pooling_2d</a>()</code> <br><code><a href="layer_global_average_pooling_3d.html">layer_global_average_pooling_3d</a>()</code> <br><code><a href="layer_global_max_pooling_1d.html">layer_global_max_pooling_1d</a>()</code> <br><code><a href="layer_global_max_pooling_2d.html">layer_global_max_pooling_2d</a>()</code> <br><code><a href="layer_global_max_pooling_3d.html">layer_global_max_pooling_3d</a>()</code> <br><code><a href="layer_group_normalization.html">layer_group_normalization</a>()</code> <br><code><a href="layer_group_query_attention.html">layer_group_query_attention</a>()</code> <br><code><a href="layer_gru.html">layer_gru</a>()</code> <br><code><a href="layer_hashed_crossing.html">layer_hashed_crossing</a>()</code> <br><code><a href="layer_hashing.html">layer_hashing</a>()</code> <br><code><a href="layer_identity.html">layer_identity</a>()</code> <br><code><a href="layer_integer_lookup.html">layer_integer_lookup</a>()</code> <br><code><a href="layer_jax_model_wrapper.html">layer_jax_model_wrapper</a>()</code> <br><code><a href="layer_lambda.html">layer_lambda</a>()</code> <br><code><a href="layer_layer_normalization.html">layer_layer_normalization</a>()</code> <br><code><a href="layer_lstm.html">layer_lstm</a>()</code> <br><code><a href="layer_masking.html">layer_masking</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_max_pooling_1d.html">layer_max_pooling_1d</a>()</code> <br><code><a href="layer_max_pooling_2d.html">layer_max_pooling_2d</a>()</code> <br><code><a href="layer_max_pooling_3d.html">layer_max_pooling_3d</a>()</code> <br><code><a href="layer_maximum.html">layer_maximum</a>()</code> <br><code><a href="layer_mel_spectrogram.html">layer_mel_spectrogram</a>()</code> <br><code><a href="layer_minimum.html">layer_minimum</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_multi_head_attention.html">layer_multi_head_attention</a>()</code> <br><code><a href="layer_multiply.html">layer_multiply</a>()</code> <br><code><a href="layer_normalization.html">layer_normalization</a>()</code> <br><code><a href="layer_permute.html">layer_permute</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_brightness.html">layer_random_brightness</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_contrast.html">layer_random_contrast</a>()</code> <br><code><a href="layer_random_crop.html">layer_random_crop</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_flip.html">layer_random_flip</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_rotation.html">layer_random_rotation</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_random_translation.html">layer_random_translation</a>()</code> <br><code><a href="layer_random_zoom.html">layer_random_zoom</a>()</code> <br><code><a href="layer_repeat_vector.html">layer_repeat_vector</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_reshape.html">layer_reshape</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_rms_normalization.html">layer_rms_normalization</a>()</code> <br><code><a href="layer_rnn.html">layer_rnn</a>()</code> <br><code><a href="layer_separable_conv_1d.html">layer_separable_conv_1d</a>()</code> <br><code><a href="layer_separable_conv_2d.html">layer_separable_conv_2d</a>()</code> <br><code><a href="layer_simple_rnn.html">layer_simple_rnn</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br><code><a href="layer_spatial_dropout_1d.html">layer_spatial_dropout_1d</a>()</code> <br><code><a href="layer_spatial_dropout_2d.html">layer_spatial_dropout_2d</a>()</code> <br><code><a href="layer_spatial_dropout_3d.html">layer_spatial_dropout_3d</a>()</code> <br><code><a href="layer_spectral_normalization.html">layer_spectral_normalization</a>()</code> <br><code><a href="layer_stft_spectrogram.html">layer_stft_spectrogram</a>()</code> <br><code><a href="layer_string_lookup.html">layer_string_lookup</a>()</code> <br><code><a href="layer_subtract.html">layer_subtract</a>()</code> <br><code><a href="layer_text_vectorization.html">layer_text_vectorization</a>()</code> <br><code><a href="layer_tfsm.html">layer_tfsm</a>()</code> <br><code><a href="layer_time_distributed.html">layer_time_distributed</a>()</code> <br><code><a href="layer_torch_module_wrapper.html">layer_torch_module_wrapper</a>()</code> <br><code><a href="layer_unit_normalization.html">layer_unit_normalization</a>()</code> <br><code><a href="layer_upsampling_1d.html">layer_upsampling_1d</a>()</code> <br><code><a href="layer_upsampling_2d.html">layer_upsampling_2d</a>()</code> <br><code><a href="layer_upsampling_3d.html">layer_upsampling_3d</a>()</code> <br><code><a href="layer_zero_padding_1d.html">layer_zero_padding_1d</a>()</code> <br><code><a href="layer_zero_padding_2d.html">layer_zero_padding_2d</a>()</code> <br><code><a href="layer_zero_padding_3d.html">layer_zero_padding_3d</a>()</code> <br><code><a href="rnn_cell_gru.html">rnn_cell_gru</a>()</code> <br><code><a href="rnn_cell_lstm.html">rnn_cell_lstm</a>()</code> <br><code><a href="rnn_cell_simple.html">rnn_cell_simple</a>()</code> <br><code><a href="rnn_cells_stack.html">rnn_cells_stack</a>()</code> <br></p></div>
+<p>Other layers: <br><code><a href="Layer.html">Layer</a>()</code> <br><code><a href="layer_activation.html">layer_activation</a>()</code> <br><code><a href="layer_activation_elu.html">layer_activation_elu</a>()</code> <br><code><a href="layer_activation_leaky_relu.html">layer_activation_leaky_relu</a>()</code> <br><code><a href="layer_activation_parametric_relu.html">layer_activation_parametric_relu</a>()</code> <br><code><a href="layer_activation_relu.html">layer_activation_relu</a>()</code> <br><code><a href="layer_activation_softmax.html">layer_activation_softmax</a>()</code> <br><code><a href="layer_activity_regularization.html">layer_activity_regularization</a>()</code> <br><code><a href="layer_add.html">layer_add</a>()</code> <br><code><a href="layer_additive_attention.html">layer_additive_attention</a>()</code> <br><code><a href="layer_alpha_dropout.html">layer_alpha_dropout</a>()</code> <br><code><a href="layer_attention.html">layer_attention</a>()</code> <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_average.html">layer_average</a>()</code> <br><code><a href="layer_average_pooling_1d.html">layer_average_pooling_1d</a>()</code> <br><code><a href="layer_average_pooling_2d.html">layer_average_pooling_2d</a>()</code> <br><code><a href="layer_average_pooling_3d.html">layer_average_pooling_3d</a>()</code> <br><code><a href="layer_batch_normalization.html">layer_batch_normalization</a>()</code> <br><code><a href="layer_bidirectional.html">layer_bidirectional</a>()</code> <br><code><a href="layer_category_encoding.html">layer_category_encoding</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_concatenate.html">layer_concatenate</a>()</code> <br><code><a href="layer_conv_1d.html">layer_conv_1d</a>()</code> <br><code><a href="layer_conv_1d_transpose.html">layer_conv_1d_transpose</a>()</code> <br><code><a href="layer_conv_2d.html">layer_conv_2d</a>()</code> <br><code><a href="layer_conv_2d_transpose.html">layer_conv_2d_transpose</a>()</code> <br><code><a href="layer_conv_3d.html">layer_conv_3d</a>()</code> <br><code><a href="layer_conv_3d_transpose.html">layer_conv_3d_transpose</a>()</code> <br><code><a href="layer_conv_lstm_1d.html">layer_conv_lstm_1d</a>()</code> <br><code><a href="layer_conv_lstm_2d.html">layer_conv_lstm_2d</a>()</code> <br><code><a href="layer_conv_lstm_3d.html">layer_conv_lstm_3d</a>()</code> <br><code><a href="layer_cropping_1d.html">layer_cropping_1d</a>()</code> <br><code><a href="layer_cropping_2d.html">layer_cropping_2d</a>()</code> <br><code><a href="layer_cropping_3d.html">layer_cropping_3d</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_dense.html">layer_dense</a>()</code> <br><code><a href="layer_depthwise_conv_1d.html">layer_depthwise_conv_1d</a>()</code> <br><code><a href="layer_depthwise_conv_2d.html">layer_depthwise_conv_2d</a>()</code> <br><code><a href="layer_discretization.html">layer_discretization</a>()</code> <br><code><a href="layer_dropout.html">layer_dropout</a>()</code> <br><code><a href="layer_einsum_dense.html">layer_einsum_dense</a>()</code> <br><code><a href="layer_embedding.html">layer_embedding</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_feature_space.html">layer_feature_space</a>()</code> <br><code><a href="layer_flatten.html">layer_flatten</a>()</code> <br><code><a href="layer_flax_module_wrapper.html">layer_flax_module_wrapper</a>()</code> <br><code><a href="layer_gaussian_dropout.html">layer_gaussian_dropout</a>()</code> <br><code><a href="layer_gaussian_noise.html">layer_gaussian_noise</a>()</code> <br><code><a href="layer_global_average_pooling_1d.html">layer_global_average_pooling_1d</a>()</code> <br><code><a href="layer_global_average_pooling_2d.html">layer_global_average_pooling_2d</a>()</code> <br><code><a href="layer_global_average_pooling_3d.html">layer_global_average_pooling_3d</a>()</code> <br><code><a href="layer_global_max_pooling_1d.html">layer_global_max_pooling_1d</a>()</code> <br><code><a href="layer_global_max_pooling_2d.html">layer_global_max_pooling_2d</a>()</code> <br><code><a href="layer_global_max_pooling_3d.html">layer_global_max_pooling_3d</a>()</code> <br><code><a href="layer_group_normalization.html">layer_group_normalization</a>()</code> <br><code><a href="layer_group_query_attention.html">layer_group_query_attention</a>()</code> <br><code><a href="layer_gru.html">layer_gru</a>()</code> <br><code><a href="layer_hashed_crossing.html">layer_hashed_crossing</a>()</code> <br><code><a href="layer_hashing.html">layer_hashing</a>()</code> <br><code><a href="layer_identity.html">layer_identity</a>()</code> <br><code><a href="layer_integer_lookup.html">layer_integer_lookup</a>()</code> <br><code><a href="layer_jax_model_wrapper.html">layer_jax_model_wrapper</a>()</code> <br><code><a href="layer_lambda.html">layer_lambda</a>()</code> <br><code><a href="layer_layer_normalization.html">layer_layer_normalization</a>()</code> <br><code><a href="layer_lstm.html">layer_lstm</a>()</code> <br><code><a href="layer_masking.html">layer_masking</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_max_pooling_1d.html">layer_max_pooling_1d</a>()</code> <br><code><a href="layer_max_pooling_2d.html">layer_max_pooling_2d</a>()</code> <br><code><a href="layer_max_pooling_3d.html">layer_max_pooling_3d</a>()</code> <br><code><a href="layer_maximum.html">layer_maximum</a>()</code> <br><code><a href="layer_mel_spectrogram.html">layer_mel_spectrogram</a>()</code> <br><code><a href="layer_minimum.html">layer_minimum</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_multi_head_attention.html">layer_multi_head_attention</a>()</code> <br><code><a href="layer_multiply.html">layer_multiply</a>()</code> <br><code><a href="layer_normalization.html">layer_normalization</a>()</code> <br><code><a href="layer_permute.html">layer_permute</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_brightness.html">layer_random_brightness</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_contrast.html">layer_random_contrast</a>()</code> <br><code><a href="layer_random_crop.html">layer_random_crop</a>()</code> <br><code><a href="layer_random_elastic_transform.html">layer_random_elastic_transform</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_flip.html">layer_random_flip</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_rotation.html">layer_random_rotation</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_random_translation.html">layer_random_translation</a>()</code> <br><code><a href="layer_random_zoom.html">layer_random_zoom</a>()</code> <br><code><a href="layer_repeat_vector.html">layer_repeat_vector</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_reshape.html">layer_reshape</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_rms_normalization.html">layer_rms_normalization</a>()</code> <br><code><a href="layer_rnn.html">layer_rnn</a>()</code> <br><code><a href="layer_separable_conv_1d.html">layer_separable_conv_1d</a>()</code> <br><code><a href="layer_separable_conv_2d.html">layer_separable_conv_2d</a>()</code> <br><code><a href="layer_simple_rnn.html">layer_simple_rnn</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br><code><a href="layer_spatial_dropout_1d.html">layer_spatial_dropout_1d</a>()</code> <br><code><a href="layer_spatial_dropout_2d.html">layer_spatial_dropout_2d</a>()</code> <br><code><a href="layer_spatial_dropout_3d.html">layer_spatial_dropout_3d</a>()</code> <br><code><a href="layer_spectral_normalization.html">layer_spectral_normalization</a>()</code> <br><code><a href="layer_stft_spectrogram.html">layer_stft_spectrogram</a>()</code> <br><code><a href="layer_string_lookup.html">layer_string_lookup</a>()</code> <br><code><a href="layer_subtract.html">layer_subtract</a>()</code> <br><code><a href="layer_text_vectorization.html">layer_text_vectorization</a>()</code> <br><code><a href="layer_tfsm.html">layer_tfsm</a>()</code> <br><code><a href="layer_time_distributed.html">layer_time_distributed</a>()</code> <br><code><a href="layer_torch_module_wrapper.html">layer_torch_module_wrapper</a>()</code> <br><code><a href="layer_unit_normalization.html">layer_unit_normalization</a>()</code> <br><code><a href="layer_upsampling_1d.html">layer_upsampling_1d</a>()</code> <br><code><a href="layer_upsampling_2d.html">layer_upsampling_2d</a>()</code> <br><code><a href="layer_upsampling_3d.html">layer_upsampling_3d</a>()</code> <br><code><a href="layer_zero_padding_1d.html">layer_zero_padding_1d</a>()</code> <br><code><a href="layer_zero_padding_2d.html">layer_zero_padding_2d</a>()</code> <br><code><a href="layer_zero_padding_3d.html">layer_zero_padding_3d</a>()</code> <br><code><a href="rnn_cell_gru.html">rnn_cell_gru</a>()</code> <br><code><a href="rnn_cell_lstm.html">rnn_cell_lstm</a>()</code> <br><code><a href="rnn_cell_simple.html">rnn_cell_simple</a>()</code> <br><code><a href="rnn_cells_stack.html">rnn_cells_stack</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -157,7 +157,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/layer_dropout.html b/docs/dev/reference/layer_dropout.html
index 0a8bd61dc..80b7d64dd 100644
--- a/docs/dev/reference/layer_dropout.html
+++ b/docs/dev/reference/layer_dropout.html
@@ -30,7 +30,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -145,7 +145,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
     <div class="dont-index">
 <ul><li><p><a href="https://keras.io/api/layers/regularization_layers/dropout#dropout-class" class="external-link">https://keras.io/api/layers/regularization_layers/dropout#dropout-class</a></p></li>
 </ul><p>Other regularization layers: <br><code><a href="layer_activity_regularization.html">layer_activity_regularization</a>()</code> <br><code><a href="layer_alpha_dropout.html">layer_alpha_dropout</a>()</code> <br><code><a href="layer_gaussian_dropout.html">layer_gaussian_dropout</a>()</code> <br><code><a href="layer_gaussian_noise.html">layer_gaussian_noise</a>()</code> <br><code><a href="layer_spatial_dropout_1d.html">layer_spatial_dropout_1d</a>()</code> <br><code><a href="layer_spatial_dropout_2d.html">layer_spatial_dropout_2d</a>()</code> <br><code><a href="layer_spatial_dropout_3d.html">layer_spatial_dropout_3d</a>()</code> <br></p>
-<p>Other layers: <br><code><a href="Layer.html">Layer</a>()</code> <br><code><a href="layer_activation.html">layer_activation</a>()</code> <br><code><a href="layer_activation_elu.html">layer_activation_elu</a>()</code> <br><code><a href="layer_activation_leaky_relu.html">layer_activation_leaky_relu</a>()</code> <br><code><a href="layer_activation_parametric_relu.html">layer_activation_parametric_relu</a>()</code> <br><code><a href="layer_activation_relu.html">layer_activation_relu</a>()</code> <br><code><a href="layer_activation_softmax.html">layer_activation_softmax</a>()</code> <br><code><a href="layer_activity_regularization.html">layer_activity_regularization</a>()</code> <br><code><a href="layer_add.html">layer_add</a>()</code> <br><code><a href="layer_additive_attention.html">layer_additive_attention</a>()</code> <br><code><a href="layer_alpha_dropout.html">layer_alpha_dropout</a>()</code> <br><code><a href="layer_attention.html">layer_attention</a>()</code> <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_average.html">layer_average</a>()</code> <br><code><a href="layer_average_pooling_1d.html">layer_average_pooling_1d</a>()</code> <br><code><a href="layer_average_pooling_2d.html">layer_average_pooling_2d</a>()</code> <br><code><a href="layer_average_pooling_3d.html">layer_average_pooling_3d</a>()</code> <br><code><a href="layer_batch_normalization.html">layer_batch_normalization</a>()</code> <br><code><a href="layer_bidirectional.html">layer_bidirectional</a>()</code> <br><code><a href="layer_category_encoding.html">layer_category_encoding</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_concatenate.html">layer_concatenate</a>()</code> <br><code><a href="layer_conv_1d.html">layer_conv_1d</a>()</code> <br><code><a href="layer_conv_1d_transpose.html">layer_conv_1d_transpose</a>()</code> <br><code><a href="layer_conv_2d.html">layer_conv_2d</a>()</code> <br><code><a href="layer_conv_2d_transpose.html">layer_conv_2d_transpose</a>()</code> <br><code><a href="layer_conv_3d.html">layer_conv_3d</a>()</code> <br><code><a href="layer_conv_3d_transpose.html">layer_conv_3d_transpose</a>()</code> <br><code><a href="layer_conv_lstm_1d.html">layer_conv_lstm_1d</a>()</code> <br><code><a href="layer_conv_lstm_2d.html">layer_conv_lstm_2d</a>()</code> <br><code><a href="layer_conv_lstm_3d.html">layer_conv_lstm_3d</a>()</code> <br><code><a href="layer_cropping_1d.html">layer_cropping_1d</a>()</code> <br><code><a href="layer_cropping_2d.html">layer_cropping_2d</a>()</code> <br><code><a href="layer_cropping_3d.html">layer_cropping_3d</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_dense.html">layer_dense</a>()</code> <br><code><a href="layer_depthwise_conv_1d.html">layer_depthwise_conv_1d</a>()</code> <br><code><a href="layer_depthwise_conv_2d.html">layer_depthwise_conv_2d</a>()</code> <br><code><a href="layer_discretization.html">layer_discretization</a>()</code> <br><code><a href="layer_dot.html">layer_dot</a>()</code> <br><code><a href="layer_einsum_dense.html">layer_einsum_dense</a>()</code> <br><code><a href="layer_embedding.html">layer_embedding</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_feature_space.html">layer_feature_space</a>()</code> <br><code><a href="layer_flatten.html">layer_flatten</a>()</code> <br><code><a href="layer_flax_module_wrapper.html">layer_flax_module_wrapper</a>()</code> <br><code><a href="layer_gaussian_dropout.html">layer_gaussian_dropout</a>()</code> <br><code><a href="layer_gaussian_noise.html">layer_gaussian_noise</a>()</code> <br><code><a href="layer_global_average_pooling_1d.html">layer_global_average_pooling_1d</a>()</code> <br><code><a href="layer_global_average_pooling_2d.html">layer_global_average_pooling_2d</a>()</code> <br><code><a href="layer_global_average_pooling_3d.html">layer_global_average_pooling_3d</a>()</code> <br><code><a href="layer_global_max_pooling_1d.html">layer_global_max_pooling_1d</a>()</code> <br><code><a href="layer_global_max_pooling_2d.html">layer_global_max_pooling_2d</a>()</code> <br><code><a href="layer_global_max_pooling_3d.html">layer_global_max_pooling_3d</a>()</code> <br><code><a href="layer_group_normalization.html">layer_group_normalization</a>()</code> <br><code><a href="layer_group_query_attention.html">layer_group_query_attention</a>()</code> <br><code><a href="layer_gru.html">layer_gru</a>()</code> <br><code><a href="layer_hashed_crossing.html">layer_hashed_crossing</a>()</code> <br><code><a href="layer_hashing.html">layer_hashing</a>()</code> <br><code><a href="layer_identity.html">layer_identity</a>()</code> <br><code><a href="layer_integer_lookup.html">layer_integer_lookup</a>()</code> <br><code><a href="layer_jax_model_wrapper.html">layer_jax_model_wrapper</a>()</code> <br><code><a href="layer_lambda.html">layer_lambda</a>()</code> <br><code><a href="layer_layer_normalization.html">layer_layer_normalization</a>()</code> <br><code><a href="layer_lstm.html">layer_lstm</a>()</code> <br><code><a href="layer_masking.html">layer_masking</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_max_pooling_1d.html">layer_max_pooling_1d</a>()</code> <br><code><a href="layer_max_pooling_2d.html">layer_max_pooling_2d</a>()</code> <br><code><a href="layer_max_pooling_3d.html">layer_max_pooling_3d</a>()</code> <br><code><a href="layer_maximum.html">layer_maximum</a>()</code> <br><code><a href="layer_mel_spectrogram.html">layer_mel_spectrogram</a>()</code> <br><code><a href="layer_minimum.html">layer_minimum</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_multi_head_attention.html">layer_multi_head_attention</a>()</code> <br><code><a href="layer_multiply.html">layer_multiply</a>()</code> <br><code><a href="layer_normalization.html">layer_normalization</a>()</code> <br><code><a href="layer_permute.html">layer_permute</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_brightness.html">layer_random_brightness</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_contrast.html">layer_random_contrast</a>()</code> <br><code><a href="layer_random_crop.html">layer_random_crop</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_flip.html">layer_random_flip</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_rotation.html">layer_random_rotation</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_random_translation.html">layer_random_translation</a>()</code> <br><code><a href="layer_random_zoom.html">layer_random_zoom</a>()</code> <br><code><a href="layer_repeat_vector.html">layer_repeat_vector</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_reshape.html">layer_reshape</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_rms_normalization.html">layer_rms_normalization</a>()</code> <br><code><a href="layer_rnn.html">layer_rnn</a>()</code> <br><code><a href="layer_separable_conv_1d.html">layer_separable_conv_1d</a>()</code> <br><code><a href="layer_separable_conv_2d.html">layer_separable_conv_2d</a>()</code> <br><code><a href="layer_simple_rnn.html">layer_simple_rnn</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br><code><a href="layer_spatial_dropout_1d.html">layer_spatial_dropout_1d</a>()</code> <br><code><a href="layer_spatial_dropout_2d.html">layer_spatial_dropout_2d</a>()</code> <br><code><a href="layer_spatial_dropout_3d.html">layer_spatial_dropout_3d</a>()</code> <br><code><a href="layer_spectral_normalization.html">layer_spectral_normalization</a>()</code> <br><code><a href="layer_stft_spectrogram.html">layer_stft_spectrogram</a>()</code> <br><code><a href="layer_string_lookup.html">layer_string_lookup</a>()</code> <br><code><a href="layer_subtract.html">layer_subtract</a>()</code> <br><code><a href="layer_text_vectorization.html">layer_text_vectorization</a>()</code> <br><code><a href="layer_tfsm.html">layer_tfsm</a>()</code> <br><code><a href="layer_time_distributed.html">layer_time_distributed</a>()</code> <br><code><a href="layer_torch_module_wrapper.html">layer_torch_module_wrapper</a>()</code> <br><code><a href="layer_unit_normalization.html">layer_unit_normalization</a>()</code> <br><code><a href="layer_upsampling_1d.html">layer_upsampling_1d</a>()</code> <br><code><a href="layer_upsampling_2d.html">layer_upsampling_2d</a>()</code> <br><code><a href="layer_upsampling_3d.html">layer_upsampling_3d</a>()</code> <br><code><a href="layer_zero_padding_1d.html">layer_zero_padding_1d</a>()</code> <br><code><a href="layer_zero_padding_2d.html">layer_zero_padding_2d</a>()</code> <br><code><a href="layer_zero_padding_3d.html">layer_zero_padding_3d</a>()</code> <br><code><a href="rnn_cell_gru.html">rnn_cell_gru</a>()</code> <br><code><a href="rnn_cell_lstm.html">rnn_cell_lstm</a>()</code> <br><code><a href="rnn_cell_simple.html">rnn_cell_simple</a>()</code> <br><code><a href="rnn_cells_stack.html">rnn_cells_stack</a>()</code> <br></p></div>
+<p>Other layers: <br><code><a href="Layer.html">Layer</a>()</code> <br><code><a href="layer_activation.html">layer_activation</a>()</code> <br><code><a href="layer_activation_elu.html">layer_activation_elu</a>()</code> <br><code><a href="layer_activation_leaky_relu.html">layer_activation_leaky_relu</a>()</code> <br><code><a href="layer_activation_parametric_relu.html">layer_activation_parametric_relu</a>()</code> <br><code><a href="layer_activation_relu.html">layer_activation_relu</a>()</code> <br><code><a href="layer_activation_softmax.html">layer_activation_softmax</a>()</code> <br><code><a href="layer_activity_regularization.html">layer_activity_regularization</a>()</code> <br><code><a href="layer_add.html">layer_add</a>()</code> <br><code><a href="layer_additive_attention.html">layer_additive_attention</a>()</code> <br><code><a href="layer_alpha_dropout.html">layer_alpha_dropout</a>()</code> <br><code><a href="layer_attention.html">layer_attention</a>()</code> <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_average.html">layer_average</a>()</code> <br><code><a href="layer_average_pooling_1d.html">layer_average_pooling_1d</a>()</code> <br><code><a href="layer_average_pooling_2d.html">layer_average_pooling_2d</a>()</code> <br><code><a href="layer_average_pooling_3d.html">layer_average_pooling_3d</a>()</code> <br><code><a href="layer_batch_normalization.html">layer_batch_normalization</a>()</code> <br><code><a href="layer_bidirectional.html">layer_bidirectional</a>()</code> <br><code><a href="layer_category_encoding.html">layer_category_encoding</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_concatenate.html">layer_concatenate</a>()</code> <br><code><a href="layer_conv_1d.html">layer_conv_1d</a>()</code> <br><code><a href="layer_conv_1d_transpose.html">layer_conv_1d_transpose</a>()</code> <br><code><a href="layer_conv_2d.html">layer_conv_2d</a>()</code> <br><code><a href="layer_conv_2d_transpose.html">layer_conv_2d_transpose</a>()</code> <br><code><a href="layer_conv_3d.html">layer_conv_3d</a>()</code> <br><code><a href="layer_conv_3d_transpose.html">layer_conv_3d_transpose</a>()</code> <br><code><a href="layer_conv_lstm_1d.html">layer_conv_lstm_1d</a>()</code> <br><code><a href="layer_conv_lstm_2d.html">layer_conv_lstm_2d</a>()</code> <br><code><a href="layer_conv_lstm_3d.html">layer_conv_lstm_3d</a>()</code> <br><code><a href="layer_cropping_1d.html">layer_cropping_1d</a>()</code> <br><code><a href="layer_cropping_2d.html">layer_cropping_2d</a>()</code> <br><code><a href="layer_cropping_3d.html">layer_cropping_3d</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_dense.html">layer_dense</a>()</code> <br><code><a href="layer_depthwise_conv_1d.html">layer_depthwise_conv_1d</a>()</code> <br><code><a href="layer_depthwise_conv_2d.html">layer_depthwise_conv_2d</a>()</code> <br><code><a href="layer_discretization.html">layer_discretization</a>()</code> <br><code><a href="layer_dot.html">layer_dot</a>()</code> <br><code><a href="layer_einsum_dense.html">layer_einsum_dense</a>()</code> <br><code><a href="layer_embedding.html">layer_embedding</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_feature_space.html">layer_feature_space</a>()</code> <br><code><a href="layer_flatten.html">layer_flatten</a>()</code> <br><code><a href="layer_flax_module_wrapper.html">layer_flax_module_wrapper</a>()</code> <br><code><a href="layer_gaussian_dropout.html">layer_gaussian_dropout</a>()</code> <br><code><a href="layer_gaussian_noise.html">layer_gaussian_noise</a>()</code> <br><code><a href="layer_global_average_pooling_1d.html">layer_global_average_pooling_1d</a>()</code> <br><code><a href="layer_global_average_pooling_2d.html">layer_global_average_pooling_2d</a>()</code> <br><code><a href="layer_global_average_pooling_3d.html">layer_global_average_pooling_3d</a>()</code> <br><code><a href="layer_global_max_pooling_1d.html">layer_global_max_pooling_1d</a>()</code> <br><code><a href="layer_global_max_pooling_2d.html">layer_global_max_pooling_2d</a>()</code> <br><code><a href="layer_global_max_pooling_3d.html">layer_global_max_pooling_3d</a>()</code> <br><code><a href="layer_group_normalization.html">layer_group_normalization</a>()</code> <br><code><a href="layer_group_query_attention.html">layer_group_query_attention</a>()</code> <br><code><a href="layer_gru.html">layer_gru</a>()</code> <br><code><a href="layer_hashed_crossing.html">layer_hashed_crossing</a>()</code> <br><code><a href="layer_hashing.html">layer_hashing</a>()</code> <br><code><a href="layer_identity.html">layer_identity</a>()</code> <br><code><a href="layer_integer_lookup.html">layer_integer_lookup</a>()</code> <br><code><a href="layer_jax_model_wrapper.html">layer_jax_model_wrapper</a>()</code> <br><code><a href="layer_lambda.html">layer_lambda</a>()</code> <br><code><a href="layer_layer_normalization.html">layer_layer_normalization</a>()</code> <br><code><a href="layer_lstm.html">layer_lstm</a>()</code> <br><code><a href="layer_masking.html">layer_masking</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_max_pooling_1d.html">layer_max_pooling_1d</a>()</code> <br><code><a href="layer_max_pooling_2d.html">layer_max_pooling_2d</a>()</code> <br><code><a href="layer_max_pooling_3d.html">layer_max_pooling_3d</a>()</code> <br><code><a href="layer_maximum.html">layer_maximum</a>()</code> <br><code><a href="layer_mel_spectrogram.html">layer_mel_spectrogram</a>()</code> <br><code><a href="layer_minimum.html">layer_minimum</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_multi_head_attention.html">layer_multi_head_attention</a>()</code> <br><code><a href="layer_multiply.html">layer_multiply</a>()</code> <br><code><a href="layer_normalization.html">layer_normalization</a>()</code> <br><code><a href="layer_permute.html">layer_permute</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_brightness.html">layer_random_brightness</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_contrast.html">layer_random_contrast</a>()</code> <br><code><a href="layer_random_crop.html">layer_random_crop</a>()</code> <br><code><a href="layer_random_elastic_transform.html">layer_random_elastic_transform</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_flip.html">layer_random_flip</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_rotation.html">layer_random_rotation</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_random_translation.html">layer_random_translation</a>()</code> <br><code><a href="layer_random_zoom.html">layer_random_zoom</a>()</code> <br><code><a href="layer_repeat_vector.html">layer_repeat_vector</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_reshape.html">layer_reshape</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_rms_normalization.html">layer_rms_normalization</a>()</code> <br><code><a href="layer_rnn.html">layer_rnn</a>()</code> <br><code><a href="layer_separable_conv_1d.html">layer_separable_conv_1d</a>()</code> <br><code><a href="layer_separable_conv_2d.html">layer_separable_conv_2d</a>()</code> <br><code><a href="layer_simple_rnn.html">layer_simple_rnn</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br><code><a href="layer_spatial_dropout_1d.html">layer_spatial_dropout_1d</a>()</code> <br><code><a href="layer_spatial_dropout_2d.html">layer_spatial_dropout_2d</a>()</code> <br><code><a href="layer_spatial_dropout_3d.html">layer_spatial_dropout_3d</a>()</code> <br><code><a href="layer_spectral_normalization.html">layer_spectral_normalization</a>()</code> <br><code><a href="layer_stft_spectrogram.html">layer_stft_spectrogram</a>()</code> <br><code><a href="layer_string_lookup.html">layer_string_lookup</a>()</code> <br><code><a href="layer_subtract.html">layer_subtract</a>()</code> <br><code><a href="layer_text_vectorization.html">layer_text_vectorization</a>()</code> <br><code><a href="layer_tfsm.html">layer_tfsm</a>()</code> <br><code><a href="layer_time_distributed.html">layer_time_distributed</a>()</code> <br><code><a href="layer_torch_module_wrapper.html">layer_torch_module_wrapper</a>()</code> <br><code><a href="layer_unit_normalization.html">layer_unit_normalization</a>()</code> <br><code><a href="layer_upsampling_1d.html">layer_upsampling_1d</a>()</code> <br><code><a href="layer_upsampling_2d.html">layer_upsampling_2d</a>()</code> <br><code><a href="layer_upsampling_3d.html">layer_upsampling_3d</a>()</code> <br><code><a href="layer_zero_padding_1d.html">layer_zero_padding_1d</a>()</code> <br><code><a href="layer_zero_padding_2d.html">layer_zero_padding_2d</a>()</code> <br><code><a href="layer_zero_padding_3d.html">layer_zero_padding_3d</a>()</code> <br><code><a href="rnn_cell_gru.html">rnn_cell_gru</a>()</code> <br><code><a href="rnn_cell_lstm.html">rnn_cell_lstm</a>()</code> <br><code><a href="rnn_cell_simple.html">rnn_cell_simple</a>()</code> <br><code><a href="rnn_cells_stack.html">rnn_cells_stack</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -157,7 +157,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/layer_einsum_dense.html b/docs/dev/reference/layer_einsum_dense.html
index ab8983973..4e9ff2d47 100644
--- a/docs/dev/reference/layer_einsum_dense.html
+++ b/docs/dev/reference/layer_einsum_dense.html
@@ -8,7 +8,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -72,6 +72,7 @@ <h2 id="ref-usage">Usage<a class="anchor" aria-label="anchor" href="#ref-usage">
 <span>  kernel_constraint <span class="op">=</span> <span class="cn">NULL</span>,</span>
 <span>  bias_constraint <span class="op">=</span> <span class="cn">NULL</span>,</span>
 <span>  lora_rank <span class="op">=</span> <span class="cn">NULL</span>,</span>
+<span>  lora_alpha <span class="op">=</span> <span class="cn">NULL</span>,</span>
 <span>  <span class="va">...</span></span>
 <span><span class="op">)</span></span></code></pre></div>
     </div>
@@ -152,6 +153,12 @@ <h2 id="arguments">Arguments<a class="anchor" aria-label="anchor" href="#argumen
 <code>EinsumDense</code> layer by calling <code>layer$enable_lora(rank)</code>.</p></dd>
 
 
+<dt id="arg-lora-alpha">lora_alpha<a class="anchor" aria-label="anchor" href="#arg-lora-alpha"></a></dt>
+<dd><p>Optional integer. Scales the low-rank adaptation delta during the forward
+pass. The delta is scaled by <code>lora_alpha / lora_rank</code>, letting you tune the
+LoRA adjustment strength independently of <code>lora_rank</code>.</p></dd>
+
+
 <dt id="arg--">...<a class="anchor" aria-label="anchor" href="#arg--"></a></dt>
 <dd><p>Base layer keyword arguments, such as <code>name</code> and <code>dtype</code>.</p></dd>
 
@@ -231,7 +238,7 @@ <h2 id="readonly-properties-">Readonly properties:<a class="anchor" aria-label="
     <div class="section level2">
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
     <div class="dont-index"><p>Other core layers: <br><code><a href="layer_dense.html">layer_dense</a>()</code> <br><code><a href="layer_embedding.html">layer_embedding</a>()</code> <br><code><a href="layer_identity.html">layer_identity</a>()</code> <br><code><a href="layer_lambda.html">layer_lambda</a>()</code> <br><code><a href="layer_masking.html">layer_masking</a>()</code> <br></p>
-<p>Other layers: <br><code><a href="Layer.html">Layer</a>()</code> <br><code><a href="layer_activation.html">layer_activation</a>()</code> <br><code><a href="layer_activation_elu.html">layer_activation_elu</a>()</code> <br><code><a href="layer_activation_leaky_relu.html">layer_activation_leaky_relu</a>()</code> <br><code><a href="layer_activation_parametric_relu.html">layer_activation_parametric_relu</a>()</code> <br><code><a href="layer_activation_relu.html">layer_activation_relu</a>()</code> <br><code><a href="layer_activation_softmax.html">layer_activation_softmax</a>()</code> <br><code><a href="layer_activity_regularization.html">layer_activity_regularization</a>()</code> <br><code><a href="layer_add.html">layer_add</a>()</code> <br><code><a href="layer_additive_attention.html">layer_additive_attention</a>()</code> <br><code><a href="layer_alpha_dropout.html">layer_alpha_dropout</a>()</code> <br><code><a href="layer_attention.html">layer_attention</a>()</code> <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_average.html">layer_average</a>()</code> <br><code><a href="layer_average_pooling_1d.html">layer_average_pooling_1d</a>()</code> <br><code><a href="layer_average_pooling_2d.html">layer_average_pooling_2d</a>()</code> <br><code><a href="layer_average_pooling_3d.html">layer_average_pooling_3d</a>()</code> <br><code><a href="layer_batch_normalization.html">layer_batch_normalization</a>()</code> <br><code><a href="layer_bidirectional.html">layer_bidirectional</a>()</code> <br><code><a href="layer_category_encoding.html">layer_category_encoding</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_concatenate.html">layer_concatenate</a>()</code> <br><code><a href="layer_conv_1d.html">layer_conv_1d</a>()</code> <br><code><a href="layer_conv_1d_transpose.html">layer_conv_1d_transpose</a>()</code> <br><code><a href="layer_conv_2d.html">layer_conv_2d</a>()</code> <br><code><a href="layer_conv_2d_transpose.html">layer_conv_2d_transpose</a>()</code> <br><code><a href="layer_conv_3d.html">layer_conv_3d</a>()</code> <br><code><a href="layer_conv_3d_transpose.html">layer_conv_3d_transpose</a>()</code> <br><code><a href="layer_conv_lstm_1d.html">layer_conv_lstm_1d</a>()</code> <br><code><a href="layer_conv_lstm_2d.html">layer_conv_lstm_2d</a>()</code> <br><code><a href="layer_conv_lstm_3d.html">layer_conv_lstm_3d</a>()</code> <br><code><a href="layer_cropping_1d.html">layer_cropping_1d</a>()</code> <br><code><a href="layer_cropping_2d.html">layer_cropping_2d</a>()</code> <br><code><a href="layer_cropping_3d.html">layer_cropping_3d</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_dense.html">layer_dense</a>()</code> <br><code><a href="layer_depthwise_conv_1d.html">layer_depthwise_conv_1d</a>()</code> <br><code><a href="layer_depthwise_conv_2d.html">layer_depthwise_conv_2d</a>()</code> <br><code><a href="layer_discretization.html">layer_discretization</a>()</code> <br><code><a href="layer_dot.html">layer_dot</a>()</code> <br><code><a href="layer_dropout.html">layer_dropout</a>()</code> <br><code><a href="layer_embedding.html">layer_embedding</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_feature_space.html">layer_feature_space</a>()</code> <br><code><a href="layer_flatten.html">layer_flatten</a>()</code> <br><code><a href="layer_flax_module_wrapper.html">layer_flax_module_wrapper</a>()</code> <br><code><a href="layer_gaussian_dropout.html">layer_gaussian_dropout</a>()</code> <br><code><a href="layer_gaussian_noise.html">layer_gaussian_noise</a>()</code> <br><code><a href="layer_global_average_pooling_1d.html">layer_global_average_pooling_1d</a>()</code> <br><code><a href="layer_global_average_pooling_2d.html">layer_global_average_pooling_2d</a>()</code> <br><code><a href="layer_global_average_pooling_3d.html">layer_global_average_pooling_3d</a>()</code> <br><code><a href="layer_global_max_pooling_1d.html">layer_global_max_pooling_1d</a>()</code> <br><code><a href="layer_global_max_pooling_2d.html">layer_global_max_pooling_2d</a>()</code> <br><code><a href="layer_global_max_pooling_3d.html">layer_global_max_pooling_3d</a>()</code> <br><code><a href="layer_group_normalization.html">layer_group_normalization</a>()</code> <br><code><a href="layer_group_query_attention.html">layer_group_query_attention</a>()</code> <br><code><a href="layer_gru.html">layer_gru</a>()</code> <br><code><a href="layer_hashed_crossing.html">layer_hashed_crossing</a>()</code> <br><code><a href="layer_hashing.html">layer_hashing</a>()</code> <br><code><a href="layer_identity.html">layer_identity</a>()</code> <br><code><a href="layer_integer_lookup.html">layer_integer_lookup</a>()</code> <br><code><a href="layer_jax_model_wrapper.html">layer_jax_model_wrapper</a>()</code> <br><code><a href="layer_lambda.html">layer_lambda</a>()</code> <br><code><a href="layer_layer_normalization.html">layer_layer_normalization</a>()</code> <br><code><a href="layer_lstm.html">layer_lstm</a>()</code> <br><code><a href="layer_masking.html">layer_masking</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_max_pooling_1d.html">layer_max_pooling_1d</a>()</code> <br><code><a href="layer_max_pooling_2d.html">layer_max_pooling_2d</a>()</code> <br><code><a href="layer_max_pooling_3d.html">layer_max_pooling_3d</a>()</code> <br><code><a href="layer_maximum.html">layer_maximum</a>()</code> <br><code><a href="layer_mel_spectrogram.html">layer_mel_spectrogram</a>()</code> <br><code><a href="layer_minimum.html">layer_minimum</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_multi_head_attention.html">layer_multi_head_attention</a>()</code> <br><code><a href="layer_multiply.html">layer_multiply</a>()</code> <br><code><a href="layer_normalization.html">layer_normalization</a>()</code> <br><code><a href="layer_permute.html">layer_permute</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_brightness.html">layer_random_brightness</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_contrast.html">layer_random_contrast</a>()</code> <br><code><a href="layer_random_crop.html">layer_random_crop</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_flip.html">layer_random_flip</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_rotation.html">layer_random_rotation</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_random_translation.html">layer_random_translation</a>()</code> <br><code><a href="layer_random_zoom.html">layer_random_zoom</a>()</code> <br><code><a href="layer_repeat_vector.html">layer_repeat_vector</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_reshape.html">layer_reshape</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_rms_normalization.html">layer_rms_normalization</a>()</code> <br><code><a href="layer_rnn.html">layer_rnn</a>()</code> <br><code><a href="layer_separable_conv_1d.html">layer_separable_conv_1d</a>()</code> <br><code><a href="layer_separable_conv_2d.html">layer_separable_conv_2d</a>()</code> <br><code><a href="layer_simple_rnn.html">layer_simple_rnn</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br><code><a href="layer_spatial_dropout_1d.html">layer_spatial_dropout_1d</a>()</code> <br><code><a href="layer_spatial_dropout_2d.html">layer_spatial_dropout_2d</a>()</code> <br><code><a href="layer_spatial_dropout_3d.html">layer_spatial_dropout_3d</a>()</code> <br><code><a href="layer_spectral_normalization.html">layer_spectral_normalization</a>()</code> <br><code><a href="layer_stft_spectrogram.html">layer_stft_spectrogram</a>()</code> <br><code><a href="layer_string_lookup.html">layer_string_lookup</a>()</code> <br><code><a href="layer_subtract.html">layer_subtract</a>()</code> <br><code><a href="layer_text_vectorization.html">layer_text_vectorization</a>()</code> <br><code><a href="layer_tfsm.html">layer_tfsm</a>()</code> <br><code><a href="layer_time_distributed.html">layer_time_distributed</a>()</code> <br><code><a href="layer_torch_module_wrapper.html">layer_torch_module_wrapper</a>()</code> <br><code><a href="layer_unit_normalization.html">layer_unit_normalization</a>()</code> <br><code><a href="layer_upsampling_1d.html">layer_upsampling_1d</a>()</code> <br><code><a href="layer_upsampling_2d.html">layer_upsampling_2d</a>()</code> <br><code><a href="layer_upsampling_3d.html">layer_upsampling_3d</a>()</code> <br><code><a href="layer_zero_padding_1d.html">layer_zero_padding_1d</a>()</code> <br><code><a href="layer_zero_padding_2d.html">layer_zero_padding_2d</a>()</code> <br><code><a href="layer_zero_padding_3d.html">layer_zero_padding_3d</a>()</code> <br><code><a href="rnn_cell_gru.html">rnn_cell_gru</a>()</code> <br><code><a href="rnn_cell_lstm.html">rnn_cell_lstm</a>()</code> <br><code><a href="rnn_cell_simple.html">rnn_cell_simple</a>()</code> <br><code><a href="rnn_cells_stack.html">rnn_cells_stack</a>()</code> <br></p></div>
+<p>Other layers: <br><code><a href="Layer.html">Layer</a>()</code> <br><code><a href="layer_activation.html">layer_activation</a>()</code> <br><code><a href="layer_activation_elu.html">layer_activation_elu</a>()</code> <br><code><a href="layer_activation_leaky_relu.html">layer_activation_leaky_relu</a>()</code> <br><code><a href="layer_activation_parametric_relu.html">layer_activation_parametric_relu</a>()</code> <br><code><a href="layer_activation_relu.html">layer_activation_relu</a>()</code> <br><code><a href="layer_activation_softmax.html">layer_activation_softmax</a>()</code> <br><code><a href="layer_activity_regularization.html">layer_activity_regularization</a>()</code> <br><code><a href="layer_add.html">layer_add</a>()</code> <br><code><a href="layer_additive_attention.html">layer_additive_attention</a>()</code> <br><code><a href="layer_alpha_dropout.html">layer_alpha_dropout</a>()</code> <br><code><a href="layer_attention.html">layer_attention</a>()</code> <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_average.html">layer_average</a>()</code> <br><code><a href="layer_average_pooling_1d.html">layer_average_pooling_1d</a>()</code> <br><code><a href="layer_average_pooling_2d.html">layer_average_pooling_2d</a>()</code> <br><code><a href="layer_average_pooling_3d.html">layer_average_pooling_3d</a>()</code> <br><code><a href="layer_batch_normalization.html">layer_batch_normalization</a>()</code> <br><code><a href="layer_bidirectional.html">layer_bidirectional</a>()</code> <br><code><a href="layer_category_encoding.html">layer_category_encoding</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_concatenate.html">layer_concatenate</a>()</code> <br><code><a href="layer_conv_1d.html">layer_conv_1d</a>()</code> <br><code><a href="layer_conv_1d_transpose.html">layer_conv_1d_transpose</a>()</code> <br><code><a href="layer_conv_2d.html">layer_conv_2d</a>()</code> <br><code><a href="layer_conv_2d_transpose.html">layer_conv_2d_transpose</a>()</code> <br><code><a href="layer_conv_3d.html">layer_conv_3d</a>()</code> <br><code><a href="layer_conv_3d_transpose.html">layer_conv_3d_transpose</a>()</code> <br><code><a href="layer_conv_lstm_1d.html">layer_conv_lstm_1d</a>()</code> <br><code><a href="layer_conv_lstm_2d.html">layer_conv_lstm_2d</a>()</code> <br><code><a href="layer_conv_lstm_3d.html">layer_conv_lstm_3d</a>()</code> <br><code><a href="layer_cropping_1d.html">layer_cropping_1d</a>()</code> <br><code><a href="layer_cropping_2d.html">layer_cropping_2d</a>()</code> <br><code><a href="layer_cropping_3d.html">layer_cropping_3d</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_dense.html">layer_dense</a>()</code> <br><code><a href="layer_depthwise_conv_1d.html">layer_depthwise_conv_1d</a>()</code> <br><code><a href="layer_depthwise_conv_2d.html">layer_depthwise_conv_2d</a>()</code> <br><code><a href="layer_discretization.html">layer_discretization</a>()</code> <br><code><a href="layer_dot.html">layer_dot</a>()</code> <br><code><a href="layer_dropout.html">layer_dropout</a>()</code> <br><code><a href="layer_embedding.html">layer_embedding</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_feature_space.html">layer_feature_space</a>()</code> <br><code><a href="layer_flatten.html">layer_flatten</a>()</code> <br><code><a href="layer_flax_module_wrapper.html">layer_flax_module_wrapper</a>()</code> <br><code><a href="layer_gaussian_dropout.html">layer_gaussian_dropout</a>()</code> <br><code><a href="layer_gaussian_noise.html">layer_gaussian_noise</a>()</code> <br><code><a href="layer_global_average_pooling_1d.html">layer_global_average_pooling_1d</a>()</code> <br><code><a href="layer_global_average_pooling_2d.html">layer_global_average_pooling_2d</a>()</code> <br><code><a href="layer_global_average_pooling_3d.html">layer_global_average_pooling_3d</a>()</code> <br><code><a href="layer_global_max_pooling_1d.html">layer_global_max_pooling_1d</a>()</code> <br><code><a href="layer_global_max_pooling_2d.html">layer_global_max_pooling_2d</a>()</code> <br><code><a href="layer_global_max_pooling_3d.html">layer_global_max_pooling_3d</a>()</code> <br><code><a href="layer_group_normalization.html">layer_group_normalization</a>()</code> <br><code><a href="layer_group_query_attention.html">layer_group_query_attention</a>()</code> <br><code><a href="layer_gru.html">layer_gru</a>()</code> <br><code><a href="layer_hashed_crossing.html">layer_hashed_crossing</a>()</code> <br><code><a href="layer_hashing.html">layer_hashing</a>()</code> <br><code><a href="layer_identity.html">layer_identity</a>()</code> <br><code><a href="layer_integer_lookup.html">layer_integer_lookup</a>()</code> <br><code><a href="layer_jax_model_wrapper.html">layer_jax_model_wrapper</a>()</code> <br><code><a href="layer_lambda.html">layer_lambda</a>()</code> <br><code><a href="layer_layer_normalization.html">layer_layer_normalization</a>()</code> <br><code><a href="layer_lstm.html">layer_lstm</a>()</code> <br><code><a href="layer_masking.html">layer_masking</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_max_pooling_1d.html">layer_max_pooling_1d</a>()</code> <br><code><a href="layer_max_pooling_2d.html">layer_max_pooling_2d</a>()</code> <br><code><a href="layer_max_pooling_3d.html">layer_max_pooling_3d</a>()</code> <br><code><a href="layer_maximum.html">layer_maximum</a>()</code> <br><code><a href="layer_mel_spectrogram.html">layer_mel_spectrogram</a>()</code> <br><code><a href="layer_minimum.html">layer_minimum</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_multi_head_attention.html">layer_multi_head_attention</a>()</code> <br><code><a href="layer_multiply.html">layer_multiply</a>()</code> <br><code><a href="layer_normalization.html">layer_normalization</a>()</code> <br><code><a href="layer_permute.html">layer_permute</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_brightness.html">layer_random_brightness</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_contrast.html">layer_random_contrast</a>()</code> <br><code><a href="layer_random_crop.html">layer_random_crop</a>()</code> <br><code><a href="layer_random_elastic_transform.html">layer_random_elastic_transform</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_flip.html">layer_random_flip</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_rotation.html">layer_random_rotation</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_random_translation.html">layer_random_translation</a>()</code> <br><code><a href="layer_random_zoom.html">layer_random_zoom</a>()</code> <br><code><a href="layer_repeat_vector.html">layer_repeat_vector</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_reshape.html">layer_reshape</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_rms_normalization.html">layer_rms_normalization</a>()</code> <br><code><a href="layer_rnn.html">layer_rnn</a>()</code> <br><code><a href="layer_separable_conv_1d.html">layer_separable_conv_1d</a>()</code> <br><code><a href="layer_separable_conv_2d.html">layer_separable_conv_2d</a>()</code> <br><code><a href="layer_simple_rnn.html">layer_simple_rnn</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br><code><a href="layer_spatial_dropout_1d.html">layer_spatial_dropout_1d</a>()</code> <br><code><a href="layer_spatial_dropout_2d.html">layer_spatial_dropout_2d</a>()</code> <br><code><a href="layer_spatial_dropout_3d.html">layer_spatial_dropout_3d</a>()</code> <br><code><a href="layer_spectral_normalization.html">layer_spectral_normalization</a>()</code> <br><code><a href="layer_stft_spectrogram.html">layer_stft_spectrogram</a>()</code> <br><code><a href="layer_string_lookup.html">layer_string_lookup</a>()</code> <br><code><a href="layer_subtract.html">layer_subtract</a>()</code> <br><code><a href="layer_text_vectorization.html">layer_text_vectorization</a>()</code> <br><code><a href="layer_tfsm.html">layer_tfsm</a>()</code> <br><code><a href="layer_time_distributed.html">layer_time_distributed</a>()</code> <br><code><a href="layer_torch_module_wrapper.html">layer_torch_module_wrapper</a>()</code> <br><code><a href="layer_unit_normalization.html">layer_unit_normalization</a>()</code> <br><code><a href="layer_upsampling_1d.html">layer_upsampling_1d</a>()</code> <br><code><a href="layer_upsampling_2d.html">layer_upsampling_2d</a>()</code> <br><code><a href="layer_upsampling_3d.html">layer_upsampling_3d</a>()</code> <br><code><a href="layer_zero_padding_1d.html">layer_zero_padding_1d</a>()</code> <br><code><a href="layer_zero_padding_2d.html">layer_zero_padding_2d</a>()</code> <br><code><a href="layer_zero_padding_3d.html">layer_zero_padding_3d</a>()</code> <br><code><a href="rnn_cell_gru.html">rnn_cell_gru</a>()</code> <br><code><a href="rnn_cell_lstm.html">rnn_cell_lstm</a>()</code> <br><code><a href="rnn_cell_simple.html">rnn_cell_simple</a>()</code> <br><code><a href="rnn_cells_stack.html">rnn_cells_stack</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -243,7 +250,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/layer_embedding.html b/docs/dev/reference/layer_embedding.html
index 40bd0fd2e..0ca38e948 100644
--- a/docs/dev/reference/layer_embedding.html
+++ b/docs/dev/reference/layer_embedding.html
@@ -10,7 +10,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -72,6 +72,7 @@ <h2 id="ref-usage">Usage<a class="anchor" aria-label="anchor" href="#ref-usage">
 <span>  mask_zero <span class="op">=</span> <span class="cn">FALSE</span>,</span>
 <span>  weights <span class="op">=</span> <span class="cn">NULL</span>,</span>
 <span>  lora_rank <span class="op">=</span> <span class="cn">NULL</span>,</span>
+<span>  lora_alpha <span class="op">=</span> <span class="cn">NULL</span>,</span>
 <span>  <span class="va">...</span></span>
 <span><span class="op">)</span></span></code></pre></div>
     </div>
@@ -138,6 +139,12 @@ <h2 id="arguments">Arguments<a class="anchor" aria-label="anchor" href="#argumen
 <code>Embedding</code> layer instance by calling <code>layer$enable_lora(rank)</code>.</p></dd>
 
 
+<dt id="arg-lora-alpha">lora_alpha<a class="anchor" aria-label="anchor" href="#arg-lora-alpha"></a></dt>
+<dd><p>Optional integer. Scales the low-rank adaptation delta during the forward
+pass. The delta is scaled by <code>lora_alpha / lora_rank</code>, letting you tune the
+LoRA adjustment strength independently of <code>lora_rank</code>.</p></dd>
+
+
 <dt id="arg--">...<a class="anchor" aria-label="anchor" href="#arg--"></a></dt>
 <dd><p>For forward/backward compatability.</p></dd>
 
@@ -183,10 +190,13 @@ <h2 id="methods">Methods<a class="anchor" aria-label="anchor" href="#methods"></
 <ul><li><p></p>
 <p></p><div class="sourceCode r"><pre><code><span><span class="fu">enable_lora</span><span class="op">(</span></span>
 <span>  <span class="va">rank</span>,</span>
+<span>  lora_alpha <span class="op">=</span> <span class="cn">NULL</span>,</span>
 <span>  a_initializer <span class="op">=</span> <span class="st">'he_uniform'</span>,</span>
 <span>  b_initializer <span class="op">=</span> <span class="st">'zeros'</span></span>
 <span><span class="op">)</span></span></code></pre><p></p></div></li>
 <li><p></p>
+<p></p><div class="sourceCode r"><pre><code><span><span class="fu">compute_output_spec</span><span class="op">(</span><span class="va">...</span><span class="op">)</span></span></code></pre><p></p></div></li>
+<li><p></p>
 <p></p><div class="sourceCode r"><pre><code><span><span class="fu">quantize</span><span class="op">(</span><span class="va">mode</span>, type_check <span class="op">=</span> <span class="cn">TRUE</span><span class="op">)</span></span></code></pre><p></p></div></li>
 <li><p></p>
 <p></p><div class="sourceCode r"><pre><code><span><span class="fu">quantized_build</span><span class="op">(</span><span class="va">input_shape</span>, <span class="va">mode</span><span class="op">)</span></span></code></pre><p></p></div></li>
@@ -203,7 +213,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
     <div class="dont-index">
 <ul><li><p><a href="https://keras.io/api/layers/core_layers/embedding#embedding-class" class="external-link">https://keras.io/api/layers/core_layers/embedding#embedding-class</a></p></li>
 </ul><p>Other core layers: <br><code><a href="layer_dense.html">layer_dense</a>()</code> <br><code><a href="layer_einsum_dense.html">layer_einsum_dense</a>()</code> <br><code><a href="layer_identity.html">layer_identity</a>()</code> <br><code><a href="layer_lambda.html">layer_lambda</a>()</code> <br><code><a href="layer_masking.html">layer_masking</a>()</code> <br></p>
-<p>Other layers: <br><code><a href="Layer.html">Layer</a>()</code> <br><code><a href="layer_activation.html">layer_activation</a>()</code> <br><code><a href="layer_activation_elu.html">layer_activation_elu</a>()</code> <br><code><a href="layer_activation_leaky_relu.html">layer_activation_leaky_relu</a>()</code> <br><code><a href="layer_activation_parametric_relu.html">layer_activation_parametric_relu</a>()</code> <br><code><a href="layer_activation_relu.html">layer_activation_relu</a>()</code> <br><code><a href="layer_activation_softmax.html">layer_activation_softmax</a>()</code> <br><code><a href="layer_activity_regularization.html">layer_activity_regularization</a>()</code> <br><code><a href="layer_add.html">layer_add</a>()</code> <br><code><a href="layer_additive_attention.html">layer_additive_attention</a>()</code> <br><code><a href="layer_alpha_dropout.html">layer_alpha_dropout</a>()</code> <br><code><a href="layer_attention.html">layer_attention</a>()</code> <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_average.html">layer_average</a>()</code> <br><code><a href="layer_average_pooling_1d.html">layer_average_pooling_1d</a>()</code> <br><code><a href="layer_average_pooling_2d.html">layer_average_pooling_2d</a>()</code> <br><code><a href="layer_average_pooling_3d.html">layer_average_pooling_3d</a>()</code> <br><code><a href="layer_batch_normalization.html">layer_batch_normalization</a>()</code> <br><code><a href="layer_bidirectional.html">layer_bidirectional</a>()</code> <br><code><a href="layer_category_encoding.html">layer_category_encoding</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_concatenate.html">layer_concatenate</a>()</code> <br><code><a href="layer_conv_1d.html">layer_conv_1d</a>()</code> <br><code><a href="layer_conv_1d_transpose.html">layer_conv_1d_transpose</a>()</code> <br><code><a href="layer_conv_2d.html">layer_conv_2d</a>()</code> <br><code><a href="layer_conv_2d_transpose.html">layer_conv_2d_transpose</a>()</code> <br><code><a href="layer_conv_3d.html">layer_conv_3d</a>()</code> <br><code><a href="layer_conv_3d_transpose.html">layer_conv_3d_transpose</a>()</code> <br><code><a href="layer_conv_lstm_1d.html">layer_conv_lstm_1d</a>()</code> <br><code><a href="layer_conv_lstm_2d.html">layer_conv_lstm_2d</a>()</code> <br><code><a href="layer_conv_lstm_3d.html">layer_conv_lstm_3d</a>()</code> <br><code><a href="layer_cropping_1d.html">layer_cropping_1d</a>()</code> <br><code><a href="layer_cropping_2d.html">layer_cropping_2d</a>()</code> <br><code><a href="layer_cropping_3d.html">layer_cropping_3d</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_dense.html">layer_dense</a>()</code> <br><code><a href="layer_depthwise_conv_1d.html">layer_depthwise_conv_1d</a>()</code> <br><code><a href="layer_depthwise_conv_2d.html">layer_depthwise_conv_2d</a>()</code> <br><code><a href="layer_discretization.html">layer_discretization</a>()</code> <br><code><a href="layer_dot.html">layer_dot</a>()</code> <br><code><a href="layer_dropout.html">layer_dropout</a>()</code> <br><code><a href="layer_einsum_dense.html">layer_einsum_dense</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_feature_space.html">layer_feature_space</a>()</code> <br><code><a href="layer_flatten.html">layer_flatten</a>()</code> <br><code><a href="layer_flax_module_wrapper.html">layer_flax_module_wrapper</a>()</code> <br><code><a href="layer_gaussian_dropout.html">layer_gaussian_dropout</a>()</code> <br><code><a href="layer_gaussian_noise.html">layer_gaussian_noise</a>()</code> <br><code><a href="layer_global_average_pooling_1d.html">layer_global_average_pooling_1d</a>()</code> <br><code><a href="layer_global_average_pooling_2d.html">layer_global_average_pooling_2d</a>()</code> <br><code><a href="layer_global_average_pooling_3d.html">layer_global_average_pooling_3d</a>()</code> <br><code><a href="layer_global_max_pooling_1d.html">layer_global_max_pooling_1d</a>()</code> <br><code><a href="layer_global_max_pooling_2d.html">layer_global_max_pooling_2d</a>()</code> <br><code><a href="layer_global_max_pooling_3d.html">layer_global_max_pooling_3d</a>()</code> <br><code><a href="layer_group_normalization.html">layer_group_normalization</a>()</code> <br><code><a href="layer_group_query_attention.html">layer_group_query_attention</a>()</code> <br><code><a href="layer_gru.html">layer_gru</a>()</code> <br><code><a href="layer_hashed_crossing.html">layer_hashed_crossing</a>()</code> <br><code><a href="layer_hashing.html">layer_hashing</a>()</code> <br><code><a href="layer_identity.html">layer_identity</a>()</code> <br><code><a href="layer_integer_lookup.html">layer_integer_lookup</a>()</code> <br><code><a href="layer_jax_model_wrapper.html">layer_jax_model_wrapper</a>()</code> <br><code><a href="layer_lambda.html">layer_lambda</a>()</code> <br><code><a href="layer_layer_normalization.html">layer_layer_normalization</a>()</code> <br><code><a href="layer_lstm.html">layer_lstm</a>()</code> <br><code><a href="layer_masking.html">layer_masking</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_max_pooling_1d.html">layer_max_pooling_1d</a>()</code> <br><code><a href="layer_max_pooling_2d.html">layer_max_pooling_2d</a>()</code> <br><code><a href="layer_max_pooling_3d.html">layer_max_pooling_3d</a>()</code> <br><code><a href="layer_maximum.html">layer_maximum</a>()</code> <br><code><a href="layer_mel_spectrogram.html">layer_mel_spectrogram</a>()</code> <br><code><a href="layer_minimum.html">layer_minimum</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_multi_head_attention.html">layer_multi_head_attention</a>()</code> <br><code><a href="layer_multiply.html">layer_multiply</a>()</code> <br><code><a href="layer_normalization.html">layer_normalization</a>()</code> <br><code><a href="layer_permute.html">layer_permute</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_brightness.html">layer_random_brightness</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_contrast.html">layer_random_contrast</a>()</code> <br><code><a href="layer_random_crop.html">layer_random_crop</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_flip.html">layer_random_flip</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_rotation.html">layer_random_rotation</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_random_translation.html">layer_random_translation</a>()</code> <br><code><a href="layer_random_zoom.html">layer_random_zoom</a>()</code> <br><code><a href="layer_repeat_vector.html">layer_repeat_vector</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_reshape.html">layer_reshape</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_rms_normalization.html">layer_rms_normalization</a>()</code> <br><code><a href="layer_rnn.html">layer_rnn</a>()</code> <br><code><a href="layer_separable_conv_1d.html">layer_separable_conv_1d</a>()</code> <br><code><a href="layer_separable_conv_2d.html">layer_separable_conv_2d</a>()</code> <br><code><a href="layer_simple_rnn.html">layer_simple_rnn</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br><code><a href="layer_spatial_dropout_1d.html">layer_spatial_dropout_1d</a>()</code> <br><code><a href="layer_spatial_dropout_2d.html">layer_spatial_dropout_2d</a>()</code> <br><code><a href="layer_spatial_dropout_3d.html">layer_spatial_dropout_3d</a>()</code> <br><code><a href="layer_spectral_normalization.html">layer_spectral_normalization</a>()</code> <br><code><a href="layer_stft_spectrogram.html">layer_stft_spectrogram</a>()</code> <br><code><a href="layer_string_lookup.html">layer_string_lookup</a>()</code> <br><code><a href="layer_subtract.html">layer_subtract</a>()</code> <br><code><a href="layer_text_vectorization.html">layer_text_vectorization</a>()</code> <br><code><a href="layer_tfsm.html">layer_tfsm</a>()</code> <br><code><a href="layer_time_distributed.html">layer_time_distributed</a>()</code> <br><code><a href="layer_torch_module_wrapper.html">layer_torch_module_wrapper</a>()</code> <br><code><a href="layer_unit_normalization.html">layer_unit_normalization</a>()</code> <br><code><a href="layer_upsampling_1d.html">layer_upsampling_1d</a>()</code> <br><code><a href="layer_upsampling_2d.html">layer_upsampling_2d</a>()</code> <br><code><a href="layer_upsampling_3d.html">layer_upsampling_3d</a>()</code> <br><code><a href="layer_zero_padding_1d.html">layer_zero_padding_1d</a>()</code> <br><code><a href="layer_zero_padding_2d.html">layer_zero_padding_2d</a>()</code> <br><code><a href="layer_zero_padding_3d.html">layer_zero_padding_3d</a>()</code> <br><code><a href="rnn_cell_gru.html">rnn_cell_gru</a>()</code> <br><code><a href="rnn_cell_lstm.html">rnn_cell_lstm</a>()</code> <br><code><a href="rnn_cell_simple.html">rnn_cell_simple</a>()</code> <br><code><a href="rnn_cells_stack.html">rnn_cells_stack</a>()</code> <br></p></div>
+<p>Other layers: <br><code><a href="Layer.html">Layer</a>()</code> <br><code><a href="layer_activation.html">layer_activation</a>()</code> <br><code><a href="layer_activation_elu.html">layer_activation_elu</a>()</code> <br><code><a href="layer_activation_leaky_relu.html">layer_activation_leaky_relu</a>()</code> <br><code><a href="layer_activation_parametric_relu.html">layer_activation_parametric_relu</a>()</code> <br><code><a href="layer_activation_relu.html">layer_activation_relu</a>()</code> <br><code><a href="layer_activation_softmax.html">layer_activation_softmax</a>()</code> <br><code><a href="layer_activity_regularization.html">layer_activity_regularization</a>()</code> <br><code><a href="layer_add.html">layer_add</a>()</code> <br><code><a href="layer_additive_attention.html">layer_additive_attention</a>()</code> <br><code><a href="layer_alpha_dropout.html">layer_alpha_dropout</a>()</code> <br><code><a href="layer_attention.html">layer_attention</a>()</code> <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_average.html">layer_average</a>()</code> <br><code><a href="layer_average_pooling_1d.html">layer_average_pooling_1d</a>()</code> <br><code><a href="layer_average_pooling_2d.html">layer_average_pooling_2d</a>()</code> <br><code><a href="layer_average_pooling_3d.html">layer_average_pooling_3d</a>()</code> <br><code><a href="layer_batch_normalization.html">layer_batch_normalization</a>()</code> <br><code><a href="layer_bidirectional.html">layer_bidirectional</a>()</code> <br><code><a href="layer_category_encoding.html">layer_category_encoding</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_concatenate.html">layer_concatenate</a>()</code> <br><code><a href="layer_conv_1d.html">layer_conv_1d</a>()</code> <br><code><a href="layer_conv_1d_transpose.html">layer_conv_1d_transpose</a>()</code> <br><code><a href="layer_conv_2d.html">layer_conv_2d</a>()</code> <br><code><a href="layer_conv_2d_transpose.html">layer_conv_2d_transpose</a>()</code> <br><code><a href="layer_conv_3d.html">layer_conv_3d</a>()</code> <br><code><a href="layer_conv_3d_transpose.html">layer_conv_3d_transpose</a>()</code> <br><code><a href="layer_conv_lstm_1d.html">layer_conv_lstm_1d</a>()</code> <br><code><a href="layer_conv_lstm_2d.html">layer_conv_lstm_2d</a>()</code> <br><code><a href="layer_conv_lstm_3d.html">layer_conv_lstm_3d</a>()</code> <br><code><a href="layer_cropping_1d.html">layer_cropping_1d</a>()</code> <br><code><a href="layer_cropping_2d.html">layer_cropping_2d</a>()</code> <br><code><a href="layer_cropping_3d.html">layer_cropping_3d</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_dense.html">layer_dense</a>()</code> <br><code><a href="layer_depthwise_conv_1d.html">layer_depthwise_conv_1d</a>()</code> <br><code><a href="layer_depthwise_conv_2d.html">layer_depthwise_conv_2d</a>()</code> <br><code><a href="layer_discretization.html">layer_discretization</a>()</code> <br><code><a href="layer_dot.html">layer_dot</a>()</code> <br><code><a href="layer_dropout.html">layer_dropout</a>()</code> <br><code><a href="layer_einsum_dense.html">layer_einsum_dense</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_feature_space.html">layer_feature_space</a>()</code> <br><code><a href="layer_flatten.html">layer_flatten</a>()</code> <br><code><a href="layer_flax_module_wrapper.html">layer_flax_module_wrapper</a>()</code> <br><code><a href="layer_gaussian_dropout.html">layer_gaussian_dropout</a>()</code> <br><code><a href="layer_gaussian_noise.html">layer_gaussian_noise</a>()</code> <br><code><a href="layer_global_average_pooling_1d.html">layer_global_average_pooling_1d</a>()</code> <br><code><a href="layer_global_average_pooling_2d.html">layer_global_average_pooling_2d</a>()</code> <br><code><a href="layer_global_average_pooling_3d.html">layer_global_average_pooling_3d</a>()</code> <br><code><a href="layer_global_max_pooling_1d.html">layer_global_max_pooling_1d</a>()</code> <br><code><a href="layer_global_max_pooling_2d.html">layer_global_max_pooling_2d</a>()</code> <br><code><a href="layer_global_max_pooling_3d.html">layer_global_max_pooling_3d</a>()</code> <br><code><a href="layer_group_normalization.html">layer_group_normalization</a>()</code> <br><code><a href="layer_group_query_attention.html">layer_group_query_attention</a>()</code> <br><code><a href="layer_gru.html">layer_gru</a>()</code> <br><code><a href="layer_hashed_crossing.html">layer_hashed_crossing</a>()</code> <br><code><a href="layer_hashing.html">layer_hashing</a>()</code> <br><code><a href="layer_identity.html">layer_identity</a>()</code> <br><code><a href="layer_integer_lookup.html">layer_integer_lookup</a>()</code> <br><code><a href="layer_jax_model_wrapper.html">layer_jax_model_wrapper</a>()</code> <br><code><a href="layer_lambda.html">layer_lambda</a>()</code> <br><code><a href="layer_layer_normalization.html">layer_layer_normalization</a>()</code> <br><code><a href="layer_lstm.html">layer_lstm</a>()</code> <br><code><a href="layer_masking.html">layer_masking</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_max_pooling_1d.html">layer_max_pooling_1d</a>()</code> <br><code><a href="layer_max_pooling_2d.html">layer_max_pooling_2d</a>()</code> <br><code><a href="layer_max_pooling_3d.html">layer_max_pooling_3d</a>()</code> <br><code><a href="layer_maximum.html">layer_maximum</a>()</code> <br><code><a href="layer_mel_spectrogram.html">layer_mel_spectrogram</a>()</code> <br><code><a href="layer_minimum.html">layer_minimum</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_multi_head_attention.html">layer_multi_head_attention</a>()</code> <br><code><a href="layer_multiply.html">layer_multiply</a>()</code> <br><code><a href="layer_normalization.html">layer_normalization</a>()</code> <br><code><a href="layer_permute.html">layer_permute</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_brightness.html">layer_random_brightness</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_contrast.html">layer_random_contrast</a>()</code> <br><code><a href="layer_random_crop.html">layer_random_crop</a>()</code> <br><code><a href="layer_random_elastic_transform.html">layer_random_elastic_transform</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_flip.html">layer_random_flip</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_rotation.html">layer_random_rotation</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_random_translation.html">layer_random_translation</a>()</code> <br><code><a href="layer_random_zoom.html">layer_random_zoom</a>()</code> <br><code><a href="layer_repeat_vector.html">layer_repeat_vector</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_reshape.html">layer_reshape</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_rms_normalization.html">layer_rms_normalization</a>()</code> <br><code><a href="layer_rnn.html">layer_rnn</a>()</code> <br><code><a href="layer_separable_conv_1d.html">layer_separable_conv_1d</a>()</code> <br><code><a href="layer_separable_conv_2d.html">layer_separable_conv_2d</a>()</code> <br><code><a href="layer_simple_rnn.html">layer_simple_rnn</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br><code><a href="layer_spatial_dropout_1d.html">layer_spatial_dropout_1d</a>()</code> <br><code><a href="layer_spatial_dropout_2d.html">layer_spatial_dropout_2d</a>()</code> <br><code><a href="layer_spatial_dropout_3d.html">layer_spatial_dropout_3d</a>()</code> <br><code><a href="layer_spectral_normalization.html">layer_spectral_normalization</a>()</code> <br><code><a href="layer_stft_spectrogram.html">layer_stft_spectrogram</a>()</code> <br><code><a href="layer_string_lookup.html">layer_string_lookup</a>()</code> <br><code><a href="layer_subtract.html">layer_subtract</a>()</code> <br><code><a href="layer_text_vectorization.html">layer_text_vectorization</a>()</code> <br><code><a href="layer_tfsm.html">layer_tfsm</a>()</code> <br><code><a href="layer_time_distributed.html">layer_time_distributed</a>()</code> <br><code><a href="layer_torch_module_wrapper.html">layer_torch_module_wrapper</a>()</code> <br><code><a href="layer_unit_normalization.html">layer_unit_normalization</a>()</code> <br><code><a href="layer_upsampling_1d.html">layer_upsampling_1d</a>()</code> <br><code><a href="layer_upsampling_2d.html">layer_upsampling_2d</a>()</code> <br><code><a href="layer_upsampling_3d.html">layer_upsampling_3d</a>()</code> <br><code><a href="layer_zero_padding_1d.html">layer_zero_padding_1d</a>()</code> <br><code><a href="layer_zero_padding_2d.html">layer_zero_padding_2d</a>()</code> <br><code><a href="layer_zero_padding_3d.html">layer_zero_padding_3d</a>()</code> <br><code><a href="rnn_cell_gru.html">rnn_cell_gru</a>()</code> <br><code><a href="rnn_cell_lstm.html">rnn_cell_lstm</a>()</code> <br><code><a href="rnn_cell_simple.html">rnn_cell_simple</a>()</code> <br><code><a href="rnn_cells_stack.html">rnn_cells_stack</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -215,7 +225,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/layer_equalization.html b/docs/dev/reference/layer_equalization.html
index 04e8535ab..218eb941b 100644
--- a/docs/dev/reference/layer_equalization.html
+++ b/docs/dev/reference/layer_equalization.html
@@ -60,7 +60,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -201,9 +201,9 @@ <h2 id="examples">Examples<a class="anchor" aria-label="anchor" href="#examples"
     </div>
     <div class="section level2">
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
-    <div class="dont-index"><p>Other image preprocessing layers: <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br></p>
-<p>Other preprocessing layers: <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_category_encoding.html">layer_category_encoding</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_discretization.html">layer_discretization</a>()</code> <br><code><a href="layer_feature_space.html">layer_feature_space</a>()</code> <br><code><a href="layer_hashed_crossing.html">layer_hashed_crossing</a>()</code> <br><code><a href="layer_hashing.html">layer_hashing</a>()</code> <br><code><a href="layer_integer_lookup.html">layer_integer_lookup</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_mel_spectrogram.html">layer_mel_spectrogram</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_normalization.html">layer_normalization</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_brightness.html">layer_random_brightness</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_contrast.html">layer_random_contrast</a>()</code> <br><code><a href="layer_random_crop.html">layer_random_crop</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_flip.html">layer_random_flip</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_rotation.html">layer_random_rotation</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_random_translation.html">layer_random_translation</a>()</code> <br><code><a href="layer_random_zoom.html">layer_random_zoom</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br><code><a href="layer_stft_spectrogram.html">layer_stft_spectrogram</a>()</code> <br><code><a href="layer_string_lookup.html">layer_string_lookup</a>()</code> <br><code><a href="layer_text_vectorization.html">layer_text_vectorization</a>()</code> <br></p>
-<p>Other layers: <br><code><a href="Layer.html">Layer</a>()</code> <br><code><a href="layer_activation.html">layer_activation</a>()</code> <br><code><a href="layer_activation_elu.html">layer_activation_elu</a>()</code> <br><code><a href="layer_activation_leaky_relu.html">layer_activation_leaky_relu</a>()</code> <br><code><a href="layer_activation_parametric_relu.html">layer_activation_parametric_relu</a>()</code> <br><code><a href="layer_activation_relu.html">layer_activation_relu</a>()</code> <br><code><a href="layer_activation_softmax.html">layer_activation_softmax</a>()</code> <br><code><a href="layer_activity_regularization.html">layer_activity_regularization</a>()</code> <br><code><a href="layer_add.html">layer_add</a>()</code> <br><code><a href="layer_additive_attention.html">layer_additive_attention</a>()</code> <br><code><a href="layer_alpha_dropout.html">layer_alpha_dropout</a>()</code> <br><code><a href="layer_attention.html">layer_attention</a>()</code> <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_average.html">layer_average</a>()</code> <br><code><a href="layer_average_pooling_1d.html">layer_average_pooling_1d</a>()</code> <br><code><a href="layer_average_pooling_2d.html">layer_average_pooling_2d</a>()</code> <br><code><a href="layer_average_pooling_3d.html">layer_average_pooling_3d</a>()</code> <br><code><a href="layer_batch_normalization.html">layer_batch_normalization</a>()</code> <br><code><a href="layer_bidirectional.html">layer_bidirectional</a>()</code> <br><code><a href="layer_category_encoding.html">layer_category_encoding</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_concatenate.html">layer_concatenate</a>()</code> <br><code><a href="layer_conv_1d.html">layer_conv_1d</a>()</code> <br><code><a href="layer_conv_1d_transpose.html">layer_conv_1d_transpose</a>()</code> <br><code><a href="layer_conv_2d.html">layer_conv_2d</a>()</code> <br><code><a href="layer_conv_2d_transpose.html">layer_conv_2d_transpose</a>()</code> <br><code><a href="layer_conv_3d.html">layer_conv_3d</a>()</code> <br><code><a href="layer_conv_3d_transpose.html">layer_conv_3d_transpose</a>()</code> <br><code><a href="layer_conv_lstm_1d.html">layer_conv_lstm_1d</a>()</code> <br><code><a href="layer_conv_lstm_2d.html">layer_conv_lstm_2d</a>()</code> <br><code><a href="layer_conv_lstm_3d.html">layer_conv_lstm_3d</a>()</code> <br><code><a href="layer_cropping_1d.html">layer_cropping_1d</a>()</code> <br><code><a href="layer_cropping_2d.html">layer_cropping_2d</a>()</code> <br><code><a href="layer_cropping_3d.html">layer_cropping_3d</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_dense.html">layer_dense</a>()</code> <br><code><a href="layer_depthwise_conv_1d.html">layer_depthwise_conv_1d</a>()</code> <br><code><a href="layer_depthwise_conv_2d.html">layer_depthwise_conv_2d</a>()</code> <br><code><a href="layer_discretization.html">layer_discretization</a>()</code> <br><code><a href="layer_dot.html">layer_dot</a>()</code> <br><code><a href="layer_dropout.html">layer_dropout</a>()</code> <br><code><a href="layer_einsum_dense.html">layer_einsum_dense</a>()</code> <br><code><a href="layer_embedding.html">layer_embedding</a>()</code> <br><code><a href="layer_feature_space.html">layer_feature_space</a>()</code> <br><code><a href="layer_flatten.html">layer_flatten</a>()</code> <br><code><a href="layer_flax_module_wrapper.html">layer_flax_module_wrapper</a>()</code> <br><code><a href="layer_gaussian_dropout.html">layer_gaussian_dropout</a>()</code> <br><code><a href="layer_gaussian_noise.html">layer_gaussian_noise</a>()</code> <br><code><a href="layer_global_average_pooling_1d.html">layer_global_average_pooling_1d</a>()</code> <br><code><a href="layer_global_average_pooling_2d.html">layer_global_average_pooling_2d</a>()</code> <br><code><a href="layer_global_average_pooling_3d.html">layer_global_average_pooling_3d</a>()</code> <br><code><a href="layer_global_max_pooling_1d.html">layer_global_max_pooling_1d</a>()</code> <br><code><a href="layer_global_max_pooling_2d.html">layer_global_max_pooling_2d</a>()</code> <br><code><a href="layer_global_max_pooling_3d.html">layer_global_max_pooling_3d</a>()</code> <br><code><a href="layer_group_normalization.html">layer_group_normalization</a>()</code> <br><code><a href="layer_group_query_attention.html">layer_group_query_attention</a>()</code> <br><code><a href="layer_gru.html">layer_gru</a>()</code> <br><code><a href="layer_hashed_crossing.html">layer_hashed_crossing</a>()</code> <br><code><a href="layer_hashing.html">layer_hashing</a>()</code> <br><code><a href="layer_identity.html">layer_identity</a>()</code> <br><code><a href="layer_integer_lookup.html">layer_integer_lookup</a>()</code> <br><code><a href="layer_jax_model_wrapper.html">layer_jax_model_wrapper</a>()</code> <br><code><a href="layer_lambda.html">layer_lambda</a>()</code> <br><code><a href="layer_layer_normalization.html">layer_layer_normalization</a>()</code> <br><code><a href="layer_lstm.html">layer_lstm</a>()</code> <br><code><a href="layer_masking.html">layer_masking</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_max_pooling_1d.html">layer_max_pooling_1d</a>()</code> <br><code><a href="layer_max_pooling_2d.html">layer_max_pooling_2d</a>()</code> <br><code><a href="layer_max_pooling_3d.html">layer_max_pooling_3d</a>()</code> <br><code><a href="layer_maximum.html">layer_maximum</a>()</code> <br><code><a href="layer_mel_spectrogram.html">layer_mel_spectrogram</a>()</code> <br><code><a href="layer_minimum.html">layer_minimum</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_multi_head_attention.html">layer_multi_head_attention</a>()</code> <br><code><a href="layer_multiply.html">layer_multiply</a>()</code> <br><code><a href="layer_normalization.html">layer_normalization</a>()</code> <br><code><a href="layer_permute.html">layer_permute</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_brightness.html">layer_random_brightness</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_contrast.html">layer_random_contrast</a>()</code> <br><code><a href="layer_random_crop.html">layer_random_crop</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_flip.html">layer_random_flip</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_rotation.html">layer_random_rotation</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_random_translation.html">layer_random_translation</a>()</code> <br><code><a href="layer_random_zoom.html">layer_random_zoom</a>()</code> <br><code><a href="layer_repeat_vector.html">layer_repeat_vector</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_reshape.html">layer_reshape</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_rms_normalization.html">layer_rms_normalization</a>()</code> <br><code><a href="layer_rnn.html">layer_rnn</a>()</code> <br><code><a href="layer_separable_conv_1d.html">layer_separable_conv_1d</a>()</code> <br><code><a href="layer_separable_conv_2d.html">layer_separable_conv_2d</a>()</code> <br><code><a href="layer_simple_rnn.html">layer_simple_rnn</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br><code><a href="layer_spatial_dropout_1d.html">layer_spatial_dropout_1d</a>()</code> <br><code><a href="layer_spatial_dropout_2d.html">layer_spatial_dropout_2d</a>()</code> <br><code><a href="layer_spatial_dropout_3d.html">layer_spatial_dropout_3d</a>()</code> <br><code><a href="layer_spectral_normalization.html">layer_spectral_normalization</a>()</code> <br><code><a href="layer_stft_spectrogram.html">layer_stft_spectrogram</a>()</code> <br><code><a href="layer_string_lookup.html">layer_string_lookup</a>()</code> <br><code><a href="layer_subtract.html">layer_subtract</a>()</code> <br><code><a href="layer_text_vectorization.html">layer_text_vectorization</a>()</code> <br><code><a href="layer_tfsm.html">layer_tfsm</a>()</code> <br><code><a href="layer_time_distributed.html">layer_time_distributed</a>()</code> <br><code><a href="layer_torch_module_wrapper.html">layer_torch_module_wrapper</a>()</code> <br><code><a href="layer_unit_normalization.html">layer_unit_normalization</a>()</code> <br><code><a href="layer_upsampling_1d.html">layer_upsampling_1d</a>()</code> <br><code><a href="layer_upsampling_2d.html">layer_upsampling_2d</a>()</code> <br><code><a href="layer_upsampling_3d.html">layer_upsampling_3d</a>()</code> <br><code><a href="layer_zero_padding_1d.html">layer_zero_padding_1d</a>()</code> <br><code><a href="layer_zero_padding_2d.html">layer_zero_padding_2d</a>()</code> <br><code><a href="layer_zero_padding_3d.html">layer_zero_padding_3d</a>()</code> <br><code><a href="rnn_cell_gru.html">rnn_cell_gru</a>()</code> <br><code><a href="rnn_cell_lstm.html">rnn_cell_lstm</a>()</code> <br><code><a href="rnn_cell_simple.html">rnn_cell_simple</a>()</code> <br><code><a href="rnn_cells_stack.html">rnn_cells_stack</a>()</code> <br></p></div>
+    <div class="dont-index"><p>Other image preprocessing layers: <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_elastic_transform.html">layer_random_elastic_transform</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br></p>
+<p>Other preprocessing layers: <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_category_encoding.html">layer_category_encoding</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_discretization.html">layer_discretization</a>()</code> <br><code><a href="layer_feature_space.html">layer_feature_space</a>()</code> <br><code><a href="layer_hashed_crossing.html">layer_hashed_crossing</a>()</code> <br><code><a href="layer_hashing.html">layer_hashing</a>()</code> <br><code><a href="layer_integer_lookup.html">layer_integer_lookup</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_mel_spectrogram.html">layer_mel_spectrogram</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_normalization.html">layer_normalization</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_brightness.html">layer_random_brightness</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_contrast.html">layer_random_contrast</a>()</code> <br><code><a href="layer_random_crop.html">layer_random_crop</a>()</code> <br><code><a href="layer_random_elastic_transform.html">layer_random_elastic_transform</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_flip.html">layer_random_flip</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_rotation.html">layer_random_rotation</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_random_translation.html">layer_random_translation</a>()</code> <br><code><a href="layer_random_zoom.html">layer_random_zoom</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br><code><a href="layer_stft_spectrogram.html">layer_stft_spectrogram</a>()</code> <br><code><a href="layer_string_lookup.html">layer_string_lookup</a>()</code> <br><code><a href="layer_text_vectorization.html">layer_text_vectorization</a>()</code> <br></p>
+<p>Other layers: <br><code><a href="Layer.html">Layer</a>()</code> <br><code><a href="layer_activation.html">layer_activation</a>()</code> <br><code><a href="layer_activation_elu.html">layer_activation_elu</a>()</code> <br><code><a href="layer_activation_leaky_relu.html">layer_activation_leaky_relu</a>()</code> <br><code><a href="layer_activation_parametric_relu.html">layer_activation_parametric_relu</a>()</code> <br><code><a href="layer_activation_relu.html">layer_activation_relu</a>()</code> <br><code><a href="layer_activation_softmax.html">layer_activation_softmax</a>()</code> <br><code><a href="layer_activity_regularization.html">layer_activity_regularization</a>()</code> <br><code><a href="layer_add.html">layer_add</a>()</code> <br><code><a href="layer_additive_attention.html">layer_additive_attention</a>()</code> <br><code><a href="layer_alpha_dropout.html">layer_alpha_dropout</a>()</code> <br><code><a href="layer_attention.html">layer_attention</a>()</code> <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_average.html">layer_average</a>()</code> <br><code><a href="layer_average_pooling_1d.html">layer_average_pooling_1d</a>()</code> <br><code><a href="layer_average_pooling_2d.html">layer_average_pooling_2d</a>()</code> <br><code><a href="layer_average_pooling_3d.html">layer_average_pooling_3d</a>()</code> <br><code><a href="layer_batch_normalization.html">layer_batch_normalization</a>()</code> <br><code><a href="layer_bidirectional.html">layer_bidirectional</a>()</code> <br><code><a href="layer_category_encoding.html">layer_category_encoding</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_concatenate.html">layer_concatenate</a>()</code> <br><code><a href="layer_conv_1d.html">layer_conv_1d</a>()</code> <br><code><a href="layer_conv_1d_transpose.html">layer_conv_1d_transpose</a>()</code> <br><code><a href="layer_conv_2d.html">layer_conv_2d</a>()</code> <br><code><a href="layer_conv_2d_transpose.html">layer_conv_2d_transpose</a>()</code> <br><code><a href="layer_conv_3d.html">layer_conv_3d</a>()</code> <br><code><a href="layer_conv_3d_transpose.html">layer_conv_3d_transpose</a>()</code> <br><code><a href="layer_conv_lstm_1d.html">layer_conv_lstm_1d</a>()</code> <br><code><a href="layer_conv_lstm_2d.html">layer_conv_lstm_2d</a>()</code> <br><code><a href="layer_conv_lstm_3d.html">layer_conv_lstm_3d</a>()</code> <br><code><a href="layer_cropping_1d.html">layer_cropping_1d</a>()</code> <br><code><a href="layer_cropping_2d.html">layer_cropping_2d</a>()</code> <br><code><a href="layer_cropping_3d.html">layer_cropping_3d</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_dense.html">layer_dense</a>()</code> <br><code><a href="layer_depthwise_conv_1d.html">layer_depthwise_conv_1d</a>()</code> <br><code><a href="layer_depthwise_conv_2d.html">layer_depthwise_conv_2d</a>()</code> <br><code><a href="layer_discretization.html">layer_discretization</a>()</code> <br><code><a href="layer_dot.html">layer_dot</a>()</code> <br><code><a href="layer_dropout.html">layer_dropout</a>()</code> <br><code><a href="layer_einsum_dense.html">layer_einsum_dense</a>()</code> <br><code><a href="layer_embedding.html">layer_embedding</a>()</code> <br><code><a href="layer_feature_space.html">layer_feature_space</a>()</code> <br><code><a href="layer_flatten.html">layer_flatten</a>()</code> <br><code><a href="layer_flax_module_wrapper.html">layer_flax_module_wrapper</a>()</code> <br><code><a href="layer_gaussian_dropout.html">layer_gaussian_dropout</a>()</code> <br><code><a href="layer_gaussian_noise.html">layer_gaussian_noise</a>()</code> <br><code><a href="layer_global_average_pooling_1d.html">layer_global_average_pooling_1d</a>()</code> <br><code><a href="layer_global_average_pooling_2d.html">layer_global_average_pooling_2d</a>()</code> <br><code><a href="layer_global_average_pooling_3d.html">layer_global_average_pooling_3d</a>()</code> <br><code><a href="layer_global_max_pooling_1d.html">layer_global_max_pooling_1d</a>()</code> <br><code><a href="layer_global_max_pooling_2d.html">layer_global_max_pooling_2d</a>()</code> <br><code><a href="layer_global_max_pooling_3d.html">layer_global_max_pooling_3d</a>()</code> <br><code><a href="layer_group_normalization.html">layer_group_normalization</a>()</code> <br><code><a href="layer_group_query_attention.html">layer_group_query_attention</a>()</code> <br><code><a href="layer_gru.html">layer_gru</a>()</code> <br><code><a href="layer_hashed_crossing.html">layer_hashed_crossing</a>()</code> <br><code><a href="layer_hashing.html">layer_hashing</a>()</code> <br><code><a href="layer_identity.html">layer_identity</a>()</code> <br><code><a href="layer_integer_lookup.html">layer_integer_lookup</a>()</code> <br><code><a href="layer_jax_model_wrapper.html">layer_jax_model_wrapper</a>()</code> <br><code><a href="layer_lambda.html">layer_lambda</a>()</code> <br><code><a href="layer_layer_normalization.html">layer_layer_normalization</a>()</code> <br><code><a href="layer_lstm.html">layer_lstm</a>()</code> <br><code><a href="layer_masking.html">layer_masking</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_max_pooling_1d.html">layer_max_pooling_1d</a>()</code> <br><code><a href="layer_max_pooling_2d.html">layer_max_pooling_2d</a>()</code> <br><code><a href="layer_max_pooling_3d.html">layer_max_pooling_3d</a>()</code> <br><code><a href="layer_maximum.html">layer_maximum</a>()</code> <br><code><a href="layer_mel_spectrogram.html">layer_mel_spectrogram</a>()</code> <br><code><a href="layer_minimum.html">layer_minimum</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_multi_head_attention.html">layer_multi_head_attention</a>()</code> <br><code><a href="layer_multiply.html">layer_multiply</a>()</code> <br><code><a href="layer_normalization.html">layer_normalization</a>()</code> <br><code><a href="layer_permute.html">layer_permute</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_brightness.html">layer_random_brightness</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_contrast.html">layer_random_contrast</a>()</code> <br><code><a href="layer_random_crop.html">layer_random_crop</a>()</code> <br><code><a href="layer_random_elastic_transform.html">layer_random_elastic_transform</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_flip.html">layer_random_flip</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_rotation.html">layer_random_rotation</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_random_translation.html">layer_random_translation</a>()</code> <br><code><a href="layer_random_zoom.html">layer_random_zoom</a>()</code> <br><code><a href="layer_repeat_vector.html">layer_repeat_vector</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_reshape.html">layer_reshape</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_rms_normalization.html">layer_rms_normalization</a>()</code> <br><code><a href="layer_rnn.html">layer_rnn</a>()</code> <br><code><a href="layer_separable_conv_1d.html">layer_separable_conv_1d</a>()</code> <br><code><a href="layer_separable_conv_2d.html">layer_separable_conv_2d</a>()</code> <br><code><a href="layer_simple_rnn.html">layer_simple_rnn</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br><code><a href="layer_spatial_dropout_1d.html">layer_spatial_dropout_1d</a>()</code> <br><code><a href="layer_spatial_dropout_2d.html">layer_spatial_dropout_2d</a>()</code> <br><code><a href="layer_spatial_dropout_3d.html">layer_spatial_dropout_3d</a>()</code> <br><code><a href="layer_spectral_normalization.html">layer_spectral_normalization</a>()</code> <br><code><a href="layer_stft_spectrogram.html">layer_stft_spectrogram</a>()</code> <br><code><a href="layer_string_lookup.html">layer_string_lookup</a>()</code> <br><code><a href="layer_subtract.html">layer_subtract</a>()</code> <br><code><a href="layer_text_vectorization.html">layer_text_vectorization</a>()</code> <br><code><a href="layer_tfsm.html">layer_tfsm</a>()</code> <br><code><a href="layer_time_distributed.html">layer_time_distributed</a>()</code> <br><code><a href="layer_torch_module_wrapper.html">layer_torch_module_wrapper</a>()</code> <br><code><a href="layer_unit_normalization.html">layer_unit_normalization</a>()</code> <br><code><a href="layer_upsampling_1d.html">layer_upsampling_1d</a>()</code> <br><code><a href="layer_upsampling_2d.html">layer_upsampling_2d</a>()</code> <br><code><a href="layer_upsampling_3d.html">layer_upsampling_3d</a>()</code> <br><code><a href="layer_zero_padding_1d.html">layer_zero_padding_1d</a>()</code> <br><code><a href="layer_zero_padding_2d.html">layer_zero_padding_2d</a>()</code> <br><code><a href="layer_zero_padding_3d.html">layer_zero_padding_3d</a>()</code> <br><code><a href="rnn_cell_gru.html">rnn_cell_gru</a>()</code> <br><code><a href="rnn_cell_lstm.html">rnn_cell_lstm</a>()</code> <br><code><a href="rnn_cell_simple.html">rnn_cell_simple</a>()</code> <br><code><a href="rnn_cells_stack.html">rnn_cells_stack</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -215,7 +215,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/layer_feature_space.html b/docs/dev/reference/layer_feature_space.html
index c477ba761..d777230c3 100644
--- a/docs/dev/reference/layer_feature_space.html
+++ b/docs/dev/reference/layer_feature_space.html
@@ -106,7 +106,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -457,8 +457,8 @@ <h2 id="examples">Examples<a class="anchor" aria-label="anchor" href="#examples"
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
     <div class="dont-index">
 <ul><li><p><a href="https://keras.io/api/utils/feature_space#featurespace-class" class="external-link">https://keras.io/api/utils/feature_space#featurespace-class</a></p></li>
-</ul><p>Other preprocessing layers: <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_category_encoding.html">layer_category_encoding</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_discretization.html">layer_discretization</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_hashed_crossing.html">layer_hashed_crossing</a>()</code> <br><code><a href="layer_hashing.html">layer_hashing</a>()</code> <br><code><a href="layer_integer_lookup.html">layer_integer_lookup</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_mel_spectrogram.html">layer_mel_spectrogram</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_normalization.html">layer_normalization</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_brightness.html">layer_random_brightness</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_contrast.html">layer_random_contrast</a>()</code> <br><code><a href="layer_random_crop.html">layer_random_crop</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_flip.html">layer_random_flip</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_rotation.html">layer_random_rotation</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_random_translation.html">layer_random_translation</a>()</code> <br><code><a href="layer_random_zoom.html">layer_random_zoom</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br><code><a href="layer_stft_spectrogram.html">layer_stft_spectrogram</a>()</code> <br><code><a href="layer_string_lookup.html">layer_string_lookup</a>()</code> <br><code><a href="layer_text_vectorization.html">layer_text_vectorization</a>()</code> <br></p>
-<p>Other layers: <br><code><a href="Layer.html">Layer</a>()</code> <br><code><a href="layer_activation.html">layer_activation</a>()</code> <br><code><a href="layer_activation_elu.html">layer_activation_elu</a>()</code> <br><code><a href="layer_activation_leaky_relu.html">layer_activation_leaky_relu</a>()</code> <br><code><a href="layer_activation_parametric_relu.html">layer_activation_parametric_relu</a>()</code> <br><code><a href="layer_activation_relu.html">layer_activation_relu</a>()</code> <br><code><a href="layer_activation_softmax.html">layer_activation_softmax</a>()</code> <br><code><a href="layer_activity_regularization.html">layer_activity_regularization</a>()</code> <br><code><a href="layer_add.html">layer_add</a>()</code> <br><code><a href="layer_additive_attention.html">layer_additive_attention</a>()</code> <br><code><a href="layer_alpha_dropout.html">layer_alpha_dropout</a>()</code> <br><code><a href="layer_attention.html">layer_attention</a>()</code> <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_average.html">layer_average</a>()</code> <br><code><a href="layer_average_pooling_1d.html">layer_average_pooling_1d</a>()</code> <br><code><a href="layer_average_pooling_2d.html">layer_average_pooling_2d</a>()</code> <br><code><a href="layer_average_pooling_3d.html">layer_average_pooling_3d</a>()</code> <br><code><a href="layer_batch_normalization.html">layer_batch_normalization</a>()</code> <br><code><a href="layer_bidirectional.html">layer_bidirectional</a>()</code> <br><code><a href="layer_category_encoding.html">layer_category_encoding</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_concatenate.html">layer_concatenate</a>()</code> <br><code><a href="layer_conv_1d.html">layer_conv_1d</a>()</code> <br><code><a href="layer_conv_1d_transpose.html">layer_conv_1d_transpose</a>()</code> <br><code><a href="layer_conv_2d.html">layer_conv_2d</a>()</code> <br><code><a href="layer_conv_2d_transpose.html">layer_conv_2d_transpose</a>()</code> <br><code><a href="layer_conv_3d.html">layer_conv_3d</a>()</code> <br><code><a href="layer_conv_3d_transpose.html">layer_conv_3d_transpose</a>()</code> <br><code><a href="layer_conv_lstm_1d.html">layer_conv_lstm_1d</a>()</code> <br><code><a href="layer_conv_lstm_2d.html">layer_conv_lstm_2d</a>()</code> <br><code><a href="layer_conv_lstm_3d.html">layer_conv_lstm_3d</a>()</code> <br><code><a href="layer_cropping_1d.html">layer_cropping_1d</a>()</code> <br><code><a href="layer_cropping_2d.html">layer_cropping_2d</a>()</code> <br><code><a href="layer_cropping_3d.html">layer_cropping_3d</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_dense.html">layer_dense</a>()</code> <br><code><a href="layer_depthwise_conv_1d.html">layer_depthwise_conv_1d</a>()</code> <br><code><a href="layer_depthwise_conv_2d.html">layer_depthwise_conv_2d</a>()</code> <br><code><a href="layer_discretization.html">layer_discretization</a>()</code> <br><code><a href="layer_dot.html">layer_dot</a>()</code> <br><code><a href="layer_dropout.html">layer_dropout</a>()</code> <br><code><a href="layer_einsum_dense.html">layer_einsum_dense</a>()</code> <br><code><a href="layer_embedding.html">layer_embedding</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_flatten.html">layer_flatten</a>()</code> <br><code><a href="layer_flax_module_wrapper.html">layer_flax_module_wrapper</a>()</code> <br><code><a href="layer_gaussian_dropout.html">layer_gaussian_dropout</a>()</code> <br><code><a href="layer_gaussian_noise.html">layer_gaussian_noise</a>()</code> <br><code><a href="layer_global_average_pooling_1d.html">layer_global_average_pooling_1d</a>()</code> <br><code><a href="layer_global_average_pooling_2d.html">layer_global_average_pooling_2d</a>()</code> <br><code><a href="layer_global_average_pooling_3d.html">layer_global_average_pooling_3d</a>()</code> <br><code><a href="layer_global_max_pooling_1d.html">layer_global_max_pooling_1d</a>()</code> <br><code><a href="layer_global_max_pooling_2d.html">layer_global_max_pooling_2d</a>()</code> <br><code><a href="layer_global_max_pooling_3d.html">layer_global_max_pooling_3d</a>()</code> <br><code><a href="layer_group_normalization.html">layer_group_normalization</a>()</code> <br><code><a href="layer_group_query_attention.html">layer_group_query_attention</a>()</code> <br><code><a href="layer_gru.html">layer_gru</a>()</code> <br><code><a href="layer_hashed_crossing.html">layer_hashed_crossing</a>()</code> <br><code><a href="layer_hashing.html">layer_hashing</a>()</code> <br><code><a href="layer_identity.html">layer_identity</a>()</code> <br><code><a href="layer_integer_lookup.html">layer_integer_lookup</a>()</code> <br><code><a href="layer_jax_model_wrapper.html">layer_jax_model_wrapper</a>()</code> <br><code><a href="layer_lambda.html">layer_lambda</a>()</code> <br><code><a href="layer_layer_normalization.html">layer_layer_normalization</a>()</code> <br><code><a href="layer_lstm.html">layer_lstm</a>()</code> <br><code><a href="layer_masking.html">layer_masking</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_max_pooling_1d.html">layer_max_pooling_1d</a>()</code> <br><code><a href="layer_max_pooling_2d.html">layer_max_pooling_2d</a>()</code> <br><code><a href="layer_max_pooling_3d.html">layer_max_pooling_3d</a>()</code> <br><code><a href="layer_maximum.html">layer_maximum</a>()</code> <br><code><a href="layer_mel_spectrogram.html">layer_mel_spectrogram</a>()</code> <br><code><a href="layer_minimum.html">layer_minimum</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_multi_head_attention.html">layer_multi_head_attention</a>()</code> <br><code><a href="layer_multiply.html">layer_multiply</a>()</code> <br><code><a href="layer_normalization.html">layer_normalization</a>()</code> <br><code><a href="layer_permute.html">layer_permute</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_brightness.html">layer_random_brightness</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_contrast.html">layer_random_contrast</a>()</code> <br><code><a href="layer_random_crop.html">layer_random_crop</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_flip.html">layer_random_flip</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_rotation.html">layer_random_rotation</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_random_translation.html">layer_random_translation</a>()</code> <br><code><a href="layer_random_zoom.html">layer_random_zoom</a>()</code> <br><code><a href="layer_repeat_vector.html">layer_repeat_vector</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_reshape.html">layer_reshape</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_rms_normalization.html">layer_rms_normalization</a>()</code> <br><code><a href="layer_rnn.html">layer_rnn</a>()</code> <br><code><a href="layer_separable_conv_1d.html">layer_separable_conv_1d</a>()</code> <br><code><a href="layer_separable_conv_2d.html">layer_separable_conv_2d</a>()</code> <br><code><a href="layer_simple_rnn.html">layer_simple_rnn</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br><code><a href="layer_spatial_dropout_1d.html">layer_spatial_dropout_1d</a>()</code> <br><code><a href="layer_spatial_dropout_2d.html">layer_spatial_dropout_2d</a>()</code> <br><code><a href="layer_spatial_dropout_3d.html">layer_spatial_dropout_3d</a>()</code> <br><code><a href="layer_spectral_normalization.html">layer_spectral_normalization</a>()</code> <br><code><a href="layer_stft_spectrogram.html">layer_stft_spectrogram</a>()</code> <br><code><a href="layer_string_lookup.html">layer_string_lookup</a>()</code> <br><code><a href="layer_subtract.html">layer_subtract</a>()</code> <br><code><a href="layer_text_vectorization.html">layer_text_vectorization</a>()</code> <br><code><a href="layer_tfsm.html">layer_tfsm</a>()</code> <br><code><a href="layer_time_distributed.html">layer_time_distributed</a>()</code> <br><code><a href="layer_torch_module_wrapper.html">layer_torch_module_wrapper</a>()</code> <br><code><a href="layer_unit_normalization.html">layer_unit_normalization</a>()</code> <br><code><a href="layer_upsampling_1d.html">layer_upsampling_1d</a>()</code> <br><code><a href="layer_upsampling_2d.html">layer_upsampling_2d</a>()</code> <br><code><a href="layer_upsampling_3d.html">layer_upsampling_3d</a>()</code> <br><code><a href="layer_zero_padding_1d.html">layer_zero_padding_1d</a>()</code> <br><code><a href="layer_zero_padding_2d.html">layer_zero_padding_2d</a>()</code> <br><code><a href="layer_zero_padding_3d.html">layer_zero_padding_3d</a>()</code> <br><code><a href="rnn_cell_gru.html">rnn_cell_gru</a>()</code> <br><code><a href="rnn_cell_lstm.html">rnn_cell_lstm</a>()</code> <br><code><a href="rnn_cell_simple.html">rnn_cell_simple</a>()</code> <br><code><a href="rnn_cells_stack.html">rnn_cells_stack</a>()</code> <br></p>
+</ul><p>Other preprocessing layers: <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_category_encoding.html">layer_category_encoding</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_discretization.html">layer_discretization</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_hashed_crossing.html">layer_hashed_crossing</a>()</code> <br><code><a href="layer_hashing.html">layer_hashing</a>()</code> <br><code><a href="layer_integer_lookup.html">layer_integer_lookup</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_mel_spectrogram.html">layer_mel_spectrogram</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_normalization.html">layer_normalization</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_brightness.html">layer_random_brightness</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_contrast.html">layer_random_contrast</a>()</code> <br><code><a href="layer_random_crop.html">layer_random_crop</a>()</code> <br><code><a href="layer_random_elastic_transform.html">layer_random_elastic_transform</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_flip.html">layer_random_flip</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_rotation.html">layer_random_rotation</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_random_translation.html">layer_random_translation</a>()</code> <br><code><a href="layer_random_zoom.html">layer_random_zoom</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br><code><a href="layer_stft_spectrogram.html">layer_stft_spectrogram</a>()</code> <br><code><a href="layer_string_lookup.html">layer_string_lookup</a>()</code> <br><code><a href="layer_text_vectorization.html">layer_text_vectorization</a>()</code> <br></p>
+<p>Other layers: <br><code><a href="Layer.html">Layer</a>()</code> <br><code><a href="layer_activation.html">layer_activation</a>()</code> <br><code><a href="layer_activation_elu.html">layer_activation_elu</a>()</code> <br><code><a href="layer_activation_leaky_relu.html">layer_activation_leaky_relu</a>()</code> <br><code><a href="layer_activation_parametric_relu.html">layer_activation_parametric_relu</a>()</code> <br><code><a href="layer_activation_relu.html">layer_activation_relu</a>()</code> <br><code><a href="layer_activation_softmax.html">layer_activation_softmax</a>()</code> <br><code><a href="layer_activity_regularization.html">layer_activity_regularization</a>()</code> <br><code><a href="layer_add.html">layer_add</a>()</code> <br><code><a href="layer_additive_attention.html">layer_additive_attention</a>()</code> <br><code><a href="layer_alpha_dropout.html">layer_alpha_dropout</a>()</code> <br><code><a href="layer_attention.html">layer_attention</a>()</code> <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_average.html">layer_average</a>()</code> <br><code><a href="layer_average_pooling_1d.html">layer_average_pooling_1d</a>()</code> <br><code><a href="layer_average_pooling_2d.html">layer_average_pooling_2d</a>()</code> <br><code><a href="layer_average_pooling_3d.html">layer_average_pooling_3d</a>()</code> <br><code><a href="layer_batch_normalization.html">layer_batch_normalization</a>()</code> <br><code><a href="layer_bidirectional.html">layer_bidirectional</a>()</code> <br><code><a href="layer_category_encoding.html">layer_category_encoding</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_concatenate.html">layer_concatenate</a>()</code> <br><code><a href="layer_conv_1d.html">layer_conv_1d</a>()</code> <br><code><a href="layer_conv_1d_transpose.html">layer_conv_1d_transpose</a>()</code> <br><code><a href="layer_conv_2d.html">layer_conv_2d</a>()</code> <br><code><a href="layer_conv_2d_transpose.html">layer_conv_2d_transpose</a>()</code> <br><code><a href="layer_conv_3d.html">layer_conv_3d</a>()</code> <br><code><a href="layer_conv_3d_transpose.html">layer_conv_3d_transpose</a>()</code> <br><code><a href="layer_conv_lstm_1d.html">layer_conv_lstm_1d</a>()</code> <br><code><a href="layer_conv_lstm_2d.html">layer_conv_lstm_2d</a>()</code> <br><code><a href="layer_conv_lstm_3d.html">layer_conv_lstm_3d</a>()</code> <br><code><a href="layer_cropping_1d.html">layer_cropping_1d</a>()</code> <br><code><a href="layer_cropping_2d.html">layer_cropping_2d</a>()</code> <br><code><a href="layer_cropping_3d.html">layer_cropping_3d</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_dense.html">layer_dense</a>()</code> <br><code><a href="layer_depthwise_conv_1d.html">layer_depthwise_conv_1d</a>()</code> <br><code><a href="layer_depthwise_conv_2d.html">layer_depthwise_conv_2d</a>()</code> <br><code><a href="layer_discretization.html">layer_discretization</a>()</code> <br><code><a href="layer_dot.html">layer_dot</a>()</code> <br><code><a href="layer_dropout.html">layer_dropout</a>()</code> <br><code><a href="layer_einsum_dense.html">layer_einsum_dense</a>()</code> <br><code><a href="layer_embedding.html">layer_embedding</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_flatten.html">layer_flatten</a>()</code> <br><code><a href="layer_flax_module_wrapper.html">layer_flax_module_wrapper</a>()</code> <br><code><a href="layer_gaussian_dropout.html">layer_gaussian_dropout</a>()</code> <br><code><a href="layer_gaussian_noise.html">layer_gaussian_noise</a>()</code> <br><code><a href="layer_global_average_pooling_1d.html">layer_global_average_pooling_1d</a>()</code> <br><code><a href="layer_global_average_pooling_2d.html">layer_global_average_pooling_2d</a>()</code> <br><code><a href="layer_global_average_pooling_3d.html">layer_global_average_pooling_3d</a>()</code> <br><code><a href="layer_global_max_pooling_1d.html">layer_global_max_pooling_1d</a>()</code> <br><code><a href="layer_global_max_pooling_2d.html">layer_global_max_pooling_2d</a>()</code> <br><code><a href="layer_global_max_pooling_3d.html">layer_global_max_pooling_3d</a>()</code> <br><code><a href="layer_group_normalization.html">layer_group_normalization</a>()</code> <br><code><a href="layer_group_query_attention.html">layer_group_query_attention</a>()</code> <br><code><a href="layer_gru.html">layer_gru</a>()</code> <br><code><a href="layer_hashed_crossing.html">layer_hashed_crossing</a>()</code> <br><code><a href="layer_hashing.html">layer_hashing</a>()</code> <br><code><a href="layer_identity.html">layer_identity</a>()</code> <br><code><a href="layer_integer_lookup.html">layer_integer_lookup</a>()</code> <br><code><a href="layer_jax_model_wrapper.html">layer_jax_model_wrapper</a>()</code> <br><code><a href="layer_lambda.html">layer_lambda</a>()</code> <br><code><a href="layer_layer_normalization.html">layer_layer_normalization</a>()</code> <br><code><a href="layer_lstm.html">layer_lstm</a>()</code> <br><code><a href="layer_masking.html">layer_masking</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_max_pooling_1d.html">layer_max_pooling_1d</a>()</code> <br><code><a href="layer_max_pooling_2d.html">layer_max_pooling_2d</a>()</code> <br><code><a href="layer_max_pooling_3d.html">layer_max_pooling_3d</a>()</code> <br><code><a href="layer_maximum.html">layer_maximum</a>()</code> <br><code><a href="layer_mel_spectrogram.html">layer_mel_spectrogram</a>()</code> <br><code><a href="layer_minimum.html">layer_minimum</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_multi_head_attention.html">layer_multi_head_attention</a>()</code> <br><code><a href="layer_multiply.html">layer_multiply</a>()</code> <br><code><a href="layer_normalization.html">layer_normalization</a>()</code> <br><code><a href="layer_permute.html">layer_permute</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_brightness.html">layer_random_brightness</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_contrast.html">layer_random_contrast</a>()</code> <br><code><a href="layer_random_crop.html">layer_random_crop</a>()</code> <br><code><a href="layer_random_elastic_transform.html">layer_random_elastic_transform</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_flip.html">layer_random_flip</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_rotation.html">layer_random_rotation</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_random_translation.html">layer_random_translation</a>()</code> <br><code><a href="layer_random_zoom.html">layer_random_zoom</a>()</code> <br><code><a href="layer_repeat_vector.html">layer_repeat_vector</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_reshape.html">layer_reshape</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_rms_normalization.html">layer_rms_normalization</a>()</code> <br><code><a href="layer_rnn.html">layer_rnn</a>()</code> <br><code><a href="layer_separable_conv_1d.html">layer_separable_conv_1d</a>()</code> <br><code><a href="layer_separable_conv_2d.html">layer_separable_conv_2d</a>()</code> <br><code><a href="layer_simple_rnn.html">layer_simple_rnn</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br><code><a href="layer_spatial_dropout_1d.html">layer_spatial_dropout_1d</a>()</code> <br><code><a href="layer_spatial_dropout_2d.html">layer_spatial_dropout_2d</a>()</code> <br><code><a href="layer_spatial_dropout_3d.html">layer_spatial_dropout_3d</a>()</code> <br><code><a href="layer_spectral_normalization.html">layer_spectral_normalization</a>()</code> <br><code><a href="layer_stft_spectrogram.html">layer_stft_spectrogram</a>()</code> <br><code><a href="layer_string_lookup.html">layer_string_lookup</a>()</code> <br><code><a href="layer_subtract.html">layer_subtract</a>()</code> <br><code><a href="layer_text_vectorization.html">layer_text_vectorization</a>()</code> <br><code><a href="layer_tfsm.html">layer_tfsm</a>()</code> <br><code><a href="layer_time_distributed.html">layer_time_distributed</a>()</code> <br><code><a href="layer_torch_module_wrapper.html">layer_torch_module_wrapper</a>()</code> <br><code><a href="layer_unit_normalization.html">layer_unit_normalization</a>()</code> <br><code><a href="layer_upsampling_1d.html">layer_upsampling_1d</a>()</code> <br><code><a href="layer_upsampling_2d.html">layer_upsampling_2d</a>()</code> <br><code><a href="layer_upsampling_3d.html">layer_upsampling_3d</a>()</code> <br><code><a href="layer_zero_padding_1d.html">layer_zero_padding_1d</a>()</code> <br><code><a href="layer_zero_padding_2d.html">layer_zero_padding_2d</a>()</code> <br><code><a href="layer_zero_padding_3d.html">layer_zero_padding_3d</a>()</code> <br><code><a href="rnn_cell_gru.html">rnn_cell_gru</a>()</code> <br><code><a href="rnn_cell_lstm.html">rnn_cell_lstm</a>()</code> <br><code><a href="rnn_cell_simple.html">rnn_cell_simple</a>()</code> <br><code><a href="rnn_cells_stack.html">rnn_cells_stack</a>()</code> <br></p>
 <p>Other utils: <br><code><a href="audio_dataset_from_directory.html">audio_dataset_from_directory</a>()</code> <br><code><a href="clear_session.html">clear_session</a>()</code> <br><code><a href="config_disable_interactive_logging.html">config_disable_interactive_logging</a>()</code> <br><code><a href="config_disable_traceback_filtering.html">config_disable_traceback_filtering</a>()</code> <br><code><a href="config_enable_interactive_logging.html">config_enable_interactive_logging</a>()</code> <br><code><a href="config_enable_traceback_filtering.html">config_enable_traceback_filtering</a>()</code> <br><code><a href="config_is_interactive_logging_enabled.html">config_is_interactive_logging_enabled</a>()</code> <br><code><a href="config_is_traceback_filtering_enabled.html">config_is_traceback_filtering_enabled</a>()</code> <br><code><a href="get_file.html">get_file</a>()</code> <br><code><a href="get_source_inputs.html">get_source_inputs</a>()</code> <br><code><a href="image_array_save.html">image_array_save</a>()</code> <br><code><a href="image_dataset_from_directory.html">image_dataset_from_directory</a>()</code> <br><code><a href="image_from_array.html">image_from_array</a>()</code> <br><code><a href="image_load.html">image_load</a>()</code> <br><code><a href="image_smart_resize.html">image_smart_resize</a>()</code> <br><code><a href="image_to_array.html">image_to_array</a>()</code> <br><code><a href="normalize.html">normalize</a>()</code> <br><code><a href="pad_sequences.html">pad_sequences</a>()</code> <br><code><a href="set_random_seed.html">set_random_seed</a>()</code> <br><code><a href="split_dataset.html">split_dataset</a>()</code> <br><code><a href="text_dataset_from_directory.html">text_dataset_from_directory</a>()</code> <br><code><a href="timeseries_dataset_from_array.html">timeseries_dataset_from_array</a>()</code> <br><code><a href="to_categorical.html">to_categorical</a>()</code> <br><code><a href="zip_lists.html">zip_lists</a>()</code> <br></p></div>
     </div>
 
@@ -471,7 +471,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/layer_flatten.html b/docs/dev/reference/layer_flatten.html
index 8ed74a29a..e3e52b7e9 100644
--- a/docs/dev/reference/layer_flatten.html
+++ b/docs/dev/reference/layer_flatten.html
@@ -8,7 +8,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -112,7 +112,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
     <div class="dont-index">
 <ul><li><p><a href="https://keras.io/api/layers/reshaping_layers/flatten#flatten-class" class="external-link">https://keras.io/api/layers/reshaping_layers/flatten#flatten-class</a></p></li>
 </ul><p>Other reshaping layers: <br><code><a href="layer_cropping_1d.html">layer_cropping_1d</a>()</code> <br><code><a href="layer_cropping_2d.html">layer_cropping_2d</a>()</code> <br><code><a href="layer_cropping_3d.html">layer_cropping_3d</a>()</code> <br><code><a href="layer_permute.html">layer_permute</a>()</code> <br><code><a href="layer_repeat_vector.html">layer_repeat_vector</a>()</code> <br><code><a href="layer_reshape.html">layer_reshape</a>()</code> <br><code><a href="layer_upsampling_1d.html">layer_upsampling_1d</a>()</code> <br><code><a href="layer_upsampling_2d.html">layer_upsampling_2d</a>()</code> <br><code><a href="layer_upsampling_3d.html">layer_upsampling_3d</a>()</code> <br><code><a href="layer_zero_padding_1d.html">layer_zero_padding_1d</a>()</code> <br><code><a href="layer_zero_padding_2d.html">layer_zero_padding_2d</a>()</code> <br><code><a href="layer_zero_padding_3d.html">layer_zero_padding_3d</a>()</code> <br></p>
-<p>Other layers: <br><code><a href="Layer.html">Layer</a>()</code> <br><code><a href="layer_activation.html">layer_activation</a>()</code> <br><code><a href="layer_activation_elu.html">layer_activation_elu</a>()</code> <br><code><a href="layer_activation_leaky_relu.html">layer_activation_leaky_relu</a>()</code> <br><code><a href="layer_activation_parametric_relu.html">layer_activation_parametric_relu</a>()</code> <br><code><a href="layer_activation_relu.html">layer_activation_relu</a>()</code> <br><code><a href="layer_activation_softmax.html">layer_activation_softmax</a>()</code> <br><code><a href="layer_activity_regularization.html">layer_activity_regularization</a>()</code> <br><code><a href="layer_add.html">layer_add</a>()</code> <br><code><a href="layer_additive_attention.html">layer_additive_attention</a>()</code> <br><code><a href="layer_alpha_dropout.html">layer_alpha_dropout</a>()</code> <br><code><a href="layer_attention.html">layer_attention</a>()</code> <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_average.html">layer_average</a>()</code> <br><code><a href="layer_average_pooling_1d.html">layer_average_pooling_1d</a>()</code> <br><code><a href="layer_average_pooling_2d.html">layer_average_pooling_2d</a>()</code> <br><code><a href="layer_average_pooling_3d.html">layer_average_pooling_3d</a>()</code> <br><code><a href="layer_batch_normalization.html">layer_batch_normalization</a>()</code> <br><code><a href="layer_bidirectional.html">layer_bidirectional</a>()</code> <br><code><a href="layer_category_encoding.html">layer_category_encoding</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_concatenate.html">layer_concatenate</a>()</code> <br><code><a href="layer_conv_1d.html">layer_conv_1d</a>()</code> <br><code><a href="layer_conv_1d_transpose.html">layer_conv_1d_transpose</a>()</code> <br><code><a href="layer_conv_2d.html">layer_conv_2d</a>()</code> <br><code><a href="layer_conv_2d_transpose.html">layer_conv_2d_transpose</a>()</code> <br><code><a href="layer_conv_3d.html">layer_conv_3d</a>()</code> <br><code><a href="layer_conv_3d_transpose.html">layer_conv_3d_transpose</a>()</code> <br><code><a href="layer_conv_lstm_1d.html">layer_conv_lstm_1d</a>()</code> <br><code><a href="layer_conv_lstm_2d.html">layer_conv_lstm_2d</a>()</code> <br><code><a href="layer_conv_lstm_3d.html">layer_conv_lstm_3d</a>()</code> <br><code><a href="layer_cropping_1d.html">layer_cropping_1d</a>()</code> <br><code><a href="layer_cropping_2d.html">layer_cropping_2d</a>()</code> <br><code><a href="layer_cropping_3d.html">layer_cropping_3d</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_dense.html">layer_dense</a>()</code> <br><code><a href="layer_depthwise_conv_1d.html">layer_depthwise_conv_1d</a>()</code> <br><code><a href="layer_depthwise_conv_2d.html">layer_depthwise_conv_2d</a>()</code> <br><code><a href="layer_discretization.html">layer_discretization</a>()</code> <br><code><a href="layer_dot.html">layer_dot</a>()</code> <br><code><a href="layer_dropout.html">layer_dropout</a>()</code> <br><code><a href="layer_einsum_dense.html">layer_einsum_dense</a>()</code> <br><code><a href="layer_embedding.html">layer_embedding</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_feature_space.html">layer_feature_space</a>()</code> <br><code><a href="layer_flax_module_wrapper.html">layer_flax_module_wrapper</a>()</code> <br><code><a href="layer_gaussian_dropout.html">layer_gaussian_dropout</a>()</code> <br><code><a href="layer_gaussian_noise.html">layer_gaussian_noise</a>()</code> <br><code><a href="layer_global_average_pooling_1d.html">layer_global_average_pooling_1d</a>()</code> <br><code><a href="layer_global_average_pooling_2d.html">layer_global_average_pooling_2d</a>()</code> <br><code><a href="layer_global_average_pooling_3d.html">layer_global_average_pooling_3d</a>()</code> <br><code><a href="layer_global_max_pooling_1d.html">layer_global_max_pooling_1d</a>()</code> <br><code><a href="layer_global_max_pooling_2d.html">layer_global_max_pooling_2d</a>()</code> <br><code><a href="layer_global_max_pooling_3d.html">layer_global_max_pooling_3d</a>()</code> <br><code><a href="layer_group_normalization.html">layer_group_normalization</a>()</code> <br><code><a href="layer_group_query_attention.html">layer_group_query_attention</a>()</code> <br><code><a href="layer_gru.html">layer_gru</a>()</code> <br><code><a href="layer_hashed_crossing.html">layer_hashed_crossing</a>()</code> <br><code><a href="layer_hashing.html">layer_hashing</a>()</code> <br><code><a href="layer_identity.html">layer_identity</a>()</code> <br><code><a href="layer_integer_lookup.html">layer_integer_lookup</a>()</code> <br><code><a href="layer_jax_model_wrapper.html">layer_jax_model_wrapper</a>()</code> <br><code><a href="layer_lambda.html">layer_lambda</a>()</code> <br><code><a href="layer_layer_normalization.html">layer_layer_normalization</a>()</code> <br><code><a href="layer_lstm.html">layer_lstm</a>()</code> <br><code><a href="layer_masking.html">layer_masking</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_max_pooling_1d.html">layer_max_pooling_1d</a>()</code> <br><code><a href="layer_max_pooling_2d.html">layer_max_pooling_2d</a>()</code> <br><code><a href="layer_max_pooling_3d.html">layer_max_pooling_3d</a>()</code> <br><code><a href="layer_maximum.html">layer_maximum</a>()</code> <br><code><a href="layer_mel_spectrogram.html">layer_mel_spectrogram</a>()</code> <br><code><a href="layer_minimum.html">layer_minimum</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_multi_head_attention.html">layer_multi_head_attention</a>()</code> <br><code><a href="layer_multiply.html">layer_multiply</a>()</code> <br><code><a href="layer_normalization.html">layer_normalization</a>()</code> <br><code><a href="layer_permute.html">layer_permute</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_brightness.html">layer_random_brightness</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_contrast.html">layer_random_contrast</a>()</code> <br><code><a href="layer_random_crop.html">layer_random_crop</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_flip.html">layer_random_flip</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_rotation.html">layer_random_rotation</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_random_translation.html">layer_random_translation</a>()</code> <br><code><a href="layer_random_zoom.html">layer_random_zoom</a>()</code> <br><code><a href="layer_repeat_vector.html">layer_repeat_vector</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_reshape.html">layer_reshape</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_rms_normalization.html">layer_rms_normalization</a>()</code> <br><code><a href="layer_rnn.html">layer_rnn</a>()</code> <br><code><a href="layer_separable_conv_1d.html">layer_separable_conv_1d</a>()</code> <br><code><a href="layer_separable_conv_2d.html">layer_separable_conv_2d</a>()</code> <br><code><a href="layer_simple_rnn.html">layer_simple_rnn</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br><code><a href="layer_spatial_dropout_1d.html">layer_spatial_dropout_1d</a>()</code> <br><code><a href="layer_spatial_dropout_2d.html">layer_spatial_dropout_2d</a>()</code> <br><code><a href="layer_spatial_dropout_3d.html">layer_spatial_dropout_3d</a>()</code> <br><code><a href="layer_spectral_normalization.html">layer_spectral_normalization</a>()</code> <br><code><a href="layer_stft_spectrogram.html">layer_stft_spectrogram</a>()</code> <br><code><a href="layer_string_lookup.html">layer_string_lookup</a>()</code> <br><code><a href="layer_subtract.html">layer_subtract</a>()</code> <br><code><a href="layer_text_vectorization.html">layer_text_vectorization</a>()</code> <br><code><a href="layer_tfsm.html">layer_tfsm</a>()</code> <br><code><a href="layer_time_distributed.html">layer_time_distributed</a>()</code> <br><code><a href="layer_torch_module_wrapper.html">layer_torch_module_wrapper</a>()</code> <br><code><a href="layer_unit_normalization.html">layer_unit_normalization</a>()</code> <br><code><a href="layer_upsampling_1d.html">layer_upsampling_1d</a>()</code> <br><code><a href="layer_upsampling_2d.html">layer_upsampling_2d</a>()</code> <br><code><a href="layer_upsampling_3d.html">layer_upsampling_3d</a>()</code> <br><code><a href="layer_zero_padding_1d.html">layer_zero_padding_1d</a>()</code> <br><code><a href="layer_zero_padding_2d.html">layer_zero_padding_2d</a>()</code> <br><code><a href="layer_zero_padding_3d.html">layer_zero_padding_3d</a>()</code> <br><code><a href="rnn_cell_gru.html">rnn_cell_gru</a>()</code> <br><code><a href="rnn_cell_lstm.html">rnn_cell_lstm</a>()</code> <br><code><a href="rnn_cell_simple.html">rnn_cell_simple</a>()</code> <br><code><a href="rnn_cells_stack.html">rnn_cells_stack</a>()</code> <br></p></div>
+<p>Other layers: <br><code><a href="Layer.html">Layer</a>()</code> <br><code><a href="layer_activation.html">layer_activation</a>()</code> <br><code><a href="layer_activation_elu.html">layer_activation_elu</a>()</code> <br><code><a href="layer_activation_leaky_relu.html">layer_activation_leaky_relu</a>()</code> <br><code><a href="layer_activation_parametric_relu.html">layer_activation_parametric_relu</a>()</code> <br><code><a href="layer_activation_relu.html">layer_activation_relu</a>()</code> <br><code><a href="layer_activation_softmax.html">layer_activation_softmax</a>()</code> <br><code><a href="layer_activity_regularization.html">layer_activity_regularization</a>()</code> <br><code><a href="layer_add.html">layer_add</a>()</code> <br><code><a href="layer_additive_attention.html">layer_additive_attention</a>()</code> <br><code><a href="layer_alpha_dropout.html">layer_alpha_dropout</a>()</code> <br><code><a href="layer_attention.html">layer_attention</a>()</code> <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_average.html">layer_average</a>()</code> <br><code><a href="layer_average_pooling_1d.html">layer_average_pooling_1d</a>()</code> <br><code><a href="layer_average_pooling_2d.html">layer_average_pooling_2d</a>()</code> <br><code><a href="layer_average_pooling_3d.html">layer_average_pooling_3d</a>()</code> <br><code><a href="layer_batch_normalization.html">layer_batch_normalization</a>()</code> <br><code><a href="layer_bidirectional.html">layer_bidirectional</a>()</code> <br><code><a href="layer_category_encoding.html">layer_category_encoding</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_concatenate.html">layer_concatenate</a>()</code> <br><code><a href="layer_conv_1d.html">layer_conv_1d</a>()</code> <br><code><a href="layer_conv_1d_transpose.html">layer_conv_1d_transpose</a>()</code> <br><code><a href="layer_conv_2d.html">layer_conv_2d</a>()</code> <br><code><a href="layer_conv_2d_transpose.html">layer_conv_2d_transpose</a>()</code> <br><code><a href="layer_conv_3d.html">layer_conv_3d</a>()</code> <br><code><a href="layer_conv_3d_transpose.html">layer_conv_3d_transpose</a>()</code> <br><code><a href="layer_conv_lstm_1d.html">layer_conv_lstm_1d</a>()</code> <br><code><a href="layer_conv_lstm_2d.html">layer_conv_lstm_2d</a>()</code> <br><code><a href="layer_conv_lstm_3d.html">layer_conv_lstm_3d</a>()</code> <br><code><a href="layer_cropping_1d.html">layer_cropping_1d</a>()</code> <br><code><a href="layer_cropping_2d.html">layer_cropping_2d</a>()</code> <br><code><a href="layer_cropping_3d.html">layer_cropping_3d</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_dense.html">layer_dense</a>()</code> <br><code><a href="layer_depthwise_conv_1d.html">layer_depthwise_conv_1d</a>()</code> <br><code><a href="layer_depthwise_conv_2d.html">layer_depthwise_conv_2d</a>()</code> <br><code><a href="layer_discretization.html">layer_discretization</a>()</code> <br><code><a href="layer_dot.html">layer_dot</a>()</code> <br><code><a href="layer_dropout.html">layer_dropout</a>()</code> <br><code><a href="layer_einsum_dense.html">layer_einsum_dense</a>()</code> <br><code><a href="layer_embedding.html">layer_embedding</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_feature_space.html">layer_feature_space</a>()</code> <br><code><a href="layer_flax_module_wrapper.html">layer_flax_module_wrapper</a>()</code> <br><code><a href="layer_gaussian_dropout.html">layer_gaussian_dropout</a>()</code> <br><code><a href="layer_gaussian_noise.html">layer_gaussian_noise</a>()</code> <br><code><a href="layer_global_average_pooling_1d.html">layer_global_average_pooling_1d</a>()</code> <br><code><a href="layer_global_average_pooling_2d.html">layer_global_average_pooling_2d</a>()</code> <br><code><a href="layer_global_average_pooling_3d.html">layer_global_average_pooling_3d</a>()</code> <br><code><a href="layer_global_max_pooling_1d.html">layer_global_max_pooling_1d</a>()</code> <br><code><a href="layer_global_max_pooling_2d.html">layer_global_max_pooling_2d</a>()</code> <br><code><a href="layer_global_max_pooling_3d.html">layer_global_max_pooling_3d</a>()</code> <br><code><a href="layer_group_normalization.html">layer_group_normalization</a>()</code> <br><code><a href="layer_group_query_attention.html">layer_group_query_attention</a>()</code> <br><code><a href="layer_gru.html">layer_gru</a>()</code> <br><code><a href="layer_hashed_crossing.html">layer_hashed_crossing</a>()</code> <br><code><a href="layer_hashing.html">layer_hashing</a>()</code> <br><code><a href="layer_identity.html">layer_identity</a>()</code> <br><code><a href="layer_integer_lookup.html">layer_integer_lookup</a>()</code> <br><code><a href="layer_jax_model_wrapper.html">layer_jax_model_wrapper</a>()</code> <br><code><a href="layer_lambda.html">layer_lambda</a>()</code> <br><code><a href="layer_layer_normalization.html">layer_layer_normalization</a>()</code> <br><code><a href="layer_lstm.html">layer_lstm</a>()</code> <br><code><a href="layer_masking.html">layer_masking</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_max_pooling_1d.html">layer_max_pooling_1d</a>()</code> <br><code><a href="layer_max_pooling_2d.html">layer_max_pooling_2d</a>()</code> <br><code><a href="layer_max_pooling_3d.html">layer_max_pooling_3d</a>()</code> <br><code><a href="layer_maximum.html">layer_maximum</a>()</code> <br><code><a href="layer_mel_spectrogram.html">layer_mel_spectrogram</a>()</code> <br><code><a href="layer_minimum.html">layer_minimum</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_multi_head_attention.html">layer_multi_head_attention</a>()</code> <br><code><a href="layer_multiply.html">layer_multiply</a>()</code> <br><code><a href="layer_normalization.html">layer_normalization</a>()</code> <br><code><a href="layer_permute.html">layer_permute</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_brightness.html">layer_random_brightness</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_contrast.html">layer_random_contrast</a>()</code> <br><code><a href="layer_random_crop.html">layer_random_crop</a>()</code> <br><code><a href="layer_random_elastic_transform.html">layer_random_elastic_transform</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_flip.html">layer_random_flip</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_rotation.html">layer_random_rotation</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_random_translation.html">layer_random_translation</a>()</code> <br><code><a href="layer_random_zoom.html">layer_random_zoom</a>()</code> <br><code><a href="layer_repeat_vector.html">layer_repeat_vector</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_reshape.html">layer_reshape</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_rms_normalization.html">layer_rms_normalization</a>()</code> <br><code><a href="layer_rnn.html">layer_rnn</a>()</code> <br><code><a href="layer_separable_conv_1d.html">layer_separable_conv_1d</a>()</code> <br><code><a href="layer_separable_conv_2d.html">layer_separable_conv_2d</a>()</code> <br><code><a href="layer_simple_rnn.html">layer_simple_rnn</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br><code><a href="layer_spatial_dropout_1d.html">layer_spatial_dropout_1d</a>()</code> <br><code><a href="layer_spatial_dropout_2d.html">layer_spatial_dropout_2d</a>()</code> <br><code><a href="layer_spatial_dropout_3d.html">layer_spatial_dropout_3d</a>()</code> <br><code><a href="layer_spectral_normalization.html">layer_spectral_normalization</a>()</code> <br><code><a href="layer_stft_spectrogram.html">layer_stft_spectrogram</a>()</code> <br><code><a href="layer_string_lookup.html">layer_string_lookup</a>()</code> <br><code><a href="layer_subtract.html">layer_subtract</a>()</code> <br><code><a href="layer_text_vectorization.html">layer_text_vectorization</a>()</code> <br><code><a href="layer_tfsm.html">layer_tfsm</a>()</code> <br><code><a href="layer_time_distributed.html">layer_time_distributed</a>()</code> <br><code><a href="layer_torch_module_wrapper.html">layer_torch_module_wrapper</a>()</code> <br><code><a href="layer_unit_normalization.html">layer_unit_normalization</a>()</code> <br><code><a href="layer_upsampling_1d.html">layer_upsampling_1d</a>()</code> <br><code><a href="layer_upsampling_2d.html">layer_upsampling_2d</a>()</code> <br><code><a href="layer_upsampling_3d.html">layer_upsampling_3d</a>()</code> <br><code><a href="layer_zero_padding_1d.html">layer_zero_padding_1d</a>()</code> <br><code><a href="layer_zero_padding_2d.html">layer_zero_padding_2d</a>()</code> <br><code><a href="layer_zero_padding_3d.html">layer_zero_padding_3d</a>()</code> <br><code><a href="rnn_cell_gru.html">rnn_cell_gru</a>()</code> <br><code><a href="rnn_cell_lstm.html">rnn_cell_lstm</a>()</code> <br><code><a href="rnn_cell_simple.html">rnn_cell_simple</a>()</code> <br><code><a href="rnn_cells_stack.html">rnn_cells_stack</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -124,7 +124,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/layer_flax_module_wrapper.html b/docs/dev/reference/layer_flax_module_wrapper.html
index 970898df2..08bf0fe11 100644
--- a/docs/dev/reference/layer_flax_module_wrapper.html
+++ b/docs/dev/reference/layer_flax_module_wrapper.html
@@ -178,7 +178,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -363,7 +363,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
     <div class="dont-index">
 <ul><li><p><a href="https://www.tensorflow.org/api_docs/python/tf/keras/layers/FlaxLayer" class="external-link">https://www.tensorflow.org/api_docs/python/tf/keras/layers/FlaxLayer</a></p></li>
 </ul><p>Other wrapping layers: <br><code><a href="layer_jax_model_wrapper.html">layer_jax_model_wrapper</a>()</code> <br><code><a href="layer_torch_module_wrapper.html">layer_torch_module_wrapper</a>()</code> <br></p>
-<p>Other layers: <br><code><a href="Layer.html">Layer</a>()</code> <br><code><a href="layer_activation.html">layer_activation</a>()</code> <br><code><a href="layer_activation_elu.html">layer_activation_elu</a>()</code> <br><code><a href="layer_activation_leaky_relu.html">layer_activation_leaky_relu</a>()</code> <br><code><a href="layer_activation_parametric_relu.html">layer_activation_parametric_relu</a>()</code> <br><code><a href="layer_activation_relu.html">layer_activation_relu</a>()</code> <br><code><a href="layer_activation_softmax.html">layer_activation_softmax</a>()</code> <br><code><a href="layer_activity_regularization.html">layer_activity_regularization</a>()</code> <br><code><a href="layer_add.html">layer_add</a>()</code> <br><code><a href="layer_additive_attention.html">layer_additive_attention</a>()</code> <br><code><a href="layer_alpha_dropout.html">layer_alpha_dropout</a>()</code> <br><code><a href="layer_attention.html">layer_attention</a>()</code> <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_average.html">layer_average</a>()</code> <br><code><a href="layer_average_pooling_1d.html">layer_average_pooling_1d</a>()</code> <br><code><a href="layer_average_pooling_2d.html">layer_average_pooling_2d</a>()</code> <br><code><a href="layer_average_pooling_3d.html">layer_average_pooling_3d</a>()</code> <br><code><a href="layer_batch_normalization.html">layer_batch_normalization</a>()</code> <br><code><a href="layer_bidirectional.html">layer_bidirectional</a>()</code> <br><code><a href="layer_category_encoding.html">layer_category_encoding</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_concatenate.html">layer_concatenate</a>()</code> <br><code><a href="layer_conv_1d.html">layer_conv_1d</a>()</code> <br><code><a href="layer_conv_1d_transpose.html">layer_conv_1d_transpose</a>()</code> <br><code><a href="layer_conv_2d.html">layer_conv_2d</a>()</code> <br><code><a href="layer_conv_2d_transpose.html">layer_conv_2d_transpose</a>()</code> <br><code><a href="layer_conv_3d.html">layer_conv_3d</a>()</code> <br><code><a href="layer_conv_3d_transpose.html">layer_conv_3d_transpose</a>()</code> <br><code><a href="layer_conv_lstm_1d.html">layer_conv_lstm_1d</a>()</code> <br><code><a href="layer_conv_lstm_2d.html">layer_conv_lstm_2d</a>()</code> <br><code><a href="layer_conv_lstm_3d.html">layer_conv_lstm_3d</a>()</code> <br><code><a href="layer_cropping_1d.html">layer_cropping_1d</a>()</code> <br><code><a href="layer_cropping_2d.html">layer_cropping_2d</a>()</code> <br><code><a href="layer_cropping_3d.html">layer_cropping_3d</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_dense.html">layer_dense</a>()</code> <br><code><a href="layer_depthwise_conv_1d.html">layer_depthwise_conv_1d</a>()</code> <br><code><a href="layer_depthwise_conv_2d.html">layer_depthwise_conv_2d</a>()</code> <br><code><a href="layer_discretization.html">layer_discretization</a>()</code> <br><code><a href="layer_dot.html">layer_dot</a>()</code> <br><code><a href="layer_dropout.html">layer_dropout</a>()</code> <br><code><a href="layer_einsum_dense.html">layer_einsum_dense</a>()</code> <br><code><a href="layer_embedding.html">layer_embedding</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_feature_space.html">layer_feature_space</a>()</code> <br><code><a href="layer_flatten.html">layer_flatten</a>()</code> <br><code><a href="layer_gaussian_dropout.html">layer_gaussian_dropout</a>()</code> <br><code><a href="layer_gaussian_noise.html">layer_gaussian_noise</a>()</code> <br><code><a href="layer_global_average_pooling_1d.html">layer_global_average_pooling_1d</a>()</code> <br><code><a href="layer_global_average_pooling_2d.html">layer_global_average_pooling_2d</a>()</code> <br><code><a href="layer_global_average_pooling_3d.html">layer_global_average_pooling_3d</a>()</code> <br><code><a href="layer_global_max_pooling_1d.html">layer_global_max_pooling_1d</a>()</code> <br><code><a href="layer_global_max_pooling_2d.html">layer_global_max_pooling_2d</a>()</code> <br><code><a href="layer_global_max_pooling_3d.html">layer_global_max_pooling_3d</a>()</code> <br><code><a href="layer_group_normalization.html">layer_group_normalization</a>()</code> <br><code><a href="layer_group_query_attention.html">layer_group_query_attention</a>()</code> <br><code><a href="layer_gru.html">layer_gru</a>()</code> <br><code><a href="layer_hashed_crossing.html">layer_hashed_crossing</a>()</code> <br><code><a href="layer_hashing.html">layer_hashing</a>()</code> <br><code><a href="layer_identity.html">layer_identity</a>()</code> <br><code><a href="layer_integer_lookup.html">layer_integer_lookup</a>()</code> <br><code><a href="layer_jax_model_wrapper.html">layer_jax_model_wrapper</a>()</code> <br><code><a href="layer_lambda.html">layer_lambda</a>()</code> <br><code><a href="layer_layer_normalization.html">layer_layer_normalization</a>()</code> <br><code><a href="layer_lstm.html">layer_lstm</a>()</code> <br><code><a href="layer_masking.html">layer_masking</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_max_pooling_1d.html">layer_max_pooling_1d</a>()</code> <br><code><a href="layer_max_pooling_2d.html">layer_max_pooling_2d</a>()</code> <br><code><a href="layer_max_pooling_3d.html">layer_max_pooling_3d</a>()</code> <br><code><a href="layer_maximum.html">layer_maximum</a>()</code> <br><code><a href="layer_mel_spectrogram.html">layer_mel_spectrogram</a>()</code> <br><code><a href="layer_minimum.html">layer_minimum</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_multi_head_attention.html">layer_multi_head_attention</a>()</code> <br><code><a href="layer_multiply.html">layer_multiply</a>()</code> <br><code><a href="layer_normalization.html">layer_normalization</a>()</code> <br><code><a href="layer_permute.html">layer_permute</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_brightness.html">layer_random_brightness</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_contrast.html">layer_random_contrast</a>()</code> <br><code><a href="layer_random_crop.html">layer_random_crop</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_flip.html">layer_random_flip</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_rotation.html">layer_random_rotation</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_random_translation.html">layer_random_translation</a>()</code> <br><code><a href="layer_random_zoom.html">layer_random_zoom</a>()</code> <br><code><a href="layer_repeat_vector.html">layer_repeat_vector</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_reshape.html">layer_reshape</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_rms_normalization.html">layer_rms_normalization</a>()</code> <br><code><a href="layer_rnn.html">layer_rnn</a>()</code> <br><code><a href="layer_separable_conv_1d.html">layer_separable_conv_1d</a>()</code> <br><code><a href="layer_separable_conv_2d.html">layer_separable_conv_2d</a>()</code> <br><code><a href="layer_simple_rnn.html">layer_simple_rnn</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br><code><a href="layer_spatial_dropout_1d.html">layer_spatial_dropout_1d</a>()</code> <br><code><a href="layer_spatial_dropout_2d.html">layer_spatial_dropout_2d</a>()</code> <br><code><a href="layer_spatial_dropout_3d.html">layer_spatial_dropout_3d</a>()</code> <br><code><a href="layer_spectral_normalization.html">layer_spectral_normalization</a>()</code> <br><code><a href="layer_stft_spectrogram.html">layer_stft_spectrogram</a>()</code> <br><code><a href="layer_string_lookup.html">layer_string_lookup</a>()</code> <br><code><a href="layer_subtract.html">layer_subtract</a>()</code> <br><code><a href="layer_text_vectorization.html">layer_text_vectorization</a>()</code> <br><code><a href="layer_tfsm.html">layer_tfsm</a>()</code> <br><code><a href="layer_time_distributed.html">layer_time_distributed</a>()</code> <br><code><a href="layer_torch_module_wrapper.html">layer_torch_module_wrapper</a>()</code> <br><code><a href="layer_unit_normalization.html">layer_unit_normalization</a>()</code> <br><code><a href="layer_upsampling_1d.html">layer_upsampling_1d</a>()</code> <br><code><a href="layer_upsampling_2d.html">layer_upsampling_2d</a>()</code> <br><code><a href="layer_upsampling_3d.html">layer_upsampling_3d</a>()</code> <br><code><a href="layer_zero_padding_1d.html">layer_zero_padding_1d</a>()</code> <br><code><a href="layer_zero_padding_2d.html">layer_zero_padding_2d</a>()</code> <br><code><a href="layer_zero_padding_3d.html">layer_zero_padding_3d</a>()</code> <br><code><a href="rnn_cell_gru.html">rnn_cell_gru</a>()</code> <br><code><a href="rnn_cell_lstm.html">rnn_cell_lstm</a>()</code> <br><code><a href="rnn_cell_simple.html">rnn_cell_simple</a>()</code> <br><code><a href="rnn_cells_stack.html">rnn_cells_stack</a>()</code> <br></p></div>
+<p>Other layers: <br><code><a href="Layer.html">Layer</a>()</code> <br><code><a href="layer_activation.html">layer_activation</a>()</code> <br><code><a href="layer_activation_elu.html">layer_activation_elu</a>()</code> <br><code><a href="layer_activation_leaky_relu.html">layer_activation_leaky_relu</a>()</code> <br><code><a href="layer_activation_parametric_relu.html">layer_activation_parametric_relu</a>()</code> <br><code><a href="layer_activation_relu.html">layer_activation_relu</a>()</code> <br><code><a href="layer_activation_softmax.html">layer_activation_softmax</a>()</code> <br><code><a href="layer_activity_regularization.html">layer_activity_regularization</a>()</code> <br><code><a href="layer_add.html">layer_add</a>()</code> <br><code><a href="layer_additive_attention.html">layer_additive_attention</a>()</code> <br><code><a href="layer_alpha_dropout.html">layer_alpha_dropout</a>()</code> <br><code><a href="layer_attention.html">layer_attention</a>()</code> <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_average.html">layer_average</a>()</code> <br><code><a href="layer_average_pooling_1d.html">layer_average_pooling_1d</a>()</code> <br><code><a href="layer_average_pooling_2d.html">layer_average_pooling_2d</a>()</code> <br><code><a href="layer_average_pooling_3d.html">layer_average_pooling_3d</a>()</code> <br><code><a href="layer_batch_normalization.html">layer_batch_normalization</a>()</code> <br><code><a href="layer_bidirectional.html">layer_bidirectional</a>()</code> <br><code><a href="layer_category_encoding.html">layer_category_encoding</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_concatenate.html">layer_concatenate</a>()</code> <br><code><a href="layer_conv_1d.html">layer_conv_1d</a>()</code> <br><code><a href="layer_conv_1d_transpose.html">layer_conv_1d_transpose</a>()</code> <br><code><a href="layer_conv_2d.html">layer_conv_2d</a>()</code> <br><code><a href="layer_conv_2d_transpose.html">layer_conv_2d_transpose</a>()</code> <br><code><a href="layer_conv_3d.html">layer_conv_3d</a>()</code> <br><code><a href="layer_conv_3d_transpose.html">layer_conv_3d_transpose</a>()</code> <br><code><a href="layer_conv_lstm_1d.html">layer_conv_lstm_1d</a>()</code> <br><code><a href="layer_conv_lstm_2d.html">layer_conv_lstm_2d</a>()</code> <br><code><a href="layer_conv_lstm_3d.html">layer_conv_lstm_3d</a>()</code> <br><code><a href="layer_cropping_1d.html">layer_cropping_1d</a>()</code> <br><code><a href="layer_cropping_2d.html">layer_cropping_2d</a>()</code> <br><code><a href="layer_cropping_3d.html">layer_cropping_3d</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_dense.html">layer_dense</a>()</code> <br><code><a href="layer_depthwise_conv_1d.html">layer_depthwise_conv_1d</a>()</code> <br><code><a href="layer_depthwise_conv_2d.html">layer_depthwise_conv_2d</a>()</code> <br><code><a href="layer_discretization.html">layer_discretization</a>()</code> <br><code><a href="layer_dot.html">layer_dot</a>()</code> <br><code><a href="layer_dropout.html">layer_dropout</a>()</code> <br><code><a href="layer_einsum_dense.html">layer_einsum_dense</a>()</code> <br><code><a href="layer_embedding.html">layer_embedding</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_feature_space.html">layer_feature_space</a>()</code> <br><code><a href="layer_flatten.html">layer_flatten</a>()</code> <br><code><a href="layer_gaussian_dropout.html">layer_gaussian_dropout</a>()</code> <br><code><a href="layer_gaussian_noise.html">layer_gaussian_noise</a>()</code> <br><code><a href="layer_global_average_pooling_1d.html">layer_global_average_pooling_1d</a>()</code> <br><code><a href="layer_global_average_pooling_2d.html">layer_global_average_pooling_2d</a>()</code> <br><code><a href="layer_global_average_pooling_3d.html">layer_global_average_pooling_3d</a>()</code> <br><code><a href="layer_global_max_pooling_1d.html">layer_global_max_pooling_1d</a>()</code> <br><code><a href="layer_global_max_pooling_2d.html">layer_global_max_pooling_2d</a>()</code> <br><code><a href="layer_global_max_pooling_3d.html">layer_global_max_pooling_3d</a>()</code> <br><code><a href="layer_group_normalization.html">layer_group_normalization</a>()</code> <br><code><a href="layer_group_query_attention.html">layer_group_query_attention</a>()</code> <br><code><a href="layer_gru.html">layer_gru</a>()</code> <br><code><a href="layer_hashed_crossing.html">layer_hashed_crossing</a>()</code> <br><code><a href="layer_hashing.html">layer_hashing</a>()</code> <br><code><a href="layer_identity.html">layer_identity</a>()</code> <br><code><a href="layer_integer_lookup.html">layer_integer_lookup</a>()</code> <br><code><a href="layer_jax_model_wrapper.html">layer_jax_model_wrapper</a>()</code> <br><code><a href="layer_lambda.html">layer_lambda</a>()</code> <br><code><a href="layer_layer_normalization.html">layer_layer_normalization</a>()</code> <br><code><a href="layer_lstm.html">layer_lstm</a>()</code> <br><code><a href="layer_masking.html">layer_masking</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_max_pooling_1d.html">layer_max_pooling_1d</a>()</code> <br><code><a href="layer_max_pooling_2d.html">layer_max_pooling_2d</a>()</code> <br><code><a href="layer_max_pooling_3d.html">layer_max_pooling_3d</a>()</code> <br><code><a href="layer_maximum.html">layer_maximum</a>()</code> <br><code><a href="layer_mel_spectrogram.html">layer_mel_spectrogram</a>()</code> <br><code><a href="layer_minimum.html">layer_minimum</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_multi_head_attention.html">layer_multi_head_attention</a>()</code> <br><code><a href="layer_multiply.html">layer_multiply</a>()</code> <br><code><a href="layer_normalization.html">layer_normalization</a>()</code> <br><code><a href="layer_permute.html">layer_permute</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_brightness.html">layer_random_brightness</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_contrast.html">layer_random_contrast</a>()</code> <br><code><a href="layer_random_crop.html">layer_random_crop</a>()</code> <br><code><a href="layer_random_elastic_transform.html">layer_random_elastic_transform</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_flip.html">layer_random_flip</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_rotation.html">layer_random_rotation</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_random_translation.html">layer_random_translation</a>()</code> <br><code><a href="layer_random_zoom.html">layer_random_zoom</a>()</code> <br><code><a href="layer_repeat_vector.html">layer_repeat_vector</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_reshape.html">layer_reshape</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_rms_normalization.html">layer_rms_normalization</a>()</code> <br><code><a href="layer_rnn.html">layer_rnn</a>()</code> <br><code><a href="layer_separable_conv_1d.html">layer_separable_conv_1d</a>()</code> <br><code><a href="layer_separable_conv_2d.html">layer_separable_conv_2d</a>()</code> <br><code><a href="layer_simple_rnn.html">layer_simple_rnn</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br><code><a href="layer_spatial_dropout_1d.html">layer_spatial_dropout_1d</a>()</code> <br><code><a href="layer_spatial_dropout_2d.html">layer_spatial_dropout_2d</a>()</code> <br><code><a href="layer_spatial_dropout_3d.html">layer_spatial_dropout_3d</a>()</code> <br><code><a href="layer_spectral_normalization.html">layer_spectral_normalization</a>()</code> <br><code><a href="layer_stft_spectrogram.html">layer_stft_spectrogram</a>()</code> <br><code><a href="layer_string_lookup.html">layer_string_lookup</a>()</code> <br><code><a href="layer_subtract.html">layer_subtract</a>()</code> <br><code><a href="layer_text_vectorization.html">layer_text_vectorization</a>()</code> <br><code><a href="layer_tfsm.html">layer_tfsm</a>()</code> <br><code><a href="layer_time_distributed.html">layer_time_distributed</a>()</code> <br><code><a href="layer_torch_module_wrapper.html">layer_torch_module_wrapper</a>()</code> <br><code><a href="layer_unit_normalization.html">layer_unit_normalization</a>()</code> <br><code><a href="layer_upsampling_1d.html">layer_upsampling_1d</a>()</code> <br><code><a href="layer_upsampling_2d.html">layer_upsampling_2d</a>()</code> <br><code><a href="layer_upsampling_3d.html">layer_upsampling_3d</a>()</code> <br><code><a href="layer_zero_padding_1d.html">layer_zero_padding_1d</a>()</code> <br><code><a href="layer_zero_padding_2d.html">layer_zero_padding_2d</a>()</code> <br><code><a href="layer_zero_padding_3d.html">layer_zero_padding_3d</a>()</code> <br><code><a href="rnn_cell_gru.html">rnn_cell_gru</a>()</code> <br><code><a href="rnn_cell_lstm.html">rnn_cell_lstm</a>()</code> <br><code><a href="rnn_cell_simple.html">rnn_cell_simple</a>()</code> <br><code><a href="rnn_cells_stack.html">rnn_cells_stack</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -375,7 +375,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/layer_gaussian_dropout.html b/docs/dev/reference/layer_gaussian_dropout.html
index 599136b78..3221a2f65 100644
--- a/docs/dev/reference/layer_gaussian_dropout.html
+++ b/docs/dev/reference/layer_gaussian_dropout.html
@@ -8,7 +8,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -105,7 +105,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
     <div class="dont-index">
 <ul><li><p><a href="https://keras.io/api/layers/regularization_layers/gaussian_dropout#gaussiandropout-class" class="external-link">https://keras.io/api/layers/regularization_layers/gaussian_dropout#gaussiandropout-class</a></p></li>
 </ul><p>Other regularization layers: <br><code><a href="layer_activity_regularization.html">layer_activity_regularization</a>()</code> <br><code><a href="layer_alpha_dropout.html">layer_alpha_dropout</a>()</code> <br><code><a href="layer_dropout.html">layer_dropout</a>()</code> <br><code><a href="layer_gaussian_noise.html">layer_gaussian_noise</a>()</code> <br><code><a href="layer_spatial_dropout_1d.html">layer_spatial_dropout_1d</a>()</code> <br><code><a href="layer_spatial_dropout_2d.html">layer_spatial_dropout_2d</a>()</code> <br><code><a href="layer_spatial_dropout_3d.html">layer_spatial_dropout_3d</a>()</code> <br></p>
-<p>Other layers: <br><code><a href="Layer.html">Layer</a>()</code> <br><code><a href="layer_activation.html">layer_activation</a>()</code> <br><code><a href="layer_activation_elu.html">layer_activation_elu</a>()</code> <br><code><a href="layer_activation_leaky_relu.html">layer_activation_leaky_relu</a>()</code> <br><code><a href="layer_activation_parametric_relu.html">layer_activation_parametric_relu</a>()</code> <br><code><a href="layer_activation_relu.html">layer_activation_relu</a>()</code> <br><code><a href="layer_activation_softmax.html">layer_activation_softmax</a>()</code> <br><code><a href="layer_activity_regularization.html">layer_activity_regularization</a>()</code> <br><code><a href="layer_add.html">layer_add</a>()</code> <br><code><a href="layer_additive_attention.html">layer_additive_attention</a>()</code> <br><code><a href="layer_alpha_dropout.html">layer_alpha_dropout</a>()</code> <br><code><a href="layer_attention.html">layer_attention</a>()</code> <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_average.html">layer_average</a>()</code> <br><code><a href="layer_average_pooling_1d.html">layer_average_pooling_1d</a>()</code> <br><code><a href="layer_average_pooling_2d.html">layer_average_pooling_2d</a>()</code> <br><code><a href="layer_average_pooling_3d.html">layer_average_pooling_3d</a>()</code> <br><code><a href="layer_batch_normalization.html">layer_batch_normalization</a>()</code> <br><code><a href="layer_bidirectional.html">layer_bidirectional</a>()</code> <br><code><a href="layer_category_encoding.html">layer_category_encoding</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_concatenate.html">layer_concatenate</a>()</code> <br><code><a href="layer_conv_1d.html">layer_conv_1d</a>()</code> <br><code><a href="layer_conv_1d_transpose.html">layer_conv_1d_transpose</a>()</code> <br><code><a href="layer_conv_2d.html">layer_conv_2d</a>()</code> <br><code><a href="layer_conv_2d_transpose.html">layer_conv_2d_transpose</a>()</code> <br><code><a href="layer_conv_3d.html">layer_conv_3d</a>()</code> <br><code><a href="layer_conv_3d_transpose.html">layer_conv_3d_transpose</a>()</code> <br><code><a href="layer_conv_lstm_1d.html">layer_conv_lstm_1d</a>()</code> <br><code><a href="layer_conv_lstm_2d.html">layer_conv_lstm_2d</a>()</code> <br><code><a href="layer_conv_lstm_3d.html">layer_conv_lstm_3d</a>()</code> <br><code><a href="layer_cropping_1d.html">layer_cropping_1d</a>()</code> <br><code><a href="layer_cropping_2d.html">layer_cropping_2d</a>()</code> <br><code><a href="layer_cropping_3d.html">layer_cropping_3d</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_dense.html">layer_dense</a>()</code> <br><code><a href="layer_depthwise_conv_1d.html">layer_depthwise_conv_1d</a>()</code> <br><code><a href="layer_depthwise_conv_2d.html">layer_depthwise_conv_2d</a>()</code> <br><code><a href="layer_discretization.html">layer_discretization</a>()</code> <br><code><a href="layer_dot.html">layer_dot</a>()</code> <br><code><a href="layer_dropout.html">layer_dropout</a>()</code> <br><code><a href="layer_einsum_dense.html">layer_einsum_dense</a>()</code> <br><code><a href="layer_embedding.html">layer_embedding</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_feature_space.html">layer_feature_space</a>()</code> <br><code><a href="layer_flatten.html">layer_flatten</a>()</code> <br><code><a href="layer_flax_module_wrapper.html">layer_flax_module_wrapper</a>()</code> <br><code><a href="layer_gaussian_noise.html">layer_gaussian_noise</a>()</code> <br><code><a href="layer_global_average_pooling_1d.html">layer_global_average_pooling_1d</a>()</code> <br><code><a href="layer_global_average_pooling_2d.html">layer_global_average_pooling_2d</a>()</code> <br><code><a href="layer_global_average_pooling_3d.html">layer_global_average_pooling_3d</a>()</code> <br><code><a href="layer_global_max_pooling_1d.html">layer_global_max_pooling_1d</a>()</code> <br><code><a href="layer_global_max_pooling_2d.html">layer_global_max_pooling_2d</a>()</code> <br><code><a href="layer_global_max_pooling_3d.html">layer_global_max_pooling_3d</a>()</code> <br><code><a href="layer_group_normalization.html">layer_group_normalization</a>()</code> <br><code><a href="layer_group_query_attention.html">layer_group_query_attention</a>()</code> <br><code><a href="layer_gru.html">layer_gru</a>()</code> <br><code><a href="layer_hashed_crossing.html">layer_hashed_crossing</a>()</code> <br><code><a href="layer_hashing.html">layer_hashing</a>()</code> <br><code><a href="layer_identity.html">layer_identity</a>()</code> <br><code><a href="layer_integer_lookup.html">layer_integer_lookup</a>()</code> <br><code><a href="layer_jax_model_wrapper.html">layer_jax_model_wrapper</a>()</code> <br><code><a href="layer_lambda.html">layer_lambda</a>()</code> <br><code><a href="layer_layer_normalization.html">layer_layer_normalization</a>()</code> <br><code><a href="layer_lstm.html">layer_lstm</a>()</code> <br><code><a href="layer_masking.html">layer_masking</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_max_pooling_1d.html">layer_max_pooling_1d</a>()</code> <br><code><a href="layer_max_pooling_2d.html">layer_max_pooling_2d</a>()</code> <br><code><a href="layer_max_pooling_3d.html">layer_max_pooling_3d</a>()</code> <br><code><a href="layer_maximum.html">layer_maximum</a>()</code> <br><code><a href="layer_mel_spectrogram.html">layer_mel_spectrogram</a>()</code> <br><code><a href="layer_minimum.html">layer_minimum</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_multi_head_attention.html">layer_multi_head_attention</a>()</code> <br><code><a href="layer_multiply.html">layer_multiply</a>()</code> <br><code><a href="layer_normalization.html">layer_normalization</a>()</code> <br><code><a href="layer_permute.html">layer_permute</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_brightness.html">layer_random_brightness</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_contrast.html">layer_random_contrast</a>()</code> <br><code><a href="layer_random_crop.html">layer_random_crop</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_flip.html">layer_random_flip</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_rotation.html">layer_random_rotation</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_random_translation.html">layer_random_translation</a>()</code> <br><code><a href="layer_random_zoom.html">layer_random_zoom</a>()</code> <br><code><a href="layer_repeat_vector.html">layer_repeat_vector</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_reshape.html">layer_reshape</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_rms_normalization.html">layer_rms_normalization</a>()</code> <br><code><a href="layer_rnn.html">layer_rnn</a>()</code> <br><code><a href="layer_separable_conv_1d.html">layer_separable_conv_1d</a>()</code> <br><code><a href="layer_separable_conv_2d.html">layer_separable_conv_2d</a>()</code> <br><code><a href="layer_simple_rnn.html">layer_simple_rnn</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br><code><a href="layer_spatial_dropout_1d.html">layer_spatial_dropout_1d</a>()</code> <br><code><a href="layer_spatial_dropout_2d.html">layer_spatial_dropout_2d</a>()</code> <br><code><a href="layer_spatial_dropout_3d.html">layer_spatial_dropout_3d</a>()</code> <br><code><a href="layer_spectral_normalization.html">layer_spectral_normalization</a>()</code> <br><code><a href="layer_stft_spectrogram.html">layer_stft_spectrogram</a>()</code> <br><code><a href="layer_string_lookup.html">layer_string_lookup</a>()</code> <br><code><a href="layer_subtract.html">layer_subtract</a>()</code> <br><code><a href="layer_text_vectorization.html">layer_text_vectorization</a>()</code> <br><code><a href="layer_tfsm.html">layer_tfsm</a>()</code> <br><code><a href="layer_time_distributed.html">layer_time_distributed</a>()</code> <br><code><a href="layer_torch_module_wrapper.html">layer_torch_module_wrapper</a>()</code> <br><code><a href="layer_unit_normalization.html">layer_unit_normalization</a>()</code> <br><code><a href="layer_upsampling_1d.html">layer_upsampling_1d</a>()</code> <br><code><a href="layer_upsampling_2d.html">layer_upsampling_2d</a>()</code> <br><code><a href="layer_upsampling_3d.html">layer_upsampling_3d</a>()</code> <br><code><a href="layer_zero_padding_1d.html">layer_zero_padding_1d</a>()</code> <br><code><a href="layer_zero_padding_2d.html">layer_zero_padding_2d</a>()</code> <br><code><a href="layer_zero_padding_3d.html">layer_zero_padding_3d</a>()</code> <br><code><a href="rnn_cell_gru.html">rnn_cell_gru</a>()</code> <br><code><a href="rnn_cell_lstm.html">rnn_cell_lstm</a>()</code> <br><code><a href="rnn_cell_simple.html">rnn_cell_simple</a>()</code> <br><code><a href="rnn_cells_stack.html">rnn_cells_stack</a>()</code> <br></p></div>
+<p>Other layers: <br><code><a href="Layer.html">Layer</a>()</code> <br><code><a href="layer_activation.html">layer_activation</a>()</code> <br><code><a href="layer_activation_elu.html">layer_activation_elu</a>()</code> <br><code><a href="layer_activation_leaky_relu.html">layer_activation_leaky_relu</a>()</code> <br><code><a href="layer_activation_parametric_relu.html">layer_activation_parametric_relu</a>()</code> <br><code><a href="layer_activation_relu.html">layer_activation_relu</a>()</code> <br><code><a href="layer_activation_softmax.html">layer_activation_softmax</a>()</code> <br><code><a href="layer_activity_regularization.html">layer_activity_regularization</a>()</code> <br><code><a href="layer_add.html">layer_add</a>()</code> <br><code><a href="layer_additive_attention.html">layer_additive_attention</a>()</code> <br><code><a href="layer_alpha_dropout.html">layer_alpha_dropout</a>()</code> <br><code><a href="layer_attention.html">layer_attention</a>()</code> <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_average.html">layer_average</a>()</code> <br><code><a href="layer_average_pooling_1d.html">layer_average_pooling_1d</a>()</code> <br><code><a href="layer_average_pooling_2d.html">layer_average_pooling_2d</a>()</code> <br><code><a href="layer_average_pooling_3d.html">layer_average_pooling_3d</a>()</code> <br><code><a href="layer_batch_normalization.html">layer_batch_normalization</a>()</code> <br><code><a href="layer_bidirectional.html">layer_bidirectional</a>()</code> <br><code><a href="layer_category_encoding.html">layer_category_encoding</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_concatenate.html">layer_concatenate</a>()</code> <br><code><a href="layer_conv_1d.html">layer_conv_1d</a>()</code> <br><code><a href="layer_conv_1d_transpose.html">layer_conv_1d_transpose</a>()</code> <br><code><a href="layer_conv_2d.html">layer_conv_2d</a>()</code> <br><code><a href="layer_conv_2d_transpose.html">layer_conv_2d_transpose</a>()</code> <br><code><a href="layer_conv_3d.html">layer_conv_3d</a>()</code> <br><code><a href="layer_conv_3d_transpose.html">layer_conv_3d_transpose</a>()</code> <br><code><a href="layer_conv_lstm_1d.html">layer_conv_lstm_1d</a>()</code> <br><code><a href="layer_conv_lstm_2d.html">layer_conv_lstm_2d</a>()</code> <br><code><a href="layer_conv_lstm_3d.html">layer_conv_lstm_3d</a>()</code> <br><code><a href="layer_cropping_1d.html">layer_cropping_1d</a>()</code> <br><code><a href="layer_cropping_2d.html">layer_cropping_2d</a>()</code> <br><code><a href="layer_cropping_3d.html">layer_cropping_3d</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_dense.html">layer_dense</a>()</code> <br><code><a href="layer_depthwise_conv_1d.html">layer_depthwise_conv_1d</a>()</code> <br><code><a href="layer_depthwise_conv_2d.html">layer_depthwise_conv_2d</a>()</code> <br><code><a href="layer_discretization.html">layer_discretization</a>()</code> <br><code><a href="layer_dot.html">layer_dot</a>()</code> <br><code><a href="layer_dropout.html">layer_dropout</a>()</code> <br><code><a href="layer_einsum_dense.html">layer_einsum_dense</a>()</code> <br><code><a href="layer_embedding.html">layer_embedding</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_feature_space.html">layer_feature_space</a>()</code> <br><code><a href="layer_flatten.html">layer_flatten</a>()</code> <br><code><a href="layer_flax_module_wrapper.html">layer_flax_module_wrapper</a>()</code> <br><code><a href="layer_gaussian_noise.html">layer_gaussian_noise</a>()</code> <br><code><a href="layer_global_average_pooling_1d.html">layer_global_average_pooling_1d</a>()</code> <br><code><a href="layer_global_average_pooling_2d.html">layer_global_average_pooling_2d</a>()</code> <br><code><a href="layer_global_average_pooling_3d.html">layer_global_average_pooling_3d</a>()</code> <br><code><a href="layer_global_max_pooling_1d.html">layer_global_max_pooling_1d</a>()</code> <br><code><a href="layer_global_max_pooling_2d.html">layer_global_max_pooling_2d</a>()</code> <br><code><a href="layer_global_max_pooling_3d.html">layer_global_max_pooling_3d</a>()</code> <br><code><a href="layer_group_normalization.html">layer_group_normalization</a>()</code> <br><code><a href="layer_group_query_attention.html">layer_group_query_attention</a>()</code> <br><code><a href="layer_gru.html">layer_gru</a>()</code> <br><code><a href="layer_hashed_crossing.html">layer_hashed_crossing</a>()</code> <br><code><a href="layer_hashing.html">layer_hashing</a>()</code> <br><code><a href="layer_identity.html">layer_identity</a>()</code> <br><code><a href="layer_integer_lookup.html">layer_integer_lookup</a>()</code> <br><code><a href="layer_jax_model_wrapper.html">layer_jax_model_wrapper</a>()</code> <br><code><a href="layer_lambda.html">layer_lambda</a>()</code> <br><code><a href="layer_layer_normalization.html">layer_layer_normalization</a>()</code> <br><code><a href="layer_lstm.html">layer_lstm</a>()</code> <br><code><a href="layer_masking.html">layer_masking</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_max_pooling_1d.html">layer_max_pooling_1d</a>()</code> <br><code><a href="layer_max_pooling_2d.html">layer_max_pooling_2d</a>()</code> <br><code><a href="layer_max_pooling_3d.html">layer_max_pooling_3d</a>()</code> <br><code><a href="layer_maximum.html">layer_maximum</a>()</code> <br><code><a href="layer_mel_spectrogram.html">layer_mel_spectrogram</a>()</code> <br><code><a href="layer_minimum.html">layer_minimum</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_multi_head_attention.html">layer_multi_head_attention</a>()</code> <br><code><a href="layer_multiply.html">layer_multiply</a>()</code> <br><code><a href="layer_normalization.html">layer_normalization</a>()</code> <br><code><a href="layer_permute.html">layer_permute</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_brightness.html">layer_random_brightness</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_contrast.html">layer_random_contrast</a>()</code> <br><code><a href="layer_random_crop.html">layer_random_crop</a>()</code> <br><code><a href="layer_random_elastic_transform.html">layer_random_elastic_transform</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_flip.html">layer_random_flip</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_rotation.html">layer_random_rotation</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_random_translation.html">layer_random_translation</a>()</code> <br><code><a href="layer_random_zoom.html">layer_random_zoom</a>()</code> <br><code><a href="layer_repeat_vector.html">layer_repeat_vector</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_reshape.html">layer_reshape</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_rms_normalization.html">layer_rms_normalization</a>()</code> <br><code><a href="layer_rnn.html">layer_rnn</a>()</code> <br><code><a href="layer_separable_conv_1d.html">layer_separable_conv_1d</a>()</code> <br><code><a href="layer_separable_conv_2d.html">layer_separable_conv_2d</a>()</code> <br><code><a href="layer_simple_rnn.html">layer_simple_rnn</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br><code><a href="layer_spatial_dropout_1d.html">layer_spatial_dropout_1d</a>()</code> <br><code><a href="layer_spatial_dropout_2d.html">layer_spatial_dropout_2d</a>()</code> <br><code><a href="layer_spatial_dropout_3d.html">layer_spatial_dropout_3d</a>()</code> <br><code><a href="layer_spectral_normalization.html">layer_spectral_normalization</a>()</code> <br><code><a href="layer_stft_spectrogram.html">layer_stft_spectrogram</a>()</code> <br><code><a href="layer_string_lookup.html">layer_string_lookup</a>()</code> <br><code><a href="layer_subtract.html">layer_subtract</a>()</code> <br><code><a href="layer_text_vectorization.html">layer_text_vectorization</a>()</code> <br><code><a href="layer_tfsm.html">layer_tfsm</a>()</code> <br><code><a href="layer_time_distributed.html">layer_time_distributed</a>()</code> <br><code><a href="layer_torch_module_wrapper.html">layer_torch_module_wrapper</a>()</code> <br><code><a href="layer_unit_normalization.html">layer_unit_normalization</a>()</code> <br><code><a href="layer_upsampling_1d.html">layer_upsampling_1d</a>()</code> <br><code><a href="layer_upsampling_2d.html">layer_upsampling_2d</a>()</code> <br><code><a href="layer_upsampling_3d.html">layer_upsampling_3d</a>()</code> <br><code><a href="layer_zero_padding_1d.html">layer_zero_padding_1d</a>()</code> <br><code><a href="layer_zero_padding_2d.html">layer_zero_padding_2d</a>()</code> <br><code><a href="layer_zero_padding_3d.html">layer_zero_padding_3d</a>()</code> <br><code><a href="rnn_cell_gru.html">rnn_cell_gru</a>()</code> <br><code><a href="rnn_cell_lstm.html">rnn_cell_lstm</a>()</code> <br><code><a href="rnn_cell_simple.html">rnn_cell_simple</a>()</code> <br><code><a href="rnn_cells_stack.html">rnn_cells_stack</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -117,7 +117,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/layer_gaussian_noise.html b/docs/dev/reference/layer_gaussian_noise.html
index 36dcbfc05..c88b7a5c5 100644
--- a/docs/dev/reference/layer_gaussian_noise.html
+++ b/docs/dev/reference/layer_gaussian_noise.html
@@ -16,7 +16,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -115,7 +115,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
     <div class="dont-index">
 <ul><li><p><a href="https://keras.io/api/layers/regularization_layers/gaussian_noise#gaussiannoise-class" class="external-link">https://keras.io/api/layers/regularization_layers/gaussian_noise#gaussiannoise-class</a></p></li>
 </ul><p>Other regularization layers: <br><code><a href="layer_activity_regularization.html">layer_activity_regularization</a>()</code> <br><code><a href="layer_alpha_dropout.html">layer_alpha_dropout</a>()</code> <br><code><a href="layer_dropout.html">layer_dropout</a>()</code> <br><code><a href="layer_gaussian_dropout.html">layer_gaussian_dropout</a>()</code> <br><code><a href="layer_spatial_dropout_1d.html">layer_spatial_dropout_1d</a>()</code> <br><code><a href="layer_spatial_dropout_2d.html">layer_spatial_dropout_2d</a>()</code> <br><code><a href="layer_spatial_dropout_3d.html">layer_spatial_dropout_3d</a>()</code> <br></p>
-<p>Other layers: <br><code><a href="Layer.html">Layer</a>()</code> <br><code><a href="layer_activation.html">layer_activation</a>()</code> <br><code><a href="layer_activation_elu.html">layer_activation_elu</a>()</code> <br><code><a href="layer_activation_leaky_relu.html">layer_activation_leaky_relu</a>()</code> <br><code><a href="layer_activation_parametric_relu.html">layer_activation_parametric_relu</a>()</code> <br><code><a href="layer_activation_relu.html">layer_activation_relu</a>()</code> <br><code><a href="layer_activation_softmax.html">layer_activation_softmax</a>()</code> <br><code><a href="layer_activity_regularization.html">layer_activity_regularization</a>()</code> <br><code><a href="layer_add.html">layer_add</a>()</code> <br><code><a href="layer_additive_attention.html">layer_additive_attention</a>()</code> <br><code><a href="layer_alpha_dropout.html">layer_alpha_dropout</a>()</code> <br><code><a href="layer_attention.html">layer_attention</a>()</code> <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_average.html">layer_average</a>()</code> <br><code><a href="layer_average_pooling_1d.html">layer_average_pooling_1d</a>()</code> <br><code><a href="layer_average_pooling_2d.html">layer_average_pooling_2d</a>()</code> <br><code><a href="layer_average_pooling_3d.html">layer_average_pooling_3d</a>()</code> <br><code><a href="layer_batch_normalization.html">layer_batch_normalization</a>()</code> <br><code><a href="layer_bidirectional.html">layer_bidirectional</a>()</code> <br><code><a href="layer_category_encoding.html">layer_category_encoding</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_concatenate.html">layer_concatenate</a>()</code> <br><code><a href="layer_conv_1d.html">layer_conv_1d</a>()</code> <br><code><a href="layer_conv_1d_transpose.html">layer_conv_1d_transpose</a>()</code> <br><code><a href="layer_conv_2d.html">layer_conv_2d</a>()</code> <br><code><a href="layer_conv_2d_transpose.html">layer_conv_2d_transpose</a>()</code> <br><code><a href="layer_conv_3d.html">layer_conv_3d</a>()</code> <br><code><a href="layer_conv_3d_transpose.html">layer_conv_3d_transpose</a>()</code> <br><code><a href="layer_conv_lstm_1d.html">layer_conv_lstm_1d</a>()</code> <br><code><a href="layer_conv_lstm_2d.html">layer_conv_lstm_2d</a>()</code> <br><code><a href="layer_conv_lstm_3d.html">layer_conv_lstm_3d</a>()</code> <br><code><a href="layer_cropping_1d.html">layer_cropping_1d</a>()</code> <br><code><a href="layer_cropping_2d.html">layer_cropping_2d</a>()</code> <br><code><a href="layer_cropping_3d.html">layer_cropping_3d</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_dense.html">layer_dense</a>()</code> <br><code><a href="layer_depthwise_conv_1d.html">layer_depthwise_conv_1d</a>()</code> <br><code><a href="layer_depthwise_conv_2d.html">layer_depthwise_conv_2d</a>()</code> <br><code><a href="layer_discretization.html">layer_discretization</a>()</code> <br><code><a href="layer_dot.html">layer_dot</a>()</code> <br><code><a href="layer_dropout.html">layer_dropout</a>()</code> <br><code><a href="layer_einsum_dense.html">layer_einsum_dense</a>()</code> <br><code><a href="layer_embedding.html">layer_embedding</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_feature_space.html">layer_feature_space</a>()</code> <br><code><a href="layer_flatten.html">layer_flatten</a>()</code> <br><code><a href="layer_flax_module_wrapper.html">layer_flax_module_wrapper</a>()</code> <br><code><a href="layer_gaussian_dropout.html">layer_gaussian_dropout</a>()</code> <br><code><a href="layer_global_average_pooling_1d.html">layer_global_average_pooling_1d</a>()</code> <br><code><a href="layer_global_average_pooling_2d.html">layer_global_average_pooling_2d</a>()</code> <br><code><a href="layer_global_average_pooling_3d.html">layer_global_average_pooling_3d</a>()</code> <br><code><a href="layer_global_max_pooling_1d.html">layer_global_max_pooling_1d</a>()</code> <br><code><a href="layer_global_max_pooling_2d.html">layer_global_max_pooling_2d</a>()</code> <br><code><a href="layer_global_max_pooling_3d.html">layer_global_max_pooling_3d</a>()</code> <br><code><a href="layer_group_normalization.html">layer_group_normalization</a>()</code> <br><code><a href="layer_group_query_attention.html">layer_group_query_attention</a>()</code> <br><code><a href="layer_gru.html">layer_gru</a>()</code> <br><code><a href="layer_hashed_crossing.html">layer_hashed_crossing</a>()</code> <br><code><a href="layer_hashing.html">layer_hashing</a>()</code> <br><code><a href="layer_identity.html">layer_identity</a>()</code> <br><code><a href="layer_integer_lookup.html">layer_integer_lookup</a>()</code> <br><code><a href="layer_jax_model_wrapper.html">layer_jax_model_wrapper</a>()</code> <br><code><a href="layer_lambda.html">layer_lambda</a>()</code> <br><code><a href="layer_layer_normalization.html">layer_layer_normalization</a>()</code> <br><code><a href="layer_lstm.html">layer_lstm</a>()</code> <br><code><a href="layer_masking.html">layer_masking</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_max_pooling_1d.html">layer_max_pooling_1d</a>()</code> <br><code><a href="layer_max_pooling_2d.html">layer_max_pooling_2d</a>()</code> <br><code><a href="layer_max_pooling_3d.html">layer_max_pooling_3d</a>()</code> <br><code><a href="layer_maximum.html">layer_maximum</a>()</code> <br><code><a href="layer_mel_spectrogram.html">layer_mel_spectrogram</a>()</code> <br><code><a href="layer_minimum.html">layer_minimum</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_multi_head_attention.html">layer_multi_head_attention</a>()</code> <br><code><a href="layer_multiply.html">layer_multiply</a>()</code> <br><code><a href="layer_normalization.html">layer_normalization</a>()</code> <br><code><a href="layer_permute.html">layer_permute</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_brightness.html">layer_random_brightness</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_contrast.html">layer_random_contrast</a>()</code> <br><code><a href="layer_random_crop.html">layer_random_crop</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_flip.html">layer_random_flip</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_rotation.html">layer_random_rotation</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_random_translation.html">layer_random_translation</a>()</code> <br><code><a href="layer_random_zoom.html">layer_random_zoom</a>()</code> <br><code><a href="layer_repeat_vector.html">layer_repeat_vector</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_reshape.html">layer_reshape</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_rms_normalization.html">layer_rms_normalization</a>()</code> <br><code><a href="layer_rnn.html">layer_rnn</a>()</code> <br><code><a href="layer_separable_conv_1d.html">layer_separable_conv_1d</a>()</code> <br><code><a href="layer_separable_conv_2d.html">layer_separable_conv_2d</a>()</code> <br><code><a href="layer_simple_rnn.html">layer_simple_rnn</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br><code><a href="layer_spatial_dropout_1d.html">layer_spatial_dropout_1d</a>()</code> <br><code><a href="layer_spatial_dropout_2d.html">layer_spatial_dropout_2d</a>()</code> <br><code><a href="layer_spatial_dropout_3d.html">layer_spatial_dropout_3d</a>()</code> <br><code><a href="layer_spectral_normalization.html">layer_spectral_normalization</a>()</code> <br><code><a href="layer_stft_spectrogram.html">layer_stft_spectrogram</a>()</code> <br><code><a href="layer_string_lookup.html">layer_string_lookup</a>()</code> <br><code><a href="layer_subtract.html">layer_subtract</a>()</code> <br><code><a href="layer_text_vectorization.html">layer_text_vectorization</a>()</code> <br><code><a href="layer_tfsm.html">layer_tfsm</a>()</code> <br><code><a href="layer_time_distributed.html">layer_time_distributed</a>()</code> <br><code><a href="layer_torch_module_wrapper.html">layer_torch_module_wrapper</a>()</code> <br><code><a href="layer_unit_normalization.html">layer_unit_normalization</a>()</code> <br><code><a href="layer_upsampling_1d.html">layer_upsampling_1d</a>()</code> <br><code><a href="layer_upsampling_2d.html">layer_upsampling_2d</a>()</code> <br><code><a href="layer_upsampling_3d.html">layer_upsampling_3d</a>()</code> <br><code><a href="layer_zero_padding_1d.html">layer_zero_padding_1d</a>()</code> <br><code><a href="layer_zero_padding_2d.html">layer_zero_padding_2d</a>()</code> <br><code><a href="layer_zero_padding_3d.html">layer_zero_padding_3d</a>()</code> <br><code><a href="rnn_cell_gru.html">rnn_cell_gru</a>()</code> <br><code><a href="rnn_cell_lstm.html">rnn_cell_lstm</a>()</code> <br><code><a href="rnn_cell_simple.html">rnn_cell_simple</a>()</code> <br><code><a href="rnn_cells_stack.html">rnn_cells_stack</a>()</code> <br></p></div>
+<p>Other layers: <br><code><a href="Layer.html">Layer</a>()</code> <br><code><a href="layer_activation.html">layer_activation</a>()</code> <br><code><a href="layer_activation_elu.html">layer_activation_elu</a>()</code> <br><code><a href="layer_activation_leaky_relu.html">layer_activation_leaky_relu</a>()</code> <br><code><a href="layer_activation_parametric_relu.html">layer_activation_parametric_relu</a>()</code> <br><code><a href="layer_activation_relu.html">layer_activation_relu</a>()</code> <br><code><a href="layer_activation_softmax.html">layer_activation_softmax</a>()</code> <br><code><a href="layer_activity_regularization.html">layer_activity_regularization</a>()</code> <br><code><a href="layer_add.html">layer_add</a>()</code> <br><code><a href="layer_additive_attention.html">layer_additive_attention</a>()</code> <br><code><a href="layer_alpha_dropout.html">layer_alpha_dropout</a>()</code> <br><code><a href="layer_attention.html">layer_attention</a>()</code> <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_average.html">layer_average</a>()</code> <br><code><a href="layer_average_pooling_1d.html">layer_average_pooling_1d</a>()</code> <br><code><a href="layer_average_pooling_2d.html">layer_average_pooling_2d</a>()</code> <br><code><a href="layer_average_pooling_3d.html">layer_average_pooling_3d</a>()</code> <br><code><a href="layer_batch_normalization.html">layer_batch_normalization</a>()</code> <br><code><a href="layer_bidirectional.html">layer_bidirectional</a>()</code> <br><code><a href="layer_category_encoding.html">layer_category_encoding</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_concatenate.html">layer_concatenate</a>()</code> <br><code><a href="layer_conv_1d.html">layer_conv_1d</a>()</code> <br><code><a href="layer_conv_1d_transpose.html">layer_conv_1d_transpose</a>()</code> <br><code><a href="layer_conv_2d.html">layer_conv_2d</a>()</code> <br><code><a href="layer_conv_2d_transpose.html">layer_conv_2d_transpose</a>()</code> <br><code><a href="layer_conv_3d.html">layer_conv_3d</a>()</code> <br><code><a href="layer_conv_3d_transpose.html">layer_conv_3d_transpose</a>()</code> <br><code><a href="layer_conv_lstm_1d.html">layer_conv_lstm_1d</a>()</code> <br><code><a href="layer_conv_lstm_2d.html">layer_conv_lstm_2d</a>()</code> <br><code><a href="layer_conv_lstm_3d.html">layer_conv_lstm_3d</a>()</code> <br><code><a href="layer_cropping_1d.html">layer_cropping_1d</a>()</code> <br><code><a href="layer_cropping_2d.html">layer_cropping_2d</a>()</code> <br><code><a href="layer_cropping_3d.html">layer_cropping_3d</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_dense.html">layer_dense</a>()</code> <br><code><a href="layer_depthwise_conv_1d.html">layer_depthwise_conv_1d</a>()</code> <br><code><a href="layer_depthwise_conv_2d.html">layer_depthwise_conv_2d</a>()</code> <br><code><a href="layer_discretization.html">layer_discretization</a>()</code> <br><code><a href="layer_dot.html">layer_dot</a>()</code> <br><code><a href="layer_dropout.html">layer_dropout</a>()</code> <br><code><a href="layer_einsum_dense.html">layer_einsum_dense</a>()</code> <br><code><a href="layer_embedding.html">layer_embedding</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_feature_space.html">layer_feature_space</a>()</code> <br><code><a href="layer_flatten.html">layer_flatten</a>()</code> <br><code><a href="layer_flax_module_wrapper.html">layer_flax_module_wrapper</a>()</code> <br><code><a href="layer_gaussian_dropout.html">layer_gaussian_dropout</a>()</code> <br><code><a href="layer_global_average_pooling_1d.html">layer_global_average_pooling_1d</a>()</code> <br><code><a href="layer_global_average_pooling_2d.html">layer_global_average_pooling_2d</a>()</code> <br><code><a href="layer_global_average_pooling_3d.html">layer_global_average_pooling_3d</a>()</code> <br><code><a href="layer_global_max_pooling_1d.html">layer_global_max_pooling_1d</a>()</code> <br><code><a href="layer_global_max_pooling_2d.html">layer_global_max_pooling_2d</a>()</code> <br><code><a href="layer_global_max_pooling_3d.html">layer_global_max_pooling_3d</a>()</code> <br><code><a href="layer_group_normalization.html">layer_group_normalization</a>()</code> <br><code><a href="layer_group_query_attention.html">layer_group_query_attention</a>()</code> <br><code><a href="layer_gru.html">layer_gru</a>()</code> <br><code><a href="layer_hashed_crossing.html">layer_hashed_crossing</a>()</code> <br><code><a href="layer_hashing.html">layer_hashing</a>()</code> <br><code><a href="layer_identity.html">layer_identity</a>()</code> <br><code><a href="layer_integer_lookup.html">layer_integer_lookup</a>()</code> <br><code><a href="layer_jax_model_wrapper.html">layer_jax_model_wrapper</a>()</code> <br><code><a href="layer_lambda.html">layer_lambda</a>()</code> <br><code><a href="layer_layer_normalization.html">layer_layer_normalization</a>()</code> <br><code><a href="layer_lstm.html">layer_lstm</a>()</code> <br><code><a href="layer_masking.html">layer_masking</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_max_pooling_1d.html">layer_max_pooling_1d</a>()</code> <br><code><a href="layer_max_pooling_2d.html">layer_max_pooling_2d</a>()</code> <br><code><a href="layer_max_pooling_3d.html">layer_max_pooling_3d</a>()</code> <br><code><a href="layer_maximum.html">layer_maximum</a>()</code> <br><code><a href="layer_mel_spectrogram.html">layer_mel_spectrogram</a>()</code> <br><code><a href="layer_minimum.html">layer_minimum</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_multi_head_attention.html">layer_multi_head_attention</a>()</code> <br><code><a href="layer_multiply.html">layer_multiply</a>()</code> <br><code><a href="layer_normalization.html">layer_normalization</a>()</code> <br><code><a href="layer_permute.html">layer_permute</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_brightness.html">layer_random_brightness</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_contrast.html">layer_random_contrast</a>()</code> <br><code><a href="layer_random_crop.html">layer_random_crop</a>()</code> <br><code><a href="layer_random_elastic_transform.html">layer_random_elastic_transform</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_flip.html">layer_random_flip</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_rotation.html">layer_random_rotation</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_random_translation.html">layer_random_translation</a>()</code> <br><code><a href="layer_random_zoom.html">layer_random_zoom</a>()</code> <br><code><a href="layer_repeat_vector.html">layer_repeat_vector</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_reshape.html">layer_reshape</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_rms_normalization.html">layer_rms_normalization</a>()</code> <br><code><a href="layer_rnn.html">layer_rnn</a>()</code> <br><code><a href="layer_separable_conv_1d.html">layer_separable_conv_1d</a>()</code> <br><code><a href="layer_separable_conv_2d.html">layer_separable_conv_2d</a>()</code> <br><code><a href="layer_simple_rnn.html">layer_simple_rnn</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br><code><a href="layer_spatial_dropout_1d.html">layer_spatial_dropout_1d</a>()</code> <br><code><a href="layer_spatial_dropout_2d.html">layer_spatial_dropout_2d</a>()</code> <br><code><a href="layer_spatial_dropout_3d.html">layer_spatial_dropout_3d</a>()</code> <br><code><a href="layer_spectral_normalization.html">layer_spectral_normalization</a>()</code> <br><code><a href="layer_stft_spectrogram.html">layer_stft_spectrogram</a>()</code> <br><code><a href="layer_string_lookup.html">layer_string_lookup</a>()</code> <br><code><a href="layer_subtract.html">layer_subtract</a>()</code> <br><code><a href="layer_text_vectorization.html">layer_text_vectorization</a>()</code> <br><code><a href="layer_tfsm.html">layer_tfsm</a>()</code> <br><code><a href="layer_time_distributed.html">layer_time_distributed</a>()</code> <br><code><a href="layer_torch_module_wrapper.html">layer_torch_module_wrapper</a>()</code> <br><code><a href="layer_unit_normalization.html">layer_unit_normalization</a>()</code> <br><code><a href="layer_upsampling_1d.html">layer_upsampling_1d</a>()</code> <br><code><a href="layer_upsampling_2d.html">layer_upsampling_2d</a>()</code> <br><code><a href="layer_upsampling_3d.html">layer_upsampling_3d</a>()</code> <br><code><a href="layer_zero_padding_1d.html">layer_zero_padding_1d</a>()</code> <br><code><a href="layer_zero_padding_2d.html">layer_zero_padding_2d</a>()</code> <br><code><a href="layer_zero_padding_3d.html">layer_zero_padding_3d</a>()</code> <br><code><a href="rnn_cell_gru.html">rnn_cell_gru</a>()</code> <br><code><a href="rnn_cell_lstm.html">rnn_cell_lstm</a>()</code> <br><code><a href="rnn_cell_simple.html">rnn_cell_simple</a>()</code> <br><code><a href="rnn_cells_stack.html">rnn_cells_stack</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -127,7 +127,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/layer_global_average_pooling_1d.html b/docs/dev/reference/layer_global_average_pooling_1d.html
index 150aee0b5..20b5524c0 100644
--- a/docs/dev/reference/layer_global_average_pooling_1d.html
+++ b/docs/dev/reference/layer_global_average_pooling_1d.html
@@ -8,7 +8,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -147,7 +147,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
     <div class="dont-index">
 <ul><li><p><a href="https://keras.io/api/layers/pooling_layers/global_average_pooling1d#globalaveragepooling1d-class" class="external-link">https://keras.io/api/layers/pooling_layers/global_average_pooling1d#globalaveragepooling1d-class</a></p></li>
 </ul><p>Other pooling layers: <br><code><a href="layer_average_pooling_1d.html">layer_average_pooling_1d</a>()</code> <br><code><a href="layer_average_pooling_2d.html">layer_average_pooling_2d</a>()</code> <br><code><a href="layer_average_pooling_3d.html">layer_average_pooling_3d</a>()</code> <br><code><a href="layer_global_average_pooling_2d.html">layer_global_average_pooling_2d</a>()</code> <br><code><a href="layer_global_average_pooling_3d.html">layer_global_average_pooling_3d</a>()</code> <br><code><a href="layer_global_max_pooling_1d.html">layer_global_max_pooling_1d</a>()</code> <br><code><a href="layer_global_max_pooling_2d.html">layer_global_max_pooling_2d</a>()</code> <br><code><a href="layer_global_max_pooling_3d.html">layer_global_max_pooling_3d</a>()</code> <br><code><a href="layer_max_pooling_1d.html">layer_max_pooling_1d</a>()</code> <br><code><a href="layer_max_pooling_2d.html">layer_max_pooling_2d</a>()</code> <br><code><a href="layer_max_pooling_3d.html">layer_max_pooling_3d</a>()</code> <br></p>
-<p>Other layers: <br><code><a href="Layer.html">Layer</a>()</code> <br><code><a href="layer_activation.html">layer_activation</a>()</code> <br><code><a href="layer_activation_elu.html">layer_activation_elu</a>()</code> <br><code><a href="layer_activation_leaky_relu.html">layer_activation_leaky_relu</a>()</code> <br><code><a href="layer_activation_parametric_relu.html">layer_activation_parametric_relu</a>()</code> <br><code><a href="layer_activation_relu.html">layer_activation_relu</a>()</code> <br><code><a href="layer_activation_softmax.html">layer_activation_softmax</a>()</code> <br><code><a href="layer_activity_regularization.html">layer_activity_regularization</a>()</code> <br><code><a href="layer_add.html">layer_add</a>()</code> <br><code><a href="layer_additive_attention.html">layer_additive_attention</a>()</code> <br><code><a href="layer_alpha_dropout.html">layer_alpha_dropout</a>()</code> <br><code><a href="layer_attention.html">layer_attention</a>()</code> <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_average.html">layer_average</a>()</code> <br><code><a href="layer_average_pooling_1d.html">layer_average_pooling_1d</a>()</code> <br><code><a href="layer_average_pooling_2d.html">layer_average_pooling_2d</a>()</code> <br><code><a href="layer_average_pooling_3d.html">layer_average_pooling_3d</a>()</code> <br><code><a href="layer_batch_normalization.html">layer_batch_normalization</a>()</code> <br><code><a href="layer_bidirectional.html">layer_bidirectional</a>()</code> <br><code><a href="layer_category_encoding.html">layer_category_encoding</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_concatenate.html">layer_concatenate</a>()</code> <br><code><a href="layer_conv_1d.html">layer_conv_1d</a>()</code> <br><code><a href="layer_conv_1d_transpose.html">layer_conv_1d_transpose</a>()</code> <br><code><a href="layer_conv_2d.html">layer_conv_2d</a>()</code> <br><code><a href="layer_conv_2d_transpose.html">layer_conv_2d_transpose</a>()</code> <br><code><a href="layer_conv_3d.html">layer_conv_3d</a>()</code> <br><code><a href="layer_conv_3d_transpose.html">layer_conv_3d_transpose</a>()</code> <br><code><a href="layer_conv_lstm_1d.html">layer_conv_lstm_1d</a>()</code> <br><code><a href="layer_conv_lstm_2d.html">layer_conv_lstm_2d</a>()</code> <br><code><a href="layer_conv_lstm_3d.html">layer_conv_lstm_3d</a>()</code> <br><code><a href="layer_cropping_1d.html">layer_cropping_1d</a>()</code> <br><code><a href="layer_cropping_2d.html">layer_cropping_2d</a>()</code> <br><code><a href="layer_cropping_3d.html">layer_cropping_3d</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_dense.html">layer_dense</a>()</code> <br><code><a href="layer_depthwise_conv_1d.html">layer_depthwise_conv_1d</a>()</code> <br><code><a href="layer_depthwise_conv_2d.html">layer_depthwise_conv_2d</a>()</code> <br><code><a href="layer_discretization.html">layer_discretization</a>()</code> <br><code><a href="layer_dot.html">layer_dot</a>()</code> <br><code><a href="layer_dropout.html">layer_dropout</a>()</code> <br><code><a href="layer_einsum_dense.html">layer_einsum_dense</a>()</code> <br><code><a href="layer_embedding.html">layer_embedding</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_feature_space.html">layer_feature_space</a>()</code> <br><code><a href="layer_flatten.html">layer_flatten</a>()</code> <br><code><a href="layer_flax_module_wrapper.html">layer_flax_module_wrapper</a>()</code> <br><code><a href="layer_gaussian_dropout.html">layer_gaussian_dropout</a>()</code> <br><code><a href="layer_gaussian_noise.html">layer_gaussian_noise</a>()</code> <br><code><a href="layer_global_average_pooling_2d.html">layer_global_average_pooling_2d</a>()</code> <br><code><a href="layer_global_average_pooling_3d.html">layer_global_average_pooling_3d</a>()</code> <br><code><a href="layer_global_max_pooling_1d.html">layer_global_max_pooling_1d</a>()</code> <br><code><a href="layer_global_max_pooling_2d.html">layer_global_max_pooling_2d</a>()</code> <br><code><a href="layer_global_max_pooling_3d.html">layer_global_max_pooling_3d</a>()</code> <br><code><a href="layer_group_normalization.html">layer_group_normalization</a>()</code> <br><code><a href="layer_group_query_attention.html">layer_group_query_attention</a>()</code> <br><code><a href="layer_gru.html">layer_gru</a>()</code> <br><code><a href="layer_hashed_crossing.html">layer_hashed_crossing</a>()</code> <br><code><a href="layer_hashing.html">layer_hashing</a>()</code> <br><code><a href="layer_identity.html">layer_identity</a>()</code> <br><code><a href="layer_integer_lookup.html">layer_integer_lookup</a>()</code> <br><code><a href="layer_jax_model_wrapper.html">layer_jax_model_wrapper</a>()</code> <br><code><a href="layer_lambda.html">layer_lambda</a>()</code> <br><code><a href="layer_layer_normalization.html">layer_layer_normalization</a>()</code> <br><code><a href="layer_lstm.html">layer_lstm</a>()</code> <br><code><a href="layer_masking.html">layer_masking</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_max_pooling_1d.html">layer_max_pooling_1d</a>()</code> <br><code><a href="layer_max_pooling_2d.html">layer_max_pooling_2d</a>()</code> <br><code><a href="layer_max_pooling_3d.html">layer_max_pooling_3d</a>()</code> <br><code><a href="layer_maximum.html">layer_maximum</a>()</code> <br><code><a href="layer_mel_spectrogram.html">layer_mel_spectrogram</a>()</code> <br><code><a href="layer_minimum.html">layer_minimum</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_multi_head_attention.html">layer_multi_head_attention</a>()</code> <br><code><a href="layer_multiply.html">layer_multiply</a>()</code> <br><code><a href="layer_normalization.html">layer_normalization</a>()</code> <br><code><a href="layer_permute.html">layer_permute</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_brightness.html">layer_random_brightness</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_contrast.html">layer_random_contrast</a>()</code> <br><code><a href="layer_random_crop.html">layer_random_crop</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_flip.html">layer_random_flip</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_rotation.html">layer_random_rotation</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_random_translation.html">layer_random_translation</a>()</code> <br><code><a href="layer_random_zoom.html">layer_random_zoom</a>()</code> <br><code><a href="layer_repeat_vector.html">layer_repeat_vector</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_reshape.html">layer_reshape</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_rms_normalization.html">layer_rms_normalization</a>()</code> <br><code><a href="layer_rnn.html">layer_rnn</a>()</code> <br><code><a href="layer_separable_conv_1d.html">layer_separable_conv_1d</a>()</code> <br><code><a href="layer_separable_conv_2d.html">layer_separable_conv_2d</a>()</code> <br><code><a href="layer_simple_rnn.html">layer_simple_rnn</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br><code><a href="layer_spatial_dropout_1d.html">layer_spatial_dropout_1d</a>()</code> <br><code><a href="layer_spatial_dropout_2d.html">layer_spatial_dropout_2d</a>()</code> <br><code><a href="layer_spatial_dropout_3d.html">layer_spatial_dropout_3d</a>()</code> <br><code><a href="layer_spectral_normalization.html">layer_spectral_normalization</a>()</code> <br><code><a href="layer_stft_spectrogram.html">layer_stft_spectrogram</a>()</code> <br><code><a href="layer_string_lookup.html">layer_string_lookup</a>()</code> <br><code><a href="layer_subtract.html">layer_subtract</a>()</code> <br><code><a href="layer_text_vectorization.html">layer_text_vectorization</a>()</code> <br><code><a href="layer_tfsm.html">layer_tfsm</a>()</code> <br><code><a href="layer_time_distributed.html">layer_time_distributed</a>()</code> <br><code><a href="layer_torch_module_wrapper.html">layer_torch_module_wrapper</a>()</code> <br><code><a href="layer_unit_normalization.html">layer_unit_normalization</a>()</code> <br><code><a href="layer_upsampling_1d.html">layer_upsampling_1d</a>()</code> <br><code><a href="layer_upsampling_2d.html">layer_upsampling_2d</a>()</code> <br><code><a href="layer_upsampling_3d.html">layer_upsampling_3d</a>()</code> <br><code><a href="layer_zero_padding_1d.html">layer_zero_padding_1d</a>()</code> <br><code><a href="layer_zero_padding_2d.html">layer_zero_padding_2d</a>()</code> <br><code><a href="layer_zero_padding_3d.html">layer_zero_padding_3d</a>()</code> <br><code><a href="rnn_cell_gru.html">rnn_cell_gru</a>()</code> <br><code><a href="rnn_cell_lstm.html">rnn_cell_lstm</a>()</code> <br><code><a href="rnn_cell_simple.html">rnn_cell_simple</a>()</code> <br><code><a href="rnn_cells_stack.html">rnn_cells_stack</a>()</code> <br></p></div>
+<p>Other layers: <br><code><a href="Layer.html">Layer</a>()</code> <br><code><a href="layer_activation.html">layer_activation</a>()</code> <br><code><a href="layer_activation_elu.html">layer_activation_elu</a>()</code> <br><code><a href="layer_activation_leaky_relu.html">layer_activation_leaky_relu</a>()</code> <br><code><a href="layer_activation_parametric_relu.html">layer_activation_parametric_relu</a>()</code> <br><code><a href="layer_activation_relu.html">layer_activation_relu</a>()</code> <br><code><a href="layer_activation_softmax.html">layer_activation_softmax</a>()</code> <br><code><a href="layer_activity_regularization.html">layer_activity_regularization</a>()</code> <br><code><a href="layer_add.html">layer_add</a>()</code> <br><code><a href="layer_additive_attention.html">layer_additive_attention</a>()</code> <br><code><a href="layer_alpha_dropout.html">layer_alpha_dropout</a>()</code> <br><code><a href="layer_attention.html">layer_attention</a>()</code> <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_average.html">layer_average</a>()</code> <br><code><a href="layer_average_pooling_1d.html">layer_average_pooling_1d</a>()</code> <br><code><a href="layer_average_pooling_2d.html">layer_average_pooling_2d</a>()</code> <br><code><a href="layer_average_pooling_3d.html">layer_average_pooling_3d</a>()</code> <br><code><a href="layer_batch_normalization.html">layer_batch_normalization</a>()</code> <br><code><a href="layer_bidirectional.html">layer_bidirectional</a>()</code> <br><code><a href="layer_category_encoding.html">layer_category_encoding</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_concatenate.html">layer_concatenate</a>()</code> <br><code><a href="layer_conv_1d.html">layer_conv_1d</a>()</code> <br><code><a href="layer_conv_1d_transpose.html">layer_conv_1d_transpose</a>()</code> <br><code><a href="layer_conv_2d.html">layer_conv_2d</a>()</code> <br><code><a href="layer_conv_2d_transpose.html">layer_conv_2d_transpose</a>()</code> <br><code><a href="layer_conv_3d.html">layer_conv_3d</a>()</code> <br><code><a href="layer_conv_3d_transpose.html">layer_conv_3d_transpose</a>()</code> <br><code><a href="layer_conv_lstm_1d.html">layer_conv_lstm_1d</a>()</code> <br><code><a href="layer_conv_lstm_2d.html">layer_conv_lstm_2d</a>()</code> <br><code><a href="layer_conv_lstm_3d.html">layer_conv_lstm_3d</a>()</code> <br><code><a href="layer_cropping_1d.html">layer_cropping_1d</a>()</code> <br><code><a href="layer_cropping_2d.html">layer_cropping_2d</a>()</code> <br><code><a href="layer_cropping_3d.html">layer_cropping_3d</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_dense.html">layer_dense</a>()</code> <br><code><a href="layer_depthwise_conv_1d.html">layer_depthwise_conv_1d</a>()</code> <br><code><a href="layer_depthwise_conv_2d.html">layer_depthwise_conv_2d</a>()</code> <br><code><a href="layer_discretization.html">layer_discretization</a>()</code> <br><code><a href="layer_dot.html">layer_dot</a>()</code> <br><code><a href="layer_dropout.html">layer_dropout</a>()</code> <br><code><a href="layer_einsum_dense.html">layer_einsum_dense</a>()</code> <br><code><a href="layer_embedding.html">layer_embedding</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_feature_space.html">layer_feature_space</a>()</code> <br><code><a href="layer_flatten.html">layer_flatten</a>()</code> <br><code><a href="layer_flax_module_wrapper.html">layer_flax_module_wrapper</a>()</code> <br><code><a href="layer_gaussian_dropout.html">layer_gaussian_dropout</a>()</code> <br><code><a href="layer_gaussian_noise.html">layer_gaussian_noise</a>()</code> <br><code><a href="layer_global_average_pooling_2d.html">layer_global_average_pooling_2d</a>()</code> <br><code><a href="layer_global_average_pooling_3d.html">layer_global_average_pooling_3d</a>()</code> <br><code><a href="layer_global_max_pooling_1d.html">layer_global_max_pooling_1d</a>()</code> <br><code><a href="layer_global_max_pooling_2d.html">layer_global_max_pooling_2d</a>()</code> <br><code><a href="layer_global_max_pooling_3d.html">layer_global_max_pooling_3d</a>()</code> <br><code><a href="layer_group_normalization.html">layer_group_normalization</a>()</code> <br><code><a href="layer_group_query_attention.html">layer_group_query_attention</a>()</code> <br><code><a href="layer_gru.html">layer_gru</a>()</code> <br><code><a href="layer_hashed_crossing.html">layer_hashed_crossing</a>()</code> <br><code><a href="layer_hashing.html">layer_hashing</a>()</code> <br><code><a href="layer_identity.html">layer_identity</a>()</code> <br><code><a href="layer_integer_lookup.html">layer_integer_lookup</a>()</code> <br><code><a href="layer_jax_model_wrapper.html">layer_jax_model_wrapper</a>()</code> <br><code><a href="layer_lambda.html">layer_lambda</a>()</code> <br><code><a href="layer_layer_normalization.html">layer_layer_normalization</a>()</code> <br><code><a href="layer_lstm.html">layer_lstm</a>()</code> <br><code><a href="layer_masking.html">layer_masking</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_max_pooling_1d.html">layer_max_pooling_1d</a>()</code> <br><code><a href="layer_max_pooling_2d.html">layer_max_pooling_2d</a>()</code> <br><code><a href="layer_max_pooling_3d.html">layer_max_pooling_3d</a>()</code> <br><code><a href="layer_maximum.html">layer_maximum</a>()</code> <br><code><a href="layer_mel_spectrogram.html">layer_mel_spectrogram</a>()</code> <br><code><a href="layer_minimum.html">layer_minimum</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_multi_head_attention.html">layer_multi_head_attention</a>()</code> <br><code><a href="layer_multiply.html">layer_multiply</a>()</code> <br><code><a href="layer_normalization.html">layer_normalization</a>()</code> <br><code><a href="layer_permute.html">layer_permute</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_brightness.html">layer_random_brightness</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_contrast.html">layer_random_contrast</a>()</code> <br><code><a href="layer_random_crop.html">layer_random_crop</a>()</code> <br><code><a href="layer_random_elastic_transform.html">layer_random_elastic_transform</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_flip.html">layer_random_flip</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_rotation.html">layer_random_rotation</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_random_translation.html">layer_random_translation</a>()</code> <br><code><a href="layer_random_zoom.html">layer_random_zoom</a>()</code> <br><code><a href="layer_repeat_vector.html">layer_repeat_vector</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_reshape.html">layer_reshape</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_rms_normalization.html">layer_rms_normalization</a>()</code> <br><code><a href="layer_rnn.html">layer_rnn</a>()</code> <br><code><a href="layer_separable_conv_1d.html">layer_separable_conv_1d</a>()</code> <br><code><a href="layer_separable_conv_2d.html">layer_separable_conv_2d</a>()</code> <br><code><a href="layer_simple_rnn.html">layer_simple_rnn</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br><code><a href="layer_spatial_dropout_1d.html">layer_spatial_dropout_1d</a>()</code> <br><code><a href="layer_spatial_dropout_2d.html">layer_spatial_dropout_2d</a>()</code> <br><code><a href="layer_spatial_dropout_3d.html">layer_spatial_dropout_3d</a>()</code> <br><code><a href="layer_spectral_normalization.html">layer_spectral_normalization</a>()</code> <br><code><a href="layer_stft_spectrogram.html">layer_stft_spectrogram</a>()</code> <br><code><a href="layer_string_lookup.html">layer_string_lookup</a>()</code> <br><code><a href="layer_subtract.html">layer_subtract</a>()</code> <br><code><a href="layer_text_vectorization.html">layer_text_vectorization</a>()</code> <br><code><a href="layer_tfsm.html">layer_tfsm</a>()</code> <br><code><a href="layer_time_distributed.html">layer_time_distributed</a>()</code> <br><code><a href="layer_torch_module_wrapper.html">layer_torch_module_wrapper</a>()</code> <br><code><a href="layer_unit_normalization.html">layer_unit_normalization</a>()</code> <br><code><a href="layer_upsampling_1d.html">layer_upsampling_1d</a>()</code> <br><code><a href="layer_upsampling_2d.html">layer_upsampling_2d</a>()</code> <br><code><a href="layer_upsampling_3d.html">layer_upsampling_3d</a>()</code> <br><code><a href="layer_zero_padding_1d.html">layer_zero_padding_1d</a>()</code> <br><code><a href="layer_zero_padding_2d.html">layer_zero_padding_2d</a>()</code> <br><code><a href="layer_zero_padding_3d.html">layer_zero_padding_3d</a>()</code> <br><code><a href="rnn_cell_gru.html">rnn_cell_gru</a>()</code> <br><code><a href="rnn_cell_lstm.html">rnn_cell_lstm</a>()</code> <br><code><a href="rnn_cell_simple.html">rnn_cell_simple</a>()</code> <br><code><a href="rnn_cells_stack.html">rnn_cells_stack</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -159,7 +159,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/layer_global_average_pooling_2d.html b/docs/dev/reference/layer_global_average_pooling_2d.html
index 3ebd81027..da3f5449c 100644
--- a/docs/dev/reference/layer_global_average_pooling_2d.html
+++ b/docs/dev/reference/layer_global_average_pooling_2d.html
@@ -8,7 +8,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -141,7 +141,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
     <div class="dont-index">
 <ul><li><p><a href="https://keras.io/api/layers/pooling_layers/global_average_pooling2d#globalaveragepooling2d-class" class="external-link">https://keras.io/api/layers/pooling_layers/global_average_pooling2d#globalaveragepooling2d-class</a></p></li>
 </ul><p>Other pooling layers: <br><code><a href="layer_average_pooling_1d.html">layer_average_pooling_1d</a>()</code> <br><code><a href="layer_average_pooling_2d.html">layer_average_pooling_2d</a>()</code> <br><code><a href="layer_average_pooling_3d.html">layer_average_pooling_3d</a>()</code> <br><code><a href="layer_global_average_pooling_1d.html">layer_global_average_pooling_1d</a>()</code> <br><code><a href="layer_global_average_pooling_3d.html">layer_global_average_pooling_3d</a>()</code> <br><code><a href="layer_global_max_pooling_1d.html">layer_global_max_pooling_1d</a>()</code> <br><code><a href="layer_global_max_pooling_2d.html">layer_global_max_pooling_2d</a>()</code> <br><code><a href="layer_global_max_pooling_3d.html">layer_global_max_pooling_3d</a>()</code> <br><code><a href="layer_max_pooling_1d.html">layer_max_pooling_1d</a>()</code> <br><code><a href="layer_max_pooling_2d.html">layer_max_pooling_2d</a>()</code> <br><code><a href="layer_max_pooling_3d.html">layer_max_pooling_3d</a>()</code> <br></p>
-<p>Other layers: <br><code><a href="Layer.html">Layer</a>()</code> <br><code><a href="layer_activation.html">layer_activation</a>()</code> <br><code><a href="layer_activation_elu.html">layer_activation_elu</a>()</code> <br><code><a href="layer_activation_leaky_relu.html">layer_activation_leaky_relu</a>()</code> <br><code><a href="layer_activation_parametric_relu.html">layer_activation_parametric_relu</a>()</code> <br><code><a href="layer_activation_relu.html">layer_activation_relu</a>()</code> <br><code><a href="layer_activation_softmax.html">layer_activation_softmax</a>()</code> <br><code><a href="layer_activity_regularization.html">layer_activity_regularization</a>()</code> <br><code><a href="layer_add.html">layer_add</a>()</code> <br><code><a href="layer_additive_attention.html">layer_additive_attention</a>()</code> <br><code><a href="layer_alpha_dropout.html">layer_alpha_dropout</a>()</code> <br><code><a href="layer_attention.html">layer_attention</a>()</code> <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_average.html">layer_average</a>()</code> <br><code><a href="layer_average_pooling_1d.html">layer_average_pooling_1d</a>()</code> <br><code><a href="layer_average_pooling_2d.html">layer_average_pooling_2d</a>()</code> <br><code><a href="layer_average_pooling_3d.html">layer_average_pooling_3d</a>()</code> <br><code><a href="layer_batch_normalization.html">layer_batch_normalization</a>()</code> <br><code><a href="layer_bidirectional.html">layer_bidirectional</a>()</code> <br><code><a href="layer_category_encoding.html">layer_category_encoding</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_concatenate.html">layer_concatenate</a>()</code> <br><code><a href="layer_conv_1d.html">layer_conv_1d</a>()</code> <br><code><a href="layer_conv_1d_transpose.html">layer_conv_1d_transpose</a>()</code> <br><code><a href="layer_conv_2d.html">layer_conv_2d</a>()</code> <br><code><a href="layer_conv_2d_transpose.html">layer_conv_2d_transpose</a>()</code> <br><code><a href="layer_conv_3d.html">layer_conv_3d</a>()</code> <br><code><a href="layer_conv_3d_transpose.html">layer_conv_3d_transpose</a>()</code> <br><code><a href="layer_conv_lstm_1d.html">layer_conv_lstm_1d</a>()</code> <br><code><a href="layer_conv_lstm_2d.html">layer_conv_lstm_2d</a>()</code> <br><code><a href="layer_conv_lstm_3d.html">layer_conv_lstm_3d</a>()</code> <br><code><a href="layer_cropping_1d.html">layer_cropping_1d</a>()</code> <br><code><a href="layer_cropping_2d.html">layer_cropping_2d</a>()</code> <br><code><a href="layer_cropping_3d.html">layer_cropping_3d</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_dense.html">layer_dense</a>()</code> <br><code><a href="layer_depthwise_conv_1d.html">layer_depthwise_conv_1d</a>()</code> <br><code><a href="layer_depthwise_conv_2d.html">layer_depthwise_conv_2d</a>()</code> <br><code><a href="layer_discretization.html">layer_discretization</a>()</code> <br><code><a href="layer_dot.html">layer_dot</a>()</code> <br><code><a href="layer_dropout.html">layer_dropout</a>()</code> <br><code><a href="layer_einsum_dense.html">layer_einsum_dense</a>()</code> <br><code><a href="layer_embedding.html">layer_embedding</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_feature_space.html">layer_feature_space</a>()</code> <br><code><a href="layer_flatten.html">layer_flatten</a>()</code> <br><code><a href="layer_flax_module_wrapper.html">layer_flax_module_wrapper</a>()</code> <br><code><a href="layer_gaussian_dropout.html">layer_gaussian_dropout</a>()</code> <br><code><a href="layer_gaussian_noise.html">layer_gaussian_noise</a>()</code> <br><code><a href="layer_global_average_pooling_1d.html">layer_global_average_pooling_1d</a>()</code> <br><code><a href="layer_global_average_pooling_3d.html">layer_global_average_pooling_3d</a>()</code> <br><code><a href="layer_global_max_pooling_1d.html">layer_global_max_pooling_1d</a>()</code> <br><code><a href="layer_global_max_pooling_2d.html">layer_global_max_pooling_2d</a>()</code> <br><code><a href="layer_global_max_pooling_3d.html">layer_global_max_pooling_3d</a>()</code> <br><code><a href="layer_group_normalization.html">layer_group_normalization</a>()</code> <br><code><a href="layer_group_query_attention.html">layer_group_query_attention</a>()</code> <br><code><a href="layer_gru.html">layer_gru</a>()</code> <br><code><a href="layer_hashed_crossing.html">layer_hashed_crossing</a>()</code> <br><code><a href="layer_hashing.html">layer_hashing</a>()</code> <br><code><a href="layer_identity.html">layer_identity</a>()</code> <br><code><a href="layer_integer_lookup.html">layer_integer_lookup</a>()</code> <br><code><a href="layer_jax_model_wrapper.html">layer_jax_model_wrapper</a>()</code> <br><code><a href="layer_lambda.html">layer_lambda</a>()</code> <br><code><a href="layer_layer_normalization.html">layer_layer_normalization</a>()</code> <br><code><a href="layer_lstm.html">layer_lstm</a>()</code> <br><code><a href="layer_masking.html">layer_masking</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_max_pooling_1d.html">layer_max_pooling_1d</a>()</code> <br><code><a href="layer_max_pooling_2d.html">layer_max_pooling_2d</a>()</code> <br><code><a href="layer_max_pooling_3d.html">layer_max_pooling_3d</a>()</code> <br><code><a href="layer_maximum.html">layer_maximum</a>()</code> <br><code><a href="layer_mel_spectrogram.html">layer_mel_spectrogram</a>()</code> <br><code><a href="layer_minimum.html">layer_minimum</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_multi_head_attention.html">layer_multi_head_attention</a>()</code> <br><code><a href="layer_multiply.html">layer_multiply</a>()</code> <br><code><a href="layer_normalization.html">layer_normalization</a>()</code> <br><code><a href="layer_permute.html">layer_permute</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_brightness.html">layer_random_brightness</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_contrast.html">layer_random_contrast</a>()</code> <br><code><a href="layer_random_crop.html">layer_random_crop</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_flip.html">layer_random_flip</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_rotation.html">layer_random_rotation</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_random_translation.html">layer_random_translation</a>()</code> <br><code><a href="layer_random_zoom.html">layer_random_zoom</a>()</code> <br><code><a href="layer_repeat_vector.html">layer_repeat_vector</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_reshape.html">layer_reshape</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_rms_normalization.html">layer_rms_normalization</a>()</code> <br><code><a href="layer_rnn.html">layer_rnn</a>()</code> <br><code><a href="layer_separable_conv_1d.html">layer_separable_conv_1d</a>()</code> <br><code><a href="layer_separable_conv_2d.html">layer_separable_conv_2d</a>()</code> <br><code><a href="layer_simple_rnn.html">layer_simple_rnn</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br><code><a href="layer_spatial_dropout_1d.html">layer_spatial_dropout_1d</a>()</code> <br><code><a href="layer_spatial_dropout_2d.html">layer_spatial_dropout_2d</a>()</code> <br><code><a href="layer_spatial_dropout_3d.html">layer_spatial_dropout_3d</a>()</code> <br><code><a href="layer_spectral_normalization.html">layer_spectral_normalization</a>()</code> <br><code><a href="layer_stft_spectrogram.html">layer_stft_spectrogram</a>()</code> <br><code><a href="layer_string_lookup.html">layer_string_lookup</a>()</code> <br><code><a href="layer_subtract.html">layer_subtract</a>()</code> <br><code><a href="layer_text_vectorization.html">layer_text_vectorization</a>()</code> <br><code><a href="layer_tfsm.html">layer_tfsm</a>()</code> <br><code><a href="layer_time_distributed.html">layer_time_distributed</a>()</code> <br><code><a href="layer_torch_module_wrapper.html">layer_torch_module_wrapper</a>()</code> <br><code><a href="layer_unit_normalization.html">layer_unit_normalization</a>()</code> <br><code><a href="layer_upsampling_1d.html">layer_upsampling_1d</a>()</code> <br><code><a href="layer_upsampling_2d.html">layer_upsampling_2d</a>()</code> <br><code><a href="layer_upsampling_3d.html">layer_upsampling_3d</a>()</code> <br><code><a href="layer_zero_padding_1d.html">layer_zero_padding_1d</a>()</code> <br><code><a href="layer_zero_padding_2d.html">layer_zero_padding_2d</a>()</code> <br><code><a href="layer_zero_padding_3d.html">layer_zero_padding_3d</a>()</code> <br><code><a href="rnn_cell_gru.html">rnn_cell_gru</a>()</code> <br><code><a href="rnn_cell_lstm.html">rnn_cell_lstm</a>()</code> <br><code><a href="rnn_cell_simple.html">rnn_cell_simple</a>()</code> <br><code><a href="rnn_cells_stack.html">rnn_cells_stack</a>()</code> <br></p></div>
+<p>Other layers: <br><code><a href="Layer.html">Layer</a>()</code> <br><code><a href="layer_activation.html">layer_activation</a>()</code> <br><code><a href="layer_activation_elu.html">layer_activation_elu</a>()</code> <br><code><a href="layer_activation_leaky_relu.html">layer_activation_leaky_relu</a>()</code> <br><code><a href="layer_activation_parametric_relu.html">layer_activation_parametric_relu</a>()</code> <br><code><a href="layer_activation_relu.html">layer_activation_relu</a>()</code> <br><code><a href="layer_activation_softmax.html">layer_activation_softmax</a>()</code> <br><code><a href="layer_activity_regularization.html">layer_activity_regularization</a>()</code> <br><code><a href="layer_add.html">layer_add</a>()</code> <br><code><a href="layer_additive_attention.html">layer_additive_attention</a>()</code> <br><code><a href="layer_alpha_dropout.html">layer_alpha_dropout</a>()</code> <br><code><a href="layer_attention.html">layer_attention</a>()</code> <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_average.html">layer_average</a>()</code> <br><code><a href="layer_average_pooling_1d.html">layer_average_pooling_1d</a>()</code> <br><code><a href="layer_average_pooling_2d.html">layer_average_pooling_2d</a>()</code> <br><code><a href="layer_average_pooling_3d.html">layer_average_pooling_3d</a>()</code> <br><code><a href="layer_batch_normalization.html">layer_batch_normalization</a>()</code> <br><code><a href="layer_bidirectional.html">layer_bidirectional</a>()</code> <br><code><a href="layer_category_encoding.html">layer_category_encoding</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_concatenate.html">layer_concatenate</a>()</code> <br><code><a href="layer_conv_1d.html">layer_conv_1d</a>()</code> <br><code><a href="layer_conv_1d_transpose.html">layer_conv_1d_transpose</a>()</code> <br><code><a href="layer_conv_2d.html">layer_conv_2d</a>()</code> <br><code><a href="layer_conv_2d_transpose.html">layer_conv_2d_transpose</a>()</code> <br><code><a href="layer_conv_3d.html">layer_conv_3d</a>()</code> <br><code><a href="layer_conv_3d_transpose.html">layer_conv_3d_transpose</a>()</code> <br><code><a href="layer_conv_lstm_1d.html">layer_conv_lstm_1d</a>()</code> <br><code><a href="layer_conv_lstm_2d.html">layer_conv_lstm_2d</a>()</code> <br><code><a href="layer_conv_lstm_3d.html">layer_conv_lstm_3d</a>()</code> <br><code><a href="layer_cropping_1d.html">layer_cropping_1d</a>()</code> <br><code><a href="layer_cropping_2d.html">layer_cropping_2d</a>()</code> <br><code><a href="layer_cropping_3d.html">layer_cropping_3d</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_dense.html">layer_dense</a>()</code> <br><code><a href="layer_depthwise_conv_1d.html">layer_depthwise_conv_1d</a>()</code> <br><code><a href="layer_depthwise_conv_2d.html">layer_depthwise_conv_2d</a>()</code> <br><code><a href="layer_discretization.html">layer_discretization</a>()</code> <br><code><a href="layer_dot.html">layer_dot</a>()</code> <br><code><a href="layer_dropout.html">layer_dropout</a>()</code> <br><code><a href="layer_einsum_dense.html">layer_einsum_dense</a>()</code> <br><code><a href="layer_embedding.html">layer_embedding</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_feature_space.html">layer_feature_space</a>()</code> <br><code><a href="layer_flatten.html">layer_flatten</a>()</code> <br><code><a href="layer_flax_module_wrapper.html">layer_flax_module_wrapper</a>()</code> <br><code><a href="layer_gaussian_dropout.html">layer_gaussian_dropout</a>()</code> <br><code><a href="layer_gaussian_noise.html">layer_gaussian_noise</a>()</code> <br><code><a href="layer_global_average_pooling_1d.html">layer_global_average_pooling_1d</a>()</code> <br><code><a href="layer_global_average_pooling_3d.html">layer_global_average_pooling_3d</a>()</code> <br><code><a href="layer_global_max_pooling_1d.html">layer_global_max_pooling_1d</a>()</code> <br><code><a href="layer_global_max_pooling_2d.html">layer_global_max_pooling_2d</a>()</code> <br><code><a href="layer_global_max_pooling_3d.html">layer_global_max_pooling_3d</a>()</code> <br><code><a href="layer_group_normalization.html">layer_group_normalization</a>()</code> <br><code><a href="layer_group_query_attention.html">layer_group_query_attention</a>()</code> <br><code><a href="layer_gru.html">layer_gru</a>()</code> <br><code><a href="layer_hashed_crossing.html">layer_hashed_crossing</a>()</code> <br><code><a href="layer_hashing.html">layer_hashing</a>()</code> <br><code><a href="layer_identity.html">layer_identity</a>()</code> <br><code><a href="layer_integer_lookup.html">layer_integer_lookup</a>()</code> <br><code><a href="layer_jax_model_wrapper.html">layer_jax_model_wrapper</a>()</code> <br><code><a href="layer_lambda.html">layer_lambda</a>()</code> <br><code><a href="layer_layer_normalization.html">layer_layer_normalization</a>()</code> <br><code><a href="layer_lstm.html">layer_lstm</a>()</code> <br><code><a href="layer_masking.html">layer_masking</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_max_pooling_1d.html">layer_max_pooling_1d</a>()</code> <br><code><a href="layer_max_pooling_2d.html">layer_max_pooling_2d</a>()</code> <br><code><a href="layer_max_pooling_3d.html">layer_max_pooling_3d</a>()</code> <br><code><a href="layer_maximum.html">layer_maximum</a>()</code> <br><code><a href="layer_mel_spectrogram.html">layer_mel_spectrogram</a>()</code> <br><code><a href="layer_minimum.html">layer_minimum</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_multi_head_attention.html">layer_multi_head_attention</a>()</code> <br><code><a href="layer_multiply.html">layer_multiply</a>()</code> <br><code><a href="layer_normalization.html">layer_normalization</a>()</code> <br><code><a href="layer_permute.html">layer_permute</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_brightness.html">layer_random_brightness</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_contrast.html">layer_random_contrast</a>()</code> <br><code><a href="layer_random_crop.html">layer_random_crop</a>()</code> <br><code><a href="layer_random_elastic_transform.html">layer_random_elastic_transform</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_flip.html">layer_random_flip</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_rotation.html">layer_random_rotation</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_random_translation.html">layer_random_translation</a>()</code> <br><code><a href="layer_random_zoom.html">layer_random_zoom</a>()</code> <br><code><a href="layer_repeat_vector.html">layer_repeat_vector</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_reshape.html">layer_reshape</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_rms_normalization.html">layer_rms_normalization</a>()</code> <br><code><a href="layer_rnn.html">layer_rnn</a>()</code> <br><code><a href="layer_separable_conv_1d.html">layer_separable_conv_1d</a>()</code> <br><code><a href="layer_separable_conv_2d.html">layer_separable_conv_2d</a>()</code> <br><code><a href="layer_simple_rnn.html">layer_simple_rnn</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br><code><a href="layer_spatial_dropout_1d.html">layer_spatial_dropout_1d</a>()</code> <br><code><a href="layer_spatial_dropout_2d.html">layer_spatial_dropout_2d</a>()</code> <br><code><a href="layer_spatial_dropout_3d.html">layer_spatial_dropout_3d</a>()</code> <br><code><a href="layer_spectral_normalization.html">layer_spectral_normalization</a>()</code> <br><code><a href="layer_stft_spectrogram.html">layer_stft_spectrogram</a>()</code> <br><code><a href="layer_string_lookup.html">layer_string_lookup</a>()</code> <br><code><a href="layer_subtract.html">layer_subtract</a>()</code> <br><code><a href="layer_text_vectorization.html">layer_text_vectorization</a>()</code> <br><code><a href="layer_tfsm.html">layer_tfsm</a>()</code> <br><code><a href="layer_time_distributed.html">layer_time_distributed</a>()</code> <br><code><a href="layer_torch_module_wrapper.html">layer_torch_module_wrapper</a>()</code> <br><code><a href="layer_unit_normalization.html">layer_unit_normalization</a>()</code> <br><code><a href="layer_upsampling_1d.html">layer_upsampling_1d</a>()</code> <br><code><a href="layer_upsampling_2d.html">layer_upsampling_2d</a>()</code> <br><code><a href="layer_upsampling_3d.html">layer_upsampling_3d</a>()</code> <br><code><a href="layer_zero_padding_1d.html">layer_zero_padding_1d</a>()</code> <br><code><a href="layer_zero_padding_2d.html">layer_zero_padding_2d</a>()</code> <br><code><a href="layer_zero_padding_3d.html">layer_zero_padding_3d</a>()</code> <br><code><a href="rnn_cell_gru.html">rnn_cell_gru</a>()</code> <br><code><a href="rnn_cell_lstm.html">rnn_cell_lstm</a>()</code> <br><code><a href="rnn_cell_simple.html">rnn_cell_simple</a>()</code> <br><code><a href="rnn_cells_stack.html">rnn_cells_stack</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -153,7 +153,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/layer_global_average_pooling_3d.html b/docs/dev/reference/layer_global_average_pooling_3d.html
index bc2e1317c..21da9c5b7 100644
--- a/docs/dev/reference/layer_global_average_pooling_3d.html
+++ b/docs/dev/reference/layer_global_average_pooling_3d.html
@@ -8,7 +8,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -142,7 +142,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
     <div class="dont-index">
 <ul><li><p><a href="https://keras.io/api/layers/pooling_layers/global_average_pooling3d#globalaveragepooling3d-class" class="external-link">https://keras.io/api/layers/pooling_layers/global_average_pooling3d#globalaveragepooling3d-class</a></p></li>
 </ul><p>Other pooling layers: <br><code><a href="layer_average_pooling_1d.html">layer_average_pooling_1d</a>()</code> <br><code><a href="layer_average_pooling_2d.html">layer_average_pooling_2d</a>()</code> <br><code><a href="layer_average_pooling_3d.html">layer_average_pooling_3d</a>()</code> <br><code><a href="layer_global_average_pooling_1d.html">layer_global_average_pooling_1d</a>()</code> <br><code><a href="layer_global_average_pooling_2d.html">layer_global_average_pooling_2d</a>()</code> <br><code><a href="layer_global_max_pooling_1d.html">layer_global_max_pooling_1d</a>()</code> <br><code><a href="layer_global_max_pooling_2d.html">layer_global_max_pooling_2d</a>()</code> <br><code><a href="layer_global_max_pooling_3d.html">layer_global_max_pooling_3d</a>()</code> <br><code><a href="layer_max_pooling_1d.html">layer_max_pooling_1d</a>()</code> <br><code><a href="layer_max_pooling_2d.html">layer_max_pooling_2d</a>()</code> <br><code><a href="layer_max_pooling_3d.html">layer_max_pooling_3d</a>()</code> <br></p>
-<p>Other layers: <br><code><a href="Layer.html">Layer</a>()</code> <br><code><a href="layer_activation.html">layer_activation</a>()</code> <br><code><a href="layer_activation_elu.html">layer_activation_elu</a>()</code> <br><code><a href="layer_activation_leaky_relu.html">layer_activation_leaky_relu</a>()</code> <br><code><a href="layer_activation_parametric_relu.html">layer_activation_parametric_relu</a>()</code> <br><code><a href="layer_activation_relu.html">layer_activation_relu</a>()</code> <br><code><a href="layer_activation_softmax.html">layer_activation_softmax</a>()</code> <br><code><a href="layer_activity_regularization.html">layer_activity_regularization</a>()</code> <br><code><a href="layer_add.html">layer_add</a>()</code> <br><code><a href="layer_additive_attention.html">layer_additive_attention</a>()</code> <br><code><a href="layer_alpha_dropout.html">layer_alpha_dropout</a>()</code> <br><code><a href="layer_attention.html">layer_attention</a>()</code> <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_average.html">layer_average</a>()</code> <br><code><a href="layer_average_pooling_1d.html">layer_average_pooling_1d</a>()</code> <br><code><a href="layer_average_pooling_2d.html">layer_average_pooling_2d</a>()</code> <br><code><a href="layer_average_pooling_3d.html">layer_average_pooling_3d</a>()</code> <br><code><a href="layer_batch_normalization.html">layer_batch_normalization</a>()</code> <br><code><a href="layer_bidirectional.html">layer_bidirectional</a>()</code> <br><code><a href="layer_category_encoding.html">layer_category_encoding</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_concatenate.html">layer_concatenate</a>()</code> <br><code><a href="layer_conv_1d.html">layer_conv_1d</a>()</code> <br><code><a href="layer_conv_1d_transpose.html">layer_conv_1d_transpose</a>()</code> <br><code><a href="layer_conv_2d.html">layer_conv_2d</a>()</code> <br><code><a href="layer_conv_2d_transpose.html">layer_conv_2d_transpose</a>()</code> <br><code><a href="layer_conv_3d.html">layer_conv_3d</a>()</code> <br><code><a href="layer_conv_3d_transpose.html">layer_conv_3d_transpose</a>()</code> <br><code><a href="layer_conv_lstm_1d.html">layer_conv_lstm_1d</a>()</code> <br><code><a href="layer_conv_lstm_2d.html">layer_conv_lstm_2d</a>()</code> <br><code><a href="layer_conv_lstm_3d.html">layer_conv_lstm_3d</a>()</code> <br><code><a href="layer_cropping_1d.html">layer_cropping_1d</a>()</code> <br><code><a href="layer_cropping_2d.html">layer_cropping_2d</a>()</code> <br><code><a href="layer_cropping_3d.html">layer_cropping_3d</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_dense.html">layer_dense</a>()</code> <br><code><a href="layer_depthwise_conv_1d.html">layer_depthwise_conv_1d</a>()</code> <br><code><a href="layer_depthwise_conv_2d.html">layer_depthwise_conv_2d</a>()</code> <br><code><a href="layer_discretization.html">layer_discretization</a>()</code> <br><code><a href="layer_dot.html">layer_dot</a>()</code> <br><code><a href="layer_dropout.html">layer_dropout</a>()</code> <br><code><a href="layer_einsum_dense.html">layer_einsum_dense</a>()</code> <br><code><a href="layer_embedding.html">layer_embedding</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_feature_space.html">layer_feature_space</a>()</code> <br><code><a href="layer_flatten.html">layer_flatten</a>()</code> <br><code><a href="layer_flax_module_wrapper.html">layer_flax_module_wrapper</a>()</code> <br><code><a href="layer_gaussian_dropout.html">layer_gaussian_dropout</a>()</code> <br><code><a href="layer_gaussian_noise.html">layer_gaussian_noise</a>()</code> <br><code><a href="layer_global_average_pooling_1d.html">layer_global_average_pooling_1d</a>()</code> <br><code><a href="layer_global_average_pooling_2d.html">layer_global_average_pooling_2d</a>()</code> <br><code><a href="layer_global_max_pooling_1d.html">layer_global_max_pooling_1d</a>()</code> <br><code><a href="layer_global_max_pooling_2d.html">layer_global_max_pooling_2d</a>()</code> <br><code><a href="layer_global_max_pooling_3d.html">layer_global_max_pooling_3d</a>()</code> <br><code><a href="layer_group_normalization.html">layer_group_normalization</a>()</code> <br><code><a href="layer_group_query_attention.html">layer_group_query_attention</a>()</code> <br><code><a href="layer_gru.html">layer_gru</a>()</code> <br><code><a href="layer_hashed_crossing.html">layer_hashed_crossing</a>()</code> <br><code><a href="layer_hashing.html">layer_hashing</a>()</code> <br><code><a href="layer_identity.html">layer_identity</a>()</code> <br><code><a href="layer_integer_lookup.html">layer_integer_lookup</a>()</code> <br><code><a href="layer_jax_model_wrapper.html">layer_jax_model_wrapper</a>()</code> <br><code><a href="layer_lambda.html">layer_lambda</a>()</code> <br><code><a href="layer_layer_normalization.html">layer_layer_normalization</a>()</code> <br><code><a href="layer_lstm.html">layer_lstm</a>()</code> <br><code><a href="layer_masking.html">layer_masking</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_max_pooling_1d.html">layer_max_pooling_1d</a>()</code> <br><code><a href="layer_max_pooling_2d.html">layer_max_pooling_2d</a>()</code> <br><code><a href="layer_max_pooling_3d.html">layer_max_pooling_3d</a>()</code> <br><code><a href="layer_maximum.html">layer_maximum</a>()</code> <br><code><a href="layer_mel_spectrogram.html">layer_mel_spectrogram</a>()</code> <br><code><a href="layer_minimum.html">layer_minimum</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_multi_head_attention.html">layer_multi_head_attention</a>()</code> <br><code><a href="layer_multiply.html">layer_multiply</a>()</code> <br><code><a href="layer_normalization.html">layer_normalization</a>()</code> <br><code><a href="layer_permute.html">layer_permute</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_brightness.html">layer_random_brightness</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_contrast.html">layer_random_contrast</a>()</code> <br><code><a href="layer_random_crop.html">layer_random_crop</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_flip.html">layer_random_flip</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_rotation.html">layer_random_rotation</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_random_translation.html">layer_random_translation</a>()</code> <br><code><a href="layer_random_zoom.html">layer_random_zoom</a>()</code> <br><code><a href="layer_repeat_vector.html">layer_repeat_vector</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_reshape.html">layer_reshape</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_rms_normalization.html">layer_rms_normalization</a>()</code> <br><code><a href="layer_rnn.html">layer_rnn</a>()</code> <br><code><a href="layer_separable_conv_1d.html">layer_separable_conv_1d</a>()</code> <br><code><a href="layer_separable_conv_2d.html">layer_separable_conv_2d</a>()</code> <br><code><a href="layer_simple_rnn.html">layer_simple_rnn</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br><code><a href="layer_spatial_dropout_1d.html">layer_spatial_dropout_1d</a>()</code> <br><code><a href="layer_spatial_dropout_2d.html">layer_spatial_dropout_2d</a>()</code> <br><code><a href="layer_spatial_dropout_3d.html">layer_spatial_dropout_3d</a>()</code> <br><code><a href="layer_spectral_normalization.html">layer_spectral_normalization</a>()</code> <br><code><a href="layer_stft_spectrogram.html">layer_stft_spectrogram</a>()</code> <br><code><a href="layer_string_lookup.html">layer_string_lookup</a>()</code> <br><code><a href="layer_subtract.html">layer_subtract</a>()</code> <br><code><a href="layer_text_vectorization.html">layer_text_vectorization</a>()</code> <br><code><a href="layer_tfsm.html">layer_tfsm</a>()</code> <br><code><a href="layer_time_distributed.html">layer_time_distributed</a>()</code> <br><code><a href="layer_torch_module_wrapper.html">layer_torch_module_wrapper</a>()</code> <br><code><a href="layer_unit_normalization.html">layer_unit_normalization</a>()</code> <br><code><a href="layer_upsampling_1d.html">layer_upsampling_1d</a>()</code> <br><code><a href="layer_upsampling_2d.html">layer_upsampling_2d</a>()</code> <br><code><a href="layer_upsampling_3d.html">layer_upsampling_3d</a>()</code> <br><code><a href="layer_zero_padding_1d.html">layer_zero_padding_1d</a>()</code> <br><code><a href="layer_zero_padding_2d.html">layer_zero_padding_2d</a>()</code> <br><code><a href="layer_zero_padding_3d.html">layer_zero_padding_3d</a>()</code> <br><code><a href="rnn_cell_gru.html">rnn_cell_gru</a>()</code> <br><code><a href="rnn_cell_lstm.html">rnn_cell_lstm</a>()</code> <br><code><a href="rnn_cell_simple.html">rnn_cell_simple</a>()</code> <br><code><a href="rnn_cells_stack.html">rnn_cells_stack</a>()</code> <br></p></div>
+<p>Other layers: <br><code><a href="Layer.html">Layer</a>()</code> <br><code><a href="layer_activation.html">layer_activation</a>()</code> <br><code><a href="layer_activation_elu.html">layer_activation_elu</a>()</code> <br><code><a href="layer_activation_leaky_relu.html">layer_activation_leaky_relu</a>()</code> <br><code><a href="layer_activation_parametric_relu.html">layer_activation_parametric_relu</a>()</code> <br><code><a href="layer_activation_relu.html">layer_activation_relu</a>()</code> <br><code><a href="layer_activation_softmax.html">layer_activation_softmax</a>()</code> <br><code><a href="layer_activity_regularization.html">layer_activity_regularization</a>()</code> <br><code><a href="layer_add.html">layer_add</a>()</code> <br><code><a href="layer_additive_attention.html">layer_additive_attention</a>()</code> <br><code><a href="layer_alpha_dropout.html">layer_alpha_dropout</a>()</code> <br><code><a href="layer_attention.html">layer_attention</a>()</code> <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_average.html">layer_average</a>()</code> <br><code><a href="layer_average_pooling_1d.html">layer_average_pooling_1d</a>()</code> <br><code><a href="layer_average_pooling_2d.html">layer_average_pooling_2d</a>()</code> <br><code><a href="layer_average_pooling_3d.html">layer_average_pooling_3d</a>()</code> <br><code><a href="layer_batch_normalization.html">layer_batch_normalization</a>()</code> <br><code><a href="layer_bidirectional.html">layer_bidirectional</a>()</code> <br><code><a href="layer_category_encoding.html">layer_category_encoding</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_concatenate.html">layer_concatenate</a>()</code> <br><code><a href="layer_conv_1d.html">layer_conv_1d</a>()</code> <br><code><a href="layer_conv_1d_transpose.html">layer_conv_1d_transpose</a>()</code> <br><code><a href="layer_conv_2d.html">layer_conv_2d</a>()</code> <br><code><a href="layer_conv_2d_transpose.html">layer_conv_2d_transpose</a>()</code> <br><code><a href="layer_conv_3d.html">layer_conv_3d</a>()</code> <br><code><a href="layer_conv_3d_transpose.html">layer_conv_3d_transpose</a>()</code> <br><code><a href="layer_conv_lstm_1d.html">layer_conv_lstm_1d</a>()</code> <br><code><a href="layer_conv_lstm_2d.html">layer_conv_lstm_2d</a>()</code> <br><code><a href="layer_conv_lstm_3d.html">layer_conv_lstm_3d</a>()</code> <br><code><a href="layer_cropping_1d.html">layer_cropping_1d</a>()</code> <br><code><a href="layer_cropping_2d.html">layer_cropping_2d</a>()</code> <br><code><a href="layer_cropping_3d.html">layer_cropping_3d</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_dense.html">layer_dense</a>()</code> <br><code><a href="layer_depthwise_conv_1d.html">layer_depthwise_conv_1d</a>()</code> <br><code><a href="layer_depthwise_conv_2d.html">layer_depthwise_conv_2d</a>()</code> <br><code><a href="layer_discretization.html">layer_discretization</a>()</code> <br><code><a href="layer_dot.html">layer_dot</a>()</code> <br><code><a href="layer_dropout.html">layer_dropout</a>()</code> <br><code><a href="layer_einsum_dense.html">layer_einsum_dense</a>()</code> <br><code><a href="layer_embedding.html">layer_embedding</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_feature_space.html">layer_feature_space</a>()</code> <br><code><a href="layer_flatten.html">layer_flatten</a>()</code> <br><code><a href="layer_flax_module_wrapper.html">layer_flax_module_wrapper</a>()</code> <br><code><a href="layer_gaussian_dropout.html">layer_gaussian_dropout</a>()</code> <br><code><a href="layer_gaussian_noise.html">layer_gaussian_noise</a>()</code> <br><code><a href="layer_global_average_pooling_1d.html">layer_global_average_pooling_1d</a>()</code> <br><code><a href="layer_global_average_pooling_2d.html">layer_global_average_pooling_2d</a>()</code> <br><code><a href="layer_global_max_pooling_1d.html">layer_global_max_pooling_1d</a>()</code> <br><code><a href="layer_global_max_pooling_2d.html">layer_global_max_pooling_2d</a>()</code> <br><code><a href="layer_global_max_pooling_3d.html">layer_global_max_pooling_3d</a>()</code> <br><code><a href="layer_group_normalization.html">layer_group_normalization</a>()</code> <br><code><a href="layer_group_query_attention.html">layer_group_query_attention</a>()</code> <br><code><a href="layer_gru.html">layer_gru</a>()</code> <br><code><a href="layer_hashed_crossing.html">layer_hashed_crossing</a>()</code> <br><code><a href="layer_hashing.html">layer_hashing</a>()</code> <br><code><a href="layer_identity.html">layer_identity</a>()</code> <br><code><a href="layer_integer_lookup.html">layer_integer_lookup</a>()</code> <br><code><a href="layer_jax_model_wrapper.html">layer_jax_model_wrapper</a>()</code> <br><code><a href="layer_lambda.html">layer_lambda</a>()</code> <br><code><a href="layer_layer_normalization.html">layer_layer_normalization</a>()</code> <br><code><a href="layer_lstm.html">layer_lstm</a>()</code> <br><code><a href="layer_masking.html">layer_masking</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_max_pooling_1d.html">layer_max_pooling_1d</a>()</code> <br><code><a href="layer_max_pooling_2d.html">layer_max_pooling_2d</a>()</code> <br><code><a href="layer_max_pooling_3d.html">layer_max_pooling_3d</a>()</code> <br><code><a href="layer_maximum.html">layer_maximum</a>()</code> <br><code><a href="layer_mel_spectrogram.html">layer_mel_spectrogram</a>()</code> <br><code><a href="layer_minimum.html">layer_minimum</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_multi_head_attention.html">layer_multi_head_attention</a>()</code> <br><code><a href="layer_multiply.html">layer_multiply</a>()</code> <br><code><a href="layer_normalization.html">layer_normalization</a>()</code> <br><code><a href="layer_permute.html">layer_permute</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_brightness.html">layer_random_brightness</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_contrast.html">layer_random_contrast</a>()</code> <br><code><a href="layer_random_crop.html">layer_random_crop</a>()</code> <br><code><a href="layer_random_elastic_transform.html">layer_random_elastic_transform</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_flip.html">layer_random_flip</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_rotation.html">layer_random_rotation</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_random_translation.html">layer_random_translation</a>()</code> <br><code><a href="layer_random_zoom.html">layer_random_zoom</a>()</code> <br><code><a href="layer_repeat_vector.html">layer_repeat_vector</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_reshape.html">layer_reshape</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_rms_normalization.html">layer_rms_normalization</a>()</code> <br><code><a href="layer_rnn.html">layer_rnn</a>()</code> <br><code><a href="layer_separable_conv_1d.html">layer_separable_conv_1d</a>()</code> <br><code><a href="layer_separable_conv_2d.html">layer_separable_conv_2d</a>()</code> <br><code><a href="layer_simple_rnn.html">layer_simple_rnn</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br><code><a href="layer_spatial_dropout_1d.html">layer_spatial_dropout_1d</a>()</code> <br><code><a href="layer_spatial_dropout_2d.html">layer_spatial_dropout_2d</a>()</code> <br><code><a href="layer_spatial_dropout_3d.html">layer_spatial_dropout_3d</a>()</code> <br><code><a href="layer_spectral_normalization.html">layer_spectral_normalization</a>()</code> <br><code><a href="layer_stft_spectrogram.html">layer_stft_spectrogram</a>()</code> <br><code><a href="layer_string_lookup.html">layer_string_lookup</a>()</code> <br><code><a href="layer_subtract.html">layer_subtract</a>()</code> <br><code><a href="layer_text_vectorization.html">layer_text_vectorization</a>()</code> <br><code><a href="layer_tfsm.html">layer_tfsm</a>()</code> <br><code><a href="layer_time_distributed.html">layer_time_distributed</a>()</code> <br><code><a href="layer_torch_module_wrapper.html">layer_torch_module_wrapper</a>()</code> <br><code><a href="layer_unit_normalization.html">layer_unit_normalization</a>()</code> <br><code><a href="layer_upsampling_1d.html">layer_upsampling_1d</a>()</code> <br><code><a href="layer_upsampling_2d.html">layer_upsampling_2d</a>()</code> <br><code><a href="layer_upsampling_3d.html">layer_upsampling_3d</a>()</code> <br><code><a href="layer_zero_padding_1d.html">layer_zero_padding_1d</a>()</code> <br><code><a href="layer_zero_padding_2d.html">layer_zero_padding_2d</a>()</code> <br><code><a href="layer_zero_padding_3d.html">layer_zero_padding_3d</a>()</code> <br><code><a href="rnn_cell_gru.html">rnn_cell_gru</a>()</code> <br><code><a href="rnn_cell_lstm.html">rnn_cell_lstm</a>()</code> <br><code><a href="rnn_cell_simple.html">rnn_cell_simple</a>()</code> <br><code><a href="rnn_cells_stack.html">rnn_cells_stack</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -154,7 +154,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/layer_global_max_pooling_1d.html b/docs/dev/reference/layer_global_max_pooling_1d.html
index 5ca75ae54..0569a5a62 100644
--- a/docs/dev/reference/layer_global_max_pooling_1d.html
+++ b/docs/dev/reference/layer_global_max_pooling_1d.html
@@ -8,7 +8,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -135,7 +135,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
     <div class="dont-index">
 <ul><li><p><a href="https://keras.io/api/layers/pooling_layers/global_max_pooling1d#globalmaxpooling1d-class" class="external-link">https://keras.io/api/layers/pooling_layers/global_max_pooling1d#globalmaxpooling1d-class</a></p></li>
 </ul><p>Other pooling layers: <br><code><a href="layer_average_pooling_1d.html">layer_average_pooling_1d</a>()</code> <br><code><a href="layer_average_pooling_2d.html">layer_average_pooling_2d</a>()</code> <br><code><a href="layer_average_pooling_3d.html">layer_average_pooling_3d</a>()</code> <br><code><a href="layer_global_average_pooling_1d.html">layer_global_average_pooling_1d</a>()</code> <br><code><a href="layer_global_average_pooling_2d.html">layer_global_average_pooling_2d</a>()</code> <br><code><a href="layer_global_average_pooling_3d.html">layer_global_average_pooling_3d</a>()</code> <br><code><a href="layer_global_max_pooling_2d.html">layer_global_max_pooling_2d</a>()</code> <br><code><a href="layer_global_max_pooling_3d.html">layer_global_max_pooling_3d</a>()</code> <br><code><a href="layer_max_pooling_1d.html">layer_max_pooling_1d</a>()</code> <br><code><a href="layer_max_pooling_2d.html">layer_max_pooling_2d</a>()</code> <br><code><a href="layer_max_pooling_3d.html">layer_max_pooling_3d</a>()</code> <br></p>
-<p>Other layers: <br><code><a href="Layer.html">Layer</a>()</code> <br><code><a href="layer_activation.html">layer_activation</a>()</code> <br><code><a href="layer_activation_elu.html">layer_activation_elu</a>()</code> <br><code><a href="layer_activation_leaky_relu.html">layer_activation_leaky_relu</a>()</code> <br><code><a href="layer_activation_parametric_relu.html">layer_activation_parametric_relu</a>()</code> <br><code><a href="layer_activation_relu.html">layer_activation_relu</a>()</code> <br><code><a href="layer_activation_softmax.html">layer_activation_softmax</a>()</code> <br><code><a href="layer_activity_regularization.html">layer_activity_regularization</a>()</code> <br><code><a href="layer_add.html">layer_add</a>()</code> <br><code><a href="layer_additive_attention.html">layer_additive_attention</a>()</code> <br><code><a href="layer_alpha_dropout.html">layer_alpha_dropout</a>()</code> <br><code><a href="layer_attention.html">layer_attention</a>()</code> <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_average.html">layer_average</a>()</code> <br><code><a href="layer_average_pooling_1d.html">layer_average_pooling_1d</a>()</code> <br><code><a href="layer_average_pooling_2d.html">layer_average_pooling_2d</a>()</code> <br><code><a href="layer_average_pooling_3d.html">layer_average_pooling_3d</a>()</code> <br><code><a href="layer_batch_normalization.html">layer_batch_normalization</a>()</code> <br><code><a href="layer_bidirectional.html">layer_bidirectional</a>()</code> <br><code><a href="layer_category_encoding.html">layer_category_encoding</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_concatenate.html">layer_concatenate</a>()</code> <br><code><a href="layer_conv_1d.html">layer_conv_1d</a>()</code> <br><code><a href="layer_conv_1d_transpose.html">layer_conv_1d_transpose</a>()</code> <br><code><a href="layer_conv_2d.html">layer_conv_2d</a>()</code> <br><code><a href="layer_conv_2d_transpose.html">layer_conv_2d_transpose</a>()</code> <br><code><a href="layer_conv_3d.html">layer_conv_3d</a>()</code> <br><code><a href="layer_conv_3d_transpose.html">layer_conv_3d_transpose</a>()</code> <br><code><a href="layer_conv_lstm_1d.html">layer_conv_lstm_1d</a>()</code> <br><code><a href="layer_conv_lstm_2d.html">layer_conv_lstm_2d</a>()</code> <br><code><a href="layer_conv_lstm_3d.html">layer_conv_lstm_3d</a>()</code> <br><code><a href="layer_cropping_1d.html">layer_cropping_1d</a>()</code> <br><code><a href="layer_cropping_2d.html">layer_cropping_2d</a>()</code> <br><code><a href="layer_cropping_3d.html">layer_cropping_3d</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_dense.html">layer_dense</a>()</code> <br><code><a href="layer_depthwise_conv_1d.html">layer_depthwise_conv_1d</a>()</code> <br><code><a href="layer_depthwise_conv_2d.html">layer_depthwise_conv_2d</a>()</code> <br><code><a href="layer_discretization.html">layer_discretization</a>()</code> <br><code><a href="layer_dot.html">layer_dot</a>()</code> <br><code><a href="layer_dropout.html">layer_dropout</a>()</code> <br><code><a href="layer_einsum_dense.html">layer_einsum_dense</a>()</code> <br><code><a href="layer_embedding.html">layer_embedding</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_feature_space.html">layer_feature_space</a>()</code> <br><code><a href="layer_flatten.html">layer_flatten</a>()</code> <br><code><a href="layer_flax_module_wrapper.html">layer_flax_module_wrapper</a>()</code> <br><code><a href="layer_gaussian_dropout.html">layer_gaussian_dropout</a>()</code> <br><code><a href="layer_gaussian_noise.html">layer_gaussian_noise</a>()</code> <br><code><a href="layer_global_average_pooling_1d.html">layer_global_average_pooling_1d</a>()</code> <br><code><a href="layer_global_average_pooling_2d.html">layer_global_average_pooling_2d</a>()</code> <br><code><a href="layer_global_average_pooling_3d.html">layer_global_average_pooling_3d</a>()</code> <br><code><a href="layer_global_max_pooling_2d.html">layer_global_max_pooling_2d</a>()</code> <br><code><a href="layer_global_max_pooling_3d.html">layer_global_max_pooling_3d</a>()</code> <br><code><a href="layer_group_normalization.html">layer_group_normalization</a>()</code> <br><code><a href="layer_group_query_attention.html">layer_group_query_attention</a>()</code> <br><code><a href="layer_gru.html">layer_gru</a>()</code> <br><code><a href="layer_hashed_crossing.html">layer_hashed_crossing</a>()</code> <br><code><a href="layer_hashing.html">layer_hashing</a>()</code> <br><code><a href="layer_identity.html">layer_identity</a>()</code> <br><code><a href="layer_integer_lookup.html">layer_integer_lookup</a>()</code> <br><code><a href="layer_jax_model_wrapper.html">layer_jax_model_wrapper</a>()</code> <br><code><a href="layer_lambda.html">layer_lambda</a>()</code> <br><code><a href="layer_layer_normalization.html">layer_layer_normalization</a>()</code> <br><code><a href="layer_lstm.html">layer_lstm</a>()</code> <br><code><a href="layer_masking.html">layer_masking</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_max_pooling_1d.html">layer_max_pooling_1d</a>()</code> <br><code><a href="layer_max_pooling_2d.html">layer_max_pooling_2d</a>()</code> <br><code><a href="layer_max_pooling_3d.html">layer_max_pooling_3d</a>()</code> <br><code><a href="layer_maximum.html">layer_maximum</a>()</code> <br><code><a href="layer_mel_spectrogram.html">layer_mel_spectrogram</a>()</code> <br><code><a href="layer_minimum.html">layer_minimum</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_multi_head_attention.html">layer_multi_head_attention</a>()</code> <br><code><a href="layer_multiply.html">layer_multiply</a>()</code> <br><code><a href="layer_normalization.html">layer_normalization</a>()</code> <br><code><a href="layer_permute.html">layer_permute</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_brightness.html">layer_random_brightness</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_contrast.html">layer_random_contrast</a>()</code> <br><code><a href="layer_random_crop.html">layer_random_crop</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_flip.html">layer_random_flip</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_rotation.html">layer_random_rotation</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_random_translation.html">layer_random_translation</a>()</code> <br><code><a href="layer_random_zoom.html">layer_random_zoom</a>()</code> <br><code><a href="layer_repeat_vector.html">layer_repeat_vector</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_reshape.html">layer_reshape</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_rms_normalization.html">layer_rms_normalization</a>()</code> <br><code><a href="layer_rnn.html">layer_rnn</a>()</code> <br><code><a href="layer_separable_conv_1d.html">layer_separable_conv_1d</a>()</code> <br><code><a href="layer_separable_conv_2d.html">layer_separable_conv_2d</a>()</code> <br><code><a href="layer_simple_rnn.html">layer_simple_rnn</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br><code><a href="layer_spatial_dropout_1d.html">layer_spatial_dropout_1d</a>()</code> <br><code><a href="layer_spatial_dropout_2d.html">layer_spatial_dropout_2d</a>()</code> <br><code><a href="layer_spatial_dropout_3d.html">layer_spatial_dropout_3d</a>()</code> <br><code><a href="layer_spectral_normalization.html">layer_spectral_normalization</a>()</code> <br><code><a href="layer_stft_spectrogram.html">layer_stft_spectrogram</a>()</code> <br><code><a href="layer_string_lookup.html">layer_string_lookup</a>()</code> <br><code><a href="layer_subtract.html">layer_subtract</a>()</code> <br><code><a href="layer_text_vectorization.html">layer_text_vectorization</a>()</code> <br><code><a href="layer_tfsm.html">layer_tfsm</a>()</code> <br><code><a href="layer_time_distributed.html">layer_time_distributed</a>()</code> <br><code><a href="layer_torch_module_wrapper.html">layer_torch_module_wrapper</a>()</code> <br><code><a href="layer_unit_normalization.html">layer_unit_normalization</a>()</code> <br><code><a href="layer_upsampling_1d.html">layer_upsampling_1d</a>()</code> <br><code><a href="layer_upsampling_2d.html">layer_upsampling_2d</a>()</code> <br><code><a href="layer_upsampling_3d.html">layer_upsampling_3d</a>()</code> <br><code><a href="layer_zero_padding_1d.html">layer_zero_padding_1d</a>()</code> <br><code><a href="layer_zero_padding_2d.html">layer_zero_padding_2d</a>()</code> <br><code><a href="layer_zero_padding_3d.html">layer_zero_padding_3d</a>()</code> <br><code><a href="rnn_cell_gru.html">rnn_cell_gru</a>()</code> <br><code><a href="rnn_cell_lstm.html">rnn_cell_lstm</a>()</code> <br><code><a href="rnn_cell_simple.html">rnn_cell_simple</a>()</code> <br><code><a href="rnn_cells_stack.html">rnn_cells_stack</a>()</code> <br></p></div>
+<p>Other layers: <br><code><a href="Layer.html">Layer</a>()</code> <br><code><a href="layer_activation.html">layer_activation</a>()</code> <br><code><a href="layer_activation_elu.html">layer_activation_elu</a>()</code> <br><code><a href="layer_activation_leaky_relu.html">layer_activation_leaky_relu</a>()</code> <br><code><a href="layer_activation_parametric_relu.html">layer_activation_parametric_relu</a>()</code> <br><code><a href="layer_activation_relu.html">layer_activation_relu</a>()</code> <br><code><a href="layer_activation_softmax.html">layer_activation_softmax</a>()</code> <br><code><a href="layer_activity_regularization.html">layer_activity_regularization</a>()</code> <br><code><a href="layer_add.html">layer_add</a>()</code> <br><code><a href="layer_additive_attention.html">layer_additive_attention</a>()</code> <br><code><a href="layer_alpha_dropout.html">layer_alpha_dropout</a>()</code> <br><code><a href="layer_attention.html">layer_attention</a>()</code> <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_average.html">layer_average</a>()</code> <br><code><a href="layer_average_pooling_1d.html">layer_average_pooling_1d</a>()</code> <br><code><a href="layer_average_pooling_2d.html">layer_average_pooling_2d</a>()</code> <br><code><a href="layer_average_pooling_3d.html">layer_average_pooling_3d</a>()</code> <br><code><a href="layer_batch_normalization.html">layer_batch_normalization</a>()</code> <br><code><a href="layer_bidirectional.html">layer_bidirectional</a>()</code> <br><code><a href="layer_category_encoding.html">layer_category_encoding</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_concatenate.html">layer_concatenate</a>()</code> <br><code><a href="layer_conv_1d.html">layer_conv_1d</a>()</code> <br><code><a href="layer_conv_1d_transpose.html">layer_conv_1d_transpose</a>()</code> <br><code><a href="layer_conv_2d.html">layer_conv_2d</a>()</code> <br><code><a href="layer_conv_2d_transpose.html">layer_conv_2d_transpose</a>()</code> <br><code><a href="layer_conv_3d.html">layer_conv_3d</a>()</code> <br><code><a href="layer_conv_3d_transpose.html">layer_conv_3d_transpose</a>()</code> <br><code><a href="layer_conv_lstm_1d.html">layer_conv_lstm_1d</a>()</code> <br><code><a href="layer_conv_lstm_2d.html">layer_conv_lstm_2d</a>()</code> <br><code><a href="layer_conv_lstm_3d.html">layer_conv_lstm_3d</a>()</code> <br><code><a href="layer_cropping_1d.html">layer_cropping_1d</a>()</code> <br><code><a href="layer_cropping_2d.html">layer_cropping_2d</a>()</code> <br><code><a href="layer_cropping_3d.html">layer_cropping_3d</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_dense.html">layer_dense</a>()</code> <br><code><a href="layer_depthwise_conv_1d.html">layer_depthwise_conv_1d</a>()</code> <br><code><a href="layer_depthwise_conv_2d.html">layer_depthwise_conv_2d</a>()</code> <br><code><a href="layer_discretization.html">layer_discretization</a>()</code> <br><code><a href="layer_dot.html">layer_dot</a>()</code> <br><code><a href="layer_dropout.html">layer_dropout</a>()</code> <br><code><a href="layer_einsum_dense.html">layer_einsum_dense</a>()</code> <br><code><a href="layer_embedding.html">layer_embedding</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_feature_space.html">layer_feature_space</a>()</code> <br><code><a href="layer_flatten.html">layer_flatten</a>()</code> <br><code><a href="layer_flax_module_wrapper.html">layer_flax_module_wrapper</a>()</code> <br><code><a href="layer_gaussian_dropout.html">layer_gaussian_dropout</a>()</code> <br><code><a href="layer_gaussian_noise.html">layer_gaussian_noise</a>()</code> <br><code><a href="layer_global_average_pooling_1d.html">layer_global_average_pooling_1d</a>()</code> <br><code><a href="layer_global_average_pooling_2d.html">layer_global_average_pooling_2d</a>()</code> <br><code><a href="layer_global_average_pooling_3d.html">layer_global_average_pooling_3d</a>()</code> <br><code><a href="layer_global_max_pooling_2d.html">layer_global_max_pooling_2d</a>()</code> <br><code><a href="layer_global_max_pooling_3d.html">layer_global_max_pooling_3d</a>()</code> <br><code><a href="layer_group_normalization.html">layer_group_normalization</a>()</code> <br><code><a href="layer_group_query_attention.html">layer_group_query_attention</a>()</code> <br><code><a href="layer_gru.html">layer_gru</a>()</code> <br><code><a href="layer_hashed_crossing.html">layer_hashed_crossing</a>()</code> <br><code><a href="layer_hashing.html">layer_hashing</a>()</code> <br><code><a href="layer_identity.html">layer_identity</a>()</code> <br><code><a href="layer_integer_lookup.html">layer_integer_lookup</a>()</code> <br><code><a href="layer_jax_model_wrapper.html">layer_jax_model_wrapper</a>()</code> <br><code><a href="layer_lambda.html">layer_lambda</a>()</code> <br><code><a href="layer_layer_normalization.html">layer_layer_normalization</a>()</code> <br><code><a href="layer_lstm.html">layer_lstm</a>()</code> <br><code><a href="layer_masking.html">layer_masking</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_max_pooling_1d.html">layer_max_pooling_1d</a>()</code> <br><code><a href="layer_max_pooling_2d.html">layer_max_pooling_2d</a>()</code> <br><code><a href="layer_max_pooling_3d.html">layer_max_pooling_3d</a>()</code> <br><code><a href="layer_maximum.html">layer_maximum</a>()</code> <br><code><a href="layer_mel_spectrogram.html">layer_mel_spectrogram</a>()</code> <br><code><a href="layer_minimum.html">layer_minimum</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_multi_head_attention.html">layer_multi_head_attention</a>()</code> <br><code><a href="layer_multiply.html">layer_multiply</a>()</code> <br><code><a href="layer_normalization.html">layer_normalization</a>()</code> <br><code><a href="layer_permute.html">layer_permute</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_brightness.html">layer_random_brightness</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_contrast.html">layer_random_contrast</a>()</code> <br><code><a href="layer_random_crop.html">layer_random_crop</a>()</code> <br><code><a href="layer_random_elastic_transform.html">layer_random_elastic_transform</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_flip.html">layer_random_flip</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_rotation.html">layer_random_rotation</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_random_translation.html">layer_random_translation</a>()</code> <br><code><a href="layer_random_zoom.html">layer_random_zoom</a>()</code> <br><code><a href="layer_repeat_vector.html">layer_repeat_vector</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_reshape.html">layer_reshape</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_rms_normalization.html">layer_rms_normalization</a>()</code> <br><code><a href="layer_rnn.html">layer_rnn</a>()</code> <br><code><a href="layer_separable_conv_1d.html">layer_separable_conv_1d</a>()</code> <br><code><a href="layer_separable_conv_2d.html">layer_separable_conv_2d</a>()</code> <br><code><a href="layer_simple_rnn.html">layer_simple_rnn</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br><code><a href="layer_spatial_dropout_1d.html">layer_spatial_dropout_1d</a>()</code> <br><code><a href="layer_spatial_dropout_2d.html">layer_spatial_dropout_2d</a>()</code> <br><code><a href="layer_spatial_dropout_3d.html">layer_spatial_dropout_3d</a>()</code> <br><code><a href="layer_spectral_normalization.html">layer_spectral_normalization</a>()</code> <br><code><a href="layer_stft_spectrogram.html">layer_stft_spectrogram</a>()</code> <br><code><a href="layer_string_lookup.html">layer_string_lookup</a>()</code> <br><code><a href="layer_subtract.html">layer_subtract</a>()</code> <br><code><a href="layer_text_vectorization.html">layer_text_vectorization</a>()</code> <br><code><a href="layer_tfsm.html">layer_tfsm</a>()</code> <br><code><a href="layer_time_distributed.html">layer_time_distributed</a>()</code> <br><code><a href="layer_torch_module_wrapper.html">layer_torch_module_wrapper</a>()</code> <br><code><a href="layer_unit_normalization.html">layer_unit_normalization</a>()</code> <br><code><a href="layer_upsampling_1d.html">layer_upsampling_1d</a>()</code> <br><code><a href="layer_upsampling_2d.html">layer_upsampling_2d</a>()</code> <br><code><a href="layer_upsampling_3d.html">layer_upsampling_3d</a>()</code> <br><code><a href="layer_zero_padding_1d.html">layer_zero_padding_1d</a>()</code> <br><code><a href="layer_zero_padding_2d.html">layer_zero_padding_2d</a>()</code> <br><code><a href="layer_zero_padding_3d.html">layer_zero_padding_3d</a>()</code> <br><code><a href="rnn_cell_gru.html">rnn_cell_gru</a>()</code> <br><code><a href="rnn_cell_lstm.html">rnn_cell_lstm</a>()</code> <br><code><a href="rnn_cell_simple.html">rnn_cell_simple</a>()</code> <br><code><a href="rnn_cells_stack.html">rnn_cells_stack</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -147,7 +147,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/layer_global_max_pooling_2d.html b/docs/dev/reference/layer_global_max_pooling_2d.html
index 8e129bc61..64a3d86bf 100644
--- a/docs/dev/reference/layer_global_max_pooling_2d.html
+++ b/docs/dev/reference/layer_global_max_pooling_2d.html
@@ -8,7 +8,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -136,7 +136,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
     <div class="dont-index">
 <ul><li><p><a href="https://keras.io/api/layers/pooling_layers/global_max_pooling2d#globalmaxpooling2d-class" class="external-link">https://keras.io/api/layers/pooling_layers/global_max_pooling2d#globalmaxpooling2d-class</a></p></li>
 </ul><p>Other pooling layers: <br><code><a href="layer_average_pooling_1d.html">layer_average_pooling_1d</a>()</code> <br><code><a href="layer_average_pooling_2d.html">layer_average_pooling_2d</a>()</code> <br><code><a href="layer_average_pooling_3d.html">layer_average_pooling_3d</a>()</code> <br><code><a href="layer_global_average_pooling_1d.html">layer_global_average_pooling_1d</a>()</code> <br><code><a href="layer_global_average_pooling_2d.html">layer_global_average_pooling_2d</a>()</code> <br><code><a href="layer_global_average_pooling_3d.html">layer_global_average_pooling_3d</a>()</code> <br><code><a href="layer_global_max_pooling_1d.html">layer_global_max_pooling_1d</a>()</code> <br><code><a href="layer_global_max_pooling_3d.html">layer_global_max_pooling_3d</a>()</code> <br><code><a href="layer_max_pooling_1d.html">layer_max_pooling_1d</a>()</code> <br><code><a href="layer_max_pooling_2d.html">layer_max_pooling_2d</a>()</code> <br><code><a href="layer_max_pooling_3d.html">layer_max_pooling_3d</a>()</code> <br></p>
-<p>Other layers: <br><code><a href="Layer.html">Layer</a>()</code> <br><code><a href="layer_activation.html">layer_activation</a>()</code> <br><code><a href="layer_activation_elu.html">layer_activation_elu</a>()</code> <br><code><a href="layer_activation_leaky_relu.html">layer_activation_leaky_relu</a>()</code> <br><code><a href="layer_activation_parametric_relu.html">layer_activation_parametric_relu</a>()</code> <br><code><a href="layer_activation_relu.html">layer_activation_relu</a>()</code> <br><code><a href="layer_activation_softmax.html">layer_activation_softmax</a>()</code> <br><code><a href="layer_activity_regularization.html">layer_activity_regularization</a>()</code> <br><code><a href="layer_add.html">layer_add</a>()</code> <br><code><a href="layer_additive_attention.html">layer_additive_attention</a>()</code> <br><code><a href="layer_alpha_dropout.html">layer_alpha_dropout</a>()</code> <br><code><a href="layer_attention.html">layer_attention</a>()</code> <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_average.html">layer_average</a>()</code> <br><code><a href="layer_average_pooling_1d.html">layer_average_pooling_1d</a>()</code> <br><code><a href="layer_average_pooling_2d.html">layer_average_pooling_2d</a>()</code> <br><code><a href="layer_average_pooling_3d.html">layer_average_pooling_3d</a>()</code> <br><code><a href="layer_batch_normalization.html">layer_batch_normalization</a>()</code> <br><code><a href="layer_bidirectional.html">layer_bidirectional</a>()</code> <br><code><a href="layer_category_encoding.html">layer_category_encoding</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_concatenate.html">layer_concatenate</a>()</code> <br><code><a href="layer_conv_1d.html">layer_conv_1d</a>()</code> <br><code><a href="layer_conv_1d_transpose.html">layer_conv_1d_transpose</a>()</code> <br><code><a href="layer_conv_2d.html">layer_conv_2d</a>()</code> <br><code><a href="layer_conv_2d_transpose.html">layer_conv_2d_transpose</a>()</code> <br><code><a href="layer_conv_3d.html">layer_conv_3d</a>()</code> <br><code><a href="layer_conv_3d_transpose.html">layer_conv_3d_transpose</a>()</code> <br><code><a href="layer_conv_lstm_1d.html">layer_conv_lstm_1d</a>()</code> <br><code><a href="layer_conv_lstm_2d.html">layer_conv_lstm_2d</a>()</code> <br><code><a href="layer_conv_lstm_3d.html">layer_conv_lstm_3d</a>()</code> <br><code><a href="layer_cropping_1d.html">layer_cropping_1d</a>()</code> <br><code><a href="layer_cropping_2d.html">layer_cropping_2d</a>()</code> <br><code><a href="layer_cropping_3d.html">layer_cropping_3d</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_dense.html">layer_dense</a>()</code> <br><code><a href="layer_depthwise_conv_1d.html">layer_depthwise_conv_1d</a>()</code> <br><code><a href="layer_depthwise_conv_2d.html">layer_depthwise_conv_2d</a>()</code> <br><code><a href="layer_discretization.html">layer_discretization</a>()</code> <br><code><a href="layer_dot.html">layer_dot</a>()</code> <br><code><a href="layer_dropout.html">layer_dropout</a>()</code> <br><code><a href="layer_einsum_dense.html">layer_einsum_dense</a>()</code> <br><code><a href="layer_embedding.html">layer_embedding</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_feature_space.html">layer_feature_space</a>()</code> <br><code><a href="layer_flatten.html">layer_flatten</a>()</code> <br><code><a href="layer_flax_module_wrapper.html">layer_flax_module_wrapper</a>()</code> <br><code><a href="layer_gaussian_dropout.html">layer_gaussian_dropout</a>()</code> <br><code><a href="layer_gaussian_noise.html">layer_gaussian_noise</a>()</code> <br><code><a href="layer_global_average_pooling_1d.html">layer_global_average_pooling_1d</a>()</code> <br><code><a href="layer_global_average_pooling_2d.html">layer_global_average_pooling_2d</a>()</code> <br><code><a href="layer_global_average_pooling_3d.html">layer_global_average_pooling_3d</a>()</code> <br><code><a href="layer_global_max_pooling_1d.html">layer_global_max_pooling_1d</a>()</code> <br><code><a href="layer_global_max_pooling_3d.html">layer_global_max_pooling_3d</a>()</code> <br><code><a href="layer_group_normalization.html">layer_group_normalization</a>()</code> <br><code><a href="layer_group_query_attention.html">layer_group_query_attention</a>()</code> <br><code><a href="layer_gru.html">layer_gru</a>()</code> <br><code><a href="layer_hashed_crossing.html">layer_hashed_crossing</a>()</code> <br><code><a href="layer_hashing.html">layer_hashing</a>()</code> <br><code><a href="layer_identity.html">layer_identity</a>()</code> <br><code><a href="layer_integer_lookup.html">layer_integer_lookup</a>()</code> <br><code><a href="layer_jax_model_wrapper.html">layer_jax_model_wrapper</a>()</code> <br><code><a href="layer_lambda.html">layer_lambda</a>()</code> <br><code><a href="layer_layer_normalization.html">layer_layer_normalization</a>()</code> <br><code><a href="layer_lstm.html">layer_lstm</a>()</code> <br><code><a href="layer_masking.html">layer_masking</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_max_pooling_1d.html">layer_max_pooling_1d</a>()</code> <br><code><a href="layer_max_pooling_2d.html">layer_max_pooling_2d</a>()</code> <br><code><a href="layer_max_pooling_3d.html">layer_max_pooling_3d</a>()</code> <br><code><a href="layer_maximum.html">layer_maximum</a>()</code> <br><code><a href="layer_mel_spectrogram.html">layer_mel_spectrogram</a>()</code> <br><code><a href="layer_minimum.html">layer_minimum</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_multi_head_attention.html">layer_multi_head_attention</a>()</code> <br><code><a href="layer_multiply.html">layer_multiply</a>()</code> <br><code><a href="layer_normalization.html">layer_normalization</a>()</code> <br><code><a href="layer_permute.html">layer_permute</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_brightness.html">layer_random_brightness</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_contrast.html">layer_random_contrast</a>()</code> <br><code><a href="layer_random_crop.html">layer_random_crop</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_flip.html">layer_random_flip</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_rotation.html">layer_random_rotation</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_random_translation.html">layer_random_translation</a>()</code> <br><code><a href="layer_random_zoom.html">layer_random_zoom</a>()</code> <br><code><a href="layer_repeat_vector.html">layer_repeat_vector</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_reshape.html">layer_reshape</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_rms_normalization.html">layer_rms_normalization</a>()</code> <br><code><a href="layer_rnn.html">layer_rnn</a>()</code> <br><code><a href="layer_separable_conv_1d.html">layer_separable_conv_1d</a>()</code> <br><code><a href="layer_separable_conv_2d.html">layer_separable_conv_2d</a>()</code> <br><code><a href="layer_simple_rnn.html">layer_simple_rnn</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br><code><a href="layer_spatial_dropout_1d.html">layer_spatial_dropout_1d</a>()</code> <br><code><a href="layer_spatial_dropout_2d.html">layer_spatial_dropout_2d</a>()</code> <br><code><a href="layer_spatial_dropout_3d.html">layer_spatial_dropout_3d</a>()</code> <br><code><a href="layer_spectral_normalization.html">layer_spectral_normalization</a>()</code> <br><code><a href="layer_stft_spectrogram.html">layer_stft_spectrogram</a>()</code> <br><code><a href="layer_string_lookup.html">layer_string_lookup</a>()</code> <br><code><a href="layer_subtract.html">layer_subtract</a>()</code> <br><code><a href="layer_text_vectorization.html">layer_text_vectorization</a>()</code> <br><code><a href="layer_tfsm.html">layer_tfsm</a>()</code> <br><code><a href="layer_time_distributed.html">layer_time_distributed</a>()</code> <br><code><a href="layer_torch_module_wrapper.html">layer_torch_module_wrapper</a>()</code> <br><code><a href="layer_unit_normalization.html">layer_unit_normalization</a>()</code> <br><code><a href="layer_upsampling_1d.html">layer_upsampling_1d</a>()</code> <br><code><a href="layer_upsampling_2d.html">layer_upsampling_2d</a>()</code> <br><code><a href="layer_upsampling_3d.html">layer_upsampling_3d</a>()</code> <br><code><a href="layer_zero_padding_1d.html">layer_zero_padding_1d</a>()</code> <br><code><a href="layer_zero_padding_2d.html">layer_zero_padding_2d</a>()</code> <br><code><a href="layer_zero_padding_3d.html">layer_zero_padding_3d</a>()</code> <br><code><a href="rnn_cell_gru.html">rnn_cell_gru</a>()</code> <br><code><a href="rnn_cell_lstm.html">rnn_cell_lstm</a>()</code> <br><code><a href="rnn_cell_simple.html">rnn_cell_simple</a>()</code> <br><code><a href="rnn_cells_stack.html">rnn_cells_stack</a>()</code> <br></p></div>
+<p>Other layers: <br><code><a href="Layer.html">Layer</a>()</code> <br><code><a href="layer_activation.html">layer_activation</a>()</code> <br><code><a href="layer_activation_elu.html">layer_activation_elu</a>()</code> <br><code><a href="layer_activation_leaky_relu.html">layer_activation_leaky_relu</a>()</code> <br><code><a href="layer_activation_parametric_relu.html">layer_activation_parametric_relu</a>()</code> <br><code><a href="layer_activation_relu.html">layer_activation_relu</a>()</code> <br><code><a href="layer_activation_softmax.html">layer_activation_softmax</a>()</code> <br><code><a href="layer_activity_regularization.html">layer_activity_regularization</a>()</code> <br><code><a href="layer_add.html">layer_add</a>()</code> <br><code><a href="layer_additive_attention.html">layer_additive_attention</a>()</code> <br><code><a href="layer_alpha_dropout.html">layer_alpha_dropout</a>()</code> <br><code><a href="layer_attention.html">layer_attention</a>()</code> <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_average.html">layer_average</a>()</code> <br><code><a href="layer_average_pooling_1d.html">layer_average_pooling_1d</a>()</code> <br><code><a href="layer_average_pooling_2d.html">layer_average_pooling_2d</a>()</code> <br><code><a href="layer_average_pooling_3d.html">layer_average_pooling_3d</a>()</code> <br><code><a href="layer_batch_normalization.html">layer_batch_normalization</a>()</code> <br><code><a href="layer_bidirectional.html">layer_bidirectional</a>()</code> <br><code><a href="layer_category_encoding.html">layer_category_encoding</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_concatenate.html">layer_concatenate</a>()</code> <br><code><a href="layer_conv_1d.html">layer_conv_1d</a>()</code> <br><code><a href="layer_conv_1d_transpose.html">layer_conv_1d_transpose</a>()</code> <br><code><a href="layer_conv_2d.html">layer_conv_2d</a>()</code> <br><code><a href="layer_conv_2d_transpose.html">layer_conv_2d_transpose</a>()</code> <br><code><a href="layer_conv_3d.html">layer_conv_3d</a>()</code> <br><code><a href="layer_conv_3d_transpose.html">layer_conv_3d_transpose</a>()</code> <br><code><a href="layer_conv_lstm_1d.html">layer_conv_lstm_1d</a>()</code> <br><code><a href="layer_conv_lstm_2d.html">layer_conv_lstm_2d</a>()</code> <br><code><a href="layer_conv_lstm_3d.html">layer_conv_lstm_3d</a>()</code> <br><code><a href="layer_cropping_1d.html">layer_cropping_1d</a>()</code> <br><code><a href="layer_cropping_2d.html">layer_cropping_2d</a>()</code> <br><code><a href="layer_cropping_3d.html">layer_cropping_3d</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_dense.html">layer_dense</a>()</code> <br><code><a href="layer_depthwise_conv_1d.html">layer_depthwise_conv_1d</a>()</code> <br><code><a href="layer_depthwise_conv_2d.html">layer_depthwise_conv_2d</a>()</code> <br><code><a href="layer_discretization.html">layer_discretization</a>()</code> <br><code><a href="layer_dot.html">layer_dot</a>()</code> <br><code><a href="layer_dropout.html">layer_dropout</a>()</code> <br><code><a href="layer_einsum_dense.html">layer_einsum_dense</a>()</code> <br><code><a href="layer_embedding.html">layer_embedding</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_feature_space.html">layer_feature_space</a>()</code> <br><code><a href="layer_flatten.html">layer_flatten</a>()</code> <br><code><a href="layer_flax_module_wrapper.html">layer_flax_module_wrapper</a>()</code> <br><code><a href="layer_gaussian_dropout.html">layer_gaussian_dropout</a>()</code> <br><code><a href="layer_gaussian_noise.html">layer_gaussian_noise</a>()</code> <br><code><a href="layer_global_average_pooling_1d.html">layer_global_average_pooling_1d</a>()</code> <br><code><a href="layer_global_average_pooling_2d.html">layer_global_average_pooling_2d</a>()</code> <br><code><a href="layer_global_average_pooling_3d.html">layer_global_average_pooling_3d</a>()</code> <br><code><a href="layer_global_max_pooling_1d.html">layer_global_max_pooling_1d</a>()</code> <br><code><a href="layer_global_max_pooling_3d.html">layer_global_max_pooling_3d</a>()</code> <br><code><a href="layer_group_normalization.html">layer_group_normalization</a>()</code> <br><code><a href="layer_group_query_attention.html">layer_group_query_attention</a>()</code> <br><code><a href="layer_gru.html">layer_gru</a>()</code> <br><code><a href="layer_hashed_crossing.html">layer_hashed_crossing</a>()</code> <br><code><a href="layer_hashing.html">layer_hashing</a>()</code> <br><code><a href="layer_identity.html">layer_identity</a>()</code> <br><code><a href="layer_integer_lookup.html">layer_integer_lookup</a>()</code> <br><code><a href="layer_jax_model_wrapper.html">layer_jax_model_wrapper</a>()</code> <br><code><a href="layer_lambda.html">layer_lambda</a>()</code> <br><code><a href="layer_layer_normalization.html">layer_layer_normalization</a>()</code> <br><code><a href="layer_lstm.html">layer_lstm</a>()</code> <br><code><a href="layer_masking.html">layer_masking</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_max_pooling_1d.html">layer_max_pooling_1d</a>()</code> <br><code><a href="layer_max_pooling_2d.html">layer_max_pooling_2d</a>()</code> <br><code><a href="layer_max_pooling_3d.html">layer_max_pooling_3d</a>()</code> <br><code><a href="layer_maximum.html">layer_maximum</a>()</code> <br><code><a href="layer_mel_spectrogram.html">layer_mel_spectrogram</a>()</code> <br><code><a href="layer_minimum.html">layer_minimum</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_multi_head_attention.html">layer_multi_head_attention</a>()</code> <br><code><a href="layer_multiply.html">layer_multiply</a>()</code> <br><code><a href="layer_normalization.html">layer_normalization</a>()</code> <br><code><a href="layer_permute.html">layer_permute</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_brightness.html">layer_random_brightness</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_contrast.html">layer_random_contrast</a>()</code> <br><code><a href="layer_random_crop.html">layer_random_crop</a>()</code> <br><code><a href="layer_random_elastic_transform.html">layer_random_elastic_transform</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_flip.html">layer_random_flip</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_rotation.html">layer_random_rotation</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_random_translation.html">layer_random_translation</a>()</code> <br><code><a href="layer_random_zoom.html">layer_random_zoom</a>()</code> <br><code><a href="layer_repeat_vector.html">layer_repeat_vector</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_reshape.html">layer_reshape</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_rms_normalization.html">layer_rms_normalization</a>()</code> <br><code><a href="layer_rnn.html">layer_rnn</a>()</code> <br><code><a href="layer_separable_conv_1d.html">layer_separable_conv_1d</a>()</code> <br><code><a href="layer_separable_conv_2d.html">layer_separable_conv_2d</a>()</code> <br><code><a href="layer_simple_rnn.html">layer_simple_rnn</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br><code><a href="layer_spatial_dropout_1d.html">layer_spatial_dropout_1d</a>()</code> <br><code><a href="layer_spatial_dropout_2d.html">layer_spatial_dropout_2d</a>()</code> <br><code><a href="layer_spatial_dropout_3d.html">layer_spatial_dropout_3d</a>()</code> <br><code><a href="layer_spectral_normalization.html">layer_spectral_normalization</a>()</code> <br><code><a href="layer_stft_spectrogram.html">layer_stft_spectrogram</a>()</code> <br><code><a href="layer_string_lookup.html">layer_string_lookup</a>()</code> <br><code><a href="layer_subtract.html">layer_subtract</a>()</code> <br><code><a href="layer_text_vectorization.html">layer_text_vectorization</a>()</code> <br><code><a href="layer_tfsm.html">layer_tfsm</a>()</code> <br><code><a href="layer_time_distributed.html">layer_time_distributed</a>()</code> <br><code><a href="layer_torch_module_wrapper.html">layer_torch_module_wrapper</a>()</code> <br><code><a href="layer_unit_normalization.html">layer_unit_normalization</a>()</code> <br><code><a href="layer_upsampling_1d.html">layer_upsampling_1d</a>()</code> <br><code><a href="layer_upsampling_2d.html">layer_upsampling_2d</a>()</code> <br><code><a href="layer_upsampling_3d.html">layer_upsampling_3d</a>()</code> <br><code><a href="layer_zero_padding_1d.html">layer_zero_padding_1d</a>()</code> <br><code><a href="layer_zero_padding_2d.html">layer_zero_padding_2d</a>()</code> <br><code><a href="layer_zero_padding_3d.html">layer_zero_padding_3d</a>()</code> <br><code><a href="rnn_cell_gru.html">rnn_cell_gru</a>()</code> <br><code><a href="rnn_cell_lstm.html">rnn_cell_lstm</a>()</code> <br><code><a href="rnn_cell_simple.html">rnn_cell_simple</a>()</code> <br><code><a href="rnn_cells_stack.html">rnn_cells_stack</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -148,7 +148,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/layer_global_max_pooling_3d.html b/docs/dev/reference/layer_global_max_pooling_3d.html
index 19d04c067..fb41cb3fb 100644
--- a/docs/dev/reference/layer_global_max_pooling_3d.html
+++ b/docs/dev/reference/layer_global_max_pooling_3d.html
@@ -8,7 +8,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -137,7 +137,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
     <div class="dont-index">
 <ul><li><p><a href="https://keras.io/api/layers/pooling_layers/global_max_pooling3d#globalmaxpooling3d-class" class="external-link">https://keras.io/api/layers/pooling_layers/global_max_pooling3d#globalmaxpooling3d-class</a></p></li>
 </ul><p>Other pooling layers: <br><code><a href="layer_average_pooling_1d.html">layer_average_pooling_1d</a>()</code> <br><code><a href="layer_average_pooling_2d.html">layer_average_pooling_2d</a>()</code> <br><code><a href="layer_average_pooling_3d.html">layer_average_pooling_3d</a>()</code> <br><code><a href="layer_global_average_pooling_1d.html">layer_global_average_pooling_1d</a>()</code> <br><code><a href="layer_global_average_pooling_2d.html">layer_global_average_pooling_2d</a>()</code> <br><code><a href="layer_global_average_pooling_3d.html">layer_global_average_pooling_3d</a>()</code> <br><code><a href="layer_global_max_pooling_1d.html">layer_global_max_pooling_1d</a>()</code> <br><code><a href="layer_global_max_pooling_2d.html">layer_global_max_pooling_2d</a>()</code> <br><code><a href="layer_max_pooling_1d.html">layer_max_pooling_1d</a>()</code> <br><code><a href="layer_max_pooling_2d.html">layer_max_pooling_2d</a>()</code> <br><code><a href="layer_max_pooling_3d.html">layer_max_pooling_3d</a>()</code> <br></p>
-<p>Other layers: <br><code><a href="Layer.html">Layer</a>()</code> <br><code><a href="layer_activation.html">layer_activation</a>()</code> <br><code><a href="layer_activation_elu.html">layer_activation_elu</a>()</code> <br><code><a href="layer_activation_leaky_relu.html">layer_activation_leaky_relu</a>()</code> <br><code><a href="layer_activation_parametric_relu.html">layer_activation_parametric_relu</a>()</code> <br><code><a href="layer_activation_relu.html">layer_activation_relu</a>()</code> <br><code><a href="layer_activation_softmax.html">layer_activation_softmax</a>()</code> <br><code><a href="layer_activity_regularization.html">layer_activity_regularization</a>()</code> <br><code><a href="layer_add.html">layer_add</a>()</code> <br><code><a href="layer_additive_attention.html">layer_additive_attention</a>()</code> <br><code><a href="layer_alpha_dropout.html">layer_alpha_dropout</a>()</code> <br><code><a href="layer_attention.html">layer_attention</a>()</code> <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_average.html">layer_average</a>()</code> <br><code><a href="layer_average_pooling_1d.html">layer_average_pooling_1d</a>()</code> <br><code><a href="layer_average_pooling_2d.html">layer_average_pooling_2d</a>()</code> <br><code><a href="layer_average_pooling_3d.html">layer_average_pooling_3d</a>()</code> <br><code><a href="layer_batch_normalization.html">layer_batch_normalization</a>()</code> <br><code><a href="layer_bidirectional.html">layer_bidirectional</a>()</code> <br><code><a href="layer_category_encoding.html">layer_category_encoding</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_concatenate.html">layer_concatenate</a>()</code> <br><code><a href="layer_conv_1d.html">layer_conv_1d</a>()</code> <br><code><a href="layer_conv_1d_transpose.html">layer_conv_1d_transpose</a>()</code> <br><code><a href="layer_conv_2d.html">layer_conv_2d</a>()</code> <br><code><a href="layer_conv_2d_transpose.html">layer_conv_2d_transpose</a>()</code> <br><code><a href="layer_conv_3d.html">layer_conv_3d</a>()</code> <br><code><a href="layer_conv_3d_transpose.html">layer_conv_3d_transpose</a>()</code> <br><code><a href="layer_conv_lstm_1d.html">layer_conv_lstm_1d</a>()</code> <br><code><a href="layer_conv_lstm_2d.html">layer_conv_lstm_2d</a>()</code> <br><code><a href="layer_conv_lstm_3d.html">layer_conv_lstm_3d</a>()</code> <br><code><a href="layer_cropping_1d.html">layer_cropping_1d</a>()</code> <br><code><a href="layer_cropping_2d.html">layer_cropping_2d</a>()</code> <br><code><a href="layer_cropping_3d.html">layer_cropping_3d</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_dense.html">layer_dense</a>()</code> <br><code><a href="layer_depthwise_conv_1d.html">layer_depthwise_conv_1d</a>()</code> <br><code><a href="layer_depthwise_conv_2d.html">layer_depthwise_conv_2d</a>()</code> <br><code><a href="layer_discretization.html">layer_discretization</a>()</code> <br><code><a href="layer_dot.html">layer_dot</a>()</code> <br><code><a href="layer_dropout.html">layer_dropout</a>()</code> <br><code><a href="layer_einsum_dense.html">layer_einsum_dense</a>()</code> <br><code><a href="layer_embedding.html">layer_embedding</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_feature_space.html">layer_feature_space</a>()</code> <br><code><a href="layer_flatten.html">layer_flatten</a>()</code> <br><code><a href="layer_flax_module_wrapper.html">layer_flax_module_wrapper</a>()</code> <br><code><a href="layer_gaussian_dropout.html">layer_gaussian_dropout</a>()</code> <br><code><a href="layer_gaussian_noise.html">layer_gaussian_noise</a>()</code> <br><code><a href="layer_global_average_pooling_1d.html">layer_global_average_pooling_1d</a>()</code> <br><code><a href="layer_global_average_pooling_2d.html">layer_global_average_pooling_2d</a>()</code> <br><code><a href="layer_global_average_pooling_3d.html">layer_global_average_pooling_3d</a>()</code> <br><code><a href="layer_global_max_pooling_1d.html">layer_global_max_pooling_1d</a>()</code> <br><code><a href="layer_global_max_pooling_2d.html">layer_global_max_pooling_2d</a>()</code> <br><code><a href="layer_group_normalization.html">layer_group_normalization</a>()</code> <br><code><a href="layer_group_query_attention.html">layer_group_query_attention</a>()</code> <br><code><a href="layer_gru.html">layer_gru</a>()</code> <br><code><a href="layer_hashed_crossing.html">layer_hashed_crossing</a>()</code> <br><code><a href="layer_hashing.html">layer_hashing</a>()</code> <br><code><a href="layer_identity.html">layer_identity</a>()</code> <br><code><a href="layer_integer_lookup.html">layer_integer_lookup</a>()</code> <br><code><a href="layer_jax_model_wrapper.html">layer_jax_model_wrapper</a>()</code> <br><code><a href="layer_lambda.html">layer_lambda</a>()</code> <br><code><a href="layer_layer_normalization.html">layer_layer_normalization</a>()</code> <br><code><a href="layer_lstm.html">layer_lstm</a>()</code> <br><code><a href="layer_masking.html">layer_masking</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_max_pooling_1d.html">layer_max_pooling_1d</a>()</code> <br><code><a href="layer_max_pooling_2d.html">layer_max_pooling_2d</a>()</code> <br><code><a href="layer_max_pooling_3d.html">layer_max_pooling_3d</a>()</code> <br><code><a href="layer_maximum.html">layer_maximum</a>()</code> <br><code><a href="layer_mel_spectrogram.html">layer_mel_spectrogram</a>()</code> <br><code><a href="layer_minimum.html">layer_minimum</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_multi_head_attention.html">layer_multi_head_attention</a>()</code> <br><code><a href="layer_multiply.html">layer_multiply</a>()</code> <br><code><a href="layer_normalization.html">layer_normalization</a>()</code> <br><code><a href="layer_permute.html">layer_permute</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_brightness.html">layer_random_brightness</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_contrast.html">layer_random_contrast</a>()</code> <br><code><a href="layer_random_crop.html">layer_random_crop</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_flip.html">layer_random_flip</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_rotation.html">layer_random_rotation</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_random_translation.html">layer_random_translation</a>()</code> <br><code><a href="layer_random_zoom.html">layer_random_zoom</a>()</code> <br><code><a href="layer_repeat_vector.html">layer_repeat_vector</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_reshape.html">layer_reshape</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_rms_normalization.html">layer_rms_normalization</a>()</code> <br><code><a href="layer_rnn.html">layer_rnn</a>()</code> <br><code><a href="layer_separable_conv_1d.html">layer_separable_conv_1d</a>()</code> <br><code><a href="layer_separable_conv_2d.html">layer_separable_conv_2d</a>()</code> <br><code><a href="layer_simple_rnn.html">layer_simple_rnn</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br><code><a href="layer_spatial_dropout_1d.html">layer_spatial_dropout_1d</a>()</code> <br><code><a href="layer_spatial_dropout_2d.html">layer_spatial_dropout_2d</a>()</code> <br><code><a href="layer_spatial_dropout_3d.html">layer_spatial_dropout_3d</a>()</code> <br><code><a href="layer_spectral_normalization.html">layer_spectral_normalization</a>()</code> <br><code><a href="layer_stft_spectrogram.html">layer_stft_spectrogram</a>()</code> <br><code><a href="layer_string_lookup.html">layer_string_lookup</a>()</code> <br><code><a href="layer_subtract.html">layer_subtract</a>()</code> <br><code><a href="layer_text_vectorization.html">layer_text_vectorization</a>()</code> <br><code><a href="layer_tfsm.html">layer_tfsm</a>()</code> <br><code><a href="layer_time_distributed.html">layer_time_distributed</a>()</code> <br><code><a href="layer_torch_module_wrapper.html">layer_torch_module_wrapper</a>()</code> <br><code><a href="layer_unit_normalization.html">layer_unit_normalization</a>()</code> <br><code><a href="layer_upsampling_1d.html">layer_upsampling_1d</a>()</code> <br><code><a href="layer_upsampling_2d.html">layer_upsampling_2d</a>()</code> <br><code><a href="layer_upsampling_3d.html">layer_upsampling_3d</a>()</code> <br><code><a href="layer_zero_padding_1d.html">layer_zero_padding_1d</a>()</code> <br><code><a href="layer_zero_padding_2d.html">layer_zero_padding_2d</a>()</code> <br><code><a href="layer_zero_padding_3d.html">layer_zero_padding_3d</a>()</code> <br><code><a href="rnn_cell_gru.html">rnn_cell_gru</a>()</code> <br><code><a href="rnn_cell_lstm.html">rnn_cell_lstm</a>()</code> <br><code><a href="rnn_cell_simple.html">rnn_cell_simple</a>()</code> <br><code><a href="rnn_cells_stack.html">rnn_cells_stack</a>()</code> <br></p></div>
+<p>Other layers: <br><code><a href="Layer.html">Layer</a>()</code> <br><code><a href="layer_activation.html">layer_activation</a>()</code> <br><code><a href="layer_activation_elu.html">layer_activation_elu</a>()</code> <br><code><a href="layer_activation_leaky_relu.html">layer_activation_leaky_relu</a>()</code> <br><code><a href="layer_activation_parametric_relu.html">layer_activation_parametric_relu</a>()</code> <br><code><a href="layer_activation_relu.html">layer_activation_relu</a>()</code> <br><code><a href="layer_activation_softmax.html">layer_activation_softmax</a>()</code> <br><code><a href="layer_activity_regularization.html">layer_activity_regularization</a>()</code> <br><code><a href="layer_add.html">layer_add</a>()</code> <br><code><a href="layer_additive_attention.html">layer_additive_attention</a>()</code> <br><code><a href="layer_alpha_dropout.html">layer_alpha_dropout</a>()</code> <br><code><a href="layer_attention.html">layer_attention</a>()</code> <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_average.html">layer_average</a>()</code> <br><code><a href="layer_average_pooling_1d.html">layer_average_pooling_1d</a>()</code> <br><code><a href="layer_average_pooling_2d.html">layer_average_pooling_2d</a>()</code> <br><code><a href="layer_average_pooling_3d.html">layer_average_pooling_3d</a>()</code> <br><code><a href="layer_batch_normalization.html">layer_batch_normalization</a>()</code> <br><code><a href="layer_bidirectional.html">layer_bidirectional</a>()</code> <br><code><a href="layer_category_encoding.html">layer_category_encoding</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_concatenate.html">layer_concatenate</a>()</code> <br><code><a href="layer_conv_1d.html">layer_conv_1d</a>()</code> <br><code><a href="layer_conv_1d_transpose.html">layer_conv_1d_transpose</a>()</code> <br><code><a href="layer_conv_2d.html">layer_conv_2d</a>()</code> <br><code><a href="layer_conv_2d_transpose.html">layer_conv_2d_transpose</a>()</code> <br><code><a href="layer_conv_3d.html">layer_conv_3d</a>()</code> <br><code><a href="layer_conv_3d_transpose.html">layer_conv_3d_transpose</a>()</code> <br><code><a href="layer_conv_lstm_1d.html">layer_conv_lstm_1d</a>()</code> <br><code><a href="layer_conv_lstm_2d.html">layer_conv_lstm_2d</a>()</code> <br><code><a href="layer_conv_lstm_3d.html">layer_conv_lstm_3d</a>()</code> <br><code><a href="layer_cropping_1d.html">layer_cropping_1d</a>()</code> <br><code><a href="layer_cropping_2d.html">layer_cropping_2d</a>()</code> <br><code><a href="layer_cropping_3d.html">layer_cropping_3d</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_dense.html">layer_dense</a>()</code> <br><code><a href="layer_depthwise_conv_1d.html">layer_depthwise_conv_1d</a>()</code> <br><code><a href="layer_depthwise_conv_2d.html">layer_depthwise_conv_2d</a>()</code> <br><code><a href="layer_discretization.html">layer_discretization</a>()</code> <br><code><a href="layer_dot.html">layer_dot</a>()</code> <br><code><a href="layer_dropout.html">layer_dropout</a>()</code> <br><code><a href="layer_einsum_dense.html">layer_einsum_dense</a>()</code> <br><code><a href="layer_embedding.html">layer_embedding</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_feature_space.html">layer_feature_space</a>()</code> <br><code><a href="layer_flatten.html">layer_flatten</a>()</code> <br><code><a href="layer_flax_module_wrapper.html">layer_flax_module_wrapper</a>()</code> <br><code><a href="layer_gaussian_dropout.html">layer_gaussian_dropout</a>()</code> <br><code><a href="layer_gaussian_noise.html">layer_gaussian_noise</a>()</code> <br><code><a href="layer_global_average_pooling_1d.html">layer_global_average_pooling_1d</a>()</code> <br><code><a href="layer_global_average_pooling_2d.html">layer_global_average_pooling_2d</a>()</code> <br><code><a href="layer_global_average_pooling_3d.html">layer_global_average_pooling_3d</a>()</code> <br><code><a href="layer_global_max_pooling_1d.html">layer_global_max_pooling_1d</a>()</code> <br><code><a href="layer_global_max_pooling_2d.html">layer_global_max_pooling_2d</a>()</code> <br><code><a href="layer_group_normalization.html">layer_group_normalization</a>()</code> <br><code><a href="layer_group_query_attention.html">layer_group_query_attention</a>()</code> <br><code><a href="layer_gru.html">layer_gru</a>()</code> <br><code><a href="layer_hashed_crossing.html">layer_hashed_crossing</a>()</code> <br><code><a href="layer_hashing.html">layer_hashing</a>()</code> <br><code><a href="layer_identity.html">layer_identity</a>()</code> <br><code><a href="layer_integer_lookup.html">layer_integer_lookup</a>()</code> <br><code><a href="layer_jax_model_wrapper.html">layer_jax_model_wrapper</a>()</code> <br><code><a href="layer_lambda.html">layer_lambda</a>()</code> <br><code><a href="layer_layer_normalization.html">layer_layer_normalization</a>()</code> <br><code><a href="layer_lstm.html">layer_lstm</a>()</code> <br><code><a href="layer_masking.html">layer_masking</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_max_pooling_1d.html">layer_max_pooling_1d</a>()</code> <br><code><a href="layer_max_pooling_2d.html">layer_max_pooling_2d</a>()</code> <br><code><a href="layer_max_pooling_3d.html">layer_max_pooling_3d</a>()</code> <br><code><a href="layer_maximum.html">layer_maximum</a>()</code> <br><code><a href="layer_mel_spectrogram.html">layer_mel_spectrogram</a>()</code> <br><code><a href="layer_minimum.html">layer_minimum</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_multi_head_attention.html">layer_multi_head_attention</a>()</code> <br><code><a href="layer_multiply.html">layer_multiply</a>()</code> <br><code><a href="layer_normalization.html">layer_normalization</a>()</code> <br><code><a href="layer_permute.html">layer_permute</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_brightness.html">layer_random_brightness</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_contrast.html">layer_random_contrast</a>()</code> <br><code><a href="layer_random_crop.html">layer_random_crop</a>()</code> <br><code><a href="layer_random_elastic_transform.html">layer_random_elastic_transform</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_flip.html">layer_random_flip</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_rotation.html">layer_random_rotation</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_random_translation.html">layer_random_translation</a>()</code> <br><code><a href="layer_random_zoom.html">layer_random_zoom</a>()</code> <br><code><a href="layer_repeat_vector.html">layer_repeat_vector</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_reshape.html">layer_reshape</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_rms_normalization.html">layer_rms_normalization</a>()</code> <br><code><a href="layer_rnn.html">layer_rnn</a>()</code> <br><code><a href="layer_separable_conv_1d.html">layer_separable_conv_1d</a>()</code> <br><code><a href="layer_separable_conv_2d.html">layer_separable_conv_2d</a>()</code> <br><code><a href="layer_simple_rnn.html">layer_simple_rnn</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br><code><a href="layer_spatial_dropout_1d.html">layer_spatial_dropout_1d</a>()</code> <br><code><a href="layer_spatial_dropout_2d.html">layer_spatial_dropout_2d</a>()</code> <br><code><a href="layer_spatial_dropout_3d.html">layer_spatial_dropout_3d</a>()</code> <br><code><a href="layer_spectral_normalization.html">layer_spectral_normalization</a>()</code> <br><code><a href="layer_stft_spectrogram.html">layer_stft_spectrogram</a>()</code> <br><code><a href="layer_string_lookup.html">layer_string_lookup</a>()</code> <br><code><a href="layer_subtract.html">layer_subtract</a>()</code> <br><code><a href="layer_text_vectorization.html">layer_text_vectorization</a>()</code> <br><code><a href="layer_tfsm.html">layer_tfsm</a>()</code> <br><code><a href="layer_time_distributed.html">layer_time_distributed</a>()</code> <br><code><a href="layer_torch_module_wrapper.html">layer_torch_module_wrapper</a>()</code> <br><code><a href="layer_unit_normalization.html">layer_unit_normalization</a>()</code> <br><code><a href="layer_upsampling_1d.html">layer_upsampling_1d</a>()</code> <br><code><a href="layer_upsampling_2d.html">layer_upsampling_2d</a>()</code> <br><code><a href="layer_upsampling_3d.html">layer_upsampling_3d</a>()</code> <br><code><a href="layer_zero_padding_1d.html">layer_zero_padding_1d</a>()</code> <br><code><a href="layer_zero_padding_2d.html">layer_zero_padding_2d</a>()</code> <br><code><a href="layer_zero_padding_3d.html">layer_zero_padding_3d</a>()</code> <br><code><a href="rnn_cell_gru.html">rnn_cell_gru</a>()</code> <br><code><a href="rnn_cell_lstm.html">rnn_cell_lstm</a>()</code> <br><code><a href="rnn_cell_simple.html">rnn_cell_simple</a>()</code> <br><code><a href="rnn_cells_stack.html">rnn_cells_stack</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -149,7 +149,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/layer_group_normalization.html b/docs/dev/reference/layer_group_normalization.html
index 797c8ba65..7897b0f4e 100644
--- a/docs/dev/reference/layer_group_normalization.html
+++ b/docs/dev/reference/layer_group_normalization.html
@@ -30,7 +30,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -210,7 +210,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
     <div class="dont-index">
 <ul><li><p><a href="https://keras.io/api/layers/normalization_layers/group_normalization#groupnormalization-class" class="external-link">https://keras.io/api/layers/normalization_layers/group_normalization#groupnormalization-class</a></p></li>
 </ul><p>Other normalization layers: <br><code><a href="layer_batch_normalization.html">layer_batch_normalization</a>()</code> <br><code><a href="layer_layer_normalization.html">layer_layer_normalization</a>()</code> <br><code><a href="layer_rms_normalization.html">layer_rms_normalization</a>()</code> <br><code><a href="layer_spectral_normalization.html">layer_spectral_normalization</a>()</code> <br><code><a href="layer_unit_normalization.html">layer_unit_normalization</a>()</code> <br></p>
-<p>Other layers: <br><code><a href="Layer.html">Layer</a>()</code> <br><code><a href="layer_activation.html">layer_activation</a>()</code> <br><code><a href="layer_activation_elu.html">layer_activation_elu</a>()</code> <br><code><a href="layer_activation_leaky_relu.html">layer_activation_leaky_relu</a>()</code> <br><code><a href="layer_activation_parametric_relu.html">layer_activation_parametric_relu</a>()</code> <br><code><a href="layer_activation_relu.html">layer_activation_relu</a>()</code> <br><code><a href="layer_activation_softmax.html">layer_activation_softmax</a>()</code> <br><code><a href="layer_activity_regularization.html">layer_activity_regularization</a>()</code> <br><code><a href="layer_add.html">layer_add</a>()</code> <br><code><a href="layer_additive_attention.html">layer_additive_attention</a>()</code> <br><code><a href="layer_alpha_dropout.html">layer_alpha_dropout</a>()</code> <br><code><a href="layer_attention.html">layer_attention</a>()</code> <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_average.html">layer_average</a>()</code> <br><code><a href="layer_average_pooling_1d.html">layer_average_pooling_1d</a>()</code> <br><code><a href="layer_average_pooling_2d.html">layer_average_pooling_2d</a>()</code> <br><code><a href="layer_average_pooling_3d.html">layer_average_pooling_3d</a>()</code> <br><code><a href="layer_batch_normalization.html">layer_batch_normalization</a>()</code> <br><code><a href="layer_bidirectional.html">layer_bidirectional</a>()</code> <br><code><a href="layer_category_encoding.html">layer_category_encoding</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_concatenate.html">layer_concatenate</a>()</code> <br><code><a href="layer_conv_1d.html">layer_conv_1d</a>()</code> <br><code><a href="layer_conv_1d_transpose.html">layer_conv_1d_transpose</a>()</code> <br><code><a href="layer_conv_2d.html">layer_conv_2d</a>()</code> <br><code><a href="layer_conv_2d_transpose.html">layer_conv_2d_transpose</a>()</code> <br><code><a href="layer_conv_3d.html">layer_conv_3d</a>()</code> <br><code><a href="layer_conv_3d_transpose.html">layer_conv_3d_transpose</a>()</code> <br><code><a href="layer_conv_lstm_1d.html">layer_conv_lstm_1d</a>()</code> <br><code><a href="layer_conv_lstm_2d.html">layer_conv_lstm_2d</a>()</code> <br><code><a href="layer_conv_lstm_3d.html">layer_conv_lstm_3d</a>()</code> <br><code><a href="layer_cropping_1d.html">layer_cropping_1d</a>()</code> <br><code><a href="layer_cropping_2d.html">layer_cropping_2d</a>()</code> <br><code><a href="layer_cropping_3d.html">layer_cropping_3d</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_dense.html">layer_dense</a>()</code> <br><code><a href="layer_depthwise_conv_1d.html">layer_depthwise_conv_1d</a>()</code> <br><code><a href="layer_depthwise_conv_2d.html">layer_depthwise_conv_2d</a>()</code> <br><code><a href="layer_discretization.html">layer_discretization</a>()</code> <br><code><a href="layer_dot.html">layer_dot</a>()</code> <br><code><a href="layer_dropout.html">layer_dropout</a>()</code> <br><code><a href="layer_einsum_dense.html">layer_einsum_dense</a>()</code> <br><code><a href="layer_embedding.html">layer_embedding</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_feature_space.html">layer_feature_space</a>()</code> <br><code><a href="layer_flatten.html">layer_flatten</a>()</code> <br><code><a href="layer_flax_module_wrapper.html">layer_flax_module_wrapper</a>()</code> <br><code><a href="layer_gaussian_dropout.html">layer_gaussian_dropout</a>()</code> <br><code><a href="layer_gaussian_noise.html">layer_gaussian_noise</a>()</code> <br><code><a href="layer_global_average_pooling_1d.html">layer_global_average_pooling_1d</a>()</code> <br><code><a href="layer_global_average_pooling_2d.html">layer_global_average_pooling_2d</a>()</code> <br><code><a href="layer_global_average_pooling_3d.html">layer_global_average_pooling_3d</a>()</code> <br><code><a href="layer_global_max_pooling_1d.html">layer_global_max_pooling_1d</a>()</code> <br><code><a href="layer_global_max_pooling_2d.html">layer_global_max_pooling_2d</a>()</code> <br><code><a href="layer_global_max_pooling_3d.html">layer_global_max_pooling_3d</a>()</code> <br><code><a href="layer_group_query_attention.html">layer_group_query_attention</a>()</code> <br><code><a href="layer_gru.html">layer_gru</a>()</code> <br><code><a href="layer_hashed_crossing.html">layer_hashed_crossing</a>()</code> <br><code><a href="layer_hashing.html">layer_hashing</a>()</code> <br><code><a href="layer_identity.html">layer_identity</a>()</code> <br><code><a href="layer_integer_lookup.html">layer_integer_lookup</a>()</code> <br><code><a href="layer_jax_model_wrapper.html">layer_jax_model_wrapper</a>()</code> <br><code><a href="layer_lambda.html">layer_lambda</a>()</code> <br><code><a href="layer_layer_normalization.html">layer_layer_normalization</a>()</code> <br><code><a href="layer_lstm.html">layer_lstm</a>()</code> <br><code><a href="layer_masking.html">layer_masking</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_max_pooling_1d.html">layer_max_pooling_1d</a>()</code> <br><code><a href="layer_max_pooling_2d.html">layer_max_pooling_2d</a>()</code> <br><code><a href="layer_max_pooling_3d.html">layer_max_pooling_3d</a>()</code> <br><code><a href="layer_maximum.html">layer_maximum</a>()</code> <br><code><a href="layer_mel_spectrogram.html">layer_mel_spectrogram</a>()</code> <br><code><a href="layer_minimum.html">layer_minimum</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_multi_head_attention.html">layer_multi_head_attention</a>()</code> <br><code><a href="layer_multiply.html">layer_multiply</a>()</code> <br><code><a href="layer_normalization.html">layer_normalization</a>()</code> <br><code><a href="layer_permute.html">layer_permute</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_brightness.html">layer_random_brightness</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_contrast.html">layer_random_contrast</a>()</code> <br><code><a href="layer_random_crop.html">layer_random_crop</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_flip.html">layer_random_flip</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_rotation.html">layer_random_rotation</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_random_translation.html">layer_random_translation</a>()</code> <br><code><a href="layer_random_zoom.html">layer_random_zoom</a>()</code> <br><code><a href="layer_repeat_vector.html">layer_repeat_vector</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_reshape.html">layer_reshape</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_rms_normalization.html">layer_rms_normalization</a>()</code> <br><code><a href="layer_rnn.html">layer_rnn</a>()</code> <br><code><a href="layer_separable_conv_1d.html">layer_separable_conv_1d</a>()</code> <br><code><a href="layer_separable_conv_2d.html">layer_separable_conv_2d</a>()</code> <br><code><a href="layer_simple_rnn.html">layer_simple_rnn</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br><code><a href="layer_spatial_dropout_1d.html">layer_spatial_dropout_1d</a>()</code> <br><code><a href="layer_spatial_dropout_2d.html">layer_spatial_dropout_2d</a>()</code> <br><code><a href="layer_spatial_dropout_3d.html">layer_spatial_dropout_3d</a>()</code> <br><code><a href="layer_spectral_normalization.html">layer_spectral_normalization</a>()</code> <br><code><a href="layer_stft_spectrogram.html">layer_stft_spectrogram</a>()</code> <br><code><a href="layer_string_lookup.html">layer_string_lookup</a>()</code> <br><code><a href="layer_subtract.html">layer_subtract</a>()</code> <br><code><a href="layer_text_vectorization.html">layer_text_vectorization</a>()</code> <br><code><a href="layer_tfsm.html">layer_tfsm</a>()</code> <br><code><a href="layer_time_distributed.html">layer_time_distributed</a>()</code> <br><code><a href="layer_torch_module_wrapper.html">layer_torch_module_wrapper</a>()</code> <br><code><a href="layer_unit_normalization.html">layer_unit_normalization</a>()</code> <br><code><a href="layer_upsampling_1d.html">layer_upsampling_1d</a>()</code> <br><code><a href="layer_upsampling_2d.html">layer_upsampling_2d</a>()</code> <br><code><a href="layer_upsampling_3d.html">layer_upsampling_3d</a>()</code> <br><code><a href="layer_zero_padding_1d.html">layer_zero_padding_1d</a>()</code> <br><code><a href="layer_zero_padding_2d.html">layer_zero_padding_2d</a>()</code> <br><code><a href="layer_zero_padding_3d.html">layer_zero_padding_3d</a>()</code> <br><code><a href="rnn_cell_gru.html">rnn_cell_gru</a>()</code> <br><code><a href="rnn_cell_lstm.html">rnn_cell_lstm</a>()</code> <br><code><a href="rnn_cell_simple.html">rnn_cell_simple</a>()</code> <br><code><a href="rnn_cells_stack.html">rnn_cells_stack</a>()</code> <br></p></div>
+<p>Other layers: <br><code><a href="Layer.html">Layer</a>()</code> <br><code><a href="layer_activation.html">layer_activation</a>()</code> <br><code><a href="layer_activation_elu.html">layer_activation_elu</a>()</code> <br><code><a href="layer_activation_leaky_relu.html">layer_activation_leaky_relu</a>()</code> <br><code><a href="layer_activation_parametric_relu.html">layer_activation_parametric_relu</a>()</code> <br><code><a href="layer_activation_relu.html">layer_activation_relu</a>()</code> <br><code><a href="layer_activation_softmax.html">layer_activation_softmax</a>()</code> <br><code><a href="layer_activity_regularization.html">layer_activity_regularization</a>()</code> <br><code><a href="layer_add.html">layer_add</a>()</code> <br><code><a href="layer_additive_attention.html">layer_additive_attention</a>()</code> <br><code><a href="layer_alpha_dropout.html">layer_alpha_dropout</a>()</code> <br><code><a href="layer_attention.html">layer_attention</a>()</code> <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_average.html">layer_average</a>()</code> <br><code><a href="layer_average_pooling_1d.html">layer_average_pooling_1d</a>()</code> <br><code><a href="layer_average_pooling_2d.html">layer_average_pooling_2d</a>()</code> <br><code><a href="layer_average_pooling_3d.html">layer_average_pooling_3d</a>()</code> <br><code><a href="layer_batch_normalization.html">layer_batch_normalization</a>()</code> <br><code><a href="layer_bidirectional.html">layer_bidirectional</a>()</code> <br><code><a href="layer_category_encoding.html">layer_category_encoding</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_concatenate.html">layer_concatenate</a>()</code> <br><code><a href="layer_conv_1d.html">layer_conv_1d</a>()</code> <br><code><a href="layer_conv_1d_transpose.html">layer_conv_1d_transpose</a>()</code> <br><code><a href="layer_conv_2d.html">layer_conv_2d</a>()</code> <br><code><a href="layer_conv_2d_transpose.html">layer_conv_2d_transpose</a>()</code> <br><code><a href="layer_conv_3d.html">layer_conv_3d</a>()</code> <br><code><a href="layer_conv_3d_transpose.html">layer_conv_3d_transpose</a>()</code> <br><code><a href="layer_conv_lstm_1d.html">layer_conv_lstm_1d</a>()</code> <br><code><a href="layer_conv_lstm_2d.html">layer_conv_lstm_2d</a>()</code> <br><code><a href="layer_conv_lstm_3d.html">layer_conv_lstm_3d</a>()</code> <br><code><a href="layer_cropping_1d.html">layer_cropping_1d</a>()</code> <br><code><a href="layer_cropping_2d.html">layer_cropping_2d</a>()</code> <br><code><a href="layer_cropping_3d.html">layer_cropping_3d</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_dense.html">layer_dense</a>()</code> <br><code><a href="layer_depthwise_conv_1d.html">layer_depthwise_conv_1d</a>()</code> <br><code><a href="layer_depthwise_conv_2d.html">layer_depthwise_conv_2d</a>()</code> <br><code><a href="layer_discretization.html">layer_discretization</a>()</code> <br><code><a href="layer_dot.html">layer_dot</a>()</code> <br><code><a href="layer_dropout.html">layer_dropout</a>()</code> <br><code><a href="layer_einsum_dense.html">layer_einsum_dense</a>()</code> <br><code><a href="layer_embedding.html">layer_embedding</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_feature_space.html">layer_feature_space</a>()</code> <br><code><a href="layer_flatten.html">layer_flatten</a>()</code> <br><code><a href="layer_flax_module_wrapper.html">layer_flax_module_wrapper</a>()</code> <br><code><a href="layer_gaussian_dropout.html">layer_gaussian_dropout</a>()</code> <br><code><a href="layer_gaussian_noise.html">layer_gaussian_noise</a>()</code> <br><code><a href="layer_global_average_pooling_1d.html">layer_global_average_pooling_1d</a>()</code> <br><code><a href="layer_global_average_pooling_2d.html">layer_global_average_pooling_2d</a>()</code> <br><code><a href="layer_global_average_pooling_3d.html">layer_global_average_pooling_3d</a>()</code> <br><code><a href="layer_global_max_pooling_1d.html">layer_global_max_pooling_1d</a>()</code> <br><code><a href="layer_global_max_pooling_2d.html">layer_global_max_pooling_2d</a>()</code> <br><code><a href="layer_global_max_pooling_3d.html">layer_global_max_pooling_3d</a>()</code> <br><code><a href="layer_group_query_attention.html">layer_group_query_attention</a>()</code> <br><code><a href="layer_gru.html">layer_gru</a>()</code> <br><code><a href="layer_hashed_crossing.html">layer_hashed_crossing</a>()</code> <br><code><a href="layer_hashing.html">layer_hashing</a>()</code> <br><code><a href="layer_identity.html">layer_identity</a>()</code> <br><code><a href="layer_integer_lookup.html">layer_integer_lookup</a>()</code> <br><code><a href="layer_jax_model_wrapper.html">layer_jax_model_wrapper</a>()</code> <br><code><a href="layer_lambda.html">layer_lambda</a>()</code> <br><code><a href="layer_layer_normalization.html">layer_layer_normalization</a>()</code> <br><code><a href="layer_lstm.html">layer_lstm</a>()</code> <br><code><a href="layer_masking.html">layer_masking</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_max_pooling_1d.html">layer_max_pooling_1d</a>()</code> <br><code><a href="layer_max_pooling_2d.html">layer_max_pooling_2d</a>()</code> <br><code><a href="layer_max_pooling_3d.html">layer_max_pooling_3d</a>()</code> <br><code><a href="layer_maximum.html">layer_maximum</a>()</code> <br><code><a href="layer_mel_spectrogram.html">layer_mel_spectrogram</a>()</code> <br><code><a href="layer_minimum.html">layer_minimum</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_multi_head_attention.html">layer_multi_head_attention</a>()</code> <br><code><a href="layer_multiply.html">layer_multiply</a>()</code> <br><code><a href="layer_normalization.html">layer_normalization</a>()</code> <br><code><a href="layer_permute.html">layer_permute</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_brightness.html">layer_random_brightness</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_contrast.html">layer_random_contrast</a>()</code> <br><code><a href="layer_random_crop.html">layer_random_crop</a>()</code> <br><code><a href="layer_random_elastic_transform.html">layer_random_elastic_transform</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_flip.html">layer_random_flip</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_rotation.html">layer_random_rotation</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_random_translation.html">layer_random_translation</a>()</code> <br><code><a href="layer_random_zoom.html">layer_random_zoom</a>()</code> <br><code><a href="layer_repeat_vector.html">layer_repeat_vector</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_reshape.html">layer_reshape</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_rms_normalization.html">layer_rms_normalization</a>()</code> <br><code><a href="layer_rnn.html">layer_rnn</a>()</code> <br><code><a href="layer_separable_conv_1d.html">layer_separable_conv_1d</a>()</code> <br><code><a href="layer_separable_conv_2d.html">layer_separable_conv_2d</a>()</code> <br><code><a href="layer_simple_rnn.html">layer_simple_rnn</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br><code><a href="layer_spatial_dropout_1d.html">layer_spatial_dropout_1d</a>()</code> <br><code><a href="layer_spatial_dropout_2d.html">layer_spatial_dropout_2d</a>()</code> <br><code><a href="layer_spatial_dropout_3d.html">layer_spatial_dropout_3d</a>()</code> <br><code><a href="layer_spectral_normalization.html">layer_spectral_normalization</a>()</code> <br><code><a href="layer_stft_spectrogram.html">layer_stft_spectrogram</a>()</code> <br><code><a href="layer_string_lookup.html">layer_string_lookup</a>()</code> <br><code><a href="layer_subtract.html">layer_subtract</a>()</code> <br><code><a href="layer_text_vectorization.html">layer_text_vectorization</a>()</code> <br><code><a href="layer_tfsm.html">layer_tfsm</a>()</code> <br><code><a href="layer_time_distributed.html">layer_time_distributed</a>()</code> <br><code><a href="layer_torch_module_wrapper.html">layer_torch_module_wrapper</a>()</code> <br><code><a href="layer_unit_normalization.html">layer_unit_normalization</a>()</code> <br><code><a href="layer_upsampling_1d.html">layer_upsampling_1d</a>()</code> <br><code><a href="layer_upsampling_2d.html">layer_upsampling_2d</a>()</code> <br><code><a href="layer_upsampling_3d.html">layer_upsampling_3d</a>()</code> <br><code><a href="layer_zero_padding_1d.html">layer_zero_padding_1d</a>()</code> <br><code><a href="layer_zero_padding_2d.html">layer_zero_padding_2d</a>()</code> <br><code><a href="layer_zero_padding_3d.html">layer_zero_padding_3d</a>()</code> <br><code><a href="rnn_cell_gru.html">rnn_cell_gru</a>()</code> <br><code><a href="rnn_cell_lstm.html">rnn_cell_lstm</a>()</code> <br><code><a href="rnn_cell_simple.html">rnn_cell_simple</a>()</code> <br><code><a href="rnn_cells_stack.html">rnn_cells_stack</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -222,7 +222,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/layer_group_query_attention.html b/docs/dev/reference/layer_group_query_attention.html
index 3bfca311d..19d8524de 100644
--- a/docs/dev/reference/layer_group_query_attention.html
+++ b/docs/dev/reference/layer_group_query_attention.html
@@ -30,7 +30,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -225,7 +225,7 @@ <h2 id="call-arguments">Call Arguments<a class="anchor" aria-label="anchor" href
     <div class="section level2">
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
     <div class="dont-index"><p>Other attention layers: <br><code><a href="layer_additive_attention.html">layer_additive_attention</a>()</code> <br><code><a href="layer_attention.html">layer_attention</a>()</code> <br><code><a href="layer_multi_head_attention.html">layer_multi_head_attention</a>()</code> <br></p>
-<p>Other layers: <br><code><a href="Layer.html">Layer</a>()</code> <br><code><a href="layer_activation.html">layer_activation</a>()</code> <br><code><a href="layer_activation_elu.html">layer_activation_elu</a>()</code> <br><code><a href="layer_activation_leaky_relu.html">layer_activation_leaky_relu</a>()</code> <br><code><a href="layer_activation_parametric_relu.html">layer_activation_parametric_relu</a>()</code> <br><code><a href="layer_activation_relu.html">layer_activation_relu</a>()</code> <br><code><a href="layer_activation_softmax.html">layer_activation_softmax</a>()</code> <br><code><a href="layer_activity_regularization.html">layer_activity_regularization</a>()</code> <br><code><a href="layer_add.html">layer_add</a>()</code> <br><code><a href="layer_additive_attention.html">layer_additive_attention</a>()</code> <br><code><a href="layer_alpha_dropout.html">layer_alpha_dropout</a>()</code> <br><code><a href="layer_attention.html">layer_attention</a>()</code> <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_average.html">layer_average</a>()</code> <br><code><a href="layer_average_pooling_1d.html">layer_average_pooling_1d</a>()</code> <br><code><a href="layer_average_pooling_2d.html">layer_average_pooling_2d</a>()</code> <br><code><a href="layer_average_pooling_3d.html">layer_average_pooling_3d</a>()</code> <br><code><a href="layer_batch_normalization.html">layer_batch_normalization</a>()</code> <br><code><a href="layer_bidirectional.html">layer_bidirectional</a>()</code> <br><code><a href="layer_category_encoding.html">layer_category_encoding</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_concatenate.html">layer_concatenate</a>()</code> <br><code><a href="layer_conv_1d.html">layer_conv_1d</a>()</code> <br><code><a href="layer_conv_1d_transpose.html">layer_conv_1d_transpose</a>()</code> <br><code><a href="layer_conv_2d.html">layer_conv_2d</a>()</code> <br><code><a href="layer_conv_2d_transpose.html">layer_conv_2d_transpose</a>()</code> <br><code><a href="layer_conv_3d.html">layer_conv_3d</a>()</code> <br><code><a href="layer_conv_3d_transpose.html">layer_conv_3d_transpose</a>()</code> <br><code><a href="layer_conv_lstm_1d.html">layer_conv_lstm_1d</a>()</code> <br><code><a href="layer_conv_lstm_2d.html">layer_conv_lstm_2d</a>()</code> <br><code><a href="layer_conv_lstm_3d.html">layer_conv_lstm_3d</a>()</code> <br><code><a href="layer_cropping_1d.html">layer_cropping_1d</a>()</code> <br><code><a href="layer_cropping_2d.html">layer_cropping_2d</a>()</code> <br><code><a href="layer_cropping_3d.html">layer_cropping_3d</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_dense.html">layer_dense</a>()</code> <br><code><a href="layer_depthwise_conv_1d.html">layer_depthwise_conv_1d</a>()</code> <br><code><a href="layer_depthwise_conv_2d.html">layer_depthwise_conv_2d</a>()</code> <br><code><a href="layer_discretization.html">layer_discretization</a>()</code> <br><code><a href="layer_dot.html">layer_dot</a>()</code> <br><code><a href="layer_dropout.html">layer_dropout</a>()</code> <br><code><a href="layer_einsum_dense.html">layer_einsum_dense</a>()</code> <br><code><a href="layer_embedding.html">layer_embedding</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_feature_space.html">layer_feature_space</a>()</code> <br><code><a href="layer_flatten.html">layer_flatten</a>()</code> <br><code><a href="layer_flax_module_wrapper.html">layer_flax_module_wrapper</a>()</code> <br><code><a href="layer_gaussian_dropout.html">layer_gaussian_dropout</a>()</code> <br><code><a href="layer_gaussian_noise.html">layer_gaussian_noise</a>()</code> <br><code><a href="layer_global_average_pooling_1d.html">layer_global_average_pooling_1d</a>()</code> <br><code><a href="layer_global_average_pooling_2d.html">layer_global_average_pooling_2d</a>()</code> <br><code><a href="layer_global_average_pooling_3d.html">layer_global_average_pooling_3d</a>()</code> <br><code><a href="layer_global_max_pooling_1d.html">layer_global_max_pooling_1d</a>()</code> <br><code><a href="layer_global_max_pooling_2d.html">layer_global_max_pooling_2d</a>()</code> <br><code><a href="layer_global_max_pooling_3d.html">layer_global_max_pooling_3d</a>()</code> <br><code><a href="layer_group_normalization.html">layer_group_normalization</a>()</code> <br><code><a href="layer_gru.html">layer_gru</a>()</code> <br><code><a href="layer_hashed_crossing.html">layer_hashed_crossing</a>()</code> <br><code><a href="layer_hashing.html">layer_hashing</a>()</code> <br><code><a href="layer_identity.html">layer_identity</a>()</code> <br><code><a href="layer_integer_lookup.html">layer_integer_lookup</a>()</code> <br><code><a href="layer_jax_model_wrapper.html">layer_jax_model_wrapper</a>()</code> <br><code><a href="layer_lambda.html">layer_lambda</a>()</code> <br><code><a href="layer_layer_normalization.html">layer_layer_normalization</a>()</code> <br><code><a href="layer_lstm.html">layer_lstm</a>()</code> <br><code><a href="layer_masking.html">layer_masking</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_max_pooling_1d.html">layer_max_pooling_1d</a>()</code> <br><code><a href="layer_max_pooling_2d.html">layer_max_pooling_2d</a>()</code> <br><code><a href="layer_max_pooling_3d.html">layer_max_pooling_3d</a>()</code> <br><code><a href="layer_maximum.html">layer_maximum</a>()</code> <br><code><a href="layer_mel_spectrogram.html">layer_mel_spectrogram</a>()</code> <br><code><a href="layer_minimum.html">layer_minimum</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_multi_head_attention.html">layer_multi_head_attention</a>()</code> <br><code><a href="layer_multiply.html">layer_multiply</a>()</code> <br><code><a href="layer_normalization.html">layer_normalization</a>()</code> <br><code><a href="layer_permute.html">layer_permute</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_brightness.html">layer_random_brightness</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_contrast.html">layer_random_contrast</a>()</code> <br><code><a href="layer_random_crop.html">layer_random_crop</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_flip.html">layer_random_flip</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_rotation.html">layer_random_rotation</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_random_translation.html">layer_random_translation</a>()</code> <br><code><a href="layer_random_zoom.html">layer_random_zoom</a>()</code> <br><code><a href="layer_repeat_vector.html">layer_repeat_vector</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_reshape.html">layer_reshape</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_rms_normalization.html">layer_rms_normalization</a>()</code> <br><code><a href="layer_rnn.html">layer_rnn</a>()</code> <br><code><a href="layer_separable_conv_1d.html">layer_separable_conv_1d</a>()</code> <br><code><a href="layer_separable_conv_2d.html">layer_separable_conv_2d</a>()</code> <br><code><a href="layer_simple_rnn.html">layer_simple_rnn</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br><code><a href="layer_spatial_dropout_1d.html">layer_spatial_dropout_1d</a>()</code> <br><code><a href="layer_spatial_dropout_2d.html">layer_spatial_dropout_2d</a>()</code> <br><code><a href="layer_spatial_dropout_3d.html">layer_spatial_dropout_3d</a>()</code> <br><code><a href="layer_spectral_normalization.html">layer_spectral_normalization</a>()</code> <br><code><a href="layer_stft_spectrogram.html">layer_stft_spectrogram</a>()</code> <br><code><a href="layer_string_lookup.html">layer_string_lookup</a>()</code> <br><code><a href="layer_subtract.html">layer_subtract</a>()</code> <br><code><a href="layer_text_vectorization.html">layer_text_vectorization</a>()</code> <br><code><a href="layer_tfsm.html">layer_tfsm</a>()</code> <br><code><a href="layer_time_distributed.html">layer_time_distributed</a>()</code> <br><code><a href="layer_torch_module_wrapper.html">layer_torch_module_wrapper</a>()</code> <br><code><a href="layer_unit_normalization.html">layer_unit_normalization</a>()</code> <br><code><a href="layer_upsampling_1d.html">layer_upsampling_1d</a>()</code> <br><code><a href="layer_upsampling_2d.html">layer_upsampling_2d</a>()</code> <br><code><a href="layer_upsampling_3d.html">layer_upsampling_3d</a>()</code> <br><code><a href="layer_zero_padding_1d.html">layer_zero_padding_1d</a>()</code> <br><code><a href="layer_zero_padding_2d.html">layer_zero_padding_2d</a>()</code> <br><code><a href="layer_zero_padding_3d.html">layer_zero_padding_3d</a>()</code> <br><code><a href="rnn_cell_gru.html">rnn_cell_gru</a>()</code> <br><code><a href="rnn_cell_lstm.html">rnn_cell_lstm</a>()</code> <br><code><a href="rnn_cell_simple.html">rnn_cell_simple</a>()</code> <br><code><a href="rnn_cells_stack.html">rnn_cells_stack</a>()</code> <br></p></div>
+<p>Other layers: <br><code><a href="Layer.html">Layer</a>()</code> <br><code><a href="layer_activation.html">layer_activation</a>()</code> <br><code><a href="layer_activation_elu.html">layer_activation_elu</a>()</code> <br><code><a href="layer_activation_leaky_relu.html">layer_activation_leaky_relu</a>()</code> <br><code><a href="layer_activation_parametric_relu.html">layer_activation_parametric_relu</a>()</code> <br><code><a href="layer_activation_relu.html">layer_activation_relu</a>()</code> <br><code><a href="layer_activation_softmax.html">layer_activation_softmax</a>()</code> <br><code><a href="layer_activity_regularization.html">layer_activity_regularization</a>()</code> <br><code><a href="layer_add.html">layer_add</a>()</code> <br><code><a href="layer_additive_attention.html">layer_additive_attention</a>()</code> <br><code><a href="layer_alpha_dropout.html">layer_alpha_dropout</a>()</code> <br><code><a href="layer_attention.html">layer_attention</a>()</code> <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_average.html">layer_average</a>()</code> <br><code><a href="layer_average_pooling_1d.html">layer_average_pooling_1d</a>()</code> <br><code><a href="layer_average_pooling_2d.html">layer_average_pooling_2d</a>()</code> <br><code><a href="layer_average_pooling_3d.html">layer_average_pooling_3d</a>()</code> <br><code><a href="layer_batch_normalization.html">layer_batch_normalization</a>()</code> <br><code><a href="layer_bidirectional.html">layer_bidirectional</a>()</code> <br><code><a href="layer_category_encoding.html">layer_category_encoding</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_concatenate.html">layer_concatenate</a>()</code> <br><code><a href="layer_conv_1d.html">layer_conv_1d</a>()</code> <br><code><a href="layer_conv_1d_transpose.html">layer_conv_1d_transpose</a>()</code> <br><code><a href="layer_conv_2d.html">layer_conv_2d</a>()</code> <br><code><a href="layer_conv_2d_transpose.html">layer_conv_2d_transpose</a>()</code> <br><code><a href="layer_conv_3d.html">layer_conv_3d</a>()</code> <br><code><a href="layer_conv_3d_transpose.html">layer_conv_3d_transpose</a>()</code> <br><code><a href="layer_conv_lstm_1d.html">layer_conv_lstm_1d</a>()</code> <br><code><a href="layer_conv_lstm_2d.html">layer_conv_lstm_2d</a>()</code> <br><code><a href="layer_conv_lstm_3d.html">layer_conv_lstm_3d</a>()</code> <br><code><a href="layer_cropping_1d.html">layer_cropping_1d</a>()</code> <br><code><a href="layer_cropping_2d.html">layer_cropping_2d</a>()</code> <br><code><a href="layer_cropping_3d.html">layer_cropping_3d</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_dense.html">layer_dense</a>()</code> <br><code><a href="layer_depthwise_conv_1d.html">layer_depthwise_conv_1d</a>()</code> <br><code><a href="layer_depthwise_conv_2d.html">layer_depthwise_conv_2d</a>()</code> <br><code><a href="layer_discretization.html">layer_discretization</a>()</code> <br><code><a href="layer_dot.html">layer_dot</a>()</code> <br><code><a href="layer_dropout.html">layer_dropout</a>()</code> <br><code><a href="layer_einsum_dense.html">layer_einsum_dense</a>()</code> <br><code><a href="layer_embedding.html">layer_embedding</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_feature_space.html">layer_feature_space</a>()</code> <br><code><a href="layer_flatten.html">layer_flatten</a>()</code> <br><code><a href="layer_flax_module_wrapper.html">layer_flax_module_wrapper</a>()</code> <br><code><a href="layer_gaussian_dropout.html">layer_gaussian_dropout</a>()</code> <br><code><a href="layer_gaussian_noise.html">layer_gaussian_noise</a>()</code> <br><code><a href="layer_global_average_pooling_1d.html">layer_global_average_pooling_1d</a>()</code> <br><code><a href="layer_global_average_pooling_2d.html">layer_global_average_pooling_2d</a>()</code> <br><code><a href="layer_global_average_pooling_3d.html">layer_global_average_pooling_3d</a>()</code> <br><code><a href="layer_global_max_pooling_1d.html">layer_global_max_pooling_1d</a>()</code> <br><code><a href="layer_global_max_pooling_2d.html">layer_global_max_pooling_2d</a>()</code> <br><code><a href="layer_global_max_pooling_3d.html">layer_global_max_pooling_3d</a>()</code> <br><code><a href="layer_group_normalization.html">layer_group_normalization</a>()</code> <br><code><a href="layer_gru.html">layer_gru</a>()</code> <br><code><a href="layer_hashed_crossing.html">layer_hashed_crossing</a>()</code> <br><code><a href="layer_hashing.html">layer_hashing</a>()</code> <br><code><a href="layer_identity.html">layer_identity</a>()</code> <br><code><a href="layer_integer_lookup.html">layer_integer_lookup</a>()</code> <br><code><a href="layer_jax_model_wrapper.html">layer_jax_model_wrapper</a>()</code> <br><code><a href="layer_lambda.html">layer_lambda</a>()</code> <br><code><a href="layer_layer_normalization.html">layer_layer_normalization</a>()</code> <br><code><a href="layer_lstm.html">layer_lstm</a>()</code> <br><code><a href="layer_masking.html">layer_masking</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_max_pooling_1d.html">layer_max_pooling_1d</a>()</code> <br><code><a href="layer_max_pooling_2d.html">layer_max_pooling_2d</a>()</code> <br><code><a href="layer_max_pooling_3d.html">layer_max_pooling_3d</a>()</code> <br><code><a href="layer_maximum.html">layer_maximum</a>()</code> <br><code><a href="layer_mel_spectrogram.html">layer_mel_spectrogram</a>()</code> <br><code><a href="layer_minimum.html">layer_minimum</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_multi_head_attention.html">layer_multi_head_attention</a>()</code> <br><code><a href="layer_multiply.html">layer_multiply</a>()</code> <br><code><a href="layer_normalization.html">layer_normalization</a>()</code> <br><code><a href="layer_permute.html">layer_permute</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_brightness.html">layer_random_brightness</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_contrast.html">layer_random_contrast</a>()</code> <br><code><a href="layer_random_crop.html">layer_random_crop</a>()</code> <br><code><a href="layer_random_elastic_transform.html">layer_random_elastic_transform</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_flip.html">layer_random_flip</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_rotation.html">layer_random_rotation</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_random_translation.html">layer_random_translation</a>()</code> <br><code><a href="layer_random_zoom.html">layer_random_zoom</a>()</code> <br><code><a href="layer_repeat_vector.html">layer_repeat_vector</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_reshape.html">layer_reshape</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_rms_normalization.html">layer_rms_normalization</a>()</code> <br><code><a href="layer_rnn.html">layer_rnn</a>()</code> <br><code><a href="layer_separable_conv_1d.html">layer_separable_conv_1d</a>()</code> <br><code><a href="layer_separable_conv_2d.html">layer_separable_conv_2d</a>()</code> <br><code><a href="layer_simple_rnn.html">layer_simple_rnn</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br><code><a href="layer_spatial_dropout_1d.html">layer_spatial_dropout_1d</a>()</code> <br><code><a href="layer_spatial_dropout_2d.html">layer_spatial_dropout_2d</a>()</code> <br><code><a href="layer_spatial_dropout_3d.html">layer_spatial_dropout_3d</a>()</code> <br><code><a href="layer_spectral_normalization.html">layer_spectral_normalization</a>()</code> <br><code><a href="layer_stft_spectrogram.html">layer_stft_spectrogram</a>()</code> <br><code><a href="layer_string_lookup.html">layer_string_lookup</a>()</code> <br><code><a href="layer_subtract.html">layer_subtract</a>()</code> <br><code><a href="layer_text_vectorization.html">layer_text_vectorization</a>()</code> <br><code><a href="layer_tfsm.html">layer_tfsm</a>()</code> <br><code><a href="layer_time_distributed.html">layer_time_distributed</a>()</code> <br><code><a href="layer_torch_module_wrapper.html">layer_torch_module_wrapper</a>()</code> <br><code><a href="layer_unit_normalization.html">layer_unit_normalization</a>()</code> <br><code><a href="layer_upsampling_1d.html">layer_upsampling_1d</a>()</code> <br><code><a href="layer_upsampling_2d.html">layer_upsampling_2d</a>()</code> <br><code><a href="layer_upsampling_3d.html">layer_upsampling_3d</a>()</code> <br><code><a href="layer_zero_padding_1d.html">layer_zero_padding_1d</a>()</code> <br><code><a href="layer_zero_padding_2d.html">layer_zero_padding_2d</a>()</code> <br><code><a href="layer_zero_padding_3d.html">layer_zero_padding_3d</a>()</code> <br><code><a href="rnn_cell_gru.html">rnn_cell_gru</a>()</code> <br><code><a href="rnn_cell_lstm.html">rnn_cell_lstm</a>()</code> <br><code><a href="rnn_cell_simple.html">rnn_cell_simple</a>()</code> <br><code><a href="rnn_cells_stack.html">rnn_cells_stack</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -237,7 +237,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/layer_gru.html b/docs/dev/reference/layer_gru.html
index 0e554e99e..ae1c9dfcb 100644
--- a/docs/dev/reference/layer_gru.html
+++ b/docs/dev/reference/layer_gru.html
@@ -98,7 +98,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -391,7 +391,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 <ul><li><p><a href="https://keras.io/api/layers/recurrent_layers/gru#gru-class" class="external-link">https://keras.io/api/layers/recurrent_layers/gru#gru-class</a></p></li>
 </ul><p>Other gru rnn layers: <br><code><a href="rnn_cell_gru.html">rnn_cell_gru</a>()</code> <br></p>
 <p>Other rnn layers: <br><code><a href="layer_bidirectional.html">layer_bidirectional</a>()</code> <br><code><a href="layer_conv_lstm_1d.html">layer_conv_lstm_1d</a>()</code> <br><code><a href="layer_conv_lstm_2d.html">layer_conv_lstm_2d</a>()</code> <br><code><a href="layer_conv_lstm_3d.html">layer_conv_lstm_3d</a>()</code> <br><code><a href="layer_lstm.html">layer_lstm</a>()</code> <br><code><a href="layer_rnn.html">layer_rnn</a>()</code> <br><code><a href="layer_simple_rnn.html">layer_simple_rnn</a>()</code> <br><code><a href="layer_time_distributed.html">layer_time_distributed</a>()</code> <br><code><a href="rnn_cell_gru.html">rnn_cell_gru</a>()</code> <br><code><a href="rnn_cell_lstm.html">rnn_cell_lstm</a>()</code> <br><code><a href="rnn_cell_simple.html">rnn_cell_simple</a>()</code> <br><code><a href="rnn_cells_stack.html">rnn_cells_stack</a>()</code> <br></p>
-<p>Other layers: <br><code><a href="Layer.html">Layer</a>()</code> <br><code><a href="layer_activation.html">layer_activation</a>()</code> <br><code><a href="layer_activation_elu.html">layer_activation_elu</a>()</code> <br><code><a href="layer_activation_leaky_relu.html">layer_activation_leaky_relu</a>()</code> <br><code><a href="layer_activation_parametric_relu.html">layer_activation_parametric_relu</a>()</code> <br><code><a href="layer_activation_relu.html">layer_activation_relu</a>()</code> <br><code><a href="layer_activation_softmax.html">layer_activation_softmax</a>()</code> <br><code><a href="layer_activity_regularization.html">layer_activity_regularization</a>()</code> <br><code><a href="layer_add.html">layer_add</a>()</code> <br><code><a href="layer_additive_attention.html">layer_additive_attention</a>()</code> <br><code><a href="layer_alpha_dropout.html">layer_alpha_dropout</a>()</code> <br><code><a href="layer_attention.html">layer_attention</a>()</code> <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_average.html">layer_average</a>()</code> <br><code><a href="layer_average_pooling_1d.html">layer_average_pooling_1d</a>()</code> <br><code><a href="layer_average_pooling_2d.html">layer_average_pooling_2d</a>()</code> <br><code><a href="layer_average_pooling_3d.html">layer_average_pooling_3d</a>()</code> <br><code><a href="layer_batch_normalization.html">layer_batch_normalization</a>()</code> <br><code><a href="layer_bidirectional.html">layer_bidirectional</a>()</code> <br><code><a href="layer_category_encoding.html">layer_category_encoding</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_concatenate.html">layer_concatenate</a>()</code> <br><code><a href="layer_conv_1d.html">layer_conv_1d</a>()</code> <br><code><a href="layer_conv_1d_transpose.html">layer_conv_1d_transpose</a>()</code> <br><code><a href="layer_conv_2d.html">layer_conv_2d</a>()</code> <br><code><a href="layer_conv_2d_transpose.html">layer_conv_2d_transpose</a>()</code> <br><code><a href="layer_conv_3d.html">layer_conv_3d</a>()</code> <br><code><a href="layer_conv_3d_transpose.html">layer_conv_3d_transpose</a>()</code> <br><code><a href="layer_conv_lstm_1d.html">layer_conv_lstm_1d</a>()</code> <br><code><a href="layer_conv_lstm_2d.html">layer_conv_lstm_2d</a>()</code> <br><code><a href="layer_conv_lstm_3d.html">layer_conv_lstm_3d</a>()</code> <br><code><a href="layer_cropping_1d.html">layer_cropping_1d</a>()</code> <br><code><a href="layer_cropping_2d.html">layer_cropping_2d</a>()</code> <br><code><a href="layer_cropping_3d.html">layer_cropping_3d</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_dense.html">layer_dense</a>()</code> <br><code><a href="layer_depthwise_conv_1d.html">layer_depthwise_conv_1d</a>()</code> <br><code><a href="layer_depthwise_conv_2d.html">layer_depthwise_conv_2d</a>()</code> <br><code><a href="layer_discretization.html">layer_discretization</a>()</code> <br><code><a href="layer_dot.html">layer_dot</a>()</code> <br><code><a href="layer_dropout.html">layer_dropout</a>()</code> <br><code><a href="layer_einsum_dense.html">layer_einsum_dense</a>()</code> <br><code><a href="layer_embedding.html">layer_embedding</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_feature_space.html">layer_feature_space</a>()</code> <br><code><a href="layer_flatten.html">layer_flatten</a>()</code> <br><code><a href="layer_flax_module_wrapper.html">layer_flax_module_wrapper</a>()</code> <br><code><a href="layer_gaussian_dropout.html">layer_gaussian_dropout</a>()</code> <br><code><a href="layer_gaussian_noise.html">layer_gaussian_noise</a>()</code> <br><code><a href="layer_global_average_pooling_1d.html">layer_global_average_pooling_1d</a>()</code> <br><code><a href="layer_global_average_pooling_2d.html">layer_global_average_pooling_2d</a>()</code> <br><code><a href="layer_global_average_pooling_3d.html">layer_global_average_pooling_3d</a>()</code> <br><code><a href="layer_global_max_pooling_1d.html">layer_global_max_pooling_1d</a>()</code> <br><code><a href="layer_global_max_pooling_2d.html">layer_global_max_pooling_2d</a>()</code> <br><code><a href="layer_global_max_pooling_3d.html">layer_global_max_pooling_3d</a>()</code> <br><code><a href="layer_group_normalization.html">layer_group_normalization</a>()</code> <br><code><a href="layer_group_query_attention.html">layer_group_query_attention</a>()</code> <br><code><a href="layer_hashed_crossing.html">layer_hashed_crossing</a>()</code> <br><code><a href="layer_hashing.html">layer_hashing</a>()</code> <br><code><a href="layer_identity.html">layer_identity</a>()</code> <br><code><a href="layer_integer_lookup.html">layer_integer_lookup</a>()</code> <br><code><a href="layer_jax_model_wrapper.html">layer_jax_model_wrapper</a>()</code> <br><code><a href="layer_lambda.html">layer_lambda</a>()</code> <br><code><a href="layer_layer_normalization.html">layer_layer_normalization</a>()</code> <br><code><a href="layer_lstm.html">layer_lstm</a>()</code> <br><code><a href="layer_masking.html">layer_masking</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_max_pooling_1d.html">layer_max_pooling_1d</a>()</code> <br><code><a href="layer_max_pooling_2d.html">layer_max_pooling_2d</a>()</code> <br><code><a href="layer_max_pooling_3d.html">layer_max_pooling_3d</a>()</code> <br><code><a href="layer_maximum.html">layer_maximum</a>()</code> <br><code><a href="layer_mel_spectrogram.html">layer_mel_spectrogram</a>()</code> <br><code><a href="layer_minimum.html">layer_minimum</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_multi_head_attention.html">layer_multi_head_attention</a>()</code> <br><code><a href="layer_multiply.html">layer_multiply</a>()</code> <br><code><a href="layer_normalization.html">layer_normalization</a>()</code> <br><code><a href="layer_permute.html">layer_permute</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_brightness.html">layer_random_brightness</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_contrast.html">layer_random_contrast</a>()</code> <br><code><a href="layer_random_crop.html">layer_random_crop</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_flip.html">layer_random_flip</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_rotation.html">layer_random_rotation</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_random_translation.html">layer_random_translation</a>()</code> <br><code><a href="layer_random_zoom.html">layer_random_zoom</a>()</code> <br><code><a href="layer_repeat_vector.html">layer_repeat_vector</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_reshape.html">layer_reshape</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_rms_normalization.html">layer_rms_normalization</a>()</code> <br><code><a href="layer_rnn.html">layer_rnn</a>()</code> <br><code><a href="layer_separable_conv_1d.html">layer_separable_conv_1d</a>()</code> <br><code><a href="layer_separable_conv_2d.html">layer_separable_conv_2d</a>()</code> <br><code><a href="layer_simple_rnn.html">layer_simple_rnn</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br><code><a href="layer_spatial_dropout_1d.html">layer_spatial_dropout_1d</a>()</code> <br><code><a href="layer_spatial_dropout_2d.html">layer_spatial_dropout_2d</a>()</code> <br><code><a href="layer_spatial_dropout_3d.html">layer_spatial_dropout_3d</a>()</code> <br><code><a href="layer_spectral_normalization.html">layer_spectral_normalization</a>()</code> <br><code><a href="layer_stft_spectrogram.html">layer_stft_spectrogram</a>()</code> <br><code><a href="layer_string_lookup.html">layer_string_lookup</a>()</code> <br><code><a href="layer_subtract.html">layer_subtract</a>()</code> <br><code><a href="layer_text_vectorization.html">layer_text_vectorization</a>()</code> <br><code><a href="layer_tfsm.html">layer_tfsm</a>()</code> <br><code><a href="layer_time_distributed.html">layer_time_distributed</a>()</code> <br><code><a href="layer_torch_module_wrapper.html">layer_torch_module_wrapper</a>()</code> <br><code><a href="layer_unit_normalization.html">layer_unit_normalization</a>()</code> <br><code><a href="layer_upsampling_1d.html">layer_upsampling_1d</a>()</code> <br><code><a href="layer_upsampling_2d.html">layer_upsampling_2d</a>()</code> <br><code><a href="layer_upsampling_3d.html">layer_upsampling_3d</a>()</code> <br><code><a href="layer_zero_padding_1d.html">layer_zero_padding_1d</a>()</code> <br><code><a href="layer_zero_padding_2d.html">layer_zero_padding_2d</a>()</code> <br><code><a href="layer_zero_padding_3d.html">layer_zero_padding_3d</a>()</code> <br><code><a href="rnn_cell_gru.html">rnn_cell_gru</a>()</code> <br><code><a href="rnn_cell_lstm.html">rnn_cell_lstm</a>()</code> <br><code><a href="rnn_cell_simple.html">rnn_cell_simple</a>()</code> <br><code><a href="rnn_cells_stack.html">rnn_cells_stack</a>()</code> <br></p></div>
+<p>Other layers: <br><code><a href="Layer.html">Layer</a>()</code> <br><code><a href="layer_activation.html">layer_activation</a>()</code> <br><code><a href="layer_activation_elu.html">layer_activation_elu</a>()</code> <br><code><a href="layer_activation_leaky_relu.html">layer_activation_leaky_relu</a>()</code> <br><code><a href="layer_activation_parametric_relu.html">layer_activation_parametric_relu</a>()</code> <br><code><a href="layer_activation_relu.html">layer_activation_relu</a>()</code> <br><code><a href="layer_activation_softmax.html">layer_activation_softmax</a>()</code> <br><code><a href="layer_activity_regularization.html">layer_activity_regularization</a>()</code> <br><code><a href="layer_add.html">layer_add</a>()</code> <br><code><a href="layer_additive_attention.html">layer_additive_attention</a>()</code> <br><code><a href="layer_alpha_dropout.html">layer_alpha_dropout</a>()</code> <br><code><a href="layer_attention.html">layer_attention</a>()</code> <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_average.html">layer_average</a>()</code> <br><code><a href="layer_average_pooling_1d.html">layer_average_pooling_1d</a>()</code> <br><code><a href="layer_average_pooling_2d.html">layer_average_pooling_2d</a>()</code> <br><code><a href="layer_average_pooling_3d.html">layer_average_pooling_3d</a>()</code> <br><code><a href="layer_batch_normalization.html">layer_batch_normalization</a>()</code> <br><code><a href="layer_bidirectional.html">layer_bidirectional</a>()</code> <br><code><a href="layer_category_encoding.html">layer_category_encoding</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_concatenate.html">layer_concatenate</a>()</code> <br><code><a href="layer_conv_1d.html">layer_conv_1d</a>()</code> <br><code><a href="layer_conv_1d_transpose.html">layer_conv_1d_transpose</a>()</code> <br><code><a href="layer_conv_2d.html">layer_conv_2d</a>()</code> <br><code><a href="layer_conv_2d_transpose.html">layer_conv_2d_transpose</a>()</code> <br><code><a href="layer_conv_3d.html">layer_conv_3d</a>()</code> <br><code><a href="layer_conv_3d_transpose.html">layer_conv_3d_transpose</a>()</code> <br><code><a href="layer_conv_lstm_1d.html">layer_conv_lstm_1d</a>()</code> <br><code><a href="layer_conv_lstm_2d.html">layer_conv_lstm_2d</a>()</code> <br><code><a href="layer_conv_lstm_3d.html">layer_conv_lstm_3d</a>()</code> <br><code><a href="layer_cropping_1d.html">layer_cropping_1d</a>()</code> <br><code><a href="layer_cropping_2d.html">layer_cropping_2d</a>()</code> <br><code><a href="layer_cropping_3d.html">layer_cropping_3d</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_dense.html">layer_dense</a>()</code> <br><code><a href="layer_depthwise_conv_1d.html">layer_depthwise_conv_1d</a>()</code> <br><code><a href="layer_depthwise_conv_2d.html">layer_depthwise_conv_2d</a>()</code> <br><code><a href="layer_discretization.html">layer_discretization</a>()</code> <br><code><a href="layer_dot.html">layer_dot</a>()</code> <br><code><a href="layer_dropout.html">layer_dropout</a>()</code> <br><code><a href="layer_einsum_dense.html">layer_einsum_dense</a>()</code> <br><code><a href="layer_embedding.html">layer_embedding</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_feature_space.html">layer_feature_space</a>()</code> <br><code><a href="layer_flatten.html">layer_flatten</a>()</code> <br><code><a href="layer_flax_module_wrapper.html">layer_flax_module_wrapper</a>()</code> <br><code><a href="layer_gaussian_dropout.html">layer_gaussian_dropout</a>()</code> <br><code><a href="layer_gaussian_noise.html">layer_gaussian_noise</a>()</code> <br><code><a href="layer_global_average_pooling_1d.html">layer_global_average_pooling_1d</a>()</code> <br><code><a href="layer_global_average_pooling_2d.html">layer_global_average_pooling_2d</a>()</code> <br><code><a href="layer_global_average_pooling_3d.html">layer_global_average_pooling_3d</a>()</code> <br><code><a href="layer_global_max_pooling_1d.html">layer_global_max_pooling_1d</a>()</code> <br><code><a href="layer_global_max_pooling_2d.html">layer_global_max_pooling_2d</a>()</code> <br><code><a href="layer_global_max_pooling_3d.html">layer_global_max_pooling_3d</a>()</code> <br><code><a href="layer_group_normalization.html">layer_group_normalization</a>()</code> <br><code><a href="layer_group_query_attention.html">layer_group_query_attention</a>()</code> <br><code><a href="layer_hashed_crossing.html">layer_hashed_crossing</a>()</code> <br><code><a href="layer_hashing.html">layer_hashing</a>()</code> <br><code><a href="layer_identity.html">layer_identity</a>()</code> <br><code><a href="layer_integer_lookup.html">layer_integer_lookup</a>()</code> <br><code><a href="layer_jax_model_wrapper.html">layer_jax_model_wrapper</a>()</code> <br><code><a href="layer_lambda.html">layer_lambda</a>()</code> <br><code><a href="layer_layer_normalization.html">layer_layer_normalization</a>()</code> <br><code><a href="layer_lstm.html">layer_lstm</a>()</code> <br><code><a href="layer_masking.html">layer_masking</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_max_pooling_1d.html">layer_max_pooling_1d</a>()</code> <br><code><a href="layer_max_pooling_2d.html">layer_max_pooling_2d</a>()</code> <br><code><a href="layer_max_pooling_3d.html">layer_max_pooling_3d</a>()</code> <br><code><a href="layer_maximum.html">layer_maximum</a>()</code> <br><code><a href="layer_mel_spectrogram.html">layer_mel_spectrogram</a>()</code> <br><code><a href="layer_minimum.html">layer_minimum</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_multi_head_attention.html">layer_multi_head_attention</a>()</code> <br><code><a href="layer_multiply.html">layer_multiply</a>()</code> <br><code><a href="layer_normalization.html">layer_normalization</a>()</code> <br><code><a href="layer_permute.html">layer_permute</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_brightness.html">layer_random_brightness</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_contrast.html">layer_random_contrast</a>()</code> <br><code><a href="layer_random_crop.html">layer_random_crop</a>()</code> <br><code><a href="layer_random_elastic_transform.html">layer_random_elastic_transform</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_flip.html">layer_random_flip</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_rotation.html">layer_random_rotation</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_random_translation.html">layer_random_translation</a>()</code> <br><code><a href="layer_random_zoom.html">layer_random_zoom</a>()</code> <br><code><a href="layer_repeat_vector.html">layer_repeat_vector</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_reshape.html">layer_reshape</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_rms_normalization.html">layer_rms_normalization</a>()</code> <br><code><a href="layer_rnn.html">layer_rnn</a>()</code> <br><code><a href="layer_separable_conv_1d.html">layer_separable_conv_1d</a>()</code> <br><code><a href="layer_separable_conv_2d.html">layer_separable_conv_2d</a>()</code> <br><code><a href="layer_simple_rnn.html">layer_simple_rnn</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br><code><a href="layer_spatial_dropout_1d.html">layer_spatial_dropout_1d</a>()</code> <br><code><a href="layer_spatial_dropout_2d.html">layer_spatial_dropout_2d</a>()</code> <br><code><a href="layer_spatial_dropout_3d.html">layer_spatial_dropout_3d</a>()</code> <br><code><a href="layer_spectral_normalization.html">layer_spectral_normalization</a>()</code> <br><code><a href="layer_stft_spectrogram.html">layer_stft_spectrogram</a>()</code> <br><code><a href="layer_string_lookup.html">layer_string_lookup</a>()</code> <br><code><a href="layer_subtract.html">layer_subtract</a>()</code> <br><code><a href="layer_text_vectorization.html">layer_text_vectorization</a>()</code> <br><code><a href="layer_tfsm.html">layer_tfsm</a>()</code> <br><code><a href="layer_time_distributed.html">layer_time_distributed</a>()</code> <br><code><a href="layer_torch_module_wrapper.html">layer_torch_module_wrapper</a>()</code> <br><code><a href="layer_unit_normalization.html">layer_unit_normalization</a>()</code> <br><code><a href="layer_upsampling_1d.html">layer_upsampling_1d</a>()</code> <br><code><a href="layer_upsampling_2d.html">layer_upsampling_2d</a>()</code> <br><code><a href="layer_upsampling_3d.html">layer_upsampling_3d</a>()</code> <br><code><a href="layer_zero_padding_1d.html">layer_zero_padding_1d</a>()</code> <br><code><a href="layer_zero_padding_2d.html">layer_zero_padding_2d</a>()</code> <br><code><a href="layer_zero_padding_3d.html">layer_zero_padding_3d</a>()</code> <br><code><a href="rnn_cell_gru.html">rnn_cell_gru</a>()</code> <br><code><a href="rnn_cell_lstm.html">rnn_cell_lstm</a>()</code> <br><code><a href="rnn_cell_simple.html">rnn_cell_simple</a>()</code> <br><code><a href="rnn_cells_stack.html">rnn_cells_stack</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -403,7 +403,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/layer_hashed_crossing.html b/docs/dev/reference/layer_hashed_crossing.html
index db2f5c9ea..b579f9d71 100644
--- a/docs/dev/reference/layer_hashed_crossing.html
+++ b/docs/dev/reference/layer_hashed_crossing.html
@@ -36,7 +36,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -185,8 +185,8 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
     <div class="dont-index">
 <ul><li><p><a href="https://keras.io/api/layers/preprocessing_layers/categorical/hashed_crossing#hashedcrossing-class" class="external-link">https://keras.io/api/layers/preprocessing_layers/categorical/hashed_crossing#hashedcrossing-class</a></p></li>
 </ul><p>Other categorical features preprocessing layers: <br><code><a href="layer_category_encoding.html">layer_category_encoding</a>()</code> <br><code><a href="layer_hashing.html">layer_hashing</a>()</code> <br><code><a href="layer_integer_lookup.html">layer_integer_lookup</a>()</code> <br><code><a href="layer_string_lookup.html">layer_string_lookup</a>()</code> <br></p>
-<p>Other preprocessing layers: <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_category_encoding.html">layer_category_encoding</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_discretization.html">layer_discretization</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_feature_space.html">layer_feature_space</a>()</code> <br><code><a href="layer_hashing.html">layer_hashing</a>()</code> <br><code><a href="layer_integer_lookup.html">layer_integer_lookup</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_mel_spectrogram.html">layer_mel_spectrogram</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_normalization.html">layer_normalization</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_brightness.html">layer_random_brightness</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_contrast.html">layer_random_contrast</a>()</code> <br><code><a href="layer_random_crop.html">layer_random_crop</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_flip.html">layer_random_flip</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_rotation.html">layer_random_rotation</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_random_translation.html">layer_random_translation</a>()</code> <br><code><a href="layer_random_zoom.html">layer_random_zoom</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br><code><a href="layer_stft_spectrogram.html">layer_stft_spectrogram</a>()</code> <br><code><a href="layer_string_lookup.html">layer_string_lookup</a>()</code> <br><code><a href="layer_text_vectorization.html">layer_text_vectorization</a>()</code> <br></p>
-<p>Other layers: <br><code><a href="Layer.html">Layer</a>()</code> <br><code><a href="layer_activation.html">layer_activation</a>()</code> <br><code><a href="layer_activation_elu.html">layer_activation_elu</a>()</code> <br><code><a href="layer_activation_leaky_relu.html">layer_activation_leaky_relu</a>()</code> <br><code><a href="layer_activation_parametric_relu.html">layer_activation_parametric_relu</a>()</code> <br><code><a href="layer_activation_relu.html">layer_activation_relu</a>()</code> <br><code><a href="layer_activation_softmax.html">layer_activation_softmax</a>()</code> <br><code><a href="layer_activity_regularization.html">layer_activity_regularization</a>()</code> <br><code><a href="layer_add.html">layer_add</a>()</code> <br><code><a href="layer_additive_attention.html">layer_additive_attention</a>()</code> <br><code><a href="layer_alpha_dropout.html">layer_alpha_dropout</a>()</code> <br><code><a href="layer_attention.html">layer_attention</a>()</code> <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_average.html">layer_average</a>()</code> <br><code><a href="layer_average_pooling_1d.html">layer_average_pooling_1d</a>()</code> <br><code><a href="layer_average_pooling_2d.html">layer_average_pooling_2d</a>()</code> <br><code><a href="layer_average_pooling_3d.html">layer_average_pooling_3d</a>()</code> <br><code><a href="layer_batch_normalization.html">layer_batch_normalization</a>()</code> <br><code><a href="layer_bidirectional.html">layer_bidirectional</a>()</code> <br><code><a href="layer_category_encoding.html">layer_category_encoding</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_concatenate.html">layer_concatenate</a>()</code> <br><code><a href="layer_conv_1d.html">layer_conv_1d</a>()</code> <br><code><a href="layer_conv_1d_transpose.html">layer_conv_1d_transpose</a>()</code> <br><code><a href="layer_conv_2d.html">layer_conv_2d</a>()</code> <br><code><a href="layer_conv_2d_transpose.html">layer_conv_2d_transpose</a>()</code> <br><code><a href="layer_conv_3d.html">layer_conv_3d</a>()</code> <br><code><a href="layer_conv_3d_transpose.html">layer_conv_3d_transpose</a>()</code> <br><code><a href="layer_conv_lstm_1d.html">layer_conv_lstm_1d</a>()</code> <br><code><a href="layer_conv_lstm_2d.html">layer_conv_lstm_2d</a>()</code> <br><code><a href="layer_conv_lstm_3d.html">layer_conv_lstm_3d</a>()</code> <br><code><a href="layer_cropping_1d.html">layer_cropping_1d</a>()</code> <br><code><a href="layer_cropping_2d.html">layer_cropping_2d</a>()</code> <br><code><a href="layer_cropping_3d.html">layer_cropping_3d</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_dense.html">layer_dense</a>()</code> <br><code><a href="layer_depthwise_conv_1d.html">layer_depthwise_conv_1d</a>()</code> <br><code><a href="layer_depthwise_conv_2d.html">layer_depthwise_conv_2d</a>()</code> <br><code><a href="layer_discretization.html">layer_discretization</a>()</code> <br><code><a href="layer_dot.html">layer_dot</a>()</code> <br><code><a href="layer_dropout.html">layer_dropout</a>()</code> <br><code><a href="layer_einsum_dense.html">layer_einsum_dense</a>()</code> <br><code><a href="layer_embedding.html">layer_embedding</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_feature_space.html">layer_feature_space</a>()</code> <br><code><a href="layer_flatten.html">layer_flatten</a>()</code> <br><code><a href="layer_flax_module_wrapper.html">layer_flax_module_wrapper</a>()</code> <br><code><a href="layer_gaussian_dropout.html">layer_gaussian_dropout</a>()</code> <br><code><a href="layer_gaussian_noise.html">layer_gaussian_noise</a>()</code> <br><code><a href="layer_global_average_pooling_1d.html">layer_global_average_pooling_1d</a>()</code> <br><code><a href="layer_global_average_pooling_2d.html">layer_global_average_pooling_2d</a>()</code> <br><code><a href="layer_global_average_pooling_3d.html">layer_global_average_pooling_3d</a>()</code> <br><code><a href="layer_global_max_pooling_1d.html">layer_global_max_pooling_1d</a>()</code> <br><code><a href="layer_global_max_pooling_2d.html">layer_global_max_pooling_2d</a>()</code> <br><code><a href="layer_global_max_pooling_3d.html">layer_global_max_pooling_3d</a>()</code> <br><code><a href="layer_group_normalization.html">layer_group_normalization</a>()</code> <br><code><a href="layer_group_query_attention.html">layer_group_query_attention</a>()</code> <br><code><a href="layer_gru.html">layer_gru</a>()</code> <br><code><a href="layer_hashing.html">layer_hashing</a>()</code> <br><code><a href="layer_identity.html">layer_identity</a>()</code> <br><code><a href="layer_integer_lookup.html">layer_integer_lookup</a>()</code> <br><code><a href="layer_jax_model_wrapper.html">layer_jax_model_wrapper</a>()</code> <br><code><a href="layer_lambda.html">layer_lambda</a>()</code> <br><code><a href="layer_layer_normalization.html">layer_layer_normalization</a>()</code> <br><code><a href="layer_lstm.html">layer_lstm</a>()</code> <br><code><a href="layer_masking.html">layer_masking</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_max_pooling_1d.html">layer_max_pooling_1d</a>()</code> <br><code><a href="layer_max_pooling_2d.html">layer_max_pooling_2d</a>()</code> <br><code><a href="layer_max_pooling_3d.html">layer_max_pooling_3d</a>()</code> <br><code><a href="layer_maximum.html">layer_maximum</a>()</code> <br><code><a href="layer_mel_spectrogram.html">layer_mel_spectrogram</a>()</code> <br><code><a href="layer_minimum.html">layer_minimum</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_multi_head_attention.html">layer_multi_head_attention</a>()</code> <br><code><a href="layer_multiply.html">layer_multiply</a>()</code> <br><code><a href="layer_normalization.html">layer_normalization</a>()</code> <br><code><a href="layer_permute.html">layer_permute</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_brightness.html">layer_random_brightness</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_contrast.html">layer_random_contrast</a>()</code> <br><code><a href="layer_random_crop.html">layer_random_crop</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_flip.html">layer_random_flip</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_rotation.html">layer_random_rotation</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_random_translation.html">layer_random_translation</a>()</code> <br><code><a href="layer_random_zoom.html">layer_random_zoom</a>()</code> <br><code><a href="layer_repeat_vector.html">layer_repeat_vector</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_reshape.html">layer_reshape</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_rms_normalization.html">layer_rms_normalization</a>()</code> <br><code><a href="layer_rnn.html">layer_rnn</a>()</code> <br><code><a href="layer_separable_conv_1d.html">layer_separable_conv_1d</a>()</code> <br><code><a href="layer_separable_conv_2d.html">layer_separable_conv_2d</a>()</code> <br><code><a href="layer_simple_rnn.html">layer_simple_rnn</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br><code><a href="layer_spatial_dropout_1d.html">layer_spatial_dropout_1d</a>()</code> <br><code><a href="layer_spatial_dropout_2d.html">layer_spatial_dropout_2d</a>()</code> <br><code><a href="layer_spatial_dropout_3d.html">layer_spatial_dropout_3d</a>()</code> <br><code><a href="layer_spectral_normalization.html">layer_spectral_normalization</a>()</code> <br><code><a href="layer_stft_spectrogram.html">layer_stft_spectrogram</a>()</code> <br><code><a href="layer_string_lookup.html">layer_string_lookup</a>()</code> <br><code><a href="layer_subtract.html">layer_subtract</a>()</code> <br><code><a href="layer_text_vectorization.html">layer_text_vectorization</a>()</code> <br><code><a href="layer_tfsm.html">layer_tfsm</a>()</code> <br><code><a href="layer_time_distributed.html">layer_time_distributed</a>()</code> <br><code><a href="layer_torch_module_wrapper.html">layer_torch_module_wrapper</a>()</code> <br><code><a href="layer_unit_normalization.html">layer_unit_normalization</a>()</code> <br><code><a href="layer_upsampling_1d.html">layer_upsampling_1d</a>()</code> <br><code><a href="layer_upsampling_2d.html">layer_upsampling_2d</a>()</code> <br><code><a href="layer_upsampling_3d.html">layer_upsampling_3d</a>()</code> <br><code><a href="layer_zero_padding_1d.html">layer_zero_padding_1d</a>()</code> <br><code><a href="layer_zero_padding_2d.html">layer_zero_padding_2d</a>()</code> <br><code><a href="layer_zero_padding_3d.html">layer_zero_padding_3d</a>()</code> <br><code><a href="rnn_cell_gru.html">rnn_cell_gru</a>()</code> <br><code><a href="rnn_cell_lstm.html">rnn_cell_lstm</a>()</code> <br><code><a href="rnn_cell_simple.html">rnn_cell_simple</a>()</code> <br><code><a href="rnn_cells_stack.html">rnn_cells_stack</a>()</code> <br></p></div>
+<p>Other preprocessing layers: <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_category_encoding.html">layer_category_encoding</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_discretization.html">layer_discretization</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_feature_space.html">layer_feature_space</a>()</code> <br><code><a href="layer_hashing.html">layer_hashing</a>()</code> <br><code><a href="layer_integer_lookup.html">layer_integer_lookup</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_mel_spectrogram.html">layer_mel_spectrogram</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_normalization.html">layer_normalization</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_brightness.html">layer_random_brightness</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_contrast.html">layer_random_contrast</a>()</code> <br><code><a href="layer_random_crop.html">layer_random_crop</a>()</code> <br><code><a href="layer_random_elastic_transform.html">layer_random_elastic_transform</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_flip.html">layer_random_flip</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_rotation.html">layer_random_rotation</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_random_translation.html">layer_random_translation</a>()</code> <br><code><a href="layer_random_zoom.html">layer_random_zoom</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br><code><a href="layer_stft_spectrogram.html">layer_stft_spectrogram</a>()</code> <br><code><a href="layer_string_lookup.html">layer_string_lookup</a>()</code> <br><code><a href="layer_text_vectorization.html">layer_text_vectorization</a>()</code> <br></p>
+<p>Other layers: <br><code><a href="Layer.html">Layer</a>()</code> <br><code><a href="layer_activation.html">layer_activation</a>()</code> <br><code><a href="layer_activation_elu.html">layer_activation_elu</a>()</code> <br><code><a href="layer_activation_leaky_relu.html">layer_activation_leaky_relu</a>()</code> <br><code><a href="layer_activation_parametric_relu.html">layer_activation_parametric_relu</a>()</code> <br><code><a href="layer_activation_relu.html">layer_activation_relu</a>()</code> <br><code><a href="layer_activation_softmax.html">layer_activation_softmax</a>()</code> <br><code><a href="layer_activity_regularization.html">layer_activity_regularization</a>()</code> <br><code><a href="layer_add.html">layer_add</a>()</code> <br><code><a href="layer_additive_attention.html">layer_additive_attention</a>()</code> <br><code><a href="layer_alpha_dropout.html">layer_alpha_dropout</a>()</code> <br><code><a href="layer_attention.html">layer_attention</a>()</code> <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_average.html">layer_average</a>()</code> <br><code><a href="layer_average_pooling_1d.html">layer_average_pooling_1d</a>()</code> <br><code><a href="layer_average_pooling_2d.html">layer_average_pooling_2d</a>()</code> <br><code><a href="layer_average_pooling_3d.html">layer_average_pooling_3d</a>()</code> <br><code><a href="layer_batch_normalization.html">layer_batch_normalization</a>()</code> <br><code><a href="layer_bidirectional.html">layer_bidirectional</a>()</code> <br><code><a href="layer_category_encoding.html">layer_category_encoding</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_concatenate.html">layer_concatenate</a>()</code> <br><code><a href="layer_conv_1d.html">layer_conv_1d</a>()</code> <br><code><a href="layer_conv_1d_transpose.html">layer_conv_1d_transpose</a>()</code> <br><code><a href="layer_conv_2d.html">layer_conv_2d</a>()</code> <br><code><a href="layer_conv_2d_transpose.html">layer_conv_2d_transpose</a>()</code> <br><code><a href="layer_conv_3d.html">layer_conv_3d</a>()</code> <br><code><a href="layer_conv_3d_transpose.html">layer_conv_3d_transpose</a>()</code> <br><code><a href="layer_conv_lstm_1d.html">layer_conv_lstm_1d</a>()</code> <br><code><a href="layer_conv_lstm_2d.html">layer_conv_lstm_2d</a>()</code> <br><code><a href="layer_conv_lstm_3d.html">layer_conv_lstm_3d</a>()</code> <br><code><a href="layer_cropping_1d.html">layer_cropping_1d</a>()</code> <br><code><a href="layer_cropping_2d.html">layer_cropping_2d</a>()</code> <br><code><a href="layer_cropping_3d.html">layer_cropping_3d</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_dense.html">layer_dense</a>()</code> <br><code><a href="layer_depthwise_conv_1d.html">layer_depthwise_conv_1d</a>()</code> <br><code><a href="layer_depthwise_conv_2d.html">layer_depthwise_conv_2d</a>()</code> <br><code><a href="layer_discretization.html">layer_discretization</a>()</code> <br><code><a href="layer_dot.html">layer_dot</a>()</code> <br><code><a href="layer_dropout.html">layer_dropout</a>()</code> <br><code><a href="layer_einsum_dense.html">layer_einsum_dense</a>()</code> <br><code><a href="layer_embedding.html">layer_embedding</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_feature_space.html">layer_feature_space</a>()</code> <br><code><a href="layer_flatten.html">layer_flatten</a>()</code> <br><code><a href="layer_flax_module_wrapper.html">layer_flax_module_wrapper</a>()</code> <br><code><a href="layer_gaussian_dropout.html">layer_gaussian_dropout</a>()</code> <br><code><a href="layer_gaussian_noise.html">layer_gaussian_noise</a>()</code> <br><code><a href="layer_global_average_pooling_1d.html">layer_global_average_pooling_1d</a>()</code> <br><code><a href="layer_global_average_pooling_2d.html">layer_global_average_pooling_2d</a>()</code> <br><code><a href="layer_global_average_pooling_3d.html">layer_global_average_pooling_3d</a>()</code> <br><code><a href="layer_global_max_pooling_1d.html">layer_global_max_pooling_1d</a>()</code> <br><code><a href="layer_global_max_pooling_2d.html">layer_global_max_pooling_2d</a>()</code> <br><code><a href="layer_global_max_pooling_3d.html">layer_global_max_pooling_3d</a>()</code> <br><code><a href="layer_group_normalization.html">layer_group_normalization</a>()</code> <br><code><a href="layer_group_query_attention.html">layer_group_query_attention</a>()</code> <br><code><a href="layer_gru.html">layer_gru</a>()</code> <br><code><a href="layer_hashing.html">layer_hashing</a>()</code> <br><code><a href="layer_identity.html">layer_identity</a>()</code> <br><code><a href="layer_integer_lookup.html">layer_integer_lookup</a>()</code> <br><code><a href="layer_jax_model_wrapper.html">layer_jax_model_wrapper</a>()</code> <br><code><a href="layer_lambda.html">layer_lambda</a>()</code> <br><code><a href="layer_layer_normalization.html">layer_layer_normalization</a>()</code> <br><code><a href="layer_lstm.html">layer_lstm</a>()</code> <br><code><a href="layer_masking.html">layer_masking</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_max_pooling_1d.html">layer_max_pooling_1d</a>()</code> <br><code><a href="layer_max_pooling_2d.html">layer_max_pooling_2d</a>()</code> <br><code><a href="layer_max_pooling_3d.html">layer_max_pooling_3d</a>()</code> <br><code><a href="layer_maximum.html">layer_maximum</a>()</code> <br><code><a href="layer_mel_spectrogram.html">layer_mel_spectrogram</a>()</code> <br><code><a href="layer_minimum.html">layer_minimum</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_multi_head_attention.html">layer_multi_head_attention</a>()</code> <br><code><a href="layer_multiply.html">layer_multiply</a>()</code> <br><code><a href="layer_normalization.html">layer_normalization</a>()</code> <br><code><a href="layer_permute.html">layer_permute</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_brightness.html">layer_random_brightness</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_contrast.html">layer_random_contrast</a>()</code> <br><code><a href="layer_random_crop.html">layer_random_crop</a>()</code> <br><code><a href="layer_random_elastic_transform.html">layer_random_elastic_transform</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_flip.html">layer_random_flip</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_rotation.html">layer_random_rotation</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_random_translation.html">layer_random_translation</a>()</code> <br><code><a href="layer_random_zoom.html">layer_random_zoom</a>()</code> <br><code><a href="layer_repeat_vector.html">layer_repeat_vector</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_reshape.html">layer_reshape</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_rms_normalization.html">layer_rms_normalization</a>()</code> <br><code><a href="layer_rnn.html">layer_rnn</a>()</code> <br><code><a href="layer_separable_conv_1d.html">layer_separable_conv_1d</a>()</code> <br><code><a href="layer_separable_conv_2d.html">layer_separable_conv_2d</a>()</code> <br><code><a href="layer_simple_rnn.html">layer_simple_rnn</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br><code><a href="layer_spatial_dropout_1d.html">layer_spatial_dropout_1d</a>()</code> <br><code><a href="layer_spatial_dropout_2d.html">layer_spatial_dropout_2d</a>()</code> <br><code><a href="layer_spatial_dropout_3d.html">layer_spatial_dropout_3d</a>()</code> <br><code><a href="layer_spectral_normalization.html">layer_spectral_normalization</a>()</code> <br><code><a href="layer_stft_spectrogram.html">layer_stft_spectrogram</a>()</code> <br><code><a href="layer_string_lookup.html">layer_string_lookup</a>()</code> <br><code><a href="layer_subtract.html">layer_subtract</a>()</code> <br><code><a href="layer_text_vectorization.html">layer_text_vectorization</a>()</code> <br><code><a href="layer_tfsm.html">layer_tfsm</a>()</code> <br><code><a href="layer_time_distributed.html">layer_time_distributed</a>()</code> <br><code><a href="layer_torch_module_wrapper.html">layer_torch_module_wrapper</a>()</code> <br><code><a href="layer_unit_normalization.html">layer_unit_normalization</a>()</code> <br><code><a href="layer_upsampling_1d.html">layer_upsampling_1d</a>()</code> <br><code><a href="layer_upsampling_2d.html">layer_upsampling_2d</a>()</code> <br><code><a href="layer_upsampling_3d.html">layer_upsampling_3d</a>()</code> <br><code><a href="layer_zero_padding_1d.html">layer_zero_padding_1d</a>()</code> <br><code><a href="layer_zero_padding_2d.html">layer_zero_padding_2d</a>()</code> <br><code><a href="layer_zero_padding_3d.html">layer_zero_padding_3d</a>()</code> <br><code><a href="rnn_cell_gru.html">rnn_cell_gru</a>()</code> <br><code><a href="rnn_cell_lstm.html">rnn_cell_lstm</a>()</code> <br><code><a href="rnn_cell_simple.html">rnn_cell_simple</a>()</code> <br><code><a href="rnn_cells_stack.html">rnn_cells_stack</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -198,7 +198,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/layer_hashing.html b/docs/dev/reference/layer_hashing.html
index c44236da6..204df400f 100644
--- a/docs/dev/reference/layer_hashing.html
+++ b/docs/dev/reference/layer_hashing.html
@@ -152,7 +152,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -371,8 +371,8 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
     <div class="dont-index">
 <ul><li><p><a href="https://keras.io/api/layers/preprocessing_layers/categorical/hashing#hashing-class" class="external-link">https://keras.io/api/layers/preprocessing_layers/categorical/hashing#hashing-class</a></p></li>
 </ul><p>Other categorical features preprocessing layers: <br><code><a href="layer_category_encoding.html">layer_category_encoding</a>()</code> <br><code><a href="layer_hashed_crossing.html">layer_hashed_crossing</a>()</code> <br><code><a href="layer_integer_lookup.html">layer_integer_lookup</a>()</code> <br><code><a href="layer_string_lookup.html">layer_string_lookup</a>()</code> <br></p>
-<p>Other preprocessing layers: <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_category_encoding.html">layer_category_encoding</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_discretization.html">layer_discretization</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_feature_space.html">layer_feature_space</a>()</code> <br><code><a href="layer_hashed_crossing.html">layer_hashed_crossing</a>()</code> <br><code><a href="layer_integer_lookup.html">layer_integer_lookup</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_mel_spectrogram.html">layer_mel_spectrogram</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_normalization.html">layer_normalization</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_brightness.html">layer_random_brightness</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_contrast.html">layer_random_contrast</a>()</code> <br><code><a href="layer_random_crop.html">layer_random_crop</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_flip.html">layer_random_flip</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_rotation.html">layer_random_rotation</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_random_translation.html">layer_random_translation</a>()</code> <br><code><a href="layer_random_zoom.html">layer_random_zoom</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br><code><a href="layer_stft_spectrogram.html">layer_stft_spectrogram</a>()</code> <br><code><a href="layer_string_lookup.html">layer_string_lookup</a>()</code> <br><code><a href="layer_text_vectorization.html">layer_text_vectorization</a>()</code> <br></p>
-<p>Other layers: <br><code><a href="Layer.html">Layer</a>()</code> <br><code><a href="layer_activation.html">layer_activation</a>()</code> <br><code><a href="layer_activation_elu.html">layer_activation_elu</a>()</code> <br><code><a href="layer_activation_leaky_relu.html">layer_activation_leaky_relu</a>()</code> <br><code><a href="layer_activation_parametric_relu.html">layer_activation_parametric_relu</a>()</code> <br><code><a href="layer_activation_relu.html">layer_activation_relu</a>()</code> <br><code><a href="layer_activation_softmax.html">layer_activation_softmax</a>()</code> <br><code><a href="layer_activity_regularization.html">layer_activity_regularization</a>()</code> <br><code><a href="layer_add.html">layer_add</a>()</code> <br><code><a href="layer_additive_attention.html">layer_additive_attention</a>()</code> <br><code><a href="layer_alpha_dropout.html">layer_alpha_dropout</a>()</code> <br><code><a href="layer_attention.html">layer_attention</a>()</code> <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_average.html">layer_average</a>()</code> <br><code><a href="layer_average_pooling_1d.html">layer_average_pooling_1d</a>()</code> <br><code><a href="layer_average_pooling_2d.html">layer_average_pooling_2d</a>()</code> <br><code><a href="layer_average_pooling_3d.html">layer_average_pooling_3d</a>()</code> <br><code><a href="layer_batch_normalization.html">layer_batch_normalization</a>()</code> <br><code><a href="layer_bidirectional.html">layer_bidirectional</a>()</code> <br><code><a href="layer_category_encoding.html">layer_category_encoding</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_concatenate.html">layer_concatenate</a>()</code> <br><code><a href="layer_conv_1d.html">layer_conv_1d</a>()</code> <br><code><a href="layer_conv_1d_transpose.html">layer_conv_1d_transpose</a>()</code> <br><code><a href="layer_conv_2d.html">layer_conv_2d</a>()</code> <br><code><a href="layer_conv_2d_transpose.html">layer_conv_2d_transpose</a>()</code> <br><code><a href="layer_conv_3d.html">layer_conv_3d</a>()</code> <br><code><a href="layer_conv_3d_transpose.html">layer_conv_3d_transpose</a>()</code> <br><code><a href="layer_conv_lstm_1d.html">layer_conv_lstm_1d</a>()</code> <br><code><a href="layer_conv_lstm_2d.html">layer_conv_lstm_2d</a>()</code> <br><code><a href="layer_conv_lstm_3d.html">layer_conv_lstm_3d</a>()</code> <br><code><a href="layer_cropping_1d.html">layer_cropping_1d</a>()</code> <br><code><a href="layer_cropping_2d.html">layer_cropping_2d</a>()</code> <br><code><a href="layer_cropping_3d.html">layer_cropping_3d</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_dense.html">layer_dense</a>()</code> <br><code><a href="layer_depthwise_conv_1d.html">layer_depthwise_conv_1d</a>()</code> <br><code><a href="layer_depthwise_conv_2d.html">layer_depthwise_conv_2d</a>()</code> <br><code><a href="layer_discretization.html">layer_discretization</a>()</code> <br><code><a href="layer_dot.html">layer_dot</a>()</code> <br><code><a href="layer_dropout.html">layer_dropout</a>()</code> <br><code><a href="layer_einsum_dense.html">layer_einsum_dense</a>()</code> <br><code><a href="layer_embedding.html">layer_embedding</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_feature_space.html">layer_feature_space</a>()</code> <br><code><a href="layer_flatten.html">layer_flatten</a>()</code> <br><code><a href="layer_flax_module_wrapper.html">layer_flax_module_wrapper</a>()</code> <br><code><a href="layer_gaussian_dropout.html">layer_gaussian_dropout</a>()</code> <br><code><a href="layer_gaussian_noise.html">layer_gaussian_noise</a>()</code> <br><code><a href="layer_global_average_pooling_1d.html">layer_global_average_pooling_1d</a>()</code> <br><code><a href="layer_global_average_pooling_2d.html">layer_global_average_pooling_2d</a>()</code> <br><code><a href="layer_global_average_pooling_3d.html">layer_global_average_pooling_3d</a>()</code> <br><code><a href="layer_global_max_pooling_1d.html">layer_global_max_pooling_1d</a>()</code> <br><code><a href="layer_global_max_pooling_2d.html">layer_global_max_pooling_2d</a>()</code> <br><code><a href="layer_global_max_pooling_3d.html">layer_global_max_pooling_3d</a>()</code> <br><code><a href="layer_group_normalization.html">layer_group_normalization</a>()</code> <br><code><a href="layer_group_query_attention.html">layer_group_query_attention</a>()</code> <br><code><a href="layer_gru.html">layer_gru</a>()</code> <br><code><a href="layer_hashed_crossing.html">layer_hashed_crossing</a>()</code> <br><code><a href="layer_identity.html">layer_identity</a>()</code> <br><code><a href="layer_integer_lookup.html">layer_integer_lookup</a>()</code> <br><code><a href="layer_jax_model_wrapper.html">layer_jax_model_wrapper</a>()</code> <br><code><a href="layer_lambda.html">layer_lambda</a>()</code> <br><code><a href="layer_layer_normalization.html">layer_layer_normalization</a>()</code> <br><code><a href="layer_lstm.html">layer_lstm</a>()</code> <br><code><a href="layer_masking.html">layer_masking</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_max_pooling_1d.html">layer_max_pooling_1d</a>()</code> <br><code><a href="layer_max_pooling_2d.html">layer_max_pooling_2d</a>()</code> <br><code><a href="layer_max_pooling_3d.html">layer_max_pooling_3d</a>()</code> <br><code><a href="layer_maximum.html">layer_maximum</a>()</code> <br><code><a href="layer_mel_spectrogram.html">layer_mel_spectrogram</a>()</code> <br><code><a href="layer_minimum.html">layer_minimum</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_multi_head_attention.html">layer_multi_head_attention</a>()</code> <br><code><a href="layer_multiply.html">layer_multiply</a>()</code> <br><code><a href="layer_normalization.html">layer_normalization</a>()</code> <br><code><a href="layer_permute.html">layer_permute</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_brightness.html">layer_random_brightness</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_contrast.html">layer_random_contrast</a>()</code> <br><code><a href="layer_random_crop.html">layer_random_crop</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_flip.html">layer_random_flip</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_rotation.html">layer_random_rotation</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_random_translation.html">layer_random_translation</a>()</code> <br><code><a href="layer_random_zoom.html">layer_random_zoom</a>()</code> <br><code><a href="layer_repeat_vector.html">layer_repeat_vector</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_reshape.html">layer_reshape</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_rms_normalization.html">layer_rms_normalization</a>()</code> <br><code><a href="layer_rnn.html">layer_rnn</a>()</code> <br><code><a href="layer_separable_conv_1d.html">layer_separable_conv_1d</a>()</code> <br><code><a href="layer_separable_conv_2d.html">layer_separable_conv_2d</a>()</code> <br><code><a href="layer_simple_rnn.html">layer_simple_rnn</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br><code><a href="layer_spatial_dropout_1d.html">layer_spatial_dropout_1d</a>()</code> <br><code><a href="layer_spatial_dropout_2d.html">layer_spatial_dropout_2d</a>()</code> <br><code><a href="layer_spatial_dropout_3d.html">layer_spatial_dropout_3d</a>()</code> <br><code><a href="layer_spectral_normalization.html">layer_spectral_normalization</a>()</code> <br><code><a href="layer_stft_spectrogram.html">layer_stft_spectrogram</a>()</code> <br><code><a href="layer_string_lookup.html">layer_string_lookup</a>()</code> <br><code><a href="layer_subtract.html">layer_subtract</a>()</code> <br><code><a href="layer_text_vectorization.html">layer_text_vectorization</a>()</code> <br><code><a href="layer_tfsm.html">layer_tfsm</a>()</code> <br><code><a href="layer_time_distributed.html">layer_time_distributed</a>()</code> <br><code><a href="layer_torch_module_wrapper.html">layer_torch_module_wrapper</a>()</code> <br><code><a href="layer_unit_normalization.html">layer_unit_normalization</a>()</code> <br><code><a href="layer_upsampling_1d.html">layer_upsampling_1d</a>()</code> <br><code><a href="layer_upsampling_2d.html">layer_upsampling_2d</a>()</code> <br><code><a href="layer_upsampling_3d.html">layer_upsampling_3d</a>()</code> <br><code><a href="layer_zero_padding_1d.html">layer_zero_padding_1d</a>()</code> <br><code><a href="layer_zero_padding_2d.html">layer_zero_padding_2d</a>()</code> <br><code><a href="layer_zero_padding_3d.html">layer_zero_padding_3d</a>()</code> <br><code><a href="rnn_cell_gru.html">rnn_cell_gru</a>()</code> <br><code><a href="rnn_cell_lstm.html">rnn_cell_lstm</a>()</code> <br><code><a href="rnn_cell_simple.html">rnn_cell_simple</a>()</code> <br><code><a href="rnn_cells_stack.html">rnn_cells_stack</a>()</code> <br></p></div>
+<p>Other preprocessing layers: <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_category_encoding.html">layer_category_encoding</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_discretization.html">layer_discretization</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_feature_space.html">layer_feature_space</a>()</code> <br><code><a href="layer_hashed_crossing.html">layer_hashed_crossing</a>()</code> <br><code><a href="layer_integer_lookup.html">layer_integer_lookup</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_mel_spectrogram.html">layer_mel_spectrogram</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_normalization.html">layer_normalization</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_brightness.html">layer_random_brightness</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_contrast.html">layer_random_contrast</a>()</code> <br><code><a href="layer_random_crop.html">layer_random_crop</a>()</code> <br><code><a href="layer_random_elastic_transform.html">layer_random_elastic_transform</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_flip.html">layer_random_flip</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_rotation.html">layer_random_rotation</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_random_translation.html">layer_random_translation</a>()</code> <br><code><a href="layer_random_zoom.html">layer_random_zoom</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br><code><a href="layer_stft_spectrogram.html">layer_stft_spectrogram</a>()</code> <br><code><a href="layer_string_lookup.html">layer_string_lookup</a>()</code> <br><code><a href="layer_text_vectorization.html">layer_text_vectorization</a>()</code> <br></p>
+<p>Other layers: <br><code><a href="Layer.html">Layer</a>()</code> <br><code><a href="layer_activation.html">layer_activation</a>()</code> <br><code><a href="layer_activation_elu.html">layer_activation_elu</a>()</code> <br><code><a href="layer_activation_leaky_relu.html">layer_activation_leaky_relu</a>()</code> <br><code><a href="layer_activation_parametric_relu.html">layer_activation_parametric_relu</a>()</code> <br><code><a href="layer_activation_relu.html">layer_activation_relu</a>()</code> <br><code><a href="layer_activation_softmax.html">layer_activation_softmax</a>()</code> <br><code><a href="layer_activity_regularization.html">layer_activity_regularization</a>()</code> <br><code><a href="layer_add.html">layer_add</a>()</code> <br><code><a href="layer_additive_attention.html">layer_additive_attention</a>()</code> <br><code><a href="layer_alpha_dropout.html">layer_alpha_dropout</a>()</code> <br><code><a href="layer_attention.html">layer_attention</a>()</code> <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_average.html">layer_average</a>()</code> <br><code><a href="layer_average_pooling_1d.html">layer_average_pooling_1d</a>()</code> <br><code><a href="layer_average_pooling_2d.html">layer_average_pooling_2d</a>()</code> <br><code><a href="layer_average_pooling_3d.html">layer_average_pooling_3d</a>()</code> <br><code><a href="layer_batch_normalization.html">layer_batch_normalization</a>()</code> <br><code><a href="layer_bidirectional.html">layer_bidirectional</a>()</code> <br><code><a href="layer_category_encoding.html">layer_category_encoding</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_concatenate.html">layer_concatenate</a>()</code> <br><code><a href="layer_conv_1d.html">layer_conv_1d</a>()</code> <br><code><a href="layer_conv_1d_transpose.html">layer_conv_1d_transpose</a>()</code> <br><code><a href="layer_conv_2d.html">layer_conv_2d</a>()</code> <br><code><a href="layer_conv_2d_transpose.html">layer_conv_2d_transpose</a>()</code> <br><code><a href="layer_conv_3d.html">layer_conv_3d</a>()</code> <br><code><a href="layer_conv_3d_transpose.html">layer_conv_3d_transpose</a>()</code> <br><code><a href="layer_conv_lstm_1d.html">layer_conv_lstm_1d</a>()</code> <br><code><a href="layer_conv_lstm_2d.html">layer_conv_lstm_2d</a>()</code> <br><code><a href="layer_conv_lstm_3d.html">layer_conv_lstm_3d</a>()</code> <br><code><a href="layer_cropping_1d.html">layer_cropping_1d</a>()</code> <br><code><a href="layer_cropping_2d.html">layer_cropping_2d</a>()</code> <br><code><a href="layer_cropping_3d.html">layer_cropping_3d</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_dense.html">layer_dense</a>()</code> <br><code><a href="layer_depthwise_conv_1d.html">layer_depthwise_conv_1d</a>()</code> <br><code><a href="layer_depthwise_conv_2d.html">layer_depthwise_conv_2d</a>()</code> <br><code><a href="layer_discretization.html">layer_discretization</a>()</code> <br><code><a href="layer_dot.html">layer_dot</a>()</code> <br><code><a href="layer_dropout.html">layer_dropout</a>()</code> <br><code><a href="layer_einsum_dense.html">layer_einsum_dense</a>()</code> <br><code><a href="layer_embedding.html">layer_embedding</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_feature_space.html">layer_feature_space</a>()</code> <br><code><a href="layer_flatten.html">layer_flatten</a>()</code> <br><code><a href="layer_flax_module_wrapper.html">layer_flax_module_wrapper</a>()</code> <br><code><a href="layer_gaussian_dropout.html">layer_gaussian_dropout</a>()</code> <br><code><a href="layer_gaussian_noise.html">layer_gaussian_noise</a>()</code> <br><code><a href="layer_global_average_pooling_1d.html">layer_global_average_pooling_1d</a>()</code> <br><code><a href="layer_global_average_pooling_2d.html">layer_global_average_pooling_2d</a>()</code> <br><code><a href="layer_global_average_pooling_3d.html">layer_global_average_pooling_3d</a>()</code> <br><code><a href="layer_global_max_pooling_1d.html">layer_global_max_pooling_1d</a>()</code> <br><code><a href="layer_global_max_pooling_2d.html">layer_global_max_pooling_2d</a>()</code> <br><code><a href="layer_global_max_pooling_3d.html">layer_global_max_pooling_3d</a>()</code> <br><code><a href="layer_group_normalization.html">layer_group_normalization</a>()</code> <br><code><a href="layer_group_query_attention.html">layer_group_query_attention</a>()</code> <br><code><a href="layer_gru.html">layer_gru</a>()</code> <br><code><a href="layer_hashed_crossing.html">layer_hashed_crossing</a>()</code> <br><code><a href="layer_identity.html">layer_identity</a>()</code> <br><code><a href="layer_integer_lookup.html">layer_integer_lookup</a>()</code> <br><code><a href="layer_jax_model_wrapper.html">layer_jax_model_wrapper</a>()</code> <br><code><a href="layer_lambda.html">layer_lambda</a>()</code> <br><code><a href="layer_layer_normalization.html">layer_layer_normalization</a>()</code> <br><code><a href="layer_lstm.html">layer_lstm</a>()</code> <br><code><a href="layer_masking.html">layer_masking</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_max_pooling_1d.html">layer_max_pooling_1d</a>()</code> <br><code><a href="layer_max_pooling_2d.html">layer_max_pooling_2d</a>()</code> <br><code><a href="layer_max_pooling_3d.html">layer_max_pooling_3d</a>()</code> <br><code><a href="layer_maximum.html">layer_maximum</a>()</code> <br><code><a href="layer_mel_spectrogram.html">layer_mel_spectrogram</a>()</code> <br><code><a href="layer_minimum.html">layer_minimum</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_multi_head_attention.html">layer_multi_head_attention</a>()</code> <br><code><a href="layer_multiply.html">layer_multiply</a>()</code> <br><code><a href="layer_normalization.html">layer_normalization</a>()</code> <br><code><a href="layer_permute.html">layer_permute</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_brightness.html">layer_random_brightness</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_contrast.html">layer_random_contrast</a>()</code> <br><code><a href="layer_random_crop.html">layer_random_crop</a>()</code> <br><code><a href="layer_random_elastic_transform.html">layer_random_elastic_transform</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_flip.html">layer_random_flip</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_rotation.html">layer_random_rotation</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_random_translation.html">layer_random_translation</a>()</code> <br><code><a href="layer_random_zoom.html">layer_random_zoom</a>()</code> <br><code><a href="layer_repeat_vector.html">layer_repeat_vector</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_reshape.html">layer_reshape</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_rms_normalization.html">layer_rms_normalization</a>()</code> <br><code><a href="layer_rnn.html">layer_rnn</a>()</code> <br><code><a href="layer_separable_conv_1d.html">layer_separable_conv_1d</a>()</code> <br><code><a href="layer_separable_conv_2d.html">layer_separable_conv_2d</a>()</code> <br><code><a href="layer_simple_rnn.html">layer_simple_rnn</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br><code><a href="layer_spatial_dropout_1d.html">layer_spatial_dropout_1d</a>()</code> <br><code><a href="layer_spatial_dropout_2d.html">layer_spatial_dropout_2d</a>()</code> <br><code><a href="layer_spatial_dropout_3d.html">layer_spatial_dropout_3d</a>()</code> <br><code><a href="layer_spectral_normalization.html">layer_spectral_normalization</a>()</code> <br><code><a href="layer_stft_spectrogram.html">layer_stft_spectrogram</a>()</code> <br><code><a href="layer_string_lookup.html">layer_string_lookup</a>()</code> <br><code><a href="layer_subtract.html">layer_subtract</a>()</code> <br><code><a href="layer_text_vectorization.html">layer_text_vectorization</a>()</code> <br><code><a href="layer_tfsm.html">layer_tfsm</a>()</code> <br><code><a href="layer_time_distributed.html">layer_time_distributed</a>()</code> <br><code><a href="layer_torch_module_wrapper.html">layer_torch_module_wrapper</a>()</code> <br><code><a href="layer_unit_normalization.html">layer_unit_normalization</a>()</code> <br><code><a href="layer_upsampling_1d.html">layer_upsampling_1d</a>()</code> <br><code><a href="layer_upsampling_2d.html">layer_upsampling_2d</a>()</code> <br><code><a href="layer_upsampling_3d.html">layer_upsampling_3d</a>()</code> <br><code><a href="layer_zero_padding_1d.html">layer_zero_padding_1d</a>()</code> <br><code><a href="layer_zero_padding_2d.html">layer_zero_padding_2d</a>()</code> <br><code><a href="layer_zero_padding_3d.html">layer_zero_padding_3d</a>()</code> <br><code><a href="rnn_cell_gru.html">rnn_cell_gru</a>()</code> <br><code><a href="rnn_cell_lstm.html">rnn_cell_lstm</a>()</code> <br><code><a href="rnn_cell_simple.html">rnn_cell_simple</a>()</code> <br><code><a href="rnn_cells_stack.html">rnn_cells_stack</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -384,7 +384,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/layer_identity.html b/docs/dev/reference/layer_identity.html
index 78367ee09..a06f9317c 100644
--- a/docs/dev/reference/layer_identity.html
+++ b/docs/dev/reference/layer_identity.html
@@ -10,7 +10,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -89,7 +89,7 @@ <h2 id="value">Value<a class="anchor" aria-label="anchor" href="#value"></a></h2
     <div class="section level2">
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
     <div class="dont-index"><p>Other core layers: <br><code><a href="layer_dense.html">layer_dense</a>()</code> <br><code><a href="layer_einsum_dense.html">layer_einsum_dense</a>()</code> <br><code><a href="layer_embedding.html">layer_embedding</a>()</code> <br><code><a href="layer_lambda.html">layer_lambda</a>()</code> <br><code><a href="layer_masking.html">layer_masking</a>()</code> <br></p>
-<p>Other layers: <br><code><a href="Layer.html">Layer</a>()</code> <br><code><a href="layer_activation.html">layer_activation</a>()</code> <br><code><a href="layer_activation_elu.html">layer_activation_elu</a>()</code> <br><code><a href="layer_activation_leaky_relu.html">layer_activation_leaky_relu</a>()</code> <br><code><a href="layer_activation_parametric_relu.html">layer_activation_parametric_relu</a>()</code> <br><code><a href="layer_activation_relu.html">layer_activation_relu</a>()</code> <br><code><a href="layer_activation_softmax.html">layer_activation_softmax</a>()</code> <br><code><a href="layer_activity_regularization.html">layer_activity_regularization</a>()</code> <br><code><a href="layer_add.html">layer_add</a>()</code> <br><code><a href="layer_additive_attention.html">layer_additive_attention</a>()</code> <br><code><a href="layer_alpha_dropout.html">layer_alpha_dropout</a>()</code> <br><code><a href="layer_attention.html">layer_attention</a>()</code> <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_average.html">layer_average</a>()</code> <br><code><a href="layer_average_pooling_1d.html">layer_average_pooling_1d</a>()</code> <br><code><a href="layer_average_pooling_2d.html">layer_average_pooling_2d</a>()</code> <br><code><a href="layer_average_pooling_3d.html">layer_average_pooling_3d</a>()</code> <br><code><a href="layer_batch_normalization.html">layer_batch_normalization</a>()</code> <br><code><a href="layer_bidirectional.html">layer_bidirectional</a>()</code> <br><code><a href="layer_category_encoding.html">layer_category_encoding</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_concatenate.html">layer_concatenate</a>()</code> <br><code><a href="layer_conv_1d.html">layer_conv_1d</a>()</code> <br><code><a href="layer_conv_1d_transpose.html">layer_conv_1d_transpose</a>()</code> <br><code><a href="layer_conv_2d.html">layer_conv_2d</a>()</code> <br><code><a href="layer_conv_2d_transpose.html">layer_conv_2d_transpose</a>()</code> <br><code><a href="layer_conv_3d.html">layer_conv_3d</a>()</code> <br><code><a href="layer_conv_3d_transpose.html">layer_conv_3d_transpose</a>()</code> <br><code><a href="layer_conv_lstm_1d.html">layer_conv_lstm_1d</a>()</code> <br><code><a href="layer_conv_lstm_2d.html">layer_conv_lstm_2d</a>()</code> <br><code><a href="layer_conv_lstm_3d.html">layer_conv_lstm_3d</a>()</code> <br><code><a href="layer_cropping_1d.html">layer_cropping_1d</a>()</code> <br><code><a href="layer_cropping_2d.html">layer_cropping_2d</a>()</code> <br><code><a href="layer_cropping_3d.html">layer_cropping_3d</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_dense.html">layer_dense</a>()</code> <br><code><a href="layer_depthwise_conv_1d.html">layer_depthwise_conv_1d</a>()</code> <br><code><a href="layer_depthwise_conv_2d.html">layer_depthwise_conv_2d</a>()</code> <br><code><a href="layer_discretization.html">layer_discretization</a>()</code> <br><code><a href="layer_dot.html">layer_dot</a>()</code> <br><code><a href="layer_dropout.html">layer_dropout</a>()</code> <br><code><a href="layer_einsum_dense.html">layer_einsum_dense</a>()</code> <br><code><a href="layer_embedding.html">layer_embedding</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_feature_space.html">layer_feature_space</a>()</code> <br><code><a href="layer_flatten.html">layer_flatten</a>()</code> <br><code><a href="layer_flax_module_wrapper.html">layer_flax_module_wrapper</a>()</code> <br><code><a href="layer_gaussian_dropout.html">layer_gaussian_dropout</a>()</code> <br><code><a href="layer_gaussian_noise.html">layer_gaussian_noise</a>()</code> <br><code><a href="layer_global_average_pooling_1d.html">layer_global_average_pooling_1d</a>()</code> <br><code><a href="layer_global_average_pooling_2d.html">layer_global_average_pooling_2d</a>()</code> <br><code><a href="layer_global_average_pooling_3d.html">layer_global_average_pooling_3d</a>()</code> <br><code><a href="layer_global_max_pooling_1d.html">layer_global_max_pooling_1d</a>()</code> <br><code><a href="layer_global_max_pooling_2d.html">layer_global_max_pooling_2d</a>()</code> <br><code><a href="layer_global_max_pooling_3d.html">layer_global_max_pooling_3d</a>()</code> <br><code><a href="layer_group_normalization.html">layer_group_normalization</a>()</code> <br><code><a href="layer_group_query_attention.html">layer_group_query_attention</a>()</code> <br><code><a href="layer_gru.html">layer_gru</a>()</code> <br><code><a href="layer_hashed_crossing.html">layer_hashed_crossing</a>()</code> <br><code><a href="layer_hashing.html">layer_hashing</a>()</code> <br><code><a href="layer_integer_lookup.html">layer_integer_lookup</a>()</code> <br><code><a href="layer_jax_model_wrapper.html">layer_jax_model_wrapper</a>()</code> <br><code><a href="layer_lambda.html">layer_lambda</a>()</code> <br><code><a href="layer_layer_normalization.html">layer_layer_normalization</a>()</code> <br><code><a href="layer_lstm.html">layer_lstm</a>()</code> <br><code><a href="layer_masking.html">layer_masking</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_max_pooling_1d.html">layer_max_pooling_1d</a>()</code> <br><code><a href="layer_max_pooling_2d.html">layer_max_pooling_2d</a>()</code> <br><code><a href="layer_max_pooling_3d.html">layer_max_pooling_3d</a>()</code> <br><code><a href="layer_maximum.html">layer_maximum</a>()</code> <br><code><a href="layer_mel_spectrogram.html">layer_mel_spectrogram</a>()</code> <br><code><a href="layer_minimum.html">layer_minimum</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_multi_head_attention.html">layer_multi_head_attention</a>()</code> <br><code><a href="layer_multiply.html">layer_multiply</a>()</code> <br><code><a href="layer_normalization.html">layer_normalization</a>()</code> <br><code><a href="layer_permute.html">layer_permute</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_brightness.html">layer_random_brightness</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_contrast.html">layer_random_contrast</a>()</code> <br><code><a href="layer_random_crop.html">layer_random_crop</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_flip.html">layer_random_flip</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_rotation.html">layer_random_rotation</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_random_translation.html">layer_random_translation</a>()</code> <br><code><a href="layer_random_zoom.html">layer_random_zoom</a>()</code> <br><code><a href="layer_repeat_vector.html">layer_repeat_vector</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_reshape.html">layer_reshape</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_rms_normalization.html">layer_rms_normalization</a>()</code> <br><code><a href="layer_rnn.html">layer_rnn</a>()</code> <br><code><a href="layer_separable_conv_1d.html">layer_separable_conv_1d</a>()</code> <br><code><a href="layer_separable_conv_2d.html">layer_separable_conv_2d</a>()</code> <br><code><a href="layer_simple_rnn.html">layer_simple_rnn</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br><code><a href="layer_spatial_dropout_1d.html">layer_spatial_dropout_1d</a>()</code> <br><code><a href="layer_spatial_dropout_2d.html">layer_spatial_dropout_2d</a>()</code> <br><code><a href="layer_spatial_dropout_3d.html">layer_spatial_dropout_3d</a>()</code> <br><code><a href="layer_spectral_normalization.html">layer_spectral_normalization</a>()</code> <br><code><a href="layer_stft_spectrogram.html">layer_stft_spectrogram</a>()</code> <br><code><a href="layer_string_lookup.html">layer_string_lookup</a>()</code> <br><code><a href="layer_subtract.html">layer_subtract</a>()</code> <br><code><a href="layer_text_vectorization.html">layer_text_vectorization</a>()</code> <br><code><a href="layer_tfsm.html">layer_tfsm</a>()</code> <br><code><a href="layer_time_distributed.html">layer_time_distributed</a>()</code> <br><code><a href="layer_torch_module_wrapper.html">layer_torch_module_wrapper</a>()</code> <br><code><a href="layer_unit_normalization.html">layer_unit_normalization</a>()</code> <br><code><a href="layer_upsampling_1d.html">layer_upsampling_1d</a>()</code> <br><code><a href="layer_upsampling_2d.html">layer_upsampling_2d</a>()</code> <br><code><a href="layer_upsampling_3d.html">layer_upsampling_3d</a>()</code> <br><code><a href="layer_zero_padding_1d.html">layer_zero_padding_1d</a>()</code> <br><code><a href="layer_zero_padding_2d.html">layer_zero_padding_2d</a>()</code> <br><code><a href="layer_zero_padding_3d.html">layer_zero_padding_3d</a>()</code> <br><code><a href="rnn_cell_gru.html">rnn_cell_gru</a>()</code> <br><code><a href="rnn_cell_lstm.html">rnn_cell_lstm</a>()</code> <br><code><a href="rnn_cell_simple.html">rnn_cell_simple</a>()</code> <br><code><a href="rnn_cells_stack.html">rnn_cells_stack</a>()</code> <br></p></div>
+<p>Other layers: <br><code><a href="Layer.html">Layer</a>()</code> <br><code><a href="layer_activation.html">layer_activation</a>()</code> <br><code><a href="layer_activation_elu.html">layer_activation_elu</a>()</code> <br><code><a href="layer_activation_leaky_relu.html">layer_activation_leaky_relu</a>()</code> <br><code><a href="layer_activation_parametric_relu.html">layer_activation_parametric_relu</a>()</code> <br><code><a href="layer_activation_relu.html">layer_activation_relu</a>()</code> <br><code><a href="layer_activation_softmax.html">layer_activation_softmax</a>()</code> <br><code><a href="layer_activity_regularization.html">layer_activity_regularization</a>()</code> <br><code><a href="layer_add.html">layer_add</a>()</code> <br><code><a href="layer_additive_attention.html">layer_additive_attention</a>()</code> <br><code><a href="layer_alpha_dropout.html">layer_alpha_dropout</a>()</code> <br><code><a href="layer_attention.html">layer_attention</a>()</code> <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_average.html">layer_average</a>()</code> <br><code><a href="layer_average_pooling_1d.html">layer_average_pooling_1d</a>()</code> <br><code><a href="layer_average_pooling_2d.html">layer_average_pooling_2d</a>()</code> <br><code><a href="layer_average_pooling_3d.html">layer_average_pooling_3d</a>()</code> <br><code><a href="layer_batch_normalization.html">layer_batch_normalization</a>()</code> <br><code><a href="layer_bidirectional.html">layer_bidirectional</a>()</code> <br><code><a href="layer_category_encoding.html">layer_category_encoding</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_concatenate.html">layer_concatenate</a>()</code> <br><code><a href="layer_conv_1d.html">layer_conv_1d</a>()</code> <br><code><a href="layer_conv_1d_transpose.html">layer_conv_1d_transpose</a>()</code> <br><code><a href="layer_conv_2d.html">layer_conv_2d</a>()</code> <br><code><a href="layer_conv_2d_transpose.html">layer_conv_2d_transpose</a>()</code> <br><code><a href="layer_conv_3d.html">layer_conv_3d</a>()</code> <br><code><a href="layer_conv_3d_transpose.html">layer_conv_3d_transpose</a>()</code> <br><code><a href="layer_conv_lstm_1d.html">layer_conv_lstm_1d</a>()</code> <br><code><a href="layer_conv_lstm_2d.html">layer_conv_lstm_2d</a>()</code> <br><code><a href="layer_conv_lstm_3d.html">layer_conv_lstm_3d</a>()</code> <br><code><a href="layer_cropping_1d.html">layer_cropping_1d</a>()</code> <br><code><a href="layer_cropping_2d.html">layer_cropping_2d</a>()</code> <br><code><a href="layer_cropping_3d.html">layer_cropping_3d</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_dense.html">layer_dense</a>()</code> <br><code><a href="layer_depthwise_conv_1d.html">layer_depthwise_conv_1d</a>()</code> <br><code><a href="layer_depthwise_conv_2d.html">layer_depthwise_conv_2d</a>()</code> <br><code><a href="layer_discretization.html">layer_discretization</a>()</code> <br><code><a href="layer_dot.html">layer_dot</a>()</code> <br><code><a href="layer_dropout.html">layer_dropout</a>()</code> <br><code><a href="layer_einsum_dense.html">layer_einsum_dense</a>()</code> <br><code><a href="layer_embedding.html">layer_embedding</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_feature_space.html">layer_feature_space</a>()</code> <br><code><a href="layer_flatten.html">layer_flatten</a>()</code> <br><code><a href="layer_flax_module_wrapper.html">layer_flax_module_wrapper</a>()</code> <br><code><a href="layer_gaussian_dropout.html">layer_gaussian_dropout</a>()</code> <br><code><a href="layer_gaussian_noise.html">layer_gaussian_noise</a>()</code> <br><code><a href="layer_global_average_pooling_1d.html">layer_global_average_pooling_1d</a>()</code> <br><code><a href="layer_global_average_pooling_2d.html">layer_global_average_pooling_2d</a>()</code> <br><code><a href="layer_global_average_pooling_3d.html">layer_global_average_pooling_3d</a>()</code> <br><code><a href="layer_global_max_pooling_1d.html">layer_global_max_pooling_1d</a>()</code> <br><code><a href="layer_global_max_pooling_2d.html">layer_global_max_pooling_2d</a>()</code> <br><code><a href="layer_global_max_pooling_3d.html">layer_global_max_pooling_3d</a>()</code> <br><code><a href="layer_group_normalization.html">layer_group_normalization</a>()</code> <br><code><a href="layer_group_query_attention.html">layer_group_query_attention</a>()</code> <br><code><a href="layer_gru.html">layer_gru</a>()</code> <br><code><a href="layer_hashed_crossing.html">layer_hashed_crossing</a>()</code> <br><code><a href="layer_hashing.html">layer_hashing</a>()</code> <br><code><a href="layer_integer_lookup.html">layer_integer_lookup</a>()</code> <br><code><a href="layer_jax_model_wrapper.html">layer_jax_model_wrapper</a>()</code> <br><code><a href="layer_lambda.html">layer_lambda</a>()</code> <br><code><a href="layer_layer_normalization.html">layer_layer_normalization</a>()</code> <br><code><a href="layer_lstm.html">layer_lstm</a>()</code> <br><code><a href="layer_masking.html">layer_masking</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_max_pooling_1d.html">layer_max_pooling_1d</a>()</code> <br><code><a href="layer_max_pooling_2d.html">layer_max_pooling_2d</a>()</code> <br><code><a href="layer_max_pooling_3d.html">layer_max_pooling_3d</a>()</code> <br><code><a href="layer_maximum.html">layer_maximum</a>()</code> <br><code><a href="layer_mel_spectrogram.html">layer_mel_spectrogram</a>()</code> <br><code><a href="layer_minimum.html">layer_minimum</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_multi_head_attention.html">layer_multi_head_attention</a>()</code> <br><code><a href="layer_multiply.html">layer_multiply</a>()</code> <br><code><a href="layer_normalization.html">layer_normalization</a>()</code> <br><code><a href="layer_permute.html">layer_permute</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_brightness.html">layer_random_brightness</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_contrast.html">layer_random_contrast</a>()</code> <br><code><a href="layer_random_crop.html">layer_random_crop</a>()</code> <br><code><a href="layer_random_elastic_transform.html">layer_random_elastic_transform</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_flip.html">layer_random_flip</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_rotation.html">layer_random_rotation</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_random_translation.html">layer_random_translation</a>()</code> <br><code><a href="layer_random_zoom.html">layer_random_zoom</a>()</code> <br><code><a href="layer_repeat_vector.html">layer_repeat_vector</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_reshape.html">layer_reshape</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_rms_normalization.html">layer_rms_normalization</a>()</code> <br><code><a href="layer_rnn.html">layer_rnn</a>()</code> <br><code><a href="layer_separable_conv_1d.html">layer_separable_conv_1d</a>()</code> <br><code><a href="layer_separable_conv_2d.html">layer_separable_conv_2d</a>()</code> <br><code><a href="layer_simple_rnn.html">layer_simple_rnn</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br><code><a href="layer_spatial_dropout_1d.html">layer_spatial_dropout_1d</a>()</code> <br><code><a href="layer_spatial_dropout_2d.html">layer_spatial_dropout_2d</a>()</code> <br><code><a href="layer_spatial_dropout_3d.html">layer_spatial_dropout_3d</a>()</code> <br><code><a href="layer_spectral_normalization.html">layer_spectral_normalization</a>()</code> <br><code><a href="layer_stft_spectrogram.html">layer_stft_spectrogram</a>()</code> <br><code><a href="layer_string_lookup.html">layer_string_lookup</a>()</code> <br><code><a href="layer_subtract.html">layer_subtract</a>()</code> <br><code><a href="layer_text_vectorization.html">layer_text_vectorization</a>()</code> <br><code><a href="layer_tfsm.html">layer_tfsm</a>()</code> <br><code><a href="layer_time_distributed.html">layer_time_distributed</a>()</code> <br><code><a href="layer_torch_module_wrapper.html">layer_torch_module_wrapper</a>()</code> <br><code><a href="layer_unit_normalization.html">layer_unit_normalization</a>()</code> <br><code><a href="layer_upsampling_1d.html">layer_upsampling_1d</a>()</code> <br><code><a href="layer_upsampling_2d.html">layer_upsampling_2d</a>()</code> <br><code><a href="layer_upsampling_3d.html">layer_upsampling_3d</a>()</code> <br><code><a href="layer_zero_padding_1d.html">layer_zero_padding_1d</a>()</code> <br><code><a href="layer_zero_padding_2d.html">layer_zero_padding_2d</a>()</code> <br><code><a href="layer_zero_padding_3d.html">layer_zero_padding_3d</a>()</code> <br><code><a href="rnn_cell_gru.html">rnn_cell_gru</a>()</code> <br><code><a href="rnn_cell_lstm.html">rnn_cell_lstm</a>()</code> <br><code><a href="rnn_cell_simple.html">rnn_cell_simple</a>()</code> <br><code><a href="rnn_cells_stack.html">rnn_cells_stack</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -101,7 +101,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/layer_input.html b/docs/dev/reference/layer_input.html
index ec7383516..7cd47e3a1 100644
--- a/docs/dev/reference/layer_input.html
+++ b/docs/dev/reference/layer_input.html
@@ -10,7 +10,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -150,7 +150,7 @@ <h2 id="value">Value<a class="anchor" aria-label="anchor" href="#value"></a></h2
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/layer_integer_lookup.html b/docs/dev/reference/layer_integer_lookup.html
index 0a6a76a25..cf477e554 100644
--- a/docs/dev/reference/layer_integer_lookup.html
+++ b/docs/dev/reference/layer_integer_lookup.html
@@ -84,7 +84,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -494,8 +494,8 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
     <div class="dont-index">
 <ul><li><p><a href="https://keras.io/api/layers/preprocessing_layers/categorical/integer_lookup#integerlookup-class" class="external-link">https://keras.io/api/layers/preprocessing_layers/categorical/integer_lookup#integerlookup-class</a></p></li>
 </ul><p>Other categorical features preprocessing layers: <br><code><a href="layer_category_encoding.html">layer_category_encoding</a>()</code> <br><code><a href="layer_hashed_crossing.html">layer_hashed_crossing</a>()</code> <br><code><a href="layer_hashing.html">layer_hashing</a>()</code> <br><code><a href="layer_string_lookup.html">layer_string_lookup</a>()</code> <br></p>
-<p>Other preprocessing layers: <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_category_encoding.html">layer_category_encoding</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_discretization.html">layer_discretization</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_feature_space.html">layer_feature_space</a>()</code> <br><code><a href="layer_hashed_crossing.html">layer_hashed_crossing</a>()</code> <br><code><a href="layer_hashing.html">layer_hashing</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_mel_spectrogram.html">layer_mel_spectrogram</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_normalization.html">layer_normalization</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_brightness.html">layer_random_brightness</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_contrast.html">layer_random_contrast</a>()</code> <br><code><a href="layer_random_crop.html">layer_random_crop</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_flip.html">layer_random_flip</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_rotation.html">layer_random_rotation</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_random_translation.html">layer_random_translation</a>()</code> <br><code><a href="layer_random_zoom.html">layer_random_zoom</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br><code><a href="layer_stft_spectrogram.html">layer_stft_spectrogram</a>()</code> <br><code><a href="layer_string_lookup.html">layer_string_lookup</a>()</code> <br><code><a href="layer_text_vectorization.html">layer_text_vectorization</a>()</code> <br></p>
-<p>Other layers: <br><code><a href="Layer.html">Layer</a>()</code> <br><code><a href="layer_activation.html">layer_activation</a>()</code> <br><code><a href="layer_activation_elu.html">layer_activation_elu</a>()</code> <br><code><a href="layer_activation_leaky_relu.html">layer_activation_leaky_relu</a>()</code> <br><code><a href="layer_activation_parametric_relu.html">layer_activation_parametric_relu</a>()</code> <br><code><a href="layer_activation_relu.html">layer_activation_relu</a>()</code> <br><code><a href="layer_activation_softmax.html">layer_activation_softmax</a>()</code> <br><code><a href="layer_activity_regularization.html">layer_activity_regularization</a>()</code> <br><code><a href="layer_add.html">layer_add</a>()</code> <br><code><a href="layer_additive_attention.html">layer_additive_attention</a>()</code> <br><code><a href="layer_alpha_dropout.html">layer_alpha_dropout</a>()</code> <br><code><a href="layer_attention.html">layer_attention</a>()</code> <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_average.html">layer_average</a>()</code> <br><code><a href="layer_average_pooling_1d.html">layer_average_pooling_1d</a>()</code> <br><code><a href="layer_average_pooling_2d.html">layer_average_pooling_2d</a>()</code> <br><code><a href="layer_average_pooling_3d.html">layer_average_pooling_3d</a>()</code> <br><code><a href="layer_batch_normalization.html">layer_batch_normalization</a>()</code> <br><code><a href="layer_bidirectional.html">layer_bidirectional</a>()</code> <br><code><a href="layer_category_encoding.html">layer_category_encoding</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_concatenate.html">layer_concatenate</a>()</code> <br><code><a href="layer_conv_1d.html">layer_conv_1d</a>()</code> <br><code><a href="layer_conv_1d_transpose.html">layer_conv_1d_transpose</a>()</code> <br><code><a href="layer_conv_2d.html">layer_conv_2d</a>()</code> <br><code><a href="layer_conv_2d_transpose.html">layer_conv_2d_transpose</a>()</code> <br><code><a href="layer_conv_3d.html">layer_conv_3d</a>()</code> <br><code><a href="layer_conv_3d_transpose.html">layer_conv_3d_transpose</a>()</code> <br><code><a href="layer_conv_lstm_1d.html">layer_conv_lstm_1d</a>()</code> <br><code><a href="layer_conv_lstm_2d.html">layer_conv_lstm_2d</a>()</code> <br><code><a href="layer_conv_lstm_3d.html">layer_conv_lstm_3d</a>()</code> <br><code><a href="layer_cropping_1d.html">layer_cropping_1d</a>()</code> <br><code><a href="layer_cropping_2d.html">layer_cropping_2d</a>()</code> <br><code><a href="layer_cropping_3d.html">layer_cropping_3d</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_dense.html">layer_dense</a>()</code> <br><code><a href="layer_depthwise_conv_1d.html">layer_depthwise_conv_1d</a>()</code> <br><code><a href="layer_depthwise_conv_2d.html">layer_depthwise_conv_2d</a>()</code> <br><code><a href="layer_discretization.html">layer_discretization</a>()</code> <br><code><a href="layer_dot.html">layer_dot</a>()</code> <br><code><a href="layer_dropout.html">layer_dropout</a>()</code> <br><code><a href="layer_einsum_dense.html">layer_einsum_dense</a>()</code> <br><code><a href="layer_embedding.html">layer_embedding</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_feature_space.html">layer_feature_space</a>()</code> <br><code><a href="layer_flatten.html">layer_flatten</a>()</code> <br><code><a href="layer_flax_module_wrapper.html">layer_flax_module_wrapper</a>()</code> <br><code><a href="layer_gaussian_dropout.html">layer_gaussian_dropout</a>()</code> <br><code><a href="layer_gaussian_noise.html">layer_gaussian_noise</a>()</code> <br><code><a href="layer_global_average_pooling_1d.html">layer_global_average_pooling_1d</a>()</code> <br><code><a href="layer_global_average_pooling_2d.html">layer_global_average_pooling_2d</a>()</code> <br><code><a href="layer_global_average_pooling_3d.html">layer_global_average_pooling_3d</a>()</code> <br><code><a href="layer_global_max_pooling_1d.html">layer_global_max_pooling_1d</a>()</code> <br><code><a href="layer_global_max_pooling_2d.html">layer_global_max_pooling_2d</a>()</code> <br><code><a href="layer_global_max_pooling_3d.html">layer_global_max_pooling_3d</a>()</code> <br><code><a href="layer_group_normalization.html">layer_group_normalization</a>()</code> <br><code><a href="layer_group_query_attention.html">layer_group_query_attention</a>()</code> <br><code><a href="layer_gru.html">layer_gru</a>()</code> <br><code><a href="layer_hashed_crossing.html">layer_hashed_crossing</a>()</code> <br><code><a href="layer_hashing.html">layer_hashing</a>()</code> <br><code><a href="layer_identity.html">layer_identity</a>()</code> <br><code><a href="layer_jax_model_wrapper.html">layer_jax_model_wrapper</a>()</code> <br><code><a href="layer_lambda.html">layer_lambda</a>()</code> <br><code><a href="layer_layer_normalization.html">layer_layer_normalization</a>()</code> <br><code><a href="layer_lstm.html">layer_lstm</a>()</code> <br><code><a href="layer_masking.html">layer_masking</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_max_pooling_1d.html">layer_max_pooling_1d</a>()</code> <br><code><a href="layer_max_pooling_2d.html">layer_max_pooling_2d</a>()</code> <br><code><a href="layer_max_pooling_3d.html">layer_max_pooling_3d</a>()</code> <br><code><a href="layer_maximum.html">layer_maximum</a>()</code> <br><code><a href="layer_mel_spectrogram.html">layer_mel_spectrogram</a>()</code> <br><code><a href="layer_minimum.html">layer_minimum</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_multi_head_attention.html">layer_multi_head_attention</a>()</code> <br><code><a href="layer_multiply.html">layer_multiply</a>()</code> <br><code><a href="layer_normalization.html">layer_normalization</a>()</code> <br><code><a href="layer_permute.html">layer_permute</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_brightness.html">layer_random_brightness</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_contrast.html">layer_random_contrast</a>()</code> <br><code><a href="layer_random_crop.html">layer_random_crop</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_flip.html">layer_random_flip</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_rotation.html">layer_random_rotation</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_random_translation.html">layer_random_translation</a>()</code> <br><code><a href="layer_random_zoom.html">layer_random_zoom</a>()</code> <br><code><a href="layer_repeat_vector.html">layer_repeat_vector</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_reshape.html">layer_reshape</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_rms_normalization.html">layer_rms_normalization</a>()</code> <br><code><a href="layer_rnn.html">layer_rnn</a>()</code> <br><code><a href="layer_separable_conv_1d.html">layer_separable_conv_1d</a>()</code> <br><code><a href="layer_separable_conv_2d.html">layer_separable_conv_2d</a>()</code> <br><code><a href="layer_simple_rnn.html">layer_simple_rnn</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br><code><a href="layer_spatial_dropout_1d.html">layer_spatial_dropout_1d</a>()</code> <br><code><a href="layer_spatial_dropout_2d.html">layer_spatial_dropout_2d</a>()</code> <br><code><a href="layer_spatial_dropout_3d.html">layer_spatial_dropout_3d</a>()</code> <br><code><a href="layer_spectral_normalization.html">layer_spectral_normalization</a>()</code> <br><code><a href="layer_stft_spectrogram.html">layer_stft_spectrogram</a>()</code> <br><code><a href="layer_string_lookup.html">layer_string_lookup</a>()</code> <br><code><a href="layer_subtract.html">layer_subtract</a>()</code> <br><code><a href="layer_text_vectorization.html">layer_text_vectorization</a>()</code> <br><code><a href="layer_tfsm.html">layer_tfsm</a>()</code> <br><code><a href="layer_time_distributed.html">layer_time_distributed</a>()</code> <br><code><a href="layer_torch_module_wrapper.html">layer_torch_module_wrapper</a>()</code> <br><code><a href="layer_unit_normalization.html">layer_unit_normalization</a>()</code> <br><code><a href="layer_upsampling_1d.html">layer_upsampling_1d</a>()</code> <br><code><a href="layer_upsampling_2d.html">layer_upsampling_2d</a>()</code> <br><code><a href="layer_upsampling_3d.html">layer_upsampling_3d</a>()</code> <br><code><a href="layer_zero_padding_1d.html">layer_zero_padding_1d</a>()</code> <br><code><a href="layer_zero_padding_2d.html">layer_zero_padding_2d</a>()</code> <br><code><a href="layer_zero_padding_3d.html">layer_zero_padding_3d</a>()</code> <br><code><a href="rnn_cell_gru.html">rnn_cell_gru</a>()</code> <br><code><a href="rnn_cell_lstm.html">rnn_cell_lstm</a>()</code> <br><code><a href="rnn_cell_simple.html">rnn_cell_simple</a>()</code> <br><code><a href="rnn_cells_stack.html">rnn_cells_stack</a>()</code> <br></p></div>
+<p>Other preprocessing layers: <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_category_encoding.html">layer_category_encoding</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_discretization.html">layer_discretization</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_feature_space.html">layer_feature_space</a>()</code> <br><code><a href="layer_hashed_crossing.html">layer_hashed_crossing</a>()</code> <br><code><a href="layer_hashing.html">layer_hashing</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_mel_spectrogram.html">layer_mel_spectrogram</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_normalization.html">layer_normalization</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_brightness.html">layer_random_brightness</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_contrast.html">layer_random_contrast</a>()</code> <br><code><a href="layer_random_crop.html">layer_random_crop</a>()</code> <br><code><a href="layer_random_elastic_transform.html">layer_random_elastic_transform</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_flip.html">layer_random_flip</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_rotation.html">layer_random_rotation</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_random_translation.html">layer_random_translation</a>()</code> <br><code><a href="layer_random_zoom.html">layer_random_zoom</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br><code><a href="layer_stft_spectrogram.html">layer_stft_spectrogram</a>()</code> <br><code><a href="layer_string_lookup.html">layer_string_lookup</a>()</code> <br><code><a href="layer_text_vectorization.html">layer_text_vectorization</a>()</code> <br></p>
+<p>Other layers: <br><code><a href="Layer.html">Layer</a>()</code> <br><code><a href="layer_activation.html">layer_activation</a>()</code> <br><code><a href="layer_activation_elu.html">layer_activation_elu</a>()</code> <br><code><a href="layer_activation_leaky_relu.html">layer_activation_leaky_relu</a>()</code> <br><code><a href="layer_activation_parametric_relu.html">layer_activation_parametric_relu</a>()</code> <br><code><a href="layer_activation_relu.html">layer_activation_relu</a>()</code> <br><code><a href="layer_activation_softmax.html">layer_activation_softmax</a>()</code> <br><code><a href="layer_activity_regularization.html">layer_activity_regularization</a>()</code> <br><code><a href="layer_add.html">layer_add</a>()</code> <br><code><a href="layer_additive_attention.html">layer_additive_attention</a>()</code> <br><code><a href="layer_alpha_dropout.html">layer_alpha_dropout</a>()</code> <br><code><a href="layer_attention.html">layer_attention</a>()</code> <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_average.html">layer_average</a>()</code> <br><code><a href="layer_average_pooling_1d.html">layer_average_pooling_1d</a>()</code> <br><code><a href="layer_average_pooling_2d.html">layer_average_pooling_2d</a>()</code> <br><code><a href="layer_average_pooling_3d.html">layer_average_pooling_3d</a>()</code> <br><code><a href="layer_batch_normalization.html">layer_batch_normalization</a>()</code> <br><code><a href="layer_bidirectional.html">layer_bidirectional</a>()</code> <br><code><a href="layer_category_encoding.html">layer_category_encoding</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_concatenate.html">layer_concatenate</a>()</code> <br><code><a href="layer_conv_1d.html">layer_conv_1d</a>()</code> <br><code><a href="layer_conv_1d_transpose.html">layer_conv_1d_transpose</a>()</code> <br><code><a href="layer_conv_2d.html">layer_conv_2d</a>()</code> <br><code><a href="layer_conv_2d_transpose.html">layer_conv_2d_transpose</a>()</code> <br><code><a href="layer_conv_3d.html">layer_conv_3d</a>()</code> <br><code><a href="layer_conv_3d_transpose.html">layer_conv_3d_transpose</a>()</code> <br><code><a href="layer_conv_lstm_1d.html">layer_conv_lstm_1d</a>()</code> <br><code><a href="layer_conv_lstm_2d.html">layer_conv_lstm_2d</a>()</code> <br><code><a href="layer_conv_lstm_3d.html">layer_conv_lstm_3d</a>()</code> <br><code><a href="layer_cropping_1d.html">layer_cropping_1d</a>()</code> <br><code><a href="layer_cropping_2d.html">layer_cropping_2d</a>()</code> <br><code><a href="layer_cropping_3d.html">layer_cropping_3d</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_dense.html">layer_dense</a>()</code> <br><code><a href="layer_depthwise_conv_1d.html">layer_depthwise_conv_1d</a>()</code> <br><code><a href="layer_depthwise_conv_2d.html">layer_depthwise_conv_2d</a>()</code> <br><code><a href="layer_discretization.html">layer_discretization</a>()</code> <br><code><a href="layer_dot.html">layer_dot</a>()</code> <br><code><a href="layer_dropout.html">layer_dropout</a>()</code> <br><code><a href="layer_einsum_dense.html">layer_einsum_dense</a>()</code> <br><code><a href="layer_embedding.html">layer_embedding</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_feature_space.html">layer_feature_space</a>()</code> <br><code><a href="layer_flatten.html">layer_flatten</a>()</code> <br><code><a href="layer_flax_module_wrapper.html">layer_flax_module_wrapper</a>()</code> <br><code><a href="layer_gaussian_dropout.html">layer_gaussian_dropout</a>()</code> <br><code><a href="layer_gaussian_noise.html">layer_gaussian_noise</a>()</code> <br><code><a href="layer_global_average_pooling_1d.html">layer_global_average_pooling_1d</a>()</code> <br><code><a href="layer_global_average_pooling_2d.html">layer_global_average_pooling_2d</a>()</code> <br><code><a href="layer_global_average_pooling_3d.html">layer_global_average_pooling_3d</a>()</code> <br><code><a href="layer_global_max_pooling_1d.html">layer_global_max_pooling_1d</a>()</code> <br><code><a href="layer_global_max_pooling_2d.html">layer_global_max_pooling_2d</a>()</code> <br><code><a href="layer_global_max_pooling_3d.html">layer_global_max_pooling_3d</a>()</code> <br><code><a href="layer_group_normalization.html">layer_group_normalization</a>()</code> <br><code><a href="layer_group_query_attention.html">layer_group_query_attention</a>()</code> <br><code><a href="layer_gru.html">layer_gru</a>()</code> <br><code><a href="layer_hashed_crossing.html">layer_hashed_crossing</a>()</code> <br><code><a href="layer_hashing.html">layer_hashing</a>()</code> <br><code><a href="layer_identity.html">layer_identity</a>()</code> <br><code><a href="layer_jax_model_wrapper.html">layer_jax_model_wrapper</a>()</code> <br><code><a href="layer_lambda.html">layer_lambda</a>()</code> <br><code><a href="layer_layer_normalization.html">layer_layer_normalization</a>()</code> <br><code><a href="layer_lstm.html">layer_lstm</a>()</code> <br><code><a href="layer_masking.html">layer_masking</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_max_pooling_1d.html">layer_max_pooling_1d</a>()</code> <br><code><a href="layer_max_pooling_2d.html">layer_max_pooling_2d</a>()</code> <br><code><a href="layer_max_pooling_3d.html">layer_max_pooling_3d</a>()</code> <br><code><a href="layer_maximum.html">layer_maximum</a>()</code> <br><code><a href="layer_mel_spectrogram.html">layer_mel_spectrogram</a>()</code> <br><code><a href="layer_minimum.html">layer_minimum</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_multi_head_attention.html">layer_multi_head_attention</a>()</code> <br><code><a href="layer_multiply.html">layer_multiply</a>()</code> <br><code><a href="layer_normalization.html">layer_normalization</a>()</code> <br><code><a href="layer_permute.html">layer_permute</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_brightness.html">layer_random_brightness</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_contrast.html">layer_random_contrast</a>()</code> <br><code><a href="layer_random_crop.html">layer_random_crop</a>()</code> <br><code><a href="layer_random_elastic_transform.html">layer_random_elastic_transform</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_flip.html">layer_random_flip</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_rotation.html">layer_random_rotation</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_random_translation.html">layer_random_translation</a>()</code> <br><code><a href="layer_random_zoom.html">layer_random_zoom</a>()</code> <br><code><a href="layer_repeat_vector.html">layer_repeat_vector</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_reshape.html">layer_reshape</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_rms_normalization.html">layer_rms_normalization</a>()</code> <br><code><a href="layer_rnn.html">layer_rnn</a>()</code> <br><code><a href="layer_separable_conv_1d.html">layer_separable_conv_1d</a>()</code> <br><code><a href="layer_separable_conv_2d.html">layer_separable_conv_2d</a>()</code> <br><code><a href="layer_simple_rnn.html">layer_simple_rnn</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br><code><a href="layer_spatial_dropout_1d.html">layer_spatial_dropout_1d</a>()</code> <br><code><a href="layer_spatial_dropout_2d.html">layer_spatial_dropout_2d</a>()</code> <br><code><a href="layer_spatial_dropout_3d.html">layer_spatial_dropout_3d</a>()</code> <br><code><a href="layer_spectral_normalization.html">layer_spectral_normalization</a>()</code> <br><code><a href="layer_stft_spectrogram.html">layer_stft_spectrogram</a>()</code> <br><code><a href="layer_string_lookup.html">layer_string_lookup</a>()</code> <br><code><a href="layer_subtract.html">layer_subtract</a>()</code> <br><code><a href="layer_text_vectorization.html">layer_text_vectorization</a>()</code> <br><code><a href="layer_tfsm.html">layer_tfsm</a>()</code> <br><code><a href="layer_time_distributed.html">layer_time_distributed</a>()</code> <br><code><a href="layer_torch_module_wrapper.html">layer_torch_module_wrapper</a>()</code> <br><code><a href="layer_unit_normalization.html">layer_unit_normalization</a>()</code> <br><code><a href="layer_upsampling_1d.html">layer_upsampling_1d</a>()</code> <br><code><a href="layer_upsampling_2d.html">layer_upsampling_2d</a>()</code> <br><code><a href="layer_upsampling_3d.html">layer_upsampling_3d</a>()</code> <br><code><a href="layer_zero_padding_1d.html">layer_zero_padding_1d</a>()</code> <br><code><a href="layer_zero_padding_2d.html">layer_zero_padding_2d</a>()</code> <br><code><a href="layer_zero_padding_3d.html">layer_zero_padding_3d</a>()</code> <br><code><a href="rnn_cell_gru.html">rnn_cell_gru</a>()</code> <br><code><a href="rnn_cell_lstm.html">rnn_cell_lstm</a>()</code> <br><code><a href="rnn_cell_simple.html">rnn_cell_simple</a>()</code> <br><code><a href="rnn_cells_stack.html">rnn_cells_stack</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -507,7 +507,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/layer_jax_model_wrapper.html b/docs/dev/reference/layer_jax_model_wrapper.html
index 40d1ffd73..b0556b2b2 100644
--- a/docs/dev/reference/layer_jax_model_wrapper.html
+++ b/docs/dev/reference/layer_jax_model_wrapper.html
@@ -10,7 +10,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -304,7 +304,7 @@ <h3 id="usage-with-haiku-modules">Usage with Haiku modules<a class="anchor" aria
     <div class="section level2">
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
     <div class="dont-index"><p>Other wrapping layers: <br><code><a href="layer_flax_module_wrapper.html">layer_flax_module_wrapper</a>()</code> <br><code><a href="layer_torch_module_wrapper.html">layer_torch_module_wrapper</a>()</code> <br></p>
-<p>Other layers: <br><code><a href="Layer.html">Layer</a>()</code> <br><code><a href="layer_activation.html">layer_activation</a>()</code> <br><code><a href="layer_activation_elu.html">layer_activation_elu</a>()</code> <br><code><a href="layer_activation_leaky_relu.html">layer_activation_leaky_relu</a>()</code> <br><code><a href="layer_activation_parametric_relu.html">layer_activation_parametric_relu</a>()</code> <br><code><a href="layer_activation_relu.html">layer_activation_relu</a>()</code> <br><code><a href="layer_activation_softmax.html">layer_activation_softmax</a>()</code> <br><code><a href="layer_activity_regularization.html">layer_activity_regularization</a>()</code> <br><code><a href="layer_add.html">layer_add</a>()</code> <br><code><a href="layer_additive_attention.html">layer_additive_attention</a>()</code> <br><code><a href="layer_alpha_dropout.html">layer_alpha_dropout</a>()</code> <br><code><a href="layer_attention.html">layer_attention</a>()</code> <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_average.html">layer_average</a>()</code> <br><code><a href="layer_average_pooling_1d.html">layer_average_pooling_1d</a>()</code> <br><code><a href="layer_average_pooling_2d.html">layer_average_pooling_2d</a>()</code> <br><code><a href="layer_average_pooling_3d.html">layer_average_pooling_3d</a>()</code> <br><code><a href="layer_batch_normalization.html">layer_batch_normalization</a>()</code> <br><code><a href="layer_bidirectional.html">layer_bidirectional</a>()</code> <br><code><a href="layer_category_encoding.html">layer_category_encoding</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_concatenate.html">layer_concatenate</a>()</code> <br><code><a href="layer_conv_1d.html">layer_conv_1d</a>()</code> <br><code><a href="layer_conv_1d_transpose.html">layer_conv_1d_transpose</a>()</code> <br><code><a href="layer_conv_2d.html">layer_conv_2d</a>()</code> <br><code><a href="layer_conv_2d_transpose.html">layer_conv_2d_transpose</a>()</code> <br><code><a href="layer_conv_3d.html">layer_conv_3d</a>()</code> <br><code><a href="layer_conv_3d_transpose.html">layer_conv_3d_transpose</a>()</code> <br><code><a href="layer_conv_lstm_1d.html">layer_conv_lstm_1d</a>()</code> <br><code><a href="layer_conv_lstm_2d.html">layer_conv_lstm_2d</a>()</code> <br><code><a href="layer_conv_lstm_3d.html">layer_conv_lstm_3d</a>()</code> <br><code><a href="layer_cropping_1d.html">layer_cropping_1d</a>()</code> <br><code><a href="layer_cropping_2d.html">layer_cropping_2d</a>()</code> <br><code><a href="layer_cropping_3d.html">layer_cropping_3d</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_dense.html">layer_dense</a>()</code> <br><code><a href="layer_depthwise_conv_1d.html">layer_depthwise_conv_1d</a>()</code> <br><code><a href="layer_depthwise_conv_2d.html">layer_depthwise_conv_2d</a>()</code> <br><code><a href="layer_discretization.html">layer_discretization</a>()</code> <br><code><a href="layer_dot.html">layer_dot</a>()</code> <br><code><a href="layer_dropout.html">layer_dropout</a>()</code> <br><code><a href="layer_einsum_dense.html">layer_einsum_dense</a>()</code> <br><code><a href="layer_embedding.html">layer_embedding</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_feature_space.html">layer_feature_space</a>()</code> <br><code><a href="layer_flatten.html">layer_flatten</a>()</code> <br><code><a href="layer_flax_module_wrapper.html">layer_flax_module_wrapper</a>()</code> <br><code><a href="layer_gaussian_dropout.html">layer_gaussian_dropout</a>()</code> <br><code><a href="layer_gaussian_noise.html">layer_gaussian_noise</a>()</code> <br><code><a href="layer_global_average_pooling_1d.html">layer_global_average_pooling_1d</a>()</code> <br><code><a href="layer_global_average_pooling_2d.html">layer_global_average_pooling_2d</a>()</code> <br><code><a href="layer_global_average_pooling_3d.html">layer_global_average_pooling_3d</a>()</code> <br><code><a href="layer_global_max_pooling_1d.html">layer_global_max_pooling_1d</a>()</code> <br><code><a href="layer_global_max_pooling_2d.html">layer_global_max_pooling_2d</a>()</code> <br><code><a href="layer_global_max_pooling_3d.html">layer_global_max_pooling_3d</a>()</code> <br><code><a href="layer_group_normalization.html">layer_group_normalization</a>()</code> <br><code><a href="layer_group_query_attention.html">layer_group_query_attention</a>()</code> <br><code><a href="layer_gru.html">layer_gru</a>()</code> <br><code><a href="layer_hashed_crossing.html">layer_hashed_crossing</a>()</code> <br><code><a href="layer_hashing.html">layer_hashing</a>()</code> <br><code><a href="layer_identity.html">layer_identity</a>()</code> <br><code><a href="layer_integer_lookup.html">layer_integer_lookup</a>()</code> <br><code><a href="layer_lambda.html">layer_lambda</a>()</code> <br><code><a href="layer_layer_normalization.html">layer_layer_normalization</a>()</code> <br><code><a href="layer_lstm.html">layer_lstm</a>()</code> <br><code><a href="layer_masking.html">layer_masking</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_max_pooling_1d.html">layer_max_pooling_1d</a>()</code> <br><code><a href="layer_max_pooling_2d.html">layer_max_pooling_2d</a>()</code> <br><code><a href="layer_max_pooling_3d.html">layer_max_pooling_3d</a>()</code> <br><code><a href="layer_maximum.html">layer_maximum</a>()</code> <br><code><a href="layer_mel_spectrogram.html">layer_mel_spectrogram</a>()</code> <br><code><a href="layer_minimum.html">layer_minimum</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_multi_head_attention.html">layer_multi_head_attention</a>()</code> <br><code><a href="layer_multiply.html">layer_multiply</a>()</code> <br><code><a href="layer_normalization.html">layer_normalization</a>()</code> <br><code><a href="layer_permute.html">layer_permute</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_brightness.html">layer_random_brightness</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_contrast.html">layer_random_contrast</a>()</code> <br><code><a href="layer_random_crop.html">layer_random_crop</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_flip.html">layer_random_flip</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_rotation.html">layer_random_rotation</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_random_translation.html">layer_random_translation</a>()</code> <br><code><a href="layer_random_zoom.html">layer_random_zoom</a>()</code> <br><code><a href="layer_repeat_vector.html">layer_repeat_vector</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_reshape.html">layer_reshape</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_rms_normalization.html">layer_rms_normalization</a>()</code> <br><code><a href="layer_rnn.html">layer_rnn</a>()</code> <br><code><a href="layer_separable_conv_1d.html">layer_separable_conv_1d</a>()</code> <br><code><a href="layer_separable_conv_2d.html">layer_separable_conv_2d</a>()</code> <br><code><a href="layer_simple_rnn.html">layer_simple_rnn</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br><code><a href="layer_spatial_dropout_1d.html">layer_spatial_dropout_1d</a>()</code> <br><code><a href="layer_spatial_dropout_2d.html">layer_spatial_dropout_2d</a>()</code> <br><code><a href="layer_spatial_dropout_3d.html">layer_spatial_dropout_3d</a>()</code> <br><code><a href="layer_spectral_normalization.html">layer_spectral_normalization</a>()</code> <br><code><a href="layer_stft_spectrogram.html">layer_stft_spectrogram</a>()</code> <br><code><a href="layer_string_lookup.html">layer_string_lookup</a>()</code> <br><code><a href="layer_subtract.html">layer_subtract</a>()</code> <br><code><a href="layer_text_vectorization.html">layer_text_vectorization</a>()</code> <br><code><a href="layer_tfsm.html">layer_tfsm</a>()</code> <br><code><a href="layer_time_distributed.html">layer_time_distributed</a>()</code> <br><code><a href="layer_torch_module_wrapper.html">layer_torch_module_wrapper</a>()</code> <br><code><a href="layer_unit_normalization.html">layer_unit_normalization</a>()</code> <br><code><a href="layer_upsampling_1d.html">layer_upsampling_1d</a>()</code> <br><code><a href="layer_upsampling_2d.html">layer_upsampling_2d</a>()</code> <br><code><a href="layer_upsampling_3d.html">layer_upsampling_3d</a>()</code> <br><code><a href="layer_zero_padding_1d.html">layer_zero_padding_1d</a>()</code> <br><code><a href="layer_zero_padding_2d.html">layer_zero_padding_2d</a>()</code> <br><code><a href="layer_zero_padding_3d.html">layer_zero_padding_3d</a>()</code> <br><code><a href="rnn_cell_gru.html">rnn_cell_gru</a>()</code> <br><code><a href="rnn_cell_lstm.html">rnn_cell_lstm</a>()</code> <br><code><a href="rnn_cell_simple.html">rnn_cell_simple</a>()</code> <br><code><a href="rnn_cells_stack.html">rnn_cells_stack</a>()</code> <br></p></div>
+<p>Other layers: <br><code><a href="Layer.html">Layer</a>()</code> <br><code><a href="layer_activation.html">layer_activation</a>()</code> <br><code><a href="layer_activation_elu.html">layer_activation_elu</a>()</code> <br><code><a href="layer_activation_leaky_relu.html">layer_activation_leaky_relu</a>()</code> <br><code><a href="layer_activation_parametric_relu.html">layer_activation_parametric_relu</a>()</code> <br><code><a href="layer_activation_relu.html">layer_activation_relu</a>()</code> <br><code><a href="layer_activation_softmax.html">layer_activation_softmax</a>()</code> <br><code><a href="layer_activity_regularization.html">layer_activity_regularization</a>()</code> <br><code><a href="layer_add.html">layer_add</a>()</code> <br><code><a href="layer_additive_attention.html">layer_additive_attention</a>()</code> <br><code><a href="layer_alpha_dropout.html">layer_alpha_dropout</a>()</code> <br><code><a href="layer_attention.html">layer_attention</a>()</code> <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_average.html">layer_average</a>()</code> <br><code><a href="layer_average_pooling_1d.html">layer_average_pooling_1d</a>()</code> <br><code><a href="layer_average_pooling_2d.html">layer_average_pooling_2d</a>()</code> <br><code><a href="layer_average_pooling_3d.html">layer_average_pooling_3d</a>()</code> <br><code><a href="layer_batch_normalization.html">layer_batch_normalization</a>()</code> <br><code><a href="layer_bidirectional.html">layer_bidirectional</a>()</code> <br><code><a href="layer_category_encoding.html">layer_category_encoding</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_concatenate.html">layer_concatenate</a>()</code> <br><code><a href="layer_conv_1d.html">layer_conv_1d</a>()</code> <br><code><a href="layer_conv_1d_transpose.html">layer_conv_1d_transpose</a>()</code> <br><code><a href="layer_conv_2d.html">layer_conv_2d</a>()</code> <br><code><a href="layer_conv_2d_transpose.html">layer_conv_2d_transpose</a>()</code> <br><code><a href="layer_conv_3d.html">layer_conv_3d</a>()</code> <br><code><a href="layer_conv_3d_transpose.html">layer_conv_3d_transpose</a>()</code> <br><code><a href="layer_conv_lstm_1d.html">layer_conv_lstm_1d</a>()</code> <br><code><a href="layer_conv_lstm_2d.html">layer_conv_lstm_2d</a>()</code> <br><code><a href="layer_conv_lstm_3d.html">layer_conv_lstm_3d</a>()</code> <br><code><a href="layer_cropping_1d.html">layer_cropping_1d</a>()</code> <br><code><a href="layer_cropping_2d.html">layer_cropping_2d</a>()</code> <br><code><a href="layer_cropping_3d.html">layer_cropping_3d</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_dense.html">layer_dense</a>()</code> <br><code><a href="layer_depthwise_conv_1d.html">layer_depthwise_conv_1d</a>()</code> <br><code><a href="layer_depthwise_conv_2d.html">layer_depthwise_conv_2d</a>()</code> <br><code><a href="layer_discretization.html">layer_discretization</a>()</code> <br><code><a href="layer_dot.html">layer_dot</a>()</code> <br><code><a href="layer_dropout.html">layer_dropout</a>()</code> <br><code><a href="layer_einsum_dense.html">layer_einsum_dense</a>()</code> <br><code><a href="layer_embedding.html">layer_embedding</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_feature_space.html">layer_feature_space</a>()</code> <br><code><a href="layer_flatten.html">layer_flatten</a>()</code> <br><code><a href="layer_flax_module_wrapper.html">layer_flax_module_wrapper</a>()</code> <br><code><a href="layer_gaussian_dropout.html">layer_gaussian_dropout</a>()</code> <br><code><a href="layer_gaussian_noise.html">layer_gaussian_noise</a>()</code> <br><code><a href="layer_global_average_pooling_1d.html">layer_global_average_pooling_1d</a>()</code> <br><code><a href="layer_global_average_pooling_2d.html">layer_global_average_pooling_2d</a>()</code> <br><code><a href="layer_global_average_pooling_3d.html">layer_global_average_pooling_3d</a>()</code> <br><code><a href="layer_global_max_pooling_1d.html">layer_global_max_pooling_1d</a>()</code> <br><code><a href="layer_global_max_pooling_2d.html">layer_global_max_pooling_2d</a>()</code> <br><code><a href="layer_global_max_pooling_3d.html">layer_global_max_pooling_3d</a>()</code> <br><code><a href="layer_group_normalization.html">layer_group_normalization</a>()</code> <br><code><a href="layer_group_query_attention.html">layer_group_query_attention</a>()</code> <br><code><a href="layer_gru.html">layer_gru</a>()</code> <br><code><a href="layer_hashed_crossing.html">layer_hashed_crossing</a>()</code> <br><code><a href="layer_hashing.html">layer_hashing</a>()</code> <br><code><a href="layer_identity.html">layer_identity</a>()</code> <br><code><a href="layer_integer_lookup.html">layer_integer_lookup</a>()</code> <br><code><a href="layer_lambda.html">layer_lambda</a>()</code> <br><code><a href="layer_layer_normalization.html">layer_layer_normalization</a>()</code> <br><code><a href="layer_lstm.html">layer_lstm</a>()</code> <br><code><a href="layer_masking.html">layer_masking</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_max_pooling_1d.html">layer_max_pooling_1d</a>()</code> <br><code><a href="layer_max_pooling_2d.html">layer_max_pooling_2d</a>()</code> <br><code><a href="layer_max_pooling_3d.html">layer_max_pooling_3d</a>()</code> <br><code><a href="layer_maximum.html">layer_maximum</a>()</code> <br><code><a href="layer_mel_spectrogram.html">layer_mel_spectrogram</a>()</code> <br><code><a href="layer_minimum.html">layer_minimum</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_multi_head_attention.html">layer_multi_head_attention</a>()</code> <br><code><a href="layer_multiply.html">layer_multiply</a>()</code> <br><code><a href="layer_normalization.html">layer_normalization</a>()</code> <br><code><a href="layer_permute.html">layer_permute</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_brightness.html">layer_random_brightness</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_contrast.html">layer_random_contrast</a>()</code> <br><code><a href="layer_random_crop.html">layer_random_crop</a>()</code> <br><code><a href="layer_random_elastic_transform.html">layer_random_elastic_transform</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_flip.html">layer_random_flip</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_rotation.html">layer_random_rotation</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_random_translation.html">layer_random_translation</a>()</code> <br><code><a href="layer_random_zoom.html">layer_random_zoom</a>()</code> <br><code><a href="layer_repeat_vector.html">layer_repeat_vector</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_reshape.html">layer_reshape</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_rms_normalization.html">layer_rms_normalization</a>()</code> <br><code><a href="layer_rnn.html">layer_rnn</a>()</code> <br><code><a href="layer_separable_conv_1d.html">layer_separable_conv_1d</a>()</code> <br><code><a href="layer_separable_conv_2d.html">layer_separable_conv_2d</a>()</code> <br><code><a href="layer_simple_rnn.html">layer_simple_rnn</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br><code><a href="layer_spatial_dropout_1d.html">layer_spatial_dropout_1d</a>()</code> <br><code><a href="layer_spatial_dropout_2d.html">layer_spatial_dropout_2d</a>()</code> <br><code><a href="layer_spatial_dropout_3d.html">layer_spatial_dropout_3d</a>()</code> <br><code><a href="layer_spectral_normalization.html">layer_spectral_normalization</a>()</code> <br><code><a href="layer_stft_spectrogram.html">layer_stft_spectrogram</a>()</code> <br><code><a href="layer_string_lookup.html">layer_string_lookup</a>()</code> <br><code><a href="layer_subtract.html">layer_subtract</a>()</code> <br><code><a href="layer_text_vectorization.html">layer_text_vectorization</a>()</code> <br><code><a href="layer_tfsm.html">layer_tfsm</a>()</code> <br><code><a href="layer_time_distributed.html">layer_time_distributed</a>()</code> <br><code><a href="layer_torch_module_wrapper.html">layer_torch_module_wrapper</a>()</code> <br><code><a href="layer_unit_normalization.html">layer_unit_normalization</a>()</code> <br><code><a href="layer_upsampling_1d.html">layer_upsampling_1d</a>()</code> <br><code><a href="layer_upsampling_2d.html">layer_upsampling_2d</a>()</code> <br><code><a href="layer_upsampling_3d.html">layer_upsampling_3d</a>()</code> <br><code><a href="layer_zero_padding_1d.html">layer_zero_padding_1d</a>()</code> <br><code><a href="layer_zero_padding_2d.html">layer_zero_padding_2d</a>()</code> <br><code><a href="layer_zero_padding_3d.html">layer_zero_padding_3d</a>()</code> <br><code><a href="rnn_cell_gru.html">rnn_cell_gru</a>()</code> <br><code><a href="rnn_cell_lstm.html">rnn_cell_lstm</a>()</code> <br><code><a href="rnn_cell_simple.html">rnn_cell_simple</a>()</code> <br><code><a href="rnn_cells_stack.html">rnn_cells_stack</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -316,7 +316,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/layer_lambda.html b/docs/dev/reference/layer_lambda.html
index b458c111b..8622e23be 100644
--- a/docs/dev/reference/layer_lambda.html
+++ b/docs/dev/reference/layer_lambda.html
@@ -16,7 +16,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -144,7 +144,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
     <div class="dont-index">
 <ul><li><p><a href="https://keras.io/api/layers/core_layers/lambda#lambda-class" class="external-link">https://keras.io/api/layers/core_layers/lambda#lambda-class</a></p></li>
 </ul><p>Other core layers: <br><code><a href="layer_dense.html">layer_dense</a>()</code> <br><code><a href="layer_einsum_dense.html">layer_einsum_dense</a>()</code> <br><code><a href="layer_embedding.html">layer_embedding</a>()</code> <br><code><a href="layer_identity.html">layer_identity</a>()</code> <br><code><a href="layer_masking.html">layer_masking</a>()</code> <br></p>
-<p>Other layers: <br><code><a href="Layer.html">Layer</a>()</code> <br><code><a href="layer_activation.html">layer_activation</a>()</code> <br><code><a href="layer_activation_elu.html">layer_activation_elu</a>()</code> <br><code><a href="layer_activation_leaky_relu.html">layer_activation_leaky_relu</a>()</code> <br><code><a href="layer_activation_parametric_relu.html">layer_activation_parametric_relu</a>()</code> <br><code><a href="layer_activation_relu.html">layer_activation_relu</a>()</code> <br><code><a href="layer_activation_softmax.html">layer_activation_softmax</a>()</code> <br><code><a href="layer_activity_regularization.html">layer_activity_regularization</a>()</code> <br><code><a href="layer_add.html">layer_add</a>()</code> <br><code><a href="layer_additive_attention.html">layer_additive_attention</a>()</code> <br><code><a href="layer_alpha_dropout.html">layer_alpha_dropout</a>()</code> <br><code><a href="layer_attention.html">layer_attention</a>()</code> <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_average.html">layer_average</a>()</code> <br><code><a href="layer_average_pooling_1d.html">layer_average_pooling_1d</a>()</code> <br><code><a href="layer_average_pooling_2d.html">layer_average_pooling_2d</a>()</code> <br><code><a href="layer_average_pooling_3d.html">layer_average_pooling_3d</a>()</code> <br><code><a href="layer_batch_normalization.html">layer_batch_normalization</a>()</code> <br><code><a href="layer_bidirectional.html">layer_bidirectional</a>()</code> <br><code><a href="layer_category_encoding.html">layer_category_encoding</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_concatenate.html">layer_concatenate</a>()</code> <br><code><a href="layer_conv_1d.html">layer_conv_1d</a>()</code> <br><code><a href="layer_conv_1d_transpose.html">layer_conv_1d_transpose</a>()</code> <br><code><a href="layer_conv_2d.html">layer_conv_2d</a>()</code> <br><code><a href="layer_conv_2d_transpose.html">layer_conv_2d_transpose</a>()</code> <br><code><a href="layer_conv_3d.html">layer_conv_3d</a>()</code> <br><code><a href="layer_conv_3d_transpose.html">layer_conv_3d_transpose</a>()</code> <br><code><a href="layer_conv_lstm_1d.html">layer_conv_lstm_1d</a>()</code> <br><code><a href="layer_conv_lstm_2d.html">layer_conv_lstm_2d</a>()</code> <br><code><a href="layer_conv_lstm_3d.html">layer_conv_lstm_3d</a>()</code> <br><code><a href="layer_cropping_1d.html">layer_cropping_1d</a>()</code> <br><code><a href="layer_cropping_2d.html">layer_cropping_2d</a>()</code> <br><code><a href="layer_cropping_3d.html">layer_cropping_3d</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_dense.html">layer_dense</a>()</code> <br><code><a href="layer_depthwise_conv_1d.html">layer_depthwise_conv_1d</a>()</code> <br><code><a href="layer_depthwise_conv_2d.html">layer_depthwise_conv_2d</a>()</code> <br><code><a href="layer_discretization.html">layer_discretization</a>()</code> <br><code><a href="layer_dot.html">layer_dot</a>()</code> <br><code><a href="layer_dropout.html">layer_dropout</a>()</code> <br><code><a href="layer_einsum_dense.html">layer_einsum_dense</a>()</code> <br><code><a href="layer_embedding.html">layer_embedding</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_feature_space.html">layer_feature_space</a>()</code> <br><code><a href="layer_flatten.html">layer_flatten</a>()</code> <br><code><a href="layer_flax_module_wrapper.html">layer_flax_module_wrapper</a>()</code> <br><code><a href="layer_gaussian_dropout.html">layer_gaussian_dropout</a>()</code> <br><code><a href="layer_gaussian_noise.html">layer_gaussian_noise</a>()</code> <br><code><a href="layer_global_average_pooling_1d.html">layer_global_average_pooling_1d</a>()</code> <br><code><a href="layer_global_average_pooling_2d.html">layer_global_average_pooling_2d</a>()</code> <br><code><a href="layer_global_average_pooling_3d.html">layer_global_average_pooling_3d</a>()</code> <br><code><a href="layer_global_max_pooling_1d.html">layer_global_max_pooling_1d</a>()</code> <br><code><a href="layer_global_max_pooling_2d.html">layer_global_max_pooling_2d</a>()</code> <br><code><a href="layer_global_max_pooling_3d.html">layer_global_max_pooling_3d</a>()</code> <br><code><a href="layer_group_normalization.html">layer_group_normalization</a>()</code> <br><code><a href="layer_group_query_attention.html">layer_group_query_attention</a>()</code> <br><code><a href="layer_gru.html">layer_gru</a>()</code> <br><code><a href="layer_hashed_crossing.html">layer_hashed_crossing</a>()</code> <br><code><a href="layer_hashing.html">layer_hashing</a>()</code> <br><code><a href="layer_identity.html">layer_identity</a>()</code> <br><code><a href="layer_integer_lookup.html">layer_integer_lookup</a>()</code> <br><code><a href="layer_jax_model_wrapper.html">layer_jax_model_wrapper</a>()</code> <br><code><a href="layer_layer_normalization.html">layer_layer_normalization</a>()</code> <br><code><a href="layer_lstm.html">layer_lstm</a>()</code> <br><code><a href="layer_masking.html">layer_masking</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_max_pooling_1d.html">layer_max_pooling_1d</a>()</code> <br><code><a href="layer_max_pooling_2d.html">layer_max_pooling_2d</a>()</code> <br><code><a href="layer_max_pooling_3d.html">layer_max_pooling_3d</a>()</code> <br><code><a href="layer_maximum.html">layer_maximum</a>()</code> <br><code><a href="layer_mel_spectrogram.html">layer_mel_spectrogram</a>()</code> <br><code><a href="layer_minimum.html">layer_minimum</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_multi_head_attention.html">layer_multi_head_attention</a>()</code> <br><code><a href="layer_multiply.html">layer_multiply</a>()</code> <br><code><a href="layer_normalization.html">layer_normalization</a>()</code> <br><code><a href="layer_permute.html">layer_permute</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_brightness.html">layer_random_brightness</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_contrast.html">layer_random_contrast</a>()</code> <br><code><a href="layer_random_crop.html">layer_random_crop</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_flip.html">layer_random_flip</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_rotation.html">layer_random_rotation</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_random_translation.html">layer_random_translation</a>()</code> <br><code><a href="layer_random_zoom.html">layer_random_zoom</a>()</code> <br><code><a href="layer_repeat_vector.html">layer_repeat_vector</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_reshape.html">layer_reshape</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_rms_normalization.html">layer_rms_normalization</a>()</code> <br><code><a href="layer_rnn.html">layer_rnn</a>()</code> <br><code><a href="layer_separable_conv_1d.html">layer_separable_conv_1d</a>()</code> <br><code><a href="layer_separable_conv_2d.html">layer_separable_conv_2d</a>()</code> <br><code><a href="layer_simple_rnn.html">layer_simple_rnn</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br><code><a href="layer_spatial_dropout_1d.html">layer_spatial_dropout_1d</a>()</code> <br><code><a href="layer_spatial_dropout_2d.html">layer_spatial_dropout_2d</a>()</code> <br><code><a href="layer_spatial_dropout_3d.html">layer_spatial_dropout_3d</a>()</code> <br><code><a href="layer_spectral_normalization.html">layer_spectral_normalization</a>()</code> <br><code><a href="layer_stft_spectrogram.html">layer_stft_spectrogram</a>()</code> <br><code><a href="layer_string_lookup.html">layer_string_lookup</a>()</code> <br><code><a href="layer_subtract.html">layer_subtract</a>()</code> <br><code><a href="layer_text_vectorization.html">layer_text_vectorization</a>()</code> <br><code><a href="layer_tfsm.html">layer_tfsm</a>()</code> <br><code><a href="layer_time_distributed.html">layer_time_distributed</a>()</code> <br><code><a href="layer_torch_module_wrapper.html">layer_torch_module_wrapper</a>()</code> <br><code><a href="layer_unit_normalization.html">layer_unit_normalization</a>()</code> <br><code><a href="layer_upsampling_1d.html">layer_upsampling_1d</a>()</code> <br><code><a href="layer_upsampling_2d.html">layer_upsampling_2d</a>()</code> <br><code><a href="layer_upsampling_3d.html">layer_upsampling_3d</a>()</code> <br><code><a href="layer_zero_padding_1d.html">layer_zero_padding_1d</a>()</code> <br><code><a href="layer_zero_padding_2d.html">layer_zero_padding_2d</a>()</code> <br><code><a href="layer_zero_padding_3d.html">layer_zero_padding_3d</a>()</code> <br><code><a href="rnn_cell_gru.html">rnn_cell_gru</a>()</code> <br><code><a href="rnn_cell_lstm.html">rnn_cell_lstm</a>()</code> <br><code><a href="rnn_cell_simple.html">rnn_cell_simple</a>()</code> <br><code><a href="rnn_cells_stack.html">rnn_cells_stack</a>()</code> <br></p></div>
+<p>Other layers: <br><code><a href="Layer.html">Layer</a>()</code> <br><code><a href="layer_activation.html">layer_activation</a>()</code> <br><code><a href="layer_activation_elu.html">layer_activation_elu</a>()</code> <br><code><a href="layer_activation_leaky_relu.html">layer_activation_leaky_relu</a>()</code> <br><code><a href="layer_activation_parametric_relu.html">layer_activation_parametric_relu</a>()</code> <br><code><a href="layer_activation_relu.html">layer_activation_relu</a>()</code> <br><code><a href="layer_activation_softmax.html">layer_activation_softmax</a>()</code> <br><code><a href="layer_activity_regularization.html">layer_activity_regularization</a>()</code> <br><code><a href="layer_add.html">layer_add</a>()</code> <br><code><a href="layer_additive_attention.html">layer_additive_attention</a>()</code> <br><code><a href="layer_alpha_dropout.html">layer_alpha_dropout</a>()</code> <br><code><a href="layer_attention.html">layer_attention</a>()</code> <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_average.html">layer_average</a>()</code> <br><code><a href="layer_average_pooling_1d.html">layer_average_pooling_1d</a>()</code> <br><code><a href="layer_average_pooling_2d.html">layer_average_pooling_2d</a>()</code> <br><code><a href="layer_average_pooling_3d.html">layer_average_pooling_3d</a>()</code> <br><code><a href="layer_batch_normalization.html">layer_batch_normalization</a>()</code> <br><code><a href="layer_bidirectional.html">layer_bidirectional</a>()</code> <br><code><a href="layer_category_encoding.html">layer_category_encoding</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_concatenate.html">layer_concatenate</a>()</code> <br><code><a href="layer_conv_1d.html">layer_conv_1d</a>()</code> <br><code><a href="layer_conv_1d_transpose.html">layer_conv_1d_transpose</a>()</code> <br><code><a href="layer_conv_2d.html">layer_conv_2d</a>()</code> <br><code><a href="layer_conv_2d_transpose.html">layer_conv_2d_transpose</a>()</code> <br><code><a href="layer_conv_3d.html">layer_conv_3d</a>()</code> <br><code><a href="layer_conv_3d_transpose.html">layer_conv_3d_transpose</a>()</code> <br><code><a href="layer_conv_lstm_1d.html">layer_conv_lstm_1d</a>()</code> <br><code><a href="layer_conv_lstm_2d.html">layer_conv_lstm_2d</a>()</code> <br><code><a href="layer_conv_lstm_3d.html">layer_conv_lstm_3d</a>()</code> <br><code><a href="layer_cropping_1d.html">layer_cropping_1d</a>()</code> <br><code><a href="layer_cropping_2d.html">layer_cropping_2d</a>()</code> <br><code><a href="layer_cropping_3d.html">layer_cropping_3d</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_dense.html">layer_dense</a>()</code> <br><code><a href="layer_depthwise_conv_1d.html">layer_depthwise_conv_1d</a>()</code> <br><code><a href="layer_depthwise_conv_2d.html">layer_depthwise_conv_2d</a>()</code> <br><code><a href="layer_discretization.html">layer_discretization</a>()</code> <br><code><a href="layer_dot.html">layer_dot</a>()</code> <br><code><a href="layer_dropout.html">layer_dropout</a>()</code> <br><code><a href="layer_einsum_dense.html">layer_einsum_dense</a>()</code> <br><code><a href="layer_embedding.html">layer_embedding</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_feature_space.html">layer_feature_space</a>()</code> <br><code><a href="layer_flatten.html">layer_flatten</a>()</code> <br><code><a href="layer_flax_module_wrapper.html">layer_flax_module_wrapper</a>()</code> <br><code><a href="layer_gaussian_dropout.html">layer_gaussian_dropout</a>()</code> <br><code><a href="layer_gaussian_noise.html">layer_gaussian_noise</a>()</code> <br><code><a href="layer_global_average_pooling_1d.html">layer_global_average_pooling_1d</a>()</code> <br><code><a href="layer_global_average_pooling_2d.html">layer_global_average_pooling_2d</a>()</code> <br><code><a href="layer_global_average_pooling_3d.html">layer_global_average_pooling_3d</a>()</code> <br><code><a href="layer_global_max_pooling_1d.html">layer_global_max_pooling_1d</a>()</code> <br><code><a href="layer_global_max_pooling_2d.html">layer_global_max_pooling_2d</a>()</code> <br><code><a href="layer_global_max_pooling_3d.html">layer_global_max_pooling_3d</a>()</code> <br><code><a href="layer_group_normalization.html">layer_group_normalization</a>()</code> <br><code><a href="layer_group_query_attention.html">layer_group_query_attention</a>()</code> <br><code><a href="layer_gru.html">layer_gru</a>()</code> <br><code><a href="layer_hashed_crossing.html">layer_hashed_crossing</a>()</code> <br><code><a href="layer_hashing.html">layer_hashing</a>()</code> <br><code><a href="layer_identity.html">layer_identity</a>()</code> <br><code><a href="layer_integer_lookup.html">layer_integer_lookup</a>()</code> <br><code><a href="layer_jax_model_wrapper.html">layer_jax_model_wrapper</a>()</code> <br><code><a href="layer_layer_normalization.html">layer_layer_normalization</a>()</code> <br><code><a href="layer_lstm.html">layer_lstm</a>()</code> <br><code><a href="layer_masking.html">layer_masking</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_max_pooling_1d.html">layer_max_pooling_1d</a>()</code> <br><code><a href="layer_max_pooling_2d.html">layer_max_pooling_2d</a>()</code> <br><code><a href="layer_max_pooling_3d.html">layer_max_pooling_3d</a>()</code> <br><code><a href="layer_maximum.html">layer_maximum</a>()</code> <br><code><a href="layer_mel_spectrogram.html">layer_mel_spectrogram</a>()</code> <br><code><a href="layer_minimum.html">layer_minimum</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_multi_head_attention.html">layer_multi_head_attention</a>()</code> <br><code><a href="layer_multiply.html">layer_multiply</a>()</code> <br><code><a href="layer_normalization.html">layer_normalization</a>()</code> <br><code><a href="layer_permute.html">layer_permute</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_brightness.html">layer_random_brightness</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_contrast.html">layer_random_contrast</a>()</code> <br><code><a href="layer_random_crop.html">layer_random_crop</a>()</code> <br><code><a href="layer_random_elastic_transform.html">layer_random_elastic_transform</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_flip.html">layer_random_flip</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_rotation.html">layer_random_rotation</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_random_translation.html">layer_random_translation</a>()</code> <br><code><a href="layer_random_zoom.html">layer_random_zoom</a>()</code> <br><code><a href="layer_repeat_vector.html">layer_repeat_vector</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_reshape.html">layer_reshape</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_rms_normalization.html">layer_rms_normalization</a>()</code> <br><code><a href="layer_rnn.html">layer_rnn</a>()</code> <br><code><a href="layer_separable_conv_1d.html">layer_separable_conv_1d</a>()</code> <br><code><a href="layer_separable_conv_2d.html">layer_separable_conv_2d</a>()</code> <br><code><a href="layer_simple_rnn.html">layer_simple_rnn</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br><code><a href="layer_spatial_dropout_1d.html">layer_spatial_dropout_1d</a>()</code> <br><code><a href="layer_spatial_dropout_2d.html">layer_spatial_dropout_2d</a>()</code> <br><code><a href="layer_spatial_dropout_3d.html">layer_spatial_dropout_3d</a>()</code> <br><code><a href="layer_spectral_normalization.html">layer_spectral_normalization</a>()</code> <br><code><a href="layer_stft_spectrogram.html">layer_stft_spectrogram</a>()</code> <br><code><a href="layer_string_lookup.html">layer_string_lookup</a>()</code> <br><code><a href="layer_subtract.html">layer_subtract</a>()</code> <br><code><a href="layer_text_vectorization.html">layer_text_vectorization</a>()</code> <br><code><a href="layer_tfsm.html">layer_tfsm</a>()</code> <br><code><a href="layer_time_distributed.html">layer_time_distributed</a>()</code> <br><code><a href="layer_torch_module_wrapper.html">layer_torch_module_wrapper</a>()</code> <br><code><a href="layer_unit_normalization.html">layer_unit_normalization</a>()</code> <br><code><a href="layer_upsampling_1d.html">layer_upsampling_1d</a>()</code> <br><code><a href="layer_upsampling_2d.html">layer_upsampling_2d</a>()</code> <br><code><a href="layer_upsampling_3d.html">layer_upsampling_3d</a>()</code> <br><code><a href="layer_zero_padding_1d.html">layer_zero_padding_1d</a>()</code> <br><code><a href="layer_zero_padding_2d.html">layer_zero_padding_2d</a>()</code> <br><code><a href="layer_zero_padding_3d.html">layer_zero_padding_3d</a>()</code> <br><code><a href="rnn_cell_gru.html">rnn_cell_gru</a>()</code> <br><code><a href="rnn_cell_lstm.html">rnn_cell_lstm</a>()</code> <br><code><a href="rnn_cell_simple.html">rnn_cell_simple</a>()</code> <br><code><a href="rnn_cells_stack.html">rnn_cells_stack</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -156,7 +156,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/layer_layer_normalization.html b/docs/dev/reference/layer_layer_normalization.html
index 994c2aa3b..5a1923ed6 100644
--- a/docs/dev/reference/layer_layer_normalization.html
+++ b/docs/dev/reference/layer_layer_normalization.html
@@ -102,7 +102,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -201,7 +201,6 @@ <h2 id="ref-usage">Usage<a class="anchor" aria-label="anchor" href="#ref-usage">
 <span>  epsilon <span class="op">=</span> <span class="fl">0.001</span>,</span>
 <span>  center <span class="op">=</span> <span class="cn">TRUE</span>,</span>
 <span>  scale <span class="op">=</span> <span class="cn">TRUE</span>,</span>
-<span>  rms_scaling <span class="op">=</span> <span class="cn">FALSE</span>,</span>
 <span>  beta_initializer <span class="op">=</span> <span class="st">"zeros"</span>,</span>
 <span>  gamma_initializer <span class="op">=</span> <span class="st">"ones"</span>,</span>
 <span>  beta_regularizer <span class="op">=</span> <span class="cn">NULL</span>,</span>
@@ -244,15 +243,6 @@ <h2 id="arguments">Arguments<a class="anchor" aria-label="anchor" href="#argumen
 Defaults to <code>TRUE</code>.</p></dd>
 
 
-<dt id="arg-rms-scaling">rms_scaling<a class="anchor" aria-label="anchor" href="#arg-rms-scaling"></a></dt>
-<dd><p>If <code>TRUE</code>, <code>center</code> and <code>scale</code> are ignored, and the
-inputs are scaled by <code>gamma</code> and the inverse square root
-of the square of all inputs. This is an approximate and faster
-approach that avoids ever computing the mean of the input. Note that
-this <em>isn't</em> equivalent to the computation that the
-<code>layer_rms_normalization</code> layer performs.</p></dd>
-
-
 <dt id="arg-beta-initializer">beta_initializer<a class="anchor" aria-label="anchor" href="#arg-beta-initializer"></a></dt>
 <dd><p>Initializer for the beta weight. Defaults to zeros.</p></dd>
 
@@ -304,7 +294,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
     <div class="dont-index">
 <ul><li><p><a href="https://keras.io/api/layers/normalization_layers/layer_normalization#layernormalization-class" class="external-link">https://keras.io/api/layers/normalization_layers/layer_normalization#layernormalization-class</a></p></li>
 </ul><p>Other normalization layers: <br><code><a href="layer_batch_normalization.html">layer_batch_normalization</a>()</code> <br><code><a href="layer_group_normalization.html">layer_group_normalization</a>()</code> <br><code><a href="layer_rms_normalization.html">layer_rms_normalization</a>()</code> <br><code><a href="layer_spectral_normalization.html">layer_spectral_normalization</a>()</code> <br><code><a href="layer_unit_normalization.html">layer_unit_normalization</a>()</code> <br></p>
-<p>Other layers: <br><code><a href="Layer.html">Layer</a>()</code> <br><code><a href="layer_activation.html">layer_activation</a>()</code> <br><code><a href="layer_activation_elu.html">layer_activation_elu</a>()</code> <br><code><a href="layer_activation_leaky_relu.html">layer_activation_leaky_relu</a>()</code> <br><code><a href="layer_activation_parametric_relu.html">layer_activation_parametric_relu</a>()</code> <br><code><a href="layer_activation_relu.html">layer_activation_relu</a>()</code> <br><code><a href="layer_activation_softmax.html">layer_activation_softmax</a>()</code> <br><code><a href="layer_activity_regularization.html">layer_activity_regularization</a>()</code> <br><code><a href="layer_add.html">layer_add</a>()</code> <br><code><a href="layer_additive_attention.html">layer_additive_attention</a>()</code> <br><code><a href="layer_alpha_dropout.html">layer_alpha_dropout</a>()</code> <br><code><a href="layer_attention.html">layer_attention</a>()</code> <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_average.html">layer_average</a>()</code> <br><code><a href="layer_average_pooling_1d.html">layer_average_pooling_1d</a>()</code> <br><code><a href="layer_average_pooling_2d.html">layer_average_pooling_2d</a>()</code> <br><code><a href="layer_average_pooling_3d.html">layer_average_pooling_3d</a>()</code> <br><code><a href="layer_batch_normalization.html">layer_batch_normalization</a>()</code> <br><code><a href="layer_bidirectional.html">layer_bidirectional</a>()</code> <br><code><a href="layer_category_encoding.html">layer_category_encoding</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_concatenate.html">layer_concatenate</a>()</code> <br><code><a href="layer_conv_1d.html">layer_conv_1d</a>()</code> <br><code><a href="layer_conv_1d_transpose.html">layer_conv_1d_transpose</a>()</code> <br><code><a href="layer_conv_2d.html">layer_conv_2d</a>()</code> <br><code><a href="layer_conv_2d_transpose.html">layer_conv_2d_transpose</a>()</code> <br><code><a href="layer_conv_3d.html">layer_conv_3d</a>()</code> <br><code><a href="layer_conv_3d_transpose.html">layer_conv_3d_transpose</a>()</code> <br><code><a href="layer_conv_lstm_1d.html">layer_conv_lstm_1d</a>()</code> <br><code><a href="layer_conv_lstm_2d.html">layer_conv_lstm_2d</a>()</code> <br><code><a href="layer_conv_lstm_3d.html">layer_conv_lstm_3d</a>()</code> <br><code><a href="layer_cropping_1d.html">layer_cropping_1d</a>()</code> <br><code><a href="layer_cropping_2d.html">layer_cropping_2d</a>()</code> <br><code><a href="layer_cropping_3d.html">layer_cropping_3d</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_dense.html">layer_dense</a>()</code> <br><code><a href="layer_depthwise_conv_1d.html">layer_depthwise_conv_1d</a>()</code> <br><code><a href="layer_depthwise_conv_2d.html">layer_depthwise_conv_2d</a>()</code> <br><code><a href="layer_discretization.html">layer_discretization</a>()</code> <br><code><a href="layer_dot.html">layer_dot</a>()</code> <br><code><a href="layer_dropout.html">layer_dropout</a>()</code> <br><code><a href="layer_einsum_dense.html">layer_einsum_dense</a>()</code> <br><code><a href="layer_embedding.html">layer_embedding</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_feature_space.html">layer_feature_space</a>()</code> <br><code><a href="layer_flatten.html">layer_flatten</a>()</code> <br><code><a href="layer_flax_module_wrapper.html">layer_flax_module_wrapper</a>()</code> <br><code><a href="layer_gaussian_dropout.html">layer_gaussian_dropout</a>()</code> <br><code><a href="layer_gaussian_noise.html">layer_gaussian_noise</a>()</code> <br><code><a href="layer_global_average_pooling_1d.html">layer_global_average_pooling_1d</a>()</code> <br><code><a href="layer_global_average_pooling_2d.html">layer_global_average_pooling_2d</a>()</code> <br><code><a href="layer_global_average_pooling_3d.html">layer_global_average_pooling_3d</a>()</code> <br><code><a href="layer_global_max_pooling_1d.html">layer_global_max_pooling_1d</a>()</code> <br><code><a href="layer_global_max_pooling_2d.html">layer_global_max_pooling_2d</a>()</code> <br><code><a href="layer_global_max_pooling_3d.html">layer_global_max_pooling_3d</a>()</code> <br><code><a href="layer_group_normalization.html">layer_group_normalization</a>()</code> <br><code><a href="layer_group_query_attention.html">layer_group_query_attention</a>()</code> <br><code><a href="layer_gru.html">layer_gru</a>()</code> <br><code><a href="layer_hashed_crossing.html">layer_hashed_crossing</a>()</code> <br><code><a href="layer_hashing.html">layer_hashing</a>()</code> <br><code><a href="layer_identity.html">layer_identity</a>()</code> <br><code><a href="layer_integer_lookup.html">layer_integer_lookup</a>()</code> <br><code><a href="layer_jax_model_wrapper.html">layer_jax_model_wrapper</a>()</code> <br><code><a href="layer_lambda.html">layer_lambda</a>()</code> <br><code><a href="layer_lstm.html">layer_lstm</a>()</code> <br><code><a href="layer_masking.html">layer_masking</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_max_pooling_1d.html">layer_max_pooling_1d</a>()</code> <br><code><a href="layer_max_pooling_2d.html">layer_max_pooling_2d</a>()</code> <br><code><a href="layer_max_pooling_3d.html">layer_max_pooling_3d</a>()</code> <br><code><a href="layer_maximum.html">layer_maximum</a>()</code> <br><code><a href="layer_mel_spectrogram.html">layer_mel_spectrogram</a>()</code> <br><code><a href="layer_minimum.html">layer_minimum</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_multi_head_attention.html">layer_multi_head_attention</a>()</code> <br><code><a href="layer_multiply.html">layer_multiply</a>()</code> <br><code><a href="layer_normalization.html">layer_normalization</a>()</code> <br><code><a href="layer_permute.html">layer_permute</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_brightness.html">layer_random_brightness</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_contrast.html">layer_random_contrast</a>()</code> <br><code><a href="layer_random_crop.html">layer_random_crop</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_flip.html">layer_random_flip</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_rotation.html">layer_random_rotation</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_random_translation.html">layer_random_translation</a>()</code> <br><code><a href="layer_random_zoom.html">layer_random_zoom</a>()</code> <br><code><a href="layer_repeat_vector.html">layer_repeat_vector</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_reshape.html">layer_reshape</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_rms_normalization.html">layer_rms_normalization</a>()</code> <br><code><a href="layer_rnn.html">layer_rnn</a>()</code> <br><code><a href="layer_separable_conv_1d.html">layer_separable_conv_1d</a>()</code> <br><code><a href="layer_separable_conv_2d.html">layer_separable_conv_2d</a>()</code> <br><code><a href="layer_simple_rnn.html">layer_simple_rnn</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br><code><a href="layer_spatial_dropout_1d.html">layer_spatial_dropout_1d</a>()</code> <br><code><a href="layer_spatial_dropout_2d.html">layer_spatial_dropout_2d</a>()</code> <br><code><a href="layer_spatial_dropout_3d.html">layer_spatial_dropout_3d</a>()</code> <br><code><a href="layer_spectral_normalization.html">layer_spectral_normalization</a>()</code> <br><code><a href="layer_stft_spectrogram.html">layer_stft_spectrogram</a>()</code> <br><code><a href="layer_string_lookup.html">layer_string_lookup</a>()</code> <br><code><a href="layer_subtract.html">layer_subtract</a>()</code> <br><code><a href="layer_text_vectorization.html">layer_text_vectorization</a>()</code> <br><code><a href="layer_tfsm.html">layer_tfsm</a>()</code> <br><code><a href="layer_time_distributed.html">layer_time_distributed</a>()</code> <br><code><a href="layer_torch_module_wrapper.html">layer_torch_module_wrapper</a>()</code> <br><code><a href="layer_unit_normalization.html">layer_unit_normalization</a>()</code> <br><code><a href="layer_upsampling_1d.html">layer_upsampling_1d</a>()</code> <br><code><a href="layer_upsampling_2d.html">layer_upsampling_2d</a>()</code> <br><code><a href="layer_upsampling_3d.html">layer_upsampling_3d</a>()</code> <br><code><a href="layer_zero_padding_1d.html">layer_zero_padding_1d</a>()</code> <br><code><a href="layer_zero_padding_2d.html">layer_zero_padding_2d</a>()</code> <br><code><a href="layer_zero_padding_3d.html">layer_zero_padding_3d</a>()</code> <br><code><a href="rnn_cell_gru.html">rnn_cell_gru</a>()</code> <br><code><a href="rnn_cell_lstm.html">rnn_cell_lstm</a>()</code> <br><code><a href="rnn_cell_simple.html">rnn_cell_simple</a>()</code> <br><code><a href="rnn_cells_stack.html">rnn_cells_stack</a>()</code> <br></p></div>
+<p>Other layers: <br><code><a href="Layer.html">Layer</a>()</code> <br><code><a href="layer_activation.html">layer_activation</a>()</code> <br><code><a href="layer_activation_elu.html">layer_activation_elu</a>()</code> <br><code><a href="layer_activation_leaky_relu.html">layer_activation_leaky_relu</a>()</code> <br><code><a href="layer_activation_parametric_relu.html">layer_activation_parametric_relu</a>()</code> <br><code><a href="layer_activation_relu.html">layer_activation_relu</a>()</code> <br><code><a href="layer_activation_softmax.html">layer_activation_softmax</a>()</code> <br><code><a href="layer_activity_regularization.html">layer_activity_regularization</a>()</code> <br><code><a href="layer_add.html">layer_add</a>()</code> <br><code><a href="layer_additive_attention.html">layer_additive_attention</a>()</code> <br><code><a href="layer_alpha_dropout.html">layer_alpha_dropout</a>()</code> <br><code><a href="layer_attention.html">layer_attention</a>()</code> <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_average.html">layer_average</a>()</code> <br><code><a href="layer_average_pooling_1d.html">layer_average_pooling_1d</a>()</code> <br><code><a href="layer_average_pooling_2d.html">layer_average_pooling_2d</a>()</code> <br><code><a href="layer_average_pooling_3d.html">layer_average_pooling_3d</a>()</code> <br><code><a href="layer_batch_normalization.html">layer_batch_normalization</a>()</code> <br><code><a href="layer_bidirectional.html">layer_bidirectional</a>()</code> <br><code><a href="layer_category_encoding.html">layer_category_encoding</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_concatenate.html">layer_concatenate</a>()</code> <br><code><a href="layer_conv_1d.html">layer_conv_1d</a>()</code> <br><code><a href="layer_conv_1d_transpose.html">layer_conv_1d_transpose</a>()</code> <br><code><a href="layer_conv_2d.html">layer_conv_2d</a>()</code> <br><code><a href="layer_conv_2d_transpose.html">layer_conv_2d_transpose</a>()</code> <br><code><a href="layer_conv_3d.html">layer_conv_3d</a>()</code> <br><code><a href="layer_conv_3d_transpose.html">layer_conv_3d_transpose</a>()</code> <br><code><a href="layer_conv_lstm_1d.html">layer_conv_lstm_1d</a>()</code> <br><code><a href="layer_conv_lstm_2d.html">layer_conv_lstm_2d</a>()</code> <br><code><a href="layer_conv_lstm_3d.html">layer_conv_lstm_3d</a>()</code> <br><code><a href="layer_cropping_1d.html">layer_cropping_1d</a>()</code> <br><code><a href="layer_cropping_2d.html">layer_cropping_2d</a>()</code> <br><code><a href="layer_cropping_3d.html">layer_cropping_3d</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_dense.html">layer_dense</a>()</code> <br><code><a href="layer_depthwise_conv_1d.html">layer_depthwise_conv_1d</a>()</code> <br><code><a href="layer_depthwise_conv_2d.html">layer_depthwise_conv_2d</a>()</code> <br><code><a href="layer_discretization.html">layer_discretization</a>()</code> <br><code><a href="layer_dot.html">layer_dot</a>()</code> <br><code><a href="layer_dropout.html">layer_dropout</a>()</code> <br><code><a href="layer_einsum_dense.html">layer_einsum_dense</a>()</code> <br><code><a href="layer_embedding.html">layer_embedding</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_feature_space.html">layer_feature_space</a>()</code> <br><code><a href="layer_flatten.html">layer_flatten</a>()</code> <br><code><a href="layer_flax_module_wrapper.html">layer_flax_module_wrapper</a>()</code> <br><code><a href="layer_gaussian_dropout.html">layer_gaussian_dropout</a>()</code> <br><code><a href="layer_gaussian_noise.html">layer_gaussian_noise</a>()</code> <br><code><a href="layer_global_average_pooling_1d.html">layer_global_average_pooling_1d</a>()</code> <br><code><a href="layer_global_average_pooling_2d.html">layer_global_average_pooling_2d</a>()</code> <br><code><a href="layer_global_average_pooling_3d.html">layer_global_average_pooling_3d</a>()</code> <br><code><a href="layer_global_max_pooling_1d.html">layer_global_max_pooling_1d</a>()</code> <br><code><a href="layer_global_max_pooling_2d.html">layer_global_max_pooling_2d</a>()</code> <br><code><a href="layer_global_max_pooling_3d.html">layer_global_max_pooling_3d</a>()</code> <br><code><a href="layer_group_normalization.html">layer_group_normalization</a>()</code> <br><code><a href="layer_group_query_attention.html">layer_group_query_attention</a>()</code> <br><code><a href="layer_gru.html">layer_gru</a>()</code> <br><code><a href="layer_hashed_crossing.html">layer_hashed_crossing</a>()</code> <br><code><a href="layer_hashing.html">layer_hashing</a>()</code> <br><code><a href="layer_identity.html">layer_identity</a>()</code> <br><code><a href="layer_integer_lookup.html">layer_integer_lookup</a>()</code> <br><code><a href="layer_jax_model_wrapper.html">layer_jax_model_wrapper</a>()</code> <br><code><a href="layer_lambda.html">layer_lambda</a>()</code> <br><code><a href="layer_lstm.html">layer_lstm</a>()</code> <br><code><a href="layer_masking.html">layer_masking</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_max_pooling_1d.html">layer_max_pooling_1d</a>()</code> <br><code><a href="layer_max_pooling_2d.html">layer_max_pooling_2d</a>()</code> <br><code><a href="layer_max_pooling_3d.html">layer_max_pooling_3d</a>()</code> <br><code><a href="layer_maximum.html">layer_maximum</a>()</code> <br><code><a href="layer_mel_spectrogram.html">layer_mel_spectrogram</a>()</code> <br><code><a href="layer_minimum.html">layer_minimum</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_multi_head_attention.html">layer_multi_head_attention</a>()</code> <br><code><a href="layer_multiply.html">layer_multiply</a>()</code> <br><code><a href="layer_normalization.html">layer_normalization</a>()</code> <br><code><a href="layer_permute.html">layer_permute</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_brightness.html">layer_random_brightness</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_contrast.html">layer_random_contrast</a>()</code> <br><code><a href="layer_random_crop.html">layer_random_crop</a>()</code> <br><code><a href="layer_random_elastic_transform.html">layer_random_elastic_transform</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_flip.html">layer_random_flip</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_rotation.html">layer_random_rotation</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_random_translation.html">layer_random_translation</a>()</code> <br><code><a href="layer_random_zoom.html">layer_random_zoom</a>()</code> <br><code><a href="layer_repeat_vector.html">layer_repeat_vector</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_reshape.html">layer_reshape</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_rms_normalization.html">layer_rms_normalization</a>()</code> <br><code><a href="layer_rnn.html">layer_rnn</a>()</code> <br><code><a href="layer_separable_conv_1d.html">layer_separable_conv_1d</a>()</code> <br><code><a href="layer_separable_conv_2d.html">layer_separable_conv_2d</a>()</code> <br><code><a href="layer_simple_rnn.html">layer_simple_rnn</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br><code><a href="layer_spatial_dropout_1d.html">layer_spatial_dropout_1d</a>()</code> <br><code><a href="layer_spatial_dropout_2d.html">layer_spatial_dropout_2d</a>()</code> <br><code><a href="layer_spatial_dropout_3d.html">layer_spatial_dropout_3d</a>()</code> <br><code><a href="layer_spectral_normalization.html">layer_spectral_normalization</a>()</code> <br><code><a href="layer_stft_spectrogram.html">layer_stft_spectrogram</a>()</code> <br><code><a href="layer_string_lookup.html">layer_string_lookup</a>()</code> <br><code><a href="layer_subtract.html">layer_subtract</a>()</code> <br><code><a href="layer_text_vectorization.html">layer_text_vectorization</a>()</code> <br><code><a href="layer_tfsm.html">layer_tfsm</a>()</code> <br><code><a href="layer_time_distributed.html">layer_time_distributed</a>()</code> <br><code><a href="layer_torch_module_wrapper.html">layer_torch_module_wrapper</a>()</code> <br><code><a href="layer_unit_normalization.html">layer_unit_normalization</a>()</code> <br><code><a href="layer_upsampling_1d.html">layer_upsampling_1d</a>()</code> <br><code><a href="layer_upsampling_2d.html">layer_upsampling_2d</a>()</code> <br><code><a href="layer_upsampling_3d.html">layer_upsampling_3d</a>()</code> <br><code><a href="layer_zero_padding_1d.html">layer_zero_padding_1d</a>()</code> <br><code><a href="layer_zero_padding_2d.html">layer_zero_padding_2d</a>()</code> <br><code><a href="layer_zero_padding_3d.html">layer_zero_padding_3d</a>()</code> <br><code><a href="rnn_cell_gru.html">rnn_cell_gru</a>()</code> <br><code><a href="rnn_cell_lstm.html">rnn_cell_lstm</a>()</code> <br><code><a href="rnn_cell_simple.html">rnn_cell_simple</a>()</code> <br><code><a href="rnn_cells_stack.html">rnn_cells_stack</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -316,7 +306,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/layer_lstm.html b/docs/dev/reference/layer_lstm.html
index 99b5c68c2..c43d883e5 100644
--- a/docs/dev/reference/layer_lstm.html
+++ b/docs/dev/reference/layer_lstm.html
@@ -88,7 +88,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -375,7 +375,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 <ul><li><p><a href="https://keras.io/api/layers/recurrent_layers/lstm#lstm-class" class="external-link">https://keras.io/api/layers/recurrent_layers/lstm#lstm-class</a></p></li>
 </ul><p>Other lstm rnn layers: <br><code><a href="rnn_cell_lstm.html">rnn_cell_lstm</a>()</code> <br></p>
 <p>Other rnn layers: <br><code><a href="layer_bidirectional.html">layer_bidirectional</a>()</code> <br><code><a href="layer_conv_lstm_1d.html">layer_conv_lstm_1d</a>()</code> <br><code><a href="layer_conv_lstm_2d.html">layer_conv_lstm_2d</a>()</code> <br><code><a href="layer_conv_lstm_3d.html">layer_conv_lstm_3d</a>()</code> <br><code><a href="layer_gru.html">layer_gru</a>()</code> <br><code><a href="layer_rnn.html">layer_rnn</a>()</code> <br><code><a href="layer_simple_rnn.html">layer_simple_rnn</a>()</code> <br><code><a href="layer_time_distributed.html">layer_time_distributed</a>()</code> <br><code><a href="rnn_cell_gru.html">rnn_cell_gru</a>()</code> <br><code><a href="rnn_cell_lstm.html">rnn_cell_lstm</a>()</code> <br><code><a href="rnn_cell_simple.html">rnn_cell_simple</a>()</code> <br><code><a href="rnn_cells_stack.html">rnn_cells_stack</a>()</code> <br></p>
-<p>Other layers: <br><code><a href="Layer.html">Layer</a>()</code> <br><code><a href="layer_activation.html">layer_activation</a>()</code> <br><code><a href="layer_activation_elu.html">layer_activation_elu</a>()</code> <br><code><a href="layer_activation_leaky_relu.html">layer_activation_leaky_relu</a>()</code> <br><code><a href="layer_activation_parametric_relu.html">layer_activation_parametric_relu</a>()</code> <br><code><a href="layer_activation_relu.html">layer_activation_relu</a>()</code> <br><code><a href="layer_activation_softmax.html">layer_activation_softmax</a>()</code> <br><code><a href="layer_activity_regularization.html">layer_activity_regularization</a>()</code> <br><code><a href="layer_add.html">layer_add</a>()</code> <br><code><a href="layer_additive_attention.html">layer_additive_attention</a>()</code> <br><code><a href="layer_alpha_dropout.html">layer_alpha_dropout</a>()</code> <br><code><a href="layer_attention.html">layer_attention</a>()</code> <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_average.html">layer_average</a>()</code> <br><code><a href="layer_average_pooling_1d.html">layer_average_pooling_1d</a>()</code> <br><code><a href="layer_average_pooling_2d.html">layer_average_pooling_2d</a>()</code> <br><code><a href="layer_average_pooling_3d.html">layer_average_pooling_3d</a>()</code> <br><code><a href="layer_batch_normalization.html">layer_batch_normalization</a>()</code> <br><code><a href="layer_bidirectional.html">layer_bidirectional</a>()</code> <br><code><a href="layer_category_encoding.html">layer_category_encoding</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_concatenate.html">layer_concatenate</a>()</code> <br><code><a href="layer_conv_1d.html">layer_conv_1d</a>()</code> <br><code><a href="layer_conv_1d_transpose.html">layer_conv_1d_transpose</a>()</code> <br><code><a href="layer_conv_2d.html">layer_conv_2d</a>()</code> <br><code><a href="layer_conv_2d_transpose.html">layer_conv_2d_transpose</a>()</code> <br><code><a href="layer_conv_3d.html">layer_conv_3d</a>()</code> <br><code><a href="layer_conv_3d_transpose.html">layer_conv_3d_transpose</a>()</code> <br><code><a href="layer_conv_lstm_1d.html">layer_conv_lstm_1d</a>()</code> <br><code><a href="layer_conv_lstm_2d.html">layer_conv_lstm_2d</a>()</code> <br><code><a href="layer_conv_lstm_3d.html">layer_conv_lstm_3d</a>()</code> <br><code><a href="layer_cropping_1d.html">layer_cropping_1d</a>()</code> <br><code><a href="layer_cropping_2d.html">layer_cropping_2d</a>()</code> <br><code><a href="layer_cropping_3d.html">layer_cropping_3d</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_dense.html">layer_dense</a>()</code> <br><code><a href="layer_depthwise_conv_1d.html">layer_depthwise_conv_1d</a>()</code> <br><code><a href="layer_depthwise_conv_2d.html">layer_depthwise_conv_2d</a>()</code> <br><code><a href="layer_discretization.html">layer_discretization</a>()</code> <br><code><a href="layer_dot.html">layer_dot</a>()</code> <br><code><a href="layer_dropout.html">layer_dropout</a>()</code> <br><code><a href="layer_einsum_dense.html">layer_einsum_dense</a>()</code> <br><code><a href="layer_embedding.html">layer_embedding</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_feature_space.html">layer_feature_space</a>()</code> <br><code><a href="layer_flatten.html">layer_flatten</a>()</code> <br><code><a href="layer_flax_module_wrapper.html">layer_flax_module_wrapper</a>()</code> <br><code><a href="layer_gaussian_dropout.html">layer_gaussian_dropout</a>()</code> <br><code><a href="layer_gaussian_noise.html">layer_gaussian_noise</a>()</code> <br><code><a href="layer_global_average_pooling_1d.html">layer_global_average_pooling_1d</a>()</code> <br><code><a href="layer_global_average_pooling_2d.html">layer_global_average_pooling_2d</a>()</code> <br><code><a href="layer_global_average_pooling_3d.html">layer_global_average_pooling_3d</a>()</code> <br><code><a href="layer_global_max_pooling_1d.html">layer_global_max_pooling_1d</a>()</code> <br><code><a href="layer_global_max_pooling_2d.html">layer_global_max_pooling_2d</a>()</code> <br><code><a href="layer_global_max_pooling_3d.html">layer_global_max_pooling_3d</a>()</code> <br><code><a href="layer_group_normalization.html">layer_group_normalization</a>()</code> <br><code><a href="layer_group_query_attention.html">layer_group_query_attention</a>()</code> <br><code><a href="layer_gru.html">layer_gru</a>()</code> <br><code><a href="layer_hashed_crossing.html">layer_hashed_crossing</a>()</code> <br><code><a href="layer_hashing.html">layer_hashing</a>()</code> <br><code><a href="layer_identity.html">layer_identity</a>()</code> <br><code><a href="layer_integer_lookup.html">layer_integer_lookup</a>()</code> <br><code><a href="layer_jax_model_wrapper.html">layer_jax_model_wrapper</a>()</code> <br><code><a href="layer_lambda.html">layer_lambda</a>()</code> <br><code><a href="layer_layer_normalization.html">layer_layer_normalization</a>()</code> <br><code><a href="layer_masking.html">layer_masking</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_max_pooling_1d.html">layer_max_pooling_1d</a>()</code> <br><code><a href="layer_max_pooling_2d.html">layer_max_pooling_2d</a>()</code> <br><code><a href="layer_max_pooling_3d.html">layer_max_pooling_3d</a>()</code> <br><code><a href="layer_maximum.html">layer_maximum</a>()</code> <br><code><a href="layer_mel_spectrogram.html">layer_mel_spectrogram</a>()</code> <br><code><a href="layer_minimum.html">layer_minimum</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_multi_head_attention.html">layer_multi_head_attention</a>()</code> <br><code><a href="layer_multiply.html">layer_multiply</a>()</code> <br><code><a href="layer_normalization.html">layer_normalization</a>()</code> <br><code><a href="layer_permute.html">layer_permute</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_brightness.html">layer_random_brightness</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_contrast.html">layer_random_contrast</a>()</code> <br><code><a href="layer_random_crop.html">layer_random_crop</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_flip.html">layer_random_flip</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_rotation.html">layer_random_rotation</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_random_translation.html">layer_random_translation</a>()</code> <br><code><a href="layer_random_zoom.html">layer_random_zoom</a>()</code> <br><code><a href="layer_repeat_vector.html">layer_repeat_vector</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_reshape.html">layer_reshape</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_rms_normalization.html">layer_rms_normalization</a>()</code> <br><code><a href="layer_rnn.html">layer_rnn</a>()</code> <br><code><a href="layer_separable_conv_1d.html">layer_separable_conv_1d</a>()</code> <br><code><a href="layer_separable_conv_2d.html">layer_separable_conv_2d</a>()</code> <br><code><a href="layer_simple_rnn.html">layer_simple_rnn</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br><code><a href="layer_spatial_dropout_1d.html">layer_spatial_dropout_1d</a>()</code> <br><code><a href="layer_spatial_dropout_2d.html">layer_spatial_dropout_2d</a>()</code> <br><code><a href="layer_spatial_dropout_3d.html">layer_spatial_dropout_3d</a>()</code> <br><code><a href="layer_spectral_normalization.html">layer_spectral_normalization</a>()</code> <br><code><a href="layer_stft_spectrogram.html">layer_stft_spectrogram</a>()</code> <br><code><a href="layer_string_lookup.html">layer_string_lookup</a>()</code> <br><code><a href="layer_subtract.html">layer_subtract</a>()</code> <br><code><a href="layer_text_vectorization.html">layer_text_vectorization</a>()</code> <br><code><a href="layer_tfsm.html">layer_tfsm</a>()</code> <br><code><a href="layer_time_distributed.html">layer_time_distributed</a>()</code> <br><code><a href="layer_torch_module_wrapper.html">layer_torch_module_wrapper</a>()</code> <br><code><a href="layer_unit_normalization.html">layer_unit_normalization</a>()</code> <br><code><a href="layer_upsampling_1d.html">layer_upsampling_1d</a>()</code> <br><code><a href="layer_upsampling_2d.html">layer_upsampling_2d</a>()</code> <br><code><a href="layer_upsampling_3d.html">layer_upsampling_3d</a>()</code> <br><code><a href="layer_zero_padding_1d.html">layer_zero_padding_1d</a>()</code> <br><code><a href="layer_zero_padding_2d.html">layer_zero_padding_2d</a>()</code> <br><code><a href="layer_zero_padding_3d.html">layer_zero_padding_3d</a>()</code> <br><code><a href="rnn_cell_gru.html">rnn_cell_gru</a>()</code> <br><code><a href="rnn_cell_lstm.html">rnn_cell_lstm</a>()</code> <br><code><a href="rnn_cell_simple.html">rnn_cell_simple</a>()</code> <br><code><a href="rnn_cells_stack.html">rnn_cells_stack</a>()</code> <br></p></div>
+<p>Other layers: <br><code><a href="Layer.html">Layer</a>()</code> <br><code><a href="layer_activation.html">layer_activation</a>()</code> <br><code><a href="layer_activation_elu.html">layer_activation_elu</a>()</code> <br><code><a href="layer_activation_leaky_relu.html">layer_activation_leaky_relu</a>()</code> <br><code><a href="layer_activation_parametric_relu.html">layer_activation_parametric_relu</a>()</code> <br><code><a href="layer_activation_relu.html">layer_activation_relu</a>()</code> <br><code><a href="layer_activation_softmax.html">layer_activation_softmax</a>()</code> <br><code><a href="layer_activity_regularization.html">layer_activity_regularization</a>()</code> <br><code><a href="layer_add.html">layer_add</a>()</code> <br><code><a href="layer_additive_attention.html">layer_additive_attention</a>()</code> <br><code><a href="layer_alpha_dropout.html">layer_alpha_dropout</a>()</code> <br><code><a href="layer_attention.html">layer_attention</a>()</code> <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_average.html">layer_average</a>()</code> <br><code><a href="layer_average_pooling_1d.html">layer_average_pooling_1d</a>()</code> <br><code><a href="layer_average_pooling_2d.html">layer_average_pooling_2d</a>()</code> <br><code><a href="layer_average_pooling_3d.html">layer_average_pooling_3d</a>()</code> <br><code><a href="layer_batch_normalization.html">layer_batch_normalization</a>()</code> <br><code><a href="layer_bidirectional.html">layer_bidirectional</a>()</code> <br><code><a href="layer_category_encoding.html">layer_category_encoding</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_concatenate.html">layer_concatenate</a>()</code> <br><code><a href="layer_conv_1d.html">layer_conv_1d</a>()</code> <br><code><a href="layer_conv_1d_transpose.html">layer_conv_1d_transpose</a>()</code> <br><code><a href="layer_conv_2d.html">layer_conv_2d</a>()</code> <br><code><a href="layer_conv_2d_transpose.html">layer_conv_2d_transpose</a>()</code> <br><code><a href="layer_conv_3d.html">layer_conv_3d</a>()</code> <br><code><a href="layer_conv_3d_transpose.html">layer_conv_3d_transpose</a>()</code> <br><code><a href="layer_conv_lstm_1d.html">layer_conv_lstm_1d</a>()</code> <br><code><a href="layer_conv_lstm_2d.html">layer_conv_lstm_2d</a>()</code> <br><code><a href="layer_conv_lstm_3d.html">layer_conv_lstm_3d</a>()</code> <br><code><a href="layer_cropping_1d.html">layer_cropping_1d</a>()</code> <br><code><a href="layer_cropping_2d.html">layer_cropping_2d</a>()</code> <br><code><a href="layer_cropping_3d.html">layer_cropping_3d</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_dense.html">layer_dense</a>()</code> <br><code><a href="layer_depthwise_conv_1d.html">layer_depthwise_conv_1d</a>()</code> <br><code><a href="layer_depthwise_conv_2d.html">layer_depthwise_conv_2d</a>()</code> <br><code><a href="layer_discretization.html">layer_discretization</a>()</code> <br><code><a href="layer_dot.html">layer_dot</a>()</code> <br><code><a href="layer_dropout.html">layer_dropout</a>()</code> <br><code><a href="layer_einsum_dense.html">layer_einsum_dense</a>()</code> <br><code><a href="layer_embedding.html">layer_embedding</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_feature_space.html">layer_feature_space</a>()</code> <br><code><a href="layer_flatten.html">layer_flatten</a>()</code> <br><code><a href="layer_flax_module_wrapper.html">layer_flax_module_wrapper</a>()</code> <br><code><a href="layer_gaussian_dropout.html">layer_gaussian_dropout</a>()</code> <br><code><a href="layer_gaussian_noise.html">layer_gaussian_noise</a>()</code> <br><code><a href="layer_global_average_pooling_1d.html">layer_global_average_pooling_1d</a>()</code> <br><code><a href="layer_global_average_pooling_2d.html">layer_global_average_pooling_2d</a>()</code> <br><code><a href="layer_global_average_pooling_3d.html">layer_global_average_pooling_3d</a>()</code> <br><code><a href="layer_global_max_pooling_1d.html">layer_global_max_pooling_1d</a>()</code> <br><code><a href="layer_global_max_pooling_2d.html">layer_global_max_pooling_2d</a>()</code> <br><code><a href="layer_global_max_pooling_3d.html">layer_global_max_pooling_3d</a>()</code> <br><code><a href="layer_group_normalization.html">layer_group_normalization</a>()</code> <br><code><a href="layer_group_query_attention.html">layer_group_query_attention</a>()</code> <br><code><a href="layer_gru.html">layer_gru</a>()</code> <br><code><a href="layer_hashed_crossing.html">layer_hashed_crossing</a>()</code> <br><code><a href="layer_hashing.html">layer_hashing</a>()</code> <br><code><a href="layer_identity.html">layer_identity</a>()</code> <br><code><a href="layer_integer_lookup.html">layer_integer_lookup</a>()</code> <br><code><a href="layer_jax_model_wrapper.html">layer_jax_model_wrapper</a>()</code> <br><code><a href="layer_lambda.html">layer_lambda</a>()</code> <br><code><a href="layer_layer_normalization.html">layer_layer_normalization</a>()</code> <br><code><a href="layer_masking.html">layer_masking</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_max_pooling_1d.html">layer_max_pooling_1d</a>()</code> <br><code><a href="layer_max_pooling_2d.html">layer_max_pooling_2d</a>()</code> <br><code><a href="layer_max_pooling_3d.html">layer_max_pooling_3d</a>()</code> <br><code><a href="layer_maximum.html">layer_maximum</a>()</code> <br><code><a href="layer_mel_spectrogram.html">layer_mel_spectrogram</a>()</code> <br><code><a href="layer_minimum.html">layer_minimum</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_multi_head_attention.html">layer_multi_head_attention</a>()</code> <br><code><a href="layer_multiply.html">layer_multiply</a>()</code> <br><code><a href="layer_normalization.html">layer_normalization</a>()</code> <br><code><a href="layer_permute.html">layer_permute</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_brightness.html">layer_random_brightness</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_contrast.html">layer_random_contrast</a>()</code> <br><code><a href="layer_random_crop.html">layer_random_crop</a>()</code> <br><code><a href="layer_random_elastic_transform.html">layer_random_elastic_transform</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_flip.html">layer_random_flip</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_rotation.html">layer_random_rotation</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_random_translation.html">layer_random_translation</a>()</code> <br><code><a href="layer_random_zoom.html">layer_random_zoom</a>()</code> <br><code><a href="layer_repeat_vector.html">layer_repeat_vector</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_reshape.html">layer_reshape</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_rms_normalization.html">layer_rms_normalization</a>()</code> <br><code><a href="layer_rnn.html">layer_rnn</a>()</code> <br><code><a href="layer_separable_conv_1d.html">layer_separable_conv_1d</a>()</code> <br><code><a href="layer_separable_conv_2d.html">layer_separable_conv_2d</a>()</code> <br><code><a href="layer_simple_rnn.html">layer_simple_rnn</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br><code><a href="layer_spatial_dropout_1d.html">layer_spatial_dropout_1d</a>()</code> <br><code><a href="layer_spatial_dropout_2d.html">layer_spatial_dropout_2d</a>()</code> <br><code><a href="layer_spatial_dropout_3d.html">layer_spatial_dropout_3d</a>()</code> <br><code><a href="layer_spectral_normalization.html">layer_spectral_normalization</a>()</code> <br><code><a href="layer_stft_spectrogram.html">layer_stft_spectrogram</a>()</code> <br><code><a href="layer_string_lookup.html">layer_string_lookup</a>()</code> <br><code><a href="layer_subtract.html">layer_subtract</a>()</code> <br><code><a href="layer_text_vectorization.html">layer_text_vectorization</a>()</code> <br><code><a href="layer_tfsm.html">layer_tfsm</a>()</code> <br><code><a href="layer_time_distributed.html">layer_time_distributed</a>()</code> <br><code><a href="layer_torch_module_wrapper.html">layer_torch_module_wrapper</a>()</code> <br><code><a href="layer_unit_normalization.html">layer_unit_normalization</a>()</code> <br><code><a href="layer_upsampling_1d.html">layer_upsampling_1d</a>()</code> <br><code><a href="layer_upsampling_2d.html">layer_upsampling_2d</a>()</code> <br><code><a href="layer_upsampling_3d.html">layer_upsampling_3d</a>()</code> <br><code><a href="layer_zero_padding_1d.html">layer_zero_padding_1d</a>()</code> <br><code><a href="layer_zero_padding_2d.html">layer_zero_padding_2d</a>()</code> <br><code><a href="layer_zero_padding_3d.html">layer_zero_padding_3d</a>()</code> <br><code><a href="rnn_cell_gru.html">rnn_cell_gru</a>()</code> <br><code><a href="rnn_cell_lstm.html">rnn_cell_lstm</a>()</code> <br><code><a href="rnn_cell_simple.html">rnn_cell_simple</a>()</code> <br><code><a href="rnn_cells_stack.html">rnn_cells_stack</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -387,7 +387,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/layer_masking.html b/docs/dev/reference/layer_masking.html
index 3c5de166b..9d5424b36 100644
--- a/docs/dev/reference/layer_masking.html
+++ b/docs/dev/reference/layer_masking.html
@@ -18,7 +18,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -131,7 +131,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
     <div class="dont-index">
 <ul><li><p><a href="https://keras.io/api/layers/core_layers/masking#masking-class" class="external-link">https://keras.io/api/layers/core_layers/masking#masking-class</a></p></li>
 </ul><p>Other core layers: <br><code><a href="layer_dense.html">layer_dense</a>()</code> <br><code><a href="layer_einsum_dense.html">layer_einsum_dense</a>()</code> <br><code><a href="layer_embedding.html">layer_embedding</a>()</code> <br><code><a href="layer_identity.html">layer_identity</a>()</code> <br><code><a href="layer_lambda.html">layer_lambda</a>()</code> <br></p>
-<p>Other layers: <br><code><a href="Layer.html">Layer</a>()</code> <br><code><a href="layer_activation.html">layer_activation</a>()</code> <br><code><a href="layer_activation_elu.html">layer_activation_elu</a>()</code> <br><code><a href="layer_activation_leaky_relu.html">layer_activation_leaky_relu</a>()</code> <br><code><a href="layer_activation_parametric_relu.html">layer_activation_parametric_relu</a>()</code> <br><code><a href="layer_activation_relu.html">layer_activation_relu</a>()</code> <br><code><a href="layer_activation_softmax.html">layer_activation_softmax</a>()</code> <br><code><a href="layer_activity_regularization.html">layer_activity_regularization</a>()</code> <br><code><a href="layer_add.html">layer_add</a>()</code> <br><code><a href="layer_additive_attention.html">layer_additive_attention</a>()</code> <br><code><a href="layer_alpha_dropout.html">layer_alpha_dropout</a>()</code> <br><code><a href="layer_attention.html">layer_attention</a>()</code> <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_average.html">layer_average</a>()</code> <br><code><a href="layer_average_pooling_1d.html">layer_average_pooling_1d</a>()</code> <br><code><a href="layer_average_pooling_2d.html">layer_average_pooling_2d</a>()</code> <br><code><a href="layer_average_pooling_3d.html">layer_average_pooling_3d</a>()</code> <br><code><a href="layer_batch_normalization.html">layer_batch_normalization</a>()</code> <br><code><a href="layer_bidirectional.html">layer_bidirectional</a>()</code> <br><code><a href="layer_category_encoding.html">layer_category_encoding</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_concatenate.html">layer_concatenate</a>()</code> <br><code><a href="layer_conv_1d.html">layer_conv_1d</a>()</code> <br><code><a href="layer_conv_1d_transpose.html">layer_conv_1d_transpose</a>()</code> <br><code><a href="layer_conv_2d.html">layer_conv_2d</a>()</code> <br><code><a href="layer_conv_2d_transpose.html">layer_conv_2d_transpose</a>()</code> <br><code><a href="layer_conv_3d.html">layer_conv_3d</a>()</code> <br><code><a href="layer_conv_3d_transpose.html">layer_conv_3d_transpose</a>()</code> <br><code><a href="layer_conv_lstm_1d.html">layer_conv_lstm_1d</a>()</code> <br><code><a href="layer_conv_lstm_2d.html">layer_conv_lstm_2d</a>()</code> <br><code><a href="layer_conv_lstm_3d.html">layer_conv_lstm_3d</a>()</code> <br><code><a href="layer_cropping_1d.html">layer_cropping_1d</a>()</code> <br><code><a href="layer_cropping_2d.html">layer_cropping_2d</a>()</code> <br><code><a href="layer_cropping_3d.html">layer_cropping_3d</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_dense.html">layer_dense</a>()</code> <br><code><a href="layer_depthwise_conv_1d.html">layer_depthwise_conv_1d</a>()</code> <br><code><a href="layer_depthwise_conv_2d.html">layer_depthwise_conv_2d</a>()</code> <br><code><a href="layer_discretization.html">layer_discretization</a>()</code> <br><code><a href="layer_dot.html">layer_dot</a>()</code> <br><code><a href="layer_dropout.html">layer_dropout</a>()</code> <br><code><a href="layer_einsum_dense.html">layer_einsum_dense</a>()</code> <br><code><a href="layer_embedding.html">layer_embedding</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_feature_space.html">layer_feature_space</a>()</code> <br><code><a href="layer_flatten.html">layer_flatten</a>()</code> <br><code><a href="layer_flax_module_wrapper.html">layer_flax_module_wrapper</a>()</code> <br><code><a href="layer_gaussian_dropout.html">layer_gaussian_dropout</a>()</code> <br><code><a href="layer_gaussian_noise.html">layer_gaussian_noise</a>()</code> <br><code><a href="layer_global_average_pooling_1d.html">layer_global_average_pooling_1d</a>()</code> <br><code><a href="layer_global_average_pooling_2d.html">layer_global_average_pooling_2d</a>()</code> <br><code><a href="layer_global_average_pooling_3d.html">layer_global_average_pooling_3d</a>()</code> <br><code><a href="layer_global_max_pooling_1d.html">layer_global_max_pooling_1d</a>()</code> <br><code><a href="layer_global_max_pooling_2d.html">layer_global_max_pooling_2d</a>()</code> <br><code><a href="layer_global_max_pooling_3d.html">layer_global_max_pooling_3d</a>()</code> <br><code><a href="layer_group_normalization.html">layer_group_normalization</a>()</code> <br><code><a href="layer_group_query_attention.html">layer_group_query_attention</a>()</code> <br><code><a href="layer_gru.html">layer_gru</a>()</code> <br><code><a href="layer_hashed_crossing.html">layer_hashed_crossing</a>()</code> <br><code><a href="layer_hashing.html">layer_hashing</a>()</code> <br><code><a href="layer_identity.html">layer_identity</a>()</code> <br><code><a href="layer_integer_lookup.html">layer_integer_lookup</a>()</code> <br><code><a href="layer_jax_model_wrapper.html">layer_jax_model_wrapper</a>()</code> <br><code><a href="layer_lambda.html">layer_lambda</a>()</code> <br><code><a href="layer_layer_normalization.html">layer_layer_normalization</a>()</code> <br><code><a href="layer_lstm.html">layer_lstm</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_max_pooling_1d.html">layer_max_pooling_1d</a>()</code> <br><code><a href="layer_max_pooling_2d.html">layer_max_pooling_2d</a>()</code> <br><code><a href="layer_max_pooling_3d.html">layer_max_pooling_3d</a>()</code> <br><code><a href="layer_maximum.html">layer_maximum</a>()</code> <br><code><a href="layer_mel_spectrogram.html">layer_mel_spectrogram</a>()</code> <br><code><a href="layer_minimum.html">layer_minimum</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_multi_head_attention.html">layer_multi_head_attention</a>()</code> <br><code><a href="layer_multiply.html">layer_multiply</a>()</code> <br><code><a href="layer_normalization.html">layer_normalization</a>()</code> <br><code><a href="layer_permute.html">layer_permute</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_brightness.html">layer_random_brightness</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_contrast.html">layer_random_contrast</a>()</code> <br><code><a href="layer_random_crop.html">layer_random_crop</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_flip.html">layer_random_flip</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_rotation.html">layer_random_rotation</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_random_translation.html">layer_random_translation</a>()</code> <br><code><a href="layer_random_zoom.html">layer_random_zoom</a>()</code> <br><code><a href="layer_repeat_vector.html">layer_repeat_vector</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_reshape.html">layer_reshape</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_rms_normalization.html">layer_rms_normalization</a>()</code> <br><code><a href="layer_rnn.html">layer_rnn</a>()</code> <br><code><a href="layer_separable_conv_1d.html">layer_separable_conv_1d</a>()</code> <br><code><a href="layer_separable_conv_2d.html">layer_separable_conv_2d</a>()</code> <br><code><a href="layer_simple_rnn.html">layer_simple_rnn</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br><code><a href="layer_spatial_dropout_1d.html">layer_spatial_dropout_1d</a>()</code> <br><code><a href="layer_spatial_dropout_2d.html">layer_spatial_dropout_2d</a>()</code> <br><code><a href="layer_spatial_dropout_3d.html">layer_spatial_dropout_3d</a>()</code> <br><code><a href="layer_spectral_normalization.html">layer_spectral_normalization</a>()</code> <br><code><a href="layer_stft_spectrogram.html">layer_stft_spectrogram</a>()</code> <br><code><a href="layer_string_lookup.html">layer_string_lookup</a>()</code> <br><code><a href="layer_subtract.html">layer_subtract</a>()</code> <br><code><a href="layer_text_vectorization.html">layer_text_vectorization</a>()</code> <br><code><a href="layer_tfsm.html">layer_tfsm</a>()</code> <br><code><a href="layer_time_distributed.html">layer_time_distributed</a>()</code> <br><code><a href="layer_torch_module_wrapper.html">layer_torch_module_wrapper</a>()</code> <br><code><a href="layer_unit_normalization.html">layer_unit_normalization</a>()</code> <br><code><a href="layer_upsampling_1d.html">layer_upsampling_1d</a>()</code> <br><code><a href="layer_upsampling_2d.html">layer_upsampling_2d</a>()</code> <br><code><a href="layer_upsampling_3d.html">layer_upsampling_3d</a>()</code> <br><code><a href="layer_zero_padding_1d.html">layer_zero_padding_1d</a>()</code> <br><code><a href="layer_zero_padding_2d.html">layer_zero_padding_2d</a>()</code> <br><code><a href="layer_zero_padding_3d.html">layer_zero_padding_3d</a>()</code> <br><code><a href="rnn_cell_gru.html">rnn_cell_gru</a>()</code> <br><code><a href="rnn_cell_lstm.html">rnn_cell_lstm</a>()</code> <br><code><a href="rnn_cell_simple.html">rnn_cell_simple</a>()</code> <br><code><a href="rnn_cells_stack.html">rnn_cells_stack</a>()</code> <br></p></div>
+<p>Other layers: <br><code><a href="Layer.html">Layer</a>()</code> <br><code><a href="layer_activation.html">layer_activation</a>()</code> <br><code><a href="layer_activation_elu.html">layer_activation_elu</a>()</code> <br><code><a href="layer_activation_leaky_relu.html">layer_activation_leaky_relu</a>()</code> <br><code><a href="layer_activation_parametric_relu.html">layer_activation_parametric_relu</a>()</code> <br><code><a href="layer_activation_relu.html">layer_activation_relu</a>()</code> <br><code><a href="layer_activation_softmax.html">layer_activation_softmax</a>()</code> <br><code><a href="layer_activity_regularization.html">layer_activity_regularization</a>()</code> <br><code><a href="layer_add.html">layer_add</a>()</code> <br><code><a href="layer_additive_attention.html">layer_additive_attention</a>()</code> <br><code><a href="layer_alpha_dropout.html">layer_alpha_dropout</a>()</code> <br><code><a href="layer_attention.html">layer_attention</a>()</code> <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_average.html">layer_average</a>()</code> <br><code><a href="layer_average_pooling_1d.html">layer_average_pooling_1d</a>()</code> <br><code><a href="layer_average_pooling_2d.html">layer_average_pooling_2d</a>()</code> <br><code><a href="layer_average_pooling_3d.html">layer_average_pooling_3d</a>()</code> <br><code><a href="layer_batch_normalization.html">layer_batch_normalization</a>()</code> <br><code><a href="layer_bidirectional.html">layer_bidirectional</a>()</code> <br><code><a href="layer_category_encoding.html">layer_category_encoding</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_concatenate.html">layer_concatenate</a>()</code> <br><code><a href="layer_conv_1d.html">layer_conv_1d</a>()</code> <br><code><a href="layer_conv_1d_transpose.html">layer_conv_1d_transpose</a>()</code> <br><code><a href="layer_conv_2d.html">layer_conv_2d</a>()</code> <br><code><a href="layer_conv_2d_transpose.html">layer_conv_2d_transpose</a>()</code> <br><code><a href="layer_conv_3d.html">layer_conv_3d</a>()</code> <br><code><a href="layer_conv_3d_transpose.html">layer_conv_3d_transpose</a>()</code> <br><code><a href="layer_conv_lstm_1d.html">layer_conv_lstm_1d</a>()</code> <br><code><a href="layer_conv_lstm_2d.html">layer_conv_lstm_2d</a>()</code> <br><code><a href="layer_conv_lstm_3d.html">layer_conv_lstm_3d</a>()</code> <br><code><a href="layer_cropping_1d.html">layer_cropping_1d</a>()</code> <br><code><a href="layer_cropping_2d.html">layer_cropping_2d</a>()</code> <br><code><a href="layer_cropping_3d.html">layer_cropping_3d</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_dense.html">layer_dense</a>()</code> <br><code><a href="layer_depthwise_conv_1d.html">layer_depthwise_conv_1d</a>()</code> <br><code><a href="layer_depthwise_conv_2d.html">layer_depthwise_conv_2d</a>()</code> <br><code><a href="layer_discretization.html">layer_discretization</a>()</code> <br><code><a href="layer_dot.html">layer_dot</a>()</code> <br><code><a href="layer_dropout.html">layer_dropout</a>()</code> <br><code><a href="layer_einsum_dense.html">layer_einsum_dense</a>()</code> <br><code><a href="layer_embedding.html">layer_embedding</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_feature_space.html">layer_feature_space</a>()</code> <br><code><a href="layer_flatten.html">layer_flatten</a>()</code> <br><code><a href="layer_flax_module_wrapper.html">layer_flax_module_wrapper</a>()</code> <br><code><a href="layer_gaussian_dropout.html">layer_gaussian_dropout</a>()</code> <br><code><a href="layer_gaussian_noise.html">layer_gaussian_noise</a>()</code> <br><code><a href="layer_global_average_pooling_1d.html">layer_global_average_pooling_1d</a>()</code> <br><code><a href="layer_global_average_pooling_2d.html">layer_global_average_pooling_2d</a>()</code> <br><code><a href="layer_global_average_pooling_3d.html">layer_global_average_pooling_3d</a>()</code> <br><code><a href="layer_global_max_pooling_1d.html">layer_global_max_pooling_1d</a>()</code> <br><code><a href="layer_global_max_pooling_2d.html">layer_global_max_pooling_2d</a>()</code> <br><code><a href="layer_global_max_pooling_3d.html">layer_global_max_pooling_3d</a>()</code> <br><code><a href="layer_group_normalization.html">layer_group_normalization</a>()</code> <br><code><a href="layer_group_query_attention.html">layer_group_query_attention</a>()</code> <br><code><a href="layer_gru.html">layer_gru</a>()</code> <br><code><a href="layer_hashed_crossing.html">layer_hashed_crossing</a>()</code> <br><code><a href="layer_hashing.html">layer_hashing</a>()</code> <br><code><a href="layer_identity.html">layer_identity</a>()</code> <br><code><a href="layer_integer_lookup.html">layer_integer_lookup</a>()</code> <br><code><a href="layer_jax_model_wrapper.html">layer_jax_model_wrapper</a>()</code> <br><code><a href="layer_lambda.html">layer_lambda</a>()</code> <br><code><a href="layer_layer_normalization.html">layer_layer_normalization</a>()</code> <br><code><a href="layer_lstm.html">layer_lstm</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_max_pooling_1d.html">layer_max_pooling_1d</a>()</code> <br><code><a href="layer_max_pooling_2d.html">layer_max_pooling_2d</a>()</code> <br><code><a href="layer_max_pooling_3d.html">layer_max_pooling_3d</a>()</code> <br><code><a href="layer_maximum.html">layer_maximum</a>()</code> <br><code><a href="layer_mel_spectrogram.html">layer_mel_spectrogram</a>()</code> <br><code><a href="layer_minimum.html">layer_minimum</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_multi_head_attention.html">layer_multi_head_attention</a>()</code> <br><code><a href="layer_multiply.html">layer_multiply</a>()</code> <br><code><a href="layer_normalization.html">layer_normalization</a>()</code> <br><code><a href="layer_permute.html">layer_permute</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_brightness.html">layer_random_brightness</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_contrast.html">layer_random_contrast</a>()</code> <br><code><a href="layer_random_crop.html">layer_random_crop</a>()</code> <br><code><a href="layer_random_elastic_transform.html">layer_random_elastic_transform</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_flip.html">layer_random_flip</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_rotation.html">layer_random_rotation</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_random_translation.html">layer_random_translation</a>()</code> <br><code><a href="layer_random_zoom.html">layer_random_zoom</a>()</code> <br><code><a href="layer_repeat_vector.html">layer_repeat_vector</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_reshape.html">layer_reshape</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_rms_normalization.html">layer_rms_normalization</a>()</code> <br><code><a href="layer_rnn.html">layer_rnn</a>()</code> <br><code><a href="layer_separable_conv_1d.html">layer_separable_conv_1d</a>()</code> <br><code><a href="layer_separable_conv_2d.html">layer_separable_conv_2d</a>()</code> <br><code><a href="layer_simple_rnn.html">layer_simple_rnn</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br><code><a href="layer_spatial_dropout_1d.html">layer_spatial_dropout_1d</a>()</code> <br><code><a href="layer_spatial_dropout_2d.html">layer_spatial_dropout_2d</a>()</code> <br><code><a href="layer_spatial_dropout_3d.html">layer_spatial_dropout_3d</a>()</code> <br><code><a href="layer_spectral_normalization.html">layer_spectral_normalization</a>()</code> <br><code><a href="layer_stft_spectrogram.html">layer_stft_spectrogram</a>()</code> <br><code><a href="layer_string_lookup.html">layer_string_lookup</a>()</code> <br><code><a href="layer_subtract.html">layer_subtract</a>()</code> <br><code><a href="layer_text_vectorization.html">layer_text_vectorization</a>()</code> <br><code><a href="layer_tfsm.html">layer_tfsm</a>()</code> <br><code><a href="layer_time_distributed.html">layer_time_distributed</a>()</code> <br><code><a href="layer_torch_module_wrapper.html">layer_torch_module_wrapper</a>()</code> <br><code><a href="layer_unit_normalization.html">layer_unit_normalization</a>()</code> <br><code><a href="layer_upsampling_1d.html">layer_upsampling_1d</a>()</code> <br><code><a href="layer_upsampling_2d.html">layer_upsampling_2d</a>()</code> <br><code><a href="layer_upsampling_3d.html">layer_upsampling_3d</a>()</code> <br><code><a href="layer_zero_padding_1d.html">layer_zero_padding_1d</a>()</code> <br><code><a href="layer_zero_padding_2d.html">layer_zero_padding_2d</a>()</code> <br><code><a href="layer_zero_padding_3d.html">layer_zero_padding_3d</a>()</code> <br><code><a href="rnn_cell_gru.html">rnn_cell_gru</a>()</code> <br><code><a href="rnn_cell_lstm.html">rnn_cell_lstm</a>()</code> <br><code><a href="rnn_cell_simple.html">rnn_cell_simple</a>()</code> <br><code><a href="rnn_cells_stack.html">rnn_cells_stack</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -143,7 +143,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/layer_max_num_bounding_boxes.html b/docs/dev/reference/layer_max_num_bounding_boxes.html
index a5dc29470..e97a468b9 100644
--- a/docs/dev/reference/layer_max_num_bounding_boxes.html
+++ b/docs/dev/reference/layer_max_num_bounding_boxes.html
@@ -8,7 +8,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -94,9 +94,9 @@ <h2 id="value">Value<a class="anchor" aria-label="anchor" href="#value"></a></h2
 </ul></div>
     <div class="section level2">
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
-    <div class="dont-index"><p>Other image preprocessing layers: <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br></p>
-<p>Other preprocessing layers: <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_category_encoding.html">layer_category_encoding</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_discretization.html">layer_discretization</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_feature_space.html">layer_feature_space</a>()</code> <br><code><a href="layer_hashed_crossing.html">layer_hashed_crossing</a>()</code> <br><code><a href="layer_hashing.html">layer_hashing</a>()</code> <br><code><a href="layer_integer_lookup.html">layer_integer_lookup</a>()</code> <br><code><a href="layer_mel_spectrogram.html">layer_mel_spectrogram</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_normalization.html">layer_normalization</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_brightness.html">layer_random_brightness</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_contrast.html">layer_random_contrast</a>()</code> <br><code><a href="layer_random_crop.html">layer_random_crop</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_flip.html">layer_random_flip</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_rotation.html">layer_random_rotation</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_random_translation.html">layer_random_translation</a>()</code> <br><code><a href="layer_random_zoom.html">layer_random_zoom</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br><code><a href="layer_stft_spectrogram.html">layer_stft_spectrogram</a>()</code> <br><code><a href="layer_string_lookup.html">layer_string_lookup</a>()</code> <br><code><a href="layer_text_vectorization.html">layer_text_vectorization</a>()</code> <br></p>
-<p>Other layers: <br><code><a href="Layer.html">Layer</a>()</code> <br><code><a href="layer_activation.html">layer_activation</a>()</code> <br><code><a href="layer_activation_elu.html">layer_activation_elu</a>()</code> <br><code><a href="layer_activation_leaky_relu.html">layer_activation_leaky_relu</a>()</code> <br><code><a href="layer_activation_parametric_relu.html">layer_activation_parametric_relu</a>()</code> <br><code><a href="layer_activation_relu.html">layer_activation_relu</a>()</code> <br><code><a href="layer_activation_softmax.html">layer_activation_softmax</a>()</code> <br><code><a href="layer_activity_regularization.html">layer_activity_regularization</a>()</code> <br><code><a href="layer_add.html">layer_add</a>()</code> <br><code><a href="layer_additive_attention.html">layer_additive_attention</a>()</code> <br><code><a href="layer_alpha_dropout.html">layer_alpha_dropout</a>()</code> <br><code><a href="layer_attention.html">layer_attention</a>()</code> <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_average.html">layer_average</a>()</code> <br><code><a href="layer_average_pooling_1d.html">layer_average_pooling_1d</a>()</code> <br><code><a href="layer_average_pooling_2d.html">layer_average_pooling_2d</a>()</code> <br><code><a href="layer_average_pooling_3d.html">layer_average_pooling_3d</a>()</code> <br><code><a href="layer_batch_normalization.html">layer_batch_normalization</a>()</code> <br><code><a href="layer_bidirectional.html">layer_bidirectional</a>()</code> <br><code><a href="layer_category_encoding.html">layer_category_encoding</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_concatenate.html">layer_concatenate</a>()</code> <br><code><a href="layer_conv_1d.html">layer_conv_1d</a>()</code> <br><code><a href="layer_conv_1d_transpose.html">layer_conv_1d_transpose</a>()</code> <br><code><a href="layer_conv_2d.html">layer_conv_2d</a>()</code> <br><code><a href="layer_conv_2d_transpose.html">layer_conv_2d_transpose</a>()</code> <br><code><a href="layer_conv_3d.html">layer_conv_3d</a>()</code> <br><code><a href="layer_conv_3d_transpose.html">layer_conv_3d_transpose</a>()</code> <br><code><a href="layer_conv_lstm_1d.html">layer_conv_lstm_1d</a>()</code> <br><code><a href="layer_conv_lstm_2d.html">layer_conv_lstm_2d</a>()</code> <br><code><a href="layer_conv_lstm_3d.html">layer_conv_lstm_3d</a>()</code> <br><code><a href="layer_cropping_1d.html">layer_cropping_1d</a>()</code> <br><code><a href="layer_cropping_2d.html">layer_cropping_2d</a>()</code> <br><code><a href="layer_cropping_3d.html">layer_cropping_3d</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_dense.html">layer_dense</a>()</code> <br><code><a href="layer_depthwise_conv_1d.html">layer_depthwise_conv_1d</a>()</code> <br><code><a href="layer_depthwise_conv_2d.html">layer_depthwise_conv_2d</a>()</code> <br><code><a href="layer_discretization.html">layer_discretization</a>()</code> <br><code><a href="layer_dot.html">layer_dot</a>()</code> <br><code><a href="layer_dropout.html">layer_dropout</a>()</code> <br><code><a href="layer_einsum_dense.html">layer_einsum_dense</a>()</code> <br><code><a href="layer_embedding.html">layer_embedding</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_feature_space.html">layer_feature_space</a>()</code> <br><code><a href="layer_flatten.html">layer_flatten</a>()</code> <br><code><a href="layer_flax_module_wrapper.html">layer_flax_module_wrapper</a>()</code> <br><code><a href="layer_gaussian_dropout.html">layer_gaussian_dropout</a>()</code> <br><code><a href="layer_gaussian_noise.html">layer_gaussian_noise</a>()</code> <br><code><a href="layer_global_average_pooling_1d.html">layer_global_average_pooling_1d</a>()</code> <br><code><a href="layer_global_average_pooling_2d.html">layer_global_average_pooling_2d</a>()</code> <br><code><a href="layer_global_average_pooling_3d.html">layer_global_average_pooling_3d</a>()</code> <br><code><a href="layer_global_max_pooling_1d.html">layer_global_max_pooling_1d</a>()</code> <br><code><a href="layer_global_max_pooling_2d.html">layer_global_max_pooling_2d</a>()</code> <br><code><a href="layer_global_max_pooling_3d.html">layer_global_max_pooling_3d</a>()</code> <br><code><a href="layer_group_normalization.html">layer_group_normalization</a>()</code> <br><code><a href="layer_group_query_attention.html">layer_group_query_attention</a>()</code> <br><code><a href="layer_gru.html">layer_gru</a>()</code> <br><code><a href="layer_hashed_crossing.html">layer_hashed_crossing</a>()</code> <br><code><a href="layer_hashing.html">layer_hashing</a>()</code> <br><code><a href="layer_identity.html">layer_identity</a>()</code> <br><code><a href="layer_integer_lookup.html">layer_integer_lookup</a>()</code> <br><code><a href="layer_jax_model_wrapper.html">layer_jax_model_wrapper</a>()</code> <br><code><a href="layer_lambda.html">layer_lambda</a>()</code> <br><code><a href="layer_layer_normalization.html">layer_layer_normalization</a>()</code> <br><code><a href="layer_lstm.html">layer_lstm</a>()</code> <br><code><a href="layer_masking.html">layer_masking</a>()</code> <br><code><a href="layer_max_pooling_1d.html">layer_max_pooling_1d</a>()</code> <br><code><a href="layer_max_pooling_2d.html">layer_max_pooling_2d</a>()</code> <br><code><a href="layer_max_pooling_3d.html">layer_max_pooling_3d</a>()</code> <br><code><a href="layer_maximum.html">layer_maximum</a>()</code> <br><code><a href="layer_mel_spectrogram.html">layer_mel_spectrogram</a>()</code> <br><code><a href="layer_minimum.html">layer_minimum</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_multi_head_attention.html">layer_multi_head_attention</a>()</code> <br><code><a href="layer_multiply.html">layer_multiply</a>()</code> <br><code><a href="layer_normalization.html">layer_normalization</a>()</code> <br><code><a href="layer_permute.html">layer_permute</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_brightness.html">layer_random_brightness</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_contrast.html">layer_random_contrast</a>()</code> <br><code><a href="layer_random_crop.html">layer_random_crop</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_flip.html">layer_random_flip</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_rotation.html">layer_random_rotation</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_random_translation.html">layer_random_translation</a>()</code> <br><code><a href="layer_random_zoom.html">layer_random_zoom</a>()</code> <br><code><a href="layer_repeat_vector.html">layer_repeat_vector</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_reshape.html">layer_reshape</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_rms_normalization.html">layer_rms_normalization</a>()</code> <br><code><a href="layer_rnn.html">layer_rnn</a>()</code> <br><code><a href="layer_separable_conv_1d.html">layer_separable_conv_1d</a>()</code> <br><code><a href="layer_separable_conv_2d.html">layer_separable_conv_2d</a>()</code> <br><code><a href="layer_simple_rnn.html">layer_simple_rnn</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br><code><a href="layer_spatial_dropout_1d.html">layer_spatial_dropout_1d</a>()</code> <br><code><a href="layer_spatial_dropout_2d.html">layer_spatial_dropout_2d</a>()</code> <br><code><a href="layer_spatial_dropout_3d.html">layer_spatial_dropout_3d</a>()</code> <br><code><a href="layer_spectral_normalization.html">layer_spectral_normalization</a>()</code> <br><code><a href="layer_stft_spectrogram.html">layer_stft_spectrogram</a>()</code> <br><code><a href="layer_string_lookup.html">layer_string_lookup</a>()</code> <br><code><a href="layer_subtract.html">layer_subtract</a>()</code> <br><code><a href="layer_text_vectorization.html">layer_text_vectorization</a>()</code> <br><code><a href="layer_tfsm.html">layer_tfsm</a>()</code> <br><code><a href="layer_time_distributed.html">layer_time_distributed</a>()</code> <br><code><a href="layer_torch_module_wrapper.html">layer_torch_module_wrapper</a>()</code> <br><code><a href="layer_unit_normalization.html">layer_unit_normalization</a>()</code> <br><code><a href="layer_upsampling_1d.html">layer_upsampling_1d</a>()</code> <br><code><a href="layer_upsampling_2d.html">layer_upsampling_2d</a>()</code> <br><code><a href="layer_upsampling_3d.html">layer_upsampling_3d</a>()</code> <br><code><a href="layer_zero_padding_1d.html">layer_zero_padding_1d</a>()</code> <br><code><a href="layer_zero_padding_2d.html">layer_zero_padding_2d</a>()</code> <br><code><a href="layer_zero_padding_3d.html">layer_zero_padding_3d</a>()</code> <br><code><a href="rnn_cell_gru.html">rnn_cell_gru</a>()</code> <br><code><a href="rnn_cell_lstm.html">rnn_cell_lstm</a>()</code> <br><code><a href="rnn_cell_simple.html">rnn_cell_simple</a>()</code> <br><code><a href="rnn_cells_stack.html">rnn_cells_stack</a>()</code> <br></p></div>
+    <div class="dont-index"><p>Other image preprocessing layers: <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_elastic_transform.html">layer_random_elastic_transform</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br></p>
+<p>Other preprocessing layers: <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_category_encoding.html">layer_category_encoding</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_discretization.html">layer_discretization</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_feature_space.html">layer_feature_space</a>()</code> <br><code><a href="layer_hashed_crossing.html">layer_hashed_crossing</a>()</code> <br><code><a href="layer_hashing.html">layer_hashing</a>()</code> <br><code><a href="layer_integer_lookup.html">layer_integer_lookup</a>()</code> <br><code><a href="layer_mel_spectrogram.html">layer_mel_spectrogram</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_normalization.html">layer_normalization</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_brightness.html">layer_random_brightness</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_contrast.html">layer_random_contrast</a>()</code> <br><code><a href="layer_random_crop.html">layer_random_crop</a>()</code> <br><code><a href="layer_random_elastic_transform.html">layer_random_elastic_transform</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_flip.html">layer_random_flip</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_rotation.html">layer_random_rotation</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_random_translation.html">layer_random_translation</a>()</code> <br><code><a href="layer_random_zoom.html">layer_random_zoom</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br><code><a href="layer_stft_spectrogram.html">layer_stft_spectrogram</a>()</code> <br><code><a href="layer_string_lookup.html">layer_string_lookup</a>()</code> <br><code><a href="layer_text_vectorization.html">layer_text_vectorization</a>()</code> <br></p>
+<p>Other layers: <br><code><a href="Layer.html">Layer</a>()</code> <br><code><a href="layer_activation.html">layer_activation</a>()</code> <br><code><a href="layer_activation_elu.html">layer_activation_elu</a>()</code> <br><code><a href="layer_activation_leaky_relu.html">layer_activation_leaky_relu</a>()</code> <br><code><a href="layer_activation_parametric_relu.html">layer_activation_parametric_relu</a>()</code> <br><code><a href="layer_activation_relu.html">layer_activation_relu</a>()</code> <br><code><a href="layer_activation_softmax.html">layer_activation_softmax</a>()</code> <br><code><a href="layer_activity_regularization.html">layer_activity_regularization</a>()</code> <br><code><a href="layer_add.html">layer_add</a>()</code> <br><code><a href="layer_additive_attention.html">layer_additive_attention</a>()</code> <br><code><a href="layer_alpha_dropout.html">layer_alpha_dropout</a>()</code> <br><code><a href="layer_attention.html">layer_attention</a>()</code> <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_average.html">layer_average</a>()</code> <br><code><a href="layer_average_pooling_1d.html">layer_average_pooling_1d</a>()</code> <br><code><a href="layer_average_pooling_2d.html">layer_average_pooling_2d</a>()</code> <br><code><a href="layer_average_pooling_3d.html">layer_average_pooling_3d</a>()</code> <br><code><a href="layer_batch_normalization.html">layer_batch_normalization</a>()</code> <br><code><a href="layer_bidirectional.html">layer_bidirectional</a>()</code> <br><code><a href="layer_category_encoding.html">layer_category_encoding</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_concatenate.html">layer_concatenate</a>()</code> <br><code><a href="layer_conv_1d.html">layer_conv_1d</a>()</code> <br><code><a href="layer_conv_1d_transpose.html">layer_conv_1d_transpose</a>()</code> <br><code><a href="layer_conv_2d.html">layer_conv_2d</a>()</code> <br><code><a href="layer_conv_2d_transpose.html">layer_conv_2d_transpose</a>()</code> <br><code><a href="layer_conv_3d.html">layer_conv_3d</a>()</code> <br><code><a href="layer_conv_3d_transpose.html">layer_conv_3d_transpose</a>()</code> <br><code><a href="layer_conv_lstm_1d.html">layer_conv_lstm_1d</a>()</code> <br><code><a href="layer_conv_lstm_2d.html">layer_conv_lstm_2d</a>()</code> <br><code><a href="layer_conv_lstm_3d.html">layer_conv_lstm_3d</a>()</code> <br><code><a href="layer_cropping_1d.html">layer_cropping_1d</a>()</code> <br><code><a href="layer_cropping_2d.html">layer_cropping_2d</a>()</code> <br><code><a href="layer_cropping_3d.html">layer_cropping_3d</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_dense.html">layer_dense</a>()</code> <br><code><a href="layer_depthwise_conv_1d.html">layer_depthwise_conv_1d</a>()</code> <br><code><a href="layer_depthwise_conv_2d.html">layer_depthwise_conv_2d</a>()</code> <br><code><a href="layer_discretization.html">layer_discretization</a>()</code> <br><code><a href="layer_dot.html">layer_dot</a>()</code> <br><code><a href="layer_dropout.html">layer_dropout</a>()</code> <br><code><a href="layer_einsum_dense.html">layer_einsum_dense</a>()</code> <br><code><a href="layer_embedding.html">layer_embedding</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_feature_space.html">layer_feature_space</a>()</code> <br><code><a href="layer_flatten.html">layer_flatten</a>()</code> <br><code><a href="layer_flax_module_wrapper.html">layer_flax_module_wrapper</a>()</code> <br><code><a href="layer_gaussian_dropout.html">layer_gaussian_dropout</a>()</code> <br><code><a href="layer_gaussian_noise.html">layer_gaussian_noise</a>()</code> <br><code><a href="layer_global_average_pooling_1d.html">layer_global_average_pooling_1d</a>()</code> <br><code><a href="layer_global_average_pooling_2d.html">layer_global_average_pooling_2d</a>()</code> <br><code><a href="layer_global_average_pooling_3d.html">layer_global_average_pooling_3d</a>()</code> <br><code><a href="layer_global_max_pooling_1d.html">layer_global_max_pooling_1d</a>()</code> <br><code><a href="layer_global_max_pooling_2d.html">layer_global_max_pooling_2d</a>()</code> <br><code><a href="layer_global_max_pooling_3d.html">layer_global_max_pooling_3d</a>()</code> <br><code><a href="layer_group_normalization.html">layer_group_normalization</a>()</code> <br><code><a href="layer_group_query_attention.html">layer_group_query_attention</a>()</code> <br><code><a href="layer_gru.html">layer_gru</a>()</code> <br><code><a href="layer_hashed_crossing.html">layer_hashed_crossing</a>()</code> <br><code><a href="layer_hashing.html">layer_hashing</a>()</code> <br><code><a href="layer_identity.html">layer_identity</a>()</code> <br><code><a href="layer_integer_lookup.html">layer_integer_lookup</a>()</code> <br><code><a href="layer_jax_model_wrapper.html">layer_jax_model_wrapper</a>()</code> <br><code><a href="layer_lambda.html">layer_lambda</a>()</code> <br><code><a href="layer_layer_normalization.html">layer_layer_normalization</a>()</code> <br><code><a href="layer_lstm.html">layer_lstm</a>()</code> <br><code><a href="layer_masking.html">layer_masking</a>()</code> <br><code><a href="layer_max_pooling_1d.html">layer_max_pooling_1d</a>()</code> <br><code><a href="layer_max_pooling_2d.html">layer_max_pooling_2d</a>()</code> <br><code><a href="layer_max_pooling_3d.html">layer_max_pooling_3d</a>()</code> <br><code><a href="layer_maximum.html">layer_maximum</a>()</code> <br><code><a href="layer_mel_spectrogram.html">layer_mel_spectrogram</a>()</code> <br><code><a href="layer_minimum.html">layer_minimum</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_multi_head_attention.html">layer_multi_head_attention</a>()</code> <br><code><a href="layer_multiply.html">layer_multiply</a>()</code> <br><code><a href="layer_normalization.html">layer_normalization</a>()</code> <br><code><a href="layer_permute.html">layer_permute</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_brightness.html">layer_random_brightness</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_contrast.html">layer_random_contrast</a>()</code> <br><code><a href="layer_random_crop.html">layer_random_crop</a>()</code> <br><code><a href="layer_random_elastic_transform.html">layer_random_elastic_transform</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_flip.html">layer_random_flip</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_rotation.html">layer_random_rotation</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_random_translation.html">layer_random_translation</a>()</code> <br><code><a href="layer_random_zoom.html">layer_random_zoom</a>()</code> <br><code><a href="layer_repeat_vector.html">layer_repeat_vector</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_reshape.html">layer_reshape</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_rms_normalization.html">layer_rms_normalization</a>()</code> <br><code><a href="layer_rnn.html">layer_rnn</a>()</code> <br><code><a href="layer_separable_conv_1d.html">layer_separable_conv_1d</a>()</code> <br><code><a href="layer_separable_conv_2d.html">layer_separable_conv_2d</a>()</code> <br><code><a href="layer_simple_rnn.html">layer_simple_rnn</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br><code><a href="layer_spatial_dropout_1d.html">layer_spatial_dropout_1d</a>()</code> <br><code><a href="layer_spatial_dropout_2d.html">layer_spatial_dropout_2d</a>()</code> <br><code><a href="layer_spatial_dropout_3d.html">layer_spatial_dropout_3d</a>()</code> <br><code><a href="layer_spectral_normalization.html">layer_spectral_normalization</a>()</code> <br><code><a href="layer_stft_spectrogram.html">layer_stft_spectrogram</a>()</code> <br><code><a href="layer_string_lookup.html">layer_string_lookup</a>()</code> <br><code><a href="layer_subtract.html">layer_subtract</a>()</code> <br><code><a href="layer_text_vectorization.html">layer_text_vectorization</a>()</code> <br><code><a href="layer_tfsm.html">layer_tfsm</a>()</code> <br><code><a href="layer_time_distributed.html">layer_time_distributed</a>()</code> <br><code><a href="layer_torch_module_wrapper.html">layer_torch_module_wrapper</a>()</code> <br><code><a href="layer_unit_normalization.html">layer_unit_normalization</a>()</code> <br><code><a href="layer_upsampling_1d.html">layer_upsampling_1d</a>()</code> <br><code><a href="layer_upsampling_2d.html">layer_upsampling_2d</a>()</code> <br><code><a href="layer_upsampling_3d.html">layer_upsampling_3d</a>()</code> <br><code><a href="layer_zero_padding_1d.html">layer_zero_padding_1d</a>()</code> <br><code><a href="layer_zero_padding_2d.html">layer_zero_padding_2d</a>()</code> <br><code><a href="layer_zero_padding_3d.html">layer_zero_padding_3d</a>()</code> <br><code><a href="rnn_cell_gru.html">rnn_cell_gru</a>()</code> <br><code><a href="rnn_cell_lstm.html">rnn_cell_lstm</a>()</code> <br><code><a href="rnn_cell_simple.html">rnn_cell_simple</a>()</code> <br><code><a href="rnn_cells_stack.html">rnn_cells_stack</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -108,7 +108,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/layer_max_pooling_1d.html b/docs/dev/reference/layer_max_pooling_1d.html
index 45227a26a..10303ae0e 100644
--- a/docs/dev/reference/layer_max_pooling_1d.html
+++ b/docs/dev/reference/layer_max_pooling_1d.html
@@ -18,7 +18,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -198,7 +198,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
     <div class="dont-index">
 <ul><li><p><a href="https://keras.io/api/layers/pooling_layers/max_pooling1d#maxpooling1d-class" class="external-link">https://keras.io/api/layers/pooling_layers/max_pooling1d#maxpooling1d-class</a></p></li>
 </ul><p>Other pooling layers: <br><code><a href="layer_average_pooling_1d.html">layer_average_pooling_1d</a>()</code> <br><code><a href="layer_average_pooling_2d.html">layer_average_pooling_2d</a>()</code> <br><code><a href="layer_average_pooling_3d.html">layer_average_pooling_3d</a>()</code> <br><code><a href="layer_global_average_pooling_1d.html">layer_global_average_pooling_1d</a>()</code> <br><code><a href="layer_global_average_pooling_2d.html">layer_global_average_pooling_2d</a>()</code> <br><code><a href="layer_global_average_pooling_3d.html">layer_global_average_pooling_3d</a>()</code> <br><code><a href="layer_global_max_pooling_1d.html">layer_global_max_pooling_1d</a>()</code> <br><code><a href="layer_global_max_pooling_2d.html">layer_global_max_pooling_2d</a>()</code> <br><code><a href="layer_global_max_pooling_3d.html">layer_global_max_pooling_3d</a>()</code> <br><code><a href="layer_max_pooling_2d.html">layer_max_pooling_2d</a>()</code> <br><code><a href="layer_max_pooling_3d.html">layer_max_pooling_3d</a>()</code> <br></p>
-<p>Other layers: <br><code><a href="Layer.html">Layer</a>()</code> <br><code><a href="layer_activation.html">layer_activation</a>()</code> <br><code><a href="layer_activation_elu.html">layer_activation_elu</a>()</code> <br><code><a href="layer_activation_leaky_relu.html">layer_activation_leaky_relu</a>()</code> <br><code><a href="layer_activation_parametric_relu.html">layer_activation_parametric_relu</a>()</code> <br><code><a href="layer_activation_relu.html">layer_activation_relu</a>()</code> <br><code><a href="layer_activation_softmax.html">layer_activation_softmax</a>()</code> <br><code><a href="layer_activity_regularization.html">layer_activity_regularization</a>()</code> <br><code><a href="layer_add.html">layer_add</a>()</code> <br><code><a href="layer_additive_attention.html">layer_additive_attention</a>()</code> <br><code><a href="layer_alpha_dropout.html">layer_alpha_dropout</a>()</code> <br><code><a href="layer_attention.html">layer_attention</a>()</code> <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_average.html">layer_average</a>()</code> <br><code><a href="layer_average_pooling_1d.html">layer_average_pooling_1d</a>()</code> <br><code><a href="layer_average_pooling_2d.html">layer_average_pooling_2d</a>()</code> <br><code><a href="layer_average_pooling_3d.html">layer_average_pooling_3d</a>()</code> <br><code><a href="layer_batch_normalization.html">layer_batch_normalization</a>()</code> <br><code><a href="layer_bidirectional.html">layer_bidirectional</a>()</code> <br><code><a href="layer_category_encoding.html">layer_category_encoding</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_concatenate.html">layer_concatenate</a>()</code> <br><code><a href="layer_conv_1d.html">layer_conv_1d</a>()</code> <br><code><a href="layer_conv_1d_transpose.html">layer_conv_1d_transpose</a>()</code> <br><code><a href="layer_conv_2d.html">layer_conv_2d</a>()</code> <br><code><a href="layer_conv_2d_transpose.html">layer_conv_2d_transpose</a>()</code> <br><code><a href="layer_conv_3d.html">layer_conv_3d</a>()</code> <br><code><a href="layer_conv_3d_transpose.html">layer_conv_3d_transpose</a>()</code> <br><code><a href="layer_conv_lstm_1d.html">layer_conv_lstm_1d</a>()</code> <br><code><a href="layer_conv_lstm_2d.html">layer_conv_lstm_2d</a>()</code> <br><code><a href="layer_conv_lstm_3d.html">layer_conv_lstm_3d</a>()</code> <br><code><a href="layer_cropping_1d.html">layer_cropping_1d</a>()</code> <br><code><a href="layer_cropping_2d.html">layer_cropping_2d</a>()</code> <br><code><a href="layer_cropping_3d.html">layer_cropping_3d</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_dense.html">layer_dense</a>()</code> <br><code><a href="layer_depthwise_conv_1d.html">layer_depthwise_conv_1d</a>()</code> <br><code><a href="layer_depthwise_conv_2d.html">layer_depthwise_conv_2d</a>()</code> <br><code><a href="layer_discretization.html">layer_discretization</a>()</code> <br><code><a href="layer_dot.html">layer_dot</a>()</code> <br><code><a href="layer_dropout.html">layer_dropout</a>()</code> <br><code><a href="layer_einsum_dense.html">layer_einsum_dense</a>()</code> <br><code><a href="layer_embedding.html">layer_embedding</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_feature_space.html">layer_feature_space</a>()</code> <br><code><a href="layer_flatten.html">layer_flatten</a>()</code> <br><code><a href="layer_flax_module_wrapper.html">layer_flax_module_wrapper</a>()</code> <br><code><a href="layer_gaussian_dropout.html">layer_gaussian_dropout</a>()</code> <br><code><a href="layer_gaussian_noise.html">layer_gaussian_noise</a>()</code> <br><code><a href="layer_global_average_pooling_1d.html">layer_global_average_pooling_1d</a>()</code> <br><code><a href="layer_global_average_pooling_2d.html">layer_global_average_pooling_2d</a>()</code> <br><code><a href="layer_global_average_pooling_3d.html">layer_global_average_pooling_3d</a>()</code> <br><code><a href="layer_global_max_pooling_1d.html">layer_global_max_pooling_1d</a>()</code> <br><code><a href="layer_global_max_pooling_2d.html">layer_global_max_pooling_2d</a>()</code> <br><code><a href="layer_global_max_pooling_3d.html">layer_global_max_pooling_3d</a>()</code> <br><code><a href="layer_group_normalization.html">layer_group_normalization</a>()</code> <br><code><a href="layer_group_query_attention.html">layer_group_query_attention</a>()</code> <br><code><a href="layer_gru.html">layer_gru</a>()</code> <br><code><a href="layer_hashed_crossing.html">layer_hashed_crossing</a>()</code> <br><code><a href="layer_hashing.html">layer_hashing</a>()</code> <br><code><a href="layer_identity.html">layer_identity</a>()</code> <br><code><a href="layer_integer_lookup.html">layer_integer_lookup</a>()</code> <br><code><a href="layer_jax_model_wrapper.html">layer_jax_model_wrapper</a>()</code> <br><code><a href="layer_lambda.html">layer_lambda</a>()</code> <br><code><a href="layer_layer_normalization.html">layer_layer_normalization</a>()</code> <br><code><a href="layer_lstm.html">layer_lstm</a>()</code> <br><code><a href="layer_masking.html">layer_masking</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_max_pooling_2d.html">layer_max_pooling_2d</a>()</code> <br><code><a href="layer_max_pooling_3d.html">layer_max_pooling_3d</a>()</code> <br><code><a href="layer_maximum.html">layer_maximum</a>()</code> <br><code><a href="layer_mel_spectrogram.html">layer_mel_spectrogram</a>()</code> <br><code><a href="layer_minimum.html">layer_minimum</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_multi_head_attention.html">layer_multi_head_attention</a>()</code> <br><code><a href="layer_multiply.html">layer_multiply</a>()</code> <br><code><a href="layer_normalization.html">layer_normalization</a>()</code> <br><code><a href="layer_permute.html">layer_permute</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_brightness.html">layer_random_brightness</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_contrast.html">layer_random_contrast</a>()</code> <br><code><a href="layer_random_crop.html">layer_random_crop</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_flip.html">layer_random_flip</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_rotation.html">layer_random_rotation</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_random_translation.html">layer_random_translation</a>()</code> <br><code><a href="layer_random_zoom.html">layer_random_zoom</a>()</code> <br><code><a href="layer_repeat_vector.html">layer_repeat_vector</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_reshape.html">layer_reshape</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_rms_normalization.html">layer_rms_normalization</a>()</code> <br><code><a href="layer_rnn.html">layer_rnn</a>()</code> <br><code><a href="layer_separable_conv_1d.html">layer_separable_conv_1d</a>()</code> <br><code><a href="layer_separable_conv_2d.html">layer_separable_conv_2d</a>()</code> <br><code><a href="layer_simple_rnn.html">layer_simple_rnn</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br><code><a href="layer_spatial_dropout_1d.html">layer_spatial_dropout_1d</a>()</code> <br><code><a href="layer_spatial_dropout_2d.html">layer_spatial_dropout_2d</a>()</code> <br><code><a href="layer_spatial_dropout_3d.html">layer_spatial_dropout_3d</a>()</code> <br><code><a href="layer_spectral_normalization.html">layer_spectral_normalization</a>()</code> <br><code><a href="layer_stft_spectrogram.html">layer_stft_spectrogram</a>()</code> <br><code><a href="layer_string_lookup.html">layer_string_lookup</a>()</code> <br><code><a href="layer_subtract.html">layer_subtract</a>()</code> <br><code><a href="layer_text_vectorization.html">layer_text_vectorization</a>()</code> <br><code><a href="layer_tfsm.html">layer_tfsm</a>()</code> <br><code><a href="layer_time_distributed.html">layer_time_distributed</a>()</code> <br><code><a href="layer_torch_module_wrapper.html">layer_torch_module_wrapper</a>()</code> <br><code><a href="layer_unit_normalization.html">layer_unit_normalization</a>()</code> <br><code><a href="layer_upsampling_1d.html">layer_upsampling_1d</a>()</code> <br><code><a href="layer_upsampling_2d.html">layer_upsampling_2d</a>()</code> <br><code><a href="layer_upsampling_3d.html">layer_upsampling_3d</a>()</code> <br><code><a href="layer_zero_padding_1d.html">layer_zero_padding_1d</a>()</code> <br><code><a href="layer_zero_padding_2d.html">layer_zero_padding_2d</a>()</code> <br><code><a href="layer_zero_padding_3d.html">layer_zero_padding_3d</a>()</code> <br><code><a href="rnn_cell_gru.html">rnn_cell_gru</a>()</code> <br><code><a href="rnn_cell_lstm.html">rnn_cell_lstm</a>()</code> <br><code><a href="rnn_cell_simple.html">rnn_cell_simple</a>()</code> <br><code><a href="rnn_cells_stack.html">rnn_cells_stack</a>()</code> <br></p></div>
+<p>Other layers: <br><code><a href="Layer.html">Layer</a>()</code> <br><code><a href="layer_activation.html">layer_activation</a>()</code> <br><code><a href="layer_activation_elu.html">layer_activation_elu</a>()</code> <br><code><a href="layer_activation_leaky_relu.html">layer_activation_leaky_relu</a>()</code> <br><code><a href="layer_activation_parametric_relu.html">layer_activation_parametric_relu</a>()</code> <br><code><a href="layer_activation_relu.html">layer_activation_relu</a>()</code> <br><code><a href="layer_activation_softmax.html">layer_activation_softmax</a>()</code> <br><code><a href="layer_activity_regularization.html">layer_activity_regularization</a>()</code> <br><code><a href="layer_add.html">layer_add</a>()</code> <br><code><a href="layer_additive_attention.html">layer_additive_attention</a>()</code> <br><code><a href="layer_alpha_dropout.html">layer_alpha_dropout</a>()</code> <br><code><a href="layer_attention.html">layer_attention</a>()</code> <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_average.html">layer_average</a>()</code> <br><code><a href="layer_average_pooling_1d.html">layer_average_pooling_1d</a>()</code> <br><code><a href="layer_average_pooling_2d.html">layer_average_pooling_2d</a>()</code> <br><code><a href="layer_average_pooling_3d.html">layer_average_pooling_3d</a>()</code> <br><code><a href="layer_batch_normalization.html">layer_batch_normalization</a>()</code> <br><code><a href="layer_bidirectional.html">layer_bidirectional</a>()</code> <br><code><a href="layer_category_encoding.html">layer_category_encoding</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_concatenate.html">layer_concatenate</a>()</code> <br><code><a href="layer_conv_1d.html">layer_conv_1d</a>()</code> <br><code><a href="layer_conv_1d_transpose.html">layer_conv_1d_transpose</a>()</code> <br><code><a href="layer_conv_2d.html">layer_conv_2d</a>()</code> <br><code><a href="layer_conv_2d_transpose.html">layer_conv_2d_transpose</a>()</code> <br><code><a href="layer_conv_3d.html">layer_conv_3d</a>()</code> <br><code><a href="layer_conv_3d_transpose.html">layer_conv_3d_transpose</a>()</code> <br><code><a href="layer_conv_lstm_1d.html">layer_conv_lstm_1d</a>()</code> <br><code><a href="layer_conv_lstm_2d.html">layer_conv_lstm_2d</a>()</code> <br><code><a href="layer_conv_lstm_3d.html">layer_conv_lstm_3d</a>()</code> <br><code><a href="layer_cropping_1d.html">layer_cropping_1d</a>()</code> <br><code><a href="layer_cropping_2d.html">layer_cropping_2d</a>()</code> <br><code><a href="layer_cropping_3d.html">layer_cropping_3d</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_dense.html">layer_dense</a>()</code> <br><code><a href="layer_depthwise_conv_1d.html">layer_depthwise_conv_1d</a>()</code> <br><code><a href="layer_depthwise_conv_2d.html">layer_depthwise_conv_2d</a>()</code> <br><code><a href="layer_discretization.html">layer_discretization</a>()</code> <br><code><a href="layer_dot.html">layer_dot</a>()</code> <br><code><a href="layer_dropout.html">layer_dropout</a>()</code> <br><code><a href="layer_einsum_dense.html">layer_einsum_dense</a>()</code> <br><code><a href="layer_embedding.html">layer_embedding</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_feature_space.html">layer_feature_space</a>()</code> <br><code><a href="layer_flatten.html">layer_flatten</a>()</code> <br><code><a href="layer_flax_module_wrapper.html">layer_flax_module_wrapper</a>()</code> <br><code><a href="layer_gaussian_dropout.html">layer_gaussian_dropout</a>()</code> <br><code><a href="layer_gaussian_noise.html">layer_gaussian_noise</a>()</code> <br><code><a href="layer_global_average_pooling_1d.html">layer_global_average_pooling_1d</a>()</code> <br><code><a href="layer_global_average_pooling_2d.html">layer_global_average_pooling_2d</a>()</code> <br><code><a href="layer_global_average_pooling_3d.html">layer_global_average_pooling_3d</a>()</code> <br><code><a href="layer_global_max_pooling_1d.html">layer_global_max_pooling_1d</a>()</code> <br><code><a href="layer_global_max_pooling_2d.html">layer_global_max_pooling_2d</a>()</code> <br><code><a href="layer_global_max_pooling_3d.html">layer_global_max_pooling_3d</a>()</code> <br><code><a href="layer_group_normalization.html">layer_group_normalization</a>()</code> <br><code><a href="layer_group_query_attention.html">layer_group_query_attention</a>()</code> <br><code><a href="layer_gru.html">layer_gru</a>()</code> <br><code><a href="layer_hashed_crossing.html">layer_hashed_crossing</a>()</code> <br><code><a href="layer_hashing.html">layer_hashing</a>()</code> <br><code><a href="layer_identity.html">layer_identity</a>()</code> <br><code><a href="layer_integer_lookup.html">layer_integer_lookup</a>()</code> <br><code><a href="layer_jax_model_wrapper.html">layer_jax_model_wrapper</a>()</code> <br><code><a href="layer_lambda.html">layer_lambda</a>()</code> <br><code><a href="layer_layer_normalization.html">layer_layer_normalization</a>()</code> <br><code><a href="layer_lstm.html">layer_lstm</a>()</code> <br><code><a href="layer_masking.html">layer_masking</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_max_pooling_2d.html">layer_max_pooling_2d</a>()</code> <br><code><a href="layer_max_pooling_3d.html">layer_max_pooling_3d</a>()</code> <br><code><a href="layer_maximum.html">layer_maximum</a>()</code> <br><code><a href="layer_mel_spectrogram.html">layer_mel_spectrogram</a>()</code> <br><code><a href="layer_minimum.html">layer_minimum</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_multi_head_attention.html">layer_multi_head_attention</a>()</code> <br><code><a href="layer_multiply.html">layer_multiply</a>()</code> <br><code><a href="layer_normalization.html">layer_normalization</a>()</code> <br><code><a href="layer_permute.html">layer_permute</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_brightness.html">layer_random_brightness</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_contrast.html">layer_random_contrast</a>()</code> <br><code><a href="layer_random_crop.html">layer_random_crop</a>()</code> <br><code><a href="layer_random_elastic_transform.html">layer_random_elastic_transform</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_flip.html">layer_random_flip</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_rotation.html">layer_random_rotation</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_random_translation.html">layer_random_translation</a>()</code> <br><code><a href="layer_random_zoom.html">layer_random_zoom</a>()</code> <br><code><a href="layer_repeat_vector.html">layer_repeat_vector</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_reshape.html">layer_reshape</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_rms_normalization.html">layer_rms_normalization</a>()</code> <br><code><a href="layer_rnn.html">layer_rnn</a>()</code> <br><code><a href="layer_separable_conv_1d.html">layer_separable_conv_1d</a>()</code> <br><code><a href="layer_separable_conv_2d.html">layer_separable_conv_2d</a>()</code> <br><code><a href="layer_simple_rnn.html">layer_simple_rnn</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br><code><a href="layer_spatial_dropout_1d.html">layer_spatial_dropout_1d</a>()</code> <br><code><a href="layer_spatial_dropout_2d.html">layer_spatial_dropout_2d</a>()</code> <br><code><a href="layer_spatial_dropout_3d.html">layer_spatial_dropout_3d</a>()</code> <br><code><a href="layer_spectral_normalization.html">layer_spectral_normalization</a>()</code> <br><code><a href="layer_stft_spectrogram.html">layer_stft_spectrogram</a>()</code> <br><code><a href="layer_string_lookup.html">layer_string_lookup</a>()</code> <br><code><a href="layer_subtract.html">layer_subtract</a>()</code> <br><code><a href="layer_text_vectorization.html">layer_text_vectorization</a>()</code> <br><code><a href="layer_tfsm.html">layer_tfsm</a>()</code> <br><code><a href="layer_time_distributed.html">layer_time_distributed</a>()</code> <br><code><a href="layer_torch_module_wrapper.html">layer_torch_module_wrapper</a>()</code> <br><code><a href="layer_unit_normalization.html">layer_unit_normalization</a>()</code> <br><code><a href="layer_upsampling_1d.html">layer_upsampling_1d</a>()</code> <br><code><a href="layer_upsampling_2d.html">layer_upsampling_2d</a>()</code> <br><code><a href="layer_upsampling_3d.html">layer_upsampling_3d</a>()</code> <br><code><a href="layer_zero_padding_1d.html">layer_zero_padding_1d</a>()</code> <br><code><a href="layer_zero_padding_2d.html">layer_zero_padding_2d</a>()</code> <br><code><a href="layer_zero_padding_3d.html">layer_zero_padding_3d</a>()</code> <br><code><a href="rnn_cell_gru.html">rnn_cell_gru</a>()</code> <br><code><a href="rnn_cell_lstm.html">rnn_cell_lstm</a>()</code> <br><code><a href="rnn_cell_simple.html">rnn_cell_simple</a>()</code> <br><code><a href="rnn_cells_stack.html">rnn_cells_stack</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -210,7 +210,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/layer_max_pooling_2d.html b/docs/dev/reference/layer_max_pooling_2d.html
index 4d8cd83cd..d0af17c5c 100644
--- a/docs/dev/reference/layer_max_pooling_2d.html
+++ b/docs/dev/reference/layer_max_pooling_2d.html
@@ -26,7 +26,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -226,7 +226,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
     <div class="dont-index">
 <ul><li><p><a href="https://keras.io/api/layers/pooling_layers/max_pooling2d#maxpooling2d-class" class="external-link">https://keras.io/api/layers/pooling_layers/max_pooling2d#maxpooling2d-class</a></p></li>
 </ul><p>Other pooling layers: <br><code><a href="layer_average_pooling_1d.html">layer_average_pooling_1d</a>()</code> <br><code><a href="layer_average_pooling_2d.html">layer_average_pooling_2d</a>()</code> <br><code><a href="layer_average_pooling_3d.html">layer_average_pooling_3d</a>()</code> <br><code><a href="layer_global_average_pooling_1d.html">layer_global_average_pooling_1d</a>()</code> <br><code><a href="layer_global_average_pooling_2d.html">layer_global_average_pooling_2d</a>()</code> <br><code><a href="layer_global_average_pooling_3d.html">layer_global_average_pooling_3d</a>()</code> <br><code><a href="layer_global_max_pooling_1d.html">layer_global_max_pooling_1d</a>()</code> <br><code><a href="layer_global_max_pooling_2d.html">layer_global_max_pooling_2d</a>()</code> <br><code><a href="layer_global_max_pooling_3d.html">layer_global_max_pooling_3d</a>()</code> <br><code><a href="layer_max_pooling_1d.html">layer_max_pooling_1d</a>()</code> <br><code><a href="layer_max_pooling_3d.html">layer_max_pooling_3d</a>()</code> <br></p>
-<p>Other layers: <br><code><a href="Layer.html">Layer</a>()</code> <br><code><a href="layer_activation.html">layer_activation</a>()</code> <br><code><a href="layer_activation_elu.html">layer_activation_elu</a>()</code> <br><code><a href="layer_activation_leaky_relu.html">layer_activation_leaky_relu</a>()</code> <br><code><a href="layer_activation_parametric_relu.html">layer_activation_parametric_relu</a>()</code> <br><code><a href="layer_activation_relu.html">layer_activation_relu</a>()</code> <br><code><a href="layer_activation_softmax.html">layer_activation_softmax</a>()</code> <br><code><a href="layer_activity_regularization.html">layer_activity_regularization</a>()</code> <br><code><a href="layer_add.html">layer_add</a>()</code> <br><code><a href="layer_additive_attention.html">layer_additive_attention</a>()</code> <br><code><a href="layer_alpha_dropout.html">layer_alpha_dropout</a>()</code> <br><code><a href="layer_attention.html">layer_attention</a>()</code> <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_average.html">layer_average</a>()</code> <br><code><a href="layer_average_pooling_1d.html">layer_average_pooling_1d</a>()</code> <br><code><a href="layer_average_pooling_2d.html">layer_average_pooling_2d</a>()</code> <br><code><a href="layer_average_pooling_3d.html">layer_average_pooling_3d</a>()</code> <br><code><a href="layer_batch_normalization.html">layer_batch_normalization</a>()</code> <br><code><a href="layer_bidirectional.html">layer_bidirectional</a>()</code> <br><code><a href="layer_category_encoding.html">layer_category_encoding</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_concatenate.html">layer_concatenate</a>()</code> <br><code><a href="layer_conv_1d.html">layer_conv_1d</a>()</code> <br><code><a href="layer_conv_1d_transpose.html">layer_conv_1d_transpose</a>()</code> <br><code><a href="layer_conv_2d.html">layer_conv_2d</a>()</code> <br><code><a href="layer_conv_2d_transpose.html">layer_conv_2d_transpose</a>()</code> <br><code><a href="layer_conv_3d.html">layer_conv_3d</a>()</code> <br><code><a href="layer_conv_3d_transpose.html">layer_conv_3d_transpose</a>()</code> <br><code><a href="layer_conv_lstm_1d.html">layer_conv_lstm_1d</a>()</code> <br><code><a href="layer_conv_lstm_2d.html">layer_conv_lstm_2d</a>()</code> <br><code><a href="layer_conv_lstm_3d.html">layer_conv_lstm_3d</a>()</code> <br><code><a href="layer_cropping_1d.html">layer_cropping_1d</a>()</code> <br><code><a href="layer_cropping_2d.html">layer_cropping_2d</a>()</code> <br><code><a href="layer_cropping_3d.html">layer_cropping_3d</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_dense.html">layer_dense</a>()</code> <br><code><a href="layer_depthwise_conv_1d.html">layer_depthwise_conv_1d</a>()</code> <br><code><a href="layer_depthwise_conv_2d.html">layer_depthwise_conv_2d</a>()</code> <br><code><a href="layer_discretization.html">layer_discretization</a>()</code> <br><code><a href="layer_dot.html">layer_dot</a>()</code> <br><code><a href="layer_dropout.html">layer_dropout</a>()</code> <br><code><a href="layer_einsum_dense.html">layer_einsum_dense</a>()</code> <br><code><a href="layer_embedding.html">layer_embedding</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_feature_space.html">layer_feature_space</a>()</code> <br><code><a href="layer_flatten.html">layer_flatten</a>()</code> <br><code><a href="layer_flax_module_wrapper.html">layer_flax_module_wrapper</a>()</code> <br><code><a href="layer_gaussian_dropout.html">layer_gaussian_dropout</a>()</code> <br><code><a href="layer_gaussian_noise.html">layer_gaussian_noise</a>()</code> <br><code><a href="layer_global_average_pooling_1d.html">layer_global_average_pooling_1d</a>()</code> <br><code><a href="layer_global_average_pooling_2d.html">layer_global_average_pooling_2d</a>()</code> <br><code><a href="layer_global_average_pooling_3d.html">layer_global_average_pooling_3d</a>()</code> <br><code><a href="layer_global_max_pooling_1d.html">layer_global_max_pooling_1d</a>()</code> <br><code><a href="layer_global_max_pooling_2d.html">layer_global_max_pooling_2d</a>()</code> <br><code><a href="layer_global_max_pooling_3d.html">layer_global_max_pooling_3d</a>()</code> <br><code><a href="layer_group_normalization.html">layer_group_normalization</a>()</code> <br><code><a href="layer_group_query_attention.html">layer_group_query_attention</a>()</code> <br><code><a href="layer_gru.html">layer_gru</a>()</code> <br><code><a href="layer_hashed_crossing.html">layer_hashed_crossing</a>()</code> <br><code><a href="layer_hashing.html">layer_hashing</a>()</code> <br><code><a href="layer_identity.html">layer_identity</a>()</code> <br><code><a href="layer_integer_lookup.html">layer_integer_lookup</a>()</code> <br><code><a href="layer_jax_model_wrapper.html">layer_jax_model_wrapper</a>()</code> <br><code><a href="layer_lambda.html">layer_lambda</a>()</code> <br><code><a href="layer_layer_normalization.html">layer_layer_normalization</a>()</code> <br><code><a href="layer_lstm.html">layer_lstm</a>()</code> <br><code><a href="layer_masking.html">layer_masking</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_max_pooling_1d.html">layer_max_pooling_1d</a>()</code> <br><code><a href="layer_max_pooling_3d.html">layer_max_pooling_3d</a>()</code> <br><code><a href="layer_maximum.html">layer_maximum</a>()</code> <br><code><a href="layer_mel_spectrogram.html">layer_mel_spectrogram</a>()</code> <br><code><a href="layer_minimum.html">layer_minimum</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_multi_head_attention.html">layer_multi_head_attention</a>()</code> <br><code><a href="layer_multiply.html">layer_multiply</a>()</code> <br><code><a href="layer_normalization.html">layer_normalization</a>()</code> <br><code><a href="layer_permute.html">layer_permute</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_brightness.html">layer_random_brightness</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_contrast.html">layer_random_contrast</a>()</code> <br><code><a href="layer_random_crop.html">layer_random_crop</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_flip.html">layer_random_flip</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_rotation.html">layer_random_rotation</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_random_translation.html">layer_random_translation</a>()</code> <br><code><a href="layer_random_zoom.html">layer_random_zoom</a>()</code> <br><code><a href="layer_repeat_vector.html">layer_repeat_vector</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_reshape.html">layer_reshape</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_rms_normalization.html">layer_rms_normalization</a>()</code> <br><code><a href="layer_rnn.html">layer_rnn</a>()</code> <br><code><a href="layer_separable_conv_1d.html">layer_separable_conv_1d</a>()</code> <br><code><a href="layer_separable_conv_2d.html">layer_separable_conv_2d</a>()</code> <br><code><a href="layer_simple_rnn.html">layer_simple_rnn</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br><code><a href="layer_spatial_dropout_1d.html">layer_spatial_dropout_1d</a>()</code> <br><code><a href="layer_spatial_dropout_2d.html">layer_spatial_dropout_2d</a>()</code> <br><code><a href="layer_spatial_dropout_3d.html">layer_spatial_dropout_3d</a>()</code> <br><code><a href="layer_spectral_normalization.html">layer_spectral_normalization</a>()</code> <br><code><a href="layer_stft_spectrogram.html">layer_stft_spectrogram</a>()</code> <br><code><a href="layer_string_lookup.html">layer_string_lookup</a>()</code> <br><code><a href="layer_subtract.html">layer_subtract</a>()</code> <br><code><a href="layer_text_vectorization.html">layer_text_vectorization</a>()</code> <br><code><a href="layer_tfsm.html">layer_tfsm</a>()</code> <br><code><a href="layer_time_distributed.html">layer_time_distributed</a>()</code> <br><code><a href="layer_torch_module_wrapper.html">layer_torch_module_wrapper</a>()</code> <br><code><a href="layer_unit_normalization.html">layer_unit_normalization</a>()</code> <br><code><a href="layer_upsampling_1d.html">layer_upsampling_1d</a>()</code> <br><code><a href="layer_upsampling_2d.html">layer_upsampling_2d</a>()</code> <br><code><a href="layer_upsampling_3d.html">layer_upsampling_3d</a>()</code> <br><code><a href="layer_zero_padding_1d.html">layer_zero_padding_1d</a>()</code> <br><code><a href="layer_zero_padding_2d.html">layer_zero_padding_2d</a>()</code> <br><code><a href="layer_zero_padding_3d.html">layer_zero_padding_3d</a>()</code> <br><code><a href="rnn_cell_gru.html">rnn_cell_gru</a>()</code> <br><code><a href="rnn_cell_lstm.html">rnn_cell_lstm</a>()</code> <br><code><a href="rnn_cell_simple.html">rnn_cell_simple</a>()</code> <br><code><a href="rnn_cells_stack.html">rnn_cells_stack</a>()</code> <br></p></div>
+<p>Other layers: <br><code><a href="Layer.html">Layer</a>()</code> <br><code><a href="layer_activation.html">layer_activation</a>()</code> <br><code><a href="layer_activation_elu.html">layer_activation_elu</a>()</code> <br><code><a href="layer_activation_leaky_relu.html">layer_activation_leaky_relu</a>()</code> <br><code><a href="layer_activation_parametric_relu.html">layer_activation_parametric_relu</a>()</code> <br><code><a href="layer_activation_relu.html">layer_activation_relu</a>()</code> <br><code><a href="layer_activation_softmax.html">layer_activation_softmax</a>()</code> <br><code><a href="layer_activity_regularization.html">layer_activity_regularization</a>()</code> <br><code><a href="layer_add.html">layer_add</a>()</code> <br><code><a href="layer_additive_attention.html">layer_additive_attention</a>()</code> <br><code><a href="layer_alpha_dropout.html">layer_alpha_dropout</a>()</code> <br><code><a href="layer_attention.html">layer_attention</a>()</code> <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_average.html">layer_average</a>()</code> <br><code><a href="layer_average_pooling_1d.html">layer_average_pooling_1d</a>()</code> <br><code><a href="layer_average_pooling_2d.html">layer_average_pooling_2d</a>()</code> <br><code><a href="layer_average_pooling_3d.html">layer_average_pooling_3d</a>()</code> <br><code><a href="layer_batch_normalization.html">layer_batch_normalization</a>()</code> <br><code><a href="layer_bidirectional.html">layer_bidirectional</a>()</code> <br><code><a href="layer_category_encoding.html">layer_category_encoding</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_concatenate.html">layer_concatenate</a>()</code> <br><code><a href="layer_conv_1d.html">layer_conv_1d</a>()</code> <br><code><a href="layer_conv_1d_transpose.html">layer_conv_1d_transpose</a>()</code> <br><code><a href="layer_conv_2d.html">layer_conv_2d</a>()</code> <br><code><a href="layer_conv_2d_transpose.html">layer_conv_2d_transpose</a>()</code> <br><code><a href="layer_conv_3d.html">layer_conv_3d</a>()</code> <br><code><a href="layer_conv_3d_transpose.html">layer_conv_3d_transpose</a>()</code> <br><code><a href="layer_conv_lstm_1d.html">layer_conv_lstm_1d</a>()</code> <br><code><a href="layer_conv_lstm_2d.html">layer_conv_lstm_2d</a>()</code> <br><code><a href="layer_conv_lstm_3d.html">layer_conv_lstm_3d</a>()</code> <br><code><a href="layer_cropping_1d.html">layer_cropping_1d</a>()</code> <br><code><a href="layer_cropping_2d.html">layer_cropping_2d</a>()</code> <br><code><a href="layer_cropping_3d.html">layer_cropping_3d</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_dense.html">layer_dense</a>()</code> <br><code><a href="layer_depthwise_conv_1d.html">layer_depthwise_conv_1d</a>()</code> <br><code><a href="layer_depthwise_conv_2d.html">layer_depthwise_conv_2d</a>()</code> <br><code><a href="layer_discretization.html">layer_discretization</a>()</code> <br><code><a href="layer_dot.html">layer_dot</a>()</code> <br><code><a href="layer_dropout.html">layer_dropout</a>()</code> <br><code><a href="layer_einsum_dense.html">layer_einsum_dense</a>()</code> <br><code><a href="layer_embedding.html">layer_embedding</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_feature_space.html">layer_feature_space</a>()</code> <br><code><a href="layer_flatten.html">layer_flatten</a>()</code> <br><code><a href="layer_flax_module_wrapper.html">layer_flax_module_wrapper</a>()</code> <br><code><a href="layer_gaussian_dropout.html">layer_gaussian_dropout</a>()</code> <br><code><a href="layer_gaussian_noise.html">layer_gaussian_noise</a>()</code> <br><code><a href="layer_global_average_pooling_1d.html">layer_global_average_pooling_1d</a>()</code> <br><code><a href="layer_global_average_pooling_2d.html">layer_global_average_pooling_2d</a>()</code> <br><code><a href="layer_global_average_pooling_3d.html">layer_global_average_pooling_3d</a>()</code> <br><code><a href="layer_global_max_pooling_1d.html">layer_global_max_pooling_1d</a>()</code> <br><code><a href="layer_global_max_pooling_2d.html">layer_global_max_pooling_2d</a>()</code> <br><code><a href="layer_global_max_pooling_3d.html">layer_global_max_pooling_3d</a>()</code> <br><code><a href="layer_group_normalization.html">layer_group_normalization</a>()</code> <br><code><a href="layer_group_query_attention.html">layer_group_query_attention</a>()</code> <br><code><a href="layer_gru.html">layer_gru</a>()</code> <br><code><a href="layer_hashed_crossing.html">layer_hashed_crossing</a>()</code> <br><code><a href="layer_hashing.html">layer_hashing</a>()</code> <br><code><a href="layer_identity.html">layer_identity</a>()</code> <br><code><a href="layer_integer_lookup.html">layer_integer_lookup</a>()</code> <br><code><a href="layer_jax_model_wrapper.html">layer_jax_model_wrapper</a>()</code> <br><code><a href="layer_lambda.html">layer_lambda</a>()</code> <br><code><a href="layer_layer_normalization.html">layer_layer_normalization</a>()</code> <br><code><a href="layer_lstm.html">layer_lstm</a>()</code> <br><code><a href="layer_masking.html">layer_masking</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_max_pooling_1d.html">layer_max_pooling_1d</a>()</code> <br><code><a href="layer_max_pooling_3d.html">layer_max_pooling_3d</a>()</code> <br><code><a href="layer_maximum.html">layer_maximum</a>()</code> <br><code><a href="layer_mel_spectrogram.html">layer_mel_spectrogram</a>()</code> <br><code><a href="layer_minimum.html">layer_minimum</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_multi_head_attention.html">layer_multi_head_attention</a>()</code> <br><code><a href="layer_multiply.html">layer_multiply</a>()</code> <br><code><a href="layer_normalization.html">layer_normalization</a>()</code> <br><code><a href="layer_permute.html">layer_permute</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_brightness.html">layer_random_brightness</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_contrast.html">layer_random_contrast</a>()</code> <br><code><a href="layer_random_crop.html">layer_random_crop</a>()</code> <br><code><a href="layer_random_elastic_transform.html">layer_random_elastic_transform</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_flip.html">layer_random_flip</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_rotation.html">layer_random_rotation</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_random_translation.html">layer_random_translation</a>()</code> <br><code><a href="layer_random_zoom.html">layer_random_zoom</a>()</code> <br><code><a href="layer_repeat_vector.html">layer_repeat_vector</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_reshape.html">layer_reshape</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_rms_normalization.html">layer_rms_normalization</a>()</code> <br><code><a href="layer_rnn.html">layer_rnn</a>()</code> <br><code><a href="layer_separable_conv_1d.html">layer_separable_conv_1d</a>()</code> <br><code><a href="layer_separable_conv_2d.html">layer_separable_conv_2d</a>()</code> <br><code><a href="layer_simple_rnn.html">layer_simple_rnn</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br><code><a href="layer_spatial_dropout_1d.html">layer_spatial_dropout_1d</a>()</code> <br><code><a href="layer_spatial_dropout_2d.html">layer_spatial_dropout_2d</a>()</code> <br><code><a href="layer_spatial_dropout_3d.html">layer_spatial_dropout_3d</a>()</code> <br><code><a href="layer_spectral_normalization.html">layer_spectral_normalization</a>()</code> <br><code><a href="layer_stft_spectrogram.html">layer_stft_spectrogram</a>()</code> <br><code><a href="layer_string_lookup.html">layer_string_lookup</a>()</code> <br><code><a href="layer_subtract.html">layer_subtract</a>()</code> <br><code><a href="layer_text_vectorization.html">layer_text_vectorization</a>()</code> <br><code><a href="layer_tfsm.html">layer_tfsm</a>()</code> <br><code><a href="layer_time_distributed.html">layer_time_distributed</a>()</code> <br><code><a href="layer_torch_module_wrapper.html">layer_torch_module_wrapper</a>()</code> <br><code><a href="layer_unit_normalization.html">layer_unit_normalization</a>()</code> <br><code><a href="layer_upsampling_1d.html">layer_upsampling_1d</a>()</code> <br><code><a href="layer_upsampling_2d.html">layer_upsampling_2d</a>()</code> <br><code><a href="layer_upsampling_3d.html">layer_upsampling_3d</a>()</code> <br><code><a href="layer_zero_padding_1d.html">layer_zero_padding_1d</a>()</code> <br><code><a href="layer_zero_padding_2d.html">layer_zero_padding_2d</a>()</code> <br><code><a href="layer_zero_padding_3d.html">layer_zero_padding_3d</a>()</code> <br><code><a href="rnn_cell_gru.html">rnn_cell_gru</a>()</code> <br><code><a href="rnn_cell_lstm.html">rnn_cell_lstm</a>()</code> <br><code><a href="rnn_cell_simple.html">rnn_cell_simple</a>()</code> <br><code><a href="rnn_cells_stack.html">rnn_cells_stack</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -238,7 +238,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/layer_max_pooling_3d.html b/docs/dev/reference/layer_max_pooling_3d.html
index b4a9a76ab..5a15f2df5 100644
--- a/docs/dev/reference/layer_max_pooling_3d.html
+++ b/docs/dev/reference/layer_max_pooling_3d.html
@@ -14,7 +14,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -174,7 +174,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
     <div class="dont-index">
 <ul><li><p><a href="https://keras.io/api/layers/pooling_layers/max_pooling3d#maxpooling3d-class" class="external-link">https://keras.io/api/layers/pooling_layers/max_pooling3d#maxpooling3d-class</a></p></li>
 </ul><p>Other pooling layers: <br><code><a href="layer_average_pooling_1d.html">layer_average_pooling_1d</a>()</code> <br><code><a href="layer_average_pooling_2d.html">layer_average_pooling_2d</a>()</code> <br><code><a href="layer_average_pooling_3d.html">layer_average_pooling_3d</a>()</code> <br><code><a href="layer_global_average_pooling_1d.html">layer_global_average_pooling_1d</a>()</code> <br><code><a href="layer_global_average_pooling_2d.html">layer_global_average_pooling_2d</a>()</code> <br><code><a href="layer_global_average_pooling_3d.html">layer_global_average_pooling_3d</a>()</code> <br><code><a href="layer_global_max_pooling_1d.html">layer_global_max_pooling_1d</a>()</code> <br><code><a href="layer_global_max_pooling_2d.html">layer_global_max_pooling_2d</a>()</code> <br><code><a href="layer_global_max_pooling_3d.html">layer_global_max_pooling_3d</a>()</code> <br><code><a href="layer_max_pooling_1d.html">layer_max_pooling_1d</a>()</code> <br><code><a href="layer_max_pooling_2d.html">layer_max_pooling_2d</a>()</code> <br></p>
-<p>Other layers: <br><code><a href="Layer.html">Layer</a>()</code> <br><code><a href="layer_activation.html">layer_activation</a>()</code> <br><code><a href="layer_activation_elu.html">layer_activation_elu</a>()</code> <br><code><a href="layer_activation_leaky_relu.html">layer_activation_leaky_relu</a>()</code> <br><code><a href="layer_activation_parametric_relu.html">layer_activation_parametric_relu</a>()</code> <br><code><a href="layer_activation_relu.html">layer_activation_relu</a>()</code> <br><code><a href="layer_activation_softmax.html">layer_activation_softmax</a>()</code> <br><code><a href="layer_activity_regularization.html">layer_activity_regularization</a>()</code> <br><code><a href="layer_add.html">layer_add</a>()</code> <br><code><a href="layer_additive_attention.html">layer_additive_attention</a>()</code> <br><code><a href="layer_alpha_dropout.html">layer_alpha_dropout</a>()</code> <br><code><a href="layer_attention.html">layer_attention</a>()</code> <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_average.html">layer_average</a>()</code> <br><code><a href="layer_average_pooling_1d.html">layer_average_pooling_1d</a>()</code> <br><code><a href="layer_average_pooling_2d.html">layer_average_pooling_2d</a>()</code> <br><code><a href="layer_average_pooling_3d.html">layer_average_pooling_3d</a>()</code> <br><code><a href="layer_batch_normalization.html">layer_batch_normalization</a>()</code> <br><code><a href="layer_bidirectional.html">layer_bidirectional</a>()</code> <br><code><a href="layer_category_encoding.html">layer_category_encoding</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_concatenate.html">layer_concatenate</a>()</code> <br><code><a href="layer_conv_1d.html">layer_conv_1d</a>()</code> <br><code><a href="layer_conv_1d_transpose.html">layer_conv_1d_transpose</a>()</code> <br><code><a href="layer_conv_2d.html">layer_conv_2d</a>()</code> <br><code><a href="layer_conv_2d_transpose.html">layer_conv_2d_transpose</a>()</code> <br><code><a href="layer_conv_3d.html">layer_conv_3d</a>()</code> <br><code><a href="layer_conv_3d_transpose.html">layer_conv_3d_transpose</a>()</code> <br><code><a href="layer_conv_lstm_1d.html">layer_conv_lstm_1d</a>()</code> <br><code><a href="layer_conv_lstm_2d.html">layer_conv_lstm_2d</a>()</code> <br><code><a href="layer_conv_lstm_3d.html">layer_conv_lstm_3d</a>()</code> <br><code><a href="layer_cropping_1d.html">layer_cropping_1d</a>()</code> <br><code><a href="layer_cropping_2d.html">layer_cropping_2d</a>()</code> <br><code><a href="layer_cropping_3d.html">layer_cropping_3d</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_dense.html">layer_dense</a>()</code> <br><code><a href="layer_depthwise_conv_1d.html">layer_depthwise_conv_1d</a>()</code> <br><code><a href="layer_depthwise_conv_2d.html">layer_depthwise_conv_2d</a>()</code> <br><code><a href="layer_discretization.html">layer_discretization</a>()</code> <br><code><a href="layer_dot.html">layer_dot</a>()</code> <br><code><a href="layer_dropout.html">layer_dropout</a>()</code> <br><code><a href="layer_einsum_dense.html">layer_einsum_dense</a>()</code> <br><code><a href="layer_embedding.html">layer_embedding</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_feature_space.html">layer_feature_space</a>()</code> <br><code><a href="layer_flatten.html">layer_flatten</a>()</code> <br><code><a href="layer_flax_module_wrapper.html">layer_flax_module_wrapper</a>()</code> <br><code><a href="layer_gaussian_dropout.html">layer_gaussian_dropout</a>()</code> <br><code><a href="layer_gaussian_noise.html">layer_gaussian_noise</a>()</code> <br><code><a href="layer_global_average_pooling_1d.html">layer_global_average_pooling_1d</a>()</code> <br><code><a href="layer_global_average_pooling_2d.html">layer_global_average_pooling_2d</a>()</code> <br><code><a href="layer_global_average_pooling_3d.html">layer_global_average_pooling_3d</a>()</code> <br><code><a href="layer_global_max_pooling_1d.html">layer_global_max_pooling_1d</a>()</code> <br><code><a href="layer_global_max_pooling_2d.html">layer_global_max_pooling_2d</a>()</code> <br><code><a href="layer_global_max_pooling_3d.html">layer_global_max_pooling_3d</a>()</code> <br><code><a href="layer_group_normalization.html">layer_group_normalization</a>()</code> <br><code><a href="layer_group_query_attention.html">layer_group_query_attention</a>()</code> <br><code><a href="layer_gru.html">layer_gru</a>()</code> <br><code><a href="layer_hashed_crossing.html">layer_hashed_crossing</a>()</code> <br><code><a href="layer_hashing.html">layer_hashing</a>()</code> <br><code><a href="layer_identity.html">layer_identity</a>()</code> <br><code><a href="layer_integer_lookup.html">layer_integer_lookup</a>()</code> <br><code><a href="layer_jax_model_wrapper.html">layer_jax_model_wrapper</a>()</code> <br><code><a href="layer_lambda.html">layer_lambda</a>()</code> <br><code><a href="layer_layer_normalization.html">layer_layer_normalization</a>()</code> <br><code><a href="layer_lstm.html">layer_lstm</a>()</code> <br><code><a href="layer_masking.html">layer_masking</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_max_pooling_1d.html">layer_max_pooling_1d</a>()</code> <br><code><a href="layer_max_pooling_2d.html">layer_max_pooling_2d</a>()</code> <br><code><a href="layer_maximum.html">layer_maximum</a>()</code> <br><code><a href="layer_mel_spectrogram.html">layer_mel_spectrogram</a>()</code> <br><code><a href="layer_minimum.html">layer_minimum</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_multi_head_attention.html">layer_multi_head_attention</a>()</code> <br><code><a href="layer_multiply.html">layer_multiply</a>()</code> <br><code><a href="layer_normalization.html">layer_normalization</a>()</code> <br><code><a href="layer_permute.html">layer_permute</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_brightness.html">layer_random_brightness</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_contrast.html">layer_random_contrast</a>()</code> <br><code><a href="layer_random_crop.html">layer_random_crop</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_flip.html">layer_random_flip</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_rotation.html">layer_random_rotation</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_random_translation.html">layer_random_translation</a>()</code> <br><code><a href="layer_random_zoom.html">layer_random_zoom</a>()</code> <br><code><a href="layer_repeat_vector.html">layer_repeat_vector</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_reshape.html">layer_reshape</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_rms_normalization.html">layer_rms_normalization</a>()</code> <br><code><a href="layer_rnn.html">layer_rnn</a>()</code> <br><code><a href="layer_separable_conv_1d.html">layer_separable_conv_1d</a>()</code> <br><code><a href="layer_separable_conv_2d.html">layer_separable_conv_2d</a>()</code> <br><code><a href="layer_simple_rnn.html">layer_simple_rnn</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br><code><a href="layer_spatial_dropout_1d.html">layer_spatial_dropout_1d</a>()</code> <br><code><a href="layer_spatial_dropout_2d.html">layer_spatial_dropout_2d</a>()</code> <br><code><a href="layer_spatial_dropout_3d.html">layer_spatial_dropout_3d</a>()</code> <br><code><a href="layer_spectral_normalization.html">layer_spectral_normalization</a>()</code> <br><code><a href="layer_stft_spectrogram.html">layer_stft_spectrogram</a>()</code> <br><code><a href="layer_string_lookup.html">layer_string_lookup</a>()</code> <br><code><a href="layer_subtract.html">layer_subtract</a>()</code> <br><code><a href="layer_text_vectorization.html">layer_text_vectorization</a>()</code> <br><code><a href="layer_tfsm.html">layer_tfsm</a>()</code> <br><code><a href="layer_time_distributed.html">layer_time_distributed</a>()</code> <br><code><a href="layer_torch_module_wrapper.html">layer_torch_module_wrapper</a>()</code> <br><code><a href="layer_unit_normalization.html">layer_unit_normalization</a>()</code> <br><code><a href="layer_upsampling_1d.html">layer_upsampling_1d</a>()</code> <br><code><a href="layer_upsampling_2d.html">layer_upsampling_2d</a>()</code> <br><code><a href="layer_upsampling_3d.html">layer_upsampling_3d</a>()</code> <br><code><a href="layer_zero_padding_1d.html">layer_zero_padding_1d</a>()</code> <br><code><a href="layer_zero_padding_2d.html">layer_zero_padding_2d</a>()</code> <br><code><a href="layer_zero_padding_3d.html">layer_zero_padding_3d</a>()</code> <br><code><a href="rnn_cell_gru.html">rnn_cell_gru</a>()</code> <br><code><a href="rnn_cell_lstm.html">rnn_cell_lstm</a>()</code> <br><code><a href="rnn_cell_simple.html">rnn_cell_simple</a>()</code> <br><code><a href="rnn_cells_stack.html">rnn_cells_stack</a>()</code> <br></p></div>
+<p>Other layers: <br><code><a href="Layer.html">Layer</a>()</code> <br><code><a href="layer_activation.html">layer_activation</a>()</code> <br><code><a href="layer_activation_elu.html">layer_activation_elu</a>()</code> <br><code><a href="layer_activation_leaky_relu.html">layer_activation_leaky_relu</a>()</code> <br><code><a href="layer_activation_parametric_relu.html">layer_activation_parametric_relu</a>()</code> <br><code><a href="layer_activation_relu.html">layer_activation_relu</a>()</code> <br><code><a href="layer_activation_softmax.html">layer_activation_softmax</a>()</code> <br><code><a href="layer_activity_regularization.html">layer_activity_regularization</a>()</code> <br><code><a href="layer_add.html">layer_add</a>()</code> <br><code><a href="layer_additive_attention.html">layer_additive_attention</a>()</code> <br><code><a href="layer_alpha_dropout.html">layer_alpha_dropout</a>()</code> <br><code><a href="layer_attention.html">layer_attention</a>()</code> <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_average.html">layer_average</a>()</code> <br><code><a href="layer_average_pooling_1d.html">layer_average_pooling_1d</a>()</code> <br><code><a href="layer_average_pooling_2d.html">layer_average_pooling_2d</a>()</code> <br><code><a href="layer_average_pooling_3d.html">layer_average_pooling_3d</a>()</code> <br><code><a href="layer_batch_normalization.html">layer_batch_normalization</a>()</code> <br><code><a href="layer_bidirectional.html">layer_bidirectional</a>()</code> <br><code><a href="layer_category_encoding.html">layer_category_encoding</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_concatenate.html">layer_concatenate</a>()</code> <br><code><a href="layer_conv_1d.html">layer_conv_1d</a>()</code> <br><code><a href="layer_conv_1d_transpose.html">layer_conv_1d_transpose</a>()</code> <br><code><a href="layer_conv_2d.html">layer_conv_2d</a>()</code> <br><code><a href="layer_conv_2d_transpose.html">layer_conv_2d_transpose</a>()</code> <br><code><a href="layer_conv_3d.html">layer_conv_3d</a>()</code> <br><code><a href="layer_conv_3d_transpose.html">layer_conv_3d_transpose</a>()</code> <br><code><a href="layer_conv_lstm_1d.html">layer_conv_lstm_1d</a>()</code> <br><code><a href="layer_conv_lstm_2d.html">layer_conv_lstm_2d</a>()</code> <br><code><a href="layer_conv_lstm_3d.html">layer_conv_lstm_3d</a>()</code> <br><code><a href="layer_cropping_1d.html">layer_cropping_1d</a>()</code> <br><code><a href="layer_cropping_2d.html">layer_cropping_2d</a>()</code> <br><code><a href="layer_cropping_3d.html">layer_cropping_3d</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_dense.html">layer_dense</a>()</code> <br><code><a href="layer_depthwise_conv_1d.html">layer_depthwise_conv_1d</a>()</code> <br><code><a href="layer_depthwise_conv_2d.html">layer_depthwise_conv_2d</a>()</code> <br><code><a href="layer_discretization.html">layer_discretization</a>()</code> <br><code><a href="layer_dot.html">layer_dot</a>()</code> <br><code><a href="layer_dropout.html">layer_dropout</a>()</code> <br><code><a href="layer_einsum_dense.html">layer_einsum_dense</a>()</code> <br><code><a href="layer_embedding.html">layer_embedding</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_feature_space.html">layer_feature_space</a>()</code> <br><code><a href="layer_flatten.html">layer_flatten</a>()</code> <br><code><a href="layer_flax_module_wrapper.html">layer_flax_module_wrapper</a>()</code> <br><code><a href="layer_gaussian_dropout.html">layer_gaussian_dropout</a>()</code> <br><code><a href="layer_gaussian_noise.html">layer_gaussian_noise</a>()</code> <br><code><a href="layer_global_average_pooling_1d.html">layer_global_average_pooling_1d</a>()</code> <br><code><a href="layer_global_average_pooling_2d.html">layer_global_average_pooling_2d</a>()</code> <br><code><a href="layer_global_average_pooling_3d.html">layer_global_average_pooling_3d</a>()</code> <br><code><a href="layer_global_max_pooling_1d.html">layer_global_max_pooling_1d</a>()</code> <br><code><a href="layer_global_max_pooling_2d.html">layer_global_max_pooling_2d</a>()</code> <br><code><a href="layer_global_max_pooling_3d.html">layer_global_max_pooling_3d</a>()</code> <br><code><a href="layer_group_normalization.html">layer_group_normalization</a>()</code> <br><code><a href="layer_group_query_attention.html">layer_group_query_attention</a>()</code> <br><code><a href="layer_gru.html">layer_gru</a>()</code> <br><code><a href="layer_hashed_crossing.html">layer_hashed_crossing</a>()</code> <br><code><a href="layer_hashing.html">layer_hashing</a>()</code> <br><code><a href="layer_identity.html">layer_identity</a>()</code> <br><code><a href="layer_integer_lookup.html">layer_integer_lookup</a>()</code> <br><code><a href="layer_jax_model_wrapper.html">layer_jax_model_wrapper</a>()</code> <br><code><a href="layer_lambda.html">layer_lambda</a>()</code> <br><code><a href="layer_layer_normalization.html">layer_layer_normalization</a>()</code> <br><code><a href="layer_lstm.html">layer_lstm</a>()</code> <br><code><a href="layer_masking.html">layer_masking</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_max_pooling_1d.html">layer_max_pooling_1d</a>()</code> <br><code><a href="layer_max_pooling_2d.html">layer_max_pooling_2d</a>()</code> <br><code><a href="layer_maximum.html">layer_maximum</a>()</code> <br><code><a href="layer_mel_spectrogram.html">layer_mel_spectrogram</a>()</code> <br><code><a href="layer_minimum.html">layer_minimum</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_multi_head_attention.html">layer_multi_head_attention</a>()</code> <br><code><a href="layer_multiply.html">layer_multiply</a>()</code> <br><code><a href="layer_normalization.html">layer_normalization</a>()</code> <br><code><a href="layer_permute.html">layer_permute</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_brightness.html">layer_random_brightness</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_contrast.html">layer_random_contrast</a>()</code> <br><code><a href="layer_random_crop.html">layer_random_crop</a>()</code> <br><code><a href="layer_random_elastic_transform.html">layer_random_elastic_transform</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_flip.html">layer_random_flip</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_rotation.html">layer_random_rotation</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_random_translation.html">layer_random_translation</a>()</code> <br><code><a href="layer_random_zoom.html">layer_random_zoom</a>()</code> <br><code><a href="layer_repeat_vector.html">layer_repeat_vector</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_reshape.html">layer_reshape</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_rms_normalization.html">layer_rms_normalization</a>()</code> <br><code><a href="layer_rnn.html">layer_rnn</a>()</code> <br><code><a href="layer_separable_conv_1d.html">layer_separable_conv_1d</a>()</code> <br><code><a href="layer_separable_conv_2d.html">layer_separable_conv_2d</a>()</code> <br><code><a href="layer_simple_rnn.html">layer_simple_rnn</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br><code><a href="layer_spatial_dropout_1d.html">layer_spatial_dropout_1d</a>()</code> <br><code><a href="layer_spatial_dropout_2d.html">layer_spatial_dropout_2d</a>()</code> <br><code><a href="layer_spatial_dropout_3d.html">layer_spatial_dropout_3d</a>()</code> <br><code><a href="layer_spectral_normalization.html">layer_spectral_normalization</a>()</code> <br><code><a href="layer_stft_spectrogram.html">layer_stft_spectrogram</a>()</code> <br><code><a href="layer_string_lookup.html">layer_string_lookup</a>()</code> <br><code><a href="layer_subtract.html">layer_subtract</a>()</code> <br><code><a href="layer_text_vectorization.html">layer_text_vectorization</a>()</code> <br><code><a href="layer_tfsm.html">layer_tfsm</a>()</code> <br><code><a href="layer_time_distributed.html">layer_time_distributed</a>()</code> <br><code><a href="layer_torch_module_wrapper.html">layer_torch_module_wrapper</a>()</code> <br><code><a href="layer_unit_normalization.html">layer_unit_normalization</a>()</code> <br><code><a href="layer_upsampling_1d.html">layer_upsampling_1d</a>()</code> <br><code><a href="layer_upsampling_2d.html">layer_upsampling_2d</a>()</code> <br><code><a href="layer_upsampling_3d.html">layer_upsampling_3d</a>()</code> <br><code><a href="layer_zero_padding_1d.html">layer_zero_padding_1d</a>()</code> <br><code><a href="layer_zero_padding_2d.html">layer_zero_padding_2d</a>()</code> <br><code><a href="layer_zero_padding_3d.html">layer_zero_padding_3d</a>()</code> <br><code><a href="rnn_cell_gru.html">rnn_cell_gru</a>()</code> <br><code><a href="rnn_cell_lstm.html">rnn_cell_lstm</a>()</code> <br><code><a href="rnn_cell_simple.html">rnn_cell_simple</a>()</code> <br><code><a href="rnn_cells_stack.html">rnn_cells_stack</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -186,7 +186,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/layer_maximum.html b/docs/dev/reference/layer_maximum.html
index 2877d747a..035a24911 100644
--- a/docs/dev/reference/layer_maximum.html
+++ b/docs/dev/reference/layer_maximum.html
@@ -10,7 +10,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -107,7 +107,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
     <div class="dont-index">
 <ul><li><p><a href="https://keras.io/api/layers/merging_layers/maximum#maximum-class" class="external-link">https://keras.io/api/layers/merging_layers/maximum#maximum-class</a></p></li>
 </ul><p>Other merging layers: <br><code><a href="layer_add.html">layer_add</a>()</code> <br><code><a href="layer_average.html">layer_average</a>()</code> <br><code><a href="layer_concatenate.html">layer_concatenate</a>()</code> <br><code><a href="layer_dot.html">layer_dot</a>()</code> <br><code><a href="layer_minimum.html">layer_minimum</a>()</code> <br><code><a href="layer_multiply.html">layer_multiply</a>()</code> <br><code><a href="layer_subtract.html">layer_subtract</a>()</code> <br></p>
-<p>Other layers: <br><code><a href="Layer.html">Layer</a>()</code> <br><code><a href="layer_activation.html">layer_activation</a>()</code> <br><code><a href="layer_activation_elu.html">layer_activation_elu</a>()</code> <br><code><a href="layer_activation_leaky_relu.html">layer_activation_leaky_relu</a>()</code> <br><code><a href="layer_activation_parametric_relu.html">layer_activation_parametric_relu</a>()</code> <br><code><a href="layer_activation_relu.html">layer_activation_relu</a>()</code> <br><code><a href="layer_activation_softmax.html">layer_activation_softmax</a>()</code> <br><code><a href="layer_activity_regularization.html">layer_activity_regularization</a>()</code> <br><code><a href="layer_add.html">layer_add</a>()</code> <br><code><a href="layer_additive_attention.html">layer_additive_attention</a>()</code> <br><code><a href="layer_alpha_dropout.html">layer_alpha_dropout</a>()</code> <br><code><a href="layer_attention.html">layer_attention</a>()</code> <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_average.html">layer_average</a>()</code> <br><code><a href="layer_average_pooling_1d.html">layer_average_pooling_1d</a>()</code> <br><code><a href="layer_average_pooling_2d.html">layer_average_pooling_2d</a>()</code> <br><code><a href="layer_average_pooling_3d.html">layer_average_pooling_3d</a>()</code> <br><code><a href="layer_batch_normalization.html">layer_batch_normalization</a>()</code> <br><code><a href="layer_bidirectional.html">layer_bidirectional</a>()</code> <br><code><a href="layer_category_encoding.html">layer_category_encoding</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_concatenate.html">layer_concatenate</a>()</code> <br><code><a href="layer_conv_1d.html">layer_conv_1d</a>()</code> <br><code><a href="layer_conv_1d_transpose.html">layer_conv_1d_transpose</a>()</code> <br><code><a href="layer_conv_2d.html">layer_conv_2d</a>()</code> <br><code><a href="layer_conv_2d_transpose.html">layer_conv_2d_transpose</a>()</code> <br><code><a href="layer_conv_3d.html">layer_conv_3d</a>()</code> <br><code><a href="layer_conv_3d_transpose.html">layer_conv_3d_transpose</a>()</code> <br><code><a href="layer_conv_lstm_1d.html">layer_conv_lstm_1d</a>()</code> <br><code><a href="layer_conv_lstm_2d.html">layer_conv_lstm_2d</a>()</code> <br><code><a href="layer_conv_lstm_3d.html">layer_conv_lstm_3d</a>()</code> <br><code><a href="layer_cropping_1d.html">layer_cropping_1d</a>()</code> <br><code><a href="layer_cropping_2d.html">layer_cropping_2d</a>()</code> <br><code><a href="layer_cropping_3d.html">layer_cropping_3d</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_dense.html">layer_dense</a>()</code> <br><code><a href="layer_depthwise_conv_1d.html">layer_depthwise_conv_1d</a>()</code> <br><code><a href="layer_depthwise_conv_2d.html">layer_depthwise_conv_2d</a>()</code> <br><code><a href="layer_discretization.html">layer_discretization</a>()</code> <br><code><a href="layer_dot.html">layer_dot</a>()</code> <br><code><a href="layer_dropout.html">layer_dropout</a>()</code> <br><code><a href="layer_einsum_dense.html">layer_einsum_dense</a>()</code> <br><code><a href="layer_embedding.html">layer_embedding</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_feature_space.html">layer_feature_space</a>()</code> <br><code><a href="layer_flatten.html">layer_flatten</a>()</code> <br><code><a href="layer_flax_module_wrapper.html">layer_flax_module_wrapper</a>()</code> <br><code><a href="layer_gaussian_dropout.html">layer_gaussian_dropout</a>()</code> <br><code><a href="layer_gaussian_noise.html">layer_gaussian_noise</a>()</code> <br><code><a href="layer_global_average_pooling_1d.html">layer_global_average_pooling_1d</a>()</code> <br><code><a href="layer_global_average_pooling_2d.html">layer_global_average_pooling_2d</a>()</code> <br><code><a href="layer_global_average_pooling_3d.html">layer_global_average_pooling_3d</a>()</code> <br><code><a href="layer_global_max_pooling_1d.html">layer_global_max_pooling_1d</a>()</code> <br><code><a href="layer_global_max_pooling_2d.html">layer_global_max_pooling_2d</a>()</code> <br><code><a href="layer_global_max_pooling_3d.html">layer_global_max_pooling_3d</a>()</code> <br><code><a href="layer_group_normalization.html">layer_group_normalization</a>()</code> <br><code><a href="layer_group_query_attention.html">layer_group_query_attention</a>()</code> <br><code><a href="layer_gru.html">layer_gru</a>()</code> <br><code><a href="layer_hashed_crossing.html">layer_hashed_crossing</a>()</code> <br><code><a href="layer_hashing.html">layer_hashing</a>()</code> <br><code><a href="layer_identity.html">layer_identity</a>()</code> <br><code><a href="layer_integer_lookup.html">layer_integer_lookup</a>()</code> <br><code><a href="layer_jax_model_wrapper.html">layer_jax_model_wrapper</a>()</code> <br><code><a href="layer_lambda.html">layer_lambda</a>()</code> <br><code><a href="layer_layer_normalization.html">layer_layer_normalization</a>()</code> <br><code><a href="layer_lstm.html">layer_lstm</a>()</code> <br><code><a href="layer_masking.html">layer_masking</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_max_pooling_1d.html">layer_max_pooling_1d</a>()</code> <br><code><a href="layer_max_pooling_2d.html">layer_max_pooling_2d</a>()</code> <br><code><a href="layer_max_pooling_3d.html">layer_max_pooling_3d</a>()</code> <br><code><a href="layer_mel_spectrogram.html">layer_mel_spectrogram</a>()</code> <br><code><a href="layer_minimum.html">layer_minimum</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_multi_head_attention.html">layer_multi_head_attention</a>()</code> <br><code><a href="layer_multiply.html">layer_multiply</a>()</code> <br><code><a href="layer_normalization.html">layer_normalization</a>()</code> <br><code><a href="layer_permute.html">layer_permute</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_brightness.html">layer_random_brightness</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_contrast.html">layer_random_contrast</a>()</code> <br><code><a href="layer_random_crop.html">layer_random_crop</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_flip.html">layer_random_flip</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_rotation.html">layer_random_rotation</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_random_translation.html">layer_random_translation</a>()</code> <br><code><a href="layer_random_zoom.html">layer_random_zoom</a>()</code> <br><code><a href="layer_repeat_vector.html">layer_repeat_vector</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_reshape.html">layer_reshape</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_rms_normalization.html">layer_rms_normalization</a>()</code> <br><code><a href="layer_rnn.html">layer_rnn</a>()</code> <br><code><a href="layer_separable_conv_1d.html">layer_separable_conv_1d</a>()</code> <br><code><a href="layer_separable_conv_2d.html">layer_separable_conv_2d</a>()</code> <br><code><a href="layer_simple_rnn.html">layer_simple_rnn</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br><code><a href="layer_spatial_dropout_1d.html">layer_spatial_dropout_1d</a>()</code> <br><code><a href="layer_spatial_dropout_2d.html">layer_spatial_dropout_2d</a>()</code> <br><code><a href="layer_spatial_dropout_3d.html">layer_spatial_dropout_3d</a>()</code> <br><code><a href="layer_spectral_normalization.html">layer_spectral_normalization</a>()</code> <br><code><a href="layer_stft_spectrogram.html">layer_stft_spectrogram</a>()</code> <br><code><a href="layer_string_lookup.html">layer_string_lookup</a>()</code> <br><code><a href="layer_subtract.html">layer_subtract</a>()</code> <br><code><a href="layer_text_vectorization.html">layer_text_vectorization</a>()</code> <br><code><a href="layer_tfsm.html">layer_tfsm</a>()</code> <br><code><a href="layer_time_distributed.html">layer_time_distributed</a>()</code> <br><code><a href="layer_torch_module_wrapper.html">layer_torch_module_wrapper</a>()</code> <br><code><a href="layer_unit_normalization.html">layer_unit_normalization</a>()</code> <br><code><a href="layer_upsampling_1d.html">layer_upsampling_1d</a>()</code> <br><code><a href="layer_upsampling_2d.html">layer_upsampling_2d</a>()</code> <br><code><a href="layer_upsampling_3d.html">layer_upsampling_3d</a>()</code> <br><code><a href="layer_zero_padding_1d.html">layer_zero_padding_1d</a>()</code> <br><code><a href="layer_zero_padding_2d.html">layer_zero_padding_2d</a>()</code> <br><code><a href="layer_zero_padding_3d.html">layer_zero_padding_3d</a>()</code> <br><code><a href="rnn_cell_gru.html">rnn_cell_gru</a>()</code> <br><code><a href="rnn_cell_lstm.html">rnn_cell_lstm</a>()</code> <br><code><a href="rnn_cell_simple.html">rnn_cell_simple</a>()</code> <br><code><a href="rnn_cells_stack.html">rnn_cells_stack</a>()</code> <br></p></div>
+<p>Other layers: <br><code><a href="Layer.html">Layer</a>()</code> <br><code><a href="layer_activation.html">layer_activation</a>()</code> <br><code><a href="layer_activation_elu.html">layer_activation_elu</a>()</code> <br><code><a href="layer_activation_leaky_relu.html">layer_activation_leaky_relu</a>()</code> <br><code><a href="layer_activation_parametric_relu.html">layer_activation_parametric_relu</a>()</code> <br><code><a href="layer_activation_relu.html">layer_activation_relu</a>()</code> <br><code><a href="layer_activation_softmax.html">layer_activation_softmax</a>()</code> <br><code><a href="layer_activity_regularization.html">layer_activity_regularization</a>()</code> <br><code><a href="layer_add.html">layer_add</a>()</code> <br><code><a href="layer_additive_attention.html">layer_additive_attention</a>()</code> <br><code><a href="layer_alpha_dropout.html">layer_alpha_dropout</a>()</code> <br><code><a href="layer_attention.html">layer_attention</a>()</code> <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_average.html">layer_average</a>()</code> <br><code><a href="layer_average_pooling_1d.html">layer_average_pooling_1d</a>()</code> <br><code><a href="layer_average_pooling_2d.html">layer_average_pooling_2d</a>()</code> <br><code><a href="layer_average_pooling_3d.html">layer_average_pooling_3d</a>()</code> <br><code><a href="layer_batch_normalization.html">layer_batch_normalization</a>()</code> <br><code><a href="layer_bidirectional.html">layer_bidirectional</a>()</code> <br><code><a href="layer_category_encoding.html">layer_category_encoding</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_concatenate.html">layer_concatenate</a>()</code> <br><code><a href="layer_conv_1d.html">layer_conv_1d</a>()</code> <br><code><a href="layer_conv_1d_transpose.html">layer_conv_1d_transpose</a>()</code> <br><code><a href="layer_conv_2d.html">layer_conv_2d</a>()</code> <br><code><a href="layer_conv_2d_transpose.html">layer_conv_2d_transpose</a>()</code> <br><code><a href="layer_conv_3d.html">layer_conv_3d</a>()</code> <br><code><a href="layer_conv_3d_transpose.html">layer_conv_3d_transpose</a>()</code> <br><code><a href="layer_conv_lstm_1d.html">layer_conv_lstm_1d</a>()</code> <br><code><a href="layer_conv_lstm_2d.html">layer_conv_lstm_2d</a>()</code> <br><code><a href="layer_conv_lstm_3d.html">layer_conv_lstm_3d</a>()</code> <br><code><a href="layer_cropping_1d.html">layer_cropping_1d</a>()</code> <br><code><a href="layer_cropping_2d.html">layer_cropping_2d</a>()</code> <br><code><a href="layer_cropping_3d.html">layer_cropping_3d</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_dense.html">layer_dense</a>()</code> <br><code><a href="layer_depthwise_conv_1d.html">layer_depthwise_conv_1d</a>()</code> <br><code><a href="layer_depthwise_conv_2d.html">layer_depthwise_conv_2d</a>()</code> <br><code><a href="layer_discretization.html">layer_discretization</a>()</code> <br><code><a href="layer_dot.html">layer_dot</a>()</code> <br><code><a href="layer_dropout.html">layer_dropout</a>()</code> <br><code><a href="layer_einsum_dense.html">layer_einsum_dense</a>()</code> <br><code><a href="layer_embedding.html">layer_embedding</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_feature_space.html">layer_feature_space</a>()</code> <br><code><a href="layer_flatten.html">layer_flatten</a>()</code> <br><code><a href="layer_flax_module_wrapper.html">layer_flax_module_wrapper</a>()</code> <br><code><a href="layer_gaussian_dropout.html">layer_gaussian_dropout</a>()</code> <br><code><a href="layer_gaussian_noise.html">layer_gaussian_noise</a>()</code> <br><code><a href="layer_global_average_pooling_1d.html">layer_global_average_pooling_1d</a>()</code> <br><code><a href="layer_global_average_pooling_2d.html">layer_global_average_pooling_2d</a>()</code> <br><code><a href="layer_global_average_pooling_3d.html">layer_global_average_pooling_3d</a>()</code> <br><code><a href="layer_global_max_pooling_1d.html">layer_global_max_pooling_1d</a>()</code> <br><code><a href="layer_global_max_pooling_2d.html">layer_global_max_pooling_2d</a>()</code> <br><code><a href="layer_global_max_pooling_3d.html">layer_global_max_pooling_3d</a>()</code> <br><code><a href="layer_group_normalization.html">layer_group_normalization</a>()</code> <br><code><a href="layer_group_query_attention.html">layer_group_query_attention</a>()</code> <br><code><a href="layer_gru.html">layer_gru</a>()</code> <br><code><a href="layer_hashed_crossing.html">layer_hashed_crossing</a>()</code> <br><code><a href="layer_hashing.html">layer_hashing</a>()</code> <br><code><a href="layer_identity.html">layer_identity</a>()</code> <br><code><a href="layer_integer_lookup.html">layer_integer_lookup</a>()</code> <br><code><a href="layer_jax_model_wrapper.html">layer_jax_model_wrapper</a>()</code> <br><code><a href="layer_lambda.html">layer_lambda</a>()</code> <br><code><a href="layer_layer_normalization.html">layer_layer_normalization</a>()</code> <br><code><a href="layer_lstm.html">layer_lstm</a>()</code> <br><code><a href="layer_masking.html">layer_masking</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_max_pooling_1d.html">layer_max_pooling_1d</a>()</code> <br><code><a href="layer_max_pooling_2d.html">layer_max_pooling_2d</a>()</code> <br><code><a href="layer_max_pooling_3d.html">layer_max_pooling_3d</a>()</code> <br><code><a href="layer_mel_spectrogram.html">layer_mel_spectrogram</a>()</code> <br><code><a href="layer_minimum.html">layer_minimum</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_multi_head_attention.html">layer_multi_head_attention</a>()</code> <br><code><a href="layer_multiply.html">layer_multiply</a>()</code> <br><code><a href="layer_normalization.html">layer_normalization</a>()</code> <br><code><a href="layer_permute.html">layer_permute</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_brightness.html">layer_random_brightness</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_contrast.html">layer_random_contrast</a>()</code> <br><code><a href="layer_random_crop.html">layer_random_crop</a>()</code> <br><code><a href="layer_random_elastic_transform.html">layer_random_elastic_transform</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_flip.html">layer_random_flip</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_rotation.html">layer_random_rotation</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_random_translation.html">layer_random_translation</a>()</code> <br><code><a href="layer_random_zoom.html">layer_random_zoom</a>()</code> <br><code><a href="layer_repeat_vector.html">layer_repeat_vector</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_reshape.html">layer_reshape</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_rms_normalization.html">layer_rms_normalization</a>()</code> <br><code><a href="layer_rnn.html">layer_rnn</a>()</code> <br><code><a href="layer_separable_conv_1d.html">layer_separable_conv_1d</a>()</code> <br><code><a href="layer_separable_conv_2d.html">layer_separable_conv_2d</a>()</code> <br><code><a href="layer_simple_rnn.html">layer_simple_rnn</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br><code><a href="layer_spatial_dropout_1d.html">layer_spatial_dropout_1d</a>()</code> <br><code><a href="layer_spatial_dropout_2d.html">layer_spatial_dropout_2d</a>()</code> <br><code><a href="layer_spatial_dropout_3d.html">layer_spatial_dropout_3d</a>()</code> <br><code><a href="layer_spectral_normalization.html">layer_spectral_normalization</a>()</code> <br><code><a href="layer_stft_spectrogram.html">layer_stft_spectrogram</a>()</code> <br><code><a href="layer_string_lookup.html">layer_string_lookup</a>()</code> <br><code><a href="layer_subtract.html">layer_subtract</a>()</code> <br><code><a href="layer_text_vectorization.html">layer_text_vectorization</a>()</code> <br><code><a href="layer_tfsm.html">layer_tfsm</a>()</code> <br><code><a href="layer_time_distributed.html">layer_time_distributed</a>()</code> <br><code><a href="layer_torch_module_wrapper.html">layer_torch_module_wrapper</a>()</code> <br><code><a href="layer_unit_normalization.html">layer_unit_normalization</a>()</code> <br><code><a href="layer_upsampling_1d.html">layer_upsampling_1d</a>()</code> <br><code><a href="layer_upsampling_2d.html">layer_upsampling_2d</a>()</code> <br><code><a href="layer_upsampling_3d.html">layer_upsampling_3d</a>()</code> <br><code><a href="layer_zero_padding_1d.html">layer_zero_padding_1d</a>()</code> <br><code><a href="layer_zero_padding_2d.html">layer_zero_padding_2d</a>()</code> <br><code><a href="layer_zero_padding_3d.html">layer_zero_padding_3d</a>()</code> <br><code><a href="rnn_cell_gru.html">rnn_cell_gru</a>()</code> <br><code><a href="rnn_cell_lstm.html">rnn_cell_lstm</a>()</code> <br><code><a href="rnn_cell_simple.html">rnn_cell_simple</a>()</code> <br><code><a href="rnn_cells_stack.html">rnn_cells_stack</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -119,7 +119,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/layer_mel_spectrogram.html b/docs/dev/reference/layer_mel_spectrogram.html
index 38f9f2b4c..4925274a8 100644
--- a/docs/dev/reference/layer_mel_spectrogram.html
+++ b/docs/dev/reference/layer_mel_spectrogram.html
@@ -30,7 +30,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -230,8 +230,8 @@ <h2 id="output-shape">Output Shape<a class="anchor" aria-label="anchor" href="#o
     <div class="section level2">
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
     <div class="dont-index"><p>Other audio preprocessing layers: <br><code><a href="layer_stft_spectrogram.html">layer_stft_spectrogram</a>()</code> <br></p>
-<p>Other preprocessing layers: <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_category_encoding.html">layer_category_encoding</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_discretization.html">layer_discretization</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_feature_space.html">layer_feature_space</a>()</code> <br><code><a href="layer_hashed_crossing.html">layer_hashed_crossing</a>()</code> <br><code><a href="layer_hashing.html">layer_hashing</a>()</code> <br><code><a href="layer_integer_lookup.html">layer_integer_lookup</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_normalization.html">layer_normalization</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_brightness.html">layer_random_brightness</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_contrast.html">layer_random_contrast</a>()</code> <br><code><a href="layer_random_crop.html">layer_random_crop</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_flip.html">layer_random_flip</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_rotation.html">layer_random_rotation</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_random_translation.html">layer_random_translation</a>()</code> <br><code><a href="layer_random_zoom.html">layer_random_zoom</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br><code><a href="layer_stft_spectrogram.html">layer_stft_spectrogram</a>()</code> <br><code><a href="layer_string_lookup.html">layer_string_lookup</a>()</code> <br><code><a href="layer_text_vectorization.html">layer_text_vectorization</a>()</code> <br></p>
-<p>Other layers: <br><code><a href="Layer.html">Layer</a>()</code> <br><code><a href="layer_activation.html">layer_activation</a>()</code> <br><code><a href="layer_activation_elu.html">layer_activation_elu</a>()</code> <br><code><a href="layer_activation_leaky_relu.html">layer_activation_leaky_relu</a>()</code> <br><code><a href="layer_activation_parametric_relu.html">layer_activation_parametric_relu</a>()</code> <br><code><a href="layer_activation_relu.html">layer_activation_relu</a>()</code> <br><code><a href="layer_activation_softmax.html">layer_activation_softmax</a>()</code> <br><code><a href="layer_activity_regularization.html">layer_activity_regularization</a>()</code> <br><code><a href="layer_add.html">layer_add</a>()</code> <br><code><a href="layer_additive_attention.html">layer_additive_attention</a>()</code> <br><code><a href="layer_alpha_dropout.html">layer_alpha_dropout</a>()</code> <br><code><a href="layer_attention.html">layer_attention</a>()</code> <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_average.html">layer_average</a>()</code> <br><code><a href="layer_average_pooling_1d.html">layer_average_pooling_1d</a>()</code> <br><code><a href="layer_average_pooling_2d.html">layer_average_pooling_2d</a>()</code> <br><code><a href="layer_average_pooling_3d.html">layer_average_pooling_3d</a>()</code> <br><code><a href="layer_batch_normalization.html">layer_batch_normalization</a>()</code> <br><code><a href="layer_bidirectional.html">layer_bidirectional</a>()</code> <br><code><a href="layer_category_encoding.html">layer_category_encoding</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_concatenate.html">layer_concatenate</a>()</code> <br><code><a href="layer_conv_1d.html">layer_conv_1d</a>()</code> <br><code><a href="layer_conv_1d_transpose.html">layer_conv_1d_transpose</a>()</code> <br><code><a href="layer_conv_2d.html">layer_conv_2d</a>()</code> <br><code><a href="layer_conv_2d_transpose.html">layer_conv_2d_transpose</a>()</code> <br><code><a href="layer_conv_3d.html">layer_conv_3d</a>()</code> <br><code><a href="layer_conv_3d_transpose.html">layer_conv_3d_transpose</a>()</code> <br><code><a href="layer_conv_lstm_1d.html">layer_conv_lstm_1d</a>()</code> <br><code><a href="layer_conv_lstm_2d.html">layer_conv_lstm_2d</a>()</code> <br><code><a href="layer_conv_lstm_3d.html">layer_conv_lstm_3d</a>()</code> <br><code><a href="layer_cropping_1d.html">layer_cropping_1d</a>()</code> <br><code><a href="layer_cropping_2d.html">layer_cropping_2d</a>()</code> <br><code><a href="layer_cropping_3d.html">layer_cropping_3d</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_dense.html">layer_dense</a>()</code> <br><code><a href="layer_depthwise_conv_1d.html">layer_depthwise_conv_1d</a>()</code> <br><code><a href="layer_depthwise_conv_2d.html">layer_depthwise_conv_2d</a>()</code> <br><code><a href="layer_discretization.html">layer_discretization</a>()</code> <br><code><a href="layer_dot.html">layer_dot</a>()</code> <br><code><a href="layer_dropout.html">layer_dropout</a>()</code> <br><code><a href="layer_einsum_dense.html">layer_einsum_dense</a>()</code> <br><code><a href="layer_embedding.html">layer_embedding</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_feature_space.html">layer_feature_space</a>()</code> <br><code><a href="layer_flatten.html">layer_flatten</a>()</code> <br><code><a href="layer_flax_module_wrapper.html">layer_flax_module_wrapper</a>()</code> <br><code><a href="layer_gaussian_dropout.html">layer_gaussian_dropout</a>()</code> <br><code><a href="layer_gaussian_noise.html">layer_gaussian_noise</a>()</code> <br><code><a href="layer_global_average_pooling_1d.html">layer_global_average_pooling_1d</a>()</code> <br><code><a href="layer_global_average_pooling_2d.html">layer_global_average_pooling_2d</a>()</code> <br><code><a href="layer_global_average_pooling_3d.html">layer_global_average_pooling_3d</a>()</code> <br><code><a href="layer_global_max_pooling_1d.html">layer_global_max_pooling_1d</a>()</code> <br><code><a href="layer_global_max_pooling_2d.html">layer_global_max_pooling_2d</a>()</code> <br><code><a href="layer_global_max_pooling_3d.html">layer_global_max_pooling_3d</a>()</code> <br><code><a href="layer_group_normalization.html">layer_group_normalization</a>()</code> <br><code><a href="layer_group_query_attention.html">layer_group_query_attention</a>()</code> <br><code><a href="layer_gru.html">layer_gru</a>()</code> <br><code><a href="layer_hashed_crossing.html">layer_hashed_crossing</a>()</code> <br><code><a href="layer_hashing.html">layer_hashing</a>()</code> <br><code><a href="layer_identity.html">layer_identity</a>()</code> <br><code><a href="layer_integer_lookup.html">layer_integer_lookup</a>()</code> <br><code><a href="layer_jax_model_wrapper.html">layer_jax_model_wrapper</a>()</code> <br><code><a href="layer_lambda.html">layer_lambda</a>()</code> <br><code><a href="layer_layer_normalization.html">layer_layer_normalization</a>()</code> <br><code><a href="layer_lstm.html">layer_lstm</a>()</code> <br><code><a href="layer_masking.html">layer_masking</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_max_pooling_1d.html">layer_max_pooling_1d</a>()</code> <br><code><a href="layer_max_pooling_2d.html">layer_max_pooling_2d</a>()</code> <br><code><a href="layer_max_pooling_3d.html">layer_max_pooling_3d</a>()</code> <br><code><a href="layer_maximum.html">layer_maximum</a>()</code> <br><code><a href="layer_minimum.html">layer_minimum</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_multi_head_attention.html">layer_multi_head_attention</a>()</code> <br><code><a href="layer_multiply.html">layer_multiply</a>()</code> <br><code><a href="layer_normalization.html">layer_normalization</a>()</code> <br><code><a href="layer_permute.html">layer_permute</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_brightness.html">layer_random_brightness</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_contrast.html">layer_random_contrast</a>()</code> <br><code><a href="layer_random_crop.html">layer_random_crop</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_flip.html">layer_random_flip</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_rotation.html">layer_random_rotation</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_random_translation.html">layer_random_translation</a>()</code> <br><code><a href="layer_random_zoom.html">layer_random_zoom</a>()</code> <br><code><a href="layer_repeat_vector.html">layer_repeat_vector</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_reshape.html">layer_reshape</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_rms_normalization.html">layer_rms_normalization</a>()</code> <br><code><a href="layer_rnn.html">layer_rnn</a>()</code> <br><code><a href="layer_separable_conv_1d.html">layer_separable_conv_1d</a>()</code> <br><code><a href="layer_separable_conv_2d.html">layer_separable_conv_2d</a>()</code> <br><code><a href="layer_simple_rnn.html">layer_simple_rnn</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br><code><a href="layer_spatial_dropout_1d.html">layer_spatial_dropout_1d</a>()</code> <br><code><a href="layer_spatial_dropout_2d.html">layer_spatial_dropout_2d</a>()</code> <br><code><a href="layer_spatial_dropout_3d.html">layer_spatial_dropout_3d</a>()</code> <br><code><a href="layer_spectral_normalization.html">layer_spectral_normalization</a>()</code> <br><code><a href="layer_stft_spectrogram.html">layer_stft_spectrogram</a>()</code> <br><code><a href="layer_string_lookup.html">layer_string_lookup</a>()</code> <br><code><a href="layer_subtract.html">layer_subtract</a>()</code> <br><code><a href="layer_text_vectorization.html">layer_text_vectorization</a>()</code> <br><code><a href="layer_tfsm.html">layer_tfsm</a>()</code> <br><code><a href="layer_time_distributed.html">layer_time_distributed</a>()</code> <br><code><a href="layer_torch_module_wrapper.html">layer_torch_module_wrapper</a>()</code> <br><code><a href="layer_unit_normalization.html">layer_unit_normalization</a>()</code> <br><code><a href="layer_upsampling_1d.html">layer_upsampling_1d</a>()</code> <br><code><a href="layer_upsampling_2d.html">layer_upsampling_2d</a>()</code> <br><code><a href="layer_upsampling_3d.html">layer_upsampling_3d</a>()</code> <br><code><a href="layer_zero_padding_1d.html">layer_zero_padding_1d</a>()</code> <br><code><a href="layer_zero_padding_2d.html">layer_zero_padding_2d</a>()</code> <br><code><a href="layer_zero_padding_3d.html">layer_zero_padding_3d</a>()</code> <br><code><a href="rnn_cell_gru.html">rnn_cell_gru</a>()</code> <br><code><a href="rnn_cell_lstm.html">rnn_cell_lstm</a>()</code> <br><code><a href="rnn_cell_simple.html">rnn_cell_simple</a>()</code> <br><code><a href="rnn_cells_stack.html">rnn_cells_stack</a>()</code> <br></p></div>
+<p>Other preprocessing layers: <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_category_encoding.html">layer_category_encoding</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_discretization.html">layer_discretization</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_feature_space.html">layer_feature_space</a>()</code> <br><code><a href="layer_hashed_crossing.html">layer_hashed_crossing</a>()</code> <br><code><a href="layer_hashing.html">layer_hashing</a>()</code> <br><code><a href="layer_integer_lookup.html">layer_integer_lookup</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_normalization.html">layer_normalization</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_brightness.html">layer_random_brightness</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_contrast.html">layer_random_contrast</a>()</code> <br><code><a href="layer_random_crop.html">layer_random_crop</a>()</code> <br><code><a href="layer_random_elastic_transform.html">layer_random_elastic_transform</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_flip.html">layer_random_flip</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_rotation.html">layer_random_rotation</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_random_translation.html">layer_random_translation</a>()</code> <br><code><a href="layer_random_zoom.html">layer_random_zoom</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br><code><a href="layer_stft_spectrogram.html">layer_stft_spectrogram</a>()</code> <br><code><a href="layer_string_lookup.html">layer_string_lookup</a>()</code> <br><code><a href="layer_text_vectorization.html">layer_text_vectorization</a>()</code> <br></p>
+<p>Other layers: <br><code><a href="Layer.html">Layer</a>()</code> <br><code><a href="layer_activation.html">layer_activation</a>()</code> <br><code><a href="layer_activation_elu.html">layer_activation_elu</a>()</code> <br><code><a href="layer_activation_leaky_relu.html">layer_activation_leaky_relu</a>()</code> <br><code><a href="layer_activation_parametric_relu.html">layer_activation_parametric_relu</a>()</code> <br><code><a href="layer_activation_relu.html">layer_activation_relu</a>()</code> <br><code><a href="layer_activation_softmax.html">layer_activation_softmax</a>()</code> <br><code><a href="layer_activity_regularization.html">layer_activity_regularization</a>()</code> <br><code><a href="layer_add.html">layer_add</a>()</code> <br><code><a href="layer_additive_attention.html">layer_additive_attention</a>()</code> <br><code><a href="layer_alpha_dropout.html">layer_alpha_dropout</a>()</code> <br><code><a href="layer_attention.html">layer_attention</a>()</code> <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_average.html">layer_average</a>()</code> <br><code><a href="layer_average_pooling_1d.html">layer_average_pooling_1d</a>()</code> <br><code><a href="layer_average_pooling_2d.html">layer_average_pooling_2d</a>()</code> <br><code><a href="layer_average_pooling_3d.html">layer_average_pooling_3d</a>()</code> <br><code><a href="layer_batch_normalization.html">layer_batch_normalization</a>()</code> <br><code><a href="layer_bidirectional.html">layer_bidirectional</a>()</code> <br><code><a href="layer_category_encoding.html">layer_category_encoding</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_concatenate.html">layer_concatenate</a>()</code> <br><code><a href="layer_conv_1d.html">layer_conv_1d</a>()</code> <br><code><a href="layer_conv_1d_transpose.html">layer_conv_1d_transpose</a>()</code> <br><code><a href="layer_conv_2d.html">layer_conv_2d</a>()</code> <br><code><a href="layer_conv_2d_transpose.html">layer_conv_2d_transpose</a>()</code> <br><code><a href="layer_conv_3d.html">layer_conv_3d</a>()</code> <br><code><a href="layer_conv_3d_transpose.html">layer_conv_3d_transpose</a>()</code> <br><code><a href="layer_conv_lstm_1d.html">layer_conv_lstm_1d</a>()</code> <br><code><a href="layer_conv_lstm_2d.html">layer_conv_lstm_2d</a>()</code> <br><code><a href="layer_conv_lstm_3d.html">layer_conv_lstm_3d</a>()</code> <br><code><a href="layer_cropping_1d.html">layer_cropping_1d</a>()</code> <br><code><a href="layer_cropping_2d.html">layer_cropping_2d</a>()</code> <br><code><a href="layer_cropping_3d.html">layer_cropping_3d</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_dense.html">layer_dense</a>()</code> <br><code><a href="layer_depthwise_conv_1d.html">layer_depthwise_conv_1d</a>()</code> <br><code><a href="layer_depthwise_conv_2d.html">layer_depthwise_conv_2d</a>()</code> <br><code><a href="layer_discretization.html">layer_discretization</a>()</code> <br><code><a href="layer_dot.html">layer_dot</a>()</code> <br><code><a href="layer_dropout.html">layer_dropout</a>()</code> <br><code><a href="layer_einsum_dense.html">layer_einsum_dense</a>()</code> <br><code><a href="layer_embedding.html">layer_embedding</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_feature_space.html">layer_feature_space</a>()</code> <br><code><a href="layer_flatten.html">layer_flatten</a>()</code> <br><code><a href="layer_flax_module_wrapper.html">layer_flax_module_wrapper</a>()</code> <br><code><a href="layer_gaussian_dropout.html">layer_gaussian_dropout</a>()</code> <br><code><a href="layer_gaussian_noise.html">layer_gaussian_noise</a>()</code> <br><code><a href="layer_global_average_pooling_1d.html">layer_global_average_pooling_1d</a>()</code> <br><code><a href="layer_global_average_pooling_2d.html">layer_global_average_pooling_2d</a>()</code> <br><code><a href="layer_global_average_pooling_3d.html">layer_global_average_pooling_3d</a>()</code> <br><code><a href="layer_global_max_pooling_1d.html">layer_global_max_pooling_1d</a>()</code> <br><code><a href="layer_global_max_pooling_2d.html">layer_global_max_pooling_2d</a>()</code> <br><code><a href="layer_global_max_pooling_3d.html">layer_global_max_pooling_3d</a>()</code> <br><code><a href="layer_group_normalization.html">layer_group_normalization</a>()</code> <br><code><a href="layer_group_query_attention.html">layer_group_query_attention</a>()</code> <br><code><a href="layer_gru.html">layer_gru</a>()</code> <br><code><a href="layer_hashed_crossing.html">layer_hashed_crossing</a>()</code> <br><code><a href="layer_hashing.html">layer_hashing</a>()</code> <br><code><a href="layer_identity.html">layer_identity</a>()</code> <br><code><a href="layer_integer_lookup.html">layer_integer_lookup</a>()</code> <br><code><a href="layer_jax_model_wrapper.html">layer_jax_model_wrapper</a>()</code> <br><code><a href="layer_lambda.html">layer_lambda</a>()</code> <br><code><a href="layer_layer_normalization.html">layer_layer_normalization</a>()</code> <br><code><a href="layer_lstm.html">layer_lstm</a>()</code> <br><code><a href="layer_masking.html">layer_masking</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_max_pooling_1d.html">layer_max_pooling_1d</a>()</code> <br><code><a href="layer_max_pooling_2d.html">layer_max_pooling_2d</a>()</code> <br><code><a href="layer_max_pooling_3d.html">layer_max_pooling_3d</a>()</code> <br><code><a href="layer_maximum.html">layer_maximum</a>()</code> <br><code><a href="layer_minimum.html">layer_minimum</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_multi_head_attention.html">layer_multi_head_attention</a>()</code> <br><code><a href="layer_multiply.html">layer_multiply</a>()</code> <br><code><a href="layer_normalization.html">layer_normalization</a>()</code> <br><code><a href="layer_permute.html">layer_permute</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_brightness.html">layer_random_brightness</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_contrast.html">layer_random_contrast</a>()</code> <br><code><a href="layer_random_crop.html">layer_random_crop</a>()</code> <br><code><a href="layer_random_elastic_transform.html">layer_random_elastic_transform</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_flip.html">layer_random_flip</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_rotation.html">layer_random_rotation</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_random_translation.html">layer_random_translation</a>()</code> <br><code><a href="layer_random_zoom.html">layer_random_zoom</a>()</code> <br><code><a href="layer_repeat_vector.html">layer_repeat_vector</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_reshape.html">layer_reshape</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_rms_normalization.html">layer_rms_normalization</a>()</code> <br><code><a href="layer_rnn.html">layer_rnn</a>()</code> <br><code><a href="layer_separable_conv_1d.html">layer_separable_conv_1d</a>()</code> <br><code><a href="layer_separable_conv_2d.html">layer_separable_conv_2d</a>()</code> <br><code><a href="layer_simple_rnn.html">layer_simple_rnn</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br><code><a href="layer_spatial_dropout_1d.html">layer_spatial_dropout_1d</a>()</code> <br><code><a href="layer_spatial_dropout_2d.html">layer_spatial_dropout_2d</a>()</code> <br><code><a href="layer_spatial_dropout_3d.html">layer_spatial_dropout_3d</a>()</code> <br><code><a href="layer_spectral_normalization.html">layer_spectral_normalization</a>()</code> <br><code><a href="layer_stft_spectrogram.html">layer_stft_spectrogram</a>()</code> <br><code><a href="layer_string_lookup.html">layer_string_lookup</a>()</code> <br><code><a href="layer_subtract.html">layer_subtract</a>()</code> <br><code><a href="layer_text_vectorization.html">layer_text_vectorization</a>()</code> <br><code><a href="layer_tfsm.html">layer_tfsm</a>()</code> <br><code><a href="layer_time_distributed.html">layer_time_distributed</a>()</code> <br><code><a href="layer_torch_module_wrapper.html">layer_torch_module_wrapper</a>()</code> <br><code><a href="layer_unit_normalization.html">layer_unit_normalization</a>()</code> <br><code><a href="layer_upsampling_1d.html">layer_upsampling_1d</a>()</code> <br><code><a href="layer_upsampling_2d.html">layer_upsampling_2d</a>()</code> <br><code><a href="layer_upsampling_3d.html">layer_upsampling_3d</a>()</code> <br><code><a href="layer_zero_padding_1d.html">layer_zero_padding_1d</a>()</code> <br><code><a href="layer_zero_padding_2d.html">layer_zero_padding_2d</a>()</code> <br><code><a href="layer_zero_padding_3d.html">layer_zero_padding_3d</a>()</code> <br><code><a href="rnn_cell_gru.html">rnn_cell_gru</a>()</code> <br><code><a href="rnn_cell_lstm.html">rnn_cell_lstm</a>()</code> <br><code><a href="rnn_cell_simple.html">rnn_cell_simple</a>()</code> <br><code><a href="rnn_cells_stack.html">rnn_cells_stack</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -243,7 +243,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/layer_minimum.html b/docs/dev/reference/layer_minimum.html
index 8121c88f1..373950436 100644
--- a/docs/dev/reference/layer_minimum.html
+++ b/docs/dev/reference/layer_minimum.html
@@ -10,7 +10,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -107,7 +107,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
     <div class="dont-index">
 <ul><li><p><a href="https://keras.io/api/layers/merging_layers/minimum#minimum-class" class="external-link">https://keras.io/api/layers/merging_layers/minimum#minimum-class</a></p></li>
 </ul><p>Other merging layers: <br><code><a href="layer_add.html">layer_add</a>()</code> <br><code><a href="layer_average.html">layer_average</a>()</code> <br><code><a href="layer_concatenate.html">layer_concatenate</a>()</code> <br><code><a href="layer_dot.html">layer_dot</a>()</code> <br><code><a href="layer_maximum.html">layer_maximum</a>()</code> <br><code><a href="layer_multiply.html">layer_multiply</a>()</code> <br><code><a href="layer_subtract.html">layer_subtract</a>()</code> <br></p>
-<p>Other layers: <br><code><a href="Layer.html">Layer</a>()</code> <br><code><a href="layer_activation.html">layer_activation</a>()</code> <br><code><a href="layer_activation_elu.html">layer_activation_elu</a>()</code> <br><code><a href="layer_activation_leaky_relu.html">layer_activation_leaky_relu</a>()</code> <br><code><a href="layer_activation_parametric_relu.html">layer_activation_parametric_relu</a>()</code> <br><code><a href="layer_activation_relu.html">layer_activation_relu</a>()</code> <br><code><a href="layer_activation_softmax.html">layer_activation_softmax</a>()</code> <br><code><a href="layer_activity_regularization.html">layer_activity_regularization</a>()</code> <br><code><a href="layer_add.html">layer_add</a>()</code> <br><code><a href="layer_additive_attention.html">layer_additive_attention</a>()</code> <br><code><a href="layer_alpha_dropout.html">layer_alpha_dropout</a>()</code> <br><code><a href="layer_attention.html">layer_attention</a>()</code> <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_average.html">layer_average</a>()</code> <br><code><a href="layer_average_pooling_1d.html">layer_average_pooling_1d</a>()</code> <br><code><a href="layer_average_pooling_2d.html">layer_average_pooling_2d</a>()</code> <br><code><a href="layer_average_pooling_3d.html">layer_average_pooling_3d</a>()</code> <br><code><a href="layer_batch_normalization.html">layer_batch_normalization</a>()</code> <br><code><a href="layer_bidirectional.html">layer_bidirectional</a>()</code> <br><code><a href="layer_category_encoding.html">layer_category_encoding</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_concatenate.html">layer_concatenate</a>()</code> <br><code><a href="layer_conv_1d.html">layer_conv_1d</a>()</code> <br><code><a href="layer_conv_1d_transpose.html">layer_conv_1d_transpose</a>()</code> <br><code><a href="layer_conv_2d.html">layer_conv_2d</a>()</code> <br><code><a href="layer_conv_2d_transpose.html">layer_conv_2d_transpose</a>()</code> <br><code><a href="layer_conv_3d.html">layer_conv_3d</a>()</code> <br><code><a href="layer_conv_3d_transpose.html">layer_conv_3d_transpose</a>()</code> <br><code><a href="layer_conv_lstm_1d.html">layer_conv_lstm_1d</a>()</code> <br><code><a href="layer_conv_lstm_2d.html">layer_conv_lstm_2d</a>()</code> <br><code><a href="layer_conv_lstm_3d.html">layer_conv_lstm_3d</a>()</code> <br><code><a href="layer_cropping_1d.html">layer_cropping_1d</a>()</code> <br><code><a href="layer_cropping_2d.html">layer_cropping_2d</a>()</code> <br><code><a href="layer_cropping_3d.html">layer_cropping_3d</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_dense.html">layer_dense</a>()</code> <br><code><a href="layer_depthwise_conv_1d.html">layer_depthwise_conv_1d</a>()</code> <br><code><a href="layer_depthwise_conv_2d.html">layer_depthwise_conv_2d</a>()</code> <br><code><a href="layer_discretization.html">layer_discretization</a>()</code> <br><code><a href="layer_dot.html">layer_dot</a>()</code> <br><code><a href="layer_dropout.html">layer_dropout</a>()</code> <br><code><a href="layer_einsum_dense.html">layer_einsum_dense</a>()</code> <br><code><a href="layer_embedding.html">layer_embedding</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_feature_space.html">layer_feature_space</a>()</code> <br><code><a href="layer_flatten.html">layer_flatten</a>()</code> <br><code><a href="layer_flax_module_wrapper.html">layer_flax_module_wrapper</a>()</code> <br><code><a href="layer_gaussian_dropout.html">layer_gaussian_dropout</a>()</code> <br><code><a href="layer_gaussian_noise.html">layer_gaussian_noise</a>()</code> <br><code><a href="layer_global_average_pooling_1d.html">layer_global_average_pooling_1d</a>()</code> <br><code><a href="layer_global_average_pooling_2d.html">layer_global_average_pooling_2d</a>()</code> <br><code><a href="layer_global_average_pooling_3d.html">layer_global_average_pooling_3d</a>()</code> <br><code><a href="layer_global_max_pooling_1d.html">layer_global_max_pooling_1d</a>()</code> <br><code><a href="layer_global_max_pooling_2d.html">layer_global_max_pooling_2d</a>()</code> <br><code><a href="layer_global_max_pooling_3d.html">layer_global_max_pooling_3d</a>()</code> <br><code><a href="layer_group_normalization.html">layer_group_normalization</a>()</code> <br><code><a href="layer_group_query_attention.html">layer_group_query_attention</a>()</code> <br><code><a href="layer_gru.html">layer_gru</a>()</code> <br><code><a href="layer_hashed_crossing.html">layer_hashed_crossing</a>()</code> <br><code><a href="layer_hashing.html">layer_hashing</a>()</code> <br><code><a href="layer_identity.html">layer_identity</a>()</code> <br><code><a href="layer_integer_lookup.html">layer_integer_lookup</a>()</code> <br><code><a href="layer_jax_model_wrapper.html">layer_jax_model_wrapper</a>()</code> <br><code><a href="layer_lambda.html">layer_lambda</a>()</code> <br><code><a href="layer_layer_normalization.html">layer_layer_normalization</a>()</code> <br><code><a href="layer_lstm.html">layer_lstm</a>()</code> <br><code><a href="layer_masking.html">layer_masking</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_max_pooling_1d.html">layer_max_pooling_1d</a>()</code> <br><code><a href="layer_max_pooling_2d.html">layer_max_pooling_2d</a>()</code> <br><code><a href="layer_max_pooling_3d.html">layer_max_pooling_3d</a>()</code> <br><code><a href="layer_maximum.html">layer_maximum</a>()</code> <br><code><a href="layer_mel_spectrogram.html">layer_mel_spectrogram</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_multi_head_attention.html">layer_multi_head_attention</a>()</code> <br><code><a href="layer_multiply.html">layer_multiply</a>()</code> <br><code><a href="layer_normalization.html">layer_normalization</a>()</code> <br><code><a href="layer_permute.html">layer_permute</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_brightness.html">layer_random_brightness</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_contrast.html">layer_random_contrast</a>()</code> <br><code><a href="layer_random_crop.html">layer_random_crop</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_flip.html">layer_random_flip</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_rotation.html">layer_random_rotation</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_random_translation.html">layer_random_translation</a>()</code> <br><code><a href="layer_random_zoom.html">layer_random_zoom</a>()</code> <br><code><a href="layer_repeat_vector.html">layer_repeat_vector</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_reshape.html">layer_reshape</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_rms_normalization.html">layer_rms_normalization</a>()</code> <br><code><a href="layer_rnn.html">layer_rnn</a>()</code> <br><code><a href="layer_separable_conv_1d.html">layer_separable_conv_1d</a>()</code> <br><code><a href="layer_separable_conv_2d.html">layer_separable_conv_2d</a>()</code> <br><code><a href="layer_simple_rnn.html">layer_simple_rnn</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br><code><a href="layer_spatial_dropout_1d.html">layer_spatial_dropout_1d</a>()</code> <br><code><a href="layer_spatial_dropout_2d.html">layer_spatial_dropout_2d</a>()</code> <br><code><a href="layer_spatial_dropout_3d.html">layer_spatial_dropout_3d</a>()</code> <br><code><a href="layer_spectral_normalization.html">layer_spectral_normalization</a>()</code> <br><code><a href="layer_stft_spectrogram.html">layer_stft_spectrogram</a>()</code> <br><code><a href="layer_string_lookup.html">layer_string_lookup</a>()</code> <br><code><a href="layer_subtract.html">layer_subtract</a>()</code> <br><code><a href="layer_text_vectorization.html">layer_text_vectorization</a>()</code> <br><code><a href="layer_tfsm.html">layer_tfsm</a>()</code> <br><code><a href="layer_time_distributed.html">layer_time_distributed</a>()</code> <br><code><a href="layer_torch_module_wrapper.html">layer_torch_module_wrapper</a>()</code> <br><code><a href="layer_unit_normalization.html">layer_unit_normalization</a>()</code> <br><code><a href="layer_upsampling_1d.html">layer_upsampling_1d</a>()</code> <br><code><a href="layer_upsampling_2d.html">layer_upsampling_2d</a>()</code> <br><code><a href="layer_upsampling_3d.html">layer_upsampling_3d</a>()</code> <br><code><a href="layer_zero_padding_1d.html">layer_zero_padding_1d</a>()</code> <br><code><a href="layer_zero_padding_2d.html">layer_zero_padding_2d</a>()</code> <br><code><a href="layer_zero_padding_3d.html">layer_zero_padding_3d</a>()</code> <br><code><a href="rnn_cell_gru.html">rnn_cell_gru</a>()</code> <br><code><a href="rnn_cell_lstm.html">rnn_cell_lstm</a>()</code> <br><code><a href="rnn_cell_simple.html">rnn_cell_simple</a>()</code> <br><code><a href="rnn_cells_stack.html">rnn_cells_stack</a>()</code> <br></p></div>
+<p>Other layers: <br><code><a href="Layer.html">Layer</a>()</code> <br><code><a href="layer_activation.html">layer_activation</a>()</code> <br><code><a href="layer_activation_elu.html">layer_activation_elu</a>()</code> <br><code><a href="layer_activation_leaky_relu.html">layer_activation_leaky_relu</a>()</code> <br><code><a href="layer_activation_parametric_relu.html">layer_activation_parametric_relu</a>()</code> <br><code><a href="layer_activation_relu.html">layer_activation_relu</a>()</code> <br><code><a href="layer_activation_softmax.html">layer_activation_softmax</a>()</code> <br><code><a href="layer_activity_regularization.html">layer_activity_regularization</a>()</code> <br><code><a href="layer_add.html">layer_add</a>()</code> <br><code><a href="layer_additive_attention.html">layer_additive_attention</a>()</code> <br><code><a href="layer_alpha_dropout.html">layer_alpha_dropout</a>()</code> <br><code><a href="layer_attention.html">layer_attention</a>()</code> <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_average.html">layer_average</a>()</code> <br><code><a href="layer_average_pooling_1d.html">layer_average_pooling_1d</a>()</code> <br><code><a href="layer_average_pooling_2d.html">layer_average_pooling_2d</a>()</code> <br><code><a href="layer_average_pooling_3d.html">layer_average_pooling_3d</a>()</code> <br><code><a href="layer_batch_normalization.html">layer_batch_normalization</a>()</code> <br><code><a href="layer_bidirectional.html">layer_bidirectional</a>()</code> <br><code><a href="layer_category_encoding.html">layer_category_encoding</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_concatenate.html">layer_concatenate</a>()</code> <br><code><a href="layer_conv_1d.html">layer_conv_1d</a>()</code> <br><code><a href="layer_conv_1d_transpose.html">layer_conv_1d_transpose</a>()</code> <br><code><a href="layer_conv_2d.html">layer_conv_2d</a>()</code> <br><code><a href="layer_conv_2d_transpose.html">layer_conv_2d_transpose</a>()</code> <br><code><a href="layer_conv_3d.html">layer_conv_3d</a>()</code> <br><code><a href="layer_conv_3d_transpose.html">layer_conv_3d_transpose</a>()</code> <br><code><a href="layer_conv_lstm_1d.html">layer_conv_lstm_1d</a>()</code> <br><code><a href="layer_conv_lstm_2d.html">layer_conv_lstm_2d</a>()</code> <br><code><a href="layer_conv_lstm_3d.html">layer_conv_lstm_3d</a>()</code> <br><code><a href="layer_cropping_1d.html">layer_cropping_1d</a>()</code> <br><code><a href="layer_cropping_2d.html">layer_cropping_2d</a>()</code> <br><code><a href="layer_cropping_3d.html">layer_cropping_3d</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_dense.html">layer_dense</a>()</code> <br><code><a href="layer_depthwise_conv_1d.html">layer_depthwise_conv_1d</a>()</code> <br><code><a href="layer_depthwise_conv_2d.html">layer_depthwise_conv_2d</a>()</code> <br><code><a href="layer_discretization.html">layer_discretization</a>()</code> <br><code><a href="layer_dot.html">layer_dot</a>()</code> <br><code><a href="layer_dropout.html">layer_dropout</a>()</code> <br><code><a href="layer_einsum_dense.html">layer_einsum_dense</a>()</code> <br><code><a href="layer_embedding.html">layer_embedding</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_feature_space.html">layer_feature_space</a>()</code> <br><code><a href="layer_flatten.html">layer_flatten</a>()</code> <br><code><a href="layer_flax_module_wrapper.html">layer_flax_module_wrapper</a>()</code> <br><code><a href="layer_gaussian_dropout.html">layer_gaussian_dropout</a>()</code> <br><code><a href="layer_gaussian_noise.html">layer_gaussian_noise</a>()</code> <br><code><a href="layer_global_average_pooling_1d.html">layer_global_average_pooling_1d</a>()</code> <br><code><a href="layer_global_average_pooling_2d.html">layer_global_average_pooling_2d</a>()</code> <br><code><a href="layer_global_average_pooling_3d.html">layer_global_average_pooling_3d</a>()</code> <br><code><a href="layer_global_max_pooling_1d.html">layer_global_max_pooling_1d</a>()</code> <br><code><a href="layer_global_max_pooling_2d.html">layer_global_max_pooling_2d</a>()</code> <br><code><a href="layer_global_max_pooling_3d.html">layer_global_max_pooling_3d</a>()</code> <br><code><a href="layer_group_normalization.html">layer_group_normalization</a>()</code> <br><code><a href="layer_group_query_attention.html">layer_group_query_attention</a>()</code> <br><code><a href="layer_gru.html">layer_gru</a>()</code> <br><code><a href="layer_hashed_crossing.html">layer_hashed_crossing</a>()</code> <br><code><a href="layer_hashing.html">layer_hashing</a>()</code> <br><code><a href="layer_identity.html">layer_identity</a>()</code> <br><code><a href="layer_integer_lookup.html">layer_integer_lookup</a>()</code> <br><code><a href="layer_jax_model_wrapper.html">layer_jax_model_wrapper</a>()</code> <br><code><a href="layer_lambda.html">layer_lambda</a>()</code> <br><code><a href="layer_layer_normalization.html">layer_layer_normalization</a>()</code> <br><code><a href="layer_lstm.html">layer_lstm</a>()</code> <br><code><a href="layer_masking.html">layer_masking</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_max_pooling_1d.html">layer_max_pooling_1d</a>()</code> <br><code><a href="layer_max_pooling_2d.html">layer_max_pooling_2d</a>()</code> <br><code><a href="layer_max_pooling_3d.html">layer_max_pooling_3d</a>()</code> <br><code><a href="layer_maximum.html">layer_maximum</a>()</code> <br><code><a href="layer_mel_spectrogram.html">layer_mel_spectrogram</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_multi_head_attention.html">layer_multi_head_attention</a>()</code> <br><code><a href="layer_multiply.html">layer_multiply</a>()</code> <br><code><a href="layer_normalization.html">layer_normalization</a>()</code> <br><code><a href="layer_permute.html">layer_permute</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_brightness.html">layer_random_brightness</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_contrast.html">layer_random_contrast</a>()</code> <br><code><a href="layer_random_crop.html">layer_random_crop</a>()</code> <br><code><a href="layer_random_elastic_transform.html">layer_random_elastic_transform</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_flip.html">layer_random_flip</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_rotation.html">layer_random_rotation</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_random_translation.html">layer_random_translation</a>()</code> <br><code><a href="layer_random_zoom.html">layer_random_zoom</a>()</code> <br><code><a href="layer_repeat_vector.html">layer_repeat_vector</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_reshape.html">layer_reshape</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_rms_normalization.html">layer_rms_normalization</a>()</code> <br><code><a href="layer_rnn.html">layer_rnn</a>()</code> <br><code><a href="layer_separable_conv_1d.html">layer_separable_conv_1d</a>()</code> <br><code><a href="layer_separable_conv_2d.html">layer_separable_conv_2d</a>()</code> <br><code><a href="layer_simple_rnn.html">layer_simple_rnn</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br><code><a href="layer_spatial_dropout_1d.html">layer_spatial_dropout_1d</a>()</code> <br><code><a href="layer_spatial_dropout_2d.html">layer_spatial_dropout_2d</a>()</code> <br><code><a href="layer_spatial_dropout_3d.html">layer_spatial_dropout_3d</a>()</code> <br><code><a href="layer_spectral_normalization.html">layer_spectral_normalization</a>()</code> <br><code><a href="layer_stft_spectrogram.html">layer_stft_spectrogram</a>()</code> <br><code><a href="layer_string_lookup.html">layer_string_lookup</a>()</code> <br><code><a href="layer_subtract.html">layer_subtract</a>()</code> <br><code><a href="layer_text_vectorization.html">layer_text_vectorization</a>()</code> <br><code><a href="layer_tfsm.html">layer_tfsm</a>()</code> <br><code><a href="layer_time_distributed.html">layer_time_distributed</a>()</code> <br><code><a href="layer_torch_module_wrapper.html">layer_torch_module_wrapper</a>()</code> <br><code><a href="layer_unit_normalization.html">layer_unit_normalization</a>()</code> <br><code><a href="layer_upsampling_1d.html">layer_upsampling_1d</a>()</code> <br><code><a href="layer_upsampling_2d.html">layer_upsampling_2d</a>()</code> <br><code><a href="layer_upsampling_3d.html">layer_upsampling_3d</a>()</code> <br><code><a href="layer_zero_padding_1d.html">layer_zero_padding_1d</a>()</code> <br><code><a href="layer_zero_padding_2d.html">layer_zero_padding_2d</a>()</code> <br><code><a href="layer_zero_padding_3d.html">layer_zero_padding_3d</a>()</code> <br><code><a href="rnn_cell_gru.html">rnn_cell_gru</a>()</code> <br><code><a href="rnn_cell_lstm.html">rnn_cell_lstm</a>()</code> <br><code><a href="rnn_cell_simple.html">rnn_cell_simple</a>()</code> <br><code><a href="rnn_cells_stack.html">rnn_cells_stack</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -119,7 +119,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/layer_mix_up.html b/docs/dev/reference/layer_mix_up.html
index cf7d39a5b..1994afe6f 100644
--- a/docs/dev/reference/layer_mix_up.html
+++ b/docs/dev/reference/layer_mix_up.html
@@ -8,7 +8,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -110,9 +110,9 @@ <h2 id="examples">Examples<a class="anchor" aria-label="anchor" href="#examples"
     </div>
     <div class="section level2">
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
-    <div class="dont-index"><p>Other image preprocessing layers: <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br></p>
-<p>Other preprocessing layers: <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_category_encoding.html">layer_category_encoding</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_discretization.html">layer_discretization</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_feature_space.html">layer_feature_space</a>()</code> <br><code><a href="layer_hashed_crossing.html">layer_hashed_crossing</a>()</code> <br><code><a href="layer_hashing.html">layer_hashing</a>()</code> <br><code><a href="layer_integer_lookup.html">layer_integer_lookup</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_mel_spectrogram.html">layer_mel_spectrogram</a>()</code> <br><code><a href="layer_normalization.html">layer_normalization</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_brightness.html">layer_random_brightness</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_contrast.html">layer_random_contrast</a>()</code> <br><code><a href="layer_random_crop.html">layer_random_crop</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_flip.html">layer_random_flip</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_rotation.html">layer_random_rotation</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_random_translation.html">layer_random_translation</a>()</code> <br><code><a href="layer_random_zoom.html">layer_random_zoom</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br><code><a href="layer_stft_spectrogram.html">layer_stft_spectrogram</a>()</code> <br><code><a href="layer_string_lookup.html">layer_string_lookup</a>()</code> <br><code><a href="layer_text_vectorization.html">layer_text_vectorization</a>()</code> <br></p>
-<p>Other layers: <br><code><a href="Layer.html">Layer</a>()</code> <br><code><a href="layer_activation.html">layer_activation</a>()</code> <br><code><a href="layer_activation_elu.html">layer_activation_elu</a>()</code> <br><code><a href="layer_activation_leaky_relu.html">layer_activation_leaky_relu</a>()</code> <br><code><a href="layer_activation_parametric_relu.html">layer_activation_parametric_relu</a>()</code> <br><code><a href="layer_activation_relu.html">layer_activation_relu</a>()</code> <br><code><a href="layer_activation_softmax.html">layer_activation_softmax</a>()</code> <br><code><a href="layer_activity_regularization.html">layer_activity_regularization</a>()</code> <br><code><a href="layer_add.html">layer_add</a>()</code> <br><code><a href="layer_additive_attention.html">layer_additive_attention</a>()</code> <br><code><a href="layer_alpha_dropout.html">layer_alpha_dropout</a>()</code> <br><code><a href="layer_attention.html">layer_attention</a>()</code> <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_average.html">layer_average</a>()</code> <br><code><a href="layer_average_pooling_1d.html">layer_average_pooling_1d</a>()</code> <br><code><a href="layer_average_pooling_2d.html">layer_average_pooling_2d</a>()</code> <br><code><a href="layer_average_pooling_3d.html">layer_average_pooling_3d</a>()</code> <br><code><a href="layer_batch_normalization.html">layer_batch_normalization</a>()</code> <br><code><a href="layer_bidirectional.html">layer_bidirectional</a>()</code> <br><code><a href="layer_category_encoding.html">layer_category_encoding</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_concatenate.html">layer_concatenate</a>()</code> <br><code><a href="layer_conv_1d.html">layer_conv_1d</a>()</code> <br><code><a href="layer_conv_1d_transpose.html">layer_conv_1d_transpose</a>()</code> <br><code><a href="layer_conv_2d.html">layer_conv_2d</a>()</code> <br><code><a href="layer_conv_2d_transpose.html">layer_conv_2d_transpose</a>()</code> <br><code><a href="layer_conv_3d.html">layer_conv_3d</a>()</code> <br><code><a href="layer_conv_3d_transpose.html">layer_conv_3d_transpose</a>()</code> <br><code><a href="layer_conv_lstm_1d.html">layer_conv_lstm_1d</a>()</code> <br><code><a href="layer_conv_lstm_2d.html">layer_conv_lstm_2d</a>()</code> <br><code><a href="layer_conv_lstm_3d.html">layer_conv_lstm_3d</a>()</code> <br><code><a href="layer_cropping_1d.html">layer_cropping_1d</a>()</code> <br><code><a href="layer_cropping_2d.html">layer_cropping_2d</a>()</code> <br><code><a href="layer_cropping_3d.html">layer_cropping_3d</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_dense.html">layer_dense</a>()</code> <br><code><a href="layer_depthwise_conv_1d.html">layer_depthwise_conv_1d</a>()</code> <br><code><a href="layer_depthwise_conv_2d.html">layer_depthwise_conv_2d</a>()</code> <br><code><a href="layer_discretization.html">layer_discretization</a>()</code> <br><code><a href="layer_dot.html">layer_dot</a>()</code> <br><code><a href="layer_dropout.html">layer_dropout</a>()</code> <br><code><a href="layer_einsum_dense.html">layer_einsum_dense</a>()</code> <br><code><a href="layer_embedding.html">layer_embedding</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_feature_space.html">layer_feature_space</a>()</code> <br><code><a href="layer_flatten.html">layer_flatten</a>()</code> <br><code><a href="layer_flax_module_wrapper.html">layer_flax_module_wrapper</a>()</code> <br><code><a href="layer_gaussian_dropout.html">layer_gaussian_dropout</a>()</code> <br><code><a href="layer_gaussian_noise.html">layer_gaussian_noise</a>()</code> <br><code><a href="layer_global_average_pooling_1d.html">layer_global_average_pooling_1d</a>()</code> <br><code><a href="layer_global_average_pooling_2d.html">layer_global_average_pooling_2d</a>()</code> <br><code><a href="layer_global_average_pooling_3d.html">layer_global_average_pooling_3d</a>()</code> <br><code><a href="layer_global_max_pooling_1d.html">layer_global_max_pooling_1d</a>()</code> <br><code><a href="layer_global_max_pooling_2d.html">layer_global_max_pooling_2d</a>()</code> <br><code><a href="layer_global_max_pooling_3d.html">layer_global_max_pooling_3d</a>()</code> <br><code><a href="layer_group_normalization.html">layer_group_normalization</a>()</code> <br><code><a href="layer_group_query_attention.html">layer_group_query_attention</a>()</code> <br><code><a href="layer_gru.html">layer_gru</a>()</code> <br><code><a href="layer_hashed_crossing.html">layer_hashed_crossing</a>()</code> <br><code><a href="layer_hashing.html">layer_hashing</a>()</code> <br><code><a href="layer_identity.html">layer_identity</a>()</code> <br><code><a href="layer_integer_lookup.html">layer_integer_lookup</a>()</code> <br><code><a href="layer_jax_model_wrapper.html">layer_jax_model_wrapper</a>()</code> <br><code><a href="layer_lambda.html">layer_lambda</a>()</code> <br><code><a href="layer_layer_normalization.html">layer_layer_normalization</a>()</code> <br><code><a href="layer_lstm.html">layer_lstm</a>()</code> <br><code><a href="layer_masking.html">layer_masking</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_max_pooling_1d.html">layer_max_pooling_1d</a>()</code> <br><code><a href="layer_max_pooling_2d.html">layer_max_pooling_2d</a>()</code> <br><code><a href="layer_max_pooling_3d.html">layer_max_pooling_3d</a>()</code> <br><code><a href="layer_maximum.html">layer_maximum</a>()</code> <br><code><a href="layer_mel_spectrogram.html">layer_mel_spectrogram</a>()</code> <br><code><a href="layer_minimum.html">layer_minimum</a>()</code> <br><code><a href="layer_multi_head_attention.html">layer_multi_head_attention</a>()</code> <br><code><a href="layer_multiply.html">layer_multiply</a>()</code> <br><code><a href="layer_normalization.html">layer_normalization</a>()</code> <br><code><a href="layer_permute.html">layer_permute</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_brightness.html">layer_random_brightness</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_contrast.html">layer_random_contrast</a>()</code> <br><code><a href="layer_random_crop.html">layer_random_crop</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_flip.html">layer_random_flip</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_rotation.html">layer_random_rotation</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_random_translation.html">layer_random_translation</a>()</code> <br><code><a href="layer_random_zoom.html">layer_random_zoom</a>()</code> <br><code><a href="layer_repeat_vector.html">layer_repeat_vector</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_reshape.html">layer_reshape</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_rms_normalization.html">layer_rms_normalization</a>()</code> <br><code><a href="layer_rnn.html">layer_rnn</a>()</code> <br><code><a href="layer_separable_conv_1d.html">layer_separable_conv_1d</a>()</code> <br><code><a href="layer_separable_conv_2d.html">layer_separable_conv_2d</a>()</code> <br><code><a href="layer_simple_rnn.html">layer_simple_rnn</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br><code><a href="layer_spatial_dropout_1d.html">layer_spatial_dropout_1d</a>()</code> <br><code><a href="layer_spatial_dropout_2d.html">layer_spatial_dropout_2d</a>()</code> <br><code><a href="layer_spatial_dropout_3d.html">layer_spatial_dropout_3d</a>()</code> <br><code><a href="layer_spectral_normalization.html">layer_spectral_normalization</a>()</code> <br><code><a href="layer_stft_spectrogram.html">layer_stft_spectrogram</a>()</code> <br><code><a href="layer_string_lookup.html">layer_string_lookup</a>()</code> <br><code><a href="layer_subtract.html">layer_subtract</a>()</code> <br><code><a href="layer_text_vectorization.html">layer_text_vectorization</a>()</code> <br><code><a href="layer_tfsm.html">layer_tfsm</a>()</code> <br><code><a href="layer_time_distributed.html">layer_time_distributed</a>()</code> <br><code><a href="layer_torch_module_wrapper.html">layer_torch_module_wrapper</a>()</code> <br><code><a href="layer_unit_normalization.html">layer_unit_normalization</a>()</code> <br><code><a href="layer_upsampling_1d.html">layer_upsampling_1d</a>()</code> <br><code><a href="layer_upsampling_2d.html">layer_upsampling_2d</a>()</code> <br><code><a href="layer_upsampling_3d.html">layer_upsampling_3d</a>()</code> <br><code><a href="layer_zero_padding_1d.html">layer_zero_padding_1d</a>()</code> <br><code><a href="layer_zero_padding_2d.html">layer_zero_padding_2d</a>()</code> <br><code><a href="layer_zero_padding_3d.html">layer_zero_padding_3d</a>()</code> <br><code><a href="rnn_cell_gru.html">rnn_cell_gru</a>()</code> <br><code><a href="rnn_cell_lstm.html">rnn_cell_lstm</a>()</code> <br><code><a href="rnn_cell_simple.html">rnn_cell_simple</a>()</code> <br><code><a href="rnn_cells_stack.html">rnn_cells_stack</a>()</code> <br></p></div>
+    <div class="dont-index"><p>Other image preprocessing layers: <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_elastic_transform.html">layer_random_elastic_transform</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br></p>
+<p>Other preprocessing layers: <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_category_encoding.html">layer_category_encoding</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_discretization.html">layer_discretization</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_feature_space.html">layer_feature_space</a>()</code> <br><code><a href="layer_hashed_crossing.html">layer_hashed_crossing</a>()</code> <br><code><a href="layer_hashing.html">layer_hashing</a>()</code> <br><code><a href="layer_integer_lookup.html">layer_integer_lookup</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_mel_spectrogram.html">layer_mel_spectrogram</a>()</code> <br><code><a href="layer_normalization.html">layer_normalization</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_brightness.html">layer_random_brightness</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_contrast.html">layer_random_contrast</a>()</code> <br><code><a href="layer_random_crop.html">layer_random_crop</a>()</code> <br><code><a href="layer_random_elastic_transform.html">layer_random_elastic_transform</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_flip.html">layer_random_flip</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_rotation.html">layer_random_rotation</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_random_translation.html">layer_random_translation</a>()</code> <br><code><a href="layer_random_zoom.html">layer_random_zoom</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br><code><a href="layer_stft_spectrogram.html">layer_stft_spectrogram</a>()</code> <br><code><a href="layer_string_lookup.html">layer_string_lookup</a>()</code> <br><code><a href="layer_text_vectorization.html">layer_text_vectorization</a>()</code> <br></p>
+<p>Other layers: <br><code><a href="Layer.html">Layer</a>()</code> <br><code><a href="layer_activation.html">layer_activation</a>()</code> <br><code><a href="layer_activation_elu.html">layer_activation_elu</a>()</code> <br><code><a href="layer_activation_leaky_relu.html">layer_activation_leaky_relu</a>()</code> <br><code><a href="layer_activation_parametric_relu.html">layer_activation_parametric_relu</a>()</code> <br><code><a href="layer_activation_relu.html">layer_activation_relu</a>()</code> <br><code><a href="layer_activation_softmax.html">layer_activation_softmax</a>()</code> <br><code><a href="layer_activity_regularization.html">layer_activity_regularization</a>()</code> <br><code><a href="layer_add.html">layer_add</a>()</code> <br><code><a href="layer_additive_attention.html">layer_additive_attention</a>()</code> <br><code><a href="layer_alpha_dropout.html">layer_alpha_dropout</a>()</code> <br><code><a href="layer_attention.html">layer_attention</a>()</code> <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_average.html">layer_average</a>()</code> <br><code><a href="layer_average_pooling_1d.html">layer_average_pooling_1d</a>()</code> <br><code><a href="layer_average_pooling_2d.html">layer_average_pooling_2d</a>()</code> <br><code><a href="layer_average_pooling_3d.html">layer_average_pooling_3d</a>()</code> <br><code><a href="layer_batch_normalization.html">layer_batch_normalization</a>()</code> <br><code><a href="layer_bidirectional.html">layer_bidirectional</a>()</code> <br><code><a href="layer_category_encoding.html">layer_category_encoding</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_concatenate.html">layer_concatenate</a>()</code> <br><code><a href="layer_conv_1d.html">layer_conv_1d</a>()</code> <br><code><a href="layer_conv_1d_transpose.html">layer_conv_1d_transpose</a>()</code> <br><code><a href="layer_conv_2d.html">layer_conv_2d</a>()</code> <br><code><a href="layer_conv_2d_transpose.html">layer_conv_2d_transpose</a>()</code> <br><code><a href="layer_conv_3d.html">layer_conv_3d</a>()</code> <br><code><a href="layer_conv_3d_transpose.html">layer_conv_3d_transpose</a>()</code> <br><code><a href="layer_conv_lstm_1d.html">layer_conv_lstm_1d</a>()</code> <br><code><a href="layer_conv_lstm_2d.html">layer_conv_lstm_2d</a>()</code> <br><code><a href="layer_conv_lstm_3d.html">layer_conv_lstm_3d</a>()</code> <br><code><a href="layer_cropping_1d.html">layer_cropping_1d</a>()</code> <br><code><a href="layer_cropping_2d.html">layer_cropping_2d</a>()</code> <br><code><a href="layer_cropping_3d.html">layer_cropping_3d</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_dense.html">layer_dense</a>()</code> <br><code><a href="layer_depthwise_conv_1d.html">layer_depthwise_conv_1d</a>()</code> <br><code><a href="layer_depthwise_conv_2d.html">layer_depthwise_conv_2d</a>()</code> <br><code><a href="layer_discretization.html">layer_discretization</a>()</code> <br><code><a href="layer_dot.html">layer_dot</a>()</code> <br><code><a href="layer_dropout.html">layer_dropout</a>()</code> <br><code><a href="layer_einsum_dense.html">layer_einsum_dense</a>()</code> <br><code><a href="layer_embedding.html">layer_embedding</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_feature_space.html">layer_feature_space</a>()</code> <br><code><a href="layer_flatten.html">layer_flatten</a>()</code> <br><code><a href="layer_flax_module_wrapper.html">layer_flax_module_wrapper</a>()</code> <br><code><a href="layer_gaussian_dropout.html">layer_gaussian_dropout</a>()</code> <br><code><a href="layer_gaussian_noise.html">layer_gaussian_noise</a>()</code> <br><code><a href="layer_global_average_pooling_1d.html">layer_global_average_pooling_1d</a>()</code> <br><code><a href="layer_global_average_pooling_2d.html">layer_global_average_pooling_2d</a>()</code> <br><code><a href="layer_global_average_pooling_3d.html">layer_global_average_pooling_3d</a>()</code> <br><code><a href="layer_global_max_pooling_1d.html">layer_global_max_pooling_1d</a>()</code> <br><code><a href="layer_global_max_pooling_2d.html">layer_global_max_pooling_2d</a>()</code> <br><code><a href="layer_global_max_pooling_3d.html">layer_global_max_pooling_3d</a>()</code> <br><code><a href="layer_group_normalization.html">layer_group_normalization</a>()</code> <br><code><a href="layer_group_query_attention.html">layer_group_query_attention</a>()</code> <br><code><a href="layer_gru.html">layer_gru</a>()</code> <br><code><a href="layer_hashed_crossing.html">layer_hashed_crossing</a>()</code> <br><code><a href="layer_hashing.html">layer_hashing</a>()</code> <br><code><a href="layer_identity.html">layer_identity</a>()</code> <br><code><a href="layer_integer_lookup.html">layer_integer_lookup</a>()</code> <br><code><a href="layer_jax_model_wrapper.html">layer_jax_model_wrapper</a>()</code> <br><code><a href="layer_lambda.html">layer_lambda</a>()</code> <br><code><a href="layer_layer_normalization.html">layer_layer_normalization</a>()</code> <br><code><a href="layer_lstm.html">layer_lstm</a>()</code> <br><code><a href="layer_masking.html">layer_masking</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_max_pooling_1d.html">layer_max_pooling_1d</a>()</code> <br><code><a href="layer_max_pooling_2d.html">layer_max_pooling_2d</a>()</code> <br><code><a href="layer_max_pooling_3d.html">layer_max_pooling_3d</a>()</code> <br><code><a href="layer_maximum.html">layer_maximum</a>()</code> <br><code><a href="layer_mel_spectrogram.html">layer_mel_spectrogram</a>()</code> <br><code><a href="layer_minimum.html">layer_minimum</a>()</code> <br><code><a href="layer_multi_head_attention.html">layer_multi_head_attention</a>()</code> <br><code><a href="layer_multiply.html">layer_multiply</a>()</code> <br><code><a href="layer_normalization.html">layer_normalization</a>()</code> <br><code><a href="layer_permute.html">layer_permute</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_brightness.html">layer_random_brightness</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_contrast.html">layer_random_contrast</a>()</code> <br><code><a href="layer_random_crop.html">layer_random_crop</a>()</code> <br><code><a href="layer_random_elastic_transform.html">layer_random_elastic_transform</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_flip.html">layer_random_flip</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_rotation.html">layer_random_rotation</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_random_translation.html">layer_random_translation</a>()</code> <br><code><a href="layer_random_zoom.html">layer_random_zoom</a>()</code> <br><code><a href="layer_repeat_vector.html">layer_repeat_vector</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_reshape.html">layer_reshape</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_rms_normalization.html">layer_rms_normalization</a>()</code> <br><code><a href="layer_rnn.html">layer_rnn</a>()</code> <br><code><a href="layer_separable_conv_1d.html">layer_separable_conv_1d</a>()</code> <br><code><a href="layer_separable_conv_2d.html">layer_separable_conv_2d</a>()</code> <br><code><a href="layer_simple_rnn.html">layer_simple_rnn</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br><code><a href="layer_spatial_dropout_1d.html">layer_spatial_dropout_1d</a>()</code> <br><code><a href="layer_spatial_dropout_2d.html">layer_spatial_dropout_2d</a>()</code> <br><code><a href="layer_spatial_dropout_3d.html">layer_spatial_dropout_3d</a>()</code> <br><code><a href="layer_spectral_normalization.html">layer_spectral_normalization</a>()</code> <br><code><a href="layer_stft_spectrogram.html">layer_stft_spectrogram</a>()</code> <br><code><a href="layer_string_lookup.html">layer_string_lookup</a>()</code> <br><code><a href="layer_subtract.html">layer_subtract</a>()</code> <br><code><a href="layer_text_vectorization.html">layer_text_vectorization</a>()</code> <br><code><a href="layer_tfsm.html">layer_tfsm</a>()</code> <br><code><a href="layer_time_distributed.html">layer_time_distributed</a>()</code> <br><code><a href="layer_torch_module_wrapper.html">layer_torch_module_wrapper</a>()</code> <br><code><a href="layer_unit_normalization.html">layer_unit_normalization</a>()</code> <br><code><a href="layer_upsampling_1d.html">layer_upsampling_1d</a>()</code> <br><code><a href="layer_upsampling_2d.html">layer_upsampling_2d</a>()</code> <br><code><a href="layer_upsampling_3d.html">layer_upsampling_3d</a>()</code> <br><code><a href="layer_zero_padding_1d.html">layer_zero_padding_1d</a>()</code> <br><code><a href="layer_zero_padding_2d.html">layer_zero_padding_2d</a>()</code> <br><code><a href="layer_zero_padding_3d.html">layer_zero_padding_3d</a>()</code> <br><code><a href="rnn_cell_gru.html">rnn_cell_gru</a>()</code> <br><code><a href="rnn_cell_lstm.html">rnn_cell_lstm</a>()</code> <br><code><a href="rnn_cell_simple.html">rnn_cell_simple</a>()</code> <br><code><a href="rnn_cells_stack.html">rnn_cells_stack</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -124,7 +124,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/layer_multi_head_attention.html b/docs/dev/reference/layer_multi_head_attention.html
index 9d9bba86d..cdc451d0a 100644
--- a/docs/dev/reference/layer_multi_head_attention.html
+++ b/docs/dev/reference/layer_multi_head_attention.html
@@ -40,7 +40,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -277,7 +277,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
     <div class="dont-index">
 <ul><li><p><a href="https://keras.io/api/layers/attention_layers/multi_head_attention#multiheadattention-class" class="external-link">https://keras.io/api/layers/attention_layers/multi_head_attention#multiheadattention-class</a></p></li>
 </ul><p>Other attention layers: <br><code><a href="layer_additive_attention.html">layer_additive_attention</a>()</code> <br><code><a href="layer_attention.html">layer_attention</a>()</code> <br><code><a href="layer_group_query_attention.html">layer_group_query_attention</a>()</code> <br></p>
-<p>Other layers: <br><code><a href="Layer.html">Layer</a>()</code> <br><code><a href="layer_activation.html">layer_activation</a>()</code> <br><code><a href="layer_activation_elu.html">layer_activation_elu</a>()</code> <br><code><a href="layer_activation_leaky_relu.html">layer_activation_leaky_relu</a>()</code> <br><code><a href="layer_activation_parametric_relu.html">layer_activation_parametric_relu</a>()</code> <br><code><a href="layer_activation_relu.html">layer_activation_relu</a>()</code> <br><code><a href="layer_activation_softmax.html">layer_activation_softmax</a>()</code> <br><code><a href="layer_activity_regularization.html">layer_activity_regularization</a>()</code> <br><code><a href="layer_add.html">layer_add</a>()</code> <br><code><a href="layer_additive_attention.html">layer_additive_attention</a>()</code> <br><code><a href="layer_alpha_dropout.html">layer_alpha_dropout</a>()</code> <br><code><a href="layer_attention.html">layer_attention</a>()</code> <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_average.html">layer_average</a>()</code> <br><code><a href="layer_average_pooling_1d.html">layer_average_pooling_1d</a>()</code> <br><code><a href="layer_average_pooling_2d.html">layer_average_pooling_2d</a>()</code> <br><code><a href="layer_average_pooling_3d.html">layer_average_pooling_3d</a>()</code> <br><code><a href="layer_batch_normalization.html">layer_batch_normalization</a>()</code> <br><code><a href="layer_bidirectional.html">layer_bidirectional</a>()</code> <br><code><a href="layer_category_encoding.html">layer_category_encoding</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_concatenate.html">layer_concatenate</a>()</code> <br><code><a href="layer_conv_1d.html">layer_conv_1d</a>()</code> <br><code><a href="layer_conv_1d_transpose.html">layer_conv_1d_transpose</a>()</code> <br><code><a href="layer_conv_2d.html">layer_conv_2d</a>()</code> <br><code><a href="layer_conv_2d_transpose.html">layer_conv_2d_transpose</a>()</code> <br><code><a href="layer_conv_3d.html">layer_conv_3d</a>()</code> <br><code><a href="layer_conv_3d_transpose.html">layer_conv_3d_transpose</a>()</code> <br><code><a href="layer_conv_lstm_1d.html">layer_conv_lstm_1d</a>()</code> <br><code><a href="layer_conv_lstm_2d.html">layer_conv_lstm_2d</a>()</code> <br><code><a href="layer_conv_lstm_3d.html">layer_conv_lstm_3d</a>()</code> <br><code><a href="layer_cropping_1d.html">layer_cropping_1d</a>()</code> <br><code><a href="layer_cropping_2d.html">layer_cropping_2d</a>()</code> <br><code><a href="layer_cropping_3d.html">layer_cropping_3d</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_dense.html">layer_dense</a>()</code> <br><code><a href="layer_depthwise_conv_1d.html">layer_depthwise_conv_1d</a>()</code> <br><code><a href="layer_depthwise_conv_2d.html">layer_depthwise_conv_2d</a>()</code> <br><code><a href="layer_discretization.html">layer_discretization</a>()</code> <br><code><a href="layer_dot.html">layer_dot</a>()</code> <br><code><a href="layer_dropout.html">layer_dropout</a>()</code> <br><code><a href="layer_einsum_dense.html">layer_einsum_dense</a>()</code> <br><code><a href="layer_embedding.html">layer_embedding</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_feature_space.html">layer_feature_space</a>()</code> <br><code><a href="layer_flatten.html">layer_flatten</a>()</code> <br><code><a href="layer_flax_module_wrapper.html">layer_flax_module_wrapper</a>()</code> <br><code><a href="layer_gaussian_dropout.html">layer_gaussian_dropout</a>()</code> <br><code><a href="layer_gaussian_noise.html">layer_gaussian_noise</a>()</code> <br><code><a href="layer_global_average_pooling_1d.html">layer_global_average_pooling_1d</a>()</code> <br><code><a href="layer_global_average_pooling_2d.html">layer_global_average_pooling_2d</a>()</code> <br><code><a href="layer_global_average_pooling_3d.html">layer_global_average_pooling_3d</a>()</code> <br><code><a href="layer_global_max_pooling_1d.html">layer_global_max_pooling_1d</a>()</code> <br><code><a href="layer_global_max_pooling_2d.html">layer_global_max_pooling_2d</a>()</code> <br><code><a href="layer_global_max_pooling_3d.html">layer_global_max_pooling_3d</a>()</code> <br><code><a href="layer_group_normalization.html">layer_group_normalization</a>()</code> <br><code><a href="layer_group_query_attention.html">layer_group_query_attention</a>()</code> <br><code><a href="layer_gru.html">layer_gru</a>()</code> <br><code><a href="layer_hashed_crossing.html">layer_hashed_crossing</a>()</code> <br><code><a href="layer_hashing.html">layer_hashing</a>()</code> <br><code><a href="layer_identity.html">layer_identity</a>()</code> <br><code><a href="layer_integer_lookup.html">layer_integer_lookup</a>()</code> <br><code><a href="layer_jax_model_wrapper.html">layer_jax_model_wrapper</a>()</code> <br><code><a href="layer_lambda.html">layer_lambda</a>()</code> <br><code><a href="layer_layer_normalization.html">layer_layer_normalization</a>()</code> <br><code><a href="layer_lstm.html">layer_lstm</a>()</code> <br><code><a href="layer_masking.html">layer_masking</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_max_pooling_1d.html">layer_max_pooling_1d</a>()</code> <br><code><a href="layer_max_pooling_2d.html">layer_max_pooling_2d</a>()</code> <br><code><a href="layer_max_pooling_3d.html">layer_max_pooling_3d</a>()</code> <br><code><a href="layer_maximum.html">layer_maximum</a>()</code> <br><code><a href="layer_mel_spectrogram.html">layer_mel_spectrogram</a>()</code> <br><code><a href="layer_minimum.html">layer_minimum</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_multiply.html">layer_multiply</a>()</code> <br><code><a href="layer_normalization.html">layer_normalization</a>()</code> <br><code><a href="layer_permute.html">layer_permute</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_brightness.html">layer_random_brightness</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_contrast.html">layer_random_contrast</a>()</code> <br><code><a href="layer_random_crop.html">layer_random_crop</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_flip.html">layer_random_flip</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_rotation.html">layer_random_rotation</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_random_translation.html">layer_random_translation</a>()</code> <br><code><a href="layer_random_zoom.html">layer_random_zoom</a>()</code> <br><code><a href="layer_repeat_vector.html">layer_repeat_vector</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_reshape.html">layer_reshape</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_rms_normalization.html">layer_rms_normalization</a>()</code> <br><code><a href="layer_rnn.html">layer_rnn</a>()</code> <br><code><a href="layer_separable_conv_1d.html">layer_separable_conv_1d</a>()</code> <br><code><a href="layer_separable_conv_2d.html">layer_separable_conv_2d</a>()</code> <br><code><a href="layer_simple_rnn.html">layer_simple_rnn</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br><code><a href="layer_spatial_dropout_1d.html">layer_spatial_dropout_1d</a>()</code> <br><code><a href="layer_spatial_dropout_2d.html">layer_spatial_dropout_2d</a>()</code> <br><code><a href="layer_spatial_dropout_3d.html">layer_spatial_dropout_3d</a>()</code> <br><code><a href="layer_spectral_normalization.html">layer_spectral_normalization</a>()</code> <br><code><a href="layer_stft_spectrogram.html">layer_stft_spectrogram</a>()</code> <br><code><a href="layer_string_lookup.html">layer_string_lookup</a>()</code> <br><code><a href="layer_subtract.html">layer_subtract</a>()</code> <br><code><a href="layer_text_vectorization.html">layer_text_vectorization</a>()</code> <br><code><a href="layer_tfsm.html">layer_tfsm</a>()</code> <br><code><a href="layer_time_distributed.html">layer_time_distributed</a>()</code> <br><code><a href="layer_torch_module_wrapper.html">layer_torch_module_wrapper</a>()</code> <br><code><a href="layer_unit_normalization.html">layer_unit_normalization</a>()</code> <br><code><a href="layer_upsampling_1d.html">layer_upsampling_1d</a>()</code> <br><code><a href="layer_upsampling_2d.html">layer_upsampling_2d</a>()</code> <br><code><a href="layer_upsampling_3d.html">layer_upsampling_3d</a>()</code> <br><code><a href="layer_zero_padding_1d.html">layer_zero_padding_1d</a>()</code> <br><code><a href="layer_zero_padding_2d.html">layer_zero_padding_2d</a>()</code> <br><code><a href="layer_zero_padding_3d.html">layer_zero_padding_3d</a>()</code> <br><code><a href="rnn_cell_gru.html">rnn_cell_gru</a>()</code> <br><code><a href="rnn_cell_lstm.html">rnn_cell_lstm</a>()</code> <br><code><a href="rnn_cell_simple.html">rnn_cell_simple</a>()</code> <br><code><a href="rnn_cells_stack.html">rnn_cells_stack</a>()</code> <br></p></div>
+<p>Other layers: <br><code><a href="Layer.html">Layer</a>()</code> <br><code><a href="layer_activation.html">layer_activation</a>()</code> <br><code><a href="layer_activation_elu.html">layer_activation_elu</a>()</code> <br><code><a href="layer_activation_leaky_relu.html">layer_activation_leaky_relu</a>()</code> <br><code><a href="layer_activation_parametric_relu.html">layer_activation_parametric_relu</a>()</code> <br><code><a href="layer_activation_relu.html">layer_activation_relu</a>()</code> <br><code><a href="layer_activation_softmax.html">layer_activation_softmax</a>()</code> <br><code><a href="layer_activity_regularization.html">layer_activity_regularization</a>()</code> <br><code><a href="layer_add.html">layer_add</a>()</code> <br><code><a href="layer_additive_attention.html">layer_additive_attention</a>()</code> <br><code><a href="layer_alpha_dropout.html">layer_alpha_dropout</a>()</code> <br><code><a href="layer_attention.html">layer_attention</a>()</code> <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_average.html">layer_average</a>()</code> <br><code><a href="layer_average_pooling_1d.html">layer_average_pooling_1d</a>()</code> <br><code><a href="layer_average_pooling_2d.html">layer_average_pooling_2d</a>()</code> <br><code><a href="layer_average_pooling_3d.html">layer_average_pooling_3d</a>()</code> <br><code><a href="layer_batch_normalization.html">layer_batch_normalization</a>()</code> <br><code><a href="layer_bidirectional.html">layer_bidirectional</a>()</code> <br><code><a href="layer_category_encoding.html">layer_category_encoding</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_concatenate.html">layer_concatenate</a>()</code> <br><code><a href="layer_conv_1d.html">layer_conv_1d</a>()</code> <br><code><a href="layer_conv_1d_transpose.html">layer_conv_1d_transpose</a>()</code> <br><code><a href="layer_conv_2d.html">layer_conv_2d</a>()</code> <br><code><a href="layer_conv_2d_transpose.html">layer_conv_2d_transpose</a>()</code> <br><code><a href="layer_conv_3d.html">layer_conv_3d</a>()</code> <br><code><a href="layer_conv_3d_transpose.html">layer_conv_3d_transpose</a>()</code> <br><code><a href="layer_conv_lstm_1d.html">layer_conv_lstm_1d</a>()</code> <br><code><a href="layer_conv_lstm_2d.html">layer_conv_lstm_2d</a>()</code> <br><code><a href="layer_conv_lstm_3d.html">layer_conv_lstm_3d</a>()</code> <br><code><a href="layer_cropping_1d.html">layer_cropping_1d</a>()</code> <br><code><a href="layer_cropping_2d.html">layer_cropping_2d</a>()</code> <br><code><a href="layer_cropping_3d.html">layer_cropping_3d</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_dense.html">layer_dense</a>()</code> <br><code><a href="layer_depthwise_conv_1d.html">layer_depthwise_conv_1d</a>()</code> <br><code><a href="layer_depthwise_conv_2d.html">layer_depthwise_conv_2d</a>()</code> <br><code><a href="layer_discretization.html">layer_discretization</a>()</code> <br><code><a href="layer_dot.html">layer_dot</a>()</code> <br><code><a href="layer_dropout.html">layer_dropout</a>()</code> <br><code><a href="layer_einsum_dense.html">layer_einsum_dense</a>()</code> <br><code><a href="layer_embedding.html">layer_embedding</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_feature_space.html">layer_feature_space</a>()</code> <br><code><a href="layer_flatten.html">layer_flatten</a>()</code> <br><code><a href="layer_flax_module_wrapper.html">layer_flax_module_wrapper</a>()</code> <br><code><a href="layer_gaussian_dropout.html">layer_gaussian_dropout</a>()</code> <br><code><a href="layer_gaussian_noise.html">layer_gaussian_noise</a>()</code> <br><code><a href="layer_global_average_pooling_1d.html">layer_global_average_pooling_1d</a>()</code> <br><code><a href="layer_global_average_pooling_2d.html">layer_global_average_pooling_2d</a>()</code> <br><code><a href="layer_global_average_pooling_3d.html">layer_global_average_pooling_3d</a>()</code> <br><code><a href="layer_global_max_pooling_1d.html">layer_global_max_pooling_1d</a>()</code> <br><code><a href="layer_global_max_pooling_2d.html">layer_global_max_pooling_2d</a>()</code> <br><code><a href="layer_global_max_pooling_3d.html">layer_global_max_pooling_3d</a>()</code> <br><code><a href="layer_group_normalization.html">layer_group_normalization</a>()</code> <br><code><a href="layer_group_query_attention.html">layer_group_query_attention</a>()</code> <br><code><a href="layer_gru.html">layer_gru</a>()</code> <br><code><a href="layer_hashed_crossing.html">layer_hashed_crossing</a>()</code> <br><code><a href="layer_hashing.html">layer_hashing</a>()</code> <br><code><a href="layer_identity.html">layer_identity</a>()</code> <br><code><a href="layer_integer_lookup.html">layer_integer_lookup</a>()</code> <br><code><a href="layer_jax_model_wrapper.html">layer_jax_model_wrapper</a>()</code> <br><code><a href="layer_lambda.html">layer_lambda</a>()</code> <br><code><a href="layer_layer_normalization.html">layer_layer_normalization</a>()</code> <br><code><a href="layer_lstm.html">layer_lstm</a>()</code> <br><code><a href="layer_masking.html">layer_masking</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_max_pooling_1d.html">layer_max_pooling_1d</a>()</code> <br><code><a href="layer_max_pooling_2d.html">layer_max_pooling_2d</a>()</code> <br><code><a href="layer_max_pooling_3d.html">layer_max_pooling_3d</a>()</code> <br><code><a href="layer_maximum.html">layer_maximum</a>()</code> <br><code><a href="layer_mel_spectrogram.html">layer_mel_spectrogram</a>()</code> <br><code><a href="layer_minimum.html">layer_minimum</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_multiply.html">layer_multiply</a>()</code> <br><code><a href="layer_normalization.html">layer_normalization</a>()</code> <br><code><a href="layer_permute.html">layer_permute</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_brightness.html">layer_random_brightness</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_contrast.html">layer_random_contrast</a>()</code> <br><code><a href="layer_random_crop.html">layer_random_crop</a>()</code> <br><code><a href="layer_random_elastic_transform.html">layer_random_elastic_transform</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_flip.html">layer_random_flip</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_rotation.html">layer_random_rotation</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_random_translation.html">layer_random_translation</a>()</code> <br><code><a href="layer_random_zoom.html">layer_random_zoom</a>()</code> <br><code><a href="layer_repeat_vector.html">layer_repeat_vector</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_reshape.html">layer_reshape</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_rms_normalization.html">layer_rms_normalization</a>()</code> <br><code><a href="layer_rnn.html">layer_rnn</a>()</code> <br><code><a href="layer_separable_conv_1d.html">layer_separable_conv_1d</a>()</code> <br><code><a href="layer_separable_conv_2d.html">layer_separable_conv_2d</a>()</code> <br><code><a href="layer_simple_rnn.html">layer_simple_rnn</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br><code><a href="layer_spatial_dropout_1d.html">layer_spatial_dropout_1d</a>()</code> <br><code><a href="layer_spatial_dropout_2d.html">layer_spatial_dropout_2d</a>()</code> <br><code><a href="layer_spatial_dropout_3d.html">layer_spatial_dropout_3d</a>()</code> <br><code><a href="layer_spectral_normalization.html">layer_spectral_normalization</a>()</code> <br><code><a href="layer_stft_spectrogram.html">layer_stft_spectrogram</a>()</code> <br><code><a href="layer_string_lookup.html">layer_string_lookup</a>()</code> <br><code><a href="layer_subtract.html">layer_subtract</a>()</code> <br><code><a href="layer_text_vectorization.html">layer_text_vectorization</a>()</code> <br><code><a href="layer_tfsm.html">layer_tfsm</a>()</code> <br><code><a href="layer_time_distributed.html">layer_time_distributed</a>()</code> <br><code><a href="layer_torch_module_wrapper.html">layer_torch_module_wrapper</a>()</code> <br><code><a href="layer_unit_normalization.html">layer_unit_normalization</a>()</code> <br><code><a href="layer_upsampling_1d.html">layer_upsampling_1d</a>()</code> <br><code><a href="layer_upsampling_2d.html">layer_upsampling_2d</a>()</code> <br><code><a href="layer_upsampling_3d.html">layer_upsampling_3d</a>()</code> <br><code><a href="layer_zero_padding_1d.html">layer_zero_padding_1d</a>()</code> <br><code><a href="layer_zero_padding_2d.html">layer_zero_padding_2d</a>()</code> <br><code><a href="layer_zero_padding_3d.html">layer_zero_padding_3d</a>()</code> <br><code><a href="rnn_cell_gru.html">rnn_cell_gru</a>()</code> <br><code><a href="rnn_cell_lstm.html">rnn_cell_lstm</a>()</code> <br><code><a href="rnn_cell_simple.html">rnn_cell_simple</a>()</code> <br><code><a href="rnn_cells_stack.html">rnn_cells_stack</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -289,7 +289,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/layer_multiply.html b/docs/dev/reference/layer_multiply.html
index fe2841631..bdeb69fe5 100644
--- a/docs/dev/reference/layer_multiply.html
+++ b/docs/dev/reference/layer_multiply.html
@@ -10,7 +10,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -107,7 +107,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
     <div class="dont-index">
 <ul><li><p><a href="https://keras.io/api/layers/merging_layers/multiply#multiply-class" class="external-link">https://keras.io/api/layers/merging_layers/multiply#multiply-class</a></p></li>
 </ul><p>Other merging layers: <br><code><a href="layer_add.html">layer_add</a>()</code> <br><code><a href="layer_average.html">layer_average</a>()</code> <br><code><a href="layer_concatenate.html">layer_concatenate</a>()</code> <br><code><a href="layer_dot.html">layer_dot</a>()</code> <br><code><a href="layer_maximum.html">layer_maximum</a>()</code> <br><code><a href="layer_minimum.html">layer_minimum</a>()</code> <br><code><a href="layer_subtract.html">layer_subtract</a>()</code> <br></p>
-<p>Other layers: <br><code><a href="Layer.html">Layer</a>()</code> <br><code><a href="layer_activation.html">layer_activation</a>()</code> <br><code><a href="layer_activation_elu.html">layer_activation_elu</a>()</code> <br><code><a href="layer_activation_leaky_relu.html">layer_activation_leaky_relu</a>()</code> <br><code><a href="layer_activation_parametric_relu.html">layer_activation_parametric_relu</a>()</code> <br><code><a href="layer_activation_relu.html">layer_activation_relu</a>()</code> <br><code><a href="layer_activation_softmax.html">layer_activation_softmax</a>()</code> <br><code><a href="layer_activity_regularization.html">layer_activity_regularization</a>()</code> <br><code><a href="layer_add.html">layer_add</a>()</code> <br><code><a href="layer_additive_attention.html">layer_additive_attention</a>()</code> <br><code><a href="layer_alpha_dropout.html">layer_alpha_dropout</a>()</code> <br><code><a href="layer_attention.html">layer_attention</a>()</code> <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_average.html">layer_average</a>()</code> <br><code><a href="layer_average_pooling_1d.html">layer_average_pooling_1d</a>()</code> <br><code><a href="layer_average_pooling_2d.html">layer_average_pooling_2d</a>()</code> <br><code><a href="layer_average_pooling_3d.html">layer_average_pooling_3d</a>()</code> <br><code><a href="layer_batch_normalization.html">layer_batch_normalization</a>()</code> <br><code><a href="layer_bidirectional.html">layer_bidirectional</a>()</code> <br><code><a href="layer_category_encoding.html">layer_category_encoding</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_concatenate.html">layer_concatenate</a>()</code> <br><code><a href="layer_conv_1d.html">layer_conv_1d</a>()</code> <br><code><a href="layer_conv_1d_transpose.html">layer_conv_1d_transpose</a>()</code> <br><code><a href="layer_conv_2d.html">layer_conv_2d</a>()</code> <br><code><a href="layer_conv_2d_transpose.html">layer_conv_2d_transpose</a>()</code> <br><code><a href="layer_conv_3d.html">layer_conv_3d</a>()</code> <br><code><a href="layer_conv_3d_transpose.html">layer_conv_3d_transpose</a>()</code> <br><code><a href="layer_conv_lstm_1d.html">layer_conv_lstm_1d</a>()</code> <br><code><a href="layer_conv_lstm_2d.html">layer_conv_lstm_2d</a>()</code> <br><code><a href="layer_conv_lstm_3d.html">layer_conv_lstm_3d</a>()</code> <br><code><a href="layer_cropping_1d.html">layer_cropping_1d</a>()</code> <br><code><a href="layer_cropping_2d.html">layer_cropping_2d</a>()</code> <br><code><a href="layer_cropping_3d.html">layer_cropping_3d</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_dense.html">layer_dense</a>()</code> <br><code><a href="layer_depthwise_conv_1d.html">layer_depthwise_conv_1d</a>()</code> <br><code><a href="layer_depthwise_conv_2d.html">layer_depthwise_conv_2d</a>()</code> <br><code><a href="layer_discretization.html">layer_discretization</a>()</code> <br><code><a href="layer_dot.html">layer_dot</a>()</code> <br><code><a href="layer_dropout.html">layer_dropout</a>()</code> <br><code><a href="layer_einsum_dense.html">layer_einsum_dense</a>()</code> <br><code><a href="layer_embedding.html">layer_embedding</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_feature_space.html">layer_feature_space</a>()</code> <br><code><a href="layer_flatten.html">layer_flatten</a>()</code> <br><code><a href="layer_flax_module_wrapper.html">layer_flax_module_wrapper</a>()</code> <br><code><a href="layer_gaussian_dropout.html">layer_gaussian_dropout</a>()</code> <br><code><a href="layer_gaussian_noise.html">layer_gaussian_noise</a>()</code> <br><code><a href="layer_global_average_pooling_1d.html">layer_global_average_pooling_1d</a>()</code> <br><code><a href="layer_global_average_pooling_2d.html">layer_global_average_pooling_2d</a>()</code> <br><code><a href="layer_global_average_pooling_3d.html">layer_global_average_pooling_3d</a>()</code> <br><code><a href="layer_global_max_pooling_1d.html">layer_global_max_pooling_1d</a>()</code> <br><code><a href="layer_global_max_pooling_2d.html">layer_global_max_pooling_2d</a>()</code> <br><code><a href="layer_global_max_pooling_3d.html">layer_global_max_pooling_3d</a>()</code> <br><code><a href="layer_group_normalization.html">layer_group_normalization</a>()</code> <br><code><a href="layer_group_query_attention.html">layer_group_query_attention</a>()</code> <br><code><a href="layer_gru.html">layer_gru</a>()</code> <br><code><a href="layer_hashed_crossing.html">layer_hashed_crossing</a>()</code> <br><code><a href="layer_hashing.html">layer_hashing</a>()</code> <br><code><a href="layer_identity.html">layer_identity</a>()</code> <br><code><a href="layer_integer_lookup.html">layer_integer_lookup</a>()</code> <br><code><a href="layer_jax_model_wrapper.html">layer_jax_model_wrapper</a>()</code> <br><code><a href="layer_lambda.html">layer_lambda</a>()</code> <br><code><a href="layer_layer_normalization.html">layer_layer_normalization</a>()</code> <br><code><a href="layer_lstm.html">layer_lstm</a>()</code> <br><code><a href="layer_masking.html">layer_masking</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_max_pooling_1d.html">layer_max_pooling_1d</a>()</code> <br><code><a href="layer_max_pooling_2d.html">layer_max_pooling_2d</a>()</code> <br><code><a href="layer_max_pooling_3d.html">layer_max_pooling_3d</a>()</code> <br><code><a href="layer_maximum.html">layer_maximum</a>()</code> <br><code><a href="layer_mel_spectrogram.html">layer_mel_spectrogram</a>()</code> <br><code><a href="layer_minimum.html">layer_minimum</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_multi_head_attention.html">layer_multi_head_attention</a>()</code> <br><code><a href="layer_normalization.html">layer_normalization</a>()</code> <br><code><a href="layer_permute.html">layer_permute</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_brightness.html">layer_random_brightness</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_contrast.html">layer_random_contrast</a>()</code> <br><code><a href="layer_random_crop.html">layer_random_crop</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_flip.html">layer_random_flip</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_rotation.html">layer_random_rotation</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_random_translation.html">layer_random_translation</a>()</code> <br><code><a href="layer_random_zoom.html">layer_random_zoom</a>()</code> <br><code><a href="layer_repeat_vector.html">layer_repeat_vector</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_reshape.html">layer_reshape</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_rms_normalization.html">layer_rms_normalization</a>()</code> <br><code><a href="layer_rnn.html">layer_rnn</a>()</code> <br><code><a href="layer_separable_conv_1d.html">layer_separable_conv_1d</a>()</code> <br><code><a href="layer_separable_conv_2d.html">layer_separable_conv_2d</a>()</code> <br><code><a href="layer_simple_rnn.html">layer_simple_rnn</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br><code><a href="layer_spatial_dropout_1d.html">layer_spatial_dropout_1d</a>()</code> <br><code><a href="layer_spatial_dropout_2d.html">layer_spatial_dropout_2d</a>()</code> <br><code><a href="layer_spatial_dropout_3d.html">layer_spatial_dropout_3d</a>()</code> <br><code><a href="layer_spectral_normalization.html">layer_spectral_normalization</a>()</code> <br><code><a href="layer_stft_spectrogram.html">layer_stft_spectrogram</a>()</code> <br><code><a href="layer_string_lookup.html">layer_string_lookup</a>()</code> <br><code><a href="layer_subtract.html">layer_subtract</a>()</code> <br><code><a href="layer_text_vectorization.html">layer_text_vectorization</a>()</code> <br><code><a href="layer_tfsm.html">layer_tfsm</a>()</code> <br><code><a href="layer_time_distributed.html">layer_time_distributed</a>()</code> <br><code><a href="layer_torch_module_wrapper.html">layer_torch_module_wrapper</a>()</code> <br><code><a href="layer_unit_normalization.html">layer_unit_normalization</a>()</code> <br><code><a href="layer_upsampling_1d.html">layer_upsampling_1d</a>()</code> <br><code><a href="layer_upsampling_2d.html">layer_upsampling_2d</a>()</code> <br><code><a href="layer_upsampling_3d.html">layer_upsampling_3d</a>()</code> <br><code><a href="layer_zero_padding_1d.html">layer_zero_padding_1d</a>()</code> <br><code><a href="layer_zero_padding_2d.html">layer_zero_padding_2d</a>()</code> <br><code><a href="layer_zero_padding_3d.html">layer_zero_padding_3d</a>()</code> <br><code><a href="rnn_cell_gru.html">rnn_cell_gru</a>()</code> <br><code><a href="rnn_cell_lstm.html">rnn_cell_lstm</a>()</code> <br><code><a href="rnn_cell_simple.html">rnn_cell_simple</a>()</code> <br><code><a href="rnn_cells_stack.html">rnn_cells_stack</a>()</code> <br></p></div>
+<p>Other layers: <br><code><a href="Layer.html">Layer</a>()</code> <br><code><a href="layer_activation.html">layer_activation</a>()</code> <br><code><a href="layer_activation_elu.html">layer_activation_elu</a>()</code> <br><code><a href="layer_activation_leaky_relu.html">layer_activation_leaky_relu</a>()</code> <br><code><a href="layer_activation_parametric_relu.html">layer_activation_parametric_relu</a>()</code> <br><code><a href="layer_activation_relu.html">layer_activation_relu</a>()</code> <br><code><a href="layer_activation_softmax.html">layer_activation_softmax</a>()</code> <br><code><a href="layer_activity_regularization.html">layer_activity_regularization</a>()</code> <br><code><a href="layer_add.html">layer_add</a>()</code> <br><code><a href="layer_additive_attention.html">layer_additive_attention</a>()</code> <br><code><a href="layer_alpha_dropout.html">layer_alpha_dropout</a>()</code> <br><code><a href="layer_attention.html">layer_attention</a>()</code> <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_average.html">layer_average</a>()</code> <br><code><a href="layer_average_pooling_1d.html">layer_average_pooling_1d</a>()</code> <br><code><a href="layer_average_pooling_2d.html">layer_average_pooling_2d</a>()</code> <br><code><a href="layer_average_pooling_3d.html">layer_average_pooling_3d</a>()</code> <br><code><a href="layer_batch_normalization.html">layer_batch_normalization</a>()</code> <br><code><a href="layer_bidirectional.html">layer_bidirectional</a>()</code> <br><code><a href="layer_category_encoding.html">layer_category_encoding</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_concatenate.html">layer_concatenate</a>()</code> <br><code><a href="layer_conv_1d.html">layer_conv_1d</a>()</code> <br><code><a href="layer_conv_1d_transpose.html">layer_conv_1d_transpose</a>()</code> <br><code><a href="layer_conv_2d.html">layer_conv_2d</a>()</code> <br><code><a href="layer_conv_2d_transpose.html">layer_conv_2d_transpose</a>()</code> <br><code><a href="layer_conv_3d.html">layer_conv_3d</a>()</code> <br><code><a href="layer_conv_3d_transpose.html">layer_conv_3d_transpose</a>()</code> <br><code><a href="layer_conv_lstm_1d.html">layer_conv_lstm_1d</a>()</code> <br><code><a href="layer_conv_lstm_2d.html">layer_conv_lstm_2d</a>()</code> <br><code><a href="layer_conv_lstm_3d.html">layer_conv_lstm_3d</a>()</code> <br><code><a href="layer_cropping_1d.html">layer_cropping_1d</a>()</code> <br><code><a href="layer_cropping_2d.html">layer_cropping_2d</a>()</code> <br><code><a href="layer_cropping_3d.html">layer_cropping_3d</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_dense.html">layer_dense</a>()</code> <br><code><a href="layer_depthwise_conv_1d.html">layer_depthwise_conv_1d</a>()</code> <br><code><a href="layer_depthwise_conv_2d.html">layer_depthwise_conv_2d</a>()</code> <br><code><a href="layer_discretization.html">layer_discretization</a>()</code> <br><code><a href="layer_dot.html">layer_dot</a>()</code> <br><code><a href="layer_dropout.html">layer_dropout</a>()</code> <br><code><a href="layer_einsum_dense.html">layer_einsum_dense</a>()</code> <br><code><a href="layer_embedding.html">layer_embedding</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_feature_space.html">layer_feature_space</a>()</code> <br><code><a href="layer_flatten.html">layer_flatten</a>()</code> <br><code><a href="layer_flax_module_wrapper.html">layer_flax_module_wrapper</a>()</code> <br><code><a href="layer_gaussian_dropout.html">layer_gaussian_dropout</a>()</code> <br><code><a href="layer_gaussian_noise.html">layer_gaussian_noise</a>()</code> <br><code><a href="layer_global_average_pooling_1d.html">layer_global_average_pooling_1d</a>()</code> <br><code><a href="layer_global_average_pooling_2d.html">layer_global_average_pooling_2d</a>()</code> <br><code><a href="layer_global_average_pooling_3d.html">layer_global_average_pooling_3d</a>()</code> <br><code><a href="layer_global_max_pooling_1d.html">layer_global_max_pooling_1d</a>()</code> <br><code><a href="layer_global_max_pooling_2d.html">layer_global_max_pooling_2d</a>()</code> <br><code><a href="layer_global_max_pooling_3d.html">layer_global_max_pooling_3d</a>()</code> <br><code><a href="layer_group_normalization.html">layer_group_normalization</a>()</code> <br><code><a href="layer_group_query_attention.html">layer_group_query_attention</a>()</code> <br><code><a href="layer_gru.html">layer_gru</a>()</code> <br><code><a href="layer_hashed_crossing.html">layer_hashed_crossing</a>()</code> <br><code><a href="layer_hashing.html">layer_hashing</a>()</code> <br><code><a href="layer_identity.html">layer_identity</a>()</code> <br><code><a href="layer_integer_lookup.html">layer_integer_lookup</a>()</code> <br><code><a href="layer_jax_model_wrapper.html">layer_jax_model_wrapper</a>()</code> <br><code><a href="layer_lambda.html">layer_lambda</a>()</code> <br><code><a href="layer_layer_normalization.html">layer_layer_normalization</a>()</code> <br><code><a href="layer_lstm.html">layer_lstm</a>()</code> <br><code><a href="layer_masking.html">layer_masking</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_max_pooling_1d.html">layer_max_pooling_1d</a>()</code> <br><code><a href="layer_max_pooling_2d.html">layer_max_pooling_2d</a>()</code> <br><code><a href="layer_max_pooling_3d.html">layer_max_pooling_3d</a>()</code> <br><code><a href="layer_maximum.html">layer_maximum</a>()</code> <br><code><a href="layer_mel_spectrogram.html">layer_mel_spectrogram</a>()</code> <br><code><a href="layer_minimum.html">layer_minimum</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_multi_head_attention.html">layer_multi_head_attention</a>()</code> <br><code><a href="layer_normalization.html">layer_normalization</a>()</code> <br><code><a href="layer_permute.html">layer_permute</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_brightness.html">layer_random_brightness</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_contrast.html">layer_random_contrast</a>()</code> <br><code><a href="layer_random_crop.html">layer_random_crop</a>()</code> <br><code><a href="layer_random_elastic_transform.html">layer_random_elastic_transform</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_flip.html">layer_random_flip</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_rotation.html">layer_random_rotation</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_random_translation.html">layer_random_translation</a>()</code> <br><code><a href="layer_random_zoom.html">layer_random_zoom</a>()</code> <br><code><a href="layer_repeat_vector.html">layer_repeat_vector</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_reshape.html">layer_reshape</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_rms_normalization.html">layer_rms_normalization</a>()</code> <br><code><a href="layer_rnn.html">layer_rnn</a>()</code> <br><code><a href="layer_separable_conv_1d.html">layer_separable_conv_1d</a>()</code> <br><code><a href="layer_separable_conv_2d.html">layer_separable_conv_2d</a>()</code> <br><code><a href="layer_simple_rnn.html">layer_simple_rnn</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br><code><a href="layer_spatial_dropout_1d.html">layer_spatial_dropout_1d</a>()</code> <br><code><a href="layer_spatial_dropout_2d.html">layer_spatial_dropout_2d</a>()</code> <br><code><a href="layer_spatial_dropout_3d.html">layer_spatial_dropout_3d</a>()</code> <br><code><a href="layer_spectral_normalization.html">layer_spectral_normalization</a>()</code> <br><code><a href="layer_stft_spectrogram.html">layer_stft_spectrogram</a>()</code> <br><code><a href="layer_string_lookup.html">layer_string_lookup</a>()</code> <br><code><a href="layer_subtract.html">layer_subtract</a>()</code> <br><code><a href="layer_text_vectorization.html">layer_text_vectorization</a>()</code> <br><code><a href="layer_tfsm.html">layer_tfsm</a>()</code> <br><code><a href="layer_time_distributed.html">layer_time_distributed</a>()</code> <br><code><a href="layer_torch_module_wrapper.html">layer_torch_module_wrapper</a>()</code> <br><code><a href="layer_unit_normalization.html">layer_unit_normalization</a>()</code> <br><code><a href="layer_upsampling_1d.html">layer_upsampling_1d</a>()</code> <br><code><a href="layer_upsampling_2d.html">layer_upsampling_2d</a>()</code> <br><code><a href="layer_upsampling_3d.html">layer_upsampling_3d</a>()</code> <br><code><a href="layer_zero_padding_1d.html">layer_zero_padding_1d</a>()</code> <br><code><a href="layer_zero_padding_2d.html">layer_zero_padding_2d</a>()</code> <br><code><a href="layer_zero_padding_3d.html">layer_zero_padding_3d</a>()</code> <br><code><a href="rnn_cell_gru.html">rnn_cell_gru</a>()</code> <br><code><a href="rnn_cell_lstm.html">rnn_cell_lstm</a>()</code> <br><code><a href="rnn_cell_simple.html">rnn_cell_simple</a>()</code> <br><code><a href="rnn_cells_stack.html">rnn_cells_stack</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -119,7 +119,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/layer_normalization.html b/docs/dev/reference/layer_normalization.html
index f1a9454cf..7c3b98616 100644
--- a/docs/dev/reference/layer_normalization.html
+++ b/docs/dev/reference/layer_normalization.html
@@ -22,7 +22,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -193,8 +193,8 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
     <div class="dont-index">
 <ul><li><p><a href="https://keras.io/api/layers/preprocessing_layers/numerical/normalization#normalization-class" class="external-link">https://keras.io/api/layers/preprocessing_layers/numerical/normalization#normalization-class</a></p></li>
 </ul><p>Other numerical features preprocessing layers: <br><code><a href="layer_discretization.html">layer_discretization</a>()</code> <br></p>
-<p>Other preprocessing layers: <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_category_encoding.html">layer_category_encoding</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_discretization.html">layer_discretization</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_feature_space.html">layer_feature_space</a>()</code> <br><code><a href="layer_hashed_crossing.html">layer_hashed_crossing</a>()</code> <br><code><a href="layer_hashing.html">layer_hashing</a>()</code> <br><code><a href="layer_integer_lookup.html">layer_integer_lookup</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_mel_spectrogram.html">layer_mel_spectrogram</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_brightness.html">layer_random_brightness</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_contrast.html">layer_random_contrast</a>()</code> <br><code><a href="layer_random_crop.html">layer_random_crop</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_flip.html">layer_random_flip</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_rotation.html">layer_random_rotation</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_random_translation.html">layer_random_translation</a>()</code> <br><code><a href="layer_random_zoom.html">layer_random_zoom</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br><code><a href="layer_stft_spectrogram.html">layer_stft_spectrogram</a>()</code> <br><code><a href="layer_string_lookup.html">layer_string_lookup</a>()</code> <br><code><a href="layer_text_vectorization.html">layer_text_vectorization</a>()</code> <br></p>
-<p>Other layers: <br><code><a href="Layer.html">Layer</a>()</code> <br><code><a href="layer_activation.html">layer_activation</a>()</code> <br><code><a href="layer_activation_elu.html">layer_activation_elu</a>()</code> <br><code><a href="layer_activation_leaky_relu.html">layer_activation_leaky_relu</a>()</code> <br><code><a href="layer_activation_parametric_relu.html">layer_activation_parametric_relu</a>()</code> <br><code><a href="layer_activation_relu.html">layer_activation_relu</a>()</code> <br><code><a href="layer_activation_softmax.html">layer_activation_softmax</a>()</code> <br><code><a href="layer_activity_regularization.html">layer_activity_regularization</a>()</code> <br><code><a href="layer_add.html">layer_add</a>()</code> <br><code><a href="layer_additive_attention.html">layer_additive_attention</a>()</code> <br><code><a href="layer_alpha_dropout.html">layer_alpha_dropout</a>()</code> <br><code><a href="layer_attention.html">layer_attention</a>()</code> <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_average.html">layer_average</a>()</code> <br><code><a href="layer_average_pooling_1d.html">layer_average_pooling_1d</a>()</code> <br><code><a href="layer_average_pooling_2d.html">layer_average_pooling_2d</a>()</code> <br><code><a href="layer_average_pooling_3d.html">layer_average_pooling_3d</a>()</code> <br><code><a href="layer_batch_normalization.html">layer_batch_normalization</a>()</code> <br><code><a href="layer_bidirectional.html">layer_bidirectional</a>()</code> <br><code><a href="layer_category_encoding.html">layer_category_encoding</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_concatenate.html">layer_concatenate</a>()</code> <br><code><a href="layer_conv_1d.html">layer_conv_1d</a>()</code> <br><code><a href="layer_conv_1d_transpose.html">layer_conv_1d_transpose</a>()</code> <br><code><a href="layer_conv_2d.html">layer_conv_2d</a>()</code> <br><code><a href="layer_conv_2d_transpose.html">layer_conv_2d_transpose</a>()</code> <br><code><a href="layer_conv_3d.html">layer_conv_3d</a>()</code> <br><code><a href="layer_conv_3d_transpose.html">layer_conv_3d_transpose</a>()</code> <br><code><a href="layer_conv_lstm_1d.html">layer_conv_lstm_1d</a>()</code> <br><code><a href="layer_conv_lstm_2d.html">layer_conv_lstm_2d</a>()</code> <br><code><a href="layer_conv_lstm_3d.html">layer_conv_lstm_3d</a>()</code> <br><code><a href="layer_cropping_1d.html">layer_cropping_1d</a>()</code> <br><code><a href="layer_cropping_2d.html">layer_cropping_2d</a>()</code> <br><code><a href="layer_cropping_3d.html">layer_cropping_3d</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_dense.html">layer_dense</a>()</code> <br><code><a href="layer_depthwise_conv_1d.html">layer_depthwise_conv_1d</a>()</code> <br><code><a href="layer_depthwise_conv_2d.html">layer_depthwise_conv_2d</a>()</code> <br><code><a href="layer_discretization.html">layer_discretization</a>()</code> <br><code><a href="layer_dot.html">layer_dot</a>()</code> <br><code><a href="layer_dropout.html">layer_dropout</a>()</code> <br><code><a href="layer_einsum_dense.html">layer_einsum_dense</a>()</code> <br><code><a href="layer_embedding.html">layer_embedding</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_feature_space.html">layer_feature_space</a>()</code> <br><code><a href="layer_flatten.html">layer_flatten</a>()</code> <br><code><a href="layer_flax_module_wrapper.html">layer_flax_module_wrapper</a>()</code> <br><code><a href="layer_gaussian_dropout.html">layer_gaussian_dropout</a>()</code> <br><code><a href="layer_gaussian_noise.html">layer_gaussian_noise</a>()</code> <br><code><a href="layer_global_average_pooling_1d.html">layer_global_average_pooling_1d</a>()</code> <br><code><a href="layer_global_average_pooling_2d.html">layer_global_average_pooling_2d</a>()</code> <br><code><a href="layer_global_average_pooling_3d.html">layer_global_average_pooling_3d</a>()</code> <br><code><a href="layer_global_max_pooling_1d.html">layer_global_max_pooling_1d</a>()</code> <br><code><a href="layer_global_max_pooling_2d.html">layer_global_max_pooling_2d</a>()</code> <br><code><a href="layer_global_max_pooling_3d.html">layer_global_max_pooling_3d</a>()</code> <br><code><a href="layer_group_normalization.html">layer_group_normalization</a>()</code> <br><code><a href="layer_group_query_attention.html">layer_group_query_attention</a>()</code> <br><code><a href="layer_gru.html">layer_gru</a>()</code> <br><code><a href="layer_hashed_crossing.html">layer_hashed_crossing</a>()</code> <br><code><a href="layer_hashing.html">layer_hashing</a>()</code> <br><code><a href="layer_identity.html">layer_identity</a>()</code> <br><code><a href="layer_integer_lookup.html">layer_integer_lookup</a>()</code> <br><code><a href="layer_jax_model_wrapper.html">layer_jax_model_wrapper</a>()</code> <br><code><a href="layer_lambda.html">layer_lambda</a>()</code> <br><code><a href="layer_layer_normalization.html">layer_layer_normalization</a>()</code> <br><code><a href="layer_lstm.html">layer_lstm</a>()</code> <br><code><a href="layer_masking.html">layer_masking</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_max_pooling_1d.html">layer_max_pooling_1d</a>()</code> <br><code><a href="layer_max_pooling_2d.html">layer_max_pooling_2d</a>()</code> <br><code><a href="layer_max_pooling_3d.html">layer_max_pooling_3d</a>()</code> <br><code><a href="layer_maximum.html">layer_maximum</a>()</code> <br><code><a href="layer_mel_spectrogram.html">layer_mel_spectrogram</a>()</code> <br><code><a href="layer_minimum.html">layer_minimum</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_multi_head_attention.html">layer_multi_head_attention</a>()</code> <br><code><a href="layer_multiply.html">layer_multiply</a>()</code> <br><code><a href="layer_permute.html">layer_permute</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_brightness.html">layer_random_brightness</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_contrast.html">layer_random_contrast</a>()</code> <br><code><a href="layer_random_crop.html">layer_random_crop</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_flip.html">layer_random_flip</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_rotation.html">layer_random_rotation</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_random_translation.html">layer_random_translation</a>()</code> <br><code><a href="layer_random_zoom.html">layer_random_zoom</a>()</code> <br><code><a href="layer_repeat_vector.html">layer_repeat_vector</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_reshape.html">layer_reshape</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_rms_normalization.html">layer_rms_normalization</a>()</code> <br><code><a href="layer_rnn.html">layer_rnn</a>()</code> <br><code><a href="layer_separable_conv_1d.html">layer_separable_conv_1d</a>()</code> <br><code><a href="layer_separable_conv_2d.html">layer_separable_conv_2d</a>()</code> <br><code><a href="layer_simple_rnn.html">layer_simple_rnn</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br><code><a href="layer_spatial_dropout_1d.html">layer_spatial_dropout_1d</a>()</code> <br><code><a href="layer_spatial_dropout_2d.html">layer_spatial_dropout_2d</a>()</code> <br><code><a href="layer_spatial_dropout_3d.html">layer_spatial_dropout_3d</a>()</code> <br><code><a href="layer_spectral_normalization.html">layer_spectral_normalization</a>()</code> <br><code><a href="layer_stft_spectrogram.html">layer_stft_spectrogram</a>()</code> <br><code><a href="layer_string_lookup.html">layer_string_lookup</a>()</code> <br><code><a href="layer_subtract.html">layer_subtract</a>()</code> <br><code><a href="layer_text_vectorization.html">layer_text_vectorization</a>()</code> <br><code><a href="layer_tfsm.html">layer_tfsm</a>()</code> <br><code><a href="layer_time_distributed.html">layer_time_distributed</a>()</code> <br><code><a href="layer_torch_module_wrapper.html">layer_torch_module_wrapper</a>()</code> <br><code><a href="layer_unit_normalization.html">layer_unit_normalization</a>()</code> <br><code><a href="layer_upsampling_1d.html">layer_upsampling_1d</a>()</code> <br><code><a href="layer_upsampling_2d.html">layer_upsampling_2d</a>()</code> <br><code><a href="layer_upsampling_3d.html">layer_upsampling_3d</a>()</code> <br><code><a href="layer_zero_padding_1d.html">layer_zero_padding_1d</a>()</code> <br><code><a href="layer_zero_padding_2d.html">layer_zero_padding_2d</a>()</code> <br><code><a href="layer_zero_padding_3d.html">layer_zero_padding_3d</a>()</code> <br><code><a href="rnn_cell_gru.html">rnn_cell_gru</a>()</code> <br><code><a href="rnn_cell_lstm.html">rnn_cell_lstm</a>()</code> <br><code><a href="rnn_cell_simple.html">rnn_cell_simple</a>()</code> <br><code><a href="rnn_cells_stack.html">rnn_cells_stack</a>()</code> <br></p></div>
+<p>Other preprocessing layers: <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_category_encoding.html">layer_category_encoding</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_discretization.html">layer_discretization</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_feature_space.html">layer_feature_space</a>()</code> <br><code><a href="layer_hashed_crossing.html">layer_hashed_crossing</a>()</code> <br><code><a href="layer_hashing.html">layer_hashing</a>()</code> <br><code><a href="layer_integer_lookup.html">layer_integer_lookup</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_mel_spectrogram.html">layer_mel_spectrogram</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_brightness.html">layer_random_brightness</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_contrast.html">layer_random_contrast</a>()</code> <br><code><a href="layer_random_crop.html">layer_random_crop</a>()</code> <br><code><a href="layer_random_elastic_transform.html">layer_random_elastic_transform</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_flip.html">layer_random_flip</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_rotation.html">layer_random_rotation</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_random_translation.html">layer_random_translation</a>()</code> <br><code><a href="layer_random_zoom.html">layer_random_zoom</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br><code><a href="layer_stft_spectrogram.html">layer_stft_spectrogram</a>()</code> <br><code><a href="layer_string_lookup.html">layer_string_lookup</a>()</code> <br><code><a href="layer_text_vectorization.html">layer_text_vectorization</a>()</code> <br></p>
+<p>Other layers: <br><code><a href="Layer.html">Layer</a>()</code> <br><code><a href="layer_activation.html">layer_activation</a>()</code> <br><code><a href="layer_activation_elu.html">layer_activation_elu</a>()</code> <br><code><a href="layer_activation_leaky_relu.html">layer_activation_leaky_relu</a>()</code> <br><code><a href="layer_activation_parametric_relu.html">layer_activation_parametric_relu</a>()</code> <br><code><a href="layer_activation_relu.html">layer_activation_relu</a>()</code> <br><code><a href="layer_activation_softmax.html">layer_activation_softmax</a>()</code> <br><code><a href="layer_activity_regularization.html">layer_activity_regularization</a>()</code> <br><code><a href="layer_add.html">layer_add</a>()</code> <br><code><a href="layer_additive_attention.html">layer_additive_attention</a>()</code> <br><code><a href="layer_alpha_dropout.html">layer_alpha_dropout</a>()</code> <br><code><a href="layer_attention.html">layer_attention</a>()</code> <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_average.html">layer_average</a>()</code> <br><code><a href="layer_average_pooling_1d.html">layer_average_pooling_1d</a>()</code> <br><code><a href="layer_average_pooling_2d.html">layer_average_pooling_2d</a>()</code> <br><code><a href="layer_average_pooling_3d.html">layer_average_pooling_3d</a>()</code> <br><code><a href="layer_batch_normalization.html">layer_batch_normalization</a>()</code> <br><code><a href="layer_bidirectional.html">layer_bidirectional</a>()</code> <br><code><a href="layer_category_encoding.html">layer_category_encoding</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_concatenate.html">layer_concatenate</a>()</code> <br><code><a href="layer_conv_1d.html">layer_conv_1d</a>()</code> <br><code><a href="layer_conv_1d_transpose.html">layer_conv_1d_transpose</a>()</code> <br><code><a href="layer_conv_2d.html">layer_conv_2d</a>()</code> <br><code><a href="layer_conv_2d_transpose.html">layer_conv_2d_transpose</a>()</code> <br><code><a href="layer_conv_3d.html">layer_conv_3d</a>()</code> <br><code><a href="layer_conv_3d_transpose.html">layer_conv_3d_transpose</a>()</code> <br><code><a href="layer_conv_lstm_1d.html">layer_conv_lstm_1d</a>()</code> <br><code><a href="layer_conv_lstm_2d.html">layer_conv_lstm_2d</a>()</code> <br><code><a href="layer_conv_lstm_3d.html">layer_conv_lstm_3d</a>()</code> <br><code><a href="layer_cropping_1d.html">layer_cropping_1d</a>()</code> <br><code><a href="layer_cropping_2d.html">layer_cropping_2d</a>()</code> <br><code><a href="layer_cropping_3d.html">layer_cropping_3d</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_dense.html">layer_dense</a>()</code> <br><code><a href="layer_depthwise_conv_1d.html">layer_depthwise_conv_1d</a>()</code> <br><code><a href="layer_depthwise_conv_2d.html">layer_depthwise_conv_2d</a>()</code> <br><code><a href="layer_discretization.html">layer_discretization</a>()</code> <br><code><a href="layer_dot.html">layer_dot</a>()</code> <br><code><a href="layer_dropout.html">layer_dropout</a>()</code> <br><code><a href="layer_einsum_dense.html">layer_einsum_dense</a>()</code> <br><code><a href="layer_embedding.html">layer_embedding</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_feature_space.html">layer_feature_space</a>()</code> <br><code><a href="layer_flatten.html">layer_flatten</a>()</code> <br><code><a href="layer_flax_module_wrapper.html">layer_flax_module_wrapper</a>()</code> <br><code><a href="layer_gaussian_dropout.html">layer_gaussian_dropout</a>()</code> <br><code><a href="layer_gaussian_noise.html">layer_gaussian_noise</a>()</code> <br><code><a href="layer_global_average_pooling_1d.html">layer_global_average_pooling_1d</a>()</code> <br><code><a href="layer_global_average_pooling_2d.html">layer_global_average_pooling_2d</a>()</code> <br><code><a href="layer_global_average_pooling_3d.html">layer_global_average_pooling_3d</a>()</code> <br><code><a href="layer_global_max_pooling_1d.html">layer_global_max_pooling_1d</a>()</code> <br><code><a href="layer_global_max_pooling_2d.html">layer_global_max_pooling_2d</a>()</code> <br><code><a href="layer_global_max_pooling_3d.html">layer_global_max_pooling_3d</a>()</code> <br><code><a href="layer_group_normalization.html">layer_group_normalization</a>()</code> <br><code><a href="layer_group_query_attention.html">layer_group_query_attention</a>()</code> <br><code><a href="layer_gru.html">layer_gru</a>()</code> <br><code><a href="layer_hashed_crossing.html">layer_hashed_crossing</a>()</code> <br><code><a href="layer_hashing.html">layer_hashing</a>()</code> <br><code><a href="layer_identity.html">layer_identity</a>()</code> <br><code><a href="layer_integer_lookup.html">layer_integer_lookup</a>()</code> <br><code><a href="layer_jax_model_wrapper.html">layer_jax_model_wrapper</a>()</code> <br><code><a href="layer_lambda.html">layer_lambda</a>()</code> <br><code><a href="layer_layer_normalization.html">layer_layer_normalization</a>()</code> <br><code><a href="layer_lstm.html">layer_lstm</a>()</code> <br><code><a href="layer_masking.html">layer_masking</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_max_pooling_1d.html">layer_max_pooling_1d</a>()</code> <br><code><a href="layer_max_pooling_2d.html">layer_max_pooling_2d</a>()</code> <br><code><a href="layer_max_pooling_3d.html">layer_max_pooling_3d</a>()</code> <br><code><a href="layer_maximum.html">layer_maximum</a>()</code> <br><code><a href="layer_mel_spectrogram.html">layer_mel_spectrogram</a>()</code> <br><code><a href="layer_minimum.html">layer_minimum</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_multi_head_attention.html">layer_multi_head_attention</a>()</code> <br><code><a href="layer_multiply.html">layer_multiply</a>()</code> <br><code><a href="layer_permute.html">layer_permute</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_brightness.html">layer_random_brightness</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_contrast.html">layer_random_contrast</a>()</code> <br><code><a href="layer_random_crop.html">layer_random_crop</a>()</code> <br><code><a href="layer_random_elastic_transform.html">layer_random_elastic_transform</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_flip.html">layer_random_flip</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_rotation.html">layer_random_rotation</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_random_translation.html">layer_random_translation</a>()</code> <br><code><a href="layer_random_zoom.html">layer_random_zoom</a>()</code> <br><code><a href="layer_repeat_vector.html">layer_repeat_vector</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_reshape.html">layer_reshape</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_rms_normalization.html">layer_rms_normalization</a>()</code> <br><code><a href="layer_rnn.html">layer_rnn</a>()</code> <br><code><a href="layer_separable_conv_1d.html">layer_separable_conv_1d</a>()</code> <br><code><a href="layer_separable_conv_2d.html">layer_separable_conv_2d</a>()</code> <br><code><a href="layer_simple_rnn.html">layer_simple_rnn</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br><code><a href="layer_spatial_dropout_1d.html">layer_spatial_dropout_1d</a>()</code> <br><code><a href="layer_spatial_dropout_2d.html">layer_spatial_dropout_2d</a>()</code> <br><code><a href="layer_spatial_dropout_3d.html">layer_spatial_dropout_3d</a>()</code> <br><code><a href="layer_spectral_normalization.html">layer_spectral_normalization</a>()</code> <br><code><a href="layer_stft_spectrogram.html">layer_stft_spectrogram</a>()</code> <br><code><a href="layer_string_lookup.html">layer_string_lookup</a>()</code> <br><code><a href="layer_subtract.html">layer_subtract</a>()</code> <br><code><a href="layer_text_vectorization.html">layer_text_vectorization</a>()</code> <br><code><a href="layer_tfsm.html">layer_tfsm</a>()</code> <br><code><a href="layer_time_distributed.html">layer_time_distributed</a>()</code> <br><code><a href="layer_torch_module_wrapper.html">layer_torch_module_wrapper</a>()</code> <br><code><a href="layer_unit_normalization.html">layer_unit_normalization</a>()</code> <br><code><a href="layer_upsampling_1d.html">layer_upsampling_1d</a>()</code> <br><code><a href="layer_upsampling_2d.html">layer_upsampling_2d</a>()</code> <br><code><a href="layer_upsampling_3d.html">layer_upsampling_3d</a>()</code> <br><code><a href="layer_zero_padding_1d.html">layer_zero_padding_1d</a>()</code> <br><code><a href="layer_zero_padding_2d.html">layer_zero_padding_2d</a>()</code> <br><code><a href="layer_zero_padding_3d.html">layer_zero_padding_3d</a>()</code> <br><code><a href="rnn_cell_gru.html">rnn_cell_gru</a>()</code> <br><code><a href="rnn_cell_lstm.html">rnn_cell_lstm</a>()</code> <br><code><a href="rnn_cell_simple.html">rnn_cell_simple</a>()</code> <br><code><a href="rnn_cells_stack.html">rnn_cells_stack</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -206,7 +206,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/layer_permute.html b/docs/dev/reference/layer_permute.html
index 60d0edbb3..ecc8513c3 100644
--- a/docs/dev/reference/layer_permute.html
+++ b/docs/dev/reference/layer_permute.html
@@ -8,7 +8,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -112,7 +112,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
     <div class="dont-index">
 <ul><li><p><a href="https://keras.io/api/layers/reshaping_layers/permute#permute-class" class="external-link">https://keras.io/api/layers/reshaping_layers/permute#permute-class</a></p></li>
 </ul><p>Other reshaping layers: <br><code><a href="layer_cropping_1d.html">layer_cropping_1d</a>()</code> <br><code><a href="layer_cropping_2d.html">layer_cropping_2d</a>()</code> <br><code><a href="layer_cropping_3d.html">layer_cropping_3d</a>()</code> <br><code><a href="layer_flatten.html">layer_flatten</a>()</code> <br><code><a href="layer_repeat_vector.html">layer_repeat_vector</a>()</code> <br><code><a href="layer_reshape.html">layer_reshape</a>()</code> <br><code><a href="layer_upsampling_1d.html">layer_upsampling_1d</a>()</code> <br><code><a href="layer_upsampling_2d.html">layer_upsampling_2d</a>()</code> <br><code><a href="layer_upsampling_3d.html">layer_upsampling_3d</a>()</code> <br><code><a href="layer_zero_padding_1d.html">layer_zero_padding_1d</a>()</code> <br><code><a href="layer_zero_padding_2d.html">layer_zero_padding_2d</a>()</code> <br><code><a href="layer_zero_padding_3d.html">layer_zero_padding_3d</a>()</code> <br></p>
-<p>Other layers: <br><code><a href="Layer.html">Layer</a>()</code> <br><code><a href="layer_activation.html">layer_activation</a>()</code> <br><code><a href="layer_activation_elu.html">layer_activation_elu</a>()</code> <br><code><a href="layer_activation_leaky_relu.html">layer_activation_leaky_relu</a>()</code> <br><code><a href="layer_activation_parametric_relu.html">layer_activation_parametric_relu</a>()</code> <br><code><a href="layer_activation_relu.html">layer_activation_relu</a>()</code> <br><code><a href="layer_activation_softmax.html">layer_activation_softmax</a>()</code> <br><code><a href="layer_activity_regularization.html">layer_activity_regularization</a>()</code> <br><code><a href="layer_add.html">layer_add</a>()</code> <br><code><a href="layer_additive_attention.html">layer_additive_attention</a>()</code> <br><code><a href="layer_alpha_dropout.html">layer_alpha_dropout</a>()</code> <br><code><a href="layer_attention.html">layer_attention</a>()</code> <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_average.html">layer_average</a>()</code> <br><code><a href="layer_average_pooling_1d.html">layer_average_pooling_1d</a>()</code> <br><code><a href="layer_average_pooling_2d.html">layer_average_pooling_2d</a>()</code> <br><code><a href="layer_average_pooling_3d.html">layer_average_pooling_3d</a>()</code> <br><code><a href="layer_batch_normalization.html">layer_batch_normalization</a>()</code> <br><code><a href="layer_bidirectional.html">layer_bidirectional</a>()</code> <br><code><a href="layer_category_encoding.html">layer_category_encoding</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_concatenate.html">layer_concatenate</a>()</code> <br><code><a href="layer_conv_1d.html">layer_conv_1d</a>()</code> <br><code><a href="layer_conv_1d_transpose.html">layer_conv_1d_transpose</a>()</code> <br><code><a href="layer_conv_2d.html">layer_conv_2d</a>()</code> <br><code><a href="layer_conv_2d_transpose.html">layer_conv_2d_transpose</a>()</code> <br><code><a href="layer_conv_3d.html">layer_conv_3d</a>()</code> <br><code><a href="layer_conv_3d_transpose.html">layer_conv_3d_transpose</a>()</code> <br><code><a href="layer_conv_lstm_1d.html">layer_conv_lstm_1d</a>()</code> <br><code><a href="layer_conv_lstm_2d.html">layer_conv_lstm_2d</a>()</code> <br><code><a href="layer_conv_lstm_3d.html">layer_conv_lstm_3d</a>()</code> <br><code><a href="layer_cropping_1d.html">layer_cropping_1d</a>()</code> <br><code><a href="layer_cropping_2d.html">layer_cropping_2d</a>()</code> <br><code><a href="layer_cropping_3d.html">layer_cropping_3d</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_dense.html">layer_dense</a>()</code> <br><code><a href="layer_depthwise_conv_1d.html">layer_depthwise_conv_1d</a>()</code> <br><code><a href="layer_depthwise_conv_2d.html">layer_depthwise_conv_2d</a>()</code> <br><code><a href="layer_discretization.html">layer_discretization</a>()</code> <br><code><a href="layer_dot.html">layer_dot</a>()</code> <br><code><a href="layer_dropout.html">layer_dropout</a>()</code> <br><code><a href="layer_einsum_dense.html">layer_einsum_dense</a>()</code> <br><code><a href="layer_embedding.html">layer_embedding</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_feature_space.html">layer_feature_space</a>()</code> <br><code><a href="layer_flatten.html">layer_flatten</a>()</code> <br><code><a href="layer_flax_module_wrapper.html">layer_flax_module_wrapper</a>()</code> <br><code><a href="layer_gaussian_dropout.html">layer_gaussian_dropout</a>()</code> <br><code><a href="layer_gaussian_noise.html">layer_gaussian_noise</a>()</code> <br><code><a href="layer_global_average_pooling_1d.html">layer_global_average_pooling_1d</a>()</code> <br><code><a href="layer_global_average_pooling_2d.html">layer_global_average_pooling_2d</a>()</code> <br><code><a href="layer_global_average_pooling_3d.html">layer_global_average_pooling_3d</a>()</code> <br><code><a href="layer_global_max_pooling_1d.html">layer_global_max_pooling_1d</a>()</code> <br><code><a href="layer_global_max_pooling_2d.html">layer_global_max_pooling_2d</a>()</code> <br><code><a href="layer_global_max_pooling_3d.html">layer_global_max_pooling_3d</a>()</code> <br><code><a href="layer_group_normalization.html">layer_group_normalization</a>()</code> <br><code><a href="layer_group_query_attention.html">layer_group_query_attention</a>()</code> <br><code><a href="layer_gru.html">layer_gru</a>()</code> <br><code><a href="layer_hashed_crossing.html">layer_hashed_crossing</a>()</code> <br><code><a href="layer_hashing.html">layer_hashing</a>()</code> <br><code><a href="layer_identity.html">layer_identity</a>()</code> <br><code><a href="layer_integer_lookup.html">layer_integer_lookup</a>()</code> <br><code><a href="layer_jax_model_wrapper.html">layer_jax_model_wrapper</a>()</code> <br><code><a href="layer_lambda.html">layer_lambda</a>()</code> <br><code><a href="layer_layer_normalization.html">layer_layer_normalization</a>()</code> <br><code><a href="layer_lstm.html">layer_lstm</a>()</code> <br><code><a href="layer_masking.html">layer_masking</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_max_pooling_1d.html">layer_max_pooling_1d</a>()</code> <br><code><a href="layer_max_pooling_2d.html">layer_max_pooling_2d</a>()</code> <br><code><a href="layer_max_pooling_3d.html">layer_max_pooling_3d</a>()</code> <br><code><a href="layer_maximum.html">layer_maximum</a>()</code> <br><code><a href="layer_mel_spectrogram.html">layer_mel_spectrogram</a>()</code> <br><code><a href="layer_minimum.html">layer_minimum</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_multi_head_attention.html">layer_multi_head_attention</a>()</code> <br><code><a href="layer_multiply.html">layer_multiply</a>()</code> <br><code><a href="layer_normalization.html">layer_normalization</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_brightness.html">layer_random_brightness</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_contrast.html">layer_random_contrast</a>()</code> <br><code><a href="layer_random_crop.html">layer_random_crop</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_flip.html">layer_random_flip</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_rotation.html">layer_random_rotation</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_random_translation.html">layer_random_translation</a>()</code> <br><code><a href="layer_random_zoom.html">layer_random_zoom</a>()</code> <br><code><a href="layer_repeat_vector.html">layer_repeat_vector</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_reshape.html">layer_reshape</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_rms_normalization.html">layer_rms_normalization</a>()</code> <br><code><a href="layer_rnn.html">layer_rnn</a>()</code> <br><code><a href="layer_separable_conv_1d.html">layer_separable_conv_1d</a>()</code> <br><code><a href="layer_separable_conv_2d.html">layer_separable_conv_2d</a>()</code> <br><code><a href="layer_simple_rnn.html">layer_simple_rnn</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br><code><a href="layer_spatial_dropout_1d.html">layer_spatial_dropout_1d</a>()</code> <br><code><a href="layer_spatial_dropout_2d.html">layer_spatial_dropout_2d</a>()</code> <br><code><a href="layer_spatial_dropout_3d.html">layer_spatial_dropout_3d</a>()</code> <br><code><a href="layer_spectral_normalization.html">layer_spectral_normalization</a>()</code> <br><code><a href="layer_stft_spectrogram.html">layer_stft_spectrogram</a>()</code> <br><code><a href="layer_string_lookup.html">layer_string_lookup</a>()</code> <br><code><a href="layer_subtract.html">layer_subtract</a>()</code> <br><code><a href="layer_text_vectorization.html">layer_text_vectorization</a>()</code> <br><code><a href="layer_tfsm.html">layer_tfsm</a>()</code> <br><code><a href="layer_time_distributed.html">layer_time_distributed</a>()</code> <br><code><a href="layer_torch_module_wrapper.html">layer_torch_module_wrapper</a>()</code> <br><code><a href="layer_unit_normalization.html">layer_unit_normalization</a>()</code> <br><code><a href="layer_upsampling_1d.html">layer_upsampling_1d</a>()</code> <br><code><a href="layer_upsampling_2d.html">layer_upsampling_2d</a>()</code> <br><code><a href="layer_upsampling_3d.html">layer_upsampling_3d</a>()</code> <br><code><a href="layer_zero_padding_1d.html">layer_zero_padding_1d</a>()</code> <br><code><a href="layer_zero_padding_2d.html">layer_zero_padding_2d</a>()</code> <br><code><a href="layer_zero_padding_3d.html">layer_zero_padding_3d</a>()</code> <br><code><a href="rnn_cell_gru.html">rnn_cell_gru</a>()</code> <br><code><a href="rnn_cell_lstm.html">rnn_cell_lstm</a>()</code> <br><code><a href="rnn_cell_simple.html">rnn_cell_simple</a>()</code> <br><code><a href="rnn_cells_stack.html">rnn_cells_stack</a>()</code> <br></p></div>
+<p>Other layers: <br><code><a href="Layer.html">Layer</a>()</code> <br><code><a href="layer_activation.html">layer_activation</a>()</code> <br><code><a href="layer_activation_elu.html">layer_activation_elu</a>()</code> <br><code><a href="layer_activation_leaky_relu.html">layer_activation_leaky_relu</a>()</code> <br><code><a href="layer_activation_parametric_relu.html">layer_activation_parametric_relu</a>()</code> <br><code><a href="layer_activation_relu.html">layer_activation_relu</a>()</code> <br><code><a href="layer_activation_softmax.html">layer_activation_softmax</a>()</code> <br><code><a href="layer_activity_regularization.html">layer_activity_regularization</a>()</code> <br><code><a href="layer_add.html">layer_add</a>()</code> <br><code><a href="layer_additive_attention.html">layer_additive_attention</a>()</code> <br><code><a href="layer_alpha_dropout.html">layer_alpha_dropout</a>()</code> <br><code><a href="layer_attention.html">layer_attention</a>()</code> <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_average.html">layer_average</a>()</code> <br><code><a href="layer_average_pooling_1d.html">layer_average_pooling_1d</a>()</code> <br><code><a href="layer_average_pooling_2d.html">layer_average_pooling_2d</a>()</code> <br><code><a href="layer_average_pooling_3d.html">layer_average_pooling_3d</a>()</code> <br><code><a href="layer_batch_normalization.html">layer_batch_normalization</a>()</code> <br><code><a href="layer_bidirectional.html">layer_bidirectional</a>()</code> <br><code><a href="layer_category_encoding.html">layer_category_encoding</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_concatenate.html">layer_concatenate</a>()</code> <br><code><a href="layer_conv_1d.html">layer_conv_1d</a>()</code> <br><code><a href="layer_conv_1d_transpose.html">layer_conv_1d_transpose</a>()</code> <br><code><a href="layer_conv_2d.html">layer_conv_2d</a>()</code> <br><code><a href="layer_conv_2d_transpose.html">layer_conv_2d_transpose</a>()</code> <br><code><a href="layer_conv_3d.html">layer_conv_3d</a>()</code> <br><code><a href="layer_conv_3d_transpose.html">layer_conv_3d_transpose</a>()</code> <br><code><a href="layer_conv_lstm_1d.html">layer_conv_lstm_1d</a>()</code> <br><code><a href="layer_conv_lstm_2d.html">layer_conv_lstm_2d</a>()</code> <br><code><a href="layer_conv_lstm_3d.html">layer_conv_lstm_3d</a>()</code> <br><code><a href="layer_cropping_1d.html">layer_cropping_1d</a>()</code> <br><code><a href="layer_cropping_2d.html">layer_cropping_2d</a>()</code> <br><code><a href="layer_cropping_3d.html">layer_cropping_3d</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_dense.html">layer_dense</a>()</code> <br><code><a href="layer_depthwise_conv_1d.html">layer_depthwise_conv_1d</a>()</code> <br><code><a href="layer_depthwise_conv_2d.html">layer_depthwise_conv_2d</a>()</code> <br><code><a href="layer_discretization.html">layer_discretization</a>()</code> <br><code><a href="layer_dot.html">layer_dot</a>()</code> <br><code><a href="layer_dropout.html">layer_dropout</a>()</code> <br><code><a href="layer_einsum_dense.html">layer_einsum_dense</a>()</code> <br><code><a href="layer_embedding.html">layer_embedding</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_feature_space.html">layer_feature_space</a>()</code> <br><code><a href="layer_flatten.html">layer_flatten</a>()</code> <br><code><a href="layer_flax_module_wrapper.html">layer_flax_module_wrapper</a>()</code> <br><code><a href="layer_gaussian_dropout.html">layer_gaussian_dropout</a>()</code> <br><code><a href="layer_gaussian_noise.html">layer_gaussian_noise</a>()</code> <br><code><a href="layer_global_average_pooling_1d.html">layer_global_average_pooling_1d</a>()</code> <br><code><a href="layer_global_average_pooling_2d.html">layer_global_average_pooling_2d</a>()</code> <br><code><a href="layer_global_average_pooling_3d.html">layer_global_average_pooling_3d</a>()</code> <br><code><a href="layer_global_max_pooling_1d.html">layer_global_max_pooling_1d</a>()</code> <br><code><a href="layer_global_max_pooling_2d.html">layer_global_max_pooling_2d</a>()</code> <br><code><a href="layer_global_max_pooling_3d.html">layer_global_max_pooling_3d</a>()</code> <br><code><a href="layer_group_normalization.html">layer_group_normalization</a>()</code> <br><code><a href="layer_group_query_attention.html">layer_group_query_attention</a>()</code> <br><code><a href="layer_gru.html">layer_gru</a>()</code> <br><code><a href="layer_hashed_crossing.html">layer_hashed_crossing</a>()</code> <br><code><a href="layer_hashing.html">layer_hashing</a>()</code> <br><code><a href="layer_identity.html">layer_identity</a>()</code> <br><code><a href="layer_integer_lookup.html">layer_integer_lookup</a>()</code> <br><code><a href="layer_jax_model_wrapper.html">layer_jax_model_wrapper</a>()</code> <br><code><a href="layer_lambda.html">layer_lambda</a>()</code> <br><code><a href="layer_layer_normalization.html">layer_layer_normalization</a>()</code> <br><code><a href="layer_lstm.html">layer_lstm</a>()</code> <br><code><a href="layer_masking.html">layer_masking</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_max_pooling_1d.html">layer_max_pooling_1d</a>()</code> <br><code><a href="layer_max_pooling_2d.html">layer_max_pooling_2d</a>()</code> <br><code><a href="layer_max_pooling_3d.html">layer_max_pooling_3d</a>()</code> <br><code><a href="layer_maximum.html">layer_maximum</a>()</code> <br><code><a href="layer_mel_spectrogram.html">layer_mel_spectrogram</a>()</code> <br><code><a href="layer_minimum.html">layer_minimum</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_multi_head_attention.html">layer_multi_head_attention</a>()</code> <br><code><a href="layer_multiply.html">layer_multiply</a>()</code> <br><code><a href="layer_normalization.html">layer_normalization</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_brightness.html">layer_random_brightness</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_contrast.html">layer_random_contrast</a>()</code> <br><code><a href="layer_random_crop.html">layer_random_crop</a>()</code> <br><code><a href="layer_random_elastic_transform.html">layer_random_elastic_transform</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_flip.html">layer_random_flip</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_rotation.html">layer_random_rotation</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_random_translation.html">layer_random_translation</a>()</code> <br><code><a href="layer_random_zoom.html">layer_random_zoom</a>()</code> <br><code><a href="layer_repeat_vector.html">layer_repeat_vector</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_reshape.html">layer_reshape</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_rms_normalization.html">layer_rms_normalization</a>()</code> <br><code><a href="layer_rnn.html">layer_rnn</a>()</code> <br><code><a href="layer_separable_conv_1d.html">layer_separable_conv_1d</a>()</code> <br><code><a href="layer_separable_conv_2d.html">layer_separable_conv_2d</a>()</code> <br><code><a href="layer_simple_rnn.html">layer_simple_rnn</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br><code><a href="layer_spatial_dropout_1d.html">layer_spatial_dropout_1d</a>()</code> <br><code><a href="layer_spatial_dropout_2d.html">layer_spatial_dropout_2d</a>()</code> <br><code><a href="layer_spatial_dropout_3d.html">layer_spatial_dropout_3d</a>()</code> <br><code><a href="layer_spectral_normalization.html">layer_spectral_normalization</a>()</code> <br><code><a href="layer_stft_spectrogram.html">layer_stft_spectrogram</a>()</code> <br><code><a href="layer_string_lookup.html">layer_string_lookup</a>()</code> <br><code><a href="layer_subtract.html">layer_subtract</a>()</code> <br><code><a href="layer_text_vectorization.html">layer_text_vectorization</a>()</code> <br><code><a href="layer_tfsm.html">layer_tfsm</a>()</code> <br><code><a href="layer_time_distributed.html">layer_time_distributed</a>()</code> <br><code><a href="layer_torch_module_wrapper.html">layer_torch_module_wrapper</a>()</code> <br><code><a href="layer_unit_normalization.html">layer_unit_normalization</a>()</code> <br><code><a href="layer_upsampling_1d.html">layer_upsampling_1d</a>()</code> <br><code><a href="layer_upsampling_2d.html">layer_upsampling_2d</a>()</code> <br><code><a href="layer_upsampling_3d.html">layer_upsampling_3d</a>()</code> <br><code><a href="layer_zero_padding_1d.html">layer_zero_padding_1d</a>()</code> <br><code><a href="layer_zero_padding_2d.html">layer_zero_padding_2d</a>()</code> <br><code><a href="layer_zero_padding_3d.html">layer_zero_padding_3d</a>()</code> <br><code><a href="rnn_cell_gru.html">rnn_cell_gru</a>()</code> <br><code><a href="rnn_cell_lstm.html">rnn_cell_lstm</a>()</code> <br><code><a href="rnn_cell_simple.html">rnn_cell_simple</a>()</code> <br><code><a href="rnn_cells_stack.html">rnn_cells_stack</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -124,7 +124,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/layer_pipeline.html b/docs/dev/reference/layer_pipeline.html
index 692b47a94..bca4c2701 100644
--- a/docs/dev/reference/layer_pipeline.html
+++ b/docs/dev/reference/layer_pipeline.html
@@ -34,7 +34,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -141,7 +141,7 @@ <h2 id="examples">Examples<a class="anchor" aria-label="anchor" href="#examples"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/layer_rand_augment.html b/docs/dev/reference/layer_rand_augment.html
index 408579ffc..445306f0f 100644
--- a/docs/dev/reference/layer_rand_augment.html
+++ b/docs/dev/reference/layer_rand_augment.html
@@ -12,7 +12,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -130,9 +130,9 @@ <h2 id="references">References<a class="anchor" aria-label="anchor" href="#refer
 </ul></div>
     <div class="section level2">
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
-    <div class="dont-index"><p>Other image preprocessing layers: <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br></p>
-<p>Other preprocessing layers: <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_category_encoding.html">layer_category_encoding</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_discretization.html">layer_discretization</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_feature_space.html">layer_feature_space</a>()</code> <br><code><a href="layer_hashed_crossing.html">layer_hashed_crossing</a>()</code> <br><code><a href="layer_hashing.html">layer_hashing</a>()</code> <br><code><a href="layer_integer_lookup.html">layer_integer_lookup</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_mel_spectrogram.html">layer_mel_spectrogram</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_normalization.html">layer_normalization</a>()</code> <br><code><a href="layer_random_brightness.html">layer_random_brightness</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_contrast.html">layer_random_contrast</a>()</code> <br><code><a href="layer_random_crop.html">layer_random_crop</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_flip.html">layer_random_flip</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_rotation.html">layer_random_rotation</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_random_translation.html">layer_random_translation</a>()</code> <br><code><a href="layer_random_zoom.html">layer_random_zoom</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br><code><a href="layer_stft_spectrogram.html">layer_stft_spectrogram</a>()</code> <br><code><a href="layer_string_lookup.html">layer_string_lookup</a>()</code> <br><code><a href="layer_text_vectorization.html">layer_text_vectorization</a>()</code> <br></p>
-<p>Other layers: <br><code><a href="Layer.html">Layer</a>()</code> <br><code><a href="layer_activation.html">layer_activation</a>()</code> <br><code><a href="layer_activation_elu.html">layer_activation_elu</a>()</code> <br><code><a href="layer_activation_leaky_relu.html">layer_activation_leaky_relu</a>()</code> <br><code><a href="layer_activation_parametric_relu.html">layer_activation_parametric_relu</a>()</code> <br><code><a href="layer_activation_relu.html">layer_activation_relu</a>()</code> <br><code><a href="layer_activation_softmax.html">layer_activation_softmax</a>()</code> <br><code><a href="layer_activity_regularization.html">layer_activity_regularization</a>()</code> <br><code><a href="layer_add.html">layer_add</a>()</code> <br><code><a href="layer_additive_attention.html">layer_additive_attention</a>()</code> <br><code><a href="layer_alpha_dropout.html">layer_alpha_dropout</a>()</code> <br><code><a href="layer_attention.html">layer_attention</a>()</code> <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_average.html">layer_average</a>()</code> <br><code><a href="layer_average_pooling_1d.html">layer_average_pooling_1d</a>()</code> <br><code><a href="layer_average_pooling_2d.html">layer_average_pooling_2d</a>()</code> <br><code><a href="layer_average_pooling_3d.html">layer_average_pooling_3d</a>()</code> <br><code><a href="layer_batch_normalization.html">layer_batch_normalization</a>()</code> <br><code><a href="layer_bidirectional.html">layer_bidirectional</a>()</code> <br><code><a href="layer_category_encoding.html">layer_category_encoding</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_concatenate.html">layer_concatenate</a>()</code> <br><code><a href="layer_conv_1d.html">layer_conv_1d</a>()</code> <br><code><a href="layer_conv_1d_transpose.html">layer_conv_1d_transpose</a>()</code> <br><code><a href="layer_conv_2d.html">layer_conv_2d</a>()</code> <br><code><a href="layer_conv_2d_transpose.html">layer_conv_2d_transpose</a>()</code> <br><code><a href="layer_conv_3d.html">layer_conv_3d</a>()</code> <br><code><a href="layer_conv_3d_transpose.html">layer_conv_3d_transpose</a>()</code> <br><code><a href="layer_conv_lstm_1d.html">layer_conv_lstm_1d</a>()</code> <br><code><a href="layer_conv_lstm_2d.html">layer_conv_lstm_2d</a>()</code> <br><code><a href="layer_conv_lstm_3d.html">layer_conv_lstm_3d</a>()</code> <br><code><a href="layer_cropping_1d.html">layer_cropping_1d</a>()</code> <br><code><a href="layer_cropping_2d.html">layer_cropping_2d</a>()</code> <br><code><a href="layer_cropping_3d.html">layer_cropping_3d</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_dense.html">layer_dense</a>()</code> <br><code><a href="layer_depthwise_conv_1d.html">layer_depthwise_conv_1d</a>()</code> <br><code><a href="layer_depthwise_conv_2d.html">layer_depthwise_conv_2d</a>()</code> <br><code><a href="layer_discretization.html">layer_discretization</a>()</code> <br><code><a href="layer_dot.html">layer_dot</a>()</code> <br><code><a href="layer_dropout.html">layer_dropout</a>()</code> <br><code><a href="layer_einsum_dense.html">layer_einsum_dense</a>()</code> <br><code><a href="layer_embedding.html">layer_embedding</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_feature_space.html">layer_feature_space</a>()</code> <br><code><a href="layer_flatten.html">layer_flatten</a>()</code> <br><code><a href="layer_flax_module_wrapper.html">layer_flax_module_wrapper</a>()</code> <br><code><a href="layer_gaussian_dropout.html">layer_gaussian_dropout</a>()</code> <br><code><a href="layer_gaussian_noise.html">layer_gaussian_noise</a>()</code> <br><code><a href="layer_global_average_pooling_1d.html">layer_global_average_pooling_1d</a>()</code> <br><code><a href="layer_global_average_pooling_2d.html">layer_global_average_pooling_2d</a>()</code> <br><code><a href="layer_global_average_pooling_3d.html">layer_global_average_pooling_3d</a>()</code> <br><code><a href="layer_global_max_pooling_1d.html">layer_global_max_pooling_1d</a>()</code> <br><code><a href="layer_global_max_pooling_2d.html">layer_global_max_pooling_2d</a>()</code> <br><code><a href="layer_global_max_pooling_3d.html">layer_global_max_pooling_3d</a>()</code> <br><code><a href="layer_group_normalization.html">layer_group_normalization</a>()</code> <br><code><a href="layer_group_query_attention.html">layer_group_query_attention</a>()</code> <br><code><a href="layer_gru.html">layer_gru</a>()</code> <br><code><a href="layer_hashed_crossing.html">layer_hashed_crossing</a>()</code> <br><code><a href="layer_hashing.html">layer_hashing</a>()</code> <br><code><a href="layer_identity.html">layer_identity</a>()</code> <br><code><a href="layer_integer_lookup.html">layer_integer_lookup</a>()</code> <br><code><a href="layer_jax_model_wrapper.html">layer_jax_model_wrapper</a>()</code> <br><code><a href="layer_lambda.html">layer_lambda</a>()</code> <br><code><a href="layer_layer_normalization.html">layer_layer_normalization</a>()</code> <br><code><a href="layer_lstm.html">layer_lstm</a>()</code> <br><code><a href="layer_masking.html">layer_masking</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_max_pooling_1d.html">layer_max_pooling_1d</a>()</code> <br><code><a href="layer_max_pooling_2d.html">layer_max_pooling_2d</a>()</code> <br><code><a href="layer_max_pooling_3d.html">layer_max_pooling_3d</a>()</code> <br><code><a href="layer_maximum.html">layer_maximum</a>()</code> <br><code><a href="layer_mel_spectrogram.html">layer_mel_spectrogram</a>()</code> <br><code><a href="layer_minimum.html">layer_minimum</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_multi_head_attention.html">layer_multi_head_attention</a>()</code> <br><code><a href="layer_multiply.html">layer_multiply</a>()</code> <br><code><a href="layer_normalization.html">layer_normalization</a>()</code> <br><code><a href="layer_permute.html">layer_permute</a>()</code> <br><code><a href="layer_random_brightness.html">layer_random_brightness</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_contrast.html">layer_random_contrast</a>()</code> <br><code><a href="layer_random_crop.html">layer_random_crop</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_flip.html">layer_random_flip</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_rotation.html">layer_random_rotation</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_random_translation.html">layer_random_translation</a>()</code> <br><code><a href="layer_random_zoom.html">layer_random_zoom</a>()</code> <br><code><a href="layer_repeat_vector.html">layer_repeat_vector</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_reshape.html">layer_reshape</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_rms_normalization.html">layer_rms_normalization</a>()</code> <br><code><a href="layer_rnn.html">layer_rnn</a>()</code> <br><code><a href="layer_separable_conv_1d.html">layer_separable_conv_1d</a>()</code> <br><code><a href="layer_separable_conv_2d.html">layer_separable_conv_2d</a>()</code> <br><code><a href="layer_simple_rnn.html">layer_simple_rnn</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br><code><a href="layer_spatial_dropout_1d.html">layer_spatial_dropout_1d</a>()</code> <br><code><a href="layer_spatial_dropout_2d.html">layer_spatial_dropout_2d</a>()</code> <br><code><a href="layer_spatial_dropout_3d.html">layer_spatial_dropout_3d</a>()</code> <br><code><a href="layer_spectral_normalization.html">layer_spectral_normalization</a>()</code> <br><code><a href="layer_stft_spectrogram.html">layer_stft_spectrogram</a>()</code> <br><code><a href="layer_string_lookup.html">layer_string_lookup</a>()</code> <br><code><a href="layer_subtract.html">layer_subtract</a>()</code> <br><code><a href="layer_text_vectorization.html">layer_text_vectorization</a>()</code> <br><code><a href="layer_tfsm.html">layer_tfsm</a>()</code> <br><code><a href="layer_time_distributed.html">layer_time_distributed</a>()</code> <br><code><a href="layer_torch_module_wrapper.html">layer_torch_module_wrapper</a>()</code> <br><code><a href="layer_unit_normalization.html">layer_unit_normalization</a>()</code> <br><code><a href="layer_upsampling_1d.html">layer_upsampling_1d</a>()</code> <br><code><a href="layer_upsampling_2d.html">layer_upsampling_2d</a>()</code> <br><code><a href="layer_upsampling_3d.html">layer_upsampling_3d</a>()</code> <br><code><a href="layer_zero_padding_1d.html">layer_zero_padding_1d</a>()</code> <br><code><a href="layer_zero_padding_2d.html">layer_zero_padding_2d</a>()</code> <br><code><a href="layer_zero_padding_3d.html">layer_zero_padding_3d</a>()</code> <br><code><a href="rnn_cell_gru.html">rnn_cell_gru</a>()</code> <br><code><a href="rnn_cell_lstm.html">rnn_cell_lstm</a>()</code> <br><code><a href="rnn_cell_simple.html">rnn_cell_simple</a>()</code> <br><code><a href="rnn_cells_stack.html">rnn_cells_stack</a>()</code> <br></p></div>
+    <div class="dont-index"><p>Other image preprocessing layers: <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_elastic_transform.html">layer_random_elastic_transform</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br></p>
+<p>Other preprocessing layers: <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_category_encoding.html">layer_category_encoding</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_discretization.html">layer_discretization</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_feature_space.html">layer_feature_space</a>()</code> <br><code><a href="layer_hashed_crossing.html">layer_hashed_crossing</a>()</code> <br><code><a href="layer_hashing.html">layer_hashing</a>()</code> <br><code><a href="layer_integer_lookup.html">layer_integer_lookup</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_mel_spectrogram.html">layer_mel_spectrogram</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_normalization.html">layer_normalization</a>()</code> <br><code><a href="layer_random_brightness.html">layer_random_brightness</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_contrast.html">layer_random_contrast</a>()</code> <br><code><a href="layer_random_crop.html">layer_random_crop</a>()</code> <br><code><a href="layer_random_elastic_transform.html">layer_random_elastic_transform</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_flip.html">layer_random_flip</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_rotation.html">layer_random_rotation</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_random_translation.html">layer_random_translation</a>()</code> <br><code><a href="layer_random_zoom.html">layer_random_zoom</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br><code><a href="layer_stft_spectrogram.html">layer_stft_spectrogram</a>()</code> <br><code><a href="layer_string_lookup.html">layer_string_lookup</a>()</code> <br><code><a href="layer_text_vectorization.html">layer_text_vectorization</a>()</code> <br></p>
+<p>Other layers: <br><code><a href="Layer.html">Layer</a>()</code> <br><code><a href="layer_activation.html">layer_activation</a>()</code> <br><code><a href="layer_activation_elu.html">layer_activation_elu</a>()</code> <br><code><a href="layer_activation_leaky_relu.html">layer_activation_leaky_relu</a>()</code> <br><code><a href="layer_activation_parametric_relu.html">layer_activation_parametric_relu</a>()</code> <br><code><a href="layer_activation_relu.html">layer_activation_relu</a>()</code> <br><code><a href="layer_activation_softmax.html">layer_activation_softmax</a>()</code> <br><code><a href="layer_activity_regularization.html">layer_activity_regularization</a>()</code> <br><code><a href="layer_add.html">layer_add</a>()</code> <br><code><a href="layer_additive_attention.html">layer_additive_attention</a>()</code> <br><code><a href="layer_alpha_dropout.html">layer_alpha_dropout</a>()</code> <br><code><a href="layer_attention.html">layer_attention</a>()</code> <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_average.html">layer_average</a>()</code> <br><code><a href="layer_average_pooling_1d.html">layer_average_pooling_1d</a>()</code> <br><code><a href="layer_average_pooling_2d.html">layer_average_pooling_2d</a>()</code> <br><code><a href="layer_average_pooling_3d.html">layer_average_pooling_3d</a>()</code> <br><code><a href="layer_batch_normalization.html">layer_batch_normalization</a>()</code> <br><code><a href="layer_bidirectional.html">layer_bidirectional</a>()</code> <br><code><a href="layer_category_encoding.html">layer_category_encoding</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_concatenate.html">layer_concatenate</a>()</code> <br><code><a href="layer_conv_1d.html">layer_conv_1d</a>()</code> <br><code><a href="layer_conv_1d_transpose.html">layer_conv_1d_transpose</a>()</code> <br><code><a href="layer_conv_2d.html">layer_conv_2d</a>()</code> <br><code><a href="layer_conv_2d_transpose.html">layer_conv_2d_transpose</a>()</code> <br><code><a href="layer_conv_3d.html">layer_conv_3d</a>()</code> <br><code><a href="layer_conv_3d_transpose.html">layer_conv_3d_transpose</a>()</code> <br><code><a href="layer_conv_lstm_1d.html">layer_conv_lstm_1d</a>()</code> <br><code><a href="layer_conv_lstm_2d.html">layer_conv_lstm_2d</a>()</code> <br><code><a href="layer_conv_lstm_3d.html">layer_conv_lstm_3d</a>()</code> <br><code><a href="layer_cropping_1d.html">layer_cropping_1d</a>()</code> <br><code><a href="layer_cropping_2d.html">layer_cropping_2d</a>()</code> <br><code><a href="layer_cropping_3d.html">layer_cropping_3d</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_dense.html">layer_dense</a>()</code> <br><code><a href="layer_depthwise_conv_1d.html">layer_depthwise_conv_1d</a>()</code> <br><code><a href="layer_depthwise_conv_2d.html">layer_depthwise_conv_2d</a>()</code> <br><code><a href="layer_discretization.html">layer_discretization</a>()</code> <br><code><a href="layer_dot.html">layer_dot</a>()</code> <br><code><a href="layer_dropout.html">layer_dropout</a>()</code> <br><code><a href="layer_einsum_dense.html">layer_einsum_dense</a>()</code> <br><code><a href="layer_embedding.html">layer_embedding</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_feature_space.html">layer_feature_space</a>()</code> <br><code><a href="layer_flatten.html">layer_flatten</a>()</code> <br><code><a href="layer_flax_module_wrapper.html">layer_flax_module_wrapper</a>()</code> <br><code><a href="layer_gaussian_dropout.html">layer_gaussian_dropout</a>()</code> <br><code><a href="layer_gaussian_noise.html">layer_gaussian_noise</a>()</code> <br><code><a href="layer_global_average_pooling_1d.html">layer_global_average_pooling_1d</a>()</code> <br><code><a href="layer_global_average_pooling_2d.html">layer_global_average_pooling_2d</a>()</code> <br><code><a href="layer_global_average_pooling_3d.html">layer_global_average_pooling_3d</a>()</code> <br><code><a href="layer_global_max_pooling_1d.html">layer_global_max_pooling_1d</a>()</code> <br><code><a href="layer_global_max_pooling_2d.html">layer_global_max_pooling_2d</a>()</code> <br><code><a href="layer_global_max_pooling_3d.html">layer_global_max_pooling_3d</a>()</code> <br><code><a href="layer_group_normalization.html">layer_group_normalization</a>()</code> <br><code><a href="layer_group_query_attention.html">layer_group_query_attention</a>()</code> <br><code><a href="layer_gru.html">layer_gru</a>()</code> <br><code><a href="layer_hashed_crossing.html">layer_hashed_crossing</a>()</code> <br><code><a href="layer_hashing.html">layer_hashing</a>()</code> <br><code><a href="layer_identity.html">layer_identity</a>()</code> <br><code><a href="layer_integer_lookup.html">layer_integer_lookup</a>()</code> <br><code><a href="layer_jax_model_wrapper.html">layer_jax_model_wrapper</a>()</code> <br><code><a href="layer_lambda.html">layer_lambda</a>()</code> <br><code><a href="layer_layer_normalization.html">layer_layer_normalization</a>()</code> <br><code><a href="layer_lstm.html">layer_lstm</a>()</code> <br><code><a href="layer_masking.html">layer_masking</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_max_pooling_1d.html">layer_max_pooling_1d</a>()</code> <br><code><a href="layer_max_pooling_2d.html">layer_max_pooling_2d</a>()</code> <br><code><a href="layer_max_pooling_3d.html">layer_max_pooling_3d</a>()</code> <br><code><a href="layer_maximum.html">layer_maximum</a>()</code> <br><code><a href="layer_mel_spectrogram.html">layer_mel_spectrogram</a>()</code> <br><code><a href="layer_minimum.html">layer_minimum</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_multi_head_attention.html">layer_multi_head_attention</a>()</code> <br><code><a href="layer_multiply.html">layer_multiply</a>()</code> <br><code><a href="layer_normalization.html">layer_normalization</a>()</code> <br><code><a href="layer_permute.html">layer_permute</a>()</code> <br><code><a href="layer_random_brightness.html">layer_random_brightness</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_contrast.html">layer_random_contrast</a>()</code> <br><code><a href="layer_random_crop.html">layer_random_crop</a>()</code> <br><code><a href="layer_random_elastic_transform.html">layer_random_elastic_transform</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_flip.html">layer_random_flip</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_rotation.html">layer_random_rotation</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_random_translation.html">layer_random_translation</a>()</code> <br><code><a href="layer_random_zoom.html">layer_random_zoom</a>()</code> <br><code><a href="layer_repeat_vector.html">layer_repeat_vector</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_reshape.html">layer_reshape</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_rms_normalization.html">layer_rms_normalization</a>()</code> <br><code><a href="layer_rnn.html">layer_rnn</a>()</code> <br><code><a href="layer_separable_conv_1d.html">layer_separable_conv_1d</a>()</code> <br><code><a href="layer_separable_conv_2d.html">layer_separable_conv_2d</a>()</code> <br><code><a href="layer_simple_rnn.html">layer_simple_rnn</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br><code><a href="layer_spatial_dropout_1d.html">layer_spatial_dropout_1d</a>()</code> <br><code><a href="layer_spatial_dropout_2d.html">layer_spatial_dropout_2d</a>()</code> <br><code><a href="layer_spatial_dropout_3d.html">layer_spatial_dropout_3d</a>()</code> <br><code><a href="layer_spectral_normalization.html">layer_spectral_normalization</a>()</code> <br><code><a href="layer_stft_spectrogram.html">layer_stft_spectrogram</a>()</code> <br><code><a href="layer_string_lookup.html">layer_string_lookup</a>()</code> <br><code><a href="layer_subtract.html">layer_subtract</a>()</code> <br><code><a href="layer_text_vectorization.html">layer_text_vectorization</a>()</code> <br><code><a href="layer_tfsm.html">layer_tfsm</a>()</code> <br><code><a href="layer_time_distributed.html">layer_time_distributed</a>()</code> <br><code><a href="layer_torch_module_wrapper.html">layer_torch_module_wrapper</a>()</code> <br><code><a href="layer_unit_normalization.html">layer_unit_normalization</a>()</code> <br><code><a href="layer_upsampling_1d.html">layer_upsampling_1d</a>()</code> <br><code><a href="layer_upsampling_2d.html">layer_upsampling_2d</a>()</code> <br><code><a href="layer_upsampling_3d.html">layer_upsampling_3d</a>()</code> <br><code><a href="layer_zero_padding_1d.html">layer_zero_padding_1d</a>()</code> <br><code><a href="layer_zero_padding_2d.html">layer_zero_padding_2d</a>()</code> <br><code><a href="layer_zero_padding_3d.html">layer_zero_padding_3d</a>()</code> <br><code><a href="rnn_cell_gru.html">rnn_cell_gru</a>()</code> <br><code><a href="rnn_cell_lstm.html">rnn_cell_lstm</a>()</code> <br><code><a href="rnn_cell_simple.html">rnn_cell_simple</a>()</code> <br><code><a href="rnn_cells_stack.html">rnn_cells_stack</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -144,7 +144,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/layer_random_brightness.html b/docs/dev/reference/layer_random_brightness.html
index 00f93e681..68bd17b84 100644
--- a/docs/dev/reference/layer_random_brightness.html
+++ b/docs/dev/reference/layer_random_brightness.html
@@ -16,7 +16,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -163,8 +163,8 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
     <div class="dont-index">
 <ul><li><p><a href="https://keras.io/api/layers/preprocessing_layers/image_augmentation/random_brightness#randombrightness-class" class="external-link">https://keras.io/api/layers/preprocessing_layers/image_augmentation/random_brightness#randombrightness-class</a></p></li>
 </ul><p>Other image augmentation layers: <br><code><a href="layer_random_contrast.html">layer_random_contrast</a>()</code> <br><code><a href="layer_random_crop.html">layer_random_crop</a>()</code> <br><code><a href="layer_random_flip.html">layer_random_flip</a>()</code> <br><code><a href="layer_random_rotation.html">layer_random_rotation</a>()</code> <br><code><a href="layer_random_translation.html">layer_random_translation</a>()</code> <br><code><a href="layer_random_zoom.html">layer_random_zoom</a>()</code> <br></p>
-<p>Other preprocessing layers: <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_category_encoding.html">layer_category_encoding</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_discretization.html">layer_discretization</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_feature_space.html">layer_feature_space</a>()</code> <br><code><a href="layer_hashed_crossing.html">layer_hashed_crossing</a>()</code> <br><code><a href="layer_hashing.html">layer_hashing</a>()</code> <br><code><a href="layer_integer_lookup.html">layer_integer_lookup</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_mel_spectrogram.html">layer_mel_spectrogram</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_normalization.html">layer_normalization</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_contrast.html">layer_random_contrast</a>()</code> <br><code><a href="layer_random_crop.html">layer_random_crop</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_flip.html">layer_random_flip</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_rotation.html">layer_random_rotation</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_random_translation.html">layer_random_translation</a>()</code> <br><code><a href="layer_random_zoom.html">layer_random_zoom</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br><code><a href="layer_stft_spectrogram.html">layer_stft_spectrogram</a>()</code> <br><code><a href="layer_string_lookup.html">layer_string_lookup</a>()</code> <br><code><a href="layer_text_vectorization.html">layer_text_vectorization</a>()</code> <br></p>
-<p>Other layers: <br><code><a href="Layer.html">Layer</a>()</code> <br><code><a href="layer_activation.html">layer_activation</a>()</code> <br><code><a href="layer_activation_elu.html">layer_activation_elu</a>()</code> <br><code><a href="layer_activation_leaky_relu.html">layer_activation_leaky_relu</a>()</code> <br><code><a href="layer_activation_parametric_relu.html">layer_activation_parametric_relu</a>()</code> <br><code><a href="layer_activation_relu.html">layer_activation_relu</a>()</code> <br><code><a href="layer_activation_softmax.html">layer_activation_softmax</a>()</code> <br><code><a href="layer_activity_regularization.html">layer_activity_regularization</a>()</code> <br><code><a href="layer_add.html">layer_add</a>()</code> <br><code><a href="layer_additive_attention.html">layer_additive_attention</a>()</code> <br><code><a href="layer_alpha_dropout.html">layer_alpha_dropout</a>()</code> <br><code><a href="layer_attention.html">layer_attention</a>()</code> <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_average.html">layer_average</a>()</code> <br><code><a href="layer_average_pooling_1d.html">layer_average_pooling_1d</a>()</code> <br><code><a href="layer_average_pooling_2d.html">layer_average_pooling_2d</a>()</code> <br><code><a href="layer_average_pooling_3d.html">layer_average_pooling_3d</a>()</code> <br><code><a href="layer_batch_normalization.html">layer_batch_normalization</a>()</code> <br><code><a href="layer_bidirectional.html">layer_bidirectional</a>()</code> <br><code><a href="layer_category_encoding.html">layer_category_encoding</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_concatenate.html">layer_concatenate</a>()</code> <br><code><a href="layer_conv_1d.html">layer_conv_1d</a>()</code> <br><code><a href="layer_conv_1d_transpose.html">layer_conv_1d_transpose</a>()</code> <br><code><a href="layer_conv_2d.html">layer_conv_2d</a>()</code> <br><code><a href="layer_conv_2d_transpose.html">layer_conv_2d_transpose</a>()</code> <br><code><a href="layer_conv_3d.html">layer_conv_3d</a>()</code> <br><code><a href="layer_conv_3d_transpose.html">layer_conv_3d_transpose</a>()</code> <br><code><a href="layer_conv_lstm_1d.html">layer_conv_lstm_1d</a>()</code> <br><code><a href="layer_conv_lstm_2d.html">layer_conv_lstm_2d</a>()</code> <br><code><a href="layer_conv_lstm_3d.html">layer_conv_lstm_3d</a>()</code> <br><code><a href="layer_cropping_1d.html">layer_cropping_1d</a>()</code> <br><code><a href="layer_cropping_2d.html">layer_cropping_2d</a>()</code> <br><code><a href="layer_cropping_3d.html">layer_cropping_3d</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_dense.html">layer_dense</a>()</code> <br><code><a href="layer_depthwise_conv_1d.html">layer_depthwise_conv_1d</a>()</code> <br><code><a href="layer_depthwise_conv_2d.html">layer_depthwise_conv_2d</a>()</code> <br><code><a href="layer_discretization.html">layer_discretization</a>()</code> <br><code><a href="layer_dot.html">layer_dot</a>()</code> <br><code><a href="layer_dropout.html">layer_dropout</a>()</code> <br><code><a href="layer_einsum_dense.html">layer_einsum_dense</a>()</code> <br><code><a href="layer_embedding.html">layer_embedding</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_feature_space.html">layer_feature_space</a>()</code> <br><code><a href="layer_flatten.html">layer_flatten</a>()</code> <br><code><a href="layer_flax_module_wrapper.html">layer_flax_module_wrapper</a>()</code> <br><code><a href="layer_gaussian_dropout.html">layer_gaussian_dropout</a>()</code> <br><code><a href="layer_gaussian_noise.html">layer_gaussian_noise</a>()</code> <br><code><a href="layer_global_average_pooling_1d.html">layer_global_average_pooling_1d</a>()</code> <br><code><a href="layer_global_average_pooling_2d.html">layer_global_average_pooling_2d</a>()</code> <br><code><a href="layer_global_average_pooling_3d.html">layer_global_average_pooling_3d</a>()</code> <br><code><a href="layer_global_max_pooling_1d.html">layer_global_max_pooling_1d</a>()</code> <br><code><a href="layer_global_max_pooling_2d.html">layer_global_max_pooling_2d</a>()</code> <br><code><a href="layer_global_max_pooling_3d.html">layer_global_max_pooling_3d</a>()</code> <br><code><a href="layer_group_normalization.html">layer_group_normalization</a>()</code> <br><code><a href="layer_group_query_attention.html">layer_group_query_attention</a>()</code> <br><code><a href="layer_gru.html">layer_gru</a>()</code> <br><code><a href="layer_hashed_crossing.html">layer_hashed_crossing</a>()</code> <br><code><a href="layer_hashing.html">layer_hashing</a>()</code> <br><code><a href="layer_identity.html">layer_identity</a>()</code> <br><code><a href="layer_integer_lookup.html">layer_integer_lookup</a>()</code> <br><code><a href="layer_jax_model_wrapper.html">layer_jax_model_wrapper</a>()</code> <br><code><a href="layer_lambda.html">layer_lambda</a>()</code> <br><code><a href="layer_layer_normalization.html">layer_layer_normalization</a>()</code> <br><code><a href="layer_lstm.html">layer_lstm</a>()</code> <br><code><a href="layer_masking.html">layer_masking</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_max_pooling_1d.html">layer_max_pooling_1d</a>()</code> <br><code><a href="layer_max_pooling_2d.html">layer_max_pooling_2d</a>()</code> <br><code><a href="layer_max_pooling_3d.html">layer_max_pooling_3d</a>()</code> <br><code><a href="layer_maximum.html">layer_maximum</a>()</code> <br><code><a href="layer_mel_spectrogram.html">layer_mel_spectrogram</a>()</code> <br><code><a href="layer_minimum.html">layer_minimum</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_multi_head_attention.html">layer_multi_head_attention</a>()</code> <br><code><a href="layer_multiply.html">layer_multiply</a>()</code> <br><code><a href="layer_normalization.html">layer_normalization</a>()</code> <br><code><a href="layer_permute.html">layer_permute</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_contrast.html">layer_random_contrast</a>()</code> <br><code><a href="layer_random_crop.html">layer_random_crop</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_flip.html">layer_random_flip</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_rotation.html">layer_random_rotation</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_random_translation.html">layer_random_translation</a>()</code> <br><code><a href="layer_random_zoom.html">layer_random_zoom</a>()</code> <br><code><a href="layer_repeat_vector.html">layer_repeat_vector</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_reshape.html">layer_reshape</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_rms_normalization.html">layer_rms_normalization</a>()</code> <br><code><a href="layer_rnn.html">layer_rnn</a>()</code> <br><code><a href="layer_separable_conv_1d.html">layer_separable_conv_1d</a>()</code> <br><code><a href="layer_separable_conv_2d.html">layer_separable_conv_2d</a>()</code> <br><code><a href="layer_simple_rnn.html">layer_simple_rnn</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br><code><a href="layer_spatial_dropout_1d.html">layer_spatial_dropout_1d</a>()</code> <br><code><a href="layer_spatial_dropout_2d.html">layer_spatial_dropout_2d</a>()</code> <br><code><a href="layer_spatial_dropout_3d.html">layer_spatial_dropout_3d</a>()</code> <br><code><a href="layer_spectral_normalization.html">layer_spectral_normalization</a>()</code> <br><code><a href="layer_stft_spectrogram.html">layer_stft_spectrogram</a>()</code> <br><code><a href="layer_string_lookup.html">layer_string_lookup</a>()</code> <br><code><a href="layer_subtract.html">layer_subtract</a>()</code> <br><code><a href="layer_text_vectorization.html">layer_text_vectorization</a>()</code> <br><code><a href="layer_tfsm.html">layer_tfsm</a>()</code> <br><code><a href="layer_time_distributed.html">layer_time_distributed</a>()</code> <br><code><a href="layer_torch_module_wrapper.html">layer_torch_module_wrapper</a>()</code> <br><code><a href="layer_unit_normalization.html">layer_unit_normalization</a>()</code> <br><code><a href="layer_upsampling_1d.html">layer_upsampling_1d</a>()</code> <br><code><a href="layer_upsampling_2d.html">layer_upsampling_2d</a>()</code> <br><code><a href="layer_upsampling_3d.html">layer_upsampling_3d</a>()</code> <br><code><a href="layer_zero_padding_1d.html">layer_zero_padding_1d</a>()</code> <br><code><a href="layer_zero_padding_2d.html">layer_zero_padding_2d</a>()</code> <br><code><a href="layer_zero_padding_3d.html">layer_zero_padding_3d</a>()</code> <br><code><a href="rnn_cell_gru.html">rnn_cell_gru</a>()</code> <br><code><a href="rnn_cell_lstm.html">rnn_cell_lstm</a>()</code> <br><code><a href="rnn_cell_simple.html">rnn_cell_simple</a>()</code> <br><code><a href="rnn_cells_stack.html">rnn_cells_stack</a>()</code> <br></p></div>
+<p>Other preprocessing layers: <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_category_encoding.html">layer_category_encoding</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_discretization.html">layer_discretization</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_feature_space.html">layer_feature_space</a>()</code> <br><code><a href="layer_hashed_crossing.html">layer_hashed_crossing</a>()</code> <br><code><a href="layer_hashing.html">layer_hashing</a>()</code> <br><code><a href="layer_integer_lookup.html">layer_integer_lookup</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_mel_spectrogram.html">layer_mel_spectrogram</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_normalization.html">layer_normalization</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_contrast.html">layer_random_contrast</a>()</code> <br><code><a href="layer_random_crop.html">layer_random_crop</a>()</code> <br><code><a href="layer_random_elastic_transform.html">layer_random_elastic_transform</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_flip.html">layer_random_flip</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_rotation.html">layer_random_rotation</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_random_translation.html">layer_random_translation</a>()</code> <br><code><a href="layer_random_zoom.html">layer_random_zoom</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br><code><a href="layer_stft_spectrogram.html">layer_stft_spectrogram</a>()</code> <br><code><a href="layer_string_lookup.html">layer_string_lookup</a>()</code> <br><code><a href="layer_text_vectorization.html">layer_text_vectorization</a>()</code> <br></p>
+<p>Other layers: <br><code><a href="Layer.html">Layer</a>()</code> <br><code><a href="layer_activation.html">layer_activation</a>()</code> <br><code><a href="layer_activation_elu.html">layer_activation_elu</a>()</code> <br><code><a href="layer_activation_leaky_relu.html">layer_activation_leaky_relu</a>()</code> <br><code><a href="layer_activation_parametric_relu.html">layer_activation_parametric_relu</a>()</code> <br><code><a href="layer_activation_relu.html">layer_activation_relu</a>()</code> <br><code><a href="layer_activation_softmax.html">layer_activation_softmax</a>()</code> <br><code><a href="layer_activity_regularization.html">layer_activity_regularization</a>()</code> <br><code><a href="layer_add.html">layer_add</a>()</code> <br><code><a href="layer_additive_attention.html">layer_additive_attention</a>()</code> <br><code><a href="layer_alpha_dropout.html">layer_alpha_dropout</a>()</code> <br><code><a href="layer_attention.html">layer_attention</a>()</code> <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_average.html">layer_average</a>()</code> <br><code><a href="layer_average_pooling_1d.html">layer_average_pooling_1d</a>()</code> <br><code><a href="layer_average_pooling_2d.html">layer_average_pooling_2d</a>()</code> <br><code><a href="layer_average_pooling_3d.html">layer_average_pooling_3d</a>()</code> <br><code><a href="layer_batch_normalization.html">layer_batch_normalization</a>()</code> <br><code><a href="layer_bidirectional.html">layer_bidirectional</a>()</code> <br><code><a href="layer_category_encoding.html">layer_category_encoding</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_concatenate.html">layer_concatenate</a>()</code> <br><code><a href="layer_conv_1d.html">layer_conv_1d</a>()</code> <br><code><a href="layer_conv_1d_transpose.html">layer_conv_1d_transpose</a>()</code> <br><code><a href="layer_conv_2d.html">layer_conv_2d</a>()</code> <br><code><a href="layer_conv_2d_transpose.html">layer_conv_2d_transpose</a>()</code> <br><code><a href="layer_conv_3d.html">layer_conv_3d</a>()</code> <br><code><a href="layer_conv_3d_transpose.html">layer_conv_3d_transpose</a>()</code> <br><code><a href="layer_conv_lstm_1d.html">layer_conv_lstm_1d</a>()</code> <br><code><a href="layer_conv_lstm_2d.html">layer_conv_lstm_2d</a>()</code> <br><code><a href="layer_conv_lstm_3d.html">layer_conv_lstm_3d</a>()</code> <br><code><a href="layer_cropping_1d.html">layer_cropping_1d</a>()</code> <br><code><a href="layer_cropping_2d.html">layer_cropping_2d</a>()</code> <br><code><a href="layer_cropping_3d.html">layer_cropping_3d</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_dense.html">layer_dense</a>()</code> <br><code><a href="layer_depthwise_conv_1d.html">layer_depthwise_conv_1d</a>()</code> <br><code><a href="layer_depthwise_conv_2d.html">layer_depthwise_conv_2d</a>()</code> <br><code><a href="layer_discretization.html">layer_discretization</a>()</code> <br><code><a href="layer_dot.html">layer_dot</a>()</code> <br><code><a href="layer_dropout.html">layer_dropout</a>()</code> <br><code><a href="layer_einsum_dense.html">layer_einsum_dense</a>()</code> <br><code><a href="layer_embedding.html">layer_embedding</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_feature_space.html">layer_feature_space</a>()</code> <br><code><a href="layer_flatten.html">layer_flatten</a>()</code> <br><code><a href="layer_flax_module_wrapper.html">layer_flax_module_wrapper</a>()</code> <br><code><a href="layer_gaussian_dropout.html">layer_gaussian_dropout</a>()</code> <br><code><a href="layer_gaussian_noise.html">layer_gaussian_noise</a>()</code> <br><code><a href="layer_global_average_pooling_1d.html">layer_global_average_pooling_1d</a>()</code> <br><code><a href="layer_global_average_pooling_2d.html">layer_global_average_pooling_2d</a>()</code> <br><code><a href="layer_global_average_pooling_3d.html">layer_global_average_pooling_3d</a>()</code> <br><code><a href="layer_global_max_pooling_1d.html">layer_global_max_pooling_1d</a>()</code> <br><code><a href="layer_global_max_pooling_2d.html">layer_global_max_pooling_2d</a>()</code> <br><code><a href="layer_global_max_pooling_3d.html">layer_global_max_pooling_3d</a>()</code> <br><code><a href="layer_group_normalization.html">layer_group_normalization</a>()</code> <br><code><a href="layer_group_query_attention.html">layer_group_query_attention</a>()</code> <br><code><a href="layer_gru.html">layer_gru</a>()</code> <br><code><a href="layer_hashed_crossing.html">layer_hashed_crossing</a>()</code> <br><code><a href="layer_hashing.html">layer_hashing</a>()</code> <br><code><a href="layer_identity.html">layer_identity</a>()</code> <br><code><a href="layer_integer_lookup.html">layer_integer_lookup</a>()</code> <br><code><a href="layer_jax_model_wrapper.html">layer_jax_model_wrapper</a>()</code> <br><code><a href="layer_lambda.html">layer_lambda</a>()</code> <br><code><a href="layer_layer_normalization.html">layer_layer_normalization</a>()</code> <br><code><a href="layer_lstm.html">layer_lstm</a>()</code> <br><code><a href="layer_masking.html">layer_masking</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_max_pooling_1d.html">layer_max_pooling_1d</a>()</code> <br><code><a href="layer_max_pooling_2d.html">layer_max_pooling_2d</a>()</code> <br><code><a href="layer_max_pooling_3d.html">layer_max_pooling_3d</a>()</code> <br><code><a href="layer_maximum.html">layer_maximum</a>()</code> <br><code><a href="layer_mel_spectrogram.html">layer_mel_spectrogram</a>()</code> <br><code><a href="layer_minimum.html">layer_minimum</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_multi_head_attention.html">layer_multi_head_attention</a>()</code> <br><code><a href="layer_multiply.html">layer_multiply</a>()</code> <br><code><a href="layer_normalization.html">layer_normalization</a>()</code> <br><code><a href="layer_permute.html">layer_permute</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_contrast.html">layer_random_contrast</a>()</code> <br><code><a href="layer_random_crop.html">layer_random_crop</a>()</code> <br><code><a href="layer_random_elastic_transform.html">layer_random_elastic_transform</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_flip.html">layer_random_flip</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_rotation.html">layer_random_rotation</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_random_translation.html">layer_random_translation</a>()</code> <br><code><a href="layer_random_zoom.html">layer_random_zoom</a>()</code> <br><code><a href="layer_repeat_vector.html">layer_repeat_vector</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_reshape.html">layer_reshape</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_rms_normalization.html">layer_rms_normalization</a>()</code> <br><code><a href="layer_rnn.html">layer_rnn</a>()</code> <br><code><a href="layer_separable_conv_1d.html">layer_separable_conv_1d</a>()</code> <br><code><a href="layer_separable_conv_2d.html">layer_separable_conv_2d</a>()</code> <br><code><a href="layer_simple_rnn.html">layer_simple_rnn</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br><code><a href="layer_spatial_dropout_1d.html">layer_spatial_dropout_1d</a>()</code> <br><code><a href="layer_spatial_dropout_2d.html">layer_spatial_dropout_2d</a>()</code> <br><code><a href="layer_spatial_dropout_3d.html">layer_spatial_dropout_3d</a>()</code> <br><code><a href="layer_spectral_normalization.html">layer_spectral_normalization</a>()</code> <br><code><a href="layer_stft_spectrogram.html">layer_stft_spectrogram</a>()</code> <br><code><a href="layer_string_lookup.html">layer_string_lookup</a>()</code> <br><code><a href="layer_subtract.html">layer_subtract</a>()</code> <br><code><a href="layer_text_vectorization.html">layer_text_vectorization</a>()</code> <br><code><a href="layer_tfsm.html">layer_tfsm</a>()</code> <br><code><a href="layer_time_distributed.html">layer_time_distributed</a>()</code> <br><code><a href="layer_torch_module_wrapper.html">layer_torch_module_wrapper</a>()</code> <br><code><a href="layer_unit_normalization.html">layer_unit_normalization</a>()</code> <br><code><a href="layer_upsampling_1d.html">layer_upsampling_1d</a>()</code> <br><code><a href="layer_upsampling_2d.html">layer_upsampling_2d</a>()</code> <br><code><a href="layer_upsampling_3d.html">layer_upsampling_3d</a>()</code> <br><code><a href="layer_zero_padding_1d.html">layer_zero_padding_1d</a>()</code> <br><code><a href="layer_zero_padding_2d.html">layer_zero_padding_2d</a>()</code> <br><code><a href="layer_zero_padding_3d.html">layer_zero_padding_3d</a>()</code> <br><code><a href="rnn_cell_gru.html">rnn_cell_gru</a>()</code> <br><code><a href="rnn_cell_lstm.html">rnn_cell_lstm</a>()</code> <br><code><a href="rnn_cell_simple.html">rnn_cell_simple</a>()</code> <br><code><a href="rnn_cells_stack.html">rnn_cells_stack</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -176,7 +176,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/layer_random_color_degeneration.html b/docs/dev/reference/layer_random_color_degeneration.html
index 64605ecbc..9488ddfbf 100644
--- a/docs/dev/reference/layer_random_color_degeneration.html
+++ b/docs/dev/reference/layer_random_color_degeneration.html
@@ -12,7 +12,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -122,9 +122,9 @@ <h2 id="arguments">Arguments<a class="anchor" aria-label="anchor" href="#argumen
 </dl></div>
     <div class="section level2">
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
-    <div class="dont-index"><p>Other image preprocessing layers: <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br></p>
-<p>Other preprocessing layers: <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_category_encoding.html">layer_category_encoding</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_discretization.html">layer_discretization</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_feature_space.html">layer_feature_space</a>()</code> <br><code><a href="layer_hashed_crossing.html">layer_hashed_crossing</a>()</code> <br><code><a href="layer_hashing.html">layer_hashing</a>()</code> <br><code><a href="layer_integer_lookup.html">layer_integer_lookup</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_mel_spectrogram.html">layer_mel_spectrogram</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_normalization.html">layer_normalization</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_brightness.html">layer_random_brightness</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_contrast.html">layer_random_contrast</a>()</code> <br><code><a href="layer_random_crop.html">layer_random_crop</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_flip.html">layer_random_flip</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_rotation.html">layer_random_rotation</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_random_translation.html">layer_random_translation</a>()</code> <br><code><a href="layer_random_zoom.html">layer_random_zoom</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br><code><a href="layer_stft_spectrogram.html">layer_stft_spectrogram</a>()</code> <br><code><a href="layer_string_lookup.html">layer_string_lookup</a>()</code> <br><code><a href="layer_text_vectorization.html">layer_text_vectorization</a>()</code> <br></p>
-<p>Other layers: <br><code><a href="Layer.html">Layer</a>()</code> <br><code><a href="layer_activation.html">layer_activation</a>()</code> <br><code><a href="layer_activation_elu.html">layer_activation_elu</a>()</code> <br><code><a href="layer_activation_leaky_relu.html">layer_activation_leaky_relu</a>()</code> <br><code><a href="layer_activation_parametric_relu.html">layer_activation_parametric_relu</a>()</code> <br><code><a href="layer_activation_relu.html">layer_activation_relu</a>()</code> <br><code><a href="layer_activation_softmax.html">layer_activation_softmax</a>()</code> <br><code><a href="layer_activity_regularization.html">layer_activity_regularization</a>()</code> <br><code><a href="layer_add.html">layer_add</a>()</code> <br><code><a href="layer_additive_attention.html">layer_additive_attention</a>()</code> <br><code><a href="layer_alpha_dropout.html">layer_alpha_dropout</a>()</code> <br><code><a href="layer_attention.html">layer_attention</a>()</code> <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_average.html">layer_average</a>()</code> <br><code><a href="layer_average_pooling_1d.html">layer_average_pooling_1d</a>()</code> <br><code><a href="layer_average_pooling_2d.html">layer_average_pooling_2d</a>()</code> <br><code><a href="layer_average_pooling_3d.html">layer_average_pooling_3d</a>()</code> <br><code><a href="layer_batch_normalization.html">layer_batch_normalization</a>()</code> <br><code><a href="layer_bidirectional.html">layer_bidirectional</a>()</code> <br><code><a href="layer_category_encoding.html">layer_category_encoding</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_concatenate.html">layer_concatenate</a>()</code> <br><code><a href="layer_conv_1d.html">layer_conv_1d</a>()</code> <br><code><a href="layer_conv_1d_transpose.html">layer_conv_1d_transpose</a>()</code> <br><code><a href="layer_conv_2d.html">layer_conv_2d</a>()</code> <br><code><a href="layer_conv_2d_transpose.html">layer_conv_2d_transpose</a>()</code> <br><code><a href="layer_conv_3d.html">layer_conv_3d</a>()</code> <br><code><a href="layer_conv_3d_transpose.html">layer_conv_3d_transpose</a>()</code> <br><code><a href="layer_conv_lstm_1d.html">layer_conv_lstm_1d</a>()</code> <br><code><a href="layer_conv_lstm_2d.html">layer_conv_lstm_2d</a>()</code> <br><code><a href="layer_conv_lstm_3d.html">layer_conv_lstm_3d</a>()</code> <br><code><a href="layer_cropping_1d.html">layer_cropping_1d</a>()</code> <br><code><a href="layer_cropping_2d.html">layer_cropping_2d</a>()</code> <br><code><a href="layer_cropping_3d.html">layer_cropping_3d</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_dense.html">layer_dense</a>()</code> <br><code><a href="layer_depthwise_conv_1d.html">layer_depthwise_conv_1d</a>()</code> <br><code><a href="layer_depthwise_conv_2d.html">layer_depthwise_conv_2d</a>()</code> <br><code><a href="layer_discretization.html">layer_discretization</a>()</code> <br><code><a href="layer_dot.html">layer_dot</a>()</code> <br><code><a href="layer_dropout.html">layer_dropout</a>()</code> <br><code><a href="layer_einsum_dense.html">layer_einsum_dense</a>()</code> <br><code><a href="layer_embedding.html">layer_embedding</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_feature_space.html">layer_feature_space</a>()</code> <br><code><a href="layer_flatten.html">layer_flatten</a>()</code> <br><code><a href="layer_flax_module_wrapper.html">layer_flax_module_wrapper</a>()</code> <br><code><a href="layer_gaussian_dropout.html">layer_gaussian_dropout</a>()</code> <br><code><a href="layer_gaussian_noise.html">layer_gaussian_noise</a>()</code> <br><code><a href="layer_global_average_pooling_1d.html">layer_global_average_pooling_1d</a>()</code> <br><code><a href="layer_global_average_pooling_2d.html">layer_global_average_pooling_2d</a>()</code> <br><code><a href="layer_global_average_pooling_3d.html">layer_global_average_pooling_3d</a>()</code> <br><code><a href="layer_global_max_pooling_1d.html">layer_global_max_pooling_1d</a>()</code> <br><code><a href="layer_global_max_pooling_2d.html">layer_global_max_pooling_2d</a>()</code> <br><code><a href="layer_global_max_pooling_3d.html">layer_global_max_pooling_3d</a>()</code> <br><code><a href="layer_group_normalization.html">layer_group_normalization</a>()</code> <br><code><a href="layer_group_query_attention.html">layer_group_query_attention</a>()</code> <br><code><a href="layer_gru.html">layer_gru</a>()</code> <br><code><a href="layer_hashed_crossing.html">layer_hashed_crossing</a>()</code> <br><code><a href="layer_hashing.html">layer_hashing</a>()</code> <br><code><a href="layer_identity.html">layer_identity</a>()</code> <br><code><a href="layer_integer_lookup.html">layer_integer_lookup</a>()</code> <br><code><a href="layer_jax_model_wrapper.html">layer_jax_model_wrapper</a>()</code> <br><code><a href="layer_lambda.html">layer_lambda</a>()</code> <br><code><a href="layer_layer_normalization.html">layer_layer_normalization</a>()</code> <br><code><a href="layer_lstm.html">layer_lstm</a>()</code> <br><code><a href="layer_masking.html">layer_masking</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_max_pooling_1d.html">layer_max_pooling_1d</a>()</code> <br><code><a href="layer_max_pooling_2d.html">layer_max_pooling_2d</a>()</code> <br><code><a href="layer_max_pooling_3d.html">layer_max_pooling_3d</a>()</code> <br><code><a href="layer_maximum.html">layer_maximum</a>()</code> <br><code><a href="layer_mel_spectrogram.html">layer_mel_spectrogram</a>()</code> <br><code><a href="layer_minimum.html">layer_minimum</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_multi_head_attention.html">layer_multi_head_attention</a>()</code> <br><code><a href="layer_multiply.html">layer_multiply</a>()</code> <br><code><a href="layer_normalization.html">layer_normalization</a>()</code> <br><code><a href="layer_permute.html">layer_permute</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_brightness.html">layer_random_brightness</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_contrast.html">layer_random_contrast</a>()</code> <br><code><a href="layer_random_crop.html">layer_random_crop</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_flip.html">layer_random_flip</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_rotation.html">layer_random_rotation</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_random_translation.html">layer_random_translation</a>()</code> <br><code><a href="layer_random_zoom.html">layer_random_zoom</a>()</code> <br><code><a href="layer_repeat_vector.html">layer_repeat_vector</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_reshape.html">layer_reshape</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_rms_normalization.html">layer_rms_normalization</a>()</code> <br><code><a href="layer_rnn.html">layer_rnn</a>()</code> <br><code><a href="layer_separable_conv_1d.html">layer_separable_conv_1d</a>()</code> <br><code><a href="layer_separable_conv_2d.html">layer_separable_conv_2d</a>()</code> <br><code><a href="layer_simple_rnn.html">layer_simple_rnn</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br><code><a href="layer_spatial_dropout_1d.html">layer_spatial_dropout_1d</a>()</code> <br><code><a href="layer_spatial_dropout_2d.html">layer_spatial_dropout_2d</a>()</code> <br><code><a href="layer_spatial_dropout_3d.html">layer_spatial_dropout_3d</a>()</code> <br><code><a href="layer_spectral_normalization.html">layer_spectral_normalization</a>()</code> <br><code><a href="layer_stft_spectrogram.html">layer_stft_spectrogram</a>()</code> <br><code><a href="layer_string_lookup.html">layer_string_lookup</a>()</code> <br><code><a href="layer_subtract.html">layer_subtract</a>()</code> <br><code><a href="layer_text_vectorization.html">layer_text_vectorization</a>()</code> <br><code><a href="layer_tfsm.html">layer_tfsm</a>()</code> <br><code><a href="layer_time_distributed.html">layer_time_distributed</a>()</code> <br><code><a href="layer_torch_module_wrapper.html">layer_torch_module_wrapper</a>()</code> <br><code><a href="layer_unit_normalization.html">layer_unit_normalization</a>()</code> <br><code><a href="layer_upsampling_1d.html">layer_upsampling_1d</a>()</code> <br><code><a href="layer_upsampling_2d.html">layer_upsampling_2d</a>()</code> <br><code><a href="layer_upsampling_3d.html">layer_upsampling_3d</a>()</code> <br><code><a href="layer_zero_padding_1d.html">layer_zero_padding_1d</a>()</code> <br><code><a href="layer_zero_padding_2d.html">layer_zero_padding_2d</a>()</code> <br><code><a href="layer_zero_padding_3d.html">layer_zero_padding_3d</a>()</code> <br><code><a href="rnn_cell_gru.html">rnn_cell_gru</a>()</code> <br><code><a href="rnn_cell_lstm.html">rnn_cell_lstm</a>()</code> <br><code><a href="rnn_cell_simple.html">rnn_cell_simple</a>()</code> <br><code><a href="rnn_cells_stack.html">rnn_cells_stack</a>()</code> <br></p></div>
+    <div class="dont-index"><p>Other image preprocessing layers: <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_elastic_transform.html">layer_random_elastic_transform</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br></p>
+<p>Other preprocessing layers: <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_category_encoding.html">layer_category_encoding</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_discretization.html">layer_discretization</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_feature_space.html">layer_feature_space</a>()</code> <br><code><a href="layer_hashed_crossing.html">layer_hashed_crossing</a>()</code> <br><code><a href="layer_hashing.html">layer_hashing</a>()</code> <br><code><a href="layer_integer_lookup.html">layer_integer_lookup</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_mel_spectrogram.html">layer_mel_spectrogram</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_normalization.html">layer_normalization</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_brightness.html">layer_random_brightness</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_contrast.html">layer_random_contrast</a>()</code> <br><code><a href="layer_random_crop.html">layer_random_crop</a>()</code> <br><code><a href="layer_random_elastic_transform.html">layer_random_elastic_transform</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_flip.html">layer_random_flip</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_rotation.html">layer_random_rotation</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_random_translation.html">layer_random_translation</a>()</code> <br><code><a href="layer_random_zoom.html">layer_random_zoom</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br><code><a href="layer_stft_spectrogram.html">layer_stft_spectrogram</a>()</code> <br><code><a href="layer_string_lookup.html">layer_string_lookup</a>()</code> <br><code><a href="layer_text_vectorization.html">layer_text_vectorization</a>()</code> <br></p>
+<p>Other layers: <br><code><a href="Layer.html">Layer</a>()</code> <br><code><a href="layer_activation.html">layer_activation</a>()</code> <br><code><a href="layer_activation_elu.html">layer_activation_elu</a>()</code> <br><code><a href="layer_activation_leaky_relu.html">layer_activation_leaky_relu</a>()</code> <br><code><a href="layer_activation_parametric_relu.html">layer_activation_parametric_relu</a>()</code> <br><code><a href="layer_activation_relu.html">layer_activation_relu</a>()</code> <br><code><a href="layer_activation_softmax.html">layer_activation_softmax</a>()</code> <br><code><a href="layer_activity_regularization.html">layer_activity_regularization</a>()</code> <br><code><a href="layer_add.html">layer_add</a>()</code> <br><code><a href="layer_additive_attention.html">layer_additive_attention</a>()</code> <br><code><a href="layer_alpha_dropout.html">layer_alpha_dropout</a>()</code> <br><code><a href="layer_attention.html">layer_attention</a>()</code> <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_average.html">layer_average</a>()</code> <br><code><a href="layer_average_pooling_1d.html">layer_average_pooling_1d</a>()</code> <br><code><a href="layer_average_pooling_2d.html">layer_average_pooling_2d</a>()</code> <br><code><a href="layer_average_pooling_3d.html">layer_average_pooling_3d</a>()</code> <br><code><a href="layer_batch_normalization.html">layer_batch_normalization</a>()</code> <br><code><a href="layer_bidirectional.html">layer_bidirectional</a>()</code> <br><code><a href="layer_category_encoding.html">layer_category_encoding</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_concatenate.html">layer_concatenate</a>()</code> <br><code><a href="layer_conv_1d.html">layer_conv_1d</a>()</code> <br><code><a href="layer_conv_1d_transpose.html">layer_conv_1d_transpose</a>()</code> <br><code><a href="layer_conv_2d.html">layer_conv_2d</a>()</code> <br><code><a href="layer_conv_2d_transpose.html">layer_conv_2d_transpose</a>()</code> <br><code><a href="layer_conv_3d.html">layer_conv_3d</a>()</code> <br><code><a href="layer_conv_3d_transpose.html">layer_conv_3d_transpose</a>()</code> <br><code><a href="layer_conv_lstm_1d.html">layer_conv_lstm_1d</a>()</code> <br><code><a href="layer_conv_lstm_2d.html">layer_conv_lstm_2d</a>()</code> <br><code><a href="layer_conv_lstm_3d.html">layer_conv_lstm_3d</a>()</code> <br><code><a href="layer_cropping_1d.html">layer_cropping_1d</a>()</code> <br><code><a href="layer_cropping_2d.html">layer_cropping_2d</a>()</code> <br><code><a href="layer_cropping_3d.html">layer_cropping_3d</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_dense.html">layer_dense</a>()</code> <br><code><a href="layer_depthwise_conv_1d.html">layer_depthwise_conv_1d</a>()</code> <br><code><a href="layer_depthwise_conv_2d.html">layer_depthwise_conv_2d</a>()</code> <br><code><a href="layer_discretization.html">layer_discretization</a>()</code> <br><code><a href="layer_dot.html">layer_dot</a>()</code> <br><code><a href="layer_dropout.html">layer_dropout</a>()</code> <br><code><a href="layer_einsum_dense.html">layer_einsum_dense</a>()</code> <br><code><a href="layer_embedding.html">layer_embedding</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_feature_space.html">layer_feature_space</a>()</code> <br><code><a href="layer_flatten.html">layer_flatten</a>()</code> <br><code><a href="layer_flax_module_wrapper.html">layer_flax_module_wrapper</a>()</code> <br><code><a href="layer_gaussian_dropout.html">layer_gaussian_dropout</a>()</code> <br><code><a href="layer_gaussian_noise.html">layer_gaussian_noise</a>()</code> <br><code><a href="layer_global_average_pooling_1d.html">layer_global_average_pooling_1d</a>()</code> <br><code><a href="layer_global_average_pooling_2d.html">layer_global_average_pooling_2d</a>()</code> <br><code><a href="layer_global_average_pooling_3d.html">layer_global_average_pooling_3d</a>()</code> <br><code><a href="layer_global_max_pooling_1d.html">layer_global_max_pooling_1d</a>()</code> <br><code><a href="layer_global_max_pooling_2d.html">layer_global_max_pooling_2d</a>()</code> <br><code><a href="layer_global_max_pooling_3d.html">layer_global_max_pooling_3d</a>()</code> <br><code><a href="layer_group_normalization.html">layer_group_normalization</a>()</code> <br><code><a href="layer_group_query_attention.html">layer_group_query_attention</a>()</code> <br><code><a href="layer_gru.html">layer_gru</a>()</code> <br><code><a href="layer_hashed_crossing.html">layer_hashed_crossing</a>()</code> <br><code><a href="layer_hashing.html">layer_hashing</a>()</code> <br><code><a href="layer_identity.html">layer_identity</a>()</code> <br><code><a href="layer_integer_lookup.html">layer_integer_lookup</a>()</code> <br><code><a href="layer_jax_model_wrapper.html">layer_jax_model_wrapper</a>()</code> <br><code><a href="layer_lambda.html">layer_lambda</a>()</code> <br><code><a href="layer_layer_normalization.html">layer_layer_normalization</a>()</code> <br><code><a href="layer_lstm.html">layer_lstm</a>()</code> <br><code><a href="layer_masking.html">layer_masking</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_max_pooling_1d.html">layer_max_pooling_1d</a>()</code> <br><code><a href="layer_max_pooling_2d.html">layer_max_pooling_2d</a>()</code> <br><code><a href="layer_max_pooling_3d.html">layer_max_pooling_3d</a>()</code> <br><code><a href="layer_maximum.html">layer_maximum</a>()</code> <br><code><a href="layer_mel_spectrogram.html">layer_mel_spectrogram</a>()</code> <br><code><a href="layer_minimum.html">layer_minimum</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_multi_head_attention.html">layer_multi_head_attention</a>()</code> <br><code><a href="layer_multiply.html">layer_multiply</a>()</code> <br><code><a href="layer_normalization.html">layer_normalization</a>()</code> <br><code><a href="layer_permute.html">layer_permute</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_brightness.html">layer_random_brightness</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_contrast.html">layer_random_contrast</a>()</code> <br><code><a href="layer_random_crop.html">layer_random_crop</a>()</code> <br><code><a href="layer_random_elastic_transform.html">layer_random_elastic_transform</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_flip.html">layer_random_flip</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_rotation.html">layer_random_rotation</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_random_translation.html">layer_random_translation</a>()</code> <br><code><a href="layer_random_zoom.html">layer_random_zoom</a>()</code> <br><code><a href="layer_repeat_vector.html">layer_repeat_vector</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_reshape.html">layer_reshape</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_rms_normalization.html">layer_rms_normalization</a>()</code> <br><code><a href="layer_rnn.html">layer_rnn</a>()</code> <br><code><a href="layer_separable_conv_1d.html">layer_separable_conv_1d</a>()</code> <br><code><a href="layer_separable_conv_2d.html">layer_separable_conv_2d</a>()</code> <br><code><a href="layer_simple_rnn.html">layer_simple_rnn</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br><code><a href="layer_spatial_dropout_1d.html">layer_spatial_dropout_1d</a>()</code> <br><code><a href="layer_spatial_dropout_2d.html">layer_spatial_dropout_2d</a>()</code> <br><code><a href="layer_spatial_dropout_3d.html">layer_spatial_dropout_3d</a>()</code> <br><code><a href="layer_spectral_normalization.html">layer_spectral_normalization</a>()</code> <br><code><a href="layer_stft_spectrogram.html">layer_stft_spectrogram</a>()</code> <br><code><a href="layer_string_lookup.html">layer_string_lookup</a>()</code> <br><code><a href="layer_subtract.html">layer_subtract</a>()</code> <br><code><a href="layer_text_vectorization.html">layer_text_vectorization</a>()</code> <br><code><a href="layer_tfsm.html">layer_tfsm</a>()</code> <br><code><a href="layer_time_distributed.html">layer_time_distributed</a>()</code> <br><code><a href="layer_torch_module_wrapper.html">layer_torch_module_wrapper</a>()</code> <br><code><a href="layer_unit_normalization.html">layer_unit_normalization</a>()</code> <br><code><a href="layer_upsampling_1d.html">layer_upsampling_1d</a>()</code> <br><code><a href="layer_upsampling_2d.html">layer_upsampling_2d</a>()</code> <br><code><a href="layer_upsampling_3d.html">layer_upsampling_3d</a>()</code> <br><code><a href="layer_zero_padding_1d.html">layer_zero_padding_1d</a>()</code> <br><code><a href="layer_zero_padding_2d.html">layer_zero_padding_2d</a>()</code> <br><code><a href="layer_zero_padding_3d.html">layer_zero_padding_3d</a>()</code> <br><code><a href="rnn_cell_gru.html">rnn_cell_gru</a>()</code> <br><code><a href="rnn_cell_lstm.html">rnn_cell_lstm</a>()</code> <br><code><a href="rnn_cell_simple.html">rnn_cell_simple</a>()</code> <br><code><a href="rnn_cells_stack.html">rnn_cells_stack</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -136,7 +136,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/layer_random_color_jitter.html b/docs/dev/reference/layer_random_color_jitter.html
index bdbc5d6b2..667f74671 100644
--- a/docs/dev/reference/layer_random_color_jitter.html
+++ b/docs/dev/reference/layer_random_color_jitter.html
@@ -10,7 +10,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -155,9 +155,9 @@ <h2 id="arguments">Arguments<a class="anchor" aria-label="anchor" href="#argumen
 </dl></div>
     <div class="section level2">
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
-    <div class="dont-index"><p>Other image preprocessing layers: <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br></p>
-<p>Other preprocessing layers: <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_category_encoding.html">layer_category_encoding</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_discretization.html">layer_discretization</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_feature_space.html">layer_feature_space</a>()</code> <br><code><a href="layer_hashed_crossing.html">layer_hashed_crossing</a>()</code> <br><code><a href="layer_hashing.html">layer_hashing</a>()</code> <br><code><a href="layer_integer_lookup.html">layer_integer_lookup</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_mel_spectrogram.html">layer_mel_spectrogram</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_normalization.html">layer_normalization</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_brightness.html">layer_random_brightness</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_contrast.html">layer_random_contrast</a>()</code> <br><code><a href="layer_random_crop.html">layer_random_crop</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_flip.html">layer_random_flip</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_rotation.html">layer_random_rotation</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_random_translation.html">layer_random_translation</a>()</code> <br><code><a href="layer_random_zoom.html">layer_random_zoom</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br><code><a href="layer_stft_spectrogram.html">layer_stft_spectrogram</a>()</code> <br><code><a href="layer_string_lookup.html">layer_string_lookup</a>()</code> <br><code><a href="layer_text_vectorization.html">layer_text_vectorization</a>()</code> <br></p>
-<p>Other layers: <br><code><a href="Layer.html">Layer</a>()</code> <br><code><a href="layer_activation.html">layer_activation</a>()</code> <br><code><a href="layer_activation_elu.html">layer_activation_elu</a>()</code> <br><code><a href="layer_activation_leaky_relu.html">layer_activation_leaky_relu</a>()</code> <br><code><a href="layer_activation_parametric_relu.html">layer_activation_parametric_relu</a>()</code> <br><code><a href="layer_activation_relu.html">layer_activation_relu</a>()</code> <br><code><a href="layer_activation_softmax.html">layer_activation_softmax</a>()</code> <br><code><a href="layer_activity_regularization.html">layer_activity_regularization</a>()</code> <br><code><a href="layer_add.html">layer_add</a>()</code> <br><code><a href="layer_additive_attention.html">layer_additive_attention</a>()</code> <br><code><a href="layer_alpha_dropout.html">layer_alpha_dropout</a>()</code> <br><code><a href="layer_attention.html">layer_attention</a>()</code> <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_average.html">layer_average</a>()</code> <br><code><a href="layer_average_pooling_1d.html">layer_average_pooling_1d</a>()</code> <br><code><a href="layer_average_pooling_2d.html">layer_average_pooling_2d</a>()</code> <br><code><a href="layer_average_pooling_3d.html">layer_average_pooling_3d</a>()</code> <br><code><a href="layer_batch_normalization.html">layer_batch_normalization</a>()</code> <br><code><a href="layer_bidirectional.html">layer_bidirectional</a>()</code> <br><code><a href="layer_category_encoding.html">layer_category_encoding</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_concatenate.html">layer_concatenate</a>()</code> <br><code><a href="layer_conv_1d.html">layer_conv_1d</a>()</code> <br><code><a href="layer_conv_1d_transpose.html">layer_conv_1d_transpose</a>()</code> <br><code><a href="layer_conv_2d.html">layer_conv_2d</a>()</code> <br><code><a href="layer_conv_2d_transpose.html">layer_conv_2d_transpose</a>()</code> <br><code><a href="layer_conv_3d.html">layer_conv_3d</a>()</code> <br><code><a href="layer_conv_3d_transpose.html">layer_conv_3d_transpose</a>()</code> <br><code><a href="layer_conv_lstm_1d.html">layer_conv_lstm_1d</a>()</code> <br><code><a href="layer_conv_lstm_2d.html">layer_conv_lstm_2d</a>()</code> <br><code><a href="layer_conv_lstm_3d.html">layer_conv_lstm_3d</a>()</code> <br><code><a href="layer_cropping_1d.html">layer_cropping_1d</a>()</code> <br><code><a href="layer_cropping_2d.html">layer_cropping_2d</a>()</code> <br><code><a href="layer_cropping_3d.html">layer_cropping_3d</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_dense.html">layer_dense</a>()</code> <br><code><a href="layer_depthwise_conv_1d.html">layer_depthwise_conv_1d</a>()</code> <br><code><a href="layer_depthwise_conv_2d.html">layer_depthwise_conv_2d</a>()</code> <br><code><a href="layer_discretization.html">layer_discretization</a>()</code> <br><code><a href="layer_dot.html">layer_dot</a>()</code> <br><code><a href="layer_dropout.html">layer_dropout</a>()</code> <br><code><a href="layer_einsum_dense.html">layer_einsum_dense</a>()</code> <br><code><a href="layer_embedding.html">layer_embedding</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_feature_space.html">layer_feature_space</a>()</code> <br><code><a href="layer_flatten.html">layer_flatten</a>()</code> <br><code><a href="layer_flax_module_wrapper.html">layer_flax_module_wrapper</a>()</code> <br><code><a href="layer_gaussian_dropout.html">layer_gaussian_dropout</a>()</code> <br><code><a href="layer_gaussian_noise.html">layer_gaussian_noise</a>()</code> <br><code><a href="layer_global_average_pooling_1d.html">layer_global_average_pooling_1d</a>()</code> <br><code><a href="layer_global_average_pooling_2d.html">layer_global_average_pooling_2d</a>()</code> <br><code><a href="layer_global_average_pooling_3d.html">layer_global_average_pooling_3d</a>()</code> <br><code><a href="layer_global_max_pooling_1d.html">layer_global_max_pooling_1d</a>()</code> <br><code><a href="layer_global_max_pooling_2d.html">layer_global_max_pooling_2d</a>()</code> <br><code><a href="layer_global_max_pooling_3d.html">layer_global_max_pooling_3d</a>()</code> <br><code><a href="layer_group_normalization.html">layer_group_normalization</a>()</code> <br><code><a href="layer_group_query_attention.html">layer_group_query_attention</a>()</code> <br><code><a href="layer_gru.html">layer_gru</a>()</code> <br><code><a href="layer_hashed_crossing.html">layer_hashed_crossing</a>()</code> <br><code><a href="layer_hashing.html">layer_hashing</a>()</code> <br><code><a href="layer_identity.html">layer_identity</a>()</code> <br><code><a href="layer_integer_lookup.html">layer_integer_lookup</a>()</code> <br><code><a href="layer_jax_model_wrapper.html">layer_jax_model_wrapper</a>()</code> <br><code><a href="layer_lambda.html">layer_lambda</a>()</code> <br><code><a href="layer_layer_normalization.html">layer_layer_normalization</a>()</code> <br><code><a href="layer_lstm.html">layer_lstm</a>()</code> <br><code><a href="layer_masking.html">layer_masking</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_max_pooling_1d.html">layer_max_pooling_1d</a>()</code> <br><code><a href="layer_max_pooling_2d.html">layer_max_pooling_2d</a>()</code> <br><code><a href="layer_max_pooling_3d.html">layer_max_pooling_3d</a>()</code> <br><code><a href="layer_maximum.html">layer_maximum</a>()</code> <br><code><a href="layer_mel_spectrogram.html">layer_mel_spectrogram</a>()</code> <br><code><a href="layer_minimum.html">layer_minimum</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_multi_head_attention.html">layer_multi_head_attention</a>()</code> <br><code><a href="layer_multiply.html">layer_multiply</a>()</code> <br><code><a href="layer_normalization.html">layer_normalization</a>()</code> <br><code><a href="layer_permute.html">layer_permute</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_brightness.html">layer_random_brightness</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_contrast.html">layer_random_contrast</a>()</code> <br><code><a href="layer_random_crop.html">layer_random_crop</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_flip.html">layer_random_flip</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_rotation.html">layer_random_rotation</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_random_translation.html">layer_random_translation</a>()</code> <br><code><a href="layer_random_zoom.html">layer_random_zoom</a>()</code> <br><code><a href="layer_repeat_vector.html">layer_repeat_vector</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_reshape.html">layer_reshape</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_rms_normalization.html">layer_rms_normalization</a>()</code> <br><code><a href="layer_rnn.html">layer_rnn</a>()</code> <br><code><a href="layer_separable_conv_1d.html">layer_separable_conv_1d</a>()</code> <br><code><a href="layer_separable_conv_2d.html">layer_separable_conv_2d</a>()</code> <br><code><a href="layer_simple_rnn.html">layer_simple_rnn</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br><code><a href="layer_spatial_dropout_1d.html">layer_spatial_dropout_1d</a>()</code> <br><code><a href="layer_spatial_dropout_2d.html">layer_spatial_dropout_2d</a>()</code> <br><code><a href="layer_spatial_dropout_3d.html">layer_spatial_dropout_3d</a>()</code> <br><code><a href="layer_spectral_normalization.html">layer_spectral_normalization</a>()</code> <br><code><a href="layer_stft_spectrogram.html">layer_stft_spectrogram</a>()</code> <br><code><a href="layer_string_lookup.html">layer_string_lookup</a>()</code> <br><code><a href="layer_subtract.html">layer_subtract</a>()</code> <br><code><a href="layer_text_vectorization.html">layer_text_vectorization</a>()</code> <br><code><a href="layer_tfsm.html">layer_tfsm</a>()</code> <br><code><a href="layer_time_distributed.html">layer_time_distributed</a>()</code> <br><code><a href="layer_torch_module_wrapper.html">layer_torch_module_wrapper</a>()</code> <br><code><a href="layer_unit_normalization.html">layer_unit_normalization</a>()</code> <br><code><a href="layer_upsampling_1d.html">layer_upsampling_1d</a>()</code> <br><code><a href="layer_upsampling_2d.html">layer_upsampling_2d</a>()</code> <br><code><a href="layer_upsampling_3d.html">layer_upsampling_3d</a>()</code> <br><code><a href="layer_zero_padding_1d.html">layer_zero_padding_1d</a>()</code> <br><code><a href="layer_zero_padding_2d.html">layer_zero_padding_2d</a>()</code> <br><code><a href="layer_zero_padding_3d.html">layer_zero_padding_3d</a>()</code> <br><code><a href="rnn_cell_gru.html">rnn_cell_gru</a>()</code> <br><code><a href="rnn_cell_lstm.html">rnn_cell_lstm</a>()</code> <br><code><a href="rnn_cell_simple.html">rnn_cell_simple</a>()</code> <br><code><a href="rnn_cells_stack.html">rnn_cells_stack</a>()</code> <br></p></div>
+    <div class="dont-index"><p>Other image preprocessing layers: <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_elastic_transform.html">layer_random_elastic_transform</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br></p>
+<p>Other preprocessing layers: <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_category_encoding.html">layer_category_encoding</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_discretization.html">layer_discretization</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_feature_space.html">layer_feature_space</a>()</code> <br><code><a href="layer_hashed_crossing.html">layer_hashed_crossing</a>()</code> <br><code><a href="layer_hashing.html">layer_hashing</a>()</code> <br><code><a href="layer_integer_lookup.html">layer_integer_lookup</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_mel_spectrogram.html">layer_mel_spectrogram</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_normalization.html">layer_normalization</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_brightness.html">layer_random_brightness</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_contrast.html">layer_random_contrast</a>()</code> <br><code><a href="layer_random_crop.html">layer_random_crop</a>()</code> <br><code><a href="layer_random_elastic_transform.html">layer_random_elastic_transform</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_flip.html">layer_random_flip</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_rotation.html">layer_random_rotation</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_random_translation.html">layer_random_translation</a>()</code> <br><code><a href="layer_random_zoom.html">layer_random_zoom</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br><code><a href="layer_stft_spectrogram.html">layer_stft_spectrogram</a>()</code> <br><code><a href="layer_string_lookup.html">layer_string_lookup</a>()</code> <br><code><a href="layer_text_vectorization.html">layer_text_vectorization</a>()</code> <br></p>
+<p>Other layers: <br><code><a href="Layer.html">Layer</a>()</code> <br><code><a href="layer_activation.html">layer_activation</a>()</code> <br><code><a href="layer_activation_elu.html">layer_activation_elu</a>()</code> <br><code><a href="layer_activation_leaky_relu.html">layer_activation_leaky_relu</a>()</code> <br><code><a href="layer_activation_parametric_relu.html">layer_activation_parametric_relu</a>()</code> <br><code><a href="layer_activation_relu.html">layer_activation_relu</a>()</code> <br><code><a href="layer_activation_softmax.html">layer_activation_softmax</a>()</code> <br><code><a href="layer_activity_regularization.html">layer_activity_regularization</a>()</code> <br><code><a href="layer_add.html">layer_add</a>()</code> <br><code><a href="layer_additive_attention.html">layer_additive_attention</a>()</code> <br><code><a href="layer_alpha_dropout.html">layer_alpha_dropout</a>()</code> <br><code><a href="layer_attention.html">layer_attention</a>()</code> <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_average.html">layer_average</a>()</code> <br><code><a href="layer_average_pooling_1d.html">layer_average_pooling_1d</a>()</code> <br><code><a href="layer_average_pooling_2d.html">layer_average_pooling_2d</a>()</code> <br><code><a href="layer_average_pooling_3d.html">layer_average_pooling_3d</a>()</code> <br><code><a href="layer_batch_normalization.html">layer_batch_normalization</a>()</code> <br><code><a href="layer_bidirectional.html">layer_bidirectional</a>()</code> <br><code><a href="layer_category_encoding.html">layer_category_encoding</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_concatenate.html">layer_concatenate</a>()</code> <br><code><a href="layer_conv_1d.html">layer_conv_1d</a>()</code> <br><code><a href="layer_conv_1d_transpose.html">layer_conv_1d_transpose</a>()</code> <br><code><a href="layer_conv_2d.html">layer_conv_2d</a>()</code> <br><code><a href="layer_conv_2d_transpose.html">layer_conv_2d_transpose</a>()</code> <br><code><a href="layer_conv_3d.html">layer_conv_3d</a>()</code> <br><code><a href="layer_conv_3d_transpose.html">layer_conv_3d_transpose</a>()</code> <br><code><a href="layer_conv_lstm_1d.html">layer_conv_lstm_1d</a>()</code> <br><code><a href="layer_conv_lstm_2d.html">layer_conv_lstm_2d</a>()</code> <br><code><a href="layer_conv_lstm_3d.html">layer_conv_lstm_3d</a>()</code> <br><code><a href="layer_cropping_1d.html">layer_cropping_1d</a>()</code> <br><code><a href="layer_cropping_2d.html">layer_cropping_2d</a>()</code> <br><code><a href="layer_cropping_3d.html">layer_cropping_3d</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_dense.html">layer_dense</a>()</code> <br><code><a href="layer_depthwise_conv_1d.html">layer_depthwise_conv_1d</a>()</code> <br><code><a href="layer_depthwise_conv_2d.html">layer_depthwise_conv_2d</a>()</code> <br><code><a href="layer_discretization.html">layer_discretization</a>()</code> <br><code><a href="layer_dot.html">layer_dot</a>()</code> <br><code><a href="layer_dropout.html">layer_dropout</a>()</code> <br><code><a href="layer_einsum_dense.html">layer_einsum_dense</a>()</code> <br><code><a href="layer_embedding.html">layer_embedding</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_feature_space.html">layer_feature_space</a>()</code> <br><code><a href="layer_flatten.html">layer_flatten</a>()</code> <br><code><a href="layer_flax_module_wrapper.html">layer_flax_module_wrapper</a>()</code> <br><code><a href="layer_gaussian_dropout.html">layer_gaussian_dropout</a>()</code> <br><code><a href="layer_gaussian_noise.html">layer_gaussian_noise</a>()</code> <br><code><a href="layer_global_average_pooling_1d.html">layer_global_average_pooling_1d</a>()</code> <br><code><a href="layer_global_average_pooling_2d.html">layer_global_average_pooling_2d</a>()</code> <br><code><a href="layer_global_average_pooling_3d.html">layer_global_average_pooling_3d</a>()</code> <br><code><a href="layer_global_max_pooling_1d.html">layer_global_max_pooling_1d</a>()</code> <br><code><a href="layer_global_max_pooling_2d.html">layer_global_max_pooling_2d</a>()</code> <br><code><a href="layer_global_max_pooling_3d.html">layer_global_max_pooling_3d</a>()</code> <br><code><a href="layer_group_normalization.html">layer_group_normalization</a>()</code> <br><code><a href="layer_group_query_attention.html">layer_group_query_attention</a>()</code> <br><code><a href="layer_gru.html">layer_gru</a>()</code> <br><code><a href="layer_hashed_crossing.html">layer_hashed_crossing</a>()</code> <br><code><a href="layer_hashing.html">layer_hashing</a>()</code> <br><code><a href="layer_identity.html">layer_identity</a>()</code> <br><code><a href="layer_integer_lookup.html">layer_integer_lookup</a>()</code> <br><code><a href="layer_jax_model_wrapper.html">layer_jax_model_wrapper</a>()</code> <br><code><a href="layer_lambda.html">layer_lambda</a>()</code> <br><code><a href="layer_layer_normalization.html">layer_layer_normalization</a>()</code> <br><code><a href="layer_lstm.html">layer_lstm</a>()</code> <br><code><a href="layer_masking.html">layer_masking</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_max_pooling_1d.html">layer_max_pooling_1d</a>()</code> <br><code><a href="layer_max_pooling_2d.html">layer_max_pooling_2d</a>()</code> <br><code><a href="layer_max_pooling_3d.html">layer_max_pooling_3d</a>()</code> <br><code><a href="layer_maximum.html">layer_maximum</a>()</code> <br><code><a href="layer_mel_spectrogram.html">layer_mel_spectrogram</a>()</code> <br><code><a href="layer_minimum.html">layer_minimum</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_multi_head_attention.html">layer_multi_head_attention</a>()</code> <br><code><a href="layer_multiply.html">layer_multiply</a>()</code> <br><code><a href="layer_normalization.html">layer_normalization</a>()</code> <br><code><a href="layer_permute.html">layer_permute</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_brightness.html">layer_random_brightness</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_contrast.html">layer_random_contrast</a>()</code> <br><code><a href="layer_random_crop.html">layer_random_crop</a>()</code> <br><code><a href="layer_random_elastic_transform.html">layer_random_elastic_transform</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_flip.html">layer_random_flip</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_rotation.html">layer_random_rotation</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_random_translation.html">layer_random_translation</a>()</code> <br><code><a href="layer_random_zoom.html">layer_random_zoom</a>()</code> <br><code><a href="layer_repeat_vector.html">layer_repeat_vector</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_reshape.html">layer_reshape</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_rms_normalization.html">layer_rms_normalization</a>()</code> <br><code><a href="layer_rnn.html">layer_rnn</a>()</code> <br><code><a href="layer_separable_conv_1d.html">layer_separable_conv_1d</a>()</code> <br><code><a href="layer_separable_conv_2d.html">layer_separable_conv_2d</a>()</code> <br><code><a href="layer_simple_rnn.html">layer_simple_rnn</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br><code><a href="layer_spatial_dropout_1d.html">layer_spatial_dropout_1d</a>()</code> <br><code><a href="layer_spatial_dropout_2d.html">layer_spatial_dropout_2d</a>()</code> <br><code><a href="layer_spatial_dropout_3d.html">layer_spatial_dropout_3d</a>()</code> <br><code><a href="layer_spectral_normalization.html">layer_spectral_normalization</a>()</code> <br><code><a href="layer_stft_spectrogram.html">layer_stft_spectrogram</a>()</code> <br><code><a href="layer_string_lookup.html">layer_string_lookup</a>()</code> <br><code><a href="layer_subtract.html">layer_subtract</a>()</code> <br><code><a href="layer_text_vectorization.html">layer_text_vectorization</a>()</code> <br><code><a href="layer_tfsm.html">layer_tfsm</a>()</code> <br><code><a href="layer_time_distributed.html">layer_time_distributed</a>()</code> <br><code><a href="layer_torch_module_wrapper.html">layer_torch_module_wrapper</a>()</code> <br><code><a href="layer_unit_normalization.html">layer_unit_normalization</a>()</code> <br><code><a href="layer_upsampling_1d.html">layer_upsampling_1d</a>()</code> <br><code><a href="layer_upsampling_2d.html">layer_upsampling_2d</a>()</code> <br><code><a href="layer_upsampling_3d.html">layer_upsampling_3d</a>()</code> <br><code><a href="layer_zero_padding_1d.html">layer_zero_padding_1d</a>()</code> <br><code><a href="layer_zero_padding_2d.html">layer_zero_padding_2d</a>()</code> <br><code><a href="layer_zero_padding_3d.html">layer_zero_padding_3d</a>()</code> <br><code><a href="rnn_cell_gru.html">rnn_cell_gru</a>()</code> <br><code><a href="rnn_cell_lstm.html">rnn_cell_lstm</a>()</code> <br><code><a href="rnn_cell_simple.html">rnn_cell_simple</a>()</code> <br><code><a href="rnn_cells_stack.html">rnn_cells_stack</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -169,7 +169,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/layer_random_contrast.html b/docs/dev/reference/layer_random_contrast.html
index 4a169e9ab..d34653b26 100644
--- a/docs/dev/reference/layer_random_contrast.html
+++ b/docs/dev/reference/layer_random_contrast.html
@@ -28,7 +28,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -155,8 +155,8 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
     <div class="dont-index">
 <ul><li><p><a href="https://keras.io/api/layers/preprocessing_layers/image_augmentation/random_contrast#randomcontrast-class" class="external-link">https://keras.io/api/layers/preprocessing_layers/image_augmentation/random_contrast#randomcontrast-class</a></p></li>
 </ul><p>Other image augmentation layers: <br><code><a href="layer_random_brightness.html">layer_random_brightness</a>()</code> <br><code><a href="layer_random_crop.html">layer_random_crop</a>()</code> <br><code><a href="layer_random_flip.html">layer_random_flip</a>()</code> <br><code><a href="layer_random_rotation.html">layer_random_rotation</a>()</code> <br><code><a href="layer_random_translation.html">layer_random_translation</a>()</code> <br><code><a href="layer_random_zoom.html">layer_random_zoom</a>()</code> <br></p>
-<p>Other preprocessing layers: <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_category_encoding.html">layer_category_encoding</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_discretization.html">layer_discretization</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_feature_space.html">layer_feature_space</a>()</code> <br><code><a href="layer_hashed_crossing.html">layer_hashed_crossing</a>()</code> <br><code><a href="layer_hashing.html">layer_hashing</a>()</code> <br><code><a href="layer_integer_lookup.html">layer_integer_lookup</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_mel_spectrogram.html">layer_mel_spectrogram</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_normalization.html">layer_normalization</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_brightness.html">layer_random_brightness</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_crop.html">layer_random_crop</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_flip.html">layer_random_flip</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_rotation.html">layer_random_rotation</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_random_translation.html">layer_random_translation</a>()</code> <br><code><a href="layer_random_zoom.html">layer_random_zoom</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br><code><a href="layer_stft_spectrogram.html">layer_stft_spectrogram</a>()</code> <br><code><a href="layer_string_lookup.html">layer_string_lookup</a>()</code> <br><code><a href="layer_text_vectorization.html">layer_text_vectorization</a>()</code> <br></p>
-<p>Other layers: <br><code><a href="Layer.html">Layer</a>()</code> <br><code><a href="layer_activation.html">layer_activation</a>()</code> <br><code><a href="layer_activation_elu.html">layer_activation_elu</a>()</code> <br><code><a href="layer_activation_leaky_relu.html">layer_activation_leaky_relu</a>()</code> <br><code><a href="layer_activation_parametric_relu.html">layer_activation_parametric_relu</a>()</code> <br><code><a href="layer_activation_relu.html">layer_activation_relu</a>()</code> <br><code><a href="layer_activation_softmax.html">layer_activation_softmax</a>()</code> <br><code><a href="layer_activity_regularization.html">layer_activity_regularization</a>()</code> <br><code><a href="layer_add.html">layer_add</a>()</code> <br><code><a href="layer_additive_attention.html">layer_additive_attention</a>()</code> <br><code><a href="layer_alpha_dropout.html">layer_alpha_dropout</a>()</code> <br><code><a href="layer_attention.html">layer_attention</a>()</code> <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_average.html">layer_average</a>()</code> <br><code><a href="layer_average_pooling_1d.html">layer_average_pooling_1d</a>()</code> <br><code><a href="layer_average_pooling_2d.html">layer_average_pooling_2d</a>()</code> <br><code><a href="layer_average_pooling_3d.html">layer_average_pooling_3d</a>()</code> <br><code><a href="layer_batch_normalization.html">layer_batch_normalization</a>()</code> <br><code><a href="layer_bidirectional.html">layer_bidirectional</a>()</code> <br><code><a href="layer_category_encoding.html">layer_category_encoding</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_concatenate.html">layer_concatenate</a>()</code> <br><code><a href="layer_conv_1d.html">layer_conv_1d</a>()</code> <br><code><a href="layer_conv_1d_transpose.html">layer_conv_1d_transpose</a>()</code> <br><code><a href="layer_conv_2d.html">layer_conv_2d</a>()</code> <br><code><a href="layer_conv_2d_transpose.html">layer_conv_2d_transpose</a>()</code> <br><code><a href="layer_conv_3d.html">layer_conv_3d</a>()</code> <br><code><a href="layer_conv_3d_transpose.html">layer_conv_3d_transpose</a>()</code> <br><code><a href="layer_conv_lstm_1d.html">layer_conv_lstm_1d</a>()</code> <br><code><a href="layer_conv_lstm_2d.html">layer_conv_lstm_2d</a>()</code> <br><code><a href="layer_conv_lstm_3d.html">layer_conv_lstm_3d</a>()</code> <br><code><a href="layer_cropping_1d.html">layer_cropping_1d</a>()</code> <br><code><a href="layer_cropping_2d.html">layer_cropping_2d</a>()</code> <br><code><a href="layer_cropping_3d.html">layer_cropping_3d</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_dense.html">layer_dense</a>()</code> <br><code><a href="layer_depthwise_conv_1d.html">layer_depthwise_conv_1d</a>()</code> <br><code><a href="layer_depthwise_conv_2d.html">layer_depthwise_conv_2d</a>()</code> <br><code><a href="layer_discretization.html">layer_discretization</a>()</code> <br><code><a href="layer_dot.html">layer_dot</a>()</code> <br><code><a href="layer_dropout.html">layer_dropout</a>()</code> <br><code><a href="layer_einsum_dense.html">layer_einsum_dense</a>()</code> <br><code><a href="layer_embedding.html">layer_embedding</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_feature_space.html">layer_feature_space</a>()</code> <br><code><a href="layer_flatten.html">layer_flatten</a>()</code> <br><code><a href="layer_flax_module_wrapper.html">layer_flax_module_wrapper</a>()</code> <br><code><a href="layer_gaussian_dropout.html">layer_gaussian_dropout</a>()</code> <br><code><a href="layer_gaussian_noise.html">layer_gaussian_noise</a>()</code> <br><code><a href="layer_global_average_pooling_1d.html">layer_global_average_pooling_1d</a>()</code> <br><code><a href="layer_global_average_pooling_2d.html">layer_global_average_pooling_2d</a>()</code> <br><code><a href="layer_global_average_pooling_3d.html">layer_global_average_pooling_3d</a>()</code> <br><code><a href="layer_global_max_pooling_1d.html">layer_global_max_pooling_1d</a>()</code> <br><code><a href="layer_global_max_pooling_2d.html">layer_global_max_pooling_2d</a>()</code> <br><code><a href="layer_global_max_pooling_3d.html">layer_global_max_pooling_3d</a>()</code> <br><code><a href="layer_group_normalization.html">layer_group_normalization</a>()</code> <br><code><a href="layer_group_query_attention.html">layer_group_query_attention</a>()</code> <br><code><a href="layer_gru.html">layer_gru</a>()</code> <br><code><a href="layer_hashed_crossing.html">layer_hashed_crossing</a>()</code> <br><code><a href="layer_hashing.html">layer_hashing</a>()</code> <br><code><a href="layer_identity.html">layer_identity</a>()</code> <br><code><a href="layer_integer_lookup.html">layer_integer_lookup</a>()</code> <br><code><a href="layer_jax_model_wrapper.html">layer_jax_model_wrapper</a>()</code> <br><code><a href="layer_lambda.html">layer_lambda</a>()</code> <br><code><a href="layer_layer_normalization.html">layer_layer_normalization</a>()</code> <br><code><a href="layer_lstm.html">layer_lstm</a>()</code> <br><code><a href="layer_masking.html">layer_masking</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_max_pooling_1d.html">layer_max_pooling_1d</a>()</code> <br><code><a href="layer_max_pooling_2d.html">layer_max_pooling_2d</a>()</code> <br><code><a href="layer_max_pooling_3d.html">layer_max_pooling_3d</a>()</code> <br><code><a href="layer_maximum.html">layer_maximum</a>()</code> <br><code><a href="layer_mel_spectrogram.html">layer_mel_spectrogram</a>()</code> <br><code><a href="layer_minimum.html">layer_minimum</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_multi_head_attention.html">layer_multi_head_attention</a>()</code> <br><code><a href="layer_multiply.html">layer_multiply</a>()</code> <br><code><a href="layer_normalization.html">layer_normalization</a>()</code> <br><code><a href="layer_permute.html">layer_permute</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_brightness.html">layer_random_brightness</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_crop.html">layer_random_crop</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_flip.html">layer_random_flip</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_rotation.html">layer_random_rotation</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_random_translation.html">layer_random_translation</a>()</code> <br><code><a href="layer_random_zoom.html">layer_random_zoom</a>()</code> <br><code><a href="layer_repeat_vector.html">layer_repeat_vector</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_reshape.html">layer_reshape</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_rms_normalization.html">layer_rms_normalization</a>()</code> <br><code><a href="layer_rnn.html">layer_rnn</a>()</code> <br><code><a href="layer_separable_conv_1d.html">layer_separable_conv_1d</a>()</code> <br><code><a href="layer_separable_conv_2d.html">layer_separable_conv_2d</a>()</code> <br><code><a href="layer_simple_rnn.html">layer_simple_rnn</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br><code><a href="layer_spatial_dropout_1d.html">layer_spatial_dropout_1d</a>()</code> <br><code><a href="layer_spatial_dropout_2d.html">layer_spatial_dropout_2d</a>()</code> <br><code><a href="layer_spatial_dropout_3d.html">layer_spatial_dropout_3d</a>()</code> <br><code><a href="layer_spectral_normalization.html">layer_spectral_normalization</a>()</code> <br><code><a href="layer_stft_spectrogram.html">layer_stft_spectrogram</a>()</code> <br><code><a href="layer_string_lookup.html">layer_string_lookup</a>()</code> <br><code><a href="layer_subtract.html">layer_subtract</a>()</code> <br><code><a href="layer_text_vectorization.html">layer_text_vectorization</a>()</code> <br><code><a href="layer_tfsm.html">layer_tfsm</a>()</code> <br><code><a href="layer_time_distributed.html">layer_time_distributed</a>()</code> <br><code><a href="layer_torch_module_wrapper.html">layer_torch_module_wrapper</a>()</code> <br><code><a href="layer_unit_normalization.html">layer_unit_normalization</a>()</code> <br><code><a href="layer_upsampling_1d.html">layer_upsampling_1d</a>()</code> <br><code><a href="layer_upsampling_2d.html">layer_upsampling_2d</a>()</code> <br><code><a href="layer_upsampling_3d.html">layer_upsampling_3d</a>()</code> <br><code><a href="layer_zero_padding_1d.html">layer_zero_padding_1d</a>()</code> <br><code><a href="layer_zero_padding_2d.html">layer_zero_padding_2d</a>()</code> <br><code><a href="layer_zero_padding_3d.html">layer_zero_padding_3d</a>()</code> <br><code><a href="rnn_cell_gru.html">rnn_cell_gru</a>()</code> <br><code><a href="rnn_cell_lstm.html">rnn_cell_lstm</a>()</code> <br><code><a href="rnn_cell_simple.html">rnn_cell_simple</a>()</code> <br><code><a href="rnn_cells_stack.html">rnn_cells_stack</a>()</code> <br></p></div>
+<p>Other preprocessing layers: <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_category_encoding.html">layer_category_encoding</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_discretization.html">layer_discretization</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_feature_space.html">layer_feature_space</a>()</code> <br><code><a href="layer_hashed_crossing.html">layer_hashed_crossing</a>()</code> <br><code><a href="layer_hashing.html">layer_hashing</a>()</code> <br><code><a href="layer_integer_lookup.html">layer_integer_lookup</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_mel_spectrogram.html">layer_mel_spectrogram</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_normalization.html">layer_normalization</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_brightness.html">layer_random_brightness</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_crop.html">layer_random_crop</a>()</code> <br><code><a href="layer_random_elastic_transform.html">layer_random_elastic_transform</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_flip.html">layer_random_flip</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_rotation.html">layer_random_rotation</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_random_translation.html">layer_random_translation</a>()</code> <br><code><a href="layer_random_zoom.html">layer_random_zoom</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br><code><a href="layer_stft_spectrogram.html">layer_stft_spectrogram</a>()</code> <br><code><a href="layer_string_lookup.html">layer_string_lookup</a>()</code> <br><code><a href="layer_text_vectorization.html">layer_text_vectorization</a>()</code> <br></p>
+<p>Other layers: <br><code><a href="Layer.html">Layer</a>()</code> <br><code><a href="layer_activation.html">layer_activation</a>()</code> <br><code><a href="layer_activation_elu.html">layer_activation_elu</a>()</code> <br><code><a href="layer_activation_leaky_relu.html">layer_activation_leaky_relu</a>()</code> <br><code><a href="layer_activation_parametric_relu.html">layer_activation_parametric_relu</a>()</code> <br><code><a href="layer_activation_relu.html">layer_activation_relu</a>()</code> <br><code><a href="layer_activation_softmax.html">layer_activation_softmax</a>()</code> <br><code><a href="layer_activity_regularization.html">layer_activity_regularization</a>()</code> <br><code><a href="layer_add.html">layer_add</a>()</code> <br><code><a href="layer_additive_attention.html">layer_additive_attention</a>()</code> <br><code><a href="layer_alpha_dropout.html">layer_alpha_dropout</a>()</code> <br><code><a href="layer_attention.html">layer_attention</a>()</code> <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_average.html">layer_average</a>()</code> <br><code><a href="layer_average_pooling_1d.html">layer_average_pooling_1d</a>()</code> <br><code><a href="layer_average_pooling_2d.html">layer_average_pooling_2d</a>()</code> <br><code><a href="layer_average_pooling_3d.html">layer_average_pooling_3d</a>()</code> <br><code><a href="layer_batch_normalization.html">layer_batch_normalization</a>()</code> <br><code><a href="layer_bidirectional.html">layer_bidirectional</a>()</code> <br><code><a href="layer_category_encoding.html">layer_category_encoding</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_concatenate.html">layer_concatenate</a>()</code> <br><code><a href="layer_conv_1d.html">layer_conv_1d</a>()</code> <br><code><a href="layer_conv_1d_transpose.html">layer_conv_1d_transpose</a>()</code> <br><code><a href="layer_conv_2d.html">layer_conv_2d</a>()</code> <br><code><a href="layer_conv_2d_transpose.html">layer_conv_2d_transpose</a>()</code> <br><code><a href="layer_conv_3d.html">layer_conv_3d</a>()</code> <br><code><a href="layer_conv_3d_transpose.html">layer_conv_3d_transpose</a>()</code> <br><code><a href="layer_conv_lstm_1d.html">layer_conv_lstm_1d</a>()</code> <br><code><a href="layer_conv_lstm_2d.html">layer_conv_lstm_2d</a>()</code> <br><code><a href="layer_conv_lstm_3d.html">layer_conv_lstm_3d</a>()</code> <br><code><a href="layer_cropping_1d.html">layer_cropping_1d</a>()</code> <br><code><a href="layer_cropping_2d.html">layer_cropping_2d</a>()</code> <br><code><a href="layer_cropping_3d.html">layer_cropping_3d</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_dense.html">layer_dense</a>()</code> <br><code><a href="layer_depthwise_conv_1d.html">layer_depthwise_conv_1d</a>()</code> <br><code><a href="layer_depthwise_conv_2d.html">layer_depthwise_conv_2d</a>()</code> <br><code><a href="layer_discretization.html">layer_discretization</a>()</code> <br><code><a href="layer_dot.html">layer_dot</a>()</code> <br><code><a href="layer_dropout.html">layer_dropout</a>()</code> <br><code><a href="layer_einsum_dense.html">layer_einsum_dense</a>()</code> <br><code><a href="layer_embedding.html">layer_embedding</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_feature_space.html">layer_feature_space</a>()</code> <br><code><a href="layer_flatten.html">layer_flatten</a>()</code> <br><code><a href="layer_flax_module_wrapper.html">layer_flax_module_wrapper</a>()</code> <br><code><a href="layer_gaussian_dropout.html">layer_gaussian_dropout</a>()</code> <br><code><a href="layer_gaussian_noise.html">layer_gaussian_noise</a>()</code> <br><code><a href="layer_global_average_pooling_1d.html">layer_global_average_pooling_1d</a>()</code> <br><code><a href="layer_global_average_pooling_2d.html">layer_global_average_pooling_2d</a>()</code> <br><code><a href="layer_global_average_pooling_3d.html">layer_global_average_pooling_3d</a>()</code> <br><code><a href="layer_global_max_pooling_1d.html">layer_global_max_pooling_1d</a>()</code> <br><code><a href="layer_global_max_pooling_2d.html">layer_global_max_pooling_2d</a>()</code> <br><code><a href="layer_global_max_pooling_3d.html">layer_global_max_pooling_3d</a>()</code> <br><code><a href="layer_group_normalization.html">layer_group_normalization</a>()</code> <br><code><a href="layer_group_query_attention.html">layer_group_query_attention</a>()</code> <br><code><a href="layer_gru.html">layer_gru</a>()</code> <br><code><a href="layer_hashed_crossing.html">layer_hashed_crossing</a>()</code> <br><code><a href="layer_hashing.html">layer_hashing</a>()</code> <br><code><a href="layer_identity.html">layer_identity</a>()</code> <br><code><a href="layer_integer_lookup.html">layer_integer_lookup</a>()</code> <br><code><a href="layer_jax_model_wrapper.html">layer_jax_model_wrapper</a>()</code> <br><code><a href="layer_lambda.html">layer_lambda</a>()</code> <br><code><a href="layer_layer_normalization.html">layer_layer_normalization</a>()</code> <br><code><a href="layer_lstm.html">layer_lstm</a>()</code> <br><code><a href="layer_masking.html">layer_masking</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_max_pooling_1d.html">layer_max_pooling_1d</a>()</code> <br><code><a href="layer_max_pooling_2d.html">layer_max_pooling_2d</a>()</code> <br><code><a href="layer_max_pooling_3d.html">layer_max_pooling_3d</a>()</code> <br><code><a href="layer_maximum.html">layer_maximum</a>()</code> <br><code><a href="layer_mel_spectrogram.html">layer_mel_spectrogram</a>()</code> <br><code><a href="layer_minimum.html">layer_minimum</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_multi_head_attention.html">layer_multi_head_attention</a>()</code> <br><code><a href="layer_multiply.html">layer_multiply</a>()</code> <br><code><a href="layer_normalization.html">layer_normalization</a>()</code> <br><code><a href="layer_permute.html">layer_permute</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_brightness.html">layer_random_brightness</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_crop.html">layer_random_crop</a>()</code> <br><code><a href="layer_random_elastic_transform.html">layer_random_elastic_transform</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_flip.html">layer_random_flip</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_rotation.html">layer_random_rotation</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_random_translation.html">layer_random_translation</a>()</code> <br><code><a href="layer_random_zoom.html">layer_random_zoom</a>()</code> <br><code><a href="layer_repeat_vector.html">layer_repeat_vector</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_reshape.html">layer_reshape</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_rms_normalization.html">layer_rms_normalization</a>()</code> <br><code><a href="layer_rnn.html">layer_rnn</a>()</code> <br><code><a href="layer_separable_conv_1d.html">layer_separable_conv_1d</a>()</code> <br><code><a href="layer_separable_conv_2d.html">layer_separable_conv_2d</a>()</code> <br><code><a href="layer_simple_rnn.html">layer_simple_rnn</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br><code><a href="layer_spatial_dropout_1d.html">layer_spatial_dropout_1d</a>()</code> <br><code><a href="layer_spatial_dropout_2d.html">layer_spatial_dropout_2d</a>()</code> <br><code><a href="layer_spatial_dropout_3d.html">layer_spatial_dropout_3d</a>()</code> <br><code><a href="layer_spectral_normalization.html">layer_spectral_normalization</a>()</code> <br><code><a href="layer_stft_spectrogram.html">layer_stft_spectrogram</a>()</code> <br><code><a href="layer_string_lookup.html">layer_string_lookup</a>()</code> <br><code><a href="layer_subtract.html">layer_subtract</a>()</code> <br><code><a href="layer_text_vectorization.html">layer_text_vectorization</a>()</code> <br><code><a href="layer_tfsm.html">layer_tfsm</a>()</code> <br><code><a href="layer_time_distributed.html">layer_time_distributed</a>()</code> <br><code><a href="layer_torch_module_wrapper.html">layer_torch_module_wrapper</a>()</code> <br><code><a href="layer_unit_normalization.html">layer_unit_normalization</a>()</code> <br><code><a href="layer_upsampling_1d.html">layer_upsampling_1d</a>()</code> <br><code><a href="layer_upsampling_2d.html">layer_upsampling_2d</a>()</code> <br><code><a href="layer_upsampling_3d.html">layer_upsampling_3d</a>()</code> <br><code><a href="layer_zero_padding_1d.html">layer_zero_padding_1d</a>()</code> <br><code><a href="layer_zero_padding_2d.html">layer_zero_padding_2d</a>()</code> <br><code><a href="layer_zero_padding_3d.html">layer_zero_padding_3d</a>()</code> <br><code><a href="rnn_cell_gru.html">rnn_cell_gru</a>()</code> <br><code><a href="rnn_cell_lstm.html">rnn_cell_lstm</a>()</code> <br><code><a href="rnn_cell_simple.html">rnn_cell_simple</a>()</code> <br><code><a href="rnn_cells_stack.html">rnn_cells_stack</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -168,7 +168,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/layer_random_crop.html b/docs/dev/reference/layer_random_crop.html
index d7b559dbb..cae8d6d6c 100644
--- a/docs/dev/reference/layer_random_crop.html
+++ b/docs/dev/reference/layer_random_crop.html
@@ -32,7 +32,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -170,8 +170,8 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
     <div class="dont-index">
 <ul><li><p><a href="https://keras.io/api/layers/preprocessing_layers/image_augmentation/random_crop#randomcrop-class" class="external-link">https://keras.io/api/layers/preprocessing_layers/image_augmentation/random_crop#randomcrop-class</a></p></li>
 </ul><p>Other image augmentation layers: <br><code><a href="layer_random_brightness.html">layer_random_brightness</a>()</code> <br><code><a href="layer_random_contrast.html">layer_random_contrast</a>()</code> <br><code><a href="layer_random_flip.html">layer_random_flip</a>()</code> <br><code><a href="layer_random_rotation.html">layer_random_rotation</a>()</code> <br><code><a href="layer_random_translation.html">layer_random_translation</a>()</code> <br><code><a href="layer_random_zoom.html">layer_random_zoom</a>()</code> <br></p>
-<p>Other preprocessing layers: <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_category_encoding.html">layer_category_encoding</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_discretization.html">layer_discretization</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_feature_space.html">layer_feature_space</a>()</code> <br><code><a href="layer_hashed_crossing.html">layer_hashed_crossing</a>()</code> <br><code><a href="layer_hashing.html">layer_hashing</a>()</code> <br><code><a href="layer_integer_lookup.html">layer_integer_lookup</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_mel_spectrogram.html">layer_mel_spectrogram</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_normalization.html">layer_normalization</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_brightness.html">layer_random_brightness</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_contrast.html">layer_random_contrast</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_flip.html">layer_random_flip</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_rotation.html">layer_random_rotation</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_random_translation.html">layer_random_translation</a>()</code> <br><code><a href="layer_random_zoom.html">layer_random_zoom</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br><code><a href="layer_stft_spectrogram.html">layer_stft_spectrogram</a>()</code> <br><code><a href="layer_string_lookup.html">layer_string_lookup</a>()</code> <br><code><a href="layer_text_vectorization.html">layer_text_vectorization</a>()</code> <br></p>
-<p>Other layers: <br><code><a href="Layer.html">Layer</a>()</code> <br><code><a href="layer_activation.html">layer_activation</a>()</code> <br><code><a href="layer_activation_elu.html">layer_activation_elu</a>()</code> <br><code><a href="layer_activation_leaky_relu.html">layer_activation_leaky_relu</a>()</code> <br><code><a href="layer_activation_parametric_relu.html">layer_activation_parametric_relu</a>()</code> <br><code><a href="layer_activation_relu.html">layer_activation_relu</a>()</code> <br><code><a href="layer_activation_softmax.html">layer_activation_softmax</a>()</code> <br><code><a href="layer_activity_regularization.html">layer_activity_regularization</a>()</code> <br><code><a href="layer_add.html">layer_add</a>()</code> <br><code><a href="layer_additive_attention.html">layer_additive_attention</a>()</code> <br><code><a href="layer_alpha_dropout.html">layer_alpha_dropout</a>()</code> <br><code><a href="layer_attention.html">layer_attention</a>()</code> <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_average.html">layer_average</a>()</code> <br><code><a href="layer_average_pooling_1d.html">layer_average_pooling_1d</a>()</code> <br><code><a href="layer_average_pooling_2d.html">layer_average_pooling_2d</a>()</code> <br><code><a href="layer_average_pooling_3d.html">layer_average_pooling_3d</a>()</code> <br><code><a href="layer_batch_normalization.html">layer_batch_normalization</a>()</code> <br><code><a href="layer_bidirectional.html">layer_bidirectional</a>()</code> <br><code><a href="layer_category_encoding.html">layer_category_encoding</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_concatenate.html">layer_concatenate</a>()</code> <br><code><a href="layer_conv_1d.html">layer_conv_1d</a>()</code> <br><code><a href="layer_conv_1d_transpose.html">layer_conv_1d_transpose</a>()</code> <br><code><a href="layer_conv_2d.html">layer_conv_2d</a>()</code> <br><code><a href="layer_conv_2d_transpose.html">layer_conv_2d_transpose</a>()</code> <br><code><a href="layer_conv_3d.html">layer_conv_3d</a>()</code> <br><code><a href="layer_conv_3d_transpose.html">layer_conv_3d_transpose</a>()</code> <br><code><a href="layer_conv_lstm_1d.html">layer_conv_lstm_1d</a>()</code> <br><code><a href="layer_conv_lstm_2d.html">layer_conv_lstm_2d</a>()</code> <br><code><a href="layer_conv_lstm_3d.html">layer_conv_lstm_3d</a>()</code> <br><code><a href="layer_cropping_1d.html">layer_cropping_1d</a>()</code> <br><code><a href="layer_cropping_2d.html">layer_cropping_2d</a>()</code> <br><code><a href="layer_cropping_3d.html">layer_cropping_3d</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_dense.html">layer_dense</a>()</code> <br><code><a href="layer_depthwise_conv_1d.html">layer_depthwise_conv_1d</a>()</code> <br><code><a href="layer_depthwise_conv_2d.html">layer_depthwise_conv_2d</a>()</code> <br><code><a href="layer_discretization.html">layer_discretization</a>()</code> <br><code><a href="layer_dot.html">layer_dot</a>()</code> <br><code><a href="layer_dropout.html">layer_dropout</a>()</code> <br><code><a href="layer_einsum_dense.html">layer_einsum_dense</a>()</code> <br><code><a href="layer_embedding.html">layer_embedding</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_feature_space.html">layer_feature_space</a>()</code> <br><code><a href="layer_flatten.html">layer_flatten</a>()</code> <br><code><a href="layer_flax_module_wrapper.html">layer_flax_module_wrapper</a>()</code> <br><code><a href="layer_gaussian_dropout.html">layer_gaussian_dropout</a>()</code> <br><code><a href="layer_gaussian_noise.html">layer_gaussian_noise</a>()</code> <br><code><a href="layer_global_average_pooling_1d.html">layer_global_average_pooling_1d</a>()</code> <br><code><a href="layer_global_average_pooling_2d.html">layer_global_average_pooling_2d</a>()</code> <br><code><a href="layer_global_average_pooling_3d.html">layer_global_average_pooling_3d</a>()</code> <br><code><a href="layer_global_max_pooling_1d.html">layer_global_max_pooling_1d</a>()</code> <br><code><a href="layer_global_max_pooling_2d.html">layer_global_max_pooling_2d</a>()</code> <br><code><a href="layer_global_max_pooling_3d.html">layer_global_max_pooling_3d</a>()</code> <br><code><a href="layer_group_normalization.html">layer_group_normalization</a>()</code> <br><code><a href="layer_group_query_attention.html">layer_group_query_attention</a>()</code> <br><code><a href="layer_gru.html">layer_gru</a>()</code> <br><code><a href="layer_hashed_crossing.html">layer_hashed_crossing</a>()</code> <br><code><a href="layer_hashing.html">layer_hashing</a>()</code> <br><code><a href="layer_identity.html">layer_identity</a>()</code> <br><code><a href="layer_integer_lookup.html">layer_integer_lookup</a>()</code> <br><code><a href="layer_jax_model_wrapper.html">layer_jax_model_wrapper</a>()</code> <br><code><a href="layer_lambda.html">layer_lambda</a>()</code> <br><code><a href="layer_layer_normalization.html">layer_layer_normalization</a>()</code> <br><code><a href="layer_lstm.html">layer_lstm</a>()</code> <br><code><a href="layer_masking.html">layer_masking</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_max_pooling_1d.html">layer_max_pooling_1d</a>()</code> <br><code><a href="layer_max_pooling_2d.html">layer_max_pooling_2d</a>()</code> <br><code><a href="layer_max_pooling_3d.html">layer_max_pooling_3d</a>()</code> <br><code><a href="layer_maximum.html">layer_maximum</a>()</code> <br><code><a href="layer_mel_spectrogram.html">layer_mel_spectrogram</a>()</code> <br><code><a href="layer_minimum.html">layer_minimum</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_multi_head_attention.html">layer_multi_head_attention</a>()</code> <br><code><a href="layer_multiply.html">layer_multiply</a>()</code> <br><code><a href="layer_normalization.html">layer_normalization</a>()</code> <br><code><a href="layer_permute.html">layer_permute</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_brightness.html">layer_random_brightness</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_contrast.html">layer_random_contrast</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_flip.html">layer_random_flip</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_rotation.html">layer_random_rotation</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_random_translation.html">layer_random_translation</a>()</code> <br><code><a href="layer_random_zoom.html">layer_random_zoom</a>()</code> <br><code><a href="layer_repeat_vector.html">layer_repeat_vector</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_reshape.html">layer_reshape</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_rms_normalization.html">layer_rms_normalization</a>()</code> <br><code><a href="layer_rnn.html">layer_rnn</a>()</code> <br><code><a href="layer_separable_conv_1d.html">layer_separable_conv_1d</a>()</code> <br><code><a href="layer_separable_conv_2d.html">layer_separable_conv_2d</a>()</code> <br><code><a href="layer_simple_rnn.html">layer_simple_rnn</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br><code><a href="layer_spatial_dropout_1d.html">layer_spatial_dropout_1d</a>()</code> <br><code><a href="layer_spatial_dropout_2d.html">layer_spatial_dropout_2d</a>()</code> <br><code><a href="layer_spatial_dropout_3d.html">layer_spatial_dropout_3d</a>()</code> <br><code><a href="layer_spectral_normalization.html">layer_spectral_normalization</a>()</code> <br><code><a href="layer_stft_spectrogram.html">layer_stft_spectrogram</a>()</code> <br><code><a href="layer_string_lookup.html">layer_string_lookup</a>()</code> <br><code><a href="layer_subtract.html">layer_subtract</a>()</code> <br><code><a href="layer_text_vectorization.html">layer_text_vectorization</a>()</code> <br><code><a href="layer_tfsm.html">layer_tfsm</a>()</code> <br><code><a href="layer_time_distributed.html">layer_time_distributed</a>()</code> <br><code><a href="layer_torch_module_wrapper.html">layer_torch_module_wrapper</a>()</code> <br><code><a href="layer_unit_normalization.html">layer_unit_normalization</a>()</code> <br><code><a href="layer_upsampling_1d.html">layer_upsampling_1d</a>()</code> <br><code><a href="layer_upsampling_2d.html">layer_upsampling_2d</a>()</code> <br><code><a href="layer_upsampling_3d.html">layer_upsampling_3d</a>()</code> <br><code><a href="layer_zero_padding_1d.html">layer_zero_padding_1d</a>()</code> <br><code><a href="layer_zero_padding_2d.html">layer_zero_padding_2d</a>()</code> <br><code><a href="layer_zero_padding_3d.html">layer_zero_padding_3d</a>()</code> <br><code><a href="rnn_cell_gru.html">rnn_cell_gru</a>()</code> <br><code><a href="rnn_cell_lstm.html">rnn_cell_lstm</a>()</code> <br><code><a href="rnn_cell_simple.html">rnn_cell_simple</a>()</code> <br><code><a href="rnn_cells_stack.html">rnn_cells_stack</a>()</code> <br></p></div>
+<p>Other preprocessing layers: <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_category_encoding.html">layer_category_encoding</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_discretization.html">layer_discretization</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_feature_space.html">layer_feature_space</a>()</code> <br><code><a href="layer_hashed_crossing.html">layer_hashed_crossing</a>()</code> <br><code><a href="layer_hashing.html">layer_hashing</a>()</code> <br><code><a href="layer_integer_lookup.html">layer_integer_lookup</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_mel_spectrogram.html">layer_mel_spectrogram</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_normalization.html">layer_normalization</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_brightness.html">layer_random_brightness</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_contrast.html">layer_random_contrast</a>()</code> <br><code><a href="layer_random_elastic_transform.html">layer_random_elastic_transform</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_flip.html">layer_random_flip</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_rotation.html">layer_random_rotation</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_random_translation.html">layer_random_translation</a>()</code> <br><code><a href="layer_random_zoom.html">layer_random_zoom</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br><code><a href="layer_stft_spectrogram.html">layer_stft_spectrogram</a>()</code> <br><code><a href="layer_string_lookup.html">layer_string_lookup</a>()</code> <br><code><a href="layer_text_vectorization.html">layer_text_vectorization</a>()</code> <br></p>
+<p>Other layers: <br><code><a href="Layer.html">Layer</a>()</code> <br><code><a href="layer_activation.html">layer_activation</a>()</code> <br><code><a href="layer_activation_elu.html">layer_activation_elu</a>()</code> <br><code><a href="layer_activation_leaky_relu.html">layer_activation_leaky_relu</a>()</code> <br><code><a href="layer_activation_parametric_relu.html">layer_activation_parametric_relu</a>()</code> <br><code><a href="layer_activation_relu.html">layer_activation_relu</a>()</code> <br><code><a href="layer_activation_softmax.html">layer_activation_softmax</a>()</code> <br><code><a href="layer_activity_regularization.html">layer_activity_regularization</a>()</code> <br><code><a href="layer_add.html">layer_add</a>()</code> <br><code><a href="layer_additive_attention.html">layer_additive_attention</a>()</code> <br><code><a href="layer_alpha_dropout.html">layer_alpha_dropout</a>()</code> <br><code><a href="layer_attention.html">layer_attention</a>()</code> <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_average.html">layer_average</a>()</code> <br><code><a href="layer_average_pooling_1d.html">layer_average_pooling_1d</a>()</code> <br><code><a href="layer_average_pooling_2d.html">layer_average_pooling_2d</a>()</code> <br><code><a href="layer_average_pooling_3d.html">layer_average_pooling_3d</a>()</code> <br><code><a href="layer_batch_normalization.html">layer_batch_normalization</a>()</code> <br><code><a href="layer_bidirectional.html">layer_bidirectional</a>()</code> <br><code><a href="layer_category_encoding.html">layer_category_encoding</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_concatenate.html">layer_concatenate</a>()</code> <br><code><a href="layer_conv_1d.html">layer_conv_1d</a>()</code> <br><code><a href="layer_conv_1d_transpose.html">layer_conv_1d_transpose</a>()</code> <br><code><a href="layer_conv_2d.html">layer_conv_2d</a>()</code> <br><code><a href="layer_conv_2d_transpose.html">layer_conv_2d_transpose</a>()</code> <br><code><a href="layer_conv_3d.html">layer_conv_3d</a>()</code> <br><code><a href="layer_conv_3d_transpose.html">layer_conv_3d_transpose</a>()</code> <br><code><a href="layer_conv_lstm_1d.html">layer_conv_lstm_1d</a>()</code> <br><code><a href="layer_conv_lstm_2d.html">layer_conv_lstm_2d</a>()</code> <br><code><a href="layer_conv_lstm_3d.html">layer_conv_lstm_3d</a>()</code> <br><code><a href="layer_cropping_1d.html">layer_cropping_1d</a>()</code> <br><code><a href="layer_cropping_2d.html">layer_cropping_2d</a>()</code> <br><code><a href="layer_cropping_3d.html">layer_cropping_3d</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_dense.html">layer_dense</a>()</code> <br><code><a href="layer_depthwise_conv_1d.html">layer_depthwise_conv_1d</a>()</code> <br><code><a href="layer_depthwise_conv_2d.html">layer_depthwise_conv_2d</a>()</code> <br><code><a href="layer_discretization.html">layer_discretization</a>()</code> <br><code><a href="layer_dot.html">layer_dot</a>()</code> <br><code><a href="layer_dropout.html">layer_dropout</a>()</code> <br><code><a href="layer_einsum_dense.html">layer_einsum_dense</a>()</code> <br><code><a href="layer_embedding.html">layer_embedding</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_feature_space.html">layer_feature_space</a>()</code> <br><code><a href="layer_flatten.html">layer_flatten</a>()</code> <br><code><a href="layer_flax_module_wrapper.html">layer_flax_module_wrapper</a>()</code> <br><code><a href="layer_gaussian_dropout.html">layer_gaussian_dropout</a>()</code> <br><code><a href="layer_gaussian_noise.html">layer_gaussian_noise</a>()</code> <br><code><a href="layer_global_average_pooling_1d.html">layer_global_average_pooling_1d</a>()</code> <br><code><a href="layer_global_average_pooling_2d.html">layer_global_average_pooling_2d</a>()</code> <br><code><a href="layer_global_average_pooling_3d.html">layer_global_average_pooling_3d</a>()</code> <br><code><a href="layer_global_max_pooling_1d.html">layer_global_max_pooling_1d</a>()</code> <br><code><a href="layer_global_max_pooling_2d.html">layer_global_max_pooling_2d</a>()</code> <br><code><a href="layer_global_max_pooling_3d.html">layer_global_max_pooling_3d</a>()</code> <br><code><a href="layer_group_normalization.html">layer_group_normalization</a>()</code> <br><code><a href="layer_group_query_attention.html">layer_group_query_attention</a>()</code> <br><code><a href="layer_gru.html">layer_gru</a>()</code> <br><code><a href="layer_hashed_crossing.html">layer_hashed_crossing</a>()</code> <br><code><a href="layer_hashing.html">layer_hashing</a>()</code> <br><code><a href="layer_identity.html">layer_identity</a>()</code> <br><code><a href="layer_integer_lookup.html">layer_integer_lookup</a>()</code> <br><code><a href="layer_jax_model_wrapper.html">layer_jax_model_wrapper</a>()</code> <br><code><a href="layer_lambda.html">layer_lambda</a>()</code> <br><code><a href="layer_layer_normalization.html">layer_layer_normalization</a>()</code> <br><code><a href="layer_lstm.html">layer_lstm</a>()</code> <br><code><a href="layer_masking.html">layer_masking</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_max_pooling_1d.html">layer_max_pooling_1d</a>()</code> <br><code><a href="layer_max_pooling_2d.html">layer_max_pooling_2d</a>()</code> <br><code><a href="layer_max_pooling_3d.html">layer_max_pooling_3d</a>()</code> <br><code><a href="layer_maximum.html">layer_maximum</a>()</code> <br><code><a href="layer_mel_spectrogram.html">layer_mel_spectrogram</a>()</code> <br><code><a href="layer_minimum.html">layer_minimum</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_multi_head_attention.html">layer_multi_head_attention</a>()</code> <br><code><a href="layer_multiply.html">layer_multiply</a>()</code> <br><code><a href="layer_normalization.html">layer_normalization</a>()</code> <br><code><a href="layer_permute.html">layer_permute</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_brightness.html">layer_random_brightness</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_contrast.html">layer_random_contrast</a>()</code> <br><code><a href="layer_random_elastic_transform.html">layer_random_elastic_transform</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_flip.html">layer_random_flip</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_rotation.html">layer_random_rotation</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_random_translation.html">layer_random_translation</a>()</code> <br><code><a href="layer_random_zoom.html">layer_random_zoom</a>()</code> <br><code><a href="layer_repeat_vector.html">layer_repeat_vector</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_reshape.html">layer_reshape</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_rms_normalization.html">layer_rms_normalization</a>()</code> <br><code><a href="layer_rnn.html">layer_rnn</a>()</code> <br><code><a href="layer_separable_conv_1d.html">layer_separable_conv_1d</a>()</code> <br><code><a href="layer_separable_conv_2d.html">layer_separable_conv_2d</a>()</code> <br><code><a href="layer_simple_rnn.html">layer_simple_rnn</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br><code><a href="layer_spatial_dropout_1d.html">layer_spatial_dropout_1d</a>()</code> <br><code><a href="layer_spatial_dropout_2d.html">layer_spatial_dropout_2d</a>()</code> <br><code><a href="layer_spatial_dropout_3d.html">layer_spatial_dropout_3d</a>()</code> <br><code><a href="layer_spectral_normalization.html">layer_spectral_normalization</a>()</code> <br><code><a href="layer_stft_spectrogram.html">layer_stft_spectrogram</a>()</code> <br><code><a href="layer_string_lookup.html">layer_string_lookup</a>()</code> <br><code><a href="layer_subtract.html">layer_subtract</a>()</code> <br><code><a href="layer_text_vectorization.html">layer_text_vectorization</a>()</code> <br><code><a href="layer_tfsm.html">layer_tfsm</a>()</code> <br><code><a href="layer_time_distributed.html">layer_time_distributed</a>()</code> <br><code><a href="layer_torch_module_wrapper.html">layer_torch_module_wrapper</a>()</code> <br><code><a href="layer_unit_normalization.html">layer_unit_normalization</a>()</code> <br><code><a href="layer_upsampling_1d.html">layer_upsampling_1d</a>()</code> <br><code><a href="layer_upsampling_2d.html">layer_upsampling_2d</a>()</code> <br><code><a href="layer_upsampling_3d.html">layer_upsampling_3d</a>()</code> <br><code><a href="layer_zero_padding_1d.html">layer_zero_padding_1d</a>()</code> <br><code><a href="layer_zero_padding_2d.html">layer_zero_padding_2d</a>()</code> <br><code><a href="layer_zero_padding_3d.html">layer_zero_padding_3d</a>()</code> <br><code><a href="rnn_cell_gru.html">rnn_cell_gru</a>()</code> <br><code><a href="rnn_cell_lstm.html">rnn_cell_lstm</a>()</code> <br><code><a href="rnn_cell_simple.html">rnn_cell_simple</a>()</code> <br><code><a href="rnn_cells_stack.html">rnn_cells_stack</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -183,7 +183,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/layer_random_elastic_transform.html b/docs/dev/reference/layer_random_elastic_transform.html
new file mode 100644
index 000000000..bd9a1ddcf
--- /dev/null
+++ b/docs/dev/reference/layer_random_elastic_transform.html
@@ -0,0 +1,197 @@
+<!DOCTYPE html>
+<!-- Generated by pkgdown: do not edit by hand --><html lang="en"><head><meta http-equiv="Content-Type" content="text/html; charset=UTF-8"><meta charset="utf-8"><meta http-equiv="X-UA-Compatible" content="IE=edge"><meta name="viewport" content="width=device-width, initial-scale=1, shrink-to-fit=no"><title>A preprocessing layer that applies random elastic transformations. — layer_random_elastic_transform • keras3</title><!-- favicons --><link rel="icon" type="image/png" sizes="96x96" href="../favicon-96x96.png"><link rel="icon" type="”image/svg+xml”" href="../favicon.svg"><link rel="apple-touch-icon" sizes="180x180" href="../apple-touch-icon.png"><link rel="icon" sizes="any" href="../favicon.ico"><link rel="manifest" href="../site.webmanifest"><script src="../deps/jquery-3.6.0/jquery-3.6.0.min.js"></script><meta name="viewport" content="width=device-width, initial-scale=1, shrink-to-fit=no"><link href="../deps/bootstrap-5.3.1/bootstrap.min.css" rel="stylesheet"><script src="../deps/bootstrap-5.3.1/bootstrap.bundle.min.js"></script><link href="../deps/font-awesome-6.5.2/css/all.min.css" rel="stylesheet"><link href="../deps/font-awesome-6.5.2/css/v4-shims.min.css" rel="stylesheet"><script src="../deps/headroom-0.11.0/headroom.min.js"></script><script src="../deps/headroom-0.11.0/jQuery.headroom.min.js"></script><script src="../deps/bootstrap-toc-1.0.1/bootstrap-toc.min.js"></script><script src="../deps/clipboard.js-2.0.11/clipboard.min.js"></script><script src="../deps/search-1.0.0/autocomplete.jquery.min.js"></script><script src="../deps/search-1.0.0/fuse.min.js"></script><script src="../deps/search-1.0.0/mark.min.js"></script><!-- pkgdown --><script src="../pkgdown.js"></script><link href="../extra.css" rel="stylesheet"><meta property="og:title" content="A preprocessing layer that applies random elastic transformations. — layer_random_elastic_transform"><meta name="description" content="This layer distorts input images by applying elastic deformations,
+simulating a physically realistic transformation. The magnitude of the
+distortion is controlled by the scale parameter, while the factor
+determines the probability of applying the transformation."><meta property="og:description" content="This layer distorts input images by applying elastic deformations,
+simulating a physically realistic transformation. The magnitude of the
+distortion is controlled by the scale parameter, while the factor
+determines the probability of applying the transformation."><meta name="robots" content="noindex"><!-- Google Tag Manager --><script>(function(w,d,s,l,i){w[l]=w[l]||[];w[l].push({'gtm.start': new Date().getTime(),event:'gtm.js'});var f=d.getElementsByTagName(s)[0],j=d.createElement(s),dl=l!='dataLayer'?'&l='+l:'';j.async=true;j.src='https://www.googletagmanager.com/gtm.js?id='+i+dl;f.parentNode.insertBefore(j,f);})(window,document,'script','dataLayer','GTM-KHBDBW7');</script><!-- End Google Tag Manager --></head><body>
+    <a href="#main" class="visually-hidden-focusable">Skip to contents</a>
+    <!-- Google Tag Manager (noscript) --> <noscript><iframe src="https://www.googletagmanager.com/ns.html?id=GTM-KHBDBW7" height="0" width="0" style="display:none;visibility:hidden"></iframe></noscript> <!-- End Google Tag Manager (noscript) -->
+
+
+    <nav class="navbar navbar-expand-lg fixed-top bg-primary" data-bs-theme="inverse" aria-label="Site navigation"><div class="container">
+
+    <a class="navbar-brand me-2" href="../index.html">keras3</a>
+
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
+
+
+    <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
+      <span class="navbar-toggler-icon"></span>
+    </button>
+
+    <div id="navbar" class="collapse navbar-collapse ms-3">
+      <ul class="navbar-nav me-auto"><li class="nav-item"><a class="nav-link" href="../articles/getting_started.html">Getting Started</a></li>
+<li class="nav-item dropdown">
+  <button class="nav-link dropdown-toggle" type="button" id="dropdown-guides" data-bs-toggle="dropdown" aria-expanded="false" aria-haspopup="true">Guides</button>
+  <ul class="dropdown-menu" aria-labelledby="dropdown-guides"><li><h6 class="dropdown-header" data-toc-skip>Model definition</h6></li>
+    <li><a class="dropdown-item" href="../articles/sequential_model.html">Sequential Model</a></li>
+    <li><a class="dropdown-item" href="../articles/functional_api.html">Functional API</a></li>
+    <li><h6 class="dropdown-header" data-toc-skip>Extending and customizing</h6></li>
+    <li><a class="dropdown-item" href="../articles/training_with_built_in_methods.html">Training &amp; evaluation with the built-in methods</a></li>
+    <li><a class="dropdown-item" href="../articles/custom_train_step_in_tensorflow.html">Customizing `fit()` with Tensorflow</a></li>
+    <li><a class="dropdown-item" href="../articles/writing_your_own_callbacks.html">Writing your own callbacks</a></li>
+    <li><a class="dropdown-item" href="../articles/making_new_layers_and_models_via_subclassing.html">Making new layers and models via subclassing</a></li>
+    <li><a class="dropdown-item" href="../articles/writing_a_custom_training_loop_in_tensorflow.html">Writing a training loop from scratch in TensorFlow</a></li>
+    <li><a class="dropdown-item" href="../articles/serialization_and_saving.html">Serialization and Saving</a></li>
+    <li><h6 class="dropdown-header" data-toc-skip>Other topics</h6></li>
+    <li><a class="dropdown-item" href="../articles/transfer_learning.html">Transfer learning and fine tuning</a></li>
+    <li><a class="dropdown-item" href="../articles/distributed_training_with_tensorflow.html">Distributed training with TensorFlow</a></li>
+    <li><a class="dropdown-item" href="../articles/distribution.html">Distributed training with Jax</a></li>
+  </ul></li>
+<li class="nav-item"><a class="nav-link" href="../articles/examples/index.html">Examples</a></li>
+<li class="active nav-item"><a class="nav-link" href="../reference/index.html">Reference</a></li>
+<li class="nav-item"><a class="nav-link" href="../news/index.html">News</a></li>
+      </ul><ul class="navbar-nav"><li class="nav-item"><form class="form-inline" role="search">
+ <input class="form-control" type="search" name="search-input" id="search-input" autocomplete="off" aria-label="Search site" placeholder="Search for" data-search-index="../search.json"></form></li>
+<li class="nav-item"><a class="external-link nav-link" href="https://github.com/rstudio/keras3/" aria-label="GitHub"><span class="fa fab fa-github fa-lg"></span></a></li>
+      </ul></div>
+
+
+  </div>
+</nav><div class="container template-reference-topic">
+<div class="row">
+  <main id="main" class="col-md-9"><div class="page-header">
+
+      <h1>A preprocessing layer that applies random elastic transformations.</h1>
+      <small class="dont-index">Source: <a href="https://github.com/rstudio/keras3/blob/HEAD/R/layers-preprocessing.R" class="external-link"><code>R/layers-preprocessing.R</code></a></small>
+      <div class="d-none name"><code>layer_random_elastic_transform.Rd</code></div>
+    </div>
+
+    <div class="ref-description section level2">
+    <p>This layer distorts input images by applying elastic deformations,
+simulating a physically realistic transformation. The magnitude of the
+distortion is controlled by the <code>scale</code> parameter, while the <code>factor</code>
+determines the probability of applying the transformation.</p>
+    </div>
+
+    <div class="section level2">
+    <h2 id="ref-usage">Usage<a class="anchor" aria-label="anchor" href="#ref-usage"></a></h2>
+    <div class="sourceCode"><pre class="sourceCode r"><code><span><span class="fu">layer_random_elastic_transform</span><span class="op">(</span></span>
+<span>  <span class="va">object</span>,</span>
+<span>  factor <span class="op">=</span> <span class="fl">1</span>,</span>
+<span>  scale <span class="op">=</span> <span class="fl">1</span>,</span>
+<span>  interpolation <span class="op">=</span> <span class="st">"bilinear"</span>,</span>
+<span>  fill_mode <span class="op">=</span> <span class="st">"reflect"</span>,</span>
+<span>  fill_value <span class="op">=</span> <span class="fl">0</span>,</span>
+<span>  value_range <span class="op">=</span> <span class="fu"><a href="https://rdrr.io/r/base/list.html" class="external-link">list</a></span><span class="op">(</span><span class="fl">0L</span>, <span class="fl">255L</span><span class="op">)</span>,</span>
+<span>  seed <span class="op">=</span> <span class="cn">NULL</span>,</span>
+<span>  data_format <span class="op">=</span> <span class="cn">NULL</span>,</span>
+<span>  <span class="va">...</span></span>
+<span><span class="op">)</span></span></code></pre></div>
+    </div>
+
+    <div class="section level2">
+    <h2 id="arguments">Arguments<a class="anchor" aria-label="anchor" href="#arguments"></a></h2>
+
+
+<dl><dt id="arg-object">object<a class="anchor" aria-label="anchor" href="#arg-object"></a></dt>
+<dd><p>Object to compose the layer with. A tensor, array, or sequential model.</p></dd>
+
+
+<dt id="arg-factor">factor<a class="anchor" aria-label="anchor" href="#arg-factor"></a></dt>
+<dd><p>A single float or a tuple of two floats.
+<code>factor</code> controls the probability of applying the transformation.</p><ul><li><p><code>factor = 0.0</code> ensures no transformation is applied.</p></li>
+<li><p><code>factor = 1.0</code> means the transformation is always applied.</p></li>
+<li><p>If a tuple <code>(min, max)</code> is provided, a probability value
+is sampled between <code>min</code> and <code>max</code> for each image.</p></li>
+<li><p>If a single float is provided, a probability is sampled
+between <code>0.0</code> and the given float.
+Default is <code>1.0</code>.</p></li>
+</ul></dd>
+
+
+<dt id="arg-scale">scale<a class="anchor" aria-label="anchor" href="#arg-scale"></a></dt>
+<dd><p>A float or a tuple of two floats defining the magnitude of
+the distortion applied.</p><ul><li><p>If a tuple <code>(min, max)</code> is provided, a random scale value is
+sampled within this range.</p></li>
+<li><p>If a single float is provided, a random scale value is sampled
+between <code>0.0</code> and the given float.
+Default is <code>1.0</code>.</p></li>
+</ul></dd>
+
+
+<dt id="arg-interpolation">interpolation<a class="anchor" aria-label="anchor" href="#arg-interpolation"></a></dt>
+<dd><p>Interpolation mode. Supported values: <code>"nearest"</code>,
+<code>"bilinear"</code>.</p></dd>
+
+
+<dt id="arg-fill-mode">fill_mode<a class="anchor" aria-label="anchor" href="#arg-fill-mode"></a></dt>
+<dd><p>Points outside the boundaries of the input are filled
+according to the given mode. Available methods are <code>"constant"</code>,
+<code>"nearest"</code>, <code>"wrap"</code> and <code>"reflect"</code>. Defaults to <code>"reflect"</code>.</p><ul><li><p><code>"reflect"</code>: <code>(d c b a | a b c d | d c b a)</code>
+The input is extended by reflecting about the edge of the last
+pixel.</p></li>
+<li><p><code>"constant"</code>: <code>(k k k k | a b c d | k k k k)</code>
+The input is extended by filling all values beyond
+the edge with the same constant value <code>k</code> specified by
+<code>fill_value</code>.</p></li>
+<li><p><code>"wrap"</code>: <code>(a b c d | a b c d | a b c d)</code>
+The input is extended by wrapping around to the opposite edge.</p></li>
+<li><p><code>"nearest"</code>: <code>(a a a a | a b c d | d d d d)</code>
+The input is extended by the nearest pixel.
+When using the torch backend, <code>"reflect"</code> is redirected to
+<code>"mirror"</code> because torch does not support <code>"reflect"</code>.
+The torch backend also does not support <code>"wrap"</code>.</p></li>
+</ul></dd>
+
+
+<dt id="arg-fill-value">fill_value<a class="anchor" aria-label="anchor" href="#arg-fill-value"></a></dt>
+<dd><p>A float representing the value to fill outside the boundaries when
+<code>fill_mode = "constant"</code>.</p></dd>
+
+
+<dt id="arg-value-range">value_range<a class="anchor" aria-label="anchor" href="#arg-value-range"></a></dt>
+<dd><p>The range of values the incoming images will have.
+Represented as a two-number tuple written <code>[low, high]</code>. This is
+typically either <code>[0, 1]</code> or <code>[0, 255]</code> depending on how your
+preprocessing pipeline is set up.</p></dd>
+
+
+<dt id="arg-seed">seed<a class="anchor" aria-label="anchor" href="#arg-seed"></a></dt>
+<dd><p>Integer. Used to create a random seed.</p></dd>
+
+
+<dt id="arg-data-format">data_format<a class="anchor" aria-label="anchor" href="#arg-data-format"></a></dt>
+<dd><p>string, either <code>"channels_last"</code> or <code>"channels_first"</code>.
+The ordering of the dimensions in the inputs. <code>"channels_last"</code>
+corresponds to inputs with shape <code>(batch, height, width, channels)</code>
+while <code>"channels_first"</code> corresponds to inputs with shape
+<code>(batch, channels, height, width)</code>. It defaults to the
+<code>image_data_format</code> value found in your Keras config file at
+<code>~/.keras/keras.json</code>. If you never set it, then it will be
+<code>"channels_last"</code>.</p></dd>
+
+
+<dt id="arg--">...<a class="anchor" aria-label="anchor" href="#arg--"></a></dt>
+<dd><p>For forward/backward compatability.</p></dd>
+
+</dl></div>
+    <div class="section level2">
+    <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
+    <div class="dont-index"><p>Other image preprocessing layers: <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br></p>
+<p>Other preprocessing layers: <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_category_encoding.html">layer_category_encoding</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_discretization.html">layer_discretization</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_feature_space.html">layer_feature_space</a>()</code> <br><code><a href="layer_hashed_crossing.html">layer_hashed_crossing</a>()</code> <br><code><a href="layer_hashing.html">layer_hashing</a>()</code> <br><code><a href="layer_integer_lookup.html">layer_integer_lookup</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_mel_spectrogram.html">layer_mel_spectrogram</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_normalization.html">layer_normalization</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_brightness.html">layer_random_brightness</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_contrast.html">layer_random_contrast</a>()</code> <br><code><a href="layer_random_crop.html">layer_random_crop</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_flip.html">layer_random_flip</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_rotation.html">layer_random_rotation</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_random_translation.html">layer_random_translation</a>()</code> <br><code><a href="layer_random_zoom.html">layer_random_zoom</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br><code><a href="layer_stft_spectrogram.html">layer_stft_spectrogram</a>()</code> <br><code><a href="layer_string_lookup.html">layer_string_lookup</a>()</code> <br><code><a href="layer_text_vectorization.html">layer_text_vectorization</a>()</code> <br></p>
+<p>Other layers: <br><code><a href="Layer.html">Layer</a>()</code> <br><code><a href="layer_activation.html">layer_activation</a>()</code> <br><code><a href="layer_activation_elu.html">layer_activation_elu</a>()</code> <br><code><a href="layer_activation_leaky_relu.html">layer_activation_leaky_relu</a>()</code> <br><code><a href="layer_activation_parametric_relu.html">layer_activation_parametric_relu</a>()</code> <br><code><a href="layer_activation_relu.html">layer_activation_relu</a>()</code> <br><code><a href="layer_activation_softmax.html">layer_activation_softmax</a>()</code> <br><code><a href="layer_activity_regularization.html">layer_activity_regularization</a>()</code> <br><code><a href="layer_add.html">layer_add</a>()</code> <br><code><a href="layer_additive_attention.html">layer_additive_attention</a>()</code> <br><code><a href="layer_alpha_dropout.html">layer_alpha_dropout</a>()</code> <br><code><a href="layer_attention.html">layer_attention</a>()</code> <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_average.html">layer_average</a>()</code> <br><code><a href="layer_average_pooling_1d.html">layer_average_pooling_1d</a>()</code> <br><code><a href="layer_average_pooling_2d.html">layer_average_pooling_2d</a>()</code> <br><code><a href="layer_average_pooling_3d.html">layer_average_pooling_3d</a>()</code> <br><code><a href="layer_batch_normalization.html">layer_batch_normalization</a>()</code> <br><code><a href="layer_bidirectional.html">layer_bidirectional</a>()</code> <br><code><a href="layer_category_encoding.html">layer_category_encoding</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_concatenate.html">layer_concatenate</a>()</code> <br><code><a href="layer_conv_1d.html">layer_conv_1d</a>()</code> <br><code><a href="layer_conv_1d_transpose.html">layer_conv_1d_transpose</a>()</code> <br><code><a href="layer_conv_2d.html">layer_conv_2d</a>()</code> <br><code><a href="layer_conv_2d_transpose.html">layer_conv_2d_transpose</a>()</code> <br><code><a href="layer_conv_3d.html">layer_conv_3d</a>()</code> <br><code><a href="layer_conv_3d_transpose.html">layer_conv_3d_transpose</a>()</code> <br><code><a href="layer_conv_lstm_1d.html">layer_conv_lstm_1d</a>()</code> <br><code><a href="layer_conv_lstm_2d.html">layer_conv_lstm_2d</a>()</code> <br><code><a href="layer_conv_lstm_3d.html">layer_conv_lstm_3d</a>()</code> <br><code><a href="layer_cropping_1d.html">layer_cropping_1d</a>()</code> <br><code><a href="layer_cropping_2d.html">layer_cropping_2d</a>()</code> <br><code><a href="layer_cropping_3d.html">layer_cropping_3d</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_dense.html">layer_dense</a>()</code> <br><code><a href="layer_depthwise_conv_1d.html">layer_depthwise_conv_1d</a>()</code> <br><code><a href="layer_depthwise_conv_2d.html">layer_depthwise_conv_2d</a>()</code> <br><code><a href="layer_discretization.html">layer_discretization</a>()</code> <br><code><a href="layer_dot.html">layer_dot</a>()</code> <br><code><a href="layer_dropout.html">layer_dropout</a>()</code> <br><code><a href="layer_einsum_dense.html">layer_einsum_dense</a>()</code> <br><code><a href="layer_embedding.html">layer_embedding</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_feature_space.html">layer_feature_space</a>()</code> <br><code><a href="layer_flatten.html">layer_flatten</a>()</code> <br><code><a href="layer_flax_module_wrapper.html">layer_flax_module_wrapper</a>()</code> <br><code><a href="layer_gaussian_dropout.html">layer_gaussian_dropout</a>()</code> <br><code><a href="layer_gaussian_noise.html">layer_gaussian_noise</a>()</code> <br><code><a href="layer_global_average_pooling_1d.html">layer_global_average_pooling_1d</a>()</code> <br><code><a href="layer_global_average_pooling_2d.html">layer_global_average_pooling_2d</a>()</code> <br><code><a href="layer_global_average_pooling_3d.html">layer_global_average_pooling_3d</a>()</code> <br><code><a href="layer_global_max_pooling_1d.html">layer_global_max_pooling_1d</a>()</code> <br><code><a href="layer_global_max_pooling_2d.html">layer_global_max_pooling_2d</a>()</code> <br><code><a href="layer_global_max_pooling_3d.html">layer_global_max_pooling_3d</a>()</code> <br><code><a href="layer_group_normalization.html">layer_group_normalization</a>()</code> <br><code><a href="layer_group_query_attention.html">layer_group_query_attention</a>()</code> <br><code><a href="layer_gru.html">layer_gru</a>()</code> <br><code><a href="layer_hashed_crossing.html">layer_hashed_crossing</a>()</code> <br><code><a href="layer_hashing.html">layer_hashing</a>()</code> <br><code><a href="layer_identity.html">layer_identity</a>()</code> <br><code><a href="layer_integer_lookup.html">layer_integer_lookup</a>()</code> <br><code><a href="layer_jax_model_wrapper.html">layer_jax_model_wrapper</a>()</code> <br><code><a href="layer_lambda.html">layer_lambda</a>()</code> <br><code><a href="layer_layer_normalization.html">layer_layer_normalization</a>()</code> <br><code><a href="layer_lstm.html">layer_lstm</a>()</code> <br><code><a href="layer_masking.html">layer_masking</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_max_pooling_1d.html">layer_max_pooling_1d</a>()</code> <br><code><a href="layer_max_pooling_2d.html">layer_max_pooling_2d</a>()</code> <br><code><a href="layer_max_pooling_3d.html">layer_max_pooling_3d</a>()</code> <br><code><a href="layer_maximum.html">layer_maximum</a>()</code> <br><code><a href="layer_mel_spectrogram.html">layer_mel_spectrogram</a>()</code> <br><code><a href="layer_minimum.html">layer_minimum</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_multi_head_attention.html">layer_multi_head_attention</a>()</code> <br><code><a href="layer_multiply.html">layer_multiply</a>()</code> <br><code><a href="layer_normalization.html">layer_normalization</a>()</code> <br><code><a href="layer_permute.html">layer_permute</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_brightness.html">layer_random_brightness</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_contrast.html">layer_random_contrast</a>()</code> <br><code><a href="layer_random_crop.html">layer_random_crop</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_flip.html">layer_random_flip</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_rotation.html">layer_random_rotation</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_random_translation.html">layer_random_translation</a>()</code> <br><code><a href="layer_random_zoom.html">layer_random_zoom</a>()</code> <br><code><a href="layer_repeat_vector.html">layer_repeat_vector</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_reshape.html">layer_reshape</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_rms_normalization.html">layer_rms_normalization</a>()</code> <br><code><a href="layer_rnn.html">layer_rnn</a>()</code> <br><code><a href="layer_separable_conv_1d.html">layer_separable_conv_1d</a>()</code> <br><code><a href="layer_separable_conv_2d.html">layer_separable_conv_2d</a>()</code> <br><code><a href="layer_simple_rnn.html">layer_simple_rnn</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br><code><a href="layer_spatial_dropout_1d.html">layer_spatial_dropout_1d</a>()</code> <br><code><a href="layer_spatial_dropout_2d.html">layer_spatial_dropout_2d</a>()</code> <br><code><a href="layer_spatial_dropout_3d.html">layer_spatial_dropout_3d</a>()</code> <br><code><a href="layer_spectral_normalization.html">layer_spectral_normalization</a>()</code> <br><code><a href="layer_stft_spectrogram.html">layer_stft_spectrogram</a>()</code> <br><code><a href="layer_string_lookup.html">layer_string_lookup</a>()</code> <br><code><a href="layer_subtract.html">layer_subtract</a>()</code> <br><code><a href="layer_text_vectorization.html">layer_text_vectorization</a>()</code> <br><code><a href="layer_tfsm.html">layer_tfsm</a>()</code> <br><code><a href="layer_time_distributed.html">layer_time_distributed</a>()</code> <br><code><a href="layer_torch_module_wrapper.html">layer_torch_module_wrapper</a>()</code> <br><code><a href="layer_unit_normalization.html">layer_unit_normalization</a>()</code> <br><code><a href="layer_upsampling_1d.html">layer_upsampling_1d</a>()</code> <br><code><a href="layer_upsampling_2d.html">layer_upsampling_2d</a>()</code> <br><code><a href="layer_upsampling_3d.html">layer_upsampling_3d</a>()</code> <br><code><a href="layer_zero_padding_1d.html">layer_zero_padding_1d</a>()</code> <br><code><a href="layer_zero_padding_2d.html">layer_zero_padding_2d</a>()</code> <br><code><a href="layer_zero_padding_3d.html">layer_zero_padding_3d</a>()</code> <br><code><a href="rnn_cell_gru.html">rnn_cell_gru</a>()</code> <br><code><a href="rnn_cell_lstm.html">rnn_cell_lstm</a>()</code> <br><code><a href="rnn_cell_simple.html">rnn_cell_simple</a>()</code> <br><code><a href="rnn_cells_stack.html">rnn_cells_stack</a>()</code> <br></p></div>
+    </div>
+
+  </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
+    </nav></aside></div>
+
+
+    <footer><div class="pkgdown-footer-left">
+  <p>Developed by Tomasz Kalinowski, JJ Allaire, François Chollet, Posit Software, PBC, Google.</p>
+</div>
+
+<div class="pkgdown-footer-right">
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
+</div>
+
+    </footer></div>
+
+
+
+
+
+  </body></html>
+
diff --git a/docs/dev/reference/layer_random_erasing.html b/docs/dev/reference/layer_random_erasing.html
index 2feee009a..0b7b39b38 100644
--- a/docs/dev/reference/layer_random_erasing.html
+++ b/docs/dev/reference/layer_random_erasing.html
@@ -12,7 +12,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -143,9 +143,9 @@ <h2 id="references">References<a class="anchor" aria-label="anchor" href="#refer
 </ul></div>
     <div class="section level2">
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
-    <div class="dont-index"><p>Other image preprocessing layers: <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br></p>
-<p>Other preprocessing layers: <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_category_encoding.html">layer_category_encoding</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_discretization.html">layer_discretization</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_feature_space.html">layer_feature_space</a>()</code> <br><code><a href="layer_hashed_crossing.html">layer_hashed_crossing</a>()</code> <br><code><a href="layer_hashing.html">layer_hashing</a>()</code> <br><code><a href="layer_integer_lookup.html">layer_integer_lookup</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_mel_spectrogram.html">layer_mel_spectrogram</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_normalization.html">layer_normalization</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_brightness.html">layer_random_brightness</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_contrast.html">layer_random_contrast</a>()</code> <br><code><a href="layer_random_crop.html">layer_random_crop</a>()</code> <br><code><a href="layer_random_flip.html">layer_random_flip</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_rotation.html">layer_random_rotation</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_random_translation.html">layer_random_translation</a>()</code> <br><code><a href="layer_random_zoom.html">layer_random_zoom</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br><code><a href="layer_stft_spectrogram.html">layer_stft_spectrogram</a>()</code> <br><code><a href="layer_string_lookup.html">layer_string_lookup</a>()</code> <br><code><a href="layer_text_vectorization.html">layer_text_vectorization</a>()</code> <br></p>
-<p>Other layers: <br><code><a href="Layer.html">Layer</a>()</code> <br><code><a href="layer_activation.html">layer_activation</a>()</code> <br><code><a href="layer_activation_elu.html">layer_activation_elu</a>()</code> <br><code><a href="layer_activation_leaky_relu.html">layer_activation_leaky_relu</a>()</code> <br><code><a href="layer_activation_parametric_relu.html">layer_activation_parametric_relu</a>()</code> <br><code><a href="layer_activation_relu.html">layer_activation_relu</a>()</code> <br><code><a href="layer_activation_softmax.html">layer_activation_softmax</a>()</code> <br><code><a href="layer_activity_regularization.html">layer_activity_regularization</a>()</code> <br><code><a href="layer_add.html">layer_add</a>()</code> <br><code><a href="layer_additive_attention.html">layer_additive_attention</a>()</code> <br><code><a href="layer_alpha_dropout.html">layer_alpha_dropout</a>()</code> <br><code><a href="layer_attention.html">layer_attention</a>()</code> <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_average.html">layer_average</a>()</code> <br><code><a href="layer_average_pooling_1d.html">layer_average_pooling_1d</a>()</code> <br><code><a href="layer_average_pooling_2d.html">layer_average_pooling_2d</a>()</code> <br><code><a href="layer_average_pooling_3d.html">layer_average_pooling_3d</a>()</code> <br><code><a href="layer_batch_normalization.html">layer_batch_normalization</a>()</code> <br><code><a href="layer_bidirectional.html">layer_bidirectional</a>()</code> <br><code><a href="layer_category_encoding.html">layer_category_encoding</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_concatenate.html">layer_concatenate</a>()</code> <br><code><a href="layer_conv_1d.html">layer_conv_1d</a>()</code> <br><code><a href="layer_conv_1d_transpose.html">layer_conv_1d_transpose</a>()</code> <br><code><a href="layer_conv_2d.html">layer_conv_2d</a>()</code> <br><code><a href="layer_conv_2d_transpose.html">layer_conv_2d_transpose</a>()</code> <br><code><a href="layer_conv_3d.html">layer_conv_3d</a>()</code> <br><code><a href="layer_conv_3d_transpose.html">layer_conv_3d_transpose</a>()</code> <br><code><a href="layer_conv_lstm_1d.html">layer_conv_lstm_1d</a>()</code> <br><code><a href="layer_conv_lstm_2d.html">layer_conv_lstm_2d</a>()</code> <br><code><a href="layer_conv_lstm_3d.html">layer_conv_lstm_3d</a>()</code> <br><code><a href="layer_cropping_1d.html">layer_cropping_1d</a>()</code> <br><code><a href="layer_cropping_2d.html">layer_cropping_2d</a>()</code> <br><code><a href="layer_cropping_3d.html">layer_cropping_3d</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_dense.html">layer_dense</a>()</code> <br><code><a href="layer_depthwise_conv_1d.html">layer_depthwise_conv_1d</a>()</code> <br><code><a href="layer_depthwise_conv_2d.html">layer_depthwise_conv_2d</a>()</code> <br><code><a href="layer_discretization.html">layer_discretization</a>()</code> <br><code><a href="layer_dot.html">layer_dot</a>()</code> <br><code><a href="layer_dropout.html">layer_dropout</a>()</code> <br><code><a href="layer_einsum_dense.html">layer_einsum_dense</a>()</code> <br><code><a href="layer_embedding.html">layer_embedding</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_feature_space.html">layer_feature_space</a>()</code> <br><code><a href="layer_flatten.html">layer_flatten</a>()</code> <br><code><a href="layer_flax_module_wrapper.html">layer_flax_module_wrapper</a>()</code> <br><code><a href="layer_gaussian_dropout.html">layer_gaussian_dropout</a>()</code> <br><code><a href="layer_gaussian_noise.html">layer_gaussian_noise</a>()</code> <br><code><a href="layer_global_average_pooling_1d.html">layer_global_average_pooling_1d</a>()</code> <br><code><a href="layer_global_average_pooling_2d.html">layer_global_average_pooling_2d</a>()</code> <br><code><a href="layer_global_average_pooling_3d.html">layer_global_average_pooling_3d</a>()</code> <br><code><a href="layer_global_max_pooling_1d.html">layer_global_max_pooling_1d</a>()</code> <br><code><a href="layer_global_max_pooling_2d.html">layer_global_max_pooling_2d</a>()</code> <br><code><a href="layer_global_max_pooling_3d.html">layer_global_max_pooling_3d</a>()</code> <br><code><a href="layer_group_normalization.html">layer_group_normalization</a>()</code> <br><code><a href="layer_group_query_attention.html">layer_group_query_attention</a>()</code> <br><code><a href="layer_gru.html">layer_gru</a>()</code> <br><code><a href="layer_hashed_crossing.html">layer_hashed_crossing</a>()</code> <br><code><a href="layer_hashing.html">layer_hashing</a>()</code> <br><code><a href="layer_identity.html">layer_identity</a>()</code> <br><code><a href="layer_integer_lookup.html">layer_integer_lookup</a>()</code> <br><code><a href="layer_jax_model_wrapper.html">layer_jax_model_wrapper</a>()</code> <br><code><a href="layer_lambda.html">layer_lambda</a>()</code> <br><code><a href="layer_layer_normalization.html">layer_layer_normalization</a>()</code> <br><code><a href="layer_lstm.html">layer_lstm</a>()</code> <br><code><a href="layer_masking.html">layer_masking</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_max_pooling_1d.html">layer_max_pooling_1d</a>()</code> <br><code><a href="layer_max_pooling_2d.html">layer_max_pooling_2d</a>()</code> <br><code><a href="layer_max_pooling_3d.html">layer_max_pooling_3d</a>()</code> <br><code><a href="layer_maximum.html">layer_maximum</a>()</code> <br><code><a href="layer_mel_spectrogram.html">layer_mel_spectrogram</a>()</code> <br><code><a href="layer_minimum.html">layer_minimum</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_multi_head_attention.html">layer_multi_head_attention</a>()</code> <br><code><a href="layer_multiply.html">layer_multiply</a>()</code> <br><code><a href="layer_normalization.html">layer_normalization</a>()</code> <br><code><a href="layer_permute.html">layer_permute</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_brightness.html">layer_random_brightness</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_contrast.html">layer_random_contrast</a>()</code> <br><code><a href="layer_random_crop.html">layer_random_crop</a>()</code> <br><code><a href="layer_random_flip.html">layer_random_flip</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_rotation.html">layer_random_rotation</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_random_translation.html">layer_random_translation</a>()</code> <br><code><a href="layer_random_zoom.html">layer_random_zoom</a>()</code> <br><code><a href="layer_repeat_vector.html">layer_repeat_vector</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_reshape.html">layer_reshape</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_rms_normalization.html">layer_rms_normalization</a>()</code> <br><code><a href="layer_rnn.html">layer_rnn</a>()</code> <br><code><a href="layer_separable_conv_1d.html">layer_separable_conv_1d</a>()</code> <br><code><a href="layer_separable_conv_2d.html">layer_separable_conv_2d</a>()</code> <br><code><a href="layer_simple_rnn.html">layer_simple_rnn</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br><code><a href="layer_spatial_dropout_1d.html">layer_spatial_dropout_1d</a>()</code> <br><code><a href="layer_spatial_dropout_2d.html">layer_spatial_dropout_2d</a>()</code> <br><code><a href="layer_spatial_dropout_3d.html">layer_spatial_dropout_3d</a>()</code> <br><code><a href="layer_spectral_normalization.html">layer_spectral_normalization</a>()</code> <br><code><a href="layer_stft_spectrogram.html">layer_stft_spectrogram</a>()</code> <br><code><a href="layer_string_lookup.html">layer_string_lookup</a>()</code> <br><code><a href="layer_subtract.html">layer_subtract</a>()</code> <br><code><a href="layer_text_vectorization.html">layer_text_vectorization</a>()</code> <br><code><a href="layer_tfsm.html">layer_tfsm</a>()</code> <br><code><a href="layer_time_distributed.html">layer_time_distributed</a>()</code> <br><code><a href="layer_torch_module_wrapper.html">layer_torch_module_wrapper</a>()</code> <br><code><a href="layer_unit_normalization.html">layer_unit_normalization</a>()</code> <br><code><a href="layer_upsampling_1d.html">layer_upsampling_1d</a>()</code> <br><code><a href="layer_upsampling_2d.html">layer_upsampling_2d</a>()</code> <br><code><a href="layer_upsampling_3d.html">layer_upsampling_3d</a>()</code> <br><code><a href="layer_zero_padding_1d.html">layer_zero_padding_1d</a>()</code> <br><code><a href="layer_zero_padding_2d.html">layer_zero_padding_2d</a>()</code> <br><code><a href="layer_zero_padding_3d.html">layer_zero_padding_3d</a>()</code> <br><code><a href="rnn_cell_gru.html">rnn_cell_gru</a>()</code> <br><code><a href="rnn_cell_lstm.html">rnn_cell_lstm</a>()</code> <br><code><a href="rnn_cell_simple.html">rnn_cell_simple</a>()</code> <br><code><a href="rnn_cells_stack.html">rnn_cells_stack</a>()</code> <br></p></div>
+    <div class="dont-index"><p>Other image preprocessing layers: <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_elastic_transform.html">layer_random_elastic_transform</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br></p>
+<p>Other preprocessing layers: <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_category_encoding.html">layer_category_encoding</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_discretization.html">layer_discretization</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_feature_space.html">layer_feature_space</a>()</code> <br><code><a href="layer_hashed_crossing.html">layer_hashed_crossing</a>()</code> <br><code><a href="layer_hashing.html">layer_hashing</a>()</code> <br><code><a href="layer_integer_lookup.html">layer_integer_lookup</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_mel_spectrogram.html">layer_mel_spectrogram</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_normalization.html">layer_normalization</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_brightness.html">layer_random_brightness</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_contrast.html">layer_random_contrast</a>()</code> <br><code><a href="layer_random_crop.html">layer_random_crop</a>()</code> <br><code><a href="layer_random_elastic_transform.html">layer_random_elastic_transform</a>()</code> <br><code><a href="layer_random_flip.html">layer_random_flip</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_rotation.html">layer_random_rotation</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_random_translation.html">layer_random_translation</a>()</code> <br><code><a href="layer_random_zoom.html">layer_random_zoom</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br><code><a href="layer_stft_spectrogram.html">layer_stft_spectrogram</a>()</code> <br><code><a href="layer_string_lookup.html">layer_string_lookup</a>()</code> <br><code><a href="layer_text_vectorization.html">layer_text_vectorization</a>()</code> <br></p>
+<p>Other layers: <br><code><a href="Layer.html">Layer</a>()</code> <br><code><a href="layer_activation.html">layer_activation</a>()</code> <br><code><a href="layer_activation_elu.html">layer_activation_elu</a>()</code> <br><code><a href="layer_activation_leaky_relu.html">layer_activation_leaky_relu</a>()</code> <br><code><a href="layer_activation_parametric_relu.html">layer_activation_parametric_relu</a>()</code> <br><code><a href="layer_activation_relu.html">layer_activation_relu</a>()</code> <br><code><a href="layer_activation_softmax.html">layer_activation_softmax</a>()</code> <br><code><a href="layer_activity_regularization.html">layer_activity_regularization</a>()</code> <br><code><a href="layer_add.html">layer_add</a>()</code> <br><code><a href="layer_additive_attention.html">layer_additive_attention</a>()</code> <br><code><a href="layer_alpha_dropout.html">layer_alpha_dropout</a>()</code> <br><code><a href="layer_attention.html">layer_attention</a>()</code> <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_average.html">layer_average</a>()</code> <br><code><a href="layer_average_pooling_1d.html">layer_average_pooling_1d</a>()</code> <br><code><a href="layer_average_pooling_2d.html">layer_average_pooling_2d</a>()</code> <br><code><a href="layer_average_pooling_3d.html">layer_average_pooling_3d</a>()</code> <br><code><a href="layer_batch_normalization.html">layer_batch_normalization</a>()</code> <br><code><a href="layer_bidirectional.html">layer_bidirectional</a>()</code> <br><code><a href="layer_category_encoding.html">layer_category_encoding</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_concatenate.html">layer_concatenate</a>()</code> <br><code><a href="layer_conv_1d.html">layer_conv_1d</a>()</code> <br><code><a href="layer_conv_1d_transpose.html">layer_conv_1d_transpose</a>()</code> <br><code><a href="layer_conv_2d.html">layer_conv_2d</a>()</code> <br><code><a href="layer_conv_2d_transpose.html">layer_conv_2d_transpose</a>()</code> <br><code><a href="layer_conv_3d.html">layer_conv_3d</a>()</code> <br><code><a href="layer_conv_3d_transpose.html">layer_conv_3d_transpose</a>()</code> <br><code><a href="layer_conv_lstm_1d.html">layer_conv_lstm_1d</a>()</code> <br><code><a href="layer_conv_lstm_2d.html">layer_conv_lstm_2d</a>()</code> <br><code><a href="layer_conv_lstm_3d.html">layer_conv_lstm_3d</a>()</code> <br><code><a href="layer_cropping_1d.html">layer_cropping_1d</a>()</code> <br><code><a href="layer_cropping_2d.html">layer_cropping_2d</a>()</code> <br><code><a href="layer_cropping_3d.html">layer_cropping_3d</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_dense.html">layer_dense</a>()</code> <br><code><a href="layer_depthwise_conv_1d.html">layer_depthwise_conv_1d</a>()</code> <br><code><a href="layer_depthwise_conv_2d.html">layer_depthwise_conv_2d</a>()</code> <br><code><a href="layer_discretization.html">layer_discretization</a>()</code> <br><code><a href="layer_dot.html">layer_dot</a>()</code> <br><code><a href="layer_dropout.html">layer_dropout</a>()</code> <br><code><a href="layer_einsum_dense.html">layer_einsum_dense</a>()</code> <br><code><a href="layer_embedding.html">layer_embedding</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_feature_space.html">layer_feature_space</a>()</code> <br><code><a href="layer_flatten.html">layer_flatten</a>()</code> <br><code><a href="layer_flax_module_wrapper.html">layer_flax_module_wrapper</a>()</code> <br><code><a href="layer_gaussian_dropout.html">layer_gaussian_dropout</a>()</code> <br><code><a href="layer_gaussian_noise.html">layer_gaussian_noise</a>()</code> <br><code><a href="layer_global_average_pooling_1d.html">layer_global_average_pooling_1d</a>()</code> <br><code><a href="layer_global_average_pooling_2d.html">layer_global_average_pooling_2d</a>()</code> <br><code><a href="layer_global_average_pooling_3d.html">layer_global_average_pooling_3d</a>()</code> <br><code><a href="layer_global_max_pooling_1d.html">layer_global_max_pooling_1d</a>()</code> <br><code><a href="layer_global_max_pooling_2d.html">layer_global_max_pooling_2d</a>()</code> <br><code><a href="layer_global_max_pooling_3d.html">layer_global_max_pooling_3d</a>()</code> <br><code><a href="layer_group_normalization.html">layer_group_normalization</a>()</code> <br><code><a href="layer_group_query_attention.html">layer_group_query_attention</a>()</code> <br><code><a href="layer_gru.html">layer_gru</a>()</code> <br><code><a href="layer_hashed_crossing.html">layer_hashed_crossing</a>()</code> <br><code><a href="layer_hashing.html">layer_hashing</a>()</code> <br><code><a href="layer_identity.html">layer_identity</a>()</code> <br><code><a href="layer_integer_lookup.html">layer_integer_lookup</a>()</code> <br><code><a href="layer_jax_model_wrapper.html">layer_jax_model_wrapper</a>()</code> <br><code><a href="layer_lambda.html">layer_lambda</a>()</code> <br><code><a href="layer_layer_normalization.html">layer_layer_normalization</a>()</code> <br><code><a href="layer_lstm.html">layer_lstm</a>()</code> <br><code><a href="layer_masking.html">layer_masking</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_max_pooling_1d.html">layer_max_pooling_1d</a>()</code> <br><code><a href="layer_max_pooling_2d.html">layer_max_pooling_2d</a>()</code> <br><code><a href="layer_max_pooling_3d.html">layer_max_pooling_3d</a>()</code> <br><code><a href="layer_maximum.html">layer_maximum</a>()</code> <br><code><a href="layer_mel_spectrogram.html">layer_mel_spectrogram</a>()</code> <br><code><a href="layer_minimum.html">layer_minimum</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_multi_head_attention.html">layer_multi_head_attention</a>()</code> <br><code><a href="layer_multiply.html">layer_multiply</a>()</code> <br><code><a href="layer_normalization.html">layer_normalization</a>()</code> <br><code><a href="layer_permute.html">layer_permute</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_brightness.html">layer_random_brightness</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_contrast.html">layer_random_contrast</a>()</code> <br><code><a href="layer_random_crop.html">layer_random_crop</a>()</code> <br><code><a href="layer_random_elastic_transform.html">layer_random_elastic_transform</a>()</code> <br><code><a href="layer_random_flip.html">layer_random_flip</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_rotation.html">layer_random_rotation</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_random_translation.html">layer_random_translation</a>()</code> <br><code><a href="layer_random_zoom.html">layer_random_zoom</a>()</code> <br><code><a href="layer_repeat_vector.html">layer_repeat_vector</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_reshape.html">layer_reshape</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_rms_normalization.html">layer_rms_normalization</a>()</code> <br><code><a href="layer_rnn.html">layer_rnn</a>()</code> <br><code><a href="layer_separable_conv_1d.html">layer_separable_conv_1d</a>()</code> <br><code><a href="layer_separable_conv_2d.html">layer_separable_conv_2d</a>()</code> <br><code><a href="layer_simple_rnn.html">layer_simple_rnn</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br><code><a href="layer_spatial_dropout_1d.html">layer_spatial_dropout_1d</a>()</code> <br><code><a href="layer_spatial_dropout_2d.html">layer_spatial_dropout_2d</a>()</code> <br><code><a href="layer_spatial_dropout_3d.html">layer_spatial_dropout_3d</a>()</code> <br><code><a href="layer_spectral_normalization.html">layer_spectral_normalization</a>()</code> <br><code><a href="layer_stft_spectrogram.html">layer_stft_spectrogram</a>()</code> <br><code><a href="layer_string_lookup.html">layer_string_lookup</a>()</code> <br><code><a href="layer_subtract.html">layer_subtract</a>()</code> <br><code><a href="layer_text_vectorization.html">layer_text_vectorization</a>()</code> <br><code><a href="layer_tfsm.html">layer_tfsm</a>()</code> <br><code><a href="layer_time_distributed.html">layer_time_distributed</a>()</code> <br><code><a href="layer_torch_module_wrapper.html">layer_torch_module_wrapper</a>()</code> <br><code><a href="layer_unit_normalization.html">layer_unit_normalization</a>()</code> <br><code><a href="layer_upsampling_1d.html">layer_upsampling_1d</a>()</code> <br><code><a href="layer_upsampling_2d.html">layer_upsampling_2d</a>()</code> <br><code><a href="layer_upsampling_3d.html">layer_upsampling_3d</a>()</code> <br><code><a href="layer_zero_padding_1d.html">layer_zero_padding_1d</a>()</code> <br><code><a href="layer_zero_padding_2d.html">layer_zero_padding_2d</a>()</code> <br><code><a href="layer_zero_padding_3d.html">layer_zero_padding_3d</a>()</code> <br><code><a href="rnn_cell_gru.html">rnn_cell_gru</a>()</code> <br><code><a href="rnn_cell_lstm.html">rnn_cell_lstm</a>()</code> <br><code><a href="rnn_cell_simple.html">rnn_cell_simple</a>()</code> <br><code><a href="rnn_cells_stack.html">rnn_cells_stack</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -157,7 +157,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/layer_random_flip.html b/docs/dev/reference/layer_random_flip.html
index 4583343c6..52554edc9 100644
--- a/docs/dev/reference/layer_random_flip.html
+++ b/docs/dev/reference/layer_random_flip.html
@@ -22,7 +22,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -148,8 +148,8 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
     <div class="dont-index">
 <ul><li><p><a href="https://keras.io/api/layers/preprocessing_layers/image_augmentation/random_flip#randomflip-class" class="external-link">https://keras.io/api/layers/preprocessing_layers/image_augmentation/random_flip#randomflip-class</a></p></li>
 </ul><p>Other image augmentation layers: <br><code><a href="layer_random_brightness.html">layer_random_brightness</a>()</code> <br><code><a href="layer_random_contrast.html">layer_random_contrast</a>()</code> <br><code><a href="layer_random_crop.html">layer_random_crop</a>()</code> <br><code><a href="layer_random_rotation.html">layer_random_rotation</a>()</code> <br><code><a href="layer_random_translation.html">layer_random_translation</a>()</code> <br><code><a href="layer_random_zoom.html">layer_random_zoom</a>()</code> <br></p>
-<p>Other preprocessing layers: <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_category_encoding.html">layer_category_encoding</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_discretization.html">layer_discretization</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_feature_space.html">layer_feature_space</a>()</code> <br><code><a href="layer_hashed_crossing.html">layer_hashed_crossing</a>()</code> <br><code><a href="layer_hashing.html">layer_hashing</a>()</code> <br><code><a href="layer_integer_lookup.html">layer_integer_lookup</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_mel_spectrogram.html">layer_mel_spectrogram</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_normalization.html">layer_normalization</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_brightness.html">layer_random_brightness</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_contrast.html">layer_random_contrast</a>()</code> <br><code><a href="layer_random_crop.html">layer_random_crop</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_rotation.html">layer_random_rotation</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_random_translation.html">layer_random_translation</a>()</code> <br><code><a href="layer_random_zoom.html">layer_random_zoom</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br><code><a href="layer_stft_spectrogram.html">layer_stft_spectrogram</a>()</code> <br><code><a href="layer_string_lookup.html">layer_string_lookup</a>()</code> <br><code><a href="layer_text_vectorization.html">layer_text_vectorization</a>()</code> <br></p>
-<p>Other layers: <br><code><a href="Layer.html">Layer</a>()</code> <br><code><a href="layer_activation.html">layer_activation</a>()</code> <br><code><a href="layer_activation_elu.html">layer_activation_elu</a>()</code> <br><code><a href="layer_activation_leaky_relu.html">layer_activation_leaky_relu</a>()</code> <br><code><a href="layer_activation_parametric_relu.html">layer_activation_parametric_relu</a>()</code> <br><code><a href="layer_activation_relu.html">layer_activation_relu</a>()</code> <br><code><a href="layer_activation_softmax.html">layer_activation_softmax</a>()</code> <br><code><a href="layer_activity_regularization.html">layer_activity_regularization</a>()</code> <br><code><a href="layer_add.html">layer_add</a>()</code> <br><code><a href="layer_additive_attention.html">layer_additive_attention</a>()</code> <br><code><a href="layer_alpha_dropout.html">layer_alpha_dropout</a>()</code> <br><code><a href="layer_attention.html">layer_attention</a>()</code> <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_average.html">layer_average</a>()</code> <br><code><a href="layer_average_pooling_1d.html">layer_average_pooling_1d</a>()</code> <br><code><a href="layer_average_pooling_2d.html">layer_average_pooling_2d</a>()</code> <br><code><a href="layer_average_pooling_3d.html">layer_average_pooling_3d</a>()</code> <br><code><a href="layer_batch_normalization.html">layer_batch_normalization</a>()</code> <br><code><a href="layer_bidirectional.html">layer_bidirectional</a>()</code> <br><code><a href="layer_category_encoding.html">layer_category_encoding</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_concatenate.html">layer_concatenate</a>()</code> <br><code><a href="layer_conv_1d.html">layer_conv_1d</a>()</code> <br><code><a href="layer_conv_1d_transpose.html">layer_conv_1d_transpose</a>()</code> <br><code><a href="layer_conv_2d.html">layer_conv_2d</a>()</code> <br><code><a href="layer_conv_2d_transpose.html">layer_conv_2d_transpose</a>()</code> <br><code><a href="layer_conv_3d.html">layer_conv_3d</a>()</code> <br><code><a href="layer_conv_3d_transpose.html">layer_conv_3d_transpose</a>()</code> <br><code><a href="layer_conv_lstm_1d.html">layer_conv_lstm_1d</a>()</code> <br><code><a href="layer_conv_lstm_2d.html">layer_conv_lstm_2d</a>()</code> <br><code><a href="layer_conv_lstm_3d.html">layer_conv_lstm_3d</a>()</code> <br><code><a href="layer_cropping_1d.html">layer_cropping_1d</a>()</code> <br><code><a href="layer_cropping_2d.html">layer_cropping_2d</a>()</code> <br><code><a href="layer_cropping_3d.html">layer_cropping_3d</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_dense.html">layer_dense</a>()</code> <br><code><a href="layer_depthwise_conv_1d.html">layer_depthwise_conv_1d</a>()</code> <br><code><a href="layer_depthwise_conv_2d.html">layer_depthwise_conv_2d</a>()</code> <br><code><a href="layer_discretization.html">layer_discretization</a>()</code> <br><code><a href="layer_dot.html">layer_dot</a>()</code> <br><code><a href="layer_dropout.html">layer_dropout</a>()</code> <br><code><a href="layer_einsum_dense.html">layer_einsum_dense</a>()</code> <br><code><a href="layer_embedding.html">layer_embedding</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_feature_space.html">layer_feature_space</a>()</code> <br><code><a href="layer_flatten.html">layer_flatten</a>()</code> <br><code><a href="layer_flax_module_wrapper.html">layer_flax_module_wrapper</a>()</code> <br><code><a href="layer_gaussian_dropout.html">layer_gaussian_dropout</a>()</code> <br><code><a href="layer_gaussian_noise.html">layer_gaussian_noise</a>()</code> <br><code><a href="layer_global_average_pooling_1d.html">layer_global_average_pooling_1d</a>()</code> <br><code><a href="layer_global_average_pooling_2d.html">layer_global_average_pooling_2d</a>()</code> <br><code><a href="layer_global_average_pooling_3d.html">layer_global_average_pooling_3d</a>()</code> <br><code><a href="layer_global_max_pooling_1d.html">layer_global_max_pooling_1d</a>()</code> <br><code><a href="layer_global_max_pooling_2d.html">layer_global_max_pooling_2d</a>()</code> <br><code><a href="layer_global_max_pooling_3d.html">layer_global_max_pooling_3d</a>()</code> <br><code><a href="layer_group_normalization.html">layer_group_normalization</a>()</code> <br><code><a href="layer_group_query_attention.html">layer_group_query_attention</a>()</code> <br><code><a href="layer_gru.html">layer_gru</a>()</code> <br><code><a href="layer_hashed_crossing.html">layer_hashed_crossing</a>()</code> <br><code><a href="layer_hashing.html">layer_hashing</a>()</code> <br><code><a href="layer_identity.html">layer_identity</a>()</code> <br><code><a href="layer_integer_lookup.html">layer_integer_lookup</a>()</code> <br><code><a href="layer_jax_model_wrapper.html">layer_jax_model_wrapper</a>()</code> <br><code><a href="layer_lambda.html">layer_lambda</a>()</code> <br><code><a href="layer_layer_normalization.html">layer_layer_normalization</a>()</code> <br><code><a href="layer_lstm.html">layer_lstm</a>()</code> <br><code><a href="layer_masking.html">layer_masking</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_max_pooling_1d.html">layer_max_pooling_1d</a>()</code> <br><code><a href="layer_max_pooling_2d.html">layer_max_pooling_2d</a>()</code> <br><code><a href="layer_max_pooling_3d.html">layer_max_pooling_3d</a>()</code> <br><code><a href="layer_maximum.html">layer_maximum</a>()</code> <br><code><a href="layer_mel_spectrogram.html">layer_mel_spectrogram</a>()</code> <br><code><a href="layer_minimum.html">layer_minimum</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_multi_head_attention.html">layer_multi_head_attention</a>()</code> <br><code><a href="layer_multiply.html">layer_multiply</a>()</code> <br><code><a href="layer_normalization.html">layer_normalization</a>()</code> <br><code><a href="layer_permute.html">layer_permute</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_brightness.html">layer_random_brightness</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_contrast.html">layer_random_contrast</a>()</code> <br><code><a href="layer_random_crop.html">layer_random_crop</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_rotation.html">layer_random_rotation</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_random_translation.html">layer_random_translation</a>()</code> <br><code><a href="layer_random_zoom.html">layer_random_zoom</a>()</code> <br><code><a href="layer_repeat_vector.html">layer_repeat_vector</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_reshape.html">layer_reshape</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_rms_normalization.html">layer_rms_normalization</a>()</code> <br><code><a href="layer_rnn.html">layer_rnn</a>()</code> <br><code><a href="layer_separable_conv_1d.html">layer_separable_conv_1d</a>()</code> <br><code><a href="layer_separable_conv_2d.html">layer_separable_conv_2d</a>()</code> <br><code><a href="layer_simple_rnn.html">layer_simple_rnn</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br><code><a href="layer_spatial_dropout_1d.html">layer_spatial_dropout_1d</a>()</code> <br><code><a href="layer_spatial_dropout_2d.html">layer_spatial_dropout_2d</a>()</code> <br><code><a href="layer_spatial_dropout_3d.html">layer_spatial_dropout_3d</a>()</code> <br><code><a href="layer_spectral_normalization.html">layer_spectral_normalization</a>()</code> <br><code><a href="layer_stft_spectrogram.html">layer_stft_spectrogram</a>()</code> <br><code><a href="layer_string_lookup.html">layer_string_lookup</a>()</code> <br><code><a href="layer_subtract.html">layer_subtract</a>()</code> <br><code><a href="layer_text_vectorization.html">layer_text_vectorization</a>()</code> <br><code><a href="layer_tfsm.html">layer_tfsm</a>()</code> <br><code><a href="layer_time_distributed.html">layer_time_distributed</a>()</code> <br><code><a href="layer_torch_module_wrapper.html">layer_torch_module_wrapper</a>()</code> <br><code><a href="layer_unit_normalization.html">layer_unit_normalization</a>()</code> <br><code><a href="layer_upsampling_1d.html">layer_upsampling_1d</a>()</code> <br><code><a href="layer_upsampling_2d.html">layer_upsampling_2d</a>()</code> <br><code><a href="layer_upsampling_3d.html">layer_upsampling_3d</a>()</code> <br><code><a href="layer_zero_padding_1d.html">layer_zero_padding_1d</a>()</code> <br><code><a href="layer_zero_padding_2d.html">layer_zero_padding_2d</a>()</code> <br><code><a href="layer_zero_padding_3d.html">layer_zero_padding_3d</a>()</code> <br><code><a href="rnn_cell_gru.html">rnn_cell_gru</a>()</code> <br><code><a href="rnn_cell_lstm.html">rnn_cell_lstm</a>()</code> <br><code><a href="rnn_cell_simple.html">rnn_cell_simple</a>()</code> <br><code><a href="rnn_cells_stack.html">rnn_cells_stack</a>()</code> <br></p></div>
+<p>Other preprocessing layers: <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_category_encoding.html">layer_category_encoding</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_discretization.html">layer_discretization</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_feature_space.html">layer_feature_space</a>()</code> <br><code><a href="layer_hashed_crossing.html">layer_hashed_crossing</a>()</code> <br><code><a href="layer_hashing.html">layer_hashing</a>()</code> <br><code><a href="layer_integer_lookup.html">layer_integer_lookup</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_mel_spectrogram.html">layer_mel_spectrogram</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_normalization.html">layer_normalization</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_brightness.html">layer_random_brightness</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_contrast.html">layer_random_contrast</a>()</code> <br><code><a href="layer_random_crop.html">layer_random_crop</a>()</code> <br><code><a href="layer_random_elastic_transform.html">layer_random_elastic_transform</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_rotation.html">layer_random_rotation</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_random_translation.html">layer_random_translation</a>()</code> <br><code><a href="layer_random_zoom.html">layer_random_zoom</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br><code><a href="layer_stft_spectrogram.html">layer_stft_spectrogram</a>()</code> <br><code><a href="layer_string_lookup.html">layer_string_lookup</a>()</code> <br><code><a href="layer_text_vectorization.html">layer_text_vectorization</a>()</code> <br></p>
+<p>Other layers: <br><code><a href="Layer.html">Layer</a>()</code> <br><code><a href="layer_activation.html">layer_activation</a>()</code> <br><code><a href="layer_activation_elu.html">layer_activation_elu</a>()</code> <br><code><a href="layer_activation_leaky_relu.html">layer_activation_leaky_relu</a>()</code> <br><code><a href="layer_activation_parametric_relu.html">layer_activation_parametric_relu</a>()</code> <br><code><a href="layer_activation_relu.html">layer_activation_relu</a>()</code> <br><code><a href="layer_activation_softmax.html">layer_activation_softmax</a>()</code> <br><code><a href="layer_activity_regularization.html">layer_activity_regularization</a>()</code> <br><code><a href="layer_add.html">layer_add</a>()</code> <br><code><a href="layer_additive_attention.html">layer_additive_attention</a>()</code> <br><code><a href="layer_alpha_dropout.html">layer_alpha_dropout</a>()</code> <br><code><a href="layer_attention.html">layer_attention</a>()</code> <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_average.html">layer_average</a>()</code> <br><code><a href="layer_average_pooling_1d.html">layer_average_pooling_1d</a>()</code> <br><code><a href="layer_average_pooling_2d.html">layer_average_pooling_2d</a>()</code> <br><code><a href="layer_average_pooling_3d.html">layer_average_pooling_3d</a>()</code> <br><code><a href="layer_batch_normalization.html">layer_batch_normalization</a>()</code> <br><code><a href="layer_bidirectional.html">layer_bidirectional</a>()</code> <br><code><a href="layer_category_encoding.html">layer_category_encoding</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_concatenate.html">layer_concatenate</a>()</code> <br><code><a href="layer_conv_1d.html">layer_conv_1d</a>()</code> <br><code><a href="layer_conv_1d_transpose.html">layer_conv_1d_transpose</a>()</code> <br><code><a href="layer_conv_2d.html">layer_conv_2d</a>()</code> <br><code><a href="layer_conv_2d_transpose.html">layer_conv_2d_transpose</a>()</code> <br><code><a href="layer_conv_3d.html">layer_conv_3d</a>()</code> <br><code><a href="layer_conv_3d_transpose.html">layer_conv_3d_transpose</a>()</code> <br><code><a href="layer_conv_lstm_1d.html">layer_conv_lstm_1d</a>()</code> <br><code><a href="layer_conv_lstm_2d.html">layer_conv_lstm_2d</a>()</code> <br><code><a href="layer_conv_lstm_3d.html">layer_conv_lstm_3d</a>()</code> <br><code><a href="layer_cropping_1d.html">layer_cropping_1d</a>()</code> <br><code><a href="layer_cropping_2d.html">layer_cropping_2d</a>()</code> <br><code><a href="layer_cropping_3d.html">layer_cropping_3d</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_dense.html">layer_dense</a>()</code> <br><code><a href="layer_depthwise_conv_1d.html">layer_depthwise_conv_1d</a>()</code> <br><code><a href="layer_depthwise_conv_2d.html">layer_depthwise_conv_2d</a>()</code> <br><code><a href="layer_discretization.html">layer_discretization</a>()</code> <br><code><a href="layer_dot.html">layer_dot</a>()</code> <br><code><a href="layer_dropout.html">layer_dropout</a>()</code> <br><code><a href="layer_einsum_dense.html">layer_einsum_dense</a>()</code> <br><code><a href="layer_embedding.html">layer_embedding</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_feature_space.html">layer_feature_space</a>()</code> <br><code><a href="layer_flatten.html">layer_flatten</a>()</code> <br><code><a href="layer_flax_module_wrapper.html">layer_flax_module_wrapper</a>()</code> <br><code><a href="layer_gaussian_dropout.html">layer_gaussian_dropout</a>()</code> <br><code><a href="layer_gaussian_noise.html">layer_gaussian_noise</a>()</code> <br><code><a href="layer_global_average_pooling_1d.html">layer_global_average_pooling_1d</a>()</code> <br><code><a href="layer_global_average_pooling_2d.html">layer_global_average_pooling_2d</a>()</code> <br><code><a href="layer_global_average_pooling_3d.html">layer_global_average_pooling_3d</a>()</code> <br><code><a href="layer_global_max_pooling_1d.html">layer_global_max_pooling_1d</a>()</code> <br><code><a href="layer_global_max_pooling_2d.html">layer_global_max_pooling_2d</a>()</code> <br><code><a href="layer_global_max_pooling_3d.html">layer_global_max_pooling_3d</a>()</code> <br><code><a href="layer_group_normalization.html">layer_group_normalization</a>()</code> <br><code><a href="layer_group_query_attention.html">layer_group_query_attention</a>()</code> <br><code><a href="layer_gru.html">layer_gru</a>()</code> <br><code><a href="layer_hashed_crossing.html">layer_hashed_crossing</a>()</code> <br><code><a href="layer_hashing.html">layer_hashing</a>()</code> <br><code><a href="layer_identity.html">layer_identity</a>()</code> <br><code><a href="layer_integer_lookup.html">layer_integer_lookup</a>()</code> <br><code><a href="layer_jax_model_wrapper.html">layer_jax_model_wrapper</a>()</code> <br><code><a href="layer_lambda.html">layer_lambda</a>()</code> <br><code><a href="layer_layer_normalization.html">layer_layer_normalization</a>()</code> <br><code><a href="layer_lstm.html">layer_lstm</a>()</code> <br><code><a href="layer_masking.html">layer_masking</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_max_pooling_1d.html">layer_max_pooling_1d</a>()</code> <br><code><a href="layer_max_pooling_2d.html">layer_max_pooling_2d</a>()</code> <br><code><a href="layer_max_pooling_3d.html">layer_max_pooling_3d</a>()</code> <br><code><a href="layer_maximum.html">layer_maximum</a>()</code> <br><code><a href="layer_mel_spectrogram.html">layer_mel_spectrogram</a>()</code> <br><code><a href="layer_minimum.html">layer_minimum</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_multi_head_attention.html">layer_multi_head_attention</a>()</code> <br><code><a href="layer_multiply.html">layer_multiply</a>()</code> <br><code><a href="layer_normalization.html">layer_normalization</a>()</code> <br><code><a href="layer_permute.html">layer_permute</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_brightness.html">layer_random_brightness</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_contrast.html">layer_random_contrast</a>()</code> <br><code><a href="layer_random_crop.html">layer_random_crop</a>()</code> <br><code><a href="layer_random_elastic_transform.html">layer_random_elastic_transform</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_rotation.html">layer_random_rotation</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_random_translation.html">layer_random_translation</a>()</code> <br><code><a href="layer_random_zoom.html">layer_random_zoom</a>()</code> <br><code><a href="layer_repeat_vector.html">layer_repeat_vector</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_reshape.html">layer_reshape</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_rms_normalization.html">layer_rms_normalization</a>()</code> <br><code><a href="layer_rnn.html">layer_rnn</a>()</code> <br><code><a href="layer_separable_conv_1d.html">layer_separable_conv_1d</a>()</code> <br><code><a href="layer_separable_conv_2d.html">layer_separable_conv_2d</a>()</code> <br><code><a href="layer_simple_rnn.html">layer_simple_rnn</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br><code><a href="layer_spatial_dropout_1d.html">layer_spatial_dropout_1d</a>()</code> <br><code><a href="layer_spatial_dropout_2d.html">layer_spatial_dropout_2d</a>()</code> <br><code><a href="layer_spatial_dropout_3d.html">layer_spatial_dropout_3d</a>()</code> <br><code><a href="layer_spectral_normalization.html">layer_spectral_normalization</a>()</code> <br><code><a href="layer_stft_spectrogram.html">layer_stft_spectrogram</a>()</code> <br><code><a href="layer_string_lookup.html">layer_string_lookup</a>()</code> <br><code><a href="layer_subtract.html">layer_subtract</a>()</code> <br><code><a href="layer_text_vectorization.html">layer_text_vectorization</a>()</code> <br><code><a href="layer_tfsm.html">layer_tfsm</a>()</code> <br><code><a href="layer_time_distributed.html">layer_time_distributed</a>()</code> <br><code><a href="layer_torch_module_wrapper.html">layer_torch_module_wrapper</a>()</code> <br><code><a href="layer_unit_normalization.html">layer_unit_normalization</a>()</code> <br><code><a href="layer_upsampling_1d.html">layer_upsampling_1d</a>()</code> <br><code><a href="layer_upsampling_2d.html">layer_upsampling_2d</a>()</code> <br><code><a href="layer_upsampling_3d.html">layer_upsampling_3d</a>()</code> <br><code><a href="layer_zero_padding_1d.html">layer_zero_padding_1d</a>()</code> <br><code><a href="layer_zero_padding_2d.html">layer_zero_padding_2d</a>()</code> <br><code><a href="layer_zero_padding_3d.html">layer_zero_padding_3d</a>()</code> <br><code><a href="rnn_cell_gru.html">rnn_cell_gru</a>()</code> <br><code><a href="rnn_cell_lstm.html">rnn_cell_lstm</a>()</code> <br><code><a href="rnn_cell_simple.html">rnn_cell_simple</a>()</code> <br><code><a href="rnn_cells_stack.html">rnn_cells_stack</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -161,7 +161,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/layer_random_gaussian_blur.html b/docs/dev/reference/layer_random_gaussian_blur.html
index def45c756..1ec9de7ca 100644
--- a/docs/dev/reference/layer_random_gaussian_blur.html
+++ b/docs/dev/reference/layer_random_gaussian_blur.html
@@ -12,7 +12,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -135,9 +135,9 @@ <h2 id="arguments">Arguments<a class="anchor" aria-label="anchor" href="#argumen
 </dl></div>
     <div class="section level2">
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
-    <div class="dont-index"><p>Other image preprocessing layers: <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br></p>
-<p>Other preprocessing layers: <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_category_encoding.html">layer_category_encoding</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_discretization.html">layer_discretization</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_feature_space.html">layer_feature_space</a>()</code> <br><code><a href="layer_hashed_crossing.html">layer_hashed_crossing</a>()</code> <br><code><a href="layer_hashing.html">layer_hashing</a>()</code> <br><code><a href="layer_integer_lookup.html">layer_integer_lookup</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_mel_spectrogram.html">layer_mel_spectrogram</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_normalization.html">layer_normalization</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_brightness.html">layer_random_brightness</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_contrast.html">layer_random_contrast</a>()</code> <br><code><a href="layer_random_crop.html">layer_random_crop</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_flip.html">layer_random_flip</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_rotation.html">layer_random_rotation</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_random_translation.html">layer_random_translation</a>()</code> <br><code><a href="layer_random_zoom.html">layer_random_zoom</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br><code><a href="layer_stft_spectrogram.html">layer_stft_spectrogram</a>()</code> <br><code><a href="layer_string_lookup.html">layer_string_lookup</a>()</code> <br><code><a href="layer_text_vectorization.html">layer_text_vectorization</a>()</code> <br></p>
-<p>Other layers: <br><code><a href="Layer.html">Layer</a>()</code> <br><code><a href="layer_activation.html">layer_activation</a>()</code> <br><code><a href="layer_activation_elu.html">layer_activation_elu</a>()</code> <br><code><a href="layer_activation_leaky_relu.html">layer_activation_leaky_relu</a>()</code> <br><code><a href="layer_activation_parametric_relu.html">layer_activation_parametric_relu</a>()</code> <br><code><a href="layer_activation_relu.html">layer_activation_relu</a>()</code> <br><code><a href="layer_activation_softmax.html">layer_activation_softmax</a>()</code> <br><code><a href="layer_activity_regularization.html">layer_activity_regularization</a>()</code> <br><code><a href="layer_add.html">layer_add</a>()</code> <br><code><a href="layer_additive_attention.html">layer_additive_attention</a>()</code> <br><code><a href="layer_alpha_dropout.html">layer_alpha_dropout</a>()</code> <br><code><a href="layer_attention.html">layer_attention</a>()</code> <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_average.html">layer_average</a>()</code> <br><code><a href="layer_average_pooling_1d.html">layer_average_pooling_1d</a>()</code> <br><code><a href="layer_average_pooling_2d.html">layer_average_pooling_2d</a>()</code> <br><code><a href="layer_average_pooling_3d.html">layer_average_pooling_3d</a>()</code> <br><code><a href="layer_batch_normalization.html">layer_batch_normalization</a>()</code> <br><code><a href="layer_bidirectional.html">layer_bidirectional</a>()</code> <br><code><a href="layer_category_encoding.html">layer_category_encoding</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_concatenate.html">layer_concatenate</a>()</code> <br><code><a href="layer_conv_1d.html">layer_conv_1d</a>()</code> <br><code><a href="layer_conv_1d_transpose.html">layer_conv_1d_transpose</a>()</code> <br><code><a href="layer_conv_2d.html">layer_conv_2d</a>()</code> <br><code><a href="layer_conv_2d_transpose.html">layer_conv_2d_transpose</a>()</code> <br><code><a href="layer_conv_3d.html">layer_conv_3d</a>()</code> <br><code><a href="layer_conv_3d_transpose.html">layer_conv_3d_transpose</a>()</code> <br><code><a href="layer_conv_lstm_1d.html">layer_conv_lstm_1d</a>()</code> <br><code><a href="layer_conv_lstm_2d.html">layer_conv_lstm_2d</a>()</code> <br><code><a href="layer_conv_lstm_3d.html">layer_conv_lstm_3d</a>()</code> <br><code><a href="layer_cropping_1d.html">layer_cropping_1d</a>()</code> <br><code><a href="layer_cropping_2d.html">layer_cropping_2d</a>()</code> <br><code><a href="layer_cropping_3d.html">layer_cropping_3d</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_dense.html">layer_dense</a>()</code> <br><code><a href="layer_depthwise_conv_1d.html">layer_depthwise_conv_1d</a>()</code> <br><code><a href="layer_depthwise_conv_2d.html">layer_depthwise_conv_2d</a>()</code> <br><code><a href="layer_discretization.html">layer_discretization</a>()</code> <br><code><a href="layer_dot.html">layer_dot</a>()</code> <br><code><a href="layer_dropout.html">layer_dropout</a>()</code> <br><code><a href="layer_einsum_dense.html">layer_einsum_dense</a>()</code> <br><code><a href="layer_embedding.html">layer_embedding</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_feature_space.html">layer_feature_space</a>()</code> <br><code><a href="layer_flatten.html">layer_flatten</a>()</code> <br><code><a href="layer_flax_module_wrapper.html">layer_flax_module_wrapper</a>()</code> <br><code><a href="layer_gaussian_dropout.html">layer_gaussian_dropout</a>()</code> <br><code><a href="layer_gaussian_noise.html">layer_gaussian_noise</a>()</code> <br><code><a href="layer_global_average_pooling_1d.html">layer_global_average_pooling_1d</a>()</code> <br><code><a href="layer_global_average_pooling_2d.html">layer_global_average_pooling_2d</a>()</code> <br><code><a href="layer_global_average_pooling_3d.html">layer_global_average_pooling_3d</a>()</code> <br><code><a href="layer_global_max_pooling_1d.html">layer_global_max_pooling_1d</a>()</code> <br><code><a href="layer_global_max_pooling_2d.html">layer_global_max_pooling_2d</a>()</code> <br><code><a href="layer_global_max_pooling_3d.html">layer_global_max_pooling_3d</a>()</code> <br><code><a href="layer_group_normalization.html">layer_group_normalization</a>()</code> <br><code><a href="layer_group_query_attention.html">layer_group_query_attention</a>()</code> <br><code><a href="layer_gru.html">layer_gru</a>()</code> <br><code><a href="layer_hashed_crossing.html">layer_hashed_crossing</a>()</code> <br><code><a href="layer_hashing.html">layer_hashing</a>()</code> <br><code><a href="layer_identity.html">layer_identity</a>()</code> <br><code><a href="layer_integer_lookup.html">layer_integer_lookup</a>()</code> <br><code><a href="layer_jax_model_wrapper.html">layer_jax_model_wrapper</a>()</code> <br><code><a href="layer_lambda.html">layer_lambda</a>()</code> <br><code><a href="layer_layer_normalization.html">layer_layer_normalization</a>()</code> <br><code><a href="layer_lstm.html">layer_lstm</a>()</code> <br><code><a href="layer_masking.html">layer_masking</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_max_pooling_1d.html">layer_max_pooling_1d</a>()</code> <br><code><a href="layer_max_pooling_2d.html">layer_max_pooling_2d</a>()</code> <br><code><a href="layer_max_pooling_3d.html">layer_max_pooling_3d</a>()</code> <br><code><a href="layer_maximum.html">layer_maximum</a>()</code> <br><code><a href="layer_mel_spectrogram.html">layer_mel_spectrogram</a>()</code> <br><code><a href="layer_minimum.html">layer_minimum</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_multi_head_attention.html">layer_multi_head_attention</a>()</code> <br><code><a href="layer_multiply.html">layer_multiply</a>()</code> <br><code><a href="layer_normalization.html">layer_normalization</a>()</code> <br><code><a href="layer_permute.html">layer_permute</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_brightness.html">layer_random_brightness</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_contrast.html">layer_random_contrast</a>()</code> <br><code><a href="layer_random_crop.html">layer_random_crop</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_flip.html">layer_random_flip</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_rotation.html">layer_random_rotation</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_random_translation.html">layer_random_translation</a>()</code> <br><code><a href="layer_random_zoom.html">layer_random_zoom</a>()</code> <br><code><a href="layer_repeat_vector.html">layer_repeat_vector</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_reshape.html">layer_reshape</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_rms_normalization.html">layer_rms_normalization</a>()</code> <br><code><a href="layer_rnn.html">layer_rnn</a>()</code> <br><code><a href="layer_separable_conv_1d.html">layer_separable_conv_1d</a>()</code> <br><code><a href="layer_separable_conv_2d.html">layer_separable_conv_2d</a>()</code> <br><code><a href="layer_simple_rnn.html">layer_simple_rnn</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br><code><a href="layer_spatial_dropout_1d.html">layer_spatial_dropout_1d</a>()</code> <br><code><a href="layer_spatial_dropout_2d.html">layer_spatial_dropout_2d</a>()</code> <br><code><a href="layer_spatial_dropout_3d.html">layer_spatial_dropout_3d</a>()</code> <br><code><a href="layer_spectral_normalization.html">layer_spectral_normalization</a>()</code> <br><code><a href="layer_stft_spectrogram.html">layer_stft_spectrogram</a>()</code> <br><code><a href="layer_string_lookup.html">layer_string_lookup</a>()</code> <br><code><a href="layer_subtract.html">layer_subtract</a>()</code> <br><code><a href="layer_text_vectorization.html">layer_text_vectorization</a>()</code> <br><code><a href="layer_tfsm.html">layer_tfsm</a>()</code> <br><code><a href="layer_time_distributed.html">layer_time_distributed</a>()</code> <br><code><a href="layer_torch_module_wrapper.html">layer_torch_module_wrapper</a>()</code> <br><code><a href="layer_unit_normalization.html">layer_unit_normalization</a>()</code> <br><code><a href="layer_upsampling_1d.html">layer_upsampling_1d</a>()</code> <br><code><a href="layer_upsampling_2d.html">layer_upsampling_2d</a>()</code> <br><code><a href="layer_upsampling_3d.html">layer_upsampling_3d</a>()</code> <br><code><a href="layer_zero_padding_1d.html">layer_zero_padding_1d</a>()</code> <br><code><a href="layer_zero_padding_2d.html">layer_zero_padding_2d</a>()</code> <br><code><a href="layer_zero_padding_3d.html">layer_zero_padding_3d</a>()</code> <br><code><a href="rnn_cell_gru.html">rnn_cell_gru</a>()</code> <br><code><a href="rnn_cell_lstm.html">rnn_cell_lstm</a>()</code> <br><code><a href="rnn_cell_simple.html">rnn_cell_simple</a>()</code> <br><code><a href="rnn_cells_stack.html">rnn_cells_stack</a>()</code> <br></p></div>
+    <div class="dont-index"><p>Other image preprocessing layers: <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_elastic_transform.html">layer_random_elastic_transform</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br></p>
+<p>Other preprocessing layers: <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_category_encoding.html">layer_category_encoding</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_discretization.html">layer_discretization</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_feature_space.html">layer_feature_space</a>()</code> <br><code><a href="layer_hashed_crossing.html">layer_hashed_crossing</a>()</code> <br><code><a href="layer_hashing.html">layer_hashing</a>()</code> <br><code><a href="layer_integer_lookup.html">layer_integer_lookup</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_mel_spectrogram.html">layer_mel_spectrogram</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_normalization.html">layer_normalization</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_brightness.html">layer_random_brightness</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_contrast.html">layer_random_contrast</a>()</code> <br><code><a href="layer_random_crop.html">layer_random_crop</a>()</code> <br><code><a href="layer_random_elastic_transform.html">layer_random_elastic_transform</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_flip.html">layer_random_flip</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_rotation.html">layer_random_rotation</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_random_translation.html">layer_random_translation</a>()</code> <br><code><a href="layer_random_zoom.html">layer_random_zoom</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br><code><a href="layer_stft_spectrogram.html">layer_stft_spectrogram</a>()</code> <br><code><a href="layer_string_lookup.html">layer_string_lookup</a>()</code> <br><code><a href="layer_text_vectorization.html">layer_text_vectorization</a>()</code> <br></p>
+<p>Other layers: <br><code><a href="Layer.html">Layer</a>()</code> <br><code><a href="layer_activation.html">layer_activation</a>()</code> <br><code><a href="layer_activation_elu.html">layer_activation_elu</a>()</code> <br><code><a href="layer_activation_leaky_relu.html">layer_activation_leaky_relu</a>()</code> <br><code><a href="layer_activation_parametric_relu.html">layer_activation_parametric_relu</a>()</code> <br><code><a href="layer_activation_relu.html">layer_activation_relu</a>()</code> <br><code><a href="layer_activation_softmax.html">layer_activation_softmax</a>()</code> <br><code><a href="layer_activity_regularization.html">layer_activity_regularization</a>()</code> <br><code><a href="layer_add.html">layer_add</a>()</code> <br><code><a href="layer_additive_attention.html">layer_additive_attention</a>()</code> <br><code><a href="layer_alpha_dropout.html">layer_alpha_dropout</a>()</code> <br><code><a href="layer_attention.html">layer_attention</a>()</code> <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_average.html">layer_average</a>()</code> <br><code><a href="layer_average_pooling_1d.html">layer_average_pooling_1d</a>()</code> <br><code><a href="layer_average_pooling_2d.html">layer_average_pooling_2d</a>()</code> <br><code><a href="layer_average_pooling_3d.html">layer_average_pooling_3d</a>()</code> <br><code><a href="layer_batch_normalization.html">layer_batch_normalization</a>()</code> <br><code><a href="layer_bidirectional.html">layer_bidirectional</a>()</code> <br><code><a href="layer_category_encoding.html">layer_category_encoding</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_concatenate.html">layer_concatenate</a>()</code> <br><code><a href="layer_conv_1d.html">layer_conv_1d</a>()</code> <br><code><a href="layer_conv_1d_transpose.html">layer_conv_1d_transpose</a>()</code> <br><code><a href="layer_conv_2d.html">layer_conv_2d</a>()</code> <br><code><a href="layer_conv_2d_transpose.html">layer_conv_2d_transpose</a>()</code> <br><code><a href="layer_conv_3d.html">layer_conv_3d</a>()</code> <br><code><a href="layer_conv_3d_transpose.html">layer_conv_3d_transpose</a>()</code> <br><code><a href="layer_conv_lstm_1d.html">layer_conv_lstm_1d</a>()</code> <br><code><a href="layer_conv_lstm_2d.html">layer_conv_lstm_2d</a>()</code> <br><code><a href="layer_conv_lstm_3d.html">layer_conv_lstm_3d</a>()</code> <br><code><a href="layer_cropping_1d.html">layer_cropping_1d</a>()</code> <br><code><a href="layer_cropping_2d.html">layer_cropping_2d</a>()</code> <br><code><a href="layer_cropping_3d.html">layer_cropping_3d</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_dense.html">layer_dense</a>()</code> <br><code><a href="layer_depthwise_conv_1d.html">layer_depthwise_conv_1d</a>()</code> <br><code><a href="layer_depthwise_conv_2d.html">layer_depthwise_conv_2d</a>()</code> <br><code><a href="layer_discretization.html">layer_discretization</a>()</code> <br><code><a href="layer_dot.html">layer_dot</a>()</code> <br><code><a href="layer_dropout.html">layer_dropout</a>()</code> <br><code><a href="layer_einsum_dense.html">layer_einsum_dense</a>()</code> <br><code><a href="layer_embedding.html">layer_embedding</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_feature_space.html">layer_feature_space</a>()</code> <br><code><a href="layer_flatten.html">layer_flatten</a>()</code> <br><code><a href="layer_flax_module_wrapper.html">layer_flax_module_wrapper</a>()</code> <br><code><a href="layer_gaussian_dropout.html">layer_gaussian_dropout</a>()</code> <br><code><a href="layer_gaussian_noise.html">layer_gaussian_noise</a>()</code> <br><code><a href="layer_global_average_pooling_1d.html">layer_global_average_pooling_1d</a>()</code> <br><code><a href="layer_global_average_pooling_2d.html">layer_global_average_pooling_2d</a>()</code> <br><code><a href="layer_global_average_pooling_3d.html">layer_global_average_pooling_3d</a>()</code> <br><code><a href="layer_global_max_pooling_1d.html">layer_global_max_pooling_1d</a>()</code> <br><code><a href="layer_global_max_pooling_2d.html">layer_global_max_pooling_2d</a>()</code> <br><code><a href="layer_global_max_pooling_3d.html">layer_global_max_pooling_3d</a>()</code> <br><code><a href="layer_group_normalization.html">layer_group_normalization</a>()</code> <br><code><a href="layer_group_query_attention.html">layer_group_query_attention</a>()</code> <br><code><a href="layer_gru.html">layer_gru</a>()</code> <br><code><a href="layer_hashed_crossing.html">layer_hashed_crossing</a>()</code> <br><code><a href="layer_hashing.html">layer_hashing</a>()</code> <br><code><a href="layer_identity.html">layer_identity</a>()</code> <br><code><a href="layer_integer_lookup.html">layer_integer_lookup</a>()</code> <br><code><a href="layer_jax_model_wrapper.html">layer_jax_model_wrapper</a>()</code> <br><code><a href="layer_lambda.html">layer_lambda</a>()</code> <br><code><a href="layer_layer_normalization.html">layer_layer_normalization</a>()</code> <br><code><a href="layer_lstm.html">layer_lstm</a>()</code> <br><code><a href="layer_masking.html">layer_masking</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_max_pooling_1d.html">layer_max_pooling_1d</a>()</code> <br><code><a href="layer_max_pooling_2d.html">layer_max_pooling_2d</a>()</code> <br><code><a href="layer_max_pooling_3d.html">layer_max_pooling_3d</a>()</code> <br><code><a href="layer_maximum.html">layer_maximum</a>()</code> <br><code><a href="layer_mel_spectrogram.html">layer_mel_spectrogram</a>()</code> <br><code><a href="layer_minimum.html">layer_minimum</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_multi_head_attention.html">layer_multi_head_attention</a>()</code> <br><code><a href="layer_multiply.html">layer_multiply</a>()</code> <br><code><a href="layer_normalization.html">layer_normalization</a>()</code> <br><code><a href="layer_permute.html">layer_permute</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_brightness.html">layer_random_brightness</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_contrast.html">layer_random_contrast</a>()</code> <br><code><a href="layer_random_crop.html">layer_random_crop</a>()</code> <br><code><a href="layer_random_elastic_transform.html">layer_random_elastic_transform</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_flip.html">layer_random_flip</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_rotation.html">layer_random_rotation</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_random_translation.html">layer_random_translation</a>()</code> <br><code><a href="layer_random_zoom.html">layer_random_zoom</a>()</code> <br><code><a href="layer_repeat_vector.html">layer_repeat_vector</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_reshape.html">layer_reshape</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_rms_normalization.html">layer_rms_normalization</a>()</code> <br><code><a href="layer_rnn.html">layer_rnn</a>()</code> <br><code><a href="layer_separable_conv_1d.html">layer_separable_conv_1d</a>()</code> <br><code><a href="layer_separable_conv_2d.html">layer_separable_conv_2d</a>()</code> <br><code><a href="layer_simple_rnn.html">layer_simple_rnn</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br><code><a href="layer_spatial_dropout_1d.html">layer_spatial_dropout_1d</a>()</code> <br><code><a href="layer_spatial_dropout_2d.html">layer_spatial_dropout_2d</a>()</code> <br><code><a href="layer_spatial_dropout_3d.html">layer_spatial_dropout_3d</a>()</code> <br><code><a href="layer_spectral_normalization.html">layer_spectral_normalization</a>()</code> <br><code><a href="layer_stft_spectrogram.html">layer_stft_spectrogram</a>()</code> <br><code><a href="layer_string_lookup.html">layer_string_lookup</a>()</code> <br><code><a href="layer_subtract.html">layer_subtract</a>()</code> <br><code><a href="layer_text_vectorization.html">layer_text_vectorization</a>()</code> <br><code><a href="layer_tfsm.html">layer_tfsm</a>()</code> <br><code><a href="layer_time_distributed.html">layer_time_distributed</a>()</code> <br><code><a href="layer_torch_module_wrapper.html">layer_torch_module_wrapper</a>()</code> <br><code><a href="layer_unit_normalization.html">layer_unit_normalization</a>()</code> <br><code><a href="layer_upsampling_1d.html">layer_upsampling_1d</a>()</code> <br><code><a href="layer_upsampling_2d.html">layer_upsampling_2d</a>()</code> <br><code><a href="layer_upsampling_3d.html">layer_upsampling_3d</a>()</code> <br><code><a href="layer_zero_padding_1d.html">layer_zero_padding_1d</a>()</code> <br><code><a href="layer_zero_padding_2d.html">layer_zero_padding_2d</a>()</code> <br><code><a href="layer_zero_padding_3d.html">layer_zero_padding_3d</a>()</code> <br><code><a href="rnn_cell_gru.html">rnn_cell_gru</a>()</code> <br><code><a href="rnn_cell_lstm.html">rnn_cell_lstm</a>()</code> <br><code><a href="rnn_cell_simple.html">rnn_cell_simple</a>()</code> <br><code><a href="rnn_cells_stack.html">rnn_cells_stack</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -149,7 +149,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/layer_random_grayscale.html b/docs/dev/reference/layer_random_grayscale.html
index b9a8f2cf4..b3aae6aa3 100644
--- a/docs/dev/reference/layer_random_grayscale.html
+++ b/docs/dev/reference/layer_random_grayscale.html
@@ -26,7 +26,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -141,9 +141,9 @@ <h2 id="output-shape">Output Shape<a class="anchor" aria-label="anchor" href="#o
     </div>
     <div class="section level2">
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
-    <div class="dont-index"><p>Other image preprocessing layers: <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br></p>
-<p>Other preprocessing layers: <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_category_encoding.html">layer_category_encoding</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_discretization.html">layer_discretization</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_feature_space.html">layer_feature_space</a>()</code> <br><code><a href="layer_hashed_crossing.html">layer_hashed_crossing</a>()</code> <br><code><a href="layer_hashing.html">layer_hashing</a>()</code> <br><code><a href="layer_integer_lookup.html">layer_integer_lookup</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_mel_spectrogram.html">layer_mel_spectrogram</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_normalization.html">layer_normalization</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_brightness.html">layer_random_brightness</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_contrast.html">layer_random_contrast</a>()</code> <br><code><a href="layer_random_crop.html">layer_random_crop</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_flip.html">layer_random_flip</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_rotation.html">layer_random_rotation</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_random_translation.html">layer_random_translation</a>()</code> <br><code><a href="layer_random_zoom.html">layer_random_zoom</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br><code><a href="layer_stft_spectrogram.html">layer_stft_spectrogram</a>()</code> <br><code><a href="layer_string_lookup.html">layer_string_lookup</a>()</code> <br><code><a href="layer_text_vectorization.html">layer_text_vectorization</a>()</code> <br></p>
-<p>Other layers: <br><code><a href="Layer.html">Layer</a>()</code> <br><code><a href="layer_activation.html">layer_activation</a>()</code> <br><code><a href="layer_activation_elu.html">layer_activation_elu</a>()</code> <br><code><a href="layer_activation_leaky_relu.html">layer_activation_leaky_relu</a>()</code> <br><code><a href="layer_activation_parametric_relu.html">layer_activation_parametric_relu</a>()</code> <br><code><a href="layer_activation_relu.html">layer_activation_relu</a>()</code> <br><code><a href="layer_activation_softmax.html">layer_activation_softmax</a>()</code> <br><code><a href="layer_activity_regularization.html">layer_activity_regularization</a>()</code> <br><code><a href="layer_add.html">layer_add</a>()</code> <br><code><a href="layer_additive_attention.html">layer_additive_attention</a>()</code> <br><code><a href="layer_alpha_dropout.html">layer_alpha_dropout</a>()</code> <br><code><a href="layer_attention.html">layer_attention</a>()</code> <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_average.html">layer_average</a>()</code> <br><code><a href="layer_average_pooling_1d.html">layer_average_pooling_1d</a>()</code> <br><code><a href="layer_average_pooling_2d.html">layer_average_pooling_2d</a>()</code> <br><code><a href="layer_average_pooling_3d.html">layer_average_pooling_3d</a>()</code> <br><code><a href="layer_batch_normalization.html">layer_batch_normalization</a>()</code> <br><code><a href="layer_bidirectional.html">layer_bidirectional</a>()</code> <br><code><a href="layer_category_encoding.html">layer_category_encoding</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_concatenate.html">layer_concatenate</a>()</code> <br><code><a href="layer_conv_1d.html">layer_conv_1d</a>()</code> <br><code><a href="layer_conv_1d_transpose.html">layer_conv_1d_transpose</a>()</code> <br><code><a href="layer_conv_2d.html">layer_conv_2d</a>()</code> <br><code><a href="layer_conv_2d_transpose.html">layer_conv_2d_transpose</a>()</code> <br><code><a href="layer_conv_3d.html">layer_conv_3d</a>()</code> <br><code><a href="layer_conv_3d_transpose.html">layer_conv_3d_transpose</a>()</code> <br><code><a href="layer_conv_lstm_1d.html">layer_conv_lstm_1d</a>()</code> <br><code><a href="layer_conv_lstm_2d.html">layer_conv_lstm_2d</a>()</code> <br><code><a href="layer_conv_lstm_3d.html">layer_conv_lstm_3d</a>()</code> <br><code><a href="layer_cropping_1d.html">layer_cropping_1d</a>()</code> <br><code><a href="layer_cropping_2d.html">layer_cropping_2d</a>()</code> <br><code><a href="layer_cropping_3d.html">layer_cropping_3d</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_dense.html">layer_dense</a>()</code> <br><code><a href="layer_depthwise_conv_1d.html">layer_depthwise_conv_1d</a>()</code> <br><code><a href="layer_depthwise_conv_2d.html">layer_depthwise_conv_2d</a>()</code> <br><code><a href="layer_discretization.html">layer_discretization</a>()</code> <br><code><a href="layer_dot.html">layer_dot</a>()</code> <br><code><a href="layer_dropout.html">layer_dropout</a>()</code> <br><code><a href="layer_einsum_dense.html">layer_einsum_dense</a>()</code> <br><code><a href="layer_embedding.html">layer_embedding</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_feature_space.html">layer_feature_space</a>()</code> <br><code><a href="layer_flatten.html">layer_flatten</a>()</code> <br><code><a href="layer_flax_module_wrapper.html">layer_flax_module_wrapper</a>()</code> <br><code><a href="layer_gaussian_dropout.html">layer_gaussian_dropout</a>()</code> <br><code><a href="layer_gaussian_noise.html">layer_gaussian_noise</a>()</code> <br><code><a href="layer_global_average_pooling_1d.html">layer_global_average_pooling_1d</a>()</code> <br><code><a href="layer_global_average_pooling_2d.html">layer_global_average_pooling_2d</a>()</code> <br><code><a href="layer_global_average_pooling_3d.html">layer_global_average_pooling_3d</a>()</code> <br><code><a href="layer_global_max_pooling_1d.html">layer_global_max_pooling_1d</a>()</code> <br><code><a href="layer_global_max_pooling_2d.html">layer_global_max_pooling_2d</a>()</code> <br><code><a href="layer_global_max_pooling_3d.html">layer_global_max_pooling_3d</a>()</code> <br><code><a href="layer_group_normalization.html">layer_group_normalization</a>()</code> <br><code><a href="layer_group_query_attention.html">layer_group_query_attention</a>()</code> <br><code><a href="layer_gru.html">layer_gru</a>()</code> <br><code><a href="layer_hashed_crossing.html">layer_hashed_crossing</a>()</code> <br><code><a href="layer_hashing.html">layer_hashing</a>()</code> <br><code><a href="layer_identity.html">layer_identity</a>()</code> <br><code><a href="layer_integer_lookup.html">layer_integer_lookup</a>()</code> <br><code><a href="layer_jax_model_wrapper.html">layer_jax_model_wrapper</a>()</code> <br><code><a href="layer_lambda.html">layer_lambda</a>()</code> <br><code><a href="layer_layer_normalization.html">layer_layer_normalization</a>()</code> <br><code><a href="layer_lstm.html">layer_lstm</a>()</code> <br><code><a href="layer_masking.html">layer_masking</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_max_pooling_1d.html">layer_max_pooling_1d</a>()</code> <br><code><a href="layer_max_pooling_2d.html">layer_max_pooling_2d</a>()</code> <br><code><a href="layer_max_pooling_3d.html">layer_max_pooling_3d</a>()</code> <br><code><a href="layer_maximum.html">layer_maximum</a>()</code> <br><code><a href="layer_mel_spectrogram.html">layer_mel_spectrogram</a>()</code> <br><code><a href="layer_minimum.html">layer_minimum</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_multi_head_attention.html">layer_multi_head_attention</a>()</code> <br><code><a href="layer_multiply.html">layer_multiply</a>()</code> <br><code><a href="layer_normalization.html">layer_normalization</a>()</code> <br><code><a href="layer_permute.html">layer_permute</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_brightness.html">layer_random_brightness</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_contrast.html">layer_random_contrast</a>()</code> <br><code><a href="layer_random_crop.html">layer_random_crop</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_flip.html">layer_random_flip</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_rotation.html">layer_random_rotation</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_random_translation.html">layer_random_translation</a>()</code> <br><code><a href="layer_random_zoom.html">layer_random_zoom</a>()</code> <br><code><a href="layer_repeat_vector.html">layer_repeat_vector</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_reshape.html">layer_reshape</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_rms_normalization.html">layer_rms_normalization</a>()</code> <br><code><a href="layer_rnn.html">layer_rnn</a>()</code> <br><code><a href="layer_separable_conv_1d.html">layer_separable_conv_1d</a>()</code> <br><code><a href="layer_separable_conv_2d.html">layer_separable_conv_2d</a>()</code> <br><code><a href="layer_simple_rnn.html">layer_simple_rnn</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br><code><a href="layer_spatial_dropout_1d.html">layer_spatial_dropout_1d</a>()</code> <br><code><a href="layer_spatial_dropout_2d.html">layer_spatial_dropout_2d</a>()</code> <br><code><a href="layer_spatial_dropout_3d.html">layer_spatial_dropout_3d</a>()</code> <br><code><a href="layer_spectral_normalization.html">layer_spectral_normalization</a>()</code> <br><code><a href="layer_stft_spectrogram.html">layer_stft_spectrogram</a>()</code> <br><code><a href="layer_string_lookup.html">layer_string_lookup</a>()</code> <br><code><a href="layer_subtract.html">layer_subtract</a>()</code> <br><code><a href="layer_text_vectorization.html">layer_text_vectorization</a>()</code> <br><code><a href="layer_tfsm.html">layer_tfsm</a>()</code> <br><code><a href="layer_time_distributed.html">layer_time_distributed</a>()</code> <br><code><a href="layer_torch_module_wrapper.html">layer_torch_module_wrapper</a>()</code> <br><code><a href="layer_unit_normalization.html">layer_unit_normalization</a>()</code> <br><code><a href="layer_upsampling_1d.html">layer_upsampling_1d</a>()</code> <br><code><a href="layer_upsampling_2d.html">layer_upsampling_2d</a>()</code> <br><code><a href="layer_upsampling_3d.html">layer_upsampling_3d</a>()</code> <br><code><a href="layer_zero_padding_1d.html">layer_zero_padding_1d</a>()</code> <br><code><a href="layer_zero_padding_2d.html">layer_zero_padding_2d</a>()</code> <br><code><a href="layer_zero_padding_3d.html">layer_zero_padding_3d</a>()</code> <br><code><a href="rnn_cell_gru.html">rnn_cell_gru</a>()</code> <br><code><a href="rnn_cell_lstm.html">rnn_cell_lstm</a>()</code> <br><code><a href="rnn_cell_simple.html">rnn_cell_simple</a>()</code> <br><code><a href="rnn_cells_stack.html">rnn_cells_stack</a>()</code> <br></p></div>
+    <div class="dont-index"><p>Other image preprocessing layers: <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_elastic_transform.html">layer_random_elastic_transform</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br></p>
+<p>Other preprocessing layers: <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_category_encoding.html">layer_category_encoding</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_discretization.html">layer_discretization</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_feature_space.html">layer_feature_space</a>()</code> <br><code><a href="layer_hashed_crossing.html">layer_hashed_crossing</a>()</code> <br><code><a href="layer_hashing.html">layer_hashing</a>()</code> <br><code><a href="layer_integer_lookup.html">layer_integer_lookup</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_mel_spectrogram.html">layer_mel_spectrogram</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_normalization.html">layer_normalization</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_brightness.html">layer_random_brightness</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_contrast.html">layer_random_contrast</a>()</code> <br><code><a href="layer_random_crop.html">layer_random_crop</a>()</code> <br><code><a href="layer_random_elastic_transform.html">layer_random_elastic_transform</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_flip.html">layer_random_flip</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_rotation.html">layer_random_rotation</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_random_translation.html">layer_random_translation</a>()</code> <br><code><a href="layer_random_zoom.html">layer_random_zoom</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br><code><a href="layer_stft_spectrogram.html">layer_stft_spectrogram</a>()</code> <br><code><a href="layer_string_lookup.html">layer_string_lookup</a>()</code> <br><code><a href="layer_text_vectorization.html">layer_text_vectorization</a>()</code> <br></p>
+<p>Other layers: <br><code><a href="Layer.html">Layer</a>()</code> <br><code><a href="layer_activation.html">layer_activation</a>()</code> <br><code><a href="layer_activation_elu.html">layer_activation_elu</a>()</code> <br><code><a href="layer_activation_leaky_relu.html">layer_activation_leaky_relu</a>()</code> <br><code><a href="layer_activation_parametric_relu.html">layer_activation_parametric_relu</a>()</code> <br><code><a href="layer_activation_relu.html">layer_activation_relu</a>()</code> <br><code><a href="layer_activation_softmax.html">layer_activation_softmax</a>()</code> <br><code><a href="layer_activity_regularization.html">layer_activity_regularization</a>()</code> <br><code><a href="layer_add.html">layer_add</a>()</code> <br><code><a href="layer_additive_attention.html">layer_additive_attention</a>()</code> <br><code><a href="layer_alpha_dropout.html">layer_alpha_dropout</a>()</code> <br><code><a href="layer_attention.html">layer_attention</a>()</code> <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_average.html">layer_average</a>()</code> <br><code><a href="layer_average_pooling_1d.html">layer_average_pooling_1d</a>()</code> <br><code><a href="layer_average_pooling_2d.html">layer_average_pooling_2d</a>()</code> <br><code><a href="layer_average_pooling_3d.html">layer_average_pooling_3d</a>()</code> <br><code><a href="layer_batch_normalization.html">layer_batch_normalization</a>()</code> <br><code><a href="layer_bidirectional.html">layer_bidirectional</a>()</code> <br><code><a href="layer_category_encoding.html">layer_category_encoding</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_concatenate.html">layer_concatenate</a>()</code> <br><code><a href="layer_conv_1d.html">layer_conv_1d</a>()</code> <br><code><a href="layer_conv_1d_transpose.html">layer_conv_1d_transpose</a>()</code> <br><code><a href="layer_conv_2d.html">layer_conv_2d</a>()</code> <br><code><a href="layer_conv_2d_transpose.html">layer_conv_2d_transpose</a>()</code> <br><code><a href="layer_conv_3d.html">layer_conv_3d</a>()</code> <br><code><a href="layer_conv_3d_transpose.html">layer_conv_3d_transpose</a>()</code> <br><code><a href="layer_conv_lstm_1d.html">layer_conv_lstm_1d</a>()</code> <br><code><a href="layer_conv_lstm_2d.html">layer_conv_lstm_2d</a>()</code> <br><code><a href="layer_conv_lstm_3d.html">layer_conv_lstm_3d</a>()</code> <br><code><a href="layer_cropping_1d.html">layer_cropping_1d</a>()</code> <br><code><a href="layer_cropping_2d.html">layer_cropping_2d</a>()</code> <br><code><a href="layer_cropping_3d.html">layer_cropping_3d</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_dense.html">layer_dense</a>()</code> <br><code><a href="layer_depthwise_conv_1d.html">layer_depthwise_conv_1d</a>()</code> <br><code><a href="layer_depthwise_conv_2d.html">layer_depthwise_conv_2d</a>()</code> <br><code><a href="layer_discretization.html">layer_discretization</a>()</code> <br><code><a href="layer_dot.html">layer_dot</a>()</code> <br><code><a href="layer_dropout.html">layer_dropout</a>()</code> <br><code><a href="layer_einsum_dense.html">layer_einsum_dense</a>()</code> <br><code><a href="layer_embedding.html">layer_embedding</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_feature_space.html">layer_feature_space</a>()</code> <br><code><a href="layer_flatten.html">layer_flatten</a>()</code> <br><code><a href="layer_flax_module_wrapper.html">layer_flax_module_wrapper</a>()</code> <br><code><a href="layer_gaussian_dropout.html">layer_gaussian_dropout</a>()</code> <br><code><a href="layer_gaussian_noise.html">layer_gaussian_noise</a>()</code> <br><code><a href="layer_global_average_pooling_1d.html">layer_global_average_pooling_1d</a>()</code> <br><code><a href="layer_global_average_pooling_2d.html">layer_global_average_pooling_2d</a>()</code> <br><code><a href="layer_global_average_pooling_3d.html">layer_global_average_pooling_3d</a>()</code> <br><code><a href="layer_global_max_pooling_1d.html">layer_global_max_pooling_1d</a>()</code> <br><code><a href="layer_global_max_pooling_2d.html">layer_global_max_pooling_2d</a>()</code> <br><code><a href="layer_global_max_pooling_3d.html">layer_global_max_pooling_3d</a>()</code> <br><code><a href="layer_group_normalization.html">layer_group_normalization</a>()</code> <br><code><a href="layer_group_query_attention.html">layer_group_query_attention</a>()</code> <br><code><a href="layer_gru.html">layer_gru</a>()</code> <br><code><a href="layer_hashed_crossing.html">layer_hashed_crossing</a>()</code> <br><code><a href="layer_hashing.html">layer_hashing</a>()</code> <br><code><a href="layer_identity.html">layer_identity</a>()</code> <br><code><a href="layer_integer_lookup.html">layer_integer_lookup</a>()</code> <br><code><a href="layer_jax_model_wrapper.html">layer_jax_model_wrapper</a>()</code> <br><code><a href="layer_lambda.html">layer_lambda</a>()</code> <br><code><a href="layer_layer_normalization.html">layer_layer_normalization</a>()</code> <br><code><a href="layer_lstm.html">layer_lstm</a>()</code> <br><code><a href="layer_masking.html">layer_masking</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_max_pooling_1d.html">layer_max_pooling_1d</a>()</code> <br><code><a href="layer_max_pooling_2d.html">layer_max_pooling_2d</a>()</code> <br><code><a href="layer_max_pooling_3d.html">layer_max_pooling_3d</a>()</code> <br><code><a href="layer_maximum.html">layer_maximum</a>()</code> <br><code><a href="layer_mel_spectrogram.html">layer_mel_spectrogram</a>()</code> <br><code><a href="layer_minimum.html">layer_minimum</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_multi_head_attention.html">layer_multi_head_attention</a>()</code> <br><code><a href="layer_multiply.html">layer_multiply</a>()</code> <br><code><a href="layer_normalization.html">layer_normalization</a>()</code> <br><code><a href="layer_permute.html">layer_permute</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_brightness.html">layer_random_brightness</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_contrast.html">layer_random_contrast</a>()</code> <br><code><a href="layer_random_crop.html">layer_random_crop</a>()</code> <br><code><a href="layer_random_elastic_transform.html">layer_random_elastic_transform</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_flip.html">layer_random_flip</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_rotation.html">layer_random_rotation</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_random_translation.html">layer_random_translation</a>()</code> <br><code><a href="layer_random_zoom.html">layer_random_zoom</a>()</code> <br><code><a href="layer_repeat_vector.html">layer_repeat_vector</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_reshape.html">layer_reshape</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_rms_normalization.html">layer_rms_normalization</a>()</code> <br><code><a href="layer_rnn.html">layer_rnn</a>()</code> <br><code><a href="layer_separable_conv_1d.html">layer_separable_conv_1d</a>()</code> <br><code><a href="layer_separable_conv_2d.html">layer_separable_conv_2d</a>()</code> <br><code><a href="layer_simple_rnn.html">layer_simple_rnn</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br><code><a href="layer_spatial_dropout_1d.html">layer_spatial_dropout_1d</a>()</code> <br><code><a href="layer_spatial_dropout_2d.html">layer_spatial_dropout_2d</a>()</code> <br><code><a href="layer_spatial_dropout_3d.html">layer_spatial_dropout_3d</a>()</code> <br><code><a href="layer_spectral_normalization.html">layer_spectral_normalization</a>()</code> <br><code><a href="layer_stft_spectrogram.html">layer_stft_spectrogram</a>()</code> <br><code><a href="layer_string_lookup.html">layer_string_lookup</a>()</code> <br><code><a href="layer_subtract.html">layer_subtract</a>()</code> <br><code><a href="layer_text_vectorization.html">layer_text_vectorization</a>()</code> <br><code><a href="layer_tfsm.html">layer_tfsm</a>()</code> <br><code><a href="layer_time_distributed.html">layer_time_distributed</a>()</code> <br><code><a href="layer_torch_module_wrapper.html">layer_torch_module_wrapper</a>()</code> <br><code><a href="layer_unit_normalization.html">layer_unit_normalization</a>()</code> <br><code><a href="layer_upsampling_1d.html">layer_upsampling_1d</a>()</code> <br><code><a href="layer_upsampling_2d.html">layer_upsampling_2d</a>()</code> <br><code><a href="layer_upsampling_3d.html">layer_upsampling_3d</a>()</code> <br><code><a href="layer_zero_padding_1d.html">layer_zero_padding_1d</a>()</code> <br><code><a href="layer_zero_padding_2d.html">layer_zero_padding_2d</a>()</code> <br><code><a href="layer_zero_padding_3d.html">layer_zero_padding_3d</a>()</code> <br><code><a href="rnn_cell_gru.html">rnn_cell_gru</a>()</code> <br><code><a href="rnn_cell_lstm.html">rnn_cell_lstm</a>()</code> <br><code><a href="rnn_cell_simple.html">rnn_cell_simple</a>()</code> <br><code><a href="rnn_cells_stack.html">rnn_cells_stack</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -155,7 +155,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/layer_random_hue.html b/docs/dev/reference/layer_random_hue.html
index a4780c604..6d7f9eb2a 100644
--- a/docs/dev/reference/layer_random_hue.html
+++ b/docs/dev/reference/layer_random_hue.html
@@ -14,7 +14,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -131,9 +131,9 @@ <h2 id="examples">Examples<a class="anchor" aria-label="anchor" href="#examples"
     </div>
     <div class="section level2">
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
-    <div class="dont-index"><p>Other image preprocessing layers: <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br></p>
-<p>Other preprocessing layers: <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_category_encoding.html">layer_category_encoding</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_discretization.html">layer_discretization</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_feature_space.html">layer_feature_space</a>()</code> <br><code><a href="layer_hashed_crossing.html">layer_hashed_crossing</a>()</code> <br><code><a href="layer_hashing.html">layer_hashing</a>()</code> <br><code><a href="layer_integer_lookup.html">layer_integer_lookup</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_mel_spectrogram.html">layer_mel_spectrogram</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_normalization.html">layer_normalization</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_brightness.html">layer_random_brightness</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_contrast.html">layer_random_contrast</a>()</code> <br><code><a href="layer_random_crop.html">layer_random_crop</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_flip.html">layer_random_flip</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_rotation.html">layer_random_rotation</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_random_translation.html">layer_random_translation</a>()</code> <br><code><a href="layer_random_zoom.html">layer_random_zoom</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br><code><a href="layer_stft_spectrogram.html">layer_stft_spectrogram</a>()</code> <br><code><a href="layer_string_lookup.html">layer_string_lookup</a>()</code> <br><code><a href="layer_text_vectorization.html">layer_text_vectorization</a>()</code> <br></p>
-<p>Other layers: <br><code><a href="Layer.html">Layer</a>()</code> <br><code><a href="layer_activation.html">layer_activation</a>()</code> <br><code><a href="layer_activation_elu.html">layer_activation_elu</a>()</code> <br><code><a href="layer_activation_leaky_relu.html">layer_activation_leaky_relu</a>()</code> <br><code><a href="layer_activation_parametric_relu.html">layer_activation_parametric_relu</a>()</code> <br><code><a href="layer_activation_relu.html">layer_activation_relu</a>()</code> <br><code><a href="layer_activation_softmax.html">layer_activation_softmax</a>()</code> <br><code><a href="layer_activity_regularization.html">layer_activity_regularization</a>()</code> <br><code><a href="layer_add.html">layer_add</a>()</code> <br><code><a href="layer_additive_attention.html">layer_additive_attention</a>()</code> <br><code><a href="layer_alpha_dropout.html">layer_alpha_dropout</a>()</code> <br><code><a href="layer_attention.html">layer_attention</a>()</code> <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_average.html">layer_average</a>()</code> <br><code><a href="layer_average_pooling_1d.html">layer_average_pooling_1d</a>()</code> <br><code><a href="layer_average_pooling_2d.html">layer_average_pooling_2d</a>()</code> <br><code><a href="layer_average_pooling_3d.html">layer_average_pooling_3d</a>()</code> <br><code><a href="layer_batch_normalization.html">layer_batch_normalization</a>()</code> <br><code><a href="layer_bidirectional.html">layer_bidirectional</a>()</code> <br><code><a href="layer_category_encoding.html">layer_category_encoding</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_concatenate.html">layer_concatenate</a>()</code> <br><code><a href="layer_conv_1d.html">layer_conv_1d</a>()</code> <br><code><a href="layer_conv_1d_transpose.html">layer_conv_1d_transpose</a>()</code> <br><code><a href="layer_conv_2d.html">layer_conv_2d</a>()</code> <br><code><a href="layer_conv_2d_transpose.html">layer_conv_2d_transpose</a>()</code> <br><code><a href="layer_conv_3d.html">layer_conv_3d</a>()</code> <br><code><a href="layer_conv_3d_transpose.html">layer_conv_3d_transpose</a>()</code> <br><code><a href="layer_conv_lstm_1d.html">layer_conv_lstm_1d</a>()</code> <br><code><a href="layer_conv_lstm_2d.html">layer_conv_lstm_2d</a>()</code> <br><code><a href="layer_conv_lstm_3d.html">layer_conv_lstm_3d</a>()</code> <br><code><a href="layer_cropping_1d.html">layer_cropping_1d</a>()</code> <br><code><a href="layer_cropping_2d.html">layer_cropping_2d</a>()</code> <br><code><a href="layer_cropping_3d.html">layer_cropping_3d</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_dense.html">layer_dense</a>()</code> <br><code><a href="layer_depthwise_conv_1d.html">layer_depthwise_conv_1d</a>()</code> <br><code><a href="layer_depthwise_conv_2d.html">layer_depthwise_conv_2d</a>()</code> <br><code><a href="layer_discretization.html">layer_discretization</a>()</code> <br><code><a href="layer_dot.html">layer_dot</a>()</code> <br><code><a href="layer_dropout.html">layer_dropout</a>()</code> <br><code><a href="layer_einsum_dense.html">layer_einsum_dense</a>()</code> <br><code><a href="layer_embedding.html">layer_embedding</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_feature_space.html">layer_feature_space</a>()</code> <br><code><a href="layer_flatten.html">layer_flatten</a>()</code> <br><code><a href="layer_flax_module_wrapper.html">layer_flax_module_wrapper</a>()</code> <br><code><a href="layer_gaussian_dropout.html">layer_gaussian_dropout</a>()</code> <br><code><a href="layer_gaussian_noise.html">layer_gaussian_noise</a>()</code> <br><code><a href="layer_global_average_pooling_1d.html">layer_global_average_pooling_1d</a>()</code> <br><code><a href="layer_global_average_pooling_2d.html">layer_global_average_pooling_2d</a>()</code> <br><code><a href="layer_global_average_pooling_3d.html">layer_global_average_pooling_3d</a>()</code> <br><code><a href="layer_global_max_pooling_1d.html">layer_global_max_pooling_1d</a>()</code> <br><code><a href="layer_global_max_pooling_2d.html">layer_global_max_pooling_2d</a>()</code> <br><code><a href="layer_global_max_pooling_3d.html">layer_global_max_pooling_3d</a>()</code> <br><code><a href="layer_group_normalization.html">layer_group_normalization</a>()</code> <br><code><a href="layer_group_query_attention.html">layer_group_query_attention</a>()</code> <br><code><a href="layer_gru.html">layer_gru</a>()</code> <br><code><a href="layer_hashed_crossing.html">layer_hashed_crossing</a>()</code> <br><code><a href="layer_hashing.html">layer_hashing</a>()</code> <br><code><a href="layer_identity.html">layer_identity</a>()</code> <br><code><a href="layer_integer_lookup.html">layer_integer_lookup</a>()</code> <br><code><a href="layer_jax_model_wrapper.html">layer_jax_model_wrapper</a>()</code> <br><code><a href="layer_lambda.html">layer_lambda</a>()</code> <br><code><a href="layer_layer_normalization.html">layer_layer_normalization</a>()</code> <br><code><a href="layer_lstm.html">layer_lstm</a>()</code> <br><code><a href="layer_masking.html">layer_masking</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_max_pooling_1d.html">layer_max_pooling_1d</a>()</code> <br><code><a href="layer_max_pooling_2d.html">layer_max_pooling_2d</a>()</code> <br><code><a href="layer_max_pooling_3d.html">layer_max_pooling_3d</a>()</code> <br><code><a href="layer_maximum.html">layer_maximum</a>()</code> <br><code><a href="layer_mel_spectrogram.html">layer_mel_spectrogram</a>()</code> <br><code><a href="layer_minimum.html">layer_minimum</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_multi_head_attention.html">layer_multi_head_attention</a>()</code> <br><code><a href="layer_multiply.html">layer_multiply</a>()</code> <br><code><a href="layer_normalization.html">layer_normalization</a>()</code> <br><code><a href="layer_permute.html">layer_permute</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_brightness.html">layer_random_brightness</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_contrast.html">layer_random_contrast</a>()</code> <br><code><a href="layer_random_crop.html">layer_random_crop</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_flip.html">layer_random_flip</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_rotation.html">layer_random_rotation</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_random_translation.html">layer_random_translation</a>()</code> <br><code><a href="layer_random_zoom.html">layer_random_zoom</a>()</code> <br><code><a href="layer_repeat_vector.html">layer_repeat_vector</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_reshape.html">layer_reshape</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_rms_normalization.html">layer_rms_normalization</a>()</code> <br><code><a href="layer_rnn.html">layer_rnn</a>()</code> <br><code><a href="layer_separable_conv_1d.html">layer_separable_conv_1d</a>()</code> <br><code><a href="layer_separable_conv_2d.html">layer_separable_conv_2d</a>()</code> <br><code><a href="layer_simple_rnn.html">layer_simple_rnn</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br><code><a href="layer_spatial_dropout_1d.html">layer_spatial_dropout_1d</a>()</code> <br><code><a href="layer_spatial_dropout_2d.html">layer_spatial_dropout_2d</a>()</code> <br><code><a href="layer_spatial_dropout_3d.html">layer_spatial_dropout_3d</a>()</code> <br><code><a href="layer_spectral_normalization.html">layer_spectral_normalization</a>()</code> <br><code><a href="layer_stft_spectrogram.html">layer_stft_spectrogram</a>()</code> <br><code><a href="layer_string_lookup.html">layer_string_lookup</a>()</code> <br><code><a href="layer_subtract.html">layer_subtract</a>()</code> <br><code><a href="layer_text_vectorization.html">layer_text_vectorization</a>()</code> <br><code><a href="layer_tfsm.html">layer_tfsm</a>()</code> <br><code><a href="layer_time_distributed.html">layer_time_distributed</a>()</code> <br><code><a href="layer_torch_module_wrapper.html">layer_torch_module_wrapper</a>()</code> <br><code><a href="layer_unit_normalization.html">layer_unit_normalization</a>()</code> <br><code><a href="layer_upsampling_1d.html">layer_upsampling_1d</a>()</code> <br><code><a href="layer_upsampling_2d.html">layer_upsampling_2d</a>()</code> <br><code><a href="layer_upsampling_3d.html">layer_upsampling_3d</a>()</code> <br><code><a href="layer_zero_padding_1d.html">layer_zero_padding_1d</a>()</code> <br><code><a href="layer_zero_padding_2d.html">layer_zero_padding_2d</a>()</code> <br><code><a href="layer_zero_padding_3d.html">layer_zero_padding_3d</a>()</code> <br><code><a href="rnn_cell_gru.html">rnn_cell_gru</a>()</code> <br><code><a href="rnn_cell_lstm.html">rnn_cell_lstm</a>()</code> <br><code><a href="rnn_cell_simple.html">rnn_cell_simple</a>()</code> <br><code><a href="rnn_cells_stack.html">rnn_cells_stack</a>()</code> <br></p></div>
+    <div class="dont-index"><p>Other image preprocessing layers: <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_elastic_transform.html">layer_random_elastic_transform</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br></p>
+<p>Other preprocessing layers: <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_category_encoding.html">layer_category_encoding</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_discretization.html">layer_discretization</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_feature_space.html">layer_feature_space</a>()</code> <br><code><a href="layer_hashed_crossing.html">layer_hashed_crossing</a>()</code> <br><code><a href="layer_hashing.html">layer_hashing</a>()</code> <br><code><a href="layer_integer_lookup.html">layer_integer_lookup</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_mel_spectrogram.html">layer_mel_spectrogram</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_normalization.html">layer_normalization</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_brightness.html">layer_random_brightness</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_contrast.html">layer_random_contrast</a>()</code> <br><code><a href="layer_random_crop.html">layer_random_crop</a>()</code> <br><code><a href="layer_random_elastic_transform.html">layer_random_elastic_transform</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_flip.html">layer_random_flip</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_rotation.html">layer_random_rotation</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_random_translation.html">layer_random_translation</a>()</code> <br><code><a href="layer_random_zoom.html">layer_random_zoom</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br><code><a href="layer_stft_spectrogram.html">layer_stft_spectrogram</a>()</code> <br><code><a href="layer_string_lookup.html">layer_string_lookup</a>()</code> <br><code><a href="layer_text_vectorization.html">layer_text_vectorization</a>()</code> <br></p>
+<p>Other layers: <br><code><a href="Layer.html">Layer</a>()</code> <br><code><a href="layer_activation.html">layer_activation</a>()</code> <br><code><a href="layer_activation_elu.html">layer_activation_elu</a>()</code> <br><code><a href="layer_activation_leaky_relu.html">layer_activation_leaky_relu</a>()</code> <br><code><a href="layer_activation_parametric_relu.html">layer_activation_parametric_relu</a>()</code> <br><code><a href="layer_activation_relu.html">layer_activation_relu</a>()</code> <br><code><a href="layer_activation_softmax.html">layer_activation_softmax</a>()</code> <br><code><a href="layer_activity_regularization.html">layer_activity_regularization</a>()</code> <br><code><a href="layer_add.html">layer_add</a>()</code> <br><code><a href="layer_additive_attention.html">layer_additive_attention</a>()</code> <br><code><a href="layer_alpha_dropout.html">layer_alpha_dropout</a>()</code> <br><code><a href="layer_attention.html">layer_attention</a>()</code> <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_average.html">layer_average</a>()</code> <br><code><a href="layer_average_pooling_1d.html">layer_average_pooling_1d</a>()</code> <br><code><a href="layer_average_pooling_2d.html">layer_average_pooling_2d</a>()</code> <br><code><a href="layer_average_pooling_3d.html">layer_average_pooling_3d</a>()</code> <br><code><a href="layer_batch_normalization.html">layer_batch_normalization</a>()</code> <br><code><a href="layer_bidirectional.html">layer_bidirectional</a>()</code> <br><code><a href="layer_category_encoding.html">layer_category_encoding</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_concatenate.html">layer_concatenate</a>()</code> <br><code><a href="layer_conv_1d.html">layer_conv_1d</a>()</code> <br><code><a href="layer_conv_1d_transpose.html">layer_conv_1d_transpose</a>()</code> <br><code><a href="layer_conv_2d.html">layer_conv_2d</a>()</code> <br><code><a href="layer_conv_2d_transpose.html">layer_conv_2d_transpose</a>()</code> <br><code><a href="layer_conv_3d.html">layer_conv_3d</a>()</code> <br><code><a href="layer_conv_3d_transpose.html">layer_conv_3d_transpose</a>()</code> <br><code><a href="layer_conv_lstm_1d.html">layer_conv_lstm_1d</a>()</code> <br><code><a href="layer_conv_lstm_2d.html">layer_conv_lstm_2d</a>()</code> <br><code><a href="layer_conv_lstm_3d.html">layer_conv_lstm_3d</a>()</code> <br><code><a href="layer_cropping_1d.html">layer_cropping_1d</a>()</code> <br><code><a href="layer_cropping_2d.html">layer_cropping_2d</a>()</code> <br><code><a href="layer_cropping_3d.html">layer_cropping_3d</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_dense.html">layer_dense</a>()</code> <br><code><a href="layer_depthwise_conv_1d.html">layer_depthwise_conv_1d</a>()</code> <br><code><a href="layer_depthwise_conv_2d.html">layer_depthwise_conv_2d</a>()</code> <br><code><a href="layer_discretization.html">layer_discretization</a>()</code> <br><code><a href="layer_dot.html">layer_dot</a>()</code> <br><code><a href="layer_dropout.html">layer_dropout</a>()</code> <br><code><a href="layer_einsum_dense.html">layer_einsum_dense</a>()</code> <br><code><a href="layer_embedding.html">layer_embedding</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_feature_space.html">layer_feature_space</a>()</code> <br><code><a href="layer_flatten.html">layer_flatten</a>()</code> <br><code><a href="layer_flax_module_wrapper.html">layer_flax_module_wrapper</a>()</code> <br><code><a href="layer_gaussian_dropout.html">layer_gaussian_dropout</a>()</code> <br><code><a href="layer_gaussian_noise.html">layer_gaussian_noise</a>()</code> <br><code><a href="layer_global_average_pooling_1d.html">layer_global_average_pooling_1d</a>()</code> <br><code><a href="layer_global_average_pooling_2d.html">layer_global_average_pooling_2d</a>()</code> <br><code><a href="layer_global_average_pooling_3d.html">layer_global_average_pooling_3d</a>()</code> <br><code><a href="layer_global_max_pooling_1d.html">layer_global_max_pooling_1d</a>()</code> <br><code><a href="layer_global_max_pooling_2d.html">layer_global_max_pooling_2d</a>()</code> <br><code><a href="layer_global_max_pooling_3d.html">layer_global_max_pooling_3d</a>()</code> <br><code><a href="layer_group_normalization.html">layer_group_normalization</a>()</code> <br><code><a href="layer_group_query_attention.html">layer_group_query_attention</a>()</code> <br><code><a href="layer_gru.html">layer_gru</a>()</code> <br><code><a href="layer_hashed_crossing.html">layer_hashed_crossing</a>()</code> <br><code><a href="layer_hashing.html">layer_hashing</a>()</code> <br><code><a href="layer_identity.html">layer_identity</a>()</code> <br><code><a href="layer_integer_lookup.html">layer_integer_lookup</a>()</code> <br><code><a href="layer_jax_model_wrapper.html">layer_jax_model_wrapper</a>()</code> <br><code><a href="layer_lambda.html">layer_lambda</a>()</code> <br><code><a href="layer_layer_normalization.html">layer_layer_normalization</a>()</code> <br><code><a href="layer_lstm.html">layer_lstm</a>()</code> <br><code><a href="layer_masking.html">layer_masking</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_max_pooling_1d.html">layer_max_pooling_1d</a>()</code> <br><code><a href="layer_max_pooling_2d.html">layer_max_pooling_2d</a>()</code> <br><code><a href="layer_max_pooling_3d.html">layer_max_pooling_3d</a>()</code> <br><code><a href="layer_maximum.html">layer_maximum</a>()</code> <br><code><a href="layer_mel_spectrogram.html">layer_mel_spectrogram</a>()</code> <br><code><a href="layer_minimum.html">layer_minimum</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_multi_head_attention.html">layer_multi_head_attention</a>()</code> <br><code><a href="layer_multiply.html">layer_multiply</a>()</code> <br><code><a href="layer_normalization.html">layer_normalization</a>()</code> <br><code><a href="layer_permute.html">layer_permute</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_brightness.html">layer_random_brightness</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_contrast.html">layer_random_contrast</a>()</code> <br><code><a href="layer_random_crop.html">layer_random_crop</a>()</code> <br><code><a href="layer_random_elastic_transform.html">layer_random_elastic_transform</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_flip.html">layer_random_flip</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_rotation.html">layer_random_rotation</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_random_translation.html">layer_random_translation</a>()</code> <br><code><a href="layer_random_zoom.html">layer_random_zoom</a>()</code> <br><code><a href="layer_repeat_vector.html">layer_repeat_vector</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_reshape.html">layer_reshape</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_rms_normalization.html">layer_rms_normalization</a>()</code> <br><code><a href="layer_rnn.html">layer_rnn</a>()</code> <br><code><a href="layer_separable_conv_1d.html">layer_separable_conv_1d</a>()</code> <br><code><a href="layer_separable_conv_2d.html">layer_separable_conv_2d</a>()</code> <br><code><a href="layer_simple_rnn.html">layer_simple_rnn</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br><code><a href="layer_spatial_dropout_1d.html">layer_spatial_dropout_1d</a>()</code> <br><code><a href="layer_spatial_dropout_2d.html">layer_spatial_dropout_2d</a>()</code> <br><code><a href="layer_spatial_dropout_3d.html">layer_spatial_dropout_3d</a>()</code> <br><code><a href="layer_spectral_normalization.html">layer_spectral_normalization</a>()</code> <br><code><a href="layer_stft_spectrogram.html">layer_stft_spectrogram</a>()</code> <br><code><a href="layer_string_lookup.html">layer_string_lookup</a>()</code> <br><code><a href="layer_subtract.html">layer_subtract</a>()</code> <br><code><a href="layer_text_vectorization.html">layer_text_vectorization</a>()</code> <br><code><a href="layer_tfsm.html">layer_tfsm</a>()</code> <br><code><a href="layer_time_distributed.html">layer_time_distributed</a>()</code> <br><code><a href="layer_torch_module_wrapper.html">layer_torch_module_wrapper</a>()</code> <br><code><a href="layer_unit_normalization.html">layer_unit_normalization</a>()</code> <br><code><a href="layer_upsampling_1d.html">layer_upsampling_1d</a>()</code> <br><code><a href="layer_upsampling_2d.html">layer_upsampling_2d</a>()</code> <br><code><a href="layer_upsampling_3d.html">layer_upsampling_3d</a>()</code> <br><code><a href="layer_zero_padding_1d.html">layer_zero_padding_1d</a>()</code> <br><code><a href="layer_zero_padding_2d.html">layer_zero_padding_2d</a>()</code> <br><code><a href="layer_zero_padding_3d.html">layer_zero_padding_3d</a>()</code> <br><code><a href="rnn_cell_gru.html">rnn_cell_gru</a>()</code> <br><code><a href="rnn_cell_lstm.html">rnn_cell_lstm</a>()</code> <br><code><a href="rnn_cell_simple.html">rnn_cell_simple</a>()</code> <br><code><a href="rnn_cells_stack.html">rnn_cells_stack</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -145,7 +145,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/layer_random_invert.html b/docs/dev/reference/layer_random_invert.html
index c26d665e1..f7109aa2e 100644
--- a/docs/dev/reference/layer_random_invert.html
+++ b/docs/dev/reference/layer_random_invert.html
@@ -16,7 +16,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -127,9 +127,9 @@ <h2 id="arguments">Arguments<a class="anchor" aria-label="anchor" href="#argumen
 </dl></div>
     <div class="section level2">
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
-    <div class="dont-index"><p>Other image preprocessing layers: <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br></p>
-<p>Other preprocessing layers: <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_category_encoding.html">layer_category_encoding</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_discretization.html">layer_discretization</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_feature_space.html">layer_feature_space</a>()</code> <br><code><a href="layer_hashed_crossing.html">layer_hashed_crossing</a>()</code> <br><code><a href="layer_hashing.html">layer_hashing</a>()</code> <br><code><a href="layer_integer_lookup.html">layer_integer_lookup</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_mel_spectrogram.html">layer_mel_spectrogram</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_normalization.html">layer_normalization</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_brightness.html">layer_random_brightness</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_contrast.html">layer_random_contrast</a>()</code> <br><code><a href="layer_random_crop.html">layer_random_crop</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_flip.html">layer_random_flip</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_rotation.html">layer_random_rotation</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_random_translation.html">layer_random_translation</a>()</code> <br><code><a href="layer_random_zoom.html">layer_random_zoom</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br><code><a href="layer_stft_spectrogram.html">layer_stft_spectrogram</a>()</code> <br><code><a href="layer_string_lookup.html">layer_string_lookup</a>()</code> <br><code><a href="layer_text_vectorization.html">layer_text_vectorization</a>()</code> <br></p>
-<p>Other layers: <br><code><a href="Layer.html">Layer</a>()</code> <br><code><a href="layer_activation.html">layer_activation</a>()</code> <br><code><a href="layer_activation_elu.html">layer_activation_elu</a>()</code> <br><code><a href="layer_activation_leaky_relu.html">layer_activation_leaky_relu</a>()</code> <br><code><a href="layer_activation_parametric_relu.html">layer_activation_parametric_relu</a>()</code> <br><code><a href="layer_activation_relu.html">layer_activation_relu</a>()</code> <br><code><a href="layer_activation_softmax.html">layer_activation_softmax</a>()</code> <br><code><a href="layer_activity_regularization.html">layer_activity_regularization</a>()</code> <br><code><a href="layer_add.html">layer_add</a>()</code> <br><code><a href="layer_additive_attention.html">layer_additive_attention</a>()</code> <br><code><a href="layer_alpha_dropout.html">layer_alpha_dropout</a>()</code> <br><code><a href="layer_attention.html">layer_attention</a>()</code> <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_average.html">layer_average</a>()</code> <br><code><a href="layer_average_pooling_1d.html">layer_average_pooling_1d</a>()</code> <br><code><a href="layer_average_pooling_2d.html">layer_average_pooling_2d</a>()</code> <br><code><a href="layer_average_pooling_3d.html">layer_average_pooling_3d</a>()</code> <br><code><a href="layer_batch_normalization.html">layer_batch_normalization</a>()</code> <br><code><a href="layer_bidirectional.html">layer_bidirectional</a>()</code> <br><code><a href="layer_category_encoding.html">layer_category_encoding</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_concatenate.html">layer_concatenate</a>()</code> <br><code><a href="layer_conv_1d.html">layer_conv_1d</a>()</code> <br><code><a href="layer_conv_1d_transpose.html">layer_conv_1d_transpose</a>()</code> <br><code><a href="layer_conv_2d.html">layer_conv_2d</a>()</code> <br><code><a href="layer_conv_2d_transpose.html">layer_conv_2d_transpose</a>()</code> <br><code><a href="layer_conv_3d.html">layer_conv_3d</a>()</code> <br><code><a href="layer_conv_3d_transpose.html">layer_conv_3d_transpose</a>()</code> <br><code><a href="layer_conv_lstm_1d.html">layer_conv_lstm_1d</a>()</code> <br><code><a href="layer_conv_lstm_2d.html">layer_conv_lstm_2d</a>()</code> <br><code><a href="layer_conv_lstm_3d.html">layer_conv_lstm_3d</a>()</code> <br><code><a href="layer_cropping_1d.html">layer_cropping_1d</a>()</code> <br><code><a href="layer_cropping_2d.html">layer_cropping_2d</a>()</code> <br><code><a href="layer_cropping_3d.html">layer_cropping_3d</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_dense.html">layer_dense</a>()</code> <br><code><a href="layer_depthwise_conv_1d.html">layer_depthwise_conv_1d</a>()</code> <br><code><a href="layer_depthwise_conv_2d.html">layer_depthwise_conv_2d</a>()</code> <br><code><a href="layer_discretization.html">layer_discretization</a>()</code> <br><code><a href="layer_dot.html">layer_dot</a>()</code> <br><code><a href="layer_dropout.html">layer_dropout</a>()</code> <br><code><a href="layer_einsum_dense.html">layer_einsum_dense</a>()</code> <br><code><a href="layer_embedding.html">layer_embedding</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_feature_space.html">layer_feature_space</a>()</code> <br><code><a href="layer_flatten.html">layer_flatten</a>()</code> <br><code><a href="layer_flax_module_wrapper.html">layer_flax_module_wrapper</a>()</code> <br><code><a href="layer_gaussian_dropout.html">layer_gaussian_dropout</a>()</code> <br><code><a href="layer_gaussian_noise.html">layer_gaussian_noise</a>()</code> <br><code><a href="layer_global_average_pooling_1d.html">layer_global_average_pooling_1d</a>()</code> <br><code><a href="layer_global_average_pooling_2d.html">layer_global_average_pooling_2d</a>()</code> <br><code><a href="layer_global_average_pooling_3d.html">layer_global_average_pooling_3d</a>()</code> <br><code><a href="layer_global_max_pooling_1d.html">layer_global_max_pooling_1d</a>()</code> <br><code><a href="layer_global_max_pooling_2d.html">layer_global_max_pooling_2d</a>()</code> <br><code><a href="layer_global_max_pooling_3d.html">layer_global_max_pooling_3d</a>()</code> <br><code><a href="layer_group_normalization.html">layer_group_normalization</a>()</code> <br><code><a href="layer_group_query_attention.html">layer_group_query_attention</a>()</code> <br><code><a href="layer_gru.html">layer_gru</a>()</code> <br><code><a href="layer_hashed_crossing.html">layer_hashed_crossing</a>()</code> <br><code><a href="layer_hashing.html">layer_hashing</a>()</code> <br><code><a href="layer_identity.html">layer_identity</a>()</code> <br><code><a href="layer_integer_lookup.html">layer_integer_lookup</a>()</code> <br><code><a href="layer_jax_model_wrapper.html">layer_jax_model_wrapper</a>()</code> <br><code><a href="layer_lambda.html">layer_lambda</a>()</code> <br><code><a href="layer_layer_normalization.html">layer_layer_normalization</a>()</code> <br><code><a href="layer_lstm.html">layer_lstm</a>()</code> <br><code><a href="layer_masking.html">layer_masking</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_max_pooling_1d.html">layer_max_pooling_1d</a>()</code> <br><code><a href="layer_max_pooling_2d.html">layer_max_pooling_2d</a>()</code> <br><code><a href="layer_max_pooling_3d.html">layer_max_pooling_3d</a>()</code> <br><code><a href="layer_maximum.html">layer_maximum</a>()</code> <br><code><a href="layer_mel_spectrogram.html">layer_mel_spectrogram</a>()</code> <br><code><a href="layer_minimum.html">layer_minimum</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_multi_head_attention.html">layer_multi_head_attention</a>()</code> <br><code><a href="layer_multiply.html">layer_multiply</a>()</code> <br><code><a href="layer_normalization.html">layer_normalization</a>()</code> <br><code><a href="layer_permute.html">layer_permute</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_brightness.html">layer_random_brightness</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_contrast.html">layer_random_contrast</a>()</code> <br><code><a href="layer_random_crop.html">layer_random_crop</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_flip.html">layer_random_flip</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_rotation.html">layer_random_rotation</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_random_translation.html">layer_random_translation</a>()</code> <br><code><a href="layer_random_zoom.html">layer_random_zoom</a>()</code> <br><code><a href="layer_repeat_vector.html">layer_repeat_vector</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_reshape.html">layer_reshape</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_rms_normalization.html">layer_rms_normalization</a>()</code> <br><code><a href="layer_rnn.html">layer_rnn</a>()</code> <br><code><a href="layer_separable_conv_1d.html">layer_separable_conv_1d</a>()</code> <br><code><a href="layer_separable_conv_2d.html">layer_separable_conv_2d</a>()</code> <br><code><a href="layer_simple_rnn.html">layer_simple_rnn</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br><code><a href="layer_spatial_dropout_1d.html">layer_spatial_dropout_1d</a>()</code> <br><code><a href="layer_spatial_dropout_2d.html">layer_spatial_dropout_2d</a>()</code> <br><code><a href="layer_spatial_dropout_3d.html">layer_spatial_dropout_3d</a>()</code> <br><code><a href="layer_spectral_normalization.html">layer_spectral_normalization</a>()</code> <br><code><a href="layer_stft_spectrogram.html">layer_stft_spectrogram</a>()</code> <br><code><a href="layer_string_lookup.html">layer_string_lookup</a>()</code> <br><code><a href="layer_subtract.html">layer_subtract</a>()</code> <br><code><a href="layer_text_vectorization.html">layer_text_vectorization</a>()</code> <br><code><a href="layer_tfsm.html">layer_tfsm</a>()</code> <br><code><a href="layer_time_distributed.html">layer_time_distributed</a>()</code> <br><code><a href="layer_torch_module_wrapper.html">layer_torch_module_wrapper</a>()</code> <br><code><a href="layer_unit_normalization.html">layer_unit_normalization</a>()</code> <br><code><a href="layer_upsampling_1d.html">layer_upsampling_1d</a>()</code> <br><code><a href="layer_upsampling_2d.html">layer_upsampling_2d</a>()</code> <br><code><a href="layer_upsampling_3d.html">layer_upsampling_3d</a>()</code> <br><code><a href="layer_zero_padding_1d.html">layer_zero_padding_1d</a>()</code> <br><code><a href="layer_zero_padding_2d.html">layer_zero_padding_2d</a>()</code> <br><code><a href="layer_zero_padding_3d.html">layer_zero_padding_3d</a>()</code> <br><code><a href="rnn_cell_gru.html">rnn_cell_gru</a>()</code> <br><code><a href="rnn_cell_lstm.html">rnn_cell_lstm</a>()</code> <br><code><a href="rnn_cell_simple.html">rnn_cell_simple</a>()</code> <br><code><a href="rnn_cells_stack.html">rnn_cells_stack</a>()</code> <br></p></div>
+    <div class="dont-index"><p>Other image preprocessing layers: <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_elastic_transform.html">layer_random_elastic_transform</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br></p>
+<p>Other preprocessing layers: <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_category_encoding.html">layer_category_encoding</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_discretization.html">layer_discretization</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_feature_space.html">layer_feature_space</a>()</code> <br><code><a href="layer_hashed_crossing.html">layer_hashed_crossing</a>()</code> <br><code><a href="layer_hashing.html">layer_hashing</a>()</code> <br><code><a href="layer_integer_lookup.html">layer_integer_lookup</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_mel_spectrogram.html">layer_mel_spectrogram</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_normalization.html">layer_normalization</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_brightness.html">layer_random_brightness</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_contrast.html">layer_random_contrast</a>()</code> <br><code><a href="layer_random_crop.html">layer_random_crop</a>()</code> <br><code><a href="layer_random_elastic_transform.html">layer_random_elastic_transform</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_flip.html">layer_random_flip</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_rotation.html">layer_random_rotation</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_random_translation.html">layer_random_translation</a>()</code> <br><code><a href="layer_random_zoom.html">layer_random_zoom</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br><code><a href="layer_stft_spectrogram.html">layer_stft_spectrogram</a>()</code> <br><code><a href="layer_string_lookup.html">layer_string_lookup</a>()</code> <br><code><a href="layer_text_vectorization.html">layer_text_vectorization</a>()</code> <br></p>
+<p>Other layers: <br><code><a href="Layer.html">Layer</a>()</code> <br><code><a href="layer_activation.html">layer_activation</a>()</code> <br><code><a href="layer_activation_elu.html">layer_activation_elu</a>()</code> <br><code><a href="layer_activation_leaky_relu.html">layer_activation_leaky_relu</a>()</code> <br><code><a href="layer_activation_parametric_relu.html">layer_activation_parametric_relu</a>()</code> <br><code><a href="layer_activation_relu.html">layer_activation_relu</a>()</code> <br><code><a href="layer_activation_softmax.html">layer_activation_softmax</a>()</code> <br><code><a href="layer_activity_regularization.html">layer_activity_regularization</a>()</code> <br><code><a href="layer_add.html">layer_add</a>()</code> <br><code><a href="layer_additive_attention.html">layer_additive_attention</a>()</code> <br><code><a href="layer_alpha_dropout.html">layer_alpha_dropout</a>()</code> <br><code><a href="layer_attention.html">layer_attention</a>()</code> <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_average.html">layer_average</a>()</code> <br><code><a href="layer_average_pooling_1d.html">layer_average_pooling_1d</a>()</code> <br><code><a href="layer_average_pooling_2d.html">layer_average_pooling_2d</a>()</code> <br><code><a href="layer_average_pooling_3d.html">layer_average_pooling_3d</a>()</code> <br><code><a href="layer_batch_normalization.html">layer_batch_normalization</a>()</code> <br><code><a href="layer_bidirectional.html">layer_bidirectional</a>()</code> <br><code><a href="layer_category_encoding.html">layer_category_encoding</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_concatenate.html">layer_concatenate</a>()</code> <br><code><a href="layer_conv_1d.html">layer_conv_1d</a>()</code> <br><code><a href="layer_conv_1d_transpose.html">layer_conv_1d_transpose</a>()</code> <br><code><a href="layer_conv_2d.html">layer_conv_2d</a>()</code> <br><code><a href="layer_conv_2d_transpose.html">layer_conv_2d_transpose</a>()</code> <br><code><a href="layer_conv_3d.html">layer_conv_3d</a>()</code> <br><code><a href="layer_conv_3d_transpose.html">layer_conv_3d_transpose</a>()</code> <br><code><a href="layer_conv_lstm_1d.html">layer_conv_lstm_1d</a>()</code> <br><code><a href="layer_conv_lstm_2d.html">layer_conv_lstm_2d</a>()</code> <br><code><a href="layer_conv_lstm_3d.html">layer_conv_lstm_3d</a>()</code> <br><code><a href="layer_cropping_1d.html">layer_cropping_1d</a>()</code> <br><code><a href="layer_cropping_2d.html">layer_cropping_2d</a>()</code> <br><code><a href="layer_cropping_3d.html">layer_cropping_3d</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_dense.html">layer_dense</a>()</code> <br><code><a href="layer_depthwise_conv_1d.html">layer_depthwise_conv_1d</a>()</code> <br><code><a href="layer_depthwise_conv_2d.html">layer_depthwise_conv_2d</a>()</code> <br><code><a href="layer_discretization.html">layer_discretization</a>()</code> <br><code><a href="layer_dot.html">layer_dot</a>()</code> <br><code><a href="layer_dropout.html">layer_dropout</a>()</code> <br><code><a href="layer_einsum_dense.html">layer_einsum_dense</a>()</code> <br><code><a href="layer_embedding.html">layer_embedding</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_feature_space.html">layer_feature_space</a>()</code> <br><code><a href="layer_flatten.html">layer_flatten</a>()</code> <br><code><a href="layer_flax_module_wrapper.html">layer_flax_module_wrapper</a>()</code> <br><code><a href="layer_gaussian_dropout.html">layer_gaussian_dropout</a>()</code> <br><code><a href="layer_gaussian_noise.html">layer_gaussian_noise</a>()</code> <br><code><a href="layer_global_average_pooling_1d.html">layer_global_average_pooling_1d</a>()</code> <br><code><a href="layer_global_average_pooling_2d.html">layer_global_average_pooling_2d</a>()</code> <br><code><a href="layer_global_average_pooling_3d.html">layer_global_average_pooling_3d</a>()</code> <br><code><a href="layer_global_max_pooling_1d.html">layer_global_max_pooling_1d</a>()</code> <br><code><a href="layer_global_max_pooling_2d.html">layer_global_max_pooling_2d</a>()</code> <br><code><a href="layer_global_max_pooling_3d.html">layer_global_max_pooling_3d</a>()</code> <br><code><a href="layer_group_normalization.html">layer_group_normalization</a>()</code> <br><code><a href="layer_group_query_attention.html">layer_group_query_attention</a>()</code> <br><code><a href="layer_gru.html">layer_gru</a>()</code> <br><code><a href="layer_hashed_crossing.html">layer_hashed_crossing</a>()</code> <br><code><a href="layer_hashing.html">layer_hashing</a>()</code> <br><code><a href="layer_identity.html">layer_identity</a>()</code> <br><code><a href="layer_integer_lookup.html">layer_integer_lookup</a>()</code> <br><code><a href="layer_jax_model_wrapper.html">layer_jax_model_wrapper</a>()</code> <br><code><a href="layer_lambda.html">layer_lambda</a>()</code> <br><code><a href="layer_layer_normalization.html">layer_layer_normalization</a>()</code> <br><code><a href="layer_lstm.html">layer_lstm</a>()</code> <br><code><a href="layer_masking.html">layer_masking</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_max_pooling_1d.html">layer_max_pooling_1d</a>()</code> <br><code><a href="layer_max_pooling_2d.html">layer_max_pooling_2d</a>()</code> <br><code><a href="layer_max_pooling_3d.html">layer_max_pooling_3d</a>()</code> <br><code><a href="layer_maximum.html">layer_maximum</a>()</code> <br><code><a href="layer_mel_spectrogram.html">layer_mel_spectrogram</a>()</code> <br><code><a href="layer_minimum.html">layer_minimum</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_multi_head_attention.html">layer_multi_head_attention</a>()</code> <br><code><a href="layer_multiply.html">layer_multiply</a>()</code> <br><code><a href="layer_normalization.html">layer_normalization</a>()</code> <br><code><a href="layer_permute.html">layer_permute</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_brightness.html">layer_random_brightness</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_contrast.html">layer_random_contrast</a>()</code> <br><code><a href="layer_random_crop.html">layer_random_crop</a>()</code> <br><code><a href="layer_random_elastic_transform.html">layer_random_elastic_transform</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_flip.html">layer_random_flip</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_rotation.html">layer_random_rotation</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_random_translation.html">layer_random_translation</a>()</code> <br><code><a href="layer_random_zoom.html">layer_random_zoom</a>()</code> <br><code><a href="layer_repeat_vector.html">layer_repeat_vector</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_reshape.html">layer_reshape</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_rms_normalization.html">layer_rms_normalization</a>()</code> <br><code><a href="layer_rnn.html">layer_rnn</a>()</code> <br><code><a href="layer_separable_conv_1d.html">layer_separable_conv_1d</a>()</code> <br><code><a href="layer_separable_conv_2d.html">layer_separable_conv_2d</a>()</code> <br><code><a href="layer_simple_rnn.html">layer_simple_rnn</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br><code><a href="layer_spatial_dropout_1d.html">layer_spatial_dropout_1d</a>()</code> <br><code><a href="layer_spatial_dropout_2d.html">layer_spatial_dropout_2d</a>()</code> <br><code><a href="layer_spatial_dropout_3d.html">layer_spatial_dropout_3d</a>()</code> <br><code><a href="layer_spectral_normalization.html">layer_spectral_normalization</a>()</code> <br><code><a href="layer_stft_spectrogram.html">layer_stft_spectrogram</a>()</code> <br><code><a href="layer_string_lookup.html">layer_string_lookup</a>()</code> <br><code><a href="layer_subtract.html">layer_subtract</a>()</code> <br><code><a href="layer_text_vectorization.html">layer_text_vectorization</a>()</code> <br><code><a href="layer_tfsm.html">layer_tfsm</a>()</code> <br><code><a href="layer_time_distributed.html">layer_time_distributed</a>()</code> <br><code><a href="layer_torch_module_wrapper.html">layer_torch_module_wrapper</a>()</code> <br><code><a href="layer_unit_normalization.html">layer_unit_normalization</a>()</code> <br><code><a href="layer_upsampling_1d.html">layer_upsampling_1d</a>()</code> <br><code><a href="layer_upsampling_2d.html">layer_upsampling_2d</a>()</code> <br><code><a href="layer_upsampling_3d.html">layer_upsampling_3d</a>()</code> <br><code><a href="layer_zero_padding_1d.html">layer_zero_padding_1d</a>()</code> <br><code><a href="layer_zero_padding_2d.html">layer_zero_padding_2d</a>()</code> <br><code><a href="layer_zero_padding_3d.html">layer_zero_padding_3d</a>()</code> <br><code><a href="rnn_cell_gru.html">rnn_cell_gru</a>()</code> <br><code><a href="rnn_cell_lstm.html">rnn_cell_lstm</a>()</code> <br><code><a href="rnn_cell_simple.html">rnn_cell_simple</a>()</code> <br><code><a href="rnn_cells_stack.html">rnn_cells_stack</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -141,7 +141,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/layer_random_perspective.html b/docs/dev/reference/layer_random_perspective.html
index 9d75544a5..696a5b94d 100644
--- a/docs/dev/reference/layer_random_perspective.html
+++ b/docs/dev/reference/layer_random_perspective.html
@@ -12,7 +12,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -135,9 +135,9 @@ <h2 id="arguments">Arguments<a class="anchor" aria-label="anchor" href="#argumen
 </dl></div>
     <div class="section level2">
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
-    <div class="dont-index"><p>Other image preprocessing layers: <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br></p>
-<p>Other preprocessing layers: <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_category_encoding.html">layer_category_encoding</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_discretization.html">layer_discretization</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_feature_space.html">layer_feature_space</a>()</code> <br><code><a href="layer_hashed_crossing.html">layer_hashed_crossing</a>()</code> <br><code><a href="layer_hashing.html">layer_hashing</a>()</code> <br><code><a href="layer_integer_lookup.html">layer_integer_lookup</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_mel_spectrogram.html">layer_mel_spectrogram</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_normalization.html">layer_normalization</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_brightness.html">layer_random_brightness</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_contrast.html">layer_random_contrast</a>()</code> <br><code><a href="layer_random_crop.html">layer_random_crop</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_flip.html">layer_random_flip</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_rotation.html">layer_random_rotation</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_random_translation.html">layer_random_translation</a>()</code> <br><code><a href="layer_random_zoom.html">layer_random_zoom</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br><code><a href="layer_stft_spectrogram.html">layer_stft_spectrogram</a>()</code> <br><code><a href="layer_string_lookup.html">layer_string_lookup</a>()</code> <br><code><a href="layer_text_vectorization.html">layer_text_vectorization</a>()</code> <br></p>
-<p>Other layers: <br><code><a href="Layer.html">Layer</a>()</code> <br><code><a href="layer_activation.html">layer_activation</a>()</code> <br><code><a href="layer_activation_elu.html">layer_activation_elu</a>()</code> <br><code><a href="layer_activation_leaky_relu.html">layer_activation_leaky_relu</a>()</code> <br><code><a href="layer_activation_parametric_relu.html">layer_activation_parametric_relu</a>()</code> <br><code><a href="layer_activation_relu.html">layer_activation_relu</a>()</code> <br><code><a href="layer_activation_softmax.html">layer_activation_softmax</a>()</code> <br><code><a href="layer_activity_regularization.html">layer_activity_regularization</a>()</code> <br><code><a href="layer_add.html">layer_add</a>()</code> <br><code><a href="layer_additive_attention.html">layer_additive_attention</a>()</code> <br><code><a href="layer_alpha_dropout.html">layer_alpha_dropout</a>()</code> <br><code><a href="layer_attention.html">layer_attention</a>()</code> <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_average.html">layer_average</a>()</code> <br><code><a href="layer_average_pooling_1d.html">layer_average_pooling_1d</a>()</code> <br><code><a href="layer_average_pooling_2d.html">layer_average_pooling_2d</a>()</code> <br><code><a href="layer_average_pooling_3d.html">layer_average_pooling_3d</a>()</code> <br><code><a href="layer_batch_normalization.html">layer_batch_normalization</a>()</code> <br><code><a href="layer_bidirectional.html">layer_bidirectional</a>()</code> <br><code><a href="layer_category_encoding.html">layer_category_encoding</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_concatenate.html">layer_concatenate</a>()</code> <br><code><a href="layer_conv_1d.html">layer_conv_1d</a>()</code> <br><code><a href="layer_conv_1d_transpose.html">layer_conv_1d_transpose</a>()</code> <br><code><a href="layer_conv_2d.html">layer_conv_2d</a>()</code> <br><code><a href="layer_conv_2d_transpose.html">layer_conv_2d_transpose</a>()</code> <br><code><a href="layer_conv_3d.html">layer_conv_3d</a>()</code> <br><code><a href="layer_conv_3d_transpose.html">layer_conv_3d_transpose</a>()</code> <br><code><a href="layer_conv_lstm_1d.html">layer_conv_lstm_1d</a>()</code> <br><code><a href="layer_conv_lstm_2d.html">layer_conv_lstm_2d</a>()</code> <br><code><a href="layer_conv_lstm_3d.html">layer_conv_lstm_3d</a>()</code> <br><code><a href="layer_cropping_1d.html">layer_cropping_1d</a>()</code> <br><code><a href="layer_cropping_2d.html">layer_cropping_2d</a>()</code> <br><code><a href="layer_cropping_3d.html">layer_cropping_3d</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_dense.html">layer_dense</a>()</code> <br><code><a href="layer_depthwise_conv_1d.html">layer_depthwise_conv_1d</a>()</code> <br><code><a href="layer_depthwise_conv_2d.html">layer_depthwise_conv_2d</a>()</code> <br><code><a href="layer_discretization.html">layer_discretization</a>()</code> <br><code><a href="layer_dot.html">layer_dot</a>()</code> <br><code><a href="layer_dropout.html">layer_dropout</a>()</code> <br><code><a href="layer_einsum_dense.html">layer_einsum_dense</a>()</code> <br><code><a href="layer_embedding.html">layer_embedding</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_feature_space.html">layer_feature_space</a>()</code> <br><code><a href="layer_flatten.html">layer_flatten</a>()</code> <br><code><a href="layer_flax_module_wrapper.html">layer_flax_module_wrapper</a>()</code> <br><code><a href="layer_gaussian_dropout.html">layer_gaussian_dropout</a>()</code> <br><code><a href="layer_gaussian_noise.html">layer_gaussian_noise</a>()</code> <br><code><a href="layer_global_average_pooling_1d.html">layer_global_average_pooling_1d</a>()</code> <br><code><a href="layer_global_average_pooling_2d.html">layer_global_average_pooling_2d</a>()</code> <br><code><a href="layer_global_average_pooling_3d.html">layer_global_average_pooling_3d</a>()</code> <br><code><a href="layer_global_max_pooling_1d.html">layer_global_max_pooling_1d</a>()</code> <br><code><a href="layer_global_max_pooling_2d.html">layer_global_max_pooling_2d</a>()</code> <br><code><a href="layer_global_max_pooling_3d.html">layer_global_max_pooling_3d</a>()</code> <br><code><a href="layer_group_normalization.html">layer_group_normalization</a>()</code> <br><code><a href="layer_group_query_attention.html">layer_group_query_attention</a>()</code> <br><code><a href="layer_gru.html">layer_gru</a>()</code> <br><code><a href="layer_hashed_crossing.html">layer_hashed_crossing</a>()</code> <br><code><a href="layer_hashing.html">layer_hashing</a>()</code> <br><code><a href="layer_identity.html">layer_identity</a>()</code> <br><code><a href="layer_integer_lookup.html">layer_integer_lookup</a>()</code> <br><code><a href="layer_jax_model_wrapper.html">layer_jax_model_wrapper</a>()</code> <br><code><a href="layer_lambda.html">layer_lambda</a>()</code> <br><code><a href="layer_layer_normalization.html">layer_layer_normalization</a>()</code> <br><code><a href="layer_lstm.html">layer_lstm</a>()</code> <br><code><a href="layer_masking.html">layer_masking</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_max_pooling_1d.html">layer_max_pooling_1d</a>()</code> <br><code><a href="layer_max_pooling_2d.html">layer_max_pooling_2d</a>()</code> <br><code><a href="layer_max_pooling_3d.html">layer_max_pooling_3d</a>()</code> <br><code><a href="layer_maximum.html">layer_maximum</a>()</code> <br><code><a href="layer_mel_spectrogram.html">layer_mel_spectrogram</a>()</code> <br><code><a href="layer_minimum.html">layer_minimum</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_multi_head_attention.html">layer_multi_head_attention</a>()</code> <br><code><a href="layer_multiply.html">layer_multiply</a>()</code> <br><code><a href="layer_normalization.html">layer_normalization</a>()</code> <br><code><a href="layer_permute.html">layer_permute</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_brightness.html">layer_random_brightness</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_contrast.html">layer_random_contrast</a>()</code> <br><code><a href="layer_random_crop.html">layer_random_crop</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_flip.html">layer_random_flip</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_rotation.html">layer_random_rotation</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_random_translation.html">layer_random_translation</a>()</code> <br><code><a href="layer_random_zoom.html">layer_random_zoom</a>()</code> <br><code><a href="layer_repeat_vector.html">layer_repeat_vector</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_reshape.html">layer_reshape</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_rms_normalization.html">layer_rms_normalization</a>()</code> <br><code><a href="layer_rnn.html">layer_rnn</a>()</code> <br><code><a href="layer_separable_conv_1d.html">layer_separable_conv_1d</a>()</code> <br><code><a href="layer_separable_conv_2d.html">layer_separable_conv_2d</a>()</code> <br><code><a href="layer_simple_rnn.html">layer_simple_rnn</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br><code><a href="layer_spatial_dropout_1d.html">layer_spatial_dropout_1d</a>()</code> <br><code><a href="layer_spatial_dropout_2d.html">layer_spatial_dropout_2d</a>()</code> <br><code><a href="layer_spatial_dropout_3d.html">layer_spatial_dropout_3d</a>()</code> <br><code><a href="layer_spectral_normalization.html">layer_spectral_normalization</a>()</code> <br><code><a href="layer_stft_spectrogram.html">layer_stft_spectrogram</a>()</code> <br><code><a href="layer_string_lookup.html">layer_string_lookup</a>()</code> <br><code><a href="layer_subtract.html">layer_subtract</a>()</code> <br><code><a href="layer_text_vectorization.html">layer_text_vectorization</a>()</code> <br><code><a href="layer_tfsm.html">layer_tfsm</a>()</code> <br><code><a href="layer_time_distributed.html">layer_time_distributed</a>()</code> <br><code><a href="layer_torch_module_wrapper.html">layer_torch_module_wrapper</a>()</code> <br><code><a href="layer_unit_normalization.html">layer_unit_normalization</a>()</code> <br><code><a href="layer_upsampling_1d.html">layer_upsampling_1d</a>()</code> <br><code><a href="layer_upsampling_2d.html">layer_upsampling_2d</a>()</code> <br><code><a href="layer_upsampling_3d.html">layer_upsampling_3d</a>()</code> <br><code><a href="layer_zero_padding_1d.html">layer_zero_padding_1d</a>()</code> <br><code><a href="layer_zero_padding_2d.html">layer_zero_padding_2d</a>()</code> <br><code><a href="layer_zero_padding_3d.html">layer_zero_padding_3d</a>()</code> <br><code><a href="rnn_cell_gru.html">rnn_cell_gru</a>()</code> <br><code><a href="rnn_cell_lstm.html">rnn_cell_lstm</a>()</code> <br><code><a href="rnn_cell_simple.html">rnn_cell_simple</a>()</code> <br><code><a href="rnn_cells_stack.html">rnn_cells_stack</a>()</code> <br></p></div>
+    <div class="dont-index"><p>Other image preprocessing layers: <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_elastic_transform.html">layer_random_elastic_transform</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br></p>
+<p>Other preprocessing layers: <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_category_encoding.html">layer_category_encoding</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_discretization.html">layer_discretization</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_feature_space.html">layer_feature_space</a>()</code> <br><code><a href="layer_hashed_crossing.html">layer_hashed_crossing</a>()</code> <br><code><a href="layer_hashing.html">layer_hashing</a>()</code> <br><code><a href="layer_integer_lookup.html">layer_integer_lookup</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_mel_spectrogram.html">layer_mel_spectrogram</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_normalization.html">layer_normalization</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_brightness.html">layer_random_brightness</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_contrast.html">layer_random_contrast</a>()</code> <br><code><a href="layer_random_crop.html">layer_random_crop</a>()</code> <br><code><a href="layer_random_elastic_transform.html">layer_random_elastic_transform</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_flip.html">layer_random_flip</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_rotation.html">layer_random_rotation</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_random_translation.html">layer_random_translation</a>()</code> <br><code><a href="layer_random_zoom.html">layer_random_zoom</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br><code><a href="layer_stft_spectrogram.html">layer_stft_spectrogram</a>()</code> <br><code><a href="layer_string_lookup.html">layer_string_lookup</a>()</code> <br><code><a href="layer_text_vectorization.html">layer_text_vectorization</a>()</code> <br></p>
+<p>Other layers: <br><code><a href="Layer.html">Layer</a>()</code> <br><code><a href="layer_activation.html">layer_activation</a>()</code> <br><code><a href="layer_activation_elu.html">layer_activation_elu</a>()</code> <br><code><a href="layer_activation_leaky_relu.html">layer_activation_leaky_relu</a>()</code> <br><code><a href="layer_activation_parametric_relu.html">layer_activation_parametric_relu</a>()</code> <br><code><a href="layer_activation_relu.html">layer_activation_relu</a>()</code> <br><code><a href="layer_activation_softmax.html">layer_activation_softmax</a>()</code> <br><code><a href="layer_activity_regularization.html">layer_activity_regularization</a>()</code> <br><code><a href="layer_add.html">layer_add</a>()</code> <br><code><a href="layer_additive_attention.html">layer_additive_attention</a>()</code> <br><code><a href="layer_alpha_dropout.html">layer_alpha_dropout</a>()</code> <br><code><a href="layer_attention.html">layer_attention</a>()</code> <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_average.html">layer_average</a>()</code> <br><code><a href="layer_average_pooling_1d.html">layer_average_pooling_1d</a>()</code> <br><code><a href="layer_average_pooling_2d.html">layer_average_pooling_2d</a>()</code> <br><code><a href="layer_average_pooling_3d.html">layer_average_pooling_3d</a>()</code> <br><code><a href="layer_batch_normalization.html">layer_batch_normalization</a>()</code> <br><code><a href="layer_bidirectional.html">layer_bidirectional</a>()</code> <br><code><a href="layer_category_encoding.html">layer_category_encoding</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_concatenate.html">layer_concatenate</a>()</code> <br><code><a href="layer_conv_1d.html">layer_conv_1d</a>()</code> <br><code><a href="layer_conv_1d_transpose.html">layer_conv_1d_transpose</a>()</code> <br><code><a href="layer_conv_2d.html">layer_conv_2d</a>()</code> <br><code><a href="layer_conv_2d_transpose.html">layer_conv_2d_transpose</a>()</code> <br><code><a href="layer_conv_3d.html">layer_conv_3d</a>()</code> <br><code><a href="layer_conv_3d_transpose.html">layer_conv_3d_transpose</a>()</code> <br><code><a href="layer_conv_lstm_1d.html">layer_conv_lstm_1d</a>()</code> <br><code><a href="layer_conv_lstm_2d.html">layer_conv_lstm_2d</a>()</code> <br><code><a href="layer_conv_lstm_3d.html">layer_conv_lstm_3d</a>()</code> <br><code><a href="layer_cropping_1d.html">layer_cropping_1d</a>()</code> <br><code><a href="layer_cropping_2d.html">layer_cropping_2d</a>()</code> <br><code><a href="layer_cropping_3d.html">layer_cropping_3d</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_dense.html">layer_dense</a>()</code> <br><code><a href="layer_depthwise_conv_1d.html">layer_depthwise_conv_1d</a>()</code> <br><code><a href="layer_depthwise_conv_2d.html">layer_depthwise_conv_2d</a>()</code> <br><code><a href="layer_discretization.html">layer_discretization</a>()</code> <br><code><a href="layer_dot.html">layer_dot</a>()</code> <br><code><a href="layer_dropout.html">layer_dropout</a>()</code> <br><code><a href="layer_einsum_dense.html">layer_einsum_dense</a>()</code> <br><code><a href="layer_embedding.html">layer_embedding</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_feature_space.html">layer_feature_space</a>()</code> <br><code><a href="layer_flatten.html">layer_flatten</a>()</code> <br><code><a href="layer_flax_module_wrapper.html">layer_flax_module_wrapper</a>()</code> <br><code><a href="layer_gaussian_dropout.html">layer_gaussian_dropout</a>()</code> <br><code><a href="layer_gaussian_noise.html">layer_gaussian_noise</a>()</code> <br><code><a href="layer_global_average_pooling_1d.html">layer_global_average_pooling_1d</a>()</code> <br><code><a href="layer_global_average_pooling_2d.html">layer_global_average_pooling_2d</a>()</code> <br><code><a href="layer_global_average_pooling_3d.html">layer_global_average_pooling_3d</a>()</code> <br><code><a href="layer_global_max_pooling_1d.html">layer_global_max_pooling_1d</a>()</code> <br><code><a href="layer_global_max_pooling_2d.html">layer_global_max_pooling_2d</a>()</code> <br><code><a href="layer_global_max_pooling_3d.html">layer_global_max_pooling_3d</a>()</code> <br><code><a href="layer_group_normalization.html">layer_group_normalization</a>()</code> <br><code><a href="layer_group_query_attention.html">layer_group_query_attention</a>()</code> <br><code><a href="layer_gru.html">layer_gru</a>()</code> <br><code><a href="layer_hashed_crossing.html">layer_hashed_crossing</a>()</code> <br><code><a href="layer_hashing.html">layer_hashing</a>()</code> <br><code><a href="layer_identity.html">layer_identity</a>()</code> <br><code><a href="layer_integer_lookup.html">layer_integer_lookup</a>()</code> <br><code><a href="layer_jax_model_wrapper.html">layer_jax_model_wrapper</a>()</code> <br><code><a href="layer_lambda.html">layer_lambda</a>()</code> <br><code><a href="layer_layer_normalization.html">layer_layer_normalization</a>()</code> <br><code><a href="layer_lstm.html">layer_lstm</a>()</code> <br><code><a href="layer_masking.html">layer_masking</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_max_pooling_1d.html">layer_max_pooling_1d</a>()</code> <br><code><a href="layer_max_pooling_2d.html">layer_max_pooling_2d</a>()</code> <br><code><a href="layer_max_pooling_3d.html">layer_max_pooling_3d</a>()</code> <br><code><a href="layer_maximum.html">layer_maximum</a>()</code> <br><code><a href="layer_mel_spectrogram.html">layer_mel_spectrogram</a>()</code> <br><code><a href="layer_minimum.html">layer_minimum</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_multi_head_attention.html">layer_multi_head_attention</a>()</code> <br><code><a href="layer_multiply.html">layer_multiply</a>()</code> <br><code><a href="layer_normalization.html">layer_normalization</a>()</code> <br><code><a href="layer_permute.html">layer_permute</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_brightness.html">layer_random_brightness</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_contrast.html">layer_random_contrast</a>()</code> <br><code><a href="layer_random_crop.html">layer_random_crop</a>()</code> <br><code><a href="layer_random_elastic_transform.html">layer_random_elastic_transform</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_flip.html">layer_random_flip</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_rotation.html">layer_random_rotation</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_random_translation.html">layer_random_translation</a>()</code> <br><code><a href="layer_random_zoom.html">layer_random_zoom</a>()</code> <br><code><a href="layer_repeat_vector.html">layer_repeat_vector</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_reshape.html">layer_reshape</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_rms_normalization.html">layer_rms_normalization</a>()</code> <br><code><a href="layer_rnn.html">layer_rnn</a>()</code> <br><code><a href="layer_separable_conv_1d.html">layer_separable_conv_1d</a>()</code> <br><code><a href="layer_separable_conv_2d.html">layer_separable_conv_2d</a>()</code> <br><code><a href="layer_simple_rnn.html">layer_simple_rnn</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br><code><a href="layer_spatial_dropout_1d.html">layer_spatial_dropout_1d</a>()</code> <br><code><a href="layer_spatial_dropout_2d.html">layer_spatial_dropout_2d</a>()</code> <br><code><a href="layer_spatial_dropout_3d.html">layer_spatial_dropout_3d</a>()</code> <br><code><a href="layer_spectral_normalization.html">layer_spectral_normalization</a>()</code> <br><code><a href="layer_stft_spectrogram.html">layer_stft_spectrogram</a>()</code> <br><code><a href="layer_string_lookup.html">layer_string_lookup</a>()</code> <br><code><a href="layer_subtract.html">layer_subtract</a>()</code> <br><code><a href="layer_text_vectorization.html">layer_text_vectorization</a>()</code> <br><code><a href="layer_tfsm.html">layer_tfsm</a>()</code> <br><code><a href="layer_time_distributed.html">layer_time_distributed</a>()</code> <br><code><a href="layer_torch_module_wrapper.html">layer_torch_module_wrapper</a>()</code> <br><code><a href="layer_unit_normalization.html">layer_unit_normalization</a>()</code> <br><code><a href="layer_upsampling_1d.html">layer_upsampling_1d</a>()</code> <br><code><a href="layer_upsampling_2d.html">layer_upsampling_2d</a>()</code> <br><code><a href="layer_upsampling_3d.html">layer_upsampling_3d</a>()</code> <br><code><a href="layer_zero_padding_1d.html">layer_zero_padding_1d</a>()</code> <br><code><a href="layer_zero_padding_2d.html">layer_zero_padding_2d</a>()</code> <br><code><a href="layer_zero_padding_3d.html">layer_zero_padding_3d</a>()</code> <br><code><a href="rnn_cell_gru.html">rnn_cell_gru</a>()</code> <br><code><a href="rnn_cell_lstm.html">rnn_cell_lstm</a>()</code> <br><code><a href="rnn_cell_simple.html">rnn_cell_simple</a>()</code> <br><code><a href="rnn_cells_stack.html">rnn_cells_stack</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -149,7 +149,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/layer_random_posterization.html b/docs/dev/reference/layer_random_posterization.html
index 5f0485bc4..78b68e47f 100644
--- a/docs/dev/reference/layer_random_posterization.html
+++ b/docs/dev/reference/layer_random_posterization.html
@@ -8,7 +8,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -114,9 +114,9 @@ <h2 id="references">References<a class="anchor" aria-label="anchor" href="#refer
 </ul></div>
     <div class="section level2">
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
-    <div class="dont-index"><p>Other image preprocessing layers: <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br></p>
-<p>Other preprocessing layers: <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_category_encoding.html">layer_category_encoding</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_discretization.html">layer_discretization</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_feature_space.html">layer_feature_space</a>()</code> <br><code><a href="layer_hashed_crossing.html">layer_hashed_crossing</a>()</code> <br><code><a href="layer_hashing.html">layer_hashing</a>()</code> <br><code><a href="layer_integer_lookup.html">layer_integer_lookup</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_mel_spectrogram.html">layer_mel_spectrogram</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_normalization.html">layer_normalization</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_brightness.html">layer_random_brightness</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_contrast.html">layer_random_contrast</a>()</code> <br><code><a href="layer_random_crop.html">layer_random_crop</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_flip.html">layer_random_flip</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_rotation.html">layer_random_rotation</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_random_translation.html">layer_random_translation</a>()</code> <br><code><a href="layer_random_zoom.html">layer_random_zoom</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br><code><a href="layer_stft_spectrogram.html">layer_stft_spectrogram</a>()</code> <br><code><a href="layer_string_lookup.html">layer_string_lookup</a>()</code> <br><code><a href="layer_text_vectorization.html">layer_text_vectorization</a>()</code> <br></p>
-<p>Other layers: <br><code><a href="Layer.html">Layer</a>()</code> <br><code><a href="layer_activation.html">layer_activation</a>()</code> <br><code><a href="layer_activation_elu.html">layer_activation_elu</a>()</code> <br><code><a href="layer_activation_leaky_relu.html">layer_activation_leaky_relu</a>()</code> <br><code><a href="layer_activation_parametric_relu.html">layer_activation_parametric_relu</a>()</code> <br><code><a href="layer_activation_relu.html">layer_activation_relu</a>()</code> <br><code><a href="layer_activation_softmax.html">layer_activation_softmax</a>()</code> <br><code><a href="layer_activity_regularization.html">layer_activity_regularization</a>()</code> <br><code><a href="layer_add.html">layer_add</a>()</code> <br><code><a href="layer_additive_attention.html">layer_additive_attention</a>()</code> <br><code><a href="layer_alpha_dropout.html">layer_alpha_dropout</a>()</code> <br><code><a href="layer_attention.html">layer_attention</a>()</code> <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_average.html">layer_average</a>()</code> <br><code><a href="layer_average_pooling_1d.html">layer_average_pooling_1d</a>()</code> <br><code><a href="layer_average_pooling_2d.html">layer_average_pooling_2d</a>()</code> <br><code><a href="layer_average_pooling_3d.html">layer_average_pooling_3d</a>()</code> <br><code><a href="layer_batch_normalization.html">layer_batch_normalization</a>()</code> <br><code><a href="layer_bidirectional.html">layer_bidirectional</a>()</code> <br><code><a href="layer_category_encoding.html">layer_category_encoding</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_concatenate.html">layer_concatenate</a>()</code> <br><code><a href="layer_conv_1d.html">layer_conv_1d</a>()</code> <br><code><a href="layer_conv_1d_transpose.html">layer_conv_1d_transpose</a>()</code> <br><code><a href="layer_conv_2d.html">layer_conv_2d</a>()</code> <br><code><a href="layer_conv_2d_transpose.html">layer_conv_2d_transpose</a>()</code> <br><code><a href="layer_conv_3d.html">layer_conv_3d</a>()</code> <br><code><a href="layer_conv_3d_transpose.html">layer_conv_3d_transpose</a>()</code> <br><code><a href="layer_conv_lstm_1d.html">layer_conv_lstm_1d</a>()</code> <br><code><a href="layer_conv_lstm_2d.html">layer_conv_lstm_2d</a>()</code> <br><code><a href="layer_conv_lstm_3d.html">layer_conv_lstm_3d</a>()</code> <br><code><a href="layer_cropping_1d.html">layer_cropping_1d</a>()</code> <br><code><a href="layer_cropping_2d.html">layer_cropping_2d</a>()</code> <br><code><a href="layer_cropping_3d.html">layer_cropping_3d</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_dense.html">layer_dense</a>()</code> <br><code><a href="layer_depthwise_conv_1d.html">layer_depthwise_conv_1d</a>()</code> <br><code><a href="layer_depthwise_conv_2d.html">layer_depthwise_conv_2d</a>()</code> <br><code><a href="layer_discretization.html">layer_discretization</a>()</code> <br><code><a href="layer_dot.html">layer_dot</a>()</code> <br><code><a href="layer_dropout.html">layer_dropout</a>()</code> <br><code><a href="layer_einsum_dense.html">layer_einsum_dense</a>()</code> <br><code><a href="layer_embedding.html">layer_embedding</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_feature_space.html">layer_feature_space</a>()</code> <br><code><a href="layer_flatten.html">layer_flatten</a>()</code> <br><code><a href="layer_flax_module_wrapper.html">layer_flax_module_wrapper</a>()</code> <br><code><a href="layer_gaussian_dropout.html">layer_gaussian_dropout</a>()</code> <br><code><a href="layer_gaussian_noise.html">layer_gaussian_noise</a>()</code> <br><code><a href="layer_global_average_pooling_1d.html">layer_global_average_pooling_1d</a>()</code> <br><code><a href="layer_global_average_pooling_2d.html">layer_global_average_pooling_2d</a>()</code> <br><code><a href="layer_global_average_pooling_3d.html">layer_global_average_pooling_3d</a>()</code> <br><code><a href="layer_global_max_pooling_1d.html">layer_global_max_pooling_1d</a>()</code> <br><code><a href="layer_global_max_pooling_2d.html">layer_global_max_pooling_2d</a>()</code> <br><code><a href="layer_global_max_pooling_3d.html">layer_global_max_pooling_3d</a>()</code> <br><code><a href="layer_group_normalization.html">layer_group_normalization</a>()</code> <br><code><a href="layer_group_query_attention.html">layer_group_query_attention</a>()</code> <br><code><a href="layer_gru.html">layer_gru</a>()</code> <br><code><a href="layer_hashed_crossing.html">layer_hashed_crossing</a>()</code> <br><code><a href="layer_hashing.html">layer_hashing</a>()</code> <br><code><a href="layer_identity.html">layer_identity</a>()</code> <br><code><a href="layer_integer_lookup.html">layer_integer_lookup</a>()</code> <br><code><a href="layer_jax_model_wrapper.html">layer_jax_model_wrapper</a>()</code> <br><code><a href="layer_lambda.html">layer_lambda</a>()</code> <br><code><a href="layer_layer_normalization.html">layer_layer_normalization</a>()</code> <br><code><a href="layer_lstm.html">layer_lstm</a>()</code> <br><code><a href="layer_masking.html">layer_masking</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_max_pooling_1d.html">layer_max_pooling_1d</a>()</code> <br><code><a href="layer_max_pooling_2d.html">layer_max_pooling_2d</a>()</code> <br><code><a href="layer_max_pooling_3d.html">layer_max_pooling_3d</a>()</code> <br><code><a href="layer_maximum.html">layer_maximum</a>()</code> <br><code><a href="layer_mel_spectrogram.html">layer_mel_spectrogram</a>()</code> <br><code><a href="layer_minimum.html">layer_minimum</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_multi_head_attention.html">layer_multi_head_attention</a>()</code> <br><code><a href="layer_multiply.html">layer_multiply</a>()</code> <br><code><a href="layer_normalization.html">layer_normalization</a>()</code> <br><code><a href="layer_permute.html">layer_permute</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_brightness.html">layer_random_brightness</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_contrast.html">layer_random_contrast</a>()</code> <br><code><a href="layer_random_crop.html">layer_random_crop</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_flip.html">layer_random_flip</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_rotation.html">layer_random_rotation</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_random_translation.html">layer_random_translation</a>()</code> <br><code><a href="layer_random_zoom.html">layer_random_zoom</a>()</code> <br><code><a href="layer_repeat_vector.html">layer_repeat_vector</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_reshape.html">layer_reshape</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_rms_normalization.html">layer_rms_normalization</a>()</code> <br><code><a href="layer_rnn.html">layer_rnn</a>()</code> <br><code><a href="layer_separable_conv_1d.html">layer_separable_conv_1d</a>()</code> <br><code><a href="layer_separable_conv_2d.html">layer_separable_conv_2d</a>()</code> <br><code><a href="layer_simple_rnn.html">layer_simple_rnn</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br><code><a href="layer_spatial_dropout_1d.html">layer_spatial_dropout_1d</a>()</code> <br><code><a href="layer_spatial_dropout_2d.html">layer_spatial_dropout_2d</a>()</code> <br><code><a href="layer_spatial_dropout_3d.html">layer_spatial_dropout_3d</a>()</code> <br><code><a href="layer_spectral_normalization.html">layer_spectral_normalization</a>()</code> <br><code><a href="layer_stft_spectrogram.html">layer_stft_spectrogram</a>()</code> <br><code><a href="layer_string_lookup.html">layer_string_lookup</a>()</code> <br><code><a href="layer_subtract.html">layer_subtract</a>()</code> <br><code><a href="layer_text_vectorization.html">layer_text_vectorization</a>()</code> <br><code><a href="layer_tfsm.html">layer_tfsm</a>()</code> <br><code><a href="layer_time_distributed.html">layer_time_distributed</a>()</code> <br><code><a href="layer_torch_module_wrapper.html">layer_torch_module_wrapper</a>()</code> <br><code><a href="layer_unit_normalization.html">layer_unit_normalization</a>()</code> <br><code><a href="layer_upsampling_1d.html">layer_upsampling_1d</a>()</code> <br><code><a href="layer_upsampling_2d.html">layer_upsampling_2d</a>()</code> <br><code><a href="layer_upsampling_3d.html">layer_upsampling_3d</a>()</code> <br><code><a href="layer_zero_padding_1d.html">layer_zero_padding_1d</a>()</code> <br><code><a href="layer_zero_padding_2d.html">layer_zero_padding_2d</a>()</code> <br><code><a href="layer_zero_padding_3d.html">layer_zero_padding_3d</a>()</code> <br><code><a href="rnn_cell_gru.html">rnn_cell_gru</a>()</code> <br><code><a href="rnn_cell_lstm.html">rnn_cell_lstm</a>()</code> <br><code><a href="rnn_cell_simple.html">rnn_cell_simple</a>()</code> <br><code><a href="rnn_cells_stack.html">rnn_cells_stack</a>()</code> <br></p></div>
+    <div class="dont-index"><p>Other image preprocessing layers: <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_elastic_transform.html">layer_random_elastic_transform</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br></p>
+<p>Other preprocessing layers: <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_category_encoding.html">layer_category_encoding</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_discretization.html">layer_discretization</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_feature_space.html">layer_feature_space</a>()</code> <br><code><a href="layer_hashed_crossing.html">layer_hashed_crossing</a>()</code> <br><code><a href="layer_hashing.html">layer_hashing</a>()</code> <br><code><a href="layer_integer_lookup.html">layer_integer_lookup</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_mel_spectrogram.html">layer_mel_spectrogram</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_normalization.html">layer_normalization</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_brightness.html">layer_random_brightness</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_contrast.html">layer_random_contrast</a>()</code> <br><code><a href="layer_random_crop.html">layer_random_crop</a>()</code> <br><code><a href="layer_random_elastic_transform.html">layer_random_elastic_transform</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_flip.html">layer_random_flip</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_rotation.html">layer_random_rotation</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_random_translation.html">layer_random_translation</a>()</code> <br><code><a href="layer_random_zoom.html">layer_random_zoom</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br><code><a href="layer_stft_spectrogram.html">layer_stft_spectrogram</a>()</code> <br><code><a href="layer_string_lookup.html">layer_string_lookup</a>()</code> <br><code><a href="layer_text_vectorization.html">layer_text_vectorization</a>()</code> <br></p>
+<p>Other layers: <br><code><a href="Layer.html">Layer</a>()</code> <br><code><a href="layer_activation.html">layer_activation</a>()</code> <br><code><a href="layer_activation_elu.html">layer_activation_elu</a>()</code> <br><code><a href="layer_activation_leaky_relu.html">layer_activation_leaky_relu</a>()</code> <br><code><a href="layer_activation_parametric_relu.html">layer_activation_parametric_relu</a>()</code> <br><code><a href="layer_activation_relu.html">layer_activation_relu</a>()</code> <br><code><a href="layer_activation_softmax.html">layer_activation_softmax</a>()</code> <br><code><a href="layer_activity_regularization.html">layer_activity_regularization</a>()</code> <br><code><a href="layer_add.html">layer_add</a>()</code> <br><code><a href="layer_additive_attention.html">layer_additive_attention</a>()</code> <br><code><a href="layer_alpha_dropout.html">layer_alpha_dropout</a>()</code> <br><code><a href="layer_attention.html">layer_attention</a>()</code> <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_average.html">layer_average</a>()</code> <br><code><a href="layer_average_pooling_1d.html">layer_average_pooling_1d</a>()</code> <br><code><a href="layer_average_pooling_2d.html">layer_average_pooling_2d</a>()</code> <br><code><a href="layer_average_pooling_3d.html">layer_average_pooling_3d</a>()</code> <br><code><a href="layer_batch_normalization.html">layer_batch_normalization</a>()</code> <br><code><a href="layer_bidirectional.html">layer_bidirectional</a>()</code> <br><code><a href="layer_category_encoding.html">layer_category_encoding</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_concatenate.html">layer_concatenate</a>()</code> <br><code><a href="layer_conv_1d.html">layer_conv_1d</a>()</code> <br><code><a href="layer_conv_1d_transpose.html">layer_conv_1d_transpose</a>()</code> <br><code><a href="layer_conv_2d.html">layer_conv_2d</a>()</code> <br><code><a href="layer_conv_2d_transpose.html">layer_conv_2d_transpose</a>()</code> <br><code><a href="layer_conv_3d.html">layer_conv_3d</a>()</code> <br><code><a href="layer_conv_3d_transpose.html">layer_conv_3d_transpose</a>()</code> <br><code><a href="layer_conv_lstm_1d.html">layer_conv_lstm_1d</a>()</code> <br><code><a href="layer_conv_lstm_2d.html">layer_conv_lstm_2d</a>()</code> <br><code><a href="layer_conv_lstm_3d.html">layer_conv_lstm_3d</a>()</code> <br><code><a href="layer_cropping_1d.html">layer_cropping_1d</a>()</code> <br><code><a href="layer_cropping_2d.html">layer_cropping_2d</a>()</code> <br><code><a href="layer_cropping_3d.html">layer_cropping_3d</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_dense.html">layer_dense</a>()</code> <br><code><a href="layer_depthwise_conv_1d.html">layer_depthwise_conv_1d</a>()</code> <br><code><a href="layer_depthwise_conv_2d.html">layer_depthwise_conv_2d</a>()</code> <br><code><a href="layer_discretization.html">layer_discretization</a>()</code> <br><code><a href="layer_dot.html">layer_dot</a>()</code> <br><code><a href="layer_dropout.html">layer_dropout</a>()</code> <br><code><a href="layer_einsum_dense.html">layer_einsum_dense</a>()</code> <br><code><a href="layer_embedding.html">layer_embedding</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_feature_space.html">layer_feature_space</a>()</code> <br><code><a href="layer_flatten.html">layer_flatten</a>()</code> <br><code><a href="layer_flax_module_wrapper.html">layer_flax_module_wrapper</a>()</code> <br><code><a href="layer_gaussian_dropout.html">layer_gaussian_dropout</a>()</code> <br><code><a href="layer_gaussian_noise.html">layer_gaussian_noise</a>()</code> <br><code><a href="layer_global_average_pooling_1d.html">layer_global_average_pooling_1d</a>()</code> <br><code><a href="layer_global_average_pooling_2d.html">layer_global_average_pooling_2d</a>()</code> <br><code><a href="layer_global_average_pooling_3d.html">layer_global_average_pooling_3d</a>()</code> <br><code><a href="layer_global_max_pooling_1d.html">layer_global_max_pooling_1d</a>()</code> <br><code><a href="layer_global_max_pooling_2d.html">layer_global_max_pooling_2d</a>()</code> <br><code><a href="layer_global_max_pooling_3d.html">layer_global_max_pooling_3d</a>()</code> <br><code><a href="layer_group_normalization.html">layer_group_normalization</a>()</code> <br><code><a href="layer_group_query_attention.html">layer_group_query_attention</a>()</code> <br><code><a href="layer_gru.html">layer_gru</a>()</code> <br><code><a href="layer_hashed_crossing.html">layer_hashed_crossing</a>()</code> <br><code><a href="layer_hashing.html">layer_hashing</a>()</code> <br><code><a href="layer_identity.html">layer_identity</a>()</code> <br><code><a href="layer_integer_lookup.html">layer_integer_lookup</a>()</code> <br><code><a href="layer_jax_model_wrapper.html">layer_jax_model_wrapper</a>()</code> <br><code><a href="layer_lambda.html">layer_lambda</a>()</code> <br><code><a href="layer_layer_normalization.html">layer_layer_normalization</a>()</code> <br><code><a href="layer_lstm.html">layer_lstm</a>()</code> <br><code><a href="layer_masking.html">layer_masking</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_max_pooling_1d.html">layer_max_pooling_1d</a>()</code> <br><code><a href="layer_max_pooling_2d.html">layer_max_pooling_2d</a>()</code> <br><code><a href="layer_max_pooling_3d.html">layer_max_pooling_3d</a>()</code> <br><code><a href="layer_maximum.html">layer_maximum</a>()</code> <br><code><a href="layer_mel_spectrogram.html">layer_mel_spectrogram</a>()</code> <br><code><a href="layer_minimum.html">layer_minimum</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_multi_head_attention.html">layer_multi_head_attention</a>()</code> <br><code><a href="layer_multiply.html">layer_multiply</a>()</code> <br><code><a href="layer_normalization.html">layer_normalization</a>()</code> <br><code><a href="layer_permute.html">layer_permute</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_brightness.html">layer_random_brightness</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_contrast.html">layer_random_contrast</a>()</code> <br><code><a href="layer_random_crop.html">layer_random_crop</a>()</code> <br><code><a href="layer_random_elastic_transform.html">layer_random_elastic_transform</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_flip.html">layer_random_flip</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_rotation.html">layer_random_rotation</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_random_translation.html">layer_random_translation</a>()</code> <br><code><a href="layer_random_zoom.html">layer_random_zoom</a>()</code> <br><code><a href="layer_repeat_vector.html">layer_repeat_vector</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_reshape.html">layer_reshape</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_rms_normalization.html">layer_rms_normalization</a>()</code> <br><code><a href="layer_rnn.html">layer_rnn</a>()</code> <br><code><a href="layer_separable_conv_1d.html">layer_separable_conv_1d</a>()</code> <br><code><a href="layer_separable_conv_2d.html">layer_separable_conv_2d</a>()</code> <br><code><a href="layer_simple_rnn.html">layer_simple_rnn</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br><code><a href="layer_spatial_dropout_1d.html">layer_spatial_dropout_1d</a>()</code> <br><code><a href="layer_spatial_dropout_2d.html">layer_spatial_dropout_2d</a>()</code> <br><code><a href="layer_spatial_dropout_3d.html">layer_spatial_dropout_3d</a>()</code> <br><code><a href="layer_spectral_normalization.html">layer_spectral_normalization</a>()</code> <br><code><a href="layer_stft_spectrogram.html">layer_stft_spectrogram</a>()</code> <br><code><a href="layer_string_lookup.html">layer_string_lookup</a>()</code> <br><code><a href="layer_subtract.html">layer_subtract</a>()</code> <br><code><a href="layer_text_vectorization.html">layer_text_vectorization</a>()</code> <br><code><a href="layer_tfsm.html">layer_tfsm</a>()</code> <br><code><a href="layer_time_distributed.html">layer_time_distributed</a>()</code> <br><code><a href="layer_torch_module_wrapper.html">layer_torch_module_wrapper</a>()</code> <br><code><a href="layer_unit_normalization.html">layer_unit_normalization</a>()</code> <br><code><a href="layer_upsampling_1d.html">layer_upsampling_1d</a>()</code> <br><code><a href="layer_upsampling_2d.html">layer_upsampling_2d</a>()</code> <br><code><a href="layer_upsampling_3d.html">layer_upsampling_3d</a>()</code> <br><code><a href="layer_zero_padding_1d.html">layer_zero_padding_1d</a>()</code> <br><code><a href="layer_zero_padding_2d.html">layer_zero_padding_2d</a>()</code> <br><code><a href="layer_zero_padding_3d.html">layer_zero_padding_3d</a>()</code> <br><code><a href="rnn_cell_gru.html">rnn_cell_gru</a>()</code> <br><code><a href="rnn_cell_lstm.html">rnn_cell_lstm</a>()</code> <br><code><a href="rnn_cell_simple.html">rnn_cell_simple</a>()</code> <br><code><a href="rnn_cells_stack.html">rnn_cells_stack</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -128,7 +128,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/layer_random_rotation.html b/docs/dev/reference/layer_random_rotation.html
index 157fa7b79..a9d5be681 100644
--- a/docs/dev/reference/layer_random_rotation.html
+++ b/docs/dev/reference/layer_random_rotation.html
@@ -26,7 +26,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -192,8 +192,8 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
     <div class="dont-index">
 <ul><li><p><a href="https://keras.io/api/layers/preprocessing_layers/image_augmentation/random_rotation#randomrotation-class" class="external-link">https://keras.io/api/layers/preprocessing_layers/image_augmentation/random_rotation#randomrotation-class</a></p></li>
 </ul><p>Other image augmentation layers: <br><code><a href="layer_random_brightness.html">layer_random_brightness</a>()</code> <br><code><a href="layer_random_contrast.html">layer_random_contrast</a>()</code> <br><code><a href="layer_random_crop.html">layer_random_crop</a>()</code> <br><code><a href="layer_random_flip.html">layer_random_flip</a>()</code> <br><code><a href="layer_random_translation.html">layer_random_translation</a>()</code> <br><code><a href="layer_random_zoom.html">layer_random_zoom</a>()</code> <br></p>
-<p>Other preprocessing layers: <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_category_encoding.html">layer_category_encoding</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_discretization.html">layer_discretization</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_feature_space.html">layer_feature_space</a>()</code> <br><code><a href="layer_hashed_crossing.html">layer_hashed_crossing</a>()</code> <br><code><a href="layer_hashing.html">layer_hashing</a>()</code> <br><code><a href="layer_integer_lookup.html">layer_integer_lookup</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_mel_spectrogram.html">layer_mel_spectrogram</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_normalization.html">layer_normalization</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_brightness.html">layer_random_brightness</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_contrast.html">layer_random_contrast</a>()</code> <br><code><a href="layer_random_crop.html">layer_random_crop</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_flip.html">layer_random_flip</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_random_translation.html">layer_random_translation</a>()</code> <br><code><a href="layer_random_zoom.html">layer_random_zoom</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br><code><a href="layer_stft_spectrogram.html">layer_stft_spectrogram</a>()</code> <br><code><a href="layer_string_lookup.html">layer_string_lookup</a>()</code> <br><code><a href="layer_text_vectorization.html">layer_text_vectorization</a>()</code> <br></p>
-<p>Other layers: <br><code><a href="Layer.html">Layer</a>()</code> <br><code><a href="layer_activation.html">layer_activation</a>()</code> <br><code><a href="layer_activation_elu.html">layer_activation_elu</a>()</code> <br><code><a href="layer_activation_leaky_relu.html">layer_activation_leaky_relu</a>()</code> <br><code><a href="layer_activation_parametric_relu.html">layer_activation_parametric_relu</a>()</code> <br><code><a href="layer_activation_relu.html">layer_activation_relu</a>()</code> <br><code><a href="layer_activation_softmax.html">layer_activation_softmax</a>()</code> <br><code><a href="layer_activity_regularization.html">layer_activity_regularization</a>()</code> <br><code><a href="layer_add.html">layer_add</a>()</code> <br><code><a href="layer_additive_attention.html">layer_additive_attention</a>()</code> <br><code><a href="layer_alpha_dropout.html">layer_alpha_dropout</a>()</code> <br><code><a href="layer_attention.html">layer_attention</a>()</code> <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_average.html">layer_average</a>()</code> <br><code><a href="layer_average_pooling_1d.html">layer_average_pooling_1d</a>()</code> <br><code><a href="layer_average_pooling_2d.html">layer_average_pooling_2d</a>()</code> <br><code><a href="layer_average_pooling_3d.html">layer_average_pooling_3d</a>()</code> <br><code><a href="layer_batch_normalization.html">layer_batch_normalization</a>()</code> <br><code><a href="layer_bidirectional.html">layer_bidirectional</a>()</code> <br><code><a href="layer_category_encoding.html">layer_category_encoding</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_concatenate.html">layer_concatenate</a>()</code> <br><code><a href="layer_conv_1d.html">layer_conv_1d</a>()</code> <br><code><a href="layer_conv_1d_transpose.html">layer_conv_1d_transpose</a>()</code> <br><code><a href="layer_conv_2d.html">layer_conv_2d</a>()</code> <br><code><a href="layer_conv_2d_transpose.html">layer_conv_2d_transpose</a>()</code> <br><code><a href="layer_conv_3d.html">layer_conv_3d</a>()</code> <br><code><a href="layer_conv_3d_transpose.html">layer_conv_3d_transpose</a>()</code> <br><code><a href="layer_conv_lstm_1d.html">layer_conv_lstm_1d</a>()</code> <br><code><a href="layer_conv_lstm_2d.html">layer_conv_lstm_2d</a>()</code> <br><code><a href="layer_conv_lstm_3d.html">layer_conv_lstm_3d</a>()</code> <br><code><a href="layer_cropping_1d.html">layer_cropping_1d</a>()</code> <br><code><a href="layer_cropping_2d.html">layer_cropping_2d</a>()</code> <br><code><a href="layer_cropping_3d.html">layer_cropping_3d</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_dense.html">layer_dense</a>()</code> <br><code><a href="layer_depthwise_conv_1d.html">layer_depthwise_conv_1d</a>()</code> <br><code><a href="layer_depthwise_conv_2d.html">layer_depthwise_conv_2d</a>()</code> <br><code><a href="layer_discretization.html">layer_discretization</a>()</code> <br><code><a href="layer_dot.html">layer_dot</a>()</code> <br><code><a href="layer_dropout.html">layer_dropout</a>()</code> <br><code><a href="layer_einsum_dense.html">layer_einsum_dense</a>()</code> <br><code><a href="layer_embedding.html">layer_embedding</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_feature_space.html">layer_feature_space</a>()</code> <br><code><a href="layer_flatten.html">layer_flatten</a>()</code> <br><code><a href="layer_flax_module_wrapper.html">layer_flax_module_wrapper</a>()</code> <br><code><a href="layer_gaussian_dropout.html">layer_gaussian_dropout</a>()</code> <br><code><a href="layer_gaussian_noise.html">layer_gaussian_noise</a>()</code> <br><code><a href="layer_global_average_pooling_1d.html">layer_global_average_pooling_1d</a>()</code> <br><code><a href="layer_global_average_pooling_2d.html">layer_global_average_pooling_2d</a>()</code> <br><code><a href="layer_global_average_pooling_3d.html">layer_global_average_pooling_3d</a>()</code> <br><code><a href="layer_global_max_pooling_1d.html">layer_global_max_pooling_1d</a>()</code> <br><code><a href="layer_global_max_pooling_2d.html">layer_global_max_pooling_2d</a>()</code> <br><code><a href="layer_global_max_pooling_3d.html">layer_global_max_pooling_3d</a>()</code> <br><code><a href="layer_group_normalization.html">layer_group_normalization</a>()</code> <br><code><a href="layer_group_query_attention.html">layer_group_query_attention</a>()</code> <br><code><a href="layer_gru.html">layer_gru</a>()</code> <br><code><a href="layer_hashed_crossing.html">layer_hashed_crossing</a>()</code> <br><code><a href="layer_hashing.html">layer_hashing</a>()</code> <br><code><a href="layer_identity.html">layer_identity</a>()</code> <br><code><a href="layer_integer_lookup.html">layer_integer_lookup</a>()</code> <br><code><a href="layer_jax_model_wrapper.html">layer_jax_model_wrapper</a>()</code> <br><code><a href="layer_lambda.html">layer_lambda</a>()</code> <br><code><a href="layer_layer_normalization.html">layer_layer_normalization</a>()</code> <br><code><a href="layer_lstm.html">layer_lstm</a>()</code> <br><code><a href="layer_masking.html">layer_masking</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_max_pooling_1d.html">layer_max_pooling_1d</a>()</code> <br><code><a href="layer_max_pooling_2d.html">layer_max_pooling_2d</a>()</code> <br><code><a href="layer_max_pooling_3d.html">layer_max_pooling_3d</a>()</code> <br><code><a href="layer_maximum.html">layer_maximum</a>()</code> <br><code><a href="layer_mel_spectrogram.html">layer_mel_spectrogram</a>()</code> <br><code><a href="layer_minimum.html">layer_minimum</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_multi_head_attention.html">layer_multi_head_attention</a>()</code> <br><code><a href="layer_multiply.html">layer_multiply</a>()</code> <br><code><a href="layer_normalization.html">layer_normalization</a>()</code> <br><code><a href="layer_permute.html">layer_permute</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_brightness.html">layer_random_brightness</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_contrast.html">layer_random_contrast</a>()</code> <br><code><a href="layer_random_crop.html">layer_random_crop</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_flip.html">layer_random_flip</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_random_translation.html">layer_random_translation</a>()</code> <br><code><a href="layer_random_zoom.html">layer_random_zoom</a>()</code> <br><code><a href="layer_repeat_vector.html">layer_repeat_vector</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_reshape.html">layer_reshape</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_rms_normalization.html">layer_rms_normalization</a>()</code> <br><code><a href="layer_rnn.html">layer_rnn</a>()</code> <br><code><a href="layer_separable_conv_1d.html">layer_separable_conv_1d</a>()</code> <br><code><a href="layer_separable_conv_2d.html">layer_separable_conv_2d</a>()</code> <br><code><a href="layer_simple_rnn.html">layer_simple_rnn</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br><code><a href="layer_spatial_dropout_1d.html">layer_spatial_dropout_1d</a>()</code> <br><code><a href="layer_spatial_dropout_2d.html">layer_spatial_dropout_2d</a>()</code> <br><code><a href="layer_spatial_dropout_3d.html">layer_spatial_dropout_3d</a>()</code> <br><code><a href="layer_spectral_normalization.html">layer_spectral_normalization</a>()</code> <br><code><a href="layer_stft_spectrogram.html">layer_stft_spectrogram</a>()</code> <br><code><a href="layer_string_lookup.html">layer_string_lookup</a>()</code> <br><code><a href="layer_subtract.html">layer_subtract</a>()</code> <br><code><a href="layer_text_vectorization.html">layer_text_vectorization</a>()</code> <br><code><a href="layer_tfsm.html">layer_tfsm</a>()</code> <br><code><a href="layer_time_distributed.html">layer_time_distributed</a>()</code> <br><code><a href="layer_torch_module_wrapper.html">layer_torch_module_wrapper</a>()</code> <br><code><a href="layer_unit_normalization.html">layer_unit_normalization</a>()</code> <br><code><a href="layer_upsampling_1d.html">layer_upsampling_1d</a>()</code> <br><code><a href="layer_upsampling_2d.html">layer_upsampling_2d</a>()</code> <br><code><a href="layer_upsampling_3d.html">layer_upsampling_3d</a>()</code> <br><code><a href="layer_zero_padding_1d.html">layer_zero_padding_1d</a>()</code> <br><code><a href="layer_zero_padding_2d.html">layer_zero_padding_2d</a>()</code> <br><code><a href="layer_zero_padding_3d.html">layer_zero_padding_3d</a>()</code> <br><code><a href="rnn_cell_gru.html">rnn_cell_gru</a>()</code> <br><code><a href="rnn_cell_lstm.html">rnn_cell_lstm</a>()</code> <br><code><a href="rnn_cell_simple.html">rnn_cell_simple</a>()</code> <br><code><a href="rnn_cells_stack.html">rnn_cells_stack</a>()</code> <br></p></div>
+<p>Other preprocessing layers: <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_category_encoding.html">layer_category_encoding</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_discretization.html">layer_discretization</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_feature_space.html">layer_feature_space</a>()</code> <br><code><a href="layer_hashed_crossing.html">layer_hashed_crossing</a>()</code> <br><code><a href="layer_hashing.html">layer_hashing</a>()</code> <br><code><a href="layer_integer_lookup.html">layer_integer_lookup</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_mel_spectrogram.html">layer_mel_spectrogram</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_normalization.html">layer_normalization</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_brightness.html">layer_random_brightness</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_contrast.html">layer_random_contrast</a>()</code> <br><code><a href="layer_random_crop.html">layer_random_crop</a>()</code> <br><code><a href="layer_random_elastic_transform.html">layer_random_elastic_transform</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_flip.html">layer_random_flip</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_random_translation.html">layer_random_translation</a>()</code> <br><code><a href="layer_random_zoom.html">layer_random_zoom</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br><code><a href="layer_stft_spectrogram.html">layer_stft_spectrogram</a>()</code> <br><code><a href="layer_string_lookup.html">layer_string_lookup</a>()</code> <br><code><a href="layer_text_vectorization.html">layer_text_vectorization</a>()</code> <br></p>
+<p>Other layers: <br><code><a href="Layer.html">Layer</a>()</code> <br><code><a href="layer_activation.html">layer_activation</a>()</code> <br><code><a href="layer_activation_elu.html">layer_activation_elu</a>()</code> <br><code><a href="layer_activation_leaky_relu.html">layer_activation_leaky_relu</a>()</code> <br><code><a href="layer_activation_parametric_relu.html">layer_activation_parametric_relu</a>()</code> <br><code><a href="layer_activation_relu.html">layer_activation_relu</a>()</code> <br><code><a href="layer_activation_softmax.html">layer_activation_softmax</a>()</code> <br><code><a href="layer_activity_regularization.html">layer_activity_regularization</a>()</code> <br><code><a href="layer_add.html">layer_add</a>()</code> <br><code><a href="layer_additive_attention.html">layer_additive_attention</a>()</code> <br><code><a href="layer_alpha_dropout.html">layer_alpha_dropout</a>()</code> <br><code><a href="layer_attention.html">layer_attention</a>()</code> <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_average.html">layer_average</a>()</code> <br><code><a href="layer_average_pooling_1d.html">layer_average_pooling_1d</a>()</code> <br><code><a href="layer_average_pooling_2d.html">layer_average_pooling_2d</a>()</code> <br><code><a href="layer_average_pooling_3d.html">layer_average_pooling_3d</a>()</code> <br><code><a href="layer_batch_normalization.html">layer_batch_normalization</a>()</code> <br><code><a href="layer_bidirectional.html">layer_bidirectional</a>()</code> <br><code><a href="layer_category_encoding.html">layer_category_encoding</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_concatenate.html">layer_concatenate</a>()</code> <br><code><a href="layer_conv_1d.html">layer_conv_1d</a>()</code> <br><code><a href="layer_conv_1d_transpose.html">layer_conv_1d_transpose</a>()</code> <br><code><a href="layer_conv_2d.html">layer_conv_2d</a>()</code> <br><code><a href="layer_conv_2d_transpose.html">layer_conv_2d_transpose</a>()</code> <br><code><a href="layer_conv_3d.html">layer_conv_3d</a>()</code> <br><code><a href="layer_conv_3d_transpose.html">layer_conv_3d_transpose</a>()</code> <br><code><a href="layer_conv_lstm_1d.html">layer_conv_lstm_1d</a>()</code> <br><code><a href="layer_conv_lstm_2d.html">layer_conv_lstm_2d</a>()</code> <br><code><a href="layer_conv_lstm_3d.html">layer_conv_lstm_3d</a>()</code> <br><code><a href="layer_cropping_1d.html">layer_cropping_1d</a>()</code> <br><code><a href="layer_cropping_2d.html">layer_cropping_2d</a>()</code> <br><code><a href="layer_cropping_3d.html">layer_cropping_3d</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_dense.html">layer_dense</a>()</code> <br><code><a href="layer_depthwise_conv_1d.html">layer_depthwise_conv_1d</a>()</code> <br><code><a href="layer_depthwise_conv_2d.html">layer_depthwise_conv_2d</a>()</code> <br><code><a href="layer_discretization.html">layer_discretization</a>()</code> <br><code><a href="layer_dot.html">layer_dot</a>()</code> <br><code><a href="layer_dropout.html">layer_dropout</a>()</code> <br><code><a href="layer_einsum_dense.html">layer_einsum_dense</a>()</code> <br><code><a href="layer_embedding.html">layer_embedding</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_feature_space.html">layer_feature_space</a>()</code> <br><code><a href="layer_flatten.html">layer_flatten</a>()</code> <br><code><a href="layer_flax_module_wrapper.html">layer_flax_module_wrapper</a>()</code> <br><code><a href="layer_gaussian_dropout.html">layer_gaussian_dropout</a>()</code> <br><code><a href="layer_gaussian_noise.html">layer_gaussian_noise</a>()</code> <br><code><a href="layer_global_average_pooling_1d.html">layer_global_average_pooling_1d</a>()</code> <br><code><a href="layer_global_average_pooling_2d.html">layer_global_average_pooling_2d</a>()</code> <br><code><a href="layer_global_average_pooling_3d.html">layer_global_average_pooling_3d</a>()</code> <br><code><a href="layer_global_max_pooling_1d.html">layer_global_max_pooling_1d</a>()</code> <br><code><a href="layer_global_max_pooling_2d.html">layer_global_max_pooling_2d</a>()</code> <br><code><a href="layer_global_max_pooling_3d.html">layer_global_max_pooling_3d</a>()</code> <br><code><a href="layer_group_normalization.html">layer_group_normalization</a>()</code> <br><code><a href="layer_group_query_attention.html">layer_group_query_attention</a>()</code> <br><code><a href="layer_gru.html">layer_gru</a>()</code> <br><code><a href="layer_hashed_crossing.html">layer_hashed_crossing</a>()</code> <br><code><a href="layer_hashing.html">layer_hashing</a>()</code> <br><code><a href="layer_identity.html">layer_identity</a>()</code> <br><code><a href="layer_integer_lookup.html">layer_integer_lookup</a>()</code> <br><code><a href="layer_jax_model_wrapper.html">layer_jax_model_wrapper</a>()</code> <br><code><a href="layer_lambda.html">layer_lambda</a>()</code> <br><code><a href="layer_layer_normalization.html">layer_layer_normalization</a>()</code> <br><code><a href="layer_lstm.html">layer_lstm</a>()</code> <br><code><a href="layer_masking.html">layer_masking</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_max_pooling_1d.html">layer_max_pooling_1d</a>()</code> <br><code><a href="layer_max_pooling_2d.html">layer_max_pooling_2d</a>()</code> <br><code><a href="layer_max_pooling_3d.html">layer_max_pooling_3d</a>()</code> <br><code><a href="layer_maximum.html">layer_maximum</a>()</code> <br><code><a href="layer_mel_spectrogram.html">layer_mel_spectrogram</a>()</code> <br><code><a href="layer_minimum.html">layer_minimum</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_multi_head_attention.html">layer_multi_head_attention</a>()</code> <br><code><a href="layer_multiply.html">layer_multiply</a>()</code> <br><code><a href="layer_normalization.html">layer_normalization</a>()</code> <br><code><a href="layer_permute.html">layer_permute</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_brightness.html">layer_random_brightness</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_contrast.html">layer_random_contrast</a>()</code> <br><code><a href="layer_random_crop.html">layer_random_crop</a>()</code> <br><code><a href="layer_random_elastic_transform.html">layer_random_elastic_transform</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_flip.html">layer_random_flip</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_random_translation.html">layer_random_translation</a>()</code> <br><code><a href="layer_random_zoom.html">layer_random_zoom</a>()</code> <br><code><a href="layer_repeat_vector.html">layer_repeat_vector</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_reshape.html">layer_reshape</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_rms_normalization.html">layer_rms_normalization</a>()</code> <br><code><a href="layer_rnn.html">layer_rnn</a>()</code> <br><code><a href="layer_separable_conv_1d.html">layer_separable_conv_1d</a>()</code> <br><code><a href="layer_separable_conv_2d.html">layer_separable_conv_2d</a>()</code> <br><code><a href="layer_simple_rnn.html">layer_simple_rnn</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br><code><a href="layer_spatial_dropout_1d.html">layer_spatial_dropout_1d</a>()</code> <br><code><a href="layer_spatial_dropout_2d.html">layer_spatial_dropout_2d</a>()</code> <br><code><a href="layer_spatial_dropout_3d.html">layer_spatial_dropout_3d</a>()</code> <br><code><a href="layer_spectral_normalization.html">layer_spectral_normalization</a>()</code> <br><code><a href="layer_stft_spectrogram.html">layer_stft_spectrogram</a>()</code> <br><code><a href="layer_string_lookup.html">layer_string_lookup</a>()</code> <br><code><a href="layer_subtract.html">layer_subtract</a>()</code> <br><code><a href="layer_text_vectorization.html">layer_text_vectorization</a>()</code> <br><code><a href="layer_tfsm.html">layer_tfsm</a>()</code> <br><code><a href="layer_time_distributed.html">layer_time_distributed</a>()</code> <br><code><a href="layer_torch_module_wrapper.html">layer_torch_module_wrapper</a>()</code> <br><code><a href="layer_unit_normalization.html">layer_unit_normalization</a>()</code> <br><code><a href="layer_upsampling_1d.html">layer_upsampling_1d</a>()</code> <br><code><a href="layer_upsampling_2d.html">layer_upsampling_2d</a>()</code> <br><code><a href="layer_upsampling_3d.html">layer_upsampling_3d</a>()</code> <br><code><a href="layer_zero_padding_1d.html">layer_zero_padding_1d</a>()</code> <br><code><a href="layer_zero_padding_2d.html">layer_zero_padding_2d</a>()</code> <br><code><a href="layer_zero_padding_3d.html">layer_zero_padding_3d</a>()</code> <br><code><a href="rnn_cell_gru.html">rnn_cell_gru</a>()</code> <br><code><a href="rnn_cell_lstm.html">rnn_cell_lstm</a>()</code> <br><code><a href="rnn_cell_simple.html">rnn_cell_simple</a>()</code> <br><code><a href="rnn_cells_stack.html">rnn_cells_stack</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -205,7 +205,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/layer_random_saturation.html b/docs/dev/reference/layer_random_saturation.html
index 971c6c20b..68dc7c776 100644
--- a/docs/dev/reference/layer_random_saturation.html
+++ b/docs/dev/reference/layer_random_saturation.html
@@ -10,7 +10,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -126,9 +126,9 @@ <h2 id="examples">Examples<a class="anchor" aria-label="anchor" href="#examples"
     </div>
     <div class="section level2">
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
-    <div class="dont-index"><p>Other image preprocessing layers: <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br></p>
-<p>Other preprocessing layers: <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_category_encoding.html">layer_category_encoding</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_discretization.html">layer_discretization</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_feature_space.html">layer_feature_space</a>()</code> <br><code><a href="layer_hashed_crossing.html">layer_hashed_crossing</a>()</code> <br><code><a href="layer_hashing.html">layer_hashing</a>()</code> <br><code><a href="layer_integer_lookup.html">layer_integer_lookup</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_mel_spectrogram.html">layer_mel_spectrogram</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_normalization.html">layer_normalization</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_brightness.html">layer_random_brightness</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_contrast.html">layer_random_contrast</a>()</code> <br><code><a href="layer_random_crop.html">layer_random_crop</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_flip.html">layer_random_flip</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_rotation.html">layer_random_rotation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_random_translation.html">layer_random_translation</a>()</code> <br><code><a href="layer_random_zoom.html">layer_random_zoom</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br><code><a href="layer_stft_spectrogram.html">layer_stft_spectrogram</a>()</code> <br><code><a href="layer_string_lookup.html">layer_string_lookup</a>()</code> <br><code><a href="layer_text_vectorization.html">layer_text_vectorization</a>()</code> <br></p>
-<p>Other layers: <br><code><a href="Layer.html">Layer</a>()</code> <br><code><a href="layer_activation.html">layer_activation</a>()</code> <br><code><a href="layer_activation_elu.html">layer_activation_elu</a>()</code> <br><code><a href="layer_activation_leaky_relu.html">layer_activation_leaky_relu</a>()</code> <br><code><a href="layer_activation_parametric_relu.html">layer_activation_parametric_relu</a>()</code> <br><code><a href="layer_activation_relu.html">layer_activation_relu</a>()</code> <br><code><a href="layer_activation_softmax.html">layer_activation_softmax</a>()</code> <br><code><a href="layer_activity_regularization.html">layer_activity_regularization</a>()</code> <br><code><a href="layer_add.html">layer_add</a>()</code> <br><code><a href="layer_additive_attention.html">layer_additive_attention</a>()</code> <br><code><a href="layer_alpha_dropout.html">layer_alpha_dropout</a>()</code> <br><code><a href="layer_attention.html">layer_attention</a>()</code> <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_average.html">layer_average</a>()</code> <br><code><a href="layer_average_pooling_1d.html">layer_average_pooling_1d</a>()</code> <br><code><a href="layer_average_pooling_2d.html">layer_average_pooling_2d</a>()</code> <br><code><a href="layer_average_pooling_3d.html">layer_average_pooling_3d</a>()</code> <br><code><a href="layer_batch_normalization.html">layer_batch_normalization</a>()</code> <br><code><a href="layer_bidirectional.html">layer_bidirectional</a>()</code> <br><code><a href="layer_category_encoding.html">layer_category_encoding</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_concatenate.html">layer_concatenate</a>()</code> <br><code><a href="layer_conv_1d.html">layer_conv_1d</a>()</code> <br><code><a href="layer_conv_1d_transpose.html">layer_conv_1d_transpose</a>()</code> <br><code><a href="layer_conv_2d.html">layer_conv_2d</a>()</code> <br><code><a href="layer_conv_2d_transpose.html">layer_conv_2d_transpose</a>()</code> <br><code><a href="layer_conv_3d.html">layer_conv_3d</a>()</code> <br><code><a href="layer_conv_3d_transpose.html">layer_conv_3d_transpose</a>()</code> <br><code><a href="layer_conv_lstm_1d.html">layer_conv_lstm_1d</a>()</code> <br><code><a href="layer_conv_lstm_2d.html">layer_conv_lstm_2d</a>()</code> <br><code><a href="layer_conv_lstm_3d.html">layer_conv_lstm_3d</a>()</code> <br><code><a href="layer_cropping_1d.html">layer_cropping_1d</a>()</code> <br><code><a href="layer_cropping_2d.html">layer_cropping_2d</a>()</code> <br><code><a href="layer_cropping_3d.html">layer_cropping_3d</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_dense.html">layer_dense</a>()</code> <br><code><a href="layer_depthwise_conv_1d.html">layer_depthwise_conv_1d</a>()</code> <br><code><a href="layer_depthwise_conv_2d.html">layer_depthwise_conv_2d</a>()</code> <br><code><a href="layer_discretization.html">layer_discretization</a>()</code> <br><code><a href="layer_dot.html">layer_dot</a>()</code> <br><code><a href="layer_dropout.html">layer_dropout</a>()</code> <br><code><a href="layer_einsum_dense.html">layer_einsum_dense</a>()</code> <br><code><a href="layer_embedding.html">layer_embedding</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_feature_space.html">layer_feature_space</a>()</code> <br><code><a href="layer_flatten.html">layer_flatten</a>()</code> <br><code><a href="layer_flax_module_wrapper.html">layer_flax_module_wrapper</a>()</code> <br><code><a href="layer_gaussian_dropout.html">layer_gaussian_dropout</a>()</code> <br><code><a href="layer_gaussian_noise.html">layer_gaussian_noise</a>()</code> <br><code><a href="layer_global_average_pooling_1d.html">layer_global_average_pooling_1d</a>()</code> <br><code><a href="layer_global_average_pooling_2d.html">layer_global_average_pooling_2d</a>()</code> <br><code><a href="layer_global_average_pooling_3d.html">layer_global_average_pooling_3d</a>()</code> <br><code><a href="layer_global_max_pooling_1d.html">layer_global_max_pooling_1d</a>()</code> <br><code><a href="layer_global_max_pooling_2d.html">layer_global_max_pooling_2d</a>()</code> <br><code><a href="layer_global_max_pooling_3d.html">layer_global_max_pooling_3d</a>()</code> <br><code><a href="layer_group_normalization.html">layer_group_normalization</a>()</code> <br><code><a href="layer_group_query_attention.html">layer_group_query_attention</a>()</code> <br><code><a href="layer_gru.html">layer_gru</a>()</code> <br><code><a href="layer_hashed_crossing.html">layer_hashed_crossing</a>()</code> <br><code><a href="layer_hashing.html">layer_hashing</a>()</code> <br><code><a href="layer_identity.html">layer_identity</a>()</code> <br><code><a href="layer_integer_lookup.html">layer_integer_lookup</a>()</code> <br><code><a href="layer_jax_model_wrapper.html">layer_jax_model_wrapper</a>()</code> <br><code><a href="layer_lambda.html">layer_lambda</a>()</code> <br><code><a href="layer_layer_normalization.html">layer_layer_normalization</a>()</code> <br><code><a href="layer_lstm.html">layer_lstm</a>()</code> <br><code><a href="layer_masking.html">layer_masking</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_max_pooling_1d.html">layer_max_pooling_1d</a>()</code> <br><code><a href="layer_max_pooling_2d.html">layer_max_pooling_2d</a>()</code> <br><code><a href="layer_max_pooling_3d.html">layer_max_pooling_3d</a>()</code> <br><code><a href="layer_maximum.html">layer_maximum</a>()</code> <br><code><a href="layer_mel_spectrogram.html">layer_mel_spectrogram</a>()</code> <br><code><a href="layer_minimum.html">layer_minimum</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_multi_head_attention.html">layer_multi_head_attention</a>()</code> <br><code><a href="layer_multiply.html">layer_multiply</a>()</code> <br><code><a href="layer_normalization.html">layer_normalization</a>()</code> <br><code><a href="layer_permute.html">layer_permute</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_brightness.html">layer_random_brightness</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_contrast.html">layer_random_contrast</a>()</code> <br><code><a href="layer_random_crop.html">layer_random_crop</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_flip.html">layer_random_flip</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_rotation.html">layer_random_rotation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_random_translation.html">layer_random_translation</a>()</code> <br><code><a href="layer_random_zoom.html">layer_random_zoom</a>()</code> <br><code><a href="layer_repeat_vector.html">layer_repeat_vector</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_reshape.html">layer_reshape</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_rms_normalization.html">layer_rms_normalization</a>()</code> <br><code><a href="layer_rnn.html">layer_rnn</a>()</code> <br><code><a href="layer_separable_conv_1d.html">layer_separable_conv_1d</a>()</code> <br><code><a href="layer_separable_conv_2d.html">layer_separable_conv_2d</a>()</code> <br><code><a href="layer_simple_rnn.html">layer_simple_rnn</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br><code><a href="layer_spatial_dropout_1d.html">layer_spatial_dropout_1d</a>()</code> <br><code><a href="layer_spatial_dropout_2d.html">layer_spatial_dropout_2d</a>()</code> <br><code><a href="layer_spatial_dropout_3d.html">layer_spatial_dropout_3d</a>()</code> <br><code><a href="layer_spectral_normalization.html">layer_spectral_normalization</a>()</code> <br><code><a href="layer_stft_spectrogram.html">layer_stft_spectrogram</a>()</code> <br><code><a href="layer_string_lookup.html">layer_string_lookup</a>()</code> <br><code><a href="layer_subtract.html">layer_subtract</a>()</code> <br><code><a href="layer_text_vectorization.html">layer_text_vectorization</a>()</code> <br><code><a href="layer_tfsm.html">layer_tfsm</a>()</code> <br><code><a href="layer_time_distributed.html">layer_time_distributed</a>()</code> <br><code><a href="layer_torch_module_wrapper.html">layer_torch_module_wrapper</a>()</code> <br><code><a href="layer_unit_normalization.html">layer_unit_normalization</a>()</code> <br><code><a href="layer_upsampling_1d.html">layer_upsampling_1d</a>()</code> <br><code><a href="layer_upsampling_2d.html">layer_upsampling_2d</a>()</code> <br><code><a href="layer_upsampling_3d.html">layer_upsampling_3d</a>()</code> <br><code><a href="layer_zero_padding_1d.html">layer_zero_padding_1d</a>()</code> <br><code><a href="layer_zero_padding_2d.html">layer_zero_padding_2d</a>()</code> <br><code><a href="layer_zero_padding_3d.html">layer_zero_padding_3d</a>()</code> <br><code><a href="rnn_cell_gru.html">rnn_cell_gru</a>()</code> <br><code><a href="rnn_cell_lstm.html">rnn_cell_lstm</a>()</code> <br><code><a href="rnn_cell_simple.html">rnn_cell_simple</a>()</code> <br><code><a href="rnn_cells_stack.html">rnn_cells_stack</a>()</code> <br></p></div>
+    <div class="dont-index"><p>Other image preprocessing layers: <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_elastic_transform.html">layer_random_elastic_transform</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br></p>
+<p>Other preprocessing layers: <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_category_encoding.html">layer_category_encoding</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_discretization.html">layer_discretization</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_feature_space.html">layer_feature_space</a>()</code> <br><code><a href="layer_hashed_crossing.html">layer_hashed_crossing</a>()</code> <br><code><a href="layer_hashing.html">layer_hashing</a>()</code> <br><code><a href="layer_integer_lookup.html">layer_integer_lookup</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_mel_spectrogram.html">layer_mel_spectrogram</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_normalization.html">layer_normalization</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_brightness.html">layer_random_brightness</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_contrast.html">layer_random_contrast</a>()</code> <br><code><a href="layer_random_crop.html">layer_random_crop</a>()</code> <br><code><a href="layer_random_elastic_transform.html">layer_random_elastic_transform</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_flip.html">layer_random_flip</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_rotation.html">layer_random_rotation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_random_translation.html">layer_random_translation</a>()</code> <br><code><a href="layer_random_zoom.html">layer_random_zoom</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br><code><a href="layer_stft_spectrogram.html">layer_stft_spectrogram</a>()</code> <br><code><a href="layer_string_lookup.html">layer_string_lookup</a>()</code> <br><code><a href="layer_text_vectorization.html">layer_text_vectorization</a>()</code> <br></p>
+<p>Other layers: <br><code><a href="Layer.html">Layer</a>()</code> <br><code><a href="layer_activation.html">layer_activation</a>()</code> <br><code><a href="layer_activation_elu.html">layer_activation_elu</a>()</code> <br><code><a href="layer_activation_leaky_relu.html">layer_activation_leaky_relu</a>()</code> <br><code><a href="layer_activation_parametric_relu.html">layer_activation_parametric_relu</a>()</code> <br><code><a href="layer_activation_relu.html">layer_activation_relu</a>()</code> <br><code><a href="layer_activation_softmax.html">layer_activation_softmax</a>()</code> <br><code><a href="layer_activity_regularization.html">layer_activity_regularization</a>()</code> <br><code><a href="layer_add.html">layer_add</a>()</code> <br><code><a href="layer_additive_attention.html">layer_additive_attention</a>()</code> <br><code><a href="layer_alpha_dropout.html">layer_alpha_dropout</a>()</code> <br><code><a href="layer_attention.html">layer_attention</a>()</code> <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_average.html">layer_average</a>()</code> <br><code><a href="layer_average_pooling_1d.html">layer_average_pooling_1d</a>()</code> <br><code><a href="layer_average_pooling_2d.html">layer_average_pooling_2d</a>()</code> <br><code><a href="layer_average_pooling_3d.html">layer_average_pooling_3d</a>()</code> <br><code><a href="layer_batch_normalization.html">layer_batch_normalization</a>()</code> <br><code><a href="layer_bidirectional.html">layer_bidirectional</a>()</code> <br><code><a href="layer_category_encoding.html">layer_category_encoding</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_concatenate.html">layer_concatenate</a>()</code> <br><code><a href="layer_conv_1d.html">layer_conv_1d</a>()</code> <br><code><a href="layer_conv_1d_transpose.html">layer_conv_1d_transpose</a>()</code> <br><code><a href="layer_conv_2d.html">layer_conv_2d</a>()</code> <br><code><a href="layer_conv_2d_transpose.html">layer_conv_2d_transpose</a>()</code> <br><code><a href="layer_conv_3d.html">layer_conv_3d</a>()</code> <br><code><a href="layer_conv_3d_transpose.html">layer_conv_3d_transpose</a>()</code> <br><code><a href="layer_conv_lstm_1d.html">layer_conv_lstm_1d</a>()</code> <br><code><a href="layer_conv_lstm_2d.html">layer_conv_lstm_2d</a>()</code> <br><code><a href="layer_conv_lstm_3d.html">layer_conv_lstm_3d</a>()</code> <br><code><a href="layer_cropping_1d.html">layer_cropping_1d</a>()</code> <br><code><a href="layer_cropping_2d.html">layer_cropping_2d</a>()</code> <br><code><a href="layer_cropping_3d.html">layer_cropping_3d</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_dense.html">layer_dense</a>()</code> <br><code><a href="layer_depthwise_conv_1d.html">layer_depthwise_conv_1d</a>()</code> <br><code><a href="layer_depthwise_conv_2d.html">layer_depthwise_conv_2d</a>()</code> <br><code><a href="layer_discretization.html">layer_discretization</a>()</code> <br><code><a href="layer_dot.html">layer_dot</a>()</code> <br><code><a href="layer_dropout.html">layer_dropout</a>()</code> <br><code><a href="layer_einsum_dense.html">layer_einsum_dense</a>()</code> <br><code><a href="layer_embedding.html">layer_embedding</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_feature_space.html">layer_feature_space</a>()</code> <br><code><a href="layer_flatten.html">layer_flatten</a>()</code> <br><code><a href="layer_flax_module_wrapper.html">layer_flax_module_wrapper</a>()</code> <br><code><a href="layer_gaussian_dropout.html">layer_gaussian_dropout</a>()</code> <br><code><a href="layer_gaussian_noise.html">layer_gaussian_noise</a>()</code> <br><code><a href="layer_global_average_pooling_1d.html">layer_global_average_pooling_1d</a>()</code> <br><code><a href="layer_global_average_pooling_2d.html">layer_global_average_pooling_2d</a>()</code> <br><code><a href="layer_global_average_pooling_3d.html">layer_global_average_pooling_3d</a>()</code> <br><code><a href="layer_global_max_pooling_1d.html">layer_global_max_pooling_1d</a>()</code> <br><code><a href="layer_global_max_pooling_2d.html">layer_global_max_pooling_2d</a>()</code> <br><code><a href="layer_global_max_pooling_3d.html">layer_global_max_pooling_3d</a>()</code> <br><code><a href="layer_group_normalization.html">layer_group_normalization</a>()</code> <br><code><a href="layer_group_query_attention.html">layer_group_query_attention</a>()</code> <br><code><a href="layer_gru.html">layer_gru</a>()</code> <br><code><a href="layer_hashed_crossing.html">layer_hashed_crossing</a>()</code> <br><code><a href="layer_hashing.html">layer_hashing</a>()</code> <br><code><a href="layer_identity.html">layer_identity</a>()</code> <br><code><a href="layer_integer_lookup.html">layer_integer_lookup</a>()</code> <br><code><a href="layer_jax_model_wrapper.html">layer_jax_model_wrapper</a>()</code> <br><code><a href="layer_lambda.html">layer_lambda</a>()</code> <br><code><a href="layer_layer_normalization.html">layer_layer_normalization</a>()</code> <br><code><a href="layer_lstm.html">layer_lstm</a>()</code> <br><code><a href="layer_masking.html">layer_masking</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_max_pooling_1d.html">layer_max_pooling_1d</a>()</code> <br><code><a href="layer_max_pooling_2d.html">layer_max_pooling_2d</a>()</code> <br><code><a href="layer_max_pooling_3d.html">layer_max_pooling_3d</a>()</code> <br><code><a href="layer_maximum.html">layer_maximum</a>()</code> <br><code><a href="layer_mel_spectrogram.html">layer_mel_spectrogram</a>()</code> <br><code><a href="layer_minimum.html">layer_minimum</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_multi_head_attention.html">layer_multi_head_attention</a>()</code> <br><code><a href="layer_multiply.html">layer_multiply</a>()</code> <br><code><a href="layer_normalization.html">layer_normalization</a>()</code> <br><code><a href="layer_permute.html">layer_permute</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_brightness.html">layer_random_brightness</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_contrast.html">layer_random_contrast</a>()</code> <br><code><a href="layer_random_crop.html">layer_random_crop</a>()</code> <br><code><a href="layer_random_elastic_transform.html">layer_random_elastic_transform</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_flip.html">layer_random_flip</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_rotation.html">layer_random_rotation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_random_translation.html">layer_random_translation</a>()</code> <br><code><a href="layer_random_zoom.html">layer_random_zoom</a>()</code> <br><code><a href="layer_repeat_vector.html">layer_repeat_vector</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_reshape.html">layer_reshape</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_rms_normalization.html">layer_rms_normalization</a>()</code> <br><code><a href="layer_rnn.html">layer_rnn</a>()</code> <br><code><a href="layer_separable_conv_1d.html">layer_separable_conv_1d</a>()</code> <br><code><a href="layer_separable_conv_2d.html">layer_separable_conv_2d</a>()</code> <br><code><a href="layer_simple_rnn.html">layer_simple_rnn</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br><code><a href="layer_spatial_dropout_1d.html">layer_spatial_dropout_1d</a>()</code> <br><code><a href="layer_spatial_dropout_2d.html">layer_spatial_dropout_2d</a>()</code> <br><code><a href="layer_spatial_dropout_3d.html">layer_spatial_dropout_3d</a>()</code> <br><code><a href="layer_spectral_normalization.html">layer_spectral_normalization</a>()</code> <br><code><a href="layer_stft_spectrogram.html">layer_stft_spectrogram</a>()</code> <br><code><a href="layer_string_lookup.html">layer_string_lookup</a>()</code> <br><code><a href="layer_subtract.html">layer_subtract</a>()</code> <br><code><a href="layer_text_vectorization.html">layer_text_vectorization</a>()</code> <br><code><a href="layer_tfsm.html">layer_tfsm</a>()</code> <br><code><a href="layer_time_distributed.html">layer_time_distributed</a>()</code> <br><code><a href="layer_torch_module_wrapper.html">layer_torch_module_wrapper</a>()</code> <br><code><a href="layer_unit_normalization.html">layer_unit_normalization</a>()</code> <br><code><a href="layer_upsampling_1d.html">layer_upsampling_1d</a>()</code> <br><code><a href="layer_upsampling_2d.html">layer_upsampling_2d</a>()</code> <br><code><a href="layer_upsampling_3d.html">layer_upsampling_3d</a>()</code> <br><code><a href="layer_zero_padding_1d.html">layer_zero_padding_1d</a>()</code> <br><code><a href="layer_zero_padding_2d.html">layer_zero_padding_2d</a>()</code> <br><code><a href="layer_zero_padding_3d.html">layer_zero_padding_3d</a>()</code> <br><code><a href="rnn_cell_gru.html">rnn_cell_gru</a>()</code> <br><code><a href="rnn_cell_lstm.html">rnn_cell_lstm</a>()</code> <br><code><a href="rnn_cell_simple.html">rnn_cell_simple</a>()</code> <br><code><a href="rnn_cells_stack.html">rnn_cells_stack</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -140,7 +140,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/layer_random_sharpness.html b/docs/dev/reference/layer_random_sharpness.html
index dfa4ee859..df1565536 100644
--- a/docs/dev/reference/layer_random_sharpness.html
+++ b/docs/dev/reference/layer_random_sharpness.html
@@ -12,7 +12,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -122,9 +122,9 @@ <h2 id="arguments">Arguments<a class="anchor" aria-label="anchor" href="#argumen
 </dl></div>
     <div class="section level2">
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
-    <div class="dont-index"><p>Other image preprocessing layers: <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br></p>
-<p>Other preprocessing layers: <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_category_encoding.html">layer_category_encoding</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_discretization.html">layer_discretization</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_feature_space.html">layer_feature_space</a>()</code> <br><code><a href="layer_hashed_crossing.html">layer_hashed_crossing</a>()</code> <br><code><a href="layer_hashing.html">layer_hashing</a>()</code> <br><code><a href="layer_integer_lookup.html">layer_integer_lookup</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_mel_spectrogram.html">layer_mel_spectrogram</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_normalization.html">layer_normalization</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_brightness.html">layer_random_brightness</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_contrast.html">layer_random_contrast</a>()</code> <br><code><a href="layer_random_crop.html">layer_random_crop</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_flip.html">layer_random_flip</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_rotation.html">layer_random_rotation</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_random_translation.html">layer_random_translation</a>()</code> <br><code><a href="layer_random_zoom.html">layer_random_zoom</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br><code><a href="layer_stft_spectrogram.html">layer_stft_spectrogram</a>()</code> <br><code><a href="layer_string_lookup.html">layer_string_lookup</a>()</code> <br><code><a href="layer_text_vectorization.html">layer_text_vectorization</a>()</code> <br></p>
-<p>Other layers: <br><code><a href="Layer.html">Layer</a>()</code> <br><code><a href="layer_activation.html">layer_activation</a>()</code> <br><code><a href="layer_activation_elu.html">layer_activation_elu</a>()</code> <br><code><a href="layer_activation_leaky_relu.html">layer_activation_leaky_relu</a>()</code> <br><code><a href="layer_activation_parametric_relu.html">layer_activation_parametric_relu</a>()</code> <br><code><a href="layer_activation_relu.html">layer_activation_relu</a>()</code> <br><code><a href="layer_activation_softmax.html">layer_activation_softmax</a>()</code> <br><code><a href="layer_activity_regularization.html">layer_activity_regularization</a>()</code> <br><code><a href="layer_add.html">layer_add</a>()</code> <br><code><a href="layer_additive_attention.html">layer_additive_attention</a>()</code> <br><code><a href="layer_alpha_dropout.html">layer_alpha_dropout</a>()</code> <br><code><a href="layer_attention.html">layer_attention</a>()</code> <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_average.html">layer_average</a>()</code> <br><code><a href="layer_average_pooling_1d.html">layer_average_pooling_1d</a>()</code> <br><code><a href="layer_average_pooling_2d.html">layer_average_pooling_2d</a>()</code> <br><code><a href="layer_average_pooling_3d.html">layer_average_pooling_3d</a>()</code> <br><code><a href="layer_batch_normalization.html">layer_batch_normalization</a>()</code> <br><code><a href="layer_bidirectional.html">layer_bidirectional</a>()</code> <br><code><a href="layer_category_encoding.html">layer_category_encoding</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_concatenate.html">layer_concatenate</a>()</code> <br><code><a href="layer_conv_1d.html">layer_conv_1d</a>()</code> <br><code><a href="layer_conv_1d_transpose.html">layer_conv_1d_transpose</a>()</code> <br><code><a href="layer_conv_2d.html">layer_conv_2d</a>()</code> <br><code><a href="layer_conv_2d_transpose.html">layer_conv_2d_transpose</a>()</code> <br><code><a href="layer_conv_3d.html">layer_conv_3d</a>()</code> <br><code><a href="layer_conv_3d_transpose.html">layer_conv_3d_transpose</a>()</code> <br><code><a href="layer_conv_lstm_1d.html">layer_conv_lstm_1d</a>()</code> <br><code><a href="layer_conv_lstm_2d.html">layer_conv_lstm_2d</a>()</code> <br><code><a href="layer_conv_lstm_3d.html">layer_conv_lstm_3d</a>()</code> <br><code><a href="layer_cropping_1d.html">layer_cropping_1d</a>()</code> <br><code><a href="layer_cropping_2d.html">layer_cropping_2d</a>()</code> <br><code><a href="layer_cropping_3d.html">layer_cropping_3d</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_dense.html">layer_dense</a>()</code> <br><code><a href="layer_depthwise_conv_1d.html">layer_depthwise_conv_1d</a>()</code> <br><code><a href="layer_depthwise_conv_2d.html">layer_depthwise_conv_2d</a>()</code> <br><code><a href="layer_discretization.html">layer_discretization</a>()</code> <br><code><a href="layer_dot.html">layer_dot</a>()</code> <br><code><a href="layer_dropout.html">layer_dropout</a>()</code> <br><code><a href="layer_einsum_dense.html">layer_einsum_dense</a>()</code> <br><code><a href="layer_embedding.html">layer_embedding</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_feature_space.html">layer_feature_space</a>()</code> <br><code><a href="layer_flatten.html">layer_flatten</a>()</code> <br><code><a href="layer_flax_module_wrapper.html">layer_flax_module_wrapper</a>()</code> <br><code><a href="layer_gaussian_dropout.html">layer_gaussian_dropout</a>()</code> <br><code><a href="layer_gaussian_noise.html">layer_gaussian_noise</a>()</code> <br><code><a href="layer_global_average_pooling_1d.html">layer_global_average_pooling_1d</a>()</code> <br><code><a href="layer_global_average_pooling_2d.html">layer_global_average_pooling_2d</a>()</code> <br><code><a href="layer_global_average_pooling_3d.html">layer_global_average_pooling_3d</a>()</code> <br><code><a href="layer_global_max_pooling_1d.html">layer_global_max_pooling_1d</a>()</code> <br><code><a href="layer_global_max_pooling_2d.html">layer_global_max_pooling_2d</a>()</code> <br><code><a href="layer_global_max_pooling_3d.html">layer_global_max_pooling_3d</a>()</code> <br><code><a href="layer_group_normalization.html">layer_group_normalization</a>()</code> <br><code><a href="layer_group_query_attention.html">layer_group_query_attention</a>()</code> <br><code><a href="layer_gru.html">layer_gru</a>()</code> <br><code><a href="layer_hashed_crossing.html">layer_hashed_crossing</a>()</code> <br><code><a href="layer_hashing.html">layer_hashing</a>()</code> <br><code><a href="layer_identity.html">layer_identity</a>()</code> <br><code><a href="layer_integer_lookup.html">layer_integer_lookup</a>()</code> <br><code><a href="layer_jax_model_wrapper.html">layer_jax_model_wrapper</a>()</code> <br><code><a href="layer_lambda.html">layer_lambda</a>()</code> <br><code><a href="layer_layer_normalization.html">layer_layer_normalization</a>()</code> <br><code><a href="layer_lstm.html">layer_lstm</a>()</code> <br><code><a href="layer_masking.html">layer_masking</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_max_pooling_1d.html">layer_max_pooling_1d</a>()</code> <br><code><a href="layer_max_pooling_2d.html">layer_max_pooling_2d</a>()</code> <br><code><a href="layer_max_pooling_3d.html">layer_max_pooling_3d</a>()</code> <br><code><a href="layer_maximum.html">layer_maximum</a>()</code> <br><code><a href="layer_mel_spectrogram.html">layer_mel_spectrogram</a>()</code> <br><code><a href="layer_minimum.html">layer_minimum</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_multi_head_attention.html">layer_multi_head_attention</a>()</code> <br><code><a href="layer_multiply.html">layer_multiply</a>()</code> <br><code><a href="layer_normalization.html">layer_normalization</a>()</code> <br><code><a href="layer_permute.html">layer_permute</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_brightness.html">layer_random_brightness</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_contrast.html">layer_random_contrast</a>()</code> <br><code><a href="layer_random_crop.html">layer_random_crop</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_flip.html">layer_random_flip</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_rotation.html">layer_random_rotation</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_random_translation.html">layer_random_translation</a>()</code> <br><code><a href="layer_random_zoom.html">layer_random_zoom</a>()</code> <br><code><a href="layer_repeat_vector.html">layer_repeat_vector</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_reshape.html">layer_reshape</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_rms_normalization.html">layer_rms_normalization</a>()</code> <br><code><a href="layer_rnn.html">layer_rnn</a>()</code> <br><code><a href="layer_separable_conv_1d.html">layer_separable_conv_1d</a>()</code> <br><code><a href="layer_separable_conv_2d.html">layer_separable_conv_2d</a>()</code> <br><code><a href="layer_simple_rnn.html">layer_simple_rnn</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br><code><a href="layer_spatial_dropout_1d.html">layer_spatial_dropout_1d</a>()</code> <br><code><a href="layer_spatial_dropout_2d.html">layer_spatial_dropout_2d</a>()</code> <br><code><a href="layer_spatial_dropout_3d.html">layer_spatial_dropout_3d</a>()</code> <br><code><a href="layer_spectral_normalization.html">layer_spectral_normalization</a>()</code> <br><code><a href="layer_stft_spectrogram.html">layer_stft_spectrogram</a>()</code> <br><code><a href="layer_string_lookup.html">layer_string_lookup</a>()</code> <br><code><a href="layer_subtract.html">layer_subtract</a>()</code> <br><code><a href="layer_text_vectorization.html">layer_text_vectorization</a>()</code> <br><code><a href="layer_tfsm.html">layer_tfsm</a>()</code> <br><code><a href="layer_time_distributed.html">layer_time_distributed</a>()</code> <br><code><a href="layer_torch_module_wrapper.html">layer_torch_module_wrapper</a>()</code> <br><code><a href="layer_unit_normalization.html">layer_unit_normalization</a>()</code> <br><code><a href="layer_upsampling_1d.html">layer_upsampling_1d</a>()</code> <br><code><a href="layer_upsampling_2d.html">layer_upsampling_2d</a>()</code> <br><code><a href="layer_upsampling_3d.html">layer_upsampling_3d</a>()</code> <br><code><a href="layer_zero_padding_1d.html">layer_zero_padding_1d</a>()</code> <br><code><a href="layer_zero_padding_2d.html">layer_zero_padding_2d</a>()</code> <br><code><a href="layer_zero_padding_3d.html">layer_zero_padding_3d</a>()</code> <br><code><a href="rnn_cell_gru.html">rnn_cell_gru</a>()</code> <br><code><a href="rnn_cell_lstm.html">rnn_cell_lstm</a>()</code> <br><code><a href="rnn_cell_simple.html">rnn_cell_simple</a>()</code> <br><code><a href="rnn_cells_stack.html">rnn_cells_stack</a>()</code> <br></p></div>
+    <div class="dont-index"><p>Other image preprocessing layers: <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_elastic_transform.html">layer_random_elastic_transform</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br></p>
+<p>Other preprocessing layers: <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_category_encoding.html">layer_category_encoding</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_discretization.html">layer_discretization</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_feature_space.html">layer_feature_space</a>()</code> <br><code><a href="layer_hashed_crossing.html">layer_hashed_crossing</a>()</code> <br><code><a href="layer_hashing.html">layer_hashing</a>()</code> <br><code><a href="layer_integer_lookup.html">layer_integer_lookup</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_mel_spectrogram.html">layer_mel_spectrogram</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_normalization.html">layer_normalization</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_brightness.html">layer_random_brightness</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_contrast.html">layer_random_contrast</a>()</code> <br><code><a href="layer_random_crop.html">layer_random_crop</a>()</code> <br><code><a href="layer_random_elastic_transform.html">layer_random_elastic_transform</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_flip.html">layer_random_flip</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_rotation.html">layer_random_rotation</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_random_translation.html">layer_random_translation</a>()</code> <br><code><a href="layer_random_zoom.html">layer_random_zoom</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br><code><a href="layer_stft_spectrogram.html">layer_stft_spectrogram</a>()</code> <br><code><a href="layer_string_lookup.html">layer_string_lookup</a>()</code> <br><code><a href="layer_text_vectorization.html">layer_text_vectorization</a>()</code> <br></p>
+<p>Other layers: <br><code><a href="Layer.html">Layer</a>()</code> <br><code><a href="layer_activation.html">layer_activation</a>()</code> <br><code><a href="layer_activation_elu.html">layer_activation_elu</a>()</code> <br><code><a href="layer_activation_leaky_relu.html">layer_activation_leaky_relu</a>()</code> <br><code><a href="layer_activation_parametric_relu.html">layer_activation_parametric_relu</a>()</code> <br><code><a href="layer_activation_relu.html">layer_activation_relu</a>()</code> <br><code><a href="layer_activation_softmax.html">layer_activation_softmax</a>()</code> <br><code><a href="layer_activity_regularization.html">layer_activity_regularization</a>()</code> <br><code><a href="layer_add.html">layer_add</a>()</code> <br><code><a href="layer_additive_attention.html">layer_additive_attention</a>()</code> <br><code><a href="layer_alpha_dropout.html">layer_alpha_dropout</a>()</code> <br><code><a href="layer_attention.html">layer_attention</a>()</code> <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_average.html">layer_average</a>()</code> <br><code><a href="layer_average_pooling_1d.html">layer_average_pooling_1d</a>()</code> <br><code><a href="layer_average_pooling_2d.html">layer_average_pooling_2d</a>()</code> <br><code><a href="layer_average_pooling_3d.html">layer_average_pooling_3d</a>()</code> <br><code><a href="layer_batch_normalization.html">layer_batch_normalization</a>()</code> <br><code><a href="layer_bidirectional.html">layer_bidirectional</a>()</code> <br><code><a href="layer_category_encoding.html">layer_category_encoding</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_concatenate.html">layer_concatenate</a>()</code> <br><code><a href="layer_conv_1d.html">layer_conv_1d</a>()</code> <br><code><a href="layer_conv_1d_transpose.html">layer_conv_1d_transpose</a>()</code> <br><code><a href="layer_conv_2d.html">layer_conv_2d</a>()</code> <br><code><a href="layer_conv_2d_transpose.html">layer_conv_2d_transpose</a>()</code> <br><code><a href="layer_conv_3d.html">layer_conv_3d</a>()</code> <br><code><a href="layer_conv_3d_transpose.html">layer_conv_3d_transpose</a>()</code> <br><code><a href="layer_conv_lstm_1d.html">layer_conv_lstm_1d</a>()</code> <br><code><a href="layer_conv_lstm_2d.html">layer_conv_lstm_2d</a>()</code> <br><code><a href="layer_conv_lstm_3d.html">layer_conv_lstm_3d</a>()</code> <br><code><a href="layer_cropping_1d.html">layer_cropping_1d</a>()</code> <br><code><a href="layer_cropping_2d.html">layer_cropping_2d</a>()</code> <br><code><a href="layer_cropping_3d.html">layer_cropping_3d</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_dense.html">layer_dense</a>()</code> <br><code><a href="layer_depthwise_conv_1d.html">layer_depthwise_conv_1d</a>()</code> <br><code><a href="layer_depthwise_conv_2d.html">layer_depthwise_conv_2d</a>()</code> <br><code><a href="layer_discretization.html">layer_discretization</a>()</code> <br><code><a href="layer_dot.html">layer_dot</a>()</code> <br><code><a href="layer_dropout.html">layer_dropout</a>()</code> <br><code><a href="layer_einsum_dense.html">layer_einsum_dense</a>()</code> <br><code><a href="layer_embedding.html">layer_embedding</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_feature_space.html">layer_feature_space</a>()</code> <br><code><a href="layer_flatten.html">layer_flatten</a>()</code> <br><code><a href="layer_flax_module_wrapper.html">layer_flax_module_wrapper</a>()</code> <br><code><a href="layer_gaussian_dropout.html">layer_gaussian_dropout</a>()</code> <br><code><a href="layer_gaussian_noise.html">layer_gaussian_noise</a>()</code> <br><code><a href="layer_global_average_pooling_1d.html">layer_global_average_pooling_1d</a>()</code> <br><code><a href="layer_global_average_pooling_2d.html">layer_global_average_pooling_2d</a>()</code> <br><code><a href="layer_global_average_pooling_3d.html">layer_global_average_pooling_3d</a>()</code> <br><code><a href="layer_global_max_pooling_1d.html">layer_global_max_pooling_1d</a>()</code> <br><code><a href="layer_global_max_pooling_2d.html">layer_global_max_pooling_2d</a>()</code> <br><code><a href="layer_global_max_pooling_3d.html">layer_global_max_pooling_3d</a>()</code> <br><code><a href="layer_group_normalization.html">layer_group_normalization</a>()</code> <br><code><a href="layer_group_query_attention.html">layer_group_query_attention</a>()</code> <br><code><a href="layer_gru.html">layer_gru</a>()</code> <br><code><a href="layer_hashed_crossing.html">layer_hashed_crossing</a>()</code> <br><code><a href="layer_hashing.html">layer_hashing</a>()</code> <br><code><a href="layer_identity.html">layer_identity</a>()</code> <br><code><a href="layer_integer_lookup.html">layer_integer_lookup</a>()</code> <br><code><a href="layer_jax_model_wrapper.html">layer_jax_model_wrapper</a>()</code> <br><code><a href="layer_lambda.html">layer_lambda</a>()</code> <br><code><a href="layer_layer_normalization.html">layer_layer_normalization</a>()</code> <br><code><a href="layer_lstm.html">layer_lstm</a>()</code> <br><code><a href="layer_masking.html">layer_masking</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_max_pooling_1d.html">layer_max_pooling_1d</a>()</code> <br><code><a href="layer_max_pooling_2d.html">layer_max_pooling_2d</a>()</code> <br><code><a href="layer_max_pooling_3d.html">layer_max_pooling_3d</a>()</code> <br><code><a href="layer_maximum.html">layer_maximum</a>()</code> <br><code><a href="layer_mel_spectrogram.html">layer_mel_spectrogram</a>()</code> <br><code><a href="layer_minimum.html">layer_minimum</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_multi_head_attention.html">layer_multi_head_attention</a>()</code> <br><code><a href="layer_multiply.html">layer_multiply</a>()</code> <br><code><a href="layer_normalization.html">layer_normalization</a>()</code> <br><code><a href="layer_permute.html">layer_permute</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_brightness.html">layer_random_brightness</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_contrast.html">layer_random_contrast</a>()</code> <br><code><a href="layer_random_crop.html">layer_random_crop</a>()</code> <br><code><a href="layer_random_elastic_transform.html">layer_random_elastic_transform</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_flip.html">layer_random_flip</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_rotation.html">layer_random_rotation</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_random_translation.html">layer_random_translation</a>()</code> <br><code><a href="layer_random_zoom.html">layer_random_zoom</a>()</code> <br><code><a href="layer_repeat_vector.html">layer_repeat_vector</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_reshape.html">layer_reshape</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_rms_normalization.html">layer_rms_normalization</a>()</code> <br><code><a href="layer_rnn.html">layer_rnn</a>()</code> <br><code><a href="layer_separable_conv_1d.html">layer_separable_conv_1d</a>()</code> <br><code><a href="layer_separable_conv_2d.html">layer_separable_conv_2d</a>()</code> <br><code><a href="layer_simple_rnn.html">layer_simple_rnn</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br><code><a href="layer_spatial_dropout_1d.html">layer_spatial_dropout_1d</a>()</code> <br><code><a href="layer_spatial_dropout_2d.html">layer_spatial_dropout_2d</a>()</code> <br><code><a href="layer_spatial_dropout_3d.html">layer_spatial_dropout_3d</a>()</code> <br><code><a href="layer_spectral_normalization.html">layer_spectral_normalization</a>()</code> <br><code><a href="layer_stft_spectrogram.html">layer_stft_spectrogram</a>()</code> <br><code><a href="layer_string_lookup.html">layer_string_lookup</a>()</code> <br><code><a href="layer_subtract.html">layer_subtract</a>()</code> <br><code><a href="layer_text_vectorization.html">layer_text_vectorization</a>()</code> <br><code><a href="layer_tfsm.html">layer_tfsm</a>()</code> <br><code><a href="layer_time_distributed.html">layer_time_distributed</a>()</code> <br><code><a href="layer_torch_module_wrapper.html">layer_torch_module_wrapper</a>()</code> <br><code><a href="layer_unit_normalization.html">layer_unit_normalization</a>()</code> <br><code><a href="layer_upsampling_1d.html">layer_upsampling_1d</a>()</code> <br><code><a href="layer_upsampling_2d.html">layer_upsampling_2d</a>()</code> <br><code><a href="layer_upsampling_3d.html">layer_upsampling_3d</a>()</code> <br><code><a href="layer_zero_padding_1d.html">layer_zero_padding_1d</a>()</code> <br><code><a href="layer_zero_padding_2d.html">layer_zero_padding_2d</a>()</code> <br><code><a href="layer_zero_padding_3d.html">layer_zero_padding_3d</a>()</code> <br><code><a href="rnn_cell_gru.html">rnn_cell_gru</a>()</code> <br><code><a href="rnn_cell_lstm.html">rnn_cell_lstm</a>()</code> <br><code><a href="rnn_cell_simple.html">rnn_cell_simple</a>()</code> <br><code><a href="rnn_cells_stack.html">rnn_cells_stack</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -136,7 +136,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/layer_random_shear.html b/docs/dev/reference/layer_random_shear.html
index 0d7f9174c..680cd2542 100644
--- a/docs/dev/reference/layer_random_shear.html
+++ b/docs/dev/reference/layer_random_shear.html
@@ -18,7 +18,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -162,9 +162,9 @@ <h2 id="arguments">Arguments<a class="anchor" aria-label="anchor" href="#argumen
 </dl></div>
     <div class="section level2">
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
-    <div class="dont-index"><p>Other image preprocessing layers: <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br></p>
-<p>Other preprocessing layers: <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_category_encoding.html">layer_category_encoding</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_discretization.html">layer_discretization</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_feature_space.html">layer_feature_space</a>()</code> <br><code><a href="layer_hashed_crossing.html">layer_hashed_crossing</a>()</code> <br><code><a href="layer_hashing.html">layer_hashing</a>()</code> <br><code><a href="layer_integer_lookup.html">layer_integer_lookup</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_mel_spectrogram.html">layer_mel_spectrogram</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_normalization.html">layer_normalization</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_brightness.html">layer_random_brightness</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_contrast.html">layer_random_contrast</a>()</code> <br><code><a href="layer_random_crop.html">layer_random_crop</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_flip.html">layer_random_flip</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_rotation.html">layer_random_rotation</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_translation.html">layer_random_translation</a>()</code> <br><code><a href="layer_random_zoom.html">layer_random_zoom</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br><code><a href="layer_stft_spectrogram.html">layer_stft_spectrogram</a>()</code> <br><code><a href="layer_string_lookup.html">layer_string_lookup</a>()</code> <br><code><a href="layer_text_vectorization.html">layer_text_vectorization</a>()</code> <br></p>
-<p>Other layers: <br><code><a href="Layer.html">Layer</a>()</code> <br><code><a href="layer_activation.html">layer_activation</a>()</code> <br><code><a href="layer_activation_elu.html">layer_activation_elu</a>()</code> <br><code><a href="layer_activation_leaky_relu.html">layer_activation_leaky_relu</a>()</code> <br><code><a href="layer_activation_parametric_relu.html">layer_activation_parametric_relu</a>()</code> <br><code><a href="layer_activation_relu.html">layer_activation_relu</a>()</code> <br><code><a href="layer_activation_softmax.html">layer_activation_softmax</a>()</code> <br><code><a href="layer_activity_regularization.html">layer_activity_regularization</a>()</code> <br><code><a href="layer_add.html">layer_add</a>()</code> <br><code><a href="layer_additive_attention.html">layer_additive_attention</a>()</code> <br><code><a href="layer_alpha_dropout.html">layer_alpha_dropout</a>()</code> <br><code><a href="layer_attention.html">layer_attention</a>()</code> <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_average.html">layer_average</a>()</code> <br><code><a href="layer_average_pooling_1d.html">layer_average_pooling_1d</a>()</code> <br><code><a href="layer_average_pooling_2d.html">layer_average_pooling_2d</a>()</code> <br><code><a href="layer_average_pooling_3d.html">layer_average_pooling_3d</a>()</code> <br><code><a href="layer_batch_normalization.html">layer_batch_normalization</a>()</code> <br><code><a href="layer_bidirectional.html">layer_bidirectional</a>()</code> <br><code><a href="layer_category_encoding.html">layer_category_encoding</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_concatenate.html">layer_concatenate</a>()</code> <br><code><a href="layer_conv_1d.html">layer_conv_1d</a>()</code> <br><code><a href="layer_conv_1d_transpose.html">layer_conv_1d_transpose</a>()</code> <br><code><a href="layer_conv_2d.html">layer_conv_2d</a>()</code> <br><code><a href="layer_conv_2d_transpose.html">layer_conv_2d_transpose</a>()</code> <br><code><a href="layer_conv_3d.html">layer_conv_3d</a>()</code> <br><code><a href="layer_conv_3d_transpose.html">layer_conv_3d_transpose</a>()</code> <br><code><a href="layer_conv_lstm_1d.html">layer_conv_lstm_1d</a>()</code> <br><code><a href="layer_conv_lstm_2d.html">layer_conv_lstm_2d</a>()</code> <br><code><a href="layer_conv_lstm_3d.html">layer_conv_lstm_3d</a>()</code> <br><code><a href="layer_cropping_1d.html">layer_cropping_1d</a>()</code> <br><code><a href="layer_cropping_2d.html">layer_cropping_2d</a>()</code> <br><code><a href="layer_cropping_3d.html">layer_cropping_3d</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_dense.html">layer_dense</a>()</code> <br><code><a href="layer_depthwise_conv_1d.html">layer_depthwise_conv_1d</a>()</code> <br><code><a href="layer_depthwise_conv_2d.html">layer_depthwise_conv_2d</a>()</code> <br><code><a href="layer_discretization.html">layer_discretization</a>()</code> <br><code><a href="layer_dot.html">layer_dot</a>()</code> <br><code><a href="layer_dropout.html">layer_dropout</a>()</code> <br><code><a href="layer_einsum_dense.html">layer_einsum_dense</a>()</code> <br><code><a href="layer_embedding.html">layer_embedding</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_feature_space.html">layer_feature_space</a>()</code> <br><code><a href="layer_flatten.html">layer_flatten</a>()</code> <br><code><a href="layer_flax_module_wrapper.html">layer_flax_module_wrapper</a>()</code> <br><code><a href="layer_gaussian_dropout.html">layer_gaussian_dropout</a>()</code> <br><code><a href="layer_gaussian_noise.html">layer_gaussian_noise</a>()</code> <br><code><a href="layer_global_average_pooling_1d.html">layer_global_average_pooling_1d</a>()</code> <br><code><a href="layer_global_average_pooling_2d.html">layer_global_average_pooling_2d</a>()</code> <br><code><a href="layer_global_average_pooling_3d.html">layer_global_average_pooling_3d</a>()</code> <br><code><a href="layer_global_max_pooling_1d.html">layer_global_max_pooling_1d</a>()</code> <br><code><a href="layer_global_max_pooling_2d.html">layer_global_max_pooling_2d</a>()</code> <br><code><a href="layer_global_max_pooling_3d.html">layer_global_max_pooling_3d</a>()</code> <br><code><a href="layer_group_normalization.html">layer_group_normalization</a>()</code> <br><code><a href="layer_group_query_attention.html">layer_group_query_attention</a>()</code> <br><code><a href="layer_gru.html">layer_gru</a>()</code> <br><code><a href="layer_hashed_crossing.html">layer_hashed_crossing</a>()</code> <br><code><a href="layer_hashing.html">layer_hashing</a>()</code> <br><code><a href="layer_identity.html">layer_identity</a>()</code> <br><code><a href="layer_integer_lookup.html">layer_integer_lookup</a>()</code> <br><code><a href="layer_jax_model_wrapper.html">layer_jax_model_wrapper</a>()</code> <br><code><a href="layer_lambda.html">layer_lambda</a>()</code> <br><code><a href="layer_layer_normalization.html">layer_layer_normalization</a>()</code> <br><code><a href="layer_lstm.html">layer_lstm</a>()</code> <br><code><a href="layer_masking.html">layer_masking</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_max_pooling_1d.html">layer_max_pooling_1d</a>()</code> <br><code><a href="layer_max_pooling_2d.html">layer_max_pooling_2d</a>()</code> <br><code><a href="layer_max_pooling_3d.html">layer_max_pooling_3d</a>()</code> <br><code><a href="layer_maximum.html">layer_maximum</a>()</code> <br><code><a href="layer_mel_spectrogram.html">layer_mel_spectrogram</a>()</code> <br><code><a href="layer_minimum.html">layer_minimum</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_multi_head_attention.html">layer_multi_head_attention</a>()</code> <br><code><a href="layer_multiply.html">layer_multiply</a>()</code> <br><code><a href="layer_normalization.html">layer_normalization</a>()</code> <br><code><a href="layer_permute.html">layer_permute</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_brightness.html">layer_random_brightness</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_contrast.html">layer_random_contrast</a>()</code> <br><code><a href="layer_random_crop.html">layer_random_crop</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_flip.html">layer_random_flip</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_rotation.html">layer_random_rotation</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_translation.html">layer_random_translation</a>()</code> <br><code><a href="layer_random_zoom.html">layer_random_zoom</a>()</code> <br><code><a href="layer_repeat_vector.html">layer_repeat_vector</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_reshape.html">layer_reshape</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_rms_normalization.html">layer_rms_normalization</a>()</code> <br><code><a href="layer_rnn.html">layer_rnn</a>()</code> <br><code><a href="layer_separable_conv_1d.html">layer_separable_conv_1d</a>()</code> <br><code><a href="layer_separable_conv_2d.html">layer_separable_conv_2d</a>()</code> <br><code><a href="layer_simple_rnn.html">layer_simple_rnn</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br><code><a href="layer_spatial_dropout_1d.html">layer_spatial_dropout_1d</a>()</code> <br><code><a href="layer_spatial_dropout_2d.html">layer_spatial_dropout_2d</a>()</code> <br><code><a href="layer_spatial_dropout_3d.html">layer_spatial_dropout_3d</a>()</code> <br><code><a href="layer_spectral_normalization.html">layer_spectral_normalization</a>()</code> <br><code><a href="layer_stft_spectrogram.html">layer_stft_spectrogram</a>()</code> <br><code><a href="layer_string_lookup.html">layer_string_lookup</a>()</code> <br><code><a href="layer_subtract.html">layer_subtract</a>()</code> <br><code><a href="layer_text_vectorization.html">layer_text_vectorization</a>()</code> <br><code><a href="layer_tfsm.html">layer_tfsm</a>()</code> <br><code><a href="layer_time_distributed.html">layer_time_distributed</a>()</code> <br><code><a href="layer_torch_module_wrapper.html">layer_torch_module_wrapper</a>()</code> <br><code><a href="layer_unit_normalization.html">layer_unit_normalization</a>()</code> <br><code><a href="layer_upsampling_1d.html">layer_upsampling_1d</a>()</code> <br><code><a href="layer_upsampling_2d.html">layer_upsampling_2d</a>()</code> <br><code><a href="layer_upsampling_3d.html">layer_upsampling_3d</a>()</code> <br><code><a href="layer_zero_padding_1d.html">layer_zero_padding_1d</a>()</code> <br><code><a href="layer_zero_padding_2d.html">layer_zero_padding_2d</a>()</code> <br><code><a href="layer_zero_padding_3d.html">layer_zero_padding_3d</a>()</code> <br><code><a href="rnn_cell_gru.html">rnn_cell_gru</a>()</code> <br><code><a href="rnn_cell_lstm.html">rnn_cell_lstm</a>()</code> <br><code><a href="rnn_cell_simple.html">rnn_cell_simple</a>()</code> <br><code><a href="rnn_cells_stack.html">rnn_cells_stack</a>()</code> <br></p></div>
+    <div class="dont-index"><p>Other image preprocessing layers: <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_elastic_transform.html">layer_random_elastic_transform</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br></p>
+<p>Other preprocessing layers: <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_category_encoding.html">layer_category_encoding</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_discretization.html">layer_discretization</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_feature_space.html">layer_feature_space</a>()</code> <br><code><a href="layer_hashed_crossing.html">layer_hashed_crossing</a>()</code> <br><code><a href="layer_hashing.html">layer_hashing</a>()</code> <br><code><a href="layer_integer_lookup.html">layer_integer_lookup</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_mel_spectrogram.html">layer_mel_spectrogram</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_normalization.html">layer_normalization</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_brightness.html">layer_random_brightness</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_contrast.html">layer_random_contrast</a>()</code> <br><code><a href="layer_random_crop.html">layer_random_crop</a>()</code> <br><code><a href="layer_random_elastic_transform.html">layer_random_elastic_transform</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_flip.html">layer_random_flip</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_rotation.html">layer_random_rotation</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_translation.html">layer_random_translation</a>()</code> <br><code><a href="layer_random_zoom.html">layer_random_zoom</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br><code><a href="layer_stft_spectrogram.html">layer_stft_spectrogram</a>()</code> <br><code><a href="layer_string_lookup.html">layer_string_lookup</a>()</code> <br><code><a href="layer_text_vectorization.html">layer_text_vectorization</a>()</code> <br></p>
+<p>Other layers: <br><code><a href="Layer.html">Layer</a>()</code> <br><code><a href="layer_activation.html">layer_activation</a>()</code> <br><code><a href="layer_activation_elu.html">layer_activation_elu</a>()</code> <br><code><a href="layer_activation_leaky_relu.html">layer_activation_leaky_relu</a>()</code> <br><code><a href="layer_activation_parametric_relu.html">layer_activation_parametric_relu</a>()</code> <br><code><a href="layer_activation_relu.html">layer_activation_relu</a>()</code> <br><code><a href="layer_activation_softmax.html">layer_activation_softmax</a>()</code> <br><code><a href="layer_activity_regularization.html">layer_activity_regularization</a>()</code> <br><code><a href="layer_add.html">layer_add</a>()</code> <br><code><a href="layer_additive_attention.html">layer_additive_attention</a>()</code> <br><code><a href="layer_alpha_dropout.html">layer_alpha_dropout</a>()</code> <br><code><a href="layer_attention.html">layer_attention</a>()</code> <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_average.html">layer_average</a>()</code> <br><code><a href="layer_average_pooling_1d.html">layer_average_pooling_1d</a>()</code> <br><code><a href="layer_average_pooling_2d.html">layer_average_pooling_2d</a>()</code> <br><code><a href="layer_average_pooling_3d.html">layer_average_pooling_3d</a>()</code> <br><code><a href="layer_batch_normalization.html">layer_batch_normalization</a>()</code> <br><code><a href="layer_bidirectional.html">layer_bidirectional</a>()</code> <br><code><a href="layer_category_encoding.html">layer_category_encoding</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_concatenate.html">layer_concatenate</a>()</code> <br><code><a href="layer_conv_1d.html">layer_conv_1d</a>()</code> <br><code><a href="layer_conv_1d_transpose.html">layer_conv_1d_transpose</a>()</code> <br><code><a href="layer_conv_2d.html">layer_conv_2d</a>()</code> <br><code><a href="layer_conv_2d_transpose.html">layer_conv_2d_transpose</a>()</code> <br><code><a href="layer_conv_3d.html">layer_conv_3d</a>()</code> <br><code><a href="layer_conv_3d_transpose.html">layer_conv_3d_transpose</a>()</code> <br><code><a href="layer_conv_lstm_1d.html">layer_conv_lstm_1d</a>()</code> <br><code><a href="layer_conv_lstm_2d.html">layer_conv_lstm_2d</a>()</code> <br><code><a href="layer_conv_lstm_3d.html">layer_conv_lstm_3d</a>()</code> <br><code><a href="layer_cropping_1d.html">layer_cropping_1d</a>()</code> <br><code><a href="layer_cropping_2d.html">layer_cropping_2d</a>()</code> <br><code><a href="layer_cropping_3d.html">layer_cropping_3d</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_dense.html">layer_dense</a>()</code> <br><code><a href="layer_depthwise_conv_1d.html">layer_depthwise_conv_1d</a>()</code> <br><code><a href="layer_depthwise_conv_2d.html">layer_depthwise_conv_2d</a>()</code> <br><code><a href="layer_discretization.html">layer_discretization</a>()</code> <br><code><a href="layer_dot.html">layer_dot</a>()</code> <br><code><a href="layer_dropout.html">layer_dropout</a>()</code> <br><code><a href="layer_einsum_dense.html">layer_einsum_dense</a>()</code> <br><code><a href="layer_embedding.html">layer_embedding</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_feature_space.html">layer_feature_space</a>()</code> <br><code><a href="layer_flatten.html">layer_flatten</a>()</code> <br><code><a href="layer_flax_module_wrapper.html">layer_flax_module_wrapper</a>()</code> <br><code><a href="layer_gaussian_dropout.html">layer_gaussian_dropout</a>()</code> <br><code><a href="layer_gaussian_noise.html">layer_gaussian_noise</a>()</code> <br><code><a href="layer_global_average_pooling_1d.html">layer_global_average_pooling_1d</a>()</code> <br><code><a href="layer_global_average_pooling_2d.html">layer_global_average_pooling_2d</a>()</code> <br><code><a href="layer_global_average_pooling_3d.html">layer_global_average_pooling_3d</a>()</code> <br><code><a href="layer_global_max_pooling_1d.html">layer_global_max_pooling_1d</a>()</code> <br><code><a href="layer_global_max_pooling_2d.html">layer_global_max_pooling_2d</a>()</code> <br><code><a href="layer_global_max_pooling_3d.html">layer_global_max_pooling_3d</a>()</code> <br><code><a href="layer_group_normalization.html">layer_group_normalization</a>()</code> <br><code><a href="layer_group_query_attention.html">layer_group_query_attention</a>()</code> <br><code><a href="layer_gru.html">layer_gru</a>()</code> <br><code><a href="layer_hashed_crossing.html">layer_hashed_crossing</a>()</code> <br><code><a href="layer_hashing.html">layer_hashing</a>()</code> <br><code><a href="layer_identity.html">layer_identity</a>()</code> <br><code><a href="layer_integer_lookup.html">layer_integer_lookup</a>()</code> <br><code><a href="layer_jax_model_wrapper.html">layer_jax_model_wrapper</a>()</code> <br><code><a href="layer_lambda.html">layer_lambda</a>()</code> <br><code><a href="layer_layer_normalization.html">layer_layer_normalization</a>()</code> <br><code><a href="layer_lstm.html">layer_lstm</a>()</code> <br><code><a href="layer_masking.html">layer_masking</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_max_pooling_1d.html">layer_max_pooling_1d</a>()</code> <br><code><a href="layer_max_pooling_2d.html">layer_max_pooling_2d</a>()</code> <br><code><a href="layer_max_pooling_3d.html">layer_max_pooling_3d</a>()</code> <br><code><a href="layer_maximum.html">layer_maximum</a>()</code> <br><code><a href="layer_mel_spectrogram.html">layer_mel_spectrogram</a>()</code> <br><code><a href="layer_minimum.html">layer_minimum</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_multi_head_attention.html">layer_multi_head_attention</a>()</code> <br><code><a href="layer_multiply.html">layer_multiply</a>()</code> <br><code><a href="layer_normalization.html">layer_normalization</a>()</code> <br><code><a href="layer_permute.html">layer_permute</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_brightness.html">layer_random_brightness</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_contrast.html">layer_random_contrast</a>()</code> <br><code><a href="layer_random_crop.html">layer_random_crop</a>()</code> <br><code><a href="layer_random_elastic_transform.html">layer_random_elastic_transform</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_flip.html">layer_random_flip</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_rotation.html">layer_random_rotation</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_translation.html">layer_random_translation</a>()</code> <br><code><a href="layer_random_zoom.html">layer_random_zoom</a>()</code> <br><code><a href="layer_repeat_vector.html">layer_repeat_vector</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_reshape.html">layer_reshape</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_rms_normalization.html">layer_rms_normalization</a>()</code> <br><code><a href="layer_rnn.html">layer_rnn</a>()</code> <br><code><a href="layer_separable_conv_1d.html">layer_separable_conv_1d</a>()</code> <br><code><a href="layer_separable_conv_2d.html">layer_separable_conv_2d</a>()</code> <br><code><a href="layer_simple_rnn.html">layer_simple_rnn</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br><code><a href="layer_spatial_dropout_1d.html">layer_spatial_dropout_1d</a>()</code> <br><code><a href="layer_spatial_dropout_2d.html">layer_spatial_dropout_2d</a>()</code> <br><code><a href="layer_spatial_dropout_3d.html">layer_spatial_dropout_3d</a>()</code> <br><code><a href="layer_spectral_normalization.html">layer_spectral_normalization</a>()</code> <br><code><a href="layer_stft_spectrogram.html">layer_stft_spectrogram</a>()</code> <br><code><a href="layer_string_lookup.html">layer_string_lookup</a>()</code> <br><code><a href="layer_subtract.html">layer_subtract</a>()</code> <br><code><a href="layer_text_vectorization.html">layer_text_vectorization</a>()</code> <br><code><a href="layer_tfsm.html">layer_tfsm</a>()</code> <br><code><a href="layer_time_distributed.html">layer_time_distributed</a>()</code> <br><code><a href="layer_torch_module_wrapper.html">layer_torch_module_wrapper</a>()</code> <br><code><a href="layer_unit_normalization.html">layer_unit_normalization</a>()</code> <br><code><a href="layer_upsampling_1d.html">layer_upsampling_1d</a>()</code> <br><code><a href="layer_upsampling_2d.html">layer_upsampling_2d</a>()</code> <br><code><a href="layer_upsampling_3d.html">layer_upsampling_3d</a>()</code> <br><code><a href="layer_zero_padding_1d.html">layer_zero_padding_1d</a>()</code> <br><code><a href="layer_zero_padding_2d.html">layer_zero_padding_2d</a>()</code> <br><code><a href="layer_zero_padding_3d.html">layer_zero_padding_3d</a>()</code> <br><code><a href="rnn_cell_gru.html">rnn_cell_gru</a>()</code> <br><code><a href="rnn_cell_lstm.html">rnn_cell_lstm</a>()</code> <br><code><a href="rnn_cell_simple.html">rnn_cell_simple</a>()</code> <br><code><a href="rnn_cells_stack.html">rnn_cells_stack</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -176,7 +176,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/layer_random_translation.html b/docs/dev/reference/layer_random_translation.html
index 762ded316..739c35c1f 100644
--- a/docs/dev/reference/layer_random_translation.html
+++ b/docs/dev/reference/layer_random_translation.html
@@ -16,7 +16,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -194,8 +194,8 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
     <div class="dont-index">
 <ul><li><p><a href="https://keras.io/api/layers/preprocessing_layers/image_augmentation/random_translation#randomtranslation-class" class="external-link">https://keras.io/api/layers/preprocessing_layers/image_augmentation/random_translation#randomtranslation-class</a></p></li>
 </ul><p>Other image augmentation layers: <br><code><a href="layer_random_brightness.html">layer_random_brightness</a>()</code> <br><code><a href="layer_random_contrast.html">layer_random_contrast</a>()</code> <br><code><a href="layer_random_crop.html">layer_random_crop</a>()</code> <br><code><a href="layer_random_flip.html">layer_random_flip</a>()</code> <br><code><a href="layer_random_rotation.html">layer_random_rotation</a>()</code> <br><code><a href="layer_random_zoom.html">layer_random_zoom</a>()</code> <br></p>
-<p>Other preprocessing layers: <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_category_encoding.html">layer_category_encoding</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_discretization.html">layer_discretization</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_feature_space.html">layer_feature_space</a>()</code> <br><code><a href="layer_hashed_crossing.html">layer_hashed_crossing</a>()</code> <br><code><a href="layer_hashing.html">layer_hashing</a>()</code> <br><code><a href="layer_integer_lookup.html">layer_integer_lookup</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_mel_spectrogram.html">layer_mel_spectrogram</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_normalization.html">layer_normalization</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_brightness.html">layer_random_brightness</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_contrast.html">layer_random_contrast</a>()</code> <br><code><a href="layer_random_crop.html">layer_random_crop</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_flip.html">layer_random_flip</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_rotation.html">layer_random_rotation</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_random_zoom.html">layer_random_zoom</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br><code><a href="layer_stft_spectrogram.html">layer_stft_spectrogram</a>()</code> <br><code><a href="layer_string_lookup.html">layer_string_lookup</a>()</code> <br><code><a href="layer_text_vectorization.html">layer_text_vectorization</a>()</code> <br></p>
-<p>Other layers: <br><code><a href="Layer.html">Layer</a>()</code> <br><code><a href="layer_activation.html">layer_activation</a>()</code> <br><code><a href="layer_activation_elu.html">layer_activation_elu</a>()</code> <br><code><a href="layer_activation_leaky_relu.html">layer_activation_leaky_relu</a>()</code> <br><code><a href="layer_activation_parametric_relu.html">layer_activation_parametric_relu</a>()</code> <br><code><a href="layer_activation_relu.html">layer_activation_relu</a>()</code> <br><code><a href="layer_activation_softmax.html">layer_activation_softmax</a>()</code> <br><code><a href="layer_activity_regularization.html">layer_activity_regularization</a>()</code> <br><code><a href="layer_add.html">layer_add</a>()</code> <br><code><a href="layer_additive_attention.html">layer_additive_attention</a>()</code> <br><code><a href="layer_alpha_dropout.html">layer_alpha_dropout</a>()</code> <br><code><a href="layer_attention.html">layer_attention</a>()</code> <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_average.html">layer_average</a>()</code> <br><code><a href="layer_average_pooling_1d.html">layer_average_pooling_1d</a>()</code> <br><code><a href="layer_average_pooling_2d.html">layer_average_pooling_2d</a>()</code> <br><code><a href="layer_average_pooling_3d.html">layer_average_pooling_3d</a>()</code> <br><code><a href="layer_batch_normalization.html">layer_batch_normalization</a>()</code> <br><code><a href="layer_bidirectional.html">layer_bidirectional</a>()</code> <br><code><a href="layer_category_encoding.html">layer_category_encoding</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_concatenate.html">layer_concatenate</a>()</code> <br><code><a href="layer_conv_1d.html">layer_conv_1d</a>()</code> <br><code><a href="layer_conv_1d_transpose.html">layer_conv_1d_transpose</a>()</code> <br><code><a href="layer_conv_2d.html">layer_conv_2d</a>()</code> <br><code><a href="layer_conv_2d_transpose.html">layer_conv_2d_transpose</a>()</code> <br><code><a href="layer_conv_3d.html">layer_conv_3d</a>()</code> <br><code><a href="layer_conv_3d_transpose.html">layer_conv_3d_transpose</a>()</code> <br><code><a href="layer_conv_lstm_1d.html">layer_conv_lstm_1d</a>()</code> <br><code><a href="layer_conv_lstm_2d.html">layer_conv_lstm_2d</a>()</code> <br><code><a href="layer_conv_lstm_3d.html">layer_conv_lstm_3d</a>()</code> <br><code><a href="layer_cropping_1d.html">layer_cropping_1d</a>()</code> <br><code><a href="layer_cropping_2d.html">layer_cropping_2d</a>()</code> <br><code><a href="layer_cropping_3d.html">layer_cropping_3d</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_dense.html">layer_dense</a>()</code> <br><code><a href="layer_depthwise_conv_1d.html">layer_depthwise_conv_1d</a>()</code> <br><code><a href="layer_depthwise_conv_2d.html">layer_depthwise_conv_2d</a>()</code> <br><code><a href="layer_discretization.html">layer_discretization</a>()</code> <br><code><a href="layer_dot.html">layer_dot</a>()</code> <br><code><a href="layer_dropout.html">layer_dropout</a>()</code> <br><code><a href="layer_einsum_dense.html">layer_einsum_dense</a>()</code> <br><code><a href="layer_embedding.html">layer_embedding</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_feature_space.html">layer_feature_space</a>()</code> <br><code><a href="layer_flatten.html">layer_flatten</a>()</code> <br><code><a href="layer_flax_module_wrapper.html">layer_flax_module_wrapper</a>()</code> <br><code><a href="layer_gaussian_dropout.html">layer_gaussian_dropout</a>()</code> <br><code><a href="layer_gaussian_noise.html">layer_gaussian_noise</a>()</code> <br><code><a href="layer_global_average_pooling_1d.html">layer_global_average_pooling_1d</a>()</code> <br><code><a href="layer_global_average_pooling_2d.html">layer_global_average_pooling_2d</a>()</code> <br><code><a href="layer_global_average_pooling_3d.html">layer_global_average_pooling_3d</a>()</code> <br><code><a href="layer_global_max_pooling_1d.html">layer_global_max_pooling_1d</a>()</code> <br><code><a href="layer_global_max_pooling_2d.html">layer_global_max_pooling_2d</a>()</code> <br><code><a href="layer_global_max_pooling_3d.html">layer_global_max_pooling_3d</a>()</code> <br><code><a href="layer_group_normalization.html">layer_group_normalization</a>()</code> <br><code><a href="layer_group_query_attention.html">layer_group_query_attention</a>()</code> <br><code><a href="layer_gru.html">layer_gru</a>()</code> <br><code><a href="layer_hashed_crossing.html">layer_hashed_crossing</a>()</code> <br><code><a href="layer_hashing.html">layer_hashing</a>()</code> <br><code><a href="layer_identity.html">layer_identity</a>()</code> <br><code><a href="layer_integer_lookup.html">layer_integer_lookup</a>()</code> <br><code><a href="layer_jax_model_wrapper.html">layer_jax_model_wrapper</a>()</code> <br><code><a href="layer_lambda.html">layer_lambda</a>()</code> <br><code><a href="layer_layer_normalization.html">layer_layer_normalization</a>()</code> <br><code><a href="layer_lstm.html">layer_lstm</a>()</code> <br><code><a href="layer_masking.html">layer_masking</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_max_pooling_1d.html">layer_max_pooling_1d</a>()</code> <br><code><a href="layer_max_pooling_2d.html">layer_max_pooling_2d</a>()</code> <br><code><a href="layer_max_pooling_3d.html">layer_max_pooling_3d</a>()</code> <br><code><a href="layer_maximum.html">layer_maximum</a>()</code> <br><code><a href="layer_mel_spectrogram.html">layer_mel_spectrogram</a>()</code> <br><code><a href="layer_minimum.html">layer_minimum</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_multi_head_attention.html">layer_multi_head_attention</a>()</code> <br><code><a href="layer_multiply.html">layer_multiply</a>()</code> <br><code><a href="layer_normalization.html">layer_normalization</a>()</code> <br><code><a href="layer_permute.html">layer_permute</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_brightness.html">layer_random_brightness</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_contrast.html">layer_random_contrast</a>()</code> <br><code><a href="layer_random_crop.html">layer_random_crop</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_flip.html">layer_random_flip</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_rotation.html">layer_random_rotation</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_random_zoom.html">layer_random_zoom</a>()</code> <br><code><a href="layer_repeat_vector.html">layer_repeat_vector</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_reshape.html">layer_reshape</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_rms_normalization.html">layer_rms_normalization</a>()</code> <br><code><a href="layer_rnn.html">layer_rnn</a>()</code> <br><code><a href="layer_separable_conv_1d.html">layer_separable_conv_1d</a>()</code> <br><code><a href="layer_separable_conv_2d.html">layer_separable_conv_2d</a>()</code> <br><code><a href="layer_simple_rnn.html">layer_simple_rnn</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br><code><a href="layer_spatial_dropout_1d.html">layer_spatial_dropout_1d</a>()</code> <br><code><a href="layer_spatial_dropout_2d.html">layer_spatial_dropout_2d</a>()</code> <br><code><a href="layer_spatial_dropout_3d.html">layer_spatial_dropout_3d</a>()</code> <br><code><a href="layer_spectral_normalization.html">layer_spectral_normalization</a>()</code> <br><code><a href="layer_stft_spectrogram.html">layer_stft_spectrogram</a>()</code> <br><code><a href="layer_string_lookup.html">layer_string_lookup</a>()</code> <br><code><a href="layer_subtract.html">layer_subtract</a>()</code> <br><code><a href="layer_text_vectorization.html">layer_text_vectorization</a>()</code> <br><code><a href="layer_tfsm.html">layer_tfsm</a>()</code> <br><code><a href="layer_time_distributed.html">layer_time_distributed</a>()</code> <br><code><a href="layer_torch_module_wrapper.html">layer_torch_module_wrapper</a>()</code> <br><code><a href="layer_unit_normalization.html">layer_unit_normalization</a>()</code> <br><code><a href="layer_upsampling_1d.html">layer_upsampling_1d</a>()</code> <br><code><a href="layer_upsampling_2d.html">layer_upsampling_2d</a>()</code> <br><code><a href="layer_upsampling_3d.html">layer_upsampling_3d</a>()</code> <br><code><a href="layer_zero_padding_1d.html">layer_zero_padding_1d</a>()</code> <br><code><a href="layer_zero_padding_2d.html">layer_zero_padding_2d</a>()</code> <br><code><a href="layer_zero_padding_3d.html">layer_zero_padding_3d</a>()</code> <br><code><a href="rnn_cell_gru.html">rnn_cell_gru</a>()</code> <br><code><a href="rnn_cell_lstm.html">rnn_cell_lstm</a>()</code> <br><code><a href="rnn_cell_simple.html">rnn_cell_simple</a>()</code> <br><code><a href="rnn_cells_stack.html">rnn_cells_stack</a>()</code> <br></p></div>
+<p>Other preprocessing layers: <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_category_encoding.html">layer_category_encoding</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_discretization.html">layer_discretization</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_feature_space.html">layer_feature_space</a>()</code> <br><code><a href="layer_hashed_crossing.html">layer_hashed_crossing</a>()</code> <br><code><a href="layer_hashing.html">layer_hashing</a>()</code> <br><code><a href="layer_integer_lookup.html">layer_integer_lookup</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_mel_spectrogram.html">layer_mel_spectrogram</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_normalization.html">layer_normalization</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_brightness.html">layer_random_brightness</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_contrast.html">layer_random_contrast</a>()</code> <br><code><a href="layer_random_crop.html">layer_random_crop</a>()</code> <br><code><a href="layer_random_elastic_transform.html">layer_random_elastic_transform</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_flip.html">layer_random_flip</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_rotation.html">layer_random_rotation</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_random_zoom.html">layer_random_zoom</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br><code><a href="layer_stft_spectrogram.html">layer_stft_spectrogram</a>()</code> <br><code><a href="layer_string_lookup.html">layer_string_lookup</a>()</code> <br><code><a href="layer_text_vectorization.html">layer_text_vectorization</a>()</code> <br></p>
+<p>Other layers: <br><code><a href="Layer.html">Layer</a>()</code> <br><code><a href="layer_activation.html">layer_activation</a>()</code> <br><code><a href="layer_activation_elu.html">layer_activation_elu</a>()</code> <br><code><a href="layer_activation_leaky_relu.html">layer_activation_leaky_relu</a>()</code> <br><code><a href="layer_activation_parametric_relu.html">layer_activation_parametric_relu</a>()</code> <br><code><a href="layer_activation_relu.html">layer_activation_relu</a>()</code> <br><code><a href="layer_activation_softmax.html">layer_activation_softmax</a>()</code> <br><code><a href="layer_activity_regularization.html">layer_activity_regularization</a>()</code> <br><code><a href="layer_add.html">layer_add</a>()</code> <br><code><a href="layer_additive_attention.html">layer_additive_attention</a>()</code> <br><code><a href="layer_alpha_dropout.html">layer_alpha_dropout</a>()</code> <br><code><a href="layer_attention.html">layer_attention</a>()</code> <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_average.html">layer_average</a>()</code> <br><code><a href="layer_average_pooling_1d.html">layer_average_pooling_1d</a>()</code> <br><code><a href="layer_average_pooling_2d.html">layer_average_pooling_2d</a>()</code> <br><code><a href="layer_average_pooling_3d.html">layer_average_pooling_3d</a>()</code> <br><code><a href="layer_batch_normalization.html">layer_batch_normalization</a>()</code> <br><code><a href="layer_bidirectional.html">layer_bidirectional</a>()</code> <br><code><a href="layer_category_encoding.html">layer_category_encoding</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_concatenate.html">layer_concatenate</a>()</code> <br><code><a href="layer_conv_1d.html">layer_conv_1d</a>()</code> <br><code><a href="layer_conv_1d_transpose.html">layer_conv_1d_transpose</a>()</code> <br><code><a href="layer_conv_2d.html">layer_conv_2d</a>()</code> <br><code><a href="layer_conv_2d_transpose.html">layer_conv_2d_transpose</a>()</code> <br><code><a href="layer_conv_3d.html">layer_conv_3d</a>()</code> <br><code><a href="layer_conv_3d_transpose.html">layer_conv_3d_transpose</a>()</code> <br><code><a href="layer_conv_lstm_1d.html">layer_conv_lstm_1d</a>()</code> <br><code><a href="layer_conv_lstm_2d.html">layer_conv_lstm_2d</a>()</code> <br><code><a href="layer_conv_lstm_3d.html">layer_conv_lstm_3d</a>()</code> <br><code><a href="layer_cropping_1d.html">layer_cropping_1d</a>()</code> <br><code><a href="layer_cropping_2d.html">layer_cropping_2d</a>()</code> <br><code><a href="layer_cropping_3d.html">layer_cropping_3d</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_dense.html">layer_dense</a>()</code> <br><code><a href="layer_depthwise_conv_1d.html">layer_depthwise_conv_1d</a>()</code> <br><code><a href="layer_depthwise_conv_2d.html">layer_depthwise_conv_2d</a>()</code> <br><code><a href="layer_discretization.html">layer_discretization</a>()</code> <br><code><a href="layer_dot.html">layer_dot</a>()</code> <br><code><a href="layer_dropout.html">layer_dropout</a>()</code> <br><code><a href="layer_einsum_dense.html">layer_einsum_dense</a>()</code> <br><code><a href="layer_embedding.html">layer_embedding</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_feature_space.html">layer_feature_space</a>()</code> <br><code><a href="layer_flatten.html">layer_flatten</a>()</code> <br><code><a href="layer_flax_module_wrapper.html">layer_flax_module_wrapper</a>()</code> <br><code><a href="layer_gaussian_dropout.html">layer_gaussian_dropout</a>()</code> <br><code><a href="layer_gaussian_noise.html">layer_gaussian_noise</a>()</code> <br><code><a href="layer_global_average_pooling_1d.html">layer_global_average_pooling_1d</a>()</code> <br><code><a href="layer_global_average_pooling_2d.html">layer_global_average_pooling_2d</a>()</code> <br><code><a href="layer_global_average_pooling_3d.html">layer_global_average_pooling_3d</a>()</code> <br><code><a href="layer_global_max_pooling_1d.html">layer_global_max_pooling_1d</a>()</code> <br><code><a href="layer_global_max_pooling_2d.html">layer_global_max_pooling_2d</a>()</code> <br><code><a href="layer_global_max_pooling_3d.html">layer_global_max_pooling_3d</a>()</code> <br><code><a href="layer_group_normalization.html">layer_group_normalization</a>()</code> <br><code><a href="layer_group_query_attention.html">layer_group_query_attention</a>()</code> <br><code><a href="layer_gru.html">layer_gru</a>()</code> <br><code><a href="layer_hashed_crossing.html">layer_hashed_crossing</a>()</code> <br><code><a href="layer_hashing.html">layer_hashing</a>()</code> <br><code><a href="layer_identity.html">layer_identity</a>()</code> <br><code><a href="layer_integer_lookup.html">layer_integer_lookup</a>()</code> <br><code><a href="layer_jax_model_wrapper.html">layer_jax_model_wrapper</a>()</code> <br><code><a href="layer_lambda.html">layer_lambda</a>()</code> <br><code><a href="layer_layer_normalization.html">layer_layer_normalization</a>()</code> <br><code><a href="layer_lstm.html">layer_lstm</a>()</code> <br><code><a href="layer_masking.html">layer_masking</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_max_pooling_1d.html">layer_max_pooling_1d</a>()</code> <br><code><a href="layer_max_pooling_2d.html">layer_max_pooling_2d</a>()</code> <br><code><a href="layer_max_pooling_3d.html">layer_max_pooling_3d</a>()</code> <br><code><a href="layer_maximum.html">layer_maximum</a>()</code> <br><code><a href="layer_mel_spectrogram.html">layer_mel_spectrogram</a>()</code> <br><code><a href="layer_minimum.html">layer_minimum</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_multi_head_attention.html">layer_multi_head_attention</a>()</code> <br><code><a href="layer_multiply.html">layer_multiply</a>()</code> <br><code><a href="layer_normalization.html">layer_normalization</a>()</code> <br><code><a href="layer_permute.html">layer_permute</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_brightness.html">layer_random_brightness</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_contrast.html">layer_random_contrast</a>()</code> <br><code><a href="layer_random_crop.html">layer_random_crop</a>()</code> <br><code><a href="layer_random_elastic_transform.html">layer_random_elastic_transform</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_flip.html">layer_random_flip</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_rotation.html">layer_random_rotation</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_random_zoom.html">layer_random_zoom</a>()</code> <br><code><a href="layer_repeat_vector.html">layer_repeat_vector</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_reshape.html">layer_reshape</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_rms_normalization.html">layer_rms_normalization</a>()</code> <br><code><a href="layer_rnn.html">layer_rnn</a>()</code> <br><code><a href="layer_separable_conv_1d.html">layer_separable_conv_1d</a>()</code> <br><code><a href="layer_separable_conv_2d.html">layer_separable_conv_2d</a>()</code> <br><code><a href="layer_simple_rnn.html">layer_simple_rnn</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br><code><a href="layer_spatial_dropout_1d.html">layer_spatial_dropout_1d</a>()</code> <br><code><a href="layer_spatial_dropout_2d.html">layer_spatial_dropout_2d</a>()</code> <br><code><a href="layer_spatial_dropout_3d.html">layer_spatial_dropout_3d</a>()</code> <br><code><a href="layer_spectral_normalization.html">layer_spectral_normalization</a>()</code> <br><code><a href="layer_stft_spectrogram.html">layer_stft_spectrogram</a>()</code> <br><code><a href="layer_string_lookup.html">layer_string_lookup</a>()</code> <br><code><a href="layer_subtract.html">layer_subtract</a>()</code> <br><code><a href="layer_text_vectorization.html">layer_text_vectorization</a>()</code> <br><code><a href="layer_tfsm.html">layer_tfsm</a>()</code> <br><code><a href="layer_time_distributed.html">layer_time_distributed</a>()</code> <br><code><a href="layer_torch_module_wrapper.html">layer_torch_module_wrapper</a>()</code> <br><code><a href="layer_unit_normalization.html">layer_unit_normalization</a>()</code> <br><code><a href="layer_upsampling_1d.html">layer_upsampling_1d</a>()</code> <br><code><a href="layer_upsampling_2d.html">layer_upsampling_2d</a>()</code> <br><code><a href="layer_upsampling_3d.html">layer_upsampling_3d</a>()</code> <br><code><a href="layer_zero_padding_1d.html">layer_zero_padding_1d</a>()</code> <br><code><a href="layer_zero_padding_2d.html">layer_zero_padding_2d</a>()</code> <br><code><a href="layer_zero_padding_3d.html">layer_zero_padding_3d</a>()</code> <br><code><a href="rnn_cell_gru.html">rnn_cell_gru</a>()</code> <br><code><a href="rnn_cell_lstm.html">rnn_cell_lstm</a>()</code> <br><code><a href="rnn_cell_simple.html">rnn_cell_simple</a>()</code> <br><code><a href="rnn_cells_stack.html">rnn_cells_stack</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -207,7 +207,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/layer_random_zoom.html b/docs/dev/reference/layer_random_zoom.html
index a554b6102..3a67c623a 100644
--- a/docs/dev/reference/layer_random_zoom.html
+++ b/docs/dev/reference/layer_random_zoom.html
@@ -16,7 +16,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -200,8 +200,8 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
     <div class="dont-index">
 <ul><li><p><a href="https://keras.io/api/layers/preprocessing_layers/image_augmentation/random_zoom#randomzoom-class" class="external-link">https://keras.io/api/layers/preprocessing_layers/image_augmentation/random_zoom#randomzoom-class</a></p></li>
 </ul><p>Other image augmentation layers: <br><code><a href="layer_random_brightness.html">layer_random_brightness</a>()</code> <br><code><a href="layer_random_contrast.html">layer_random_contrast</a>()</code> <br><code><a href="layer_random_crop.html">layer_random_crop</a>()</code> <br><code><a href="layer_random_flip.html">layer_random_flip</a>()</code> <br><code><a href="layer_random_rotation.html">layer_random_rotation</a>()</code> <br><code><a href="layer_random_translation.html">layer_random_translation</a>()</code> <br></p>
-<p>Other preprocessing layers: <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_category_encoding.html">layer_category_encoding</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_discretization.html">layer_discretization</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_feature_space.html">layer_feature_space</a>()</code> <br><code><a href="layer_hashed_crossing.html">layer_hashed_crossing</a>()</code> <br><code><a href="layer_hashing.html">layer_hashing</a>()</code> <br><code><a href="layer_integer_lookup.html">layer_integer_lookup</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_mel_spectrogram.html">layer_mel_spectrogram</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_normalization.html">layer_normalization</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_brightness.html">layer_random_brightness</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_contrast.html">layer_random_contrast</a>()</code> <br><code><a href="layer_random_crop.html">layer_random_crop</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_flip.html">layer_random_flip</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_rotation.html">layer_random_rotation</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_random_translation.html">layer_random_translation</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br><code><a href="layer_stft_spectrogram.html">layer_stft_spectrogram</a>()</code> <br><code><a href="layer_string_lookup.html">layer_string_lookup</a>()</code> <br><code><a href="layer_text_vectorization.html">layer_text_vectorization</a>()</code> <br></p>
-<p>Other layers: <br><code><a href="Layer.html">Layer</a>()</code> <br><code><a href="layer_activation.html">layer_activation</a>()</code> <br><code><a href="layer_activation_elu.html">layer_activation_elu</a>()</code> <br><code><a href="layer_activation_leaky_relu.html">layer_activation_leaky_relu</a>()</code> <br><code><a href="layer_activation_parametric_relu.html">layer_activation_parametric_relu</a>()</code> <br><code><a href="layer_activation_relu.html">layer_activation_relu</a>()</code> <br><code><a href="layer_activation_softmax.html">layer_activation_softmax</a>()</code> <br><code><a href="layer_activity_regularization.html">layer_activity_regularization</a>()</code> <br><code><a href="layer_add.html">layer_add</a>()</code> <br><code><a href="layer_additive_attention.html">layer_additive_attention</a>()</code> <br><code><a href="layer_alpha_dropout.html">layer_alpha_dropout</a>()</code> <br><code><a href="layer_attention.html">layer_attention</a>()</code> <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_average.html">layer_average</a>()</code> <br><code><a href="layer_average_pooling_1d.html">layer_average_pooling_1d</a>()</code> <br><code><a href="layer_average_pooling_2d.html">layer_average_pooling_2d</a>()</code> <br><code><a href="layer_average_pooling_3d.html">layer_average_pooling_3d</a>()</code> <br><code><a href="layer_batch_normalization.html">layer_batch_normalization</a>()</code> <br><code><a href="layer_bidirectional.html">layer_bidirectional</a>()</code> <br><code><a href="layer_category_encoding.html">layer_category_encoding</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_concatenate.html">layer_concatenate</a>()</code> <br><code><a href="layer_conv_1d.html">layer_conv_1d</a>()</code> <br><code><a href="layer_conv_1d_transpose.html">layer_conv_1d_transpose</a>()</code> <br><code><a href="layer_conv_2d.html">layer_conv_2d</a>()</code> <br><code><a href="layer_conv_2d_transpose.html">layer_conv_2d_transpose</a>()</code> <br><code><a href="layer_conv_3d.html">layer_conv_3d</a>()</code> <br><code><a href="layer_conv_3d_transpose.html">layer_conv_3d_transpose</a>()</code> <br><code><a href="layer_conv_lstm_1d.html">layer_conv_lstm_1d</a>()</code> <br><code><a href="layer_conv_lstm_2d.html">layer_conv_lstm_2d</a>()</code> <br><code><a href="layer_conv_lstm_3d.html">layer_conv_lstm_3d</a>()</code> <br><code><a href="layer_cropping_1d.html">layer_cropping_1d</a>()</code> <br><code><a href="layer_cropping_2d.html">layer_cropping_2d</a>()</code> <br><code><a href="layer_cropping_3d.html">layer_cropping_3d</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_dense.html">layer_dense</a>()</code> <br><code><a href="layer_depthwise_conv_1d.html">layer_depthwise_conv_1d</a>()</code> <br><code><a href="layer_depthwise_conv_2d.html">layer_depthwise_conv_2d</a>()</code> <br><code><a href="layer_discretization.html">layer_discretization</a>()</code> <br><code><a href="layer_dot.html">layer_dot</a>()</code> <br><code><a href="layer_dropout.html">layer_dropout</a>()</code> <br><code><a href="layer_einsum_dense.html">layer_einsum_dense</a>()</code> <br><code><a href="layer_embedding.html">layer_embedding</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_feature_space.html">layer_feature_space</a>()</code> <br><code><a href="layer_flatten.html">layer_flatten</a>()</code> <br><code><a href="layer_flax_module_wrapper.html">layer_flax_module_wrapper</a>()</code> <br><code><a href="layer_gaussian_dropout.html">layer_gaussian_dropout</a>()</code> <br><code><a href="layer_gaussian_noise.html">layer_gaussian_noise</a>()</code> <br><code><a href="layer_global_average_pooling_1d.html">layer_global_average_pooling_1d</a>()</code> <br><code><a href="layer_global_average_pooling_2d.html">layer_global_average_pooling_2d</a>()</code> <br><code><a href="layer_global_average_pooling_3d.html">layer_global_average_pooling_3d</a>()</code> <br><code><a href="layer_global_max_pooling_1d.html">layer_global_max_pooling_1d</a>()</code> <br><code><a href="layer_global_max_pooling_2d.html">layer_global_max_pooling_2d</a>()</code> <br><code><a href="layer_global_max_pooling_3d.html">layer_global_max_pooling_3d</a>()</code> <br><code><a href="layer_group_normalization.html">layer_group_normalization</a>()</code> <br><code><a href="layer_group_query_attention.html">layer_group_query_attention</a>()</code> <br><code><a href="layer_gru.html">layer_gru</a>()</code> <br><code><a href="layer_hashed_crossing.html">layer_hashed_crossing</a>()</code> <br><code><a href="layer_hashing.html">layer_hashing</a>()</code> <br><code><a href="layer_identity.html">layer_identity</a>()</code> <br><code><a href="layer_integer_lookup.html">layer_integer_lookup</a>()</code> <br><code><a href="layer_jax_model_wrapper.html">layer_jax_model_wrapper</a>()</code> <br><code><a href="layer_lambda.html">layer_lambda</a>()</code> <br><code><a href="layer_layer_normalization.html">layer_layer_normalization</a>()</code> <br><code><a href="layer_lstm.html">layer_lstm</a>()</code> <br><code><a href="layer_masking.html">layer_masking</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_max_pooling_1d.html">layer_max_pooling_1d</a>()</code> <br><code><a href="layer_max_pooling_2d.html">layer_max_pooling_2d</a>()</code> <br><code><a href="layer_max_pooling_3d.html">layer_max_pooling_3d</a>()</code> <br><code><a href="layer_maximum.html">layer_maximum</a>()</code> <br><code><a href="layer_mel_spectrogram.html">layer_mel_spectrogram</a>()</code> <br><code><a href="layer_minimum.html">layer_minimum</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_multi_head_attention.html">layer_multi_head_attention</a>()</code> <br><code><a href="layer_multiply.html">layer_multiply</a>()</code> <br><code><a href="layer_normalization.html">layer_normalization</a>()</code> <br><code><a href="layer_permute.html">layer_permute</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_brightness.html">layer_random_brightness</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_contrast.html">layer_random_contrast</a>()</code> <br><code><a href="layer_random_crop.html">layer_random_crop</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_flip.html">layer_random_flip</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_rotation.html">layer_random_rotation</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_random_translation.html">layer_random_translation</a>()</code> <br><code><a href="layer_repeat_vector.html">layer_repeat_vector</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_reshape.html">layer_reshape</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_rms_normalization.html">layer_rms_normalization</a>()</code> <br><code><a href="layer_rnn.html">layer_rnn</a>()</code> <br><code><a href="layer_separable_conv_1d.html">layer_separable_conv_1d</a>()</code> <br><code><a href="layer_separable_conv_2d.html">layer_separable_conv_2d</a>()</code> <br><code><a href="layer_simple_rnn.html">layer_simple_rnn</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br><code><a href="layer_spatial_dropout_1d.html">layer_spatial_dropout_1d</a>()</code> <br><code><a href="layer_spatial_dropout_2d.html">layer_spatial_dropout_2d</a>()</code> <br><code><a href="layer_spatial_dropout_3d.html">layer_spatial_dropout_3d</a>()</code> <br><code><a href="layer_spectral_normalization.html">layer_spectral_normalization</a>()</code> <br><code><a href="layer_stft_spectrogram.html">layer_stft_spectrogram</a>()</code> <br><code><a href="layer_string_lookup.html">layer_string_lookup</a>()</code> <br><code><a href="layer_subtract.html">layer_subtract</a>()</code> <br><code><a href="layer_text_vectorization.html">layer_text_vectorization</a>()</code> <br><code><a href="layer_tfsm.html">layer_tfsm</a>()</code> <br><code><a href="layer_time_distributed.html">layer_time_distributed</a>()</code> <br><code><a href="layer_torch_module_wrapper.html">layer_torch_module_wrapper</a>()</code> <br><code><a href="layer_unit_normalization.html">layer_unit_normalization</a>()</code> <br><code><a href="layer_upsampling_1d.html">layer_upsampling_1d</a>()</code> <br><code><a href="layer_upsampling_2d.html">layer_upsampling_2d</a>()</code> <br><code><a href="layer_upsampling_3d.html">layer_upsampling_3d</a>()</code> <br><code><a href="layer_zero_padding_1d.html">layer_zero_padding_1d</a>()</code> <br><code><a href="layer_zero_padding_2d.html">layer_zero_padding_2d</a>()</code> <br><code><a href="layer_zero_padding_3d.html">layer_zero_padding_3d</a>()</code> <br><code><a href="rnn_cell_gru.html">rnn_cell_gru</a>()</code> <br><code><a href="rnn_cell_lstm.html">rnn_cell_lstm</a>()</code> <br><code><a href="rnn_cell_simple.html">rnn_cell_simple</a>()</code> <br><code><a href="rnn_cells_stack.html">rnn_cells_stack</a>()</code> <br></p></div>
+<p>Other preprocessing layers: <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_category_encoding.html">layer_category_encoding</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_discretization.html">layer_discretization</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_feature_space.html">layer_feature_space</a>()</code> <br><code><a href="layer_hashed_crossing.html">layer_hashed_crossing</a>()</code> <br><code><a href="layer_hashing.html">layer_hashing</a>()</code> <br><code><a href="layer_integer_lookup.html">layer_integer_lookup</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_mel_spectrogram.html">layer_mel_spectrogram</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_normalization.html">layer_normalization</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_brightness.html">layer_random_brightness</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_contrast.html">layer_random_contrast</a>()</code> <br><code><a href="layer_random_crop.html">layer_random_crop</a>()</code> <br><code><a href="layer_random_elastic_transform.html">layer_random_elastic_transform</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_flip.html">layer_random_flip</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_rotation.html">layer_random_rotation</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_random_translation.html">layer_random_translation</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br><code><a href="layer_stft_spectrogram.html">layer_stft_spectrogram</a>()</code> <br><code><a href="layer_string_lookup.html">layer_string_lookup</a>()</code> <br><code><a href="layer_text_vectorization.html">layer_text_vectorization</a>()</code> <br></p>
+<p>Other layers: <br><code><a href="Layer.html">Layer</a>()</code> <br><code><a href="layer_activation.html">layer_activation</a>()</code> <br><code><a href="layer_activation_elu.html">layer_activation_elu</a>()</code> <br><code><a href="layer_activation_leaky_relu.html">layer_activation_leaky_relu</a>()</code> <br><code><a href="layer_activation_parametric_relu.html">layer_activation_parametric_relu</a>()</code> <br><code><a href="layer_activation_relu.html">layer_activation_relu</a>()</code> <br><code><a href="layer_activation_softmax.html">layer_activation_softmax</a>()</code> <br><code><a href="layer_activity_regularization.html">layer_activity_regularization</a>()</code> <br><code><a href="layer_add.html">layer_add</a>()</code> <br><code><a href="layer_additive_attention.html">layer_additive_attention</a>()</code> <br><code><a href="layer_alpha_dropout.html">layer_alpha_dropout</a>()</code> <br><code><a href="layer_attention.html">layer_attention</a>()</code> <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_average.html">layer_average</a>()</code> <br><code><a href="layer_average_pooling_1d.html">layer_average_pooling_1d</a>()</code> <br><code><a href="layer_average_pooling_2d.html">layer_average_pooling_2d</a>()</code> <br><code><a href="layer_average_pooling_3d.html">layer_average_pooling_3d</a>()</code> <br><code><a href="layer_batch_normalization.html">layer_batch_normalization</a>()</code> <br><code><a href="layer_bidirectional.html">layer_bidirectional</a>()</code> <br><code><a href="layer_category_encoding.html">layer_category_encoding</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_concatenate.html">layer_concatenate</a>()</code> <br><code><a href="layer_conv_1d.html">layer_conv_1d</a>()</code> <br><code><a href="layer_conv_1d_transpose.html">layer_conv_1d_transpose</a>()</code> <br><code><a href="layer_conv_2d.html">layer_conv_2d</a>()</code> <br><code><a href="layer_conv_2d_transpose.html">layer_conv_2d_transpose</a>()</code> <br><code><a href="layer_conv_3d.html">layer_conv_3d</a>()</code> <br><code><a href="layer_conv_3d_transpose.html">layer_conv_3d_transpose</a>()</code> <br><code><a href="layer_conv_lstm_1d.html">layer_conv_lstm_1d</a>()</code> <br><code><a href="layer_conv_lstm_2d.html">layer_conv_lstm_2d</a>()</code> <br><code><a href="layer_conv_lstm_3d.html">layer_conv_lstm_3d</a>()</code> <br><code><a href="layer_cropping_1d.html">layer_cropping_1d</a>()</code> <br><code><a href="layer_cropping_2d.html">layer_cropping_2d</a>()</code> <br><code><a href="layer_cropping_3d.html">layer_cropping_3d</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_dense.html">layer_dense</a>()</code> <br><code><a href="layer_depthwise_conv_1d.html">layer_depthwise_conv_1d</a>()</code> <br><code><a href="layer_depthwise_conv_2d.html">layer_depthwise_conv_2d</a>()</code> <br><code><a href="layer_discretization.html">layer_discretization</a>()</code> <br><code><a href="layer_dot.html">layer_dot</a>()</code> <br><code><a href="layer_dropout.html">layer_dropout</a>()</code> <br><code><a href="layer_einsum_dense.html">layer_einsum_dense</a>()</code> <br><code><a href="layer_embedding.html">layer_embedding</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_feature_space.html">layer_feature_space</a>()</code> <br><code><a href="layer_flatten.html">layer_flatten</a>()</code> <br><code><a href="layer_flax_module_wrapper.html">layer_flax_module_wrapper</a>()</code> <br><code><a href="layer_gaussian_dropout.html">layer_gaussian_dropout</a>()</code> <br><code><a href="layer_gaussian_noise.html">layer_gaussian_noise</a>()</code> <br><code><a href="layer_global_average_pooling_1d.html">layer_global_average_pooling_1d</a>()</code> <br><code><a href="layer_global_average_pooling_2d.html">layer_global_average_pooling_2d</a>()</code> <br><code><a href="layer_global_average_pooling_3d.html">layer_global_average_pooling_3d</a>()</code> <br><code><a href="layer_global_max_pooling_1d.html">layer_global_max_pooling_1d</a>()</code> <br><code><a href="layer_global_max_pooling_2d.html">layer_global_max_pooling_2d</a>()</code> <br><code><a href="layer_global_max_pooling_3d.html">layer_global_max_pooling_3d</a>()</code> <br><code><a href="layer_group_normalization.html">layer_group_normalization</a>()</code> <br><code><a href="layer_group_query_attention.html">layer_group_query_attention</a>()</code> <br><code><a href="layer_gru.html">layer_gru</a>()</code> <br><code><a href="layer_hashed_crossing.html">layer_hashed_crossing</a>()</code> <br><code><a href="layer_hashing.html">layer_hashing</a>()</code> <br><code><a href="layer_identity.html">layer_identity</a>()</code> <br><code><a href="layer_integer_lookup.html">layer_integer_lookup</a>()</code> <br><code><a href="layer_jax_model_wrapper.html">layer_jax_model_wrapper</a>()</code> <br><code><a href="layer_lambda.html">layer_lambda</a>()</code> <br><code><a href="layer_layer_normalization.html">layer_layer_normalization</a>()</code> <br><code><a href="layer_lstm.html">layer_lstm</a>()</code> <br><code><a href="layer_masking.html">layer_masking</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_max_pooling_1d.html">layer_max_pooling_1d</a>()</code> <br><code><a href="layer_max_pooling_2d.html">layer_max_pooling_2d</a>()</code> <br><code><a href="layer_max_pooling_3d.html">layer_max_pooling_3d</a>()</code> <br><code><a href="layer_maximum.html">layer_maximum</a>()</code> <br><code><a href="layer_mel_spectrogram.html">layer_mel_spectrogram</a>()</code> <br><code><a href="layer_minimum.html">layer_minimum</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_multi_head_attention.html">layer_multi_head_attention</a>()</code> <br><code><a href="layer_multiply.html">layer_multiply</a>()</code> <br><code><a href="layer_normalization.html">layer_normalization</a>()</code> <br><code><a href="layer_permute.html">layer_permute</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_brightness.html">layer_random_brightness</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_contrast.html">layer_random_contrast</a>()</code> <br><code><a href="layer_random_crop.html">layer_random_crop</a>()</code> <br><code><a href="layer_random_elastic_transform.html">layer_random_elastic_transform</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_flip.html">layer_random_flip</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_rotation.html">layer_random_rotation</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_random_translation.html">layer_random_translation</a>()</code> <br><code><a href="layer_repeat_vector.html">layer_repeat_vector</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_reshape.html">layer_reshape</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_rms_normalization.html">layer_rms_normalization</a>()</code> <br><code><a href="layer_rnn.html">layer_rnn</a>()</code> <br><code><a href="layer_separable_conv_1d.html">layer_separable_conv_1d</a>()</code> <br><code><a href="layer_separable_conv_2d.html">layer_separable_conv_2d</a>()</code> <br><code><a href="layer_simple_rnn.html">layer_simple_rnn</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br><code><a href="layer_spatial_dropout_1d.html">layer_spatial_dropout_1d</a>()</code> <br><code><a href="layer_spatial_dropout_2d.html">layer_spatial_dropout_2d</a>()</code> <br><code><a href="layer_spatial_dropout_3d.html">layer_spatial_dropout_3d</a>()</code> <br><code><a href="layer_spectral_normalization.html">layer_spectral_normalization</a>()</code> <br><code><a href="layer_stft_spectrogram.html">layer_stft_spectrogram</a>()</code> <br><code><a href="layer_string_lookup.html">layer_string_lookup</a>()</code> <br><code><a href="layer_subtract.html">layer_subtract</a>()</code> <br><code><a href="layer_text_vectorization.html">layer_text_vectorization</a>()</code> <br><code><a href="layer_tfsm.html">layer_tfsm</a>()</code> <br><code><a href="layer_time_distributed.html">layer_time_distributed</a>()</code> <br><code><a href="layer_torch_module_wrapper.html">layer_torch_module_wrapper</a>()</code> <br><code><a href="layer_unit_normalization.html">layer_unit_normalization</a>()</code> <br><code><a href="layer_upsampling_1d.html">layer_upsampling_1d</a>()</code> <br><code><a href="layer_upsampling_2d.html">layer_upsampling_2d</a>()</code> <br><code><a href="layer_upsampling_3d.html">layer_upsampling_3d</a>()</code> <br><code><a href="layer_zero_padding_1d.html">layer_zero_padding_1d</a>()</code> <br><code><a href="layer_zero_padding_2d.html">layer_zero_padding_2d</a>()</code> <br><code><a href="layer_zero_padding_3d.html">layer_zero_padding_3d</a>()</code> <br><code><a href="rnn_cell_gru.html">rnn_cell_gru</a>()</code> <br><code><a href="rnn_cell_lstm.html">rnn_cell_lstm</a>()</code> <br><code><a href="rnn_cell_simple.html">rnn_cell_simple</a>()</code> <br><code><a href="rnn_cells_stack.html">rnn_cells_stack</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -213,7 +213,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/layer_repeat_vector.html b/docs/dev/reference/layer_repeat_vector.html
index 72dacbcd1..647a37787 100644
--- a/docs/dev/reference/layer_repeat_vector.html
+++ b/docs/dev/reference/layer_repeat_vector.html
@@ -8,7 +8,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -108,7 +108,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
     <div class="dont-index">
 <ul><li><p><a href="https://keras.io/api/layers/reshaping_layers/repeat_vector#repeatvector-class" class="external-link">https://keras.io/api/layers/reshaping_layers/repeat_vector#repeatvector-class</a></p></li>
 </ul><p>Other reshaping layers: <br><code><a href="layer_cropping_1d.html">layer_cropping_1d</a>()</code> <br><code><a href="layer_cropping_2d.html">layer_cropping_2d</a>()</code> <br><code><a href="layer_cropping_3d.html">layer_cropping_3d</a>()</code> <br><code><a href="layer_flatten.html">layer_flatten</a>()</code> <br><code><a href="layer_permute.html">layer_permute</a>()</code> <br><code><a href="layer_reshape.html">layer_reshape</a>()</code> <br><code><a href="layer_upsampling_1d.html">layer_upsampling_1d</a>()</code> <br><code><a href="layer_upsampling_2d.html">layer_upsampling_2d</a>()</code> <br><code><a href="layer_upsampling_3d.html">layer_upsampling_3d</a>()</code> <br><code><a href="layer_zero_padding_1d.html">layer_zero_padding_1d</a>()</code> <br><code><a href="layer_zero_padding_2d.html">layer_zero_padding_2d</a>()</code> <br><code><a href="layer_zero_padding_3d.html">layer_zero_padding_3d</a>()</code> <br></p>
-<p>Other layers: <br><code><a href="Layer.html">Layer</a>()</code> <br><code><a href="layer_activation.html">layer_activation</a>()</code> <br><code><a href="layer_activation_elu.html">layer_activation_elu</a>()</code> <br><code><a href="layer_activation_leaky_relu.html">layer_activation_leaky_relu</a>()</code> <br><code><a href="layer_activation_parametric_relu.html">layer_activation_parametric_relu</a>()</code> <br><code><a href="layer_activation_relu.html">layer_activation_relu</a>()</code> <br><code><a href="layer_activation_softmax.html">layer_activation_softmax</a>()</code> <br><code><a href="layer_activity_regularization.html">layer_activity_regularization</a>()</code> <br><code><a href="layer_add.html">layer_add</a>()</code> <br><code><a href="layer_additive_attention.html">layer_additive_attention</a>()</code> <br><code><a href="layer_alpha_dropout.html">layer_alpha_dropout</a>()</code> <br><code><a href="layer_attention.html">layer_attention</a>()</code> <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_average.html">layer_average</a>()</code> <br><code><a href="layer_average_pooling_1d.html">layer_average_pooling_1d</a>()</code> <br><code><a href="layer_average_pooling_2d.html">layer_average_pooling_2d</a>()</code> <br><code><a href="layer_average_pooling_3d.html">layer_average_pooling_3d</a>()</code> <br><code><a href="layer_batch_normalization.html">layer_batch_normalization</a>()</code> <br><code><a href="layer_bidirectional.html">layer_bidirectional</a>()</code> <br><code><a href="layer_category_encoding.html">layer_category_encoding</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_concatenate.html">layer_concatenate</a>()</code> <br><code><a href="layer_conv_1d.html">layer_conv_1d</a>()</code> <br><code><a href="layer_conv_1d_transpose.html">layer_conv_1d_transpose</a>()</code> <br><code><a href="layer_conv_2d.html">layer_conv_2d</a>()</code> <br><code><a href="layer_conv_2d_transpose.html">layer_conv_2d_transpose</a>()</code> <br><code><a href="layer_conv_3d.html">layer_conv_3d</a>()</code> <br><code><a href="layer_conv_3d_transpose.html">layer_conv_3d_transpose</a>()</code> <br><code><a href="layer_conv_lstm_1d.html">layer_conv_lstm_1d</a>()</code> <br><code><a href="layer_conv_lstm_2d.html">layer_conv_lstm_2d</a>()</code> <br><code><a href="layer_conv_lstm_3d.html">layer_conv_lstm_3d</a>()</code> <br><code><a href="layer_cropping_1d.html">layer_cropping_1d</a>()</code> <br><code><a href="layer_cropping_2d.html">layer_cropping_2d</a>()</code> <br><code><a href="layer_cropping_3d.html">layer_cropping_3d</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_dense.html">layer_dense</a>()</code> <br><code><a href="layer_depthwise_conv_1d.html">layer_depthwise_conv_1d</a>()</code> <br><code><a href="layer_depthwise_conv_2d.html">layer_depthwise_conv_2d</a>()</code> <br><code><a href="layer_discretization.html">layer_discretization</a>()</code> <br><code><a href="layer_dot.html">layer_dot</a>()</code> <br><code><a href="layer_dropout.html">layer_dropout</a>()</code> <br><code><a href="layer_einsum_dense.html">layer_einsum_dense</a>()</code> <br><code><a href="layer_embedding.html">layer_embedding</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_feature_space.html">layer_feature_space</a>()</code> <br><code><a href="layer_flatten.html">layer_flatten</a>()</code> <br><code><a href="layer_flax_module_wrapper.html">layer_flax_module_wrapper</a>()</code> <br><code><a href="layer_gaussian_dropout.html">layer_gaussian_dropout</a>()</code> <br><code><a href="layer_gaussian_noise.html">layer_gaussian_noise</a>()</code> <br><code><a href="layer_global_average_pooling_1d.html">layer_global_average_pooling_1d</a>()</code> <br><code><a href="layer_global_average_pooling_2d.html">layer_global_average_pooling_2d</a>()</code> <br><code><a href="layer_global_average_pooling_3d.html">layer_global_average_pooling_3d</a>()</code> <br><code><a href="layer_global_max_pooling_1d.html">layer_global_max_pooling_1d</a>()</code> <br><code><a href="layer_global_max_pooling_2d.html">layer_global_max_pooling_2d</a>()</code> <br><code><a href="layer_global_max_pooling_3d.html">layer_global_max_pooling_3d</a>()</code> <br><code><a href="layer_group_normalization.html">layer_group_normalization</a>()</code> <br><code><a href="layer_group_query_attention.html">layer_group_query_attention</a>()</code> <br><code><a href="layer_gru.html">layer_gru</a>()</code> <br><code><a href="layer_hashed_crossing.html">layer_hashed_crossing</a>()</code> <br><code><a href="layer_hashing.html">layer_hashing</a>()</code> <br><code><a href="layer_identity.html">layer_identity</a>()</code> <br><code><a href="layer_integer_lookup.html">layer_integer_lookup</a>()</code> <br><code><a href="layer_jax_model_wrapper.html">layer_jax_model_wrapper</a>()</code> <br><code><a href="layer_lambda.html">layer_lambda</a>()</code> <br><code><a href="layer_layer_normalization.html">layer_layer_normalization</a>()</code> <br><code><a href="layer_lstm.html">layer_lstm</a>()</code> <br><code><a href="layer_masking.html">layer_masking</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_max_pooling_1d.html">layer_max_pooling_1d</a>()</code> <br><code><a href="layer_max_pooling_2d.html">layer_max_pooling_2d</a>()</code> <br><code><a href="layer_max_pooling_3d.html">layer_max_pooling_3d</a>()</code> <br><code><a href="layer_maximum.html">layer_maximum</a>()</code> <br><code><a href="layer_mel_spectrogram.html">layer_mel_spectrogram</a>()</code> <br><code><a href="layer_minimum.html">layer_minimum</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_multi_head_attention.html">layer_multi_head_attention</a>()</code> <br><code><a href="layer_multiply.html">layer_multiply</a>()</code> <br><code><a href="layer_normalization.html">layer_normalization</a>()</code> <br><code><a href="layer_permute.html">layer_permute</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_brightness.html">layer_random_brightness</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_contrast.html">layer_random_contrast</a>()</code> <br><code><a href="layer_random_crop.html">layer_random_crop</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_flip.html">layer_random_flip</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_rotation.html">layer_random_rotation</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_random_translation.html">layer_random_translation</a>()</code> <br><code><a href="layer_random_zoom.html">layer_random_zoom</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_reshape.html">layer_reshape</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_rms_normalization.html">layer_rms_normalization</a>()</code> <br><code><a href="layer_rnn.html">layer_rnn</a>()</code> <br><code><a href="layer_separable_conv_1d.html">layer_separable_conv_1d</a>()</code> <br><code><a href="layer_separable_conv_2d.html">layer_separable_conv_2d</a>()</code> <br><code><a href="layer_simple_rnn.html">layer_simple_rnn</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br><code><a href="layer_spatial_dropout_1d.html">layer_spatial_dropout_1d</a>()</code> <br><code><a href="layer_spatial_dropout_2d.html">layer_spatial_dropout_2d</a>()</code> <br><code><a href="layer_spatial_dropout_3d.html">layer_spatial_dropout_3d</a>()</code> <br><code><a href="layer_spectral_normalization.html">layer_spectral_normalization</a>()</code> <br><code><a href="layer_stft_spectrogram.html">layer_stft_spectrogram</a>()</code> <br><code><a href="layer_string_lookup.html">layer_string_lookup</a>()</code> <br><code><a href="layer_subtract.html">layer_subtract</a>()</code> <br><code><a href="layer_text_vectorization.html">layer_text_vectorization</a>()</code> <br><code><a href="layer_tfsm.html">layer_tfsm</a>()</code> <br><code><a href="layer_time_distributed.html">layer_time_distributed</a>()</code> <br><code><a href="layer_torch_module_wrapper.html">layer_torch_module_wrapper</a>()</code> <br><code><a href="layer_unit_normalization.html">layer_unit_normalization</a>()</code> <br><code><a href="layer_upsampling_1d.html">layer_upsampling_1d</a>()</code> <br><code><a href="layer_upsampling_2d.html">layer_upsampling_2d</a>()</code> <br><code><a href="layer_upsampling_3d.html">layer_upsampling_3d</a>()</code> <br><code><a href="layer_zero_padding_1d.html">layer_zero_padding_1d</a>()</code> <br><code><a href="layer_zero_padding_2d.html">layer_zero_padding_2d</a>()</code> <br><code><a href="layer_zero_padding_3d.html">layer_zero_padding_3d</a>()</code> <br><code><a href="rnn_cell_gru.html">rnn_cell_gru</a>()</code> <br><code><a href="rnn_cell_lstm.html">rnn_cell_lstm</a>()</code> <br><code><a href="rnn_cell_simple.html">rnn_cell_simple</a>()</code> <br><code><a href="rnn_cells_stack.html">rnn_cells_stack</a>()</code> <br></p></div>
+<p>Other layers: <br><code><a href="Layer.html">Layer</a>()</code> <br><code><a href="layer_activation.html">layer_activation</a>()</code> <br><code><a href="layer_activation_elu.html">layer_activation_elu</a>()</code> <br><code><a href="layer_activation_leaky_relu.html">layer_activation_leaky_relu</a>()</code> <br><code><a href="layer_activation_parametric_relu.html">layer_activation_parametric_relu</a>()</code> <br><code><a href="layer_activation_relu.html">layer_activation_relu</a>()</code> <br><code><a href="layer_activation_softmax.html">layer_activation_softmax</a>()</code> <br><code><a href="layer_activity_regularization.html">layer_activity_regularization</a>()</code> <br><code><a href="layer_add.html">layer_add</a>()</code> <br><code><a href="layer_additive_attention.html">layer_additive_attention</a>()</code> <br><code><a href="layer_alpha_dropout.html">layer_alpha_dropout</a>()</code> <br><code><a href="layer_attention.html">layer_attention</a>()</code> <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_average.html">layer_average</a>()</code> <br><code><a href="layer_average_pooling_1d.html">layer_average_pooling_1d</a>()</code> <br><code><a href="layer_average_pooling_2d.html">layer_average_pooling_2d</a>()</code> <br><code><a href="layer_average_pooling_3d.html">layer_average_pooling_3d</a>()</code> <br><code><a href="layer_batch_normalization.html">layer_batch_normalization</a>()</code> <br><code><a href="layer_bidirectional.html">layer_bidirectional</a>()</code> <br><code><a href="layer_category_encoding.html">layer_category_encoding</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_concatenate.html">layer_concatenate</a>()</code> <br><code><a href="layer_conv_1d.html">layer_conv_1d</a>()</code> <br><code><a href="layer_conv_1d_transpose.html">layer_conv_1d_transpose</a>()</code> <br><code><a href="layer_conv_2d.html">layer_conv_2d</a>()</code> <br><code><a href="layer_conv_2d_transpose.html">layer_conv_2d_transpose</a>()</code> <br><code><a href="layer_conv_3d.html">layer_conv_3d</a>()</code> <br><code><a href="layer_conv_3d_transpose.html">layer_conv_3d_transpose</a>()</code> <br><code><a href="layer_conv_lstm_1d.html">layer_conv_lstm_1d</a>()</code> <br><code><a href="layer_conv_lstm_2d.html">layer_conv_lstm_2d</a>()</code> <br><code><a href="layer_conv_lstm_3d.html">layer_conv_lstm_3d</a>()</code> <br><code><a href="layer_cropping_1d.html">layer_cropping_1d</a>()</code> <br><code><a href="layer_cropping_2d.html">layer_cropping_2d</a>()</code> <br><code><a href="layer_cropping_3d.html">layer_cropping_3d</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_dense.html">layer_dense</a>()</code> <br><code><a href="layer_depthwise_conv_1d.html">layer_depthwise_conv_1d</a>()</code> <br><code><a href="layer_depthwise_conv_2d.html">layer_depthwise_conv_2d</a>()</code> <br><code><a href="layer_discretization.html">layer_discretization</a>()</code> <br><code><a href="layer_dot.html">layer_dot</a>()</code> <br><code><a href="layer_dropout.html">layer_dropout</a>()</code> <br><code><a href="layer_einsum_dense.html">layer_einsum_dense</a>()</code> <br><code><a href="layer_embedding.html">layer_embedding</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_feature_space.html">layer_feature_space</a>()</code> <br><code><a href="layer_flatten.html">layer_flatten</a>()</code> <br><code><a href="layer_flax_module_wrapper.html">layer_flax_module_wrapper</a>()</code> <br><code><a href="layer_gaussian_dropout.html">layer_gaussian_dropout</a>()</code> <br><code><a href="layer_gaussian_noise.html">layer_gaussian_noise</a>()</code> <br><code><a href="layer_global_average_pooling_1d.html">layer_global_average_pooling_1d</a>()</code> <br><code><a href="layer_global_average_pooling_2d.html">layer_global_average_pooling_2d</a>()</code> <br><code><a href="layer_global_average_pooling_3d.html">layer_global_average_pooling_3d</a>()</code> <br><code><a href="layer_global_max_pooling_1d.html">layer_global_max_pooling_1d</a>()</code> <br><code><a href="layer_global_max_pooling_2d.html">layer_global_max_pooling_2d</a>()</code> <br><code><a href="layer_global_max_pooling_3d.html">layer_global_max_pooling_3d</a>()</code> <br><code><a href="layer_group_normalization.html">layer_group_normalization</a>()</code> <br><code><a href="layer_group_query_attention.html">layer_group_query_attention</a>()</code> <br><code><a href="layer_gru.html">layer_gru</a>()</code> <br><code><a href="layer_hashed_crossing.html">layer_hashed_crossing</a>()</code> <br><code><a href="layer_hashing.html">layer_hashing</a>()</code> <br><code><a href="layer_identity.html">layer_identity</a>()</code> <br><code><a href="layer_integer_lookup.html">layer_integer_lookup</a>()</code> <br><code><a href="layer_jax_model_wrapper.html">layer_jax_model_wrapper</a>()</code> <br><code><a href="layer_lambda.html">layer_lambda</a>()</code> <br><code><a href="layer_layer_normalization.html">layer_layer_normalization</a>()</code> <br><code><a href="layer_lstm.html">layer_lstm</a>()</code> <br><code><a href="layer_masking.html">layer_masking</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_max_pooling_1d.html">layer_max_pooling_1d</a>()</code> <br><code><a href="layer_max_pooling_2d.html">layer_max_pooling_2d</a>()</code> <br><code><a href="layer_max_pooling_3d.html">layer_max_pooling_3d</a>()</code> <br><code><a href="layer_maximum.html">layer_maximum</a>()</code> <br><code><a href="layer_mel_spectrogram.html">layer_mel_spectrogram</a>()</code> <br><code><a href="layer_minimum.html">layer_minimum</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_multi_head_attention.html">layer_multi_head_attention</a>()</code> <br><code><a href="layer_multiply.html">layer_multiply</a>()</code> <br><code><a href="layer_normalization.html">layer_normalization</a>()</code> <br><code><a href="layer_permute.html">layer_permute</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_brightness.html">layer_random_brightness</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_contrast.html">layer_random_contrast</a>()</code> <br><code><a href="layer_random_crop.html">layer_random_crop</a>()</code> <br><code><a href="layer_random_elastic_transform.html">layer_random_elastic_transform</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_flip.html">layer_random_flip</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_rotation.html">layer_random_rotation</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_random_translation.html">layer_random_translation</a>()</code> <br><code><a href="layer_random_zoom.html">layer_random_zoom</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_reshape.html">layer_reshape</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_rms_normalization.html">layer_rms_normalization</a>()</code> <br><code><a href="layer_rnn.html">layer_rnn</a>()</code> <br><code><a href="layer_separable_conv_1d.html">layer_separable_conv_1d</a>()</code> <br><code><a href="layer_separable_conv_2d.html">layer_separable_conv_2d</a>()</code> <br><code><a href="layer_simple_rnn.html">layer_simple_rnn</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br><code><a href="layer_spatial_dropout_1d.html">layer_spatial_dropout_1d</a>()</code> <br><code><a href="layer_spatial_dropout_2d.html">layer_spatial_dropout_2d</a>()</code> <br><code><a href="layer_spatial_dropout_3d.html">layer_spatial_dropout_3d</a>()</code> <br><code><a href="layer_spectral_normalization.html">layer_spectral_normalization</a>()</code> <br><code><a href="layer_stft_spectrogram.html">layer_stft_spectrogram</a>()</code> <br><code><a href="layer_string_lookup.html">layer_string_lookup</a>()</code> <br><code><a href="layer_subtract.html">layer_subtract</a>()</code> <br><code><a href="layer_text_vectorization.html">layer_text_vectorization</a>()</code> <br><code><a href="layer_tfsm.html">layer_tfsm</a>()</code> <br><code><a href="layer_time_distributed.html">layer_time_distributed</a>()</code> <br><code><a href="layer_torch_module_wrapper.html">layer_torch_module_wrapper</a>()</code> <br><code><a href="layer_unit_normalization.html">layer_unit_normalization</a>()</code> <br><code><a href="layer_upsampling_1d.html">layer_upsampling_1d</a>()</code> <br><code><a href="layer_upsampling_2d.html">layer_upsampling_2d</a>()</code> <br><code><a href="layer_upsampling_3d.html">layer_upsampling_3d</a>()</code> <br><code><a href="layer_zero_padding_1d.html">layer_zero_padding_1d</a>()</code> <br><code><a href="layer_zero_padding_2d.html">layer_zero_padding_2d</a>()</code> <br><code><a href="layer_zero_padding_3d.html">layer_zero_padding_3d</a>()</code> <br><code><a href="rnn_cell_gru.html">rnn_cell_gru</a>()</code> <br><code><a href="rnn_cell_lstm.html">rnn_cell_lstm</a>()</code> <br><code><a href="rnn_cell_simple.html">rnn_cell_simple</a>()</code> <br><code><a href="rnn_cells_stack.html">rnn_cells_stack</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -120,7 +120,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/layer_rescaling.html b/docs/dev/reference/layer_rescaling.html
index 4d1522bcb..b6b6ac51a 100644
--- a/docs/dev/reference/layer_rescaling.html
+++ b/docs/dev/reference/layer_rescaling.html
@@ -34,7 +34,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -107,11 +107,15 @@ <h2 id="arguments">Arguments<a class="anchor" aria-label="anchor" href="#argumen
 
 
 <dt id="arg-scale">scale<a class="anchor" aria-label="anchor" href="#arg-scale"></a></dt>
-<dd><p>Float, the scale to apply to the inputs.</p></dd>
+<dd><p>Numeric scalar, vector, or array. The scale to apply to the inputs. If
+scalar, the same scale is applied to every feature or channel; if a vector
+or array, scaling is applied per channel.</p></dd>
 
 
 <dt id="arg-offset">offset<a class="anchor" aria-label="anchor" href="#arg-offset"></a></dt>
-<dd><p>Float, the offset to apply to the inputs.</p></dd>
+<dd><p>Numeric scalar, vector, or array. The offset to apply to the inputs. If
+scalar, the same offset is applied to every feature or channel; if a vector
+or array, the shift is applied per channel.</p></dd>
 
 
 <dt id="arg--">...<a class="anchor" aria-label="anchor" href="#arg--"></a></dt>
@@ -131,9 +135,9 @@ <h2 id="value">Value<a class="anchor" aria-label="anchor" href="#value"></a></h2
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
     <div class="dont-index">
 <ul><li><p><a href="https://keras.io/api/layers/preprocessing_layers/image_preprocessing/rescaling#rescaling-class" class="external-link">https://keras.io/api/layers/preprocessing_layers/image_preprocessing/rescaling#rescaling-class</a></p></li>
-</ul><p>Other image preprocessing layers: <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br></p>
-<p>Other preprocessing layers: <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_category_encoding.html">layer_category_encoding</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_discretization.html">layer_discretization</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_feature_space.html">layer_feature_space</a>()</code> <br><code><a href="layer_hashed_crossing.html">layer_hashed_crossing</a>()</code> <br><code><a href="layer_hashing.html">layer_hashing</a>()</code> <br><code><a href="layer_integer_lookup.html">layer_integer_lookup</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_mel_spectrogram.html">layer_mel_spectrogram</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_normalization.html">layer_normalization</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_brightness.html">layer_random_brightness</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_contrast.html">layer_random_contrast</a>()</code> <br><code><a href="layer_random_crop.html">layer_random_crop</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_flip.html">layer_random_flip</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_rotation.html">layer_random_rotation</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_random_translation.html">layer_random_translation</a>()</code> <br><code><a href="layer_random_zoom.html">layer_random_zoom</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br><code><a href="layer_stft_spectrogram.html">layer_stft_spectrogram</a>()</code> <br><code><a href="layer_string_lookup.html">layer_string_lookup</a>()</code> <br><code><a href="layer_text_vectorization.html">layer_text_vectorization</a>()</code> <br></p>
-<p>Other layers: <br><code><a href="Layer.html">Layer</a>()</code> <br><code><a href="layer_activation.html">layer_activation</a>()</code> <br><code><a href="layer_activation_elu.html">layer_activation_elu</a>()</code> <br><code><a href="layer_activation_leaky_relu.html">layer_activation_leaky_relu</a>()</code> <br><code><a href="layer_activation_parametric_relu.html">layer_activation_parametric_relu</a>()</code> <br><code><a href="layer_activation_relu.html">layer_activation_relu</a>()</code> <br><code><a href="layer_activation_softmax.html">layer_activation_softmax</a>()</code> <br><code><a href="layer_activity_regularization.html">layer_activity_regularization</a>()</code> <br><code><a href="layer_add.html">layer_add</a>()</code> <br><code><a href="layer_additive_attention.html">layer_additive_attention</a>()</code> <br><code><a href="layer_alpha_dropout.html">layer_alpha_dropout</a>()</code> <br><code><a href="layer_attention.html">layer_attention</a>()</code> <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_average.html">layer_average</a>()</code> <br><code><a href="layer_average_pooling_1d.html">layer_average_pooling_1d</a>()</code> <br><code><a href="layer_average_pooling_2d.html">layer_average_pooling_2d</a>()</code> <br><code><a href="layer_average_pooling_3d.html">layer_average_pooling_3d</a>()</code> <br><code><a href="layer_batch_normalization.html">layer_batch_normalization</a>()</code> <br><code><a href="layer_bidirectional.html">layer_bidirectional</a>()</code> <br><code><a href="layer_category_encoding.html">layer_category_encoding</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_concatenate.html">layer_concatenate</a>()</code> <br><code><a href="layer_conv_1d.html">layer_conv_1d</a>()</code> <br><code><a href="layer_conv_1d_transpose.html">layer_conv_1d_transpose</a>()</code> <br><code><a href="layer_conv_2d.html">layer_conv_2d</a>()</code> <br><code><a href="layer_conv_2d_transpose.html">layer_conv_2d_transpose</a>()</code> <br><code><a href="layer_conv_3d.html">layer_conv_3d</a>()</code> <br><code><a href="layer_conv_3d_transpose.html">layer_conv_3d_transpose</a>()</code> <br><code><a href="layer_conv_lstm_1d.html">layer_conv_lstm_1d</a>()</code> <br><code><a href="layer_conv_lstm_2d.html">layer_conv_lstm_2d</a>()</code> <br><code><a href="layer_conv_lstm_3d.html">layer_conv_lstm_3d</a>()</code> <br><code><a href="layer_cropping_1d.html">layer_cropping_1d</a>()</code> <br><code><a href="layer_cropping_2d.html">layer_cropping_2d</a>()</code> <br><code><a href="layer_cropping_3d.html">layer_cropping_3d</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_dense.html">layer_dense</a>()</code> <br><code><a href="layer_depthwise_conv_1d.html">layer_depthwise_conv_1d</a>()</code> <br><code><a href="layer_depthwise_conv_2d.html">layer_depthwise_conv_2d</a>()</code> <br><code><a href="layer_discretization.html">layer_discretization</a>()</code> <br><code><a href="layer_dot.html">layer_dot</a>()</code> <br><code><a href="layer_dropout.html">layer_dropout</a>()</code> <br><code><a href="layer_einsum_dense.html">layer_einsum_dense</a>()</code> <br><code><a href="layer_embedding.html">layer_embedding</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_feature_space.html">layer_feature_space</a>()</code> <br><code><a href="layer_flatten.html">layer_flatten</a>()</code> <br><code><a href="layer_flax_module_wrapper.html">layer_flax_module_wrapper</a>()</code> <br><code><a href="layer_gaussian_dropout.html">layer_gaussian_dropout</a>()</code> <br><code><a href="layer_gaussian_noise.html">layer_gaussian_noise</a>()</code> <br><code><a href="layer_global_average_pooling_1d.html">layer_global_average_pooling_1d</a>()</code> <br><code><a href="layer_global_average_pooling_2d.html">layer_global_average_pooling_2d</a>()</code> <br><code><a href="layer_global_average_pooling_3d.html">layer_global_average_pooling_3d</a>()</code> <br><code><a href="layer_global_max_pooling_1d.html">layer_global_max_pooling_1d</a>()</code> <br><code><a href="layer_global_max_pooling_2d.html">layer_global_max_pooling_2d</a>()</code> <br><code><a href="layer_global_max_pooling_3d.html">layer_global_max_pooling_3d</a>()</code> <br><code><a href="layer_group_normalization.html">layer_group_normalization</a>()</code> <br><code><a href="layer_group_query_attention.html">layer_group_query_attention</a>()</code> <br><code><a href="layer_gru.html">layer_gru</a>()</code> <br><code><a href="layer_hashed_crossing.html">layer_hashed_crossing</a>()</code> <br><code><a href="layer_hashing.html">layer_hashing</a>()</code> <br><code><a href="layer_identity.html">layer_identity</a>()</code> <br><code><a href="layer_integer_lookup.html">layer_integer_lookup</a>()</code> <br><code><a href="layer_jax_model_wrapper.html">layer_jax_model_wrapper</a>()</code> <br><code><a href="layer_lambda.html">layer_lambda</a>()</code> <br><code><a href="layer_layer_normalization.html">layer_layer_normalization</a>()</code> <br><code><a href="layer_lstm.html">layer_lstm</a>()</code> <br><code><a href="layer_masking.html">layer_masking</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_max_pooling_1d.html">layer_max_pooling_1d</a>()</code> <br><code><a href="layer_max_pooling_2d.html">layer_max_pooling_2d</a>()</code> <br><code><a href="layer_max_pooling_3d.html">layer_max_pooling_3d</a>()</code> <br><code><a href="layer_maximum.html">layer_maximum</a>()</code> <br><code><a href="layer_mel_spectrogram.html">layer_mel_spectrogram</a>()</code> <br><code><a href="layer_minimum.html">layer_minimum</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_multi_head_attention.html">layer_multi_head_attention</a>()</code> <br><code><a href="layer_multiply.html">layer_multiply</a>()</code> <br><code><a href="layer_normalization.html">layer_normalization</a>()</code> <br><code><a href="layer_permute.html">layer_permute</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_brightness.html">layer_random_brightness</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_contrast.html">layer_random_contrast</a>()</code> <br><code><a href="layer_random_crop.html">layer_random_crop</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_flip.html">layer_random_flip</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_rotation.html">layer_random_rotation</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_random_translation.html">layer_random_translation</a>()</code> <br><code><a href="layer_random_zoom.html">layer_random_zoom</a>()</code> <br><code><a href="layer_repeat_vector.html">layer_repeat_vector</a>()</code> <br><code><a href="layer_reshape.html">layer_reshape</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_rms_normalization.html">layer_rms_normalization</a>()</code> <br><code><a href="layer_rnn.html">layer_rnn</a>()</code> <br><code><a href="layer_separable_conv_1d.html">layer_separable_conv_1d</a>()</code> <br><code><a href="layer_separable_conv_2d.html">layer_separable_conv_2d</a>()</code> <br><code><a href="layer_simple_rnn.html">layer_simple_rnn</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br><code><a href="layer_spatial_dropout_1d.html">layer_spatial_dropout_1d</a>()</code> <br><code><a href="layer_spatial_dropout_2d.html">layer_spatial_dropout_2d</a>()</code> <br><code><a href="layer_spatial_dropout_3d.html">layer_spatial_dropout_3d</a>()</code> <br><code><a href="layer_spectral_normalization.html">layer_spectral_normalization</a>()</code> <br><code><a href="layer_stft_spectrogram.html">layer_stft_spectrogram</a>()</code> <br><code><a href="layer_string_lookup.html">layer_string_lookup</a>()</code> <br><code><a href="layer_subtract.html">layer_subtract</a>()</code> <br><code><a href="layer_text_vectorization.html">layer_text_vectorization</a>()</code> <br><code><a href="layer_tfsm.html">layer_tfsm</a>()</code> <br><code><a href="layer_time_distributed.html">layer_time_distributed</a>()</code> <br><code><a href="layer_torch_module_wrapper.html">layer_torch_module_wrapper</a>()</code> <br><code><a href="layer_unit_normalization.html">layer_unit_normalization</a>()</code> <br><code><a href="layer_upsampling_1d.html">layer_upsampling_1d</a>()</code> <br><code><a href="layer_upsampling_2d.html">layer_upsampling_2d</a>()</code> <br><code><a href="layer_upsampling_3d.html">layer_upsampling_3d</a>()</code> <br><code><a href="layer_zero_padding_1d.html">layer_zero_padding_1d</a>()</code> <br><code><a href="layer_zero_padding_2d.html">layer_zero_padding_2d</a>()</code> <br><code><a href="layer_zero_padding_3d.html">layer_zero_padding_3d</a>()</code> <br><code><a href="rnn_cell_gru.html">rnn_cell_gru</a>()</code> <br><code><a href="rnn_cell_lstm.html">rnn_cell_lstm</a>()</code> <br><code><a href="rnn_cell_simple.html">rnn_cell_simple</a>()</code> <br><code><a href="rnn_cells_stack.html">rnn_cells_stack</a>()</code> <br></p></div>
+</ul><p>Other image preprocessing layers: <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_elastic_transform.html">layer_random_elastic_transform</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br></p>
+<p>Other preprocessing layers: <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_category_encoding.html">layer_category_encoding</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_discretization.html">layer_discretization</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_feature_space.html">layer_feature_space</a>()</code> <br><code><a href="layer_hashed_crossing.html">layer_hashed_crossing</a>()</code> <br><code><a href="layer_hashing.html">layer_hashing</a>()</code> <br><code><a href="layer_integer_lookup.html">layer_integer_lookup</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_mel_spectrogram.html">layer_mel_spectrogram</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_normalization.html">layer_normalization</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_brightness.html">layer_random_brightness</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_contrast.html">layer_random_contrast</a>()</code> <br><code><a href="layer_random_crop.html">layer_random_crop</a>()</code> <br><code><a href="layer_random_elastic_transform.html">layer_random_elastic_transform</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_flip.html">layer_random_flip</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_rotation.html">layer_random_rotation</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_random_translation.html">layer_random_translation</a>()</code> <br><code><a href="layer_random_zoom.html">layer_random_zoom</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br><code><a href="layer_stft_spectrogram.html">layer_stft_spectrogram</a>()</code> <br><code><a href="layer_string_lookup.html">layer_string_lookup</a>()</code> <br><code><a href="layer_text_vectorization.html">layer_text_vectorization</a>()</code> <br></p>
+<p>Other layers: <br><code><a href="Layer.html">Layer</a>()</code> <br><code><a href="layer_activation.html">layer_activation</a>()</code> <br><code><a href="layer_activation_elu.html">layer_activation_elu</a>()</code> <br><code><a href="layer_activation_leaky_relu.html">layer_activation_leaky_relu</a>()</code> <br><code><a href="layer_activation_parametric_relu.html">layer_activation_parametric_relu</a>()</code> <br><code><a href="layer_activation_relu.html">layer_activation_relu</a>()</code> <br><code><a href="layer_activation_softmax.html">layer_activation_softmax</a>()</code> <br><code><a href="layer_activity_regularization.html">layer_activity_regularization</a>()</code> <br><code><a href="layer_add.html">layer_add</a>()</code> <br><code><a href="layer_additive_attention.html">layer_additive_attention</a>()</code> <br><code><a href="layer_alpha_dropout.html">layer_alpha_dropout</a>()</code> <br><code><a href="layer_attention.html">layer_attention</a>()</code> <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_average.html">layer_average</a>()</code> <br><code><a href="layer_average_pooling_1d.html">layer_average_pooling_1d</a>()</code> <br><code><a href="layer_average_pooling_2d.html">layer_average_pooling_2d</a>()</code> <br><code><a href="layer_average_pooling_3d.html">layer_average_pooling_3d</a>()</code> <br><code><a href="layer_batch_normalization.html">layer_batch_normalization</a>()</code> <br><code><a href="layer_bidirectional.html">layer_bidirectional</a>()</code> <br><code><a href="layer_category_encoding.html">layer_category_encoding</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_concatenate.html">layer_concatenate</a>()</code> <br><code><a href="layer_conv_1d.html">layer_conv_1d</a>()</code> <br><code><a href="layer_conv_1d_transpose.html">layer_conv_1d_transpose</a>()</code> <br><code><a href="layer_conv_2d.html">layer_conv_2d</a>()</code> <br><code><a href="layer_conv_2d_transpose.html">layer_conv_2d_transpose</a>()</code> <br><code><a href="layer_conv_3d.html">layer_conv_3d</a>()</code> <br><code><a href="layer_conv_3d_transpose.html">layer_conv_3d_transpose</a>()</code> <br><code><a href="layer_conv_lstm_1d.html">layer_conv_lstm_1d</a>()</code> <br><code><a href="layer_conv_lstm_2d.html">layer_conv_lstm_2d</a>()</code> <br><code><a href="layer_conv_lstm_3d.html">layer_conv_lstm_3d</a>()</code> <br><code><a href="layer_cropping_1d.html">layer_cropping_1d</a>()</code> <br><code><a href="layer_cropping_2d.html">layer_cropping_2d</a>()</code> <br><code><a href="layer_cropping_3d.html">layer_cropping_3d</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_dense.html">layer_dense</a>()</code> <br><code><a href="layer_depthwise_conv_1d.html">layer_depthwise_conv_1d</a>()</code> <br><code><a href="layer_depthwise_conv_2d.html">layer_depthwise_conv_2d</a>()</code> <br><code><a href="layer_discretization.html">layer_discretization</a>()</code> <br><code><a href="layer_dot.html">layer_dot</a>()</code> <br><code><a href="layer_dropout.html">layer_dropout</a>()</code> <br><code><a href="layer_einsum_dense.html">layer_einsum_dense</a>()</code> <br><code><a href="layer_embedding.html">layer_embedding</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_feature_space.html">layer_feature_space</a>()</code> <br><code><a href="layer_flatten.html">layer_flatten</a>()</code> <br><code><a href="layer_flax_module_wrapper.html">layer_flax_module_wrapper</a>()</code> <br><code><a href="layer_gaussian_dropout.html">layer_gaussian_dropout</a>()</code> <br><code><a href="layer_gaussian_noise.html">layer_gaussian_noise</a>()</code> <br><code><a href="layer_global_average_pooling_1d.html">layer_global_average_pooling_1d</a>()</code> <br><code><a href="layer_global_average_pooling_2d.html">layer_global_average_pooling_2d</a>()</code> <br><code><a href="layer_global_average_pooling_3d.html">layer_global_average_pooling_3d</a>()</code> <br><code><a href="layer_global_max_pooling_1d.html">layer_global_max_pooling_1d</a>()</code> <br><code><a href="layer_global_max_pooling_2d.html">layer_global_max_pooling_2d</a>()</code> <br><code><a href="layer_global_max_pooling_3d.html">layer_global_max_pooling_3d</a>()</code> <br><code><a href="layer_group_normalization.html">layer_group_normalization</a>()</code> <br><code><a href="layer_group_query_attention.html">layer_group_query_attention</a>()</code> <br><code><a href="layer_gru.html">layer_gru</a>()</code> <br><code><a href="layer_hashed_crossing.html">layer_hashed_crossing</a>()</code> <br><code><a href="layer_hashing.html">layer_hashing</a>()</code> <br><code><a href="layer_identity.html">layer_identity</a>()</code> <br><code><a href="layer_integer_lookup.html">layer_integer_lookup</a>()</code> <br><code><a href="layer_jax_model_wrapper.html">layer_jax_model_wrapper</a>()</code> <br><code><a href="layer_lambda.html">layer_lambda</a>()</code> <br><code><a href="layer_layer_normalization.html">layer_layer_normalization</a>()</code> <br><code><a href="layer_lstm.html">layer_lstm</a>()</code> <br><code><a href="layer_masking.html">layer_masking</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_max_pooling_1d.html">layer_max_pooling_1d</a>()</code> <br><code><a href="layer_max_pooling_2d.html">layer_max_pooling_2d</a>()</code> <br><code><a href="layer_max_pooling_3d.html">layer_max_pooling_3d</a>()</code> <br><code><a href="layer_maximum.html">layer_maximum</a>()</code> <br><code><a href="layer_mel_spectrogram.html">layer_mel_spectrogram</a>()</code> <br><code><a href="layer_minimum.html">layer_minimum</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_multi_head_attention.html">layer_multi_head_attention</a>()</code> <br><code><a href="layer_multiply.html">layer_multiply</a>()</code> <br><code><a href="layer_normalization.html">layer_normalization</a>()</code> <br><code><a href="layer_permute.html">layer_permute</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_brightness.html">layer_random_brightness</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_contrast.html">layer_random_contrast</a>()</code> <br><code><a href="layer_random_crop.html">layer_random_crop</a>()</code> <br><code><a href="layer_random_elastic_transform.html">layer_random_elastic_transform</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_flip.html">layer_random_flip</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_rotation.html">layer_random_rotation</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_random_translation.html">layer_random_translation</a>()</code> <br><code><a href="layer_random_zoom.html">layer_random_zoom</a>()</code> <br><code><a href="layer_repeat_vector.html">layer_repeat_vector</a>()</code> <br><code><a href="layer_reshape.html">layer_reshape</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_rms_normalization.html">layer_rms_normalization</a>()</code> <br><code><a href="layer_rnn.html">layer_rnn</a>()</code> <br><code><a href="layer_separable_conv_1d.html">layer_separable_conv_1d</a>()</code> <br><code><a href="layer_separable_conv_2d.html">layer_separable_conv_2d</a>()</code> <br><code><a href="layer_simple_rnn.html">layer_simple_rnn</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br><code><a href="layer_spatial_dropout_1d.html">layer_spatial_dropout_1d</a>()</code> <br><code><a href="layer_spatial_dropout_2d.html">layer_spatial_dropout_2d</a>()</code> <br><code><a href="layer_spatial_dropout_3d.html">layer_spatial_dropout_3d</a>()</code> <br><code><a href="layer_spectral_normalization.html">layer_spectral_normalization</a>()</code> <br><code><a href="layer_stft_spectrogram.html">layer_stft_spectrogram</a>()</code> <br><code><a href="layer_string_lookup.html">layer_string_lookup</a>()</code> <br><code><a href="layer_subtract.html">layer_subtract</a>()</code> <br><code><a href="layer_text_vectorization.html">layer_text_vectorization</a>()</code> <br><code><a href="layer_tfsm.html">layer_tfsm</a>()</code> <br><code><a href="layer_time_distributed.html">layer_time_distributed</a>()</code> <br><code><a href="layer_torch_module_wrapper.html">layer_torch_module_wrapper</a>()</code> <br><code><a href="layer_unit_normalization.html">layer_unit_normalization</a>()</code> <br><code><a href="layer_upsampling_1d.html">layer_upsampling_1d</a>()</code> <br><code><a href="layer_upsampling_2d.html">layer_upsampling_2d</a>()</code> <br><code><a href="layer_upsampling_3d.html">layer_upsampling_3d</a>()</code> <br><code><a href="layer_zero_padding_1d.html">layer_zero_padding_1d</a>()</code> <br><code><a href="layer_zero_padding_2d.html">layer_zero_padding_2d</a>()</code> <br><code><a href="layer_zero_padding_3d.html">layer_zero_padding_3d</a>()</code> <br><code><a href="rnn_cell_gru.html">rnn_cell_gru</a>()</code> <br><code><a href="rnn_cell_lstm.html">rnn_cell_lstm</a>()</code> <br><code><a href="rnn_cell_simple.html">rnn_cell_simple</a>()</code> <br><code><a href="rnn_cells_stack.html">rnn_cells_stack</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -145,7 +149,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/layer_reshape.html b/docs/dev/reference/layer_reshape.html
index 39c9fa5ea..f49df313e 100644
--- a/docs/dev/reference/layer_reshape.html
+++ b/docs/dev/reference/layer_reshape.html
@@ -8,7 +8,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -117,7 +117,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
     <div class="dont-index">
 <ul><li><p><a href="https://keras.io/api/layers/reshaping_layers/reshape#reshape-class" class="external-link">https://keras.io/api/layers/reshaping_layers/reshape#reshape-class</a></p></li>
 </ul><p>Other reshaping layers: <br><code><a href="layer_cropping_1d.html">layer_cropping_1d</a>()</code> <br><code><a href="layer_cropping_2d.html">layer_cropping_2d</a>()</code> <br><code><a href="layer_cropping_3d.html">layer_cropping_3d</a>()</code> <br><code><a href="layer_flatten.html">layer_flatten</a>()</code> <br><code><a href="layer_permute.html">layer_permute</a>()</code> <br><code><a href="layer_repeat_vector.html">layer_repeat_vector</a>()</code> <br><code><a href="layer_upsampling_1d.html">layer_upsampling_1d</a>()</code> <br><code><a href="layer_upsampling_2d.html">layer_upsampling_2d</a>()</code> <br><code><a href="layer_upsampling_3d.html">layer_upsampling_3d</a>()</code> <br><code><a href="layer_zero_padding_1d.html">layer_zero_padding_1d</a>()</code> <br><code><a href="layer_zero_padding_2d.html">layer_zero_padding_2d</a>()</code> <br><code><a href="layer_zero_padding_3d.html">layer_zero_padding_3d</a>()</code> <br></p>
-<p>Other layers: <br><code><a href="Layer.html">Layer</a>()</code> <br><code><a href="layer_activation.html">layer_activation</a>()</code> <br><code><a href="layer_activation_elu.html">layer_activation_elu</a>()</code> <br><code><a href="layer_activation_leaky_relu.html">layer_activation_leaky_relu</a>()</code> <br><code><a href="layer_activation_parametric_relu.html">layer_activation_parametric_relu</a>()</code> <br><code><a href="layer_activation_relu.html">layer_activation_relu</a>()</code> <br><code><a href="layer_activation_softmax.html">layer_activation_softmax</a>()</code> <br><code><a href="layer_activity_regularization.html">layer_activity_regularization</a>()</code> <br><code><a href="layer_add.html">layer_add</a>()</code> <br><code><a href="layer_additive_attention.html">layer_additive_attention</a>()</code> <br><code><a href="layer_alpha_dropout.html">layer_alpha_dropout</a>()</code> <br><code><a href="layer_attention.html">layer_attention</a>()</code> <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_average.html">layer_average</a>()</code> <br><code><a href="layer_average_pooling_1d.html">layer_average_pooling_1d</a>()</code> <br><code><a href="layer_average_pooling_2d.html">layer_average_pooling_2d</a>()</code> <br><code><a href="layer_average_pooling_3d.html">layer_average_pooling_3d</a>()</code> <br><code><a href="layer_batch_normalization.html">layer_batch_normalization</a>()</code> <br><code><a href="layer_bidirectional.html">layer_bidirectional</a>()</code> <br><code><a href="layer_category_encoding.html">layer_category_encoding</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_concatenate.html">layer_concatenate</a>()</code> <br><code><a href="layer_conv_1d.html">layer_conv_1d</a>()</code> <br><code><a href="layer_conv_1d_transpose.html">layer_conv_1d_transpose</a>()</code> <br><code><a href="layer_conv_2d.html">layer_conv_2d</a>()</code> <br><code><a href="layer_conv_2d_transpose.html">layer_conv_2d_transpose</a>()</code> <br><code><a href="layer_conv_3d.html">layer_conv_3d</a>()</code> <br><code><a href="layer_conv_3d_transpose.html">layer_conv_3d_transpose</a>()</code> <br><code><a href="layer_conv_lstm_1d.html">layer_conv_lstm_1d</a>()</code> <br><code><a href="layer_conv_lstm_2d.html">layer_conv_lstm_2d</a>()</code> <br><code><a href="layer_conv_lstm_3d.html">layer_conv_lstm_3d</a>()</code> <br><code><a href="layer_cropping_1d.html">layer_cropping_1d</a>()</code> <br><code><a href="layer_cropping_2d.html">layer_cropping_2d</a>()</code> <br><code><a href="layer_cropping_3d.html">layer_cropping_3d</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_dense.html">layer_dense</a>()</code> <br><code><a href="layer_depthwise_conv_1d.html">layer_depthwise_conv_1d</a>()</code> <br><code><a href="layer_depthwise_conv_2d.html">layer_depthwise_conv_2d</a>()</code> <br><code><a href="layer_discretization.html">layer_discretization</a>()</code> <br><code><a href="layer_dot.html">layer_dot</a>()</code> <br><code><a href="layer_dropout.html">layer_dropout</a>()</code> <br><code><a href="layer_einsum_dense.html">layer_einsum_dense</a>()</code> <br><code><a href="layer_embedding.html">layer_embedding</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_feature_space.html">layer_feature_space</a>()</code> <br><code><a href="layer_flatten.html">layer_flatten</a>()</code> <br><code><a href="layer_flax_module_wrapper.html">layer_flax_module_wrapper</a>()</code> <br><code><a href="layer_gaussian_dropout.html">layer_gaussian_dropout</a>()</code> <br><code><a href="layer_gaussian_noise.html">layer_gaussian_noise</a>()</code> <br><code><a href="layer_global_average_pooling_1d.html">layer_global_average_pooling_1d</a>()</code> <br><code><a href="layer_global_average_pooling_2d.html">layer_global_average_pooling_2d</a>()</code> <br><code><a href="layer_global_average_pooling_3d.html">layer_global_average_pooling_3d</a>()</code> <br><code><a href="layer_global_max_pooling_1d.html">layer_global_max_pooling_1d</a>()</code> <br><code><a href="layer_global_max_pooling_2d.html">layer_global_max_pooling_2d</a>()</code> <br><code><a href="layer_global_max_pooling_3d.html">layer_global_max_pooling_3d</a>()</code> <br><code><a href="layer_group_normalization.html">layer_group_normalization</a>()</code> <br><code><a href="layer_group_query_attention.html">layer_group_query_attention</a>()</code> <br><code><a href="layer_gru.html">layer_gru</a>()</code> <br><code><a href="layer_hashed_crossing.html">layer_hashed_crossing</a>()</code> <br><code><a href="layer_hashing.html">layer_hashing</a>()</code> <br><code><a href="layer_identity.html">layer_identity</a>()</code> <br><code><a href="layer_integer_lookup.html">layer_integer_lookup</a>()</code> <br><code><a href="layer_jax_model_wrapper.html">layer_jax_model_wrapper</a>()</code> <br><code><a href="layer_lambda.html">layer_lambda</a>()</code> <br><code><a href="layer_layer_normalization.html">layer_layer_normalization</a>()</code> <br><code><a href="layer_lstm.html">layer_lstm</a>()</code> <br><code><a href="layer_masking.html">layer_masking</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_max_pooling_1d.html">layer_max_pooling_1d</a>()</code> <br><code><a href="layer_max_pooling_2d.html">layer_max_pooling_2d</a>()</code> <br><code><a href="layer_max_pooling_3d.html">layer_max_pooling_3d</a>()</code> <br><code><a href="layer_maximum.html">layer_maximum</a>()</code> <br><code><a href="layer_mel_spectrogram.html">layer_mel_spectrogram</a>()</code> <br><code><a href="layer_minimum.html">layer_minimum</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_multi_head_attention.html">layer_multi_head_attention</a>()</code> <br><code><a href="layer_multiply.html">layer_multiply</a>()</code> <br><code><a href="layer_normalization.html">layer_normalization</a>()</code> <br><code><a href="layer_permute.html">layer_permute</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_brightness.html">layer_random_brightness</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_contrast.html">layer_random_contrast</a>()</code> <br><code><a href="layer_random_crop.html">layer_random_crop</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_flip.html">layer_random_flip</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_rotation.html">layer_random_rotation</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_random_translation.html">layer_random_translation</a>()</code> <br><code><a href="layer_random_zoom.html">layer_random_zoom</a>()</code> <br><code><a href="layer_repeat_vector.html">layer_repeat_vector</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_rms_normalization.html">layer_rms_normalization</a>()</code> <br><code><a href="layer_rnn.html">layer_rnn</a>()</code> <br><code><a href="layer_separable_conv_1d.html">layer_separable_conv_1d</a>()</code> <br><code><a href="layer_separable_conv_2d.html">layer_separable_conv_2d</a>()</code> <br><code><a href="layer_simple_rnn.html">layer_simple_rnn</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br><code><a href="layer_spatial_dropout_1d.html">layer_spatial_dropout_1d</a>()</code> <br><code><a href="layer_spatial_dropout_2d.html">layer_spatial_dropout_2d</a>()</code> <br><code><a href="layer_spatial_dropout_3d.html">layer_spatial_dropout_3d</a>()</code> <br><code><a href="layer_spectral_normalization.html">layer_spectral_normalization</a>()</code> <br><code><a href="layer_stft_spectrogram.html">layer_stft_spectrogram</a>()</code> <br><code><a href="layer_string_lookup.html">layer_string_lookup</a>()</code> <br><code><a href="layer_subtract.html">layer_subtract</a>()</code> <br><code><a href="layer_text_vectorization.html">layer_text_vectorization</a>()</code> <br><code><a href="layer_tfsm.html">layer_tfsm</a>()</code> <br><code><a href="layer_time_distributed.html">layer_time_distributed</a>()</code> <br><code><a href="layer_torch_module_wrapper.html">layer_torch_module_wrapper</a>()</code> <br><code><a href="layer_unit_normalization.html">layer_unit_normalization</a>()</code> <br><code><a href="layer_upsampling_1d.html">layer_upsampling_1d</a>()</code> <br><code><a href="layer_upsampling_2d.html">layer_upsampling_2d</a>()</code> <br><code><a href="layer_upsampling_3d.html">layer_upsampling_3d</a>()</code> <br><code><a href="layer_zero_padding_1d.html">layer_zero_padding_1d</a>()</code> <br><code><a href="layer_zero_padding_2d.html">layer_zero_padding_2d</a>()</code> <br><code><a href="layer_zero_padding_3d.html">layer_zero_padding_3d</a>()</code> <br><code><a href="rnn_cell_gru.html">rnn_cell_gru</a>()</code> <br><code><a href="rnn_cell_lstm.html">rnn_cell_lstm</a>()</code> <br><code><a href="rnn_cell_simple.html">rnn_cell_simple</a>()</code> <br><code><a href="rnn_cells_stack.html">rnn_cells_stack</a>()</code> <br></p></div>
+<p>Other layers: <br><code><a href="Layer.html">Layer</a>()</code> <br><code><a href="layer_activation.html">layer_activation</a>()</code> <br><code><a href="layer_activation_elu.html">layer_activation_elu</a>()</code> <br><code><a href="layer_activation_leaky_relu.html">layer_activation_leaky_relu</a>()</code> <br><code><a href="layer_activation_parametric_relu.html">layer_activation_parametric_relu</a>()</code> <br><code><a href="layer_activation_relu.html">layer_activation_relu</a>()</code> <br><code><a href="layer_activation_softmax.html">layer_activation_softmax</a>()</code> <br><code><a href="layer_activity_regularization.html">layer_activity_regularization</a>()</code> <br><code><a href="layer_add.html">layer_add</a>()</code> <br><code><a href="layer_additive_attention.html">layer_additive_attention</a>()</code> <br><code><a href="layer_alpha_dropout.html">layer_alpha_dropout</a>()</code> <br><code><a href="layer_attention.html">layer_attention</a>()</code> <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_average.html">layer_average</a>()</code> <br><code><a href="layer_average_pooling_1d.html">layer_average_pooling_1d</a>()</code> <br><code><a href="layer_average_pooling_2d.html">layer_average_pooling_2d</a>()</code> <br><code><a href="layer_average_pooling_3d.html">layer_average_pooling_3d</a>()</code> <br><code><a href="layer_batch_normalization.html">layer_batch_normalization</a>()</code> <br><code><a href="layer_bidirectional.html">layer_bidirectional</a>()</code> <br><code><a href="layer_category_encoding.html">layer_category_encoding</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_concatenate.html">layer_concatenate</a>()</code> <br><code><a href="layer_conv_1d.html">layer_conv_1d</a>()</code> <br><code><a href="layer_conv_1d_transpose.html">layer_conv_1d_transpose</a>()</code> <br><code><a href="layer_conv_2d.html">layer_conv_2d</a>()</code> <br><code><a href="layer_conv_2d_transpose.html">layer_conv_2d_transpose</a>()</code> <br><code><a href="layer_conv_3d.html">layer_conv_3d</a>()</code> <br><code><a href="layer_conv_3d_transpose.html">layer_conv_3d_transpose</a>()</code> <br><code><a href="layer_conv_lstm_1d.html">layer_conv_lstm_1d</a>()</code> <br><code><a href="layer_conv_lstm_2d.html">layer_conv_lstm_2d</a>()</code> <br><code><a href="layer_conv_lstm_3d.html">layer_conv_lstm_3d</a>()</code> <br><code><a href="layer_cropping_1d.html">layer_cropping_1d</a>()</code> <br><code><a href="layer_cropping_2d.html">layer_cropping_2d</a>()</code> <br><code><a href="layer_cropping_3d.html">layer_cropping_3d</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_dense.html">layer_dense</a>()</code> <br><code><a href="layer_depthwise_conv_1d.html">layer_depthwise_conv_1d</a>()</code> <br><code><a href="layer_depthwise_conv_2d.html">layer_depthwise_conv_2d</a>()</code> <br><code><a href="layer_discretization.html">layer_discretization</a>()</code> <br><code><a href="layer_dot.html">layer_dot</a>()</code> <br><code><a href="layer_dropout.html">layer_dropout</a>()</code> <br><code><a href="layer_einsum_dense.html">layer_einsum_dense</a>()</code> <br><code><a href="layer_embedding.html">layer_embedding</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_feature_space.html">layer_feature_space</a>()</code> <br><code><a href="layer_flatten.html">layer_flatten</a>()</code> <br><code><a href="layer_flax_module_wrapper.html">layer_flax_module_wrapper</a>()</code> <br><code><a href="layer_gaussian_dropout.html">layer_gaussian_dropout</a>()</code> <br><code><a href="layer_gaussian_noise.html">layer_gaussian_noise</a>()</code> <br><code><a href="layer_global_average_pooling_1d.html">layer_global_average_pooling_1d</a>()</code> <br><code><a href="layer_global_average_pooling_2d.html">layer_global_average_pooling_2d</a>()</code> <br><code><a href="layer_global_average_pooling_3d.html">layer_global_average_pooling_3d</a>()</code> <br><code><a href="layer_global_max_pooling_1d.html">layer_global_max_pooling_1d</a>()</code> <br><code><a href="layer_global_max_pooling_2d.html">layer_global_max_pooling_2d</a>()</code> <br><code><a href="layer_global_max_pooling_3d.html">layer_global_max_pooling_3d</a>()</code> <br><code><a href="layer_group_normalization.html">layer_group_normalization</a>()</code> <br><code><a href="layer_group_query_attention.html">layer_group_query_attention</a>()</code> <br><code><a href="layer_gru.html">layer_gru</a>()</code> <br><code><a href="layer_hashed_crossing.html">layer_hashed_crossing</a>()</code> <br><code><a href="layer_hashing.html">layer_hashing</a>()</code> <br><code><a href="layer_identity.html">layer_identity</a>()</code> <br><code><a href="layer_integer_lookup.html">layer_integer_lookup</a>()</code> <br><code><a href="layer_jax_model_wrapper.html">layer_jax_model_wrapper</a>()</code> <br><code><a href="layer_lambda.html">layer_lambda</a>()</code> <br><code><a href="layer_layer_normalization.html">layer_layer_normalization</a>()</code> <br><code><a href="layer_lstm.html">layer_lstm</a>()</code> <br><code><a href="layer_masking.html">layer_masking</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_max_pooling_1d.html">layer_max_pooling_1d</a>()</code> <br><code><a href="layer_max_pooling_2d.html">layer_max_pooling_2d</a>()</code> <br><code><a href="layer_max_pooling_3d.html">layer_max_pooling_3d</a>()</code> <br><code><a href="layer_maximum.html">layer_maximum</a>()</code> <br><code><a href="layer_mel_spectrogram.html">layer_mel_spectrogram</a>()</code> <br><code><a href="layer_minimum.html">layer_minimum</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_multi_head_attention.html">layer_multi_head_attention</a>()</code> <br><code><a href="layer_multiply.html">layer_multiply</a>()</code> <br><code><a href="layer_normalization.html">layer_normalization</a>()</code> <br><code><a href="layer_permute.html">layer_permute</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_brightness.html">layer_random_brightness</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_contrast.html">layer_random_contrast</a>()</code> <br><code><a href="layer_random_crop.html">layer_random_crop</a>()</code> <br><code><a href="layer_random_elastic_transform.html">layer_random_elastic_transform</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_flip.html">layer_random_flip</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_rotation.html">layer_random_rotation</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_random_translation.html">layer_random_translation</a>()</code> <br><code><a href="layer_random_zoom.html">layer_random_zoom</a>()</code> <br><code><a href="layer_repeat_vector.html">layer_repeat_vector</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_rms_normalization.html">layer_rms_normalization</a>()</code> <br><code><a href="layer_rnn.html">layer_rnn</a>()</code> <br><code><a href="layer_separable_conv_1d.html">layer_separable_conv_1d</a>()</code> <br><code><a href="layer_separable_conv_2d.html">layer_separable_conv_2d</a>()</code> <br><code><a href="layer_simple_rnn.html">layer_simple_rnn</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br><code><a href="layer_spatial_dropout_1d.html">layer_spatial_dropout_1d</a>()</code> <br><code><a href="layer_spatial_dropout_2d.html">layer_spatial_dropout_2d</a>()</code> <br><code><a href="layer_spatial_dropout_3d.html">layer_spatial_dropout_3d</a>()</code> <br><code><a href="layer_spectral_normalization.html">layer_spectral_normalization</a>()</code> <br><code><a href="layer_stft_spectrogram.html">layer_stft_spectrogram</a>()</code> <br><code><a href="layer_string_lookup.html">layer_string_lookup</a>()</code> <br><code><a href="layer_subtract.html">layer_subtract</a>()</code> <br><code><a href="layer_text_vectorization.html">layer_text_vectorization</a>()</code> <br><code><a href="layer_tfsm.html">layer_tfsm</a>()</code> <br><code><a href="layer_time_distributed.html">layer_time_distributed</a>()</code> <br><code><a href="layer_torch_module_wrapper.html">layer_torch_module_wrapper</a>()</code> <br><code><a href="layer_unit_normalization.html">layer_unit_normalization</a>()</code> <br><code><a href="layer_upsampling_1d.html">layer_upsampling_1d</a>()</code> <br><code><a href="layer_upsampling_2d.html">layer_upsampling_2d</a>()</code> <br><code><a href="layer_upsampling_3d.html">layer_upsampling_3d</a>()</code> <br><code><a href="layer_zero_padding_1d.html">layer_zero_padding_1d</a>()</code> <br><code><a href="layer_zero_padding_2d.html">layer_zero_padding_2d</a>()</code> <br><code><a href="layer_zero_padding_3d.html">layer_zero_padding_3d</a>()</code> <br><code><a href="rnn_cell_gru.html">rnn_cell_gru</a>()</code> <br><code><a href="rnn_cell_lstm.html">rnn_cell_lstm</a>()</code> <br><code><a href="rnn_cell_simple.html">rnn_cell_simple</a>()</code> <br><code><a href="rnn_cells_stack.html">rnn_cells_stack</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -129,7 +129,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/layer_resizing.html b/docs/dev/reference/layer_resizing.html
index 7d98026df..215c75c7a 100644
--- a/docs/dev/reference/layer_resizing.html
+++ b/docs/dev/reference/layer_resizing.html
@@ -14,7 +14,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -180,9 +180,9 @@ <h2 id="output-shape">Output Shape<a class="anchor" aria-label="anchor" href="#o
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
     <div class="dont-index">
 <ul><li><p><a href="https://keras.io/api/layers/preprocessing_layers/image_preprocessing/resizing#resizing-class" class="external-link">https://keras.io/api/layers/preprocessing_layers/image_preprocessing/resizing#resizing-class</a></p></li>
-</ul><p>Other image preprocessing layers: <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br></p>
-<p>Other preprocessing layers: <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_category_encoding.html">layer_category_encoding</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_discretization.html">layer_discretization</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_feature_space.html">layer_feature_space</a>()</code> <br><code><a href="layer_hashed_crossing.html">layer_hashed_crossing</a>()</code> <br><code><a href="layer_hashing.html">layer_hashing</a>()</code> <br><code><a href="layer_integer_lookup.html">layer_integer_lookup</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_mel_spectrogram.html">layer_mel_spectrogram</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_normalization.html">layer_normalization</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_brightness.html">layer_random_brightness</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_contrast.html">layer_random_contrast</a>()</code> <br><code><a href="layer_random_crop.html">layer_random_crop</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_flip.html">layer_random_flip</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_rotation.html">layer_random_rotation</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_random_translation.html">layer_random_translation</a>()</code> <br><code><a href="layer_random_zoom.html">layer_random_zoom</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br><code><a href="layer_stft_spectrogram.html">layer_stft_spectrogram</a>()</code> <br><code><a href="layer_string_lookup.html">layer_string_lookup</a>()</code> <br><code><a href="layer_text_vectorization.html">layer_text_vectorization</a>()</code> <br></p>
-<p>Other layers: <br><code><a href="Layer.html">Layer</a>()</code> <br><code><a href="layer_activation.html">layer_activation</a>()</code> <br><code><a href="layer_activation_elu.html">layer_activation_elu</a>()</code> <br><code><a href="layer_activation_leaky_relu.html">layer_activation_leaky_relu</a>()</code> <br><code><a href="layer_activation_parametric_relu.html">layer_activation_parametric_relu</a>()</code> <br><code><a href="layer_activation_relu.html">layer_activation_relu</a>()</code> <br><code><a href="layer_activation_softmax.html">layer_activation_softmax</a>()</code> <br><code><a href="layer_activity_regularization.html">layer_activity_regularization</a>()</code> <br><code><a href="layer_add.html">layer_add</a>()</code> <br><code><a href="layer_additive_attention.html">layer_additive_attention</a>()</code> <br><code><a href="layer_alpha_dropout.html">layer_alpha_dropout</a>()</code> <br><code><a href="layer_attention.html">layer_attention</a>()</code> <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_average.html">layer_average</a>()</code> <br><code><a href="layer_average_pooling_1d.html">layer_average_pooling_1d</a>()</code> <br><code><a href="layer_average_pooling_2d.html">layer_average_pooling_2d</a>()</code> <br><code><a href="layer_average_pooling_3d.html">layer_average_pooling_3d</a>()</code> <br><code><a href="layer_batch_normalization.html">layer_batch_normalization</a>()</code> <br><code><a href="layer_bidirectional.html">layer_bidirectional</a>()</code> <br><code><a href="layer_category_encoding.html">layer_category_encoding</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_concatenate.html">layer_concatenate</a>()</code> <br><code><a href="layer_conv_1d.html">layer_conv_1d</a>()</code> <br><code><a href="layer_conv_1d_transpose.html">layer_conv_1d_transpose</a>()</code> <br><code><a href="layer_conv_2d.html">layer_conv_2d</a>()</code> <br><code><a href="layer_conv_2d_transpose.html">layer_conv_2d_transpose</a>()</code> <br><code><a href="layer_conv_3d.html">layer_conv_3d</a>()</code> <br><code><a href="layer_conv_3d_transpose.html">layer_conv_3d_transpose</a>()</code> <br><code><a href="layer_conv_lstm_1d.html">layer_conv_lstm_1d</a>()</code> <br><code><a href="layer_conv_lstm_2d.html">layer_conv_lstm_2d</a>()</code> <br><code><a href="layer_conv_lstm_3d.html">layer_conv_lstm_3d</a>()</code> <br><code><a href="layer_cropping_1d.html">layer_cropping_1d</a>()</code> <br><code><a href="layer_cropping_2d.html">layer_cropping_2d</a>()</code> <br><code><a href="layer_cropping_3d.html">layer_cropping_3d</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_dense.html">layer_dense</a>()</code> <br><code><a href="layer_depthwise_conv_1d.html">layer_depthwise_conv_1d</a>()</code> <br><code><a href="layer_depthwise_conv_2d.html">layer_depthwise_conv_2d</a>()</code> <br><code><a href="layer_discretization.html">layer_discretization</a>()</code> <br><code><a href="layer_dot.html">layer_dot</a>()</code> <br><code><a href="layer_dropout.html">layer_dropout</a>()</code> <br><code><a href="layer_einsum_dense.html">layer_einsum_dense</a>()</code> <br><code><a href="layer_embedding.html">layer_embedding</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_feature_space.html">layer_feature_space</a>()</code> <br><code><a href="layer_flatten.html">layer_flatten</a>()</code> <br><code><a href="layer_flax_module_wrapper.html">layer_flax_module_wrapper</a>()</code> <br><code><a href="layer_gaussian_dropout.html">layer_gaussian_dropout</a>()</code> <br><code><a href="layer_gaussian_noise.html">layer_gaussian_noise</a>()</code> <br><code><a href="layer_global_average_pooling_1d.html">layer_global_average_pooling_1d</a>()</code> <br><code><a href="layer_global_average_pooling_2d.html">layer_global_average_pooling_2d</a>()</code> <br><code><a href="layer_global_average_pooling_3d.html">layer_global_average_pooling_3d</a>()</code> <br><code><a href="layer_global_max_pooling_1d.html">layer_global_max_pooling_1d</a>()</code> <br><code><a href="layer_global_max_pooling_2d.html">layer_global_max_pooling_2d</a>()</code> <br><code><a href="layer_global_max_pooling_3d.html">layer_global_max_pooling_3d</a>()</code> <br><code><a href="layer_group_normalization.html">layer_group_normalization</a>()</code> <br><code><a href="layer_group_query_attention.html">layer_group_query_attention</a>()</code> <br><code><a href="layer_gru.html">layer_gru</a>()</code> <br><code><a href="layer_hashed_crossing.html">layer_hashed_crossing</a>()</code> <br><code><a href="layer_hashing.html">layer_hashing</a>()</code> <br><code><a href="layer_identity.html">layer_identity</a>()</code> <br><code><a href="layer_integer_lookup.html">layer_integer_lookup</a>()</code> <br><code><a href="layer_jax_model_wrapper.html">layer_jax_model_wrapper</a>()</code> <br><code><a href="layer_lambda.html">layer_lambda</a>()</code> <br><code><a href="layer_layer_normalization.html">layer_layer_normalization</a>()</code> <br><code><a href="layer_lstm.html">layer_lstm</a>()</code> <br><code><a href="layer_masking.html">layer_masking</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_max_pooling_1d.html">layer_max_pooling_1d</a>()</code> <br><code><a href="layer_max_pooling_2d.html">layer_max_pooling_2d</a>()</code> <br><code><a href="layer_max_pooling_3d.html">layer_max_pooling_3d</a>()</code> <br><code><a href="layer_maximum.html">layer_maximum</a>()</code> <br><code><a href="layer_mel_spectrogram.html">layer_mel_spectrogram</a>()</code> <br><code><a href="layer_minimum.html">layer_minimum</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_multi_head_attention.html">layer_multi_head_attention</a>()</code> <br><code><a href="layer_multiply.html">layer_multiply</a>()</code> <br><code><a href="layer_normalization.html">layer_normalization</a>()</code> <br><code><a href="layer_permute.html">layer_permute</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_brightness.html">layer_random_brightness</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_contrast.html">layer_random_contrast</a>()</code> <br><code><a href="layer_random_crop.html">layer_random_crop</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_flip.html">layer_random_flip</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_rotation.html">layer_random_rotation</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_random_translation.html">layer_random_translation</a>()</code> <br><code><a href="layer_random_zoom.html">layer_random_zoom</a>()</code> <br><code><a href="layer_repeat_vector.html">layer_repeat_vector</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_reshape.html">layer_reshape</a>()</code> <br><code><a href="layer_rms_normalization.html">layer_rms_normalization</a>()</code> <br><code><a href="layer_rnn.html">layer_rnn</a>()</code> <br><code><a href="layer_separable_conv_1d.html">layer_separable_conv_1d</a>()</code> <br><code><a href="layer_separable_conv_2d.html">layer_separable_conv_2d</a>()</code> <br><code><a href="layer_simple_rnn.html">layer_simple_rnn</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br><code><a href="layer_spatial_dropout_1d.html">layer_spatial_dropout_1d</a>()</code> <br><code><a href="layer_spatial_dropout_2d.html">layer_spatial_dropout_2d</a>()</code> <br><code><a href="layer_spatial_dropout_3d.html">layer_spatial_dropout_3d</a>()</code> <br><code><a href="layer_spectral_normalization.html">layer_spectral_normalization</a>()</code> <br><code><a href="layer_stft_spectrogram.html">layer_stft_spectrogram</a>()</code> <br><code><a href="layer_string_lookup.html">layer_string_lookup</a>()</code> <br><code><a href="layer_subtract.html">layer_subtract</a>()</code> <br><code><a href="layer_text_vectorization.html">layer_text_vectorization</a>()</code> <br><code><a href="layer_tfsm.html">layer_tfsm</a>()</code> <br><code><a href="layer_time_distributed.html">layer_time_distributed</a>()</code> <br><code><a href="layer_torch_module_wrapper.html">layer_torch_module_wrapper</a>()</code> <br><code><a href="layer_unit_normalization.html">layer_unit_normalization</a>()</code> <br><code><a href="layer_upsampling_1d.html">layer_upsampling_1d</a>()</code> <br><code><a href="layer_upsampling_2d.html">layer_upsampling_2d</a>()</code> <br><code><a href="layer_upsampling_3d.html">layer_upsampling_3d</a>()</code> <br><code><a href="layer_zero_padding_1d.html">layer_zero_padding_1d</a>()</code> <br><code><a href="layer_zero_padding_2d.html">layer_zero_padding_2d</a>()</code> <br><code><a href="layer_zero_padding_3d.html">layer_zero_padding_3d</a>()</code> <br><code><a href="rnn_cell_gru.html">rnn_cell_gru</a>()</code> <br><code><a href="rnn_cell_lstm.html">rnn_cell_lstm</a>()</code> <br><code><a href="rnn_cell_simple.html">rnn_cell_simple</a>()</code> <br><code><a href="rnn_cells_stack.html">rnn_cells_stack</a>()</code> <br></p></div>
+</ul><p>Other image preprocessing layers: <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_elastic_transform.html">layer_random_elastic_transform</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br></p>
+<p>Other preprocessing layers: <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_category_encoding.html">layer_category_encoding</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_discretization.html">layer_discretization</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_feature_space.html">layer_feature_space</a>()</code> <br><code><a href="layer_hashed_crossing.html">layer_hashed_crossing</a>()</code> <br><code><a href="layer_hashing.html">layer_hashing</a>()</code> <br><code><a href="layer_integer_lookup.html">layer_integer_lookup</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_mel_spectrogram.html">layer_mel_spectrogram</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_normalization.html">layer_normalization</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_brightness.html">layer_random_brightness</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_contrast.html">layer_random_contrast</a>()</code> <br><code><a href="layer_random_crop.html">layer_random_crop</a>()</code> <br><code><a href="layer_random_elastic_transform.html">layer_random_elastic_transform</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_flip.html">layer_random_flip</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_rotation.html">layer_random_rotation</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_random_translation.html">layer_random_translation</a>()</code> <br><code><a href="layer_random_zoom.html">layer_random_zoom</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br><code><a href="layer_stft_spectrogram.html">layer_stft_spectrogram</a>()</code> <br><code><a href="layer_string_lookup.html">layer_string_lookup</a>()</code> <br><code><a href="layer_text_vectorization.html">layer_text_vectorization</a>()</code> <br></p>
+<p>Other layers: <br><code><a href="Layer.html">Layer</a>()</code> <br><code><a href="layer_activation.html">layer_activation</a>()</code> <br><code><a href="layer_activation_elu.html">layer_activation_elu</a>()</code> <br><code><a href="layer_activation_leaky_relu.html">layer_activation_leaky_relu</a>()</code> <br><code><a href="layer_activation_parametric_relu.html">layer_activation_parametric_relu</a>()</code> <br><code><a href="layer_activation_relu.html">layer_activation_relu</a>()</code> <br><code><a href="layer_activation_softmax.html">layer_activation_softmax</a>()</code> <br><code><a href="layer_activity_regularization.html">layer_activity_regularization</a>()</code> <br><code><a href="layer_add.html">layer_add</a>()</code> <br><code><a href="layer_additive_attention.html">layer_additive_attention</a>()</code> <br><code><a href="layer_alpha_dropout.html">layer_alpha_dropout</a>()</code> <br><code><a href="layer_attention.html">layer_attention</a>()</code> <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_average.html">layer_average</a>()</code> <br><code><a href="layer_average_pooling_1d.html">layer_average_pooling_1d</a>()</code> <br><code><a href="layer_average_pooling_2d.html">layer_average_pooling_2d</a>()</code> <br><code><a href="layer_average_pooling_3d.html">layer_average_pooling_3d</a>()</code> <br><code><a href="layer_batch_normalization.html">layer_batch_normalization</a>()</code> <br><code><a href="layer_bidirectional.html">layer_bidirectional</a>()</code> <br><code><a href="layer_category_encoding.html">layer_category_encoding</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_concatenate.html">layer_concatenate</a>()</code> <br><code><a href="layer_conv_1d.html">layer_conv_1d</a>()</code> <br><code><a href="layer_conv_1d_transpose.html">layer_conv_1d_transpose</a>()</code> <br><code><a href="layer_conv_2d.html">layer_conv_2d</a>()</code> <br><code><a href="layer_conv_2d_transpose.html">layer_conv_2d_transpose</a>()</code> <br><code><a href="layer_conv_3d.html">layer_conv_3d</a>()</code> <br><code><a href="layer_conv_3d_transpose.html">layer_conv_3d_transpose</a>()</code> <br><code><a href="layer_conv_lstm_1d.html">layer_conv_lstm_1d</a>()</code> <br><code><a href="layer_conv_lstm_2d.html">layer_conv_lstm_2d</a>()</code> <br><code><a href="layer_conv_lstm_3d.html">layer_conv_lstm_3d</a>()</code> <br><code><a href="layer_cropping_1d.html">layer_cropping_1d</a>()</code> <br><code><a href="layer_cropping_2d.html">layer_cropping_2d</a>()</code> <br><code><a href="layer_cropping_3d.html">layer_cropping_3d</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_dense.html">layer_dense</a>()</code> <br><code><a href="layer_depthwise_conv_1d.html">layer_depthwise_conv_1d</a>()</code> <br><code><a href="layer_depthwise_conv_2d.html">layer_depthwise_conv_2d</a>()</code> <br><code><a href="layer_discretization.html">layer_discretization</a>()</code> <br><code><a href="layer_dot.html">layer_dot</a>()</code> <br><code><a href="layer_dropout.html">layer_dropout</a>()</code> <br><code><a href="layer_einsum_dense.html">layer_einsum_dense</a>()</code> <br><code><a href="layer_embedding.html">layer_embedding</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_feature_space.html">layer_feature_space</a>()</code> <br><code><a href="layer_flatten.html">layer_flatten</a>()</code> <br><code><a href="layer_flax_module_wrapper.html">layer_flax_module_wrapper</a>()</code> <br><code><a href="layer_gaussian_dropout.html">layer_gaussian_dropout</a>()</code> <br><code><a href="layer_gaussian_noise.html">layer_gaussian_noise</a>()</code> <br><code><a href="layer_global_average_pooling_1d.html">layer_global_average_pooling_1d</a>()</code> <br><code><a href="layer_global_average_pooling_2d.html">layer_global_average_pooling_2d</a>()</code> <br><code><a href="layer_global_average_pooling_3d.html">layer_global_average_pooling_3d</a>()</code> <br><code><a href="layer_global_max_pooling_1d.html">layer_global_max_pooling_1d</a>()</code> <br><code><a href="layer_global_max_pooling_2d.html">layer_global_max_pooling_2d</a>()</code> <br><code><a href="layer_global_max_pooling_3d.html">layer_global_max_pooling_3d</a>()</code> <br><code><a href="layer_group_normalization.html">layer_group_normalization</a>()</code> <br><code><a href="layer_group_query_attention.html">layer_group_query_attention</a>()</code> <br><code><a href="layer_gru.html">layer_gru</a>()</code> <br><code><a href="layer_hashed_crossing.html">layer_hashed_crossing</a>()</code> <br><code><a href="layer_hashing.html">layer_hashing</a>()</code> <br><code><a href="layer_identity.html">layer_identity</a>()</code> <br><code><a href="layer_integer_lookup.html">layer_integer_lookup</a>()</code> <br><code><a href="layer_jax_model_wrapper.html">layer_jax_model_wrapper</a>()</code> <br><code><a href="layer_lambda.html">layer_lambda</a>()</code> <br><code><a href="layer_layer_normalization.html">layer_layer_normalization</a>()</code> <br><code><a href="layer_lstm.html">layer_lstm</a>()</code> <br><code><a href="layer_masking.html">layer_masking</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_max_pooling_1d.html">layer_max_pooling_1d</a>()</code> <br><code><a href="layer_max_pooling_2d.html">layer_max_pooling_2d</a>()</code> <br><code><a href="layer_max_pooling_3d.html">layer_max_pooling_3d</a>()</code> <br><code><a href="layer_maximum.html">layer_maximum</a>()</code> <br><code><a href="layer_mel_spectrogram.html">layer_mel_spectrogram</a>()</code> <br><code><a href="layer_minimum.html">layer_minimum</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_multi_head_attention.html">layer_multi_head_attention</a>()</code> <br><code><a href="layer_multiply.html">layer_multiply</a>()</code> <br><code><a href="layer_normalization.html">layer_normalization</a>()</code> <br><code><a href="layer_permute.html">layer_permute</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_brightness.html">layer_random_brightness</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_contrast.html">layer_random_contrast</a>()</code> <br><code><a href="layer_random_crop.html">layer_random_crop</a>()</code> <br><code><a href="layer_random_elastic_transform.html">layer_random_elastic_transform</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_flip.html">layer_random_flip</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_rotation.html">layer_random_rotation</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_random_translation.html">layer_random_translation</a>()</code> <br><code><a href="layer_random_zoom.html">layer_random_zoom</a>()</code> <br><code><a href="layer_repeat_vector.html">layer_repeat_vector</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_reshape.html">layer_reshape</a>()</code> <br><code><a href="layer_rms_normalization.html">layer_rms_normalization</a>()</code> <br><code><a href="layer_rnn.html">layer_rnn</a>()</code> <br><code><a href="layer_separable_conv_1d.html">layer_separable_conv_1d</a>()</code> <br><code><a href="layer_separable_conv_2d.html">layer_separable_conv_2d</a>()</code> <br><code><a href="layer_simple_rnn.html">layer_simple_rnn</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br><code><a href="layer_spatial_dropout_1d.html">layer_spatial_dropout_1d</a>()</code> <br><code><a href="layer_spatial_dropout_2d.html">layer_spatial_dropout_2d</a>()</code> <br><code><a href="layer_spatial_dropout_3d.html">layer_spatial_dropout_3d</a>()</code> <br><code><a href="layer_spectral_normalization.html">layer_spectral_normalization</a>()</code> <br><code><a href="layer_stft_spectrogram.html">layer_stft_spectrogram</a>()</code> <br><code><a href="layer_string_lookup.html">layer_string_lookup</a>()</code> <br><code><a href="layer_subtract.html">layer_subtract</a>()</code> <br><code><a href="layer_text_vectorization.html">layer_text_vectorization</a>()</code> <br><code><a href="layer_tfsm.html">layer_tfsm</a>()</code> <br><code><a href="layer_time_distributed.html">layer_time_distributed</a>()</code> <br><code><a href="layer_torch_module_wrapper.html">layer_torch_module_wrapper</a>()</code> <br><code><a href="layer_unit_normalization.html">layer_unit_normalization</a>()</code> <br><code><a href="layer_upsampling_1d.html">layer_upsampling_1d</a>()</code> <br><code><a href="layer_upsampling_2d.html">layer_upsampling_2d</a>()</code> <br><code><a href="layer_upsampling_3d.html">layer_upsampling_3d</a>()</code> <br><code><a href="layer_zero_padding_1d.html">layer_zero_padding_1d</a>()</code> <br><code><a href="layer_zero_padding_2d.html">layer_zero_padding_2d</a>()</code> <br><code><a href="layer_zero_padding_3d.html">layer_zero_padding_3d</a>()</code> <br><code><a href="rnn_cell_gru.html">rnn_cell_gru</a>()</code> <br><code><a href="rnn_cell_lstm.html">rnn_cell_lstm</a>()</code> <br><code><a href="rnn_cell_simple.html">rnn_cell_simple</a>()</code> <br><code><a href="rnn_cells_stack.html">rnn_cells_stack</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -194,7 +194,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/layer_rms_normalization.html b/docs/dev/reference/layer_rms_normalization.html
index 1bce4250e..8938c3e64 100644
--- a/docs/dev/reference/layer_rms_normalization.html
+++ b/docs/dev/reference/layer_rms_normalization.html
@@ -48,7 +48,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -143,7 +143,7 @@ <h2 id="arguments">Arguments<a class="anchor" aria-label="anchor" href="#argumen
     <div class="section level2">
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
     <div class="dont-index"><p>Other normalization layers: <br><code><a href="layer_batch_normalization.html">layer_batch_normalization</a>()</code> <br><code><a href="layer_group_normalization.html">layer_group_normalization</a>()</code> <br><code><a href="layer_layer_normalization.html">layer_layer_normalization</a>()</code> <br><code><a href="layer_spectral_normalization.html">layer_spectral_normalization</a>()</code> <br><code><a href="layer_unit_normalization.html">layer_unit_normalization</a>()</code> <br></p>
-<p>Other layers: <br><code><a href="Layer.html">Layer</a>()</code> <br><code><a href="layer_activation.html">layer_activation</a>()</code> <br><code><a href="layer_activation_elu.html">layer_activation_elu</a>()</code> <br><code><a href="layer_activation_leaky_relu.html">layer_activation_leaky_relu</a>()</code> <br><code><a href="layer_activation_parametric_relu.html">layer_activation_parametric_relu</a>()</code> <br><code><a href="layer_activation_relu.html">layer_activation_relu</a>()</code> <br><code><a href="layer_activation_softmax.html">layer_activation_softmax</a>()</code> <br><code><a href="layer_activity_regularization.html">layer_activity_regularization</a>()</code> <br><code><a href="layer_add.html">layer_add</a>()</code> <br><code><a href="layer_additive_attention.html">layer_additive_attention</a>()</code> <br><code><a href="layer_alpha_dropout.html">layer_alpha_dropout</a>()</code> <br><code><a href="layer_attention.html">layer_attention</a>()</code> <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_average.html">layer_average</a>()</code> <br><code><a href="layer_average_pooling_1d.html">layer_average_pooling_1d</a>()</code> <br><code><a href="layer_average_pooling_2d.html">layer_average_pooling_2d</a>()</code> <br><code><a href="layer_average_pooling_3d.html">layer_average_pooling_3d</a>()</code> <br><code><a href="layer_batch_normalization.html">layer_batch_normalization</a>()</code> <br><code><a href="layer_bidirectional.html">layer_bidirectional</a>()</code> <br><code><a href="layer_category_encoding.html">layer_category_encoding</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_concatenate.html">layer_concatenate</a>()</code> <br><code><a href="layer_conv_1d.html">layer_conv_1d</a>()</code> <br><code><a href="layer_conv_1d_transpose.html">layer_conv_1d_transpose</a>()</code> <br><code><a href="layer_conv_2d.html">layer_conv_2d</a>()</code> <br><code><a href="layer_conv_2d_transpose.html">layer_conv_2d_transpose</a>()</code> <br><code><a href="layer_conv_3d.html">layer_conv_3d</a>()</code> <br><code><a href="layer_conv_3d_transpose.html">layer_conv_3d_transpose</a>()</code> <br><code><a href="layer_conv_lstm_1d.html">layer_conv_lstm_1d</a>()</code> <br><code><a href="layer_conv_lstm_2d.html">layer_conv_lstm_2d</a>()</code> <br><code><a href="layer_conv_lstm_3d.html">layer_conv_lstm_3d</a>()</code> <br><code><a href="layer_cropping_1d.html">layer_cropping_1d</a>()</code> <br><code><a href="layer_cropping_2d.html">layer_cropping_2d</a>()</code> <br><code><a href="layer_cropping_3d.html">layer_cropping_3d</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_dense.html">layer_dense</a>()</code> <br><code><a href="layer_depthwise_conv_1d.html">layer_depthwise_conv_1d</a>()</code> <br><code><a href="layer_depthwise_conv_2d.html">layer_depthwise_conv_2d</a>()</code> <br><code><a href="layer_discretization.html">layer_discretization</a>()</code> <br><code><a href="layer_dot.html">layer_dot</a>()</code> <br><code><a href="layer_dropout.html">layer_dropout</a>()</code> <br><code><a href="layer_einsum_dense.html">layer_einsum_dense</a>()</code> <br><code><a href="layer_embedding.html">layer_embedding</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_feature_space.html">layer_feature_space</a>()</code> <br><code><a href="layer_flatten.html">layer_flatten</a>()</code> <br><code><a href="layer_flax_module_wrapper.html">layer_flax_module_wrapper</a>()</code> <br><code><a href="layer_gaussian_dropout.html">layer_gaussian_dropout</a>()</code> <br><code><a href="layer_gaussian_noise.html">layer_gaussian_noise</a>()</code> <br><code><a href="layer_global_average_pooling_1d.html">layer_global_average_pooling_1d</a>()</code> <br><code><a href="layer_global_average_pooling_2d.html">layer_global_average_pooling_2d</a>()</code> <br><code><a href="layer_global_average_pooling_3d.html">layer_global_average_pooling_3d</a>()</code> <br><code><a href="layer_global_max_pooling_1d.html">layer_global_max_pooling_1d</a>()</code> <br><code><a href="layer_global_max_pooling_2d.html">layer_global_max_pooling_2d</a>()</code> <br><code><a href="layer_global_max_pooling_3d.html">layer_global_max_pooling_3d</a>()</code> <br><code><a href="layer_group_normalization.html">layer_group_normalization</a>()</code> <br><code><a href="layer_group_query_attention.html">layer_group_query_attention</a>()</code> <br><code><a href="layer_gru.html">layer_gru</a>()</code> <br><code><a href="layer_hashed_crossing.html">layer_hashed_crossing</a>()</code> <br><code><a href="layer_hashing.html">layer_hashing</a>()</code> <br><code><a href="layer_identity.html">layer_identity</a>()</code> <br><code><a href="layer_integer_lookup.html">layer_integer_lookup</a>()</code> <br><code><a href="layer_jax_model_wrapper.html">layer_jax_model_wrapper</a>()</code> <br><code><a href="layer_lambda.html">layer_lambda</a>()</code> <br><code><a href="layer_layer_normalization.html">layer_layer_normalization</a>()</code> <br><code><a href="layer_lstm.html">layer_lstm</a>()</code> <br><code><a href="layer_masking.html">layer_masking</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_max_pooling_1d.html">layer_max_pooling_1d</a>()</code> <br><code><a href="layer_max_pooling_2d.html">layer_max_pooling_2d</a>()</code> <br><code><a href="layer_max_pooling_3d.html">layer_max_pooling_3d</a>()</code> <br><code><a href="layer_maximum.html">layer_maximum</a>()</code> <br><code><a href="layer_mel_spectrogram.html">layer_mel_spectrogram</a>()</code> <br><code><a href="layer_minimum.html">layer_minimum</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_multi_head_attention.html">layer_multi_head_attention</a>()</code> <br><code><a href="layer_multiply.html">layer_multiply</a>()</code> <br><code><a href="layer_normalization.html">layer_normalization</a>()</code> <br><code><a href="layer_permute.html">layer_permute</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_brightness.html">layer_random_brightness</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_contrast.html">layer_random_contrast</a>()</code> <br><code><a href="layer_random_crop.html">layer_random_crop</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_flip.html">layer_random_flip</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_rotation.html">layer_random_rotation</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_random_translation.html">layer_random_translation</a>()</code> <br><code><a href="layer_random_zoom.html">layer_random_zoom</a>()</code> <br><code><a href="layer_repeat_vector.html">layer_repeat_vector</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_reshape.html">layer_reshape</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_rnn.html">layer_rnn</a>()</code> <br><code><a href="layer_separable_conv_1d.html">layer_separable_conv_1d</a>()</code> <br><code><a href="layer_separable_conv_2d.html">layer_separable_conv_2d</a>()</code> <br><code><a href="layer_simple_rnn.html">layer_simple_rnn</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br><code><a href="layer_spatial_dropout_1d.html">layer_spatial_dropout_1d</a>()</code> <br><code><a href="layer_spatial_dropout_2d.html">layer_spatial_dropout_2d</a>()</code> <br><code><a href="layer_spatial_dropout_3d.html">layer_spatial_dropout_3d</a>()</code> <br><code><a href="layer_spectral_normalization.html">layer_spectral_normalization</a>()</code> <br><code><a href="layer_stft_spectrogram.html">layer_stft_spectrogram</a>()</code> <br><code><a href="layer_string_lookup.html">layer_string_lookup</a>()</code> <br><code><a href="layer_subtract.html">layer_subtract</a>()</code> <br><code><a href="layer_text_vectorization.html">layer_text_vectorization</a>()</code> <br><code><a href="layer_tfsm.html">layer_tfsm</a>()</code> <br><code><a href="layer_time_distributed.html">layer_time_distributed</a>()</code> <br><code><a href="layer_torch_module_wrapper.html">layer_torch_module_wrapper</a>()</code> <br><code><a href="layer_unit_normalization.html">layer_unit_normalization</a>()</code> <br><code><a href="layer_upsampling_1d.html">layer_upsampling_1d</a>()</code> <br><code><a href="layer_upsampling_2d.html">layer_upsampling_2d</a>()</code> <br><code><a href="layer_upsampling_3d.html">layer_upsampling_3d</a>()</code> <br><code><a href="layer_zero_padding_1d.html">layer_zero_padding_1d</a>()</code> <br><code><a href="layer_zero_padding_2d.html">layer_zero_padding_2d</a>()</code> <br><code><a href="layer_zero_padding_3d.html">layer_zero_padding_3d</a>()</code> <br><code><a href="rnn_cell_gru.html">rnn_cell_gru</a>()</code> <br><code><a href="rnn_cell_lstm.html">rnn_cell_lstm</a>()</code> <br><code><a href="rnn_cell_simple.html">rnn_cell_simple</a>()</code> <br><code><a href="rnn_cells_stack.html">rnn_cells_stack</a>()</code> <br></p></div>
+<p>Other layers: <br><code><a href="Layer.html">Layer</a>()</code> <br><code><a href="layer_activation.html">layer_activation</a>()</code> <br><code><a href="layer_activation_elu.html">layer_activation_elu</a>()</code> <br><code><a href="layer_activation_leaky_relu.html">layer_activation_leaky_relu</a>()</code> <br><code><a href="layer_activation_parametric_relu.html">layer_activation_parametric_relu</a>()</code> <br><code><a href="layer_activation_relu.html">layer_activation_relu</a>()</code> <br><code><a href="layer_activation_softmax.html">layer_activation_softmax</a>()</code> <br><code><a href="layer_activity_regularization.html">layer_activity_regularization</a>()</code> <br><code><a href="layer_add.html">layer_add</a>()</code> <br><code><a href="layer_additive_attention.html">layer_additive_attention</a>()</code> <br><code><a href="layer_alpha_dropout.html">layer_alpha_dropout</a>()</code> <br><code><a href="layer_attention.html">layer_attention</a>()</code> <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_average.html">layer_average</a>()</code> <br><code><a href="layer_average_pooling_1d.html">layer_average_pooling_1d</a>()</code> <br><code><a href="layer_average_pooling_2d.html">layer_average_pooling_2d</a>()</code> <br><code><a href="layer_average_pooling_3d.html">layer_average_pooling_3d</a>()</code> <br><code><a href="layer_batch_normalization.html">layer_batch_normalization</a>()</code> <br><code><a href="layer_bidirectional.html">layer_bidirectional</a>()</code> <br><code><a href="layer_category_encoding.html">layer_category_encoding</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_concatenate.html">layer_concatenate</a>()</code> <br><code><a href="layer_conv_1d.html">layer_conv_1d</a>()</code> <br><code><a href="layer_conv_1d_transpose.html">layer_conv_1d_transpose</a>()</code> <br><code><a href="layer_conv_2d.html">layer_conv_2d</a>()</code> <br><code><a href="layer_conv_2d_transpose.html">layer_conv_2d_transpose</a>()</code> <br><code><a href="layer_conv_3d.html">layer_conv_3d</a>()</code> <br><code><a href="layer_conv_3d_transpose.html">layer_conv_3d_transpose</a>()</code> <br><code><a href="layer_conv_lstm_1d.html">layer_conv_lstm_1d</a>()</code> <br><code><a href="layer_conv_lstm_2d.html">layer_conv_lstm_2d</a>()</code> <br><code><a href="layer_conv_lstm_3d.html">layer_conv_lstm_3d</a>()</code> <br><code><a href="layer_cropping_1d.html">layer_cropping_1d</a>()</code> <br><code><a href="layer_cropping_2d.html">layer_cropping_2d</a>()</code> <br><code><a href="layer_cropping_3d.html">layer_cropping_3d</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_dense.html">layer_dense</a>()</code> <br><code><a href="layer_depthwise_conv_1d.html">layer_depthwise_conv_1d</a>()</code> <br><code><a href="layer_depthwise_conv_2d.html">layer_depthwise_conv_2d</a>()</code> <br><code><a href="layer_discretization.html">layer_discretization</a>()</code> <br><code><a href="layer_dot.html">layer_dot</a>()</code> <br><code><a href="layer_dropout.html">layer_dropout</a>()</code> <br><code><a href="layer_einsum_dense.html">layer_einsum_dense</a>()</code> <br><code><a href="layer_embedding.html">layer_embedding</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_feature_space.html">layer_feature_space</a>()</code> <br><code><a href="layer_flatten.html">layer_flatten</a>()</code> <br><code><a href="layer_flax_module_wrapper.html">layer_flax_module_wrapper</a>()</code> <br><code><a href="layer_gaussian_dropout.html">layer_gaussian_dropout</a>()</code> <br><code><a href="layer_gaussian_noise.html">layer_gaussian_noise</a>()</code> <br><code><a href="layer_global_average_pooling_1d.html">layer_global_average_pooling_1d</a>()</code> <br><code><a href="layer_global_average_pooling_2d.html">layer_global_average_pooling_2d</a>()</code> <br><code><a href="layer_global_average_pooling_3d.html">layer_global_average_pooling_3d</a>()</code> <br><code><a href="layer_global_max_pooling_1d.html">layer_global_max_pooling_1d</a>()</code> <br><code><a href="layer_global_max_pooling_2d.html">layer_global_max_pooling_2d</a>()</code> <br><code><a href="layer_global_max_pooling_3d.html">layer_global_max_pooling_3d</a>()</code> <br><code><a href="layer_group_normalization.html">layer_group_normalization</a>()</code> <br><code><a href="layer_group_query_attention.html">layer_group_query_attention</a>()</code> <br><code><a href="layer_gru.html">layer_gru</a>()</code> <br><code><a href="layer_hashed_crossing.html">layer_hashed_crossing</a>()</code> <br><code><a href="layer_hashing.html">layer_hashing</a>()</code> <br><code><a href="layer_identity.html">layer_identity</a>()</code> <br><code><a href="layer_integer_lookup.html">layer_integer_lookup</a>()</code> <br><code><a href="layer_jax_model_wrapper.html">layer_jax_model_wrapper</a>()</code> <br><code><a href="layer_lambda.html">layer_lambda</a>()</code> <br><code><a href="layer_layer_normalization.html">layer_layer_normalization</a>()</code> <br><code><a href="layer_lstm.html">layer_lstm</a>()</code> <br><code><a href="layer_masking.html">layer_masking</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_max_pooling_1d.html">layer_max_pooling_1d</a>()</code> <br><code><a href="layer_max_pooling_2d.html">layer_max_pooling_2d</a>()</code> <br><code><a href="layer_max_pooling_3d.html">layer_max_pooling_3d</a>()</code> <br><code><a href="layer_maximum.html">layer_maximum</a>()</code> <br><code><a href="layer_mel_spectrogram.html">layer_mel_spectrogram</a>()</code> <br><code><a href="layer_minimum.html">layer_minimum</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_multi_head_attention.html">layer_multi_head_attention</a>()</code> <br><code><a href="layer_multiply.html">layer_multiply</a>()</code> <br><code><a href="layer_normalization.html">layer_normalization</a>()</code> <br><code><a href="layer_permute.html">layer_permute</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_brightness.html">layer_random_brightness</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_contrast.html">layer_random_contrast</a>()</code> <br><code><a href="layer_random_crop.html">layer_random_crop</a>()</code> <br><code><a href="layer_random_elastic_transform.html">layer_random_elastic_transform</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_flip.html">layer_random_flip</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_rotation.html">layer_random_rotation</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_random_translation.html">layer_random_translation</a>()</code> <br><code><a href="layer_random_zoom.html">layer_random_zoom</a>()</code> <br><code><a href="layer_repeat_vector.html">layer_repeat_vector</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_reshape.html">layer_reshape</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_rnn.html">layer_rnn</a>()</code> <br><code><a href="layer_separable_conv_1d.html">layer_separable_conv_1d</a>()</code> <br><code><a href="layer_separable_conv_2d.html">layer_separable_conv_2d</a>()</code> <br><code><a href="layer_simple_rnn.html">layer_simple_rnn</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br><code><a href="layer_spatial_dropout_1d.html">layer_spatial_dropout_1d</a>()</code> <br><code><a href="layer_spatial_dropout_2d.html">layer_spatial_dropout_2d</a>()</code> <br><code><a href="layer_spatial_dropout_3d.html">layer_spatial_dropout_3d</a>()</code> <br><code><a href="layer_spectral_normalization.html">layer_spectral_normalization</a>()</code> <br><code><a href="layer_stft_spectrogram.html">layer_stft_spectrogram</a>()</code> <br><code><a href="layer_string_lookup.html">layer_string_lookup</a>()</code> <br><code><a href="layer_subtract.html">layer_subtract</a>()</code> <br><code><a href="layer_text_vectorization.html">layer_text_vectorization</a>()</code> <br><code><a href="layer_tfsm.html">layer_tfsm</a>()</code> <br><code><a href="layer_time_distributed.html">layer_time_distributed</a>()</code> <br><code><a href="layer_torch_module_wrapper.html">layer_torch_module_wrapper</a>()</code> <br><code><a href="layer_unit_normalization.html">layer_unit_normalization</a>()</code> <br><code><a href="layer_upsampling_1d.html">layer_upsampling_1d</a>()</code> <br><code><a href="layer_upsampling_2d.html">layer_upsampling_2d</a>()</code> <br><code><a href="layer_upsampling_3d.html">layer_upsampling_3d</a>()</code> <br><code><a href="layer_zero_padding_1d.html">layer_zero_padding_1d</a>()</code> <br><code><a href="layer_zero_padding_2d.html">layer_zero_padding_2d</a>()</code> <br><code><a href="layer_zero_padding_3d.html">layer_zero_padding_3d</a>()</code> <br><code><a href="rnn_cell_gru.html">rnn_cell_gru</a>()</code> <br><code><a href="rnn_cell_lstm.html">rnn_cell_lstm</a>()</code> <br><code><a href="rnn_cell_simple.html">rnn_cell_simple</a>()</code> <br><code><a href="rnn_cells_stack.html">rnn_cells_stack</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -155,7 +155,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/layer_rnn.html b/docs/dev/reference/layer_rnn.html
index c9584e479..0b41dbda0 100644
--- a/docs/dev/reference/layer_rnn.html
+++ b/docs/dev/reference/layer_rnn.html
@@ -8,7 +8,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -243,7 +243,6 @@ <h2 id="examples">Examples<a class="anchor" aria-label="anchor" href="#examples"
 <span>      initializer <span class="op">=</span> <span class="st">'uniform'</span>,</span>
 <span>      name <span class="op">=</span> <span class="st">'recurrent_kernel'</span></span>
 <span>    <span class="op">)</span></span>
-<span>    <span class="va">self</span><span class="op">$</span><span class="va">built</span> <span class="op">&lt;-</span> <span class="cn">TRUE</span></span>
 <span>  <span class="op">}</span>,</span>
 <span></span>
 <span>  call <span class="op">=</span> <span class="kw">function</span><span class="op">(</span><span class="va">inputs</span>, <span class="va">states</span><span class="op">)</span> <span class="op">{</span></span>
@@ -269,7 +268,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 <ul><li><p><a href="https://keras.io/api/layers/recurrent_layers/rnn#rnn-class" class="external-link">https://keras.io/api/layers/recurrent_layers/rnn#rnn-class</a></p></li>
 </ul><p>Other rnn cells: <br><code><a href="rnn_cell_gru.html">rnn_cell_gru</a>()</code> <br><code><a href="rnn_cell_lstm.html">rnn_cell_lstm</a>()</code> <br><code><a href="rnn_cell_simple.html">rnn_cell_simple</a>()</code> <br></p>
 <p>Other rnn layers: <br><code><a href="layer_bidirectional.html">layer_bidirectional</a>()</code> <br><code><a href="layer_conv_lstm_1d.html">layer_conv_lstm_1d</a>()</code> <br><code><a href="layer_conv_lstm_2d.html">layer_conv_lstm_2d</a>()</code> <br><code><a href="layer_conv_lstm_3d.html">layer_conv_lstm_3d</a>()</code> <br><code><a href="layer_gru.html">layer_gru</a>()</code> <br><code><a href="layer_lstm.html">layer_lstm</a>()</code> <br><code><a href="layer_simple_rnn.html">layer_simple_rnn</a>()</code> <br><code><a href="layer_time_distributed.html">layer_time_distributed</a>()</code> <br><code><a href="rnn_cell_gru.html">rnn_cell_gru</a>()</code> <br><code><a href="rnn_cell_lstm.html">rnn_cell_lstm</a>()</code> <br><code><a href="rnn_cell_simple.html">rnn_cell_simple</a>()</code> <br><code><a href="rnn_cells_stack.html">rnn_cells_stack</a>()</code> <br></p>
-<p>Other layers: <br><code><a href="Layer.html">Layer</a>()</code> <br><code><a href="layer_activation.html">layer_activation</a>()</code> <br><code><a href="layer_activation_elu.html">layer_activation_elu</a>()</code> <br><code><a href="layer_activation_leaky_relu.html">layer_activation_leaky_relu</a>()</code> <br><code><a href="layer_activation_parametric_relu.html">layer_activation_parametric_relu</a>()</code> <br><code><a href="layer_activation_relu.html">layer_activation_relu</a>()</code> <br><code><a href="layer_activation_softmax.html">layer_activation_softmax</a>()</code> <br><code><a href="layer_activity_regularization.html">layer_activity_regularization</a>()</code> <br><code><a href="layer_add.html">layer_add</a>()</code> <br><code><a href="layer_additive_attention.html">layer_additive_attention</a>()</code> <br><code><a href="layer_alpha_dropout.html">layer_alpha_dropout</a>()</code> <br><code><a href="layer_attention.html">layer_attention</a>()</code> <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_average.html">layer_average</a>()</code> <br><code><a href="layer_average_pooling_1d.html">layer_average_pooling_1d</a>()</code> <br><code><a href="layer_average_pooling_2d.html">layer_average_pooling_2d</a>()</code> <br><code><a href="layer_average_pooling_3d.html">layer_average_pooling_3d</a>()</code> <br><code><a href="layer_batch_normalization.html">layer_batch_normalization</a>()</code> <br><code><a href="layer_bidirectional.html">layer_bidirectional</a>()</code> <br><code><a href="layer_category_encoding.html">layer_category_encoding</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_concatenate.html">layer_concatenate</a>()</code> <br><code><a href="layer_conv_1d.html">layer_conv_1d</a>()</code> <br><code><a href="layer_conv_1d_transpose.html">layer_conv_1d_transpose</a>()</code> <br><code><a href="layer_conv_2d.html">layer_conv_2d</a>()</code> <br><code><a href="layer_conv_2d_transpose.html">layer_conv_2d_transpose</a>()</code> <br><code><a href="layer_conv_3d.html">layer_conv_3d</a>()</code> <br><code><a href="layer_conv_3d_transpose.html">layer_conv_3d_transpose</a>()</code> <br><code><a href="layer_conv_lstm_1d.html">layer_conv_lstm_1d</a>()</code> <br><code><a href="layer_conv_lstm_2d.html">layer_conv_lstm_2d</a>()</code> <br><code><a href="layer_conv_lstm_3d.html">layer_conv_lstm_3d</a>()</code> <br><code><a href="layer_cropping_1d.html">layer_cropping_1d</a>()</code> <br><code><a href="layer_cropping_2d.html">layer_cropping_2d</a>()</code> <br><code><a href="layer_cropping_3d.html">layer_cropping_3d</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_dense.html">layer_dense</a>()</code> <br><code><a href="layer_depthwise_conv_1d.html">layer_depthwise_conv_1d</a>()</code> <br><code><a href="layer_depthwise_conv_2d.html">layer_depthwise_conv_2d</a>()</code> <br><code><a href="layer_discretization.html">layer_discretization</a>()</code> <br><code><a href="layer_dot.html">layer_dot</a>()</code> <br><code><a href="layer_dropout.html">layer_dropout</a>()</code> <br><code><a href="layer_einsum_dense.html">layer_einsum_dense</a>()</code> <br><code><a href="layer_embedding.html">layer_embedding</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_feature_space.html">layer_feature_space</a>()</code> <br><code><a href="layer_flatten.html">layer_flatten</a>()</code> <br><code><a href="layer_flax_module_wrapper.html">layer_flax_module_wrapper</a>()</code> <br><code><a href="layer_gaussian_dropout.html">layer_gaussian_dropout</a>()</code> <br><code><a href="layer_gaussian_noise.html">layer_gaussian_noise</a>()</code> <br><code><a href="layer_global_average_pooling_1d.html">layer_global_average_pooling_1d</a>()</code> <br><code><a href="layer_global_average_pooling_2d.html">layer_global_average_pooling_2d</a>()</code> <br><code><a href="layer_global_average_pooling_3d.html">layer_global_average_pooling_3d</a>()</code> <br><code><a href="layer_global_max_pooling_1d.html">layer_global_max_pooling_1d</a>()</code> <br><code><a href="layer_global_max_pooling_2d.html">layer_global_max_pooling_2d</a>()</code> <br><code><a href="layer_global_max_pooling_3d.html">layer_global_max_pooling_3d</a>()</code> <br><code><a href="layer_group_normalization.html">layer_group_normalization</a>()</code> <br><code><a href="layer_group_query_attention.html">layer_group_query_attention</a>()</code> <br><code><a href="layer_gru.html">layer_gru</a>()</code> <br><code><a href="layer_hashed_crossing.html">layer_hashed_crossing</a>()</code> <br><code><a href="layer_hashing.html">layer_hashing</a>()</code> <br><code><a href="layer_identity.html">layer_identity</a>()</code> <br><code><a href="layer_integer_lookup.html">layer_integer_lookup</a>()</code> <br><code><a href="layer_jax_model_wrapper.html">layer_jax_model_wrapper</a>()</code> <br><code><a href="layer_lambda.html">layer_lambda</a>()</code> <br><code><a href="layer_layer_normalization.html">layer_layer_normalization</a>()</code> <br><code><a href="layer_lstm.html">layer_lstm</a>()</code> <br><code><a href="layer_masking.html">layer_masking</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_max_pooling_1d.html">layer_max_pooling_1d</a>()</code> <br><code><a href="layer_max_pooling_2d.html">layer_max_pooling_2d</a>()</code> <br><code><a href="layer_max_pooling_3d.html">layer_max_pooling_3d</a>()</code> <br><code><a href="layer_maximum.html">layer_maximum</a>()</code> <br><code><a href="layer_mel_spectrogram.html">layer_mel_spectrogram</a>()</code> <br><code><a href="layer_minimum.html">layer_minimum</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_multi_head_attention.html">layer_multi_head_attention</a>()</code> <br><code><a href="layer_multiply.html">layer_multiply</a>()</code> <br><code><a href="layer_normalization.html">layer_normalization</a>()</code> <br><code><a href="layer_permute.html">layer_permute</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_brightness.html">layer_random_brightness</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_contrast.html">layer_random_contrast</a>()</code> <br><code><a href="layer_random_crop.html">layer_random_crop</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_flip.html">layer_random_flip</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_rotation.html">layer_random_rotation</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_random_translation.html">layer_random_translation</a>()</code> <br><code><a href="layer_random_zoom.html">layer_random_zoom</a>()</code> <br><code><a href="layer_repeat_vector.html">layer_repeat_vector</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_reshape.html">layer_reshape</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_rms_normalization.html">layer_rms_normalization</a>()</code> <br><code><a href="layer_separable_conv_1d.html">layer_separable_conv_1d</a>()</code> <br><code><a href="layer_separable_conv_2d.html">layer_separable_conv_2d</a>()</code> <br><code><a href="layer_simple_rnn.html">layer_simple_rnn</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br><code><a href="layer_spatial_dropout_1d.html">layer_spatial_dropout_1d</a>()</code> <br><code><a href="layer_spatial_dropout_2d.html">layer_spatial_dropout_2d</a>()</code> <br><code><a href="layer_spatial_dropout_3d.html">layer_spatial_dropout_3d</a>()</code> <br><code><a href="layer_spectral_normalization.html">layer_spectral_normalization</a>()</code> <br><code><a href="layer_stft_spectrogram.html">layer_stft_spectrogram</a>()</code> <br><code><a href="layer_string_lookup.html">layer_string_lookup</a>()</code> <br><code><a href="layer_subtract.html">layer_subtract</a>()</code> <br><code><a href="layer_text_vectorization.html">layer_text_vectorization</a>()</code> <br><code><a href="layer_tfsm.html">layer_tfsm</a>()</code> <br><code><a href="layer_time_distributed.html">layer_time_distributed</a>()</code> <br><code><a href="layer_torch_module_wrapper.html">layer_torch_module_wrapper</a>()</code> <br><code><a href="layer_unit_normalization.html">layer_unit_normalization</a>()</code> <br><code><a href="layer_upsampling_1d.html">layer_upsampling_1d</a>()</code> <br><code><a href="layer_upsampling_2d.html">layer_upsampling_2d</a>()</code> <br><code><a href="layer_upsampling_3d.html">layer_upsampling_3d</a>()</code> <br><code><a href="layer_zero_padding_1d.html">layer_zero_padding_1d</a>()</code> <br><code><a href="layer_zero_padding_2d.html">layer_zero_padding_2d</a>()</code> <br><code><a href="layer_zero_padding_3d.html">layer_zero_padding_3d</a>()</code> <br><code><a href="rnn_cell_gru.html">rnn_cell_gru</a>()</code> <br><code><a href="rnn_cell_lstm.html">rnn_cell_lstm</a>()</code> <br><code><a href="rnn_cell_simple.html">rnn_cell_simple</a>()</code> <br><code><a href="rnn_cells_stack.html">rnn_cells_stack</a>()</code> <br></p></div>
+<p>Other layers: <br><code><a href="Layer.html">Layer</a>()</code> <br><code><a href="layer_activation.html">layer_activation</a>()</code> <br><code><a href="layer_activation_elu.html">layer_activation_elu</a>()</code> <br><code><a href="layer_activation_leaky_relu.html">layer_activation_leaky_relu</a>()</code> <br><code><a href="layer_activation_parametric_relu.html">layer_activation_parametric_relu</a>()</code> <br><code><a href="layer_activation_relu.html">layer_activation_relu</a>()</code> <br><code><a href="layer_activation_softmax.html">layer_activation_softmax</a>()</code> <br><code><a href="layer_activity_regularization.html">layer_activity_regularization</a>()</code> <br><code><a href="layer_add.html">layer_add</a>()</code> <br><code><a href="layer_additive_attention.html">layer_additive_attention</a>()</code> <br><code><a href="layer_alpha_dropout.html">layer_alpha_dropout</a>()</code> <br><code><a href="layer_attention.html">layer_attention</a>()</code> <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_average.html">layer_average</a>()</code> <br><code><a href="layer_average_pooling_1d.html">layer_average_pooling_1d</a>()</code> <br><code><a href="layer_average_pooling_2d.html">layer_average_pooling_2d</a>()</code> <br><code><a href="layer_average_pooling_3d.html">layer_average_pooling_3d</a>()</code> <br><code><a href="layer_batch_normalization.html">layer_batch_normalization</a>()</code> <br><code><a href="layer_bidirectional.html">layer_bidirectional</a>()</code> <br><code><a href="layer_category_encoding.html">layer_category_encoding</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_concatenate.html">layer_concatenate</a>()</code> <br><code><a href="layer_conv_1d.html">layer_conv_1d</a>()</code> <br><code><a href="layer_conv_1d_transpose.html">layer_conv_1d_transpose</a>()</code> <br><code><a href="layer_conv_2d.html">layer_conv_2d</a>()</code> <br><code><a href="layer_conv_2d_transpose.html">layer_conv_2d_transpose</a>()</code> <br><code><a href="layer_conv_3d.html">layer_conv_3d</a>()</code> <br><code><a href="layer_conv_3d_transpose.html">layer_conv_3d_transpose</a>()</code> <br><code><a href="layer_conv_lstm_1d.html">layer_conv_lstm_1d</a>()</code> <br><code><a href="layer_conv_lstm_2d.html">layer_conv_lstm_2d</a>()</code> <br><code><a href="layer_conv_lstm_3d.html">layer_conv_lstm_3d</a>()</code> <br><code><a href="layer_cropping_1d.html">layer_cropping_1d</a>()</code> <br><code><a href="layer_cropping_2d.html">layer_cropping_2d</a>()</code> <br><code><a href="layer_cropping_3d.html">layer_cropping_3d</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_dense.html">layer_dense</a>()</code> <br><code><a href="layer_depthwise_conv_1d.html">layer_depthwise_conv_1d</a>()</code> <br><code><a href="layer_depthwise_conv_2d.html">layer_depthwise_conv_2d</a>()</code> <br><code><a href="layer_discretization.html">layer_discretization</a>()</code> <br><code><a href="layer_dot.html">layer_dot</a>()</code> <br><code><a href="layer_dropout.html">layer_dropout</a>()</code> <br><code><a href="layer_einsum_dense.html">layer_einsum_dense</a>()</code> <br><code><a href="layer_embedding.html">layer_embedding</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_feature_space.html">layer_feature_space</a>()</code> <br><code><a href="layer_flatten.html">layer_flatten</a>()</code> <br><code><a href="layer_flax_module_wrapper.html">layer_flax_module_wrapper</a>()</code> <br><code><a href="layer_gaussian_dropout.html">layer_gaussian_dropout</a>()</code> <br><code><a href="layer_gaussian_noise.html">layer_gaussian_noise</a>()</code> <br><code><a href="layer_global_average_pooling_1d.html">layer_global_average_pooling_1d</a>()</code> <br><code><a href="layer_global_average_pooling_2d.html">layer_global_average_pooling_2d</a>()</code> <br><code><a href="layer_global_average_pooling_3d.html">layer_global_average_pooling_3d</a>()</code> <br><code><a href="layer_global_max_pooling_1d.html">layer_global_max_pooling_1d</a>()</code> <br><code><a href="layer_global_max_pooling_2d.html">layer_global_max_pooling_2d</a>()</code> <br><code><a href="layer_global_max_pooling_3d.html">layer_global_max_pooling_3d</a>()</code> <br><code><a href="layer_group_normalization.html">layer_group_normalization</a>()</code> <br><code><a href="layer_group_query_attention.html">layer_group_query_attention</a>()</code> <br><code><a href="layer_gru.html">layer_gru</a>()</code> <br><code><a href="layer_hashed_crossing.html">layer_hashed_crossing</a>()</code> <br><code><a href="layer_hashing.html">layer_hashing</a>()</code> <br><code><a href="layer_identity.html">layer_identity</a>()</code> <br><code><a href="layer_integer_lookup.html">layer_integer_lookup</a>()</code> <br><code><a href="layer_jax_model_wrapper.html">layer_jax_model_wrapper</a>()</code> <br><code><a href="layer_lambda.html">layer_lambda</a>()</code> <br><code><a href="layer_layer_normalization.html">layer_layer_normalization</a>()</code> <br><code><a href="layer_lstm.html">layer_lstm</a>()</code> <br><code><a href="layer_masking.html">layer_masking</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_max_pooling_1d.html">layer_max_pooling_1d</a>()</code> <br><code><a href="layer_max_pooling_2d.html">layer_max_pooling_2d</a>()</code> <br><code><a href="layer_max_pooling_3d.html">layer_max_pooling_3d</a>()</code> <br><code><a href="layer_maximum.html">layer_maximum</a>()</code> <br><code><a href="layer_mel_spectrogram.html">layer_mel_spectrogram</a>()</code> <br><code><a href="layer_minimum.html">layer_minimum</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_multi_head_attention.html">layer_multi_head_attention</a>()</code> <br><code><a href="layer_multiply.html">layer_multiply</a>()</code> <br><code><a href="layer_normalization.html">layer_normalization</a>()</code> <br><code><a href="layer_permute.html">layer_permute</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_brightness.html">layer_random_brightness</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_contrast.html">layer_random_contrast</a>()</code> <br><code><a href="layer_random_crop.html">layer_random_crop</a>()</code> <br><code><a href="layer_random_elastic_transform.html">layer_random_elastic_transform</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_flip.html">layer_random_flip</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_rotation.html">layer_random_rotation</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_random_translation.html">layer_random_translation</a>()</code> <br><code><a href="layer_random_zoom.html">layer_random_zoom</a>()</code> <br><code><a href="layer_repeat_vector.html">layer_repeat_vector</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_reshape.html">layer_reshape</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_rms_normalization.html">layer_rms_normalization</a>()</code> <br><code><a href="layer_separable_conv_1d.html">layer_separable_conv_1d</a>()</code> <br><code><a href="layer_separable_conv_2d.html">layer_separable_conv_2d</a>()</code> <br><code><a href="layer_simple_rnn.html">layer_simple_rnn</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br><code><a href="layer_spatial_dropout_1d.html">layer_spatial_dropout_1d</a>()</code> <br><code><a href="layer_spatial_dropout_2d.html">layer_spatial_dropout_2d</a>()</code> <br><code><a href="layer_spatial_dropout_3d.html">layer_spatial_dropout_3d</a>()</code> <br><code><a href="layer_spectral_normalization.html">layer_spectral_normalization</a>()</code> <br><code><a href="layer_stft_spectrogram.html">layer_stft_spectrogram</a>()</code> <br><code><a href="layer_string_lookup.html">layer_string_lookup</a>()</code> <br><code><a href="layer_subtract.html">layer_subtract</a>()</code> <br><code><a href="layer_text_vectorization.html">layer_text_vectorization</a>()</code> <br><code><a href="layer_tfsm.html">layer_tfsm</a>()</code> <br><code><a href="layer_time_distributed.html">layer_time_distributed</a>()</code> <br><code><a href="layer_torch_module_wrapper.html">layer_torch_module_wrapper</a>()</code> <br><code><a href="layer_unit_normalization.html">layer_unit_normalization</a>()</code> <br><code><a href="layer_upsampling_1d.html">layer_upsampling_1d</a>()</code> <br><code><a href="layer_upsampling_2d.html">layer_upsampling_2d</a>()</code> <br><code><a href="layer_upsampling_3d.html">layer_upsampling_3d</a>()</code> <br><code><a href="layer_zero_padding_1d.html">layer_zero_padding_1d</a>()</code> <br><code><a href="layer_zero_padding_2d.html">layer_zero_padding_2d</a>()</code> <br><code><a href="layer_zero_padding_3d.html">layer_zero_padding_3d</a>()</code> <br><code><a href="rnn_cell_gru.html">rnn_cell_gru</a>()</code> <br><code><a href="rnn_cell_lstm.html">rnn_cell_lstm</a>()</code> <br><code><a href="rnn_cell_simple.html">rnn_cell_simple</a>()</code> <br><code><a href="rnn_cells_stack.html">rnn_cells_stack</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -281,7 +280,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/layer_separable_conv_1d.html b/docs/dev/reference/layer_separable_conv_1d.html
index 731c46989..4c88d2d7b 100644
--- a/docs/dev/reference/layer_separable_conv_1d.html
+++ b/docs/dev/reference/layer_separable_conv_1d.html
@@ -16,7 +16,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -249,7 +249,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
     <div class="dont-index">
 <ul><li><p><a href="https://keras.io/api/layers/convolution_layers/separable_convolution1d#separableconv1d-class" class="external-link">https://keras.io/api/layers/convolution_layers/separable_convolution1d#separableconv1d-class</a></p></li>
 </ul><p>Other convolutional layers: <br><code><a href="layer_conv_1d.html">layer_conv_1d</a>()</code> <br><code><a href="layer_conv_1d_transpose.html">layer_conv_1d_transpose</a>()</code> <br><code><a href="layer_conv_2d.html">layer_conv_2d</a>()</code> <br><code><a href="layer_conv_2d_transpose.html">layer_conv_2d_transpose</a>()</code> <br><code><a href="layer_conv_3d.html">layer_conv_3d</a>()</code> <br><code><a href="layer_conv_3d_transpose.html">layer_conv_3d_transpose</a>()</code> <br><code><a href="layer_depthwise_conv_1d.html">layer_depthwise_conv_1d</a>()</code> <br><code><a href="layer_depthwise_conv_2d.html">layer_depthwise_conv_2d</a>()</code> <br><code><a href="layer_separable_conv_2d.html">layer_separable_conv_2d</a>()</code> <br></p>
-<p>Other layers: <br><code><a href="Layer.html">Layer</a>()</code> <br><code><a href="layer_activation.html">layer_activation</a>()</code> <br><code><a href="layer_activation_elu.html">layer_activation_elu</a>()</code> <br><code><a href="layer_activation_leaky_relu.html">layer_activation_leaky_relu</a>()</code> <br><code><a href="layer_activation_parametric_relu.html">layer_activation_parametric_relu</a>()</code> <br><code><a href="layer_activation_relu.html">layer_activation_relu</a>()</code> <br><code><a href="layer_activation_softmax.html">layer_activation_softmax</a>()</code> <br><code><a href="layer_activity_regularization.html">layer_activity_regularization</a>()</code> <br><code><a href="layer_add.html">layer_add</a>()</code> <br><code><a href="layer_additive_attention.html">layer_additive_attention</a>()</code> <br><code><a href="layer_alpha_dropout.html">layer_alpha_dropout</a>()</code> <br><code><a href="layer_attention.html">layer_attention</a>()</code> <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_average.html">layer_average</a>()</code> <br><code><a href="layer_average_pooling_1d.html">layer_average_pooling_1d</a>()</code> <br><code><a href="layer_average_pooling_2d.html">layer_average_pooling_2d</a>()</code> <br><code><a href="layer_average_pooling_3d.html">layer_average_pooling_3d</a>()</code> <br><code><a href="layer_batch_normalization.html">layer_batch_normalization</a>()</code> <br><code><a href="layer_bidirectional.html">layer_bidirectional</a>()</code> <br><code><a href="layer_category_encoding.html">layer_category_encoding</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_concatenate.html">layer_concatenate</a>()</code> <br><code><a href="layer_conv_1d.html">layer_conv_1d</a>()</code> <br><code><a href="layer_conv_1d_transpose.html">layer_conv_1d_transpose</a>()</code> <br><code><a href="layer_conv_2d.html">layer_conv_2d</a>()</code> <br><code><a href="layer_conv_2d_transpose.html">layer_conv_2d_transpose</a>()</code> <br><code><a href="layer_conv_3d.html">layer_conv_3d</a>()</code> <br><code><a href="layer_conv_3d_transpose.html">layer_conv_3d_transpose</a>()</code> <br><code><a href="layer_conv_lstm_1d.html">layer_conv_lstm_1d</a>()</code> <br><code><a href="layer_conv_lstm_2d.html">layer_conv_lstm_2d</a>()</code> <br><code><a href="layer_conv_lstm_3d.html">layer_conv_lstm_3d</a>()</code> <br><code><a href="layer_cropping_1d.html">layer_cropping_1d</a>()</code> <br><code><a href="layer_cropping_2d.html">layer_cropping_2d</a>()</code> <br><code><a href="layer_cropping_3d.html">layer_cropping_3d</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_dense.html">layer_dense</a>()</code> <br><code><a href="layer_depthwise_conv_1d.html">layer_depthwise_conv_1d</a>()</code> <br><code><a href="layer_depthwise_conv_2d.html">layer_depthwise_conv_2d</a>()</code> <br><code><a href="layer_discretization.html">layer_discretization</a>()</code> <br><code><a href="layer_dot.html">layer_dot</a>()</code> <br><code><a href="layer_dropout.html">layer_dropout</a>()</code> <br><code><a href="layer_einsum_dense.html">layer_einsum_dense</a>()</code> <br><code><a href="layer_embedding.html">layer_embedding</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_feature_space.html">layer_feature_space</a>()</code> <br><code><a href="layer_flatten.html">layer_flatten</a>()</code> <br><code><a href="layer_flax_module_wrapper.html">layer_flax_module_wrapper</a>()</code> <br><code><a href="layer_gaussian_dropout.html">layer_gaussian_dropout</a>()</code> <br><code><a href="layer_gaussian_noise.html">layer_gaussian_noise</a>()</code> <br><code><a href="layer_global_average_pooling_1d.html">layer_global_average_pooling_1d</a>()</code> <br><code><a href="layer_global_average_pooling_2d.html">layer_global_average_pooling_2d</a>()</code> <br><code><a href="layer_global_average_pooling_3d.html">layer_global_average_pooling_3d</a>()</code> <br><code><a href="layer_global_max_pooling_1d.html">layer_global_max_pooling_1d</a>()</code> <br><code><a href="layer_global_max_pooling_2d.html">layer_global_max_pooling_2d</a>()</code> <br><code><a href="layer_global_max_pooling_3d.html">layer_global_max_pooling_3d</a>()</code> <br><code><a href="layer_group_normalization.html">layer_group_normalization</a>()</code> <br><code><a href="layer_group_query_attention.html">layer_group_query_attention</a>()</code> <br><code><a href="layer_gru.html">layer_gru</a>()</code> <br><code><a href="layer_hashed_crossing.html">layer_hashed_crossing</a>()</code> <br><code><a href="layer_hashing.html">layer_hashing</a>()</code> <br><code><a href="layer_identity.html">layer_identity</a>()</code> <br><code><a href="layer_integer_lookup.html">layer_integer_lookup</a>()</code> <br><code><a href="layer_jax_model_wrapper.html">layer_jax_model_wrapper</a>()</code> <br><code><a href="layer_lambda.html">layer_lambda</a>()</code> <br><code><a href="layer_layer_normalization.html">layer_layer_normalization</a>()</code> <br><code><a href="layer_lstm.html">layer_lstm</a>()</code> <br><code><a href="layer_masking.html">layer_masking</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_max_pooling_1d.html">layer_max_pooling_1d</a>()</code> <br><code><a href="layer_max_pooling_2d.html">layer_max_pooling_2d</a>()</code> <br><code><a href="layer_max_pooling_3d.html">layer_max_pooling_3d</a>()</code> <br><code><a href="layer_maximum.html">layer_maximum</a>()</code> <br><code><a href="layer_mel_spectrogram.html">layer_mel_spectrogram</a>()</code> <br><code><a href="layer_minimum.html">layer_minimum</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_multi_head_attention.html">layer_multi_head_attention</a>()</code> <br><code><a href="layer_multiply.html">layer_multiply</a>()</code> <br><code><a href="layer_normalization.html">layer_normalization</a>()</code> <br><code><a href="layer_permute.html">layer_permute</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_brightness.html">layer_random_brightness</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_contrast.html">layer_random_contrast</a>()</code> <br><code><a href="layer_random_crop.html">layer_random_crop</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_flip.html">layer_random_flip</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_rotation.html">layer_random_rotation</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_random_translation.html">layer_random_translation</a>()</code> <br><code><a href="layer_random_zoom.html">layer_random_zoom</a>()</code> <br><code><a href="layer_repeat_vector.html">layer_repeat_vector</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_reshape.html">layer_reshape</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_rms_normalization.html">layer_rms_normalization</a>()</code> <br><code><a href="layer_rnn.html">layer_rnn</a>()</code> <br><code><a href="layer_separable_conv_2d.html">layer_separable_conv_2d</a>()</code> <br><code><a href="layer_simple_rnn.html">layer_simple_rnn</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br><code><a href="layer_spatial_dropout_1d.html">layer_spatial_dropout_1d</a>()</code> <br><code><a href="layer_spatial_dropout_2d.html">layer_spatial_dropout_2d</a>()</code> <br><code><a href="layer_spatial_dropout_3d.html">layer_spatial_dropout_3d</a>()</code> <br><code><a href="layer_spectral_normalization.html">layer_spectral_normalization</a>()</code> <br><code><a href="layer_stft_spectrogram.html">layer_stft_spectrogram</a>()</code> <br><code><a href="layer_string_lookup.html">layer_string_lookup</a>()</code> <br><code><a href="layer_subtract.html">layer_subtract</a>()</code> <br><code><a href="layer_text_vectorization.html">layer_text_vectorization</a>()</code> <br><code><a href="layer_tfsm.html">layer_tfsm</a>()</code> <br><code><a href="layer_time_distributed.html">layer_time_distributed</a>()</code> <br><code><a href="layer_torch_module_wrapper.html">layer_torch_module_wrapper</a>()</code> <br><code><a href="layer_unit_normalization.html">layer_unit_normalization</a>()</code> <br><code><a href="layer_upsampling_1d.html">layer_upsampling_1d</a>()</code> <br><code><a href="layer_upsampling_2d.html">layer_upsampling_2d</a>()</code> <br><code><a href="layer_upsampling_3d.html">layer_upsampling_3d</a>()</code> <br><code><a href="layer_zero_padding_1d.html">layer_zero_padding_1d</a>()</code> <br><code><a href="layer_zero_padding_2d.html">layer_zero_padding_2d</a>()</code> <br><code><a href="layer_zero_padding_3d.html">layer_zero_padding_3d</a>()</code> <br><code><a href="rnn_cell_gru.html">rnn_cell_gru</a>()</code> <br><code><a href="rnn_cell_lstm.html">rnn_cell_lstm</a>()</code> <br><code><a href="rnn_cell_simple.html">rnn_cell_simple</a>()</code> <br><code><a href="rnn_cells_stack.html">rnn_cells_stack</a>()</code> <br></p></div>
+<p>Other layers: <br><code><a href="Layer.html">Layer</a>()</code> <br><code><a href="layer_activation.html">layer_activation</a>()</code> <br><code><a href="layer_activation_elu.html">layer_activation_elu</a>()</code> <br><code><a href="layer_activation_leaky_relu.html">layer_activation_leaky_relu</a>()</code> <br><code><a href="layer_activation_parametric_relu.html">layer_activation_parametric_relu</a>()</code> <br><code><a href="layer_activation_relu.html">layer_activation_relu</a>()</code> <br><code><a href="layer_activation_softmax.html">layer_activation_softmax</a>()</code> <br><code><a href="layer_activity_regularization.html">layer_activity_regularization</a>()</code> <br><code><a href="layer_add.html">layer_add</a>()</code> <br><code><a href="layer_additive_attention.html">layer_additive_attention</a>()</code> <br><code><a href="layer_alpha_dropout.html">layer_alpha_dropout</a>()</code> <br><code><a href="layer_attention.html">layer_attention</a>()</code> <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_average.html">layer_average</a>()</code> <br><code><a href="layer_average_pooling_1d.html">layer_average_pooling_1d</a>()</code> <br><code><a href="layer_average_pooling_2d.html">layer_average_pooling_2d</a>()</code> <br><code><a href="layer_average_pooling_3d.html">layer_average_pooling_3d</a>()</code> <br><code><a href="layer_batch_normalization.html">layer_batch_normalization</a>()</code> <br><code><a href="layer_bidirectional.html">layer_bidirectional</a>()</code> <br><code><a href="layer_category_encoding.html">layer_category_encoding</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_concatenate.html">layer_concatenate</a>()</code> <br><code><a href="layer_conv_1d.html">layer_conv_1d</a>()</code> <br><code><a href="layer_conv_1d_transpose.html">layer_conv_1d_transpose</a>()</code> <br><code><a href="layer_conv_2d.html">layer_conv_2d</a>()</code> <br><code><a href="layer_conv_2d_transpose.html">layer_conv_2d_transpose</a>()</code> <br><code><a href="layer_conv_3d.html">layer_conv_3d</a>()</code> <br><code><a href="layer_conv_3d_transpose.html">layer_conv_3d_transpose</a>()</code> <br><code><a href="layer_conv_lstm_1d.html">layer_conv_lstm_1d</a>()</code> <br><code><a href="layer_conv_lstm_2d.html">layer_conv_lstm_2d</a>()</code> <br><code><a href="layer_conv_lstm_3d.html">layer_conv_lstm_3d</a>()</code> <br><code><a href="layer_cropping_1d.html">layer_cropping_1d</a>()</code> <br><code><a href="layer_cropping_2d.html">layer_cropping_2d</a>()</code> <br><code><a href="layer_cropping_3d.html">layer_cropping_3d</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_dense.html">layer_dense</a>()</code> <br><code><a href="layer_depthwise_conv_1d.html">layer_depthwise_conv_1d</a>()</code> <br><code><a href="layer_depthwise_conv_2d.html">layer_depthwise_conv_2d</a>()</code> <br><code><a href="layer_discretization.html">layer_discretization</a>()</code> <br><code><a href="layer_dot.html">layer_dot</a>()</code> <br><code><a href="layer_dropout.html">layer_dropout</a>()</code> <br><code><a href="layer_einsum_dense.html">layer_einsum_dense</a>()</code> <br><code><a href="layer_embedding.html">layer_embedding</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_feature_space.html">layer_feature_space</a>()</code> <br><code><a href="layer_flatten.html">layer_flatten</a>()</code> <br><code><a href="layer_flax_module_wrapper.html">layer_flax_module_wrapper</a>()</code> <br><code><a href="layer_gaussian_dropout.html">layer_gaussian_dropout</a>()</code> <br><code><a href="layer_gaussian_noise.html">layer_gaussian_noise</a>()</code> <br><code><a href="layer_global_average_pooling_1d.html">layer_global_average_pooling_1d</a>()</code> <br><code><a href="layer_global_average_pooling_2d.html">layer_global_average_pooling_2d</a>()</code> <br><code><a href="layer_global_average_pooling_3d.html">layer_global_average_pooling_3d</a>()</code> <br><code><a href="layer_global_max_pooling_1d.html">layer_global_max_pooling_1d</a>()</code> <br><code><a href="layer_global_max_pooling_2d.html">layer_global_max_pooling_2d</a>()</code> <br><code><a href="layer_global_max_pooling_3d.html">layer_global_max_pooling_3d</a>()</code> <br><code><a href="layer_group_normalization.html">layer_group_normalization</a>()</code> <br><code><a href="layer_group_query_attention.html">layer_group_query_attention</a>()</code> <br><code><a href="layer_gru.html">layer_gru</a>()</code> <br><code><a href="layer_hashed_crossing.html">layer_hashed_crossing</a>()</code> <br><code><a href="layer_hashing.html">layer_hashing</a>()</code> <br><code><a href="layer_identity.html">layer_identity</a>()</code> <br><code><a href="layer_integer_lookup.html">layer_integer_lookup</a>()</code> <br><code><a href="layer_jax_model_wrapper.html">layer_jax_model_wrapper</a>()</code> <br><code><a href="layer_lambda.html">layer_lambda</a>()</code> <br><code><a href="layer_layer_normalization.html">layer_layer_normalization</a>()</code> <br><code><a href="layer_lstm.html">layer_lstm</a>()</code> <br><code><a href="layer_masking.html">layer_masking</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_max_pooling_1d.html">layer_max_pooling_1d</a>()</code> <br><code><a href="layer_max_pooling_2d.html">layer_max_pooling_2d</a>()</code> <br><code><a href="layer_max_pooling_3d.html">layer_max_pooling_3d</a>()</code> <br><code><a href="layer_maximum.html">layer_maximum</a>()</code> <br><code><a href="layer_mel_spectrogram.html">layer_mel_spectrogram</a>()</code> <br><code><a href="layer_minimum.html">layer_minimum</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_multi_head_attention.html">layer_multi_head_attention</a>()</code> <br><code><a href="layer_multiply.html">layer_multiply</a>()</code> <br><code><a href="layer_normalization.html">layer_normalization</a>()</code> <br><code><a href="layer_permute.html">layer_permute</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_brightness.html">layer_random_brightness</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_contrast.html">layer_random_contrast</a>()</code> <br><code><a href="layer_random_crop.html">layer_random_crop</a>()</code> <br><code><a href="layer_random_elastic_transform.html">layer_random_elastic_transform</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_flip.html">layer_random_flip</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_rotation.html">layer_random_rotation</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_random_translation.html">layer_random_translation</a>()</code> <br><code><a href="layer_random_zoom.html">layer_random_zoom</a>()</code> <br><code><a href="layer_repeat_vector.html">layer_repeat_vector</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_reshape.html">layer_reshape</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_rms_normalization.html">layer_rms_normalization</a>()</code> <br><code><a href="layer_rnn.html">layer_rnn</a>()</code> <br><code><a href="layer_separable_conv_2d.html">layer_separable_conv_2d</a>()</code> <br><code><a href="layer_simple_rnn.html">layer_simple_rnn</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br><code><a href="layer_spatial_dropout_1d.html">layer_spatial_dropout_1d</a>()</code> <br><code><a href="layer_spatial_dropout_2d.html">layer_spatial_dropout_2d</a>()</code> <br><code><a href="layer_spatial_dropout_3d.html">layer_spatial_dropout_3d</a>()</code> <br><code><a href="layer_spectral_normalization.html">layer_spectral_normalization</a>()</code> <br><code><a href="layer_stft_spectrogram.html">layer_stft_spectrogram</a>()</code> <br><code><a href="layer_string_lookup.html">layer_string_lookup</a>()</code> <br><code><a href="layer_subtract.html">layer_subtract</a>()</code> <br><code><a href="layer_text_vectorization.html">layer_text_vectorization</a>()</code> <br><code><a href="layer_tfsm.html">layer_tfsm</a>()</code> <br><code><a href="layer_time_distributed.html">layer_time_distributed</a>()</code> <br><code><a href="layer_torch_module_wrapper.html">layer_torch_module_wrapper</a>()</code> <br><code><a href="layer_unit_normalization.html">layer_unit_normalization</a>()</code> <br><code><a href="layer_upsampling_1d.html">layer_upsampling_1d</a>()</code> <br><code><a href="layer_upsampling_2d.html">layer_upsampling_2d</a>()</code> <br><code><a href="layer_upsampling_3d.html">layer_upsampling_3d</a>()</code> <br><code><a href="layer_zero_padding_1d.html">layer_zero_padding_1d</a>()</code> <br><code><a href="layer_zero_padding_2d.html">layer_zero_padding_2d</a>()</code> <br><code><a href="layer_zero_padding_3d.html">layer_zero_padding_3d</a>()</code> <br><code><a href="rnn_cell_gru.html">rnn_cell_gru</a>()</code> <br><code><a href="rnn_cell_lstm.html">rnn_cell_lstm</a>()</code> <br><code><a href="rnn_cell_simple.html">rnn_cell_simple</a>()</code> <br><code><a href="rnn_cells_stack.html">rnn_cells_stack</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -261,7 +261,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/layer_separable_conv_2d.html b/docs/dev/reference/layer_separable_conv_2d.html
index 6b3bdacdb..dbfd526ac 100644
--- a/docs/dev/reference/layer_separable_conv_2d.html
+++ b/docs/dev/reference/layer_separable_conv_2d.html
@@ -16,7 +16,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -250,7 +250,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
     <div class="dont-index">
 <ul><li><p><a href="https://keras.io/api/layers/convolution_layers/separable_convolution2d#separableconv2d-class" class="external-link">https://keras.io/api/layers/convolution_layers/separable_convolution2d#separableconv2d-class</a></p></li>
 </ul><p>Other convolutional layers: <br><code><a href="layer_conv_1d.html">layer_conv_1d</a>()</code> <br><code><a href="layer_conv_1d_transpose.html">layer_conv_1d_transpose</a>()</code> <br><code><a href="layer_conv_2d.html">layer_conv_2d</a>()</code> <br><code><a href="layer_conv_2d_transpose.html">layer_conv_2d_transpose</a>()</code> <br><code><a href="layer_conv_3d.html">layer_conv_3d</a>()</code> <br><code><a href="layer_conv_3d_transpose.html">layer_conv_3d_transpose</a>()</code> <br><code><a href="layer_depthwise_conv_1d.html">layer_depthwise_conv_1d</a>()</code> <br><code><a href="layer_depthwise_conv_2d.html">layer_depthwise_conv_2d</a>()</code> <br><code><a href="layer_separable_conv_1d.html">layer_separable_conv_1d</a>()</code> <br></p>
-<p>Other layers: <br><code><a href="Layer.html">Layer</a>()</code> <br><code><a href="layer_activation.html">layer_activation</a>()</code> <br><code><a href="layer_activation_elu.html">layer_activation_elu</a>()</code> <br><code><a href="layer_activation_leaky_relu.html">layer_activation_leaky_relu</a>()</code> <br><code><a href="layer_activation_parametric_relu.html">layer_activation_parametric_relu</a>()</code> <br><code><a href="layer_activation_relu.html">layer_activation_relu</a>()</code> <br><code><a href="layer_activation_softmax.html">layer_activation_softmax</a>()</code> <br><code><a href="layer_activity_regularization.html">layer_activity_regularization</a>()</code> <br><code><a href="layer_add.html">layer_add</a>()</code> <br><code><a href="layer_additive_attention.html">layer_additive_attention</a>()</code> <br><code><a href="layer_alpha_dropout.html">layer_alpha_dropout</a>()</code> <br><code><a href="layer_attention.html">layer_attention</a>()</code> <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_average.html">layer_average</a>()</code> <br><code><a href="layer_average_pooling_1d.html">layer_average_pooling_1d</a>()</code> <br><code><a href="layer_average_pooling_2d.html">layer_average_pooling_2d</a>()</code> <br><code><a href="layer_average_pooling_3d.html">layer_average_pooling_3d</a>()</code> <br><code><a href="layer_batch_normalization.html">layer_batch_normalization</a>()</code> <br><code><a href="layer_bidirectional.html">layer_bidirectional</a>()</code> <br><code><a href="layer_category_encoding.html">layer_category_encoding</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_concatenate.html">layer_concatenate</a>()</code> <br><code><a href="layer_conv_1d.html">layer_conv_1d</a>()</code> <br><code><a href="layer_conv_1d_transpose.html">layer_conv_1d_transpose</a>()</code> <br><code><a href="layer_conv_2d.html">layer_conv_2d</a>()</code> <br><code><a href="layer_conv_2d_transpose.html">layer_conv_2d_transpose</a>()</code> <br><code><a href="layer_conv_3d.html">layer_conv_3d</a>()</code> <br><code><a href="layer_conv_3d_transpose.html">layer_conv_3d_transpose</a>()</code> <br><code><a href="layer_conv_lstm_1d.html">layer_conv_lstm_1d</a>()</code> <br><code><a href="layer_conv_lstm_2d.html">layer_conv_lstm_2d</a>()</code> <br><code><a href="layer_conv_lstm_3d.html">layer_conv_lstm_3d</a>()</code> <br><code><a href="layer_cropping_1d.html">layer_cropping_1d</a>()</code> <br><code><a href="layer_cropping_2d.html">layer_cropping_2d</a>()</code> <br><code><a href="layer_cropping_3d.html">layer_cropping_3d</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_dense.html">layer_dense</a>()</code> <br><code><a href="layer_depthwise_conv_1d.html">layer_depthwise_conv_1d</a>()</code> <br><code><a href="layer_depthwise_conv_2d.html">layer_depthwise_conv_2d</a>()</code> <br><code><a href="layer_discretization.html">layer_discretization</a>()</code> <br><code><a href="layer_dot.html">layer_dot</a>()</code> <br><code><a href="layer_dropout.html">layer_dropout</a>()</code> <br><code><a href="layer_einsum_dense.html">layer_einsum_dense</a>()</code> <br><code><a href="layer_embedding.html">layer_embedding</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_feature_space.html">layer_feature_space</a>()</code> <br><code><a href="layer_flatten.html">layer_flatten</a>()</code> <br><code><a href="layer_flax_module_wrapper.html">layer_flax_module_wrapper</a>()</code> <br><code><a href="layer_gaussian_dropout.html">layer_gaussian_dropout</a>()</code> <br><code><a href="layer_gaussian_noise.html">layer_gaussian_noise</a>()</code> <br><code><a href="layer_global_average_pooling_1d.html">layer_global_average_pooling_1d</a>()</code> <br><code><a href="layer_global_average_pooling_2d.html">layer_global_average_pooling_2d</a>()</code> <br><code><a href="layer_global_average_pooling_3d.html">layer_global_average_pooling_3d</a>()</code> <br><code><a href="layer_global_max_pooling_1d.html">layer_global_max_pooling_1d</a>()</code> <br><code><a href="layer_global_max_pooling_2d.html">layer_global_max_pooling_2d</a>()</code> <br><code><a href="layer_global_max_pooling_3d.html">layer_global_max_pooling_3d</a>()</code> <br><code><a href="layer_group_normalization.html">layer_group_normalization</a>()</code> <br><code><a href="layer_group_query_attention.html">layer_group_query_attention</a>()</code> <br><code><a href="layer_gru.html">layer_gru</a>()</code> <br><code><a href="layer_hashed_crossing.html">layer_hashed_crossing</a>()</code> <br><code><a href="layer_hashing.html">layer_hashing</a>()</code> <br><code><a href="layer_identity.html">layer_identity</a>()</code> <br><code><a href="layer_integer_lookup.html">layer_integer_lookup</a>()</code> <br><code><a href="layer_jax_model_wrapper.html">layer_jax_model_wrapper</a>()</code> <br><code><a href="layer_lambda.html">layer_lambda</a>()</code> <br><code><a href="layer_layer_normalization.html">layer_layer_normalization</a>()</code> <br><code><a href="layer_lstm.html">layer_lstm</a>()</code> <br><code><a href="layer_masking.html">layer_masking</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_max_pooling_1d.html">layer_max_pooling_1d</a>()</code> <br><code><a href="layer_max_pooling_2d.html">layer_max_pooling_2d</a>()</code> <br><code><a href="layer_max_pooling_3d.html">layer_max_pooling_3d</a>()</code> <br><code><a href="layer_maximum.html">layer_maximum</a>()</code> <br><code><a href="layer_mel_spectrogram.html">layer_mel_spectrogram</a>()</code> <br><code><a href="layer_minimum.html">layer_minimum</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_multi_head_attention.html">layer_multi_head_attention</a>()</code> <br><code><a href="layer_multiply.html">layer_multiply</a>()</code> <br><code><a href="layer_normalization.html">layer_normalization</a>()</code> <br><code><a href="layer_permute.html">layer_permute</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_brightness.html">layer_random_brightness</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_contrast.html">layer_random_contrast</a>()</code> <br><code><a href="layer_random_crop.html">layer_random_crop</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_flip.html">layer_random_flip</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_rotation.html">layer_random_rotation</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_random_translation.html">layer_random_translation</a>()</code> <br><code><a href="layer_random_zoom.html">layer_random_zoom</a>()</code> <br><code><a href="layer_repeat_vector.html">layer_repeat_vector</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_reshape.html">layer_reshape</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_rms_normalization.html">layer_rms_normalization</a>()</code> <br><code><a href="layer_rnn.html">layer_rnn</a>()</code> <br><code><a href="layer_separable_conv_1d.html">layer_separable_conv_1d</a>()</code> <br><code><a href="layer_simple_rnn.html">layer_simple_rnn</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br><code><a href="layer_spatial_dropout_1d.html">layer_spatial_dropout_1d</a>()</code> <br><code><a href="layer_spatial_dropout_2d.html">layer_spatial_dropout_2d</a>()</code> <br><code><a href="layer_spatial_dropout_3d.html">layer_spatial_dropout_3d</a>()</code> <br><code><a href="layer_spectral_normalization.html">layer_spectral_normalization</a>()</code> <br><code><a href="layer_stft_spectrogram.html">layer_stft_spectrogram</a>()</code> <br><code><a href="layer_string_lookup.html">layer_string_lookup</a>()</code> <br><code><a href="layer_subtract.html">layer_subtract</a>()</code> <br><code><a href="layer_text_vectorization.html">layer_text_vectorization</a>()</code> <br><code><a href="layer_tfsm.html">layer_tfsm</a>()</code> <br><code><a href="layer_time_distributed.html">layer_time_distributed</a>()</code> <br><code><a href="layer_torch_module_wrapper.html">layer_torch_module_wrapper</a>()</code> <br><code><a href="layer_unit_normalization.html">layer_unit_normalization</a>()</code> <br><code><a href="layer_upsampling_1d.html">layer_upsampling_1d</a>()</code> <br><code><a href="layer_upsampling_2d.html">layer_upsampling_2d</a>()</code> <br><code><a href="layer_upsampling_3d.html">layer_upsampling_3d</a>()</code> <br><code><a href="layer_zero_padding_1d.html">layer_zero_padding_1d</a>()</code> <br><code><a href="layer_zero_padding_2d.html">layer_zero_padding_2d</a>()</code> <br><code><a href="layer_zero_padding_3d.html">layer_zero_padding_3d</a>()</code> <br><code><a href="rnn_cell_gru.html">rnn_cell_gru</a>()</code> <br><code><a href="rnn_cell_lstm.html">rnn_cell_lstm</a>()</code> <br><code><a href="rnn_cell_simple.html">rnn_cell_simple</a>()</code> <br><code><a href="rnn_cells_stack.html">rnn_cells_stack</a>()</code> <br></p></div>
+<p>Other layers: <br><code><a href="Layer.html">Layer</a>()</code> <br><code><a href="layer_activation.html">layer_activation</a>()</code> <br><code><a href="layer_activation_elu.html">layer_activation_elu</a>()</code> <br><code><a href="layer_activation_leaky_relu.html">layer_activation_leaky_relu</a>()</code> <br><code><a href="layer_activation_parametric_relu.html">layer_activation_parametric_relu</a>()</code> <br><code><a href="layer_activation_relu.html">layer_activation_relu</a>()</code> <br><code><a href="layer_activation_softmax.html">layer_activation_softmax</a>()</code> <br><code><a href="layer_activity_regularization.html">layer_activity_regularization</a>()</code> <br><code><a href="layer_add.html">layer_add</a>()</code> <br><code><a href="layer_additive_attention.html">layer_additive_attention</a>()</code> <br><code><a href="layer_alpha_dropout.html">layer_alpha_dropout</a>()</code> <br><code><a href="layer_attention.html">layer_attention</a>()</code> <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_average.html">layer_average</a>()</code> <br><code><a href="layer_average_pooling_1d.html">layer_average_pooling_1d</a>()</code> <br><code><a href="layer_average_pooling_2d.html">layer_average_pooling_2d</a>()</code> <br><code><a href="layer_average_pooling_3d.html">layer_average_pooling_3d</a>()</code> <br><code><a href="layer_batch_normalization.html">layer_batch_normalization</a>()</code> <br><code><a href="layer_bidirectional.html">layer_bidirectional</a>()</code> <br><code><a href="layer_category_encoding.html">layer_category_encoding</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_concatenate.html">layer_concatenate</a>()</code> <br><code><a href="layer_conv_1d.html">layer_conv_1d</a>()</code> <br><code><a href="layer_conv_1d_transpose.html">layer_conv_1d_transpose</a>()</code> <br><code><a href="layer_conv_2d.html">layer_conv_2d</a>()</code> <br><code><a href="layer_conv_2d_transpose.html">layer_conv_2d_transpose</a>()</code> <br><code><a href="layer_conv_3d.html">layer_conv_3d</a>()</code> <br><code><a href="layer_conv_3d_transpose.html">layer_conv_3d_transpose</a>()</code> <br><code><a href="layer_conv_lstm_1d.html">layer_conv_lstm_1d</a>()</code> <br><code><a href="layer_conv_lstm_2d.html">layer_conv_lstm_2d</a>()</code> <br><code><a href="layer_conv_lstm_3d.html">layer_conv_lstm_3d</a>()</code> <br><code><a href="layer_cropping_1d.html">layer_cropping_1d</a>()</code> <br><code><a href="layer_cropping_2d.html">layer_cropping_2d</a>()</code> <br><code><a href="layer_cropping_3d.html">layer_cropping_3d</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_dense.html">layer_dense</a>()</code> <br><code><a href="layer_depthwise_conv_1d.html">layer_depthwise_conv_1d</a>()</code> <br><code><a href="layer_depthwise_conv_2d.html">layer_depthwise_conv_2d</a>()</code> <br><code><a href="layer_discretization.html">layer_discretization</a>()</code> <br><code><a href="layer_dot.html">layer_dot</a>()</code> <br><code><a href="layer_dropout.html">layer_dropout</a>()</code> <br><code><a href="layer_einsum_dense.html">layer_einsum_dense</a>()</code> <br><code><a href="layer_embedding.html">layer_embedding</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_feature_space.html">layer_feature_space</a>()</code> <br><code><a href="layer_flatten.html">layer_flatten</a>()</code> <br><code><a href="layer_flax_module_wrapper.html">layer_flax_module_wrapper</a>()</code> <br><code><a href="layer_gaussian_dropout.html">layer_gaussian_dropout</a>()</code> <br><code><a href="layer_gaussian_noise.html">layer_gaussian_noise</a>()</code> <br><code><a href="layer_global_average_pooling_1d.html">layer_global_average_pooling_1d</a>()</code> <br><code><a href="layer_global_average_pooling_2d.html">layer_global_average_pooling_2d</a>()</code> <br><code><a href="layer_global_average_pooling_3d.html">layer_global_average_pooling_3d</a>()</code> <br><code><a href="layer_global_max_pooling_1d.html">layer_global_max_pooling_1d</a>()</code> <br><code><a href="layer_global_max_pooling_2d.html">layer_global_max_pooling_2d</a>()</code> <br><code><a href="layer_global_max_pooling_3d.html">layer_global_max_pooling_3d</a>()</code> <br><code><a href="layer_group_normalization.html">layer_group_normalization</a>()</code> <br><code><a href="layer_group_query_attention.html">layer_group_query_attention</a>()</code> <br><code><a href="layer_gru.html">layer_gru</a>()</code> <br><code><a href="layer_hashed_crossing.html">layer_hashed_crossing</a>()</code> <br><code><a href="layer_hashing.html">layer_hashing</a>()</code> <br><code><a href="layer_identity.html">layer_identity</a>()</code> <br><code><a href="layer_integer_lookup.html">layer_integer_lookup</a>()</code> <br><code><a href="layer_jax_model_wrapper.html">layer_jax_model_wrapper</a>()</code> <br><code><a href="layer_lambda.html">layer_lambda</a>()</code> <br><code><a href="layer_layer_normalization.html">layer_layer_normalization</a>()</code> <br><code><a href="layer_lstm.html">layer_lstm</a>()</code> <br><code><a href="layer_masking.html">layer_masking</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_max_pooling_1d.html">layer_max_pooling_1d</a>()</code> <br><code><a href="layer_max_pooling_2d.html">layer_max_pooling_2d</a>()</code> <br><code><a href="layer_max_pooling_3d.html">layer_max_pooling_3d</a>()</code> <br><code><a href="layer_maximum.html">layer_maximum</a>()</code> <br><code><a href="layer_mel_spectrogram.html">layer_mel_spectrogram</a>()</code> <br><code><a href="layer_minimum.html">layer_minimum</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_multi_head_attention.html">layer_multi_head_attention</a>()</code> <br><code><a href="layer_multiply.html">layer_multiply</a>()</code> <br><code><a href="layer_normalization.html">layer_normalization</a>()</code> <br><code><a href="layer_permute.html">layer_permute</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_brightness.html">layer_random_brightness</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_contrast.html">layer_random_contrast</a>()</code> <br><code><a href="layer_random_crop.html">layer_random_crop</a>()</code> <br><code><a href="layer_random_elastic_transform.html">layer_random_elastic_transform</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_flip.html">layer_random_flip</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_rotation.html">layer_random_rotation</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_random_translation.html">layer_random_translation</a>()</code> <br><code><a href="layer_random_zoom.html">layer_random_zoom</a>()</code> <br><code><a href="layer_repeat_vector.html">layer_repeat_vector</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_reshape.html">layer_reshape</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_rms_normalization.html">layer_rms_normalization</a>()</code> <br><code><a href="layer_rnn.html">layer_rnn</a>()</code> <br><code><a href="layer_separable_conv_1d.html">layer_separable_conv_1d</a>()</code> <br><code><a href="layer_simple_rnn.html">layer_simple_rnn</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br><code><a href="layer_spatial_dropout_1d.html">layer_spatial_dropout_1d</a>()</code> <br><code><a href="layer_spatial_dropout_2d.html">layer_spatial_dropout_2d</a>()</code> <br><code><a href="layer_spatial_dropout_3d.html">layer_spatial_dropout_3d</a>()</code> <br><code><a href="layer_spectral_normalization.html">layer_spectral_normalization</a>()</code> <br><code><a href="layer_stft_spectrogram.html">layer_stft_spectrogram</a>()</code> <br><code><a href="layer_string_lookup.html">layer_string_lookup</a>()</code> <br><code><a href="layer_subtract.html">layer_subtract</a>()</code> <br><code><a href="layer_text_vectorization.html">layer_text_vectorization</a>()</code> <br><code><a href="layer_tfsm.html">layer_tfsm</a>()</code> <br><code><a href="layer_time_distributed.html">layer_time_distributed</a>()</code> <br><code><a href="layer_torch_module_wrapper.html">layer_torch_module_wrapper</a>()</code> <br><code><a href="layer_unit_normalization.html">layer_unit_normalization</a>()</code> <br><code><a href="layer_upsampling_1d.html">layer_upsampling_1d</a>()</code> <br><code><a href="layer_upsampling_2d.html">layer_upsampling_2d</a>()</code> <br><code><a href="layer_upsampling_3d.html">layer_upsampling_3d</a>()</code> <br><code><a href="layer_zero_padding_1d.html">layer_zero_padding_1d</a>()</code> <br><code><a href="layer_zero_padding_2d.html">layer_zero_padding_2d</a>()</code> <br><code><a href="layer_zero_padding_3d.html">layer_zero_padding_3d</a>()</code> <br><code><a href="rnn_cell_gru.html">rnn_cell_gru</a>()</code> <br><code><a href="rnn_cell_lstm.html">rnn_cell_lstm</a>()</code> <br><code><a href="rnn_cell_simple.html">rnn_cell_simple</a>()</code> <br><code><a href="rnn_cells_stack.html">rnn_cells_stack</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -262,7 +262,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/layer_simple_rnn.html b/docs/dev/reference/layer_simple_rnn.html
index 6ea9dae39..ef17015ac 100644
--- a/docs/dev/reference/layer_simple_rnn.html
+++ b/docs/dev/reference/layer_simple_rnn.html
@@ -8,7 +8,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -255,7 +255,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 <ul><li><p><a href="https://keras.io/api/layers/recurrent_layers/simple_rnn#simplernn-class" class="external-link">https://keras.io/api/layers/recurrent_layers/simple_rnn#simplernn-class</a></p></li>
 </ul><p>Other simple rnn layers: <br><code><a href="rnn_cell_simple.html">rnn_cell_simple</a>()</code> <br></p>
 <p>Other rnn layers: <br><code><a href="layer_bidirectional.html">layer_bidirectional</a>()</code> <br><code><a href="layer_conv_lstm_1d.html">layer_conv_lstm_1d</a>()</code> <br><code><a href="layer_conv_lstm_2d.html">layer_conv_lstm_2d</a>()</code> <br><code><a href="layer_conv_lstm_3d.html">layer_conv_lstm_3d</a>()</code> <br><code><a href="layer_gru.html">layer_gru</a>()</code> <br><code><a href="layer_lstm.html">layer_lstm</a>()</code> <br><code><a href="layer_rnn.html">layer_rnn</a>()</code> <br><code><a href="layer_time_distributed.html">layer_time_distributed</a>()</code> <br><code><a href="rnn_cell_gru.html">rnn_cell_gru</a>()</code> <br><code><a href="rnn_cell_lstm.html">rnn_cell_lstm</a>()</code> <br><code><a href="rnn_cell_simple.html">rnn_cell_simple</a>()</code> <br><code><a href="rnn_cells_stack.html">rnn_cells_stack</a>()</code> <br></p>
-<p>Other layers: <br><code><a href="Layer.html">Layer</a>()</code> <br><code><a href="layer_activation.html">layer_activation</a>()</code> <br><code><a href="layer_activation_elu.html">layer_activation_elu</a>()</code> <br><code><a href="layer_activation_leaky_relu.html">layer_activation_leaky_relu</a>()</code> <br><code><a href="layer_activation_parametric_relu.html">layer_activation_parametric_relu</a>()</code> <br><code><a href="layer_activation_relu.html">layer_activation_relu</a>()</code> <br><code><a href="layer_activation_softmax.html">layer_activation_softmax</a>()</code> <br><code><a href="layer_activity_regularization.html">layer_activity_regularization</a>()</code> <br><code><a href="layer_add.html">layer_add</a>()</code> <br><code><a href="layer_additive_attention.html">layer_additive_attention</a>()</code> <br><code><a href="layer_alpha_dropout.html">layer_alpha_dropout</a>()</code> <br><code><a href="layer_attention.html">layer_attention</a>()</code> <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_average.html">layer_average</a>()</code> <br><code><a href="layer_average_pooling_1d.html">layer_average_pooling_1d</a>()</code> <br><code><a href="layer_average_pooling_2d.html">layer_average_pooling_2d</a>()</code> <br><code><a href="layer_average_pooling_3d.html">layer_average_pooling_3d</a>()</code> <br><code><a href="layer_batch_normalization.html">layer_batch_normalization</a>()</code> <br><code><a href="layer_bidirectional.html">layer_bidirectional</a>()</code> <br><code><a href="layer_category_encoding.html">layer_category_encoding</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_concatenate.html">layer_concatenate</a>()</code> <br><code><a href="layer_conv_1d.html">layer_conv_1d</a>()</code> <br><code><a href="layer_conv_1d_transpose.html">layer_conv_1d_transpose</a>()</code> <br><code><a href="layer_conv_2d.html">layer_conv_2d</a>()</code> <br><code><a href="layer_conv_2d_transpose.html">layer_conv_2d_transpose</a>()</code> <br><code><a href="layer_conv_3d.html">layer_conv_3d</a>()</code> <br><code><a href="layer_conv_3d_transpose.html">layer_conv_3d_transpose</a>()</code> <br><code><a href="layer_conv_lstm_1d.html">layer_conv_lstm_1d</a>()</code> <br><code><a href="layer_conv_lstm_2d.html">layer_conv_lstm_2d</a>()</code> <br><code><a href="layer_conv_lstm_3d.html">layer_conv_lstm_3d</a>()</code> <br><code><a href="layer_cropping_1d.html">layer_cropping_1d</a>()</code> <br><code><a href="layer_cropping_2d.html">layer_cropping_2d</a>()</code> <br><code><a href="layer_cropping_3d.html">layer_cropping_3d</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_dense.html">layer_dense</a>()</code> <br><code><a href="layer_depthwise_conv_1d.html">layer_depthwise_conv_1d</a>()</code> <br><code><a href="layer_depthwise_conv_2d.html">layer_depthwise_conv_2d</a>()</code> <br><code><a href="layer_discretization.html">layer_discretization</a>()</code> <br><code><a href="layer_dot.html">layer_dot</a>()</code> <br><code><a href="layer_dropout.html">layer_dropout</a>()</code> <br><code><a href="layer_einsum_dense.html">layer_einsum_dense</a>()</code> <br><code><a href="layer_embedding.html">layer_embedding</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_feature_space.html">layer_feature_space</a>()</code> <br><code><a href="layer_flatten.html">layer_flatten</a>()</code> <br><code><a href="layer_flax_module_wrapper.html">layer_flax_module_wrapper</a>()</code> <br><code><a href="layer_gaussian_dropout.html">layer_gaussian_dropout</a>()</code> <br><code><a href="layer_gaussian_noise.html">layer_gaussian_noise</a>()</code> <br><code><a href="layer_global_average_pooling_1d.html">layer_global_average_pooling_1d</a>()</code> <br><code><a href="layer_global_average_pooling_2d.html">layer_global_average_pooling_2d</a>()</code> <br><code><a href="layer_global_average_pooling_3d.html">layer_global_average_pooling_3d</a>()</code> <br><code><a href="layer_global_max_pooling_1d.html">layer_global_max_pooling_1d</a>()</code> <br><code><a href="layer_global_max_pooling_2d.html">layer_global_max_pooling_2d</a>()</code> <br><code><a href="layer_global_max_pooling_3d.html">layer_global_max_pooling_3d</a>()</code> <br><code><a href="layer_group_normalization.html">layer_group_normalization</a>()</code> <br><code><a href="layer_group_query_attention.html">layer_group_query_attention</a>()</code> <br><code><a href="layer_gru.html">layer_gru</a>()</code> <br><code><a href="layer_hashed_crossing.html">layer_hashed_crossing</a>()</code> <br><code><a href="layer_hashing.html">layer_hashing</a>()</code> <br><code><a href="layer_identity.html">layer_identity</a>()</code> <br><code><a href="layer_integer_lookup.html">layer_integer_lookup</a>()</code> <br><code><a href="layer_jax_model_wrapper.html">layer_jax_model_wrapper</a>()</code> <br><code><a href="layer_lambda.html">layer_lambda</a>()</code> <br><code><a href="layer_layer_normalization.html">layer_layer_normalization</a>()</code> <br><code><a href="layer_lstm.html">layer_lstm</a>()</code> <br><code><a href="layer_masking.html">layer_masking</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_max_pooling_1d.html">layer_max_pooling_1d</a>()</code> <br><code><a href="layer_max_pooling_2d.html">layer_max_pooling_2d</a>()</code> <br><code><a href="layer_max_pooling_3d.html">layer_max_pooling_3d</a>()</code> <br><code><a href="layer_maximum.html">layer_maximum</a>()</code> <br><code><a href="layer_mel_spectrogram.html">layer_mel_spectrogram</a>()</code> <br><code><a href="layer_minimum.html">layer_minimum</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_multi_head_attention.html">layer_multi_head_attention</a>()</code> <br><code><a href="layer_multiply.html">layer_multiply</a>()</code> <br><code><a href="layer_normalization.html">layer_normalization</a>()</code> <br><code><a href="layer_permute.html">layer_permute</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_brightness.html">layer_random_brightness</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_contrast.html">layer_random_contrast</a>()</code> <br><code><a href="layer_random_crop.html">layer_random_crop</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_flip.html">layer_random_flip</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_rotation.html">layer_random_rotation</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_random_translation.html">layer_random_translation</a>()</code> <br><code><a href="layer_random_zoom.html">layer_random_zoom</a>()</code> <br><code><a href="layer_repeat_vector.html">layer_repeat_vector</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_reshape.html">layer_reshape</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_rms_normalization.html">layer_rms_normalization</a>()</code> <br><code><a href="layer_rnn.html">layer_rnn</a>()</code> <br><code><a href="layer_separable_conv_1d.html">layer_separable_conv_1d</a>()</code> <br><code><a href="layer_separable_conv_2d.html">layer_separable_conv_2d</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br><code><a href="layer_spatial_dropout_1d.html">layer_spatial_dropout_1d</a>()</code> <br><code><a href="layer_spatial_dropout_2d.html">layer_spatial_dropout_2d</a>()</code> <br><code><a href="layer_spatial_dropout_3d.html">layer_spatial_dropout_3d</a>()</code> <br><code><a href="layer_spectral_normalization.html">layer_spectral_normalization</a>()</code> <br><code><a href="layer_stft_spectrogram.html">layer_stft_spectrogram</a>()</code> <br><code><a href="layer_string_lookup.html">layer_string_lookup</a>()</code> <br><code><a href="layer_subtract.html">layer_subtract</a>()</code> <br><code><a href="layer_text_vectorization.html">layer_text_vectorization</a>()</code> <br><code><a href="layer_tfsm.html">layer_tfsm</a>()</code> <br><code><a href="layer_time_distributed.html">layer_time_distributed</a>()</code> <br><code><a href="layer_torch_module_wrapper.html">layer_torch_module_wrapper</a>()</code> <br><code><a href="layer_unit_normalization.html">layer_unit_normalization</a>()</code> <br><code><a href="layer_upsampling_1d.html">layer_upsampling_1d</a>()</code> <br><code><a href="layer_upsampling_2d.html">layer_upsampling_2d</a>()</code> <br><code><a href="layer_upsampling_3d.html">layer_upsampling_3d</a>()</code> <br><code><a href="layer_zero_padding_1d.html">layer_zero_padding_1d</a>()</code> <br><code><a href="layer_zero_padding_2d.html">layer_zero_padding_2d</a>()</code> <br><code><a href="layer_zero_padding_3d.html">layer_zero_padding_3d</a>()</code> <br><code><a href="rnn_cell_gru.html">rnn_cell_gru</a>()</code> <br><code><a href="rnn_cell_lstm.html">rnn_cell_lstm</a>()</code> <br><code><a href="rnn_cell_simple.html">rnn_cell_simple</a>()</code> <br><code><a href="rnn_cells_stack.html">rnn_cells_stack</a>()</code> <br></p></div>
+<p>Other layers: <br><code><a href="Layer.html">Layer</a>()</code> <br><code><a href="layer_activation.html">layer_activation</a>()</code> <br><code><a href="layer_activation_elu.html">layer_activation_elu</a>()</code> <br><code><a href="layer_activation_leaky_relu.html">layer_activation_leaky_relu</a>()</code> <br><code><a href="layer_activation_parametric_relu.html">layer_activation_parametric_relu</a>()</code> <br><code><a href="layer_activation_relu.html">layer_activation_relu</a>()</code> <br><code><a href="layer_activation_softmax.html">layer_activation_softmax</a>()</code> <br><code><a href="layer_activity_regularization.html">layer_activity_regularization</a>()</code> <br><code><a href="layer_add.html">layer_add</a>()</code> <br><code><a href="layer_additive_attention.html">layer_additive_attention</a>()</code> <br><code><a href="layer_alpha_dropout.html">layer_alpha_dropout</a>()</code> <br><code><a href="layer_attention.html">layer_attention</a>()</code> <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_average.html">layer_average</a>()</code> <br><code><a href="layer_average_pooling_1d.html">layer_average_pooling_1d</a>()</code> <br><code><a href="layer_average_pooling_2d.html">layer_average_pooling_2d</a>()</code> <br><code><a href="layer_average_pooling_3d.html">layer_average_pooling_3d</a>()</code> <br><code><a href="layer_batch_normalization.html">layer_batch_normalization</a>()</code> <br><code><a href="layer_bidirectional.html">layer_bidirectional</a>()</code> <br><code><a href="layer_category_encoding.html">layer_category_encoding</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_concatenate.html">layer_concatenate</a>()</code> <br><code><a href="layer_conv_1d.html">layer_conv_1d</a>()</code> <br><code><a href="layer_conv_1d_transpose.html">layer_conv_1d_transpose</a>()</code> <br><code><a href="layer_conv_2d.html">layer_conv_2d</a>()</code> <br><code><a href="layer_conv_2d_transpose.html">layer_conv_2d_transpose</a>()</code> <br><code><a href="layer_conv_3d.html">layer_conv_3d</a>()</code> <br><code><a href="layer_conv_3d_transpose.html">layer_conv_3d_transpose</a>()</code> <br><code><a href="layer_conv_lstm_1d.html">layer_conv_lstm_1d</a>()</code> <br><code><a href="layer_conv_lstm_2d.html">layer_conv_lstm_2d</a>()</code> <br><code><a href="layer_conv_lstm_3d.html">layer_conv_lstm_3d</a>()</code> <br><code><a href="layer_cropping_1d.html">layer_cropping_1d</a>()</code> <br><code><a href="layer_cropping_2d.html">layer_cropping_2d</a>()</code> <br><code><a href="layer_cropping_3d.html">layer_cropping_3d</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_dense.html">layer_dense</a>()</code> <br><code><a href="layer_depthwise_conv_1d.html">layer_depthwise_conv_1d</a>()</code> <br><code><a href="layer_depthwise_conv_2d.html">layer_depthwise_conv_2d</a>()</code> <br><code><a href="layer_discretization.html">layer_discretization</a>()</code> <br><code><a href="layer_dot.html">layer_dot</a>()</code> <br><code><a href="layer_dropout.html">layer_dropout</a>()</code> <br><code><a href="layer_einsum_dense.html">layer_einsum_dense</a>()</code> <br><code><a href="layer_embedding.html">layer_embedding</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_feature_space.html">layer_feature_space</a>()</code> <br><code><a href="layer_flatten.html">layer_flatten</a>()</code> <br><code><a href="layer_flax_module_wrapper.html">layer_flax_module_wrapper</a>()</code> <br><code><a href="layer_gaussian_dropout.html">layer_gaussian_dropout</a>()</code> <br><code><a href="layer_gaussian_noise.html">layer_gaussian_noise</a>()</code> <br><code><a href="layer_global_average_pooling_1d.html">layer_global_average_pooling_1d</a>()</code> <br><code><a href="layer_global_average_pooling_2d.html">layer_global_average_pooling_2d</a>()</code> <br><code><a href="layer_global_average_pooling_3d.html">layer_global_average_pooling_3d</a>()</code> <br><code><a href="layer_global_max_pooling_1d.html">layer_global_max_pooling_1d</a>()</code> <br><code><a href="layer_global_max_pooling_2d.html">layer_global_max_pooling_2d</a>()</code> <br><code><a href="layer_global_max_pooling_3d.html">layer_global_max_pooling_3d</a>()</code> <br><code><a href="layer_group_normalization.html">layer_group_normalization</a>()</code> <br><code><a href="layer_group_query_attention.html">layer_group_query_attention</a>()</code> <br><code><a href="layer_gru.html">layer_gru</a>()</code> <br><code><a href="layer_hashed_crossing.html">layer_hashed_crossing</a>()</code> <br><code><a href="layer_hashing.html">layer_hashing</a>()</code> <br><code><a href="layer_identity.html">layer_identity</a>()</code> <br><code><a href="layer_integer_lookup.html">layer_integer_lookup</a>()</code> <br><code><a href="layer_jax_model_wrapper.html">layer_jax_model_wrapper</a>()</code> <br><code><a href="layer_lambda.html">layer_lambda</a>()</code> <br><code><a href="layer_layer_normalization.html">layer_layer_normalization</a>()</code> <br><code><a href="layer_lstm.html">layer_lstm</a>()</code> <br><code><a href="layer_masking.html">layer_masking</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_max_pooling_1d.html">layer_max_pooling_1d</a>()</code> <br><code><a href="layer_max_pooling_2d.html">layer_max_pooling_2d</a>()</code> <br><code><a href="layer_max_pooling_3d.html">layer_max_pooling_3d</a>()</code> <br><code><a href="layer_maximum.html">layer_maximum</a>()</code> <br><code><a href="layer_mel_spectrogram.html">layer_mel_spectrogram</a>()</code> <br><code><a href="layer_minimum.html">layer_minimum</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_multi_head_attention.html">layer_multi_head_attention</a>()</code> <br><code><a href="layer_multiply.html">layer_multiply</a>()</code> <br><code><a href="layer_normalization.html">layer_normalization</a>()</code> <br><code><a href="layer_permute.html">layer_permute</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_brightness.html">layer_random_brightness</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_contrast.html">layer_random_contrast</a>()</code> <br><code><a href="layer_random_crop.html">layer_random_crop</a>()</code> <br><code><a href="layer_random_elastic_transform.html">layer_random_elastic_transform</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_flip.html">layer_random_flip</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_rotation.html">layer_random_rotation</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_random_translation.html">layer_random_translation</a>()</code> <br><code><a href="layer_random_zoom.html">layer_random_zoom</a>()</code> <br><code><a href="layer_repeat_vector.html">layer_repeat_vector</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_reshape.html">layer_reshape</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_rms_normalization.html">layer_rms_normalization</a>()</code> <br><code><a href="layer_rnn.html">layer_rnn</a>()</code> <br><code><a href="layer_separable_conv_1d.html">layer_separable_conv_1d</a>()</code> <br><code><a href="layer_separable_conv_2d.html">layer_separable_conv_2d</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br><code><a href="layer_spatial_dropout_1d.html">layer_spatial_dropout_1d</a>()</code> <br><code><a href="layer_spatial_dropout_2d.html">layer_spatial_dropout_2d</a>()</code> <br><code><a href="layer_spatial_dropout_3d.html">layer_spatial_dropout_3d</a>()</code> <br><code><a href="layer_spectral_normalization.html">layer_spectral_normalization</a>()</code> <br><code><a href="layer_stft_spectrogram.html">layer_stft_spectrogram</a>()</code> <br><code><a href="layer_string_lookup.html">layer_string_lookup</a>()</code> <br><code><a href="layer_subtract.html">layer_subtract</a>()</code> <br><code><a href="layer_text_vectorization.html">layer_text_vectorization</a>()</code> <br><code><a href="layer_tfsm.html">layer_tfsm</a>()</code> <br><code><a href="layer_time_distributed.html">layer_time_distributed</a>()</code> <br><code><a href="layer_torch_module_wrapper.html">layer_torch_module_wrapper</a>()</code> <br><code><a href="layer_unit_normalization.html">layer_unit_normalization</a>()</code> <br><code><a href="layer_upsampling_1d.html">layer_upsampling_1d</a>()</code> <br><code><a href="layer_upsampling_2d.html">layer_upsampling_2d</a>()</code> <br><code><a href="layer_upsampling_3d.html">layer_upsampling_3d</a>()</code> <br><code><a href="layer_zero_padding_1d.html">layer_zero_padding_1d</a>()</code> <br><code><a href="layer_zero_padding_2d.html">layer_zero_padding_2d</a>()</code> <br><code><a href="layer_zero_padding_3d.html">layer_zero_padding_3d</a>()</code> <br><code><a href="rnn_cell_gru.html">rnn_cell_gru</a>()</code> <br><code><a href="rnn_cell_lstm.html">rnn_cell_lstm</a>()</code> <br><code><a href="rnn_cell_simple.html">rnn_cell_simple</a>()</code> <br><code><a href="rnn_cells_stack.html">rnn_cells_stack</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -267,7 +267,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/layer_solarization.html b/docs/dev/reference/layer_solarization.html
index bfc8edd31..21dece7c9 100644
--- a/docs/dev/reference/layer_solarization.html
+++ b/docs/dev/reference/layer_solarization.html
@@ -12,7 +12,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -136,9 +136,9 @@ <h2 id="examples">Examples<a class="anchor" aria-label="anchor" href="#examples"
     </div>
     <div class="section level2">
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
-    <div class="dont-index"><p>Other image preprocessing layers: <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br></p>
-<p>Other preprocessing layers: <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_category_encoding.html">layer_category_encoding</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_discretization.html">layer_discretization</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_feature_space.html">layer_feature_space</a>()</code> <br><code><a href="layer_hashed_crossing.html">layer_hashed_crossing</a>()</code> <br><code><a href="layer_hashing.html">layer_hashing</a>()</code> <br><code><a href="layer_integer_lookup.html">layer_integer_lookup</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_mel_spectrogram.html">layer_mel_spectrogram</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_normalization.html">layer_normalization</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_brightness.html">layer_random_brightness</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_contrast.html">layer_random_contrast</a>()</code> <br><code><a href="layer_random_crop.html">layer_random_crop</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_flip.html">layer_random_flip</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_rotation.html">layer_random_rotation</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_random_translation.html">layer_random_translation</a>()</code> <br><code><a href="layer_random_zoom.html">layer_random_zoom</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_stft_spectrogram.html">layer_stft_spectrogram</a>()</code> <br><code><a href="layer_string_lookup.html">layer_string_lookup</a>()</code> <br><code><a href="layer_text_vectorization.html">layer_text_vectorization</a>()</code> <br></p>
-<p>Other layers: <br><code><a href="Layer.html">Layer</a>()</code> <br><code><a href="layer_activation.html">layer_activation</a>()</code> <br><code><a href="layer_activation_elu.html">layer_activation_elu</a>()</code> <br><code><a href="layer_activation_leaky_relu.html">layer_activation_leaky_relu</a>()</code> <br><code><a href="layer_activation_parametric_relu.html">layer_activation_parametric_relu</a>()</code> <br><code><a href="layer_activation_relu.html">layer_activation_relu</a>()</code> <br><code><a href="layer_activation_softmax.html">layer_activation_softmax</a>()</code> <br><code><a href="layer_activity_regularization.html">layer_activity_regularization</a>()</code> <br><code><a href="layer_add.html">layer_add</a>()</code> <br><code><a href="layer_additive_attention.html">layer_additive_attention</a>()</code> <br><code><a href="layer_alpha_dropout.html">layer_alpha_dropout</a>()</code> <br><code><a href="layer_attention.html">layer_attention</a>()</code> <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_average.html">layer_average</a>()</code> <br><code><a href="layer_average_pooling_1d.html">layer_average_pooling_1d</a>()</code> <br><code><a href="layer_average_pooling_2d.html">layer_average_pooling_2d</a>()</code> <br><code><a href="layer_average_pooling_3d.html">layer_average_pooling_3d</a>()</code> <br><code><a href="layer_batch_normalization.html">layer_batch_normalization</a>()</code> <br><code><a href="layer_bidirectional.html">layer_bidirectional</a>()</code> <br><code><a href="layer_category_encoding.html">layer_category_encoding</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_concatenate.html">layer_concatenate</a>()</code> <br><code><a href="layer_conv_1d.html">layer_conv_1d</a>()</code> <br><code><a href="layer_conv_1d_transpose.html">layer_conv_1d_transpose</a>()</code> <br><code><a href="layer_conv_2d.html">layer_conv_2d</a>()</code> <br><code><a href="layer_conv_2d_transpose.html">layer_conv_2d_transpose</a>()</code> <br><code><a href="layer_conv_3d.html">layer_conv_3d</a>()</code> <br><code><a href="layer_conv_3d_transpose.html">layer_conv_3d_transpose</a>()</code> <br><code><a href="layer_conv_lstm_1d.html">layer_conv_lstm_1d</a>()</code> <br><code><a href="layer_conv_lstm_2d.html">layer_conv_lstm_2d</a>()</code> <br><code><a href="layer_conv_lstm_3d.html">layer_conv_lstm_3d</a>()</code> <br><code><a href="layer_cropping_1d.html">layer_cropping_1d</a>()</code> <br><code><a href="layer_cropping_2d.html">layer_cropping_2d</a>()</code> <br><code><a href="layer_cropping_3d.html">layer_cropping_3d</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_dense.html">layer_dense</a>()</code> <br><code><a href="layer_depthwise_conv_1d.html">layer_depthwise_conv_1d</a>()</code> <br><code><a href="layer_depthwise_conv_2d.html">layer_depthwise_conv_2d</a>()</code> <br><code><a href="layer_discretization.html">layer_discretization</a>()</code> <br><code><a href="layer_dot.html">layer_dot</a>()</code> <br><code><a href="layer_dropout.html">layer_dropout</a>()</code> <br><code><a href="layer_einsum_dense.html">layer_einsum_dense</a>()</code> <br><code><a href="layer_embedding.html">layer_embedding</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_feature_space.html">layer_feature_space</a>()</code> <br><code><a href="layer_flatten.html">layer_flatten</a>()</code> <br><code><a href="layer_flax_module_wrapper.html">layer_flax_module_wrapper</a>()</code> <br><code><a href="layer_gaussian_dropout.html">layer_gaussian_dropout</a>()</code> <br><code><a href="layer_gaussian_noise.html">layer_gaussian_noise</a>()</code> <br><code><a href="layer_global_average_pooling_1d.html">layer_global_average_pooling_1d</a>()</code> <br><code><a href="layer_global_average_pooling_2d.html">layer_global_average_pooling_2d</a>()</code> <br><code><a href="layer_global_average_pooling_3d.html">layer_global_average_pooling_3d</a>()</code> <br><code><a href="layer_global_max_pooling_1d.html">layer_global_max_pooling_1d</a>()</code> <br><code><a href="layer_global_max_pooling_2d.html">layer_global_max_pooling_2d</a>()</code> <br><code><a href="layer_global_max_pooling_3d.html">layer_global_max_pooling_3d</a>()</code> <br><code><a href="layer_group_normalization.html">layer_group_normalization</a>()</code> <br><code><a href="layer_group_query_attention.html">layer_group_query_attention</a>()</code> <br><code><a href="layer_gru.html">layer_gru</a>()</code> <br><code><a href="layer_hashed_crossing.html">layer_hashed_crossing</a>()</code> <br><code><a href="layer_hashing.html">layer_hashing</a>()</code> <br><code><a href="layer_identity.html">layer_identity</a>()</code> <br><code><a href="layer_integer_lookup.html">layer_integer_lookup</a>()</code> <br><code><a href="layer_jax_model_wrapper.html">layer_jax_model_wrapper</a>()</code> <br><code><a href="layer_lambda.html">layer_lambda</a>()</code> <br><code><a href="layer_layer_normalization.html">layer_layer_normalization</a>()</code> <br><code><a href="layer_lstm.html">layer_lstm</a>()</code> <br><code><a href="layer_masking.html">layer_masking</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_max_pooling_1d.html">layer_max_pooling_1d</a>()</code> <br><code><a href="layer_max_pooling_2d.html">layer_max_pooling_2d</a>()</code> <br><code><a href="layer_max_pooling_3d.html">layer_max_pooling_3d</a>()</code> <br><code><a href="layer_maximum.html">layer_maximum</a>()</code> <br><code><a href="layer_mel_spectrogram.html">layer_mel_spectrogram</a>()</code> <br><code><a href="layer_minimum.html">layer_minimum</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_multi_head_attention.html">layer_multi_head_attention</a>()</code> <br><code><a href="layer_multiply.html">layer_multiply</a>()</code> <br><code><a href="layer_normalization.html">layer_normalization</a>()</code> <br><code><a href="layer_permute.html">layer_permute</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_brightness.html">layer_random_brightness</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_contrast.html">layer_random_contrast</a>()</code> <br><code><a href="layer_random_crop.html">layer_random_crop</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_flip.html">layer_random_flip</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_rotation.html">layer_random_rotation</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_random_translation.html">layer_random_translation</a>()</code> <br><code><a href="layer_random_zoom.html">layer_random_zoom</a>()</code> <br><code><a href="layer_repeat_vector.html">layer_repeat_vector</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_reshape.html">layer_reshape</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_rms_normalization.html">layer_rms_normalization</a>()</code> <br><code><a href="layer_rnn.html">layer_rnn</a>()</code> <br><code><a href="layer_separable_conv_1d.html">layer_separable_conv_1d</a>()</code> <br><code><a href="layer_separable_conv_2d.html">layer_separable_conv_2d</a>()</code> <br><code><a href="layer_simple_rnn.html">layer_simple_rnn</a>()</code> <br><code><a href="layer_spatial_dropout_1d.html">layer_spatial_dropout_1d</a>()</code> <br><code><a href="layer_spatial_dropout_2d.html">layer_spatial_dropout_2d</a>()</code> <br><code><a href="layer_spatial_dropout_3d.html">layer_spatial_dropout_3d</a>()</code> <br><code><a href="layer_spectral_normalization.html">layer_spectral_normalization</a>()</code> <br><code><a href="layer_stft_spectrogram.html">layer_stft_spectrogram</a>()</code> <br><code><a href="layer_string_lookup.html">layer_string_lookup</a>()</code> <br><code><a href="layer_subtract.html">layer_subtract</a>()</code> <br><code><a href="layer_text_vectorization.html">layer_text_vectorization</a>()</code> <br><code><a href="layer_tfsm.html">layer_tfsm</a>()</code> <br><code><a href="layer_time_distributed.html">layer_time_distributed</a>()</code> <br><code><a href="layer_torch_module_wrapper.html">layer_torch_module_wrapper</a>()</code> <br><code><a href="layer_unit_normalization.html">layer_unit_normalization</a>()</code> <br><code><a href="layer_upsampling_1d.html">layer_upsampling_1d</a>()</code> <br><code><a href="layer_upsampling_2d.html">layer_upsampling_2d</a>()</code> <br><code><a href="layer_upsampling_3d.html">layer_upsampling_3d</a>()</code> <br><code><a href="layer_zero_padding_1d.html">layer_zero_padding_1d</a>()</code> <br><code><a href="layer_zero_padding_2d.html">layer_zero_padding_2d</a>()</code> <br><code><a href="layer_zero_padding_3d.html">layer_zero_padding_3d</a>()</code> <br><code><a href="rnn_cell_gru.html">rnn_cell_gru</a>()</code> <br><code><a href="rnn_cell_lstm.html">rnn_cell_lstm</a>()</code> <br><code><a href="rnn_cell_simple.html">rnn_cell_simple</a>()</code> <br><code><a href="rnn_cells_stack.html">rnn_cells_stack</a>()</code> <br></p></div>
+    <div class="dont-index"><p>Other image preprocessing layers: <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_elastic_transform.html">layer_random_elastic_transform</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br></p>
+<p>Other preprocessing layers: <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_category_encoding.html">layer_category_encoding</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_discretization.html">layer_discretization</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_feature_space.html">layer_feature_space</a>()</code> <br><code><a href="layer_hashed_crossing.html">layer_hashed_crossing</a>()</code> <br><code><a href="layer_hashing.html">layer_hashing</a>()</code> <br><code><a href="layer_integer_lookup.html">layer_integer_lookup</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_mel_spectrogram.html">layer_mel_spectrogram</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_normalization.html">layer_normalization</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_brightness.html">layer_random_brightness</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_contrast.html">layer_random_contrast</a>()</code> <br><code><a href="layer_random_crop.html">layer_random_crop</a>()</code> <br><code><a href="layer_random_elastic_transform.html">layer_random_elastic_transform</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_flip.html">layer_random_flip</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_rotation.html">layer_random_rotation</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_random_translation.html">layer_random_translation</a>()</code> <br><code><a href="layer_random_zoom.html">layer_random_zoom</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_stft_spectrogram.html">layer_stft_spectrogram</a>()</code> <br><code><a href="layer_string_lookup.html">layer_string_lookup</a>()</code> <br><code><a href="layer_text_vectorization.html">layer_text_vectorization</a>()</code> <br></p>
+<p>Other layers: <br><code><a href="Layer.html">Layer</a>()</code> <br><code><a href="layer_activation.html">layer_activation</a>()</code> <br><code><a href="layer_activation_elu.html">layer_activation_elu</a>()</code> <br><code><a href="layer_activation_leaky_relu.html">layer_activation_leaky_relu</a>()</code> <br><code><a href="layer_activation_parametric_relu.html">layer_activation_parametric_relu</a>()</code> <br><code><a href="layer_activation_relu.html">layer_activation_relu</a>()</code> <br><code><a href="layer_activation_softmax.html">layer_activation_softmax</a>()</code> <br><code><a href="layer_activity_regularization.html">layer_activity_regularization</a>()</code> <br><code><a href="layer_add.html">layer_add</a>()</code> <br><code><a href="layer_additive_attention.html">layer_additive_attention</a>()</code> <br><code><a href="layer_alpha_dropout.html">layer_alpha_dropout</a>()</code> <br><code><a href="layer_attention.html">layer_attention</a>()</code> <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_average.html">layer_average</a>()</code> <br><code><a href="layer_average_pooling_1d.html">layer_average_pooling_1d</a>()</code> <br><code><a href="layer_average_pooling_2d.html">layer_average_pooling_2d</a>()</code> <br><code><a href="layer_average_pooling_3d.html">layer_average_pooling_3d</a>()</code> <br><code><a href="layer_batch_normalization.html">layer_batch_normalization</a>()</code> <br><code><a href="layer_bidirectional.html">layer_bidirectional</a>()</code> <br><code><a href="layer_category_encoding.html">layer_category_encoding</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_concatenate.html">layer_concatenate</a>()</code> <br><code><a href="layer_conv_1d.html">layer_conv_1d</a>()</code> <br><code><a href="layer_conv_1d_transpose.html">layer_conv_1d_transpose</a>()</code> <br><code><a href="layer_conv_2d.html">layer_conv_2d</a>()</code> <br><code><a href="layer_conv_2d_transpose.html">layer_conv_2d_transpose</a>()</code> <br><code><a href="layer_conv_3d.html">layer_conv_3d</a>()</code> <br><code><a href="layer_conv_3d_transpose.html">layer_conv_3d_transpose</a>()</code> <br><code><a href="layer_conv_lstm_1d.html">layer_conv_lstm_1d</a>()</code> <br><code><a href="layer_conv_lstm_2d.html">layer_conv_lstm_2d</a>()</code> <br><code><a href="layer_conv_lstm_3d.html">layer_conv_lstm_3d</a>()</code> <br><code><a href="layer_cropping_1d.html">layer_cropping_1d</a>()</code> <br><code><a href="layer_cropping_2d.html">layer_cropping_2d</a>()</code> <br><code><a href="layer_cropping_3d.html">layer_cropping_3d</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_dense.html">layer_dense</a>()</code> <br><code><a href="layer_depthwise_conv_1d.html">layer_depthwise_conv_1d</a>()</code> <br><code><a href="layer_depthwise_conv_2d.html">layer_depthwise_conv_2d</a>()</code> <br><code><a href="layer_discretization.html">layer_discretization</a>()</code> <br><code><a href="layer_dot.html">layer_dot</a>()</code> <br><code><a href="layer_dropout.html">layer_dropout</a>()</code> <br><code><a href="layer_einsum_dense.html">layer_einsum_dense</a>()</code> <br><code><a href="layer_embedding.html">layer_embedding</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_feature_space.html">layer_feature_space</a>()</code> <br><code><a href="layer_flatten.html">layer_flatten</a>()</code> <br><code><a href="layer_flax_module_wrapper.html">layer_flax_module_wrapper</a>()</code> <br><code><a href="layer_gaussian_dropout.html">layer_gaussian_dropout</a>()</code> <br><code><a href="layer_gaussian_noise.html">layer_gaussian_noise</a>()</code> <br><code><a href="layer_global_average_pooling_1d.html">layer_global_average_pooling_1d</a>()</code> <br><code><a href="layer_global_average_pooling_2d.html">layer_global_average_pooling_2d</a>()</code> <br><code><a href="layer_global_average_pooling_3d.html">layer_global_average_pooling_3d</a>()</code> <br><code><a href="layer_global_max_pooling_1d.html">layer_global_max_pooling_1d</a>()</code> <br><code><a href="layer_global_max_pooling_2d.html">layer_global_max_pooling_2d</a>()</code> <br><code><a href="layer_global_max_pooling_3d.html">layer_global_max_pooling_3d</a>()</code> <br><code><a href="layer_group_normalization.html">layer_group_normalization</a>()</code> <br><code><a href="layer_group_query_attention.html">layer_group_query_attention</a>()</code> <br><code><a href="layer_gru.html">layer_gru</a>()</code> <br><code><a href="layer_hashed_crossing.html">layer_hashed_crossing</a>()</code> <br><code><a href="layer_hashing.html">layer_hashing</a>()</code> <br><code><a href="layer_identity.html">layer_identity</a>()</code> <br><code><a href="layer_integer_lookup.html">layer_integer_lookup</a>()</code> <br><code><a href="layer_jax_model_wrapper.html">layer_jax_model_wrapper</a>()</code> <br><code><a href="layer_lambda.html">layer_lambda</a>()</code> <br><code><a href="layer_layer_normalization.html">layer_layer_normalization</a>()</code> <br><code><a href="layer_lstm.html">layer_lstm</a>()</code> <br><code><a href="layer_masking.html">layer_masking</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_max_pooling_1d.html">layer_max_pooling_1d</a>()</code> <br><code><a href="layer_max_pooling_2d.html">layer_max_pooling_2d</a>()</code> <br><code><a href="layer_max_pooling_3d.html">layer_max_pooling_3d</a>()</code> <br><code><a href="layer_maximum.html">layer_maximum</a>()</code> <br><code><a href="layer_mel_spectrogram.html">layer_mel_spectrogram</a>()</code> <br><code><a href="layer_minimum.html">layer_minimum</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_multi_head_attention.html">layer_multi_head_attention</a>()</code> <br><code><a href="layer_multiply.html">layer_multiply</a>()</code> <br><code><a href="layer_normalization.html">layer_normalization</a>()</code> <br><code><a href="layer_permute.html">layer_permute</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_brightness.html">layer_random_brightness</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_contrast.html">layer_random_contrast</a>()</code> <br><code><a href="layer_random_crop.html">layer_random_crop</a>()</code> <br><code><a href="layer_random_elastic_transform.html">layer_random_elastic_transform</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_flip.html">layer_random_flip</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_rotation.html">layer_random_rotation</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_random_translation.html">layer_random_translation</a>()</code> <br><code><a href="layer_random_zoom.html">layer_random_zoom</a>()</code> <br><code><a href="layer_repeat_vector.html">layer_repeat_vector</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_reshape.html">layer_reshape</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_rms_normalization.html">layer_rms_normalization</a>()</code> <br><code><a href="layer_rnn.html">layer_rnn</a>()</code> <br><code><a href="layer_separable_conv_1d.html">layer_separable_conv_1d</a>()</code> <br><code><a href="layer_separable_conv_2d.html">layer_separable_conv_2d</a>()</code> <br><code><a href="layer_simple_rnn.html">layer_simple_rnn</a>()</code> <br><code><a href="layer_spatial_dropout_1d.html">layer_spatial_dropout_1d</a>()</code> <br><code><a href="layer_spatial_dropout_2d.html">layer_spatial_dropout_2d</a>()</code> <br><code><a href="layer_spatial_dropout_3d.html">layer_spatial_dropout_3d</a>()</code> <br><code><a href="layer_spectral_normalization.html">layer_spectral_normalization</a>()</code> <br><code><a href="layer_stft_spectrogram.html">layer_stft_spectrogram</a>()</code> <br><code><a href="layer_string_lookup.html">layer_string_lookup</a>()</code> <br><code><a href="layer_subtract.html">layer_subtract</a>()</code> <br><code><a href="layer_text_vectorization.html">layer_text_vectorization</a>()</code> <br><code><a href="layer_tfsm.html">layer_tfsm</a>()</code> <br><code><a href="layer_time_distributed.html">layer_time_distributed</a>()</code> <br><code><a href="layer_torch_module_wrapper.html">layer_torch_module_wrapper</a>()</code> <br><code><a href="layer_unit_normalization.html">layer_unit_normalization</a>()</code> <br><code><a href="layer_upsampling_1d.html">layer_upsampling_1d</a>()</code> <br><code><a href="layer_upsampling_2d.html">layer_upsampling_2d</a>()</code> <br><code><a href="layer_upsampling_3d.html">layer_upsampling_3d</a>()</code> <br><code><a href="layer_zero_padding_1d.html">layer_zero_padding_1d</a>()</code> <br><code><a href="layer_zero_padding_2d.html">layer_zero_padding_2d</a>()</code> <br><code><a href="layer_zero_padding_3d.html">layer_zero_padding_3d</a>()</code> <br><code><a href="rnn_cell_gru.html">rnn_cell_gru</a>()</code> <br><code><a href="rnn_cell_lstm.html">rnn_cell_lstm</a>()</code> <br><code><a href="rnn_cell_simple.html">rnn_cell_simple</a>()</code> <br><code><a href="rnn_cells_stack.html">rnn_cells_stack</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -150,7 +150,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/layer_spatial_dropout_1d.html b/docs/dev/reference/layer_spatial_dropout_1d.html
index ad3df12c6..50c1d95bf 100644
--- a/docs/dev/reference/layer_spatial_dropout_1d.html
+++ b/docs/dev/reference/layer_spatial_dropout_1d.html
@@ -20,7 +20,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -140,7 +140,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 <ul><li><p><a href="https://keras.io/api/layers/regularization_layers/spatial_dropout1d#spatialdropout1d-class" class="external-link">https://keras.io/api/layers/regularization_layers/spatial_dropout1d#spatialdropout1d-class</a></p></li>
 </ul><p>Other spatial dropout regularization layers: <br><code><a href="layer_spatial_dropout_2d.html">layer_spatial_dropout_2d</a>()</code> <br><code><a href="layer_spatial_dropout_3d.html">layer_spatial_dropout_3d</a>()</code> <br></p>
 <p>Other regularization layers: <br><code><a href="layer_activity_regularization.html">layer_activity_regularization</a>()</code> <br><code><a href="layer_alpha_dropout.html">layer_alpha_dropout</a>()</code> <br><code><a href="layer_dropout.html">layer_dropout</a>()</code> <br><code><a href="layer_gaussian_dropout.html">layer_gaussian_dropout</a>()</code> <br><code><a href="layer_gaussian_noise.html">layer_gaussian_noise</a>()</code> <br><code><a href="layer_spatial_dropout_2d.html">layer_spatial_dropout_2d</a>()</code> <br><code><a href="layer_spatial_dropout_3d.html">layer_spatial_dropout_3d</a>()</code> <br></p>
-<p>Other layers: <br><code><a href="Layer.html">Layer</a>()</code> <br><code><a href="layer_activation.html">layer_activation</a>()</code> <br><code><a href="layer_activation_elu.html">layer_activation_elu</a>()</code> <br><code><a href="layer_activation_leaky_relu.html">layer_activation_leaky_relu</a>()</code> <br><code><a href="layer_activation_parametric_relu.html">layer_activation_parametric_relu</a>()</code> <br><code><a href="layer_activation_relu.html">layer_activation_relu</a>()</code> <br><code><a href="layer_activation_softmax.html">layer_activation_softmax</a>()</code> <br><code><a href="layer_activity_regularization.html">layer_activity_regularization</a>()</code> <br><code><a href="layer_add.html">layer_add</a>()</code> <br><code><a href="layer_additive_attention.html">layer_additive_attention</a>()</code> <br><code><a href="layer_alpha_dropout.html">layer_alpha_dropout</a>()</code> <br><code><a href="layer_attention.html">layer_attention</a>()</code> <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_average.html">layer_average</a>()</code> <br><code><a href="layer_average_pooling_1d.html">layer_average_pooling_1d</a>()</code> <br><code><a href="layer_average_pooling_2d.html">layer_average_pooling_2d</a>()</code> <br><code><a href="layer_average_pooling_3d.html">layer_average_pooling_3d</a>()</code> <br><code><a href="layer_batch_normalization.html">layer_batch_normalization</a>()</code> <br><code><a href="layer_bidirectional.html">layer_bidirectional</a>()</code> <br><code><a href="layer_category_encoding.html">layer_category_encoding</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_concatenate.html">layer_concatenate</a>()</code> <br><code><a href="layer_conv_1d.html">layer_conv_1d</a>()</code> <br><code><a href="layer_conv_1d_transpose.html">layer_conv_1d_transpose</a>()</code> <br><code><a href="layer_conv_2d.html">layer_conv_2d</a>()</code> <br><code><a href="layer_conv_2d_transpose.html">layer_conv_2d_transpose</a>()</code> <br><code><a href="layer_conv_3d.html">layer_conv_3d</a>()</code> <br><code><a href="layer_conv_3d_transpose.html">layer_conv_3d_transpose</a>()</code> <br><code><a href="layer_conv_lstm_1d.html">layer_conv_lstm_1d</a>()</code> <br><code><a href="layer_conv_lstm_2d.html">layer_conv_lstm_2d</a>()</code> <br><code><a href="layer_conv_lstm_3d.html">layer_conv_lstm_3d</a>()</code> <br><code><a href="layer_cropping_1d.html">layer_cropping_1d</a>()</code> <br><code><a href="layer_cropping_2d.html">layer_cropping_2d</a>()</code> <br><code><a href="layer_cropping_3d.html">layer_cropping_3d</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_dense.html">layer_dense</a>()</code> <br><code><a href="layer_depthwise_conv_1d.html">layer_depthwise_conv_1d</a>()</code> <br><code><a href="layer_depthwise_conv_2d.html">layer_depthwise_conv_2d</a>()</code> <br><code><a href="layer_discretization.html">layer_discretization</a>()</code> <br><code><a href="layer_dot.html">layer_dot</a>()</code> <br><code><a href="layer_dropout.html">layer_dropout</a>()</code> <br><code><a href="layer_einsum_dense.html">layer_einsum_dense</a>()</code> <br><code><a href="layer_embedding.html">layer_embedding</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_feature_space.html">layer_feature_space</a>()</code> <br><code><a href="layer_flatten.html">layer_flatten</a>()</code> <br><code><a href="layer_flax_module_wrapper.html">layer_flax_module_wrapper</a>()</code> <br><code><a href="layer_gaussian_dropout.html">layer_gaussian_dropout</a>()</code> <br><code><a href="layer_gaussian_noise.html">layer_gaussian_noise</a>()</code> <br><code><a href="layer_global_average_pooling_1d.html">layer_global_average_pooling_1d</a>()</code> <br><code><a href="layer_global_average_pooling_2d.html">layer_global_average_pooling_2d</a>()</code> <br><code><a href="layer_global_average_pooling_3d.html">layer_global_average_pooling_3d</a>()</code> <br><code><a href="layer_global_max_pooling_1d.html">layer_global_max_pooling_1d</a>()</code> <br><code><a href="layer_global_max_pooling_2d.html">layer_global_max_pooling_2d</a>()</code> <br><code><a href="layer_global_max_pooling_3d.html">layer_global_max_pooling_3d</a>()</code> <br><code><a href="layer_group_normalization.html">layer_group_normalization</a>()</code> <br><code><a href="layer_group_query_attention.html">layer_group_query_attention</a>()</code> <br><code><a href="layer_gru.html">layer_gru</a>()</code> <br><code><a href="layer_hashed_crossing.html">layer_hashed_crossing</a>()</code> <br><code><a href="layer_hashing.html">layer_hashing</a>()</code> <br><code><a href="layer_identity.html">layer_identity</a>()</code> <br><code><a href="layer_integer_lookup.html">layer_integer_lookup</a>()</code> <br><code><a href="layer_jax_model_wrapper.html">layer_jax_model_wrapper</a>()</code> <br><code><a href="layer_lambda.html">layer_lambda</a>()</code> <br><code><a href="layer_layer_normalization.html">layer_layer_normalization</a>()</code> <br><code><a href="layer_lstm.html">layer_lstm</a>()</code> <br><code><a href="layer_masking.html">layer_masking</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_max_pooling_1d.html">layer_max_pooling_1d</a>()</code> <br><code><a href="layer_max_pooling_2d.html">layer_max_pooling_2d</a>()</code> <br><code><a href="layer_max_pooling_3d.html">layer_max_pooling_3d</a>()</code> <br><code><a href="layer_maximum.html">layer_maximum</a>()</code> <br><code><a href="layer_mel_spectrogram.html">layer_mel_spectrogram</a>()</code> <br><code><a href="layer_minimum.html">layer_minimum</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_multi_head_attention.html">layer_multi_head_attention</a>()</code> <br><code><a href="layer_multiply.html">layer_multiply</a>()</code> <br><code><a href="layer_normalization.html">layer_normalization</a>()</code> <br><code><a href="layer_permute.html">layer_permute</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_brightness.html">layer_random_brightness</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_contrast.html">layer_random_contrast</a>()</code> <br><code><a href="layer_random_crop.html">layer_random_crop</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_flip.html">layer_random_flip</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_rotation.html">layer_random_rotation</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_random_translation.html">layer_random_translation</a>()</code> <br><code><a href="layer_random_zoom.html">layer_random_zoom</a>()</code> <br><code><a href="layer_repeat_vector.html">layer_repeat_vector</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_reshape.html">layer_reshape</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_rms_normalization.html">layer_rms_normalization</a>()</code> <br><code><a href="layer_rnn.html">layer_rnn</a>()</code> <br><code><a href="layer_separable_conv_1d.html">layer_separable_conv_1d</a>()</code> <br><code><a href="layer_separable_conv_2d.html">layer_separable_conv_2d</a>()</code> <br><code><a href="layer_simple_rnn.html">layer_simple_rnn</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br><code><a href="layer_spatial_dropout_2d.html">layer_spatial_dropout_2d</a>()</code> <br><code><a href="layer_spatial_dropout_3d.html">layer_spatial_dropout_3d</a>()</code> <br><code><a href="layer_spectral_normalization.html">layer_spectral_normalization</a>()</code> <br><code><a href="layer_stft_spectrogram.html">layer_stft_spectrogram</a>()</code> <br><code><a href="layer_string_lookup.html">layer_string_lookup</a>()</code> <br><code><a href="layer_subtract.html">layer_subtract</a>()</code> <br><code><a href="layer_text_vectorization.html">layer_text_vectorization</a>()</code> <br><code><a href="layer_tfsm.html">layer_tfsm</a>()</code> <br><code><a href="layer_time_distributed.html">layer_time_distributed</a>()</code> <br><code><a href="layer_torch_module_wrapper.html">layer_torch_module_wrapper</a>()</code> <br><code><a href="layer_unit_normalization.html">layer_unit_normalization</a>()</code> <br><code><a href="layer_upsampling_1d.html">layer_upsampling_1d</a>()</code> <br><code><a href="layer_upsampling_2d.html">layer_upsampling_2d</a>()</code> <br><code><a href="layer_upsampling_3d.html">layer_upsampling_3d</a>()</code> <br><code><a href="layer_zero_padding_1d.html">layer_zero_padding_1d</a>()</code> <br><code><a href="layer_zero_padding_2d.html">layer_zero_padding_2d</a>()</code> <br><code><a href="layer_zero_padding_3d.html">layer_zero_padding_3d</a>()</code> <br><code><a href="rnn_cell_gru.html">rnn_cell_gru</a>()</code> <br><code><a href="rnn_cell_lstm.html">rnn_cell_lstm</a>()</code> <br><code><a href="rnn_cell_simple.html">rnn_cell_simple</a>()</code> <br><code><a href="rnn_cells_stack.html">rnn_cells_stack</a>()</code> <br></p></div>
+<p>Other layers: <br><code><a href="Layer.html">Layer</a>()</code> <br><code><a href="layer_activation.html">layer_activation</a>()</code> <br><code><a href="layer_activation_elu.html">layer_activation_elu</a>()</code> <br><code><a href="layer_activation_leaky_relu.html">layer_activation_leaky_relu</a>()</code> <br><code><a href="layer_activation_parametric_relu.html">layer_activation_parametric_relu</a>()</code> <br><code><a href="layer_activation_relu.html">layer_activation_relu</a>()</code> <br><code><a href="layer_activation_softmax.html">layer_activation_softmax</a>()</code> <br><code><a href="layer_activity_regularization.html">layer_activity_regularization</a>()</code> <br><code><a href="layer_add.html">layer_add</a>()</code> <br><code><a href="layer_additive_attention.html">layer_additive_attention</a>()</code> <br><code><a href="layer_alpha_dropout.html">layer_alpha_dropout</a>()</code> <br><code><a href="layer_attention.html">layer_attention</a>()</code> <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_average.html">layer_average</a>()</code> <br><code><a href="layer_average_pooling_1d.html">layer_average_pooling_1d</a>()</code> <br><code><a href="layer_average_pooling_2d.html">layer_average_pooling_2d</a>()</code> <br><code><a href="layer_average_pooling_3d.html">layer_average_pooling_3d</a>()</code> <br><code><a href="layer_batch_normalization.html">layer_batch_normalization</a>()</code> <br><code><a href="layer_bidirectional.html">layer_bidirectional</a>()</code> <br><code><a href="layer_category_encoding.html">layer_category_encoding</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_concatenate.html">layer_concatenate</a>()</code> <br><code><a href="layer_conv_1d.html">layer_conv_1d</a>()</code> <br><code><a href="layer_conv_1d_transpose.html">layer_conv_1d_transpose</a>()</code> <br><code><a href="layer_conv_2d.html">layer_conv_2d</a>()</code> <br><code><a href="layer_conv_2d_transpose.html">layer_conv_2d_transpose</a>()</code> <br><code><a href="layer_conv_3d.html">layer_conv_3d</a>()</code> <br><code><a href="layer_conv_3d_transpose.html">layer_conv_3d_transpose</a>()</code> <br><code><a href="layer_conv_lstm_1d.html">layer_conv_lstm_1d</a>()</code> <br><code><a href="layer_conv_lstm_2d.html">layer_conv_lstm_2d</a>()</code> <br><code><a href="layer_conv_lstm_3d.html">layer_conv_lstm_3d</a>()</code> <br><code><a href="layer_cropping_1d.html">layer_cropping_1d</a>()</code> <br><code><a href="layer_cropping_2d.html">layer_cropping_2d</a>()</code> <br><code><a href="layer_cropping_3d.html">layer_cropping_3d</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_dense.html">layer_dense</a>()</code> <br><code><a href="layer_depthwise_conv_1d.html">layer_depthwise_conv_1d</a>()</code> <br><code><a href="layer_depthwise_conv_2d.html">layer_depthwise_conv_2d</a>()</code> <br><code><a href="layer_discretization.html">layer_discretization</a>()</code> <br><code><a href="layer_dot.html">layer_dot</a>()</code> <br><code><a href="layer_dropout.html">layer_dropout</a>()</code> <br><code><a href="layer_einsum_dense.html">layer_einsum_dense</a>()</code> <br><code><a href="layer_embedding.html">layer_embedding</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_feature_space.html">layer_feature_space</a>()</code> <br><code><a href="layer_flatten.html">layer_flatten</a>()</code> <br><code><a href="layer_flax_module_wrapper.html">layer_flax_module_wrapper</a>()</code> <br><code><a href="layer_gaussian_dropout.html">layer_gaussian_dropout</a>()</code> <br><code><a href="layer_gaussian_noise.html">layer_gaussian_noise</a>()</code> <br><code><a href="layer_global_average_pooling_1d.html">layer_global_average_pooling_1d</a>()</code> <br><code><a href="layer_global_average_pooling_2d.html">layer_global_average_pooling_2d</a>()</code> <br><code><a href="layer_global_average_pooling_3d.html">layer_global_average_pooling_3d</a>()</code> <br><code><a href="layer_global_max_pooling_1d.html">layer_global_max_pooling_1d</a>()</code> <br><code><a href="layer_global_max_pooling_2d.html">layer_global_max_pooling_2d</a>()</code> <br><code><a href="layer_global_max_pooling_3d.html">layer_global_max_pooling_3d</a>()</code> <br><code><a href="layer_group_normalization.html">layer_group_normalization</a>()</code> <br><code><a href="layer_group_query_attention.html">layer_group_query_attention</a>()</code> <br><code><a href="layer_gru.html">layer_gru</a>()</code> <br><code><a href="layer_hashed_crossing.html">layer_hashed_crossing</a>()</code> <br><code><a href="layer_hashing.html">layer_hashing</a>()</code> <br><code><a href="layer_identity.html">layer_identity</a>()</code> <br><code><a href="layer_integer_lookup.html">layer_integer_lookup</a>()</code> <br><code><a href="layer_jax_model_wrapper.html">layer_jax_model_wrapper</a>()</code> <br><code><a href="layer_lambda.html">layer_lambda</a>()</code> <br><code><a href="layer_layer_normalization.html">layer_layer_normalization</a>()</code> <br><code><a href="layer_lstm.html">layer_lstm</a>()</code> <br><code><a href="layer_masking.html">layer_masking</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_max_pooling_1d.html">layer_max_pooling_1d</a>()</code> <br><code><a href="layer_max_pooling_2d.html">layer_max_pooling_2d</a>()</code> <br><code><a href="layer_max_pooling_3d.html">layer_max_pooling_3d</a>()</code> <br><code><a href="layer_maximum.html">layer_maximum</a>()</code> <br><code><a href="layer_mel_spectrogram.html">layer_mel_spectrogram</a>()</code> <br><code><a href="layer_minimum.html">layer_minimum</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_multi_head_attention.html">layer_multi_head_attention</a>()</code> <br><code><a href="layer_multiply.html">layer_multiply</a>()</code> <br><code><a href="layer_normalization.html">layer_normalization</a>()</code> <br><code><a href="layer_permute.html">layer_permute</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_brightness.html">layer_random_brightness</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_contrast.html">layer_random_contrast</a>()</code> <br><code><a href="layer_random_crop.html">layer_random_crop</a>()</code> <br><code><a href="layer_random_elastic_transform.html">layer_random_elastic_transform</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_flip.html">layer_random_flip</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_rotation.html">layer_random_rotation</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_random_translation.html">layer_random_translation</a>()</code> <br><code><a href="layer_random_zoom.html">layer_random_zoom</a>()</code> <br><code><a href="layer_repeat_vector.html">layer_repeat_vector</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_reshape.html">layer_reshape</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_rms_normalization.html">layer_rms_normalization</a>()</code> <br><code><a href="layer_rnn.html">layer_rnn</a>()</code> <br><code><a href="layer_separable_conv_1d.html">layer_separable_conv_1d</a>()</code> <br><code><a href="layer_separable_conv_2d.html">layer_separable_conv_2d</a>()</code> <br><code><a href="layer_simple_rnn.html">layer_simple_rnn</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br><code><a href="layer_spatial_dropout_2d.html">layer_spatial_dropout_2d</a>()</code> <br><code><a href="layer_spatial_dropout_3d.html">layer_spatial_dropout_3d</a>()</code> <br><code><a href="layer_spectral_normalization.html">layer_spectral_normalization</a>()</code> <br><code><a href="layer_stft_spectrogram.html">layer_stft_spectrogram</a>()</code> <br><code><a href="layer_string_lookup.html">layer_string_lookup</a>()</code> <br><code><a href="layer_subtract.html">layer_subtract</a>()</code> <br><code><a href="layer_text_vectorization.html">layer_text_vectorization</a>()</code> <br><code><a href="layer_tfsm.html">layer_tfsm</a>()</code> <br><code><a href="layer_time_distributed.html">layer_time_distributed</a>()</code> <br><code><a href="layer_torch_module_wrapper.html">layer_torch_module_wrapper</a>()</code> <br><code><a href="layer_unit_normalization.html">layer_unit_normalization</a>()</code> <br><code><a href="layer_upsampling_1d.html">layer_upsampling_1d</a>()</code> <br><code><a href="layer_upsampling_2d.html">layer_upsampling_2d</a>()</code> <br><code><a href="layer_upsampling_3d.html">layer_upsampling_3d</a>()</code> <br><code><a href="layer_zero_padding_1d.html">layer_zero_padding_1d</a>()</code> <br><code><a href="layer_zero_padding_2d.html">layer_zero_padding_2d</a>()</code> <br><code><a href="layer_zero_padding_3d.html">layer_zero_padding_3d</a>()</code> <br><code><a href="rnn_cell_gru.html">rnn_cell_gru</a>()</code> <br><code><a href="rnn_cell_lstm.html">rnn_cell_lstm</a>()</code> <br><code><a href="rnn_cell_simple.html">rnn_cell_simple</a>()</code> <br><code><a href="rnn_cells_stack.html">rnn_cells_stack</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -152,7 +152,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/layer_spatial_dropout_2d.html b/docs/dev/reference/layer_spatial_dropout_2d.html
index 1a649d131..6d1c16c45 100644
--- a/docs/dev/reference/layer_spatial_dropout_2d.html
+++ b/docs/dev/reference/layer_spatial_dropout_2d.html
@@ -20,7 +20,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -159,7 +159,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 <ul><li><p><a href="https://keras.io/api/layers/regularization_layers/spatial_dropout2d#spatialdropout2d-class" class="external-link">https://keras.io/api/layers/regularization_layers/spatial_dropout2d#spatialdropout2d-class</a></p></li>
 </ul><p>Other spatial dropout regularization layers: <br><code><a href="layer_spatial_dropout_1d.html">layer_spatial_dropout_1d</a>()</code> <br><code><a href="layer_spatial_dropout_3d.html">layer_spatial_dropout_3d</a>()</code> <br></p>
 <p>Other regularization layers: <br><code><a href="layer_activity_regularization.html">layer_activity_regularization</a>()</code> <br><code><a href="layer_alpha_dropout.html">layer_alpha_dropout</a>()</code> <br><code><a href="layer_dropout.html">layer_dropout</a>()</code> <br><code><a href="layer_gaussian_dropout.html">layer_gaussian_dropout</a>()</code> <br><code><a href="layer_gaussian_noise.html">layer_gaussian_noise</a>()</code> <br><code><a href="layer_spatial_dropout_1d.html">layer_spatial_dropout_1d</a>()</code> <br><code><a href="layer_spatial_dropout_3d.html">layer_spatial_dropout_3d</a>()</code> <br></p>
-<p>Other layers: <br><code><a href="Layer.html">Layer</a>()</code> <br><code><a href="layer_activation.html">layer_activation</a>()</code> <br><code><a href="layer_activation_elu.html">layer_activation_elu</a>()</code> <br><code><a href="layer_activation_leaky_relu.html">layer_activation_leaky_relu</a>()</code> <br><code><a href="layer_activation_parametric_relu.html">layer_activation_parametric_relu</a>()</code> <br><code><a href="layer_activation_relu.html">layer_activation_relu</a>()</code> <br><code><a href="layer_activation_softmax.html">layer_activation_softmax</a>()</code> <br><code><a href="layer_activity_regularization.html">layer_activity_regularization</a>()</code> <br><code><a href="layer_add.html">layer_add</a>()</code> <br><code><a href="layer_additive_attention.html">layer_additive_attention</a>()</code> <br><code><a href="layer_alpha_dropout.html">layer_alpha_dropout</a>()</code> <br><code><a href="layer_attention.html">layer_attention</a>()</code> <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_average.html">layer_average</a>()</code> <br><code><a href="layer_average_pooling_1d.html">layer_average_pooling_1d</a>()</code> <br><code><a href="layer_average_pooling_2d.html">layer_average_pooling_2d</a>()</code> <br><code><a href="layer_average_pooling_3d.html">layer_average_pooling_3d</a>()</code> <br><code><a href="layer_batch_normalization.html">layer_batch_normalization</a>()</code> <br><code><a href="layer_bidirectional.html">layer_bidirectional</a>()</code> <br><code><a href="layer_category_encoding.html">layer_category_encoding</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_concatenate.html">layer_concatenate</a>()</code> <br><code><a href="layer_conv_1d.html">layer_conv_1d</a>()</code> <br><code><a href="layer_conv_1d_transpose.html">layer_conv_1d_transpose</a>()</code> <br><code><a href="layer_conv_2d.html">layer_conv_2d</a>()</code> <br><code><a href="layer_conv_2d_transpose.html">layer_conv_2d_transpose</a>()</code> <br><code><a href="layer_conv_3d.html">layer_conv_3d</a>()</code> <br><code><a href="layer_conv_3d_transpose.html">layer_conv_3d_transpose</a>()</code> <br><code><a href="layer_conv_lstm_1d.html">layer_conv_lstm_1d</a>()</code> <br><code><a href="layer_conv_lstm_2d.html">layer_conv_lstm_2d</a>()</code> <br><code><a href="layer_conv_lstm_3d.html">layer_conv_lstm_3d</a>()</code> <br><code><a href="layer_cropping_1d.html">layer_cropping_1d</a>()</code> <br><code><a href="layer_cropping_2d.html">layer_cropping_2d</a>()</code> <br><code><a href="layer_cropping_3d.html">layer_cropping_3d</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_dense.html">layer_dense</a>()</code> <br><code><a href="layer_depthwise_conv_1d.html">layer_depthwise_conv_1d</a>()</code> <br><code><a href="layer_depthwise_conv_2d.html">layer_depthwise_conv_2d</a>()</code> <br><code><a href="layer_discretization.html">layer_discretization</a>()</code> <br><code><a href="layer_dot.html">layer_dot</a>()</code> <br><code><a href="layer_dropout.html">layer_dropout</a>()</code> <br><code><a href="layer_einsum_dense.html">layer_einsum_dense</a>()</code> <br><code><a href="layer_embedding.html">layer_embedding</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_feature_space.html">layer_feature_space</a>()</code> <br><code><a href="layer_flatten.html">layer_flatten</a>()</code> <br><code><a href="layer_flax_module_wrapper.html">layer_flax_module_wrapper</a>()</code> <br><code><a href="layer_gaussian_dropout.html">layer_gaussian_dropout</a>()</code> <br><code><a href="layer_gaussian_noise.html">layer_gaussian_noise</a>()</code> <br><code><a href="layer_global_average_pooling_1d.html">layer_global_average_pooling_1d</a>()</code> <br><code><a href="layer_global_average_pooling_2d.html">layer_global_average_pooling_2d</a>()</code> <br><code><a href="layer_global_average_pooling_3d.html">layer_global_average_pooling_3d</a>()</code> <br><code><a href="layer_global_max_pooling_1d.html">layer_global_max_pooling_1d</a>()</code> <br><code><a href="layer_global_max_pooling_2d.html">layer_global_max_pooling_2d</a>()</code> <br><code><a href="layer_global_max_pooling_3d.html">layer_global_max_pooling_3d</a>()</code> <br><code><a href="layer_group_normalization.html">layer_group_normalization</a>()</code> <br><code><a href="layer_group_query_attention.html">layer_group_query_attention</a>()</code> <br><code><a href="layer_gru.html">layer_gru</a>()</code> <br><code><a href="layer_hashed_crossing.html">layer_hashed_crossing</a>()</code> <br><code><a href="layer_hashing.html">layer_hashing</a>()</code> <br><code><a href="layer_identity.html">layer_identity</a>()</code> <br><code><a href="layer_integer_lookup.html">layer_integer_lookup</a>()</code> <br><code><a href="layer_jax_model_wrapper.html">layer_jax_model_wrapper</a>()</code> <br><code><a href="layer_lambda.html">layer_lambda</a>()</code> <br><code><a href="layer_layer_normalization.html">layer_layer_normalization</a>()</code> <br><code><a href="layer_lstm.html">layer_lstm</a>()</code> <br><code><a href="layer_masking.html">layer_masking</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_max_pooling_1d.html">layer_max_pooling_1d</a>()</code> <br><code><a href="layer_max_pooling_2d.html">layer_max_pooling_2d</a>()</code> <br><code><a href="layer_max_pooling_3d.html">layer_max_pooling_3d</a>()</code> <br><code><a href="layer_maximum.html">layer_maximum</a>()</code> <br><code><a href="layer_mel_spectrogram.html">layer_mel_spectrogram</a>()</code> <br><code><a href="layer_minimum.html">layer_minimum</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_multi_head_attention.html">layer_multi_head_attention</a>()</code> <br><code><a href="layer_multiply.html">layer_multiply</a>()</code> <br><code><a href="layer_normalization.html">layer_normalization</a>()</code> <br><code><a href="layer_permute.html">layer_permute</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_brightness.html">layer_random_brightness</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_contrast.html">layer_random_contrast</a>()</code> <br><code><a href="layer_random_crop.html">layer_random_crop</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_flip.html">layer_random_flip</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_rotation.html">layer_random_rotation</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_random_translation.html">layer_random_translation</a>()</code> <br><code><a href="layer_random_zoom.html">layer_random_zoom</a>()</code> <br><code><a href="layer_repeat_vector.html">layer_repeat_vector</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_reshape.html">layer_reshape</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_rms_normalization.html">layer_rms_normalization</a>()</code> <br><code><a href="layer_rnn.html">layer_rnn</a>()</code> <br><code><a href="layer_separable_conv_1d.html">layer_separable_conv_1d</a>()</code> <br><code><a href="layer_separable_conv_2d.html">layer_separable_conv_2d</a>()</code> <br><code><a href="layer_simple_rnn.html">layer_simple_rnn</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br><code><a href="layer_spatial_dropout_1d.html">layer_spatial_dropout_1d</a>()</code> <br><code><a href="layer_spatial_dropout_3d.html">layer_spatial_dropout_3d</a>()</code> <br><code><a href="layer_spectral_normalization.html">layer_spectral_normalization</a>()</code> <br><code><a href="layer_stft_spectrogram.html">layer_stft_spectrogram</a>()</code> <br><code><a href="layer_string_lookup.html">layer_string_lookup</a>()</code> <br><code><a href="layer_subtract.html">layer_subtract</a>()</code> <br><code><a href="layer_text_vectorization.html">layer_text_vectorization</a>()</code> <br><code><a href="layer_tfsm.html">layer_tfsm</a>()</code> <br><code><a href="layer_time_distributed.html">layer_time_distributed</a>()</code> <br><code><a href="layer_torch_module_wrapper.html">layer_torch_module_wrapper</a>()</code> <br><code><a href="layer_unit_normalization.html">layer_unit_normalization</a>()</code> <br><code><a href="layer_upsampling_1d.html">layer_upsampling_1d</a>()</code> <br><code><a href="layer_upsampling_2d.html">layer_upsampling_2d</a>()</code> <br><code><a href="layer_upsampling_3d.html">layer_upsampling_3d</a>()</code> <br><code><a href="layer_zero_padding_1d.html">layer_zero_padding_1d</a>()</code> <br><code><a href="layer_zero_padding_2d.html">layer_zero_padding_2d</a>()</code> <br><code><a href="layer_zero_padding_3d.html">layer_zero_padding_3d</a>()</code> <br><code><a href="rnn_cell_gru.html">rnn_cell_gru</a>()</code> <br><code><a href="rnn_cell_lstm.html">rnn_cell_lstm</a>()</code> <br><code><a href="rnn_cell_simple.html">rnn_cell_simple</a>()</code> <br><code><a href="rnn_cells_stack.html">rnn_cells_stack</a>()</code> <br></p></div>
+<p>Other layers: <br><code><a href="Layer.html">Layer</a>()</code> <br><code><a href="layer_activation.html">layer_activation</a>()</code> <br><code><a href="layer_activation_elu.html">layer_activation_elu</a>()</code> <br><code><a href="layer_activation_leaky_relu.html">layer_activation_leaky_relu</a>()</code> <br><code><a href="layer_activation_parametric_relu.html">layer_activation_parametric_relu</a>()</code> <br><code><a href="layer_activation_relu.html">layer_activation_relu</a>()</code> <br><code><a href="layer_activation_softmax.html">layer_activation_softmax</a>()</code> <br><code><a href="layer_activity_regularization.html">layer_activity_regularization</a>()</code> <br><code><a href="layer_add.html">layer_add</a>()</code> <br><code><a href="layer_additive_attention.html">layer_additive_attention</a>()</code> <br><code><a href="layer_alpha_dropout.html">layer_alpha_dropout</a>()</code> <br><code><a href="layer_attention.html">layer_attention</a>()</code> <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_average.html">layer_average</a>()</code> <br><code><a href="layer_average_pooling_1d.html">layer_average_pooling_1d</a>()</code> <br><code><a href="layer_average_pooling_2d.html">layer_average_pooling_2d</a>()</code> <br><code><a href="layer_average_pooling_3d.html">layer_average_pooling_3d</a>()</code> <br><code><a href="layer_batch_normalization.html">layer_batch_normalization</a>()</code> <br><code><a href="layer_bidirectional.html">layer_bidirectional</a>()</code> <br><code><a href="layer_category_encoding.html">layer_category_encoding</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_concatenate.html">layer_concatenate</a>()</code> <br><code><a href="layer_conv_1d.html">layer_conv_1d</a>()</code> <br><code><a href="layer_conv_1d_transpose.html">layer_conv_1d_transpose</a>()</code> <br><code><a href="layer_conv_2d.html">layer_conv_2d</a>()</code> <br><code><a href="layer_conv_2d_transpose.html">layer_conv_2d_transpose</a>()</code> <br><code><a href="layer_conv_3d.html">layer_conv_3d</a>()</code> <br><code><a href="layer_conv_3d_transpose.html">layer_conv_3d_transpose</a>()</code> <br><code><a href="layer_conv_lstm_1d.html">layer_conv_lstm_1d</a>()</code> <br><code><a href="layer_conv_lstm_2d.html">layer_conv_lstm_2d</a>()</code> <br><code><a href="layer_conv_lstm_3d.html">layer_conv_lstm_3d</a>()</code> <br><code><a href="layer_cropping_1d.html">layer_cropping_1d</a>()</code> <br><code><a href="layer_cropping_2d.html">layer_cropping_2d</a>()</code> <br><code><a href="layer_cropping_3d.html">layer_cropping_3d</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_dense.html">layer_dense</a>()</code> <br><code><a href="layer_depthwise_conv_1d.html">layer_depthwise_conv_1d</a>()</code> <br><code><a href="layer_depthwise_conv_2d.html">layer_depthwise_conv_2d</a>()</code> <br><code><a href="layer_discretization.html">layer_discretization</a>()</code> <br><code><a href="layer_dot.html">layer_dot</a>()</code> <br><code><a href="layer_dropout.html">layer_dropout</a>()</code> <br><code><a href="layer_einsum_dense.html">layer_einsum_dense</a>()</code> <br><code><a href="layer_embedding.html">layer_embedding</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_feature_space.html">layer_feature_space</a>()</code> <br><code><a href="layer_flatten.html">layer_flatten</a>()</code> <br><code><a href="layer_flax_module_wrapper.html">layer_flax_module_wrapper</a>()</code> <br><code><a href="layer_gaussian_dropout.html">layer_gaussian_dropout</a>()</code> <br><code><a href="layer_gaussian_noise.html">layer_gaussian_noise</a>()</code> <br><code><a href="layer_global_average_pooling_1d.html">layer_global_average_pooling_1d</a>()</code> <br><code><a href="layer_global_average_pooling_2d.html">layer_global_average_pooling_2d</a>()</code> <br><code><a href="layer_global_average_pooling_3d.html">layer_global_average_pooling_3d</a>()</code> <br><code><a href="layer_global_max_pooling_1d.html">layer_global_max_pooling_1d</a>()</code> <br><code><a href="layer_global_max_pooling_2d.html">layer_global_max_pooling_2d</a>()</code> <br><code><a href="layer_global_max_pooling_3d.html">layer_global_max_pooling_3d</a>()</code> <br><code><a href="layer_group_normalization.html">layer_group_normalization</a>()</code> <br><code><a href="layer_group_query_attention.html">layer_group_query_attention</a>()</code> <br><code><a href="layer_gru.html">layer_gru</a>()</code> <br><code><a href="layer_hashed_crossing.html">layer_hashed_crossing</a>()</code> <br><code><a href="layer_hashing.html">layer_hashing</a>()</code> <br><code><a href="layer_identity.html">layer_identity</a>()</code> <br><code><a href="layer_integer_lookup.html">layer_integer_lookup</a>()</code> <br><code><a href="layer_jax_model_wrapper.html">layer_jax_model_wrapper</a>()</code> <br><code><a href="layer_lambda.html">layer_lambda</a>()</code> <br><code><a href="layer_layer_normalization.html">layer_layer_normalization</a>()</code> <br><code><a href="layer_lstm.html">layer_lstm</a>()</code> <br><code><a href="layer_masking.html">layer_masking</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_max_pooling_1d.html">layer_max_pooling_1d</a>()</code> <br><code><a href="layer_max_pooling_2d.html">layer_max_pooling_2d</a>()</code> <br><code><a href="layer_max_pooling_3d.html">layer_max_pooling_3d</a>()</code> <br><code><a href="layer_maximum.html">layer_maximum</a>()</code> <br><code><a href="layer_mel_spectrogram.html">layer_mel_spectrogram</a>()</code> <br><code><a href="layer_minimum.html">layer_minimum</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_multi_head_attention.html">layer_multi_head_attention</a>()</code> <br><code><a href="layer_multiply.html">layer_multiply</a>()</code> <br><code><a href="layer_normalization.html">layer_normalization</a>()</code> <br><code><a href="layer_permute.html">layer_permute</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_brightness.html">layer_random_brightness</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_contrast.html">layer_random_contrast</a>()</code> <br><code><a href="layer_random_crop.html">layer_random_crop</a>()</code> <br><code><a href="layer_random_elastic_transform.html">layer_random_elastic_transform</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_flip.html">layer_random_flip</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_rotation.html">layer_random_rotation</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_random_translation.html">layer_random_translation</a>()</code> <br><code><a href="layer_random_zoom.html">layer_random_zoom</a>()</code> <br><code><a href="layer_repeat_vector.html">layer_repeat_vector</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_reshape.html">layer_reshape</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_rms_normalization.html">layer_rms_normalization</a>()</code> <br><code><a href="layer_rnn.html">layer_rnn</a>()</code> <br><code><a href="layer_separable_conv_1d.html">layer_separable_conv_1d</a>()</code> <br><code><a href="layer_separable_conv_2d.html">layer_separable_conv_2d</a>()</code> <br><code><a href="layer_simple_rnn.html">layer_simple_rnn</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br><code><a href="layer_spatial_dropout_1d.html">layer_spatial_dropout_1d</a>()</code> <br><code><a href="layer_spatial_dropout_3d.html">layer_spatial_dropout_3d</a>()</code> <br><code><a href="layer_spectral_normalization.html">layer_spectral_normalization</a>()</code> <br><code><a href="layer_stft_spectrogram.html">layer_stft_spectrogram</a>()</code> <br><code><a href="layer_string_lookup.html">layer_string_lookup</a>()</code> <br><code><a href="layer_subtract.html">layer_subtract</a>()</code> <br><code><a href="layer_text_vectorization.html">layer_text_vectorization</a>()</code> <br><code><a href="layer_tfsm.html">layer_tfsm</a>()</code> <br><code><a href="layer_time_distributed.html">layer_time_distributed</a>()</code> <br><code><a href="layer_torch_module_wrapper.html">layer_torch_module_wrapper</a>()</code> <br><code><a href="layer_unit_normalization.html">layer_unit_normalization</a>()</code> <br><code><a href="layer_upsampling_1d.html">layer_upsampling_1d</a>()</code> <br><code><a href="layer_upsampling_2d.html">layer_upsampling_2d</a>()</code> <br><code><a href="layer_upsampling_3d.html">layer_upsampling_3d</a>()</code> <br><code><a href="layer_zero_padding_1d.html">layer_zero_padding_1d</a>()</code> <br><code><a href="layer_zero_padding_2d.html">layer_zero_padding_2d</a>()</code> <br><code><a href="layer_zero_padding_3d.html">layer_zero_padding_3d</a>()</code> <br><code><a href="rnn_cell_gru.html">rnn_cell_gru</a>()</code> <br><code><a href="rnn_cell_lstm.html">rnn_cell_lstm</a>()</code> <br><code><a href="rnn_cell_simple.html">rnn_cell_simple</a>()</code> <br><code><a href="rnn_cells_stack.html">rnn_cells_stack</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -171,7 +171,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/layer_spatial_dropout_3d.html b/docs/dev/reference/layer_spatial_dropout_3d.html
index f9a5078ca..53a4a066e 100644
--- a/docs/dev/reference/layer_spatial_dropout_3d.html
+++ b/docs/dev/reference/layer_spatial_dropout_3d.html
@@ -20,7 +20,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -159,7 +159,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 <ul><li><p><a href="https://keras.io/api/layers/regularization_layers/spatial_dropout3d#spatialdropout3d-class" class="external-link">https://keras.io/api/layers/regularization_layers/spatial_dropout3d#spatialdropout3d-class</a></p></li>
 </ul><p>Other spatial dropout regularization layers: <br><code><a href="layer_spatial_dropout_1d.html">layer_spatial_dropout_1d</a>()</code> <br><code><a href="layer_spatial_dropout_2d.html">layer_spatial_dropout_2d</a>()</code> <br></p>
 <p>Other regularization layers: <br><code><a href="layer_activity_regularization.html">layer_activity_regularization</a>()</code> <br><code><a href="layer_alpha_dropout.html">layer_alpha_dropout</a>()</code> <br><code><a href="layer_dropout.html">layer_dropout</a>()</code> <br><code><a href="layer_gaussian_dropout.html">layer_gaussian_dropout</a>()</code> <br><code><a href="layer_gaussian_noise.html">layer_gaussian_noise</a>()</code> <br><code><a href="layer_spatial_dropout_1d.html">layer_spatial_dropout_1d</a>()</code> <br><code><a href="layer_spatial_dropout_2d.html">layer_spatial_dropout_2d</a>()</code> <br></p>
-<p>Other layers: <br><code><a href="Layer.html">Layer</a>()</code> <br><code><a href="layer_activation.html">layer_activation</a>()</code> <br><code><a href="layer_activation_elu.html">layer_activation_elu</a>()</code> <br><code><a href="layer_activation_leaky_relu.html">layer_activation_leaky_relu</a>()</code> <br><code><a href="layer_activation_parametric_relu.html">layer_activation_parametric_relu</a>()</code> <br><code><a href="layer_activation_relu.html">layer_activation_relu</a>()</code> <br><code><a href="layer_activation_softmax.html">layer_activation_softmax</a>()</code> <br><code><a href="layer_activity_regularization.html">layer_activity_regularization</a>()</code> <br><code><a href="layer_add.html">layer_add</a>()</code> <br><code><a href="layer_additive_attention.html">layer_additive_attention</a>()</code> <br><code><a href="layer_alpha_dropout.html">layer_alpha_dropout</a>()</code> <br><code><a href="layer_attention.html">layer_attention</a>()</code> <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_average.html">layer_average</a>()</code> <br><code><a href="layer_average_pooling_1d.html">layer_average_pooling_1d</a>()</code> <br><code><a href="layer_average_pooling_2d.html">layer_average_pooling_2d</a>()</code> <br><code><a href="layer_average_pooling_3d.html">layer_average_pooling_3d</a>()</code> <br><code><a href="layer_batch_normalization.html">layer_batch_normalization</a>()</code> <br><code><a href="layer_bidirectional.html">layer_bidirectional</a>()</code> <br><code><a href="layer_category_encoding.html">layer_category_encoding</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_concatenate.html">layer_concatenate</a>()</code> <br><code><a href="layer_conv_1d.html">layer_conv_1d</a>()</code> <br><code><a href="layer_conv_1d_transpose.html">layer_conv_1d_transpose</a>()</code> <br><code><a href="layer_conv_2d.html">layer_conv_2d</a>()</code> <br><code><a href="layer_conv_2d_transpose.html">layer_conv_2d_transpose</a>()</code> <br><code><a href="layer_conv_3d.html">layer_conv_3d</a>()</code> <br><code><a href="layer_conv_3d_transpose.html">layer_conv_3d_transpose</a>()</code> <br><code><a href="layer_conv_lstm_1d.html">layer_conv_lstm_1d</a>()</code> <br><code><a href="layer_conv_lstm_2d.html">layer_conv_lstm_2d</a>()</code> <br><code><a href="layer_conv_lstm_3d.html">layer_conv_lstm_3d</a>()</code> <br><code><a href="layer_cropping_1d.html">layer_cropping_1d</a>()</code> <br><code><a href="layer_cropping_2d.html">layer_cropping_2d</a>()</code> <br><code><a href="layer_cropping_3d.html">layer_cropping_3d</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_dense.html">layer_dense</a>()</code> <br><code><a href="layer_depthwise_conv_1d.html">layer_depthwise_conv_1d</a>()</code> <br><code><a href="layer_depthwise_conv_2d.html">layer_depthwise_conv_2d</a>()</code> <br><code><a href="layer_discretization.html">layer_discretization</a>()</code> <br><code><a href="layer_dot.html">layer_dot</a>()</code> <br><code><a href="layer_dropout.html">layer_dropout</a>()</code> <br><code><a href="layer_einsum_dense.html">layer_einsum_dense</a>()</code> <br><code><a href="layer_embedding.html">layer_embedding</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_feature_space.html">layer_feature_space</a>()</code> <br><code><a href="layer_flatten.html">layer_flatten</a>()</code> <br><code><a href="layer_flax_module_wrapper.html">layer_flax_module_wrapper</a>()</code> <br><code><a href="layer_gaussian_dropout.html">layer_gaussian_dropout</a>()</code> <br><code><a href="layer_gaussian_noise.html">layer_gaussian_noise</a>()</code> <br><code><a href="layer_global_average_pooling_1d.html">layer_global_average_pooling_1d</a>()</code> <br><code><a href="layer_global_average_pooling_2d.html">layer_global_average_pooling_2d</a>()</code> <br><code><a href="layer_global_average_pooling_3d.html">layer_global_average_pooling_3d</a>()</code> <br><code><a href="layer_global_max_pooling_1d.html">layer_global_max_pooling_1d</a>()</code> <br><code><a href="layer_global_max_pooling_2d.html">layer_global_max_pooling_2d</a>()</code> <br><code><a href="layer_global_max_pooling_3d.html">layer_global_max_pooling_3d</a>()</code> <br><code><a href="layer_group_normalization.html">layer_group_normalization</a>()</code> <br><code><a href="layer_group_query_attention.html">layer_group_query_attention</a>()</code> <br><code><a href="layer_gru.html">layer_gru</a>()</code> <br><code><a href="layer_hashed_crossing.html">layer_hashed_crossing</a>()</code> <br><code><a href="layer_hashing.html">layer_hashing</a>()</code> <br><code><a href="layer_identity.html">layer_identity</a>()</code> <br><code><a href="layer_integer_lookup.html">layer_integer_lookup</a>()</code> <br><code><a href="layer_jax_model_wrapper.html">layer_jax_model_wrapper</a>()</code> <br><code><a href="layer_lambda.html">layer_lambda</a>()</code> <br><code><a href="layer_layer_normalization.html">layer_layer_normalization</a>()</code> <br><code><a href="layer_lstm.html">layer_lstm</a>()</code> <br><code><a href="layer_masking.html">layer_masking</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_max_pooling_1d.html">layer_max_pooling_1d</a>()</code> <br><code><a href="layer_max_pooling_2d.html">layer_max_pooling_2d</a>()</code> <br><code><a href="layer_max_pooling_3d.html">layer_max_pooling_3d</a>()</code> <br><code><a href="layer_maximum.html">layer_maximum</a>()</code> <br><code><a href="layer_mel_spectrogram.html">layer_mel_spectrogram</a>()</code> <br><code><a href="layer_minimum.html">layer_minimum</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_multi_head_attention.html">layer_multi_head_attention</a>()</code> <br><code><a href="layer_multiply.html">layer_multiply</a>()</code> <br><code><a href="layer_normalization.html">layer_normalization</a>()</code> <br><code><a href="layer_permute.html">layer_permute</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_brightness.html">layer_random_brightness</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_contrast.html">layer_random_contrast</a>()</code> <br><code><a href="layer_random_crop.html">layer_random_crop</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_flip.html">layer_random_flip</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_rotation.html">layer_random_rotation</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_random_translation.html">layer_random_translation</a>()</code> <br><code><a href="layer_random_zoom.html">layer_random_zoom</a>()</code> <br><code><a href="layer_repeat_vector.html">layer_repeat_vector</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_reshape.html">layer_reshape</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_rms_normalization.html">layer_rms_normalization</a>()</code> <br><code><a href="layer_rnn.html">layer_rnn</a>()</code> <br><code><a href="layer_separable_conv_1d.html">layer_separable_conv_1d</a>()</code> <br><code><a href="layer_separable_conv_2d.html">layer_separable_conv_2d</a>()</code> <br><code><a href="layer_simple_rnn.html">layer_simple_rnn</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br><code><a href="layer_spatial_dropout_1d.html">layer_spatial_dropout_1d</a>()</code> <br><code><a href="layer_spatial_dropout_2d.html">layer_spatial_dropout_2d</a>()</code> <br><code><a href="layer_spectral_normalization.html">layer_spectral_normalization</a>()</code> <br><code><a href="layer_stft_spectrogram.html">layer_stft_spectrogram</a>()</code> <br><code><a href="layer_string_lookup.html">layer_string_lookup</a>()</code> <br><code><a href="layer_subtract.html">layer_subtract</a>()</code> <br><code><a href="layer_text_vectorization.html">layer_text_vectorization</a>()</code> <br><code><a href="layer_tfsm.html">layer_tfsm</a>()</code> <br><code><a href="layer_time_distributed.html">layer_time_distributed</a>()</code> <br><code><a href="layer_torch_module_wrapper.html">layer_torch_module_wrapper</a>()</code> <br><code><a href="layer_unit_normalization.html">layer_unit_normalization</a>()</code> <br><code><a href="layer_upsampling_1d.html">layer_upsampling_1d</a>()</code> <br><code><a href="layer_upsampling_2d.html">layer_upsampling_2d</a>()</code> <br><code><a href="layer_upsampling_3d.html">layer_upsampling_3d</a>()</code> <br><code><a href="layer_zero_padding_1d.html">layer_zero_padding_1d</a>()</code> <br><code><a href="layer_zero_padding_2d.html">layer_zero_padding_2d</a>()</code> <br><code><a href="layer_zero_padding_3d.html">layer_zero_padding_3d</a>()</code> <br><code><a href="rnn_cell_gru.html">rnn_cell_gru</a>()</code> <br><code><a href="rnn_cell_lstm.html">rnn_cell_lstm</a>()</code> <br><code><a href="rnn_cell_simple.html">rnn_cell_simple</a>()</code> <br><code><a href="rnn_cells_stack.html">rnn_cells_stack</a>()</code> <br></p></div>
+<p>Other layers: <br><code><a href="Layer.html">Layer</a>()</code> <br><code><a href="layer_activation.html">layer_activation</a>()</code> <br><code><a href="layer_activation_elu.html">layer_activation_elu</a>()</code> <br><code><a href="layer_activation_leaky_relu.html">layer_activation_leaky_relu</a>()</code> <br><code><a href="layer_activation_parametric_relu.html">layer_activation_parametric_relu</a>()</code> <br><code><a href="layer_activation_relu.html">layer_activation_relu</a>()</code> <br><code><a href="layer_activation_softmax.html">layer_activation_softmax</a>()</code> <br><code><a href="layer_activity_regularization.html">layer_activity_regularization</a>()</code> <br><code><a href="layer_add.html">layer_add</a>()</code> <br><code><a href="layer_additive_attention.html">layer_additive_attention</a>()</code> <br><code><a href="layer_alpha_dropout.html">layer_alpha_dropout</a>()</code> <br><code><a href="layer_attention.html">layer_attention</a>()</code> <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_average.html">layer_average</a>()</code> <br><code><a href="layer_average_pooling_1d.html">layer_average_pooling_1d</a>()</code> <br><code><a href="layer_average_pooling_2d.html">layer_average_pooling_2d</a>()</code> <br><code><a href="layer_average_pooling_3d.html">layer_average_pooling_3d</a>()</code> <br><code><a href="layer_batch_normalization.html">layer_batch_normalization</a>()</code> <br><code><a href="layer_bidirectional.html">layer_bidirectional</a>()</code> <br><code><a href="layer_category_encoding.html">layer_category_encoding</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_concatenate.html">layer_concatenate</a>()</code> <br><code><a href="layer_conv_1d.html">layer_conv_1d</a>()</code> <br><code><a href="layer_conv_1d_transpose.html">layer_conv_1d_transpose</a>()</code> <br><code><a href="layer_conv_2d.html">layer_conv_2d</a>()</code> <br><code><a href="layer_conv_2d_transpose.html">layer_conv_2d_transpose</a>()</code> <br><code><a href="layer_conv_3d.html">layer_conv_3d</a>()</code> <br><code><a href="layer_conv_3d_transpose.html">layer_conv_3d_transpose</a>()</code> <br><code><a href="layer_conv_lstm_1d.html">layer_conv_lstm_1d</a>()</code> <br><code><a href="layer_conv_lstm_2d.html">layer_conv_lstm_2d</a>()</code> <br><code><a href="layer_conv_lstm_3d.html">layer_conv_lstm_3d</a>()</code> <br><code><a href="layer_cropping_1d.html">layer_cropping_1d</a>()</code> <br><code><a href="layer_cropping_2d.html">layer_cropping_2d</a>()</code> <br><code><a href="layer_cropping_3d.html">layer_cropping_3d</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_dense.html">layer_dense</a>()</code> <br><code><a href="layer_depthwise_conv_1d.html">layer_depthwise_conv_1d</a>()</code> <br><code><a href="layer_depthwise_conv_2d.html">layer_depthwise_conv_2d</a>()</code> <br><code><a href="layer_discretization.html">layer_discretization</a>()</code> <br><code><a href="layer_dot.html">layer_dot</a>()</code> <br><code><a href="layer_dropout.html">layer_dropout</a>()</code> <br><code><a href="layer_einsum_dense.html">layer_einsum_dense</a>()</code> <br><code><a href="layer_embedding.html">layer_embedding</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_feature_space.html">layer_feature_space</a>()</code> <br><code><a href="layer_flatten.html">layer_flatten</a>()</code> <br><code><a href="layer_flax_module_wrapper.html">layer_flax_module_wrapper</a>()</code> <br><code><a href="layer_gaussian_dropout.html">layer_gaussian_dropout</a>()</code> <br><code><a href="layer_gaussian_noise.html">layer_gaussian_noise</a>()</code> <br><code><a href="layer_global_average_pooling_1d.html">layer_global_average_pooling_1d</a>()</code> <br><code><a href="layer_global_average_pooling_2d.html">layer_global_average_pooling_2d</a>()</code> <br><code><a href="layer_global_average_pooling_3d.html">layer_global_average_pooling_3d</a>()</code> <br><code><a href="layer_global_max_pooling_1d.html">layer_global_max_pooling_1d</a>()</code> <br><code><a href="layer_global_max_pooling_2d.html">layer_global_max_pooling_2d</a>()</code> <br><code><a href="layer_global_max_pooling_3d.html">layer_global_max_pooling_3d</a>()</code> <br><code><a href="layer_group_normalization.html">layer_group_normalization</a>()</code> <br><code><a href="layer_group_query_attention.html">layer_group_query_attention</a>()</code> <br><code><a href="layer_gru.html">layer_gru</a>()</code> <br><code><a href="layer_hashed_crossing.html">layer_hashed_crossing</a>()</code> <br><code><a href="layer_hashing.html">layer_hashing</a>()</code> <br><code><a href="layer_identity.html">layer_identity</a>()</code> <br><code><a href="layer_integer_lookup.html">layer_integer_lookup</a>()</code> <br><code><a href="layer_jax_model_wrapper.html">layer_jax_model_wrapper</a>()</code> <br><code><a href="layer_lambda.html">layer_lambda</a>()</code> <br><code><a href="layer_layer_normalization.html">layer_layer_normalization</a>()</code> <br><code><a href="layer_lstm.html">layer_lstm</a>()</code> <br><code><a href="layer_masking.html">layer_masking</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_max_pooling_1d.html">layer_max_pooling_1d</a>()</code> <br><code><a href="layer_max_pooling_2d.html">layer_max_pooling_2d</a>()</code> <br><code><a href="layer_max_pooling_3d.html">layer_max_pooling_3d</a>()</code> <br><code><a href="layer_maximum.html">layer_maximum</a>()</code> <br><code><a href="layer_mel_spectrogram.html">layer_mel_spectrogram</a>()</code> <br><code><a href="layer_minimum.html">layer_minimum</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_multi_head_attention.html">layer_multi_head_attention</a>()</code> <br><code><a href="layer_multiply.html">layer_multiply</a>()</code> <br><code><a href="layer_normalization.html">layer_normalization</a>()</code> <br><code><a href="layer_permute.html">layer_permute</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_brightness.html">layer_random_brightness</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_contrast.html">layer_random_contrast</a>()</code> <br><code><a href="layer_random_crop.html">layer_random_crop</a>()</code> <br><code><a href="layer_random_elastic_transform.html">layer_random_elastic_transform</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_flip.html">layer_random_flip</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_rotation.html">layer_random_rotation</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_random_translation.html">layer_random_translation</a>()</code> <br><code><a href="layer_random_zoom.html">layer_random_zoom</a>()</code> <br><code><a href="layer_repeat_vector.html">layer_repeat_vector</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_reshape.html">layer_reshape</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_rms_normalization.html">layer_rms_normalization</a>()</code> <br><code><a href="layer_rnn.html">layer_rnn</a>()</code> <br><code><a href="layer_separable_conv_1d.html">layer_separable_conv_1d</a>()</code> <br><code><a href="layer_separable_conv_2d.html">layer_separable_conv_2d</a>()</code> <br><code><a href="layer_simple_rnn.html">layer_simple_rnn</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br><code><a href="layer_spatial_dropout_1d.html">layer_spatial_dropout_1d</a>()</code> <br><code><a href="layer_spatial_dropout_2d.html">layer_spatial_dropout_2d</a>()</code> <br><code><a href="layer_spectral_normalization.html">layer_spectral_normalization</a>()</code> <br><code><a href="layer_stft_spectrogram.html">layer_stft_spectrogram</a>()</code> <br><code><a href="layer_string_lookup.html">layer_string_lookup</a>()</code> <br><code><a href="layer_subtract.html">layer_subtract</a>()</code> <br><code><a href="layer_text_vectorization.html">layer_text_vectorization</a>()</code> <br><code><a href="layer_tfsm.html">layer_tfsm</a>()</code> <br><code><a href="layer_time_distributed.html">layer_time_distributed</a>()</code> <br><code><a href="layer_torch_module_wrapper.html">layer_torch_module_wrapper</a>()</code> <br><code><a href="layer_unit_normalization.html">layer_unit_normalization</a>()</code> <br><code><a href="layer_upsampling_1d.html">layer_upsampling_1d</a>()</code> <br><code><a href="layer_upsampling_2d.html">layer_upsampling_2d</a>()</code> <br><code><a href="layer_upsampling_3d.html">layer_upsampling_3d</a>()</code> <br><code><a href="layer_zero_padding_1d.html">layer_zero_padding_1d</a>()</code> <br><code><a href="layer_zero_padding_2d.html">layer_zero_padding_2d</a>()</code> <br><code><a href="layer_zero_padding_3d.html">layer_zero_padding_3d</a>()</code> <br><code><a href="rnn_cell_gru.html">rnn_cell_gru</a>()</code> <br><code><a href="rnn_cell_lstm.html">rnn_cell_lstm</a>()</code> <br><code><a href="rnn_cell_simple.html">rnn_cell_simple</a>()</code> <br><code><a href="rnn_cells_stack.html">rnn_cells_stack</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -171,7 +171,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/layer_spectral_normalization.html b/docs/dev/reference/layer_spectral_normalization.html
index 80a728fa9..403e09060 100644
--- a/docs/dev/reference/layer_spectral_normalization.html
+++ b/docs/dev/reference/layer_spectral_normalization.html
@@ -10,7 +10,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -123,7 +123,7 @@ <h2 id="reference">Reference<a class="anchor" aria-label="anchor" href="#referen
     <div class="section level2">
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
     <div class="dont-index"><p>Other normalization layers: <br><code><a href="layer_batch_normalization.html">layer_batch_normalization</a>()</code> <br><code><a href="layer_group_normalization.html">layer_group_normalization</a>()</code> <br><code><a href="layer_layer_normalization.html">layer_layer_normalization</a>()</code> <br><code><a href="layer_rms_normalization.html">layer_rms_normalization</a>()</code> <br><code><a href="layer_unit_normalization.html">layer_unit_normalization</a>()</code> <br></p>
-<p>Other layers: <br><code><a href="Layer.html">Layer</a>()</code> <br><code><a href="layer_activation.html">layer_activation</a>()</code> <br><code><a href="layer_activation_elu.html">layer_activation_elu</a>()</code> <br><code><a href="layer_activation_leaky_relu.html">layer_activation_leaky_relu</a>()</code> <br><code><a href="layer_activation_parametric_relu.html">layer_activation_parametric_relu</a>()</code> <br><code><a href="layer_activation_relu.html">layer_activation_relu</a>()</code> <br><code><a href="layer_activation_softmax.html">layer_activation_softmax</a>()</code> <br><code><a href="layer_activity_regularization.html">layer_activity_regularization</a>()</code> <br><code><a href="layer_add.html">layer_add</a>()</code> <br><code><a href="layer_additive_attention.html">layer_additive_attention</a>()</code> <br><code><a href="layer_alpha_dropout.html">layer_alpha_dropout</a>()</code> <br><code><a href="layer_attention.html">layer_attention</a>()</code> <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_average.html">layer_average</a>()</code> <br><code><a href="layer_average_pooling_1d.html">layer_average_pooling_1d</a>()</code> <br><code><a href="layer_average_pooling_2d.html">layer_average_pooling_2d</a>()</code> <br><code><a href="layer_average_pooling_3d.html">layer_average_pooling_3d</a>()</code> <br><code><a href="layer_batch_normalization.html">layer_batch_normalization</a>()</code> <br><code><a href="layer_bidirectional.html">layer_bidirectional</a>()</code> <br><code><a href="layer_category_encoding.html">layer_category_encoding</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_concatenate.html">layer_concatenate</a>()</code> <br><code><a href="layer_conv_1d.html">layer_conv_1d</a>()</code> <br><code><a href="layer_conv_1d_transpose.html">layer_conv_1d_transpose</a>()</code> <br><code><a href="layer_conv_2d.html">layer_conv_2d</a>()</code> <br><code><a href="layer_conv_2d_transpose.html">layer_conv_2d_transpose</a>()</code> <br><code><a href="layer_conv_3d.html">layer_conv_3d</a>()</code> <br><code><a href="layer_conv_3d_transpose.html">layer_conv_3d_transpose</a>()</code> <br><code><a href="layer_conv_lstm_1d.html">layer_conv_lstm_1d</a>()</code> <br><code><a href="layer_conv_lstm_2d.html">layer_conv_lstm_2d</a>()</code> <br><code><a href="layer_conv_lstm_3d.html">layer_conv_lstm_3d</a>()</code> <br><code><a href="layer_cropping_1d.html">layer_cropping_1d</a>()</code> <br><code><a href="layer_cropping_2d.html">layer_cropping_2d</a>()</code> <br><code><a href="layer_cropping_3d.html">layer_cropping_3d</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_dense.html">layer_dense</a>()</code> <br><code><a href="layer_depthwise_conv_1d.html">layer_depthwise_conv_1d</a>()</code> <br><code><a href="layer_depthwise_conv_2d.html">layer_depthwise_conv_2d</a>()</code> <br><code><a href="layer_discretization.html">layer_discretization</a>()</code> <br><code><a href="layer_dot.html">layer_dot</a>()</code> <br><code><a href="layer_dropout.html">layer_dropout</a>()</code> <br><code><a href="layer_einsum_dense.html">layer_einsum_dense</a>()</code> <br><code><a href="layer_embedding.html">layer_embedding</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_feature_space.html">layer_feature_space</a>()</code> <br><code><a href="layer_flatten.html">layer_flatten</a>()</code> <br><code><a href="layer_flax_module_wrapper.html">layer_flax_module_wrapper</a>()</code> <br><code><a href="layer_gaussian_dropout.html">layer_gaussian_dropout</a>()</code> <br><code><a href="layer_gaussian_noise.html">layer_gaussian_noise</a>()</code> <br><code><a href="layer_global_average_pooling_1d.html">layer_global_average_pooling_1d</a>()</code> <br><code><a href="layer_global_average_pooling_2d.html">layer_global_average_pooling_2d</a>()</code> <br><code><a href="layer_global_average_pooling_3d.html">layer_global_average_pooling_3d</a>()</code> <br><code><a href="layer_global_max_pooling_1d.html">layer_global_max_pooling_1d</a>()</code> <br><code><a href="layer_global_max_pooling_2d.html">layer_global_max_pooling_2d</a>()</code> <br><code><a href="layer_global_max_pooling_3d.html">layer_global_max_pooling_3d</a>()</code> <br><code><a href="layer_group_normalization.html">layer_group_normalization</a>()</code> <br><code><a href="layer_group_query_attention.html">layer_group_query_attention</a>()</code> <br><code><a href="layer_gru.html">layer_gru</a>()</code> <br><code><a href="layer_hashed_crossing.html">layer_hashed_crossing</a>()</code> <br><code><a href="layer_hashing.html">layer_hashing</a>()</code> <br><code><a href="layer_identity.html">layer_identity</a>()</code> <br><code><a href="layer_integer_lookup.html">layer_integer_lookup</a>()</code> <br><code><a href="layer_jax_model_wrapper.html">layer_jax_model_wrapper</a>()</code> <br><code><a href="layer_lambda.html">layer_lambda</a>()</code> <br><code><a href="layer_layer_normalization.html">layer_layer_normalization</a>()</code> <br><code><a href="layer_lstm.html">layer_lstm</a>()</code> <br><code><a href="layer_masking.html">layer_masking</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_max_pooling_1d.html">layer_max_pooling_1d</a>()</code> <br><code><a href="layer_max_pooling_2d.html">layer_max_pooling_2d</a>()</code> <br><code><a href="layer_max_pooling_3d.html">layer_max_pooling_3d</a>()</code> <br><code><a href="layer_maximum.html">layer_maximum</a>()</code> <br><code><a href="layer_mel_spectrogram.html">layer_mel_spectrogram</a>()</code> <br><code><a href="layer_minimum.html">layer_minimum</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_multi_head_attention.html">layer_multi_head_attention</a>()</code> <br><code><a href="layer_multiply.html">layer_multiply</a>()</code> <br><code><a href="layer_normalization.html">layer_normalization</a>()</code> <br><code><a href="layer_permute.html">layer_permute</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_brightness.html">layer_random_brightness</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_contrast.html">layer_random_contrast</a>()</code> <br><code><a href="layer_random_crop.html">layer_random_crop</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_flip.html">layer_random_flip</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_rotation.html">layer_random_rotation</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_random_translation.html">layer_random_translation</a>()</code> <br><code><a href="layer_random_zoom.html">layer_random_zoom</a>()</code> <br><code><a href="layer_repeat_vector.html">layer_repeat_vector</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_reshape.html">layer_reshape</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_rms_normalization.html">layer_rms_normalization</a>()</code> <br><code><a href="layer_rnn.html">layer_rnn</a>()</code> <br><code><a href="layer_separable_conv_1d.html">layer_separable_conv_1d</a>()</code> <br><code><a href="layer_separable_conv_2d.html">layer_separable_conv_2d</a>()</code> <br><code><a href="layer_simple_rnn.html">layer_simple_rnn</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br><code><a href="layer_spatial_dropout_1d.html">layer_spatial_dropout_1d</a>()</code> <br><code><a href="layer_spatial_dropout_2d.html">layer_spatial_dropout_2d</a>()</code> <br><code><a href="layer_spatial_dropout_3d.html">layer_spatial_dropout_3d</a>()</code> <br><code><a href="layer_stft_spectrogram.html">layer_stft_spectrogram</a>()</code> <br><code><a href="layer_string_lookup.html">layer_string_lookup</a>()</code> <br><code><a href="layer_subtract.html">layer_subtract</a>()</code> <br><code><a href="layer_text_vectorization.html">layer_text_vectorization</a>()</code> <br><code><a href="layer_tfsm.html">layer_tfsm</a>()</code> <br><code><a href="layer_time_distributed.html">layer_time_distributed</a>()</code> <br><code><a href="layer_torch_module_wrapper.html">layer_torch_module_wrapper</a>()</code> <br><code><a href="layer_unit_normalization.html">layer_unit_normalization</a>()</code> <br><code><a href="layer_upsampling_1d.html">layer_upsampling_1d</a>()</code> <br><code><a href="layer_upsampling_2d.html">layer_upsampling_2d</a>()</code> <br><code><a href="layer_upsampling_3d.html">layer_upsampling_3d</a>()</code> <br><code><a href="layer_zero_padding_1d.html">layer_zero_padding_1d</a>()</code> <br><code><a href="layer_zero_padding_2d.html">layer_zero_padding_2d</a>()</code> <br><code><a href="layer_zero_padding_3d.html">layer_zero_padding_3d</a>()</code> <br><code><a href="rnn_cell_gru.html">rnn_cell_gru</a>()</code> <br><code><a href="rnn_cell_lstm.html">rnn_cell_lstm</a>()</code> <br><code><a href="rnn_cell_simple.html">rnn_cell_simple</a>()</code> <br><code><a href="rnn_cells_stack.html">rnn_cells_stack</a>()</code> <br></p></div>
+<p>Other layers: <br><code><a href="Layer.html">Layer</a>()</code> <br><code><a href="layer_activation.html">layer_activation</a>()</code> <br><code><a href="layer_activation_elu.html">layer_activation_elu</a>()</code> <br><code><a href="layer_activation_leaky_relu.html">layer_activation_leaky_relu</a>()</code> <br><code><a href="layer_activation_parametric_relu.html">layer_activation_parametric_relu</a>()</code> <br><code><a href="layer_activation_relu.html">layer_activation_relu</a>()</code> <br><code><a href="layer_activation_softmax.html">layer_activation_softmax</a>()</code> <br><code><a href="layer_activity_regularization.html">layer_activity_regularization</a>()</code> <br><code><a href="layer_add.html">layer_add</a>()</code> <br><code><a href="layer_additive_attention.html">layer_additive_attention</a>()</code> <br><code><a href="layer_alpha_dropout.html">layer_alpha_dropout</a>()</code> <br><code><a href="layer_attention.html">layer_attention</a>()</code> <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_average.html">layer_average</a>()</code> <br><code><a href="layer_average_pooling_1d.html">layer_average_pooling_1d</a>()</code> <br><code><a href="layer_average_pooling_2d.html">layer_average_pooling_2d</a>()</code> <br><code><a href="layer_average_pooling_3d.html">layer_average_pooling_3d</a>()</code> <br><code><a href="layer_batch_normalization.html">layer_batch_normalization</a>()</code> <br><code><a href="layer_bidirectional.html">layer_bidirectional</a>()</code> <br><code><a href="layer_category_encoding.html">layer_category_encoding</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_concatenate.html">layer_concatenate</a>()</code> <br><code><a href="layer_conv_1d.html">layer_conv_1d</a>()</code> <br><code><a href="layer_conv_1d_transpose.html">layer_conv_1d_transpose</a>()</code> <br><code><a href="layer_conv_2d.html">layer_conv_2d</a>()</code> <br><code><a href="layer_conv_2d_transpose.html">layer_conv_2d_transpose</a>()</code> <br><code><a href="layer_conv_3d.html">layer_conv_3d</a>()</code> <br><code><a href="layer_conv_3d_transpose.html">layer_conv_3d_transpose</a>()</code> <br><code><a href="layer_conv_lstm_1d.html">layer_conv_lstm_1d</a>()</code> <br><code><a href="layer_conv_lstm_2d.html">layer_conv_lstm_2d</a>()</code> <br><code><a href="layer_conv_lstm_3d.html">layer_conv_lstm_3d</a>()</code> <br><code><a href="layer_cropping_1d.html">layer_cropping_1d</a>()</code> <br><code><a href="layer_cropping_2d.html">layer_cropping_2d</a>()</code> <br><code><a href="layer_cropping_3d.html">layer_cropping_3d</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_dense.html">layer_dense</a>()</code> <br><code><a href="layer_depthwise_conv_1d.html">layer_depthwise_conv_1d</a>()</code> <br><code><a href="layer_depthwise_conv_2d.html">layer_depthwise_conv_2d</a>()</code> <br><code><a href="layer_discretization.html">layer_discretization</a>()</code> <br><code><a href="layer_dot.html">layer_dot</a>()</code> <br><code><a href="layer_dropout.html">layer_dropout</a>()</code> <br><code><a href="layer_einsum_dense.html">layer_einsum_dense</a>()</code> <br><code><a href="layer_embedding.html">layer_embedding</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_feature_space.html">layer_feature_space</a>()</code> <br><code><a href="layer_flatten.html">layer_flatten</a>()</code> <br><code><a href="layer_flax_module_wrapper.html">layer_flax_module_wrapper</a>()</code> <br><code><a href="layer_gaussian_dropout.html">layer_gaussian_dropout</a>()</code> <br><code><a href="layer_gaussian_noise.html">layer_gaussian_noise</a>()</code> <br><code><a href="layer_global_average_pooling_1d.html">layer_global_average_pooling_1d</a>()</code> <br><code><a href="layer_global_average_pooling_2d.html">layer_global_average_pooling_2d</a>()</code> <br><code><a href="layer_global_average_pooling_3d.html">layer_global_average_pooling_3d</a>()</code> <br><code><a href="layer_global_max_pooling_1d.html">layer_global_max_pooling_1d</a>()</code> <br><code><a href="layer_global_max_pooling_2d.html">layer_global_max_pooling_2d</a>()</code> <br><code><a href="layer_global_max_pooling_3d.html">layer_global_max_pooling_3d</a>()</code> <br><code><a href="layer_group_normalization.html">layer_group_normalization</a>()</code> <br><code><a href="layer_group_query_attention.html">layer_group_query_attention</a>()</code> <br><code><a href="layer_gru.html">layer_gru</a>()</code> <br><code><a href="layer_hashed_crossing.html">layer_hashed_crossing</a>()</code> <br><code><a href="layer_hashing.html">layer_hashing</a>()</code> <br><code><a href="layer_identity.html">layer_identity</a>()</code> <br><code><a href="layer_integer_lookup.html">layer_integer_lookup</a>()</code> <br><code><a href="layer_jax_model_wrapper.html">layer_jax_model_wrapper</a>()</code> <br><code><a href="layer_lambda.html">layer_lambda</a>()</code> <br><code><a href="layer_layer_normalization.html">layer_layer_normalization</a>()</code> <br><code><a href="layer_lstm.html">layer_lstm</a>()</code> <br><code><a href="layer_masking.html">layer_masking</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_max_pooling_1d.html">layer_max_pooling_1d</a>()</code> <br><code><a href="layer_max_pooling_2d.html">layer_max_pooling_2d</a>()</code> <br><code><a href="layer_max_pooling_3d.html">layer_max_pooling_3d</a>()</code> <br><code><a href="layer_maximum.html">layer_maximum</a>()</code> <br><code><a href="layer_mel_spectrogram.html">layer_mel_spectrogram</a>()</code> <br><code><a href="layer_minimum.html">layer_minimum</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_multi_head_attention.html">layer_multi_head_attention</a>()</code> <br><code><a href="layer_multiply.html">layer_multiply</a>()</code> <br><code><a href="layer_normalization.html">layer_normalization</a>()</code> <br><code><a href="layer_permute.html">layer_permute</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_brightness.html">layer_random_brightness</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_contrast.html">layer_random_contrast</a>()</code> <br><code><a href="layer_random_crop.html">layer_random_crop</a>()</code> <br><code><a href="layer_random_elastic_transform.html">layer_random_elastic_transform</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_flip.html">layer_random_flip</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_rotation.html">layer_random_rotation</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_random_translation.html">layer_random_translation</a>()</code> <br><code><a href="layer_random_zoom.html">layer_random_zoom</a>()</code> <br><code><a href="layer_repeat_vector.html">layer_repeat_vector</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_reshape.html">layer_reshape</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_rms_normalization.html">layer_rms_normalization</a>()</code> <br><code><a href="layer_rnn.html">layer_rnn</a>()</code> <br><code><a href="layer_separable_conv_1d.html">layer_separable_conv_1d</a>()</code> <br><code><a href="layer_separable_conv_2d.html">layer_separable_conv_2d</a>()</code> <br><code><a href="layer_simple_rnn.html">layer_simple_rnn</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br><code><a href="layer_spatial_dropout_1d.html">layer_spatial_dropout_1d</a>()</code> <br><code><a href="layer_spatial_dropout_2d.html">layer_spatial_dropout_2d</a>()</code> <br><code><a href="layer_spatial_dropout_3d.html">layer_spatial_dropout_3d</a>()</code> <br><code><a href="layer_stft_spectrogram.html">layer_stft_spectrogram</a>()</code> <br><code><a href="layer_string_lookup.html">layer_string_lookup</a>()</code> <br><code><a href="layer_subtract.html">layer_subtract</a>()</code> <br><code><a href="layer_text_vectorization.html">layer_text_vectorization</a>()</code> <br><code><a href="layer_tfsm.html">layer_tfsm</a>()</code> <br><code><a href="layer_time_distributed.html">layer_time_distributed</a>()</code> <br><code><a href="layer_torch_module_wrapper.html">layer_torch_module_wrapper</a>()</code> <br><code><a href="layer_unit_normalization.html">layer_unit_normalization</a>()</code> <br><code><a href="layer_upsampling_1d.html">layer_upsampling_1d</a>()</code> <br><code><a href="layer_upsampling_2d.html">layer_upsampling_2d</a>()</code> <br><code><a href="layer_upsampling_3d.html">layer_upsampling_3d</a>()</code> <br><code><a href="layer_zero_padding_1d.html">layer_zero_padding_1d</a>()</code> <br><code><a href="layer_zero_padding_2d.html">layer_zero_padding_2d</a>()</code> <br><code><a href="layer_zero_padding_3d.html">layer_zero_padding_3d</a>()</code> <br><code><a href="rnn_cell_gru.html">rnn_cell_gru</a>()</code> <br><code><a href="rnn_cell_lstm.html">rnn_cell_lstm</a>()</code> <br><code><a href="rnn_cell_simple.html">rnn_cell_simple</a>()</code> <br><code><a href="rnn_cells_stack.html">rnn_cells_stack</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -135,7 +135,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/layer_stft_spectrogram.html b/docs/dev/reference/layer_stft_spectrogram.html
index 3b27dfed8..cfdeca72c 100644
--- a/docs/dev/reference/layer_stft_spectrogram.html
+++ b/docs/dev/reference/layer_stft_spectrogram.html
@@ -24,7 +24,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -259,8 +259,8 @@ <h2 id="output-shape">Output Shape<a class="anchor" aria-label="anchor" href="#o
     <div class="section level2">
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
     <div class="dont-index"><p>Other audio preprocessing layers: <br><code><a href="layer_mel_spectrogram.html">layer_mel_spectrogram</a>()</code> <br></p>
-<p>Other preprocessing layers: <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_category_encoding.html">layer_category_encoding</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_discretization.html">layer_discretization</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_feature_space.html">layer_feature_space</a>()</code> <br><code><a href="layer_hashed_crossing.html">layer_hashed_crossing</a>()</code> <br><code><a href="layer_hashing.html">layer_hashing</a>()</code> <br><code><a href="layer_integer_lookup.html">layer_integer_lookup</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_mel_spectrogram.html">layer_mel_spectrogram</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_normalization.html">layer_normalization</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_brightness.html">layer_random_brightness</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_contrast.html">layer_random_contrast</a>()</code> <br><code><a href="layer_random_crop.html">layer_random_crop</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_flip.html">layer_random_flip</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_rotation.html">layer_random_rotation</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_random_translation.html">layer_random_translation</a>()</code> <br><code><a href="layer_random_zoom.html">layer_random_zoom</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br><code><a href="layer_string_lookup.html">layer_string_lookup</a>()</code> <br><code><a href="layer_text_vectorization.html">layer_text_vectorization</a>()</code> <br></p>
-<p>Other layers: <br><code><a href="Layer.html">Layer</a>()</code> <br><code><a href="layer_activation.html">layer_activation</a>()</code> <br><code><a href="layer_activation_elu.html">layer_activation_elu</a>()</code> <br><code><a href="layer_activation_leaky_relu.html">layer_activation_leaky_relu</a>()</code> <br><code><a href="layer_activation_parametric_relu.html">layer_activation_parametric_relu</a>()</code> <br><code><a href="layer_activation_relu.html">layer_activation_relu</a>()</code> <br><code><a href="layer_activation_softmax.html">layer_activation_softmax</a>()</code> <br><code><a href="layer_activity_regularization.html">layer_activity_regularization</a>()</code> <br><code><a href="layer_add.html">layer_add</a>()</code> <br><code><a href="layer_additive_attention.html">layer_additive_attention</a>()</code> <br><code><a href="layer_alpha_dropout.html">layer_alpha_dropout</a>()</code> <br><code><a href="layer_attention.html">layer_attention</a>()</code> <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_average.html">layer_average</a>()</code> <br><code><a href="layer_average_pooling_1d.html">layer_average_pooling_1d</a>()</code> <br><code><a href="layer_average_pooling_2d.html">layer_average_pooling_2d</a>()</code> <br><code><a href="layer_average_pooling_3d.html">layer_average_pooling_3d</a>()</code> <br><code><a href="layer_batch_normalization.html">layer_batch_normalization</a>()</code> <br><code><a href="layer_bidirectional.html">layer_bidirectional</a>()</code> <br><code><a href="layer_category_encoding.html">layer_category_encoding</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_concatenate.html">layer_concatenate</a>()</code> <br><code><a href="layer_conv_1d.html">layer_conv_1d</a>()</code> <br><code><a href="layer_conv_1d_transpose.html">layer_conv_1d_transpose</a>()</code> <br><code><a href="layer_conv_2d.html">layer_conv_2d</a>()</code> <br><code><a href="layer_conv_2d_transpose.html">layer_conv_2d_transpose</a>()</code> <br><code><a href="layer_conv_3d.html">layer_conv_3d</a>()</code> <br><code><a href="layer_conv_3d_transpose.html">layer_conv_3d_transpose</a>()</code> <br><code><a href="layer_conv_lstm_1d.html">layer_conv_lstm_1d</a>()</code> <br><code><a href="layer_conv_lstm_2d.html">layer_conv_lstm_2d</a>()</code> <br><code><a href="layer_conv_lstm_3d.html">layer_conv_lstm_3d</a>()</code> <br><code><a href="layer_cropping_1d.html">layer_cropping_1d</a>()</code> <br><code><a href="layer_cropping_2d.html">layer_cropping_2d</a>()</code> <br><code><a href="layer_cropping_3d.html">layer_cropping_3d</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_dense.html">layer_dense</a>()</code> <br><code><a href="layer_depthwise_conv_1d.html">layer_depthwise_conv_1d</a>()</code> <br><code><a href="layer_depthwise_conv_2d.html">layer_depthwise_conv_2d</a>()</code> <br><code><a href="layer_discretization.html">layer_discretization</a>()</code> <br><code><a href="layer_dot.html">layer_dot</a>()</code> <br><code><a href="layer_dropout.html">layer_dropout</a>()</code> <br><code><a href="layer_einsum_dense.html">layer_einsum_dense</a>()</code> <br><code><a href="layer_embedding.html">layer_embedding</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_feature_space.html">layer_feature_space</a>()</code> <br><code><a href="layer_flatten.html">layer_flatten</a>()</code> <br><code><a href="layer_flax_module_wrapper.html">layer_flax_module_wrapper</a>()</code> <br><code><a href="layer_gaussian_dropout.html">layer_gaussian_dropout</a>()</code> <br><code><a href="layer_gaussian_noise.html">layer_gaussian_noise</a>()</code> <br><code><a href="layer_global_average_pooling_1d.html">layer_global_average_pooling_1d</a>()</code> <br><code><a href="layer_global_average_pooling_2d.html">layer_global_average_pooling_2d</a>()</code> <br><code><a href="layer_global_average_pooling_3d.html">layer_global_average_pooling_3d</a>()</code> <br><code><a href="layer_global_max_pooling_1d.html">layer_global_max_pooling_1d</a>()</code> <br><code><a href="layer_global_max_pooling_2d.html">layer_global_max_pooling_2d</a>()</code> <br><code><a href="layer_global_max_pooling_3d.html">layer_global_max_pooling_3d</a>()</code> <br><code><a href="layer_group_normalization.html">layer_group_normalization</a>()</code> <br><code><a href="layer_group_query_attention.html">layer_group_query_attention</a>()</code> <br><code><a href="layer_gru.html">layer_gru</a>()</code> <br><code><a href="layer_hashed_crossing.html">layer_hashed_crossing</a>()</code> <br><code><a href="layer_hashing.html">layer_hashing</a>()</code> <br><code><a href="layer_identity.html">layer_identity</a>()</code> <br><code><a href="layer_integer_lookup.html">layer_integer_lookup</a>()</code> <br><code><a href="layer_jax_model_wrapper.html">layer_jax_model_wrapper</a>()</code> <br><code><a href="layer_lambda.html">layer_lambda</a>()</code> <br><code><a href="layer_layer_normalization.html">layer_layer_normalization</a>()</code> <br><code><a href="layer_lstm.html">layer_lstm</a>()</code> <br><code><a href="layer_masking.html">layer_masking</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_max_pooling_1d.html">layer_max_pooling_1d</a>()</code> <br><code><a href="layer_max_pooling_2d.html">layer_max_pooling_2d</a>()</code> <br><code><a href="layer_max_pooling_3d.html">layer_max_pooling_3d</a>()</code> <br><code><a href="layer_maximum.html">layer_maximum</a>()</code> <br><code><a href="layer_mel_spectrogram.html">layer_mel_spectrogram</a>()</code> <br><code><a href="layer_minimum.html">layer_minimum</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_multi_head_attention.html">layer_multi_head_attention</a>()</code> <br><code><a href="layer_multiply.html">layer_multiply</a>()</code> <br><code><a href="layer_normalization.html">layer_normalization</a>()</code> <br><code><a href="layer_permute.html">layer_permute</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_brightness.html">layer_random_brightness</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_contrast.html">layer_random_contrast</a>()</code> <br><code><a href="layer_random_crop.html">layer_random_crop</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_flip.html">layer_random_flip</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_rotation.html">layer_random_rotation</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_random_translation.html">layer_random_translation</a>()</code> <br><code><a href="layer_random_zoom.html">layer_random_zoom</a>()</code> <br><code><a href="layer_repeat_vector.html">layer_repeat_vector</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_reshape.html">layer_reshape</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_rms_normalization.html">layer_rms_normalization</a>()</code> <br><code><a href="layer_rnn.html">layer_rnn</a>()</code> <br><code><a href="layer_separable_conv_1d.html">layer_separable_conv_1d</a>()</code> <br><code><a href="layer_separable_conv_2d.html">layer_separable_conv_2d</a>()</code> <br><code><a href="layer_simple_rnn.html">layer_simple_rnn</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br><code><a href="layer_spatial_dropout_1d.html">layer_spatial_dropout_1d</a>()</code> <br><code><a href="layer_spatial_dropout_2d.html">layer_spatial_dropout_2d</a>()</code> <br><code><a href="layer_spatial_dropout_3d.html">layer_spatial_dropout_3d</a>()</code> <br><code><a href="layer_spectral_normalization.html">layer_spectral_normalization</a>()</code> <br><code><a href="layer_string_lookup.html">layer_string_lookup</a>()</code> <br><code><a href="layer_subtract.html">layer_subtract</a>()</code> <br><code><a href="layer_text_vectorization.html">layer_text_vectorization</a>()</code> <br><code><a href="layer_tfsm.html">layer_tfsm</a>()</code> <br><code><a href="layer_time_distributed.html">layer_time_distributed</a>()</code> <br><code><a href="layer_torch_module_wrapper.html">layer_torch_module_wrapper</a>()</code> <br><code><a href="layer_unit_normalization.html">layer_unit_normalization</a>()</code> <br><code><a href="layer_upsampling_1d.html">layer_upsampling_1d</a>()</code> <br><code><a href="layer_upsampling_2d.html">layer_upsampling_2d</a>()</code> <br><code><a href="layer_upsampling_3d.html">layer_upsampling_3d</a>()</code> <br><code><a href="layer_zero_padding_1d.html">layer_zero_padding_1d</a>()</code> <br><code><a href="layer_zero_padding_2d.html">layer_zero_padding_2d</a>()</code> <br><code><a href="layer_zero_padding_3d.html">layer_zero_padding_3d</a>()</code> <br><code><a href="rnn_cell_gru.html">rnn_cell_gru</a>()</code> <br><code><a href="rnn_cell_lstm.html">rnn_cell_lstm</a>()</code> <br><code><a href="rnn_cell_simple.html">rnn_cell_simple</a>()</code> <br><code><a href="rnn_cells_stack.html">rnn_cells_stack</a>()</code> <br></p></div>
+<p>Other preprocessing layers: <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_category_encoding.html">layer_category_encoding</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_discretization.html">layer_discretization</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_feature_space.html">layer_feature_space</a>()</code> <br><code><a href="layer_hashed_crossing.html">layer_hashed_crossing</a>()</code> <br><code><a href="layer_hashing.html">layer_hashing</a>()</code> <br><code><a href="layer_integer_lookup.html">layer_integer_lookup</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_mel_spectrogram.html">layer_mel_spectrogram</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_normalization.html">layer_normalization</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_brightness.html">layer_random_brightness</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_contrast.html">layer_random_contrast</a>()</code> <br><code><a href="layer_random_crop.html">layer_random_crop</a>()</code> <br><code><a href="layer_random_elastic_transform.html">layer_random_elastic_transform</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_flip.html">layer_random_flip</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_rotation.html">layer_random_rotation</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_random_translation.html">layer_random_translation</a>()</code> <br><code><a href="layer_random_zoom.html">layer_random_zoom</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br><code><a href="layer_string_lookup.html">layer_string_lookup</a>()</code> <br><code><a href="layer_text_vectorization.html">layer_text_vectorization</a>()</code> <br></p>
+<p>Other layers: <br><code><a href="Layer.html">Layer</a>()</code> <br><code><a href="layer_activation.html">layer_activation</a>()</code> <br><code><a href="layer_activation_elu.html">layer_activation_elu</a>()</code> <br><code><a href="layer_activation_leaky_relu.html">layer_activation_leaky_relu</a>()</code> <br><code><a href="layer_activation_parametric_relu.html">layer_activation_parametric_relu</a>()</code> <br><code><a href="layer_activation_relu.html">layer_activation_relu</a>()</code> <br><code><a href="layer_activation_softmax.html">layer_activation_softmax</a>()</code> <br><code><a href="layer_activity_regularization.html">layer_activity_regularization</a>()</code> <br><code><a href="layer_add.html">layer_add</a>()</code> <br><code><a href="layer_additive_attention.html">layer_additive_attention</a>()</code> <br><code><a href="layer_alpha_dropout.html">layer_alpha_dropout</a>()</code> <br><code><a href="layer_attention.html">layer_attention</a>()</code> <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_average.html">layer_average</a>()</code> <br><code><a href="layer_average_pooling_1d.html">layer_average_pooling_1d</a>()</code> <br><code><a href="layer_average_pooling_2d.html">layer_average_pooling_2d</a>()</code> <br><code><a href="layer_average_pooling_3d.html">layer_average_pooling_3d</a>()</code> <br><code><a href="layer_batch_normalization.html">layer_batch_normalization</a>()</code> <br><code><a href="layer_bidirectional.html">layer_bidirectional</a>()</code> <br><code><a href="layer_category_encoding.html">layer_category_encoding</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_concatenate.html">layer_concatenate</a>()</code> <br><code><a href="layer_conv_1d.html">layer_conv_1d</a>()</code> <br><code><a href="layer_conv_1d_transpose.html">layer_conv_1d_transpose</a>()</code> <br><code><a href="layer_conv_2d.html">layer_conv_2d</a>()</code> <br><code><a href="layer_conv_2d_transpose.html">layer_conv_2d_transpose</a>()</code> <br><code><a href="layer_conv_3d.html">layer_conv_3d</a>()</code> <br><code><a href="layer_conv_3d_transpose.html">layer_conv_3d_transpose</a>()</code> <br><code><a href="layer_conv_lstm_1d.html">layer_conv_lstm_1d</a>()</code> <br><code><a href="layer_conv_lstm_2d.html">layer_conv_lstm_2d</a>()</code> <br><code><a href="layer_conv_lstm_3d.html">layer_conv_lstm_3d</a>()</code> <br><code><a href="layer_cropping_1d.html">layer_cropping_1d</a>()</code> <br><code><a href="layer_cropping_2d.html">layer_cropping_2d</a>()</code> <br><code><a href="layer_cropping_3d.html">layer_cropping_3d</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_dense.html">layer_dense</a>()</code> <br><code><a href="layer_depthwise_conv_1d.html">layer_depthwise_conv_1d</a>()</code> <br><code><a href="layer_depthwise_conv_2d.html">layer_depthwise_conv_2d</a>()</code> <br><code><a href="layer_discretization.html">layer_discretization</a>()</code> <br><code><a href="layer_dot.html">layer_dot</a>()</code> <br><code><a href="layer_dropout.html">layer_dropout</a>()</code> <br><code><a href="layer_einsum_dense.html">layer_einsum_dense</a>()</code> <br><code><a href="layer_embedding.html">layer_embedding</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_feature_space.html">layer_feature_space</a>()</code> <br><code><a href="layer_flatten.html">layer_flatten</a>()</code> <br><code><a href="layer_flax_module_wrapper.html">layer_flax_module_wrapper</a>()</code> <br><code><a href="layer_gaussian_dropout.html">layer_gaussian_dropout</a>()</code> <br><code><a href="layer_gaussian_noise.html">layer_gaussian_noise</a>()</code> <br><code><a href="layer_global_average_pooling_1d.html">layer_global_average_pooling_1d</a>()</code> <br><code><a href="layer_global_average_pooling_2d.html">layer_global_average_pooling_2d</a>()</code> <br><code><a href="layer_global_average_pooling_3d.html">layer_global_average_pooling_3d</a>()</code> <br><code><a href="layer_global_max_pooling_1d.html">layer_global_max_pooling_1d</a>()</code> <br><code><a href="layer_global_max_pooling_2d.html">layer_global_max_pooling_2d</a>()</code> <br><code><a href="layer_global_max_pooling_3d.html">layer_global_max_pooling_3d</a>()</code> <br><code><a href="layer_group_normalization.html">layer_group_normalization</a>()</code> <br><code><a href="layer_group_query_attention.html">layer_group_query_attention</a>()</code> <br><code><a href="layer_gru.html">layer_gru</a>()</code> <br><code><a href="layer_hashed_crossing.html">layer_hashed_crossing</a>()</code> <br><code><a href="layer_hashing.html">layer_hashing</a>()</code> <br><code><a href="layer_identity.html">layer_identity</a>()</code> <br><code><a href="layer_integer_lookup.html">layer_integer_lookup</a>()</code> <br><code><a href="layer_jax_model_wrapper.html">layer_jax_model_wrapper</a>()</code> <br><code><a href="layer_lambda.html">layer_lambda</a>()</code> <br><code><a href="layer_layer_normalization.html">layer_layer_normalization</a>()</code> <br><code><a href="layer_lstm.html">layer_lstm</a>()</code> <br><code><a href="layer_masking.html">layer_masking</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_max_pooling_1d.html">layer_max_pooling_1d</a>()</code> <br><code><a href="layer_max_pooling_2d.html">layer_max_pooling_2d</a>()</code> <br><code><a href="layer_max_pooling_3d.html">layer_max_pooling_3d</a>()</code> <br><code><a href="layer_maximum.html">layer_maximum</a>()</code> <br><code><a href="layer_mel_spectrogram.html">layer_mel_spectrogram</a>()</code> <br><code><a href="layer_minimum.html">layer_minimum</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_multi_head_attention.html">layer_multi_head_attention</a>()</code> <br><code><a href="layer_multiply.html">layer_multiply</a>()</code> <br><code><a href="layer_normalization.html">layer_normalization</a>()</code> <br><code><a href="layer_permute.html">layer_permute</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_brightness.html">layer_random_brightness</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_contrast.html">layer_random_contrast</a>()</code> <br><code><a href="layer_random_crop.html">layer_random_crop</a>()</code> <br><code><a href="layer_random_elastic_transform.html">layer_random_elastic_transform</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_flip.html">layer_random_flip</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_rotation.html">layer_random_rotation</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_random_translation.html">layer_random_translation</a>()</code> <br><code><a href="layer_random_zoom.html">layer_random_zoom</a>()</code> <br><code><a href="layer_repeat_vector.html">layer_repeat_vector</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_reshape.html">layer_reshape</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_rms_normalization.html">layer_rms_normalization</a>()</code> <br><code><a href="layer_rnn.html">layer_rnn</a>()</code> <br><code><a href="layer_separable_conv_1d.html">layer_separable_conv_1d</a>()</code> <br><code><a href="layer_separable_conv_2d.html">layer_separable_conv_2d</a>()</code> <br><code><a href="layer_simple_rnn.html">layer_simple_rnn</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br><code><a href="layer_spatial_dropout_1d.html">layer_spatial_dropout_1d</a>()</code> <br><code><a href="layer_spatial_dropout_2d.html">layer_spatial_dropout_2d</a>()</code> <br><code><a href="layer_spatial_dropout_3d.html">layer_spatial_dropout_3d</a>()</code> <br><code><a href="layer_spectral_normalization.html">layer_spectral_normalization</a>()</code> <br><code><a href="layer_string_lookup.html">layer_string_lookup</a>()</code> <br><code><a href="layer_subtract.html">layer_subtract</a>()</code> <br><code><a href="layer_text_vectorization.html">layer_text_vectorization</a>()</code> <br><code><a href="layer_tfsm.html">layer_tfsm</a>()</code> <br><code><a href="layer_time_distributed.html">layer_time_distributed</a>()</code> <br><code><a href="layer_torch_module_wrapper.html">layer_torch_module_wrapper</a>()</code> <br><code><a href="layer_unit_normalization.html">layer_unit_normalization</a>()</code> <br><code><a href="layer_upsampling_1d.html">layer_upsampling_1d</a>()</code> <br><code><a href="layer_upsampling_2d.html">layer_upsampling_2d</a>()</code> <br><code><a href="layer_upsampling_3d.html">layer_upsampling_3d</a>()</code> <br><code><a href="layer_zero_padding_1d.html">layer_zero_padding_1d</a>()</code> <br><code><a href="layer_zero_padding_2d.html">layer_zero_padding_2d</a>()</code> <br><code><a href="layer_zero_padding_3d.html">layer_zero_padding_3d</a>()</code> <br><code><a href="rnn_cell_gru.html">rnn_cell_gru</a>()</code> <br><code><a href="rnn_cell_lstm.html">rnn_cell_lstm</a>()</code> <br><code><a href="rnn_cell_simple.html">rnn_cell_simple</a>()</code> <br><code><a href="rnn_cells_stack.html">rnn_cells_stack</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -272,7 +272,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/layer_string_lookup.html b/docs/dev/reference/layer_string_lookup.html
index 20fec02c2..f9d60b5b0 100644
--- a/docs/dev/reference/layer_string_lookup.html
+++ b/docs/dev/reference/layer_string_lookup.html
@@ -82,7 +82,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -478,8 +478,8 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
     <div class="dont-index">
 <ul><li><p><a href="https://keras.io/api/layers/preprocessing_layers/categorical/string_lookup#stringlookup-class" class="external-link">https://keras.io/api/layers/preprocessing_layers/categorical/string_lookup#stringlookup-class</a></p></li>
 </ul><p>Other categorical features preprocessing layers: <br><code><a href="layer_category_encoding.html">layer_category_encoding</a>()</code> <br><code><a href="layer_hashed_crossing.html">layer_hashed_crossing</a>()</code> <br><code><a href="layer_hashing.html">layer_hashing</a>()</code> <br><code><a href="layer_integer_lookup.html">layer_integer_lookup</a>()</code> <br></p>
-<p>Other preprocessing layers: <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_category_encoding.html">layer_category_encoding</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_discretization.html">layer_discretization</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_feature_space.html">layer_feature_space</a>()</code> <br><code><a href="layer_hashed_crossing.html">layer_hashed_crossing</a>()</code> <br><code><a href="layer_hashing.html">layer_hashing</a>()</code> <br><code><a href="layer_integer_lookup.html">layer_integer_lookup</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_mel_spectrogram.html">layer_mel_spectrogram</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_normalization.html">layer_normalization</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_brightness.html">layer_random_brightness</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_contrast.html">layer_random_contrast</a>()</code> <br><code><a href="layer_random_crop.html">layer_random_crop</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_flip.html">layer_random_flip</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_rotation.html">layer_random_rotation</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_random_translation.html">layer_random_translation</a>()</code> <br><code><a href="layer_random_zoom.html">layer_random_zoom</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br><code><a href="layer_stft_spectrogram.html">layer_stft_spectrogram</a>()</code> <br><code><a href="layer_text_vectorization.html">layer_text_vectorization</a>()</code> <br></p>
-<p>Other layers: <br><code><a href="Layer.html">Layer</a>()</code> <br><code><a href="layer_activation.html">layer_activation</a>()</code> <br><code><a href="layer_activation_elu.html">layer_activation_elu</a>()</code> <br><code><a href="layer_activation_leaky_relu.html">layer_activation_leaky_relu</a>()</code> <br><code><a href="layer_activation_parametric_relu.html">layer_activation_parametric_relu</a>()</code> <br><code><a href="layer_activation_relu.html">layer_activation_relu</a>()</code> <br><code><a href="layer_activation_softmax.html">layer_activation_softmax</a>()</code> <br><code><a href="layer_activity_regularization.html">layer_activity_regularization</a>()</code> <br><code><a href="layer_add.html">layer_add</a>()</code> <br><code><a href="layer_additive_attention.html">layer_additive_attention</a>()</code> <br><code><a href="layer_alpha_dropout.html">layer_alpha_dropout</a>()</code> <br><code><a href="layer_attention.html">layer_attention</a>()</code> <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_average.html">layer_average</a>()</code> <br><code><a href="layer_average_pooling_1d.html">layer_average_pooling_1d</a>()</code> <br><code><a href="layer_average_pooling_2d.html">layer_average_pooling_2d</a>()</code> <br><code><a href="layer_average_pooling_3d.html">layer_average_pooling_3d</a>()</code> <br><code><a href="layer_batch_normalization.html">layer_batch_normalization</a>()</code> <br><code><a href="layer_bidirectional.html">layer_bidirectional</a>()</code> <br><code><a href="layer_category_encoding.html">layer_category_encoding</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_concatenate.html">layer_concatenate</a>()</code> <br><code><a href="layer_conv_1d.html">layer_conv_1d</a>()</code> <br><code><a href="layer_conv_1d_transpose.html">layer_conv_1d_transpose</a>()</code> <br><code><a href="layer_conv_2d.html">layer_conv_2d</a>()</code> <br><code><a href="layer_conv_2d_transpose.html">layer_conv_2d_transpose</a>()</code> <br><code><a href="layer_conv_3d.html">layer_conv_3d</a>()</code> <br><code><a href="layer_conv_3d_transpose.html">layer_conv_3d_transpose</a>()</code> <br><code><a href="layer_conv_lstm_1d.html">layer_conv_lstm_1d</a>()</code> <br><code><a href="layer_conv_lstm_2d.html">layer_conv_lstm_2d</a>()</code> <br><code><a href="layer_conv_lstm_3d.html">layer_conv_lstm_3d</a>()</code> <br><code><a href="layer_cropping_1d.html">layer_cropping_1d</a>()</code> <br><code><a href="layer_cropping_2d.html">layer_cropping_2d</a>()</code> <br><code><a href="layer_cropping_3d.html">layer_cropping_3d</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_dense.html">layer_dense</a>()</code> <br><code><a href="layer_depthwise_conv_1d.html">layer_depthwise_conv_1d</a>()</code> <br><code><a href="layer_depthwise_conv_2d.html">layer_depthwise_conv_2d</a>()</code> <br><code><a href="layer_discretization.html">layer_discretization</a>()</code> <br><code><a href="layer_dot.html">layer_dot</a>()</code> <br><code><a href="layer_dropout.html">layer_dropout</a>()</code> <br><code><a href="layer_einsum_dense.html">layer_einsum_dense</a>()</code> <br><code><a href="layer_embedding.html">layer_embedding</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_feature_space.html">layer_feature_space</a>()</code> <br><code><a href="layer_flatten.html">layer_flatten</a>()</code> <br><code><a href="layer_flax_module_wrapper.html">layer_flax_module_wrapper</a>()</code> <br><code><a href="layer_gaussian_dropout.html">layer_gaussian_dropout</a>()</code> <br><code><a href="layer_gaussian_noise.html">layer_gaussian_noise</a>()</code> <br><code><a href="layer_global_average_pooling_1d.html">layer_global_average_pooling_1d</a>()</code> <br><code><a href="layer_global_average_pooling_2d.html">layer_global_average_pooling_2d</a>()</code> <br><code><a href="layer_global_average_pooling_3d.html">layer_global_average_pooling_3d</a>()</code> <br><code><a href="layer_global_max_pooling_1d.html">layer_global_max_pooling_1d</a>()</code> <br><code><a href="layer_global_max_pooling_2d.html">layer_global_max_pooling_2d</a>()</code> <br><code><a href="layer_global_max_pooling_3d.html">layer_global_max_pooling_3d</a>()</code> <br><code><a href="layer_group_normalization.html">layer_group_normalization</a>()</code> <br><code><a href="layer_group_query_attention.html">layer_group_query_attention</a>()</code> <br><code><a href="layer_gru.html">layer_gru</a>()</code> <br><code><a href="layer_hashed_crossing.html">layer_hashed_crossing</a>()</code> <br><code><a href="layer_hashing.html">layer_hashing</a>()</code> <br><code><a href="layer_identity.html">layer_identity</a>()</code> <br><code><a href="layer_integer_lookup.html">layer_integer_lookup</a>()</code> <br><code><a href="layer_jax_model_wrapper.html">layer_jax_model_wrapper</a>()</code> <br><code><a href="layer_lambda.html">layer_lambda</a>()</code> <br><code><a href="layer_layer_normalization.html">layer_layer_normalization</a>()</code> <br><code><a href="layer_lstm.html">layer_lstm</a>()</code> <br><code><a href="layer_masking.html">layer_masking</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_max_pooling_1d.html">layer_max_pooling_1d</a>()</code> <br><code><a href="layer_max_pooling_2d.html">layer_max_pooling_2d</a>()</code> <br><code><a href="layer_max_pooling_3d.html">layer_max_pooling_3d</a>()</code> <br><code><a href="layer_maximum.html">layer_maximum</a>()</code> <br><code><a href="layer_mel_spectrogram.html">layer_mel_spectrogram</a>()</code> <br><code><a href="layer_minimum.html">layer_minimum</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_multi_head_attention.html">layer_multi_head_attention</a>()</code> <br><code><a href="layer_multiply.html">layer_multiply</a>()</code> <br><code><a href="layer_normalization.html">layer_normalization</a>()</code> <br><code><a href="layer_permute.html">layer_permute</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_brightness.html">layer_random_brightness</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_contrast.html">layer_random_contrast</a>()</code> <br><code><a href="layer_random_crop.html">layer_random_crop</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_flip.html">layer_random_flip</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_rotation.html">layer_random_rotation</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_random_translation.html">layer_random_translation</a>()</code> <br><code><a href="layer_random_zoom.html">layer_random_zoom</a>()</code> <br><code><a href="layer_repeat_vector.html">layer_repeat_vector</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_reshape.html">layer_reshape</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_rms_normalization.html">layer_rms_normalization</a>()</code> <br><code><a href="layer_rnn.html">layer_rnn</a>()</code> <br><code><a href="layer_separable_conv_1d.html">layer_separable_conv_1d</a>()</code> <br><code><a href="layer_separable_conv_2d.html">layer_separable_conv_2d</a>()</code> <br><code><a href="layer_simple_rnn.html">layer_simple_rnn</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br><code><a href="layer_spatial_dropout_1d.html">layer_spatial_dropout_1d</a>()</code> <br><code><a href="layer_spatial_dropout_2d.html">layer_spatial_dropout_2d</a>()</code> <br><code><a href="layer_spatial_dropout_3d.html">layer_spatial_dropout_3d</a>()</code> <br><code><a href="layer_spectral_normalization.html">layer_spectral_normalization</a>()</code> <br><code><a href="layer_stft_spectrogram.html">layer_stft_spectrogram</a>()</code> <br><code><a href="layer_subtract.html">layer_subtract</a>()</code> <br><code><a href="layer_text_vectorization.html">layer_text_vectorization</a>()</code> <br><code><a href="layer_tfsm.html">layer_tfsm</a>()</code> <br><code><a href="layer_time_distributed.html">layer_time_distributed</a>()</code> <br><code><a href="layer_torch_module_wrapper.html">layer_torch_module_wrapper</a>()</code> <br><code><a href="layer_unit_normalization.html">layer_unit_normalization</a>()</code> <br><code><a href="layer_upsampling_1d.html">layer_upsampling_1d</a>()</code> <br><code><a href="layer_upsampling_2d.html">layer_upsampling_2d</a>()</code> <br><code><a href="layer_upsampling_3d.html">layer_upsampling_3d</a>()</code> <br><code><a href="layer_zero_padding_1d.html">layer_zero_padding_1d</a>()</code> <br><code><a href="layer_zero_padding_2d.html">layer_zero_padding_2d</a>()</code> <br><code><a href="layer_zero_padding_3d.html">layer_zero_padding_3d</a>()</code> <br><code><a href="rnn_cell_gru.html">rnn_cell_gru</a>()</code> <br><code><a href="rnn_cell_lstm.html">rnn_cell_lstm</a>()</code> <br><code><a href="rnn_cell_simple.html">rnn_cell_simple</a>()</code> <br><code><a href="rnn_cells_stack.html">rnn_cells_stack</a>()</code> <br></p></div>
+<p>Other preprocessing layers: <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_category_encoding.html">layer_category_encoding</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_discretization.html">layer_discretization</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_feature_space.html">layer_feature_space</a>()</code> <br><code><a href="layer_hashed_crossing.html">layer_hashed_crossing</a>()</code> <br><code><a href="layer_hashing.html">layer_hashing</a>()</code> <br><code><a href="layer_integer_lookup.html">layer_integer_lookup</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_mel_spectrogram.html">layer_mel_spectrogram</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_normalization.html">layer_normalization</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_brightness.html">layer_random_brightness</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_contrast.html">layer_random_contrast</a>()</code> <br><code><a href="layer_random_crop.html">layer_random_crop</a>()</code> <br><code><a href="layer_random_elastic_transform.html">layer_random_elastic_transform</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_flip.html">layer_random_flip</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_rotation.html">layer_random_rotation</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_random_translation.html">layer_random_translation</a>()</code> <br><code><a href="layer_random_zoom.html">layer_random_zoom</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br><code><a href="layer_stft_spectrogram.html">layer_stft_spectrogram</a>()</code> <br><code><a href="layer_text_vectorization.html">layer_text_vectorization</a>()</code> <br></p>
+<p>Other layers: <br><code><a href="Layer.html">Layer</a>()</code> <br><code><a href="layer_activation.html">layer_activation</a>()</code> <br><code><a href="layer_activation_elu.html">layer_activation_elu</a>()</code> <br><code><a href="layer_activation_leaky_relu.html">layer_activation_leaky_relu</a>()</code> <br><code><a href="layer_activation_parametric_relu.html">layer_activation_parametric_relu</a>()</code> <br><code><a href="layer_activation_relu.html">layer_activation_relu</a>()</code> <br><code><a href="layer_activation_softmax.html">layer_activation_softmax</a>()</code> <br><code><a href="layer_activity_regularization.html">layer_activity_regularization</a>()</code> <br><code><a href="layer_add.html">layer_add</a>()</code> <br><code><a href="layer_additive_attention.html">layer_additive_attention</a>()</code> <br><code><a href="layer_alpha_dropout.html">layer_alpha_dropout</a>()</code> <br><code><a href="layer_attention.html">layer_attention</a>()</code> <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_average.html">layer_average</a>()</code> <br><code><a href="layer_average_pooling_1d.html">layer_average_pooling_1d</a>()</code> <br><code><a href="layer_average_pooling_2d.html">layer_average_pooling_2d</a>()</code> <br><code><a href="layer_average_pooling_3d.html">layer_average_pooling_3d</a>()</code> <br><code><a href="layer_batch_normalization.html">layer_batch_normalization</a>()</code> <br><code><a href="layer_bidirectional.html">layer_bidirectional</a>()</code> <br><code><a href="layer_category_encoding.html">layer_category_encoding</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_concatenate.html">layer_concatenate</a>()</code> <br><code><a href="layer_conv_1d.html">layer_conv_1d</a>()</code> <br><code><a href="layer_conv_1d_transpose.html">layer_conv_1d_transpose</a>()</code> <br><code><a href="layer_conv_2d.html">layer_conv_2d</a>()</code> <br><code><a href="layer_conv_2d_transpose.html">layer_conv_2d_transpose</a>()</code> <br><code><a href="layer_conv_3d.html">layer_conv_3d</a>()</code> <br><code><a href="layer_conv_3d_transpose.html">layer_conv_3d_transpose</a>()</code> <br><code><a href="layer_conv_lstm_1d.html">layer_conv_lstm_1d</a>()</code> <br><code><a href="layer_conv_lstm_2d.html">layer_conv_lstm_2d</a>()</code> <br><code><a href="layer_conv_lstm_3d.html">layer_conv_lstm_3d</a>()</code> <br><code><a href="layer_cropping_1d.html">layer_cropping_1d</a>()</code> <br><code><a href="layer_cropping_2d.html">layer_cropping_2d</a>()</code> <br><code><a href="layer_cropping_3d.html">layer_cropping_3d</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_dense.html">layer_dense</a>()</code> <br><code><a href="layer_depthwise_conv_1d.html">layer_depthwise_conv_1d</a>()</code> <br><code><a href="layer_depthwise_conv_2d.html">layer_depthwise_conv_2d</a>()</code> <br><code><a href="layer_discretization.html">layer_discretization</a>()</code> <br><code><a href="layer_dot.html">layer_dot</a>()</code> <br><code><a href="layer_dropout.html">layer_dropout</a>()</code> <br><code><a href="layer_einsum_dense.html">layer_einsum_dense</a>()</code> <br><code><a href="layer_embedding.html">layer_embedding</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_feature_space.html">layer_feature_space</a>()</code> <br><code><a href="layer_flatten.html">layer_flatten</a>()</code> <br><code><a href="layer_flax_module_wrapper.html">layer_flax_module_wrapper</a>()</code> <br><code><a href="layer_gaussian_dropout.html">layer_gaussian_dropout</a>()</code> <br><code><a href="layer_gaussian_noise.html">layer_gaussian_noise</a>()</code> <br><code><a href="layer_global_average_pooling_1d.html">layer_global_average_pooling_1d</a>()</code> <br><code><a href="layer_global_average_pooling_2d.html">layer_global_average_pooling_2d</a>()</code> <br><code><a href="layer_global_average_pooling_3d.html">layer_global_average_pooling_3d</a>()</code> <br><code><a href="layer_global_max_pooling_1d.html">layer_global_max_pooling_1d</a>()</code> <br><code><a href="layer_global_max_pooling_2d.html">layer_global_max_pooling_2d</a>()</code> <br><code><a href="layer_global_max_pooling_3d.html">layer_global_max_pooling_3d</a>()</code> <br><code><a href="layer_group_normalization.html">layer_group_normalization</a>()</code> <br><code><a href="layer_group_query_attention.html">layer_group_query_attention</a>()</code> <br><code><a href="layer_gru.html">layer_gru</a>()</code> <br><code><a href="layer_hashed_crossing.html">layer_hashed_crossing</a>()</code> <br><code><a href="layer_hashing.html">layer_hashing</a>()</code> <br><code><a href="layer_identity.html">layer_identity</a>()</code> <br><code><a href="layer_integer_lookup.html">layer_integer_lookup</a>()</code> <br><code><a href="layer_jax_model_wrapper.html">layer_jax_model_wrapper</a>()</code> <br><code><a href="layer_lambda.html">layer_lambda</a>()</code> <br><code><a href="layer_layer_normalization.html">layer_layer_normalization</a>()</code> <br><code><a href="layer_lstm.html">layer_lstm</a>()</code> <br><code><a href="layer_masking.html">layer_masking</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_max_pooling_1d.html">layer_max_pooling_1d</a>()</code> <br><code><a href="layer_max_pooling_2d.html">layer_max_pooling_2d</a>()</code> <br><code><a href="layer_max_pooling_3d.html">layer_max_pooling_3d</a>()</code> <br><code><a href="layer_maximum.html">layer_maximum</a>()</code> <br><code><a href="layer_mel_spectrogram.html">layer_mel_spectrogram</a>()</code> <br><code><a href="layer_minimum.html">layer_minimum</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_multi_head_attention.html">layer_multi_head_attention</a>()</code> <br><code><a href="layer_multiply.html">layer_multiply</a>()</code> <br><code><a href="layer_normalization.html">layer_normalization</a>()</code> <br><code><a href="layer_permute.html">layer_permute</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_brightness.html">layer_random_brightness</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_contrast.html">layer_random_contrast</a>()</code> <br><code><a href="layer_random_crop.html">layer_random_crop</a>()</code> <br><code><a href="layer_random_elastic_transform.html">layer_random_elastic_transform</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_flip.html">layer_random_flip</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_rotation.html">layer_random_rotation</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_random_translation.html">layer_random_translation</a>()</code> <br><code><a href="layer_random_zoom.html">layer_random_zoom</a>()</code> <br><code><a href="layer_repeat_vector.html">layer_repeat_vector</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_reshape.html">layer_reshape</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_rms_normalization.html">layer_rms_normalization</a>()</code> <br><code><a href="layer_rnn.html">layer_rnn</a>()</code> <br><code><a href="layer_separable_conv_1d.html">layer_separable_conv_1d</a>()</code> <br><code><a href="layer_separable_conv_2d.html">layer_separable_conv_2d</a>()</code> <br><code><a href="layer_simple_rnn.html">layer_simple_rnn</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br><code><a href="layer_spatial_dropout_1d.html">layer_spatial_dropout_1d</a>()</code> <br><code><a href="layer_spatial_dropout_2d.html">layer_spatial_dropout_2d</a>()</code> <br><code><a href="layer_spatial_dropout_3d.html">layer_spatial_dropout_3d</a>()</code> <br><code><a href="layer_spectral_normalization.html">layer_spectral_normalization</a>()</code> <br><code><a href="layer_stft_spectrogram.html">layer_stft_spectrogram</a>()</code> <br><code><a href="layer_subtract.html">layer_subtract</a>()</code> <br><code><a href="layer_text_vectorization.html">layer_text_vectorization</a>()</code> <br><code><a href="layer_tfsm.html">layer_tfsm</a>()</code> <br><code><a href="layer_time_distributed.html">layer_time_distributed</a>()</code> <br><code><a href="layer_torch_module_wrapper.html">layer_torch_module_wrapper</a>()</code> <br><code><a href="layer_unit_normalization.html">layer_unit_normalization</a>()</code> <br><code><a href="layer_upsampling_1d.html">layer_upsampling_1d</a>()</code> <br><code><a href="layer_upsampling_2d.html">layer_upsampling_2d</a>()</code> <br><code><a href="layer_upsampling_3d.html">layer_upsampling_3d</a>()</code> <br><code><a href="layer_zero_padding_1d.html">layer_zero_padding_1d</a>()</code> <br><code><a href="layer_zero_padding_2d.html">layer_zero_padding_2d</a>()</code> <br><code><a href="layer_zero_padding_3d.html">layer_zero_padding_3d</a>()</code> <br><code><a href="rnn_cell_gru.html">rnn_cell_gru</a>()</code> <br><code><a href="rnn_cell_lstm.html">rnn_cell_lstm</a>()</code> <br><code><a href="rnn_cell_simple.html">rnn_cell_simple</a>()</code> <br><code><a href="rnn_cells_stack.html">rnn_cells_stack</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -491,7 +491,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/layer_subtract.html b/docs/dev/reference/layer_subtract.html
index 0021eed31..ba1ed6393 100644
--- a/docs/dev/reference/layer_subtract.html
+++ b/docs/dev/reference/layer_subtract.html
@@ -12,7 +12,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -109,7 +109,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
     <div class="dont-index">
 <ul><li><p><a href="https://keras.io/api/layers/merging_layers/subtract#subtract-class" class="external-link">https://keras.io/api/layers/merging_layers/subtract#subtract-class</a></p></li>
 </ul><p>Other merging layers: <br><code><a href="layer_add.html">layer_add</a>()</code> <br><code><a href="layer_average.html">layer_average</a>()</code> <br><code><a href="layer_concatenate.html">layer_concatenate</a>()</code> <br><code><a href="layer_dot.html">layer_dot</a>()</code> <br><code><a href="layer_maximum.html">layer_maximum</a>()</code> <br><code><a href="layer_minimum.html">layer_minimum</a>()</code> <br><code><a href="layer_multiply.html">layer_multiply</a>()</code> <br></p>
-<p>Other layers: <br><code><a href="Layer.html">Layer</a>()</code> <br><code><a href="layer_activation.html">layer_activation</a>()</code> <br><code><a href="layer_activation_elu.html">layer_activation_elu</a>()</code> <br><code><a href="layer_activation_leaky_relu.html">layer_activation_leaky_relu</a>()</code> <br><code><a href="layer_activation_parametric_relu.html">layer_activation_parametric_relu</a>()</code> <br><code><a href="layer_activation_relu.html">layer_activation_relu</a>()</code> <br><code><a href="layer_activation_softmax.html">layer_activation_softmax</a>()</code> <br><code><a href="layer_activity_regularization.html">layer_activity_regularization</a>()</code> <br><code><a href="layer_add.html">layer_add</a>()</code> <br><code><a href="layer_additive_attention.html">layer_additive_attention</a>()</code> <br><code><a href="layer_alpha_dropout.html">layer_alpha_dropout</a>()</code> <br><code><a href="layer_attention.html">layer_attention</a>()</code> <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_average.html">layer_average</a>()</code> <br><code><a href="layer_average_pooling_1d.html">layer_average_pooling_1d</a>()</code> <br><code><a href="layer_average_pooling_2d.html">layer_average_pooling_2d</a>()</code> <br><code><a href="layer_average_pooling_3d.html">layer_average_pooling_3d</a>()</code> <br><code><a href="layer_batch_normalization.html">layer_batch_normalization</a>()</code> <br><code><a href="layer_bidirectional.html">layer_bidirectional</a>()</code> <br><code><a href="layer_category_encoding.html">layer_category_encoding</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_concatenate.html">layer_concatenate</a>()</code> <br><code><a href="layer_conv_1d.html">layer_conv_1d</a>()</code> <br><code><a href="layer_conv_1d_transpose.html">layer_conv_1d_transpose</a>()</code> <br><code><a href="layer_conv_2d.html">layer_conv_2d</a>()</code> <br><code><a href="layer_conv_2d_transpose.html">layer_conv_2d_transpose</a>()</code> <br><code><a href="layer_conv_3d.html">layer_conv_3d</a>()</code> <br><code><a href="layer_conv_3d_transpose.html">layer_conv_3d_transpose</a>()</code> <br><code><a href="layer_conv_lstm_1d.html">layer_conv_lstm_1d</a>()</code> <br><code><a href="layer_conv_lstm_2d.html">layer_conv_lstm_2d</a>()</code> <br><code><a href="layer_conv_lstm_3d.html">layer_conv_lstm_3d</a>()</code> <br><code><a href="layer_cropping_1d.html">layer_cropping_1d</a>()</code> <br><code><a href="layer_cropping_2d.html">layer_cropping_2d</a>()</code> <br><code><a href="layer_cropping_3d.html">layer_cropping_3d</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_dense.html">layer_dense</a>()</code> <br><code><a href="layer_depthwise_conv_1d.html">layer_depthwise_conv_1d</a>()</code> <br><code><a href="layer_depthwise_conv_2d.html">layer_depthwise_conv_2d</a>()</code> <br><code><a href="layer_discretization.html">layer_discretization</a>()</code> <br><code><a href="layer_dot.html">layer_dot</a>()</code> <br><code><a href="layer_dropout.html">layer_dropout</a>()</code> <br><code><a href="layer_einsum_dense.html">layer_einsum_dense</a>()</code> <br><code><a href="layer_embedding.html">layer_embedding</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_feature_space.html">layer_feature_space</a>()</code> <br><code><a href="layer_flatten.html">layer_flatten</a>()</code> <br><code><a href="layer_flax_module_wrapper.html">layer_flax_module_wrapper</a>()</code> <br><code><a href="layer_gaussian_dropout.html">layer_gaussian_dropout</a>()</code> <br><code><a href="layer_gaussian_noise.html">layer_gaussian_noise</a>()</code> <br><code><a href="layer_global_average_pooling_1d.html">layer_global_average_pooling_1d</a>()</code> <br><code><a href="layer_global_average_pooling_2d.html">layer_global_average_pooling_2d</a>()</code> <br><code><a href="layer_global_average_pooling_3d.html">layer_global_average_pooling_3d</a>()</code> <br><code><a href="layer_global_max_pooling_1d.html">layer_global_max_pooling_1d</a>()</code> <br><code><a href="layer_global_max_pooling_2d.html">layer_global_max_pooling_2d</a>()</code> <br><code><a href="layer_global_max_pooling_3d.html">layer_global_max_pooling_3d</a>()</code> <br><code><a href="layer_group_normalization.html">layer_group_normalization</a>()</code> <br><code><a href="layer_group_query_attention.html">layer_group_query_attention</a>()</code> <br><code><a href="layer_gru.html">layer_gru</a>()</code> <br><code><a href="layer_hashed_crossing.html">layer_hashed_crossing</a>()</code> <br><code><a href="layer_hashing.html">layer_hashing</a>()</code> <br><code><a href="layer_identity.html">layer_identity</a>()</code> <br><code><a href="layer_integer_lookup.html">layer_integer_lookup</a>()</code> <br><code><a href="layer_jax_model_wrapper.html">layer_jax_model_wrapper</a>()</code> <br><code><a href="layer_lambda.html">layer_lambda</a>()</code> <br><code><a href="layer_layer_normalization.html">layer_layer_normalization</a>()</code> <br><code><a href="layer_lstm.html">layer_lstm</a>()</code> <br><code><a href="layer_masking.html">layer_masking</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_max_pooling_1d.html">layer_max_pooling_1d</a>()</code> <br><code><a href="layer_max_pooling_2d.html">layer_max_pooling_2d</a>()</code> <br><code><a href="layer_max_pooling_3d.html">layer_max_pooling_3d</a>()</code> <br><code><a href="layer_maximum.html">layer_maximum</a>()</code> <br><code><a href="layer_mel_spectrogram.html">layer_mel_spectrogram</a>()</code> <br><code><a href="layer_minimum.html">layer_minimum</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_multi_head_attention.html">layer_multi_head_attention</a>()</code> <br><code><a href="layer_multiply.html">layer_multiply</a>()</code> <br><code><a href="layer_normalization.html">layer_normalization</a>()</code> <br><code><a href="layer_permute.html">layer_permute</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_brightness.html">layer_random_brightness</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_contrast.html">layer_random_contrast</a>()</code> <br><code><a href="layer_random_crop.html">layer_random_crop</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_flip.html">layer_random_flip</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_rotation.html">layer_random_rotation</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_random_translation.html">layer_random_translation</a>()</code> <br><code><a href="layer_random_zoom.html">layer_random_zoom</a>()</code> <br><code><a href="layer_repeat_vector.html">layer_repeat_vector</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_reshape.html">layer_reshape</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_rms_normalization.html">layer_rms_normalization</a>()</code> <br><code><a href="layer_rnn.html">layer_rnn</a>()</code> <br><code><a href="layer_separable_conv_1d.html">layer_separable_conv_1d</a>()</code> <br><code><a href="layer_separable_conv_2d.html">layer_separable_conv_2d</a>()</code> <br><code><a href="layer_simple_rnn.html">layer_simple_rnn</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br><code><a href="layer_spatial_dropout_1d.html">layer_spatial_dropout_1d</a>()</code> <br><code><a href="layer_spatial_dropout_2d.html">layer_spatial_dropout_2d</a>()</code> <br><code><a href="layer_spatial_dropout_3d.html">layer_spatial_dropout_3d</a>()</code> <br><code><a href="layer_spectral_normalization.html">layer_spectral_normalization</a>()</code> <br><code><a href="layer_stft_spectrogram.html">layer_stft_spectrogram</a>()</code> <br><code><a href="layer_string_lookup.html">layer_string_lookup</a>()</code> <br><code><a href="layer_text_vectorization.html">layer_text_vectorization</a>()</code> <br><code><a href="layer_tfsm.html">layer_tfsm</a>()</code> <br><code><a href="layer_time_distributed.html">layer_time_distributed</a>()</code> <br><code><a href="layer_torch_module_wrapper.html">layer_torch_module_wrapper</a>()</code> <br><code><a href="layer_unit_normalization.html">layer_unit_normalization</a>()</code> <br><code><a href="layer_upsampling_1d.html">layer_upsampling_1d</a>()</code> <br><code><a href="layer_upsampling_2d.html">layer_upsampling_2d</a>()</code> <br><code><a href="layer_upsampling_3d.html">layer_upsampling_3d</a>()</code> <br><code><a href="layer_zero_padding_1d.html">layer_zero_padding_1d</a>()</code> <br><code><a href="layer_zero_padding_2d.html">layer_zero_padding_2d</a>()</code> <br><code><a href="layer_zero_padding_3d.html">layer_zero_padding_3d</a>()</code> <br><code><a href="rnn_cell_gru.html">rnn_cell_gru</a>()</code> <br><code><a href="rnn_cell_lstm.html">rnn_cell_lstm</a>()</code> <br><code><a href="rnn_cell_simple.html">rnn_cell_simple</a>()</code> <br><code><a href="rnn_cells_stack.html">rnn_cells_stack</a>()</code> <br></p></div>
+<p>Other layers: <br><code><a href="Layer.html">Layer</a>()</code> <br><code><a href="layer_activation.html">layer_activation</a>()</code> <br><code><a href="layer_activation_elu.html">layer_activation_elu</a>()</code> <br><code><a href="layer_activation_leaky_relu.html">layer_activation_leaky_relu</a>()</code> <br><code><a href="layer_activation_parametric_relu.html">layer_activation_parametric_relu</a>()</code> <br><code><a href="layer_activation_relu.html">layer_activation_relu</a>()</code> <br><code><a href="layer_activation_softmax.html">layer_activation_softmax</a>()</code> <br><code><a href="layer_activity_regularization.html">layer_activity_regularization</a>()</code> <br><code><a href="layer_add.html">layer_add</a>()</code> <br><code><a href="layer_additive_attention.html">layer_additive_attention</a>()</code> <br><code><a href="layer_alpha_dropout.html">layer_alpha_dropout</a>()</code> <br><code><a href="layer_attention.html">layer_attention</a>()</code> <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_average.html">layer_average</a>()</code> <br><code><a href="layer_average_pooling_1d.html">layer_average_pooling_1d</a>()</code> <br><code><a href="layer_average_pooling_2d.html">layer_average_pooling_2d</a>()</code> <br><code><a href="layer_average_pooling_3d.html">layer_average_pooling_3d</a>()</code> <br><code><a href="layer_batch_normalization.html">layer_batch_normalization</a>()</code> <br><code><a href="layer_bidirectional.html">layer_bidirectional</a>()</code> <br><code><a href="layer_category_encoding.html">layer_category_encoding</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_concatenate.html">layer_concatenate</a>()</code> <br><code><a href="layer_conv_1d.html">layer_conv_1d</a>()</code> <br><code><a href="layer_conv_1d_transpose.html">layer_conv_1d_transpose</a>()</code> <br><code><a href="layer_conv_2d.html">layer_conv_2d</a>()</code> <br><code><a href="layer_conv_2d_transpose.html">layer_conv_2d_transpose</a>()</code> <br><code><a href="layer_conv_3d.html">layer_conv_3d</a>()</code> <br><code><a href="layer_conv_3d_transpose.html">layer_conv_3d_transpose</a>()</code> <br><code><a href="layer_conv_lstm_1d.html">layer_conv_lstm_1d</a>()</code> <br><code><a href="layer_conv_lstm_2d.html">layer_conv_lstm_2d</a>()</code> <br><code><a href="layer_conv_lstm_3d.html">layer_conv_lstm_3d</a>()</code> <br><code><a href="layer_cropping_1d.html">layer_cropping_1d</a>()</code> <br><code><a href="layer_cropping_2d.html">layer_cropping_2d</a>()</code> <br><code><a href="layer_cropping_3d.html">layer_cropping_3d</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_dense.html">layer_dense</a>()</code> <br><code><a href="layer_depthwise_conv_1d.html">layer_depthwise_conv_1d</a>()</code> <br><code><a href="layer_depthwise_conv_2d.html">layer_depthwise_conv_2d</a>()</code> <br><code><a href="layer_discretization.html">layer_discretization</a>()</code> <br><code><a href="layer_dot.html">layer_dot</a>()</code> <br><code><a href="layer_dropout.html">layer_dropout</a>()</code> <br><code><a href="layer_einsum_dense.html">layer_einsum_dense</a>()</code> <br><code><a href="layer_embedding.html">layer_embedding</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_feature_space.html">layer_feature_space</a>()</code> <br><code><a href="layer_flatten.html">layer_flatten</a>()</code> <br><code><a href="layer_flax_module_wrapper.html">layer_flax_module_wrapper</a>()</code> <br><code><a href="layer_gaussian_dropout.html">layer_gaussian_dropout</a>()</code> <br><code><a href="layer_gaussian_noise.html">layer_gaussian_noise</a>()</code> <br><code><a href="layer_global_average_pooling_1d.html">layer_global_average_pooling_1d</a>()</code> <br><code><a href="layer_global_average_pooling_2d.html">layer_global_average_pooling_2d</a>()</code> <br><code><a href="layer_global_average_pooling_3d.html">layer_global_average_pooling_3d</a>()</code> <br><code><a href="layer_global_max_pooling_1d.html">layer_global_max_pooling_1d</a>()</code> <br><code><a href="layer_global_max_pooling_2d.html">layer_global_max_pooling_2d</a>()</code> <br><code><a href="layer_global_max_pooling_3d.html">layer_global_max_pooling_3d</a>()</code> <br><code><a href="layer_group_normalization.html">layer_group_normalization</a>()</code> <br><code><a href="layer_group_query_attention.html">layer_group_query_attention</a>()</code> <br><code><a href="layer_gru.html">layer_gru</a>()</code> <br><code><a href="layer_hashed_crossing.html">layer_hashed_crossing</a>()</code> <br><code><a href="layer_hashing.html">layer_hashing</a>()</code> <br><code><a href="layer_identity.html">layer_identity</a>()</code> <br><code><a href="layer_integer_lookup.html">layer_integer_lookup</a>()</code> <br><code><a href="layer_jax_model_wrapper.html">layer_jax_model_wrapper</a>()</code> <br><code><a href="layer_lambda.html">layer_lambda</a>()</code> <br><code><a href="layer_layer_normalization.html">layer_layer_normalization</a>()</code> <br><code><a href="layer_lstm.html">layer_lstm</a>()</code> <br><code><a href="layer_masking.html">layer_masking</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_max_pooling_1d.html">layer_max_pooling_1d</a>()</code> <br><code><a href="layer_max_pooling_2d.html">layer_max_pooling_2d</a>()</code> <br><code><a href="layer_max_pooling_3d.html">layer_max_pooling_3d</a>()</code> <br><code><a href="layer_maximum.html">layer_maximum</a>()</code> <br><code><a href="layer_mel_spectrogram.html">layer_mel_spectrogram</a>()</code> <br><code><a href="layer_minimum.html">layer_minimum</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_multi_head_attention.html">layer_multi_head_attention</a>()</code> <br><code><a href="layer_multiply.html">layer_multiply</a>()</code> <br><code><a href="layer_normalization.html">layer_normalization</a>()</code> <br><code><a href="layer_permute.html">layer_permute</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_brightness.html">layer_random_brightness</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_contrast.html">layer_random_contrast</a>()</code> <br><code><a href="layer_random_crop.html">layer_random_crop</a>()</code> <br><code><a href="layer_random_elastic_transform.html">layer_random_elastic_transform</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_flip.html">layer_random_flip</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_rotation.html">layer_random_rotation</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_random_translation.html">layer_random_translation</a>()</code> <br><code><a href="layer_random_zoom.html">layer_random_zoom</a>()</code> <br><code><a href="layer_repeat_vector.html">layer_repeat_vector</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_reshape.html">layer_reshape</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_rms_normalization.html">layer_rms_normalization</a>()</code> <br><code><a href="layer_rnn.html">layer_rnn</a>()</code> <br><code><a href="layer_separable_conv_1d.html">layer_separable_conv_1d</a>()</code> <br><code><a href="layer_separable_conv_2d.html">layer_separable_conv_2d</a>()</code> <br><code><a href="layer_simple_rnn.html">layer_simple_rnn</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br><code><a href="layer_spatial_dropout_1d.html">layer_spatial_dropout_1d</a>()</code> <br><code><a href="layer_spatial_dropout_2d.html">layer_spatial_dropout_2d</a>()</code> <br><code><a href="layer_spatial_dropout_3d.html">layer_spatial_dropout_3d</a>()</code> <br><code><a href="layer_spectral_normalization.html">layer_spectral_normalization</a>()</code> <br><code><a href="layer_stft_spectrogram.html">layer_stft_spectrogram</a>()</code> <br><code><a href="layer_string_lookup.html">layer_string_lookup</a>()</code> <br><code><a href="layer_text_vectorization.html">layer_text_vectorization</a>()</code> <br><code><a href="layer_tfsm.html">layer_tfsm</a>()</code> <br><code><a href="layer_time_distributed.html">layer_time_distributed</a>()</code> <br><code><a href="layer_torch_module_wrapper.html">layer_torch_module_wrapper</a>()</code> <br><code><a href="layer_unit_normalization.html">layer_unit_normalization</a>()</code> <br><code><a href="layer_upsampling_1d.html">layer_upsampling_1d</a>()</code> <br><code><a href="layer_upsampling_2d.html">layer_upsampling_2d</a>()</code> <br><code><a href="layer_upsampling_3d.html">layer_upsampling_3d</a>()</code> <br><code><a href="layer_zero_padding_1d.html">layer_zero_padding_1d</a>()</code> <br><code><a href="layer_zero_padding_2d.html">layer_zero_padding_2d</a>()</code> <br><code><a href="layer_zero_padding_3d.html">layer_zero_padding_3d</a>()</code> <br><code><a href="rnn_cell_gru.html">rnn_cell_gru</a>()</code> <br><code><a href="rnn_cell_lstm.html">rnn_cell_lstm</a>()</code> <br><code><a href="rnn_cell_simple.html">rnn_cell_simple</a>()</code> <br><code><a href="rnn_cells_stack.html">rnn_cells_stack</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -121,7 +121,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/layer_text_vectorization.html b/docs/dev/reference/layer_text_vectorization.html
index 18ad31ee9..b4bd60aa4 100644
--- a/docs/dev/reference/layer_text_vectorization.html
+++ b/docs/dev/reference/layer_text_vectorization.html
@@ -106,7 +106,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -418,8 +418,8 @@ <h2 id="examples">Examples<a class="anchor" aria-label="anchor" href="#examples"
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
     <div class="dont-index">
 <ul><li><p><a href="https://keras.io/api/layers/preprocessing_layers/text/text_vectorization#textvectorization-class" class="external-link">https://keras.io/api/layers/preprocessing_layers/text/text_vectorization#textvectorization-class</a></p></li>
-</ul><p>Other preprocessing layers: <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_category_encoding.html">layer_category_encoding</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_discretization.html">layer_discretization</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_feature_space.html">layer_feature_space</a>()</code> <br><code><a href="layer_hashed_crossing.html">layer_hashed_crossing</a>()</code> <br><code><a href="layer_hashing.html">layer_hashing</a>()</code> <br><code><a href="layer_integer_lookup.html">layer_integer_lookup</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_mel_spectrogram.html">layer_mel_spectrogram</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_normalization.html">layer_normalization</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_brightness.html">layer_random_brightness</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_contrast.html">layer_random_contrast</a>()</code> <br><code><a href="layer_random_crop.html">layer_random_crop</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_flip.html">layer_random_flip</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_rotation.html">layer_random_rotation</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_random_translation.html">layer_random_translation</a>()</code> <br><code><a href="layer_random_zoom.html">layer_random_zoom</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br><code><a href="layer_stft_spectrogram.html">layer_stft_spectrogram</a>()</code> <br><code><a href="layer_string_lookup.html">layer_string_lookup</a>()</code> <br></p>
-<p>Other layers: <br><code><a href="Layer.html">Layer</a>()</code> <br><code><a href="layer_activation.html">layer_activation</a>()</code> <br><code><a href="layer_activation_elu.html">layer_activation_elu</a>()</code> <br><code><a href="layer_activation_leaky_relu.html">layer_activation_leaky_relu</a>()</code> <br><code><a href="layer_activation_parametric_relu.html">layer_activation_parametric_relu</a>()</code> <br><code><a href="layer_activation_relu.html">layer_activation_relu</a>()</code> <br><code><a href="layer_activation_softmax.html">layer_activation_softmax</a>()</code> <br><code><a href="layer_activity_regularization.html">layer_activity_regularization</a>()</code> <br><code><a href="layer_add.html">layer_add</a>()</code> <br><code><a href="layer_additive_attention.html">layer_additive_attention</a>()</code> <br><code><a href="layer_alpha_dropout.html">layer_alpha_dropout</a>()</code> <br><code><a href="layer_attention.html">layer_attention</a>()</code> <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_average.html">layer_average</a>()</code> <br><code><a href="layer_average_pooling_1d.html">layer_average_pooling_1d</a>()</code> <br><code><a href="layer_average_pooling_2d.html">layer_average_pooling_2d</a>()</code> <br><code><a href="layer_average_pooling_3d.html">layer_average_pooling_3d</a>()</code> <br><code><a href="layer_batch_normalization.html">layer_batch_normalization</a>()</code> <br><code><a href="layer_bidirectional.html">layer_bidirectional</a>()</code> <br><code><a href="layer_category_encoding.html">layer_category_encoding</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_concatenate.html">layer_concatenate</a>()</code> <br><code><a href="layer_conv_1d.html">layer_conv_1d</a>()</code> <br><code><a href="layer_conv_1d_transpose.html">layer_conv_1d_transpose</a>()</code> <br><code><a href="layer_conv_2d.html">layer_conv_2d</a>()</code> <br><code><a href="layer_conv_2d_transpose.html">layer_conv_2d_transpose</a>()</code> <br><code><a href="layer_conv_3d.html">layer_conv_3d</a>()</code> <br><code><a href="layer_conv_3d_transpose.html">layer_conv_3d_transpose</a>()</code> <br><code><a href="layer_conv_lstm_1d.html">layer_conv_lstm_1d</a>()</code> <br><code><a href="layer_conv_lstm_2d.html">layer_conv_lstm_2d</a>()</code> <br><code><a href="layer_conv_lstm_3d.html">layer_conv_lstm_3d</a>()</code> <br><code><a href="layer_cropping_1d.html">layer_cropping_1d</a>()</code> <br><code><a href="layer_cropping_2d.html">layer_cropping_2d</a>()</code> <br><code><a href="layer_cropping_3d.html">layer_cropping_3d</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_dense.html">layer_dense</a>()</code> <br><code><a href="layer_depthwise_conv_1d.html">layer_depthwise_conv_1d</a>()</code> <br><code><a href="layer_depthwise_conv_2d.html">layer_depthwise_conv_2d</a>()</code> <br><code><a href="layer_discretization.html">layer_discretization</a>()</code> <br><code><a href="layer_dot.html">layer_dot</a>()</code> <br><code><a href="layer_dropout.html">layer_dropout</a>()</code> <br><code><a href="layer_einsum_dense.html">layer_einsum_dense</a>()</code> <br><code><a href="layer_embedding.html">layer_embedding</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_feature_space.html">layer_feature_space</a>()</code> <br><code><a href="layer_flatten.html">layer_flatten</a>()</code> <br><code><a href="layer_flax_module_wrapper.html">layer_flax_module_wrapper</a>()</code> <br><code><a href="layer_gaussian_dropout.html">layer_gaussian_dropout</a>()</code> <br><code><a href="layer_gaussian_noise.html">layer_gaussian_noise</a>()</code> <br><code><a href="layer_global_average_pooling_1d.html">layer_global_average_pooling_1d</a>()</code> <br><code><a href="layer_global_average_pooling_2d.html">layer_global_average_pooling_2d</a>()</code> <br><code><a href="layer_global_average_pooling_3d.html">layer_global_average_pooling_3d</a>()</code> <br><code><a href="layer_global_max_pooling_1d.html">layer_global_max_pooling_1d</a>()</code> <br><code><a href="layer_global_max_pooling_2d.html">layer_global_max_pooling_2d</a>()</code> <br><code><a href="layer_global_max_pooling_3d.html">layer_global_max_pooling_3d</a>()</code> <br><code><a href="layer_group_normalization.html">layer_group_normalization</a>()</code> <br><code><a href="layer_group_query_attention.html">layer_group_query_attention</a>()</code> <br><code><a href="layer_gru.html">layer_gru</a>()</code> <br><code><a href="layer_hashed_crossing.html">layer_hashed_crossing</a>()</code> <br><code><a href="layer_hashing.html">layer_hashing</a>()</code> <br><code><a href="layer_identity.html">layer_identity</a>()</code> <br><code><a href="layer_integer_lookup.html">layer_integer_lookup</a>()</code> <br><code><a href="layer_jax_model_wrapper.html">layer_jax_model_wrapper</a>()</code> <br><code><a href="layer_lambda.html">layer_lambda</a>()</code> <br><code><a href="layer_layer_normalization.html">layer_layer_normalization</a>()</code> <br><code><a href="layer_lstm.html">layer_lstm</a>()</code> <br><code><a href="layer_masking.html">layer_masking</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_max_pooling_1d.html">layer_max_pooling_1d</a>()</code> <br><code><a href="layer_max_pooling_2d.html">layer_max_pooling_2d</a>()</code> <br><code><a href="layer_max_pooling_3d.html">layer_max_pooling_3d</a>()</code> <br><code><a href="layer_maximum.html">layer_maximum</a>()</code> <br><code><a href="layer_mel_spectrogram.html">layer_mel_spectrogram</a>()</code> <br><code><a href="layer_minimum.html">layer_minimum</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_multi_head_attention.html">layer_multi_head_attention</a>()</code> <br><code><a href="layer_multiply.html">layer_multiply</a>()</code> <br><code><a href="layer_normalization.html">layer_normalization</a>()</code> <br><code><a href="layer_permute.html">layer_permute</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_brightness.html">layer_random_brightness</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_contrast.html">layer_random_contrast</a>()</code> <br><code><a href="layer_random_crop.html">layer_random_crop</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_flip.html">layer_random_flip</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_rotation.html">layer_random_rotation</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_random_translation.html">layer_random_translation</a>()</code> <br><code><a href="layer_random_zoom.html">layer_random_zoom</a>()</code> <br><code><a href="layer_repeat_vector.html">layer_repeat_vector</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_reshape.html">layer_reshape</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_rms_normalization.html">layer_rms_normalization</a>()</code> <br><code><a href="layer_rnn.html">layer_rnn</a>()</code> <br><code><a href="layer_separable_conv_1d.html">layer_separable_conv_1d</a>()</code> <br><code><a href="layer_separable_conv_2d.html">layer_separable_conv_2d</a>()</code> <br><code><a href="layer_simple_rnn.html">layer_simple_rnn</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br><code><a href="layer_spatial_dropout_1d.html">layer_spatial_dropout_1d</a>()</code> <br><code><a href="layer_spatial_dropout_2d.html">layer_spatial_dropout_2d</a>()</code> <br><code><a href="layer_spatial_dropout_3d.html">layer_spatial_dropout_3d</a>()</code> <br><code><a href="layer_spectral_normalization.html">layer_spectral_normalization</a>()</code> <br><code><a href="layer_stft_spectrogram.html">layer_stft_spectrogram</a>()</code> <br><code><a href="layer_string_lookup.html">layer_string_lookup</a>()</code> <br><code><a href="layer_subtract.html">layer_subtract</a>()</code> <br><code><a href="layer_tfsm.html">layer_tfsm</a>()</code> <br><code><a href="layer_time_distributed.html">layer_time_distributed</a>()</code> <br><code><a href="layer_torch_module_wrapper.html">layer_torch_module_wrapper</a>()</code> <br><code><a href="layer_unit_normalization.html">layer_unit_normalization</a>()</code> <br><code><a href="layer_upsampling_1d.html">layer_upsampling_1d</a>()</code> <br><code><a href="layer_upsampling_2d.html">layer_upsampling_2d</a>()</code> <br><code><a href="layer_upsampling_3d.html">layer_upsampling_3d</a>()</code> <br><code><a href="layer_zero_padding_1d.html">layer_zero_padding_1d</a>()</code> <br><code><a href="layer_zero_padding_2d.html">layer_zero_padding_2d</a>()</code> <br><code><a href="layer_zero_padding_3d.html">layer_zero_padding_3d</a>()</code> <br><code><a href="rnn_cell_gru.html">rnn_cell_gru</a>()</code> <br><code><a href="rnn_cell_lstm.html">rnn_cell_lstm</a>()</code> <br><code><a href="rnn_cell_simple.html">rnn_cell_simple</a>()</code> <br><code><a href="rnn_cells_stack.html">rnn_cells_stack</a>()</code> <br></p></div>
+</ul><p>Other preprocessing layers: <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_category_encoding.html">layer_category_encoding</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_discretization.html">layer_discretization</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_feature_space.html">layer_feature_space</a>()</code> <br><code><a href="layer_hashed_crossing.html">layer_hashed_crossing</a>()</code> <br><code><a href="layer_hashing.html">layer_hashing</a>()</code> <br><code><a href="layer_integer_lookup.html">layer_integer_lookup</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_mel_spectrogram.html">layer_mel_spectrogram</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_normalization.html">layer_normalization</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_brightness.html">layer_random_brightness</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_contrast.html">layer_random_contrast</a>()</code> <br><code><a href="layer_random_crop.html">layer_random_crop</a>()</code> <br><code><a href="layer_random_elastic_transform.html">layer_random_elastic_transform</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_flip.html">layer_random_flip</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_rotation.html">layer_random_rotation</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_random_translation.html">layer_random_translation</a>()</code> <br><code><a href="layer_random_zoom.html">layer_random_zoom</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br><code><a href="layer_stft_spectrogram.html">layer_stft_spectrogram</a>()</code> <br><code><a href="layer_string_lookup.html">layer_string_lookup</a>()</code> <br></p>
+<p>Other layers: <br><code><a href="Layer.html">Layer</a>()</code> <br><code><a href="layer_activation.html">layer_activation</a>()</code> <br><code><a href="layer_activation_elu.html">layer_activation_elu</a>()</code> <br><code><a href="layer_activation_leaky_relu.html">layer_activation_leaky_relu</a>()</code> <br><code><a href="layer_activation_parametric_relu.html">layer_activation_parametric_relu</a>()</code> <br><code><a href="layer_activation_relu.html">layer_activation_relu</a>()</code> <br><code><a href="layer_activation_softmax.html">layer_activation_softmax</a>()</code> <br><code><a href="layer_activity_regularization.html">layer_activity_regularization</a>()</code> <br><code><a href="layer_add.html">layer_add</a>()</code> <br><code><a href="layer_additive_attention.html">layer_additive_attention</a>()</code> <br><code><a href="layer_alpha_dropout.html">layer_alpha_dropout</a>()</code> <br><code><a href="layer_attention.html">layer_attention</a>()</code> <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_average.html">layer_average</a>()</code> <br><code><a href="layer_average_pooling_1d.html">layer_average_pooling_1d</a>()</code> <br><code><a href="layer_average_pooling_2d.html">layer_average_pooling_2d</a>()</code> <br><code><a href="layer_average_pooling_3d.html">layer_average_pooling_3d</a>()</code> <br><code><a href="layer_batch_normalization.html">layer_batch_normalization</a>()</code> <br><code><a href="layer_bidirectional.html">layer_bidirectional</a>()</code> <br><code><a href="layer_category_encoding.html">layer_category_encoding</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_concatenate.html">layer_concatenate</a>()</code> <br><code><a href="layer_conv_1d.html">layer_conv_1d</a>()</code> <br><code><a href="layer_conv_1d_transpose.html">layer_conv_1d_transpose</a>()</code> <br><code><a href="layer_conv_2d.html">layer_conv_2d</a>()</code> <br><code><a href="layer_conv_2d_transpose.html">layer_conv_2d_transpose</a>()</code> <br><code><a href="layer_conv_3d.html">layer_conv_3d</a>()</code> <br><code><a href="layer_conv_3d_transpose.html">layer_conv_3d_transpose</a>()</code> <br><code><a href="layer_conv_lstm_1d.html">layer_conv_lstm_1d</a>()</code> <br><code><a href="layer_conv_lstm_2d.html">layer_conv_lstm_2d</a>()</code> <br><code><a href="layer_conv_lstm_3d.html">layer_conv_lstm_3d</a>()</code> <br><code><a href="layer_cropping_1d.html">layer_cropping_1d</a>()</code> <br><code><a href="layer_cropping_2d.html">layer_cropping_2d</a>()</code> <br><code><a href="layer_cropping_3d.html">layer_cropping_3d</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_dense.html">layer_dense</a>()</code> <br><code><a href="layer_depthwise_conv_1d.html">layer_depthwise_conv_1d</a>()</code> <br><code><a href="layer_depthwise_conv_2d.html">layer_depthwise_conv_2d</a>()</code> <br><code><a href="layer_discretization.html">layer_discretization</a>()</code> <br><code><a href="layer_dot.html">layer_dot</a>()</code> <br><code><a href="layer_dropout.html">layer_dropout</a>()</code> <br><code><a href="layer_einsum_dense.html">layer_einsum_dense</a>()</code> <br><code><a href="layer_embedding.html">layer_embedding</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_feature_space.html">layer_feature_space</a>()</code> <br><code><a href="layer_flatten.html">layer_flatten</a>()</code> <br><code><a href="layer_flax_module_wrapper.html">layer_flax_module_wrapper</a>()</code> <br><code><a href="layer_gaussian_dropout.html">layer_gaussian_dropout</a>()</code> <br><code><a href="layer_gaussian_noise.html">layer_gaussian_noise</a>()</code> <br><code><a href="layer_global_average_pooling_1d.html">layer_global_average_pooling_1d</a>()</code> <br><code><a href="layer_global_average_pooling_2d.html">layer_global_average_pooling_2d</a>()</code> <br><code><a href="layer_global_average_pooling_3d.html">layer_global_average_pooling_3d</a>()</code> <br><code><a href="layer_global_max_pooling_1d.html">layer_global_max_pooling_1d</a>()</code> <br><code><a href="layer_global_max_pooling_2d.html">layer_global_max_pooling_2d</a>()</code> <br><code><a href="layer_global_max_pooling_3d.html">layer_global_max_pooling_3d</a>()</code> <br><code><a href="layer_group_normalization.html">layer_group_normalization</a>()</code> <br><code><a href="layer_group_query_attention.html">layer_group_query_attention</a>()</code> <br><code><a href="layer_gru.html">layer_gru</a>()</code> <br><code><a href="layer_hashed_crossing.html">layer_hashed_crossing</a>()</code> <br><code><a href="layer_hashing.html">layer_hashing</a>()</code> <br><code><a href="layer_identity.html">layer_identity</a>()</code> <br><code><a href="layer_integer_lookup.html">layer_integer_lookup</a>()</code> <br><code><a href="layer_jax_model_wrapper.html">layer_jax_model_wrapper</a>()</code> <br><code><a href="layer_lambda.html">layer_lambda</a>()</code> <br><code><a href="layer_layer_normalization.html">layer_layer_normalization</a>()</code> <br><code><a href="layer_lstm.html">layer_lstm</a>()</code> <br><code><a href="layer_masking.html">layer_masking</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_max_pooling_1d.html">layer_max_pooling_1d</a>()</code> <br><code><a href="layer_max_pooling_2d.html">layer_max_pooling_2d</a>()</code> <br><code><a href="layer_max_pooling_3d.html">layer_max_pooling_3d</a>()</code> <br><code><a href="layer_maximum.html">layer_maximum</a>()</code> <br><code><a href="layer_mel_spectrogram.html">layer_mel_spectrogram</a>()</code> <br><code><a href="layer_minimum.html">layer_minimum</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_multi_head_attention.html">layer_multi_head_attention</a>()</code> <br><code><a href="layer_multiply.html">layer_multiply</a>()</code> <br><code><a href="layer_normalization.html">layer_normalization</a>()</code> <br><code><a href="layer_permute.html">layer_permute</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_brightness.html">layer_random_brightness</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_contrast.html">layer_random_contrast</a>()</code> <br><code><a href="layer_random_crop.html">layer_random_crop</a>()</code> <br><code><a href="layer_random_elastic_transform.html">layer_random_elastic_transform</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_flip.html">layer_random_flip</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_rotation.html">layer_random_rotation</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_random_translation.html">layer_random_translation</a>()</code> <br><code><a href="layer_random_zoom.html">layer_random_zoom</a>()</code> <br><code><a href="layer_repeat_vector.html">layer_repeat_vector</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_reshape.html">layer_reshape</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_rms_normalization.html">layer_rms_normalization</a>()</code> <br><code><a href="layer_rnn.html">layer_rnn</a>()</code> <br><code><a href="layer_separable_conv_1d.html">layer_separable_conv_1d</a>()</code> <br><code><a href="layer_separable_conv_2d.html">layer_separable_conv_2d</a>()</code> <br><code><a href="layer_simple_rnn.html">layer_simple_rnn</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br><code><a href="layer_spatial_dropout_1d.html">layer_spatial_dropout_1d</a>()</code> <br><code><a href="layer_spatial_dropout_2d.html">layer_spatial_dropout_2d</a>()</code> <br><code><a href="layer_spatial_dropout_3d.html">layer_spatial_dropout_3d</a>()</code> <br><code><a href="layer_spectral_normalization.html">layer_spectral_normalization</a>()</code> <br><code><a href="layer_stft_spectrogram.html">layer_stft_spectrogram</a>()</code> <br><code><a href="layer_string_lookup.html">layer_string_lookup</a>()</code> <br><code><a href="layer_subtract.html">layer_subtract</a>()</code> <br><code><a href="layer_tfsm.html">layer_tfsm</a>()</code> <br><code><a href="layer_time_distributed.html">layer_time_distributed</a>()</code> <br><code><a href="layer_torch_module_wrapper.html">layer_torch_module_wrapper</a>()</code> <br><code><a href="layer_unit_normalization.html">layer_unit_normalization</a>()</code> <br><code><a href="layer_upsampling_1d.html">layer_upsampling_1d</a>()</code> <br><code><a href="layer_upsampling_2d.html">layer_upsampling_2d</a>()</code> <br><code><a href="layer_upsampling_3d.html">layer_upsampling_3d</a>()</code> <br><code><a href="layer_zero_padding_1d.html">layer_zero_padding_1d</a>()</code> <br><code><a href="layer_zero_padding_2d.html">layer_zero_padding_2d</a>()</code> <br><code><a href="layer_zero_padding_3d.html">layer_zero_padding_3d</a>()</code> <br><code><a href="rnn_cell_gru.html">rnn_cell_gru</a>()</code> <br><code><a href="rnn_cell_lstm.html">rnn_cell_lstm</a>()</code> <br><code><a href="rnn_cell_simple.html">rnn_cell_simple</a>()</code> <br><code><a href="rnn_cells_stack.html">rnn_cells_stack</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -431,7 +431,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/layer_tfsm.html b/docs/dev/reference/layer_tfsm.html
index 7f0e8e889..c89aaf65c 100644
--- a/docs/dev/reference/layer_tfsm.html
+++ b/docs/dev/reference/layer_tfsm.html
@@ -8,7 +8,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -126,8 +126,8 @@ <h2 id="examples">Examples<a class="anchor" aria-label="anchor" href="#examples"
 <span id="cb1-5"><a href="#cb1-5" tabindex="-1"></a><span class="do">## Output Type:</span></span>
 <span id="cb1-6"><a href="#cb1-6" tabindex="-1"></a><span class="do">##   TensorSpec(shape=(None, 10), dtype=tf.float32, name=None)</span></span>
 <span id="cb1-7"><a href="#cb1-7" tabindex="-1"></a><span class="do">## Captures:</span></span>
-<span id="cb1-8"><a href="#cb1-8" tabindex="-1"></a><span class="do">##   133047893322704: TensorSpec(shape=(), dtype=tf.resource, name=None)</span></span>
-<span id="cb1-9"><a href="#cb1-9" tabindex="-1"></a><span class="do">##   133047893325008: TensorSpec(shape=(), dtype=tf.resource, name=None)</span></span>
+<span id="cb1-8"><a href="#cb1-8" tabindex="-1"></a><span class="do">##   131671756274576: TensorSpec(shape=(), dtype=tf.resource, name=None)</span></span>
+<span id="cb1-9"><a href="#cb1-9" tabindex="-1"></a><span class="do">##   131671756265936: TensorSpec(shape=(), dtype=tf.resource, name=None)</span></span>
 <span id="cb1-10"><a href="#cb1-10" tabindex="-1"></a></span></code></pre><p></p></div>
 <p></p><div class="sourceCode r"><pre><code><span><span class="va">reloaded_layer</span> <span class="op">&lt;-</span> <span class="fu"><a href="../reference/layer_tfsm.html">layer_tfsm</a></span><span class="op">(</span>filepath <span class="op">=</span> <span class="st">"path/to/artifact"</span><span class="op">)</span></span>
 <span><span class="va">input</span> <span class="op">&lt;-</span> <span class="fu"><a href="../reference/random_normal.html">random_normal</a></span><span class="op">(</span><span class="fu"><a href="https://rdrr.io/r/base/c.html" class="external-link">c</a></span><span class="op">(</span><span class="fl">2</span>, <span class="fl">784</span><span class="op">)</span><span class="op">)</span></span>
@@ -151,7 +151,7 @@ <h2 id="examples">Examples<a class="anchor" aria-label="anchor" href="#examples"
 </ul></div>
     <div class="section level2">
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
-    <div class="dont-index"><p>Other layers: <br><code><a href="Layer.html">Layer</a>()</code> <br><code><a href="layer_activation.html">layer_activation</a>()</code> <br><code><a href="layer_activation_elu.html">layer_activation_elu</a>()</code> <br><code><a href="layer_activation_leaky_relu.html">layer_activation_leaky_relu</a>()</code> <br><code><a href="layer_activation_parametric_relu.html">layer_activation_parametric_relu</a>()</code> <br><code><a href="layer_activation_relu.html">layer_activation_relu</a>()</code> <br><code><a href="layer_activation_softmax.html">layer_activation_softmax</a>()</code> <br><code><a href="layer_activity_regularization.html">layer_activity_regularization</a>()</code> <br><code><a href="layer_add.html">layer_add</a>()</code> <br><code><a href="layer_additive_attention.html">layer_additive_attention</a>()</code> <br><code><a href="layer_alpha_dropout.html">layer_alpha_dropout</a>()</code> <br><code><a href="layer_attention.html">layer_attention</a>()</code> <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_average.html">layer_average</a>()</code> <br><code><a href="layer_average_pooling_1d.html">layer_average_pooling_1d</a>()</code> <br><code><a href="layer_average_pooling_2d.html">layer_average_pooling_2d</a>()</code> <br><code><a href="layer_average_pooling_3d.html">layer_average_pooling_3d</a>()</code> <br><code><a href="layer_batch_normalization.html">layer_batch_normalization</a>()</code> <br><code><a href="layer_bidirectional.html">layer_bidirectional</a>()</code> <br><code><a href="layer_category_encoding.html">layer_category_encoding</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_concatenate.html">layer_concatenate</a>()</code> <br><code><a href="layer_conv_1d.html">layer_conv_1d</a>()</code> <br><code><a href="layer_conv_1d_transpose.html">layer_conv_1d_transpose</a>()</code> <br><code><a href="layer_conv_2d.html">layer_conv_2d</a>()</code> <br><code><a href="layer_conv_2d_transpose.html">layer_conv_2d_transpose</a>()</code> <br><code><a href="layer_conv_3d.html">layer_conv_3d</a>()</code> <br><code><a href="layer_conv_3d_transpose.html">layer_conv_3d_transpose</a>()</code> <br><code><a href="layer_conv_lstm_1d.html">layer_conv_lstm_1d</a>()</code> <br><code><a href="layer_conv_lstm_2d.html">layer_conv_lstm_2d</a>()</code> <br><code><a href="layer_conv_lstm_3d.html">layer_conv_lstm_3d</a>()</code> <br><code><a href="layer_cropping_1d.html">layer_cropping_1d</a>()</code> <br><code><a href="layer_cropping_2d.html">layer_cropping_2d</a>()</code> <br><code><a href="layer_cropping_3d.html">layer_cropping_3d</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_dense.html">layer_dense</a>()</code> <br><code><a href="layer_depthwise_conv_1d.html">layer_depthwise_conv_1d</a>()</code> <br><code><a href="layer_depthwise_conv_2d.html">layer_depthwise_conv_2d</a>()</code> <br><code><a href="layer_discretization.html">layer_discretization</a>()</code> <br><code><a href="layer_dot.html">layer_dot</a>()</code> <br><code><a href="layer_dropout.html">layer_dropout</a>()</code> <br><code><a href="layer_einsum_dense.html">layer_einsum_dense</a>()</code> <br><code><a href="layer_embedding.html">layer_embedding</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_feature_space.html">layer_feature_space</a>()</code> <br><code><a href="layer_flatten.html">layer_flatten</a>()</code> <br><code><a href="layer_flax_module_wrapper.html">layer_flax_module_wrapper</a>()</code> <br><code><a href="layer_gaussian_dropout.html">layer_gaussian_dropout</a>()</code> <br><code><a href="layer_gaussian_noise.html">layer_gaussian_noise</a>()</code> <br><code><a href="layer_global_average_pooling_1d.html">layer_global_average_pooling_1d</a>()</code> <br><code><a href="layer_global_average_pooling_2d.html">layer_global_average_pooling_2d</a>()</code> <br><code><a href="layer_global_average_pooling_3d.html">layer_global_average_pooling_3d</a>()</code> <br><code><a href="layer_global_max_pooling_1d.html">layer_global_max_pooling_1d</a>()</code> <br><code><a href="layer_global_max_pooling_2d.html">layer_global_max_pooling_2d</a>()</code> <br><code><a href="layer_global_max_pooling_3d.html">layer_global_max_pooling_3d</a>()</code> <br><code><a href="layer_group_normalization.html">layer_group_normalization</a>()</code> <br><code><a href="layer_group_query_attention.html">layer_group_query_attention</a>()</code> <br><code><a href="layer_gru.html">layer_gru</a>()</code> <br><code><a href="layer_hashed_crossing.html">layer_hashed_crossing</a>()</code> <br><code><a href="layer_hashing.html">layer_hashing</a>()</code> <br><code><a href="layer_identity.html">layer_identity</a>()</code> <br><code><a href="layer_integer_lookup.html">layer_integer_lookup</a>()</code> <br><code><a href="layer_jax_model_wrapper.html">layer_jax_model_wrapper</a>()</code> <br><code><a href="layer_lambda.html">layer_lambda</a>()</code> <br><code><a href="layer_layer_normalization.html">layer_layer_normalization</a>()</code> <br><code><a href="layer_lstm.html">layer_lstm</a>()</code> <br><code><a href="layer_masking.html">layer_masking</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_max_pooling_1d.html">layer_max_pooling_1d</a>()</code> <br><code><a href="layer_max_pooling_2d.html">layer_max_pooling_2d</a>()</code> <br><code><a href="layer_max_pooling_3d.html">layer_max_pooling_3d</a>()</code> <br><code><a href="layer_maximum.html">layer_maximum</a>()</code> <br><code><a href="layer_mel_spectrogram.html">layer_mel_spectrogram</a>()</code> <br><code><a href="layer_minimum.html">layer_minimum</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_multi_head_attention.html">layer_multi_head_attention</a>()</code> <br><code><a href="layer_multiply.html">layer_multiply</a>()</code> <br><code><a href="layer_normalization.html">layer_normalization</a>()</code> <br><code><a href="layer_permute.html">layer_permute</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_brightness.html">layer_random_brightness</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_contrast.html">layer_random_contrast</a>()</code> <br><code><a href="layer_random_crop.html">layer_random_crop</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_flip.html">layer_random_flip</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_rotation.html">layer_random_rotation</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_random_translation.html">layer_random_translation</a>()</code> <br><code><a href="layer_random_zoom.html">layer_random_zoom</a>()</code> <br><code><a href="layer_repeat_vector.html">layer_repeat_vector</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_reshape.html">layer_reshape</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_rms_normalization.html">layer_rms_normalization</a>()</code> <br><code><a href="layer_rnn.html">layer_rnn</a>()</code> <br><code><a href="layer_separable_conv_1d.html">layer_separable_conv_1d</a>()</code> <br><code><a href="layer_separable_conv_2d.html">layer_separable_conv_2d</a>()</code> <br><code><a href="layer_simple_rnn.html">layer_simple_rnn</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br><code><a href="layer_spatial_dropout_1d.html">layer_spatial_dropout_1d</a>()</code> <br><code><a href="layer_spatial_dropout_2d.html">layer_spatial_dropout_2d</a>()</code> <br><code><a href="layer_spatial_dropout_3d.html">layer_spatial_dropout_3d</a>()</code> <br><code><a href="layer_spectral_normalization.html">layer_spectral_normalization</a>()</code> <br><code><a href="layer_stft_spectrogram.html">layer_stft_spectrogram</a>()</code> <br><code><a href="layer_string_lookup.html">layer_string_lookup</a>()</code> <br><code><a href="layer_subtract.html">layer_subtract</a>()</code> <br><code><a href="layer_text_vectorization.html">layer_text_vectorization</a>()</code> <br><code><a href="layer_time_distributed.html">layer_time_distributed</a>()</code> <br><code><a href="layer_torch_module_wrapper.html">layer_torch_module_wrapper</a>()</code> <br><code><a href="layer_unit_normalization.html">layer_unit_normalization</a>()</code> <br><code><a href="layer_upsampling_1d.html">layer_upsampling_1d</a>()</code> <br><code><a href="layer_upsampling_2d.html">layer_upsampling_2d</a>()</code> <br><code><a href="layer_upsampling_3d.html">layer_upsampling_3d</a>()</code> <br><code><a href="layer_zero_padding_1d.html">layer_zero_padding_1d</a>()</code> <br><code><a href="layer_zero_padding_2d.html">layer_zero_padding_2d</a>()</code> <br><code><a href="layer_zero_padding_3d.html">layer_zero_padding_3d</a>()</code> <br><code><a href="rnn_cell_gru.html">rnn_cell_gru</a>()</code> <br><code><a href="rnn_cell_lstm.html">rnn_cell_lstm</a>()</code> <br><code><a href="rnn_cell_simple.html">rnn_cell_simple</a>()</code> <br><code><a href="rnn_cells_stack.html">rnn_cells_stack</a>()</code> <br></p>
+    <div class="dont-index"><p>Other layers: <br><code><a href="Layer.html">Layer</a>()</code> <br><code><a href="layer_activation.html">layer_activation</a>()</code> <br><code><a href="layer_activation_elu.html">layer_activation_elu</a>()</code> <br><code><a href="layer_activation_leaky_relu.html">layer_activation_leaky_relu</a>()</code> <br><code><a href="layer_activation_parametric_relu.html">layer_activation_parametric_relu</a>()</code> <br><code><a href="layer_activation_relu.html">layer_activation_relu</a>()</code> <br><code><a href="layer_activation_softmax.html">layer_activation_softmax</a>()</code> <br><code><a href="layer_activity_regularization.html">layer_activity_regularization</a>()</code> <br><code><a href="layer_add.html">layer_add</a>()</code> <br><code><a href="layer_additive_attention.html">layer_additive_attention</a>()</code> <br><code><a href="layer_alpha_dropout.html">layer_alpha_dropout</a>()</code> <br><code><a href="layer_attention.html">layer_attention</a>()</code> <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_average.html">layer_average</a>()</code> <br><code><a href="layer_average_pooling_1d.html">layer_average_pooling_1d</a>()</code> <br><code><a href="layer_average_pooling_2d.html">layer_average_pooling_2d</a>()</code> <br><code><a href="layer_average_pooling_3d.html">layer_average_pooling_3d</a>()</code> <br><code><a href="layer_batch_normalization.html">layer_batch_normalization</a>()</code> <br><code><a href="layer_bidirectional.html">layer_bidirectional</a>()</code> <br><code><a href="layer_category_encoding.html">layer_category_encoding</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_concatenate.html">layer_concatenate</a>()</code> <br><code><a href="layer_conv_1d.html">layer_conv_1d</a>()</code> <br><code><a href="layer_conv_1d_transpose.html">layer_conv_1d_transpose</a>()</code> <br><code><a href="layer_conv_2d.html">layer_conv_2d</a>()</code> <br><code><a href="layer_conv_2d_transpose.html">layer_conv_2d_transpose</a>()</code> <br><code><a href="layer_conv_3d.html">layer_conv_3d</a>()</code> <br><code><a href="layer_conv_3d_transpose.html">layer_conv_3d_transpose</a>()</code> <br><code><a href="layer_conv_lstm_1d.html">layer_conv_lstm_1d</a>()</code> <br><code><a href="layer_conv_lstm_2d.html">layer_conv_lstm_2d</a>()</code> <br><code><a href="layer_conv_lstm_3d.html">layer_conv_lstm_3d</a>()</code> <br><code><a href="layer_cropping_1d.html">layer_cropping_1d</a>()</code> <br><code><a href="layer_cropping_2d.html">layer_cropping_2d</a>()</code> <br><code><a href="layer_cropping_3d.html">layer_cropping_3d</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_dense.html">layer_dense</a>()</code> <br><code><a href="layer_depthwise_conv_1d.html">layer_depthwise_conv_1d</a>()</code> <br><code><a href="layer_depthwise_conv_2d.html">layer_depthwise_conv_2d</a>()</code> <br><code><a href="layer_discretization.html">layer_discretization</a>()</code> <br><code><a href="layer_dot.html">layer_dot</a>()</code> <br><code><a href="layer_dropout.html">layer_dropout</a>()</code> <br><code><a href="layer_einsum_dense.html">layer_einsum_dense</a>()</code> <br><code><a href="layer_embedding.html">layer_embedding</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_feature_space.html">layer_feature_space</a>()</code> <br><code><a href="layer_flatten.html">layer_flatten</a>()</code> <br><code><a href="layer_flax_module_wrapper.html">layer_flax_module_wrapper</a>()</code> <br><code><a href="layer_gaussian_dropout.html">layer_gaussian_dropout</a>()</code> <br><code><a href="layer_gaussian_noise.html">layer_gaussian_noise</a>()</code> <br><code><a href="layer_global_average_pooling_1d.html">layer_global_average_pooling_1d</a>()</code> <br><code><a href="layer_global_average_pooling_2d.html">layer_global_average_pooling_2d</a>()</code> <br><code><a href="layer_global_average_pooling_3d.html">layer_global_average_pooling_3d</a>()</code> <br><code><a href="layer_global_max_pooling_1d.html">layer_global_max_pooling_1d</a>()</code> <br><code><a href="layer_global_max_pooling_2d.html">layer_global_max_pooling_2d</a>()</code> <br><code><a href="layer_global_max_pooling_3d.html">layer_global_max_pooling_3d</a>()</code> <br><code><a href="layer_group_normalization.html">layer_group_normalization</a>()</code> <br><code><a href="layer_group_query_attention.html">layer_group_query_attention</a>()</code> <br><code><a href="layer_gru.html">layer_gru</a>()</code> <br><code><a href="layer_hashed_crossing.html">layer_hashed_crossing</a>()</code> <br><code><a href="layer_hashing.html">layer_hashing</a>()</code> <br><code><a href="layer_identity.html">layer_identity</a>()</code> <br><code><a href="layer_integer_lookup.html">layer_integer_lookup</a>()</code> <br><code><a href="layer_jax_model_wrapper.html">layer_jax_model_wrapper</a>()</code> <br><code><a href="layer_lambda.html">layer_lambda</a>()</code> <br><code><a href="layer_layer_normalization.html">layer_layer_normalization</a>()</code> <br><code><a href="layer_lstm.html">layer_lstm</a>()</code> <br><code><a href="layer_masking.html">layer_masking</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_max_pooling_1d.html">layer_max_pooling_1d</a>()</code> <br><code><a href="layer_max_pooling_2d.html">layer_max_pooling_2d</a>()</code> <br><code><a href="layer_max_pooling_3d.html">layer_max_pooling_3d</a>()</code> <br><code><a href="layer_maximum.html">layer_maximum</a>()</code> <br><code><a href="layer_mel_spectrogram.html">layer_mel_spectrogram</a>()</code> <br><code><a href="layer_minimum.html">layer_minimum</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_multi_head_attention.html">layer_multi_head_attention</a>()</code> <br><code><a href="layer_multiply.html">layer_multiply</a>()</code> <br><code><a href="layer_normalization.html">layer_normalization</a>()</code> <br><code><a href="layer_permute.html">layer_permute</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_brightness.html">layer_random_brightness</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_contrast.html">layer_random_contrast</a>()</code> <br><code><a href="layer_random_crop.html">layer_random_crop</a>()</code> <br><code><a href="layer_random_elastic_transform.html">layer_random_elastic_transform</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_flip.html">layer_random_flip</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_rotation.html">layer_random_rotation</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_random_translation.html">layer_random_translation</a>()</code> <br><code><a href="layer_random_zoom.html">layer_random_zoom</a>()</code> <br><code><a href="layer_repeat_vector.html">layer_repeat_vector</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_reshape.html">layer_reshape</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_rms_normalization.html">layer_rms_normalization</a>()</code> <br><code><a href="layer_rnn.html">layer_rnn</a>()</code> <br><code><a href="layer_separable_conv_1d.html">layer_separable_conv_1d</a>()</code> <br><code><a href="layer_separable_conv_2d.html">layer_separable_conv_2d</a>()</code> <br><code><a href="layer_simple_rnn.html">layer_simple_rnn</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br><code><a href="layer_spatial_dropout_1d.html">layer_spatial_dropout_1d</a>()</code> <br><code><a href="layer_spatial_dropout_2d.html">layer_spatial_dropout_2d</a>()</code> <br><code><a href="layer_spatial_dropout_3d.html">layer_spatial_dropout_3d</a>()</code> <br><code><a href="layer_spectral_normalization.html">layer_spectral_normalization</a>()</code> <br><code><a href="layer_stft_spectrogram.html">layer_stft_spectrogram</a>()</code> <br><code><a href="layer_string_lookup.html">layer_string_lookup</a>()</code> <br><code><a href="layer_subtract.html">layer_subtract</a>()</code> <br><code><a href="layer_text_vectorization.html">layer_text_vectorization</a>()</code> <br><code><a href="layer_time_distributed.html">layer_time_distributed</a>()</code> <br><code><a href="layer_torch_module_wrapper.html">layer_torch_module_wrapper</a>()</code> <br><code><a href="layer_unit_normalization.html">layer_unit_normalization</a>()</code> <br><code><a href="layer_upsampling_1d.html">layer_upsampling_1d</a>()</code> <br><code><a href="layer_upsampling_2d.html">layer_upsampling_2d</a>()</code> <br><code><a href="layer_upsampling_3d.html">layer_upsampling_3d</a>()</code> <br><code><a href="layer_zero_padding_1d.html">layer_zero_padding_1d</a>()</code> <br><code><a href="layer_zero_padding_2d.html">layer_zero_padding_2d</a>()</code> <br><code><a href="layer_zero_padding_3d.html">layer_zero_padding_3d</a>()</code> <br><code><a href="rnn_cell_gru.html">rnn_cell_gru</a>()</code> <br><code><a href="rnn_cell_lstm.html">rnn_cell_lstm</a>()</code> <br><code><a href="rnn_cell_simple.html">rnn_cell_simple</a>()</code> <br><code><a href="rnn_cells_stack.html">rnn_cells_stack</a>()</code> <br></p>
 <p>Other saving and loading functions: <br><code><a href="export_savedmodel.keras.src.models.model.Model.html">export_savedmodel.keras.src.models.model.Model</a>()</code> <br><code><a href="load_model.html">load_model</a>()</code> <br><code><a href="load_model_weights.html">load_model_weights</a>()</code> <br><code><a href="register_keras_serializable.html">register_keras_serializable</a>()</code> <br><code><a href="save_model.html">save_model</a>()</code> <br><code><a href="save_model_config.html">save_model_config</a>()</code> <br><code><a href="save_model_weights.html">save_model_weights</a>()</code> <br><code><a href="with_custom_object_scope.html">with_custom_object_scope</a>()</code> <br></p></div>
     </div>
 
@@ -164,7 +164,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/layer_time_distributed.html b/docs/dev/reference/layer_time_distributed.html
index bb21ec9ee..ed8105be6 100644
--- a/docs/dev/reference/layer_time_distributed.html
+++ b/docs/dev/reference/layer_time_distributed.html
@@ -40,7 +40,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -150,7 +150,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
     <div class="dont-index">
 <ul><li><p><a href="https://keras.io/api/layers/recurrent_layers/time_distributed#timedistributed-class" class="external-link">https://keras.io/api/layers/recurrent_layers/time_distributed#timedistributed-class</a></p></li>
 </ul><p>Other rnn layers: <br><code><a href="layer_bidirectional.html">layer_bidirectional</a>()</code> <br><code><a href="layer_conv_lstm_1d.html">layer_conv_lstm_1d</a>()</code> <br><code><a href="layer_conv_lstm_2d.html">layer_conv_lstm_2d</a>()</code> <br><code><a href="layer_conv_lstm_3d.html">layer_conv_lstm_3d</a>()</code> <br><code><a href="layer_gru.html">layer_gru</a>()</code> <br><code><a href="layer_lstm.html">layer_lstm</a>()</code> <br><code><a href="layer_rnn.html">layer_rnn</a>()</code> <br><code><a href="layer_simple_rnn.html">layer_simple_rnn</a>()</code> <br><code><a href="rnn_cell_gru.html">rnn_cell_gru</a>()</code> <br><code><a href="rnn_cell_lstm.html">rnn_cell_lstm</a>()</code> <br><code><a href="rnn_cell_simple.html">rnn_cell_simple</a>()</code> <br><code><a href="rnn_cells_stack.html">rnn_cells_stack</a>()</code> <br></p>
-<p>Other layers: <br><code><a href="Layer.html">Layer</a>()</code> <br><code><a href="layer_activation.html">layer_activation</a>()</code> <br><code><a href="layer_activation_elu.html">layer_activation_elu</a>()</code> <br><code><a href="layer_activation_leaky_relu.html">layer_activation_leaky_relu</a>()</code> <br><code><a href="layer_activation_parametric_relu.html">layer_activation_parametric_relu</a>()</code> <br><code><a href="layer_activation_relu.html">layer_activation_relu</a>()</code> <br><code><a href="layer_activation_softmax.html">layer_activation_softmax</a>()</code> <br><code><a href="layer_activity_regularization.html">layer_activity_regularization</a>()</code> <br><code><a href="layer_add.html">layer_add</a>()</code> <br><code><a href="layer_additive_attention.html">layer_additive_attention</a>()</code> <br><code><a href="layer_alpha_dropout.html">layer_alpha_dropout</a>()</code> <br><code><a href="layer_attention.html">layer_attention</a>()</code> <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_average.html">layer_average</a>()</code> <br><code><a href="layer_average_pooling_1d.html">layer_average_pooling_1d</a>()</code> <br><code><a href="layer_average_pooling_2d.html">layer_average_pooling_2d</a>()</code> <br><code><a href="layer_average_pooling_3d.html">layer_average_pooling_3d</a>()</code> <br><code><a href="layer_batch_normalization.html">layer_batch_normalization</a>()</code> <br><code><a href="layer_bidirectional.html">layer_bidirectional</a>()</code> <br><code><a href="layer_category_encoding.html">layer_category_encoding</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_concatenate.html">layer_concatenate</a>()</code> <br><code><a href="layer_conv_1d.html">layer_conv_1d</a>()</code> <br><code><a href="layer_conv_1d_transpose.html">layer_conv_1d_transpose</a>()</code> <br><code><a href="layer_conv_2d.html">layer_conv_2d</a>()</code> <br><code><a href="layer_conv_2d_transpose.html">layer_conv_2d_transpose</a>()</code> <br><code><a href="layer_conv_3d.html">layer_conv_3d</a>()</code> <br><code><a href="layer_conv_3d_transpose.html">layer_conv_3d_transpose</a>()</code> <br><code><a href="layer_conv_lstm_1d.html">layer_conv_lstm_1d</a>()</code> <br><code><a href="layer_conv_lstm_2d.html">layer_conv_lstm_2d</a>()</code> <br><code><a href="layer_conv_lstm_3d.html">layer_conv_lstm_3d</a>()</code> <br><code><a href="layer_cropping_1d.html">layer_cropping_1d</a>()</code> <br><code><a href="layer_cropping_2d.html">layer_cropping_2d</a>()</code> <br><code><a href="layer_cropping_3d.html">layer_cropping_3d</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_dense.html">layer_dense</a>()</code> <br><code><a href="layer_depthwise_conv_1d.html">layer_depthwise_conv_1d</a>()</code> <br><code><a href="layer_depthwise_conv_2d.html">layer_depthwise_conv_2d</a>()</code> <br><code><a href="layer_discretization.html">layer_discretization</a>()</code> <br><code><a href="layer_dot.html">layer_dot</a>()</code> <br><code><a href="layer_dropout.html">layer_dropout</a>()</code> <br><code><a href="layer_einsum_dense.html">layer_einsum_dense</a>()</code> <br><code><a href="layer_embedding.html">layer_embedding</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_feature_space.html">layer_feature_space</a>()</code> <br><code><a href="layer_flatten.html">layer_flatten</a>()</code> <br><code><a href="layer_flax_module_wrapper.html">layer_flax_module_wrapper</a>()</code> <br><code><a href="layer_gaussian_dropout.html">layer_gaussian_dropout</a>()</code> <br><code><a href="layer_gaussian_noise.html">layer_gaussian_noise</a>()</code> <br><code><a href="layer_global_average_pooling_1d.html">layer_global_average_pooling_1d</a>()</code> <br><code><a href="layer_global_average_pooling_2d.html">layer_global_average_pooling_2d</a>()</code> <br><code><a href="layer_global_average_pooling_3d.html">layer_global_average_pooling_3d</a>()</code> <br><code><a href="layer_global_max_pooling_1d.html">layer_global_max_pooling_1d</a>()</code> <br><code><a href="layer_global_max_pooling_2d.html">layer_global_max_pooling_2d</a>()</code> <br><code><a href="layer_global_max_pooling_3d.html">layer_global_max_pooling_3d</a>()</code> <br><code><a href="layer_group_normalization.html">layer_group_normalization</a>()</code> <br><code><a href="layer_group_query_attention.html">layer_group_query_attention</a>()</code> <br><code><a href="layer_gru.html">layer_gru</a>()</code> <br><code><a href="layer_hashed_crossing.html">layer_hashed_crossing</a>()</code> <br><code><a href="layer_hashing.html">layer_hashing</a>()</code> <br><code><a href="layer_identity.html">layer_identity</a>()</code> <br><code><a href="layer_integer_lookup.html">layer_integer_lookup</a>()</code> <br><code><a href="layer_jax_model_wrapper.html">layer_jax_model_wrapper</a>()</code> <br><code><a href="layer_lambda.html">layer_lambda</a>()</code> <br><code><a href="layer_layer_normalization.html">layer_layer_normalization</a>()</code> <br><code><a href="layer_lstm.html">layer_lstm</a>()</code> <br><code><a href="layer_masking.html">layer_masking</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_max_pooling_1d.html">layer_max_pooling_1d</a>()</code> <br><code><a href="layer_max_pooling_2d.html">layer_max_pooling_2d</a>()</code> <br><code><a href="layer_max_pooling_3d.html">layer_max_pooling_3d</a>()</code> <br><code><a href="layer_maximum.html">layer_maximum</a>()</code> <br><code><a href="layer_mel_spectrogram.html">layer_mel_spectrogram</a>()</code> <br><code><a href="layer_minimum.html">layer_minimum</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_multi_head_attention.html">layer_multi_head_attention</a>()</code> <br><code><a href="layer_multiply.html">layer_multiply</a>()</code> <br><code><a href="layer_normalization.html">layer_normalization</a>()</code> <br><code><a href="layer_permute.html">layer_permute</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_brightness.html">layer_random_brightness</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_contrast.html">layer_random_contrast</a>()</code> <br><code><a href="layer_random_crop.html">layer_random_crop</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_flip.html">layer_random_flip</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_rotation.html">layer_random_rotation</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_random_translation.html">layer_random_translation</a>()</code> <br><code><a href="layer_random_zoom.html">layer_random_zoom</a>()</code> <br><code><a href="layer_repeat_vector.html">layer_repeat_vector</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_reshape.html">layer_reshape</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_rms_normalization.html">layer_rms_normalization</a>()</code> <br><code><a href="layer_rnn.html">layer_rnn</a>()</code> <br><code><a href="layer_separable_conv_1d.html">layer_separable_conv_1d</a>()</code> <br><code><a href="layer_separable_conv_2d.html">layer_separable_conv_2d</a>()</code> <br><code><a href="layer_simple_rnn.html">layer_simple_rnn</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br><code><a href="layer_spatial_dropout_1d.html">layer_spatial_dropout_1d</a>()</code> <br><code><a href="layer_spatial_dropout_2d.html">layer_spatial_dropout_2d</a>()</code> <br><code><a href="layer_spatial_dropout_3d.html">layer_spatial_dropout_3d</a>()</code> <br><code><a href="layer_spectral_normalization.html">layer_spectral_normalization</a>()</code> <br><code><a href="layer_stft_spectrogram.html">layer_stft_spectrogram</a>()</code> <br><code><a href="layer_string_lookup.html">layer_string_lookup</a>()</code> <br><code><a href="layer_subtract.html">layer_subtract</a>()</code> <br><code><a href="layer_text_vectorization.html">layer_text_vectorization</a>()</code> <br><code><a href="layer_tfsm.html">layer_tfsm</a>()</code> <br><code><a href="layer_torch_module_wrapper.html">layer_torch_module_wrapper</a>()</code> <br><code><a href="layer_unit_normalization.html">layer_unit_normalization</a>()</code> <br><code><a href="layer_upsampling_1d.html">layer_upsampling_1d</a>()</code> <br><code><a href="layer_upsampling_2d.html">layer_upsampling_2d</a>()</code> <br><code><a href="layer_upsampling_3d.html">layer_upsampling_3d</a>()</code> <br><code><a href="layer_zero_padding_1d.html">layer_zero_padding_1d</a>()</code> <br><code><a href="layer_zero_padding_2d.html">layer_zero_padding_2d</a>()</code> <br><code><a href="layer_zero_padding_3d.html">layer_zero_padding_3d</a>()</code> <br><code><a href="rnn_cell_gru.html">rnn_cell_gru</a>()</code> <br><code><a href="rnn_cell_lstm.html">rnn_cell_lstm</a>()</code> <br><code><a href="rnn_cell_simple.html">rnn_cell_simple</a>()</code> <br><code><a href="rnn_cells_stack.html">rnn_cells_stack</a>()</code> <br></p></div>
+<p>Other layers: <br><code><a href="Layer.html">Layer</a>()</code> <br><code><a href="layer_activation.html">layer_activation</a>()</code> <br><code><a href="layer_activation_elu.html">layer_activation_elu</a>()</code> <br><code><a href="layer_activation_leaky_relu.html">layer_activation_leaky_relu</a>()</code> <br><code><a href="layer_activation_parametric_relu.html">layer_activation_parametric_relu</a>()</code> <br><code><a href="layer_activation_relu.html">layer_activation_relu</a>()</code> <br><code><a href="layer_activation_softmax.html">layer_activation_softmax</a>()</code> <br><code><a href="layer_activity_regularization.html">layer_activity_regularization</a>()</code> <br><code><a href="layer_add.html">layer_add</a>()</code> <br><code><a href="layer_additive_attention.html">layer_additive_attention</a>()</code> <br><code><a href="layer_alpha_dropout.html">layer_alpha_dropout</a>()</code> <br><code><a href="layer_attention.html">layer_attention</a>()</code> <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_average.html">layer_average</a>()</code> <br><code><a href="layer_average_pooling_1d.html">layer_average_pooling_1d</a>()</code> <br><code><a href="layer_average_pooling_2d.html">layer_average_pooling_2d</a>()</code> <br><code><a href="layer_average_pooling_3d.html">layer_average_pooling_3d</a>()</code> <br><code><a href="layer_batch_normalization.html">layer_batch_normalization</a>()</code> <br><code><a href="layer_bidirectional.html">layer_bidirectional</a>()</code> <br><code><a href="layer_category_encoding.html">layer_category_encoding</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_concatenate.html">layer_concatenate</a>()</code> <br><code><a href="layer_conv_1d.html">layer_conv_1d</a>()</code> <br><code><a href="layer_conv_1d_transpose.html">layer_conv_1d_transpose</a>()</code> <br><code><a href="layer_conv_2d.html">layer_conv_2d</a>()</code> <br><code><a href="layer_conv_2d_transpose.html">layer_conv_2d_transpose</a>()</code> <br><code><a href="layer_conv_3d.html">layer_conv_3d</a>()</code> <br><code><a href="layer_conv_3d_transpose.html">layer_conv_3d_transpose</a>()</code> <br><code><a href="layer_conv_lstm_1d.html">layer_conv_lstm_1d</a>()</code> <br><code><a href="layer_conv_lstm_2d.html">layer_conv_lstm_2d</a>()</code> <br><code><a href="layer_conv_lstm_3d.html">layer_conv_lstm_3d</a>()</code> <br><code><a href="layer_cropping_1d.html">layer_cropping_1d</a>()</code> <br><code><a href="layer_cropping_2d.html">layer_cropping_2d</a>()</code> <br><code><a href="layer_cropping_3d.html">layer_cropping_3d</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_dense.html">layer_dense</a>()</code> <br><code><a href="layer_depthwise_conv_1d.html">layer_depthwise_conv_1d</a>()</code> <br><code><a href="layer_depthwise_conv_2d.html">layer_depthwise_conv_2d</a>()</code> <br><code><a href="layer_discretization.html">layer_discretization</a>()</code> <br><code><a href="layer_dot.html">layer_dot</a>()</code> <br><code><a href="layer_dropout.html">layer_dropout</a>()</code> <br><code><a href="layer_einsum_dense.html">layer_einsum_dense</a>()</code> <br><code><a href="layer_embedding.html">layer_embedding</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_feature_space.html">layer_feature_space</a>()</code> <br><code><a href="layer_flatten.html">layer_flatten</a>()</code> <br><code><a href="layer_flax_module_wrapper.html">layer_flax_module_wrapper</a>()</code> <br><code><a href="layer_gaussian_dropout.html">layer_gaussian_dropout</a>()</code> <br><code><a href="layer_gaussian_noise.html">layer_gaussian_noise</a>()</code> <br><code><a href="layer_global_average_pooling_1d.html">layer_global_average_pooling_1d</a>()</code> <br><code><a href="layer_global_average_pooling_2d.html">layer_global_average_pooling_2d</a>()</code> <br><code><a href="layer_global_average_pooling_3d.html">layer_global_average_pooling_3d</a>()</code> <br><code><a href="layer_global_max_pooling_1d.html">layer_global_max_pooling_1d</a>()</code> <br><code><a href="layer_global_max_pooling_2d.html">layer_global_max_pooling_2d</a>()</code> <br><code><a href="layer_global_max_pooling_3d.html">layer_global_max_pooling_3d</a>()</code> <br><code><a href="layer_group_normalization.html">layer_group_normalization</a>()</code> <br><code><a href="layer_group_query_attention.html">layer_group_query_attention</a>()</code> <br><code><a href="layer_gru.html">layer_gru</a>()</code> <br><code><a href="layer_hashed_crossing.html">layer_hashed_crossing</a>()</code> <br><code><a href="layer_hashing.html">layer_hashing</a>()</code> <br><code><a href="layer_identity.html">layer_identity</a>()</code> <br><code><a href="layer_integer_lookup.html">layer_integer_lookup</a>()</code> <br><code><a href="layer_jax_model_wrapper.html">layer_jax_model_wrapper</a>()</code> <br><code><a href="layer_lambda.html">layer_lambda</a>()</code> <br><code><a href="layer_layer_normalization.html">layer_layer_normalization</a>()</code> <br><code><a href="layer_lstm.html">layer_lstm</a>()</code> <br><code><a href="layer_masking.html">layer_masking</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_max_pooling_1d.html">layer_max_pooling_1d</a>()</code> <br><code><a href="layer_max_pooling_2d.html">layer_max_pooling_2d</a>()</code> <br><code><a href="layer_max_pooling_3d.html">layer_max_pooling_3d</a>()</code> <br><code><a href="layer_maximum.html">layer_maximum</a>()</code> <br><code><a href="layer_mel_spectrogram.html">layer_mel_spectrogram</a>()</code> <br><code><a href="layer_minimum.html">layer_minimum</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_multi_head_attention.html">layer_multi_head_attention</a>()</code> <br><code><a href="layer_multiply.html">layer_multiply</a>()</code> <br><code><a href="layer_normalization.html">layer_normalization</a>()</code> <br><code><a href="layer_permute.html">layer_permute</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_brightness.html">layer_random_brightness</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_contrast.html">layer_random_contrast</a>()</code> <br><code><a href="layer_random_crop.html">layer_random_crop</a>()</code> <br><code><a href="layer_random_elastic_transform.html">layer_random_elastic_transform</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_flip.html">layer_random_flip</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_rotation.html">layer_random_rotation</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_random_translation.html">layer_random_translation</a>()</code> <br><code><a href="layer_random_zoom.html">layer_random_zoom</a>()</code> <br><code><a href="layer_repeat_vector.html">layer_repeat_vector</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_reshape.html">layer_reshape</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_rms_normalization.html">layer_rms_normalization</a>()</code> <br><code><a href="layer_rnn.html">layer_rnn</a>()</code> <br><code><a href="layer_separable_conv_1d.html">layer_separable_conv_1d</a>()</code> <br><code><a href="layer_separable_conv_2d.html">layer_separable_conv_2d</a>()</code> <br><code><a href="layer_simple_rnn.html">layer_simple_rnn</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br><code><a href="layer_spatial_dropout_1d.html">layer_spatial_dropout_1d</a>()</code> <br><code><a href="layer_spatial_dropout_2d.html">layer_spatial_dropout_2d</a>()</code> <br><code><a href="layer_spatial_dropout_3d.html">layer_spatial_dropout_3d</a>()</code> <br><code><a href="layer_spectral_normalization.html">layer_spectral_normalization</a>()</code> <br><code><a href="layer_stft_spectrogram.html">layer_stft_spectrogram</a>()</code> <br><code><a href="layer_string_lookup.html">layer_string_lookup</a>()</code> <br><code><a href="layer_subtract.html">layer_subtract</a>()</code> <br><code><a href="layer_text_vectorization.html">layer_text_vectorization</a>()</code> <br><code><a href="layer_tfsm.html">layer_tfsm</a>()</code> <br><code><a href="layer_torch_module_wrapper.html">layer_torch_module_wrapper</a>()</code> <br><code><a href="layer_unit_normalization.html">layer_unit_normalization</a>()</code> <br><code><a href="layer_upsampling_1d.html">layer_upsampling_1d</a>()</code> <br><code><a href="layer_upsampling_2d.html">layer_upsampling_2d</a>()</code> <br><code><a href="layer_upsampling_3d.html">layer_upsampling_3d</a>()</code> <br><code><a href="layer_zero_padding_1d.html">layer_zero_padding_1d</a>()</code> <br><code><a href="layer_zero_padding_2d.html">layer_zero_padding_2d</a>()</code> <br><code><a href="layer_zero_padding_3d.html">layer_zero_padding_3d</a>()</code> <br><code><a href="rnn_cell_gru.html">rnn_cell_gru</a>()</code> <br><code><a href="rnn_cell_lstm.html">rnn_cell_lstm</a>()</code> <br><code><a href="rnn_cell_simple.html">rnn_cell_simple</a>()</code> <br><code><a href="rnn_cells_stack.html">rnn_cells_stack</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -162,7 +162,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/layer_torch_module_wrapper.html b/docs/dev/reference/layer_torch_module_wrapper.html
index ed0bc484b..8dd85b8f4 100644
--- a/docs/dev/reference/layer_torch_module_wrapper.html
+++ b/docs/dev/reference/layer_torch_module_wrapper.html
@@ -16,7 +16,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -71,7 +71,13 @@ <h1>Torch module wrapper layer.</h1>
 
     <div class="section level2">
     <h2 id="ref-usage">Usage<a class="anchor" aria-label="anchor" href="#ref-usage"></a></h2>
-    <div class="sourceCode"><pre class="sourceCode r"><code><span><span class="fu">layer_torch_module_wrapper</span><span class="op">(</span><span class="va">object</span>, <span class="va">module</span>, name <span class="op">=</span> <span class="cn">NULL</span>, <span class="va">...</span><span class="op">)</span></span></code></pre></div>
+    <div class="sourceCode"><pre class="sourceCode r"><code><span><span class="fu">layer_torch_module_wrapper</span><span class="op">(</span></span>
+<span>  <span class="va">object</span>,</span>
+<span>  <span class="va">module</span>,</span>
+<span>  output_shape <span class="op">=</span> <span class="cn">NULL</span>,</span>
+<span>  name <span class="op">=</span> <span class="cn">NULL</span>,</span>
+<span>  <span class="va">...</span></span>
+<span><span class="op">)</span></span></code></pre></div>
     </div>
 
     <div class="section level2">
@@ -89,6 +95,10 @@ <h2 id="arguments">Arguments<a class="anchor" aria-label="anchor" href="#argumen
 it once).</p></dd>
 
 
+<dt id="arg-output-shape">output_shape<a class="anchor" aria-label="anchor" href="#arg-output-shape"></a></dt>
+<dd><p>Shape of the output from this layer. Helps Keras infer shapes.</p></dd>
+
+
 <dt id="arg-name">name<a class="anchor" aria-label="anchor" href="#arg-name"></a></dt>
 <dd><p>The name of the layer (string).</p></dd>
 
@@ -166,7 +176,7 @@ <h2 id="example">Example<a class="anchor" aria-label="anchor" href="#example"></
     <div class="section level2">
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
     <div class="dont-index"><p>Other wrapping layers: <br><code><a href="layer_flax_module_wrapper.html">layer_flax_module_wrapper</a>()</code> <br><code><a href="layer_jax_model_wrapper.html">layer_jax_model_wrapper</a>()</code> <br></p>
-<p>Other layers: <br><code><a href="Layer.html">Layer</a>()</code> <br><code><a href="layer_activation.html">layer_activation</a>()</code> <br><code><a href="layer_activation_elu.html">layer_activation_elu</a>()</code> <br><code><a href="layer_activation_leaky_relu.html">layer_activation_leaky_relu</a>()</code> <br><code><a href="layer_activation_parametric_relu.html">layer_activation_parametric_relu</a>()</code> <br><code><a href="layer_activation_relu.html">layer_activation_relu</a>()</code> <br><code><a href="layer_activation_softmax.html">layer_activation_softmax</a>()</code> <br><code><a href="layer_activity_regularization.html">layer_activity_regularization</a>()</code> <br><code><a href="layer_add.html">layer_add</a>()</code> <br><code><a href="layer_additive_attention.html">layer_additive_attention</a>()</code> <br><code><a href="layer_alpha_dropout.html">layer_alpha_dropout</a>()</code> <br><code><a href="layer_attention.html">layer_attention</a>()</code> <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_average.html">layer_average</a>()</code> <br><code><a href="layer_average_pooling_1d.html">layer_average_pooling_1d</a>()</code> <br><code><a href="layer_average_pooling_2d.html">layer_average_pooling_2d</a>()</code> <br><code><a href="layer_average_pooling_3d.html">layer_average_pooling_3d</a>()</code> <br><code><a href="layer_batch_normalization.html">layer_batch_normalization</a>()</code> <br><code><a href="layer_bidirectional.html">layer_bidirectional</a>()</code> <br><code><a href="layer_category_encoding.html">layer_category_encoding</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_concatenate.html">layer_concatenate</a>()</code> <br><code><a href="layer_conv_1d.html">layer_conv_1d</a>()</code> <br><code><a href="layer_conv_1d_transpose.html">layer_conv_1d_transpose</a>()</code> <br><code><a href="layer_conv_2d.html">layer_conv_2d</a>()</code> <br><code><a href="layer_conv_2d_transpose.html">layer_conv_2d_transpose</a>()</code> <br><code><a href="layer_conv_3d.html">layer_conv_3d</a>()</code> <br><code><a href="layer_conv_3d_transpose.html">layer_conv_3d_transpose</a>()</code> <br><code><a href="layer_conv_lstm_1d.html">layer_conv_lstm_1d</a>()</code> <br><code><a href="layer_conv_lstm_2d.html">layer_conv_lstm_2d</a>()</code> <br><code><a href="layer_conv_lstm_3d.html">layer_conv_lstm_3d</a>()</code> <br><code><a href="layer_cropping_1d.html">layer_cropping_1d</a>()</code> <br><code><a href="layer_cropping_2d.html">layer_cropping_2d</a>()</code> <br><code><a href="layer_cropping_3d.html">layer_cropping_3d</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_dense.html">layer_dense</a>()</code> <br><code><a href="layer_depthwise_conv_1d.html">layer_depthwise_conv_1d</a>()</code> <br><code><a href="layer_depthwise_conv_2d.html">layer_depthwise_conv_2d</a>()</code> <br><code><a href="layer_discretization.html">layer_discretization</a>()</code> <br><code><a href="layer_dot.html">layer_dot</a>()</code> <br><code><a href="layer_dropout.html">layer_dropout</a>()</code> <br><code><a href="layer_einsum_dense.html">layer_einsum_dense</a>()</code> <br><code><a href="layer_embedding.html">layer_embedding</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_feature_space.html">layer_feature_space</a>()</code> <br><code><a href="layer_flatten.html">layer_flatten</a>()</code> <br><code><a href="layer_flax_module_wrapper.html">layer_flax_module_wrapper</a>()</code> <br><code><a href="layer_gaussian_dropout.html">layer_gaussian_dropout</a>()</code> <br><code><a href="layer_gaussian_noise.html">layer_gaussian_noise</a>()</code> <br><code><a href="layer_global_average_pooling_1d.html">layer_global_average_pooling_1d</a>()</code> <br><code><a href="layer_global_average_pooling_2d.html">layer_global_average_pooling_2d</a>()</code> <br><code><a href="layer_global_average_pooling_3d.html">layer_global_average_pooling_3d</a>()</code> <br><code><a href="layer_global_max_pooling_1d.html">layer_global_max_pooling_1d</a>()</code> <br><code><a href="layer_global_max_pooling_2d.html">layer_global_max_pooling_2d</a>()</code> <br><code><a href="layer_global_max_pooling_3d.html">layer_global_max_pooling_3d</a>()</code> <br><code><a href="layer_group_normalization.html">layer_group_normalization</a>()</code> <br><code><a href="layer_group_query_attention.html">layer_group_query_attention</a>()</code> <br><code><a href="layer_gru.html">layer_gru</a>()</code> <br><code><a href="layer_hashed_crossing.html">layer_hashed_crossing</a>()</code> <br><code><a href="layer_hashing.html">layer_hashing</a>()</code> <br><code><a href="layer_identity.html">layer_identity</a>()</code> <br><code><a href="layer_integer_lookup.html">layer_integer_lookup</a>()</code> <br><code><a href="layer_jax_model_wrapper.html">layer_jax_model_wrapper</a>()</code> <br><code><a href="layer_lambda.html">layer_lambda</a>()</code> <br><code><a href="layer_layer_normalization.html">layer_layer_normalization</a>()</code> <br><code><a href="layer_lstm.html">layer_lstm</a>()</code> <br><code><a href="layer_masking.html">layer_masking</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_max_pooling_1d.html">layer_max_pooling_1d</a>()</code> <br><code><a href="layer_max_pooling_2d.html">layer_max_pooling_2d</a>()</code> <br><code><a href="layer_max_pooling_3d.html">layer_max_pooling_3d</a>()</code> <br><code><a href="layer_maximum.html">layer_maximum</a>()</code> <br><code><a href="layer_mel_spectrogram.html">layer_mel_spectrogram</a>()</code> <br><code><a href="layer_minimum.html">layer_minimum</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_multi_head_attention.html">layer_multi_head_attention</a>()</code> <br><code><a href="layer_multiply.html">layer_multiply</a>()</code> <br><code><a href="layer_normalization.html">layer_normalization</a>()</code> <br><code><a href="layer_permute.html">layer_permute</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_brightness.html">layer_random_brightness</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_contrast.html">layer_random_contrast</a>()</code> <br><code><a href="layer_random_crop.html">layer_random_crop</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_flip.html">layer_random_flip</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_rotation.html">layer_random_rotation</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_random_translation.html">layer_random_translation</a>()</code> <br><code><a href="layer_random_zoom.html">layer_random_zoom</a>()</code> <br><code><a href="layer_repeat_vector.html">layer_repeat_vector</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_reshape.html">layer_reshape</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_rms_normalization.html">layer_rms_normalization</a>()</code> <br><code><a href="layer_rnn.html">layer_rnn</a>()</code> <br><code><a href="layer_separable_conv_1d.html">layer_separable_conv_1d</a>()</code> <br><code><a href="layer_separable_conv_2d.html">layer_separable_conv_2d</a>()</code> <br><code><a href="layer_simple_rnn.html">layer_simple_rnn</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br><code><a href="layer_spatial_dropout_1d.html">layer_spatial_dropout_1d</a>()</code> <br><code><a href="layer_spatial_dropout_2d.html">layer_spatial_dropout_2d</a>()</code> <br><code><a href="layer_spatial_dropout_3d.html">layer_spatial_dropout_3d</a>()</code> <br><code><a href="layer_spectral_normalization.html">layer_spectral_normalization</a>()</code> <br><code><a href="layer_stft_spectrogram.html">layer_stft_spectrogram</a>()</code> <br><code><a href="layer_string_lookup.html">layer_string_lookup</a>()</code> <br><code><a href="layer_subtract.html">layer_subtract</a>()</code> <br><code><a href="layer_text_vectorization.html">layer_text_vectorization</a>()</code> <br><code><a href="layer_tfsm.html">layer_tfsm</a>()</code> <br><code><a href="layer_time_distributed.html">layer_time_distributed</a>()</code> <br><code><a href="layer_unit_normalization.html">layer_unit_normalization</a>()</code> <br><code><a href="layer_upsampling_1d.html">layer_upsampling_1d</a>()</code> <br><code><a href="layer_upsampling_2d.html">layer_upsampling_2d</a>()</code> <br><code><a href="layer_upsampling_3d.html">layer_upsampling_3d</a>()</code> <br><code><a href="layer_zero_padding_1d.html">layer_zero_padding_1d</a>()</code> <br><code><a href="layer_zero_padding_2d.html">layer_zero_padding_2d</a>()</code> <br><code><a href="layer_zero_padding_3d.html">layer_zero_padding_3d</a>()</code> <br><code><a href="rnn_cell_gru.html">rnn_cell_gru</a>()</code> <br><code><a href="rnn_cell_lstm.html">rnn_cell_lstm</a>()</code> <br><code><a href="rnn_cell_simple.html">rnn_cell_simple</a>()</code> <br><code><a href="rnn_cells_stack.html">rnn_cells_stack</a>()</code> <br></p></div>
+<p>Other layers: <br><code><a href="Layer.html">Layer</a>()</code> <br><code><a href="layer_activation.html">layer_activation</a>()</code> <br><code><a href="layer_activation_elu.html">layer_activation_elu</a>()</code> <br><code><a href="layer_activation_leaky_relu.html">layer_activation_leaky_relu</a>()</code> <br><code><a href="layer_activation_parametric_relu.html">layer_activation_parametric_relu</a>()</code> <br><code><a href="layer_activation_relu.html">layer_activation_relu</a>()</code> <br><code><a href="layer_activation_softmax.html">layer_activation_softmax</a>()</code> <br><code><a href="layer_activity_regularization.html">layer_activity_regularization</a>()</code> <br><code><a href="layer_add.html">layer_add</a>()</code> <br><code><a href="layer_additive_attention.html">layer_additive_attention</a>()</code> <br><code><a href="layer_alpha_dropout.html">layer_alpha_dropout</a>()</code> <br><code><a href="layer_attention.html">layer_attention</a>()</code> <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_average.html">layer_average</a>()</code> <br><code><a href="layer_average_pooling_1d.html">layer_average_pooling_1d</a>()</code> <br><code><a href="layer_average_pooling_2d.html">layer_average_pooling_2d</a>()</code> <br><code><a href="layer_average_pooling_3d.html">layer_average_pooling_3d</a>()</code> <br><code><a href="layer_batch_normalization.html">layer_batch_normalization</a>()</code> <br><code><a href="layer_bidirectional.html">layer_bidirectional</a>()</code> <br><code><a href="layer_category_encoding.html">layer_category_encoding</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_concatenate.html">layer_concatenate</a>()</code> <br><code><a href="layer_conv_1d.html">layer_conv_1d</a>()</code> <br><code><a href="layer_conv_1d_transpose.html">layer_conv_1d_transpose</a>()</code> <br><code><a href="layer_conv_2d.html">layer_conv_2d</a>()</code> <br><code><a href="layer_conv_2d_transpose.html">layer_conv_2d_transpose</a>()</code> <br><code><a href="layer_conv_3d.html">layer_conv_3d</a>()</code> <br><code><a href="layer_conv_3d_transpose.html">layer_conv_3d_transpose</a>()</code> <br><code><a href="layer_conv_lstm_1d.html">layer_conv_lstm_1d</a>()</code> <br><code><a href="layer_conv_lstm_2d.html">layer_conv_lstm_2d</a>()</code> <br><code><a href="layer_conv_lstm_3d.html">layer_conv_lstm_3d</a>()</code> <br><code><a href="layer_cropping_1d.html">layer_cropping_1d</a>()</code> <br><code><a href="layer_cropping_2d.html">layer_cropping_2d</a>()</code> <br><code><a href="layer_cropping_3d.html">layer_cropping_3d</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_dense.html">layer_dense</a>()</code> <br><code><a href="layer_depthwise_conv_1d.html">layer_depthwise_conv_1d</a>()</code> <br><code><a href="layer_depthwise_conv_2d.html">layer_depthwise_conv_2d</a>()</code> <br><code><a href="layer_discretization.html">layer_discretization</a>()</code> <br><code><a href="layer_dot.html">layer_dot</a>()</code> <br><code><a href="layer_dropout.html">layer_dropout</a>()</code> <br><code><a href="layer_einsum_dense.html">layer_einsum_dense</a>()</code> <br><code><a href="layer_embedding.html">layer_embedding</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_feature_space.html">layer_feature_space</a>()</code> <br><code><a href="layer_flatten.html">layer_flatten</a>()</code> <br><code><a href="layer_flax_module_wrapper.html">layer_flax_module_wrapper</a>()</code> <br><code><a href="layer_gaussian_dropout.html">layer_gaussian_dropout</a>()</code> <br><code><a href="layer_gaussian_noise.html">layer_gaussian_noise</a>()</code> <br><code><a href="layer_global_average_pooling_1d.html">layer_global_average_pooling_1d</a>()</code> <br><code><a href="layer_global_average_pooling_2d.html">layer_global_average_pooling_2d</a>()</code> <br><code><a href="layer_global_average_pooling_3d.html">layer_global_average_pooling_3d</a>()</code> <br><code><a href="layer_global_max_pooling_1d.html">layer_global_max_pooling_1d</a>()</code> <br><code><a href="layer_global_max_pooling_2d.html">layer_global_max_pooling_2d</a>()</code> <br><code><a href="layer_global_max_pooling_3d.html">layer_global_max_pooling_3d</a>()</code> <br><code><a href="layer_group_normalization.html">layer_group_normalization</a>()</code> <br><code><a href="layer_group_query_attention.html">layer_group_query_attention</a>()</code> <br><code><a href="layer_gru.html">layer_gru</a>()</code> <br><code><a href="layer_hashed_crossing.html">layer_hashed_crossing</a>()</code> <br><code><a href="layer_hashing.html">layer_hashing</a>()</code> <br><code><a href="layer_identity.html">layer_identity</a>()</code> <br><code><a href="layer_integer_lookup.html">layer_integer_lookup</a>()</code> <br><code><a href="layer_jax_model_wrapper.html">layer_jax_model_wrapper</a>()</code> <br><code><a href="layer_lambda.html">layer_lambda</a>()</code> <br><code><a href="layer_layer_normalization.html">layer_layer_normalization</a>()</code> <br><code><a href="layer_lstm.html">layer_lstm</a>()</code> <br><code><a href="layer_masking.html">layer_masking</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_max_pooling_1d.html">layer_max_pooling_1d</a>()</code> <br><code><a href="layer_max_pooling_2d.html">layer_max_pooling_2d</a>()</code> <br><code><a href="layer_max_pooling_3d.html">layer_max_pooling_3d</a>()</code> <br><code><a href="layer_maximum.html">layer_maximum</a>()</code> <br><code><a href="layer_mel_spectrogram.html">layer_mel_spectrogram</a>()</code> <br><code><a href="layer_minimum.html">layer_minimum</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_multi_head_attention.html">layer_multi_head_attention</a>()</code> <br><code><a href="layer_multiply.html">layer_multiply</a>()</code> <br><code><a href="layer_normalization.html">layer_normalization</a>()</code> <br><code><a href="layer_permute.html">layer_permute</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_brightness.html">layer_random_brightness</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_contrast.html">layer_random_contrast</a>()</code> <br><code><a href="layer_random_crop.html">layer_random_crop</a>()</code> <br><code><a href="layer_random_elastic_transform.html">layer_random_elastic_transform</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_flip.html">layer_random_flip</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_rotation.html">layer_random_rotation</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_random_translation.html">layer_random_translation</a>()</code> <br><code><a href="layer_random_zoom.html">layer_random_zoom</a>()</code> <br><code><a href="layer_repeat_vector.html">layer_repeat_vector</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_reshape.html">layer_reshape</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_rms_normalization.html">layer_rms_normalization</a>()</code> <br><code><a href="layer_rnn.html">layer_rnn</a>()</code> <br><code><a href="layer_separable_conv_1d.html">layer_separable_conv_1d</a>()</code> <br><code><a href="layer_separable_conv_2d.html">layer_separable_conv_2d</a>()</code> <br><code><a href="layer_simple_rnn.html">layer_simple_rnn</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br><code><a href="layer_spatial_dropout_1d.html">layer_spatial_dropout_1d</a>()</code> <br><code><a href="layer_spatial_dropout_2d.html">layer_spatial_dropout_2d</a>()</code> <br><code><a href="layer_spatial_dropout_3d.html">layer_spatial_dropout_3d</a>()</code> <br><code><a href="layer_spectral_normalization.html">layer_spectral_normalization</a>()</code> <br><code><a href="layer_stft_spectrogram.html">layer_stft_spectrogram</a>()</code> <br><code><a href="layer_string_lookup.html">layer_string_lookup</a>()</code> <br><code><a href="layer_subtract.html">layer_subtract</a>()</code> <br><code><a href="layer_text_vectorization.html">layer_text_vectorization</a>()</code> <br><code><a href="layer_tfsm.html">layer_tfsm</a>()</code> <br><code><a href="layer_time_distributed.html">layer_time_distributed</a>()</code> <br><code><a href="layer_unit_normalization.html">layer_unit_normalization</a>()</code> <br><code><a href="layer_upsampling_1d.html">layer_upsampling_1d</a>()</code> <br><code><a href="layer_upsampling_2d.html">layer_upsampling_2d</a>()</code> <br><code><a href="layer_upsampling_3d.html">layer_upsampling_3d</a>()</code> <br><code><a href="layer_zero_padding_1d.html">layer_zero_padding_1d</a>()</code> <br><code><a href="layer_zero_padding_2d.html">layer_zero_padding_2d</a>()</code> <br><code><a href="layer_zero_padding_3d.html">layer_zero_padding_3d</a>()</code> <br><code><a href="rnn_cell_gru.html">rnn_cell_gru</a>()</code> <br><code><a href="rnn_cell_lstm.html">rnn_cell_lstm</a>()</code> <br><code><a href="rnn_cell_simple.html">rnn_cell_simple</a>()</code> <br><code><a href="rnn_cells_stack.html">rnn_cells_stack</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -178,7 +188,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/layer_unit_normalization.html b/docs/dev/reference/layer_unit_normalization.html
index 108e33b77..74103e4df 100644
--- a/docs/dev/reference/layer_unit_normalization.html
+++ b/docs/dev/reference/layer_unit_normalization.html
@@ -10,7 +10,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -106,7 +106,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
     <div class="dont-index">
 <ul><li><p><a href="https://keras.io/api/layers/normalization_layers/unit_normalization#unitnormalization-class" class="external-link">https://keras.io/api/layers/normalization_layers/unit_normalization#unitnormalization-class</a></p></li>
 </ul><p>Other normalization layers: <br><code><a href="layer_batch_normalization.html">layer_batch_normalization</a>()</code> <br><code><a href="layer_group_normalization.html">layer_group_normalization</a>()</code> <br><code><a href="layer_layer_normalization.html">layer_layer_normalization</a>()</code> <br><code><a href="layer_rms_normalization.html">layer_rms_normalization</a>()</code> <br><code><a href="layer_spectral_normalization.html">layer_spectral_normalization</a>()</code> <br></p>
-<p>Other layers: <br><code><a href="Layer.html">Layer</a>()</code> <br><code><a href="layer_activation.html">layer_activation</a>()</code> <br><code><a href="layer_activation_elu.html">layer_activation_elu</a>()</code> <br><code><a href="layer_activation_leaky_relu.html">layer_activation_leaky_relu</a>()</code> <br><code><a href="layer_activation_parametric_relu.html">layer_activation_parametric_relu</a>()</code> <br><code><a href="layer_activation_relu.html">layer_activation_relu</a>()</code> <br><code><a href="layer_activation_softmax.html">layer_activation_softmax</a>()</code> <br><code><a href="layer_activity_regularization.html">layer_activity_regularization</a>()</code> <br><code><a href="layer_add.html">layer_add</a>()</code> <br><code><a href="layer_additive_attention.html">layer_additive_attention</a>()</code> <br><code><a href="layer_alpha_dropout.html">layer_alpha_dropout</a>()</code> <br><code><a href="layer_attention.html">layer_attention</a>()</code> <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_average.html">layer_average</a>()</code> <br><code><a href="layer_average_pooling_1d.html">layer_average_pooling_1d</a>()</code> <br><code><a href="layer_average_pooling_2d.html">layer_average_pooling_2d</a>()</code> <br><code><a href="layer_average_pooling_3d.html">layer_average_pooling_3d</a>()</code> <br><code><a href="layer_batch_normalization.html">layer_batch_normalization</a>()</code> <br><code><a href="layer_bidirectional.html">layer_bidirectional</a>()</code> <br><code><a href="layer_category_encoding.html">layer_category_encoding</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_concatenate.html">layer_concatenate</a>()</code> <br><code><a href="layer_conv_1d.html">layer_conv_1d</a>()</code> <br><code><a href="layer_conv_1d_transpose.html">layer_conv_1d_transpose</a>()</code> <br><code><a href="layer_conv_2d.html">layer_conv_2d</a>()</code> <br><code><a href="layer_conv_2d_transpose.html">layer_conv_2d_transpose</a>()</code> <br><code><a href="layer_conv_3d.html">layer_conv_3d</a>()</code> <br><code><a href="layer_conv_3d_transpose.html">layer_conv_3d_transpose</a>()</code> <br><code><a href="layer_conv_lstm_1d.html">layer_conv_lstm_1d</a>()</code> <br><code><a href="layer_conv_lstm_2d.html">layer_conv_lstm_2d</a>()</code> <br><code><a href="layer_conv_lstm_3d.html">layer_conv_lstm_3d</a>()</code> <br><code><a href="layer_cropping_1d.html">layer_cropping_1d</a>()</code> <br><code><a href="layer_cropping_2d.html">layer_cropping_2d</a>()</code> <br><code><a href="layer_cropping_3d.html">layer_cropping_3d</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_dense.html">layer_dense</a>()</code> <br><code><a href="layer_depthwise_conv_1d.html">layer_depthwise_conv_1d</a>()</code> <br><code><a href="layer_depthwise_conv_2d.html">layer_depthwise_conv_2d</a>()</code> <br><code><a href="layer_discretization.html">layer_discretization</a>()</code> <br><code><a href="layer_dot.html">layer_dot</a>()</code> <br><code><a href="layer_dropout.html">layer_dropout</a>()</code> <br><code><a href="layer_einsum_dense.html">layer_einsum_dense</a>()</code> <br><code><a href="layer_embedding.html">layer_embedding</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_feature_space.html">layer_feature_space</a>()</code> <br><code><a href="layer_flatten.html">layer_flatten</a>()</code> <br><code><a href="layer_flax_module_wrapper.html">layer_flax_module_wrapper</a>()</code> <br><code><a href="layer_gaussian_dropout.html">layer_gaussian_dropout</a>()</code> <br><code><a href="layer_gaussian_noise.html">layer_gaussian_noise</a>()</code> <br><code><a href="layer_global_average_pooling_1d.html">layer_global_average_pooling_1d</a>()</code> <br><code><a href="layer_global_average_pooling_2d.html">layer_global_average_pooling_2d</a>()</code> <br><code><a href="layer_global_average_pooling_3d.html">layer_global_average_pooling_3d</a>()</code> <br><code><a href="layer_global_max_pooling_1d.html">layer_global_max_pooling_1d</a>()</code> <br><code><a href="layer_global_max_pooling_2d.html">layer_global_max_pooling_2d</a>()</code> <br><code><a href="layer_global_max_pooling_3d.html">layer_global_max_pooling_3d</a>()</code> <br><code><a href="layer_group_normalization.html">layer_group_normalization</a>()</code> <br><code><a href="layer_group_query_attention.html">layer_group_query_attention</a>()</code> <br><code><a href="layer_gru.html">layer_gru</a>()</code> <br><code><a href="layer_hashed_crossing.html">layer_hashed_crossing</a>()</code> <br><code><a href="layer_hashing.html">layer_hashing</a>()</code> <br><code><a href="layer_identity.html">layer_identity</a>()</code> <br><code><a href="layer_integer_lookup.html">layer_integer_lookup</a>()</code> <br><code><a href="layer_jax_model_wrapper.html">layer_jax_model_wrapper</a>()</code> <br><code><a href="layer_lambda.html">layer_lambda</a>()</code> <br><code><a href="layer_layer_normalization.html">layer_layer_normalization</a>()</code> <br><code><a href="layer_lstm.html">layer_lstm</a>()</code> <br><code><a href="layer_masking.html">layer_masking</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_max_pooling_1d.html">layer_max_pooling_1d</a>()</code> <br><code><a href="layer_max_pooling_2d.html">layer_max_pooling_2d</a>()</code> <br><code><a href="layer_max_pooling_3d.html">layer_max_pooling_3d</a>()</code> <br><code><a href="layer_maximum.html">layer_maximum</a>()</code> <br><code><a href="layer_mel_spectrogram.html">layer_mel_spectrogram</a>()</code> <br><code><a href="layer_minimum.html">layer_minimum</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_multi_head_attention.html">layer_multi_head_attention</a>()</code> <br><code><a href="layer_multiply.html">layer_multiply</a>()</code> <br><code><a href="layer_normalization.html">layer_normalization</a>()</code> <br><code><a href="layer_permute.html">layer_permute</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_brightness.html">layer_random_brightness</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_contrast.html">layer_random_contrast</a>()</code> <br><code><a href="layer_random_crop.html">layer_random_crop</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_flip.html">layer_random_flip</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_rotation.html">layer_random_rotation</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_random_translation.html">layer_random_translation</a>()</code> <br><code><a href="layer_random_zoom.html">layer_random_zoom</a>()</code> <br><code><a href="layer_repeat_vector.html">layer_repeat_vector</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_reshape.html">layer_reshape</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_rms_normalization.html">layer_rms_normalization</a>()</code> <br><code><a href="layer_rnn.html">layer_rnn</a>()</code> <br><code><a href="layer_separable_conv_1d.html">layer_separable_conv_1d</a>()</code> <br><code><a href="layer_separable_conv_2d.html">layer_separable_conv_2d</a>()</code> <br><code><a href="layer_simple_rnn.html">layer_simple_rnn</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br><code><a href="layer_spatial_dropout_1d.html">layer_spatial_dropout_1d</a>()</code> <br><code><a href="layer_spatial_dropout_2d.html">layer_spatial_dropout_2d</a>()</code> <br><code><a href="layer_spatial_dropout_3d.html">layer_spatial_dropout_3d</a>()</code> <br><code><a href="layer_spectral_normalization.html">layer_spectral_normalization</a>()</code> <br><code><a href="layer_stft_spectrogram.html">layer_stft_spectrogram</a>()</code> <br><code><a href="layer_string_lookup.html">layer_string_lookup</a>()</code> <br><code><a href="layer_subtract.html">layer_subtract</a>()</code> <br><code><a href="layer_text_vectorization.html">layer_text_vectorization</a>()</code> <br><code><a href="layer_tfsm.html">layer_tfsm</a>()</code> <br><code><a href="layer_time_distributed.html">layer_time_distributed</a>()</code> <br><code><a href="layer_torch_module_wrapper.html">layer_torch_module_wrapper</a>()</code> <br><code><a href="layer_upsampling_1d.html">layer_upsampling_1d</a>()</code> <br><code><a href="layer_upsampling_2d.html">layer_upsampling_2d</a>()</code> <br><code><a href="layer_upsampling_3d.html">layer_upsampling_3d</a>()</code> <br><code><a href="layer_zero_padding_1d.html">layer_zero_padding_1d</a>()</code> <br><code><a href="layer_zero_padding_2d.html">layer_zero_padding_2d</a>()</code> <br><code><a href="layer_zero_padding_3d.html">layer_zero_padding_3d</a>()</code> <br><code><a href="rnn_cell_gru.html">rnn_cell_gru</a>()</code> <br><code><a href="rnn_cell_lstm.html">rnn_cell_lstm</a>()</code> <br><code><a href="rnn_cell_simple.html">rnn_cell_simple</a>()</code> <br><code><a href="rnn_cells_stack.html">rnn_cells_stack</a>()</code> <br></p></div>
+<p>Other layers: <br><code><a href="Layer.html">Layer</a>()</code> <br><code><a href="layer_activation.html">layer_activation</a>()</code> <br><code><a href="layer_activation_elu.html">layer_activation_elu</a>()</code> <br><code><a href="layer_activation_leaky_relu.html">layer_activation_leaky_relu</a>()</code> <br><code><a href="layer_activation_parametric_relu.html">layer_activation_parametric_relu</a>()</code> <br><code><a href="layer_activation_relu.html">layer_activation_relu</a>()</code> <br><code><a href="layer_activation_softmax.html">layer_activation_softmax</a>()</code> <br><code><a href="layer_activity_regularization.html">layer_activity_regularization</a>()</code> <br><code><a href="layer_add.html">layer_add</a>()</code> <br><code><a href="layer_additive_attention.html">layer_additive_attention</a>()</code> <br><code><a href="layer_alpha_dropout.html">layer_alpha_dropout</a>()</code> <br><code><a href="layer_attention.html">layer_attention</a>()</code> <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_average.html">layer_average</a>()</code> <br><code><a href="layer_average_pooling_1d.html">layer_average_pooling_1d</a>()</code> <br><code><a href="layer_average_pooling_2d.html">layer_average_pooling_2d</a>()</code> <br><code><a href="layer_average_pooling_3d.html">layer_average_pooling_3d</a>()</code> <br><code><a href="layer_batch_normalization.html">layer_batch_normalization</a>()</code> <br><code><a href="layer_bidirectional.html">layer_bidirectional</a>()</code> <br><code><a href="layer_category_encoding.html">layer_category_encoding</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_concatenate.html">layer_concatenate</a>()</code> <br><code><a href="layer_conv_1d.html">layer_conv_1d</a>()</code> <br><code><a href="layer_conv_1d_transpose.html">layer_conv_1d_transpose</a>()</code> <br><code><a href="layer_conv_2d.html">layer_conv_2d</a>()</code> <br><code><a href="layer_conv_2d_transpose.html">layer_conv_2d_transpose</a>()</code> <br><code><a href="layer_conv_3d.html">layer_conv_3d</a>()</code> <br><code><a href="layer_conv_3d_transpose.html">layer_conv_3d_transpose</a>()</code> <br><code><a href="layer_conv_lstm_1d.html">layer_conv_lstm_1d</a>()</code> <br><code><a href="layer_conv_lstm_2d.html">layer_conv_lstm_2d</a>()</code> <br><code><a href="layer_conv_lstm_3d.html">layer_conv_lstm_3d</a>()</code> <br><code><a href="layer_cropping_1d.html">layer_cropping_1d</a>()</code> <br><code><a href="layer_cropping_2d.html">layer_cropping_2d</a>()</code> <br><code><a href="layer_cropping_3d.html">layer_cropping_3d</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_dense.html">layer_dense</a>()</code> <br><code><a href="layer_depthwise_conv_1d.html">layer_depthwise_conv_1d</a>()</code> <br><code><a href="layer_depthwise_conv_2d.html">layer_depthwise_conv_2d</a>()</code> <br><code><a href="layer_discretization.html">layer_discretization</a>()</code> <br><code><a href="layer_dot.html">layer_dot</a>()</code> <br><code><a href="layer_dropout.html">layer_dropout</a>()</code> <br><code><a href="layer_einsum_dense.html">layer_einsum_dense</a>()</code> <br><code><a href="layer_embedding.html">layer_embedding</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_feature_space.html">layer_feature_space</a>()</code> <br><code><a href="layer_flatten.html">layer_flatten</a>()</code> <br><code><a href="layer_flax_module_wrapper.html">layer_flax_module_wrapper</a>()</code> <br><code><a href="layer_gaussian_dropout.html">layer_gaussian_dropout</a>()</code> <br><code><a href="layer_gaussian_noise.html">layer_gaussian_noise</a>()</code> <br><code><a href="layer_global_average_pooling_1d.html">layer_global_average_pooling_1d</a>()</code> <br><code><a href="layer_global_average_pooling_2d.html">layer_global_average_pooling_2d</a>()</code> <br><code><a href="layer_global_average_pooling_3d.html">layer_global_average_pooling_3d</a>()</code> <br><code><a href="layer_global_max_pooling_1d.html">layer_global_max_pooling_1d</a>()</code> <br><code><a href="layer_global_max_pooling_2d.html">layer_global_max_pooling_2d</a>()</code> <br><code><a href="layer_global_max_pooling_3d.html">layer_global_max_pooling_3d</a>()</code> <br><code><a href="layer_group_normalization.html">layer_group_normalization</a>()</code> <br><code><a href="layer_group_query_attention.html">layer_group_query_attention</a>()</code> <br><code><a href="layer_gru.html">layer_gru</a>()</code> <br><code><a href="layer_hashed_crossing.html">layer_hashed_crossing</a>()</code> <br><code><a href="layer_hashing.html">layer_hashing</a>()</code> <br><code><a href="layer_identity.html">layer_identity</a>()</code> <br><code><a href="layer_integer_lookup.html">layer_integer_lookup</a>()</code> <br><code><a href="layer_jax_model_wrapper.html">layer_jax_model_wrapper</a>()</code> <br><code><a href="layer_lambda.html">layer_lambda</a>()</code> <br><code><a href="layer_layer_normalization.html">layer_layer_normalization</a>()</code> <br><code><a href="layer_lstm.html">layer_lstm</a>()</code> <br><code><a href="layer_masking.html">layer_masking</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_max_pooling_1d.html">layer_max_pooling_1d</a>()</code> <br><code><a href="layer_max_pooling_2d.html">layer_max_pooling_2d</a>()</code> <br><code><a href="layer_max_pooling_3d.html">layer_max_pooling_3d</a>()</code> <br><code><a href="layer_maximum.html">layer_maximum</a>()</code> <br><code><a href="layer_mel_spectrogram.html">layer_mel_spectrogram</a>()</code> <br><code><a href="layer_minimum.html">layer_minimum</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_multi_head_attention.html">layer_multi_head_attention</a>()</code> <br><code><a href="layer_multiply.html">layer_multiply</a>()</code> <br><code><a href="layer_normalization.html">layer_normalization</a>()</code> <br><code><a href="layer_permute.html">layer_permute</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_brightness.html">layer_random_brightness</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_contrast.html">layer_random_contrast</a>()</code> <br><code><a href="layer_random_crop.html">layer_random_crop</a>()</code> <br><code><a href="layer_random_elastic_transform.html">layer_random_elastic_transform</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_flip.html">layer_random_flip</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_rotation.html">layer_random_rotation</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_random_translation.html">layer_random_translation</a>()</code> <br><code><a href="layer_random_zoom.html">layer_random_zoom</a>()</code> <br><code><a href="layer_repeat_vector.html">layer_repeat_vector</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_reshape.html">layer_reshape</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_rms_normalization.html">layer_rms_normalization</a>()</code> <br><code><a href="layer_rnn.html">layer_rnn</a>()</code> <br><code><a href="layer_separable_conv_1d.html">layer_separable_conv_1d</a>()</code> <br><code><a href="layer_separable_conv_2d.html">layer_separable_conv_2d</a>()</code> <br><code><a href="layer_simple_rnn.html">layer_simple_rnn</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br><code><a href="layer_spatial_dropout_1d.html">layer_spatial_dropout_1d</a>()</code> <br><code><a href="layer_spatial_dropout_2d.html">layer_spatial_dropout_2d</a>()</code> <br><code><a href="layer_spatial_dropout_3d.html">layer_spatial_dropout_3d</a>()</code> <br><code><a href="layer_spectral_normalization.html">layer_spectral_normalization</a>()</code> <br><code><a href="layer_stft_spectrogram.html">layer_stft_spectrogram</a>()</code> <br><code><a href="layer_string_lookup.html">layer_string_lookup</a>()</code> <br><code><a href="layer_subtract.html">layer_subtract</a>()</code> <br><code><a href="layer_text_vectorization.html">layer_text_vectorization</a>()</code> <br><code><a href="layer_tfsm.html">layer_tfsm</a>()</code> <br><code><a href="layer_time_distributed.html">layer_time_distributed</a>()</code> <br><code><a href="layer_torch_module_wrapper.html">layer_torch_module_wrapper</a>()</code> <br><code><a href="layer_upsampling_1d.html">layer_upsampling_1d</a>()</code> <br><code><a href="layer_upsampling_2d.html">layer_upsampling_2d</a>()</code> <br><code><a href="layer_upsampling_3d.html">layer_upsampling_3d</a>()</code> <br><code><a href="layer_zero_padding_1d.html">layer_zero_padding_1d</a>()</code> <br><code><a href="layer_zero_padding_2d.html">layer_zero_padding_2d</a>()</code> <br><code><a href="layer_zero_padding_3d.html">layer_zero_padding_3d</a>()</code> <br><code><a href="rnn_cell_gru.html">rnn_cell_gru</a>()</code> <br><code><a href="rnn_cell_lstm.html">rnn_cell_lstm</a>()</code> <br><code><a href="rnn_cell_simple.html">rnn_cell_simple</a>()</code> <br><code><a href="rnn_cells_stack.html">rnn_cells_stack</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -118,7 +118,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/layer_upsampling_1d.html b/docs/dev/reference/layer_upsampling_1d.html
index c5e8b24bb..a0b077b92 100644
--- a/docs/dev/reference/layer_upsampling_1d.html
+++ b/docs/dev/reference/layer_upsampling_1d.html
@@ -8,7 +8,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -126,7 +126,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
     <div class="dont-index">
 <ul><li><p><a href="https://keras.io/api/layers/reshaping_layers/up_sampling1d#upsampling1d-class" class="external-link">https://keras.io/api/layers/reshaping_layers/up_sampling1d#upsampling1d-class</a></p></li>
 </ul><p>Other reshaping layers: <br><code><a href="layer_cropping_1d.html">layer_cropping_1d</a>()</code> <br><code><a href="layer_cropping_2d.html">layer_cropping_2d</a>()</code> <br><code><a href="layer_cropping_3d.html">layer_cropping_3d</a>()</code> <br><code><a href="layer_flatten.html">layer_flatten</a>()</code> <br><code><a href="layer_permute.html">layer_permute</a>()</code> <br><code><a href="layer_repeat_vector.html">layer_repeat_vector</a>()</code> <br><code><a href="layer_reshape.html">layer_reshape</a>()</code> <br><code><a href="layer_upsampling_2d.html">layer_upsampling_2d</a>()</code> <br><code><a href="layer_upsampling_3d.html">layer_upsampling_3d</a>()</code> <br><code><a href="layer_zero_padding_1d.html">layer_zero_padding_1d</a>()</code> <br><code><a href="layer_zero_padding_2d.html">layer_zero_padding_2d</a>()</code> <br><code><a href="layer_zero_padding_3d.html">layer_zero_padding_3d</a>()</code> <br></p>
-<p>Other layers: <br><code><a href="Layer.html">Layer</a>()</code> <br><code><a href="layer_activation.html">layer_activation</a>()</code> <br><code><a href="layer_activation_elu.html">layer_activation_elu</a>()</code> <br><code><a href="layer_activation_leaky_relu.html">layer_activation_leaky_relu</a>()</code> <br><code><a href="layer_activation_parametric_relu.html">layer_activation_parametric_relu</a>()</code> <br><code><a href="layer_activation_relu.html">layer_activation_relu</a>()</code> <br><code><a href="layer_activation_softmax.html">layer_activation_softmax</a>()</code> <br><code><a href="layer_activity_regularization.html">layer_activity_regularization</a>()</code> <br><code><a href="layer_add.html">layer_add</a>()</code> <br><code><a href="layer_additive_attention.html">layer_additive_attention</a>()</code> <br><code><a href="layer_alpha_dropout.html">layer_alpha_dropout</a>()</code> <br><code><a href="layer_attention.html">layer_attention</a>()</code> <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_average.html">layer_average</a>()</code> <br><code><a href="layer_average_pooling_1d.html">layer_average_pooling_1d</a>()</code> <br><code><a href="layer_average_pooling_2d.html">layer_average_pooling_2d</a>()</code> <br><code><a href="layer_average_pooling_3d.html">layer_average_pooling_3d</a>()</code> <br><code><a href="layer_batch_normalization.html">layer_batch_normalization</a>()</code> <br><code><a href="layer_bidirectional.html">layer_bidirectional</a>()</code> <br><code><a href="layer_category_encoding.html">layer_category_encoding</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_concatenate.html">layer_concatenate</a>()</code> <br><code><a href="layer_conv_1d.html">layer_conv_1d</a>()</code> <br><code><a href="layer_conv_1d_transpose.html">layer_conv_1d_transpose</a>()</code> <br><code><a href="layer_conv_2d.html">layer_conv_2d</a>()</code> <br><code><a href="layer_conv_2d_transpose.html">layer_conv_2d_transpose</a>()</code> <br><code><a href="layer_conv_3d.html">layer_conv_3d</a>()</code> <br><code><a href="layer_conv_3d_transpose.html">layer_conv_3d_transpose</a>()</code> <br><code><a href="layer_conv_lstm_1d.html">layer_conv_lstm_1d</a>()</code> <br><code><a href="layer_conv_lstm_2d.html">layer_conv_lstm_2d</a>()</code> <br><code><a href="layer_conv_lstm_3d.html">layer_conv_lstm_3d</a>()</code> <br><code><a href="layer_cropping_1d.html">layer_cropping_1d</a>()</code> <br><code><a href="layer_cropping_2d.html">layer_cropping_2d</a>()</code> <br><code><a href="layer_cropping_3d.html">layer_cropping_3d</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_dense.html">layer_dense</a>()</code> <br><code><a href="layer_depthwise_conv_1d.html">layer_depthwise_conv_1d</a>()</code> <br><code><a href="layer_depthwise_conv_2d.html">layer_depthwise_conv_2d</a>()</code> <br><code><a href="layer_discretization.html">layer_discretization</a>()</code> <br><code><a href="layer_dot.html">layer_dot</a>()</code> <br><code><a href="layer_dropout.html">layer_dropout</a>()</code> <br><code><a href="layer_einsum_dense.html">layer_einsum_dense</a>()</code> <br><code><a href="layer_embedding.html">layer_embedding</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_feature_space.html">layer_feature_space</a>()</code> <br><code><a href="layer_flatten.html">layer_flatten</a>()</code> <br><code><a href="layer_flax_module_wrapper.html">layer_flax_module_wrapper</a>()</code> <br><code><a href="layer_gaussian_dropout.html">layer_gaussian_dropout</a>()</code> <br><code><a href="layer_gaussian_noise.html">layer_gaussian_noise</a>()</code> <br><code><a href="layer_global_average_pooling_1d.html">layer_global_average_pooling_1d</a>()</code> <br><code><a href="layer_global_average_pooling_2d.html">layer_global_average_pooling_2d</a>()</code> <br><code><a href="layer_global_average_pooling_3d.html">layer_global_average_pooling_3d</a>()</code> <br><code><a href="layer_global_max_pooling_1d.html">layer_global_max_pooling_1d</a>()</code> <br><code><a href="layer_global_max_pooling_2d.html">layer_global_max_pooling_2d</a>()</code> <br><code><a href="layer_global_max_pooling_3d.html">layer_global_max_pooling_3d</a>()</code> <br><code><a href="layer_group_normalization.html">layer_group_normalization</a>()</code> <br><code><a href="layer_group_query_attention.html">layer_group_query_attention</a>()</code> <br><code><a href="layer_gru.html">layer_gru</a>()</code> <br><code><a href="layer_hashed_crossing.html">layer_hashed_crossing</a>()</code> <br><code><a href="layer_hashing.html">layer_hashing</a>()</code> <br><code><a href="layer_identity.html">layer_identity</a>()</code> <br><code><a href="layer_integer_lookup.html">layer_integer_lookup</a>()</code> <br><code><a href="layer_jax_model_wrapper.html">layer_jax_model_wrapper</a>()</code> <br><code><a href="layer_lambda.html">layer_lambda</a>()</code> <br><code><a href="layer_layer_normalization.html">layer_layer_normalization</a>()</code> <br><code><a href="layer_lstm.html">layer_lstm</a>()</code> <br><code><a href="layer_masking.html">layer_masking</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_max_pooling_1d.html">layer_max_pooling_1d</a>()</code> <br><code><a href="layer_max_pooling_2d.html">layer_max_pooling_2d</a>()</code> <br><code><a href="layer_max_pooling_3d.html">layer_max_pooling_3d</a>()</code> <br><code><a href="layer_maximum.html">layer_maximum</a>()</code> <br><code><a href="layer_mel_spectrogram.html">layer_mel_spectrogram</a>()</code> <br><code><a href="layer_minimum.html">layer_minimum</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_multi_head_attention.html">layer_multi_head_attention</a>()</code> <br><code><a href="layer_multiply.html">layer_multiply</a>()</code> <br><code><a href="layer_normalization.html">layer_normalization</a>()</code> <br><code><a href="layer_permute.html">layer_permute</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_brightness.html">layer_random_brightness</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_contrast.html">layer_random_contrast</a>()</code> <br><code><a href="layer_random_crop.html">layer_random_crop</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_flip.html">layer_random_flip</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_rotation.html">layer_random_rotation</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_random_translation.html">layer_random_translation</a>()</code> <br><code><a href="layer_random_zoom.html">layer_random_zoom</a>()</code> <br><code><a href="layer_repeat_vector.html">layer_repeat_vector</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_reshape.html">layer_reshape</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_rms_normalization.html">layer_rms_normalization</a>()</code> <br><code><a href="layer_rnn.html">layer_rnn</a>()</code> <br><code><a href="layer_separable_conv_1d.html">layer_separable_conv_1d</a>()</code> <br><code><a href="layer_separable_conv_2d.html">layer_separable_conv_2d</a>()</code> <br><code><a href="layer_simple_rnn.html">layer_simple_rnn</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br><code><a href="layer_spatial_dropout_1d.html">layer_spatial_dropout_1d</a>()</code> <br><code><a href="layer_spatial_dropout_2d.html">layer_spatial_dropout_2d</a>()</code> <br><code><a href="layer_spatial_dropout_3d.html">layer_spatial_dropout_3d</a>()</code> <br><code><a href="layer_spectral_normalization.html">layer_spectral_normalization</a>()</code> <br><code><a href="layer_stft_spectrogram.html">layer_stft_spectrogram</a>()</code> <br><code><a href="layer_string_lookup.html">layer_string_lookup</a>()</code> <br><code><a href="layer_subtract.html">layer_subtract</a>()</code> <br><code><a href="layer_text_vectorization.html">layer_text_vectorization</a>()</code> <br><code><a href="layer_tfsm.html">layer_tfsm</a>()</code> <br><code><a href="layer_time_distributed.html">layer_time_distributed</a>()</code> <br><code><a href="layer_torch_module_wrapper.html">layer_torch_module_wrapper</a>()</code> <br><code><a href="layer_unit_normalization.html">layer_unit_normalization</a>()</code> <br><code><a href="layer_upsampling_2d.html">layer_upsampling_2d</a>()</code> <br><code><a href="layer_upsampling_3d.html">layer_upsampling_3d</a>()</code> <br><code><a href="layer_zero_padding_1d.html">layer_zero_padding_1d</a>()</code> <br><code><a href="layer_zero_padding_2d.html">layer_zero_padding_2d</a>()</code> <br><code><a href="layer_zero_padding_3d.html">layer_zero_padding_3d</a>()</code> <br><code><a href="rnn_cell_gru.html">rnn_cell_gru</a>()</code> <br><code><a href="rnn_cell_lstm.html">rnn_cell_lstm</a>()</code> <br><code><a href="rnn_cell_simple.html">rnn_cell_simple</a>()</code> <br><code><a href="rnn_cells_stack.html">rnn_cells_stack</a>()</code> <br></p></div>
+<p>Other layers: <br><code><a href="Layer.html">Layer</a>()</code> <br><code><a href="layer_activation.html">layer_activation</a>()</code> <br><code><a href="layer_activation_elu.html">layer_activation_elu</a>()</code> <br><code><a href="layer_activation_leaky_relu.html">layer_activation_leaky_relu</a>()</code> <br><code><a href="layer_activation_parametric_relu.html">layer_activation_parametric_relu</a>()</code> <br><code><a href="layer_activation_relu.html">layer_activation_relu</a>()</code> <br><code><a href="layer_activation_softmax.html">layer_activation_softmax</a>()</code> <br><code><a href="layer_activity_regularization.html">layer_activity_regularization</a>()</code> <br><code><a href="layer_add.html">layer_add</a>()</code> <br><code><a href="layer_additive_attention.html">layer_additive_attention</a>()</code> <br><code><a href="layer_alpha_dropout.html">layer_alpha_dropout</a>()</code> <br><code><a href="layer_attention.html">layer_attention</a>()</code> <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_average.html">layer_average</a>()</code> <br><code><a href="layer_average_pooling_1d.html">layer_average_pooling_1d</a>()</code> <br><code><a href="layer_average_pooling_2d.html">layer_average_pooling_2d</a>()</code> <br><code><a href="layer_average_pooling_3d.html">layer_average_pooling_3d</a>()</code> <br><code><a href="layer_batch_normalization.html">layer_batch_normalization</a>()</code> <br><code><a href="layer_bidirectional.html">layer_bidirectional</a>()</code> <br><code><a href="layer_category_encoding.html">layer_category_encoding</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_concatenate.html">layer_concatenate</a>()</code> <br><code><a href="layer_conv_1d.html">layer_conv_1d</a>()</code> <br><code><a href="layer_conv_1d_transpose.html">layer_conv_1d_transpose</a>()</code> <br><code><a href="layer_conv_2d.html">layer_conv_2d</a>()</code> <br><code><a href="layer_conv_2d_transpose.html">layer_conv_2d_transpose</a>()</code> <br><code><a href="layer_conv_3d.html">layer_conv_3d</a>()</code> <br><code><a href="layer_conv_3d_transpose.html">layer_conv_3d_transpose</a>()</code> <br><code><a href="layer_conv_lstm_1d.html">layer_conv_lstm_1d</a>()</code> <br><code><a href="layer_conv_lstm_2d.html">layer_conv_lstm_2d</a>()</code> <br><code><a href="layer_conv_lstm_3d.html">layer_conv_lstm_3d</a>()</code> <br><code><a href="layer_cropping_1d.html">layer_cropping_1d</a>()</code> <br><code><a href="layer_cropping_2d.html">layer_cropping_2d</a>()</code> <br><code><a href="layer_cropping_3d.html">layer_cropping_3d</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_dense.html">layer_dense</a>()</code> <br><code><a href="layer_depthwise_conv_1d.html">layer_depthwise_conv_1d</a>()</code> <br><code><a href="layer_depthwise_conv_2d.html">layer_depthwise_conv_2d</a>()</code> <br><code><a href="layer_discretization.html">layer_discretization</a>()</code> <br><code><a href="layer_dot.html">layer_dot</a>()</code> <br><code><a href="layer_dropout.html">layer_dropout</a>()</code> <br><code><a href="layer_einsum_dense.html">layer_einsum_dense</a>()</code> <br><code><a href="layer_embedding.html">layer_embedding</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_feature_space.html">layer_feature_space</a>()</code> <br><code><a href="layer_flatten.html">layer_flatten</a>()</code> <br><code><a href="layer_flax_module_wrapper.html">layer_flax_module_wrapper</a>()</code> <br><code><a href="layer_gaussian_dropout.html">layer_gaussian_dropout</a>()</code> <br><code><a href="layer_gaussian_noise.html">layer_gaussian_noise</a>()</code> <br><code><a href="layer_global_average_pooling_1d.html">layer_global_average_pooling_1d</a>()</code> <br><code><a href="layer_global_average_pooling_2d.html">layer_global_average_pooling_2d</a>()</code> <br><code><a href="layer_global_average_pooling_3d.html">layer_global_average_pooling_3d</a>()</code> <br><code><a href="layer_global_max_pooling_1d.html">layer_global_max_pooling_1d</a>()</code> <br><code><a href="layer_global_max_pooling_2d.html">layer_global_max_pooling_2d</a>()</code> <br><code><a href="layer_global_max_pooling_3d.html">layer_global_max_pooling_3d</a>()</code> <br><code><a href="layer_group_normalization.html">layer_group_normalization</a>()</code> <br><code><a href="layer_group_query_attention.html">layer_group_query_attention</a>()</code> <br><code><a href="layer_gru.html">layer_gru</a>()</code> <br><code><a href="layer_hashed_crossing.html">layer_hashed_crossing</a>()</code> <br><code><a href="layer_hashing.html">layer_hashing</a>()</code> <br><code><a href="layer_identity.html">layer_identity</a>()</code> <br><code><a href="layer_integer_lookup.html">layer_integer_lookup</a>()</code> <br><code><a href="layer_jax_model_wrapper.html">layer_jax_model_wrapper</a>()</code> <br><code><a href="layer_lambda.html">layer_lambda</a>()</code> <br><code><a href="layer_layer_normalization.html">layer_layer_normalization</a>()</code> <br><code><a href="layer_lstm.html">layer_lstm</a>()</code> <br><code><a href="layer_masking.html">layer_masking</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_max_pooling_1d.html">layer_max_pooling_1d</a>()</code> <br><code><a href="layer_max_pooling_2d.html">layer_max_pooling_2d</a>()</code> <br><code><a href="layer_max_pooling_3d.html">layer_max_pooling_3d</a>()</code> <br><code><a href="layer_maximum.html">layer_maximum</a>()</code> <br><code><a href="layer_mel_spectrogram.html">layer_mel_spectrogram</a>()</code> <br><code><a href="layer_minimum.html">layer_minimum</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_multi_head_attention.html">layer_multi_head_attention</a>()</code> <br><code><a href="layer_multiply.html">layer_multiply</a>()</code> <br><code><a href="layer_normalization.html">layer_normalization</a>()</code> <br><code><a href="layer_permute.html">layer_permute</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_brightness.html">layer_random_brightness</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_contrast.html">layer_random_contrast</a>()</code> <br><code><a href="layer_random_crop.html">layer_random_crop</a>()</code> <br><code><a href="layer_random_elastic_transform.html">layer_random_elastic_transform</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_flip.html">layer_random_flip</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_rotation.html">layer_random_rotation</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_random_translation.html">layer_random_translation</a>()</code> <br><code><a href="layer_random_zoom.html">layer_random_zoom</a>()</code> <br><code><a href="layer_repeat_vector.html">layer_repeat_vector</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_reshape.html">layer_reshape</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_rms_normalization.html">layer_rms_normalization</a>()</code> <br><code><a href="layer_rnn.html">layer_rnn</a>()</code> <br><code><a href="layer_separable_conv_1d.html">layer_separable_conv_1d</a>()</code> <br><code><a href="layer_separable_conv_2d.html">layer_separable_conv_2d</a>()</code> <br><code><a href="layer_simple_rnn.html">layer_simple_rnn</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br><code><a href="layer_spatial_dropout_1d.html">layer_spatial_dropout_1d</a>()</code> <br><code><a href="layer_spatial_dropout_2d.html">layer_spatial_dropout_2d</a>()</code> <br><code><a href="layer_spatial_dropout_3d.html">layer_spatial_dropout_3d</a>()</code> <br><code><a href="layer_spectral_normalization.html">layer_spectral_normalization</a>()</code> <br><code><a href="layer_stft_spectrogram.html">layer_stft_spectrogram</a>()</code> <br><code><a href="layer_string_lookup.html">layer_string_lookup</a>()</code> <br><code><a href="layer_subtract.html">layer_subtract</a>()</code> <br><code><a href="layer_text_vectorization.html">layer_text_vectorization</a>()</code> <br><code><a href="layer_tfsm.html">layer_tfsm</a>()</code> <br><code><a href="layer_time_distributed.html">layer_time_distributed</a>()</code> <br><code><a href="layer_torch_module_wrapper.html">layer_torch_module_wrapper</a>()</code> <br><code><a href="layer_unit_normalization.html">layer_unit_normalization</a>()</code> <br><code><a href="layer_upsampling_2d.html">layer_upsampling_2d</a>()</code> <br><code><a href="layer_upsampling_3d.html">layer_upsampling_3d</a>()</code> <br><code><a href="layer_zero_padding_1d.html">layer_zero_padding_1d</a>()</code> <br><code><a href="layer_zero_padding_2d.html">layer_zero_padding_2d</a>()</code> <br><code><a href="layer_zero_padding_3d.html">layer_zero_padding_3d</a>()</code> <br><code><a href="rnn_cell_gru.html">rnn_cell_gru</a>()</code> <br><code><a href="rnn_cell_lstm.html">rnn_cell_lstm</a>()</code> <br><code><a href="rnn_cell_simple.html">rnn_cell_simple</a>()</code> <br><code><a href="rnn_cells_stack.html">rnn_cells_stack</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -138,7 +138,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/layer_upsampling_2d.html b/docs/dev/reference/layer_upsampling_2d.html
index 24e8c9eae..ce6fde429 100644
--- a/docs/dev/reference/layer_upsampling_2d.html
+++ b/docs/dev/reference/layer_upsampling_2d.html
@@ -12,7 +12,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -170,7 +170,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
     <div class="dont-index">
 <ul><li><p><a href="https://keras.io/api/layers/reshaping_layers/up_sampling2d#upsampling2d-class" class="external-link">https://keras.io/api/layers/reshaping_layers/up_sampling2d#upsampling2d-class</a></p></li>
 </ul><p>Other reshaping layers: <br><code><a href="layer_cropping_1d.html">layer_cropping_1d</a>()</code> <br><code><a href="layer_cropping_2d.html">layer_cropping_2d</a>()</code> <br><code><a href="layer_cropping_3d.html">layer_cropping_3d</a>()</code> <br><code><a href="layer_flatten.html">layer_flatten</a>()</code> <br><code><a href="layer_permute.html">layer_permute</a>()</code> <br><code><a href="layer_repeat_vector.html">layer_repeat_vector</a>()</code> <br><code><a href="layer_reshape.html">layer_reshape</a>()</code> <br><code><a href="layer_upsampling_1d.html">layer_upsampling_1d</a>()</code> <br><code><a href="layer_upsampling_3d.html">layer_upsampling_3d</a>()</code> <br><code><a href="layer_zero_padding_1d.html">layer_zero_padding_1d</a>()</code> <br><code><a href="layer_zero_padding_2d.html">layer_zero_padding_2d</a>()</code> <br><code><a href="layer_zero_padding_3d.html">layer_zero_padding_3d</a>()</code> <br></p>
-<p>Other layers: <br><code><a href="Layer.html">Layer</a>()</code> <br><code><a href="layer_activation.html">layer_activation</a>()</code> <br><code><a href="layer_activation_elu.html">layer_activation_elu</a>()</code> <br><code><a href="layer_activation_leaky_relu.html">layer_activation_leaky_relu</a>()</code> <br><code><a href="layer_activation_parametric_relu.html">layer_activation_parametric_relu</a>()</code> <br><code><a href="layer_activation_relu.html">layer_activation_relu</a>()</code> <br><code><a href="layer_activation_softmax.html">layer_activation_softmax</a>()</code> <br><code><a href="layer_activity_regularization.html">layer_activity_regularization</a>()</code> <br><code><a href="layer_add.html">layer_add</a>()</code> <br><code><a href="layer_additive_attention.html">layer_additive_attention</a>()</code> <br><code><a href="layer_alpha_dropout.html">layer_alpha_dropout</a>()</code> <br><code><a href="layer_attention.html">layer_attention</a>()</code> <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_average.html">layer_average</a>()</code> <br><code><a href="layer_average_pooling_1d.html">layer_average_pooling_1d</a>()</code> <br><code><a href="layer_average_pooling_2d.html">layer_average_pooling_2d</a>()</code> <br><code><a href="layer_average_pooling_3d.html">layer_average_pooling_3d</a>()</code> <br><code><a href="layer_batch_normalization.html">layer_batch_normalization</a>()</code> <br><code><a href="layer_bidirectional.html">layer_bidirectional</a>()</code> <br><code><a href="layer_category_encoding.html">layer_category_encoding</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_concatenate.html">layer_concatenate</a>()</code> <br><code><a href="layer_conv_1d.html">layer_conv_1d</a>()</code> <br><code><a href="layer_conv_1d_transpose.html">layer_conv_1d_transpose</a>()</code> <br><code><a href="layer_conv_2d.html">layer_conv_2d</a>()</code> <br><code><a href="layer_conv_2d_transpose.html">layer_conv_2d_transpose</a>()</code> <br><code><a href="layer_conv_3d.html">layer_conv_3d</a>()</code> <br><code><a href="layer_conv_3d_transpose.html">layer_conv_3d_transpose</a>()</code> <br><code><a href="layer_conv_lstm_1d.html">layer_conv_lstm_1d</a>()</code> <br><code><a href="layer_conv_lstm_2d.html">layer_conv_lstm_2d</a>()</code> <br><code><a href="layer_conv_lstm_3d.html">layer_conv_lstm_3d</a>()</code> <br><code><a href="layer_cropping_1d.html">layer_cropping_1d</a>()</code> <br><code><a href="layer_cropping_2d.html">layer_cropping_2d</a>()</code> <br><code><a href="layer_cropping_3d.html">layer_cropping_3d</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_dense.html">layer_dense</a>()</code> <br><code><a href="layer_depthwise_conv_1d.html">layer_depthwise_conv_1d</a>()</code> <br><code><a href="layer_depthwise_conv_2d.html">layer_depthwise_conv_2d</a>()</code> <br><code><a href="layer_discretization.html">layer_discretization</a>()</code> <br><code><a href="layer_dot.html">layer_dot</a>()</code> <br><code><a href="layer_dropout.html">layer_dropout</a>()</code> <br><code><a href="layer_einsum_dense.html">layer_einsum_dense</a>()</code> <br><code><a href="layer_embedding.html">layer_embedding</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_feature_space.html">layer_feature_space</a>()</code> <br><code><a href="layer_flatten.html">layer_flatten</a>()</code> <br><code><a href="layer_flax_module_wrapper.html">layer_flax_module_wrapper</a>()</code> <br><code><a href="layer_gaussian_dropout.html">layer_gaussian_dropout</a>()</code> <br><code><a href="layer_gaussian_noise.html">layer_gaussian_noise</a>()</code> <br><code><a href="layer_global_average_pooling_1d.html">layer_global_average_pooling_1d</a>()</code> <br><code><a href="layer_global_average_pooling_2d.html">layer_global_average_pooling_2d</a>()</code> <br><code><a href="layer_global_average_pooling_3d.html">layer_global_average_pooling_3d</a>()</code> <br><code><a href="layer_global_max_pooling_1d.html">layer_global_max_pooling_1d</a>()</code> <br><code><a href="layer_global_max_pooling_2d.html">layer_global_max_pooling_2d</a>()</code> <br><code><a href="layer_global_max_pooling_3d.html">layer_global_max_pooling_3d</a>()</code> <br><code><a href="layer_group_normalization.html">layer_group_normalization</a>()</code> <br><code><a href="layer_group_query_attention.html">layer_group_query_attention</a>()</code> <br><code><a href="layer_gru.html">layer_gru</a>()</code> <br><code><a href="layer_hashed_crossing.html">layer_hashed_crossing</a>()</code> <br><code><a href="layer_hashing.html">layer_hashing</a>()</code> <br><code><a href="layer_identity.html">layer_identity</a>()</code> <br><code><a href="layer_integer_lookup.html">layer_integer_lookup</a>()</code> <br><code><a href="layer_jax_model_wrapper.html">layer_jax_model_wrapper</a>()</code> <br><code><a href="layer_lambda.html">layer_lambda</a>()</code> <br><code><a href="layer_layer_normalization.html">layer_layer_normalization</a>()</code> <br><code><a href="layer_lstm.html">layer_lstm</a>()</code> <br><code><a href="layer_masking.html">layer_masking</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_max_pooling_1d.html">layer_max_pooling_1d</a>()</code> <br><code><a href="layer_max_pooling_2d.html">layer_max_pooling_2d</a>()</code> <br><code><a href="layer_max_pooling_3d.html">layer_max_pooling_3d</a>()</code> <br><code><a href="layer_maximum.html">layer_maximum</a>()</code> <br><code><a href="layer_mel_spectrogram.html">layer_mel_spectrogram</a>()</code> <br><code><a href="layer_minimum.html">layer_minimum</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_multi_head_attention.html">layer_multi_head_attention</a>()</code> <br><code><a href="layer_multiply.html">layer_multiply</a>()</code> <br><code><a href="layer_normalization.html">layer_normalization</a>()</code> <br><code><a href="layer_permute.html">layer_permute</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_brightness.html">layer_random_brightness</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_contrast.html">layer_random_contrast</a>()</code> <br><code><a href="layer_random_crop.html">layer_random_crop</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_flip.html">layer_random_flip</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_rotation.html">layer_random_rotation</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_random_translation.html">layer_random_translation</a>()</code> <br><code><a href="layer_random_zoom.html">layer_random_zoom</a>()</code> <br><code><a href="layer_repeat_vector.html">layer_repeat_vector</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_reshape.html">layer_reshape</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_rms_normalization.html">layer_rms_normalization</a>()</code> <br><code><a href="layer_rnn.html">layer_rnn</a>()</code> <br><code><a href="layer_separable_conv_1d.html">layer_separable_conv_1d</a>()</code> <br><code><a href="layer_separable_conv_2d.html">layer_separable_conv_2d</a>()</code> <br><code><a href="layer_simple_rnn.html">layer_simple_rnn</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br><code><a href="layer_spatial_dropout_1d.html">layer_spatial_dropout_1d</a>()</code> <br><code><a href="layer_spatial_dropout_2d.html">layer_spatial_dropout_2d</a>()</code> <br><code><a href="layer_spatial_dropout_3d.html">layer_spatial_dropout_3d</a>()</code> <br><code><a href="layer_spectral_normalization.html">layer_spectral_normalization</a>()</code> <br><code><a href="layer_stft_spectrogram.html">layer_stft_spectrogram</a>()</code> <br><code><a href="layer_string_lookup.html">layer_string_lookup</a>()</code> <br><code><a href="layer_subtract.html">layer_subtract</a>()</code> <br><code><a href="layer_text_vectorization.html">layer_text_vectorization</a>()</code> <br><code><a href="layer_tfsm.html">layer_tfsm</a>()</code> <br><code><a href="layer_time_distributed.html">layer_time_distributed</a>()</code> <br><code><a href="layer_torch_module_wrapper.html">layer_torch_module_wrapper</a>()</code> <br><code><a href="layer_unit_normalization.html">layer_unit_normalization</a>()</code> <br><code><a href="layer_upsampling_1d.html">layer_upsampling_1d</a>()</code> <br><code><a href="layer_upsampling_3d.html">layer_upsampling_3d</a>()</code> <br><code><a href="layer_zero_padding_1d.html">layer_zero_padding_1d</a>()</code> <br><code><a href="layer_zero_padding_2d.html">layer_zero_padding_2d</a>()</code> <br><code><a href="layer_zero_padding_3d.html">layer_zero_padding_3d</a>()</code> <br><code><a href="rnn_cell_gru.html">rnn_cell_gru</a>()</code> <br><code><a href="rnn_cell_lstm.html">rnn_cell_lstm</a>()</code> <br><code><a href="rnn_cell_simple.html">rnn_cell_simple</a>()</code> <br><code><a href="rnn_cells_stack.html">rnn_cells_stack</a>()</code> <br></p></div>
+<p>Other layers: <br><code><a href="Layer.html">Layer</a>()</code> <br><code><a href="layer_activation.html">layer_activation</a>()</code> <br><code><a href="layer_activation_elu.html">layer_activation_elu</a>()</code> <br><code><a href="layer_activation_leaky_relu.html">layer_activation_leaky_relu</a>()</code> <br><code><a href="layer_activation_parametric_relu.html">layer_activation_parametric_relu</a>()</code> <br><code><a href="layer_activation_relu.html">layer_activation_relu</a>()</code> <br><code><a href="layer_activation_softmax.html">layer_activation_softmax</a>()</code> <br><code><a href="layer_activity_regularization.html">layer_activity_regularization</a>()</code> <br><code><a href="layer_add.html">layer_add</a>()</code> <br><code><a href="layer_additive_attention.html">layer_additive_attention</a>()</code> <br><code><a href="layer_alpha_dropout.html">layer_alpha_dropout</a>()</code> <br><code><a href="layer_attention.html">layer_attention</a>()</code> <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_average.html">layer_average</a>()</code> <br><code><a href="layer_average_pooling_1d.html">layer_average_pooling_1d</a>()</code> <br><code><a href="layer_average_pooling_2d.html">layer_average_pooling_2d</a>()</code> <br><code><a href="layer_average_pooling_3d.html">layer_average_pooling_3d</a>()</code> <br><code><a href="layer_batch_normalization.html">layer_batch_normalization</a>()</code> <br><code><a href="layer_bidirectional.html">layer_bidirectional</a>()</code> <br><code><a href="layer_category_encoding.html">layer_category_encoding</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_concatenate.html">layer_concatenate</a>()</code> <br><code><a href="layer_conv_1d.html">layer_conv_1d</a>()</code> <br><code><a href="layer_conv_1d_transpose.html">layer_conv_1d_transpose</a>()</code> <br><code><a href="layer_conv_2d.html">layer_conv_2d</a>()</code> <br><code><a href="layer_conv_2d_transpose.html">layer_conv_2d_transpose</a>()</code> <br><code><a href="layer_conv_3d.html">layer_conv_3d</a>()</code> <br><code><a href="layer_conv_3d_transpose.html">layer_conv_3d_transpose</a>()</code> <br><code><a href="layer_conv_lstm_1d.html">layer_conv_lstm_1d</a>()</code> <br><code><a href="layer_conv_lstm_2d.html">layer_conv_lstm_2d</a>()</code> <br><code><a href="layer_conv_lstm_3d.html">layer_conv_lstm_3d</a>()</code> <br><code><a href="layer_cropping_1d.html">layer_cropping_1d</a>()</code> <br><code><a href="layer_cropping_2d.html">layer_cropping_2d</a>()</code> <br><code><a href="layer_cropping_3d.html">layer_cropping_3d</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_dense.html">layer_dense</a>()</code> <br><code><a href="layer_depthwise_conv_1d.html">layer_depthwise_conv_1d</a>()</code> <br><code><a href="layer_depthwise_conv_2d.html">layer_depthwise_conv_2d</a>()</code> <br><code><a href="layer_discretization.html">layer_discretization</a>()</code> <br><code><a href="layer_dot.html">layer_dot</a>()</code> <br><code><a href="layer_dropout.html">layer_dropout</a>()</code> <br><code><a href="layer_einsum_dense.html">layer_einsum_dense</a>()</code> <br><code><a href="layer_embedding.html">layer_embedding</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_feature_space.html">layer_feature_space</a>()</code> <br><code><a href="layer_flatten.html">layer_flatten</a>()</code> <br><code><a href="layer_flax_module_wrapper.html">layer_flax_module_wrapper</a>()</code> <br><code><a href="layer_gaussian_dropout.html">layer_gaussian_dropout</a>()</code> <br><code><a href="layer_gaussian_noise.html">layer_gaussian_noise</a>()</code> <br><code><a href="layer_global_average_pooling_1d.html">layer_global_average_pooling_1d</a>()</code> <br><code><a href="layer_global_average_pooling_2d.html">layer_global_average_pooling_2d</a>()</code> <br><code><a href="layer_global_average_pooling_3d.html">layer_global_average_pooling_3d</a>()</code> <br><code><a href="layer_global_max_pooling_1d.html">layer_global_max_pooling_1d</a>()</code> <br><code><a href="layer_global_max_pooling_2d.html">layer_global_max_pooling_2d</a>()</code> <br><code><a href="layer_global_max_pooling_3d.html">layer_global_max_pooling_3d</a>()</code> <br><code><a href="layer_group_normalization.html">layer_group_normalization</a>()</code> <br><code><a href="layer_group_query_attention.html">layer_group_query_attention</a>()</code> <br><code><a href="layer_gru.html">layer_gru</a>()</code> <br><code><a href="layer_hashed_crossing.html">layer_hashed_crossing</a>()</code> <br><code><a href="layer_hashing.html">layer_hashing</a>()</code> <br><code><a href="layer_identity.html">layer_identity</a>()</code> <br><code><a href="layer_integer_lookup.html">layer_integer_lookup</a>()</code> <br><code><a href="layer_jax_model_wrapper.html">layer_jax_model_wrapper</a>()</code> <br><code><a href="layer_lambda.html">layer_lambda</a>()</code> <br><code><a href="layer_layer_normalization.html">layer_layer_normalization</a>()</code> <br><code><a href="layer_lstm.html">layer_lstm</a>()</code> <br><code><a href="layer_masking.html">layer_masking</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_max_pooling_1d.html">layer_max_pooling_1d</a>()</code> <br><code><a href="layer_max_pooling_2d.html">layer_max_pooling_2d</a>()</code> <br><code><a href="layer_max_pooling_3d.html">layer_max_pooling_3d</a>()</code> <br><code><a href="layer_maximum.html">layer_maximum</a>()</code> <br><code><a href="layer_mel_spectrogram.html">layer_mel_spectrogram</a>()</code> <br><code><a href="layer_minimum.html">layer_minimum</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_multi_head_attention.html">layer_multi_head_attention</a>()</code> <br><code><a href="layer_multiply.html">layer_multiply</a>()</code> <br><code><a href="layer_normalization.html">layer_normalization</a>()</code> <br><code><a href="layer_permute.html">layer_permute</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_brightness.html">layer_random_brightness</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_contrast.html">layer_random_contrast</a>()</code> <br><code><a href="layer_random_crop.html">layer_random_crop</a>()</code> <br><code><a href="layer_random_elastic_transform.html">layer_random_elastic_transform</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_flip.html">layer_random_flip</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_rotation.html">layer_random_rotation</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_random_translation.html">layer_random_translation</a>()</code> <br><code><a href="layer_random_zoom.html">layer_random_zoom</a>()</code> <br><code><a href="layer_repeat_vector.html">layer_repeat_vector</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_reshape.html">layer_reshape</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_rms_normalization.html">layer_rms_normalization</a>()</code> <br><code><a href="layer_rnn.html">layer_rnn</a>()</code> <br><code><a href="layer_separable_conv_1d.html">layer_separable_conv_1d</a>()</code> <br><code><a href="layer_separable_conv_2d.html">layer_separable_conv_2d</a>()</code> <br><code><a href="layer_simple_rnn.html">layer_simple_rnn</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br><code><a href="layer_spatial_dropout_1d.html">layer_spatial_dropout_1d</a>()</code> <br><code><a href="layer_spatial_dropout_2d.html">layer_spatial_dropout_2d</a>()</code> <br><code><a href="layer_spatial_dropout_3d.html">layer_spatial_dropout_3d</a>()</code> <br><code><a href="layer_spectral_normalization.html">layer_spectral_normalization</a>()</code> <br><code><a href="layer_stft_spectrogram.html">layer_stft_spectrogram</a>()</code> <br><code><a href="layer_string_lookup.html">layer_string_lookup</a>()</code> <br><code><a href="layer_subtract.html">layer_subtract</a>()</code> <br><code><a href="layer_text_vectorization.html">layer_text_vectorization</a>()</code> <br><code><a href="layer_tfsm.html">layer_tfsm</a>()</code> <br><code><a href="layer_time_distributed.html">layer_time_distributed</a>()</code> <br><code><a href="layer_torch_module_wrapper.html">layer_torch_module_wrapper</a>()</code> <br><code><a href="layer_unit_normalization.html">layer_unit_normalization</a>()</code> <br><code><a href="layer_upsampling_1d.html">layer_upsampling_1d</a>()</code> <br><code><a href="layer_upsampling_3d.html">layer_upsampling_3d</a>()</code> <br><code><a href="layer_zero_padding_1d.html">layer_zero_padding_1d</a>()</code> <br><code><a href="layer_zero_padding_2d.html">layer_zero_padding_2d</a>()</code> <br><code><a href="layer_zero_padding_3d.html">layer_zero_padding_3d</a>()</code> <br><code><a href="rnn_cell_gru.html">rnn_cell_gru</a>()</code> <br><code><a href="rnn_cell_lstm.html">rnn_cell_lstm</a>()</code> <br><code><a href="rnn_cell_simple.html">rnn_cell_simple</a>()</code> <br><code><a href="rnn_cells_stack.html">rnn_cells_stack</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -182,7 +182,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/layer_upsampling_3d.html b/docs/dev/reference/layer_upsampling_3d.html
index 943c930d5..72e733d1c 100644
--- a/docs/dev/reference/layer_upsampling_3d.html
+++ b/docs/dev/reference/layer_upsampling_3d.html
@@ -10,7 +10,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -133,7 +133,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
     <div class="dont-index">
 <ul><li><p><a href="https://keras.io/api/layers/reshaping_layers/up_sampling3d#upsampling3d-class" class="external-link">https://keras.io/api/layers/reshaping_layers/up_sampling3d#upsampling3d-class</a></p></li>
 </ul><p>Other reshaping layers: <br><code><a href="layer_cropping_1d.html">layer_cropping_1d</a>()</code> <br><code><a href="layer_cropping_2d.html">layer_cropping_2d</a>()</code> <br><code><a href="layer_cropping_3d.html">layer_cropping_3d</a>()</code> <br><code><a href="layer_flatten.html">layer_flatten</a>()</code> <br><code><a href="layer_permute.html">layer_permute</a>()</code> <br><code><a href="layer_repeat_vector.html">layer_repeat_vector</a>()</code> <br><code><a href="layer_reshape.html">layer_reshape</a>()</code> <br><code><a href="layer_upsampling_1d.html">layer_upsampling_1d</a>()</code> <br><code><a href="layer_upsampling_2d.html">layer_upsampling_2d</a>()</code> <br><code><a href="layer_zero_padding_1d.html">layer_zero_padding_1d</a>()</code> <br><code><a href="layer_zero_padding_2d.html">layer_zero_padding_2d</a>()</code> <br><code><a href="layer_zero_padding_3d.html">layer_zero_padding_3d</a>()</code> <br></p>
-<p>Other layers: <br><code><a href="Layer.html">Layer</a>()</code> <br><code><a href="layer_activation.html">layer_activation</a>()</code> <br><code><a href="layer_activation_elu.html">layer_activation_elu</a>()</code> <br><code><a href="layer_activation_leaky_relu.html">layer_activation_leaky_relu</a>()</code> <br><code><a href="layer_activation_parametric_relu.html">layer_activation_parametric_relu</a>()</code> <br><code><a href="layer_activation_relu.html">layer_activation_relu</a>()</code> <br><code><a href="layer_activation_softmax.html">layer_activation_softmax</a>()</code> <br><code><a href="layer_activity_regularization.html">layer_activity_regularization</a>()</code> <br><code><a href="layer_add.html">layer_add</a>()</code> <br><code><a href="layer_additive_attention.html">layer_additive_attention</a>()</code> <br><code><a href="layer_alpha_dropout.html">layer_alpha_dropout</a>()</code> <br><code><a href="layer_attention.html">layer_attention</a>()</code> <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_average.html">layer_average</a>()</code> <br><code><a href="layer_average_pooling_1d.html">layer_average_pooling_1d</a>()</code> <br><code><a href="layer_average_pooling_2d.html">layer_average_pooling_2d</a>()</code> <br><code><a href="layer_average_pooling_3d.html">layer_average_pooling_3d</a>()</code> <br><code><a href="layer_batch_normalization.html">layer_batch_normalization</a>()</code> <br><code><a href="layer_bidirectional.html">layer_bidirectional</a>()</code> <br><code><a href="layer_category_encoding.html">layer_category_encoding</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_concatenate.html">layer_concatenate</a>()</code> <br><code><a href="layer_conv_1d.html">layer_conv_1d</a>()</code> <br><code><a href="layer_conv_1d_transpose.html">layer_conv_1d_transpose</a>()</code> <br><code><a href="layer_conv_2d.html">layer_conv_2d</a>()</code> <br><code><a href="layer_conv_2d_transpose.html">layer_conv_2d_transpose</a>()</code> <br><code><a href="layer_conv_3d.html">layer_conv_3d</a>()</code> <br><code><a href="layer_conv_3d_transpose.html">layer_conv_3d_transpose</a>()</code> <br><code><a href="layer_conv_lstm_1d.html">layer_conv_lstm_1d</a>()</code> <br><code><a href="layer_conv_lstm_2d.html">layer_conv_lstm_2d</a>()</code> <br><code><a href="layer_conv_lstm_3d.html">layer_conv_lstm_3d</a>()</code> <br><code><a href="layer_cropping_1d.html">layer_cropping_1d</a>()</code> <br><code><a href="layer_cropping_2d.html">layer_cropping_2d</a>()</code> <br><code><a href="layer_cropping_3d.html">layer_cropping_3d</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_dense.html">layer_dense</a>()</code> <br><code><a href="layer_depthwise_conv_1d.html">layer_depthwise_conv_1d</a>()</code> <br><code><a href="layer_depthwise_conv_2d.html">layer_depthwise_conv_2d</a>()</code> <br><code><a href="layer_discretization.html">layer_discretization</a>()</code> <br><code><a href="layer_dot.html">layer_dot</a>()</code> <br><code><a href="layer_dropout.html">layer_dropout</a>()</code> <br><code><a href="layer_einsum_dense.html">layer_einsum_dense</a>()</code> <br><code><a href="layer_embedding.html">layer_embedding</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_feature_space.html">layer_feature_space</a>()</code> <br><code><a href="layer_flatten.html">layer_flatten</a>()</code> <br><code><a href="layer_flax_module_wrapper.html">layer_flax_module_wrapper</a>()</code> <br><code><a href="layer_gaussian_dropout.html">layer_gaussian_dropout</a>()</code> <br><code><a href="layer_gaussian_noise.html">layer_gaussian_noise</a>()</code> <br><code><a href="layer_global_average_pooling_1d.html">layer_global_average_pooling_1d</a>()</code> <br><code><a href="layer_global_average_pooling_2d.html">layer_global_average_pooling_2d</a>()</code> <br><code><a href="layer_global_average_pooling_3d.html">layer_global_average_pooling_3d</a>()</code> <br><code><a href="layer_global_max_pooling_1d.html">layer_global_max_pooling_1d</a>()</code> <br><code><a href="layer_global_max_pooling_2d.html">layer_global_max_pooling_2d</a>()</code> <br><code><a href="layer_global_max_pooling_3d.html">layer_global_max_pooling_3d</a>()</code> <br><code><a href="layer_group_normalization.html">layer_group_normalization</a>()</code> <br><code><a href="layer_group_query_attention.html">layer_group_query_attention</a>()</code> <br><code><a href="layer_gru.html">layer_gru</a>()</code> <br><code><a href="layer_hashed_crossing.html">layer_hashed_crossing</a>()</code> <br><code><a href="layer_hashing.html">layer_hashing</a>()</code> <br><code><a href="layer_identity.html">layer_identity</a>()</code> <br><code><a href="layer_integer_lookup.html">layer_integer_lookup</a>()</code> <br><code><a href="layer_jax_model_wrapper.html">layer_jax_model_wrapper</a>()</code> <br><code><a href="layer_lambda.html">layer_lambda</a>()</code> <br><code><a href="layer_layer_normalization.html">layer_layer_normalization</a>()</code> <br><code><a href="layer_lstm.html">layer_lstm</a>()</code> <br><code><a href="layer_masking.html">layer_masking</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_max_pooling_1d.html">layer_max_pooling_1d</a>()</code> <br><code><a href="layer_max_pooling_2d.html">layer_max_pooling_2d</a>()</code> <br><code><a href="layer_max_pooling_3d.html">layer_max_pooling_3d</a>()</code> <br><code><a href="layer_maximum.html">layer_maximum</a>()</code> <br><code><a href="layer_mel_spectrogram.html">layer_mel_spectrogram</a>()</code> <br><code><a href="layer_minimum.html">layer_minimum</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_multi_head_attention.html">layer_multi_head_attention</a>()</code> <br><code><a href="layer_multiply.html">layer_multiply</a>()</code> <br><code><a href="layer_normalization.html">layer_normalization</a>()</code> <br><code><a href="layer_permute.html">layer_permute</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_brightness.html">layer_random_brightness</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_contrast.html">layer_random_contrast</a>()</code> <br><code><a href="layer_random_crop.html">layer_random_crop</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_flip.html">layer_random_flip</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_rotation.html">layer_random_rotation</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_random_translation.html">layer_random_translation</a>()</code> <br><code><a href="layer_random_zoom.html">layer_random_zoom</a>()</code> <br><code><a href="layer_repeat_vector.html">layer_repeat_vector</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_reshape.html">layer_reshape</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_rms_normalization.html">layer_rms_normalization</a>()</code> <br><code><a href="layer_rnn.html">layer_rnn</a>()</code> <br><code><a href="layer_separable_conv_1d.html">layer_separable_conv_1d</a>()</code> <br><code><a href="layer_separable_conv_2d.html">layer_separable_conv_2d</a>()</code> <br><code><a href="layer_simple_rnn.html">layer_simple_rnn</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br><code><a href="layer_spatial_dropout_1d.html">layer_spatial_dropout_1d</a>()</code> <br><code><a href="layer_spatial_dropout_2d.html">layer_spatial_dropout_2d</a>()</code> <br><code><a href="layer_spatial_dropout_3d.html">layer_spatial_dropout_3d</a>()</code> <br><code><a href="layer_spectral_normalization.html">layer_spectral_normalization</a>()</code> <br><code><a href="layer_stft_spectrogram.html">layer_stft_spectrogram</a>()</code> <br><code><a href="layer_string_lookup.html">layer_string_lookup</a>()</code> <br><code><a href="layer_subtract.html">layer_subtract</a>()</code> <br><code><a href="layer_text_vectorization.html">layer_text_vectorization</a>()</code> <br><code><a href="layer_tfsm.html">layer_tfsm</a>()</code> <br><code><a href="layer_time_distributed.html">layer_time_distributed</a>()</code> <br><code><a href="layer_torch_module_wrapper.html">layer_torch_module_wrapper</a>()</code> <br><code><a href="layer_unit_normalization.html">layer_unit_normalization</a>()</code> <br><code><a href="layer_upsampling_1d.html">layer_upsampling_1d</a>()</code> <br><code><a href="layer_upsampling_2d.html">layer_upsampling_2d</a>()</code> <br><code><a href="layer_zero_padding_1d.html">layer_zero_padding_1d</a>()</code> <br><code><a href="layer_zero_padding_2d.html">layer_zero_padding_2d</a>()</code> <br><code><a href="layer_zero_padding_3d.html">layer_zero_padding_3d</a>()</code> <br><code><a href="rnn_cell_gru.html">rnn_cell_gru</a>()</code> <br><code><a href="rnn_cell_lstm.html">rnn_cell_lstm</a>()</code> <br><code><a href="rnn_cell_simple.html">rnn_cell_simple</a>()</code> <br><code><a href="rnn_cells_stack.html">rnn_cells_stack</a>()</code> <br></p></div>
+<p>Other layers: <br><code><a href="Layer.html">Layer</a>()</code> <br><code><a href="layer_activation.html">layer_activation</a>()</code> <br><code><a href="layer_activation_elu.html">layer_activation_elu</a>()</code> <br><code><a href="layer_activation_leaky_relu.html">layer_activation_leaky_relu</a>()</code> <br><code><a href="layer_activation_parametric_relu.html">layer_activation_parametric_relu</a>()</code> <br><code><a href="layer_activation_relu.html">layer_activation_relu</a>()</code> <br><code><a href="layer_activation_softmax.html">layer_activation_softmax</a>()</code> <br><code><a href="layer_activity_regularization.html">layer_activity_regularization</a>()</code> <br><code><a href="layer_add.html">layer_add</a>()</code> <br><code><a href="layer_additive_attention.html">layer_additive_attention</a>()</code> <br><code><a href="layer_alpha_dropout.html">layer_alpha_dropout</a>()</code> <br><code><a href="layer_attention.html">layer_attention</a>()</code> <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_average.html">layer_average</a>()</code> <br><code><a href="layer_average_pooling_1d.html">layer_average_pooling_1d</a>()</code> <br><code><a href="layer_average_pooling_2d.html">layer_average_pooling_2d</a>()</code> <br><code><a href="layer_average_pooling_3d.html">layer_average_pooling_3d</a>()</code> <br><code><a href="layer_batch_normalization.html">layer_batch_normalization</a>()</code> <br><code><a href="layer_bidirectional.html">layer_bidirectional</a>()</code> <br><code><a href="layer_category_encoding.html">layer_category_encoding</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_concatenate.html">layer_concatenate</a>()</code> <br><code><a href="layer_conv_1d.html">layer_conv_1d</a>()</code> <br><code><a href="layer_conv_1d_transpose.html">layer_conv_1d_transpose</a>()</code> <br><code><a href="layer_conv_2d.html">layer_conv_2d</a>()</code> <br><code><a href="layer_conv_2d_transpose.html">layer_conv_2d_transpose</a>()</code> <br><code><a href="layer_conv_3d.html">layer_conv_3d</a>()</code> <br><code><a href="layer_conv_3d_transpose.html">layer_conv_3d_transpose</a>()</code> <br><code><a href="layer_conv_lstm_1d.html">layer_conv_lstm_1d</a>()</code> <br><code><a href="layer_conv_lstm_2d.html">layer_conv_lstm_2d</a>()</code> <br><code><a href="layer_conv_lstm_3d.html">layer_conv_lstm_3d</a>()</code> <br><code><a href="layer_cropping_1d.html">layer_cropping_1d</a>()</code> <br><code><a href="layer_cropping_2d.html">layer_cropping_2d</a>()</code> <br><code><a href="layer_cropping_3d.html">layer_cropping_3d</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_dense.html">layer_dense</a>()</code> <br><code><a href="layer_depthwise_conv_1d.html">layer_depthwise_conv_1d</a>()</code> <br><code><a href="layer_depthwise_conv_2d.html">layer_depthwise_conv_2d</a>()</code> <br><code><a href="layer_discretization.html">layer_discretization</a>()</code> <br><code><a href="layer_dot.html">layer_dot</a>()</code> <br><code><a href="layer_dropout.html">layer_dropout</a>()</code> <br><code><a href="layer_einsum_dense.html">layer_einsum_dense</a>()</code> <br><code><a href="layer_embedding.html">layer_embedding</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_feature_space.html">layer_feature_space</a>()</code> <br><code><a href="layer_flatten.html">layer_flatten</a>()</code> <br><code><a href="layer_flax_module_wrapper.html">layer_flax_module_wrapper</a>()</code> <br><code><a href="layer_gaussian_dropout.html">layer_gaussian_dropout</a>()</code> <br><code><a href="layer_gaussian_noise.html">layer_gaussian_noise</a>()</code> <br><code><a href="layer_global_average_pooling_1d.html">layer_global_average_pooling_1d</a>()</code> <br><code><a href="layer_global_average_pooling_2d.html">layer_global_average_pooling_2d</a>()</code> <br><code><a href="layer_global_average_pooling_3d.html">layer_global_average_pooling_3d</a>()</code> <br><code><a href="layer_global_max_pooling_1d.html">layer_global_max_pooling_1d</a>()</code> <br><code><a href="layer_global_max_pooling_2d.html">layer_global_max_pooling_2d</a>()</code> <br><code><a href="layer_global_max_pooling_3d.html">layer_global_max_pooling_3d</a>()</code> <br><code><a href="layer_group_normalization.html">layer_group_normalization</a>()</code> <br><code><a href="layer_group_query_attention.html">layer_group_query_attention</a>()</code> <br><code><a href="layer_gru.html">layer_gru</a>()</code> <br><code><a href="layer_hashed_crossing.html">layer_hashed_crossing</a>()</code> <br><code><a href="layer_hashing.html">layer_hashing</a>()</code> <br><code><a href="layer_identity.html">layer_identity</a>()</code> <br><code><a href="layer_integer_lookup.html">layer_integer_lookup</a>()</code> <br><code><a href="layer_jax_model_wrapper.html">layer_jax_model_wrapper</a>()</code> <br><code><a href="layer_lambda.html">layer_lambda</a>()</code> <br><code><a href="layer_layer_normalization.html">layer_layer_normalization</a>()</code> <br><code><a href="layer_lstm.html">layer_lstm</a>()</code> <br><code><a href="layer_masking.html">layer_masking</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_max_pooling_1d.html">layer_max_pooling_1d</a>()</code> <br><code><a href="layer_max_pooling_2d.html">layer_max_pooling_2d</a>()</code> <br><code><a href="layer_max_pooling_3d.html">layer_max_pooling_3d</a>()</code> <br><code><a href="layer_maximum.html">layer_maximum</a>()</code> <br><code><a href="layer_mel_spectrogram.html">layer_mel_spectrogram</a>()</code> <br><code><a href="layer_minimum.html">layer_minimum</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_multi_head_attention.html">layer_multi_head_attention</a>()</code> <br><code><a href="layer_multiply.html">layer_multiply</a>()</code> <br><code><a href="layer_normalization.html">layer_normalization</a>()</code> <br><code><a href="layer_permute.html">layer_permute</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_brightness.html">layer_random_brightness</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_contrast.html">layer_random_contrast</a>()</code> <br><code><a href="layer_random_crop.html">layer_random_crop</a>()</code> <br><code><a href="layer_random_elastic_transform.html">layer_random_elastic_transform</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_flip.html">layer_random_flip</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_rotation.html">layer_random_rotation</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_random_translation.html">layer_random_translation</a>()</code> <br><code><a href="layer_random_zoom.html">layer_random_zoom</a>()</code> <br><code><a href="layer_repeat_vector.html">layer_repeat_vector</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_reshape.html">layer_reshape</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_rms_normalization.html">layer_rms_normalization</a>()</code> <br><code><a href="layer_rnn.html">layer_rnn</a>()</code> <br><code><a href="layer_separable_conv_1d.html">layer_separable_conv_1d</a>()</code> <br><code><a href="layer_separable_conv_2d.html">layer_separable_conv_2d</a>()</code> <br><code><a href="layer_simple_rnn.html">layer_simple_rnn</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br><code><a href="layer_spatial_dropout_1d.html">layer_spatial_dropout_1d</a>()</code> <br><code><a href="layer_spatial_dropout_2d.html">layer_spatial_dropout_2d</a>()</code> <br><code><a href="layer_spatial_dropout_3d.html">layer_spatial_dropout_3d</a>()</code> <br><code><a href="layer_spectral_normalization.html">layer_spectral_normalization</a>()</code> <br><code><a href="layer_stft_spectrogram.html">layer_stft_spectrogram</a>()</code> <br><code><a href="layer_string_lookup.html">layer_string_lookup</a>()</code> <br><code><a href="layer_subtract.html">layer_subtract</a>()</code> <br><code><a href="layer_text_vectorization.html">layer_text_vectorization</a>()</code> <br><code><a href="layer_tfsm.html">layer_tfsm</a>()</code> <br><code><a href="layer_time_distributed.html">layer_time_distributed</a>()</code> <br><code><a href="layer_torch_module_wrapper.html">layer_torch_module_wrapper</a>()</code> <br><code><a href="layer_unit_normalization.html">layer_unit_normalization</a>()</code> <br><code><a href="layer_upsampling_1d.html">layer_upsampling_1d</a>()</code> <br><code><a href="layer_upsampling_2d.html">layer_upsampling_2d</a>()</code> <br><code><a href="layer_zero_padding_1d.html">layer_zero_padding_1d</a>()</code> <br><code><a href="layer_zero_padding_2d.html">layer_zero_padding_2d</a>()</code> <br><code><a href="layer_zero_padding_3d.html">layer_zero_padding_3d</a>()</code> <br><code><a href="rnn_cell_gru.html">rnn_cell_gru</a>()</code> <br><code><a href="rnn_cell_lstm.html">rnn_cell_lstm</a>()</code> <br><code><a href="rnn_cell_simple.html">rnn_cell_simple</a>()</code> <br><code><a href="rnn_cells_stack.html">rnn_cells_stack</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -145,7 +145,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/layer_zero_padding_1d.html b/docs/dev/reference/layer_zero_padding_1d.html
index 16d26a605..1a73b7dcf 100644
--- a/docs/dev/reference/layer_zero_padding_1d.html
+++ b/docs/dev/reference/layer_zero_padding_1d.html
@@ -8,7 +8,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -152,7 +152,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
     <div class="dont-index">
 <ul><li><p><a href="https://keras.io/api/layers/reshaping_layers/zero_padding1d#zeropadding1d-class" class="external-link">https://keras.io/api/layers/reshaping_layers/zero_padding1d#zeropadding1d-class</a></p></li>
 </ul><p>Other reshaping layers: <br><code><a href="layer_cropping_1d.html">layer_cropping_1d</a>()</code> <br><code><a href="layer_cropping_2d.html">layer_cropping_2d</a>()</code> <br><code><a href="layer_cropping_3d.html">layer_cropping_3d</a>()</code> <br><code><a href="layer_flatten.html">layer_flatten</a>()</code> <br><code><a href="layer_permute.html">layer_permute</a>()</code> <br><code><a href="layer_repeat_vector.html">layer_repeat_vector</a>()</code> <br><code><a href="layer_reshape.html">layer_reshape</a>()</code> <br><code><a href="layer_upsampling_1d.html">layer_upsampling_1d</a>()</code> <br><code><a href="layer_upsampling_2d.html">layer_upsampling_2d</a>()</code> <br><code><a href="layer_upsampling_3d.html">layer_upsampling_3d</a>()</code> <br><code><a href="layer_zero_padding_2d.html">layer_zero_padding_2d</a>()</code> <br><code><a href="layer_zero_padding_3d.html">layer_zero_padding_3d</a>()</code> <br></p>
-<p>Other layers: <br><code><a href="Layer.html">Layer</a>()</code> <br><code><a href="layer_activation.html">layer_activation</a>()</code> <br><code><a href="layer_activation_elu.html">layer_activation_elu</a>()</code> <br><code><a href="layer_activation_leaky_relu.html">layer_activation_leaky_relu</a>()</code> <br><code><a href="layer_activation_parametric_relu.html">layer_activation_parametric_relu</a>()</code> <br><code><a href="layer_activation_relu.html">layer_activation_relu</a>()</code> <br><code><a href="layer_activation_softmax.html">layer_activation_softmax</a>()</code> <br><code><a href="layer_activity_regularization.html">layer_activity_regularization</a>()</code> <br><code><a href="layer_add.html">layer_add</a>()</code> <br><code><a href="layer_additive_attention.html">layer_additive_attention</a>()</code> <br><code><a href="layer_alpha_dropout.html">layer_alpha_dropout</a>()</code> <br><code><a href="layer_attention.html">layer_attention</a>()</code> <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_average.html">layer_average</a>()</code> <br><code><a href="layer_average_pooling_1d.html">layer_average_pooling_1d</a>()</code> <br><code><a href="layer_average_pooling_2d.html">layer_average_pooling_2d</a>()</code> <br><code><a href="layer_average_pooling_3d.html">layer_average_pooling_3d</a>()</code> <br><code><a href="layer_batch_normalization.html">layer_batch_normalization</a>()</code> <br><code><a href="layer_bidirectional.html">layer_bidirectional</a>()</code> <br><code><a href="layer_category_encoding.html">layer_category_encoding</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_concatenate.html">layer_concatenate</a>()</code> <br><code><a href="layer_conv_1d.html">layer_conv_1d</a>()</code> <br><code><a href="layer_conv_1d_transpose.html">layer_conv_1d_transpose</a>()</code> <br><code><a href="layer_conv_2d.html">layer_conv_2d</a>()</code> <br><code><a href="layer_conv_2d_transpose.html">layer_conv_2d_transpose</a>()</code> <br><code><a href="layer_conv_3d.html">layer_conv_3d</a>()</code> <br><code><a href="layer_conv_3d_transpose.html">layer_conv_3d_transpose</a>()</code> <br><code><a href="layer_conv_lstm_1d.html">layer_conv_lstm_1d</a>()</code> <br><code><a href="layer_conv_lstm_2d.html">layer_conv_lstm_2d</a>()</code> <br><code><a href="layer_conv_lstm_3d.html">layer_conv_lstm_3d</a>()</code> <br><code><a href="layer_cropping_1d.html">layer_cropping_1d</a>()</code> <br><code><a href="layer_cropping_2d.html">layer_cropping_2d</a>()</code> <br><code><a href="layer_cropping_3d.html">layer_cropping_3d</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_dense.html">layer_dense</a>()</code> <br><code><a href="layer_depthwise_conv_1d.html">layer_depthwise_conv_1d</a>()</code> <br><code><a href="layer_depthwise_conv_2d.html">layer_depthwise_conv_2d</a>()</code> <br><code><a href="layer_discretization.html">layer_discretization</a>()</code> <br><code><a href="layer_dot.html">layer_dot</a>()</code> <br><code><a href="layer_dropout.html">layer_dropout</a>()</code> <br><code><a href="layer_einsum_dense.html">layer_einsum_dense</a>()</code> <br><code><a href="layer_embedding.html">layer_embedding</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_feature_space.html">layer_feature_space</a>()</code> <br><code><a href="layer_flatten.html">layer_flatten</a>()</code> <br><code><a href="layer_flax_module_wrapper.html">layer_flax_module_wrapper</a>()</code> <br><code><a href="layer_gaussian_dropout.html">layer_gaussian_dropout</a>()</code> <br><code><a href="layer_gaussian_noise.html">layer_gaussian_noise</a>()</code> <br><code><a href="layer_global_average_pooling_1d.html">layer_global_average_pooling_1d</a>()</code> <br><code><a href="layer_global_average_pooling_2d.html">layer_global_average_pooling_2d</a>()</code> <br><code><a href="layer_global_average_pooling_3d.html">layer_global_average_pooling_3d</a>()</code> <br><code><a href="layer_global_max_pooling_1d.html">layer_global_max_pooling_1d</a>()</code> <br><code><a href="layer_global_max_pooling_2d.html">layer_global_max_pooling_2d</a>()</code> <br><code><a href="layer_global_max_pooling_3d.html">layer_global_max_pooling_3d</a>()</code> <br><code><a href="layer_group_normalization.html">layer_group_normalization</a>()</code> <br><code><a href="layer_group_query_attention.html">layer_group_query_attention</a>()</code> <br><code><a href="layer_gru.html">layer_gru</a>()</code> <br><code><a href="layer_hashed_crossing.html">layer_hashed_crossing</a>()</code> <br><code><a href="layer_hashing.html">layer_hashing</a>()</code> <br><code><a href="layer_identity.html">layer_identity</a>()</code> <br><code><a href="layer_integer_lookup.html">layer_integer_lookup</a>()</code> <br><code><a href="layer_jax_model_wrapper.html">layer_jax_model_wrapper</a>()</code> <br><code><a href="layer_lambda.html">layer_lambda</a>()</code> <br><code><a href="layer_layer_normalization.html">layer_layer_normalization</a>()</code> <br><code><a href="layer_lstm.html">layer_lstm</a>()</code> <br><code><a href="layer_masking.html">layer_masking</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_max_pooling_1d.html">layer_max_pooling_1d</a>()</code> <br><code><a href="layer_max_pooling_2d.html">layer_max_pooling_2d</a>()</code> <br><code><a href="layer_max_pooling_3d.html">layer_max_pooling_3d</a>()</code> <br><code><a href="layer_maximum.html">layer_maximum</a>()</code> <br><code><a href="layer_mel_spectrogram.html">layer_mel_spectrogram</a>()</code> <br><code><a href="layer_minimum.html">layer_minimum</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_multi_head_attention.html">layer_multi_head_attention</a>()</code> <br><code><a href="layer_multiply.html">layer_multiply</a>()</code> <br><code><a href="layer_normalization.html">layer_normalization</a>()</code> <br><code><a href="layer_permute.html">layer_permute</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_brightness.html">layer_random_brightness</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_contrast.html">layer_random_contrast</a>()</code> <br><code><a href="layer_random_crop.html">layer_random_crop</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_flip.html">layer_random_flip</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_rotation.html">layer_random_rotation</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_random_translation.html">layer_random_translation</a>()</code> <br><code><a href="layer_random_zoom.html">layer_random_zoom</a>()</code> <br><code><a href="layer_repeat_vector.html">layer_repeat_vector</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_reshape.html">layer_reshape</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_rms_normalization.html">layer_rms_normalization</a>()</code> <br><code><a href="layer_rnn.html">layer_rnn</a>()</code> <br><code><a href="layer_separable_conv_1d.html">layer_separable_conv_1d</a>()</code> <br><code><a href="layer_separable_conv_2d.html">layer_separable_conv_2d</a>()</code> <br><code><a href="layer_simple_rnn.html">layer_simple_rnn</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br><code><a href="layer_spatial_dropout_1d.html">layer_spatial_dropout_1d</a>()</code> <br><code><a href="layer_spatial_dropout_2d.html">layer_spatial_dropout_2d</a>()</code> <br><code><a href="layer_spatial_dropout_3d.html">layer_spatial_dropout_3d</a>()</code> <br><code><a href="layer_spectral_normalization.html">layer_spectral_normalization</a>()</code> <br><code><a href="layer_stft_spectrogram.html">layer_stft_spectrogram</a>()</code> <br><code><a href="layer_string_lookup.html">layer_string_lookup</a>()</code> <br><code><a href="layer_subtract.html">layer_subtract</a>()</code> <br><code><a href="layer_text_vectorization.html">layer_text_vectorization</a>()</code> <br><code><a href="layer_tfsm.html">layer_tfsm</a>()</code> <br><code><a href="layer_time_distributed.html">layer_time_distributed</a>()</code> <br><code><a href="layer_torch_module_wrapper.html">layer_torch_module_wrapper</a>()</code> <br><code><a href="layer_unit_normalization.html">layer_unit_normalization</a>()</code> <br><code><a href="layer_upsampling_1d.html">layer_upsampling_1d</a>()</code> <br><code><a href="layer_upsampling_2d.html">layer_upsampling_2d</a>()</code> <br><code><a href="layer_upsampling_3d.html">layer_upsampling_3d</a>()</code> <br><code><a href="layer_zero_padding_2d.html">layer_zero_padding_2d</a>()</code> <br><code><a href="layer_zero_padding_3d.html">layer_zero_padding_3d</a>()</code> <br><code><a href="rnn_cell_gru.html">rnn_cell_gru</a>()</code> <br><code><a href="rnn_cell_lstm.html">rnn_cell_lstm</a>()</code> <br><code><a href="rnn_cell_simple.html">rnn_cell_simple</a>()</code> <br><code><a href="rnn_cells_stack.html">rnn_cells_stack</a>()</code> <br></p></div>
+<p>Other layers: <br><code><a href="Layer.html">Layer</a>()</code> <br><code><a href="layer_activation.html">layer_activation</a>()</code> <br><code><a href="layer_activation_elu.html">layer_activation_elu</a>()</code> <br><code><a href="layer_activation_leaky_relu.html">layer_activation_leaky_relu</a>()</code> <br><code><a href="layer_activation_parametric_relu.html">layer_activation_parametric_relu</a>()</code> <br><code><a href="layer_activation_relu.html">layer_activation_relu</a>()</code> <br><code><a href="layer_activation_softmax.html">layer_activation_softmax</a>()</code> <br><code><a href="layer_activity_regularization.html">layer_activity_regularization</a>()</code> <br><code><a href="layer_add.html">layer_add</a>()</code> <br><code><a href="layer_additive_attention.html">layer_additive_attention</a>()</code> <br><code><a href="layer_alpha_dropout.html">layer_alpha_dropout</a>()</code> <br><code><a href="layer_attention.html">layer_attention</a>()</code> <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_average.html">layer_average</a>()</code> <br><code><a href="layer_average_pooling_1d.html">layer_average_pooling_1d</a>()</code> <br><code><a href="layer_average_pooling_2d.html">layer_average_pooling_2d</a>()</code> <br><code><a href="layer_average_pooling_3d.html">layer_average_pooling_3d</a>()</code> <br><code><a href="layer_batch_normalization.html">layer_batch_normalization</a>()</code> <br><code><a href="layer_bidirectional.html">layer_bidirectional</a>()</code> <br><code><a href="layer_category_encoding.html">layer_category_encoding</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_concatenate.html">layer_concatenate</a>()</code> <br><code><a href="layer_conv_1d.html">layer_conv_1d</a>()</code> <br><code><a href="layer_conv_1d_transpose.html">layer_conv_1d_transpose</a>()</code> <br><code><a href="layer_conv_2d.html">layer_conv_2d</a>()</code> <br><code><a href="layer_conv_2d_transpose.html">layer_conv_2d_transpose</a>()</code> <br><code><a href="layer_conv_3d.html">layer_conv_3d</a>()</code> <br><code><a href="layer_conv_3d_transpose.html">layer_conv_3d_transpose</a>()</code> <br><code><a href="layer_conv_lstm_1d.html">layer_conv_lstm_1d</a>()</code> <br><code><a href="layer_conv_lstm_2d.html">layer_conv_lstm_2d</a>()</code> <br><code><a href="layer_conv_lstm_3d.html">layer_conv_lstm_3d</a>()</code> <br><code><a href="layer_cropping_1d.html">layer_cropping_1d</a>()</code> <br><code><a href="layer_cropping_2d.html">layer_cropping_2d</a>()</code> <br><code><a href="layer_cropping_3d.html">layer_cropping_3d</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_dense.html">layer_dense</a>()</code> <br><code><a href="layer_depthwise_conv_1d.html">layer_depthwise_conv_1d</a>()</code> <br><code><a href="layer_depthwise_conv_2d.html">layer_depthwise_conv_2d</a>()</code> <br><code><a href="layer_discretization.html">layer_discretization</a>()</code> <br><code><a href="layer_dot.html">layer_dot</a>()</code> <br><code><a href="layer_dropout.html">layer_dropout</a>()</code> <br><code><a href="layer_einsum_dense.html">layer_einsum_dense</a>()</code> <br><code><a href="layer_embedding.html">layer_embedding</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_feature_space.html">layer_feature_space</a>()</code> <br><code><a href="layer_flatten.html">layer_flatten</a>()</code> <br><code><a href="layer_flax_module_wrapper.html">layer_flax_module_wrapper</a>()</code> <br><code><a href="layer_gaussian_dropout.html">layer_gaussian_dropout</a>()</code> <br><code><a href="layer_gaussian_noise.html">layer_gaussian_noise</a>()</code> <br><code><a href="layer_global_average_pooling_1d.html">layer_global_average_pooling_1d</a>()</code> <br><code><a href="layer_global_average_pooling_2d.html">layer_global_average_pooling_2d</a>()</code> <br><code><a href="layer_global_average_pooling_3d.html">layer_global_average_pooling_3d</a>()</code> <br><code><a href="layer_global_max_pooling_1d.html">layer_global_max_pooling_1d</a>()</code> <br><code><a href="layer_global_max_pooling_2d.html">layer_global_max_pooling_2d</a>()</code> <br><code><a href="layer_global_max_pooling_3d.html">layer_global_max_pooling_3d</a>()</code> <br><code><a href="layer_group_normalization.html">layer_group_normalization</a>()</code> <br><code><a href="layer_group_query_attention.html">layer_group_query_attention</a>()</code> <br><code><a href="layer_gru.html">layer_gru</a>()</code> <br><code><a href="layer_hashed_crossing.html">layer_hashed_crossing</a>()</code> <br><code><a href="layer_hashing.html">layer_hashing</a>()</code> <br><code><a href="layer_identity.html">layer_identity</a>()</code> <br><code><a href="layer_integer_lookup.html">layer_integer_lookup</a>()</code> <br><code><a href="layer_jax_model_wrapper.html">layer_jax_model_wrapper</a>()</code> <br><code><a href="layer_lambda.html">layer_lambda</a>()</code> <br><code><a href="layer_layer_normalization.html">layer_layer_normalization</a>()</code> <br><code><a href="layer_lstm.html">layer_lstm</a>()</code> <br><code><a href="layer_masking.html">layer_masking</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_max_pooling_1d.html">layer_max_pooling_1d</a>()</code> <br><code><a href="layer_max_pooling_2d.html">layer_max_pooling_2d</a>()</code> <br><code><a href="layer_max_pooling_3d.html">layer_max_pooling_3d</a>()</code> <br><code><a href="layer_maximum.html">layer_maximum</a>()</code> <br><code><a href="layer_mel_spectrogram.html">layer_mel_spectrogram</a>()</code> <br><code><a href="layer_minimum.html">layer_minimum</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_multi_head_attention.html">layer_multi_head_attention</a>()</code> <br><code><a href="layer_multiply.html">layer_multiply</a>()</code> <br><code><a href="layer_normalization.html">layer_normalization</a>()</code> <br><code><a href="layer_permute.html">layer_permute</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_brightness.html">layer_random_brightness</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_contrast.html">layer_random_contrast</a>()</code> <br><code><a href="layer_random_crop.html">layer_random_crop</a>()</code> <br><code><a href="layer_random_elastic_transform.html">layer_random_elastic_transform</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_flip.html">layer_random_flip</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_rotation.html">layer_random_rotation</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_random_translation.html">layer_random_translation</a>()</code> <br><code><a href="layer_random_zoom.html">layer_random_zoom</a>()</code> <br><code><a href="layer_repeat_vector.html">layer_repeat_vector</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_reshape.html">layer_reshape</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_rms_normalization.html">layer_rms_normalization</a>()</code> <br><code><a href="layer_rnn.html">layer_rnn</a>()</code> <br><code><a href="layer_separable_conv_1d.html">layer_separable_conv_1d</a>()</code> <br><code><a href="layer_separable_conv_2d.html">layer_separable_conv_2d</a>()</code> <br><code><a href="layer_simple_rnn.html">layer_simple_rnn</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br><code><a href="layer_spatial_dropout_1d.html">layer_spatial_dropout_1d</a>()</code> <br><code><a href="layer_spatial_dropout_2d.html">layer_spatial_dropout_2d</a>()</code> <br><code><a href="layer_spatial_dropout_3d.html">layer_spatial_dropout_3d</a>()</code> <br><code><a href="layer_spectral_normalization.html">layer_spectral_normalization</a>()</code> <br><code><a href="layer_stft_spectrogram.html">layer_stft_spectrogram</a>()</code> <br><code><a href="layer_string_lookup.html">layer_string_lookup</a>()</code> <br><code><a href="layer_subtract.html">layer_subtract</a>()</code> <br><code><a href="layer_text_vectorization.html">layer_text_vectorization</a>()</code> <br><code><a href="layer_tfsm.html">layer_tfsm</a>()</code> <br><code><a href="layer_time_distributed.html">layer_time_distributed</a>()</code> <br><code><a href="layer_torch_module_wrapper.html">layer_torch_module_wrapper</a>()</code> <br><code><a href="layer_unit_normalization.html">layer_unit_normalization</a>()</code> <br><code><a href="layer_upsampling_1d.html">layer_upsampling_1d</a>()</code> <br><code><a href="layer_upsampling_2d.html">layer_upsampling_2d</a>()</code> <br><code><a href="layer_upsampling_3d.html">layer_upsampling_3d</a>()</code> <br><code><a href="layer_zero_padding_2d.html">layer_zero_padding_2d</a>()</code> <br><code><a href="layer_zero_padding_3d.html">layer_zero_padding_3d</a>()</code> <br><code><a href="rnn_cell_gru.html">rnn_cell_gru</a>()</code> <br><code><a href="rnn_cell_lstm.html">rnn_cell_lstm</a>()</code> <br><code><a href="rnn_cell_simple.html">rnn_cell_simple</a>()</code> <br><code><a href="rnn_cells_stack.html">rnn_cells_stack</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -164,7 +164,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/layer_zero_padding_2d.html b/docs/dev/reference/layer_zero_padding_2d.html
index 431897677..5f393a20f 100644
--- a/docs/dev/reference/layer_zero_padding_2d.html
+++ b/docs/dev/reference/layer_zero_padding_2d.html
@@ -10,7 +10,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -155,7 +155,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
     <div class="dont-index">
 <ul><li><p><a href="https://keras.io/api/layers/reshaping_layers/zero_padding2d#zeropadding2d-class" class="external-link">https://keras.io/api/layers/reshaping_layers/zero_padding2d#zeropadding2d-class</a></p></li>
 </ul><p>Other reshaping layers: <br><code><a href="layer_cropping_1d.html">layer_cropping_1d</a>()</code> <br><code><a href="layer_cropping_2d.html">layer_cropping_2d</a>()</code> <br><code><a href="layer_cropping_3d.html">layer_cropping_3d</a>()</code> <br><code><a href="layer_flatten.html">layer_flatten</a>()</code> <br><code><a href="layer_permute.html">layer_permute</a>()</code> <br><code><a href="layer_repeat_vector.html">layer_repeat_vector</a>()</code> <br><code><a href="layer_reshape.html">layer_reshape</a>()</code> <br><code><a href="layer_upsampling_1d.html">layer_upsampling_1d</a>()</code> <br><code><a href="layer_upsampling_2d.html">layer_upsampling_2d</a>()</code> <br><code><a href="layer_upsampling_3d.html">layer_upsampling_3d</a>()</code> <br><code><a href="layer_zero_padding_1d.html">layer_zero_padding_1d</a>()</code> <br><code><a href="layer_zero_padding_3d.html">layer_zero_padding_3d</a>()</code> <br></p>
-<p>Other layers: <br><code><a href="Layer.html">Layer</a>()</code> <br><code><a href="layer_activation.html">layer_activation</a>()</code> <br><code><a href="layer_activation_elu.html">layer_activation_elu</a>()</code> <br><code><a href="layer_activation_leaky_relu.html">layer_activation_leaky_relu</a>()</code> <br><code><a href="layer_activation_parametric_relu.html">layer_activation_parametric_relu</a>()</code> <br><code><a href="layer_activation_relu.html">layer_activation_relu</a>()</code> <br><code><a href="layer_activation_softmax.html">layer_activation_softmax</a>()</code> <br><code><a href="layer_activity_regularization.html">layer_activity_regularization</a>()</code> <br><code><a href="layer_add.html">layer_add</a>()</code> <br><code><a href="layer_additive_attention.html">layer_additive_attention</a>()</code> <br><code><a href="layer_alpha_dropout.html">layer_alpha_dropout</a>()</code> <br><code><a href="layer_attention.html">layer_attention</a>()</code> <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_average.html">layer_average</a>()</code> <br><code><a href="layer_average_pooling_1d.html">layer_average_pooling_1d</a>()</code> <br><code><a href="layer_average_pooling_2d.html">layer_average_pooling_2d</a>()</code> <br><code><a href="layer_average_pooling_3d.html">layer_average_pooling_3d</a>()</code> <br><code><a href="layer_batch_normalization.html">layer_batch_normalization</a>()</code> <br><code><a href="layer_bidirectional.html">layer_bidirectional</a>()</code> <br><code><a href="layer_category_encoding.html">layer_category_encoding</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_concatenate.html">layer_concatenate</a>()</code> <br><code><a href="layer_conv_1d.html">layer_conv_1d</a>()</code> <br><code><a href="layer_conv_1d_transpose.html">layer_conv_1d_transpose</a>()</code> <br><code><a href="layer_conv_2d.html">layer_conv_2d</a>()</code> <br><code><a href="layer_conv_2d_transpose.html">layer_conv_2d_transpose</a>()</code> <br><code><a href="layer_conv_3d.html">layer_conv_3d</a>()</code> <br><code><a href="layer_conv_3d_transpose.html">layer_conv_3d_transpose</a>()</code> <br><code><a href="layer_conv_lstm_1d.html">layer_conv_lstm_1d</a>()</code> <br><code><a href="layer_conv_lstm_2d.html">layer_conv_lstm_2d</a>()</code> <br><code><a href="layer_conv_lstm_3d.html">layer_conv_lstm_3d</a>()</code> <br><code><a href="layer_cropping_1d.html">layer_cropping_1d</a>()</code> <br><code><a href="layer_cropping_2d.html">layer_cropping_2d</a>()</code> <br><code><a href="layer_cropping_3d.html">layer_cropping_3d</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_dense.html">layer_dense</a>()</code> <br><code><a href="layer_depthwise_conv_1d.html">layer_depthwise_conv_1d</a>()</code> <br><code><a href="layer_depthwise_conv_2d.html">layer_depthwise_conv_2d</a>()</code> <br><code><a href="layer_discretization.html">layer_discretization</a>()</code> <br><code><a href="layer_dot.html">layer_dot</a>()</code> <br><code><a href="layer_dropout.html">layer_dropout</a>()</code> <br><code><a href="layer_einsum_dense.html">layer_einsum_dense</a>()</code> <br><code><a href="layer_embedding.html">layer_embedding</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_feature_space.html">layer_feature_space</a>()</code> <br><code><a href="layer_flatten.html">layer_flatten</a>()</code> <br><code><a href="layer_flax_module_wrapper.html">layer_flax_module_wrapper</a>()</code> <br><code><a href="layer_gaussian_dropout.html">layer_gaussian_dropout</a>()</code> <br><code><a href="layer_gaussian_noise.html">layer_gaussian_noise</a>()</code> <br><code><a href="layer_global_average_pooling_1d.html">layer_global_average_pooling_1d</a>()</code> <br><code><a href="layer_global_average_pooling_2d.html">layer_global_average_pooling_2d</a>()</code> <br><code><a href="layer_global_average_pooling_3d.html">layer_global_average_pooling_3d</a>()</code> <br><code><a href="layer_global_max_pooling_1d.html">layer_global_max_pooling_1d</a>()</code> <br><code><a href="layer_global_max_pooling_2d.html">layer_global_max_pooling_2d</a>()</code> <br><code><a href="layer_global_max_pooling_3d.html">layer_global_max_pooling_3d</a>()</code> <br><code><a href="layer_group_normalization.html">layer_group_normalization</a>()</code> <br><code><a href="layer_group_query_attention.html">layer_group_query_attention</a>()</code> <br><code><a href="layer_gru.html">layer_gru</a>()</code> <br><code><a href="layer_hashed_crossing.html">layer_hashed_crossing</a>()</code> <br><code><a href="layer_hashing.html">layer_hashing</a>()</code> <br><code><a href="layer_identity.html">layer_identity</a>()</code> <br><code><a href="layer_integer_lookup.html">layer_integer_lookup</a>()</code> <br><code><a href="layer_jax_model_wrapper.html">layer_jax_model_wrapper</a>()</code> <br><code><a href="layer_lambda.html">layer_lambda</a>()</code> <br><code><a href="layer_layer_normalization.html">layer_layer_normalization</a>()</code> <br><code><a href="layer_lstm.html">layer_lstm</a>()</code> <br><code><a href="layer_masking.html">layer_masking</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_max_pooling_1d.html">layer_max_pooling_1d</a>()</code> <br><code><a href="layer_max_pooling_2d.html">layer_max_pooling_2d</a>()</code> <br><code><a href="layer_max_pooling_3d.html">layer_max_pooling_3d</a>()</code> <br><code><a href="layer_maximum.html">layer_maximum</a>()</code> <br><code><a href="layer_mel_spectrogram.html">layer_mel_spectrogram</a>()</code> <br><code><a href="layer_minimum.html">layer_minimum</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_multi_head_attention.html">layer_multi_head_attention</a>()</code> <br><code><a href="layer_multiply.html">layer_multiply</a>()</code> <br><code><a href="layer_normalization.html">layer_normalization</a>()</code> <br><code><a href="layer_permute.html">layer_permute</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_brightness.html">layer_random_brightness</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_contrast.html">layer_random_contrast</a>()</code> <br><code><a href="layer_random_crop.html">layer_random_crop</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_flip.html">layer_random_flip</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_rotation.html">layer_random_rotation</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_random_translation.html">layer_random_translation</a>()</code> <br><code><a href="layer_random_zoom.html">layer_random_zoom</a>()</code> <br><code><a href="layer_repeat_vector.html">layer_repeat_vector</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_reshape.html">layer_reshape</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_rms_normalization.html">layer_rms_normalization</a>()</code> <br><code><a href="layer_rnn.html">layer_rnn</a>()</code> <br><code><a href="layer_separable_conv_1d.html">layer_separable_conv_1d</a>()</code> <br><code><a href="layer_separable_conv_2d.html">layer_separable_conv_2d</a>()</code> <br><code><a href="layer_simple_rnn.html">layer_simple_rnn</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br><code><a href="layer_spatial_dropout_1d.html">layer_spatial_dropout_1d</a>()</code> <br><code><a href="layer_spatial_dropout_2d.html">layer_spatial_dropout_2d</a>()</code> <br><code><a href="layer_spatial_dropout_3d.html">layer_spatial_dropout_3d</a>()</code> <br><code><a href="layer_spectral_normalization.html">layer_spectral_normalization</a>()</code> <br><code><a href="layer_stft_spectrogram.html">layer_stft_spectrogram</a>()</code> <br><code><a href="layer_string_lookup.html">layer_string_lookup</a>()</code> <br><code><a href="layer_subtract.html">layer_subtract</a>()</code> <br><code><a href="layer_text_vectorization.html">layer_text_vectorization</a>()</code> <br><code><a href="layer_tfsm.html">layer_tfsm</a>()</code> <br><code><a href="layer_time_distributed.html">layer_time_distributed</a>()</code> <br><code><a href="layer_torch_module_wrapper.html">layer_torch_module_wrapper</a>()</code> <br><code><a href="layer_unit_normalization.html">layer_unit_normalization</a>()</code> <br><code><a href="layer_upsampling_1d.html">layer_upsampling_1d</a>()</code> <br><code><a href="layer_upsampling_2d.html">layer_upsampling_2d</a>()</code> <br><code><a href="layer_upsampling_3d.html">layer_upsampling_3d</a>()</code> <br><code><a href="layer_zero_padding_1d.html">layer_zero_padding_1d</a>()</code> <br><code><a href="layer_zero_padding_3d.html">layer_zero_padding_3d</a>()</code> <br><code><a href="rnn_cell_gru.html">rnn_cell_gru</a>()</code> <br><code><a href="rnn_cell_lstm.html">rnn_cell_lstm</a>()</code> <br><code><a href="rnn_cell_simple.html">rnn_cell_simple</a>()</code> <br><code><a href="rnn_cells_stack.html">rnn_cells_stack</a>()</code> <br></p></div>
+<p>Other layers: <br><code><a href="Layer.html">Layer</a>()</code> <br><code><a href="layer_activation.html">layer_activation</a>()</code> <br><code><a href="layer_activation_elu.html">layer_activation_elu</a>()</code> <br><code><a href="layer_activation_leaky_relu.html">layer_activation_leaky_relu</a>()</code> <br><code><a href="layer_activation_parametric_relu.html">layer_activation_parametric_relu</a>()</code> <br><code><a href="layer_activation_relu.html">layer_activation_relu</a>()</code> <br><code><a href="layer_activation_softmax.html">layer_activation_softmax</a>()</code> <br><code><a href="layer_activity_regularization.html">layer_activity_regularization</a>()</code> <br><code><a href="layer_add.html">layer_add</a>()</code> <br><code><a href="layer_additive_attention.html">layer_additive_attention</a>()</code> <br><code><a href="layer_alpha_dropout.html">layer_alpha_dropout</a>()</code> <br><code><a href="layer_attention.html">layer_attention</a>()</code> <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_average.html">layer_average</a>()</code> <br><code><a href="layer_average_pooling_1d.html">layer_average_pooling_1d</a>()</code> <br><code><a href="layer_average_pooling_2d.html">layer_average_pooling_2d</a>()</code> <br><code><a href="layer_average_pooling_3d.html">layer_average_pooling_3d</a>()</code> <br><code><a href="layer_batch_normalization.html">layer_batch_normalization</a>()</code> <br><code><a href="layer_bidirectional.html">layer_bidirectional</a>()</code> <br><code><a href="layer_category_encoding.html">layer_category_encoding</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_concatenate.html">layer_concatenate</a>()</code> <br><code><a href="layer_conv_1d.html">layer_conv_1d</a>()</code> <br><code><a href="layer_conv_1d_transpose.html">layer_conv_1d_transpose</a>()</code> <br><code><a href="layer_conv_2d.html">layer_conv_2d</a>()</code> <br><code><a href="layer_conv_2d_transpose.html">layer_conv_2d_transpose</a>()</code> <br><code><a href="layer_conv_3d.html">layer_conv_3d</a>()</code> <br><code><a href="layer_conv_3d_transpose.html">layer_conv_3d_transpose</a>()</code> <br><code><a href="layer_conv_lstm_1d.html">layer_conv_lstm_1d</a>()</code> <br><code><a href="layer_conv_lstm_2d.html">layer_conv_lstm_2d</a>()</code> <br><code><a href="layer_conv_lstm_3d.html">layer_conv_lstm_3d</a>()</code> <br><code><a href="layer_cropping_1d.html">layer_cropping_1d</a>()</code> <br><code><a href="layer_cropping_2d.html">layer_cropping_2d</a>()</code> <br><code><a href="layer_cropping_3d.html">layer_cropping_3d</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_dense.html">layer_dense</a>()</code> <br><code><a href="layer_depthwise_conv_1d.html">layer_depthwise_conv_1d</a>()</code> <br><code><a href="layer_depthwise_conv_2d.html">layer_depthwise_conv_2d</a>()</code> <br><code><a href="layer_discretization.html">layer_discretization</a>()</code> <br><code><a href="layer_dot.html">layer_dot</a>()</code> <br><code><a href="layer_dropout.html">layer_dropout</a>()</code> <br><code><a href="layer_einsum_dense.html">layer_einsum_dense</a>()</code> <br><code><a href="layer_embedding.html">layer_embedding</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_feature_space.html">layer_feature_space</a>()</code> <br><code><a href="layer_flatten.html">layer_flatten</a>()</code> <br><code><a href="layer_flax_module_wrapper.html">layer_flax_module_wrapper</a>()</code> <br><code><a href="layer_gaussian_dropout.html">layer_gaussian_dropout</a>()</code> <br><code><a href="layer_gaussian_noise.html">layer_gaussian_noise</a>()</code> <br><code><a href="layer_global_average_pooling_1d.html">layer_global_average_pooling_1d</a>()</code> <br><code><a href="layer_global_average_pooling_2d.html">layer_global_average_pooling_2d</a>()</code> <br><code><a href="layer_global_average_pooling_3d.html">layer_global_average_pooling_3d</a>()</code> <br><code><a href="layer_global_max_pooling_1d.html">layer_global_max_pooling_1d</a>()</code> <br><code><a href="layer_global_max_pooling_2d.html">layer_global_max_pooling_2d</a>()</code> <br><code><a href="layer_global_max_pooling_3d.html">layer_global_max_pooling_3d</a>()</code> <br><code><a href="layer_group_normalization.html">layer_group_normalization</a>()</code> <br><code><a href="layer_group_query_attention.html">layer_group_query_attention</a>()</code> <br><code><a href="layer_gru.html">layer_gru</a>()</code> <br><code><a href="layer_hashed_crossing.html">layer_hashed_crossing</a>()</code> <br><code><a href="layer_hashing.html">layer_hashing</a>()</code> <br><code><a href="layer_identity.html">layer_identity</a>()</code> <br><code><a href="layer_integer_lookup.html">layer_integer_lookup</a>()</code> <br><code><a href="layer_jax_model_wrapper.html">layer_jax_model_wrapper</a>()</code> <br><code><a href="layer_lambda.html">layer_lambda</a>()</code> <br><code><a href="layer_layer_normalization.html">layer_layer_normalization</a>()</code> <br><code><a href="layer_lstm.html">layer_lstm</a>()</code> <br><code><a href="layer_masking.html">layer_masking</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_max_pooling_1d.html">layer_max_pooling_1d</a>()</code> <br><code><a href="layer_max_pooling_2d.html">layer_max_pooling_2d</a>()</code> <br><code><a href="layer_max_pooling_3d.html">layer_max_pooling_3d</a>()</code> <br><code><a href="layer_maximum.html">layer_maximum</a>()</code> <br><code><a href="layer_mel_spectrogram.html">layer_mel_spectrogram</a>()</code> <br><code><a href="layer_minimum.html">layer_minimum</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_multi_head_attention.html">layer_multi_head_attention</a>()</code> <br><code><a href="layer_multiply.html">layer_multiply</a>()</code> <br><code><a href="layer_normalization.html">layer_normalization</a>()</code> <br><code><a href="layer_permute.html">layer_permute</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_brightness.html">layer_random_brightness</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_contrast.html">layer_random_contrast</a>()</code> <br><code><a href="layer_random_crop.html">layer_random_crop</a>()</code> <br><code><a href="layer_random_elastic_transform.html">layer_random_elastic_transform</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_flip.html">layer_random_flip</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_rotation.html">layer_random_rotation</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_random_translation.html">layer_random_translation</a>()</code> <br><code><a href="layer_random_zoom.html">layer_random_zoom</a>()</code> <br><code><a href="layer_repeat_vector.html">layer_repeat_vector</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_reshape.html">layer_reshape</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_rms_normalization.html">layer_rms_normalization</a>()</code> <br><code><a href="layer_rnn.html">layer_rnn</a>()</code> <br><code><a href="layer_separable_conv_1d.html">layer_separable_conv_1d</a>()</code> <br><code><a href="layer_separable_conv_2d.html">layer_separable_conv_2d</a>()</code> <br><code><a href="layer_simple_rnn.html">layer_simple_rnn</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br><code><a href="layer_spatial_dropout_1d.html">layer_spatial_dropout_1d</a>()</code> <br><code><a href="layer_spatial_dropout_2d.html">layer_spatial_dropout_2d</a>()</code> <br><code><a href="layer_spatial_dropout_3d.html">layer_spatial_dropout_3d</a>()</code> <br><code><a href="layer_spectral_normalization.html">layer_spectral_normalization</a>()</code> <br><code><a href="layer_stft_spectrogram.html">layer_stft_spectrogram</a>()</code> <br><code><a href="layer_string_lookup.html">layer_string_lookup</a>()</code> <br><code><a href="layer_subtract.html">layer_subtract</a>()</code> <br><code><a href="layer_text_vectorization.html">layer_text_vectorization</a>()</code> <br><code><a href="layer_tfsm.html">layer_tfsm</a>()</code> <br><code><a href="layer_time_distributed.html">layer_time_distributed</a>()</code> <br><code><a href="layer_torch_module_wrapper.html">layer_torch_module_wrapper</a>()</code> <br><code><a href="layer_unit_normalization.html">layer_unit_normalization</a>()</code> <br><code><a href="layer_upsampling_1d.html">layer_upsampling_1d</a>()</code> <br><code><a href="layer_upsampling_2d.html">layer_upsampling_2d</a>()</code> <br><code><a href="layer_upsampling_3d.html">layer_upsampling_3d</a>()</code> <br><code><a href="layer_zero_padding_1d.html">layer_zero_padding_1d</a>()</code> <br><code><a href="layer_zero_padding_3d.html">layer_zero_padding_3d</a>()</code> <br><code><a href="rnn_cell_gru.html">rnn_cell_gru</a>()</code> <br><code><a href="rnn_cell_lstm.html">rnn_cell_lstm</a>()</code> <br><code><a href="rnn_cell_simple.html">rnn_cell_simple</a>()</code> <br><code><a href="rnn_cells_stack.html">rnn_cells_stack</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -167,7 +167,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/layer_zero_padding_3d.html b/docs/dev/reference/layer_zero_padding_3d.html
index ecfaf4233..ec3196913 100644
--- a/docs/dev/reference/layer_zero_padding_3d.html
+++ b/docs/dev/reference/layer_zero_padding_3d.html
@@ -8,7 +8,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -147,7 +147,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
     <div class="dont-index">
 <ul><li><p><a href="https://keras.io/api/layers/reshaping_layers/zero_padding3d#zeropadding3d-class" class="external-link">https://keras.io/api/layers/reshaping_layers/zero_padding3d#zeropadding3d-class</a></p></li>
 </ul><p>Other reshaping layers: <br><code><a href="layer_cropping_1d.html">layer_cropping_1d</a>()</code> <br><code><a href="layer_cropping_2d.html">layer_cropping_2d</a>()</code> <br><code><a href="layer_cropping_3d.html">layer_cropping_3d</a>()</code> <br><code><a href="layer_flatten.html">layer_flatten</a>()</code> <br><code><a href="layer_permute.html">layer_permute</a>()</code> <br><code><a href="layer_repeat_vector.html">layer_repeat_vector</a>()</code> <br><code><a href="layer_reshape.html">layer_reshape</a>()</code> <br><code><a href="layer_upsampling_1d.html">layer_upsampling_1d</a>()</code> <br><code><a href="layer_upsampling_2d.html">layer_upsampling_2d</a>()</code> <br><code><a href="layer_upsampling_3d.html">layer_upsampling_3d</a>()</code> <br><code><a href="layer_zero_padding_1d.html">layer_zero_padding_1d</a>()</code> <br><code><a href="layer_zero_padding_2d.html">layer_zero_padding_2d</a>()</code> <br></p>
-<p>Other layers: <br><code><a href="Layer.html">Layer</a>()</code> <br><code><a href="layer_activation.html">layer_activation</a>()</code> <br><code><a href="layer_activation_elu.html">layer_activation_elu</a>()</code> <br><code><a href="layer_activation_leaky_relu.html">layer_activation_leaky_relu</a>()</code> <br><code><a href="layer_activation_parametric_relu.html">layer_activation_parametric_relu</a>()</code> <br><code><a href="layer_activation_relu.html">layer_activation_relu</a>()</code> <br><code><a href="layer_activation_softmax.html">layer_activation_softmax</a>()</code> <br><code><a href="layer_activity_regularization.html">layer_activity_regularization</a>()</code> <br><code><a href="layer_add.html">layer_add</a>()</code> <br><code><a href="layer_additive_attention.html">layer_additive_attention</a>()</code> <br><code><a href="layer_alpha_dropout.html">layer_alpha_dropout</a>()</code> <br><code><a href="layer_attention.html">layer_attention</a>()</code> <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_average.html">layer_average</a>()</code> <br><code><a href="layer_average_pooling_1d.html">layer_average_pooling_1d</a>()</code> <br><code><a href="layer_average_pooling_2d.html">layer_average_pooling_2d</a>()</code> <br><code><a href="layer_average_pooling_3d.html">layer_average_pooling_3d</a>()</code> <br><code><a href="layer_batch_normalization.html">layer_batch_normalization</a>()</code> <br><code><a href="layer_bidirectional.html">layer_bidirectional</a>()</code> <br><code><a href="layer_category_encoding.html">layer_category_encoding</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_concatenate.html">layer_concatenate</a>()</code> <br><code><a href="layer_conv_1d.html">layer_conv_1d</a>()</code> <br><code><a href="layer_conv_1d_transpose.html">layer_conv_1d_transpose</a>()</code> <br><code><a href="layer_conv_2d.html">layer_conv_2d</a>()</code> <br><code><a href="layer_conv_2d_transpose.html">layer_conv_2d_transpose</a>()</code> <br><code><a href="layer_conv_3d.html">layer_conv_3d</a>()</code> <br><code><a href="layer_conv_3d_transpose.html">layer_conv_3d_transpose</a>()</code> <br><code><a href="layer_conv_lstm_1d.html">layer_conv_lstm_1d</a>()</code> <br><code><a href="layer_conv_lstm_2d.html">layer_conv_lstm_2d</a>()</code> <br><code><a href="layer_conv_lstm_3d.html">layer_conv_lstm_3d</a>()</code> <br><code><a href="layer_cropping_1d.html">layer_cropping_1d</a>()</code> <br><code><a href="layer_cropping_2d.html">layer_cropping_2d</a>()</code> <br><code><a href="layer_cropping_3d.html">layer_cropping_3d</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_dense.html">layer_dense</a>()</code> <br><code><a href="layer_depthwise_conv_1d.html">layer_depthwise_conv_1d</a>()</code> <br><code><a href="layer_depthwise_conv_2d.html">layer_depthwise_conv_2d</a>()</code> <br><code><a href="layer_discretization.html">layer_discretization</a>()</code> <br><code><a href="layer_dot.html">layer_dot</a>()</code> <br><code><a href="layer_dropout.html">layer_dropout</a>()</code> <br><code><a href="layer_einsum_dense.html">layer_einsum_dense</a>()</code> <br><code><a href="layer_embedding.html">layer_embedding</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_feature_space.html">layer_feature_space</a>()</code> <br><code><a href="layer_flatten.html">layer_flatten</a>()</code> <br><code><a href="layer_flax_module_wrapper.html">layer_flax_module_wrapper</a>()</code> <br><code><a href="layer_gaussian_dropout.html">layer_gaussian_dropout</a>()</code> <br><code><a href="layer_gaussian_noise.html">layer_gaussian_noise</a>()</code> <br><code><a href="layer_global_average_pooling_1d.html">layer_global_average_pooling_1d</a>()</code> <br><code><a href="layer_global_average_pooling_2d.html">layer_global_average_pooling_2d</a>()</code> <br><code><a href="layer_global_average_pooling_3d.html">layer_global_average_pooling_3d</a>()</code> <br><code><a href="layer_global_max_pooling_1d.html">layer_global_max_pooling_1d</a>()</code> <br><code><a href="layer_global_max_pooling_2d.html">layer_global_max_pooling_2d</a>()</code> <br><code><a href="layer_global_max_pooling_3d.html">layer_global_max_pooling_3d</a>()</code> <br><code><a href="layer_group_normalization.html">layer_group_normalization</a>()</code> <br><code><a href="layer_group_query_attention.html">layer_group_query_attention</a>()</code> <br><code><a href="layer_gru.html">layer_gru</a>()</code> <br><code><a href="layer_hashed_crossing.html">layer_hashed_crossing</a>()</code> <br><code><a href="layer_hashing.html">layer_hashing</a>()</code> <br><code><a href="layer_identity.html">layer_identity</a>()</code> <br><code><a href="layer_integer_lookup.html">layer_integer_lookup</a>()</code> <br><code><a href="layer_jax_model_wrapper.html">layer_jax_model_wrapper</a>()</code> <br><code><a href="layer_lambda.html">layer_lambda</a>()</code> <br><code><a href="layer_layer_normalization.html">layer_layer_normalization</a>()</code> <br><code><a href="layer_lstm.html">layer_lstm</a>()</code> <br><code><a href="layer_masking.html">layer_masking</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_max_pooling_1d.html">layer_max_pooling_1d</a>()</code> <br><code><a href="layer_max_pooling_2d.html">layer_max_pooling_2d</a>()</code> <br><code><a href="layer_max_pooling_3d.html">layer_max_pooling_3d</a>()</code> <br><code><a href="layer_maximum.html">layer_maximum</a>()</code> <br><code><a href="layer_mel_spectrogram.html">layer_mel_spectrogram</a>()</code> <br><code><a href="layer_minimum.html">layer_minimum</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_multi_head_attention.html">layer_multi_head_attention</a>()</code> <br><code><a href="layer_multiply.html">layer_multiply</a>()</code> <br><code><a href="layer_normalization.html">layer_normalization</a>()</code> <br><code><a href="layer_permute.html">layer_permute</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_brightness.html">layer_random_brightness</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_contrast.html">layer_random_contrast</a>()</code> <br><code><a href="layer_random_crop.html">layer_random_crop</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_flip.html">layer_random_flip</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_rotation.html">layer_random_rotation</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_random_translation.html">layer_random_translation</a>()</code> <br><code><a href="layer_random_zoom.html">layer_random_zoom</a>()</code> <br><code><a href="layer_repeat_vector.html">layer_repeat_vector</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_reshape.html">layer_reshape</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_rms_normalization.html">layer_rms_normalization</a>()</code> <br><code><a href="layer_rnn.html">layer_rnn</a>()</code> <br><code><a href="layer_separable_conv_1d.html">layer_separable_conv_1d</a>()</code> <br><code><a href="layer_separable_conv_2d.html">layer_separable_conv_2d</a>()</code> <br><code><a href="layer_simple_rnn.html">layer_simple_rnn</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br><code><a href="layer_spatial_dropout_1d.html">layer_spatial_dropout_1d</a>()</code> <br><code><a href="layer_spatial_dropout_2d.html">layer_spatial_dropout_2d</a>()</code> <br><code><a href="layer_spatial_dropout_3d.html">layer_spatial_dropout_3d</a>()</code> <br><code><a href="layer_spectral_normalization.html">layer_spectral_normalization</a>()</code> <br><code><a href="layer_stft_spectrogram.html">layer_stft_spectrogram</a>()</code> <br><code><a href="layer_string_lookup.html">layer_string_lookup</a>()</code> <br><code><a href="layer_subtract.html">layer_subtract</a>()</code> <br><code><a href="layer_text_vectorization.html">layer_text_vectorization</a>()</code> <br><code><a href="layer_tfsm.html">layer_tfsm</a>()</code> <br><code><a href="layer_time_distributed.html">layer_time_distributed</a>()</code> <br><code><a href="layer_torch_module_wrapper.html">layer_torch_module_wrapper</a>()</code> <br><code><a href="layer_unit_normalization.html">layer_unit_normalization</a>()</code> <br><code><a href="layer_upsampling_1d.html">layer_upsampling_1d</a>()</code> <br><code><a href="layer_upsampling_2d.html">layer_upsampling_2d</a>()</code> <br><code><a href="layer_upsampling_3d.html">layer_upsampling_3d</a>()</code> <br><code><a href="layer_zero_padding_1d.html">layer_zero_padding_1d</a>()</code> <br><code><a href="layer_zero_padding_2d.html">layer_zero_padding_2d</a>()</code> <br><code><a href="rnn_cell_gru.html">rnn_cell_gru</a>()</code> <br><code><a href="rnn_cell_lstm.html">rnn_cell_lstm</a>()</code> <br><code><a href="rnn_cell_simple.html">rnn_cell_simple</a>()</code> <br><code><a href="rnn_cells_stack.html">rnn_cells_stack</a>()</code> <br></p></div>
+<p>Other layers: <br><code><a href="Layer.html">Layer</a>()</code> <br><code><a href="layer_activation.html">layer_activation</a>()</code> <br><code><a href="layer_activation_elu.html">layer_activation_elu</a>()</code> <br><code><a href="layer_activation_leaky_relu.html">layer_activation_leaky_relu</a>()</code> <br><code><a href="layer_activation_parametric_relu.html">layer_activation_parametric_relu</a>()</code> <br><code><a href="layer_activation_relu.html">layer_activation_relu</a>()</code> <br><code><a href="layer_activation_softmax.html">layer_activation_softmax</a>()</code> <br><code><a href="layer_activity_regularization.html">layer_activity_regularization</a>()</code> <br><code><a href="layer_add.html">layer_add</a>()</code> <br><code><a href="layer_additive_attention.html">layer_additive_attention</a>()</code> <br><code><a href="layer_alpha_dropout.html">layer_alpha_dropout</a>()</code> <br><code><a href="layer_attention.html">layer_attention</a>()</code> <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_average.html">layer_average</a>()</code> <br><code><a href="layer_average_pooling_1d.html">layer_average_pooling_1d</a>()</code> <br><code><a href="layer_average_pooling_2d.html">layer_average_pooling_2d</a>()</code> <br><code><a href="layer_average_pooling_3d.html">layer_average_pooling_3d</a>()</code> <br><code><a href="layer_batch_normalization.html">layer_batch_normalization</a>()</code> <br><code><a href="layer_bidirectional.html">layer_bidirectional</a>()</code> <br><code><a href="layer_category_encoding.html">layer_category_encoding</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_concatenate.html">layer_concatenate</a>()</code> <br><code><a href="layer_conv_1d.html">layer_conv_1d</a>()</code> <br><code><a href="layer_conv_1d_transpose.html">layer_conv_1d_transpose</a>()</code> <br><code><a href="layer_conv_2d.html">layer_conv_2d</a>()</code> <br><code><a href="layer_conv_2d_transpose.html">layer_conv_2d_transpose</a>()</code> <br><code><a href="layer_conv_3d.html">layer_conv_3d</a>()</code> <br><code><a href="layer_conv_3d_transpose.html">layer_conv_3d_transpose</a>()</code> <br><code><a href="layer_conv_lstm_1d.html">layer_conv_lstm_1d</a>()</code> <br><code><a href="layer_conv_lstm_2d.html">layer_conv_lstm_2d</a>()</code> <br><code><a href="layer_conv_lstm_3d.html">layer_conv_lstm_3d</a>()</code> <br><code><a href="layer_cropping_1d.html">layer_cropping_1d</a>()</code> <br><code><a href="layer_cropping_2d.html">layer_cropping_2d</a>()</code> <br><code><a href="layer_cropping_3d.html">layer_cropping_3d</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_dense.html">layer_dense</a>()</code> <br><code><a href="layer_depthwise_conv_1d.html">layer_depthwise_conv_1d</a>()</code> <br><code><a href="layer_depthwise_conv_2d.html">layer_depthwise_conv_2d</a>()</code> <br><code><a href="layer_discretization.html">layer_discretization</a>()</code> <br><code><a href="layer_dot.html">layer_dot</a>()</code> <br><code><a href="layer_dropout.html">layer_dropout</a>()</code> <br><code><a href="layer_einsum_dense.html">layer_einsum_dense</a>()</code> <br><code><a href="layer_embedding.html">layer_embedding</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_feature_space.html">layer_feature_space</a>()</code> <br><code><a href="layer_flatten.html">layer_flatten</a>()</code> <br><code><a href="layer_flax_module_wrapper.html">layer_flax_module_wrapper</a>()</code> <br><code><a href="layer_gaussian_dropout.html">layer_gaussian_dropout</a>()</code> <br><code><a href="layer_gaussian_noise.html">layer_gaussian_noise</a>()</code> <br><code><a href="layer_global_average_pooling_1d.html">layer_global_average_pooling_1d</a>()</code> <br><code><a href="layer_global_average_pooling_2d.html">layer_global_average_pooling_2d</a>()</code> <br><code><a href="layer_global_average_pooling_3d.html">layer_global_average_pooling_3d</a>()</code> <br><code><a href="layer_global_max_pooling_1d.html">layer_global_max_pooling_1d</a>()</code> <br><code><a href="layer_global_max_pooling_2d.html">layer_global_max_pooling_2d</a>()</code> <br><code><a href="layer_global_max_pooling_3d.html">layer_global_max_pooling_3d</a>()</code> <br><code><a href="layer_group_normalization.html">layer_group_normalization</a>()</code> <br><code><a href="layer_group_query_attention.html">layer_group_query_attention</a>()</code> <br><code><a href="layer_gru.html">layer_gru</a>()</code> <br><code><a href="layer_hashed_crossing.html">layer_hashed_crossing</a>()</code> <br><code><a href="layer_hashing.html">layer_hashing</a>()</code> <br><code><a href="layer_identity.html">layer_identity</a>()</code> <br><code><a href="layer_integer_lookup.html">layer_integer_lookup</a>()</code> <br><code><a href="layer_jax_model_wrapper.html">layer_jax_model_wrapper</a>()</code> <br><code><a href="layer_lambda.html">layer_lambda</a>()</code> <br><code><a href="layer_layer_normalization.html">layer_layer_normalization</a>()</code> <br><code><a href="layer_lstm.html">layer_lstm</a>()</code> <br><code><a href="layer_masking.html">layer_masking</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_max_pooling_1d.html">layer_max_pooling_1d</a>()</code> <br><code><a href="layer_max_pooling_2d.html">layer_max_pooling_2d</a>()</code> <br><code><a href="layer_max_pooling_3d.html">layer_max_pooling_3d</a>()</code> <br><code><a href="layer_maximum.html">layer_maximum</a>()</code> <br><code><a href="layer_mel_spectrogram.html">layer_mel_spectrogram</a>()</code> <br><code><a href="layer_minimum.html">layer_minimum</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_multi_head_attention.html">layer_multi_head_attention</a>()</code> <br><code><a href="layer_multiply.html">layer_multiply</a>()</code> <br><code><a href="layer_normalization.html">layer_normalization</a>()</code> <br><code><a href="layer_permute.html">layer_permute</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_brightness.html">layer_random_brightness</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_contrast.html">layer_random_contrast</a>()</code> <br><code><a href="layer_random_crop.html">layer_random_crop</a>()</code> <br><code><a href="layer_random_elastic_transform.html">layer_random_elastic_transform</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_flip.html">layer_random_flip</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_rotation.html">layer_random_rotation</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_random_translation.html">layer_random_translation</a>()</code> <br><code><a href="layer_random_zoom.html">layer_random_zoom</a>()</code> <br><code><a href="layer_repeat_vector.html">layer_repeat_vector</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_reshape.html">layer_reshape</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_rms_normalization.html">layer_rms_normalization</a>()</code> <br><code><a href="layer_rnn.html">layer_rnn</a>()</code> <br><code><a href="layer_separable_conv_1d.html">layer_separable_conv_1d</a>()</code> <br><code><a href="layer_separable_conv_2d.html">layer_separable_conv_2d</a>()</code> <br><code><a href="layer_simple_rnn.html">layer_simple_rnn</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br><code><a href="layer_spatial_dropout_1d.html">layer_spatial_dropout_1d</a>()</code> <br><code><a href="layer_spatial_dropout_2d.html">layer_spatial_dropout_2d</a>()</code> <br><code><a href="layer_spatial_dropout_3d.html">layer_spatial_dropout_3d</a>()</code> <br><code><a href="layer_spectral_normalization.html">layer_spectral_normalization</a>()</code> <br><code><a href="layer_stft_spectrogram.html">layer_stft_spectrogram</a>()</code> <br><code><a href="layer_string_lookup.html">layer_string_lookup</a>()</code> <br><code><a href="layer_subtract.html">layer_subtract</a>()</code> <br><code><a href="layer_text_vectorization.html">layer_text_vectorization</a>()</code> <br><code><a href="layer_tfsm.html">layer_tfsm</a>()</code> <br><code><a href="layer_time_distributed.html">layer_time_distributed</a>()</code> <br><code><a href="layer_torch_module_wrapper.html">layer_torch_module_wrapper</a>()</code> <br><code><a href="layer_unit_normalization.html">layer_unit_normalization</a>()</code> <br><code><a href="layer_upsampling_1d.html">layer_upsampling_1d</a>()</code> <br><code><a href="layer_upsampling_2d.html">layer_upsampling_2d</a>()</code> <br><code><a href="layer_upsampling_3d.html">layer_upsampling_3d</a>()</code> <br><code><a href="layer_zero_padding_1d.html">layer_zero_padding_1d</a>()</code> <br><code><a href="layer_zero_padding_2d.html">layer_zero_padding_2d</a>()</code> <br><code><a href="rnn_cell_gru.html">rnn_cell_gru</a>()</code> <br><code><a href="rnn_cell_lstm.html">rnn_cell_lstm</a>()</code> <br><code><a href="rnn_cell_simple.html">rnn_cell_simple</a>()</code> <br><code><a href="rnn_cells_stack.html">rnn_cells_stack</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -159,7 +159,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/learning_rate_schedule_cosine_decay.html b/docs/dev/reference/learning_rate_schedule_cosine_decay.html
index 8cd2e7344..428292720 100644
--- a/docs/dev/reference/learning_rate_schedule_cosine_decay.html
+++ b/docs/dev/reference/learning_rate_schedule_cosine_decay.html
@@ -124,7 +124,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -295,7 +295,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/learning_rate_schedule_cosine_decay_restarts.html b/docs/dev/reference/learning_rate_schedule_cosine_decay_restarts.html
index a77e8ce60..6d10a3468 100644
--- a/docs/dev/reference/learning_rate_schedule_cosine_decay_restarts.html
+++ b/docs/dev/reference/learning_rate_schedule_cosine_decay_restarts.html
@@ -34,7 +34,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -173,7 +173,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/learning_rate_schedule_exponential_decay.html b/docs/dev/reference/learning_rate_schedule_exponential_decay.html
index 708cd8bcd..4dc8054d2 100644
--- a/docs/dev/reference/learning_rate_schedule_exponential_decay.html
+++ b/docs/dev/reference/learning_rate_schedule_exponential_decay.html
@@ -38,7 +38,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -182,7 +182,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/learning_rate_schedule_inverse_time_decay.html b/docs/dev/reference/learning_rate_schedule_inverse_time_decay.html
index dd0bd54d7..fecb07ac6 100644
--- a/docs/dev/reference/learning_rate_schedule_inverse_time_decay.html
+++ b/docs/dev/reference/learning_rate_schedule_inverse_time_decay.html
@@ -48,7 +48,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -193,7 +193,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/learning_rate_schedule_piecewise_constant_decay.html b/docs/dev/reference/learning_rate_schedule_piecewise_constant_decay.html
index bcce47fab..76f756b75 100644
--- a/docs/dev/reference/learning_rate_schedule_piecewise_constant_decay.html
+++ b/docs/dev/reference/learning_rate_schedule_piecewise_constant_decay.html
@@ -12,7 +12,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -142,7 +142,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/learning_rate_schedule_polynomial_decay.html b/docs/dev/reference/learning_rate_schedule_polynomial_decay.html
index d2f69c58d..da5859d0f 100644
--- a/docs/dev/reference/learning_rate_schedule_polynomial_decay.html
+++ b/docs/dev/reference/learning_rate_schedule_polynomial_decay.html
@@ -66,7 +66,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -232,7 +232,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/load_model.html b/docs/dev/reference/load_model.html
index 5490f02b6..25886550a 100644
--- a/docs/dev/reference/load_model.html
+++ b/docs/dev/reference/load_model.html
@@ -8,7 +8,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -128,7 +128,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/load_model_weights.html b/docs/dev/reference/load_model_weights.html
index 7fb546bd3..da8821ddd 100644
--- a/docs/dev/reference/load_model_weights.html
+++ b/docs/dev/reference/load_model_weights.html
@@ -1,9 +1,13 @@
 <!DOCTYPE html>
-<!-- Generated by pkgdown: do not edit by hand --><html lang="en"><head><meta http-equiv="Content-Type" content="text/html; charset=UTF-8"><meta charset="utf-8"><meta http-equiv="X-UA-Compatible" content="IE=edge"><meta name="viewport" content="width=device-width, initial-scale=1, shrink-to-fit=no"><title>Load weights from a file saved via save_model_weights(). — load_model_weights • keras3</title><!-- favicons --><link rel="icon" type="image/png" sizes="96x96" href="../favicon-96x96.png"><link rel="icon" type="”image/svg+xml”" href="../favicon.svg"><link rel="apple-touch-icon" sizes="180x180" href="../apple-touch-icon.png"><link rel="icon" sizes="any" href="../favicon.ico"><link rel="manifest" href="../site.webmanifest"><script src="../deps/jquery-3.6.0/jquery-3.6.0.min.js"></script><meta name="viewport" content="width=device-width, initial-scale=1, shrink-to-fit=no"><link href="../deps/bootstrap-5.3.1/bootstrap.min.css" rel="stylesheet"><script src="../deps/bootstrap-5.3.1/bootstrap.bundle.min.js"></script><link href="../deps/font-awesome-6.5.2/css/all.min.css" rel="stylesheet"><link href="../deps/font-awesome-6.5.2/css/v4-shims.min.css" rel="stylesheet"><script src="../deps/headroom-0.11.0/headroom.min.js"></script><script src="../deps/headroom-0.11.0/jQuery.headroom.min.js"></script><script src="../deps/bootstrap-toc-1.0.1/bootstrap-toc.min.js"></script><script src="../deps/clipboard.js-2.0.11/clipboard.min.js"></script><script src="../deps/search-1.0.0/autocomplete.jquery.min.js"></script><script src="../deps/search-1.0.0/fuse.min.js"></script><script src="../deps/search-1.0.0/mark.min.js"></script><!-- pkgdown --><script src="../pkgdown.js"></script><link href="../extra.css" rel="stylesheet"><meta property="og:title" content="Load weights from a file saved via save_model_weights(). — load_model_weights"><meta name="description" content="Weights are loaded based on the network's
+<!-- Generated by pkgdown: do not edit by hand --><html lang="en"><head><meta http-equiv="Content-Type" content="text/html; charset=UTF-8"><meta charset="utf-8"><meta http-equiv="X-UA-Compatible" content="IE=edge"><meta name="viewport" content="width=device-width, initial-scale=1, shrink-to-fit=no"><title>Load the weights from a single file or sharded files. — load_model_weights • keras3</title><!-- favicons --><link rel="icon" type="image/png" sizes="96x96" href="../favicon-96x96.png"><link rel="icon" type="”image/svg+xml”" href="../favicon.svg"><link rel="apple-touch-icon" sizes="180x180" href="../apple-touch-icon.png"><link rel="icon" sizes="any" href="../favicon.ico"><link rel="manifest" href="../site.webmanifest"><script src="../deps/jquery-3.6.0/jquery-3.6.0.min.js"></script><meta name="viewport" content="width=device-width, initial-scale=1, shrink-to-fit=no"><link href="../deps/bootstrap-5.3.1/bootstrap.min.css" rel="stylesheet"><script src="../deps/bootstrap-5.3.1/bootstrap.bundle.min.js"></script><link href="../deps/font-awesome-6.5.2/css/all.min.css" rel="stylesheet"><link href="../deps/font-awesome-6.5.2/css/v4-shims.min.css" rel="stylesheet"><script src="../deps/headroom-0.11.0/headroom.min.js"></script><script src="../deps/headroom-0.11.0/jQuery.headroom.min.js"></script><script src="../deps/bootstrap-toc-1.0.1/bootstrap-toc.min.js"></script><script src="../deps/clipboard.js-2.0.11/clipboard.min.js"></script><script src="../deps/search-1.0.0/autocomplete.jquery.min.js"></script><script src="../deps/search-1.0.0/fuse.min.js"></script><script src="../deps/search-1.0.0/mark.min.js"></script><!-- pkgdown --><script src="../pkgdown.js"></script><link href="../extra.css" rel="stylesheet"><meta property="og:title" content="Load the weights from a single file or sharded files. — load_model_weights"><meta name="description" content="Weights are loaded based on the network's
 topology. This means the architecture should be the same as when the
 weights were saved. Note that layers that don't have weights are not
 taken into account in the topological ordering, so adding or removing
 layers is fine as long as they don't have weights.
+Sharding
+When loading sharded weights, specify a filepath ending in
+&quot;.weights.json&quot; (the configuration file), with the corresponding shard files
+(*_xxxxx.weights.h5) located alongside it.
 Partial weight loading
 If you have modified your model, for instance by adding a new layer
 (with weights) or by changing the shape of the weights of a layer,
@@ -15,6 +19,10 @@
 weights were saved. Note that layers that don't have weights are not
 taken into account in the topological ordering, so adding or removing
 layers is fine as long as they don't have weights.
+Sharding
+When loading sharded weights, specify a filepath ending in
+&quot;.weights.json&quot; (the configuration file), with the corresponding shard files
+(*_xxxxx.weights.h5) located alongside it.
 Partial weight loading
 If you have modified your model, for instance by adding a new layer
 (with weights) or by changing the shape of the weights of a layer,
@@ -30,7 +38,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -70,7 +78,7 @@
 <div class="row">
   <main id="main" class="col-md-9"><div class="page-header">
 
-      <h1>Load weights from a file saved via <code>save_model_weights()</code>.</h1>
+      <h1>Load the weights from a single file or sharded files.</h1>
       <small class="dont-index">Source: <a href="https://github.com/rstudio/keras3/blob/HEAD/R/model-persistence.R" class="external-link"><code>R/model-persistence.R</code></a></small>
       <div class="d-none name"><code>load_model_weights.Rd</code></div>
     </div>
@@ -81,6 +89,10 @@ <h1>Load weights from a file saved via <code>save_model_weights()</code>.</h1>
 weights were saved. Note that layers that don't have weights are not
 taken into account in the topological ordering, so adding or removing
 layers is fine as long as they don't have weights.</p>
+<p><strong>Sharding</strong></p>
+<p>When loading sharded weights, specify a <code>filepath</code> ending in
+<code>".weights.json"</code> (the configuration file), with the corresponding shard files
+(<code>*_xxxxx.weights.h5</code>) located alongside it.</p>
 <p><strong>Partial weight loading</strong></p>
 <p>If you have modified your model, for instance by adding a new layer
 (with weights) or by changing the shape of the weights of a layer,
@@ -104,9 +116,9 @@ <h2 id="arguments">Arguments<a class="anchor" aria-label="anchor" href="#argumen
 
 
 <dt id="arg-filepath">filepath<a class="anchor" aria-label="anchor" href="#arg-filepath"></a></dt>
-<dd><p>String, path to the weights file to load.
-It can either be a <code>.weights.h5</code> file
-or a legacy <code>.h5</code> weights file.</p></dd>
+<dd><p>Path or path-like object to the weights. Accepts <code>.weights.h5</code>, legacy <code>.h5</code>,
+or sharded weights through a <code>.weights.json</code> manifest sitting alongside the
+shard files (<code>*_xxxxx.weights.h5</code>).</p></dd>
 
 
 <dt id="arg-skip-mismatch">skip_mismatch<a class="anchor" aria-label="anchor" href="#arg-skip-mismatch"></a></dt>
@@ -123,6 +135,11 @@ <h2 id="arguments">Arguments<a class="anchor" aria-label="anchor" href="#argumen
     <h2 id="value">Value<a class="anchor" aria-label="anchor" href="#value"></a></h2>
     <p>This is called primarily for side effects. <code>model</code> is returned,
 invisibly, to enable usage with the pipe.</p>
+    </div>
+    <div class="section level2">
+    <h2 id="examples">Examples<a class="anchor" aria-label="anchor" href="#examples"></a></h2>
+    <p></p><div class="sourceCode r"><pre><code><span><span class="va">model</span> <span class="op">|&gt;</span> <span class="fu"><a href="../reference/load_model_weights.html">load_model_weights</a></span><span class="op">(</span><span class="st">"model.weights.h5"</span><span class="op">)</span></span>
+<span><span class="va">model</span> <span class="op">|&gt;</span> <span class="fu"><a href="../reference/load_model_weights.html">load_model_weights</a></span><span class="op">(</span><span class="st">"model.weights.json"</span><span class="op">)</span></span></code></pre><p></p></div>
     </div>
     <div class="section level2">
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
@@ -140,7 +157,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/loss_binary_crossentropy.html b/docs/dev/reference/loss_binary_crossentropy.html
index aff2db0e7..9292af724 100644
--- a/docs/dev/reference/loss_binary_crossentropy.html
+++ b/docs/dev/reference/loss_binary_crossentropy.html
@@ -26,7 +26,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -217,7 +217,7 @@ <h2 id="examples">Examples<a class="anchor" aria-label="anchor" href="#examples"
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
     <div class="dont-index">
 <ul><li><p><a href="https://keras.io/api/losses/probabilistic_losses#binarycrossentropy-class" class="external-link">https://keras.io/api/losses/probabilistic_losses#binarycrossentropy-class</a></p></li>
-</ul><p>Other losses: <br><code><a href="Loss.html">Loss</a>()</code> <br><code><a href="loss_binary_focal_crossentropy.html">loss_binary_focal_crossentropy</a>()</code> <br><code><a href="loss_categorical_crossentropy.html">loss_categorical_crossentropy</a>()</code> <br><code><a href="loss_categorical_focal_crossentropy.html">loss_categorical_focal_crossentropy</a>()</code> <br><code><a href="loss_categorical_hinge.html">loss_categorical_hinge</a>()</code> <br><code><a href="loss_circle.html">loss_circle</a>()</code> <br><code><a href="loss_cosine_similarity.html">loss_cosine_similarity</a>()</code> <br><code><a href="loss_ctc.html">loss_ctc</a>()</code> <br><code><a href="loss_dice.html">loss_dice</a>()</code> <br><code><a href="loss_hinge.html">loss_hinge</a>()</code> <br><code><a href="loss_huber.html">loss_huber</a>()</code> <br><code><a href="loss_kl_divergence.html">loss_kl_divergence</a>()</code> <br><code><a href="loss_log_cosh.html">loss_log_cosh</a>()</code> <br><code><a href="loss_mean_absolute_error.html">loss_mean_absolute_error</a>()</code> <br><code><a href="loss_mean_absolute_percentage_error.html">loss_mean_absolute_percentage_error</a>()</code> <br><code><a href="loss_mean_squared_error.html">loss_mean_squared_error</a>()</code> <br><code><a href="loss_mean_squared_logarithmic_error.html">loss_mean_squared_logarithmic_error</a>()</code> <br><code><a href="loss_poisson.html">loss_poisson</a>()</code> <br><code><a href="loss_sparse_categorical_crossentropy.html">loss_sparse_categorical_crossentropy</a>()</code> <br><code><a href="loss_squared_hinge.html">loss_squared_hinge</a>()</code> <br><code><a href="loss_tversky.html">loss_tversky</a>()</code> <br><code><a href="metric_binary_crossentropy.html">metric_binary_crossentropy</a>()</code> <br><code><a href="metric_binary_focal_crossentropy.html">metric_binary_focal_crossentropy</a>()</code> <br><code><a href="metric_categorical_crossentropy.html">metric_categorical_crossentropy</a>()</code> <br><code><a href="metric_categorical_focal_crossentropy.html">metric_categorical_focal_crossentropy</a>()</code> <br><code><a href="metric_categorical_hinge.html">metric_categorical_hinge</a>()</code> <br><code><a href="metric_hinge.html">metric_hinge</a>()</code> <br><code><a href="metric_huber.html">metric_huber</a>()</code> <br><code><a href="metric_kl_divergence.html">metric_kl_divergence</a>()</code> <br><code><a href="metric_log_cosh.html">metric_log_cosh</a>()</code> <br><code><a href="metric_mean_absolute_error.html">metric_mean_absolute_error</a>()</code> <br><code><a href="metric_mean_absolute_percentage_error.html">metric_mean_absolute_percentage_error</a>()</code> <br><code><a href="metric_mean_squared_error.html">metric_mean_squared_error</a>()</code> <br><code><a href="metric_mean_squared_logarithmic_error.html">metric_mean_squared_logarithmic_error</a>()</code> <br><code><a href="metric_poisson.html">metric_poisson</a>()</code> <br><code><a href="metric_sparse_categorical_crossentropy.html">metric_sparse_categorical_crossentropy</a>()</code> <br><code><a href="metric_squared_hinge.html">metric_squared_hinge</a>()</code> <br></p></div>
+</ul><p>Other losses: <br><code><a href="Loss.html">Loss</a>()</code> <br><code><a href="loss_binary_focal_crossentropy.html">loss_binary_focal_crossentropy</a>()</code> <br><code><a href="loss_categorical_crossentropy.html">loss_categorical_crossentropy</a>()</code> <br><code><a href="loss_categorical_focal_crossentropy.html">loss_categorical_focal_crossentropy</a>()</code> <br><code><a href="loss_categorical_generalized_cross_entropy.html">loss_categorical_generalized_cross_entropy</a>()</code> <br><code><a href="loss_categorical_hinge.html">loss_categorical_hinge</a>()</code> <br><code><a href="loss_circle.html">loss_circle</a>()</code> <br><code><a href="loss_cosine_similarity.html">loss_cosine_similarity</a>()</code> <br><code><a href="loss_ctc.html">loss_ctc</a>()</code> <br><code><a href="loss_dice.html">loss_dice</a>()</code> <br><code><a href="loss_hinge.html">loss_hinge</a>()</code> <br><code><a href="loss_huber.html">loss_huber</a>()</code> <br><code><a href="loss_kl_divergence.html">loss_kl_divergence</a>()</code> <br><code><a href="loss_log_cosh.html">loss_log_cosh</a>()</code> <br><code><a href="loss_mean_absolute_error.html">loss_mean_absolute_error</a>()</code> <br><code><a href="loss_mean_absolute_percentage_error.html">loss_mean_absolute_percentage_error</a>()</code> <br><code><a href="loss_mean_squared_error.html">loss_mean_squared_error</a>()</code> <br><code><a href="loss_mean_squared_logarithmic_error.html">loss_mean_squared_logarithmic_error</a>()</code> <br><code><a href="loss_poisson.html">loss_poisson</a>()</code> <br><code><a href="loss_sparse_categorical_crossentropy.html">loss_sparse_categorical_crossentropy</a>()</code> <br><code><a href="loss_squared_hinge.html">loss_squared_hinge</a>()</code> <br><code><a href="loss_tversky.html">loss_tversky</a>()</code> <br><code><a href="metric_binary_crossentropy.html">metric_binary_crossentropy</a>()</code> <br><code><a href="metric_binary_focal_crossentropy.html">metric_binary_focal_crossentropy</a>()</code> <br><code><a href="metric_categorical_crossentropy.html">metric_categorical_crossentropy</a>()</code> <br><code><a href="metric_categorical_focal_crossentropy.html">metric_categorical_focal_crossentropy</a>()</code> <br><code><a href="metric_categorical_hinge.html">metric_categorical_hinge</a>()</code> <br><code><a href="metric_hinge.html">metric_hinge</a>()</code> <br><code><a href="metric_huber.html">metric_huber</a>()</code> <br><code><a href="metric_kl_divergence.html">metric_kl_divergence</a>()</code> <br><code><a href="metric_log_cosh.html">metric_log_cosh</a>()</code> <br><code><a href="metric_mean_absolute_error.html">metric_mean_absolute_error</a>()</code> <br><code><a href="metric_mean_absolute_percentage_error.html">metric_mean_absolute_percentage_error</a>()</code> <br><code><a href="metric_mean_squared_error.html">metric_mean_squared_error</a>()</code> <br><code><a href="metric_mean_squared_logarithmic_error.html">metric_mean_squared_logarithmic_error</a>()</code> <br><code><a href="metric_poisson.html">metric_poisson</a>()</code> <br><code><a href="metric_sparse_categorical_crossentropy.html">metric_sparse_categorical_crossentropy</a>()</code> <br><code><a href="metric_squared_hinge.html">metric_squared_hinge</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -229,7 +229,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/loss_binary_focal_crossentropy.html b/docs/dev/reference/loss_binary_focal_crossentropy.html
index b0843614a..a0425951c 100644
--- a/docs/dev/reference/loss_binary_focal_crossentropy.html
+++ b/docs/dev/reference/loss_binary_focal_crossentropy.html
@@ -64,7 +64,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -322,7 +322,7 @@ <h2 id="examples">Examples<a class="anchor" aria-label="anchor" href="#examples"
     </div>
     <div class="section level2">
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
-    <div class="dont-index"><p>Other losses: <br><code><a href="Loss.html">Loss</a>()</code> <br><code><a href="loss_binary_crossentropy.html">loss_binary_crossentropy</a>()</code> <br><code><a href="loss_categorical_crossentropy.html">loss_categorical_crossentropy</a>()</code> <br><code><a href="loss_categorical_focal_crossentropy.html">loss_categorical_focal_crossentropy</a>()</code> <br><code><a href="loss_categorical_hinge.html">loss_categorical_hinge</a>()</code> <br><code><a href="loss_circle.html">loss_circle</a>()</code> <br><code><a href="loss_cosine_similarity.html">loss_cosine_similarity</a>()</code> <br><code><a href="loss_ctc.html">loss_ctc</a>()</code> <br><code><a href="loss_dice.html">loss_dice</a>()</code> <br><code><a href="loss_hinge.html">loss_hinge</a>()</code> <br><code><a href="loss_huber.html">loss_huber</a>()</code> <br><code><a href="loss_kl_divergence.html">loss_kl_divergence</a>()</code> <br><code><a href="loss_log_cosh.html">loss_log_cosh</a>()</code> <br><code><a href="loss_mean_absolute_error.html">loss_mean_absolute_error</a>()</code> <br><code><a href="loss_mean_absolute_percentage_error.html">loss_mean_absolute_percentage_error</a>()</code> <br><code><a href="loss_mean_squared_error.html">loss_mean_squared_error</a>()</code> <br><code><a href="loss_mean_squared_logarithmic_error.html">loss_mean_squared_logarithmic_error</a>()</code> <br><code><a href="loss_poisson.html">loss_poisson</a>()</code> <br><code><a href="loss_sparse_categorical_crossentropy.html">loss_sparse_categorical_crossentropy</a>()</code> <br><code><a href="loss_squared_hinge.html">loss_squared_hinge</a>()</code> <br><code><a href="loss_tversky.html">loss_tversky</a>()</code> <br><code><a href="metric_binary_crossentropy.html">metric_binary_crossentropy</a>()</code> <br><code><a href="metric_binary_focal_crossentropy.html">metric_binary_focal_crossentropy</a>()</code> <br><code><a href="metric_categorical_crossentropy.html">metric_categorical_crossentropy</a>()</code> <br><code><a href="metric_categorical_focal_crossentropy.html">metric_categorical_focal_crossentropy</a>()</code> <br><code><a href="metric_categorical_hinge.html">metric_categorical_hinge</a>()</code> <br><code><a href="metric_hinge.html">metric_hinge</a>()</code> <br><code><a href="metric_huber.html">metric_huber</a>()</code> <br><code><a href="metric_kl_divergence.html">metric_kl_divergence</a>()</code> <br><code><a href="metric_log_cosh.html">metric_log_cosh</a>()</code> <br><code><a href="metric_mean_absolute_error.html">metric_mean_absolute_error</a>()</code> <br><code><a href="metric_mean_absolute_percentage_error.html">metric_mean_absolute_percentage_error</a>()</code> <br><code><a href="metric_mean_squared_error.html">metric_mean_squared_error</a>()</code> <br><code><a href="metric_mean_squared_logarithmic_error.html">metric_mean_squared_logarithmic_error</a>()</code> <br><code><a href="metric_poisson.html">metric_poisson</a>()</code> <br><code><a href="metric_sparse_categorical_crossentropy.html">metric_sparse_categorical_crossentropy</a>()</code> <br><code><a href="metric_squared_hinge.html">metric_squared_hinge</a>()</code> <br></p></div>
+    <div class="dont-index"><p>Other losses: <br><code><a href="Loss.html">Loss</a>()</code> <br><code><a href="loss_binary_crossentropy.html">loss_binary_crossentropy</a>()</code> <br><code><a href="loss_categorical_crossentropy.html">loss_categorical_crossentropy</a>()</code> <br><code><a href="loss_categorical_focal_crossentropy.html">loss_categorical_focal_crossentropy</a>()</code> <br><code><a href="loss_categorical_generalized_cross_entropy.html">loss_categorical_generalized_cross_entropy</a>()</code> <br><code><a href="loss_categorical_hinge.html">loss_categorical_hinge</a>()</code> <br><code><a href="loss_circle.html">loss_circle</a>()</code> <br><code><a href="loss_cosine_similarity.html">loss_cosine_similarity</a>()</code> <br><code><a href="loss_ctc.html">loss_ctc</a>()</code> <br><code><a href="loss_dice.html">loss_dice</a>()</code> <br><code><a href="loss_hinge.html">loss_hinge</a>()</code> <br><code><a href="loss_huber.html">loss_huber</a>()</code> <br><code><a href="loss_kl_divergence.html">loss_kl_divergence</a>()</code> <br><code><a href="loss_log_cosh.html">loss_log_cosh</a>()</code> <br><code><a href="loss_mean_absolute_error.html">loss_mean_absolute_error</a>()</code> <br><code><a href="loss_mean_absolute_percentage_error.html">loss_mean_absolute_percentage_error</a>()</code> <br><code><a href="loss_mean_squared_error.html">loss_mean_squared_error</a>()</code> <br><code><a href="loss_mean_squared_logarithmic_error.html">loss_mean_squared_logarithmic_error</a>()</code> <br><code><a href="loss_poisson.html">loss_poisson</a>()</code> <br><code><a href="loss_sparse_categorical_crossentropy.html">loss_sparse_categorical_crossentropy</a>()</code> <br><code><a href="loss_squared_hinge.html">loss_squared_hinge</a>()</code> <br><code><a href="loss_tversky.html">loss_tversky</a>()</code> <br><code><a href="metric_binary_crossentropy.html">metric_binary_crossentropy</a>()</code> <br><code><a href="metric_binary_focal_crossentropy.html">metric_binary_focal_crossentropy</a>()</code> <br><code><a href="metric_categorical_crossentropy.html">metric_categorical_crossentropy</a>()</code> <br><code><a href="metric_categorical_focal_crossentropy.html">metric_categorical_focal_crossentropy</a>()</code> <br><code><a href="metric_categorical_hinge.html">metric_categorical_hinge</a>()</code> <br><code><a href="metric_hinge.html">metric_hinge</a>()</code> <br><code><a href="metric_huber.html">metric_huber</a>()</code> <br><code><a href="metric_kl_divergence.html">metric_kl_divergence</a>()</code> <br><code><a href="metric_log_cosh.html">metric_log_cosh</a>()</code> <br><code><a href="metric_mean_absolute_error.html">metric_mean_absolute_error</a>()</code> <br><code><a href="metric_mean_absolute_percentage_error.html">metric_mean_absolute_percentage_error</a>()</code> <br><code><a href="metric_mean_squared_error.html">metric_mean_squared_error</a>()</code> <br><code><a href="metric_mean_squared_logarithmic_error.html">metric_mean_squared_logarithmic_error</a>()</code> <br><code><a href="metric_poisson.html">metric_poisson</a>()</code> <br><code><a href="metric_sparse_categorical_crossentropy.html">metric_sparse_categorical_crossentropy</a>()</code> <br><code><a href="metric_squared_hinge.html">metric_squared_hinge</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -334,7 +334,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/loss_categorical_crossentropy.html b/docs/dev/reference/loss_categorical_crossentropy.html
index 9f63ef73c..8013c3dff 100644
--- a/docs/dev/reference/loss_categorical_crossentropy.html
+++ b/docs/dev/reference/loss_categorical_crossentropy.html
@@ -18,7 +18,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -185,7 +185,7 @@ <h2 id="examples">Examples<a class="anchor" aria-label="anchor" href="#examples"
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
     <div class="dont-index">
 <ul><li><p><a href="https://keras.io/api/losses/probabilistic_losses#categoricalcrossentropy-class" class="external-link">https://keras.io/api/losses/probabilistic_losses#categoricalcrossentropy-class</a></p></li>
-</ul><p>Other losses: <br><code><a href="Loss.html">Loss</a>()</code> <br><code><a href="loss_binary_crossentropy.html">loss_binary_crossentropy</a>()</code> <br><code><a href="loss_binary_focal_crossentropy.html">loss_binary_focal_crossentropy</a>()</code> <br><code><a href="loss_categorical_focal_crossentropy.html">loss_categorical_focal_crossentropy</a>()</code> <br><code><a href="loss_categorical_hinge.html">loss_categorical_hinge</a>()</code> <br><code><a href="loss_circle.html">loss_circle</a>()</code> <br><code><a href="loss_cosine_similarity.html">loss_cosine_similarity</a>()</code> <br><code><a href="loss_ctc.html">loss_ctc</a>()</code> <br><code><a href="loss_dice.html">loss_dice</a>()</code> <br><code><a href="loss_hinge.html">loss_hinge</a>()</code> <br><code><a href="loss_huber.html">loss_huber</a>()</code> <br><code><a href="loss_kl_divergence.html">loss_kl_divergence</a>()</code> <br><code><a href="loss_log_cosh.html">loss_log_cosh</a>()</code> <br><code><a href="loss_mean_absolute_error.html">loss_mean_absolute_error</a>()</code> <br><code><a href="loss_mean_absolute_percentage_error.html">loss_mean_absolute_percentage_error</a>()</code> <br><code><a href="loss_mean_squared_error.html">loss_mean_squared_error</a>()</code> <br><code><a href="loss_mean_squared_logarithmic_error.html">loss_mean_squared_logarithmic_error</a>()</code> <br><code><a href="loss_poisson.html">loss_poisson</a>()</code> <br><code><a href="loss_sparse_categorical_crossentropy.html">loss_sparse_categorical_crossentropy</a>()</code> <br><code><a href="loss_squared_hinge.html">loss_squared_hinge</a>()</code> <br><code><a href="loss_tversky.html">loss_tversky</a>()</code> <br><code><a href="metric_binary_crossentropy.html">metric_binary_crossentropy</a>()</code> <br><code><a href="metric_binary_focal_crossentropy.html">metric_binary_focal_crossentropy</a>()</code> <br><code><a href="metric_categorical_crossentropy.html">metric_categorical_crossentropy</a>()</code> <br><code><a href="metric_categorical_focal_crossentropy.html">metric_categorical_focal_crossentropy</a>()</code> <br><code><a href="metric_categorical_hinge.html">metric_categorical_hinge</a>()</code> <br><code><a href="metric_hinge.html">metric_hinge</a>()</code> <br><code><a href="metric_huber.html">metric_huber</a>()</code> <br><code><a href="metric_kl_divergence.html">metric_kl_divergence</a>()</code> <br><code><a href="metric_log_cosh.html">metric_log_cosh</a>()</code> <br><code><a href="metric_mean_absolute_error.html">metric_mean_absolute_error</a>()</code> <br><code><a href="metric_mean_absolute_percentage_error.html">metric_mean_absolute_percentage_error</a>()</code> <br><code><a href="metric_mean_squared_error.html">metric_mean_squared_error</a>()</code> <br><code><a href="metric_mean_squared_logarithmic_error.html">metric_mean_squared_logarithmic_error</a>()</code> <br><code><a href="metric_poisson.html">metric_poisson</a>()</code> <br><code><a href="metric_sparse_categorical_crossentropy.html">metric_sparse_categorical_crossentropy</a>()</code> <br><code><a href="metric_squared_hinge.html">metric_squared_hinge</a>()</code> <br></p></div>
+</ul><p>Other losses: <br><code><a href="Loss.html">Loss</a>()</code> <br><code><a href="loss_binary_crossentropy.html">loss_binary_crossentropy</a>()</code> <br><code><a href="loss_binary_focal_crossentropy.html">loss_binary_focal_crossentropy</a>()</code> <br><code><a href="loss_categorical_focal_crossentropy.html">loss_categorical_focal_crossentropy</a>()</code> <br><code><a href="loss_categorical_generalized_cross_entropy.html">loss_categorical_generalized_cross_entropy</a>()</code> <br><code><a href="loss_categorical_hinge.html">loss_categorical_hinge</a>()</code> <br><code><a href="loss_circle.html">loss_circle</a>()</code> <br><code><a href="loss_cosine_similarity.html">loss_cosine_similarity</a>()</code> <br><code><a href="loss_ctc.html">loss_ctc</a>()</code> <br><code><a href="loss_dice.html">loss_dice</a>()</code> <br><code><a href="loss_hinge.html">loss_hinge</a>()</code> <br><code><a href="loss_huber.html">loss_huber</a>()</code> <br><code><a href="loss_kl_divergence.html">loss_kl_divergence</a>()</code> <br><code><a href="loss_log_cosh.html">loss_log_cosh</a>()</code> <br><code><a href="loss_mean_absolute_error.html">loss_mean_absolute_error</a>()</code> <br><code><a href="loss_mean_absolute_percentage_error.html">loss_mean_absolute_percentage_error</a>()</code> <br><code><a href="loss_mean_squared_error.html">loss_mean_squared_error</a>()</code> <br><code><a href="loss_mean_squared_logarithmic_error.html">loss_mean_squared_logarithmic_error</a>()</code> <br><code><a href="loss_poisson.html">loss_poisson</a>()</code> <br><code><a href="loss_sparse_categorical_crossentropy.html">loss_sparse_categorical_crossentropy</a>()</code> <br><code><a href="loss_squared_hinge.html">loss_squared_hinge</a>()</code> <br><code><a href="loss_tversky.html">loss_tversky</a>()</code> <br><code><a href="metric_binary_crossentropy.html">metric_binary_crossentropy</a>()</code> <br><code><a href="metric_binary_focal_crossentropy.html">metric_binary_focal_crossentropy</a>()</code> <br><code><a href="metric_categorical_crossentropy.html">metric_categorical_crossentropy</a>()</code> <br><code><a href="metric_categorical_focal_crossentropy.html">metric_categorical_focal_crossentropy</a>()</code> <br><code><a href="metric_categorical_hinge.html">metric_categorical_hinge</a>()</code> <br><code><a href="metric_hinge.html">metric_hinge</a>()</code> <br><code><a href="metric_huber.html">metric_huber</a>()</code> <br><code><a href="metric_kl_divergence.html">metric_kl_divergence</a>()</code> <br><code><a href="metric_log_cosh.html">metric_log_cosh</a>()</code> <br><code><a href="metric_mean_absolute_error.html">metric_mean_absolute_error</a>()</code> <br><code><a href="metric_mean_absolute_percentage_error.html">metric_mean_absolute_percentage_error</a>()</code> <br><code><a href="metric_mean_squared_error.html">metric_mean_squared_error</a>()</code> <br><code><a href="metric_mean_squared_logarithmic_error.html">metric_mean_squared_logarithmic_error</a>()</code> <br><code><a href="metric_poisson.html">metric_poisson</a>()</code> <br><code><a href="metric_sparse_categorical_crossentropy.html">metric_sparse_categorical_crossentropy</a>()</code> <br><code><a href="metric_squared_hinge.html">metric_squared_hinge</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -197,7 +197,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/loss_categorical_focal_crossentropy.html b/docs/dev/reference/loss_categorical_focal_crossentropy.html
index dc4f2f066..104ddf7a9 100644
--- a/docs/dev/reference/loss_categorical_focal_crossentropy.html
+++ b/docs/dev/reference/loss_categorical_focal_crossentropy.html
@@ -64,7 +64,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -270,7 +270,7 @@ <h2 id="examples">Examples<a class="anchor" aria-label="anchor" href="#examples"
     </div>
     <div class="section level2">
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
-    <div class="dont-index"><p>Other losses: <br><code><a href="Loss.html">Loss</a>()</code> <br><code><a href="loss_binary_crossentropy.html">loss_binary_crossentropy</a>()</code> <br><code><a href="loss_binary_focal_crossentropy.html">loss_binary_focal_crossentropy</a>()</code> <br><code><a href="loss_categorical_crossentropy.html">loss_categorical_crossentropy</a>()</code> <br><code><a href="loss_categorical_hinge.html">loss_categorical_hinge</a>()</code> <br><code><a href="loss_circle.html">loss_circle</a>()</code> <br><code><a href="loss_cosine_similarity.html">loss_cosine_similarity</a>()</code> <br><code><a href="loss_ctc.html">loss_ctc</a>()</code> <br><code><a href="loss_dice.html">loss_dice</a>()</code> <br><code><a href="loss_hinge.html">loss_hinge</a>()</code> <br><code><a href="loss_huber.html">loss_huber</a>()</code> <br><code><a href="loss_kl_divergence.html">loss_kl_divergence</a>()</code> <br><code><a href="loss_log_cosh.html">loss_log_cosh</a>()</code> <br><code><a href="loss_mean_absolute_error.html">loss_mean_absolute_error</a>()</code> <br><code><a href="loss_mean_absolute_percentage_error.html">loss_mean_absolute_percentage_error</a>()</code> <br><code><a href="loss_mean_squared_error.html">loss_mean_squared_error</a>()</code> <br><code><a href="loss_mean_squared_logarithmic_error.html">loss_mean_squared_logarithmic_error</a>()</code> <br><code><a href="loss_poisson.html">loss_poisson</a>()</code> <br><code><a href="loss_sparse_categorical_crossentropy.html">loss_sparse_categorical_crossentropy</a>()</code> <br><code><a href="loss_squared_hinge.html">loss_squared_hinge</a>()</code> <br><code><a href="loss_tversky.html">loss_tversky</a>()</code> <br><code><a href="metric_binary_crossentropy.html">metric_binary_crossentropy</a>()</code> <br><code><a href="metric_binary_focal_crossentropy.html">metric_binary_focal_crossentropy</a>()</code> <br><code><a href="metric_categorical_crossentropy.html">metric_categorical_crossentropy</a>()</code> <br><code><a href="metric_categorical_focal_crossentropy.html">metric_categorical_focal_crossentropy</a>()</code> <br><code><a href="metric_categorical_hinge.html">metric_categorical_hinge</a>()</code> <br><code><a href="metric_hinge.html">metric_hinge</a>()</code> <br><code><a href="metric_huber.html">metric_huber</a>()</code> <br><code><a href="metric_kl_divergence.html">metric_kl_divergence</a>()</code> <br><code><a href="metric_log_cosh.html">metric_log_cosh</a>()</code> <br><code><a href="metric_mean_absolute_error.html">metric_mean_absolute_error</a>()</code> <br><code><a href="metric_mean_absolute_percentage_error.html">metric_mean_absolute_percentage_error</a>()</code> <br><code><a href="metric_mean_squared_error.html">metric_mean_squared_error</a>()</code> <br><code><a href="metric_mean_squared_logarithmic_error.html">metric_mean_squared_logarithmic_error</a>()</code> <br><code><a href="metric_poisson.html">metric_poisson</a>()</code> <br><code><a href="metric_sparse_categorical_crossentropy.html">metric_sparse_categorical_crossentropy</a>()</code> <br><code><a href="metric_squared_hinge.html">metric_squared_hinge</a>()</code> <br></p></div>
+    <div class="dont-index"><p>Other losses: <br><code><a href="Loss.html">Loss</a>()</code> <br><code><a href="loss_binary_crossentropy.html">loss_binary_crossentropy</a>()</code> <br><code><a href="loss_binary_focal_crossentropy.html">loss_binary_focal_crossentropy</a>()</code> <br><code><a href="loss_categorical_crossentropy.html">loss_categorical_crossentropy</a>()</code> <br><code><a href="loss_categorical_generalized_cross_entropy.html">loss_categorical_generalized_cross_entropy</a>()</code> <br><code><a href="loss_categorical_hinge.html">loss_categorical_hinge</a>()</code> <br><code><a href="loss_circle.html">loss_circle</a>()</code> <br><code><a href="loss_cosine_similarity.html">loss_cosine_similarity</a>()</code> <br><code><a href="loss_ctc.html">loss_ctc</a>()</code> <br><code><a href="loss_dice.html">loss_dice</a>()</code> <br><code><a href="loss_hinge.html">loss_hinge</a>()</code> <br><code><a href="loss_huber.html">loss_huber</a>()</code> <br><code><a href="loss_kl_divergence.html">loss_kl_divergence</a>()</code> <br><code><a href="loss_log_cosh.html">loss_log_cosh</a>()</code> <br><code><a href="loss_mean_absolute_error.html">loss_mean_absolute_error</a>()</code> <br><code><a href="loss_mean_absolute_percentage_error.html">loss_mean_absolute_percentage_error</a>()</code> <br><code><a href="loss_mean_squared_error.html">loss_mean_squared_error</a>()</code> <br><code><a href="loss_mean_squared_logarithmic_error.html">loss_mean_squared_logarithmic_error</a>()</code> <br><code><a href="loss_poisson.html">loss_poisson</a>()</code> <br><code><a href="loss_sparse_categorical_crossentropy.html">loss_sparse_categorical_crossentropy</a>()</code> <br><code><a href="loss_squared_hinge.html">loss_squared_hinge</a>()</code> <br><code><a href="loss_tversky.html">loss_tversky</a>()</code> <br><code><a href="metric_binary_crossentropy.html">metric_binary_crossentropy</a>()</code> <br><code><a href="metric_binary_focal_crossentropy.html">metric_binary_focal_crossentropy</a>()</code> <br><code><a href="metric_categorical_crossentropy.html">metric_categorical_crossentropy</a>()</code> <br><code><a href="metric_categorical_focal_crossentropy.html">metric_categorical_focal_crossentropy</a>()</code> <br><code><a href="metric_categorical_hinge.html">metric_categorical_hinge</a>()</code> <br><code><a href="metric_hinge.html">metric_hinge</a>()</code> <br><code><a href="metric_huber.html">metric_huber</a>()</code> <br><code><a href="metric_kl_divergence.html">metric_kl_divergence</a>()</code> <br><code><a href="metric_log_cosh.html">metric_log_cosh</a>()</code> <br><code><a href="metric_mean_absolute_error.html">metric_mean_absolute_error</a>()</code> <br><code><a href="metric_mean_absolute_percentage_error.html">metric_mean_absolute_percentage_error</a>()</code> <br><code><a href="metric_mean_squared_error.html">metric_mean_squared_error</a>()</code> <br><code><a href="metric_mean_squared_logarithmic_error.html">metric_mean_squared_logarithmic_error</a>()</code> <br><code><a href="metric_poisson.html">metric_poisson</a>()</code> <br><code><a href="metric_sparse_categorical_crossentropy.html">metric_sparse_categorical_crossentropy</a>()</code> <br><code><a href="metric_squared_hinge.html">metric_squared_hinge</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -282,7 +282,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/loss_categorical_generalized_cross_entropy.html b/docs/dev/reference/loss_categorical_generalized_cross_entropy.html
new file mode 100644
index 000000000..3d087de4b
--- /dev/null
+++ b/docs/dev/reference/loss_categorical_generalized_cross_entropy.html
@@ -0,0 +1,173 @@
+<!DOCTYPE html>
+<!-- Generated by pkgdown: do not edit by hand --><html lang="en"><head><meta http-equiv="Content-Type" content="text/html; charset=UTF-8"><meta charset="utf-8"><meta http-equiv="X-UA-Compatible" content="IE=edge"><meta name="viewport" content="width=device-width, initial-scale=1, shrink-to-fit=no"><title>Computes the generalized cross entropy loss. — loss_categorical_generalized_cross_entropy • keras3</title><!-- favicons --><link rel="icon" type="image/png" sizes="96x96" href="../favicon-96x96.png"><link rel="icon" type="”image/svg+xml”" href="../favicon.svg"><link rel="apple-touch-icon" sizes="180x180" href="../apple-touch-icon.png"><link rel="icon" sizes="any" href="../favicon.ico"><link rel="manifest" href="../site.webmanifest"><script src="../deps/jquery-3.6.0/jquery-3.6.0.min.js"></script><meta name="viewport" content="width=device-width, initial-scale=1, shrink-to-fit=no"><link href="../deps/bootstrap-5.3.1/bootstrap.min.css" rel="stylesheet"><script src="../deps/bootstrap-5.3.1/bootstrap.bundle.min.js"></script><link href="../deps/font-awesome-6.5.2/css/all.min.css" rel="stylesheet"><link href="../deps/font-awesome-6.5.2/css/v4-shims.min.css" rel="stylesheet"><script src="../deps/headroom-0.11.0/headroom.min.js"></script><script src="../deps/headroom-0.11.0/jQuery.headroom.min.js"></script><script src="../deps/bootstrap-toc-1.0.1/bootstrap-toc.min.js"></script><script src="../deps/clipboard.js-2.0.11/clipboard.min.js"></script><script src="../deps/search-1.0.0/autocomplete.jquery.min.js"></script><script src="../deps/search-1.0.0/fuse.min.js"></script><script src="../deps/search-1.0.0/mark.min.js"></script><!-- pkgdown --><script src="../pkgdown.js"></script><link href="../extra.css" rel="stylesheet"><meta property="og:title" content="Computes the generalized cross entropy loss. — loss_categorical_generalized_cross_entropy"><meta name="description" content="The generalized cross entropy (GCE) loss offers robustness to noisy labels by
+interpolating between categorical cross entropy (q -&amp;gt; 0) and mean absolute
+error (q -&amp;gt; 1). For a true-class probability p and noise parameter q,
+the loss is loss = (1 - p^q) / q."><meta property="og:description" content="The generalized cross entropy (GCE) loss offers robustness to noisy labels by
+interpolating between categorical cross entropy (q -&amp;gt; 0) and mean absolute
+error (q -&amp;gt; 1). For a true-class probability p and noise parameter q,
+the loss is loss = (1 - p^q) / q."><meta name="robots" content="noindex"><!-- Google Tag Manager --><script>(function(w,d,s,l,i){w[l]=w[l]||[];w[l].push({'gtm.start': new Date().getTime(),event:'gtm.js'});var f=d.getElementsByTagName(s)[0],j=d.createElement(s),dl=l!='dataLayer'?'&l='+l:'';j.async=true;j.src='https://www.googletagmanager.com/gtm.js?id='+i+dl;f.parentNode.insertBefore(j,f);})(window,document,'script','dataLayer','GTM-KHBDBW7');</script><!-- End Google Tag Manager --></head><body>
+    <a href="#main" class="visually-hidden-focusable">Skip to contents</a>
+    <!-- Google Tag Manager (noscript) --> <noscript><iframe src="https://www.googletagmanager.com/ns.html?id=GTM-KHBDBW7" height="0" width="0" style="display:none;visibility:hidden"></iframe></noscript> <!-- End Google Tag Manager (noscript) -->
+
+
+    <nav class="navbar navbar-expand-lg fixed-top bg-primary" data-bs-theme="inverse" aria-label="Site navigation"><div class="container">
+
+    <a class="navbar-brand me-2" href="../index.html">keras3</a>
+
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
+
+
+    <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
+      <span class="navbar-toggler-icon"></span>
+    </button>
+
+    <div id="navbar" class="collapse navbar-collapse ms-3">
+      <ul class="navbar-nav me-auto"><li class="nav-item"><a class="nav-link" href="../articles/getting_started.html">Getting Started</a></li>
+<li class="nav-item dropdown">
+  <button class="nav-link dropdown-toggle" type="button" id="dropdown-guides" data-bs-toggle="dropdown" aria-expanded="false" aria-haspopup="true">Guides</button>
+  <ul class="dropdown-menu" aria-labelledby="dropdown-guides"><li><h6 class="dropdown-header" data-toc-skip>Model definition</h6></li>
+    <li><a class="dropdown-item" href="../articles/sequential_model.html">Sequential Model</a></li>
+    <li><a class="dropdown-item" href="../articles/functional_api.html">Functional API</a></li>
+    <li><h6 class="dropdown-header" data-toc-skip>Extending and customizing</h6></li>
+    <li><a class="dropdown-item" href="../articles/training_with_built_in_methods.html">Training &amp; evaluation with the built-in methods</a></li>
+    <li><a class="dropdown-item" href="../articles/custom_train_step_in_tensorflow.html">Customizing `fit()` with Tensorflow</a></li>
+    <li><a class="dropdown-item" href="../articles/writing_your_own_callbacks.html">Writing your own callbacks</a></li>
+    <li><a class="dropdown-item" href="../articles/making_new_layers_and_models_via_subclassing.html">Making new layers and models via subclassing</a></li>
+    <li><a class="dropdown-item" href="../articles/writing_a_custom_training_loop_in_tensorflow.html">Writing a training loop from scratch in TensorFlow</a></li>
+    <li><a class="dropdown-item" href="../articles/serialization_and_saving.html">Serialization and Saving</a></li>
+    <li><h6 class="dropdown-header" data-toc-skip>Other topics</h6></li>
+    <li><a class="dropdown-item" href="../articles/transfer_learning.html">Transfer learning and fine tuning</a></li>
+    <li><a class="dropdown-item" href="../articles/distributed_training_with_tensorflow.html">Distributed training with TensorFlow</a></li>
+    <li><a class="dropdown-item" href="../articles/distribution.html">Distributed training with Jax</a></li>
+  </ul></li>
+<li class="nav-item"><a class="nav-link" href="../articles/examples/index.html">Examples</a></li>
+<li class="active nav-item"><a class="nav-link" href="../reference/index.html">Reference</a></li>
+<li class="nav-item"><a class="nav-link" href="../news/index.html">News</a></li>
+      </ul><ul class="navbar-nav"><li class="nav-item"><form class="form-inline" role="search">
+ <input class="form-control" type="search" name="search-input" id="search-input" autocomplete="off" aria-label="Search site" placeholder="Search for" data-search-index="../search.json"></form></li>
+<li class="nav-item"><a class="external-link nav-link" href="https://github.com/rstudio/keras3/" aria-label="GitHub"><span class="fa fab fa-github fa-lg"></span></a></li>
+      </ul></div>
+
+
+  </div>
+</nav><div class="container template-reference-topic">
+<div class="row">
+  <main id="main" class="col-md-9"><div class="page-header">
+
+      <h1>Computes the generalized cross entropy loss.</h1>
+      <small class="dont-index">Source: <a href="https://github.com/rstudio/keras3/blob/HEAD/R/losses.R" class="external-link"><code>R/losses.R</code></a></small>
+      <div class="d-none name"><code>loss_categorical_generalized_cross_entropy.Rd</code></div>
+    </div>
+
+    <div class="ref-description section level2">
+    <p>The generalized cross entropy (GCE) loss offers robustness to noisy labels by
+interpolating between categorical cross entropy (<code>q -&gt; 0</code>) and mean absolute
+error (<code>q -&gt; 1</code>). For a true-class probability <code>p</code> and noise parameter <code>q</code>,
+the loss is <code>loss = (1 - p^q) / q</code>.</p>
+    </div>
+
+    <div class="section level2">
+    <h2 id="ref-usage">Usage<a class="anchor" aria-label="anchor" href="#ref-usage"></a></h2>
+    <div class="sourceCode"><pre class="sourceCode r"><code><span><span class="fu">loss_categorical_generalized_cross_entropy</span><span class="op">(</span></span>
+<span>  <span class="va">y_true</span>,</span>
+<span>  <span class="va">y_pred</span>,</span>
+<span>  q <span class="op">=</span> <span class="fl">0.5</span>,</span>
+<span>  <span class="va">...</span>,</span>
+<span>  reduction <span class="op">=</span> <span class="st">"sum_over_batch_size"</span>,</span>
+<span>  name <span class="op">=</span> <span class="st">"categorical_generalized_cross_entropy"</span>,</span>
+<span>  dtype <span class="op">=</span> <span class="cn">NULL</span></span>
+<span><span class="op">)</span></span></code></pre></div>
+    </div>
+
+    <div class="section level2">
+    <h2 id="arguments">Arguments<a class="anchor" aria-label="anchor" href="#arguments"></a></h2>
+
+
+<dl><dt id="arg-y-true">y_true<a class="anchor" aria-label="anchor" href="#arg-y-true"></a></dt>
+<dd><p>Integer class indices with shape <code>(batch_size)</code> or <code>(batch_size, 1)</code>.</p></dd>
+
+
+<dt id="arg-y-pred">y_pred<a class="anchor" aria-label="anchor" href="#arg-y-pred"></a></dt>
+<dd><p>Predicted class probabilities with shape <code>(batch_size, num_classes)</code>.</p></dd>
+
+
+<dt id="arg-q">q<a class="anchor" aria-label="anchor" href="#arg-q"></a></dt>
+<dd><p>Float in <code>(0, 1)</code>. Controls the transition between cross entropy and mean
+absolute error. Defaults to <code>0.5</code>.</p><ul><li><p>As <code>q</code> approaches <code>0</code>: behaves like categorical cross entropy.</p></li>
+<li><p>As <code>q</code> approaches <code>1</code>: behaves like mean absolute error.</p></li>
+</ul></dd>
+
+
+<dt id="arg--">...<a class="anchor" aria-label="anchor" href="#arg--"></a></dt>
+<dd><p>For forward/backward compatibility.</p></dd>
+
+
+<dt id="arg-reduction">reduction<a class="anchor" aria-label="anchor" href="#arg-reduction"></a></dt>
+<dd><p>Type of reduction to apply to the loss. In almost all cases
+this should be <code>"sum_over_batch_size"</code>. Supported options are
+<code>"sum"</code>, <code>"sum_over_batch_size"</code>, <code>"mean"</code>,
+<code>"mean_with_sample_weight"</code> or <code>NULL</code>. <code>"sum"</code> sums the loss,
+<code>"sum_over_batch_size"</code> and <code>"mean"</code> sum the loss and divide by the
+sample size, and <code>"mean_with_sample_weight"</code> sums the loss and
+divides by the sum of the sample weights. <code>"none"</code> and <code>NULL</code>
+perform no aggregation. Defaults to <code>"sum_over_batch_size"</code>.</p></dd>
+
+
+<dt id="arg-name">name<a class="anchor" aria-label="anchor" href="#arg-name"></a></dt>
+<dd><p>Optional name for the loss instance.</p></dd>
+
+
+<dt id="arg-dtype">dtype<a class="anchor" aria-label="anchor" href="#arg-dtype"></a></dt>
+<dd><p>Dtype used for loss computations. Defaults to <code><a href="config_floatx.html">config_floatx()</a></code> (the global
+float type).</p></dd>
+
+</dl></div>
+    <div class="section level2">
+    <h2 id="value">Value<a class="anchor" aria-label="anchor" href="#value"></a></h2>
+    <p>Generalized cross entropy loss value(s).</p>
+    </div>
+    <div class="section level2">
+    <h2 id="references">References<a class="anchor" aria-label="anchor" href="#references"></a></h2>
+
+<ul><li><p>Zhang &amp; Sabuncu (2018), "Generalized Cross Entropy Loss for Training Deep
+Neural Networks with Noisy Labels"</p></li>
+</ul></div>
+    <div class="section level2">
+    <h2 id="examples">Examples<a class="anchor" aria-label="anchor" href="#examples"></a></h2>
+    <p></p><div class="sourceCode r"><pre><code><span><span class="va">y_true</span> <span class="op">&lt;-</span> <span class="fu"><a href="https://rdrr.io/r/base/c.html" class="external-link">c</a></span><span class="op">(</span><span class="fl">0L</span>, <span class="fl">1L</span>, <span class="fl">0L</span>, <span class="fl">1L</span><span class="op">)</span></span>
+<span><span class="va">y_pred</span> <span class="op">&lt;-</span> <span class="fu"><a href="https://rdrr.io/r/base/cbind.html" class="external-link">rbind</a></span><span class="op">(</span></span>
+<span>  <span class="fu"><a href="https://rdrr.io/r/base/c.html" class="external-link">c</a></span><span class="op">(</span><span class="fl">0.7</span>, <span class="fl">0.3</span><span class="op">)</span>,</span>
+<span>  <span class="fu"><a href="https://rdrr.io/r/base/c.html" class="external-link">c</a></span><span class="op">(</span><span class="fl">0.2</span>, <span class="fl">0.8</span><span class="op">)</span>,</span>
+<span>  <span class="fu"><a href="https://rdrr.io/r/base/c.html" class="external-link">c</a></span><span class="op">(</span><span class="fl">0.6</span>, <span class="fl">0.4</span><span class="op">)</span>,</span>
+<span>  <span class="fu"><a href="https://rdrr.io/r/base/c.html" class="external-link">c</a></span><span class="op">(</span><span class="fl">0.4</span>, <span class="fl">0.6</span><span class="op">)</span></span>
+<span><span class="op">)</span></span>
+<span><span class="va">gce</span> <span class="op">&lt;-</span> <span class="fu"><a href="../reference/loss_categorical_generalized_cross_entropy.html">loss_categorical_generalized_cross_entropy</a></span><span class="op">(</span>q <span class="op">=</span> <span class="fl">0.7</span><span class="op">)</span></span>
+<span><span class="fu">gce</span><span class="op">(</span><span class="va">y_true</span>, <span class="va">y_pred</span><span class="op">)</span></span></code></pre><p></p></div>
+<p></p><div class="sourceCode"><pre><code><span id="cb1-1"><a href="#cb1-1" tabindex="-1"></a><span class="do">## tf.Tensor(0.34529287, shape=(), dtype=float32)</span></span>
+<span id="cb1-2"><a href="#cb1-2" tabindex="-1"></a></span></code></pre><p></p></div>
+    </div>
+    <div class="section level2">
+    <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
+    <div class="dont-index"><p>Other losses: <br><code><a href="Loss.html">Loss</a>()</code> <br><code><a href="loss_binary_crossentropy.html">loss_binary_crossentropy</a>()</code> <br><code><a href="loss_binary_focal_crossentropy.html">loss_binary_focal_crossentropy</a>()</code> <br><code><a href="loss_categorical_crossentropy.html">loss_categorical_crossentropy</a>()</code> <br><code><a href="loss_categorical_focal_crossentropy.html">loss_categorical_focal_crossentropy</a>()</code> <br><code><a href="loss_categorical_hinge.html">loss_categorical_hinge</a>()</code> <br><code><a href="loss_circle.html">loss_circle</a>()</code> <br><code><a href="loss_cosine_similarity.html">loss_cosine_similarity</a>()</code> <br><code><a href="loss_ctc.html">loss_ctc</a>()</code> <br><code><a href="loss_dice.html">loss_dice</a>()</code> <br><code><a href="loss_hinge.html">loss_hinge</a>()</code> <br><code><a href="loss_huber.html">loss_huber</a>()</code> <br><code><a href="loss_kl_divergence.html">loss_kl_divergence</a>()</code> <br><code><a href="loss_log_cosh.html">loss_log_cosh</a>()</code> <br><code><a href="loss_mean_absolute_error.html">loss_mean_absolute_error</a>()</code> <br><code><a href="loss_mean_absolute_percentage_error.html">loss_mean_absolute_percentage_error</a>()</code> <br><code><a href="loss_mean_squared_error.html">loss_mean_squared_error</a>()</code> <br><code><a href="loss_mean_squared_logarithmic_error.html">loss_mean_squared_logarithmic_error</a>()</code> <br><code><a href="loss_poisson.html">loss_poisson</a>()</code> <br><code><a href="loss_sparse_categorical_crossentropy.html">loss_sparse_categorical_crossentropy</a>()</code> <br><code><a href="loss_squared_hinge.html">loss_squared_hinge</a>()</code> <br><code><a href="loss_tversky.html">loss_tversky</a>()</code> <br><code><a href="metric_binary_crossentropy.html">metric_binary_crossentropy</a>()</code> <br><code><a href="metric_binary_focal_crossentropy.html">metric_binary_focal_crossentropy</a>()</code> <br><code><a href="metric_categorical_crossentropy.html">metric_categorical_crossentropy</a>()</code> <br><code><a href="metric_categorical_focal_crossentropy.html">metric_categorical_focal_crossentropy</a>()</code> <br><code><a href="metric_categorical_hinge.html">metric_categorical_hinge</a>()</code> <br><code><a href="metric_hinge.html">metric_hinge</a>()</code> <br><code><a href="metric_huber.html">metric_huber</a>()</code> <br><code><a href="metric_kl_divergence.html">metric_kl_divergence</a>()</code> <br><code><a href="metric_log_cosh.html">metric_log_cosh</a>()</code> <br><code><a href="metric_mean_absolute_error.html">metric_mean_absolute_error</a>()</code> <br><code><a href="metric_mean_absolute_percentage_error.html">metric_mean_absolute_percentage_error</a>()</code> <br><code><a href="metric_mean_squared_error.html">metric_mean_squared_error</a>()</code> <br><code><a href="metric_mean_squared_logarithmic_error.html">metric_mean_squared_logarithmic_error</a>()</code> <br><code><a href="metric_poisson.html">metric_poisson</a>()</code> <br><code><a href="metric_sparse_categorical_crossentropy.html">metric_sparse_categorical_crossentropy</a>()</code> <br><code><a href="metric_squared_hinge.html">metric_squared_hinge</a>()</code> <br></p></div>
+    </div>
+
+  </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
+    </nav></aside></div>
+
+
+    <footer><div class="pkgdown-footer-left">
+  <p>Developed by Tomasz Kalinowski, JJ Allaire, François Chollet, Posit Software, PBC, Google.</p>
+</div>
+
+<div class="pkgdown-footer-right">
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
+</div>
+
+    </footer></div>
+
+
+
+
+
+  </body></html>
+
diff --git a/docs/dev/reference/loss_categorical_hinge.html b/docs/dev/reference/loss_categorical_hinge.html
index b892434bf..0775edc3e 100644
--- a/docs/dev/reference/loss_categorical_hinge.html
+++ b/docs/dev/reference/loss_categorical_hinge.html
@@ -14,7 +14,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -132,7 +132,7 @@ <h2 id="examples">Examples<a class="anchor" aria-label="anchor" href="#examples"
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
     <div class="dont-index">
 <ul><li><p><a href="https://keras.io/api/losses/hinge_losses#categoricalhinge-class" class="external-link">https://keras.io/api/losses/hinge_losses#categoricalhinge-class</a></p></li>
-</ul><p>Other losses: <br><code><a href="Loss.html">Loss</a>()</code> <br><code><a href="loss_binary_crossentropy.html">loss_binary_crossentropy</a>()</code> <br><code><a href="loss_binary_focal_crossentropy.html">loss_binary_focal_crossentropy</a>()</code> <br><code><a href="loss_categorical_crossentropy.html">loss_categorical_crossentropy</a>()</code> <br><code><a href="loss_categorical_focal_crossentropy.html">loss_categorical_focal_crossentropy</a>()</code> <br><code><a href="loss_circle.html">loss_circle</a>()</code> <br><code><a href="loss_cosine_similarity.html">loss_cosine_similarity</a>()</code> <br><code><a href="loss_ctc.html">loss_ctc</a>()</code> <br><code><a href="loss_dice.html">loss_dice</a>()</code> <br><code><a href="loss_hinge.html">loss_hinge</a>()</code> <br><code><a href="loss_huber.html">loss_huber</a>()</code> <br><code><a href="loss_kl_divergence.html">loss_kl_divergence</a>()</code> <br><code><a href="loss_log_cosh.html">loss_log_cosh</a>()</code> <br><code><a href="loss_mean_absolute_error.html">loss_mean_absolute_error</a>()</code> <br><code><a href="loss_mean_absolute_percentage_error.html">loss_mean_absolute_percentage_error</a>()</code> <br><code><a href="loss_mean_squared_error.html">loss_mean_squared_error</a>()</code> <br><code><a href="loss_mean_squared_logarithmic_error.html">loss_mean_squared_logarithmic_error</a>()</code> <br><code><a href="loss_poisson.html">loss_poisson</a>()</code> <br><code><a href="loss_sparse_categorical_crossentropy.html">loss_sparse_categorical_crossentropy</a>()</code> <br><code><a href="loss_squared_hinge.html">loss_squared_hinge</a>()</code> <br><code><a href="loss_tversky.html">loss_tversky</a>()</code> <br><code><a href="metric_binary_crossentropy.html">metric_binary_crossentropy</a>()</code> <br><code><a href="metric_binary_focal_crossentropy.html">metric_binary_focal_crossentropy</a>()</code> <br><code><a href="metric_categorical_crossentropy.html">metric_categorical_crossentropy</a>()</code> <br><code><a href="metric_categorical_focal_crossentropy.html">metric_categorical_focal_crossentropy</a>()</code> <br><code><a href="metric_categorical_hinge.html">metric_categorical_hinge</a>()</code> <br><code><a href="metric_hinge.html">metric_hinge</a>()</code> <br><code><a href="metric_huber.html">metric_huber</a>()</code> <br><code><a href="metric_kl_divergence.html">metric_kl_divergence</a>()</code> <br><code><a href="metric_log_cosh.html">metric_log_cosh</a>()</code> <br><code><a href="metric_mean_absolute_error.html">metric_mean_absolute_error</a>()</code> <br><code><a href="metric_mean_absolute_percentage_error.html">metric_mean_absolute_percentage_error</a>()</code> <br><code><a href="metric_mean_squared_error.html">metric_mean_squared_error</a>()</code> <br><code><a href="metric_mean_squared_logarithmic_error.html">metric_mean_squared_logarithmic_error</a>()</code> <br><code><a href="metric_poisson.html">metric_poisson</a>()</code> <br><code><a href="metric_sparse_categorical_crossentropy.html">metric_sparse_categorical_crossentropy</a>()</code> <br><code><a href="metric_squared_hinge.html">metric_squared_hinge</a>()</code> <br></p></div>
+</ul><p>Other losses: <br><code><a href="Loss.html">Loss</a>()</code> <br><code><a href="loss_binary_crossentropy.html">loss_binary_crossentropy</a>()</code> <br><code><a href="loss_binary_focal_crossentropy.html">loss_binary_focal_crossentropy</a>()</code> <br><code><a href="loss_categorical_crossentropy.html">loss_categorical_crossentropy</a>()</code> <br><code><a href="loss_categorical_focal_crossentropy.html">loss_categorical_focal_crossentropy</a>()</code> <br><code><a href="loss_categorical_generalized_cross_entropy.html">loss_categorical_generalized_cross_entropy</a>()</code> <br><code><a href="loss_circle.html">loss_circle</a>()</code> <br><code><a href="loss_cosine_similarity.html">loss_cosine_similarity</a>()</code> <br><code><a href="loss_ctc.html">loss_ctc</a>()</code> <br><code><a href="loss_dice.html">loss_dice</a>()</code> <br><code><a href="loss_hinge.html">loss_hinge</a>()</code> <br><code><a href="loss_huber.html">loss_huber</a>()</code> <br><code><a href="loss_kl_divergence.html">loss_kl_divergence</a>()</code> <br><code><a href="loss_log_cosh.html">loss_log_cosh</a>()</code> <br><code><a href="loss_mean_absolute_error.html">loss_mean_absolute_error</a>()</code> <br><code><a href="loss_mean_absolute_percentage_error.html">loss_mean_absolute_percentage_error</a>()</code> <br><code><a href="loss_mean_squared_error.html">loss_mean_squared_error</a>()</code> <br><code><a href="loss_mean_squared_logarithmic_error.html">loss_mean_squared_logarithmic_error</a>()</code> <br><code><a href="loss_poisson.html">loss_poisson</a>()</code> <br><code><a href="loss_sparse_categorical_crossentropy.html">loss_sparse_categorical_crossentropy</a>()</code> <br><code><a href="loss_squared_hinge.html">loss_squared_hinge</a>()</code> <br><code><a href="loss_tversky.html">loss_tversky</a>()</code> <br><code><a href="metric_binary_crossentropy.html">metric_binary_crossentropy</a>()</code> <br><code><a href="metric_binary_focal_crossentropy.html">metric_binary_focal_crossentropy</a>()</code> <br><code><a href="metric_categorical_crossentropy.html">metric_categorical_crossentropy</a>()</code> <br><code><a href="metric_categorical_focal_crossentropy.html">metric_categorical_focal_crossentropy</a>()</code> <br><code><a href="metric_categorical_hinge.html">metric_categorical_hinge</a>()</code> <br><code><a href="metric_hinge.html">metric_hinge</a>()</code> <br><code><a href="metric_huber.html">metric_huber</a>()</code> <br><code><a href="metric_kl_divergence.html">metric_kl_divergence</a>()</code> <br><code><a href="metric_log_cosh.html">metric_log_cosh</a>()</code> <br><code><a href="metric_mean_absolute_error.html">metric_mean_absolute_error</a>()</code> <br><code><a href="metric_mean_absolute_percentage_error.html">metric_mean_absolute_percentage_error</a>()</code> <br><code><a href="metric_mean_squared_error.html">metric_mean_squared_error</a>()</code> <br><code><a href="metric_mean_squared_logarithmic_error.html">metric_mean_squared_logarithmic_error</a>()</code> <br><code><a href="metric_poisson.html">metric_poisson</a>()</code> <br><code><a href="metric_sparse_categorical_crossentropy.html">metric_sparse_categorical_crossentropy</a>()</code> <br><code><a href="metric_squared_hinge.html">metric_squared_hinge</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -144,7 +144,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/loss_circle.html b/docs/dev/reference/loss_circle.html
index bf6e401fa..4e82a67b9 100644
--- a/docs/dev/reference/loss_circle.html
+++ b/docs/dev/reference/loss_circle.html
@@ -24,7 +24,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -187,7 +187,7 @@ <h2 id="reference">Reference<a class="anchor" aria-label="anchor" href="#referen
 </ul></div>
     <div class="section level2">
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
-    <div class="dont-index"><p>Other losses: <br><code><a href="Loss.html">Loss</a>()</code> <br><code><a href="loss_binary_crossentropy.html">loss_binary_crossentropy</a>()</code> <br><code><a href="loss_binary_focal_crossentropy.html">loss_binary_focal_crossentropy</a>()</code> <br><code><a href="loss_categorical_crossentropy.html">loss_categorical_crossentropy</a>()</code> <br><code><a href="loss_categorical_focal_crossentropy.html">loss_categorical_focal_crossentropy</a>()</code> <br><code><a href="loss_categorical_hinge.html">loss_categorical_hinge</a>()</code> <br><code><a href="loss_cosine_similarity.html">loss_cosine_similarity</a>()</code> <br><code><a href="loss_ctc.html">loss_ctc</a>()</code> <br><code><a href="loss_dice.html">loss_dice</a>()</code> <br><code><a href="loss_hinge.html">loss_hinge</a>()</code> <br><code><a href="loss_huber.html">loss_huber</a>()</code> <br><code><a href="loss_kl_divergence.html">loss_kl_divergence</a>()</code> <br><code><a href="loss_log_cosh.html">loss_log_cosh</a>()</code> <br><code><a href="loss_mean_absolute_error.html">loss_mean_absolute_error</a>()</code> <br><code><a href="loss_mean_absolute_percentage_error.html">loss_mean_absolute_percentage_error</a>()</code> <br><code><a href="loss_mean_squared_error.html">loss_mean_squared_error</a>()</code> <br><code><a href="loss_mean_squared_logarithmic_error.html">loss_mean_squared_logarithmic_error</a>()</code> <br><code><a href="loss_poisson.html">loss_poisson</a>()</code> <br><code><a href="loss_sparse_categorical_crossentropy.html">loss_sparse_categorical_crossentropy</a>()</code> <br><code><a href="loss_squared_hinge.html">loss_squared_hinge</a>()</code> <br><code><a href="loss_tversky.html">loss_tversky</a>()</code> <br><code><a href="metric_binary_crossentropy.html">metric_binary_crossentropy</a>()</code> <br><code><a href="metric_binary_focal_crossentropy.html">metric_binary_focal_crossentropy</a>()</code> <br><code><a href="metric_categorical_crossentropy.html">metric_categorical_crossentropy</a>()</code> <br><code><a href="metric_categorical_focal_crossentropy.html">metric_categorical_focal_crossentropy</a>()</code> <br><code><a href="metric_categorical_hinge.html">metric_categorical_hinge</a>()</code> <br><code><a href="metric_hinge.html">metric_hinge</a>()</code> <br><code><a href="metric_huber.html">metric_huber</a>()</code> <br><code><a href="metric_kl_divergence.html">metric_kl_divergence</a>()</code> <br><code><a href="metric_log_cosh.html">metric_log_cosh</a>()</code> <br><code><a href="metric_mean_absolute_error.html">metric_mean_absolute_error</a>()</code> <br><code><a href="metric_mean_absolute_percentage_error.html">metric_mean_absolute_percentage_error</a>()</code> <br><code><a href="metric_mean_squared_error.html">metric_mean_squared_error</a>()</code> <br><code><a href="metric_mean_squared_logarithmic_error.html">metric_mean_squared_logarithmic_error</a>()</code> <br><code><a href="metric_poisson.html">metric_poisson</a>()</code> <br><code><a href="metric_sparse_categorical_crossentropy.html">metric_sparse_categorical_crossentropy</a>()</code> <br><code><a href="metric_squared_hinge.html">metric_squared_hinge</a>()</code> <br></p></div>
+    <div class="dont-index"><p>Other losses: <br><code><a href="Loss.html">Loss</a>()</code> <br><code><a href="loss_binary_crossentropy.html">loss_binary_crossentropy</a>()</code> <br><code><a href="loss_binary_focal_crossentropy.html">loss_binary_focal_crossentropy</a>()</code> <br><code><a href="loss_categorical_crossentropy.html">loss_categorical_crossentropy</a>()</code> <br><code><a href="loss_categorical_focal_crossentropy.html">loss_categorical_focal_crossentropy</a>()</code> <br><code><a href="loss_categorical_generalized_cross_entropy.html">loss_categorical_generalized_cross_entropy</a>()</code> <br><code><a href="loss_categorical_hinge.html">loss_categorical_hinge</a>()</code> <br><code><a href="loss_cosine_similarity.html">loss_cosine_similarity</a>()</code> <br><code><a href="loss_ctc.html">loss_ctc</a>()</code> <br><code><a href="loss_dice.html">loss_dice</a>()</code> <br><code><a href="loss_hinge.html">loss_hinge</a>()</code> <br><code><a href="loss_huber.html">loss_huber</a>()</code> <br><code><a href="loss_kl_divergence.html">loss_kl_divergence</a>()</code> <br><code><a href="loss_log_cosh.html">loss_log_cosh</a>()</code> <br><code><a href="loss_mean_absolute_error.html">loss_mean_absolute_error</a>()</code> <br><code><a href="loss_mean_absolute_percentage_error.html">loss_mean_absolute_percentage_error</a>()</code> <br><code><a href="loss_mean_squared_error.html">loss_mean_squared_error</a>()</code> <br><code><a href="loss_mean_squared_logarithmic_error.html">loss_mean_squared_logarithmic_error</a>()</code> <br><code><a href="loss_poisson.html">loss_poisson</a>()</code> <br><code><a href="loss_sparse_categorical_crossentropy.html">loss_sparse_categorical_crossentropy</a>()</code> <br><code><a href="loss_squared_hinge.html">loss_squared_hinge</a>()</code> <br><code><a href="loss_tversky.html">loss_tversky</a>()</code> <br><code><a href="metric_binary_crossentropy.html">metric_binary_crossentropy</a>()</code> <br><code><a href="metric_binary_focal_crossentropy.html">metric_binary_focal_crossentropy</a>()</code> <br><code><a href="metric_categorical_crossentropy.html">metric_categorical_crossentropy</a>()</code> <br><code><a href="metric_categorical_focal_crossentropy.html">metric_categorical_focal_crossentropy</a>()</code> <br><code><a href="metric_categorical_hinge.html">metric_categorical_hinge</a>()</code> <br><code><a href="metric_hinge.html">metric_hinge</a>()</code> <br><code><a href="metric_huber.html">metric_huber</a>()</code> <br><code><a href="metric_kl_divergence.html">metric_kl_divergence</a>()</code> <br><code><a href="metric_log_cosh.html">metric_log_cosh</a>()</code> <br><code><a href="metric_mean_absolute_error.html">metric_mean_absolute_error</a>()</code> <br><code><a href="metric_mean_absolute_percentage_error.html">metric_mean_absolute_percentage_error</a>()</code> <br><code><a href="metric_mean_squared_error.html">metric_mean_squared_error</a>()</code> <br><code><a href="metric_mean_squared_logarithmic_error.html">metric_mean_squared_logarithmic_error</a>()</code> <br><code><a href="metric_poisson.html">metric_poisson</a>()</code> <br><code><a href="metric_sparse_categorical_crossentropy.html">metric_sparse_categorical_crossentropy</a>()</code> <br><code><a href="metric_squared_hinge.html">metric_squared_hinge</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -199,7 +199,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/loss_cosine_similarity.html b/docs/dev/reference/loss_cosine_similarity.html
index b89ce00d5..d80a569fa 100644
--- a/docs/dev/reference/loss_cosine_similarity.html
+++ b/docs/dev/reference/loss_cosine_similarity.html
@@ -26,7 +26,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -157,7 +157,7 @@ <h2 id="examples">Examples<a class="anchor" aria-label="anchor" href="#examples"
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
     <div class="dont-index">
 <ul><li><p><a href="https://keras.io/api/losses/regression_losses#cosinesimilarity-class" class="external-link">https://keras.io/api/losses/regression_losses#cosinesimilarity-class</a></p></li>
-</ul><p>Other losses: <br><code><a href="Loss.html">Loss</a>()</code> <br><code><a href="loss_binary_crossentropy.html">loss_binary_crossentropy</a>()</code> <br><code><a href="loss_binary_focal_crossentropy.html">loss_binary_focal_crossentropy</a>()</code> <br><code><a href="loss_categorical_crossentropy.html">loss_categorical_crossentropy</a>()</code> <br><code><a href="loss_categorical_focal_crossentropy.html">loss_categorical_focal_crossentropy</a>()</code> <br><code><a href="loss_categorical_hinge.html">loss_categorical_hinge</a>()</code> <br><code><a href="loss_circle.html">loss_circle</a>()</code> <br><code><a href="loss_ctc.html">loss_ctc</a>()</code> <br><code><a href="loss_dice.html">loss_dice</a>()</code> <br><code><a href="loss_hinge.html">loss_hinge</a>()</code> <br><code><a href="loss_huber.html">loss_huber</a>()</code> <br><code><a href="loss_kl_divergence.html">loss_kl_divergence</a>()</code> <br><code><a href="loss_log_cosh.html">loss_log_cosh</a>()</code> <br><code><a href="loss_mean_absolute_error.html">loss_mean_absolute_error</a>()</code> <br><code><a href="loss_mean_absolute_percentage_error.html">loss_mean_absolute_percentage_error</a>()</code> <br><code><a href="loss_mean_squared_error.html">loss_mean_squared_error</a>()</code> <br><code><a href="loss_mean_squared_logarithmic_error.html">loss_mean_squared_logarithmic_error</a>()</code> <br><code><a href="loss_poisson.html">loss_poisson</a>()</code> <br><code><a href="loss_sparse_categorical_crossentropy.html">loss_sparse_categorical_crossentropy</a>()</code> <br><code><a href="loss_squared_hinge.html">loss_squared_hinge</a>()</code> <br><code><a href="loss_tversky.html">loss_tversky</a>()</code> <br><code><a href="metric_binary_crossentropy.html">metric_binary_crossentropy</a>()</code> <br><code><a href="metric_binary_focal_crossentropy.html">metric_binary_focal_crossentropy</a>()</code> <br><code><a href="metric_categorical_crossentropy.html">metric_categorical_crossentropy</a>()</code> <br><code><a href="metric_categorical_focal_crossentropy.html">metric_categorical_focal_crossentropy</a>()</code> <br><code><a href="metric_categorical_hinge.html">metric_categorical_hinge</a>()</code> <br><code><a href="metric_hinge.html">metric_hinge</a>()</code> <br><code><a href="metric_huber.html">metric_huber</a>()</code> <br><code><a href="metric_kl_divergence.html">metric_kl_divergence</a>()</code> <br><code><a href="metric_log_cosh.html">metric_log_cosh</a>()</code> <br><code><a href="metric_mean_absolute_error.html">metric_mean_absolute_error</a>()</code> <br><code><a href="metric_mean_absolute_percentage_error.html">metric_mean_absolute_percentage_error</a>()</code> <br><code><a href="metric_mean_squared_error.html">metric_mean_squared_error</a>()</code> <br><code><a href="metric_mean_squared_logarithmic_error.html">metric_mean_squared_logarithmic_error</a>()</code> <br><code><a href="metric_poisson.html">metric_poisson</a>()</code> <br><code><a href="metric_sparse_categorical_crossentropy.html">metric_sparse_categorical_crossentropy</a>()</code> <br><code><a href="metric_squared_hinge.html">metric_squared_hinge</a>()</code> <br></p></div>
+</ul><p>Other losses: <br><code><a href="Loss.html">Loss</a>()</code> <br><code><a href="loss_binary_crossentropy.html">loss_binary_crossentropy</a>()</code> <br><code><a href="loss_binary_focal_crossentropy.html">loss_binary_focal_crossentropy</a>()</code> <br><code><a href="loss_categorical_crossentropy.html">loss_categorical_crossentropy</a>()</code> <br><code><a href="loss_categorical_focal_crossentropy.html">loss_categorical_focal_crossentropy</a>()</code> <br><code><a href="loss_categorical_generalized_cross_entropy.html">loss_categorical_generalized_cross_entropy</a>()</code> <br><code><a href="loss_categorical_hinge.html">loss_categorical_hinge</a>()</code> <br><code><a href="loss_circle.html">loss_circle</a>()</code> <br><code><a href="loss_ctc.html">loss_ctc</a>()</code> <br><code><a href="loss_dice.html">loss_dice</a>()</code> <br><code><a href="loss_hinge.html">loss_hinge</a>()</code> <br><code><a href="loss_huber.html">loss_huber</a>()</code> <br><code><a href="loss_kl_divergence.html">loss_kl_divergence</a>()</code> <br><code><a href="loss_log_cosh.html">loss_log_cosh</a>()</code> <br><code><a href="loss_mean_absolute_error.html">loss_mean_absolute_error</a>()</code> <br><code><a href="loss_mean_absolute_percentage_error.html">loss_mean_absolute_percentage_error</a>()</code> <br><code><a href="loss_mean_squared_error.html">loss_mean_squared_error</a>()</code> <br><code><a href="loss_mean_squared_logarithmic_error.html">loss_mean_squared_logarithmic_error</a>()</code> <br><code><a href="loss_poisson.html">loss_poisson</a>()</code> <br><code><a href="loss_sparse_categorical_crossentropy.html">loss_sparse_categorical_crossentropy</a>()</code> <br><code><a href="loss_squared_hinge.html">loss_squared_hinge</a>()</code> <br><code><a href="loss_tversky.html">loss_tversky</a>()</code> <br><code><a href="metric_binary_crossentropy.html">metric_binary_crossentropy</a>()</code> <br><code><a href="metric_binary_focal_crossentropy.html">metric_binary_focal_crossentropy</a>()</code> <br><code><a href="metric_categorical_crossentropy.html">metric_categorical_crossentropy</a>()</code> <br><code><a href="metric_categorical_focal_crossentropy.html">metric_categorical_focal_crossentropy</a>()</code> <br><code><a href="metric_categorical_hinge.html">metric_categorical_hinge</a>()</code> <br><code><a href="metric_hinge.html">metric_hinge</a>()</code> <br><code><a href="metric_huber.html">metric_huber</a>()</code> <br><code><a href="metric_kl_divergence.html">metric_kl_divergence</a>()</code> <br><code><a href="metric_log_cosh.html">metric_log_cosh</a>()</code> <br><code><a href="metric_mean_absolute_error.html">metric_mean_absolute_error</a>()</code> <br><code><a href="metric_mean_absolute_percentage_error.html">metric_mean_absolute_percentage_error</a>()</code> <br><code><a href="metric_mean_squared_error.html">metric_mean_squared_error</a>()</code> <br><code><a href="metric_mean_squared_logarithmic_error.html">metric_mean_squared_logarithmic_error</a>()</code> <br><code><a href="metric_poisson.html">metric_poisson</a>()</code> <br><code><a href="metric_sparse_categorical_crossentropy.html">metric_sparse_categorical_crossentropy</a>()</code> <br><code><a href="metric_squared_hinge.html">metric_squared_hinge</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -169,7 +169,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/loss_ctc.html b/docs/dev/reference/loss_ctc.html
index 110e16ca7..cc314a9da 100644
--- a/docs/dev/reference/loss_ctc.html
+++ b/docs/dev/reference/loss_ctc.html
@@ -8,7 +8,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -118,7 +118,7 @@ <h2 id="value">Value<a class="anchor" aria-label="anchor" href="#value"></a></h2
     </div>
     <div class="section level2">
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
-    <div class="dont-index"><p>Other losses: <br><code><a href="Loss.html">Loss</a>()</code> <br><code><a href="loss_binary_crossentropy.html">loss_binary_crossentropy</a>()</code> <br><code><a href="loss_binary_focal_crossentropy.html">loss_binary_focal_crossentropy</a>()</code> <br><code><a href="loss_categorical_crossentropy.html">loss_categorical_crossentropy</a>()</code> <br><code><a href="loss_categorical_focal_crossentropy.html">loss_categorical_focal_crossentropy</a>()</code> <br><code><a href="loss_categorical_hinge.html">loss_categorical_hinge</a>()</code> <br><code><a href="loss_circle.html">loss_circle</a>()</code> <br><code><a href="loss_cosine_similarity.html">loss_cosine_similarity</a>()</code> <br><code><a href="loss_dice.html">loss_dice</a>()</code> <br><code><a href="loss_hinge.html">loss_hinge</a>()</code> <br><code><a href="loss_huber.html">loss_huber</a>()</code> <br><code><a href="loss_kl_divergence.html">loss_kl_divergence</a>()</code> <br><code><a href="loss_log_cosh.html">loss_log_cosh</a>()</code> <br><code><a href="loss_mean_absolute_error.html">loss_mean_absolute_error</a>()</code> <br><code><a href="loss_mean_absolute_percentage_error.html">loss_mean_absolute_percentage_error</a>()</code> <br><code><a href="loss_mean_squared_error.html">loss_mean_squared_error</a>()</code> <br><code><a href="loss_mean_squared_logarithmic_error.html">loss_mean_squared_logarithmic_error</a>()</code> <br><code><a href="loss_poisson.html">loss_poisson</a>()</code> <br><code><a href="loss_sparse_categorical_crossentropy.html">loss_sparse_categorical_crossentropy</a>()</code> <br><code><a href="loss_squared_hinge.html">loss_squared_hinge</a>()</code> <br><code><a href="loss_tversky.html">loss_tversky</a>()</code> <br><code><a href="metric_binary_crossentropy.html">metric_binary_crossentropy</a>()</code> <br><code><a href="metric_binary_focal_crossentropy.html">metric_binary_focal_crossentropy</a>()</code> <br><code><a href="metric_categorical_crossentropy.html">metric_categorical_crossentropy</a>()</code> <br><code><a href="metric_categorical_focal_crossentropy.html">metric_categorical_focal_crossentropy</a>()</code> <br><code><a href="metric_categorical_hinge.html">metric_categorical_hinge</a>()</code> <br><code><a href="metric_hinge.html">metric_hinge</a>()</code> <br><code><a href="metric_huber.html">metric_huber</a>()</code> <br><code><a href="metric_kl_divergence.html">metric_kl_divergence</a>()</code> <br><code><a href="metric_log_cosh.html">metric_log_cosh</a>()</code> <br><code><a href="metric_mean_absolute_error.html">metric_mean_absolute_error</a>()</code> <br><code><a href="metric_mean_absolute_percentage_error.html">metric_mean_absolute_percentage_error</a>()</code> <br><code><a href="metric_mean_squared_error.html">metric_mean_squared_error</a>()</code> <br><code><a href="metric_mean_squared_logarithmic_error.html">metric_mean_squared_logarithmic_error</a>()</code> <br><code><a href="metric_poisson.html">metric_poisson</a>()</code> <br><code><a href="metric_sparse_categorical_crossentropy.html">metric_sparse_categorical_crossentropy</a>()</code> <br><code><a href="metric_squared_hinge.html">metric_squared_hinge</a>()</code> <br></p></div>
+    <div class="dont-index"><p>Other losses: <br><code><a href="Loss.html">Loss</a>()</code> <br><code><a href="loss_binary_crossentropy.html">loss_binary_crossentropy</a>()</code> <br><code><a href="loss_binary_focal_crossentropy.html">loss_binary_focal_crossentropy</a>()</code> <br><code><a href="loss_categorical_crossentropy.html">loss_categorical_crossentropy</a>()</code> <br><code><a href="loss_categorical_focal_crossentropy.html">loss_categorical_focal_crossentropy</a>()</code> <br><code><a href="loss_categorical_generalized_cross_entropy.html">loss_categorical_generalized_cross_entropy</a>()</code> <br><code><a href="loss_categorical_hinge.html">loss_categorical_hinge</a>()</code> <br><code><a href="loss_circle.html">loss_circle</a>()</code> <br><code><a href="loss_cosine_similarity.html">loss_cosine_similarity</a>()</code> <br><code><a href="loss_dice.html">loss_dice</a>()</code> <br><code><a href="loss_hinge.html">loss_hinge</a>()</code> <br><code><a href="loss_huber.html">loss_huber</a>()</code> <br><code><a href="loss_kl_divergence.html">loss_kl_divergence</a>()</code> <br><code><a href="loss_log_cosh.html">loss_log_cosh</a>()</code> <br><code><a href="loss_mean_absolute_error.html">loss_mean_absolute_error</a>()</code> <br><code><a href="loss_mean_absolute_percentage_error.html">loss_mean_absolute_percentage_error</a>()</code> <br><code><a href="loss_mean_squared_error.html">loss_mean_squared_error</a>()</code> <br><code><a href="loss_mean_squared_logarithmic_error.html">loss_mean_squared_logarithmic_error</a>()</code> <br><code><a href="loss_poisson.html">loss_poisson</a>()</code> <br><code><a href="loss_sparse_categorical_crossentropy.html">loss_sparse_categorical_crossentropy</a>()</code> <br><code><a href="loss_squared_hinge.html">loss_squared_hinge</a>()</code> <br><code><a href="loss_tversky.html">loss_tversky</a>()</code> <br><code><a href="metric_binary_crossentropy.html">metric_binary_crossentropy</a>()</code> <br><code><a href="metric_binary_focal_crossentropy.html">metric_binary_focal_crossentropy</a>()</code> <br><code><a href="metric_categorical_crossentropy.html">metric_categorical_crossentropy</a>()</code> <br><code><a href="metric_categorical_focal_crossentropy.html">metric_categorical_focal_crossentropy</a>()</code> <br><code><a href="metric_categorical_hinge.html">metric_categorical_hinge</a>()</code> <br><code><a href="metric_hinge.html">metric_hinge</a>()</code> <br><code><a href="metric_huber.html">metric_huber</a>()</code> <br><code><a href="metric_kl_divergence.html">metric_kl_divergence</a>()</code> <br><code><a href="metric_log_cosh.html">metric_log_cosh</a>()</code> <br><code><a href="metric_mean_absolute_error.html">metric_mean_absolute_error</a>()</code> <br><code><a href="metric_mean_absolute_percentage_error.html">metric_mean_absolute_percentage_error</a>()</code> <br><code><a href="metric_mean_squared_error.html">metric_mean_squared_error</a>()</code> <br><code><a href="metric_mean_squared_logarithmic_error.html">metric_mean_squared_logarithmic_error</a>()</code> <br><code><a href="metric_poisson.html">metric_poisson</a>()</code> <br><code><a href="metric_sparse_categorical_crossentropy.html">metric_sparse_categorical_crossentropy</a>()</code> <br><code><a href="metric_squared_hinge.html">metric_squared_hinge</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -130,7 +130,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/loss_dice.html b/docs/dev/reference/loss_dice.html
index 18d2f302e..b9380dfde 100644
--- a/docs/dev/reference/loss_dice.html
+++ b/docs/dev/reference/loss_dice.html
@@ -18,7 +18,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -140,12 +140,14 @@ <h2 id="example">Example<a class="anchor" aria-label="anchor" href="#example"></
 <span>                  <span class="fl">1</span>,   <span class="fl">0</span>, <span class="fl">1</span>, <span class="fl">0.9</span><span class="op">)</span>, dim <span class="op">=</span> <span class="fu"><a href="https://rdrr.io/r/base/c.html" class="external-link">c</a></span><span class="op">(</span><span class="fl">2</span>, <span class="fl">2</span>, <span class="fl">2</span>, <span class="fl">1</span><span class="op">)</span><span class="op">)</span></span>
 <span></span>
 <span><span class="va">axis</span> <span class="op">&lt;-</span> <span class="fu"><a href="https://rdrr.io/r/base/c.html" class="external-link">c</a></span><span class="op">(</span><span class="fl">2</span>, <span class="fl">3</span>, <span class="fl">4</span><span class="op">)</span></span>
-<span><span class="va">loss</span> <span class="op">&lt;-</span> <span class="fu"><a href="../reference/loss_dice.html">loss_dice</a></span><span class="op">(</span><span class="va">y_true</span>, <span class="va">y_pred</span>, axis <span class="op">=</span> <span class="va">axis</span><span class="op">)</span></span>
+<span><span class="va">loss_fn</span> <span class="op">&lt;-</span> <span class="fu"><a href="../reference/loss_dice.html">loss_dice</a></span><span class="op">(</span>axis <span class="op">=</span> <span class="va">axis</span>, reduction <span class="op">=</span> <span class="cn">NULL</span><span class="op">)</span></span>
+<span><span class="va">loss</span> <span class="op">&lt;-</span> <span class="fu">loss_fn</span><span class="op">(</span><span class="va">y_true</span>, <span class="va">y_pred</span><span class="op">)</span></span>
 <span><span class="fu"><a href="https://rdrr.io/r/base/stopifnot.html" class="external-link">stopifnot</a></span><span class="op">(</span><span class="fu"><a href="../reference/shape.html">shape</a></span><span class="op">(</span><span class="va">loss</span><span class="op">)</span> <span class="op">==</span> <span class="fu"><a href="../reference/shape.html">shape</a></span><span class="op">(</span><span class="fl">2</span><span class="op">)</span><span class="op">)</span></span>
 <span><span class="va">loss</span></span></code></pre><p></p></div>
 <p></p><div class="sourceCode"><pre><code><span id="cb1-1"><a href="#cb1-1" tabindex="-1"></a><span class="do">## tf.Tensor([0.5        0.75757575], shape=(2), dtype=float32)</span></span>
 <span id="cb1-2"><a href="#cb1-2" tabindex="-1"></a></span></code></pre><p></p></div>
-<p></p><div class="sourceCode r"><pre><code><span><span class="va">loss</span> <span class="op">=</span> <span class="fu"><a href="../reference/loss_dice.html">loss_dice</a></span><span class="op">(</span><span class="va">y_true</span>, <span class="va">y_pred</span><span class="op">)</span></span>
+<p></p><div class="sourceCode r"><pre><code><span><span class="va">loss_fn</span> <span class="op">&lt;-</span> <span class="fu"><a href="../reference/loss_dice.html">loss_dice</a></span><span class="op">(</span><span class="op">)</span></span>
+<span><span class="va">loss</span> <span class="op">&lt;-</span> <span class="fu">loss_fn</span><span class="op">(</span><span class="va">y_true</span>, <span class="va">y_pred</span><span class="op">)</span></span>
 <span><span class="fu"><a href="https://rdrr.io/r/base/stopifnot.html" class="external-link">stopifnot</a></span><span class="op">(</span><span class="fu"><a href="../reference/shape.html">shape</a></span><span class="op">(</span><span class="va">loss</span><span class="op">)</span> <span class="op">==</span> <span class="fu"><a href="../reference/shape.html">shape</a></span><span class="op">(</span><span class="op">)</span><span class="op">)</span></span>
 <span><span class="va">loss</span></span></code></pre><p></p></div>
 <p></p><div class="sourceCode"><pre><code><span id="cb1-1"><a href="#cb1-1" tabindex="-1"></a><span class="do">## tf.Tensor(0.6164384, shape=(), dtype=float32)</span></span>
@@ -153,7 +155,7 @@ <h2 id="example">Example<a class="anchor" aria-label="anchor" href="#example"></
     </div>
     <div class="section level2">
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
-    <div class="dont-index"><p>Other losses: <br><code><a href="Loss.html">Loss</a>()</code> <br><code><a href="loss_binary_crossentropy.html">loss_binary_crossentropy</a>()</code> <br><code><a href="loss_binary_focal_crossentropy.html">loss_binary_focal_crossentropy</a>()</code> <br><code><a href="loss_categorical_crossentropy.html">loss_categorical_crossentropy</a>()</code> <br><code><a href="loss_categorical_focal_crossentropy.html">loss_categorical_focal_crossentropy</a>()</code> <br><code><a href="loss_categorical_hinge.html">loss_categorical_hinge</a>()</code> <br><code><a href="loss_circle.html">loss_circle</a>()</code> <br><code><a href="loss_cosine_similarity.html">loss_cosine_similarity</a>()</code> <br><code><a href="loss_ctc.html">loss_ctc</a>()</code> <br><code><a href="loss_hinge.html">loss_hinge</a>()</code> <br><code><a href="loss_huber.html">loss_huber</a>()</code> <br><code><a href="loss_kl_divergence.html">loss_kl_divergence</a>()</code> <br><code><a href="loss_log_cosh.html">loss_log_cosh</a>()</code> <br><code><a href="loss_mean_absolute_error.html">loss_mean_absolute_error</a>()</code> <br><code><a href="loss_mean_absolute_percentage_error.html">loss_mean_absolute_percentage_error</a>()</code> <br><code><a href="loss_mean_squared_error.html">loss_mean_squared_error</a>()</code> <br><code><a href="loss_mean_squared_logarithmic_error.html">loss_mean_squared_logarithmic_error</a>()</code> <br><code><a href="loss_poisson.html">loss_poisson</a>()</code> <br><code><a href="loss_sparse_categorical_crossentropy.html">loss_sparse_categorical_crossentropy</a>()</code> <br><code><a href="loss_squared_hinge.html">loss_squared_hinge</a>()</code> <br><code><a href="loss_tversky.html">loss_tversky</a>()</code> <br><code><a href="metric_binary_crossentropy.html">metric_binary_crossentropy</a>()</code> <br><code><a href="metric_binary_focal_crossentropy.html">metric_binary_focal_crossentropy</a>()</code> <br><code><a href="metric_categorical_crossentropy.html">metric_categorical_crossentropy</a>()</code> <br><code><a href="metric_categorical_focal_crossentropy.html">metric_categorical_focal_crossentropy</a>()</code> <br><code><a href="metric_categorical_hinge.html">metric_categorical_hinge</a>()</code> <br><code><a href="metric_hinge.html">metric_hinge</a>()</code> <br><code><a href="metric_huber.html">metric_huber</a>()</code> <br><code><a href="metric_kl_divergence.html">metric_kl_divergence</a>()</code> <br><code><a href="metric_log_cosh.html">metric_log_cosh</a>()</code> <br><code><a href="metric_mean_absolute_error.html">metric_mean_absolute_error</a>()</code> <br><code><a href="metric_mean_absolute_percentage_error.html">metric_mean_absolute_percentage_error</a>()</code> <br><code><a href="metric_mean_squared_error.html">metric_mean_squared_error</a>()</code> <br><code><a href="metric_mean_squared_logarithmic_error.html">metric_mean_squared_logarithmic_error</a>()</code> <br><code><a href="metric_poisson.html">metric_poisson</a>()</code> <br><code><a href="metric_sparse_categorical_crossentropy.html">metric_sparse_categorical_crossentropy</a>()</code> <br><code><a href="metric_squared_hinge.html">metric_squared_hinge</a>()</code> <br></p></div>
+    <div class="dont-index"><p>Other losses: <br><code><a href="Loss.html">Loss</a>()</code> <br><code><a href="loss_binary_crossentropy.html">loss_binary_crossentropy</a>()</code> <br><code><a href="loss_binary_focal_crossentropy.html">loss_binary_focal_crossentropy</a>()</code> <br><code><a href="loss_categorical_crossentropy.html">loss_categorical_crossentropy</a>()</code> <br><code><a href="loss_categorical_focal_crossentropy.html">loss_categorical_focal_crossentropy</a>()</code> <br><code><a href="loss_categorical_generalized_cross_entropy.html">loss_categorical_generalized_cross_entropy</a>()</code> <br><code><a href="loss_categorical_hinge.html">loss_categorical_hinge</a>()</code> <br><code><a href="loss_circle.html">loss_circle</a>()</code> <br><code><a href="loss_cosine_similarity.html">loss_cosine_similarity</a>()</code> <br><code><a href="loss_ctc.html">loss_ctc</a>()</code> <br><code><a href="loss_hinge.html">loss_hinge</a>()</code> <br><code><a href="loss_huber.html">loss_huber</a>()</code> <br><code><a href="loss_kl_divergence.html">loss_kl_divergence</a>()</code> <br><code><a href="loss_log_cosh.html">loss_log_cosh</a>()</code> <br><code><a href="loss_mean_absolute_error.html">loss_mean_absolute_error</a>()</code> <br><code><a href="loss_mean_absolute_percentage_error.html">loss_mean_absolute_percentage_error</a>()</code> <br><code><a href="loss_mean_squared_error.html">loss_mean_squared_error</a>()</code> <br><code><a href="loss_mean_squared_logarithmic_error.html">loss_mean_squared_logarithmic_error</a>()</code> <br><code><a href="loss_poisson.html">loss_poisson</a>()</code> <br><code><a href="loss_sparse_categorical_crossentropy.html">loss_sparse_categorical_crossentropy</a>()</code> <br><code><a href="loss_squared_hinge.html">loss_squared_hinge</a>()</code> <br><code><a href="loss_tversky.html">loss_tversky</a>()</code> <br><code><a href="metric_binary_crossentropy.html">metric_binary_crossentropy</a>()</code> <br><code><a href="metric_binary_focal_crossentropy.html">metric_binary_focal_crossentropy</a>()</code> <br><code><a href="metric_categorical_crossentropy.html">metric_categorical_crossentropy</a>()</code> <br><code><a href="metric_categorical_focal_crossentropy.html">metric_categorical_focal_crossentropy</a>()</code> <br><code><a href="metric_categorical_hinge.html">metric_categorical_hinge</a>()</code> <br><code><a href="metric_hinge.html">metric_hinge</a>()</code> <br><code><a href="metric_huber.html">metric_huber</a>()</code> <br><code><a href="metric_kl_divergence.html">metric_kl_divergence</a>()</code> <br><code><a href="metric_log_cosh.html">metric_log_cosh</a>()</code> <br><code><a href="metric_mean_absolute_error.html">metric_mean_absolute_error</a>()</code> <br><code><a href="metric_mean_absolute_percentage_error.html">metric_mean_absolute_percentage_error</a>()</code> <br><code><a href="metric_mean_squared_error.html">metric_mean_squared_error</a>()</code> <br><code><a href="metric_mean_squared_logarithmic_error.html">metric_mean_squared_logarithmic_error</a>()</code> <br><code><a href="metric_poisson.html">metric_poisson</a>()</code> <br><code><a href="metric_sparse_categorical_crossentropy.html">metric_sparse_categorical_crossentropy</a>()</code> <br><code><a href="metric_squared_hinge.html">metric_squared_hinge</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -165,7 +167,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/loss_hinge.html b/docs/dev/reference/loss_hinge.html
index b887b0aa1..770a99a6b 100644
--- a/docs/dev/reference/loss_hinge.html
+++ b/docs/dev/reference/loss_hinge.html
@@ -16,7 +16,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -138,7 +138,7 @@ <h2 id="examples">Examples<a class="anchor" aria-label="anchor" href="#examples"
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
     <div class="dont-index">
 <ul><li><p><a href="https://keras.io/api/losses/hinge_losses#hinge-class" class="external-link">https://keras.io/api/losses/hinge_losses#hinge-class</a></p></li>
-</ul><p>Other losses: <br><code><a href="Loss.html">Loss</a>()</code> <br><code><a href="loss_binary_crossentropy.html">loss_binary_crossentropy</a>()</code> <br><code><a href="loss_binary_focal_crossentropy.html">loss_binary_focal_crossentropy</a>()</code> <br><code><a href="loss_categorical_crossentropy.html">loss_categorical_crossentropy</a>()</code> <br><code><a href="loss_categorical_focal_crossentropy.html">loss_categorical_focal_crossentropy</a>()</code> <br><code><a href="loss_categorical_hinge.html">loss_categorical_hinge</a>()</code> <br><code><a href="loss_circle.html">loss_circle</a>()</code> <br><code><a href="loss_cosine_similarity.html">loss_cosine_similarity</a>()</code> <br><code><a href="loss_ctc.html">loss_ctc</a>()</code> <br><code><a href="loss_dice.html">loss_dice</a>()</code> <br><code><a href="loss_huber.html">loss_huber</a>()</code> <br><code><a href="loss_kl_divergence.html">loss_kl_divergence</a>()</code> <br><code><a href="loss_log_cosh.html">loss_log_cosh</a>()</code> <br><code><a href="loss_mean_absolute_error.html">loss_mean_absolute_error</a>()</code> <br><code><a href="loss_mean_absolute_percentage_error.html">loss_mean_absolute_percentage_error</a>()</code> <br><code><a href="loss_mean_squared_error.html">loss_mean_squared_error</a>()</code> <br><code><a href="loss_mean_squared_logarithmic_error.html">loss_mean_squared_logarithmic_error</a>()</code> <br><code><a href="loss_poisson.html">loss_poisson</a>()</code> <br><code><a href="loss_sparse_categorical_crossentropy.html">loss_sparse_categorical_crossentropy</a>()</code> <br><code><a href="loss_squared_hinge.html">loss_squared_hinge</a>()</code> <br><code><a href="loss_tversky.html">loss_tversky</a>()</code> <br><code><a href="metric_binary_crossentropy.html">metric_binary_crossentropy</a>()</code> <br><code><a href="metric_binary_focal_crossentropy.html">metric_binary_focal_crossentropy</a>()</code> <br><code><a href="metric_categorical_crossentropy.html">metric_categorical_crossentropy</a>()</code> <br><code><a href="metric_categorical_focal_crossentropy.html">metric_categorical_focal_crossentropy</a>()</code> <br><code><a href="metric_categorical_hinge.html">metric_categorical_hinge</a>()</code> <br><code><a href="metric_hinge.html">metric_hinge</a>()</code> <br><code><a href="metric_huber.html">metric_huber</a>()</code> <br><code><a href="metric_kl_divergence.html">metric_kl_divergence</a>()</code> <br><code><a href="metric_log_cosh.html">metric_log_cosh</a>()</code> <br><code><a href="metric_mean_absolute_error.html">metric_mean_absolute_error</a>()</code> <br><code><a href="metric_mean_absolute_percentage_error.html">metric_mean_absolute_percentage_error</a>()</code> <br><code><a href="metric_mean_squared_error.html">metric_mean_squared_error</a>()</code> <br><code><a href="metric_mean_squared_logarithmic_error.html">metric_mean_squared_logarithmic_error</a>()</code> <br><code><a href="metric_poisson.html">metric_poisson</a>()</code> <br><code><a href="metric_sparse_categorical_crossentropy.html">metric_sparse_categorical_crossentropy</a>()</code> <br><code><a href="metric_squared_hinge.html">metric_squared_hinge</a>()</code> <br></p></div>
+</ul><p>Other losses: <br><code><a href="Loss.html">Loss</a>()</code> <br><code><a href="loss_binary_crossentropy.html">loss_binary_crossentropy</a>()</code> <br><code><a href="loss_binary_focal_crossentropy.html">loss_binary_focal_crossentropy</a>()</code> <br><code><a href="loss_categorical_crossentropy.html">loss_categorical_crossentropy</a>()</code> <br><code><a href="loss_categorical_focal_crossentropy.html">loss_categorical_focal_crossentropy</a>()</code> <br><code><a href="loss_categorical_generalized_cross_entropy.html">loss_categorical_generalized_cross_entropy</a>()</code> <br><code><a href="loss_categorical_hinge.html">loss_categorical_hinge</a>()</code> <br><code><a href="loss_circle.html">loss_circle</a>()</code> <br><code><a href="loss_cosine_similarity.html">loss_cosine_similarity</a>()</code> <br><code><a href="loss_ctc.html">loss_ctc</a>()</code> <br><code><a href="loss_dice.html">loss_dice</a>()</code> <br><code><a href="loss_huber.html">loss_huber</a>()</code> <br><code><a href="loss_kl_divergence.html">loss_kl_divergence</a>()</code> <br><code><a href="loss_log_cosh.html">loss_log_cosh</a>()</code> <br><code><a href="loss_mean_absolute_error.html">loss_mean_absolute_error</a>()</code> <br><code><a href="loss_mean_absolute_percentage_error.html">loss_mean_absolute_percentage_error</a>()</code> <br><code><a href="loss_mean_squared_error.html">loss_mean_squared_error</a>()</code> <br><code><a href="loss_mean_squared_logarithmic_error.html">loss_mean_squared_logarithmic_error</a>()</code> <br><code><a href="loss_poisson.html">loss_poisson</a>()</code> <br><code><a href="loss_sparse_categorical_crossentropy.html">loss_sparse_categorical_crossentropy</a>()</code> <br><code><a href="loss_squared_hinge.html">loss_squared_hinge</a>()</code> <br><code><a href="loss_tversky.html">loss_tversky</a>()</code> <br><code><a href="metric_binary_crossentropy.html">metric_binary_crossentropy</a>()</code> <br><code><a href="metric_binary_focal_crossentropy.html">metric_binary_focal_crossentropy</a>()</code> <br><code><a href="metric_categorical_crossentropy.html">metric_categorical_crossentropy</a>()</code> <br><code><a href="metric_categorical_focal_crossentropy.html">metric_categorical_focal_crossentropy</a>()</code> <br><code><a href="metric_categorical_hinge.html">metric_categorical_hinge</a>()</code> <br><code><a href="metric_hinge.html">metric_hinge</a>()</code> <br><code><a href="metric_huber.html">metric_huber</a>()</code> <br><code><a href="metric_kl_divergence.html">metric_kl_divergence</a>()</code> <br><code><a href="metric_log_cosh.html">metric_log_cosh</a>()</code> <br><code><a href="metric_mean_absolute_error.html">metric_mean_absolute_error</a>()</code> <br><code><a href="metric_mean_absolute_percentage_error.html">metric_mean_absolute_percentage_error</a>()</code> <br><code><a href="metric_mean_squared_error.html">metric_mean_squared_error</a>()</code> <br><code><a href="metric_mean_squared_logarithmic_error.html">metric_mean_squared_logarithmic_error</a>()</code> <br><code><a href="metric_poisson.html">metric_poisson</a>()</code> <br><code><a href="metric_sparse_categorical_crossentropy.html">metric_sparse_categorical_crossentropy</a>()</code> <br><code><a href="metric_squared_hinge.html">metric_squared_hinge</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -150,7 +150,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/loss_huber.html b/docs/dev/reference/loss_huber.html
index 1f0cb4e18..b4d3db082 100644
--- a/docs/dev/reference/loss_huber.html
+++ b/docs/dev/reference/loss_huber.html
@@ -28,7 +28,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -152,7 +152,7 @@ <h2 id="examples">Examples<a class="anchor" aria-label="anchor" href="#examples"
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
     <div class="dont-index">
 <ul><li><p><a href="https://keras.io/api/losses/regression_losses#huber-class" class="external-link">https://keras.io/api/losses/regression_losses#huber-class</a></p></li>
-</ul><p>Other losses: <br><code><a href="Loss.html">Loss</a>()</code> <br><code><a href="loss_binary_crossentropy.html">loss_binary_crossentropy</a>()</code> <br><code><a href="loss_binary_focal_crossentropy.html">loss_binary_focal_crossentropy</a>()</code> <br><code><a href="loss_categorical_crossentropy.html">loss_categorical_crossentropy</a>()</code> <br><code><a href="loss_categorical_focal_crossentropy.html">loss_categorical_focal_crossentropy</a>()</code> <br><code><a href="loss_categorical_hinge.html">loss_categorical_hinge</a>()</code> <br><code><a href="loss_circle.html">loss_circle</a>()</code> <br><code><a href="loss_cosine_similarity.html">loss_cosine_similarity</a>()</code> <br><code><a href="loss_ctc.html">loss_ctc</a>()</code> <br><code><a href="loss_dice.html">loss_dice</a>()</code> <br><code><a href="loss_hinge.html">loss_hinge</a>()</code> <br><code><a href="loss_kl_divergence.html">loss_kl_divergence</a>()</code> <br><code><a href="loss_log_cosh.html">loss_log_cosh</a>()</code> <br><code><a href="loss_mean_absolute_error.html">loss_mean_absolute_error</a>()</code> <br><code><a href="loss_mean_absolute_percentage_error.html">loss_mean_absolute_percentage_error</a>()</code> <br><code><a href="loss_mean_squared_error.html">loss_mean_squared_error</a>()</code> <br><code><a href="loss_mean_squared_logarithmic_error.html">loss_mean_squared_logarithmic_error</a>()</code> <br><code><a href="loss_poisson.html">loss_poisson</a>()</code> <br><code><a href="loss_sparse_categorical_crossentropy.html">loss_sparse_categorical_crossentropy</a>()</code> <br><code><a href="loss_squared_hinge.html">loss_squared_hinge</a>()</code> <br><code><a href="loss_tversky.html">loss_tversky</a>()</code> <br><code><a href="metric_binary_crossentropy.html">metric_binary_crossentropy</a>()</code> <br><code><a href="metric_binary_focal_crossentropy.html">metric_binary_focal_crossentropy</a>()</code> <br><code><a href="metric_categorical_crossentropy.html">metric_categorical_crossentropy</a>()</code> <br><code><a href="metric_categorical_focal_crossentropy.html">metric_categorical_focal_crossentropy</a>()</code> <br><code><a href="metric_categorical_hinge.html">metric_categorical_hinge</a>()</code> <br><code><a href="metric_hinge.html">metric_hinge</a>()</code> <br><code><a href="metric_huber.html">metric_huber</a>()</code> <br><code><a href="metric_kl_divergence.html">metric_kl_divergence</a>()</code> <br><code><a href="metric_log_cosh.html">metric_log_cosh</a>()</code> <br><code><a href="metric_mean_absolute_error.html">metric_mean_absolute_error</a>()</code> <br><code><a href="metric_mean_absolute_percentage_error.html">metric_mean_absolute_percentage_error</a>()</code> <br><code><a href="metric_mean_squared_error.html">metric_mean_squared_error</a>()</code> <br><code><a href="metric_mean_squared_logarithmic_error.html">metric_mean_squared_logarithmic_error</a>()</code> <br><code><a href="metric_poisson.html">metric_poisson</a>()</code> <br><code><a href="metric_sparse_categorical_crossentropy.html">metric_sparse_categorical_crossentropy</a>()</code> <br><code><a href="metric_squared_hinge.html">metric_squared_hinge</a>()</code> <br></p></div>
+</ul><p>Other losses: <br><code><a href="Loss.html">Loss</a>()</code> <br><code><a href="loss_binary_crossentropy.html">loss_binary_crossentropy</a>()</code> <br><code><a href="loss_binary_focal_crossentropy.html">loss_binary_focal_crossentropy</a>()</code> <br><code><a href="loss_categorical_crossentropy.html">loss_categorical_crossentropy</a>()</code> <br><code><a href="loss_categorical_focal_crossentropy.html">loss_categorical_focal_crossentropy</a>()</code> <br><code><a href="loss_categorical_generalized_cross_entropy.html">loss_categorical_generalized_cross_entropy</a>()</code> <br><code><a href="loss_categorical_hinge.html">loss_categorical_hinge</a>()</code> <br><code><a href="loss_circle.html">loss_circle</a>()</code> <br><code><a href="loss_cosine_similarity.html">loss_cosine_similarity</a>()</code> <br><code><a href="loss_ctc.html">loss_ctc</a>()</code> <br><code><a href="loss_dice.html">loss_dice</a>()</code> <br><code><a href="loss_hinge.html">loss_hinge</a>()</code> <br><code><a href="loss_kl_divergence.html">loss_kl_divergence</a>()</code> <br><code><a href="loss_log_cosh.html">loss_log_cosh</a>()</code> <br><code><a href="loss_mean_absolute_error.html">loss_mean_absolute_error</a>()</code> <br><code><a href="loss_mean_absolute_percentage_error.html">loss_mean_absolute_percentage_error</a>()</code> <br><code><a href="loss_mean_squared_error.html">loss_mean_squared_error</a>()</code> <br><code><a href="loss_mean_squared_logarithmic_error.html">loss_mean_squared_logarithmic_error</a>()</code> <br><code><a href="loss_poisson.html">loss_poisson</a>()</code> <br><code><a href="loss_sparse_categorical_crossentropy.html">loss_sparse_categorical_crossentropy</a>()</code> <br><code><a href="loss_squared_hinge.html">loss_squared_hinge</a>()</code> <br><code><a href="loss_tversky.html">loss_tversky</a>()</code> <br><code><a href="metric_binary_crossentropy.html">metric_binary_crossentropy</a>()</code> <br><code><a href="metric_binary_focal_crossentropy.html">metric_binary_focal_crossentropy</a>()</code> <br><code><a href="metric_categorical_crossentropy.html">metric_categorical_crossentropy</a>()</code> <br><code><a href="metric_categorical_focal_crossentropy.html">metric_categorical_focal_crossentropy</a>()</code> <br><code><a href="metric_categorical_hinge.html">metric_categorical_hinge</a>()</code> <br><code><a href="metric_hinge.html">metric_hinge</a>()</code> <br><code><a href="metric_huber.html">metric_huber</a>()</code> <br><code><a href="metric_kl_divergence.html">metric_kl_divergence</a>()</code> <br><code><a href="metric_log_cosh.html">metric_log_cosh</a>()</code> <br><code><a href="metric_mean_absolute_error.html">metric_mean_absolute_error</a>()</code> <br><code><a href="metric_mean_absolute_percentage_error.html">metric_mean_absolute_percentage_error</a>()</code> <br><code><a href="metric_mean_squared_error.html">metric_mean_squared_error</a>()</code> <br><code><a href="metric_mean_squared_logarithmic_error.html">metric_mean_squared_logarithmic_error</a>()</code> <br><code><a href="metric_poisson.html">metric_poisson</a>()</code> <br><code><a href="metric_sparse_categorical_crossentropy.html">metric_sparse_categorical_crossentropy</a>()</code> <br><code><a href="metric_squared_hinge.html">metric_squared_hinge</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -164,7 +164,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/loss_kl_divergence.html b/docs/dev/reference/loss_kl_divergence.html
index f24ae7727..1fa6b5063 100644
--- a/docs/dev/reference/loss_kl_divergence.html
+++ b/docs/dev/reference/loss_kl_divergence.html
@@ -18,7 +18,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -139,7 +139,7 @@ <h2 id="examples">Examples<a class="anchor" aria-label="anchor" href="#examples"
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
     <div class="dont-index">
 <ul><li><p><a href="https://keras.io/api/losses/probabilistic_losses#kldivergence-class" class="external-link">https://keras.io/api/losses/probabilistic_losses#kldivergence-class</a></p></li>
-</ul><p>Other losses: <br><code><a href="Loss.html">Loss</a>()</code> <br><code><a href="loss_binary_crossentropy.html">loss_binary_crossentropy</a>()</code> <br><code><a href="loss_binary_focal_crossentropy.html">loss_binary_focal_crossentropy</a>()</code> <br><code><a href="loss_categorical_crossentropy.html">loss_categorical_crossentropy</a>()</code> <br><code><a href="loss_categorical_focal_crossentropy.html">loss_categorical_focal_crossentropy</a>()</code> <br><code><a href="loss_categorical_hinge.html">loss_categorical_hinge</a>()</code> <br><code><a href="loss_circle.html">loss_circle</a>()</code> <br><code><a href="loss_cosine_similarity.html">loss_cosine_similarity</a>()</code> <br><code><a href="loss_ctc.html">loss_ctc</a>()</code> <br><code><a href="loss_dice.html">loss_dice</a>()</code> <br><code><a href="loss_hinge.html">loss_hinge</a>()</code> <br><code><a href="loss_huber.html">loss_huber</a>()</code> <br><code><a href="loss_log_cosh.html">loss_log_cosh</a>()</code> <br><code><a href="loss_mean_absolute_error.html">loss_mean_absolute_error</a>()</code> <br><code><a href="loss_mean_absolute_percentage_error.html">loss_mean_absolute_percentage_error</a>()</code> <br><code><a href="loss_mean_squared_error.html">loss_mean_squared_error</a>()</code> <br><code><a href="loss_mean_squared_logarithmic_error.html">loss_mean_squared_logarithmic_error</a>()</code> <br><code><a href="loss_poisson.html">loss_poisson</a>()</code> <br><code><a href="loss_sparse_categorical_crossentropy.html">loss_sparse_categorical_crossentropy</a>()</code> <br><code><a href="loss_squared_hinge.html">loss_squared_hinge</a>()</code> <br><code><a href="loss_tversky.html">loss_tversky</a>()</code> <br><code><a href="metric_binary_crossentropy.html">metric_binary_crossentropy</a>()</code> <br><code><a href="metric_binary_focal_crossentropy.html">metric_binary_focal_crossentropy</a>()</code> <br><code><a href="metric_categorical_crossentropy.html">metric_categorical_crossentropy</a>()</code> <br><code><a href="metric_categorical_focal_crossentropy.html">metric_categorical_focal_crossentropy</a>()</code> <br><code><a href="metric_categorical_hinge.html">metric_categorical_hinge</a>()</code> <br><code><a href="metric_hinge.html">metric_hinge</a>()</code> <br><code><a href="metric_huber.html">metric_huber</a>()</code> <br><code><a href="metric_kl_divergence.html">metric_kl_divergence</a>()</code> <br><code><a href="metric_log_cosh.html">metric_log_cosh</a>()</code> <br><code><a href="metric_mean_absolute_error.html">metric_mean_absolute_error</a>()</code> <br><code><a href="metric_mean_absolute_percentage_error.html">metric_mean_absolute_percentage_error</a>()</code> <br><code><a href="metric_mean_squared_error.html">metric_mean_squared_error</a>()</code> <br><code><a href="metric_mean_squared_logarithmic_error.html">metric_mean_squared_logarithmic_error</a>()</code> <br><code><a href="metric_poisson.html">metric_poisson</a>()</code> <br><code><a href="metric_sparse_categorical_crossentropy.html">metric_sparse_categorical_crossentropy</a>()</code> <br><code><a href="metric_squared_hinge.html">metric_squared_hinge</a>()</code> <br></p></div>
+</ul><p>Other losses: <br><code><a href="Loss.html">Loss</a>()</code> <br><code><a href="loss_binary_crossentropy.html">loss_binary_crossentropy</a>()</code> <br><code><a href="loss_binary_focal_crossentropy.html">loss_binary_focal_crossentropy</a>()</code> <br><code><a href="loss_categorical_crossentropy.html">loss_categorical_crossentropy</a>()</code> <br><code><a href="loss_categorical_focal_crossentropy.html">loss_categorical_focal_crossentropy</a>()</code> <br><code><a href="loss_categorical_generalized_cross_entropy.html">loss_categorical_generalized_cross_entropy</a>()</code> <br><code><a href="loss_categorical_hinge.html">loss_categorical_hinge</a>()</code> <br><code><a href="loss_circle.html">loss_circle</a>()</code> <br><code><a href="loss_cosine_similarity.html">loss_cosine_similarity</a>()</code> <br><code><a href="loss_ctc.html">loss_ctc</a>()</code> <br><code><a href="loss_dice.html">loss_dice</a>()</code> <br><code><a href="loss_hinge.html">loss_hinge</a>()</code> <br><code><a href="loss_huber.html">loss_huber</a>()</code> <br><code><a href="loss_log_cosh.html">loss_log_cosh</a>()</code> <br><code><a href="loss_mean_absolute_error.html">loss_mean_absolute_error</a>()</code> <br><code><a href="loss_mean_absolute_percentage_error.html">loss_mean_absolute_percentage_error</a>()</code> <br><code><a href="loss_mean_squared_error.html">loss_mean_squared_error</a>()</code> <br><code><a href="loss_mean_squared_logarithmic_error.html">loss_mean_squared_logarithmic_error</a>()</code> <br><code><a href="loss_poisson.html">loss_poisson</a>()</code> <br><code><a href="loss_sparse_categorical_crossentropy.html">loss_sparse_categorical_crossentropy</a>()</code> <br><code><a href="loss_squared_hinge.html">loss_squared_hinge</a>()</code> <br><code><a href="loss_tversky.html">loss_tversky</a>()</code> <br><code><a href="metric_binary_crossentropy.html">metric_binary_crossentropy</a>()</code> <br><code><a href="metric_binary_focal_crossentropy.html">metric_binary_focal_crossentropy</a>()</code> <br><code><a href="metric_categorical_crossentropy.html">metric_categorical_crossentropy</a>()</code> <br><code><a href="metric_categorical_focal_crossentropy.html">metric_categorical_focal_crossentropy</a>()</code> <br><code><a href="metric_categorical_hinge.html">metric_categorical_hinge</a>()</code> <br><code><a href="metric_hinge.html">metric_hinge</a>()</code> <br><code><a href="metric_huber.html">metric_huber</a>()</code> <br><code><a href="metric_kl_divergence.html">metric_kl_divergence</a>()</code> <br><code><a href="metric_log_cosh.html">metric_log_cosh</a>()</code> <br><code><a href="metric_mean_absolute_error.html">metric_mean_absolute_error</a>()</code> <br><code><a href="metric_mean_absolute_percentage_error.html">metric_mean_absolute_percentage_error</a>()</code> <br><code><a href="metric_mean_squared_error.html">metric_mean_squared_error</a>()</code> <br><code><a href="metric_mean_squared_logarithmic_error.html">metric_mean_squared_logarithmic_error</a>()</code> <br><code><a href="metric_poisson.html">metric_poisson</a>()</code> <br><code><a href="metric_sparse_categorical_crossentropy.html">metric_sparse_categorical_crossentropy</a>()</code> <br><code><a href="metric_squared_hinge.html">metric_squared_hinge</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -151,7 +151,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/loss_log_cosh.html b/docs/dev/reference/loss_log_cosh.html
index 89388ba77..bd4e88ffd 100644
--- a/docs/dev/reference/loss_log_cosh.html
+++ b/docs/dev/reference/loss_log_cosh.html
@@ -20,7 +20,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -135,7 +135,7 @@ <h2 id="examples">Examples<a class="anchor" aria-label="anchor" href="#examples"
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
     <div class="dont-index">
 <ul><li><p><a href="https://keras.io/api/losses/regression_losses#logcosh-class" class="external-link">https://keras.io/api/losses/regression_losses#logcosh-class</a></p></li>
-</ul><p>Other losses: <br><code><a href="Loss.html">Loss</a>()</code> <br><code><a href="loss_binary_crossentropy.html">loss_binary_crossentropy</a>()</code> <br><code><a href="loss_binary_focal_crossentropy.html">loss_binary_focal_crossentropy</a>()</code> <br><code><a href="loss_categorical_crossentropy.html">loss_categorical_crossentropy</a>()</code> <br><code><a href="loss_categorical_focal_crossentropy.html">loss_categorical_focal_crossentropy</a>()</code> <br><code><a href="loss_categorical_hinge.html">loss_categorical_hinge</a>()</code> <br><code><a href="loss_circle.html">loss_circle</a>()</code> <br><code><a href="loss_cosine_similarity.html">loss_cosine_similarity</a>()</code> <br><code><a href="loss_ctc.html">loss_ctc</a>()</code> <br><code><a href="loss_dice.html">loss_dice</a>()</code> <br><code><a href="loss_hinge.html">loss_hinge</a>()</code> <br><code><a href="loss_huber.html">loss_huber</a>()</code> <br><code><a href="loss_kl_divergence.html">loss_kl_divergence</a>()</code> <br><code><a href="loss_mean_absolute_error.html">loss_mean_absolute_error</a>()</code> <br><code><a href="loss_mean_absolute_percentage_error.html">loss_mean_absolute_percentage_error</a>()</code> <br><code><a href="loss_mean_squared_error.html">loss_mean_squared_error</a>()</code> <br><code><a href="loss_mean_squared_logarithmic_error.html">loss_mean_squared_logarithmic_error</a>()</code> <br><code><a href="loss_poisson.html">loss_poisson</a>()</code> <br><code><a href="loss_sparse_categorical_crossentropy.html">loss_sparse_categorical_crossentropy</a>()</code> <br><code><a href="loss_squared_hinge.html">loss_squared_hinge</a>()</code> <br><code><a href="loss_tversky.html">loss_tversky</a>()</code> <br><code><a href="metric_binary_crossentropy.html">metric_binary_crossentropy</a>()</code> <br><code><a href="metric_binary_focal_crossentropy.html">metric_binary_focal_crossentropy</a>()</code> <br><code><a href="metric_categorical_crossentropy.html">metric_categorical_crossentropy</a>()</code> <br><code><a href="metric_categorical_focal_crossentropy.html">metric_categorical_focal_crossentropy</a>()</code> <br><code><a href="metric_categorical_hinge.html">metric_categorical_hinge</a>()</code> <br><code><a href="metric_hinge.html">metric_hinge</a>()</code> <br><code><a href="metric_huber.html">metric_huber</a>()</code> <br><code><a href="metric_kl_divergence.html">metric_kl_divergence</a>()</code> <br><code><a href="metric_log_cosh.html">metric_log_cosh</a>()</code> <br><code><a href="metric_mean_absolute_error.html">metric_mean_absolute_error</a>()</code> <br><code><a href="metric_mean_absolute_percentage_error.html">metric_mean_absolute_percentage_error</a>()</code> <br><code><a href="metric_mean_squared_error.html">metric_mean_squared_error</a>()</code> <br><code><a href="metric_mean_squared_logarithmic_error.html">metric_mean_squared_logarithmic_error</a>()</code> <br><code><a href="metric_poisson.html">metric_poisson</a>()</code> <br><code><a href="metric_sparse_categorical_crossentropy.html">metric_sparse_categorical_crossentropy</a>()</code> <br><code><a href="metric_squared_hinge.html">metric_squared_hinge</a>()</code> <br></p></div>
+</ul><p>Other losses: <br><code><a href="Loss.html">Loss</a>()</code> <br><code><a href="loss_binary_crossentropy.html">loss_binary_crossentropy</a>()</code> <br><code><a href="loss_binary_focal_crossentropy.html">loss_binary_focal_crossentropy</a>()</code> <br><code><a href="loss_categorical_crossentropy.html">loss_categorical_crossentropy</a>()</code> <br><code><a href="loss_categorical_focal_crossentropy.html">loss_categorical_focal_crossentropy</a>()</code> <br><code><a href="loss_categorical_generalized_cross_entropy.html">loss_categorical_generalized_cross_entropy</a>()</code> <br><code><a href="loss_categorical_hinge.html">loss_categorical_hinge</a>()</code> <br><code><a href="loss_circle.html">loss_circle</a>()</code> <br><code><a href="loss_cosine_similarity.html">loss_cosine_similarity</a>()</code> <br><code><a href="loss_ctc.html">loss_ctc</a>()</code> <br><code><a href="loss_dice.html">loss_dice</a>()</code> <br><code><a href="loss_hinge.html">loss_hinge</a>()</code> <br><code><a href="loss_huber.html">loss_huber</a>()</code> <br><code><a href="loss_kl_divergence.html">loss_kl_divergence</a>()</code> <br><code><a href="loss_mean_absolute_error.html">loss_mean_absolute_error</a>()</code> <br><code><a href="loss_mean_absolute_percentage_error.html">loss_mean_absolute_percentage_error</a>()</code> <br><code><a href="loss_mean_squared_error.html">loss_mean_squared_error</a>()</code> <br><code><a href="loss_mean_squared_logarithmic_error.html">loss_mean_squared_logarithmic_error</a>()</code> <br><code><a href="loss_poisson.html">loss_poisson</a>()</code> <br><code><a href="loss_sparse_categorical_crossentropy.html">loss_sparse_categorical_crossentropy</a>()</code> <br><code><a href="loss_squared_hinge.html">loss_squared_hinge</a>()</code> <br><code><a href="loss_tversky.html">loss_tversky</a>()</code> <br><code><a href="metric_binary_crossentropy.html">metric_binary_crossentropy</a>()</code> <br><code><a href="metric_binary_focal_crossentropy.html">metric_binary_focal_crossentropy</a>()</code> <br><code><a href="metric_categorical_crossentropy.html">metric_categorical_crossentropy</a>()</code> <br><code><a href="metric_categorical_focal_crossentropy.html">metric_categorical_focal_crossentropy</a>()</code> <br><code><a href="metric_categorical_hinge.html">metric_categorical_hinge</a>()</code> <br><code><a href="metric_hinge.html">metric_hinge</a>()</code> <br><code><a href="metric_huber.html">metric_huber</a>()</code> <br><code><a href="metric_kl_divergence.html">metric_kl_divergence</a>()</code> <br><code><a href="metric_log_cosh.html">metric_log_cosh</a>()</code> <br><code><a href="metric_mean_absolute_error.html">metric_mean_absolute_error</a>()</code> <br><code><a href="metric_mean_absolute_percentage_error.html">metric_mean_absolute_percentage_error</a>()</code> <br><code><a href="metric_mean_squared_error.html">metric_mean_squared_error</a>()</code> <br><code><a href="metric_mean_squared_logarithmic_error.html">metric_mean_squared_logarithmic_error</a>()</code> <br><code><a href="metric_poisson.html">metric_poisson</a>()</code> <br><code><a href="metric_sparse_categorical_crossentropy.html">metric_sparse_categorical_crossentropy</a>()</code> <br><code><a href="metric_squared_hinge.html">metric_squared_hinge</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -147,7 +147,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/loss_mean_absolute_error.html b/docs/dev/reference/loss_mean_absolute_error.html
index 3980b68bc..590d35af9 100644
--- a/docs/dev/reference/loss_mean_absolute_error.html
+++ b/docs/dev/reference/loss_mean_absolute_error.html
@@ -12,7 +12,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -127,7 +127,7 @@ <h2 id="examples">Examples<a class="anchor" aria-label="anchor" href="#examples"
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
     <div class="dont-index">
 <ul><li><p><a href="https://keras.io/api/losses/regression_losses#meanabsoluteerror-class" class="external-link">https://keras.io/api/losses/regression_losses#meanabsoluteerror-class</a></p></li>
-</ul><p>Other losses: <br><code><a href="Loss.html">Loss</a>()</code> <br><code><a href="loss_binary_crossentropy.html">loss_binary_crossentropy</a>()</code> <br><code><a href="loss_binary_focal_crossentropy.html">loss_binary_focal_crossentropy</a>()</code> <br><code><a href="loss_categorical_crossentropy.html">loss_categorical_crossentropy</a>()</code> <br><code><a href="loss_categorical_focal_crossentropy.html">loss_categorical_focal_crossentropy</a>()</code> <br><code><a href="loss_categorical_hinge.html">loss_categorical_hinge</a>()</code> <br><code><a href="loss_circle.html">loss_circle</a>()</code> <br><code><a href="loss_cosine_similarity.html">loss_cosine_similarity</a>()</code> <br><code><a href="loss_ctc.html">loss_ctc</a>()</code> <br><code><a href="loss_dice.html">loss_dice</a>()</code> <br><code><a href="loss_hinge.html">loss_hinge</a>()</code> <br><code><a href="loss_huber.html">loss_huber</a>()</code> <br><code><a href="loss_kl_divergence.html">loss_kl_divergence</a>()</code> <br><code><a href="loss_log_cosh.html">loss_log_cosh</a>()</code> <br><code><a href="loss_mean_absolute_percentage_error.html">loss_mean_absolute_percentage_error</a>()</code> <br><code><a href="loss_mean_squared_error.html">loss_mean_squared_error</a>()</code> <br><code><a href="loss_mean_squared_logarithmic_error.html">loss_mean_squared_logarithmic_error</a>()</code> <br><code><a href="loss_poisson.html">loss_poisson</a>()</code> <br><code><a href="loss_sparse_categorical_crossentropy.html">loss_sparse_categorical_crossentropy</a>()</code> <br><code><a href="loss_squared_hinge.html">loss_squared_hinge</a>()</code> <br><code><a href="loss_tversky.html">loss_tversky</a>()</code> <br><code><a href="metric_binary_crossentropy.html">metric_binary_crossentropy</a>()</code> <br><code><a href="metric_binary_focal_crossentropy.html">metric_binary_focal_crossentropy</a>()</code> <br><code><a href="metric_categorical_crossentropy.html">metric_categorical_crossentropy</a>()</code> <br><code><a href="metric_categorical_focal_crossentropy.html">metric_categorical_focal_crossentropy</a>()</code> <br><code><a href="metric_categorical_hinge.html">metric_categorical_hinge</a>()</code> <br><code><a href="metric_hinge.html">metric_hinge</a>()</code> <br><code><a href="metric_huber.html">metric_huber</a>()</code> <br><code><a href="metric_kl_divergence.html">metric_kl_divergence</a>()</code> <br><code><a href="metric_log_cosh.html">metric_log_cosh</a>()</code> <br><code><a href="metric_mean_absolute_error.html">metric_mean_absolute_error</a>()</code> <br><code><a href="metric_mean_absolute_percentage_error.html">metric_mean_absolute_percentage_error</a>()</code> <br><code><a href="metric_mean_squared_error.html">metric_mean_squared_error</a>()</code> <br><code><a href="metric_mean_squared_logarithmic_error.html">metric_mean_squared_logarithmic_error</a>()</code> <br><code><a href="metric_poisson.html">metric_poisson</a>()</code> <br><code><a href="metric_sparse_categorical_crossentropy.html">metric_sparse_categorical_crossentropy</a>()</code> <br><code><a href="metric_squared_hinge.html">metric_squared_hinge</a>()</code> <br></p></div>
+</ul><p>Other losses: <br><code><a href="Loss.html">Loss</a>()</code> <br><code><a href="loss_binary_crossentropy.html">loss_binary_crossentropy</a>()</code> <br><code><a href="loss_binary_focal_crossentropy.html">loss_binary_focal_crossentropy</a>()</code> <br><code><a href="loss_categorical_crossentropy.html">loss_categorical_crossentropy</a>()</code> <br><code><a href="loss_categorical_focal_crossentropy.html">loss_categorical_focal_crossentropy</a>()</code> <br><code><a href="loss_categorical_generalized_cross_entropy.html">loss_categorical_generalized_cross_entropy</a>()</code> <br><code><a href="loss_categorical_hinge.html">loss_categorical_hinge</a>()</code> <br><code><a href="loss_circle.html">loss_circle</a>()</code> <br><code><a href="loss_cosine_similarity.html">loss_cosine_similarity</a>()</code> <br><code><a href="loss_ctc.html">loss_ctc</a>()</code> <br><code><a href="loss_dice.html">loss_dice</a>()</code> <br><code><a href="loss_hinge.html">loss_hinge</a>()</code> <br><code><a href="loss_huber.html">loss_huber</a>()</code> <br><code><a href="loss_kl_divergence.html">loss_kl_divergence</a>()</code> <br><code><a href="loss_log_cosh.html">loss_log_cosh</a>()</code> <br><code><a href="loss_mean_absolute_percentage_error.html">loss_mean_absolute_percentage_error</a>()</code> <br><code><a href="loss_mean_squared_error.html">loss_mean_squared_error</a>()</code> <br><code><a href="loss_mean_squared_logarithmic_error.html">loss_mean_squared_logarithmic_error</a>()</code> <br><code><a href="loss_poisson.html">loss_poisson</a>()</code> <br><code><a href="loss_sparse_categorical_crossentropy.html">loss_sparse_categorical_crossentropy</a>()</code> <br><code><a href="loss_squared_hinge.html">loss_squared_hinge</a>()</code> <br><code><a href="loss_tversky.html">loss_tversky</a>()</code> <br><code><a href="metric_binary_crossentropy.html">metric_binary_crossentropy</a>()</code> <br><code><a href="metric_binary_focal_crossentropy.html">metric_binary_focal_crossentropy</a>()</code> <br><code><a href="metric_categorical_crossentropy.html">metric_categorical_crossentropy</a>()</code> <br><code><a href="metric_categorical_focal_crossentropy.html">metric_categorical_focal_crossentropy</a>()</code> <br><code><a href="metric_categorical_hinge.html">metric_categorical_hinge</a>()</code> <br><code><a href="metric_hinge.html">metric_hinge</a>()</code> <br><code><a href="metric_huber.html">metric_huber</a>()</code> <br><code><a href="metric_kl_divergence.html">metric_kl_divergence</a>()</code> <br><code><a href="metric_log_cosh.html">metric_log_cosh</a>()</code> <br><code><a href="metric_mean_absolute_error.html">metric_mean_absolute_error</a>()</code> <br><code><a href="metric_mean_absolute_percentage_error.html">metric_mean_absolute_percentage_error</a>()</code> <br><code><a href="metric_mean_squared_error.html">metric_mean_squared_error</a>()</code> <br><code><a href="metric_mean_squared_logarithmic_error.html">metric_mean_squared_logarithmic_error</a>()</code> <br><code><a href="metric_poisson.html">metric_poisson</a>()</code> <br><code><a href="metric_sparse_categorical_crossentropy.html">metric_sparse_categorical_crossentropy</a>()</code> <br><code><a href="metric_squared_hinge.html">metric_squared_hinge</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -139,7 +139,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/loss_mean_absolute_percentage_error.html b/docs/dev/reference/loss_mean_absolute_percentage_error.html
index 7f392c448..e0818cee8 100644
--- a/docs/dev/reference/loss_mean_absolute_percentage_error.html
+++ b/docs/dev/reference/loss_mean_absolute_percentage_error.html
@@ -20,7 +20,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -139,7 +139,7 @@ <h2 id="examples">Examples<a class="anchor" aria-label="anchor" href="#examples"
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
     <div class="dont-index">
 <ul><li><p><a href="https://keras.io/api/losses/regression_losses#meanabsolutepercentageerror-class" class="external-link">https://keras.io/api/losses/regression_losses#meanabsolutepercentageerror-class</a></p></li>
-</ul><p>Other losses: <br><code><a href="Loss.html">Loss</a>()</code> <br><code><a href="loss_binary_crossentropy.html">loss_binary_crossentropy</a>()</code> <br><code><a href="loss_binary_focal_crossentropy.html">loss_binary_focal_crossentropy</a>()</code> <br><code><a href="loss_categorical_crossentropy.html">loss_categorical_crossentropy</a>()</code> <br><code><a href="loss_categorical_focal_crossentropy.html">loss_categorical_focal_crossentropy</a>()</code> <br><code><a href="loss_categorical_hinge.html">loss_categorical_hinge</a>()</code> <br><code><a href="loss_circle.html">loss_circle</a>()</code> <br><code><a href="loss_cosine_similarity.html">loss_cosine_similarity</a>()</code> <br><code><a href="loss_ctc.html">loss_ctc</a>()</code> <br><code><a href="loss_dice.html">loss_dice</a>()</code> <br><code><a href="loss_hinge.html">loss_hinge</a>()</code> <br><code><a href="loss_huber.html">loss_huber</a>()</code> <br><code><a href="loss_kl_divergence.html">loss_kl_divergence</a>()</code> <br><code><a href="loss_log_cosh.html">loss_log_cosh</a>()</code> <br><code><a href="loss_mean_absolute_error.html">loss_mean_absolute_error</a>()</code> <br><code><a href="loss_mean_squared_error.html">loss_mean_squared_error</a>()</code> <br><code><a href="loss_mean_squared_logarithmic_error.html">loss_mean_squared_logarithmic_error</a>()</code> <br><code><a href="loss_poisson.html">loss_poisson</a>()</code> <br><code><a href="loss_sparse_categorical_crossentropy.html">loss_sparse_categorical_crossentropy</a>()</code> <br><code><a href="loss_squared_hinge.html">loss_squared_hinge</a>()</code> <br><code><a href="loss_tversky.html">loss_tversky</a>()</code> <br><code><a href="metric_binary_crossentropy.html">metric_binary_crossentropy</a>()</code> <br><code><a href="metric_binary_focal_crossentropy.html">metric_binary_focal_crossentropy</a>()</code> <br><code><a href="metric_categorical_crossentropy.html">metric_categorical_crossentropy</a>()</code> <br><code><a href="metric_categorical_focal_crossentropy.html">metric_categorical_focal_crossentropy</a>()</code> <br><code><a href="metric_categorical_hinge.html">metric_categorical_hinge</a>()</code> <br><code><a href="metric_hinge.html">metric_hinge</a>()</code> <br><code><a href="metric_huber.html">metric_huber</a>()</code> <br><code><a href="metric_kl_divergence.html">metric_kl_divergence</a>()</code> <br><code><a href="metric_log_cosh.html">metric_log_cosh</a>()</code> <br><code><a href="metric_mean_absolute_error.html">metric_mean_absolute_error</a>()</code> <br><code><a href="metric_mean_absolute_percentage_error.html">metric_mean_absolute_percentage_error</a>()</code> <br><code><a href="metric_mean_squared_error.html">metric_mean_squared_error</a>()</code> <br><code><a href="metric_mean_squared_logarithmic_error.html">metric_mean_squared_logarithmic_error</a>()</code> <br><code><a href="metric_poisson.html">metric_poisson</a>()</code> <br><code><a href="metric_sparse_categorical_crossentropy.html">metric_sparse_categorical_crossentropy</a>()</code> <br><code><a href="metric_squared_hinge.html">metric_squared_hinge</a>()</code> <br></p></div>
+</ul><p>Other losses: <br><code><a href="Loss.html">Loss</a>()</code> <br><code><a href="loss_binary_crossentropy.html">loss_binary_crossentropy</a>()</code> <br><code><a href="loss_binary_focal_crossentropy.html">loss_binary_focal_crossentropy</a>()</code> <br><code><a href="loss_categorical_crossentropy.html">loss_categorical_crossentropy</a>()</code> <br><code><a href="loss_categorical_focal_crossentropy.html">loss_categorical_focal_crossentropy</a>()</code> <br><code><a href="loss_categorical_generalized_cross_entropy.html">loss_categorical_generalized_cross_entropy</a>()</code> <br><code><a href="loss_categorical_hinge.html">loss_categorical_hinge</a>()</code> <br><code><a href="loss_circle.html">loss_circle</a>()</code> <br><code><a href="loss_cosine_similarity.html">loss_cosine_similarity</a>()</code> <br><code><a href="loss_ctc.html">loss_ctc</a>()</code> <br><code><a href="loss_dice.html">loss_dice</a>()</code> <br><code><a href="loss_hinge.html">loss_hinge</a>()</code> <br><code><a href="loss_huber.html">loss_huber</a>()</code> <br><code><a href="loss_kl_divergence.html">loss_kl_divergence</a>()</code> <br><code><a href="loss_log_cosh.html">loss_log_cosh</a>()</code> <br><code><a href="loss_mean_absolute_error.html">loss_mean_absolute_error</a>()</code> <br><code><a href="loss_mean_squared_error.html">loss_mean_squared_error</a>()</code> <br><code><a href="loss_mean_squared_logarithmic_error.html">loss_mean_squared_logarithmic_error</a>()</code> <br><code><a href="loss_poisson.html">loss_poisson</a>()</code> <br><code><a href="loss_sparse_categorical_crossentropy.html">loss_sparse_categorical_crossentropy</a>()</code> <br><code><a href="loss_squared_hinge.html">loss_squared_hinge</a>()</code> <br><code><a href="loss_tversky.html">loss_tversky</a>()</code> <br><code><a href="metric_binary_crossentropy.html">metric_binary_crossentropy</a>()</code> <br><code><a href="metric_binary_focal_crossentropy.html">metric_binary_focal_crossentropy</a>()</code> <br><code><a href="metric_categorical_crossentropy.html">metric_categorical_crossentropy</a>()</code> <br><code><a href="metric_categorical_focal_crossentropy.html">metric_categorical_focal_crossentropy</a>()</code> <br><code><a href="metric_categorical_hinge.html">metric_categorical_hinge</a>()</code> <br><code><a href="metric_hinge.html">metric_hinge</a>()</code> <br><code><a href="metric_huber.html">metric_huber</a>()</code> <br><code><a href="metric_kl_divergence.html">metric_kl_divergence</a>()</code> <br><code><a href="metric_log_cosh.html">metric_log_cosh</a>()</code> <br><code><a href="metric_mean_absolute_error.html">metric_mean_absolute_error</a>()</code> <br><code><a href="metric_mean_absolute_percentage_error.html">metric_mean_absolute_percentage_error</a>()</code> <br><code><a href="metric_mean_squared_error.html">metric_mean_squared_error</a>()</code> <br><code><a href="metric_mean_squared_logarithmic_error.html">metric_mean_squared_logarithmic_error</a>()</code> <br><code><a href="metric_poisson.html">metric_poisson</a>()</code> <br><code><a href="metric_sparse_categorical_crossentropy.html">metric_sparse_categorical_crossentropy</a>()</code> <br><code><a href="metric_squared_hinge.html">metric_squared_hinge</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -151,7 +151,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/loss_mean_squared_error.html b/docs/dev/reference/loss_mean_squared_error.html
index e7b24da75..64dddc558 100644
--- a/docs/dev/reference/loss_mean_squared_error.html
+++ b/docs/dev/reference/loss_mean_squared_error.html
@@ -12,7 +12,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -127,7 +127,7 @@ <h2 id="examples">Examples<a class="anchor" aria-label="anchor" href="#examples"
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
     <div class="dont-index">
 <ul><li><p><a href="https://keras.io/api/losses/regression_losses#meansquarederror-class" class="external-link">https://keras.io/api/losses/regression_losses#meansquarederror-class</a></p></li>
-</ul><p>Other losses: <br><code><a href="Loss.html">Loss</a>()</code> <br><code><a href="loss_binary_crossentropy.html">loss_binary_crossentropy</a>()</code> <br><code><a href="loss_binary_focal_crossentropy.html">loss_binary_focal_crossentropy</a>()</code> <br><code><a href="loss_categorical_crossentropy.html">loss_categorical_crossentropy</a>()</code> <br><code><a href="loss_categorical_focal_crossentropy.html">loss_categorical_focal_crossentropy</a>()</code> <br><code><a href="loss_categorical_hinge.html">loss_categorical_hinge</a>()</code> <br><code><a href="loss_circle.html">loss_circle</a>()</code> <br><code><a href="loss_cosine_similarity.html">loss_cosine_similarity</a>()</code> <br><code><a href="loss_ctc.html">loss_ctc</a>()</code> <br><code><a href="loss_dice.html">loss_dice</a>()</code> <br><code><a href="loss_hinge.html">loss_hinge</a>()</code> <br><code><a href="loss_huber.html">loss_huber</a>()</code> <br><code><a href="loss_kl_divergence.html">loss_kl_divergence</a>()</code> <br><code><a href="loss_log_cosh.html">loss_log_cosh</a>()</code> <br><code><a href="loss_mean_absolute_error.html">loss_mean_absolute_error</a>()</code> <br><code><a href="loss_mean_absolute_percentage_error.html">loss_mean_absolute_percentage_error</a>()</code> <br><code><a href="loss_mean_squared_logarithmic_error.html">loss_mean_squared_logarithmic_error</a>()</code> <br><code><a href="loss_poisson.html">loss_poisson</a>()</code> <br><code><a href="loss_sparse_categorical_crossentropy.html">loss_sparse_categorical_crossentropy</a>()</code> <br><code><a href="loss_squared_hinge.html">loss_squared_hinge</a>()</code> <br><code><a href="loss_tversky.html">loss_tversky</a>()</code> <br><code><a href="metric_binary_crossentropy.html">metric_binary_crossentropy</a>()</code> <br><code><a href="metric_binary_focal_crossentropy.html">metric_binary_focal_crossentropy</a>()</code> <br><code><a href="metric_categorical_crossentropy.html">metric_categorical_crossentropy</a>()</code> <br><code><a href="metric_categorical_focal_crossentropy.html">metric_categorical_focal_crossentropy</a>()</code> <br><code><a href="metric_categorical_hinge.html">metric_categorical_hinge</a>()</code> <br><code><a href="metric_hinge.html">metric_hinge</a>()</code> <br><code><a href="metric_huber.html">metric_huber</a>()</code> <br><code><a href="metric_kl_divergence.html">metric_kl_divergence</a>()</code> <br><code><a href="metric_log_cosh.html">metric_log_cosh</a>()</code> <br><code><a href="metric_mean_absolute_error.html">metric_mean_absolute_error</a>()</code> <br><code><a href="metric_mean_absolute_percentage_error.html">metric_mean_absolute_percentage_error</a>()</code> <br><code><a href="metric_mean_squared_error.html">metric_mean_squared_error</a>()</code> <br><code><a href="metric_mean_squared_logarithmic_error.html">metric_mean_squared_logarithmic_error</a>()</code> <br><code><a href="metric_poisson.html">metric_poisson</a>()</code> <br><code><a href="metric_sparse_categorical_crossentropy.html">metric_sparse_categorical_crossentropy</a>()</code> <br><code><a href="metric_squared_hinge.html">metric_squared_hinge</a>()</code> <br></p></div>
+</ul><p>Other losses: <br><code><a href="Loss.html">Loss</a>()</code> <br><code><a href="loss_binary_crossentropy.html">loss_binary_crossentropy</a>()</code> <br><code><a href="loss_binary_focal_crossentropy.html">loss_binary_focal_crossentropy</a>()</code> <br><code><a href="loss_categorical_crossentropy.html">loss_categorical_crossentropy</a>()</code> <br><code><a href="loss_categorical_focal_crossentropy.html">loss_categorical_focal_crossentropy</a>()</code> <br><code><a href="loss_categorical_generalized_cross_entropy.html">loss_categorical_generalized_cross_entropy</a>()</code> <br><code><a href="loss_categorical_hinge.html">loss_categorical_hinge</a>()</code> <br><code><a href="loss_circle.html">loss_circle</a>()</code> <br><code><a href="loss_cosine_similarity.html">loss_cosine_similarity</a>()</code> <br><code><a href="loss_ctc.html">loss_ctc</a>()</code> <br><code><a href="loss_dice.html">loss_dice</a>()</code> <br><code><a href="loss_hinge.html">loss_hinge</a>()</code> <br><code><a href="loss_huber.html">loss_huber</a>()</code> <br><code><a href="loss_kl_divergence.html">loss_kl_divergence</a>()</code> <br><code><a href="loss_log_cosh.html">loss_log_cosh</a>()</code> <br><code><a href="loss_mean_absolute_error.html">loss_mean_absolute_error</a>()</code> <br><code><a href="loss_mean_absolute_percentage_error.html">loss_mean_absolute_percentage_error</a>()</code> <br><code><a href="loss_mean_squared_logarithmic_error.html">loss_mean_squared_logarithmic_error</a>()</code> <br><code><a href="loss_poisson.html">loss_poisson</a>()</code> <br><code><a href="loss_sparse_categorical_crossentropy.html">loss_sparse_categorical_crossentropy</a>()</code> <br><code><a href="loss_squared_hinge.html">loss_squared_hinge</a>()</code> <br><code><a href="loss_tversky.html">loss_tversky</a>()</code> <br><code><a href="metric_binary_crossentropy.html">metric_binary_crossentropy</a>()</code> <br><code><a href="metric_binary_focal_crossentropy.html">metric_binary_focal_crossentropy</a>()</code> <br><code><a href="metric_categorical_crossentropy.html">metric_categorical_crossentropy</a>()</code> <br><code><a href="metric_categorical_focal_crossentropy.html">metric_categorical_focal_crossentropy</a>()</code> <br><code><a href="metric_categorical_hinge.html">metric_categorical_hinge</a>()</code> <br><code><a href="metric_hinge.html">metric_hinge</a>()</code> <br><code><a href="metric_huber.html">metric_huber</a>()</code> <br><code><a href="metric_kl_divergence.html">metric_kl_divergence</a>()</code> <br><code><a href="metric_log_cosh.html">metric_log_cosh</a>()</code> <br><code><a href="metric_mean_absolute_error.html">metric_mean_absolute_error</a>()</code> <br><code><a href="metric_mean_absolute_percentage_error.html">metric_mean_absolute_percentage_error</a>()</code> <br><code><a href="metric_mean_squared_error.html">metric_mean_squared_error</a>()</code> <br><code><a href="metric_mean_squared_logarithmic_error.html">metric_mean_squared_logarithmic_error</a>()</code> <br><code><a href="metric_poisson.html">metric_poisson</a>()</code> <br><code><a href="metric_sparse_categorical_crossentropy.html">metric_sparse_categorical_crossentropy</a>()</code> <br><code><a href="metric_squared_hinge.html">metric_squared_hinge</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -139,7 +139,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/loss_mean_squared_logarithmic_error.html b/docs/dev/reference/loss_mean_squared_logarithmic_error.html
index e8d0595de..909541d61 100644
--- a/docs/dev/reference/loss_mean_squared_logarithmic_error.html
+++ b/docs/dev/reference/loss_mean_squared_logarithmic_error.html
@@ -18,7 +18,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -136,7 +136,7 @@ <h2 id="examples">Examples<a class="anchor" aria-label="anchor" href="#examples"
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
     <div class="dont-index">
 <ul><li><p><a href="https://keras.io/api/losses/regression_losses#meansquaredlogarithmicerror-class" class="external-link">https://keras.io/api/losses/regression_losses#meansquaredlogarithmicerror-class</a></p></li>
-</ul><p>Other losses: <br><code><a href="Loss.html">Loss</a>()</code> <br><code><a href="loss_binary_crossentropy.html">loss_binary_crossentropy</a>()</code> <br><code><a href="loss_binary_focal_crossentropy.html">loss_binary_focal_crossentropy</a>()</code> <br><code><a href="loss_categorical_crossentropy.html">loss_categorical_crossentropy</a>()</code> <br><code><a href="loss_categorical_focal_crossentropy.html">loss_categorical_focal_crossentropy</a>()</code> <br><code><a href="loss_categorical_hinge.html">loss_categorical_hinge</a>()</code> <br><code><a href="loss_circle.html">loss_circle</a>()</code> <br><code><a href="loss_cosine_similarity.html">loss_cosine_similarity</a>()</code> <br><code><a href="loss_ctc.html">loss_ctc</a>()</code> <br><code><a href="loss_dice.html">loss_dice</a>()</code> <br><code><a href="loss_hinge.html">loss_hinge</a>()</code> <br><code><a href="loss_huber.html">loss_huber</a>()</code> <br><code><a href="loss_kl_divergence.html">loss_kl_divergence</a>()</code> <br><code><a href="loss_log_cosh.html">loss_log_cosh</a>()</code> <br><code><a href="loss_mean_absolute_error.html">loss_mean_absolute_error</a>()</code> <br><code><a href="loss_mean_absolute_percentage_error.html">loss_mean_absolute_percentage_error</a>()</code> <br><code><a href="loss_mean_squared_error.html">loss_mean_squared_error</a>()</code> <br><code><a href="loss_poisson.html">loss_poisson</a>()</code> <br><code><a href="loss_sparse_categorical_crossentropy.html">loss_sparse_categorical_crossentropy</a>()</code> <br><code><a href="loss_squared_hinge.html">loss_squared_hinge</a>()</code> <br><code><a href="loss_tversky.html">loss_tversky</a>()</code> <br><code><a href="metric_binary_crossentropy.html">metric_binary_crossentropy</a>()</code> <br><code><a href="metric_binary_focal_crossentropy.html">metric_binary_focal_crossentropy</a>()</code> <br><code><a href="metric_categorical_crossentropy.html">metric_categorical_crossentropy</a>()</code> <br><code><a href="metric_categorical_focal_crossentropy.html">metric_categorical_focal_crossentropy</a>()</code> <br><code><a href="metric_categorical_hinge.html">metric_categorical_hinge</a>()</code> <br><code><a href="metric_hinge.html">metric_hinge</a>()</code> <br><code><a href="metric_huber.html">metric_huber</a>()</code> <br><code><a href="metric_kl_divergence.html">metric_kl_divergence</a>()</code> <br><code><a href="metric_log_cosh.html">metric_log_cosh</a>()</code> <br><code><a href="metric_mean_absolute_error.html">metric_mean_absolute_error</a>()</code> <br><code><a href="metric_mean_absolute_percentage_error.html">metric_mean_absolute_percentage_error</a>()</code> <br><code><a href="metric_mean_squared_error.html">metric_mean_squared_error</a>()</code> <br><code><a href="metric_mean_squared_logarithmic_error.html">metric_mean_squared_logarithmic_error</a>()</code> <br><code><a href="metric_poisson.html">metric_poisson</a>()</code> <br><code><a href="metric_sparse_categorical_crossentropy.html">metric_sparse_categorical_crossentropy</a>()</code> <br><code><a href="metric_squared_hinge.html">metric_squared_hinge</a>()</code> <br></p></div>
+</ul><p>Other losses: <br><code><a href="Loss.html">Loss</a>()</code> <br><code><a href="loss_binary_crossentropy.html">loss_binary_crossentropy</a>()</code> <br><code><a href="loss_binary_focal_crossentropy.html">loss_binary_focal_crossentropy</a>()</code> <br><code><a href="loss_categorical_crossentropy.html">loss_categorical_crossentropy</a>()</code> <br><code><a href="loss_categorical_focal_crossentropy.html">loss_categorical_focal_crossentropy</a>()</code> <br><code><a href="loss_categorical_generalized_cross_entropy.html">loss_categorical_generalized_cross_entropy</a>()</code> <br><code><a href="loss_categorical_hinge.html">loss_categorical_hinge</a>()</code> <br><code><a href="loss_circle.html">loss_circle</a>()</code> <br><code><a href="loss_cosine_similarity.html">loss_cosine_similarity</a>()</code> <br><code><a href="loss_ctc.html">loss_ctc</a>()</code> <br><code><a href="loss_dice.html">loss_dice</a>()</code> <br><code><a href="loss_hinge.html">loss_hinge</a>()</code> <br><code><a href="loss_huber.html">loss_huber</a>()</code> <br><code><a href="loss_kl_divergence.html">loss_kl_divergence</a>()</code> <br><code><a href="loss_log_cosh.html">loss_log_cosh</a>()</code> <br><code><a href="loss_mean_absolute_error.html">loss_mean_absolute_error</a>()</code> <br><code><a href="loss_mean_absolute_percentage_error.html">loss_mean_absolute_percentage_error</a>()</code> <br><code><a href="loss_mean_squared_error.html">loss_mean_squared_error</a>()</code> <br><code><a href="loss_poisson.html">loss_poisson</a>()</code> <br><code><a href="loss_sparse_categorical_crossentropy.html">loss_sparse_categorical_crossentropy</a>()</code> <br><code><a href="loss_squared_hinge.html">loss_squared_hinge</a>()</code> <br><code><a href="loss_tversky.html">loss_tversky</a>()</code> <br><code><a href="metric_binary_crossentropy.html">metric_binary_crossentropy</a>()</code> <br><code><a href="metric_binary_focal_crossentropy.html">metric_binary_focal_crossentropy</a>()</code> <br><code><a href="metric_categorical_crossentropy.html">metric_categorical_crossentropy</a>()</code> <br><code><a href="metric_categorical_focal_crossentropy.html">metric_categorical_focal_crossentropy</a>()</code> <br><code><a href="metric_categorical_hinge.html">metric_categorical_hinge</a>()</code> <br><code><a href="metric_hinge.html">metric_hinge</a>()</code> <br><code><a href="metric_huber.html">metric_huber</a>()</code> <br><code><a href="metric_kl_divergence.html">metric_kl_divergence</a>()</code> <br><code><a href="metric_log_cosh.html">metric_log_cosh</a>()</code> <br><code><a href="metric_mean_absolute_error.html">metric_mean_absolute_error</a>()</code> <br><code><a href="metric_mean_absolute_percentage_error.html">metric_mean_absolute_percentage_error</a>()</code> <br><code><a href="metric_mean_squared_error.html">metric_mean_squared_error</a>()</code> <br><code><a href="metric_mean_squared_logarithmic_error.html">metric_mean_squared_logarithmic_error</a>()</code> <br><code><a href="metric_poisson.html">metric_poisson</a>()</code> <br><code><a href="metric_sparse_categorical_crossentropy.html">metric_sparse_categorical_crossentropy</a>()</code> <br><code><a href="metric_squared_hinge.html">metric_squared_hinge</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -148,7 +148,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/loss_poisson.html b/docs/dev/reference/loss_poisson.html
index 294af79f8..b7925ab86 100644
--- a/docs/dev/reference/loss_poisson.html
+++ b/docs/dev/reference/loss_poisson.html
@@ -12,7 +12,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -130,7 +130,7 @@ <h2 id="examples">Examples<a class="anchor" aria-label="anchor" href="#examples"
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
     <div class="dont-index">
 <ul><li><p><a href="https://keras.io/api/losses/probabilistic_losses#poisson-class" class="external-link">https://keras.io/api/losses/probabilistic_losses#poisson-class</a></p></li>
-</ul><p>Other losses: <br><code><a href="Loss.html">Loss</a>()</code> <br><code><a href="loss_binary_crossentropy.html">loss_binary_crossentropy</a>()</code> <br><code><a href="loss_binary_focal_crossentropy.html">loss_binary_focal_crossentropy</a>()</code> <br><code><a href="loss_categorical_crossentropy.html">loss_categorical_crossentropy</a>()</code> <br><code><a href="loss_categorical_focal_crossentropy.html">loss_categorical_focal_crossentropy</a>()</code> <br><code><a href="loss_categorical_hinge.html">loss_categorical_hinge</a>()</code> <br><code><a href="loss_circle.html">loss_circle</a>()</code> <br><code><a href="loss_cosine_similarity.html">loss_cosine_similarity</a>()</code> <br><code><a href="loss_ctc.html">loss_ctc</a>()</code> <br><code><a href="loss_dice.html">loss_dice</a>()</code> <br><code><a href="loss_hinge.html">loss_hinge</a>()</code> <br><code><a href="loss_huber.html">loss_huber</a>()</code> <br><code><a href="loss_kl_divergence.html">loss_kl_divergence</a>()</code> <br><code><a href="loss_log_cosh.html">loss_log_cosh</a>()</code> <br><code><a href="loss_mean_absolute_error.html">loss_mean_absolute_error</a>()</code> <br><code><a href="loss_mean_absolute_percentage_error.html">loss_mean_absolute_percentage_error</a>()</code> <br><code><a href="loss_mean_squared_error.html">loss_mean_squared_error</a>()</code> <br><code><a href="loss_mean_squared_logarithmic_error.html">loss_mean_squared_logarithmic_error</a>()</code> <br><code><a href="loss_sparse_categorical_crossentropy.html">loss_sparse_categorical_crossentropy</a>()</code> <br><code><a href="loss_squared_hinge.html">loss_squared_hinge</a>()</code> <br><code><a href="loss_tversky.html">loss_tversky</a>()</code> <br><code><a href="metric_binary_crossentropy.html">metric_binary_crossentropy</a>()</code> <br><code><a href="metric_binary_focal_crossentropy.html">metric_binary_focal_crossentropy</a>()</code> <br><code><a href="metric_categorical_crossentropy.html">metric_categorical_crossentropy</a>()</code> <br><code><a href="metric_categorical_focal_crossentropy.html">metric_categorical_focal_crossentropy</a>()</code> <br><code><a href="metric_categorical_hinge.html">metric_categorical_hinge</a>()</code> <br><code><a href="metric_hinge.html">metric_hinge</a>()</code> <br><code><a href="metric_huber.html">metric_huber</a>()</code> <br><code><a href="metric_kl_divergence.html">metric_kl_divergence</a>()</code> <br><code><a href="metric_log_cosh.html">metric_log_cosh</a>()</code> <br><code><a href="metric_mean_absolute_error.html">metric_mean_absolute_error</a>()</code> <br><code><a href="metric_mean_absolute_percentage_error.html">metric_mean_absolute_percentage_error</a>()</code> <br><code><a href="metric_mean_squared_error.html">metric_mean_squared_error</a>()</code> <br><code><a href="metric_mean_squared_logarithmic_error.html">metric_mean_squared_logarithmic_error</a>()</code> <br><code><a href="metric_poisson.html">metric_poisson</a>()</code> <br><code><a href="metric_sparse_categorical_crossentropy.html">metric_sparse_categorical_crossentropy</a>()</code> <br><code><a href="metric_squared_hinge.html">metric_squared_hinge</a>()</code> <br></p></div>
+</ul><p>Other losses: <br><code><a href="Loss.html">Loss</a>()</code> <br><code><a href="loss_binary_crossentropy.html">loss_binary_crossentropy</a>()</code> <br><code><a href="loss_binary_focal_crossentropy.html">loss_binary_focal_crossentropy</a>()</code> <br><code><a href="loss_categorical_crossentropy.html">loss_categorical_crossentropy</a>()</code> <br><code><a href="loss_categorical_focal_crossentropy.html">loss_categorical_focal_crossentropy</a>()</code> <br><code><a href="loss_categorical_generalized_cross_entropy.html">loss_categorical_generalized_cross_entropy</a>()</code> <br><code><a href="loss_categorical_hinge.html">loss_categorical_hinge</a>()</code> <br><code><a href="loss_circle.html">loss_circle</a>()</code> <br><code><a href="loss_cosine_similarity.html">loss_cosine_similarity</a>()</code> <br><code><a href="loss_ctc.html">loss_ctc</a>()</code> <br><code><a href="loss_dice.html">loss_dice</a>()</code> <br><code><a href="loss_hinge.html">loss_hinge</a>()</code> <br><code><a href="loss_huber.html">loss_huber</a>()</code> <br><code><a href="loss_kl_divergence.html">loss_kl_divergence</a>()</code> <br><code><a href="loss_log_cosh.html">loss_log_cosh</a>()</code> <br><code><a href="loss_mean_absolute_error.html">loss_mean_absolute_error</a>()</code> <br><code><a href="loss_mean_absolute_percentage_error.html">loss_mean_absolute_percentage_error</a>()</code> <br><code><a href="loss_mean_squared_error.html">loss_mean_squared_error</a>()</code> <br><code><a href="loss_mean_squared_logarithmic_error.html">loss_mean_squared_logarithmic_error</a>()</code> <br><code><a href="loss_sparse_categorical_crossentropy.html">loss_sparse_categorical_crossentropy</a>()</code> <br><code><a href="loss_squared_hinge.html">loss_squared_hinge</a>()</code> <br><code><a href="loss_tversky.html">loss_tversky</a>()</code> <br><code><a href="metric_binary_crossentropy.html">metric_binary_crossentropy</a>()</code> <br><code><a href="metric_binary_focal_crossentropy.html">metric_binary_focal_crossentropy</a>()</code> <br><code><a href="metric_categorical_crossentropy.html">metric_categorical_crossentropy</a>()</code> <br><code><a href="metric_categorical_focal_crossentropy.html">metric_categorical_focal_crossentropy</a>()</code> <br><code><a href="metric_categorical_hinge.html">metric_categorical_hinge</a>()</code> <br><code><a href="metric_hinge.html">metric_hinge</a>()</code> <br><code><a href="metric_huber.html">metric_huber</a>()</code> <br><code><a href="metric_kl_divergence.html">metric_kl_divergence</a>()</code> <br><code><a href="metric_log_cosh.html">metric_log_cosh</a>()</code> <br><code><a href="metric_mean_absolute_error.html">metric_mean_absolute_error</a>()</code> <br><code><a href="metric_mean_absolute_percentage_error.html">metric_mean_absolute_percentage_error</a>()</code> <br><code><a href="metric_mean_squared_error.html">metric_mean_squared_error</a>()</code> <br><code><a href="metric_mean_squared_logarithmic_error.html">metric_mean_squared_logarithmic_error</a>()</code> <br><code><a href="metric_poisson.html">metric_poisson</a>()</code> <br><code><a href="metric_sparse_categorical_crossentropy.html">metric_sparse_categorical_crossentropy</a>()</code> <br><code><a href="metric_squared_hinge.html">metric_squared_hinge</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -142,7 +142,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/loss_sparse_categorical_crossentropy.html b/docs/dev/reference/loss_sparse_categorical_crossentropy.html
index abf25e0b3..c5b3f3bd8 100644
--- a/docs/dev/reference/loss_sparse_categorical_crossentropy.html
+++ b/docs/dev/reference/loss_sparse_categorical_crossentropy.html
@@ -26,7 +26,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -162,36 +162,33 @@ <h2 id="value">Value<a class="anchor" aria-label="anchor" href="#value"></a></h2
     </div>
     <div class="section level2">
     <h2 id="examples">Examples<a class="anchor" aria-label="anchor" href="#examples"></a></h2>
-    <p></p><div class="sourceCode r"><pre><code><span><span class="va">y_true</span> <span class="op">&lt;-</span> <span class="fu"><a href="https://rdrr.io/r/base/c.html" class="external-link">c</a></span><span class="op">(</span><span class="fl">1</span>, <span class="fl">2</span><span class="op">)</span></span>
-<span><span class="va">y_pred</span> <span class="op">&lt;-</span> <span class="fu"><a href="https://rdrr.io/r/base/cbind.html" class="external-link">rbind</a></span><span class="op">(</span><span class="fu"><a href="https://rdrr.io/r/base/c.html" class="external-link">c</a></span><span class="op">(</span><span class="fl">0.05</span>, <span class="fl">0.95</span>, <span class="fl">0</span><span class="op">)</span>, <span class="fu"><a href="https://rdrr.io/r/base/c.html" class="external-link">c</a></span><span class="op">(</span><span class="fl">0.1</span>, <span class="fl">0.8</span>, <span class="fl">0.1</span><span class="op">)</span><span class="op">)</span></span>
+    <p></p><div class="sourceCode r"><pre><code><span><span class="va">y_true</span> <span class="op">&lt;-</span> <span class="fu"><a href="../reference/op_array.html">op_array</a></span><span class="op">(</span><span class="fu"><a href="https://rdrr.io/r/base/c.html" class="external-link">c</a></span><span class="op">(</span><span class="fl">1L</span>, <span class="fl">2L</span><span class="op">)</span><span class="op">)</span></span>
+<span><span class="va">y_pred</span> <span class="op">&lt;-</span> <span class="fu"><a href="../reference/op_array.html">op_array</a></span><span class="op">(</span><span class="fu"><a href="https://rdrr.io/r/base/cbind.html" class="external-link">rbind</a></span><span class="op">(</span><span class="fu"><a href="https://rdrr.io/r/base/c.html" class="external-link">c</a></span><span class="op">(</span><span class="fl">0.05</span>, <span class="fl">0.95</span>, <span class="fl">0</span><span class="op">)</span>, <span class="fu"><a href="https://rdrr.io/r/base/c.html" class="external-link">c</a></span><span class="op">(</span><span class="fl">0.1</span>, <span class="fl">0.8</span>, <span class="fl">0.1</span><span class="op">)</span><span class="op">)</span><span class="op">)</span></span>
 <span><span class="va">loss</span> <span class="op">&lt;-</span> <span class="fu"><a href="../reference/loss_sparse_categorical_crossentropy.html">loss_sparse_categorical_crossentropy</a></span><span class="op">(</span><span class="va">y_true</span>, <span class="va">y_pred</span><span class="op">)</span></span>
 <span><span class="va">loss</span></span></code></pre><p></p></div>
-<p></p><div class="sourceCode"><pre><code><span id="cb1-1"><a href="#cb1-1" tabindex="-1"></a><span class="do">## tf.Tensor([0.05129344 2.3025851 ], shape=(2), dtype=float32)</span></span>
+<p></p><div class="sourceCode"><pre><code><span id="cb1-1"><a href="#cb1-1" tabindex="-1"></a><span class="do">## tf.Tensor([0.05129339 2.30258509], shape=(2), dtype=float64)</span></span>
 <span id="cb1-2"><a href="#cb1-2" tabindex="-1"></a></span></code></pre><p></p></div>
-<p></p><div class="sourceCode r"><pre><code><span><span class="va">y_true</span> <span class="op">&lt;-</span> <span class="fu"><a href="https://rdrr.io/r/base/c.html" class="external-link">c</a></span><span class="op">(</span><span class="fl">1</span>, <span class="fl">2</span><span class="op">)</span></span>
-<span><span class="va">y_pred</span> <span class="op">&lt;-</span> <span class="fu"><a href="https://rdrr.io/r/base/cbind.html" class="external-link">rbind</a></span><span class="op">(</span><span class="fu"><a href="https://rdrr.io/r/base/c.html" class="external-link">c</a></span><span class="op">(</span><span class="fl">0.05</span>, <span class="fl">0.95</span>, <span class="fl">0</span><span class="op">)</span>, <span class="fu"><a href="https://rdrr.io/r/base/c.html" class="external-link">c</a></span><span class="op">(</span><span class="fl">0.1</span>, <span class="fl">0.8</span>, <span class="fl">0.1</span><span class="op">)</span><span class="op">)</span></span>
+<p></p><div class="sourceCode r"><pre><code><span><span class="va">y_true</span> <span class="op">&lt;-</span> <span class="fu"><a href="../reference/op_array.html">op_array</a></span><span class="op">(</span><span class="fu"><a href="https://rdrr.io/r/base/c.html" class="external-link">c</a></span><span class="op">(</span><span class="fl">1L</span>, <span class="fl">2L</span><span class="op">)</span><span class="op">)</span></span>
+<span><span class="va">y_pred</span> <span class="op">&lt;-</span> <span class="fu"><a href="../reference/op_array.html">op_array</a></span><span class="op">(</span><span class="fu"><a href="https://rdrr.io/r/base/cbind.html" class="external-link">rbind</a></span><span class="op">(</span><span class="fu"><a href="https://rdrr.io/r/base/c.html" class="external-link">c</a></span><span class="op">(</span><span class="fl">0.05</span>, <span class="fl">0.95</span>, <span class="fl">0</span><span class="op">)</span>, <span class="fu"><a href="https://rdrr.io/r/base/c.html" class="external-link">c</a></span><span class="op">(</span><span class="fl">0.1</span>, <span class="fl">0.8</span>, <span class="fl">0.1</span><span class="op">)</span><span class="op">)</span><span class="op">)</span></span>
 <span><span class="co"># Using 'auto'/'sum_over_batch_size' reduction type.</span></span>
 <span><span class="va">scce</span> <span class="op">&lt;-</span> <span class="fu"><a href="../reference/loss_sparse_categorical_crossentropy.html">loss_sparse_categorical_crossentropy</a></span><span class="op">(</span><span class="op">)</span></span>
-<span><span class="fu">scce</span><span class="op">(</span><span class="fu"><a href="../reference/op_array.html">op_array</a></span><span class="op">(</span><span class="va">y_true</span><span class="op">)</span>, <span class="fu"><a href="../reference/op_array.html">op_array</a></span><span class="op">(</span><span class="va">y_pred</span><span class="op">)</span><span class="op">)</span></span></code></pre><p></p></div>
+<span><span class="fu">scce</span><span class="op">(</span><span class="va">y_true</span>, <span class="va">y_pred</span><span class="op">)</span></span></code></pre><p></p></div>
 <p></p><div class="sourceCode"><pre><code><span id="cb1-1"><a href="#cb1-1" tabindex="-1"></a><span class="do">## tf.Tensor(1.1769392, shape=(), dtype=float32)</span></span>
 <span id="cb1-2"><a href="#cb1-2" tabindex="-1"></a></span></code></pre><p></p></div>
-<p></p><div class="sourceCode r"><pre><code><span><span class="co"># 1.177</span></span></code></pre><p></p></div>
 <p></p><div class="sourceCode r"><pre><code><span><span class="co"># Calling with 'sample_weight'.</span></span>
-<span><span class="fu">scce</span><span class="op">(</span><span class="fu"><a href="../reference/op_array.html">op_array</a></span><span class="op">(</span><span class="va">y_true</span><span class="op">)</span>, <span class="fu"><a href="../reference/op_array.html">op_array</a></span><span class="op">(</span><span class="va">y_pred</span><span class="op">)</span>, sample_weight <span class="op">=</span> <span class="fu"><a href="../reference/op_array.html">op_array</a></span><span class="op">(</span><span class="fu"><a href="https://rdrr.io/r/base/c.html" class="external-link">c</a></span><span class="op">(</span><span class="fl">0.3</span>, <span class="fl">0.7</span><span class="op">)</span><span class="op">)</span><span class="op">)</span></span></code></pre><p></p></div>
+<span><span class="fu">scce</span><span class="op">(</span><span class="va">y_true</span>, <span class="va">y_pred</span>, sample_weight <span class="op">=</span> <span class="fu"><a href="../reference/op_array.html">op_array</a></span><span class="op">(</span><span class="fu"><a href="https://rdrr.io/r/base/c.html" class="external-link">c</a></span><span class="op">(</span><span class="fl">0.3</span>, <span class="fl">0.7</span><span class="op">)</span><span class="op">)</span><span class="op">)</span></span></code></pre><p></p></div>
 <p></p><div class="sourceCode"><pre><code><span id="cb1-1"><a href="#cb1-1" tabindex="-1"></a><span class="do">## tf.Tensor(0.8135988, shape=(), dtype=float32)</span></span>
 <span id="cb1-2"><a href="#cb1-2" tabindex="-1"></a></span></code></pre><p></p></div>
 <p></p><div class="sourceCode r"><pre><code><span><span class="co"># Using 'sum' reduction type.</span></span>
 <span><span class="va">scce</span> <span class="op">&lt;-</span> <span class="fu"><a href="../reference/loss_sparse_categorical_crossentropy.html">loss_sparse_categorical_crossentropy</a></span><span class="op">(</span>reduction<span class="op">=</span><span class="st">"sum"</span><span class="op">)</span></span>
-<span><span class="fu">scce</span><span class="op">(</span><span class="fu"><a href="../reference/op_array.html">op_array</a></span><span class="op">(</span><span class="va">y_true</span><span class="op">)</span>, <span class="fu"><a href="../reference/op_array.html">op_array</a></span><span class="op">(</span><span class="va">y_pred</span><span class="op">)</span><span class="op">)</span></span></code></pre><p></p></div>
+<span><span class="fu">scce</span><span class="op">(</span><span class="va">y_true</span>, <span class="va">y_pred</span><span class="op">)</span></span></code></pre><p></p></div>
 <p></p><div class="sourceCode"><pre><code><span id="cb1-1"><a href="#cb1-1" tabindex="-1"></a><span class="do">## tf.Tensor(2.3538785, shape=(), dtype=float32)</span></span>
 <span id="cb1-2"><a href="#cb1-2" tabindex="-1"></a></span></code></pre><p></p></div>
-<p></p><div class="sourceCode r"><pre><code><span><span class="co"># 2.354</span></span></code></pre><p></p></div>
 <p></p><div class="sourceCode r"><pre><code><span><span class="co"># Using 'none' reduction type.</span></span>
 <span><span class="va">scce</span> <span class="op">&lt;-</span> <span class="fu"><a href="../reference/loss_sparse_categorical_crossentropy.html">loss_sparse_categorical_crossentropy</a></span><span class="op">(</span>reduction<span class="op">=</span><span class="cn">NULL</span><span class="op">)</span></span>
-<span><span class="fu">scce</span><span class="op">(</span><span class="fu"><a href="../reference/op_array.html">op_array</a></span><span class="op">(</span><span class="va">y_true</span><span class="op">)</span>, <span class="fu"><a href="../reference/op_array.html">op_array</a></span><span class="op">(</span><span class="va">y_pred</span><span class="op">)</span><span class="op">)</span></span></code></pre><p></p></div>
+<span><span class="fu">scce</span><span class="op">(</span><span class="va">y_true</span>, <span class="va">y_pred</span><span class="op">)</span></span></code></pre><p></p></div>
 <p></p><div class="sourceCode"><pre><code><span id="cb1-1"><a href="#cb1-1" tabindex="-1"></a><span class="do">## tf.Tensor([0.05129344 2.3025851 ], shape=(2), dtype=float32)</span></span>
 <span id="cb1-2"><a href="#cb1-2" tabindex="-1"></a></span></code></pre><p></p></div>
-<p></p><div class="sourceCode r"><pre><code><span><span class="co"># array([0.0513, 2.303], dtype=float32)</span></span></code></pre><p></p></div>
 <p>Usage with the <code><a href="https://generics.r-lib.org/reference/compile.html" class="external-link">compile()</a></code> API:</p>
 <p></p><div class="sourceCode r"><pre><code><span><span class="va">model</span> <span class="op"><a href="../reference/pipe.html">%&gt;%</a></span> <span class="fu"><a href="https://generics.r-lib.org/reference/compile.html" class="external-link">compile</a></span><span class="op">(</span>optimizer <span class="op">=</span> <span class="st">'sgd'</span>,</span>
 <span>                  loss <span class="op">=</span> <span class="fu"><a href="../reference/loss_sparse_categorical_crossentropy.html">loss_sparse_categorical_crossentropy</a></span><span class="op">(</span><span class="op">)</span><span class="op">)</span></span></code></pre><p></p></div>
@@ -200,7 +197,7 @@ <h2 id="examples">Examples<a class="anchor" aria-label="anchor" href="#examples"
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
     <div class="dont-index">
 <ul><li><p><a href="https://keras.io/api/losses/probabilistic_losses#sparsecategoricalcrossentropy-class" class="external-link">https://keras.io/api/losses/probabilistic_losses#sparsecategoricalcrossentropy-class</a></p></li>
-</ul><p>Other losses: <br><code><a href="Loss.html">Loss</a>()</code> <br><code><a href="loss_binary_crossentropy.html">loss_binary_crossentropy</a>()</code> <br><code><a href="loss_binary_focal_crossentropy.html">loss_binary_focal_crossentropy</a>()</code> <br><code><a href="loss_categorical_crossentropy.html">loss_categorical_crossentropy</a>()</code> <br><code><a href="loss_categorical_focal_crossentropy.html">loss_categorical_focal_crossentropy</a>()</code> <br><code><a href="loss_categorical_hinge.html">loss_categorical_hinge</a>()</code> <br><code><a href="loss_circle.html">loss_circle</a>()</code> <br><code><a href="loss_cosine_similarity.html">loss_cosine_similarity</a>()</code> <br><code><a href="loss_ctc.html">loss_ctc</a>()</code> <br><code><a href="loss_dice.html">loss_dice</a>()</code> <br><code><a href="loss_hinge.html">loss_hinge</a>()</code> <br><code><a href="loss_huber.html">loss_huber</a>()</code> <br><code><a href="loss_kl_divergence.html">loss_kl_divergence</a>()</code> <br><code><a href="loss_log_cosh.html">loss_log_cosh</a>()</code> <br><code><a href="loss_mean_absolute_error.html">loss_mean_absolute_error</a>()</code> <br><code><a href="loss_mean_absolute_percentage_error.html">loss_mean_absolute_percentage_error</a>()</code> <br><code><a href="loss_mean_squared_error.html">loss_mean_squared_error</a>()</code> <br><code><a href="loss_mean_squared_logarithmic_error.html">loss_mean_squared_logarithmic_error</a>()</code> <br><code><a href="loss_poisson.html">loss_poisson</a>()</code> <br><code><a href="loss_squared_hinge.html">loss_squared_hinge</a>()</code> <br><code><a href="loss_tversky.html">loss_tversky</a>()</code> <br><code><a href="metric_binary_crossentropy.html">metric_binary_crossentropy</a>()</code> <br><code><a href="metric_binary_focal_crossentropy.html">metric_binary_focal_crossentropy</a>()</code> <br><code><a href="metric_categorical_crossentropy.html">metric_categorical_crossentropy</a>()</code> <br><code><a href="metric_categorical_focal_crossentropy.html">metric_categorical_focal_crossentropy</a>()</code> <br><code><a href="metric_categorical_hinge.html">metric_categorical_hinge</a>()</code> <br><code><a href="metric_hinge.html">metric_hinge</a>()</code> <br><code><a href="metric_huber.html">metric_huber</a>()</code> <br><code><a href="metric_kl_divergence.html">metric_kl_divergence</a>()</code> <br><code><a href="metric_log_cosh.html">metric_log_cosh</a>()</code> <br><code><a href="metric_mean_absolute_error.html">metric_mean_absolute_error</a>()</code> <br><code><a href="metric_mean_absolute_percentage_error.html">metric_mean_absolute_percentage_error</a>()</code> <br><code><a href="metric_mean_squared_error.html">metric_mean_squared_error</a>()</code> <br><code><a href="metric_mean_squared_logarithmic_error.html">metric_mean_squared_logarithmic_error</a>()</code> <br><code><a href="metric_poisson.html">metric_poisson</a>()</code> <br><code><a href="metric_sparse_categorical_crossentropy.html">metric_sparse_categorical_crossentropy</a>()</code> <br><code><a href="metric_squared_hinge.html">metric_squared_hinge</a>()</code> <br></p></div>
+</ul><p>Other losses: <br><code><a href="Loss.html">Loss</a>()</code> <br><code><a href="loss_binary_crossentropy.html">loss_binary_crossentropy</a>()</code> <br><code><a href="loss_binary_focal_crossentropy.html">loss_binary_focal_crossentropy</a>()</code> <br><code><a href="loss_categorical_crossentropy.html">loss_categorical_crossentropy</a>()</code> <br><code><a href="loss_categorical_focal_crossentropy.html">loss_categorical_focal_crossentropy</a>()</code> <br><code><a href="loss_categorical_generalized_cross_entropy.html">loss_categorical_generalized_cross_entropy</a>()</code> <br><code><a href="loss_categorical_hinge.html">loss_categorical_hinge</a>()</code> <br><code><a href="loss_circle.html">loss_circle</a>()</code> <br><code><a href="loss_cosine_similarity.html">loss_cosine_similarity</a>()</code> <br><code><a href="loss_ctc.html">loss_ctc</a>()</code> <br><code><a href="loss_dice.html">loss_dice</a>()</code> <br><code><a href="loss_hinge.html">loss_hinge</a>()</code> <br><code><a href="loss_huber.html">loss_huber</a>()</code> <br><code><a href="loss_kl_divergence.html">loss_kl_divergence</a>()</code> <br><code><a href="loss_log_cosh.html">loss_log_cosh</a>()</code> <br><code><a href="loss_mean_absolute_error.html">loss_mean_absolute_error</a>()</code> <br><code><a href="loss_mean_absolute_percentage_error.html">loss_mean_absolute_percentage_error</a>()</code> <br><code><a href="loss_mean_squared_error.html">loss_mean_squared_error</a>()</code> <br><code><a href="loss_mean_squared_logarithmic_error.html">loss_mean_squared_logarithmic_error</a>()</code> <br><code><a href="loss_poisson.html">loss_poisson</a>()</code> <br><code><a href="loss_squared_hinge.html">loss_squared_hinge</a>()</code> <br><code><a href="loss_tversky.html">loss_tversky</a>()</code> <br><code><a href="metric_binary_crossentropy.html">metric_binary_crossentropy</a>()</code> <br><code><a href="metric_binary_focal_crossentropy.html">metric_binary_focal_crossentropy</a>()</code> <br><code><a href="metric_categorical_crossentropy.html">metric_categorical_crossentropy</a>()</code> <br><code><a href="metric_categorical_focal_crossentropy.html">metric_categorical_focal_crossentropy</a>()</code> <br><code><a href="metric_categorical_hinge.html">metric_categorical_hinge</a>()</code> <br><code><a href="metric_hinge.html">metric_hinge</a>()</code> <br><code><a href="metric_huber.html">metric_huber</a>()</code> <br><code><a href="metric_kl_divergence.html">metric_kl_divergence</a>()</code> <br><code><a href="metric_log_cosh.html">metric_log_cosh</a>()</code> <br><code><a href="metric_mean_absolute_error.html">metric_mean_absolute_error</a>()</code> <br><code><a href="metric_mean_absolute_percentage_error.html">metric_mean_absolute_percentage_error</a>()</code> <br><code><a href="metric_mean_squared_error.html">metric_mean_squared_error</a>()</code> <br><code><a href="metric_mean_squared_logarithmic_error.html">metric_mean_squared_logarithmic_error</a>()</code> <br><code><a href="metric_poisson.html">metric_poisson</a>()</code> <br><code><a href="metric_sparse_categorical_crossentropy.html">metric_sparse_categorical_crossentropy</a>()</code> <br><code><a href="metric_squared_hinge.html">metric_squared_hinge</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -212,7 +209,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/loss_squared_hinge.html b/docs/dev/reference/loss_squared_hinge.html
index 9785b15ea..d7c59eb3f 100644
--- a/docs/dev/reference/loss_squared_hinge.html
+++ b/docs/dev/reference/loss_squared_hinge.html
@@ -16,7 +16,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -135,7 +135,7 @@ <h2 id="examples">Examples<a class="anchor" aria-label="anchor" href="#examples"
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
     <div class="dont-index">
 <ul><li><p><a href="https://keras.io/api/losses/hinge_losses#squaredhinge-class" class="external-link">https://keras.io/api/losses/hinge_losses#squaredhinge-class</a></p></li>
-</ul><p>Other losses: <br><code><a href="Loss.html">Loss</a>()</code> <br><code><a href="loss_binary_crossentropy.html">loss_binary_crossentropy</a>()</code> <br><code><a href="loss_binary_focal_crossentropy.html">loss_binary_focal_crossentropy</a>()</code> <br><code><a href="loss_categorical_crossentropy.html">loss_categorical_crossentropy</a>()</code> <br><code><a href="loss_categorical_focal_crossentropy.html">loss_categorical_focal_crossentropy</a>()</code> <br><code><a href="loss_categorical_hinge.html">loss_categorical_hinge</a>()</code> <br><code><a href="loss_circle.html">loss_circle</a>()</code> <br><code><a href="loss_cosine_similarity.html">loss_cosine_similarity</a>()</code> <br><code><a href="loss_ctc.html">loss_ctc</a>()</code> <br><code><a href="loss_dice.html">loss_dice</a>()</code> <br><code><a href="loss_hinge.html">loss_hinge</a>()</code> <br><code><a href="loss_huber.html">loss_huber</a>()</code> <br><code><a href="loss_kl_divergence.html">loss_kl_divergence</a>()</code> <br><code><a href="loss_log_cosh.html">loss_log_cosh</a>()</code> <br><code><a href="loss_mean_absolute_error.html">loss_mean_absolute_error</a>()</code> <br><code><a href="loss_mean_absolute_percentage_error.html">loss_mean_absolute_percentage_error</a>()</code> <br><code><a href="loss_mean_squared_error.html">loss_mean_squared_error</a>()</code> <br><code><a href="loss_mean_squared_logarithmic_error.html">loss_mean_squared_logarithmic_error</a>()</code> <br><code><a href="loss_poisson.html">loss_poisson</a>()</code> <br><code><a href="loss_sparse_categorical_crossentropy.html">loss_sparse_categorical_crossentropy</a>()</code> <br><code><a href="loss_tversky.html">loss_tversky</a>()</code> <br><code><a href="metric_binary_crossentropy.html">metric_binary_crossentropy</a>()</code> <br><code><a href="metric_binary_focal_crossentropy.html">metric_binary_focal_crossentropy</a>()</code> <br><code><a href="metric_categorical_crossentropy.html">metric_categorical_crossentropy</a>()</code> <br><code><a href="metric_categorical_focal_crossentropy.html">metric_categorical_focal_crossentropy</a>()</code> <br><code><a href="metric_categorical_hinge.html">metric_categorical_hinge</a>()</code> <br><code><a href="metric_hinge.html">metric_hinge</a>()</code> <br><code><a href="metric_huber.html">metric_huber</a>()</code> <br><code><a href="metric_kl_divergence.html">metric_kl_divergence</a>()</code> <br><code><a href="metric_log_cosh.html">metric_log_cosh</a>()</code> <br><code><a href="metric_mean_absolute_error.html">metric_mean_absolute_error</a>()</code> <br><code><a href="metric_mean_absolute_percentage_error.html">metric_mean_absolute_percentage_error</a>()</code> <br><code><a href="metric_mean_squared_error.html">metric_mean_squared_error</a>()</code> <br><code><a href="metric_mean_squared_logarithmic_error.html">metric_mean_squared_logarithmic_error</a>()</code> <br><code><a href="metric_poisson.html">metric_poisson</a>()</code> <br><code><a href="metric_sparse_categorical_crossentropy.html">metric_sparse_categorical_crossentropy</a>()</code> <br><code><a href="metric_squared_hinge.html">metric_squared_hinge</a>()</code> <br></p></div>
+</ul><p>Other losses: <br><code><a href="Loss.html">Loss</a>()</code> <br><code><a href="loss_binary_crossentropy.html">loss_binary_crossentropy</a>()</code> <br><code><a href="loss_binary_focal_crossentropy.html">loss_binary_focal_crossentropy</a>()</code> <br><code><a href="loss_categorical_crossentropy.html">loss_categorical_crossentropy</a>()</code> <br><code><a href="loss_categorical_focal_crossentropy.html">loss_categorical_focal_crossentropy</a>()</code> <br><code><a href="loss_categorical_generalized_cross_entropy.html">loss_categorical_generalized_cross_entropy</a>()</code> <br><code><a href="loss_categorical_hinge.html">loss_categorical_hinge</a>()</code> <br><code><a href="loss_circle.html">loss_circle</a>()</code> <br><code><a href="loss_cosine_similarity.html">loss_cosine_similarity</a>()</code> <br><code><a href="loss_ctc.html">loss_ctc</a>()</code> <br><code><a href="loss_dice.html">loss_dice</a>()</code> <br><code><a href="loss_hinge.html">loss_hinge</a>()</code> <br><code><a href="loss_huber.html">loss_huber</a>()</code> <br><code><a href="loss_kl_divergence.html">loss_kl_divergence</a>()</code> <br><code><a href="loss_log_cosh.html">loss_log_cosh</a>()</code> <br><code><a href="loss_mean_absolute_error.html">loss_mean_absolute_error</a>()</code> <br><code><a href="loss_mean_absolute_percentage_error.html">loss_mean_absolute_percentage_error</a>()</code> <br><code><a href="loss_mean_squared_error.html">loss_mean_squared_error</a>()</code> <br><code><a href="loss_mean_squared_logarithmic_error.html">loss_mean_squared_logarithmic_error</a>()</code> <br><code><a href="loss_poisson.html">loss_poisson</a>()</code> <br><code><a href="loss_sparse_categorical_crossentropy.html">loss_sparse_categorical_crossentropy</a>()</code> <br><code><a href="loss_tversky.html">loss_tversky</a>()</code> <br><code><a href="metric_binary_crossentropy.html">metric_binary_crossentropy</a>()</code> <br><code><a href="metric_binary_focal_crossentropy.html">metric_binary_focal_crossentropy</a>()</code> <br><code><a href="metric_categorical_crossentropy.html">metric_categorical_crossentropy</a>()</code> <br><code><a href="metric_categorical_focal_crossentropy.html">metric_categorical_focal_crossentropy</a>()</code> <br><code><a href="metric_categorical_hinge.html">metric_categorical_hinge</a>()</code> <br><code><a href="metric_hinge.html">metric_hinge</a>()</code> <br><code><a href="metric_huber.html">metric_huber</a>()</code> <br><code><a href="metric_kl_divergence.html">metric_kl_divergence</a>()</code> <br><code><a href="metric_log_cosh.html">metric_log_cosh</a>()</code> <br><code><a href="metric_mean_absolute_error.html">metric_mean_absolute_error</a>()</code> <br><code><a href="metric_mean_absolute_percentage_error.html">metric_mean_absolute_percentage_error</a>()</code> <br><code><a href="metric_mean_squared_error.html">metric_mean_squared_error</a>()</code> <br><code><a href="metric_mean_squared_logarithmic_error.html">metric_mean_squared_logarithmic_error</a>()</code> <br><code><a href="metric_poisson.html">metric_poisson</a>()</code> <br><code><a href="metric_sparse_categorical_crossentropy.html">metric_sparse_categorical_crossentropy</a>()</code> <br><code><a href="metric_squared_hinge.html">metric_squared_hinge</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -147,7 +147,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/loss_tversky.html b/docs/dev/reference/loss_tversky.html
index 39a531b34..fa410f84a 100644
--- a/docs/dev/reference/loss_tversky.html
+++ b/docs/dev/reference/loss_tversky.html
@@ -22,7 +22,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -157,7 +157,7 @@ <h2 id="reference">Reference<a class="anchor" aria-label="anchor" href="#referen
 </ul></div>
     <div class="section level2">
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
-    <div class="dont-index"><p>Other losses: <br><code><a href="Loss.html">Loss</a>()</code> <br><code><a href="loss_binary_crossentropy.html">loss_binary_crossentropy</a>()</code> <br><code><a href="loss_binary_focal_crossentropy.html">loss_binary_focal_crossentropy</a>()</code> <br><code><a href="loss_categorical_crossentropy.html">loss_categorical_crossentropy</a>()</code> <br><code><a href="loss_categorical_focal_crossentropy.html">loss_categorical_focal_crossentropy</a>()</code> <br><code><a href="loss_categorical_hinge.html">loss_categorical_hinge</a>()</code> <br><code><a href="loss_circle.html">loss_circle</a>()</code> <br><code><a href="loss_cosine_similarity.html">loss_cosine_similarity</a>()</code> <br><code><a href="loss_ctc.html">loss_ctc</a>()</code> <br><code><a href="loss_dice.html">loss_dice</a>()</code> <br><code><a href="loss_hinge.html">loss_hinge</a>()</code> <br><code><a href="loss_huber.html">loss_huber</a>()</code> <br><code><a href="loss_kl_divergence.html">loss_kl_divergence</a>()</code> <br><code><a href="loss_log_cosh.html">loss_log_cosh</a>()</code> <br><code><a href="loss_mean_absolute_error.html">loss_mean_absolute_error</a>()</code> <br><code><a href="loss_mean_absolute_percentage_error.html">loss_mean_absolute_percentage_error</a>()</code> <br><code><a href="loss_mean_squared_error.html">loss_mean_squared_error</a>()</code> <br><code><a href="loss_mean_squared_logarithmic_error.html">loss_mean_squared_logarithmic_error</a>()</code> <br><code><a href="loss_poisson.html">loss_poisson</a>()</code> <br><code><a href="loss_sparse_categorical_crossentropy.html">loss_sparse_categorical_crossentropy</a>()</code> <br><code><a href="loss_squared_hinge.html">loss_squared_hinge</a>()</code> <br><code><a href="metric_binary_crossentropy.html">metric_binary_crossentropy</a>()</code> <br><code><a href="metric_binary_focal_crossentropy.html">metric_binary_focal_crossentropy</a>()</code> <br><code><a href="metric_categorical_crossentropy.html">metric_categorical_crossentropy</a>()</code> <br><code><a href="metric_categorical_focal_crossentropy.html">metric_categorical_focal_crossentropy</a>()</code> <br><code><a href="metric_categorical_hinge.html">metric_categorical_hinge</a>()</code> <br><code><a href="metric_hinge.html">metric_hinge</a>()</code> <br><code><a href="metric_huber.html">metric_huber</a>()</code> <br><code><a href="metric_kl_divergence.html">metric_kl_divergence</a>()</code> <br><code><a href="metric_log_cosh.html">metric_log_cosh</a>()</code> <br><code><a href="metric_mean_absolute_error.html">metric_mean_absolute_error</a>()</code> <br><code><a href="metric_mean_absolute_percentage_error.html">metric_mean_absolute_percentage_error</a>()</code> <br><code><a href="metric_mean_squared_error.html">metric_mean_squared_error</a>()</code> <br><code><a href="metric_mean_squared_logarithmic_error.html">metric_mean_squared_logarithmic_error</a>()</code> <br><code><a href="metric_poisson.html">metric_poisson</a>()</code> <br><code><a href="metric_sparse_categorical_crossentropy.html">metric_sparse_categorical_crossentropy</a>()</code> <br><code><a href="metric_squared_hinge.html">metric_squared_hinge</a>()</code> <br></p></div>
+    <div class="dont-index"><p>Other losses: <br><code><a href="Loss.html">Loss</a>()</code> <br><code><a href="loss_binary_crossentropy.html">loss_binary_crossentropy</a>()</code> <br><code><a href="loss_binary_focal_crossentropy.html">loss_binary_focal_crossentropy</a>()</code> <br><code><a href="loss_categorical_crossentropy.html">loss_categorical_crossentropy</a>()</code> <br><code><a href="loss_categorical_focal_crossentropy.html">loss_categorical_focal_crossentropy</a>()</code> <br><code><a href="loss_categorical_generalized_cross_entropy.html">loss_categorical_generalized_cross_entropy</a>()</code> <br><code><a href="loss_categorical_hinge.html">loss_categorical_hinge</a>()</code> <br><code><a href="loss_circle.html">loss_circle</a>()</code> <br><code><a href="loss_cosine_similarity.html">loss_cosine_similarity</a>()</code> <br><code><a href="loss_ctc.html">loss_ctc</a>()</code> <br><code><a href="loss_dice.html">loss_dice</a>()</code> <br><code><a href="loss_hinge.html">loss_hinge</a>()</code> <br><code><a href="loss_huber.html">loss_huber</a>()</code> <br><code><a href="loss_kl_divergence.html">loss_kl_divergence</a>()</code> <br><code><a href="loss_log_cosh.html">loss_log_cosh</a>()</code> <br><code><a href="loss_mean_absolute_error.html">loss_mean_absolute_error</a>()</code> <br><code><a href="loss_mean_absolute_percentage_error.html">loss_mean_absolute_percentage_error</a>()</code> <br><code><a href="loss_mean_squared_error.html">loss_mean_squared_error</a>()</code> <br><code><a href="loss_mean_squared_logarithmic_error.html">loss_mean_squared_logarithmic_error</a>()</code> <br><code><a href="loss_poisson.html">loss_poisson</a>()</code> <br><code><a href="loss_sparse_categorical_crossentropy.html">loss_sparse_categorical_crossentropy</a>()</code> <br><code><a href="loss_squared_hinge.html">loss_squared_hinge</a>()</code> <br><code><a href="metric_binary_crossentropy.html">metric_binary_crossentropy</a>()</code> <br><code><a href="metric_binary_focal_crossentropy.html">metric_binary_focal_crossentropy</a>()</code> <br><code><a href="metric_categorical_crossentropy.html">metric_categorical_crossentropy</a>()</code> <br><code><a href="metric_categorical_focal_crossentropy.html">metric_categorical_focal_crossentropy</a>()</code> <br><code><a href="metric_categorical_hinge.html">metric_categorical_hinge</a>()</code> <br><code><a href="metric_hinge.html">metric_hinge</a>()</code> <br><code><a href="metric_huber.html">metric_huber</a>()</code> <br><code><a href="metric_kl_divergence.html">metric_kl_divergence</a>()</code> <br><code><a href="metric_log_cosh.html">metric_log_cosh</a>()</code> <br><code><a href="metric_mean_absolute_error.html">metric_mean_absolute_error</a>()</code> <br><code><a href="metric_mean_absolute_percentage_error.html">metric_mean_absolute_percentage_error</a>()</code> <br><code><a href="metric_mean_squared_error.html">metric_mean_squared_error</a>()</code> <br><code><a href="metric_mean_squared_logarithmic_error.html">metric_mean_squared_logarithmic_error</a>()</code> <br><code><a href="metric_poisson.html">metric_poisson</a>()</code> <br><code><a href="metric_sparse_categorical_crossentropy.html">metric_sparse_categorical_crossentropy</a>()</code> <br><code><a href="metric_squared_hinge.html">metric_squared_hinge</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -169,7 +169,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/mark_active.html b/docs/dev/reference/mark_active.html
index 9d2621392..911609d43 100644
--- a/docs/dev/reference/mark_active.html
+++ b/docs/dev/reference/mark_active.html
@@ -10,7 +10,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -89,7 +89,7 @@ <h2 id="value">Value<a class="anchor" aria-label="anchor" href="#value"></a></h2
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/metric_auc.html b/docs/dev/reference/metric_auc.html
index 44de562d3..2f2e990d1 100644
--- a/docs/dev/reference/metric_auc.html
+++ b/docs/dev/reference/metric_auc.html
@@ -68,7 +68,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -305,7 +305,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/metric_binary_accuracy.html b/docs/dev/reference/metric_binary_accuracy.html
index 2d78af4c8..cac0cacb9 100644
--- a/docs/dev/reference/metric_binary_accuracy.html
+++ b/docs/dev/reference/metric_binary_accuracy.html
@@ -18,7 +18,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -159,7 +159,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/metric_binary_crossentropy.html b/docs/dev/reference/metric_binary_crossentropy.html
index 4419d167a..5a1919c90 100644
--- a/docs/dev/reference/metric_binary_crossentropy.html
+++ b/docs/dev/reference/metric_binary_crossentropy.html
@@ -10,7 +10,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -148,7 +148,7 @@ <h2 id="examples">Examples<a class="anchor" aria-label="anchor" href="#examples"
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
     <div class="dont-index">
 <ul><li><p><a href="https://keras.io/api/metrics/probabilistic_metrics#binarycrossentropy-class" class="external-link">https://keras.io/api/metrics/probabilistic_metrics#binarycrossentropy-class</a></p></li>
-</ul><p>Other losses: <br><code><a href="Loss.html">Loss</a>()</code> <br><code><a href="loss_binary_crossentropy.html">loss_binary_crossentropy</a>()</code> <br><code><a href="loss_binary_focal_crossentropy.html">loss_binary_focal_crossentropy</a>()</code> <br><code><a href="loss_categorical_crossentropy.html">loss_categorical_crossentropy</a>()</code> <br><code><a href="loss_categorical_focal_crossentropy.html">loss_categorical_focal_crossentropy</a>()</code> <br><code><a href="loss_categorical_hinge.html">loss_categorical_hinge</a>()</code> <br><code><a href="loss_circle.html">loss_circle</a>()</code> <br><code><a href="loss_cosine_similarity.html">loss_cosine_similarity</a>()</code> <br><code><a href="loss_ctc.html">loss_ctc</a>()</code> <br><code><a href="loss_dice.html">loss_dice</a>()</code> <br><code><a href="loss_hinge.html">loss_hinge</a>()</code> <br><code><a href="loss_huber.html">loss_huber</a>()</code> <br><code><a href="loss_kl_divergence.html">loss_kl_divergence</a>()</code> <br><code><a href="loss_log_cosh.html">loss_log_cosh</a>()</code> <br><code><a href="loss_mean_absolute_error.html">loss_mean_absolute_error</a>()</code> <br><code><a href="loss_mean_absolute_percentage_error.html">loss_mean_absolute_percentage_error</a>()</code> <br><code><a href="loss_mean_squared_error.html">loss_mean_squared_error</a>()</code> <br><code><a href="loss_mean_squared_logarithmic_error.html">loss_mean_squared_logarithmic_error</a>()</code> <br><code><a href="loss_poisson.html">loss_poisson</a>()</code> <br><code><a href="loss_sparse_categorical_crossentropy.html">loss_sparse_categorical_crossentropy</a>()</code> <br><code><a href="loss_squared_hinge.html">loss_squared_hinge</a>()</code> <br><code><a href="loss_tversky.html">loss_tversky</a>()</code> <br><code><a href="metric_binary_focal_crossentropy.html">metric_binary_focal_crossentropy</a>()</code> <br><code><a href="metric_categorical_crossentropy.html">metric_categorical_crossentropy</a>()</code> <br><code><a href="metric_categorical_focal_crossentropy.html">metric_categorical_focal_crossentropy</a>()</code> <br><code><a href="metric_categorical_hinge.html">metric_categorical_hinge</a>()</code> <br><code><a href="metric_hinge.html">metric_hinge</a>()</code> <br><code><a href="metric_huber.html">metric_huber</a>()</code> <br><code><a href="metric_kl_divergence.html">metric_kl_divergence</a>()</code> <br><code><a href="metric_log_cosh.html">metric_log_cosh</a>()</code> <br><code><a href="metric_mean_absolute_error.html">metric_mean_absolute_error</a>()</code> <br><code><a href="metric_mean_absolute_percentage_error.html">metric_mean_absolute_percentage_error</a>()</code> <br><code><a href="metric_mean_squared_error.html">metric_mean_squared_error</a>()</code> <br><code><a href="metric_mean_squared_logarithmic_error.html">metric_mean_squared_logarithmic_error</a>()</code> <br><code><a href="metric_poisson.html">metric_poisson</a>()</code> <br><code><a href="metric_sparse_categorical_crossentropy.html">metric_sparse_categorical_crossentropy</a>()</code> <br><code><a href="metric_squared_hinge.html">metric_squared_hinge</a>()</code> <br></p>
+</ul><p>Other losses: <br><code><a href="Loss.html">Loss</a>()</code> <br><code><a href="loss_binary_crossentropy.html">loss_binary_crossentropy</a>()</code> <br><code><a href="loss_binary_focal_crossentropy.html">loss_binary_focal_crossentropy</a>()</code> <br><code><a href="loss_categorical_crossentropy.html">loss_categorical_crossentropy</a>()</code> <br><code><a href="loss_categorical_focal_crossentropy.html">loss_categorical_focal_crossentropy</a>()</code> <br><code><a href="loss_categorical_generalized_cross_entropy.html">loss_categorical_generalized_cross_entropy</a>()</code> <br><code><a href="loss_categorical_hinge.html">loss_categorical_hinge</a>()</code> <br><code><a href="loss_circle.html">loss_circle</a>()</code> <br><code><a href="loss_cosine_similarity.html">loss_cosine_similarity</a>()</code> <br><code><a href="loss_ctc.html">loss_ctc</a>()</code> <br><code><a href="loss_dice.html">loss_dice</a>()</code> <br><code><a href="loss_hinge.html">loss_hinge</a>()</code> <br><code><a href="loss_huber.html">loss_huber</a>()</code> <br><code><a href="loss_kl_divergence.html">loss_kl_divergence</a>()</code> <br><code><a href="loss_log_cosh.html">loss_log_cosh</a>()</code> <br><code><a href="loss_mean_absolute_error.html">loss_mean_absolute_error</a>()</code> <br><code><a href="loss_mean_absolute_percentage_error.html">loss_mean_absolute_percentage_error</a>()</code> <br><code><a href="loss_mean_squared_error.html">loss_mean_squared_error</a>()</code> <br><code><a href="loss_mean_squared_logarithmic_error.html">loss_mean_squared_logarithmic_error</a>()</code> <br><code><a href="loss_poisson.html">loss_poisson</a>()</code> <br><code><a href="loss_sparse_categorical_crossentropy.html">loss_sparse_categorical_crossentropy</a>()</code> <br><code><a href="loss_squared_hinge.html">loss_squared_hinge</a>()</code> <br><code><a href="loss_tversky.html">loss_tversky</a>()</code> <br><code><a href="metric_binary_focal_crossentropy.html">metric_binary_focal_crossentropy</a>()</code> <br><code><a href="metric_categorical_crossentropy.html">metric_categorical_crossentropy</a>()</code> <br><code><a href="metric_categorical_focal_crossentropy.html">metric_categorical_focal_crossentropy</a>()</code> <br><code><a href="metric_categorical_hinge.html">metric_categorical_hinge</a>()</code> <br><code><a href="metric_hinge.html">metric_hinge</a>()</code> <br><code><a href="metric_huber.html">metric_huber</a>()</code> <br><code><a href="metric_kl_divergence.html">metric_kl_divergence</a>()</code> <br><code><a href="metric_log_cosh.html">metric_log_cosh</a>()</code> <br><code><a href="metric_mean_absolute_error.html">metric_mean_absolute_error</a>()</code> <br><code><a href="metric_mean_absolute_percentage_error.html">metric_mean_absolute_percentage_error</a>()</code> <br><code><a href="metric_mean_squared_error.html">metric_mean_squared_error</a>()</code> <br><code><a href="metric_mean_squared_logarithmic_error.html">metric_mean_squared_logarithmic_error</a>()</code> <br><code><a href="metric_poisson.html">metric_poisson</a>()</code> <br><code><a href="metric_sparse_categorical_crossentropy.html">metric_sparse_categorical_crossentropy</a>()</code> <br><code><a href="metric_squared_hinge.html">metric_squared_hinge</a>()</code> <br></p>
 <p>Other metrics: <br><code><a href="Metric.html">Metric</a>()</code> <br><code><a href="custom_metric.html">custom_metric</a>()</code> <br><code><a href="metric_auc.html">metric_auc</a>()</code> <br><code><a href="metric_binary_accuracy.html">metric_binary_accuracy</a>()</code> <br><code><a href="metric_binary_focal_crossentropy.html">metric_binary_focal_crossentropy</a>()</code> <br><code><a href="metric_binary_iou.html">metric_binary_iou</a>()</code> <br><code><a href="metric_categorical_accuracy.html">metric_categorical_accuracy</a>()</code> <br><code><a href="metric_categorical_crossentropy.html">metric_categorical_crossentropy</a>()</code> <br><code><a href="metric_categorical_focal_crossentropy.html">metric_categorical_focal_crossentropy</a>()</code> <br><code><a href="metric_categorical_hinge.html">metric_categorical_hinge</a>()</code> <br><code><a href="metric_concordance_correlation.html">metric_concordance_correlation</a>()</code> <br><code><a href="metric_cosine_similarity.html">metric_cosine_similarity</a>()</code> <br><code><a href="metric_f1_score.html">metric_f1_score</a>()</code> <br><code><a href="metric_false_negatives.html">metric_false_negatives</a>()</code> <br><code><a href="metric_false_positives.html">metric_false_positives</a>()</code> <br><code><a href="metric_fbeta_score.html">metric_fbeta_score</a>()</code> <br><code><a href="metric_hinge.html">metric_hinge</a>()</code> <br><code><a href="metric_huber.html">metric_huber</a>()</code> <br><code><a href="metric_iou.html">metric_iou</a>()</code> <br><code><a href="metric_kl_divergence.html">metric_kl_divergence</a>()</code> <br><code><a href="metric_log_cosh.html">metric_log_cosh</a>()</code> <br><code><a href="metric_log_cosh_error.html">metric_log_cosh_error</a>()</code> <br><code><a href="metric_mean.html">metric_mean</a>()</code> <br><code><a href="metric_mean_absolute_error.html">metric_mean_absolute_error</a>()</code> <br><code><a href="metric_mean_absolute_percentage_error.html">metric_mean_absolute_percentage_error</a>()</code> <br><code><a href="metric_mean_iou.html">metric_mean_iou</a>()</code> <br><code><a href="metric_mean_squared_error.html">metric_mean_squared_error</a>()</code> <br><code><a href="metric_mean_squared_logarithmic_error.html">metric_mean_squared_logarithmic_error</a>()</code> <br><code><a href="metric_mean_wrapper.html">metric_mean_wrapper</a>()</code> <br><code><a href="metric_one_hot_iou.html">metric_one_hot_iou</a>()</code> <br><code><a href="metric_one_hot_mean_iou.html">metric_one_hot_mean_iou</a>()</code> <br><code><a href="metric_pearson_correlation.html">metric_pearson_correlation</a>()</code> <br><code><a href="metric_poisson.html">metric_poisson</a>()</code> <br><code><a href="metric_precision.html">metric_precision</a>()</code> <br><code><a href="metric_precision_at_recall.html">metric_precision_at_recall</a>()</code> <br><code><a href="metric_r2_score.html">metric_r2_score</a>()</code> <br><code><a href="metric_recall.html">metric_recall</a>()</code> <br><code><a href="metric_recall_at_precision.html">metric_recall_at_precision</a>()</code> <br><code><a href="metric_root_mean_squared_error.html">metric_root_mean_squared_error</a>()</code> <br><code><a href="metric_sensitivity_at_specificity.html">metric_sensitivity_at_specificity</a>()</code> <br><code><a href="metric_sparse_categorical_accuracy.html">metric_sparse_categorical_accuracy</a>()</code> <br><code><a href="metric_sparse_categorical_crossentropy.html">metric_sparse_categorical_crossentropy</a>()</code> <br><code><a href="metric_sparse_top_k_categorical_accuracy.html">metric_sparse_top_k_categorical_accuracy</a>()</code> <br><code><a href="metric_specificity_at_sensitivity.html">metric_specificity_at_sensitivity</a>()</code> <br><code><a href="metric_squared_hinge.html">metric_squared_hinge</a>()</code> <br><code><a href="metric_sum.html">metric_sum</a>()</code> <br><code><a href="metric_top_k_categorical_accuracy.html">metric_top_k_categorical_accuracy</a>()</code> <br><code><a href="metric_true_negatives.html">metric_true_negatives</a>()</code> <br><code><a href="metric_true_positives.html">metric_true_positives</a>()</code> <br></p>
 <p>Other probabilistic metrics: <br><code><a href="metric_categorical_crossentropy.html">metric_categorical_crossentropy</a>()</code> <br><code><a href="metric_kl_divergence.html">metric_kl_divergence</a>()</code> <br><code><a href="metric_poisson.html">metric_poisson</a>()</code> <br><code><a href="metric_sparse_categorical_crossentropy.html">metric_sparse_categorical_crossentropy</a>()</code> <br></p></div>
     </div>
@@ -162,7 +162,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/metric_binary_focal_crossentropy.html b/docs/dev/reference/metric_binary_focal_crossentropy.html
index 4e13b2a29..e4a85eb5b 100644
--- a/docs/dev/reference/metric_binary_focal_crossentropy.html
+++ b/docs/dev/reference/metric_binary_focal_crossentropy.html
@@ -30,7 +30,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -154,16 +154,11 @@ <h2 id="value">Value<a class="anchor" aria-label="anchor" href="#value"></a></h2
     </div>
     <div class="section level2">
     <h2 id="examples">Examples<a class="anchor" aria-label="anchor" href="#examples"></a></h2>
-    <p></p><div class="sourceCode r"><pre><code><span><span class="va">y_true</span> <span class="op">&lt;-</span> <span class="fu"><a href="https://rdrr.io/r/base/cbind.html" class="external-link">rbind</a></span><span class="op">(</span><span class="fu"><a href="https://rdrr.io/r/base/c.html" class="external-link">c</a></span><span class="op">(</span><span class="fl">0</span>, <span class="fl">1</span><span class="op">)</span>, <span class="fu"><a href="https://rdrr.io/r/base/c.html" class="external-link">c</a></span><span class="op">(</span><span class="fl">0</span>, <span class="fl">0</span><span class="op">)</span><span class="op">)</span></span>
-<span><span class="va">y_pred</span> <span class="op">&lt;-</span> <span class="fu"><a href="https://rdrr.io/r/base/cbind.html" class="external-link">rbind</a></span><span class="op">(</span><span class="fu"><a href="https://rdrr.io/r/base/c.html" class="external-link">c</a></span><span class="op">(</span><span class="fl">0.6</span>, <span class="fl">0.4</span><span class="op">)</span>, <span class="fu"><a href="https://rdrr.io/r/base/c.html" class="external-link">c</a></span><span class="op">(</span><span class="fl">0.4</span>, <span class="fl">0.6</span><span class="op">)</span><span class="op">)</span></span>
-<span><span class="va">loss</span> <span class="op">&lt;-</span> <span class="fu"><a href="../reference/loss_binary_focal_crossentropy.html">loss_binary_focal_crossentropy</a></span><span class="op">(</span><span class="va">y_true</span>, <span class="va">y_pred</span>, gamma<span class="op">=</span><span class="fl">2</span><span class="op">)</span></span>
-<span><span class="va">loss</span></span></code></pre><p></p></div>
-<p></p><div class="sourceCode"><pre><code><span id="cb1-1"><a href="#cb1-1" tabindex="-1"></a><span class="do">## tf.Tensor([0.32986468 0.20579839], shape=(2), dtype=float32)</span></span>
-<span id="cb1-2"><a href="#cb1-2" tabindex="-1"></a></span></code></pre><p></p></div>
+    <p></p><div class="sourceCode {r}"><pre><code></code></pre><p></p></div>
     </div>
     <div class="section level2">
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
-    <div class="dont-index"><p>Other losses: <br><code><a href="Loss.html">Loss</a>()</code> <br><code><a href="loss_binary_crossentropy.html">loss_binary_crossentropy</a>()</code> <br><code><a href="loss_binary_focal_crossentropy.html">loss_binary_focal_crossentropy</a>()</code> <br><code><a href="loss_categorical_crossentropy.html">loss_categorical_crossentropy</a>()</code> <br><code><a href="loss_categorical_focal_crossentropy.html">loss_categorical_focal_crossentropy</a>()</code> <br><code><a href="loss_categorical_hinge.html">loss_categorical_hinge</a>()</code> <br><code><a href="loss_circle.html">loss_circle</a>()</code> <br><code><a href="loss_cosine_similarity.html">loss_cosine_similarity</a>()</code> <br><code><a href="loss_ctc.html">loss_ctc</a>()</code> <br><code><a href="loss_dice.html">loss_dice</a>()</code> <br><code><a href="loss_hinge.html">loss_hinge</a>()</code> <br><code><a href="loss_huber.html">loss_huber</a>()</code> <br><code><a href="loss_kl_divergence.html">loss_kl_divergence</a>()</code> <br><code><a href="loss_log_cosh.html">loss_log_cosh</a>()</code> <br><code><a href="loss_mean_absolute_error.html">loss_mean_absolute_error</a>()</code> <br><code><a href="loss_mean_absolute_percentage_error.html">loss_mean_absolute_percentage_error</a>()</code> <br><code><a href="loss_mean_squared_error.html">loss_mean_squared_error</a>()</code> <br><code><a href="loss_mean_squared_logarithmic_error.html">loss_mean_squared_logarithmic_error</a>()</code> <br><code><a href="loss_poisson.html">loss_poisson</a>()</code> <br><code><a href="loss_sparse_categorical_crossentropy.html">loss_sparse_categorical_crossentropy</a>()</code> <br><code><a href="loss_squared_hinge.html">loss_squared_hinge</a>()</code> <br><code><a href="loss_tversky.html">loss_tversky</a>()</code> <br><code><a href="metric_binary_crossentropy.html">metric_binary_crossentropy</a>()</code> <br><code><a href="metric_categorical_crossentropy.html">metric_categorical_crossentropy</a>()</code> <br><code><a href="metric_categorical_focal_crossentropy.html">metric_categorical_focal_crossentropy</a>()</code> <br><code><a href="metric_categorical_hinge.html">metric_categorical_hinge</a>()</code> <br><code><a href="metric_hinge.html">metric_hinge</a>()</code> <br><code><a href="metric_huber.html">metric_huber</a>()</code> <br><code><a href="metric_kl_divergence.html">metric_kl_divergence</a>()</code> <br><code><a href="metric_log_cosh.html">metric_log_cosh</a>()</code> <br><code><a href="metric_mean_absolute_error.html">metric_mean_absolute_error</a>()</code> <br><code><a href="metric_mean_absolute_percentage_error.html">metric_mean_absolute_percentage_error</a>()</code> <br><code><a href="metric_mean_squared_error.html">metric_mean_squared_error</a>()</code> <br><code><a href="metric_mean_squared_logarithmic_error.html">metric_mean_squared_logarithmic_error</a>()</code> <br><code><a href="metric_poisson.html">metric_poisson</a>()</code> <br><code><a href="metric_sparse_categorical_crossentropy.html">metric_sparse_categorical_crossentropy</a>()</code> <br><code><a href="metric_squared_hinge.html">metric_squared_hinge</a>()</code> <br></p>
+    <div class="dont-index"><p>Other losses: <br><code><a href="Loss.html">Loss</a>()</code> <br><code><a href="loss_binary_crossentropy.html">loss_binary_crossentropy</a>()</code> <br><code><a href="loss_binary_focal_crossentropy.html">loss_binary_focal_crossentropy</a>()</code> <br><code><a href="loss_categorical_crossentropy.html">loss_categorical_crossentropy</a>()</code> <br><code><a href="loss_categorical_focal_crossentropy.html">loss_categorical_focal_crossentropy</a>()</code> <br><code><a href="loss_categorical_generalized_cross_entropy.html">loss_categorical_generalized_cross_entropy</a>()</code> <br><code><a href="loss_categorical_hinge.html">loss_categorical_hinge</a>()</code> <br><code><a href="loss_circle.html">loss_circle</a>()</code> <br><code><a href="loss_cosine_similarity.html">loss_cosine_similarity</a>()</code> <br><code><a href="loss_ctc.html">loss_ctc</a>()</code> <br><code><a href="loss_dice.html">loss_dice</a>()</code> <br><code><a href="loss_hinge.html">loss_hinge</a>()</code> <br><code><a href="loss_huber.html">loss_huber</a>()</code> <br><code><a href="loss_kl_divergence.html">loss_kl_divergence</a>()</code> <br><code><a href="loss_log_cosh.html">loss_log_cosh</a>()</code> <br><code><a href="loss_mean_absolute_error.html">loss_mean_absolute_error</a>()</code> <br><code><a href="loss_mean_absolute_percentage_error.html">loss_mean_absolute_percentage_error</a>()</code> <br><code><a href="loss_mean_squared_error.html">loss_mean_squared_error</a>()</code> <br><code><a href="loss_mean_squared_logarithmic_error.html">loss_mean_squared_logarithmic_error</a>()</code> <br><code><a href="loss_poisson.html">loss_poisson</a>()</code> <br><code><a href="loss_sparse_categorical_crossentropy.html">loss_sparse_categorical_crossentropy</a>()</code> <br><code><a href="loss_squared_hinge.html">loss_squared_hinge</a>()</code> <br><code><a href="loss_tversky.html">loss_tversky</a>()</code> <br><code><a href="metric_binary_crossentropy.html">metric_binary_crossentropy</a>()</code> <br><code><a href="metric_categorical_crossentropy.html">metric_categorical_crossentropy</a>()</code> <br><code><a href="metric_categorical_focal_crossentropy.html">metric_categorical_focal_crossentropy</a>()</code> <br><code><a href="metric_categorical_hinge.html">metric_categorical_hinge</a>()</code> <br><code><a href="metric_hinge.html">metric_hinge</a>()</code> <br><code><a href="metric_huber.html">metric_huber</a>()</code> <br><code><a href="metric_kl_divergence.html">metric_kl_divergence</a>()</code> <br><code><a href="metric_log_cosh.html">metric_log_cosh</a>()</code> <br><code><a href="metric_mean_absolute_error.html">metric_mean_absolute_error</a>()</code> <br><code><a href="metric_mean_absolute_percentage_error.html">metric_mean_absolute_percentage_error</a>()</code> <br><code><a href="metric_mean_squared_error.html">metric_mean_squared_error</a>()</code> <br><code><a href="metric_mean_squared_logarithmic_error.html">metric_mean_squared_logarithmic_error</a>()</code> <br><code><a href="metric_poisson.html">metric_poisson</a>()</code> <br><code><a href="metric_sparse_categorical_crossentropy.html">metric_sparse_categorical_crossentropy</a>()</code> <br><code><a href="metric_squared_hinge.html">metric_squared_hinge</a>()</code> <br></p>
 <p>Other metrics: <br><code><a href="Metric.html">Metric</a>()</code> <br><code><a href="custom_metric.html">custom_metric</a>()</code> <br><code><a href="metric_auc.html">metric_auc</a>()</code> <br><code><a href="metric_binary_accuracy.html">metric_binary_accuracy</a>()</code> <br><code><a href="metric_binary_crossentropy.html">metric_binary_crossentropy</a>()</code> <br><code><a href="metric_binary_iou.html">metric_binary_iou</a>()</code> <br><code><a href="metric_categorical_accuracy.html">metric_categorical_accuracy</a>()</code> <br><code><a href="metric_categorical_crossentropy.html">metric_categorical_crossentropy</a>()</code> <br><code><a href="metric_categorical_focal_crossentropy.html">metric_categorical_focal_crossentropy</a>()</code> <br><code><a href="metric_categorical_hinge.html">metric_categorical_hinge</a>()</code> <br><code><a href="metric_concordance_correlation.html">metric_concordance_correlation</a>()</code> <br><code><a href="metric_cosine_similarity.html">metric_cosine_similarity</a>()</code> <br><code><a href="metric_f1_score.html">metric_f1_score</a>()</code> <br><code><a href="metric_false_negatives.html">metric_false_negatives</a>()</code> <br><code><a href="metric_false_positives.html">metric_false_positives</a>()</code> <br><code><a href="metric_fbeta_score.html">metric_fbeta_score</a>()</code> <br><code><a href="metric_hinge.html">metric_hinge</a>()</code> <br><code><a href="metric_huber.html">metric_huber</a>()</code> <br><code><a href="metric_iou.html">metric_iou</a>()</code> <br><code><a href="metric_kl_divergence.html">metric_kl_divergence</a>()</code> <br><code><a href="metric_log_cosh.html">metric_log_cosh</a>()</code> <br><code><a href="metric_log_cosh_error.html">metric_log_cosh_error</a>()</code> <br><code><a href="metric_mean.html">metric_mean</a>()</code> <br><code><a href="metric_mean_absolute_error.html">metric_mean_absolute_error</a>()</code> <br><code><a href="metric_mean_absolute_percentage_error.html">metric_mean_absolute_percentage_error</a>()</code> <br><code><a href="metric_mean_iou.html">metric_mean_iou</a>()</code> <br><code><a href="metric_mean_squared_error.html">metric_mean_squared_error</a>()</code> <br><code><a href="metric_mean_squared_logarithmic_error.html">metric_mean_squared_logarithmic_error</a>()</code> <br><code><a href="metric_mean_wrapper.html">metric_mean_wrapper</a>()</code> <br><code><a href="metric_one_hot_iou.html">metric_one_hot_iou</a>()</code> <br><code><a href="metric_one_hot_mean_iou.html">metric_one_hot_mean_iou</a>()</code> <br><code><a href="metric_pearson_correlation.html">metric_pearson_correlation</a>()</code> <br><code><a href="metric_poisson.html">metric_poisson</a>()</code> <br><code><a href="metric_precision.html">metric_precision</a>()</code> <br><code><a href="metric_precision_at_recall.html">metric_precision_at_recall</a>()</code> <br><code><a href="metric_r2_score.html">metric_r2_score</a>()</code> <br><code><a href="metric_recall.html">metric_recall</a>()</code> <br><code><a href="metric_recall_at_precision.html">metric_recall_at_precision</a>()</code> <br><code><a href="metric_root_mean_squared_error.html">metric_root_mean_squared_error</a>()</code> <br><code><a href="metric_sensitivity_at_specificity.html">metric_sensitivity_at_specificity</a>()</code> <br><code><a href="metric_sparse_categorical_accuracy.html">metric_sparse_categorical_accuracy</a>()</code> <br><code><a href="metric_sparse_categorical_crossentropy.html">metric_sparse_categorical_crossentropy</a>()</code> <br><code><a href="metric_sparse_top_k_categorical_accuracy.html">metric_sparse_top_k_categorical_accuracy</a>()</code> <br><code><a href="metric_specificity_at_sensitivity.html">metric_specificity_at_sensitivity</a>()</code> <br><code><a href="metric_squared_hinge.html">metric_squared_hinge</a>()</code> <br><code><a href="metric_sum.html">metric_sum</a>()</code> <br><code><a href="metric_top_k_categorical_accuracy.html">metric_top_k_categorical_accuracy</a>()</code> <br><code><a href="metric_true_negatives.html">metric_true_negatives</a>()</code> <br><code><a href="metric_true_positives.html">metric_true_positives</a>()</code> <br></p></div>
     </div>
 
@@ -176,7 +171,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/metric_binary_iou.html b/docs/dev/reference/metric_binary_iou.html
index a73361335..ddcf12c38 100644
--- a/docs/dev/reference/metric_binary_iou.html
+++ b/docs/dev/reference/metric_binary_iou.html
@@ -38,7 +38,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -199,7 +199,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/metric_categorical_accuracy.html b/docs/dev/reference/metric_categorical_accuracy.html
index 754313294..6c9ad95fd 100644
--- a/docs/dev/reference/metric_categorical_accuracy.html
+++ b/docs/dev/reference/metric_categorical_accuracy.html
@@ -28,7 +28,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -169,7 +169,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/metric_categorical_crossentropy.html b/docs/dev/reference/metric_categorical_crossentropy.html
index 5ac78df4d..e7693a645 100644
--- a/docs/dev/reference/metric_categorical_crossentropy.html
+++ b/docs/dev/reference/metric_categorical_crossentropy.html
@@ -14,7 +14,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -166,7 +166,7 @@ <h2 id="examples">Examples<a class="anchor" aria-label="anchor" href="#examples"
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
     <div class="dont-index">
 <ul><li><p><a href="https://keras.io/api/metrics/probabilistic_metrics#categoricalcrossentropy-class" class="external-link">https://keras.io/api/metrics/probabilistic_metrics#categoricalcrossentropy-class</a></p></li>
-</ul><p>Other losses: <br><code><a href="Loss.html">Loss</a>()</code> <br><code><a href="loss_binary_crossentropy.html">loss_binary_crossentropy</a>()</code> <br><code><a href="loss_binary_focal_crossentropy.html">loss_binary_focal_crossentropy</a>()</code> <br><code><a href="loss_categorical_crossentropy.html">loss_categorical_crossentropy</a>()</code> <br><code><a href="loss_categorical_focal_crossentropy.html">loss_categorical_focal_crossentropy</a>()</code> <br><code><a href="loss_categorical_hinge.html">loss_categorical_hinge</a>()</code> <br><code><a href="loss_circle.html">loss_circle</a>()</code> <br><code><a href="loss_cosine_similarity.html">loss_cosine_similarity</a>()</code> <br><code><a href="loss_ctc.html">loss_ctc</a>()</code> <br><code><a href="loss_dice.html">loss_dice</a>()</code> <br><code><a href="loss_hinge.html">loss_hinge</a>()</code> <br><code><a href="loss_huber.html">loss_huber</a>()</code> <br><code><a href="loss_kl_divergence.html">loss_kl_divergence</a>()</code> <br><code><a href="loss_log_cosh.html">loss_log_cosh</a>()</code> <br><code><a href="loss_mean_absolute_error.html">loss_mean_absolute_error</a>()</code> <br><code><a href="loss_mean_absolute_percentage_error.html">loss_mean_absolute_percentage_error</a>()</code> <br><code><a href="loss_mean_squared_error.html">loss_mean_squared_error</a>()</code> <br><code><a href="loss_mean_squared_logarithmic_error.html">loss_mean_squared_logarithmic_error</a>()</code> <br><code><a href="loss_poisson.html">loss_poisson</a>()</code> <br><code><a href="loss_sparse_categorical_crossentropy.html">loss_sparse_categorical_crossentropy</a>()</code> <br><code><a href="loss_squared_hinge.html">loss_squared_hinge</a>()</code> <br><code><a href="loss_tversky.html">loss_tversky</a>()</code> <br><code><a href="metric_binary_crossentropy.html">metric_binary_crossentropy</a>()</code> <br><code><a href="metric_binary_focal_crossentropy.html">metric_binary_focal_crossentropy</a>()</code> <br><code><a href="metric_categorical_focal_crossentropy.html">metric_categorical_focal_crossentropy</a>()</code> <br><code><a href="metric_categorical_hinge.html">metric_categorical_hinge</a>()</code> <br><code><a href="metric_hinge.html">metric_hinge</a>()</code> <br><code><a href="metric_huber.html">metric_huber</a>()</code> <br><code><a href="metric_kl_divergence.html">metric_kl_divergence</a>()</code> <br><code><a href="metric_log_cosh.html">metric_log_cosh</a>()</code> <br><code><a href="metric_mean_absolute_error.html">metric_mean_absolute_error</a>()</code> <br><code><a href="metric_mean_absolute_percentage_error.html">metric_mean_absolute_percentage_error</a>()</code> <br><code><a href="metric_mean_squared_error.html">metric_mean_squared_error</a>()</code> <br><code><a href="metric_mean_squared_logarithmic_error.html">metric_mean_squared_logarithmic_error</a>()</code> <br><code><a href="metric_poisson.html">metric_poisson</a>()</code> <br><code><a href="metric_sparse_categorical_crossentropy.html">metric_sparse_categorical_crossentropy</a>()</code> <br><code><a href="metric_squared_hinge.html">metric_squared_hinge</a>()</code> <br></p>
+</ul><p>Other losses: <br><code><a href="Loss.html">Loss</a>()</code> <br><code><a href="loss_binary_crossentropy.html">loss_binary_crossentropy</a>()</code> <br><code><a href="loss_binary_focal_crossentropy.html">loss_binary_focal_crossentropy</a>()</code> <br><code><a href="loss_categorical_crossentropy.html">loss_categorical_crossentropy</a>()</code> <br><code><a href="loss_categorical_focal_crossentropy.html">loss_categorical_focal_crossentropy</a>()</code> <br><code><a href="loss_categorical_generalized_cross_entropy.html">loss_categorical_generalized_cross_entropy</a>()</code> <br><code><a href="loss_categorical_hinge.html">loss_categorical_hinge</a>()</code> <br><code><a href="loss_circle.html">loss_circle</a>()</code> <br><code><a href="loss_cosine_similarity.html">loss_cosine_similarity</a>()</code> <br><code><a href="loss_ctc.html">loss_ctc</a>()</code> <br><code><a href="loss_dice.html">loss_dice</a>()</code> <br><code><a href="loss_hinge.html">loss_hinge</a>()</code> <br><code><a href="loss_huber.html">loss_huber</a>()</code> <br><code><a href="loss_kl_divergence.html">loss_kl_divergence</a>()</code> <br><code><a href="loss_log_cosh.html">loss_log_cosh</a>()</code> <br><code><a href="loss_mean_absolute_error.html">loss_mean_absolute_error</a>()</code> <br><code><a href="loss_mean_absolute_percentage_error.html">loss_mean_absolute_percentage_error</a>()</code> <br><code><a href="loss_mean_squared_error.html">loss_mean_squared_error</a>()</code> <br><code><a href="loss_mean_squared_logarithmic_error.html">loss_mean_squared_logarithmic_error</a>()</code> <br><code><a href="loss_poisson.html">loss_poisson</a>()</code> <br><code><a href="loss_sparse_categorical_crossentropy.html">loss_sparse_categorical_crossentropy</a>()</code> <br><code><a href="loss_squared_hinge.html">loss_squared_hinge</a>()</code> <br><code><a href="loss_tversky.html">loss_tversky</a>()</code> <br><code><a href="metric_binary_crossentropy.html">metric_binary_crossentropy</a>()</code> <br><code><a href="metric_binary_focal_crossentropy.html">metric_binary_focal_crossentropy</a>()</code> <br><code><a href="metric_categorical_focal_crossentropy.html">metric_categorical_focal_crossentropy</a>()</code> <br><code><a href="metric_categorical_hinge.html">metric_categorical_hinge</a>()</code> <br><code><a href="metric_hinge.html">metric_hinge</a>()</code> <br><code><a href="metric_huber.html">metric_huber</a>()</code> <br><code><a href="metric_kl_divergence.html">metric_kl_divergence</a>()</code> <br><code><a href="metric_log_cosh.html">metric_log_cosh</a>()</code> <br><code><a href="metric_mean_absolute_error.html">metric_mean_absolute_error</a>()</code> <br><code><a href="metric_mean_absolute_percentage_error.html">metric_mean_absolute_percentage_error</a>()</code> <br><code><a href="metric_mean_squared_error.html">metric_mean_squared_error</a>()</code> <br><code><a href="metric_mean_squared_logarithmic_error.html">metric_mean_squared_logarithmic_error</a>()</code> <br><code><a href="metric_poisson.html">metric_poisson</a>()</code> <br><code><a href="metric_sparse_categorical_crossentropy.html">metric_sparse_categorical_crossentropy</a>()</code> <br><code><a href="metric_squared_hinge.html">metric_squared_hinge</a>()</code> <br></p>
 <p>Other metrics: <br><code><a href="Metric.html">Metric</a>()</code> <br><code><a href="custom_metric.html">custom_metric</a>()</code> <br><code><a href="metric_auc.html">metric_auc</a>()</code> <br><code><a href="metric_binary_accuracy.html">metric_binary_accuracy</a>()</code> <br><code><a href="metric_binary_crossentropy.html">metric_binary_crossentropy</a>()</code> <br><code><a href="metric_binary_focal_crossentropy.html">metric_binary_focal_crossentropy</a>()</code> <br><code><a href="metric_binary_iou.html">metric_binary_iou</a>()</code> <br><code><a href="metric_categorical_accuracy.html">metric_categorical_accuracy</a>()</code> <br><code><a href="metric_categorical_focal_crossentropy.html">metric_categorical_focal_crossentropy</a>()</code> <br><code><a href="metric_categorical_hinge.html">metric_categorical_hinge</a>()</code> <br><code><a href="metric_concordance_correlation.html">metric_concordance_correlation</a>()</code> <br><code><a href="metric_cosine_similarity.html">metric_cosine_similarity</a>()</code> <br><code><a href="metric_f1_score.html">metric_f1_score</a>()</code> <br><code><a href="metric_false_negatives.html">metric_false_negatives</a>()</code> <br><code><a href="metric_false_positives.html">metric_false_positives</a>()</code> <br><code><a href="metric_fbeta_score.html">metric_fbeta_score</a>()</code> <br><code><a href="metric_hinge.html">metric_hinge</a>()</code> <br><code><a href="metric_huber.html">metric_huber</a>()</code> <br><code><a href="metric_iou.html">metric_iou</a>()</code> <br><code><a href="metric_kl_divergence.html">metric_kl_divergence</a>()</code> <br><code><a href="metric_log_cosh.html">metric_log_cosh</a>()</code> <br><code><a href="metric_log_cosh_error.html">metric_log_cosh_error</a>()</code> <br><code><a href="metric_mean.html">metric_mean</a>()</code> <br><code><a href="metric_mean_absolute_error.html">metric_mean_absolute_error</a>()</code> <br><code><a href="metric_mean_absolute_percentage_error.html">metric_mean_absolute_percentage_error</a>()</code> <br><code><a href="metric_mean_iou.html">metric_mean_iou</a>()</code> <br><code><a href="metric_mean_squared_error.html">metric_mean_squared_error</a>()</code> <br><code><a href="metric_mean_squared_logarithmic_error.html">metric_mean_squared_logarithmic_error</a>()</code> <br><code><a href="metric_mean_wrapper.html">metric_mean_wrapper</a>()</code> <br><code><a href="metric_one_hot_iou.html">metric_one_hot_iou</a>()</code> <br><code><a href="metric_one_hot_mean_iou.html">metric_one_hot_mean_iou</a>()</code> <br><code><a href="metric_pearson_correlation.html">metric_pearson_correlation</a>()</code> <br><code><a href="metric_poisson.html">metric_poisson</a>()</code> <br><code><a href="metric_precision.html">metric_precision</a>()</code> <br><code><a href="metric_precision_at_recall.html">metric_precision_at_recall</a>()</code> <br><code><a href="metric_r2_score.html">metric_r2_score</a>()</code> <br><code><a href="metric_recall.html">metric_recall</a>()</code> <br><code><a href="metric_recall_at_precision.html">metric_recall_at_precision</a>()</code> <br><code><a href="metric_root_mean_squared_error.html">metric_root_mean_squared_error</a>()</code> <br><code><a href="metric_sensitivity_at_specificity.html">metric_sensitivity_at_specificity</a>()</code> <br><code><a href="metric_sparse_categorical_accuracy.html">metric_sparse_categorical_accuracy</a>()</code> <br><code><a href="metric_sparse_categorical_crossentropy.html">metric_sparse_categorical_crossentropy</a>()</code> <br><code><a href="metric_sparse_top_k_categorical_accuracy.html">metric_sparse_top_k_categorical_accuracy</a>()</code> <br><code><a href="metric_specificity_at_sensitivity.html">metric_specificity_at_sensitivity</a>()</code> <br><code><a href="metric_squared_hinge.html">metric_squared_hinge</a>()</code> <br><code><a href="metric_sum.html">metric_sum</a>()</code> <br><code><a href="metric_top_k_categorical_accuracy.html">metric_top_k_categorical_accuracy</a>()</code> <br><code><a href="metric_true_negatives.html">metric_true_negatives</a>()</code> <br><code><a href="metric_true_positives.html">metric_true_positives</a>()</code> <br></p>
 <p>Other probabilistic metrics: <br><code><a href="metric_binary_crossentropy.html">metric_binary_crossentropy</a>()</code> <br><code><a href="metric_kl_divergence.html">metric_kl_divergence</a>()</code> <br><code><a href="metric_poisson.html">metric_poisson</a>()</code> <br><code><a href="metric_sparse_categorical_crossentropy.html">metric_sparse_categorical_crossentropy</a>()</code> <br></p></div>
     </div>
@@ -180,7 +180,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/metric_categorical_focal_crossentropy.html b/docs/dev/reference/metric_categorical_focal_crossentropy.html
index 4197130aa..17e7ea157 100644
--- a/docs/dev/reference/metric_categorical_focal_crossentropy.html
+++ b/docs/dev/reference/metric_categorical_focal_crossentropy.html
@@ -8,7 +8,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -128,7 +128,7 @@ <h2 id="examples">Examples<a class="anchor" aria-label="anchor" href="#examples"
     </div>
     <div class="section level2">
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
-    <div class="dont-index"><p>Other losses: <br><code><a href="Loss.html">Loss</a>()</code> <br><code><a href="loss_binary_crossentropy.html">loss_binary_crossentropy</a>()</code> <br><code><a href="loss_binary_focal_crossentropy.html">loss_binary_focal_crossentropy</a>()</code> <br><code><a href="loss_categorical_crossentropy.html">loss_categorical_crossentropy</a>()</code> <br><code><a href="loss_categorical_focal_crossentropy.html">loss_categorical_focal_crossentropy</a>()</code> <br><code><a href="loss_categorical_hinge.html">loss_categorical_hinge</a>()</code> <br><code><a href="loss_circle.html">loss_circle</a>()</code> <br><code><a href="loss_cosine_similarity.html">loss_cosine_similarity</a>()</code> <br><code><a href="loss_ctc.html">loss_ctc</a>()</code> <br><code><a href="loss_dice.html">loss_dice</a>()</code> <br><code><a href="loss_hinge.html">loss_hinge</a>()</code> <br><code><a href="loss_huber.html">loss_huber</a>()</code> <br><code><a href="loss_kl_divergence.html">loss_kl_divergence</a>()</code> <br><code><a href="loss_log_cosh.html">loss_log_cosh</a>()</code> <br><code><a href="loss_mean_absolute_error.html">loss_mean_absolute_error</a>()</code> <br><code><a href="loss_mean_absolute_percentage_error.html">loss_mean_absolute_percentage_error</a>()</code> <br><code><a href="loss_mean_squared_error.html">loss_mean_squared_error</a>()</code> <br><code><a href="loss_mean_squared_logarithmic_error.html">loss_mean_squared_logarithmic_error</a>()</code> <br><code><a href="loss_poisson.html">loss_poisson</a>()</code> <br><code><a href="loss_sparse_categorical_crossentropy.html">loss_sparse_categorical_crossentropy</a>()</code> <br><code><a href="loss_squared_hinge.html">loss_squared_hinge</a>()</code> <br><code><a href="loss_tversky.html">loss_tversky</a>()</code> <br><code><a href="metric_binary_crossentropy.html">metric_binary_crossentropy</a>()</code> <br><code><a href="metric_binary_focal_crossentropy.html">metric_binary_focal_crossentropy</a>()</code> <br><code><a href="metric_categorical_crossentropy.html">metric_categorical_crossentropy</a>()</code> <br><code><a href="metric_categorical_hinge.html">metric_categorical_hinge</a>()</code> <br><code><a href="metric_hinge.html">metric_hinge</a>()</code> <br><code><a href="metric_huber.html">metric_huber</a>()</code> <br><code><a href="metric_kl_divergence.html">metric_kl_divergence</a>()</code> <br><code><a href="metric_log_cosh.html">metric_log_cosh</a>()</code> <br><code><a href="metric_mean_absolute_error.html">metric_mean_absolute_error</a>()</code> <br><code><a href="metric_mean_absolute_percentage_error.html">metric_mean_absolute_percentage_error</a>()</code> <br><code><a href="metric_mean_squared_error.html">metric_mean_squared_error</a>()</code> <br><code><a href="metric_mean_squared_logarithmic_error.html">metric_mean_squared_logarithmic_error</a>()</code> <br><code><a href="metric_poisson.html">metric_poisson</a>()</code> <br><code><a href="metric_sparse_categorical_crossentropy.html">metric_sparse_categorical_crossentropy</a>()</code> <br><code><a href="metric_squared_hinge.html">metric_squared_hinge</a>()</code> <br></p>
+    <div class="dont-index"><p>Other losses: <br><code><a href="Loss.html">Loss</a>()</code> <br><code><a href="loss_binary_crossentropy.html">loss_binary_crossentropy</a>()</code> <br><code><a href="loss_binary_focal_crossentropy.html">loss_binary_focal_crossentropy</a>()</code> <br><code><a href="loss_categorical_crossentropy.html">loss_categorical_crossentropy</a>()</code> <br><code><a href="loss_categorical_focal_crossentropy.html">loss_categorical_focal_crossentropy</a>()</code> <br><code><a href="loss_categorical_generalized_cross_entropy.html">loss_categorical_generalized_cross_entropy</a>()</code> <br><code><a href="loss_categorical_hinge.html">loss_categorical_hinge</a>()</code> <br><code><a href="loss_circle.html">loss_circle</a>()</code> <br><code><a href="loss_cosine_similarity.html">loss_cosine_similarity</a>()</code> <br><code><a href="loss_ctc.html">loss_ctc</a>()</code> <br><code><a href="loss_dice.html">loss_dice</a>()</code> <br><code><a href="loss_hinge.html">loss_hinge</a>()</code> <br><code><a href="loss_huber.html">loss_huber</a>()</code> <br><code><a href="loss_kl_divergence.html">loss_kl_divergence</a>()</code> <br><code><a href="loss_log_cosh.html">loss_log_cosh</a>()</code> <br><code><a href="loss_mean_absolute_error.html">loss_mean_absolute_error</a>()</code> <br><code><a href="loss_mean_absolute_percentage_error.html">loss_mean_absolute_percentage_error</a>()</code> <br><code><a href="loss_mean_squared_error.html">loss_mean_squared_error</a>()</code> <br><code><a href="loss_mean_squared_logarithmic_error.html">loss_mean_squared_logarithmic_error</a>()</code> <br><code><a href="loss_poisson.html">loss_poisson</a>()</code> <br><code><a href="loss_sparse_categorical_crossentropy.html">loss_sparse_categorical_crossentropy</a>()</code> <br><code><a href="loss_squared_hinge.html">loss_squared_hinge</a>()</code> <br><code><a href="loss_tversky.html">loss_tversky</a>()</code> <br><code><a href="metric_binary_crossentropy.html">metric_binary_crossentropy</a>()</code> <br><code><a href="metric_binary_focal_crossentropy.html">metric_binary_focal_crossentropy</a>()</code> <br><code><a href="metric_categorical_crossentropy.html">metric_categorical_crossentropy</a>()</code> <br><code><a href="metric_categorical_hinge.html">metric_categorical_hinge</a>()</code> <br><code><a href="metric_hinge.html">metric_hinge</a>()</code> <br><code><a href="metric_huber.html">metric_huber</a>()</code> <br><code><a href="metric_kl_divergence.html">metric_kl_divergence</a>()</code> <br><code><a href="metric_log_cosh.html">metric_log_cosh</a>()</code> <br><code><a href="metric_mean_absolute_error.html">metric_mean_absolute_error</a>()</code> <br><code><a href="metric_mean_absolute_percentage_error.html">metric_mean_absolute_percentage_error</a>()</code> <br><code><a href="metric_mean_squared_error.html">metric_mean_squared_error</a>()</code> <br><code><a href="metric_mean_squared_logarithmic_error.html">metric_mean_squared_logarithmic_error</a>()</code> <br><code><a href="metric_poisson.html">metric_poisson</a>()</code> <br><code><a href="metric_sparse_categorical_crossentropy.html">metric_sparse_categorical_crossentropy</a>()</code> <br><code><a href="metric_squared_hinge.html">metric_squared_hinge</a>()</code> <br></p>
 <p>Other metrics: <br><code><a href="Metric.html">Metric</a>()</code> <br><code><a href="custom_metric.html">custom_metric</a>()</code> <br><code><a href="metric_auc.html">metric_auc</a>()</code> <br><code><a href="metric_binary_accuracy.html">metric_binary_accuracy</a>()</code> <br><code><a href="metric_binary_crossentropy.html">metric_binary_crossentropy</a>()</code> <br><code><a href="metric_binary_focal_crossentropy.html">metric_binary_focal_crossentropy</a>()</code> <br><code><a href="metric_binary_iou.html">metric_binary_iou</a>()</code> <br><code><a href="metric_categorical_accuracy.html">metric_categorical_accuracy</a>()</code> <br><code><a href="metric_categorical_crossentropy.html">metric_categorical_crossentropy</a>()</code> <br><code><a href="metric_categorical_hinge.html">metric_categorical_hinge</a>()</code> <br><code><a href="metric_concordance_correlation.html">metric_concordance_correlation</a>()</code> <br><code><a href="metric_cosine_similarity.html">metric_cosine_similarity</a>()</code> <br><code><a href="metric_f1_score.html">metric_f1_score</a>()</code> <br><code><a href="metric_false_negatives.html">metric_false_negatives</a>()</code> <br><code><a href="metric_false_positives.html">metric_false_positives</a>()</code> <br><code><a href="metric_fbeta_score.html">metric_fbeta_score</a>()</code> <br><code><a href="metric_hinge.html">metric_hinge</a>()</code> <br><code><a href="metric_huber.html">metric_huber</a>()</code> <br><code><a href="metric_iou.html">metric_iou</a>()</code> <br><code><a href="metric_kl_divergence.html">metric_kl_divergence</a>()</code> <br><code><a href="metric_log_cosh.html">metric_log_cosh</a>()</code> <br><code><a href="metric_log_cosh_error.html">metric_log_cosh_error</a>()</code> <br><code><a href="metric_mean.html">metric_mean</a>()</code> <br><code><a href="metric_mean_absolute_error.html">metric_mean_absolute_error</a>()</code> <br><code><a href="metric_mean_absolute_percentage_error.html">metric_mean_absolute_percentage_error</a>()</code> <br><code><a href="metric_mean_iou.html">metric_mean_iou</a>()</code> <br><code><a href="metric_mean_squared_error.html">metric_mean_squared_error</a>()</code> <br><code><a href="metric_mean_squared_logarithmic_error.html">metric_mean_squared_logarithmic_error</a>()</code> <br><code><a href="metric_mean_wrapper.html">metric_mean_wrapper</a>()</code> <br><code><a href="metric_one_hot_iou.html">metric_one_hot_iou</a>()</code> <br><code><a href="metric_one_hot_mean_iou.html">metric_one_hot_mean_iou</a>()</code> <br><code><a href="metric_pearson_correlation.html">metric_pearson_correlation</a>()</code> <br><code><a href="metric_poisson.html">metric_poisson</a>()</code> <br><code><a href="metric_precision.html">metric_precision</a>()</code> <br><code><a href="metric_precision_at_recall.html">metric_precision_at_recall</a>()</code> <br><code><a href="metric_r2_score.html">metric_r2_score</a>()</code> <br><code><a href="metric_recall.html">metric_recall</a>()</code> <br><code><a href="metric_recall_at_precision.html">metric_recall_at_precision</a>()</code> <br><code><a href="metric_root_mean_squared_error.html">metric_root_mean_squared_error</a>()</code> <br><code><a href="metric_sensitivity_at_specificity.html">metric_sensitivity_at_specificity</a>()</code> <br><code><a href="metric_sparse_categorical_accuracy.html">metric_sparse_categorical_accuracy</a>()</code> <br><code><a href="metric_sparse_categorical_crossentropy.html">metric_sparse_categorical_crossentropy</a>()</code> <br><code><a href="metric_sparse_top_k_categorical_accuracy.html">metric_sparse_top_k_categorical_accuracy</a>()</code> <br><code><a href="metric_specificity_at_sensitivity.html">metric_specificity_at_sensitivity</a>()</code> <br><code><a href="metric_squared_hinge.html">metric_squared_hinge</a>()</code> <br><code><a href="metric_sum.html">metric_sum</a>()</code> <br><code><a href="metric_top_k_categorical_accuracy.html">metric_top_k_categorical_accuracy</a>()</code> <br><code><a href="metric_true_negatives.html">metric_true_negatives</a>()</code> <br><code><a href="metric_true_positives.html">metric_true_positives</a>()</code> <br></p></div>
     </div>
 
@@ -141,7 +141,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/metric_categorical_hinge.html b/docs/dev/reference/metric_categorical_hinge.html
index 3e33a0329..05994fc3f 100644
--- a/docs/dev/reference/metric_categorical_hinge.html
+++ b/docs/dev/reference/metric_categorical_hinge.html
@@ -14,7 +14,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -125,7 +125,7 @@ <h2 id="usage">Usage<a class="anchor" aria-label="anchor" href="#usage"></a></h2
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
     <div class="dont-index">
 <ul><li><p><a href="https://keras.io/api/metrics/hinge_metrics#categoricalhinge-class" class="external-link">https://keras.io/api/metrics/hinge_metrics#categoricalhinge-class</a></p></li>
-</ul><p>Other losses: <br><code><a href="Loss.html">Loss</a>()</code> <br><code><a href="loss_binary_crossentropy.html">loss_binary_crossentropy</a>()</code> <br><code><a href="loss_binary_focal_crossentropy.html">loss_binary_focal_crossentropy</a>()</code> <br><code><a href="loss_categorical_crossentropy.html">loss_categorical_crossentropy</a>()</code> <br><code><a href="loss_categorical_focal_crossentropy.html">loss_categorical_focal_crossentropy</a>()</code> <br><code><a href="loss_categorical_hinge.html">loss_categorical_hinge</a>()</code> <br><code><a href="loss_circle.html">loss_circle</a>()</code> <br><code><a href="loss_cosine_similarity.html">loss_cosine_similarity</a>()</code> <br><code><a href="loss_ctc.html">loss_ctc</a>()</code> <br><code><a href="loss_dice.html">loss_dice</a>()</code> <br><code><a href="loss_hinge.html">loss_hinge</a>()</code> <br><code><a href="loss_huber.html">loss_huber</a>()</code> <br><code><a href="loss_kl_divergence.html">loss_kl_divergence</a>()</code> <br><code><a href="loss_log_cosh.html">loss_log_cosh</a>()</code> <br><code><a href="loss_mean_absolute_error.html">loss_mean_absolute_error</a>()</code> <br><code><a href="loss_mean_absolute_percentage_error.html">loss_mean_absolute_percentage_error</a>()</code> <br><code><a href="loss_mean_squared_error.html">loss_mean_squared_error</a>()</code> <br><code><a href="loss_mean_squared_logarithmic_error.html">loss_mean_squared_logarithmic_error</a>()</code> <br><code><a href="loss_poisson.html">loss_poisson</a>()</code> <br><code><a href="loss_sparse_categorical_crossentropy.html">loss_sparse_categorical_crossentropy</a>()</code> <br><code><a href="loss_squared_hinge.html">loss_squared_hinge</a>()</code> <br><code><a href="loss_tversky.html">loss_tversky</a>()</code> <br><code><a href="metric_binary_crossentropy.html">metric_binary_crossentropy</a>()</code> <br><code><a href="metric_binary_focal_crossentropy.html">metric_binary_focal_crossentropy</a>()</code> <br><code><a href="metric_categorical_crossentropy.html">metric_categorical_crossentropy</a>()</code> <br><code><a href="metric_categorical_focal_crossentropy.html">metric_categorical_focal_crossentropy</a>()</code> <br><code><a href="metric_hinge.html">metric_hinge</a>()</code> <br><code><a href="metric_huber.html">metric_huber</a>()</code> <br><code><a href="metric_kl_divergence.html">metric_kl_divergence</a>()</code> <br><code><a href="metric_log_cosh.html">metric_log_cosh</a>()</code> <br><code><a href="metric_mean_absolute_error.html">metric_mean_absolute_error</a>()</code> <br><code><a href="metric_mean_absolute_percentage_error.html">metric_mean_absolute_percentage_error</a>()</code> <br><code><a href="metric_mean_squared_error.html">metric_mean_squared_error</a>()</code> <br><code><a href="metric_mean_squared_logarithmic_error.html">metric_mean_squared_logarithmic_error</a>()</code> <br><code><a href="metric_poisson.html">metric_poisson</a>()</code> <br><code><a href="metric_sparse_categorical_crossentropy.html">metric_sparse_categorical_crossentropy</a>()</code> <br><code><a href="metric_squared_hinge.html">metric_squared_hinge</a>()</code> <br></p>
+</ul><p>Other losses: <br><code><a href="Loss.html">Loss</a>()</code> <br><code><a href="loss_binary_crossentropy.html">loss_binary_crossentropy</a>()</code> <br><code><a href="loss_binary_focal_crossentropy.html">loss_binary_focal_crossentropy</a>()</code> <br><code><a href="loss_categorical_crossentropy.html">loss_categorical_crossentropy</a>()</code> <br><code><a href="loss_categorical_focal_crossentropy.html">loss_categorical_focal_crossentropy</a>()</code> <br><code><a href="loss_categorical_generalized_cross_entropy.html">loss_categorical_generalized_cross_entropy</a>()</code> <br><code><a href="loss_categorical_hinge.html">loss_categorical_hinge</a>()</code> <br><code><a href="loss_circle.html">loss_circle</a>()</code> <br><code><a href="loss_cosine_similarity.html">loss_cosine_similarity</a>()</code> <br><code><a href="loss_ctc.html">loss_ctc</a>()</code> <br><code><a href="loss_dice.html">loss_dice</a>()</code> <br><code><a href="loss_hinge.html">loss_hinge</a>()</code> <br><code><a href="loss_huber.html">loss_huber</a>()</code> <br><code><a href="loss_kl_divergence.html">loss_kl_divergence</a>()</code> <br><code><a href="loss_log_cosh.html">loss_log_cosh</a>()</code> <br><code><a href="loss_mean_absolute_error.html">loss_mean_absolute_error</a>()</code> <br><code><a href="loss_mean_absolute_percentage_error.html">loss_mean_absolute_percentage_error</a>()</code> <br><code><a href="loss_mean_squared_error.html">loss_mean_squared_error</a>()</code> <br><code><a href="loss_mean_squared_logarithmic_error.html">loss_mean_squared_logarithmic_error</a>()</code> <br><code><a href="loss_poisson.html">loss_poisson</a>()</code> <br><code><a href="loss_sparse_categorical_crossentropy.html">loss_sparse_categorical_crossentropy</a>()</code> <br><code><a href="loss_squared_hinge.html">loss_squared_hinge</a>()</code> <br><code><a href="loss_tversky.html">loss_tversky</a>()</code> <br><code><a href="metric_binary_crossentropy.html">metric_binary_crossentropy</a>()</code> <br><code><a href="metric_binary_focal_crossentropy.html">metric_binary_focal_crossentropy</a>()</code> <br><code><a href="metric_categorical_crossentropy.html">metric_categorical_crossentropy</a>()</code> <br><code><a href="metric_categorical_focal_crossentropy.html">metric_categorical_focal_crossentropy</a>()</code> <br><code><a href="metric_hinge.html">metric_hinge</a>()</code> <br><code><a href="metric_huber.html">metric_huber</a>()</code> <br><code><a href="metric_kl_divergence.html">metric_kl_divergence</a>()</code> <br><code><a href="metric_log_cosh.html">metric_log_cosh</a>()</code> <br><code><a href="metric_mean_absolute_error.html">metric_mean_absolute_error</a>()</code> <br><code><a href="metric_mean_absolute_percentage_error.html">metric_mean_absolute_percentage_error</a>()</code> <br><code><a href="metric_mean_squared_error.html">metric_mean_squared_error</a>()</code> <br><code><a href="metric_mean_squared_logarithmic_error.html">metric_mean_squared_logarithmic_error</a>()</code> <br><code><a href="metric_poisson.html">metric_poisson</a>()</code> <br><code><a href="metric_sparse_categorical_crossentropy.html">metric_sparse_categorical_crossentropy</a>()</code> <br><code><a href="metric_squared_hinge.html">metric_squared_hinge</a>()</code> <br></p>
 <p>Other metrics: <br><code><a href="Metric.html">Metric</a>()</code> <br><code><a href="custom_metric.html">custom_metric</a>()</code> <br><code><a href="metric_auc.html">metric_auc</a>()</code> <br><code><a href="metric_binary_accuracy.html">metric_binary_accuracy</a>()</code> <br><code><a href="metric_binary_crossentropy.html">metric_binary_crossentropy</a>()</code> <br><code><a href="metric_binary_focal_crossentropy.html">metric_binary_focal_crossentropy</a>()</code> <br><code><a href="metric_binary_iou.html">metric_binary_iou</a>()</code> <br><code><a href="metric_categorical_accuracy.html">metric_categorical_accuracy</a>()</code> <br><code><a href="metric_categorical_crossentropy.html">metric_categorical_crossentropy</a>()</code> <br><code><a href="metric_categorical_focal_crossentropy.html">metric_categorical_focal_crossentropy</a>()</code> <br><code><a href="metric_concordance_correlation.html">metric_concordance_correlation</a>()</code> <br><code><a href="metric_cosine_similarity.html">metric_cosine_similarity</a>()</code> <br><code><a href="metric_f1_score.html">metric_f1_score</a>()</code> <br><code><a href="metric_false_negatives.html">metric_false_negatives</a>()</code> <br><code><a href="metric_false_positives.html">metric_false_positives</a>()</code> <br><code><a href="metric_fbeta_score.html">metric_fbeta_score</a>()</code> <br><code><a href="metric_hinge.html">metric_hinge</a>()</code> <br><code><a href="metric_huber.html">metric_huber</a>()</code> <br><code><a href="metric_iou.html">metric_iou</a>()</code> <br><code><a href="metric_kl_divergence.html">metric_kl_divergence</a>()</code> <br><code><a href="metric_log_cosh.html">metric_log_cosh</a>()</code> <br><code><a href="metric_log_cosh_error.html">metric_log_cosh_error</a>()</code> <br><code><a href="metric_mean.html">metric_mean</a>()</code> <br><code><a href="metric_mean_absolute_error.html">metric_mean_absolute_error</a>()</code> <br><code><a href="metric_mean_absolute_percentage_error.html">metric_mean_absolute_percentage_error</a>()</code> <br><code><a href="metric_mean_iou.html">metric_mean_iou</a>()</code> <br><code><a href="metric_mean_squared_error.html">metric_mean_squared_error</a>()</code> <br><code><a href="metric_mean_squared_logarithmic_error.html">metric_mean_squared_logarithmic_error</a>()</code> <br><code><a href="metric_mean_wrapper.html">metric_mean_wrapper</a>()</code> <br><code><a href="metric_one_hot_iou.html">metric_one_hot_iou</a>()</code> <br><code><a href="metric_one_hot_mean_iou.html">metric_one_hot_mean_iou</a>()</code> <br><code><a href="metric_pearson_correlation.html">metric_pearson_correlation</a>()</code> <br><code><a href="metric_poisson.html">metric_poisson</a>()</code> <br><code><a href="metric_precision.html">metric_precision</a>()</code> <br><code><a href="metric_precision_at_recall.html">metric_precision_at_recall</a>()</code> <br><code><a href="metric_r2_score.html">metric_r2_score</a>()</code> <br><code><a href="metric_recall.html">metric_recall</a>()</code> <br><code><a href="metric_recall_at_precision.html">metric_recall_at_precision</a>()</code> <br><code><a href="metric_root_mean_squared_error.html">metric_root_mean_squared_error</a>()</code> <br><code><a href="metric_sensitivity_at_specificity.html">metric_sensitivity_at_specificity</a>()</code> <br><code><a href="metric_sparse_categorical_accuracy.html">metric_sparse_categorical_accuracy</a>()</code> <br><code><a href="metric_sparse_categorical_crossentropy.html">metric_sparse_categorical_crossentropy</a>()</code> <br><code><a href="metric_sparse_top_k_categorical_accuracy.html">metric_sparse_top_k_categorical_accuracy</a>()</code> <br><code><a href="metric_specificity_at_sensitivity.html">metric_specificity_at_sensitivity</a>()</code> <br><code><a href="metric_squared_hinge.html">metric_squared_hinge</a>()</code> <br><code><a href="metric_sum.html">metric_sum</a>()</code> <br><code><a href="metric_top_k_categorical_accuracy.html">metric_top_k_categorical_accuracy</a>()</code> <br><code><a href="metric_true_negatives.html">metric_true_negatives</a>()</code> <br><code><a href="metric_true_positives.html">metric_true_positives</a>()</code> <br></p>
 <p>Other hinge metrics: <br><code><a href="metric_hinge.html">metric_hinge</a>()</code> <br><code><a href="metric_squared_hinge.html">metric_squared_hinge</a>()</code> <br></p></div>
     </div>
@@ -139,7 +139,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/metric_concordance_correlation.html b/docs/dev/reference/metric_concordance_correlation.html
index 484b61e6b..c33fb0b76 100644
--- a/docs/dev/reference/metric_concordance_correlation.html
+++ b/docs/dev/reference/metric_concordance_correlation.html
@@ -34,7 +34,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -169,7 +169,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/metric_cosine_similarity.html b/docs/dev/reference/metric_cosine_similarity.html
index 3f2adad6a..8c68d5b03 100644
--- a/docs/dev/reference/metric_cosine_similarity.html
+++ b/docs/dev/reference/metric_cosine_similarity.html
@@ -18,7 +18,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -145,7 +145,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/metric_f1_score.html b/docs/dev/reference/metric_f1_score.html
index 48bc9ff65..47364a8aa 100644
--- a/docs/dev/reference/metric_f1_score.html
+++ b/docs/dev/reference/metric_f1_score.html
@@ -18,7 +18,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -162,7 +162,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/metric_false_negatives.html b/docs/dev/reference/metric_false_negatives.html
index cb363183b..83a5ea78b 100644
--- a/docs/dev/reference/metric_false_negatives.html
+++ b/docs/dev/reference/metric_false_negatives.html
@@ -16,7 +16,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -138,7 +138,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/metric_false_positives.html b/docs/dev/reference/metric_false_positives.html
index 32867c46d..186dc8cc7 100644
--- a/docs/dev/reference/metric_false_positives.html
+++ b/docs/dev/reference/metric_false_positives.html
@@ -16,7 +16,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -137,7 +137,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/metric_fbeta_score.html b/docs/dev/reference/metric_fbeta_score.html
index 631be6584..e8e0da51d 100644
--- a/docs/dev/reference/metric_fbeta_score.html
+++ b/docs/dev/reference/metric_fbeta_score.html
@@ -20,7 +20,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -172,7 +172,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/metric_hinge.html b/docs/dev/reference/metric_hinge.html
index a12f2975f..b919dbbcf 100644
--- a/docs/dev/reference/metric_hinge.html
+++ b/docs/dev/reference/metric_hinge.html
@@ -16,7 +16,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -126,7 +126,7 @@ <h2 id="usage">Usage<a class="anchor" aria-label="anchor" href="#usage"></a></h2
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
     <div class="dont-index">
 <ul><li><p><a href="https://keras.io/api/metrics/hinge_metrics#hinge-class" class="external-link">https://keras.io/api/metrics/hinge_metrics#hinge-class</a></p></li>
-</ul><p>Other losses: <br><code><a href="Loss.html">Loss</a>()</code> <br><code><a href="loss_binary_crossentropy.html">loss_binary_crossentropy</a>()</code> <br><code><a href="loss_binary_focal_crossentropy.html">loss_binary_focal_crossentropy</a>()</code> <br><code><a href="loss_categorical_crossentropy.html">loss_categorical_crossentropy</a>()</code> <br><code><a href="loss_categorical_focal_crossentropy.html">loss_categorical_focal_crossentropy</a>()</code> <br><code><a href="loss_categorical_hinge.html">loss_categorical_hinge</a>()</code> <br><code><a href="loss_circle.html">loss_circle</a>()</code> <br><code><a href="loss_cosine_similarity.html">loss_cosine_similarity</a>()</code> <br><code><a href="loss_ctc.html">loss_ctc</a>()</code> <br><code><a href="loss_dice.html">loss_dice</a>()</code> <br><code><a href="loss_hinge.html">loss_hinge</a>()</code> <br><code><a href="loss_huber.html">loss_huber</a>()</code> <br><code><a href="loss_kl_divergence.html">loss_kl_divergence</a>()</code> <br><code><a href="loss_log_cosh.html">loss_log_cosh</a>()</code> <br><code><a href="loss_mean_absolute_error.html">loss_mean_absolute_error</a>()</code> <br><code><a href="loss_mean_absolute_percentage_error.html">loss_mean_absolute_percentage_error</a>()</code> <br><code><a href="loss_mean_squared_error.html">loss_mean_squared_error</a>()</code> <br><code><a href="loss_mean_squared_logarithmic_error.html">loss_mean_squared_logarithmic_error</a>()</code> <br><code><a href="loss_poisson.html">loss_poisson</a>()</code> <br><code><a href="loss_sparse_categorical_crossentropy.html">loss_sparse_categorical_crossentropy</a>()</code> <br><code><a href="loss_squared_hinge.html">loss_squared_hinge</a>()</code> <br><code><a href="loss_tversky.html">loss_tversky</a>()</code> <br><code><a href="metric_binary_crossentropy.html">metric_binary_crossentropy</a>()</code> <br><code><a href="metric_binary_focal_crossentropy.html">metric_binary_focal_crossentropy</a>()</code> <br><code><a href="metric_categorical_crossentropy.html">metric_categorical_crossentropy</a>()</code> <br><code><a href="metric_categorical_focal_crossentropy.html">metric_categorical_focal_crossentropy</a>()</code> <br><code><a href="metric_categorical_hinge.html">metric_categorical_hinge</a>()</code> <br><code><a href="metric_huber.html">metric_huber</a>()</code> <br><code><a href="metric_kl_divergence.html">metric_kl_divergence</a>()</code> <br><code><a href="metric_log_cosh.html">metric_log_cosh</a>()</code> <br><code><a href="metric_mean_absolute_error.html">metric_mean_absolute_error</a>()</code> <br><code><a href="metric_mean_absolute_percentage_error.html">metric_mean_absolute_percentage_error</a>()</code> <br><code><a href="metric_mean_squared_error.html">metric_mean_squared_error</a>()</code> <br><code><a href="metric_mean_squared_logarithmic_error.html">metric_mean_squared_logarithmic_error</a>()</code> <br><code><a href="metric_poisson.html">metric_poisson</a>()</code> <br><code><a href="metric_sparse_categorical_crossentropy.html">metric_sparse_categorical_crossentropy</a>()</code> <br><code><a href="metric_squared_hinge.html">metric_squared_hinge</a>()</code> <br></p>
+</ul><p>Other losses: <br><code><a href="Loss.html">Loss</a>()</code> <br><code><a href="loss_binary_crossentropy.html">loss_binary_crossentropy</a>()</code> <br><code><a href="loss_binary_focal_crossentropy.html">loss_binary_focal_crossentropy</a>()</code> <br><code><a href="loss_categorical_crossentropy.html">loss_categorical_crossentropy</a>()</code> <br><code><a href="loss_categorical_focal_crossentropy.html">loss_categorical_focal_crossentropy</a>()</code> <br><code><a href="loss_categorical_generalized_cross_entropy.html">loss_categorical_generalized_cross_entropy</a>()</code> <br><code><a href="loss_categorical_hinge.html">loss_categorical_hinge</a>()</code> <br><code><a href="loss_circle.html">loss_circle</a>()</code> <br><code><a href="loss_cosine_similarity.html">loss_cosine_similarity</a>()</code> <br><code><a href="loss_ctc.html">loss_ctc</a>()</code> <br><code><a href="loss_dice.html">loss_dice</a>()</code> <br><code><a href="loss_hinge.html">loss_hinge</a>()</code> <br><code><a href="loss_huber.html">loss_huber</a>()</code> <br><code><a href="loss_kl_divergence.html">loss_kl_divergence</a>()</code> <br><code><a href="loss_log_cosh.html">loss_log_cosh</a>()</code> <br><code><a href="loss_mean_absolute_error.html">loss_mean_absolute_error</a>()</code> <br><code><a href="loss_mean_absolute_percentage_error.html">loss_mean_absolute_percentage_error</a>()</code> <br><code><a href="loss_mean_squared_error.html">loss_mean_squared_error</a>()</code> <br><code><a href="loss_mean_squared_logarithmic_error.html">loss_mean_squared_logarithmic_error</a>()</code> <br><code><a href="loss_poisson.html">loss_poisson</a>()</code> <br><code><a href="loss_sparse_categorical_crossentropy.html">loss_sparse_categorical_crossentropy</a>()</code> <br><code><a href="loss_squared_hinge.html">loss_squared_hinge</a>()</code> <br><code><a href="loss_tversky.html">loss_tversky</a>()</code> <br><code><a href="metric_binary_crossentropy.html">metric_binary_crossentropy</a>()</code> <br><code><a href="metric_binary_focal_crossentropy.html">metric_binary_focal_crossentropy</a>()</code> <br><code><a href="metric_categorical_crossentropy.html">metric_categorical_crossentropy</a>()</code> <br><code><a href="metric_categorical_focal_crossentropy.html">metric_categorical_focal_crossentropy</a>()</code> <br><code><a href="metric_categorical_hinge.html">metric_categorical_hinge</a>()</code> <br><code><a href="metric_huber.html">metric_huber</a>()</code> <br><code><a href="metric_kl_divergence.html">metric_kl_divergence</a>()</code> <br><code><a href="metric_log_cosh.html">metric_log_cosh</a>()</code> <br><code><a href="metric_mean_absolute_error.html">metric_mean_absolute_error</a>()</code> <br><code><a href="metric_mean_absolute_percentage_error.html">metric_mean_absolute_percentage_error</a>()</code> <br><code><a href="metric_mean_squared_error.html">metric_mean_squared_error</a>()</code> <br><code><a href="metric_mean_squared_logarithmic_error.html">metric_mean_squared_logarithmic_error</a>()</code> <br><code><a href="metric_poisson.html">metric_poisson</a>()</code> <br><code><a href="metric_sparse_categorical_crossentropy.html">metric_sparse_categorical_crossentropy</a>()</code> <br><code><a href="metric_squared_hinge.html">metric_squared_hinge</a>()</code> <br></p>
 <p>Other metrics: <br><code><a href="Metric.html">Metric</a>()</code> <br><code><a href="custom_metric.html">custom_metric</a>()</code> <br><code><a href="metric_auc.html">metric_auc</a>()</code> <br><code><a href="metric_binary_accuracy.html">metric_binary_accuracy</a>()</code> <br><code><a href="metric_binary_crossentropy.html">metric_binary_crossentropy</a>()</code> <br><code><a href="metric_binary_focal_crossentropy.html">metric_binary_focal_crossentropy</a>()</code> <br><code><a href="metric_binary_iou.html">metric_binary_iou</a>()</code> <br><code><a href="metric_categorical_accuracy.html">metric_categorical_accuracy</a>()</code> <br><code><a href="metric_categorical_crossentropy.html">metric_categorical_crossentropy</a>()</code> <br><code><a href="metric_categorical_focal_crossentropy.html">metric_categorical_focal_crossentropy</a>()</code> <br><code><a href="metric_categorical_hinge.html">metric_categorical_hinge</a>()</code> <br><code><a href="metric_concordance_correlation.html">metric_concordance_correlation</a>()</code> <br><code><a href="metric_cosine_similarity.html">metric_cosine_similarity</a>()</code> <br><code><a href="metric_f1_score.html">metric_f1_score</a>()</code> <br><code><a href="metric_false_negatives.html">metric_false_negatives</a>()</code> <br><code><a href="metric_false_positives.html">metric_false_positives</a>()</code> <br><code><a href="metric_fbeta_score.html">metric_fbeta_score</a>()</code> <br><code><a href="metric_huber.html">metric_huber</a>()</code> <br><code><a href="metric_iou.html">metric_iou</a>()</code> <br><code><a href="metric_kl_divergence.html">metric_kl_divergence</a>()</code> <br><code><a href="metric_log_cosh.html">metric_log_cosh</a>()</code> <br><code><a href="metric_log_cosh_error.html">metric_log_cosh_error</a>()</code> <br><code><a href="metric_mean.html">metric_mean</a>()</code> <br><code><a href="metric_mean_absolute_error.html">metric_mean_absolute_error</a>()</code> <br><code><a href="metric_mean_absolute_percentage_error.html">metric_mean_absolute_percentage_error</a>()</code> <br><code><a href="metric_mean_iou.html">metric_mean_iou</a>()</code> <br><code><a href="metric_mean_squared_error.html">metric_mean_squared_error</a>()</code> <br><code><a href="metric_mean_squared_logarithmic_error.html">metric_mean_squared_logarithmic_error</a>()</code> <br><code><a href="metric_mean_wrapper.html">metric_mean_wrapper</a>()</code> <br><code><a href="metric_one_hot_iou.html">metric_one_hot_iou</a>()</code> <br><code><a href="metric_one_hot_mean_iou.html">metric_one_hot_mean_iou</a>()</code> <br><code><a href="metric_pearson_correlation.html">metric_pearson_correlation</a>()</code> <br><code><a href="metric_poisson.html">metric_poisson</a>()</code> <br><code><a href="metric_precision.html">metric_precision</a>()</code> <br><code><a href="metric_precision_at_recall.html">metric_precision_at_recall</a>()</code> <br><code><a href="metric_r2_score.html">metric_r2_score</a>()</code> <br><code><a href="metric_recall.html">metric_recall</a>()</code> <br><code><a href="metric_recall_at_precision.html">metric_recall_at_precision</a>()</code> <br><code><a href="metric_root_mean_squared_error.html">metric_root_mean_squared_error</a>()</code> <br><code><a href="metric_sensitivity_at_specificity.html">metric_sensitivity_at_specificity</a>()</code> <br><code><a href="metric_sparse_categorical_accuracy.html">metric_sparse_categorical_accuracy</a>()</code> <br><code><a href="metric_sparse_categorical_crossentropy.html">metric_sparse_categorical_crossentropy</a>()</code> <br><code><a href="metric_sparse_top_k_categorical_accuracy.html">metric_sparse_top_k_categorical_accuracy</a>()</code> <br><code><a href="metric_specificity_at_sensitivity.html">metric_specificity_at_sensitivity</a>()</code> <br><code><a href="metric_squared_hinge.html">metric_squared_hinge</a>()</code> <br><code><a href="metric_sum.html">metric_sum</a>()</code> <br><code><a href="metric_top_k_categorical_accuracy.html">metric_top_k_categorical_accuracy</a>()</code> <br><code><a href="metric_true_negatives.html">metric_true_negatives</a>()</code> <br><code><a href="metric_true_positives.html">metric_true_positives</a>()</code> <br></p>
 <p>Other hinge metrics: <br><code><a href="metric_categorical_hinge.html">metric_categorical_hinge</a>()</code> <br><code><a href="metric_squared_hinge.html">metric_squared_hinge</a>()</code> <br></p></div>
     </div>
@@ -140,7 +140,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/metric_huber.html b/docs/dev/reference/metric_huber.html
index 87e5ec412..c52ca2765 100644
--- a/docs/dev/reference/metric_huber.html
+++ b/docs/dev/reference/metric_huber.html
@@ -28,7 +28,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -120,7 +120,7 @@ <h2 id="examples">Examples<a class="anchor" aria-label="anchor" href="#examples"
     </div>
     <div class="section level2">
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
-    <div class="dont-index"><p>Other losses: <br><code><a href="Loss.html">Loss</a>()</code> <br><code><a href="loss_binary_crossentropy.html">loss_binary_crossentropy</a>()</code> <br><code><a href="loss_binary_focal_crossentropy.html">loss_binary_focal_crossentropy</a>()</code> <br><code><a href="loss_categorical_crossentropy.html">loss_categorical_crossentropy</a>()</code> <br><code><a href="loss_categorical_focal_crossentropy.html">loss_categorical_focal_crossentropy</a>()</code> <br><code><a href="loss_categorical_hinge.html">loss_categorical_hinge</a>()</code> <br><code><a href="loss_circle.html">loss_circle</a>()</code> <br><code><a href="loss_cosine_similarity.html">loss_cosine_similarity</a>()</code> <br><code><a href="loss_ctc.html">loss_ctc</a>()</code> <br><code><a href="loss_dice.html">loss_dice</a>()</code> <br><code><a href="loss_hinge.html">loss_hinge</a>()</code> <br><code><a href="loss_huber.html">loss_huber</a>()</code> <br><code><a href="loss_kl_divergence.html">loss_kl_divergence</a>()</code> <br><code><a href="loss_log_cosh.html">loss_log_cosh</a>()</code> <br><code><a href="loss_mean_absolute_error.html">loss_mean_absolute_error</a>()</code> <br><code><a href="loss_mean_absolute_percentage_error.html">loss_mean_absolute_percentage_error</a>()</code> <br><code><a href="loss_mean_squared_error.html">loss_mean_squared_error</a>()</code> <br><code><a href="loss_mean_squared_logarithmic_error.html">loss_mean_squared_logarithmic_error</a>()</code> <br><code><a href="loss_poisson.html">loss_poisson</a>()</code> <br><code><a href="loss_sparse_categorical_crossentropy.html">loss_sparse_categorical_crossentropy</a>()</code> <br><code><a href="loss_squared_hinge.html">loss_squared_hinge</a>()</code> <br><code><a href="loss_tversky.html">loss_tversky</a>()</code> <br><code><a href="metric_binary_crossentropy.html">metric_binary_crossentropy</a>()</code> <br><code><a href="metric_binary_focal_crossentropy.html">metric_binary_focal_crossentropy</a>()</code> <br><code><a href="metric_categorical_crossentropy.html">metric_categorical_crossentropy</a>()</code> <br><code><a href="metric_categorical_focal_crossentropy.html">metric_categorical_focal_crossentropy</a>()</code> <br><code><a href="metric_categorical_hinge.html">metric_categorical_hinge</a>()</code> <br><code><a href="metric_hinge.html">metric_hinge</a>()</code> <br><code><a href="metric_kl_divergence.html">metric_kl_divergence</a>()</code> <br><code><a href="metric_log_cosh.html">metric_log_cosh</a>()</code> <br><code><a href="metric_mean_absolute_error.html">metric_mean_absolute_error</a>()</code> <br><code><a href="metric_mean_absolute_percentage_error.html">metric_mean_absolute_percentage_error</a>()</code> <br><code><a href="metric_mean_squared_error.html">metric_mean_squared_error</a>()</code> <br><code><a href="metric_mean_squared_logarithmic_error.html">metric_mean_squared_logarithmic_error</a>()</code> <br><code><a href="metric_poisson.html">metric_poisson</a>()</code> <br><code><a href="metric_sparse_categorical_crossentropy.html">metric_sparse_categorical_crossentropy</a>()</code> <br><code><a href="metric_squared_hinge.html">metric_squared_hinge</a>()</code> <br></p>
+    <div class="dont-index"><p>Other losses: <br><code><a href="Loss.html">Loss</a>()</code> <br><code><a href="loss_binary_crossentropy.html">loss_binary_crossentropy</a>()</code> <br><code><a href="loss_binary_focal_crossentropy.html">loss_binary_focal_crossentropy</a>()</code> <br><code><a href="loss_categorical_crossentropy.html">loss_categorical_crossentropy</a>()</code> <br><code><a href="loss_categorical_focal_crossentropy.html">loss_categorical_focal_crossentropy</a>()</code> <br><code><a href="loss_categorical_generalized_cross_entropy.html">loss_categorical_generalized_cross_entropy</a>()</code> <br><code><a href="loss_categorical_hinge.html">loss_categorical_hinge</a>()</code> <br><code><a href="loss_circle.html">loss_circle</a>()</code> <br><code><a href="loss_cosine_similarity.html">loss_cosine_similarity</a>()</code> <br><code><a href="loss_ctc.html">loss_ctc</a>()</code> <br><code><a href="loss_dice.html">loss_dice</a>()</code> <br><code><a href="loss_hinge.html">loss_hinge</a>()</code> <br><code><a href="loss_huber.html">loss_huber</a>()</code> <br><code><a href="loss_kl_divergence.html">loss_kl_divergence</a>()</code> <br><code><a href="loss_log_cosh.html">loss_log_cosh</a>()</code> <br><code><a href="loss_mean_absolute_error.html">loss_mean_absolute_error</a>()</code> <br><code><a href="loss_mean_absolute_percentage_error.html">loss_mean_absolute_percentage_error</a>()</code> <br><code><a href="loss_mean_squared_error.html">loss_mean_squared_error</a>()</code> <br><code><a href="loss_mean_squared_logarithmic_error.html">loss_mean_squared_logarithmic_error</a>()</code> <br><code><a href="loss_poisson.html">loss_poisson</a>()</code> <br><code><a href="loss_sparse_categorical_crossentropy.html">loss_sparse_categorical_crossentropy</a>()</code> <br><code><a href="loss_squared_hinge.html">loss_squared_hinge</a>()</code> <br><code><a href="loss_tversky.html">loss_tversky</a>()</code> <br><code><a href="metric_binary_crossentropy.html">metric_binary_crossentropy</a>()</code> <br><code><a href="metric_binary_focal_crossentropy.html">metric_binary_focal_crossentropy</a>()</code> <br><code><a href="metric_categorical_crossentropy.html">metric_categorical_crossentropy</a>()</code> <br><code><a href="metric_categorical_focal_crossentropy.html">metric_categorical_focal_crossentropy</a>()</code> <br><code><a href="metric_categorical_hinge.html">metric_categorical_hinge</a>()</code> <br><code><a href="metric_hinge.html">metric_hinge</a>()</code> <br><code><a href="metric_kl_divergence.html">metric_kl_divergence</a>()</code> <br><code><a href="metric_log_cosh.html">metric_log_cosh</a>()</code> <br><code><a href="metric_mean_absolute_error.html">metric_mean_absolute_error</a>()</code> <br><code><a href="metric_mean_absolute_percentage_error.html">metric_mean_absolute_percentage_error</a>()</code> <br><code><a href="metric_mean_squared_error.html">metric_mean_squared_error</a>()</code> <br><code><a href="metric_mean_squared_logarithmic_error.html">metric_mean_squared_logarithmic_error</a>()</code> <br><code><a href="metric_poisson.html">metric_poisson</a>()</code> <br><code><a href="metric_sparse_categorical_crossentropy.html">metric_sparse_categorical_crossentropy</a>()</code> <br><code><a href="metric_squared_hinge.html">metric_squared_hinge</a>()</code> <br></p>
 <p>Other metrics: <br><code><a href="Metric.html">Metric</a>()</code> <br><code><a href="custom_metric.html">custom_metric</a>()</code> <br><code><a href="metric_auc.html">metric_auc</a>()</code> <br><code><a href="metric_binary_accuracy.html">metric_binary_accuracy</a>()</code> <br><code><a href="metric_binary_crossentropy.html">metric_binary_crossentropy</a>()</code> <br><code><a href="metric_binary_focal_crossentropy.html">metric_binary_focal_crossentropy</a>()</code> <br><code><a href="metric_binary_iou.html">metric_binary_iou</a>()</code> <br><code><a href="metric_categorical_accuracy.html">metric_categorical_accuracy</a>()</code> <br><code><a href="metric_categorical_crossentropy.html">metric_categorical_crossentropy</a>()</code> <br><code><a href="metric_categorical_focal_crossentropy.html">metric_categorical_focal_crossentropy</a>()</code> <br><code><a href="metric_categorical_hinge.html">metric_categorical_hinge</a>()</code> <br><code><a href="metric_concordance_correlation.html">metric_concordance_correlation</a>()</code> <br><code><a href="metric_cosine_similarity.html">metric_cosine_similarity</a>()</code> <br><code><a href="metric_f1_score.html">metric_f1_score</a>()</code> <br><code><a href="metric_false_negatives.html">metric_false_negatives</a>()</code> <br><code><a href="metric_false_positives.html">metric_false_positives</a>()</code> <br><code><a href="metric_fbeta_score.html">metric_fbeta_score</a>()</code> <br><code><a href="metric_hinge.html">metric_hinge</a>()</code> <br><code><a href="metric_iou.html">metric_iou</a>()</code> <br><code><a href="metric_kl_divergence.html">metric_kl_divergence</a>()</code> <br><code><a href="metric_log_cosh.html">metric_log_cosh</a>()</code> <br><code><a href="metric_log_cosh_error.html">metric_log_cosh_error</a>()</code> <br><code><a href="metric_mean.html">metric_mean</a>()</code> <br><code><a href="metric_mean_absolute_error.html">metric_mean_absolute_error</a>()</code> <br><code><a href="metric_mean_absolute_percentage_error.html">metric_mean_absolute_percentage_error</a>()</code> <br><code><a href="metric_mean_iou.html">metric_mean_iou</a>()</code> <br><code><a href="metric_mean_squared_error.html">metric_mean_squared_error</a>()</code> <br><code><a href="metric_mean_squared_logarithmic_error.html">metric_mean_squared_logarithmic_error</a>()</code> <br><code><a href="metric_mean_wrapper.html">metric_mean_wrapper</a>()</code> <br><code><a href="metric_one_hot_iou.html">metric_one_hot_iou</a>()</code> <br><code><a href="metric_one_hot_mean_iou.html">metric_one_hot_mean_iou</a>()</code> <br><code><a href="metric_pearson_correlation.html">metric_pearson_correlation</a>()</code> <br><code><a href="metric_poisson.html">metric_poisson</a>()</code> <br><code><a href="metric_precision.html">metric_precision</a>()</code> <br><code><a href="metric_precision_at_recall.html">metric_precision_at_recall</a>()</code> <br><code><a href="metric_r2_score.html">metric_r2_score</a>()</code> <br><code><a href="metric_recall.html">metric_recall</a>()</code> <br><code><a href="metric_recall_at_precision.html">metric_recall_at_precision</a>()</code> <br><code><a href="metric_root_mean_squared_error.html">metric_root_mean_squared_error</a>()</code> <br><code><a href="metric_sensitivity_at_specificity.html">metric_sensitivity_at_specificity</a>()</code> <br><code><a href="metric_sparse_categorical_accuracy.html">metric_sparse_categorical_accuracy</a>()</code> <br><code><a href="metric_sparse_categorical_crossentropy.html">metric_sparse_categorical_crossentropy</a>()</code> <br><code><a href="metric_sparse_top_k_categorical_accuracy.html">metric_sparse_top_k_categorical_accuracy</a>()</code> <br><code><a href="metric_specificity_at_sensitivity.html">metric_specificity_at_sensitivity</a>()</code> <br><code><a href="metric_squared_hinge.html">metric_squared_hinge</a>()</code> <br><code><a href="metric_sum.html">metric_sum</a>()</code> <br><code><a href="metric_top_k_categorical_accuracy.html">metric_top_k_categorical_accuracy</a>()</code> <br><code><a href="metric_true_negatives.html">metric_true_negatives</a>()</code> <br><code><a href="metric_true_positives.html">metric_true_positives</a>()</code> <br></p></div>
     </div>
 
@@ -133,7 +133,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/metric_iou.html b/docs/dev/reference/metric_iou.html
index 91c61d8f1..133279f32 100644
--- a/docs/dev/reference/metric_iou.html
+++ b/docs/dev/reference/metric_iou.html
@@ -32,7 +32,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -207,7 +207,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/metric_kl_divergence.html b/docs/dev/reference/metric_kl_divergence.html
index 9236a4b87..5ec8b7800 100644
--- a/docs/dev/reference/metric_kl_divergence.html
+++ b/docs/dev/reference/metric_kl_divergence.html
@@ -18,7 +18,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -131,7 +131,7 @@ <h2 id="usage">Usage<a class="anchor" aria-label="anchor" href="#usage"></a></h2
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
     <div class="dont-index">
 <ul><li><p><a href="https://keras.io/api/metrics/probabilistic_metrics#kldivergence-class" class="external-link">https://keras.io/api/metrics/probabilistic_metrics#kldivergence-class</a></p></li>
-</ul><p>Other losses: <br><code><a href="Loss.html">Loss</a>()</code> <br><code><a href="loss_binary_crossentropy.html">loss_binary_crossentropy</a>()</code> <br><code><a href="loss_binary_focal_crossentropy.html">loss_binary_focal_crossentropy</a>()</code> <br><code><a href="loss_categorical_crossentropy.html">loss_categorical_crossentropy</a>()</code> <br><code><a href="loss_categorical_focal_crossentropy.html">loss_categorical_focal_crossentropy</a>()</code> <br><code><a href="loss_categorical_hinge.html">loss_categorical_hinge</a>()</code> <br><code><a href="loss_circle.html">loss_circle</a>()</code> <br><code><a href="loss_cosine_similarity.html">loss_cosine_similarity</a>()</code> <br><code><a href="loss_ctc.html">loss_ctc</a>()</code> <br><code><a href="loss_dice.html">loss_dice</a>()</code> <br><code><a href="loss_hinge.html">loss_hinge</a>()</code> <br><code><a href="loss_huber.html">loss_huber</a>()</code> <br><code><a href="loss_kl_divergence.html">loss_kl_divergence</a>()</code> <br><code><a href="loss_log_cosh.html">loss_log_cosh</a>()</code> <br><code><a href="loss_mean_absolute_error.html">loss_mean_absolute_error</a>()</code> <br><code><a href="loss_mean_absolute_percentage_error.html">loss_mean_absolute_percentage_error</a>()</code> <br><code><a href="loss_mean_squared_error.html">loss_mean_squared_error</a>()</code> <br><code><a href="loss_mean_squared_logarithmic_error.html">loss_mean_squared_logarithmic_error</a>()</code> <br><code><a href="loss_poisson.html">loss_poisson</a>()</code> <br><code><a href="loss_sparse_categorical_crossentropy.html">loss_sparse_categorical_crossentropy</a>()</code> <br><code><a href="loss_squared_hinge.html">loss_squared_hinge</a>()</code> <br><code><a href="loss_tversky.html">loss_tversky</a>()</code> <br><code><a href="metric_binary_crossentropy.html">metric_binary_crossentropy</a>()</code> <br><code><a href="metric_binary_focal_crossentropy.html">metric_binary_focal_crossentropy</a>()</code> <br><code><a href="metric_categorical_crossentropy.html">metric_categorical_crossentropy</a>()</code> <br><code><a href="metric_categorical_focal_crossentropy.html">metric_categorical_focal_crossentropy</a>()</code> <br><code><a href="metric_categorical_hinge.html">metric_categorical_hinge</a>()</code> <br><code><a href="metric_hinge.html">metric_hinge</a>()</code> <br><code><a href="metric_huber.html">metric_huber</a>()</code> <br><code><a href="metric_log_cosh.html">metric_log_cosh</a>()</code> <br><code><a href="metric_mean_absolute_error.html">metric_mean_absolute_error</a>()</code> <br><code><a href="metric_mean_absolute_percentage_error.html">metric_mean_absolute_percentage_error</a>()</code> <br><code><a href="metric_mean_squared_error.html">metric_mean_squared_error</a>()</code> <br><code><a href="metric_mean_squared_logarithmic_error.html">metric_mean_squared_logarithmic_error</a>()</code> <br><code><a href="metric_poisson.html">metric_poisson</a>()</code> <br><code><a href="metric_sparse_categorical_crossentropy.html">metric_sparse_categorical_crossentropy</a>()</code> <br><code><a href="metric_squared_hinge.html">metric_squared_hinge</a>()</code> <br></p>
+</ul><p>Other losses: <br><code><a href="Loss.html">Loss</a>()</code> <br><code><a href="loss_binary_crossentropy.html">loss_binary_crossentropy</a>()</code> <br><code><a href="loss_binary_focal_crossentropy.html">loss_binary_focal_crossentropy</a>()</code> <br><code><a href="loss_categorical_crossentropy.html">loss_categorical_crossentropy</a>()</code> <br><code><a href="loss_categorical_focal_crossentropy.html">loss_categorical_focal_crossentropy</a>()</code> <br><code><a href="loss_categorical_generalized_cross_entropy.html">loss_categorical_generalized_cross_entropy</a>()</code> <br><code><a href="loss_categorical_hinge.html">loss_categorical_hinge</a>()</code> <br><code><a href="loss_circle.html">loss_circle</a>()</code> <br><code><a href="loss_cosine_similarity.html">loss_cosine_similarity</a>()</code> <br><code><a href="loss_ctc.html">loss_ctc</a>()</code> <br><code><a href="loss_dice.html">loss_dice</a>()</code> <br><code><a href="loss_hinge.html">loss_hinge</a>()</code> <br><code><a href="loss_huber.html">loss_huber</a>()</code> <br><code><a href="loss_kl_divergence.html">loss_kl_divergence</a>()</code> <br><code><a href="loss_log_cosh.html">loss_log_cosh</a>()</code> <br><code><a href="loss_mean_absolute_error.html">loss_mean_absolute_error</a>()</code> <br><code><a href="loss_mean_absolute_percentage_error.html">loss_mean_absolute_percentage_error</a>()</code> <br><code><a href="loss_mean_squared_error.html">loss_mean_squared_error</a>()</code> <br><code><a href="loss_mean_squared_logarithmic_error.html">loss_mean_squared_logarithmic_error</a>()</code> <br><code><a href="loss_poisson.html">loss_poisson</a>()</code> <br><code><a href="loss_sparse_categorical_crossentropy.html">loss_sparse_categorical_crossentropy</a>()</code> <br><code><a href="loss_squared_hinge.html">loss_squared_hinge</a>()</code> <br><code><a href="loss_tversky.html">loss_tversky</a>()</code> <br><code><a href="metric_binary_crossentropy.html">metric_binary_crossentropy</a>()</code> <br><code><a href="metric_binary_focal_crossentropy.html">metric_binary_focal_crossentropy</a>()</code> <br><code><a href="metric_categorical_crossentropy.html">metric_categorical_crossentropy</a>()</code> <br><code><a href="metric_categorical_focal_crossentropy.html">metric_categorical_focal_crossentropy</a>()</code> <br><code><a href="metric_categorical_hinge.html">metric_categorical_hinge</a>()</code> <br><code><a href="metric_hinge.html">metric_hinge</a>()</code> <br><code><a href="metric_huber.html">metric_huber</a>()</code> <br><code><a href="metric_log_cosh.html">metric_log_cosh</a>()</code> <br><code><a href="metric_mean_absolute_error.html">metric_mean_absolute_error</a>()</code> <br><code><a href="metric_mean_absolute_percentage_error.html">metric_mean_absolute_percentage_error</a>()</code> <br><code><a href="metric_mean_squared_error.html">metric_mean_squared_error</a>()</code> <br><code><a href="metric_mean_squared_logarithmic_error.html">metric_mean_squared_logarithmic_error</a>()</code> <br><code><a href="metric_poisson.html">metric_poisson</a>()</code> <br><code><a href="metric_sparse_categorical_crossentropy.html">metric_sparse_categorical_crossentropy</a>()</code> <br><code><a href="metric_squared_hinge.html">metric_squared_hinge</a>()</code> <br></p>
 <p>Other metrics: <br><code><a href="Metric.html">Metric</a>()</code> <br><code><a href="custom_metric.html">custom_metric</a>()</code> <br><code><a href="metric_auc.html">metric_auc</a>()</code> <br><code><a href="metric_binary_accuracy.html">metric_binary_accuracy</a>()</code> <br><code><a href="metric_binary_crossentropy.html">metric_binary_crossentropy</a>()</code> <br><code><a href="metric_binary_focal_crossentropy.html">metric_binary_focal_crossentropy</a>()</code> <br><code><a href="metric_binary_iou.html">metric_binary_iou</a>()</code> <br><code><a href="metric_categorical_accuracy.html">metric_categorical_accuracy</a>()</code> <br><code><a href="metric_categorical_crossentropy.html">metric_categorical_crossentropy</a>()</code> <br><code><a href="metric_categorical_focal_crossentropy.html">metric_categorical_focal_crossentropy</a>()</code> <br><code><a href="metric_categorical_hinge.html">metric_categorical_hinge</a>()</code> <br><code><a href="metric_concordance_correlation.html">metric_concordance_correlation</a>()</code> <br><code><a href="metric_cosine_similarity.html">metric_cosine_similarity</a>()</code> <br><code><a href="metric_f1_score.html">metric_f1_score</a>()</code> <br><code><a href="metric_false_negatives.html">metric_false_negatives</a>()</code> <br><code><a href="metric_false_positives.html">metric_false_positives</a>()</code> <br><code><a href="metric_fbeta_score.html">metric_fbeta_score</a>()</code> <br><code><a href="metric_hinge.html">metric_hinge</a>()</code> <br><code><a href="metric_huber.html">metric_huber</a>()</code> <br><code><a href="metric_iou.html">metric_iou</a>()</code> <br><code><a href="metric_log_cosh.html">metric_log_cosh</a>()</code> <br><code><a href="metric_log_cosh_error.html">metric_log_cosh_error</a>()</code> <br><code><a href="metric_mean.html">metric_mean</a>()</code> <br><code><a href="metric_mean_absolute_error.html">metric_mean_absolute_error</a>()</code> <br><code><a href="metric_mean_absolute_percentage_error.html">metric_mean_absolute_percentage_error</a>()</code> <br><code><a href="metric_mean_iou.html">metric_mean_iou</a>()</code> <br><code><a href="metric_mean_squared_error.html">metric_mean_squared_error</a>()</code> <br><code><a href="metric_mean_squared_logarithmic_error.html">metric_mean_squared_logarithmic_error</a>()</code> <br><code><a href="metric_mean_wrapper.html">metric_mean_wrapper</a>()</code> <br><code><a href="metric_one_hot_iou.html">metric_one_hot_iou</a>()</code> <br><code><a href="metric_one_hot_mean_iou.html">metric_one_hot_mean_iou</a>()</code> <br><code><a href="metric_pearson_correlation.html">metric_pearson_correlation</a>()</code> <br><code><a href="metric_poisson.html">metric_poisson</a>()</code> <br><code><a href="metric_precision.html">metric_precision</a>()</code> <br><code><a href="metric_precision_at_recall.html">metric_precision_at_recall</a>()</code> <br><code><a href="metric_r2_score.html">metric_r2_score</a>()</code> <br><code><a href="metric_recall.html">metric_recall</a>()</code> <br><code><a href="metric_recall_at_precision.html">metric_recall_at_precision</a>()</code> <br><code><a href="metric_root_mean_squared_error.html">metric_root_mean_squared_error</a>()</code> <br><code><a href="metric_sensitivity_at_specificity.html">metric_sensitivity_at_specificity</a>()</code> <br><code><a href="metric_sparse_categorical_accuracy.html">metric_sparse_categorical_accuracy</a>()</code> <br><code><a href="metric_sparse_categorical_crossentropy.html">metric_sparse_categorical_crossentropy</a>()</code> <br><code><a href="metric_sparse_top_k_categorical_accuracy.html">metric_sparse_top_k_categorical_accuracy</a>()</code> <br><code><a href="metric_specificity_at_sensitivity.html">metric_specificity_at_sensitivity</a>()</code> <br><code><a href="metric_squared_hinge.html">metric_squared_hinge</a>()</code> <br><code><a href="metric_sum.html">metric_sum</a>()</code> <br><code><a href="metric_top_k_categorical_accuracy.html">metric_top_k_categorical_accuracy</a>()</code> <br><code><a href="metric_true_negatives.html">metric_true_negatives</a>()</code> <br><code><a href="metric_true_positives.html">metric_true_positives</a>()</code> <br></p>
 <p>Other probabilistic metrics: <br><code><a href="metric_binary_crossentropy.html">metric_binary_crossentropy</a>()</code> <br><code><a href="metric_categorical_crossentropy.html">metric_categorical_crossentropy</a>()</code> <br><code><a href="metric_poisson.html">metric_poisson</a>()</code> <br><code><a href="metric_sparse_categorical_crossentropy.html">metric_sparse_categorical_crossentropy</a>()</code> <br></p></div>
     </div>
@@ -145,7 +145,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/metric_log_cosh.html b/docs/dev/reference/metric_log_cosh.html
index 0acf7c0b8..fcc63f17b 100644
--- a/docs/dev/reference/metric_log_cosh.html
+++ b/docs/dev/reference/metric_log_cosh.html
@@ -20,7 +20,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -106,7 +106,7 @@ <h2 id="examples">Examples<a class="anchor" aria-label="anchor" href="#examples"
     </div>
     <div class="section level2">
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
-    <div class="dont-index"><p>Other losses: <br><code><a href="Loss.html">Loss</a>()</code> <br><code><a href="loss_binary_crossentropy.html">loss_binary_crossentropy</a>()</code> <br><code><a href="loss_binary_focal_crossentropy.html">loss_binary_focal_crossentropy</a>()</code> <br><code><a href="loss_categorical_crossentropy.html">loss_categorical_crossentropy</a>()</code> <br><code><a href="loss_categorical_focal_crossentropy.html">loss_categorical_focal_crossentropy</a>()</code> <br><code><a href="loss_categorical_hinge.html">loss_categorical_hinge</a>()</code> <br><code><a href="loss_circle.html">loss_circle</a>()</code> <br><code><a href="loss_cosine_similarity.html">loss_cosine_similarity</a>()</code> <br><code><a href="loss_ctc.html">loss_ctc</a>()</code> <br><code><a href="loss_dice.html">loss_dice</a>()</code> <br><code><a href="loss_hinge.html">loss_hinge</a>()</code> <br><code><a href="loss_huber.html">loss_huber</a>()</code> <br><code><a href="loss_kl_divergence.html">loss_kl_divergence</a>()</code> <br><code><a href="loss_log_cosh.html">loss_log_cosh</a>()</code> <br><code><a href="loss_mean_absolute_error.html">loss_mean_absolute_error</a>()</code> <br><code><a href="loss_mean_absolute_percentage_error.html">loss_mean_absolute_percentage_error</a>()</code> <br><code><a href="loss_mean_squared_error.html">loss_mean_squared_error</a>()</code> <br><code><a href="loss_mean_squared_logarithmic_error.html">loss_mean_squared_logarithmic_error</a>()</code> <br><code><a href="loss_poisson.html">loss_poisson</a>()</code> <br><code><a href="loss_sparse_categorical_crossentropy.html">loss_sparse_categorical_crossentropy</a>()</code> <br><code><a href="loss_squared_hinge.html">loss_squared_hinge</a>()</code> <br><code><a href="loss_tversky.html">loss_tversky</a>()</code> <br><code><a href="metric_binary_crossentropy.html">metric_binary_crossentropy</a>()</code> <br><code><a href="metric_binary_focal_crossentropy.html">metric_binary_focal_crossentropy</a>()</code> <br><code><a href="metric_categorical_crossentropy.html">metric_categorical_crossentropy</a>()</code> <br><code><a href="metric_categorical_focal_crossentropy.html">metric_categorical_focal_crossentropy</a>()</code> <br><code><a href="metric_categorical_hinge.html">metric_categorical_hinge</a>()</code> <br><code><a href="metric_hinge.html">metric_hinge</a>()</code> <br><code><a href="metric_huber.html">metric_huber</a>()</code> <br><code><a href="metric_kl_divergence.html">metric_kl_divergence</a>()</code> <br><code><a href="metric_mean_absolute_error.html">metric_mean_absolute_error</a>()</code> <br><code><a href="metric_mean_absolute_percentage_error.html">metric_mean_absolute_percentage_error</a>()</code> <br><code><a href="metric_mean_squared_error.html">metric_mean_squared_error</a>()</code> <br><code><a href="metric_mean_squared_logarithmic_error.html">metric_mean_squared_logarithmic_error</a>()</code> <br><code><a href="metric_poisson.html">metric_poisson</a>()</code> <br><code><a href="metric_sparse_categorical_crossentropy.html">metric_sparse_categorical_crossentropy</a>()</code> <br><code><a href="metric_squared_hinge.html">metric_squared_hinge</a>()</code> <br></p>
+    <div class="dont-index"><p>Other losses: <br><code><a href="Loss.html">Loss</a>()</code> <br><code><a href="loss_binary_crossentropy.html">loss_binary_crossentropy</a>()</code> <br><code><a href="loss_binary_focal_crossentropy.html">loss_binary_focal_crossentropy</a>()</code> <br><code><a href="loss_categorical_crossentropy.html">loss_categorical_crossentropy</a>()</code> <br><code><a href="loss_categorical_focal_crossentropy.html">loss_categorical_focal_crossentropy</a>()</code> <br><code><a href="loss_categorical_generalized_cross_entropy.html">loss_categorical_generalized_cross_entropy</a>()</code> <br><code><a href="loss_categorical_hinge.html">loss_categorical_hinge</a>()</code> <br><code><a href="loss_circle.html">loss_circle</a>()</code> <br><code><a href="loss_cosine_similarity.html">loss_cosine_similarity</a>()</code> <br><code><a href="loss_ctc.html">loss_ctc</a>()</code> <br><code><a href="loss_dice.html">loss_dice</a>()</code> <br><code><a href="loss_hinge.html">loss_hinge</a>()</code> <br><code><a href="loss_huber.html">loss_huber</a>()</code> <br><code><a href="loss_kl_divergence.html">loss_kl_divergence</a>()</code> <br><code><a href="loss_log_cosh.html">loss_log_cosh</a>()</code> <br><code><a href="loss_mean_absolute_error.html">loss_mean_absolute_error</a>()</code> <br><code><a href="loss_mean_absolute_percentage_error.html">loss_mean_absolute_percentage_error</a>()</code> <br><code><a href="loss_mean_squared_error.html">loss_mean_squared_error</a>()</code> <br><code><a href="loss_mean_squared_logarithmic_error.html">loss_mean_squared_logarithmic_error</a>()</code> <br><code><a href="loss_poisson.html">loss_poisson</a>()</code> <br><code><a href="loss_sparse_categorical_crossentropy.html">loss_sparse_categorical_crossentropy</a>()</code> <br><code><a href="loss_squared_hinge.html">loss_squared_hinge</a>()</code> <br><code><a href="loss_tversky.html">loss_tversky</a>()</code> <br><code><a href="metric_binary_crossentropy.html">metric_binary_crossentropy</a>()</code> <br><code><a href="metric_binary_focal_crossentropy.html">metric_binary_focal_crossentropy</a>()</code> <br><code><a href="metric_categorical_crossentropy.html">metric_categorical_crossentropy</a>()</code> <br><code><a href="metric_categorical_focal_crossentropy.html">metric_categorical_focal_crossentropy</a>()</code> <br><code><a href="metric_categorical_hinge.html">metric_categorical_hinge</a>()</code> <br><code><a href="metric_hinge.html">metric_hinge</a>()</code> <br><code><a href="metric_huber.html">metric_huber</a>()</code> <br><code><a href="metric_kl_divergence.html">metric_kl_divergence</a>()</code> <br><code><a href="metric_mean_absolute_error.html">metric_mean_absolute_error</a>()</code> <br><code><a href="metric_mean_absolute_percentage_error.html">metric_mean_absolute_percentage_error</a>()</code> <br><code><a href="metric_mean_squared_error.html">metric_mean_squared_error</a>()</code> <br><code><a href="metric_mean_squared_logarithmic_error.html">metric_mean_squared_logarithmic_error</a>()</code> <br><code><a href="metric_poisson.html">metric_poisson</a>()</code> <br><code><a href="metric_sparse_categorical_crossentropy.html">metric_sparse_categorical_crossentropy</a>()</code> <br><code><a href="metric_squared_hinge.html">metric_squared_hinge</a>()</code> <br></p>
 <p>Other metrics: <br><code><a href="Metric.html">Metric</a>()</code> <br><code><a href="custom_metric.html">custom_metric</a>()</code> <br><code><a href="metric_auc.html">metric_auc</a>()</code> <br><code><a href="metric_binary_accuracy.html">metric_binary_accuracy</a>()</code> <br><code><a href="metric_binary_crossentropy.html">metric_binary_crossentropy</a>()</code> <br><code><a href="metric_binary_focal_crossentropy.html">metric_binary_focal_crossentropy</a>()</code> <br><code><a href="metric_binary_iou.html">metric_binary_iou</a>()</code> <br><code><a href="metric_categorical_accuracy.html">metric_categorical_accuracy</a>()</code> <br><code><a href="metric_categorical_crossentropy.html">metric_categorical_crossentropy</a>()</code> <br><code><a href="metric_categorical_focal_crossentropy.html">metric_categorical_focal_crossentropy</a>()</code> <br><code><a href="metric_categorical_hinge.html">metric_categorical_hinge</a>()</code> <br><code><a href="metric_concordance_correlation.html">metric_concordance_correlation</a>()</code> <br><code><a href="metric_cosine_similarity.html">metric_cosine_similarity</a>()</code> <br><code><a href="metric_f1_score.html">metric_f1_score</a>()</code> <br><code><a href="metric_false_negatives.html">metric_false_negatives</a>()</code> <br><code><a href="metric_false_positives.html">metric_false_positives</a>()</code> <br><code><a href="metric_fbeta_score.html">metric_fbeta_score</a>()</code> <br><code><a href="metric_hinge.html">metric_hinge</a>()</code> <br><code><a href="metric_huber.html">metric_huber</a>()</code> <br><code><a href="metric_iou.html">metric_iou</a>()</code> <br><code><a href="metric_kl_divergence.html">metric_kl_divergence</a>()</code> <br><code><a href="metric_log_cosh_error.html">metric_log_cosh_error</a>()</code> <br><code><a href="metric_mean.html">metric_mean</a>()</code> <br><code><a href="metric_mean_absolute_error.html">metric_mean_absolute_error</a>()</code> <br><code><a href="metric_mean_absolute_percentage_error.html">metric_mean_absolute_percentage_error</a>()</code> <br><code><a href="metric_mean_iou.html">metric_mean_iou</a>()</code> <br><code><a href="metric_mean_squared_error.html">metric_mean_squared_error</a>()</code> <br><code><a href="metric_mean_squared_logarithmic_error.html">metric_mean_squared_logarithmic_error</a>()</code> <br><code><a href="metric_mean_wrapper.html">metric_mean_wrapper</a>()</code> <br><code><a href="metric_one_hot_iou.html">metric_one_hot_iou</a>()</code> <br><code><a href="metric_one_hot_mean_iou.html">metric_one_hot_mean_iou</a>()</code> <br><code><a href="metric_pearson_correlation.html">metric_pearson_correlation</a>()</code> <br><code><a href="metric_poisson.html">metric_poisson</a>()</code> <br><code><a href="metric_precision.html">metric_precision</a>()</code> <br><code><a href="metric_precision_at_recall.html">metric_precision_at_recall</a>()</code> <br><code><a href="metric_r2_score.html">metric_r2_score</a>()</code> <br><code><a href="metric_recall.html">metric_recall</a>()</code> <br><code><a href="metric_recall_at_precision.html">metric_recall_at_precision</a>()</code> <br><code><a href="metric_root_mean_squared_error.html">metric_root_mean_squared_error</a>()</code> <br><code><a href="metric_sensitivity_at_specificity.html">metric_sensitivity_at_specificity</a>()</code> <br><code><a href="metric_sparse_categorical_accuracy.html">metric_sparse_categorical_accuracy</a>()</code> <br><code><a href="metric_sparse_categorical_crossentropy.html">metric_sparse_categorical_crossentropy</a>()</code> <br><code><a href="metric_sparse_top_k_categorical_accuracy.html">metric_sparse_top_k_categorical_accuracy</a>()</code> <br><code><a href="metric_specificity_at_sensitivity.html">metric_specificity_at_sensitivity</a>()</code> <br><code><a href="metric_squared_hinge.html">metric_squared_hinge</a>()</code> <br><code><a href="metric_sum.html">metric_sum</a>()</code> <br><code><a href="metric_top_k_categorical_accuracy.html">metric_top_k_categorical_accuracy</a>()</code> <br><code><a href="metric_true_negatives.html">metric_true_negatives</a>()</code> <br><code><a href="metric_true_positives.html">metric_true_positives</a>()</code> <br></p></div>
     </div>
 
@@ -119,7 +119,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/metric_log_cosh_error.html b/docs/dev/reference/metric_log_cosh_error.html
index 93512aeec..736b733a6 100644
--- a/docs/dev/reference/metric_log_cosh_error.html
+++ b/docs/dev/reference/metric_log_cosh_error.html
@@ -14,7 +14,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -128,7 +128,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/metric_mean.html b/docs/dev/reference/metric_mean.html
index d7bd9e954..0022a60ba 100644
--- a/docs/dev/reference/metric_mean.html
+++ b/docs/dev/reference/metric_mean.html
@@ -14,7 +14,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -128,7 +128,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/metric_mean_absolute_error.html b/docs/dev/reference/metric_mean_absolute_error.html
index 1e38f1c5a..a17ae03fd 100644
--- a/docs/dev/reference/metric_mean_absolute_error.html
+++ b/docs/dev/reference/metric_mean_absolute_error.html
@@ -12,7 +12,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -129,7 +129,7 @@ <h2 id="examples">Examples<a class="anchor" aria-label="anchor" href="#examples"
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
     <div class="dont-index">
 <ul><li><p><a href="https://keras.io/api/metrics/regression_metrics#meanabsoluteerror-class" class="external-link">https://keras.io/api/metrics/regression_metrics#meanabsoluteerror-class</a></p></li>
-</ul><p>Other losses: <br><code><a href="Loss.html">Loss</a>()</code> <br><code><a href="loss_binary_crossentropy.html">loss_binary_crossentropy</a>()</code> <br><code><a href="loss_binary_focal_crossentropy.html">loss_binary_focal_crossentropy</a>()</code> <br><code><a href="loss_categorical_crossentropy.html">loss_categorical_crossentropy</a>()</code> <br><code><a href="loss_categorical_focal_crossentropy.html">loss_categorical_focal_crossentropy</a>()</code> <br><code><a href="loss_categorical_hinge.html">loss_categorical_hinge</a>()</code> <br><code><a href="loss_circle.html">loss_circle</a>()</code> <br><code><a href="loss_cosine_similarity.html">loss_cosine_similarity</a>()</code> <br><code><a href="loss_ctc.html">loss_ctc</a>()</code> <br><code><a href="loss_dice.html">loss_dice</a>()</code> <br><code><a href="loss_hinge.html">loss_hinge</a>()</code> <br><code><a href="loss_huber.html">loss_huber</a>()</code> <br><code><a href="loss_kl_divergence.html">loss_kl_divergence</a>()</code> <br><code><a href="loss_log_cosh.html">loss_log_cosh</a>()</code> <br><code><a href="loss_mean_absolute_error.html">loss_mean_absolute_error</a>()</code> <br><code><a href="loss_mean_absolute_percentage_error.html">loss_mean_absolute_percentage_error</a>()</code> <br><code><a href="loss_mean_squared_error.html">loss_mean_squared_error</a>()</code> <br><code><a href="loss_mean_squared_logarithmic_error.html">loss_mean_squared_logarithmic_error</a>()</code> <br><code><a href="loss_poisson.html">loss_poisson</a>()</code> <br><code><a href="loss_sparse_categorical_crossentropy.html">loss_sparse_categorical_crossentropy</a>()</code> <br><code><a href="loss_squared_hinge.html">loss_squared_hinge</a>()</code> <br><code><a href="loss_tversky.html">loss_tversky</a>()</code> <br><code><a href="metric_binary_crossentropy.html">metric_binary_crossentropy</a>()</code> <br><code><a href="metric_binary_focal_crossentropy.html">metric_binary_focal_crossentropy</a>()</code> <br><code><a href="metric_categorical_crossentropy.html">metric_categorical_crossentropy</a>()</code> <br><code><a href="metric_categorical_focal_crossentropy.html">metric_categorical_focal_crossentropy</a>()</code> <br><code><a href="metric_categorical_hinge.html">metric_categorical_hinge</a>()</code> <br><code><a href="metric_hinge.html">metric_hinge</a>()</code> <br><code><a href="metric_huber.html">metric_huber</a>()</code> <br><code><a href="metric_kl_divergence.html">metric_kl_divergence</a>()</code> <br><code><a href="metric_log_cosh.html">metric_log_cosh</a>()</code> <br><code><a href="metric_mean_absolute_percentage_error.html">metric_mean_absolute_percentage_error</a>()</code> <br><code><a href="metric_mean_squared_error.html">metric_mean_squared_error</a>()</code> <br><code><a href="metric_mean_squared_logarithmic_error.html">metric_mean_squared_logarithmic_error</a>()</code> <br><code><a href="metric_poisson.html">metric_poisson</a>()</code> <br><code><a href="metric_sparse_categorical_crossentropy.html">metric_sparse_categorical_crossentropy</a>()</code> <br><code><a href="metric_squared_hinge.html">metric_squared_hinge</a>()</code> <br></p>
+</ul><p>Other losses: <br><code><a href="Loss.html">Loss</a>()</code> <br><code><a href="loss_binary_crossentropy.html">loss_binary_crossentropy</a>()</code> <br><code><a href="loss_binary_focal_crossentropy.html">loss_binary_focal_crossentropy</a>()</code> <br><code><a href="loss_categorical_crossentropy.html">loss_categorical_crossentropy</a>()</code> <br><code><a href="loss_categorical_focal_crossentropy.html">loss_categorical_focal_crossentropy</a>()</code> <br><code><a href="loss_categorical_generalized_cross_entropy.html">loss_categorical_generalized_cross_entropy</a>()</code> <br><code><a href="loss_categorical_hinge.html">loss_categorical_hinge</a>()</code> <br><code><a href="loss_circle.html">loss_circle</a>()</code> <br><code><a href="loss_cosine_similarity.html">loss_cosine_similarity</a>()</code> <br><code><a href="loss_ctc.html">loss_ctc</a>()</code> <br><code><a href="loss_dice.html">loss_dice</a>()</code> <br><code><a href="loss_hinge.html">loss_hinge</a>()</code> <br><code><a href="loss_huber.html">loss_huber</a>()</code> <br><code><a href="loss_kl_divergence.html">loss_kl_divergence</a>()</code> <br><code><a href="loss_log_cosh.html">loss_log_cosh</a>()</code> <br><code><a href="loss_mean_absolute_error.html">loss_mean_absolute_error</a>()</code> <br><code><a href="loss_mean_absolute_percentage_error.html">loss_mean_absolute_percentage_error</a>()</code> <br><code><a href="loss_mean_squared_error.html">loss_mean_squared_error</a>()</code> <br><code><a href="loss_mean_squared_logarithmic_error.html">loss_mean_squared_logarithmic_error</a>()</code> <br><code><a href="loss_poisson.html">loss_poisson</a>()</code> <br><code><a href="loss_sparse_categorical_crossentropy.html">loss_sparse_categorical_crossentropy</a>()</code> <br><code><a href="loss_squared_hinge.html">loss_squared_hinge</a>()</code> <br><code><a href="loss_tversky.html">loss_tversky</a>()</code> <br><code><a href="metric_binary_crossentropy.html">metric_binary_crossentropy</a>()</code> <br><code><a href="metric_binary_focal_crossentropy.html">metric_binary_focal_crossentropy</a>()</code> <br><code><a href="metric_categorical_crossentropy.html">metric_categorical_crossentropy</a>()</code> <br><code><a href="metric_categorical_focal_crossentropy.html">metric_categorical_focal_crossentropy</a>()</code> <br><code><a href="metric_categorical_hinge.html">metric_categorical_hinge</a>()</code> <br><code><a href="metric_hinge.html">metric_hinge</a>()</code> <br><code><a href="metric_huber.html">metric_huber</a>()</code> <br><code><a href="metric_kl_divergence.html">metric_kl_divergence</a>()</code> <br><code><a href="metric_log_cosh.html">metric_log_cosh</a>()</code> <br><code><a href="metric_mean_absolute_percentage_error.html">metric_mean_absolute_percentage_error</a>()</code> <br><code><a href="metric_mean_squared_error.html">metric_mean_squared_error</a>()</code> <br><code><a href="metric_mean_squared_logarithmic_error.html">metric_mean_squared_logarithmic_error</a>()</code> <br><code><a href="metric_poisson.html">metric_poisson</a>()</code> <br><code><a href="metric_sparse_categorical_crossentropy.html">metric_sparse_categorical_crossentropy</a>()</code> <br><code><a href="metric_squared_hinge.html">metric_squared_hinge</a>()</code> <br></p>
 <p>Other metrics: <br><code><a href="Metric.html">Metric</a>()</code> <br><code><a href="custom_metric.html">custom_metric</a>()</code> <br><code><a href="metric_auc.html">metric_auc</a>()</code> <br><code><a href="metric_binary_accuracy.html">metric_binary_accuracy</a>()</code> <br><code><a href="metric_binary_crossentropy.html">metric_binary_crossentropy</a>()</code> <br><code><a href="metric_binary_focal_crossentropy.html">metric_binary_focal_crossentropy</a>()</code> <br><code><a href="metric_binary_iou.html">metric_binary_iou</a>()</code> <br><code><a href="metric_categorical_accuracy.html">metric_categorical_accuracy</a>()</code> <br><code><a href="metric_categorical_crossentropy.html">metric_categorical_crossentropy</a>()</code> <br><code><a href="metric_categorical_focal_crossentropy.html">metric_categorical_focal_crossentropy</a>()</code> <br><code><a href="metric_categorical_hinge.html">metric_categorical_hinge</a>()</code> <br><code><a href="metric_concordance_correlation.html">metric_concordance_correlation</a>()</code> <br><code><a href="metric_cosine_similarity.html">metric_cosine_similarity</a>()</code> <br><code><a href="metric_f1_score.html">metric_f1_score</a>()</code> <br><code><a href="metric_false_negatives.html">metric_false_negatives</a>()</code> <br><code><a href="metric_false_positives.html">metric_false_positives</a>()</code> <br><code><a href="metric_fbeta_score.html">metric_fbeta_score</a>()</code> <br><code><a href="metric_hinge.html">metric_hinge</a>()</code> <br><code><a href="metric_huber.html">metric_huber</a>()</code> <br><code><a href="metric_iou.html">metric_iou</a>()</code> <br><code><a href="metric_kl_divergence.html">metric_kl_divergence</a>()</code> <br><code><a href="metric_log_cosh.html">metric_log_cosh</a>()</code> <br><code><a href="metric_log_cosh_error.html">metric_log_cosh_error</a>()</code> <br><code><a href="metric_mean.html">metric_mean</a>()</code> <br><code><a href="metric_mean_absolute_percentage_error.html">metric_mean_absolute_percentage_error</a>()</code> <br><code><a href="metric_mean_iou.html">metric_mean_iou</a>()</code> <br><code><a href="metric_mean_squared_error.html">metric_mean_squared_error</a>()</code> <br><code><a href="metric_mean_squared_logarithmic_error.html">metric_mean_squared_logarithmic_error</a>()</code> <br><code><a href="metric_mean_wrapper.html">metric_mean_wrapper</a>()</code> <br><code><a href="metric_one_hot_iou.html">metric_one_hot_iou</a>()</code> <br><code><a href="metric_one_hot_mean_iou.html">metric_one_hot_mean_iou</a>()</code> <br><code><a href="metric_pearson_correlation.html">metric_pearson_correlation</a>()</code> <br><code><a href="metric_poisson.html">metric_poisson</a>()</code> <br><code><a href="metric_precision.html">metric_precision</a>()</code> <br><code><a href="metric_precision_at_recall.html">metric_precision_at_recall</a>()</code> <br><code><a href="metric_r2_score.html">metric_r2_score</a>()</code> <br><code><a href="metric_recall.html">metric_recall</a>()</code> <br><code><a href="metric_recall_at_precision.html">metric_recall_at_precision</a>()</code> <br><code><a href="metric_root_mean_squared_error.html">metric_root_mean_squared_error</a>()</code> <br><code><a href="metric_sensitivity_at_specificity.html">metric_sensitivity_at_specificity</a>()</code> <br><code><a href="metric_sparse_categorical_accuracy.html">metric_sparse_categorical_accuracy</a>()</code> <br><code><a href="metric_sparse_categorical_crossentropy.html">metric_sparse_categorical_crossentropy</a>()</code> <br><code><a href="metric_sparse_top_k_categorical_accuracy.html">metric_sparse_top_k_categorical_accuracy</a>()</code> <br><code><a href="metric_specificity_at_sensitivity.html">metric_specificity_at_sensitivity</a>()</code> <br><code><a href="metric_squared_hinge.html">metric_squared_hinge</a>()</code> <br><code><a href="metric_sum.html">metric_sum</a>()</code> <br><code><a href="metric_top_k_categorical_accuracy.html">metric_top_k_categorical_accuracy</a>()</code> <br><code><a href="metric_true_negatives.html">metric_true_negatives</a>()</code> <br><code><a href="metric_true_positives.html">metric_true_positives</a>()</code> <br></p>
 <p>Other regression metrics: <br><code><a href="metric_concordance_correlation.html">metric_concordance_correlation</a>()</code> <br><code><a href="metric_cosine_similarity.html">metric_cosine_similarity</a>()</code> <br><code><a href="metric_log_cosh_error.html">metric_log_cosh_error</a>()</code> <br><code><a href="metric_mean_absolute_percentage_error.html">metric_mean_absolute_percentage_error</a>()</code> <br><code><a href="metric_mean_squared_error.html">metric_mean_squared_error</a>()</code> <br><code><a href="metric_mean_squared_logarithmic_error.html">metric_mean_squared_logarithmic_error</a>()</code> <br><code><a href="metric_pearson_correlation.html">metric_pearson_correlation</a>()</code> <br><code><a href="metric_r2_score.html">metric_r2_score</a>()</code> <br><code><a href="metric_root_mean_squared_error.html">metric_root_mean_squared_error</a>()</code> <br></p></div>
     </div>
@@ -143,7 +143,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/metric_mean_absolute_percentage_error.html b/docs/dev/reference/metric_mean_absolute_percentage_error.html
index 4f35d302e..b36658a3c 100644
--- a/docs/dev/reference/metric_mean_absolute_percentage_error.html
+++ b/docs/dev/reference/metric_mean_absolute_percentage_error.html
@@ -18,7 +18,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -120,13 +120,13 @@ <h2 id="examples">Examples<a class="anchor" aria-label="anchor" href="#examples"
 <p></p><div class="sourceCode r"><pre><code><span><span class="va">m</span> <span class="op">&lt;-</span> <span class="fu"><a href="../reference/metric_mean_absolute_percentage_error.html">metric_mean_absolute_percentage_error</a></span><span class="op">(</span><span class="op">)</span></span>
 <span><span class="va">m</span><span class="op">$</span><span class="fu">update_state</span><span class="op">(</span><span class="fu"><a href="https://rdrr.io/r/base/cbind.html" class="external-link">rbind</a></span><span class="op">(</span><span class="fu"><a href="https://rdrr.io/r/base/c.html" class="external-link">c</a></span><span class="op">(</span><span class="fl">0</span>, <span class="fl">1</span><span class="op">)</span>, <span class="fu"><a href="https://rdrr.io/r/base/c.html" class="external-link">c</a></span><span class="op">(</span><span class="fl">0</span>, <span class="fl">0</span><span class="op">)</span><span class="op">)</span>, <span class="fu"><a href="https://rdrr.io/r/base/cbind.html" class="external-link">rbind</a></span><span class="op">(</span><span class="fu"><a href="https://rdrr.io/r/base/c.html" class="external-link">c</a></span><span class="op">(</span><span class="fl">1</span>, <span class="fl">1</span><span class="op">)</span>, <span class="fu"><a href="https://rdrr.io/r/base/c.html" class="external-link">c</a></span><span class="op">(</span><span class="fl">0</span>, <span class="fl">0</span><span class="op">)</span><span class="op">)</span><span class="op">)</span></span>
 <span><span class="va">m</span><span class="op">$</span><span class="fu">result</span><span class="op">(</span><span class="op">)</span></span></code></pre><p></p></div>
-<p></p><div class="sourceCode"><pre><code><span id="cb1-1"><a href="#cb1-1" tabindex="-1"></a><span class="do">## tf.Tensor(250000000.0, shape=(), dtype=float32)</span></span>
+<p></p><div class="sourceCode"><pre><code><span id="cb1-1"><a href="#cb1-1" tabindex="-1"></a><span class="do">## tf.Tensor(2.5e+08, shape=(), dtype=float32)</span></span>
 <span id="cb1-2"><a href="#cb1-2" tabindex="-1"></a></span></code></pre><p></p></div>
 <p></p><div class="sourceCode r"><pre><code><span><span class="va">m</span><span class="op">$</span><span class="fu">reset_state</span><span class="op">(</span><span class="op">)</span></span>
 <span><span class="va">m</span><span class="op">$</span><span class="fu">update_state</span><span class="op">(</span><span class="fu"><a href="https://rdrr.io/r/base/cbind.html" class="external-link">rbind</a></span><span class="op">(</span><span class="fu"><a href="https://rdrr.io/r/base/c.html" class="external-link">c</a></span><span class="op">(</span><span class="fl">0</span>, <span class="fl">1</span><span class="op">)</span>, <span class="fu"><a href="https://rdrr.io/r/base/c.html" class="external-link">c</a></span><span class="op">(</span><span class="fl">0</span>, <span class="fl">0</span><span class="op">)</span><span class="op">)</span>, <span class="fu"><a href="https://rdrr.io/r/base/cbind.html" class="external-link">rbind</a></span><span class="op">(</span><span class="fu"><a href="https://rdrr.io/r/base/c.html" class="external-link">c</a></span><span class="op">(</span><span class="fl">1</span>, <span class="fl">1</span><span class="op">)</span>, <span class="fu"><a href="https://rdrr.io/r/base/c.html" class="external-link">c</a></span><span class="op">(</span><span class="fl">0</span>, <span class="fl">0</span><span class="op">)</span><span class="op">)</span>,</span>
 <span>               sample_weight <span class="op">=</span> <span class="fu"><a href="https://rdrr.io/r/base/c.html" class="external-link">c</a></span><span class="op">(</span><span class="fl">1</span>, <span class="fl">0</span><span class="op">)</span><span class="op">)</span></span>
 <span><span class="va">m</span><span class="op">$</span><span class="fu">result</span><span class="op">(</span><span class="op">)</span></span></code></pre><p></p></div>
-<p></p><div class="sourceCode"><pre><code><span id="cb1-1"><a href="#cb1-1" tabindex="-1"></a><span class="do">## tf.Tensor(500000000.0, shape=(), dtype=float32)</span></span>
+<p></p><div class="sourceCode"><pre><code><span id="cb1-1"><a href="#cb1-1" tabindex="-1"></a><span class="do">## tf.Tensor(5e+08, shape=(), dtype=float32)</span></span>
 <span id="cb1-2"><a href="#cb1-2" tabindex="-1"></a></span></code></pre><p></p></div>
 <p>Usage with <code><a href="https://generics.r-lib.org/reference/compile.html" class="external-link">compile()</a></code> API:</p>
 <p></p><div class="sourceCode r"><pre><code><span><span class="va">model</span> <span class="op"><a href="../reference/pipe.html">%&gt;%</a></span> <span class="fu"><a href="https://generics.r-lib.org/reference/compile.html" class="external-link">compile</a></span><span class="op">(</span></span>
@@ -138,7 +138,7 @@ <h2 id="examples">Examples<a class="anchor" aria-label="anchor" href="#examples"
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
     <div class="dont-index">
 <ul><li><p><a href="https://keras.io/api/metrics/regression_metrics#meanabsolutepercentageerror-class" class="external-link">https://keras.io/api/metrics/regression_metrics#meanabsolutepercentageerror-class</a></p></li>
-</ul><p>Other losses: <br><code><a href="Loss.html">Loss</a>()</code> <br><code><a href="loss_binary_crossentropy.html">loss_binary_crossentropy</a>()</code> <br><code><a href="loss_binary_focal_crossentropy.html">loss_binary_focal_crossentropy</a>()</code> <br><code><a href="loss_categorical_crossentropy.html">loss_categorical_crossentropy</a>()</code> <br><code><a href="loss_categorical_focal_crossentropy.html">loss_categorical_focal_crossentropy</a>()</code> <br><code><a href="loss_categorical_hinge.html">loss_categorical_hinge</a>()</code> <br><code><a href="loss_circle.html">loss_circle</a>()</code> <br><code><a href="loss_cosine_similarity.html">loss_cosine_similarity</a>()</code> <br><code><a href="loss_ctc.html">loss_ctc</a>()</code> <br><code><a href="loss_dice.html">loss_dice</a>()</code> <br><code><a href="loss_hinge.html">loss_hinge</a>()</code> <br><code><a href="loss_huber.html">loss_huber</a>()</code> <br><code><a href="loss_kl_divergence.html">loss_kl_divergence</a>()</code> <br><code><a href="loss_log_cosh.html">loss_log_cosh</a>()</code> <br><code><a href="loss_mean_absolute_error.html">loss_mean_absolute_error</a>()</code> <br><code><a href="loss_mean_absolute_percentage_error.html">loss_mean_absolute_percentage_error</a>()</code> <br><code><a href="loss_mean_squared_error.html">loss_mean_squared_error</a>()</code> <br><code><a href="loss_mean_squared_logarithmic_error.html">loss_mean_squared_logarithmic_error</a>()</code> <br><code><a href="loss_poisson.html">loss_poisson</a>()</code> <br><code><a href="loss_sparse_categorical_crossentropy.html">loss_sparse_categorical_crossentropy</a>()</code> <br><code><a href="loss_squared_hinge.html">loss_squared_hinge</a>()</code> <br><code><a href="loss_tversky.html">loss_tversky</a>()</code> <br><code><a href="metric_binary_crossentropy.html">metric_binary_crossentropy</a>()</code> <br><code><a href="metric_binary_focal_crossentropy.html">metric_binary_focal_crossentropy</a>()</code> <br><code><a href="metric_categorical_crossentropy.html">metric_categorical_crossentropy</a>()</code> <br><code><a href="metric_categorical_focal_crossentropy.html">metric_categorical_focal_crossentropy</a>()</code> <br><code><a href="metric_categorical_hinge.html">metric_categorical_hinge</a>()</code> <br><code><a href="metric_hinge.html">metric_hinge</a>()</code> <br><code><a href="metric_huber.html">metric_huber</a>()</code> <br><code><a href="metric_kl_divergence.html">metric_kl_divergence</a>()</code> <br><code><a href="metric_log_cosh.html">metric_log_cosh</a>()</code> <br><code><a href="metric_mean_absolute_error.html">metric_mean_absolute_error</a>()</code> <br><code><a href="metric_mean_squared_error.html">metric_mean_squared_error</a>()</code> <br><code><a href="metric_mean_squared_logarithmic_error.html">metric_mean_squared_logarithmic_error</a>()</code> <br><code><a href="metric_poisson.html">metric_poisson</a>()</code> <br><code><a href="metric_sparse_categorical_crossentropy.html">metric_sparse_categorical_crossentropy</a>()</code> <br><code><a href="metric_squared_hinge.html">metric_squared_hinge</a>()</code> <br></p>
+</ul><p>Other losses: <br><code><a href="Loss.html">Loss</a>()</code> <br><code><a href="loss_binary_crossentropy.html">loss_binary_crossentropy</a>()</code> <br><code><a href="loss_binary_focal_crossentropy.html">loss_binary_focal_crossentropy</a>()</code> <br><code><a href="loss_categorical_crossentropy.html">loss_categorical_crossentropy</a>()</code> <br><code><a href="loss_categorical_focal_crossentropy.html">loss_categorical_focal_crossentropy</a>()</code> <br><code><a href="loss_categorical_generalized_cross_entropy.html">loss_categorical_generalized_cross_entropy</a>()</code> <br><code><a href="loss_categorical_hinge.html">loss_categorical_hinge</a>()</code> <br><code><a href="loss_circle.html">loss_circle</a>()</code> <br><code><a href="loss_cosine_similarity.html">loss_cosine_similarity</a>()</code> <br><code><a href="loss_ctc.html">loss_ctc</a>()</code> <br><code><a href="loss_dice.html">loss_dice</a>()</code> <br><code><a href="loss_hinge.html">loss_hinge</a>()</code> <br><code><a href="loss_huber.html">loss_huber</a>()</code> <br><code><a href="loss_kl_divergence.html">loss_kl_divergence</a>()</code> <br><code><a href="loss_log_cosh.html">loss_log_cosh</a>()</code> <br><code><a href="loss_mean_absolute_error.html">loss_mean_absolute_error</a>()</code> <br><code><a href="loss_mean_absolute_percentage_error.html">loss_mean_absolute_percentage_error</a>()</code> <br><code><a href="loss_mean_squared_error.html">loss_mean_squared_error</a>()</code> <br><code><a href="loss_mean_squared_logarithmic_error.html">loss_mean_squared_logarithmic_error</a>()</code> <br><code><a href="loss_poisson.html">loss_poisson</a>()</code> <br><code><a href="loss_sparse_categorical_crossentropy.html">loss_sparse_categorical_crossentropy</a>()</code> <br><code><a href="loss_squared_hinge.html">loss_squared_hinge</a>()</code> <br><code><a href="loss_tversky.html">loss_tversky</a>()</code> <br><code><a href="metric_binary_crossentropy.html">metric_binary_crossentropy</a>()</code> <br><code><a href="metric_binary_focal_crossentropy.html">metric_binary_focal_crossentropy</a>()</code> <br><code><a href="metric_categorical_crossentropy.html">metric_categorical_crossentropy</a>()</code> <br><code><a href="metric_categorical_focal_crossentropy.html">metric_categorical_focal_crossentropy</a>()</code> <br><code><a href="metric_categorical_hinge.html">metric_categorical_hinge</a>()</code> <br><code><a href="metric_hinge.html">metric_hinge</a>()</code> <br><code><a href="metric_huber.html">metric_huber</a>()</code> <br><code><a href="metric_kl_divergence.html">metric_kl_divergence</a>()</code> <br><code><a href="metric_log_cosh.html">metric_log_cosh</a>()</code> <br><code><a href="metric_mean_absolute_error.html">metric_mean_absolute_error</a>()</code> <br><code><a href="metric_mean_squared_error.html">metric_mean_squared_error</a>()</code> <br><code><a href="metric_mean_squared_logarithmic_error.html">metric_mean_squared_logarithmic_error</a>()</code> <br><code><a href="metric_poisson.html">metric_poisson</a>()</code> <br><code><a href="metric_sparse_categorical_crossentropy.html">metric_sparse_categorical_crossentropy</a>()</code> <br><code><a href="metric_squared_hinge.html">metric_squared_hinge</a>()</code> <br></p>
 <p>Other metrics: <br><code><a href="Metric.html">Metric</a>()</code> <br><code><a href="custom_metric.html">custom_metric</a>()</code> <br><code><a href="metric_auc.html">metric_auc</a>()</code> <br><code><a href="metric_binary_accuracy.html">metric_binary_accuracy</a>()</code> <br><code><a href="metric_binary_crossentropy.html">metric_binary_crossentropy</a>()</code> <br><code><a href="metric_binary_focal_crossentropy.html">metric_binary_focal_crossentropy</a>()</code> <br><code><a href="metric_binary_iou.html">metric_binary_iou</a>()</code> <br><code><a href="metric_categorical_accuracy.html">metric_categorical_accuracy</a>()</code> <br><code><a href="metric_categorical_crossentropy.html">metric_categorical_crossentropy</a>()</code> <br><code><a href="metric_categorical_focal_crossentropy.html">metric_categorical_focal_crossentropy</a>()</code> <br><code><a href="metric_categorical_hinge.html">metric_categorical_hinge</a>()</code> <br><code><a href="metric_concordance_correlation.html">metric_concordance_correlation</a>()</code> <br><code><a href="metric_cosine_similarity.html">metric_cosine_similarity</a>()</code> <br><code><a href="metric_f1_score.html">metric_f1_score</a>()</code> <br><code><a href="metric_false_negatives.html">metric_false_negatives</a>()</code> <br><code><a href="metric_false_positives.html">metric_false_positives</a>()</code> <br><code><a href="metric_fbeta_score.html">metric_fbeta_score</a>()</code> <br><code><a href="metric_hinge.html">metric_hinge</a>()</code> <br><code><a href="metric_huber.html">metric_huber</a>()</code> <br><code><a href="metric_iou.html">metric_iou</a>()</code> <br><code><a href="metric_kl_divergence.html">metric_kl_divergence</a>()</code> <br><code><a href="metric_log_cosh.html">metric_log_cosh</a>()</code> <br><code><a href="metric_log_cosh_error.html">metric_log_cosh_error</a>()</code> <br><code><a href="metric_mean.html">metric_mean</a>()</code> <br><code><a href="metric_mean_absolute_error.html">metric_mean_absolute_error</a>()</code> <br><code><a href="metric_mean_iou.html">metric_mean_iou</a>()</code> <br><code><a href="metric_mean_squared_error.html">metric_mean_squared_error</a>()</code> <br><code><a href="metric_mean_squared_logarithmic_error.html">metric_mean_squared_logarithmic_error</a>()</code> <br><code><a href="metric_mean_wrapper.html">metric_mean_wrapper</a>()</code> <br><code><a href="metric_one_hot_iou.html">metric_one_hot_iou</a>()</code> <br><code><a href="metric_one_hot_mean_iou.html">metric_one_hot_mean_iou</a>()</code> <br><code><a href="metric_pearson_correlation.html">metric_pearson_correlation</a>()</code> <br><code><a href="metric_poisson.html">metric_poisson</a>()</code> <br><code><a href="metric_precision.html">metric_precision</a>()</code> <br><code><a href="metric_precision_at_recall.html">metric_precision_at_recall</a>()</code> <br><code><a href="metric_r2_score.html">metric_r2_score</a>()</code> <br><code><a href="metric_recall.html">metric_recall</a>()</code> <br><code><a href="metric_recall_at_precision.html">metric_recall_at_precision</a>()</code> <br><code><a href="metric_root_mean_squared_error.html">metric_root_mean_squared_error</a>()</code> <br><code><a href="metric_sensitivity_at_specificity.html">metric_sensitivity_at_specificity</a>()</code> <br><code><a href="metric_sparse_categorical_accuracy.html">metric_sparse_categorical_accuracy</a>()</code> <br><code><a href="metric_sparse_categorical_crossentropy.html">metric_sparse_categorical_crossentropy</a>()</code> <br><code><a href="metric_sparse_top_k_categorical_accuracy.html">metric_sparse_top_k_categorical_accuracy</a>()</code> <br><code><a href="metric_specificity_at_sensitivity.html">metric_specificity_at_sensitivity</a>()</code> <br><code><a href="metric_squared_hinge.html">metric_squared_hinge</a>()</code> <br><code><a href="metric_sum.html">metric_sum</a>()</code> <br><code><a href="metric_top_k_categorical_accuracy.html">metric_top_k_categorical_accuracy</a>()</code> <br><code><a href="metric_true_negatives.html">metric_true_negatives</a>()</code> <br><code><a href="metric_true_positives.html">metric_true_positives</a>()</code> <br></p>
 <p>Other regression metrics: <br><code><a href="metric_concordance_correlation.html">metric_concordance_correlation</a>()</code> <br><code><a href="metric_cosine_similarity.html">metric_cosine_similarity</a>()</code> <br><code><a href="metric_log_cosh_error.html">metric_log_cosh_error</a>()</code> <br><code><a href="metric_mean_absolute_error.html">metric_mean_absolute_error</a>()</code> <br><code><a href="metric_mean_squared_error.html">metric_mean_squared_error</a>()</code> <br><code><a href="metric_mean_squared_logarithmic_error.html">metric_mean_squared_logarithmic_error</a>()</code> <br><code><a href="metric_pearson_correlation.html">metric_pearson_correlation</a>()</code> <br><code><a href="metric_r2_score.html">metric_r2_score</a>()</code> <br><code><a href="metric_root_mean_squared_error.html">metric_root_mean_squared_error</a>()</code> <br></p></div>
     </div>
@@ -152,7 +152,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/metric_mean_iou.html b/docs/dev/reference/metric_mean_iou.html
index da1435798..7bfcb8639 100644
--- a/docs/dev/reference/metric_mean_iou.html
+++ b/docs/dev/reference/metric_mean_iou.html
@@ -28,7 +28,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -202,7 +202,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/metric_mean_squared_error.html b/docs/dev/reference/metric_mean_squared_error.html
index e14292714..474d174b0 100644
--- a/docs/dev/reference/metric_mean_squared_error.html
+++ b/docs/dev/reference/metric_mean_squared_error.html
@@ -12,7 +12,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -117,7 +117,7 @@ <h2 id="examples">Examples<a class="anchor" aria-label="anchor" href="#examples"
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
     <div class="dont-index">
 <ul><li><p><a href="https://keras.io/api/metrics/regression_metrics#meansquarederror-class" class="external-link">https://keras.io/api/metrics/regression_metrics#meansquarederror-class</a></p></li>
-</ul><p>Other losses: <br><code><a href="Loss.html">Loss</a>()</code> <br><code><a href="loss_binary_crossentropy.html">loss_binary_crossentropy</a>()</code> <br><code><a href="loss_binary_focal_crossentropy.html">loss_binary_focal_crossentropy</a>()</code> <br><code><a href="loss_categorical_crossentropy.html">loss_categorical_crossentropy</a>()</code> <br><code><a href="loss_categorical_focal_crossentropy.html">loss_categorical_focal_crossentropy</a>()</code> <br><code><a href="loss_categorical_hinge.html">loss_categorical_hinge</a>()</code> <br><code><a href="loss_circle.html">loss_circle</a>()</code> <br><code><a href="loss_cosine_similarity.html">loss_cosine_similarity</a>()</code> <br><code><a href="loss_ctc.html">loss_ctc</a>()</code> <br><code><a href="loss_dice.html">loss_dice</a>()</code> <br><code><a href="loss_hinge.html">loss_hinge</a>()</code> <br><code><a href="loss_huber.html">loss_huber</a>()</code> <br><code><a href="loss_kl_divergence.html">loss_kl_divergence</a>()</code> <br><code><a href="loss_log_cosh.html">loss_log_cosh</a>()</code> <br><code><a href="loss_mean_absolute_error.html">loss_mean_absolute_error</a>()</code> <br><code><a href="loss_mean_absolute_percentage_error.html">loss_mean_absolute_percentage_error</a>()</code> <br><code><a href="loss_mean_squared_error.html">loss_mean_squared_error</a>()</code> <br><code><a href="loss_mean_squared_logarithmic_error.html">loss_mean_squared_logarithmic_error</a>()</code> <br><code><a href="loss_poisson.html">loss_poisson</a>()</code> <br><code><a href="loss_sparse_categorical_crossentropy.html">loss_sparse_categorical_crossentropy</a>()</code> <br><code><a href="loss_squared_hinge.html">loss_squared_hinge</a>()</code> <br><code><a href="loss_tversky.html">loss_tversky</a>()</code> <br><code><a href="metric_binary_crossentropy.html">metric_binary_crossentropy</a>()</code> <br><code><a href="metric_binary_focal_crossentropy.html">metric_binary_focal_crossentropy</a>()</code> <br><code><a href="metric_categorical_crossentropy.html">metric_categorical_crossentropy</a>()</code> <br><code><a href="metric_categorical_focal_crossentropy.html">metric_categorical_focal_crossentropy</a>()</code> <br><code><a href="metric_categorical_hinge.html">metric_categorical_hinge</a>()</code> <br><code><a href="metric_hinge.html">metric_hinge</a>()</code> <br><code><a href="metric_huber.html">metric_huber</a>()</code> <br><code><a href="metric_kl_divergence.html">metric_kl_divergence</a>()</code> <br><code><a href="metric_log_cosh.html">metric_log_cosh</a>()</code> <br><code><a href="metric_mean_absolute_error.html">metric_mean_absolute_error</a>()</code> <br><code><a href="metric_mean_absolute_percentage_error.html">metric_mean_absolute_percentage_error</a>()</code> <br><code><a href="metric_mean_squared_logarithmic_error.html">metric_mean_squared_logarithmic_error</a>()</code> <br><code><a href="metric_poisson.html">metric_poisson</a>()</code> <br><code><a href="metric_sparse_categorical_crossentropy.html">metric_sparse_categorical_crossentropy</a>()</code> <br><code><a href="metric_squared_hinge.html">metric_squared_hinge</a>()</code> <br></p>
+</ul><p>Other losses: <br><code><a href="Loss.html">Loss</a>()</code> <br><code><a href="loss_binary_crossentropy.html">loss_binary_crossentropy</a>()</code> <br><code><a href="loss_binary_focal_crossentropy.html">loss_binary_focal_crossentropy</a>()</code> <br><code><a href="loss_categorical_crossentropy.html">loss_categorical_crossentropy</a>()</code> <br><code><a href="loss_categorical_focal_crossentropy.html">loss_categorical_focal_crossentropy</a>()</code> <br><code><a href="loss_categorical_generalized_cross_entropy.html">loss_categorical_generalized_cross_entropy</a>()</code> <br><code><a href="loss_categorical_hinge.html">loss_categorical_hinge</a>()</code> <br><code><a href="loss_circle.html">loss_circle</a>()</code> <br><code><a href="loss_cosine_similarity.html">loss_cosine_similarity</a>()</code> <br><code><a href="loss_ctc.html">loss_ctc</a>()</code> <br><code><a href="loss_dice.html">loss_dice</a>()</code> <br><code><a href="loss_hinge.html">loss_hinge</a>()</code> <br><code><a href="loss_huber.html">loss_huber</a>()</code> <br><code><a href="loss_kl_divergence.html">loss_kl_divergence</a>()</code> <br><code><a href="loss_log_cosh.html">loss_log_cosh</a>()</code> <br><code><a href="loss_mean_absolute_error.html">loss_mean_absolute_error</a>()</code> <br><code><a href="loss_mean_absolute_percentage_error.html">loss_mean_absolute_percentage_error</a>()</code> <br><code><a href="loss_mean_squared_error.html">loss_mean_squared_error</a>()</code> <br><code><a href="loss_mean_squared_logarithmic_error.html">loss_mean_squared_logarithmic_error</a>()</code> <br><code><a href="loss_poisson.html">loss_poisson</a>()</code> <br><code><a href="loss_sparse_categorical_crossentropy.html">loss_sparse_categorical_crossentropy</a>()</code> <br><code><a href="loss_squared_hinge.html">loss_squared_hinge</a>()</code> <br><code><a href="loss_tversky.html">loss_tversky</a>()</code> <br><code><a href="metric_binary_crossentropy.html">metric_binary_crossentropy</a>()</code> <br><code><a href="metric_binary_focal_crossentropy.html">metric_binary_focal_crossentropy</a>()</code> <br><code><a href="metric_categorical_crossentropy.html">metric_categorical_crossentropy</a>()</code> <br><code><a href="metric_categorical_focal_crossentropy.html">metric_categorical_focal_crossentropy</a>()</code> <br><code><a href="metric_categorical_hinge.html">metric_categorical_hinge</a>()</code> <br><code><a href="metric_hinge.html">metric_hinge</a>()</code> <br><code><a href="metric_huber.html">metric_huber</a>()</code> <br><code><a href="metric_kl_divergence.html">metric_kl_divergence</a>()</code> <br><code><a href="metric_log_cosh.html">metric_log_cosh</a>()</code> <br><code><a href="metric_mean_absolute_error.html">metric_mean_absolute_error</a>()</code> <br><code><a href="metric_mean_absolute_percentage_error.html">metric_mean_absolute_percentage_error</a>()</code> <br><code><a href="metric_mean_squared_logarithmic_error.html">metric_mean_squared_logarithmic_error</a>()</code> <br><code><a href="metric_poisson.html">metric_poisson</a>()</code> <br><code><a href="metric_sparse_categorical_crossentropy.html">metric_sparse_categorical_crossentropy</a>()</code> <br><code><a href="metric_squared_hinge.html">metric_squared_hinge</a>()</code> <br></p>
 <p>Other metrics: <br><code><a href="Metric.html">Metric</a>()</code> <br><code><a href="custom_metric.html">custom_metric</a>()</code> <br><code><a href="metric_auc.html">metric_auc</a>()</code> <br><code><a href="metric_binary_accuracy.html">metric_binary_accuracy</a>()</code> <br><code><a href="metric_binary_crossentropy.html">metric_binary_crossentropy</a>()</code> <br><code><a href="metric_binary_focal_crossentropy.html">metric_binary_focal_crossentropy</a>()</code> <br><code><a href="metric_binary_iou.html">metric_binary_iou</a>()</code> <br><code><a href="metric_categorical_accuracy.html">metric_categorical_accuracy</a>()</code> <br><code><a href="metric_categorical_crossentropy.html">metric_categorical_crossentropy</a>()</code> <br><code><a href="metric_categorical_focal_crossentropy.html">metric_categorical_focal_crossentropy</a>()</code> <br><code><a href="metric_categorical_hinge.html">metric_categorical_hinge</a>()</code> <br><code><a href="metric_concordance_correlation.html">metric_concordance_correlation</a>()</code> <br><code><a href="metric_cosine_similarity.html">metric_cosine_similarity</a>()</code> <br><code><a href="metric_f1_score.html">metric_f1_score</a>()</code> <br><code><a href="metric_false_negatives.html">metric_false_negatives</a>()</code> <br><code><a href="metric_false_positives.html">metric_false_positives</a>()</code> <br><code><a href="metric_fbeta_score.html">metric_fbeta_score</a>()</code> <br><code><a href="metric_hinge.html">metric_hinge</a>()</code> <br><code><a href="metric_huber.html">metric_huber</a>()</code> <br><code><a href="metric_iou.html">metric_iou</a>()</code> <br><code><a href="metric_kl_divergence.html">metric_kl_divergence</a>()</code> <br><code><a href="metric_log_cosh.html">metric_log_cosh</a>()</code> <br><code><a href="metric_log_cosh_error.html">metric_log_cosh_error</a>()</code> <br><code><a href="metric_mean.html">metric_mean</a>()</code> <br><code><a href="metric_mean_absolute_error.html">metric_mean_absolute_error</a>()</code> <br><code><a href="metric_mean_absolute_percentage_error.html">metric_mean_absolute_percentage_error</a>()</code> <br><code><a href="metric_mean_iou.html">metric_mean_iou</a>()</code> <br><code><a href="metric_mean_squared_logarithmic_error.html">metric_mean_squared_logarithmic_error</a>()</code> <br><code><a href="metric_mean_wrapper.html">metric_mean_wrapper</a>()</code> <br><code><a href="metric_one_hot_iou.html">metric_one_hot_iou</a>()</code> <br><code><a href="metric_one_hot_mean_iou.html">metric_one_hot_mean_iou</a>()</code> <br><code><a href="metric_pearson_correlation.html">metric_pearson_correlation</a>()</code> <br><code><a href="metric_poisson.html">metric_poisson</a>()</code> <br><code><a href="metric_precision.html">metric_precision</a>()</code> <br><code><a href="metric_precision_at_recall.html">metric_precision_at_recall</a>()</code> <br><code><a href="metric_r2_score.html">metric_r2_score</a>()</code> <br><code><a href="metric_recall.html">metric_recall</a>()</code> <br><code><a href="metric_recall_at_precision.html">metric_recall_at_precision</a>()</code> <br><code><a href="metric_root_mean_squared_error.html">metric_root_mean_squared_error</a>()</code> <br><code><a href="metric_sensitivity_at_specificity.html">metric_sensitivity_at_specificity</a>()</code> <br><code><a href="metric_sparse_categorical_accuracy.html">metric_sparse_categorical_accuracy</a>()</code> <br><code><a href="metric_sparse_categorical_crossentropy.html">metric_sparse_categorical_crossentropy</a>()</code> <br><code><a href="metric_sparse_top_k_categorical_accuracy.html">metric_sparse_top_k_categorical_accuracy</a>()</code> <br><code><a href="metric_specificity_at_sensitivity.html">metric_specificity_at_sensitivity</a>()</code> <br><code><a href="metric_squared_hinge.html">metric_squared_hinge</a>()</code> <br><code><a href="metric_sum.html">metric_sum</a>()</code> <br><code><a href="metric_top_k_categorical_accuracy.html">metric_top_k_categorical_accuracy</a>()</code> <br><code><a href="metric_true_negatives.html">metric_true_negatives</a>()</code> <br><code><a href="metric_true_positives.html">metric_true_positives</a>()</code> <br></p>
 <p>Other regression metrics: <br><code><a href="metric_concordance_correlation.html">metric_concordance_correlation</a>()</code> <br><code><a href="metric_cosine_similarity.html">metric_cosine_similarity</a>()</code> <br><code><a href="metric_log_cosh_error.html">metric_log_cosh_error</a>()</code> <br><code><a href="metric_mean_absolute_error.html">metric_mean_absolute_error</a>()</code> <br><code><a href="metric_mean_absolute_percentage_error.html">metric_mean_absolute_percentage_error</a>()</code> <br><code><a href="metric_mean_squared_logarithmic_error.html">metric_mean_squared_logarithmic_error</a>()</code> <br><code><a href="metric_pearson_correlation.html">metric_pearson_correlation</a>()</code> <br><code><a href="metric_r2_score.html">metric_r2_score</a>()</code> <br><code><a href="metric_root_mean_squared_error.html">metric_root_mean_squared_error</a>()</code> <br></p></div>
     </div>
@@ -131,7 +131,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/metric_mean_squared_logarithmic_error.html b/docs/dev/reference/metric_mean_squared_logarithmic_error.html
index 2061661c1..6ff95ba6b 100644
--- a/docs/dev/reference/metric_mean_squared_logarithmic_error.html
+++ b/docs/dev/reference/metric_mean_squared_logarithmic_error.html
@@ -18,7 +18,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -138,7 +138,7 @@ <h2 id="examples">Examples<a class="anchor" aria-label="anchor" href="#examples"
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
     <div class="dont-index">
 <ul><li><p><a href="https://keras.io/api/metrics/regression_metrics#meansquaredlogarithmicerror-class" class="external-link">https://keras.io/api/metrics/regression_metrics#meansquaredlogarithmicerror-class</a></p></li>
-</ul><p>Other losses: <br><code><a href="Loss.html">Loss</a>()</code> <br><code><a href="loss_binary_crossentropy.html">loss_binary_crossentropy</a>()</code> <br><code><a href="loss_binary_focal_crossentropy.html">loss_binary_focal_crossentropy</a>()</code> <br><code><a href="loss_categorical_crossentropy.html">loss_categorical_crossentropy</a>()</code> <br><code><a href="loss_categorical_focal_crossentropy.html">loss_categorical_focal_crossentropy</a>()</code> <br><code><a href="loss_categorical_hinge.html">loss_categorical_hinge</a>()</code> <br><code><a href="loss_circle.html">loss_circle</a>()</code> <br><code><a href="loss_cosine_similarity.html">loss_cosine_similarity</a>()</code> <br><code><a href="loss_ctc.html">loss_ctc</a>()</code> <br><code><a href="loss_dice.html">loss_dice</a>()</code> <br><code><a href="loss_hinge.html">loss_hinge</a>()</code> <br><code><a href="loss_huber.html">loss_huber</a>()</code> <br><code><a href="loss_kl_divergence.html">loss_kl_divergence</a>()</code> <br><code><a href="loss_log_cosh.html">loss_log_cosh</a>()</code> <br><code><a href="loss_mean_absolute_error.html">loss_mean_absolute_error</a>()</code> <br><code><a href="loss_mean_absolute_percentage_error.html">loss_mean_absolute_percentage_error</a>()</code> <br><code><a href="loss_mean_squared_error.html">loss_mean_squared_error</a>()</code> <br><code><a href="loss_mean_squared_logarithmic_error.html">loss_mean_squared_logarithmic_error</a>()</code> <br><code><a href="loss_poisson.html">loss_poisson</a>()</code> <br><code><a href="loss_sparse_categorical_crossentropy.html">loss_sparse_categorical_crossentropy</a>()</code> <br><code><a href="loss_squared_hinge.html">loss_squared_hinge</a>()</code> <br><code><a href="loss_tversky.html">loss_tversky</a>()</code> <br><code><a href="metric_binary_crossentropy.html">metric_binary_crossentropy</a>()</code> <br><code><a href="metric_binary_focal_crossentropy.html">metric_binary_focal_crossentropy</a>()</code> <br><code><a href="metric_categorical_crossentropy.html">metric_categorical_crossentropy</a>()</code> <br><code><a href="metric_categorical_focal_crossentropy.html">metric_categorical_focal_crossentropy</a>()</code> <br><code><a href="metric_categorical_hinge.html">metric_categorical_hinge</a>()</code> <br><code><a href="metric_hinge.html">metric_hinge</a>()</code> <br><code><a href="metric_huber.html">metric_huber</a>()</code> <br><code><a href="metric_kl_divergence.html">metric_kl_divergence</a>()</code> <br><code><a href="metric_log_cosh.html">metric_log_cosh</a>()</code> <br><code><a href="metric_mean_absolute_error.html">metric_mean_absolute_error</a>()</code> <br><code><a href="metric_mean_absolute_percentage_error.html">metric_mean_absolute_percentage_error</a>()</code> <br><code><a href="metric_mean_squared_error.html">metric_mean_squared_error</a>()</code> <br><code><a href="metric_poisson.html">metric_poisson</a>()</code> <br><code><a href="metric_sparse_categorical_crossentropy.html">metric_sparse_categorical_crossentropy</a>()</code> <br><code><a href="metric_squared_hinge.html">metric_squared_hinge</a>()</code> <br></p>
+</ul><p>Other losses: <br><code><a href="Loss.html">Loss</a>()</code> <br><code><a href="loss_binary_crossentropy.html">loss_binary_crossentropy</a>()</code> <br><code><a href="loss_binary_focal_crossentropy.html">loss_binary_focal_crossentropy</a>()</code> <br><code><a href="loss_categorical_crossentropy.html">loss_categorical_crossentropy</a>()</code> <br><code><a href="loss_categorical_focal_crossentropy.html">loss_categorical_focal_crossentropy</a>()</code> <br><code><a href="loss_categorical_generalized_cross_entropy.html">loss_categorical_generalized_cross_entropy</a>()</code> <br><code><a href="loss_categorical_hinge.html">loss_categorical_hinge</a>()</code> <br><code><a href="loss_circle.html">loss_circle</a>()</code> <br><code><a href="loss_cosine_similarity.html">loss_cosine_similarity</a>()</code> <br><code><a href="loss_ctc.html">loss_ctc</a>()</code> <br><code><a href="loss_dice.html">loss_dice</a>()</code> <br><code><a href="loss_hinge.html">loss_hinge</a>()</code> <br><code><a href="loss_huber.html">loss_huber</a>()</code> <br><code><a href="loss_kl_divergence.html">loss_kl_divergence</a>()</code> <br><code><a href="loss_log_cosh.html">loss_log_cosh</a>()</code> <br><code><a href="loss_mean_absolute_error.html">loss_mean_absolute_error</a>()</code> <br><code><a href="loss_mean_absolute_percentage_error.html">loss_mean_absolute_percentage_error</a>()</code> <br><code><a href="loss_mean_squared_error.html">loss_mean_squared_error</a>()</code> <br><code><a href="loss_mean_squared_logarithmic_error.html">loss_mean_squared_logarithmic_error</a>()</code> <br><code><a href="loss_poisson.html">loss_poisson</a>()</code> <br><code><a href="loss_sparse_categorical_crossentropy.html">loss_sparse_categorical_crossentropy</a>()</code> <br><code><a href="loss_squared_hinge.html">loss_squared_hinge</a>()</code> <br><code><a href="loss_tversky.html">loss_tversky</a>()</code> <br><code><a href="metric_binary_crossentropy.html">metric_binary_crossentropy</a>()</code> <br><code><a href="metric_binary_focal_crossentropy.html">metric_binary_focal_crossentropy</a>()</code> <br><code><a href="metric_categorical_crossentropy.html">metric_categorical_crossentropy</a>()</code> <br><code><a href="metric_categorical_focal_crossentropy.html">metric_categorical_focal_crossentropy</a>()</code> <br><code><a href="metric_categorical_hinge.html">metric_categorical_hinge</a>()</code> <br><code><a href="metric_hinge.html">metric_hinge</a>()</code> <br><code><a href="metric_huber.html">metric_huber</a>()</code> <br><code><a href="metric_kl_divergence.html">metric_kl_divergence</a>()</code> <br><code><a href="metric_log_cosh.html">metric_log_cosh</a>()</code> <br><code><a href="metric_mean_absolute_error.html">metric_mean_absolute_error</a>()</code> <br><code><a href="metric_mean_absolute_percentage_error.html">metric_mean_absolute_percentage_error</a>()</code> <br><code><a href="metric_mean_squared_error.html">metric_mean_squared_error</a>()</code> <br><code><a href="metric_poisson.html">metric_poisson</a>()</code> <br><code><a href="metric_sparse_categorical_crossentropy.html">metric_sparse_categorical_crossentropy</a>()</code> <br><code><a href="metric_squared_hinge.html">metric_squared_hinge</a>()</code> <br></p>
 <p>Other metrics: <br><code><a href="Metric.html">Metric</a>()</code> <br><code><a href="custom_metric.html">custom_metric</a>()</code> <br><code><a href="metric_auc.html">metric_auc</a>()</code> <br><code><a href="metric_binary_accuracy.html">metric_binary_accuracy</a>()</code> <br><code><a href="metric_binary_crossentropy.html">metric_binary_crossentropy</a>()</code> <br><code><a href="metric_binary_focal_crossentropy.html">metric_binary_focal_crossentropy</a>()</code> <br><code><a href="metric_binary_iou.html">metric_binary_iou</a>()</code> <br><code><a href="metric_categorical_accuracy.html">metric_categorical_accuracy</a>()</code> <br><code><a href="metric_categorical_crossentropy.html">metric_categorical_crossentropy</a>()</code> <br><code><a href="metric_categorical_focal_crossentropy.html">metric_categorical_focal_crossentropy</a>()</code> <br><code><a href="metric_categorical_hinge.html">metric_categorical_hinge</a>()</code> <br><code><a href="metric_concordance_correlation.html">metric_concordance_correlation</a>()</code> <br><code><a href="metric_cosine_similarity.html">metric_cosine_similarity</a>()</code> <br><code><a href="metric_f1_score.html">metric_f1_score</a>()</code> <br><code><a href="metric_false_negatives.html">metric_false_negatives</a>()</code> <br><code><a href="metric_false_positives.html">metric_false_positives</a>()</code> <br><code><a href="metric_fbeta_score.html">metric_fbeta_score</a>()</code> <br><code><a href="metric_hinge.html">metric_hinge</a>()</code> <br><code><a href="metric_huber.html">metric_huber</a>()</code> <br><code><a href="metric_iou.html">metric_iou</a>()</code> <br><code><a href="metric_kl_divergence.html">metric_kl_divergence</a>()</code> <br><code><a href="metric_log_cosh.html">metric_log_cosh</a>()</code> <br><code><a href="metric_log_cosh_error.html">metric_log_cosh_error</a>()</code> <br><code><a href="metric_mean.html">metric_mean</a>()</code> <br><code><a href="metric_mean_absolute_error.html">metric_mean_absolute_error</a>()</code> <br><code><a href="metric_mean_absolute_percentage_error.html">metric_mean_absolute_percentage_error</a>()</code> <br><code><a href="metric_mean_iou.html">metric_mean_iou</a>()</code> <br><code><a href="metric_mean_squared_error.html">metric_mean_squared_error</a>()</code> <br><code><a href="metric_mean_wrapper.html">metric_mean_wrapper</a>()</code> <br><code><a href="metric_one_hot_iou.html">metric_one_hot_iou</a>()</code> <br><code><a href="metric_one_hot_mean_iou.html">metric_one_hot_mean_iou</a>()</code> <br><code><a href="metric_pearson_correlation.html">metric_pearson_correlation</a>()</code> <br><code><a href="metric_poisson.html">metric_poisson</a>()</code> <br><code><a href="metric_precision.html">metric_precision</a>()</code> <br><code><a href="metric_precision_at_recall.html">metric_precision_at_recall</a>()</code> <br><code><a href="metric_r2_score.html">metric_r2_score</a>()</code> <br><code><a href="metric_recall.html">metric_recall</a>()</code> <br><code><a href="metric_recall_at_precision.html">metric_recall_at_precision</a>()</code> <br><code><a href="metric_root_mean_squared_error.html">metric_root_mean_squared_error</a>()</code> <br><code><a href="metric_sensitivity_at_specificity.html">metric_sensitivity_at_specificity</a>()</code> <br><code><a href="metric_sparse_categorical_accuracy.html">metric_sparse_categorical_accuracy</a>()</code> <br><code><a href="metric_sparse_categorical_crossentropy.html">metric_sparse_categorical_crossentropy</a>()</code> <br><code><a href="metric_sparse_top_k_categorical_accuracy.html">metric_sparse_top_k_categorical_accuracy</a>()</code> <br><code><a href="metric_specificity_at_sensitivity.html">metric_specificity_at_sensitivity</a>()</code> <br><code><a href="metric_squared_hinge.html">metric_squared_hinge</a>()</code> <br><code><a href="metric_sum.html">metric_sum</a>()</code> <br><code><a href="metric_top_k_categorical_accuracy.html">metric_top_k_categorical_accuracy</a>()</code> <br><code><a href="metric_true_negatives.html">metric_true_negatives</a>()</code> <br><code><a href="metric_true_positives.html">metric_true_positives</a>()</code> <br></p>
 <p>Other regression metrics: <br><code><a href="metric_concordance_correlation.html">metric_concordance_correlation</a>()</code> <br><code><a href="metric_cosine_similarity.html">metric_cosine_similarity</a>()</code> <br><code><a href="metric_log_cosh_error.html">metric_log_cosh_error</a>()</code> <br><code><a href="metric_mean_absolute_error.html">metric_mean_absolute_error</a>()</code> <br><code><a href="metric_mean_absolute_percentage_error.html">metric_mean_absolute_percentage_error</a>()</code> <br><code><a href="metric_mean_squared_error.html">metric_mean_squared_error</a>()</code> <br><code><a href="metric_pearson_correlation.html">metric_pearson_correlation</a>()</code> <br><code><a href="metric_r2_score.html">metric_r2_score</a>()</code> <br><code><a href="metric_root_mean_squared_error.html">metric_root_mean_squared_error</a>()</code> <br></p></div>
     </div>
@@ -152,7 +152,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/metric_mean_wrapper.html b/docs/dev/reference/metric_mean_wrapper.html
index b9dae1573..1ecf2f941 100644
--- a/docs/dev/reference/metric_mean_wrapper.html
+++ b/docs/dev/reference/metric_mean_wrapper.html
@@ -38,7 +38,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -147,7 +147,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/metric_one_hot_iou.html b/docs/dev/reference/metric_one_hot_iou.html
index 580d18e53..abc00d0e6 100644
--- a/docs/dev/reference/metric_one_hot_iou.html
+++ b/docs/dev/reference/metric_one_hot_iou.html
@@ -46,7 +46,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -227,7 +227,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/metric_one_hot_mean_iou.html b/docs/dev/reference/metric_one_hot_mean_iou.html
index d411c519c..8fdd7d6d4 100644
--- a/docs/dev/reference/metric_one_hot_mean_iou.html
+++ b/docs/dev/reference/metric_one_hot_mean_iou.html
@@ -48,7 +48,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -213,7 +213,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/metric_pearson_correlation.html b/docs/dev/reference/metric_pearson_correlation.html
index fff75cdc7..c50124832 100644
--- a/docs/dev/reference/metric_pearson_correlation.html
+++ b/docs/dev/reference/metric_pearson_correlation.html
@@ -26,7 +26,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -164,7 +164,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/metric_poisson.html b/docs/dev/reference/metric_poisson.html
index 0ef96de39..a10db6783 100644
--- a/docs/dev/reference/metric_poisson.html
+++ b/docs/dev/reference/metric_poisson.html
@@ -12,7 +12,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -124,7 +124,7 @@ <h2 id="examples">Examples<a class="anchor" aria-label="anchor" href="#examples"
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
     <div class="dont-index">
 <ul><li><p><a href="https://keras.io/api/metrics/probabilistic_metrics#poisson-class" class="external-link">https://keras.io/api/metrics/probabilistic_metrics#poisson-class</a></p></li>
-</ul><p>Other losses: <br><code><a href="Loss.html">Loss</a>()</code> <br><code><a href="loss_binary_crossentropy.html">loss_binary_crossentropy</a>()</code> <br><code><a href="loss_binary_focal_crossentropy.html">loss_binary_focal_crossentropy</a>()</code> <br><code><a href="loss_categorical_crossentropy.html">loss_categorical_crossentropy</a>()</code> <br><code><a href="loss_categorical_focal_crossentropy.html">loss_categorical_focal_crossentropy</a>()</code> <br><code><a href="loss_categorical_hinge.html">loss_categorical_hinge</a>()</code> <br><code><a href="loss_circle.html">loss_circle</a>()</code> <br><code><a href="loss_cosine_similarity.html">loss_cosine_similarity</a>()</code> <br><code><a href="loss_ctc.html">loss_ctc</a>()</code> <br><code><a href="loss_dice.html">loss_dice</a>()</code> <br><code><a href="loss_hinge.html">loss_hinge</a>()</code> <br><code><a href="loss_huber.html">loss_huber</a>()</code> <br><code><a href="loss_kl_divergence.html">loss_kl_divergence</a>()</code> <br><code><a href="loss_log_cosh.html">loss_log_cosh</a>()</code> <br><code><a href="loss_mean_absolute_error.html">loss_mean_absolute_error</a>()</code> <br><code><a href="loss_mean_absolute_percentage_error.html">loss_mean_absolute_percentage_error</a>()</code> <br><code><a href="loss_mean_squared_error.html">loss_mean_squared_error</a>()</code> <br><code><a href="loss_mean_squared_logarithmic_error.html">loss_mean_squared_logarithmic_error</a>()</code> <br><code><a href="loss_poisson.html">loss_poisson</a>()</code> <br><code><a href="loss_sparse_categorical_crossentropy.html">loss_sparse_categorical_crossentropy</a>()</code> <br><code><a href="loss_squared_hinge.html">loss_squared_hinge</a>()</code> <br><code><a href="loss_tversky.html">loss_tversky</a>()</code> <br><code><a href="metric_binary_crossentropy.html">metric_binary_crossentropy</a>()</code> <br><code><a href="metric_binary_focal_crossentropy.html">metric_binary_focal_crossentropy</a>()</code> <br><code><a href="metric_categorical_crossentropy.html">metric_categorical_crossentropy</a>()</code> <br><code><a href="metric_categorical_focal_crossentropy.html">metric_categorical_focal_crossentropy</a>()</code> <br><code><a href="metric_categorical_hinge.html">metric_categorical_hinge</a>()</code> <br><code><a href="metric_hinge.html">metric_hinge</a>()</code> <br><code><a href="metric_huber.html">metric_huber</a>()</code> <br><code><a href="metric_kl_divergence.html">metric_kl_divergence</a>()</code> <br><code><a href="metric_log_cosh.html">metric_log_cosh</a>()</code> <br><code><a href="metric_mean_absolute_error.html">metric_mean_absolute_error</a>()</code> <br><code><a href="metric_mean_absolute_percentage_error.html">metric_mean_absolute_percentage_error</a>()</code> <br><code><a href="metric_mean_squared_error.html">metric_mean_squared_error</a>()</code> <br><code><a href="metric_mean_squared_logarithmic_error.html">metric_mean_squared_logarithmic_error</a>()</code> <br><code><a href="metric_sparse_categorical_crossentropy.html">metric_sparse_categorical_crossentropy</a>()</code> <br><code><a href="metric_squared_hinge.html">metric_squared_hinge</a>()</code> <br></p>
+</ul><p>Other losses: <br><code><a href="Loss.html">Loss</a>()</code> <br><code><a href="loss_binary_crossentropy.html">loss_binary_crossentropy</a>()</code> <br><code><a href="loss_binary_focal_crossentropy.html">loss_binary_focal_crossentropy</a>()</code> <br><code><a href="loss_categorical_crossentropy.html">loss_categorical_crossentropy</a>()</code> <br><code><a href="loss_categorical_focal_crossentropy.html">loss_categorical_focal_crossentropy</a>()</code> <br><code><a href="loss_categorical_generalized_cross_entropy.html">loss_categorical_generalized_cross_entropy</a>()</code> <br><code><a href="loss_categorical_hinge.html">loss_categorical_hinge</a>()</code> <br><code><a href="loss_circle.html">loss_circle</a>()</code> <br><code><a href="loss_cosine_similarity.html">loss_cosine_similarity</a>()</code> <br><code><a href="loss_ctc.html">loss_ctc</a>()</code> <br><code><a href="loss_dice.html">loss_dice</a>()</code> <br><code><a href="loss_hinge.html">loss_hinge</a>()</code> <br><code><a href="loss_huber.html">loss_huber</a>()</code> <br><code><a href="loss_kl_divergence.html">loss_kl_divergence</a>()</code> <br><code><a href="loss_log_cosh.html">loss_log_cosh</a>()</code> <br><code><a href="loss_mean_absolute_error.html">loss_mean_absolute_error</a>()</code> <br><code><a href="loss_mean_absolute_percentage_error.html">loss_mean_absolute_percentage_error</a>()</code> <br><code><a href="loss_mean_squared_error.html">loss_mean_squared_error</a>()</code> <br><code><a href="loss_mean_squared_logarithmic_error.html">loss_mean_squared_logarithmic_error</a>()</code> <br><code><a href="loss_poisson.html">loss_poisson</a>()</code> <br><code><a href="loss_sparse_categorical_crossentropy.html">loss_sparse_categorical_crossentropy</a>()</code> <br><code><a href="loss_squared_hinge.html">loss_squared_hinge</a>()</code> <br><code><a href="loss_tversky.html">loss_tversky</a>()</code> <br><code><a href="metric_binary_crossentropy.html">metric_binary_crossentropy</a>()</code> <br><code><a href="metric_binary_focal_crossentropy.html">metric_binary_focal_crossentropy</a>()</code> <br><code><a href="metric_categorical_crossentropy.html">metric_categorical_crossentropy</a>()</code> <br><code><a href="metric_categorical_focal_crossentropy.html">metric_categorical_focal_crossentropy</a>()</code> <br><code><a href="metric_categorical_hinge.html">metric_categorical_hinge</a>()</code> <br><code><a href="metric_hinge.html">metric_hinge</a>()</code> <br><code><a href="metric_huber.html">metric_huber</a>()</code> <br><code><a href="metric_kl_divergence.html">metric_kl_divergence</a>()</code> <br><code><a href="metric_log_cosh.html">metric_log_cosh</a>()</code> <br><code><a href="metric_mean_absolute_error.html">metric_mean_absolute_error</a>()</code> <br><code><a href="metric_mean_absolute_percentage_error.html">metric_mean_absolute_percentage_error</a>()</code> <br><code><a href="metric_mean_squared_error.html">metric_mean_squared_error</a>()</code> <br><code><a href="metric_mean_squared_logarithmic_error.html">metric_mean_squared_logarithmic_error</a>()</code> <br><code><a href="metric_sparse_categorical_crossentropy.html">metric_sparse_categorical_crossentropy</a>()</code> <br><code><a href="metric_squared_hinge.html">metric_squared_hinge</a>()</code> <br></p>
 <p>Other metrics: <br><code><a href="Metric.html">Metric</a>()</code> <br><code><a href="custom_metric.html">custom_metric</a>()</code> <br><code><a href="metric_auc.html">metric_auc</a>()</code> <br><code><a href="metric_binary_accuracy.html">metric_binary_accuracy</a>()</code> <br><code><a href="metric_binary_crossentropy.html">metric_binary_crossentropy</a>()</code> <br><code><a href="metric_binary_focal_crossentropy.html">metric_binary_focal_crossentropy</a>()</code> <br><code><a href="metric_binary_iou.html">metric_binary_iou</a>()</code> <br><code><a href="metric_categorical_accuracy.html">metric_categorical_accuracy</a>()</code> <br><code><a href="metric_categorical_crossentropy.html">metric_categorical_crossentropy</a>()</code> <br><code><a href="metric_categorical_focal_crossentropy.html">metric_categorical_focal_crossentropy</a>()</code> <br><code><a href="metric_categorical_hinge.html">metric_categorical_hinge</a>()</code> <br><code><a href="metric_concordance_correlation.html">metric_concordance_correlation</a>()</code> <br><code><a href="metric_cosine_similarity.html">metric_cosine_similarity</a>()</code> <br><code><a href="metric_f1_score.html">metric_f1_score</a>()</code> <br><code><a href="metric_false_negatives.html">metric_false_negatives</a>()</code> <br><code><a href="metric_false_positives.html">metric_false_positives</a>()</code> <br><code><a href="metric_fbeta_score.html">metric_fbeta_score</a>()</code> <br><code><a href="metric_hinge.html">metric_hinge</a>()</code> <br><code><a href="metric_huber.html">metric_huber</a>()</code> <br><code><a href="metric_iou.html">metric_iou</a>()</code> <br><code><a href="metric_kl_divergence.html">metric_kl_divergence</a>()</code> <br><code><a href="metric_log_cosh.html">metric_log_cosh</a>()</code> <br><code><a href="metric_log_cosh_error.html">metric_log_cosh_error</a>()</code> <br><code><a href="metric_mean.html">metric_mean</a>()</code> <br><code><a href="metric_mean_absolute_error.html">metric_mean_absolute_error</a>()</code> <br><code><a href="metric_mean_absolute_percentage_error.html">metric_mean_absolute_percentage_error</a>()</code> <br><code><a href="metric_mean_iou.html">metric_mean_iou</a>()</code> <br><code><a href="metric_mean_squared_error.html">metric_mean_squared_error</a>()</code> <br><code><a href="metric_mean_squared_logarithmic_error.html">metric_mean_squared_logarithmic_error</a>()</code> <br><code><a href="metric_mean_wrapper.html">metric_mean_wrapper</a>()</code> <br><code><a href="metric_one_hot_iou.html">metric_one_hot_iou</a>()</code> <br><code><a href="metric_one_hot_mean_iou.html">metric_one_hot_mean_iou</a>()</code> <br><code><a href="metric_pearson_correlation.html">metric_pearson_correlation</a>()</code> <br><code><a href="metric_precision.html">metric_precision</a>()</code> <br><code><a href="metric_precision_at_recall.html">metric_precision_at_recall</a>()</code> <br><code><a href="metric_r2_score.html">metric_r2_score</a>()</code> <br><code><a href="metric_recall.html">metric_recall</a>()</code> <br><code><a href="metric_recall_at_precision.html">metric_recall_at_precision</a>()</code> <br><code><a href="metric_root_mean_squared_error.html">metric_root_mean_squared_error</a>()</code> <br><code><a href="metric_sensitivity_at_specificity.html">metric_sensitivity_at_specificity</a>()</code> <br><code><a href="metric_sparse_categorical_accuracy.html">metric_sparse_categorical_accuracy</a>()</code> <br><code><a href="metric_sparse_categorical_crossentropy.html">metric_sparse_categorical_crossentropy</a>()</code> <br><code><a href="metric_sparse_top_k_categorical_accuracy.html">metric_sparse_top_k_categorical_accuracy</a>()</code> <br><code><a href="metric_specificity_at_sensitivity.html">metric_specificity_at_sensitivity</a>()</code> <br><code><a href="metric_squared_hinge.html">metric_squared_hinge</a>()</code> <br><code><a href="metric_sum.html">metric_sum</a>()</code> <br><code><a href="metric_top_k_categorical_accuracy.html">metric_top_k_categorical_accuracy</a>()</code> <br><code><a href="metric_true_negatives.html">metric_true_negatives</a>()</code> <br><code><a href="metric_true_positives.html">metric_true_positives</a>()</code> <br></p>
 <p>Other probabilistic metrics: <br><code><a href="metric_binary_crossentropy.html">metric_binary_crossentropy</a>()</code> <br><code><a href="metric_categorical_crossentropy.html">metric_categorical_crossentropy</a>()</code> <br><code><a href="metric_kl_divergence.html">metric_kl_divergence</a>()</code> <br><code><a href="metric_sparse_categorical_crossentropy.html">metric_sparse_categorical_crossentropy</a>()</code> <br></p></div>
     </div>
@@ -138,7 +138,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/metric_precision.html b/docs/dev/reference/metric_precision.html
index 3ba81b157..4d27ead07 100644
--- a/docs/dev/reference/metric_precision.html
+++ b/docs/dev/reference/metric_precision.html
@@ -34,7 +34,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -213,7 +213,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/metric_precision_at_recall.html b/docs/dev/reference/metric_precision_at_recall.html
index 500cdc661..073acdfcb 100644
--- a/docs/dev/reference/metric_precision_at_recall.html
+++ b/docs/dev/reference/metric_precision_at_recall.html
@@ -26,7 +26,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -173,7 +173,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/metric_r2_score.html b/docs/dev/reference/metric_r2_score.html
index aee4508ee..f7b250c85 100644
--- a/docs/dev/reference/metric_r2_score.html
+++ b/docs/dev/reference/metric_r2_score.html
@@ -40,7 +40,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -175,7 +175,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/metric_recall.html b/docs/dev/reference/metric_recall.html
index 80e3def11..282399ff5 100644
--- a/docs/dev/reference/metric_recall.html
+++ b/docs/dev/reference/metric_recall.html
@@ -30,7 +30,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -193,7 +193,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/metric_recall_at_precision.html b/docs/dev/reference/metric_recall_at_precision.html
index d7461c736..2f8370d07 100644
--- a/docs/dev/reference/metric_recall_at_precision.html
+++ b/docs/dev/reference/metric_recall_at_precision.html
@@ -30,7 +30,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -177,7 +177,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/metric_root_mean_squared_error.html b/docs/dev/reference/metric_root_mean_squared_error.html
index 42aab6ebe..cea25a189 100644
--- a/docs/dev/reference/metric_root_mean_squared_error.html
+++ b/docs/dev/reference/metric_root_mean_squared_error.html
@@ -12,7 +12,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -130,7 +130,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/metric_sensitivity_at_specificity.html b/docs/dev/reference/metric_sensitivity_at_specificity.html
index 081629d1f..ee8876494 100644
--- a/docs/dev/reference/metric_sensitivity_at_specificity.html
+++ b/docs/dev/reference/metric_sensitivity_at_specificity.html
@@ -40,7 +40,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -194,7 +194,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/metric_sparse_categorical_accuracy.html b/docs/dev/reference/metric_sparse_categorical_accuracy.html
index 1b94caf71..e254ff6ac 100644
--- a/docs/dev/reference/metric_sparse_categorical_accuracy.html
+++ b/docs/dev/reference/metric_sparse_categorical_accuracy.html
@@ -26,7 +26,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -162,7 +162,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/metric_sparse_categorical_crossentropy.html b/docs/dev/reference/metric_sparse_categorical_crossentropy.html
index 5963d10ba..fb52a6881 100644
--- a/docs/dev/reference/metric_sparse_categorical_crossentropy.html
+++ b/docs/dev/reference/metric_sparse_categorical_crossentropy.html
@@ -18,7 +18,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -164,7 +164,7 @@ <h2 id="examples">Examples<a class="anchor" aria-label="anchor" href="#examples"
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
     <div class="dont-index">
 <ul><li><p><a href="https://keras.io/api/metrics/probabilistic_metrics#sparsecategoricalcrossentropy-class" class="external-link">https://keras.io/api/metrics/probabilistic_metrics#sparsecategoricalcrossentropy-class</a></p></li>
-</ul><p>Other losses: <br><code><a href="Loss.html">Loss</a>()</code> <br><code><a href="loss_binary_crossentropy.html">loss_binary_crossentropy</a>()</code> <br><code><a href="loss_binary_focal_crossentropy.html">loss_binary_focal_crossentropy</a>()</code> <br><code><a href="loss_categorical_crossentropy.html">loss_categorical_crossentropy</a>()</code> <br><code><a href="loss_categorical_focal_crossentropy.html">loss_categorical_focal_crossentropy</a>()</code> <br><code><a href="loss_categorical_hinge.html">loss_categorical_hinge</a>()</code> <br><code><a href="loss_circle.html">loss_circle</a>()</code> <br><code><a href="loss_cosine_similarity.html">loss_cosine_similarity</a>()</code> <br><code><a href="loss_ctc.html">loss_ctc</a>()</code> <br><code><a href="loss_dice.html">loss_dice</a>()</code> <br><code><a href="loss_hinge.html">loss_hinge</a>()</code> <br><code><a href="loss_huber.html">loss_huber</a>()</code> <br><code><a href="loss_kl_divergence.html">loss_kl_divergence</a>()</code> <br><code><a href="loss_log_cosh.html">loss_log_cosh</a>()</code> <br><code><a href="loss_mean_absolute_error.html">loss_mean_absolute_error</a>()</code> <br><code><a href="loss_mean_absolute_percentage_error.html">loss_mean_absolute_percentage_error</a>()</code> <br><code><a href="loss_mean_squared_error.html">loss_mean_squared_error</a>()</code> <br><code><a href="loss_mean_squared_logarithmic_error.html">loss_mean_squared_logarithmic_error</a>()</code> <br><code><a href="loss_poisson.html">loss_poisson</a>()</code> <br><code><a href="loss_sparse_categorical_crossentropy.html">loss_sparse_categorical_crossentropy</a>()</code> <br><code><a href="loss_squared_hinge.html">loss_squared_hinge</a>()</code> <br><code><a href="loss_tversky.html">loss_tversky</a>()</code> <br><code><a href="metric_binary_crossentropy.html">metric_binary_crossentropy</a>()</code> <br><code><a href="metric_binary_focal_crossentropy.html">metric_binary_focal_crossentropy</a>()</code> <br><code><a href="metric_categorical_crossentropy.html">metric_categorical_crossentropy</a>()</code> <br><code><a href="metric_categorical_focal_crossentropy.html">metric_categorical_focal_crossentropy</a>()</code> <br><code><a href="metric_categorical_hinge.html">metric_categorical_hinge</a>()</code> <br><code><a href="metric_hinge.html">metric_hinge</a>()</code> <br><code><a href="metric_huber.html">metric_huber</a>()</code> <br><code><a href="metric_kl_divergence.html">metric_kl_divergence</a>()</code> <br><code><a href="metric_log_cosh.html">metric_log_cosh</a>()</code> <br><code><a href="metric_mean_absolute_error.html">metric_mean_absolute_error</a>()</code> <br><code><a href="metric_mean_absolute_percentage_error.html">metric_mean_absolute_percentage_error</a>()</code> <br><code><a href="metric_mean_squared_error.html">metric_mean_squared_error</a>()</code> <br><code><a href="metric_mean_squared_logarithmic_error.html">metric_mean_squared_logarithmic_error</a>()</code> <br><code><a href="metric_poisson.html">metric_poisson</a>()</code> <br><code><a href="metric_squared_hinge.html">metric_squared_hinge</a>()</code> <br></p>
+</ul><p>Other losses: <br><code><a href="Loss.html">Loss</a>()</code> <br><code><a href="loss_binary_crossentropy.html">loss_binary_crossentropy</a>()</code> <br><code><a href="loss_binary_focal_crossentropy.html">loss_binary_focal_crossentropy</a>()</code> <br><code><a href="loss_categorical_crossentropy.html">loss_categorical_crossentropy</a>()</code> <br><code><a href="loss_categorical_focal_crossentropy.html">loss_categorical_focal_crossentropy</a>()</code> <br><code><a href="loss_categorical_generalized_cross_entropy.html">loss_categorical_generalized_cross_entropy</a>()</code> <br><code><a href="loss_categorical_hinge.html">loss_categorical_hinge</a>()</code> <br><code><a href="loss_circle.html">loss_circle</a>()</code> <br><code><a href="loss_cosine_similarity.html">loss_cosine_similarity</a>()</code> <br><code><a href="loss_ctc.html">loss_ctc</a>()</code> <br><code><a href="loss_dice.html">loss_dice</a>()</code> <br><code><a href="loss_hinge.html">loss_hinge</a>()</code> <br><code><a href="loss_huber.html">loss_huber</a>()</code> <br><code><a href="loss_kl_divergence.html">loss_kl_divergence</a>()</code> <br><code><a href="loss_log_cosh.html">loss_log_cosh</a>()</code> <br><code><a href="loss_mean_absolute_error.html">loss_mean_absolute_error</a>()</code> <br><code><a href="loss_mean_absolute_percentage_error.html">loss_mean_absolute_percentage_error</a>()</code> <br><code><a href="loss_mean_squared_error.html">loss_mean_squared_error</a>()</code> <br><code><a href="loss_mean_squared_logarithmic_error.html">loss_mean_squared_logarithmic_error</a>()</code> <br><code><a href="loss_poisson.html">loss_poisson</a>()</code> <br><code><a href="loss_sparse_categorical_crossentropy.html">loss_sparse_categorical_crossentropy</a>()</code> <br><code><a href="loss_squared_hinge.html">loss_squared_hinge</a>()</code> <br><code><a href="loss_tversky.html">loss_tversky</a>()</code> <br><code><a href="metric_binary_crossentropy.html">metric_binary_crossentropy</a>()</code> <br><code><a href="metric_binary_focal_crossentropy.html">metric_binary_focal_crossentropy</a>()</code> <br><code><a href="metric_categorical_crossentropy.html">metric_categorical_crossentropy</a>()</code> <br><code><a href="metric_categorical_focal_crossentropy.html">metric_categorical_focal_crossentropy</a>()</code> <br><code><a href="metric_categorical_hinge.html">metric_categorical_hinge</a>()</code> <br><code><a href="metric_hinge.html">metric_hinge</a>()</code> <br><code><a href="metric_huber.html">metric_huber</a>()</code> <br><code><a href="metric_kl_divergence.html">metric_kl_divergence</a>()</code> <br><code><a href="metric_log_cosh.html">metric_log_cosh</a>()</code> <br><code><a href="metric_mean_absolute_error.html">metric_mean_absolute_error</a>()</code> <br><code><a href="metric_mean_absolute_percentage_error.html">metric_mean_absolute_percentage_error</a>()</code> <br><code><a href="metric_mean_squared_error.html">metric_mean_squared_error</a>()</code> <br><code><a href="metric_mean_squared_logarithmic_error.html">metric_mean_squared_logarithmic_error</a>()</code> <br><code><a href="metric_poisson.html">metric_poisson</a>()</code> <br><code><a href="metric_squared_hinge.html">metric_squared_hinge</a>()</code> <br></p>
 <p>Other metrics: <br><code><a href="Metric.html">Metric</a>()</code> <br><code><a href="custom_metric.html">custom_metric</a>()</code> <br><code><a href="metric_auc.html">metric_auc</a>()</code> <br><code><a href="metric_binary_accuracy.html">metric_binary_accuracy</a>()</code> <br><code><a href="metric_binary_crossentropy.html">metric_binary_crossentropy</a>()</code> <br><code><a href="metric_binary_focal_crossentropy.html">metric_binary_focal_crossentropy</a>()</code> <br><code><a href="metric_binary_iou.html">metric_binary_iou</a>()</code> <br><code><a href="metric_categorical_accuracy.html">metric_categorical_accuracy</a>()</code> <br><code><a href="metric_categorical_crossentropy.html">metric_categorical_crossentropy</a>()</code> <br><code><a href="metric_categorical_focal_crossentropy.html">metric_categorical_focal_crossentropy</a>()</code> <br><code><a href="metric_categorical_hinge.html">metric_categorical_hinge</a>()</code> <br><code><a href="metric_concordance_correlation.html">metric_concordance_correlation</a>()</code> <br><code><a href="metric_cosine_similarity.html">metric_cosine_similarity</a>()</code> <br><code><a href="metric_f1_score.html">metric_f1_score</a>()</code> <br><code><a href="metric_false_negatives.html">metric_false_negatives</a>()</code> <br><code><a href="metric_false_positives.html">metric_false_positives</a>()</code> <br><code><a href="metric_fbeta_score.html">metric_fbeta_score</a>()</code> <br><code><a href="metric_hinge.html">metric_hinge</a>()</code> <br><code><a href="metric_huber.html">metric_huber</a>()</code> <br><code><a href="metric_iou.html">metric_iou</a>()</code> <br><code><a href="metric_kl_divergence.html">metric_kl_divergence</a>()</code> <br><code><a href="metric_log_cosh.html">metric_log_cosh</a>()</code> <br><code><a href="metric_log_cosh_error.html">metric_log_cosh_error</a>()</code> <br><code><a href="metric_mean.html">metric_mean</a>()</code> <br><code><a href="metric_mean_absolute_error.html">metric_mean_absolute_error</a>()</code> <br><code><a href="metric_mean_absolute_percentage_error.html">metric_mean_absolute_percentage_error</a>()</code> <br><code><a href="metric_mean_iou.html">metric_mean_iou</a>()</code> <br><code><a href="metric_mean_squared_error.html">metric_mean_squared_error</a>()</code> <br><code><a href="metric_mean_squared_logarithmic_error.html">metric_mean_squared_logarithmic_error</a>()</code> <br><code><a href="metric_mean_wrapper.html">metric_mean_wrapper</a>()</code> <br><code><a href="metric_one_hot_iou.html">metric_one_hot_iou</a>()</code> <br><code><a href="metric_one_hot_mean_iou.html">metric_one_hot_mean_iou</a>()</code> <br><code><a href="metric_pearson_correlation.html">metric_pearson_correlation</a>()</code> <br><code><a href="metric_poisson.html">metric_poisson</a>()</code> <br><code><a href="metric_precision.html">metric_precision</a>()</code> <br><code><a href="metric_precision_at_recall.html">metric_precision_at_recall</a>()</code> <br><code><a href="metric_r2_score.html">metric_r2_score</a>()</code> <br><code><a href="metric_recall.html">metric_recall</a>()</code> <br><code><a href="metric_recall_at_precision.html">metric_recall_at_precision</a>()</code> <br><code><a href="metric_root_mean_squared_error.html">metric_root_mean_squared_error</a>()</code> <br><code><a href="metric_sensitivity_at_specificity.html">metric_sensitivity_at_specificity</a>()</code> <br><code><a href="metric_sparse_categorical_accuracy.html">metric_sparse_categorical_accuracy</a>()</code> <br><code><a href="metric_sparse_top_k_categorical_accuracy.html">metric_sparse_top_k_categorical_accuracy</a>()</code> <br><code><a href="metric_specificity_at_sensitivity.html">metric_specificity_at_sensitivity</a>()</code> <br><code><a href="metric_squared_hinge.html">metric_squared_hinge</a>()</code> <br><code><a href="metric_sum.html">metric_sum</a>()</code> <br><code><a href="metric_top_k_categorical_accuracy.html">metric_top_k_categorical_accuracy</a>()</code> <br><code><a href="metric_true_negatives.html">metric_true_negatives</a>()</code> <br><code><a href="metric_true_positives.html">metric_true_positives</a>()</code> <br></p>
 <p>Other probabilistic metrics: <br><code><a href="metric_binary_crossentropy.html">metric_binary_crossentropy</a>()</code> <br><code><a href="metric_categorical_crossentropy.html">metric_categorical_crossentropy</a>()</code> <br><code><a href="metric_kl_divergence.html">metric_kl_divergence</a>()</code> <br><code><a href="metric_poisson.html">metric_poisson</a>()</code> <br></p></div>
     </div>
@@ -178,7 +178,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/metric_sparse_top_k_categorical_accuracy.html b/docs/dev/reference/metric_sparse_top_k_categorical_accuracy.html
index 186c929ae..51211d06f 100644
--- a/docs/dev/reference/metric_sparse_top_k_categorical_accuracy.html
+++ b/docs/dev/reference/metric_sparse_top_k_categorical_accuracy.html
@@ -46,7 +46,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -213,7 +213,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/metric_specificity_at_sensitivity.html b/docs/dev/reference/metric_specificity_at_sensitivity.html
index bb52c1f88..2f8aff347 100644
--- a/docs/dev/reference/metric_specificity_at_sensitivity.html
+++ b/docs/dev/reference/metric_specificity_at_sensitivity.html
@@ -40,7 +40,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -194,7 +194,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/metric_squared_hinge.html b/docs/dev/reference/metric_squared_hinge.html
index a300d40ba..bf9e6313a 100644
--- a/docs/dev/reference/metric_squared_hinge.html
+++ b/docs/dev/reference/metric_squared_hinge.html
@@ -16,7 +16,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -126,7 +126,7 @@ <h2 id="usage">Usage<a class="anchor" aria-label="anchor" href="#usage"></a></h2
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
     <div class="dont-index">
 <ul><li><p><a href="https://keras.io/api/metrics/hinge_metrics#squaredhinge-class" class="external-link">https://keras.io/api/metrics/hinge_metrics#squaredhinge-class</a></p></li>
-</ul><p>Other losses: <br><code><a href="Loss.html">Loss</a>()</code> <br><code><a href="loss_binary_crossentropy.html">loss_binary_crossentropy</a>()</code> <br><code><a href="loss_binary_focal_crossentropy.html">loss_binary_focal_crossentropy</a>()</code> <br><code><a href="loss_categorical_crossentropy.html">loss_categorical_crossentropy</a>()</code> <br><code><a href="loss_categorical_focal_crossentropy.html">loss_categorical_focal_crossentropy</a>()</code> <br><code><a href="loss_categorical_hinge.html">loss_categorical_hinge</a>()</code> <br><code><a href="loss_circle.html">loss_circle</a>()</code> <br><code><a href="loss_cosine_similarity.html">loss_cosine_similarity</a>()</code> <br><code><a href="loss_ctc.html">loss_ctc</a>()</code> <br><code><a href="loss_dice.html">loss_dice</a>()</code> <br><code><a href="loss_hinge.html">loss_hinge</a>()</code> <br><code><a href="loss_huber.html">loss_huber</a>()</code> <br><code><a href="loss_kl_divergence.html">loss_kl_divergence</a>()</code> <br><code><a href="loss_log_cosh.html">loss_log_cosh</a>()</code> <br><code><a href="loss_mean_absolute_error.html">loss_mean_absolute_error</a>()</code> <br><code><a href="loss_mean_absolute_percentage_error.html">loss_mean_absolute_percentage_error</a>()</code> <br><code><a href="loss_mean_squared_error.html">loss_mean_squared_error</a>()</code> <br><code><a href="loss_mean_squared_logarithmic_error.html">loss_mean_squared_logarithmic_error</a>()</code> <br><code><a href="loss_poisson.html">loss_poisson</a>()</code> <br><code><a href="loss_sparse_categorical_crossentropy.html">loss_sparse_categorical_crossentropy</a>()</code> <br><code><a href="loss_squared_hinge.html">loss_squared_hinge</a>()</code> <br><code><a href="loss_tversky.html">loss_tversky</a>()</code> <br><code><a href="metric_binary_crossentropy.html">metric_binary_crossentropy</a>()</code> <br><code><a href="metric_binary_focal_crossentropy.html">metric_binary_focal_crossentropy</a>()</code> <br><code><a href="metric_categorical_crossentropy.html">metric_categorical_crossentropy</a>()</code> <br><code><a href="metric_categorical_focal_crossentropy.html">metric_categorical_focal_crossentropy</a>()</code> <br><code><a href="metric_categorical_hinge.html">metric_categorical_hinge</a>()</code> <br><code><a href="metric_hinge.html">metric_hinge</a>()</code> <br><code><a href="metric_huber.html">metric_huber</a>()</code> <br><code><a href="metric_kl_divergence.html">metric_kl_divergence</a>()</code> <br><code><a href="metric_log_cosh.html">metric_log_cosh</a>()</code> <br><code><a href="metric_mean_absolute_error.html">metric_mean_absolute_error</a>()</code> <br><code><a href="metric_mean_absolute_percentage_error.html">metric_mean_absolute_percentage_error</a>()</code> <br><code><a href="metric_mean_squared_error.html">metric_mean_squared_error</a>()</code> <br><code><a href="metric_mean_squared_logarithmic_error.html">metric_mean_squared_logarithmic_error</a>()</code> <br><code><a href="metric_poisson.html">metric_poisson</a>()</code> <br><code><a href="metric_sparse_categorical_crossentropy.html">metric_sparse_categorical_crossentropy</a>()</code> <br></p>
+</ul><p>Other losses: <br><code><a href="Loss.html">Loss</a>()</code> <br><code><a href="loss_binary_crossentropy.html">loss_binary_crossentropy</a>()</code> <br><code><a href="loss_binary_focal_crossentropy.html">loss_binary_focal_crossentropy</a>()</code> <br><code><a href="loss_categorical_crossentropy.html">loss_categorical_crossentropy</a>()</code> <br><code><a href="loss_categorical_focal_crossentropy.html">loss_categorical_focal_crossentropy</a>()</code> <br><code><a href="loss_categorical_generalized_cross_entropy.html">loss_categorical_generalized_cross_entropy</a>()</code> <br><code><a href="loss_categorical_hinge.html">loss_categorical_hinge</a>()</code> <br><code><a href="loss_circle.html">loss_circle</a>()</code> <br><code><a href="loss_cosine_similarity.html">loss_cosine_similarity</a>()</code> <br><code><a href="loss_ctc.html">loss_ctc</a>()</code> <br><code><a href="loss_dice.html">loss_dice</a>()</code> <br><code><a href="loss_hinge.html">loss_hinge</a>()</code> <br><code><a href="loss_huber.html">loss_huber</a>()</code> <br><code><a href="loss_kl_divergence.html">loss_kl_divergence</a>()</code> <br><code><a href="loss_log_cosh.html">loss_log_cosh</a>()</code> <br><code><a href="loss_mean_absolute_error.html">loss_mean_absolute_error</a>()</code> <br><code><a href="loss_mean_absolute_percentage_error.html">loss_mean_absolute_percentage_error</a>()</code> <br><code><a href="loss_mean_squared_error.html">loss_mean_squared_error</a>()</code> <br><code><a href="loss_mean_squared_logarithmic_error.html">loss_mean_squared_logarithmic_error</a>()</code> <br><code><a href="loss_poisson.html">loss_poisson</a>()</code> <br><code><a href="loss_sparse_categorical_crossentropy.html">loss_sparse_categorical_crossentropy</a>()</code> <br><code><a href="loss_squared_hinge.html">loss_squared_hinge</a>()</code> <br><code><a href="loss_tversky.html">loss_tversky</a>()</code> <br><code><a href="metric_binary_crossentropy.html">metric_binary_crossentropy</a>()</code> <br><code><a href="metric_binary_focal_crossentropy.html">metric_binary_focal_crossentropy</a>()</code> <br><code><a href="metric_categorical_crossentropy.html">metric_categorical_crossentropy</a>()</code> <br><code><a href="metric_categorical_focal_crossentropy.html">metric_categorical_focal_crossentropy</a>()</code> <br><code><a href="metric_categorical_hinge.html">metric_categorical_hinge</a>()</code> <br><code><a href="metric_hinge.html">metric_hinge</a>()</code> <br><code><a href="metric_huber.html">metric_huber</a>()</code> <br><code><a href="metric_kl_divergence.html">metric_kl_divergence</a>()</code> <br><code><a href="metric_log_cosh.html">metric_log_cosh</a>()</code> <br><code><a href="metric_mean_absolute_error.html">metric_mean_absolute_error</a>()</code> <br><code><a href="metric_mean_absolute_percentage_error.html">metric_mean_absolute_percentage_error</a>()</code> <br><code><a href="metric_mean_squared_error.html">metric_mean_squared_error</a>()</code> <br><code><a href="metric_mean_squared_logarithmic_error.html">metric_mean_squared_logarithmic_error</a>()</code> <br><code><a href="metric_poisson.html">metric_poisson</a>()</code> <br><code><a href="metric_sparse_categorical_crossentropy.html">metric_sparse_categorical_crossentropy</a>()</code> <br></p>
 <p>Other metrics: <br><code><a href="Metric.html">Metric</a>()</code> <br><code><a href="custom_metric.html">custom_metric</a>()</code> <br><code><a href="metric_auc.html">metric_auc</a>()</code> <br><code><a href="metric_binary_accuracy.html">metric_binary_accuracy</a>()</code> <br><code><a href="metric_binary_crossentropy.html">metric_binary_crossentropy</a>()</code> <br><code><a href="metric_binary_focal_crossentropy.html">metric_binary_focal_crossentropy</a>()</code> <br><code><a href="metric_binary_iou.html">metric_binary_iou</a>()</code> <br><code><a href="metric_categorical_accuracy.html">metric_categorical_accuracy</a>()</code> <br><code><a href="metric_categorical_crossentropy.html">metric_categorical_crossentropy</a>()</code> <br><code><a href="metric_categorical_focal_crossentropy.html">metric_categorical_focal_crossentropy</a>()</code> <br><code><a href="metric_categorical_hinge.html">metric_categorical_hinge</a>()</code> <br><code><a href="metric_concordance_correlation.html">metric_concordance_correlation</a>()</code> <br><code><a href="metric_cosine_similarity.html">metric_cosine_similarity</a>()</code> <br><code><a href="metric_f1_score.html">metric_f1_score</a>()</code> <br><code><a href="metric_false_negatives.html">metric_false_negatives</a>()</code> <br><code><a href="metric_false_positives.html">metric_false_positives</a>()</code> <br><code><a href="metric_fbeta_score.html">metric_fbeta_score</a>()</code> <br><code><a href="metric_hinge.html">metric_hinge</a>()</code> <br><code><a href="metric_huber.html">metric_huber</a>()</code> <br><code><a href="metric_iou.html">metric_iou</a>()</code> <br><code><a href="metric_kl_divergence.html">metric_kl_divergence</a>()</code> <br><code><a href="metric_log_cosh.html">metric_log_cosh</a>()</code> <br><code><a href="metric_log_cosh_error.html">metric_log_cosh_error</a>()</code> <br><code><a href="metric_mean.html">metric_mean</a>()</code> <br><code><a href="metric_mean_absolute_error.html">metric_mean_absolute_error</a>()</code> <br><code><a href="metric_mean_absolute_percentage_error.html">metric_mean_absolute_percentage_error</a>()</code> <br><code><a href="metric_mean_iou.html">metric_mean_iou</a>()</code> <br><code><a href="metric_mean_squared_error.html">metric_mean_squared_error</a>()</code> <br><code><a href="metric_mean_squared_logarithmic_error.html">metric_mean_squared_logarithmic_error</a>()</code> <br><code><a href="metric_mean_wrapper.html">metric_mean_wrapper</a>()</code> <br><code><a href="metric_one_hot_iou.html">metric_one_hot_iou</a>()</code> <br><code><a href="metric_one_hot_mean_iou.html">metric_one_hot_mean_iou</a>()</code> <br><code><a href="metric_pearson_correlation.html">metric_pearson_correlation</a>()</code> <br><code><a href="metric_poisson.html">metric_poisson</a>()</code> <br><code><a href="metric_precision.html">metric_precision</a>()</code> <br><code><a href="metric_precision_at_recall.html">metric_precision_at_recall</a>()</code> <br><code><a href="metric_r2_score.html">metric_r2_score</a>()</code> <br><code><a href="metric_recall.html">metric_recall</a>()</code> <br><code><a href="metric_recall_at_precision.html">metric_recall_at_precision</a>()</code> <br><code><a href="metric_root_mean_squared_error.html">metric_root_mean_squared_error</a>()</code> <br><code><a href="metric_sensitivity_at_specificity.html">metric_sensitivity_at_specificity</a>()</code> <br><code><a href="metric_sparse_categorical_accuracy.html">metric_sparse_categorical_accuracy</a>()</code> <br><code><a href="metric_sparse_categorical_crossentropy.html">metric_sparse_categorical_crossentropy</a>()</code> <br><code><a href="metric_sparse_top_k_categorical_accuracy.html">metric_sparse_top_k_categorical_accuracy</a>()</code> <br><code><a href="metric_specificity_at_sensitivity.html">metric_specificity_at_sensitivity</a>()</code> <br><code><a href="metric_sum.html">metric_sum</a>()</code> <br><code><a href="metric_top_k_categorical_accuracy.html">metric_top_k_categorical_accuracy</a>()</code> <br><code><a href="metric_true_negatives.html">metric_true_negatives</a>()</code> <br><code><a href="metric_true_positives.html">metric_true_positives</a>()</code> <br></p>
 <p>Other hinge metrics: <br><code><a href="metric_categorical_hinge.html">metric_categorical_hinge</a>()</code> <br><code><a href="metric_hinge.html">metric_hinge</a>()</code> <br></p></div>
     </div>
@@ -140,7 +140,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/metric_sum.html b/docs/dev/reference/metric_sum.html
index df54dae3a..a568206a2 100644
--- a/docs/dev/reference/metric_sum.html
+++ b/docs/dev/reference/metric_sum.html
@@ -14,7 +14,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -121,7 +121,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/metric_top_k_categorical_accuracy.html b/docs/dev/reference/metric_top_k_categorical_accuracy.html
index e0813646f..17ac0705e 100644
--- a/docs/dev/reference/metric_top_k_categorical_accuracy.html
+++ b/docs/dev/reference/metric_top_k_categorical_accuracy.html
@@ -8,7 +8,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -147,7 +147,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/metric_true_negatives.html b/docs/dev/reference/metric_true_negatives.html
index 9f80258ed..5211f9540 100644
--- a/docs/dev/reference/metric_true_negatives.html
+++ b/docs/dev/reference/metric_true_negatives.html
@@ -16,7 +16,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -137,7 +137,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/metric_true_positives.html b/docs/dev/reference/metric_true_positives.html
index 1b5568d03..e9f7c44fb 100644
--- a/docs/dev/reference/metric_true_positives.html
+++ b/docs/dev/reference/metric_true_positives.html
@@ -16,7 +16,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -137,7 +137,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/multi-assign.html b/docs/dev/reference/multi-assign.html
index 80190a6c0..aaabc8b83 100644
--- a/docs/dev/reference/multi-assign.html
+++ b/docs/dev/reference/multi-assign.html
@@ -8,7 +8,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -65,7 +65,7 @@ <h2 id="ref-usage">Usage<a class="anchor" aria-label="anchor" href="#ref-usage">
     <div class="section level2">
     <h2 id="value">Value<a class="anchor" aria-label="anchor" href="#value"></a></h2>
     <p>The right-hand-side argument, <code>value</code>, invisibly. This called
-primarily for it's side-effect of assigning symbols in the current frame.</p>
+primarily for its side-effect of assigning symbols in the current frame.</p>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -77,7 +77,7 @@ <h2 id="value">Value<a class="anchor" aria-label="anchor" href="#value"></a></h2
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/new_callback_class.html b/docs/dev/reference/new_callback_class.html
index bcf6df64f..09e82b0dc 100644
--- a/docs/dev/reference/new_callback_class.html
+++ b/docs/dev/reference/new_callback_class.html
@@ -10,7 +10,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -300,7 +300,7 @@ <h2 id="value">Value<a class="anchor" aria-label="anchor" href="#value"></a></h2
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/new_layer_class.html b/docs/dev/reference/new_layer_class.html
index a6a17fc81..876996e0f 100644
--- a/docs/dev/reference/new_layer_class.html
+++ b/docs/dev/reference/new_layer_class.html
@@ -10,7 +10,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -137,7 +137,7 @@ <h2 id="value">Value<a class="anchor" aria-label="anchor" href="#value"></a></h2
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/new_learning_rate_schedule_class.html b/docs/dev/reference/new_learning_rate_schedule_class.html
index 0d326402f..716ed1ad1 100644
--- a/docs/dev/reference/new_learning_rate_schedule_class.html
+++ b/docs/dev/reference/new_learning_rate_schedule_class.html
@@ -10,7 +10,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -123,7 +123,7 @@ <h2 id="value">Value<a class="anchor" aria-label="anchor" href="#value"></a></h2
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/new_loss_class.html b/docs/dev/reference/new_loss_class.html
index 5c912f509..c331769a2 100644
--- a/docs/dev/reference/new_loss_class.html
+++ b/docs/dev/reference/new_loss_class.html
@@ -10,7 +10,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -125,7 +125,7 @@ <h2 id="value">Value<a class="anchor" aria-label="anchor" href="#value"></a></h2
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/new_metric_class.html b/docs/dev/reference/new_metric_class.html
index 658290c4c..96df0ffb9 100644
--- a/docs/dev/reference/new_metric_class.html
+++ b/docs/dev/reference/new_metric_class.html
@@ -10,7 +10,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -123,7 +123,7 @@ <h2 id="value">Value<a class="anchor" aria-label="anchor" href="#value"></a></h2
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/new_model_class.html b/docs/dev/reference/new_model_class.html
index 2ec3bdc25..9a3da1adf 100644
--- a/docs/dev/reference/new_model_class.html
+++ b/docs/dev/reference/new_model_class.html
@@ -10,7 +10,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -127,7 +127,7 @@ <h2 id="value">Value<a class="anchor" aria-label="anchor" href="#value"></a></h2
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/newaxis.html b/docs/dev/reference/newaxis.html
index 3c4298c3b..79647a1cd 100644
--- a/docs/dev/reference/newaxis.html
+++ b/docs/dev/reference/newaxis.html
@@ -10,7 +10,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -93,7 +93,7 @@ <h2 id="details">Details<a class="anchor" aria-label="anchor" href="#details"></
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/normalize.html b/docs/dev/reference/normalize.html
index b40444dd4..a00bdb808 100644
--- a/docs/dev/reference/normalize.html
+++ b/docs/dev/reference/normalize.html
@@ -10,7 +10,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -102,7 +102,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/op_abs.html b/docs/dev/reference/op_abs.html
index 8dd916c49..023c6e13f 100644
--- a/docs/dev/reference/op_abs.html
+++ b/docs/dev/reference/op_abs.html
@@ -8,7 +8,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -83,8 +83,8 @@ <h2 id="example">Example<a class="anchor" aria-label="anchor" href="#example"></
     </div>
     <div class="section level2">
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
-    <div class="dont-index"><p>Other numpy ops: <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p>
-<p>Other ops: <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
+    <div class="dont-index"><p>Other numpy ops: <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p>
+<p>Other ops: <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_layer_normalization.html">op_layer_normalization</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparse_sigmoid.html">op_sparse_sigmoid</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -96,7 +96,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/op_add.html b/docs/dev/reference/op_add.html
index 5831f8c44..2e1532cc6 100644
--- a/docs/dev/reference/op_add.html
+++ b/docs/dev/reference/op_add.html
@@ -8,7 +8,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -110,8 +110,8 @@ <h2 id="examples">Examples<a class="anchor" aria-label="anchor" href="#examples"
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
     <div class="dont-index">
 <ul><li><p><a href="https://keras.io/api/ops/numpy#add-function" class="external-link">https://keras.io/api/ops/numpy#add-function</a></p></li>
-</ul><p>Other numpy ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p>
-<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
+</ul><p>Other numpy ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p>
+<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_layer_normalization.html">op_layer_normalization</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparse_sigmoid.html">op_sparse_sigmoid</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -123,7 +123,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/op_all.html b/docs/dev/reference/op_all.html
index 46364da99..db9ce987d 100644
--- a/docs/dev/reference/op_all.html
+++ b/docs/dev/reference/op_all.html
@@ -8,7 +8,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -112,8 +112,8 @@ <h2 id="examples">Examples<a class="anchor" aria-label="anchor" href="#examples"
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
     <div class="dont-index">
 <ul><li><p><a href="https://keras.io/api/ops/numpy#all-function" class="external-link">https://keras.io/api/ops/numpy#all-function</a></p></li>
-</ul><p>Other numpy ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p>
-<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
+</ul><p>Other numpy ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p>
+<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_layer_normalization.html">op_layer_normalization</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparse_sigmoid.html">op_sparse_sigmoid</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -125,7 +125,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/op_angle.html b/docs/dev/reference/op_angle.html
new file mode 100644
index 000000000..427cdc3cf
--- /dev/null
+++ b/docs/dev/reference/op_angle.html
@@ -0,0 +1,111 @@
+<!DOCTYPE html>
+<!-- Generated by pkgdown: do not edit by hand --><html lang="en"><head><meta http-equiv="Content-Type" content="text/html; charset=UTF-8"><meta charset="utf-8"><meta http-equiv="X-UA-Compatible" content="IE=edge"><meta name="viewport" content="width=device-width, initial-scale=1, shrink-to-fit=no"><title>Element-wise angle of a complex tensor. — op_angle • keras3</title><!-- favicons --><link rel="icon" type="image/png" sizes="96x96" href="../favicon-96x96.png"><link rel="icon" type="”image/svg+xml”" href="../favicon.svg"><link rel="apple-touch-icon" sizes="180x180" href="../apple-touch-icon.png"><link rel="icon" sizes="any" href="../favicon.ico"><link rel="manifest" href="../site.webmanifest"><script src="../deps/jquery-3.6.0/jquery-3.6.0.min.js"></script><meta name="viewport" content="width=device-width, initial-scale=1, shrink-to-fit=no"><link href="../deps/bootstrap-5.3.1/bootstrap.min.css" rel="stylesheet"><script src="../deps/bootstrap-5.3.1/bootstrap.bundle.min.js"></script><link href="../deps/font-awesome-6.5.2/css/all.min.css" rel="stylesheet"><link href="../deps/font-awesome-6.5.2/css/v4-shims.min.css" rel="stylesheet"><script src="../deps/headroom-0.11.0/headroom.min.js"></script><script src="../deps/headroom-0.11.0/jQuery.headroom.min.js"></script><script src="../deps/bootstrap-toc-1.0.1/bootstrap-toc.min.js"></script><script src="../deps/clipboard.js-2.0.11/clipboard.min.js"></script><script src="../deps/search-1.0.0/autocomplete.jquery.min.js"></script><script src="../deps/search-1.0.0/fuse.min.js"></script><script src="../deps/search-1.0.0/mark.min.js"></script><!-- pkgdown --><script src="../pkgdown.js"></script><link href="../extra.css" rel="stylesheet"><meta property="og:title" content="Element-wise angle of a complex tensor. — op_angle"><meta name="description" content="Returns the phase angle (in radians) of each element in x."><meta property="og:description" content="Returns the phase angle (in radians) of each element in x."><meta name="robots" content="noindex"><!-- Google Tag Manager --><script>(function(w,d,s,l,i){w[l]=w[l]||[];w[l].push({'gtm.start': new Date().getTime(),event:'gtm.js'});var f=d.getElementsByTagName(s)[0],j=d.createElement(s),dl=l!='dataLayer'?'&l='+l:'';j.async=true;j.src='https://www.googletagmanager.com/gtm.js?id='+i+dl;f.parentNode.insertBefore(j,f);})(window,document,'script','dataLayer','GTM-KHBDBW7');</script><!-- End Google Tag Manager --></head><body>
+    <a href="#main" class="visually-hidden-focusable">Skip to contents</a>
+    <!-- Google Tag Manager (noscript) --> <noscript><iframe src="https://www.googletagmanager.com/ns.html?id=GTM-KHBDBW7" height="0" width="0" style="display:none;visibility:hidden"></iframe></noscript> <!-- End Google Tag Manager (noscript) -->
+
+
+    <nav class="navbar navbar-expand-lg fixed-top bg-primary" data-bs-theme="inverse" aria-label="Site navigation"><div class="container">
+
+    <a class="navbar-brand me-2" href="../index.html">keras3</a>
+
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
+
+
+    <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
+      <span class="navbar-toggler-icon"></span>
+    </button>
+
+    <div id="navbar" class="collapse navbar-collapse ms-3">
+      <ul class="navbar-nav me-auto"><li class="nav-item"><a class="nav-link" href="../articles/getting_started.html">Getting Started</a></li>
+<li class="nav-item dropdown">
+  <button class="nav-link dropdown-toggle" type="button" id="dropdown-guides" data-bs-toggle="dropdown" aria-expanded="false" aria-haspopup="true">Guides</button>
+  <ul class="dropdown-menu" aria-labelledby="dropdown-guides"><li><h6 class="dropdown-header" data-toc-skip>Model definition</h6></li>
+    <li><a class="dropdown-item" href="../articles/sequential_model.html">Sequential Model</a></li>
+    <li><a class="dropdown-item" href="../articles/functional_api.html">Functional API</a></li>
+    <li><h6 class="dropdown-header" data-toc-skip>Extending and customizing</h6></li>
+    <li><a class="dropdown-item" href="../articles/training_with_built_in_methods.html">Training &amp; evaluation with the built-in methods</a></li>
+    <li><a class="dropdown-item" href="../articles/custom_train_step_in_tensorflow.html">Customizing `fit()` with Tensorflow</a></li>
+    <li><a class="dropdown-item" href="../articles/writing_your_own_callbacks.html">Writing your own callbacks</a></li>
+    <li><a class="dropdown-item" href="../articles/making_new_layers_and_models_via_subclassing.html">Making new layers and models via subclassing</a></li>
+    <li><a class="dropdown-item" href="../articles/writing_a_custom_training_loop_in_tensorflow.html">Writing a training loop from scratch in TensorFlow</a></li>
+    <li><a class="dropdown-item" href="../articles/serialization_and_saving.html">Serialization and Saving</a></li>
+    <li><h6 class="dropdown-header" data-toc-skip>Other topics</h6></li>
+    <li><a class="dropdown-item" href="../articles/transfer_learning.html">Transfer learning and fine tuning</a></li>
+    <li><a class="dropdown-item" href="../articles/distributed_training_with_tensorflow.html">Distributed training with TensorFlow</a></li>
+    <li><a class="dropdown-item" href="../articles/distribution.html">Distributed training with Jax</a></li>
+  </ul></li>
+<li class="nav-item"><a class="nav-link" href="../articles/examples/index.html">Examples</a></li>
+<li class="active nav-item"><a class="nav-link" href="../reference/index.html">Reference</a></li>
+<li class="nav-item"><a class="nav-link" href="../news/index.html">News</a></li>
+      </ul><ul class="navbar-nav"><li class="nav-item"><form class="form-inline" role="search">
+ <input class="form-control" type="search" name="search-input" id="search-input" autocomplete="off" aria-label="Search site" placeholder="Search for" data-search-index="../search.json"></form></li>
+<li class="nav-item"><a class="external-link nav-link" href="https://github.com/rstudio/keras3/" aria-label="GitHub"><span class="fa fab fa-github fa-lg"></span></a></li>
+      </ul></div>
+
+
+  </div>
+</nav><div class="container template-reference-topic">
+<div class="row">
+  <main id="main" class="col-md-9"><div class="page-header">
+
+      <h1>Element-wise angle of a complex tensor.</h1>
+      <small class="dont-index">Source: <a href="https://github.com/rstudio/keras3/blob/HEAD/R/ops.R" class="external-link"><code>R/ops.R</code></a></small>
+      <div class="d-none name"><code>op_angle.Rd</code></div>
+    </div>
+
+    <div class="ref-description section level2">
+    <p>Returns the phase angle (in radians) of each element in <code>x</code>.</p>
+    </div>
+
+    <div class="section level2">
+    <h2 id="ref-usage">Usage<a class="anchor" aria-label="anchor" href="#ref-usage"></a></h2>
+    <div class="sourceCode"><pre class="sourceCode r"><code><span><span class="fu">op_angle</span><span class="op">(</span><span class="va">x</span><span class="op">)</span></span></code></pre></div>
+    </div>
+
+    <div class="section level2">
+    <h2 id="arguments">Arguments<a class="anchor" aria-label="anchor" href="#arguments"></a></h2>
+
+
+<dl><dt id="arg-x">x<a class="anchor" aria-label="anchor" href="#arg-x"></a></dt>
+<dd><p>Input tensor. Can be real or complex.</p></dd>
+
+</dl></div>
+    <div class="section level2">
+    <h2 id="value">Value<a class="anchor" aria-label="anchor" href="#value"></a></h2>
+    <p>Tensor with the same shape as <code>x</code>, containing element-wise angles.</p>
+    </div>
+    <div class="section level2">
+    <h2 id="examples">Examples<a class="anchor" aria-label="anchor" href="#examples"></a></h2>
+    <p></p><div class="sourceCode r"><pre><code><span><span class="va">x</span> <span class="op">&lt;-</span> <span class="fu"><a href="../reference/op_convert_to_tensor.html">op_convert_to_tensor</a></span><span class="op">(</span><span class="fu"><a href="https://rdrr.io/r/base/matrix.html" class="external-link">matrix</a></span><span class="op">(</span><span class="fu"><a href="https://rdrr.io/r/base/c.html" class="external-link">c</a></span><span class="op">(</span><span class="fl">1</span> <span class="op">+</span> <span class="fl">3i</span>, <span class="fl">2</span> <span class="op">-</span> <span class="fl">5i</span>, <span class="fl">4</span> <span class="op">-</span> <span class="fl">3i</span>, <span class="fl">3</span> <span class="op">+</span> <span class="fl">2i</span><span class="op">)</span>, nrow <span class="op">=</span> <span class="fl">2</span><span class="op">)</span><span class="op">)</span></span>
+<span><span class="fu"><a href="../reference/op_angle.html">op_angle</a></span><span class="op">(</span><span class="va">x</span><span class="op">)</span></span></code></pre><p></p></div>
+<p></p><div class="sourceCode"><pre><code><span id="cb1-1"><a href="#cb1-1" tabindex="-1"></a><span class="do">## tf.Tensor(</span></span>
+<span id="cb1-2"><a href="#cb1-2" tabindex="-1"></a><span class="do">## [[ 1.24904577 -0.64350111]</span></span>
+<span id="cb1-3"><a href="#cb1-3" tabindex="-1"></a><span class="do">##  [-1.19028995  0.5880026 ]], shape=(2, 2), dtype=float64)</span></span>
+<span id="cb1-4"><a href="#cb1-4" tabindex="-1"></a></span></code></pre><p></p></div>
+    </div>
+    <div class="section level2">
+    <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
+    <div class="dont-index"><p>Other numpy ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p>
+<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_layer_normalization.html">op_layer_normalization</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparse_sigmoid.html">op_sparse_sigmoid</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
+    </div>
+
+  </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
+    </nav></aside></div>
+
+
+    <footer><div class="pkgdown-footer-left">
+  <p>Developed by Tomasz Kalinowski, JJ Allaire, François Chollet, Posit Software, PBC, Google.</p>
+</div>
+
+<div class="pkgdown-footer-right">
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
+</div>
+
+    </footer></div>
+
+
+
+
+
+  </body></html>
+
diff --git a/docs/dev/reference/op_any.html b/docs/dev/reference/op_any.html
index e9d51765e..06594ca55 100644
--- a/docs/dev/reference/op_any.html
+++ b/docs/dev/reference/op_any.html
@@ -8,7 +8,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -122,8 +122,8 @@ <h2 id="examples">Examples<a class="anchor" aria-label="anchor" href="#examples"
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
     <div class="dont-index">
 <ul><li><p><a href="https://keras.io/api/ops/numpy#any-function" class="external-link">https://keras.io/api/ops/numpy#any-function</a></p></li>
-</ul><p>Other numpy ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p>
-<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
+</ul><p>Other numpy ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p>
+<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_layer_normalization.html">op_layer_normalization</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparse_sigmoid.html">op_sparse_sigmoid</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -135,7 +135,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/op_append.html b/docs/dev/reference/op_append.html
index f0bb73881..b4c86e96e 100644
--- a/docs/dev/reference/op_append.html
+++ b/docs/dev/reference/op_append.html
@@ -8,7 +8,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -109,8 +109,8 @@ <h2 id="examples">Examples<a class="anchor" aria-label="anchor" href="#examples"
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
     <div class="dont-index">
 <ul><li><p><a href="https://keras.io/api/ops/numpy#append-function" class="external-link">https://keras.io/api/ops/numpy#append-function</a></p></li>
-</ul><p>Other numpy ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p>
-<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
+</ul><p>Other numpy ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p>
+<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_layer_normalization.html">op_layer_normalization</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparse_sigmoid.html">op_sparse_sigmoid</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -122,7 +122,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/op_arange.html b/docs/dev/reference/op_arange.html
index 4f0dac01a..2b6cb8efa 100644
--- a/docs/dev/reference/op_arange.html
+++ b/docs/dev/reference/op_arange.html
@@ -24,7 +24,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -145,8 +145,8 @@ <h2 id="examples">Examples<a class="anchor" aria-label="anchor" href="#examples"
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
     <div class="dont-index">
 <ul><li><p><a href="https://keras.io/api/ops/numpy#arange-function" class="external-link">https://keras.io/api/ops/numpy#arange-function</a></p></li>
-</ul><p>Other numpy ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p>
-<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
+</ul><p>Other numpy ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p>
+<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_layer_normalization.html">op_layer_normalization</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparse_sigmoid.html">op_sparse_sigmoid</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -158,7 +158,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/op_arccos.html b/docs/dev/reference/op_arccos.html
index 3cb868f6f..4a2e178f0 100644
--- a/docs/dev/reference/op_arccos.html
+++ b/docs/dev/reference/op_arccos.html
@@ -8,7 +8,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -86,8 +86,8 @@ <h2 id="examples">Examples<a class="anchor" aria-label="anchor" href="#examples"
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
     <div class="dont-index">
 <ul><li><p><a href="https://keras.io/api/ops/numpy#arccos-function" class="external-link">https://keras.io/api/ops/numpy#arccos-function</a></p></li>
-</ul><p>Other numpy ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p>
-<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
+</ul><p>Other numpy ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p>
+<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_layer_normalization.html">op_layer_normalization</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparse_sigmoid.html">op_sparse_sigmoid</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -99,7 +99,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/op_arccosh.html b/docs/dev/reference/op_arccosh.html
index 53a3e4183..183c30192 100644
--- a/docs/dev/reference/op_arccosh.html
+++ b/docs/dev/reference/op_arccosh.html
@@ -8,7 +8,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -83,8 +83,8 @@ <h2 id="examples">Examples<a class="anchor" aria-label="anchor" href="#examples"
     </div>
     <div class="section level2">
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
-    <div class="dont-index"><p>Other numpy ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p>
-<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
+    <div class="dont-index"><p>Other numpy ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p>
+<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_layer_normalization.html">op_layer_normalization</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparse_sigmoid.html">op_sparse_sigmoid</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -96,7 +96,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/op_arcsin.html b/docs/dev/reference/op_arcsin.html
index 89b2614dc..b217370ab 100644
--- a/docs/dev/reference/op_arcsin.html
+++ b/docs/dev/reference/op_arcsin.html
@@ -8,7 +8,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -86,8 +86,8 @@ <h2 id="examples">Examples<a class="anchor" aria-label="anchor" href="#examples"
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
     <div class="dont-index">
 <ul><li><p><a href="https://keras.io/api/ops/numpy#arcsin-function" class="external-link">https://keras.io/api/ops/numpy#arcsin-function</a></p></li>
-</ul><p>Other numpy ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p>
-<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
+</ul><p>Other numpy ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p>
+<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_layer_normalization.html">op_layer_normalization</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparse_sigmoid.html">op_sparse_sigmoid</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -99,7 +99,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/op_arcsinh.html b/docs/dev/reference/op_arcsinh.html
index 5b469dcff..c3e030aa4 100644
--- a/docs/dev/reference/op_arcsinh.html
+++ b/docs/dev/reference/op_arcsinh.html
@@ -8,7 +8,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -85,8 +85,8 @@ <h2 id="examples">Examples<a class="anchor" aria-label="anchor" href="#examples"
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
     <div class="dont-index">
 <ul><li><p><a href="https://keras.io/api/ops/numpy#arcsinh-function" class="external-link">https://keras.io/api/ops/numpy#arcsinh-function</a></p></li>
-</ul><p>Other numpy ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p>
-<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
+</ul><p>Other numpy ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p>
+<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_layer_normalization.html">op_layer_normalization</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparse_sigmoid.html">op_sparse_sigmoid</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -98,7 +98,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/op_arctan.html b/docs/dev/reference/op_arctan.html
index a1ba10575..9945641ce 100644
--- a/docs/dev/reference/op_arctan.html
+++ b/docs/dev/reference/op_arctan.html
@@ -8,7 +8,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -86,8 +86,8 @@ <h2 id="examples">Examples<a class="anchor" aria-label="anchor" href="#examples"
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
     <div class="dont-index">
 <ul><li><p><a href="https://keras.io/api/ops/numpy#arctan-function" class="external-link">https://keras.io/api/ops/numpy#arctan-function</a></p></li>
-</ul><p>Other numpy ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p>
-<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
+</ul><p>Other numpy ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p>
+<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_layer_normalization.html">op_layer_normalization</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparse_sigmoid.html">op_sparse_sigmoid</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -99,7 +99,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/op_arctan2.html b/docs/dev/reference/op_arctan2.html
index 554bee0fb..0314e5601 100644
--- a/docs/dev/reference/op_arctan2.html
+++ b/docs/dev/reference/op_arctan2.html
@@ -20,7 +20,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -123,8 +123,8 @@ <h2 id="examples">Examples<a class="anchor" aria-label="anchor" href="#examples"
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
     <div class="dont-index">
 <ul><li><p><a href="https://keras.io/api/ops/numpy#arctan2-function" class="external-link">https://keras.io/api/ops/numpy#arctan2-function</a></p></li>
-</ul><p>Other numpy ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p>
-<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
+</ul><p>Other numpy ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p>
+<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_layer_normalization.html">op_layer_normalization</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparse_sigmoid.html">op_sparse_sigmoid</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -136,7 +136,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/op_arctanh.html b/docs/dev/reference/op_arctanh.html
index a3526634c..22a3c94b8 100644
--- a/docs/dev/reference/op_arctanh.html
+++ b/docs/dev/reference/op_arctanh.html
@@ -8,7 +8,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -78,8 +78,8 @@ <h2 id="value">Value<a class="anchor" aria-label="anchor" href="#value"></a></h2
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
     <div class="dont-index">
 <ul><li><p><a href="https://keras.io/api/ops/numpy#arctanh-function" class="external-link">https://keras.io/api/ops/numpy#arctanh-function</a></p></li>
-</ul><p>Other numpy ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p>
-<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
+</ul><p>Other numpy ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p>
+<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_layer_normalization.html">op_layer_normalization</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparse_sigmoid.html">op_sparse_sigmoid</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -91,7 +91,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/op_argmax.html b/docs/dev/reference/op_argmax.html
index 6b47682fc..09179abbc 100644
--- a/docs/dev/reference/op_argmax.html
+++ b/docs/dev/reference/op_argmax.html
@@ -8,7 +8,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -120,8 +120,8 @@ <h2 id="examples">Examples<a class="anchor" aria-label="anchor" href="#examples"
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
     <div class="dont-index">
 <ul><li><p><a href="https://keras.io/api/ops/numpy#argmax-function" class="external-link">https://keras.io/api/ops/numpy#argmax-function</a></p></li>
-</ul><p>Other numpy ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p>
-<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
+</ul><p>Other numpy ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p>
+<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_layer_normalization.html">op_layer_normalization</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparse_sigmoid.html">op_sparse_sigmoid</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -133,7 +133,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/op_argmin.html b/docs/dev/reference/op_argmin.html
index 412bfae6d..5e1d32ede 100644
--- a/docs/dev/reference/op_argmin.html
+++ b/docs/dev/reference/op_argmin.html
@@ -8,7 +8,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -118,8 +118,8 @@ <h2 id="examples">Examples<a class="anchor" aria-label="anchor" href="#examples"
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
     <div class="dont-index">
 <ul><li><p><a href="https://keras.io/api/ops/numpy#argmin-function" class="external-link">https://keras.io/api/ops/numpy#argmin-function</a></p></li>
-</ul><p>Other numpy ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p>
-<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
+</ul><p>Other numpy ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p>
+<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_layer_normalization.html">op_layer_normalization</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparse_sigmoid.html">op_sparse_sigmoid</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -131,7 +131,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/op_argpartition.html b/docs/dev/reference/op_argpartition.html
index 5e77fcba2..0871ef764 100644
--- a/docs/dev/reference/op_argpartition.html
+++ b/docs/dev/reference/op_argpartition.html
@@ -44,7 +44,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -144,8 +144,8 @@ <h2 id="value">Value<a class="anchor" aria-label="anchor" href="#value"></a></h2
     </div>
     <div class="section level2">
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
-    <div class="dont-index"><p>Other numpy ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p>
-<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
+    <div class="dont-index"><p>Other numpy ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p>
+<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_layer_normalization.html">op_layer_normalization</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparse_sigmoid.html">op_sparse_sigmoid</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -157,7 +157,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/op_argsort.html b/docs/dev/reference/op_argsort.html
index 31ecab703..8b6db4b20 100644
--- a/docs/dev/reference/op_argsort.html
+++ b/docs/dev/reference/op_argsort.html
@@ -8,7 +8,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -134,8 +134,8 @@ <h2 id="examples">Examples<a class="anchor" aria-label="anchor" href="#examples"
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
     <div class="dont-index">
 <ul><li><p><a href="https://keras.io/api/ops/numpy#argsort-function" class="external-link">https://keras.io/api/ops/numpy#argsort-function</a></p></li>
-</ul><p>Other numpy ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p>
-<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
+</ul><p>Other numpy ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p>
+<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_layer_normalization.html">op_layer_normalization</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparse_sigmoid.html">op_sparse_sigmoid</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -147,7 +147,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/op_array.html b/docs/dev/reference/op_array.html
index 6c0f79c06..349359bfc 100644
--- a/docs/dev/reference/op_array.html
+++ b/docs/dev/reference/op_array.html
@@ -8,7 +8,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -94,8 +94,8 @@ <h2 id="examples">Examples<a class="anchor" aria-label="anchor" href="#examples"
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
     <div class="dont-index">
 <ul><li><p><a href="https://keras.io/api/ops/numpy#array-function" class="external-link">https://keras.io/api/ops/numpy#array-function</a></p></li>
-</ul><p>Other numpy ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p>
-<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
+</ul><p>Other numpy ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p>
+<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_layer_normalization.html">op_layer_normalization</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparse_sigmoid.html">op_sparse_sigmoid</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -107,7 +107,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/op_associative_scan.html b/docs/dev/reference/op_associative_scan.html
index 166d144ed..dcacc66eb 100644
--- a/docs/dev/reference/op_associative_scan.html
+++ b/docs/dev/reference/op_associative_scan.html
@@ -22,7 +22,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -163,7 +163,7 @@ <h2 id="examples">Examples<a class="anchor" aria-label="anchor" href="#examples"
     <div class="section level2">
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
     <div class="dont-index"><p>Other core ops: <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br></p>
-<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
+<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_layer_normalization.html">op_layer_normalization</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparse_sigmoid.html">op_sparse_sigmoid</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -175,7 +175,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/op_average.html b/docs/dev/reference/op_average.html
index d2660c1be..742f8c012 100644
--- a/docs/dev/reference/op_average.html
+++ b/docs/dev/reference/op_average.html
@@ -8,7 +8,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -133,8 +133,8 @@ <h2 id="examples">Examples<a class="anchor" aria-label="anchor" href="#examples"
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
     <div class="dont-index">
 <ul><li><p><a href="https://keras.io/api/ops/numpy#average-function" class="external-link">https://keras.io/api/ops/numpy#average-function</a></p></li>
-</ul><p>Other numpy ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p>
-<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
+</ul><p>Other numpy ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p>
+<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_layer_normalization.html">op_layer_normalization</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparse_sigmoid.html">op_sparse_sigmoid</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -146,7 +146,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/op_average_pool.html b/docs/dev/reference/op_average_pool.html
index bac403cda..8826aab9c 100644
--- a/docs/dev/reference/op_average_pool.html
+++ b/docs/dev/reference/op_average_pool.html
@@ -8,7 +8,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -120,8 +120,8 @@ <h2 id="value">Value<a class="anchor" aria-label="anchor" href="#value"></a></h2
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
     <div class="dont-index">
 <ul><li><p><a href="https://keras.io/api/ops/nn#averagepool-function" class="external-link">https://keras.io/api/ops/nn#averagepool-function</a></p></li>
-</ul><p>Other nn ops: <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br></p>
-<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
+</ul><p>Other nn ops: <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_layer_normalization.html">op_layer_normalization</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparse_sigmoid.html">op_sparse_sigmoid</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br></p>
+<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_layer_normalization.html">op_layer_normalization</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparse_sigmoid.html">op_sparse_sigmoid</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -133,7 +133,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/op_bartlett.html b/docs/dev/reference/op_bartlett.html
new file mode 100644
index 000000000..8bd1dd0db
--- /dev/null
+++ b/docs/dev/reference/op_bartlett.html
@@ -0,0 +1,108 @@
+<!DOCTYPE html>
+<!-- Generated by pkgdown: do not edit by hand --><html lang="en"><head><meta http-equiv="Content-Type" content="text/html; charset=UTF-8"><meta charset="utf-8"><meta http-equiv="X-UA-Compatible" content="IE=edge"><meta name="viewport" content="width=device-width, initial-scale=1, shrink-to-fit=no"><title>Bartlett window function. — op_bartlett • keras3</title><!-- favicons --><link rel="icon" type="image/png" sizes="96x96" href="../favicon-96x96.png"><link rel="icon" type="”image/svg+xml”" href="../favicon.svg"><link rel="apple-touch-icon" sizes="180x180" href="../apple-touch-icon.png"><link rel="icon" sizes="any" href="../favicon.ico"><link rel="manifest" href="../site.webmanifest"><script src="../deps/jquery-3.6.0/jquery-3.6.0.min.js"></script><meta name="viewport" content="width=device-width, initial-scale=1, shrink-to-fit=no"><link href="../deps/bootstrap-5.3.1/bootstrap.min.css" rel="stylesheet"><script src="../deps/bootstrap-5.3.1/bootstrap.bundle.min.js"></script><link href="../deps/font-awesome-6.5.2/css/all.min.css" rel="stylesheet"><link href="../deps/font-awesome-6.5.2/css/v4-shims.min.css" rel="stylesheet"><script src="../deps/headroom-0.11.0/headroom.min.js"></script><script src="../deps/headroom-0.11.0/jQuery.headroom.min.js"></script><script src="../deps/bootstrap-toc-1.0.1/bootstrap-toc.min.js"></script><script src="../deps/clipboard.js-2.0.11/clipboard.min.js"></script><script src="../deps/search-1.0.0/autocomplete.jquery.min.js"></script><script src="../deps/search-1.0.0/fuse.min.js"></script><script src="../deps/search-1.0.0/mark.min.js"></script><!-- pkgdown --><script src="../pkgdown.js"></script><link href="../extra.css" rel="stylesheet"><meta property="og:title" content="Bartlett window function. — op_bartlett"><meta name="description" content="The Bartlett window is a triangular window that rises then falls linearly."><meta property="og:description" content="The Bartlett window is a triangular window that rises then falls linearly."><meta name="robots" content="noindex"><!-- Google Tag Manager --><script>(function(w,d,s,l,i){w[l]=w[l]||[];w[l].push({'gtm.start': new Date().getTime(),event:'gtm.js'});var f=d.getElementsByTagName(s)[0],j=d.createElement(s),dl=l!='dataLayer'?'&l='+l:'';j.async=true;j.src='https://www.googletagmanager.com/gtm.js?id='+i+dl;f.parentNode.insertBefore(j,f);})(window,document,'script','dataLayer','GTM-KHBDBW7');</script><!-- End Google Tag Manager --></head><body>
+    <a href="#main" class="visually-hidden-focusable">Skip to contents</a>
+    <!-- Google Tag Manager (noscript) --> <noscript><iframe src="https://www.googletagmanager.com/ns.html?id=GTM-KHBDBW7" height="0" width="0" style="display:none;visibility:hidden"></iframe></noscript> <!-- End Google Tag Manager (noscript) -->
+
+
+    <nav class="navbar navbar-expand-lg fixed-top bg-primary" data-bs-theme="inverse" aria-label="Site navigation"><div class="container">
+
+    <a class="navbar-brand me-2" href="../index.html">keras3</a>
+
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
+
+
+    <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
+      <span class="navbar-toggler-icon"></span>
+    </button>
+
+    <div id="navbar" class="collapse navbar-collapse ms-3">
+      <ul class="navbar-nav me-auto"><li class="nav-item"><a class="nav-link" href="../articles/getting_started.html">Getting Started</a></li>
+<li class="nav-item dropdown">
+  <button class="nav-link dropdown-toggle" type="button" id="dropdown-guides" data-bs-toggle="dropdown" aria-expanded="false" aria-haspopup="true">Guides</button>
+  <ul class="dropdown-menu" aria-labelledby="dropdown-guides"><li><h6 class="dropdown-header" data-toc-skip>Model definition</h6></li>
+    <li><a class="dropdown-item" href="../articles/sequential_model.html">Sequential Model</a></li>
+    <li><a class="dropdown-item" href="../articles/functional_api.html">Functional API</a></li>
+    <li><h6 class="dropdown-header" data-toc-skip>Extending and customizing</h6></li>
+    <li><a class="dropdown-item" href="../articles/training_with_built_in_methods.html">Training &amp; evaluation with the built-in methods</a></li>
+    <li><a class="dropdown-item" href="../articles/custom_train_step_in_tensorflow.html">Customizing `fit()` with Tensorflow</a></li>
+    <li><a class="dropdown-item" href="../articles/writing_your_own_callbacks.html">Writing your own callbacks</a></li>
+    <li><a class="dropdown-item" href="../articles/making_new_layers_and_models_via_subclassing.html">Making new layers and models via subclassing</a></li>
+    <li><a class="dropdown-item" href="../articles/writing_a_custom_training_loop_in_tensorflow.html">Writing a training loop from scratch in TensorFlow</a></li>
+    <li><a class="dropdown-item" href="../articles/serialization_and_saving.html">Serialization and Saving</a></li>
+    <li><h6 class="dropdown-header" data-toc-skip>Other topics</h6></li>
+    <li><a class="dropdown-item" href="../articles/transfer_learning.html">Transfer learning and fine tuning</a></li>
+    <li><a class="dropdown-item" href="../articles/distributed_training_with_tensorflow.html">Distributed training with TensorFlow</a></li>
+    <li><a class="dropdown-item" href="../articles/distribution.html">Distributed training with Jax</a></li>
+  </ul></li>
+<li class="nav-item"><a class="nav-link" href="../articles/examples/index.html">Examples</a></li>
+<li class="active nav-item"><a class="nav-link" href="../reference/index.html">Reference</a></li>
+<li class="nav-item"><a class="nav-link" href="../news/index.html">News</a></li>
+      </ul><ul class="navbar-nav"><li class="nav-item"><form class="form-inline" role="search">
+ <input class="form-control" type="search" name="search-input" id="search-input" autocomplete="off" aria-label="Search site" placeholder="Search for" data-search-index="../search.json"></form></li>
+<li class="nav-item"><a class="external-link nav-link" href="https://github.com/rstudio/keras3/" aria-label="GitHub"><span class="fa fab fa-github fa-lg"></span></a></li>
+      </ul></div>
+
+
+  </div>
+</nav><div class="container template-reference-topic">
+<div class="row">
+  <main id="main" class="col-md-9"><div class="page-header">
+
+      <h1>Bartlett window function.</h1>
+      <small class="dont-index">Source: <a href="https://github.com/rstudio/keras3/blob/HEAD/R/ops-numpy.R" class="external-link"><code>R/ops-numpy.R</code></a></small>
+      <div class="d-none name"><code>op_bartlett.Rd</code></div>
+    </div>
+
+    <div class="ref-description section level2">
+    <p>The Bartlett window is a triangular window that rises then falls linearly.</p>
+    </div>
+
+    <div class="section level2">
+    <h2 id="ref-usage">Usage<a class="anchor" aria-label="anchor" href="#ref-usage"></a></h2>
+    <div class="sourceCode"><pre class="sourceCode r"><code><span><span class="fu">op_bartlett</span><span class="op">(</span><span class="va">x</span><span class="op">)</span></span></code></pre></div>
+    </div>
+
+    <div class="section level2">
+    <h2 id="arguments">Arguments<a class="anchor" aria-label="anchor" href="#arguments"></a></h2>
+
+
+<dl><dt id="arg-x">x<a class="anchor" aria-label="anchor" href="#arg-x"></a></dt>
+<dd><p>Length of the window. Must be a positive integer.</p></dd>
+
+</dl></div>
+    <div class="section level2">
+    <h2 id="value">Value<a class="anchor" aria-label="anchor" href="#value"></a></h2>
+    <p>A 1D tensor containing the window values.</p>
+    </div>
+    <div class="section level2">
+    <h2 id="examples">Examples<a class="anchor" aria-label="anchor" href="#examples"></a></h2>
+    <p></p><div class="sourceCode r"><pre><code><span><span class="fu"><a href="../reference/op_bartlett.html">op_bartlett</a></span><span class="op">(</span><span class="fl">5</span><span class="op">)</span></span></code></pre><p></p></div>
+<p></p><div class="sourceCode"><pre><code><span id="cb1-1"><a href="#cb1-1" tabindex="-1"></a><span class="do">## tf.Tensor([0.  0.5 1.  0.5 0. ], shape=(5), dtype=float32)</span></span>
+<span id="cb1-2"><a href="#cb1-2" tabindex="-1"></a></span></code></pre><p></p></div>
+    </div>
+    <div class="section level2">
+    <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
+    <div class="dont-index"><p>Other numpy ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p>
+<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_layer_normalization.html">op_layer_normalization</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparse_sigmoid.html">op_sparse_sigmoid</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
+    </div>
+
+  </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
+    </nav></aside></div>
+
+
+    <footer><div class="pkgdown-footer-left">
+  <p>Developed by Tomasz Kalinowski, JJ Allaire, François Chollet, Posit Software, PBC, Google.</p>
+</div>
+
+<div class="pkgdown-footer-right">
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
+</div>
+
+    </footer></div>
+
+
+
+
+
+  </body></html>
+
diff --git a/docs/dev/reference/op_batch_normalization.html b/docs/dev/reference/op_batch_normalization.html
index 36b6e6251..5808f9145 100644
--- a/docs/dev/reference/op_batch_normalization.html
+++ b/docs/dev/reference/op_batch_normalization.html
@@ -10,7 +10,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -137,8 +137,8 @@ <h2 id="examples">Examples<a class="anchor" aria-label="anchor" href="#examples"
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
     <div class="dont-index">
 <ul><li><p><a href="https://www.tensorflow.org/api_docs/python/tf/keras/ops/batch_normalization" class="external-link">https://www.tensorflow.org/api_docs/python/tf/keras/ops/batch_normalization</a></p></li>
-</ul><p>Other nn ops: <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br></p>
-<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
+</ul><p>Other nn ops: <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_layer_normalization.html">op_layer_normalization</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparse_sigmoid.html">op_sparse_sigmoid</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br></p>
+<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_layer_normalization.html">op_layer_normalization</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparse_sigmoid.html">op_sparse_sigmoid</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -150,7 +150,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/op_binary_crossentropy.html b/docs/dev/reference/op_binary_crossentropy.html
index 8fa29dd7c..95eecc9bd 100644
--- a/docs/dev/reference/op_binary_crossentropy.html
+++ b/docs/dev/reference/op_binary_crossentropy.html
@@ -14,7 +14,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -111,8 +111,8 @@ <h2 id="examples">Examples<a class="anchor" aria-label="anchor" href="#examples"
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
     <div class="dont-index">
 <ul><li><p><a href="https://keras.io/api/ops/nn#binarycrossentropy-function" class="external-link">https://keras.io/api/ops/nn#binarycrossentropy-function</a></p></li>
-</ul><p>Other nn ops: <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br></p>
-<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
+</ul><p>Other nn ops: <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_layer_normalization.html">op_layer_normalization</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparse_sigmoid.html">op_sparse_sigmoid</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br></p>
+<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_layer_normalization.html">op_layer_normalization</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparse_sigmoid.html">op_sparse_sigmoid</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -124,7 +124,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/op_bincount.html b/docs/dev/reference/op_bincount.html
index 1e1583b94..8f680dc63 100644
--- a/docs/dev/reference/op_bincount.html
+++ b/docs/dev/reference/op_bincount.html
@@ -8,7 +8,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -121,8 +121,8 @@ <h2 id="examples">Examples<a class="anchor" aria-label="anchor" href="#examples"
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
     <div class="dont-index">
 <ul><li><p><a href="https://keras.io/api/ops/numpy#bincount-function" class="external-link">https://keras.io/api/ops/numpy#bincount-function</a></p></li>
-</ul><p>Other numpy ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p>
-<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
+</ul><p>Other numpy ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p>
+<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_layer_normalization.html">op_layer_normalization</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparse_sigmoid.html">op_sparse_sigmoid</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -134,7 +134,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/op_bitwise_and.html b/docs/dev/reference/op_bitwise_and.html
index 2ae68e207..4021292c1 100644
--- a/docs/dev/reference/op_bitwise_and.html
+++ b/docs/dev/reference/op_bitwise_and.html
@@ -12,7 +12,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -86,8 +86,8 @@ <h2 id="value">Value<a class="anchor" aria-label="anchor" href="#value"></a></h2
     </div>
     <div class="section level2">
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
-    <div class="dont-index"><p>Other numpy ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p>
-<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
+    <div class="dont-index"><p>Other numpy ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p>
+<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_layer_normalization.html">op_layer_normalization</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparse_sigmoid.html">op_sparse_sigmoid</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -99,7 +99,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/op_bitwise_invert.html b/docs/dev/reference/op_bitwise_invert.html
index ad97b8ab2..c432fed40 100644
--- a/docs/dev/reference/op_bitwise_invert.html
+++ b/docs/dev/reference/op_bitwise_invert.html
@@ -12,7 +12,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -82,8 +82,8 @@ <h2 id="value">Value<a class="anchor" aria-label="anchor" href="#value"></a></h2
     </div>
     <div class="section level2">
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
-    <div class="dont-index"><p>Other numpy ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p>
-<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
+    <div class="dont-index"><p>Other numpy ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p>
+<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_layer_normalization.html">op_layer_normalization</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparse_sigmoid.html">op_sparse_sigmoid</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -95,7 +95,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/op_bitwise_left_shift.html b/docs/dev/reference/op_bitwise_left_shift.html
index bd9d7dd75..ab5222e44 100644
--- a/docs/dev/reference/op_bitwise_left_shift.html
+++ b/docs/dev/reference/op_bitwise_left_shift.html
@@ -12,7 +12,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -86,8 +86,8 @@ <h2 id="value">Value<a class="anchor" aria-label="anchor" href="#value"></a></h2
     </div>
     <div class="section level2">
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
-    <div class="dont-index"><p>Other numpy ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p>
-<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
+    <div class="dont-index"><p>Other numpy ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p>
+<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_layer_normalization.html">op_layer_normalization</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparse_sigmoid.html">op_sparse_sigmoid</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -99,7 +99,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/op_bitwise_not.html b/docs/dev/reference/op_bitwise_not.html
index 65684f651..ea438b321 100644
--- a/docs/dev/reference/op_bitwise_not.html
+++ b/docs/dev/reference/op_bitwise_not.html
@@ -12,7 +12,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -82,8 +82,8 @@ <h2 id="value">Value<a class="anchor" aria-label="anchor" href="#value"></a></h2
     </div>
     <div class="section level2">
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
-    <div class="dont-index"><p>Other numpy ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p>
-<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
+    <div class="dont-index"><p>Other numpy ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p>
+<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_layer_normalization.html">op_layer_normalization</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparse_sigmoid.html">op_sparse_sigmoid</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -95,7 +95,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/op_bitwise_or.html b/docs/dev/reference/op_bitwise_or.html
index f73b879a9..2082c2a94 100644
--- a/docs/dev/reference/op_bitwise_or.html
+++ b/docs/dev/reference/op_bitwise_or.html
@@ -12,7 +12,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -86,8 +86,8 @@ <h2 id="value">Value<a class="anchor" aria-label="anchor" href="#value"></a></h2
     </div>
     <div class="section level2">
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
-    <div class="dont-index"><p>Other numpy ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p>
-<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
+    <div class="dont-index"><p>Other numpy ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p>
+<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_layer_normalization.html">op_layer_normalization</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparse_sigmoid.html">op_sparse_sigmoid</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -99,7 +99,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/op_bitwise_right_shift.html b/docs/dev/reference/op_bitwise_right_shift.html
index 6e8b4c9b2..49ee25e77 100644
--- a/docs/dev/reference/op_bitwise_right_shift.html
+++ b/docs/dev/reference/op_bitwise_right_shift.html
@@ -12,7 +12,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -86,8 +86,8 @@ <h2 id="value">Value<a class="anchor" aria-label="anchor" href="#value"></a></h2
     </div>
     <div class="section level2">
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
-    <div class="dont-index"><p>Other numpy ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p>
-<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
+    <div class="dont-index"><p>Other numpy ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p>
+<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_layer_normalization.html">op_layer_normalization</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparse_sigmoid.html">op_sparse_sigmoid</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -99,7 +99,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/op_bitwise_xor.html b/docs/dev/reference/op_bitwise_xor.html
index 7495f62dc..f0e60d0b0 100644
--- a/docs/dev/reference/op_bitwise_xor.html
+++ b/docs/dev/reference/op_bitwise_xor.html
@@ -12,7 +12,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -86,8 +86,8 @@ <h2 id="value">Value<a class="anchor" aria-label="anchor" href="#value"></a></h2
     </div>
     <div class="section level2">
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
-    <div class="dont-index"><p>Other numpy ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p>
-<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
+    <div class="dont-index"><p>Other numpy ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p>
+<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_layer_normalization.html">op_layer_normalization</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparse_sigmoid.html">op_sparse_sigmoid</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -99,7 +99,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/op_blackman.html b/docs/dev/reference/op_blackman.html
new file mode 100644
index 000000000..cac2fc0df
--- /dev/null
+++ b/docs/dev/reference/op_blackman.html
@@ -0,0 +1,116 @@
+<!DOCTYPE html>
+<!-- Generated by pkgdown: do not edit by hand --><html lang="en"><head><meta http-equiv="Content-Type" content="text/html; charset=UTF-8"><meta charset="utf-8"><meta http-equiv="X-UA-Compatible" content="IE=edge"><meta name="viewport" content="width=device-width, initial-scale=1, shrink-to-fit=no"><title>Blackman window function. — op_blackman • keras3</title><!-- favicons --><link rel="icon" type="image/png" sizes="96x96" href="../favicon-96x96.png"><link rel="icon" type="”image/svg+xml”" href="../favicon.svg"><link rel="apple-touch-icon" sizes="180x180" href="../apple-touch-icon.png"><link rel="icon" sizes="any" href="../favicon.ico"><link rel="manifest" href="../site.webmanifest"><script src="../deps/jquery-3.6.0/jquery-3.6.0.min.js"></script><meta name="viewport" content="width=device-width, initial-scale=1, shrink-to-fit=no"><link href="../deps/bootstrap-5.3.1/bootstrap.min.css" rel="stylesheet"><script src="../deps/bootstrap-5.3.1/bootstrap.bundle.min.js"></script><link href="../deps/font-awesome-6.5.2/css/all.min.css" rel="stylesheet"><link href="../deps/font-awesome-6.5.2/css/v4-shims.min.css" rel="stylesheet"><script src="../deps/headroom-0.11.0/headroom.min.js"></script><script src="../deps/headroom-0.11.0/jQuery.headroom.min.js"></script><script src="../deps/bootstrap-toc-1.0.1/bootstrap-toc.min.js"></script><script src="../deps/clipboard.js-2.0.11/clipboard.min.js"></script><script src="../deps/search-1.0.0/autocomplete.jquery.min.js"></script><script src="../deps/search-1.0.0/fuse.min.js"></script><script src="../deps/search-1.0.0/mark.min.js"></script><!-- pkgdown --><script src="../pkgdown.js"></script><link href="../extra.css" rel="stylesheet"><meta property="og:title" content="Blackman window function. — op_blackman"><meta name="description" content="The Blackman window is a taper formed by using a weighted cosine."><meta property="og:description" content="The Blackman window is a taper formed by using a weighted cosine."><meta name="robots" content="noindex"><!-- Google Tag Manager --><script>(function(w,d,s,l,i){w[l]=w[l]||[];w[l].push({'gtm.start': new Date().getTime(),event:'gtm.js'});var f=d.getElementsByTagName(s)[0],j=d.createElement(s),dl=l!='dataLayer'?'&l='+l:'';j.async=true;j.src='https://www.googletagmanager.com/gtm.js?id='+i+dl;f.parentNode.insertBefore(j,f);})(window,document,'script','dataLayer','GTM-KHBDBW7');</script><!-- End Google Tag Manager --></head><body>
+    <a href="#main" class="visually-hidden-focusable">Skip to contents</a>
+    <!-- Google Tag Manager (noscript) --> <noscript><iframe src="https://www.googletagmanager.com/ns.html?id=GTM-KHBDBW7" height="0" width="0" style="display:none;visibility:hidden"></iframe></noscript> <!-- End Google Tag Manager (noscript) -->
+
+
+    <nav class="navbar navbar-expand-lg fixed-top bg-primary" data-bs-theme="inverse" aria-label="Site navigation"><div class="container">
+
+    <a class="navbar-brand me-2" href="../index.html">keras3</a>
+
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
+
+
+    <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
+      <span class="navbar-toggler-icon"></span>
+    </button>
+
+    <div id="navbar" class="collapse navbar-collapse ms-3">
+      <ul class="navbar-nav me-auto"><li class="nav-item"><a class="nav-link" href="../articles/getting_started.html">Getting Started</a></li>
+<li class="nav-item dropdown">
+  <button class="nav-link dropdown-toggle" type="button" id="dropdown-guides" data-bs-toggle="dropdown" aria-expanded="false" aria-haspopup="true">Guides</button>
+  <ul class="dropdown-menu" aria-labelledby="dropdown-guides"><li><h6 class="dropdown-header" data-toc-skip>Model definition</h6></li>
+    <li><a class="dropdown-item" href="../articles/sequential_model.html">Sequential Model</a></li>
+    <li><a class="dropdown-item" href="../articles/functional_api.html">Functional API</a></li>
+    <li><h6 class="dropdown-header" data-toc-skip>Extending and customizing</h6></li>
+    <li><a class="dropdown-item" href="../articles/training_with_built_in_methods.html">Training &amp; evaluation with the built-in methods</a></li>
+    <li><a class="dropdown-item" href="../articles/custom_train_step_in_tensorflow.html">Customizing `fit()` with Tensorflow</a></li>
+    <li><a class="dropdown-item" href="../articles/writing_your_own_callbacks.html">Writing your own callbacks</a></li>
+    <li><a class="dropdown-item" href="../articles/making_new_layers_and_models_via_subclassing.html">Making new layers and models via subclassing</a></li>
+    <li><a class="dropdown-item" href="../articles/writing_a_custom_training_loop_in_tensorflow.html">Writing a training loop from scratch in TensorFlow</a></li>
+    <li><a class="dropdown-item" href="../articles/serialization_and_saving.html">Serialization and Saving</a></li>
+    <li><h6 class="dropdown-header" data-toc-skip>Other topics</h6></li>
+    <li><a class="dropdown-item" href="../articles/transfer_learning.html">Transfer learning and fine tuning</a></li>
+    <li><a class="dropdown-item" href="../articles/distributed_training_with_tensorflow.html">Distributed training with TensorFlow</a></li>
+    <li><a class="dropdown-item" href="../articles/distribution.html">Distributed training with Jax</a></li>
+  </ul></li>
+<li class="nav-item"><a class="nav-link" href="../articles/examples/index.html">Examples</a></li>
+<li class="active nav-item"><a class="nav-link" href="../reference/index.html">Reference</a></li>
+<li class="nav-item"><a class="nav-link" href="../news/index.html">News</a></li>
+      </ul><ul class="navbar-nav"><li class="nav-item"><form class="form-inline" role="search">
+ <input class="form-control" type="search" name="search-input" id="search-input" autocomplete="off" aria-label="Search site" placeholder="Search for" data-search-index="../search.json"></form></li>
+<li class="nav-item"><a class="external-link nav-link" href="https://github.com/rstudio/keras3/" aria-label="GitHub"><span class="fa fab fa-github fa-lg"></span></a></li>
+      </ul></div>
+
+
+  </div>
+</nav><div class="container template-reference-topic">
+<div class="row">
+  <main id="main" class="col-md-9"><div class="page-header">
+
+      <h1>Blackman window function.</h1>
+      <small class="dont-index">Source: <a href="https://github.com/rstudio/keras3/blob/HEAD/R/ops-numpy.R" class="external-link"><code>R/ops-numpy.R</code></a></small>
+      <div class="d-none name"><code>op_blackman.Rd</code></div>
+    </div>
+
+    <div class="ref-description section level2">
+    <p>The Blackman window is a taper formed by using a weighted cosine.</p>
+    </div>
+
+    <div class="section level2">
+    <h2 id="ref-usage">Usage<a class="anchor" aria-label="anchor" href="#ref-usage"></a></h2>
+    <div class="sourceCode"><pre class="sourceCode r"><code><span><span class="fu">op_blackman</span><span class="op">(</span><span class="va">x</span><span class="op">)</span></span></code></pre></div>
+    </div>
+
+    <div class="section level2">
+    <h2 id="arguments">Arguments<a class="anchor" aria-label="anchor" href="#arguments"></a></h2>
+
+
+<dl><dt id="arg-x">x<a class="anchor" aria-label="anchor" href="#arg-x"></a></dt>
+<dd><p>Length of the window. Must be a positive integer.</p></dd>
+
+</dl></div>
+    <div class="section level2">
+    <h2 id="value">Value<a class="anchor" aria-label="anchor" href="#value"></a></h2>
+    <p>A 1D tensor containing the window values.</p>
+    </div>
+    <div class="section level2">
+    <h2 id="examples">Examples<a class="anchor" aria-label="anchor" href="#examples"></a></h2>
+    <p></p><div class="sourceCode r"><pre><code><span><span class="fu"><a href="../reference/op_blackman.html">op_blackman</a></span><span class="op">(</span><span class="fl">5</span><span class="op">)</span></span></code></pre><p></p></div>
+<p></p><div class="sourceCode"><pre><code><span id="cb1-1"><a href="#cb1-1" tabindex="-1"></a><span class="do">## tf.Tensor(</span></span>
+<span id="cb1-2"><a href="#cb1-2" tabindex="-1"></a><span class="do">## [-1.4901161e-08  3.4000003e-01  9.9999994e-01  3.3999997e-01</span></span>
+<span id="cb1-3"><a href="#cb1-3" tabindex="-1"></a><span class="do">##  -1.4901161e-08], shape=(5), dtype=float32)</span></span>
+<span id="cb1-4"><a href="#cb1-4" tabindex="-1"></a></span></code></pre><p></p></div>
+<p>[-1.4901161e-08  3.4000003e-01  9.9999994e-01  3.3999997e-01</p><div class="section">
+<h3 id="-e-r-e-e-e-e-a-e-">-1.4901161e-08]: R:-1.4901161e-08%20%203.4000003e-01%20%209.9999994e-01%20%203.3999997e-01%0A##%20%20-1.4901161e-08<a class="anchor" aria-label="anchor" href="#-e-r-e-e-e-e-a-e-"></a></h3>
+
+
+</div>
+
+    </div>
+    <div class="section level2">
+    <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
+    <div class="dont-index"><p>Other numpy ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p>
+<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_layer_normalization.html">op_layer_normalization</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparse_sigmoid.html">op_sparse_sigmoid</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
+    </div>
+
+  </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
+    </nav></aside></div>
+
+
+    <footer><div class="pkgdown-footer-left">
+  <p>Developed by Tomasz Kalinowski, JJ Allaire, François Chollet, Posit Software, PBC, Google.</p>
+</div>
+
+<div class="pkgdown-footer-right">
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
+</div>
+
+    </footer></div>
+
+
+
+
+
+  </body></html>
+
diff --git a/docs/dev/reference/op_broadcast_to.html b/docs/dev/reference/op_broadcast_to.html
index b54ce522b..513ff6087 100644
--- a/docs/dev/reference/op_broadcast_to.html
+++ b/docs/dev/reference/op_broadcast_to.html
@@ -8,7 +8,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -92,8 +92,8 @@ <h2 id="examples">Examples<a class="anchor" aria-label="anchor" href="#examples"
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
     <div class="dont-index">
 <ul><li><p><a href="https://keras.io/api/ops/numpy#broadcastto-function" class="external-link">https://keras.io/api/ops/numpy#broadcastto-function</a></p></li>
-</ul><p>Other numpy ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p>
-<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
+</ul><p>Other numpy ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p>
+<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_layer_normalization.html">op_layer_normalization</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparse_sigmoid.html">op_sparse_sigmoid</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -105,7 +105,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/op_cast.html b/docs/dev/reference/op_cast.html
index 716ff591f..a2fd7e6a3 100644
--- a/docs/dev/reference/op_cast.html
+++ b/docs/dev/reference/op_cast.html
@@ -8,7 +8,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -92,7 +92,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
     <div class="dont-index">
 <ul><li><p><a href="https://keras.io/api/ops/core#cast-function" class="external-link">https://keras.io/api/ops/core#cast-function</a></p></li>
 </ul><p>Other core ops: <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br></p>
-<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
+<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_layer_normalization.html">op_layer_normalization</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparse_sigmoid.html">op_sparse_sigmoid</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -104,7 +104,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/op_categorical_crossentropy.html b/docs/dev/reference/op_categorical_crossentropy.html
index 117386399..0489869b3 100644
--- a/docs/dev/reference/op_categorical_crossentropy.html
+++ b/docs/dev/reference/op_categorical_crossentropy.html
@@ -14,7 +14,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -123,8 +123,8 @@ <h2 id="examples">Examples<a class="anchor" aria-label="anchor" href="#examples"
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
     <div class="dont-index">
 <ul><li><p><a href="https://keras.io/api/ops/nn#categoricalcrossentropy-function" class="external-link">https://keras.io/api/ops/nn#categoricalcrossentropy-function</a></p></li>
-</ul><p>Other nn ops: <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br></p>
-<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
+</ul><p>Other nn ops: <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_layer_normalization.html">op_layer_normalization</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparse_sigmoid.html">op_sparse_sigmoid</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br></p>
+<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_layer_normalization.html">op_layer_normalization</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparse_sigmoid.html">op_sparse_sigmoid</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -136,7 +136,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/op_cbrt.html b/docs/dev/reference/op_cbrt.html
new file mode 100644
index 000000000..b6051b021
--- /dev/null
+++ b/docs/dev/reference/op_cbrt.html
@@ -0,0 +1,108 @@
+<!DOCTYPE html>
+<!-- Generated by pkgdown: do not edit by hand --><html lang="en"><head><meta http-equiv="Content-Type" content="text/html; charset=UTF-8"><meta charset="utf-8"><meta http-equiv="X-UA-Compatible" content="IE=edge"><meta name="viewport" content="width=device-width, initial-scale=1, shrink-to-fit=no"><title>Computes the cube root of the input tensor, element-wise. — op_cbrt • keras3</title><!-- favicons --><link rel="icon" type="image/png" sizes="96x96" href="../favicon-96x96.png"><link rel="icon" type="”image/svg+xml”" href="../favicon.svg"><link rel="apple-touch-icon" sizes="180x180" href="../apple-touch-icon.png"><link rel="icon" sizes="any" href="../favicon.ico"><link rel="manifest" href="../site.webmanifest"><script src="../deps/jquery-3.6.0/jquery-3.6.0.min.js"></script><meta name="viewport" content="width=device-width, initial-scale=1, shrink-to-fit=no"><link href="../deps/bootstrap-5.3.1/bootstrap.min.css" rel="stylesheet"><script src="../deps/bootstrap-5.3.1/bootstrap.bundle.min.js"></script><link href="../deps/font-awesome-6.5.2/css/all.min.css" rel="stylesheet"><link href="../deps/font-awesome-6.5.2/css/v4-shims.min.css" rel="stylesheet"><script src="../deps/headroom-0.11.0/headroom.min.js"></script><script src="../deps/headroom-0.11.0/jQuery.headroom.min.js"></script><script src="../deps/bootstrap-toc-1.0.1/bootstrap-toc.min.js"></script><script src="../deps/clipboard.js-2.0.11/clipboard.min.js"></script><script src="../deps/search-1.0.0/autocomplete.jquery.min.js"></script><script src="../deps/search-1.0.0/fuse.min.js"></script><script src="../deps/search-1.0.0/mark.min.js"></script><!-- pkgdown --><script src="../pkgdown.js"></script><link href="../extra.css" rel="stylesheet"><meta property="og:title" content="Computes the cube root of the input tensor, element-wise. — op_cbrt"><meta name="description" content="Returns the real-valued cube root of x, handling negative inputs in the
+real domain."><meta property="og:description" content="Returns the real-valued cube root of x, handling negative inputs in the
+real domain."><meta name="robots" content="noindex"><!-- Google Tag Manager --><script>(function(w,d,s,l,i){w[l]=w[l]||[];w[l].push({'gtm.start': new Date().getTime(),event:'gtm.js'});var f=d.getElementsByTagName(s)[0],j=d.createElement(s),dl=l!='dataLayer'?'&l='+l:'';j.async=true;j.src='https://www.googletagmanager.com/gtm.js?id='+i+dl;f.parentNode.insertBefore(j,f);})(window,document,'script','dataLayer','GTM-KHBDBW7');</script><!-- End Google Tag Manager --></head><body>
+    <a href="#main" class="visually-hidden-focusable">Skip to contents</a>
+    <!-- Google Tag Manager (noscript) --> <noscript><iframe src="https://www.googletagmanager.com/ns.html?id=GTM-KHBDBW7" height="0" width="0" style="display:none;visibility:hidden"></iframe></noscript> <!-- End Google Tag Manager (noscript) -->
+
+
+    <nav class="navbar navbar-expand-lg fixed-top bg-primary" data-bs-theme="inverse" aria-label="Site navigation"><div class="container">
+
+    <a class="navbar-brand me-2" href="../index.html">keras3</a>
+
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
+
+
+    <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
+      <span class="navbar-toggler-icon"></span>
+    </button>
+
+    <div id="navbar" class="collapse navbar-collapse ms-3">
+      <ul class="navbar-nav me-auto"><li class="nav-item"><a class="nav-link" href="../articles/getting_started.html">Getting Started</a></li>
+<li class="nav-item dropdown">
+  <button class="nav-link dropdown-toggle" type="button" id="dropdown-guides" data-bs-toggle="dropdown" aria-expanded="false" aria-haspopup="true">Guides</button>
+  <ul class="dropdown-menu" aria-labelledby="dropdown-guides"><li><h6 class="dropdown-header" data-toc-skip>Model definition</h6></li>
+    <li><a class="dropdown-item" href="../articles/sequential_model.html">Sequential Model</a></li>
+    <li><a class="dropdown-item" href="../articles/functional_api.html">Functional API</a></li>
+    <li><h6 class="dropdown-header" data-toc-skip>Extending and customizing</h6></li>
+    <li><a class="dropdown-item" href="../articles/training_with_built_in_methods.html">Training &amp; evaluation with the built-in methods</a></li>
+    <li><a class="dropdown-item" href="../articles/custom_train_step_in_tensorflow.html">Customizing `fit()` with Tensorflow</a></li>
+    <li><a class="dropdown-item" href="../articles/writing_your_own_callbacks.html">Writing your own callbacks</a></li>
+    <li><a class="dropdown-item" href="../articles/making_new_layers_and_models_via_subclassing.html">Making new layers and models via subclassing</a></li>
+    <li><a class="dropdown-item" href="../articles/writing_a_custom_training_loop_in_tensorflow.html">Writing a training loop from scratch in TensorFlow</a></li>
+    <li><a class="dropdown-item" href="../articles/serialization_and_saving.html">Serialization and Saving</a></li>
+    <li><h6 class="dropdown-header" data-toc-skip>Other topics</h6></li>
+    <li><a class="dropdown-item" href="../articles/transfer_learning.html">Transfer learning and fine tuning</a></li>
+    <li><a class="dropdown-item" href="../articles/distributed_training_with_tensorflow.html">Distributed training with TensorFlow</a></li>
+    <li><a class="dropdown-item" href="../articles/distribution.html">Distributed training with Jax</a></li>
+  </ul></li>
+<li class="nav-item"><a class="nav-link" href="../articles/examples/index.html">Examples</a></li>
+<li class="active nav-item"><a class="nav-link" href="../reference/index.html">Reference</a></li>
+<li class="nav-item"><a class="nav-link" href="../news/index.html">News</a></li>
+      </ul><ul class="navbar-nav"><li class="nav-item"><form class="form-inline" role="search">
+ <input class="form-control" type="search" name="search-input" id="search-input" autocomplete="off" aria-label="Search site" placeholder="Search for" data-search-index="../search.json"></form></li>
+<li class="nav-item"><a class="external-link nav-link" href="https://github.com/rstudio/keras3/" aria-label="GitHub"><span class="fa fab fa-github fa-lg"></span></a></li>
+      </ul></div>
+
+
+  </div>
+</nav><div class="container template-reference-topic">
+<div class="row">
+  <main id="main" class="col-md-9"><div class="page-header">
+
+      <h1>Computes the cube root of the input tensor, element-wise.</h1>
+      <small class="dont-index">Source: <a href="https://github.com/rstudio/keras3/blob/HEAD/R/ops-numpy.R" class="external-link"><code>R/ops-numpy.R</code></a></small>
+      <div class="d-none name"><code>op_cbrt.Rd</code></div>
+    </div>
+
+    <div class="ref-description section level2">
+    <p>Returns the real-valued cube root of <code>x</code>, handling negative inputs in the
+real domain.</p>
+    </div>
+
+    <div class="section level2">
+    <h2 id="ref-usage">Usage<a class="anchor" aria-label="anchor" href="#ref-usage"></a></h2>
+    <div class="sourceCode"><pre class="sourceCode r"><code><span><span class="fu">op_cbrt</span><span class="op">(</span><span class="va">x</span><span class="op">)</span></span></code></pre></div>
+    </div>
+
+    <div class="section level2">
+    <h2 id="arguments">Arguments<a class="anchor" aria-label="anchor" href="#arguments"></a></h2>
+
+
+<dl><dt id="arg-x">x<a class="anchor" aria-label="anchor" href="#arg-x"></a></dt>
+<dd><p>Input tensor.Examples</p>
+<p></p><div class="sourceCode r"><pre><code><span><span class="fu"><a href="../reference/op_cbrt.html">op_cbrt</a></span><span class="op">(</span><span class="fu"><a href="https://rdrr.io/r/base/c.html" class="external-link">c</a></span><span class="op">(</span><span class="op">-</span><span class="fl">8</span>, <span class="fl">0</span>, <span class="fl">8</span><span class="op">)</span><span class="op">)</span></span></code></pre><p></p></div>
+<p></p><div class="sourceCode"><pre><code><span id="cb1-1"><a href="#cb1-1" tabindex="-1"></a><span class="do">## tf.Tensor([-2.  0.  2.], shape=(3), dtype=float32)</span></span>
+<span id="cb1-2"><a href="#cb1-2" tabindex="-1"></a></span></code></pre><p></p></div></dd>
+
+</dl></div>
+    <div class="section level2">
+    <h2 id="value">Value<a class="anchor" aria-label="anchor" href="#value"></a></h2>
+    <p>A tensor containing the cube root of each element in <code>x</code>.</p>
+    </div>
+    <div class="section level2">
+    <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
+    <div class="dont-index"><p>Other numpy ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p>
+<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_layer_normalization.html">op_layer_normalization</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparse_sigmoid.html">op_sparse_sigmoid</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
+    </div>
+
+  </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
+    </nav></aside></div>
+
+
+    <footer><div class="pkgdown-footer-left">
+  <p>Developed by Tomasz Kalinowski, JJ Allaire, François Chollet, Posit Software, PBC, Google.</p>
+</div>
+
+<div class="pkgdown-footer-right">
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
+</div>
+
+    </footer></div>
+
+
+
+
+
+  </body></html>
+
diff --git a/docs/dev/reference/op_ceil.html b/docs/dev/reference/op_ceil.html
index b3f70333d..89c0ecf46 100644
--- a/docs/dev/reference/op_ceil.html
+++ b/docs/dev/reference/op_ceil.html
@@ -10,7 +10,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -81,8 +81,8 @@ <h2 id="value">Value<a class="anchor" aria-label="anchor" href="#value"></a></h2
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
     <div class="dont-index">
 <ul><li><p><a href="https://keras.io/api/ops/numpy#ceil-function" class="external-link">https://keras.io/api/ops/numpy#ceil-function</a></p></li>
-</ul><p>Other numpy ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p>
-<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
+</ul><p>Other numpy ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p>
+<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_layer_normalization.html">op_layer_normalization</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparse_sigmoid.html">op_sparse_sigmoid</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -94,7 +94,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/op_celu.html b/docs/dev/reference/op_celu.html
index c911e00a2..8a32fa3a3 100644
--- a/docs/dev/reference/op_celu.html
+++ b/docs/dev/reference/op_celu.html
@@ -10,7 +10,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -90,8 +90,8 @@ <h2 id="examples">Examples<a class="anchor" aria-label="anchor" href="#examples"
     </div>
     <div class="section level2">
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
-    <div class="dont-index"><p>Other nn ops: <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br></p>
-<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
+    <div class="dont-index"><p>Other nn ops: <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_layer_normalization.html">op_layer_normalization</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparse_sigmoid.html">op_sparse_sigmoid</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br></p>
+<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_layer_normalization.html">op_layer_normalization</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparse_sigmoid.html">op_sparse_sigmoid</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -103,7 +103,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/op_cholesky.html b/docs/dev/reference/op_cholesky.html
index ca5fd7fad..763793dca 100644
--- a/docs/dev/reference/op_cholesky.html
+++ b/docs/dev/reference/op_cholesky.html
@@ -8,7 +8,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -78,7 +78,7 @@ <h2 id="value">Value<a class="anchor" aria-label="anchor" href="#value"></a></h2
     <div class="section level2">
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
     <div class="dont-index"><p>Other linear algebra ops: <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br></p>
-<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
+<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_layer_normalization.html">op_layer_normalization</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparse_sigmoid.html">op_sparse_sigmoid</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -90,7 +90,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/op_clip.html b/docs/dev/reference/op_clip.html
index 18008e5a8..ac3e96174 100644
--- a/docs/dev/reference/op_clip.html
+++ b/docs/dev/reference/op_clip.html
@@ -12,7 +12,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -92,8 +92,8 @@ <h2 id="value">Value<a class="anchor" aria-label="anchor" href="#value"></a></h2
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
     <div class="dont-index">
 <ul><li><p><a href="https://keras.io/api/ops/numpy#clip-function" class="external-link">https://keras.io/api/ops/numpy#clip-function</a></p></li>
-</ul><p>Other numpy ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p>
-<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
+</ul><p>Other numpy ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p>
+<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_layer_normalization.html">op_layer_normalization</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparse_sigmoid.html">op_sparse_sigmoid</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -105,7 +105,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/op_concatenate.html b/docs/dev/reference/op_concatenate.html
index 20a21cb1f..864320dd7 100644
--- a/docs/dev/reference/op_concatenate.html
+++ b/docs/dev/reference/op_concatenate.html
@@ -8,7 +8,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -82,8 +82,8 @@ <h2 id="value">Value<a class="anchor" aria-label="anchor" href="#value"></a></h2
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
     <div class="dont-index">
 <ul><li><p><a href="https://keras.io/api/ops/numpy#concatenate-function" class="external-link">https://keras.io/api/ops/numpy#concatenate-function</a></p></li>
-</ul><p>Other numpy ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p>
-<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
+</ul><p>Other numpy ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p>
+<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_layer_normalization.html">op_layer_normalization</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparse_sigmoid.html">op_sparse_sigmoid</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -95,7 +95,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/op_cond.html b/docs/dev/reference/op_cond.html
index 34273bbdc..6ecc984ee 100644
--- a/docs/dev/reference/op_cond.html
+++ b/docs/dev/reference/op_cond.html
@@ -8,7 +8,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -128,7 +128,7 @@ <h2 id="examples">Examples<a class="anchor" aria-label="anchor" href="#examples"
     <div class="section level2">
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
     <div class="dont-index"><p>Other core ops: <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br></p>
-<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
+<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_layer_normalization.html">op_layer_normalization</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparse_sigmoid.html">op_sparse_sigmoid</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -140,7 +140,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/op_conj.html b/docs/dev/reference/op_conj.html
index ac5b6506f..17affab7a 100644
--- a/docs/dev/reference/op_conj.html
+++ b/docs/dev/reference/op_conj.html
@@ -10,7 +10,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -81,8 +81,8 @@ <h2 id="value">Value<a class="anchor" aria-label="anchor" href="#value"></a></h2
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
     <div class="dont-index">
 <ul><li><p><a href="https://keras.io/api/ops/numpy#conjugate-function" class="external-link">https://keras.io/api/ops/numpy#conjugate-function</a></p></li>
-</ul><p>Other numpy ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p>
-<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
+</ul><p>Other numpy ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p>
+<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_layer_normalization.html">op_layer_normalization</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparse_sigmoid.html">op_sparse_sigmoid</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -94,7 +94,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/op_conv.html b/docs/dev/reference/op_conv.html
index 9dc2f6f2e..56b188da8 100644
--- a/docs/dev/reference/op_conv.html
+++ b/docs/dev/reference/op_conv.html
@@ -8,7 +8,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -92,7 +92,7 @@ <h2 id="arguments">Arguments<a class="anchor" aria-label="anchor" href="#argumen
 <dd><p>int or int tuple/list of <code>len(inputs_spatial_shape)</code>,
 specifying the strides of the convolution along each spatial
 dimension. If <code>strides</code> is int, then every spatial dimension shares
-the same <code>strides</code>.</p></dd>
+the same <code>strides</code>. Defaults to <code>1</code>.</p></dd>
 
 
 <dt id="arg-padding">padding<a class="anchor" aria-label="anchor" href="#arg-padding"></a></dt>
@@ -126,8 +126,8 @@ <h2 id="value">Value<a class="anchor" aria-label="anchor" href="#value"></a></h2
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
     <div class="dont-index">
 <ul><li><p><a href="https://keras.io/api/ops/nn#conv-function" class="external-link">https://keras.io/api/ops/nn#conv-function</a></p></li>
-</ul><p>Other nn ops: <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br></p>
-<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
+</ul><p>Other nn ops: <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_layer_normalization.html">op_layer_normalization</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparse_sigmoid.html">op_sparse_sigmoid</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br></p>
+<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_layer_normalization.html">op_layer_normalization</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparse_sigmoid.html">op_sparse_sigmoid</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -139,7 +139,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/op_conv_transpose.html b/docs/dev/reference/op_conv_transpose.html
index dade36d47..0a2c7c3de 100644
--- a/docs/dev/reference/op_conv_transpose.html
+++ b/docs/dev/reference/op_conv_transpose.html
@@ -8,7 +8,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -62,7 +62,7 @@ <h2 id="ref-usage">Usage<a class="anchor" aria-label="anchor" href="#ref-usage">
     <div class="sourceCode"><pre class="sourceCode r"><code><span><span class="fu">op_conv_transpose</span><span class="op">(</span></span>
 <span>  <span class="va">inputs</span>,</span>
 <span>  <span class="va">kernel</span>,</span>
-<span>  <span class="va">strides</span>,</span>
+<span>  strides <span class="op">=</span> <span class="fl">1L</span>,</span>
 <span>  padding <span class="op">=</span> <span class="st">"valid"</span>,</span>
 <span>  output_padding <span class="op">=</span> <span class="cn">NULL</span>,</span>
 <span>  data_format <span class="op">=</span> <span class="cn">NULL</span>,</span>
@@ -137,8 +137,8 @@ <h2 id="value">Value<a class="anchor" aria-label="anchor" href="#value"></a></h2
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
     <div class="dont-index">
 <ul><li><p><a href="https://keras.io/api/ops/nn#convtranspose-function" class="external-link">https://keras.io/api/ops/nn#convtranspose-function</a></p></li>
-</ul><p>Other nn ops: <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br></p>
-<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
+</ul><p>Other nn ops: <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_layer_normalization.html">op_layer_normalization</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparse_sigmoid.html">op_sparse_sigmoid</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br></p>
+<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_layer_normalization.html">op_layer_normalization</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparse_sigmoid.html">op_sparse_sigmoid</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -150,7 +150,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/op_convert_to_numpy.html b/docs/dev/reference/op_convert_to_numpy.html
index f46a6ecac..1f4f9467f 100644
--- a/docs/dev/reference/op_convert_to_numpy.html
+++ b/docs/dev/reference/op_convert_to_numpy.html
@@ -8,7 +8,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -81,7 +81,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
     <div class="dont-index">
 <ul><li><p><a href="https://keras.io/api/ops/core#converttonumpy-function" class="external-link">https://keras.io/api/ops/core#converttonumpy-function</a></p></li>
 </ul><p>Other core ops: <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br></p>
-<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
+<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_layer_normalization.html">op_layer_normalization</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparse_sigmoid.html">op_sparse_sigmoid</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -93,7 +93,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/op_convert_to_tensor.html b/docs/dev/reference/op_convert_to_tensor.html
index a3f03c8f8..75499bc32 100644
--- a/docs/dev/reference/op_convert_to_tensor.html
+++ b/docs/dev/reference/op_convert_to_tensor.html
@@ -10,7 +10,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -110,7 +110,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 <ul><li><p><code><a href="op_array.html">op_array()</a></code></p></li>
 <li><p><a href="https://keras.io/api/ops/core#converttotensor-function" class="external-link">https://keras.io/api/ops/core#converttotensor-function</a></p></li>
 </ul><p>Other core ops: <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br></p>
-<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
+<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_layer_normalization.html">op_layer_normalization</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparse_sigmoid.html">op_sparse_sigmoid</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -122,7 +122,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/op_copy.html b/docs/dev/reference/op_copy.html
index ed66c80fb..62e578d2a 100644
--- a/docs/dev/reference/op_copy.html
+++ b/docs/dev/reference/op_copy.html
@@ -8,7 +8,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -78,8 +78,8 @@ <h2 id="value">Value<a class="anchor" aria-label="anchor" href="#value"></a></h2
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
     <div class="dont-index">
 <ul><li><p><a href="https://keras.io/api/ops/numpy#copy-function" class="external-link">https://keras.io/api/ops/numpy#copy-function</a></p></li>
-</ul><p>Other numpy ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p>
-<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
+</ul><p>Other numpy ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p>
+<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_layer_normalization.html">op_layer_normalization</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparse_sigmoid.html">op_sparse_sigmoid</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -91,7 +91,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/op_corrcoef.html b/docs/dev/reference/op_corrcoef.html
new file mode 100644
index 000000000..63aab19bb
--- /dev/null
+++ b/docs/dev/reference/op_corrcoef.html
@@ -0,0 +1,110 @@
+<!DOCTYPE html>
+<!-- Generated by pkgdown: do not edit by hand --><html lang="en"><head><meta http-equiv="Content-Type" content="text/html; charset=UTF-8"><meta charset="utf-8"><meta http-equiv="X-UA-Compatible" content="IE=edge"><meta name="viewport" content="width=device-width, initial-scale=1, shrink-to-fit=no"><title>Compute the Pearson correlation coefficient matrix. — op_corrcoef • keras3</title><!-- favicons --><link rel="icon" type="image/png" sizes="96x96" href="../favicon-96x96.png"><link rel="icon" type="”image/svg+xml”" href="../favicon.svg"><link rel="apple-touch-icon" sizes="180x180" href="../apple-touch-icon.png"><link rel="icon" sizes="any" href="../favicon.ico"><link rel="manifest" href="../site.webmanifest"><script src="../deps/jquery-3.6.0/jquery-3.6.0.min.js"></script><meta name="viewport" content="width=device-width, initial-scale=1, shrink-to-fit=no"><link href="../deps/bootstrap-5.3.1/bootstrap.min.css" rel="stylesheet"><script src="../deps/bootstrap-5.3.1/bootstrap.bundle.min.js"></script><link href="../deps/font-awesome-6.5.2/css/all.min.css" rel="stylesheet"><link href="../deps/font-awesome-6.5.2/css/v4-shims.min.css" rel="stylesheet"><script src="../deps/headroom-0.11.0/headroom.min.js"></script><script src="../deps/headroom-0.11.0/jQuery.headroom.min.js"></script><script src="../deps/bootstrap-toc-1.0.1/bootstrap-toc.min.js"></script><script src="../deps/clipboard.js-2.0.11/clipboard.min.js"></script><script src="../deps/search-1.0.0/autocomplete.jquery.min.js"></script><script src="../deps/search-1.0.0/fuse.min.js"></script><script src="../deps/search-1.0.0/mark.min.js"></script><!-- pkgdown --><script src="../pkgdown.js"></script><link href="../extra.css" rel="stylesheet"><meta property="og:title" content="Compute the Pearson correlation coefficient matrix. — op_corrcoef"><meta name="description" content="Compute the Pearson correlation coefficient matrix."><meta property="og:description" content="Compute the Pearson correlation coefficient matrix."><meta name="robots" content="noindex"><!-- Google Tag Manager --><script>(function(w,d,s,l,i){w[l]=w[l]||[];w[l].push({'gtm.start': new Date().getTime(),event:'gtm.js'});var f=d.getElementsByTagName(s)[0],j=d.createElement(s),dl=l!='dataLayer'?'&l='+l:'';j.async=true;j.src='https://www.googletagmanager.com/gtm.js?id='+i+dl;f.parentNode.insertBefore(j,f);})(window,document,'script','dataLayer','GTM-KHBDBW7');</script><!-- End Google Tag Manager --></head><body>
+    <a href="#main" class="visually-hidden-focusable">Skip to contents</a>
+    <!-- Google Tag Manager (noscript) --> <noscript><iframe src="https://www.googletagmanager.com/ns.html?id=GTM-KHBDBW7" height="0" width="0" style="display:none;visibility:hidden"></iframe></noscript> <!-- End Google Tag Manager (noscript) -->
+
+
+    <nav class="navbar navbar-expand-lg fixed-top bg-primary" data-bs-theme="inverse" aria-label="Site navigation"><div class="container">
+
+    <a class="navbar-brand me-2" href="../index.html">keras3</a>
+
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
+
+
+    <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
+      <span class="navbar-toggler-icon"></span>
+    </button>
+
+    <div id="navbar" class="collapse navbar-collapse ms-3">
+      <ul class="navbar-nav me-auto"><li class="nav-item"><a class="nav-link" href="../articles/getting_started.html">Getting Started</a></li>
+<li class="nav-item dropdown">
+  <button class="nav-link dropdown-toggle" type="button" id="dropdown-guides" data-bs-toggle="dropdown" aria-expanded="false" aria-haspopup="true">Guides</button>
+  <ul class="dropdown-menu" aria-labelledby="dropdown-guides"><li><h6 class="dropdown-header" data-toc-skip>Model definition</h6></li>
+    <li><a class="dropdown-item" href="../articles/sequential_model.html">Sequential Model</a></li>
+    <li><a class="dropdown-item" href="../articles/functional_api.html">Functional API</a></li>
+    <li><h6 class="dropdown-header" data-toc-skip>Extending and customizing</h6></li>
+    <li><a class="dropdown-item" href="../articles/training_with_built_in_methods.html">Training &amp; evaluation with the built-in methods</a></li>
+    <li><a class="dropdown-item" href="../articles/custom_train_step_in_tensorflow.html">Customizing `fit()` with Tensorflow</a></li>
+    <li><a class="dropdown-item" href="../articles/writing_your_own_callbacks.html">Writing your own callbacks</a></li>
+    <li><a class="dropdown-item" href="../articles/making_new_layers_and_models_via_subclassing.html">Making new layers and models via subclassing</a></li>
+    <li><a class="dropdown-item" href="../articles/writing_a_custom_training_loop_in_tensorflow.html">Writing a training loop from scratch in TensorFlow</a></li>
+    <li><a class="dropdown-item" href="../articles/serialization_and_saving.html">Serialization and Saving</a></li>
+    <li><h6 class="dropdown-header" data-toc-skip>Other topics</h6></li>
+    <li><a class="dropdown-item" href="../articles/transfer_learning.html">Transfer learning and fine tuning</a></li>
+    <li><a class="dropdown-item" href="../articles/distributed_training_with_tensorflow.html">Distributed training with TensorFlow</a></li>
+    <li><a class="dropdown-item" href="../articles/distribution.html">Distributed training with Jax</a></li>
+  </ul></li>
+<li class="nav-item"><a class="nav-link" href="../articles/examples/index.html">Examples</a></li>
+<li class="active nav-item"><a class="nav-link" href="../reference/index.html">Reference</a></li>
+<li class="nav-item"><a class="nav-link" href="../news/index.html">News</a></li>
+      </ul><ul class="navbar-nav"><li class="nav-item"><form class="form-inline" role="search">
+ <input class="form-control" type="search" name="search-input" id="search-input" autocomplete="off" aria-label="Search site" placeholder="Search for" data-search-index="../search.json"></form></li>
+<li class="nav-item"><a class="external-link nav-link" href="https://github.com/rstudio/keras3/" aria-label="GitHub"><span class="fa fab fa-github fa-lg"></span></a></li>
+      </ul></div>
+
+
+  </div>
+</nav><div class="container template-reference-topic">
+<div class="row">
+  <main id="main" class="col-md-9"><div class="page-header">
+
+      <h1>Compute the Pearson correlation coefficient matrix.</h1>
+      <small class="dont-index">Source: <a href="https://github.com/rstudio/keras3/blob/HEAD/R/ops-numpy.R" class="external-link"><code>R/ops-numpy.R</code></a></small>
+      <div class="d-none name"><code>op_corrcoef.Rd</code></div>
+    </div>
+
+    <div class="ref-description section level2">
+    <p>Compute the Pearson correlation coefficient matrix.</p>
+    </div>
+
+    <div class="section level2">
+    <h2 id="ref-usage">Usage<a class="anchor" aria-label="anchor" href="#ref-usage"></a></h2>
+    <div class="sourceCode"><pre class="sourceCode r"><code><span><span class="fu">op_corrcoef</span><span class="op">(</span><span class="va">x</span><span class="op">)</span></span></code></pre></div>
+    </div>
+
+    <div class="section level2">
+    <h2 id="arguments">Arguments<a class="anchor" aria-label="anchor" href="#arguments"></a></h2>
+
+
+<dl><dt id="arg-x">x<a class="anchor" aria-label="anchor" href="#arg-x"></a></dt>
+<dd><p>A 2D tensor of shape <code>(N, D)</code>, where <code>N</code> is the number of variables
+and <code>D</code> is the number of observations.Examples</p>
+<p></p><div class="sourceCode r"><pre><code><span><span class="va">x</span> <span class="op">&lt;-</span> <span class="fu"><a href="../reference/op_array.html">op_array</a></span><span class="op">(</span><span class="fu"><a href="https://rdrr.io/r/base/matrix.html" class="external-link">matrix</a></span><span class="op">(</span><span class="fu"><a href="https://rdrr.io/r/base/c.html" class="external-link">c</a></span><span class="op">(</span><span class="fl">1</span>, <span class="fl">2</span>, <span class="fl">3</span>,</span>
+<span>                       <span class="fl">2</span>, <span class="fl">3</span>, <span class="fl">4</span><span class="op">)</span>, nrow <span class="op">=</span> <span class="fl">2</span>, byrow <span class="op">=</span> <span class="cn">TRUE</span><span class="op">)</span><span class="op">)</span></span>
+<span><span class="fu"><a href="../reference/op_corrcoef.html">op_corrcoef</a></span><span class="op">(</span><span class="va">x</span><span class="op">)</span></span></code></pre><p></p></div>
+<p></p><div class="sourceCode"><pre><code><span id="cb1-1"><a href="#cb1-1" tabindex="-1"></a><span class="do">## tf.Tensor(</span></span>
+<span id="cb1-2"><a href="#cb1-2" tabindex="-1"></a><span class="do">## [[1. 1.]</span></span>
+<span id="cb1-3"><a href="#cb1-3" tabindex="-1"></a><span class="do">##  [1. 1.]], shape=(2, 2), dtype=float64)</span></span>
+<span id="cb1-4"><a href="#cb1-4" tabindex="-1"></a></span></code></pre><p></p></div></dd>
+
+</dl></div>
+    <div class="section level2">
+    <h2 id="value">Value<a class="anchor" aria-label="anchor" href="#value"></a></h2>
+    <p>A tensor of shape <code>(N, N)</code> representing the correlation matrix.</p>
+    </div>
+    <div class="section level2">
+    <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
+    <div class="dont-index"><p>Other numpy ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p>
+<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_layer_normalization.html">op_layer_normalization</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparse_sigmoid.html">op_sparse_sigmoid</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
+    </div>
+
+  </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
+    </nav></aside></div>
+
+
+    <footer><div class="pkgdown-footer-left">
+  <p>Developed by Tomasz Kalinowski, JJ Allaire, François Chollet, Posit Software, PBC, Google.</p>
+</div>
+
+<div class="pkgdown-footer-right">
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
+</div>
+
+    </footer></div>
+
+
+
+
+
+  </body></html>
+
diff --git a/docs/dev/reference/op_correlate.html b/docs/dev/reference/op_correlate.html
index 4849df660..9ff007908 100644
--- a/docs/dev/reference/op_correlate.html
+++ b/docs/dev/reference/op_correlate.html
@@ -8,7 +8,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -89,8 +89,8 @@ <h2 id="value">Value<a class="anchor" aria-label="anchor" href="#value"></a></h2
     </div>
     <div class="section level2">
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
-    <div class="dont-index"><p>Other numpy ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p>
-<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
+    <div class="dont-index"><p>Other numpy ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p>
+<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_layer_normalization.html">op_layer_normalization</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparse_sigmoid.html">op_sparse_sigmoid</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -102,7 +102,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/op_cos.html b/docs/dev/reference/op_cos.html
index b2436206a..80f29828f 100644
--- a/docs/dev/reference/op_cos.html
+++ b/docs/dev/reference/op_cos.html
@@ -8,7 +8,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -78,8 +78,8 @@ <h2 id="value">Value<a class="anchor" aria-label="anchor" href="#value"></a></h2
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
     <div class="dont-index">
 <ul><li><p><a href="https://keras.io/api/ops/numpy#cos-function" class="external-link">https://keras.io/api/ops/numpy#cos-function</a></p></li>
-</ul><p>Other numpy ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p>
-<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
+</ul><p>Other numpy ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p>
+<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_layer_normalization.html">op_layer_normalization</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparse_sigmoid.html">op_sparse_sigmoid</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -91,7 +91,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/op_cosh.html b/docs/dev/reference/op_cosh.html
index 4387a4a53..e3c168134 100644
--- a/docs/dev/reference/op_cosh.html
+++ b/docs/dev/reference/op_cosh.html
@@ -8,7 +8,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -78,8 +78,8 @@ <h2 id="value">Value<a class="anchor" aria-label="anchor" href="#value"></a></h2
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
     <div class="dont-index">
 <ul><li><p><a href="https://keras.io/api/ops/numpy#cosh-function" class="external-link">https://keras.io/api/ops/numpy#cosh-function</a></p></li>
-</ul><p>Other numpy ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p>
-<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
+</ul><p>Other numpy ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p>
+<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_layer_normalization.html">op_layer_normalization</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparse_sigmoid.html">op_sparse_sigmoid</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -91,7 +91,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/op_count_nonzero.html b/docs/dev/reference/op_count_nonzero.html
index 1023852ac..6a0cdc43a 100644
--- a/docs/dev/reference/op_count_nonzero.html
+++ b/docs/dev/reference/op_count_nonzero.html
@@ -8,7 +8,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -97,8 +97,8 @@ <h2 id="examples">Examples<a class="anchor" aria-label="anchor" href="#examples"
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
     <div class="dont-index">
 <ul><li><p><a href="https://keras.io/api/ops/numpy#countnonzero-function" class="external-link">https://keras.io/api/ops/numpy#countnonzero-function</a></p></li>
-</ul><p>Other numpy ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p>
-<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
+</ul><p>Other numpy ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p>
+<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_layer_normalization.html">op_layer_normalization</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparse_sigmoid.html">op_sparse_sigmoid</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -110,7 +110,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/op_cross.html b/docs/dev/reference/op_cross.html
index 5d82bfad1..1bdcb9899 100644
--- a/docs/dev/reference/op_cross.html
+++ b/docs/dev/reference/op_cross.html
@@ -24,7 +24,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -131,8 +131,8 @@ <h2 id="note">Note<a class="anchor" aria-label="anchor" href="#note"></a></h2>
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
     <div class="dont-index">
 <ul><li><p><a href="https://keras.io/api/ops/numpy#cross-function" class="external-link">https://keras.io/api/ops/numpy#cross-function</a></p></li>
-</ul><p>Other numpy ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p>
-<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
+</ul><p>Other numpy ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p>
+<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_layer_normalization.html">op_layer_normalization</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparse_sigmoid.html">op_sparse_sigmoid</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -144,7 +144,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/op_ctc_decode.html b/docs/dev/reference/op_ctc_decode.html
index cae3c76c8..fa5370a96 100644
--- a/docs/dev/reference/op_ctc_decode.html
+++ b/docs/dev/reference/op_ctc_decode.html
@@ -8,7 +8,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -125,8 +125,8 @@ <h2 id="value">Value<a class="anchor" aria-label="anchor" href="#value"></a></h2
 </ul></div>
     <div class="section level2">
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
-    <div class="dont-index"><p>Other numpy ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p>
-<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
+    <div class="dont-index"><p>Other numpy ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p>
+<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_layer_normalization.html">op_layer_normalization</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparse_sigmoid.html">op_sparse_sigmoid</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -138,7 +138,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/op_ctc_loss.html b/docs/dev/reference/op_ctc_loss.html
index 023675c40..bd81149c7 100644
--- a/docs/dev/reference/op_ctc_loss.html
+++ b/docs/dev/reference/op_ctc_loss.html
@@ -8,7 +8,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -97,8 +97,8 @@ <h2 id="value">Value<a class="anchor" aria-label="anchor" href="#value"></a></h2
     </div>
     <div class="section level2">
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
-    <div class="dont-index"><p>Other nn ops: <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br></p>
-<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
+    <div class="dont-index"><p>Other nn ops: <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_layer_normalization.html">op_layer_normalization</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparse_sigmoid.html">op_sparse_sigmoid</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br></p>
+<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_layer_normalization.html">op_layer_normalization</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparse_sigmoid.html">op_sparse_sigmoid</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -110,7 +110,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/op_cumprod.html b/docs/dev/reference/op_cumprod.html
index 59d007833..ed56d5db6 100644
--- a/docs/dev/reference/op_cumprod.html
+++ b/docs/dev/reference/op_cumprod.html
@@ -8,7 +8,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -87,8 +87,8 @@ <h2 id="value">Value<a class="anchor" aria-label="anchor" href="#value"></a></h2
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
     <div class="dont-index">
 <ul><li><p><a href="https://keras.io/api/ops/numpy#cumprod-function" class="external-link">https://keras.io/api/ops/numpy#cumprod-function</a></p></li>
-</ul><p>Other numpy ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p>
-<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
+</ul><p>Other numpy ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p>
+<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_layer_normalization.html">op_layer_normalization</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparse_sigmoid.html">op_sparse_sigmoid</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -100,7 +100,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/op_cumsum.html b/docs/dev/reference/op_cumsum.html
index 01e9ce367..60e305beb 100644
--- a/docs/dev/reference/op_cumsum.html
+++ b/docs/dev/reference/op_cumsum.html
@@ -8,7 +8,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -87,8 +87,8 @@ <h2 id="value">Value<a class="anchor" aria-label="anchor" href="#value"></a></h2
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
     <div class="dont-index">
 <ul><li><p><a href="https://keras.io/api/ops/numpy#cumsum-function" class="external-link">https://keras.io/api/ops/numpy#cumsum-function</a></p></li>
-</ul><p>Other numpy ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p>
-<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
+</ul><p>Other numpy ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p>
+<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_layer_normalization.html">op_layer_normalization</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparse_sigmoid.html">op_sparse_sigmoid</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -100,7 +100,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/op_custom_gradient.html b/docs/dev/reference/op_custom_gradient.html
index 2e21e666c..1b4917dad 100644
--- a/docs/dev/reference/op_custom_gradient.html
+++ b/docs/dev/reference/op_custom_gradient.html
@@ -14,7 +14,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -140,7 +140,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
     <div class="dont-index">
 <ul><li><p><a href="https://www.tensorflow.org/api_docs/python/tf/keras/ops/custom_gradient" class="external-link">https://www.tensorflow.org/api_docs/python/tf/keras/ops/custom_gradient</a></p></li>
 </ul><p>Other core ops: <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br></p>
-<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
+<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_layer_normalization.html">op_layer_normalization</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparse_sigmoid.html">op_sparse_sigmoid</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -152,7 +152,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/op_deg2rad.html b/docs/dev/reference/op_deg2rad.html
new file mode 100644
index 000000000..31b883e03
--- /dev/null
+++ b/docs/dev/reference/op_deg2rad.html
@@ -0,0 +1,111 @@
+<!DOCTYPE html>
+<!-- Generated by pkgdown: do not edit by hand --><html lang="en"><head><meta http-equiv="Content-Type" content="text/html; charset=UTF-8"><meta charset="utf-8"><meta http-equiv="X-UA-Compatible" content="IE=edge"><meta name="viewport" content="width=device-width, initial-scale=1, shrink-to-fit=no"><title>Convert angles from degrees to radians. — op_deg2rad • keras3</title><!-- favicons --><link rel="icon" type="image/png" sizes="96x96" href="../favicon-96x96.png"><link rel="icon" type="”image/svg+xml”" href="../favicon.svg"><link rel="apple-touch-icon" sizes="180x180" href="../apple-touch-icon.png"><link rel="icon" sizes="any" href="../favicon.ico"><link rel="manifest" href="../site.webmanifest"><script src="../deps/jquery-3.6.0/jquery-3.6.0.min.js"></script><meta name="viewport" content="width=device-width, initial-scale=1, shrink-to-fit=no"><link href="../deps/bootstrap-5.3.1/bootstrap.min.css" rel="stylesheet"><script src="../deps/bootstrap-5.3.1/bootstrap.bundle.min.js"></script><link href="../deps/font-awesome-6.5.2/css/all.min.css" rel="stylesheet"><link href="../deps/font-awesome-6.5.2/css/v4-shims.min.css" rel="stylesheet"><script src="../deps/headroom-0.11.0/headroom.min.js"></script><script src="../deps/headroom-0.11.0/jQuery.headroom.min.js"></script><script src="../deps/bootstrap-toc-1.0.1/bootstrap-toc.min.js"></script><script src="../deps/clipboard.js-2.0.11/clipboard.min.js"></script><script src="../deps/search-1.0.0/autocomplete.jquery.min.js"></script><script src="../deps/search-1.0.0/fuse.min.js"></script><script src="../deps/search-1.0.0/mark.min.js"></script><!-- pkgdown --><script src="../pkgdown.js"></script><link href="../extra.css" rel="stylesheet"><meta property="og:title" content="Convert angles from degrees to radians. — op_deg2rad"><meta name="description" content="The conversion is defined as:
+rad = deg * (pi / 180)."><meta property="og:description" content="The conversion is defined as:
+rad = deg * (pi / 180)."><meta name="robots" content="noindex"><!-- Google Tag Manager --><script>(function(w,d,s,l,i){w[l]=w[l]||[];w[l].push({'gtm.start': new Date().getTime(),event:'gtm.js'});var f=d.getElementsByTagName(s)[0],j=d.createElement(s),dl=l!='dataLayer'?'&l='+l:'';j.async=true;j.src='https://www.googletagmanager.com/gtm.js?id='+i+dl;f.parentNode.insertBefore(j,f);})(window,document,'script','dataLayer','GTM-KHBDBW7');</script><!-- End Google Tag Manager --></head><body>
+    <a href="#main" class="visually-hidden-focusable">Skip to contents</a>
+    <!-- Google Tag Manager (noscript) --> <noscript><iframe src="https://www.googletagmanager.com/ns.html?id=GTM-KHBDBW7" height="0" width="0" style="display:none;visibility:hidden"></iframe></noscript> <!-- End Google Tag Manager (noscript) -->
+
+
+    <nav class="navbar navbar-expand-lg fixed-top bg-primary" data-bs-theme="inverse" aria-label="Site navigation"><div class="container">
+
+    <a class="navbar-brand me-2" href="../index.html">keras3</a>
+
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
+
+
+    <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
+      <span class="navbar-toggler-icon"></span>
+    </button>
+
+    <div id="navbar" class="collapse navbar-collapse ms-3">
+      <ul class="navbar-nav me-auto"><li class="nav-item"><a class="nav-link" href="../articles/getting_started.html">Getting Started</a></li>
+<li class="nav-item dropdown">
+  <button class="nav-link dropdown-toggle" type="button" id="dropdown-guides" data-bs-toggle="dropdown" aria-expanded="false" aria-haspopup="true">Guides</button>
+  <ul class="dropdown-menu" aria-labelledby="dropdown-guides"><li><h6 class="dropdown-header" data-toc-skip>Model definition</h6></li>
+    <li><a class="dropdown-item" href="../articles/sequential_model.html">Sequential Model</a></li>
+    <li><a class="dropdown-item" href="../articles/functional_api.html">Functional API</a></li>
+    <li><h6 class="dropdown-header" data-toc-skip>Extending and customizing</h6></li>
+    <li><a class="dropdown-item" href="../articles/training_with_built_in_methods.html">Training &amp; evaluation with the built-in methods</a></li>
+    <li><a class="dropdown-item" href="../articles/custom_train_step_in_tensorflow.html">Customizing `fit()` with Tensorflow</a></li>
+    <li><a class="dropdown-item" href="../articles/writing_your_own_callbacks.html">Writing your own callbacks</a></li>
+    <li><a class="dropdown-item" href="../articles/making_new_layers_and_models_via_subclassing.html">Making new layers and models via subclassing</a></li>
+    <li><a class="dropdown-item" href="../articles/writing_a_custom_training_loop_in_tensorflow.html">Writing a training loop from scratch in TensorFlow</a></li>
+    <li><a class="dropdown-item" href="../articles/serialization_and_saving.html">Serialization and Saving</a></li>
+    <li><h6 class="dropdown-header" data-toc-skip>Other topics</h6></li>
+    <li><a class="dropdown-item" href="../articles/transfer_learning.html">Transfer learning and fine tuning</a></li>
+    <li><a class="dropdown-item" href="../articles/distributed_training_with_tensorflow.html">Distributed training with TensorFlow</a></li>
+    <li><a class="dropdown-item" href="../articles/distribution.html">Distributed training with Jax</a></li>
+  </ul></li>
+<li class="nav-item"><a class="nav-link" href="../articles/examples/index.html">Examples</a></li>
+<li class="active nav-item"><a class="nav-link" href="../reference/index.html">Reference</a></li>
+<li class="nav-item"><a class="nav-link" href="../news/index.html">News</a></li>
+      </ul><ul class="navbar-nav"><li class="nav-item"><form class="form-inline" role="search">
+ <input class="form-control" type="search" name="search-input" id="search-input" autocomplete="off" aria-label="Search site" placeholder="Search for" data-search-index="../search.json"></form></li>
+<li class="nav-item"><a class="external-link nav-link" href="https://github.com/rstudio/keras3/" aria-label="GitHub"><span class="fa fab fa-github fa-lg"></span></a></li>
+      </ul></div>
+
+
+  </div>
+</nav><div class="container template-reference-topic">
+<div class="row">
+  <main id="main" class="col-md-9"><div class="page-header">
+
+      <h1>Convert angles from degrees to radians.</h1>
+      <small class="dont-index">Source: <a href="https://github.com/rstudio/keras3/blob/HEAD/R/ops-numpy.R" class="external-link"><code>R/ops-numpy.R</code></a></small>
+      <div class="d-none name"><code>op_deg2rad.Rd</code></div>
+    </div>
+
+    <div class="ref-description section level2">
+    <p>The conversion is defined as:
+<code>rad = deg * (pi / 180)</code>.</p>
+    </div>
+
+    <div class="section level2">
+    <h2 id="ref-usage">Usage<a class="anchor" aria-label="anchor" href="#ref-usage"></a></h2>
+    <div class="sourceCode"><pre class="sourceCode r"><code><span><span class="fu">op_deg2rad</span><span class="op">(</span><span class="va">x</span><span class="op">)</span></span></code></pre></div>
+    </div>
+
+    <div class="section level2">
+    <h2 id="arguments">Arguments<a class="anchor" aria-label="anchor" href="#arguments"></a></h2>
+
+
+<dl><dt id="arg-x">x<a class="anchor" aria-label="anchor" href="#arg-x"></a></dt>
+<dd><p>Input tensor of angles in degrees.</p></dd>
+
+</dl></div>
+    <div class="section level2">
+    <h2 id="value">Value<a class="anchor" aria-label="anchor" href="#value"></a></h2>
+    <p>A tensor containing angles converted to radians.</p>
+    </div>
+    <div class="section level2">
+    <h2 id="examples">Examples<a class="anchor" aria-label="anchor" href="#examples"></a></h2>
+    <p></p><div class="sourceCode r"><pre><code><span><span class="fu"><a href="../reference/op_deg2rad.html">op_deg2rad</a></span><span class="op">(</span><span class="fu"><a href="https://rdrr.io/r/base/c.html" class="external-link">c</a></span><span class="op">(</span><span class="fl">0</span>, <span class="fl">90</span>, <span class="fl">180</span><span class="op">)</span><span class="op">)</span></span></code></pre><p></p></div>
+<p></p><div class="sourceCode"><pre><code><span id="cb1-1"><a href="#cb1-1" tabindex="-1"></a><span class="do">## tf.Tensor([0.        1.5707964 3.1415927], shape=(3), dtype=float32)</span></span>
+<span id="cb1-2"><a href="#cb1-2" tabindex="-1"></a></span></code></pre><p></p></div>
+    </div>
+    <div class="section level2">
+    <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
+    <div class="dont-index"><p>Other numpy ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p>
+<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_layer_normalization.html">op_layer_normalization</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparse_sigmoid.html">op_sparse_sigmoid</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
+    </div>
+
+  </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
+    </nav></aside></div>
+
+
+    <footer><div class="pkgdown-footer-left">
+  <p>Developed by Tomasz Kalinowski, JJ Allaire, François Chollet, Posit Software, PBC, Google.</p>
+</div>
+
+<div class="pkgdown-footer-right">
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
+</div>
+
+    </footer></div>
+
+
+
+
+
+  </body></html>
+
diff --git a/docs/dev/reference/op_depthwise_conv.html b/docs/dev/reference/op_depthwise_conv.html
index c3280e6f2..f96f7e642 100644
--- a/docs/dev/reference/op_depthwise_conv.html
+++ b/docs/dev/reference/op_depthwise_conv.html
@@ -8,7 +8,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -126,8 +126,8 @@ <h2 id="value">Value<a class="anchor" aria-label="anchor" href="#value"></a></h2
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
     <div class="dont-index">
 <ul><li><p><a href="https://keras.io/api/ops/nn#depthwiseconv-function" class="external-link">https://keras.io/api/ops/nn#depthwiseconv-function</a></p></li>
-</ul><p>Other nn ops: <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br></p>
-<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
+</ul><p>Other nn ops: <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_layer_normalization.html">op_layer_normalization</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparse_sigmoid.html">op_sparse_sigmoid</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br></p>
+<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_layer_normalization.html">op_layer_normalization</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparse_sigmoid.html">op_sparse_sigmoid</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -139,7 +139,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/op_det.html b/docs/dev/reference/op_det.html
index eea613089..19f57c71b 100644
--- a/docs/dev/reference/op_det.html
+++ b/docs/dev/reference/op_det.html
@@ -8,7 +8,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -77,7 +77,7 @@ <h2 id="value">Value<a class="anchor" aria-label="anchor" href="#value"></a></h2
     <div class="section level2">
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
     <div class="dont-index"><p>Other linear algebra ops: <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br></p>
-<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
+<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_layer_normalization.html">op_layer_normalization</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparse_sigmoid.html">op_sparse_sigmoid</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -89,7 +89,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/op_diag.html b/docs/dev/reference/op_diag.html
index 105631606..24fea0bd9 100644
--- a/docs/dev/reference/op_diag.html
+++ b/docs/dev/reference/op_diag.html
@@ -8,7 +8,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -110,8 +110,8 @@ <h2 id="examples">Examples<a class="anchor" aria-label="anchor" href="#examples"
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
     <div class="dont-index">
 <ul><li><p><a href="https://keras.io/api/ops/numpy#diag-function" class="external-link">https://keras.io/api/ops/numpy#diag-function</a></p></li>
-</ul><p>Other numpy ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p>
-<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
+</ul><p>Other numpy ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p>
+<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_layer_normalization.html">op_layer_normalization</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparse_sigmoid.html">op_sparse_sigmoid</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -123,7 +123,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/op_diagflat.html b/docs/dev/reference/op_diagflat.html
index 26d0197f2..29db8e668 100644
--- a/docs/dev/reference/op_diagflat.html
+++ b/docs/dev/reference/op_diagflat.html
@@ -8,7 +8,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -82,8 +82,8 @@ <h2 id="value">Value<a class="anchor" aria-label="anchor" href="#value"></a></h2
     </div>
     <div class="section level2">
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
-    <div class="dont-index"><p>Other numpy ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p>
-<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
+    <div class="dont-index"><p>Other numpy ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p>
+<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_layer_normalization.html">op_layer_normalization</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparse_sigmoid.html">op_sparse_sigmoid</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -95,7 +95,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/op_diagonal.html b/docs/dev/reference/op_diagonal.html
index 0e0869866..88cdbd91b 100644
--- a/docs/dev/reference/op_diagonal.html
+++ b/docs/dev/reference/op_diagonal.html
@@ -22,7 +22,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -153,8 +153,8 @@ <h2 id="examples">Examples<a class="anchor" aria-label="anchor" href="#examples"
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
     <div class="dont-index">
 <ul><li><p><a href="https://keras.io/api/ops/numpy#diagonal-function" class="external-link">https://keras.io/api/ops/numpy#diagonal-function</a></p></li>
-</ul><p>Other numpy ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p>
-<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
+</ul><p>Other numpy ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p>
+<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_layer_normalization.html">op_layer_normalization</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparse_sigmoid.html">op_sparse_sigmoid</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -166,7 +166,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/op_diff.html b/docs/dev/reference/op_diff.html
index cd22e82c6..41d502128 100644
--- a/docs/dev/reference/op_diff.html
+++ b/docs/dev/reference/op_diff.html
@@ -12,7 +12,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -111,8 +111,8 @@ <h2 id="examples">Examples<a class="anchor" aria-label="anchor" href="#examples"
     </div>
     <div class="section level2">
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
-    <div class="dont-index"><p>Other numpy ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p>
-<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
+    <div class="dont-index"><p>Other numpy ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p>
+<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_layer_normalization.html">op_layer_normalization</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparse_sigmoid.html">op_sparse_sigmoid</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -124,7 +124,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/op_digitize.html b/docs/dev/reference/op_digitize.html
index 772298235..1b7e43505 100644
--- a/docs/dev/reference/op_digitize.html
+++ b/docs/dev/reference/op_digitize.html
@@ -8,7 +8,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -97,8 +97,8 @@ <h2 id="examples">Examples<a class="anchor" aria-label="anchor" href="#examples"
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
     <div class="dont-index">
 <ul><li><p><a href="https://keras.io/api/ops/numpy#digitize-function" class="external-link">https://keras.io/api/ops/numpy#digitize-function</a></p></li>
-</ul><p>Other numpy ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p>
-<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
+</ul><p>Other numpy ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p>
+<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_layer_normalization.html">op_layer_normalization</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparse_sigmoid.html">op_sparse_sigmoid</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -110,7 +110,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/op_divide.html b/docs/dev/reference/op_divide.html
index 7a6214c85..900f4f5af 100644
--- a/docs/dev/reference/op_divide.html
+++ b/docs/dev/reference/op_divide.html
@@ -38,7 +38,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -127,8 +127,8 @@ <h2 id="example">Example<a class="anchor" aria-label="anchor" href="#example"></
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
     <div class="dont-index">
 <ul><li><p><a href="https://keras.io/api/ops/numpy#divide-function" class="external-link">https://keras.io/api/ops/numpy#divide-function</a></p></li>
-</ul><p>Other numpy ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p>
-<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
+</ul><p>Other numpy ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p>
+<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_layer_normalization.html">op_layer_normalization</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparse_sigmoid.html">op_sparse_sigmoid</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -140,7 +140,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/op_divide_no_nan.html b/docs/dev/reference/op_divide_no_nan.html
index 81974578b..11801fb4a 100644
--- a/docs/dev/reference/op_divide_no_nan.html
+++ b/docs/dev/reference/op_divide_no_nan.html
@@ -8,7 +8,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -80,8 +80,8 @@ <h2 id="value">Value<a class="anchor" aria-label="anchor" href="#value"></a></h2
     </div>
     <div class="section level2">
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
-    <div class="dont-index"><p>Other numpy ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p>
-<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
+    <div class="dont-index"><p>Other numpy ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p>
+<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_layer_normalization.html">op_layer_normalization</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparse_sigmoid.html">op_sparse_sigmoid</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -93,7 +93,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/op_dot.html b/docs/dev/reference/op_dot.html
index e1558f62e..5a5cc1ec4 100644
--- a/docs/dev/reference/op_dot.html
+++ b/docs/dev/reference/op_dot.html
@@ -32,7 +32,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -119,8 +119,8 @@ <h2 id="note">Note<a class="anchor" aria-label="anchor" href="#note"></a></h2>
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
     <div class="dont-index">
 <ul><li><p><a href="https://keras.io/api/ops/numpy#dot-function" class="external-link">https://keras.io/api/ops/numpy#dot-function</a></p></li>
-</ul><p>Other numpy ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p>
-<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
+</ul><p>Other numpy ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p>
+<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_layer_normalization.html">op_layer_normalization</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparse_sigmoid.html">op_sparse_sigmoid</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -132,7 +132,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/op_dot_product_attention.html b/docs/dev/reference/op_dot_product_attention.html
index ab4e6598e..7443b77e4 100644
--- a/docs/dev/reference/op_dot_product_attention.html
+++ b/docs/dev/reference/op_dot_product_attention.html
@@ -34,7 +34,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -103,7 +103,8 @@ <h2 id="ref-usage">Usage<a class="anchor" aria-label="anchor" href="#ref-usage">
 <span>  mask <span class="op">=</span> <span class="cn">NULL</span>,</span>
 <span>  scale <span class="op">=</span> <span class="cn">NULL</span>,</span>
 <span>  is_causal <span class="op">=</span> <span class="cn">FALSE</span>,</span>
-<span>  flash_attention <span class="op">=</span> <span class="cn">NULL</span></span>
+<span>  flash_attention <span class="op">=</span> <span class="cn">NULL</span>,</span>
+<span>  attn_logits_soft_cap <span class="op">=</span> <span class="cn">NULL</span></span>
 <span><span class="op">)</span></span></code></pre></div>
     </div>
 
@@ -154,6 +155,11 @@ <h2 id="arguments">Arguments<a class="anchor" aria-label="anchor" href="#argumen
 Typically, the inputs must be in float16 and bfloat16 dtype and the
 input layout requirements may vary depending on the backend.</p></dd>
 
+
+<dt id="arg-attn-logits-soft-cap">attn_logits_soft_cap<a class="anchor" aria-label="anchor" href="#arg-attn-logits-soft-cap"></a></dt>
+<dd><p>Optional numeric cap on the attention logits before softmax. Only supported
+on the JAX TPU backend.</p></dd>
+
 </dl></div>
     <div class="section level2">
     <h2 id="value">Value<a class="anchor" aria-label="anchor" href="#value"></a></h2>
@@ -170,8 +176,8 @@ <h2 id="examples">Examples<a class="anchor" aria-label="anchor" href="#examples"
     </div>
     <div class="section level2">
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
-    <div class="dont-index"><p>Other nn ops: <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br></p>
-<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
+    <div class="dont-index"><p>Other nn ops: <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_layer_normalization.html">op_layer_normalization</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparse_sigmoid.html">op_sparse_sigmoid</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br></p>
+<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_layer_normalization.html">op_layer_normalization</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparse_sigmoid.html">op_sparse_sigmoid</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -183,7 +189,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/op_dtype.html b/docs/dev/reference/op_dtype.html
index a25526461..8b62de9c8 100644
--- a/docs/dev/reference/op_dtype.html
+++ b/docs/dev/reference/op_dtype.html
@@ -10,7 +10,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -88,7 +88,7 @@ <h2 id="examples">Examples<a class="anchor" aria-label="anchor" href="#examples"
     <div class="section level2">
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
     <div class="dont-index"><p>Other core ops: <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br></p>
-<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
+<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_layer_normalization.html">op_layer_normalization</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparse_sigmoid.html">op_sparse_sigmoid</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -100,7 +100,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/op_eig.html b/docs/dev/reference/op_eig.html
index 4813cdca6..411a8254a 100644
--- a/docs/dev/reference/op_eig.html
+++ b/docs/dev/reference/op_eig.html
@@ -8,7 +8,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -78,7 +78,7 @@ <h2 id="value">Value<a class="anchor" aria-label="anchor" href="#value"></a></h2
     <div class="section level2">
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
     <div class="dont-index"><p>Other linear algebra ops: <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br></p>
-<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
+<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_layer_normalization.html">op_layer_normalization</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparse_sigmoid.html">op_sparse_sigmoid</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -90,7 +90,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/op_eigh.html b/docs/dev/reference/op_eigh.html
index 5db5e5810..a4f9465d3 100644
--- a/docs/dev/reference/op_eigh.html
+++ b/docs/dev/reference/op_eigh.html
@@ -8,7 +8,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -78,7 +78,7 @@ <h2 id="value">Value<a class="anchor" aria-label="anchor" href="#value"></a></h2
     <div class="section level2">
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
     <div class="dont-index"><p>Other linear algebra ops: <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br></p>
-<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
+<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_layer_normalization.html">op_layer_normalization</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparse_sigmoid.html">op_sparse_sigmoid</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -90,7 +90,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/op_einsum.html b/docs/dev/reference/op_einsum.html
index 1f21302c1..3de8f0d57 100644
--- a/docs/dev/reference/op_einsum.html
+++ b/docs/dev/reference/op_einsum.html
@@ -8,7 +8,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -75,7 +75,8 @@ <h2 id="arguments">Arguments<a class="anchor" aria-label="anchor" href="#argumen
 
 
 <dt id="arg--">...<a class="anchor" aria-label="anchor" href="#arg--"></a></dt>
-<dd><p>The operands to compute the Einstein sum of.</p></dd>
+<dd><p>The operands to compute the Einstein sum of as unnamed arguments.
+Additional named arguments are forwarded to the underlying backend.</p></dd>
 
 </dl></div>
     <div class="section level2">
@@ -144,8 +145,8 @@ <h2 id="examples">Examples<a class="anchor" aria-label="anchor" href="#examples"
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
     <div class="dont-index">
 <ul><li><p><a href="https://keras.io/api/ops/numpy#einsum-function" class="external-link">https://keras.io/api/ops/numpy#einsum-function</a></p></li>
-</ul><p>Other numpy ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p>
-<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
+</ul><p>Other numpy ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p>
+<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_layer_normalization.html">op_layer_normalization</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparse_sigmoid.html">op_sparse_sigmoid</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -157,7 +158,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/op_elu.html b/docs/dev/reference/op_elu.html
index 7d2c5d771..738e3199c 100644
--- a/docs/dev/reference/op_elu.html
+++ b/docs/dev/reference/op_elu.html
@@ -10,7 +10,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -92,8 +92,8 @@ <h2 id="examples">Examples<a class="anchor" aria-label="anchor" href="#examples"
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
     <div class="dont-index">
 <ul><li><p><a href="https://keras.io/api/ops/nn#elu-function" class="external-link">https://keras.io/api/ops/nn#elu-function</a></p></li>
-</ul><p>Other nn ops: <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br></p>
-<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
+</ul><p>Other nn ops: <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_layer_normalization.html">op_layer_normalization</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparse_sigmoid.html">op_sparse_sigmoid</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br></p>
+<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_layer_normalization.html">op_layer_normalization</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparse_sigmoid.html">op_sparse_sigmoid</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -105,7 +105,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/op_empty.html b/docs/dev/reference/op_empty.html
index ed1edf513..b23ae0801 100644
--- a/docs/dev/reference/op_empty.html
+++ b/docs/dev/reference/op_empty.html
@@ -8,7 +8,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -82,8 +82,8 @@ <h2 id="value">Value<a class="anchor" aria-label="anchor" href="#value"></a></h2
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
     <div class="dont-index">
 <ul><li><p><a href="https://keras.io/api/ops/numpy#empty-function" class="external-link">https://keras.io/api/ops/numpy#empty-function</a></p></li>
-</ul><p>Other numpy ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p>
-<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
+</ul><p>Other numpy ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p>
+<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_layer_normalization.html">op_layer_normalization</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparse_sigmoid.html">op_sparse_sigmoid</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -95,7 +95,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/op_equal.html b/docs/dev/reference/op_equal.html
index 50d4e5db2..581b26f37 100644
--- a/docs/dev/reference/op_equal.html
+++ b/docs/dev/reference/op_equal.html
@@ -38,7 +38,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -121,8 +121,8 @@ <h2 id="value">Value<a class="anchor" aria-label="anchor" href="#value"></a></h2
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
     <div class="dont-index">
 <ul><li><p><a href="https://keras.io/api/ops/numpy#equal-function" class="external-link">https://keras.io/api/ops/numpy#equal-function</a></p></li>
-</ul><p>Other numpy ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p>
-<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
+</ul><p>Other numpy ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p>
+<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_layer_normalization.html">op_layer_normalization</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparse_sigmoid.html">op_sparse_sigmoid</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -134,7 +134,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/op_erf.html b/docs/dev/reference/op_erf.html
index 97db23091..754f5b40c 100644
--- a/docs/dev/reference/op_erf.html
+++ b/docs/dev/reference/op_erf.html
@@ -8,7 +8,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -78,14 +78,14 @@ <h2 id="value">Value<a class="anchor" aria-label="anchor" href="#value"></a></h2
     <h2 id="examples">Examples<a class="anchor" aria-label="anchor" href="#examples"></a></h2>
     <p></p><div class="sourceCode r"><pre><code><span><span class="va">x</span> <span class="op">&lt;-</span> <span class="fu"><a href="../reference/op_array.html">op_array</a></span><span class="op">(</span><span class="fu"><a href="https://rdrr.io/r/base/c.html" class="external-link">c</a></span><span class="op">(</span><span class="op">-</span><span class="fl">3</span>, <span class="op">-</span><span class="fl">2</span>, <span class="op">-</span><span class="fl">1</span>, <span class="fl">0</span>, <span class="fl">1</span><span class="op">)</span><span class="op">)</span></span>
 <span><span class="fu"><a href="../reference/op_erf.html">op_erf</a></span><span class="op">(</span><span class="va">x</span><span class="op">)</span></span></code></pre><p></p></div>
-<p></p><div class="sourceCode"><pre><code><span id="cb1-1"><a href="#cb1-1" tabindex="-1"></a><span class="do">## tf.Tensor([-0.99997777 -0.9953222  -0.84270084  0.          0.84270084], shape=(5), dtype=float32)</span></span>
+<p></p><div class="sourceCode"><pre><code><span id="cb1-1"><a href="#cb1-1" tabindex="-1"></a><span class="do">## tf.Tensor([-0.999978   -0.99532235 -0.84270084  0.          0.84270084], shape=(5), dtype=float32)</span></span>
 <span id="cb1-2"><a href="#cb1-2" tabindex="-1"></a></span></code></pre><p></p></div>
 <p></p><div class="sourceCode r"><pre><code><span><span class="co"># array([-0.99998 , -0.99532, -0.842701,  0.,  0.842701], dtype=float32)</span></span></code></pre><p></p></div>
     </div>
     <div class="section level2">
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
     <div class="dont-index"><p>Other math ops: <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br></p>
-<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
+<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_layer_normalization.html">op_layer_normalization</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparse_sigmoid.html">op_sparse_sigmoid</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -97,7 +97,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/op_erfinv.html b/docs/dev/reference/op_erfinv.html
index 12f4e6eed..ed1912f98 100644
--- a/docs/dev/reference/op_erfinv.html
+++ b/docs/dev/reference/op_erfinv.html
@@ -8,7 +8,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -84,7 +84,7 @@ <h2 id="examples">Examples<a class="anchor" aria-label="anchor" href="#examples"
     <div class="section level2">
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
     <div class="dont-index"><p>Other math ops: <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br></p>
-<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
+<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_layer_normalization.html">op_layer_normalization</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparse_sigmoid.html">op_sparse_sigmoid</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -96,7 +96,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/op_exp.html b/docs/dev/reference/op_exp.html
index 92fbb8687..0993636af 100644
--- a/docs/dev/reference/op_exp.html
+++ b/docs/dev/reference/op_exp.html
@@ -8,7 +8,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -78,8 +78,8 @@ <h2 id="value">Value<a class="anchor" aria-label="anchor" href="#value"></a></h2
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
     <div class="dont-index">
 <ul><li><p><a href="https://keras.io/api/ops/numpy#exp-function" class="external-link">https://keras.io/api/ops/numpy#exp-function</a></p></li>
-</ul><p>Other numpy ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p>
-<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
+</ul><p>Other numpy ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p>
+<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_layer_normalization.html">op_layer_normalization</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparse_sigmoid.html">op_sparse_sigmoid</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -91,7 +91,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/op_exp2.html b/docs/dev/reference/op_exp2.html
index 378a43ba2..d9d96acfa 100644
--- a/docs/dev/reference/op_exp2.html
+++ b/docs/dev/reference/op_exp2.html
@@ -8,7 +8,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -76,8 +76,8 @@ <h2 id="value">Value<a class="anchor" aria-label="anchor" href="#value"></a></h2
     </div>
     <div class="section level2">
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
-    <div class="dont-index"><p>Other numpy ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p>
-<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
+    <div class="dont-index"><p>Other numpy ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p>
+<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_layer_normalization.html">op_layer_normalization</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparse_sigmoid.html">op_sparse_sigmoid</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -89,7 +89,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/op_expand_dims.html b/docs/dev/reference/op_expand_dims.html
index c387e2866..cb8c895d9 100644
--- a/docs/dev/reference/op_expand_dims.html
+++ b/docs/dev/reference/op_expand_dims.html
@@ -8,7 +8,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -83,8 +83,8 @@ <h2 id="value">Value<a class="anchor" aria-label="anchor" href="#value"></a></h2
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
     <div class="dont-index">
 <ul><li><p><a href="https://keras.io/api/ops/numpy#expanddims-function" class="external-link">https://keras.io/api/ops/numpy#expanddims-function</a></p></li>
-</ul><p>Other numpy ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p>
-<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
+</ul><p>Other numpy ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p>
+<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_layer_normalization.html">op_layer_normalization</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparse_sigmoid.html">op_sparse_sigmoid</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -96,7 +96,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/op_expm1.html b/docs/dev/reference/op_expm1.html
index 8ff51c2de..51b4c6fac 100644
--- a/docs/dev/reference/op_expm1.html
+++ b/docs/dev/reference/op_expm1.html
@@ -8,7 +8,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -78,8 +78,8 @@ <h2 id="value">Value<a class="anchor" aria-label="anchor" href="#value"></a></h2
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
     <div class="dont-index">
 <ul><li><p><a href="https://keras.io/api/ops/numpy#expm1-function" class="external-link">https://keras.io/api/ops/numpy#expm1-function</a></p></li>
-</ul><p>Other numpy ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p>
-<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
+</ul><p>Other numpy ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p>
+<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_layer_normalization.html">op_layer_normalization</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparse_sigmoid.html">op_sparse_sigmoid</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -91,7 +91,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/op_extract_sequences.html b/docs/dev/reference/op_extract_sequences.html
index fa0b7c63a..28976445f 100644
--- a/docs/dev/reference/op_extract_sequences.html
+++ b/docs/dev/reference/op_extract_sequences.html
@@ -18,7 +18,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -111,7 +111,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
     <div class="dont-index">
 <ul><li><p><a href="https://keras.io/api/ops/core#extractsequences-function" class="external-link">https://keras.io/api/ops/core#extractsequences-function</a></p></li>
 </ul><p>Other math ops: <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br></p>
-<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
+<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_layer_normalization.html">op_layer_normalization</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparse_sigmoid.html">op_sparse_sigmoid</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -123,7 +123,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/op_eye.html b/docs/dev/reference/op_eye.html
index 9bb263ed7..28a30ad6d 100644
--- a/docs/dev/reference/op_eye.html
+++ b/docs/dev/reference/op_eye.html
@@ -8,7 +8,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -92,8 +92,8 @@ <h2 id="value">Value<a class="anchor" aria-label="anchor" href="#value"></a></h2
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
     <div class="dont-index">
 <ul><li><p><a href="https://keras.io/api/ops/numpy#eye-function" class="external-link">https://keras.io/api/ops/numpy#eye-function</a></p></li>
-</ul><p>Other numpy ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p>
-<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
+</ul><p>Other numpy ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p>
+<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_layer_normalization.html">op_layer_normalization</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparse_sigmoid.html">op_sparse_sigmoid</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -105,7 +105,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/op_fft.html b/docs/dev/reference/op_fft.html
index 83c33a0de..f7db01b6d 100644
--- a/docs/dev/reference/op_fft.html
+++ b/docs/dev/reference/op_fft.html
@@ -8,7 +8,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -93,7 +93,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
     <div class="dont-index">
 <ul><li><p><a href="https://keras.io/api/ops/fft#fft-function" class="external-link">https://keras.io/api/ops/fft#fft-function</a></p></li>
 </ul><p>Other math ops: <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br></p>
-<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
+<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_layer_normalization.html">op_layer_normalization</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparse_sigmoid.html">op_sparse_sigmoid</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -105,7 +105,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/op_fft2.html b/docs/dev/reference/op_fft2.html
index bd08187b7..f5e53ec00 100644
--- a/docs/dev/reference/op_fft2.html
+++ b/docs/dev/reference/op_fft2.html
@@ -8,7 +8,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -99,7 +99,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
     <div class="dont-index">
 <ul><li><p><a href="https://keras.io/api/ops/fft#fft2-function" class="external-link">https://keras.io/api/ops/fft#fft2-function</a></p></li>
 </ul><p>Other math ops: <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br></p>
-<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
+<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_layer_normalization.html">op_layer_normalization</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparse_sigmoid.html">op_sparse_sigmoid</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -111,7 +111,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/op_flip.html b/docs/dev/reference/op_flip.html
index eb72281c1..2e59ea867 100644
--- a/docs/dev/reference/op_flip.html
+++ b/docs/dev/reference/op_flip.html
@@ -8,7 +8,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -83,8 +83,8 @@ <h2 id="value">Value<a class="anchor" aria-label="anchor" href="#value"></a></h2
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
     <div class="dont-index">
 <ul><li><p><a href="https://keras.io/api/ops/numpy#flip-function" class="external-link">https://keras.io/api/ops/numpy#flip-function</a></p></li>
-</ul><p>Other numpy ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p>
-<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
+</ul><p>Other numpy ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p>
+<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_layer_normalization.html">op_layer_normalization</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparse_sigmoid.html">op_sparse_sigmoid</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -96,7 +96,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/op_floor.html b/docs/dev/reference/op_floor.html
index d548b016e..a2eea4144 100644
--- a/docs/dev/reference/op_floor.html
+++ b/docs/dev/reference/op_floor.html
@@ -8,7 +8,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -78,8 +78,8 @@ <h2 id="value">Value<a class="anchor" aria-label="anchor" href="#value"></a></h2
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
     <div class="dont-index">
 <ul><li><p><a href="https://keras.io/api/ops/numpy#floor-function" class="external-link">https://keras.io/api/ops/numpy#floor-function</a></p></li>
-</ul><p>Other numpy ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p>
-<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
+</ul><p>Other numpy ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p>
+<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_layer_normalization.html">op_layer_normalization</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparse_sigmoid.html">op_sparse_sigmoid</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -91,7 +91,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/op_floor_divide.html b/docs/dev/reference/op_floor_divide.html
index 718afa5f6..ddc34f949 100644
--- a/docs/dev/reference/op_floor_divide.html
+++ b/docs/dev/reference/op_floor_divide.html
@@ -38,7 +38,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -119,8 +119,8 @@ <h2 id="value">Value<a class="anchor" aria-label="anchor" href="#value"></a></h2
     </div>
     <div class="section level2">
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
-    <div class="dont-index"><p>Other numpy ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p>
-<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
+    <div class="dont-index"><p>Other numpy ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p>
+<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_layer_normalization.html">op_layer_normalization</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparse_sigmoid.html">op_sparse_sigmoid</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -132,7 +132,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/op_fori_loop.html b/docs/dev/reference/op_fori_loop.html
index 6205dd658..5f8b81511 100644
--- a/docs/dev/reference/op_fori_loop.html
+++ b/docs/dev/reference/op_fori_loop.html
@@ -8,7 +8,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -102,7 +102,7 @@ <h2 id="examples">Examples<a class="anchor" aria-label="anchor" href="#examples"
     <div class="section level2">
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
     <div class="dont-index"><p>Other core ops: <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br></p>
-<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
+<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_layer_normalization.html">op_layer_normalization</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparse_sigmoid.html">op_sparse_sigmoid</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -114,7 +114,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/op_full.html b/docs/dev/reference/op_full.html
index 213b45135..b3519d7bb 100644
--- a/docs/dev/reference/op_full.html
+++ b/docs/dev/reference/op_full.html
@@ -8,7 +8,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -86,8 +86,8 @@ <h2 id="value">Value<a class="anchor" aria-label="anchor" href="#value"></a></h2
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
     <div class="dont-index">
 <ul><li><p><a href="https://keras.io/api/ops/numpy#full-function" class="external-link">https://keras.io/api/ops/numpy#full-function</a></p></li>
-</ul><p>Other numpy ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p>
-<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
+</ul><p>Other numpy ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p>
+<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_layer_normalization.html">op_layer_normalization</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparse_sigmoid.html">op_sparse_sigmoid</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -99,7 +99,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/op_full_like.html b/docs/dev/reference/op_full_like.html
index 14238b610..29b8ba612 100644
--- a/docs/dev/reference/op_full_like.html
+++ b/docs/dev/reference/op_full_like.html
@@ -8,7 +8,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -86,8 +86,8 @@ <h2 id="value">Value<a class="anchor" aria-label="anchor" href="#value"></a></h2
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
     <div class="dont-index">
 <ul><li><p><a href="https://keras.io/api/ops/numpy#fulllike-function" class="external-link">https://keras.io/api/ops/numpy#fulllike-function</a></p></li>
-</ul><p>Other numpy ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p>
-<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
+</ul><p>Other numpy ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p>
+<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_layer_normalization.html">op_layer_normalization</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparse_sigmoid.html">op_sparse_sigmoid</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -99,7 +99,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/op_gelu.html b/docs/dev/reference/op_gelu.html
index eb6aa25c8..ef9298309 100644
--- a/docs/dev/reference/op_gelu.html
+++ b/docs/dev/reference/op_gelu.html
@@ -16,7 +16,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -94,7 +94,7 @@ <h2 id="value">Value<a class="anchor" aria-label="anchor" href="#value"></a></h2
     <h2 id="examples">Examples<a class="anchor" aria-label="anchor" href="#examples"></a></h2>
     <p></p><div class="sourceCode r"><pre><code><span><span class="va">x</span> <span class="op">&lt;-</span> <span class="fu"><a href="../reference/op_array.html">op_array</a></span><span class="op">(</span><span class="fu"><a href="https://rdrr.io/r/base/c.html" class="external-link">c</a></span><span class="op">(</span><span class="op">-</span><span class="fl">1.</span>, <span class="fl">0.</span>, <span class="fl">1.</span><span class="op">)</span><span class="op">)</span></span>
 <span><span class="fu"><a href="../reference/op_gelu.html">op_gelu</a></span><span class="op">(</span><span class="va">x</span><span class="op">)</span></span></code></pre><p></p></div>
-<p></p><div class="sourceCode"><pre><code><span id="cb1-1"><a href="#cb1-1" tabindex="-1"></a><span class="do">## tf.Tensor([-0.15880796  0.          0.841192  ], shape=(3), dtype=float32)</span></span>
+<p></p><div class="sourceCode"><pre><code><span id="cb1-1"><a href="#cb1-1" tabindex="-1"></a><span class="do">## tf.Tensor([-0.158808  0.        0.841192], shape=(3), dtype=float32)</span></span>
 <span id="cb1-2"><a href="#cb1-2" tabindex="-1"></a></span></code></pre><p></p></div>
 <p></p><div class="sourceCode r"><pre><code><span><span class="fu"><a href="../reference/op_gelu.html">op_gelu</a></span><span class="op">(</span><span class="va">x</span>, <span class="cn">FALSE</span><span class="op">)</span></span></code></pre><p></p></div>
 <p></p><div class="sourceCode"><pre><code><span id="cb1-1"><a href="#cb1-1" tabindex="-1"></a><span class="do">## tf.Tensor([-0.15865526  0.          0.8413447 ], shape=(3), dtype=float32)</span></span>
@@ -109,8 +109,8 @@ <h2 id="examples">Examples<a class="anchor" aria-label="anchor" href="#examples"
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
     <div class="dont-index">
 <ul><li><p><a href="https://keras.io/api/ops/nn#gelu-function" class="external-link">https://keras.io/api/ops/nn#gelu-function</a></p></li>
-</ul><p>Other nn ops: <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br></p>
-<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
+</ul><p>Other nn ops: <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_layer_normalization.html">op_layer_normalization</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparse_sigmoid.html">op_sparse_sigmoid</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br></p>
+<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_layer_normalization.html">op_layer_normalization</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparse_sigmoid.html">op_sparse_sigmoid</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -122,7 +122,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/op_get_item.html b/docs/dev/reference/op_get_item.html
index 25b45d1cc..d933a0422 100644
--- a/docs/dev/reference/op_get_item.html
+++ b/docs/dev/reference/op_get_item.html
@@ -8,7 +8,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -86,8 +86,8 @@ <h2 id="note">Note<a class="anchor" aria-label="anchor" href="#note"></a></h2>
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
     <div class="dont-index">
 <ul><li><p><a href="https://keras.io/api/ops/numpy#getitem-function" class="external-link">https://keras.io/api/ops/numpy#getitem-function</a></p></li>
-</ul><p>Other numpy ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p>
-<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
+</ul><p>Other numpy ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p>
+<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_layer_normalization.html">op_layer_normalization</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparse_sigmoid.html">op_sparse_sigmoid</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -99,7 +99,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/op_glu.html b/docs/dev/reference/op_glu.html
index 7fb22fcfe..b6e9a603a 100644
--- a/docs/dev/reference/op_glu.html
+++ b/docs/dev/reference/op_glu.html
@@ -12,7 +12,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -93,8 +93,8 @@ <h2 id="examples">Examples<a class="anchor" aria-label="anchor" href="#examples"
     </div>
     <div class="section level2">
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
-    <div class="dont-index"><p>Other nn ops: <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br></p>
-<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
+    <div class="dont-index"><p>Other nn ops: <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_layer_normalization.html">op_layer_normalization</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparse_sigmoid.html">op_sparse_sigmoid</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br></p>
+<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_layer_normalization.html">op_layer_normalization</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparse_sigmoid.html">op_sparse_sigmoid</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -106,7 +106,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/op_greater.html b/docs/dev/reference/op_greater.html
index 13561f150..4907954d0 100644
--- a/docs/dev/reference/op_greater.html
+++ b/docs/dev/reference/op_greater.html
@@ -38,7 +38,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -121,8 +121,8 @@ <h2 id="value">Value<a class="anchor" aria-label="anchor" href="#value"></a></h2
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
     <div class="dont-index">
 <ul><li><p><a href="https://keras.io/api/ops/numpy#greater-function" class="external-link">https://keras.io/api/ops/numpy#greater-function</a></p></li>
-</ul><p>Other numpy ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p>
-<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
+</ul><p>Other numpy ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p>
+<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_layer_normalization.html">op_layer_normalization</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparse_sigmoid.html">op_sparse_sigmoid</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -134,7 +134,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/op_greater_equal.html b/docs/dev/reference/op_greater_equal.html
index 1b1bf9afc..965cb6ee8 100644
--- a/docs/dev/reference/op_greater_equal.html
+++ b/docs/dev/reference/op_greater_equal.html
@@ -38,7 +38,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -121,8 +121,8 @@ <h2 id="value">Value<a class="anchor" aria-label="anchor" href="#value"></a></h2
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
     <div class="dont-index">
 <ul><li><p><a href="https://keras.io/api/ops/numpy#greaterequal-function" class="external-link">https://keras.io/api/ops/numpy#greaterequal-function</a></p></li>
-</ul><p>Other numpy ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p>
-<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
+</ul><p>Other numpy ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p>
+<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_layer_normalization.html">op_layer_normalization</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparse_sigmoid.html">op_sparse_sigmoid</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -134,7 +134,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/op_hamming.html b/docs/dev/reference/op_hamming.html
new file mode 100644
index 000000000..a7bb6bd9d
--- /dev/null
+++ b/docs/dev/reference/op_hamming.html
@@ -0,0 +1,111 @@
+<!DOCTYPE html>
+<!-- Generated by pkgdown: do not edit by hand --><html lang="en"><head><meta http-equiv="Content-Type" content="text/html; charset=UTF-8"><meta charset="utf-8"><meta http-equiv="X-UA-Compatible" content="IE=edge"><meta name="viewport" content="width=device-width, initial-scale=1, shrink-to-fit=no"><title>Hamming window function. — op_hamming • keras3</title><!-- favicons --><link rel="icon" type="image/png" sizes="96x96" href="../favicon-96x96.png"><link rel="icon" type="”image/svg+xml”" href="../favicon.svg"><link rel="apple-touch-icon" sizes="180x180" href="../apple-touch-icon.png"><link rel="icon" sizes="any" href="../favicon.ico"><link rel="manifest" href="../site.webmanifest"><script src="../deps/jquery-3.6.0/jquery-3.6.0.min.js"></script><meta name="viewport" content="width=device-width, initial-scale=1, shrink-to-fit=no"><link href="../deps/bootstrap-5.3.1/bootstrap.min.css" rel="stylesheet"><script src="../deps/bootstrap-5.3.1/bootstrap.bundle.min.js"></script><link href="../deps/font-awesome-6.5.2/css/all.min.css" rel="stylesheet"><link href="../deps/font-awesome-6.5.2/css/v4-shims.min.css" rel="stylesheet"><script src="../deps/headroom-0.11.0/headroom.min.js"></script><script src="../deps/headroom-0.11.0/jQuery.headroom.min.js"></script><script src="../deps/bootstrap-toc-1.0.1/bootstrap-toc.min.js"></script><script src="../deps/clipboard.js-2.0.11/clipboard.min.js"></script><script src="../deps/search-1.0.0/autocomplete.jquery.min.js"></script><script src="../deps/search-1.0.0/fuse.min.js"></script><script src="../deps/search-1.0.0/mark.min.js"></script><!-- pkgdown --><script src="../pkgdown.js"></script><link href="../extra.css" rel="stylesheet"><meta property="og:title" content="Hamming window function. — op_hamming"><meta name="description" content="The Hamming window is defined as:
+w[n] = 0.54 - 0.46 * cos(2 * pi * n / (N - 1)) for 0 &amp;lt;= n &amp;lt;= N - 1."><meta property="og:description" content="The Hamming window is defined as:
+w[n] = 0.54 - 0.46 * cos(2 * pi * n / (N - 1)) for 0 &amp;lt;= n &amp;lt;= N - 1."><meta name="robots" content="noindex"><!-- Google Tag Manager --><script>(function(w,d,s,l,i){w[l]=w[l]||[];w[l].push({'gtm.start': new Date().getTime(),event:'gtm.js'});var f=d.getElementsByTagName(s)[0],j=d.createElement(s),dl=l!='dataLayer'?'&l='+l:'';j.async=true;j.src='https://www.googletagmanager.com/gtm.js?id='+i+dl;f.parentNode.insertBefore(j,f);})(window,document,'script','dataLayer','GTM-KHBDBW7');</script><!-- End Google Tag Manager --></head><body>
+    <a href="#main" class="visually-hidden-focusable">Skip to contents</a>
+    <!-- Google Tag Manager (noscript) --> <noscript><iframe src="https://www.googletagmanager.com/ns.html?id=GTM-KHBDBW7" height="0" width="0" style="display:none;visibility:hidden"></iframe></noscript> <!-- End Google Tag Manager (noscript) -->
+
+
+    <nav class="navbar navbar-expand-lg fixed-top bg-primary" data-bs-theme="inverse" aria-label="Site navigation"><div class="container">
+
+    <a class="navbar-brand me-2" href="../index.html">keras3</a>
+
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
+
+
+    <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
+      <span class="navbar-toggler-icon"></span>
+    </button>
+
+    <div id="navbar" class="collapse navbar-collapse ms-3">
+      <ul class="navbar-nav me-auto"><li class="nav-item"><a class="nav-link" href="../articles/getting_started.html">Getting Started</a></li>
+<li class="nav-item dropdown">
+  <button class="nav-link dropdown-toggle" type="button" id="dropdown-guides" data-bs-toggle="dropdown" aria-expanded="false" aria-haspopup="true">Guides</button>
+  <ul class="dropdown-menu" aria-labelledby="dropdown-guides"><li><h6 class="dropdown-header" data-toc-skip>Model definition</h6></li>
+    <li><a class="dropdown-item" href="../articles/sequential_model.html">Sequential Model</a></li>
+    <li><a class="dropdown-item" href="../articles/functional_api.html">Functional API</a></li>
+    <li><h6 class="dropdown-header" data-toc-skip>Extending and customizing</h6></li>
+    <li><a class="dropdown-item" href="../articles/training_with_built_in_methods.html">Training &amp; evaluation with the built-in methods</a></li>
+    <li><a class="dropdown-item" href="../articles/custom_train_step_in_tensorflow.html">Customizing `fit()` with Tensorflow</a></li>
+    <li><a class="dropdown-item" href="../articles/writing_your_own_callbacks.html">Writing your own callbacks</a></li>
+    <li><a class="dropdown-item" href="../articles/making_new_layers_and_models_via_subclassing.html">Making new layers and models via subclassing</a></li>
+    <li><a class="dropdown-item" href="../articles/writing_a_custom_training_loop_in_tensorflow.html">Writing a training loop from scratch in TensorFlow</a></li>
+    <li><a class="dropdown-item" href="../articles/serialization_and_saving.html">Serialization and Saving</a></li>
+    <li><h6 class="dropdown-header" data-toc-skip>Other topics</h6></li>
+    <li><a class="dropdown-item" href="../articles/transfer_learning.html">Transfer learning and fine tuning</a></li>
+    <li><a class="dropdown-item" href="../articles/distributed_training_with_tensorflow.html">Distributed training with TensorFlow</a></li>
+    <li><a class="dropdown-item" href="../articles/distribution.html">Distributed training with Jax</a></li>
+  </ul></li>
+<li class="nav-item"><a class="nav-link" href="../articles/examples/index.html">Examples</a></li>
+<li class="active nav-item"><a class="nav-link" href="../reference/index.html">Reference</a></li>
+<li class="nav-item"><a class="nav-link" href="../news/index.html">News</a></li>
+      </ul><ul class="navbar-nav"><li class="nav-item"><form class="form-inline" role="search">
+ <input class="form-control" type="search" name="search-input" id="search-input" autocomplete="off" aria-label="Search site" placeholder="Search for" data-search-index="../search.json"></form></li>
+<li class="nav-item"><a class="external-link nav-link" href="https://github.com/rstudio/keras3/" aria-label="GitHub"><span class="fa fab fa-github fa-lg"></span></a></li>
+      </ul></div>
+
+
+  </div>
+</nav><div class="container template-reference-topic">
+<div class="row">
+  <main id="main" class="col-md-9"><div class="page-header">
+
+      <h1>Hamming window function.</h1>
+      <small class="dont-index">Source: <a href="https://github.com/rstudio/keras3/blob/HEAD/R/ops-numpy.R" class="external-link"><code>R/ops-numpy.R</code></a></small>
+      <div class="d-none name"><code>op_hamming.Rd</code></div>
+    </div>
+
+    <div class="ref-description section level2">
+    <p>The Hamming window is defined as:
+<code>w[n] = 0.54 - 0.46 * cos(2 * pi * n / (N - 1))</code> for <code>0 &lt;= n &lt;= N - 1</code>.</p>
+    </div>
+
+    <div class="section level2">
+    <h2 id="ref-usage">Usage<a class="anchor" aria-label="anchor" href="#ref-usage"></a></h2>
+    <div class="sourceCode"><pre class="sourceCode r"><code><span><span class="fu">op_hamming</span><span class="op">(</span><span class="va">x</span><span class="op">)</span></span></code></pre></div>
+    </div>
+
+    <div class="section level2">
+    <h2 id="arguments">Arguments<a class="anchor" aria-label="anchor" href="#arguments"></a></h2>
+
+
+<dl><dt id="arg-x">x<a class="anchor" aria-label="anchor" href="#arg-x"></a></dt>
+<dd><p>Length of the window. Must be a positive integer.</p></dd>
+
+</dl></div>
+    <div class="section level2">
+    <h2 id="value">Value<a class="anchor" aria-label="anchor" href="#value"></a></h2>
+    <p>A 1D tensor containing the window values.</p>
+    </div>
+    <div class="section level2">
+    <h2 id="examples">Examples<a class="anchor" aria-label="anchor" href="#examples"></a></h2>
+    <p></p><div class="sourceCode r"><pre><code><span><span class="fu"><a href="../reference/op_hamming.html">op_hamming</a></span><span class="op">(</span><span class="fl">5</span><span class="op">)</span></span></code></pre><p></p></div>
+<p></p><div class="sourceCode"><pre><code><span id="cb1-1"><a href="#cb1-1" tabindex="-1"></a><span class="do">## tf.Tensor([0.08000001 0.54       1.         0.54       0.08000001], shape=(5), dtype=float32)</span></span>
+<span id="cb1-2"><a href="#cb1-2" tabindex="-1"></a></span></code></pre><p></p></div>
+    </div>
+    <div class="section level2">
+    <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
+    <div class="dont-index"><p>Other numpy ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p>
+<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_layer_normalization.html">op_layer_normalization</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparse_sigmoid.html">op_sparse_sigmoid</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
+    </div>
+
+  </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
+    </nav></aside></div>
+
+
+    <footer><div class="pkgdown-footer-left">
+  <p>Developed by Tomasz Kalinowski, JJ Allaire, François Chollet, Posit Software, PBC, Google.</p>
+</div>
+
+<div class="pkgdown-footer-right">
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
+</div>
+
+    </footer></div>
+
+
+
+
+
+  </body></html>
+
diff --git a/docs/dev/reference/op_hanning.html b/docs/dev/reference/op_hanning.html
new file mode 100644
index 000000000..aaa6606aa
--- /dev/null
+++ b/docs/dev/reference/op_hanning.html
@@ -0,0 +1,111 @@
+<!DOCTYPE html>
+<!-- Generated by pkgdown: do not edit by hand --><html lang="en"><head><meta http-equiv="Content-Type" content="text/html; charset=UTF-8"><meta charset="utf-8"><meta http-equiv="X-UA-Compatible" content="IE=edge"><meta name="viewport" content="width=device-width, initial-scale=1, shrink-to-fit=no"><title>Hanning window function. — op_hanning • keras3</title><!-- favicons --><link rel="icon" type="image/png" sizes="96x96" href="../favicon-96x96.png"><link rel="icon" type="”image/svg+xml”" href="../favicon.svg"><link rel="apple-touch-icon" sizes="180x180" href="../apple-touch-icon.png"><link rel="icon" sizes="any" href="../favicon.ico"><link rel="manifest" href="../site.webmanifest"><script src="../deps/jquery-3.6.0/jquery-3.6.0.min.js"></script><meta name="viewport" content="width=device-width, initial-scale=1, shrink-to-fit=no"><link href="../deps/bootstrap-5.3.1/bootstrap.min.css" rel="stylesheet"><script src="../deps/bootstrap-5.3.1/bootstrap.bundle.min.js"></script><link href="../deps/font-awesome-6.5.2/css/all.min.css" rel="stylesheet"><link href="../deps/font-awesome-6.5.2/css/v4-shims.min.css" rel="stylesheet"><script src="../deps/headroom-0.11.0/headroom.min.js"></script><script src="../deps/headroom-0.11.0/jQuery.headroom.min.js"></script><script src="../deps/bootstrap-toc-1.0.1/bootstrap-toc.min.js"></script><script src="../deps/clipboard.js-2.0.11/clipboard.min.js"></script><script src="../deps/search-1.0.0/autocomplete.jquery.min.js"></script><script src="../deps/search-1.0.0/fuse.min.js"></script><script src="../deps/search-1.0.0/mark.min.js"></script><!-- pkgdown --><script src="../pkgdown.js"></script><link href="../extra.css" rel="stylesheet"><meta property="og:title" content="Hanning window function. — op_hanning"><meta name="description" content="The Hanning window is defined as:
+w[n] = 0.5 - 0.5 * cos(2 * pi * n / (N - 1)) for 0 &amp;lt;= n &amp;lt;= N - 1."><meta property="og:description" content="The Hanning window is defined as:
+w[n] = 0.5 - 0.5 * cos(2 * pi * n / (N - 1)) for 0 &amp;lt;= n &amp;lt;= N - 1."><meta name="robots" content="noindex"><!-- Google Tag Manager --><script>(function(w,d,s,l,i){w[l]=w[l]||[];w[l].push({'gtm.start': new Date().getTime(),event:'gtm.js'});var f=d.getElementsByTagName(s)[0],j=d.createElement(s),dl=l!='dataLayer'?'&l='+l:'';j.async=true;j.src='https://www.googletagmanager.com/gtm.js?id='+i+dl;f.parentNode.insertBefore(j,f);})(window,document,'script','dataLayer','GTM-KHBDBW7');</script><!-- End Google Tag Manager --></head><body>
+    <a href="#main" class="visually-hidden-focusable">Skip to contents</a>
+    <!-- Google Tag Manager (noscript) --> <noscript><iframe src="https://www.googletagmanager.com/ns.html?id=GTM-KHBDBW7" height="0" width="0" style="display:none;visibility:hidden"></iframe></noscript> <!-- End Google Tag Manager (noscript) -->
+
+
+    <nav class="navbar navbar-expand-lg fixed-top bg-primary" data-bs-theme="inverse" aria-label="Site navigation"><div class="container">
+
+    <a class="navbar-brand me-2" href="../index.html">keras3</a>
+
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
+
+
+    <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
+      <span class="navbar-toggler-icon"></span>
+    </button>
+
+    <div id="navbar" class="collapse navbar-collapse ms-3">
+      <ul class="navbar-nav me-auto"><li class="nav-item"><a class="nav-link" href="../articles/getting_started.html">Getting Started</a></li>
+<li class="nav-item dropdown">
+  <button class="nav-link dropdown-toggle" type="button" id="dropdown-guides" data-bs-toggle="dropdown" aria-expanded="false" aria-haspopup="true">Guides</button>
+  <ul class="dropdown-menu" aria-labelledby="dropdown-guides"><li><h6 class="dropdown-header" data-toc-skip>Model definition</h6></li>
+    <li><a class="dropdown-item" href="../articles/sequential_model.html">Sequential Model</a></li>
+    <li><a class="dropdown-item" href="../articles/functional_api.html">Functional API</a></li>
+    <li><h6 class="dropdown-header" data-toc-skip>Extending and customizing</h6></li>
+    <li><a class="dropdown-item" href="../articles/training_with_built_in_methods.html">Training &amp; evaluation with the built-in methods</a></li>
+    <li><a class="dropdown-item" href="../articles/custom_train_step_in_tensorflow.html">Customizing `fit()` with Tensorflow</a></li>
+    <li><a class="dropdown-item" href="../articles/writing_your_own_callbacks.html">Writing your own callbacks</a></li>
+    <li><a class="dropdown-item" href="../articles/making_new_layers_and_models_via_subclassing.html">Making new layers and models via subclassing</a></li>
+    <li><a class="dropdown-item" href="../articles/writing_a_custom_training_loop_in_tensorflow.html">Writing a training loop from scratch in TensorFlow</a></li>
+    <li><a class="dropdown-item" href="../articles/serialization_and_saving.html">Serialization and Saving</a></li>
+    <li><h6 class="dropdown-header" data-toc-skip>Other topics</h6></li>
+    <li><a class="dropdown-item" href="../articles/transfer_learning.html">Transfer learning and fine tuning</a></li>
+    <li><a class="dropdown-item" href="../articles/distributed_training_with_tensorflow.html">Distributed training with TensorFlow</a></li>
+    <li><a class="dropdown-item" href="../articles/distribution.html">Distributed training with Jax</a></li>
+  </ul></li>
+<li class="nav-item"><a class="nav-link" href="../articles/examples/index.html">Examples</a></li>
+<li class="active nav-item"><a class="nav-link" href="../reference/index.html">Reference</a></li>
+<li class="nav-item"><a class="nav-link" href="../news/index.html">News</a></li>
+      </ul><ul class="navbar-nav"><li class="nav-item"><form class="form-inline" role="search">
+ <input class="form-control" type="search" name="search-input" id="search-input" autocomplete="off" aria-label="Search site" placeholder="Search for" data-search-index="../search.json"></form></li>
+<li class="nav-item"><a class="external-link nav-link" href="https://github.com/rstudio/keras3/" aria-label="GitHub"><span class="fa fab fa-github fa-lg"></span></a></li>
+      </ul></div>
+
+
+  </div>
+</nav><div class="container template-reference-topic">
+<div class="row">
+  <main id="main" class="col-md-9"><div class="page-header">
+
+      <h1>Hanning window function.</h1>
+      <small class="dont-index">Source: <a href="https://github.com/rstudio/keras3/blob/HEAD/R/ops-numpy.R" class="external-link"><code>R/ops-numpy.R</code></a></small>
+      <div class="d-none name"><code>op_hanning.Rd</code></div>
+    </div>
+
+    <div class="ref-description section level2">
+    <p>The Hanning window is defined as:
+<code>w[n] = 0.5 - 0.5 * cos(2 * pi * n / (N - 1))</code> for <code>0 &lt;= n &lt;= N - 1</code>.</p>
+    </div>
+
+    <div class="section level2">
+    <h2 id="ref-usage">Usage<a class="anchor" aria-label="anchor" href="#ref-usage"></a></h2>
+    <div class="sourceCode"><pre class="sourceCode r"><code><span><span class="fu">op_hanning</span><span class="op">(</span><span class="va">x</span><span class="op">)</span></span></code></pre></div>
+    </div>
+
+    <div class="section level2">
+    <h2 id="arguments">Arguments<a class="anchor" aria-label="anchor" href="#arguments"></a></h2>
+
+
+<dl><dt id="arg-x">x<a class="anchor" aria-label="anchor" href="#arg-x"></a></dt>
+<dd><p>Length of the window. Must be a positive integer.</p></dd>
+
+</dl></div>
+    <div class="section level2">
+    <h2 id="value">Value<a class="anchor" aria-label="anchor" href="#value"></a></h2>
+    <p>A 1D tensor containing the window values.</p>
+    </div>
+    <div class="section level2">
+    <h2 id="examples">Examples<a class="anchor" aria-label="anchor" href="#examples"></a></h2>
+    <p></p><div class="sourceCode r"><pre><code><span><span class="fu"><a href="../reference/op_hanning.html">op_hanning</a></span><span class="op">(</span><span class="fl">5</span><span class="op">)</span></span></code></pre><p></p></div>
+<p></p><div class="sourceCode"><pre><code><span id="cb1-1"><a href="#cb1-1" tabindex="-1"></a><span class="do">## tf.Tensor([0.  0.5 1.  0.5 0. ], shape=(5), dtype=float32)</span></span>
+<span id="cb1-2"><a href="#cb1-2" tabindex="-1"></a></span></code></pre><p></p></div>
+    </div>
+    <div class="section level2">
+    <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
+    <div class="dont-index"><p>Other numpy ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p>
+<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_layer_normalization.html">op_layer_normalization</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparse_sigmoid.html">op_sparse_sigmoid</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
+    </div>
+
+  </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
+    </nav></aside></div>
+
+
+    <footer><div class="pkgdown-footer-left">
+  <p>Developed by Tomasz Kalinowski, JJ Allaire, François Chollet, Posit Software, PBC, Google.</p>
+</div>
+
+<div class="pkgdown-footer-right">
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
+</div>
+
+    </footer></div>
+
+
+
+
+
+  </body></html>
+
diff --git a/docs/dev/reference/op_hard_shrink.html b/docs/dev/reference/op_hard_shrink.html
index 577958c65..bce1beba5 100644
--- a/docs/dev/reference/op_hard_shrink.html
+++ b/docs/dev/reference/op_hard_shrink.html
@@ -12,7 +12,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -93,8 +93,8 @@ <h2 id="examples">Examples<a class="anchor" aria-label="anchor" href="#examples"
     </div>
     <div class="section level2">
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
-    <div class="dont-index"><p>Other nn ops: <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br></p>
-<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
+    <div class="dont-index"><p>Other nn ops: <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_layer_normalization.html">op_layer_normalization</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparse_sigmoid.html">op_sparse_sigmoid</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br></p>
+<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_layer_normalization.html">op_layer_normalization</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparse_sigmoid.html">op_sparse_sigmoid</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -106,7 +106,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/op_hard_sigmoid.html b/docs/dev/reference/op_hard_sigmoid.html
index 71e3ec4bb..01eaf6024 100644
--- a/docs/dev/reference/op_hard_sigmoid.html
+++ b/docs/dev/reference/op_hard_sigmoid.html
@@ -10,7 +10,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -92,8 +92,8 @@ <h2 id="examples">Examples<a class="anchor" aria-label="anchor" href="#examples"
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
     <div class="dont-index">
 <ul><li><p><a href="https://keras.io/api/ops/nn#hardsigmoid-function" class="external-link">https://keras.io/api/ops/nn#hardsigmoid-function</a></p></li>
-</ul><p>Other nn ops: <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br></p>
-<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
+</ul><p>Other nn ops: <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_layer_normalization.html">op_layer_normalization</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparse_sigmoid.html">op_sparse_sigmoid</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br></p>
+<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_layer_normalization.html">op_layer_normalization</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparse_sigmoid.html">op_sparse_sigmoid</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -105,7 +105,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/op_hard_silu.html b/docs/dev/reference/op_hard_silu.html
index ff8127383..6e76184de 100644
--- a/docs/dev/reference/op_hard_silu.html
+++ b/docs/dev/reference/op_hard_silu.html
@@ -20,7 +20,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -100,8 +100,8 @@ <h2 id="examples">Examples<a class="anchor" aria-label="anchor" href="#examples"
     </div>
     <div class="section level2">
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
-    <div class="dont-index"><p>Other nn ops: <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br></p>
-<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
+    <div class="dont-index"><p>Other nn ops: <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_layer_normalization.html">op_layer_normalization</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparse_sigmoid.html">op_sparse_sigmoid</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br></p>
+<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_layer_normalization.html">op_layer_normalization</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparse_sigmoid.html">op_sparse_sigmoid</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -113,7 +113,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/op_hard_tanh.html b/docs/dev/reference/op_hard_tanh.html
index 2c70a68cd..4fb224f2f 100644
--- a/docs/dev/reference/op_hard_tanh.html
+++ b/docs/dev/reference/op_hard_tanh.html
@@ -10,7 +10,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -87,8 +87,8 @@ <h2 id="examples">Examples<a class="anchor" aria-label="anchor" href="#examples"
     </div>
     <div class="section level2">
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
-    <div class="dont-index"><p>Other nn ops: <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br></p>
-<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
+    <div class="dont-index"><p>Other nn ops: <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_layer_normalization.html">op_layer_normalization</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparse_sigmoid.html">op_sparse_sigmoid</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br></p>
+<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_layer_normalization.html">op_layer_normalization</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparse_sigmoid.html">op_sparse_sigmoid</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -100,7 +100,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/op_heaviside.html b/docs/dev/reference/op_heaviside.html
new file mode 100644
index 000000000..619b915c1
--- /dev/null
+++ b/docs/dev/reference/op_heaviside.html
@@ -0,0 +1,122 @@
+<!DOCTYPE html>
+<!-- Generated by pkgdown: do not edit by hand --><html lang="en"><head><meta http-equiv="Content-Type" content="text/html; charset=UTF-8"><meta charset="utf-8"><meta http-equiv="X-UA-Compatible" content="IE=edge"><meta name="viewport" content="width=device-width, initial-scale=1, shrink-to-fit=no"><title>Heaviside step function. — op_heaviside • keras3</title><!-- favicons --><link rel="icon" type="image/png" sizes="96x96" href="../favicon-96x96.png"><link rel="icon" type="”image/svg+xml”" href="../favicon.svg"><link rel="apple-touch-icon" sizes="180x180" href="../apple-touch-icon.png"><link rel="icon" sizes="any" href="../favicon.ico"><link rel="manifest" href="../site.webmanifest"><script src="../deps/jquery-3.6.0/jquery-3.6.0.min.js"></script><meta name="viewport" content="width=device-width, initial-scale=1, shrink-to-fit=no"><link href="../deps/bootstrap-5.3.1/bootstrap.min.css" rel="stylesheet"><script src="../deps/bootstrap-5.3.1/bootstrap.bundle.min.js"></script><link href="../deps/font-awesome-6.5.2/css/all.min.css" rel="stylesheet"><link href="../deps/font-awesome-6.5.2/css/v4-shims.min.css" rel="stylesheet"><script src="../deps/headroom-0.11.0/headroom.min.js"></script><script src="../deps/headroom-0.11.0/jQuery.headroom.min.js"></script><script src="../deps/bootstrap-toc-1.0.1/bootstrap-toc.min.js"></script><script src="../deps/clipboard.js-2.0.11/clipboard.min.js"></script><script src="../deps/search-1.0.0/autocomplete.jquery.min.js"></script><script src="../deps/search-1.0.0/fuse.min.js"></script><script src="../deps/search-1.0.0/mark.min.js"></script><!-- pkgdown --><script src="../pkgdown.js"></script><link href="../extra.css" rel="stylesheet"><meta property="og:title" content="Heaviside step function. — op_heaviside"><meta name="description" content="The Heaviside step function is defined as:
+heaviside(x1, x2) = 0 if x1 &amp;lt; 0,
+heaviside(x1, x2) = 1 if x1 &amp;gt; 0, and
+heaviside(x1, x2) = x2 if x1 == 0."><meta property="og:description" content="The Heaviside step function is defined as:
+heaviside(x1, x2) = 0 if x1 &amp;lt; 0,
+heaviside(x1, x2) = 1 if x1 &amp;gt; 0, and
+heaviside(x1, x2) = x2 if x1 == 0."><meta name="robots" content="noindex"><!-- Google Tag Manager --><script>(function(w,d,s,l,i){w[l]=w[l]||[];w[l].push({'gtm.start': new Date().getTime(),event:'gtm.js'});var f=d.getElementsByTagName(s)[0],j=d.createElement(s),dl=l!='dataLayer'?'&l='+l:'';j.async=true;j.src='https://www.googletagmanager.com/gtm.js?id='+i+dl;f.parentNode.insertBefore(j,f);})(window,document,'script','dataLayer','GTM-KHBDBW7');</script><!-- End Google Tag Manager --></head><body>
+    <a href="#main" class="visually-hidden-focusable">Skip to contents</a>
+    <!-- Google Tag Manager (noscript) --> <noscript><iframe src="https://www.googletagmanager.com/ns.html?id=GTM-KHBDBW7" height="0" width="0" style="display:none;visibility:hidden"></iframe></noscript> <!-- End Google Tag Manager (noscript) -->
+
+
+    <nav class="navbar navbar-expand-lg fixed-top bg-primary" data-bs-theme="inverse" aria-label="Site navigation"><div class="container">
+
+    <a class="navbar-brand me-2" href="../index.html">keras3</a>
+
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
+
+
+    <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
+      <span class="navbar-toggler-icon"></span>
+    </button>
+
+    <div id="navbar" class="collapse navbar-collapse ms-3">
+      <ul class="navbar-nav me-auto"><li class="nav-item"><a class="nav-link" href="../articles/getting_started.html">Getting Started</a></li>
+<li class="nav-item dropdown">
+  <button class="nav-link dropdown-toggle" type="button" id="dropdown-guides" data-bs-toggle="dropdown" aria-expanded="false" aria-haspopup="true">Guides</button>
+  <ul class="dropdown-menu" aria-labelledby="dropdown-guides"><li><h6 class="dropdown-header" data-toc-skip>Model definition</h6></li>
+    <li><a class="dropdown-item" href="../articles/sequential_model.html">Sequential Model</a></li>
+    <li><a class="dropdown-item" href="../articles/functional_api.html">Functional API</a></li>
+    <li><h6 class="dropdown-header" data-toc-skip>Extending and customizing</h6></li>
+    <li><a class="dropdown-item" href="../articles/training_with_built_in_methods.html">Training &amp; evaluation with the built-in methods</a></li>
+    <li><a class="dropdown-item" href="../articles/custom_train_step_in_tensorflow.html">Customizing `fit()` with Tensorflow</a></li>
+    <li><a class="dropdown-item" href="../articles/writing_your_own_callbacks.html">Writing your own callbacks</a></li>
+    <li><a class="dropdown-item" href="../articles/making_new_layers_and_models_via_subclassing.html">Making new layers and models via subclassing</a></li>
+    <li><a class="dropdown-item" href="../articles/writing_a_custom_training_loop_in_tensorflow.html">Writing a training loop from scratch in TensorFlow</a></li>
+    <li><a class="dropdown-item" href="../articles/serialization_and_saving.html">Serialization and Saving</a></li>
+    <li><h6 class="dropdown-header" data-toc-skip>Other topics</h6></li>
+    <li><a class="dropdown-item" href="../articles/transfer_learning.html">Transfer learning and fine tuning</a></li>
+    <li><a class="dropdown-item" href="../articles/distributed_training_with_tensorflow.html">Distributed training with TensorFlow</a></li>
+    <li><a class="dropdown-item" href="../articles/distribution.html">Distributed training with Jax</a></li>
+  </ul></li>
+<li class="nav-item"><a class="nav-link" href="../articles/examples/index.html">Examples</a></li>
+<li class="active nav-item"><a class="nav-link" href="../reference/index.html">Reference</a></li>
+<li class="nav-item"><a class="nav-link" href="../news/index.html">News</a></li>
+      </ul><ul class="navbar-nav"><li class="nav-item"><form class="form-inline" role="search">
+ <input class="form-control" type="search" name="search-input" id="search-input" autocomplete="off" aria-label="Search site" placeholder="Search for" data-search-index="../search.json"></form></li>
+<li class="nav-item"><a class="external-link nav-link" href="https://github.com/rstudio/keras3/" aria-label="GitHub"><span class="fa fab fa-github fa-lg"></span></a></li>
+      </ul></div>
+
+
+  </div>
+</nav><div class="container template-reference-topic">
+<div class="row">
+  <main id="main" class="col-md-9"><div class="page-header">
+
+      <h1>Heaviside step function.</h1>
+      <small class="dont-index">Source: <a href="https://github.com/rstudio/keras3/blob/HEAD/R/ops-numpy.R" class="external-link"><code>R/ops-numpy.R</code></a></small>
+      <div class="d-none name"><code>op_heaviside.Rd</code></div>
+    </div>
+
+    <div class="ref-description section level2">
+    <p>The Heaviside step function is defined as:
+<code>heaviside(x1, x2) = 0</code> if <code>x1 &lt; 0</code>,
+<code>heaviside(x1, x2) = 1</code> if <code>x1 &gt; 0</code>, and
+<code>heaviside(x1, x2) = x2</code> if <code>x1 == 0</code>.</p>
+    </div>
+
+    <div class="section level2">
+    <h2 id="ref-usage">Usage<a class="anchor" aria-label="anchor" href="#ref-usage"></a></h2>
+    <div class="sourceCode"><pre class="sourceCode r"><code><span><span class="fu">op_heaviside</span><span class="op">(</span><span class="va">x1</span>, <span class="va">x2</span><span class="op">)</span></span></code></pre></div>
+    </div>
+
+    <div class="section level2">
+    <h2 id="arguments">Arguments<a class="anchor" aria-label="anchor" href="#arguments"></a></h2>
+
+
+<dl><dt id="arg-x-">x1<a class="anchor" aria-label="anchor" href="#arg-x-"></a></dt>
+<dd><p>Tensor input.</p></dd>
+
+
+<dt id="arg-x-">x2<a class="anchor" aria-label="anchor" href="#arg-x-"></a></dt>
+<dd><p>Value to use when <code>x1 == 0</code>.</p></dd>
+
+</dl></div>
+    <div class="section level2">
+    <h2 id="value">Value<a class="anchor" aria-label="anchor" href="#value"></a></h2>
+    <p>A tensor broadcast from <code>x1</code> and <code>x2</code> containing <code>0</code>, <code>1</code>, or <code>x2</code>.</p>
+    </div>
+    <div class="section level2">
+    <h2 id="examples">Examples<a class="anchor" aria-label="anchor" href="#examples"></a></h2>
+    <p></p><div class="sourceCode r"><pre><code><span><span class="va">x1</span> <span class="op">&lt;-</span> <span class="fu"><a href="../reference/op_array.html">op_array</a></span><span class="op">(</span><span class="fu"><a href="https://rdrr.io/r/base/c.html" class="external-link">c</a></span><span class="op">(</span><span class="op">-</span><span class="fl">2</span>, <span class="fl">0</span>, <span class="fl">3</span><span class="op">)</span><span class="op">)</span></span>
+<span><span class="fu"><a href="../reference/op_heaviside.html">op_heaviside</a></span><span class="op">(</span><span class="va">x1</span>, <span class="fl">0.5</span><span class="op">)</span></span></code></pre><p></p></div>
+<p></p><div class="sourceCode"><pre><code><span id="cb1-1"><a href="#cb1-1" tabindex="-1"></a><span class="do">## tf.Tensor([0.  0.5 1. ], shape=(3), dtype=float32)</span></span>
+<span id="cb1-2"><a href="#cb1-2" tabindex="-1"></a></span></code></pre><p></p></div>
+    </div>
+    <div class="section level2">
+    <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
+    <div class="dont-index"><p>Other numpy ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p>
+<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_layer_normalization.html">op_layer_normalization</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparse_sigmoid.html">op_sparse_sigmoid</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
+    </div>
+
+  </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
+    </nav></aside></div>
+
+
+    <footer><div class="pkgdown-footer-left">
+  <p>Developed by Tomasz Kalinowski, JJ Allaire, François Chollet, Posit Software, PBC, Google.</p>
+</div>
+
+<div class="pkgdown-footer-right">
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
+</div>
+
+    </footer></div>
+
+
+
+
+
+  </body></html>
+
diff --git a/docs/dev/reference/op_histogram.html b/docs/dev/reference/op_histogram.html
index b1ec9a662..d86fb40b3 100644
--- a/docs/dev/reference/op_histogram.html
+++ b/docs/dev/reference/op_histogram.html
@@ -8,7 +8,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -101,8 +101,8 @@ <h2 id="examples">Examples<a class="anchor" aria-label="anchor" href="#examples"
     </div>
     <div class="section level2">
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
-    <div class="dont-index"><p>Other numpy ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p>
-<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
+    <div class="dont-index"><p>Other numpy ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p>
+<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_layer_normalization.html">op_layer_normalization</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparse_sigmoid.html">op_sparse_sigmoid</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -114,7 +114,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/op_hstack.html b/docs/dev/reference/op_hstack.html
index ffc89b0d5..30c778e6e 100644
--- a/docs/dev/reference/op_hstack.html
+++ b/docs/dev/reference/op_hstack.html
@@ -10,7 +10,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -81,8 +81,8 @@ <h2 id="value">Value<a class="anchor" aria-label="anchor" href="#value"></a></h2
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
     <div class="dont-index">
 <ul><li><p><a href="https://keras.io/api/ops/numpy#hstack-function" class="external-link">https://keras.io/api/ops/numpy#hstack-function</a></p></li>
-</ul><p>Other numpy ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p>
-<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
+</ul><p>Other numpy ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p>
+<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_layer_normalization.html">op_layer_normalization</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparse_sigmoid.html">op_sparse_sigmoid</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -94,7 +94,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/op_identity.html b/docs/dev/reference/op_identity.html
index 57d7c5ca6..ccc1293ab 100644
--- a/docs/dev/reference/op_identity.html
+++ b/docs/dev/reference/op_identity.html
@@ -10,7 +10,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -85,8 +85,8 @@ <h2 id="value">Value<a class="anchor" aria-label="anchor" href="#value"></a></h2
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
     <div class="dont-index">
 <ul><li><p><a href="https://keras.io/api/ops/numpy#identity-function" class="external-link">https://keras.io/api/ops/numpy#identity-function</a></p></li>
-</ul><p>Other numpy ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p>
-<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
+</ul><p>Other numpy ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p>
+<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_layer_normalization.html">op_layer_normalization</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparse_sigmoid.html">op_sparse_sigmoid</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -98,7 +98,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/op_ifft2.html b/docs/dev/reference/op_ifft2.html
index f58506863..27f7917e6 100644
--- a/docs/dev/reference/op_ifft2.html
+++ b/docs/dev/reference/op_ifft2.html
@@ -8,7 +8,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -95,7 +95,7 @@ <h2 id="examples">Examples<a class="anchor" aria-label="anchor" href="#examples"
     <div class="section level2">
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
     <div class="dont-index"><p>Other math ops: <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br></p>
-<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
+<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_layer_normalization.html">op_layer_normalization</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparse_sigmoid.html">op_sparse_sigmoid</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -107,7 +107,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/op_imag.html b/docs/dev/reference/op_imag.html
index f33709927..503c08399 100644
--- a/docs/dev/reference/op_imag.html
+++ b/docs/dev/reference/op_imag.html
@@ -8,7 +8,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -78,8 +78,8 @@ <h2 id="value">Value<a class="anchor" aria-label="anchor" href="#value"></a></h2
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
     <div class="dont-index">
 <ul><li><p><a href="https://keras.io/api/ops/numpy#imag-function" class="external-link">https://keras.io/api/ops/numpy#imag-function</a></p></li>
-</ul><p>Other numpy ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p>
-<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
+</ul><p>Other numpy ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p>
+<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_layer_normalization.html">op_layer_normalization</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparse_sigmoid.html">op_sparse_sigmoid</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -91,7 +91,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/op_image_affine_transform.html b/docs/dev/reference/op_image_affine_transform.html
index 90194b017..9d39af9ab 100644
--- a/docs/dev/reference/op_image_affine_transform.html
+++ b/docs/dev/reference/op_image_affine_transform.html
@@ -8,7 +8,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -165,7 +165,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 <ul><li><p><a href="https://keras.io/api/ops/image#affinetransform-function" class="external-link">https://keras.io/api/ops/image#affinetransform-function</a></p></li>
 </ul><p>Other image ops: <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br></p>
 <p>Other image utils: <br><code><a href="image_array_save.html">image_array_save</a>()</code> <br><code><a href="image_from_array.html">image_from_array</a>()</code> <br><code><a href="image_load.html">image_load</a>()</code> <br><code><a href="image_smart_resize.html">image_smart_resize</a>()</code> <br><code><a href="image_to_array.html">image_to_array</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br></p>
-<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
+<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_layer_normalization.html">op_layer_normalization</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparse_sigmoid.html">op_sparse_sigmoid</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -177,7 +177,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/op_image_crop.html b/docs/dev/reference/op_image_crop.html
index ebf06859f..f8e6d03b8 100644
--- a/docs/dev/reference/op_image_crop.html
+++ b/docs/dev/reference/op_image_crop.html
@@ -8,7 +8,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -131,7 +131,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 <ul><li><p><a href="https://www.tensorflow.org/api_docs/python/tf/keras/ops/image/crop_images" class="external-link">https://www.tensorflow.org/api_docs/python/tf/keras/ops/image/crop_images</a></p></li>
 </ul><p>Other image ops: <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br></p>
 <p>Other image utils: <br><code><a href="image_array_save.html">image_array_save</a>()</code> <br><code><a href="image_from_array.html">image_from_array</a>()</code> <br><code><a href="image_load.html">image_load</a>()</code> <br><code><a href="image_smart_resize.html">image_smart_resize</a>()</code> <br><code><a href="image_to_array.html">image_to_array</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br></p>
-<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
+<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_layer_normalization.html">op_layer_normalization</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparse_sigmoid.html">op_sparse_sigmoid</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -143,7 +143,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/op_image_elastic_transform.html b/docs/dev/reference/op_image_elastic_transform.html
new file mode 100644
index 000000000..6121841ba
--- /dev/null
+++ b/docs/dev/reference/op_image_elastic_transform.html
@@ -0,0 +1,178 @@
+<!DOCTYPE html>
+<!-- Generated by pkgdown: do not edit by hand --><html lang="en"><head><meta http-equiv="Content-Type" content="text/html; charset=UTF-8"><meta charset="utf-8"><meta http-equiv="X-UA-Compatible" content="IE=edge"><meta name="viewport" content="width=device-width, initial-scale=1, shrink-to-fit=no"><title>Applies elastic deformation to the image(s). — op_image_elastic_transform • keras3</title><!-- favicons --><link rel="icon" type="image/png" sizes="96x96" href="../favicon-96x96.png"><link rel="icon" type="”image/svg+xml”" href="../favicon.svg"><link rel="apple-touch-icon" sizes="180x180" href="../apple-touch-icon.png"><link rel="icon" sizes="any" href="../favicon.ico"><link rel="manifest" href="../site.webmanifest"><script src="../deps/jquery-3.6.0/jquery-3.6.0.min.js"></script><meta name="viewport" content="width=device-width, initial-scale=1, shrink-to-fit=no"><link href="../deps/bootstrap-5.3.1/bootstrap.min.css" rel="stylesheet"><script src="../deps/bootstrap-5.3.1/bootstrap.bundle.min.js"></script><link href="../deps/font-awesome-6.5.2/css/all.min.css" rel="stylesheet"><link href="../deps/font-awesome-6.5.2/css/v4-shims.min.css" rel="stylesheet"><script src="../deps/headroom-0.11.0/headroom.min.js"></script><script src="../deps/headroom-0.11.0/jQuery.headroom.min.js"></script><script src="../deps/bootstrap-toc-1.0.1/bootstrap-toc.min.js"></script><script src="../deps/clipboard.js-2.0.11/clipboard.min.js"></script><script src="../deps/search-1.0.0/autocomplete.jquery.min.js"></script><script src="../deps/search-1.0.0/fuse.min.js"></script><script src="../deps/search-1.0.0/mark.min.js"></script><!-- pkgdown --><script src="../pkgdown.js"></script><link href="../extra.css" rel="stylesheet"><meta property="og:title" content="Applies elastic deformation to the image(s). — op_image_elastic_transform"><meta name="description" content="Apply random elastic deformation to 3D or 4D image tensors."><meta property="og:description" content="Apply random elastic deformation to 3D or 4D image tensors."><meta name="robots" content="noindex"><!-- Google Tag Manager --><script>(function(w,d,s,l,i){w[l]=w[l]||[];w[l].push({'gtm.start': new Date().getTime(),event:'gtm.js'});var f=d.getElementsByTagName(s)[0],j=d.createElement(s),dl=l!='dataLayer'?'&l='+l:'';j.async=true;j.src='https://www.googletagmanager.com/gtm.js?id='+i+dl;f.parentNode.insertBefore(j,f);})(window,document,'script','dataLayer','GTM-KHBDBW7');</script><!-- End Google Tag Manager --></head><body>
+    <a href="#main" class="visually-hidden-focusable">Skip to contents</a>
+    <!-- Google Tag Manager (noscript) --> <noscript><iframe src="https://www.googletagmanager.com/ns.html?id=GTM-KHBDBW7" height="0" width="0" style="display:none;visibility:hidden"></iframe></noscript> <!-- End Google Tag Manager (noscript) -->
+
+
+    <nav class="navbar navbar-expand-lg fixed-top bg-primary" data-bs-theme="inverse" aria-label="Site navigation"><div class="container">
+
+    <a class="navbar-brand me-2" href="../index.html">keras3</a>
+
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
+
+
+    <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
+      <span class="navbar-toggler-icon"></span>
+    </button>
+
+    <div id="navbar" class="collapse navbar-collapse ms-3">
+      <ul class="navbar-nav me-auto"><li class="nav-item"><a class="nav-link" href="../articles/getting_started.html">Getting Started</a></li>
+<li class="nav-item dropdown">
+  <button class="nav-link dropdown-toggle" type="button" id="dropdown-guides" data-bs-toggle="dropdown" aria-expanded="false" aria-haspopup="true">Guides</button>
+  <ul class="dropdown-menu" aria-labelledby="dropdown-guides"><li><h6 class="dropdown-header" data-toc-skip>Model definition</h6></li>
+    <li><a class="dropdown-item" href="../articles/sequential_model.html">Sequential Model</a></li>
+    <li><a class="dropdown-item" href="../articles/functional_api.html">Functional API</a></li>
+    <li><h6 class="dropdown-header" data-toc-skip>Extending and customizing</h6></li>
+    <li><a class="dropdown-item" href="../articles/training_with_built_in_methods.html">Training &amp; evaluation with the built-in methods</a></li>
+    <li><a class="dropdown-item" href="../articles/custom_train_step_in_tensorflow.html">Customizing `fit()` with Tensorflow</a></li>
+    <li><a class="dropdown-item" href="../articles/writing_your_own_callbacks.html">Writing your own callbacks</a></li>
+    <li><a class="dropdown-item" href="../articles/making_new_layers_and_models_via_subclassing.html">Making new layers and models via subclassing</a></li>
+    <li><a class="dropdown-item" href="../articles/writing_a_custom_training_loop_in_tensorflow.html">Writing a training loop from scratch in TensorFlow</a></li>
+    <li><a class="dropdown-item" href="../articles/serialization_and_saving.html">Serialization and Saving</a></li>
+    <li><h6 class="dropdown-header" data-toc-skip>Other topics</h6></li>
+    <li><a class="dropdown-item" href="../articles/transfer_learning.html">Transfer learning and fine tuning</a></li>
+    <li><a class="dropdown-item" href="../articles/distributed_training_with_tensorflow.html">Distributed training with TensorFlow</a></li>
+    <li><a class="dropdown-item" href="../articles/distribution.html">Distributed training with Jax</a></li>
+  </ul></li>
+<li class="nav-item"><a class="nav-link" href="../articles/examples/index.html">Examples</a></li>
+<li class="active nav-item"><a class="nav-link" href="../reference/index.html">Reference</a></li>
+<li class="nav-item"><a class="nav-link" href="../news/index.html">News</a></li>
+      </ul><ul class="navbar-nav"><li class="nav-item"><form class="form-inline" role="search">
+ <input class="form-control" type="search" name="search-input" id="search-input" autocomplete="off" aria-label="Search site" placeholder="Search for" data-search-index="../search.json"></form></li>
+<li class="nav-item"><a class="external-link nav-link" href="https://github.com/rstudio/keras3/" aria-label="GitHub"><span class="fa fab fa-github fa-lg"></span></a></li>
+      </ul></div>
+
+
+  </div>
+</nav><div class="container template-reference-topic">
+<div class="row">
+  <main id="main" class="col-md-9"><div class="page-header">
+
+      <h1>Applies elastic deformation to the image(s).</h1>
+      <small class="dont-index">Source: <a href="https://github.com/rstudio/keras3/blob/HEAD/R/ops-image.R" class="external-link"><code>R/ops-image.R</code></a></small>
+      <div class="d-none name"><code>op_image_elastic_transform.Rd</code></div>
+    </div>
+
+    <div class="ref-description section level2">
+    <p>Apply random elastic deformation to 3D or 4D image tensors.</p>
+    </div>
+
+    <div class="section level2">
+    <h2 id="ref-usage">Usage<a class="anchor" aria-label="anchor" href="#ref-usage"></a></h2>
+    <div class="sourceCode"><pre class="sourceCode r"><code><span><span class="fu">op_image_elastic_transform</span><span class="op">(</span></span>
+<span>  <span class="va">images</span>,</span>
+<span>  alpha <span class="op">=</span> <span class="fl">20</span>,</span>
+<span>  sigma <span class="op">=</span> <span class="fl">5</span>,</span>
+<span>  interpolation <span class="op">=</span> <span class="st">"bilinear"</span>,</span>
+<span>  fill_mode <span class="op">=</span> <span class="st">"reflect"</span>,</span>
+<span>  fill_value <span class="op">=</span> <span class="fl">0</span>,</span>
+<span>  seed <span class="op">=</span> <span class="cn">NULL</span>,</span>
+<span>  data_format <span class="op">=</span> <span class="cn">NULL</span></span>
+<span><span class="op">)</span></span></code></pre></div>
+    </div>
+
+    <div class="section level2">
+    <h2 id="arguments">Arguments<a class="anchor" aria-label="anchor" href="#arguments"></a></h2>
+
+
+<dl><dt id="arg-images">images<a class="anchor" aria-label="anchor" href="#arg-images"></a></dt>
+<dd><p>Input image or batch of images. Must be 3D or 4D.</p></dd>
+
+
+<dt id="arg-alpha">alpha<a class="anchor" aria-label="anchor" href="#arg-alpha"></a></dt>
+<dd><p>Scaling factor that controls the intensity of the deformation.</p></dd>
+
+
+<dt id="arg-sigma">sigma<a class="anchor" aria-label="anchor" href="#arg-sigma"></a></dt>
+<dd><p>Standard deviation of the Gaussian filter used for
+smoothing the displacement fields.</p></dd>
+
+
+<dt id="arg-interpolation">interpolation<a class="anchor" aria-label="anchor" href="#arg-interpolation"></a></dt>
+<dd><p>Interpolation method. Available methods are <code>"nearest"</code>,
+and <code>"bilinear"</code>. Defaults to <code>"bilinear"</code>.</p></dd>
+
+
+<dt id="arg-fill-mode">fill_mode<a class="anchor" aria-label="anchor" href="#arg-fill-mode"></a></dt>
+<dd><p>Points outside the boundaries of the input are filled
+according to the given mode. Available methods are <code>"constant"</code>,
+<code>"nearest"</code>, <code>"wrap"</code> and <code>"reflect"</code>. Defaults to <code>"reflect"</code>.</p><ul><li><p><code>"reflect"</code>: <code>(d c b a | a b c d | d c b a)</code>
+The input is extended by reflecting about the edge of the last
+pixel.</p></li>
+<li><p><code>"constant"</code>: <code>(k k k k | a b c d | k k k k)</code>
+The input is extended by filling all values beyond
+the edge with the same constant value <code>k</code> specified by
+<code>fill_value</code>.</p></li>
+<li><p><code>"wrap"</code>: <code>(a b c d | a b c d | a b c d)</code>
+The input is extended by wrapping around to the opposite edge.</p></li>
+<li><p><code>"nearest"</code>: <code>(a a a a | a b c d | d d d d)</code>
+The input is extended by the nearest pixel.</p></li>
+</ul></dd>
+
+
+<dt id="arg-fill-value">fill_value<a class="anchor" aria-label="anchor" href="#arg-fill-value"></a></dt>
+<dd><p>Value used for points outside the boundaries of the input if
+<code>fill_mode="constant"</code>. Defaults to <code>0</code>.</p></dd>
+
+
+<dt id="arg-seed">seed<a class="anchor" aria-label="anchor" href="#arg-seed"></a></dt>
+<dd><p>Optional integer seed for the random number generator.</p></dd>
+
+
+<dt id="arg-data-format">data_format<a class="anchor" aria-label="anchor" href="#arg-data-format"></a></dt>
+<dd><p>A string specifying the data format of the input tensor.
+It can be either <code>"channels_last"</code> or <code>"channels_first"</code>.
+<code>"channels_last"</code> corresponds to inputs with shape
+<code>(batch, height, width, channels)</code>, while <code>"channels_first"</code>
+corresponds to inputs with shape <code>(batch, channels, height, width)</code>.
+If not specified, the value will default to
+<code>keras.config.image_data_format</code>.</p></dd>
+
+</dl></div>
+    <div class="section level2">
+    <h2 id="value">Value<a class="anchor" aria-label="anchor" href="#value"></a></h2>
+    <p>Transformed image or batch of images with elastic deformation.</p>
+    </div>
+    <div class="section level2">
+    <h2 id="examples">Examples<a class="anchor" aria-label="anchor" href="#examples"></a></h2>
+    <p></p><div class="sourceCode r"><pre><code><span><span class="va">x</span> <span class="op">&lt;-</span> <span class="fu"><a href="../reference/random_uniform.html">random_uniform</a></span><span class="op">(</span><span class="fu"><a href="https://rdrr.io/r/base/c.html" class="external-link">c</a></span><span class="op">(</span><span class="fl">2</span>, <span class="fl">64</span>, <span class="fl">80</span>, <span class="fl">3</span><span class="op">)</span><span class="op">)</span>  <span class="co"># batch of 2 RGB images</span></span>
+<span><span class="va">y</span> <span class="op">&lt;-</span> <span class="fu"><a href="../reference/op_image_elastic_transform.html">op_image_elastic_transform</a></span><span class="op">(</span><span class="va">x</span><span class="op">)</span></span>
+<span><span class="fu"><a href="../reference/op_shape.html">op_shape</a></span><span class="op">(</span><span class="va">y</span><span class="op">)</span></span></code></pre><p></p></div>
+<p></p><div class="sourceCode"><pre><code><span id="cb1-1"><a href="#cb1-1" tabindex="-1"></a><span class="do">## shape(2, 64, 80, 3)</span></span>
+<span id="cb1-2"><a href="#cb1-2" tabindex="-1"></a></span></code></pre><p></p></div>
+<p></p><div class="sourceCode r"><pre><code><span><span class="va">x</span> <span class="op">&lt;-</span> <span class="fu"><a href="../reference/random_uniform.html">random_uniform</a></span><span class="op">(</span><span class="fu"><a href="https://rdrr.io/r/base/c.html" class="external-link">c</a></span><span class="op">(</span><span class="fl">64</span>, <span class="fl">80</span>, <span class="fl">3</span><span class="op">)</span><span class="op">)</span>  <span class="co"># single RGB image</span></span>
+<span><span class="va">y</span> <span class="op">&lt;-</span> <span class="fu"><a href="../reference/op_image_elastic_transform.html">op_image_elastic_transform</a></span><span class="op">(</span><span class="va">x</span><span class="op">)</span></span>
+<span><span class="fu"><a href="../reference/op_shape.html">op_shape</a></span><span class="op">(</span><span class="va">y</span><span class="op">)</span></span></code></pre><p></p></div>
+<p></p><div class="sourceCode"><pre><code><span id="cb1-1"><a href="#cb1-1" tabindex="-1"></a><span class="do">## shape(64, 80, 3)</span></span>
+<span id="cb1-2"><a href="#cb1-2" tabindex="-1"></a></span></code></pre><p></p></div>
+<p></p><div class="sourceCode r"><pre><code><span><span class="va">x</span> <span class="op">&lt;-</span> <span class="fu"><a href="../reference/random_uniform.html">random_uniform</a></span><span class="op">(</span><span class="fu"><a href="https://rdrr.io/r/base/c.html" class="external-link">c</a></span><span class="op">(</span><span class="fl">2</span>, <span class="fl">3</span>, <span class="fl">64</span>, <span class="fl">80</span><span class="op">)</span><span class="op">)</span>  <span class="co"># batch of 2 RGB images</span></span>
+<span><span class="va">y</span> <span class="op">&lt;-</span> <span class="fu"><a href="../reference/op_image_elastic_transform.html">op_image_elastic_transform</a></span><span class="op">(</span></span>
+<span>  <span class="va">x</span>,</span>
+<span>  data_format <span class="op">=</span> <span class="st">"channels_first"</span>,</span>
+<span>  seed <span class="op">=</span> <span class="fl">123</span></span>
+<span><span class="op">)</span></span>
+<span><span class="fu"><a href="../reference/op_shape.html">op_shape</a></span><span class="op">(</span><span class="va">y</span><span class="op">)</span></span></code></pre><p></p></div>
+<p></p><div class="sourceCode"><pre><code><span id="cb1-1"><a href="#cb1-1" tabindex="-1"></a><span class="do">## shape(2, 3, 64, 80)</span></span>
+<span id="cb1-2"><a href="#cb1-2" tabindex="-1"></a></span></code></pre><p></p></div>
+    </div>
+
+  </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
+    </nav></aside></div>
+
+
+    <footer><div class="pkgdown-footer-left">
+  <p>Developed by Tomasz Kalinowski, JJ Allaire, François Chollet, Posit Software, PBC, Google.</p>
+</div>
+
+<div class="pkgdown-footer-right">
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
+</div>
+
+    </footer></div>
+
+
+
+
+
+  </body></html>
+
diff --git a/docs/dev/reference/op_image_extract_patches.html b/docs/dev/reference/op_image_extract_patches.html
index 65b8c0d36..0def8a219 100644
--- a/docs/dev/reference/op_image_extract_patches.html
+++ b/docs/dev/reference/op_image_extract_patches.html
@@ -8,7 +8,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -132,7 +132,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 <ul><li><p><a href="https://keras.io/api/ops/image#extractpatches-function" class="external-link">https://keras.io/api/ops/image#extractpatches-function</a></p></li>
 </ul><p>Other image ops: <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br></p>
 <p>Other image utils: <br><code><a href="image_array_save.html">image_array_save</a>()</code> <br><code><a href="image_from_array.html">image_from_array</a>()</code> <br><code><a href="image_load.html">image_load</a>()</code> <br><code><a href="image_smart_resize.html">image_smart_resize</a>()</code> <br><code><a href="image_to_array.html">image_to_array</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br></p>
-<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
+<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_layer_normalization.html">op_layer_normalization</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparse_sigmoid.html">op_sparse_sigmoid</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -144,7 +144,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/op_image_gaussian_blur.html b/docs/dev/reference/op_image_gaussian_blur.html
index b6694b42f..34f80cc44 100644
--- a/docs/dev/reference/op_image_gaussian_blur.html
+++ b/docs/dev/reference/op_image_gaussian_blur.html
@@ -8,7 +8,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -121,7 +121,7 @@ <h2 id="examples">Examples<a class="anchor" aria-label="anchor" href="#examples"
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
     <div class="dont-index"><p>Other image ops: <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br></p>
 <p>Other image utils: <br><code><a href="image_array_save.html">image_array_save</a>()</code> <br><code><a href="image_from_array.html">image_from_array</a>()</code> <br><code><a href="image_load.html">image_load</a>()</code> <br><code><a href="image_smart_resize.html">image_smart_resize</a>()</code> <br><code><a href="image_to_array.html">image_to_array</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br></p>
-<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
+<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_layer_normalization.html">op_layer_normalization</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparse_sigmoid.html">op_sparse_sigmoid</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -133,7 +133,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/op_image_hsv_to_rgb.html b/docs/dev/reference/op_image_hsv_to_rgb.html
index 4ba4c5d07..a8e5ff90a 100644
--- a/docs/dev/reference/op_image_hsv_to_rgb.html
+++ b/docs/dev/reference/op_image_hsv_to_rgb.html
@@ -10,7 +10,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -109,7 +109,7 @@ <h2 id="examples">Examples<a class="anchor" aria-label="anchor" href="#examples"
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
     <div class="dont-index"><p>Other image ops: <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br></p>
 <p>Other image utils: <br><code><a href="image_array_save.html">image_array_save</a>()</code> <br><code><a href="image_from_array.html">image_from_array</a>()</code> <br><code><a href="image_load.html">image_load</a>()</code> <br><code><a href="image_smart_resize.html">image_smart_resize</a>()</code> <br><code><a href="image_to_array.html">image_to_array</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br></p>
-<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
+<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_layer_normalization.html">op_layer_normalization</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparse_sigmoid.html">op_sparse_sigmoid</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -121,7 +121,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/op_image_map_coordinates.html b/docs/dev/reference/op_image_map_coordinates.html
index 3164caf6e..019b7cdac 100644
--- a/docs/dev/reference/op_image_map_coordinates.html
+++ b/docs/dev/reference/op_image_map_coordinates.html
@@ -12,7 +12,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -125,7 +125,7 @@ <h2 id="value">Value<a class="anchor" aria-label="anchor" href="#value"></a></h2
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
     <div class="dont-index"><p>Other image ops: <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br></p>
 <p>Other image utils: <br><code><a href="image_array_save.html">image_array_save</a>()</code> <br><code><a href="image_from_array.html">image_from_array</a>()</code> <br><code><a href="image_load.html">image_load</a>()</code> <br><code><a href="image_smart_resize.html">image_smart_resize</a>()</code> <br><code><a href="image_to_array.html">image_to_array</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br></p>
-<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
+<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_layer_normalization.html">op_layer_normalization</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparse_sigmoid.html">op_sparse_sigmoid</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -137,7 +137,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/op_image_pad.html b/docs/dev/reference/op_image_pad.html
index abbbd487f..e6e0f760e 100644
--- a/docs/dev/reference/op_image_pad.html
+++ b/docs/dev/reference/op_image_pad.html
@@ -8,7 +8,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -138,7 +138,7 @@ <h2 id="examples">Examples<a class="anchor" aria-label="anchor" href="#examples"
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
     <div class="dont-index"><p>Other image ops: <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br></p>
 <p>Other image utils: <br><code><a href="image_array_save.html">image_array_save</a>()</code> <br><code><a href="image_from_array.html">image_from_array</a>()</code> <br><code><a href="image_load.html">image_load</a>()</code> <br><code><a href="image_smart_resize.html">image_smart_resize</a>()</code> <br><code><a href="image_to_array.html">image_to_array</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br></p>
-<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
+<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_layer_normalization.html">op_layer_normalization</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparse_sigmoid.html">op_sparse_sigmoid</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -150,7 +150,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/op_image_perspective_transform.html b/docs/dev/reference/op_image_perspective_transform.html
index 1fdc67cee..065fe327d 100644
--- a/docs/dev/reference/op_image_perspective_transform.html
+++ b/docs/dev/reference/op_image_perspective_transform.html
@@ -8,7 +8,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -159,7 +159,7 @@ <h2 id="examples">Examples<a class="anchor" aria-label="anchor" href="#examples"
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
     <div class="dont-index"><p>Other image ops: <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br></p>
 <p>Other image utils: <br><code><a href="image_array_save.html">image_array_save</a>()</code> <br><code><a href="image_from_array.html">image_from_array</a>()</code> <br><code><a href="image_load.html">image_load</a>()</code> <br><code><a href="image_smart_resize.html">image_smart_resize</a>()</code> <br><code><a href="image_to_array.html">image_to_array</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br></p>
-<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
+<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_layer_normalization.html">op_layer_normalization</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparse_sigmoid.html">op_sparse_sigmoid</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -171,7 +171,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/op_image_resize.html b/docs/dev/reference/op_image_resize.html
index 83147189b..edfa67531 100644
--- a/docs/dev/reference/op_image_resize.html
+++ b/docs/dev/reference/op_image_resize.html
@@ -8,7 +8,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -160,7 +160,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 <ul><li><p><a href="https://keras.io/api/ops/image#resize-function" class="external-link">https://keras.io/api/ops/image#resize-function</a></p></li>
 </ul><p>Other image ops: <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br></p>
 <p>Other image utils: <br><code><a href="image_array_save.html">image_array_save</a>()</code> <br><code><a href="image_from_array.html">image_from_array</a>()</code> <br><code><a href="image_load.html">image_load</a>()</code> <br><code><a href="image_smart_resize.html">image_smart_resize</a>()</code> <br><code><a href="image_to_array.html">image_to_array</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br></p>
-<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
+<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_layer_normalization.html">op_layer_normalization</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparse_sigmoid.html">op_sparse_sigmoid</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -172,7 +172,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/op_image_rgb_to_grayscale.html b/docs/dev/reference/op_image_rgb_to_grayscale.html
index d65c51f09..36a065215 100644
--- a/docs/dev/reference/op_image_rgb_to_grayscale.html
+++ b/docs/dev/reference/op_image_rgb_to_grayscale.html
@@ -10,7 +10,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -109,7 +109,7 @@ <h2 id="examples">Examples<a class="anchor" aria-label="anchor" href="#examples"
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
     <div class="dont-index"><p>Other image ops: <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br></p>
 <p>Other image utils: <br><code><a href="image_array_save.html">image_array_save</a>()</code> <br><code><a href="image_from_array.html">image_from_array</a>()</code> <br><code><a href="image_load.html">image_load</a>()</code> <br><code><a href="image_smart_resize.html">image_smart_resize</a>()</code> <br><code><a href="image_to_array.html">image_to_array</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br></p>
-<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
+<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_layer_normalization.html">op_layer_normalization</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparse_sigmoid.html">op_sparse_sigmoid</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -121,7 +121,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/op_image_rgb_to_hsv.html b/docs/dev/reference/op_image_rgb_to_hsv.html
index fe395d27c..7a50cac58 100644
--- a/docs/dev/reference/op_image_rgb_to_hsv.html
+++ b/docs/dev/reference/op_image_rgb_to_hsv.html
@@ -14,7 +14,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -115,7 +115,7 @@ <h2 id="examples">Examples<a class="anchor" aria-label="anchor" href="#examples"
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
     <div class="dont-index"><p>Other image ops: <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br></p>
 <p>Other image utils: <br><code><a href="image_array_save.html">image_array_save</a>()</code> <br><code><a href="image_from_array.html">image_from_array</a>()</code> <br><code><a href="image_load.html">image_load</a>()</code> <br><code><a href="image_smart_resize.html">image_smart_resize</a>()</code> <br><code><a href="image_to_array.html">image_to_array</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br></p>
-<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
+<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_layer_normalization.html">op_layer_normalization</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparse_sigmoid.html">op_sparse_sigmoid</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -127,7 +127,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/op_in_top_k.html b/docs/dev/reference/op_in_top_k.html
index 5a97a7358..e4561f87d 100644
--- a/docs/dev/reference/op_in_top_k.html
+++ b/docs/dev/reference/op_in_top_k.html
@@ -8,7 +8,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -100,7 +100,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
     <div class="dont-index">
 <ul><li><p><a href="https://keras.io/api/ops/core#intopk-function" class="external-link">https://keras.io/api/ops/core#intopk-function</a></p></li>
 </ul><p>Other math ops: <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br></p>
-<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
+<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_layer_normalization.html">op_layer_normalization</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparse_sigmoid.html">op_sparse_sigmoid</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -112,7 +112,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/op_inner.html b/docs/dev/reference/op_inner.html
index ea2312a84..8667db25f 100644
--- a/docs/dev/reference/op_inner.html
+++ b/docs/dev/reference/op_inner.html
@@ -18,7 +18,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -98,8 +98,8 @@ <h2 id="value">Value<a class="anchor" aria-label="anchor" href="#value"></a></h2
     </div>
     <div class="section level2">
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
-    <div class="dont-index"><p>Other numpy ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p>
-<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
+    <div class="dont-index"><p>Other numpy ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p>
+<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_layer_normalization.html">op_layer_normalization</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparse_sigmoid.html">op_sparse_sigmoid</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -111,7 +111,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/op_inv.html b/docs/dev/reference/op_inv.html
index 43ed21f83..038d6a82e 100644
--- a/docs/dev/reference/op_inv.html
+++ b/docs/dev/reference/op_inv.html
@@ -8,7 +8,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -77,7 +77,7 @@ <h2 id="value">Value<a class="anchor" aria-label="anchor" href="#value"></a></h2
     <div class="section level2">
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
     <div class="dont-index"><p>Other linear algebra ops: <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br></p>
-<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
+<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_layer_normalization.html">op_layer_normalization</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparse_sigmoid.html">op_sparse_sigmoid</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -89,7 +89,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/op_irfft.html b/docs/dev/reference/op_irfft.html
index 92eee1f50..d535f6f13 100644
--- a/docs/dev/reference/op_irfft.html
+++ b/docs/dev/reference/op_irfft.html
@@ -28,7 +28,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -130,7 +130,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
     <div class="dont-index">
 <ul><li><p><a href="https://keras.io/api/ops/fft#irfft-function" class="external-link">https://keras.io/api/ops/fft#irfft-function</a></p></li>
 </ul><p>Other math ops: <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br></p>
-<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
+<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_layer_normalization.html">op_layer_normalization</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparse_sigmoid.html">op_sparse_sigmoid</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -142,7 +142,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/op_is_tensor.html b/docs/dev/reference/op_is_tensor.html
index e71c576ba..2f35233aa 100644
--- a/docs/dev/reference/op_is_tensor.html
+++ b/docs/dev/reference/op_is_tensor.html
@@ -8,7 +8,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -82,7 +82,7 @@ <h2 id="note">Note<a class="anchor" aria-label="anchor" href="#note"></a></h2>
     <div class="section level2">
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
     <div class="dont-index"><p>Other core ops: <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br></p>
-<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
+<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_layer_normalization.html">op_layer_normalization</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparse_sigmoid.html">op_sparse_sigmoid</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -94,7 +94,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/op_isclose.html b/docs/dev/reference/op_isclose.html
index 3f498f9e3..6b7bd9f0c 100644
--- a/docs/dev/reference/op_isclose.html
+++ b/docs/dev/reference/op_isclose.html
@@ -8,7 +8,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -94,8 +94,8 @@ <h2 id="value">Value<a class="anchor" aria-label="anchor" href="#value"></a></h2
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
     <div class="dont-index">
 <ul><li><p><a href="https://keras.io/api/ops/numpy#isclose-function" class="external-link">https://keras.io/api/ops/numpy#isclose-function</a></p></li>
-</ul><p>Other numpy ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p>
-<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
+</ul><p>Other numpy ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p>
+<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_layer_normalization.html">op_layer_normalization</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparse_sigmoid.html">op_sparse_sigmoid</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -107,7 +107,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/op_isfinite.html b/docs/dev/reference/op_isfinite.html
index f0f874a81..54ae6b2f2 100644
--- a/docs/dev/reference/op_isfinite.html
+++ b/docs/dev/reference/op_isfinite.html
@@ -12,7 +12,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -84,8 +84,8 @@ <h2 id="value">Value<a class="anchor" aria-label="anchor" href="#value"></a></h2
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
     <div class="dont-index">
 <ul><li><p><a href="https://keras.io/api/ops/numpy#isfinite-function" class="external-link">https://keras.io/api/ops/numpy#isfinite-function</a></p></li>
-</ul><p>Other numpy ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p>
-<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
+</ul><p>Other numpy ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p>
+<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_layer_normalization.html">op_layer_normalization</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparse_sigmoid.html">op_sparse_sigmoid</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -97,7 +97,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/op_isinf.html b/docs/dev/reference/op_isinf.html
index 2e0f0baac..544dbaf9c 100644
--- a/docs/dev/reference/op_isinf.html
+++ b/docs/dev/reference/op_isinf.html
@@ -8,7 +8,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -78,8 +78,8 @@ <h2 id="value">Value<a class="anchor" aria-label="anchor" href="#value"></a></h2
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
     <div class="dont-index">
 <ul><li><p><a href="https://keras.io/api/ops/numpy#isinf-function" class="external-link">https://keras.io/api/ops/numpy#isinf-function</a></p></li>
-</ul><p>Other numpy ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p>
-<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
+</ul><p>Other numpy ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p>
+<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_layer_normalization.html">op_layer_normalization</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparse_sigmoid.html">op_sparse_sigmoid</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -91,7 +91,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/op_isnan.html b/docs/dev/reference/op_isnan.html
index 65f929d19..ee7064028 100644
--- a/docs/dev/reference/op_isnan.html
+++ b/docs/dev/reference/op_isnan.html
@@ -8,7 +8,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -78,8 +78,8 @@ <h2 id="value">Value<a class="anchor" aria-label="anchor" href="#value"></a></h2
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
     <div class="dont-index">
 <ul><li><p><a href="https://keras.io/api/ops/numpy#isnan-function" class="external-link">https://keras.io/api/ops/numpy#isnan-function</a></p></li>
-</ul><p>Other numpy ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p>
-<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
+</ul><p>Other numpy ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p>
+<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_layer_normalization.html">op_layer_normalization</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparse_sigmoid.html">op_sparse_sigmoid</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -91,7 +91,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/op_istft.html b/docs/dev/reference/op_istft.html
index 65f5e1d13..04fdd3da4 100644
--- a/docs/dev/reference/op_istft.html
+++ b/docs/dev/reference/op_istft.html
@@ -10,7 +10,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -132,7 +132,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
     <div class="dont-index">
 <ul><li><p><a href="https://keras.io/api/ops/fft#istft-function" class="external-link">https://keras.io/api/ops/fft#istft-function</a></p></li>
 </ul><p>Other math ops: <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br></p>
-<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
+<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_layer_normalization.html">op_layer_normalization</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparse_sigmoid.html">op_sparse_sigmoid</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -144,7 +144,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/op_kaiser.html b/docs/dev/reference/op_kaiser.html
new file mode 100644
index 000000000..303974af9
--- /dev/null
+++ b/docs/dev/reference/op_kaiser.html
@@ -0,0 +1,118 @@
+<!DOCTYPE html>
+<!-- Generated by pkgdown: do not edit by hand --><html lang="en"><head><meta http-equiv="Content-Type" content="text/html; charset=UTF-8"><meta charset="utf-8"><meta http-equiv="X-UA-Compatible" content="IE=edge"><meta name="viewport" content="width=device-width, initial-scale=1, shrink-to-fit=no"><title>Kaiser window function. — op_kaiser • keras3</title><!-- favicons --><link rel="icon" type="image/png" sizes="96x96" href="../favicon-96x96.png"><link rel="icon" type="”image/svg+xml”" href="../favicon.svg"><link rel="apple-touch-icon" sizes="180x180" href="../apple-touch-icon.png"><link rel="icon" sizes="any" href="../favicon.ico"><link rel="manifest" href="../site.webmanifest"><script src="../deps/jquery-3.6.0/jquery-3.6.0.min.js"></script><meta name="viewport" content="width=device-width, initial-scale=1, shrink-to-fit=no"><link href="../deps/bootstrap-5.3.1/bootstrap.min.css" rel="stylesheet"><script src="../deps/bootstrap-5.3.1/bootstrap.bundle.min.js"></script><link href="../deps/font-awesome-6.5.2/css/all.min.css" rel="stylesheet"><link href="../deps/font-awesome-6.5.2/css/v4-shims.min.css" rel="stylesheet"><script src="../deps/headroom-0.11.0/headroom.min.js"></script><script src="../deps/headroom-0.11.0/jQuery.headroom.min.js"></script><script src="../deps/bootstrap-toc-1.0.1/bootstrap-toc.min.js"></script><script src="../deps/clipboard.js-2.0.11/clipboard.min.js"></script><script src="../deps/search-1.0.0/autocomplete.jquery.min.js"></script><script src="../deps/search-1.0.0/fuse.min.js"></script><script src="../deps/search-1.0.0/mark.min.js"></script><!-- pkgdown --><script src="../pkgdown.js"></script><link href="../extra.css" rel="stylesheet"><meta property="og:title" content="Kaiser window function. — op_kaiser"><meta name="description" content="The Kaiser window is defined as:
+w[n] = I0(beta * sqrt(1 - (2 * n / (N - 1) - 1)^2)) / I0(beta) where
+I0 is the modified zeroth-order Bessel function of the first kind."><meta property="og:description" content="The Kaiser window is defined as:
+w[n] = I0(beta * sqrt(1 - (2 * n / (N - 1) - 1)^2)) / I0(beta) where
+I0 is the modified zeroth-order Bessel function of the first kind."><meta name="robots" content="noindex"><!-- Google Tag Manager --><script>(function(w,d,s,l,i){w[l]=w[l]||[];w[l].push({'gtm.start': new Date().getTime(),event:'gtm.js'});var f=d.getElementsByTagName(s)[0],j=d.createElement(s),dl=l!='dataLayer'?'&l='+l:'';j.async=true;j.src='https://www.googletagmanager.com/gtm.js?id='+i+dl;f.parentNode.insertBefore(j,f);})(window,document,'script','dataLayer','GTM-KHBDBW7');</script><!-- End Google Tag Manager --></head><body>
+    <a href="#main" class="visually-hidden-focusable">Skip to contents</a>
+    <!-- Google Tag Manager (noscript) --> <noscript><iframe src="https://www.googletagmanager.com/ns.html?id=GTM-KHBDBW7" height="0" width="0" style="display:none;visibility:hidden"></iframe></noscript> <!-- End Google Tag Manager (noscript) -->
+
+
+    <nav class="navbar navbar-expand-lg fixed-top bg-primary" data-bs-theme="inverse" aria-label="Site navigation"><div class="container">
+
+    <a class="navbar-brand me-2" href="../index.html">keras3</a>
+
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
+
+
+    <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
+      <span class="navbar-toggler-icon"></span>
+    </button>
+
+    <div id="navbar" class="collapse navbar-collapse ms-3">
+      <ul class="navbar-nav me-auto"><li class="nav-item"><a class="nav-link" href="../articles/getting_started.html">Getting Started</a></li>
+<li class="nav-item dropdown">
+  <button class="nav-link dropdown-toggle" type="button" id="dropdown-guides" data-bs-toggle="dropdown" aria-expanded="false" aria-haspopup="true">Guides</button>
+  <ul class="dropdown-menu" aria-labelledby="dropdown-guides"><li><h6 class="dropdown-header" data-toc-skip>Model definition</h6></li>
+    <li><a class="dropdown-item" href="../articles/sequential_model.html">Sequential Model</a></li>
+    <li><a class="dropdown-item" href="../articles/functional_api.html">Functional API</a></li>
+    <li><h6 class="dropdown-header" data-toc-skip>Extending and customizing</h6></li>
+    <li><a class="dropdown-item" href="../articles/training_with_built_in_methods.html">Training &amp; evaluation with the built-in methods</a></li>
+    <li><a class="dropdown-item" href="../articles/custom_train_step_in_tensorflow.html">Customizing `fit()` with Tensorflow</a></li>
+    <li><a class="dropdown-item" href="../articles/writing_your_own_callbacks.html">Writing your own callbacks</a></li>
+    <li><a class="dropdown-item" href="../articles/making_new_layers_and_models_via_subclassing.html">Making new layers and models via subclassing</a></li>
+    <li><a class="dropdown-item" href="../articles/writing_a_custom_training_loop_in_tensorflow.html">Writing a training loop from scratch in TensorFlow</a></li>
+    <li><a class="dropdown-item" href="../articles/serialization_and_saving.html">Serialization and Saving</a></li>
+    <li><h6 class="dropdown-header" data-toc-skip>Other topics</h6></li>
+    <li><a class="dropdown-item" href="../articles/transfer_learning.html">Transfer learning and fine tuning</a></li>
+    <li><a class="dropdown-item" href="../articles/distributed_training_with_tensorflow.html">Distributed training with TensorFlow</a></li>
+    <li><a class="dropdown-item" href="../articles/distribution.html">Distributed training with Jax</a></li>
+  </ul></li>
+<li class="nav-item"><a class="nav-link" href="../articles/examples/index.html">Examples</a></li>
+<li class="active nav-item"><a class="nav-link" href="../reference/index.html">Reference</a></li>
+<li class="nav-item"><a class="nav-link" href="../news/index.html">News</a></li>
+      </ul><ul class="navbar-nav"><li class="nav-item"><form class="form-inline" role="search">
+ <input class="form-control" type="search" name="search-input" id="search-input" autocomplete="off" aria-label="Search site" placeholder="Search for" data-search-index="../search.json"></form></li>
+<li class="nav-item"><a class="external-link nav-link" href="https://github.com/rstudio/keras3/" aria-label="GitHub"><span class="fa fab fa-github fa-lg"></span></a></li>
+      </ul></div>
+
+
+  </div>
+</nav><div class="container template-reference-topic">
+<div class="row">
+  <main id="main" class="col-md-9"><div class="page-header">
+
+      <h1>Kaiser window function.</h1>
+      <small class="dont-index">Source: <a href="https://github.com/rstudio/keras3/blob/HEAD/R/ops-numpy.R" class="external-link"><code>R/ops-numpy.R</code></a></small>
+      <div class="d-none name"><code>op_kaiser.Rd</code></div>
+    </div>
+
+    <div class="ref-description section level2">
+    <p>The Kaiser window is defined as:
+<code>w[n] = I0(beta * sqrt(1 - (2 * n / (N - 1) - 1)^2)) / I0(beta)</code> where
+<code>I0</code> is the modified zeroth-order Bessel function of the first kind.</p>
+    </div>
+
+    <div class="section level2">
+    <h2 id="ref-usage">Usage<a class="anchor" aria-label="anchor" href="#ref-usage"></a></h2>
+    <div class="sourceCode"><pre class="sourceCode r"><code><span><span class="fu">op_kaiser</span><span class="op">(</span><span class="va">x</span>, <span class="va">beta</span><span class="op">)</span></span></code></pre></div>
+    </div>
+
+    <div class="section level2">
+    <h2 id="arguments">Arguments<a class="anchor" aria-label="anchor" href="#arguments"></a></h2>
+
+
+<dl><dt id="arg-x">x<a class="anchor" aria-label="anchor" href="#arg-x"></a></dt>
+<dd><p>Length of the window. Must be a positive integer.</p></dd>
+
+
+<dt id="arg-beta">beta<a class="anchor" aria-label="anchor" href="#arg-beta"></a></dt>
+<dd><p>Shape parameter for the window.</p></dd>
+
+</dl></div>
+    <div class="section level2">
+    <h2 id="value">Value<a class="anchor" aria-label="anchor" href="#value"></a></h2>
+    <p>A 1D tensor containing the window values.</p>
+    </div>
+    <div class="section level2">
+    <h2 id="examples">Examples<a class="anchor" aria-label="anchor" href="#examples"></a></h2>
+    <p></p><div class="sourceCode r"><pre><code><span><span class="fu"><a href="../reference/op_kaiser.html">op_kaiser</a></span><span class="op">(</span><span class="fl">5</span>, beta <span class="op">=</span> <span class="fl">14</span><span class="op">)</span></span></code></pre><p></p></div>
+<p></p><div class="sourceCode"><pre><code><span id="cb1-1"><a href="#cb1-1" tabindex="-1"></a><span class="do">## tf.Tensor([7.7268669e-06 1.6493215e-01 1.0000000e+00 1.6493215e-01 7.7268669e-06], shape=(5), dtype=float32)</span></span>
+<span id="cb1-2"><a href="#cb1-2" tabindex="-1"></a></span></code></pre><p></p></div>
+    </div>
+    <div class="section level2">
+    <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
+    <div class="dont-index"><p>Other numpy ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p>
+<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_layer_normalization.html">op_layer_normalization</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparse_sigmoid.html">op_sparse_sigmoid</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
+    </div>
+
+  </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
+    </nav></aside></div>
+
+
+    <footer><div class="pkgdown-footer-left">
+  <p>Developed by Tomasz Kalinowski, JJ Allaire, François Chollet, Posit Software, PBC, Google.</p>
+</div>
+
+<div class="pkgdown-footer-right">
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
+</div>
+
+    </footer></div>
+
+
+
+
+
+  </body></html>
+
diff --git a/docs/dev/reference/op_layer_normalization.html b/docs/dev/reference/op_layer_normalization.html
new file mode 100644
index 000000000..270e19ea4
--- /dev/null
+++ b/docs/dev/reference/op_layer_normalization.html
@@ -0,0 +1,139 @@
+<!DOCTYPE html>
+<!-- Generated by pkgdown: do not edit by hand --><html lang="en"><head><meta http-equiv="Content-Type" content="text/html; charset=UTF-8"><meta charset="utf-8"><meta http-equiv="X-UA-Compatible" content="IE=edge"><meta name="viewport" content="width=device-width, initial-scale=1, shrink-to-fit=no"><title>Layer normalization (Ba et al., 2016). — op_layer_normalization • keras3</title><!-- favicons --><link rel="icon" type="image/png" sizes="96x96" href="../favicon-96x96.png"><link rel="icon" type="”image/svg+xml”" href="../favicon.svg"><link rel="apple-touch-icon" sizes="180x180" href="../apple-touch-icon.png"><link rel="icon" sizes="any" href="../favicon.ico"><link rel="manifest" href="../site.webmanifest"><script src="../deps/jquery-3.6.0/jquery-3.6.0.min.js"></script><meta name="viewport" content="width=device-width, initial-scale=1, shrink-to-fit=no"><link href="../deps/bootstrap-5.3.1/bootstrap.min.css" rel="stylesheet"><script src="../deps/bootstrap-5.3.1/bootstrap.bundle.min.js"></script><link href="../deps/font-awesome-6.5.2/css/all.min.css" rel="stylesheet"><link href="../deps/font-awesome-6.5.2/css/v4-shims.min.css" rel="stylesheet"><script src="../deps/headroom-0.11.0/headroom.min.js"></script><script src="../deps/headroom-0.11.0/jQuery.headroom.min.js"></script><script src="../deps/bootstrap-toc-1.0.1/bootstrap-toc.min.js"></script><script src="../deps/clipboard.js-2.0.11/clipboard.min.js"></script><script src="../deps/search-1.0.0/autocomplete.jquery.min.js"></script><script src="../deps/search-1.0.0/fuse.min.js"></script><script src="../deps/search-1.0.0/mark.min.js"></script><!-- pkgdown --><script src="../pkgdown.js"></script><link href="../extra.css" rel="stylesheet"><meta property="og:title" content="Layer normalization (Ba et al., 2016). — op_layer_normalization"><meta name="description" content="Normalizes activations in x for each example independently by centering to
+mean 0 and scaling to unit variance along the specified axis."><meta property="og:description" content="Normalizes activations in x for each example independently by centering to
+mean 0 and scaling to unit variance along the specified axis."><meta name="robots" content="noindex"><!-- Google Tag Manager --><script>(function(w,d,s,l,i){w[l]=w[l]||[];w[l].push({'gtm.start': new Date().getTime(),event:'gtm.js'});var f=d.getElementsByTagName(s)[0],j=d.createElement(s),dl=l!='dataLayer'?'&l='+l:'';j.async=true;j.src='https://www.googletagmanager.com/gtm.js?id='+i+dl;f.parentNode.insertBefore(j,f);})(window,document,'script','dataLayer','GTM-KHBDBW7');</script><!-- End Google Tag Manager --></head><body>
+    <a href="#main" class="visually-hidden-focusable">Skip to contents</a>
+    <!-- Google Tag Manager (noscript) --> <noscript><iframe src="https://www.googletagmanager.com/ns.html?id=GTM-KHBDBW7" height="0" width="0" style="display:none;visibility:hidden"></iframe></noscript> <!-- End Google Tag Manager (noscript) -->
+
+
+    <nav class="navbar navbar-expand-lg fixed-top bg-primary" data-bs-theme="inverse" aria-label="Site navigation"><div class="container">
+
+    <a class="navbar-brand me-2" href="../index.html">keras3</a>
+
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
+
+
+    <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
+      <span class="navbar-toggler-icon"></span>
+    </button>
+
+    <div id="navbar" class="collapse navbar-collapse ms-3">
+      <ul class="navbar-nav me-auto"><li class="nav-item"><a class="nav-link" href="../articles/getting_started.html">Getting Started</a></li>
+<li class="nav-item dropdown">
+  <button class="nav-link dropdown-toggle" type="button" id="dropdown-guides" data-bs-toggle="dropdown" aria-expanded="false" aria-haspopup="true">Guides</button>
+  <ul class="dropdown-menu" aria-labelledby="dropdown-guides"><li><h6 class="dropdown-header" data-toc-skip>Model definition</h6></li>
+    <li><a class="dropdown-item" href="../articles/sequential_model.html">Sequential Model</a></li>
+    <li><a class="dropdown-item" href="../articles/functional_api.html">Functional API</a></li>
+    <li><h6 class="dropdown-header" data-toc-skip>Extending and customizing</h6></li>
+    <li><a class="dropdown-item" href="../articles/training_with_built_in_methods.html">Training &amp; evaluation with the built-in methods</a></li>
+    <li><a class="dropdown-item" href="../articles/custom_train_step_in_tensorflow.html">Customizing `fit()` with Tensorflow</a></li>
+    <li><a class="dropdown-item" href="../articles/writing_your_own_callbacks.html">Writing your own callbacks</a></li>
+    <li><a class="dropdown-item" href="../articles/making_new_layers_and_models_via_subclassing.html">Making new layers and models via subclassing</a></li>
+    <li><a class="dropdown-item" href="../articles/writing_a_custom_training_loop_in_tensorflow.html">Writing a training loop from scratch in TensorFlow</a></li>
+    <li><a class="dropdown-item" href="../articles/serialization_and_saving.html">Serialization and Saving</a></li>
+    <li><h6 class="dropdown-header" data-toc-skip>Other topics</h6></li>
+    <li><a class="dropdown-item" href="../articles/transfer_learning.html">Transfer learning and fine tuning</a></li>
+    <li><a class="dropdown-item" href="../articles/distributed_training_with_tensorflow.html">Distributed training with TensorFlow</a></li>
+    <li><a class="dropdown-item" href="../articles/distribution.html">Distributed training with Jax</a></li>
+  </ul></li>
+<li class="nav-item"><a class="nav-link" href="../articles/examples/index.html">Examples</a></li>
+<li class="active nav-item"><a class="nav-link" href="../reference/index.html">Reference</a></li>
+<li class="nav-item"><a class="nav-link" href="../news/index.html">News</a></li>
+      </ul><ul class="navbar-nav"><li class="nav-item"><form class="form-inline" role="search">
+ <input class="form-control" type="search" name="search-input" id="search-input" autocomplete="off" aria-label="Search site" placeholder="Search for" data-search-index="../search.json"></form></li>
+<li class="nav-item"><a class="external-link nav-link" href="https://github.com/rstudio/keras3/" aria-label="GitHub"><span class="fa fab fa-github fa-lg"></span></a></li>
+      </ul></div>
+
+
+  </div>
+</nav><div class="container template-reference-topic">
+<div class="row">
+  <main id="main" class="col-md-9"><div class="page-header">
+
+      <h1>Layer normalization (Ba et al., 2016).</h1>
+      <small class="dont-index">Source: <a href="https://github.com/rstudio/keras3/blob/HEAD/R/ops-nn.R" class="external-link"><code>R/ops-nn.R</code></a></small>
+      <div class="d-none name"><code>op_layer_normalization.Rd</code></div>
+    </div>
+
+    <div class="ref-description section level2">
+    <p>Normalizes activations in <code>x</code> for each example independently by centering to
+mean 0 and scaling to unit variance along the specified <code>axis</code>.</p>
+    </div>
+
+    <div class="section level2">
+    <h2 id="ref-usage">Usage<a class="anchor" aria-label="anchor" href="#ref-usage"></a></h2>
+    <div class="sourceCode"><pre class="sourceCode r"><code><span><span class="fu">op_layer_normalization</span><span class="op">(</span></span>
+<span>  <span class="va">x</span>,</span>
+<span>  gamma <span class="op">=</span> <span class="cn">NULL</span>,</span>
+<span>  beta <span class="op">=</span> <span class="cn">NULL</span>,</span>
+<span>  axis <span class="op">=</span> <span class="op">-</span><span class="fl">1L</span>,</span>
+<span>  epsilon <span class="op">=</span> <span class="cn">NULL</span>,</span>
+<span>  <span class="va">...</span></span>
+<span><span class="op">)</span></span></code></pre></div>
+    </div>
+
+    <div class="section level2">
+    <h2 id="arguments">Arguments<a class="anchor" aria-label="anchor" href="#arguments"></a></h2>
+
+
+<dl><dt id="arg-x">x<a class="anchor" aria-label="anchor" href="#arg-x"></a></dt>
+<dd><p>Input tensor.</p></dd>
+
+
+<dt id="arg-gamma">gamma<a class="anchor" aria-label="anchor" href="#arg-gamma"></a></dt>
+<dd><p>Optional scaling factor applied to the normalized output.</p></dd>
+
+
+<dt id="arg-beta">beta<a class="anchor" aria-label="anchor" href="#arg-beta"></a></dt>
+<dd><p>Optional offset added to the normalized output.</p></dd>
+
+
+<dt id="arg-axis">axis<a class="anchor" aria-label="anchor" href="#arg-axis"></a></dt>
+<dd><p>Axis or axes along which to compute statistics. Defaults to <code>-1</code>.</p></dd>
+
+
+<dt id="arg-epsilon">epsilon<a class="anchor" aria-label="anchor" href="#arg-epsilon"></a></dt>
+<dd><p>Small constant added to the variance for numerical stability.</p></dd>
+
+
+<dt id="arg--">...<a class="anchor" aria-label="anchor" href="#arg--"></a></dt>
+<dd><p>For forward/backward compatibility.</p></dd>
+
+</dl></div>
+    <div class="section level2">
+    <h2 id="value">Value<a class="anchor" aria-label="anchor" href="#value"></a></h2>
+    <p>Tensor with the same shape as <code>x</code> containing the normalized values.</p>
+    </div>
+    <div class="section level2">
+    <h2 id="examples">Examples<a class="anchor" aria-label="anchor" href="#examples"></a></h2>
+    <p></p><div class="sourceCode r"><pre><code><span><span class="va">x</span> <span class="op">&lt;-</span> <span class="fu"><a href="../reference/op_arange.html">op_arange</a></span><span class="op">(</span><span class="fl">5</span>, dtype <span class="op">=</span> <span class="st">"float32"</span><span class="op">)</span></span>
+<span><span class="fu"><a href="../reference/op_layer_normalization.html">op_layer_normalization</a></span><span class="op">(</span><span class="va">x</span><span class="op">)</span></span></code></pre><p></p></div>
+<p></p><div class="sourceCode"><pre><code><span id="cb1-1"><a href="#cb1-1" tabindex="-1"></a><span class="do">## tf.Tensor([-1.4142134 -0.7071067  0.         0.7071068  1.4142137], shape=(5), dtype=float32)</span></span>
+<span id="cb1-2"><a href="#cb1-2" tabindex="-1"></a></span></code></pre><p></p></div>
+    </div>
+    <div class="section level2">
+    <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
+    <div class="dont-index"><p>Other nn ops: <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparse_sigmoid.html">op_sparse_sigmoid</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br></p>
+<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparse_sigmoid.html">op_sparse_sigmoid</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
+    </div>
+
+  </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
+    </nav></aside></div>
+
+
+    <footer><div class="pkgdown-footer-left">
+  <p>Developed by Tomasz Kalinowski, JJ Allaire, François Chollet, Posit Software, PBC, Google.</p>
+</div>
+
+<div class="pkgdown-footer-right">
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
+</div>
+
+    </footer></div>
+
+
+
+
+
+  </body></html>
+
diff --git a/docs/dev/reference/op_leaky_relu.html b/docs/dev/reference/op_leaky_relu.html
index eeadead53..1ac12e36f 100644
--- a/docs/dev/reference/op_leaky_relu.html
+++ b/docs/dev/reference/op_leaky_relu.html
@@ -10,7 +10,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -98,8 +98,8 @@ <h2 id="examples">Examples<a class="anchor" aria-label="anchor" href="#examples"
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
     <div class="dont-index">
 <ul><li><p><a href="https://keras.io/api/ops/nn#leakyrelu-function" class="external-link">https://keras.io/api/ops/nn#leakyrelu-function</a></p></li>
-</ul><p>Other nn ops: <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br></p>
-<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
+</ul><p>Other nn ops: <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_layer_normalization.html">op_layer_normalization</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparse_sigmoid.html">op_sparse_sigmoid</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br></p>
+<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_layer_normalization.html">op_layer_normalization</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparse_sigmoid.html">op_sparse_sigmoid</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -111,7 +111,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/op_left_shift.html b/docs/dev/reference/op_left_shift.html
index fce097420..c937b5618 100644
--- a/docs/dev/reference/op_left_shift.html
+++ b/docs/dev/reference/op_left_shift.html
@@ -12,7 +12,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -86,8 +86,8 @@ <h2 id="value">Value<a class="anchor" aria-label="anchor" href="#value"></a></h2
     </div>
     <div class="section level2">
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
-    <div class="dont-index"><p>Other numpy ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p>
-<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
+    <div class="dont-index"><p>Other numpy ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p>
+<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_layer_normalization.html">op_layer_normalization</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparse_sigmoid.html">op_sparse_sigmoid</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -99,7 +99,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/op_less.html b/docs/dev/reference/op_less.html
index e11f24133..0127933ae 100644
--- a/docs/dev/reference/op_less.html
+++ b/docs/dev/reference/op_less.html
@@ -38,7 +38,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -121,8 +121,8 @@ <h2 id="value">Value<a class="anchor" aria-label="anchor" href="#value"></a></h2
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
     <div class="dont-index">
 <ul><li><p><a href="https://keras.io/api/ops/numpy#less-function" class="external-link">https://keras.io/api/ops/numpy#less-function</a></p></li>
-</ul><p>Other numpy ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p>
-<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
+</ul><p>Other numpy ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p>
+<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_layer_normalization.html">op_layer_normalization</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparse_sigmoid.html">op_sparse_sigmoid</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -134,7 +134,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/op_less_equal.html b/docs/dev/reference/op_less_equal.html
index 18aab0643..bcf237e3b 100644
--- a/docs/dev/reference/op_less_equal.html
+++ b/docs/dev/reference/op_less_equal.html
@@ -38,7 +38,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -121,8 +121,8 @@ <h2 id="value">Value<a class="anchor" aria-label="anchor" href="#value"></a></h2
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
     <div class="dont-index">
 <ul><li><p><a href="https://keras.io/api/ops/numpy#lessequal-function" class="external-link">https://keras.io/api/ops/numpy#lessequal-function</a></p></li>
-</ul><p>Other numpy ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p>
-<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
+</ul><p>Other numpy ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p>
+<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_layer_normalization.html">op_layer_normalization</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparse_sigmoid.html">op_sparse_sigmoid</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -134,7 +134,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/op_linspace.html b/docs/dev/reference/op_linspace.html
index ef334cbf2..fc033173e 100644
--- a/docs/dev/reference/op_linspace.html
+++ b/docs/dev/reference/op_linspace.html
@@ -12,7 +12,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -128,8 +128,8 @@ <h2 id="note">Note<a class="anchor" aria-label="anchor" href="#note"></a></h2>
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
     <div class="dont-index">
 <ul><li><p><a href="https://keras.io/api/ops/numpy#linspace-function" class="external-link">https://keras.io/api/ops/numpy#linspace-function</a></p></li>
-</ul><p>Other numpy ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p>
-<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
+</ul><p>Other numpy ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p>
+<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_layer_normalization.html">op_layer_normalization</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparse_sigmoid.html">op_sparse_sigmoid</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -141,7 +141,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/op_log.html b/docs/dev/reference/op_log.html
index f2a77b0cf..c0401e69e 100644
--- a/docs/dev/reference/op_log.html
+++ b/docs/dev/reference/op_log.html
@@ -8,7 +8,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -78,8 +78,8 @@ <h2 id="value">Value<a class="anchor" aria-label="anchor" href="#value"></a></h2
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
     <div class="dont-index">
 <ul><li><p><a href="https://keras.io/api/ops/numpy#log-function" class="external-link">https://keras.io/api/ops/numpy#log-function</a></p></li>
-</ul><p>Other numpy ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p>
-<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
+</ul><p>Other numpy ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p>
+<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_layer_normalization.html">op_layer_normalization</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparse_sigmoid.html">op_sparse_sigmoid</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -91,7 +91,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/op_log10.html b/docs/dev/reference/op_log10.html
index aabbd7a21..eaf535790 100644
--- a/docs/dev/reference/op_log10.html
+++ b/docs/dev/reference/op_log10.html
@@ -8,7 +8,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -78,8 +78,8 @@ <h2 id="value">Value<a class="anchor" aria-label="anchor" href="#value"></a></h2
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
     <div class="dont-index">
 <ul><li><p><a href="https://keras.io/api/ops/numpy#log10-function" class="external-link">https://keras.io/api/ops/numpy#log10-function</a></p></li>
-</ul><p>Other numpy ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p>
-<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
+</ul><p>Other numpy ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p>
+<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_layer_normalization.html">op_layer_normalization</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparse_sigmoid.html">op_sparse_sigmoid</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -91,7 +91,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/op_log1p.html b/docs/dev/reference/op_log1p.html
index 93b44a99e..b245be50b 100644
--- a/docs/dev/reference/op_log1p.html
+++ b/docs/dev/reference/op_log1p.html
@@ -8,7 +8,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -78,8 +78,8 @@ <h2 id="value">Value<a class="anchor" aria-label="anchor" href="#value"></a></h2
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
     <div class="dont-index">
 <ul><li><p><a href="https://keras.io/api/ops/numpy#log1p-function" class="external-link">https://keras.io/api/ops/numpy#log1p-function</a></p></li>
-</ul><p>Other numpy ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p>
-<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
+</ul><p>Other numpy ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p>
+<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_layer_normalization.html">op_layer_normalization</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparse_sigmoid.html">op_sparse_sigmoid</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -91,7 +91,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/op_log2.html b/docs/dev/reference/op_log2.html
index 4bde9531e..45bf9ff8d 100644
--- a/docs/dev/reference/op_log2.html
+++ b/docs/dev/reference/op_log2.html
@@ -8,7 +8,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -78,8 +78,8 @@ <h2 id="value">Value<a class="anchor" aria-label="anchor" href="#value"></a></h2
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
     <div class="dont-index">
 <ul><li><p><a href="https://keras.io/api/ops/numpy#log2-function" class="external-link">https://keras.io/api/ops/numpy#log2-function</a></p></li>
-</ul><p>Other numpy ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p>
-<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
+</ul><p>Other numpy ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p>
+<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_layer_normalization.html">op_layer_normalization</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparse_sigmoid.html">op_sparse_sigmoid</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -91,7 +91,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/op_log_sigmoid.html b/docs/dev/reference/op_log_sigmoid.html
index 762db3cf5..82d2846ba 100644
--- a/docs/dev/reference/op_log_sigmoid.html
+++ b/docs/dev/reference/op_log_sigmoid.html
@@ -8,7 +8,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -85,8 +85,8 @@ <h2 id="examples">Examples<a class="anchor" aria-label="anchor" href="#examples"
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
     <div class="dont-index">
 <ul><li><p><a href="https://keras.io/api/ops/nn#logsigmoid-function" class="external-link">https://keras.io/api/ops/nn#logsigmoid-function</a></p></li>
-</ul><p>Other nn ops: <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br></p>
-<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
+</ul><p>Other nn ops: <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_layer_normalization.html">op_layer_normalization</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparse_sigmoid.html">op_sparse_sigmoid</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br></p>
+<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_layer_normalization.html">op_layer_normalization</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparse_sigmoid.html">op_sparse_sigmoid</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -98,7 +98,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/op_log_softmax.html b/docs/dev/reference/op_log_softmax.html
index d2e10349c..bf6e434fe 100644
--- a/docs/dev/reference/op_log_softmax.html
+++ b/docs/dev/reference/op_log_softmax.html
@@ -10,7 +10,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -93,8 +93,8 @@ <h2 id="examples">Examples<a class="anchor" aria-label="anchor" href="#examples"
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
     <div class="dont-index">
 <ul><li><p><a href="https://keras.io/api/ops/nn#logsoftmax-function" class="external-link">https://keras.io/api/ops/nn#logsoftmax-function</a></p></li>
-</ul><p>Other nn ops: <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br></p>
-<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
+</ul><p>Other nn ops: <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_layer_normalization.html">op_layer_normalization</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparse_sigmoid.html">op_sparse_sigmoid</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br></p>
+<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_layer_normalization.html">op_layer_normalization</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparse_sigmoid.html">op_sparse_sigmoid</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -106,7 +106,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/op_logaddexp.html b/docs/dev/reference/op_logaddexp.html
index e8ab991aa..6709d7e4b 100644
--- a/docs/dev/reference/op_logaddexp.html
+++ b/docs/dev/reference/op_logaddexp.html
@@ -8,7 +8,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -83,8 +83,8 @@ <h2 id="value">Value<a class="anchor" aria-label="anchor" href="#value"></a></h2
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
     <div class="dont-index">
 <ul><li><p><a href="https://keras.io/api/ops/numpy#logaddexp-function" class="external-link">https://keras.io/api/ops/numpy#logaddexp-function</a></p></li>
-</ul><p>Other numpy ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p>
-<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
+</ul><p>Other numpy ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p>
+<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_layer_normalization.html">op_layer_normalization</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparse_sigmoid.html">op_sparse_sigmoid</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -96,7 +96,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/op_logdet.html b/docs/dev/reference/op_logdet.html
index 021c012a6..fc9c407ae 100644
--- a/docs/dev/reference/op_logdet.html
+++ b/docs/dev/reference/op_logdet.html
@@ -8,7 +8,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -76,8 +76,8 @@ <h2 id="value">Value<a class="anchor" aria-label="anchor" href="#value"></a></h2
     </div>
     <div class="section level2">
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
-    <div class="dont-index"><p>Other numpy ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p>
-<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
+    <div class="dont-index"><p>Other numpy ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p>
+<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_layer_normalization.html">op_layer_normalization</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparse_sigmoid.html">op_sparse_sigmoid</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -89,7 +89,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/op_logical_and.html b/docs/dev/reference/op_logical_and.html
index eeffdc715..275ebcdad 100644
--- a/docs/dev/reference/op_logical_and.html
+++ b/docs/dev/reference/op_logical_and.html
@@ -8,7 +8,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -86,8 +86,8 @@ <h2 id="details">Details<a class="anchor" aria-label="anchor" href="#details"></
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
     <div class="dont-index">
 <ul><li><p><a href="https://keras.io/api/ops/numpy#logicaland-function" class="external-link">https://keras.io/api/ops/numpy#logicaland-function</a></p></li>
-</ul><p>Other numpy ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p>
-<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
+</ul><p>Other numpy ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p>
+<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_layer_normalization.html">op_layer_normalization</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparse_sigmoid.html">op_sparse_sigmoid</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -99,7 +99,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/op_logical_not.html b/docs/dev/reference/op_logical_not.html
index 35af8f78d..146841c42 100644
--- a/docs/dev/reference/op_logical_not.html
+++ b/docs/dev/reference/op_logical_not.html
@@ -10,7 +10,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -81,8 +81,8 @@ <h2 id="value">Value<a class="anchor" aria-label="anchor" href="#value"></a></h2
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
     <div class="dont-index">
 <ul><li><p><a href="https://keras.io/api/ops/numpy#logicalnot-function" class="external-link">https://keras.io/api/ops/numpy#logicalnot-function</a></p></li>
-</ul><p>Other numpy ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p>
-<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
+</ul><p>Other numpy ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p>
+<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_layer_normalization.html">op_layer_normalization</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparse_sigmoid.html">op_sparse_sigmoid</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -94,7 +94,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/op_logical_or.html b/docs/dev/reference/op_logical_or.html
index 6dfb2bc61..cb89f57a8 100644
--- a/docs/dev/reference/op_logical_or.html
+++ b/docs/dev/reference/op_logical_or.html
@@ -10,7 +10,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -85,8 +85,8 @@ <h2 id="value">Value<a class="anchor" aria-label="anchor" href="#value"></a></h2
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
     <div class="dont-index">
 <ul><li><p><a href="https://keras.io/api/ops/numpy#logicalor-function" class="external-link">https://keras.io/api/ops/numpy#logicalor-function</a></p></li>
-</ul><p>Other numpy ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p>
-<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
+</ul><p>Other numpy ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p>
+<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_layer_normalization.html">op_layer_normalization</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparse_sigmoid.html">op_sparse_sigmoid</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -98,7 +98,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/op_logical_xor.html b/docs/dev/reference/op_logical_xor.html
index 3fa6fd22f..f5f360889 100644
--- a/docs/dev/reference/op_logical_xor.html
+++ b/docs/dev/reference/op_logical_xor.html
@@ -8,7 +8,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -80,8 +80,8 @@ <h2 id="value">Value<a class="anchor" aria-label="anchor" href="#value"></a></h2
     </div>
     <div class="section level2">
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
-    <div class="dont-index"><p>Other numpy ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p>
-<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
+    <div class="dont-index"><p>Other numpy ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p>
+<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_layer_normalization.html">op_layer_normalization</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparse_sigmoid.html">op_sparse_sigmoid</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -93,7 +93,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/op_logspace.html b/docs/dev/reference/op_logspace.html
index 65c225e31..46f7a4db6 100644
--- a/docs/dev/reference/op_logspace.html
+++ b/docs/dev/reference/op_logspace.html
@@ -10,7 +10,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -122,8 +122,8 @@ <h2 id="note">Note<a class="anchor" aria-label="anchor" href="#note"></a></h2>
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
     <div class="dont-index">
 <ul><li><p><a href="https://keras.io/api/ops/numpy#logspace-function" class="external-link">https://keras.io/api/ops/numpy#logspace-function</a></p></li>
-</ul><p>Other numpy ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p>
-<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
+</ul><p>Other numpy ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p>
+<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_layer_normalization.html">op_layer_normalization</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparse_sigmoid.html">op_sparse_sigmoid</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -135,7 +135,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/op_logsumexp.html b/docs/dev/reference/op_logsumexp.html
index a5604d9a5..d1093aef2 100644
--- a/docs/dev/reference/op_logsumexp.html
+++ b/docs/dev/reference/op_logsumexp.html
@@ -8,7 +8,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -98,7 +98,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
     <div class="dont-index">
 <ul><li><p><a href="https://keras.io/api/ops/core#logsumexp-function" class="external-link">https://keras.io/api/ops/core#logsumexp-function</a></p></li>
 </ul><p>Other math ops: <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br></p>
-<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
+<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_layer_normalization.html">op_layer_normalization</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparse_sigmoid.html">op_sparse_sigmoid</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -110,7 +110,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/op_lstsq.html b/docs/dev/reference/op_lstsq.html
index a1bdf306e..8b741f2b7 100644
--- a/docs/dev/reference/op_lstsq.html
+++ b/docs/dev/reference/op_lstsq.html
@@ -24,7 +24,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -124,8 +124,8 @@ <h2 id="value">Value<a class="anchor" aria-label="anchor" href="#value"></a></h2
     <div class="section level2">
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
     <div class="dont-index"><p>Other linear algebra ops: <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br></p>
-<p>Other numpy ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p>
-<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
+<p>Other numpy ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p>
+<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_layer_normalization.html">op_layer_normalization</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparse_sigmoid.html">op_sparse_sigmoid</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -137,7 +137,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/op_lu_factor.html b/docs/dev/reference/op_lu_factor.html
index 36d6f2f17..336bcb485 100644
--- a/docs/dev/reference/op_lu_factor.html
+++ b/docs/dev/reference/op_lu_factor.html
@@ -8,7 +8,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -79,7 +79,7 @@ <h2 id="value">Value<a class="anchor" aria-label="anchor" href="#value"></a></h2
     <div class="section level2">
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
     <div class="dont-index"><p>Other linear algebra ops: <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br></p>
-<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
+<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_layer_normalization.html">op_layer_normalization</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparse_sigmoid.html">op_sparse_sigmoid</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -91,7 +91,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/op_map.html b/docs/dev/reference/op_map.html
index 170cb8824..78d03877a 100644
--- a/docs/dev/reference/op_map.html
+++ b/docs/dev/reference/op_map.html
@@ -32,7 +32,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -134,7 +134,7 @@ <h2 id="examples">Examples<a class="anchor" aria-label="anchor" href="#examples"
     <div class="section level2">
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
     <div class="dont-index"><p>Other core ops: <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br></p>
-<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
+<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_layer_normalization.html">op_layer_normalization</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparse_sigmoid.html">op_sparse_sigmoid</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -146,7 +146,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/op_matmul.html b/docs/dev/reference/op_matmul.html
index 3adfce611..fbec3448c 100644
--- a/docs/dev/reference/op_matmul.html
+++ b/docs/dev/reference/op_matmul.html
@@ -30,7 +30,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -112,8 +112,8 @@ <h2 id="value">Value<a class="anchor" aria-label="anchor" href="#value"></a></h2
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
     <div class="dont-index">
 <ul><li><p><a href="https://keras.io/api/ops/numpy#matmul-function" class="external-link">https://keras.io/api/ops/numpy#matmul-function</a></p></li>
-</ul><p>Other numpy ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p>
-<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
+</ul><p>Other numpy ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p>
+<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_layer_normalization.html">op_layer_normalization</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparse_sigmoid.html">op_sparse_sigmoid</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -125,7 +125,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/op_max.html b/docs/dev/reference/op_max.html
index b712489ae..5b707237c 100644
--- a/docs/dev/reference/op_max.html
+++ b/docs/dev/reference/op_max.html
@@ -8,7 +8,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -109,8 +109,8 @@ <h2 id="examples">Examples<a class="anchor" aria-label="anchor" href="#examples"
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
     <div class="dont-index">
 <ul><li><p><a href="https://keras.io/api/ops/numpy#max-function" class="external-link">https://keras.io/api/ops/numpy#max-function</a></p></li>
-</ul><p>Other numpy ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p>
-<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
+</ul><p>Other numpy ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p>
+<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_layer_normalization.html">op_layer_normalization</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparse_sigmoid.html">op_sparse_sigmoid</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -122,7 +122,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/op_max_pool.html b/docs/dev/reference/op_max_pool.html
index 30ccb7f13..c9910d19e 100644
--- a/docs/dev/reference/op_max_pool.html
+++ b/docs/dev/reference/op_max_pool.html
@@ -8,7 +8,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -120,8 +120,8 @@ <h2 id="value">Value<a class="anchor" aria-label="anchor" href="#value"></a></h2
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
     <div class="dont-index">
 <ul><li><p><a href="https://keras.io/api/ops/nn#maxpool-function" class="external-link">https://keras.io/api/ops/nn#maxpool-function</a></p></li>
-</ul><p>Other nn ops: <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br></p>
-<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
+</ul><p>Other nn ops: <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_layer_normalization.html">op_layer_normalization</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparse_sigmoid.html">op_sparse_sigmoid</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br></p>
+<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_layer_normalization.html">op_layer_normalization</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparse_sigmoid.html">op_sparse_sigmoid</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -133,7 +133,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/op_maximum.html b/docs/dev/reference/op_maximum.html
index 7a2d986f5..36c723a71 100644
--- a/docs/dev/reference/op_maximum.html
+++ b/docs/dev/reference/op_maximum.html
@@ -8,7 +8,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -84,8 +84,8 @@ <h2 id="value">Value<a class="anchor" aria-label="anchor" href="#value"></a></h2
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
     <div class="dont-index">
 <ul><li><p><a href="https://keras.io/api/ops/numpy#maximum-function" class="external-link">https://keras.io/api/ops/numpy#maximum-function</a></p></li>
-</ul><p>Other numpy ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p>
-<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
+</ul><p>Other numpy ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p>
+<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_layer_normalization.html">op_layer_normalization</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparse_sigmoid.html">op_sparse_sigmoid</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -97,7 +97,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/op_mean.html b/docs/dev/reference/op_mean.html
index ad4eef01c..27cea97c0 100644
--- a/docs/dev/reference/op_mean.html
+++ b/docs/dev/reference/op_mean.html
@@ -8,7 +8,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -88,8 +88,8 @@ <h2 id="value">Value<a class="anchor" aria-label="anchor" href="#value"></a></h2
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
     <div class="dont-index">
 <ul><li><p><a href="https://keras.io/api/ops/numpy#mean-function" class="external-link">https://keras.io/api/ops/numpy#mean-function</a></p></li>
-</ul><p>Other numpy ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p>
-<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
+</ul><p>Other numpy ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p>
+<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_layer_normalization.html">op_layer_normalization</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparse_sigmoid.html">op_sparse_sigmoid</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -101,7 +101,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/op_median.html b/docs/dev/reference/op_median.html
index 531d26742..2d1d29edf 100644
--- a/docs/dev/reference/op_median.html
+++ b/docs/dev/reference/op_median.html
@@ -8,7 +8,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -87,8 +87,8 @@ <h2 id="value">Value<a class="anchor" aria-label="anchor" href="#value"></a></h2
     </div>
     <div class="section level2">
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
-    <div class="dont-index"><p>Other numpy ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p>
-<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
+    <div class="dont-index"><p>Other numpy ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p>
+<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_layer_normalization.html">op_layer_normalization</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparse_sigmoid.html">op_sparse_sigmoid</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -100,7 +100,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/op_meshgrid.html b/docs/dev/reference/op_meshgrid.html
index 71d772193..0297947dc 100644
--- a/docs/dev/reference/op_meshgrid.html
+++ b/docs/dev/reference/op_meshgrid.html
@@ -14,7 +14,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -116,8 +116,8 @@ <h2 id="examples">Examples<a class="anchor" aria-label="anchor" href="#examples"
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
     <div class="dont-index">
 <ul><li><p><a href="https://keras.io/api/ops/numpy#meshgrid-function" class="external-link">https://keras.io/api/ops/numpy#meshgrid-function</a></p></li>
-</ul><p>Other numpy ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p>
-<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
+</ul><p>Other numpy ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p>
+<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_layer_normalization.html">op_layer_normalization</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparse_sigmoid.html">op_sparse_sigmoid</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -129,7 +129,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/op_min.html b/docs/dev/reference/op_min.html
index 0d6ddaffe..6ea321036 100644
--- a/docs/dev/reference/op_min.html
+++ b/docs/dev/reference/op_min.html
@@ -8,7 +8,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -109,8 +109,8 @@ <h2 id="examples">Examples<a class="anchor" aria-label="anchor" href="#examples"
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
     <div class="dont-index">
 <ul><li><p><a href="https://keras.io/api/ops/numpy#min-function" class="external-link">https://keras.io/api/ops/numpy#min-function</a></p></li>
-</ul><p>Other numpy ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p>
-<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
+</ul><p>Other numpy ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p>
+<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_layer_normalization.html">op_layer_normalization</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparse_sigmoid.html">op_sparse_sigmoid</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -122,7 +122,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/op_minimum.html b/docs/dev/reference/op_minimum.html
index e6ca91a0f..76a648600 100644
--- a/docs/dev/reference/op_minimum.html
+++ b/docs/dev/reference/op_minimum.html
@@ -8,7 +8,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -84,8 +84,8 @@ <h2 id="value">Value<a class="anchor" aria-label="anchor" href="#value"></a></h2
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
     <div class="dont-index">
 <ul><li><p><a href="https://keras.io/api/ops/numpy#minimum-function" class="external-link">https://keras.io/api/ops/numpy#minimum-function</a></p></li>
-</ul><p>Other numpy ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p>
-<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
+</ul><p>Other numpy ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p>
+<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_layer_normalization.html">op_layer_normalization</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparse_sigmoid.html">op_sparse_sigmoid</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -97,7 +97,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/op_mod.html b/docs/dev/reference/op_mod.html
index e421cb02e..075589d80 100644
--- a/docs/dev/reference/op_mod.html
+++ b/docs/dev/reference/op_mod.html
@@ -38,7 +38,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -121,8 +121,8 @@ <h2 id="value">Value<a class="anchor" aria-label="anchor" href="#value"></a></h2
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
     <div class="dont-index">
 <ul><li><p><a href="https://keras.io/api/ops/numpy#mod-function" class="external-link">https://keras.io/api/ops/numpy#mod-function</a></p></li>
-</ul><p>Other numpy ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p>
-<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
+</ul><p>Other numpy ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p>
+<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_layer_normalization.html">op_layer_normalization</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparse_sigmoid.html">op_sparse_sigmoid</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -134,7 +134,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/op_moments.html b/docs/dev/reference/op_moments.html
index 8bec38b25..0f9222a35 100644
--- a/docs/dev/reference/op_moments.html
+++ b/docs/dev/reference/op_moments.html
@@ -12,7 +12,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -110,8 +110,8 @@ <h2 id="examples">Examples<a class="anchor" aria-label="anchor" href="#examples"
     </div>
     <div class="section level2">
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
-    <div class="dont-index"><p>Other nn ops: <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br></p>
-<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
+    <div class="dont-index"><p>Other nn ops: <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_layer_normalization.html">op_layer_normalization</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparse_sigmoid.html">op_sparse_sigmoid</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br></p>
+<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_layer_normalization.html">op_layer_normalization</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparse_sigmoid.html">op_sparse_sigmoid</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -123,7 +123,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/op_moveaxis.html b/docs/dev/reference/op_moveaxis.html
index 645f5e6dc..014f25e2b 100644
--- a/docs/dev/reference/op_moveaxis.html
+++ b/docs/dev/reference/op_moveaxis.html
@@ -8,7 +8,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -87,8 +87,8 @@ <h2 id="value">Value<a class="anchor" aria-label="anchor" href="#value"></a></h2
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
     <div class="dont-index">
 <ul><li><p><a href="https://keras.io/api/ops/numpy#moveaxis-function" class="external-link">https://keras.io/api/ops/numpy#moveaxis-function</a></p></li>
-</ul><p>Other numpy ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p>
-<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
+</ul><p>Other numpy ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p>
+<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_layer_normalization.html">op_layer_normalization</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparse_sigmoid.html">op_sparse_sigmoid</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -100,7 +100,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/op_multi_hot.html b/docs/dev/reference/op_multi_hot.html
index 0bd9d26e3..3178d887d 100644
--- a/docs/dev/reference/op_multi_hot.html
+++ b/docs/dev/reference/op_multi_hot.html
@@ -10,7 +10,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -116,8 +116,8 @@ <h2 id="examples">Examples<a class="anchor" aria-label="anchor" href="#examples"
     </div>
     <div class="section level2">
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
-    <div class="dont-index"><p>Other nn ops: <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br></p>
-<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
+    <div class="dont-index"><p>Other nn ops: <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_layer_normalization.html">op_layer_normalization</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparse_sigmoid.html">op_sparse_sigmoid</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br></p>
+<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_layer_normalization.html">op_layer_normalization</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparse_sigmoid.html">op_sparse_sigmoid</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -129,7 +129,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/op_multiply.html b/docs/dev/reference/op_multiply.html
index e3adcb5be..5cd778ff7 100644
--- a/docs/dev/reference/op_multiply.html
+++ b/docs/dev/reference/op_multiply.html
@@ -38,7 +38,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -121,8 +121,8 @@ <h2 id="value">Value<a class="anchor" aria-label="anchor" href="#value"></a></h2
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
     <div class="dont-index">
 <ul><li><p><a href="https://keras.io/api/ops/numpy#multiply-function" class="external-link">https://keras.io/api/ops/numpy#multiply-function</a></p></li>
-</ul><p>Other numpy ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p>
-<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
+</ul><p>Other numpy ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p>
+<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_layer_normalization.html">op_layer_normalization</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparse_sigmoid.html">op_sparse_sigmoid</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -134,7 +134,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/op_nan_to_num.html b/docs/dev/reference/op_nan_to_num.html
index 1b7fa9eb6..8844d1259 100644
--- a/docs/dev/reference/op_nan_to_num.html
+++ b/docs/dev/reference/op_nan_to_num.html
@@ -8,7 +8,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -102,8 +102,8 @@ <h2 id="example">Example<a class="anchor" aria-label="anchor" href="#example"></
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
     <div class="dont-index">
 <ul><li><p><a href="https://keras.io/api/ops/numpy#nantonum-function" class="external-link">https://keras.io/api/ops/numpy#nantonum-function</a></p></li>
-</ul><p>Other numpy ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p>
-<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
+</ul><p>Other numpy ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p>
+<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_layer_normalization.html">op_layer_normalization</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparse_sigmoid.html">op_sparse_sigmoid</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -115,7 +115,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/op_ndim.html b/docs/dev/reference/op_ndim.html
index d95278246..ef5a8436c 100644
--- a/docs/dev/reference/op_ndim.html
+++ b/docs/dev/reference/op_ndim.html
@@ -8,7 +8,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -78,8 +78,8 @@ <h2 id="value">Value<a class="anchor" aria-label="anchor" href="#value"></a></h2
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
     <div class="dont-index">
 <ul><li><p><a href="https://keras.io/api/ops/numpy#ndim-function" class="external-link">https://keras.io/api/ops/numpy#ndim-function</a></p></li>
-</ul><p>Other numpy ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p>
-<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
+</ul><p>Other numpy ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p>
+<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_layer_normalization.html">op_layer_normalization</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparse_sigmoid.html">op_sparse_sigmoid</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -91,7 +91,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/op_negative.html b/docs/dev/reference/op_negative.html
index 1c347fbef..a861ae41a 100644
--- a/docs/dev/reference/op_negative.html
+++ b/docs/dev/reference/op_negative.html
@@ -38,7 +38,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -117,8 +117,8 @@ <h2 id="value">Value<a class="anchor" aria-label="anchor" href="#value"></a></h2
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
     <div class="dont-index">
 <ul><li><p><a href="https://keras.io/api/ops/numpy#negative-function" class="external-link">https://keras.io/api/ops/numpy#negative-function</a></p></li>
-</ul><p>Other numpy ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p>
-<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
+</ul><p>Other numpy ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p>
+<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_layer_normalization.html">op_layer_normalization</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparse_sigmoid.html">op_sparse_sigmoid</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -130,7 +130,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/op_nonzero.html b/docs/dev/reference/op_nonzero.html
index 469b82960..4bff5a699 100644
--- a/docs/dev/reference/op_nonzero.html
+++ b/docs/dev/reference/op_nonzero.html
@@ -8,7 +8,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -122,8 +122,8 @@ <h3 id="example">Example<a class="anchor" aria-label="anchor" href="#example"></
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
     <div class="dont-index">
 <ul><li><p><a href="https://keras.io/api/ops/numpy#nonzero-function" class="external-link">https://keras.io/api/ops/numpy#nonzero-function</a></p></li>
-</ul><p>Other numpy ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p>
-<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
+</ul><p>Other numpy ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p>
+<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_layer_normalization.html">op_layer_normalization</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparse_sigmoid.html">op_sparse_sigmoid</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -135,7 +135,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/op_norm.html b/docs/dev/reference/op_norm.html
index e0ce8160a..386dbe486 100644
--- a/docs/dev/reference/op_norm.html
+++ b/docs/dev/reference/op_norm.html
@@ -12,7 +12,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -136,7 +136,7 @@ <h2 id="examples">Examples<a class="anchor" aria-label="anchor" href="#examples"
     <div class="section level2">
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
     <div class="dont-index"><p>Other linear algebra ops: <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br></p>
-<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
+<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_layer_normalization.html">op_layer_normalization</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparse_sigmoid.html">op_sparse_sigmoid</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -148,7 +148,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/op_normalize.html b/docs/dev/reference/op_normalize.html
index e8878a224..78f929037 100644
--- a/docs/dev/reference/op_normalize.html
+++ b/docs/dev/reference/op_normalize.html
@@ -8,7 +8,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -103,8 +103,8 @@ <h2 id="examples">Examples<a class="anchor" aria-label="anchor" href="#examples"
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
     <div class="dont-index">
 <ul><li><p><a href="https://www.tensorflow.org/api_docs/python/tf/keras/ops/normalize" class="external-link">https://www.tensorflow.org/api_docs/python/tf/keras/ops/normalize</a></p></li>
-</ul><p>Other nn ops: <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br></p>
-<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
+</ul><p>Other nn ops: <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_layer_normalization.html">op_layer_normalization</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparse_sigmoid.html">op_sparse_sigmoid</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br></p>
+<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_layer_normalization.html">op_layer_normalization</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparse_sigmoid.html">op_sparse_sigmoid</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -116,7 +116,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/op_not_equal.html b/docs/dev/reference/op_not_equal.html
index d2e4d930f..0c8cc2055 100644
--- a/docs/dev/reference/op_not_equal.html
+++ b/docs/dev/reference/op_not_equal.html
@@ -38,7 +38,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -121,8 +121,8 @@ <h2 id="value">Value<a class="anchor" aria-label="anchor" href="#value"></a></h2
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
     <div class="dont-index">
 <ul><li><p><a href="https://keras.io/api/ops/numpy#notequal-function" class="external-link">https://keras.io/api/ops/numpy#notequal-function</a></p></li>
-</ul><p>Other numpy ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p>
-<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
+</ul><p>Other numpy ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p>
+<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_layer_normalization.html">op_layer_normalization</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparse_sigmoid.html">op_sparse_sigmoid</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -134,7 +134,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/op_one_hot.html b/docs/dev/reference/op_one_hot.html
index 1eb7ef811..5dfe2a61b 100644
--- a/docs/dev/reference/op_one_hot.html
+++ b/docs/dev/reference/op_one_hot.html
@@ -14,7 +14,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -143,8 +143,8 @@ <h2 id="examples">Examples<a class="anchor" aria-label="anchor" href="#examples"
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
     <div class="dont-index">
 <ul><li><p><a href="https://keras.io/api/ops/nn#onehot-function" class="external-link">https://keras.io/api/ops/nn#onehot-function</a></p></li>
-</ul><p>Other nn ops: <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br></p>
-<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
+</ul><p>Other nn ops: <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_layer_normalization.html">op_layer_normalization</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparse_sigmoid.html">op_sparse_sigmoid</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br></p>
+<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_layer_normalization.html">op_layer_normalization</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparse_sigmoid.html">op_sparse_sigmoid</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -156,7 +156,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/op_ones.html b/docs/dev/reference/op_ones.html
index 65fe9de90..1d624ba6c 100644
--- a/docs/dev/reference/op_ones.html
+++ b/docs/dev/reference/op_ones.html
@@ -8,7 +8,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -82,8 +82,8 @@ <h2 id="value">Value<a class="anchor" aria-label="anchor" href="#value"></a></h2
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
     <div class="dont-index">
 <ul><li><p><a href="https://keras.io/api/ops/numpy#ones-function" class="external-link">https://keras.io/api/ops/numpy#ones-function</a></p></li>
-</ul><p>Other numpy ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p>
-<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
+</ul><p>Other numpy ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p>
+<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_layer_normalization.html">op_layer_normalization</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparse_sigmoid.html">op_sparse_sigmoid</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -95,7 +95,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/op_ones_like.html b/docs/dev/reference/op_ones_like.html
index 689beec5f..799f46bb5 100644
--- a/docs/dev/reference/op_ones_like.html
+++ b/docs/dev/reference/op_ones_like.html
@@ -8,7 +8,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -82,8 +82,8 @@ <h2 id="value">Value<a class="anchor" aria-label="anchor" href="#value"></a></h2
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
     <div class="dont-index">
 <ul><li><p><a href="https://keras.io/api/ops/numpy#oneslike-function" class="external-link">https://keras.io/api/ops/numpy#oneslike-function</a></p></li>
-</ul><p>Other numpy ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p>
-<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
+</ul><p>Other numpy ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p>
+<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_layer_normalization.html">op_layer_normalization</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparse_sigmoid.html">op_sparse_sigmoid</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -95,7 +95,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/op_outer.html b/docs/dev/reference/op_outer.html
index e28dbd634..2ec8dcc68 100644
--- a/docs/dev/reference/op_outer.html
+++ b/docs/dev/reference/op_outer.html
@@ -12,7 +12,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -87,8 +87,8 @@ <h2 id="value">Value<a class="anchor" aria-label="anchor" href="#value"></a></h2
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
     <div class="dont-index">
 <ul><li><p><a href="https://keras.io/api/ops/numpy#outer-function" class="external-link">https://keras.io/api/ops/numpy#outer-function</a></p></li>
-</ul><p>Other numpy ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p>
-<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
+</ul><p>Other numpy ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p>
+<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_layer_normalization.html">op_layer_normalization</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparse_sigmoid.html">op_sparse_sigmoid</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -100,7 +100,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/op_pad.html b/docs/dev/reference/op_pad.html
index 3304f0e6c..c4ea24d53 100644
--- a/docs/dev/reference/op_pad.html
+++ b/docs/dev/reference/op_pad.html
@@ -8,7 +8,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -110,8 +110,8 @@ <h2 id="note">Note<a class="anchor" aria-label="anchor" href="#note"></a></h2>
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
     <div class="dont-index">
 <ul><li><p><a href="https://keras.io/api/ops/numpy#pad-function" class="external-link">https://keras.io/api/ops/numpy#pad-function</a></p></li>
-</ul><p>Other numpy ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p>
-<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
+</ul><p>Other numpy ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p>
+<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_layer_normalization.html">op_layer_normalization</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparse_sigmoid.html">op_sparse_sigmoid</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -123,7 +123,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/op_polar.html b/docs/dev/reference/op_polar.html
index 7f6148df8..f839d007f 100644
--- a/docs/dev/reference/op_polar.html
+++ b/docs/dev/reference/op_polar.html
@@ -28,7 +28,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -119,8 +119,8 @@ <h2 id="examples">Examples<a class="anchor" aria-label="anchor" href="#examples"
     </div>
     <div class="section level2">
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
-    <div class="dont-index"><p>Other nn ops: <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br></p>
-<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
+    <div class="dont-index"><p>Other nn ops: <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_layer_normalization.html">op_layer_normalization</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparse_sigmoid.html">op_sparse_sigmoid</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br></p>
+<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_layer_normalization.html">op_layer_normalization</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparse_sigmoid.html">op_sparse_sigmoid</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -132,7 +132,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/op_power.html b/docs/dev/reference/op_power.html
index b242f0378..1f1a033f4 100644
--- a/docs/dev/reference/op_power.html
+++ b/docs/dev/reference/op_power.html
@@ -38,7 +38,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -121,8 +121,8 @@ <h2 id="value">Value<a class="anchor" aria-label="anchor" href="#value"></a></h2
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
     <div class="dont-index">
 <ul><li><p><a href="https://keras.io/api/ops/numpy#power-function" class="external-link">https://keras.io/api/ops/numpy#power-function</a></p></li>
-</ul><p>Other numpy ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p>
-<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
+</ul><p>Other numpy ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p>
+<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_layer_normalization.html">op_layer_normalization</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparse_sigmoid.html">op_sparse_sigmoid</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -134,7 +134,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/op_prod.html b/docs/dev/reference/op_prod.html
index 5e39db249..fb5655d68 100644
--- a/docs/dev/reference/op_prod.html
+++ b/docs/dev/reference/op_prod.html
@@ -8,7 +8,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -93,8 +93,8 @@ <h2 id="value">Value<a class="anchor" aria-label="anchor" href="#value"></a></h2
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
     <div class="dont-index">
 <ul><li><p><a href="https://keras.io/api/ops/numpy#prod-function" class="external-link">https://keras.io/api/ops/numpy#prod-function</a></p></li>
-</ul><p>Other numpy ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p>
-<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
+</ul><p>Other numpy ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p>
+<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_layer_normalization.html">op_layer_normalization</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparse_sigmoid.html">op_sparse_sigmoid</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -106,7 +106,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/op_psnr.html b/docs/dev/reference/op_psnr.html
index 2fd2c8d34..8be890299 100644
--- a/docs/dev/reference/op_psnr.html
+++ b/docs/dev/reference/op_psnr.html
@@ -16,7 +16,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -105,8 +105,8 @@ <h2 id="examples">Examples<a class="anchor" aria-label="anchor" href="#examples"
     </div>
     <div class="section level2">
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
-    <div class="dont-index"><p>Other nn ops: <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br></p>
-<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
+    <div class="dont-index"><p>Other nn ops: <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_layer_normalization.html">op_layer_normalization</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparse_sigmoid.html">op_sparse_sigmoid</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br></p>
+<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_layer_normalization.html">op_layer_normalization</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparse_sigmoid.html">op_sparse_sigmoid</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -118,7 +118,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/op_qr.html b/docs/dev/reference/op_qr.html
index 9c80e1320..59596bdbf 100644
--- a/docs/dev/reference/op_qr.html
+++ b/docs/dev/reference/op_qr.html
@@ -8,7 +8,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -104,7 +104,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
     <div class="dont-index">
 <ul><li><p><a href="https://keras.io/api/ops/core#qr-function" class="external-link">https://keras.io/api/ops/core#qr-function</a></p></li>
 </ul><p>Other math ops: <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br></p>
-<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
+<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_layer_normalization.html">op_layer_normalization</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparse_sigmoid.html">op_sparse_sigmoid</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -116,7 +116,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/op_quantile.html b/docs/dev/reference/op_quantile.html
index 0b299c575..9e7b3bd0e 100644
--- a/docs/dev/reference/op_quantile.html
+++ b/docs/dev/reference/op_quantile.html
@@ -8,7 +8,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -108,8 +108,8 @@ <h2 id="value">Value<a class="anchor" aria-label="anchor" href="#value"></a></h2
     </div>
     <div class="section level2">
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
-    <div class="dont-index"><p>Other numpy ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p>
-<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
+    <div class="dont-index"><p>Other numpy ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p>
+<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_layer_normalization.html">op_layer_normalization</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparse_sigmoid.html">op_sparse_sigmoid</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -121,7 +121,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/op_ravel.html b/docs/dev/reference/op_ravel.html
index 091eeeffc..71a2723ee 100644
--- a/docs/dev/reference/op_ravel.html
+++ b/docs/dev/reference/op_ravel.html
@@ -8,7 +8,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -78,8 +78,8 @@ <h2 id="value">Value<a class="anchor" aria-label="anchor" href="#value"></a></h2
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
     <div class="dont-index">
 <ul><li><p><a href="https://keras.io/api/ops/numpy#ravel-function" class="external-link">https://keras.io/api/ops/numpy#ravel-function</a></p></li>
-</ul><p>Other numpy ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p>
-<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
+</ul><p>Other numpy ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p>
+<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_layer_normalization.html">op_layer_normalization</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparse_sigmoid.html">op_sparse_sigmoid</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -91,7 +91,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/op_real.html b/docs/dev/reference/op_real.html
index dac8d73e4..a658e84d9 100644
--- a/docs/dev/reference/op_real.html
+++ b/docs/dev/reference/op_real.html
@@ -8,7 +8,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -78,8 +78,8 @@ <h2 id="value">Value<a class="anchor" aria-label="anchor" href="#value"></a></h2
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
     <div class="dont-index">
 <ul><li><p><a href="https://keras.io/api/ops/numpy#real-function" class="external-link">https://keras.io/api/ops/numpy#real-function</a></p></li>
-</ul><p>Other numpy ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p>
-<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
+</ul><p>Other numpy ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p>
+<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_layer_normalization.html">op_layer_normalization</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparse_sigmoid.html">op_sparse_sigmoid</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -91,7 +91,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/op_rearrange.html b/docs/dev/reference/op_rearrange.html
index c12913643..d57ebb2d5 100644
--- a/docs/dev/reference/op_rearrange.html
+++ b/docs/dev/reference/op_rearrange.html
@@ -8,7 +8,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -116,7 +116,7 @@ <h2 id="value">Value<a class="anchor" aria-label="anchor" href="#value"></a></h2
     <div class="section level2">
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
     <div class="dont-index"><p>Other core ops: <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br></p>
-<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
+<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_layer_normalization.html">op_layer_normalization</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparse_sigmoid.html">op_sparse_sigmoid</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -128,7 +128,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/op_reciprocal.html b/docs/dev/reference/op_reciprocal.html
index 0ad0f130a..8dc25545f 100644
--- a/docs/dev/reference/op_reciprocal.html
+++ b/docs/dev/reference/op_reciprocal.html
@@ -8,7 +8,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -78,8 +78,8 @@ <h2 id="value">Value<a class="anchor" aria-label="anchor" href="#value"></a></h2
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
     <div class="dont-index">
 <ul><li><p><a href="https://keras.io/api/ops/numpy#reciprocal-function" class="external-link">https://keras.io/api/ops/numpy#reciprocal-function</a></p></li>
-</ul><p>Other numpy ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p>
-<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
+</ul><p>Other numpy ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p>
+<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_layer_normalization.html">op_layer_normalization</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparse_sigmoid.html">op_sparse_sigmoid</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -91,7 +91,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/op_relu.html b/docs/dev/reference/op_relu.html
index b1901476d..28652c335 100644
--- a/docs/dev/reference/op_relu.html
+++ b/docs/dev/reference/op_relu.html
@@ -8,7 +8,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -88,8 +88,8 @@ <h2 id="examples">Examples<a class="anchor" aria-label="anchor" href="#examples"
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
     <div class="dont-index">
 <ul><li><p><a href="https://keras.io/api/ops/nn#relu-function" class="external-link">https://keras.io/api/ops/nn#relu-function</a></p></li>
-</ul><p>Other nn ops: <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br></p>
-<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
+</ul><p>Other nn ops: <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_layer_normalization.html">op_layer_normalization</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparse_sigmoid.html">op_sparse_sigmoid</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br></p>
+<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_layer_normalization.html">op_layer_normalization</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparse_sigmoid.html">op_sparse_sigmoid</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -101,7 +101,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/op_relu6.html b/docs/dev/reference/op_relu6.html
index 997cb6773..15d32a171 100644
--- a/docs/dev/reference/op_relu6.html
+++ b/docs/dev/reference/op_relu6.html
@@ -8,7 +8,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -88,8 +88,8 @@ <h2 id="examples">Examples<a class="anchor" aria-label="anchor" href="#examples"
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
     <div class="dont-index">
 <ul><li><p><a href="https://keras.io/api/ops/nn#relu6-function" class="external-link">https://keras.io/api/ops/nn#relu6-function</a></p></li>
-</ul><p>Other nn ops: <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br></p>
-<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
+</ul><p>Other nn ops: <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_layer_normalization.html">op_layer_normalization</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparse_sigmoid.html">op_sparse_sigmoid</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br></p>
+<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_layer_normalization.html">op_layer_normalization</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparse_sigmoid.html">op_sparse_sigmoid</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -101,7 +101,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/op_repeat.html b/docs/dev/reference/op_repeat.html
index e5da7e1ac..c20e5cda0 100644
--- a/docs/dev/reference/op_repeat.html
+++ b/docs/dev/reference/op_repeat.html
@@ -8,7 +8,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -87,8 +87,8 @@ <h2 id="value">Value<a class="anchor" aria-label="anchor" href="#value"></a></h2
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
     <div class="dont-index">
 <ul><li><p><a href="https://keras.io/api/ops/numpy#repeat-function" class="external-link">https://keras.io/api/ops/numpy#repeat-function</a></p></li>
-</ul><p>Other numpy ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p>
-<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
+</ul><p>Other numpy ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p>
+<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_layer_normalization.html">op_layer_normalization</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparse_sigmoid.html">op_sparse_sigmoid</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -100,7 +100,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/op_reshape.html b/docs/dev/reference/op_reshape.html
index 52b88bfc0..58b9adda7 100644
--- a/docs/dev/reference/op_reshape.html
+++ b/docs/dev/reference/op_reshape.html
@@ -8,7 +8,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -84,8 +84,8 @@ <h2 id="value">Value<a class="anchor" aria-label="anchor" href="#value"></a></h2
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
     <div class="dont-index">
 <ul><li><p><a href="https://keras.io/api/ops/numpy#reshape-function" class="external-link">https://keras.io/api/ops/numpy#reshape-function</a></p></li>
-</ul><p>Other numpy ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p>
-<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
+</ul><p>Other numpy ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p>
+<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_layer_normalization.html">op_layer_normalization</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparse_sigmoid.html">op_sparse_sigmoid</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -97,7 +97,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/op_rfft.html b/docs/dev/reference/op_rfft.html
index e0009e349..0ec844af7 100644
--- a/docs/dev/reference/op_rfft.html
+++ b/docs/dev/reference/op_rfft.html
@@ -24,7 +24,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -128,7 +128,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
     <div class="dont-index">
 <ul><li><p><a href="https://keras.io/api/ops/fft#rfft-function" class="external-link">https://keras.io/api/ops/fft#rfft-function</a></p></li>
 </ul><p>Other math ops: <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br></p>
-<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
+<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_layer_normalization.html">op_layer_normalization</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparse_sigmoid.html">op_sparse_sigmoid</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -140,7 +140,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/op_right_shift.html b/docs/dev/reference/op_right_shift.html
index f2defe8eb..55fb3421a 100644
--- a/docs/dev/reference/op_right_shift.html
+++ b/docs/dev/reference/op_right_shift.html
@@ -12,7 +12,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -86,8 +86,8 @@ <h2 id="value">Value<a class="anchor" aria-label="anchor" href="#value"></a></h2
     </div>
     <div class="section level2">
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
-    <div class="dont-index"><p>Other numpy ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p>
-<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
+    <div class="dont-index"><p>Other numpy ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p>
+<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_layer_normalization.html">op_layer_normalization</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparse_sigmoid.html">op_sparse_sigmoid</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -99,7 +99,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/op_rms_normalization.html b/docs/dev/reference/op_rms_normalization.html
index 5c3e9ab5b..ccd51a1c6 100644
--- a/docs/dev/reference/op_rms_normalization.html
+++ b/docs/dev/reference/op_rms_normalization.html
@@ -16,7 +16,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -71,7 +71,7 @@ <h1>Performs Root Mean Square (RMS) normalization on <code>x</code>.</h1>
 
     <div class="section level2">
     <h2 id="ref-usage">Usage<a class="anchor" aria-label="anchor" href="#ref-usage"></a></h2>
-    <div class="sourceCode"><pre class="sourceCode r"><code><span><span class="fu">op_rms_normalization</span><span class="op">(</span><span class="va">x</span>, scale <span class="op">=</span> <span class="fl">1L</span>, axis <span class="op">=</span> <span class="op">-</span><span class="fl">1L</span>, epsilon <span class="op">=</span> <span class="cn">NULL</span><span class="op">)</span></span></code></pre></div>
+    <div class="sourceCode"><pre class="sourceCode r"><code><span><span class="fu">op_rms_normalization</span><span class="op">(</span><span class="va">x</span>, scale <span class="op">=</span> <span class="cn">NULL</span>, axis <span class="op">=</span> <span class="op">-</span><span class="fl">1L</span>, epsilon <span class="op">=</span> <span class="cn">NULL</span><span class="op">)</span></span></code></pre></div>
     </div>
 
     <div class="section level2">
@@ -87,8 +87,7 @@ <h2 id="arguments">Arguments<a class="anchor" aria-label="anchor" href="#argumen
 
 
 <dt id="arg-axis">axis<a class="anchor" aria-label="anchor" href="#arg-axis"></a></dt>
-<dd><p>The axis or axes along which to perform normalization.
-Default to -1.</p></dd>
+<dd><p>The axis or axes along which to perform normalization. Defaults to <code>-1</code>.</p></dd>
 
 
 <dt id="arg-epsilon">epsilon<a class="anchor" aria-label="anchor" href="#arg-epsilon"></a></dt>
@@ -102,14 +101,35 @@ <h2 id="value">Value<a class="anchor" aria-label="anchor" href="#value"></a></h2
     </div>
     <div class="section level2">
     <h2 id="examples">Examples<a class="anchor" aria-label="anchor" href="#examples"></a></h2>
-    <p></p><div class="sourceCode python"><pre><code><span id="cb1-1"><a href="#cb1-1" tabindex="-1"></a>x <span class="op">&lt;-</span> random_uniform(c(<span class="dv">1</span>, <span class="dv">10</span>))</span>
-<span id="cb1-2"><a href="#cb1-2" tabindex="-1"></a>x_norm <span class="op">&lt;-</span> op_rms_normalization(x, scale <span class="op">=</span> <span class="dv">10</span>)</span>
-<span id="cb1-3"><a href="#cb1-3" tabindex="-1"></a>x_norm</span></code></pre><p></p></div>
+    <p></p><div class="sourceCode r"><pre><code><span><span class="va">x</span> <span class="op">&lt;-</span> <span class="fu"><a href="../reference/random_normal.html">random_normal</a></span><span class="op">(</span><span class="fu"><a href="https://rdrr.io/r/base/c.html" class="external-link">c</a></span><span class="op">(</span><span class="fl">1</span>, <span class="fl">10</span><span class="op">)</span><span class="op">)</span></span>
+<span><span class="fu"><a href="../reference/op_rms_normalization.html">op_rms_normalization</a></span><span class="op">(</span><span class="va">x</span><span class="op">)</span></span></code></pre><p></p></div>
+<p></p><div class="sourceCode"><pre><code><span id="cb1-1"><a href="#cb1-1" tabindex="-1"></a><span class="do">## tf.Tensor(</span></span>
+<span id="cb1-2"><a href="#cb1-2" tabindex="-1"></a><span class="do">## [[-1.1701705   0.9987454   0.27809578 -0.18452731  1.5731977   1.8934134</span></span>
+<span id="cb1-3"><a href="#cb1-3" tabindex="-1"></a><span class="do">##    0.10656697  1.0256069   0.6185074  -0.12678868]], shape=(1, 10), dtype=float32)</span></span>
+<span id="cb1-4"><a href="#cb1-4" tabindex="-1"></a></span></code></pre><p></p></div>
+<p></p><div class="sourceCode r"><pre><code><span><span class="fu"><a href="../reference/op_rms_normalization.html">op_rms_normalization</a></span><span class="op">(</span><span class="va">x</span>, scale <span class="op">=</span> <span class="fl">10</span><span class="op">)</span></span></code></pre><p></p></div>
+<p></p><div class="sourceCode"><pre><code><span id="cb1-1"><a href="#cb1-1" tabindex="-1"></a><span class="do">## tf.Tensor(</span></span>
+<span id="cb1-2"><a href="#cb1-2" tabindex="-1"></a><span class="do">## [[-11.701706    9.987453    2.7809577  -1.845273   15.731977   18.934134</span></span>
+<span id="cb1-3"><a href="#cb1-3" tabindex="-1"></a><span class="do">##     1.0656698  10.256069    6.185074   -1.2678868]], shape=(1, 10), dtype=float32)</span></span>
+<span id="cb1-4"><a href="#cb1-4" tabindex="-1"></a></span></code></pre><p></p></div>
+<p>[-1.1701705   0.9987454   0.27809578 -0.18452731  1.5731977   1.8934134</p><div class="section">
+<h3 id="-r-a-">0.10656697  1.0256069   0.6185074  -0.12678868]: R:-1.1701705%20%20%200.9987454%20%20%200.27809578%20-0.18452731%20%201.5731977%20%20%201.8934134%0A##%20%20%20%200.10656697%20%201.0256069%20%20%200.6185074%20%20-0.12678868<a class="anchor" aria-label="anchor" href="#-r-a-"></a></h3>
+
+
+<p>[-11.701706    9.987453    2.7809577  -1.845273   15.731977   18.934134</p>
+</div>
+
+<div class="section">
+<h3 id="-r-a--1">1.0656698  10.256069    6.185074   -1.2678868]: R:-11.701706%20%20%20%209.987453%20%20%20%202.7809577%20%20-1.845273%20%20%2015.731977%20%20%2018.934134%0A##%20%20%20%20%201.0656698%20%2010.256069%20%20%20%206.185074%20%20%20-1.2678868<a class="anchor" aria-label="anchor" href="#-r-a--1"></a></h3>
+
+
+</div>
+
     </div>
     <div class="section level2">
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
-    <div class="dont-index"><p>Other nn ops: <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br></p>
-<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
+    <div class="dont-index"><p>Other nn ops: <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_layer_normalization.html">op_layer_normalization</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparse_sigmoid.html">op_sparse_sigmoid</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br></p>
+<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_layer_normalization.html">op_layer_normalization</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparse_sigmoid.html">op_sparse_sigmoid</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -121,7 +141,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/op_roll.html b/docs/dev/reference/op_roll.html
index 945c6250a..f741bccc5 100644
--- a/docs/dev/reference/op_roll.html
+++ b/docs/dev/reference/op_roll.html
@@ -8,7 +8,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -88,8 +88,8 @@ <h2 id="value">Value<a class="anchor" aria-label="anchor" href="#value"></a></h2
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
     <div class="dont-index">
 <ul><li><p><a href="https://keras.io/api/ops/numpy#roll-function" class="external-link">https://keras.io/api/ops/numpy#roll-function</a></p></li>
-</ul><p>Other numpy ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p>
-<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
+</ul><p>Other numpy ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p>
+<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_layer_normalization.html">op_layer_normalization</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparse_sigmoid.html">op_sparse_sigmoid</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -101,7 +101,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/op_rot90.html b/docs/dev/reference/op_rot90.html
index 1971791de..6f14c1014 100644
--- a/docs/dev/reference/op_rot90.html
+++ b/docs/dev/reference/op_rot90.html
@@ -12,7 +12,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -122,8 +122,8 @@ <h2 id="examples">Examples<a class="anchor" aria-label="anchor" href="#examples"
     </div>
     <div class="section level2">
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
-    <div class="dont-index"><p>Other numpy ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p>
-<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
+    <div class="dont-index"><p>Other numpy ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p>
+<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_layer_normalization.html">op_layer_normalization</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparse_sigmoid.html">op_sparse_sigmoid</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -135,7 +135,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/op_round.html b/docs/dev/reference/op_round.html
index 0bb415b9e..5a410db21 100644
--- a/docs/dev/reference/op_round.html
+++ b/docs/dev/reference/op_round.html
@@ -8,7 +8,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -82,8 +82,8 @@ <h2 id="value">Value<a class="anchor" aria-label="anchor" href="#value"></a></h2
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
     <div class="dont-index">
 <ul><li><p><a href="https://keras.io/api/ops/numpy#round-function" class="external-link">https://keras.io/api/ops/numpy#round-function</a></p></li>
-</ul><p>Other numpy ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p>
-<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
+</ul><p>Other numpy ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p>
+<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_layer_normalization.html">op_layer_normalization</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparse_sigmoid.html">op_sparse_sigmoid</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -95,7 +95,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/op_rsqrt.html b/docs/dev/reference/op_rsqrt.html
index 4039b8c20..b5b37376f 100644
--- a/docs/dev/reference/op_rsqrt.html
+++ b/docs/dev/reference/op_rsqrt.html
@@ -8,7 +8,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -87,7 +87,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
     <div class="dont-index">
 <ul><li><p><a href="https://keras.io/api/ops/core#rsqrt-function" class="external-link">https://keras.io/api/ops/core#rsqrt-function</a></p></li>
 </ul><p>Other math ops: <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br></p>
-<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
+<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_layer_normalization.html">op_layer_normalization</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparse_sigmoid.html">op_sparse_sigmoid</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -99,7 +99,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/op_saturate_cast.html b/docs/dev/reference/op_saturate_cast.html
index 57c2a5b68..0b6dbfca3 100644
--- a/docs/dev/reference/op_saturate_cast.html
+++ b/docs/dev/reference/op_saturate_cast.html
@@ -16,7 +16,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -127,8 +127,8 @@ <h2 id="examples">Examples<a class="anchor" aria-label="anchor" href="#examples"
     </div>
     <div class="section level2">
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
-    <div class="dont-index"><p>Other numpy ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p>
-<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
+    <div class="dont-index"><p>Other numpy ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p>
+<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_layer_normalization.html">op_layer_normalization</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparse_sigmoid.html">op_sparse_sigmoid</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -140,7 +140,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/op_scan.html b/docs/dev/reference/op_scan.html
index 31fc8087e..23d9454d5 100644
--- a/docs/dev/reference/op_scan.html
+++ b/docs/dev/reference/op_scan.html
@@ -46,7 +46,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -188,7 +188,7 @@ <h2 id="examples">Examples<a class="anchor" aria-label="anchor" href="#examples"
     <div class="section level2">
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
     <div class="dont-index"><p>Other core ops: <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br></p>
-<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
+<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_layer_normalization.html">op_layer_normalization</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparse_sigmoid.html">op_sparse_sigmoid</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -200,7 +200,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/op_scatter.html b/docs/dev/reference/op_scatter.html
index afc889f2a..e55cc63fb 100644
--- a/docs/dev/reference/op_scatter.html
+++ b/docs/dev/reference/op_scatter.html
@@ -14,7 +14,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -112,7 +112,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
     <div class="dont-index">
 <ul><li><p><a href="https://keras.io/api/ops/core#scatter-function" class="external-link">https://keras.io/api/ops/core#scatter-function</a></p></li>
 </ul><p>Other core ops: <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br></p>
-<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
+<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_layer_normalization.html">op_layer_normalization</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparse_sigmoid.html">op_sparse_sigmoid</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -124,7 +124,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/op_scatter_update.html b/docs/dev/reference/op_scatter_update.html
index 8e9673105..c61b7c024 100644
--- a/docs/dev/reference/op_scatter_update.html
+++ b/docs/dev/reference/op_scatter_update.html
@@ -158,7 +158,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -307,7 +307,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
     <div class="dont-index">
 <ul><li><p><a href="https://keras.io/api/ops/core#scatterupdate-function" class="external-link">https://keras.io/api/ops/core#scatterupdate-function</a></p></li>
 </ul><p>Other core ops: <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br></p>
-<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
+<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_layer_normalization.html">op_layer_normalization</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparse_sigmoid.html">op_sparse_sigmoid</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -319,7 +319,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/op_searchsorted.html b/docs/dev/reference/op_searchsorted.html
index 74ad641b7..96ac61b3a 100644
--- a/docs/dev/reference/op_searchsorted.html
+++ b/docs/dev/reference/op_searchsorted.html
@@ -1,7 +1,7 @@
 <!DOCTYPE html>
-<!-- Generated by pkgdown: do not edit by hand --><html lang="en"><head><meta http-equiv="Content-Type" content="text/html; charset=UTF-8"><meta charset="utf-8"><meta http-equiv="X-UA-Compatible" content="IE=edge"><meta name="viewport" content="width=device-width, initial-scale=1, shrink-to-fit=no"><title>Perform a binary search — op_searchsorted • keras3</title><!-- favicons --><link rel="icon" type="image/png" sizes="96x96" href="../favicon-96x96.png"><link rel="icon" type="”image/svg+xml”" href="../favicon.svg"><link rel="apple-touch-icon" sizes="180x180" href="../apple-touch-icon.png"><link rel="icon" sizes="any" href="../favicon.ico"><link rel="manifest" href="../site.webmanifest"><script src="../deps/jquery-3.6.0/jquery-3.6.0.min.js"></script><meta name="viewport" content="width=device-width, initial-scale=1, shrink-to-fit=no"><link href="../deps/bootstrap-5.3.1/bootstrap.min.css" rel="stylesheet"><script src="../deps/bootstrap-5.3.1/bootstrap.bundle.min.js"></script><link href="../deps/font-awesome-6.5.2/css/all.min.css" rel="stylesheet"><link href="../deps/font-awesome-6.5.2/css/v4-shims.min.css" rel="stylesheet"><script src="../deps/headroom-0.11.0/headroom.min.js"></script><script src="../deps/headroom-0.11.0/jQuery.headroom.min.js"></script><script src="../deps/bootstrap-toc-1.0.1/bootstrap-toc.min.js"></script><script src="../deps/clipboard.js-2.0.11/clipboard.min.js"></script><script src="../deps/search-1.0.0/autocomplete.jquery.min.js"></script><script src="../deps/search-1.0.0/fuse.min.js"></script><script src="../deps/search-1.0.0/mark.min.js"></script><!-- pkgdown --><script src="../pkgdown.js"></script><link href="../extra.css" rel="stylesheet"><meta property="og:title" content="Perform a binary search — op_searchsorted"><meta name="description" content="Perform a binary search, returning indices for insertion of values
-into sorted_sequence that maintain the sorting order."><meta property="og:description" content="Perform a binary search, returning indices for insertion of values
-into sorted_sequence that maintain the sorting order."><meta name="robots" content="noindex"><!-- Google Tag Manager --><script>(function(w,d,s,l,i){w[l]=w[l]||[];w[l].push({'gtm.start': new Date().getTime(),event:'gtm.js'});var f=d.getElementsByTagName(s)[0],j=d.createElement(s),dl=l!='dataLayer'?'&l='+l:'';j.async=true;j.src='https://www.googletagmanager.com/gtm.js?id='+i+dl;f.parentNode.insertBefore(j,f);})(window,document,'script','dataLayer','GTM-KHBDBW7');</script><!-- End Google Tag Manager --></head><body>
+<!-- Generated by pkgdown: do not edit by hand --><html lang="en"><head><meta http-equiv="Content-Type" content="text/html; charset=UTF-8"><meta charset="utf-8"><meta http-equiv="X-UA-Compatible" content="IE=edge"><meta name="viewport" content="width=device-width, initial-scale=1, shrink-to-fit=no"><title>Perform a binary search. — op_searchsorted • keras3</title><!-- favicons --><link rel="icon" type="image/png" sizes="96x96" href="../favicon-96x96.png"><link rel="icon" type="”image/svg+xml”" href="../favicon.svg"><link rel="apple-touch-icon" sizes="180x180" href="../apple-touch-icon.png"><link rel="icon" sizes="any" href="../favicon.ico"><link rel="manifest" href="../site.webmanifest"><script src="../deps/jquery-3.6.0/jquery-3.6.0.min.js"></script><meta name="viewport" content="width=device-width, initial-scale=1, shrink-to-fit=no"><link href="../deps/bootstrap-5.3.1/bootstrap.min.css" rel="stylesheet"><script src="../deps/bootstrap-5.3.1/bootstrap.bundle.min.js"></script><link href="../deps/font-awesome-6.5.2/css/all.min.css" rel="stylesheet"><link href="../deps/font-awesome-6.5.2/css/v4-shims.min.css" rel="stylesheet"><script src="../deps/headroom-0.11.0/headroom.min.js"></script><script src="../deps/headroom-0.11.0/jQuery.headroom.min.js"></script><script src="../deps/bootstrap-toc-1.0.1/bootstrap-toc.min.js"></script><script src="../deps/clipboard.js-2.0.11/clipboard.min.js"></script><script src="../deps/search-1.0.0/autocomplete.jquery.min.js"></script><script src="../deps/search-1.0.0/fuse.min.js"></script><script src="../deps/search-1.0.0/mark.min.js"></script><!-- pkgdown --><script src="../pkgdown.js"></script><link href="../extra.css" rel="stylesheet"><meta property="og:title" content="Perform a binary search. — op_searchsorted"><meta name="description" content="Return insertion indices that keep values in sorted order when placed
+into sorted_sequence."><meta property="og:description" content="Return insertion indices that keep values in sorted order when placed
+into sorted_sequence."><meta name="robots" content="noindex"><!-- Google Tag Manager --><script>(function(w,d,s,l,i){w[l]=w[l]||[];w[l].push({'gtm.start': new Date().getTime(),event:'gtm.js'});var f=d.getElementsByTagName(s)[0],j=d.createElement(s),dl=l!='dataLayer'?'&l='+l:'';j.async=true;j.src='https://www.googletagmanager.com/gtm.js?id='+i+dl;f.parentNode.insertBefore(j,f);})(window,document,'script','dataLayer','GTM-KHBDBW7');</script><!-- End Google Tag Manager --></head><body>
     <a href="#main" class="visually-hidden-focusable">Skip to contents</a>
     <!-- Google Tag Manager (noscript) --> <noscript><iframe src="https://www.googletagmanager.com/ns.html?id=GTM-KHBDBW7" height="0" width="0" style="display:none;visibility:hidden"></iframe></noscript> <!-- End Google Tag Manager (noscript) -->
 
@@ -10,7 +10,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -50,14 +50,14 @@
 <div class="row">
   <main id="main" class="col-md-9"><div class="page-header">
 
-      <h1>Perform a binary search</h1>
+      <h1>Perform a binary search.</h1>
       <small class="dont-index">Source: <a href="https://github.com/rstudio/keras3/blob/HEAD/R/ops.R" class="external-link"><code>R/ops.R</code></a></small>
       <div class="d-none name"><code>op_searchsorted.Rd</code></div>
     </div>
 
     <div class="ref-description section level2">
-    <p>Perform a binary search, returning indices for insertion of <code>values</code>
-into <code>sorted_sequence</code> that maintain the sorting order.</p>
+    <p>Return insertion indices that keep <code>values</code> in sorted order when placed
+into <code>sorted_sequence</code>.</p>
     </div>
 
     <div class="section level2">
@@ -70,8 +70,7 @@ <h2 id="arguments">Arguments<a class="anchor" aria-label="anchor" href="#argumen
 
 
 <dl><dt id="arg-sorted-sequence">sorted_sequence<a class="anchor" aria-label="anchor" href="#arg-sorted-sequence"></a></dt>
-<dd><p>1-D input tensor, sorted along the innermost
-dimension.</p></dd>
+<dd><p>1-D input tensor that is sorted along its innermost dimension.</p></dd>
 
 
 <dt id="arg-values">values<a class="anchor" aria-label="anchor" href="#arg-values"></a></dt>
@@ -79,24 +78,27 @@ <h2 id="arguments">Arguments<a class="anchor" aria-label="anchor" href="#argumen
 
 
 <dt id="arg-side">side<a class="anchor" aria-label="anchor" href="#arg-side"></a></dt>
-<dd><p><code>'left'</code> or <code>'right'</code>, specifying the direction in which to insert
-for the equality case (tie-breaker).</p></dd>
+<dd><p>Either <code>"left"</code> or <code>"right"</code>, choosing which side to insert on ties.</p></dd>
 
 
 <dt id="arg-zero-indexed">zero_indexed<a class="anchor" aria-label="anchor" href="#arg-zero-indexed"></a></dt>
-<dd><p>If <code>TRUE</code>, the returned indices are zero-based (<code>0</code> encodes to first
+<dd><p>If <code>TRUE</code>, the returned indices are zero-based (<code>0</code> encodes the first
 position); if <code>FALSE</code> (default), the returned indices are one-based (<code>1</code>
-encodes to first position).</p></dd>
+encodes the first position).</p></dd>
 
 </dl></div>
     <div class="section level2">
     <h2 id="value">Value<a class="anchor" aria-label="anchor" href="#value"></a></h2>
-    <p>Tensor of insertion indices of same shape as <code>values</code>.</p>
+    <p>Tensor of insertion indices with the same shape as <code>values</code>.</p>
+    </div>
+    <div class="section level2">
+    <h2 id="examples">Examples<a class="anchor" aria-label="anchor" href="#examples"></a></h2>
+    <p></p><div class="sourceCode {r}"><pre><code></code></pre><p></p></div>
     </div>
     <div class="section level2">
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
     <div class="dont-index"><p>Other core ops: <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br></p>
-<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
+<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_layer_normalization.html">op_layer_normalization</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparse_sigmoid.html">op_sparse_sigmoid</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -108,7 +110,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/op_segment_max.html b/docs/dev/reference/op_segment_max.html
index 4233c33d5..dd934402a 100644
--- a/docs/dev/reference/op_segment_max.html
+++ b/docs/dev/reference/op_segment_max.html
@@ -8,7 +8,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -108,7 +108,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
     <div class="dont-index">
 <ul><li><p><a href="https://keras.io/api/ops/core#segmentmax-function" class="external-link">https://keras.io/api/ops/core#segmentmax-function</a></p></li>
 </ul><p>Other math ops: <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br></p>
-<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
+<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_layer_normalization.html">op_layer_normalization</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparse_sigmoid.html">op_sparse_sigmoid</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -120,7 +120,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/op_segment_sum.html b/docs/dev/reference/op_segment_sum.html
index a1bc86252..fe0c8ad57 100644
--- a/docs/dev/reference/op_segment_sum.html
+++ b/docs/dev/reference/op_segment_sum.html
@@ -8,7 +8,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -106,7 +106,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
     <div class="dont-index">
 <ul><li><p><a href="https://keras.io/api/ops/core#segmentsum-function" class="external-link">https://keras.io/api/ops/core#segmentsum-function</a></p></li>
 </ul><p>Other math ops: <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br></p>
-<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
+<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_layer_normalization.html">op_layer_normalization</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparse_sigmoid.html">op_sparse_sigmoid</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -118,7 +118,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/op_select.html b/docs/dev/reference/op_select.html
index e3f2a9b79..daacbad79 100644
--- a/docs/dev/reference/op_select.html
+++ b/docs/dev/reference/op_select.html
@@ -8,7 +8,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -103,8 +103,8 @@ <h2 id="examples">Examples<a class="anchor" aria-label="anchor" href="#examples"
     </div>
     <div class="section level2">
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
-    <div class="dont-index"><p>Other numpy ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p>
-<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
+    <div class="dont-index"><p>Other numpy ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p>
+<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_layer_normalization.html">op_layer_normalization</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparse_sigmoid.html">op_sparse_sigmoid</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -116,7 +116,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/op_selu.html b/docs/dev/reference/op_selu.html
index 40e0cb39b..66f5ecb44 100644
--- a/docs/dev/reference/op_selu.html
+++ b/docs/dev/reference/op_selu.html
@@ -12,7 +12,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -91,8 +91,8 @@ <h2 id="examples">Examples<a class="anchor" aria-label="anchor" href="#examples"
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
     <div class="dont-index">
 <ul><li><p><a href="https://keras.io/api/ops/nn#selu-function" class="external-link">https://keras.io/api/ops/nn#selu-function</a></p></li>
-</ul><p>Other nn ops: <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br></p>
-<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
+</ul><p>Other nn ops: <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_layer_normalization.html">op_layer_normalization</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparse_sigmoid.html">op_sparse_sigmoid</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br></p>
+<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_layer_normalization.html">op_layer_normalization</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparse_sigmoid.html">op_sparse_sigmoid</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -104,7 +104,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/op_separable_conv.html b/docs/dev/reference/op_separable_conv.html
index 2fe13e89f..cfd14df2c 100644
--- a/docs/dev/reference/op_separable_conv.html
+++ b/docs/dev/reference/op_separable_conv.html
@@ -10,7 +10,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -135,8 +135,8 @@ <h2 id="value">Value<a class="anchor" aria-label="anchor" href="#value"></a></h2
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
     <div class="dont-index">
 <ul><li><p><a href="https://keras.io/api/ops/nn#separableconv-function" class="external-link">https://keras.io/api/ops/nn#separableconv-function</a></p></li>
-</ul><p>Other nn ops: <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br></p>
-<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
+</ul><p>Other nn ops: <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_layer_normalization.html">op_layer_normalization</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparse_sigmoid.html">op_sparse_sigmoid</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br></p>
+<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_layer_normalization.html">op_layer_normalization</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparse_sigmoid.html">op_sparse_sigmoid</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -148,7 +148,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/op_shape.html b/docs/dev/reference/op_shape.html
index 1d0f96cb2..e5a58a2e0 100644
--- a/docs/dev/reference/op_shape.html
+++ b/docs/dev/reference/op_shape.html
@@ -8,7 +8,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -94,7 +94,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
     <div class="dont-index">
 <ul><li><p><a href="https://keras.io/api/ops/core#shape-function" class="external-link">https://keras.io/api/ops/core#shape-function</a></p></li>
 </ul><p>Other core ops: <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br></p>
-<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
+<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_layer_normalization.html">op_layer_normalization</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparse_sigmoid.html">op_sparse_sigmoid</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -106,7 +106,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/op_sigmoid.html b/docs/dev/reference/op_sigmoid.html
index 9ba1acab6..cf646b76c 100644
--- a/docs/dev/reference/op_sigmoid.html
+++ b/docs/dev/reference/op_sigmoid.html
@@ -8,7 +8,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -85,8 +85,8 @@ <h2 id="examples">Examples<a class="anchor" aria-label="anchor" href="#examples"
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
     <div class="dont-index">
 <ul><li><p><a href="https://keras.io/api/ops/nn#sigmoid-function" class="external-link">https://keras.io/api/ops/nn#sigmoid-function</a></p></li>
-</ul><p>Other nn ops: <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br></p>
-<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
+</ul><p>Other nn ops: <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_layer_normalization.html">op_layer_normalization</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparse_sigmoid.html">op_sparse_sigmoid</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br></p>
+<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_layer_normalization.html">op_layer_normalization</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparse_sigmoid.html">op_sparse_sigmoid</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -98,7 +98,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/op_sign.html b/docs/dev/reference/op_sign.html
index 2b900326f..48868029d 100644
--- a/docs/dev/reference/op_sign.html
+++ b/docs/dev/reference/op_sign.html
@@ -8,7 +8,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -78,8 +78,8 @@ <h2 id="value">Value<a class="anchor" aria-label="anchor" href="#value"></a></h2
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
     <div class="dont-index">
 <ul><li><p><a href="https://keras.io/api/ops/numpy#sign-function" class="external-link">https://keras.io/api/ops/numpy#sign-function</a></p></li>
-</ul><p>Other numpy ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p>
-<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
+</ul><p>Other numpy ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p>
+<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_layer_normalization.html">op_layer_normalization</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparse_sigmoid.html">op_sparse_sigmoid</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -91,7 +91,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/op_signbit.html b/docs/dev/reference/op_signbit.html
index ad620b41a..4d4d483aa 100644
--- a/docs/dev/reference/op_signbit.html
+++ b/docs/dev/reference/op_signbit.html
@@ -10,7 +10,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -79,8 +79,8 @@ <h2 id="value">Value<a class="anchor" aria-label="anchor" href="#value"></a></h2
     </div>
     <div class="section level2">
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
-    <div class="dont-index"><p>Other numpy ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p>
-<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
+    <div class="dont-index"><p>Other numpy ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p>
+<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_layer_normalization.html">op_layer_normalization</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparse_sigmoid.html">op_sparse_sigmoid</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -92,7 +92,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/op_silu.html b/docs/dev/reference/op_silu.html
index 162e6f01b..c32fd9e01 100644
--- a/docs/dev/reference/op_silu.html
+++ b/docs/dev/reference/op_silu.html
@@ -10,7 +10,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -91,8 +91,8 @@ <h2 id="examples">Examples<a class="anchor" aria-label="anchor" href="#examples"
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
     <div class="dont-index">
 <ul><li><p><a href="https://keras.io/api/ops/nn#silu-function" class="external-link">https://keras.io/api/ops/nn#silu-function</a></p></li>
-</ul><p>Other nn ops: <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br></p>
-<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
+</ul><p>Other nn ops: <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_layer_normalization.html">op_layer_normalization</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparse_sigmoid.html">op_sparse_sigmoid</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br></p>
+<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_layer_normalization.html">op_layer_normalization</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparse_sigmoid.html">op_sparse_sigmoid</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -104,7 +104,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/op_sin.html b/docs/dev/reference/op_sin.html
index a6b691d8f..5d3b8a22d 100644
--- a/docs/dev/reference/op_sin.html
+++ b/docs/dev/reference/op_sin.html
@@ -8,7 +8,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -78,8 +78,8 @@ <h2 id="value">Value<a class="anchor" aria-label="anchor" href="#value"></a></h2
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
     <div class="dont-index">
 <ul><li><p><a href="https://keras.io/api/ops/numpy#sin-function" class="external-link">https://keras.io/api/ops/numpy#sin-function</a></p></li>
-</ul><p>Other numpy ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p>
-<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
+</ul><p>Other numpy ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p>
+<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_layer_normalization.html">op_layer_normalization</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparse_sigmoid.html">op_sparse_sigmoid</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -91,7 +91,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/op_sinh.html b/docs/dev/reference/op_sinh.html
index cbfc1cab7..ad7810002 100644
--- a/docs/dev/reference/op_sinh.html
+++ b/docs/dev/reference/op_sinh.html
@@ -8,7 +8,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -78,8 +78,8 @@ <h2 id="value">Value<a class="anchor" aria-label="anchor" href="#value"></a></h2
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
     <div class="dont-index">
 <ul><li><p><a href="https://keras.io/api/ops/numpy#sinh-function" class="external-link">https://keras.io/api/ops/numpy#sinh-function</a></p></li>
-</ul><p>Other numpy ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p>
-<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
+</ul><p>Other numpy ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p>
+<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_layer_normalization.html">op_layer_normalization</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparse_sigmoid.html">op_sparse_sigmoid</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -91,7 +91,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/op_size.html b/docs/dev/reference/op_size.html
index e452823cd..d5b630e8b 100644
--- a/docs/dev/reference/op_size.html
+++ b/docs/dev/reference/op_size.html
@@ -8,7 +8,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -78,8 +78,8 @@ <h2 id="value">Value<a class="anchor" aria-label="anchor" href="#value"></a></h2
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
     <div class="dont-index">
 <ul><li><p><a href="https://keras.io/api/ops/numpy#size-function" class="external-link">https://keras.io/api/ops/numpy#size-function</a></p></li>
-</ul><p>Other numpy ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p>
-<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
+</ul><p>Other numpy ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p>
+<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_layer_normalization.html">op_layer_normalization</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparse_sigmoid.html">op_sparse_sigmoid</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -91,7 +91,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/op_slice.html b/docs/dev/reference/op_slice.html
index 3e70bff9f..aa5539966 100644
--- a/docs/dev/reference/op_slice.html
+++ b/docs/dev/reference/op_slice.html
@@ -54,7 +54,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -153,7 +153,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
     <div class="dont-index">
 <ul><li><p><a href="https://keras.io/api/ops/core#slice-function" class="external-link">https://keras.io/api/ops/core#slice-function</a></p></li>
 </ul><p>Other core ops: <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br></p>
-<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
+<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_layer_normalization.html">op_layer_normalization</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparse_sigmoid.html">op_sparse_sigmoid</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -165,7 +165,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/op_slice_update.html b/docs/dev/reference/op_slice_update.html
index 36ba18c5c..b12ad6d8e 100644
--- a/docs/dev/reference/op_slice_update.html
+++ b/docs/dev/reference/op_slice_update.html
@@ -50,7 +50,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -150,7 +150,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
     <div class="dont-index">
 <ul><li><p><a href="https://keras.io/api/ops/core#sliceupdate-function" class="external-link">https://keras.io/api/ops/core#sliceupdate-function</a></p></li>
 </ul><p>Other core ops: <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br></p>
-<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
+<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_layer_normalization.html">op_layer_normalization</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparse_sigmoid.html">op_sparse_sigmoid</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -162,7 +162,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/op_slogdet.html b/docs/dev/reference/op_slogdet.html
index 95e9ccffe..b9058f88f 100644
--- a/docs/dev/reference/op_slogdet.html
+++ b/docs/dev/reference/op_slogdet.html
@@ -8,7 +8,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -81,7 +81,7 @@ <h2 id="value">Value<a class="anchor" aria-label="anchor" href="#value"></a></h2
     <div class="section level2">
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
     <div class="dont-index"><p>Other linear algebra ops: <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br></p>
-<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
+<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_layer_normalization.html">op_layer_normalization</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparse_sigmoid.html">op_sparse_sigmoid</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -93,7 +93,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/op_soft_shrink.html b/docs/dev/reference/op_soft_shrink.html
index f75f6bc04..b66233937 100644
--- a/docs/dev/reference/op_soft_shrink.html
+++ b/docs/dev/reference/op_soft_shrink.html
@@ -14,7 +14,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -96,8 +96,8 @@ <h2 id="examples">Examples<a class="anchor" aria-label="anchor" href="#examples"
     </div>
     <div class="section level2">
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
-    <div class="dont-index"><p>Other nn ops: <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br></p>
-<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
+    <div class="dont-index"><p>Other nn ops: <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_layer_normalization.html">op_layer_normalization</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparse_sigmoid.html">op_sparse_sigmoid</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br></p>
+<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_layer_normalization.html">op_layer_normalization</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparse_sigmoid.html">op_sparse_sigmoid</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -109,7 +109,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/op_softmax.html b/docs/dev/reference/op_softmax.html
index 3df245b0d..266859030 100644
--- a/docs/dev/reference/op_softmax.html
+++ b/docs/dev/reference/op_softmax.html
@@ -18,7 +18,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -104,8 +104,8 @@ <h2 id="examples">Examples<a class="anchor" aria-label="anchor" href="#examples"
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
     <div class="dont-index">
 <ul><li><p><a href="https://keras.io/api/ops/nn#softmax-function" class="external-link">https://keras.io/api/ops/nn#softmax-function</a></p></li>
-</ul><p>Other nn ops: <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br></p>
-<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
+</ul><p>Other nn ops: <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_layer_normalization.html">op_layer_normalization</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparse_sigmoid.html">op_sparse_sigmoid</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br></p>
+<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_layer_normalization.html">op_layer_normalization</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparse_sigmoid.html">op_sparse_sigmoid</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -117,7 +117,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/op_softplus.html b/docs/dev/reference/op_softplus.html
index 7cce9e010..4a8cec714 100644
--- a/docs/dev/reference/op_softplus.html
+++ b/docs/dev/reference/op_softplus.html
@@ -10,7 +10,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -91,8 +91,8 @@ <h2 id="examples">Examples<a class="anchor" aria-label="anchor" href="#examples"
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
     <div class="dont-index">
 <ul><li><p><a href="https://keras.io/api/ops/nn#softplus-function" class="external-link">https://keras.io/api/ops/nn#softplus-function</a></p></li>
-</ul><p>Other nn ops: <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br></p>
-<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
+</ul><p>Other nn ops: <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_layer_normalization.html">op_layer_normalization</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparse_sigmoid.html">op_sparse_sigmoid</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br></p>
+<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_layer_normalization.html">op_layer_normalization</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparse_sigmoid.html">op_sparse_sigmoid</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -104,7 +104,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/op_softsign.html b/docs/dev/reference/op_softsign.html
index ccad222f3..91e015324 100644
--- a/docs/dev/reference/op_softsign.html
+++ b/docs/dev/reference/op_softsign.html
@@ -8,7 +8,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -88,8 +88,8 @@ <h2 id="examples">Examples<a class="anchor" aria-label="anchor" href="#examples"
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
     <div class="dont-index">
 <ul><li><p><a href="https://keras.io/api/ops/nn#softsign-function" class="external-link">https://keras.io/api/ops/nn#softsign-function</a></p></li>
-</ul><p>Other nn ops: <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br></p>
-<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
+</ul><p>Other nn ops: <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_layer_normalization.html">op_layer_normalization</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparse_sigmoid.html">op_sparse_sigmoid</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br></p>
+<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_layer_normalization.html">op_layer_normalization</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparse_sigmoid.html">op_sparse_sigmoid</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -101,7 +101,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/op_solve.html b/docs/dev/reference/op_solve.html
index 4ba6e5f60..970dc373f 100644
--- a/docs/dev/reference/op_solve.html
+++ b/docs/dev/reference/op_solve.html
@@ -8,7 +8,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -93,7 +93,7 @@ <h2 id="examples">Examples<a class="anchor" aria-label="anchor" href="#examples"
     <div class="section level2">
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
     <div class="dont-index"><p>Other math ops: <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br></p>
-<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
+<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_layer_normalization.html">op_layer_normalization</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparse_sigmoid.html">op_sparse_sigmoid</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -105,7 +105,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/op_solve_triangular.html b/docs/dev/reference/op_solve_triangular.html
index cac64d9b9..e66baa638 100644
--- a/docs/dev/reference/op_solve_triangular.html
+++ b/docs/dev/reference/op_solve_triangular.html
@@ -8,7 +8,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -88,7 +88,7 @@ <h2 id="value">Value<a class="anchor" aria-label="anchor" href="#value"></a></h2
     <div class="section level2">
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
     <div class="dont-index"><p>Other linear algebra ops: <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br></p>
-<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
+<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_layer_normalization.html">op_layer_normalization</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparse_sigmoid.html">op_sparse_sigmoid</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -100,7 +100,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/op_sort.html b/docs/dev/reference/op_sort.html
index 5ce874a7a..2b893efcb 100644
--- a/docs/dev/reference/op_sort.html
+++ b/docs/dev/reference/op_sort.html
@@ -8,7 +8,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -83,8 +83,8 @@ <h2 id="value">Value<a class="anchor" aria-label="anchor" href="#value"></a></h2
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
     <div class="dont-index">
 <ul><li><p><a href="https://keras.io/api/ops/numpy#sort-function" class="external-link">https://keras.io/api/ops/numpy#sort-function</a></p></li>
-</ul><p>Other numpy ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p>
-<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
+</ul><p>Other numpy ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p>
+<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_layer_normalization.html">op_layer_normalization</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparse_sigmoid.html">op_sparse_sigmoid</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -96,7 +96,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/op_sparse_categorical_crossentropy.html b/docs/dev/reference/op_sparse_categorical_crossentropy.html
index c493b1642..f766bfe42 100644
--- a/docs/dev/reference/op_sparse_categorical_crossentropy.html
+++ b/docs/dev/reference/op_sparse_categorical_crossentropy.html
@@ -14,7 +14,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -127,8 +127,8 @@ <h2 id="examples">Examples<a class="anchor" aria-label="anchor" href="#examples"
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
     <div class="dont-index">
 <ul><li><p><a href="https://keras.io/api/ops/nn#sparsecategoricalcrossentropy-function" class="external-link">https://keras.io/api/ops/nn#sparsecategoricalcrossentropy-function</a></p></li>
-</ul><p>Other nn ops: <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br></p>
-<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
+</ul><p>Other nn ops: <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_layer_normalization.html">op_layer_normalization</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparse_sigmoid.html">op_sparse_sigmoid</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br></p>
+<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_layer_normalization.html">op_layer_normalization</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparse_sigmoid.html">op_sparse_sigmoid</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -140,7 +140,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/op_sparse_plus.html b/docs/dev/reference/op_sparse_plus.html
index 01ad9f8ef..f4db11039 100644
--- a/docs/dev/reference/op_sparse_plus.html
+++ b/docs/dev/reference/op_sparse_plus.html
@@ -14,7 +14,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -92,8 +92,8 @@ <h2 id="examples">Examples<a class="anchor" aria-label="anchor" href="#examples"
     </div>
     <div class="section level2">
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
-    <div class="dont-index"><p>Other nn ops: <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br></p>
-<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
+    <div class="dont-index"><p>Other nn ops: <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_layer_normalization.html">op_layer_normalization</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_sigmoid.html">op_sparse_sigmoid</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br></p>
+<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_layer_normalization.html">op_layer_normalization</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_sigmoid.html">op_sparse_sigmoid</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -105,7 +105,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/op_sparse_sigmoid.html b/docs/dev/reference/op_sparse_sigmoid.html
new file mode 100644
index 000000000..3fc235561
--- /dev/null
+++ b/docs/dev/reference/op_sparse_sigmoid.html
@@ -0,0 +1,118 @@
+<!DOCTYPE html>
+<!-- Generated by pkgdown: do not edit by hand --><html lang="en"><head><meta http-equiv="Content-Type" content="text/html; charset=UTF-8"><meta charset="utf-8"><meta http-equiv="X-UA-Compatible" content="IE=edge"><meta name="viewport" content="width=device-width, initial-scale=1, shrink-to-fit=no"><title>Sparse sigmoid activation function. — op_sparse_sigmoid • keras3</title><!-- favicons --><link rel="icon" type="image/png" sizes="96x96" href="../favicon-96x96.png"><link rel="icon" type="”image/svg+xml”" href="../favicon.svg"><link rel="apple-touch-icon" sizes="180x180" href="../apple-touch-icon.png"><link rel="icon" sizes="any" href="../favicon.ico"><link rel="manifest" href="../site.webmanifest"><script src="../deps/jquery-3.6.0/jquery-3.6.0.min.js"></script><meta name="viewport" content="width=device-width, initial-scale=1, shrink-to-fit=no"><link href="../deps/bootstrap-5.3.1/bootstrap.min.css" rel="stylesheet"><script src="../deps/bootstrap-5.3.1/bootstrap.bundle.min.js"></script><link href="../deps/font-awesome-6.5.2/css/all.min.css" rel="stylesheet"><link href="../deps/font-awesome-6.5.2/css/v4-shims.min.css" rel="stylesheet"><script src="../deps/headroom-0.11.0/headroom.min.js"></script><script src="../deps/headroom-0.11.0/jQuery.headroom.min.js"></script><script src="../deps/bootstrap-toc-1.0.1/bootstrap-toc.min.js"></script><script src="../deps/clipboard.js-2.0.11/clipboard.min.js"></script><script src="../deps/search-1.0.0/autocomplete.jquery.min.js"></script><script src="../deps/search-1.0.0/fuse.min.js"></script><script src="../deps/search-1.0.0/mark.min.js"></script><!-- pkgdown --><script src="../pkgdown.js"></script><link href="../extra.css" rel="stylesheet"><meta property="og:title" content="Sparse sigmoid activation function. — op_sparse_sigmoid"><meta name="description" content="It is defined as
+f(x) = 0 for x &amp;lt;= -1,
+f(x) = 0.5 * (x + 1) for -1 &amp;lt; x &amp;lt; 1,
+f(x) = 1 for x &amp;gt;= 1."><meta property="og:description" content="It is defined as
+f(x) = 0 for x &amp;lt;= -1,
+f(x) = 0.5 * (x + 1) for -1 &amp;lt; x &amp;lt; 1,
+f(x) = 1 for x &amp;gt;= 1."><meta name="robots" content="noindex"><!-- Google Tag Manager --><script>(function(w,d,s,l,i){w[l]=w[l]||[];w[l].push({'gtm.start': new Date().getTime(),event:'gtm.js'});var f=d.getElementsByTagName(s)[0],j=d.createElement(s),dl=l!='dataLayer'?'&l='+l:'';j.async=true;j.src='https://www.googletagmanager.com/gtm.js?id='+i+dl;f.parentNode.insertBefore(j,f);})(window,document,'script','dataLayer','GTM-KHBDBW7');</script><!-- End Google Tag Manager --></head><body>
+    <a href="#main" class="visually-hidden-focusable">Skip to contents</a>
+    <!-- Google Tag Manager (noscript) --> <noscript><iframe src="https://www.googletagmanager.com/ns.html?id=GTM-KHBDBW7" height="0" width="0" style="display:none;visibility:hidden"></iframe></noscript> <!-- End Google Tag Manager (noscript) -->
+
+
+    <nav class="navbar navbar-expand-lg fixed-top bg-primary" data-bs-theme="inverse" aria-label="Site navigation"><div class="container">
+
+    <a class="navbar-brand me-2" href="../index.html">keras3</a>
+
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
+
+
+    <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
+      <span class="navbar-toggler-icon"></span>
+    </button>
+
+    <div id="navbar" class="collapse navbar-collapse ms-3">
+      <ul class="navbar-nav me-auto"><li class="nav-item"><a class="nav-link" href="../articles/getting_started.html">Getting Started</a></li>
+<li class="nav-item dropdown">
+  <button class="nav-link dropdown-toggle" type="button" id="dropdown-guides" data-bs-toggle="dropdown" aria-expanded="false" aria-haspopup="true">Guides</button>
+  <ul class="dropdown-menu" aria-labelledby="dropdown-guides"><li><h6 class="dropdown-header" data-toc-skip>Model definition</h6></li>
+    <li><a class="dropdown-item" href="../articles/sequential_model.html">Sequential Model</a></li>
+    <li><a class="dropdown-item" href="../articles/functional_api.html">Functional API</a></li>
+    <li><h6 class="dropdown-header" data-toc-skip>Extending and customizing</h6></li>
+    <li><a class="dropdown-item" href="../articles/training_with_built_in_methods.html">Training &amp; evaluation with the built-in methods</a></li>
+    <li><a class="dropdown-item" href="../articles/custom_train_step_in_tensorflow.html">Customizing `fit()` with Tensorflow</a></li>
+    <li><a class="dropdown-item" href="../articles/writing_your_own_callbacks.html">Writing your own callbacks</a></li>
+    <li><a class="dropdown-item" href="../articles/making_new_layers_and_models_via_subclassing.html">Making new layers and models via subclassing</a></li>
+    <li><a class="dropdown-item" href="../articles/writing_a_custom_training_loop_in_tensorflow.html">Writing a training loop from scratch in TensorFlow</a></li>
+    <li><a class="dropdown-item" href="../articles/serialization_and_saving.html">Serialization and Saving</a></li>
+    <li><h6 class="dropdown-header" data-toc-skip>Other topics</h6></li>
+    <li><a class="dropdown-item" href="../articles/transfer_learning.html">Transfer learning and fine tuning</a></li>
+    <li><a class="dropdown-item" href="../articles/distributed_training_with_tensorflow.html">Distributed training with TensorFlow</a></li>
+    <li><a class="dropdown-item" href="../articles/distribution.html">Distributed training with Jax</a></li>
+  </ul></li>
+<li class="nav-item"><a class="nav-link" href="../articles/examples/index.html">Examples</a></li>
+<li class="active nav-item"><a class="nav-link" href="../reference/index.html">Reference</a></li>
+<li class="nav-item"><a class="nav-link" href="../news/index.html">News</a></li>
+      </ul><ul class="navbar-nav"><li class="nav-item"><form class="form-inline" role="search">
+ <input class="form-control" type="search" name="search-input" id="search-input" autocomplete="off" aria-label="Search site" placeholder="Search for" data-search-index="../search.json"></form></li>
+<li class="nav-item"><a class="external-link nav-link" href="https://github.com/rstudio/keras3/" aria-label="GitHub"><span class="fa fab fa-github fa-lg"></span></a></li>
+      </ul></div>
+
+
+  </div>
+</nav><div class="container template-reference-topic">
+<div class="row">
+  <main id="main" class="col-md-9"><div class="page-header">
+
+      <h1>Sparse sigmoid activation function.</h1>
+      <small class="dont-index">Source: <a href="https://github.com/rstudio/keras3/blob/HEAD/R/ops-nn.R" class="external-link"><code>R/ops-nn.R</code></a></small>
+      <div class="d-none name"><code>op_sparse_sigmoid.Rd</code></div>
+    </div>
+
+    <div class="ref-description section level2">
+    <p>It is defined as</p>
+<p><code>f(x) = 0</code> for <code>x &lt;= -1</code>,
+<code>f(x) = 0.5 * (x + 1)</code> for <code>-1 &lt; x &lt; 1</code>,
+<code>f(x) = 1</code> for <code>x &gt;= 1</code>.</p>
+    </div>
+
+    <div class="section level2">
+    <h2 id="ref-usage">Usage<a class="anchor" aria-label="anchor" href="#ref-usage"></a></h2>
+    <div class="sourceCode"><pre class="sourceCode r"><code><span><span class="fu">op_sparse_sigmoid</span><span class="op">(</span><span class="va">x</span><span class="op">)</span></span></code></pre></div>
+    </div>
+
+    <div class="section level2">
+    <h2 id="arguments">Arguments<a class="anchor" aria-label="anchor" href="#arguments"></a></h2>
+
+
+<dl><dt id="arg-x">x<a class="anchor" aria-label="anchor" href="#arg-x"></a></dt>
+<dd><p>Input tensor.</p></dd>
+
+</dl></div>
+    <div class="section level2">
+    <h2 id="value">Value<a class="anchor" aria-label="anchor" href="#value"></a></h2>
+    <p>A tensor with the same shape as <code>x</code>.</p>
+    </div>
+    <div class="section level2">
+    <h2 id="examples">Examples<a class="anchor" aria-label="anchor" href="#examples"></a></h2>
+    <p></p><div class="sourceCode r"><pre><code><span><span class="va">x</span> <span class="op">&lt;-</span> <span class="fu"><a href="../reference/op_array.html">op_array</a></span><span class="op">(</span><span class="fu"><a href="https://rdrr.io/r/base/c.html" class="external-link">c</a></span><span class="op">(</span><span class="op">-</span><span class="fl">1.0</span>, <span class="fl">0.0</span>, <span class="fl">1.0</span><span class="op">)</span><span class="op">)</span></span>
+<span><span class="fu"><a href="../reference/op_sparse_sigmoid.html">op_sparse_sigmoid</a></span><span class="op">(</span><span class="va">x</span><span class="op">)</span></span></code></pre><p></p></div>
+<p></p><div class="sourceCode"><pre><code><span id="cb1-1"><a href="#cb1-1" tabindex="-1"></a><span class="do">## tf.Tensor([0.  0.5 1. ], shape=(3), dtype=float32)</span></span>
+<span id="cb1-2"><a href="#cb1-2" tabindex="-1"></a></span></code></pre><p></p></div>
+    </div>
+    <div class="section level2">
+    <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
+    <div class="dont-index"><p>Other nn ops: <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_layer_normalization.html">op_layer_normalization</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br></p>
+<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_layer_normalization.html">op_layer_normalization</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
+    </div>
+
+  </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
+    </nav></aside></div>
+
+
+    <footer><div class="pkgdown-footer-left">
+  <p>Developed by Tomasz Kalinowski, JJ Allaire, François Chollet, Posit Software, PBC, Google.</p>
+</div>
+
+<div class="pkgdown-footer-right">
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
+</div>
+
+    </footer></div>
+
+
+
+
+
+  </body></html>
+
diff --git a/docs/dev/reference/op_sparsemax.html b/docs/dev/reference/op_sparsemax.html
index 4c2b28b9d..326c91827 100644
--- a/docs/dev/reference/op_sparsemax.html
+++ b/docs/dev/reference/op_sparsemax.html
@@ -12,7 +12,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -94,8 +94,8 @@ <h2 id="examples">Examples<a class="anchor" aria-label="anchor" href="#examples"
     </div>
     <div class="section level2">
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
-    <div class="dont-index"><p>Other nn ops: <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br></p>
-<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
+    <div class="dont-index"><p>Other nn ops: <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_layer_normalization.html">op_layer_normalization</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparse_sigmoid.html">op_sparse_sigmoid</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br></p>
+<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_layer_normalization.html">op_layer_normalization</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparse_sigmoid.html">op_sparse_sigmoid</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -107,7 +107,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/op_split.html b/docs/dev/reference/op_split.html
index 4a471f3b6..63b7388be 100644
--- a/docs/dev/reference/op_split.html
+++ b/docs/dev/reference/op_split.html
@@ -8,7 +8,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -135,8 +135,8 @@ <h2 id="example">Example<a class="anchor" aria-label="anchor" href="#example"></
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
     <div class="dont-index">
 <ul><li><p><a href="https://keras.io/api/ops/numpy#split-function" class="external-link">https://keras.io/api/ops/numpy#split-function</a></p></li>
-</ul><p>Other numpy ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p>
-<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
+</ul><p>Other numpy ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p>
+<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_layer_normalization.html">op_layer_normalization</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparse_sigmoid.html">op_sparse_sigmoid</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -148,7 +148,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/op_sqrt.html b/docs/dev/reference/op_sqrt.html
index e2d3bd432..c4bd16574 100644
--- a/docs/dev/reference/op_sqrt.html
+++ b/docs/dev/reference/op_sqrt.html
@@ -8,7 +8,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -78,8 +78,8 @@ <h2 id="value">Value<a class="anchor" aria-label="anchor" href="#value"></a></h2
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
     <div class="dont-index">
 <ul><li><p><a href="https://keras.io/api/ops/numpy#sqrt-function" class="external-link">https://keras.io/api/ops/numpy#sqrt-function</a></p></li>
-</ul><p>Other numpy ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p>
-<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
+</ul><p>Other numpy ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p>
+<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_layer_normalization.html">op_layer_normalization</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparse_sigmoid.html">op_sparse_sigmoid</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -91,7 +91,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/op_square.html b/docs/dev/reference/op_square.html
index 071bfae18..11654cb6e 100644
--- a/docs/dev/reference/op_square.html
+++ b/docs/dev/reference/op_square.html
@@ -8,7 +8,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -78,8 +78,8 @@ <h2 id="value">Value<a class="anchor" aria-label="anchor" href="#value"></a></h2
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
     <div class="dont-index">
 <ul><li><p><a href="https://keras.io/api/ops/numpy#square-function" class="external-link">https://keras.io/api/ops/numpy#square-function</a></p></li>
-</ul><p>Other numpy ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p>
-<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
+</ul><p>Other numpy ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p>
+<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_layer_normalization.html">op_layer_normalization</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparse_sigmoid.html">op_sparse_sigmoid</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -91,7 +91,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/op_squareplus.html b/docs/dev/reference/op_squareplus.html
index 8388bb474..e14d3ddb2 100644
--- a/docs/dev/reference/op_squareplus.html
+++ b/docs/dev/reference/op_squareplus.html
@@ -10,7 +10,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -90,8 +90,8 @@ <h2 id="examples">Examples<a class="anchor" aria-label="anchor" href="#examples"
     </div>
     <div class="section level2">
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
-    <div class="dont-index"><p>Other nn ops: <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br></p>
-<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
+    <div class="dont-index"><p>Other nn ops: <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_layer_normalization.html">op_layer_normalization</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparse_sigmoid.html">op_sparse_sigmoid</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br></p>
+<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_layer_normalization.html">op_layer_normalization</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparse_sigmoid.html">op_sparse_sigmoid</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -103,7 +103,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/op_squeeze.html b/docs/dev/reference/op_squeeze.html
index ae6b67bdb..fa460e824 100644
--- a/docs/dev/reference/op_squeeze.html
+++ b/docs/dev/reference/op_squeeze.html
@@ -8,7 +8,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -83,8 +83,8 @@ <h2 id="value">Value<a class="anchor" aria-label="anchor" href="#value"></a></h2
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
     <div class="dont-index">
 <ul><li><p><a href="https://keras.io/api/ops/numpy#squeeze-function" class="external-link">https://keras.io/api/ops/numpy#squeeze-function</a></p></li>
-</ul><p>Other numpy ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p>
-<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
+</ul><p>Other numpy ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p>
+<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_layer_normalization.html">op_layer_normalization</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparse_sigmoid.html">op_sparse_sigmoid</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -96,7 +96,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/op_stack.html b/docs/dev/reference/op_stack.html
index 65be67926..475a8145a 100644
--- a/docs/dev/reference/op_stack.html
+++ b/docs/dev/reference/op_stack.html
@@ -10,7 +10,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -85,8 +85,8 @@ <h2 id="value">Value<a class="anchor" aria-label="anchor" href="#value"></a></h2
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
     <div class="dont-index">
 <ul><li><p><a href="https://keras.io/api/ops/numpy#stack-function" class="external-link">https://keras.io/api/ops/numpy#stack-function</a></p></li>
-</ul><p>Other numpy ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p>
-<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
+</ul><p>Other numpy ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p>
+<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_layer_normalization.html">op_layer_normalization</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparse_sigmoid.html">op_sparse_sigmoid</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -98,7 +98,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/op_std.html b/docs/dev/reference/op_std.html
index 0d5e0d7eb..d7d038f31 100644
--- a/docs/dev/reference/op_std.html
+++ b/docs/dev/reference/op_std.html
@@ -8,7 +8,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -89,8 +89,8 @@ <h2 id="value">Value<a class="anchor" aria-label="anchor" href="#value"></a></h2
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
     <div class="dont-index">
 <ul><li><p><a href="https://keras.io/api/ops/numpy#std-function" class="external-link">https://keras.io/api/ops/numpy#std-function</a></p></li>
-</ul><p>Other numpy ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p>
-<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
+</ul><p>Other numpy ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p>
+<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_layer_normalization.html">op_layer_normalization</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparse_sigmoid.html">op_sparse_sigmoid</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -102,7 +102,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/op_stft.html b/docs/dev/reference/op_stft.html
index eff03677c..98c05ba78 100644
--- a/docs/dev/reference/op_stft.html
+++ b/docs/dev/reference/op_stft.html
@@ -12,7 +12,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -137,7 +137,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
     <div class="dont-index">
 <ul><li><p><a href="https://keras.io/api/ops/fft#stft-function" class="external-link">https://keras.io/api/ops/fft#stft-function</a></p></li>
 </ul><p>Other math ops: <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br></p>
-<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
+<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_layer_normalization.html">op_layer_normalization</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparse_sigmoid.html">op_sparse_sigmoid</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -149,7 +149,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/op_stop_gradient.html b/docs/dev/reference/op_stop_gradient.html
index 7fd1906aa..009a8b3df 100644
--- a/docs/dev/reference/op_stop_gradient.html
+++ b/docs/dev/reference/op_stop_gradient.html
@@ -8,7 +8,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -85,7 +85,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
     <div class="dont-index">
 <ul><li><p><a href="https://keras.io/api/ops/core#stopgradient-function" class="external-link">https://keras.io/api/ops/core#stopgradient-function</a></p></li>
 </ul><p>Other core ops: <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br></p>
-<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
+<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_layer_normalization.html">op_layer_normalization</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparse_sigmoid.html">op_sparse_sigmoid</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -97,7 +97,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/op_subset.html b/docs/dev/reference/op_subset.html
index 98cf07007..73db95224 100644
--- a/docs/dev/reference/op_subset.html
+++ b/docs/dev/reference/op_subset.html
@@ -10,7 +10,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -321,11 +321,11 @@ <h2 id="examples">Examples<a class="anchor" aria-label="anchor" href="#examples"
     <div class="section level2">
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
     <div class="dont-index"><p>Other core ops: <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br></p>
-<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p>
+<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_layer_normalization.html">op_layer_normalization</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparse_sigmoid.html">op_sparse_sigmoid</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p>
 <p>Other core ops: <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br></p>
-<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p>
+<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_layer_normalization.html">op_layer_normalization</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparse_sigmoid.html">op_sparse_sigmoid</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p>
 <p>Other core ops: <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br></p>
-<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
+<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_layer_normalization.html">op_layer_normalization</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparse_sigmoid.html">op_sparse_sigmoid</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -337,7 +337,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/op_subtract.html b/docs/dev/reference/op_subtract.html
index b93b93582..c804437eb 100644
--- a/docs/dev/reference/op_subtract.html
+++ b/docs/dev/reference/op_subtract.html
@@ -30,7 +30,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -111,8 +111,8 @@ <h2 id="value">Value<a class="anchor" aria-label="anchor" href="#value"></a></h2
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
     <div class="dont-index">
 <ul><li><p><a href="https://keras.io/api/ops/numpy#subtract-function" class="external-link">https://keras.io/api/ops/numpy#subtract-function</a></p></li>
-</ul><p>Other numpy ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p>
-<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
+</ul><p>Other numpy ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p>
+<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_layer_normalization.html">op_layer_normalization</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparse_sigmoid.html">op_sparse_sigmoid</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -124,7 +124,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/op_sum.html b/docs/dev/reference/op_sum.html
index 908b15cfd..e8acaff94 100644
--- a/docs/dev/reference/op_sum.html
+++ b/docs/dev/reference/op_sum.html
@@ -8,7 +8,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -88,8 +88,8 @@ <h2 id="value">Value<a class="anchor" aria-label="anchor" href="#value"></a></h2
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
     <div class="dont-index">
 <ul><li><p><a href="https://keras.io/api/ops/numpy#sum-function" class="external-link">https://keras.io/api/ops/numpy#sum-function</a></p></li>
-</ul><p>Other numpy ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p>
-<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
+</ul><p>Other numpy ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p>
+<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_layer_normalization.html">op_layer_normalization</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparse_sigmoid.html">op_sparse_sigmoid</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -101,7 +101,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/op_svd.html b/docs/dev/reference/op_svd.html
index 605178abf..7bcc071ae 100644
--- a/docs/dev/reference/op_svd.html
+++ b/docs/dev/reference/op_svd.html
@@ -8,7 +8,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -92,7 +92,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
     <div class="dont-index">
 <ul><li><p><a href="https://www.tensorflow.org/api_docs/python/tf/keras/ops/svd" class="external-link">https://www.tensorflow.org/api_docs/python/tf/keras/ops/svd</a></p></li>
 </ul><p>Other linear algebra ops: <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br></p>
-<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
+<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_layer_normalization.html">op_layer_normalization</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparse_sigmoid.html">op_sparse_sigmoid</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -104,7 +104,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/op_swapaxes.html b/docs/dev/reference/op_swapaxes.html
index d747c3239..98c7bbffb 100644
--- a/docs/dev/reference/op_swapaxes.html
+++ b/docs/dev/reference/op_swapaxes.html
@@ -8,7 +8,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -86,8 +86,8 @@ <h2 id="value">Value<a class="anchor" aria-label="anchor" href="#value"></a></h2
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
     <div class="dont-index">
 <ul><li><p><a href="https://keras.io/api/ops/numpy#swapaxes-function" class="external-link">https://keras.io/api/ops/numpy#swapaxes-function</a></p></li>
-</ul><p>Other numpy ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p>
-<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
+</ul><p>Other numpy ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p>
+<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_layer_normalization.html">op_layer_normalization</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparse_sigmoid.html">op_sparse_sigmoid</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -99,7 +99,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/op_switch.html b/docs/dev/reference/op_switch.html
index 348aad533..2ddc2109a 100644
--- a/docs/dev/reference/op_switch.html
+++ b/docs/dev/reference/op_switch.html
@@ -20,7 +20,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -123,7 +123,7 @@ <h2 id="examples">Examples<a class="anchor" aria-label="anchor" href="#examples"
     <div class="section level2">
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
     <div class="dont-index"><p>Other core ops: <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br></p>
-<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
+<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_layer_normalization.html">op_layer_normalization</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparse_sigmoid.html">op_sparse_sigmoid</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -135,7 +135,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/op_take.html b/docs/dev/reference/op_take.html
index 24d808612..6aa860377 100644
--- a/docs/dev/reference/op_take.html
+++ b/docs/dev/reference/op_take.html
@@ -8,7 +8,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -93,8 +93,8 @@ <h2 id="value">Value<a class="anchor" aria-label="anchor" href="#value"></a></h2
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
     <div class="dont-index">
 <ul><li><p><a href="https://keras.io/api/ops/numpy#take-function" class="external-link">https://keras.io/api/ops/numpy#take-function</a></p></li>
-</ul><p>Other numpy ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p>
-<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
+</ul><p>Other numpy ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p>
+<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_layer_normalization.html">op_layer_normalization</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparse_sigmoid.html">op_sparse_sigmoid</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -106,7 +106,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/op_take_along_axis.html b/docs/dev/reference/op_take_along_axis.html
index 27afee6c1..282fe3148 100644
--- a/docs/dev/reference/op_take_along_axis.html
+++ b/docs/dev/reference/op_take_along_axis.html
@@ -8,7 +8,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -93,8 +93,8 @@ <h2 id="value">Value<a class="anchor" aria-label="anchor" href="#value"></a></h2
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
     <div class="dont-index">
 <ul><li><p><a href="https://keras.io/api/ops/numpy#takealongaxis-function" class="external-link">https://keras.io/api/ops/numpy#takealongaxis-function</a></p></li>
-</ul><p>Other numpy ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p>
-<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
+</ul><p>Other numpy ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p>
+<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_layer_normalization.html">op_layer_normalization</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparse_sigmoid.html">op_sparse_sigmoid</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -106,7 +106,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/op_tan.html b/docs/dev/reference/op_tan.html
index 18cb56785..64084244d 100644
--- a/docs/dev/reference/op_tan.html
+++ b/docs/dev/reference/op_tan.html
@@ -8,7 +8,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -78,8 +78,8 @@ <h2 id="value">Value<a class="anchor" aria-label="anchor" href="#value"></a></h2
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
     <div class="dont-index">
 <ul><li><p><a href="https://keras.io/api/ops/numpy#tan-function" class="external-link">https://keras.io/api/ops/numpy#tan-function</a></p></li>
-</ul><p>Other numpy ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p>
-<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
+</ul><p>Other numpy ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p>
+<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_layer_normalization.html">op_layer_normalization</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparse_sigmoid.html">op_sparse_sigmoid</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -91,7 +91,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/op_tanh.html b/docs/dev/reference/op_tanh.html
index 1271df9bf..5e2ca9ed9 100644
--- a/docs/dev/reference/op_tanh.html
+++ b/docs/dev/reference/op_tanh.html
@@ -8,7 +8,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -78,8 +78,8 @@ <h2 id="value">Value<a class="anchor" aria-label="anchor" href="#value"></a></h2
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
     <div class="dont-index">
 <ul><li><p><a href="https://keras.io/api/ops/nn#tanh-function" class="external-link">https://keras.io/api/ops/nn#tanh-function</a></p></li>
-</ul><p>Other numpy ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p>
-<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
+</ul><p>Other numpy ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p>
+<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_layer_normalization.html">op_layer_normalization</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparse_sigmoid.html">op_sparse_sigmoid</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -91,7 +91,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/op_tanh_shrink.html b/docs/dev/reference/op_tanh_shrink.html
index 290538401..48462bd3b 100644
--- a/docs/dev/reference/op_tanh_shrink.html
+++ b/docs/dev/reference/op_tanh_shrink.html
@@ -10,7 +10,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -87,8 +87,8 @@ <h2 id="examples">Examples<a class="anchor" aria-label="anchor" href="#examples"
     </div>
     <div class="section level2">
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
-    <div class="dont-index"><p>Other nn ops: <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br></p>
-<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
+    <div class="dont-index"><p>Other nn ops: <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_layer_normalization.html">op_layer_normalization</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparse_sigmoid.html">op_sparse_sigmoid</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br></p>
+<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_layer_normalization.html">op_layer_normalization</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparse_sigmoid.html">op_sparse_sigmoid</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -100,7 +100,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/op_tensordot.html b/docs/dev/reference/op_tensordot.html
index a3abcad90..c0fc0a63c 100644
--- a/docs/dev/reference/op_tensordot.html
+++ b/docs/dev/reference/op_tensordot.html
@@ -8,7 +8,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -92,8 +92,8 @@ <h2 id="value">Value<a class="anchor" aria-label="anchor" href="#value"></a></h2
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
     <div class="dont-index">
 <ul><li><p><a href="https://keras.io/api/ops/numpy#tensordot-function" class="external-link">https://keras.io/api/ops/numpy#tensordot-function</a></p></li>
-</ul><p>Other numpy ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p>
-<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
+</ul><p>Other numpy ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p>
+<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_layer_normalization.html">op_layer_normalization</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparse_sigmoid.html">op_sparse_sigmoid</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -105,7 +105,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/op_threshold.html b/docs/dev/reference/op_threshold.html
index 1a0b9c3c8..103608874 100644
--- a/docs/dev/reference/op_threshold.html
+++ b/docs/dev/reference/op_threshold.html
@@ -12,7 +12,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -97,8 +97,8 @@ <h2 id="examples">Examples<a class="anchor" aria-label="anchor" href="#examples"
     </div>
     <div class="section level2">
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
-    <div class="dont-index"><p>Other nn ops: <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br></p>
-<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
+    <div class="dont-index"><p>Other nn ops: <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_layer_normalization.html">op_layer_normalization</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparse_sigmoid.html">op_sparse_sigmoid</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br></p>
+<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_layer_normalization.html">op_layer_normalization</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparse_sigmoid.html">op_sparse_sigmoid</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -110,7 +110,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/op_tile.html b/docs/dev/reference/op_tile.html
index 804bc7a4e..733bc8ce3 100644
--- a/docs/dev/reference/op_tile.html
+++ b/docs/dev/reference/op_tile.html
@@ -16,7 +16,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -94,8 +94,8 @@ <h2 id="value">Value<a class="anchor" aria-label="anchor" href="#value"></a></h2
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
     <div class="dont-index">
 <ul><li><p><a href="https://keras.io/api/ops/numpy#tile-function" class="external-link">https://keras.io/api/ops/numpy#tile-function</a></p></li>
-</ul><p>Other numpy ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p>
-<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
+</ul><p>Other numpy ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p>
+<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_layer_normalization.html">op_layer_normalization</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparse_sigmoid.html">op_sparse_sigmoid</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -107,7 +107,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/op_top_k.html b/docs/dev/reference/op_top_k.html
index 0ea451507..02d4aef67 100644
--- a/docs/dev/reference/op_top_k.html
+++ b/docs/dev/reference/op_top_k.html
@@ -8,7 +8,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -114,7 +114,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
     <div class="dont-index">
 <ul><li><p><a href="https://keras.io/api/ops/core#topk-function" class="external-link">https://keras.io/api/ops/core#topk-function</a></p></li>
 </ul><p>Other math ops: <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br></p>
-<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
+<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_layer_normalization.html">op_layer_normalization</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparse_sigmoid.html">op_sparse_sigmoid</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -126,7 +126,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/op_trace.html b/docs/dev/reference/op_trace.html
index afa225f56..eb36d0d86 100644
--- a/docs/dev/reference/op_trace.html
+++ b/docs/dev/reference/op_trace.html
@@ -20,7 +20,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -113,8 +113,8 @@ <h2 id="value">Value<a class="anchor" aria-label="anchor" href="#value"></a></h2
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
     <div class="dont-index">
 <ul><li><p><a href="https://keras.io/api/ops/numpy#trace-function" class="external-link">https://keras.io/api/ops/numpy#trace-function</a></p></li>
-</ul><p>Other numpy ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p>
-<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
+</ul><p>Other numpy ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p>
+<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_layer_normalization.html">op_layer_normalization</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparse_sigmoid.html">op_sparse_sigmoid</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -126,7 +126,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/op_transpose.html b/docs/dev/reference/op_transpose.html
index f931d591e..2a4774594 100644
--- a/docs/dev/reference/op_transpose.html
+++ b/docs/dev/reference/op_transpose.html
@@ -8,7 +8,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -83,8 +83,8 @@ <h2 id="value">Value<a class="anchor" aria-label="anchor" href="#value"></a></h2
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
     <div class="dont-index">
 <ul><li><p><a href="https://keras.io/api/ops/numpy#transpose-function" class="external-link">https://keras.io/api/ops/numpy#transpose-function</a></p></li>
-</ul><p>Other numpy ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p>
-<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
+</ul><p>Other numpy ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p>
+<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_layer_normalization.html">op_layer_normalization</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparse_sigmoid.html">op_sparse_sigmoid</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -96,7 +96,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/op_tri.html b/docs/dev/reference/op_tri.html
index 52fab0cd6..f8bc67326 100644
--- a/docs/dev/reference/op_tri.html
+++ b/docs/dev/reference/op_tri.html
@@ -8,7 +8,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -93,8 +93,8 @@ <h2 id="value">Value<a class="anchor" aria-label="anchor" href="#value"></a></h2
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
     <div class="dont-index">
 <ul><li><p><a href="https://keras.io/api/ops/numpy#tri-function" class="external-link">https://keras.io/api/ops/numpy#tri-function</a></p></li>
-</ul><p>Other numpy ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p>
-<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
+</ul><p>Other numpy ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p>
+<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_layer_normalization.html">op_layer_normalization</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparse_sigmoid.html">op_sparse_sigmoid</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -106,7 +106,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/op_tril.html b/docs/dev/reference/op_tril.html
index 4aa9c54c2..e36e74563 100644
--- a/docs/dev/reference/op_tril.html
+++ b/docs/dev/reference/op_tril.html
@@ -10,7 +10,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -86,8 +86,8 @@ <h2 id="value">Value<a class="anchor" aria-label="anchor" href="#value"></a></h2
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
     <div class="dont-index">
 <ul><li><p><a href="https://keras.io/api/ops/numpy#tril-function" class="external-link">https://keras.io/api/ops/numpy#tril-function</a></p></li>
-</ul><p>Other numpy ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p>
-<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
+</ul><p>Other numpy ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p>
+<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_layer_normalization.html">op_layer_normalization</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparse_sigmoid.html">op_sparse_sigmoid</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -99,7 +99,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/op_triu.html b/docs/dev/reference/op_triu.html
index c3a2042a6..0f97022da 100644
--- a/docs/dev/reference/op_triu.html
+++ b/docs/dev/reference/op_triu.html
@@ -10,7 +10,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -86,8 +86,8 @@ <h2 id="value">Value<a class="anchor" aria-label="anchor" href="#value"></a></h2
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
     <div class="dont-index">
 <ul><li><p><a href="https://keras.io/api/ops/numpy#triu-function" class="external-link">https://keras.io/api/ops/numpy#triu-function</a></p></li>
-</ul><p>Other numpy ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p>
-<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
+</ul><p>Other numpy ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p>
+<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_layer_normalization.html">op_layer_normalization</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparse_sigmoid.html">op_sparse_sigmoid</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -99,7 +99,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/op_trunc.html b/docs/dev/reference/op_trunc.html
index 0df61fa0f..45ce81561 100644
--- a/docs/dev/reference/op_trunc.html
+++ b/docs/dev/reference/op_trunc.html
@@ -12,7 +12,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -89,8 +89,8 @@ <h2 id="examples">Examples<a class="anchor" aria-label="anchor" href="#examples"
     </div>
     <div class="section level2">
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
-    <div class="dont-index"><p>Other numpy ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p>
-<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
+    <div class="dont-index"><p>Other numpy ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p>
+<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_layer_normalization.html">op_layer_normalization</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparse_sigmoid.html">op_sparse_sigmoid</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -102,7 +102,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/op_unravel_index.html b/docs/dev/reference/op_unravel_index.html
index a1d1adeb1..0fc3be181 100644
--- a/docs/dev/reference/op_unravel_index.html
+++ b/docs/dev/reference/op_unravel_index.html
@@ -8,7 +8,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -92,8 +92,8 @@ <h2 id="examples">Examples<a class="anchor" aria-label="anchor" href="#examples"
     </div>
     <div class="section level2">
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
-    <div class="dont-index"><p>Other nn ops: <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br></p>
-<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
+    <div class="dont-index"><p>Other nn ops: <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_layer_normalization.html">op_layer_normalization</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparse_sigmoid.html">op_sparse_sigmoid</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br></p>
+<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_layer_normalization.html">op_layer_normalization</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparse_sigmoid.html">op_sparse_sigmoid</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -105,7 +105,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/op_unstack.html b/docs/dev/reference/op_unstack.html
index 287d1d1e4..ed3b69554 100644
--- a/docs/dev/reference/op_unstack.html
+++ b/docs/dev/reference/op_unstack.html
@@ -8,7 +8,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -115,7 +115,7 @@ <h2 id="examples">Examples<a class="anchor" aria-label="anchor" href="#examples"
     <div class="section level2">
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
     <div class="dont-index"><p>Other core ops: <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br></p>
-<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
+<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_layer_normalization.html">op_layer_normalization</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparse_sigmoid.html">op_sparse_sigmoid</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -127,7 +127,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/op_var.html b/docs/dev/reference/op_var.html
index 4137aaaac..5a67d58b0 100644
--- a/docs/dev/reference/op_var.html
+++ b/docs/dev/reference/op_var.html
@@ -8,7 +8,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -88,8 +88,8 @@ <h2 id="value">Value<a class="anchor" aria-label="anchor" href="#value"></a></h2
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
     <div class="dont-index">
 <ul><li><p><a href="https://keras.io/api/ops/numpy#var-function" class="external-link">https://keras.io/api/ops/numpy#var-function</a></p></li>
-</ul><p>Other numpy ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p>
-<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
+</ul><p>Other numpy ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p>
+<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_layer_normalization.html">op_layer_normalization</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparse_sigmoid.html">op_sparse_sigmoid</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -101,7 +101,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/op_vdot.html b/docs/dev/reference/op_vdot.html
index 8566636c0..33e854a39 100644
--- a/docs/dev/reference/op_vdot.html
+++ b/docs/dev/reference/op_vdot.html
@@ -12,7 +12,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -89,8 +89,8 @@ <h2 id="value">Value<a class="anchor" aria-label="anchor" href="#value"></a></h2
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
     <div class="dont-index">
 <ul><li><p><a href="https://keras.io/api/ops/numpy#vdot-function" class="external-link">https://keras.io/api/ops/numpy#vdot-function</a></p></li>
-</ul><p>Other numpy ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p>
-<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
+</ul><p>Other numpy ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p>
+<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_layer_normalization.html">op_layer_normalization</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparse_sigmoid.html">op_sparse_sigmoid</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -102,7 +102,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/op_vectorize.html b/docs/dev/reference/op_vectorize.html
index 4a4ae023c..77ff5c39f 100644
--- a/docs/dev/reference/op_vectorize.html
+++ b/docs/dev/reference/op_vectorize.html
@@ -8,7 +8,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -112,8 +112,8 @@ <h2 id="examples">Examples<a class="anchor" aria-label="anchor" href="#examples"
     </div>
     <div class="section level2">
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
-    <div class="dont-index"><p>Other numpy ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p>
-<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
+    <div class="dont-index"><p>Other numpy ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p>
+<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_layer_normalization.html">op_layer_normalization</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparse_sigmoid.html">op_sparse_sigmoid</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -125,7 +125,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/op_vectorized_map.html b/docs/dev/reference/op_vectorized_map.html
index e25d40d3c..ff289d191 100644
--- a/docs/dev/reference/op_vectorized_map.html
+++ b/docs/dev/reference/op_vectorized_map.html
@@ -12,7 +12,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -240,7 +240,7 @@ <h2 id="debugging-f">Debugging <code>f</code><a class="anchor" aria-label="ancho
     <div class="section level2">
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
     <div class="dont-index"><p>Other core ops: <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br></p>
-<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
+<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_layer_normalization.html">op_layer_normalization</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparse_sigmoid.html">op_sparse_sigmoid</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -252,7 +252,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/op_view_as_complex.html b/docs/dev/reference/op_view_as_complex.html
new file mode 100644
index 000000000..39545b01f
--- /dev/null
+++ b/docs/dev/reference/op_view_as_complex.html
@@ -0,0 +1,115 @@
+<!DOCTYPE html>
+<!-- Generated by pkgdown: do not edit by hand --><html lang="en"><head><meta http-equiv="Content-Type" content="text/html; charset=UTF-8"><meta charset="utf-8"><meta http-equiv="X-UA-Compatible" content="IE=edge"><meta name="viewport" content="width=device-width, initial-scale=1, shrink-to-fit=no"><title>Convert a real tensor with two channels into a complex tensor. — op_view_as_complex • keras3</title><!-- favicons --><link rel="icon" type="image/png" sizes="96x96" href="../favicon-96x96.png"><link rel="icon" type="”image/svg+xml”" href="../favicon.svg"><link rel="apple-touch-icon" sizes="180x180" href="../apple-touch-icon.png"><link rel="icon" sizes="any" href="../favicon.ico"><link rel="manifest" href="../site.webmanifest"><script src="../deps/jquery-3.6.0/jquery-3.6.0.min.js"></script><meta name="viewport" content="width=device-width, initial-scale=1, shrink-to-fit=no"><link href="../deps/bootstrap-5.3.1/bootstrap.min.css" rel="stylesheet"><script src="../deps/bootstrap-5.3.1/bootstrap.bundle.min.js"></script><link href="../deps/font-awesome-6.5.2/css/all.min.css" rel="stylesheet"><link href="../deps/font-awesome-6.5.2/css/v4-shims.min.css" rel="stylesheet"><script src="../deps/headroom-0.11.0/headroom.min.js"></script><script src="../deps/headroom-0.11.0/jQuery.headroom.min.js"></script><script src="../deps/bootstrap-toc-1.0.1/bootstrap-toc.min.js"></script><script src="../deps/clipboard.js-2.0.11/clipboard.min.js"></script><script src="../deps/search-1.0.0/autocomplete.jquery.min.js"></script><script src="../deps/search-1.0.0/fuse.min.js"></script><script src="../deps/search-1.0.0/mark.min.js"></script><!-- pkgdown --><script src="../pkgdown.js"></script><link href="../extra.css" rel="stylesheet"><meta property="og:title" content="Convert a real tensor with two channels into a complex tensor. — op_view_as_complex"><meta name="description" content="Expects a real-valued tensor whose last dimension has size 2, holding the
+real and imaginary parts. Returns the corresponding complex tensor with the
+last dimension removed."><meta property="og:description" content="Expects a real-valued tensor whose last dimension has size 2, holding the
+real and imaginary parts. Returns the corresponding complex tensor with the
+last dimension removed."><meta name="robots" content="noindex"><!-- Google Tag Manager --><script>(function(w,d,s,l,i){w[l]=w[l]||[];w[l].push({'gtm.start': new Date().getTime(),event:'gtm.js'});var f=d.getElementsByTagName(s)[0],j=d.createElement(s),dl=l!='dataLayer'?'&l='+l:'';j.async=true;j.src='https://www.googletagmanager.com/gtm.js?id='+i+dl;f.parentNode.insertBefore(j,f);})(window,document,'script','dataLayer','GTM-KHBDBW7');</script><!-- End Google Tag Manager --></head><body>
+    <a href="#main" class="visually-hidden-focusable">Skip to contents</a>
+    <!-- Google Tag Manager (noscript) --> <noscript><iframe src="https://www.googletagmanager.com/ns.html?id=GTM-KHBDBW7" height="0" width="0" style="display:none;visibility:hidden"></iframe></noscript> <!-- End Google Tag Manager (noscript) -->
+
+
+    <nav class="navbar navbar-expand-lg fixed-top bg-primary" data-bs-theme="inverse" aria-label="Site navigation"><div class="container">
+
+    <a class="navbar-brand me-2" href="../index.html">keras3</a>
+
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
+
+
+    <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
+      <span class="navbar-toggler-icon"></span>
+    </button>
+
+    <div id="navbar" class="collapse navbar-collapse ms-3">
+      <ul class="navbar-nav me-auto"><li class="nav-item"><a class="nav-link" href="../articles/getting_started.html">Getting Started</a></li>
+<li class="nav-item dropdown">
+  <button class="nav-link dropdown-toggle" type="button" id="dropdown-guides" data-bs-toggle="dropdown" aria-expanded="false" aria-haspopup="true">Guides</button>
+  <ul class="dropdown-menu" aria-labelledby="dropdown-guides"><li><h6 class="dropdown-header" data-toc-skip>Model definition</h6></li>
+    <li><a class="dropdown-item" href="../articles/sequential_model.html">Sequential Model</a></li>
+    <li><a class="dropdown-item" href="../articles/functional_api.html">Functional API</a></li>
+    <li><h6 class="dropdown-header" data-toc-skip>Extending and customizing</h6></li>
+    <li><a class="dropdown-item" href="../articles/training_with_built_in_methods.html">Training &amp; evaluation with the built-in methods</a></li>
+    <li><a class="dropdown-item" href="../articles/custom_train_step_in_tensorflow.html">Customizing `fit()` with Tensorflow</a></li>
+    <li><a class="dropdown-item" href="../articles/writing_your_own_callbacks.html">Writing your own callbacks</a></li>
+    <li><a class="dropdown-item" href="../articles/making_new_layers_and_models_via_subclassing.html">Making new layers and models via subclassing</a></li>
+    <li><a class="dropdown-item" href="../articles/writing_a_custom_training_loop_in_tensorflow.html">Writing a training loop from scratch in TensorFlow</a></li>
+    <li><a class="dropdown-item" href="../articles/serialization_and_saving.html">Serialization and Saving</a></li>
+    <li><h6 class="dropdown-header" data-toc-skip>Other topics</h6></li>
+    <li><a class="dropdown-item" href="../articles/transfer_learning.html">Transfer learning and fine tuning</a></li>
+    <li><a class="dropdown-item" href="../articles/distributed_training_with_tensorflow.html">Distributed training with TensorFlow</a></li>
+    <li><a class="dropdown-item" href="../articles/distribution.html">Distributed training with Jax</a></li>
+  </ul></li>
+<li class="nav-item"><a class="nav-link" href="../articles/examples/index.html">Examples</a></li>
+<li class="active nav-item"><a class="nav-link" href="../reference/index.html">Reference</a></li>
+<li class="nav-item"><a class="nav-link" href="../news/index.html">News</a></li>
+      </ul><ul class="navbar-nav"><li class="nav-item"><form class="form-inline" role="search">
+ <input class="form-control" type="search" name="search-input" id="search-input" autocomplete="off" aria-label="Search site" placeholder="Search for" data-search-index="../search.json"></form></li>
+<li class="nav-item"><a class="external-link nav-link" href="https://github.com/rstudio/keras3/" aria-label="GitHub"><span class="fa fab fa-github fa-lg"></span></a></li>
+      </ul></div>
+
+
+  </div>
+</nav><div class="container template-reference-topic">
+<div class="row">
+  <main id="main" class="col-md-9"><div class="page-header">
+
+      <h1>Convert a real tensor with two channels into a complex tensor.</h1>
+      <small class="dont-index">Source: <a href="https://github.com/rstudio/keras3/blob/HEAD/R/ops.R" class="external-link"><code>R/ops.R</code></a></small>
+      <div class="d-none name"><code>op_view_as_complex.Rd</code></div>
+    </div>
+
+    <div class="ref-description section level2">
+    <p>Expects a real-valued tensor whose last dimension has size <code>2</code>, holding the
+real and imaginary parts. Returns the corresponding complex tensor with the
+last dimension removed.</p>
+    </div>
+
+    <div class="section level2">
+    <h2 id="ref-usage">Usage<a class="anchor" aria-label="anchor" href="#ref-usage"></a></h2>
+    <div class="sourceCode"><pre class="sourceCode r"><code><span><span class="fu">op_view_as_complex</span><span class="op">(</span><span class="va">x</span><span class="op">)</span></span></code></pre></div>
+    </div>
+
+    <div class="section level2">
+    <h2 id="arguments">Arguments<a class="anchor" aria-label="anchor" href="#arguments"></a></h2>
+
+
+<dl><dt id="arg-x">x<a class="anchor" aria-label="anchor" href="#arg-x"></a></dt>
+<dd><p>Real-valued tensor whose trailing dimension encodes the complex components.</p></dd>
+
+</dl></div>
+    <div class="section level2">
+    <h2 id="value">Value<a class="anchor" aria-label="anchor" href="#value"></a></h2>
+    <p>A complex tensor with shape <code>op_shape(x)[-length(op_shape(x))]</code>.</p>
+    </div>
+    <div class="section level2">
+    <h2 id="examples">Examples<a class="anchor" aria-label="anchor" href="#examples"></a></h2>
+    <p></p><div class="sourceCode r"><pre><code><span><span class="va">x</span> <span class="op">&lt;-</span> <span class="fu"><a href="../reference/op_array.html">op_array</a></span><span class="op">(</span><span class="fu"><a href="https://rdrr.io/r/base/matrix.html" class="external-link">matrix</a></span><span class="op">(</span><span class="fu"><a href="https://rdrr.io/r/base/c.html" class="external-link">c</a></span><span class="op">(</span><span class="fl">1</span>, <span class="fl">2</span>, <span class="fl">3</span>, <span class="fl">4</span><span class="op">)</span>, nrow <span class="op">=</span> <span class="fl">2</span>, byrow <span class="op">=</span> <span class="cn">TRUE</span><span class="op">)</span><span class="op">)</span></span>
+<span><span class="fu"><a href="../reference/op_view_as_complex.html">op_view_as_complex</a></span><span class="op">(</span><span class="va">x</span><span class="op">)</span></span></code></pre><p></p></div>
+<p></p><div class="sourceCode"><pre><code><span id="cb1-1"><a href="#cb1-1" tabindex="-1"></a><span class="do">## tf.Tensor([1.+2.j 3.+4.j], shape=(2), dtype=complex64)</span></span>
+<span id="cb1-2"><a href="#cb1-2" tabindex="-1"></a></span></code></pre><p></p></div>
+    </div>
+    <div class="section level2">
+    <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
+    <div class="dont-index"><p>Other numpy ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p>
+<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_layer_normalization.html">op_layer_normalization</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparse_sigmoid.html">op_sparse_sigmoid</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
+    </div>
+
+  </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
+    </nav></aside></div>
+
+
+    <footer><div class="pkgdown-footer-left">
+  <p>Developed by Tomasz Kalinowski, JJ Allaire, François Chollet, Posit Software, PBC, Google.</p>
+</div>
+
+<div class="pkgdown-footer-right">
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
+</div>
+
+    </footer></div>
+
+
+
+
+
+  </body></html>
+
diff --git a/docs/dev/reference/op_view_as_real.html b/docs/dev/reference/op_view_as_real.html
new file mode 100644
index 000000000..0768cd900
--- /dev/null
+++ b/docs/dev/reference/op_view_as_real.html
@@ -0,0 +1,115 @@
+<!DOCTYPE html>
+<!-- Generated by pkgdown: do not edit by hand --><html lang="en"><head><meta http-equiv="Content-Type" content="text/html; charset=UTF-8"><meta charset="utf-8"><meta http-equiv="X-UA-Compatible" content="IE=edge"><meta name="viewport" content="width=device-width, initial-scale=1, shrink-to-fit=no"><title>Convert a complex tensor into a stacked real representation. — op_view_as_real • keras3</title><!-- favicons --><link rel="icon" type="image/png" sizes="96x96" href="../favicon-96x96.png"><link rel="icon" type="”image/svg+xml”" href="../favicon.svg"><link rel="apple-touch-icon" sizes="180x180" href="../apple-touch-icon.png"><link rel="icon" sizes="any" href="../favicon.ico"><link rel="manifest" href="../site.webmanifest"><script src="../deps/jquery-3.6.0/jquery-3.6.0.min.js"></script><meta name="viewport" content="width=device-width, initial-scale=1, shrink-to-fit=no"><link href="../deps/bootstrap-5.3.1/bootstrap.min.css" rel="stylesheet"><script src="../deps/bootstrap-5.3.1/bootstrap.bundle.min.js"></script><link href="../deps/font-awesome-6.5.2/css/all.min.css" rel="stylesheet"><link href="../deps/font-awesome-6.5.2/css/v4-shims.min.css" rel="stylesheet"><script src="../deps/headroom-0.11.0/headroom.min.js"></script><script src="../deps/headroom-0.11.0/jQuery.headroom.min.js"></script><script src="../deps/bootstrap-toc-1.0.1/bootstrap-toc.min.js"></script><script src="../deps/clipboard.js-2.0.11/clipboard.min.js"></script><script src="../deps/search-1.0.0/autocomplete.jquery.min.js"></script><script src="../deps/search-1.0.0/fuse.min.js"></script><script src="../deps/search-1.0.0/mark.min.js"></script><!-- pkgdown --><script src="../pkgdown.js"></script><link href="../extra.css" rel="stylesheet"><meta property="og:title" content="Convert a complex tensor into a stacked real representation. — op_view_as_real"><meta name="description" content="Produces a real-valued tensor where the last dimension gathers the real and
+imaginary parts of the complex input."><meta property="og:description" content="Produces a real-valued tensor where the last dimension gathers the real and
+imaginary parts of the complex input."><meta name="robots" content="noindex"><!-- Google Tag Manager --><script>(function(w,d,s,l,i){w[l]=w[l]||[];w[l].push({'gtm.start': new Date().getTime(),event:'gtm.js'});var f=d.getElementsByTagName(s)[0],j=d.createElement(s),dl=l!='dataLayer'?'&l='+l:'';j.async=true;j.src='https://www.googletagmanager.com/gtm.js?id='+i+dl;f.parentNode.insertBefore(j,f);})(window,document,'script','dataLayer','GTM-KHBDBW7');</script><!-- End Google Tag Manager --></head><body>
+    <a href="#main" class="visually-hidden-focusable">Skip to contents</a>
+    <!-- Google Tag Manager (noscript) --> <noscript><iframe src="https://www.googletagmanager.com/ns.html?id=GTM-KHBDBW7" height="0" width="0" style="display:none;visibility:hidden"></iframe></noscript> <!-- End Google Tag Manager (noscript) -->
+
+
+    <nav class="navbar navbar-expand-lg fixed-top bg-primary" data-bs-theme="inverse" aria-label="Site navigation"><div class="container">
+
+    <a class="navbar-brand me-2" href="../index.html">keras3</a>
+
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
+
+
+    <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
+      <span class="navbar-toggler-icon"></span>
+    </button>
+
+    <div id="navbar" class="collapse navbar-collapse ms-3">
+      <ul class="navbar-nav me-auto"><li class="nav-item"><a class="nav-link" href="../articles/getting_started.html">Getting Started</a></li>
+<li class="nav-item dropdown">
+  <button class="nav-link dropdown-toggle" type="button" id="dropdown-guides" data-bs-toggle="dropdown" aria-expanded="false" aria-haspopup="true">Guides</button>
+  <ul class="dropdown-menu" aria-labelledby="dropdown-guides"><li><h6 class="dropdown-header" data-toc-skip>Model definition</h6></li>
+    <li><a class="dropdown-item" href="../articles/sequential_model.html">Sequential Model</a></li>
+    <li><a class="dropdown-item" href="../articles/functional_api.html">Functional API</a></li>
+    <li><h6 class="dropdown-header" data-toc-skip>Extending and customizing</h6></li>
+    <li><a class="dropdown-item" href="../articles/training_with_built_in_methods.html">Training &amp; evaluation with the built-in methods</a></li>
+    <li><a class="dropdown-item" href="../articles/custom_train_step_in_tensorflow.html">Customizing `fit()` with Tensorflow</a></li>
+    <li><a class="dropdown-item" href="../articles/writing_your_own_callbacks.html">Writing your own callbacks</a></li>
+    <li><a class="dropdown-item" href="../articles/making_new_layers_and_models_via_subclassing.html">Making new layers and models via subclassing</a></li>
+    <li><a class="dropdown-item" href="../articles/writing_a_custom_training_loop_in_tensorflow.html">Writing a training loop from scratch in TensorFlow</a></li>
+    <li><a class="dropdown-item" href="../articles/serialization_and_saving.html">Serialization and Saving</a></li>
+    <li><h6 class="dropdown-header" data-toc-skip>Other topics</h6></li>
+    <li><a class="dropdown-item" href="../articles/transfer_learning.html">Transfer learning and fine tuning</a></li>
+    <li><a class="dropdown-item" href="../articles/distributed_training_with_tensorflow.html">Distributed training with TensorFlow</a></li>
+    <li><a class="dropdown-item" href="../articles/distribution.html">Distributed training with Jax</a></li>
+  </ul></li>
+<li class="nav-item"><a class="nav-link" href="../articles/examples/index.html">Examples</a></li>
+<li class="active nav-item"><a class="nav-link" href="../reference/index.html">Reference</a></li>
+<li class="nav-item"><a class="nav-link" href="../news/index.html">News</a></li>
+      </ul><ul class="navbar-nav"><li class="nav-item"><form class="form-inline" role="search">
+ <input class="form-control" type="search" name="search-input" id="search-input" autocomplete="off" aria-label="Search site" placeholder="Search for" data-search-index="../search.json"></form></li>
+<li class="nav-item"><a class="external-link nav-link" href="https://github.com/rstudio/keras3/" aria-label="GitHub"><span class="fa fab fa-github fa-lg"></span></a></li>
+      </ul></div>
+
+
+  </div>
+</nav><div class="container template-reference-topic">
+<div class="row">
+  <main id="main" class="col-md-9"><div class="page-header">
+
+      <h1>Convert a complex tensor into a stacked real representation.</h1>
+      <small class="dont-index">Source: <a href="https://github.com/rstudio/keras3/blob/HEAD/R/ops.R" class="external-link"><code>R/ops.R</code></a></small>
+      <div class="d-none name"><code>op_view_as_real.Rd</code></div>
+    </div>
+
+    <div class="ref-description section level2">
+    <p>Produces a real-valued tensor where the last dimension gathers the real and
+imaginary parts of the complex input.</p>
+    </div>
+
+    <div class="section level2">
+    <h2 id="ref-usage">Usage<a class="anchor" aria-label="anchor" href="#ref-usage"></a></h2>
+    <div class="sourceCode"><pre class="sourceCode r"><code><span><span class="fu">op_view_as_real</span><span class="op">(</span><span class="va">x</span><span class="op">)</span></span></code></pre></div>
+    </div>
+
+    <div class="section level2">
+    <h2 id="arguments">Arguments<a class="anchor" aria-label="anchor" href="#arguments"></a></h2>
+
+
+<dl><dt id="arg-x">x<a class="anchor" aria-label="anchor" href="#arg-x"></a></dt>
+<dd><p>Complex-valued tensor to be converted.</p></dd>
+
+</dl></div>
+    <div class="section level2">
+    <h2 id="value">Value<a class="anchor" aria-label="anchor" href="#value"></a></h2>
+    <p>A real tensor with shape <code>c(op_shape(x), 2)</code> containing real and imaginary parts.</p>
+    </div>
+    <div class="section level2">
+    <h2 id="examples">Examples<a class="anchor" aria-label="anchor" href="#examples"></a></h2>
+    <p></p><div class="sourceCode r"><pre><code><span><span class="va">x</span> <span class="op">&lt;-</span> <span class="fu"><a href="../reference/op_array.html">op_array</a></span><span class="op">(</span><span class="fu"><a href="https://rdrr.io/r/base/matrix.html" class="external-link">matrix</a></span><span class="op">(</span><span class="fu"><a href="https://rdrr.io/r/base/c.html" class="external-link">c</a></span><span class="op">(</span><span class="fl">1</span>, <span class="fl">2</span>, <span class="fl">3</span>, <span class="fl">4</span><span class="op">)</span>, nrow <span class="op">=</span> <span class="fl">2</span>, byrow <span class="op">=</span> <span class="cn">TRUE</span><span class="op">)</span><span class="op">)</span></span>
+<span><span class="va">z</span> <span class="op">&lt;-</span> <span class="fu"><a href="../reference/op_view_as_complex.html">op_view_as_complex</a></span><span class="op">(</span><span class="va">x</span><span class="op">)</span></span>
+<span><span class="fu"><a href="../reference/op_view_as_real.html">op_view_as_real</a></span><span class="op">(</span><span class="va">z</span><span class="op">)</span></span></code></pre><p></p></div>
+<p></p><div class="sourceCode"><pre><code><span id="cb1-1"><a href="#cb1-1" tabindex="-1"></a><span class="do">## tf.Tensor(</span></span>
+<span id="cb1-2"><a href="#cb1-2" tabindex="-1"></a><span class="do">## [[1. 2.]</span></span>
+<span id="cb1-3"><a href="#cb1-3" tabindex="-1"></a><span class="do">##  [3. 4.]], shape=(2, 2), dtype=float32)</span></span>
+<span id="cb1-4"><a href="#cb1-4" tabindex="-1"></a></span></code></pre><p></p></div>
+    </div>
+    <div class="section level2">
+    <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
+    <div class="dont-index"><p>Other numpy ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p>
+<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_layer_normalization.html">op_layer_normalization</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparse_sigmoid.html">op_sparse_sigmoid</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
+    </div>
+
+  </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
+    </nav></aside></div>
+
+
+    <footer><div class="pkgdown-footer-left">
+  <p>Developed by Tomasz Kalinowski, JJ Allaire, François Chollet, Posit Software, PBC, Google.</p>
+</div>
+
+<div class="pkgdown-footer-right">
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
+</div>
+
+    </footer></div>
+
+
+
+
+
+  </body></html>
+
diff --git a/docs/dev/reference/op_vstack.html b/docs/dev/reference/op_vstack.html
index 67e3887c1..168d21828 100644
--- a/docs/dev/reference/op_vstack.html
+++ b/docs/dev/reference/op_vstack.html
@@ -8,7 +8,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -78,8 +78,8 @@ <h2 id="value">Value<a class="anchor" aria-label="anchor" href="#value"></a></h2
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
     <div class="dont-index">
 <ul><li><p><a href="https://keras.io/api/ops/numpy#vstack-function" class="external-link">https://keras.io/api/ops/numpy#vstack-function</a></p></li>
-</ul><p>Other numpy ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p>
-<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
+</ul><p>Other numpy ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p>
+<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_layer_normalization.html">op_layer_normalization</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparse_sigmoid.html">op_sparse_sigmoid</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -91,7 +91,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/op_where.html b/docs/dev/reference/op_where.html
index 0f7241525..3d3297450 100644
--- a/docs/dev/reference/op_where.html
+++ b/docs/dev/reference/op_where.html
@@ -8,7 +8,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -87,8 +87,8 @@ <h2 id="value">Value<a class="anchor" aria-label="anchor" href="#value"></a></h2
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
     <div class="dont-index">
 <ul><li><p><a href="https://keras.io/api/ops/numpy#where-function" class="external-link">https://keras.io/api/ops/numpy#where-function</a></p></li>
-</ul><p>Other numpy ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p>
-<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
+</ul><p>Other numpy ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p>
+<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_layer_normalization.html">op_layer_normalization</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparse_sigmoid.html">op_sparse_sigmoid</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -100,7 +100,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/op_while_loop.html b/docs/dev/reference/op_while_loop.html
index 640a283c2..0e52c1b3e 100644
--- a/docs/dev/reference/op_while_loop.html
+++ b/docs/dev/reference/op_while_loop.html
@@ -8,7 +8,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -127,7 +127,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
     <div class="dont-index">
 <ul><li><p><a href="https://keras.io/api/ops/core#whileloop-function" class="external-link">https://keras.io/api/ops/core#whileloop-function</a></p></li>
 </ul><p>Other core ops: <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br></p>
-<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
+<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_layer_normalization.html">op_layer_normalization</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparse_sigmoid.html">op_sparse_sigmoid</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -139,7 +139,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/op_zeros.html b/docs/dev/reference/op_zeros.html
index 51f44d596..b555b4910 100644
--- a/docs/dev/reference/op_zeros.html
+++ b/docs/dev/reference/op_zeros.html
@@ -8,7 +8,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -82,8 +82,8 @@ <h2 id="value">Value<a class="anchor" aria-label="anchor" href="#value"></a></h2
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
     <div class="dont-index">
 <ul><li><p><a href="https://keras.io/api/ops/numpy#zeros-function" class="external-link">https://keras.io/api/ops/numpy#zeros-function</a></p></li>
-</ul><p>Other numpy ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p>
-<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
+</ul><p>Other numpy ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p>
+<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_layer_normalization.html">op_layer_normalization</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparse_sigmoid.html">op_sparse_sigmoid</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros_like.html">op_zeros_like</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -95,7 +95,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/op_zeros_like.html b/docs/dev/reference/op_zeros_like.html
index eb9b09544..642c27536 100644
--- a/docs/dev/reference/op_zeros_like.html
+++ b/docs/dev/reference/op_zeros_like.html
@@ -8,7 +8,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -82,8 +82,8 @@ <h2 id="value">Value<a class="anchor" aria-label="anchor" href="#value"></a></h2
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
     <div class="dont-index">
 <ul><li><p><a href="https://keras.io/api/ops/numpy#zeroslike-function" class="external-link">https://keras.io/api/ops/numpy#zeroslike-function</a></p></li>
-</ul><p>Other numpy ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br></p>
-<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br></p></div>
+</ul><p>Other numpy ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br></p>
+<p>Other ops: <br><code><a href="op_abs.html">op_abs</a>()</code> <br><code><a href="op_add.html">op_add</a>()</code> <br><code><a href="op_all.html">op_all</a>()</code> <br><code><a href="op_angle.html">op_angle</a>()</code> <br><code><a href="op_any.html">op_any</a>()</code> <br><code><a href="op_append.html">op_append</a>()</code> <br><code><a href="op_arange.html">op_arange</a>()</code> <br><code><a href="op_arccos.html">op_arccos</a>()</code> <br><code><a href="op_arccosh.html">op_arccosh</a>()</code> <br><code><a href="op_arcsin.html">op_arcsin</a>()</code> <br><code><a href="op_arcsinh.html">op_arcsinh</a>()</code> <br><code><a href="op_arctan.html">op_arctan</a>()</code> <br><code><a href="op_arctan2.html">op_arctan2</a>()</code> <br><code><a href="op_arctanh.html">op_arctanh</a>()</code> <br><code><a href="op_argmax.html">op_argmax</a>()</code> <br><code><a href="op_argmin.html">op_argmin</a>()</code> <br><code><a href="op_argpartition.html">op_argpartition</a>()</code> <br><code><a href="op_argsort.html">op_argsort</a>()</code> <br><code><a href="op_array.html">op_array</a>()</code> <br><code><a href="op_associative_scan.html">op_associative_scan</a>()</code> <br><code><a href="op_average.html">op_average</a>()</code> <br><code><a href="op_average_pool.html">op_average_pool</a>()</code> <br><code><a href="op_bartlett.html">op_bartlett</a>()</code> <br><code><a href="op_batch_normalization.html">op_batch_normalization</a>()</code> <br><code><a href="op_binary_crossentropy.html">op_binary_crossentropy</a>()</code> <br><code><a href="op_bincount.html">op_bincount</a>()</code> <br><code><a href="op_bitwise_and.html">op_bitwise_and</a>()</code> <br><code><a href="op_bitwise_invert.html">op_bitwise_invert</a>()</code> <br><code><a href="op_bitwise_left_shift.html">op_bitwise_left_shift</a>()</code> <br><code><a href="op_bitwise_not.html">op_bitwise_not</a>()</code> <br><code><a href="op_bitwise_or.html">op_bitwise_or</a>()</code> <br><code><a href="op_bitwise_right_shift.html">op_bitwise_right_shift</a>()</code> <br><code><a href="op_bitwise_xor.html">op_bitwise_xor</a>()</code> <br><code><a href="op_blackman.html">op_blackman</a>()</code> <br><code><a href="op_broadcast_to.html">op_broadcast_to</a>()</code> <br><code><a href="op_cast.html">op_cast</a>()</code> <br><code><a href="op_categorical_crossentropy.html">op_categorical_crossentropy</a>()</code> <br><code><a href="op_cbrt.html">op_cbrt</a>()</code> <br><code><a href="op_ceil.html">op_ceil</a>()</code> <br><code><a href="op_celu.html">op_celu</a>()</code> <br><code><a href="op_cholesky.html">op_cholesky</a>()</code> <br><code><a href="op_clip.html">op_clip</a>()</code> <br><code><a href="op_concatenate.html">op_concatenate</a>()</code> <br><code><a href="op_cond.html">op_cond</a>()</code> <br><code><a href="op_conj.html">op_conj</a>()</code> <br><code><a href="op_conv.html">op_conv</a>()</code> <br><code><a href="op_conv_transpose.html">op_conv_transpose</a>()</code> <br><code><a href="op_convert_to_numpy.html">op_convert_to_numpy</a>()</code> <br><code><a href="op_convert_to_tensor.html">op_convert_to_tensor</a>()</code> <br><code><a href="op_copy.html">op_copy</a>()</code> <br><code><a href="op_corrcoef.html">op_corrcoef</a>()</code> <br><code><a href="op_correlate.html">op_correlate</a>()</code> <br><code><a href="op_cos.html">op_cos</a>()</code> <br><code><a href="op_cosh.html">op_cosh</a>()</code> <br><code><a href="op_count_nonzero.html">op_count_nonzero</a>()</code> <br><code><a href="op_cross.html">op_cross</a>()</code> <br><code><a href="op_ctc_decode.html">op_ctc_decode</a>()</code> <br><code><a href="op_ctc_loss.html">op_ctc_loss</a>()</code> <br><code><a href="op_cumprod.html">op_cumprod</a>()</code> <br><code><a href="op_cumsum.html">op_cumsum</a>()</code> <br><code><a href="op_custom_gradient.html">op_custom_gradient</a>()</code> <br><code><a href="op_deg2rad.html">op_deg2rad</a>()</code> <br><code><a href="op_depthwise_conv.html">op_depthwise_conv</a>()</code> <br><code><a href="op_det.html">op_det</a>()</code> <br><code><a href="op_diag.html">op_diag</a>()</code> <br><code><a href="op_diagflat.html">op_diagflat</a>()</code> <br><code><a href="op_diagonal.html">op_diagonal</a>()</code> <br><code><a href="op_diff.html">op_diff</a>()</code> <br><code><a href="op_digitize.html">op_digitize</a>()</code> <br><code><a href="op_divide.html">op_divide</a>()</code> <br><code><a href="op_divide_no_nan.html">op_divide_no_nan</a>()</code> <br><code><a href="op_dot.html">op_dot</a>()</code> <br><code><a href="op_dot_product_attention.html">op_dot_product_attention</a>()</code> <br><code><a href="op_dtype.html">op_dtype</a>()</code> <br><code><a href="op_eig.html">op_eig</a>()</code> <br><code><a href="op_eigh.html">op_eigh</a>()</code> <br><code><a href="op_einsum.html">op_einsum</a>()</code> <br><code><a href="op_elu.html">op_elu</a>()</code> <br><code><a href="op_empty.html">op_empty</a>()</code> <br><code><a href="op_equal.html">op_equal</a>()</code> <br><code><a href="op_erf.html">op_erf</a>()</code> <br><code><a href="op_erfinv.html">op_erfinv</a>()</code> <br><code><a href="op_exp.html">op_exp</a>()</code> <br><code><a href="op_exp2.html">op_exp2</a>()</code> <br><code><a href="op_expand_dims.html">op_expand_dims</a>()</code> <br><code><a href="op_expm1.html">op_expm1</a>()</code> <br><code><a href="op_extract_sequences.html">op_extract_sequences</a>()</code> <br><code><a href="op_eye.html">op_eye</a>()</code> <br><code><a href="op_fft.html">op_fft</a>()</code> <br><code><a href="op_fft2.html">op_fft2</a>()</code> <br><code><a href="op_flip.html">op_flip</a>()</code> <br><code><a href="op_floor.html">op_floor</a>()</code> <br><code><a href="op_floor_divide.html">op_floor_divide</a>()</code> <br><code><a href="op_fori_loop.html">op_fori_loop</a>()</code> <br><code><a href="op_full.html">op_full</a>()</code> <br><code><a href="op_full_like.html">op_full_like</a>()</code> <br><code><a href="op_gelu.html">op_gelu</a>()</code> <br><code><a href="op_get_item.html">op_get_item</a>()</code> <br><code><a href="op_glu.html">op_glu</a>()</code> <br><code><a href="op_greater.html">op_greater</a>()</code> <br><code><a href="op_greater_equal.html">op_greater_equal</a>()</code> <br><code><a href="op_hamming.html">op_hamming</a>()</code> <br><code><a href="op_hanning.html">op_hanning</a>()</code> <br><code><a href="op_hard_shrink.html">op_hard_shrink</a>()</code> <br><code><a href="op_hard_sigmoid.html">op_hard_sigmoid</a>()</code> <br><code><a href="op_hard_silu.html">op_hard_silu</a>()</code> <br><code><a href="op_hard_tanh.html">op_hard_tanh</a>()</code> <br><code><a href="op_heaviside.html">op_heaviside</a>()</code> <br><code><a href="op_histogram.html">op_histogram</a>()</code> <br><code><a href="op_hstack.html">op_hstack</a>()</code> <br><code><a href="op_identity.html">op_identity</a>()</code> <br><code><a href="op_ifft2.html">op_ifft2</a>()</code> <br><code><a href="op_imag.html">op_imag</a>()</code> <br><code><a href="op_image_affine_transform.html">op_image_affine_transform</a>()</code> <br><code><a href="op_image_crop.html">op_image_crop</a>()</code> <br><code><a href="op_image_extract_patches.html">op_image_extract_patches</a>()</code> <br><code><a href="op_image_gaussian_blur.html">op_image_gaussian_blur</a>()</code> <br><code><a href="op_image_hsv_to_rgb.html">op_image_hsv_to_rgb</a>()</code> <br><code><a href="op_image_map_coordinates.html">op_image_map_coordinates</a>()</code> <br><code><a href="op_image_pad.html">op_image_pad</a>()</code> <br><code><a href="op_image_perspective_transform.html">op_image_perspective_transform</a>()</code> <br><code><a href="op_image_resize.html">op_image_resize</a>()</code> <br><code><a href="op_image_rgb_to_grayscale.html">op_image_rgb_to_grayscale</a>()</code> <br><code><a href="op_image_rgb_to_hsv.html">op_image_rgb_to_hsv</a>()</code> <br><code><a href="op_in_top_k.html">op_in_top_k</a>()</code> <br><code><a href="op_inner.html">op_inner</a>()</code> <br><code><a href="op_inv.html">op_inv</a>()</code> <br><code><a href="op_irfft.html">op_irfft</a>()</code> <br><code><a href="op_is_tensor.html">op_is_tensor</a>()</code> <br><code><a href="op_isclose.html">op_isclose</a>()</code> <br><code><a href="op_isfinite.html">op_isfinite</a>()</code> <br><code><a href="op_isinf.html">op_isinf</a>()</code> <br><code><a href="op_isnan.html">op_isnan</a>()</code> <br><code><a href="op_istft.html">op_istft</a>()</code> <br><code><a href="op_kaiser.html">op_kaiser</a>()</code> <br><code><a href="op_layer_normalization.html">op_layer_normalization</a>()</code> <br><code><a href="op_leaky_relu.html">op_leaky_relu</a>()</code> <br><code><a href="op_left_shift.html">op_left_shift</a>()</code> <br><code><a href="op_less.html">op_less</a>()</code> <br><code><a href="op_less_equal.html">op_less_equal</a>()</code> <br><code><a href="op_linspace.html">op_linspace</a>()</code> <br><code><a href="op_log.html">op_log</a>()</code> <br><code><a href="op_log10.html">op_log10</a>()</code> <br><code><a href="op_log1p.html">op_log1p</a>()</code> <br><code><a href="op_log2.html">op_log2</a>()</code> <br><code><a href="op_log_sigmoid.html">op_log_sigmoid</a>()</code> <br><code><a href="op_log_softmax.html">op_log_softmax</a>()</code> <br><code><a href="op_logaddexp.html">op_logaddexp</a>()</code> <br><code><a href="op_logdet.html">op_logdet</a>()</code> <br><code><a href="op_logical_and.html">op_logical_and</a>()</code> <br><code><a href="op_logical_not.html">op_logical_not</a>()</code> <br><code><a href="op_logical_or.html">op_logical_or</a>()</code> <br><code><a href="op_logical_xor.html">op_logical_xor</a>()</code> <br><code><a href="op_logspace.html">op_logspace</a>()</code> <br><code><a href="op_logsumexp.html">op_logsumexp</a>()</code> <br><code><a href="op_lstsq.html">op_lstsq</a>()</code> <br><code><a href="op_lu_factor.html">op_lu_factor</a>()</code> <br><code><a href="op_map.html">op_map</a>()</code> <br><code><a href="op_matmul.html">op_matmul</a>()</code> <br><code><a href="op_max.html">op_max</a>()</code> <br><code><a href="op_max_pool.html">op_max_pool</a>()</code> <br><code><a href="op_maximum.html">op_maximum</a>()</code> <br><code><a href="op_mean.html">op_mean</a>()</code> <br><code><a href="op_median.html">op_median</a>()</code> <br><code><a href="op_meshgrid.html">op_meshgrid</a>()</code> <br><code><a href="op_min.html">op_min</a>()</code> <br><code><a href="op_minimum.html">op_minimum</a>()</code> <br><code><a href="op_mod.html">op_mod</a>()</code> <br><code><a href="op_moments.html">op_moments</a>()</code> <br><code><a href="op_moveaxis.html">op_moveaxis</a>()</code> <br><code><a href="op_multi_hot.html">op_multi_hot</a>()</code> <br><code><a href="op_multiply.html">op_multiply</a>()</code> <br><code><a href="op_nan_to_num.html">op_nan_to_num</a>()</code> <br><code><a href="op_ndim.html">op_ndim</a>()</code> <br><code><a href="op_negative.html">op_negative</a>()</code> <br><code><a href="op_nonzero.html">op_nonzero</a>()</code> <br><code><a href="op_norm.html">op_norm</a>()</code> <br><code><a href="op_normalize.html">op_normalize</a>()</code> <br><code><a href="op_not_equal.html">op_not_equal</a>()</code> <br><code><a href="op_one_hot.html">op_one_hot</a>()</code> <br><code><a href="op_ones.html">op_ones</a>()</code> <br><code><a href="op_ones_like.html">op_ones_like</a>()</code> <br><code><a href="op_outer.html">op_outer</a>()</code> <br><code><a href="op_pad.html">op_pad</a>()</code> <br><code><a href="op_polar.html">op_polar</a>()</code> <br><code><a href="op_power.html">op_power</a>()</code> <br><code><a href="op_prod.html">op_prod</a>()</code> <br><code><a href="op_psnr.html">op_psnr</a>()</code> <br><code><a href="op_qr.html">op_qr</a>()</code> <br><code><a href="op_quantile.html">op_quantile</a>()</code> <br><code><a href="op_ravel.html">op_ravel</a>()</code> <br><code><a href="op_real.html">op_real</a>()</code> <br><code><a href="op_rearrange.html">op_rearrange</a>()</code> <br><code><a href="op_reciprocal.html">op_reciprocal</a>()</code> <br><code><a href="op_relu.html">op_relu</a>()</code> <br><code><a href="op_relu6.html">op_relu6</a>()</code> <br><code><a href="op_repeat.html">op_repeat</a>()</code> <br><code><a href="op_reshape.html">op_reshape</a>()</code> <br><code><a href="op_rfft.html">op_rfft</a>()</code> <br><code><a href="op_right_shift.html">op_right_shift</a>()</code> <br><code><a href="op_rms_normalization.html">op_rms_normalization</a>()</code> <br><code><a href="op_roll.html">op_roll</a>()</code> <br><code><a href="op_rot90.html">op_rot90</a>()</code> <br><code><a href="op_round.html">op_round</a>()</code> <br><code><a href="op_rsqrt.html">op_rsqrt</a>()</code> <br><code><a href="op_saturate_cast.html">op_saturate_cast</a>()</code> <br><code><a href="op_scan.html">op_scan</a>()</code> <br><code><a href="op_scatter.html">op_scatter</a>()</code> <br><code><a href="op_scatter_update.html">op_scatter_update</a>()</code> <br><code><a href="op_searchsorted.html">op_searchsorted</a>()</code> <br><code><a href="op_segment_max.html">op_segment_max</a>()</code> <br><code><a href="op_segment_sum.html">op_segment_sum</a>()</code> <br><code><a href="op_select.html">op_select</a>()</code> <br><code><a href="op_selu.html">op_selu</a>()</code> <br><code><a href="op_separable_conv.html">op_separable_conv</a>()</code> <br><code><a href="op_shape.html">op_shape</a>()</code> <br><code><a href="op_sigmoid.html">op_sigmoid</a>()</code> <br><code><a href="op_sign.html">op_sign</a>()</code> <br><code><a href="op_signbit.html">op_signbit</a>()</code> <br><code><a href="op_silu.html">op_silu</a>()</code> <br><code><a href="op_sin.html">op_sin</a>()</code> <br><code><a href="op_sinh.html">op_sinh</a>()</code> <br><code><a href="op_size.html">op_size</a>()</code> <br><code><a href="op_slice.html">op_slice</a>()</code> <br><code><a href="op_slice_update.html">op_slice_update</a>()</code> <br><code><a href="op_slogdet.html">op_slogdet</a>()</code> <br><code><a href="op_soft_shrink.html">op_soft_shrink</a>()</code> <br><code><a href="op_softmax.html">op_softmax</a>()</code> <br><code><a href="op_softplus.html">op_softplus</a>()</code> <br><code><a href="op_softsign.html">op_softsign</a>()</code> <br><code><a href="op_solve.html">op_solve</a>()</code> <br><code><a href="op_solve_triangular.html">op_solve_triangular</a>()</code> <br><code><a href="op_sort.html">op_sort</a>()</code> <br><code><a href="op_sparse_categorical_crossentropy.html">op_sparse_categorical_crossentropy</a>()</code> <br><code><a href="op_sparse_plus.html">op_sparse_plus</a>()</code> <br><code><a href="op_sparse_sigmoid.html">op_sparse_sigmoid</a>()</code> <br><code><a href="op_sparsemax.html">op_sparsemax</a>()</code> <br><code><a href="op_split.html">op_split</a>()</code> <br><code><a href="op_sqrt.html">op_sqrt</a>()</code> <br><code><a href="op_square.html">op_square</a>()</code> <br><code><a href="op_squareplus.html">op_squareplus</a>()</code> <br><code><a href="op_squeeze.html">op_squeeze</a>()</code> <br><code><a href="op_stack.html">op_stack</a>()</code> <br><code><a href="op_std.html">op_std</a>()</code> <br><code><a href="op_stft.html">op_stft</a>()</code> <br><code><a href="op_stop_gradient.html">op_stop_gradient</a>()</code> <br><code><a href="op_subset.html">op_subset</a>()</code> <br><code><a href="op_subtract.html">op_subtract</a>()</code> <br><code><a href="op_sum.html">op_sum</a>()</code> <br><code><a href="op_svd.html">op_svd</a>()</code> <br><code><a href="op_swapaxes.html">op_swapaxes</a>()</code> <br><code><a href="op_switch.html">op_switch</a>()</code> <br><code><a href="op_take.html">op_take</a>()</code> <br><code><a href="op_take_along_axis.html">op_take_along_axis</a>()</code> <br><code><a href="op_tan.html">op_tan</a>()</code> <br><code><a href="op_tanh.html">op_tanh</a>()</code> <br><code><a href="op_tanh_shrink.html">op_tanh_shrink</a>()</code> <br><code><a href="op_tensordot.html">op_tensordot</a>()</code> <br><code><a href="op_threshold.html">op_threshold</a>()</code> <br><code><a href="op_tile.html">op_tile</a>()</code> <br><code><a href="op_top_k.html">op_top_k</a>()</code> <br><code><a href="op_trace.html">op_trace</a>()</code> <br><code><a href="op_transpose.html">op_transpose</a>()</code> <br><code><a href="op_tri.html">op_tri</a>()</code> <br><code><a href="op_tril.html">op_tril</a>()</code> <br><code><a href="op_triu.html">op_triu</a>()</code> <br><code><a href="op_trunc.html">op_trunc</a>()</code> <br><code><a href="op_unravel_index.html">op_unravel_index</a>()</code> <br><code><a href="op_unstack.html">op_unstack</a>()</code> <br><code><a href="op_var.html">op_var</a>()</code> <br><code><a href="op_vdot.html">op_vdot</a>()</code> <br><code><a href="op_vectorize.html">op_vectorize</a>()</code> <br><code><a href="op_vectorized_map.html">op_vectorized_map</a>()</code> <br><code><a href="op_view_as_complex.html">op_view_as_complex</a>()</code> <br><code><a href="op_view_as_real.html">op_view_as_real</a>()</code> <br><code><a href="op_vstack.html">op_vstack</a>()</code> <br><code><a href="op_where.html">op_where</a>()</code> <br><code><a href="op_while_loop.html">op_while_loop</a>()</code> <br><code><a href="op_zeros.html">op_zeros</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -95,7 +95,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/optimizer_adadelta.html b/docs/dev/reference/optimizer_adadelta.html
index c5242ec48..86d418ef7 100644
--- a/docs/dev/reference/optimizer_adadelta.html
+++ b/docs/dev/reference/optimizer_adadelta.html
@@ -30,7 +30,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -225,7 +225,7 @@ <h2 id="reference">Reference<a class="anchor" aria-label="anchor" href="#referen
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
     <div class="dont-index">
 <ul><li><p><a href="https://keras.io/api/optimizers/adadelta#adadelta-class" class="external-link">https://keras.io/api/optimizers/adadelta#adadelta-class</a></p></li>
-</ul><p>Other optimizers: <br><code><a href="optimizer_adafactor.html">optimizer_adafactor</a>()</code> <br><code><a href="optimizer_adagrad.html">optimizer_adagrad</a>()</code> <br><code><a href="optimizer_adam.html">optimizer_adam</a>()</code> <br><code><a href="optimizer_adam_w.html">optimizer_adam_w</a>()</code> <br><code><a href="optimizer_adamax.html">optimizer_adamax</a>()</code> <br><code><a href="optimizer_ftrl.html">optimizer_ftrl</a>()</code> <br><code><a href="optimizer_lamb.html">optimizer_lamb</a>()</code> <br><code><a href="optimizer_lion.html">optimizer_lion</a>()</code> <br><code><a href="optimizer_loss_scale.html">optimizer_loss_scale</a>()</code> <br><code><a href="optimizer_nadam.html">optimizer_nadam</a>()</code> <br><code><a href="optimizer_rmsprop.html">optimizer_rmsprop</a>()</code> <br><code><a href="optimizer_sgd.html">optimizer_sgd</a>()</code> <br></p></div>
+</ul><p>Other optimizers: <br><code><a href="optimizer_adafactor.html">optimizer_adafactor</a>()</code> <br><code><a href="optimizer_adagrad.html">optimizer_adagrad</a>()</code> <br><code><a href="optimizer_adam.html">optimizer_adam</a>()</code> <br><code><a href="optimizer_adam_w.html">optimizer_adam_w</a>()</code> <br><code><a href="optimizer_adamax.html">optimizer_adamax</a>()</code> <br><code><a href="optimizer_ftrl.html">optimizer_ftrl</a>()</code> <br><code><a href="optimizer_lamb.html">optimizer_lamb</a>()</code> <br><code><a href="optimizer_lion.html">optimizer_lion</a>()</code> <br><code><a href="optimizer_loss_scale.html">optimizer_loss_scale</a>()</code> <br><code><a href="optimizer_muon.html">optimizer_muon</a>()</code> <br><code><a href="optimizer_nadam.html">optimizer_nadam</a>()</code> <br><code><a href="optimizer_rmsprop.html">optimizer_rmsprop</a>()</code> <br><code><a href="optimizer_sgd.html">optimizer_sgd</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -237,7 +237,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/optimizer_adafactor.html b/docs/dev/reference/optimizer_adafactor.html
index 08814a0b9..dd106912c 100644
--- a/docs/dev/reference/optimizer_adafactor.html
+++ b/docs/dev/reference/optimizer_adafactor.html
@@ -18,7 +18,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -228,7 +228,7 @@ <h2 id="reference">Reference<a class="anchor" aria-label="anchor" href="#referen
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
     <div class="dont-index">
 <ul><li><p><a href="https://keras.io/api/optimizers/adafactor#adafactor-class" class="external-link">https://keras.io/api/optimizers/adafactor#adafactor-class</a></p></li>
-</ul><p>Other optimizers: <br><code><a href="optimizer_adadelta.html">optimizer_adadelta</a>()</code> <br><code><a href="optimizer_adagrad.html">optimizer_adagrad</a>()</code> <br><code><a href="optimizer_adam.html">optimizer_adam</a>()</code> <br><code><a href="optimizer_adam_w.html">optimizer_adam_w</a>()</code> <br><code><a href="optimizer_adamax.html">optimizer_adamax</a>()</code> <br><code><a href="optimizer_ftrl.html">optimizer_ftrl</a>()</code> <br><code><a href="optimizer_lamb.html">optimizer_lamb</a>()</code> <br><code><a href="optimizer_lion.html">optimizer_lion</a>()</code> <br><code><a href="optimizer_loss_scale.html">optimizer_loss_scale</a>()</code> <br><code><a href="optimizer_nadam.html">optimizer_nadam</a>()</code> <br><code><a href="optimizer_rmsprop.html">optimizer_rmsprop</a>()</code> <br><code><a href="optimizer_sgd.html">optimizer_sgd</a>()</code> <br></p></div>
+</ul><p>Other optimizers: <br><code><a href="optimizer_adadelta.html">optimizer_adadelta</a>()</code> <br><code><a href="optimizer_adagrad.html">optimizer_adagrad</a>()</code> <br><code><a href="optimizer_adam.html">optimizer_adam</a>()</code> <br><code><a href="optimizer_adam_w.html">optimizer_adam_w</a>()</code> <br><code><a href="optimizer_adamax.html">optimizer_adamax</a>()</code> <br><code><a href="optimizer_ftrl.html">optimizer_ftrl</a>()</code> <br><code><a href="optimizer_lamb.html">optimizer_lamb</a>()</code> <br><code><a href="optimizer_lion.html">optimizer_lion</a>()</code> <br><code><a href="optimizer_loss_scale.html">optimizer_loss_scale</a>()</code> <br><code><a href="optimizer_muon.html">optimizer_muon</a>()</code> <br><code><a href="optimizer_nadam.html">optimizer_nadam</a>()</code> <br><code><a href="optimizer_rmsprop.html">optimizer_rmsprop</a>()</code> <br><code><a href="optimizer_sgd.html">optimizer_sgd</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -240,7 +240,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/optimizer_adagrad.html b/docs/dev/reference/optimizer_adagrad.html
index 7a2f985e5..e3c9174ff 100644
--- a/docs/dev/reference/optimizer_adagrad.html
+++ b/docs/dev/reference/optimizer_adagrad.html
@@ -14,7 +14,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -203,7 +203,7 @@ <h2 id="reference">Reference<a class="anchor" aria-label="anchor" href="#referen
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
     <div class="dont-index">
 <ul><li><p><a href="https://keras.io/api/optimizers/adagrad#adagrad-class" class="external-link">https://keras.io/api/optimizers/adagrad#adagrad-class</a></p></li>
-</ul><p>Other optimizers: <br><code><a href="optimizer_adadelta.html">optimizer_adadelta</a>()</code> <br><code><a href="optimizer_adafactor.html">optimizer_adafactor</a>()</code> <br><code><a href="optimizer_adam.html">optimizer_adam</a>()</code> <br><code><a href="optimizer_adam_w.html">optimizer_adam_w</a>()</code> <br><code><a href="optimizer_adamax.html">optimizer_adamax</a>()</code> <br><code><a href="optimizer_ftrl.html">optimizer_ftrl</a>()</code> <br><code><a href="optimizer_lamb.html">optimizer_lamb</a>()</code> <br><code><a href="optimizer_lion.html">optimizer_lion</a>()</code> <br><code><a href="optimizer_loss_scale.html">optimizer_loss_scale</a>()</code> <br><code><a href="optimizer_nadam.html">optimizer_nadam</a>()</code> <br><code><a href="optimizer_rmsprop.html">optimizer_rmsprop</a>()</code> <br><code><a href="optimizer_sgd.html">optimizer_sgd</a>()</code> <br></p></div>
+</ul><p>Other optimizers: <br><code><a href="optimizer_adadelta.html">optimizer_adadelta</a>()</code> <br><code><a href="optimizer_adafactor.html">optimizer_adafactor</a>()</code> <br><code><a href="optimizer_adam.html">optimizer_adam</a>()</code> <br><code><a href="optimizer_adam_w.html">optimizer_adam_w</a>()</code> <br><code><a href="optimizer_adamax.html">optimizer_adamax</a>()</code> <br><code><a href="optimizer_ftrl.html">optimizer_ftrl</a>()</code> <br><code><a href="optimizer_lamb.html">optimizer_lamb</a>()</code> <br><code><a href="optimizer_lion.html">optimizer_lion</a>()</code> <br><code><a href="optimizer_loss_scale.html">optimizer_loss_scale</a>()</code> <br><code><a href="optimizer_muon.html">optimizer_muon</a>()</code> <br><code><a href="optimizer_nadam.html">optimizer_nadam</a>()</code> <br><code><a href="optimizer_rmsprop.html">optimizer_rmsprop</a>()</code> <br><code><a href="optimizer_sgd.html">optimizer_sgd</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -215,7 +215,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/optimizer_adam.html b/docs/dev/reference/optimizer_adam.html
index ce22bcef3..8f2962732 100644
--- a/docs/dev/reference/optimizer_adam.html
+++ b/docs/dev/reference/optimizer_adam.html
@@ -22,7 +22,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -227,7 +227,7 @@ <h2 id="value">Value<a class="anchor" aria-label="anchor" href="#value"></a></h2
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
     <div class="dont-index">
 <ul><li><p><a href="https://keras.io/api/optimizers/adam#adam-class" class="external-link">https://keras.io/api/optimizers/adam#adam-class</a></p></li>
-</ul><p>Other optimizers: <br><code><a href="optimizer_adadelta.html">optimizer_adadelta</a>()</code> <br><code><a href="optimizer_adafactor.html">optimizer_adafactor</a>()</code> <br><code><a href="optimizer_adagrad.html">optimizer_adagrad</a>()</code> <br><code><a href="optimizer_adam_w.html">optimizer_adam_w</a>()</code> <br><code><a href="optimizer_adamax.html">optimizer_adamax</a>()</code> <br><code><a href="optimizer_ftrl.html">optimizer_ftrl</a>()</code> <br><code><a href="optimizer_lamb.html">optimizer_lamb</a>()</code> <br><code><a href="optimizer_lion.html">optimizer_lion</a>()</code> <br><code><a href="optimizer_loss_scale.html">optimizer_loss_scale</a>()</code> <br><code><a href="optimizer_nadam.html">optimizer_nadam</a>()</code> <br><code><a href="optimizer_rmsprop.html">optimizer_rmsprop</a>()</code> <br><code><a href="optimizer_sgd.html">optimizer_sgd</a>()</code> <br></p></div>
+</ul><p>Other optimizers: <br><code><a href="optimizer_adadelta.html">optimizer_adadelta</a>()</code> <br><code><a href="optimizer_adafactor.html">optimizer_adafactor</a>()</code> <br><code><a href="optimizer_adagrad.html">optimizer_adagrad</a>()</code> <br><code><a href="optimizer_adam_w.html">optimizer_adam_w</a>()</code> <br><code><a href="optimizer_adamax.html">optimizer_adamax</a>()</code> <br><code><a href="optimizer_ftrl.html">optimizer_ftrl</a>()</code> <br><code><a href="optimizer_lamb.html">optimizer_lamb</a>()</code> <br><code><a href="optimizer_lion.html">optimizer_lion</a>()</code> <br><code><a href="optimizer_loss_scale.html">optimizer_loss_scale</a>()</code> <br><code><a href="optimizer_muon.html">optimizer_muon</a>()</code> <br><code><a href="optimizer_nadam.html">optimizer_nadam</a>()</code> <br><code><a href="optimizer_rmsprop.html">optimizer_rmsprop</a>()</code> <br><code><a href="optimizer_sgd.html">optimizer_sgd</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -239,7 +239,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/optimizer_adam_w.html b/docs/dev/reference/optimizer_adam_w.html
index 305dd568d..0fd22bed3 100644
--- a/docs/dev/reference/optimizer_adam_w.html
+++ b/docs/dev/reference/optimizer_adam_w.html
@@ -28,7 +28,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -243,7 +243,7 @@ <h2 id="references">References<a class="anchor" aria-label="anchor" href="#refer
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
     <div class="dont-index">
 <ul><li><p><a href="https://keras.io/api/optimizers/adamw#adamw-class" class="external-link">https://keras.io/api/optimizers/adamw#adamw-class</a></p></li>
-</ul><p>Other optimizers: <br><code><a href="optimizer_adadelta.html">optimizer_adadelta</a>()</code> <br><code><a href="optimizer_adafactor.html">optimizer_adafactor</a>()</code> <br><code><a href="optimizer_adagrad.html">optimizer_adagrad</a>()</code> <br><code><a href="optimizer_adam.html">optimizer_adam</a>()</code> <br><code><a href="optimizer_adamax.html">optimizer_adamax</a>()</code> <br><code><a href="optimizer_ftrl.html">optimizer_ftrl</a>()</code> <br><code><a href="optimizer_lamb.html">optimizer_lamb</a>()</code> <br><code><a href="optimizer_lion.html">optimizer_lion</a>()</code> <br><code><a href="optimizer_loss_scale.html">optimizer_loss_scale</a>()</code> <br><code><a href="optimizer_nadam.html">optimizer_nadam</a>()</code> <br><code><a href="optimizer_rmsprop.html">optimizer_rmsprop</a>()</code> <br><code><a href="optimizer_sgd.html">optimizer_sgd</a>()</code> <br></p></div>
+</ul><p>Other optimizers: <br><code><a href="optimizer_adadelta.html">optimizer_adadelta</a>()</code> <br><code><a href="optimizer_adafactor.html">optimizer_adafactor</a>()</code> <br><code><a href="optimizer_adagrad.html">optimizer_adagrad</a>()</code> <br><code><a href="optimizer_adam.html">optimizer_adam</a>()</code> <br><code><a href="optimizer_adamax.html">optimizer_adamax</a>()</code> <br><code><a href="optimizer_ftrl.html">optimizer_ftrl</a>()</code> <br><code><a href="optimizer_lamb.html">optimizer_lamb</a>()</code> <br><code><a href="optimizer_lion.html">optimizer_lion</a>()</code> <br><code><a href="optimizer_loss_scale.html">optimizer_loss_scale</a>()</code> <br><code><a href="optimizer_muon.html">optimizer_muon</a>()</code> <br><code><a href="optimizer_nadam.html">optimizer_nadam</a>()</code> <br><code><a href="optimizer_rmsprop.html">optimizer_rmsprop</a>()</code> <br><code><a href="optimizer_sgd.html">optimizer_sgd</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -255,7 +255,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/optimizer_adamax.html b/docs/dev/reference/optimizer_adamax.html
index c7895a4b2..26fc68336 100644
--- a/docs/dev/reference/optimizer_adamax.html
+++ b/docs/dev/reference/optimizer_adamax.html
@@ -44,7 +44,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -250,7 +250,7 @@ <h2 id="reference">Reference<a class="anchor" aria-label="anchor" href="#referen
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
     <div class="dont-index">
 <ul><li><p><a href="https://keras.io/api/optimizers/adamax#adamax-class" class="external-link">https://keras.io/api/optimizers/adamax#adamax-class</a></p></li>
-</ul><p>Other optimizers: <br><code><a href="optimizer_adadelta.html">optimizer_adadelta</a>()</code> <br><code><a href="optimizer_adafactor.html">optimizer_adafactor</a>()</code> <br><code><a href="optimizer_adagrad.html">optimizer_adagrad</a>()</code> <br><code><a href="optimizer_adam.html">optimizer_adam</a>()</code> <br><code><a href="optimizer_adam_w.html">optimizer_adam_w</a>()</code> <br><code><a href="optimizer_ftrl.html">optimizer_ftrl</a>()</code> <br><code><a href="optimizer_lamb.html">optimizer_lamb</a>()</code> <br><code><a href="optimizer_lion.html">optimizer_lion</a>()</code> <br><code><a href="optimizer_loss_scale.html">optimizer_loss_scale</a>()</code> <br><code><a href="optimizer_nadam.html">optimizer_nadam</a>()</code> <br><code><a href="optimizer_rmsprop.html">optimizer_rmsprop</a>()</code> <br><code><a href="optimizer_sgd.html">optimizer_sgd</a>()</code> <br></p></div>
+</ul><p>Other optimizers: <br><code><a href="optimizer_adadelta.html">optimizer_adadelta</a>()</code> <br><code><a href="optimizer_adafactor.html">optimizer_adafactor</a>()</code> <br><code><a href="optimizer_adagrad.html">optimizer_adagrad</a>()</code> <br><code><a href="optimizer_adam.html">optimizer_adam</a>()</code> <br><code><a href="optimizer_adam_w.html">optimizer_adam_w</a>()</code> <br><code><a href="optimizer_ftrl.html">optimizer_ftrl</a>()</code> <br><code><a href="optimizer_lamb.html">optimizer_lamb</a>()</code> <br><code><a href="optimizer_lion.html">optimizer_lion</a>()</code> <br><code><a href="optimizer_loss_scale.html">optimizer_loss_scale</a>()</code> <br><code><a href="optimizer_muon.html">optimizer_muon</a>()</code> <br><code><a href="optimizer_nadam.html">optimizer_nadam</a>()</code> <br><code><a href="optimizer_rmsprop.html">optimizer_rmsprop</a>()</code> <br><code><a href="optimizer_sgd.html">optimizer_sgd</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -262,7 +262,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/optimizer_ftrl.html b/docs/dev/reference/optimizer_ftrl.html
index 618b25bda..482f18ace 100644
--- a/docs/dev/reference/optimizer_ftrl.html
+++ b/docs/dev/reference/optimizer_ftrl.html
@@ -78,7 +78,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -315,7 +315,7 @@ <h2 id="value">Value<a class="anchor" aria-label="anchor" href="#value"></a></h2
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
     <div class="dont-index">
 <ul><li><p><a href="https://keras.io/api/optimizers/ftrl#ftrl-class" class="external-link">https://keras.io/api/optimizers/ftrl#ftrl-class</a></p></li>
-</ul><p>Other optimizers: <br><code><a href="optimizer_adadelta.html">optimizer_adadelta</a>()</code> <br><code><a href="optimizer_adafactor.html">optimizer_adafactor</a>()</code> <br><code><a href="optimizer_adagrad.html">optimizer_adagrad</a>()</code> <br><code><a href="optimizer_adam.html">optimizer_adam</a>()</code> <br><code><a href="optimizer_adam_w.html">optimizer_adam_w</a>()</code> <br><code><a href="optimizer_adamax.html">optimizer_adamax</a>()</code> <br><code><a href="optimizer_lamb.html">optimizer_lamb</a>()</code> <br><code><a href="optimizer_lion.html">optimizer_lion</a>()</code> <br><code><a href="optimizer_loss_scale.html">optimizer_loss_scale</a>()</code> <br><code><a href="optimizer_nadam.html">optimizer_nadam</a>()</code> <br><code><a href="optimizer_rmsprop.html">optimizer_rmsprop</a>()</code> <br><code><a href="optimizer_sgd.html">optimizer_sgd</a>()</code> <br></p></div>
+</ul><p>Other optimizers: <br><code><a href="optimizer_adadelta.html">optimizer_adadelta</a>()</code> <br><code><a href="optimizer_adafactor.html">optimizer_adafactor</a>()</code> <br><code><a href="optimizer_adagrad.html">optimizer_adagrad</a>()</code> <br><code><a href="optimizer_adam.html">optimizer_adam</a>()</code> <br><code><a href="optimizer_adam_w.html">optimizer_adam_w</a>()</code> <br><code><a href="optimizer_adamax.html">optimizer_adamax</a>()</code> <br><code><a href="optimizer_lamb.html">optimizer_lamb</a>()</code> <br><code><a href="optimizer_lion.html">optimizer_lion</a>()</code> <br><code><a href="optimizer_loss_scale.html">optimizer_loss_scale</a>()</code> <br><code><a href="optimizer_muon.html">optimizer_muon</a>()</code> <br><code><a href="optimizer_nadam.html">optimizer_nadam</a>()</code> <br><code><a href="optimizer_rmsprop.html">optimizer_rmsprop</a>()</code> <br><code><a href="optimizer_sgd.html">optimizer_sgd</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -327,7 +327,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/optimizer_lamb.html b/docs/dev/reference/optimizer_lamb.html
index d01e445b8..f5985d018 100644
--- a/docs/dev/reference/optimizer_lamb.html
+++ b/docs/dev/reference/optimizer_lamb.html
@@ -18,7 +18,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -216,7 +216,7 @@ <h2 id="references">References<a class="anchor" aria-label="anchor" href="#refer
 </ul></div>
     <div class="section level2">
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
-    <div class="dont-index"><p>Other optimizers: <br><code><a href="optimizer_adadelta.html">optimizer_adadelta</a>()</code> <br><code><a href="optimizer_adafactor.html">optimizer_adafactor</a>()</code> <br><code><a href="optimizer_adagrad.html">optimizer_adagrad</a>()</code> <br><code><a href="optimizer_adam.html">optimizer_adam</a>()</code> <br><code><a href="optimizer_adam_w.html">optimizer_adam_w</a>()</code> <br><code><a href="optimizer_adamax.html">optimizer_adamax</a>()</code> <br><code><a href="optimizer_ftrl.html">optimizer_ftrl</a>()</code> <br><code><a href="optimizer_lion.html">optimizer_lion</a>()</code> <br><code><a href="optimizer_loss_scale.html">optimizer_loss_scale</a>()</code> <br><code><a href="optimizer_nadam.html">optimizer_nadam</a>()</code> <br><code><a href="optimizer_rmsprop.html">optimizer_rmsprop</a>()</code> <br><code><a href="optimizer_sgd.html">optimizer_sgd</a>()</code> <br></p></div>
+    <div class="dont-index"><p>Other optimizers: <br><code><a href="optimizer_adadelta.html">optimizer_adadelta</a>()</code> <br><code><a href="optimizer_adafactor.html">optimizer_adafactor</a>()</code> <br><code><a href="optimizer_adagrad.html">optimizer_adagrad</a>()</code> <br><code><a href="optimizer_adam.html">optimizer_adam</a>()</code> <br><code><a href="optimizer_adam_w.html">optimizer_adam_w</a>()</code> <br><code><a href="optimizer_adamax.html">optimizer_adamax</a>()</code> <br><code><a href="optimizer_ftrl.html">optimizer_ftrl</a>()</code> <br><code><a href="optimizer_lion.html">optimizer_lion</a>()</code> <br><code><a href="optimizer_loss_scale.html">optimizer_loss_scale</a>()</code> <br><code><a href="optimizer_muon.html">optimizer_muon</a>()</code> <br><code><a href="optimizer_nadam.html">optimizer_nadam</a>()</code> <br><code><a href="optimizer_rmsprop.html">optimizer_rmsprop</a>()</code> <br><code><a href="optimizer_sgd.html">optimizer_sgd</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -228,7 +228,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/optimizer_lion.html b/docs/dev/reference/optimizer_lion.html
index 497fc205d..549b03893 100644
--- a/docs/dev/reference/optimizer_lion.html
+++ b/docs/dev/reference/optimizer_lion.html
@@ -1,22 +1,22 @@
 <!DOCTYPE html>
 <!-- Generated by pkgdown: do not edit by hand --><html lang="en"><head><meta http-equiv="Content-Type" content="text/html; charset=UTF-8"><meta charset="utf-8"><meta http-equiv="X-UA-Compatible" content="IE=edge"><meta name="viewport" content="width=device-width, initial-scale=1, shrink-to-fit=no"><title>Optimizer that implements the Lion algorithm. — optimizer_lion • keras3</title><!-- favicons --><link rel="icon" type="image/png" sizes="96x96" href="../favicon-96x96.png"><link rel="icon" type="”image/svg+xml”" href="../favicon.svg"><link rel="apple-touch-icon" sizes="180x180" href="../apple-touch-icon.png"><link rel="icon" sizes="any" href="../favicon.ico"><link rel="manifest" href="../site.webmanifest"><script src="../deps/jquery-3.6.0/jquery-3.6.0.min.js"></script><meta name="viewport" content="width=device-width, initial-scale=1, shrink-to-fit=no"><link href="../deps/bootstrap-5.3.1/bootstrap.min.css" rel="stylesheet"><script src="../deps/bootstrap-5.3.1/bootstrap.bundle.min.js"></script><link href="../deps/font-awesome-6.5.2/css/all.min.css" rel="stylesheet"><link href="../deps/font-awesome-6.5.2/css/v4-shims.min.css" rel="stylesheet"><script src="../deps/headroom-0.11.0/headroom.min.js"></script><script src="../deps/headroom-0.11.0/jQuery.headroom.min.js"></script><script src="../deps/bootstrap-toc-1.0.1/bootstrap-toc.min.js"></script><script src="../deps/clipboard.js-2.0.11/clipboard.min.js"></script><script src="../deps/search-1.0.0/autocomplete.jquery.min.js"></script><script src="../deps/search-1.0.0/fuse.min.js"></script><script src="../deps/search-1.0.0/mark.min.js"></script><!-- pkgdown --><script src="../pkgdown.js"></script><link href="../extra.css" rel="stylesheet"><meta property="og:title" content="Optimizer that implements the Lion algorithm. — optimizer_lion"><meta name="description" content="The Lion optimizer is a stochastic-gradient-descent method that uses the
 sign operator to control the magnitude of the update, unlike other adaptive
-optimizers such as Adam that rely on second-order moments. This make
+optimizers such as Adam that rely on second-order moments. This makes
 Lion more memory-efficient as it only keeps track of the momentum. According
 to the authors (see reference), its performance gain over Adam grows with
 the batch size. Because the update of Lion is produced through the sign
 operation, resulting in a larger norm, a suitable learning rate for Lion is
 typically 3-10x smaller than that for AdamW. The weight decay for Lion
-should be in turn 3-10x larger than that for AdamW to maintain a
+should in turn be 3-10x larger than that for AdamW to maintain a
 similar strength (lr * wd)."><meta property="og:description" content="The Lion optimizer is a stochastic-gradient-descent method that uses the
 sign operator to control the magnitude of the update, unlike other adaptive
-optimizers such as Adam that rely on second-order moments. This make
+optimizers such as Adam that rely on second-order moments. This makes
 Lion more memory-efficient as it only keeps track of the momentum. According
 to the authors (see reference), its performance gain over Adam grows with
 the batch size. Because the update of Lion is produced through the sign
 operation, resulting in a larger norm, a suitable learning rate for Lion is
 typically 3-10x smaller than that for AdamW. The weight decay for Lion
-should be in turn 3-10x larger than that for AdamW to maintain a
+should in turn be 3-10x larger than that for AdamW to maintain a
 similar strength (lr * wd)."><meta name="robots" content="noindex"><!-- Google Tag Manager --><script>(function(w,d,s,l,i){w[l]=w[l]||[];w[l].push({'gtm.start': new Date().getTime(),event:'gtm.js'});var f=d.getElementsByTagName(s)[0],j=d.createElement(s),dl=l!='dataLayer'?'&l='+l:'';j.async=true;j.src='https://www.googletagmanager.com/gtm.js?id='+i+dl;f.parentNode.insertBefore(j,f);})(window,document,'script','dataLayer','GTM-KHBDBW7');</script><!-- End Google Tag Manager --></head><body>
     <a href="#main" class="visually-hidden-focusable">Skip to contents</a>
     <!-- Google Tag Manager (noscript) --> <noscript><iframe src="https://www.googletagmanager.com/ns.html?id=GTM-KHBDBW7" height="0" width="0" style="display:none;visibility:hidden"></iframe></noscript> <!-- End Google Tag Manager (noscript) -->
@@ -26,7 +26,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -74,13 +74,13 @@ <h1>Optimizer that implements the Lion algorithm.</h1>
     <div class="ref-description section level2">
     <p>The Lion optimizer is a stochastic-gradient-descent method that uses the
 sign operator to control the magnitude of the update, unlike other adaptive
-optimizers such as Adam that rely on second-order moments. This make
+optimizers such as Adam that rely on second-order moments. This makes
 Lion more memory-efficient as it only keeps track of the momentum. According
 to the authors (see reference), its performance gain over Adam grows with
 the batch size. Because the update of Lion is produced through the sign
 operation, resulting in a larger norm, a suitable learning rate for Lion is
 typically 3-10x smaller than that for AdamW. The weight decay for Lion
-should be in turn 3-10x larger than that for AdamW to maintain a
+should in turn be 3-10x larger than that for AdamW to maintain a
 similar strength (lr * wd).</p>
     </div>
 
@@ -222,7 +222,7 @@ <h2 id="references">References<a class="anchor" aria-label="anchor" href="#refer
 </ul></div>
     <div class="section level2">
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
-    <div class="dont-index"><p>Other optimizers: <br><code><a href="optimizer_adadelta.html">optimizer_adadelta</a>()</code> <br><code><a href="optimizer_adafactor.html">optimizer_adafactor</a>()</code> <br><code><a href="optimizer_adagrad.html">optimizer_adagrad</a>()</code> <br><code><a href="optimizer_adam.html">optimizer_adam</a>()</code> <br><code><a href="optimizer_adam_w.html">optimizer_adam_w</a>()</code> <br><code><a href="optimizer_adamax.html">optimizer_adamax</a>()</code> <br><code><a href="optimizer_ftrl.html">optimizer_ftrl</a>()</code> <br><code><a href="optimizer_lamb.html">optimizer_lamb</a>()</code> <br><code><a href="optimizer_loss_scale.html">optimizer_loss_scale</a>()</code> <br><code><a href="optimizer_nadam.html">optimizer_nadam</a>()</code> <br><code><a href="optimizer_rmsprop.html">optimizer_rmsprop</a>()</code> <br><code><a href="optimizer_sgd.html">optimizer_sgd</a>()</code> <br></p></div>
+    <div class="dont-index"><p>Other optimizers: <br><code><a href="optimizer_adadelta.html">optimizer_adadelta</a>()</code> <br><code><a href="optimizer_adafactor.html">optimizer_adafactor</a>()</code> <br><code><a href="optimizer_adagrad.html">optimizer_adagrad</a>()</code> <br><code><a href="optimizer_adam.html">optimizer_adam</a>()</code> <br><code><a href="optimizer_adam_w.html">optimizer_adam_w</a>()</code> <br><code><a href="optimizer_adamax.html">optimizer_adamax</a>()</code> <br><code><a href="optimizer_ftrl.html">optimizer_ftrl</a>()</code> <br><code><a href="optimizer_lamb.html">optimizer_lamb</a>()</code> <br><code><a href="optimizer_loss_scale.html">optimizer_loss_scale</a>()</code> <br><code><a href="optimizer_muon.html">optimizer_muon</a>()</code> <br><code><a href="optimizer_nadam.html">optimizer_nadam</a>()</code> <br><code><a href="optimizer_rmsprop.html">optimizer_rmsprop</a>()</code> <br><code><a href="optimizer_sgd.html">optimizer_sgd</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -234,7 +234,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/optimizer_loss_scale.html b/docs/dev/reference/optimizer_loss_scale.html
index 7981debc1..9db5a44a5 100644
--- a/docs/dev/reference/optimizer_loss_scale.html
+++ b/docs/dev/reference/optimizer_loss_scale.html
@@ -36,7 +36,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -224,7 +224,7 @@ <h2 id="value">Value<a class="anchor" aria-label="anchor" href="#value"></a></h2
     </div>
     <div class="section level2">
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
-    <div class="dont-index"><p>Other optimizers: <br><code><a href="optimizer_adadelta.html">optimizer_adadelta</a>()</code> <br><code><a href="optimizer_adafactor.html">optimizer_adafactor</a>()</code> <br><code><a href="optimizer_adagrad.html">optimizer_adagrad</a>()</code> <br><code><a href="optimizer_adam.html">optimizer_adam</a>()</code> <br><code><a href="optimizer_adam_w.html">optimizer_adam_w</a>()</code> <br><code><a href="optimizer_adamax.html">optimizer_adamax</a>()</code> <br><code><a href="optimizer_ftrl.html">optimizer_ftrl</a>()</code> <br><code><a href="optimizer_lamb.html">optimizer_lamb</a>()</code> <br><code><a href="optimizer_lion.html">optimizer_lion</a>()</code> <br><code><a href="optimizer_nadam.html">optimizer_nadam</a>()</code> <br><code><a href="optimizer_rmsprop.html">optimizer_rmsprop</a>()</code> <br><code><a href="optimizer_sgd.html">optimizer_sgd</a>()</code> <br></p></div>
+    <div class="dont-index"><p>Other optimizers: <br><code><a href="optimizer_adadelta.html">optimizer_adadelta</a>()</code> <br><code><a href="optimizer_adafactor.html">optimizer_adafactor</a>()</code> <br><code><a href="optimizer_adagrad.html">optimizer_adagrad</a>()</code> <br><code><a href="optimizer_adam.html">optimizer_adam</a>()</code> <br><code><a href="optimizer_adam_w.html">optimizer_adam_w</a>()</code> <br><code><a href="optimizer_adamax.html">optimizer_adamax</a>()</code> <br><code><a href="optimizer_ftrl.html">optimizer_ftrl</a>()</code> <br><code><a href="optimizer_lamb.html">optimizer_lamb</a>()</code> <br><code><a href="optimizer_lion.html">optimizer_lion</a>()</code> <br><code><a href="optimizer_muon.html">optimizer_muon</a>()</code> <br><code><a href="optimizer_nadam.html">optimizer_nadam</a>()</code> <br><code><a href="optimizer_rmsprop.html">optimizer_rmsprop</a>()</code> <br><code><a href="optimizer_sgd.html">optimizer_sgd</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -236,7 +236,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/optimizer_muon.html b/docs/dev/reference/optimizer_muon.html
new file mode 100644
index 000000000..4cb945b28
--- /dev/null
+++ b/docs/dev/reference/optimizer_muon.html
@@ -0,0 +1,265 @@
+<!DOCTYPE html>
+<!-- Generated by pkgdown: do not edit by hand --><html lang="en"><head><meta http-equiv="Content-Type" content="text/html; charset=UTF-8"><meta charset="utf-8"><meta http-equiv="X-UA-Compatible" content="IE=edge"><meta name="viewport" content="width=device-width, initial-scale=1, shrink-to-fit=no"><title>Optimizer that implements the Muon algorithm. — optimizer_muon • keras3</title><!-- favicons --><link rel="icon" type="image/png" sizes="96x96" href="../favicon-96x96.png"><link rel="icon" type="”image/svg+xml”" href="../favicon.svg"><link rel="apple-touch-icon" sizes="180x180" href="../apple-touch-icon.png"><link rel="icon" sizes="any" href="../favicon.ico"><link rel="manifest" href="../site.webmanifest"><script src="../deps/jquery-3.6.0/jquery-3.6.0.min.js"></script><meta name="viewport" content="width=device-width, initial-scale=1, shrink-to-fit=no"><link href="../deps/bootstrap-5.3.1/bootstrap.min.css" rel="stylesheet"><script src="../deps/bootstrap-5.3.1/bootstrap.bundle.min.js"></script><link href="../deps/font-awesome-6.5.2/css/all.min.css" rel="stylesheet"><link href="../deps/font-awesome-6.5.2/css/v4-shims.min.css" rel="stylesheet"><script src="../deps/headroom-0.11.0/headroom.min.js"></script><script src="../deps/headroom-0.11.0/jQuery.headroom.min.js"></script><script src="../deps/bootstrap-toc-1.0.1/bootstrap-toc.min.js"></script><script src="../deps/clipboard.js-2.0.11/clipboard.min.js"></script><script src="../deps/search-1.0.0/autocomplete.jquery.min.js"></script><script src="../deps/search-1.0.0/fuse.min.js"></script><script src="../deps/search-1.0.0/mark.min.js"></script><!-- pkgdown --><script src="../pkgdown.js"></script><link href="../extra.css" rel="stylesheet"><meta property="og:title" content="Optimizer that implements the Muon algorithm. — optimizer_muon"><meta name="description" content="Note that this optimizer should not be used in the following layers:
+Embedding layer
+Final output fully connected layer
+Any 0,1-D variables
+
+
+These should all be optimized using AdamW.
+The Muon optimizer can use both the Muon update step or the
+AdamW update step based on the following:
+For any variable that isn't 2D, 3D or 4D, the AdamW step
+will be used. This is not configurable.
+If the argument exclude_embeddings (defaults to TRUE) is set
+to TRUE, the AdamW step will be used.
+For any variable with a name that matches an expression
+listed in the argument exclude_layers (a list), the
+AdamW step will be used.
+Any other variable uses the Muon step.
+
+
+Typically, you only need to pass the name of your densely-connected
+output layer to exclude_layers, e.g.
+exclude_layers = &quot;output_dense&quot;."><meta property="og:description" content="Note that this optimizer should not be used in the following layers:
+Embedding layer
+Final output fully connected layer
+Any 0,1-D variables
+
+
+These should all be optimized using AdamW.
+The Muon optimizer can use both the Muon update step or the
+AdamW update step based on the following:
+For any variable that isn't 2D, 3D or 4D, the AdamW step
+will be used. This is not configurable.
+If the argument exclude_embeddings (defaults to TRUE) is set
+to TRUE, the AdamW step will be used.
+For any variable with a name that matches an expression
+listed in the argument exclude_layers (a list), the
+AdamW step will be used.
+Any other variable uses the Muon step.
+
+
+Typically, you only need to pass the name of your densely-connected
+output layer to exclude_layers, e.g.
+exclude_layers = &quot;output_dense&quot;."><meta name="robots" content="noindex"><!-- Google Tag Manager --><script>(function(w,d,s,l,i){w[l]=w[l]||[];w[l].push({'gtm.start': new Date().getTime(),event:'gtm.js'});var f=d.getElementsByTagName(s)[0],j=d.createElement(s),dl=l!='dataLayer'?'&l='+l:'';j.async=true;j.src='https://www.googletagmanager.com/gtm.js?id='+i+dl;f.parentNode.insertBefore(j,f);})(window,document,'script','dataLayer','GTM-KHBDBW7');</script><!-- End Google Tag Manager --></head><body>
+    <a href="#main" class="visually-hidden-focusable">Skip to contents</a>
+    <!-- Google Tag Manager (noscript) --> <noscript><iframe src="https://www.googletagmanager.com/ns.html?id=GTM-KHBDBW7" height="0" width="0" style="display:none;visibility:hidden"></iframe></noscript> <!-- End Google Tag Manager (noscript) -->
+
+
+    <nav class="navbar navbar-expand-lg fixed-top bg-primary" data-bs-theme="inverse" aria-label="Site navigation"><div class="container">
+
+    <a class="navbar-brand me-2" href="../index.html">keras3</a>
+
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
+
+
+    <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
+      <span class="navbar-toggler-icon"></span>
+    </button>
+
+    <div id="navbar" class="collapse navbar-collapse ms-3">
+      <ul class="navbar-nav me-auto"><li class="nav-item"><a class="nav-link" href="../articles/getting_started.html">Getting Started</a></li>
+<li class="nav-item dropdown">
+  <button class="nav-link dropdown-toggle" type="button" id="dropdown-guides" data-bs-toggle="dropdown" aria-expanded="false" aria-haspopup="true">Guides</button>
+  <ul class="dropdown-menu" aria-labelledby="dropdown-guides"><li><h6 class="dropdown-header" data-toc-skip>Model definition</h6></li>
+    <li><a class="dropdown-item" href="../articles/sequential_model.html">Sequential Model</a></li>
+    <li><a class="dropdown-item" href="../articles/functional_api.html">Functional API</a></li>
+    <li><h6 class="dropdown-header" data-toc-skip>Extending and customizing</h6></li>
+    <li><a class="dropdown-item" href="../articles/training_with_built_in_methods.html">Training &amp; evaluation with the built-in methods</a></li>
+    <li><a class="dropdown-item" href="../articles/custom_train_step_in_tensorflow.html">Customizing `fit()` with Tensorflow</a></li>
+    <li><a class="dropdown-item" href="../articles/writing_your_own_callbacks.html">Writing your own callbacks</a></li>
+    <li><a class="dropdown-item" href="../articles/making_new_layers_and_models_via_subclassing.html">Making new layers and models via subclassing</a></li>
+    <li><a class="dropdown-item" href="../articles/writing_a_custom_training_loop_in_tensorflow.html">Writing a training loop from scratch in TensorFlow</a></li>
+    <li><a class="dropdown-item" href="../articles/serialization_and_saving.html">Serialization and Saving</a></li>
+    <li><h6 class="dropdown-header" data-toc-skip>Other topics</h6></li>
+    <li><a class="dropdown-item" href="../articles/transfer_learning.html">Transfer learning and fine tuning</a></li>
+    <li><a class="dropdown-item" href="../articles/distributed_training_with_tensorflow.html">Distributed training with TensorFlow</a></li>
+    <li><a class="dropdown-item" href="../articles/distribution.html">Distributed training with Jax</a></li>
+  </ul></li>
+<li class="nav-item"><a class="nav-link" href="../articles/examples/index.html">Examples</a></li>
+<li class="active nav-item"><a class="nav-link" href="../reference/index.html">Reference</a></li>
+<li class="nav-item"><a class="nav-link" href="../news/index.html">News</a></li>
+      </ul><ul class="navbar-nav"><li class="nav-item"><form class="form-inline" role="search">
+ <input class="form-control" type="search" name="search-input" id="search-input" autocomplete="off" aria-label="Search site" placeholder="Search for" data-search-index="../search.json"></form></li>
+<li class="nav-item"><a class="external-link nav-link" href="https://github.com/rstudio/keras3/" aria-label="GitHub"><span class="fa fab fa-github fa-lg"></span></a></li>
+      </ul></div>
+
+
+  </div>
+</nav><div class="container template-reference-topic">
+<div class="row">
+  <main id="main" class="col-md-9"><div class="page-header">
+
+      <h1>Optimizer that implements the Muon algorithm.</h1>
+      <small class="dont-index">Source: <a href="https://github.com/rstudio/keras3/blob/HEAD/R/optimizers.R" class="external-link"><code>R/optimizers.R</code></a></small>
+      <div class="d-none name"><code>optimizer_muon.Rd</code></div>
+    </div>
+
+    <div class="ref-description section level2">
+    <p>Note that this optimizer should not be used in the following layers:</p><ol><li><p>Embedding layer</p></li>
+<li><p>Final output fully connected layer</p></li>
+<li><p>Any 0,1-D variables</p></li>
+</ol><p>These should all be optimized using AdamW.</p>
+<p>The Muon optimizer can use both the Muon update step or the
+AdamW update step based on the following:</p><ul><li><p>For any variable that isn't 2D, 3D or 4D, the AdamW step
+will be used. This is not configurable.</p></li>
+<li><p>If the argument <code>exclude_embeddings</code> (defaults to <code>TRUE</code>) is set
+to <code>TRUE</code>, the AdamW step will be used.</p></li>
+<li><p>For any variable with a name that matches an expression
+listed in the argument <code>exclude_layers</code> (a list), the
+AdamW step will be used.</p></li>
+<li><p>Any other variable uses the Muon step.</p></li>
+</ul><p>Typically, you only need to pass the name of your densely-connected
+output layer to <code>exclude_layers</code>, e.g.
+<code>exclude_layers = "output_dense"</code>.</p>
+    </div>
+
+    <div class="section level2">
+    <h2 id="ref-usage">Usage<a class="anchor" aria-label="anchor" href="#ref-usage"></a></h2>
+    <div class="sourceCode"><pre class="sourceCode r"><code><span><span class="fu">optimizer_muon</span><span class="op">(</span></span>
+<span>  learning_rate <span class="op">=</span> <span class="fl">0.001</span>,</span>
+<span>  adam_beta_1 <span class="op">=</span> <span class="fl">0.9</span>,</span>
+<span>  adam_beta_2 <span class="op">=</span> <span class="fl">0.999</span>,</span>
+<span>  epsilon <span class="op">=</span> <span class="fl">1e-07</span>,</span>
+<span>  weight_decay <span class="op">=</span> <span class="fl">0.1</span>,</span>
+<span>  clipnorm <span class="op">=</span> <span class="cn">NULL</span>,</span>
+<span>  clipvalue <span class="op">=</span> <span class="cn">NULL</span>,</span>
+<span>  global_clipnorm <span class="op">=</span> <span class="cn">NULL</span>,</span>
+<span>  use_ema <span class="op">=</span> <span class="cn">FALSE</span>,</span>
+<span>  ema_momentum <span class="op">=</span> <span class="fl">0.99</span>,</span>
+<span>  ema_overwrite_frequency <span class="op">=</span> <span class="cn">NULL</span>,</span>
+<span>  loss_scale_factor <span class="op">=</span> <span class="cn">NULL</span>,</span>
+<span>  gradient_accumulation_steps <span class="op">=</span> <span class="cn">NULL</span>,</span>
+<span>  name <span class="op">=</span> <span class="st">"muon"</span>,</span>
+<span>  exclude_layers <span class="op">=</span> <span class="cn">NULL</span>,</span>
+<span>  exclude_embeddings <span class="op">=</span> <span class="cn">TRUE</span>,</span>
+<span>  muon_a <span class="op">=</span> <span class="fl">3.4445</span>,</span>
+<span>  muon_b <span class="op">=</span> <span class="op">-</span><span class="fl">4.775</span>,</span>
+<span>  muon_c <span class="op">=</span> <span class="fl">2.0315</span>,</span>
+<span>  adam_lr_ratio <span class="op">=</span> <span class="fl">0.1</span>,</span>
+<span>  momentum <span class="op">=</span> <span class="fl">0.95</span>,</span>
+<span>  ns_steps <span class="op">=</span> <span class="fl">6L</span>,</span>
+<span>  nesterov <span class="op">=</span> <span class="cn">TRUE</span>,</span>
+<span>  <span class="va">...</span></span>
+<span><span class="op">)</span></span></code></pre></div>
+    </div>
+
+    <div class="section level2">
+    <h2 id="arguments">Arguments<a class="anchor" aria-label="anchor" href="#arguments"></a></h2>
+
+
+<dl><dt id="arg-learning-rate">learning_rate<a class="anchor" aria-label="anchor" href="#arg-learning-rate"></a></dt>
+<dd><p>A float,
+<code><a href="LearningRateSchedule.html">LearningRateSchedule()</a></code> instance, or
+a callable that takes no arguments and returns the actual value to
+use. The learning rate. Defaults to <code>0.001</code>.</p></dd>
+
+
+<dt id="arg-adam-beta-">adam_beta_1<a class="anchor" aria-label="anchor" href="#arg-adam-beta-"></a></dt>
+<dd><p>A float value or a constant float tensor, or a callable
+that takes no arguments and returns the actual value to use.
+The exponential decay rate for the 1st moment estimates. Defaults to
+<code>0.9</code>.</p></dd>
+
+
+<dt id="arg-adam-beta-">adam_beta_2<a class="anchor" aria-label="anchor" href="#arg-adam-beta-"></a></dt>
+<dd><p>A float value or a constant float tensor, ora callable
+that takes no arguments and returns the actual value to use.
+The exponential decay rate for the 2nd moment estimates. Defaults to
+<code>0.999</code>.</p></dd>
+
+
+<dt id="arg-epsilon">epsilon<a class="anchor" aria-label="anchor" href="#arg-epsilon"></a></dt>
+<dd><p>A small constant for numerical stability. This is
+"epsilon hat" in the Kingma and Ba paper
+(in the formula just before Section 2.1),
+not the epsilon in Algorithm 1 of the paper.
+It is used as in AdamW. Defaults to <code>1e-7</code>.</p></dd>
+
+
+<dt id="arg-name">name<a class="anchor" aria-label="anchor" href="#arg-name"></a></dt>
+<dd><p>String, name for the object</p></dd>
+
+
+<dt id="arg-exclude-layers">exclude_layers<a class="anchor" aria-label="anchor" href="#arg-exclude-layers"></a></dt>
+<dd><p>List of strings, keywords of layer names to exclude.
+All layers with keywords in their path will use AdamW.</p></dd>
+
+
+<dt id="arg-exclude-embeddings">exclude_embeddings<a class="anchor" aria-label="anchor" href="#arg-exclude-embeddings"></a></dt>
+<dd><p>Boolean value.
+If <code>TRUE</code>, embedding layers will use AdamW.</p></dd>
+
+
+<dt id="arg-muon-a">muon_a<a class="anchor" aria-label="anchor" href="#arg-muon-a"></a></dt>
+<dd><p>Float, parameter a of the muon algorithm.
+It is recommended to use the default value.</p></dd>
+
+
+<dt id="arg-muon-b">muon_b<a class="anchor" aria-label="anchor" href="#arg-muon-b"></a></dt>
+<dd><p>Float, parameter b of the muon algorithm.
+It is recommended to use the default value.</p></dd>
+
+
+<dt id="arg-muon-c">muon_c<a class="anchor" aria-label="anchor" href="#arg-muon-c"></a></dt>
+<dd><p>Float, parameter c of the muon algorithm.
+It is recommended to use the default value.</p></dd>
+
+
+<dt id="arg-adam-lr-ratio">adam_lr_ratio<a class="anchor" aria-label="anchor" href="#arg-adam-lr-ratio"></a></dt>
+<dd><p>Float, the ratio of the learning rate when
+using Adam to the main learning rate.
+it is recommended to set it to <code>0.1</code>.</p></dd>
+
+
+<dt id="arg-momentum">momentum<a class="anchor" aria-label="anchor" href="#arg-momentum"></a></dt>
+<dd><p>Float, momentum used by internal SGD.</p></dd>
+
+
+<dt id="arg-ns-steps">ns_steps<a class="anchor" aria-label="anchor" href="#arg-ns-steps"></a></dt>
+<dd><p>Integer, number of Newton-Schulz iterations to run.</p></dd>
+
+
+<dt id="arg-nesterov">nesterov<a class="anchor" aria-label="anchor" href="#arg-nesterov"></a></dt>
+<dd><p>Boolean, whether to use Nesterov-style momentum.</p></dd>
+
+
+<dt id="arg--">...<a class="anchor" aria-label="anchor" href="#arg--"></a></dt>
+<dd><p>For forward/backward compatibility.</p></dd>
+
+</dl></div>
+    <div class="section level2">
+    <h2 id="value">Value<a class="anchor" aria-label="anchor" href="#value"></a></h2>
+    <p>an <code>Optimizer</code> instance</p>
+    </div>
+    <div class="section level2">
+    <h2 id="references">References<a class="anchor" aria-label="anchor" href="#references"></a></h2>
+
+<ul><li><p><a href="https://github.com/KellerJordan/Muon" class="external-link">Original implementation</a></p></li>
+<li><p><a href="https://arxiv.org/abs/2502.16982" class="external-link">Liu et al, 2025</a></p></li>
+</ul></div>
+    <div class="section level2">
+    <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
+    <div class="dont-index"><p>Other optimizers: <br><code><a href="optimizer_adadelta.html">optimizer_adadelta</a>()</code> <br><code><a href="optimizer_adafactor.html">optimizer_adafactor</a>()</code> <br><code><a href="optimizer_adagrad.html">optimizer_adagrad</a>()</code> <br><code><a href="optimizer_adam.html">optimizer_adam</a>()</code> <br><code><a href="optimizer_adam_w.html">optimizer_adam_w</a>()</code> <br><code><a href="optimizer_adamax.html">optimizer_adamax</a>()</code> <br><code><a href="optimizer_ftrl.html">optimizer_ftrl</a>()</code> <br><code><a href="optimizer_lamb.html">optimizer_lamb</a>()</code> <br><code><a href="optimizer_lion.html">optimizer_lion</a>()</code> <br><code><a href="optimizer_loss_scale.html">optimizer_loss_scale</a>()</code> <br><code><a href="optimizer_nadam.html">optimizer_nadam</a>()</code> <br><code><a href="optimizer_rmsprop.html">optimizer_rmsprop</a>()</code> <br><code><a href="optimizer_sgd.html">optimizer_sgd</a>()</code> <br></p></div>
+    </div>
+
+  </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
+    </nav></aside></div>
+
+
+    <footer><div class="pkgdown-footer-left">
+  <p>Developed by Tomasz Kalinowski, JJ Allaire, François Chollet, Posit Software, PBC, Google.</p>
+</div>
+
+<div class="pkgdown-footer-right">
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
+</div>
+
+    </footer></div>
+
+
+
+
+
+  </body></html>
+
diff --git a/docs/dev/reference/optimizer_nadam.html b/docs/dev/reference/optimizer_nadam.html
index 2fdcd1eb4..28072e4ff 100644
--- a/docs/dev/reference/optimizer_nadam.html
+++ b/docs/dev/reference/optimizer_nadam.html
@@ -10,7 +10,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -207,7 +207,7 @@ <h2 id="reference">Reference<a class="anchor" aria-label="anchor" href="#referen
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
     <div class="dont-index">
 <ul><li><p><a href="https://keras.io/api/optimizers/Nadam#nadam-class" class="external-link">https://keras.io/api/optimizers/Nadam#nadam-class</a></p></li>
-</ul><p>Other optimizers: <br><code><a href="optimizer_adadelta.html">optimizer_adadelta</a>()</code> <br><code><a href="optimizer_adafactor.html">optimizer_adafactor</a>()</code> <br><code><a href="optimizer_adagrad.html">optimizer_adagrad</a>()</code> <br><code><a href="optimizer_adam.html">optimizer_adam</a>()</code> <br><code><a href="optimizer_adam_w.html">optimizer_adam_w</a>()</code> <br><code><a href="optimizer_adamax.html">optimizer_adamax</a>()</code> <br><code><a href="optimizer_ftrl.html">optimizer_ftrl</a>()</code> <br><code><a href="optimizer_lamb.html">optimizer_lamb</a>()</code> <br><code><a href="optimizer_lion.html">optimizer_lion</a>()</code> <br><code><a href="optimizer_loss_scale.html">optimizer_loss_scale</a>()</code> <br><code><a href="optimizer_rmsprop.html">optimizer_rmsprop</a>()</code> <br><code><a href="optimizer_sgd.html">optimizer_sgd</a>()</code> <br></p></div>
+</ul><p>Other optimizers: <br><code><a href="optimizer_adadelta.html">optimizer_adadelta</a>()</code> <br><code><a href="optimizer_adafactor.html">optimizer_adafactor</a>()</code> <br><code><a href="optimizer_adagrad.html">optimizer_adagrad</a>()</code> <br><code><a href="optimizer_adam.html">optimizer_adam</a>()</code> <br><code><a href="optimizer_adam_w.html">optimizer_adam_w</a>()</code> <br><code><a href="optimizer_adamax.html">optimizer_adamax</a>()</code> <br><code><a href="optimizer_ftrl.html">optimizer_ftrl</a>()</code> <br><code><a href="optimizer_lamb.html">optimizer_lamb</a>()</code> <br><code><a href="optimizer_lion.html">optimizer_lion</a>()</code> <br><code><a href="optimizer_loss_scale.html">optimizer_loss_scale</a>()</code> <br><code><a href="optimizer_muon.html">optimizer_muon</a>()</code> <br><code><a href="optimizer_rmsprop.html">optimizer_rmsprop</a>()</code> <br><code><a href="optimizer_sgd.html">optimizer_sgd</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -219,7 +219,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/optimizer_rmsprop.html b/docs/dev/reference/optimizer_rmsprop.html
index 5f534fb3c..482b18531 100644
--- a/docs/dev/reference/optimizer_rmsprop.html
+++ b/docs/dev/reference/optimizer_rmsprop.html
@@ -22,7 +22,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -229,7 +229,7 @@ <h2 id="reference">Reference<a class="anchor" aria-label="anchor" href="#referen
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
     <div class="dont-index">
 <ul><li><p><a href="https://keras.io/api/optimizers/rmsprop#rmsprop-class" class="external-link">https://keras.io/api/optimizers/rmsprop#rmsprop-class</a></p></li>
-</ul><p>Other optimizers: <br><code><a href="optimizer_adadelta.html">optimizer_adadelta</a>()</code> <br><code><a href="optimizer_adafactor.html">optimizer_adafactor</a>()</code> <br><code><a href="optimizer_adagrad.html">optimizer_adagrad</a>()</code> <br><code><a href="optimizer_adam.html">optimizer_adam</a>()</code> <br><code><a href="optimizer_adam_w.html">optimizer_adam_w</a>()</code> <br><code><a href="optimizer_adamax.html">optimizer_adamax</a>()</code> <br><code><a href="optimizer_ftrl.html">optimizer_ftrl</a>()</code> <br><code><a href="optimizer_lamb.html">optimizer_lamb</a>()</code> <br><code><a href="optimizer_lion.html">optimizer_lion</a>()</code> <br><code><a href="optimizer_loss_scale.html">optimizer_loss_scale</a>()</code> <br><code><a href="optimizer_nadam.html">optimizer_nadam</a>()</code> <br><code><a href="optimizer_sgd.html">optimizer_sgd</a>()</code> <br></p></div>
+</ul><p>Other optimizers: <br><code><a href="optimizer_adadelta.html">optimizer_adadelta</a>()</code> <br><code><a href="optimizer_adafactor.html">optimizer_adafactor</a>()</code> <br><code><a href="optimizer_adagrad.html">optimizer_adagrad</a>()</code> <br><code><a href="optimizer_adam.html">optimizer_adam</a>()</code> <br><code><a href="optimizer_adam_w.html">optimizer_adam_w</a>()</code> <br><code><a href="optimizer_adamax.html">optimizer_adamax</a>()</code> <br><code><a href="optimizer_ftrl.html">optimizer_ftrl</a>()</code> <br><code><a href="optimizer_lamb.html">optimizer_lamb</a>()</code> <br><code><a href="optimizer_lion.html">optimizer_lion</a>()</code> <br><code><a href="optimizer_loss_scale.html">optimizer_loss_scale</a>()</code> <br><code><a href="optimizer_muon.html">optimizer_muon</a>()</code> <br><code><a href="optimizer_nadam.html">optimizer_nadam</a>()</code> <br><code><a href="optimizer_sgd.html">optimizer_sgd</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -241,7 +241,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/optimizer_sgd.html b/docs/dev/reference/optimizer_sgd.html
index 9a9931777..6a283cfb8 100644
--- a/docs/dev/reference/optimizer_sgd.html
+++ b/docs/dev/reference/optimizer_sgd.html
@@ -28,7 +28,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -215,7 +215,7 @@ <h2 id="value">Value<a class="anchor" aria-label="anchor" href="#value"></a></h2
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
     <div class="dont-index">
 <ul><li><p><a href="https://keras.io/api/optimizers/sgd#sgd-class" class="external-link">https://keras.io/api/optimizers/sgd#sgd-class</a></p></li>
-</ul><p>Other optimizers: <br><code><a href="optimizer_adadelta.html">optimizer_adadelta</a>()</code> <br><code><a href="optimizer_adafactor.html">optimizer_adafactor</a>()</code> <br><code><a href="optimizer_adagrad.html">optimizer_adagrad</a>()</code> <br><code><a href="optimizer_adam.html">optimizer_adam</a>()</code> <br><code><a href="optimizer_adam_w.html">optimizer_adam_w</a>()</code> <br><code><a href="optimizer_adamax.html">optimizer_adamax</a>()</code> <br><code><a href="optimizer_ftrl.html">optimizer_ftrl</a>()</code> <br><code><a href="optimizer_lamb.html">optimizer_lamb</a>()</code> <br><code><a href="optimizer_lion.html">optimizer_lion</a>()</code> <br><code><a href="optimizer_loss_scale.html">optimizer_loss_scale</a>()</code> <br><code><a href="optimizer_nadam.html">optimizer_nadam</a>()</code> <br><code><a href="optimizer_rmsprop.html">optimizer_rmsprop</a>()</code> <br></p></div>
+</ul><p>Other optimizers: <br><code><a href="optimizer_adadelta.html">optimizer_adadelta</a>()</code> <br><code><a href="optimizer_adafactor.html">optimizer_adafactor</a>()</code> <br><code><a href="optimizer_adagrad.html">optimizer_adagrad</a>()</code> <br><code><a href="optimizer_adam.html">optimizer_adam</a>()</code> <br><code><a href="optimizer_adam_w.html">optimizer_adam_w</a>()</code> <br><code><a href="optimizer_adamax.html">optimizer_adamax</a>()</code> <br><code><a href="optimizer_ftrl.html">optimizer_ftrl</a>()</code> <br><code><a href="optimizer_lamb.html">optimizer_lamb</a>()</code> <br><code><a href="optimizer_lion.html">optimizer_lion</a>()</code> <br><code><a href="optimizer_loss_scale.html">optimizer_loss_scale</a>()</code> <br><code><a href="optimizer_muon.html">optimizer_muon</a>()</code> <br><code><a href="optimizer_nadam.html">optimizer_nadam</a>()</code> <br><code><a href="optimizer_rmsprop.html">optimizer_rmsprop</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -227,7 +227,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/pad_sequences.html b/docs/dev/reference/pad_sequences.html
index b416d3c33..05976a8d5 100644
--- a/docs/dev/reference/pad_sequences.html
+++ b/docs/dev/reference/pad_sequences.html
@@ -98,7 +98,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -250,7 +250,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/pipe.html b/docs/dev/reference/pipe.html
index 36d0f475c..01b999366 100644
--- a/docs/dev/reference/pipe.html
+++ b/docs/dev/reference/pipe.html
@@ -8,7 +8,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -78,7 +78,7 @@ <h2 id="value">Value<a class="anchor" aria-label="anchor" href="#value"></a></h2
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/plot.keras.src.models.model.Model.html b/docs/dev/reference/plot.keras.src.models.model.Model.html
index 0e3cb178b..2c9c4f5b4 100644
--- a/docs/dev/reference/plot.keras.src.models.model.Model.html
+++ b/docs/dev/reference/plot.keras.src.models.model.Model.html
@@ -8,7 +8,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -142,27 +142,29 @@ <h2 id="arguments">Arguments<a class="anchor" aria-label="anchor" href="#argumen
 </dl></div>
     <div class="section level2">
     <h2 id="value">Value<a class="anchor" aria-label="anchor" href="#value"></a></h2>
-    <p>Nothing, called for it's side effects.</p>
+    <p>Nothing, called for it side effects.</p>
     </div>
     <div class="section level2">
     <h2 id="raises">Raises<a class="anchor" aria-label="anchor" href="#raises"></a></h2>
     <p>ValueError: if <code>plot(model)</code> is called before the model is
-built, unless a <code>input_shape = </code> argument was supplied to
+built, unless an <code>input_shape = </code> argument was supplied to
 <code><a href="keras_model_sequential.html">keras_model_sequential()</a></code>.</p>
     </div>
     <div class="section level2">
     <h2 id="requirements">Requirements<a class="anchor" aria-label="anchor" href="#requirements"></a></h2>
 
 
-<p>This function requires pydot and graphviz.
-<code>pydot</code> is by default installed by <code><a href="install_keras.html">install_keras()</a></code>, but if you installed
-keras by other means, you can install <code>pydot</code> directly with :</p>
+<p>This function requires pydot and graphviz.</p>
+<p><code>pydot</code> is by default installed by <code><a href="install_keras.html">install_keras()</a></code>, but if you installed
+Keras by other means, you can install <code>pydot</code> directly with:</p>
 <p></p><div class="sourceCode r"><pre><code><span><span class="fu">reticulate</span><span class="fu">::</span><span class="fu"><a href="https://rstudio.github.io/reticulate/reference/py_install.html" class="external-link">py_install</a></span><span class="op">(</span><span class="st">"pydot"</span>, pip <span class="op">=</span> <span class="cn">TRUE</span><span class="op">)</span></span></code></pre><p></p></div>
 <p>You can install graphviz directly from here:
 <a href="https://graphviz.gitlab.io/download/" class="external-link">https://graphviz.gitlab.io/download/</a></p>
 <p>On most Linux platforms, can install graphviz via the package manager.
 For example, on Ubuntu/Debian you can install with</p>
 <p></p><div class="sourceCode sh"><pre><code><span id="cb1-1"><a href="#cb1-1" tabindex="-1"></a><span class="fu">sudo</span> apt install graphviz</span></code></pre><p></p></div>
+<p>On macOS you can install graphviz using <code>brew</code>:</p>
+<p></p><div class="sourceCode sh"><pre><code><span id="cb1-1"><a href="#cb1-1" tabindex="-1"></a><span class="ex">brew</span> install graphviz</span></code></pre><p></p></div>
 <p>In a conda environment, you can install graphviz with:</p>
 <p></p><div class="sourceCode r"><pre><code><span><span class="fu">reticulate</span><span class="fu">::</span><span class="fu"><a href="https://rstudio.github.io/reticulate/reference/conda-tools.html" class="external-link">conda_install</a></span><span class="op">(</span>packages <span class="op">=</span> <span class="st">"graphviz"</span><span class="op">)</span></span>
 <span><span class="co"># Restart the R session after install.</span></span></code></pre><p></p></div>
@@ -177,7 +179,7 @@ <h2 id="requirements">Requirements<a class="anchor" aria-label="anchor" href="#r
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/plot.keras_training_history.html b/docs/dev/reference/plot.keras_training_history.html
index 9024251b0..652dcfe19 100644
--- a/docs/dev/reference/plot.keras_training_history.html
+++ b/docs/dev/reference/plot.keras_training_history.html
@@ -8,7 +8,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -125,7 +125,7 @@ <h2 id="value">Value<a class="anchor" aria-label="anchor" href="#value"></a></h2
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/pop_layer.html b/docs/dev/reference/pop_layer.html
index 8e9ff1db0..b47bbce4c 100644
--- a/docs/dev/reference/pop_layer.html
+++ b/docs/dev/reference/pop_layer.html
@@ -8,7 +8,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -93,7 +93,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/predict.keras.src.models.model.Model.html b/docs/dev/reference/predict.keras.src.models.model.Model.html
index d72ee9fb4..95efc8390 100644
--- a/docs/dev/reference/predict.keras.src.models.model.Model.html
+++ b/docs/dev/reference/predict.keras.src.models.model.Model.html
@@ -8,7 +8,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -163,7 +163,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/predict_on_batch.html b/docs/dev/reference/predict_on_batch.html
index a49b7e036..ac4b3d585 100644
--- a/docs/dev/reference/predict_on_batch.html
+++ b/docs/dev/reference/predict_on_batch.html
@@ -8,7 +8,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -94,7 +94,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/process_utils.html b/docs/dev/reference/process_utils.html
index e11d46ebd..45fefe6d6 100644
--- a/docs/dev/reference/process_utils.html
+++ b/docs/dev/reference/process_utils.html
@@ -10,7 +10,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -135,7 +135,7 @@ <h2 id="ref-examples">Examples<a class="anchor" aria-label="anchor" href="#ref-e
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/quantize_weights.html b/docs/dev/reference/quantize_weights.html
index 192fb70fd..eba66915e 100644
--- a/docs/dev/reference/quantize_weights.html
+++ b/docs/dev/reference/quantize_weights.html
@@ -14,7 +14,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -107,7 +107,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/random_beta.html b/docs/dev/reference/random_beta.html
index 7f981ad57..e174753d8 100644
--- a/docs/dev/reference/random_beta.html
+++ b/docs/dev/reference/random_beta.html
@@ -10,7 +10,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -124,7 +124,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/random_binomial.html b/docs/dev/reference/random_binomial.html
index f0e651393..0898bb1a7 100644
--- a/docs/dev/reference/random_binomial.html
+++ b/docs/dev/reference/random_binomial.html
@@ -10,7 +10,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -127,7 +127,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/random_categorical.html b/docs/dev/reference/random_categorical.html
index 251cad5af..cdfa1ed6c 100644
--- a/docs/dev/reference/random_categorical.html
+++ b/docs/dev/reference/random_categorical.html
@@ -70,7 +70,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -213,7 +213,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/random_dropout.html b/docs/dev/reference/random_dropout.html
index 8623bcb11..42ae3d105 100644
--- a/docs/dev/reference/random_dropout.html
+++ b/docs/dev/reference/random_dropout.html
@@ -8,7 +8,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -100,7 +100,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/random_gamma.html b/docs/dev/reference/random_gamma.html
index d076b666d..eff0bf42a 100644
--- a/docs/dev/reference/random_gamma.html
+++ b/docs/dev/reference/random_gamma.html
@@ -8,7 +8,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -115,7 +115,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/random_integer.html b/docs/dev/reference/random_integer.html
index ef897dad8..187c88c1e 100644
--- a/docs/dev/reference/random_integer.html
+++ b/docs/dev/reference/random_integer.html
@@ -14,7 +14,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -128,7 +128,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/random_normal.html b/docs/dev/reference/random_normal.html
index 9f637afc4..e1829521b 100644
--- a/docs/dev/reference/random_normal.html
+++ b/docs/dev/reference/random_normal.html
@@ -8,7 +8,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -120,7 +120,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/random_seed_generator.html b/docs/dev/reference/random_seed_generator.html
index 87abba423..5f564390f 100644
--- a/docs/dev/reference/random_seed_generator.html
+++ b/docs/dev/reference/random_seed_generator.html
@@ -36,7 +36,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -164,7 +164,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/random_shuffle.html b/docs/dev/reference/random_shuffle.html
index 5e0d7be40..3a1b32366 100644
--- a/docs/dev/reference/random_shuffle.html
+++ b/docs/dev/reference/random_shuffle.html
@@ -8,7 +8,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -109,7 +109,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/random_truncated_normal.html b/docs/dev/reference/random_truncated_normal.html
index a08e76fa3..be70fb46e 100644
--- a/docs/dev/reference/random_truncated_normal.html
+++ b/docs/dev/reference/random_truncated_normal.html
@@ -12,7 +12,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -126,7 +126,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/random_uniform.html b/docs/dev/reference/random_uniform.html
index 5e97a3e8b..9937604ad 100644
--- a/docs/dev/reference/random_uniform.html
+++ b/docs/dev/reference/random_uniform.html
@@ -14,7 +14,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -130,7 +130,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/reexports.html b/docs/dev/reference/reexports.html
index 8d175fe79..c9669d412 100644
--- a/docs/dev/reference/reexports.html
+++ b/docs/dev/reference/reexports.html
@@ -7,7 +7,7 @@
 
 
   generics
-compile, fit
+compile, fit, generate
 
 
   magrittr
@@ -34,7 +34,7 @@
 
 
   generics
-compile, fit
+compile, fit, generate
 
 
   magrittr
@@ -62,7 +62,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -115,7 +115,7 @@ <h1>Objects exported from other packages</h1>
 
 
   <dt>generics</dt>
-<dd><p><code><a href="https://generics.r-lib.org/reference/compile.html" class="external-link">compile</a></code>, <code><a href="https://generics.r-lib.org/reference/fit.html" class="external-link">fit</a></code></p></dd>
+<dd><p><code><a href="https://generics.r-lib.org/reference/compile.html" class="external-link">compile</a></code>, <code><a href="https://generics.r-lib.org/reference/fit.html" class="external-link">fit</a></code>, <code><a href="https://generics.r-lib.org/reference/generate.html" class="external-link">generate</a></code></p></dd>
 
 
   <dt>magrittr</dt>
@@ -146,7 +146,7 @@ <h1>Objects exported from other packages</h1>
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/register_keras_serializable.html b/docs/dev/reference/register_keras_serializable.html
index a4974ed9a..addfb5bbe 100644
--- a/docs/dev/reference/register_keras_serializable.html
+++ b/docs/dev/reference/register_keras_serializable.html
@@ -22,7 +22,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -134,7 +134,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/regularizer_l1.html b/docs/dev/reference/regularizer_l1.html
index 4ddd147b3..cfc225996 100644
--- a/docs/dev/reference/regularizer_l1.html
+++ b/docs/dev/reference/regularizer_l1.html
@@ -18,7 +18,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -105,7 +105,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/regularizer_l1_l2.html b/docs/dev/reference/regularizer_l1_l2.html
index 72c196840..61f1321ad 100644
--- a/docs/dev/reference/regularizer_l1_l2.html
+++ b/docs/dev/reference/regularizer_l1_l2.html
@@ -22,7 +22,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -115,7 +115,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/regularizer_l2.html b/docs/dev/reference/regularizer_l2.html
index fb261964b..7a00dbcc8 100644
--- a/docs/dev/reference/regularizer_l2.html
+++ b/docs/dev/reference/regularizer_l2.html
@@ -18,7 +18,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -105,7 +105,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/regularizer_orthogonal.html b/docs/dev/reference/regularizer_orthogonal.html
index 39e6a7321..1d838be23 100644
--- a/docs/dev/reference/regularizer_orthogonal.html
+++ b/docs/dev/reference/regularizer_orthogonal.html
@@ -14,7 +14,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -116,7 +116,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/reset_state.html b/docs/dev/reference/reset_state.html
index e5ebb16a3..28c16c110 100644
--- a/docs/dev/reference/reset_state.html
+++ b/docs/dev/reference/reset_state.html
@@ -8,7 +8,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -91,7 +91,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/rnn_cell_gru.html b/docs/dev/reference/rnn_cell_gru.html
index af82663bb..dd2e4adc0 100644
--- a/docs/dev/reference/rnn_cell_gru.html
+++ b/docs/dev/reference/rnn_cell_gru.html
@@ -10,7 +10,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -217,7 +217,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
     <div class="dont-index"><p>Other rnn cells: <br><code><a href="layer_rnn.html">layer_rnn</a>()</code> <br><code><a href="rnn_cell_lstm.html">rnn_cell_lstm</a>()</code> <br><code><a href="rnn_cell_simple.html">rnn_cell_simple</a>()</code> <br></p>
 <p>Other gru rnn layers: <br><code><a href="layer_gru.html">layer_gru</a>()</code> <br></p>
 <p>Other rnn layers: <br><code><a href="layer_bidirectional.html">layer_bidirectional</a>()</code> <br><code><a href="layer_conv_lstm_1d.html">layer_conv_lstm_1d</a>()</code> <br><code><a href="layer_conv_lstm_2d.html">layer_conv_lstm_2d</a>()</code> <br><code><a href="layer_conv_lstm_3d.html">layer_conv_lstm_3d</a>()</code> <br><code><a href="layer_gru.html">layer_gru</a>()</code> <br><code><a href="layer_lstm.html">layer_lstm</a>()</code> <br><code><a href="layer_rnn.html">layer_rnn</a>()</code> <br><code><a href="layer_simple_rnn.html">layer_simple_rnn</a>()</code> <br><code><a href="layer_time_distributed.html">layer_time_distributed</a>()</code> <br><code><a href="rnn_cell_lstm.html">rnn_cell_lstm</a>()</code> <br><code><a href="rnn_cell_simple.html">rnn_cell_simple</a>()</code> <br><code><a href="rnn_cells_stack.html">rnn_cells_stack</a>()</code> <br></p>
-<p>Other layers: <br><code><a href="Layer.html">Layer</a>()</code> <br><code><a href="layer_activation.html">layer_activation</a>()</code> <br><code><a href="layer_activation_elu.html">layer_activation_elu</a>()</code> <br><code><a href="layer_activation_leaky_relu.html">layer_activation_leaky_relu</a>()</code> <br><code><a href="layer_activation_parametric_relu.html">layer_activation_parametric_relu</a>()</code> <br><code><a href="layer_activation_relu.html">layer_activation_relu</a>()</code> <br><code><a href="layer_activation_softmax.html">layer_activation_softmax</a>()</code> <br><code><a href="layer_activity_regularization.html">layer_activity_regularization</a>()</code> <br><code><a href="layer_add.html">layer_add</a>()</code> <br><code><a href="layer_additive_attention.html">layer_additive_attention</a>()</code> <br><code><a href="layer_alpha_dropout.html">layer_alpha_dropout</a>()</code> <br><code><a href="layer_attention.html">layer_attention</a>()</code> <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_average.html">layer_average</a>()</code> <br><code><a href="layer_average_pooling_1d.html">layer_average_pooling_1d</a>()</code> <br><code><a href="layer_average_pooling_2d.html">layer_average_pooling_2d</a>()</code> <br><code><a href="layer_average_pooling_3d.html">layer_average_pooling_3d</a>()</code> <br><code><a href="layer_batch_normalization.html">layer_batch_normalization</a>()</code> <br><code><a href="layer_bidirectional.html">layer_bidirectional</a>()</code> <br><code><a href="layer_category_encoding.html">layer_category_encoding</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_concatenate.html">layer_concatenate</a>()</code> <br><code><a href="layer_conv_1d.html">layer_conv_1d</a>()</code> <br><code><a href="layer_conv_1d_transpose.html">layer_conv_1d_transpose</a>()</code> <br><code><a href="layer_conv_2d.html">layer_conv_2d</a>()</code> <br><code><a href="layer_conv_2d_transpose.html">layer_conv_2d_transpose</a>()</code> <br><code><a href="layer_conv_3d.html">layer_conv_3d</a>()</code> <br><code><a href="layer_conv_3d_transpose.html">layer_conv_3d_transpose</a>()</code> <br><code><a href="layer_conv_lstm_1d.html">layer_conv_lstm_1d</a>()</code> <br><code><a href="layer_conv_lstm_2d.html">layer_conv_lstm_2d</a>()</code> <br><code><a href="layer_conv_lstm_3d.html">layer_conv_lstm_3d</a>()</code> <br><code><a href="layer_cropping_1d.html">layer_cropping_1d</a>()</code> <br><code><a href="layer_cropping_2d.html">layer_cropping_2d</a>()</code> <br><code><a href="layer_cropping_3d.html">layer_cropping_3d</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_dense.html">layer_dense</a>()</code> <br><code><a href="layer_depthwise_conv_1d.html">layer_depthwise_conv_1d</a>()</code> <br><code><a href="layer_depthwise_conv_2d.html">layer_depthwise_conv_2d</a>()</code> <br><code><a href="layer_discretization.html">layer_discretization</a>()</code> <br><code><a href="layer_dot.html">layer_dot</a>()</code> <br><code><a href="layer_dropout.html">layer_dropout</a>()</code> <br><code><a href="layer_einsum_dense.html">layer_einsum_dense</a>()</code> <br><code><a href="layer_embedding.html">layer_embedding</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_feature_space.html">layer_feature_space</a>()</code> <br><code><a href="layer_flatten.html">layer_flatten</a>()</code> <br><code><a href="layer_flax_module_wrapper.html">layer_flax_module_wrapper</a>()</code> <br><code><a href="layer_gaussian_dropout.html">layer_gaussian_dropout</a>()</code> <br><code><a href="layer_gaussian_noise.html">layer_gaussian_noise</a>()</code> <br><code><a href="layer_global_average_pooling_1d.html">layer_global_average_pooling_1d</a>()</code> <br><code><a href="layer_global_average_pooling_2d.html">layer_global_average_pooling_2d</a>()</code> <br><code><a href="layer_global_average_pooling_3d.html">layer_global_average_pooling_3d</a>()</code> <br><code><a href="layer_global_max_pooling_1d.html">layer_global_max_pooling_1d</a>()</code> <br><code><a href="layer_global_max_pooling_2d.html">layer_global_max_pooling_2d</a>()</code> <br><code><a href="layer_global_max_pooling_3d.html">layer_global_max_pooling_3d</a>()</code> <br><code><a href="layer_group_normalization.html">layer_group_normalization</a>()</code> <br><code><a href="layer_group_query_attention.html">layer_group_query_attention</a>()</code> <br><code><a href="layer_gru.html">layer_gru</a>()</code> <br><code><a href="layer_hashed_crossing.html">layer_hashed_crossing</a>()</code> <br><code><a href="layer_hashing.html">layer_hashing</a>()</code> <br><code><a href="layer_identity.html">layer_identity</a>()</code> <br><code><a href="layer_integer_lookup.html">layer_integer_lookup</a>()</code> <br><code><a href="layer_jax_model_wrapper.html">layer_jax_model_wrapper</a>()</code> <br><code><a href="layer_lambda.html">layer_lambda</a>()</code> <br><code><a href="layer_layer_normalization.html">layer_layer_normalization</a>()</code> <br><code><a href="layer_lstm.html">layer_lstm</a>()</code> <br><code><a href="layer_masking.html">layer_masking</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_max_pooling_1d.html">layer_max_pooling_1d</a>()</code> <br><code><a href="layer_max_pooling_2d.html">layer_max_pooling_2d</a>()</code> <br><code><a href="layer_max_pooling_3d.html">layer_max_pooling_3d</a>()</code> <br><code><a href="layer_maximum.html">layer_maximum</a>()</code> <br><code><a href="layer_mel_spectrogram.html">layer_mel_spectrogram</a>()</code> <br><code><a href="layer_minimum.html">layer_minimum</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_multi_head_attention.html">layer_multi_head_attention</a>()</code> <br><code><a href="layer_multiply.html">layer_multiply</a>()</code> <br><code><a href="layer_normalization.html">layer_normalization</a>()</code> <br><code><a href="layer_permute.html">layer_permute</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_brightness.html">layer_random_brightness</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_contrast.html">layer_random_contrast</a>()</code> <br><code><a href="layer_random_crop.html">layer_random_crop</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_flip.html">layer_random_flip</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_rotation.html">layer_random_rotation</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_random_translation.html">layer_random_translation</a>()</code> <br><code><a href="layer_random_zoom.html">layer_random_zoom</a>()</code> <br><code><a href="layer_repeat_vector.html">layer_repeat_vector</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_reshape.html">layer_reshape</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_rms_normalization.html">layer_rms_normalization</a>()</code> <br><code><a href="layer_rnn.html">layer_rnn</a>()</code> <br><code><a href="layer_separable_conv_1d.html">layer_separable_conv_1d</a>()</code> <br><code><a href="layer_separable_conv_2d.html">layer_separable_conv_2d</a>()</code> <br><code><a href="layer_simple_rnn.html">layer_simple_rnn</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br><code><a href="layer_spatial_dropout_1d.html">layer_spatial_dropout_1d</a>()</code> <br><code><a href="layer_spatial_dropout_2d.html">layer_spatial_dropout_2d</a>()</code> <br><code><a href="layer_spatial_dropout_3d.html">layer_spatial_dropout_3d</a>()</code> <br><code><a href="layer_spectral_normalization.html">layer_spectral_normalization</a>()</code> <br><code><a href="layer_stft_spectrogram.html">layer_stft_spectrogram</a>()</code> <br><code><a href="layer_string_lookup.html">layer_string_lookup</a>()</code> <br><code><a href="layer_subtract.html">layer_subtract</a>()</code> <br><code><a href="layer_text_vectorization.html">layer_text_vectorization</a>()</code> <br><code><a href="layer_tfsm.html">layer_tfsm</a>()</code> <br><code><a href="layer_time_distributed.html">layer_time_distributed</a>()</code> <br><code><a href="layer_torch_module_wrapper.html">layer_torch_module_wrapper</a>()</code> <br><code><a href="layer_unit_normalization.html">layer_unit_normalization</a>()</code> <br><code><a href="layer_upsampling_1d.html">layer_upsampling_1d</a>()</code> <br><code><a href="layer_upsampling_2d.html">layer_upsampling_2d</a>()</code> <br><code><a href="layer_upsampling_3d.html">layer_upsampling_3d</a>()</code> <br><code><a href="layer_zero_padding_1d.html">layer_zero_padding_1d</a>()</code> <br><code><a href="layer_zero_padding_2d.html">layer_zero_padding_2d</a>()</code> <br><code><a href="layer_zero_padding_3d.html">layer_zero_padding_3d</a>()</code> <br><code><a href="rnn_cell_lstm.html">rnn_cell_lstm</a>()</code> <br><code><a href="rnn_cell_simple.html">rnn_cell_simple</a>()</code> <br><code><a href="rnn_cells_stack.html">rnn_cells_stack</a>()</code> <br></p></div>
+<p>Other layers: <br><code><a href="Layer.html">Layer</a>()</code> <br><code><a href="layer_activation.html">layer_activation</a>()</code> <br><code><a href="layer_activation_elu.html">layer_activation_elu</a>()</code> <br><code><a href="layer_activation_leaky_relu.html">layer_activation_leaky_relu</a>()</code> <br><code><a href="layer_activation_parametric_relu.html">layer_activation_parametric_relu</a>()</code> <br><code><a href="layer_activation_relu.html">layer_activation_relu</a>()</code> <br><code><a href="layer_activation_softmax.html">layer_activation_softmax</a>()</code> <br><code><a href="layer_activity_regularization.html">layer_activity_regularization</a>()</code> <br><code><a href="layer_add.html">layer_add</a>()</code> <br><code><a href="layer_additive_attention.html">layer_additive_attention</a>()</code> <br><code><a href="layer_alpha_dropout.html">layer_alpha_dropout</a>()</code> <br><code><a href="layer_attention.html">layer_attention</a>()</code> <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_average.html">layer_average</a>()</code> <br><code><a href="layer_average_pooling_1d.html">layer_average_pooling_1d</a>()</code> <br><code><a href="layer_average_pooling_2d.html">layer_average_pooling_2d</a>()</code> <br><code><a href="layer_average_pooling_3d.html">layer_average_pooling_3d</a>()</code> <br><code><a href="layer_batch_normalization.html">layer_batch_normalization</a>()</code> <br><code><a href="layer_bidirectional.html">layer_bidirectional</a>()</code> <br><code><a href="layer_category_encoding.html">layer_category_encoding</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_concatenate.html">layer_concatenate</a>()</code> <br><code><a href="layer_conv_1d.html">layer_conv_1d</a>()</code> <br><code><a href="layer_conv_1d_transpose.html">layer_conv_1d_transpose</a>()</code> <br><code><a href="layer_conv_2d.html">layer_conv_2d</a>()</code> <br><code><a href="layer_conv_2d_transpose.html">layer_conv_2d_transpose</a>()</code> <br><code><a href="layer_conv_3d.html">layer_conv_3d</a>()</code> <br><code><a href="layer_conv_3d_transpose.html">layer_conv_3d_transpose</a>()</code> <br><code><a href="layer_conv_lstm_1d.html">layer_conv_lstm_1d</a>()</code> <br><code><a href="layer_conv_lstm_2d.html">layer_conv_lstm_2d</a>()</code> <br><code><a href="layer_conv_lstm_3d.html">layer_conv_lstm_3d</a>()</code> <br><code><a href="layer_cropping_1d.html">layer_cropping_1d</a>()</code> <br><code><a href="layer_cropping_2d.html">layer_cropping_2d</a>()</code> <br><code><a href="layer_cropping_3d.html">layer_cropping_3d</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_dense.html">layer_dense</a>()</code> <br><code><a href="layer_depthwise_conv_1d.html">layer_depthwise_conv_1d</a>()</code> <br><code><a href="layer_depthwise_conv_2d.html">layer_depthwise_conv_2d</a>()</code> <br><code><a href="layer_discretization.html">layer_discretization</a>()</code> <br><code><a href="layer_dot.html">layer_dot</a>()</code> <br><code><a href="layer_dropout.html">layer_dropout</a>()</code> <br><code><a href="layer_einsum_dense.html">layer_einsum_dense</a>()</code> <br><code><a href="layer_embedding.html">layer_embedding</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_feature_space.html">layer_feature_space</a>()</code> <br><code><a href="layer_flatten.html">layer_flatten</a>()</code> <br><code><a href="layer_flax_module_wrapper.html">layer_flax_module_wrapper</a>()</code> <br><code><a href="layer_gaussian_dropout.html">layer_gaussian_dropout</a>()</code> <br><code><a href="layer_gaussian_noise.html">layer_gaussian_noise</a>()</code> <br><code><a href="layer_global_average_pooling_1d.html">layer_global_average_pooling_1d</a>()</code> <br><code><a href="layer_global_average_pooling_2d.html">layer_global_average_pooling_2d</a>()</code> <br><code><a href="layer_global_average_pooling_3d.html">layer_global_average_pooling_3d</a>()</code> <br><code><a href="layer_global_max_pooling_1d.html">layer_global_max_pooling_1d</a>()</code> <br><code><a href="layer_global_max_pooling_2d.html">layer_global_max_pooling_2d</a>()</code> <br><code><a href="layer_global_max_pooling_3d.html">layer_global_max_pooling_3d</a>()</code> <br><code><a href="layer_group_normalization.html">layer_group_normalization</a>()</code> <br><code><a href="layer_group_query_attention.html">layer_group_query_attention</a>()</code> <br><code><a href="layer_gru.html">layer_gru</a>()</code> <br><code><a href="layer_hashed_crossing.html">layer_hashed_crossing</a>()</code> <br><code><a href="layer_hashing.html">layer_hashing</a>()</code> <br><code><a href="layer_identity.html">layer_identity</a>()</code> <br><code><a href="layer_integer_lookup.html">layer_integer_lookup</a>()</code> <br><code><a href="layer_jax_model_wrapper.html">layer_jax_model_wrapper</a>()</code> <br><code><a href="layer_lambda.html">layer_lambda</a>()</code> <br><code><a href="layer_layer_normalization.html">layer_layer_normalization</a>()</code> <br><code><a href="layer_lstm.html">layer_lstm</a>()</code> <br><code><a href="layer_masking.html">layer_masking</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_max_pooling_1d.html">layer_max_pooling_1d</a>()</code> <br><code><a href="layer_max_pooling_2d.html">layer_max_pooling_2d</a>()</code> <br><code><a href="layer_max_pooling_3d.html">layer_max_pooling_3d</a>()</code> <br><code><a href="layer_maximum.html">layer_maximum</a>()</code> <br><code><a href="layer_mel_spectrogram.html">layer_mel_spectrogram</a>()</code> <br><code><a href="layer_minimum.html">layer_minimum</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_multi_head_attention.html">layer_multi_head_attention</a>()</code> <br><code><a href="layer_multiply.html">layer_multiply</a>()</code> <br><code><a href="layer_normalization.html">layer_normalization</a>()</code> <br><code><a href="layer_permute.html">layer_permute</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_brightness.html">layer_random_brightness</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_contrast.html">layer_random_contrast</a>()</code> <br><code><a href="layer_random_crop.html">layer_random_crop</a>()</code> <br><code><a href="layer_random_elastic_transform.html">layer_random_elastic_transform</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_flip.html">layer_random_flip</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_rotation.html">layer_random_rotation</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_random_translation.html">layer_random_translation</a>()</code> <br><code><a href="layer_random_zoom.html">layer_random_zoom</a>()</code> <br><code><a href="layer_repeat_vector.html">layer_repeat_vector</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_reshape.html">layer_reshape</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_rms_normalization.html">layer_rms_normalization</a>()</code> <br><code><a href="layer_rnn.html">layer_rnn</a>()</code> <br><code><a href="layer_separable_conv_1d.html">layer_separable_conv_1d</a>()</code> <br><code><a href="layer_separable_conv_2d.html">layer_separable_conv_2d</a>()</code> <br><code><a href="layer_simple_rnn.html">layer_simple_rnn</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br><code><a href="layer_spatial_dropout_1d.html">layer_spatial_dropout_1d</a>()</code> <br><code><a href="layer_spatial_dropout_2d.html">layer_spatial_dropout_2d</a>()</code> <br><code><a href="layer_spatial_dropout_3d.html">layer_spatial_dropout_3d</a>()</code> <br><code><a href="layer_spectral_normalization.html">layer_spectral_normalization</a>()</code> <br><code><a href="layer_stft_spectrogram.html">layer_stft_spectrogram</a>()</code> <br><code><a href="layer_string_lookup.html">layer_string_lookup</a>()</code> <br><code><a href="layer_subtract.html">layer_subtract</a>()</code> <br><code><a href="layer_text_vectorization.html">layer_text_vectorization</a>()</code> <br><code><a href="layer_tfsm.html">layer_tfsm</a>()</code> <br><code><a href="layer_time_distributed.html">layer_time_distributed</a>()</code> <br><code><a href="layer_torch_module_wrapper.html">layer_torch_module_wrapper</a>()</code> <br><code><a href="layer_unit_normalization.html">layer_unit_normalization</a>()</code> <br><code><a href="layer_upsampling_1d.html">layer_upsampling_1d</a>()</code> <br><code><a href="layer_upsampling_2d.html">layer_upsampling_2d</a>()</code> <br><code><a href="layer_upsampling_3d.html">layer_upsampling_3d</a>()</code> <br><code><a href="layer_zero_padding_1d.html">layer_zero_padding_1d</a>()</code> <br><code><a href="layer_zero_padding_2d.html">layer_zero_padding_2d</a>()</code> <br><code><a href="layer_zero_padding_3d.html">layer_zero_padding_3d</a>()</code> <br><code><a href="rnn_cell_lstm.html">rnn_cell_lstm</a>()</code> <br><code><a href="rnn_cell_simple.html">rnn_cell_simple</a>()</code> <br><code><a href="rnn_cells_stack.html">rnn_cells_stack</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -229,7 +229,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/rnn_cell_lstm.html b/docs/dev/reference/rnn_cell_lstm.html
index 0db143624..08516b472 100644
--- a/docs/dev/reference/rnn_cell_lstm.html
+++ b/docs/dev/reference/rnn_cell_lstm.html
@@ -10,7 +10,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -205,7 +205,7 @@ <h2 id="examples">Examples<a class="anchor" aria-label="anchor" href="#examples"
 <p></p><div class="sourceCode r"><pre><code><span><span class="va">rnn</span> <span class="op">&lt;-</span> <span class="fu"><a href="../reference/layer_rnn.html">layer_rnn</a></span><span class="op">(</span>cell <span class="op">=</span> <span class="fu"><a href="../reference/rnn_cell_lstm.html">rnn_cell_lstm</a></span><span class="op">(</span><span class="fl">4</span><span class="op">)</span>,</span>
 <span>                 return_sequences <span class="op">=</span> <span class="cn">T</span>,</span>
 <span>                 return_state <span class="op">=</span> <span class="cn">T</span><span class="op">)</span></span>
-<span><span class="fu"><a href="https://rdrr.io/r/base/c.html" class="external-link">c</a></span><span class="op">(</span><span class="va">whole_sequence_output</span>, <span class="va">...final_state</span><span class="op">)</span> <span class="op"><a href="../reference/multi-assign.html">%&lt;-%</a></span> <span class="fu">rnn</span><span class="op">(</span><span class="va">inputs</span><span class="op">)</span></span>
+<span><span class="fu"><a href="https://rdrr.io/r/base/c.html" class="external-link">c</a></span><span class="op">(</span><span class="va">whole_sequence_output</span>, <span class="va">..final_state</span><span class="op">)</span> <span class="op"><a href="../reference/multi-assign.html">%&lt;-%</a></span> <span class="fu">rnn</span><span class="op">(</span><span class="va">inputs</span><span class="op">)</span></span>
 <span><span class="fu"><a href="https://rdrr.io/r/utils/str.html" class="external-link">str</a></span><span class="op">(</span><span class="va">whole_sequence_output</span><span class="op">)</span></span></code></pre><p></p></div>
 <p></p><div class="sourceCode"><pre><code><span id="cb1-1"><a href="#cb1-1" tabindex="-1"></a><span class="do">## &lt;tf.Tensor: shape=(32, 10, 4), dtype=float32, numpy=…&gt;</span></span>
 <span id="cb1-2"><a href="#cb1-2" tabindex="-1"></a></span></code></pre><p></p></div>
@@ -220,7 +220,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
     <div class="dont-index"><p>Other rnn cells: <br><code><a href="layer_rnn.html">layer_rnn</a>()</code> <br><code><a href="rnn_cell_gru.html">rnn_cell_gru</a>()</code> <br><code><a href="rnn_cell_simple.html">rnn_cell_simple</a>()</code> <br></p>
 <p>Other lstm rnn layers: <br><code><a href="layer_lstm.html">layer_lstm</a>()</code> <br></p>
 <p>Other rnn layers: <br><code><a href="layer_bidirectional.html">layer_bidirectional</a>()</code> <br><code><a href="layer_conv_lstm_1d.html">layer_conv_lstm_1d</a>()</code> <br><code><a href="layer_conv_lstm_2d.html">layer_conv_lstm_2d</a>()</code> <br><code><a href="layer_conv_lstm_3d.html">layer_conv_lstm_3d</a>()</code> <br><code><a href="layer_gru.html">layer_gru</a>()</code> <br><code><a href="layer_lstm.html">layer_lstm</a>()</code> <br><code><a href="layer_rnn.html">layer_rnn</a>()</code> <br><code><a href="layer_simple_rnn.html">layer_simple_rnn</a>()</code> <br><code><a href="layer_time_distributed.html">layer_time_distributed</a>()</code> <br><code><a href="rnn_cell_gru.html">rnn_cell_gru</a>()</code> <br><code><a href="rnn_cell_simple.html">rnn_cell_simple</a>()</code> <br><code><a href="rnn_cells_stack.html">rnn_cells_stack</a>()</code> <br></p>
-<p>Other layers: <br><code><a href="Layer.html">Layer</a>()</code> <br><code><a href="layer_activation.html">layer_activation</a>()</code> <br><code><a href="layer_activation_elu.html">layer_activation_elu</a>()</code> <br><code><a href="layer_activation_leaky_relu.html">layer_activation_leaky_relu</a>()</code> <br><code><a href="layer_activation_parametric_relu.html">layer_activation_parametric_relu</a>()</code> <br><code><a href="layer_activation_relu.html">layer_activation_relu</a>()</code> <br><code><a href="layer_activation_softmax.html">layer_activation_softmax</a>()</code> <br><code><a href="layer_activity_regularization.html">layer_activity_regularization</a>()</code> <br><code><a href="layer_add.html">layer_add</a>()</code> <br><code><a href="layer_additive_attention.html">layer_additive_attention</a>()</code> <br><code><a href="layer_alpha_dropout.html">layer_alpha_dropout</a>()</code> <br><code><a href="layer_attention.html">layer_attention</a>()</code> <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_average.html">layer_average</a>()</code> <br><code><a href="layer_average_pooling_1d.html">layer_average_pooling_1d</a>()</code> <br><code><a href="layer_average_pooling_2d.html">layer_average_pooling_2d</a>()</code> <br><code><a href="layer_average_pooling_3d.html">layer_average_pooling_3d</a>()</code> <br><code><a href="layer_batch_normalization.html">layer_batch_normalization</a>()</code> <br><code><a href="layer_bidirectional.html">layer_bidirectional</a>()</code> <br><code><a href="layer_category_encoding.html">layer_category_encoding</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_concatenate.html">layer_concatenate</a>()</code> <br><code><a href="layer_conv_1d.html">layer_conv_1d</a>()</code> <br><code><a href="layer_conv_1d_transpose.html">layer_conv_1d_transpose</a>()</code> <br><code><a href="layer_conv_2d.html">layer_conv_2d</a>()</code> <br><code><a href="layer_conv_2d_transpose.html">layer_conv_2d_transpose</a>()</code> <br><code><a href="layer_conv_3d.html">layer_conv_3d</a>()</code> <br><code><a href="layer_conv_3d_transpose.html">layer_conv_3d_transpose</a>()</code> <br><code><a href="layer_conv_lstm_1d.html">layer_conv_lstm_1d</a>()</code> <br><code><a href="layer_conv_lstm_2d.html">layer_conv_lstm_2d</a>()</code> <br><code><a href="layer_conv_lstm_3d.html">layer_conv_lstm_3d</a>()</code> <br><code><a href="layer_cropping_1d.html">layer_cropping_1d</a>()</code> <br><code><a href="layer_cropping_2d.html">layer_cropping_2d</a>()</code> <br><code><a href="layer_cropping_3d.html">layer_cropping_3d</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_dense.html">layer_dense</a>()</code> <br><code><a href="layer_depthwise_conv_1d.html">layer_depthwise_conv_1d</a>()</code> <br><code><a href="layer_depthwise_conv_2d.html">layer_depthwise_conv_2d</a>()</code> <br><code><a href="layer_discretization.html">layer_discretization</a>()</code> <br><code><a href="layer_dot.html">layer_dot</a>()</code> <br><code><a href="layer_dropout.html">layer_dropout</a>()</code> <br><code><a href="layer_einsum_dense.html">layer_einsum_dense</a>()</code> <br><code><a href="layer_embedding.html">layer_embedding</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_feature_space.html">layer_feature_space</a>()</code> <br><code><a href="layer_flatten.html">layer_flatten</a>()</code> <br><code><a href="layer_flax_module_wrapper.html">layer_flax_module_wrapper</a>()</code> <br><code><a href="layer_gaussian_dropout.html">layer_gaussian_dropout</a>()</code> <br><code><a href="layer_gaussian_noise.html">layer_gaussian_noise</a>()</code> <br><code><a href="layer_global_average_pooling_1d.html">layer_global_average_pooling_1d</a>()</code> <br><code><a href="layer_global_average_pooling_2d.html">layer_global_average_pooling_2d</a>()</code> <br><code><a href="layer_global_average_pooling_3d.html">layer_global_average_pooling_3d</a>()</code> <br><code><a href="layer_global_max_pooling_1d.html">layer_global_max_pooling_1d</a>()</code> <br><code><a href="layer_global_max_pooling_2d.html">layer_global_max_pooling_2d</a>()</code> <br><code><a href="layer_global_max_pooling_3d.html">layer_global_max_pooling_3d</a>()</code> <br><code><a href="layer_group_normalization.html">layer_group_normalization</a>()</code> <br><code><a href="layer_group_query_attention.html">layer_group_query_attention</a>()</code> <br><code><a href="layer_gru.html">layer_gru</a>()</code> <br><code><a href="layer_hashed_crossing.html">layer_hashed_crossing</a>()</code> <br><code><a href="layer_hashing.html">layer_hashing</a>()</code> <br><code><a href="layer_identity.html">layer_identity</a>()</code> <br><code><a href="layer_integer_lookup.html">layer_integer_lookup</a>()</code> <br><code><a href="layer_jax_model_wrapper.html">layer_jax_model_wrapper</a>()</code> <br><code><a href="layer_lambda.html">layer_lambda</a>()</code> <br><code><a href="layer_layer_normalization.html">layer_layer_normalization</a>()</code> <br><code><a href="layer_lstm.html">layer_lstm</a>()</code> <br><code><a href="layer_masking.html">layer_masking</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_max_pooling_1d.html">layer_max_pooling_1d</a>()</code> <br><code><a href="layer_max_pooling_2d.html">layer_max_pooling_2d</a>()</code> <br><code><a href="layer_max_pooling_3d.html">layer_max_pooling_3d</a>()</code> <br><code><a href="layer_maximum.html">layer_maximum</a>()</code> <br><code><a href="layer_mel_spectrogram.html">layer_mel_spectrogram</a>()</code> <br><code><a href="layer_minimum.html">layer_minimum</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_multi_head_attention.html">layer_multi_head_attention</a>()</code> <br><code><a href="layer_multiply.html">layer_multiply</a>()</code> <br><code><a href="layer_normalization.html">layer_normalization</a>()</code> <br><code><a href="layer_permute.html">layer_permute</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_brightness.html">layer_random_brightness</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_contrast.html">layer_random_contrast</a>()</code> <br><code><a href="layer_random_crop.html">layer_random_crop</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_flip.html">layer_random_flip</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_rotation.html">layer_random_rotation</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_random_translation.html">layer_random_translation</a>()</code> <br><code><a href="layer_random_zoom.html">layer_random_zoom</a>()</code> <br><code><a href="layer_repeat_vector.html">layer_repeat_vector</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_reshape.html">layer_reshape</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_rms_normalization.html">layer_rms_normalization</a>()</code> <br><code><a href="layer_rnn.html">layer_rnn</a>()</code> <br><code><a href="layer_separable_conv_1d.html">layer_separable_conv_1d</a>()</code> <br><code><a href="layer_separable_conv_2d.html">layer_separable_conv_2d</a>()</code> <br><code><a href="layer_simple_rnn.html">layer_simple_rnn</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br><code><a href="layer_spatial_dropout_1d.html">layer_spatial_dropout_1d</a>()</code> <br><code><a href="layer_spatial_dropout_2d.html">layer_spatial_dropout_2d</a>()</code> <br><code><a href="layer_spatial_dropout_3d.html">layer_spatial_dropout_3d</a>()</code> <br><code><a href="layer_spectral_normalization.html">layer_spectral_normalization</a>()</code> <br><code><a href="layer_stft_spectrogram.html">layer_stft_spectrogram</a>()</code> <br><code><a href="layer_string_lookup.html">layer_string_lookup</a>()</code> <br><code><a href="layer_subtract.html">layer_subtract</a>()</code> <br><code><a href="layer_text_vectorization.html">layer_text_vectorization</a>()</code> <br><code><a href="layer_tfsm.html">layer_tfsm</a>()</code> <br><code><a href="layer_time_distributed.html">layer_time_distributed</a>()</code> <br><code><a href="layer_torch_module_wrapper.html">layer_torch_module_wrapper</a>()</code> <br><code><a href="layer_unit_normalization.html">layer_unit_normalization</a>()</code> <br><code><a href="layer_upsampling_1d.html">layer_upsampling_1d</a>()</code> <br><code><a href="layer_upsampling_2d.html">layer_upsampling_2d</a>()</code> <br><code><a href="layer_upsampling_3d.html">layer_upsampling_3d</a>()</code> <br><code><a href="layer_zero_padding_1d.html">layer_zero_padding_1d</a>()</code> <br><code><a href="layer_zero_padding_2d.html">layer_zero_padding_2d</a>()</code> <br><code><a href="layer_zero_padding_3d.html">layer_zero_padding_3d</a>()</code> <br><code><a href="rnn_cell_gru.html">rnn_cell_gru</a>()</code> <br><code><a href="rnn_cell_simple.html">rnn_cell_simple</a>()</code> <br><code><a href="rnn_cells_stack.html">rnn_cells_stack</a>()</code> <br></p></div>
+<p>Other layers: <br><code><a href="Layer.html">Layer</a>()</code> <br><code><a href="layer_activation.html">layer_activation</a>()</code> <br><code><a href="layer_activation_elu.html">layer_activation_elu</a>()</code> <br><code><a href="layer_activation_leaky_relu.html">layer_activation_leaky_relu</a>()</code> <br><code><a href="layer_activation_parametric_relu.html">layer_activation_parametric_relu</a>()</code> <br><code><a href="layer_activation_relu.html">layer_activation_relu</a>()</code> <br><code><a href="layer_activation_softmax.html">layer_activation_softmax</a>()</code> <br><code><a href="layer_activity_regularization.html">layer_activity_regularization</a>()</code> <br><code><a href="layer_add.html">layer_add</a>()</code> <br><code><a href="layer_additive_attention.html">layer_additive_attention</a>()</code> <br><code><a href="layer_alpha_dropout.html">layer_alpha_dropout</a>()</code> <br><code><a href="layer_attention.html">layer_attention</a>()</code> <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_average.html">layer_average</a>()</code> <br><code><a href="layer_average_pooling_1d.html">layer_average_pooling_1d</a>()</code> <br><code><a href="layer_average_pooling_2d.html">layer_average_pooling_2d</a>()</code> <br><code><a href="layer_average_pooling_3d.html">layer_average_pooling_3d</a>()</code> <br><code><a href="layer_batch_normalization.html">layer_batch_normalization</a>()</code> <br><code><a href="layer_bidirectional.html">layer_bidirectional</a>()</code> <br><code><a href="layer_category_encoding.html">layer_category_encoding</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_concatenate.html">layer_concatenate</a>()</code> <br><code><a href="layer_conv_1d.html">layer_conv_1d</a>()</code> <br><code><a href="layer_conv_1d_transpose.html">layer_conv_1d_transpose</a>()</code> <br><code><a href="layer_conv_2d.html">layer_conv_2d</a>()</code> <br><code><a href="layer_conv_2d_transpose.html">layer_conv_2d_transpose</a>()</code> <br><code><a href="layer_conv_3d.html">layer_conv_3d</a>()</code> <br><code><a href="layer_conv_3d_transpose.html">layer_conv_3d_transpose</a>()</code> <br><code><a href="layer_conv_lstm_1d.html">layer_conv_lstm_1d</a>()</code> <br><code><a href="layer_conv_lstm_2d.html">layer_conv_lstm_2d</a>()</code> <br><code><a href="layer_conv_lstm_3d.html">layer_conv_lstm_3d</a>()</code> <br><code><a href="layer_cropping_1d.html">layer_cropping_1d</a>()</code> <br><code><a href="layer_cropping_2d.html">layer_cropping_2d</a>()</code> <br><code><a href="layer_cropping_3d.html">layer_cropping_3d</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_dense.html">layer_dense</a>()</code> <br><code><a href="layer_depthwise_conv_1d.html">layer_depthwise_conv_1d</a>()</code> <br><code><a href="layer_depthwise_conv_2d.html">layer_depthwise_conv_2d</a>()</code> <br><code><a href="layer_discretization.html">layer_discretization</a>()</code> <br><code><a href="layer_dot.html">layer_dot</a>()</code> <br><code><a href="layer_dropout.html">layer_dropout</a>()</code> <br><code><a href="layer_einsum_dense.html">layer_einsum_dense</a>()</code> <br><code><a href="layer_embedding.html">layer_embedding</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_feature_space.html">layer_feature_space</a>()</code> <br><code><a href="layer_flatten.html">layer_flatten</a>()</code> <br><code><a href="layer_flax_module_wrapper.html">layer_flax_module_wrapper</a>()</code> <br><code><a href="layer_gaussian_dropout.html">layer_gaussian_dropout</a>()</code> <br><code><a href="layer_gaussian_noise.html">layer_gaussian_noise</a>()</code> <br><code><a href="layer_global_average_pooling_1d.html">layer_global_average_pooling_1d</a>()</code> <br><code><a href="layer_global_average_pooling_2d.html">layer_global_average_pooling_2d</a>()</code> <br><code><a href="layer_global_average_pooling_3d.html">layer_global_average_pooling_3d</a>()</code> <br><code><a href="layer_global_max_pooling_1d.html">layer_global_max_pooling_1d</a>()</code> <br><code><a href="layer_global_max_pooling_2d.html">layer_global_max_pooling_2d</a>()</code> <br><code><a href="layer_global_max_pooling_3d.html">layer_global_max_pooling_3d</a>()</code> <br><code><a href="layer_group_normalization.html">layer_group_normalization</a>()</code> <br><code><a href="layer_group_query_attention.html">layer_group_query_attention</a>()</code> <br><code><a href="layer_gru.html">layer_gru</a>()</code> <br><code><a href="layer_hashed_crossing.html">layer_hashed_crossing</a>()</code> <br><code><a href="layer_hashing.html">layer_hashing</a>()</code> <br><code><a href="layer_identity.html">layer_identity</a>()</code> <br><code><a href="layer_integer_lookup.html">layer_integer_lookup</a>()</code> <br><code><a href="layer_jax_model_wrapper.html">layer_jax_model_wrapper</a>()</code> <br><code><a href="layer_lambda.html">layer_lambda</a>()</code> <br><code><a href="layer_layer_normalization.html">layer_layer_normalization</a>()</code> <br><code><a href="layer_lstm.html">layer_lstm</a>()</code> <br><code><a href="layer_masking.html">layer_masking</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_max_pooling_1d.html">layer_max_pooling_1d</a>()</code> <br><code><a href="layer_max_pooling_2d.html">layer_max_pooling_2d</a>()</code> <br><code><a href="layer_max_pooling_3d.html">layer_max_pooling_3d</a>()</code> <br><code><a href="layer_maximum.html">layer_maximum</a>()</code> <br><code><a href="layer_mel_spectrogram.html">layer_mel_spectrogram</a>()</code> <br><code><a href="layer_minimum.html">layer_minimum</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_multi_head_attention.html">layer_multi_head_attention</a>()</code> <br><code><a href="layer_multiply.html">layer_multiply</a>()</code> <br><code><a href="layer_normalization.html">layer_normalization</a>()</code> <br><code><a href="layer_permute.html">layer_permute</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_brightness.html">layer_random_brightness</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_contrast.html">layer_random_contrast</a>()</code> <br><code><a href="layer_random_crop.html">layer_random_crop</a>()</code> <br><code><a href="layer_random_elastic_transform.html">layer_random_elastic_transform</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_flip.html">layer_random_flip</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_rotation.html">layer_random_rotation</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_random_translation.html">layer_random_translation</a>()</code> <br><code><a href="layer_random_zoom.html">layer_random_zoom</a>()</code> <br><code><a href="layer_repeat_vector.html">layer_repeat_vector</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_reshape.html">layer_reshape</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_rms_normalization.html">layer_rms_normalization</a>()</code> <br><code><a href="layer_rnn.html">layer_rnn</a>()</code> <br><code><a href="layer_separable_conv_1d.html">layer_separable_conv_1d</a>()</code> <br><code><a href="layer_separable_conv_2d.html">layer_separable_conv_2d</a>()</code> <br><code><a href="layer_simple_rnn.html">layer_simple_rnn</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br><code><a href="layer_spatial_dropout_1d.html">layer_spatial_dropout_1d</a>()</code> <br><code><a href="layer_spatial_dropout_2d.html">layer_spatial_dropout_2d</a>()</code> <br><code><a href="layer_spatial_dropout_3d.html">layer_spatial_dropout_3d</a>()</code> <br><code><a href="layer_spectral_normalization.html">layer_spectral_normalization</a>()</code> <br><code><a href="layer_stft_spectrogram.html">layer_stft_spectrogram</a>()</code> <br><code><a href="layer_string_lookup.html">layer_string_lookup</a>()</code> <br><code><a href="layer_subtract.html">layer_subtract</a>()</code> <br><code><a href="layer_text_vectorization.html">layer_text_vectorization</a>()</code> <br><code><a href="layer_tfsm.html">layer_tfsm</a>()</code> <br><code><a href="layer_time_distributed.html">layer_time_distributed</a>()</code> <br><code><a href="layer_torch_module_wrapper.html">layer_torch_module_wrapper</a>()</code> <br><code><a href="layer_unit_normalization.html">layer_unit_normalization</a>()</code> <br><code><a href="layer_upsampling_1d.html">layer_upsampling_1d</a>()</code> <br><code><a href="layer_upsampling_2d.html">layer_upsampling_2d</a>()</code> <br><code><a href="layer_upsampling_3d.html">layer_upsampling_3d</a>()</code> <br><code><a href="layer_zero_padding_1d.html">layer_zero_padding_1d</a>()</code> <br><code><a href="layer_zero_padding_2d.html">layer_zero_padding_2d</a>()</code> <br><code><a href="layer_zero_padding_3d.html">layer_zero_padding_3d</a>()</code> <br><code><a href="rnn_cell_gru.html">rnn_cell_gru</a>()</code> <br><code><a href="rnn_cell_simple.html">rnn_cell_simple</a>()</code> <br><code><a href="rnn_cells_stack.html">rnn_cells_stack</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -232,7 +232,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/rnn_cell_simple.html b/docs/dev/reference/rnn_cell_simple.html
index 7ee4b51f2..f0e71b309 100644
--- a/docs/dev/reference/rnn_cell_simple.html
+++ b/docs/dev/reference/rnn_cell_simple.html
@@ -10,7 +10,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -199,7 +199,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
     <div class="dont-index"><p>Other rnn cells: <br><code><a href="layer_rnn.html">layer_rnn</a>()</code> <br><code><a href="rnn_cell_gru.html">rnn_cell_gru</a>()</code> <br><code><a href="rnn_cell_lstm.html">rnn_cell_lstm</a>()</code> <br></p>
 <p>Other simple rnn layers: <br><code><a href="layer_simple_rnn.html">layer_simple_rnn</a>()</code> <br></p>
 <p>Other rnn layers: <br><code><a href="layer_bidirectional.html">layer_bidirectional</a>()</code> <br><code><a href="layer_conv_lstm_1d.html">layer_conv_lstm_1d</a>()</code> <br><code><a href="layer_conv_lstm_2d.html">layer_conv_lstm_2d</a>()</code> <br><code><a href="layer_conv_lstm_3d.html">layer_conv_lstm_3d</a>()</code> <br><code><a href="layer_gru.html">layer_gru</a>()</code> <br><code><a href="layer_lstm.html">layer_lstm</a>()</code> <br><code><a href="layer_rnn.html">layer_rnn</a>()</code> <br><code><a href="layer_simple_rnn.html">layer_simple_rnn</a>()</code> <br><code><a href="layer_time_distributed.html">layer_time_distributed</a>()</code> <br><code><a href="rnn_cell_gru.html">rnn_cell_gru</a>()</code> <br><code><a href="rnn_cell_lstm.html">rnn_cell_lstm</a>()</code> <br><code><a href="rnn_cells_stack.html">rnn_cells_stack</a>()</code> <br></p>
-<p>Other layers: <br><code><a href="Layer.html">Layer</a>()</code> <br><code><a href="layer_activation.html">layer_activation</a>()</code> <br><code><a href="layer_activation_elu.html">layer_activation_elu</a>()</code> <br><code><a href="layer_activation_leaky_relu.html">layer_activation_leaky_relu</a>()</code> <br><code><a href="layer_activation_parametric_relu.html">layer_activation_parametric_relu</a>()</code> <br><code><a href="layer_activation_relu.html">layer_activation_relu</a>()</code> <br><code><a href="layer_activation_softmax.html">layer_activation_softmax</a>()</code> <br><code><a href="layer_activity_regularization.html">layer_activity_regularization</a>()</code> <br><code><a href="layer_add.html">layer_add</a>()</code> <br><code><a href="layer_additive_attention.html">layer_additive_attention</a>()</code> <br><code><a href="layer_alpha_dropout.html">layer_alpha_dropout</a>()</code> <br><code><a href="layer_attention.html">layer_attention</a>()</code> <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_average.html">layer_average</a>()</code> <br><code><a href="layer_average_pooling_1d.html">layer_average_pooling_1d</a>()</code> <br><code><a href="layer_average_pooling_2d.html">layer_average_pooling_2d</a>()</code> <br><code><a href="layer_average_pooling_3d.html">layer_average_pooling_3d</a>()</code> <br><code><a href="layer_batch_normalization.html">layer_batch_normalization</a>()</code> <br><code><a href="layer_bidirectional.html">layer_bidirectional</a>()</code> <br><code><a href="layer_category_encoding.html">layer_category_encoding</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_concatenate.html">layer_concatenate</a>()</code> <br><code><a href="layer_conv_1d.html">layer_conv_1d</a>()</code> <br><code><a href="layer_conv_1d_transpose.html">layer_conv_1d_transpose</a>()</code> <br><code><a href="layer_conv_2d.html">layer_conv_2d</a>()</code> <br><code><a href="layer_conv_2d_transpose.html">layer_conv_2d_transpose</a>()</code> <br><code><a href="layer_conv_3d.html">layer_conv_3d</a>()</code> <br><code><a href="layer_conv_3d_transpose.html">layer_conv_3d_transpose</a>()</code> <br><code><a href="layer_conv_lstm_1d.html">layer_conv_lstm_1d</a>()</code> <br><code><a href="layer_conv_lstm_2d.html">layer_conv_lstm_2d</a>()</code> <br><code><a href="layer_conv_lstm_3d.html">layer_conv_lstm_3d</a>()</code> <br><code><a href="layer_cropping_1d.html">layer_cropping_1d</a>()</code> <br><code><a href="layer_cropping_2d.html">layer_cropping_2d</a>()</code> <br><code><a href="layer_cropping_3d.html">layer_cropping_3d</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_dense.html">layer_dense</a>()</code> <br><code><a href="layer_depthwise_conv_1d.html">layer_depthwise_conv_1d</a>()</code> <br><code><a href="layer_depthwise_conv_2d.html">layer_depthwise_conv_2d</a>()</code> <br><code><a href="layer_discretization.html">layer_discretization</a>()</code> <br><code><a href="layer_dot.html">layer_dot</a>()</code> <br><code><a href="layer_dropout.html">layer_dropout</a>()</code> <br><code><a href="layer_einsum_dense.html">layer_einsum_dense</a>()</code> <br><code><a href="layer_embedding.html">layer_embedding</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_feature_space.html">layer_feature_space</a>()</code> <br><code><a href="layer_flatten.html">layer_flatten</a>()</code> <br><code><a href="layer_flax_module_wrapper.html">layer_flax_module_wrapper</a>()</code> <br><code><a href="layer_gaussian_dropout.html">layer_gaussian_dropout</a>()</code> <br><code><a href="layer_gaussian_noise.html">layer_gaussian_noise</a>()</code> <br><code><a href="layer_global_average_pooling_1d.html">layer_global_average_pooling_1d</a>()</code> <br><code><a href="layer_global_average_pooling_2d.html">layer_global_average_pooling_2d</a>()</code> <br><code><a href="layer_global_average_pooling_3d.html">layer_global_average_pooling_3d</a>()</code> <br><code><a href="layer_global_max_pooling_1d.html">layer_global_max_pooling_1d</a>()</code> <br><code><a href="layer_global_max_pooling_2d.html">layer_global_max_pooling_2d</a>()</code> <br><code><a href="layer_global_max_pooling_3d.html">layer_global_max_pooling_3d</a>()</code> <br><code><a href="layer_group_normalization.html">layer_group_normalization</a>()</code> <br><code><a href="layer_group_query_attention.html">layer_group_query_attention</a>()</code> <br><code><a href="layer_gru.html">layer_gru</a>()</code> <br><code><a href="layer_hashed_crossing.html">layer_hashed_crossing</a>()</code> <br><code><a href="layer_hashing.html">layer_hashing</a>()</code> <br><code><a href="layer_identity.html">layer_identity</a>()</code> <br><code><a href="layer_integer_lookup.html">layer_integer_lookup</a>()</code> <br><code><a href="layer_jax_model_wrapper.html">layer_jax_model_wrapper</a>()</code> <br><code><a href="layer_lambda.html">layer_lambda</a>()</code> <br><code><a href="layer_layer_normalization.html">layer_layer_normalization</a>()</code> <br><code><a href="layer_lstm.html">layer_lstm</a>()</code> <br><code><a href="layer_masking.html">layer_masking</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_max_pooling_1d.html">layer_max_pooling_1d</a>()</code> <br><code><a href="layer_max_pooling_2d.html">layer_max_pooling_2d</a>()</code> <br><code><a href="layer_max_pooling_3d.html">layer_max_pooling_3d</a>()</code> <br><code><a href="layer_maximum.html">layer_maximum</a>()</code> <br><code><a href="layer_mel_spectrogram.html">layer_mel_spectrogram</a>()</code> <br><code><a href="layer_minimum.html">layer_minimum</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_multi_head_attention.html">layer_multi_head_attention</a>()</code> <br><code><a href="layer_multiply.html">layer_multiply</a>()</code> <br><code><a href="layer_normalization.html">layer_normalization</a>()</code> <br><code><a href="layer_permute.html">layer_permute</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_brightness.html">layer_random_brightness</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_contrast.html">layer_random_contrast</a>()</code> <br><code><a href="layer_random_crop.html">layer_random_crop</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_flip.html">layer_random_flip</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_rotation.html">layer_random_rotation</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_random_translation.html">layer_random_translation</a>()</code> <br><code><a href="layer_random_zoom.html">layer_random_zoom</a>()</code> <br><code><a href="layer_repeat_vector.html">layer_repeat_vector</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_reshape.html">layer_reshape</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_rms_normalization.html">layer_rms_normalization</a>()</code> <br><code><a href="layer_rnn.html">layer_rnn</a>()</code> <br><code><a href="layer_separable_conv_1d.html">layer_separable_conv_1d</a>()</code> <br><code><a href="layer_separable_conv_2d.html">layer_separable_conv_2d</a>()</code> <br><code><a href="layer_simple_rnn.html">layer_simple_rnn</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br><code><a href="layer_spatial_dropout_1d.html">layer_spatial_dropout_1d</a>()</code> <br><code><a href="layer_spatial_dropout_2d.html">layer_spatial_dropout_2d</a>()</code> <br><code><a href="layer_spatial_dropout_3d.html">layer_spatial_dropout_3d</a>()</code> <br><code><a href="layer_spectral_normalization.html">layer_spectral_normalization</a>()</code> <br><code><a href="layer_stft_spectrogram.html">layer_stft_spectrogram</a>()</code> <br><code><a href="layer_string_lookup.html">layer_string_lookup</a>()</code> <br><code><a href="layer_subtract.html">layer_subtract</a>()</code> <br><code><a href="layer_text_vectorization.html">layer_text_vectorization</a>()</code> <br><code><a href="layer_tfsm.html">layer_tfsm</a>()</code> <br><code><a href="layer_time_distributed.html">layer_time_distributed</a>()</code> <br><code><a href="layer_torch_module_wrapper.html">layer_torch_module_wrapper</a>()</code> <br><code><a href="layer_unit_normalization.html">layer_unit_normalization</a>()</code> <br><code><a href="layer_upsampling_1d.html">layer_upsampling_1d</a>()</code> <br><code><a href="layer_upsampling_2d.html">layer_upsampling_2d</a>()</code> <br><code><a href="layer_upsampling_3d.html">layer_upsampling_3d</a>()</code> <br><code><a href="layer_zero_padding_1d.html">layer_zero_padding_1d</a>()</code> <br><code><a href="layer_zero_padding_2d.html">layer_zero_padding_2d</a>()</code> <br><code><a href="layer_zero_padding_3d.html">layer_zero_padding_3d</a>()</code> <br><code><a href="rnn_cell_gru.html">rnn_cell_gru</a>()</code> <br><code><a href="rnn_cell_lstm.html">rnn_cell_lstm</a>()</code> <br><code><a href="rnn_cells_stack.html">rnn_cells_stack</a>()</code> <br></p></div>
+<p>Other layers: <br><code><a href="Layer.html">Layer</a>()</code> <br><code><a href="layer_activation.html">layer_activation</a>()</code> <br><code><a href="layer_activation_elu.html">layer_activation_elu</a>()</code> <br><code><a href="layer_activation_leaky_relu.html">layer_activation_leaky_relu</a>()</code> <br><code><a href="layer_activation_parametric_relu.html">layer_activation_parametric_relu</a>()</code> <br><code><a href="layer_activation_relu.html">layer_activation_relu</a>()</code> <br><code><a href="layer_activation_softmax.html">layer_activation_softmax</a>()</code> <br><code><a href="layer_activity_regularization.html">layer_activity_regularization</a>()</code> <br><code><a href="layer_add.html">layer_add</a>()</code> <br><code><a href="layer_additive_attention.html">layer_additive_attention</a>()</code> <br><code><a href="layer_alpha_dropout.html">layer_alpha_dropout</a>()</code> <br><code><a href="layer_attention.html">layer_attention</a>()</code> <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_average.html">layer_average</a>()</code> <br><code><a href="layer_average_pooling_1d.html">layer_average_pooling_1d</a>()</code> <br><code><a href="layer_average_pooling_2d.html">layer_average_pooling_2d</a>()</code> <br><code><a href="layer_average_pooling_3d.html">layer_average_pooling_3d</a>()</code> <br><code><a href="layer_batch_normalization.html">layer_batch_normalization</a>()</code> <br><code><a href="layer_bidirectional.html">layer_bidirectional</a>()</code> <br><code><a href="layer_category_encoding.html">layer_category_encoding</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_concatenate.html">layer_concatenate</a>()</code> <br><code><a href="layer_conv_1d.html">layer_conv_1d</a>()</code> <br><code><a href="layer_conv_1d_transpose.html">layer_conv_1d_transpose</a>()</code> <br><code><a href="layer_conv_2d.html">layer_conv_2d</a>()</code> <br><code><a href="layer_conv_2d_transpose.html">layer_conv_2d_transpose</a>()</code> <br><code><a href="layer_conv_3d.html">layer_conv_3d</a>()</code> <br><code><a href="layer_conv_3d_transpose.html">layer_conv_3d_transpose</a>()</code> <br><code><a href="layer_conv_lstm_1d.html">layer_conv_lstm_1d</a>()</code> <br><code><a href="layer_conv_lstm_2d.html">layer_conv_lstm_2d</a>()</code> <br><code><a href="layer_conv_lstm_3d.html">layer_conv_lstm_3d</a>()</code> <br><code><a href="layer_cropping_1d.html">layer_cropping_1d</a>()</code> <br><code><a href="layer_cropping_2d.html">layer_cropping_2d</a>()</code> <br><code><a href="layer_cropping_3d.html">layer_cropping_3d</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_dense.html">layer_dense</a>()</code> <br><code><a href="layer_depthwise_conv_1d.html">layer_depthwise_conv_1d</a>()</code> <br><code><a href="layer_depthwise_conv_2d.html">layer_depthwise_conv_2d</a>()</code> <br><code><a href="layer_discretization.html">layer_discretization</a>()</code> <br><code><a href="layer_dot.html">layer_dot</a>()</code> <br><code><a href="layer_dropout.html">layer_dropout</a>()</code> <br><code><a href="layer_einsum_dense.html">layer_einsum_dense</a>()</code> <br><code><a href="layer_embedding.html">layer_embedding</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_feature_space.html">layer_feature_space</a>()</code> <br><code><a href="layer_flatten.html">layer_flatten</a>()</code> <br><code><a href="layer_flax_module_wrapper.html">layer_flax_module_wrapper</a>()</code> <br><code><a href="layer_gaussian_dropout.html">layer_gaussian_dropout</a>()</code> <br><code><a href="layer_gaussian_noise.html">layer_gaussian_noise</a>()</code> <br><code><a href="layer_global_average_pooling_1d.html">layer_global_average_pooling_1d</a>()</code> <br><code><a href="layer_global_average_pooling_2d.html">layer_global_average_pooling_2d</a>()</code> <br><code><a href="layer_global_average_pooling_3d.html">layer_global_average_pooling_3d</a>()</code> <br><code><a href="layer_global_max_pooling_1d.html">layer_global_max_pooling_1d</a>()</code> <br><code><a href="layer_global_max_pooling_2d.html">layer_global_max_pooling_2d</a>()</code> <br><code><a href="layer_global_max_pooling_3d.html">layer_global_max_pooling_3d</a>()</code> <br><code><a href="layer_group_normalization.html">layer_group_normalization</a>()</code> <br><code><a href="layer_group_query_attention.html">layer_group_query_attention</a>()</code> <br><code><a href="layer_gru.html">layer_gru</a>()</code> <br><code><a href="layer_hashed_crossing.html">layer_hashed_crossing</a>()</code> <br><code><a href="layer_hashing.html">layer_hashing</a>()</code> <br><code><a href="layer_identity.html">layer_identity</a>()</code> <br><code><a href="layer_integer_lookup.html">layer_integer_lookup</a>()</code> <br><code><a href="layer_jax_model_wrapper.html">layer_jax_model_wrapper</a>()</code> <br><code><a href="layer_lambda.html">layer_lambda</a>()</code> <br><code><a href="layer_layer_normalization.html">layer_layer_normalization</a>()</code> <br><code><a href="layer_lstm.html">layer_lstm</a>()</code> <br><code><a href="layer_masking.html">layer_masking</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_max_pooling_1d.html">layer_max_pooling_1d</a>()</code> <br><code><a href="layer_max_pooling_2d.html">layer_max_pooling_2d</a>()</code> <br><code><a href="layer_max_pooling_3d.html">layer_max_pooling_3d</a>()</code> <br><code><a href="layer_maximum.html">layer_maximum</a>()</code> <br><code><a href="layer_mel_spectrogram.html">layer_mel_spectrogram</a>()</code> <br><code><a href="layer_minimum.html">layer_minimum</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_multi_head_attention.html">layer_multi_head_attention</a>()</code> <br><code><a href="layer_multiply.html">layer_multiply</a>()</code> <br><code><a href="layer_normalization.html">layer_normalization</a>()</code> <br><code><a href="layer_permute.html">layer_permute</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_brightness.html">layer_random_brightness</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_contrast.html">layer_random_contrast</a>()</code> <br><code><a href="layer_random_crop.html">layer_random_crop</a>()</code> <br><code><a href="layer_random_elastic_transform.html">layer_random_elastic_transform</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_flip.html">layer_random_flip</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_rotation.html">layer_random_rotation</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_random_translation.html">layer_random_translation</a>()</code> <br><code><a href="layer_random_zoom.html">layer_random_zoom</a>()</code> <br><code><a href="layer_repeat_vector.html">layer_repeat_vector</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_reshape.html">layer_reshape</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_rms_normalization.html">layer_rms_normalization</a>()</code> <br><code><a href="layer_rnn.html">layer_rnn</a>()</code> <br><code><a href="layer_separable_conv_1d.html">layer_separable_conv_1d</a>()</code> <br><code><a href="layer_separable_conv_2d.html">layer_separable_conv_2d</a>()</code> <br><code><a href="layer_simple_rnn.html">layer_simple_rnn</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br><code><a href="layer_spatial_dropout_1d.html">layer_spatial_dropout_1d</a>()</code> <br><code><a href="layer_spatial_dropout_2d.html">layer_spatial_dropout_2d</a>()</code> <br><code><a href="layer_spatial_dropout_3d.html">layer_spatial_dropout_3d</a>()</code> <br><code><a href="layer_spectral_normalization.html">layer_spectral_normalization</a>()</code> <br><code><a href="layer_stft_spectrogram.html">layer_stft_spectrogram</a>()</code> <br><code><a href="layer_string_lookup.html">layer_string_lookup</a>()</code> <br><code><a href="layer_subtract.html">layer_subtract</a>()</code> <br><code><a href="layer_text_vectorization.html">layer_text_vectorization</a>()</code> <br><code><a href="layer_tfsm.html">layer_tfsm</a>()</code> <br><code><a href="layer_time_distributed.html">layer_time_distributed</a>()</code> <br><code><a href="layer_torch_module_wrapper.html">layer_torch_module_wrapper</a>()</code> <br><code><a href="layer_unit_normalization.html">layer_unit_normalization</a>()</code> <br><code><a href="layer_upsampling_1d.html">layer_upsampling_1d</a>()</code> <br><code><a href="layer_upsampling_2d.html">layer_upsampling_2d</a>()</code> <br><code><a href="layer_upsampling_3d.html">layer_upsampling_3d</a>()</code> <br><code><a href="layer_zero_padding_1d.html">layer_zero_padding_1d</a>()</code> <br><code><a href="layer_zero_padding_2d.html">layer_zero_padding_2d</a>()</code> <br><code><a href="layer_zero_padding_3d.html">layer_zero_padding_3d</a>()</code> <br><code><a href="rnn_cell_gru.html">rnn_cell_gru</a>()</code> <br><code><a href="rnn_cell_lstm.html">rnn_cell_lstm</a>()</code> <br><code><a href="rnn_cells_stack.html">rnn_cells_stack</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -211,7 +211,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/rnn_cells_stack.html b/docs/dev/reference/rnn_cells_stack.html
index 4c95d6210..53b1045bd 100644
--- a/docs/dev/reference/rnn_cells_stack.html
+++ b/docs/dev/reference/rnn_cells_stack.html
@@ -8,7 +8,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -99,7 +99,7 @@ <h2 id="example">Example<a class="anchor" aria-label="anchor" href="#example"></
     <div class="section level2">
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
     <div class="dont-index"><p>Other rnn layers: <br><code><a href="layer_bidirectional.html">layer_bidirectional</a>()</code> <br><code><a href="layer_conv_lstm_1d.html">layer_conv_lstm_1d</a>()</code> <br><code><a href="layer_conv_lstm_2d.html">layer_conv_lstm_2d</a>()</code> <br><code><a href="layer_conv_lstm_3d.html">layer_conv_lstm_3d</a>()</code> <br><code><a href="layer_gru.html">layer_gru</a>()</code> <br><code><a href="layer_lstm.html">layer_lstm</a>()</code> <br><code><a href="layer_rnn.html">layer_rnn</a>()</code> <br><code><a href="layer_simple_rnn.html">layer_simple_rnn</a>()</code> <br><code><a href="layer_time_distributed.html">layer_time_distributed</a>()</code> <br><code><a href="rnn_cell_gru.html">rnn_cell_gru</a>()</code> <br><code><a href="rnn_cell_lstm.html">rnn_cell_lstm</a>()</code> <br><code><a href="rnn_cell_simple.html">rnn_cell_simple</a>()</code> <br></p>
-<p>Other layers: <br><code><a href="Layer.html">Layer</a>()</code> <br><code><a href="layer_activation.html">layer_activation</a>()</code> <br><code><a href="layer_activation_elu.html">layer_activation_elu</a>()</code> <br><code><a href="layer_activation_leaky_relu.html">layer_activation_leaky_relu</a>()</code> <br><code><a href="layer_activation_parametric_relu.html">layer_activation_parametric_relu</a>()</code> <br><code><a href="layer_activation_relu.html">layer_activation_relu</a>()</code> <br><code><a href="layer_activation_softmax.html">layer_activation_softmax</a>()</code> <br><code><a href="layer_activity_regularization.html">layer_activity_regularization</a>()</code> <br><code><a href="layer_add.html">layer_add</a>()</code> <br><code><a href="layer_additive_attention.html">layer_additive_attention</a>()</code> <br><code><a href="layer_alpha_dropout.html">layer_alpha_dropout</a>()</code> <br><code><a href="layer_attention.html">layer_attention</a>()</code> <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_average.html">layer_average</a>()</code> <br><code><a href="layer_average_pooling_1d.html">layer_average_pooling_1d</a>()</code> <br><code><a href="layer_average_pooling_2d.html">layer_average_pooling_2d</a>()</code> <br><code><a href="layer_average_pooling_3d.html">layer_average_pooling_3d</a>()</code> <br><code><a href="layer_batch_normalization.html">layer_batch_normalization</a>()</code> <br><code><a href="layer_bidirectional.html">layer_bidirectional</a>()</code> <br><code><a href="layer_category_encoding.html">layer_category_encoding</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_concatenate.html">layer_concatenate</a>()</code> <br><code><a href="layer_conv_1d.html">layer_conv_1d</a>()</code> <br><code><a href="layer_conv_1d_transpose.html">layer_conv_1d_transpose</a>()</code> <br><code><a href="layer_conv_2d.html">layer_conv_2d</a>()</code> <br><code><a href="layer_conv_2d_transpose.html">layer_conv_2d_transpose</a>()</code> <br><code><a href="layer_conv_3d.html">layer_conv_3d</a>()</code> <br><code><a href="layer_conv_3d_transpose.html">layer_conv_3d_transpose</a>()</code> <br><code><a href="layer_conv_lstm_1d.html">layer_conv_lstm_1d</a>()</code> <br><code><a href="layer_conv_lstm_2d.html">layer_conv_lstm_2d</a>()</code> <br><code><a href="layer_conv_lstm_3d.html">layer_conv_lstm_3d</a>()</code> <br><code><a href="layer_cropping_1d.html">layer_cropping_1d</a>()</code> <br><code><a href="layer_cropping_2d.html">layer_cropping_2d</a>()</code> <br><code><a href="layer_cropping_3d.html">layer_cropping_3d</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_dense.html">layer_dense</a>()</code> <br><code><a href="layer_depthwise_conv_1d.html">layer_depthwise_conv_1d</a>()</code> <br><code><a href="layer_depthwise_conv_2d.html">layer_depthwise_conv_2d</a>()</code> <br><code><a href="layer_discretization.html">layer_discretization</a>()</code> <br><code><a href="layer_dot.html">layer_dot</a>()</code> <br><code><a href="layer_dropout.html">layer_dropout</a>()</code> <br><code><a href="layer_einsum_dense.html">layer_einsum_dense</a>()</code> <br><code><a href="layer_embedding.html">layer_embedding</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_feature_space.html">layer_feature_space</a>()</code> <br><code><a href="layer_flatten.html">layer_flatten</a>()</code> <br><code><a href="layer_flax_module_wrapper.html">layer_flax_module_wrapper</a>()</code> <br><code><a href="layer_gaussian_dropout.html">layer_gaussian_dropout</a>()</code> <br><code><a href="layer_gaussian_noise.html">layer_gaussian_noise</a>()</code> <br><code><a href="layer_global_average_pooling_1d.html">layer_global_average_pooling_1d</a>()</code> <br><code><a href="layer_global_average_pooling_2d.html">layer_global_average_pooling_2d</a>()</code> <br><code><a href="layer_global_average_pooling_3d.html">layer_global_average_pooling_3d</a>()</code> <br><code><a href="layer_global_max_pooling_1d.html">layer_global_max_pooling_1d</a>()</code> <br><code><a href="layer_global_max_pooling_2d.html">layer_global_max_pooling_2d</a>()</code> <br><code><a href="layer_global_max_pooling_3d.html">layer_global_max_pooling_3d</a>()</code> <br><code><a href="layer_group_normalization.html">layer_group_normalization</a>()</code> <br><code><a href="layer_group_query_attention.html">layer_group_query_attention</a>()</code> <br><code><a href="layer_gru.html">layer_gru</a>()</code> <br><code><a href="layer_hashed_crossing.html">layer_hashed_crossing</a>()</code> <br><code><a href="layer_hashing.html">layer_hashing</a>()</code> <br><code><a href="layer_identity.html">layer_identity</a>()</code> <br><code><a href="layer_integer_lookup.html">layer_integer_lookup</a>()</code> <br><code><a href="layer_jax_model_wrapper.html">layer_jax_model_wrapper</a>()</code> <br><code><a href="layer_lambda.html">layer_lambda</a>()</code> <br><code><a href="layer_layer_normalization.html">layer_layer_normalization</a>()</code> <br><code><a href="layer_lstm.html">layer_lstm</a>()</code> <br><code><a href="layer_masking.html">layer_masking</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_max_pooling_1d.html">layer_max_pooling_1d</a>()</code> <br><code><a href="layer_max_pooling_2d.html">layer_max_pooling_2d</a>()</code> <br><code><a href="layer_max_pooling_3d.html">layer_max_pooling_3d</a>()</code> <br><code><a href="layer_maximum.html">layer_maximum</a>()</code> <br><code><a href="layer_mel_spectrogram.html">layer_mel_spectrogram</a>()</code> <br><code><a href="layer_minimum.html">layer_minimum</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_multi_head_attention.html">layer_multi_head_attention</a>()</code> <br><code><a href="layer_multiply.html">layer_multiply</a>()</code> <br><code><a href="layer_normalization.html">layer_normalization</a>()</code> <br><code><a href="layer_permute.html">layer_permute</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_brightness.html">layer_random_brightness</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_contrast.html">layer_random_contrast</a>()</code> <br><code><a href="layer_random_crop.html">layer_random_crop</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_flip.html">layer_random_flip</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_rotation.html">layer_random_rotation</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_random_translation.html">layer_random_translation</a>()</code> <br><code><a href="layer_random_zoom.html">layer_random_zoom</a>()</code> <br><code><a href="layer_repeat_vector.html">layer_repeat_vector</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_reshape.html">layer_reshape</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_rms_normalization.html">layer_rms_normalization</a>()</code> <br><code><a href="layer_rnn.html">layer_rnn</a>()</code> <br><code><a href="layer_separable_conv_1d.html">layer_separable_conv_1d</a>()</code> <br><code><a href="layer_separable_conv_2d.html">layer_separable_conv_2d</a>()</code> <br><code><a href="layer_simple_rnn.html">layer_simple_rnn</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br><code><a href="layer_spatial_dropout_1d.html">layer_spatial_dropout_1d</a>()</code> <br><code><a href="layer_spatial_dropout_2d.html">layer_spatial_dropout_2d</a>()</code> <br><code><a href="layer_spatial_dropout_3d.html">layer_spatial_dropout_3d</a>()</code> <br><code><a href="layer_spectral_normalization.html">layer_spectral_normalization</a>()</code> <br><code><a href="layer_stft_spectrogram.html">layer_stft_spectrogram</a>()</code> <br><code><a href="layer_string_lookup.html">layer_string_lookup</a>()</code> <br><code><a href="layer_subtract.html">layer_subtract</a>()</code> <br><code><a href="layer_text_vectorization.html">layer_text_vectorization</a>()</code> <br><code><a href="layer_tfsm.html">layer_tfsm</a>()</code> <br><code><a href="layer_time_distributed.html">layer_time_distributed</a>()</code> <br><code><a href="layer_torch_module_wrapper.html">layer_torch_module_wrapper</a>()</code> <br><code><a href="layer_unit_normalization.html">layer_unit_normalization</a>()</code> <br><code><a href="layer_upsampling_1d.html">layer_upsampling_1d</a>()</code> <br><code><a href="layer_upsampling_2d.html">layer_upsampling_2d</a>()</code> <br><code><a href="layer_upsampling_3d.html">layer_upsampling_3d</a>()</code> <br><code><a href="layer_zero_padding_1d.html">layer_zero_padding_1d</a>()</code> <br><code><a href="layer_zero_padding_2d.html">layer_zero_padding_2d</a>()</code> <br><code><a href="layer_zero_padding_3d.html">layer_zero_padding_3d</a>()</code> <br><code><a href="rnn_cell_gru.html">rnn_cell_gru</a>()</code> <br><code><a href="rnn_cell_lstm.html">rnn_cell_lstm</a>()</code> <br><code><a href="rnn_cell_simple.html">rnn_cell_simple</a>()</code> <br></p></div>
+<p>Other layers: <br><code><a href="Layer.html">Layer</a>()</code> <br><code><a href="layer_activation.html">layer_activation</a>()</code> <br><code><a href="layer_activation_elu.html">layer_activation_elu</a>()</code> <br><code><a href="layer_activation_leaky_relu.html">layer_activation_leaky_relu</a>()</code> <br><code><a href="layer_activation_parametric_relu.html">layer_activation_parametric_relu</a>()</code> <br><code><a href="layer_activation_relu.html">layer_activation_relu</a>()</code> <br><code><a href="layer_activation_softmax.html">layer_activation_softmax</a>()</code> <br><code><a href="layer_activity_regularization.html">layer_activity_regularization</a>()</code> <br><code><a href="layer_add.html">layer_add</a>()</code> <br><code><a href="layer_additive_attention.html">layer_additive_attention</a>()</code> <br><code><a href="layer_alpha_dropout.html">layer_alpha_dropout</a>()</code> <br><code><a href="layer_attention.html">layer_attention</a>()</code> <br><code><a href="layer_aug_mix.html">layer_aug_mix</a>()</code> <br><code><a href="layer_auto_contrast.html">layer_auto_contrast</a>()</code> <br><code><a href="layer_average.html">layer_average</a>()</code> <br><code><a href="layer_average_pooling_1d.html">layer_average_pooling_1d</a>()</code> <br><code><a href="layer_average_pooling_2d.html">layer_average_pooling_2d</a>()</code> <br><code><a href="layer_average_pooling_3d.html">layer_average_pooling_3d</a>()</code> <br><code><a href="layer_batch_normalization.html">layer_batch_normalization</a>()</code> <br><code><a href="layer_bidirectional.html">layer_bidirectional</a>()</code> <br><code><a href="layer_category_encoding.html">layer_category_encoding</a>()</code> <br><code><a href="layer_center_crop.html">layer_center_crop</a>()</code> <br><code><a href="layer_concatenate.html">layer_concatenate</a>()</code> <br><code><a href="layer_conv_1d.html">layer_conv_1d</a>()</code> <br><code><a href="layer_conv_1d_transpose.html">layer_conv_1d_transpose</a>()</code> <br><code><a href="layer_conv_2d.html">layer_conv_2d</a>()</code> <br><code><a href="layer_conv_2d_transpose.html">layer_conv_2d_transpose</a>()</code> <br><code><a href="layer_conv_3d.html">layer_conv_3d</a>()</code> <br><code><a href="layer_conv_3d_transpose.html">layer_conv_3d_transpose</a>()</code> <br><code><a href="layer_conv_lstm_1d.html">layer_conv_lstm_1d</a>()</code> <br><code><a href="layer_conv_lstm_2d.html">layer_conv_lstm_2d</a>()</code> <br><code><a href="layer_conv_lstm_3d.html">layer_conv_lstm_3d</a>()</code> <br><code><a href="layer_cropping_1d.html">layer_cropping_1d</a>()</code> <br><code><a href="layer_cropping_2d.html">layer_cropping_2d</a>()</code> <br><code><a href="layer_cropping_3d.html">layer_cropping_3d</a>()</code> <br><code><a href="layer_cut_mix.html">layer_cut_mix</a>()</code> <br><code><a href="layer_dense.html">layer_dense</a>()</code> <br><code><a href="layer_depthwise_conv_1d.html">layer_depthwise_conv_1d</a>()</code> <br><code><a href="layer_depthwise_conv_2d.html">layer_depthwise_conv_2d</a>()</code> <br><code><a href="layer_discretization.html">layer_discretization</a>()</code> <br><code><a href="layer_dot.html">layer_dot</a>()</code> <br><code><a href="layer_dropout.html">layer_dropout</a>()</code> <br><code><a href="layer_einsum_dense.html">layer_einsum_dense</a>()</code> <br><code><a href="layer_embedding.html">layer_embedding</a>()</code> <br><code><a href="layer_equalization.html">layer_equalization</a>()</code> <br><code><a href="layer_feature_space.html">layer_feature_space</a>()</code> <br><code><a href="layer_flatten.html">layer_flatten</a>()</code> <br><code><a href="layer_flax_module_wrapper.html">layer_flax_module_wrapper</a>()</code> <br><code><a href="layer_gaussian_dropout.html">layer_gaussian_dropout</a>()</code> <br><code><a href="layer_gaussian_noise.html">layer_gaussian_noise</a>()</code> <br><code><a href="layer_global_average_pooling_1d.html">layer_global_average_pooling_1d</a>()</code> <br><code><a href="layer_global_average_pooling_2d.html">layer_global_average_pooling_2d</a>()</code> <br><code><a href="layer_global_average_pooling_3d.html">layer_global_average_pooling_3d</a>()</code> <br><code><a href="layer_global_max_pooling_1d.html">layer_global_max_pooling_1d</a>()</code> <br><code><a href="layer_global_max_pooling_2d.html">layer_global_max_pooling_2d</a>()</code> <br><code><a href="layer_global_max_pooling_3d.html">layer_global_max_pooling_3d</a>()</code> <br><code><a href="layer_group_normalization.html">layer_group_normalization</a>()</code> <br><code><a href="layer_group_query_attention.html">layer_group_query_attention</a>()</code> <br><code><a href="layer_gru.html">layer_gru</a>()</code> <br><code><a href="layer_hashed_crossing.html">layer_hashed_crossing</a>()</code> <br><code><a href="layer_hashing.html">layer_hashing</a>()</code> <br><code><a href="layer_identity.html">layer_identity</a>()</code> <br><code><a href="layer_integer_lookup.html">layer_integer_lookup</a>()</code> <br><code><a href="layer_jax_model_wrapper.html">layer_jax_model_wrapper</a>()</code> <br><code><a href="layer_lambda.html">layer_lambda</a>()</code> <br><code><a href="layer_layer_normalization.html">layer_layer_normalization</a>()</code> <br><code><a href="layer_lstm.html">layer_lstm</a>()</code> <br><code><a href="layer_masking.html">layer_masking</a>()</code> <br><code><a href="layer_max_num_bounding_boxes.html">layer_max_num_bounding_boxes</a>()</code> <br><code><a href="layer_max_pooling_1d.html">layer_max_pooling_1d</a>()</code> <br><code><a href="layer_max_pooling_2d.html">layer_max_pooling_2d</a>()</code> <br><code><a href="layer_max_pooling_3d.html">layer_max_pooling_3d</a>()</code> <br><code><a href="layer_maximum.html">layer_maximum</a>()</code> <br><code><a href="layer_mel_spectrogram.html">layer_mel_spectrogram</a>()</code> <br><code><a href="layer_minimum.html">layer_minimum</a>()</code> <br><code><a href="layer_mix_up.html">layer_mix_up</a>()</code> <br><code><a href="layer_multi_head_attention.html">layer_multi_head_attention</a>()</code> <br><code><a href="layer_multiply.html">layer_multiply</a>()</code> <br><code><a href="layer_normalization.html">layer_normalization</a>()</code> <br><code><a href="layer_permute.html">layer_permute</a>()</code> <br><code><a href="layer_rand_augment.html">layer_rand_augment</a>()</code> <br><code><a href="layer_random_brightness.html">layer_random_brightness</a>()</code> <br><code><a href="layer_random_color_degeneration.html">layer_random_color_degeneration</a>()</code> <br><code><a href="layer_random_color_jitter.html">layer_random_color_jitter</a>()</code> <br><code><a href="layer_random_contrast.html">layer_random_contrast</a>()</code> <br><code><a href="layer_random_crop.html">layer_random_crop</a>()</code> <br><code><a href="layer_random_elastic_transform.html">layer_random_elastic_transform</a>()</code> <br><code><a href="layer_random_erasing.html">layer_random_erasing</a>()</code> <br><code><a href="layer_random_flip.html">layer_random_flip</a>()</code> <br><code><a href="layer_random_gaussian_blur.html">layer_random_gaussian_blur</a>()</code> <br><code><a href="layer_random_grayscale.html">layer_random_grayscale</a>()</code> <br><code><a href="layer_random_hue.html">layer_random_hue</a>()</code> <br><code><a href="layer_random_invert.html">layer_random_invert</a>()</code> <br><code><a href="layer_random_perspective.html">layer_random_perspective</a>()</code> <br><code><a href="layer_random_posterization.html">layer_random_posterization</a>()</code> <br><code><a href="layer_random_rotation.html">layer_random_rotation</a>()</code> <br><code><a href="layer_random_saturation.html">layer_random_saturation</a>()</code> <br><code><a href="layer_random_sharpness.html">layer_random_sharpness</a>()</code> <br><code><a href="layer_random_shear.html">layer_random_shear</a>()</code> <br><code><a href="layer_random_translation.html">layer_random_translation</a>()</code> <br><code><a href="layer_random_zoom.html">layer_random_zoom</a>()</code> <br><code><a href="layer_repeat_vector.html">layer_repeat_vector</a>()</code> <br><code><a href="layer_rescaling.html">layer_rescaling</a>()</code> <br><code><a href="layer_reshape.html">layer_reshape</a>()</code> <br><code><a href="layer_resizing.html">layer_resizing</a>()</code> <br><code><a href="layer_rms_normalization.html">layer_rms_normalization</a>()</code> <br><code><a href="layer_rnn.html">layer_rnn</a>()</code> <br><code><a href="layer_separable_conv_1d.html">layer_separable_conv_1d</a>()</code> <br><code><a href="layer_separable_conv_2d.html">layer_separable_conv_2d</a>()</code> <br><code><a href="layer_simple_rnn.html">layer_simple_rnn</a>()</code> <br><code><a href="layer_solarization.html">layer_solarization</a>()</code> <br><code><a href="layer_spatial_dropout_1d.html">layer_spatial_dropout_1d</a>()</code> <br><code><a href="layer_spatial_dropout_2d.html">layer_spatial_dropout_2d</a>()</code> <br><code><a href="layer_spatial_dropout_3d.html">layer_spatial_dropout_3d</a>()</code> <br><code><a href="layer_spectral_normalization.html">layer_spectral_normalization</a>()</code> <br><code><a href="layer_stft_spectrogram.html">layer_stft_spectrogram</a>()</code> <br><code><a href="layer_string_lookup.html">layer_string_lookup</a>()</code> <br><code><a href="layer_subtract.html">layer_subtract</a>()</code> <br><code><a href="layer_text_vectorization.html">layer_text_vectorization</a>()</code> <br><code><a href="layer_tfsm.html">layer_tfsm</a>()</code> <br><code><a href="layer_time_distributed.html">layer_time_distributed</a>()</code> <br><code><a href="layer_torch_module_wrapper.html">layer_torch_module_wrapper</a>()</code> <br><code><a href="layer_unit_normalization.html">layer_unit_normalization</a>()</code> <br><code><a href="layer_upsampling_1d.html">layer_upsampling_1d</a>()</code> <br><code><a href="layer_upsampling_2d.html">layer_upsampling_2d</a>()</code> <br><code><a href="layer_upsampling_3d.html">layer_upsampling_3d</a>()</code> <br><code><a href="layer_zero_padding_1d.html">layer_zero_padding_1d</a>()</code> <br><code><a href="layer_zero_padding_2d.html">layer_zero_padding_2d</a>()</code> <br><code><a href="layer_zero_padding_3d.html">layer_zero_padding_3d</a>()</code> <br><code><a href="rnn_cell_gru.html">rnn_cell_gru</a>()</code> <br><code><a href="rnn_cell_lstm.html">rnn_cell_lstm</a>()</code> <br><code><a href="rnn_cell_simple.html">rnn_cell_simple</a>()</code> <br></p></div>
     </div>
 
   </main><aside class="col-md-3"><nav id="toc" aria-label="Table of contents"><h2>On this page</h2>
@@ -111,7 +111,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/save_model.html b/docs/dev/reference/save_model.html
index 2a9f6fd5b..7c04dc6c2 100644
--- a/docs/dev/reference/save_model.html
+++ b/docs/dev/reference/save_model.html
@@ -8,7 +8,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -134,7 +134,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/save_model_config.html b/docs/dev/reference/save_model_config.html
index fc3020b45..509aa752b 100644
--- a/docs/dev/reference/save_model_config.html
+++ b/docs/dev/reference/save_model_config.html
@@ -10,7 +10,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -137,7 +137,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/save_model_weights.html b/docs/dev/reference/save_model_weights.html
index 351add738..c9cf9b7c4 100644
--- a/docs/dev/reference/save_model_weights.html
+++ b/docs/dev/reference/save_model_weights.html
@@ -1,5 +1,47 @@
 <!DOCTYPE html>
-<!-- Generated by pkgdown: do not edit by hand --><html lang="en"><head><meta http-equiv="Content-Type" content="text/html; charset=UTF-8"><meta charset="utf-8"><meta http-equiv="X-UA-Compatible" content="IE=edge"><meta name="viewport" content="width=device-width, initial-scale=1, shrink-to-fit=no"><title>Saves all layer weights to a .weights.h5 file. — save_model_weights • keras3</title><!-- favicons --><link rel="icon" type="image/png" sizes="96x96" href="../favicon-96x96.png"><link rel="icon" type="”image/svg+xml”" href="../favicon.svg"><link rel="apple-touch-icon" sizes="180x180" href="../apple-touch-icon.png"><link rel="icon" sizes="any" href="../favicon.ico"><link rel="manifest" href="../site.webmanifest"><script src="../deps/jquery-3.6.0/jquery-3.6.0.min.js"></script><meta name="viewport" content="width=device-width, initial-scale=1, shrink-to-fit=no"><link href="../deps/bootstrap-5.3.1/bootstrap.min.css" rel="stylesheet"><script src="../deps/bootstrap-5.3.1/bootstrap.bundle.min.js"></script><link href="../deps/font-awesome-6.5.2/css/all.min.css" rel="stylesheet"><link href="../deps/font-awesome-6.5.2/css/v4-shims.min.css" rel="stylesheet"><script src="../deps/headroom-0.11.0/headroom.min.js"></script><script src="../deps/headroom-0.11.0/jQuery.headroom.min.js"></script><script src="../deps/bootstrap-toc-1.0.1/bootstrap-toc.min.js"></script><script src="../deps/clipboard.js-2.0.11/clipboard.min.js"></script><script src="../deps/search-1.0.0/autocomplete.jquery.min.js"></script><script src="../deps/search-1.0.0/fuse.min.js"></script><script src="../deps/search-1.0.0/mark.min.js"></script><!-- pkgdown --><script src="../pkgdown.js"></script><link href="../extra.css" rel="stylesheet"><meta property="og:title" content="Saves all layer weights to a .weights.h5 file. — save_model_weights"><meta name="description" content="Saves all layer weights to a .weights.h5 file."><meta property="og:description" content="Saves all layer weights to a .weights.h5 file."><meta name="robots" content="noindex"><!-- Google Tag Manager --><script>(function(w,d,s,l,i){w[l]=w[l]||[];w[l].push({'gtm.start': new Date().getTime(),event:'gtm.js'});var f=d.getElementsByTagName(s)[0],j=d.createElement(s),dl=l!='dataLayer'?'&l='+l:'';j.async=true;j.src='https://www.googletagmanager.com/gtm.js?id='+i+dl;f.parentNode.insertBefore(j,f);})(window,document,'script','dataLayer','GTM-KHBDBW7');</script><!-- End Google Tag Manager --></head><body>
+<!-- Generated by pkgdown: do not edit by hand --><html lang="en"><head><meta http-equiv="Content-Type" content="text/html; charset=UTF-8"><meta charset="utf-8"><meta http-equiv="X-UA-Compatible" content="IE=edge"><meta name="viewport" content="width=device-width, initial-scale=1, shrink-to-fit=no"><title>Saves all weights to a single file or sharded files. — save_model_weights • keras3</title><!-- favicons --><link rel="icon" type="image/png" sizes="96x96" href="../favicon-96x96.png"><link rel="icon" type="”image/svg+xml”" href="../favicon.svg"><link rel="apple-touch-icon" sizes="180x180" href="../apple-touch-icon.png"><link rel="icon" sizes="any" href="../favicon.ico"><link rel="manifest" href="../site.webmanifest"><script src="../deps/jquery-3.6.0/jquery-3.6.0.min.js"></script><meta name="viewport" content="width=device-width, initial-scale=1, shrink-to-fit=no"><link href="../deps/bootstrap-5.3.1/bootstrap.min.css" rel="stylesheet"><script src="../deps/bootstrap-5.3.1/bootstrap.bundle.min.js"></script><link href="../deps/font-awesome-6.5.2/css/all.min.css" rel="stylesheet"><link href="../deps/font-awesome-6.5.2/css/v4-shims.min.css" rel="stylesheet"><script src="../deps/headroom-0.11.0/headroom.min.js"></script><script src="../deps/headroom-0.11.0/jQuery.headroom.min.js"></script><script src="../deps/bootstrap-toc-1.0.1/bootstrap-toc.min.js"></script><script src="../deps/clipboard.js-2.0.11/clipboard.min.js"></script><script src="../deps/search-1.0.0/autocomplete.jquery.min.js"></script><script src="../deps/search-1.0.0/fuse.min.js"></script><script src="../deps/search-1.0.0/mark.min.js"></script><!-- pkgdown --><script src="../pkgdown.js"></script><link href="../extra.css" rel="stylesheet"><meta property="og:title" content="Saves all weights to a single file or sharded files. — save_model_weights"><meta name="description" content='By default, the weights are saved in a single .weights.h5 file. Enable
+sharding via max_shard_size to split weights across multiple files (in GB)
+and produce a .weights.json manifest that tracks shard metadata.
+The saved sharded files contain:
+*.weights.json: configuration file containing metadata and
+weight_map entries.
+*_xxxxxx.weights.h5: weight shards limited by max_shard_size.
+
+
+model &amp;lt;-
+  keras_model_sequential(input_shape = 2) |&amp;gt;
+  layer_dense(4)
+
+path_h5 &amp;lt;- tempfile(fileext = ".weights.h5")
+path_json &amp;lt;- tempfile(fileext = ".weights.json")
+
+model |&amp;gt; save_model_weights(path_h5)
+model |&amp;gt; save_model_weights(path_json, max_shard_size = 0.01)
+
+model |&amp;gt; load_model_weights(path_h5)
+model |&amp;gt; load_model_weights(path_json)
+'><meta property="og:description" content='By default, the weights are saved in a single .weights.h5 file. Enable
+sharding via max_shard_size to split weights across multiple files (in GB)
+and produce a .weights.json manifest that tracks shard metadata.
+The saved sharded files contain:
+*.weights.json: configuration file containing metadata and
+weight_map entries.
+*_xxxxxx.weights.h5: weight shards limited by max_shard_size.
+
+
+model &amp;lt;-
+  keras_model_sequential(input_shape = 2) |&amp;gt;
+  layer_dense(4)
+
+path_h5 &amp;lt;- tempfile(fileext = ".weights.h5")
+path_json &amp;lt;- tempfile(fileext = ".weights.json")
+
+model |&amp;gt; save_model_weights(path_h5)
+model |&amp;gt; save_model_weights(path_json, max_shard_size = 0.01)
+
+model |&amp;gt; load_model_weights(path_h5)
+model |&amp;gt; load_model_weights(path_json)
+'><meta name="robots" content="noindex"><!-- Google Tag Manager --><script>(function(w,d,s,l,i){w[l]=w[l]||[];w[l].push({'gtm.start': new Date().getTime(),event:'gtm.js'});var f=d.getElementsByTagName(s)[0],j=d.createElement(s),dl=l!='dataLayer'?'&l='+l:'';j.async=true;j.src='https://www.googletagmanager.com/gtm.js?id='+i+dl;f.parentNode.insertBefore(j,f);})(window,document,'script','dataLayer','GTM-KHBDBW7');</script><!-- End Google Tag Manager --></head><body>
     <a href="#main" class="visually-hidden-focusable">Skip to contents</a>
     <!-- Google Tag Manager (noscript) --> <noscript><iframe src="https://www.googletagmanager.com/ns.html?id=GTM-KHBDBW7" height="0" width="0" style="display:none;visibility:hidden"></iframe></noscript> <!-- End Google Tag Manager (noscript) -->
 
@@ -8,7 +50,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -48,18 +90,35 @@
 <div class="row">
   <main id="main" class="col-md-9"><div class="page-header">
 
-      <h1>Saves all layer weights to a <code>.weights.h5</code> file.</h1>
+      <h1>Saves all weights to a single file or sharded files.</h1>
       <small class="dont-index">Source: <a href="https://github.com/rstudio/keras3/blob/HEAD/R/model-persistence.R" class="external-link"><code>R/model-persistence.R</code></a></small>
       <div class="d-none name"><code>save_model_weights.Rd</code></div>
     </div>
 
     <div class="ref-description section level2">
-    <p>Saves all layer weights to a <code>.weights.h5</code> file.</p>
+    <p>By default, the weights are saved in a single <code>.weights.h5</code> file. Enable
+sharding via <code>max_shard_size</code> to split weights across multiple files (in GB)
+and produce a <code>.weights.json</code> manifest that tracks shard metadata.</p>
+<p>The saved sharded files contain:</p><ul><li><p><code>*.weights.json</code>: configuration file containing <code>metadata</code> and
+<code>weight_map</code> entries.</p></li>
+<li><p><code>*_xxxxxx.weights.h5</code>: weight shards limited by <code>max_shard_size</code>.</p></li>
+</ul><p></p><div class="sourceCode r"><pre><code><span><span class="va">model</span> <span class="op">&lt;-</span></span>
+<span>  <span class="fu"><a href="../reference/keras_model_sequential.html">keras_model_sequential</a></span><span class="op">(</span>input_shape <span class="op">=</span> <span class="fl">2</span><span class="op">)</span> <span class="op">|&gt;</span></span>
+<span>  <span class="fu"><a href="../reference/layer_dense.html">layer_dense</a></span><span class="op">(</span><span class="fl">4</span><span class="op">)</span></span>
+<span></span>
+<span><span class="va">path_h5</span> <span class="op">&lt;-</span> <span class="fu"><a href="https://rdrr.io/r/base/tempfile.html" class="external-link">tempfile</a></span><span class="op">(</span>fileext <span class="op">=</span> <span class="st">".weights.h5"</span><span class="op">)</span></span>
+<span><span class="va">path_json</span> <span class="op">&lt;-</span> <span class="fu"><a href="https://rdrr.io/r/base/tempfile.html" class="external-link">tempfile</a></span><span class="op">(</span>fileext <span class="op">=</span> <span class="st">".weights.json"</span><span class="op">)</span></span>
+<span></span>
+<span><span class="va">model</span> <span class="op">|&gt;</span> <span class="fu"><a href="../reference/save_model_weights.html">save_model_weights</a></span><span class="op">(</span><span class="va">path_h5</span><span class="op">)</span></span>
+<span><span class="va">model</span> <span class="op">|&gt;</span> <span class="fu"><a href="../reference/save_model_weights.html">save_model_weights</a></span><span class="op">(</span><span class="va">path_json</span>, max_shard_size <span class="op">=</span> <span class="fl">0.01</span><span class="op">)</span></span>
+<span></span>
+<span><span class="va">model</span> <span class="op">|&gt;</span> <span class="fu"><a href="../reference/load_model_weights.html">load_model_weights</a></span><span class="op">(</span><span class="va">path_h5</span><span class="op">)</span></span>
+<span><span class="va">model</span> <span class="op">|&gt;</span> <span class="fu"><a href="../reference/load_model_weights.html">load_model_weights</a></span><span class="op">(</span><span class="va">path_json</span><span class="op">)</span></span></code></pre><p></p></div>
     </div>
 
     <div class="section level2">
     <h2 id="ref-usage">Usage<a class="anchor" aria-label="anchor" href="#ref-usage"></a></h2>
-    <div class="sourceCode"><pre class="sourceCode r"><code><span><span class="fu">save_model_weights</span><span class="op">(</span><span class="va">model</span>, <span class="va">filepath</span>, overwrite <span class="op">=</span> <span class="cn">FALSE</span><span class="op">)</span></span></code></pre></div>
+    <div class="sourceCode"><pre class="sourceCode r"><code><span><span class="fu">save_model_weights</span><span class="op">(</span><span class="va">model</span>, <span class="va">filepath</span>, overwrite <span class="op">=</span> <span class="cn">FALSE</span>, max_shard_size <span class="op">=</span> <span class="cn">NULL</span><span class="op">)</span></span></code></pre></div>
     </div>
 
     <div class="section level2">
@@ -67,18 +126,22 @@ <h2 id="arguments">Arguments<a class="anchor" aria-label="anchor" href="#argumen
 
 
 <dl><dt id="arg-model">model<a class="anchor" aria-label="anchor" href="#arg-model"></a></dt>
-<dd><p>A keras Model object</p></dd>
+<dd><p>A keras Model object.</p></dd>
 
 
 <dt id="arg-filepath">filepath<a class="anchor" aria-label="anchor" href="#arg-filepath"></a></dt>
-<dd><p>string.
-Path where to save the model. Must end in <code>.weights.h5</code>.</p></dd>
+<dd><p>Path where the weights will be saved. Accepts <code>.weights.h5</code>, or when sharding
+is enabled, a <code>.weights.json</code> manifest path. If <code>.weights.h5</code> is provided
+while sharding, the filename will be overridden to end in <code>.weights.json</code>.</p></dd>
 
 
 <dt id="arg-overwrite">overwrite<a class="anchor" aria-label="anchor" href="#arg-overwrite"></a></dt>
-<dd><p>Whether we should overwrite any existing model
-at the target location, or instead ask the user
-via an interactive prompt.</p></dd>
+<dd><p>Whether to overwrite any existing weights at the target location, or instead
+ask the user via an interactive prompt.</p></dd>
+
+
+<dt id="arg-max-shard-size">max_shard_size<a class="anchor" aria-label="anchor" href="#arg-max-shard-size"></a></dt>
+<dd><p>Numeric size in GB for each sharded file. Use <code>NULL</code> to disable sharding.</p></dd>
 
 </dl></div>
     <div class="section level2">
@@ -102,7 +165,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/serialize_keras_object.html b/docs/dev/reference/serialize_keras_object.html
index c68655bcf..8efe5bcc1 100644
--- a/docs/dev/reference/serialize_keras_object.html
+++ b/docs/dev/reference/serialize_keras_object.html
@@ -14,7 +14,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -103,7 +103,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/set_random_seed.html b/docs/dev/reference/set_random_seed.html
index 25baa6274..7bdb42426 100644
--- a/docs/dev/reference/set_random_seed.html
+++ b/docs/dev/reference/set_random_seed.html
@@ -42,7 +42,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -138,7 +138,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/set_state_tree.html b/docs/dev/reference/set_state_tree.html
index 4e6268129..966b4ea31 100644
--- a/docs/dev/reference/set_state_tree.html
+++ b/docs/dev/reference/set_state_tree.html
@@ -18,7 +18,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -105,7 +105,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/shape.html b/docs/dev/reference/shape.html
index 4dbbc0852..cb125ef1d 100644
--- a/docs/dev/reference/shape.html
+++ b/docs/dev/reference/shape.html
@@ -8,7 +8,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -246,7 +246,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/split_dataset.html b/docs/dev/reference/split_dataset.html
index 8526dc3fd..31aa02b6c 100644
--- a/docs/dev/reference/split_dataset.html
+++ b/docs/dev/reference/split_dataset.html
@@ -8,7 +8,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -135,7 +135,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/summary.keras.src.models.model.Model.html b/docs/dev/reference/summary.keras.src.models.model.Model.html
index f7e4d1c89..2850cc350 100644
--- a/docs/dev/reference/summary.keras.src.models.model.Model.html
+++ b/docs/dev/reference/summary.keras.src.models.model.Model.html
@@ -8,7 +8,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -160,7 +160,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/test_on_batch.html b/docs/dev/reference/test_on_batch.html
index 17923fcd3..dbc4531b6 100644
--- a/docs/dev/reference/test_on_batch.html
+++ b/docs/dev/reference/test_on_batch.html
@@ -8,7 +8,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -112,7 +112,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/text_dataset_from_directory.html b/docs/dev/reference/text_dataset_from_directory.html
index 2958dd836..c6fef1447 100644
--- a/docs/dev/reference/text_dataset_from_directory.html
+++ b/docs/dev/reference/text_dataset_from_directory.html
@@ -32,7 +32,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -229,7 +229,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/time_distributed.html b/docs/dev/reference/time_distributed.html
index 200885e23..1d3b8899a 100644
--- a/docs/dev/reference/time_distributed.html
+++ b/docs/dev/reference/time_distributed.html
@@ -10,7 +10,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -100,7 +100,7 @@ <h2 id="value">Value<a class="anchor" aria-label="anchor" href="#value"></a></h2
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/timeseries_dataset_from_array.html b/docs/dev/reference/timeseries_dataset_from_array.html
index 4f0ed0a6f..58a64c013 100644
--- a/docs/dev/reference/timeseries_dataset_from_array.html
+++ b/docs/dev/reference/timeseries_dataset_from_array.html
@@ -14,7 +14,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -219,7 +219,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/to_categorical.html b/docs/dev/reference/to_categorical.html
index b7bd99f15..fab380232 100644
--- a/docs/dev/reference/to_categorical.html
+++ b/docs/dev/reference/to_categorical.html
@@ -8,7 +8,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -129,7 +129,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/train_on_batch.html b/docs/dev/reference/train_on_batch.html
index cc225bb1e..b513288fa 100644
--- a/docs/dev/reference/train_on_batch.html
+++ b/docs/dev/reference/train_on_batch.html
@@ -8,7 +8,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -121,7 +121,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/use_backend.html b/docs/dev/reference/use_backend.html
index 6674df7aa..f6e94b985 100644
--- a/docs/dev/reference/use_backend.html
+++ b/docs/dev/reference/use_backend.html
@@ -8,7 +8,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -104,7 +104,7 @@ <h2 id="details">Details<a class="anchor" aria-label="anchor" href="#details"></
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/with_custom_object_scope.html b/docs/dev/reference/with_custom_object_scope.html
index c619437eb..19a35021d 100644
--- a/docs/dev/reference/with_custom_object_scope.html
+++ b/docs/dev/reference/with_custom_object_scope.html
@@ -8,7 +8,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -131,7 +131,7 @@ <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/reference/zip_lists.html b/docs/dev/reference/zip_lists.html
index 5bf1f80b2..6c518db70 100644
--- a/docs/dev/reference/zip_lists.html
+++ b/docs/dev/reference/zip_lists.html
@@ -16,7 +16,7 @@
 
     <a class="navbar-brand me-2" href="../index.html">keras3</a>
 
-    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.3.0.9001</small>
+    <small class="nav-text text-danger me-auto" data-bs-toggle="tooltip" data-bs-placement="bottom" title="In-development version">1.4.0.9000</small>
 
 
     <button class="navbar-toggler" type="button" data-bs-toggle="collapse" data-bs-target="#navbar" aria-controls="navbar" aria-expanded="false" aria-label="Toggle navigation">
@@ -63,7 +63,7 @@ <h1>Zip lists</h1>
 
     <div class="ref-description section level2">
     <p>This is conceptually similar to <code><a href="https://rdrr.io/r/utils/zip.html" class="external-link">zip()</a></code> in Python, or R functions
-<code><a href="https://purrr.tidyverse.org/reference/transpose.html" class="external-link">purrr::transpose()</a></code> and <code><a href="https://rdatatable.gitlab.io/data.table/reference/transpose.html" class="external-link">data.table::transpose()</a></code> (albeit, accepting
+<code><a href="https://purrr.tidyverse.org/reference/transpose.html" class="external-link">purrr::transpose()</a></code> and <code><a href="https://rdrr.io/pkg/data.table/man/transpose.html" class="external-link">data.table::transpose()</a></code> (albeit, accepting
 elements in <code>...</code> instead of a single list), with one crucial difference: if
 the provided objects are named, then matching is done by names, not
 positions.</p>
@@ -165,7 +165,7 @@ <h2 id="ref-examples">Examples<a class="anchor" aria-label="anchor" href="#ref-e
 </div>
 
 <div class="pkgdown-footer-right">
-  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.2.</p>
+  <p>Site built with <a href="https://pkgdown.r-lib.org/" class="external-link">pkgdown</a> 2.1.3.</p>
 </div>
 
     </footer></div>
diff --git a/docs/dev/search.json b/docs/dev/search.json
index 88dc92e65..1cc1e4d61 100644
--- a/docs/dev/search.json
+++ b/docs/dev/search.json
@@ -1 +1 @@
-[{"path":"https://keras3.posit.co/dev/articles/custom_train_step_in_tensorflow.html","id":"introduction","dir":"Articles","previous_headings":"","what":"Introduction","title":"Customizing what happens in `fit()` with TensorFlow","text":"’re supervised learning, can use fit() everything works smoothly. need take control every little detail, can write training loop entirely scratch. need custom training algorithm, still want benefit convenient features fit(), callbacks, built-distribution support, step fusing? core principle Keras progressive disclosure complexity. always able get lower-level workflows gradual way. shouldn’t fall cliff high-level functionality doesn’t exactly match use case. able gain control small details retaining commensurate amount high-level convenience. need customize fit() , override training step function Model class. function called fit() every batch data. able call fit() usual – running learning algorithm. Note pattern prevent building models Functional API. can whether ’re building Sequential models, Functional API models, subclassed models. Let’s see works.","code":""},{"path":"https://keras3.posit.co/dev/articles/custom_train_step_in_tensorflow.html","id":"setup","dir":"Articles","previous_headings":"","what":"Setup","title":"Customizing what happens in `fit()` with TensorFlow","text":"","code":"library(reticulate) library(tensorflow, exclude = c(\"set_random_seed\", \"shape\")) library(keras3)"},{"path":"https://keras3.posit.co/dev/articles/custom_train_step_in_tensorflow.html","id":"a-first-simple-example","dir":"Articles","previous_headings":"","what":"A first simple example","title":"Customizing what happens in `fit()` with TensorFlow","text":"Let’s start simple example: create new class subclasses Model. just override method train_step(self, data). return dictionary mapping metric names (including loss) current value. input argument data gets passed fit training data: pass arrays, calling fit(x, y, ...), data list (x, y) pass tf_dataset, calling fit(dataset, ...), data gets yielded dataset batch. body train_step() method, implement regular training update, similar already familiar . Importantly, compute loss via self.compute_loss(), wraps loss(es) function(s) passed compile(). Similarly, call metric$update_state(y, y_pred) metrics self$metrics, update state metrics passed compile(), query results self$metrics end retrieve current value. Let’s try :","code":"CustomModel <- new_model_class(   \"CustomModel\",   train_step = function(data) {     c(x, y = NULL, sample_weight = NULL) %<-% data      with(tf$GradientTape() %as% tape, {       y_pred <- self(x, training = TRUE)       loss <- self$compute_loss(y = y, y_pred = y_pred,                                 sample_weight = sample_weight)     })      # Compute gradients     trainable_vars <- self$trainable_variables     gradients <- tape$gradient(loss, trainable_vars)      # Update weights     self$optimizer$apply(gradients, trainable_vars)      # Update metrics (includes the metric that tracks the loss)     for (metric in self$metrics) {       if (metric$name == \"loss\")         metric$update_state(loss)       else         metric$update_state(y, y_pred)     }      # Return a dict mapping metric names to current value     metrics <- lapply(self$metrics, function(m) m$result())     metrics <- setNames(metrics, sapply(self$metrics, function(m) m$name))     metrics   } ) # Construct and compile an instance of CustomModel inputs <- keras_input(shape = 32) outputs <- layer_dense(inputs, 1) model <- CustomModel(inputs, outputs) model |> compile(optimizer = \"adam\", loss = \"mse\", metrics = \"mae\")  # Just use `fit` as usual x <- random_normal(c(1000, 32)) y <- random_normal(c(1000, 1)) model |> fit(x, y, epochs = 3) ## Epoch 1/3 ## 32/32 - 1s - 23ms/step - mae: 1.4339 - loss: 3.2271 ## Epoch 2/3 ## 32/32 - 0s - 2ms/step - mae: 1.3605 - loss: 2.9034 ## Epoch 3/3 ## 32/32 - 0s - 2ms/step - mae: 1.2960 - loss: 2.6272"},{"path":"https://keras3.posit.co/dev/articles/custom_train_step_in_tensorflow.html","id":"going-lower-level","dir":"Articles","previous_headings":"","what":"Going lower-level","title":"Customizing what happens in `fit()` with TensorFlow","text":"Naturally, just skip passing loss function compile(), instead everything manually train_step. Likewise metrics. ’s lower-level example, uses compile() configure optimizer: start creating Metric instances track loss MAE score (__init__()). implement custom train_step() updates state metrics (calling update_state() ), query (via result()) return current average value, displayed progress bar pass callback. Note need call reset_states() metrics epoch! Otherwise calling result() return average since start training, whereas usually work per-epoch averages. Thankfully, framework can us: just list metric want reset metrics property model. model call reset_states() object listed beginning fit() epoch beginning call evaluate().","code":"CustomModel <- new_model_class(   \"CustomModel\",   initialize = function(...) {     super$initialize(...)     self$loss_tracker <- metric_mean(name = \"loss\")     self$mae_metric <- metric_mean_absolute_error(name = \"mae\")     self$loss_fn <- loss_mean_squared_error()   },   train_step = function(data) {     c(x, y = NULL, sample_weight = NULL) %<-% data      with(tf$GradientTape() %as% tape, {       y_pred <- self(x, training = TRUE)       loss <- self$loss_fn(y, y_pred, sample_weight = sample_weight)     })      # Compute gradients     trainable_vars <- self$trainable_variables     gradients <- tape$gradient(loss, trainable_vars)      # Update weights     self$optimizer$apply(gradients, trainable_vars)      # Compute our own metrics     self$loss_tracker$update_state(loss)     self$mae_metric$update_state(y, y_pred)      # Return a dict mapping metric names to current value     list(       loss = self$loss_tracker$result(),       mae = self$mae_metric$result()     )   },   metrics = mark_active(function() {     # We list our `Metric` objects here so that `reset_states()` can be     # called automatically at the start of each epoch     # or at the start of `evaluate()`.     list(self$loss_tracker, self$mae_metric)   }) )   # Construct and compile an instance of CustomModel inputs <- keras_input(shape = 32) outputs <- layer_dense(inputs, 1) model <- CustomModel(inputs, outputs)  # We don't pass a loss or metrics here. model |> compile(optimizer = \"adam\")  # Just use `fit` as usual x <- random_normal(c(1000, 32)) y <- random_normal(c(1000, 1)) model |> fit(x, y, epochs = 3) ## Epoch 1/3 ## 32/32 - 1s - 20ms/step - loss: 2.5170 - mae: 1.2923 ## Epoch 2/3 ## 32/32 - 0s - 2ms/step - loss: 2.2689 - mae: 1.2241 ## Epoch 3/3 ## 32/32 - 0s - 2ms/step - loss: 2.0578 - mae: 1.1633"},{"path":"https://keras3.posit.co/dev/articles/custom_train_step_in_tensorflow.html","id":"supporting-sample_weight-class_weight","dir":"Articles","previous_headings":"","what":"Supporting sample_weight & class_weight","title":"Customizing what happens in `fit()` with TensorFlow","text":"may noticed first basic example didn’t make mention sample weighting. want support fit() arguments sample_weight class_weight, ’d simply following: Unpack sample_weight data argument Pass compute_loss & update_state (course, also just apply manually don’t rely compile() losses & metrics) ’s .","code":"CustomModel <- new_model_class(   \"CustomModel\",   train_step = function(data) {     c(x, y = NULL, sample_weight = NULL) %<-% data      with(tf$GradientTape() %as% tape, {       y_pred <- self(x, training = TRUE)       loss <- self$compute_loss(y = y, y_pred = y_pred,                                 sample_weight = sample_weight)     })      # Compute gradients     trainable_vars <- self$trainable_variables     gradients <- tape$gradient(loss, trainable_vars)      # Update weights     self$optimizer$apply_gradients(zip_lists(gradients, trainable_vars))      # Update metrics (includes the metric that tracks the loss)     for (metric in self$metrics) {       if (metric$name == \"loss\") {         metric$update_state(loss)       } else {         metric$update_state(y, y_pred, sample_weight = sample_weight)       }     }      # Return a dict mapping metric names to current value     metrics <- lapply(self$metrics, function(m) m$result())     metrics <- setNames(metrics, sapply(self$metrics, function(m) m$name))     metrics   } )   # Construct and compile an instance of CustomModel inputs <- keras_input(shape = 32) outputs <- layer_dense(inputs, units = 1) model <- CustomModel(inputs, outputs) model |> compile(optimizer = \"adam\", loss = \"mse\", metrics = \"mae\")  # You can now use sample_weight argument x <- random_normal(c(1000, 32)) y <- random_normal(c(1000, 1)) sw <- random_normal(c(1000, 1)) model |> fit(x, y, sample_weight = sw, epochs = 3) ## Epoch 1/3 ## 32/32 - 1s - 23ms/step - mae: 1.3434 - loss: 0.1681 ## Epoch 2/3 ## 32/32 - 0s - 2ms/step - mae: 1.3364 - loss: 0.1394 ## Epoch 3/3 ## 32/32 - 0s - 4ms/step - mae: 1.3286 - loss: 0.1148"},{"path":"https://keras3.posit.co/dev/articles/custom_train_step_in_tensorflow.html","id":"providing-your-own-evaluation-step","dir":"Articles","previous_headings":"","what":"Providing your own evaluation step","title":"Customizing what happens in `fit()` with TensorFlow","text":"want calls model.evaluate()? override test_step exactly way. ’s looks like:","code":"CustomModel <- new_model_class(   \"CustomModel\",   test_step = function(data) {     # Unpack the data     c(x, y = NULL, sw = NULL) %<-% data     # Compute predictions     y_pred = self(x, training = FALSE)     # Updates the metrics tracking the loss     self$compute_loss(y = y, y_pred = y_pred, sample_weight = sw)     # Update the metrics.     for (metric in self$metrics) {       if (metric$name != \"loss\") {         metric$update_state(y, y_pred, sample_weight = sw)       }     }     # Return a dict mapping metric names to current value.     # Note that it will include the loss (tracked in self.metrics).     metrics <- lapply(self$metrics, function(m) m$result())     metrics <- setNames(metrics, sapply(self$metrics, function(m) m$name))     metrics   } )  # Construct an instance of CustomModel inputs <- keras_input(shape = 32) outputs <- layer_dense(inputs, 1) model <- CustomModel(inputs, outputs) model |> compile(loss = \"mse\", metrics = \"mae\")  # Evaluate with our custom test_step x <- random_normal(c(1000, 32)) y <- random_normal(c(1000, 1)) model |> evaluate(x, y) ## 32/32 - 0s - 10ms/step - mae: 1.3871 - loss: 0.0000e+00 ## $loss ## tf.Tensor(0.0, shape=(), dtype=float32) ## ## $compile_metrics ## $compile_metrics$mae ## tf.Tensor(1.3871489, shape=(), dtype=float32)"},{"path":"https://keras3.posit.co/dev/articles/custom_train_step_in_tensorflow.html","id":"wrapping-up-an-end-to-end-gan-example","dir":"Articles","previous_headings":"","what":"Wrapping up: an end-to-end GAN example","title":"Customizing what happens in `fit()` with TensorFlow","text":"Let’s walk end--end example leverages everything just learned. Let’s consider: generator network meant generate 28x28x1 images. discriminator network meant classify 28x28x1 images two classes (“fake” “real”). One optimizer . loss function train discriminator. ’s feature-complete GAN class, overriding compile() use signature, implementing entire GAN algorithm 17 lines train_step: Let’s test-drive : ideas behind deep learning simple, implementation painful?","code":"# Create the discriminator discriminator <-   keras_model_sequential(name = \"discriminator\", input_shape = c(28, 28, 1)) |>    layer_conv_2d(filters = 64, kernel_size = c(3, 3),                 strides = c(2, 2),  padding = \"same\") |>   layer_activation_leaky_relu(negative_slope = 0.2) |>    layer_conv_2d(filters = 128, kernel_size = c(3, 3),                 strides = c(2, 2), padding = \"same\") |>   layer_activation_leaky_relu(negative_slope = 0.2) |>    layer_global_max_pooling_2d() |>   layer_dense(units = 1)   # Create the generator latent_dim <- 128 generator <-   keras_model_sequential(name = \"generator\", input_shape = latent_dim) |>    layer_dense(7 * 7 * 128) |>   layer_activation_leaky_relu(negative_slope = 0.2) |>    layer_reshape(target_shape = c(7, 7, 128)) |>    layer_conv_2d_transpose(filters = 128, kernel_size = c(4, 4),                           strides = c(2, 2), padding = \"same\") |>   layer_activation_leaky_relu(negative_slope = 0.2) |>    layer_conv_2d_transpose(filters = 128, kernel_size = c(4, 4),                           strides = c(2, 2), padding = \"same\") |>   layer_activation_leaky_relu(negative_slope = 0.2) |>    layer_conv_2d(filters = 1, kernel_size = c(7, 7), padding = \"same\",                 activation = \"sigmoid\") GAN <- Model(   classname = \"GAN\",    initialize = function(discriminator, generator, latent_dim, ...) {     super$initialize(...)     self$discriminator <- discriminator     self$generator <- generator     self$latent_dim <- as.integer(latent_dim)     self$d_loss_tracker <- metric_mean(name = \"d_loss\")     self$g_loss_tracker <- metric_mean(name = \"g_loss\")   },    compile = function(d_optimizer, g_optimizer, loss_fn, ...) {     super$compile(...)     self$d_optimizer <- d_optimizer     self$g_optimizer <- g_optimizer     self$loss_fn <- loss_fn   },   metrics = active_property(function() {     list(self$d_loss_tracker, self$g_loss_tracker)   }),    train_step = function(real_images) {      # Sample random points in the latent space     batch_size <- shape(real_images)[[1]]     random_latent_vectors <-       tf$random$normal(shape(batch_size, self$latent_dim))      # Decode them to fake images     generated_images <- self$generator(random_latent_vectors)      # Combine them with real images     combined_images <- op_concatenate(list(generated_images,                                            real_images))      # Assemble labels discriminating real from fake images     labels <- op_concatenate(list(op_ones(c(batch_size, 1)),                                   op_zeros(c(batch_size, 1))))      # Add random noise to the labels - important trick!     labels %<>% `+`(tf$random$uniform(shape(.), maxval = 0.05))      # Train the discriminator     with(tf$GradientTape() %as% tape, {       predictions <- self$discriminator(combined_images)       d_loss <- self$loss_fn(labels, predictions)     })     grads <- tape$gradient(d_loss, self$discriminator$trainable_weights)     self$d_optimizer$apply_gradients(       zip_lists(grads, self$discriminator$trainable_weights))      # Sample random points in the latent space     random_latent_vectors <-       tf$random$normal(shape(batch_size, self$latent_dim))      # Assemble labels that say \"all real images\"     misleading_labels <- op_zeros(c(batch_size, 1))      # Train the generator (note that we should *not* update the weights     # of the discriminator)!     with(tf$GradientTape() %as% tape, {       predictions <- self$discriminator(self$generator(random_latent_vectors))       g_loss <- self$loss_fn(misleading_labels, predictions)     })     grads <- tape$gradient(g_loss, self$generator$trainable_weights)     self$g_optimizer$apply_gradients(       zip_lists(grads, self$generator$trainable_weights))      list(d_loss = d_loss, g_loss = g_loss)   } ) batch_size <- 64 c(c(x_train, .), c(x_test, .)) %<-% dataset_mnist() all_digits <- op_concatenate(list(x_train, x_test)) all_digits <- op_reshape(all_digits, c(-1, 28, 28, 1)) dataset <- all_digits |>   tfdatasets::tensor_slices_dataset() |>   tfdatasets::dataset_map(\\(x) op_cast(x, \"float32\") / 255) |>   tfdatasets::dataset_shuffle(buffer_size = 1024) |>   tfdatasets::dataset_batch(batch_size = batch_size)  gan <- GAN(discriminator = discriminator,            generator = generator,            latent_dim = latent_dim)  gan |> compile(   d_optimizer = optimizer_adam(learning_rate = 0.0003),   g_optimizer = optimizer_adam(learning_rate = 0.0003),   loss_fn = loss_binary_crossentropy(from_logits = TRUE) )  # To limit the execution time, we only train on 100 batches. You can train on # the entire dataset. You will need about 20 epochs to get nice results. gan |> fit(   tfdatasets::dataset_take(dataset, 100),   epochs = 1 ) ## 100/100 - 5s - 53ms/step - d_loss: 0.0000e+00 - g_loss: 0.0000e+00"},{"path":"https://keras3.posit.co/dev/articles/distributed_training_with_tensorflow.html","id":"introduction","dir":"Articles","previous_headings":"","what":"Introduction","title":"Multi-GPU distributed training with TensorFlow","text":"generally two ways distribute computation across multiple devices: Data parallelism, single model gets replicated multiple devices multiple machines. processes different batches data, merge results. exist many variants setup, differ different model replicas merge results, whether stay sync every batch whether loosely coupled, etc. Model parallelism, different parts single model run different devices, processing single batch data together. works best models naturally-parallel architecture, models feature multiple branches. guide focuses data parallelism, particular synchronous data parallelism, different replicas model stay sync batch process. Synchronicity keeps model convergence behavior identical see single-device training. Specifically, guide teaches use tf.distribute API train Keras models multiple GPUs, minimal changes code, multiple GPUs (typically 2 16) installed single machine (single host, multi-device training). common setup researchers small-scale industry workflows.","code":""},{"path":"https://keras3.posit.co/dev/articles/distributed_training_with_tensorflow.html","id":"setup","dir":"Articles","previous_headings":"","what":"Setup","title":"Multi-GPU distributed training with TensorFlow","text":"","code":"library(keras3) library(tensorflow, exclude = c(\"shape\", \"set_random_seed\")) library(tfdatasets, exclude = \"shape\")"},{"path":"https://keras3.posit.co/dev/articles/distributed_training_with_tensorflow.html","id":"single-host-multi-device-synchronous-training","dir":"Articles","previous_headings":"","what":"Single-host, multi-device synchronous training","title":"Multi-GPU distributed training with TensorFlow","text":"setup, one machine several GPUs (typically 2 16). device run copy model (called replica). simplicity, follows, ’ll assume ’re dealing 8 GPUs, loss generality. works step training: current batch data (called global batch) split 8 different sub-batches (called local batches). instance, global batch 512 samples, 8 local batches 64 samples. 8 replicas independently processes local batch: run forward pass, backward pass, outputting gradient weights respect loss model local batch. weight updates originating local gradients efficiently merged across 8 replicas. done end every step, replicas always stay sync. practice, process synchronously updating weights model replicas handled level individual weight variable. done mirrored variable object. use single-host, multi-device synchronous training Keras model, use tf$distribute$MirroredStrategy API. ’s works: Instantiate MirroredStrategy, optionally configuring specific devices want use (default strategy use GPUs available). Use strategy object open scope, within scope, create Keras objects need contain variables. Typically, means creating & compiling model inside distribution scope. cases, first call fit() may also create variables, ’s good idea put fit() call scope well. Train model via fit() usual. Importantly, recommend use tf.data.Dataset objects load data multi-device distributed workflow. Schematically, looks like : ’s simple end--end runnable example:","code":"# Create a MirroredStrategy. strategy <- tf$distribute$MirroredStrategy() cat(sprintf('Number of devices: %d\\n', strategy$num_replicas_in_sync))  # Open a strategy scope. with(startegy$scope(), {   # Everything that creates variables should be under the strategy scope.   # In general this is only model construction & `compile()`.   model <- Model(...)   model |> compile(...)    # Train the model on all available devices.   model |> fit(train_dataset, validation_data=val_dataset, ...)    # Test the model on all available devices.   model |> evaluate(test_dataset) }) get_compiled_model <- function() {   inputs <- keras_input(shape = 784)   outputs <- inputs |>     layer_dense(units = 256, activation = \"relu\") |>     layer_dense(units = 256, activation = \"relu\") |>     layer_dense(units = 10)   model <- keras_model(inputs, outputs)   model |> compile(     optimizer = optimizer_adam(),     loss = loss_sparse_categorical_crossentropy(from_logits = TRUE),     metrics = list(metric_sparse_categorical_accuracy())   )   model }  get_dataset <- function(batch_size = 64) {    c(c(x_train, y_train), c(x_test, y_test)) %<-% dataset_mnist()   x_train <- array_reshape(x_train, c(-1, 784)) / 255   x_test <- array_reshape(x_test, c(-1, 784)) / 255    # Reserve 10,000 samples for validation.   val_i <- sample.int(nrow(x_train), 10000)   x_val <- x_train[val_i,]   y_val <- y_train[val_i]   x_train = x_train[-val_i,]   y_train = y_train[-val_i]    y_train <- array_reshape(y_train, c(-1, 1))   y_val <- array_reshape(y_val, c(-1, 1))   y_test <- array_reshape(y_test, c(-1, 1))    # Prepare the training dataset.   train_dataset <- list(x_train, y_train) |>     lapply(np_array, \"float32\") |>     tensor_slices_dataset() |>     dataset_batch(batch_size)    # Prepare the validation dataset.   val_dataset <- list(x_val, y_val) |>     lapply(np_array, \"float32\") |>     tensor_slices_dataset() |>     dataset_batch(batch_size)    # Prepare the test dataset.   test_dataset <- list(x_test, y_test) |>     lapply(np_array, \"float32\") |>     tensor_slices_dataset() |>     dataset_batch(batch_size)    list(train_dataset, val_dataset, test_dataset) }  # Create a MirroredStrategy. strategy <- tf$distribute$MirroredStrategy() cat(sprintf('Number of devices: %d\\n', strategy$num_replicas_in_sync)) ## Number of devices: 2 # Open a strategy scope. with(strategy$scope(), {   # Everything that creates variables should be under the strategy scope.   # In general this is only model construction & `compile()`.   model <- get_compiled_model()    c(train_dataset, val_dataset, test_dataset) %<-% get_dataset()    # Train the model on all available devices.   model |> fit(train_dataset, epochs = 2, validation_data = val_dataset)    # Test the model on all available devices.   model |> evaluate(test_dataset)  }) ## Epoch 1/2 ## 782/782 - 7s - 9ms/step - loss: nan - sparse_categorical_accuracy: nan - val_loss: nan - val_sparse_categorical_accuracy: nan ## Epoch 2/2 ## 782/782 - 5s - 7ms/step - loss: nan - sparse_categorical_accuracy: nan - val_loss: nan - val_sparse_categorical_accuracy: nan ## 157/157 - 1s - 5ms/step - loss: nan - sparse_categorical_accuracy: nan ## $loss ## [1] NaN ## ## $sparse_categorical_accuracy ## [1] NaN"},{"path":"https://keras3.posit.co/dev/articles/distributed_training_with_tensorflow.html","id":"using-callbacks-to-ensure-fault-tolerance","dir":"Articles","previous_headings":"","what":"Using callbacks to ensure fault tolerance","title":"Multi-GPU distributed training with TensorFlow","text":"using distributed training, always make sure strategy recover failure (fault tolerance). simplest way handle pass ModelCheckpoint callback fit(), save model regular intervals (e.g. every 100 batches every epoch). can restart training saved model. ’s simple example:","code":"# Prepare a directory to store all the checkpoints. checkpoint_dir <- \"./ckpt\" if (!dir.exists(checkpoint_dir)) {   dir.create(checkpoint_dir) }  make_or_restore_model <- function() {   # Either restore the latest model, or create a fresh one   # if there is no checkpoint available.   checkpoints <- list.files(checkpoint_dir, pattern = \"ckpt-.*\\\\.keras\",                             full.names = TRUE)    if (length(checkpoints) > 0) {     checkpoint_epochs <- as.integer(sub(\"ckpt-([0-9]+)\\\\.keras\", \"\\\\1\",                                         basename(checkpoints)))     latest_checkpoint <- checkpoints[which.max(checkpoint_epochs)]     load_model(latest_checkpoint)   } else {     get_compiled_model()   } }    run_training <- function(epochs = 1) {   # Create a MirroredStrategy.   strategy <- tf$distribute$MirroredStrategy()    # Open a strategy scope and create/restore the model   with(strategy$scope(), {     model <- make_or_restore_model()      callbacks <- list(       # This callback saves a SavedModel every epoch       # We include the current epoch in the folder name.       callback_model_checkpoint(         filepath = paste0(checkpoint_dir, \"/ckpt-{epoch}.keras\"),         save_freq = \"epoch\"       ))      model |> fit(       train_dataset,       epochs = epochs,       callbacks = callbacks,       validation_data = val_dataset,       verbose = 2     )   }) }  # Running the first time creates the model run_training(epochs = 1) ## 782/782 - 5s - 7ms/step - loss: nan - sparse_categorical_accuracy: nan - val_loss: nan - val_sparse_categorical_accuracy: nan # Calling the same function again will resume from where we left off run_training(epochs = 1) ## 782/782 - 6s - 7ms/step - loss: nan - sparse_categorical_accuracy: nan - val_loss: nan - val_sparse_categorical_accuracy: nan"},{"path":"https://keras3.posit.co/dev/articles/distributed_training_with_tensorflow.html","id":"tfdata-performance-tips","dir":"Articles","previous_headings":"","what":"tf$data performance tips","title":"Multi-GPU distributed training with TensorFlow","text":"distributed training, efficiency load data can often become critical. tips make sure tf$data pipelines run fast possible. Note dataset batching creating dataset, make sure batched global batch size. instance, 8 GPUs capable running batch 64 samples, call use global batch size 512. Calling dataset_cache() call dataset_cache() dataset, data cached running first iteration data. Every subsequent iteration use cached data. cache can memory (default) local file specify. can improve performance : data expected change iteration iteration reading data remote distributed filesystem reading data local disk, data fit memory workflow significantly IO-bound (e.g. reading & decoding image files). Calling dataset_prefetch(buffer_size) almost always call dataset_prefetch(buffer_size) creating dataset. means data pipeline run asynchronously model, new samples preprocessed stored buffer current batch samples used train model. next batch prefetched GPU memory time current batch . ’s !","code":""},{"path":"https://keras3.posit.co/dev/articles/distribution.html","id":"introduction","dir":"Articles","previous_headings":"","what":"Introduction","title":"Distributed training with Keras 3","text":"Keras distribution API new interface designed facilitate distributed deep learning across variety backends like JAX, TensorFlow PyTorch. powerful API introduces suite tools enabling data model parallelism, allowing efficient scaling deep learning models multiple accelerators hosts. Whether leveraging power GPUs TPUs, API provides streamlined approach initializing distributed environments, defining device meshes, orchestrating layout tensors across computational resources. classes like DataParallel ModelParallel, abstracts complexity involved parallel computation, making easier developers accelerate machine learning workflows.","code":""},{"path":"https://keras3.posit.co/dev/articles/distribution.html","id":"how-it-works","dir":"Articles","previous_headings":"","what":"How it works","title":"Distributed training with Keras 3","text":"Keras distribution API provides global programming model allows developers compose applications operate tensors global context (working single device) automatically managing distribution across many devices. API leverages underlying framework (e.g. JAX) distribute program tensors according sharding directives procedure called single program, multiple data (SPMD) expansion. decoupling application sharding directives, API enables running application single device, multiple devices, even multiple clients, preserving global semantics.","code":""},{"path":"https://keras3.posit.co/dev/articles/distribution.html","id":"setup","dir":"Articles","previous_headings":"","what":"Setup","title":"Distributed training with Keras 3","text":"","code":"# This guide assumes there are 8 GPUs available for testing. If you don't have # 8 gpus available locally, you can set the following envvar to # make xla initialize the CPU as 8 devices, to enable local testing Sys.setenv(\"CUDA_VISIBLE_DEVICES\" = \"\") Sys.setenv(\"XLA_FLAGS\" = \"--xla_force_host_platform_device_count=8\") library(keras3)  # The distribution API is only implemented for the JAX backend for now. use_backend(\"jax\", FALSE) jax <- reticulate::import(\"jax\")  library(tfdatasets, exclude = \"shape\") # For dataset input."},{"path":"https://keras3.posit.co/dev/articles/distribution.html","id":"devicemesh-and-tensorlayout","dir":"Articles","previous_headings":"","what":"DeviceMesh and TensorLayout","title":"Distributed training with Keras 3","text":"keras$distribution$DeviceMesh class Keras distribution API represents cluster computational devices configured distributed computation. aligns similar concepts jax.sharding.Mesh tf.dtensor.Mesh, ’s used map physical devices logical mesh structure. TensorLayout class specifies tensors distributed across DeviceMesh, detailing sharding tensors along specified axes correspond names axes DeviceMesh. can find detailed concept explainers TensorFlow DTensor guide.","code":"# Retrieve the local available gpu devices. devices <- jax$devices() # \"gpu\" str(devices) ## List of 8 ##  $ :TFRT_CPU_0 ##  $ :TFRT_CPU_1 ##  $ :TFRT_CPU_2 ##  $ :TFRT_CPU_3 ##  $ :TFRT_CPU_4 ##  $ :TFRT_CPU_5 ##  $ :TFRT_CPU_6 ##  $ :TFRT_CPU_7 # Define a 2x4 device mesh with data and model parallel axes mesh <- keras$distribution$DeviceMesh(   shape = shape(2, 4),   axis_names = list(\"data\", \"model\"),   devices = devices )  # A 2D layout, which describes how a tensor is distributed across the # mesh. The layout can be visualized as a 2D grid with \"model\" as rows and # \"data\" as columns, and it is a [4, 2] grid when it mapped to the physical # devices on the mesh. layout_2d <- keras$distribution$TensorLayout(   axes = c(\"model\", \"data\"),   device_mesh = mesh )  # A 4D layout which could be used for data parallelism of an image input. replicated_layout_4d <- keras$distribution$TensorLayout(   axes = list(\"data\", NULL, NULL, NULL),   device_mesh = mesh )"},{"path":"https://keras3.posit.co/dev/articles/distribution.html","id":"distribution","dir":"Articles","previous_headings":"","what":"Distribution","title":"Distributed training with Keras 3","text":"Distribution class Keras serves foundational abstract class designed developing custom distribution strategies. encapsulates core logic needed distribute model’s variables, input data, intermediate computations across device mesh. end user, won’t interact directly class, subclasses like DataParallel ModelParallel.","code":""},{"path":"https://keras3.posit.co/dev/articles/distribution.html","id":"dataparallel","dir":"Articles","previous_headings":"","what":"DataParallel","title":"Distributed training with Keras 3","text":"DataParallel class Keras distribution API designed data parallelism strategy distributed training, model weights replicated across devices DeviceMesh, device processes portion input data. sample usage class.","code":"# Create DataParallel with list of devices. # As a shortcut, the devices can be skipped, # and Keras will detect all local available devices. # E.g. data_parallel <- DataParallel() data_parallel <- keras$distribution$DataParallel(devices = devices)  # Or you can choose to create DataParallel with a 1D `DeviceMesh`. mesh_1d <- keras$distribution$DeviceMesh(   shape = shape(8),   axis_names = list(\"data\"),   devices = devices ) data_parallel <- keras$distribution$DataParallel(device_mesh = mesh_1d)  inputs <- random_normal(c(128, 28, 28, 1)) labels <- random_normal(c(128, 10)) dataset <- tensor_slices_dataset(c(inputs, labels)) |>   dataset_batch(16)  # Set the global distribution. keras$distribution$set_distribution(data_parallel)  # Note that all the model weights from here on are replicated to # all the devices of the `DeviceMesh`. This includes the RNG # state, optimizer states, metrics, etc. The dataset fed into `model |> fit()` or # `model |> evaluate()` will be split evenly on the batch dimension, and sent to # all the devices. You don't have to do any manual aggregation of losses, # since all the computation happens in a global context. inputs <- keras_input(shape = c(28, 28, 1)) outputs <- inputs |>   layer_flatten() |>   layer_dense(units = 200, use_bias = FALSE, activation = \"relu\") |>   layer_dropout(0.4) |>   layer_dense(units = 10, activation = \"softmax\")  model <- keras_model(inputs = inputs, outputs = outputs)  model |> compile(loss = \"mse\") model |> fit(dataset, epochs = 3) ## Epoch 1/3 ## 8/8 - 0s - 40ms/step - loss: 1.1536 ## Epoch 2/3 ## 8/8 - 0s - 5ms/step - loss: 1.0540 ## Epoch 3/3 ## 8/8 - 0s - 6ms/step - loss: 1.0072 model |> evaluate(dataset) ## 8/8 - 0s - 9ms/step - loss: 0.9620 ## $loss ## [1] 0.9620273"},{"path":"https://keras3.posit.co/dev/articles/distribution.html","id":"modelparallel-and-layoutmap","dir":"Articles","previous_headings":"","what":"ModelParallel and LayoutMap","title":"Distributed training with Keras 3","text":"ModelParallel mostly useful model weights large fit single accelerator. setting allows spit model weights activation tensors across devices DeviceMesh, enable horizontal scaling large models. Unlike DataParallel model weights fully replicated, weights layout ModelParallel usually need customization best performances. introduce LayoutMap let specify TensorLayout weights intermediate tensors global perspective. LayoutMap dict-like object maps string TensorLayout instances. behaves differently normal dict string key treated regex retrieving value. class allows define naming schema TensorLayout retrieve corresponding TensorLayout instance. Typically, key used query variable$path attribute, identifier variable. shortcut, list axis names also allowed inserting value, converted TensorLayout. LayoutMap can also optionally contain DeviceMesh populate TensorLayout$device_mesh set. retrieving layout key, isn’t exact match, existing keys layout map treated regex matched input key . multiple matches, ValueError raised. matches found, NULL returned. can visualize individual weights sharded also easy change mesh structure tune computation data parallel model parallel. can adjusting shape mesh. changes needed code.","code":"mesh_2d <- keras$distribution$DeviceMesh(   shape = shape(2, 4),   axis_names = c(\"data\", \"model\"),   devices = devices ) layout_map  <- keras$distribution$LayoutMap(mesh_2d)  # The rule below means that for any weights that match with d1/kernel, it # will be sharded with model dimensions (4 devices), same for the d1/bias. # All other weights will be fully replicated. layout_map[\"d1/kernel\"] <- tuple(NULL, \"model\") layout_map[\"d1/bias\"] <- tuple(\"model\")  # You can also set the layout for the layer output like layout_map[\"d2/output\"] <- tuple(\"data\", NULL)  model_parallel <- keras$distribution$ModelParallel(   layout_map = layout_map, batch_dim_name = \"data\" )  keras$distribution$set_distribution(model_parallel)  inputs <- layer_input(shape = c(28, 28, 1)) outputs <- inputs |>   layer_flatten() |>   layer_dense(units = 200, use_bias = FALSE,               activation = \"relu\", name = \"d1\") |>   layer_dropout(0.4) |>   layer_dense(units = 10,               activation = \"softmax\",               name = \"d2\")  model <- keras_model(inputs = inputs, outputs = outputs) d1 <- get_layer(model, \"d1\") d1$kernel$value |> jax$debug$visualize_array_sharding() ## ┌───────┬───────┬───────┬───────┐ ## │       │       │       │       │ ## │       │       │       │       │ ## │       │       │       │       │ ## │       │       │       │       │ ## │CPU 0,4│CPU 1,5│CPU 2,6│CPU 3,7│ ## │       │       │       │       │ ## │       │       │       │       │ ## │       │       │       │       │ ## │       │       │       │       │ ## └───────┴───────┴───────┴───────┘ d2 <- get_layer(model, \"d2\") d2$kernel$value |> jax$debug$visualize_array_sharding() ## ┌───────────────────┐ ## │                   │ ## │                   │ ## │                   │ ## │                   │ ## │CPU 0,1,2,3,4,5,6,7│ ## │                   │ ## │                   │ ## │                   │ ## │                   │ ## └───────────────────┘ d2$bias$value |> jax$debug$visualize_array_sharding() ## ┌───────────────────┐ ## │CPU 0,1,2,3,4,5,6,7│ ## └───────────────────┘ x_batch <- dataset |>   as_iterator() |> iter_next() |>   _[[1]] |> op_convert_to_tensor()  output_array <- model(x_batch) output_array |> jax$debug$visualize_array_sharding() ## ┌─────────────┐ ## │             │ ## │ CPU 0,1,2,3 │ ## │             │ ## │             │ ## ├─────────────┤ ## │             │ ## │ CPU 4,5,6,7 │ ## │             │ ## │             │ ## └─────────────┘ # The data will be sharded across the \"data\" dimension of the method, which # has 2 devices. model |> compile(loss = \"mse\") model |> fit(dataset, epochs = 3) ## Epoch 1/3 ## 8/8 - 0s - 46ms/step - loss: 1.1676 ## Epoch 2/3 ## 8/8 - 0s - 4ms/step - loss: 1.1134 ## Epoch 3/3 ## 8/8 - 0s - 5ms/step - loss: 1.1034 model |> evaluate(dataset) ## 8/8 - 0s - 8ms/step - loss: 1.0676 ## $loss ## [1] 1.067567 full_data_parallel_mesh <- keras$distribution$DeviceMesh(   shape = shape(8, 1),   axis_names = list(\"data\", \"model\"),   devices = devices ) more_data_parallel_mesh <- keras$distribution$DeviceMesh(   shape = shape(4, 2),   axis_names = list(\"data\", \"model\"),   devices = devices ) more_model_parallel_mesh <- keras$distribution$DeviceMesh(   shape = shape(2, 4),   axis_names = list(\"data\", \"model\"),   devices = devices ) full_model_parallel_mesh <- keras$distribution$DeviceMesh(   shape = shape(1, 8),   axis_names = list(\"data\", \"model\"),   devices = devices )"},{"path":"https://keras3.posit.co/dev/articles/distribution.html","id":"further-reading","dir":"Articles","previous_headings":"ModelParallel and LayoutMap","what":"Further reading","title":"Distributed training with Keras 3","text":"JAX Distributed arrays automatic parallelization JAX sharding module TensorFlow Distributed training DTensors TensorFlow DTensor concepts Using DTensors tf.keras","code":""},{"path":"https://keras3.posit.co/dev/articles/examples/nlp/neural_machine_translation_with_transformer.html","id":"introduction","dir":"Articles > Examples > Nlp","previous_headings":"","what":"Introduction","title":"English-to-Spanish translation with a sequence-to-sequence Transformer","text":"example, ’ll build sequence--sequence Transformer model, ’ll train English--Spanish machine translation task. ’ll learn : Vectorize text using layer_text_vectorization(). Implement layer_transformer_encoder(), layer_transformer_decoder(), layer_positional_embedding(). Prepare data training sequence--sequence model. Use trained model generate translations never-seen-input sentences (sequence--sequence inference). code featured adapted book Deep Learning R, Second Edition (chapter 11: Deep learning text). present example fairly barebones, detailed explanations building block works, well theory behind Transformers, recommend reading book.","code":""},{"path":"https://keras3.posit.co/dev/articles/examples/nlp/neural_machine_translation_with_transformer.html","id":"setup","dir":"Articles > Examples > Nlp","previous_headings":"","what":"Setup","title":"English-to-Spanish translation with a sequence-to-sequence Transformer","text":"","code":"library(glue)  library(tensorflow, exclude = c(\"set_random_seed\", \"shape\")) library(keras3)"},{"path":"https://keras3.posit.co/dev/articles/examples/nlp/neural_machine_translation_with_transformer.html","id":"downloading-the-data","dir":"Articles > Examples > Nlp","previous_headings":"","what":"Downloading the data","title":"English-to-Spanish translation with a sequence-to-sequence Transformer","text":"’ll working English--Spanish translation dataset provided Anki. Let’s download :","code":"zip_path <-   \"http://storage.googleapis.com/download.tensorflow.org/data/spa-eng.zip\" |>   get_file(origin = _, extract = TRUE)  text_path <- fs::path(zip_path, \"spa-eng/spa.txt\")"},{"path":"https://keras3.posit.co/dev/articles/examples/nlp/neural_machine_translation_with_transformer.html","id":"parsing-the-data","dir":"Articles > Examples > Nlp","previous_headings":"","what":"Parsing the data","title":"English-to-Spanish translation with a sequence-to-sequence Transformer","text":"line contains English sentence corresponding Spanish sentence. English sentence source sequence Spanish one target sequence. prepend token \"[start]\" append token \"[end]\" Spanish sentence. ’s sentence pairs look like: Now, let’s split sentence pairs training set, validation set, test set.","code":"text_file <- \"spa-eng/spa.txt\" text_pairs <- text_file %>%   readr::read_tsv(col_names = c(\"english\", \"spanish\"),                   col_types = c(\"cc\")) %>%   within(spanish %<>% paste(\"[start]\", ., \"[end]\")) df <- text_pairs[sample(nrow(text_pairs), 5), ] glue::glue_data(df, r\"(   english: {english}   spanish: {spanish} )\") |> cat(sep = \"\\n\\n\") ## english: I'm staying in Italy. ## spanish: [start] Me estoy quedando en Italia. [end] ## ## english: What's so strange about that? ## spanish: [start] ¿Qué es tan extraño acerca de eso? [end] ## ## english: All of the buses are full. ## spanish: [start] Todos los bondis están llenos. [end] ## ## english: Is this where your mother works? ## spanish: [start] ¿Es aquí donde trabaja tu madre? [end] ## ## english: Take precautions. ## spanish: [start] Ten cuidado. [end] random.shuffle(text_pairs) num_val_samples = int(0.15 * len(text_pairs)) num_train_samples = len(text_pairs) - 2 * num_val_samples train_pairs = text_pairs[:num_train_samples] val_pairs = text_pairs[num_train_samples : num_train_samples + num_val_samples] test_pairs = text_pairs[num_train_samples + num_val_samples :]  print(f\"{len(text_pairs)} total pairs\") print(f\"{len(train_pairs)} training pairs\") print(f\"{len(val_pairs)} validation pairs\") print(f\"{len(test_pairs)} test pairs\") num_test_samples <- num_val_samples <-   round(0.15 * nrow(text_pairs)) num_train_samples <- nrow(text_pairs) - num_val_samples - num_test_samples  pair_group <- sample(c(   rep(\"train\", num_train_samples),   rep(\"test\", num_test_samples),   rep(\"val\", num_val_samples) ))  train_pairs <- text_pairs[pair_group == \"train\", ] test_pairs <- text_pairs[pair_group == \"test\", ] val_pairs <- text_pairs[pair_group == \"val\", ] glue(r\"(   {nrow(text_pairs)} total pairs   {nrow(train_pairs)} training pairs   {nrow(val_pairs)} validation pairs   {nrow(test_pairs)} test pairs )\", .transformer = function(text, envir) {   val <- eval(str2lang(text), envir)   prettyNum(val, big.mark = \",\") }) ## 118,493 total pairs ## 82,945 training pairs ## 17,774 validation pairs ## 17,774 test pairs"},{"path":"https://keras3.posit.co/dev/articles/examples/nlp/neural_machine_translation_with_transformer.html","id":"vectorizing-the-text-data","dir":"Articles > Examples > Nlp","previous_headings":"","what":"Vectorizing the text data","title":"English-to-Spanish translation with a sequence-to-sequence Transformer","text":"’ll use two instances layer_text_vectorization() vectorize text data (one English one Spanish), say, turn original strings integer sequences integer represents index word vocabulary. English layer use default string standardization (strip punctuation characters) splitting scheme (split whitespace), Spanish layer use custom standardization, add character \"¿\" set punctuation characters stripped. Note: production-grade machine translation model, recommend stripping punctuation characters either language. Instead, recommend turning punctuation character token, achieve providing custom split function layer_text_vectorization(). Next, ’ll format datasets. training step, model seek predict target words N+1 (beyond) using source sentence target words 1 N. , training dataset yield tuple (inputs, targets), : inputs dictionary (named list) keys (names) encoder_inputs decoder_inputs. encoder_inputs vectorized source sentence decoder_inputs target sentence “far”, say, words 0 N used predict word N+1 (beyond) target sentence. target target sentence offset one step: provides next words target sentence – model try predict. Let’s take quick look sequence shapes (batches 64 pairs, sequences 20 steps long):","code":"punctuation_regex <- \"[¡¿]|[^[:^punct:][\\\\]]\" # the regex explained: Match ¡, or ¿, or any punctuation character except ] # # [:^punct:]: is a negated POSIX character class. # [:punct:] matches any punctuation character, so [:^punct:] matches any # character that is not a punctuation character. # [^...] negates the whole character class # So [^[:^punct:]] would matche any character that is a punctuation character. # Putting this all together, [^[:^punct:][\\\\]] matches any # punctuation character except the ] character.  custom_standardization <- function(input_string) {   input_string %>%     tf$strings$lower() %>%     tf$strings$regex_replace(punctuation_regex, \"\") }  input_string <- as_tensor(\"[start] ¡corre! [end]\") custom_standardization(input_string) ## tf.Tensor(b'[start] corre [end]', shape=(), dtype=string) vocab_size <- 15000 sequence_length <- 20  # rename to eng_vectorization eng_vectorization <- layer_text_vectorization(   max_tokens = vocab_size,   output_mode = \"int\",   output_sequence_length = sequence_length )  spa_vectorization <- layer_text_vectorization(   max_tokens = vocab_size,   output_mode = \"int\",   output_sequence_length = sequence_length + 1,   standardize = custom_standardization )  adapt(eng_vectorization, train_pairs$english) adapt(spa_vectorization, train_pairs$spanish) format_pair <- function(pair) {   eng <- pair$english |> eng_vectorization()   spa <- pair$spanish |> spa_vectorization()    spa_feature <- spa@r[NA:-2]                                                   # <1>   spa_target <- spa@r[2:NA]                                                     # <2>    features <- list(encoder_inputs = eng, decoder_inputs = spa_feature)   labels <- spa_target   sample_weight <- labels != 0    tuple(features, labels, sample_weight) }  batch_size <- 64  library(tfdatasets, exclude = \"shape\") make_dataset <- function(pairs) {   tensor_slices_dataset(pairs) |>     dataset_map(format_pair, num_parallel_calls = 4) |>     dataset_cache() |>     dataset_shuffle(2048) |>     dataset_batch(batch_size) |>     dataset_prefetch(16) }  train_ds <- make_dataset(train_pairs) val_ds <- make_dataset(val_pairs) c(inputs, targets, weights) %<-% iter_next(as_iterator(train_ds)) str(inputs) ## List of 2 ##  $ encoder_inputs:<tf.Tensor: shape=(64, 20), dtype=int64, numpy=…> ##  $ decoder_inputs:<tf.Tensor: shape=(64, 20), dtype=int64, numpy=…> str(targets) ## <tf.Tensor: shape=(64, 20), dtype=int64, numpy=…>"},{"path":"https://keras3.posit.co/dev/articles/examples/nlp/neural_machine_translation_with_transformer.html","id":"building-the-model","dir":"Articles > Examples > Nlp","previous_headings":"","what":"Building the model","title":"English-to-Spanish translation with a sequence-to-sequence Transformer","text":"sequence--sequence Transformer consists TransformerEncoder TransformerDecoder chained together. make model aware word order, also use PositionalEmbedding layer. source sequence pass TransformerEncoder, produce new representation . new representation passed TransformerDecoder, together target sequence far (target words 1 N). TransformerDecoder seek predict next words target sequence (N+1 beyond). key detail makes possible causal masking (see method get_causal_attention_mask() TransformerDecoder). TransformerDecoder sees entire sequences , thus must make sure uses information target tokens 0 N predicting token N+1 (otherwise, use information future, result model used inference time). Next, assemble end--end model.","code":"layer_transformer_encoder <- Layer(   classname = \"TransformerEncoder\",   initialize = function(embed_dim, dense_dim, num_heads, ...) {     super$initialize(...)     self$embed_dim <- embed_dim     self$dense_dim <- dense_dim     self$num_heads <- num_heads     self$attention <-       layer_multi_head_attention(num_heads = num_heads,                                  key_dim = embed_dim)      self$dense_proj <- keras_model_sequential() %>%       layer_dense(dense_dim, activation = \"relu\") %>%       layer_dense(embed_dim)      self$layernorm_1 <- layer_layer_normalization()     self$layernorm_2 <- layer_layer_normalization()     self$supports_masking <- TRUE   },    call = function(inputs, mask = NULL) {     if (!is.null(mask))       mask <- mask[, NULL, ] |> op_cast(\"int32\")      inputs %>%       { self$attention(., ., attention_mask = mask) + . } %>%       self$layernorm_1() %>%       { self$dense_proj(.) + . } %>%       self$layernorm_2()   },    get_config = function() {     config <- super$get_config()     for(name in c(\"embed_dim\", \"num_heads\", \"dense_dim\"))       config[[name]] <- self[[name]]     config   } )  layer_transformer_decoder <- Layer(   classname = \"TransformerDecoder\",    initialize = function(embed_dim, latent_dim, num_heads, ...) {     super$initialize(...)     self$embed_dim <- embed_dim     self$latent_dim <- latent_dim     self$num_heads <- num_heads     self$attention_1 <- layer_multi_head_attention(num_heads = num_heads,                                                    key_dim = embed_dim)     self$attention_2 <- layer_multi_head_attention(num_heads = num_heads,                                                    key_dim = embed_dim)     self$dense_proj <- keras_model_sequential() %>%       layer_dense(latent_dim, activation = \"relu\") %>%       layer_dense(embed_dim)      self$layernorm_1 <- layer_layer_normalization()     self$layernorm_2 <- layer_layer_normalization()     self$layernorm_3 <- layer_layer_normalization()     self$supports_masking <- TRUE   },    get_config = function() {     config <- super$get_config()     for (name in c(\"embed_dim\", \"num_heads\", \"latent_dim\"))       config[[name]] <- self[[name]]     config   },     get_causal_attention_mask = function(inputs) {     c(batch_size, sequence_length, encoding_length) %<-% op_shape(inputs)      x <- op_arange(0L, sequence_length, include_end = FALSE)     i <- x[, NULL]     j <- x[NULL, ]     mask <- op_cast(i >= j, \"int32\")      repeats <- op_stack(c(batch_size, 1L, 1L))     op_tile(mask[NULL, , ], repeats)   },    call = function(inputs, mask = NULL) {     c(inputs_seq, encoder_outputs) %<-% inputs     causal_mask <- self$get_causal_attention_mask(inputs_seq)      if (is.null(mask)) {       inputs_padding_mask <- NULL       encoder_outputs_padding_mask <- NULL     } else {       c(inputs_padding_mask, encoder_outputs_padding_mask) %<-% mask     }      attention_output_1 <- self$attention_1(       query = inputs_seq,       value = inputs_seq,       key = inputs_seq,       attention_mask = causal_mask,       query_mask = inputs_padding_mask     )     out_1 <- self$layernorm_1(inputs_seq + attention_output_1)      attention_output_2 <- self$attention_2(       query = out_1,       value = encoder_outputs,       key = encoder_outputs,       query_mask = inputs_padding_mask,       key_mask = encoder_outputs_padding_mask     )     out_2 <- self$layernorm_2(out_1 + attention_output_2)      self$layernorm_3(out_2 + self$dense_proj(out_2))   } )  layer_positional_embedding <- Layer(   classname = \"PositionalEmbedding\",    initialize = function(sequence_length, vocab_size, embed_dim, ...) {     super$initialize(...)     self$token_embeddings <- layer_embedding(       input_dim = vocab_size, output_dim = embed_dim     )     self$position_embeddings <- layer_embedding(       input_dim = sequence_length, output_dim = embed_dim     )     self$sequence_length <- sequence_length     self$vocab_size <- vocab_size     self$embed_dim <- embed_dim   },    call = function(inputs) {     c(., len) %<-% op_shape(inputs) # (batch_size, seq_len)     positions <- op_arange(0, len, dtype = \"int32\", include_end = FALSE)     embedded_tokens <- self$token_embeddings(inputs)     embedded_positions <- self$position_embeddings(positions)     embedded_tokens + embedded_positions   },    compute_mask = function(inputs, mask = NULL) {     inputs != 0L   },    get_config = function() {     config <- super$get_config()     for(name in c(\"sequence_length\", \"vocab_size\", \"embed_dim\"))       config[[name]] <- self[[name]]     config   } ) embed_dim <- 256 latent_dim <- 2048 num_heads <- 8  encoder_inputs <- layer_input(shape(NA), dtype = \"int64\",                               name = \"encoder_inputs\") encoder_outputs <- encoder_inputs %>%   layer_positional_embedding(sequence_length, vocab_size, embed_dim) %>%   layer_transformer_encoder(embed_dim, latent_dim, num_heads)  encoder <- keras_model(encoder_inputs, encoder_outputs)  decoder_inputs <-  layer_input(shape(NA), dtype = \"int64\",                                name = \"decoder_inputs\") encoded_seq_inputs <- layer_input(shape(NA, embed_dim),                                   name = \"decoder_state_inputs\")  transformer_decoder <- layer_transformer_decoder(NULL,   embed_dim, latent_dim, num_heads)  decoder_outputs <- decoder_inputs %>%   layer_positional_embedding(sequence_length, vocab_size, embed_dim) %>%   { transformer_decoder(list(., encoded_seq_inputs)) } %>%   layer_dropout(0.5) %>%   layer_dense(vocab_size, activation=\"softmax\")  decoder <- keras_model(inputs = list(decoder_inputs, encoded_seq_inputs),                        outputs = decoder_outputs)  decoder_outputs <- decoder(list(decoder_inputs, encoder_outputs))  transformer <- keras_model(   inputs = list(encoder_inputs = encoder_inputs, decoder_inputs = decoder_inputs),   outputs = decoder_outputs,   name = \"transformer\" )"},{"path":"https://keras3.posit.co/dev/articles/examples/nlp/neural_machine_translation_with_transformer.html","id":"training-our-model","dir":"Articles > Examples > Nlp","previous_headings":"","what":"Training our model","title":"English-to-Spanish translation with a sequence-to-sequence Transformer","text":"’ll use accuracy quick way monitor training progress validation data. Note machine translation typically uses BLEU scores well metrics, rather accuracy. train 1 epoch, get model actually converge train least 30 epochs.","code":"epochs <- 1  # This should be at least 30 for convergence  transformer ## Model: \"transformer\" ## ┏━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━┓ ## ┃ Layer (type)        ┃ Output Shape      ┃    Param # ┃ Connected to      ┃ ## ┡━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━┩ ## │ encoder_inputs      │ (None, None)      │          0 │ -                 │ ## │ (InputLayer)        │                   │            │                   │ ## ├─────────────────────┼───────────────────┼────────────┼───────────────────┤ ## │ positional_embeddi… │ (None, None, 256) │  3,845,120 │ encoder_inputs[0… │ ## │ (PositionalEmbeddi… │                   │            │                   │ ## ├─────────────────────┼───────────────────┼────────────┼───────────────────┤ ## │ not_equal           │ (None, None)      │          0 │ encoder_inputs[0… │ ## │ (NotEqual)          │                   │            │                   │ ## ├─────────────────────┼───────────────────┼────────────┼───────────────────┤ ## │ decoder_inputs      │ (None, None)      │          0 │ -                 │ ## │ (InputLayer)        │                   │            │                   │ ## ├─────────────────────┼───────────────────┼────────────┼───────────────────┤ ## │ transformer_encoder │ (None, None, 256) │  3,155,456 │ positional_embed… │ ## │ (TransformerEncode… │                   │            │ not_equal[0][0]   │ ## ├─────────────────────┼───────────────────┼────────────┼───────────────────┤ ## │ functional_3        │ (None, None,      │ 12,959,640 │ decoder_inputs[0… │ ## │ (Functional)        │ 15000)            │            │ transformer_enco… │ ## │                     │                   │            │ not_equal[0][0]   │ ## └─────────────────────┴───────────────────┴────────────┴───────────────────┘ ##  Total params: 19,960,216 (76.14 MB) ##  Trainable params: 19,960,216 (76.14 MB) ##  Non-trainable params: 0 (0.00 B) transformer |> compile(   \"rmsprop\",   loss = loss_sparse_categorical_crossentropy(ignore_class = 0),   metrics = \"accuracy\" )  transformer |> fit(train_ds, epochs = epochs,                    validation_data = val_ds) ## 1297/1297 - 61s - 47ms/step - accuracy: 0.1590 - loss: 3.8930 - val_accuracy: 0.1533 - val_loss: 3.6121"},{"path":"https://keras3.posit.co/dev/articles/examples/nlp/neural_machine_translation_with_transformer.html","id":"decoding-test-sentences","dir":"Articles > Examples > Nlp","previous_headings":"","what":"Decoding test sentences","title":"English-to-Spanish translation with a sequence-to-sequence Transformer","text":"Finally, let’s demonstrate translate brand new English sentences. simply feed model vectorized English sentence well target token \"[start]\", repeatedly generated next token, hit token \"[end]\". 30 epochs, get results :","code":"spa_vocab <- spa_vectorization |> get_vocabulary() max_decoded_sentence_length <- 20 tf_decode_sequence <- tf_function(function(input_sentence) {   withr::local_options(tensorflow.extract.style = \"python\")    tokenized_input_sentence <- input_sentence %>%     as_tensor(shape = c(1, 1)) %>%     eng_vectorization()   spa_vocab <- as_tensor(spa_vocab)   decoded_sentence <- as_tensor(\"[start]\", shape = c(1, 1))    for (i in tf$range(as.integer(max_decoded_sentence_length))) {      tokenized_target_sentence <-       spa_vectorization(decoded_sentence)[,NA:-1]      next_token_predictions <-       transformer(list(         encoder_inputs = tokenized_input_sentence,         decoder_inputs = tokenized_target_sentence       ))      sampled_token_index <- tf$argmax(next_token_predictions[0, i, ])     sampled_token <- spa_vocab[sampled_token_index]     decoded_sentence <-       tf$strings$join(c(decoded_sentence, sampled_token),                       separator = \" \")      if (sampled_token == \"[end]\")       break   }    decoded_sentence  })  for (i in seq(20)) {      c(input_sentence, correct_translation) %<-%       test_pairs[sample.int(nrow(test_pairs), 1), ]     cat(\"-\\n\")     cat(\"English:\", input_sentence, \"\\n\")     cat(\"Correct Translation:\", tolower(correct_translation), \"\\n\")     cat(\"  Model Translation:\", input_sentence %>% as_tensor() %>%           tf_decode_sequence() %>% as.character(), \"\\n\") } English: I'm sure everything will be fine. Correct Translation: [start] estoy segura de que todo irá bien. [end]   Model Translation: [start] estoy seguro de que todo va bien [end]"},{"path":"https://keras3.posit.co/dev/articles/examples/nlp/text_classification_from_scratch.html","id":"introduction","dir":"Articles > Examples > Nlp","previous_headings":"","what":"Introduction","title":"Text classification from scratch","text":"example shows text classification starting raw text (set text files disk). demonstrate workflow IMDB sentiment classification dataset (unprocessed version). use layer_text_vectorization() word splitting & indexing.","code":""},{"path":"https://keras3.posit.co/dev/articles/examples/nlp/text_classification_from_scratch.html","id":"setup","dir":"Articles > Examples > Nlp","previous_headings":"","what":"Setup","title":"Text classification from scratch","text":"","code":"library(tensorflow, exclude = c(\"shape\", \"set_random_seed\")) library(tfdatasets, exclude = \"shape\") library(keras3)"},{"path":"https://keras3.posit.co/dev/articles/examples/nlp/text_classification_from_scratch.html","id":"load-the-data-imdb-movie-review-sentiment-classification","dir":"Articles > Examples > Nlp","previous_headings":"","what":"Load the data: IMDB movie review sentiment classification","title":"Text classification from scratch","text":"Let’s download data inspect structure. aclImdb folder contains train test subfolder: aclImdb/train/pos aclImdb/train/neg folders contain text files, represents one review (either positive negative): interested pos neg subfolders, let’s delete subfolder text files : can use utility text_dataset_from_directory() generate labeled tf_dataset object set text files disk filed class-specific folders. Let’s use generate training, validation, test datasets. validation training datasets generated two subsets train directory, 20% samples going validation dataset 80% going training dataset. validation dataset addition test dataset useful tuning hyperparameters, model architecture, test dataset used. putting model real world however, retrained using available training data (without creating validation dataset), performance maximized. using validation_split subset arguments, make sure either specify random seed, pass shuffle=FALSE, validation & training splits get overlap. Let’s preview samples:","code":"if (!dir.exists(\"datasets/aclImdb\")) {   dir.create(\"datasets\")   download.file(     \"https://ai.stanford.edu/~amaas/data/sentiment/aclImdb_v1.tar.gz\",     \"datasets/aclImdb_v1.tar.gz\"   )   untar(\"datasets/aclImdb_v1.tar.gz\", exdir = \"datasets\")   unlink(\"datasets/aclImdb/train/unsup\", recursive = TRUE) } head(list.files(\"datasets/aclImdb/test\")) ## [1] \"labeledBow.feat\" \"neg\"             \"pos\"             \"urls_neg.txt\" ## [5] \"urls_pos.txt\" head(list.files(\"datasets/aclImdb/train\")) ## [1] \"labeledBow.feat\" \"neg\"             \"pos\"             \"unsupBow.feat\" ## [5] \"urls_neg.txt\"    \"urls_pos.txt\" writeLines(strwrap(readLines(\"datasets/aclImdb/train/pos/4229_10.txt\"))) ## Don't waste time reading my review. Go out and see this ## astonishingly good episode, which may very well be the best Columbo ## ever written! Ruth Gordon is perfectly cast as the scheming yet ## charming mystery writer who murders her son-in-law to avenge his ## murder of her daughter. Columbo is his usual rumpled, befuddled and ## far-cleverer-than-he-seems self, and this particular installment ## features fantastic chemistry between Gordon and Falk. Ironically, ## this was not written by heralded creators Levinson or Link yet is ## possibly the densest, most thoroughly original and twist-laden ## Columbo plot ever. Utterly satisfying in nearly every department ## and overflowing with droll and witty dialogue and thinking. Truly ## unexpected and inventive climax tops all. 10/10...seek this one out ## on Netflix! unlink(\"datasets/aclImdb/train/unsup\", recursive = TRUE) batch_size <- 32  raw_train_ds <- text_dataset_from_directory(   \"datasets/aclImdb/train\",   batch_size = batch_size,   validation_split = 0.2,   subset = \"training\",   seed = 1337 ) ## Found 25000 files belonging to 2 classes. ## Using 20000 files for training. raw_val_ds <- text_dataset_from_directory(   \"datasets/aclImdb/train\",   batch_size = batch_size,   validation_split = 0.2,   subset = \"validation\",   seed = 1337 ) ## Found 25000 files belonging to 2 classes. ## Using 5000 files for validation. raw_test_ds <- text_dataset_from_directory(   \"datasets/aclImdb/test\",   batch_size = batch_size ) ## Found 25000 files belonging to 2 classes. cat(\"Number of batches in raw_train_ds:\", length(raw_train_ds), \"\\n\") ## Number of batches in raw_train_ds: 625 cat(\"Number of batches in raw_val_ds:\", length(raw_val_ds), \"\\n\") ## Number of batches in raw_val_ds: 157 cat(\"Number of batches in raw_test_ds:\", length(raw_test_ds), \"\\n\") ## Number of batches in raw_test_ds: 782 # It's important to take a look at your raw data to ensure your normalization # and tokenization will work as expected. We can do that by taking a few # examples from the training set and looking at them. # This is one of the places where eager execution shines: # we can just evaluate these tensors using .numpy() # instead of needing to evaluate them in a Session/Graph context. batch <- iter_next(as_iterator(raw_train_ds)) str(batch) ## List of 2 ##  $ :<tf.Tensor: shape=(32), dtype=string, numpy=…> ##  $ :<tf.Tensor: shape=(32), dtype=int32, numpy=…> c(text_batch, label_batch) %<-% batch for (i in 1:3) {   print(text_batch[i])   print(label_batch[i]) } ## tf.Tensor(b\"I have read the novel Reaper of Ben Mezrich a fews years ago and last night I accidentally came to see this adaption.<br /><br />Although it's been years since I read the story the first time, the differences between the novel and the movie are humongous. Very important elements, which made the whole thing plausible are just written out or changed to bad.<br /><br />If the plot sounds interesting to you: go and get the novel. Its much, much, much better.<br /><br />Still 4 out of 10 since it was hard to stop watching because of the great basic plot by Ben Mezrich.\", shape=(), dtype=string) ## tf.Tensor(0, shape=(), dtype=int32) ## tf.Tensor(b'After seeing all the Jesse James, Quantrill, jayhawkers,etc films in the fifties, it is quite a thrill to see this film with a new perspective by director Ang Lee. The scene of the attack of Lawrence, Kansas is awesome. The romantic relationship between Jewel and Toby Mcguire turns out to be one of the best parts and Jonathan Rhys-Meyers is outstanding as the bad guy. All the time this film makes you feel the horror of war, and the desperate situation of the main characters who do not know if they are going to survive the next hours. Definitely worth seeing.', shape=(), dtype=string) ## tf.Tensor(1, shape=(), dtype=int32) ## tf.Tensor(b'AG was an excellent presentation of drama, suspense and thriller that is so rare to American TV. Sheriff Lucas gave many a viewer the willies. We rooted for Caleb as he strove to resist the overtures of Sheriff Lucas. We became engrossed and fearful upon learning of the unthinkable connection between these two characters. The manipulations which weekly gave cause to fear what Lucas would do next were truly surprising. This show lived up to the \"Gothic\" moniker in ways American entertainment has so seldom attempted, much less mastered. The suits definitely made a big mistake in not supporting this show. This show puts shame to the current glut of \"reality\" shows- which are so less than satisfying viewing.The call for a DVD box set is well based. This show is quality viewing for a discerning market hungry for quality viewing. A public that is tiring of over-saturation of mind-numbing reality fare will welcome this gem of real storytelling. Bring on the DVD box set!!', shape=(), dtype=string) ## tf.Tensor(1, shape=(), dtype=int32)"},{"path":"https://keras3.posit.co/dev/articles/examples/nlp/text_classification_from_scratch.html","id":"prepare-the-data","dir":"Articles > Examples > Nlp","previous_headings":"","what":"Prepare the data","title":"Text classification from scratch","text":"particular, remove <br /> tags.","code":"# Having looked at our data above, we see that the raw text contains HTML break # tags of the form '<br />'. These tags will not be removed by the default # standardizer (which doesn't strip HTML). Because of this, we will need to # create a custom standardization function. custom_standardization_fn <- function(string_tensor) {   string_tensor |>     tf$strings$lower() |> # convert to all lowercase     tf$strings$regex_replace(\"<br />\", \" \") |> # remove '<br />' HTML tag     tf$strings$regex_replace(\"[[:punct:]]\", \"\") # remove punctuation }   # Model constants. max_features <- 20000 embedding_dim <- 128 sequence_length <- 500  # Now that we have our custom standardization, we can instantiate our text # vectorization layer. We are using this layer to normalize, split, and map # strings to integers, so we set our 'output_mode' to 'int'. # Note that we're using the default split function, # and the custom standardization defined above. # We also set an explicit maximum sequence length, since the CNNs later in our # model won't support ragged sequences. vectorize_layer <- layer_text_vectorization(   standardize = custom_standardization_fn,   max_tokens = max_features,   output_mode = \"int\",   output_sequence_length = sequence_length, )  # Now that the vectorize_layer has been created, call `adapt` on a text-only # dataset to create the vocabulary. You don't have to batch, but for very large # datasets this means you're not keeping spare copies of the dataset in memory.  # Let's make a text-only dataset (no labels): text_ds <- raw_train_ds |>   dataset_map(\\(x, y) x) # Let's call `adapt`: vectorize_layer |> adapt(text_ds)"},{"path":"https://keras3.posit.co/dev/articles/examples/nlp/text_classification_from_scratch.html","id":"two-options-to-vectorize-the-data","dir":"Articles > Examples > Nlp","previous_headings":"","what":"Two options to vectorize the data","title":"Text classification from scratch","text":"2 ways can use text vectorization layer: Option 1: Make part model, obtain model processes raw strings, like : Option 2: Apply text dataset obtain dataset word indices, feed model expects integer sequences inputs. important difference two option 2 enables asynchronous CPU processing buffering data training GPU. ’re training model GPU, probably want go option get best performance. . export model production, ’d ship model accepts raw strings input, like code snippet option 1 . can done training. last section.","code":"text_input <- keras_input(shape = c(1L), dtype = \"string\", name = 'text') x <- text_input |>   vectorize_layer() |>   layer_embedding(max_features + 1, embedding_dim) vectorize_text <- function(text, label) {   text <- text |>     op_expand_dims(-1) |>     vectorize_layer()   list(text, label) }  # Vectorize the data. train_ds <- raw_train_ds |> dataset_map(vectorize_text) val_ds   <- raw_val_ds   |> dataset_map(vectorize_text) test_ds  <- raw_test_ds  |> dataset_map(vectorize_text)  # Do async prefetching / buffering of the data for best performance on GPU. train_ds <- train_ds |>   dataset_cache() |>   dataset_prefetch(buffer_size = 10) val_ds <- val_ds |>   dataset_cache() |>   dataset_prefetch(buffer_size = 10) test_ds <- test_ds |>   dataset_cache() |>   dataset_prefetch(buffer_size = 10)"},{"path":"https://keras3.posit.co/dev/articles/examples/nlp/text_classification_from_scratch.html","id":"build-a-model","dir":"Articles > Examples > Nlp","previous_headings":"","what":"Build a model","title":"Text classification from scratch","text":"choose simple 1D convnet starting Embedding layer.","code":"# A integer input for vocab indices. inputs <- keras_input(shape = c(NA), dtype = \"int64\")  predictions <- inputs |>   # Next, we add a layer to map those vocab indices into a space of dimensionality   # 'embedding_dim'.   layer_embedding(max_features, embedding_dim) |>   layer_dropout(0.5) |>   # Conv1D + global max pooling   layer_conv_1d(128, 7, padding = \"valid\", activation = \"relu\", strides = 3) |>   layer_conv_1d(128, 7, padding = \"valid\", activation = \"relu\", strides = 3) |>   layer_global_max_pooling_1d() |>   # We add a vanilla hidden layer:   layer_dense(128, activation = \"relu\") |>   layer_dropout(0.5) |>   # We project onto a single unit output layer, and squash it with a sigmoid:   layer_dense(1, activation = \"sigmoid\", name = \"predictions\")  model <- keras_model(inputs, predictions)  summary(model) ## Model: \"functional\" ## ┏━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━┓ ## ┃ Layer (type)                    ┃ Output Shape           ┃       Param # ┃ ## ┡━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━┩ ## │ input_layer (InputLayer)        │ (None, None)           │             0 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ embedding_1 (Embedding)         │ (None, None, 128)      │     2,560,000 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ dropout (Dropout)               │ (None, None, 128)      │             0 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ conv1d (Conv1D)                 │ (None, None, 128)      │       114,816 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ conv1d_1 (Conv1D)               │ (None, None, 128)      │       114,816 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ global_max_pooling1d            │ (None, 128)            │             0 │ ## │ (GlobalMaxPooling1D)            │                        │               │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ dense (Dense)                   │ (None, 128)            │        16,512 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ dropout_1 (Dropout)             │ (None, 128)            │             0 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ predictions (Dense)             │ (None, 1)              │           129 │ ## └─────────────────────────────────┴────────────────────────┴───────────────┘ ##  Total params: 2,806,273 (10.71 MB) ##  Trainable params: 2,806,273 (10.71 MB) ##  Non-trainable params: 0 (0.00 B) # Compile the model with binary crossentropy loss and an adam optimizer. model |> compile(loss = \"binary_crossentropy\",                  optimizer = \"adam\",                  metrics = \"accuracy\")"},{"path":"https://keras3.posit.co/dev/articles/examples/nlp/text_classification_from_scratch.html","id":"train-the-model","dir":"Articles > Examples > Nlp","previous_headings":"","what":"Train the model","title":"Text classification from scratch","text":"","code":"epochs <- 3  # Fit the model using the train and test datasets. model |> fit(train_ds, validation_data = val_ds, epochs = epochs) ## Epoch 1/3 ## 625/625 - 6s - 10ms/step - accuracy: 0.6931 - loss: 0.5276 - val_accuracy: 0.8588 - val_loss: 0.3292 ## Epoch 2/3 ## 625/625 - 4s - 6ms/step - accuracy: 0.9050 - loss: 0.2397 - val_accuracy: 0.8762 - val_loss: 0.3251 ## Epoch 3/3 ## 625/625 - 3s - 4ms/step - accuracy: 0.9579 - loss: 0.1206 - val_accuracy: 0.8586 - val_loss: 0.4119"},{"path":"https://keras3.posit.co/dev/articles/examples/nlp/text_classification_from_scratch.html","id":"evaluate-the-model-on-the-test-set","dir":"Articles > Examples > Nlp","previous_headings":"","what":"Evaluate the model on the test set","title":"Text classification from scratch","text":"","code":"model |> evaluate(test_ds) ## 782/782 - 1s - 2ms/step - accuracy: 0.8437 - loss: 0.4384 ## $accuracy ## [1] 0.84368 ## ## $loss ## [1] 0.4383631"},{"path":"https://keras3.posit.co/dev/articles/examples/nlp/text_classification_from_scratch.html","id":"make-an-end-to-end-model","dir":"Articles > Examples > Nlp","previous_headings":"","what":"Make an end-to-end model","title":"Text classification from scratch","text":"want obtain model capable processing raw strings, can simply create new model (using weights just trained):","code":"# A string input inputs <- keras_input(shape = c(1), dtype = \"string\") # Turn strings into vocab indices indices <- vectorize_layer(inputs) # Turn vocab indices into predictions outputs <- model(indices)  # Our end to end model end_to_end_model <- keras_model(inputs, outputs) end_to_end_model |> compile(   loss = \"binary_crossentropy\",   optimizer = \"adam\",   metrics = c(\"accuracy\") )  # Test it with `raw_test_ds`, which yields raw strings end_to_end_model |> evaluate(raw_test_ds) ## 782/782 - 3s - 4ms/step - accuracy: 0.8437 - loss: 0.4384 ## $accuracy ## [1] 0.84368 ## ## $loss ## [1] 0.4383632"},{"path":"https://keras3.posit.co/dev/articles/examples/structured_data/imbalanced_classification.html","id":"introduction","dir":"Articles > Examples > Structured_data","previous_headings":"","what":"Introduction","title":"Imbalanced classification: credit card fraud detection","text":"example looks Kaggle Credit Card Fraud Detection dataset demonstrate train classification model data highly imbalanced classes. can download data clicking “Download” link, ’re setup kaggle API key \"~/.kaggle/kagle.json\", can run following:","code":"reticulate::py_install(\"kaggle\", pip = TRUE) reticulate::py_available(TRUE) # ensure 'kaggle' is on the PATH system(\"kaggle datasets download -d mlg-ulb/creditcardfraud\") zip::unzip(\"creditcardfraud.zip\", files = \"creditcard.csv\")"},{"path":"https://keras3.posit.co/dev/articles/examples/structured_data/imbalanced_classification.html","id":"first-load-the-data","dir":"Articles > Examples > Structured_data","previous_headings":"","what":"First, load the data","title":"Imbalanced classification: credit card fraud detection","text":"","code":"library(readr) df <- read_csv(\"creditcard.csv\", col_types = cols(   Class = col_integer(),   .default = col_double() )) tibble::glimpse(df) ## Rows: 284,807 ## Columns: 31 ## $ Time   <dbl> 0, 0, 1, 1, 2, 2, 4, 7, 7, 9, 10, 10, 10, 11, 12, 12, 12, 1… ## $ V1     <dbl> -1.3598071, 1.1918571, -1.3583541, -0.9662717, -1.1582331, … ## $ V2     <dbl> -0.07278117, 0.26615071, -1.34016307, -0.18522601, 0.877736… ## $ V3     <dbl> 2.53634674, 0.16648011, 1.77320934, 1.79299334, 1.54871785,… ## $ V4     <dbl> 1.37815522, 0.44815408, 0.37977959, -0.86329128, 0.40303393… ## $ V5     <dbl> -0.33832077, 0.06001765, -0.50319813, -0.01030888, -0.40719… ## $ V6     <dbl> 0.46238778, -0.08236081, 1.80049938, 1.24720317, 0.09592146… ## $ V7     <dbl> 0.239598554, -0.078802983, 0.791460956, 0.237608940, 0.5929… ## $ V8     <dbl> 0.098697901, 0.085101655, 0.247675787, 0.377435875, -0.2705… ## $ V9     <dbl> 0.3637870, -0.2554251, -1.5146543, -1.3870241, 0.8177393, -… ## $ V10    <dbl> 0.09079417, -0.16697441, 0.20764287, -0.05495192, 0.7530744… ## $ V11    <dbl> -0.55159953, 1.61272666, 0.62450146, -0.22648726, -0.822842… ## $ V12    <dbl> -0.61780086, 1.06523531, 0.06608369, 0.17822823, 0.53819555… ## $ V13    <dbl> -0.99138985, 0.48909502, 0.71729273, 0.50775687, 1.34585159… ## $ V14    <dbl> -0.31116935, -0.14377230, -0.16594592, -0.28792375, -1.1196… ## $ V15    <dbl> 1.468176972, 0.635558093, 2.345864949, -0.631418118, 0.1751… ## $ V16    <dbl> -0.47040053, 0.46391704, -2.89008319, -1.05964725, -0.45144… ## $ V17    <dbl> 0.207971242, -0.114804663, 1.109969379, -0.684092786, -0.23… ## $ V18    <dbl> 0.02579058, -0.18336127, -0.12135931, 1.96577500, -0.038194… ## $ V19    <dbl> 0.40399296, -0.14578304, -2.26185710, -1.23262197, 0.803486… ## $ V20    <dbl> 0.25141210, -0.06908314, 0.52497973, -0.20803778, 0.4085423… ## $ V21    <dbl> -0.018306778, -0.225775248, 0.247998153, -0.108300452, -0.0… ## $ V22    <dbl> 0.277837576, -0.638671953, 0.771679402, 0.005273597, 0.7982… ## $ V23    <dbl> -0.110473910, 0.101288021, 0.909412262, -0.190320519, -0.13… ## $ V24    <dbl> 0.06692807, -0.33984648, -0.68928096, -1.17557533, 0.141266… ## $ V25    <dbl> 0.12853936, 0.16717040, -0.32764183, 0.64737603, -0.2060095… ## $ V26    <dbl> -0.18911484, 0.12589453, -0.13909657, -0.22192884, 0.502292… ## $ V27    <dbl> 0.133558377, -0.008983099, -0.055352794, 0.062722849, 0.219… ## $ V28    <dbl> -0.021053053, 0.014724169, -0.059751841, 0.061457629, 0.215… ## $ Amount <dbl> 149.62, 2.69, 378.66, 123.50, 69.99, 3.67, 4.99, 40.80, 93.… ## $ Class  <int> 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,…"},{"path":"https://keras3.posit.co/dev/articles/examples/structured_data/imbalanced_classification.html","id":"prepare-a-validation-set","dir":"Articles > Examples > Structured_data","previous_headings":"","what":"Prepare a validation set","title":"Imbalanced classification: credit card fraud detection","text":"","code":"val_idx <- nrow(df) %>% sample.int(., round( . * 0.2)) val_df <- df[val_idx, ] train_df <- df[-val_idx, ]  cat(\"Number of training samples:\", nrow(train_df), \"\\n\") ## Number of training samples: 227846 cat(\"Number of validation samples:\", nrow(val_df), \"\\n\") ## Number of validation samples: 56961"},{"path":"https://keras3.posit.co/dev/articles/examples/structured_data/imbalanced_classification.html","id":"analyze-class-imbalance-in-the-targets","dir":"Articles > Examples > Structured_data","previous_headings":"","what":"Analyze class imbalance in the targets","title":"Imbalanced classification: credit card fraud detection","text":"","code":"counts <- table(train_df$Class) counts ## ##      0      1 ## 227451    395 cat(sprintf(\"Number of positive samples in training data: %i (%.2f%% of total)\",             counts[\"1\"], 100 * counts[\"1\"] / sum(counts))) ## Number of positive samples in training data: 395 (0.17% of total) weight_for_0 = 1 / counts[\"0\"] weight_for_1 = 1 / counts[\"1\"]"},{"path":"https://keras3.posit.co/dev/articles/examples/structured_data/imbalanced_classification.html","id":"normalize-the-data-using-training-set-statistics","dir":"Articles > Examples > Structured_data","previous_headings":"","what":"Normalize the data using training set statistics","title":"Imbalanced classification: credit card fraud detection","text":"","code":"feature_names <- colnames(train_df) %>% setdiff(\"Class\")  train_features <- as.matrix(train_df[feature_names]) train_targets <- as.matrix(train_df$Class)  val_features <- as.matrix(val_df[feature_names]) val_targets <- as.matrix(val_df$Class)  train_features %<>% scale() val_features %<>% scale(center = attr(train_features, \"scaled:center\"),                         scale = attr(train_features, \"scaled:scale\"))"},{"path":"https://keras3.posit.co/dev/articles/examples/structured_data/imbalanced_classification.html","id":"build-a-binary-classification-model","dir":"Articles > Examples > Structured_data","previous_headings":"","what":"Build a binary classification model","title":"Imbalanced classification: credit card fraud detection","text":"","code":"model <-   keras_model_sequential(input_shape = ncol(train_features)) |>   layer_dense(256, activation = \"relu\") |>   layer_dense(256, activation = \"relu\") |>   layer_dropout(0.3) |>   layer_dense(256, activation = \"relu\") |>   layer_dropout(0.3) |>   layer_dense(1, activation = \"sigmoid\")  model ## Model: \"sequential\" ## ┏━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━┓ ## ┃ Layer (type)                    ┃ Output Shape           ┃       Param # ┃ ## ┡━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━┩ ## │ dense (Dense)                   │ (None, 256)            │         7,936 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ dense_1 (Dense)                 │ (None, 256)            │        65,792 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ dropout (Dropout)               │ (None, 256)            │             0 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ dense_2 (Dense)                 │ (None, 256)            │        65,792 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ dropout_1 (Dropout)             │ (None, 256)            │             0 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ dense_3 (Dense)                 │ (None, 1)              │           257 │ ## └─────────────────────────────────┴────────────────────────┴───────────────┘ ##  Total params: 139,777 (546.00 KB) ##  Trainable params: 139,777 (546.00 KB) ##  Non-trainable params: 0 (0.00 B)"},{"path":"https://keras3.posit.co/dev/articles/examples/structured_data/imbalanced_classification.html","id":"train-the-model-with-class_weight-argument","dir":"Articles > Examples > Structured_data","previous_headings":"","what":"Train the model with class_weight argument","title":"Imbalanced classification: credit card fraud detection","text":"","code":"metrics <- list(   metric_false_negatives(name = \"fn\"),   metric_false_positives(name = \"fp\"),   metric_true_negatives(name = \"tn\"),   metric_true_positives(name = \"tp\"),   metric_precision(name = \"precision\"),   metric_recall(name = \"recall\") ) model |> compile(   optimizer = optimizer_adam(1e-2),   loss = \"binary_crossentropy\",   metrics = metrics ) callbacks <- list(   callback_model_checkpoint(\"fraud_model_at_epoch_{epoch}.keras\") )  class_weight <- list(\"0\" = weight_for_0,                      \"1\" = weight_for_1)  model |> fit(   train_features, train_targets,   validation_data = list(val_features, val_targets),   class_weight = class_weight,   batch_size = 2048,   epochs = 30,   callbacks = callbacks,   verbose = 2 ) ## Epoch 1/30 ## 112/112 - 5s - 41ms/step - fn: 36.0000 - fp: 30458.0000 - loss: 2.3118e-06 - precision: 0.0116 - recall: 0.9089 - tn: 196993.0000 - tp: 359.0000 - val_fn: 14.0000 - val_fp: 714.0000 - val_loss: 0.0906 - val_precision: 0.1041 - val_recall: 0.8557 - val_tn: 56150.0000 - val_tp: 83.0000 ## Epoch 2/30 ## 112/112 - 1s - 11ms/step - fn: 34.0000 - fp: 5698.0000 - loss: 1.4544e-06 - precision: 0.0596 - recall: 0.9139 - tn: 221753.0000 - tp: 361.0000 - val_fn: 6.0000 - val_fp: 2982.0000 - val_loss: 0.1611 - val_precision: 0.0296 - val_recall: 0.9381 - val_tn: 53882.0000 - val_tp: 91.0000 ## Epoch 3/30 ## 112/112 - 0s - 2ms/step - fn: 30.0000 - fp: 7494.0000 - loss: 1.1267e-06 - precision: 0.0464 - recall: 0.9241 - tn: 219957.0000 - tp: 365.0000 - val_fn: 9.0000 - val_fp: 2262.0000 - val_loss: 0.1346 - val_precision: 0.0374 - val_recall: 0.9072 - val_tn: 54602.0000 - val_tp: 88.0000 ## Epoch 4/30 ## 112/112 - 0s - 2ms/step - fn: 25.0000 - fp: 5976.0000 - loss: 9.9649e-07 - precision: 0.0583 - recall: 0.9367 - tn: 221475.0000 - tp: 370.0000 - val_fn: 9.0000 - val_fp: 1593.0000 - val_loss: 0.0781 - val_precision: 0.0523 - val_recall: 0.9072 - val_tn: 55271.0000 - val_tp: 88.0000 ## Epoch 5/30 ## 112/112 - 0s - 2ms/step - fn: 22.0000 - fp: 7713.0000 - loss: 9.5664e-07 - precision: 0.0461 - recall: 0.9443 - tn: 219738.0000 - tp: 373.0000 - val_fn: 10.0000 - val_fp: 613.0000 - val_loss: 0.0444 - val_precision: 0.1243 - val_recall: 0.8969 - val_tn: 56251.0000 - val_tp: 87.0000 ## Epoch 6/30 ## 112/112 - 0s - 2ms/step - fn: 18.0000 - fp: 8517.0000 - loss: 9.6051e-07 - precision: 0.0424 - recall: 0.9544 - tn: 218934.0000 - tp: 377.0000 - val_fn: 12.0000 - val_fp: 1033.0000 - val_loss: 0.0643 - val_precision: 0.0760 - val_recall: 0.8763 - val_tn: 55831.0000 - val_tp: 85.0000 ## Epoch 7/30 ## 112/112 - 0s - 2ms/step - fn: 17.0000 - fp: 6409.0000 - loss: 7.3375e-07 - precision: 0.0557 - recall: 0.9570 - tn: 221042.0000 - tp: 378.0000 - val_fn: 10.0000 - val_fp: 1549.0000 - val_loss: 0.0697 - val_precision: 0.0532 - val_recall: 0.8969 - val_tn: 55315.0000 - val_tp: 87.0000 ## Epoch 8/30 ## 112/112 - 0s - 2ms/step - fn: 14.0000 - fp: 6080.0000 - loss: 6.5853e-07 - precision: 0.0590 - recall: 0.9646 - tn: 221371.0000 - tp: 381.0000 - val_fn: 10.0000 - val_fp: 2677.0000 - val_loss: 0.1178 - val_precision: 0.0315 - val_recall: 0.8969 - val_tn: 54187.0000 - val_tp: 87.0000 ## Epoch 9/30 ## 112/112 - 0s - 2ms/step - fn: 11.0000 - fp: 6592.0000 - loss: 5.9467e-07 - precision: 0.0550 - recall: 0.9722 - tn: 220859.0000 - tp: 384.0000 - val_fn: 9.0000 - val_fp: 2094.0000 - val_loss: 0.0806 - val_precision: 0.0403 - val_recall: 0.9072 - val_tn: 54770.0000 - val_tp: 88.0000 ## Epoch 10/30 ## 112/112 - 0s - 2ms/step - fn: 7.0000 - fp: 6334.0000 - loss: 5.3129e-07 - precision: 0.0577 - recall: 0.9823 - tn: 221117.0000 - tp: 388.0000 - val_fn: 10.0000 - val_fp: 850.0000 - val_loss: 0.0396 - val_precision: 0.0928 - val_recall: 0.8969 - val_tn: 56014.0000 - val_tp: 87.0000 ## Epoch 11/30 ## 112/112 - 0s - 2ms/step - fn: 8.0000 - fp: 5922.0000 - loss: 5.0009e-07 - precision: 0.0613 - recall: 0.9797 - tn: 221529.0000 - tp: 387.0000 - val_fn: 12.0000 - val_fp: 1488.0000 - val_loss: 0.0506 - val_precision: 0.0540 - val_recall: 0.8763 - val_tn: 55376.0000 - val_tp: 85.0000 ## Epoch 12/30 ## 112/112 - 0s - 2ms/step - fn: 8.0000 - fp: 7243.0000 - loss: 5.7125e-07 - precision: 0.0507 - recall: 0.9797 - tn: 220208.0000 - tp: 387.0000 - val_fn: 11.0000 - val_fp: 1014.0000 - val_loss: 0.0414 - val_precision: 0.0782 - val_recall: 0.8866 - val_tn: 55850.0000 - val_tp: 86.0000 ## Epoch 13/30 ## 112/112 - 0s - 2ms/step - fn: 6.0000 - fp: 4141.0000 - loss: 3.7800e-07 - precision: 0.0859 - recall: 0.9848 - tn: 223310.0000 - tp: 389.0000 - val_fn: 8.0000 - val_fp: 2491.0000 - val_loss: 0.0986 - val_precision: 0.0345 - val_recall: 0.9175 - val_tn: 54373.0000 - val_tp: 89.0000 ## Epoch 14/30 ## 112/112 - 0s - 2ms/step - fn: 6.0000 - fp: 6118.0000 - loss: 5.4537e-07 - precision: 0.0598 - recall: 0.9848 - tn: 221333.0000 - tp: 389.0000 - val_fn: 9.0000 - val_fp: 2488.0000 - val_loss: 0.1019 - val_precision: 0.0342 - val_recall: 0.9072 - val_tn: 54376.0000 - val_tp: 88.0000 ## Epoch 15/30 ## 112/112 - 0s - 2ms/step - fn: 4.0000 - fp: 4646.0000 - loss: 4.0941e-07 - precision: 0.0776 - recall: 0.9899 - tn: 222805.0000 - tp: 391.0000 - val_fn: 8.0000 - val_fp: 2949.0000 - val_loss: 0.1263 - val_precision: 0.0293 - val_recall: 0.9175 - val_tn: 53915.0000 - val_tp: 89.0000 ## Epoch 16/30 ## 112/112 - 0s - 2ms/step - fn: 10.0000 - fp: 7978.0000 - loss: 8.1907e-07 - precision: 0.0460 - recall: 0.9747 - tn: 219473.0000 - tp: 385.0000 - val_fn: 8.0000 - val_fp: 3430.0000 - val_loss: 0.2915 - val_precision: 0.0253 - val_recall: 0.9175 - val_tn: 53434.0000 - val_tp: 89.0000 ## Epoch 17/30 ## 112/112 - 0s - 2ms/step - fn: 23.0000 - fp: 12673.0000 - loss: 3.1681e-06 - precision: 0.0285 - recall: 0.9418 - tn: 214778.0000 - tp: 372.0000 - val_fn: 12.0000 - val_fp: 3416.0000 - val_loss: 0.7861 - val_precision: 0.0243 - val_recall: 0.8763 - val_tn: 53448.0000 - val_tp: 85.0000 ## Epoch 18/30 ## 112/112 - 0s - 2ms/step - fn: 23.0000 - fp: 11244.0000 - loss: 2.3250e-06 - precision: 0.0320 - recall: 0.9418 - tn: 216207.0000 - tp: 372.0000 - val_fn: 11.0000 - val_fp: 1717.0000 - val_loss: 0.1895 - val_precision: 0.0477 - val_recall: 0.8866 - val_tn: 55147.0000 - val_tp: 86.0000 ## Epoch 19/30 ## 112/112 - 0s - 2ms/step - fn: 16.0000 - fp: 8824.0000 - loss: 1.3502e-06 - precision: 0.0412 - recall: 0.9595 - tn: 218627.0000 - tp: 379.0000 - val_fn: 6.0000 - val_fp: 5161.0000 - val_loss: 0.1947 - val_precision: 0.0173 - val_recall: 0.9381 - val_tn: 51703.0000 - val_tp: 91.0000 ## Epoch 20/30 ## 112/112 - 0s - 2ms/step - fn: 10.0000 - fp: 8621.0000 - loss: 1.0916e-06 - precision: 0.0427 - recall: 0.9747 - tn: 218830.0000 - tp: 385.0000 - val_fn: 12.0000 - val_fp: 1107.0000 - val_loss: 0.1877 - val_precision: 0.0713 - val_recall: 0.8763 - val_tn: 55757.0000 - val_tp: 85.0000 ## Epoch 21/30 ## 112/112 - 0s - 2ms/step - fn: 10.0000 - fp: 7096.0000 - loss: 1.0528e-06 - precision: 0.0515 - recall: 0.9747 - tn: 220355.0000 - tp: 385.0000 - val_fn: 11.0000 - val_fp: 1099.0000 - val_loss: 0.0539 - val_precision: 0.0726 - val_recall: 0.8866 - val_tn: 55765.0000 - val_tp: 86.0000 ## Epoch 22/30 ## 112/112 - 0s - 3ms/step - fn: 11.0000 - fp: 5263.0000 - loss: 5.9445e-07 - precision: 0.0680 - recall: 0.9722 - tn: 222188.0000 - tp: 384.0000 - val_fn: 10.0000 - val_fp: 1399.0000 - val_loss: 0.0632 - val_precision: 0.0585 - val_recall: 0.8969 - val_tn: 55465.0000 - val_tp: 87.0000 ## Epoch 23/30 ## 112/112 - 0s - 4ms/step - fn: 8.0000 - fp: 5412.0000 - loss: 6.1044e-07 - precision: 0.0667 - recall: 0.9797 - tn: 222039.0000 - tp: 387.0000 - val_fn: 12.0000 - val_fp: 873.0000 - val_loss: 0.0727 - val_precision: 0.0887 - val_recall: 0.8763 - val_tn: 55991.0000 - val_tp: 85.0000 ## Epoch 24/30 ## 112/112 - 0s - 3ms/step - fn: 5.0000 - fp: 3469.0000 - loss: 4.1501e-07 - precision: 0.1011 - recall: 0.9873 - tn: 223982.0000 - tp: 390.0000 - val_fn: 11.0000 - val_fp: 1369.0000 - val_loss: 0.0758 - val_precision: 0.0591 - val_recall: 0.8866 - val_tn: 55495.0000 - val_tp: 86.0000 ## Epoch 25/30 ## 112/112 - 0s - 2ms/step - fn: 7.0000 - fp: 4489.0000 - loss: 6.9404e-07 - precision: 0.0796 - recall: 0.9823 - tn: 222962.0000 - tp: 388.0000 - val_fn: 11.0000 - val_fp: 1580.0000 - val_loss: 0.1499 - val_precision: 0.0516 - val_recall: 0.8866 - val_tn: 55284.0000 - val_tp: 86.0000 ## Epoch 26/30 ## 112/112 - 0s - 2ms/step - fn: 4.0000 - fp: 3469.0000 - loss: 4.4795e-07 - precision: 0.1013 - recall: 0.9899 - tn: 223982.0000 - tp: 391.0000 - val_fn: 12.0000 - val_fp: 951.0000 - val_loss: 0.0502 - val_precision: 0.0820 - val_recall: 0.8763 - val_tn: 55913.0000 - val_tp: 85.0000 ## Epoch 27/30 ## 112/112 - 0s - 2ms/step - fn: 5.0000 - fp: 4854.0000 - loss: 7.1714e-07 - precision: 0.0744 - recall: 0.9873 - tn: 222597.0000 - tp: 390.0000 - val_fn: 13.0000 - val_fp: 1354.0000 - val_loss: 0.0818 - val_precision: 0.0584 - val_recall: 0.8660 - val_tn: 55510.0000 - val_tp: 84.0000 ## Epoch 28/30 ## 112/112 - 0s - 3ms/step - fn: 6.0000 - fp: 4904.0000 - loss: 4.4107e-07 - precision: 0.0735 - recall: 0.9848 - tn: 222547.0000 - tp: 389.0000 - val_fn: 12.0000 - val_fp: 651.0000 - val_loss: 0.0420 - val_precision: 0.1155 - val_recall: 0.8763 - val_tn: 56213.0000 - val_tp: 85.0000 ## Epoch 29/30 ## 112/112 - 0s - 3ms/step - fn: 4.0000 - fp: 3909.0000 - loss: 4.2965e-07 - precision: 0.0909 - recall: 0.9899 - tn: 223542.0000 - tp: 391.0000 - val_fn: 12.0000 - val_fp: 817.0000 - val_loss: 0.0405 - val_precision: 0.0942 - val_recall: 0.8763 - val_tn: 56047.0000 - val_tp: 85.0000 ## Epoch 30/30 ## 112/112 - 0s - 2ms/step - fn: 2.0000 - fp: 2985.0000 - loss: 2.8492e-07 - precision: 0.1163 - recall: 0.9949 - tn: 224466.0000 - tp: 393.0000 - val_fn: 12.0000 - val_fp: 1040.0000 - val_loss: 0.0593 - val_precision: 0.0756 - val_recall: 0.8763 - val_tn: 55824.0000 - val_tp: 85.0000 val_pred <- model %>%   predict(val_features) %>%   { as.integer(. > 0.5) } ## 1781/1781 - 1s - 436us/step pred_correct <- val_df$Class == val_pred cat(sprintf(\"Validation accuracy: %.2f\", mean(pred_correct))) ## Validation accuracy: 0.98 fraudulent <- val_df$Class == 1  n_fraudulent_detected <- sum(fraudulent & pred_correct) n_fraudulent_missed <- sum(fraudulent & !pred_correct) n_legitimate_flagged <- sum(!fraudulent & !pred_correct)"},{"path":"https://keras3.posit.co/dev/articles/examples/structured_data/imbalanced_classification.html","id":"conclusions","dir":"Articles > Examples > Structured_data","previous_headings":"","what":"Conclusions","title":"Imbalanced classification: credit card fraud detection","text":"end training, 56,961 validation transactions, : Correctly identifying 85 fraudulent Missing 12 fraudulent transactions cost incorrectly flagging 1,040 legitimate transactions real world, one put even higher weight class 1, reflect False Negatives costly False Positives. Next time credit card gets declined online purchase – .","code":""},{"path":"https://keras3.posit.co/dev/articles/examples/structured_data/structured_data_classification_with_feature_space.html","id":"introduction","dir":"Articles > Examples > Structured_data","previous_headings":"","what":"Introduction","title":"Structured data classification with FeatureSpace","text":"example demonstrates structured data classification (also known tabular data classification), starting raw CSV file. data includes numerical features, integer categorical features, string categorical features. use utility layer_feature_space() index, preprocess, encode features. code adapted example Structured data classification scratch. previous example managed low-level feature preprocessing encoding Keras preprocessing layers, example delegate everything layer_feature_space(), making workflow extremely quick easy.","code":""},{"path":"https://keras3.posit.co/dev/articles/examples/structured_data/structured_data_classification_with_feature_space.html","id":"the-dataset","dir":"Articles > Examples > Structured_data","previous_headings":"Introduction","what":"The dataset","title":"Structured data classification with FeatureSpace","text":"dataset provided Cleveland Clinic Foundation Heart Disease. ’s CSV file 303 rows. row contains information patient (sample), column describes attribute patient (feature). use features predict whether patient heart disease (binary classification). ’s description feature:","code":""},{"path":"https://keras3.posit.co/dev/articles/examples/structured_data/structured_data_classification_with_feature_space.html","id":"setup","dir":"Articles > Examples > Structured_data","previous_headings":"","what":"Setup","title":"Structured data classification with FeatureSpace","text":"","code":"library(readr) library(dplyr, warn.conflicts = FALSE) library(keras3) library(tensorflow, exclude = c(\"shape\", \"set_random_seed\")) library(tfdatasets, exclude = \"shape\")  conflicted::conflicts_prefer(   keras3::shape(),   keras3::set_random_seed(),   dplyr::filter(),   .quiet = TRUE )  use_backend(\"tensorflow\")"},{"path":"https://keras3.posit.co/dev/articles/examples/structured_data/structured_data_classification_with_feature_space.html","id":"preparing-the-data","dir":"Articles > Examples > Structured_data","previous_headings":"","what":"Preparing the data","title":"Structured data classification with FeatureSpace","text":"Let’s download data load Pandas dataframe: dataset includes 303 samples 14 columns per sample (13 features, plus target label) ’s preview samples: last column, “target”, indicates whether patient heart disease (1) (0). Let’s split data training validation set: Let’s generate tf_dataset objects dataframe: tf_dataset yields tuple (input, target) input dictionary (named list) features target value 0 1: Let’s batch datasets:","code":"file_url <-   \"http://storage.googleapis.com/download.tensorflow.org/data/heart.csv\" df <- read_csv(file_url, col_types = cols(   oldpeak = col_double(),   thal = col_character(),   .default = col_integer() ))  # the dataset has two malformed rows, filter them out df <- df |> filter(!thal %in% c(\"1\", \"2\")) glimpse(df) ## Rows: 301 ## Columns: 14 ## $ age      <int> 63, 67, 67, 37, 41, 56, 62, 57, 63, 53, 57, 56, 56, 44, 5… ## $ sex      <int> 1, 1, 1, 1, 0, 1, 0, 0, 1, 1, 1, 0, 1, 1, 1, 1, 1, 1, 0, … ## $ cp       <int> 1, 4, 4, 3, 2, 2, 4, 4, 4, 4, 4, 2, 3, 2, 3, 3, 2, 4, 3, … ## $ trestbps <int> 145, 160, 120, 130, 130, 120, 140, 120, 130, 140, 140, 14… ## $ chol     <int> 233, 286, 229, 250, 204, 236, 268, 354, 254, 203, 192, 29… ## $ fbs      <int> 1, 0, 0, 0, 0, 0, 0, 0, 0, 1, 0, 0, 1, 0, 1, 0, 0, 0, 0, … ## $ restecg  <int> 2, 2, 2, 0, 2, 0, 2, 0, 2, 2, 0, 2, 2, 0, 0, 0, 0, 0, 0, … ## $ thalach  <int> 150, 108, 129, 187, 172, 178, 160, 163, 147, 155, 148, 15… ## $ exang    <int> 0, 1, 1, 0, 0, 0, 0, 1, 0, 1, 0, 0, 1, 0, 0, 0, 0, 0, 0, … ## $ oldpeak  <dbl> 2.3, 1.5, 2.6, 3.5, 1.4, 0.8, 3.6, 0.6, 1.4, 3.1, 0.4, 1.… ## $ slope    <int> 3, 2, 2, 3, 1, 1, 3, 1, 2, 3, 2, 2, 2, 1, 1, 1, 3, 1, 1, … ## $ ca       <int> 0, 3, 2, 0, 0, 0, 2, 0, 1, 0, 0, 0, 1, 0, 0, 0, 0, 0, 0, … ## $ thal     <chr> \"fixed\", \"normal\", \"reversible\", \"normal\", \"normal\", \"nor… ## $ target   <int> 0, 1, 0, 0, 0, 0, 1, 0, 1, 0, 0, 0, 1, 0, 0, 0, 0, 0, 0, … df ## # A tibble: 301 × 14 ##      age   sex    cp trestbps  chol   fbs restecg thalach exang oldpeak ##    <int> <int> <int>    <int> <int> <int>   <int>   <int> <int>   <dbl> ##  1    63     1     1      145   233     1       2     150     0     2.3 ##  2    67     1     4      160   286     0       2     108     1     1.5 ##  3    67     1     4      120   229     0       2     129     1     2.6 ##  4    37     1     3      130   250     0       0     187     0     3.5 ##  5    41     0     2      130   204     0       2     172     0     1.4 ##  6    56     1     2      120   236     0       0     178     0     0.8 ##  7    62     0     4      140   268     0       2     160     0     3.6 ##  8    57     0     4      120   354     0       0     163     1     0.6 ##  9    63     1     4      130   254     0       2     147     0     1.4 ## 10    53     1     4      140   203     1       2     155     1     3.1 ## # ℹ 291 more rows ## # ℹ 4 more variables: slope <int>, ca <int>, thal <chr>, target <int> val_idx <- nrow(df) %>% sample.int(., . * 0.2) val_df <- df[val_idx, ] train_df <- df[-val_idx, ]  cat(sprintf(   \"Using %d samples for training and %d for validation\",   nrow(train_df), nrow(val_df) )) ## Using 241 samples for training and 60 for validation dataframe_to_dataset <- function(df) {   labels <- df |> pull(target) |> as.integer()   inputs <- df |> select(-target) |> as.list()    ds <- tensor_slices_dataset(list(inputs, labels)) |>     dataset_shuffle(nrow(df))    ds }  train_ds <- dataframe_to_dataset(train_df) val_ds <- dataframe_to_dataset(val_df) c(x, y) %<-% iter_next(as_iterator(train_ds)) cat(\"Input: \"); str(x) cat(\"Target: \"); str(y) ## Input: List of 13 ##  $ age     :<tf.Tensor: shape=(), dtype=int32, numpy=63> ##  $ sex     :<tf.Tensor: shape=(), dtype=int32, numpy=0> ##  $ cp      :<tf.Tensor: shape=(), dtype=int32, numpy=4> ##  $ trestbps:<tf.Tensor: shape=(), dtype=int32, numpy=124> ##  $ chol    :<tf.Tensor: shape=(), dtype=int32, numpy=197> ##  $ fbs     :<tf.Tensor: shape=(), dtype=int32, numpy=0> ##  $ restecg :<tf.Tensor: shape=(), dtype=int32, numpy=0> ##  $ thalach :<tf.Tensor: shape=(), dtype=int32, numpy=136> ##  $ exang   :<tf.Tensor: shape=(), dtype=int32, numpy=1> ##  $ oldpeak :<tf.Tensor: shape=(), dtype=float32, numpy=0.0> ##  $ slope   :<tf.Tensor: shape=(), dtype=int32, numpy=2> ##  $ ca      :<tf.Tensor: shape=(), dtype=int32, numpy=0> ##  $ thal    :<tf.Tensor: shape=(), dtype=string, numpy=b'normal'> ## Target: <tf.Tensor: shape=(), dtype=int32, numpy=0> train_ds <- train_ds |> dataset_batch(32) val_ds <- val_ds |> dataset_batch(32)"},{"path":"https://keras3.posit.co/dev/articles/examples/structured_data/structured_data_classification_with_feature_space.html","id":"configuring-a-featurespace","dir":"Articles > Examples > Structured_data","previous_headings":"","what":"Configuring a FeatureSpace","title":"Structured data classification with FeatureSpace","text":"configure feature preprocessed, instantiate layer_feature_space(), pass dictionary (named list unique names) maps name features string describes feature type. “integer categorical” features \"FBS\", one “string categorical” feature (\"thal\"), numerical features, ’d like normalize – except \"age\", ’d like discretize number bins. also use crosses argument capture feature interactions categorical features, say, create additional features represent value co-occurrences categorical features. can compute feature crosses like arbitrary sets categorical features – just tuples two features. resulting co-occurences hashed fixed-sized vector, don’t need worry whether co-occurence space large.","code":"feature_space <- layer_feature_space(   features = list(     # Categorical features encoded as integers     sex = \"integer_categorical\",     cp = \"integer_categorical\",     fbs = \"integer_categorical\",     restecg = \"integer_categorical\",     exang = \"integer_categorical\",     ca = \"integer_categorical\",     # Categorical feature encoded as string     thal = \"string_categorical\",     # Numerical features to discretize     age = \"float_discretized\",     # Numerical features to normalize     trestbps = \"float_normalized\",     chol = \"float_normalized\",     thalach = \"float_normalized\",     oldpeak = \"float_normalized\",     slope = \"float_normalized\"   ),   # We create additional features by hashing   # value co-occurrences for the   # following groups of categorical features.   crosses = list(c(\"sex\", \"age\"), c(\"thal\", \"ca\")),   # The hashing space for these co-occurrences   # wil be 32-dimensional.   crossing_dim = 32,   # Our utility will one-hot encode all categorical   # features and concat all features into a single   # vector (one vector per sample).   output_mode = \"concat\" )"},{"path":"https://keras3.posit.co/dev/articles/examples/structured_data/structured_data_classification_with_feature_space.html","id":"further-customizing-a-featurespace","dir":"Articles > Examples > Structured_data","previous_headings":"","what":"Further customizing a FeatureSpace","title":"Structured data classification with FeatureSpace","text":"Specifying feature type via string name quick easy, sometimes may want configure preprocessing feature. instance, case, categorical features don’t large set possible values – ’s handful values per feature (e.g. 1 0 feature \"FBS\"), possible values represented training set. result, don’t need reserve index represent “vocabulary” values features – default behavior. , just specify num_oov_indices=0 features tell feature preprocessor skip “vocabulary” indexing. customizations access include specifying number bins discretizing features type \"float_discretized\", dimensionality hashing space feature crossing.","code":"feature_space <- layer_feature_space(   features = list(     # Categorical features encoded as integers     sex       = feature_integer_categorical(num_oov_indices = 0),     cp        = feature_integer_categorical(num_oov_indices = 0),     fbs       = feature_integer_categorical(num_oov_indices = 0),     restecg   = feature_integer_categorical(num_oov_indices = 0),     exang     = feature_integer_categorical(num_oov_indices = 0),     ca        = feature_integer_categorical(num_oov_indices = 0),     # Categorical feature encoded as string     thal      = feature_string_categorical(num_oov_indices = 0),     # Numerical features to discretize     age       = feature_float_discretized(num_bins = 30),     # Numerical features to normalize     trestbps  = feature_float_normalized(),     chol      = feature_float_normalized(),     thalach   = feature_float_normalized(),     oldpeak   = feature_float_normalized(),     slope     = feature_float_normalized()   ),   # Specify feature cross with a custom crossing dim.   crosses = list(     feature_cross(       feature_names = c(\"sex\", \"age\"),       crossing_dim = 64     ),     feature_cross(       feature_names = c(\"thal\", \"ca\"),       crossing_dim = 16     )   ),   output_mode = \"concat\" )"},{"path":"https://keras3.posit.co/dev/articles/examples/structured_data/structured_data_classification_with_feature_space.html","id":"adapt-the-featurespace-to-the-training-data","dir":"Articles > Examples > Structured_data","previous_headings":"","what":"Adapt the FeatureSpace to the training data","title":"Structured data classification with FeatureSpace","text":"start using FeatureSpace build model, adapt training data. adapt(), FeatureSpace : Index set possible values categorical features. Compute mean variance numerical features normalize. Compute value boundaries different bins numerical features discretize. Note adapt() called tf_dataset yields dicts (named lists) feature values – labels. point, FeatureSpace can called dict raw feature values, return single concatenate vector sample, combining encoded features feature crosses.","code":"train_ds_with_no_labels <- train_ds |> dataset_map(\\(x, y) x) feature_space |> adapt(train_ds_with_no_labels) c(x, y) %<-% iter_next(as_iterator(train_ds)) preprocessed_x <- feature_space(x) preprocessed_x ## tf.Tensor( ## [[0. 0. 0. ... 0. 0. 0.] ##  [0. 0. 0. ... 0. 0. 0.] ##  [0. 0. 0. ... 0. 0. 0.] ##  ... ##  [0. 0. 0. ... 0. 0. 0.] ##  [0. 0. 0. ... 0. 0. 0.] ##  [0. 0. 0. ... 1. 0. 0.]], shape=(32, 136), dtype=float32)"},{"path":"https://keras3.posit.co/dev/articles/examples/structured_data/structured_data_classification_with_feature_space.html","id":"two-ways-to-manage-preprocessing-as-part-of-the-tf-data-pipeline-or-in-the-model-itself","dir":"Articles > Examples > Structured_data","previous_headings":"","what":"Two ways to manage preprocessing: as part of the tf.data pipeline, or in the model itself","title":"Structured data classification with FeatureSpace","text":"two ways can leverage FeatureSpace:","code":""},{"path":"https://keras3.posit.co/dev/articles/examples/structured_data/structured_data_classification_with_feature_space.html","id":"asynchronous-preprocessing-in-tf-data","dir":"Articles > Examples > Structured_data","previous_headings":"Two ways to manage preprocessing: as part of the tf.data pipeline, or in the model itself","what":"Asynchronous preprocessing in tf.data","title":"Structured data classification with FeatureSpace","text":"can make part data pipeline, model. enables asynchronous parallel preprocessing data CPU hits model. ’re training GPU TPU, want speed preprocessing. Usually, always right thing training.","code":""},{"path":"https://keras3.posit.co/dev/articles/examples/structured_data/structured_data_classification_with_feature_space.html","id":"synchronous-preprocessing-in-the-model","dir":"Articles > Examples > Structured_data","previous_headings":"Two ways to manage preprocessing: as part of the tf.data pipeline, or in the model itself","what":"Synchronous preprocessing in the model","title":"Structured data classification with FeatureSpace","text":"can make part model. means model expect dicts raw feature values, preprocessing batch done synchronously (blocking manner) rest forward pass. want end--end model can process raw feature values – keep mind model able run CPU, since types feature preprocessing (e.g. string preprocessing) GPU TPU compatible. GPU / TPU performance-sensitive settings. general, want -model preprocessing inference CPU. case, apply FeatureSpace tf.data pipeline training, inference end--end model includes FeatureSpace. Let’s create training validation dataset preprocessed batches:","code":"preprocessed_train_ds <- train_ds |>   dataset_map(\\(x, y) list(feature_space(x), y),               num_parallel_calls = tf$data$AUTOTUNE) |>   dataset_prefetch(tf$data$AUTOTUNE)  preprocessed_val_ds <- val_ds |>   dataset_map(\\(x, y) list(feature_space(x), y),               num_parallel_calls = tf$data$AUTOTUNE) |>   dataset_prefetch(tf$data$AUTOTUNE)"},{"path":"https://keras3.posit.co/dev/articles/examples/structured_data/structured_data_classification_with_feature_space.html","id":"build-a-model","dir":"Articles > Examples > Structured_data","previous_headings":"","what":"Build a model","title":"Structured data classification with FeatureSpace","text":"Time build model – rather two models: training model expects preprocessed features (one sample = one vector) inference model expects raw features (one sample = dict raw feature values)","code":"dict_inputs <- feature_space$get_inputs() encoded_features <- feature_space$get_encoded_features()  predictions <- encoded_features |>   layer_dense(32, activation=\"relu\") |>   layer_dropout(0.5) |>   layer_dense(1, activation=\"sigmoid\")  training_model <- keras_model(inputs = encoded_features,                               outputs = predictions) training_model |> compile(optimizer = \"adam\",                           loss = \"binary_crossentropy\",                           metrics = \"accuracy\")  inference_model <- keras_model(inputs = dict_inputs,                                outputs = predictions)"},{"path":"https://keras3.posit.co/dev/articles/examples/structured_data/structured_data_classification_with_feature_space.html","id":"train-the-model","dir":"Articles > Examples > Structured_data","previous_headings":"","what":"Train the model","title":"Structured data classification with FeatureSpace","text":"Let’s train model 20 epochs. Note feature preprocessing happening part tf.data pipeline, part model. quickly get 80% validation accuracy.","code":"training_model |> fit(   preprocessed_train_ds,   epochs = 20,   validation_data = preprocessed_val_ds,   verbose = 2 ) ## Epoch 1/20 ## 8/8 - 3s - 325ms/step - accuracy: 0.4481 - loss: 0.7302 - val_accuracy: 0.6000 - val_loss: 0.6799 ## Epoch 2/20 ## 8/8 - 0s - 28ms/step - accuracy: 0.5851 - loss: 0.6797 - val_accuracy: 0.7000 - val_loss: 0.6329 ## Epoch 3/20 ## 8/8 - 0s - 31ms/step - accuracy: 0.6141 - loss: 0.6458 - val_accuracy: 0.7333 - val_loss: 0.5951 ## Epoch 4/20 ## 8/8 - 0s - 32ms/step - accuracy: 0.7220 - loss: 0.5843 - val_accuracy: 0.7667 - val_loss: 0.5644 ## Epoch 5/20 ## 8/8 - 0s - 32ms/step - accuracy: 0.7552 - loss: 0.5539 - val_accuracy: 0.7667 - val_loss: 0.5367 ## Epoch 6/20 ## 8/8 - 0s - 29ms/step - accuracy: 0.7884 - loss: 0.5248 - val_accuracy: 0.7667 - val_loss: 0.5146 ## Epoch 7/20 ## 8/8 - 0s - 32ms/step - accuracy: 0.7469 - loss: 0.5084 - val_accuracy: 0.7667 - val_loss: 0.4960 ## Epoch 8/20 ## 8/8 - 0s - 30ms/step - accuracy: 0.7718 - loss: 0.4921 - val_accuracy: 0.7833 - val_loss: 0.4804 ## Epoch 9/20 ## 8/8 - 0s - 30ms/step - accuracy: 0.8216 - loss: 0.4617 - val_accuracy: 0.8000 - val_loss: 0.4668 ## Epoch 10/20 ## 8/8 - 0s - 30ms/step - accuracy: 0.8008 - loss: 0.4550 - val_accuracy: 0.8333 - val_loss: 0.4551 ## Epoch 11/20 ## 8/8 - 0s - 30ms/step - accuracy: 0.8133 - loss: 0.4299 - val_accuracy: 0.8500 - val_loss: 0.4447 ## Epoch 12/20 ## 8/8 - 0s - 29ms/step - accuracy: 0.8506 - loss: 0.3980 - val_accuracy: 0.8500 - val_loss: 0.4372 ## Epoch 13/20 ## 8/8 - 0s - 29ms/step - accuracy: 0.8008 - loss: 0.4105 - val_accuracy: 0.8500 - val_loss: 0.4308 ## Epoch 14/20 ## 8/8 - 0s - 29ms/step - accuracy: 0.8548 - loss: 0.3820 - val_accuracy: 0.8500 - val_loss: 0.4254 ## Epoch 15/20 ## 8/8 - 0s - 31ms/step - accuracy: 0.8589 - loss: 0.3669 - val_accuracy: 0.8500 - val_loss: 0.4205 ## Epoch 16/20 ## 8/8 - 0s - 31ms/step - accuracy: 0.8257 - loss: 0.3786 - val_accuracy: 0.8333 - val_loss: 0.4142 ## Epoch 17/20 ## 8/8 - 0s - 30ms/step - accuracy: 0.8548 - loss: 0.3524 - val_accuracy: 0.8333 - val_loss: 0.4105 ## Epoch 18/20 ## 8/8 - 0s - 31ms/step - accuracy: 0.8631 - loss: 0.3530 - val_accuracy: 0.8333 - val_loss: 0.4078 ## Epoch 19/20 ## 8/8 - 0s - 30ms/step - accuracy: 0.8548 - loss: 0.3440 - val_accuracy: 0.8333 - val_loss: 0.4044 ## Epoch 20/20 ## 8/8 - 0s - 30ms/step - accuracy: 0.8838 - loss: 0.3303 - val_accuracy: 0.8333 - val_loss: 0.4021"},{"path":"https://keras3.posit.co/dev/articles/examples/structured_data/structured_data_classification_with_feature_space.html","id":"inference-on-new-data-with-the-end-to-end-model","dir":"Articles > Examples > Structured_data","previous_headings":"","what":"Inference on new data with the end-to-end model","title":"Structured data classification with FeatureSpace","text":"Now, can use inference model (includes FeatureSpace) make predictions based dicts raw features values, follows:","code":"sample <- list(   age = 60,   sex = 1,   cp = 1,   trestbps = 145,   chol = 233,   fbs = 1,   restecg = 2,   thalach = 150,   exang = 0,   oldpeak = 2.3,   slope = 3,   ca = 0,   thal = \"fixed\" )  input_dict <- lapply(sample, \\(x) op_convert_to_tensor(array(x))) predictions <- inference_model |> predict(input_dict) ## 1/1 - 0s - 368ms/step glue::glue(r\"---(   This particular patient had a {(100 * predictions) |> signif(3)}% probability   of having a heart disease, as evaluated by our model. )---\") ## This particular patient had a 48.8% probability ## of having a heart disease, as evaluated by our model."},{"path":"https://keras3.posit.co/dev/articles/examples/timeseries/timeseries_anomaly_detection.html","id":"introduction","dir":"Articles > Examples > Timeseries","previous_headings":"","what":"Introduction","title":"Timeseries anomaly detection using an Autoencoder","text":"script demonstrates can use reconstruction convolutional autoencoder model detect anomalies timeseries data.","code":""},{"path":"https://keras3.posit.co/dev/articles/examples/timeseries/timeseries_anomaly_detection.html","id":"setup","dir":"Articles > Examples > Timeseries","previous_headings":"","what":"Setup","title":"Timeseries anomaly detection using an Autoencoder","text":"","code":"library(dplyr, warn.conflicts = FALSE) library(ggplot2) theme_set(theme_minimal())  library(listarrays) library(tfdatasets, exclude = c(\"shape\")) library(keras3)"},{"path":"https://keras3.posit.co/dev/articles/examples/timeseries/timeseries_anomaly_detection.html","id":"load-the-data","dir":"Articles > Examples > Timeseries","previous_headings":"","what":"Load the data","title":"Timeseries anomaly detection using an Autoencoder","text":"use Numenta Anomaly Benchmark(NAB) dataset. provides artificial timeseries data containing labeled anomalous periods behavior. Data ordered, timestamped, single-valued metrics. use art_daily_small_noise.csv file training art_daily_jumpsup.csv file testing. simplicity dataset allows us demonstrate anomaly detection.","code":"get_data <- function(url_suffix) {   url_root <- \"https://raw.githubusercontent.com/numenta/NAB/master/data/\"   url <- paste0(url_root, url_suffix)   file <- get_file(origin = url) # cache file locally   # parse csv; 2 columns with types: datetime (T), double (d)   readr::read_csv(file, col_types = \"Td\") }  df_small_noise   <- get_data(\"artificialNoAnomaly/art_daily_small_noise.csv\") ## Downloading data from https://raw.githubusercontent.com/numenta/NAB/master/data/artificialNoAnomaly/art_daily_small_noise.csv ## 
[1m     0/140187
[0m 
[37m━━━━━━━━━━━━━━━━━━━━
[0m 
[1m0s
[0m 0s/step 
[1m122880/140187
[0m 
[32m━━━━━━━━━━━━━━━━━
[0m
[37m━━━
[0m 
[1m0s
[0m 0us/step 
[1m140187/140187
[0m 
[32m━━━━━━━━━━━━━━━━━━━━
[0m
[37m
[0m 
[1m0s
[0m 0us/step df_daily_jumpsup <- get_data(\"artificialWithAnomaly/art_daily_jumpsup.csv\") ## Downloading data from https://raw.githubusercontent.com/numenta/NAB/master/data/artificialWithAnomaly/art_daily_jumpsup.csv ## 
[1m     0/140548
[0m 
[37m━━━━━━━━━━━━━━━━━━━━
[0m 
[1m0s
[0m 0s/step 
[1m114688/140548
[0m 
[32m━━━━━━━━━━━━━━━━
[0m
[37m━━━━
[0m 
[1m0s
[0m 0us/step 
[1m140548/140548
[0m 
[32m━━━━━━━━━━━━━━━━━━━━
[0m
[37m
[0m 
[1m0s
[0m 0us/step"},{"path":"https://keras3.posit.co/dev/articles/examples/timeseries/timeseries_anomaly_detection.html","id":"quick-look-at-the-data","dir":"Articles > Examples > Timeseries","previous_headings":"","what":"Quick look at the data","title":"Timeseries anomaly detection using an Autoencoder","text":"","code":"df_small_noise ## # A tibble: 4,032 × 2 ##    timestamp           value ##    <dttm>              <dbl> ##  1 2014-04-01 00:00:00  18.3 ##  2 2014-04-01 00:05:00  22.0 ##  3 2014-04-01 00:10:00  18.6 ##  4 2014-04-01 00:15:00  22.0 ##  5 2014-04-01 00:20:00  21.9 ##  6 2014-04-01 00:25:00  21.2 ##  7 2014-04-01 00:30:00  20.6 ##  8 2014-04-01 00:35:00  20.3 ##  9 2014-04-01 00:40:00  21.5 ## 10 2014-04-01 00:45:00  19.2 ## # ℹ 4,022 more rows df_daily_jumpsup ## # A tibble: 4,032 × 2 ##    timestamp           value ##    <dttm>              <dbl> ##  1 2014-04-01 00:00:00  19.8 ##  2 2014-04-01 00:05:00  20.5 ##  3 2014-04-01 00:10:00  20.0 ##  4 2014-04-01 00:15:00  21.5 ##  5 2014-04-01 00:20:00  20.2 ##  6 2014-04-01 00:25:00  19.9 ##  7 2014-04-01 00:30:00  21.7 ##  8 2014-04-01 00:35:00  20.9 ##  9 2014-04-01 00:40:00  18.4 ## 10 2014-04-01 00:45:00  18.7 ## # ℹ 4,022 more rows"},{"path":[]},{"path":"https://keras3.posit.co/dev/articles/examples/timeseries/timeseries_anomaly_detection.html","id":"timeseries-data-without-anomalies","dir":"Articles > Examples > Timeseries","previous_headings":"Visualize the data","what":"Timeseries data without anomalies","title":"Timeseries anomaly detection using an Autoencoder","text":"use following data training.","code":"plot_ts <- function(df) {   ggplot(df, aes(x = timestamp, y = value)) + geom_line() +     scale_x_datetime(date_breaks = \"1 day\", date_labels = \"%b-%d\") }  plot_ts(df_small_noise) + ggtitle(\"Without Anomaly\")"},{"path":"https://keras3.posit.co/dev/articles/examples/timeseries/timeseries_anomaly_detection.html","id":"timeseries-data-with-anomalies","dir":"Articles > Examples > Timeseries","previous_headings":"Visualize the data","what":"Timeseries data with anomalies","title":"Timeseries anomaly detection using an Autoencoder","text":"use following data testing see sudden jump data detected anomaly.","code":"plot_ts(df_daily_jumpsup) + ggtitle(\"With Anomaly\")"},{"path":"https://keras3.posit.co/dev/articles/examples/timeseries/timeseries_anomaly_detection.html","id":"prepare-training-data","dir":"Articles > Examples > Timeseries","previous_headings":"","what":"Prepare training data","title":"Timeseries anomaly detection using an Autoencoder","text":"Get data values training timeseries data file normalize value data. value every 5 mins 14 days. 24 * 60 / 5 = 288 timesteps per day 288 * 14 = 4032 data points total","code":"df_train <- df_small_noise |>   mutate(value = (value - mean(value)) / sd(value))  cat(\"Number of training samples:\", nrow(df_train), \"\\n\") ## Number of training samples: 4032"},{"path":"https://keras3.posit.co/dev/articles/examples/timeseries/timeseries_anomaly_detection.html","id":"create-sequences","dir":"Articles > Examples > Timeseries","previous_headings":"Prepare training data","what":"Create sequences","title":"Timeseries anomaly detection using an Autoencoder","text":"Create sequences combining TIME_STEPS contiguous data values training data.","code":"TIME_STEPS <- 288  as_dataset <- function(df) {   x <- as.matrix(df$value)   ds <- timeseries_dataset_from_array(x, NULL, sequence_length = TIME_STEPS)   # Because the dataset is small, cast TF Dataset to an R array for convenience.   ds |> as_array_iterator() |> iterate() |> bind_on_rows() }  x_train <- as_dataset(df_train) writeLines(sprintf(\"Training input shape: (%s)\", toString(dim(x_train)))) ## Training input shape: (3745, 288, 1)"},{"path":"https://keras3.posit.co/dev/articles/examples/timeseries/timeseries_anomaly_detection.html","id":"build-a-model","dir":"Articles > Examples > Timeseries","previous_headings":"","what":"Build a model","title":"Timeseries anomaly detection using an Autoencoder","text":"build convolutional reconstruction autoencoder model. model take input shape (batch_size, sequence_length, num_features) return output shape. case, sequence_length 288 num_features 1.","code":"model <- keras_model_sequential(input_shape = c(TIME_STEPS, 1)) |>   layer_conv_1d(     filters = 32, kernel_size = 7, padding = \"same\",     strides = 2, activation = \"relu\"   ) |>   layer_dropout(rate = 0.2) |>   layer_conv_1d(     filters = 16, kernel_size = 7, padding = \"same\",     strides = 2, activation = \"relu\"   ) |>   layer_conv_1d_transpose(     filters = 16, kernel_size = 7, padding = \"same\",     strides = 2, activation = \"relu\"   ) |>   layer_dropout(rate = 0.2) |>   layer_conv_1d_transpose(     filters = 32, kernel_size = 7, padding = \"same\",     strides = 2, activation = \"relu\"   ) |>   layer_conv_1d_transpose(filters = 1, kernel_size = 7, padding = \"same\")  model |> compile(optimizer=optimizer_adam(learning_rate=0.001), loss=\"mse\") model ## Model: \"sequential\" ## ┏━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━┓ ## ┃ Layer (type)                    ┃ Output Shape           ┃       Param # ┃ ## ┡━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━┩ ## │ conv1d (Conv1D)                 │ (None, 144, 32)        │           256 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ dropout (Dropout)               │ (None, 144, 32)        │             0 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ conv1d_1 (Conv1D)               │ (None, 72, 16)         │         3,600 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ conv1d_transpose                │ (None, 144, 16)        │         1,808 │ ## │ (Conv1DTranspose)               │                        │               │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ dropout_1 (Dropout)             │ (None, 144, 16)        │             0 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ conv1d_transpose_1              │ (None, 288, 32)        │         3,616 │ ## │ (Conv1DTranspose)               │                        │               │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ conv1d_transpose_2              │ (None, 288, 1)         │           225 │ ## │ (Conv1DTranspose)               │                        │               │ ## └─────────────────────────────────┴────────────────────────┴───────────────┘ ##  Total params: 9,505 (37.13 KB) ##  Trainable params: 9,505 (37.13 KB) ##  Non-trainable params: 0 (0.00 B)"},{"path":"https://keras3.posit.co/dev/articles/examples/timeseries/timeseries_anomaly_detection.html","id":"train-the-model","dir":"Articles > Examples > Timeseries","previous_headings":"","what":"Train the model","title":"Timeseries anomaly detection using an Autoencoder","text":"Please note using x_train input target since reconstruction model. Let’s plot training validation loss see training went.","code":"history = model |> fit(   x_train, x_train,   epochs = 50,   validation_split = 0.1,   callbacks = c(     callback_early_stopping(       monitor = \"val_loss\", patience = 5, mode = \"min\"     )   ) ) ## Epoch 1/50 ## 106/106 - 5s - 48ms/step - loss: 0.1863 - val_loss: 0.0316 ## Epoch 2/50 ## 106/106 - 0s - 2ms/step - loss: 0.0333 - val_loss: 0.0233 ## Epoch 3/50 ## 106/106 - 0s - 2ms/step - loss: 0.0248 - val_loss: 0.0201 ## Epoch 4/50 ## 106/106 - 0s - 3ms/step - loss: 0.0209 - val_loss: 0.0187 ## Epoch 5/50 ## 106/106 - 0s - 3ms/step - loss: 0.0179 - val_loss: 0.0146 ## Epoch 6/50 ## 106/106 - 0s - 3ms/step - loss: 0.0150 - val_loss: 0.0113 ## Epoch 7/50 ## 106/106 - 0s - 3ms/step - loss: 0.0127 - val_loss: 0.0094 ## Epoch 8/50 ## 106/106 - 0s - 3ms/step - loss: 0.0109 - val_loss: 0.0088 ## Epoch 9/50 ## 106/106 - 0s - 3ms/step - loss: 0.0096 - val_loss: 0.0084 ## Epoch 10/50 ## 106/106 - 0s - 3ms/step - loss: 0.0086 - val_loss: 0.0070 ## Epoch 11/50 ## 106/106 - 0s - 3ms/step - loss: 0.0078 - val_loss: 0.0061 ## Epoch 12/50 ## 106/106 - 0s - 2ms/step - loss: 0.0073 - val_loss: 0.0058 ## Epoch 13/50 ## 106/106 - 0s - 3ms/step - loss: 0.0068 - val_loss: 0.0053 ## Epoch 14/50 ## 106/106 - 0s - 3ms/step - loss: 0.0064 - val_loss: 0.0050 ## Epoch 15/50 ## 106/106 - 0s - 3ms/step - loss: 0.0061 - val_loss: 0.0046 ## Epoch 16/50 ## 106/106 - 0s - 4ms/step - loss: 0.0058 - val_loss: 0.0040 ## Epoch 17/50 ## 106/106 - 0s - 3ms/step - loss: 0.0056 - val_loss: 0.0041 ## Epoch 18/50 ## 106/106 - 0s - 3ms/step - loss: 0.0053 - val_loss: 0.0036 ## Epoch 19/50 ## 106/106 - 0s - 3ms/step - loss: 0.0050 - val_loss: 0.0038 ## Epoch 20/50 ## 106/106 - 0s - 3ms/step - loss: 0.0048 - val_loss: 0.0032 ## Epoch 21/50 ## 106/106 - 0s - 3ms/step - loss: 0.0046 - val_loss: 0.0032 ## Epoch 22/50 ## 106/106 - 0s - 3ms/step - loss: 0.0044 - val_loss: 0.0032 ## Epoch 23/50 ## 106/106 - 0s - 3ms/step - loss: 0.0042 - val_loss: 0.0033 ## Epoch 24/50 ## 106/106 - 0s - 3ms/step - loss: 0.0041 - val_loss: 0.0031 ## Epoch 25/50 ## 106/106 - 0s - 3ms/step - loss: 0.0039 - val_loss: 0.0033 ## Epoch 26/50 ## 106/106 - 0s - 3ms/step - loss: 0.0037 - val_loss: 0.0031 ## Epoch 27/50 ## 106/106 - 0s - 3ms/step - loss: 0.0035 - val_loss: 0.0032 ## Epoch 28/50 ## 106/106 - 0s - 2ms/step - loss: 0.0034 - val_loss: 0.0026 ## Epoch 29/50 ## 106/106 - 0s - 2ms/step - loss: 0.0033 - val_loss: 0.0028 ## Epoch 30/50 ## 106/106 - 0s - 2ms/step - loss: 0.0032 - val_loss: 0.0030 ## Epoch 31/50 ## 106/106 - 0s - 2ms/step - loss: 0.0031 - val_loss: 0.0024 ## Epoch 32/50 ## 106/106 - 0s - 2ms/step - loss: 0.0030 - val_loss: 0.0028 ## Epoch 33/50 ## 106/106 - 0s - 2ms/step - loss: 0.0029 - val_loss: 0.0027 ## Epoch 34/50 ## 106/106 - 0s - 3ms/step - loss: 0.0028 - val_loss: 0.0023 ## Epoch 35/50 ## 106/106 - 0s - 2ms/step - loss: 0.0027 - val_loss: 0.0024 ## Epoch 36/50 ## 106/106 - 0s - 2ms/step - loss: 0.0027 - val_loss: 0.0025 ## Epoch 37/50 ## 106/106 - 0s - 2ms/step - loss: 0.0026 - val_loss: 0.0026 ## Epoch 38/50 ## 106/106 - 0s - 2ms/step - loss: 0.0026 - val_loss: 0.0024 ## Epoch 39/50 ## 106/106 - 0s - 3ms/step - loss: 0.0025 - val_loss: 0.0023 plot(history)"},{"path":"https://keras3.posit.co/dev/articles/examples/timeseries/timeseries_anomaly_detection.html","id":"detecting-anomalies","dir":"Articles > Examples > Timeseries","previous_headings":"","what":"Detecting anomalies","title":"Timeseries anomaly detection using an Autoencoder","text":"detect anomalies determining well model can reconstruct input data. Find MAE loss training samples. Find max MAE loss value. worst model performed trying reconstruct sample. make threshold anomaly detection. reconstruction loss sample greater threshold value can infer model seeing pattern isn’t familiar . label sample anomaly.","code":"# Get train MAE loss. x_train_pred <- model |> predict(x_train) ## 118/118 - 1s - 5ms/step train_mae_loss <- apply(abs(x_train_pred - x_train), 1, mean)  hist(train_mae_loss, breaks = 50) # Get reconstruction loss threshold. threshold <- max(train_mae_loss) cat(\"Reconstruction error threshold: \", threshold, \"\\n\") ## Reconstruction error threshold:  0.03910884"},{"path":"https://keras3.posit.co/dev/articles/examples/timeseries/timeseries_anomaly_detection.html","id":"compare-recontruction","dir":"Articles > Examples > Timeseries","previous_headings":"Detecting anomalies","what":"Compare recontruction","title":"Timeseries anomaly detection using an Autoencoder","text":"Just fun, let’s see model recontructed first sample. 288 timesteps day 1 training dataset.","code":"# Checking how the first sequence is learnt plot(NULL, NULL, ylab = 'Value',      xlim = c(0, TIME_STEPS),      ylim = range(c(x_train[1,,], x_train_pred[1,,]))) lines(x_train[1,,]) lines(x_train_pred[1,,], col = 'red') legend(\"topleft\", lty = 1,        legend = c(\"actual\", \"predicted\"),        col = c(\"black\", \"red\"))"},{"path":"https://keras3.posit.co/dev/articles/examples/timeseries/timeseries_anomaly_detection.html","id":"prepare-test-data","dir":"Articles > Examples > Timeseries","previous_headings":"Detecting anomalies","what":"Prepare test data","title":"Timeseries anomaly detection using an Autoencoder","text":"","code":"df_test <- df_daily_jumpsup |>   mutate(value =            (value - mean(df_small_noise$value)) /              sd(df_small_noise$value))  df_test |> head() plot_ts(df_test) # Create sequences from test values. x_test <- as_dataset(df_test)  # Get test MAE loss. x_test_pred <- model |> predict(x_test) test_mae_loss <- apply(abs(x_test_pred - x_test), 1, mean)  hist(test_mae_loss, breaks = 50, xlab = \"test MAE loss\", ylab = \"No of samples\") # Detect all the samples which are anomalies. anomalies <- test_mae_loss > threshold cat(\"Number of anomaly samples:\", sum(anomalies), \"\\n\") cat(\"Indices of anomaly samples:\", which(anomalies), \"\\n\", fill = TRUE) ## # A tibble: 6 × 2 ##   timestamp            value ##   <dttm>               <dbl> ## 1 2014-04-01 00:00:00 -0.808 ## 2 2014-04-01 00:05:00 -0.781 ## 3 2014-04-01 00:10:00 -0.801 ## 4 2014-04-01 00:15:00 -0.746 ## 5 2014-04-01 00:20:00 -0.792 ## 6 2014-04-01 00:25:00 -0.802 ## 118/118 - 0s - 1ms/step ## Number of anomaly samples: 483 ## Indices of anomaly samples: 216 217 218 219 220 221 396 398 507 793 794 795 ## 797 798 799 974 1658 1659 1944 1945 1946 1947 2042 2046 2050 2052 2054 2056 ## 2058 2060 2061 2062 2064 2065 2066 2068 2069 2070 2072 2073 2074 2076 2077 ## 2078 2080 2081 2082 2084 2085 2086 2088 2089 2090 2092 2093 2094 2096 2097 ## 2098 2100 2101 2102 2104 2108 2110 2112 2114 2116 2118 2120 2121 2122 2124 ## 2126 2141 2521 2522 2523 2525 2546 2698 2700 2702 2703 2704 2705 2706 2707 ## 2708 2709 2710 2711 2712 2713 2714 2715 2716 2717 2718 2719 2720 2721 2722 ## 2723 2724 2725 2726 2727 2728 2729 2730 2731 2732 2733 2734 2735 2736 2737 ## 2738 2739 2740 2741 2742 2743 2744 2745 2746 2747 2748 2749 2750 2751 2752 ## 2753 2754 2755 2756 2757 2758 2759 2760 2761 2762 2763 2764 2765 2766 2767 ## 2768 2769 2770 2771 2772 2773 2774 2775 2776 2777 2778 2779 2780 2781 2782 ## 2783 2784 2785 2786 2787 2788 2789 2790 2791 2792 2793 2794 2795 2796 2797 ## 2798 2799 2800 2801 2802 2803 2804 2805 2806 2807 2808 2809 2810 2811 2812 ## 2813 2814 2815 2816 2817 2818 2819 2820 2821 2822 2823 2824 2825 2826 2827 ## 2828 2829 2830 2831 2832 2833 2834 2835 2836 2837 2838 2839 2840 2841 2842 ## 2843 2844 2845 2846 2847 2848 2849 2850 2851 2852 2853 2854 2855 2856 2857 ## 2858 2859 2860 2861 2862 2863 2864 2865 2866 2867 2868 2869 2870 2871 2872 ## 2873 2874 2875 2876 2877 2878 2879 2880 2881 2882 2883 2884 2885 2886 2887 ## 2888 2889 2890 2891 2892 2893 2894 2895 2896 2897 2898 2899 2900 2901 2902 ## 2903 2904 2905 2906 2907 2908 2909 2910 2911 2912 2913 2914 2915 2916 2917 ## 2918 2919 2920 2921 2922 2923 2924 2925 2926 2927 2928 2929 2930 2931 2932 ## 2933 2934 2935 2936 2937 2938 2939 2940 2941 2942 2943 2944 2945 2946 2947 ## 2948 2949 2950 2951 2952 2953 2954 2955 2956 2957 2958 2959 2960 2961 2962 ## 2963 2964 2965 2966 2967 2968 2969 2970 2971 2972 2973 2974 2975 2976 2977 ## 2978 2979 2980 2981 2982 2983 2984 2985 2986 2987 2988 2989 2990 2991 2992 ## 2993 2994 2995 2996 2997 2998 2999 3000 3001 3002 3003 3004 3005 3006 3007 ## 3008 3009 3010 3011 3012 3013 3014 3015 3016 3017 3018 3019 3020 3021 3022 ## 3023 3024 3025 3026 3027 3028 3029 3030 3031 3032 3033 3034 3035 3036 3037 ## 3038 3039 3040 3041 3042 3043 3044 3045 3046 3047 3048 3049 3050 3051 3052 ## 3053 3054 3055 3056 3057 3058 3059 3060 3061 3062 3063 3064 3065 3066 3067 ## 3068 3069 3070 3071 3072 3073 3074 3075 3076 3077 3078 3079 3080 3081 3082 ## 3083 3084 3085 3086 3087 3088 3089 3090 3091 3092 3093 3094 3095 3096 3097 ## 3098 3099 3100 3101 3102"},{"path":"https://keras3.posit.co/dev/articles/examples/timeseries/timeseries_anomaly_detection.html","id":"plot-anomalies","dir":"Articles > Examples > Timeseries","previous_headings":"","what":"Plot anomalies","title":"Timeseries anomaly detection using an Autoencoder","text":"now know samples data anomalies. , find corresponding timestamps original test data. using following method : Let’s say time_steps = 3 10 training values. x_train look like : 0, 1, 2 1, 2, 3 2, 3, 4 3, 4, 5 4, 5, 6 5, 6, 7 6, 7, 8 7, 8, 9 except initial final time_steps-1 data values, appear time_steps number samples. , know samples [(3, 4, 5), (4, 5, 6), (5, 6, 7)] anomalies, can say data point 5 anomaly. Let’s overlay anomalies original test data plot.","code":"is_anomaly <- test_mae_loss > threshold is_anomaly <- is_anomaly &   zoo::rollsum(is_anomaly, TIME_STEPS,                align = \"right\", na.pad = TRUE) >= TIME_STEPS  with(df_test, {   plot(value ~ timestamp, type = 'l', xaxt = 'n', las = 2)   axis.POSIXct(1, at = seq(timestamp[1], tail(timestamp, 1), by = \"days\"),                format = \"%b-%d\") })  with(df_test[which(is_anomaly),], {   points(value ~ timestamp, col = \"red\") })"},{"path":"https://keras3.posit.co/dev/articles/examples/timeseries/timeseries_classification_from_scratch.html","id":"introduction","dir":"Articles > Examples > Timeseries","previous_headings":"","what":"Introduction","title":"Timeseries classification from scratch","text":"example shows timeseries classification scratch, starting raw CSV timeseries files disk. demonstrate workflow FordA dataset UCR/UEA archive.","code":""},{"path":"https://keras3.posit.co/dev/articles/examples/timeseries/timeseries_classification_from_scratch.html","id":"setup","dir":"Articles > Examples > Timeseries","previous_headings":"","what":"Setup","title":"Timeseries classification from scratch","text":"","code":"library(keras3) use_backend(\"jax\")"},{"path":[]},{"path":"https://keras3.posit.co/dev/articles/examples/timeseries/timeseries_classification_from_scratch.html","id":"dataset-description","dir":"Articles > Examples > Timeseries","previous_headings":"Load the data: the FordA dataset","what":"Dataset description","title":"Timeseries classification from scratch","text":"dataset using called FordA. data comes UCR archive. dataset contains 3601 training instances another 1320 testing instances. timeseries corresponds measurement engine noise captured motor sensor. task, goal automatically detect presence specific issue engine. problem balanced binary classification task. full description dataset can found .","code":""},{"path":"https://keras3.posit.co/dev/articles/examples/timeseries/timeseries_classification_from_scratch.html","id":"read-the-tsv-data","dir":"Articles > Examples > Timeseries","previous_headings":"Load the data: the FordA dataset","what":"Read the TSV data","title":"Timeseries classification from scratch","text":"use FordA_TRAIN file training FordA_TEST file testing. simplicity dataset allows us demonstrate effectively use ConvNets timeseries classification. file, first column corresponds label.","code":"get_data <- function(path) {   if(path |> startsWith(\"https://\"))     path <- get_file(origin = path)  # cache file locally    data <- readr::read_tsv(     path, col_names = FALSE,     # Each row is: one integer (the label),     # followed by 500 doubles (the timeseries)     col_types = paste0(\"i\", strrep(\"d\", 500))   )    y <- as.matrix(data[[1]])   x <- as.matrix(data[,-1])   dimnames(x) <- dimnames(y) <- NULL    list(x, y) }  root_url <- \"https://raw.githubusercontent.com/hfawaz/cd-diagram/master/FordA/\" c(x_train, y_train) %<-% get_data(paste0(root_url, \"FordA_TRAIN.tsv\")) ## Downloading data from https://raw.githubusercontent.com/hfawaz/cd-diagram/master/FordA/FordA_TRAIN.tsv ## 
[1m       0/20094049
[0m 
[37m━━━━━━━━━━━━━━━━━━━━
[0m 
[1m0s
[0m 0s/step 
[1m  114688/20094049
[0m 
[37m━━━━━━━━━━━━━━━━━━━━
[0m 
[1m11s
[0m 1us/step 
[1m  458752/20094049
[0m 
[37m━━━━━━━━━━━━━━━━━━━━
[0m 
[1m4s
[0m 0us/step  
[1m  925696/20094049
[0m 
[37m━━━━━━━━━━━━━━━━━━━━
[0m 
[1m3s
[0m 0us/step 
[1m 1458176/20094049
[0m 
[32m━
[0m
[37m━━━━━━━━━━━━━━━━━━━
[0m 
[1m2s
[0m 0us/step 
[1m 1867776/20094049
[0m 
[32m━
[0m
[37m━━━━━━━━━━━━━━━━━━━
[0m 
[1m2s
[0m 0us/step 
[1m 2375680/20094049
[0m 
[32m━━
[0m
[37m━━━━━━━━━━━━━━━━━━
[0m 
[1m2s
[0m 0us/step 
[1m 2834432/20094049
[0m 
[32m━━
[0m
[37m━━━━━━━━━━━━━━━━━━
[0m 
[1m2s
[0m 0us/step 
[1m 3407872/20094049
[0m 
[32m━━━
[0m
[37m━━━━━━━━━━━━━━━━━
[0m 
[1m2s
[0m 0us/step 
[1m 4014080/20094049
[0m 
[32m━━━
[0m
[37m━━━━━━━━━━━━━━━━━
[0m 
[1m1s
[0m 0us/step 
[1m 4751360/20094049
[0m 
[32m━━━━
[0m
[37m━━━━━━━━━━━━━━━━
[0m 
[1m1s
[0m 0us/step 
[1m 5939200/20094049
[0m 
[32m━━━━━
[0m
[37m━━━━━━━━━━━━━━━
[0m 
[1m1s
[0m 0us/step 
[1m 7675904/20094049
[0m 
[32m━━━━━━━
[0m
[37m━━━━━━━━━━━━━
[0m 
[1m1s
[0m 0us/step 
[1m 9625600/20094049
[0m 
[32m━━━━━━━━━
[0m
[37m━━━━━━━━━━━
[0m 
[1m0s
[0m 0us/step 
[1m11698176/20094049
[0m 
[32m━━━━━━━━━━━
[0m
[37m━━━━━━━━━
[0m 
[1m0s
[0m 0us/step 
[1m13975552/20094049
[0m 
[32m━━━━━━━━━━━━━
[0m
[37m━━━━━━━
[0m 
[1m0s
[0m 0us/step 
[1m16367616/20094049
[0m 
[32m━━━━━━━━━━━━━━━━
[0m
[37m━━━━
[0m 
[1m0s
[0m 0us/step 
[1m18743296/20094049
[0m 
[32m━━━━━━━━━━━━━━━━━━
[0m
[37m━━
[0m 
[1m0s
[0m 0us/step 
[1m20094049/20094049
[0m 
[32m━━━━━━━━━━━━━━━━━━━━
[0m
[37m
[0m 
[1m1s
[0m 0us/step c(x_test, y_test) %<-% get_data(paste0(root_url, \"FordA_TEST.tsv\")) ## Downloading data from https://raw.githubusercontent.com/hfawaz/cd-diagram/master/FordA/FordA_TEST.tsv ## 
[1m      0/7364408
[0m 
[37m━━━━━━━━━━━━━━━━━━━━
[0m 
[1m0s
[0m 0s/step 
[1m 122880/7364408
[0m 
[37m━━━━━━━━━━━━━━━━━━━━
[0m 
[1m3s
[0m 0us/step 
[1m 589824/7364408
[0m 
[32m━
[0m
[37m━━━━━━━━━━━━━━━━━━━
[0m 
[1m1s
[0m 0us/step 
[1m1277952/7364408
[0m 
[32m━━━
[0m
[37m━━━━━━━━━━━━━━━━━
[0m 
[1m0s
[0m 0us/step 
[1m1720320/7364408
[0m 
[32m━━━━
[0m
[37m━━━━━━━━━━━━━━━━
[0m 
[1m0s
[0m 0us/step 
[1m3342336/7364408
[0m 
[32m━━━━━━━━━
[0m
[37m━━━━━━━━━━━
[0m 
[1m0s
[0m 0us/step 
[1m4112384/7364408
[0m 
[32m━━━━━━━━━━━
[0m
[37m━━━━━━━━━
[0m 
[1m0s
[0m 0us/step 
[1m6004736/7364408
[0m 
[32m━━━━━━━━━━━━━━━━
[0m
[37m━━━━
[0m 
[1m0s
[0m 0us/step 
[1m6823936/7364408
[0m 
[32m━━━━━━━━━━━━━━━━━━
[0m
[37m━━
[0m 
[1m0s
[0m 0us/step 
[1m7364408/7364408
[0m 
[32m━━━━━━━━━━━━━━━━━━━━
[0m
[37m
[0m 
[1m1s
[0m 0us/step str(keras3:::named_list(   x_train, y_train,   x_test, y_test )) ## List of 4 ##  $ x_train: num [1:3601, 1:500] -0.797 0.805 0.728 -0.234 -0.171 ... ##  $ y_train: int [1:3601, 1] -1 1 -1 -1 -1 1 1 1 1 1 ... ##  $ x_test : num [1:1320, 1:500] -0.14 0.334 0.717 1.24 -1.159 ... ##  $ y_test : int [1:1320, 1] -1 -1 -1 1 -1 1 -1 -1 1 1 ..."},{"path":"https://keras3.posit.co/dev/articles/examples/timeseries/timeseries_classification_from_scratch.html","id":"visualize-the-data","dir":"Articles > Examples > Timeseries","previous_headings":"","what":"Visualize the data","title":"Timeseries classification from scratch","text":"visualize one timeseries example class dataset.","code":"plot(NULL, main = \"Timeseries Data\",      xlab = \"Timepoints\",  ylab = \"Values\",      xlim = c(1, ncol(x_test)),      ylim = range(x_test)) grid() lines(x_test[match(-1, y_test), ], col = \"blue\") lines(x_test[match( 1, y_test), ], col = \"red\") legend(\"topright\", legend=c(\"label -1\", \"label 1\"), col=c(\"blue\", \"red\"), lty=1)"},{"path":"https://keras3.posit.co/dev/articles/examples/timeseries/timeseries_classification_from_scratch.html","id":"standardize-the-data","dir":"Articles > Examples > Timeseries","previous_headings":"","what":"Standardize the data","title":"Timeseries classification from scratch","text":"timeseries already single length (500). However, values usually various ranges. ideal neural network; general seek make input values normalized. specific dataset, data already z-normalized: timeseries sample mean equal zero standard deviation equal one. type normalization common timeseries classification problems, see Bagnall et al. (2016). Note timeseries data used univariate, meaning one channel per timeseries example. therefore transform timeseries multivariate one one channel using simple reshaping via numpy. allow us construct model easily applicable multivariate time series. Finally, order use sparse_categorical_crossentropy, count number classes beforehand. Now shuffle training set using validation_split option later training. Standardize labels positive integers. expected labels 0 1.","code":"dim(x_train) <- c(dim(x_train), 1) dim(x_test) <- c(dim(x_test), 1) num_classes <- length(unique(y_train)) c(x_train, y_train) %<-% listarrays::shuffle_rows(x_train, y_train) # idx <- sample.int(nrow(x_train)) # x_train %<>% .[idx,, ,drop = FALSE] # y_train %<>% .[idx,  ,drop = FALSE] y_train[y_train == -1L] <- 0L y_test[y_test == -1L] <- 0L"},{"path":"https://keras3.posit.co/dev/articles/examples/timeseries/timeseries_classification_from_scratch.html","id":"build-a-model","dir":"Articles > Examples > Timeseries","previous_headings":"","what":"Build a model","title":"Timeseries classification from scratch","text":"build Fully Convolutional Neural Network originally proposed paper. implementation based TF 2 version provided . following hyperparameters (kernel_size, filters, usage BatchNorm) found via random search using KerasTuner. plot chunk unnamed-chunk-9","code":"make_model <- function(input_shape) {   inputs <- keras_input(input_shape)    outputs <- inputs |>     # conv1     layer_conv_1d(filters = 64, kernel_size = 3, padding = \"same\") |>     layer_batch_normalization() |>     layer_activation_relu() |>     # conv2     layer_conv_1d(filters = 64, kernel_size = 3, padding = \"same\") |>     layer_batch_normalization() |>     layer_activation_relu() |>     # conv3     layer_conv_1d(filters = 64, kernel_size = 3, padding = \"same\") |>     layer_batch_normalization() |>     layer_activation_relu() |>     # pooling     layer_global_average_pooling_1d() |>     # final output     layer_dense(num_classes, activation = \"softmax\")    keras_model(inputs, outputs) }  model <- make_model(input_shape = dim(x_train)[-1]) model ## Model: \"functional\" ## ┏━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━┳━━━━━━━┓ ## ┃ Layer (type)                ┃ Output Shape          ┃    Param # ┃ Trai… ┃ ## ┡━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━╇━━━━━━━┩ ## │ input_layer (InputLayer)    │ (None, 500, 1)        │          0 │   -   │ ## ├─────────────────────────────┼───────────────────────┼────────────┼───────┤ ## │ conv1d (Conv1D)             │ (None, 500, 64)       │        256 │   Y   │ ## ├─────────────────────────────┼───────────────────────┼────────────┼───────┤ ## │ batch_normalization         │ (None, 500, 64)       │        256 │   Y   │ ## │ (BatchNormalization)        │                       │            │       │ ## ├─────────────────────────────┼───────────────────────┼────────────┼───────┤ ## │ re_lu (ReLU)                │ (None, 500, 64)       │          0 │   -   │ ## ├─────────────────────────────┼───────────────────────┼────────────┼───────┤ ## │ conv1d_1 (Conv1D)           │ (None, 500, 64)       │     12,352 │   Y   │ ## ├─────────────────────────────┼───────────────────────┼────────────┼───────┤ ## │ batch_normalization_1       │ (None, 500, 64)       │        256 │   Y   │ ## │ (BatchNormalization)        │                       │            │       │ ## ├─────────────────────────────┼───────────────────────┼────────────┼───────┤ ## │ re_lu_1 (ReLU)              │ (None, 500, 64)       │          0 │   -   │ ## ├─────────────────────────────┼───────────────────────┼────────────┼───────┤ ## │ conv1d_2 (Conv1D)           │ (None, 500, 64)       │     12,352 │   Y   │ ## ├─────────────────────────────┼───────────────────────┼────────────┼───────┤ ## │ batch_normalization_2       │ (None, 500, 64)       │        256 │   Y   │ ## │ (BatchNormalization)        │                       │            │       │ ## ├─────────────────────────────┼───────────────────────┼────────────┼───────┤ ## │ re_lu_2 (ReLU)              │ (None, 500, 64)       │          0 │   -   │ ## ├─────────────────────────────┼───────────────────────┼────────────┼───────┤ ## │ global_average_pooling1d    │ (None, 64)            │          0 │   -   │ ## │ (GlobalAveragePooling1D)    │                       │            │       │ ## ├─────────────────────────────┼───────────────────────┼────────────┼───────┤ ## │ dense (Dense)               │ (None, 2)             │        130 │   Y   │ ## └─────────────────────────────┴───────────────────────┴────────────┴───────┘ ##  Total params: 25,858 (101.01 KB) ##  Trainable params: 25,474 (99.51 KB) ##  Non-trainable params: 384 (1.50 KB) plot(model, show_shapes = TRUE)"},{"path":"https://keras3.posit.co/dev/articles/examples/timeseries/timeseries_classification_from_scratch.html","id":"train-the-model","dir":"Articles > Examples > Timeseries","previous_headings":"","what":"Train the model","title":"Timeseries classification from scratch","text":"","code":"epochs <- 500 batch_size <- 32  callbacks <- c(   callback_model_checkpoint(     \"best_model.keras\", save_best_only = TRUE,     monitor = \"val_loss\"   ),   callback_reduce_lr_on_plateau(     monitor = \"val_loss\", factor = 0.5,     patience = 20, min_lr = 0.0001   ),   callback_early_stopping(     monitor = \"val_loss\", patience = 50,     verbose = 1   ) )   model |> compile(   optimizer = \"adam\",   loss = \"sparse_categorical_crossentropy\",   metrics = \"sparse_categorical_accuracy\" )  history <- model |> fit(   x_train, y_train,   batch_size = batch_size,   epochs = epochs,   callbacks = callbacks,   validation_split = 0.2 ) ## Epoch 1/500 ## 90/90 - 4s - 41ms/step - loss: 0.5310 - sparse_categorical_accuracy: 0.7198 - val_loss: 0.7948 - val_sparse_categorical_accuracy: 0.4896 - learning_rate: 1.0000e-03 ## Epoch 2/500 ## 90/90 - 1s - 13ms/step - loss: 0.4771 - sparse_categorical_accuracy: 0.7597 - val_loss: 0.9159 - val_sparse_categorical_accuracy: 0.4896 - learning_rate: 1.0000e-03 ## Epoch 3/500 ## 90/90 - 0s - 3ms/step - loss: 0.4717 - sparse_categorical_accuracy: 0.7573 - val_loss: 0.9025 - val_sparse_categorical_accuracy: 0.4896 - learning_rate: 1.0000e-03 ## Epoch 4/500 ## 90/90 - 0s - 2ms/step - loss: 0.4113 - sparse_categorical_accuracy: 0.8000 - val_loss: 0.8173 - val_sparse_categorical_accuracy: 0.4910 - learning_rate: 1.0000e-03 ## Epoch 5/500 ## 90/90 - 0s - 3ms/step - loss: 0.4164 - sparse_categorical_accuracy: 0.7920 - val_loss: 0.6884 - val_sparse_categorical_accuracy: 0.5146 - learning_rate: 1.0000e-03 ## Epoch 6/500 ## 90/90 - 0s - 4ms/step - loss: 0.3984 - sparse_categorical_accuracy: 0.8101 - val_loss: 0.4227 - val_sparse_categorical_accuracy: 0.8294 - learning_rate: 1.0000e-03 ## Epoch 7/500 ## 90/90 - 0s - 2ms/step - loss: 0.3878 - sparse_categorical_accuracy: 0.8132 - val_loss: 0.4749 - val_sparse_categorical_accuracy: 0.7018 - learning_rate: 1.0000e-03 ## Epoch 8/500 ## 90/90 - 0s - 3ms/step - loss: 0.3739 - sparse_categorical_accuracy: 0.8163 - val_loss: 0.3880 - val_sparse_categorical_accuracy: 0.8252 - learning_rate: 1.0000e-03 ## Epoch 9/500 ## 90/90 - 0s - 2ms/step - loss: 0.3684 - sparse_categorical_accuracy: 0.8292 - val_loss: 0.4363 - val_sparse_categorical_accuracy: 0.7767 - learning_rate: 1.0000e-03 ## Epoch 10/500 ## 90/90 - 0s - 2ms/step - loss: 0.3590 - sparse_categorical_accuracy: 0.8306 - val_loss: 0.5077 - val_sparse_categorical_accuracy: 0.7282 - learning_rate: 1.0000e-03 ## Epoch 11/500 ## 90/90 - 0s - 3ms/step - loss: 0.3638 - sparse_categorical_accuracy: 0.8247 - val_loss: 0.3574 - val_sparse_categorical_accuracy: 0.8405 - learning_rate: 1.0000e-03 ## Epoch 12/500 ## 90/90 - 0s - 2ms/step - loss: 0.3459 - sparse_categorical_accuracy: 0.8396 - val_loss: 0.8054 - val_sparse_categorical_accuracy: 0.6352 - learning_rate: 1.0000e-03 ## Epoch 13/500 ## 90/90 - 0s - 2ms/step - loss: 0.3350 - sparse_categorical_accuracy: 0.8552 - val_loss: 0.3838 - val_sparse_categorical_accuracy: 0.8155 - learning_rate: 1.0000e-03 ## Epoch 14/500 ## 90/90 - 0s - 2ms/step - loss: 0.3303 - sparse_categorical_accuracy: 0.8545 - val_loss: 0.4148 - val_sparse_categorical_accuracy: 0.7920 - learning_rate: 1.0000e-03 ## Epoch 15/500 ## 90/90 - 0s - 2ms/step - loss: 0.3228 - sparse_categorical_accuracy: 0.8507 - val_loss: 1.6007 - val_sparse_categorical_accuracy: 0.5132 - learning_rate: 1.0000e-03 ## Epoch 16/500 ## 90/90 - 0s - 2ms/step - loss: 0.3232 - sparse_categorical_accuracy: 0.8587 - val_loss: 0.5349 - val_sparse_categorical_accuracy: 0.7184 - learning_rate: 1.0000e-03 ## Epoch 17/500 ## 90/90 - 0s - 2ms/step - loss: 0.3119 - sparse_categorical_accuracy: 0.8656 - val_loss: 0.3336 - val_sparse_categorical_accuracy: 0.8669 - learning_rate: 1.0000e-03 ## Epoch 18/500 ## 90/90 - 0s - 2ms/step - loss: 0.3057 - sparse_categorical_accuracy: 0.8646 - val_loss: 0.5241 - val_sparse_categorical_accuracy: 0.7351 - learning_rate: 1.0000e-03 ## Epoch 19/500 ## 90/90 - 0s - 2ms/step - loss: 0.3042 - sparse_categorical_accuracy: 0.8691 - val_loss: 0.4023 - val_sparse_categorical_accuracy: 0.8128 - learning_rate: 1.0000e-03 ## Epoch 20/500 ## 90/90 - 0s - 2ms/step - loss: 0.3037 - sparse_categorical_accuracy: 0.8677 - val_loss: 0.4818 - val_sparse_categorical_accuracy: 0.7670 - learning_rate: 1.0000e-03 ## Epoch 21/500 ## 90/90 - 0s - 2ms/step - loss: 0.2834 - sparse_categorical_accuracy: 0.8795 - val_loss: 0.3017 - val_sparse_categorical_accuracy: 0.8696 - learning_rate: 1.0000e-03 ## Epoch 22/500 ## 90/90 - 0s - 1ms/step - loss: 0.2879 - sparse_categorical_accuracy: 0.8819 - val_loss: 0.4130 - val_sparse_categorical_accuracy: 0.8003 - learning_rate: 1.0000e-03 ## Epoch 23/500 ## 90/90 - 0s - 1ms/step - loss: 0.2778 - sparse_categorical_accuracy: 0.8844 - val_loss: 0.3149 - val_sparse_categorical_accuracy: 0.8599 - learning_rate: 1.0000e-03 ## Epoch 24/500 ## 90/90 - 0s - 2ms/step - loss: 0.2743 - sparse_categorical_accuracy: 0.8861 - val_loss: 0.3331 - val_sparse_categorical_accuracy: 0.8460 - learning_rate: 1.0000e-03 ## Epoch 25/500 ## 90/90 - 0s - 1ms/step - loss: 0.2654 - sparse_categorical_accuracy: 0.8892 - val_loss: 0.4245 - val_sparse_categorical_accuracy: 0.7892 - learning_rate: 1.0000e-03 ## Epoch 26/500 ## 90/90 - 0s - 1ms/step - loss: 0.3058 - sparse_categorical_accuracy: 0.8660 - val_loss: 0.3187 - val_sparse_categorical_accuracy: 0.8669 - learning_rate: 1.0000e-03 ## Epoch 27/500 ## 90/90 - 0s - 1ms/step - loss: 0.2619 - sparse_categorical_accuracy: 0.8965 - val_loss: 0.4151 - val_sparse_categorical_accuracy: 0.7850 - learning_rate: 1.0000e-03 ## Epoch 28/500 ## 90/90 - 0s - 1ms/step - loss: 0.2785 - sparse_categorical_accuracy: 0.8792 - val_loss: 0.3807 - val_sparse_categorical_accuracy: 0.8308 - learning_rate: 1.0000e-03 ## Epoch 29/500 ## 90/90 - 0s - 2ms/step - loss: 0.2594 - sparse_categorical_accuracy: 0.8878 - val_loss: 0.4138 - val_sparse_categorical_accuracy: 0.8128 - learning_rate: 1.0000e-03 ## Epoch 30/500 ## 90/90 - 0s - 2ms/step - loss: 0.2644 - sparse_categorical_accuracy: 0.8878 - val_loss: 0.5111 - val_sparse_categorical_accuracy: 0.7323 - learning_rate: 1.0000e-03 ## Epoch 31/500 ## 90/90 - 0s - 2ms/step - loss: 0.2653 - sparse_categorical_accuracy: 0.8924 - val_loss: 0.2788 - val_sparse_categorical_accuracy: 0.8863 - learning_rate: 1.0000e-03 ## Epoch 32/500 ## 90/90 - 0s - 2ms/step - loss: 0.2546 - sparse_categorical_accuracy: 0.8903 - val_loss: 0.2598 - val_sparse_categorical_accuracy: 0.8821 - learning_rate: 1.0000e-03 ## Epoch 33/500 ## 90/90 - 0s - 2ms/step - loss: 0.2408 - sparse_categorical_accuracy: 0.9062 - val_loss: 0.2632 - val_sparse_categorical_accuracy: 0.8793 - learning_rate: 1.0000e-03 ## Epoch 34/500 ## 90/90 - 0s - 2ms/step - loss: 0.2373 - sparse_categorical_accuracy: 0.9052 - val_loss: 0.2966 - val_sparse_categorical_accuracy: 0.8710 - learning_rate: 1.0000e-03 ## Epoch 35/500 ## 90/90 - 0s - 1ms/step - loss: 0.2560 - sparse_categorical_accuracy: 0.8903 - val_loss: 0.6435 - val_sparse_categorical_accuracy: 0.7309 - learning_rate: 1.0000e-03 ## Epoch 36/500 ## 90/90 - 0s - 1ms/step - loss: 0.2391 - sparse_categorical_accuracy: 0.9035 - val_loss: 0.2705 - val_sparse_categorical_accuracy: 0.8807 - learning_rate: 1.0000e-03 ## Epoch 37/500 ## 90/90 - 0s - 2ms/step - loss: 0.2384 - sparse_categorical_accuracy: 0.8993 - val_loss: 0.2720 - val_sparse_categorical_accuracy: 0.8738 - learning_rate: 1.0000e-03 ## Epoch 38/500 ## 90/90 - 0s - 2ms/step - loss: 0.2339 - sparse_categorical_accuracy: 0.9080 - val_loss: 0.9019 - val_sparse_categorical_accuracy: 0.7309 - learning_rate: 1.0000e-03 ## Epoch 39/500 ## 90/90 - 0s - 2ms/step - loss: 0.2404 - sparse_categorical_accuracy: 0.9038 - val_loss: 0.4007 - val_sparse_categorical_accuracy: 0.8294 - learning_rate: 1.0000e-03 ## Epoch 40/500 ## 90/90 - 0s - 2ms/step - loss: 0.2265 - sparse_categorical_accuracy: 0.9104 - val_loss: 0.6253 - val_sparse_categorical_accuracy: 0.7559 - learning_rate: 1.0000e-03 ## Epoch 41/500 ## 90/90 - 0s - 2ms/step - loss: 0.2159 - sparse_categorical_accuracy: 0.9174 - val_loss: 0.2307 - val_sparse_categorical_accuracy: 0.8974 - learning_rate: 1.0000e-03 ## Epoch 42/500 ## 90/90 - 0s - 2ms/step - loss: 0.2081 - sparse_categorical_accuracy: 0.9271 - val_loss: 0.5934 - val_sparse_categorical_accuracy: 0.7434 - learning_rate: 1.0000e-03 ## Epoch 43/500 ## 90/90 - 0s - 2ms/step - loss: 0.2135 - sparse_categorical_accuracy: 0.9153 - val_loss: 0.2809 - val_sparse_categorical_accuracy: 0.8613 - learning_rate: 1.0000e-03 ## Epoch 44/500 ## 90/90 - 0s - 2ms/step - loss: 0.2152 - sparse_categorical_accuracy: 0.9149 - val_loss: 0.3832 - val_sparse_categorical_accuracy: 0.8211 - learning_rate: 1.0000e-03 ## Epoch 45/500 ## 90/90 - 0s - 2ms/step - loss: 0.2060 - sparse_categorical_accuracy: 0.9233 - val_loss: 0.3182 - val_sparse_categorical_accuracy: 0.8682 - learning_rate: 1.0000e-03 ## Epoch 46/500 ## 90/90 - 0s - 2ms/step - loss: 0.2023 - sparse_categorical_accuracy: 0.9229 - val_loss: 0.2104 - val_sparse_categorical_accuracy: 0.9140 - learning_rate: 1.0000e-03 ## Epoch 47/500 ## 90/90 - 0s - 1ms/step - loss: 0.1855 - sparse_categorical_accuracy: 0.9309 - val_loss: 0.2882 - val_sparse_categorical_accuracy: 0.8877 - learning_rate: 1.0000e-03 ## Epoch 48/500 ## 90/90 - 0s - 1ms/step - loss: 0.1789 - sparse_categorical_accuracy: 0.9306 - val_loss: 0.4009 - val_sparse_categorical_accuracy: 0.8197 - learning_rate: 1.0000e-03 ## Epoch 49/500 ## 90/90 - 0s - 1ms/step - loss: 0.1643 - sparse_categorical_accuracy: 0.9417 - val_loss: 0.2160 - val_sparse_categorical_accuracy: 0.9071 - learning_rate: 1.0000e-03 ## Epoch 50/500 ## 90/90 - 0s - 1ms/step - loss: 0.1599 - sparse_categorical_accuracy: 0.9462 - val_loss: 0.6680 - val_sparse_categorical_accuracy: 0.6768 - learning_rate: 1.0000e-03 ## Epoch 51/500 ## 90/90 - 0s - 1ms/step - loss: 0.1623 - sparse_categorical_accuracy: 0.9420 - val_loss: 1.2733 - val_sparse_categorical_accuracy: 0.7254 - learning_rate: 1.0000e-03 ## Epoch 52/500 ## 90/90 - 0s - 2ms/step - loss: 0.1465 - sparse_categorical_accuracy: 0.9507 - val_loss: 1.1664 - val_sparse_categorical_accuracy: 0.6976 - learning_rate: 1.0000e-03 ## Epoch 53/500 ## 90/90 - 0s - 1ms/step - loss: 0.1453 - sparse_categorical_accuracy: 0.9465 - val_loss: 1.3677 - val_sparse_categorical_accuracy: 0.7074 - learning_rate: 1.0000e-03 ## Epoch 54/500 ## 90/90 - 0s - 1ms/step - loss: 0.1370 - sparse_categorical_accuracy: 0.9563 - val_loss: 0.2190 - val_sparse_categorical_accuracy: 0.9098 - learning_rate: 1.0000e-03 ## Epoch 55/500 ## 90/90 - 0s - 1ms/step - loss: 0.1319 - sparse_categorical_accuracy: 0.9563 - val_loss: 0.2376 - val_sparse_categorical_accuracy: 0.8932 - learning_rate: 1.0000e-03 ## Epoch 56/500 ## 90/90 - 0s - 2ms/step - loss: 0.1354 - sparse_categorical_accuracy: 0.9549 - val_loss: 0.1614 - val_sparse_categorical_accuracy: 0.9334 - learning_rate: 1.0000e-03 ## Epoch 57/500 ## 90/90 - 0s - 2ms/step - loss: 0.1298 - sparse_categorical_accuracy: 0.9590 - val_loss: 0.1651 - val_sparse_categorical_accuracy: 0.9376 - learning_rate: 1.0000e-03 ## Epoch 58/500 ## 90/90 - 0s - 2ms/step - loss: 0.1188 - sparse_categorical_accuracy: 0.9569 - val_loss: 0.2961 - val_sparse_categorical_accuracy: 0.8766 - learning_rate: 1.0000e-03 ## Epoch 59/500 ## 90/90 - 0s - 2ms/step - loss: 0.1295 - sparse_categorical_accuracy: 0.9545 - val_loss: 0.1905 - val_sparse_categorical_accuracy: 0.9209 - learning_rate: 1.0000e-03 ## Epoch 60/500 ## 90/90 - 0s - 1ms/step - loss: 0.1281 - sparse_categorical_accuracy: 0.9552 - val_loss: 0.1835 - val_sparse_categorical_accuracy: 0.9140 - learning_rate: 1.0000e-03 ## Epoch 61/500 ## 90/90 - 0s - 1ms/step - loss: 0.1176 - sparse_categorical_accuracy: 0.9646 - val_loss: 1.6040 - val_sparse_categorical_accuracy: 0.6741 - learning_rate: 1.0000e-03 ## Epoch 62/500 ## 90/90 - 0s - 1ms/step - loss: 0.1363 - sparse_categorical_accuracy: 0.9500 - val_loss: 2.0555 - val_sparse_categorical_accuracy: 0.6963 - learning_rate: 1.0000e-03 ## Epoch 63/500 ## 90/90 - 0s - 1ms/step - loss: 0.1358 - sparse_categorical_accuracy: 0.9483 - val_loss: 3.2565 - val_sparse_categorical_accuracy: 0.6352 - learning_rate: 1.0000e-03 ## Epoch 64/500 ## 90/90 - 0s - 2ms/step - loss: 0.1276 - sparse_categorical_accuracy: 0.9590 - val_loss: 0.3056 - val_sparse_categorical_accuracy: 0.8669 - learning_rate: 1.0000e-03 ## Epoch 65/500 ## 90/90 - 0s - 2ms/step - loss: 0.1189 - sparse_categorical_accuracy: 0.9590 - val_loss: 0.3571 - val_sparse_categorical_accuracy: 0.8530 - learning_rate: 1.0000e-03 ## Epoch 66/500 ## 90/90 - 0s - 2ms/step - loss: 0.1111 - sparse_categorical_accuracy: 0.9653 - val_loss: 0.1403 - val_sparse_categorical_accuracy: 0.9473 - learning_rate: 1.0000e-03 ## Epoch 67/500 ## 90/90 - 0s - 2ms/step - loss: 0.1120 - sparse_categorical_accuracy: 0.9635 - val_loss: 1.6044 - val_sparse_categorical_accuracy: 0.6477 - learning_rate: 1.0000e-03 ## Epoch 68/500 ## 90/90 - 0s - 2ms/step - loss: 0.1123 - sparse_categorical_accuracy: 0.9632 - val_loss: 0.6339 - val_sparse_categorical_accuracy: 0.7989 - learning_rate: 1.0000e-03 ## Epoch 69/500 ## 90/90 - 0s - 2ms/step - loss: 0.1150 - sparse_categorical_accuracy: 0.9611 - val_loss: 0.3409 - val_sparse_categorical_accuracy: 0.8585 - learning_rate: 1.0000e-03 ## Epoch 70/500 ## 90/90 - 0s - 1ms/step - loss: 0.1120 - sparse_categorical_accuracy: 0.9597 - val_loss: 0.1705 - val_sparse_categorical_accuracy: 0.9209 - learning_rate: 1.0000e-03 ## Epoch 71/500 ## 90/90 - 0s - 2ms/step - loss: 0.1070 - sparse_categorical_accuracy: 0.9646 - val_loss: 0.3976 - val_sparse_categorical_accuracy: 0.8488 - learning_rate: 1.0000e-03 ## Epoch 72/500 ## 90/90 - 0s - 2ms/step - loss: 0.1076 - sparse_categorical_accuracy: 0.9663 - val_loss: 0.1426 - val_sparse_categorical_accuracy: 0.9473 - learning_rate: 1.0000e-03 ## Epoch 73/500 ## 90/90 - 0s - 2ms/step - loss: 0.1025 - sparse_categorical_accuracy: 0.9677 - val_loss: 0.3274 - val_sparse_categorical_accuracy: 0.8613 - learning_rate: 1.0000e-03 ## Epoch 74/500 ## 90/90 - 0s - 2ms/step - loss: 0.1056 - sparse_categorical_accuracy: 0.9628 - val_loss: 0.2165 - val_sparse_categorical_accuracy: 0.9182 - learning_rate: 1.0000e-03 ## Epoch 75/500 ## 90/90 - 0s - 1ms/step - loss: 0.1303 - sparse_categorical_accuracy: 0.9524 - val_loss: 0.6197 - val_sparse_categorical_accuracy: 0.7282 - learning_rate: 1.0000e-03 ## Epoch 76/500 ## 90/90 - 0s - 1ms/step - loss: 0.1205 - sparse_categorical_accuracy: 0.9590 - val_loss: 0.2972 - val_sparse_categorical_accuracy: 0.8793 - learning_rate: 1.0000e-03 ## Epoch 77/500 ## 90/90 - 0s - 1ms/step - loss: 0.1061 - sparse_categorical_accuracy: 0.9632 - val_loss: 0.2158 - val_sparse_categorical_accuracy: 0.9182 - learning_rate: 1.0000e-03 ## Epoch 78/500 ## 90/90 - 0s - 1ms/step - loss: 0.0959 - sparse_categorical_accuracy: 0.9705 - val_loss: 0.1449 - val_sparse_categorical_accuracy: 0.9362 - learning_rate: 1.0000e-03 ## Epoch 79/500 ## 90/90 - 0s - 2ms/step - loss: 0.0995 - sparse_categorical_accuracy: 0.9674 - val_loss: 0.1231 - val_sparse_categorical_accuracy: 0.9598 - learning_rate: 1.0000e-03 ## Epoch 80/500 ## 90/90 - 0s - 1ms/step - loss: 0.1027 - sparse_categorical_accuracy: 0.9677 - val_loss: 0.5311 - val_sparse_categorical_accuracy: 0.7712 - learning_rate: 1.0000e-03 ## Epoch 81/500 ## 90/90 - 0s - 2ms/step - loss: 0.1012 - sparse_categorical_accuracy: 0.9677 - val_loss: 0.1622 - val_sparse_categorical_accuracy: 0.9348 - learning_rate: 1.0000e-03 ## Epoch 82/500 ## 90/90 - 0s - 1ms/step - loss: 0.0929 - sparse_categorical_accuracy: 0.9663 - val_loss: 0.1457 - val_sparse_categorical_accuracy: 0.9417 - learning_rate: 1.0000e-03 ## Epoch 83/500 ## 90/90 - 0s - 1ms/step - loss: 0.1008 - sparse_categorical_accuracy: 0.9663 - val_loss: 1.6042 - val_sparse_categorical_accuracy: 0.6879 - learning_rate: 1.0000e-03 ## Epoch 84/500 ## 90/90 - 0s - 1ms/step - loss: 0.0981 - sparse_categorical_accuracy: 0.9653 - val_loss: 0.1732 - val_sparse_categorical_accuracy: 0.9307 - learning_rate: 1.0000e-03 ## Epoch 85/500 ## 90/90 - 0s - 1ms/step - loss: 0.1006 - sparse_categorical_accuracy: 0.9674 - val_loss: 1.8495 - val_sparse_categorical_accuracy: 0.6976 - learning_rate: 1.0000e-03 ## Epoch 86/500 ## 90/90 - 0s - 1ms/step - loss: 0.1005 - sparse_categorical_accuracy: 0.9660 - val_loss: 0.1263 - val_sparse_categorical_accuracy: 0.9459 - learning_rate: 1.0000e-03 ## Epoch 87/500 ## 90/90 - 0s - 1ms/step - loss: 0.1054 - sparse_categorical_accuracy: 0.9632 - val_loss: 0.2135 - val_sparse_categorical_accuracy: 0.9140 - learning_rate: 1.0000e-03 ## Epoch 88/500 ## 90/90 - 0s - 2ms/step - loss: 0.1051 - sparse_categorical_accuracy: 0.9656 - val_loss: 1.3499 - val_sparse_categorical_accuracy: 0.6630 - learning_rate: 1.0000e-03 ## Epoch 89/500 ## 90/90 - 0s - 2ms/step - loss: 0.1037 - sparse_categorical_accuracy: 0.9615 - val_loss: 0.2271 - val_sparse_categorical_accuracy: 0.9029 - learning_rate: 1.0000e-03 ## Epoch 90/500 ## 90/90 - 0s - 2ms/step - loss: 0.0976 - sparse_categorical_accuracy: 0.9670 - val_loss: 0.7824 - val_sparse_categorical_accuracy: 0.7129 - learning_rate: 1.0000e-03 ## Epoch 91/500 ## 90/90 - 0s - 1ms/step - loss: 0.0995 - sparse_categorical_accuracy: 0.9677 - val_loss: 0.2369 - val_sparse_categorical_accuracy: 0.8918 - learning_rate: 1.0000e-03 ## Epoch 92/500 ## 90/90 - 0s - 2ms/step - loss: 0.0986 - sparse_categorical_accuracy: 0.9670 - val_loss: 0.1185 - val_sparse_categorical_accuracy: 0.9584 - learning_rate: 1.0000e-03 ## Epoch 93/500 ## 90/90 - 0s - 2ms/step - loss: 0.0957 - sparse_categorical_accuracy: 0.9670 - val_loss: 1.8619 - val_sparse_categorical_accuracy: 0.5270 - learning_rate: 1.0000e-03 ## Epoch 94/500 ## 90/90 - 0s - 2ms/step - loss: 0.0999 - sparse_categorical_accuracy: 0.9642 - val_loss: 0.4507 - val_sparse_categorical_accuracy: 0.8350 - learning_rate: 1.0000e-03 ## Epoch 95/500 ## 90/90 - 0s - 2ms/step - loss: 0.0939 - sparse_categorical_accuracy: 0.9653 - val_loss: 1.0529 - val_sparse_categorical_accuracy: 0.6893 - learning_rate: 1.0000e-03 ## Epoch 96/500 ## 90/90 - 0s - 2ms/step - loss: 0.1066 - sparse_categorical_accuracy: 0.9608 - val_loss: 0.1416 - val_sparse_categorical_accuracy: 0.9417 - learning_rate: 1.0000e-03 ## Epoch 97/500 ## 90/90 - 0s - 1ms/step - loss: 0.1068 - sparse_categorical_accuracy: 0.9618 - val_loss: 0.5324 - val_sparse_categorical_accuracy: 0.7961 - learning_rate: 1.0000e-03 ## Epoch 98/500 ## 90/90 - 0s - 1ms/step - loss: 0.0916 - sparse_categorical_accuracy: 0.9694 - val_loss: 0.8154 - val_sparse_categorical_accuracy: 0.7656 - learning_rate: 1.0000e-03 ## Epoch 99/500 ## 90/90 - 0s - 2ms/step - loss: 0.0942 - sparse_categorical_accuracy: 0.9639 - val_loss: 0.1314 - val_sparse_categorical_accuracy: 0.9542 - learning_rate: 1.0000e-03 ## Epoch 100/500 ## 90/90 - 0s - 1ms/step - loss: 0.0938 - sparse_categorical_accuracy: 0.9698 - val_loss: 2.2256 - val_sparse_categorical_accuracy: 0.6366 - learning_rate: 1.0000e-03 ## Epoch 101/500 ## 90/90 - 0s - 2ms/step - loss: 0.0853 - sparse_categorical_accuracy: 0.9750 - val_loss: 0.1660 - val_sparse_categorical_accuracy: 0.9376 - learning_rate: 1.0000e-03 ## Epoch 102/500 ## 90/90 - 0s - 2ms/step - loss: 0.0957 - sparse_categorical_accuracy: 0.9691 - val_loss: 0.2444 - val_sparse_categorical_accuracy: 0.8932 - learning_rate: 1.0000e-03 ## Epoch 103/500 ## 90/90 - 0s - 1ms/step - loss: 0.0949 - sparse_categorical_accuracy: 0.9660 - val_loss: 1.3661 - val_sparse_categorical_accuracy: 0.6519 - learning_rate: 1.0000e-03 ## Epoch 104/500 ## 90/90 - 0s - 1ms/step - loss: 0.1033 - sparse_categorical_accuracy: 0.9622 - val_loss: 0.2086 - val_sparse_categorical_accuracy: 0.9126 - learning_rate: 1.0000e-03 ## Epoch 105/500 ## 90/90 - 0s - 1ms/step - loss: 0.0981 - sparse_categorical_accuracy: 0.9663 - val_loss: 0.2216 - val_sparse_categorical_accuracy: 0.8988 - learning_rate: 1.0000e-03 ## Epoch 106/500 ## 90/90 - 0s - 1ms/step - loss: 0.0895 - sparse_categorical_accuracy: 0.9681 - val_loss: 2.2126 - val_sparse_categorical_accuracy: 0.6269 - learning_rate: 1.0000e-03 ## Epoch 107/500 ## 90/90 - 0s - 1ms/step - loss: 0.0939 - sparse_categorical_accuracy: 0.9670 - val_loss: 0.4274 - val_sparse_categorical_accuracy: 0.8128 - learning_rate: 1.0000e-03 ## Epoch 108/500 ## 90/90 - 0s - 1ms/step - loss: 0.0894 - sparse_categorical_accuracy: 0.9719 - val_loss: 0.2745 - val_sparse_categorical_accuracy: 0.8946 - learning_rate: 1.0000e-03 ## Epoch 109/500 ## 90/90 - 0s - 2ms/step - loss: 0.0937 - sparse_categorical_accuracy: 0.9722 - val_loss: 0.1248 - val_sparse_categorical_accuracy: 0.9487 - learning_rate: 1.0000e-03 ## Epoch 110/500 ## 90/90 - 0s - 1ms/step - loss: 0.0965 - sparse_categorical_accuracy: 0.9656 - val_loss: 0.5962 - val_sparse_categorical_accuracy: 0.7989 - learning_rate: 1.0000e-03 ## Epoch 111/500 ## 90/90 - 0s - 1ms/step - loss: 0.0825 - sparse_categorical_accuracy: 0.9726 - val_loss: 0.1671 - val_sparse_categorical_accuracy: 0.9307 - learning_rate: 1.0000e-03 ## Epoch 112/500 ## 90/90 - 0s - 1ms/step - loss: 0.0879 - sparse_categorical_accuracy: 0.9694 - val_loss: 0.1298 - val_sparse_categorical_accuracy: 0.9390 - learning_rate: 1.0000e-03 ## Epoch 113/500 ## 90/90 - 1s - 13ms/step - loss: 0.0785 - sparse_categorical_accuracy: 0.9757 - val_loss: 0.1539 - val_sparse_categorical_accuracy: 0.9348 - learning_rate: 5.0000e-04 ## Epoch 114/500 ## 90/90 - 0s - 1ms/step - loss: 0.0914 - sparse_categorical_accuracy: 0.9722 - val_loss: 0.1216 - val_sparse_categorical_accuracy: 0.9542 - learning_rate: 5.0000e-04 ## Epoch 115/500 ## 90/90 - 0s - 2ms/step - loss: 0.0754 - sparse_categorical_accuracy: 0.9740 - val_loss: 0.1167 - val_sparse_categorical_accuracy: 0.9431 - learning_rate: 5.0000e-04 ## Epoch 116/500 ## 90/90 - 0s - 1ms/step - loss: 0.0786 - sparse_categorical_accuracy: 0.9753 - val_loss: 0.1977 - val_sparse_categorical_accuracy: 0.9279 - learning_rate: 5.0000e-04 ## Epoch 117/500 ## 90/90 - 0s - 1ms/step - loss: 0.0758 - sparse_categorical_accuracy: 0.9736 - val_loss: 0.1724 - val_sparse_categorical_accuracy: 0.9431 - learning_rate: 5.0000e-04 ## Epoch 118/500 ## 90/90 - 0s - 2ms/step - loss: 0.0793 - sparse_categorical_accuracy: 0.9712 - val_loss: 0.1316 - val_sparse_categorical_accuracy: 0.9501 - learning_rate: 5.0000e-04 ## Epoch 119/500 ## 90/90 - 0s - 2ms/step - loss: 0.0781 - sparse_categorical_accuracy: 0.9750 - val_loss: 0.1054 - val_sparse_categorical_accuracy: 0.9598 - learning_rate: 5.0000e-04 ## Epoch 120/500 ## 90/90 - 0s - 2ms/step - loss: 0.0808 - sparse_categorical_accuracy: 0.9740 - val_loss: 0.1425 - val_sparse_categorical_accuracy: 0.9515 - learning_rate: 5.0000e-04 ## Epoch 121/500 ## 90/90 - 0s - 2ms/step - loss: 0.0761 - sparse_categorical_accuracy: 0.9753 - val_loss: 0.1359 - val_sparse_categorical_accuracy: 0.9431 - learning_rate: 5.0000e-04 ## Epoch 122/500 ## 90/90 - 0s - 1ms/step - loss: 0.0813 - sparse_categorical_accuracy: 0.9719 - val_loss: 0.1067 - val_sparse_categorical_accuracy: 0.9556 - learning_rate: 5.0000e-04 ## Epoch 123/500 ## 90/90 - 0s - 2ms/step - loss: 0.0778 - sparse_categorical_accuracy: 0.9767 - val_loss: 0.1328 - val_sparse_categorical_accuracy: 0.9528 - learning_rate: 5.0000e-04 ## Epoch 124/500 ## 90/90 - 0s - 2ms/step - loss: 0.0822 - sparse_categorical_accuracy: 0.9705 - val_loss: 0.1931 - val_sparse_categorical_accuracy: 0.9265 - learning_rate: 5.0000e-04 ## Epoch 125/500 ## 90/90 - 0s - 2ms/step - loss: 0.0797 - sparse_categorical_accuracy: 0.9719 - val_loss: 0.1279 - val_sparse_categorical_accuracy: 0.9459 - learning_rate: 5.0000e-04 ## Epoch 126/500 ## 90/90 - 0s - 1ms/step - loss: 0.0761 - sparse_categorical_accuracy: 0.9743 - val_loss: 0.2298 - val_sparse_categorical_accuracy: 0.9154 - learning_rate: 5.0000e-04 ## Epoch 127/500 ## 90/90 - 0s - 1ms/step - loss: 0.0740 - sparse_categorical_accuracy: 0.9802 - val_loss: 0.2257 - val_sparse_categorical_accuracy: 0.9223 - learning_rate: 5.0000e-04 ## Epoch 128/500 ## 90/90 - 0s - 2ms/step - loss: 0.0772 - sparse_categorical_accuracy: 0.9760 - val_loss: 0.1223 - val_sparse_categorical_accuracy: 0.9515 - learning_rate: 5.0000e-04 ## Epoch 129/500 ## 90/90 - 0s - 1ms/step - loss: 0.0760 - sparse_categorical_accuracy: 0.9764 - val_loss: 0.1443 - val_sparse_categorical_accuracy: 0.9501 - learning_rate: 5.0000e-04 ## Epoch 130/500 ## 90/90 - 0s - 1ms/step - loss: 0.0841 - sparse_categorical_accuracy: 0.9743 - val_loss: 0.1104 - val_sparse_categorical_accuracy: 0.9612 - learning_rate: 5.0000e-04 ## Epoch 131/500 ## 90/90 - 0s - 1ms/step - loss: 0.0697 - sparse_categorical_accuracy: 0.9785 - val_loss: 0.1062 - val_sparse_categorical_accuracy: 0.9584 - learning_rate: 5.0000e-04 ## Epoch 132/500 ## 90/90 - 0s - 1ms/step - loss: 0.0739 - sparse_categorical_accuracy: 0.9743 - val_loss: 0.1153 - val_sparse_categorical_accuracy: 0.9653 - learning_rate: 5.0000e-04 ## Epoch 133/500 ## 90/90 - 0s - 1ms/step - loss: 0.0704 - sparse_categorical_accuracy: 0.9781 - val_loss: 0.1186 - val_sparse_categorical_accuracy: 0.9473 - learning_rate: 5.0000e-04 ## Epoch 134/500 ## 90/90 - 0s - 1ms/step - loss: 0.0753 - sparse_categorical_accuracy: 0.9774 - val_loss: 0.2290 - val_sparse_categorical_accuracy: 0.9196 - learning_rate: 5.0000e-04 ## Epoch 135/500 ## 90/90 - 0s - 1ms/step - loss: 0.0734 - sparse_categorical_accuracy: 0.9743 - val_loss: 0.1498 - val_sparse_categorical_accuracy: 0.9501 - learning_rate: 5.0000e-04 ## Epoch 136/500 ## 90/90 - 0s - 1ms/step - loss: 0.0773 - sparse_categorical_accuracy: 0.9712 - val_loss: 0.1266 - val_sparse_categorical_accuracy: 0.9570 - learning_rate: 5.0000e-04 ## Epoch 137/500 ## 90/90 - 0s - 1ms/step - loss: 0.0807 - sparse_categorical_accuracy: 0.9740 - val_loss: 0.1143 - val_sparse_categorical_accuracy: 0.9639 - learning_rate: 5.0000e-04 ## Epoch 138/500 ## 90/90 - 0s - 1ms/step - loss: 0.0755 - sparse_categorical_accuracy: 0.9753 - val_loss: 0.2602 - val_sparse_categorical_accuracy: 0.9015 - learning_rate: 5.0000e-04 ## Epoch 139/500 ## 90/90 - 0s - 2ms/step - loss: 0.0807 - sparse_categorical_accuracy: 0.9750 - val_loss: 0.0994 - val_sparse_categorical_accuracy: 0.9626 - learning_rate: 5.0000e-04 ## Epoch 140/500 ## 90/90 - 0s - 1ms/step - loss: 0.0697 - sparse_categorical_accuracy: 0.9774 - val_loss: 0.2056 - val_sparse_categorical_accuracy: 0.9182 - learning_rate: 5.0000e-04 ## Epoch 141/500 ## 90/90 - 0s - 2ms/step - loss: 0.0701 - sparse_categorical_accuracy: 0.9753 - val_loss: 0.1829 - val_sparse_categorical_accuracy: 0.9362 - learning_rate: 5.0000e-04 ## Epoch 142/500 ## 90/90 - 0s - 1ms/step - loss: 0.0669 - sparse_categorical_accuracy: 0.9792 - val_loss: 0.1132 - val_sparse_categorical_accuracy: 0.9598 - learning_rate: 5.0000e-04 ## Epoch 143/500 ## 90/90 - 0s - 1ms/step - loss: 0.0751 - sparse_categorical_accuracy: 0.9767 - val_loss: 0.1953 - val_sparse_categorical_accuracy: 0.9223 - learning_rate: 5.0000e-04 ## Epoch 144/500 ## 90/90 - 0s - 1ms/step - loss: 0.0660 - sparse_categorical_accuracy: 0.9792 - val_loss: 0.1440 - val_sparse_categorical_accuracy: 0.9404 - learning_rate: 5.0000e-04 ## Epoch 145/500 ## 90/90 - 0s - 2ms/step - loss: 0.0673 - sparse_categorical_accuracy: 0.9764 - val_loss: 0.6545 - val_sparse_categorical_accuracy: 0.8100 - learning_rate: 5.0000e-04 ## Epoch 146/500 ## 90/90 - 0s - 2ms/step - loss: 0.0708 - sparse_categorical_accuracy: 0.9757 - val_loss: 0.1321 - val_sparse_categorical_accuracy: 0.9542 - learning_rate: 5.0000e-04 ## Epoch 147/500 ## 90/90 - 0s - 2ms/step - loss: 0.0733 - sparse_categorical_accuracy: 0.9764 - val_loss: 0.1479 - val_sparse_categorical_accuracy: 0.9487 - learning_rate: 5.0000e-04 ## Epoch 148/500 ## 90/90 - 0s - 2ms/step - loss: 0.0797 - sparse_categorical_accuracy: 0.9715 - val_loss: 0.3700 - val_sparse_categorical_accuracy: 0.8752 - learning_rate: 5.0000e-04 ## Epoch 149/500 ## 90/90 - 0s - 2ms/step - loss: 0.0709 - sparse_categorical_accuracy: 0.9760 - val_loss: 0.1170 - val_sparse_categorical_accuracy: 0.9584 - learning_rate: 5.0000e-04 ## Epoch 150/500 ## 90/90 - 0s - 2ms/step - loss: 0.0761 - sparse_categorical_accuracy: 0.9726 - val_loss: 0.1173 - val_sparse_categorical_accuracy: 0.9653 - learning_rate: 5.0000e-04 ## Epoch 151/500 ## 90/90 - 0s - 3ms/step - loss: 0.0728 - sparse_categorical_accuracy: 0.9719 - val_loss: 0.1596 - val_sparse_categorical_accuracy: 0.9334 - learning_rate: 5.0000e-04 ## Epoch 152/500 ## 90/90 - 0s - 2ms/step - loss: 0.0727 - sparse_categorical_accuracy: 0.9722 - val_loss: 0.1086 - val_sparse_categorical_accuracy: 0.9528 - learning_rate: 5.0000e-04 ## Epoch 153/500 ## 90/90 - 0s - 2ms/step - loss: 0.0704 - sparse_categorical_accuracy: 0.9771 - val_loss: 0.1305 - val_sparse_categorical_accuracy: 0.9487 - learning_rate: 5.0000e-04 ## Epoch 154/500 ## 90/90 - 0s - 2ms/step - loss: 0.0667 - sparse_categorical_accuracy: 0.9781 - val_loss: 0.1788 - val_sparse_categorical_accuracy: 0.9279 - learning_rate: 5.0000e-04 ## Epoch 155/500 ## 90/90 - 0s - 2ms/step - loss: 0.0702 - sparse_categorical_accuracy: 0.9764 - val_loss: 0.1109 - val_sparse_categorical_accuracy: 0.9501 - learning_rate: 5.0000e-04 ## Epoch 156/500 ## 90/90 - 0s - 1ms/step - loss: 0.0741 - sparse_categorical_accuracy: 0.9764 - val_loss: 0.1325 - val_sparse_categorical_accuracy: 0.9431 - learning_rate: 5.0000e-04 ## Epoch 157/500 ## 90/90 - 0s - 1ms/step - loss: 0.0750 - sparse_categorical_accuracy: 0.9722 - val_loss: 0.1820 - val_sparse_categorical_accuracy: 0.9348 - learning_rate: 5.0000e-04 ## Epoch 158/500 ## 90/90 - 0s - 1ms/step - loss: 0.0714 - sparse_categorical_accuracy: 0.9771 - val_loss: 0.1982 - val_sparse_categorical_accuracy: 0.9182 - learning_rate: 5.0000e-04 ## Epoch 159/500 ## 90/90 - 0s - 2ms/step - loss: 0.0732 - sparse_categorical_accuracy: 0.9750 - val_loss: 0.2841 - val_sparse_categorical_accuracy: 0.8988 - learning_rate: 5.0000e-04 ## Epoch 160/500 ## 90/90 - 0s - 1ms/step - loss: 0.0679 - sparse_categorical_accuracy: 0.9760 - val_loss: 0.1040 - val_sparse_categorical_accuracy: 0.9639 - learning_rate: 2.5000e-04 ## Epoch 161/500 ## 90/90 - 0s - 2ms/step - loss: 0.0649 - sparse_categorical_accuracy: 0.9785 - val_loss: 0.1281 - val_sparse_categorical_accuracy: 0.9459 - learning_rate: 2.5000e-04 ## Epoch 162/500 ## 90/90 - 0s - 2ms/step - loss: 0.0623 - sparse_categorical_accuracy: 0.9781 - val_loss: 0.1535 - val_sparse_categorical_accuracy: 0.9431 - learning_rate: 2.5000e-04 ## Epoch 163/500 ## 90/90 - 0s - 2ms/step - loss: 0.0588 - sparse_categorical_accuracy: 0.9813 - val_loss: 0.1049 - val_sparse_categorical_accuracy: 0.9667 - learning_rate: 2.5000e-04 ## Epoch 164/500 ## 90/90 - 0s - 2ms/step - loss: 0.0615 - sparse_categorical_accuracy: 0.9816 - val_loss: 0.1094 - val_sparse_categorical_accuracy: 0.9515 - learning_rate: 2.5000e-04 ## Epoch 165/500 ## 90/90 - 0s - 2ms/step - loss: 0.0594 - sparse_categorical_accuracy: 0.9809 - val_loss: 0.1116 - val_sparse_categorical_accuracy: 0.9515 - learning_rate: 2.5000e-04 ## Epoch 166/500 ## 90/90 - 0s - 1ms/step - loss: 0.0586 - sparse_categorical_accuracy: 0.9806 - val_loss: 0.1275 - val_sparse_categorical_accuracy: 0.9501 - learning_rate: 2.5000e-04 ## Epoch 167/500 ## 90/90 - 0s - 1ms/step - loss: 0.0605 - sparse_categorical_accuracy: 0.9795 - val_loss: 0.1516 - val_sparse_categorical_accuracy: 0.9417 - learning_rate: 2.5000e-04 ## Epoch 168/500 ## 90/90 - 0s - 1ms/step - loss: 0.0696 - sparse_categorical_accuracy: 0.9785 - val_loss: 0.1038 - val_sparse_categorical_accuracy: 0.9612 - learning_rate: 2.5000e-04 ## Epoch 169/500 ## 90/90 - 0s - 1ms/step - loss: 0.0612 - sparse_categorical_accuracy: 0.9799 - val_loss: 0.1354 - val_sparse_categorical_accuracy: 0.9598 - learning_rate: 2.5000e-04 ## Epoch 170/500 ## 90/90 - 0s - 1ms/step - loss: 0.0626 - sparse_categorical_accuracy: 0.9802 - val_loss: 0.1219 - val_sparse_categorical_accuracy: 0.9598 - learning_rate: 2.5000e-04 ## Epoch 171/500 ## 90/90 - 0s - 1ms/step - loss: 0.0582 - sparse_categorical_accuracy: 0.9816 - val_loss: 0.1627 - val_sparse_categorical_accuracy: 0.9417 - learning_rate: 2.5000e-04 ## Epoch 172/500 ## 90/90 - 0s - 2ms/step - loss: 0.0601 - sparse_categorical_accuracy: 0.9795 - val_loss: 0.1023 - val_sparse_categorical_accuracy: 0.9723 - learning_rate: 2.5000e-04 ## Epoch 173/500 ## 90/90 - 0s - 2ms/step - loss: 0.0587 - sparse_categorical_accuracy: 0.9813 - val_loss: 0.1001 - val_sparse_categorical_accuracy: 0.9639 - learning_rate: 2.5000e-04 ## Epoch 174/500 ## 90/90 - 0s - 2ms/step - loss: 0.0611 - sparse_categorical_accuracy: 0.9806 - val_loss: 0.1214 - val_sparse_categorical_accuracy: 0.9487 - learning_rate: 2.5000e-04 ## Epoch 175/500 ## 90/90 - 0s - 1ms/step - loss: 0.0630 - sparse_categorical_accuracy: 0.9788 - val_loss: 0.1159 - val_sparse_categorical_accuracy: 0.9556 - learning_rate: 2.5000e-04 ## Epoch 176/500 ## 90/90 - 0s - 1ms/step - loss: 0.0593 - sparse_categorical_accuracy: 0.9792 - val_loss: 0.1535 - val_sparse_categorical_accuracy: 0.9473 - learning_rate: 2.5000e-04 ## Epoch 177/500 ## 90/90 - 0s - 2ms/step - loss: 0.0579 - sparse_categorical_accuracy: 0.9844 - val_loss: 0.1511 - val_sparse_categorical_accuracy: 0.9473 - learning_rate: 2.5000e-04 ## Epoch 178/500 ## 90/90 - 0s - 2ms/step - loss: 0.0552 - sparse_categorical_accuracy: 0.9826 - val_loss: 0.2204 - val_sparse_categorical_accuracy: 0.9140 - learning_rate: 2.5000e-04 ## Epoch 179/500 ## 90/90 - 0s - 2ms/step - loss: 0.0572 - sparse_categorical_accuracy: 0.9799 - val_loss: 0.1048 - val_sparse_categorical_accuracy: 0.9681 - learning_rate: 2.5000e-04 ## Epoch 180/500 ## 90/90 - 0s - 2ms/step - loss: 0.0554 - sparse_categorical_accuracy: 0.9823 - val_loss: 0.1197 - val_sparse_categorical_accuracy: 0.9695 - learning_rate: 1.2500e-04 ## Epoch 181/500 ## 90/90 - 0s - 1ms/step - loss: 0.0575 - sparse_categorical_accuracy: 0.9837 - val_loss: 0.1206 - val_sparse_categorical_accuracy: 0.9653 - learning_rate: 1.2500e-04 ## Epoch 182/500 ## 90/90 - 0s - 1ms/step - loss: 0.0537 - sparse_categorical_accuracy: 0.9833 - val_loss: 0.1074 - val_sparse_categorical_accuracy: 0.9667 - learning_rate: 1.2500e-04 ## Epoch 183/500 ## 90/90 - 0s - 2ms/step - loss: 0.0571 - sparse_categorical_accuracy: 0.9809 - val_loss: 0.1052 - val_sparse_categorical_accuracy: 0.9695 - learning_rate: 1.2500e-04 ## Epoch 184/500 ## 90/90 - 0s - 2ms/step - loss: 0.0528 - sparse_categorical_accuracy: 0.9830 - val_loss: 0.1022 - val_sparse_categorical_accuracy: 0.9709 - learning_rate: 1.2500e-04 ## Epoch 185/500 ## 90/90 - 0s - 1ms/step - loss: 0.0550 - sparse_categorical_accuracy: 0.9830 - val_loss: 0.1042 - val_sparse_categorical_accuracy: 0.9653 - learning_rate: 1.2500e-04 ## Epoch 186/500 ## 90/90 - 0s - 2ms/step - loss: 0.0605 - sparse_categorical_accuracy: 0.9792 - val_loss: 0.1048 - val_sparse_categorical_accuracy: 0.9695 - learning_rate: 1.2500e-04 ## Epoch 187/500 ## 90/90 - 0s - 1ms/step - loss: 0.0539 - sparse_categorical_accuracy: 0.9847 - val_loss: 0.1090 - val_sparse_categorical_accuracy: 0.9639 - learning_rate: 1.2500e-04 ## Epoch 188/500 ## 90/90 - 0s - 2ms/step - loss: 0.0540 - sparse_categorical_accuracy: 0.9840 - val_loss: 0.1189 - val_sparse_categorical_accuracy: 0.9653 - learning_rate: 1.2500e-04 ## Epoch 189/500 ## 90/90 - 0s - 2ms/step - loss: 0.0514 - sparse_categorical_accuracy: 0.9854 - val_loss: 0.1599 - val_sparse_categorical_accuracy: 0.9431 - learning_rate: 1.2500e-04 ## Epoch 189: early stopping"},{"path":"https://keras3.posit.co/dev/articles/examples/timeseries/timeseries_classification_from_scratch.html","id":"evaluate-model-on-test-data","dir":"Articles > Examples > Timeseries","previous_headings":"","what":"Evaluate model on test data","title":"Timeseries classification from scratch","text":"","code":"model <- load_model(\"best_model.keras\")  results <- model |> evaluate(x_test, y_test) ## 42/42 - 1s - 16ms/step - loss: 0.0998 - sparse_categorical_accuracy: 0.9652 str(results) ## List of 2 ##  $ loss                       : num 0.0998 ##  $ sparse_categorical_accuracy: num 0.965 cat(   \"Test accuracy: \", results$sparse_categorical_accuracy, \"\\n\",   \"Test loss: \", results$loss, \"\\n\",   sep = \"\" ) ## Test accuracy: 0.9651515 ## Test loss: 0.0998451"},{"path":"https://keras3.posit.co/dev/articles/examples/timeseries/timeseries_classification_from_scratch.html","id":"plot-the-models-training-history","dir":"Articles > Examples > Timeseries","previous_headings":"","what":"Plot the model’s training history","title":"Timeseries classification from scratch","text":"Plot just training validation accuracy: can see training accuracy reaches almost 0.95 100 epochs. However, observing validation accuracy can see network still needs training reaches almost 0.97 validation training accuracy 200 epochs. Beyond 200th epoch, continue training, validation accuracy start decreasing training accuracy continue increasing: model starts overfitting.","code":"plot(history) plot(history, metric = \"sparse_categorical_accuracy\") +   # scale x axis to actual number of epochs run before early stopping   ggplot2::xlim(0, length(history$metrics$loss))"},{"path":"https://keras3.posit.co/dev/articles/examples/vision/autoencoder.html","id":"introduction","dir":"Articles > Examples > Vision","previous_headings":"","what":"Introduction","title":"Convolutional autoencoder for image denoising","text":"example demonstrates implement deep convolutional autoencoder image denoising, mapping noisy digits images MNIST dataset clean digits images. implementation based original blog post titled Building Autoencoders Keras François Chollet.","code":""},{"path":"https://keras3.posit.co/dev/articles/examples/vision/autoencoder.html","id":"setup","dir":"Articles > Examples > Vision","previous_headings":"","what":"Setup","title":"Convolutional autoencoder for image denoising","text":"","code":"library(keras3)  # Normalizes the supplied array and reshapes it. preprocess <- function(array) {   array_reshape(array/255, c(dim(array)[1], 28, 28, 1)) }  # Adds random noise to each image in the supplied array. noise <- function(array) {   noise_factor <- 0.4   noisy_array <- array + noise_factor * random_normal(dim(array))   op_clip(noisy_array, 0.0, 1.0) }  display <- function(array1, array2) {   n <- 2   indices <- sample.int(dim(array1)[1], n)   images1 <- as.array(array1)[indices, , , ]   images2 <- as.array(array2)[indices, , , ]    par(mfrow = c(2, n), mar = c(0, 0, 0, 0))   for (i in seq_len(n)) {     plot(as.raster(images1[i, , ]))     plot(as.raster(images2[i, , ]))   } }"},{"path":"https://keras3.posit.co/dev/articles/examples/vision/autoencoder.html","id":"prepare-the-data","dir":"Articles > Examples > Vision","previous_headings":"","what":"Prepare the data","title":"Convolutional autoencoder for image denoising","text":"","code":"# Since we only need images from the dataset to encode and decode, we # won't use the labels. c(c(train_data, .), c(test_data, .)) %<-% dataset_mnist()  # Normalize and reshape the data train_data <- preprocess(train_data) test_data <- preprocess(test_data)  # Create a copy of the data with added noise noisy_train_data <- noise(train_data) noisy_test_data <- noise(test_data)  # Display the train data and a version of it with added noise display(train_data, noisy_train_data)"},{"path":"https://keras3.posit.co/dev/articles/examples/vision/autoencoder.html","id":"build-the-autoencoder","dir":"Articles > Examples > Vision","previous_headings":"","what":"Build the autoencoder","title":"Convolutional autoencoder for image denoising","text":"going use Functional API build convolutional autoencoder. Now can train autoencoder using train_data input data target. Notice setting validation data using format. Let’s predict test dataset display original image together prediction autoencoder. Notice predictions pretty close original images, although quite . Now know autoencoder works, let’s retrain using noisy data input clean data target. want autoencoder learn denoise images. Let’s now predict noisy data display results autoencoder. Notice autoencoder amazing job removing noise input images.","code":"input <- keras_input(shape = c(28, 28, 1))  # Encoder enc <- input |>   layer_conv_2d(filters = 32, kernel_size = c(3, 3),                 activation = \"relu\", padding = \"same\") |>   layer_max_pooling_2d(pool_size = c(2, 2), padding = \"same\") |>   layer_conv_2d(filters = 32, kernel_size = c(3, 3),                 activation = \"relu\", padding = \"same\") |>   layer_max_pooling_2d(pool_size = c(2, 2), padding = \"same\")  # Decoder dec <- enc |>   layer_conv_2d_transpose(filters = 32, kernel_size = c(3, 3), strides = 2,                           activation = \"relu\", padding = \"same\") |>   layer_conv_2d_transpose(filters = 32, kernel_size = c(3, 3), strides = 2,                           activation = \"relu\", padding = \"same\") |>   layer_conv_2d(filters = 1, kernel_size = c(3, 3),                 activation = \"sigmoid\", padding = \"same\")  # Autoencoder autoencoder <- keras_model(input, dec) autoencoder |> compile(optimizer = \"adam\", loss = \"binary_crossentropy\") autoencoder |> summary() ## Model: \"functional\" ## ┏━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━┓ ## ┃ Layer (type)                    ┃ Output Shape           ┃       Param # ┃ ## ┡━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━┩ ## │ input_layer (InputLayer)        │ (None, 28, 28, 1)      │             0 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ conv2d (Conv2D)                 │ (None, 28, 28, 32)     │           320 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ max_pooling2d (MaxPooling2D)    │ (None, 14, 14, 32)     │             0 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ conv2d_1 (Conv2D)               │ (None, 14, 14, 32)     │         9,248 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ max_pooling2d_1 (MaxPooling2D)  │ (None, 7, 7, 32)       │             0 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ conv2d_transpose                │ (None, 14, 14, 32)     │         9,248 │ ## │ (Conv2DTranspose)               │                        │               │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ conv2d_transpose_1              │ (None, 28, 28, 32)     │         9,248 │ ## │ (Conv2DTranspose)               │                        │               │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ conv2d_2 (Conv2D)               │ (None, 28, 28, 1)      │           289 │ ## └─────────────────────────────────┴────────────────────────┴───────────────┘ ##  Total params: 28,353 (110.75 KB) ##  Trainable params: 28,353 (110.75 KB) ##  Non-trainable params: 0 (0.00 B) autoencoder |> fit(   x = train_data,   y = train_data,   epochs = 50,   batch_size = 128,   shuffle = TRUE,   validation_data = list(test_data, test_data), ) ## Epoch 1/50 ## 469/469 - 6s - 12ms/step - loss: 0.1371 - val_loss: 0.0727 ## Epoch 2/50 ## 469/469 - 2s - 3ms/step - loss: 0.0712 - val_loss: 0.0692 ## Epoch 3/50 ## 469/469 - 2s - 3ms/step - loss: 0.0691 - val_loss: 0.0679 ## Epoch 4/50 ## 469/469 - 2s - 4ms/step - loss: 0.0680 - val_loss: 0.0671 ## Epoch 5/50 ## 469/469 - 2s - 4ms/step - loss: 0.0672 - val_loss: 0.0665 ## Epoch 6/50 ## 469/469 - 2s - 4ms/step - loss: 0.0667 - val_loss: 0.0660 ## Epoch 7/50 ## 469/469 - 2s - 3ms/step - loss: 0.0662 - val_loss: 0.0656 ## Epoch 8/50 ## 469/469 - 2s - 3ms/step - loss: 0.0659 - val_loss: 0.0653 ## Epoch 9/50 ## 469/469 - 2s - 4ms/step - loss: 0.0656 - val_loss: 0.0650 ## Epoch 10/50 ## 469/469 - 2s - 4ms/step - loss: 0.0654 - val_loss: 0.0648 ## Epoch 11/50 ## 469/469 - 2s - 4ms/step - loss: 0.0652 - val_loss: 0.0646 ## Epoch 12/50 ## 469/469 - 2s - 4ms/step - loss: 0.0650 - val_loss: 0.0644 ## Epoch 13/50 ## 469/469 - 2s - 4ms/step - loss: 0.0648 - val_loss: 0.0643 ## Epoch 14/50 ## 469/469 - 2s - 4ms/step - loss: 0.0646 - val_loss: 0.0641 ## Epoch 15/50 ## 469/469 - 2s - 4ms/step - loss: 0.0645 - val_loss: 0.0640 ## Epoch 16/50 ## 469/469 - 2s - 4ms/step - loss: 0.0644 - val_loss: 0.0639 ## Epoch 17/50 ## 469/469 - 2s - 4ms/step - loss: 0.0643 - val_loss: 0.0638 ## Epoch 18/50 ## 469/469 - 2s - 4ms/step - loss: 0.0642 - val_loss: 0.0637 ## Epoch 19/50 ## 469/469 - 2s - 4ms/step - loss: 0.0641 - val_loss: 0.0636 ## Epoch 20/50 ## 469/469 - 2s - 4ms/step - loss: 0.0640 - val_loss: 0.0635 ## Epoch 21/50 ## 469/469 - 2s - 4ms/step - loss: 0.0639 - val_loss: 0.0635 ## Epoch 22/50 ## 469/469 - 2s - 4ms/step - loss: 0.0638 - val_loss: 0.0634 ## Epoch 23/50 ## 469/469 - 2s - 4ms/step - loss: 0.0638 - val_loss: 0.0634 ## Epoch 24/50 ## 469/469 - 2s - 4ms/step - loss: 0.0637 - val_loss: 0.0633 ## Epoch 25/50 ## 469/469 - 2s - 4ms/step - loss: 0.0637 - val_loss: 0.0633 ## Epoch 26/50 ## 469/469 - 2s - 4ms/step - loss: 0.0636 - val_loss: 0.0632 ## Epoch 27/50 ## 469/469 - 2s - 4ms/step - loss: 0.0636 - val_loss: 0.0632 ## Epoch 28/50 ## 469/469 - 2s - 4ms/step - loss: 0.0635 - val_loss: 0.0631 ## Epoch 29/50 ## 469/469 - 2s - 4ms/step - loss: 0.0635 - val_loss: 0.0631 ## Epoch 30/50 ## 469/469 - 2s - 4ms/step - loss: 0.0634 - val_loss: 0.0630 ## Epoch 31/50 ## 469/469 - 2s - 4ms/step - loss: 0.0634 - val_loss: 0.0630 ## Epoch 32/50 ## 469/469 - 2s - 4ms/step - loss: 0.0633 - val_loss: 0.0629 ## Epoch 33/50 ## 469/469 - 2s - 4ms/step - loss: 0.0633 - val_loss: 0.0629 ## Epoch 34/50 ## 469/469 - 2s - 4ms/step - loss: 0.0633 - val_loss: 0.0629 ## Epoch 35/50 ## 469/469 - 2s - 4ms/step - loss: 0.0632 - val_loss: 0.0628 ## Epoch 36/50 ## 469/469 - 2s - 4ms/step - loss: 0.0632 - val_loss: 0.0628 ## Epoch 37/50 ## 469/469 - 2s - 4ms/step - loss: 0.0632 - val_loss: 0.0628 ## Epoch 38/50 ## 469/469 - 2s - 4ms/step - loss: 0.0631 - val_loss: 0.0628 ## Epoch 39/50 ## 469/469 - 2s - 4ms/step - loss: 0.0631 - val_loss: 0.0627 ## Epoch 40/50 ## 469/469 - 2s - 4ms/step - loss: 0.0631 - val_loss: 0.0627 ## Epoch 41/50 ## 469/469 - 2s - 4ms/step - loss: 0.0630 - val_loss: 0.0627 ## Epoch 42/50 ## 469/469 - 2s - 4ms/step - loss: 0.0630 - val_loss: 0.0627 ## Epoch 43/50 ## 469/469 - 2s - 4ms/step - loss: 0.0630 - val_loss: 0.0626 ## Epoch 44/50 ## 469/469 - 2s - 4ms/step - loss: 0.0630 - val_loss: 0.0626 ## Epoch 45/50 ## 469/469 - 2s - 4ms/step - loss: 0.0629 - val_loss: 0.0626 ## Epoch 46/50 ## 469/469 - 2s - 4ms/step - loss: 0.0629 - val_loss: 0.0626 ## Epoch 47/50 ## 469/469 - 2s - 4ms/step - loss: 0.0629 - val_loss: 0.0625 ## Epoch 48/50 ## 469/469 - 2s - 4ms/step - loss: 0.0629 - val_loss: 0.0625 ## Epoch 49/50 ## 469/469 - 2s - 4ms/step - loss: 0.0628 - val_loss: 0.0625 ## Epoch 50/50 ## 469/469 - 2s - 4ms/step - loss: 0.0628 - val_loss: 0.0625 predictions <- autoencoder |> predict(test_data) ## 313/313 - 1s - 3ms/step display(test_data, predictions) autoencoder |> fit(   x = noisy_train_data,   y = train_data,   epochs = 100,   batch_size = 128,   shuffle = TRUE,   validation_data = list(noisy_test_data, test_data), ) ## Epoch 1/100 ## 469/469 - 2s - 5ms/step - loss: 0.1009 - val_loss: 0.0939 ## Epoch 2/100 ## 469/469 - 2s - 4ms/step - loss: 0.0936 - val_loss: 0.0919 ## Epoch 3/100 ## 469/469 - 2s - 4ms/step - loss: 0.0921 - val_loss: 0.0908 ## Epoch 4/100 ## 469/469 - 2s - 4ms/step - loss: 0.0911 - val_loss: 0.0900 ## Epoch 5/100 ## 469/469 - 2s - 4ms/step - loss: 0.0903 - val_loss: 0.0895 ## Epoch 6/100 ## 469/469 - 2s - 4ms/step - loss: 0.0897 - val_loss: 0.0889 ## Epoch 7/100 ## 469/469 - 2s - 4ms/step - loss: 0.0893 - val_loss: 0.0885 ## Epoch 8/100 ## 469/469 - 2s - 4ms/step - loss: 0.0889 - val_loss: 0.0881 ## Epoch 9/100 ## 469/469 - 2s - 4ms/step - loss: 0.0886 - val_loss: 0.0879 ## Epoch 10/100 ## 469/469 - 2s - 4ms/step - loss: 0.0883 - val_loss: 0.0876 ## Epoch 11/100 ## 469/469 - 2s - 4ms/step - loss: 0.0880 - val_loss: 0.0874 ## Epoch 12/100 ## 469/469 - 2s - 4ms/step - loss: 0.0878 - val_loss: 0.0872 ## Epoch 13/100 ## 469/469 - 2s - 4ms/step - loss: 0.0876 - val_loss: 0.0870 ## Epoch 14/100 ## 469/469 - 2s - 4ms/step - loss: 0.0875 - val_loss: 0.0868 ## Epoch 15/100 ## 469/469 - 2s - 4ms/step - loss: 0.0873 - val_loss: 0.0867 ## Epoch 16/100 ## 469/469 - 2s - 4ms/step - loss: 0.0872 - val_loss: 0.0866 ## Epoch 17/100 ## 469/469 - 2s - 4ms/step - loss: 0.0870 - val_loss: 0.0865 ## Epoch 18/100 ## 469/469 - 2s - 4ms/step - loss: 0.0869 - val_loss: 0.0864 ## Epoch 19/100 ## 469/469 - 2s - 4ms/step - loss: 0.0868 - val_loss: 0.0863 ## Epoch 20/100 ## 469/469 - 2s - 4ms/step - loss: 0.0867 - val_loss: 0.0862 ## Epoch 21/100 ## 469/469 - 2s - 4ms/step - loss: 0.0866 - val_loss: 0.0861 ## Epoch 22/100 ## 469/469 - 2s - 4ms/step - loss: 0.0865 - val_loss: 0.0860 ## Epoch 23/100 ## 469/469 - 2s - 4ms/step - loss: 0.0865 - val_loss: 0.0859 ## Epoch 24/100 ## 469/469 - 2s - 4ms/step - loss: 0.0864 - val_loss: 0.0859 ## Epoch 25/100 ## 469/469 - 2s - 4ms/step - loss: 0.0863 - val_loss: 0.0858 ## Epoch 26/100 ## 469/469 - 2s - 4ms/step - loss: 0.0862 - val_loss: 0.0857 ## Epoch 27/100 ## 469/469 - 2s - 4ms/step - loss: 0.0862 - val_loss: 0.0857 ## Epoch 28/100 ## 469/469 - 2s - 4ms/step - loss: 0.0861 - val_loss: 0.0856 ## Epoch 29/100 ## 469/469 - 2s - 4ms/step - loss: 0.0860 - val_loss: 0.0856 ## Epoch 30/100 ## 469/469 - 2s - 4ms/step - loss: 0.0860 - val_loss: 0.0855 ## Epoch 31/100 ## 469/469 - 2s - 4ms/step - loss: 0.0859 - val_loss: 0.0854 ## Epoch 32/100 ## 469/469 - 2s - 4ms/step - loss: 0.0859 - val_loss: 0.0854 ## Epoch 33/100 ## 469/469 - 2s - 4ms/step - loss: 0.0858 - val_loss: 0.0854 ## Epoch 34/100 ## 469/469 - 2s - 4ms/step - loss: 0.0858 - val_loss: 0.0853 ## Epoch 35/100 ## 469/469 - 2s - 4ms/step - loss: 0.0857 - val_loss: 0.0853 ## Epoch 36/100 ## 469/469 - 2s - 4ms/step - loss: 0.0857 - val_loss: 0.0852 ## Epoch 37/100 ## 469/469 - 2s - 4ms/step - loss: 0.0856 - val_loss: 0.0852 ## Epoch 38/100 ## 469/469 - 2s - 4ms/step - loss: 0.0856 - val_loss: 0.0851 ## Epoch 39/100 ## 469/469 - 2s - 4ms/step - loss: 0.0856 - val_loss: 0.0851 ## Epoch 40/100 ## 469/469 - 2s - 4ms/step - loss: 0.0855 - val_loss: 0.0851 ## Epoch 41/100 ## 469/469 - 2s - 4ms/step - loss: 0.0855 - val_loss: 0.0850 ## Epoch 42/100 ## 469/469 - 2s - 4ms/step - loss: 0.0854 - val_loss: 0.0850 ## Epoch 43/100 ## 469/469 - 2s - 4ms/step - loss: 0.0854 - val_loss: 0.0850 ## Epoch 44/100 ## 469/469 - 2s - 4ms/step - loss: 0.0854 - val_loss: 0.0850 ## Epoch 45/100 ## 469/469 - 2s - 4ms/step - loss: 0.0853 - val_loss: 0.0849 ## Epoch 46/100 ## 469/469 - 2s - 4ms/step - loss: 0.0853 - val_loss: 0.0849 ## Epoch 47/100 ## 469/469 - 2s - 4ms/step - loss: 0.0853 - val_loss: 0.0849 ## Epoch 48/100 ## 469/469 - 2s - 4ms/step - loss: 0.0853 - val_loss: 0.0849 ## Epoch 49/100 ## 469/469 - 2s - 4ms/step - loss: 0.0852 - val_loss: 0.0849 ## Epoch 50/100 ## 469/469 - 2s - 4ms/step - loss: 0.0852 - val_loss: 0.0848 ## Epoch 51/100 ## 469/469 - 2s - 4ms/step - loss: 0.0852 - val_loss: 0.0848 ## Epoch 52/100 ## 469/469 - 2s - 4ms/step - loss: 0.0852 - val_loss: 0.0848 ## Epoch 53/100 ## 469/469 - 2s - 4ms/step - loss: 0.0851 - val_loss: 0.0848 ## Epoch 54/100 ## 469/469 - 2s - 4ms/step - loss: 0.0851 - val_loss: 0.0848 ## Epoch 55/100 ## 469/469 - 2s - 4ms/step - loss: 0.0851 - val_loss: 0.0847 ## Epoch 56/100 ## 469/469 - 2s - 4ms/step - loss: 0.0851 - val_loss: 0.0847 ## Epoch 57/100 ## 469/469 - 2s - 4ms/step - loss: 0.0850 - val_loss: 0.0847 ## Epoch 58/100 ## 469/469 - 2s - 4ms/step - loss: 0.0850 - val_loss: 0.0847 ## Epoch 59/100 ## 469/469 - 2s - 4ms/step - loss: 0.0850 - val_loss: 0.0847 ## Epoch 60/100 ## 469/469 - 2s - 4ms/step - loss: 0.0850 - val_loss: 0.0847 ## Epoch 61/100 ## 469/469 - 2s - 4ms/step - loss: 0.0850 - val_loss: 0.0847 ## Epoch 62/100 ## 469/469 - 2s - 4ms/step - loss: 0.0850 - val_loss: 0.0846 ## Epoch 63/100 ## 469/469 - 2s - 4ms/step - loss: 0.0849 - val_loss: 0.0846 ## Epoch 64/100 ## 469/469 - 2s - 4ms/step - loss: 0.0849 - val_loss: 0.0846 ## Epoch 65/100 ## 469/469 - 2s - 4ms/step - loss: 0.0849 - val_loss: 0.0846 ## Epoch 66/100 ## 469/469 - 2s - 4ms/step - loss: 0.0849 - val_loss: 0.0846 ## Epoch 67/100 ## 469/469 - 2s - 4ms/step - loss: 0.0849 - val_loss: 0.0846 ## Epoch 68/100 ## 469/469 - 2s - 4ms/step - loss: 0.0849 - val_loss: 0.0845 ## Epoch 69/100 ## 469/469 - 2s - 4ms/step - loss: 0.0848 - val_loss: 0.0845 ## Epoch 70/100 ## 469/469 - 2s - 4ms/step - loss: 0.0848 - val_loss: 0.0845 ## Epoch 71/100 ## 469/469 - 2s - 4ms/step - loss: 0.0848 - val_loss: 0.0845 ## Epoch 72/100 ## 469/469 - 2s - 4ms/step - loss: 0.0848 - val_loss: 0.0845 ## Epoch 73/100 ## 469/469 - 2s - 4ms/step - loss: 0.0848 - val_loss: 0.0845 ## Epoch 74/100 ## 469/469 - 2s - 4ms/step - loss: 0.0848 - val_loss: 0.0845 ## Epoch 75/100 ## 469/469 - 2s - 4ms/step - loss: 0.0848 - val_loss: 0.0845 ## Epoch 76/100 ## 469/469 - 2s - 4ms/step - loss: 0.0848 - val_loss: 0.0845 ## Epoch 77/100 ## 469/469 - 2s - 4ms/step - loss: 0.0847 - val_loss: 0.0845 ## Epoch 78/100 ## 469/469 - 2s - 4ms/step - loss: 0.0847 - val_loss: 0.0844 ## Epoch 79/100 ## 469/469 - 2s - 4ms/step - loss: 0.0847 - val_loss: 0.0844 ## Epoch 80/100 ## 469/469 - 2s - 4ms/step - loss: 0.0847 - val_loss: 0.0844 ## Epoch 81/100 ## 469/469 - 2s - 4ms/step - loss: 0.0847 - val_loss: 0.0844 ## Epoch 82/100 ## 469/469 - 2s - 4ms/step - loss: 0.0847 - val_loss: 0.0844 ## Epoch 83/100 ## 469/469 - 2s - 4ms/step - loss: 0.0847 - val_loss: 0.0844 ## Epoch 84/100 ## 469/469 - 2s - 4ms/step - loss: 0.0847 - val_loss: 0.0844 ## Epoch 85/100 ## 469/469 - 2s - 4ms/step - loss: 0.0847 - val_loss: 0.0844 ## Epoch 86/100 ## 469/469 - 2s - 4ms/step - loss: 0.0846 - val_loss: 0.0844 ## Epoch 87/100 ## 469/469 - 2s - 4ms/step - loss: 0.0846 - val_loss: 0.0844 ## Epoch 88/100 ## 469/469 - 2s - 4ms/step - loss: 0.0846 - val_loss: 0.0844 ## Epoch 89/100 ## 469/469 - 2s - 4ms/step - loss: 0.0846 - val_loss: 0.0843 ## Epoch 90/100 ## 469/469 - 2s - 4ms/step - loss: 0.0846 - val_loss: 0.0843 ## Epoch 91/100 ## 469/469 - 2s - 4ms/step - loss: 0.0846 - val_loss: 0.0843 ## Epoch 92/100 ## 469/469 - 2s - 4ms/step - loss: 0.0846 - val_loss: 0.0843 ## Epoch 93/100 ## 469/469 - 2s - 4ms/step - loss: 0.0846 - val_loss: 0.0843 ## Epoch 94/100 ## 469/469 - 2s - 4ms/step - loss: 0.0846 - val_loss: 0.0843 ## Epoch 95/100 ## 469/469 - 2s - 4ms/step - loss: 0.0846 - val_loss: 0.0843 ## Epoch 96/100 ## 469/469 - 2s - 4ms/step - loss: 0.0846 - val_loss: 0.0843 ## Epoch 97/100 ## 469/469 - 2s - 4ms/step - loss: 0.0845 - val_loss: 0.0843 ## Epoch 98/100 ## 469/469 - 2s - 4ms/step - loss: 0.0845 - val_loss: 0.0843 ## Epoch 99/100 ## 469/469 - 2s - 4ms/step - loss: 0.0845 - val_loss: 0.0843 ## Epoch 100/100 ## 469/469 - 2s - 4ms/step - loss: 0.0845 - val_loss: 0.0843 predictions <- autoencoder |> predict(noisy_test_data) ## 313/313 - 0s - 1ms/step display(noisy_test_data, predictions)"},{"path":"https://keras3.posit.co/dev/articles/examples/vision/mnist_convnet.html","id":"setup","dir":"Articles > Examples > Vision","previous_headings":"","what":"Setup","title":"Simple MNIST convnet","text":"","code":"library(keras3)"},{"path":"https://keras3.posit.co/dev/articles/examples/vision/mnist_convnet.html","id":"prepare-the-data","dir":"Articles > Examples > Vision","previous_headings":"","what":"Prepare the data","title":"Simple MNIST convnet","text":"","code":"# Model / data parameters num_classes <- 10 input_shape <- c(28, 28, 1)  # Load the data and split it between train and test sets c(c(x_train, y_train), c(x_test, y_test)) %<-% dataset_mnist()  # Scale images to the [0, 1] range x_train <- x_train / 255 x_test <- x_test / 255 # Make sure images have shape (28, 28, 1) x_train <- op_expand_dims(x_train, -1) x_test <- op_expand_dims(x_test, -1)   dim(x_train) ## [1] 60000    28    28     1 dim(x_test) ## [1] 10000    28    28     1 # convert class vectors to binary class matrices y_train <- to_categorical(y_train, num_classes) y_test <- to_categorical(y_test, num_classes)"},{"path":"https://keras3.posit.co/dev/articles/examples/vision/mnist_convnet.html","id":"build-the-model","dir":"Articles > Examples > Vision","previous_headings":"","what":"Build the model","title":"Simple MNIST convnet","text":"","code":"model <- keras_model_sequential(input_shape = input_shape) model |>   layer_conv_2d(filters = 32, kernel_size = c(3, 3), activation = \"relu\") |>   layer_max_pooling_2d(pool_size = c(2, 2)) |>   layer_conv_2d(filters = 64, kernel_size = c(3, 3), activation = \"relu\") |>   layer_max_pooling_2d(pool_size = c(2, 2)) |>   layer_flatten() |>   layer_dropout(rate = 0.5) |>   layer_dense(units = num_classes, activation = \"softmax\")  summary(model) ## Model: \"sequential\" ## ┏━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━┓ ## ┃ Layer (type)                    ┃ Output Shape           ┃       Param # ┃ ## ┡━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━┩ ## │ conv2d (Conv2D)                 │ (None, 26, 26, 32)     │           320 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ max_pooling2d (MaxPooling2D)    │ (None, 13, 13, 32)     │             0 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ conv2d_1 (Conv2D)               │ (None, 11, 11, 64)     │        18,496 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ max_pooling2d_1 (MaxPooling2D)  │ (None, 5, 5, 64)       │             0 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ flatten (Flatten)               │ (None, 1600)           │             0 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ dropout (Dropout)               │ (None, 1600)           │             0 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ dense (Dense)                   │ (None, 10)             │        16,010 │ ## └─────────────────────────────────┴────────────────────────┴───────────────┘ ##  Total params: 34,826 (136.04 KB) ##  Trainable params: 34,826 (136.04 KB) ##  Non-trainable params: 0 (0.00 B)"},{"path":"https://keras3.posit.co/dev/articles/examples/vision/mnist_convnet.html","id":"train-the-model","dir":"Articles > Examples > Vision","previous_headings":"","what":"Train the model","title":"Simple MNIST convnet","text":"","code":"batch_size <- 128 epochs <- 15  model |> compile(   loss = \"categorical_crossentropy\",   optimizer = \"adam\",   metrics = \"accuracy\" )  model |> fit(   x_train, y_train,   batch_size = batch_size,   epochs = epochs,   validation_split = 0.1 ) ## Epoch 1/15 ## 422/422 - 5s - 13ms/step - accuracy: 0.8894 - loss: 0.3635 - val_accuracy: 0.9787 - val_loss: 0.0791 ## Epoch 2/15 ## 422/422 - 1s - 3ms/step - accuracy: 0.9665 - loss: 0.1111 - val_accuracy: 0.9848 - val_loss: 0.0550 ## Epoch 3/15 ## 422/422 - 1s - 2ms/step - accuracy: 0.9743 - loss: 0.0824 - val_accuracy: 0.9883 - val_loss: 0.0441 ## Epoch 4/15 ## 422/422 - 1s - 2ms/step - accuracy: 0.9785 - loss: 0.0695 - val_accuracy: 0.9895 - val_loss: 0.0397 ## Epoch 5/15 ## 422/422 - 1s - 2ms/step - accuracy: 0.9803 - loss: 0.0626 - val_accuracy: 0.9903 - val_loss: 0.0352 ## Epoch 6/15 ## 422/422 - 1s - 2ms/step - accuracy: 0.9823 - loss: 0.0559 - val_accuracy: 0.9912 - val_loss: 0.0332 ## Epoch 7/15 ## 422/422 - 1s - 2ms/step - accuracy: 0.9837 - loss: 0.0501 - val_accuracy: 0.9917 - val_loss: 0.0311 ## Epoch 8/15 ## 422/422 - 1s - 2ms/step - accuracy: 0.9848 - loss: 0.0481 - val_accuracy: 0.9923 - val_loss: 0.0307 ## Epoch 9/15 ## 422/422 - 1s - 2ms/step - accuracy: 0.9861 - loss: 0.0444 - val_accuracy: 0.9917 - val_loss: 0.0303 ## Epoch 10/15 ## 422/422 - 1s - 2ms/step - accuracy: 0.9862 - loss: 0.0440 - val_accuracy: 0.9913 - val_loss: 0.0298 ## Epoch 11/15 ## 422/422 - 1s - 2ms/step - accuracy: 0.9873 - loss: 0.0394 - val_accuracy: 0.9920 - val_loss: 0.0304 ## Epoch 12/15 ## 422/422 - 1s - 2ms/step - accuracy: 0.9875 - loss: 0.0375 - val_accuracy: 0.9925 - val_loss: 0.0295 ## Epoch 13/15 ## 422/422 - 1s - 2ms/step - accuracy: 0.9887 - loss: 0.0346 - val_accuracy: 0.9920 - val_loss: 0.0297 ## Epoch 14/15 ## 422/422 - 1s - 2ms/step - accuracy: 0.9889 - loss: 0.0343 - val_accuracy: 0.9923 - val_loss: 0.0285 ## Epoch 15/15 ## 422/422 - 1s - 3ms/step - accuracy: 0.9898 - loss: 0.0320 - val_accuracy: 0.9925 - val_loss: 0.0279"},{"path":"https://keras3.posit.co/dev/articles/examples/vision/mnist_convnet.html","id":"evaluate-the-trained-model","dir":"Articles > Examples > Vision","previous_headings":"","what":"Evaluate the trained model","title":"Simple MNIST convnet","text":"","code":"score <- model |> evaluate(x_test, y_test, verbose = 0) score ## $accuracy ## [1] 0.9913 ## ## $loss ## [1] 0.02430356"},{"path":"https://keras3.posit.co/dev/articles/examples/vision/mnist_siamese_graph.html","id":"introduction","dir":"Articles > Examples > Vision","previous_headings":"","what":"Introduction","title":"Train a Siamese MLP on pairs of digits from the MNIST dataset.","text":"Siamese Networks neural networks share weights two sister networks, producing embedding vectors respective inputs. supervised similarity learning, networks trained maximize contrast (distance) embeddings inputs different classes, minimizing distance embeddings similar classes, resulting embedding spaces reflect class segmentation training inputs. implementation loosely follows Hadsell-et-al.’06 [1] (see paper mode details) euclidean distance replaced subtraction layer one fully-connect (FC) layer. [1] “Dimensionality Reduction Learning Invariant Mapping” https://yann.lecun.com/exdb/publis/pdf/hadsell-chopra-lecun-06.pdf Gets 98.11% test accuracy 20 epochs. 3 seconds per epoch AMD Ryzen 7 PRO 4750U (CPU)","code":"library(keras3) contrastive_loss <- function(y_true, y_pred) {     # Contrastive loss from Hadsell-et-al.'06     # https://yann.lecun.com/exdb/publis/pdf/hadsell-chopra-lecun-06.pdf     margin = 1     margin_square = op_square(op_maximum(margin - (y_pred), 0))     op_mean((1 - y_true) * op_square(y_pred) + (y_true) * margin_square) }"},{"path":"https://keras3.posit.co/dev/articles/examples/vision/mnist_siamese_graph.html","id":"create-pairs-of-images","dir":"Articles > Examples > Vision","previous_headings":"","what":"Create pairs of images","title":"Train a Siamese MLP on pairs of digits from the MNIST dataset.","text":"train model differentiate digits different classes. example, digit 0 needs differentiated rest digits (1 9), digit 1 - 0 2 9, . carry , select N random images class (example, digit 0) pair N random images another class B (example, digit 1). , can repeat process classes digits (digit 9). paired digit 0 digits, can repeat process remaining classes rest digits (1 9).","code":"create_pairs <- function(x, y) {     # Positive and negative pair creation.     # Alternates between positive and negative pairs.     digit_indices <- tapply(1:length(y), y, list)     y1 <- y     y2 <- sapply(y, function(a) sample(0:9,1,prob=0.1+0.8*(0:9==a)))     idx1 <- 1:nrow(x)     idx2 <- sapply(as.character(y2), function(a) sample(digit_indices[[a]],1))     is_same  <- 1*(y1==y2)     list(pair1 = x[idx1,], pair2 = x[idx2,], y = is_same) }  compute_accuracy <- function(predictions, labels) {     # Compute classification accuracy with a fixed threshold on distances.     mean(labels[predictions > 0.5]) } # the data, shuffled and split between train and test sets mnist   <- dataset_mnist() x_train <- mnist$train$x y_train <- mnist$train$y x_test  <- mnist$test$x y_test  <- mnist$test$y x_train <- array_reshape(x_train, c(nrow(x_train), 784)) x_test  <- array_reshape(x_test, c(nrow(x_test), 784)) x_train <- x_train / 255 x_test  <- x_test / 255  # create training+test positive and negative pairs tr <- create_pairs(x_train, y_train) te <- create_pairs(x_test,  y_test)  names(tr) ## [1] \"pair1\" \"pair2\" \"y\""},{"path":"https://keras3.posit.co/dev/articles/examples/vision/mnist_siamese_graph.html","id":"network-definition","dir":"Articles > Examples > Vision","previous_headings":"","what":"Network definition","title":"Train a Siamese MLP on pairs of digits from the MNIST dataset.","text":"","code":"# input layers input_dim = 784 input_1 <- layer_input(shape = c(input_dim)) input_2 <- layer_input(shape = c(input_dim))  # definition of the base network that will be shared base_network <- keras_model_sequential() %>%     layer_dense(units = 128, activation = 'relu') %>%     layer_dropout(rate = 0.1) %>%     layer_dense(units = 128, activation = 'relu') %>%     layer_dropout(rate = 0.1) %>%     layer_dense(units = 128, activation = 'relu')  # because we re-use the same instance `base_network`, the weights of # the network will be shared across the two branches branch_1 <- base_network(input_1) branch_2 <- base_network(input_2)  # merging layer out <- layer_subtract(list(branch_1, branch_2)) %>%     layer_dropout(rate = 0.1) %>%     layer_dense(units = 16, activation = 'relu') %>%     layer_dense(1, activation = \"sigmoid\")  # create and compile model model <- keras_model(list(input_1, input_2), out)"},{"path":"https://keras3.posit.co/dev/articles/examples/vision/mnist_siamese_graph.html","id":"train","dir":"Articles > Examples > Vision","previous_headings":"","what":"Train","title":"Train a Siamese MLP on pairs of digits from the MNIST dataset.","text":"","code":"model %>% compile(     optimizer = \"rmsprop\",     #loss = \"binary_crossentropy\",     loss = contrastive_loss,     metrics = metric_binary_accuracy )  history <- model %>% fit(     list(tr$pair1, tr$pair2), tr$y,     batch_size = 128,     epochs = 20,     validation_data = list(         list(te$pair1, te$pair2),         te$y     ) ) ## Epoch 1/20 ## 469/469 - 6s - 13ms/step - binary_accuracy: 0.7569 - loss: 0.1640 - val_binary_accuracy: 0.8798 - val_loss: 0.0974 ## Epoch 2/20 ## 469/469 - 1s - 2ms/step - binary_accuracy: 0.8899 - loss: 0.0866 - val_binary_accuracy: 0.9227 - val_loss: 0.0626 ## Epoch 3/20 ## 469/469 - 1s - 2ms/step - binary_accuracy: 0.9233 - loss: 0.0603 - val_binary_accuracy: 0.9398 - val_loss: 0.0497 ## Epoch 4/20 ## 469/469 - 1s - 2ms/step - binary_accuracy: 0.9423 - loss: 0.0460 - val_binary_accuracy: 0.9480 - val_loss: 0.0399 ## Epoch 5/20 ## 469/469 - 1s - 2ms/step - binary_accuracy: 0.9561 - loss: 0.0356 - val_binary_accuracy: 0.9578 - val_loss: 0.0330 ## Epoch 6/20 ## 469/469 - 1s - 2ms/step - binary_accuracy: 0.9637 - loss: 0.0295 - val_binary_accuracy: 0.9613 - val_loss: 0.0313 ## Epoch 7/20 ## 469/469 - 1s - 2ms/step - binary_accuracy: 0.9686 - loss: 0.0252 - val_binary_accuracy: 0.9645 - val_loss: 0.0275 ## Epoch 8/20 ## 469/469 - 1s - 2ms/step - binary_accuracy: 0.9730 - loss: 0.0217 - val_binary_accuracy: 0.9672 - val_loss: 0.0264 ## Epoch 9/20 ## 469/469 - 1s - 2ms/step - binary_accuracy: 0.9768 - loss: 0.0190 - val_binary_accuracy: 0.9686 - val_loss: 0.0246 ## Epoch 10/20 ## 469/469 - 1s - 2ms/step - binary_accuracy: 0.9793 - loss: 0.0169 - val_binary_accuracy: 0.9701 - val_loss: 0.0242 ## Epoch 11/20 ## 469/469 - 1s - 2ms/step - binary_accuracy: 0.9805 - loss: 0.0158 - val_binary_accuracy: 0.9719 - val_loss: 0.0233 ## Epoch 12/20 ## 469/469 - 1s - 2ms/step - binary_accuracy: 0.9822 - loss: 0.0148 - val_binary_accuracy: 0.9718 - val_loss: 0.0230 ## Epoch 13/20 ## 469/469 - 1s - 2ms/step - binary_accuracy: 0.9833 - loss: 0.0135 - val_binary_accuracy: 0.9702 - val_loss: 0.0234 ## Epoch 14/20 ## 469/469 - 1s - 2ms/step - binary_accuracy: 0.9846 - loss: 0.0129 - val_binary_accuracy: 0.9730 - val_loss: 0.0220 ## Epoch 15/20 ## 469/469 - 1s - 2ms/step - binary_accuracy: 0.9857 - loss: 0.0117 - val_binary_accuracy: 0.9759 - val_loss: 0.0193 ## Epoch 16/20 ## 469/469 - 1s - 2ms/step - binary_accuracy: 0.9868 - loss: 0.0111 - val_binary_accuracy: 0.9749 - val_loss: 0.0217 ## Epoch 17/20 ## 469/469 - 1s - 2ms/step - binary_accuracy: 0.9872 - loss: 0.0106 - val_binary_accuracy: 0.9741 - val_loss: 0.0211 ## Epoch 18/20 ## 469/469 - 1s - 2ms/step - binary_accuracy: 0.9880 - loss: 0.0099 - val_binary_accuracy: 0.9746 - val_loss: 0.0201 ## Epoch 19/20 ## 469/469 - 1s - 2ms/step - binary_accuracy: 0.9879 - loss: 0.0100 - val_binary_accuracy: 0.9756 - val_loss: 0.0207 ## Epoch 20/20 ## 469/469 - 1s - 2ms/step - binary_accuracy: 0.9888 - loss: 0.0091 - val_binary_accuracy: 0.9776 - val_loss: 0.0184 plot(history)"},{"path":"https://keras3.posit.co/dev/articles/examples/vision/mnist_siamese_graph.html","id":"evaluate","dir":"Articles > Examples > Vision","previous_headings":"","what":"Evaluate","title":"Train a Siamese MLP on pairs of digits from the MNIST dataset.","text":"","code":"# compute final accuracy on training and test sets  tr_pred <- predict(model, list(tr$pair1, tr$pair2))[,1] ## 1875/1875 - 2s - 1ms/step tr_acc  <- compute_accuracy(tr_pred, tr$y) te_pred <- predict(model, list(te$pair1, te$pair2))[,1] ## 313/313 - 1s - 2ms/step te_acc  <- compute_accuracy(te_pred, te$y)  sprintf('* Accuracy on training set: %0.2f%%', (100 * tr_acc)) ## [1] \"* Accuracy on training set: 99.67%\" sprintf('* Accuracy on test set: %0.2f%%', (100 * te_acc)) ## [1] \"* Accuracy on test set: 98.25%\""},{"path":"https://keras3.posit.co/dev/articles/examples/vision/mnist_siamese_graph.html","id":"plots","dir":"Articles > Examples > Vision","previous_headings":"","what":"Plots","title":"Train a Siamese MLP on pairs of digits from the MNIST dataset.","text":"","code":"par(mfrow=c(1,1)) vioplot::vioplot( te_pred ~ te$y ) i=3 visualizePair <- function(i) {     image(rbind(matrix( te$pair1[i,],28,28)[,28:1], matrix( te$pair2[i,],28,28)[,28:1]))     title(paste(\"true:\", te$y[i],\"|  pred:\", round(te_pred[i],5))) } par(mfrow=c(3,3)) lapply(1:9, visualizePair) ## [[1]] ## NULL ## ## [[2]] ## NULL ## ## [[3]] ## NULL ## ## [[4]] ## NULL ## ## [[5]] ## NULL ## ## [[6]] ## NULL ## ## [[7]] ## NULL ## ## [[8]] ## NULL ## ## [[9]] ## NULL"},{"path":"https://keras3.posit.co/dev/articles/examples/vision/oxford_pets_image_segmentation.html","id":"download-the-data","dir":"Articles > Examples > Vision","previous_headings":"","what":"Download the data","title":"Image segmentation with a U-Net-like architecture","text":"","code":"options(timeout = 5000) download.file(   \"https://www.robots.ox.ac.uk/~vgg/data/pets/data/images.tar.gz\",   \"datasets/images.tar.gz\" ) download.file(   \"https://www.robots.ox.ac.uk/~vgg/data/pets/data/annotations.tar.gz\",   \"datasets/annotations.tar.gz\" )  untar(\"datasets/images.tar.gz\", exdir = \"datasets\") untar(\"datasets/annotations.tar.gz\", exdir = \"datasets\")"},{"path":"https://keras3.posit.co/dev/articles/examples/vision/oxford_pets_image_segmentation.html","id":"prepare-paths-of-input-images-and-target-segmentation-masks","dir":"Articles > Examples > Vision","previous_headings":"","what":"Prepare paths of input images and target segmentation masks","title":"Image segmentation with a U-Net-like architecture","text":"","code":"library(keras3) input_dir <- \"datasets/images/\" target_dir <- \"datasets/annotations/trimaps/\" img_size <- c(160, 160) num_classes <- 3 batch_size <- 32  input_img_paths <- fs::dir_ls(input_dir, glob = \"*.jpg\") |> sort() target_img_paths <- fs::dir_ls(target_dir, glob = \"*.png\") |> sort()  cat(\"Number of samples:\", length(input_img_paths), \"\\n\") ## Number of samples: 7390 for (i in 1:10) {   cat(input_img_paths[i], \"|\", target_img_paths[i], \"\\n\") } ## datasets/images/Abyssinian_1.jpg | datasets/annotations/trimaps/Abyssinian_1.png ## datasets/images/Abyssinian_10.jpg | datasets/annotations/trimaps/Abyssinian_10.png ## datasets/images/Abyssinian_100.jpg | datasets/annotations/trimaps/Abyssinian_100.png ## datasets/images/Abyssinian_101.jpg | datasets/annotations/trimaps/Abyssinian_101.png ## datasets/images/Abyssinian_102.jpg | datasets/annotations/trimaps/Abyssinian_102.png ## datasets/images/Abyssinian_103.jpg | datasets/annotations/trimaps/Abyssinian_103.png ## datasets/images/Abyssinian_104.jpg | datasets/annotations/trimaps/Abyssinian_104.png ## datasets/images/Abyssinian_105.jpg | datasets/annotations/trimaps/Abyssinian_105.png ## datasets/images/Abyssinian_106.jpg | datasets/annotations/trimaps/Abyssinian_106.png ## datasets/images/Abyssinian_107.jpg | datasets/annotations/trimaps/Abyssinian_107.png"},{"path":"https://keras3.posit.co/dev/articles/examples/vision/oxford_pets_image_segmentation.html","id":"what-does-one-input-image-and-corresponding-segmentation-mask-look-like","dir":"Articles > Examples > Vision","previous_headings":"","what":"What does one input image and corresponding segmentation mask look like?","title":"Image segmentation with a U-Net-like architecture","text":"","code":"# Display input image #10 input_img_paths[10] |>   jpeg::readJPEG() |>   as.raster() |>   plot() target_img_paths[10] |>   png::readPNG() |>   magrittr::multiply_by(255)|>   as.raster(max = 3) |>   plot()"},{"path":"https://keras3.posit.co/dev/articles/examples/vision/oxford_pets_image_segmentation.html","id":"prepare-dataset-to-load-vectorize-batches-of-data","dir":"Articles > Examples > Vision","previous_headings":"","what":"Prepare dataset to load & vectorize batches of data","title":"Image segmentation with a U-Net-like architecture","text":"","code":"library(tensorflow, exclude = c(\"shape\", \"set_random_seed\")) library(tfdatasets, exclude = \"shape\")   # Returns a tf_dataset get_dataset <- function(batch_size, img_size, input_img_paths, target_img_paths,                         max_dataset_len = NULL) {    img_size <- as.integer(img_size)    load_img_masks <- function(input_img_path, target_img_path) {     input_img <- input_img_path |>       tf$io$read_file() |>       tf$io$decode_jpeg(channels = 3) |>       tf$image$resize(img_size) |>       tf$image$convert_image_dtype(\"float32\")      target_img <- target_img_path |>       tf$io$read_file() |>       tf$io$decode_png(channels = 1) |>       tf$image$resize(img_size, method = \"nearest\") |>       tf$image$convert_image_dtype(\"uint8\")      # Ground truth labels are 1, 2, 3. Subtract one to make them 0, 1, 2:     target_img <- target_img - 1L      list(input_img, target_img)   }    if (!is.null(max_dataset_len)) {     input_img_paths <- input_img_paths[1:max_dataset_len]     target_img_paths <- target_img_paths[1:max_dataset_len]   }    list(input_img_paths, target_img_paths) |>     tensor_slices_dataset() |>     dataset_map(load_img_masks, num_parallel_calls = tf$data$AUTOTUNE)|>     dataset_batch(batch_size) }"},{"path":"https://keras3.posit.co/dev/articles/examples/vision/oxford_pets_image_segmentation.html","id":"prepare-u-net-xception-style-model","dir":"Articles > Examples > Vision","previous_headings":"","what":"Prepare U-Net Xception-style model","title":"Image segmentation with a U-Net-like architecture","text":"","code":"get_model <- function(img_size, num_classes) {    inputs <- keras_input(shape = c(img_size, 3))    ### [First half of the network: downsampling inputs] ###    # Entry block   x <- inputs |>     layer_conv_2d(filters = 32, kernel_size = 3, strides = 2, padding = \"same\") |>     layer_batch_normalization() |>     layer_activation(\"relu\")    previous_block_activation <- x  # Set aside residual    for (filters in c(64, 128, 256)) {     x <- x |>       layer_activation(\"relu\") |>       layer_separable_conv_2d(filters = filters, kernel_size = 3, padding = \"same\") |>       layer_batch_normalization() |>        layer_activation(\"relu\") |>       layer_separable_conv_2d(filters = filters, kernel_size = 3, padding = \"same\") |>       layer_batch_normalization() |>        layer_max_pooling_2d(pool_size = 3, strides = 2, padding = \"same\")      residual <- previous_block_activation |>       layer_conv_2d(filters = filters, kernel_size = 1, strides = 2, padding = \"same\")      x <- layer_add(x, residual)  # Add back residual     previous_block_activation <- x  # Set aside next residual   }    ### [Second half of the network: upsampling inputs] ###    for (filters in c(256, 128, 64, 32)) {     x <- x |>       layer_activation(\"relu\") |>       layer_conv_2d_transpose(filters = filters, kernel_size = 3, padding = \"same\") |>       layer_batch_normalization() |>        layer_activation(\"relu\") |>       layer_conv_2d_transpose(filters = filters, kernel_size = 3, padding = \"same\") |>       layer_batch_normalization() |>        layer_upsampling_2d(size = 2)      # Project residual     residual <- previous_block_activation |>       layer_upsampling_2d(size = 2) |>       layer_conv_2d(filters = filters, kernel_size = 1, padding = \"same\")      x <- layer_add(x, residual)     # Add back residual     previous_block_activation <- x  # Set aside next residual   }    # Add a per-pixel classification layer   outputs <- x |>     layer_conv_2d(num_classes, 3, activation = \"softmax\", padding = \"same\")    # Define the model   keras_model(inputs, outputs) }  # Build model model <- get_model(img_size, num_classes) summary(model) ## Model: \"functional\" ## ┏━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━┳━━━━━━━━━━━━━━━━┳━━━━━━━┓ ## ┃ Layer (type)      ┃ Output Shape    ┃   Param # ┃ Connected to   ┃ Trai… ┃ ## ┡━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━╇━━━━━━━━━━━━━━━━╇━━━━━━━┩ ## │ input_layer       │ (None, 160,     │         0 │ -              │   -   │ ## │ (InputLayer)      │ 160, 3)         │           │                │       │ ## ├───────────────────┼─────────────────┼───────────┼────────────────┼───────┤ ## │ conv2d (Conv2D)   │ (None, 80, 80,  │       896 │ input_layer[0… │   Y   │ ## │                   │ 32)             │           │                │       │ ## ├───────────────────┼─────────────────┼───────────┼────────────────┼───────┤ ## │ batch_normalizat… │ (None, 80, 80,  │       128 │ conv2d[0][0]   │   Y   │ ## │ (BatchNormalizat… │ 32)             │           │                │       │ ## ├───────────────────┼─────────────────┼───────────┼────────────────┼───────┤ ## │ activation        │ (None, 80, 80,  │         0 │ batch_normali… │   -   │ ## │ (Activation)      │ 32)             │           │                │       │ ## ├───────────────────┼─────────────────┼───────────┼────────────────┼───────┤ ## │ activation_1      │ (None, 80, 80,  │         0 │ activation[0]… │   -   │ ## │ (Activation)      │ 32)             │           │                │       │ ## ├───────────────────┼─────────────────┼───────────┼────────────────┼───────┤ ## │ separable_conv2d  │ (None, 80, 80,  │     2,400 │ activation_1[… │   Y   │ ## │ (SeparableConv2D) │ 64)             │           │                │       │ ## ├───────────────────┼─────────────────┼───────────┼────────────────┼───────┤ ## │ batch_normalizat… │ (None, 80, 80,  │       256 │ separable_con… │   Y   │ ## │ (BatchNormalizat… │ 64)             │           │                │       │ ## ├───────────────────┼─────────────────┼───────────┼────────────────┼───────┤ ## │ activation_2      │ (None, 80, 80,  │         0 │ batch_normali… │   -   │ ## │ (Activation)      │ 64)             │           │                │       │ ## ├───────────────────┼─────────────────┼───────────┼────────────────┼───────┤ ## │ separable_conv2d… │ (None, 80, 80,  │     4,736 │ activation_2[… │   Y   │ ## │ (SeparableConv2D) │ 64)             │           │                │       │ ## ├───────────────────┼─────────────────┼───────────┼────────────────┼───────┤ ## │ batch_normalizat… │ (None, 80, 80,  │       256 │ separable_con… │   Y   │ ## │ (BatchNormalizat… │ 64)             │           │                │       │ ## ├───────────────────┼─────────────────┼───────────┼────────────────┼───────┤ ## │ max_pooling2d     │ (None, 40, 40,  │         0 │ batch_normali… │   -   │ ## │ (MaxPooling2D)    │ 64)             │           │                │       │ ## ├───────────────────┼─────────────────┼───────────┼────────────────┼───────┤ ## │ conv2d_1 (Conv2D) │ (None, 40, 40,  │     2,112 │ activation[0]… │   Y   │ ## │                   │ 64)             │           │                │       │ ## ├───────────────────┼─────────────────┼───────────┼────────────────┼───────┤ ## │ add (Add)         │ (None, 40, 40,  │         0 │ max_pooling2d… │   -   │ ## │                   │ 64)             │           │ conv2d_1[0][0] │       │ ## ├───────────────────┼─────────────────┼───────────┼────────────────┼───────┤ ## │ activation_3      │ (None, 40, 40,  │         0 │ add[0][0]      │   -   │ ## │ (Activation)      │ 64)             │           │                │       │ ## ├───────────────────┼─────────────────┼───────────┼────────────────┼───────┤ ## │ separable_conv2d… │ (None, 40, 40,  │     8,896 │ activation_3[… │   Y   │ ## │ (SeparableConv2D) │ 128)            │           │                │       │ ## ├───────────────────┼─────────────────┼───────────┼────────────────┼───────┤ ## │ batch_normalizat… │ (None, 40, 40,  │       512 │ separable_con… │   Y   │ ## │ (BatchNormalizat… │ 128)            │           │                │       │ ## ├───────────────────┼─────────────────┼───────────┼────────────────┼───────┤ ## │ activation_4      │ (None, 40, 40,  │         0 │ batch_normali… │   -   │ ## │ (Activation)      │ 128)            │           │                │       │ ## ├───────────────────┼─────────────────┼───────────┼────────────────┼───────┤ ## │ separable_conv2d… │ (None, 40, 40,  │    17,664 │ activation_4[… │   Y   │ ## │ (SeparableConv2D) │ 128)            │           │                │       │ ## ├───────────────────┼─────────────────┼───────────┼────────────────┼───────┤ ## │ batch_normalizat… │ (None, 40, 40,  │       512 │ separable_con… │   Y   │ ## │ (BatchNormalizat… │ 128)            │           │                │       │ ## ├───────────────────┼─────────────────┼───────────┼────────────────┼───────┤ ## │ max_pooling2d_1   │ (None, 20, 20,  │         0 │ batch_normali… │   -   │ ## │ (MaxPooling2D)    │ 128)            │           │                │       │ ## ├───────────────────┼─────────────────┼───────────┼────────────────┼───────┤ ## │ conv2d_2 (Conv2D) │ (None, 20, 20,  │     8,320 │ add[0][0]      │   Y   │ ## │                   │ 128)            │           │                │       │ ## ├───────────────────┼─────────────────┼───────────┼────────────────┼───────┤ ## │ add_1 (Add)       │ (None, 20, 20,  │         0 │ max_pooling2d… │   -   │ ## │                   │ 128)            │           │ conv2d_2[0][0] │       │ ## ├───────────────────┼─────────────────┼───────────┼────────────────┼───────┤ ## │ activation_5      │ (None, 20, 20,  │         0 │ add_1[0][0]    │   -   │ ## │ (Activation)      │ 128)            │           │                │       │ ## ├───────────────────┼─────────────────┼───────────┼────────────────┼───────┤ ## │ separable_conv2d… │ (None, 20, 20,  │    34,176 │ activation_5[… │   Y   │ ## │ (SeparableConv2D) │ 256)            │           │                │       │ ## ├───────────────────┼─────────────────┼───────────┼────────────────┼───────┤ ## │ batch_normalizat… │ (None, 20, 20,  │     1,024 │ separable_con… │   Y   │ ## │ (BatchNormalizat… │ 256)            │           │                │       │ ## ├───────────────────┼─────────────────┼───────────┼────────────────┼───────┤ ## │ activation_6      │ (None, 20, 20,  │         0 │ batch_normali… │   -   │ ## │ (Activation)      │ 256)            │           │                │       │ ## ├───────────────────┼─────────────────┼───────────┼────────────────┼───────┤ ## │ separable_conv2d… │ (None, 20, 20,  │    68,096 │ activation_6[… │   Y   │ ## │ (SeparableConv2D) │ 256)            │           │                │       │ ## ├───────────────────┼─────────────────┼───────────┼────────────────┼───────┤ ## │ batch_normalizat… │ (None, 20, 20,  │     1,024 │ separable_con… │   Y   │ ## │ (BatchNormalizat… │ 256)            │           │                │       │ ## ├───────────────────┼─────────────────┼───────────┼────────────────┼───────┤ ## │ max_pooling2d_2   │ (None, 10, 10,  │         0 │ batch_normali… │   -   │ ## │ (MaxPooling2D)    │ 256)            │           │                │       │ ## ├───────────────────┼─────────────────┼───────────┼────────────────┼───────┤ ## │ conv2d_3 (Conv2D) │ (None, 10, 10,  │    33,024 │ add_1[0][0]    │   Y   │ ## │                   │ 256)            │           │                │       │ ## ├───────────────────┼─────────────────┼───────────┼────────────────┼───────┤ ## │ add_2 (Add)       │ (None, 10, 10,  │         0 │ max_pooling2d… │   -   │ ## │                   │ 256)            │           │ conv2d_3[0][0] │       │ ## ├───────────────────┼─────────────────┼───────────┼────────────────┼───────┤ ## │ activation_7      │ (None, 10, 10,  │         0 │ add_2[0][0]    │   -   │ ## │ (Activation)      │ 256)            │           │                │       │ ## ├───────────────────┼─────────────────┼───────────┼────────────────┼───────┤ ## │ conv2d_transpose  │ (None, 10, 10,  │   590,080 │ activation_7[… │   Y   │ ## │ (Conv2DTranspose) │ 256)            │           │                │       │ ## ├───────────────────┼─────────────────┼───────────┼────────────────┼───────┤ ## │ batch_normalizat… │ (None, 10, 10,  │     1,024 │ conv2d_transp… │   Y   │ ## │ (BatchNormalizat… │ 256)            │           │                │       │ ## ├───────────────────┼─────────────────┼───────────┼────────────────┼───────┤ ## │ activation_8      │ (None, 10, 10,  │         0 │ batch_normali… │   -   │ ## │ (Activation)      │ 256)            │           │                │       │ ## ├───────────────────┼─────────────────┼───────────┼────────────────┼───────┤ ## │ conv2d_transpose… │ (None, 10, 10,  │   590,080 │ activation_8[… │   Y   │ ## │ (Conv2DTranspose) │ 256)            │           │                │       │ ## ├───────────────────┼─────────────────┼───────────┼────────────────┼───────┤ ## │ batch_normalizat… │ (None, 10, 10,  │     1,024 │ conv2d_transp… │   Y   │ ## │ (BatchNormalizat… │ 256)            │           │                │       │ ## ├───────────────────┼─────────────────┼───────────┼────────────────┼───────┤ ## │ up_sampling2d_1   │ (None, 20, 20,  │         0 │ add_2[0][0]    │   -   │ ## │ (UpSampling2D)    │ 256)            │           │                │       │ ## ├───────────────────┼─────────────────┼───────────┼────────────────┼───────┤ ## │ up_sampling2d     │ (None, 20, 20,  │         0 │ batch_normali… │   -   │ ## │ (UpSampling2D)    │ 256)            │           │                │       │ ## ├───────────────────┼─────────────────┼───────────┼────────────────┼───────┤ ## │ conv2d_4 (Conv2D) │ (None, 20, 20,  │    65,792 │ up_sampling2d… │   Y   │ ## │                   │ 256)            │           │                │       │ ## ├───────────────────┼─────────────────┼───────────┼────────────────┼───────┤ ## │ add_3 (Add)       │ (None, 20, 20,  │         0 │ up_sampling2d… │   -   │ ## │                   │ 256)            │           │ conv2d_4[0][0] │       │ ## ├───────────────────┼─────────────────┼───────────┼────────────────┼───────┤ ## │ activation_9      │ (None, 20, 20,  │         0 │ add_3[0][0]    │   -   │ ## │ (Activation)      │ 256)            │           │                │       │ ## ├───────────────────┼─────────────────┼───────────┼────────────────┼───────┤ ## │ conv2d_transpose… │ (None, 20, 20,  │   295,040 │ activation_9[… │   Y   │ ## │ (Conv2DTranspose) │ 128)            │           │                │       │ ## ├───────────────────┼─────────────────┼───────────┼────────────────┼───────┤ ## │ batch_normalizat… │ (None, 20, 20,  │       512 │ conv2d_transp… │   Y   │ ## │ (BatchNormalizat… │ 128)            │           │                │       │ ## ├───────────────────┼─────────────────┼───────────┼────────────────┼───────┤ ## │ activation_10     │ (None, 20, 20,  │         0 │ batch_normali… │   -   │ ## │ (Activation)      │ 128)            │           │                │       │ ## ├───────────────────┼─────────────────┼───────────┼────────────────┼───────┤ ## │ conv2d_transpose… │ (None, 20, 20,  │   147,584 │ activation_10… │   Y   │ ## │ (Conv2DTranspose) │ 128)            │           │                │       │ ## ├───────────────────┼─────────────────┼───────────┼────────────────┼───────┤ ## │ batch_normalizat… │ (None, 20, 20,  │       512 │ conv2d_transp… │   Y   │ ## │ (BatchNormalizat… │ 128)            │           │                │       │ ## ├───────────────────┼─────────────────┼───────────┼────────────────┼───────┤ ## │ up_sampling2d_3   │ (None, 40, 40,  │         0 │ add_3[0][0]    │   -   │ ## │ (UpSampling2D)    │ 256)            │           │                │       │ ## ├───────────────────┼─────────────────┼───────────┼────────────────┼───────┤ ## │ up_sampling2d_2   │ (None, 40, 40,  │         0 │ batch_normali… │   -   │ ## │ (UpSampling2D)    │ 128)            │           │                │       │ ## ├───────────────────┼─────────────────┼───────────┼────────────────┼───────┤ ## │ conv2d_5 (Conv2D) │ (None, 40, 40,  │    32,896 │ up_sampling2d… │   Y   │ ## │                   │ 128)            │           │                │       │ ## ├───────────────────┼─────────────────┼───────────┼────────────────┼───────┤ ## │ add_4 (Add)       │ (None, 40, 40,  │         0 │ up_sampling2d… │   -   │ ## │                   │ 128)            │           │ conv2d_5[0][0] │       │ ## ├───────────────────┼─────────────────┼───────────┼────────────────┼───────┤ ## │ activation_11     │ (None, 40, 40,  │         0 │ add_4[0][0]    │   -   │ ## │ (Activation)      │ 128)            │           │                │       │ ## ├───────────────────┼─────────────────┼───────────┼────────────────┼───────┤ ## │ conv2d_transpose… │ (None, 40, 40,  │    73,792 │ activation_11… │   Y   │ ## │ (Conv2DTranspose) │ 64)             │           │                │       │ ## ├───────────────────┼─────────────────┼───────────┼────────────────┼───────┤ ## │ batch_normalizat… │ (None, 40, 40,  │       256 │ conv2d_transp… │   Y   │ ## │ (BatchNormalizat… │ 64)             │           │                │       │ ## ├───────────────────┼─────────────────┼───────────┼────────────────┼───────┤ ## │ activation_12     │ (None, 40, 40,  │         0 │ batch_normali… │   -   │ ## │ (Activation)      │ 64)             │           │                │       │ ## ├───────────────────┼─────────────────┼───────────┼────────────────┼───────┤ ## │ conv2d_transpose… │ (None, 40, 40,  │    36,928 │ activation_12… │   Y   │ ## │ (Conv2DTranspose) │ 64)             │           │                │       │ ## ├───────────────────┼─────────────────┼───────────┼────────────────┼───────┤ ## │ batch_normalizat… │ (None, 40, 40,  │       256 │ conv2d_transp… │   Y   │ ## │ (BatchNormalizat… │ 64)             │           │                │       │ ## ├───────────────────┼─────────────────┼───────────┼────────────────┼───────┤ ## │ up_sampling2d_5   │ (None, 80, 80,  │         0 │ add_4[0][0]    │   -   │ ## │ (UpSampling2D)    │ 128)            │           │                │       │ ## ├───────────────────┼─────────────────┼───────────┼────────────────┼───────┤ ## │ up_sampling2d_4   │ (None, 80, 80,  │         0 │ batch_normali… │   -   │ ## │ (UpSampling2D)    │ 64)             │           │                │       │ ## ├───────────────────┼─────────────────┼───────────┼────────────────┼───────┤ ## │ conv2d_6 (Conv2D) │ (None, 80, 80,  │     8,256 │ up_sampling2d… │   Y   │ ## │                   │ 64)             │           │                │       │ ## ├───────────────────┼─────────────────┼───────────┼────────────────┼───────┤ ## │ add_5 (Add)       │ (None, 80, 80,  │         0 │ up_sampling2d… │   -   │ ## │                   │ 64)             │           │ conv2d_6[0][0] │       │ ## ├───────────────────┼─────────────────┼───────────┼────────────────┼───────┤ ## │ activation_13     │ (None, 80, 80,  │         0 │ add_5[0][0]    │   -   │ ## │ (Activation)      │ 64)             │           │                │       │ ## ├───────────────────┼─────────────────┼───────────┼────────────────┼───────┤ ## │ conv2d_transpose… │ (None, 80, 80,  │    18,464 │ activation_13… │   Y   │ ## │ (Conv2DTranspose) │ 32)             │           │                │       │ ## ├───────────────────┼─────────────────┼───────────┼────────────────┼───────┤ ## │ batch_normalizat… │ (None, 80, 80,  │       128 │ conv2d_transp… │   Y   │ ## │ (BatchNormalizat… │ 32)             │           │                │       │ ## ├───────────────────┼─────────────────┼───────────┼────────────────┼───────┤ ## │ activation_14     │ (None, 80, 80,  │         0 │ batch_normali… │   -   │ ## │ (Activation)      │ 32)             │           │                │       │ ## ├───────────────────┼─────────────────┼───────────┼────────────────┼───────┤ ## │ conv2d_transpose… │ (None, 80, 80,  │     9,248 │ activation_14… │   Y   │ ## │ (Conv2DTranspose) │ 32)             │           │                │       │ ## ├───────────────────┼─────────────────┼───────────┼────────────────┼───────┤ ## │ batch_normalizat… │ (None, 80, 80,  │       128 │ conv2d_transp… │   Y   │ ## │ (BatchNormalizat… │ 32)             │           │                │       │ ## ├───────────────────┼─────────────────┼───────────┼────────────────┼───────┤ ## │ up_sampling2d_7   │ (None, 160,     │         0 │ add_5[0][0]    │   -   │ ## │ (UpSampling2D)    │ 160, 64)        │           │                │       │ ## ├───────────────────┼─────────────────┼───────────┼────────────────┼───────┤ ## │ up_sampling2d_6   │ (None, 160,     │         0 │ batch_normali… │   -   │ ## │ (UpSampling2D)    │ 160, 32)        │           │                │       │ ## ├───────────────────┼─────────────────┼───────────┼────────────────┼───────┤ ## │ conv2d_7 (Conv2D) │ (None, 160,     │     2,080 │ up_sampling2d… │   Y   │ ## │                   │ 160, 32)        │           │                │       │ ## ├───────────────────┼─────────────────┼───────────┼────────────────┼───────┤ ## │ add_6 (Add)       │ (None, 160,     │         0 │ up_sampling2d… │   -   │ ## │                   │ 160, 32)        │           │ conv2d_7[0][0] │       │ ## ├───────────────────┼─────────────────┼───────────┼────────────────┼───────┤ ## │ conv2d_8 (Conv2D) │ (None, 160,     │       867 │ add_6[0][0]    │   Y   │ ## │                   │ 160, 3)         │           │                │       │ ## └───────────────────┴─────────────────┴───────────┴────────────────┴───────┘ ##  Total params: 2,058,979 (7.85 MB) ##  Trainable params: 2,055,203 (7.84 MB) ##  Non-trainable params: 3,776 (14.75 KB)"},{"path":"https://keras3.posit.co/dev/articles/examples/vision/oxford_pets_image_segmentation.html","id":"set-aside-a-validation-split","dir":"Articles > Examples > Vision","previous_headings":"","what":"Set aside a validation split","title":"Image segmentation with a U-Net-like architecture","text":"","code":"# Split our img paths into a training and a validation set val_samples <- 1000 val_samples <- sample.int(length(input_img_paths), val_samples)  train_input_img_paths <- input_img_paths[-val_samples] train_target_img_paths <- target_img_paths[-val_samples]  val_input_img_paths <- input_img_paths[val_samples] val_target_img_paths <- target_img_paths[val_samples]  # Instantiate dataset for each split # Limit input files in `max_dataset_len` for faster epoch training time. # Remove the `max_dataset_len` arg when running with full dataset. train_dataset <- get_dataset(   batch_size,   img_size,   train_input_img_paths,   train_target_img_paths,   max_dataset_len = 1000 ) valid_dataset <- get_dataset(   batch_size, img_size, val_input_img_paths, val_target_img_paths )"},{"path":"https://keras3.posit.co/dev/articles/examples/vision/oxford_pets_image_segmentation.html","id":"train-the-model","dir":"Articles > Examples > Vision","previous_headings":"","what":"Train the model","title":"Image segmentation with a U-Net-like architecture","text":"","code":"# Configure the model for training. # We use the \"sparse\" version of categorical_crossentropy # because our target data is integers. model |> compile(   optimizer = optimizer_adam(1e-4),   loss = \"sparse_categorical_crossentropy\" )  callbacks <- list(   callback_model_checkpoint(     \"models/oxford_segmentation.keras\", save_best_only = TRUE   ) )  # Train the model, doing validation at the end of each epoch. epochs <- 50 model |> fit(     train_dataset,     epochs=epochs,     validation_data=valid_dataset,     callbacks=callbacks,     verbose=2 ) ## Epoch 1/50 ## 32/32 - 49s - 2s/step - loss: 1.4283 - val_loss: 1.5505 ## Epoch 2/50 ## 32/32 - 2s - 73ms/step - loss: 0.9223 - val_loss: 1.9896 ## Epoch 3/50 ## 32/32 - 2s - 72ms/step - loss: 0.7764 - val_loss: 2.5134 ## Epoch 4/50 ## 32/32 - 2s - 73ms/step - loss: 0.7200 - val_loss: 3.0166 ## Epoch 5/50 ## 32/32 - 2s - 74ms/step - loss: 0.6848 - val_loss: 3.2910 ## Epoch 6/50 ## 32/32 - 2s - 71ms/step - loss: 0.6557 - val_loss: 3.4529 ## Epoch 7/50 ## 32/32 - 2s - 74ms/step - loss: 0.6303 - val_loss: 3.5613 ## Epoch 8/50 ## 32/32 - 2s - 71ms/step - loss: 0.6083 - val_loss: 3.6520 ## Epoch 9/50 ## 32/32 - 2s - 68ms/step - loss: 0.5894 - val_loss: 3.7289 ## Epoch 10/50 ## 32/32 - 2s - 76ms/step - loss: 0.5726 - val_loss: 3.7945 ## Epoch 11/50 ## 32/32 - 2s - 75ms/step - loss: 0.5566 - val_loss: 3.8264 ## Epoch 12/50 ## 32/32 - 2s - 73ms/step - loss: 0.5406 - val_loss: 3.8009 ## Epoch 13/50 ## 32/32 - 2s - 73ms/step - loss: 0.5240 - val_loss: 3.7206 ## Epoch 14/50 ## 32/32 - 2s - 76ms/step - loss: 0.5060 - val_loss: 3.5849 ## Epoch 15/50 ## 32/32 - 2s - 76ms/step - loss: 0.4858 - val_loss: 3.4372 ## Epoch 16/50 ## 32/32 - 2s - 73ms/step - loss: 0.4636 - val_loss: 3.2371 ## Epoch 17/50 ## 32/32 - 2s - 78ms/step - loss: 0.4395 - val_loss: 3.0008 ## Epoch 18/50 ## 32/32 - 2s - 75ms/step - loss: 0.4140 - val_loss: 2.6982 ## Epoch 19/50 ## 32/32 - 2s - 72ms/step - loss: 0.3878 - val_loss: 2.3533 ## Epoch 20/50 ## 32/32 - 2s - 74ms/step - loss: 0.3623 - val_loss: 1.9573 ## Epoch 21/50 ## 32/32 - 2s - 73ms/step - loss: 0.3391 - val_loss: 1.6047 ## Epoch 22/50 ## 32/32 - 3s - 83ms/step - loss: 0.3202 - val_loss: 1.3140 ## Epoch 23/50 ## 32/32 - 2s - 77ms/step - loss: 0.3083 - val_loss: 1.0783 ## Epoch 24/50 ## 32/32 - 3s - 80ms/step - loss: 0.3097 - val_loss: 1.0180 ## Epoch 25/50 ## 32/32 - 2s - 77ms/step - loss: 0.3446 - val_loss: 0.9350 ## Epoch 26/50 ## 32/32 - 2s - 70ms/step - loss: 0.3603 - val_loss: 0.9866 ## Epoch 27/50 ## 32/32 - 3s - 82ms/step - loss: 0.3268 - val_loss: 0.8704 ## Epoch 28/50 ## 32/32 - 2s - 72ms/step - loss: 0.2880 - val_loss: 1.0167 ## Epoch 29/50 ## 32/32 - 2s - 75ms/step - loss: 0.2735 - val_loss: 1.1785 ## Epoch 30/50 ## 32/32 - 2s - 75ms/step - loss: 0.2686 - val_loss: 1.1817 ## Epoch 31/50 ## 32/32 - 2s - 74ms/step - loss: 0.2718 - val_loss: 1.1267 ## Epoch 32/50 ## 32/32 - 2s - 72ms/step - loss: 0.2867 - val_loss: 1.1271 ## Epoch 33/50 ## 32/32 - 2s - 75ms/step - loss: 0.3123 - val_loss: 1.2541 ## Epoch 34/50 ## 32/32 - 2s - 75ms/step - loss: 0.2997 - val_loss: 1.2340 ## Epoch 35/50 ## 32/32 - 2s - 74ms/step - loss: 0.2851 - val_loss: 1.0755 ## Epoch 36/50 ## 32/32 - 2s - 73ms/step - loss: 0.2781 - val_loss: 1.0022 ## Epoch 37/50 ## 32/32 - 2s - 75ms/step - loss: 0.2755 - val_loss: 1.5445 ## Epoch 38/50 ## 32/32 - 2s - 76ms/step - loss: 0.2629 - val_loss: 1.1026 ## Epoch 39/50 ## 32/32 - 2s - 75ms/step - loss: 0.2428 - val_loss: 1.2205 ## Epoch 40/50 ## 32/32 - 2s - 74ms/step - loss: 0.2409 - val_loss: 1.2005 ## Epoch 41/50 ## 32/32 - 2s - 74ms/step - loss: 0.2453 - val_loss: 1.2381 ## Epoch 42/50 ## 32/32 - 2s - 73ms/step - loss: 0.2363 - val_loss: 1.5562 ## Epoch 43/50 ## 32/32 - 2s - 75ms/step - loss: 0.2337 - val_loss: 1.3185 ## Epoch 44/50 ## 32/32 - 2s - 77ms/step - loss: 0.2222 - val_loss: 1.1867 ## Epoch 45/50 ## 32/32 - 2s - 77ms/step - loss: 0.2151 - val_loss: 1.0701 ## Epoch 46/50 ## 32/32 - 2s - 75ms/step - loss: 0.2099 - val_loss: 1.1109 ## Epoch 47/50 ## 32/32 - 2s - 74ms/step - loss: 0.2062 - val_loss: 1.0900 ## Epoch 48/50 ## 32/32 - 2s - 74ms/step - loss: 0.2038 - val_loss: 1.1194 ## Epoch 49/50 ## 32/32 - 2s - 71ms/step - loss: 0.1964 - val_loss: 1.0576 ## Epoch 50/50 ## 32/32 - 2s - 73ms/step - loss: 0.1871 - val_loss: 1.0851"},{"path":"https://keras3.posit.co/dev/articles/examples/vision/oxford_pets_image_segmentation.html","id":"visualize-predictions","dir":"Articles > Examples > Vision","previous_headings":"","what":"Visualize predictions","title":"Image segmentation with a U-Net-like architecture","text":"","code":"model <- load_model(\"models/oxford_segmentation.keras\") # Generate predictions for all images in the validation set val_dataset <- get_dataset(   batch_size, img_size, val_input_img_paths, val_target_img_paths ) val_preds <- predict(model, val_dataset) ## 32/32 - 4s - 117ms/step display_mask <- function(i) {   # Quick utility to display a model's prediction.   mask <- val_preds[i,,,] %>%     apply(c(1,2), which.max) %>%     array_reshape(dim = c(img_size, 1))   mask <- abind::abind(mask, mask, mask, along = 3)   plot(as.raster(mask, max = 3)) }  # Display results for validation image #10 i <- 10  par(mfrow = c(1, 3)) # Display input image input_img_paths[i] |>   jpeg::readJPEG() |>   as.raster() |>   plot()  # Display ground-truth target mask target_img_paths[i] |>   png::readPNG() |>   magrittr::multiply_by(255)|>   as.raster(max = 3) |>   plot()  # Display mask predicted by our model display_mask(i)  # Note that the model only sees inputs at 150x150."},{"path":"https://keras3.posit.co/dev/articles/functional_api.html","id":"setup","dir":"Articles","previous_headings":"","what":"Setup","title":"The Functional API","text":"","code":"library(keras3)"},{"path":"https://keras3.posit.co/dev/articles/functional_api.html","id":"introduction","dir":"Articles","previous_headings":"","what":"Introduction","title":"The Functional API","text":"Keras functional API way create models flexible sequential API. functional API can handle models non-linear topology, shared layers, even multiple inputs outputs. main idea deep learning model usually directed acyclic graph (DAG) layers. functional API way build graphs layers. Consider following model: basic graph three layers. build model using functional API, start creating input node: shape data set 784-dimensional vector. batch size always omitted since shape sample specified. , example, image input shape (32, 32, 3), use: inputs returned contains information shape dtype input data feed model. ’s shape: ’s dtype: create new node graph layers calling layer inputs object: “layer call” action like drawing arrow “inputs” layer created. ’re “passing” inputs dense layer, get x output. Let’s add layers graph layers: point, can create Model specifying inputs outputs graph layers: Let’s check model summary looks like: can also plot model graph:  , optionally, display input output shapes layer plotted graph:  figure code almost identical. code version, connection arrows replaced call operation. “graph layers” intuitive mental image deep learning model, functional API way create models closely mirrors .","code":"(input: 784-dimensional vectors)        ↧ [Dense (64 units, relu activation)]        ↧ [Dense (64 units, relu activation)]        ↧ [Dense (10 units, softmax activation)]        ↧ (output: logits of a probability distribution over 10 classes) inputs <- keras_input(shape = c(784)) # Just for demonstration purposes. img_inputs <- keras_input(shape = c(32, 32, 3)) shape(inputs) ## shape(NA, 784) inputs$dtype ## [1] \"float32\" dense <- layer_dense(units = 64, activation=\"relu\") x <- dense(inputs) outputs <- x |>   layer_dense(units = 64, activation = \"relu\") |>   layer_dense(units = 10) model <- keras_model(inputs = inputs, outputs = outputs, name = \"mnist_model\") summary(model) ## Model: \"mnist_model\" ## ┏━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━┓ ## ┃ Layer (type)                    ┃ Output Shape           ┃       Param # ┃ ## ┡━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━┩ ## │ input_layer (InputLayer)        │ (None, 784)            │             0 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ dense (Dense)                   │ (None, 64)             │        50,240 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ dense_1 (Dense)                 │ (None, 64)             │         4,160 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ dense_2 (Dense)                 │ (None, 10)             │           650 │ ## └─────────────────────────────────┴────────────────────────┴───────────────┘ ##  Total params: 55,050 (215.04 KB) ##  Trainable params: 55,050 (215.04 KB) ##  Non-trainable params: 0 (0.00 B) plot(model) plot(model, show_shapes = TRUE)"},{"path":"https://keras3.posit.co/dev/articles/functional_api.html","id":"training-evaluation-and-inference","dir":"Articles","previous_headings":"","what":"Training, evaluation, and inference","title":"The Functional API","text":"Training, evaluation, inference work exactly way models built using functional API Sequential models. Model class offers built-training loop (fit() method) built-evaluation loop (evaluate() method). Note can easily customize loops implement training routines beyond supervised learning (e.g. GANs). , load MNIST image data, reshape vectors, fit model data (monitoring performance validation split), evaluate model test data: reading, see training evaluation guide.","code":"c(c(x_train, y_train), c(x_test, y_test)) %<-% dataset_mnist()  x_train <- array_reshape(x_train, c(60000, 784)) / 255 x_test <- array_reshape(x_test, c(10000, 784)) / 255  model |> compile(   loss = loss_sparse_categorical_crossentropy(from_logits = TRUE),   optimizer = optimizer_rmsprop(),   metrics = \"accuracy\" )  history <- model |> fit(     x_train, y_train, batch_size = 64, epochs = 2, validation_split = 0.2 ) ## Epoch 1/2 ## 750/750 - 2s - 3ms/step - accuracy: 0.8979 - loss: 0.3540 - val_accuracy: 0.9448 - val_loss: 0.1903 ## Epoch 2/2 ## 750/750 - 1s - 1ms/step - accuracy: 0.9511 - loss: 0.1634 - val_accuracy: 0.9605 - val_loss: 0.1386 test_scores <- model |> evaluate(x_test, y_test, verbose=2) ## 313/313 - 1s - 2ms/step - accuracy: 0.9593 - loss: 0.1323 cat(\"Test loss:\", test_scores$loss, \"\\n\") cat(\"Test accuracy:\", test_scores$accuracy, \"\\n\") ## Test loss: 0.1323339 ## Test accuracy: 0.9593"},{"path":"https://keras3.posit.co/dev/articles/functional_api.html","id":"save-and-serialize","dir":"Articles","previous_headings":"","what":"Save and serialize","title":"The Functional API","text":"Saving model serialization work way models built using functional API Sequential models. standard way save functional model call model.save() save entire model single file. can later recreate model file, even code built model longer available. saved file includes : - model architecture - model weight values (learned training) - model training config, (passed compile()) - optimizer state, (restart training left ) details, read model serialization & saving guide.","code":"model |> save_model(\"my_model.keras\") rm(model) # Recreate the exact same model purely from the file: model <- load_model(\"my_model.keras\")"},{"path":"https://keras3.posit.co/dev/articles/functional_api.html","id":"use-the-same-graph-of-layers-to-define-multiple-models","dir":"Articles","previous_headings":"","what":"Use the same graph of layers to define multiple models","title":"The Functional API","text":"functional API, models created specifying inputs outputs graph layers. means single graph layers can used generate multiple models. example , use stack layers instantiate two models: encoder model turns image inputs 16-dimensional vectors, end--end autoencoder model training. , decoding architecture strictly symmetrical encoding architecture, output shape input shape (28, 28, 1). reverse conv_2d layer conv_2d_transpose layer, reverse max_pooling_2d layer upsampling_2d layer.","code":"encoder_input <- keras_input(shape = c(28, 28, 1), name=\"img\") encoder_output <- encoder_input |>   layer_conv_2d(16, 3, activation = \"relu\") |>   layer_conv_2d(32, 3, activation = \"relu\") |>   layer_max_pooling_2d(3) |>   layer_conv_2d(32, 3, activation = \"relu\") |>   layer_conv_2d(16, 3, activation = \"relu\") |>   layer_global_max_pooling_2d()  encoder <- keras_model(encoder_input, encoder_output, name=\"encoder\") summary(encoder) ## Model: \"encoder\" ## ┏━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━┓ ## ┃ Layer (type)                    ┃ Output Shape           ┃       Param # ┃ ## ┡━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━┩ ## │ img (InputLayer)                │ (None, 28, 28, 1)      │             0 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ conv2d (Conv2D)                 │ (None, 26, 26, 16)     │           160 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ conv2d_1 (Conv2D)               │ (None, 24, 24, 32)     │         4,640 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ max_pooling2d (MaxPooling2D)    │ (None, 8, 8, 32)       │             0 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ conv2d_2 (Conv2D)               │ (None, 6, 6, 32)       │         9,248 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ conv2d_3 (Conv2D)               │ (None, 4, 4, 16)       │         4,624 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ global_max_pooling2d            │ (None, 16)             │             0 │ ## │ (GlobalMaxPooling2D)            │                        │               │ ## └─────────────────────────────────┴────────────────────────┴───────────────┘ ##  Total params: 18,672 (72.94 KB) ##  Trainable params: 18,672 (72.94 KB) ##  Non-trainable params: 0 (0.00 B) decoder_output <- encoder_output |>   layer_reshape(c(4, 4, 1)) |>   layer_conv_2d_transpose(16, 3, activation = \"relu\") |>   layer_conv_2d_transpose(32, 3, activation = \"relu\") |>   layer_upsampling_2d(3) |>   layer_conv_2d_transpose(16, 3, activation = \"relu\") |>   layer_conv_2d_transpose(1, 3, activation = \"relu\")  autoencoder <- keras_model(encoder_input, decoder_output, name=\"autoencoder\") summary(autoencoder) ## Model: \"autoencoder\" ## ┏━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━┓ ## ┃ Layer (type)                    ┃ Output Shape           ┃       Param # ┃ ## ┡━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━┩ ## │ img (InputLayer)                │ (None, 28, 28, 1)      │             0 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ conv2d (Conv2D)                 │ (None, 26, 26, 16)     │           160 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ conv2d_1 (Conv2D)               │ (None, 24, 24, 32)     │         4,640 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ max_pooling2d (MaxPooling2D)    │ (None, 8, 8, 32)       │             0 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ conv2d_2 (Conv2D)               │ (None, 6, 6, 32)       │         9,248 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ conv2d_3 (Conv2D)               │ (None, 4, 4, 16)       │         4,624 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ global_max_pooling2d            │ (None, 16)             │             0 │ ## │ (GlobalMaxPooling2D)            │                        │               │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ reshape (Reshape)               │ (None, 4, 4, 1)        │             0 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ conv2d_transpose                │ (None, 6, 6, 16)       │           160 │ ## │ (Conv2DTranspose)               │                        │               │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ conv2d_transpose_1              │ (None, 8, 8, 32)       │         4,640 │ ## │ (Conv2DTranspose)               │                        │               │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ up_sampling2d (UpSampling2D)    │ (None, 24, 24, 32)     │             0 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ conv2d_transpose_2              │ (None, 26, 26, 16)     │         4,624 │ ## │ (Conv2DTranspose)               │                        │               │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ conv2d_transpose_3              │ (None, 28, 28, 1)      │           145 │ ## │ (Conv2DTranspose)               │                        │               │ ## └─────────────────────────────────┴────────────────────────┴───────────────┘ ##  Total params: 28,241 (110.32 KB) ##  Trainable params: 28,241 (110.32 KB) ##  Non-trainable params: 0 (0.00 B)"},{"path":"https://keras3.posit.co/dev/articles/functional_api.html","id":"all-models-are-callable-just-like-layers","dir":"Articles","previous_headings":"","what":"All models are callable, just like layers","title":"The Functional API","text":"can treat model layer invoking Input output another layer. calling model aren’t just reusing architecture model, ’re also reusing weights. see action, ’s different take autoencoder example creates encoder model, decoder model, chains two calls obtain autoencoder model: can see, model can nested: model can contain sub-models (since model just like layer). common use case model nesting ensembling. example, ’s ensemble set models single model averages predictions:","code":"encoder_input <- keras_input(shape = c(28, 28, 1), name=\"img\") encoder_output <- encoder_input |>   layer_conv_2d(16, 3, activation = \"relu\") |>   layer_conv_2d(32, 3, activation = \"relu\") |>   layer_max_pooling_2d(3) |>   layer_conv_2d(32, 3, activation = \"relu\") |>   layer_conv_2d(16, 3, activation = \"relu\") |>   layer_global_max_pooling_2d()  encoder <- keras_model(encoder_input, encoder_output, name=\"encoder\") summary(encoder) ## Model: \"encoder\" ## ┏━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━┓ ## ┃ Layer (type)                    ┃ Output Shape           ┃       Param # ┃ ## ┡━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━┩ ## │ img (InputLayer)                │ (None, 28, 28, 1)      │             0 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ conv2d_4 (Conv2D)               │ (None, 26, 26, 16)     │           160 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ conv2d_5 (Conv2D)               │ (None, 24, 24, 32)     │         4,640 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ max_pooling2d_1 (MaxPooling2D)  │ (None, 8, 8, 32)       │             0 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ conv2d_6 (Conv2D)               │ (None, 6, 6, 32)       │         9,248 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ conv2d_7 (Conv2D)               │ (None, 4, 4, 16)       │         4,624 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ global_max_pooling2d_1          │ (None, 16)             │             0 │ ## │ (GlobalMaxPooling2D)            │                        │               │ ## └─────────────────────────────────┴────────────────────────┴───────────────┘ ##  Total params: 18,672 (72.94 KB) ##  Trainable params: 18,672 (72.94 KB) ##  Non-trainable params: 0 (0.00 B) decoder_input <- keras_input(shape = c(16), name = \"encoded_img\") decoder_output <- decoder_input |>   layer_reshape(c(4, 4, 1)) |>   layer_conv_2d_transpose(16, 3, activation = \"relu\") |>   layer_conv_2d_transpose(32, 3, activation = \"relu\") |>   layer_upsampling_2d(3) |>   layer_conv_2d_transpose(16, 3, activation = \"relu\") |>   layer_conv_2d_transpose(1, 3, activation = \"relu\")  decoder <- keras_model(decoder_input, decoder_output, name = \"decoder\") summary(decoder) ## Model: \"decoder\" ## ┏━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━┓ ## ┃ Layer (type)                    ┃ Output Shape           ┃       Param # ┃ ## ┡━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━┩ ## │ encoded_img (InputLayer)        │ (None, 16)             │             0 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ reshape_1 (Reshape)             │ (None, 4, 4, 1)        │             0 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ conv2d_transpose_4              │ (None, 6, 6, 16)       │           160 │ ## │ (Conv2DTranspose)               │                        │               │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ conv2d_transpose_5              │ (None, 8, 8, 32)       │         4,640 │ ## │ (Conv2DTranspose)               │                        │               │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ up_sampling2d_1 (UpSampling2D)  │ (None, 24, 24, 32)     │             0 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ conv2d_transpose_6              │ (None, 26, 26, 16)     │         4,624 │ ## │ (Conv2DTranspose)               │                        │               │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ conv2d_transpose_7              │ (None, 28, 28, 1)      │           145 │ ## │ (Conv2DTranspose)               │                        │               │ ## └─────────────────────────────────┴────────────────────────┴───────────────┘ ##  Total params: 9,569 (37.38 KB) ##  Trainable params: 9,569 (37.38 KB) ##  Non-trainable params: 0 (0.00 B) autoencoder_input <- keras_input(shape = c(28, 28, 1), name = \"img\") encoded_img <- encoder(autoencoder_input) decoded_img <- decoder(encoded_img) autoencoder <- keras_model(autoencoder_input, decoded_img,                            name = \"autoencoder\") summary(autoencoder) ## Model: \"autoencoder\" ## ┏━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━┓ ## ┃ Layer (type)                    ┃ Output Shape           ┃       Param # ┃ ## ┡━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━┩ ## │ img (InputLayer)                │ (None, 28, 28, 1)      │             0 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ encoder (Functional)            │ (None, 16)             │        18,672 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ decoder (Functional)            │ (None, 28, 28, 1)      │         9,569 │ ## └─────────────────────────────────┴────────────────────────┴───────────────┘ ##  Total params: 28,241 (110.32 KB) ##  Trainable params: 28,241 (110.32 KB) ##  Non-trainable params: 0 (0.00 B) get_model <- function() {   inputs <- keras_input(shape = 128)   outputs <- inputs |> layer_dense(1)   keras_model(inputs, outputs) }  model1 <- get_model() model2 <- get_model() model3 <- get_model()  inputs <- keras_input(shape = 128) y1 <- model1(inputs) y2 <- model2(inputs) y3 <- model3(inputs) outputs <- layer_average(list(y1, y2, y3)) ensemble_model <- keras_model(inputs = inputs, outputs = outputs)"},{"path":[]},{"path":"https://keras3.posit.co/dev/articles/functional_api.html","id":"models-with-multiple-inputs-and-outputs","dir":"Articles","previous_headings":"Manipulate complex graph topologies","what":"Models with multiple inputs and outputs","title":"The Functional API","text":"functional API makes easy manipulate multiple inputs outputs. handled Sequential API. example, ’re building system ranking customer issue tickets priority routing correct department, model three inputs: title ticket (text input), text body ticket (text input), tags added user (categorical input) model two outputs: priority score 0 1 (scalar sigmoid output), department handle ticket (softmax output set departments). can build model lines functional API: Now plot model:  compiling model, can assign different losses output. can even assign different weights loss – modulate contribution total training loss. Since output layers different names, also specify losses loss weights corresponding layer names: Train model passing lists NumPy arrays inputs targets: calling fit Dataset object, yield either list lists like list(list(title_data, body_data, tags_data), list(priority_targets, dept_targets)) list named lists like list(list(title = title_data, body = body_data, tags = tags_data), list(priority = priority_targets, department = dept_targets)). detailed explanation, refer training evaluation guide.","code":"num_tags <- 12  # Number of unique issue tags num_words <- 10000  # Size of vocabulary obtained when preprocessing text data num_departments <- 4  # Number of departments for predictions  title_input <- # Variable-length sequence of ints   keras_input(shape(NA), name = \"title\") body_input <-  # Variable-length sequence of ints   keras_input(shape(NA), name = \"body\") tags_input <-  # Binary vectors of size `num_tags`   keras_input(shape = num_tags, name = \"tags\")  # Embed each word in the title into a 64-dimensional vector title_features <- layer_embedding(title_input, num_words, 64) # Embed each word in the text into a 64-dimensional vector body_features <- layer_embedding(body_input, num_words, 64)  # Reduce sequence of embedded words in the title # into a single 128-dimensional vector title_features <- layer_lstm(title_features, 128) # Reduce sequence of embedded words in the body # into a single 32-dimensional vector body_features <- layer_lstm(body_features, 32)  # Merge all available features into a single large vector via concatenation x <- layer_concatenate(title_features, body_features, tags_input)  # Stick a logistic regression for priority prediction on top of the features priority_pred <- layer_dense(x, 1, name = \"priority\")  # Stick a department classifier on top of the features department_pred <- layer_dense(x, num_departments, name = \"department\")  # Instantiate an end-to-end model predicting both priority and department model <- keras_model(   inputs = list(title_input, body_input, tags_input),   outputs = list(priority = priority_pred, department = department_pred) ) plot(model, show_shapes = TRUE) model |> compile(   optimizer = optimizer_rmsprop(1e-3),   loss = list(     loss_binary_crossentropy(from_logits = TRUE),     loss_categorical_crossentropy(from_logits = TRUE)   ),   loss_weights = c(1.0, 0.2) ) model |> compile(   optimizer = optimizer_rmsprop(1e-3),   loss = list(     priority = loss_binary_crossentropy(from_logits = TRUE),     department = loss_categorical_crossentropy(from_logits = TRUE)   ),   loss_weights = list(priority = 1.0, department = 0.2) ) # Dummy input data title_data <- random_integer(c(1280, 12), 0, num_words) body_data <- random_integer(c(1280, 100), 0, num_words) tags_data <- random_integer(c(1280, num_tags), 0, 2)  # Dummy target data priority_targets <- random_normal(c(1280, 1)) dept_targets <- random_integer(c(1280, num_departments), 0, 2)  model |> fit(   list(title = title_data, body = body_data, tags = tags_data),   list(priority = priority_targets, department = dept_targets),   epochs = 2,   batch_size = 32 ) ## Epoch 1/2 ## 40/40 - 4s - 99ms/step - department_loss: 2.8465 - loss: 0.7669 - priority_loss: 0.1976 ## Epoch 2/2 ## 40/40 - 0s - 6ms/step - department_loss: 2.8554 - loss: 0.7538 - priority_loss: 0.1828"},{"path":"https://keras3.posit.co/dev/articles/functional_api.html","id":"a-toy-resnet-model","dir":"Articles","previous_headings":"Manipulate complex graph topologies","what":"A toy ResNet model","title":"The Functional API","text":"addition models multiple inputs outputs, functional API makes easy manipulate non-linear connectivity topologies – models layers connected sequentially, Sequential API handle. common use case residual connections. Let’s build toy ResNet model CIFAR10 demonstrate : Plot model:  Now train model:","code":"inputs <- keras_input(shape = c(32, 32, 3), name = \"img\") block_1_output <- inputs |>   layer_conv_2d(32, kernel_size = 3, activation = \"relu\") |>   layer_conv_2d(64, kernel_size = 3, activation = \"relu\") |>   layer_max_pooling_2d(pool_size = 3)  block_2_output <- block_1_output |>   layer_conv_2d(32, kernel_size = 3, activation = \"relu\", padding = \"same\") |>   layer_conv_2d(64, kernel_size = 3, activation = \"relu\", padding = \"same\") |>   layer_add(block_1_output)  block_3_output <- block_2_output |>   layer_conv_2d(64, kernel_size = 3, activation = \"relu\", padding = \"same\") |>   layer_conv_2d(64, kernel_size = 3, activation = \"relu\", padding = \"same\") |>   layer_add(block_2_output)  outputs <- block_3_output |>   layer_conv_2d(64, 3, activation = \"relu\") |>   layer_global_average_pooling_2d() |>   layer_dense(256, activation = \"relu\") |>   layer_dropout(0.5) |>   layer_dense(10)  model <- keras_model(inputs, outputs, name = \"toy_resnet\") summary(model) ## Model: \"toy_resnet\" ## ┏━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━┓ ## ┃ Layer (type)        ┃ Output Shape      ┃    Param # ┃ Connected to      ┃ ## ┡━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━┩ ## │ img (InputLayer)    │ (None, 32, 32, 3) │          0 │ -                 │ ## ├─────────────────────┼───────────────────┼────────────┼───────────────────┤ ## │ conv2d_8 (Conv2D)   │ (None, 30, 30,    │        896 │ img[0][0]         │ ## │                     │ 32)               │            │                   │ ## ├─────────────────────┼───────────────────┼────────────┼───────────────────┤ ## │ conv2d_9 (Conv2D)   │ (None, 28, 28,    │     18,496 │ conv2d_8[0][0]    │ ## │                     │ 64)               │            │                   │ ## ├─────────────────────┼───────────────────┼────────────┼───────────────────┤ ## │ max_pooling2d_2     │ (None, 9, 9, 64)  │          0 │ conv2d_9[0][0]    │ ## │ (MaxPooling2D)      │                   │            │                   │ ## ├─────────────────────┼───────────────────┼────────────┼───────────────────┤ ## │ conv2d_10 (Conv2D)  │ (None, 9, 9, 32)  │     18,464 │ max_pooling2d_2[… │ ## ├─────────────────────┼───────────────────┼────────────┼───────────────────┤ ## │ conv2d_11 (Conv2D)  │ (None, 9, 9, 64)  │     18,496 │ conv2d_10[0][0]   │ ## ├─────────────────────┼───────────────────┼────────────┼───────────────────┤ ## │ add (Add)           │ (None, 9, 9, 64)  │          0 │ conv2d_11[0][0],  │ ## │                     │                   │            │ max_pooling2d_2[… │ ## ├─────────────────────┼───────────────────┼────────────┼───────────────────┤ ## │ conv2d_12 (Conv2D)  │ (None, 9, 9, 64)  │     36,928 │ add[0][0]         │ ## ├─────────────────────┼───────────────────┼────────────┼───────────────────┤ ## │ conv2d_13 (Conv2D)  │ (None, 9, 9, 64)  │     36,928 │ conv2d_12[0][0]   │ ## ├─────────────────────┼───────────────────┼────────────┼───────────────────┤ ## │ add_1 (Add)         │ (None, 9, 9, 64)  │          0 │ conv2d_13[0][0],  │ ## │                     │                   │            │ add[0][0]         │ ## ├─────────────────────┼───────────────────┼────────────┼───────────────────┤ ## │ conv2d_14 (Conv2D)  │ (None, 7, 7, 64)  │     36,928 │ add_1[0][0]       │ ## ├─────────────────────┼───────────────────┼────────────┼───────────────────┤ ## │ global_average_poo… │ (None, 64)        │          0 │ conv2d_14[0][0]   │ ## │ (GlobalAveragePool… │                   │            │                   │ ## ├─────────────────────┼───────────────────┼────────────┼───────────────────┤ ## │ dense_6 (Dense)     │ (None, 256)       │     16,640 │ global_average_p… │ ## ├─────────────────────┼───────────────────┼────────────┼───────────────────┤ ## │ dropout (Dropout)   │ (None, 256)       │          0 │ dense_6[0][0]     │ ## ├─────────────────────┼───────────────────┼────────────┼───────────────────┤ ## │ dense_7 (Dense)     │ (None, 10)        │      2,570 │ dropout[0][0]     │ ## └─────────────────────┴───────────────────┴────────────┴───────────────────┘ ##  Total params: 186,346 (727.91 KB) ##  Trainable params: 186,346 (727.91 KB) ##  Non-trainable params: 0 (0.00 B) plot(model, show_shapes = TRUE) c(c(x_train, y_train), c(x_test, y_test)) %<-% dataset_cifar10()  x_train <- x_train / 255.0 x_test <- x_test / 255.0  model |> compile(   optimizer = optimizer_rmsprop(1e-3),   loss = loss_sparse_categorical_crossentropy(from_logits = TRUE),   metrics = \"acc\" ) # We restrict the data to the first 1000 samples so as to limit the # guide render time. # Try to train on the entire dataset until convergence! model |> fit(   x_train[1:1000, , , ],   y_train[1:1000, ],   batch_size = 64,   epochs = 1,   validation_split = 0.2 ) ## 13/13 - 5s - 418ms/step - acc: 0.1225 - loss: 2.3000 - val_acc: 0.1150 - val_loss: 2.2942"},{"path":"https://keras3.posit.co/dev/articles/functional_api.html","id":"shared-layers","dir":"Articles","previous_headings":"","what":"Shared layers","title":"The Functional API","text":"Another good use functional API models use shared layers. Shared layers layer instances reused multiple times model – learn features correspond multiple paths graph--layers. Shared layers often used encode inputs similar spaces (say, two different pieces text feature similar vocabulary). enable sharing information across different inputs, make possible train model less data. given word seen one inputs, benefit processing inputs pass shared layer. share layer functional API, call layer instance multiple times. instance, ’s Embedding layer shared across two different text inputs:","code":"# Embedding for 1000 unique words mapped to 128-dimensional vectors shared_embedding <- layer_embedding(input_dim = 1000, output_dim = 128)  # Variable-length sequence of integers text_input_a <- keras_input(shape = shape(NA), dtype=\"int32\")  # Variable-length sequence of integers text_input_b <- keras_input(shape = shape(NA), dtype=\"int32\")  # Reuse the same layer to encode both inputs encoded_input_a <- shared_embedding(text_input_a) encoded_input_b <- shared_embedding(text_input_b)"},{"path":"https://keras3.posit.co/dev/articles/functional_api.html","id":"extract-and-reuse-nodes-in-the-graph-of-layers","dir":"Articles","previous_headings":"","what":"Extract and reuse nodes in the graph of layers","title":"The Functional API","text":"graph layers manipulating static data structure, can accessed inspected. able plot functional models images. also means can access activations intermediate layers (“nodes” graph) reuse elsewhere – useful something like feature extraction. Let’s look example. VGG19 model weights pretrained ImageNet: intermediate activations model, obtained querying graph data structure: Use features create new feature-extraction model returns values intermediate layer activations: comes handy tasks like neural style transfer, among things.","code":"vgg19 <- application_vgg19() features_list <- lapply(vgg19$layers, function(x) x$output) feat_extraction_model <- keras_model(inputs = vgg19$input,                                      outputs = features_list)  img <- random_normal(c(1, 224, 224, 3)) extracted_features <- feat_extraction_model(img)"},{"path":"https://keras3.posit.co/dev/articles/functional_api.html","id":"extend-the-api-using-custom-layers","dir":"Articles","previous_headings":"","what":"Extend the API using custom layers","title":"The Functional API","text":"keras includes wide range built-layers, example: Convolutional layers: conv_1d, conv_2d, conv_3d, conv_2d_transpose Pooling layers: max_pooling_1d, max_pooling_2d, max_pooling_3d, average_pooling_3d RNN layers: gru, lstm, conv_lstm_2d batch_normalization, dropout, embedding, etc. don’t find need, ’s easy extend API creating layers. layers subclass Layer class implement: call method, specifies computation done layer. build method, creates weights layer (just style convention since can create weights initialize, well). learn creating layers scratch, read custom layers models guide. following basic implementation layer_dense(): serialization support custom layer, define get_config() method returns constructor arguments layer instance: Optionally, implement class method from_config(cls, config) used recreating layer instance given config dictionary. default implementation from_config :","code":"custom_dense <- Layer(   classname = \"CustomDense\",   initialize = function(units = 32) {     super$initialize()     self$units <- as.integer(units)   },   build = function(input_shape) {     self$w <- self$add_weight(       shape = shape(input_shape[[2]], self$units),       initializer = \"random_normal\",       trainable = TRUE,     )     self$b <- self$add_weight(       shape = shape(self$units),             initializer=\"random_normal\",       trainable = TRUE         )   },   call = function(inputs) {     op_matmul(inputs, self$w) + self$b   } )  inputs <- keras_input(c(4)) outputs <- custom_dense(inputs, 10)  model <- keras_model(inputs, outputs) custom_dense <- Layer(   classname = \"CustomDense\",    initialize = function(units = 32, ...) {     super$initialize()     self$units <- as.integer(units)   },    build = function(input_shape) {     self$w <- self$add_weight(       shape = shape(input_shape[[2]], self$units),       initializer = \"random_normal\",       trainable = TRUE,     )     self$b <- self$add_weight(       shape = shape(self$units),             initializer=\"random_normal\",       trainable = TRUE         )   },    call = function(inputs) {     op_matmul(inputs, self$w) + self$b   },    get_config = function() {     list(units = self$units)   } )  inputs <- keras_input(c(4)) outputs <- custom_dense(inputs, 10)  model <- keras_model(inputs, outputs) config <- get_config(model)  new_model <-   from_config(config, custom_objects = list(CustomDense = custom_dense)) from_config <- function(cls, config) {   do.call(cls, config) }"},{"path":"https://keras3.posit.co/dev/articles/functional_api.html","id":"when-to-use-the-functional-api","dir":"Articles","previous_headings":"","what":"When to use the functional API","title":"The Functional API","text":"use Keras functional API create new model, just subclass Model class directly? general, functional API higher-level, easier safer, number features subclassed models support. However, model subclassing provides greater flexibility building models easily expressible directed acyclic graphs layers. example, implement Tree-RNN functional API subclass Model directly. -depth look differences functional API model subclassing, read Symbolic Imperative APIs TensorFlow 2.0?.","code":""},{"path":"https://keras3.posit.co/dev/articles/functional_api.html","id":"functional-api-strengths","dir":"Articles","previous_headings":"When to use the functional API","what":"Functional API strengths:","title":"The Functional API","text":"following properties also true Sequential models (also data structures), true subclassed models (R Python (byte)code, data structures).","code":""},{"path":"https://keras3.posit.co/dev/articles/functional_api.html","id":"less-verbose","dir":"Articles","previous_headings":"When to use the functional API > Functional API strengths:","what":"Less verbose","title":"The Functional API","text":"super$initialize(...), call = function(...), self$..., etc. Compare: subclassed version:","code":"inputs <- keras_input(shape = shape(32)) outputs <- inputs |>   layer_dense(64, activation = \"relu\") |>   layer_dense(10) mlp <- keras_model(inputs, outputs) MLP <- Model(   classname = \"MLP\",   initialize = function(...) {     super$initialize(...)     self$dense_1 <- layer_dense(units = 64, activation = \"relu\")     self$dense_2 <- layer_dense(units = 10)   },   call = function(inputs) {     inputs |>       self$dense_1() |>       self$dense_2()   } )  # Instantiate the model. mlp <- MLP() # Necessary to create the model's state. # The model doesn't have a state until it's called at least once. out <- mlp(op_zeros(c(1, 32)))"},{"path":"https://keras3.posit.co/dev/articles/functional_api.html","id":"model-validation-while-defining-its-connectivity-graph","dir":"Articles","previous_headings":"When to use the functional API > Functional API strengths:","what":"Model validation while defining its connectivity graph","title":"The Functional API","text":"functional API, input specification (shape dtype) created advance (using Input). Every time call layer, layer checks specification passed matches assumptions, raise helpful error message . guarantees model can build functional API run. debugging – convergence-related debugging – happens statically model construction execution time. similar type checking compiler.","code":""},{"path":"https://keras3.posit.co/dev/articles/functional_api.html","id":"a-functional-model-is-plottable-and-inspectable","dir":"Articles","previous_headings":"When to use the functional API > Functional API strengths:","what":"A functional model is plottable and inspectable","title":"The Functional API","text":"can plot model graph, can easily access intermediate nodes graph. example, extract reuse activations intermediate layers (seen previous example):","code":"features_list <- lapply(vgg19$layers, function(x) x$output) feat_extraction_model <- keras_model(inputs = vgg19$input,                                      outputs = features_list)"},{"path":"https://keras3.posit.co/dev/articles/functional_api.html","id":"a-functional-model-can-be-serialized-or-cloned","dir":"Articles","previous_headings":"When to use the functional API > Functional API strengths:","what":"A functional model can be serialized or cloned","title":"The Functional API","text":"functional model data structure rather piece code, safely serializable can saved single file allows recreate exact model without access original code. See serialization & saving guide. serialize subclassed model, necessary implementer specify get_config() from_config() method model level.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/articles/functional_api.html","id":"it-does-not-support-dynamic-architectures","dir":"Articles","previous_headings":"When to use the functional API > Functional API weakness:","what":"It does not support dynamic architectures","title":"The Functional API","text":"functional API treats models DAGs layers. true deep learning architectures, – example, recursive networks Tree RNNs follow assumption implemented functional API.","code":""},{"path":"https://keras3.posit.co/dev/articles/functional_api.html","id":"mix-and-match-api-styles","dir":"Articles","previous_headings":"","what":"Mix-and-match API styles","title":"The Functional API","text":"Choosing functional API Model subclassing isn’t binary decision restricts one category models. models keras API can interact , whether ’re Sequential models, functional models, subclassed models written scratch. can always use functional model Sequential model part subclassed model layer: can use subclassed layer model functional API long implements call method follows one following patterns: call(inputs, ...) – inputs tensor nested structure tensors (e.g. list tensors), ... non-tensor arguments (non-inputs). call(inputs, training = NULL, ...) – training boolean indicating whether layer behave training mode inference mode. call(inputs, mask = NULL, ...) – mask boolean mask tensor (useful RNNs, instance). call(inputs, training = NULL, mask = NULL, ...) – course, can masking training-specific behavior time. Additionally, implement get_config() method custom Layer model, functional models create still serializable cloneable. ’s quick example custom RNN, written scratch, used functional model:","code":"units <- 32 timesteps <- 10 input_dim <- 5  # Define a Functional model inputs <- keras_input(shape(NA, units)) outputs <- inputs |>   layer_global_average_pooling_1d() |>   layer_dense(units = 1)  model <- keras_model(inputs, outputs)  layer_custom_rnn <- Layer(   classname = \"CustomRNN\",   initialize = function(...) {     super$initialize(...)     self$units <- units     self$projection_1 <- layer_dense(units = units, activation = \"tanh\")     self$projection_2 <- layer_dense(units = units, activation = \"tanh\")     self$classifier <- model   },   call = function(inputs, ...) {     outputs <- list()     state <- op_zeros(c(shape(inputs)[[1]], self$units))     for (t in 1:(shape(inputs)[[2]])) {       x <- inputs[, t, ]       h <- self$projection_1(x)       y <- h + self$projection_2(state)       state <- y       outputs[[t]] <- y     }     features <- op_stack(outputs, axis = 2)     self$classifier(features)   } )  rnn <- layer_custom_rnn() out <- rnn(op_zeros(c(1, timesteps, input_dim))) units <- 32 timesteps <- 10 input_dim <- 5 batch_size <- 16  layer_custom_rnn <- Layer(   \"custom_rnn\",   initialize = function(...) {     super$initialize(...)     self$units <- units     self$projection_1 <- layer_dense(units = units, activation = \"tanh\")     self$projection_2 <- layer_dense(units = units, activation = \"tanh\")     self$classifier <- layer_dense(units = 1)   },   call = function(inputs, ...) {     outputs <- list()     state <- op_zeros(c(shape(inputs)[[1]], self$units))     for (t in 1:(shape(inputs)[[2]])) {       x <- inputs[, t, ]       h <- self$projection_1(x)       y <- h + self$projection_2(state)       state <- y       outputs[[t]] <- y     }     features <- op_stack(outputs, axis = 2)     self$classifier(features)   } )  # Note that you specify a static batch size for the inputs with the `batch_shape` # arg, because the inner computation of `layer_custom_rnn()` requires a static batch size # (when you create the `state` zeros tensor). inputs <- keras_input(batch_shape = shape(batch_size, timesteps, input_dim)) outputs <- inputs |>   layer_conv_1d(filters = 32, kernel_size = 3) |>   layer_custom_rnn()  model <- keras_model(inputs, outputs) out <- model(op_zeros(c(1, 10, 5)))"},{"path":"https://keras3.posit.co/dev/articles/getting_started.html","id":"overview","dir":"Articles","previous_headings":"","what":"Overview","title":"Getting Started with Keras","text":"Keras high-level neural networks API developed focus enabling fast experimentation. able go idea result least possible delay key good research. Keras following key features: Allows code run CPU GPU, seamlessly. User-friendly API makes easy quickly prototype deep learning models. Built-support convolutional networks (computer vision), recurrent networks (sequence processing), combination . Supports arbitrary network architectures: multi-input multi-output models, layer sharing, model sharing, etc. means Keras appropriate building essentially deep learning model, memory network neural Turing machine. website provides documentation R interface Keras. See main Keras website https://keras.io additional information project.","code":""},{"path":"https://keras3.posit.co/dev/articles/getting_started.html","id":"installation","dir":"Articles","previous_headings":"","what":"Installation","title":"Getting Started with Keras","text":"First, install keras R package: install development version : Keras R interface requires backend engine installed. TensorFlow default. provide default installation Keras TensorFlow GPU capable, GPU available. want customized installation, e.g. see documentation install_keras() installation section.","code":"install.packages(\"keras3\") remotes::install_github(\"rstudio/keras3\") keras3::install_keras(backend = \"tensorflow\")"},{"path":"https://keras3.posit.co/dev/articles/getting_started.html","id":"mnist-example","dir":"Articles","previous_headings":"","what":"MNIST Example","title":"Getting Started with Keras","text":"can learn basics Keras walking simple example: recognizing handwritten digits MNIST dataset. MNIST consists 28 x 28 grayscale images handwritten digits like :  dataset also includes labels image, telling us digit . example, labels images 5, 0, 4, 1.","code":""},{"path":"https://keras3.posit.co/dev/articles/getting_started.html","id":"preparing-the-data","dir":"Articles","previous_headings":"MNIST Example","what":"Preparing the Data","title":"Getting Started with Keras","text":"MNIST dataset included Keras can accessed using dataset_mnist() function. load dataset create variables test training data: x data 3-d array (images, width, height) grayscale values. prepare data training convert 3-d arrays matrices reshaping width height single dimension (28x28 images flattened length 784 vectors). , convert grayscale values integers ranging 0 255 floating point values ranging 0 1: Note use array_reshape() function rather dim<-() function reshape array. data re-interpreted using row-major semantics (opposed R’s default column-major semantics), turn compatible way numerical libraries called Keras interpret array dimensions. y data integer vector values ranging 0 9. prepare data training one-hot encode vectors binary class matrices using Keras to_categorical() function:","code":"library(keras3) mnist <- dataset_mnist() x_train <- mnist$train$x y_train <- mnist$train$y x_test <- mnist$test$x y_test <- mnist$test$y # reshape x_train <- array_reshape(x_train, c(nrow(x_train), 784)) x_test <- array_reshape(x_test, c(nrow(x_test), 784)) # rescale x_train <- x_train / 255 x_test <- x_test / 255 y_train <- to_categorical(y_train, 10) y_test <- to_categorical(y_test, 10)"},{"path":"https://keras3.posit.co/dev/articles/getting_started.html","id":"defining-the-model","dir":"Articles","previous_headings":"MNIST Example","what":"Defining the Model","title":"Getting Started with Keras","text":"core data structure Keras model, way organize layers. simplest type model Sequential model, linear stack layers. begin creating sequential model adding layers using pipe (|>) operator: input_shape argument first layer specifies shape input data (length 784 numeric vector representing grayscale image). final layer outputs length 10 numeric vector (probabilities digit) using softmax activation function. Use summary() function print details model: Next, compile model appropriate loss function, optimizer, metrics:","code":"model <- keras_model_sequential(input_shape = c(784)) model |>   layer_dense(units = 256, activation = 'relu') |>   layer_dropout(rate = 0.4) |>   layer_dense(units = 128, activation = 'relu') |>   layer_dropout(rate = 0.3) |>   layer_dense(units = 10, activation = 'softmax') summary(model) ## Model: \"sequential\" ## ┏━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━┓ ## ┃ Layer (type)                    ┃ Output Shape           ┃       Param # ┃ ## ┡━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━┩ ## │ dense (Dense)                   │ (None, 256)            │       200,960 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ dropout (Dropout)               │ (None, 256)            │             0 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ dense_1 (Dense)                 │ (None, 128)            │        32,896 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ dropout_1 (Dropout)             │ (None, 128)            │             0 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ dense_2 (Dense)                 │ (None, 10)             │         1,290 │ ## └─────────────────────────────────┴────────────────────────┴───────────────┘ ##  Total params: 235,146 (918.54 KB) ##  Trainable params: 235,146 (918.54 KB) ##  Non-trainable params: 0 (0.00 B) plot(model) model |> compile(   loss = 'categorical_crossentropy',   optimizer = optimizer_rmsprop(),   metrics = c('accuracy') )"},{"path":"https://keras3.posit.co/dev/articles/getting_started.html","id":"training-and-evaluation","dir":"Articles","previous_headings":"MNIST Example","what":"Training and Evaluation","title":"Getting Started with Keras","text":"Use fit() function train model 30 epochs using batches 128 images: history object returned fit() includes loss accuracy metrics can plot: Evaluate model’s performance test data: Generate predictions new data: Keras provides vocabulary building deep learning models simple, elegant, intuitive. Building question answering system, image classification model, neural Turing machine, model just straightforward.","code":"history <- model |> fit(   x_train, y_train,   epochs = 30, batch_size = 128,   validation_split = 0.2 ) plot(history) model |> evaluate(x_test, y_test) ## 313/313 - 1s - 3ms/step - accuracy: 0.9815 - loss: 0.0803 ## $accuracy ## [1] 0.9815 ## ## $loss ## [1] 0.08029716 probs <- model |> predict(x_test) ## 313/313 - 0s - 1ms/step max.col(probs) - 1L ##   [1] 7 2 1 0 4 1 4 9 5 9 0 6 9 0 1 5 9 7 3 4 9 6 6 5 4 0 7 4 0 1 3 1 3 4 7 ##  [36] 2 7 1 2 1 1 7 4 2 3 5 1 2 4 4 6 3 5 5 6 0 4 1 9 5 7 8 9 3 7 4 6 4 3 0 ##  [71] 7 0 2 9 1 7 3 2 9 7 7 6 2 7 8 4 7 3 6 1 3 6 9 3 1 4 1 7 6 9 ##  [ reached getOption(\"max.print\") -- omitted 9900 entries ]"},{"path":"https://keras3.posit.co/dev/articles/getting_started.html","id":"deep-learning-with-r-book","dir":"Articles","previous_headings":"MNIST Example","what":"Deep Learning with R Book","title":"Getting Started with Keras","text":"want comprehensive introduction Keras concepts practice deep learning, recommend Deep Learning R, 2nd Edition book Manning. book collaboration François Chollet, creator (Python) Keras, J.J. Allaire, wrote original R interface Keras, Tomasz Kalinowski, maintainer R interface Keras. book presumes significant knowledge machine learning deep learning, goes way basic theory advanced practical applications, using R interface Keras.","code":""},{"path":"https://keras3.posit.co/dev/articles/getting_started.html","id":"why-this-name-keras","dir":"Articles","previous_headings":"","what":"Why this name, Keras?","title":"Getting Started with Keras","text":"Keras (κέρας) means horn Greek. reference literary image ancient Greek Latin literature, first found Odyssey, dream spirits (Oneiroi, singular Oneiros) divided deceive men false visions, arrive Earth gate ivory, announce future come pass, arrive gate horn. ’s play words κέρας (horn) / κραίνω (fulfill), ἐλέφας (ivory) / ἐλεφαίρομαι (deceive). Keras initially developed part research effort project ONEIROS (Open-ended Neuro-Electronic Intelligent Robot Operating System). “Oneiroi beyond unravelling –can sure tale tell? men look comes pass. Two gates give passage fleeting Oneiroi; one made horn, one ivory. Oneiroi pass sawn ivory deceitful, bearing message fulfilled; come polished horn truth behind , accomplished men see .” Homer, Odyssey 19. 562 ff (Shewring translation).","code":""},{"path":"https://keras3.posit.co/dev/articles/intro_to_keras_for_engineers.html","id":"introduction","dir":"Articles","previous_headings":"","what":"Introduction","title":"Introduction to Keras for engineers","text":"Keras 3 deep learning framework works TensorFlow, JAX, PyTorch interchangeably. notebook walk key Keras 3 workflows. Let’s start installing Keras 3:","code":"install.packages(\"keras3\") keras3::install_keras()"},{"path":"https://keras3.posit.co/dev/articles/intro_to_keras_for_engineers.html","id":"setup","dir":"Articles","previous_headings":"","what":"Setup","title":"Introduction to Keras for engineers","text":"’re going using tensorflow backend – can edit string \"jax\" \"torch\" hit “Restart runtime”, whole notebook run just ! entire guide backend-agnostic.","code":"library(tensorflow, exclude = c(\"shape\", \"set_random_seed\")) library(keras3)  # Note that you must configure the backend # before calling any other keras functions. # The backend cannot be changed once the # package is imported. use_backend(\"tensorflow\")"},{"path":"https://keras3.posit.co/dev/articles/intro_to_keras_for_engineers.html","id":"a-first-example-a-mnist-convnet","dir":"Articles","previous_headings":"","what":"A first example: A MNIST convnet","title":"Introduction to Keras for engineers","text":"Let’s start Hello World ML: training convnet classify MNIST digits. ’s data: ’s model. Different model-building options Keras offers include: Sequential API (use ) Functional API (typical) Writing models via subclassing (advanced use cases) ’s model summary: use compile() method specify optimizer, loss function, metrics monitor. Note JAX TensorFlow backends, XLA compilation turned default. Let’s train evaluate model. ’ll set aside validation split 15% data training monitor generalization unseen data. training, saving model end epoch. can also save model latest state like : reload like : Next, can query predictions class probabilities predict(): ’s basics!","code":"# Load the data and split it between train and test sets c(c(x_train, y_train), c(x_test, y_test)) %<-% keras3::dataset_mnist()  # Scale images to the [0, 1] range x_train <- x_train / 255 x_test <- x_test / 255  # Make sure images have shape (28, 28, 1) x_train <- array_reshape(x_train, c(-1, 28, 28, 1)) x_test <- array_reshape(x_test, c(-1, 28, 28, 1))  dim(x_train) ## [1] 60000    28    28     1 dim(x_test) ## [1] 10000    28    28     1 # Model parameters num_classes <- 10 input_shape <- c(28, 28, 1)  model <- keras_model_sequential(input_shape = input_shape) model |>   layer_conv_2d(filters = 64, kernel_size = c(3, 3), activation = \"relu\") |>   layer_conv_2d(filters = 64, kernel_size = c(3, 3), activation = \"relu\") |>   layer_max_pooling_2d(pool_size = c(2, 2)) |>   layer_conv_2d(filters = 128, kernel_size = c(3, 3), activation = \"relu\") |>   layer_conv_2d(filters = 128, kernel_size = c(3, 3), activation = \"relu\") |>   layer_global_average_pooling_2d() |>   layer_dropout(rate = 0.5) |>   layer_dense(units = num_classes, activation = \"softmax\") summary(model) ## Model: \"sequential\" ## ┏━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━┓ ## ┃ Layer (type)                    ┃ Output Shape           ┃       Param # ┃ ## ┡━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━┩ ## │ conv2d (Conv2D)                 │ (None, 26, 26, 64)     │           640 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ conv2d_1 (Conv2D)               │ (None, 24, 24, 64)     │        36,928 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ max_pooling2d (MaxPooling2D)    │ (None, 12, 12, 64)     │             0 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ conv2d_2 (Conv2D)               │ (None, 10, 10, 128)    │        73,856 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ conv2d_3 (Conv2D)               │ (None, 8, 8, 128)      │       147,584 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ global_average_pooling2d        │ (None, 128)            │             0 │ ## │ (GlobalAveragePooling2D)        │                        │               │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ dropout (Dropout)               │ (None, 128)            │             0 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ dense (Dense)                   │ (None, 10)             │         1,290 │ ## └─────────────────────────────────┴────────────────────────┴───────────────┘ ##  Total params: 260,298 (1016.79 KB) ##  Trainable params: 260,298 (1016.79 KB) ##  Non-trainable params: 0 (0.00 B) model |> compile(   optimizer = \"adam\",   loss = \"sparse_categorical_crossentropy\",   metrics = list(     metric_sparse_categorical_accuracy(name = \"acc\")   ) ) batch_size <- 128 epochs <- 10  callbacks <- list(   callback_model_checkpoint(filepath=\"model_at_epoch_{epoch}.keras\"),   callback_early_stopping(monitor=\"val_loss\", patience=2) )  model |> fit(   x_train, y_train,   batch_size = batch_size,   epochs = epochs,   validation_split = 0.15,   callbacks = callbacks ) ## Epoch 1/10 ## 399/399 - 8s - 19ms/step - acc: 0.7477 - loss: 0.7456 - val_acc: 0.9654 - val_loss: 0.1189 ## Epoch 2/10 ## 399/399 - 3s - 7ms/step - acc: 0.9377 - loss: 0.2072 - val_acc: 0.9764 - val_loss: 0.0786 ## Epoch 3/10 ## 399/399 - 3s - 7ms/step - acc: 0.9567 - loss: 0.1474 - val_acc: 0.9820 - val_loss: 0.0623 ## Epoch 4/10 ## 399/399 - 3s - 7ms/step - acc: 0.9648 - loss: 0.1182 - val_acc: 0.9860 - val_loss: 0.0490 ## Epoch 5/10 ## 399/399 - 3s - 7ms/step - acc: 0.9707 - loss: 0.1007 - val_acc: 0.9872 - val_loss: 0.0475 ## Epoch 6/10 ## 399/399 - 3s - 7ms/step - acc: 0.9753 - loss: 0.0868 - val_acc: 0.9886 - val_loss: 0.0403 ## Epoch 7/10 ## 399/399 - 3s - 7ms/step - acc: 0.9761 - loss: 0.0790 - val_acc: 0.9890 - val_loss: 0.0408 ## Epoch 8/10 ## 399/399 - 3s - 7ms/step - acc: 0.9794 - loss: 0.0685 - val_acc: 0.9878 - val_loss: 0.0443 score <- model |> evaluate(x_test, y_test, verbose = 0) save_model(model, \"final_model.keras\", overwrite=TRUE) model <- load_model(\"final_model.keras\") predictions <- model |> predict(x_test) ## 313/313 - 1s - 2ms/step dim(predictions) ## [1] 10000    10"},{"path":"https://keras3.posit.co/dev/articles/intro_to_keras_for_engineers.html","id":"writing-cross-framework-custom-components","dir":"Articles","previous_headings":"","what":"Writing cross-framework custom components","title":"Introduction to Keras for engineers","text":"Keras enables write custom Layers, Models, Metrics, Losses, Optimizers work across TensorFlow, JAX, PyTorch codebase. Let’s take look custom layers first. op_ namespace contains: implementation NumPy API, e.g. op_stack op_matmul. set neural network specific ops absent NumPy, op_conv op_binary_crossentropy. Let’s make custom Dense layer works backends: Next, let’s make custom Dropout layer relies random_* namespace: Next, let’s write custom subclassed model uses two custom layers: Let’s compile fit :","code":"layer_my_dense <- Layer(   classname = \"MyDense\",   initialize = function(units, activation = NULL, name = NULL, ...) {     super$initialize(name = name, ...)     self$units <- units     self$activation <- activation   },   build = function(input_shape) {     input_dim <- tail(input_shape, 1)     self$w <- self$add_weight(       shape = shape(input_dim, self$units),       initializer = initializer_glorot_normal(),       name = \"kernel\",       trainable = TRUE     )     self$b <- self$add_weight(       shape = shape(self$units),       initializer = initializer_zeros(),       name = \"bias\",       trainable = TRUE     )   },   call = function(inputs) {     # Use Keras ops to create backend-agnostic layers/metrics/etc.     x <- op_matmul(inputs, self$w) + self$b     if (!is.null(self$activation))       x <- self$activation(x)     x   } ) layer_my_dropout <- Layer(   \"MyDropout\",   initialize = function(rate, name = NULL, seed = NULL, ...) {     super$initialize(name = name)     self$rate <- rate     # Use seed_generator for managing RNG state.     # It is a state element and its seed variable is     # tracked as part of `layer$variables`.     self$seed_generator <- random_seed_generator(seed)   },   call = function(inputs) {     # Use `keras3::random_*` for random ops.     random_dropout(inputs, self$rate, seed = self$seed_generator)   } ) MyModel <- Model(   \"MyModel\",   initialize = function(num_classes, ...) {     super$initialize(...)     self$conv_base <-       keras_model_sequential() |>       layer_conv_2d(64, kernel_size = c(3, 3), activation = \"relu\") |>       layer_conv_2d(64, kernel_size = c(3, 3), activation = \"relu\") |>       layer_max_pooling_2d(pool_size = c(2, 2)) |>       layer_conv_2d(128, kernel_size = c(3, 3), activation = \"relu\") |>       layer_conv_2d(128, kernel_size = c(3, 3), activation = \"relu\") |>       layer_global_average_pooling_2d()      self$dp <- layer_my_dropout(rate = 0.5)     self$dense <- layer_my_dense(units = num_classes,                                  activation = activation_softmax)   },   call = function(inputs) {     inputs |>       self$conv_base() |>       self$dp() |>       self$dense()   } ) model <- MyModel(num_classes = 10) model |> compile(   loss = loss_sparse_categorical_crossentropy(),   optimizer = optimizer_adam(learning_rate = 1e-3),   metrics = list(     metric_sparse_categorical_accuracy(name = \"acc\")   ) )  model |> fit(   x_train, y_train,   batch_size = batch_size,   epochs = 1, # For speed   validation_split = 0.15 ) ## 399/399 - 7s - 17ms/step - acc: 0.7350 - loss: 0.7737 - val_acc: 0.9276 - val_loss: 0.2389"},{"path":"https://keras3.posit.co/dev/articles/intro_to_keras_for_engineers.html","id":"training-models-on-arbitrary-data-sources","dir":"Articles","previous_headings":"","what":"Training models on arbitrary data sources","title":"Introduction to Keras for engineers","text":"Keras models can trained evaluated wide variety data sources, independently backend ’re using. includes: Arrays Dataframes TensorFlow tf_dataset objects PyTorch DataLoader objects Keras PyDataset objects work whether ’re using TensorFlow, JAX, PyTorch Keras backend. Let’s try tf_dataset:","code":"library(tfdatasets, exclude = \"shape\")  train_dataset <- list(x_train, y_train) |>   tensor_slices_dataset() |>   dataset_batch(batch_size) |>   dataset_prefetch(buffer_size = tf$data$AUTOTUNE)  test_dataset <- list(x_test, y_test) |>   tensor_slices_dataset() |>   dataset_batch(batch_size) |>   dataset_prefetch(buffer_size = tf$data$AUTOTUNE)  model <- MyModel(num_classes = 10) model |> compile(   loss = loss_sparse_categorical_crossentropy(),   optimizer = optimizer_adam(learning_rate = 1e-3),   metrics = list(     metric_sparse_categorical_accuracy(name = \"acc\")   ) )  model |> fit(train_dataset, epochs = 1, validation_data = test_dataset) ## 469/469 - 8s - 17ms/step - acc: 0.7492 - loss: 0.7477 - val_acc: 0.9125 - val_loss: 0.2952"},{"path":"https://keras3.posit.co/dev/articles/intro_to_keras_for_engineers.html","id":"further-reading","dir":"Articles","previous_headings":"","what":"Further reading","title":"Introduction to Keras for engineers","text":"concludes short overview new multi-backend capabilities Keras 3. Next, can learn :","code":""},{"path":"https://keras3.posit.co/dev/articles/intro_to_keras_for_engineers.html","id":"how-to-customize-what-happens-in-fit","dir":"Articles","previous_headings":"Further reading","what":"How to customize what happens in fit()","title":"Introduction to Keras for engineers","text":"Want implement non-standard training algorithm still want benefit power usability fit()? ’s easy customize fit() support arbitrary use cases: Customizing happens fit() TensorFlow","code":""},{"path":"https://keras3.posit.co/dev/articles/intro_to_keras_for_engineers.html","id":"how-to-write-custom-training-loops","dir":"Articles","previous_headings":"","what":"How to write custom training loops","title":"Introduction to Keras for engineers","text":"Writing training loop scratch TensorFlow","code":""},{"path":"https://keras3.posit.co/dev/articles/intro_to_keras_for_engineers.html","id":"how-to-distribute-training","dir":"Articles","previous_headings":"","what":"How to distribute training","title":"Introduction to Keras for engineers","text":"Guide distributed training TensorFlow Enjoy library! 🚀","code":""},{"path":"https://keras3.posit.co/dev/articles/intro_to_keras_for_researchers.html","id":"setup","dir":"Articles","previous_headings":"","what":"Setup","title":"Introduction to Keras for Researchers","text":"","code":"library(keras3) library(tensorflow)"},{"path":"https://keras3.posit.co/dev/articles/intro_to_keras_for_researchers.html","id":"introduction","dir":"Articles","previous_headings":"","what":"Introduction","title":"Introduction to Keras for Researchers","text":"machine learning researcher? publish NeurIPS push state---art CV NLP? guide serve first introduction core Keras & TensorFlow API concepts. guide, learn : Tensors, variables, gradients TensorFlow Creating layers subclassing [Layer] class Writing low-level training loops Tracking losses created layers via add_loss() method Tracking metrics low-level training loop Speeding execution compiled [tensorflow::tf_function()] Executing layers training inference mode Keras Functional API also see Keras API action two end--end research examples: Variational Autoencoder, Hypernetwork.","code":""},{"path":"https://keras3.posit.co/dev/articles/intro_to_keras_for_researchers.html","id":"tensors","dir":"Articles","previous_headings":"","what":"Tensors","title":"Introduction to Keras for Researchers","text":"TensorFlow infrastructure layer differentiable programming. heart, ’s framework manipulating N-dimensional arrays (tensors), much like NumPy. However, three key differences NumPy TensorFlow: TensorFlow can leverage hardware accelerators GPUs TPUs. TensorFlow can automatically compute gradient arbitrary differentiable tensor expressions. TensorFlow computation can distributed large numbers devices single machine, large number machines (potentially multiple devices ). Let’s take look object core TensorFlow: Tensor. ’s constant tensor: can get value R array calling .array(): features attributes dtype shape: common way create constant tensors via tf$ones tf$zeros: can also create random constant tensors:","code":"x <- tf$constant(rbind(c(5, 2), c(1, 3))) print(x) ## tf.Tensor( ## [[5. 2.] ##  [1. 3.]], shape=(2, 2), dtype=float64) as.array(x) ##      [,1] [,2] ## [1,]    5    2 ## [2,]    1    3 x$dtype ## tf.float64 x$shape ## TensorShape([2, 2]) tf$ones(shape = shape(2, 1)) ## tf.Tensor( ## [[1.] ##  [1.]], shape=(2, 1), dtype=float32) tf$zeros(shape = shape(2, 1)) ## tf.Tensor( ## [[0.] ##  [0.]], shape=(2, 1), dtype=float32) x <- random_normal(shape = c(2, 2), mean = 0.0, stddev = 1.0) x <- random_uniform(shape = c(2, 2), minval = 0, maxval = 10)"},{"path":"https://keras3.posit.co/dev/articles/intro_to_keras_for_researchers.html","id":"variables","dir":"Articles","previous_headings":"","what":"Variables","title":"Introduction to Keras for Researchers","text":"Variables special tensors used store mutable state (weights neural network). create Variable using initial value: update value Variable using methods $assign(value), $assign_add(increment), $assign_sub(decrement):","code":"initial_value <- random_normal(shape=c(2, 2)) a <- tf$Variable(initial_value) print(a) ## <tf.Variable 'Variable:0' shape=(2, 2) dtype=float32, numpy= ## array([[ 0.9057419 ,  0.7916686 ], ##        [ 0.28754202, -0.5408822 ]], dtype=float32)> new_value <- random_normal(shape=c(2, 2)) a$assign(new_value) ## <tf.Variable 'UnreadVariable' shape=(2, 2) dtype=float32, numpy= ## array([[-0.3405368 , -2.1463926 ], ##        [ 1.2602988 ,  0.12241419]], dtype=float32)> added_value <- random_normal(shape=c(2, 2)) a$assign_add(added_value) ## <tf.Variable 'UnreadVariable' shape=(2, 2) dtype=float32, numpy= ## array([[ 0.04820395, -2.6854615 ], ##        [ 0.23246336,  1.4535258 ]], dtype=float32)>"},{"path":"https://keras3.posit.co/dev/articles/intro_to_keras_for_researchers.html","id":"doing-math-in-tensorflow","dir":"Articles","previous_headings":"","what":"Doing math in TensorFlow","title":"Introduction to Keras for Researchers","text":"’ve used NumPy, math TensorFlow look familiar. main difference TensorFlow code can run GPU TPU.","code":"a <- random_normal(shape=c(2, 2)) b <- random_normal(shape=c(2, 2))  c <- a + b d <- tf$square(c) e <- tf$exp(d)"},{"path":"https://keras3.posit.co/dev/articles/intro_to_keras_for_researchers.html","id":"gradients","dir":"Articles","previous_headings":"","what":"Gradients","title":"Introduction to Keras for Researchers","text":"’s another big difference R: can automatically retrieve gradient differentiable expression. Just open GradientTape, start “watching” tensor via tape$watch(), compose differentiable expression using tensor input: default, variables watched automatically, don’t need manually watch : Note can compute higher-order derivatives nesting tapes:","code":"a <- random_normal(shape=c(2, 2)) b <- random_normal(shape=c(2, 2))  with(tf$GradientTape() %as% tape, {   tape$watch(a)  # Start recording the history of operations applied to `a`   c <- tf$sqrt(tf$square(a) + tf$square(b))  # Do some math using `a`   # What's the gradient of `c` with respect to `a`?   dc_da <- tape$gradient(c, a)   print(dc_da) }) ## tf.Tensor( ## [[ 0.9969011  -0.7707146 ] ##  [ 0.23378514  0.96255165]], shape=(2, 2), dtype=float32) a <- tf$Variable(a)  with(tf$GradientTape() %as% tape, {   c <- tf$sqrt(tf$square(a) + tf$square(b))   dc_da <- tape$gradient(c, a)   print(dc_da) }) ## tf.Tensor( ## [[ 0.9969011  -0.7707146 ] ##  [ 0.23378514  0.96255165]], shape=(2, 2), dtype=float32) with(tf$GradientTape() %as% outer_tape, {   with(tf$GradientTape() %as% tape, {     c <- tf$sqrt(tf$square(a) + tf$square(b))     dc_da <- tape$gradient(c, a)   })   d2c_da2 <- outer_tape$gradient(dc_da, a)   print(d2c_da2) }) ## tf.Tensor( ## [[3.3447742e-03 7.1282005e-01] ##  [5.7464113e+00 5.5013180e-02]], shape=(2, 2), dtype=float32)"},{"path":"https://keras3.posit.co/dev/articles/intro_to_keras_for_researchers.html","id":"keras-layers","dir":"Articles","previous_headings":"","what":"Keras layers","title":"Introduction to Keras for Researchers","text":"TensorFlow infrastructure layer differentiable programming, dealing tensors, variables, gradients, Keras user interface deep learning, dealing layers, models, optimizers, loss functions, metrics, . Keras serves high-level API TensorFlow: Keras makes TensorFlow simple productive. Layer class fundamental abstraction Keras. Layer encapsulates state (weights) computation (defined call method). simple layer looks like . self$add_weight() method gives shortcut creating weights: use Layer instance much like R function: weight variables (created initialize) automatically tracked weights property: many built-layers available, Dense Conv2D LSTM fancier ones like Conv3DTranspose ConvLSTM2D. smart reusing built-functionality.","code":"Linear <- new_layer_class(   \"Linear\",   initialize = function(units = 32, input_dim = 32) {     super$initialize()     self$w <- self$add_weight(       shape = shape(input_dim, units),       initializer = \"random_normal\",       trainable = TRUE     )     self$b <- self$add_weight(       shape = shape(units),       initializer = \"zeros\",       trainable = TRUE     )   },   call = function(inputs) {     tf$matmul(inputs, self$w) + self$b   } ) # Instantiate our layer. linear_layer <- Linear(units=4, input_dim=2)  # The layer can be treated as a function. # Here we call it on some data. y <- linear_layer(tf$ones(shape(2, 2))) linear_layer$weights ## [[1]] ## <KerasVariable shape=(2, 4), dtype=float32, path=linear/variable> ## ## [[2]] ## <KerasVariable shape=(4), dtype=float32, path=linear/variable_1>"},{"path":"https://keras3.posit.co/dev/articles/intro_to_keras_for_researchers.html","id":"layer-weight-creation-in-buildinput_shape","dir":"Articles","previous_headings":"","what":"Layer weight creation in build(input_shape)","title":"Introduction to Keras for Researchers","text":"’s often good idea defer weight creation build() method, don’t need specify input dim/shape layer construction time:","code":"Linear <- new_layer_class(   \"Linear\",   initialize = function(units = 32) {     super$initialize()     self$units <- units   },   build = function(input_shape) {     self$w <- self$add_weight(       shape = shape(input_shape[-1], self$units),       initializer = \"random_normal\",       trainable = TRUE     )     self$b <- self$add_weight(       shape = shape(self$units),       initializer = \"zeros\",       trainable = TRUE     )   },   call = function(inputs) {     tf$matmul(inputs, self$w) + self$b   } )  # Instantiate our layer. linear_layer <- Linear(units = 4)  # This will also call `build(input_shape)` and create the weights. y <- linear_layer(tf$ones(shape(2, 2)))"},{"path":"https://keras3.posit.co/dev/articles/intro_to_keras_for_researchers.html","id":"layer-gradients","dir":"Articles","previous_headings":"","what":"Layer gradients","title":"Introduction to Keras for Researchers","text":"can automatically retrieve gradients weights layer calling inside GradientTape. Using gradients, can update weights layer, either manually, using optimizer object. course, can modify gradients using , need .","code":"# Prepare a dataset. c(c(x_train, y_train), .) %<-% dataset_mnist()  x_train <- array_reshape(x_train, c(60000, 784)) / 255  dataset <- tfdatasets::tensor_slices_dataset(list(x_train, y_train)) %>%   tfdatasets::dataset_shuffle(buffer_size=1024) %>%   tfdatasets::dataset_batch(64)  # Instantiate our linear layer (defined above) with 10 units. linear_layer <- Linear(units = 10)  # Instantiate a logistic loss function that expects integer targets. loss_fn <- loss_sparse_categorical_crossentropy(from_logits=TRUE)  # Instantiate an optimizer. optimizer <- optimizer_sgd(learning_rate=1e-3)  # Iterate over the batches of the dataset. coro::loop(for(data in dataset) {   # Open a GradientTape.   with(tf$GradientTape() %as% tape, {     # Forward pass.     logits <- linear_layer(data[[1]])      # Loss value for this batch.     loss_value <- loss_fn(data[[2]], logits)   })    # Get gradients of the loss wrt the weights.   gradients <- tape$gradient(loss_value, linear_layer$trainable_weights)    # Update the weights of our linear layer.   optimizer$apply_gradients(zip_lists(gradients, linear_layer$trainable_weights)) }) loss_value ## tf.Tensor(1.2819729, shape=(), dtype=float32)"},{"path":"https://keras3.posit.co/dev/articles/intro_to_keras_for_researchers.html","id":"trainable-and-non-trainable-weights","dir":"Articles","previous_headings":"","what":"Trainable and non-trainable weights","title":"Introduction to Keras for Researchers","text":"Weights created layers can either trainable non-trainable. ’re exposed trainable_weights non_trainable_weights respectively. ’s layer non-trainable weight:","code":"ComputeSum <- new_layer_class(   \"ComputeSum\",   initialize = function(input_dim) {     super$initialize()     # Create a non-trainable weight.     self$total <- self$add_weight(       initializer = \"zeros\",       shape = shape(input_dim),       trainable = FALSE     )   },   call = function(inputs) {     self$total$assign_add(tf$reduce_sum(inputs, axis=0L))     self$total   } )  my_sum <- ComputeSum(input_dim = 2) x <- tf$ones(shape(2, 2))  as.array(my_sum(x)) ## [1] 2 2 as.array(my_sum(x)) ## [1] 4 4 my_sum$trainable_weights ## list()"},{"path":"https://keras3.posit.co/dev/articles/intro_to_keras_for_researchers.html","id":"layers-that-own-layers","dir":"Articles","previous_headings":"","what":"Layers that own layers","title":"Introduction to Keras for Researchers","text":"Layers can recursively nested create bigger computation blocks. layer track weights sublayers (trainable non-trainable). Note manually-created MLP equivalent following built-option:","code":"# Let's reuse the Linear class # with a `build` method that we defined above.  MLP <- new_layer_class(   \"MLP\",   initialize = function() {     super$initialize()     self$linear_1 <- Linear(units = 32)     self$linear_2 <- Linear(units = 32)     self$linear_3 <- Linear(units = 10)   },   call = function(inputs) {     x <- self$linear_1(inputs)     x <- tf$nn$relu(x)     x <- self$linear_2(x)     x <- tf$nn$relu(x)     return(self$linear_3(x))   } )  mlp <- MLP()  # The first call to the `mlp` object will create the weights. y <- mlp(tf$ones(shape=shape(3, 64)))  # Weights are recursively tracked. length(mlp$weights) ## [1] 6 mlp <- keras_model_sequential() %>%   layer_dense(units = 32, activation = \"relu\") %>%   layer_dense(units = 32, activation = \"relu\") %>%   layer_dense(units = 10)"},{"path":"https://keras3.posit.co/dev/articles/intro_to_keras_for_researchers.html","id":"tracking-losses-created-by-layers","dir":"Articles","previous_headings":"","what":"Tracking losses created by layers","title":"Introduction to Keras for Researchers","text":"Layers can create losses forward pass via add_loss() method. especially useful regularization losses. losses created sublayers recursively tracked parent layers. ’s layer creates activity regularization loss: model incorporating layer track regularization loss: losses cleared top-level layer start forward pass – don’t accumulate. layer.losses always contains losses created last forward pass. typically use losses summing computing gradients writing training loop.","code":"# A layer that creates an activity sparsity regularization loss ActivityRegularization <- new_layer_class(   \"ActivityRegularization\",   initialize = function(rate=1e-2) {     super$initialize()     self$rate <- rate   },   call = function(inputs) {     self$add_loss(self$rate * tf$reduce_sum(tf$abs(inputs)))     inputs   } ) # Let's use the loss layer in a MLP block. SparseMLP <- new_layer_class(   \"SparseMLP\",   initialize = function() {     super$initialize()     self$linear_1 <- Linear(units = 32)     self$reg <- ActivityRegularization(rate = 1e-2)     self$linear_3 <- Linear(units = 10)   },   call = function(inputs) {     x <- self$linear_1(inputs)     x <- tf$nn$relu(x)     x <- self$reg(x)     return(self$linear_3(x))   } )  mlp <- SparseMLP() y <- mlp(tf$ones(shape(10, 10)))  mlp$losses  # List containing one float32 scalar ## [[1]] ## tf.Tensor(0.18065463, shape=(), dtype=float32) # Losses correspond to the *last* forward pass. mlp <- SparseMLP() mlp(tf$ones(shape(10, 10))) ## tf.Tensor( ## [[ 0.0388482  -0.03920118  0.01624808 -0.01361975 -0.01354899  0.07107338 ##   -0.01077365  0.05688906 -0.02838149 -0.04084621] ##  [ 0.0388482  -0.03920118  0.01624808 -0.01361975 -0.01354899  0.07107338 ##   -0.01077365  0.05688906 -0.02838149 -0.04084621] ##  [ 0.0388482  -0.03920118  0.01624808 -0.01361975 -0.01354899  0.07107338 ##   -0.01077365  0.05688906 -0.02838149 -0.04084621] ##  [ 0.0388482  -0.03920118  0.01624808 -0.01361975 -0.01354899  0.07107338 ##   -0.01077365  0.05688906 -0.02838149 -0.04084621] ##  [ 0.0388482  -0.03920118  0.01624808 -0.01361975 -0.01354899  0.07107338 ##   -0.01077365  0.05688906 -0.02838149 -0.04084621] ##  [ 0.0388482  -0.03920118  0.01624808 -0.01361975 -0.01354899  0.07107338 ##   -0.01077365  0.05688906 -0.02838149 -0.04084621] ##  [ 0.0388482  -0.03920118  0.01624808 -0.01361975 -0.01354899  0.07107338 ##   -0.01077365  0.05688906 -0.02838149 -0.04084621] ##  [ 0.0388482  -0.03920118  0.01624808 -0.01361975 -0.01354899  0.07107338 ##   -0.01077365  0.05688906 -0.02838149 -0.04084621] ##  [ 0.0388482  -0.03920118  0.01624808 -0.01361975 -0.01354899  0.07107338 ##   -0.01077365  0.05688906 -0.02838149 -0.04084621] ##  [ 0.0388482  -0.03920118  0.01624808 -0.01361975 -0.01354899  0.07107338 ##   -0.01077365  0.05688906 -0.02838149 -0.04084621]], shape=(10, 10), dtype=float32) length(mlp$losses) ## [1] 1 mlp(tf$ones(shape(10, 10))) ## tf.Tensor( ## [[ 0.0388482  -0.03920118  0.01624808 -0.01361975 -0.01354899  0.07107338 ##   -0.01077365  0.05688906 -0.02838149 -0.04084621] ##  [ 0.0388482  -0.03920118  0.01624808 -0.01361975 -0.01354899  0.07107338 ##   -0.01077365  0.05688906 -0.02838149 -0.04084621] ##  [ 0.0388482  -0.03920118  0.01624808 -0.01361975 -0.01354899  0.07107338 ##   -0.01077365  0.05688906 -0.02838149 -0.04084621] ##  [ 0.0388482  -0.03920118  0.01624808 -0.01361975 -0.01354899  0.07107338 ##   -0.01077365  0.05688906 -0.02838149 -0.04084621] ##  [ 0.0388482  -0.03920118  0.01624808 -0.01361975 -0.01354899  0.07107338 ##   -0.01077365  0.05688906 -0.02838149 -0.04084621] ##  [ 0.0388482  -0.03920118  0.01624808 -0.01361975 -0.01354899  0.07107338 ##   -0.01077365  0.05688906 -0.02838149 -0.04084621] ##  [ 0.0388482  -0.03920118  0.01624808 -0.01361975 -0.01354899  0.07107338 ##   -0.01077365  0.05688906 -0.02838149 -0.04084621] ##  [ 0.0388482  -0.03920118  0.01624808 -0.01361975 -0.01354899  0.07107338 ##   -0.01077365  0.05688906 -0.02838149 -0.04084621] ##  [ 0.0388482  -0.03920118  0.01624808 -0.01361975 -0.01354899  0.07107338 ##   -0.01077365  0.05688906 -0.02838149 -0.04084621] ##  [ 0.0388482  -0.03920118  0.01624808 -0.01361975 -0.01354899  0.07107338 ##   -0.01077365  0.05688906 -0.02838149 -0.04084621]], shape=(10, 10), dtype=float32) length(mlp$losses)  # No accumulation. ## [1] 1 # Let's demonstrate how to use these losses in a training loop.  # Prepare a dataset. c(c(x_train, y_train), .) %<-% dataset_mnist() x_train <- array_reshape(x_train, c(60000, 784)) / 255  dataset <- tfdatasets::tensor_slices_dataset(list(x_train, y_train)) %>%   tfdatasets::dataset_shuffle(buffer_size=1024) %>%   tfdatasets::dataset_batch(64)  # A new MLP. mlp <- SparseMLP()  # Loss and optimizer. loss_fn <- loss_sparse_categorical_crossentropy(from_logits=TRUE) optimizer <- optimizer_sgd(learning_rate=1e-3)  coro::loop(for(data in dataset) {   x <- data[[1]]   y <- data[[2]]   with(tf$GradientTape() %as% tape, {     # Forward pass.     logits <- mlp(x)      # External loss value for this batch.     loss <- loss_fn(y, logits)      # Add the losses created during the forward pass.     loss <- loss + Reduce(`+`, mlp$losses)      # Get gradients of the loss wrt the weights.     gradients <- tape$gradient(loss, mlp$trainable_weights)      # Update the weights of our linear layer.     optimizer$apply_gradients(zip_lists(gradients, mlp$trainable_weights))   }) })"},{"path":"https://keras3.posit.co/dev/articles/intro_to_keras_for_researchers.html","id":"keeping-track-of-training-metrics","dir":"Articles","previous_headings":"","what":"Keeping track of training metrics","title":"Introduction to Keras for Researchers","text":"Keras offers broad range built-metrics, like metric_auc metric_precision_at_recall. ’s also easy create metrics lines code. use metric custom training loop, : Instantiate metric object, e.g. metric = metric_auc() Call metric$udpate_state(targets, predictions) method batch data Query result via metric$result() Reset metric’s state end epoch start evaluation via metric$reset_state() ’s simple example: can also define metrics subclassing keras.metrics.Metric. need override three functions called : Override update_state() update statistic values. Override result() return metric value. Override reset_state() reset metric initial state. example implement F1-score metric (support sample weighting). Let’s test-drive :","code":"# Instantiate a metric object accuracy <- metric_sparse_categorical_accuracy()  # Prepare our layer, loss, and optimizer. model <- keras_model_sequential() %>%   layer_dense(units = 32, activation = \"relu\") %>%   layer_dense(units = 32, activation = \"relu\") %>%   layer_dense(units = 10) loss_fn <- loss_sparse_categorical_crossentropy(from_logits = TRUE) optimizer <- optimizer_adam(learning_rate=1e-3)  for (epoch in seq_len(2)) {   coro::loop(for (data in dataset) {     x <- data[[1]]     y <- data[[2]]     with(tf$GradientTape() %as% tape, {       # Forward pass.       logits <- model(x)        # External loss value for this batch.       loss_value <- loss_fn(y, logits)     })      # Update the state of the `accuracy` metric.     accuracy$update_state(y, logits)      # Update the weights of the model to minimize the loss value.     gradients <- tape$gradient(loss_value, model$trainable_weights)     optimizer$apply_gradients(zip_lists(gradients, model$trainable_weights))    })   cat(\"Epoch:\", epoch, \"Accuracy:\", as.numeric(accuracy$result()), \"\\n\")   accuracy$reset_state() } ## Epoch: 1 Accuracy: 0.8757833 ## Epoch: 2 Accuracy: 0.93915 F1Score <- new_metric_class(   \"F1Score\",   initialize = function(self, name=\"f1_score\", dtype=\"float32\", threshold=0.5, ...) {     super$initialize(name=name, dtype=dtype, ...)     self$threshold <- threshold     self$true_positives <- self$add_weight(       name=\"tp\", dtype=dtype, initializer=\"zeros\"     )     self$false_positives <- self$add_weight(       name=\"fp\", dtype=dtype, initializer=\"zeros\"     )     self$false_negatives <- self$add_weight(       name=\"fn\", dtype=dtype, initializer=\"zeros\"     )   },   update_state = function(y_true, y_pred, sample_weight=NULL) {     y_pred <- tf$math$greater_equal(y_pred, self$threshold)     y_true <- tf$cast(y_true, tf$bool)     y_pred <- tf$cast(y_pred, tf$bool)      true_positives <- tf$cast(y_true & y_pred, self$dtype)     false_positives <- tf$cast((!y_true) & y_pred, self$dtype)     false_negatives <- tf$cast(y_true & (!y_pred), self$dtype)      if (!is.null(sample_weight)) {       sample_weight <- tf$cast(sample_weight, self$dtype)       true_positives <- true_positives * sample_weight       false_positives <- false_positives * sample_weight       false_negatives <- false_negatives * sample_weight     }      self$true_positives$assign_add(tf$reduce_sum(true_positives))     self$false_positives$assign_add(tf$reduce_sum(false_positives))     self$false_negatives$assign_add(tf$reduce_sum(false_negatives))   },    result = function() {     precision <- self$true_positives / (self$true_positives + self$false_positives)     recall <- self$true_positives / (self$true_positives + self$false_negatives)     f1_score <- 2 * precision * recall / (precision + recall)     f1_score   },    reset_state = function() {     self$true_positives$assign(0)     self$false_positives$assign(0)     self$false_negatives$assign(0)   } ) m <- F1Score() m$update_state(c(0, 1, 0, 0), c(0.3, 0.5, 0.8, 0.9)) cat(\"Intermediate result:\", as.numeric(m$result()), \"\\n\") ## Intermediate result: 0.5 m$update_state(c(1, 1, 1, 1), c(0.1, 0.7, 0.6, 0.0)) cat(\"Final result:\", as.numeric(m$result()), \"\\n\") ## Final result: 0.6"},{"path":"https://keras3.posit.co/dev/articles/intro_to_keras_for_researchers.html","id":"compiled-functions","dir":"Articles","previous_headings":"","what":"Compiled functions","title":"Introduction to Keras for Researchers","text":"Running eagerly great debugging, get better performance compiling computation static graphs. Static graphs researcher’s best friends. can compile function wrapping tf.function decorator.","code":"# Prepare our layer, loss, and optimizer. model <- keras_model_sequential() %>%   layer_dense(units = 32, activation = \"relu\") %>%   layer_dense(units = 32, activation = \"relu\") %>%   layer_dense(units = 10) loss_fn <- loss_sparse_categorical_crossentropy(from_logits = TRUE) optimizer <- optimizer_adam(learning_rate=1e-3)  # Create a training step function. train_on_batch <- tf_function(function(x, y) {   with(tf$GradientTape() %as% tape, {     # Forward pass.     logits <- model(x)     # External loss value for this batch.     loss_value <- loss_fn(y, logits)   })   # Update the weights of the model to minimize the loss value.   gradients <- tape$gradient(loss_value, model$trainable_weights)   optimizer$apply_gradients(zip_lists(gradients, model$trainable_weights))   loss_value })   # Prepare a dataset. c(c(x_train, y_train), .) %<-% dataset_mnist() x_train <- array_reshape(x_train, c(60000, 784)) / 255  dataset <- tfdatasets::tensor_slices_dataset(list(x_train, y_train)) %>%   tfdatasets::dataset_shuffle(buffer_size=1024) %>%   tfdatasets::dataset_batch(64)  i <- 0 coro::loop(for (data in dataset) {   i <- i + 1   x <- data[[1]]   y <- data[[2]]   loss <- train_on_batch(x, y)   if (i %% 100 == 0)     cat(\"Loss:\", as.numeric(loss), \"\\n\") }) ## Loss: 0.551749 ## Loss: 0.2131135 ## Loss: 0.2765952 ## Loss: 0.1296219 ## Loss: 0.2657076 ## Loss: 0.2683381 ## Loss: 0.1570166 ## Loss: 0.3139241 ## Loss: 0.08981849"},{"path":"https://keras3.posit.co/dev/articles/intro_to_keras_for_researchers.html","id":"training-mode-inference-mode","dir":"Articles","previous_headings":"","what":"Training mode & inference mode","title":"Introduction to Keras for Researchers","text":"layers, particular BatchNormalization layer Dropout layer, different behaviors training inference. layers, standard practice expose training (boolean) argument call method. exposing argument call, enable built-training evaluation loops (e.g. fit) correctly use layer training inference modes.","code":"Dropout <- new_layer_class(   \"Dropout\",   initialize = function(rate) {     super$initialize()     self$rate <- rate   },   call = function(inputs, training = NULL) {     if (!is.null(training) && training) {       return(tf$nn$dropout(inputs, rate = self$rate))     }     inputs   } )  MLPWithDropout <- new_layer_class(   \"MLPWithDropout\",   initialize = function() {     super$initialize()     self$linear_1 <- Linear(units = 32)     self$dropout <- Dropout(rate = 0.5)     self$linear_3 <- Linear(units = 10)   },   call = function(inputs, training = NULL) {     x <- self$linear_1(inputs)     x <- tf$nn$relu(x)     x <- self$dropout(x, training = training)     self$linear_3(x)   } )  mlp <- MLPWithDropout() y_train <- mlp(tf$ones(shape(2, 2)), training=TRUE) y_test <- mlp(tf$ones(shape(2, 2)), training=FALSE)"},{"path":"https://keras3.posit.co/dev/articles/intro_to_keras_for_researchers.html","id":"the-functional-api-for-model-building","dir":"Articles","previous_headings":"","what":"The Functional API for model-building","title":"Introduction to Keras for Researchers","text":"build deep learning models, don’t use object-oriented programming time. layers ’ve seen far can also composed functionally, like (call “Functional API”): Functional API tends concise subclassing, provides advantages (generally advantages functional, typed languages provide untyped OO development). However, can used define DAGs layers – recursive networks defined Layer subclasses instead. Learn Functional API . research workflows, may often find mix--matching OO models Functional models. Note Model class also features built-training & evaluation loops: fit(), predict() evaluate() (configured via compile() method). built-functions give access following built-training infrastructure features: Callbacks. can leverage built-callbacks early-stopping, model checkpointing, monitoring training TensorBoard. can also implement custom callbacks needed. Distributed training. can easily scale training multiple GPUs, TPU, even multiple machines tf.distribute API – changes code. Step fusing. steps_per_execution argument Model.compile(), can process multiple batches single tf.function call, greatly improves device utilization TPUs. won’t go details, provide simple code example . leverages built-training infrastructure implement MNIST example . can always subclass Model class (works exactly like subclassing Layer) want leverage built-training loops OO models. Just override Model$train_step() customize happens fit() retaining support built-infrastructure features outlined – callbacks, zero-code distribution support, step fusing support. may also override test_step() customize happens evaluate(), override predict_step() customize happens predict(). information, please refer guide.","code":"# We use an `Input` object to describe the shape and dtype of the inputs. # This is the deep learning equivalent of *declaring a type*. # The shape argument is per-sample; it does not include the batch size. # The functional API focused on defining per-sample transformations. # The model we create will automatically batch the per-sample transformations, # so that it can be called on batches of data. inputs <- layer_input(shape = 16, dtype = \"float32\")  # We call layers on these \"type\" objects # and they return updated types (new shapes/dtypes). outputs <- inputs %>%   Linear(units = 32) %>% # We are reusing the Linear layer we defined earlier.   Dropout(rate = 0.5) %>% # We are reusing the Dropout layer we defined earlier.   Linear(units = 10)  # A functional `Model` can be defined by specifying inputs and outputs. # A model is itself a layer like any other. model <- keras_model(inputs, outputs)  # A functional model already has weights, before being called on any data. # That's because we defined its input shape in advance (in `Input`). length(model$weights) ## [1] 4 # Let's call our model on some data, for fun. y <- model(tf$ones(shape(2, 16))) y$shape ## TensorShape([2, 10]) # You can pass a `training` argument in `__call__` # (it will get passed down to the Dropout layer). y <- model(tf$ones(shape(2, 16)), training=TRUE) inputs <- layer_input(shape = 784, dtype=\"float32\") outputs <- inputs %>%   layer_dense(units = 32, activation = \"relu\") %>%   layer_dense(units = 32, activation = \"relu\") %>%   layer_dense(units = 10) model <- keras_model(inputs, outputs)  # Specify the loss, optimizer, and metrics with `compile()`. model %>% compile(     loss = loss_sparse_categorical_crossentropy(from_logits=TRUE),     optimizer=optimizer_adam(learning_rate=1e-3),     metrics=list(metric_sparse_categorical_accuracy()), )  # Train the model with the dataset for 2 epochs. model %>% fit(dataset, epochs=2) ## Epoch 1/2 ## 938/938 - 4s - 4ms/step - loss: 0.3958 - sparse_categorical_accuracy: 0.8866 ## Epoch 2/2 ## 938/938 - 1s - 960us/step - loss: 0.1888 - sparse_categorical_accuracy: 0.9443 predictions <- model %>% predict(dataset) ## 938/938 - 1s - 1ms/step model %>% evaluate(dataset) ## 938/938 - 1s - 1ms/step - loss: 0.1763 - sparse_categorical_accuracy: 0.9454 ## $loss ## [1] 0.1763445 ## ## $sparse_categorical_accuracy ## [1] 0.9454167 CustomModel <- new_model_class(   \"CustomModel\",   initialize = function(...) {     super$initialize(...)     self$loss_tracker <- metric_mean(name=\"loss\")     self$accuracy <- metric_sparse_categorical_accuracy()     self$loss_fn <- loss_sparse_categorical_crossentropy(from_logits=TRUE)     self$optimizer <- optimizer_adam(learning_rate=1e-3)   },   train_step = function(data) {     c(x, y = NULL, sample_weight = NULL) %<-% data     with(tf$GradientTape() %as% tape, {       y_pred <- self(x, training=TRUE)       loss <- self$loss_fn(y = y, y_pred = y_pred, sample_weight=sample_weight)     })     gradients <- tape$gradient(loss, self$trainable_variables)     self$optimizer$apply_gradients(       zip_lists(gradients, self$trainable_variables)     )      # Update metrics (includes the metric that tracks the loss)     self$loss_tracker$update_state(loss)     self$accuracy$update_state(y, y_pred, sample_weight=sample_weight)     # Return a list mapping metric names to current value     list(       loss = self$loss_tracker$result(),       accuracy = self$accuracy$result()     )   },   metrics = mark_active(function() {     list(self$loss_tracker, self$accuracy)   }) )  inputs <- layer_input(shape = 784, dtype=\"float32\") outputs <- inputs %>%   layer_dense(units = 32, activation = \"relu\") %>%   layer_dense(units = 32, activation = \"relu\") %>%   layer_dense(units = 10) model <- CustomModel(inputs, outputs) model %>% compile() model %>% fit(dataset, epochs=2) ## Epoch 1/2 ## 938/938 - 2s - 2ms/step - loss: 0.3869 - sparse_categorical_accuracy: 0.8924 ## Epoch 2/2 ## 938/938 - 1s - 1ms/step - loss: 0.2163 - sparse_categorical_accuracy: 0.9370"},{"path":"https://keras3.posit.co/dev/articles/intro_to_keras_for_researchers.html","id":"end-to-end-experiment-example-1-variational-autoencoders-","dir":"Articles","previous_headings":"","what":"End-to-end experiment example 1: variational autoencoders.","title":"Introduction to Keras for Researchers","text":"things ’ve learned far: Layer encapsulates state (created __init__ build) computation (defined call). Layers can recursively nested create new, bigger computation blocks. can easily write highly hackable training loops opening GradientTape, calling model inside tape’s scope, retrieving gradients applying via optimizer. can speed training loops using @tf.function decorator. Layers can create track losses (typically regularization losses) via self.add_loss(). Let’s put things together end--end example: ’re going implement Variational AutoEncoder (VAE). ’ll train MNIST digits. VAE subclass Layer, built nested composition layers subclass Layer. feature regularization loss (KL divergence). model definition. First, Encoder class, uses Sampling layer map MNIST digit latent-space triplet (z_mean, z_log_var, z). Next, Decoder class, maps probabilistic latent space coordinates back MNIST digit. Finally, VariationalAutoEncoder composes together encoder decoder, creates KL divergence regularization loss via add_loss(). Now, let’s write training loop. training step decorated @tf.function compile super fast graph function. can see, building training type model Keras quick painless.","code":"Sampling <- new_layer_class(   \"Sampling\",   call = function(inputs) {     c(z_mean, z_log_var) %<-% inputs     batch <- op_shape(z_mean)[[1]]     dim <- op_shape(z_mean)[[2]]     epsilon <- random_normal(shape = c(batch, dim))     z_mean + op_exp(0.5 * z_log_var) * epsilon   } )  Encoder <- new_layer_class(   \"Encoder\",   initialize = function(latent_dim = 32, intermediate_dim = 64, ...) {     super$initialize(...)     self$dense_proj <- layer_dense(units = intermediate_dim, activation = \"relu\")     self$dense_mean <- layer_dense(units = latent_dim)     self$dense_log_var <- layer_dense(units = latent_dim)     self$sampling <- Sampling()   },   call = function(inputs) {     x <- self$dense_proj(inputs)     z_mean <- self$dense_mean(x)     z_log_var <- self$dense_log_var(x)     z <- self$sampling(list(z_mean, z_log_var))     list(z_mean, z_log_var, z)   } ) Decoder <- new_layer_class(   \"Decoder\",   initialize = function(original_dim, intermediate_dim = 64, ...) {     super$initialize(...)     self$dense_proj <- layer_dense(units = intermediate_dim, activation = \"relu\")     self$dense_output <- layer_dense(units = original_dim, activation = \"sigmoid\")   },   call = function(inputs) {     x <- self$dense_proj(inputs)     self$dense_output(x)   } ) VariationalAutoEncoder <- new_model_class(   \"VariationalAutoEncoder\",   initialize = function(original_dim,         intermediate_dim=64,         latent_dim=32,         name=\"autoencoder\", ...) {     super$initialize(name = name, ...)     self$original_dim <- original_dim     self$encoder <- Encoder(       latent_dim = latent_dim,       intermediate_dim = intermediate_dim     )     self$decoder <- Decoder(       original_dim = original_dim,       intermediate_dim = intermediate_dim     )   },   call = function(inputs) {     c(z_mean, z_log_var, z) %<-% self$encoder(inputs)     reconstructed <- self$decoder(z)     # Add KL divergence regularization loss.     kl_loss <- -0.5 * op_mean(       z_log_var - op_square(z_mean) - op_exp(z_log_var) + 1     )     self$add_loss(kl_loss)     reconstructed   } ) # Our model. vae <- VariationalAutoEncoder(   original_dim = 784,   intermediate_dim = 64,   latent_dim = 32 )  # Loss and optimizer. loss_fn <- loss_mean_squared_error() optimizer = optimizer_adam(learning_rate=1e-3)  # Prepare a dataset. c(c(x_train, .), .) %<-% dataset_mnist() x_train <- array_reshape(x_train, c(60000, 784)) / 255  dataset <- tfdatasets::tensor_slices_dataset(x_train) %>%   tfdatasets::dataset_shuffle(buffer_size=1024) %>%   tfdatasets::dataset_batch(32)   training_step <- tf_function(function(x) {   with(tf$GradientTape() %as% tape, {     reconstructed <- vae(x)  # Compute input reconstruction.     # Compute loss.     loss <- loss_fn(x, reconstructed)     loss <- loss + op_sum(vae$losses)  # Add KLD term.   })   # Update the weights of the VAE.   grads <- tape$gradient(loss, vae$trainable_weights)   optimizer$apply_gradients(zip_lists(grads, vae$trainable_weights))   loss })  losses <- c()  # Keep track of the losses over time. coro::loop(for(data in dataset) {   loss <- training_step(data)    # Logging.   losses[length(losses) + 1] <- as.numeric(loss)   if (length(losses) %% 100 == 0) {     cat(\"Step:\", length(losses), \"Loss:\", mean(losses), \"\\n\")   }   # Stop after 1000 steps.   # Training the model to convergence is left   # as an exercise to the reader.   if (length(losses) >= 1000) {     break   } }) ## Step: 100 Loss: 0.1270978 ## Step: 200 Loss: 0.1003238 ## Step: 300 Loss: 0.09001128 ## Step: 400 Loss: 0.08493649 ## Step: 500 Loss: 0.08171404 ## Step: 600 Loss: 0.07926706 ## Step: 700 Loss: 0.07790599 ## Step: 800 Loss: 0.07670419 ## Step: 900 Loss: 0.07570736 ## Step: 1000 Loss: 0.07476593"},{"path":"https://keras3.posit.co/dev/articles/intro_to_keras_for_researchers.html","id":"end-to-end-experiment-example-2-hypernetworks-","dir":"Articles","previous_headings":"","what":"End-to-end experiment example 2: hypernetworks.","title":"Introduction to Keras for Researchers","text":"Let’s take look another kind research experiment: hypernetworks. idea use small deep neural network (hypernetwork) generate weights larger network (main network). Let’s implement really trivial hypernetwork: ’ll use small 2-layer network generate weights larger 3-layer network. training loop. batch data: use hypernetwork generate array weight coefficients, weights_pred reshape coefficients kernel & bias tensors main_network run forward pass main_network compute actual MNIST predictions run backprop weights hypernetwork minimize final classification loss Implementing arbitrary research ideas Keras straightforward highly productive. Imagine trying 25 ideas per day (20 minutes per experiment average)! Keras designed go idea results fast possible, believe key great research. hope enjoyed quick introduction. Let us know build Keras!","code":"input_dim <- 784 classes <- 10  # This is the main network we'll actually use to predict labels. inputs <- layer_input(shape = input_dim) dense1 <- layer_dense(units = 64, activation = \"relu\") dense1$built <- TRUE  dense2 <- layer_dense(units = classes) dense2$built <- TRUE  outputs <- inputs %>% dense1() %>% dense2() main_network <- keras_model(inputs, outputs)  # This is the number of weight coefficients to generate. Each layer in the # main network requires output_dim * input_dim + output_dim coefficients. num_weights_to_generate <- (classes * 64 + classes) + (64 * input_dim + 64)  # This is the hypernetwork that generates the weights of the `main_network` above. hypernetwork <- keras_model_sequential() %>%   layer_dense(units=16, activation=\"relu\") %>%   layer_dense(units=num_weights_to_generate, activation=\"sigmoid\") # Loss and optimizer. loss_fn <- loss_sparse_categorical_crossentropy(from_logits = TRUE) optimizer <- optimizer_adam(learning_rate=1e-4)  # Prepare a dataset. c(c(x_train, y_train), .) %<-% dataset_mnist() x_train <- array_reshape(x_train, c(60000, 784)) / 255  dataset <- tfdatasets::tensor_slices_dataset(list(x_train, y_train)) %>%   tfdatasets::dataset_shuffle(buffer_size=1024) %>%   # We'll use a batch size of 1 for this experiment.   tfdatasets::dataset_batch(1)  train_step <- function(x, y) {   with(tf$GradientTape() %as% tape, {     weights_pred <- hypernetwork(x)      # Reshape them to the expected shapes for w and b for the outer model.     # Layer 1 kernel.     start_index <- 1     w1_shape <- c(input_dim, 64)     w1_coeffs <- weights_pred[, start_index:(start_index + prod(w1_shape) - 1)]     w1 <- tf$reshape(w1_coeffs, as.integer(w1_shape))     start_index <- start_index + prod(w1_shape)      # Layer 1 bias.     b1_shape <- c(64)     b1_coeffs <- weights_pred[, start_index:(start_index + prod(b1_shape) - 1)]     b1 <- tf$reshape(b1_coeffs, as.integer(b1_shape))     start_index <- start_index + prod(b1_shape)      # Layer 2 kernel.     w2_shape <- c(64, classes)     w2_coeffs <- weights_pred[, start_index:(start_index + prod(w2_shape) - 1)]     w2 <- tf$reshape(w2_coeffs, as.integer(w2_shape))     start_index <- start_index + prod(w2_shape)      # Layer 2 bias.     b2_shape <- c(classes)     b2_coeffs <- weights_pred[, start_index:(start_index + prod(b2_shape) - 1)]     b2 <- tf$reshape(b2_coeffs, as.integer(b2_shape))     start_index <- start_index + prod(b2_shape)      # Set the weight predictions as the weight variables on the outer model.     dense1$kernel <- w1     dense1$bias <- b1     dense2$kernel <- w2     dense2$bias <- b2      # Inference on the outer model.     preds <- main_network(x)     loss <- loss_fn(y, preds)   })    grads <- tape$gradient(loss, hypernetwork$trainable_weights)   optimizer$apply_gradients(zip_lists(grads, hypernetwork$trainable_weights))   loss }  losses <- c()  # Keep track of the losses over time. coro::loop(for (data in dataset) {   x <- data[[1]]   y <- data[[2]]   loss <- train_step(x, y)    # Logging.   losses[length(losses) + 1] <- as.numeric(loss)   if (length(losses) %% 100 == 0) {     cat(\"Step:\", length(losses), \"Loss:\", mean(losses), \"\\n\")   }   # Stop after 1000 steps.   # Training the model to convergence is left   # as an exercise to the reader.   if (length(losses) >= 1000) {     break   } }) ## Step: 100 Loss: 2.536778 ## Step: 200 Loss: 2.236472 ## Step: 300 Loss: 2.119417 ## Step: 400 Loss: 2.040341 ## Step: 500 Loss: 1.949125 ## Step: 600 Loss: 1.859384 ## Step: 700 Loss: 1.845726 ## Step: 800 Loss: 1.820594 ## Step: 900 Loss: 1.771334 ## Step: 1000 Loss: 1.730648"},{"path":"https://keras3.posit.co/dev/articles/making_new_layers_and_models_via_subclassing.html","id":"introduction","dir":"Articles","previous_headings":"","what":"Introduction","title":"Making new layers and models via subclassing","text":"guide cover everything need know build subclassed layers models. particular, ’ll learn following features: Layer class add_weight() method Trainable non-trainable weights build() method Making sure layers can used backend add_loss() method training argument call() mask argument call() Making sure layers can serialized Let’s dive .","code":""},{"path":"https://keras3.posit.co/dev/articles/making_new_layers_and_models_via_subclassing.html","id":"setup","dir":"Articles","previous_headings":"","what":"Setup","title":"Making new layers and models via subclassing","text":"","code":"library(keras3) library(tensorflow, exclude = c(\"set_random_seed\", \"shape\")) library(tfdatasets, exclude = \"shape\")"},{"path":"https://keras3.posit.co/dev/articles/making_new_layers_and_models_via_subclassing.html","id":"the-layer-class-the-combination-of-state-weights-and-some-computation","dir":"Articles","previous_headings":"","what":"The Layer class: the combination of state (weights) and some computation","title":"Making new layers and models via subclassing","text":"One central abstractions Keras Layer class. layer encapsulates state (layer’s “weights”) transformation inputs outputs (“call”, layer’s forward pass). ’s densely-connected layer. two state variables: variables w b. use layer calling tensor input(s), much like R function. Note weights w b automatically tracked layer upon set layer attributes:","code":"layer_linear <- Layer(\"Linear\",    initialize = function(units = 32, input_dim = 32, ...) {     super$initialize(...)     self$w <- self$add_weight(       shape = shape(input_dim, units),       initializer = \"random_normal\",       trainable = TRUE     )     self$b <- self$add_weight(       shape = shape(units),       initializer = \"zeros\",       trainable = TRUE     )   },    call = function(inputs) {     op_matmul(inputs, self$w) + self$b   } ) x <- op_ones(c(2, 2)) linear_layer <- layer_linear(units = 4, input_dim = 2) y <- linear_layer(x) print(y) ## tf.Tensor( ## [[0.02153057 0.15450525 0.0205495  0.04493225] ##  [0.02153057 0.15450525 0.0205495  0.04493225]], shape=(2, 4), dtype=float32) linear_layer$weights ## [[1]] ## <Variable path=linear/variable, shape=(2, 4), dtype=float32, value=[[-0.06251299  0.05335509  0.01485647 -0.00985784] ##  [ 0.08404355  0.10115016  0.00569303  0.05479009]]> ## ## [[2]] ## <Variable path=linear/variable_1, shape=(4), dtype=float32, value=[0. 0. 0. 0.]>"},{"path":"https://keras3.posit.co/dev/articles/making_new_layers_and_models_via_subclassing.html","id":"layers-can-have-non-trainable-weights","dir":"Articles","previous_headings":"","what":"Layers can have non-trainable weights","title":"Making new layers and models via subclassing","text":"Besides trainable weights, can add non-trainable weights layer well. weights meant taken account backpropagation, training layer. ’s add use non-trainable weight: ’s part layer$weights, gets categorized non-trainable weight:","code":"layer_compute_sum <- Layer(   \"ComputeSum\",   initialize = function(input_dim) {     super$initialize()     self$total <- self$add_weight(       initializer = \"zeros\",       shape = shape(input_dim),       trainable = FALSE     )   },   call = function(inputs) {     self$total$assign_add(op_sum(inputs, axis = 1))     self$total   } )  x <- op_ones(c(2, 2)) my_sum <- layer_compute_sum(input_dim = 2) y <- my_sum(x) print(as.array(y)) ## [1] 2 2 y <- my_sum(x) print(as.array(y)) ## [1] 4 4 cat(\"weights:\", length(my_sum$weights)) ## weights: 1 cat(\"non-trainable weights:\", length(my_sum$non_trainable_weights)) ## non-trainable weights: 1 # It's not included in the trainable weights: cat(\"trainable_weights:\", length(my_sum$trainable_weights)) ## trainable_weights: 0"},{"path":"https://keras3.posit.co/dev/articles/making_new_layers_and_models_via_subclassing.html","id":"best-practice-deferring-weight-creation-until-the-shape-of-the-inputs-is-known","dir":"Articles","previous_headings":"","what":"Best practice: deferring weight creation until the shape of the inputs is known","title":"Making new layers and models via subclassing","text":"Linear layer took input_dim argument used compute shape weights w b initialize(): many cases, may know advance size inputs, like lazily create weights value becomes known, time instantiating layer. Keras API, recommend creating layer weights build(self, inputs_shape) method layer. Like : call() method layer automatically run build first time called. now layer ’s lazy thus easier use: Implementing build() separately shown nicely separates creating weights using weights every call.","code":"layer_linear <- Layer(\"Linear\",    initialize = function(units = 32, input_dim = 32, ...) {     super$initialize(...)     self$w <- self$add_weight(       shape = shape(input_dim, units),       initializer = \"random_normal\",       trainable = TRUE     )     self$b <- self$add_weight(       shape = shape(units),       initializer = \"zeros\",       trainable = TRUE     )   },    call = function(inputs) {     op_matmul(inputs, self$w) + self$b   } ) layer_linear <- Layer(   \"Linear\",   initialize = function(units = 32, ...) {     self$units <- as.integer(units)     super$initialize(...)   },   build = function(input_shape) {     self$w <- self$add_weight(       shape = shape(tail(input_shape, 1), self$units),       initializer = \"random_normal\",       trainable = TRUE     )     self$b <- self$add_weight(       shape = shape(self$units),       initializer = \"zeros\",       trainable = TRUE     )   },   call = function(inputs) {     op_matmul(inputs, self$w) + self$b   } ) # At instantiation, we don't know on what inputs this is going to get called linear_layer <- layer_linear(units = 32)  # The layer's weights are created dynamically the first time the layer is called y <- linear_layer(x)"},{"path":"https://keras3.posit.co/dev/articles/making_new_layers_and_models_via_subclassing.html","id":"layers-are-recursively-composable","dir":"Articles","previous_headings":"","what":"Layers are recursively composable","title":"Making new layers and models via subclassing","text":"assign Layer instance attribute another Layer, outer layer start tracking weights created inner layer. recommend creating sublayers initialize() method leave first call() trigger building weights.","code":"MLPBlock <- Layer(   \"MLPBlock\",   initialize = function() {     super$initialize()     self$linear_1 <- layer_linear(units = 32)     self$linear_2 <- layer_linear(units = 32)     self$linear_3 <- layer_linear(units = 1)   },   call = function(inputs) {     inputs |>       self$linear_1() |>       activation_relu() |>       self$linear_2() |>       activation_relu() |>       self$linear_3()   } )  mlp <- MLPBlock() # The first call to the `mlp` will create the weights y <- mlp(op_ones(shape = c(3, 64)))  cat(\"weights:\", length(mlp$weights), \"\\n\") ## weights: 6 cat(\"trainable weights:\", length(mlp$trainable_weights), \"\\n\") ## trainable weights: 6"},{"path":"https://keras3.posit.co/dev/articles/making_new_layers_and_models_via_subclassing.html","id":"backend-agnostic-layers-and-backend-specific-layers","dir":"Articles","previous_headings":"","what":"Backend-agnostic layers and backend-specific layers","title":"Making new layers and models via subclassing","text":"long layer uses APIs ops namespace (ie. using functions starting op_), (Keras namespaces activations_*, random_*, layer_*), can used backend – TensorFlow, JAX, PyTorch. layers ’ve seen far guide work Keras backends. ops namespace gives access : NumPy API, e.g. op_matmul, op_sum, op_reshape, op_stack, etc. Neural networks-specific APIs op_softmax, op_conv, op_binary_crossentropy, op_relu, etc. can also use backend-native APIs layers (tf$nn functions), , layer usable backend question. instance, write following JAX-specific layer using jax$numpy: equivalent TensorFlow-specific layer: equivalent PyTorch-specific layer: cross-backend compatibility tremendously useful property, strongly recommend seek always make layers backend-agnostic leveraging Keras APIs.","code":"# keras3::install_keras(backend = c(\"jax\")) jax <- reticulate::import(\"jax\")  Linear <- new_layer_class(   ...   call = function(inputs) {     jax$numpy$matmul(inputs, self$w) + self$b   } ) library(tensorflow)  Linear <- new_layer_class(   ...   call = function(inputs) {     tf$matmul(inputs, self$w) + self$b   } ) torch <- reticulate::import(\"torch\")  Linear <- new_layer_class(   ...   call = function(inputs) {     torch$matmul(inputs, self$w) + self$b   } )"},{"path":"https://keras3.posit.co/dev/articles/making_new_layers_and_models_via_subclassing.html","id":"the-add_loss-method","dir":"Articles","previous_headings":"","what":"The add_loss() method","title":"Making new layers and models via subclassing","text":"writing call() method layer, can create loss tensors want use later, writing training loop. doable calling self$add_loss(value): losses (including created inner layer) can retrieved via layer$losses. property reset start every call top-level layer, layer$losses always contains loss values created last forward pass. addition, loss property also contains regularization losses created weights inner layer: losses meant taken account writing custom training loops. also work seamlessly fit() (get automatically summed added main loss, ):","code":"# A layer that creates an activity regularization loss layer_activity_regularization <- Layer(   \"ActivityRegularizationLayer\",   initialize = function(rate = 1e-2) {     self$rate <- as.numeric(rate)     super$initialize()   },   call = function(inputs) {     self$add_loss(self$rate * op_mean(inputs))     inputs   } ) layer_outer <- Layer(   \"OuterLayer\",   initialize = function() {     super$initialize()     self$activity_reg <- layer_activity_regularization(rate = 1e-2)   },   call = function(inputs) {     self$activity_reg(inputs)     inputs   } )  layer <- layer_outer() # No losses yet since the layer has never been called cat(\"losses:\", length(layer$losses), \"\\n\") ## losses: 0 x <- layer(op_zeros(c(1, 1))) # We created one loss value cat(\"losses:\", length(layer$losses), \"\\n\") ## losses: 1 # `layer$losses` gets reset at the start of each call x <- layer(op_zeros(c(1, 1))) # This is the loss created during the call above cat(\"losses:\", length(layer$losses), \"\\n\") ## losses: 1 layer_outer_with_kernel_regularizer <- Layer(   \"OuterLayerWithKernelRegularizer\",   initialize = function() {     super$initialize()     self$dense <- layer_dense(units = 32,                               kernel_regularizer = regularizer_l2(1e-3))   },   call = function(inputs) {     self$dense(inputs)   } )  layer <- layer_outer_with_kernel_regularizer() x <- layer(op_zeros(c(1, 1)))  # This is `1e-3 * sum(layer$dense$kernel ** 2)`, # created by the `kernel_regularizer` above. print(layer$losses) ## [[1]] ## tf.Tensor(0.002025157, shape=(), dtype=float32) inputs <- keras_input(shape = 3) outputs <- inputs |> layer_activity_regularization() model <- keras_model(inputs, outputs)  # If there is a loss passed in `compile`, the regularization # losses get added to it model |> compile(optimizer = \"adam\", loss = \"mse\") model |> fit(random_normal(c(2, 3)), random_normal(c(2, 3)), epochs = 1) ## 1/1 - 0s - 161ms/step - loss: 1.9081 # It's also possible not to pass any loss in `compile`, # since the model already has a loss to minimize, via the `add_loss` # call during the forward pass! model |> compile(optimizer = \"adam\") model |> fit(random_normal(c(2, 3)), random_normal(c(2, 3)), epochs = 1) ## 1/1 - 0s - 139ms/step - loss: 1.6613"},{"path":"https://keras3.posit.co/dev/articles/making_new_layers_and_models_via_subclassing.html","id":"you-can-optionally-enable-serialization-on-your-layers","dir":"Articles","previous_headings":"","what":"You can optionally enable serialization on your layers","title":"Making new layers and models via subclassing","text":"need custom layers serializable part Functional model, can optionally implement get_config() method: Note initialize() method base Layer class takes keyword arguments, particular name dtype. ’s good practice pass arguments parent class initialize() include layer config: need flexibility deserializing layer config, can also override from_config() class method. base implementation from_config(): learn serialization saving, see complete guide saving serializing models.","code":"layer_linear <- Layer(   \"Linear\",   initialize = function(units = 32) {     self$units <- as.integer(units)     super$initialize()   },   build = function(input_shape) {     self$w <- self$add_weight(       shape = shape(tail(input_shape, 1), self$units),       initializer = \"random_normal\",       trainable = TRUE     )     self$b <- self$add_weight(       shape = shape(self$units),       initializer = \"zeros\",       trainable = TRUE     )   },   call = function(inputs) {     op_matmul(inputs, self$w) + self$b   },   get_config = function() {     list(units = self$units)   } )  # Now you can recreate the layer from its config: layer <- layer_linear(units = 64) config <- get_config(layer) str(config) ## List of 1 ##  $ units: int 64 ##  - attr(*, \"__class__\")=<class '<r-globalenv>.Linear'> new_layer <- from_config(config) Linear <- new_layer_class(   \"Linear\",   initialize = function(units = 32, ...) {     self$units <- as.integer(units)     super$initialize(...)   },   build = function(input_shape) {     self$w <- self$add_weight(       shape = shape(tail(input_shape, 1), self$units),       initializer = \"random_normal\",       trainable = TRUE     )     self$b <- self$add_weight(       shape = shape(self$units),       initializer = \"zeros\",       trainable = TRUE     )   },   call = function(inputs) {     op_matmul(inputs, self$w) + self$b   },   get_config = function() {     list(units = self$units)   } )  layer <- Linear(units = 64) config <- get_config(layer) str(config) ## List of 1 ##  $ units: int 64 ##  - attr(*, \"__class__\")=<class '<r-globalenv>.Linear'> new_layer <- from_config(config) Layer(   ...,   from_config = function(config) {     # calling `__class__`() creates a new instance and calls initialize()     do.call(`__class__`, config)   } )"},{"path":"https://keras3.posit.co/dev/articles/making_new_layers_and_models_via_subclassing.html","id":"privileged-training-argument-in-the-call-method","dir":"Articles","previous_headings":"","what":"Privileged training argument in the call() method","title":"Making new layers and models via subclassing","text":"layers, particular BatchNormalization layer Dropout layer, different behaviors training inference. layers, standard practice expose training (boolean) argument call() method. exposing argument call(), enable built-training evaluation loops (e.g. fit()) correctly use layer training inference.","code":"layer_custom_dropout <- Layer(   \"CustomDropout\",   initialize = function(rate, ...) {     super$initialize(...)     self$rate <- rate     self$seed_generator <- random_seed_generator(1337)   },   call = function(inputs, training = NULL) {     if (isTRUE(training))       return(random_dropout(inputs, rate = self$rate,                             seed = self.seed_generator))     inputs   } )"},{"path":"https://keras3.posit.co/dev/articles/making_new_layers_and_models_via_subclassing.html","id":"privileged-mask-argument-in-the-call-method","dir":"Articles","previous_headings":"","what":"Privileged mask argument in the call() method","title":"Making new layers and models via subclassing","text":"privileged argument supported call() mask argument. find Keras RNN layers. mask boolean tensor (one boolean value per timestep input) used skip certain input timesteps processing timeseries data. Keras automatically pass correct mask argument call() layers support , mask generated prior layer. Mask-generating layers Embedding layer configured mask_zero = TRUE, Masking layer.","code":""},{"path":"https://keras3.posit.co/dev/articles/making_new_layers_and_models_via_subclassing.html","id":"the-model-class","dir":"Articles","previous_headings":"","what":"The Model class","title":"Making new layers and models via subclassing","text":"general, use Layer class define inner computation blocks, use Model class define outer model – object train. instance, ResNet50 model, several ResNet blocks subclassing Layer, single Model encompassing entire ResNet50 network. Model class API Layer, following differences: exposes built-training, evaluation, prediction loops (fit(), evaluate(), predict()). exposes list inner layers, via model$layers property. exposes saving serialization APIs (save(), save_weights()…) Effectively, Layer class corresponds refer literature “layer” (“convolution layer” “recurrent layer”) “block” (“ResNet block” “Inception block”). Meanwhile, Model class corresponds referred literature “model” (“deep learning model”) “network” (“deep neural network”). ’re wondering, “use Layer class Model class?”, ask : need call fit() ? need call save() ? , go Model. (either class just block bigger system, writing training & saving code ), use Layer. instance, take mini-resnet example , use build Model train fit(), save save_weights():","code":"ResNet <- Model(   \"ResNet\",   initialize = function(num_classes = 1000, ...) {     super$initialize(...)     self$block_1 <- layer_resnet_block()     self$block_2 <- layer_resnet_block()     self$global_pool <- layer_global_average_pooling_2d()     self$classifier <- layer_dense(num_classes)   },   call = function(inputs) {     inputs |>       self$block_1() |>       self$block_2() |>       self$global_pool() |>       self$classifier()   } )  resnet <- ResNet() dataset <- ... resnet |> fit(dataset, epochs=10) resnet |> save_model(\"filepath.keras\")"},{"path":"https://keras3.posit.co/dev/articles/making_new_layers_and_models_via_subclassing.html","id":"putting-it-all-together-an-end-to-end-example","dir":"Articles","previous_headings":"","what":"Putting it all together: an end-to-end example","title":"Making new layers and models via subclassing","text":"’s ’ve learned far: Layer encapsulate state (created initialize() build()) computation (defined call()). Layers can recursively nested create new, bigger computation blocks. Layers backend-agnostic long use Keras APIs. can use backend-native APIs (jax$numpy, torch$nn tf$nn), layer usable specific backend. Layers can create track losses (typically regularization losses) via add_loss(). outer container, thing want train, Model. Model just like Layer, added training serialization utilities. Let’s put things together end--end example: ’re going implement Variational AutoEncoder (VAE) backend-agnostic fashion – runs TensorFlow, JAX, PyTorch. ’ll train MNIST digits. VAE subclass Model, built nested composition layers subclass Layer. feature regularization loss (KL divergence). Let’s train MNIST using fit() API:","code":"layer_sampling <- Layer(   \"Sampling\",   initialize = function(...) {     super$initialize(...)     self$seed_generator <- random_seed_generator(1337)   },   call = function(inputs) {     c(z_mean, z_log_var) %<-% inputs     batch <- op_shape(z_mean)[[1]]     dim <- op_shape(z_mean)[[2]]     epsilon <- random_normal(shape = c(batch, dim),                              seed=self$seed_generator)     z_mean + op_exp(0.5 * z_log_var) * epsilon   } )  # Maps MNIST digits to a triplet (z_mean, z_log_var, z). layer_encoder <- Layer(   \"Encoder\",   initialize = function(latent_dim = 32, intermediate_dim = 64, ...) {     super$initialize(...)     self$dense_proj <-       layer_dense(units = intermediate_dim,  activation = \"relu\")     self$dense_mean <- layer_dense(units = latent_dim)     self$dense_log_var <- layer_dense(units = latent_dim)     self$sampling <- layer_sampling()   },   call = function(inputs) {     x <- self$dense_proj(inputs)     z_mean <- self$dense_mean(x)     z_log_var <- self$dense_log_var(x)     z <- self$sampling(list(z_mean, z_log_var))     list(z_mean, z_log_var, z)   } )  # Converts z, the encoded digit vector, back into a readable digit. layer_decoder <- Layer(   \"Decoder\",   initialize = function(original_dim, intermediate_dim = 64, ...) {     super$initialize(...)     self$dense_proj <-       layer_dense(units = intermediate_dim, activation = \"relu\")     self$dense_output <-       layer_dense(units = original_dim, activation = \"sigmoid\")   },   call = function(inputs) {     x <- self$dense_proj(inputs)     self$dense_output(x)   } )  # Combines the encoder and decoder into an end-to-end model for training. VariationalAutoEncoder <- Model(   \"VariationalAutoEncoder\",    initialize = function(original_dim, intermediate_dim = 64, latent_dim = 32,                         name = \"autoencoder\", ...) {     super$initialize(name = name, ...)     self$original_dim <- original_dim     self$encoder <- layer_encoder(latent_dim = latent_dim,                             intermediate_dim = intermediate_dim)     self$decoder <- layer_decoder(original_dim = original_dim,                             intermediate_dim = intermediate_dim)   },    call = function(inputs) {     c(z_mean, z_log_var, z) %<-% self$encoder(inputs)     reconstructed <- self$decoder(z)     # Add KL divergence regularization loss.     kl_loss <- -0.5 * op_mean(z_log_var - op_square(z_mean) - op_exp(z_log_var) + 1)     self$add_loss(kl_loss)     reconstructed   } ) c(c(x_train, .), .) %<-% dataset_mnist() x_train <- x_train |>   op_reshape(c(60000, 784)) |>   op_cast(\"float32\") |>   op_divide(255)  original_dim <- 784 vae <- VariationalAutoEncoder(   original_dim = 784,   intermediate_dim = 64,   latent_dim = 32 )  optimizer <- optimizer_adam(learning_rate = 1e-3) vae |> compile(optimizer, loss = loss_mean_squared_error())  vae |> fit(x_train, x_train, epochs = 2, batch_size = 64) ## Epoch 1/2 ## 938/938 - 4s - 5ms/step - loss: 0.0748 ## Epoch 2/2 ## 938/938 - 1s - 1ms/step - loss: 0.0676"},{"path":"https://keras3.posit.co/dev/articles/sequential_model.html","id":"setup","dir":"Articles","previous_headings":"","what":"Setup","title":"The Sequential model","text":"","code":"library(keras3)"},{"path":"https://keras3.posit.co/dev/articles/sequential_model.html","id":"when-to-use-a-sequential-model","dir":"Articles","previous_headings":"","what":"When to use a Sequential model","title":"The Sequential model","text":"Sequential model appropriate plain stack layers layer exactly one input tensor one output tensor. Schematically, following Sequential model: equivalent function: Sequential model appropriate : model multiple inputs multiple outputs layers multiple inputs multiple outputs need layer sharing want non-linear topology (e.g. residual connection, multi-branch model)","code":"model <- keras_model_sequential() |>   layer_dense(units = 2, activation = \"relu\", name = \"layer1\") |>   layer_dense(units = 3, activation = \"relu\", name = \"layer2\") |>   layer_dense(units = 4, name = \"layer3\")  # Call model on a test input x <- op_ones(c(3, 3)) y <- model(x) # Create 3 layers layer1 <- layer_dense(units = 2, activation=\"relu\", name=\"layer1\") layer2 <- layer_dense(units = 3, activation=\"relu\", name=\"layer2\") layer3 <- layer_dense(units = 4, name=\"layer3\")  # Call layers on a test input x <- op_ones(c(3, 3)) y <- x |> layer1() |> layer2() |> layer3()"},{"path":"https://keras3.posit.co/dev/articles/sequential_model.html","id":"creating-a-sequential-model","dir":"Articles","previous_headings":"","what":"Creating a Sequential model","title":"The Sequential model","text":"can create Sequential model piping layers keras_model_sequential() object: passing list layers keras_model_sequential(): layers accessible via layers attribute: can also create Sequential model incrementally: Note ’s also corresponding pop_layer() method remove layers: Sequential model behaves much like stack layers. Also note Sequential constructor accepts name argument, just like layer model Keras. useful annotate TensorBoard graphs semantically meaningful names.","code":"model <- keras_model_sequential() |>   layer_dense(units = 2, activation = \"relu\") |>   layer_dense(units = 3, activation = \"relu\") |>   layer_dense(units = 4) model <- keras_model_sequential(layers = list(   layer_dense(units = 2, activation = \"relu\"),   layer_dense(units = 3, activation = \"relu\"),   layer_dense(units = 4) )) model$layers ## [[1]] ## <Dense name=dense_3, built=False> ##  signature: (*args, **kwargs) ## ## [[2]] ## <Dense name=dense_4, built=False> ##  signature: (*args, **kwargs) ## ## [[3]] ## <Dense name=dense_5, built=False> ##  signature: (*args, **kwargs) model <- keras_model_sequential() model |> layer_dense(units = 2, activation=\"relu\") model |> layer_dense(units = 3, activation=\"relu\") model |> layer_dense(units = 4) model |> pop_layer() ## <Dense name=dense_8, built=False> ##  signature: (*args, **kwargs) length(model$layers)  # 2 ## [1] 2 model <- keras_model_sequential(name = \"my_sequential\") model |> layer_dense(units = 2, activation=\"relu\", name = \"layer1\") model |> layer_dense(units = 3, activation=\"relu\", name = \"layer2\") model |> layer_dense(units = 4, name = \"layer3\")"},{"path":"https://keras3.posit.co/dev/articles/sequential_model.html","id":"specifying-the-input-shape-in-advance","dir":"Articles","previous_headings":"","what":"Specifying the input shape in advance","title":"The Sequential model","text":"Generally, layers Keras need know shape inputs order able create weights. create layer like , initially, weights: creates weights first time called input, since shape weights depends shape inputs: Naturally, also applies Sequential models. instantiate Sequential model without input shape, isn’t “built”: weights (calling model$weights results error stating just ). weights created model first sees input data: model “built”, can call summary() method display contents: However, can useful building Sequential model incrementally able display summary model far, including current output shape. case, start model passing input_shape argument model, knows input shape start: Models built predefined input shape like always weights (even seeing data) always defined output shape. general, ’s recommended best practice always specify input shape Sequential model advance know .","code":"layer <- layer_dense(units = 3) layer$weights  # Empty ## list() # Call layer on a test input x <- op_ones(c(1, 4)) y <- layer(x) layer$weights  # Now it has weights, of shape (4, 3) and (3,) ## [[1]] ## <Variable path=dense_9/kernel, shape=(4, 3), dtype=float32, value=[[ 0.48581433  0.78749573  0.61015   ] ##  [ 0.7962619   0.7261175  -0.8046875 ] ##  [-0.6189915   0.37973273  0.50559556] ##  [-0.5455791  -0.60714126  0.19791973]]> ## ## [[2]] ## <Variable path=dense_9/bias, shape=(3), dtype=float32, value=[0. 0. 0.]> model <- keras_model_sequential() |>   layer_dense(units = 2, activation = \"relu\") |>   layer_dense(units = 3, activation = \"relu\") |>   layer_dense(units = 4) # No weights at this stage!  # At this point, you can't do this: # model$weights   # Call the model on a test input x <- op_ones(c(1, 4)) y <- model(x) length(model$weights) ## [1] 6 summary(model) ## Model: \"sequential_4\" ## ┏━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━┓ ## ┃ Layer (type)                    ┃ Output Shape           ┃       Param # ┃ ## ┡━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━┩ ## │ dense_10 (Dense)                │ (1, 2)                 │            10 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ dense_11 (Dense)                │ (1, 3)                 │             9 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ dense_12 (Dense)                │ (1, 4)                 │            16 │ ## └─────────────────────────────────┴────────────────────────┴───────────────┘ ##  Total params: 35 (140.00 B) ##  Trainable params: 35 (140.00 B) ##  Non-trainable params: 0 (0.00 B) model <- keras_model_sequential(input_shape = 4) |>   layer_dense(units = 2, activation = \"relu\") summary(model) ## Model: \"sequential_5\" ## ┏━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━┓ ## ┃ Layer (type)                    ┃ Output Shape           ┃       Param # ┃ ## ┡━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━┩ ## │ dense_13 (Dense)                │ (None, 2)              │            10 │ ## └─────────────────────────────────┴────────────────────────┴───────────────┘ ##  Total params: 10 (40.00 B) ##  Trainable params: 10 (40.00 B) ##  Non-trainable params: 0 (0.00 B) model$layers ## [[1]] ## <Dense name=dense_13, built=True> ##  signature: (*args, **kwargs)"},{"path":"https://keras3.posit.co/dev/articles/sequential_model.html","id":"a-common-debugging-workflow-add-layers-summary","dir":"Articles","previous_headings":"","what":"A common debugging workflow: add layers + summary()","title":"The Sequential model","text":"building new Sequential architecture, ’s useful incrementally stack layers |> frequently print model summaries. instance, enables monitor stack Conv2D MaxPooling2D layers downsampling image feature maps: practical, right? Note |> equivalent calling model$add(), modifies model -place, don’t need reassign model symbol step.","code":"model <- keras_model_sequential(input_shape = c(250, 250, 3)) |>   layer_conv_2d(filters = 32, kernel_size = 5, strides = 2, activation = \"relu\") |>   layer_conv_2d(filters = 32, kernel_size = 3, activation = \"relu\") |>   layer_max_pooling_2d(pool_size = c(3, 3))  # Can you guess what the current output shape is at this point? Probably not. # Let's just print it: summary(model) ## Model: \"sequential_6\" ## ┏━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━┓ ## ┃ Layer (type)                    ┃ Output Shape           ┃       Param # ┃ ## ┡━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━┩ ## │ conv2d (Conv2D)                 │ (None, 123, 123, 32)   │         2,432 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ conv2d_1 (Conv2D)               │ (None, 121, 121, 32)   │         9,248 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ max_pooling2d (MaxPooling2D)    │ (None, 40, 40, 32)     │             0 │ ## └─────────────────────────────────┴────────────────────────┴───────────────┘ ##  Total params: 11,680 (45.62 KB) ##  Trainable params: 11,680 (45.62 KB) ##  Non-trainable params: 0 (0.00 B) # The answer was: (40, 40, 32), so we can keep downsampling...  model |>   layer_conv_2d(filters = 32, kernel_size = 3, activation = \"relu\") |>   layer_conv_2d(filters = 32, kernel_size = 3, activation = \"relu\") |>   layer_max_pooling_2d(pool_size = 3) |>   layer_conv_2d(filters = 32, kernel_size = 3, activation = \"relu\") |>   layer_conv_2d(filters = 32, kernel_size = 3, activation = \"relu\") |>   layer_max_pooling_2d(pool_size = 2)  # And now? summary(model) ## Model: \"sequential_6\" ## ┏━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━┓ ## ┃ Layer (type)                    ┃ Output Shape           ┃       Param # ┃ ## ┡━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━┩ ## │ conv2d (Conv2D)                 │ (None, 123, 123, 32)   │         2,432 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ conv2d_1 (Conv2D)               │ (None, 121, 121, 32)   │         9,248 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ max_pooling2d (MaxPooling2D)    │ (None, 40, 40, 32)     │             0 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ conv2d_2 (Conv2D)               │ (None, 38, 38, 32)     │         9,248 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ conv2d_3 (Conv2D)               │ (None, 36, 36, 32)     │         9,248 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ max_pooling2d_1 (MaxPooling2D)  │ (None, 12, 12, 32)     │             0 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ conv2d_4 (Conv2D)               │ (None, 10, 10, 32)     │         9,248 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ conv2d_5 (Conv2D)               │ (None, 8, 8, 32)       │         9,248 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ max_pooling2d_2 (MaxPooling2D)  │ (None, 4, 4, 32)       │             0 │ ## └─────────────────────────────────┴────────────────────────┴───────────────┘ ##  Total params: 48,672 (190.12 KB) ##  Trainable params: 48,672 (190.12 KB) ##  Non-trainable params: 0 (0.00 B) # Now that we have 4x4 feature maps, time to apply global max pooling. model |>   layer_global_max_pooling_2d()  # Finally, we add a classification layer. model |>   layer_dense(units = 10, activation = \"softmax\")"},{"path":"https://keras3.posit.co/dev/articles/sequential_model.html","id":"what-to-do-once-you-have-a-model","dir":"Articles","previous_headings":"","what":"What to do once you have a model","title":"The Sequential model","text":"model architecture ready, want : Train model, evaluate , run inference. See guide training & evaluation built-loops Save model disk restore . See guide serialization & saving.","code":""},{"path":"https://keras3.posit.co/dev/articles/sequential_model.html","id":"feature-extraction-with-a-sequential-model","dir":"Articles","previous_headings":"","what":"Feature extraction with a Sequential model","title":"The Sequential model","text":"Sequential model built, behaves like Functional API model. means every layer input output attribute. attributes can used neat things, like quickly creating model extracts outputs intermediate layers Sequential model: ’s similar example extract features one layer:","code":"initial_model <- keras_model_sequential(input_shape = c(250, 250, 3)) |>   layer_conv_2d(filters = 32, kernel_size = 5, strides = 2, activation = \"relu\") |>   layer_conv_2d(filters = 32, kernel_size = 3, activation = \"relu\") |>   layer_conv_2d(filters = 32, kernel_size = 3, activation = \"relu\")   feature_extractor <- keras_model(     inputs = initial_model$inputs,     outputs = lapply(initial_model$layers, function(x) x$output), )  # Call feature extractor on test input. x <- op_ones(c(1, 250, 250, 3)) features <- feature_extractor(x) initial_model <-   keras_model_sequential(input_shape = c(250, 250, 3)) |>   layer_conv_2d(filters = 32, kernel_size = 5, strides = 2,                 activation = \"relu\") |>   layer_conv_2d(filters = 32, kernel_size = 3, activation = \"relu\",                 name = \"my_intermediate_layer\") |>   layer_conv_2d(filters = 32, kernel_size = 3, activation = \"relu\")  feature_extractor <- keras_model(   inputs = initial_model$inputs,   outputs = get_layer(initial_model, \"my_intermediate_layer\")$output, )  # Call feature extractor on test input. x <- op_ones(c(1, 250, 250, 3)) features <- feature_extractor(x)"},{"path":"https://keras3.posit.co/dev/articles/sequential_model.html","id":"transfer-learning-with-a-sequential-model","dir":"Articles","previous_headings":"","what":"Transfer learning with a Sequential model","title":"The Sequential model","text":"Transfer learning consists freezing bottom layers model training top layers. aren’t familiar , make sure read guide transfer learning. two common transfer learning blueprint involving Sequential models. First, let’s say Sequential model, want freeze layers except last one. case, can call freeze_weights(). Alternatively, can iterate model$layers set layer$trainable <- FALSE layer, except last one. Like : Another common blueprint use Sequential model stack pre-trained model freshly initialized classification layers. Like : transfer learning, probably find frequently using two patterns. ’s need know Sequential models! find building models Keras, see: Guide Functional API Guide making new Layers & Models via subclassing","code":"model <- keras_model_sequential(input_shape = 784) |>   layer_dense(units = 32, activation = \"relu\") |>   layer_dense(units = 32, activation = \"relu\") |>   layer_dense(units = 32, activation = \"relu\") |>   layer_dense(units = 10) # Presumably you would want to first load pre-trained weights. model |> load_model_weights(...) # Freeze all layers except the last one. model |> freeze_weights(from = 1, to = -2) model # note the \"Trainable\" column now visible in the summary table ## Model: \"sequential_9\" ## ┏━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━┳━━━━━━━┓ ## ┃ Layer (type)                ┃ Output Shape          ┃    Param # ┃ Trai… ┃ ## ┡━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━╇━━━━━━━┩ ## │ dense_15 (Dense)            │ (None, 32)            │     25,120 │   N   │ ## ├─────────────────────────────┼───────────────────────┼────────────┼───────┤ ## │ dense_16 (Dense)            │ (None, 32)            │      1,056 │   N   │ ## ├─────────────────────────────┼───────────────────────┼────────────┼───────┤ ## │ dense_17 (Dense)            │ (None, 32)            │      1,056 │   N   │ ## ├─────────────────────────────┼───────────────────────┼────────────┼───────┤ ## │ dense_18 (Dense)            │ (None, 10)            │        330 │   Y   │ ## └─────────────────────────────┴───────────────────────┴────────────┴───────┘ ##  Total params: 27,562 (107.66 KB) ##  Trainable params: 330 (1.29 KB) ##  Non-trainable params: 27,232 (106.38 KB) # Another way to freeze all layers except the last one. for (layer in model$layers[-length(model$layers)]) {   layer$trainable <- FALSE }  # Recompile and train (this will only update the weights of the last layer). model |> compile(...) model |> fit(...) # Load a convolutional base with pre-trained weights base_model <- application_xception(weights = 'imagenet',                                    include_top = FALSE,                                    pooling = 'avg')  # Freeze the base model freeze_weights(base_model)  # Use a Sequential model to add a trainable classifier on top model <- keras_model_sequential() |>   base_model() |>   layer_dense(1000) # Compile & train model |> compile(...) model |> fit(...)"},{"path":"https://keras3.posit.co/dev/articles/serialization_and_saving.html","id":"introduction","dir":"Articles","previous_headings":"","what":"Introduction","title":"Save, serialize, and export models","text":"Keras model consists multiple components: architecture, configuration, specifies layers model contain, ’re connected. set weights values (“state model”). optimizer (defined compiling model). set losses metrics (defined compiling model). Keras API saves pieces together unified format, marked .keras extension. zip archive consisting following: JSON-based configuration file (config.json): Records model, layer, trackables’ configuration. H5-based state file, model.weights.h5 (whole model), directory keys layers weights. metadata file JSON, storing things current Keras version. Let’s take look works.","code":""},{"path":"https://keras3.posit.co/dev/articles/serialization_and_saving.html","id":"how-to-save-and-load-a-model","dir":"Articles","previous_headings":"","what":"How to save and load a model","title":"Save, serialize, and export models","text":"10 seconds read guide, ’s need know. Saving Keras model: Loading model back: Now, let’s look details.","code":"# Get model (Sequential, Functional Model, or Model subclass) model <- ...  # The filename needs to end with the .keras extension model |> save_model('path/to/location.keras') model <- load_model('path/to/location.keras')"},{"path":"https://keras3.posit.co/dev/articles/serialization_and_saving.html","id":"setup","dir":"Articles","previous_headings":"","what":"Setup","title":"Save, serialize, and export models","text":"","code":"library(keras3)"},{"path":"https://keras3.posit.co/dev/articles/serialization_and_saving.html","id":"saving","dir":"Articles","previous_headings":"","what":"Saving","title":"Save, serialize, and export models","text":"section saving entire model single file. file include: model’s architecture/config model’s weight values (learned training) model’s compilation information (compile() called) optimizer state, (enables restart training left)","code":""},{"path":"https://keras3.posit.co/dev/articles/serialization_and_saving.html","id":"apis","dir":"Articles","previous_headings":"Saving","what":"APIs","title":"Save, serialize, and export models","text":"can save model save_model(). can load back load_model(). supported format Keras 3 “Keras v3” format, uses .keras extension. Example:","code":"get_model <- function() {   # Create a simple model.   inputs <- keras_input(shape(32))   outputs <- inputs |> layer_dense(1)   model <-  keras_model(inputs, outputs)   model |> compile(optimizer = optimizer_adam(), loss = \"mean_squared_error\")   model }  model <- get_model()  # Train the model. test_input <- random_uniform(c(128, 32)) test_target <- random_uniform(c(128, 1)) model |> fit(test_input, test_target)  # Calling `save('my_model.keras')` creates a zip archive `my_model.keras`. model |> save_model(\"my_model.keras\")  # It can be used to reconstruct the model identically. reconstructed_model <- load_model(\"my_model.keras\")  # Let's check: stopifnot(all.equal(   model |> predict(test_input),   reconstructed_model |> predict(test_input) ))"},{"path":"https://keras3.posit.co/dev/articles/serialization_and_saving.html","id":"custom-objects","dir":"Articles","previous_headings":"Saving","what":"Custom objects","title":"Save, serialize, and export models","text":"section covers basic workflows handling custom layers, functions, models Keras saving reloading. saving model includes custom objects, subclassed Layer, must define get_config() method object class. arguments passed constructor (initialize() method) custom object aren’t simple objects (anything types like ints, strings, etc.), must also explicitly deserialize arguments from_config() class method. Like : Please see Defining config methods section details examples. saved .keras file lightweight store Python code custom objects. Therefore, reload model, load_model requires access definition custom objects used one following methods: Registering custom objects (preferred), Passing custom objects directly loading, Using custom object scope examples workflow:","code":"layer_custom <- Layer(   \"CustomLayer\",   initialize = function(sublayer, ...) {     super$initialize(...)     self$sublayer <- sublayer   },   call = function(x) {     self$sublayer(x)   },   get_config = function() {     base_config <- super$get_config()     config <- list(       sublayer = serialize_keras_object(self$sublayer)     )     c(base_config, config)   },   from_config = function(cls, config) {     sublayer_config <- config$sublayer     sublayer <- deserialize_keras_object(sublayer_config)     cls(sublayer, !!!config)   } )"},{"path":"https://keras3.posit.co/dev/articles/serialization_and_saving.html","id":"registering-custom-objects-preferred","dir":"Articles","previous_headings":"Saving > Custom objects","what":"Registering custom objects (preferred)","title":"Save, serialize, and export models","text":"preferred method, custom object registration greatly simplifies saving loading code. Calling register_keras_serializable() custom object registers object globally master list, allowing Keras recognize object loading model. Let’s create custom model involving custom layer custom activation function demonstrate . Example:","code":"# Clear all previously registered custom objects set_custom_objects(clear = TRUE) ## named list() layer_custom <- Layer(   \"CustomLayer\",   initialize = function(self, factor) {     super$initialize()     self$factor = factor   },    call = function(self, x) {     x * self$factor   },    get_config = function(self) {     list(factor = self$factor)   } )  # Upon registration, you can optionally specify a package or a name. # If left blank, the package defaults to \"Custom\" and the name defaults to # the class name. register_keras_serializable(layer_custom, package = \"MyLayers\")  custom_fn <- keras3:::py_func2(function(x) x^2, name = \"custom_fn\", convert = TRUE)  register_keras_serializable(custom_fn, name=\"custom_fn\", package=\"my_package\")   # Create the model. get_model <- function() {   inputs <- keras_input(shape(4))   mid <- inputs |> layer_custom(0.5)   outputs <- mid |> layer_dense(1, activation = custom_fn)   model <- keras_model(inputs, outputs)   model |> compile(optimizer = \"rmsprop\", loss = \"mean_squared_error\")   model }   # Train the model. train_model <- function(model) {   input <- random_uniform(c(4, 4))   target <- random_uniform(c(4, 1))   model |> fit(input, target, verbose = FALSE, epochs = 1)   model }  test_input <- random_uniform(c(4, 4)) test_target <- random_uniform(c(4, 1))  model <- get_model() |> train_model() model |> save_model(\"custom_model.keras\", overwrite = TRUE)  # Now, we can simply load without worrying about our custom objects. reconstructed_model <- load_model(\"custom_model.keras\")  # Let's check: stopifnot(all.equal(   model |> predict(test_input, verbose = FALSE),   reconstructed_model |> predict(test_input, verbose = FALSE) ))"},{"path":"https://keras3.posit.co/dev/articles/serialization_and_saving.html","id":"passing-custom-objects-to-load_model","dir":"Articles","previous_headings":"Saving > Custom objects","what":"Passing custom objects to load_model()","title":"Save, serialize, and export models","text":"","code":"model <- get_model() |> train_model()  # Calling `save_model('my_model.keras')` creates a zip archive `my_model.keras`. model |> save_model(\"custom_model.keras\", overwrite = TRUE)  # Upon loading, pass a named list containing the custom objects used in the # `custom_objects` argument of `load_model()`. reconstructed_model <-  load_model(  \"custom_model.keras\",   custom_objects = list(CustomLayer = layer_custom,                         custom_fn = custom_fn), )  # Let's check: stopifnot(all.equal(   model |> predict(test_input, verbose = FALSE),   reconstructed_model |> predict(test_input, verbose = FALSE) ))"},{"path":"https://keras3.posit.co/dev/articles/serialization_and_saving.html","id":"using-a-custom-object-scope","dir":"Articles","previous_headings":"Saving > Custom objects","what":"Using a custom object scope","title":"Save, serialize, and export models","text":"code within custom object scope able recognize custom objects passed scope argument. Therefore, loading model within scope allow loading custom objects. Example:","code":"model <- get_model() |> train_model() model |> save_model(\"custom_model.keras\", overwrite = TRUE)  # Pass the custom objects dictionary to a custom object scope and place # the `keras.models.load_model()` call within the scope. custom_objects <- list(CustomLayer = layer_custom, custom_fn = custom_fn)  with_custom_object_scope(custom_objects, {   reconstructed_model <- load_model(\"custom_model.keras\") })  # Let's check: stopifnot(all.equal(   model |> predict(test_input, verbose = FALSE),   reconstructed_model |> predict(test_input, verbose = FALSE) ))"},{"path":"https://keras3.posit.co/dev/articles/serialization_and_saving.html","id":"model-serialization","dir":"Articles","previous_headings":"Saving","what":"Model serialization","title":"Save, serialize, and export models","text":"section saving model’s configuration, without state. model’s configuration (architecture) specifies layers model contains, layers connected. configuration model, model can created freshly initialized state (weights compilation information).","code":""},{"path":"https://keras3.posit.co/dev/articles/serialization_and_saving.html","id":"apis-1","dir":"Articles","previous_headings":"Saving > Model serialization","what":"APIs","title":"Save, serialize, and export models","text":"following serialization APIs available: clone_model(model): make (randomly initialized) copy model. get_config() cls.from_config(): retrieve configuration layer model, recreate model instance config, respectively. keras.models.model_to_json() keras.models.model_from_json(): similar, JSON strings. keras.saving.serialize_keras_object(): retrieve configuration arbitrary Keras object. keras.saving.deserialize_keras_object(): recreate object instance configuration.","code":""},{"path":"https://keras3.posit.co/dev/articles/serialization_and_saving.html","id":"in-memory-model-cloning","dir":"Articles","previous_headings":"Saving > Model serialization","what":"In-memory model cloning","title":"Save, serialize, and export models","text":"can -memory cloning model via clone_model(). equivalent getting config recreating model config (preserve compilation information layer weights values). Example:","code":"new_model <- clone_model(model)"},{"path":"https://keras3.posit.co/dev/articles/serialization_and_saving.html","id":"get_config-and-from_config","dir":"Articles","previous_headings":"Saving > Model serialization","what":"get_config() and from_config()","title":"Save, serialize, and export models","text":"Calling get_config(model) get_config(layer) return named list containing configuration model layer, respectively. define get_config() contain arguments needed initialize() method model layer. loading time, from_config(config) method call initialize() arguments reconstruct model layer. Layer example: Now let’s reconstruct layer using from_config() method: Sequential model example: Functional model example:","code":"layer <- layer_dense(, 3, activation=\"relu\") layer_config <- get_config(layer) str(layer_config) ## List of 12 ##  $ name              : chr \"dense_4\" ##  $ trainable         : logi TRUE ##  $ dtype             :List of 4 ##   ..$ module         : chr \"keras\" ##   ..$ class_name     : chr \"DTypePolicy\" ##   ..$ config         :List of 1 ##   .. ..$ name: chr \"float32\" ##   ..$ registered_name: NULL ##  $ units             : int 3 ##  $ activation        : chr \"relu\" ##  $ use_bias          : logi TRUE ##  $ kernel_initializer:List of 4 ##   ..$ module         : chr \"keras.initializers\" ##   ..$ class_name     : chr \"GlorotUniform\" ##   ..$ config         :List of 1 ##   .. ..$ seed: NULL ##   ..$ registered_name: NULL ##  $ bias_initializer  :List of 4 ##   ..$ module         : chr \"keras.initializers\" ##   ..$ class_name     : chr \"Zeros\" ##   ..$ config         : Named list() ##   ..$ registered_name: NULL ##  $ kernel_regularizer: NULL ##  $ bias_regularizer  : NULL ##  $ kernel_constraint : NULL ##  $ bias_constraint   : NULL ##  - attr(*, \"__class__\")=<class 'keras.src.layers.core.dense.Dense'> new_layer <- from_config(layer_config) model <- keras_model_sequential(input_shape = c(32)) |>   layer_dense(1) config <- get_config(model) new_model <- from_config(config) inputs <- keras_input(c(32)) outputs <- inputs |> layer_dense(1) model <- keras_model(inputs, outputs) config <- get_config(model) new_model <- from_config(config)"},{"path":"https://keras3.posit.co/dev/articles/serialization_and_saving.html","id":"save_model_config-and-load_model_config","dir":"Articles","previous_headings":"Saving > Model serialization","what":"save_model_config() and load_model_config()","title":"Save, serialize, and export models","text":"similar get_config / from_config, except turns model JSON file, can loaded without original model class. also specific models, isn’t meant layers. Example:","code":"model <- keras_model_sequential(input_shape = c(32)) |>   layer_dense(1) save_model_config(model, \"model_config.json\") new_model <- load_model_config(\"model_config.json\") unlink(\"model_config.json\")"},{"path":"https://keras3.posit.co/dev/articles/serialization_and_saving.html","id":"arbitrary-object-serialization-and-deserialization","dir":"Articles","previous_headings":"Saving > Model serialization","what":"Arbitrary object serialization and deserialization","title":"Save, serialize, and export models","text":"serialize_keras_object() deserialize_keras_object() APIs general-purpose APIs can used serialize deserialize Keras object custom object. foundation saving model architecture behind serialize()/deserialize() calls keras. Example: Note serialization format containing necessary information proper reconstruction: module containing name Keras module identifying module object comes class_name containing name object’s class. config information needed reconstruct object registered_name custom objects. See . Now can reconstruct regularizer.","code":"my_reg <- regularizer_l1(0.005) config <- serialize_keras_object(my_reg) str(config) ## List of 4 ##  $ module         : chr \"keras.regularizers\" ##  $ class_name     : chr \"L1\" ##  $ config         :List of 1 ##   ..$ l1: num 0.005 ##  $ registered_name: NULL new_reg <- deserialize_keras_object(config) new_reg ## <keras.src.regularizers.regularizers.L1 object> ##  signature: (x)"},{"path":"https://keras3.posit.co/dev/articles/serialization_and_saving.html","id":"model-weights-saving","dir":"Articles","previous_headings":"Saving","what":"Model weights saving","title":"Save, serialize, and export models","text":"can choose save & load model’s weights. can useful : need model inference: case won’t need restart training, don’t need compilation information optimizer state. transfer learning: case training new model reusing state prior model, don’t need compilation information prior model.","code":""},{"path":"https://keras3.posit.co/dev/articles/serialization_and_saving.html","id":"apis-for-in-memory-weight-transfer","dir":"Articles","previous_headings":"Saving > Model weights saving","what":"APIs for in-memory weight transfer","title":"Save, serialize, and export models","text":"Weights can copied different objects using get_weights() set_weights(): get_weights(<layer>): Returns list arrays weight values. set_weights(<layer>weights): Sets model/layer weights values provided (arrays). Examples: Transferring weights one layer another, memory Transferring weights one model another model compatible architecture, memory case stateless layers stateless layers change order number weights, models can compatible architectures even extra/missing stateless layers.","code":"create_layer <- function() {   layer <- layer_dense(, 64, activation = \"relu\", name = \"dense_2\")   layer$build(shape(NA, 784))   layer }  layer_1 <- create_layer() layer_2 <- create_layer()  # Copy weights from layer 1 to layer 2 layer_2 |> set_weights(get_weights(layer_1)) # Create a simple functional model inputs <- keras_input(shape=c(784), name=\"digits\") outputs <- inputs |>   layer_dense(64, activation = \"relu\", name = \"dense_1\") |>   layer_dense(64, activation = \"relu\", name = \"dense_2\") |>   layer_dense(10, name = \"predictions\") functional_model <- keras_model(inputs = inputs, outputs = outputs,                                name = \"3_layer_mlp\")  # Define a subclassed model with the same architecture SubclassedModel <- new_model_class(   \"SubclassedModel\",   initialize = function(output_dim, name = NULL) {     super$initialize(name = name)     self$output_dim <- output_dim |> as.integer()     self$dense_1 <- layer_dense(, 64, activation = \"relu\",                                 name = \"dense_1\")     self$dense_2 <- layer_dense(, 64, activation = \"relu\",                                 name = \"dense_2\")     self$dense_3 <- layer_dense(, self$output_dim,                                 name = \"predictions\")   },    call = function(inputs) {     inputs |>       self$dense_1() |>       self$dense_2() |>       self$dense_3()   },    get_config = function(self) {     list(output_dim = self$output_dim,          name = self$name)   } )   subclassed_model <- SubclassedModel(10) # Call the subclassed model once to create the weights. subclassed_model(op_ones(c(1, 784))) |> invisible()  # Copy weights from functional_model to subclassed_model. set_weights(subclassed_model, get_weights(functional_model))  stopifnot(all.equal(   get_weights(functional_model),   get_weights(subclassed_model) )) input <- keras_input(shape = c(784), name = \"digits\") output <- input |>   layer_dense(64, activation = \"relu\", name = \"dense_1\") |>   layer_dense(64, activation = \"relu\", name = \"dense_2\") |>   layer_dense(10, name = \"predictions\") functional_model <- keras_model(inputs, outputs,                                 name = \"3_layer_mlp\")  input <- keras_input(shape = c(784), name = \"digits\") output <- input |>   layer_dense(64, activation = \"relu\", name = \"dense_1\") |>   layer_dense(64, activation = \"relu\", name = \"dense_2\") |>   # Add a dropout layer, which does not contain any weights.   layer_dropout(0.5) |>   layer_dense(10, name = \"predictions\")  functional_model_with_dropout <-   keras_model(input, output, name = \"3_layer_mlp\")  set_weights(functional_model_with_dropout,             get_weights(functional_model))"},{"path":"https://keras3.posit.co/dev/articles/serialization_and_saving.html","id":"apis-for-saving-weights-to-disk-loading-them-back","dir":"Articles","previous_headings":"Saving > Model weights saving","what":"APIs for saving weights to disk & loading them back","title":"Save, serialize, and export models","text":"Weights can saved disk calling save_model_weights(filepath). filename end .weights.h5. Example: Note using freeze_weights() may result different output get_weights(layer) ordering model contains nested layers.","code":"sequential_model = keras_model_sequential(input_shape = c(784),                                           input_name = \"digits\") |>   layer_dense(64, activation = \"relu\", name = \"dense_1\") |>   layer_dense(64, activation = \"relu\", name = \"dense_2\") |>   layer_dense(10, name = \"predictions\") sequential_model |> save_model_weights(\"my_model.weights.h5\") sequential_model |> load_model_weights(\"my_model.weights.h5\")"},{"path":"https://keras3.posit.co/dev/articles/serialization_and_saving.html","id":"transfer-learning-example","dir":"Articles","previous_headings":"Saving > Model weights saving > APIs for saving weights to disk & loading them back","what":"Transfer learning example","title":"Save, serialize, and export models","text":"loading pretrained weights weights file, recommended load weights original checkpointed model, extract desired weights/layers new model. Example:","code":"create_functional_model <- function() {   inputs <- keras_input(shape = c(784), name = \"digits\")   outputs <- inputs |>     layer_dense(64, activation = \"relu\", name = \"dense_1\") |>     layer_dense(64, activation = \"relu\", name = \"dense_2\") |>     layer_dense(10, name = \"predictions\")   keras_model(inputs, outputs, name = \"3_layer_mlp\") }  functional_model <- create_functional_model() functional_model |> save_model_weights(\"pretrained.weights.h5\")  # In a separate program: pretrained_model <- create_functional_model() pretrained_model |> load_model_weights(\"pretrained.weights.h5\")  # Create a new model by extracting layers from the original model: extracted_layers <- pretrained_model$layers |> head(-1) model <- keras_model_sequential(layers = extracted_layers) |>   layer_dense(5, name = \"dense_3\") summary(model) ## Model: \"sequential_4\" ## ┏━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━┓ ## ┃ Layer (type)                    ┃ Output Shape           ┃       Param # ┃ ## ┡━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━┩ ## │ dense_1 (Dense)                 │ (None, 64)             │        50,240 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ dense_2 (Dense)                 │ (None, 64)             │         4,160 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ dense_3 (Dense)                 │ (None, 5)              │           325 │ ## └─────────────────────────────────┴────────────────────────┴───────────────┘ ##  Total params: 54,725 (213.77 KB) ##  Trainable params: 54,725 (213.77 KB) ##  Non-trainable params: 0 (0.00 B)"},{"path":[]},{"path":"https://keras3.posit.co/dev/articles/serialization_and_saving.html","id":"defining-the-config-methods","dir":"Articles","previous_headings":"Saving > Appendix: Handling custom objects","what":"Defining the config methods","title":"Save, serialize, and export models","text":"Specifications: get_config() return JSON-serializable named list order compatible Keras architecture model-saving APIs. from_config(config) (class method) return new layer model object created config. default implementation returns .call(cls, config). NOTE: constructor arguments already serializable, e.g. strings ints, non-custom Keras objects, overriding from_config() necessary. However, complex objects layers models passed initialize(), deserialization must handled explicitly either initialize overriding from_config() method. Example: Note overriding from_config unnecessary MyDense hidden_units, kernel_initializer, kernel_regularizer ints, strings, built-Keras object, respectively. means default from_config implementation cls(!!!config) work intended. complex objects, layers models passed initialize(), example, must explicitly deserialize objects. Let’s take look example model from_config override necessary. Example:","code":"layer_my_dense <- register_keras_serializable(   package = \"MyLayers\", name = \"KernelMult\",   object = Layer(     \"MyDense\",     initialize = function(units,                           ...,                           kernel_regularizer = NULL,                           kernel_initializer = NULL,                           nested_model = NULL) {       super$initialize(...)       self$hidden_units <- units       self$kernel_regularizer <- kernel_regularizer       self$kernel_initializer <- kernel_initializer       self$nested_model <- nested_model     },     get_config = function() {       config <- super$get_config()       # Update the config with the custom layer's parameters       config <- modifyList(config, list(         units = self$hidden_units,         kernel_regularizer = self$kernel_regularizer,         kernel_initializer = self$kernel_initializer,         nested_model = self$nested_model       ))       config     },     build = function(input_shape) {       input_units <- tail(input_shape, 1)       self$kernel <- self$add_weight(         name = \"kernel\",         shape = shape(input_units, self$hidden_units),         regularizer = self$kernel_regularizer,         initializer = self$kernel_initializer,       )     },     call = function(inputs) {       op_matmul(inputs, self$kernel)     }   ) )   layer <- layer_my_dense(units = 16,                         kernel_regularizer = \"l1\",                         kernel_initializer = \"ones\") layer3 <- layer_my_dense(units = 64, nested_model = layer)  config <- serialize_keras_object(layer3) str(config) ## List of 4 ##  $ module         : chr \"<r-globalenv>\" ##  $ class_name     : chr \"MyDense\" ##  $ config         :List of 5 ##   ..$ name        : chr \"my_dense_1\" ##   ..$ trainable   : logi TRUE ##   ..$ dtype       :List of 4 ##   .. ..$ module         : chr \"keras\" ##   .. ..$ class_name     : chr \"DTypePolicy\" ##   .. ..$ config         :List of 1 ##   .. .. ..$ name: chr \"float32\" ##   .. ..$ registered_name: NULL ##   ..$ units       : num 64 ##   ..$ nested_model:List of 4 ##   .. ..$ module         : chr \"<r-globalenv>\" ##   .. ..$ class_name     : chr \"MyDense\" ##   .. ..$ config         :List of 6 ##   .. .. ..$ name              : chr \"my_dense\" ##   .. .. ..$ trainable         : logi TRUE ##   .. .. ..$ dtype             :List of 4 ##   .. .. .. ..$ module         : chr \"keras\" ##   .. .. .. ..$ class_name     : chr \"DTypePolicy\" ##   .. .. .. ..$ config         :List of 1 ##   .. .. .. .. ..$ name: chr \"float32\" ##   .. .. .. ..$ registered_name: NULL ##   .. .. ..$ units             : num 16 ##   .. .. ..$ kernel_regularizer: chr \"l1\" ##   .. .. ..$ kernel_initializer: chr \"ones\" ##   .. ..$ registered_name: chr \"MyLayers>KernelMult\" ##  $ registered_name: chr \"MyLayers>KernelMult\" new_layer <- deserialize_keras_object(config) new_layer ## <MyDense name=my_dense_1, built=False> ##  signature: (*args, **kwargs) `%||%` <- \\(x, y) if(is.null(x)) y else x layer_custom_model <- register_keras_serializable(   package = \"ComplexModels\",   object = Layer(     \"CustomModel\",     initialize = function(first_layer, second_layer = NULL, ...) {       super$initialize(...)       self$first_layer <- first_layer       self$second_layer <- second_layer %||% layer_dense(, 8)     },      get_config = function() {       config <- super$get_config()       config <- modifyList(config, list(         first_layer = self$first_layer,         second_layer = self$second_layer       ))       config     },      from_config = function(config) {       config$first_layer %<>% deserialize_keras_object()       config$second_layer %<>% deserialize_keras_object()       # note that the class is available in methods under the classname symbol,       # (`CustomModel` for this class), and also under the symbol `__class__`       cls(!!!config)       # CustomModel(!!!config)     },     call = function(self, inputs) {       inputs |>         self$first_layer() |>         self$second_layer()     }   ) )  # Let's make our first layer the custom layer from the previous example (MyDense) inputs <- keras_input(c(32)) outputs <-  inputs |> layer_custom_model(first_layer=layer) model <- keras_model(inputs, outputs)  config <- get_config(model) new_model <- from_config(config)"},{"path":"https://keras3.posit.co/dev/articles/serialization_and_saving.html","id":"how-custom-objects-are-serialized","dir":"Articles","previous_headings":"Saving > Appendix: Handling custom objects","what":"How custom objects are serialized","title":"Save, serialize, and export models","text":"serialization format special key custom objects registered via register_keras_serializable(). registered_name key allows easy retrieval loading/deserialization time also allowing users add custom naming. Let’s take look config serializing custom layer MyDense defined . Example: shown, registered_name key contains lookup information Keras master list, including package MyLayers custom name KernelMult gave calling register_keras_serializables(). Take look custom class definition/registration . Note class_name key contains original name class, allowing proper re-initialization from_config. Additionally, note module key NULL since custom object.","code":"layer <- layer_my_dense(   units = 16,   kernel_regularizer = regularizer_l1_l2(l1 = 1e-5, l2 = 1e-4),   kernel_initializer = \"ones\", ) config <- serialize_keras_object(layer) str(config) ## List of 4 ##  $ module         : chr \"<r-globalenv>\" ##  $ class_name     : chr \"MyDense\" ##  $ config         :List of 6 ##   ..$ name              : chr \"my_dense_2\" ##   ..$ trainable         : logi TRUE ##   ..$ dtype             :List of 4 ##   .. ..$ module         : chr \"keras\" ##   .. ..$ class_name     : chr \"DTypePolicy\" ##   .. ..$ config         :List of 1 ##   .. .. ..$ name: chr \"float32\" ##   .. ..$ registered_name: NULL ##   ..$ units             : num 16 ##   ..$ kernel_regularizer:List of 4 ##   .. ..$ module         : chr \"keras.regularizers\" ##   .. ..$ class_name     : chr \"L1L2\" ##   .. ..$ config         :List of 2 ##   .. .. ..$ l1: num 1e-05 ##   .. .. ..$ l2: num 1e-04 ##   .. ..$ registered_name: NULL ##   ..$ kernel_initializer: chr \"ones\" ##  $ registered_name: chr \"MyLayers>KernelMult\""},{"path":"https://keras3.posit.co/dev/articles/training_with_built_in_methods.html","id":"setup","dir":"Articles","previous_headings":"","what":"Setup","title":"Training & evaluation with the built-in methods","text":"","code":"library(keras3)"},{"path":"https://keras3.posit.co/dev/articles/training_with_built_in_methods.html","id":"introduction","dir":"Articles","previous_headings":"","what":"Introduction","title":"Training & evaluation with the built-in methods","text":"guide covers training, evaluation, prediction (inference) models using built-APIs training & validation (fit(), evaluate() predict()). interested leveraging fit() specifying training step function, see Customizing happens fit() guide. interested writing training & evaluation loops scratch, see guide Writing training loop scratch. general, whether using built-loops writing , model training & evaluation works strictly way across every kind Keras model – Sequential models, models built Functional API, models written scratch via model subclassing.","code":""},{"path":"https://keras3.posit.co/dev/articles/training_with_built_in_methods.html","id":"api-overview-a-first-end-to-end-example","dir":"Articles","previous_headings":"","what":"API overview: a first end-to-end example","title":"Training & evaluation with the built-in methods","text":"passing data built-training loops model, either use: Arrays (data small fits memory) tf_dataset objects PyTorch DataLoader instances next paragraphs, ’ll use MNIST dataset NumPy arrays, order demonstrate use optimizers, losses, metrics. Afterwards, ’ll take close look options. Let’s consider following model (, build Functional API, Sequential model subclassed model well): ’s typical end--end workflow looks like, consisting : Training Validation holdout set generated original training data Evaluation test data ’ll use MNIST data example. specify training configuration (optimizer, loss, metrics): call fit(), train model slicing data “batches” size batch_size, repeatedly iterating entire dataset given number epochs. returned history object holds record loss values metric values training: evaluate model test data via evaluate(): Now, let’s review piece workflow detail.","code":"inputs <- keras_input(shape = 784, name=\"digits\") outputs <- inputs |>   layer_dense(units = 64, activation = \"relu\", name = \"dense_1\") |>   layer_dense(units = 64, activation = \"relu\", name = \"dense_2\") |>   layer_dense(units = 10, activation = \"softmax\", name = \"predictions\") model <- keras_model(inputs = inputs, outputs = outputs) summary(model) ## Model: \"functional\" ## ┏━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━┓ ## ┃ Layer (type)                    ┃ Output Shape           ┃       Param # ┃ ## ┡━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━┩ ## │ digits (InputLayer)             │ (None, 784)            │             0 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ dense_1 (Dense)                 │ (None, 64)             │        50,240 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ dense_2 (Dense)                 │ (None, 64)             │         4,160 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ predictions (Dense)             │ (None, 10)             │           650 │ ## └─────────────────────────────────┴────────────────────────┴───────────────┘ ##  Total params: 55,050 (215.04 KB) ##  Trainable params: 55,050 (215.04 KB) ##  Non-trainable params: 0 (0.00 B) c(c(x_train, y_train), c(x_test, y_test)) %<-% dataset_mnist()  # Preprocess the data (these are NumPy arrays) x_train <- array_reshape(x_train, c(60000, 784)) / 255 x_test <- array_reshape(x_test, c(10000, 784)) / 255  # Reserve 10,000 samples for validation x_val <- x_train[1:10000,] y_val <- y_train[1:10000] x_train <- x_train[-c(1:10000),] y_train <- y_train[-c(1:10000)] model |> compile(   # Optimizer   optimizer = optimizer_rmsprop(),   # Loss function to minimize   loss = loss_sparse_categorical_crossentropy(),   # List of metrics to monitor   metrics = list(metric_sparse_categorical_accuracy()) ) history <- model |> fit(   x_train, y_train,   batch_size = 64,   epochs = 2,   # We pass some validation for   # monitoring validation loss and metrics   # at the end of each epoch   validation_data = list(x_val, y_val) ) ## Epoch 1/2 ## 782/782 - 3s - 4ms/step - loss: 0.3410 - sparse_categorical_accuracy: 0.9034 - val_loss: 0.1855 - val_sparse_categorical_accuracy: 0.9460 ## Epoch 2/2 ## 782/782 - 1s - 1ms/step - loss: 0.1590 - sparse_categorical_accuracy: 0.9538 - val_loss: 0.1323 - val_sparse_categorical_accuracy: 0.9621 history ## ## Final epoch (plot to see history): ##                            loss: 0.159 ##     sparse_categorical_accuracy: 0.9538 ##                        val_loss: 0.1323 ## val_sparse_categorical_accuracy: 0.9621 # Evaluate the model on the test data using `evaluate` results <- model |> evaluate(x_test, y_test, batch_size=128) ## 79/79 - 0s - 4ms/step - loss: 0.1273 - sparse_categorical_accuracy: 0.9621 str(results) ## List of 2 ##  $ loss                       : num 0.127 ##  $ sparse_categorical_accuracy: num 0.962 # Generate predictions (probabilities -- the output of the last layer) # on new data using `predict` predictions <- model |> predict(x_test[1:2,]) ## 1/1 - 0s - 191ms/step dim(predictions) ## [1]  2 10"},{"path":"https://keras3.posit.co/dev/articles/training_with_built_in_methods.html","id":"the-compile-method-specifying-a-loss-metrics-and-an-optimizer","dir":"Articles","previous_headings":"","what":"The compile() method: specifying a loss, metrics, and an optimizer","title":"Training & evaluation with the built-in methods","text":"train model fit(), need specify loss function, optimizer, optionally, metrics monitor. pass model arguments compile() method: metrics argument list – model can number metrics. model multiple outputs, can specify different losses metrics output, can modulate contribution output total loss model. find details Passing data multi-input, multi-output models section. Note ’re satisfied default settings, many cases optimizer, loss, metrics can specified via string identifiers shortcut: later reuse, let’s put model definition compile step functions; call several times across different examples guide.","code":"model |> compile(   optimizer = optimizer_rmsprop(learning_rate = 1e-3),   loss = loss_sparse_categorical_crossentropy(),   metrics = list(metric_sparse_categorical_accuracy()) ) model |> compile(   optimizer = \"rmsprop\",   loss = \"sparse_categorical_crossentropy\",   metrics = c(\"sparse_categorical_accuracy\") ) get_uncompiled_model <- function() {   inputs <- keras_input(shape = 784, name = \"digits\")   outputs <- inputs |>     layer_dense(units = 64, activation = \"relu\", name = \"dense_1\") |>     layer_dense(units = 64, activation = \"relu\", name = \"dense_2\") |>     layer_dense(units = 10, activation = \"softmax\", name = \"predictions\")   keras_model(inputs = inputs, outputs = outputs) }  get_compiled_model <- function() {   model <- get_uncompiled_model()   model |> compile(     optimizer = \"rmsprop\",     loss = \"sparse_categorical_crossentropy\",     metrics = c(\"sparse_categorical_accuracy\")   )   model }"},{"path":"https://keras3.posit.co/dev/articles/training_with_built_in_methods.html","id":"many-built-in-optimizers-losses-and-metrics-are-available","dir":"Articles","previous_headings":"The compile() method: specifying a loss, metrics, and an optimizer","what":"Many built-in optimizers, losses, and metrics are available","title":"Training & evaluation with the built-in methods","text":"general, won’t create losses, metrics, optimizers scratch, need likely already part Keras API: Optimizers: [optimizer_sgd()] (without momentum) [optimizer_rmsprop()] [optimizer_adam()] etc. Losses: [loss_mean_squared_error()] [loss_kl_divergence()] [loss_cosine_similarity()] etc. Metrics: [metric_auc()] [metric_precision()] [metric_recall()] etc.","code":""},{"path":"https://keras3.posit.co/dev/articles/training_with_built_in_methods.html","id":"custom-losses","dir":"Articles","previous_headings":"The compile() method: specifying a loss, metrics, and an optimizer","what":"Custom losses","title":"Training & evaluation with the built-in methods","text":"need create custom loss, Keras provides three ways . first method involves creating function accepts inputs y_true y_pred. following example shows loss function computes mean squared error real data predictions: need loss function takes parameters beside y_true y_pred, can subclass Keras base Loss class using [Loss()] implement following two methods: initialize(): accept parameters pass call loss function call(y_true, y_pred): use targets (y_true) model predictions (y_pred) compute model’s loss Let’s say want use mean squared error, added term de-incentivize prediction values far 0.5 (assume categorical targets one-hot encoded take values 0 1). creates incentive model confident, may help reduce overfitting (won’t know works try!). ’s :","code":"custom_mean_squared_error <- function(y_true, y_pred) {   op_mean(op_square(y_true - y_pred), axis = -1) }  model <- get_uncompiled_model() model |> compile(optimizer = \"adam\",                  loss = custom_mean_squared_error)  # We need to one-hot encode the labels to use MSE y_train_one_hot <- op_one_hot(y_train, num_classes = 10) model |> fit(x_train, y_train_one_hot, batch_size = 64, epochs = 2) ## Epoch 1/2 ## 782/782 - 2s - 2ms/step - loss: 0.0157 ## Epoch 2/2 ## 782/782 - 1s - 1ms/step - loss: 0.0074 loss_custom_mse <- Loss(   classname = \"CustomMSE\",   initialize = function(regularization_factor = 0.1, name = \"custom_mse\") {     super$initialize(name = name)     self$regularization_factor <- regularization_factor   },   call = function(y_true, y_pred) {     mse <- op_mean(op_square(y_true - y_pred), axis = -1)     reg <- op_mean(op_square(0.5 - y_pred), axis = -1)     mse + reg * self$regularization_factor   } )  model <- get_uncompiled_model() model |> compile(optimizer=\"adam\", loss = loss_custom_mse())  y_train_one_hot <- op_one_hot(y_train, num_classes=10) model |> fit(x_train, y_train_one_hot, batch_size=64, epochs=1) ## 782/782 - 2s - 2ms/step - loss: 0.0388"},{"path":"https://keras3.posit.co/dev/articles/training_with_built_in_methods.html","id":"custom-metrics","dir":"Articles","previous_headings":"The compile() method: specifying a loss, metrics, and an optimizer","what":"Custom metrics","title":"Training & evaluation with the built-in methods","text":"need metric isn’t part API, can easily create custom metrics subclassing Keras base Metric class using [Metric()]. need implement 4 methods: initialize(), create state variables metric. update_state(y_true, y_pred, sample_weight = NULL), uses targets y_true model predictions y_pred update state variables. result(), uses state variables compute final results. reset_state(), reinitializes state metric. State update results computation kept separate (update_state() result(), respectively) cases, results computation might expensive done periodically. ’s simple example showing implement CategoricalTruePositives metric counts many samples correctly classified belonging given class:","code":"metric_categorical_true_positives <- Metric(   \"CategoricalTruePositives\",    initialize = function(name = \"categorical_true_positives\", ...) {     super$initialize(name = name, ...)     self$true_positives <- self$add_variable(shape = shape(),                                              name = \"ctp\",                                              initializer = \"zeros\")   },    update_state = function(y_true, y_pred, sample_weight = NULL) {     y_pred <- op_argmax(y_pred, axis = 2) |> op_reshape(c(-1, 1))     values <- op_cast(y_true, \"int32\") == op_cast(y_pred, \"int32\")     values <- op_cast(values, \"float32\")     if (!is.null(sample_weight)) {       sample_weight <- op_cast(sample_weight, \"float32\")       values <- op_multiply(values, sample_weight)     }     self$true_positives$assign_add(op_sum(values))   },    result = function() {     self$true_positives$value   },    reset_state = function() {     self$true_positives$assign(0.0)   } )  model <- get_uncompiled_model() model |> compile(   optimizer = optimizer_rmsprop(learning_rate = 1e-3),   loss = loss_sparse_categorical_crossentropy(),   metrics = c(metric_categorical_true_positives()) ) history <- model |> fit(x_train, y_train, batch_size = 64, epochs = 3) ## Epoch 1/3 ## 782/782 - 2s - 2ms/step - categorical_true_positives: 285017.0000 - loss: 0.3444 ## Epoch 2/3 ## 782/782 - 1s - 1ms/step - categorical_true_positives: 284268.0000 - loss: 0.1658 ## Epoch 3/3 ## 782/782 - 1s - 1ms/step - categorical_true_positives: 284173.0000 - loss: 0.1207"},{"path":"https://keras3.posit.co/dev/articles/training_with_built_in_methods.html","id":"handling-losses-and-metrics-that-dont-fit-the-standard-signature","dir":"Articles","previous_headings":"The compile() method: specifying a loss, metrics, and an optimizer","what":"Handling losses and metrics that don’t fit the standard signature","title":"Training & evaluation with the built-in methods","text":"overwhelming majority losses metrics can computed y_true y_pred, y_pred output model – . instance, regularization loss may require activation layer (targets case), activation may model output. cases, can call self$add_loss(loss_value) inside call method custom layer. Losses added way get added “main” loss training (one passed compile()). ’s simple example adds activity regularization (note activity regularization built-Keras layers – layer just sake providing concrete example): Note pass losses via add_loss(), becomes possible call compile() without loss function, since model already loss minimize. Consider following LogisticEndpoint layer: takes inputs targets & logits, tracks crossentropy loss via add_loss(). can use model two inputs (input data & targets), compiled without loss argument, like : information training multi-input models, see section Passing data multi-input, multi-output models.","code":"layer_custom_activity_regularizer <- Layer(   \"ActivityRegularization\",   call = function(inputs) {     self$add_loss(op_sum(inputs) * 0.1)     inputs  # Pass-through layer.   } )  inputs <- keras_input(shape = 784, name = \"digits\") outputs <- inputs |>   layer_dense(units = 32, activation = \"relu\", name = \"dense_1\") |>   layer_custom_activity_regularizer() |>   layer_dense(units = 64, activation = \"relu\", name = \"dense_2\") |>   layer_dense(units = 10, name = \"predictions\")  model <- keras_model(inputs = inputs, outputs = outputs) model |> compile(optimizer = optimizer_rmsprop(learning_rate = 1e-3),                  loss = loss_sparse_categorical_crossentropy(from_logits = TRUE))  # The displayed loss will be much higher than before # due to the regularization component. model |> fit(x_train, y_train, batch_size = 64, epochs = 1) ## 782/782 - 2s - 2ms/step - loss: 2.3721 layer_logistic_endpoint <- Layer(   \"LogisticEndpoint\",   initialize = function(name = NULL) {     super$initialize(name = name)     self$loss_fn <- loss_binary_crossentropy(from_logits = TRUE)   },   call = function(targets, logits, sample_weights = NULL) {     # Compute the training-time loss value and add it     # to the layer using `self.add_loss()`.     loss <- self$loss_fn(targets, logits, sample_weights)     self$add_loss(loss)      # Return the inference-time prediction tensor (for `predict()`).     op_softmax(logits)   } ) inputs <- keras_input(shape = 3, name = \"inputs\") targets <- keras_input(shape = 10, name = \"targets\")  logits <- inputs |> layer_dense(10) predictions <- layer_logistic_endpoint(name = \"predictions\")(targets, logits)  model <- keras_model(inputs = list(inputs, targets),                      outputs = predictions) model |> compile(optimizer = \"adam\")  # No loss argument!  data <- list(   inputs = random_normal(c(3, 3)),   targets = random_normal(c(3, 10)) ) model |> fit(data, epochs = 1) ## 1/1 - 0s - 460ms/step - loss: 0.9638"},{"path":"https://keras3.posit.co/dev/articles/training_with_built_in_methods.html","id":"automatically-setting-apart-a-validation-holdout-set","dir":"Articles","previous_headings":"The compile() method: specifying a loss, metrics, and an optimizer","what":"Automatically setting apart a validation holdout set","title":"Training & evaluation with the built-in methods","text":"first end--end example saw, used validation_data argument pass list arrays list(x_val, y_val) model evaluating validation loss validation metrics end epoch. ’s another option: argument validation_split allows automatically reserve part training data validation. argument value represents fraction data reserved validation, set number higher 0 lower 1. instance, validation_split = 0.2 means “use 20% data validation”, validation_split = 0.6 means “use 60% data validation”. way validation computed taking last x% samples arrays received fit() call, shuffling. Note can use validation_split training NumPy data.","code":"model <- get_compiled_model() model |> fit(x_train, y_train,              batch_size = 64,              validation_split = 0.2, epochs = 1) ## 625/625 - 2s - 3ms/step - loss: 0.3817 - sparse_categorical_accuracy: 0.8919 - val_loss: 0.1953 - val_sparse_categorical_accuracy: 0.9431"},{"path":"https://keras3.posit.co/dev/articles/training_with_built_in_methods.html","id":"training-evaluation-using-tf-dataset-objects","dir":"Articles","previous_headings":"","what":"Training & evaluation using TF Dataset objects","title":"Training & evaluation with the built-in methods","text":"past paragraphs, ’ve seen handle losses, metrics, optimizers, ’ve seen use validation_data validation_split arguments fit(), data passed arrays. Another option use iterator-like, tf.data.Dataset, PyTorch DataLoader, R generator function. Let’s take look former. tfdatasets R package containes set utilities loading preprocessing data way ’s fast scalable. complete guide creating Datasets, see tf.data documentation. can use tf.data train Keras models regardless backend ’re using – whether ’s JAX, PyTorch, TensorFlow. can pass Dataset instance directly methods fit(), evaluate(), predict(): Note Dataset reset end epoch, can reused next epoch. want run training specific number batches Dataset, can pass steps_per_epoch argument, specifies many training steps model run using Dataset moving next epoch. can also pass Dataset instance validation_data argument fit(): end epoch, model iterate validation dataset compute validation loss validation metrics. want run validation specific number batches dataset, can pass validation_steps argument, specifies many validation steps model run validation dataset interrupting validation moving next epoch: Note validation dataset reset use (always evaluating samples epoch epoch). argument validation_split (generating holdout set training data) supported training Dataset objects, since feature requires ability index samples datasets, possible general Dataset API.","code":"library(tfdatasets, exclude = \"shape\") model <- get_compiled_model()  # First, let's create a training Dataset instance. # For the sake of our example, we'll use the same MNIST data as before. train_dataset <- tensor_slices_dataset(list(x_train, y_train))  # Shuffle and slice the dataset. train_dataset <- train_dataset |>   dataset_shuffle(buffer_size=1024) |>   dataset_batch(64)  # Now we get a test dataset. test_dataset <-   tensor_slices_dataset(list(x_test, y_test)) |>   dataset_batch(64)  # Since the dataset already takes care of batching, # we don't pass a `batch_size` argument. model |> fit(train_dataset, epochs = 3) ## Epoch 1/3 ## 782/782 - 2s - 3ms/step - loss: 0.3365 - sparse_categorical_accuracy: 0.9041 ## Epoch 2/3 ## 782/782 - 1s - 1ms/step - loss: 0.1606 - sparse_categorical_accuracy: 0.9521 ## Epoch 3/3 ## 782/782 - 1s - 1ms/step - loss: 0.1178 - sparse_categorical_accuracy: 0.9646 # You can also evaluate or predict on a dataset. result <- model |> evaluate(test_dataset) ## 157/157 - 1s - 4ms/step - loss: 0.1213 - sparse_categorical_accuracy: 0.9596 result ## $loss ## [1] 0.1213005 ## ## $sparse_categorical_accuracy ## [1] 0.9596 model <- get_compiled_model()  # Prepare the training dataset train_dataset <- tensor_slices_dataset(list(x_train, y_train)) train_dataset <- train_dataset |>   dataset_shuffle(buffer_size = 1024) |>   dataset_batch(64)  # Only use the 100 batches per epoch (that's 64 * 100 samples) model |> fit(train_dataset, epochs = 3, steps_per_epoch = 100) ## Epoch 1/3 ## 100/100 - 1s - 11ms/step - loss: 0.8017 - sparse_categorical_accuracy: 0.7806 ## Epoch 2/3 ## 100/100 - 0s - 1ms/step - loss: 0.3661 - sparse_categorical_accuracy: 0.9006 ## Epoch 3/3 ## 100/100 - 0s - 1ms/step - loss: 0.3009 - sparse_categorical_accuracy: 0.9106 model <- get_compiled_model()  # Prepare the training dataset train_dataset <- tensor_slices_dataset(list(x_train, y_train)) train_dataset <- train_dataset |>   dataset_shuffle(buffer_size=1024) |>   dataset_batch(64)  # Prepare the validation dataset val_dataset <- tensor_slices_dataset(list(x_val, y_val)) val_dataset <- val_dataset |> dataset_batch(64)  model |> fit(train_dataset, epochs = 1, validation_data = val_dataset) ## 782/782 - 3s - 3ms/step - loss: 0.3428 - sparse_categorical_accuracy: 0.9022 - val_loss: 0.2337 - val_sparse_categorical_accuracy: 0.9291 model  <- get_compiled_model()  # Prepare the training dataset train_dataset <- tensor_slices_dataset(list(x_train, y_train)) train_dataset <- train_dataset |>   dataset_shuffle(buffer_size = 1024) |>   dataset_batch(64)  # Prepare the validation dataset val_dataset <- tensor_slices_dataset(list(x_val, y_val)) val_dataset <- val_dataset |> dataset_batch(64)  model %>% fit(   train_dataset,   epochs = 1,   # Only run validation using the first 10 batches of the dataset   # using the `validation_steps` argument   validation_data = val_dataset,   validation_steps = 10, ) ## 782/782 - 2s - 3ms/step - loss: 0.3391 - sparse_categorical_accuracy: 0.9035 - val_loss: 0.1997 - val_sparse_categorical_accuracy: 0.9391"},{"path":"https://keras3.posit.co/dev/articles/training_with_built_in_methods.html","id":"using-sample-weighting-and-class-weighting","dir":"Articles","previous_headings":"","what":"Using sample weighting and class weighting","title":"Training & evaluation with the built-in methods","text":"default settings weight sample decided frequency dataset. two methods weight data, independent sample frequency: Class weights Sample weights","code":""},{"path":"https://keras3.posit.co/dev/articles/training_with_built_in_methods.html","id":"class-weights","dir":"Articles","previous_headings":"Using sample weighting and class weighting","what":"Class weights","title":"Training & evaluation with the built-in methods","text":"set passing named list class_weight argument fit(). list maps class indices weight used samples belonging class. can used balance classes without resampling, train model gives importance particular class. instance, class “0” half represented class “1” data, use model |> fit(..., class_weight = c(\"0\" = 1, \"1\" = 0.5)). ’s R example use class weights sample weights give importance correct classification class #5 (digit “5” MNIST dataset).","code":"class_weight <- c(     \"0\" = 1.0,     \"1\" = 1.0,     \"2\" = 1.0,     \"3\" = 1.0,     \"4\" = 1.0,     # Set weight \"2\" for class \"5\",     # making this class 2x more important     \"5\" = 2.0,     \"6\" = 1.0,     \"7\" = 1.0,     \"8\" = 1.0,     \"9\" = 1.0 )  model <- get_compiled_model() model |> fit(x_train, y_train,              class_weight = class_weight,              batch_size = 64, epochs = 1) ## 782/782 - 2s - 2ms/step - loss: 0.3713 - sparse_categorical_accuracy: 0.9018"},{"path":"https://keras3.posit.co/dev/articles/training_with_built_in_methods.html","id":"sample-weights","dir":"Articles","previous_headings":"Using sample weighting and class weighting","what":"Sample weights","title":"Training & evaluation with the built-in methods","text":"fine grained control, building classifier, can use sample_weights. training R arrays: Pass sample_weight argument fit(). training tf_dataset sort iterator: yield (input_batch, label_batch, sample_weight_batch) tuples. “sample weights” array array numbers specify much weight sample batch computing total loss. commonly used imbalanced classification problems (idea give weight rarely-seen classes). weights used ones zeros, array can used mask loss function (entirely discarding contribution certain samples total loss). ’s matching Dataset example:","code":"sample_weight <- rep(1.0, length(y_train)) sample_weight[y_train == 5] <- 2.0  model <- get_compiled_model() model |> fit(   x_train, y_train,   sample_weight = sample_weight,   batch_size = 64, epochs = 1 ) ## 782/782 - 2s - 2ms/step - loss: 0.3740 - sparse_categorical_accuracy: 0.9015 sample_weight <- rep(1.0, length(y_train)) sample_weight[y_train == 5] <- 2.0  # Create a Dataset that includes sample weights # (3rd element in the return tuple). train_dataset <- tensor_slices_dataset(list(     x_train, y_train, sample_weight ))  # Shuffle and slice the dataset. train_dataset <- train_dataset |>   dataset_shuffle(buffer_size = 1024) |>   dataset_batch(64)  model <- get_compiled_model() model |> fit(train_dataset, epochs = 1) ## 782/782 - 2s - 3ms/step - loss: 0.3654 - sparse_categorical_accuracy: 0.9057"},{"path":"https://keras3.posit.co/dev/articles/training_with_built_in_methods.html","id":"passing-data-to-multi-input-multi-output-models","dir":"Articles","previous_headings":"","what":"Passing data to multi-input, multi-output models","title":"Training & evaluation with the built-in methods","text":"previous examples, considering model single input (tensor shape (764)) single output (prediction tensor shape (10)). models multiple inputs outputs? Consider following model, image input shape (32, 32, 3) (’s (height, width, channels)) time series input shape (NA, 10) (’s (timesteps, features)). model two outputs computed combination inputs: “score” (shape (1)) probability distribution five classes (shape (5)). Let’s plot model, can clearly see ’re (note shapes shown plot batch shapes, rather per-sample shapes). plot chunk unnamed-chunk-26 compilation time, can specify different losses different outputs, passing loss functions list: passed single loss function model, loss function applied every output (appropriate ). Likewise metrics: Since gave names output layers, also specify per-output losses metrics via named list: recommend use names 2 outputs. ’s possible give different weights different output-specific losses (instance, one might wish privilege “score” loss example, giving 2x importance class loss), using loss_weights argument: also choose compute loss certain outputs, outputs meant prediction training: Passing data multi-input multi-output model fit() works similar way specifying loss function compile: can pass lists arrays (1:1 mapping outputs received loss function) dicts mapping output names arrays. ’s Dataset use case: similarly R arrays, Dataset return tuple named lists (dicts).","code":"image_input <- keras_input(c(32, 32, 3), name = \"img_input\") timeseries_input <- keras_input(c(NA, 10), name = \"ts_input\")  x1 <- image_input |>   layer_conv_2d(filters = 3, kernel_size = c(3, 3)) |>   layer_global_max_pooling_2d()  x2 <- timeseries_input |>   layer_conv_1d(filters = 3, kernel_size = 3) |>   layer_global_max_pooling_1d()  x <- layer_concatenate(x1, x2)  score_output <- layer_dense(x, 1, name = \"score_output\") class_output <- layer_dense(x, 5, name = \"class_output\")  model <- keras_model(   inputs = list(image_input, timeseries_input),   outputs = list(score_output, class_output) ) plot(model, show_shapes = TRUE) model |> compile(   optimizer = optimizer_rmsprop(1e-3),   loss = list(     loss_mean_squared_error(),     loss_categorical_crossentropy()   ) ) model |> compile(   optimizer = optimizer_rmsprop(1e-3),   loss = list(     loss_mean_squared_error(),     loss_categorical_crossentropy()   ),   metrics = list(     list(       metric_mean_absolute_error(),       metric_mean_absolute_percentage_error()     ),     list(metric_categorical_accuracy())   ) ) model |> compile(   optimizer = optimizer_rmsprop(1e-3),   loss = list(     score_output = loss_mean_squared_error(),     class_output = loss_categorical_crossentropy()   ),   metrics = list(     score_output = list(       metric_mean_absolute_error(),       metric_mean_absolute_percentage_error()     ),     class_output = list(metric_categorical_accuracy())   ) ) model |> compile(   optimizer = optimizer_rmsprop(1e-3),   loss = list(     score_output = loss_mean_squared_error(),     class_output = loss_categorical_crossentropy()   ),   metrics = list(     score_output = list(       metric_mean_absolute_error(),       metric_mean_absolute_percentage_error()     ),     class_output = list(metric_categorical_accuracy())   ),   loss_weights = list(score_output = 2.0, class_output = 1.0) ) # loss list, positional version model |> compile(   optimizer = optimizer_rmsprop(1e-3),   loss = list(NULL, loss_categorical_crossentropy()) )  # Or loss list, named version model |> compile(   optimizer = optimizer_rmsprop(1e-3),   loss = list(class_output = loss_categorical_crossentropy()) ) model |> compile(   optimizer = optimizer_rmsprop(1e-3),   loss = list(     loss_mean_squared_error(),     loss_binary_crossentropy()   ) )  # Generate dummy data img_data <- random_normal(c(100, 32, 32, 3)) ts_data <- random_normal(c(100, 20, 10)) score_targets <- random_normal(c(100, 1)) class_targets <- random_normal(c(100, 5))  # Fit on unnamed lists (positional matching) model |> fit(     list(img_data, ts_data),     list(score_targets, class_targets),     batch_size=32,     epochs=1 ) ## 4/4 - 2s - 559ms/step - class_output_categorical_accuracy: 0.2300 - class_output_loss: 1.8162 - loss: 3.8764 - score_output_loss: 1.6820 - score_output_mean_absolute_error: 0.8826 - score_output_mean_absolute_percentage_error: 1826.5676 # Alternatively, fit on named lists (names matching) model <- keras_model(   inputs = list(image_input, timeseries_input),   outputs = list(score_output = score_output,                  class_output = class_output) ) |> compile(   optimizer = optimizer_rmsprop(1e-3),   loss = list(     loss_mean_squared_error(),     loss_binary_crossentropy()   ) )  model |> fit(   list(img_input = img_data, ts_input = ts_data),   list(score_output = score_targets, class_output = class_targets),   batch_size = 32,   epochs = 1 ) ## 4/4 - 2s - 418ms/step - class_output_loss: 4.1489 - loss: 4.5443 - score_output_loss: -4.1125e-01 train_dataset <- tensor_slices_dataset(list(   list(img_input = img_data, ts_input = ts_data),   list(score_output = score_targets, class_output = class_targets) )) train_dataset <- train_dataset |>   dataset_shuffle(buffer_size = 1024) |>   dataset_batch(64)  model |> fit(train_dataset, epochs = 1) ## 2/2 - 2s - 757ms/step - class_output_loss: 3.8547 - loss: 3.5725 - score_output_loss: -1.5209e-01"},{"path":"https://keras3.posit.co/dev/articles/training_with_built_in_methods.html","id":"using-callbacks","dir":"Articles","previous_headings":"","what":"Using callbacks","title":"Training & evaluation with the built-in methods","text":"Callbacks Keras objects called different points training (start epoch, end batch, end epoch, etc.). can used implement certain behaviors, : validation different points training (beyond built-per-epoch validation) Checkpointing model regular intervals exceeds certain accuracy threshold Changing learning rate model training seems plateauing fine-tuning top layers training seems plateauing Sending email instant message notifications training ends certain performance threshold exceeded Etc. Callbacks can passed list call fit():","code":"model <- get_compiled_model()  callbacks <- list(   callback_early_stopping(     # Stop training when `val_loss` is no longer improving     monitor = \"val_loss\",     # \"no longer improving\" being defined as \"no better than 1e-2 less\"     min_delta = 1e-2,     # \"no longer improving\" being further defined as \"for at least 2 epochs\"     patience = 2,     verbose = 1   ) ) model |> fit(   x_train,   y_train,   epochs = 20,   batch_size = 64,   callbacks = callbacks,   validation_split = 0.2, ) ## Epoch 1/20 ## 625/625 - 2s - 3ms/step - loss: 0.3695 - sparse_categorical_accuracy: 0.8961 - val_loss: 0.1873 - val_sparse_categorical_accuracy: 0.9469 ## Epoch 2/20 ## 625/625 - 1s - 2ms/step - loss: 0.1752 - sparse_categorical_accuracy: 0.9488 - val_loss: 0.1405 - val_sparse_categorical_accuracy: 0.9576 ## Epoch 3/20 ## 625/625 - 1s - 1ms/step - loss: 0.1277 - sparse_categorical_accuracy: 0.9626 - val_loss: 0.1217 - val_sparse_categorical_accuracy: 0.9644 ## Epoch 4/20 ## 625/625 - 1s - 2ms/step - loss: 0.1009 - sparse_categorical_accuracy: 0.9702 - val_loss: 0.1138 - val_sparse_categorical_accuracy: 0.9667 ## Epoch 5/20 ## 625/625 - 1s - 1ms/step - loss: 0.0823 - sparse_categorical_accuracy: 0.9758 - val_loss: 0.1086 - val_sparse_categorical_accuracy: 0.9679 ## Epoch 6/20 ## 625/625 - 1s - 1ms/step - loss: 0.0682 - sparse_categorical_accuracy: 0.9800 - val_loss: 0.1084 - val_sparse_categorical_accuracy: 0.9685 ## Epoch 7/20 ## 625/625 - 1s - 1ms/step - loss: 0.0573 - sparse_categorical_accuracy: 0.9834 - val_loss: 0.1100 - val_sparse_categorical_accuracy: 0.9698 ## Epoch 8/20 ## 625/625 - 1s - 1ms/step - loss: 0.0485 - sparse_categorical_accuracy: 0.9868 - val_loss: 0.1113 - val_sparse_categorical_accuracy: 0.9702 ## Epoch 8: early stopping"},{"path":"https://keras3.posit.co/dev/articles/training_with_built_in_methods.html","id":"many-built-in-callbacks-are-available","dir":"Articles","previous_headings":"Using callbacks","what":"Many built-in callbacks are available","title":"Training & evaluation with the built-in methods","text":"many built-callbacks already available Keras, : callback_model_checkpoint(): Periodically save model. callback_early_stopping(): Stop training training longer improving validation metrics. callback_tensorboard(): periodically write model logs can visualized TensorBoard (details section “Visualization”). callback_csv_logger(): streams loss metrics data CSV file. etc. See callbacks documentation complete list.","code":""},{"path":"https://keras3.posit.co/dev/articles/training_with_built_in_methods.html","id":"writing-your-own-callback","dir":"Articles","previous_headings":"Using callbacks","what":"Writing your own callback","title":"Training & evaluation with the built-in methods","text":"can create custom callback subclassing base [Callback()] class. callback access associated model class property self$model. Make sure read complete guide writing custom callbacks. ’s simple example saving list per-batch loss values training:","code":"callback_loss_history <- Callback(   classname = \"LossHistory\",   initialize = function(file = \"per_training_batch_losses.txt\", ...) {     super$initialize(...)     private$file <- file   },   on_train_begin = function(logs = NULL) {     private$per_batch_losses <- fastmap::faststack()   },   on_train_batch_begin = function(batch, logs = NULL) {     private$per_batch_losses$push(logs$loss)   },   on_train_end = function(logs = NULL) {     per_batch_losses <- private$per_batch_losses$as_list() |> as.numeric()     write(per_batch_losses, private$file)   } )"},{"path":"https://keras3.posit.co/dev/articles/training_with_built_in_methods.html","id":"checkpointing-models","dir":"Articles","previous_headings":"","what":"Checkpointing models","title":"Training & evaluation with the built-in methods","text":"’re training model relatively large datasets, ’s crucial save checkpoints model frequent intervals. easiest way achieve [callback_model_checkpoint()]: ModelCheckpoint callback can used implement fault-tolerance: ability restart training last saved state model case training gets randomly interrupted. ’s basic example: call also write callback saving restoring models. complete guide serialization saving, see guide saving serializing Models.","code":"model <- get_compiled_model()  callbacks <- list(   callback_model_checkpoint(     # Path where to save the model     # The two parameters below mean that we will overwrite     # the current checkpoint if and only if     # the `val_loss` score has improved.     # The saved model name will include the current epoch.     filepath = \"mymodel_{epoch}.keras\",     save_best_only = TRUE,     # Only save a model if `val_loss` has improved.     monitor = \"val_loss\",     verbose = 1   ) ) model |> fit(   x_train, y_train,   epochs = 2, batch_size = 64,   callbacks = callbacks,   validation_split = 0.2 ) ## Epoch 1/2 ## ## Epoch 1: val_loss improved from inf to 0.19488, saving model to mymodel_1.keras ## 625/625 - 2s - 3ms/step - loss: 0.3793 - sparse_categorical_accuracy: 0.8936 - val_loss: 0.1949 - val_sparse_categorical_accuracy: 0.9447 ## Epoch 2/2 ## ## Epoch 2: val_loss improved from 0.19488 to 0.14489, saving model to mymodel_2.keras ## 625/625 - 1s - 2ms/step - loss: 0.1777 - sparse_categorical_accuracy: 0.9475 - val_loss: 0.1449 - val_sparse_categorical_accuracy: 0.9590 # Prepare a directory to store all the checkpoints. checkpoint_dir <- \"./ckpt\" fs::dir_create(checkpoint_dir)  make_or_restore_model <- function() {   # Either restore the latest (best) model, or create a fresh one   # if there is no checkpoint available.   checkpoints <- Sys.glob(file.path(checkpoint_dir, \"model-loss=*.keras\"))    if (length(checkpoints) > 0) {     checkpoint_losses <- sub(\"^model-loss=([0-9.]+)\\\\.keras$\", \"\\\\1\",                              basename(checkpoints)) |> as.numeric()     best_checkpoint <- checkpoints[which.min(checkpoint_losses)]     load_model(best_checkpoint)   } else {     get_compiled_model()   } }  model <- make_or_restore_model() callbacks <- list(   # This callback saves the model every 100 batches.   # We include the training loss in the saved model name.   callback_model_checkpoint(     filepath = file.path(checkpoint_dir, \"model-loss={loss:.2f}.keras\"),     save_freq = 100   ) ) model |> fit(x_train, y_train, epochs = 1, callbacks = callbacks) ## 1563/1563 - 3s - 2ms/step - loss: 0.2932 - sparse_categorical_accuracy: 0.9145"},{"path":"https://keras3.posit.co/dev/articles/training_with_built_in_methods.html","id":"using-learning-rate-schedules","dir":"Articles","previous_headings":"","what":"Using learning rate schedules","title":"Training & evaluation with the built-in methods","text":"common pattern training deep learning models gradually reduce learning training progresses. generally known “learning rate decay”. learning decay schedule static (fixed advance, function current epoch current batch index), dynamic (responding current behavior model, particular validation loss).","code":""},{"path":"https://keras3.posit.co/dev/articles/training_with_built_in_methods.html","id":"passing-a-schedule-to-an-optimizer","dir":"Articles","previous_headings":"Using learning rate schedules","what":"Passing a schedule to an optimizer","title":"Training & evaluation with the built-in methods","text":"can easily use static learning rate decay schedule passing schedule object learning_rate argument optimizer: Several built-schedules available: ExponentialDecay, PiecewiseConstantDecay, PolynomialDecay, InverseTimeDecay.","code":"initial_learning_rate <- 0.1 lr_schedule <- learning_rate_schedule_exponential_decay(     initial_learning_rate, decay_steps=100000, decay_rate=0.96,     staircase=TRUE )  optimizer <- optimizer_rmsprop(learning_rate = lr_schedule)"},{"path":"https://keras3.posit.co/dev/articles/training_with_built_in_methods.html","id":"using-callbacks-to-implement-a-dynamic-learning-rate-schedule","dir":"Articles","previous_headings":"Using learning rate schedules","what":"Using callbacks to implement a dynamic learning rate schedule","title":"Training & evaluation with the built-in methods","text":"dynamic learning rate schedule (instance, decreasing learning rate validation loss longer improving) achieved schedule objects, since optimizer access validation metrics. However, callbacks access metrics, including validation metrics! can thus achieve pattern using callback modifies current learning rate optimizer. fact, even built-[callback_reduce_lr_on_plateau()].","code":""},{"path":"https://keras3.posit.co/dev/articles/training_with_built_in_methods.html","id":"visualizing-loss-and-metrics-during-training-with-tensorboard","dir":"Articles","previous_headings":"","what":"Visualizing loss and metrics during training with TensorBoard","title":"Training & evaluation with the built-in methods","text":"best way keep eye model training use TensorBoard – browser-based application can run locally provides : Live plots loss metrics training evaluation (optionally) Visualizations histograms layer activations (optionally) 3D visualizations embedding spaces learned layer_embedding() installed TensorFlow pip, able launch TensorBoard command line: R using:","code":"tensorboard --logdir=/full_path_to_your_logs tensorflow::tensorboard(logdir = \"/full_path_to_your_logs\")"},{"path":"https://keras3.posit.co/dev/articles/training_with_built_in_methods.html","id":"using-the-tensorboard-callback","dir":"Articles","previous_headings":"Visualizing loss and metrics during training with TensorBoard","what":"Using the TensorBoard callback","title":"Training & evaluation with the built-in methods","text":"easiest way use TensorBoard Keras model fit() method [callback_tensorboard()]. simplest case, just specify want callback write logs, ’re good go: information, see callback_tensorboard().","code":"tb_callback <- callback_tensorboard(   log_dir = \"/full_path_to_your_logs\",   histogram_freq = 0, # How often to log histogram visualizations   embeddings_freq = 0, # How often to log embedding visualizations   update_freq = \"epoch\", # How often to write logs (default: once per epoch) )"},{"path":"https://keras3.posit.co/dev/articles/transfer_learning.html","id":"setup","dir":"Articles","previous_headings":"","what":"Setup","title":"Transfer learning & fine-tuning","text":"","code":"library(keras3)"},{"path":"https://keras3.posit.co/dev/articles/transfer_learning.html","id":"introduction","dir":"Articles","previous_headings":"","what":"Introduction","title":"Transfer learning & fine-tuning","text":"Transfer learning consists taking features learned one problem, leveraging new, similar problem. instance, features model learned identify raccoon may useful kick-start model meant identify tanukis. Transfer learning usually done tasks dataset little data train full-scale model scratch. common incarnation transfer learning context deep learning following workflow: Take layers previously trained model. Freeze , avoid destroying information contain future training rounds. Add new, trainable layers top frozen layers. learn turn old features predictions new dataset. Train new layers dataset. last, optional step, fine-tuning, consists unfreezing entire model obtained (part ), re-training new data low learning rate. can potentially achieve meaningful improvements, incrementally adapting pretrained features new data. First, go Keras trainable API detail, underlies transfer learning & fine-tuning workflows. , ’ll demonstrate typical workflow taking model pretrained ImageNet dataset, retraining Kaggle “cats vs dogs” classification dataset. adapted Deep Learning Python 2016 blog post “building powerful image classification models using little data”.","code":""},{"path":"https://keras3.posit.co/dev/articles/transfer_learning.html","id":"freezing-layers-understanding-the-trainable-attribute","dir":"Articles","previous_headings":"","what":"Freezing layers: understanding the trainable attribute","title":"Transfer learning & fine-tuning","text":"Layers & models three weight attributes: weights list weights variables layer. trainable_weights list meant updated (via gradient descent) minimize loss training. non_trainable_weights list aren’t meant trained. Typically updated model forward pass. Example: Dense layer 2 trainable weights (kernel & bias) general, weights trainable weights. built-layer non-trainable weights [layer_batch_normalization()]. uses non-trainable weights keep track mean variance inputs training. learn use non-trainable weights custom layers, see guide writing new layers scratch. Example: BatchNormalization layer 2 trainable weights 2 non-trainable weights Layers & models also feature boolean attribute trainable. value can changed. Setting layer$trainable FALSE moves layer’s weights trainable non-trainable. called “freezing” layer: state frozen layer won’t updated training (either training fit() training custom loop relies trainable_weights apply gradient updates). Example: setting trainable False trainable weight becomes non-trainable, value longer updated training. confuse layer$trainable attribute argument training layer$call() (controls whether layer run forward pass inference mode training mode). information, see Keras FAQ.","code":"layer <- layer_dense(units = 3) layer$build(shape(NULL, 4))  # Create the weights  length(layer$weights) ## [1] 2 length(layer$trainable_weights) ## [1] 2 length(layer$non_trainable_weights) ## [1] 0 layer <- layer_batch_normalization() layer$build(shape(NA, 4))  # Create the weights  length(layer$weights) ## [1] 4 length(layer$trainable_weights) ## [1] 2 length(layer$non_trainable_weights) ## [1] 2 layer <- layer_dense(units = 3) layer$build(shape(NULL, 4))  # Create the weights layer$trainable <- FALSE  # Freeze the layer  length(layer$weights) ## [1] 2 length(layer$trainable_weights) ## [1] 0 length(layer$non_trainable_weights) ## [1] 2 # Make a model with 2 layers layer1 <- layer_dense(units = 3, activation = \"relu\") layer2 <- layer_dense(units = 3, activation = \"sigmoid\") model <- keras_model_sequential(input_shape = 3) |>   layer1() |>   layer2()  # Freeze the first layer layer1$trainable <- FALSE  # Keep a copy of the weights of layer1 for later reference # (get_weights() returns a list of R arrays, #  layer$weights returns a list of KerasVariables) initial_layer1_weights_values <- get_weights(layer1)  # Train the model model |> compile(optimizer = \"adam\", loss = \"mse\") model |> fit(random_normal(c(2, 3)), random_normal(c(2, 3)), epochs = 1) ## 1/1 - 0s - 487ms/step - loss: 2.8362 # Check that the weights of layer1 have not changed during training final_layer1_weights_values <- get_weights(layer1)  all.equal(initial_layer1_weights_values,           final_layer1_weights_values) ## [1] TRUE"},{"path":"https://keras3.posit.co/dev/articles/transfer_learning.html","id":"recursive-setting-of-the-trainable-attribute","dir":"Articles","previous_headings":"","what":"Recursive setting of the trainable attribute","title":"Transfer learning & fine-tuning","text":"set $trainable <- FALSE model layer sublayers, children layers become non-trainable well. Example:","code":"inner_model <- keras_model_sequential(input_shape = 3) |>   layer_dense(units = 3, activation = \"relu\") |>   layer_dense(units = 3, activation = \"relu\")  model <- keras_model_sequential(input_shape = 3) |>   inner_model() |>   layer_dense(units = 3, activation = \"sigmoid\")  model$trainable <- FALSE  # Freeze the outer model  inner_model$trainable  # All layers in `model` are now frozen ## [1] FALSE inner_model$layers[[1]]$trainable  # `trainable` is propagated recursively ## [1] FALSE"},{"path":"https://keras3.posit.co/dev/articles/transfer_learning.html","id":"the-typical-transfer-learning-workflow","dir":"Articles","previous_headings":"","what":"The typical transfer-learning workflow","title":"Transfer learning & fine-tuning","text":"leads us typical transfer learning workflow can implemented Keras: Instantiate base model load pre-trained weights . Freeze layers base model setting trainable <- FALSE. Create new model top output one (several) layers base model. Train new model new dataset. Note alternative, lightweight workflow also : Instantiate base model load pre-trained weights . Run new dataset record output one (several) layers base model. called feature extraction. Use output input data new, smaller model. key advantage second workflow run base model data, rather per epoch training. ’s lot faster & cheaper. issue second workflow, though, doesn’t allow dynamically modify input data new model training, required data augmentation, instance. Transfer learning typically used tasks new dataset little data train full-scale model scratch, scenarios data augmentation important. follows, focus first workflow. ’s first workflow looks like Keras: First, instantiate base model pre-trained weights. , freeze base model. Create new model top. Train model new data.","code":"base_model <- application_xception(   weights = 'imagenet', # Load weights pre-trained on ImageNet.   input_shape = c(150, 150, 3),   include_top = FALSE  # Do not include the ImageNet classifier at the top. ) base_model$trainable <- FALSE inputs <- keras_input(shape = c(150, 150, 3)) # We make sure that the base_model is running in inference mode here, # by passing `training <- FALSE`. This is important for fine-tuning, as you will # learn in a few paragraphs. outputs <- inputs |>   base_model(training = FALSE) |>   # Convert features of shape `base_model$output_shape[-1]` to vectors   layer_global_average_pooling_2d() |>   # A Dense classifier with a single unit (binary classification)   layer_dense(1)  model <- keras_model(inputs, outputs) model |> compile(   optimizer = optimizer_adam(),   loss = loss_binary_crossentropy(from_logits = TRUE),   metrics = list(metric_binary_accuracy()) ) model |> fit(new_dataset, epochs = 20,              callbacks = ..., validation_data = ...)"},{"path":"https://keras3.posit.co/dev/articles/transfer_learning.html","id":"fine-tuning","dir":"Articles","previous_headings":"","what":"Fine-tuning","title":"Transfer learning & fine-tuning","text":"model converged new data, can try unfreeze part base model retrain whole model end--end low learning rate. optional last step can potentially give incremental improvements. also potentially lead quick overfitting – keep mind. critical step model frozen layers trained convergence. mix randomly-initialized trainable layers trainable layers hold pre-trained features, randomly-initialized layers cause large gradient updates training, destroy pre-trained features. ’s also critical use low learning rate stage, training much larger model first round training, dataset typically small. result, risk overfitting quickly apply large weight updates. , want readapt pretrained weights incremental way. implement fine-tuning whole base model: Important note compile() trainable Calling compile() model meant “freeze” behavior model. implies trainable attribute values time model compiled preserved throughout lifetime model, compile called . Hence, change trainable value, make sure call compile() model changes taken account. Important notes BatchNormalization layer Many image models contain BatchNormalization layers. layer special case every imaginable count. things keep mind. BatchNormalization contains 2 non-trainable weights get updated training. variables tracking mean variance inputs. set bn_layer$trainable <- FALSE, BatchNormalization layer run inference mode, update mean & variance statistics. case layers general, weight trainability & inference/training modes two orthogonal concepts. two tied case BatchNormalization layer. unfreeze model contains BatchNormalization layers order fine-tuning, keep BatchNormalization layers inference mode passing training = FALSE calling base model. Otherwise updates applied non-trainable weights suddenly destroy model learned. ’ll see pattern action end--end example end guide.","code":"# Unfreeze the base model base_model$trainable <- TRUE  # It's important to recompile your model after you make any changes # to the `trainable` attribute of any inner layer, so that your changes # are take into account model |> compile(   optimizer = optimizer_adam(1e-5), # Very low learning rate   loss = loss_binary_crossentropy(from_logits = TRUE),   metrics = c(metric_binary_accuracy()) )  # Train end-to-end. Be careful to stop before you overfit! model |> fit(new_dataset, epochs = 10,              callbacks = ..., validation_data = ...)"},{"path":"https://keras3.posit.co/dev/articles/transfer_learning.html","id":"an-end-to-end-example-fine-tuning-an-image-classification-model-on-a-cats-vs--dogs-dataset","dir":"Articles","previous_headings":"","what":"An end-to-end example: fine-tuning an image classification model on a cats vs. dogs dataset","title":"Transfer learning & fine-tuning","text":"solidify concepts, let’s walk concrete end--end transfer learning & fine-tuning example. load Xception model, pre-trained ImageNet, use Kaggle “cats vs. dogs” classification dataset.","code":""},{"path":"https://keras3.posit.co/dev/articles/transfer_learning.html","id":"getting-the-data","dir":"Articles","previous_headings":"An end-to-end example: fine-tuning an image classification model on a cats vs. dogs dataset","what":"Getting the data","title":"Transfer learning & fine-tuning","text":"First, let’s fetch cats vs. dogs dataset using TFDS. dataset, ’ll probably want use utility [image_dataset_from_directory()] generate similar labeled dataset objects set images disk filed class-specific folders. Transfer learning useful working small datasets. keep dataset small, use 40% original training data (25,000 images) training, 10% validation, 10% testing. first 9 images training dataset – can see, ’re different sizes. can also see label 1 “dog” label 0 “cat”.","code":"reticulate::py_require(\"tensorflow-datasets\") tfds <- reticulate::import(\"tensorflow_datasets\")  c(train_ds, validation_ds, test_ds) %<-% tfds$load(   \"cats_vs_dogs\",   # Reserve 10% for validation and 10% for test   split = c(\"train[:40%]\", \"train[40%:50%]\", \"train[50%:60%]\"),   as_supervised = TRUE  # Include labels )  length(train_ds) ## [1] 9305 library(tfdatasets, exclude = \"shape\")  par(mfrow = c(3, 3), mar = c(1,0,1.5,0)) train_ds |>   dataset_take(9) |>   as_array_iterator() |>   iterate(function(batch) {     c(image, label) %<-% batch     plot(as.raster(image, max = 255L))     title(sprintf(       \"label: %s   size: %s\",       label, paste(dim(image), collapse = \" x \")))   })"},{"path":"https://keras3.posit.co/dev/articles/transfer_learning.html","id":"standardizing-the-data","dir":"Articles","previous_headings":"An end-to-end example: fine-tuning an image classification model on a cats vs. dogs dataset","what":"Standardizing the data","title":"Transfer learning & fine-tuning","text":"raw images variety sizes. addition, pixel consists 3 integer values 0 255 (RGB level values). isn’t great fit feeding neural network. need 2 things: Standardize fixed image size. pick 150x150. Normalize pixel values -1 1. ’ll using Normalization layer part model . general, ’s good practice develop models take raw data input, opposed models take already-preprocessed data. reason , model expects preprocessed data, time export model use elsewhere (web browser, mobile app), ’ll need reimplement exact preprocessing pipeline. gets tricky quickly. least possible amount preprocessing hitting model. , ’ll image resizing data pipeline (deep neural network can process contiguous batches data), ’ll input value scaling part model, create . Let’s resize images 150x150:","code":"resize_fn <- layer_resizing(width = 150, height = 150) resize_pair <- function(x, y) list(resize_fn(x), y)  train_ds <- train_ds |> dataset_map(resize_pair) validation_ds <- validation_ds |> dataset_map(resize_pair) test_ds <- test_ds |> dataset_map(resize_pair)"},{"path":"https://keras3.posit.co/dev/articles/transfer_learning.html","id":"using-random-data-augmentation","dir":"Articles","previous_headings":"An end-to-end example: fine-tuning an image classification model on a cats vs. dogs dataset","what":"Using random data augmentation","title":"Transfer learning & fine-tuning","text":"don’t large image dataset, ’s good practice artificially introduce sample diversity applying random yet realistic transformations training images, random horizontal flipping small random rotations. helps expose model different aspects training data slowing overfitting. Let’s batch data use prefetching optimize loading speed. Let’s visualize first image first batch looks like various random transformations:","code":"data_augmentation <- keras_model_sequential() |>   layer_random_flip(\"horizontal\") |>   layer_random_rotation(.1)  train_ds <- train_ds %>%   dataset_map(function(x, y) list(data_augmentation(x), y)) library(tensorflow, exclude = c(\"shape\", \"set_random_seed\")) batch_size <- 64  train_ds <- train_ds |>   dataset_batch(batch_size) |>   dataset_prefetch()  validation_ds <- validation_ds |>   dataset_batch(batch_size) |>   dataset_prefetch()  test_ds <- test_ds |>   dataset_batch(batch_size) |>   dataset_prefetch() batch <- train_ds |>   dataset_take(1) |>   as_iterator() |>   iter_next()  c(images, labels) %<-% batch first_image <- images[1, all_dims(), drop = TRUE] augmented_image <- data_augmentation(first_image, training = TRUE)  plot_image <- function(image, main = deparse1(substitute(image))) {   image |>     as.array() |> # convert from tensor to R array     as.raster(max = 255) |>     plot()    if(!is.null(main))     title(main) }  par(mfrow = c(2, 2), mar = c(1, 1, 1.5, 1)) plot_image(first_image) plot_image(augmented_image) plot_image(data_augmentation(first_image, training = TRUE), \"augmented 2\") plot_image(data_augmentation(first_image, training = TRUE), \"augmented 3\")"},{"path":"https://keras3.posit.co/dev/articles/transfer_learning.html","id":"build-a-model","dir":"Articles","previous_headings":"","what":"Build a model","title":"Transfer learning & fine-tuning","text":"Now let’s built model follows blueprint ’ve explained earlier. Note : add Rescaling layer scale input values (initially [0, 255] range) [-1, 1] range. add Dropout layer classification layer, regularization. make sure pass training=FALSE calling base model, runs inference mode, batchnorm statistics don’t get updated even unfreeze base model fine-tuning.","code":"base_model <- application_xception(   weights = \"imagenet\", # Load weights pre-trained on ImageNet.   input_shape = c(150, 150, 3),   include_top = FALSE, # Do not include the ImageNet classifier at the top. )  # Freeze the base_model base_model$trainable <- FALSE  # Create new model on top inputs <- keras_input(shape = c(150, 150, 3))  # Pre-trained Xception weights requires that input be scaled # from (0, 255) to a range of (-1., +1.), the rescaling layer # outputs: `(inputs * scale) + offset` scale_layer <- layer_rescaling(scale = 1 / 127.5, offset = -1) x <- scale_layer(inputs)  # The base model contains batchnorm layers. We want to keep them in inference mode # when we unfreeze the base model for fine-tuning, so we make sure that the # base_model is running in inference mode here. outputs <- x |>   base_model(training = FALSE) |>   layer_global_average_pooling_2d() |>   layer_dropout(0.2) |>   layer_dense(1)  model <- keras_model(inputs, outputs)  summary(model, show_trainable = TRUE) ## Model: \"functional_8\" ## ┏━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━┳━━━━━━━┓ ## ┃ Layer (type)                ┃ Output Shape          ┃    Param # ┃ Trai… ┃ ## ┡━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━╇━━━━━━━┩ ## │ input_layer_7 (InputLayer)  │ (None, 150, 150, 3)   │          0 │   -   │ ## ├─────────────────────────────┼───────────────────────┼────────────┼───────┤ ## │ rescaling (Rescaling)       │ (None, 150, 150, 3)   │          0 │   -   │ ## ├─────────────────────────────┼───────────────────────┼────────────┼───────┤ ## │ xception (Functional)       │ (None, 5, 5, 2048)    │ 20,861,480 │   N   │ ## ├─────────────────────────────┼───────────────────────┼────────────┼───────┤ ## │ global_average_pooling2d_1  │ (None, 2048)          │          0 │   -   │ ## │ (GlobalAveragePooling2D)    │                       │            │       │ ## ├─────────────────────────────┼───────────────────────┼────────────┼───────┤ ## │ dropout (Dropout)           │ (None, 2048)          │          0 │   -   │ ## ├─────────────────────────────┼───────────────────────┼────────────┼───────┤ ## │ dense_8 (Dense)             │ (None, 1)             │      2,049 │   Y   │ ## └─────────────────────────────┴───────────────────────┴────────────┴───────┘ ##  Total params: 20,863,529 (79.59 MB) ##  Trainable params: 2,049 (8.00 KB) ##  Non-trainable params: 20,861,480 (79.58 MB)"},{"path":"https://keras3.posit.co/dev/articles/transfer_learning.html","id":"train-the-top-layer","dir":"Articles","previous_headings":"","what":"Train the top layer","title":"Transfer learning & fine-tuning","text":"","code":"model |> compile(   optimizer = optimizer_adam(),   loss = loss_binary_crossentropy(from_logits = TRUE),   metrics = list(metric_binary_accuracy()) )  epochs <- 1 model |> fit(train_ds, epochs = epochs, validation_data = validation_ds) ## 146/146 - 53s - 360ms/step - binary_accuracy: 0.9183 - loss: 0.1887 - val_binary_accuracy: 0.9669 - val_loss: 0.0926"},{"path":"https://keras3.posit.co/dev/articles/transfer_learning.html","id":"do-a-round-of-fine-tuning-of-the-entire-model","dir":"Articles","previous_headings":"","what":"Do a round of fine-tuning of the entire model","title":"Transfer learning & fine-tuning","text":"Finally, let’s unfreeze base model train entire model end--end low learning rate. Importantly, although base model becomes trainable, still running inference mode since passed training=FALSE calling built model. means batch normalization layers inside won’t update batch statistics. , wreck havoc representations learned model far. 10 epochs, fine-tuning gains us nice improvement . Let’s evaluate model test dataset:","code":"# Unfreeze the base_model. Note that it keeps running in inference mode # since we passed `training=FALSE` when calling it. This means that # the batchnorm layers will not update their batch statistics. # This prevents the batchnorm layers from undoing all the training # we've done so far. base_model$trainable <- TRUE summary(model, show_trainable = TRUE) ## Model: \"functional_8\" ## ┏━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━┳━━━━━━━┓ ## ┃ Layer (type)                ┃ Output Shape          ┃    Param # ┃ Trai… ┃ ## ┡━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━╇━━━━━━━┩ ## │ input_layer_7 (InputLayer)  │ (None, 150, 150, 3)   │          0 │   -   │ ## ├─────────────────────────────┼───────────────────────┼────────────┼───────┤ ## │ rescaling (Rescaling)       │ (None, 150, 150, 3)   │          0 │   -   │ ## ├─────────────────────────────┼───────────────────────┼────────────┼───────┤ ## │ xception (Functional)       │ (None, 5, 5, 2048)    │ 20,861,480 │   Y   │ ## ├─────────────────────────────┼───────────────────────┼────────────┼───────┤ ## │ global_average_pooling2d_1  │ (None, 2048)          │          0 │   -   │ ## │ (GlobalAveragePooling2D)    │                       │            │       │ ## ├─────────────────────────────┼───────────────────────┼────────────┼───────┤ ## │ dropout (Dropout)           │ (None, 2048)          │          0 │   -   │ ## ├─────────────────────────────┼───────────────────────┼────────────┼───────┤ ## │ dense_8 (Dense)             │ (None, 1)             │      2,049 │   Y   │ ## └─────────────────────────────┴───────────────────────┴────────────┴───────┘ ##  Total params: 20,867,629 (79.60 MB) ##  Trainable params: 20,809,001 (79.38 MB) ##  Non-trainable params: 54,528 (213.00 KB) ##  Optimizer params: 4,100 (16.02 KB) model |> compile(   optimizer = optimizer_adam(1e-5), # Low learning rate   loss = loss_binary_crossentropy(from_logits = TRUE),   metrics = list(metric_binary_accuracy()) )  epochs <- 1 model |> fit(train_ds, epochs = epochs, validation_data = validation_ds) ## 146/146 - 97s - 664ms/step - binary_accuracy: 0.8660 - loss: 0.3213 - val_binary_accuracy: 0.9652 - val_loss: 0.1022 model |> evaluate(test_ds) ## 37/37 - 2s - 46ms/step - binary_accuracy: 0.9540 - loss: 0.1103 ## $binary_accuracy ## [1] 0.9539983 ## ## $loss ## [1] 0.1102593"},{"path":"https://keras3.posit.co/dev/articles/understanding_masking_and_padding.html","id":"setup","dir":"Articles","previous_headings":"","what":"Setup","title":"Understanding masking & padding","text":"","code":"library(keras3)"},{"path":"https://keras3.posit.co/dev/articles/understanding_masking_and_padding.html","id":"introduction","dir":"Articles","previous_headings":"","what":"Introduction","title":"Understanding masking & padding","text":"Masking way tell sequence-processing layers certain timesteps input missing, thus skipped processing data. Padding special form masking masked steps start end sequence. Padding comes need encode sequence data contiguous batches: order make sequences batch fit given standard length, necessary pad truncate sequences. Let’s take close look.","code":""},{"path":"https://keras3.posit.co/dev/articles/understanding_masking_and_padding.html","id":"padding-sequence-data","dir":"Articles","previous_headings":"","what":"Padding sequence data","title":"Understanding masking & padding","text":"processing sequence data, common individual samples different lengths. Consider following example (text tokenized words): vocabulary lookup, data might vectorized integers, e.g.: data nested list individual samples length 3, 5, 6, respectively. Since input data deep learning model must single tensor (shape e.g. (batch_size, 6, vocab_size) case), samples shorter longest item need padded placeholder value (alternatively, one might also truncate long samples padding short samples). Keras provides utility function truncate pad Python lists common length: pad_sequences.","code":"data <- list(   c(\"Hello\", \"world\", \"!\"),   c(\"How\", \"are\", \"you\", \"doing\", \"today\"),   c(\"The\", \"weather\", \"will\", \"be\", \"nice\", \"tomorrow\") ) data <- list(   c(71, 1331, 4231),   c(73, 8, 3215, 55, 927),   c(83, 91, 1, 645, 1253, 927) ) raw_inputs <- list(   c(711, 632, 71),   c(73, 8, 3215, 55, 927),   c(83, 91, 1, 645, 1253, 927) )  # By default, this will pad using 0s; it is configurable via the # \"value\" parameter. # Note that you could use \"pre\" padding (at the beginning) or # \"post\" padding (at the end). # We recommend using \"post\" padding when working with RNN layers # (in order to be able to use the # CuDNN implementation of the layers). padded_inputs <- pad_sequences(raw_inputs, padding=\"post\") padded_inputs ##      [,1] [,2] [,3] [,4] [,5] [,6] ## [1,]  711  632   71    0    0    0 ## [2,]   73    8 3215   55  927    0 ## [3,]   83   91    1  645 1253  927"},{"path":"https://keras3.posit.co/dev/articles/understanding_masking_and_padding.html","id":"masking","dir":"Articles","previous_headings":"","what":"Masking","title":"Understanding masking & padding","text":"Now samples uniform length, model must informed part data actually padding ignored. mechanism masking. three ways introduce input masks Keras models: Add layer_masking layer. Configure layer_embedding layer mask_zero=TRUE. Pass mask argument manually calling layers support argument (e.g. RNN layers).","code":""},{"path":"https://keras3.posit.co/dev/articles/understanding_masking_and_padding.html","id":"mask-generating-layers-embedding-and-masking","dir":"Articles","previous_headings":"","what":"Mask-generating layers: Embedding and Masking","title":"Understanding masking & padding","text":"hood, layers create mask tensor (2D tensor shape (batch, sequence_length)), attach tensor output returned Masking Embedding layer. can see printed result, mask 2D boolean tensor shape (batch_size, sequence_length), individual FALSE entry indicates corresponding timestep ignored processing.","code":"embedding <- layer_embedding(input_dim=5000, output_dim=16, mask_zero=TRUE) masked_output <- embedding(padded_inputs)  masked_output$`_keras_mask` ## tf.Tensor( ## [[ True  True  True False False False] ##  [ True  True  True  True  True False] ##  [ True  True  True  True  True  True]], shape=(3, 6), dtype=bool) masking_layer <- layer_masking() # Simulate the embedding lookup by expanding the 2D input to 3D, # with embedding dimension of 10. unmasked_embedding <- op_cast(     op_tile(op_expand_dims(padded_inputs, axis=-1), c(1L, 1L, 10L)),     dtype=\"float32\" )  masked_embedding <- masking_layer(unmasked_embedding) masked_embedding$`_keras_mask` ## tf.Tensor( ## [[ True  True  True False False False] ##  [ True  True  True  True  True False] ##  [ True  True  True  True  True  True]], shape=(3, 6), dtype=bool)"},{"path":"https://keras3.posit.co/dev/articles/understanding_masking_and_padding.html","id":"mask-propagation-in-the-functional-api-and-sequential-api","dir":"Articles","previous_headings":"","what":"Mask propagation in the Functional API and Sequential API","title":"Understanding masking & padding","text":"using Functional API Sequential API, mask generated Embedding Masking layer propagated network layer capable using (example, RNN layers). Keras automatically fetch mask corresponding input pass layer knows use . instance, following Sequential model, LSTM layer automatically receive mask, means ignore padded values: also case following Functional API model:","code":"model <- keras_model_sequential() %>%   layer_embedding(input_dim=5000, output_dim=16, mask_zero=TRUE) %>%   layer_lstm(units=32) inputs <- keras_input(shape = shape(NULL), dtype=\"int32\") outputs <- inputs %>%   layer_embedding(input_dim=5000, output_dim=16, mask_zero=TRUE) %>%   layer_lstm(units=32)  model <- keras_model(inputs, outputs)"},{"path":"https://keras3.posit.co/dev/articles/understanding_masking_and_padding.html","id":"passing-mask-tensors-directly-to-layers","dir":"Articles","previous_headings":"","what":"Passing mask tensors directly to layers","title":"Understanding masking & padding","text":"Layers can handle masks (LSTM layer) mask argument call method. Meanwhile, layers produce mask (e.g. Embedding) expose compute_mask(input, previous_mask) method can call. Thus, can pass output compute_mask() method mask-producing layer call method mask-consuming layer, like :","code":"MyLayer <- new_layer_class(   \"MyLayer\",   initialize = function(...) {     super$initialize(...)     self$embedding <- layer_embedding(       input_dim=5000, output_dim=16, mask_zero=TRUE     )     self$lstm <- layer_lstm(units=32)   },   call = function(inputs) {     inputs %>%       self$embedding() %>%       # Note that you could also prepare a `mask` tensor manually.       # It only needs to be a boolean tensor       # with the right shape, i.e. (batch_size, timesteps).       self$lstm(mask=self$embedding$compute_mask(inputs))   } )  layer <- MyLayer() x <- random_integer(c(32, 10), 0, 100) layer(x) ## tf.Tensor( ## [[ 0.00130048 -0.00113367 -0.00715671 ... -0.00107615 -0.00162071 ##    0.00135018] ##  [-0.004185    0.00726349  0.00520932 ...  0.00119117  0.00230441 ##    0.00174123] ##  [-0.00537032 -0.00164898 -0.00238435 ... -0.00154158 -0.0038603 ##   -0.00105811] ##  ... ##  [ 0.00622133 -0.00905907 -0.00599518 ...  0.00025823 -0.00142478 ##   -0.00125036] ##  [-0.00523904  0.00336683 -0.00299453 ...  0.00876719  0.00172074 ##    0.00903089] ##  [-0.00393721  0.00058538  0.00503809 ... -0.00203075  0.00325885 ##   -0.00299755]], shape=(32, 32), dtype=float32)"},{"path":"https://keras3.posit.co/dev/articles/understanding_masking_and_padding.html","id":"supporting-masking-in-your-custom-layers","dir":"Articles","previous_headings":"","what":"Supporting masking in your custom layers","title":"Understanding masking & padding","text":"Sometimes, may need write layers generate mask (like Embedding), layers need modify current mask. instance, layer produces tensor different time dimension input, Concatenate layer concatenates time dimension, need modify current mask downstream layers able properly take masked timesteps account. , layer implement layer.compute_mask() method, produces new mask given input current mask. example TemporalSplit layer needs modify current mask. another example CustomEmbedding layer capable generating mask input values: Note: details format limitations related masking, see serialization guide.","code":"TemporalSplit <- new_layer_class(   \"TemporalSplit\",   call = function(inputs) {     # Expect the input to be 3D and mask to be 2D, split the input tensor into 2     # subtensors along the time axis (axis 1).     op_split(inputs, 2, axis=2)   },   compute_mask = function(inputs, mask = NULL) {     # Also split the mask into 2 if it presents.     if (!is.null(mask)) {       op_split(mask, 2, axis=2)     } else {       NULL     }   } )  c(first_half, second_half) %<-% TemporalSplit(masked_embedding) first_half$`_keras_mask` ## tf.Tensor( ## [[ True  True  True] ##  [ True  True  True] ##  [ True  True  True]], shape=(3, 3), dtype=bool) second_half$`_keras_mask` ## tf.Tensor( ## [[False False False] ##  [ True  True False] ##  [ True  True  True]], shape=(3, 3), dtype=bool) CustomEmbedding <- new_layer_class(   \"CustomEmbedding\",   initialize = function(input_dim, output_dim, mask_zero=FALSE, ...) {     super$initialize(...)     self$input_dim <- as.integer(input_dim)     self$output_dim <- as.integer(output_dim)     self$mask_zero <- mask_zero   },   build = function(input_shape) {     self$embeddings <- self$add_weight(       shape=c(self$input_dim, self$output_dim),       initializer=\"random_normal\",       dtype=\"float32\"     )   },   call = function(inputs) {     inputs <- op_cast(inputs, \"int32\")     op_take(self$embeddings, inputs)   },   compute_mask = function(inputs, mask=NULL) {     if (!self$mask_zero) {       NULL     } else {       op_not_equal(inputs, 0)     }   } )  layer <- CustomEmbedding(input_dim = 10, output_dim = 32, mask_zero=TRUE) x <- random_integer(c(3, 10), 0, 9)  y <- layer(x) mask <- layer$compute_mask(x)  mask ## tf.Tensor( ## [[ True  True  True  True  True  True  True  True  True  True] ##  [ True  True  True  True  True  True  True  True False  True] ##  [ True  True  True  True False  True  True  True  True  True]], shape=(3, 10), dtype=bool)"},{"path":"https://keras3.posit.co/dev/articles/understanding_masking_and_padding.html","id":"opting-in-to-mask-propagation-on-compatible-layers","dir":"Articles","previous_headings":"","what":"Opting-in to mask propagation on compatible layers","title":"Understanding masking & padding","text":"layers don’t modify time dimension, don’t need modify current mask. However, may still want able propagate current mask, unchanged, next layer. opt-behavior. default, custom layer destroy current mask (since framework way tell whether propagating mask safe ). custom layer modify time dimension, want able propagate current input mask, set self.supports_masking = True layer constructor. case, default behavior compute_mask() just pass current mask . ’s example layer whitelisted mask propagation: can now use custom layer -mask-generating layer (like Embedding) mask-consuming layer (like LSTM), pass mask along reaches mask-consuming layer.","code":"MyActivation <- new_layer_class(   \"MyActivation\",   initialize = function(...) {     super$initialize(...)     self$supports_masking <- TRUE   },   call = function(inputs) {     op_relu(inputs)   } ) inputs <- keras_input(shape = shape(NULL), dtype=\"int32\") outputs <- inputs %>%   layer_embedding(input_dim=5000, output_dim=16, mask_zero=TRUE) %>%   MyActivation() %>%   layer_lstm(units=32)  model <- keras_model(inputs, outputs) y <- model(random_integer(c(32, 100), 0, 5000))"},{"path":"https://keras3.posit.co/dev/articles/understanding_masking_and_padding.html","id":"writing-layers-that-need-mask-information","dir":"Articles","previous_headings":"","what":"Writing layers that need mask information","title":"Understanding masking & padding","text":"layers mask consumers: accept mask argument call use determine whether skip certain time steps. write layer, can simply add mask=None argument call signature. mask associated inputs passed layer whenever available. ’s simple example : layer computes softmax time dimension (axis 1) input sequence, discarding masked timesteps.","code":"TemporalSoftmax <- new_layer_class(   \"TemporalSoftmax\",   initialize = function(...) {     super$initialize(...)     self$supports_masking <- TRUE   },   call = function(inputs, mask=NULL) {     if (is.null(mask)) {       stop(\"`TemporalSoftmax` layer requires a previous layer to support masking.\")     }     broadcast_float_mask <- op_expand_dims(op_cast(mask, \"float32\"), -1)     inputs_exp <- op_exp(inputs) * broadcast_float_mask     inputs_sum <- op_sum(inputs_exp * broadcast_float_mask, axis=-1, keepdims=TRUE)     inputs_exp / inputs_sum   } )  inputs <- keras_input(shape = shape(NULL), dtype=\"int32\") outputs <- inputs %>%   layer_embedding(input_dim=10, output_dim=32, mask_zero=TRUE) %>%   layer_dense(1) %>%   TemporalSoftmax()  model <- keras_model(inputs, outputs) y <- model(random_integer(c(32, 100), 0, 10))"},{"path":"https://keras3.posit.co/dev/articles/understanding_masking_and_padding.html","id":"summary","dir":"Articles","previous_headings":"","what":"Summary","title":"Understanding masking & padding","text":"need know padding & masking Keras. recap: “Masking” layers able know skip / ignore certain timesteps sequence inputs. layers mask-generators: Embedding can generate mask input values (mask_zero=TRUE), can Masking layer. layers mask-consumers: expose mask argument call method. case RNN layers. Functional API Sequential API, mask information propagated automatically. using layers standalone way, can pass mask arguments layers manually. can easily write layers modify current mask, generate new mask, consume mask associated inputs.","code":""},{"path":"https://keras3.posit.co/dev/articles/writing_a_custom_training_loop_in_tensorflow.html","id":"setup","dir":"Articles","previous_headings":"","what":"Setup","title":"Writing a training loop from scratch in TensorFlow","text":"","code":"library(tensorflow, exclude = c(\"shape\", \"set_random_seed\")) library(tfdatasets, exclude = \"shape\") library(keras3)  # This guide can only be run with the TensorFlow backend. use_backend(\"tensorflow\")"},{"path":"https://keras3.posit.co/dev/articles/writing_a_custom_training_loop_in_tensorflow.html","id":"introduction","dir":"Articles","previous_headings":"","what":"Introduction","title":"Writing a training loop from scratch in TensorFlow","text":"Keras provides default training evaluation loops, fit() evaluate(). usage covered guide Training & evaluation built-methods. want customize learning algorithm model still leveraging convenience fit() (instance, train GAN using fit()), can subclass Model class implement train_step() method, called repeatedly fit(). covered guide Customizing happens fit(). Now, want low-level control training & evaluation, write training & evaluation loops scratch. guide .","code":""},{"path":"https://keras3.posit.co/dev/articles/writing_a_custom_training_loop_in_tensorflow.html","id":"using-the-gradienttape-a-first-end-to-end-example","dir":"Articles","previous_headings":"","what":"Using the GradientTape: a first end-to-end example","title":"Writing a training loop from scratch in TensorFlow","text":"Calling model inside GradientTape scope enables retrieve gradients trainable weights layer respect loss value. Using optimizer instance, can use gradients update variables (can retrieve using model$trainable_weights). Let’s consider simple MNIST model: Let’s train using mini-batch gradient custom training loop. First, ’re going need optimizer, loss function, dataset: ’s training loop: open loop iterates epochs epoch, iterate dataset, batches batch, open GradientTape() scope Inside scope, call model (forward pass) compute loss Outside scope, retrieve gradients weights model regard loss Finally, use optimizer update weights model based gradients","code":"get_model <- function() {   inputs <- keras_input(shape = 784, name = \"digits\")   outputs <- inputs |>     layer_dense(units = 64, activation = \"relu\") |>     layer_dense(units = 64, activation = \"relu\") |>     layer_dense(units = 10, name = \"predictions\")   keras_model(inputs = inputs, outputs = outputs) }  model <- get_model() # Instantiate an optimizer. optimizer <- optimizer_adam(learning_rate = 1e-3) # Instantiate a loss function. loss_fn <- loss_sparse_categorical_crossentropy(from_logits = TRUE)  # Prepare the training dataset. batch_size <- 64 c(c(x_train, y_train), c(x_test, y_test)) %<-% dataset_mnist() x_train <- array_reshape(x_train, c(-1, 784)) x_test <- array_reshape(x_test, c(-1, 784))  # Reserve 10,000 samples for validation. val_i <- sample.int(nrow(x_train), 10000) x_val <- x_train[val_i,] y_val <- y_train[val_i] x_train %<>% .[-val_i,] y_train %<>% .[-val_i]  # Prepare the training dataset. train_dataset <- list(x_train, y_train) |>   tensor_slices_dataset() |>   dataset_shuffle(buffer_size = 1024) |>   dataset_batch(batch_size)  # Prepare the validation dataset. val_dataset <- list(x_val, y_val) |>   tensor_slices_dataset() |>   dataset_batch(batch_size) epochs <- 3 for (epoch in seq_len(epochs)) {   cat(\"Start of epoch \", epoch, \"\\n\")    # Iterate over the batches of the dataset.   step <- 0   iterator <- as_iterator(train_dataset)   while (!is.null(batch <- iter_next(iterator))) {     c(x_batch_train, y_batch_train) %<-% batch     step <- step + 1     # Open a GradientTape to record the operations run     # during the forward pass, which enables auto-differentiation.     with(tf$GradientTape() %as% tape, {       # Run the forward pass of the layer.       # The operations that the layer applies       # to its inputs are going to be recorded       # on the GradientTape.       logits <- model(x_batch_train, training = TRUE) # Logits for this minibatch        # Compute the loss value for this minibatch.       loss_value <- loss_fn(y_batch_train, logits)     })      # Use the gradient tape to automatically retrieve     # the gradients of the trainable variables with respect to the loss.     gradients <- tape$gradient(loss_value, model$trainable_weights)      # Run one step of gradient descent by updating     # the value of the variables to minimize the loss.     optimizer$apply(gradients, model$trainable_weights)      # Log every 200 batches.     if (step %% 200 == 0) {       cat(sprintf(\"Training loss (for one batch) at step %d: %.4f\\n\",                   step, loss_value))       cat(sprintf(\"Seen so far: %d samples\\n\", (step * batch_size)))     }   } } ## Start of epoch  1 ## Training loss (for one batch) at step 200: 1.1675 ## Seen so far: 12800 samples ## Training loss (for one batch) at step 400: 1.6450 ## Seen so far: 25600 samples ## Training loss (for one batch) at step 600: 0.8552 ## Seen so far: 38400 samples ## Start of epoch  2 ## Training loss (for one batch) at step 200: 0.6606 ## Seen so far: 12800 samples ## Training loss (for one batch) at step 400: 0.2213 ## Seen so far: 25600 samples ## Training loss (for one batch) at step 600: 0.3709 ## Seen so far: 38400 samples ## Start of epoch  3 ## Training loss (for one batch) at step 200: 0.2968 ## Seen so far: 12800 samples ## Training loss (for one batch) at step 400: 0.4781 ## Seen so far: 25600 samples ## Training loss (for one batch) at step 600: 0.2455 ## Seen so far: 38400 samples"},{"path":"https://keras3.posit.co/dev/articles/writing_a_custom_training_loop_in_tensorflow.html","id":"low-level-handling-of-metrics","dir":"Articles","previous_headings":"","what":"Low-level handling of metrics","title":"Writing a training loop from scratch in TensorFlow","text":"Let’s add metrics monitoring basic loop. can readily reuse built-metrics (custom ones wrote) training loops written scratch. ’s flow: Instantiate metric start loop Call metric$update_state() batch Call metric$result() need display current value metric Call metric$reset_state() need clear state metric (typically end epoch) Let’s use knowledge compute SparseCategoricalAccuracy validation data end epoch: ’s training & evaluation loop:","code":"# Get a fresh model model <- get_model()  # Instantiate an optimizer to train the model. optimizer <- optimizer_adam(learning_rate = 1e-3) # Instantiate a loss function. loss_fn <- loss_sparse_categorical_crossentropy(from_logits = TRUE)  # Prepare the metrics. train_acc_metric <- metric_sparse_categorical_accuracy() val_acc_metric <- metric_sparse_categorical_accuracy() epochs <- 2 time <- Sys.time() for (epoch in seq_len(epochs)) {   cat(\"Start of epoch \", epoch, \"\\n\")    # Iterate over the batches of the dataset.   step <- 0   train_dataset_iterator <- as_iterator(train_dataset)   while (!is.null(train_batch <- iter_next(train_dataset_iterator))) {     c(x_batch_train, y_batch_train) %<-% train_batch     step <- step + 1      with(tf$GradientTape() %as% tape, {       logits <- model(x_batch_train, training = TRUE)       loss_value <- loss_fn(y_batch_train, logits)     })     gradients <- tape$gradient(loss_value, model$trainable_weights)     optimizer$apply(gradients, model$trainable_weights)      # Update training metric.     train_acc_metric$update_state(y_batch_train, logits)     if (step %% 200 == 0) {       cat(sprintf(         \"Training loss (for one batch) at step %d: %.4f\\n\", step, loss_value))       cat(sprintf(\"Seen so far: %d samples \\n\", (step * batch_size)))     }   }    # Display metrics at the end of each epoch.   train_acc <- train_acc_metric$result()   cat(sprintf(\"Training acc over epoch: %.4f\\n\", train_acc))    # Reset training metrics at the end of each epoch   train_acc_metric$reset_state()    # Run a validation loop at the end of each epoch.   iterate(val_dataset, function(val_batch) {     c(x_batch_val, y_batch_val) %<-% val_batch     val_logits <- model(x_batch_val, training = FALSE)     val_acc_metric$update_state(y_batch_val, val_logits)   })   val_acc <- val_acc_metric$result()   val_acc_metric$reset_state()   cat(sprintf(\"Validation acc: %.4f\\n\", val_acc)) } ## Start of epoch  1 ## Training loss (for one batch) at step 200: 1.9564 ## Seen so far: 12800 samples ## Training loss (for one batch) at step 400: 1.3562 ## Seen so far: 25600 samples ## Training loss (for one batch) at step 600: 0.4850 ## Seen so far: 38400 samples ## Training acc over epoch: 0.7836 ## Validation acc: 0.8681 ## Start of epoch  2 ## Training loss (for one batch) at step 200: 0.7679 ## Seen so far: 12800 samples ## Training loss (for one batch) at step 400: 0.5674 ## Seen so far: 25600 samples ## Training loss (for one batch) at step 600: 0.4705 ## Seen so far: 38400 samples ## Training acc over epoch: 0.8838 ## Validation acc: 0.9049 Sys.time() - time ## Time difference of 51.62403 secs"},{"path":"https://keras3.posit.co/dev/articles/writing_a_custom_training_loop_in_tensorflow.html","id":"speeding-up-your-training-step-with-tf_function","dir":"Articles","previous_headings":"","what":"Speeding-up your training step with tf_function()","title":"Writing a training loop from scratch in TensorFlow","text":"default runtime TensorFlow 2 eager execution. , training loop executes eagerly. great debugging, graph compilation definite performance advantage. Describing computation static graph enables framework apply global performance optimizations. impossible framework constrained greedily execute one operation another, knowledge comes next. can compile static graph function takes tensors input. Just add @tf.function decorator , like : Let’s evaluation step: Now, let’s re-run training loop compiled training step: Much faster, isn’t ?","code":"train_step <- tf_function(function(x, y) {   with(tf$GradientTape() %as% tape, {     logits <- model(x, training = TRUE)     loss_value <- loss_fn(y, logits)   })   gradients <- tape$gradient(loss_value, model$trainable_weights)   optimizer$apply(gradients, model$trainable_weights)   train_acc_metric$update_state(y, logits)   invisible(NULL) # return nothing }) test_step <- tf_function(function(x, y) {   val_logits <- model(x, training=FALSE)   val_acc_metric$update_state(y, val_logits)   invisible(NULL) # return nothing }) epochs <- 2 time <- Sys.time() for (epoch in seq_len(epochs)) {   cat(\"Start of epoch \", epoch, \"\\n\")    # Iterate over the batches of the dataset.   step <- 0   while (!is.null(batch <- iter_next(iterator))) {     c(x_batch_train, y_batch_train) %<-% batch     step <- step + 1     train_step(x_batch_train, y_batch_train)      if (step %% 200 == 0) {       cat(sprintf(         \"Training loss (for one batch) at step %d: %.4f\\n\", step, loss_value       ))       cat(sprintf(\"Seen so far: %d samples \\n\", (step * batch_size)))     }   }    # Display metrics at the end of each epoch.   train_acc <- train_acc_metric$result()   cat(sprintf(\"Training acc over epoch: %.4f\\n\", train_acc))     # Reset training metrics at the end of each epoch   train_acc_metric$reset_state()    # Run a validation loop at the end of each epoch.    iterate(val_dataset, function(val_batch) {     c(x_batch_val, y_batch_val) %<-% val_batch     test_step(x_batch_val, y_batch_val)   })    val_acc <- val_acc_metric$result()   val_acc_metric$reset_state()   cat(sprintf(\"Validation acc: %.4f\\n\", val_acc)) } ## Start of epoch  1 ## Training acc over epoch: 0.0000 ## Validation acc: 0.9049 ## Start of epoch  2 ## Training acc over epoch: 0.0000 ## Validation acc: 0.9049 Sys.time() - time ## Time difference of 0.5661066 secs"},{"path":"https://keras3.posit.co/dev/articles/writing_a_custom_training_loop_in_tensorflow.html","id":"low-level-handling-of-losses-tracked-by-the-model","dir":"Articles","previous_headings":"","what":"Low-level handling of losses tracked by the model","title":"Writing a training loop from scratch in TensorFlow","text":"Layers models recursively track losses created forward pass layers call self$add_loss(value). resulting list scalar loss values available via property model$losses end forward pass. want using loss components, sum add main loss training step. Consider layer, creates activity regularization loss: Let’s build really simple model uses : ’s training step look like now:","code":"layer_activity_regularization <- Layer(   \"ActivityRegularizationLayer\",   call = function(inputs) {     self$add_loss(0.1 * op_mean(inputs))     inputs   } ) inputs <- keras_input(shape = 784, name=\"digits\") outputs <- inputs |>   layer_dense(units = 64, activation = \"relu\") |>   layer_activity_regularization() |>   layer_dense(units = 64, activation = \"relu\") |>   layer_dense(units = 10, name = \"predictions\") model <- keras_model(inputs = inputs, outputs = outputs) train_step <- tf_function(function(x, y) {   with(tf$GradientTape() %as% tape, {     logits <- model(x, training = TRUE)     loss_value <- Reduce(`+`, c(loss_fn(y, logits),                                 model$losses))   })   gradients <- tape$gradient(loss_value, model$trainable_weights)   optimizer$apply(gradients, model$trainable_weights)   train_acc_metric$update_state(y, logits)   invisible(NULL) })"},{"path":"https://keras3.posit.co/dev/articles/writing_a_custom_training_loop_in_tensorflow.html","id":"summary","dir":"Articles","previous_headings":"","what":"Summary","title":"Writing a training loop from scratch in TensorFlow","text":"Now know everything know using built-training loops writing scratch. conclude, ’s simple end--end example ties together everything ’ve learned guide: DCGAN trained MNIST digits.","code":""},{"path":"https://keras3.posit.co/dev/articles/writing_a_custom_training_loop_in_tensorflow.html","id":"end-to-end-example-a-gan-training-loop-from-scratch","dir":"Articles","previous_headings":"","what":"End-to-end example: a GAN training loop from scratch","title":"Writing a training loop from scratch in TensorFlow","text":"may familiar Generative Adversarial Networks (GANs). GANs can generate new images look almost real, learning latent distribution training dataset images (“latent space” images). GAN made two parts: “generator” model maps points latent space points image space, “discriminator” model, classifier can tell difference real images (training dataset) fake images (output generator network). GAN training loop looks like : Train discriminator. Sample batch random points latent space. Turn points fake images via “generator” model. Get batch real images combine generated images. Train “discriminator” model classify generated vs. real images. Train generator. Sample random points latent space. Turn points fake images via “generator” network. Get batch real images combine generated images. Train “generator” model “fool” discriminator classify fake images real. much detailed overview GANs works, see Deep Learning Python. Let’s implement training loop. First, create discriminator meant classify fake vs real digits: let’s create generator network, turns latent vectors outputs shape (28, 28, 1) (representing MNIST digits): ’s key bit: training loop. can see quite straightforward. training step function takes 17 lines. Let’s train GAN, repeatedly calling train_step batches images. Since discriminator generator convnets, ’re going want run code GPU. ’s ! ’ll get nice-looking fake MNIST digits just ~30s training GPU.","code":"# Create the discriminator discriminator <-   keras_model_sequential(name = \"discriminator\",                          input_shape = c(28, 28, 1)) |>   layer_conv_2d(filters = 64, kernel_size = c(3, 3),                 strides = c(2, 2), padding = \"same\") |>   layer_activation_leaky_relu(negative_slope = 0.2) |>   layer_conv_2d(filters = 128, kernel_size = c(3, 3),                 strides = c(2, 2), padding = \"same\") |>   layer_activation_leaky_relu(negative_slope = 0.2) |>   layer_global_max_pooling_2d() |>   layer_dense(units = 1)  summary(discriminator) ## Model: \"discriminator\" ## ┏━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━┓ ## ┃ Layer (type)                    ┃ Output Shape           ┃       Param # ┃ ## ┡━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━┩ ## │ conv2d (Conv2D)                 │ (None, 14, 14, 64)     │           640 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ leaky_re_lu (LeakyReLU)         │ (None, 14, 14, 64)     │             0 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ conv2d_1 (Conv2D)               │ (None, 7, 7, 128)      │        73,856 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ leaky_re_lu_1 (LeakyReLU)       │ (None, 7, 7, 128)      │             0 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ global_max_pooling2d            │ (None, 128)            │             0 │ ## │ (GlobalMaxPooling2D)            │                        │               │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ dense_6 (Dense)                 │ (None, 1)              │           129 │ ## └─────────────────────────────────┴────────────────────────┴───────────────┘ ##  Total params: 74,625 (291.50 KB) ##  Trainable params: 74,625 (291.50 KB) ##  Non-trainable params: 0 (0.00 B) latent_dim <- 128L  generator <-   keras_model_sequential(name = \"generator\",                          input_shape = latent_dim) |>   layer_dense(7 * 7 * 128) |>   layer_activation_leaky_relu(negative_slope = 0.2) |>   layer_reshape(target_shape = c(7, 7, 128)) |>   layer_conv_2d_transpose(filters = 128, kernel_size = c(4, 4),                           strides = c(2, 2), padding = \"same\") |>   layer_activation_leaky_relu(negative_slope = 0.2) |>   layer_conv_2d_transpose(filters = 128, kernel_size = c(4, 4),                           strides = c(2, 2), padding = \"same\") |>   layer_activation_leaky_relu(negative_slope = 0.2) |>   layer_conv_2d(filters = 1, kernel_size = c(7, 7), padding = \"same\",                 activation = \"sigmoid\")  summary(generator) ## Model: \"generator\" ## ┏━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━┓ ## ┃ Layer (type)                    ┃ Output Shape           ┃       Param # ┃ ## ┡━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━┩ ## │ dense_7 (Dense)                 │ (None, 6272)           │       809,088 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ leaky_re_lu_2 (LeakyReLU)       │ (None, 6272)           │             0 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ reshape (Reshape)               │ (None, 7, 7, 128)      │             0 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ conv2d_transpose                │ (None, 14, 14, 128)    │       262,272 │ ## │ (Conv2DTranspose)               │                        │               │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ leaky_re_lu_3 (LeakyReLU)       │ (None, 14, 14, 128)    │             0 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ conv2d_transpose_1              │ (None, 28, 28, 128)    │       262,272 │ ## │ (Conv2DTranspose)               │                        │               │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ leaky_re_lu_4 (LeakyReLU)       │ (None, 28, 28, 128)    │             0 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ conv2d_2 (Conv2D)               │ (None, 28, 28, 1)      │         6,273 │ ## └─────────────────────────────────┴────────────────────────┴───────────────┘ ##  Total params: 1,339,905 (5.11 MB) ##  Trainable params: 1,339,905 (5.11 MB) ##  Non-trainable params: 0 (0.00 B) # Instantiate one optimizer for the discriminator and another for the generator. d_optimizer <- optimizer_adam(learning_rate = 0.0003) g_optimizer <- optimizer_adam(learning_rate = 0.0004)  # Instantiate a loss function. loss_fn <- loss_binary_crossentropy(from_logits = TRUE)  train_step <- tf_function(function(real_images) {   # Sample random points in the latent space   c(batch_size, ...) %<-% op_shape(real_images)   random_latent_vectors <-     tf$random$normal(shape(batch_size, latent_dim))    # Decode them to fake images   generated_images <- generator(random_latent_vectors)    # Combine them with real images   combined_images <- tf$concat(list(generated_images, real_images),                                axis = 0L)    # Assemble labels discriminating real from fake images   labels <- tf$concat(list(tf$ones(shape(batch_size, 1)),                            tf$zeros(shape(batch_size, 1))),                       axis = 0L)    # Add random noise to the labels - important trick!   labels %<>% `+`(tf$random$uniform(tf$shape(.), maxval = 0.05))    # Train the discriminator   with(tf$GradientTape() %as% tape, {     predictions <- discriminator(combined_images)     d_loss <- loss_fn(labels, predictions)   })   grads <- tape$gradient(d_loss, discriminator$trainable_weights)   d_optimizer$apply(grads, discriminator$trainable_weights)    # Sample random points in the latent space   random_latent_vectors <-     tf$random$normal(shape(batch_size, latent_dim))    # Assemble labels that say \"all real images\"   misleading_labels <- tf$zeros(shape(batch_size, 1))    # Train the generator (note that we should *not* update the weights   # of the discriminator)!   with(tf$GradientTape() %as% tape, {     predictions <- discriminator(generator(random_latent_vectors))     g_loss <- loss_fn(misleading_labels, predictions)   })    grads <- tape$gradient(g_loss, generator$trainable_weights)   g_optimizer$apply(grads, generator$trainable_weights)    list(d_loss, g_loss, generated_images) }) # Prepare the dataset. We use both the training & test MNIST digits. batch_size <- 64 c(c(x_train, .), c(x_test, .)) %<-% dataset_mnist() all_digits <- op_concatenate(list(x_train, x_test)) all_digits <- op_reshape(all_digits, c(-1, 28, 28, 1)) dataset <- all_digits |>   tensor_slices_dataset() |>   dataset_map(\\(x) op_cast(x, \"float32\") / 255) |>   dataset_shuffle(buffer_size = 1024) |>   dataset_batch(batch_size = batch_size)  epochs <- 1 # In practice you need at least 20 epochs to generate nice digits. save_dir <- \"./\"  for (epoch in seq_len(epochs)) {   cat(\"Start epoch: \", epoch, \"\\n\")   step <- 0   train_iterator <- as_iterator(dataset)   while (!is.null(real_images <- iter_next(train_iterator))) {     step <- step + 1     # Train the discriminator & generator on one batch of real images.     c(d_loss, g_loss, generated_images) %<-% train_step(real_images)      # Logging.     if (step %% 200 == 0) {       # Print metrics       cat(sprintf(\"discriminator loss at step %d: %.2f\\n\", step, d_loss))       cat(sprintf(\"adversarial loss at step %d: %.2f\\n\", step, g_loss))     }      # To limit execution time we stop after 10 steps.     # Remove the lines below to actually train the model!     if (step > 10)       break   } } ## Start epoch:  1"},{"path":"https://keras3.posit.co/dev/articles/writing_your_own_callbacks.html","id":"introduction","dir":"Articles","previous_headings":"","what":"Introduction","title":"Writing your own callbacks","text":"callback powerful tool customize behavior Keras model training, evaluation, inference. Examples include keras.callbacks.TensorBoard visualize training progress results TensorBoard, keras.callbacks.ModelCheckpoint periodically save model training. guide, learn Keras callback , can , can build . provide demos simple callback applications get started.","code":""},{"path":"https://keras3.posit.co/dev/articles/writing_your_own_callbacks.html","id":"setup","dir":"Articles","previous_headings":"","what":"Setup","title":"Writing your own callbacks","text":"","code":"library(keras3)"},{"path":"https://keras3.posit.co/dev/articles/writing_your_own_callbacks.html","id":"keras-callbacks-overview","dir":"Articles","previous_headings":"","what":"Keras callbacks overview","title":"Writing your own callbacks","text":"callbacks subclass keras.callbacks.Callback class, override set methods called various stages training, testing, predicting. Callbacks useful get view internal states statistics model training. can pass list callbacks (keyword argument callbacks) following model methods: fit() evaluate() predict()","code":""},{"path":[]},{"path":[]},{"path":"https://keras3.posit.co/dev/articles/writing_your_own_callbacks.html","id":"on_traintestpredict_beginlogs-null","dir":"Articles","previous_headings":"An overview of callback methods > Global methods","what":"on_(train|test|predict)_begin(logs = NULL)","title":"Writing your own callbacks","text":"Called beginning fit/evaluate/predict.","code":""},{"path":"https://keras3.posit.co/dev/articles/writing_your_own_callbacks.html","id":"on_traintestpredict_endlogs-null","dir":"Articles","previous_headings":"An overview of callback methods > Global methods","what":"on_(train|test|predict)_end(logs = NULL)","title":"Writing your own callbacks","text":"Called end fit/evaluate/predict.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/articles/writing_your_own_callbacks.html","id":"on_traintestpredict_batch_beginbatch-logs-null","dir":"Articles","previous_headings":"An overview of callback methods > Batch-level methods for training/testing/predicting","what":"on_(train|test|predict)_batch_begin(batch, logs = NULL)","title":"Writing your own callbacks","text":"Called right processing batch training/testing/predicting.","code":""},{"path":"https://keras3.posit.co/dev/articles/writing_your_own_callbacks.html","id":"on_traintestpredict_batch_endbatch-logs-null","dir":"Articles","previous_headings":"An overview of callback methods > Batch-level methods for training/testing/predicting","what":"on_(train|test|predict)_batch_end(batch, logs = NULL)","title":"Writing your own callbacks","text":"Called end training/testing/predicting batch. Within method, logs named list containing metrics results.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/articles/writing_your_own_callbacks.html","id":"on_epoch_beginepoch-logs-null","dir":"Articles","previous_headings":"An overview of callback methods > Epoch-level methods (training only)","what":"on_epoch_begin(epoch, logs = NULL)","title":"Writing your own callbacks","text":"Called beginning epoch training.","code":""},{"path":"https://keras3.posit.co/dev/articles/writing_your_own_callbacks.html","id":"on_epoch_endepoch-logs-null","dir":"Articles","previous_headings":"An overview of callback methods > Epoch-level methods (training only)","what":"on_epoch_end(epoch, logs = NULL)","title":"Writing your own callbacks","text":"Called end epoch training.","code":""},{"path":"https://keras3.posit.co/dev/articles/writing_your_own_callbacks.html","id":"a-basic-example","dir":"Articles","previous_headings":"","what":"A basic example","title":"Writing your own callbacks","text":"Let’s take look concrete example. get started, let’s import tensorflow define simple Sequential Keras model: , load MNIST data training testing Keras datasets API: Now, define simple custom callback logs: fit/evaluate/predict starts & ends epoch starts & ends training batch starts & ends evaluation (test) batch starts & ends inference (prediction) batch starts & ends Let’s try :","code":"# Define the Keras model to add callbacks to get_model <- function() {   model <- keras_model_sequential()   model |> layer_dense(units = 1)   model |> compile(     optimizer = optimizer_rmsprop(learning_rate = 0.1),     loss = \"mean_squared_error\",     metrics = \"mean_absolute_error\"   )   model } # Load example MNIST data and pre-process it mnist <- dataset_mnist()  flatten_and_rescale <- function(x) {   x <- array_reshape(x, c(-1, 784))   x <- x / 255   x }  mnist$train$x <- flatten_and_rescale(mnist$train$x) mnist$test$x  <- flatten_and_rescale(mnist$test$x)  # limit to 1000 samples n <- 1000 mnist$train$x <- mnist$train$x[1:n,] mnist$train$y <- mnist$train$y[1:n] mnist$test$x  <- mnist$test$x[1:n,] mnist$test$y  <- mnist$test$y[1:n] show <- function(msg, logs) {   cat(glue::glue(msg, .envir = parent.frame()),       \"got logs: \", sep = \"; \")   str(logs); cat(\"\\n\") }  callback_custom <- Callback(   \"CustomCallback\",   on_train_begin         = \\(logs = NULL)        show(\"Starting training\", logs),   on_epoch_begin         = \\(epoch, logs = NULL) show(\"Start epoch {epoch} of training\", logs),   on_train_batch_begin   = \\(batch, logs = NULL) show(\"...Training: start of batch {batch}\", logs),   on_train_batch_end     = \\(batch, logs = NULL) show(\"...Training: end of batch {batch}\",  logs),   on_epoch_end           = \\(epoch, logs = NULL) show(\"End epoch {epoch} of training\", logs),   on_train_end           = \\(logs = NULL)        show(\"Stop training\", logs),     on_test_begin          = \\(logs = NULL)        show(\"Start testing\", logs),   on_test_batch_begin    = \\(batch, logs = NULL) show(\"...Evaluating: start of batch {batch}\", logs),   on_test_batch_end      = \\(batch, logs = NULL) show(\"...Evaluating: end of batch {batch}\", logs),   on_test_end            = \\(logs = NULL)        show(\"Stop testing\", logs),    on_predict_begin       = \\(logs = NULL)        show(\"Start predicting\", logs),   on_predict_end         = \\(logs = NULL)        show(\"Stop predicting\", logs),   on_predict_batch_begin = \\(batch, logs = NULL) show(\"...Predicting: start of batch {batch}\", logs),   on_predict_batch_end   = \\(batch, logs = NULL) show(\"...Predicting: end of batch {batch}\", logs), ) model <- get_model() model |> fit(   mnist$train$x, mnist$train$y,   batch_size = 128,   epochs = 2,   verbose = 0,   validation_split = 0.5,   callbacks = list(callback_custom()) ) ## Starting training; got logs:  Named list() ## ## Start epoch 1 of training; got logs:  Named list() ## ## ...Training: start of batch 1; got logs:  Named list() ## ## ...Training: end of batch 1; got logs: List of 2 ##  $ loss               : num 25.9 ##  $ mean_absolute_error: num 4.19 ## ## ...Training: start of batch 2; got logs:  Named list() ## ## ...Training: end of batch 2; got logs: List of 2 ##  $ loss               : num 433 ##  $ mean_absolute_error: num 15.5 ## ## ...Training: start of batch 3; got logs:  Named list() ## ## ...Training: end of batch 3; got logs: List of 2 ##  $ loss               : num 297 ##  $ mean_absolute_error: num 11.8 ## ## ...Training: start of batch 4; got logs:  Named list() ## ## ...Training: end of batch 4; got logs: List of 2 ##  $ loss               : num 231 ##  $ mean_absolute_error: num 9.68 ## ## Start testing; got logs:  Named list() ## ## ...Evaluating: start of batch 1; got logs:  Named list() ## ## ...Evaluating: end of batch 1; got logs: List of 2 ##  $ loss               : num 8.1 ##  $ mean_absolute_error: num 2.3 ## ## ...Evaluating: start of batch 2; got logs:  Named list() ## ## ...Evaluating: end of batch 2; got logs: List of 2 ##  $ loss               : num 7.58 ##  $ mean_absolute_error: num 2.23 ## ## ...Evaluating: start of batch 3; got logs:  Named list() ## ## ...Evaluating: end of batch 3; got logs: List of 2 ##  $ loss               : num 7.38 ##  $ mean_absolute_error: num 2.21 ## ## ...Evaluating: start of batch 4; got logs:  Named list() ## ## ...Evaluating: end of batch 4; got logs: List of 2 ##  $ loss               : num 7.3 ##  $ mean_absolute_error: num 2.21 ## ## Stop testing; got logs: List of 2 ##  $ loss               : num 7.3 ##  $ mean_absolute_error: num 2.21 ## ## End epoch 1 of training; got logs: List of 4 ##  $ loss                   : num 231 ##  $ mean_absolute_error    : num 9.68 ##  $ val_loss               : num 7.3 ##  $ val_mean_absolute_error: num 2.21 ## ## Start epoch 2 of training; got logs:  Named list() ## ## ...Training: start of batch 1; got logs:  Named list() ## ## ...Training: end of batch 1; got logs: List of 2 ##  $ loss               : num 7.44 ##  $ mean_absolute_error: num 2.27 ## ## ...Training: start of batch 2; got logs:  Named list() ## ## ...Training: end of batch 2; got logs: List of 2 ##  $ loss               : num 6.81 ##  $ mean_absolute_error: num 2.16 ## ## ...Training: start of batch 3; got logs:  Named list() ## ## ...Training: end of batch 3; got logs: List of 2 ##  $ loss               : num 6.12 ##  $ mean_absolute_error: num 2.06 ## ## ...Training: start of batch 4; got logs:  Named list() ## ## ...Training: end of batch 4; got logs: List of 2 ##  $ loss               : num 6.08 ##  $ mean_absolute_error: num 2.04 ## ## Start testing; got logs:  Named list() ## ## ...Evaluating: start of batch 1; got logs:  Named list() ## ## ...Evaluating: end of batch 1; got logs: List of 2 ##  $ loss               : num 5.54 ##  $ mean_absolute_error: num 1.92 ## ## ...Evaluating: start of batch 2; got logs:  Named list() ## ## ...Evaluating: end of batch 2; got logs: List of 2 ##  $ loss               : num 5.31 ##  $ mean_absolute_error: num 1.87 ## ## ...Evaluating: start of batch 3; got logs:  Named list() ## ## ...Evaluating: end of batch 3; got logs: List of 2 ##  $ loss               : num 5.11 ##  $ mean_absolute_error: num 1.8 ## ## ...Evaluating: start of batch 4; got logs:  Named list() ## ## ...Evaluating: end of batch 4; got logs: List of 2 ##  $ loss               : num 5.15 ##  $ mean_absolute_error: num 1.82 ## ## Stop testing; got logs: List of 2 ##  $ loss               : num 5.15 ##  $ mean_absolute_error: num 1.82 ## ## End epoch 2 of training; got logs: List of 4 ##  $ loss                   : num 6.08 ##  $ mean_absolute_error    : num 2.04 ##  $ val_loss               : num 5.15 ##  $ val_mean_absolute_error: num 1.82 ## ## Stop training; got logs: List of 4 ##  $ loss                   : num 6.08 ##  $ mean_absolute_error    : num 2.04 ##  $ val_loss               : num 5.15 ##  $ val_mean_absolute_error: num 1.82 res <- model |> evaluate(   mnist$test$x, mnist$test$y,   batch_size = 128, verbose = 0,   callbacks = list(callback_custom()) ) ## Start testing; got logs:  Named list() ## ## ...Evaluating: start of batch 1; got logs:  Named list() ## ## ...Evaluating: end of batch 1; got logs: List of 2 ##  $ loss               : num 5.2 ##  $ mean_absolute_error: num 1.84 ## ## ...Evaluating: start of batch 2; got logs:  Named list() ## ## ...Evaluating: end of batch 2; got logs: List of 2 ##  $ loss               : num 4.62 ##  $ mean_absolute_error: num 1.73 ## ## ...Evaluating: start of batch 3; got logs:  Named list() ## ## ...Evaluating: end of batch 3; got logs: List of 2 ##  $ loss               : num 4.61 ##  $ mean_absolute_error: num 1.74 ## ## ...Evaluating: start of batch 4; got logs:  Named list() ## ## ...Evaluating: end of batch 4; got logs: List of 2 ##  $ loss               : num 4.65 ##  $ mean_absolute_error: num 1.75 ## ## ...Evaluating: start of batch 5; got logs:  Named list() ## ## ...Evaluating: end of batch 5; got logs: List of 2 ##  $ loss               : num 4.84 ##  $ mean_absolute_error: num 1.77 ## ## ...Evaluating: start of batch 6; got logs:  Named list() ## ## ...Evaluating: end of batch 6; got logs: List of 2 ##  $ loss               : num 4.76 ##  $ mean_absolute_error: num 1.76 ## ## ...Evaluating: start of batch 7; got logs:  Named list() ## ## ...Evaluating: end of batch 7; got logs: List of 2 ##  $ loss               : num 4.74 ##  $ mean_absolute_error: num 1.76 ## ## ...Evaluating: start of batch 8; got logs:  Named list() ## ## ...Evaluating: end of batch 8; got logs: List of 2 ##  $ loss               : num 4.67 ##  $ mean_absolute_error: num 1.75 ## ## Stop testing; got logs: List of 2 ##  $ loss               : num 4.67 ##  $ mean_absolute_error: num 1.75 res <- model |> predict(   mnist$test$x,   batch_size = 128, verbose = 0,   callbacks = list(callback_custom()) ) ## Start predicting; got logs:  Named list() ## ## ...Predicting: start of batch 1; got logs:  Named list() ## ## ...Predicting: end of batch 1; got logs: List of 1 ##  $ outputs:<tf.Tensor: shape=(128, 1), dtype=float32, numpy=…> ## ## ...Predicting: start of batch 2; got logs:  Named list() ## ## ...Predicting: end of batch 2; got logs: List of 1 ##  $ outputs:<tf.Tensor: shape=(128, 1), dtype=float32, numpy=…> ## ## ...Predicting: start of batch 3; got logs:  Named list() ## ## ...Predicting: end of batch 3; got logs: List of 1 ##  $ outputs:<tf.Tensor: shape=(128, 1), dtype=float32, numpy=…> ## ## ...Predicting: start of batch 4; got logs:  Named list() ## ## ...Predicting: end of batch 4; got logs: List of 1 ##  $ outputs:<tf.Tensor: shape=(128, 1), dtype=float32, numpy=…> ## ## ...Predicting: start of batch 5; got logs:  Named list() ## ## ...Predicting: end of batch 5; got logs: List of 1 ##  $ outputs:<tf.Tensor: shape=(128, 1), dtype=float32, numpy=…> ## ## ...Predicting: start of batch 6; got logs:  Named list() ## ## ...Predicting: end of batch 6; got logs: List of 1 ##  $ outputs:<tf.Tensor: shape=(128, 1), dtype=float32, numpy=…> ## ## ...Predicting: start of batch 7; got logs:  Named list() ## ## ...Predicting: end of batch 7; got logs: List of 1 ##  $ outputs:<tf.Tensor: shape=(128, 1), dtype=float32, numpy=…> ## ## ...Predicting: start of batch 8; got logs:  Named list() ## ## ...Predicting: end of batch 8; got logs: List of 1 ##  $ outputs:<tf.Tensor: shape=(104, 1), dtype=float32, numpy=…> ## ## Stop predicting; got logs:  Named list()"},{"path":"https://keras3.posit.co/dev/articles/writing_your_own_callbacks.html","id":"usage-of-logs-list","dir":"Articles","previous_headings":"A basic example","what":"Usage of logs list","title":"Writing your own callbacks","text":"logs named list contains loss value, metrics end batch epoch. Example includes loss mean absolute error. information callbacks, can check Keras callback API documentation.","code":"callback_print_loss_and_mae <- Callback(   \"LossAndErrorPrintingCallback\",    on_train_batch_end = function(batch, logs = NULL)     cat(sprintf(\"Up to batch %i, the average loss is %7.2f.\\n\",                 batch,  logs$loss)),    on_test_batch_end = function(batch, logs = NULL)     cat(sprintf(\"Up to batch %i, the average loss is %7.2f.\\n\",                 batch, logs$loss)),    on_epoch_end = function(epoch, logs = NULL)     cat(sprintf(       \"The average loss for epoch %2i is %9.2f and mean absolute error is %7.2f.\\n\",       epoch, logs$loss, logs$mean_absolute_error     )) )   model <- get_model() model |> fit(   mnist$train$x, mnist$train$y,   epochs = 2, verbose = 0, batch_size = 128,   callbacks = list(callback_print_loss_and_mae()) ) ## Up to batch 1, the average loss is   25.12. ## Up to batch 2, the average loss is  398.92. ## Up to batch 3, the average loss is  274.04. ## Up to batch 4, the average loss is  208.32. ## Up to batch 5, the average loss is  168.15. ## Up to batch 6, the average loss is  141.31. ## Up to batch 7, the average loss is  122.19. ## Up to batch 8, the average loss is  110.05. ## The average loss for epoch  1 is    110.05 and mean absolute error is    5.79. ## Up to batch 1, the average loss is    4.71. ## Up to batch 2, the average loss is    4.74. ## Up to batch 3, the average loss is    4.81. ## Up to batch 4, the average loss is    5.07. ## Up to batch 5, the average loss is    5.08. ## Up to batch 6, the average loss is    5.09. ## Up to batch 7, the average loss is    5.19. ## Up to batch 8, the average loss is    5.51. ## The average loss for epoch  2 is      5.51 and mean absolute error is    1.90. res = model |> evaluate(   mnist$test$x, mnist$test$y,   verbose = 0, batch_size = 128,   callbacks = list(callback_print_loss_and_mae()) ) ## Up to batch 1, the average loss is   15.86. ## Up to batch 2, the average loss is   16.13. ## Up to batch 3, the average loss is   16.02. ## Up to batch 4, the average loss is   16.11. ## Up to batch 5, the average loss is   16.23. ## Up to batch 6, the average loss is   16.68. ## Up to batch 7, the average loss is   16.61. ## Up to batch 8, the average loss is   16.54."},{"path":"https://keras3.posit.co/dev/articles/writing_your_own_callbacks.html","id":"usage-of-selfmodel-attribute","dir":"Articles","previous_headings":"","what":"Usage of self$model attribute","title":"Writing your own callbacks","text":"addition receiving log information one methods called, callbacks access model associated current round training/evaluation/inference: self$model. things can self$model callback: Set self$model$stop_training <- TRUE immediately interrupt training. Mutate hyperparameters optimizer (available self$model$optimizer), self$model$optimizer$learning_rate. Save model period intervals. Record output model |> predict() test samples end epoch, use sanity check training. Extract visualizations intermediate features end epoch, monitor model learning time. etc. Let’s see action couple examples.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/articles/writing_your_own_callbacks.html","id":"early-stopping-at-minimum-loss","dir":"Articles","previous_headings":"Examples of Keras callback applications","what":"Early stopping at minimum loss","title":"Writing your own callbacks","text":"first example shows creation Callback stops training minimum loss reached, setting attribute self$model$stop_training (boolean). Optionally, can provide argument patience specify many epochs wait stopping reached local minimum. callback_early_stopping() provides complete general implementation.","code":"callback_early_stopping_at_min_loss <- Callback(   \"EarlyStoppingAtMinLoss\",   `__doc__` =     \"Stop training when the loss is at its min, i.e. the loss stops decreasing.      Arguments:         patience: Number of epochs to wait after min has been hit. After this         number of no improvement, training stops.     \",    initialize = function(patience = 0) {     super$initialize()     self$patience <- patience     # best_weights to store the weights at which the minimum loss occurs.     self$best_weights <- NULL   },    on_train_begin = function(logs = NULL) {     # The number of epoch it has waited when loss is no longer minimum.     self$wait <- 0     # The epoch the training stops at.     self$stopped_epoch <- 0     # Initialize the best as infinity.     self$best <- Inf   },    on_epoch_end = function(epoch, logs = NULL) {     current <- logs$loss     if (current < self$best) {       self$best <- current       self$wait <- 0L       # Record the best weights if current results is better (less).       self$best_weights <- get_weights(self$model)     } else {       add(self$wait) <- 1L       if (self$wait >= self$patience) {         self$stopped_epoch <- epoch         self$model$stop_training <- TRUE         cat(\"Restoring model weights from the end of the best epoch.\\n\")         model$set_weights(self$best_weights)       }     }   },    on_train_end = function(logs = NULL)     if (self$stopped_epoch > 0)       cat(sprintf(\"Epoch %05d: early stopping\\n\", self$stopped_epoch + 1)) ) `add<-` <- `+`   model <- get_model() model |> fit(   mnist$train$x,   mnist$train$y,   epochs = 30,   batch_size = 64,   verbose = 0,   callbacks = list(callback_print_loss_and_mae(),                    callback_early_stopping_at_min_loss()) ) ## Up to batch 1, the average loss is   30.54. ## Up to batch 2, the average loss is  513.27. ## Up to batch 3, the average loss is  352.60. ## Up to batch 4, the average loss is  266.37. ## Up to batch 5, the average loss is  214.68. ## Up to batch 6, the average loss is  179.97. ## Up to batch 7, the average loss is  155.06. ## Up to batch 8, the average loss is  136.59. ## Up to batch 9, the average loss is  121.96. ## Up to batch 10, the average loss is  110.28. ## Up to batch 11, the average loss is  100.72. ## Up to batch 12, the average loss is   92.71. ## Up to batch 13, the average loss is   85.95. ## Up to batch 14, the average loss is   80.21. ## Up to batch 15, the average loss is   75.17. ## Up to batch 16, the average loss is   72.48. ## The average loss for epoch  1 is     72.48 and mean absolute error is    4.08. ## Up to batch 1, the average loss is    7.98. ## Up to batch 2, the average loss is    9.92. ## Up to batch 3, the average loss is   12.88. ## Up to batch 4, the average loss is   16.61. ## Up to batch 5, the average loss is   20.49. ## Up to batch 6, the average loss is   26.14. ## Up to batch 7, the average loss is   30.44. ## Up to batch 8, the average loss is   33.76. ## Up to batch 9, the average loss is   36.32. ## Up to batch 10, the average loss is   35.26. ## Up to batch 11, the average loss is   34.22. ## Up to batch 12, the average loss is   33.53. ## Up to batch 13, the average loss is   32.84. ## Up to batch 14, the average loss is   31.80. ## Up to batch 15, the average loss is   31.39. ## Up to batch 16, the average loss is   31.45. ## The average loss for epoch  2 is     31.45 and mean absolute error is    4.82. ## Up to batch 1, the average loss is   39.60. ## Up to batch 2, the average loss is   41.95. ## Up to batch 3, the average loss is   41.29. ## Up to batch 4, the average loss is   36.77. ## Up to batch 5, the average loss is   32.08. ## Up to batch 6, the average loss is   28.17. ## Up to batch 7, the average loss is   25.33. ## Up to batch 8, the average loss is   23.56. ## Up to batch 9, the average loss is   22.28. ## Up to batch 10, the average loss is   21.22. ## Up to batch 11, the average loss is   20.87. ## Up to batch 12, the average loss is   22.25. ## Up to batch 13, the average loss is   25.08. ## Up to batch 14, the average loss is   27.87. ## Up to batch 15, the average loss is   31.72. ## Up to batch 16, the average loss is   33.21. ## The average loss for epoch  3 is     33.21 and mean absolute error is    4.79. ## Restoring model weights from the end of the best epoch. ## Epoch 00004: early stopping"},{"path":"https://keras3.posit.co/dev/articles/writing_your_own_callbacks.html","id":"learning-rate-scheduling","dir":"Articles","previous_headings":"Examples of Keras callback applications","what":"Learning rate scheduling","title":"Writing your own callbacks","text":"example, show custom Callback can used dynamically change learning rate optimizer course training. See keras$callbacks$LearningRateScheduler general implementations (RStudio, press F1 cursor LearningRateScheduler browser open page).","code":"callback_custom_learning_rate_scheduler <- Callback(   \"CustomLearningRateScheduler\",   `__doc__` =   \"Learning rate scheduler which sets the learning rate according to schedule.      Arguments:         schedule: a function that takes an epoch index             (integer, indexed from 0) and current learning rate             as inputs and returns a new learning rate as output (float).     \",    initialize = function(schedule) {     super$initialize()     self$schedule <- schedule   },    on_epoch_begin = function(epoch, logs = NULL) {     ## When in doubt about what types of objects are in scope (e.g., self$model)     ## use a debugger to interact with the actual objects at the console!     # browser()      if (!\"learning_rate\" %in% names(self$model$optimizer))       stop('Optimizer must have a \"learning_rate\" attribute.')      # # Get the current learning rate from model's optimizer.     # use as.numeric() to convert the keras variablea to an R numeric     lr <- as.numeric(self$model$optimizer$learning_rate)     # # Call schedule function to get the scheduled learning rate.     scheduled_lr <- self$schedule(epoch, lr)     # # Set the value back to the optimizer before this epoch starts     optimizer <- self$model$optimizer     optimizer$learning_rate <- scheduled_lr     cat(sprintf(\"\\nEpoch %03d: Learning rate is %6.4f.\\n\", epoch, scheduled_lr))   } )  LR_SCHEDULE <- tibble::tribble(   ~start_epoch, ~learning_rate,              0,            0.1,              3,           0.05,              6,           0.01,              9,          0.005,             12,          0.001,   )  last <- function(x) x[length(x)] lr_schedule <- function(epoch, learning_rate) {   \"Helper function to retrieve the scheduled learning rate based on epoch.\"   with(LR_SCHEDULE, learning_rate[last(which(epoch >= start_epoch))]) }  model <- get_model() model |> fit(   mnist$train$x,   mnist$train$y,   epochs = 14,   batch_size = 64,   verbose = 0,   callbacks = list(     callback_print_loss_and_mae(),     callback_custom_learning_rate_scheduler(lr_schedule)   ) ) ## ## Epoch 001: Learning rate is 0.1000. ## Up to batch 1, the average loss is   29.36. ## Up to batch 2, the average loss is  513.95. ## Up to batch 3, the average loss is  352.70. ## Up to batch 4, the average loss is  266.46. ## Up to batch 5, the average loss is  214.73. ## Up to batch 6, the average loss is  180.00. ## Up to batch 7, the average loss is  155.05. ## Up to batch 8, the average loss is  136.64. ## Up to batch 9, the average loss is  121.97. ## Up to batch 10, the average loss is  110.30. ## Up to batch 11, the average loss is  100.76. ## Up to batch 12, the average loss is   92.74. ## Up to batch 13, the average loss is   85.95. ## Up to batch 14, the average loss is   80.18. ## Up to batch 15, the average loss is   75.11. ## Up to batch 16, the average loss is   72.38. ## The average loss for epoch  1 is     72.38 and mean absolute error is    4.04. ## ## Epoch 002: Learning rate is 0.1000. ## Up to batch 1, the average loss is    6.95. ## Up to batch 2, the average loss is    8.71. ## Up to batch 3, the average loss is   11.42. ## Up to batch 4, the average loss is   15.15. ## Up to batch 5, the average loss is   19.28. ## Up to batch 6, the average loss is   25.54. ## Up to batch 7, the average loss is   30.38. ## Up to batch 8, the average loss is   33.95. ## Up to batch 9, the average loss is   36.58. ## Up to batch 10, the average loss is   35.46. ## Up to batch 11, the average loss is   34.34. ## Up to batch 12, the average loss is   33.51. ## Up to batch 13, the average loss is   32.67. ## Up to batch 14, the average loss is   31.54. ## Up to batch 15, the average loss is   31.05. ## Up to batch 16, the average loss is   31.09. ## The average loss for epoch  2 is     31.09 and mean absolute error is    4.77. ## ## Epoch 003: Learning rate is 0.0500. ## Up to batch 1, the average loss is   40.40. ## Up to batch 2, the average loss is   22.33. ## Up to batch 3, the average loss is   16.18. ## Up to batch 4, the average loss is   13.09. ## Up to batch 5, the average loss is   11.48. ## Up to batch 6, the average loss is   10.21. ## Up to batch 7, the average loss is    9.22. ## Up to batch 8, the average loss is    8.70. ## Up to batch 9, the average loss is    8.16. ## Up to batch 10, the average loss is    7.80. ## Up to batch 11, the average loss is    7.50. ## Up to batch 12, the average loss is    7.17. ## Up to batch 13, the average loss is    6.89. ## Up to batch 14, the average loss is    6.70. ## Up to batch 15, the average loss is    6.52. ## Up to batch 16, the average loss is    6.54. ## The average loss for epoch  3 is      6.54 and mean absolute error is    1.93. ## ## Epoch 004: Learning rate is 0.0500. ## Up to batch 1, the average loss is    8.74. ## Up to batch 2, the average loss is    8.34. ## Up to batch 3, the average loss is    9.09. ## Up to batch 4, the average loss is    9.72. ## Up to batch 5, the average loss is   10.48. ## Up to batch 6, the average loss is   11.69. ## Up to batch 7, the average loss is   11.83. ## Up to batch 8, the average loss is   11.56. ## Up to batch 9, the average loss is   11.24. ## Up to batch 10, the average loss is   10.84. ## Up to batch 11, the average loss is   10.66. ## Up to batch 12, the average loss is   10.44. ## Up to batch 13, the average loss is   10.21. ## Up to batch 14, the average loss is   10.06. ## Up to batch 15, the average loss is   10.00. ## Up to batch 16, the average loss is   10.20. ## The average loss for epoch  4 is     10.20 and mean absolute error is    2.71. ## ## Epoch 005: Learning rate is 0.0500. ## Up to batch 1, the average loss is   17.26. ## Up to batch 2, the average loss is   14.09. ## Up to batch 3, the average loss is   12.67. ## Up to batch 4, the average loss is   11.44. ## Up to batch 5, the average loss is   10.54. ## Up to batch 6, the average loss is   10.10. ## Up to batch 7, the average loss is    9.53. ## Up to batch 8, the average loss is    9.17. ## Up to batch 9, the average loss is    8.78. ## Up to batch 10, the average loss is    8.49. ## Up to batch 11, the average loss is    8.50. ## Up to batch 12, the average loss is    8.59. ## Up to batch 13, the average loss is    8.68. ## Up to batch 14, the average loss is    8.86. ## Up to batch 15, the average loss is    9.17. ## Up to batch 16, the average loss is    9.53. ## The average loss for epoch  5 is      9.53 and mean absolute error is    2.58. ## ## Epoch 006: Learning rate is 0.0100. ## Up to batch 1, the average loss is   17.04. ## Up to batch 2, the average loss is   14.85. ## Up to batch 3, the average loss is   11.53. ## Up to batch 4, the average loss is    9.65. ## Up to batch 5, the average loss is    8.44. ## Up to batch 6, the average loss is    7.50. ## Up to batch 7, the average loss is    6.74. ## Up to batch 8, the average loss is    6.56. ## Up to batch 9, the average loss is    6.18. ## Up to batch 10, the average loss is    5.87. ## Up to batch 11, the average loss is    5.63. ## Up to batch 12, the average loss is    5.45. ## Up to batch 13, the average loss is    5.23. ## Up to batch 14, the average loss is    5.12. ## Up to batch 15, the average loss is    4.96. ## Up to batch 16, the average loss is    4.91. ## The average loss for epoch  6 is      4.91 and mean absolute error is    1.67. ## ## Epoch 007: Learning rate is 0.0100. ## Up to batch 1, the average loss is    3.65. ## Up to batch 2, the average loss is    3.04. ## Up to batch 3, the average loss is    2.88. ## Up to batch 4, the average loss is    2.85. ## Up to batch 5, the average loss is    2.88. ## Up to batch 6, the average loss is    2.81. ## Up to batch 7, the average loss is    2.70. ## Up to batch 8, the average loss is    2.96. ## Up to batch 9, the average loss is    2.96. ## Up to batch 10, the average loss is    2.93. ## Up to batch 11, the average loss is    2.95. ## Up to batch 12, the average loss is    2.98. ## Up to batch 13, the average loss is    2.97. ## Up to batch 14, the average loss is    3.01. ## Up to batch 15, the average loss is    3.00. ## Up to batch 16, the average loss is    3.05. ## The average loss for epoch  7 is      3.05 and mean absolute error is    1.34. ## ## Epoch 008: Learning rate is 0.0100. ## Up to batch 1, the average loss is    3.69. ## Up to batch 2, the average loss is    3.21. ## Up to batch 3, the average loss is    3.00. ## Up to batch 4, the average loss is    2.91. ## Up to batch 5, the average loss is    2.94. ## Up to batch 6, the average loss is    2.85. ## Up to batch 7, the average loss is    2.72. ## Up to batch 8, the average loss is    2.95. ## Up to batch 9, the average loss is    2.97. ## Up to batch 10, the average loss is    2.93. ## Up to batch 11, the average loss is    2.96. ## Up to batch 12, the average loss is    2.98. ## Up to batch 13, the average loss is    2.99. ## Up to batch 14, the average loss is    3.05. ## Up to batch 15, the average loss is    3.08. ## Up to batch 16, the average loss is    3.14. ## The average loss for epoch  8 is      3.14 and mean absolute error is    1.36. ## ## Epoch 009: Learning rate is 0.0050. ## Up to batch 1, the average loss is    3.71. ## Up to batch 2, the average loss is    2.93. ## Up to batch 3, the average loss is    2.76. ## Up to batch 4, the average loss is    2.70. ## Up to batch 5, the average loss is    2.76. ## Up to batch 6, the average loss is    2.69. ## Up to batch 7, the average loss is    2.57. ## Up to batch 8, the average loss is    2.79. ## Up to batch 9, the average loss is    2.80. ## Up to batch 10, the average loss is    2.77. ## Up to batch 11, the average loss is    2.79. ## Up to batch 12, the average loss is    2.80. ## Up to batch 13, the average loss is    2.78. ## Up to batch 14, the average loss is    2.81. ## Up to batch 15, the average loss is    2.80. ## Up to batch 16, the average loss is    2.83. ## The average loss for epoch  9 is      2.83 and mean absolute error is    1.28. ## ## Epoch 010: Learning rate is 0.0050. ## Up to batch 1, the average loss is    3.02. ## Up to batch 2, the average loss is    2.69. ## Up to batch 3, the average loss is    2.58. ## Up to batch 4, the average loss is    2.57. ## Up to batch 5, the average loss is    2.65. ## Up to batch 6, the average loss is    2.60. ## Up to batch 7, the average loss is    2.48. ## Up to batch 8, the average loss is    2.72. ## Up to batch 9, the average loss is    2.74. ## Up to batch 10, the average loss is    2.71. ## Up to batch 11, the average loss is    2.74. ## Up to batch 12, the average loss is    2.75. ## Up to batch 13, the average loss is    2.74. ## Up to batch 14, the average loss is    2.77. ## Up to batch 15, the average loss is    2.77. ## Up to batch 16, the average loss is    2.80. ## The average loss for epoch 10 is      2.80 and mean absolute error is    1.27. ## ## Epoch 011: Learning rate is 0.0050. ## Up to batch 1, the average loss is    3.01. ## Up to batch 2, the average loss is    2.69. ## Up to batch 3, the average loss is    2.58. ## Up to batch 4, the average loss is    2.56. ## Up to batch 5, the average loss is    2.63. ## Up to batch 6, the average loss is    2.58. ## Up to batch 7, the average loss is    2.47. ## Up to batch 8, the average loss is    2.70. ## Up to batch 9, the average loss is    2.72. ## Up to batch 10, the average loss is    2.69. ## Up to batch 11, the average loss is    2.71. ## Up to batch 12, the average loss is    2.72. ## Up to batch 13, the average loss is    2.71. ## Up to batch 14, the average loss is    2.75. ## Up to batch 15, the average loss is    2.74. ## Up to batch 16, the average loss is    2.77. ## The average loss for epoch 11 is      2.77 and mean absolute error is    1.27. ## ## Epoch 012: Learning rate is 0.0010. ## Up to batch 1, the average loss is    2.96. ## Up to batch 2, the average loss is    2.53. ## Up to batch 3, the average loss is    2.47. ## Up to batch 4, the average loss is    2.46. ## Up to batch 5, the average loss is    2.54. ## Up to batch 6, the average loss is    2.48. ## Up to batch 7, the average loss is    2.39. ## Up to batch 8, the average loss is    2.60. ## Up to batch 9, the average loss is    2.62. ## Up to batch 10, the average loss is    2.59. ## Up to batch 11, the average loss is    2.61. ## Up to batch 12, the average loss is    2.62. ## Up to batch 13, the average loss is    2.60. ## Up to batch 14, the average loss is    2.64. ## Up to batch 15, the average loss is    2.62. ## Up to batch 16, the average loss is    2.64. ## The average loss for epoch 12 is      2.64 and mean absolute error is    1.24. ## ## Epoch 013: Learning rate is 0.0010. ## Up to batch 1, the average loss is    2.82. ## Up to batch 2, the average loss is    2.46. ## Up to batch 3, the average loss is    2.42. ## Up to batch 4, the average loss is    2.42. ## Up to batch 5, the average loss is    2.50. ## Up to batch 6, the average loss is    2.45. ## Up to batch 7, the average loss is    2.36. ## Up to batch 8, the average loss is    2.57. ## Up to batch 9, the average loss is    2.59. ## Up to batch 10, the average loss is    2.57. ## Up to batch 11, the average loss is    2.59. ## Up to batch 12, the average loss is    2.60. ## Up to batch 13, the average loss is    2.59. ## Up to batch 14, the average loss is    2.62. ## Up to batch 15, the average loss is    2.61. ## Up to batch 16, the average loss is    2.63. ## The average loss for epoch 13 is      2.63 and mean absolute error is    1.23. ## ## Epoch 014: Learning rate is 0.0010. ## Up to batch 1, the average loss is    2.79. ## Up to batch 2, the average loss is    2.44. ## Up to batch 3, the average loss is    2.40. ## Up to batch 4, the average loss is    2.41. ## Up to batch 5, the average loss is    2.49. ## Up to batch 6, the average loss is    2.44. ## Up to batch 7, the average loss is    2.34. ## Up to batch 8, the average loss is    2.56. ## Up to batch 9, the average loss is    2.58. ## Up to batch 10, the average loss is    2.56. ## Up to batch 11, the average loss is    2.58. ## Up to batch 12, the average loss is    2.59. ## Up to batch 13, the average loss is    2.58. ## Up to batch 14, the average loss is    2.61. ## Up to batch 15, the average loss is    2.60. ## Up to batch 16, the average loss is    2.62. ## The average loss for epoch 14 is      2.62 and mean absolute error is    1.23."},{"path":"https://keras3.posit.co/dev/articles/writing_your_own_callbacks.html","id":"built-in-keras-callbacks","dir":"Articles","previous_headings":"Examples of Keras callback applications","what":"Built-in Keras callbacks","title":"Writing your own callbacks","text":"sure check existing Keras callbacks reading API docs. Applications include logging CSV, saving model, visualizing metrics TensorBoard, lot !","code":""},{"path":"https://keras3.posit.co/dev/authors.html","id":null,"dir":"","previous_headings":"","what":"Authors","title":"Authors and Citation","text":"Tomasz Kalinowski. Author, copyright holder, maintainer. Daniel Falbel. Contributor, copyright holder. JJ Allaire. Author, copyright holder. François Chollet. Author, copyright holder. Posit Software, PBC. Copyright holder, funder. Google. Copyright holder, funder. Yuan Tang. Contributor, copyright holder. Wouter Van Der Bijl. Contributor, copyright holder. Martin Studer. Contributor, copyright holder. Sigrid Keydana. Contributor.","code":""},{"path":"https://keras3.posit.co/dev/authors.html","id":"citation","dir":"","previous_headings":"","what":"Citation","title":"Authors and Citation","text":"Kalinowski T, Allaire J, Chollet F (2025). keras3: R Interface 'Keras'. R package version 1.3.0.9001, https://github.com/rstudio/keras3, https://keras3.posit.co/.","code":"@Manual{,   title = {keras3: R Interface to 'Keras'},   author = {Tomasz Kalinowski and JJ Allaire and François Chollet},   year = {2025},   note = {R package version 1.3.0.9001, https://github.com/rstudio/keras3},   url = {https://keras3.posit.co/}, }"},{"path":"https://keras3.posit.co/dev/index.html","id":"r-interface-to-keras","dir":"","previous_headings":"","what":"R Interface to Keras","title":"R Interface to Keras","text":"Keras high-level neural networks API developed focus enabling fast experimentation. able go idea result least possible delay key good research. Keras following key features: Allows code run CPU GPU, seamlessly. User-friendly API makes easy quickly prototype deep learning models. Built-support convolutional networks (computer vision), recurrent networks (sequence processing), combination . Supports arbitrary network architectures: multi-input multi-output models, layer sharing, model sharing, etc. means Keras appropriate building essentially deep learning model, memory network neural Turing machine. See package website https://keras3.posit.co complete documentation.","code":""},{"path":"https://keras3.posit.co/dev/reference/Callback.html","id":null,"dir":"Reference","previous_headings":"","what":"Define a custom Callback class — Callback","title":"Define a custom Callback class — Callback","text":"Callbacks can passed keras methods fit(), evaluate(), predict() order hook various stages model training, evaluation, inference lifecycle. create custom callback, call Callback() override method associated stage interest.","code":""},{"path":"https://keras3.posit.co/dev/reference/Callback.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Define a custom Callback class — Callback","text":"","code":"Callback(   classname,   on_epoch_begin = NULL,   on_epoch_end = NULL,   on_train_begin = NULL,   on_train_end = NULL,   on_train_batch_begin = NULL,   on_train_batch_end = NULL,   on_test_begin = NULL,   on_test_end = NULL,   on_test_batch_begin = NULL,   on_test_batch_end = NULL,   on_predict_begin = NULL,   on_predict_end = NULL,   on_predict_batch_begin = NULL,   on_predict_batch_end = NULL,   ...,   public = list(),   private = list(),   inherit = NULL,   parent_env = parent.frame() )"},{"path":"https://keras3.posit.co/dev/reference/Callback.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Define a custom Callback class — Callback","text":"classname String, name custom class. (Conventionally, CamelCase). on_epoch_begin   Called start epoch. Subclasses override actions run. function called TRAIN mode. Args: epoch: Integer, index epoch. logs: Named List. Currently data passed argument method may change future. on_epoch_end   Called end epoch. Subclasses override actions run. function called TRAIN mode. Args: epoch: Integer, index epoch. logs: Named List, metric results training epoch, validation epoch validation performed. Validation result keys prefixed val_. training epoch, values Model's metrics returned. Example: list(loss = 0.2, accuracy = 0.7). on_train_begin   Called beginning training. Subclasses override actions run. Args: logs: Named list. Currently data passed argument method may change future. on_train_end   Called end training. Subclasses override actions run. Args: logs: Named list. Currently output last call on_epoch_end() passed argument method may change future. on_train_batch_begin   Called beginning training batch fit() methods. Subclasses override actions run. Note steps_per_execution argument compile Model set N, method called every N batches. Args: batch: Integer, index batch within current epoch. logs: Named list. Currently data passed argument method may change future. on_train_batch_end   Called end training batch fit() methods. Subclasses override actions run. Note steps_per_execution argument compile Model set N, method called every N batches. Args: batch: Integer, index batch within current epoch. logs: Named list. Aggregated metric results batch. on_test_begin   Called beginning evaluation validation. Subclasses override actions run. Args: logs: Named list. Currently data passed argument method may change future. on_test_end   Called end evaluation validation. Subclasses override actions run. Args: logs: Named list. Currently output last call on_test_batch_end() passed argument method may change future. on_test_batch_begin   Called beginning batch evaluate() methods. Also called beginning validation batch fit() methods, validation data provided. Subclasses override actions run. Note steps_per_execution argument compile() Model set N, method called every N batches. Args: batch: Integer, index batch within current epoch. logs: Named list. Currently data passed argument method may change future. on_test_batch_end   Called end batch evaluate() methods. Also called end validation batch fit() methods, validation data provided. Subclasses override actions run. Note steps_per_execution argument compile() Model set N, method called every N batches. Args: batch: Integer, index batch within current epoch. logs: Named list. Aggregated metric results batch. on_predict_begin   Called beginning prediction. Subclasses override actions run. Args: logs: Named list. Currently data passed argument method may change future. on_predict_end   Called end prediction. Subclasses override actions run. Args: logs: Named list. Currently data passed argument method may change future. on_predict_batch_begin   Called beginning batch predict() methods. Subclasses override actions run. Note steps_per_execution argument compile() Model set N, method called every N batches. Args: batch: Integer, index batch within current epoch. logs: Named list. Currently data passed argument method may change future. on_predict_batch_end   Called end batch predict() methods. Subclasses override actions run. Note steps_per_execution argument compile Model set N, method called every N batches. Args: batch: Integer, index batch within current epoch. logs: Named list. Aggregated metric results batch. ..., public Additional methods public members custom class. private Named list R objects (typically, functions) include instance private environments. private methods symbols scope public methods (See section \"Symbols Scope\"). instance private environment. objects private invisible Keras framework Python runtime. inherit custom class subclass. default, base keras class. parent_env R environment class methods grandparent.","code":"\\(epoch, logs = NULL) \\(epoch, logs = NULL) \\(logs = NULL) \\(logs = NULL) \\(batch, logs = NULL) \\(batch, logs=NULL) \\(logs = NULL) \\(logs = NULL) \\(batch, logs = NULL) \\(batch, logs = NULL) \\(logs = NULL) \\(logs = NULL) \\(batch, logs = NULL) \\(batch, logs = NULL)"},{"path":"https://keras3.posit.co/dev/reference/Callback.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Define a custom Callback class — Callback","text":"function returns custom Callback instances, similar builtin callback functions.","code":""},{"path":"https://keras3.posit.co/dev/reference/Callback.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Define a custom Callback class — Callback","text":"R function custom methods (public private) following symbols scope: self: Layer instance. super: Layer superclass. private: R environment specific class instance. objects defined invisible Keras framework. __class__ current class type object. also available alias symbol, value supplied Layer(classname = )","code":"training_finished <- FALSE callback_mark_finished <- Callback(\"MarkFinished\",   on_train_end = function(logs = NULL) {     training_finished <<- TRUE   } )  model <- keras_model_sequential(input_shape = c(1)) |>   layer_dense(1) model |> compile(loss = 'mean_squared_error') model |> fit(op_ones(c(1, 1)), op_ones(c(1, 1)),              callbacks = callback_mark_finished()) stopifnot(isTRUE(training_finished))"},{"path":"https://keras3.posit.co/dev/reference/Callback.html","id":"attributes-accessible-via-self-","dir":"Reference","previous_headings":"","what":"Attributes (accessible via self$)","title":"Define a custom Callback class — Callback","text":"params: Named list, Training parameters (e.g. verbosity, batch size, number epochs, ...). model: Instance Model. Reference model trained. logs named list callback methods take argument contain keys quantities relevant current batch epoch (see method-specific docstrings).","code":""},{"path":"https://keras3.posit.co/dev/reference/Callback.html","id":"symbols-in-scope","dir":"Reference","previous_headings":"","what":"Symbols in scope","title":"Define a custom Callback class — Callback","text":"R function custom methods (public private) following symbols scope: self: custom class instance. super: custom class superclass. private: R environment specific class instance. objects assigned invisible Keras framework. __class__ .symbol(classname): custom class type object.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/Constraint.html","id":null,"dir":"Reference","previous_headings":"","what":"Define a custom Constraint class — Constraint","title":"Define a custom Constraint class — Constraint","text":"Base class weight constraints. Constraint() instance works like stateless function. Users subclass Constraint class override call() method, takes single weight parameter return projected version parameter (e.g. normalized clipped). Constraints can used various Keras layers via kernel_constraint bias_constraint arguments. simple example non-negative weight constraint:     Usage layer:","code":"constraint_nonnegative <- Constraint(\"NonNegative\",   call = function(w) {     w * op_cast(w >= 0, dtype = w$dtype)   } ) weight <- op_convert_to_tensor(c(-1, 1)) constraint_nonnegative()(weight) ## tf.Tensor([-0.  1.], shape=(2), dtype=float64) layer_dense(units = 4, kernel_constraint = constraint_nonnegative()) ## <Dense name=dense, built=False> ##  signature: (*args, **kwargs)"},{"path":"https://keras3.posit.co/dev/reference/Constraint.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Define a custom Constraint class — Constraint","text":"","code":"Constraint(   classname,   call = NULL,   get_config = NULL,   ...,   public = list(),   private = list(),   inherit = NULL,   parent_env = parent.frame() )"},{"path":"https://keras3.posit.co/dev/reference/Constraint.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Define a custom Constraint class — Constraint","text":"classname String, name custom class. (Conventionally, CamelCase). call   Applies constraint input weight variable. default, inputs weight variable modified. Users override method implement projection function. Args: w: Input weight variable. Returns: Projected variable (default, returns unmodified inputs). get_config   Function returns named list object config. constraint config named list (JSON-serializable) can used reinstantiate object (via .call(<constraint_class>, <config>)). ..., public Additional methods public members custom class. private Named list R objects (typically, functions) include instance private environments. private methods symbols scope public methods (See section \"Symbols Scope\"). instance private environment. objects private invisible Keras framework Python runtime. inherit custom class subclass. default, base keras class. parent_env R environment class methods grandparent.","code":"\\(w) \\()"},{"path":"https://keras3.posit.co/dev/reference/Constraint.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Define a custom Constraint class — Constraint","text":"function returns Constraint instances, similar builtin constraint functions like constraint_maxnorm().","code":""},{"path":"https://keras3.posit.co/dev/reference/Constraint.html","id":"symbols-in-scope","dir":"Reference","previous_headings":"","what":"Symbols in scope","title":"Define a custom Constraint class — Constraint","text":"R function custom methods (public private) following symbols scope: self: custom class instance. super: custom class superclass. private: R environment specific class instance. objects assigned invisible Keras framework. __class__ .symbol(classname): custom class type object.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/Layer.html","id":null,"dir":"Reference","previous_headings":"","what":"Define a custom Layer class. — Layer","title":"Define a custom Layer class. — Layer","text":"layer callable object takes input one tensors outputs one tensors. involves computation, defined call() method, state (weight variables). State can created: initialize(), instance via self$add_weight(); optional build() method, invoked first call() layer, supplies shape(s) input(s), may known initialization time. Layers recursively composable: assign Layer instance attribute another Layer, outer layer start tracking weights created inner layer. Nested layers instantiated initialize() method build() method. Users just instantiate layer treat callable.","code":""},{"path":"https://keras3.posit.co/dev/reference/Layer.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Define a custom Layer class. — Layer","text":"","code":"Layer(   classname,   initialize = NULL,   call = NULL,   build = NULL,   get_config = NULL,   ...,   public = list(),   private = list(),   inherit = NULL,   parent_env = parent.frame() )"},{"path":"https://keras3.posit.co/dev/reference/Layer.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Define a custom Layer class. — Layer","text":"classname String, name custom class. (Conventionally, CamelCase). initialize, call, build, get_config Recommended methods implement. See description details sections. ..., public Additional methods public members custom class. private Named list R objects (typically, functions) include instance private environments. private methods symbols scope public methods (See section \"Symbols Scope\"). instance private environment. objects private invisible Keras framework Python runtime. inherit custom class subclass. default, base keras class. parent_env R environment class methods grandparent.","code":""},{"path":"https://keras3.posit.co/dev/reference/Layer.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Define a custom Layer class. — Layer","text":"composing layer constructor, similar behavior layer functions like layer_dense(). first argument returned function object, enabling initialize()ing call() layer one step composing layer pipe, like   initialize() layer instance call() , pass missing NULL value object, pass arguments initialize() name.","code":"layer_foo <- Layer(\"Foo\", ....) output <- inputs |> layer_foo() layer <- layer_dense(units = 2, activation = \"relu\") layer <- layer_dense(NULL, 2, activation = \"relu\") layer <- layer_dense(, 2, activation = \"relu\")  # then you can call() the layer in a separate step outputs <- inputs |> layer()"},{"path":"https://keras3.posit.co/dev/reference/Layer.html","id":"symbols-in-scope","dir":"Reference","previous_headings":"","what":"Symbols in scope","title":"Define a custom Layer class. — Layer","text":"R function custom methods (public private) following symbols scope: self: custom class instance. super: custom class superclass. private: R environment specific class instance. objects assigned invisible Keras framework. __class__ .symbol(classname): custom class type object.","code":""},{"path":"https://keras3.posit.co/dev/reference/Layer.html","id":"attributes","dir":"Reference","previous_headings":"","what":"Attributes","title":"Define a custom Layer class. — Layer","text":"name: name layer (string). dtype: Dtype layer's weights. Alias layer$variable_dtype. variable_dtype: Dtype layer's weights. compute_dtype: dtype layer's computations. Layers automatically cast inputs dtype, causes computations output also dtype. mixed precision used keras$mixed_precision$DTypePolicy, different variable_dtype. trainable_weights: List variables included backprop. non_trainable_weights: List variables included backprop. weights: concatenation lists trainable_weights non_trainable_weights (order). trainable: Whether layer trained (boolean), .e. whether potentially-trainable weights returned part layer$trainable_weights. input_spec: Optional (list ) InputSpec object(s) specifying constraints inputs can accepted layer. recommend custom Layers implement following methods: initialize(): Defines custom layer attributes, creates layer weights depend input shapes, using add_weight(), state. build(input_shape): method can used create weights depend shape(s) input(s), using add_weight(), state. Calling call() automatically build layer (built yet) calling build(). call(...): Method called making sure build() called. call() performs logic applying layer input arguments. Two reserved arguments can optionally use call() : training (boolean, whether call inference mode training mode). mask (boolean tensor encoding masked timesteps input, used e.g. RNN layers). typical signature method call(inputs), user optionally add training mask layer need . get_config(): Returns named list containing configuration used initialize layer. list names differ arguments initialize(), override from_config() well. method used saving layer model contains layer.","code":""},{"path":"https://keras3.posit.co/dev/reference/Layer.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Define a custom Layer class. — Layer","text":"basic example: layer two variables, w b, returns y <- (w %*% x) + b. shows implement build() call(). Variables set attributes layer tracked weights layers (layer$weights).   Besides trainable weights, updated via backpropagation training, layers can also non-trainable weights. weights meant updated manually call(). example layer computes running sum inputs:","code":"layer_simple_dense <- Layer(   \"SimpleDense\",   initialize = function(units = 32) {     super$initialize()     self$units <- units   },    # Create the state of the layer (weights)   build = function(input_shape) {     self$kernel <- self$add_weight(       shape = shape(tail(input_shape, 1), self$units),       initializer = \"glorot_uniform\",       trainable = TRUE,       name = \"kernel\"     )     self$bias = self$add_weight(       shape = shape(self$units),       initializer = \"zeros\",       trainable = TRUE,       name = \"bias\"     )   },    # Defines the computation   call = function(self, inputs) {     op_matmul(inputs, self$kernel) + self$bias   } )  # Instantiates the layer. # Supply missing `object` arg to skip invoking `call()` and instead return # the Layer instance linear_layer <- layer_simple_dense(, 4)  # This will call `build(input_shape)` and create the weights, # and then invoke `call()`. y <- linear_layer(op_ones(c(2, 2))) stopifnot(length(linear_layer$weights) == 2)  # These weights are trainable, so they're listed in `trainable_weights`: stopifnot(length(linear_layer$trainable_weights) == 2) layer_compute_sum <- Layer(   classname = \"ComputeSum\",    initialize = function(input_dim) {     super$initialize()      # Create a non-trainable weight.     self$total <- self$add_weight(       shape = shape(),       initializer = \"zeros\",       trainable = FALSE,       name = \"total\"     )   },    call = function(inputs) {     self$total$assign(self$total + op_sum(inputs))     self$total   } )  my_sum <- layer_compute_sum(, 2) x <- op_ones(c(2, 2)) y <- my_sum(x)  stopifnot(exprs = {   all.equal(my_sum$weights,               list(my_sum$total))   all.equal(my_sum$non_trainable_weights, list(my_sum$total))   all.equal(my_sum$trainable_weights,     list()) })"},{"path":"https://keras3.posit.co/dev/reference/Layer.html","id":"methods-available","dir":"Reference","previous_headings":"","what":"Methods available","title":"Define a custom Layer class. — Layer","text":"Initialize self. method typically called custom initialize() method. Example:   Args: trainable: Boolean, whether layer's variables trainable. name: String name layer. dtype: dtype layer's computations weights. Can also keras$DTypePolicy, allows computation weight dtype differ. Defaults NULL. NULL means use config_dtype_policy(), \"float32\" policy unless set different value (via config_set_dtype_policy()).    Can called inside call() method add scalar loss. Example:         Add weight variable layer. Alias add_weight().    Add weight variable layer. Args: shape: shape variable (defined shape()) Must fully-defined (NA/NULL/-1 entries). Defaults () (scalar) unspecified. initializer: Initializer object use populate initial variable value, string name built-initializer (e.g. \"random_normal\"). unspecified, defaults \"glorot_uniform\" floating-point variables \"zeros\" types (e.g. int, bool). dtype: Dtype variable create, e.g. \"float32\". unspecified, defaults layer's variable dtype (defaults \"float32\" unspecified). trainable: Boolean, whether variable trainable via backprop whether updates managed manually. Defaults TRUE. autocast: Boolean, whether autocast layers variables accessing . Defaults TRUE. regularizer: Regularizer object call apply penalty weight. penalties summed loss function optimization. Defaults NULL. constraint: Constraint object call variable optimizer update, string name built-constraint. Defaults NULL. aggregation: Optional string, one NULL, \"none\", \"mean\", \"sum\" \"only_first_replica\". Annotates variable type multi-replica aggregation used variable writing custom data parallel training loops. Defaults \"none\". name: String name variable. Useful debugging purposes. Returns: backend tensor, wrapped KerasVariable class. KerasVariable class Methods: assign(value) assign_add(value) assign_sub(value) numpy() (calling .array(<variable>) preferred) Properties/Attributes: value dtype ndim shape (calling shape(<variable>) preferred) trainable       Builds layer's states supplied config (named list args). default, method calls .call(build, config$input_shape) method, creates weights based layer's input shape supplied config. config contains information needed load layer's state, override method. Args: config: Named list containing input shape associated layer.    See description             Count total number scalars composing weights. Returns: integer count.    Returns named list layer's input shape. method returns config (named list) can used build_from_config(config) create states (e.g. Variables Lookup tables) needed layer. default, config contains input shape layer built . writing custom layer creates state unusual way, override method make sure state already created Keras attempts load value upon model loading. Returns: named list containing input shape associated layer.    Returns config object. object config named list (serializable) containing information needed re-instantiate . config expected serializable JSON, expected consist (potentially complex, nested) structure names lists consisting simple objects like strings, ints.    Return values layer$weights list R NumPy arrays.    Currently, Dense, EinsumDense Embedding layers support -place quantization via quantize() method. Example:            Enable rematerialization dynamically layer's call method. Args: layer_call: original call method layer. ...: additional args Returns: rematerialized version layer's call method.    Loads state layer. can override method take full control state layer loaded upon calling load_model(). Args: store: Named list state model loaded.    Saves state layer. can override method take full control state layer saved upon calling save_model(). Args: store: Named list state model saved.    Sets values weights list R NumPy arrays.    Call layer without side effects. Args: trainable_variables: List trainable variables model. non_trainable_variables: List non-trainable variables model. ...: Positional named arguments passed call(). return_losses: TRUE, stateless_call() return list losses created call() part return values. Returns: unnamed list. default, returns list(outputs, non_trainable_variables). return_losses = TRUE, returns list(outputs, non_trainable_variables, losses). Note: non_trainable_variables include non-trainable weights BatchNormalization statistics, also RNG seed state (random operations part layer, dropout), Metric state (metrics attached layer). elements state layer. Example:         Creates layer config. class method, meaning, R function self symbol (class instance) scope. Use __class__ classname symbol provided Layer() constructed) resolve class definition. default implementation :   method reverse get_config(), capable instantiating layer config named list. handle layer connectivity (handled Network), weights (handled set_weights()). Args: config: named list, typically output get_config(). Returns: layer instance.","code":"initialize(...,            activity_regularizer = NULL,            trainable = TRUE,            dtype = NULL,            autocast = TRUE,            name = NULL) layer_my_layer <- Layer(\"MyLayer\",   initialize = function(units, ..., dtype = NULL, name = NULL) {     super$initialize(..., dtype = dtype, name = name)     # .... finish initializing `self` instance   } ) add_loss(loss) Layer(\"MyLayer\",   ...   call = function(x) {     self$add_loss(op_sum(x))     x   } add_metric(...) add_variable(...) add_weight(shape = NULL,            initializer = NULL,            dtype = NULL,            trainable = TRUE,            autocast = TRUE,            regularizer = NULL,            constraint = NULL,            aggregation = 'none',            name = NULL) build(input_shape) build_from_config(config) call(...) compute_mask(inputs, previous_mask) compute_output_shape(...) compute_output_spec(...) count_params() get_build_config() get_config() get_weights() quantize(mode, type_check = TRUE) model$quantize(\"int8\") # quantize model in-place model |> predict(data) # faster inference quantized_build(input_shape, mode) quantized_call(...) rematerialized_call(layer_call, ...) load_own_variables(store) save_own_variables(store) set_weights(weights) stateless_call(trainable_variables, non_trainable_variables,                ..., return_losses = FALSE) model <- ... data <- ... trainable_variables <- model$trainable_variables non_trainable_variables <- model$non_trainable_variables # Call the model with zero side effects c(outputs, non_trainable_variables) %<-% model$stateless_call(     trainable_variables,     non_trainable_variables,     data ) # Attach the updated state to the model # (until you do this, the model is still in its pre-call state). purrr::walk2(   model$non_trainable_variables, non_trainable_variables,   \\(variable, value) variable$assign(value)) symbolic_call(...) from_config(config) from_config = function(config) {   do.call(`__class__`, config) }"},{"path":"https://keras3.posit.co/dev/reference/Layer.html","id":"readonly-properties-","dir":"Reference","previous_headings":"","what":"Readonly properties:","title":"Define a custom Layer class. — Layer","text":"compute_dtype dtype computations performed layer. dtype Alias layer$variable_dtype. input_dtype dtype layer inputs converted . losses List scalar losses add_loss(), regularizers sublayers. metrics List metrics. metrics_variables List metric variables. non_trainable_variables List non-trainable layer state. extends layer$non_trainable_weights include state used layer including state metrics SeedGenerators. non_trainable_weights List non-trainable weight variables layer. weights updated optimizer training. Unlike, layer$non_trainable_variables excludes metric state random seeds. trainable_variables List trainable layer state. equivalent layer$trainable_weights. trainable_weights List trainable weight variables layer. weights get updated optimizer training. path path layer. layer built yet, NULL. quantization_mode quantization mode layer, NULL quantized. variable_dtype dtype state (weights) layer. variables List layer state, including random seeds. extends layer$weights include state used layer including SeedGenerators. Note metrics variables included , use metrics_variables visit metric variables. weights List weight variables layer. Unlike, layer$variables excludes metric state random seeds. input Retrieves input tensor(s) symbolic operation. returns tensor(s) corresponding first time operation called. Returns: Input tensor list input tensors. output Retrieves output tensor(s) layer. returns tensor(s) corresponding first time operation called. Returns: Output tensor list output tensors.","code":""},{"path":"https://keras3.posit.co/dev/reference/Layer.html","id":"data-descriptors-attributes-","dir":"Reference","previous_headings":"","what":"Data descriptors (Attributes):","title":"Define a custom Layer class. — Layer","text":"dtype_policy input_spec supports_masking Whether layer supports computing mask using compute_mask. trainable Settable boolean, whether layer trainable .","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/LearningRateSchedule.html","id":null,"dir":"Reference","previous_headings":"","what":"Define a custom LearningRateSchedule class — LearningRateSchedule","title":"Define a custom LearningRateSchedule class — LearningRateSchedule","text":"Subclass Keras LearningRateSchedule base class. can use learning rate schedule modulate learning rate optimizer changes time. Several built-learning rate schedules available, learning_rate_schedule_exponential_decay() learning_rate_schedule_piecewise_constant_decay():   LearningRateSchedule() instance can passed learning_rate argument optimizer. implement schedule object, implement call method, takes step argument (scalar integer backend tensor, current training step count). Note step 0-based (.e., first step 0). Like Keras object, can also optionally make object serializable implementing get_config() from_config() methods.","code":"lr_schedule <- learning_rate_schedule_exponential_decay(   initial_learning_rate = 1e-2,   decay_steps = 10000,   decay_rate = 0.9 ) optimizer <- optimizer_sgd(learning_rate = lr_schedule)"},{"path":"https://keras3.posit.co/dev/reference/LearningRateSchedule.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Define a custom LearningRateSchedule class — LearningRateSchedule","text":"","code":"LearningRateSchedule(   classname,   call = NULL,   initialize = NULL,   get_config = NULL,   ...,   public = list(),   private = list(),   inherit = NULL,   parent_env = parent.frame() )"},{"path":"https://keras3.posit.co/dev/reference/LearningRateSchedule.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Define a custom LearningRateSchedule class — LearningRateSchedule","text":"classname String, name custom class. (Conventionally, CamelCase). call, initialize, get_config Recommended methods implement. See description details sections. ..., public Additional methods public members custom class. private Named list R objects (typically, functions) include instance private environments. private methods symbols scope public methods (See section \"Symbols Scope\"). instance private environment. objects private invisible Keras framework Python runtime. inherit custom class subclass. default, base keras class. parent_env R environment class methods grandparent.","code":""},{"path":"https://keras3.posit.co/dev/reference/LearningRateSchedule.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Define a custom LearningRateSchedule class — LearningRateSchedule","text":"function returns LearningRateSchedule instances, similar built-learning_rate_schedule_* family functions.","code":""},{"path":"https://keras3.posit.co/dev/reference/LearningRateSchedule.html","id":"example","dir":"Reference","previous_headings":"","what":"Example","title":"Define a custom LearningRateSchedule class — LearningRateSchedule","text":"","code":"my_custom_learning_rate_schedule <- LearningRateSchedule(   classname = \"MyLRSchedule\",    initialize = function(initial_learning_rate) {     self$initial_learning_rate <- initial_learning_rate   },    call = function(step) {     # note that `step` is a tensor     # and call() will be traced via tf_function() or similar.      str(step) # <KerasVariable shape=(), dtype=int64, path=SGD/iteration>      # print 'step' every 1000 steps     op_cond((step %% 1000) == 0,             \\() {tensorflow::tf$print(step); NULL},             \\() {NULL})     self$initial_learning_rate / (step + 1)   } )  optimizer <- optimizer_sgd(   learning_rate = my_custom_learning_rate_schedule(0.1) )  # You can also call schedule instances directly # (e.g., for interactive testing, or if implementing a custom optimizer) schedule <- my_custom_learning_rate_schedule(0.1) step <- keras$Variable(initializer = op_ones,                        shape = shape(),                        dtype = \"int64\") schedule(step) ## <Variable path=variable, shape=(), dtype=int64, value=1> ## tf.Tensor(0.0, shape=(), dtype=float64)"},{"path":"https://keras3.posit.co/dev/reference/LearningRateSchedule.html","id":"methods-available-","dir":"Reference","previous_headings":"","what":"Methods available:","title":"Define a custom LearningRateSchedule class — LearningRateSchedule","text":"","code":"get_config()"},{"path":"https://keras3.posit.co/dev/reference/LearningRateSchedule.html","id":"symbols-in-scope","dir":"Reference","previous_headings":"","what":"Symbols in scope","title":"Define a custom LearningRateSchedule class — LearningRateSchedule","text":"R function custom methods (public private) following symbols scope: self: custom class instance. super: custom class superclass. private: R environment specific class instance. objects assigned invisible Keras framework. __class__ .symbol(classname): custom class type object.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/Loss.html","id":null,"dir":"Reference","previous_headings":"","what":"Subclass the base Loss class — Loss","title":"Subclass the base Loss class — Loss","text":"Use define custom loss class. Note, cases need subclass Loss define custom loss: can also pass bare R function, named R function defined custom_metric(), loss function compile().","code":""},{"path":"https://keras3.posit.co/dev/reference/Loss.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Subclass the base Loss class — Loss","text":"","code":"Loss(   classname,   call = NULL,   ...,   public = list(),   private = list(),   inherit = NULL,   parent_env = parent.frame() )"},{"path":"https://keras3.posit.co/dev/reference/Loss.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Subclass the base Loss class — Loss","text":"classname String, name custom class. (Conventionally, CamelCase). call   Method implemented subclasses: Function contains logic loss calculation using y_true, y_pred. ..., public Additional methods public members custom class. private Named list R objects (typically, functions) include instance private environments. private methods symbols scope public methods (See section \"Symbols Scope\"). instance private environment. objects private invisible Keras framework Python runtime. inherit custom class subclass. default, base keras class. parent_env R environment class methods grandparent.","code":"function(y_true, y_pred)"},{"path":"https://keras3.posit.co/dev/reference/Loss.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Subclass the base Loss class — Loss","text":"function returns Loss instances, similar builtin loss functions.","code":""},{"path":"https://keras3.posit.co/dev/reference/Loss.html","id":"details","dir":"Reference","previous_headings":"","what":"Details","title":"Subclass the base Loss class — Loss","text":"Example subclass implementation:","code":"loss_custom_mse <- Loss(   classname = \"CustomMeanSquaredError\",   call = function(y_true, y_pred) {     op_mean(op_square(y_pred - y_true), axis = -1)   } )  # Usage in compile() model <- keras_model_sequential(input_shape = 10) |> layer_dense(10) model |> compile(loss = loss_custom_mse())  # Standalone usage mse <- loss_custom_mse(name = \"my_custom_mse_instance\")  y_true <- op_arange(20) |> op_reshape(c(4, 5)) y_pred <- op_arange(20) |> op_reshape(c(4, 5)) * 2 (loss <- mse(y_true, y_pred)) ## tf.Tensor(143.5, shape=(), dtype=float32) loss2 <- (y_pred - y_true)^2 |>   op_mean(axis = -1) |>   op_mean()  stopifnot(all.equal(as.array(loss), as.array(loss2)))  sample_weight <-array(c(.25, .25, 1, 1)) (weighted_loss <- mse(y_true, y_pred, sample_weight = sample_weight)) ## tf.Tensor(129.0625, shape=(), dtype=float32) weighted_loss2 <- (y_true - y_pred)^2 |>   op_mean(axis = -1) |>   op_multiply(sample_weight) |>   op_mean()  stopifnot(all.equal(as.array(weighted_loss),                     as.array(weighted_loss2)))"},{"path":"https://keras3.posit.co/dev/reference/Loss.html","id":"methods-defined-by-base-loss-class-","dir":"Reference","previous_headings":"","what":"Methods defined by base Loss class:","title":"Subclass the base Loss class — Loss","text":"Args: name: Optional name loss instance. reduction: Type reduction apply loss. almost cases \"sum_over_batch_size\". Supported options \"sum\", \"sum_over_batch_size\", \"mean\", \"mean_with_sample_weight\" NULL. \"sum\" sums loss, \"sum_over_batch_size\" \"mean\" sum loss divide sample size, \"mean_with_sample_weight\" sums loss divides sum sample weights. \"none\" NULL perform aggregation. Defaults \"sum_over_batch_size\". dtype: dtype loss's computations. Defaults NULL, means using config_floatx(). config_floatx() \"float32\" unless set different value (via config_set_floatx()). keras$DTypePolicy provided, compute_dtype utilized.    Call loss instance function, optionally sample_weight.","code":"initialize(name=NULL, reduction=\"sum_over_batch_size\", dtype=NULL) __call__(y_true, y_pred, sample_weight=NULL) get_config()"},{"path":"https://keras3.posit.co/dev/reference/Loss.html","id":"readonly-properties-","dir":"Reference","previous_headings":"","what":"Readonly properties:","title":"Subclass the base Loss class — Loss","text":"","code":"dtype"},{"path":"https://keras3.posit.co/dev/reference/Loss.html","id":"symbols-in-scope","dir":"Reference","previous_headings":"","what":"Symbols in scope","title":"Subclass the base Loss class — Loss","text":"R function custom methods (public private) following symbols scope: self: custom class instance. super: custom class superclass. private: R environment specific class instance. objects assigned invisible Keras framework. __class__ .symbol(classname): custom class type object.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/Metric.html","id":null,"dir":"Reference","previous_headings":"","what":"Subclass the base Metric class — Metric","title":"Subclass the base Metric class — Metric","text":"Metric object encapsulates metric logic state can used track model performance training. returned family metric functions start prefix metric_*, well returned custom metrics defined Metric().","code":""},{"path":"https://keras3.posit.co/dev/reference/Metric.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Subclass the base Metric class — Metric","text":"","code":"Metric(   classname,   initialize = NULL,   update_state = NULL,   result = NULL,   ...,   public = list(),   private = list(),   inherit = NULL,   parent_env = parent.frame() )"},{"path":"https://keras3.posit.co/dev/reference/Metric.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Subclass the base Metric class — Metric","text":"classname String, name custom class. (Conventionally, CamelCase). initialize, update_state, result Recommended methods implement. See description section. ..., public Additional methods public members custom class. private Named list R objects (typically, functions) include instance private environments. private methods symbols scope public methods (See section \"Symbols Scope\"). instance private environment. objects private invisible Keras framework Python runtime. inherit custom class subclass. default, base keras class. parent_env R environment class methods grandparent.","code":""},{"path":"https://keras3.posit.co/dev/reference/Metric.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Subclass the base Metric class — Metric","text":"function returns Metric instances, similar builtin metric functions.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/Metric.html","id":"usage-with-compile-","dir":"Reference","previous_headings":"","what":"Usage with compile():","title":"Subclass the base Metric class — Metric","text":"","code":"model |> compile(   optimizer = 'sgd',   loss = 'mse',   metrics = c(metric_SOME_METRIC(), metric_SOME_OTHER_METRIC()) )"},{"path":"https://keras3.posit.co/dev/reference/Metric.html","id":"standalone-usage-","dir":"Reference","previous_headings":"","what":"Standalone usage:","title":"Subclass the base Metric class — Metric","text":"","code":"m <- metric_SOME_METRIC() for (e in seq(epochs)) {   for (i in seq(train_steps)) {     c(y_true, y_pred, sample_weight = NULL) %<-% ...     m$update_state(y_true, y_pred, sample_weight)   }   cat('Final epoch result: ', as.numeric(m$result()), \"\\n\")   m$reset_state() }"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/Metric.html","id":"usage-with-compile--1","dir":"Reference","previous_headings":"","what":"Usage with compile():","title":"Subclass the base Metric class — Metric","text":"implemented subclasses (custom metrics): initialize(): state variables created method calling self$add_variable() like: self$var <- self$add_variable(...). update_state(): Updates state variables like: self$var$assign(...). result(): Computes returns scalar value named list scalar values metric state variables. Example subclass implementation:","code":"model <- keras_model_sequential() model |>   layer_dense(64, activation = \"relu\") |>   layer_dense(64, activation = \"relu\") |>   layer_dense(10, activation = \"softmax\") model |>   compile(optimizer = optimizer_rmsprop(0.01),           loss = loss_categorical_crossentropy(),           metrics = metric_categorical_accuracy())  data <- random_uniform(c(1000, 32)) labels <- random_uniform(c(1000, 10))  model |> fit(data, labels, verbose = 0) metric_binary_true_positives <- Metric(   classname = \"BinaryTruePositives\",    initialize = function(name = 'binary_true_positives', ...) {     super$initialize(name = name, ...)     self$true_positives <-       self$add_weight(shape = shape(),                       initializer = 'zeros',                       name = 'true_positives')   },    update_state = function(y_true, y_pred, sample_weight = NULL) {     y_true <- op_cast(y_true, \"bool\")     y_pred <- op_cast(y_pred, \"bool\")      values <- y_true & y_pred # `&` calls op_logical_and()     values <- op_cast(values, self$dtype)     if (!is.null(sample_weight)) {       sample_weight <- op_cast(sample_weight, self$dtype)       sample_weight <- op_broadcast_to(sample_weight, shape(values))       values <- values * sample_weight # `*` calls op_multiply()     }     self$true_positives$assign(self$true_positives + op_sum(values))   },    result = function() {     self$true_positives   } ) model <- keras_model_sequential(input_shape = 32) |> layer_dense(10) model |> compile(loss = loss_binary_crossentropy(),                  metrics = list(metric_binary_true_positives())) model |> fit(data, labels, verbose = 0)"},{"path":"https://keras3.posit.co/dev/reference/Metric.html","id":"methods-defined-by-the-base-metric-class-","dir":"Reference","previous_headings":"","what":"Methods defined by the base Metric class:","title":"Subclass the base Metric class — Metric","text":"Calling metric instance self like m(...) equivalent calling:      Initialize self. Args: name: Optional name metric instance. dtype: dtype metric's computations. Defaults NULL, means using config_floatx(). config_floatx() \"float32\" unless set different value (via config_set_floatx()). keras$DTypePolicy provided, compute_dtype utilized.          Return serializable config metric.    Reset metric state variables. function called epochs/steps, metric evaluated training.    Compute current metric value. Returns: scalar tensor, named list scalar tensors.             Accumulate statistics metric.","code":"__call__(...) function(...) {   m$update_state(...)   m$result() } initialize(dtype=NULL, name=NULL) add_variable(shape, initializer, dtype=NULL, aggregation = 'sum', name=NULL) add_weight(shape=shape(), initializer=NULL, dtype=NULL, name=NULL) get_config() reset_state() result() stateless_result(metric_variables) stateless_reset_state() stateless_update_state(metric_variables, ...) update_state(...)"},{"path":"https://keras3.posit.co/dev/reference/Metric.html","id":"readonly-properties","dir":"Reference","previous_headings":"","what":"Readonly properties","title":"Subclass the base Metric class — Metric","text":"dtype variables","code":""},{"path":"https://keras3.posit.co/dev/reference/Metric.html","id":"symbols-in-scope","dir":"Reference","previous_headings":"","what":"Symbols in scope","title":"Subclass the base Metric class — Metric","text":"R function custom methods (public private) following symbols scope: self: custom class instance. super: custom class superclass. private: R environment specific class instance. objects assigned invisible Keras framework. __class__ .symbol(classname): custom class type object.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/Model.html","id":null,"dir":"Reference","previous_headings":"","what":"Subclass the base Keras Model Class — Model","title":"Subclass the base Keras Model Class — Model","text":"advanced use cases need subclass base Model type, e.g., want override train_step() method. just want create define keras model, prefer keras_model() keras_model_sequential(). just want encapsulate custom logic state, need customize training behavior (besides calling self$add_loss() call() method), prefer Layer().","code":""},{"path":"https://keras3.posit.co/dev/reference/Model.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Subclass the base Keras Model Class — Model","text":"","code":"Model(   classname,   initialize = NULL,   call = NULL,   train_step = NULL,   predict_step = NULL,   test_step = NULL,   compute_loss = NULL,   compute_metrics = NULL,   ...,   public = list(),   private = list(),   inherit = NULL,   parent_env = parent.frame() )"},{"path":"https://keras3.posit.co/dev/reference/Model.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Subclass the base Keras Model Class — Model","text":"classname String, name custom class. (Conventionally, CamelCase). initialize, call, train_step, predict_step, test_step, compute_loss, compute_metrics Optional methods can overridden. ..., public Additional methods public members custom class. private Named list R objects (typically, functions) include instance private environments. private methods symbols scope public methods (See section \"Symbols Scope\"). instance private environment. objects private invisible Keras framework Python runtime. inherit custom class subclass. default, base keras class. parent_env R environment class methods grandparent.","code":""},{"path":"https://keras3.posit.co/dev/reference/Model.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Subclass the base Keras Model Class — Model","text":"model constructor function, can call create instance new model type.","code":""},{"path":"https://keras3.posit.co/dev/reference/Model.html","id":"symbols-in-scope","dir":"Reference","previous_headings":"","what":"Symbols in scope","title":"Subclass the base Keras Model Class — Model","text":"R function custom methods (public private) following symbols scope: self: custom class instance. super: custom class superclass. private: R environment specific class instance. objects assigned invisible Keras framework. __class__ .symbol(classname): custom class type object.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/activation_celu.html","id":null,"dir":"Reference","previous_headings":"","what":"Continuously Differentiable Exponential Linear Unit. — activation_celu","title":"Continuously Differentiable Exponential Linear Unit. — activation_celu","text":"CeLU activation function defined : celu(x) = alpha * (exp(x / alpha) - 1) x < 0,celu(x) = x x >= 0. alpha scaling parameter controls activation's shape.","code":""},{"path":"https://keras3.posit.co/dev/reference/activation_celu.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Continuously Differentiable Exponential Linear Unit. — activation_celu","text":"","code":"activation_celu(x, alpha = 1)"},{"path":"https://keras3.posit.co/dev/reference/activation_celu.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Continuously Differentiable Exponential Linear Unit. — activation_celu","text":"x Input tensor. alpha  value CeLU formulation. Defaults 1.0.","code":""},{"path":"https://keras3.posit.co/dev/reference/activation_celu.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Continuously Differentiable Exponential Linear Unit. — activation_celu","text":"tensor, result applying activation input tensor x.","code":""},{"path":"https://keras3.posit.co/dev/reference/activation_celu.html","id":"reference","dir":"Reference","previous_headings":"","what":"Reference","title":"Continuously Differentiable Exponential Linear Unit. — activation_celu","text":"Barron, J. T., 2017","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/activation_elu.html","id":null,"dir":"Reference","previous_headings":"","what":"Exponential Linear Unit. — activation_elu","title":"Exponential Linear Unit. — activation_elu","text":"exponential linear unit (ELU) alpha > 0 defined : x x > 0 alpha * exp(x) - 1 x < 0 ELUs negative values pushes mean activations closer zero. Mean activations closer zero enable faster learning bring gradient closer natural gradient. ELUs saturate negative value argument gets smaller. Saturation means small derivative decreases variation information propagated next layer.","code":""},{"path":"https://keras3.posit.co/dev/reference/activation_elu.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Exponential Linear Unit. — activation_elu","text":"","code":"activation_elu(x, alpha = 1)"},{"path":"https://keras3.posit.co/dev/reference/activation_elu.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Exponential Linear Unit. — activation_elu","text":"x Input tensor. alpha Numeric. See description details.","code":""},{"path":"https://keras3.posit.co/dev/reference/activation_elu.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Exponential Linear Unit. — activation_elu","text":"tensor, result applying activation input tensor x.","code":""},{"path":"https://keras3.posit.co/dev/reference/activation_elu.html","id":"reference","dir":"Reference","previous_headings":"","what":"Reference","title":"Exponential Linear Unit. — activation_elu","text":"Clevert et al., 2016","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/activation_exponential.html","id":null,"dir":"Reference","previous_headings":"","what":"Exponential activation function. — activation_exponential","title":"Exponential activation function. — activation_exponential","text":"Exponential activation function.","code":""},{"path":"https://keras3.posit.co/dev/reference/activation_exponential.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Exponential activation function. — activation_exponential","text":"","code":"activation_exponential(x)"},{"path":"https://keras3.posit.co/dev/reference/activation_exponential.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Exponential activation function. — activation_exponential","text":"x Input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/activation_exponential.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Exponential activation function. — activation_exponential","text":"tensor, result applying activation input tensor x.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/activation_gelu.html","id":null,"dir":"Reference","previous_headings":"","what":"Gaussian error linear unit (GELU) activation function. — activation_gelu","title":"Gaussian error linear unit (GELU) activation function. — activation_gelu","text":"Gaussian error linear unit (GELU) defined : gelu(x) = x * P(X <= x) P(X) ~ N(0, 1), .e. gelu(x) = 0.5 * x * (1 + erf(x / sqrt(2))). GELU weights inputs value, rather gating inputs sign ReLU.","code":""},{"path":"https://keras3.posit.co/dev/reference/activation_gelu.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Gaussian error linear unit (GELU) activation function. — activation_gelu","text":"","code":"activation_gelu(x, approximate = FALSE)"},{"path":"https://keras3.posit.co/dev/reference/activation_gelu.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Gaussian error linear unit (GELU) activation function. — activation_gelu","text":"x Input tensor. approximate bool, whether enable approximation.","code":""},{"path":"https://keras3.posit.co/dev/reference/activation_gelu.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Gaussian error linear unit (GELU) activation function. — activation_gelu","text":"tensor, result applying activation input tensor x.","code":""},{"path":"https://keras3.posit.co/dev/reference/activation_gelu.html","id":"reference","dir":"Reference","previous_headings":"","what":"Reference","title":"Gaussian error linear unit (GELU) activation function. — activation_gelu","text":"Hendrycks et al., 2016","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/activation_glu.html","id":null,"dir":"Reference","previous_headings":"","what":"Gated Linear Unit (GLU) activation function. — activation_glu","title":"Gated Linear Unit (GLU) activation function. — activation_glu","text":"GLU activation function defined : glu(x) = * sigmoid(b), x split two equal parts b along given axis.","code":""},{"path":"https://keras3.posit.co/dev/reference/activation_glu.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Gated Linear Unit (GLU) activation function. — activation_glu","text":"","code":"activation_glu(x, axis = -1L)"},{"path":"https://keras3.posit.co/dev/reference/activation_glu.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Gated Linear Unit (GLU) activation function. — activation_glu","text":"x Input tensor. axis axis along split input tensor. Defaults -1.","code":""},{"path":"https://keras3.posit.co/dev/reference/activation_glu.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Gated Linear Unit (GLU) activation function. — activation_glu","text":"tensor, result applying activation input tensor x.","code":""},{"path":"https://keras3.posit.co/dev/reference/activation_glu.html","id":"reference","dir":"Reference","previous_headings":"","what":"Reference","title":"Gated Linear Unit (GLU) activation function. — activation_glu","text":"Dauphin et al., 2017","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/activation_hard_shrink.html","id":null,"dir":"Reference","previous_headings":"","what":"Hard Shrink activation function. — activation_hard_shrink","title":"Hard Shrink activation function. — activation_hard_shrink","text":"defined : hard_shrink(x) = x |x| > threshold, hard_shrink(x) = 0 otherwise.","code":""},{"path":"https://keras3.posit.co/dev/reference/activation_hard_shrink.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Hard Shrink activation function. — activation_hard_shrink","text":"","code":"activation_hard_shrink(x, threshold = 0.5)"},{"path":"https://keras3.posit.co/dev/reference/activation_hard_shrink.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Hard Shrink activation function. — activation_hard_shrink","text":"x Input tensor. threshold Threshold value. Defaults 0.5.","code":""},{"path":"https://keras3.posit.co/dev/reference/activation_hard_shrink.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Hard Shrink activation function. — activation_hard_shrink","text":"tensor, result applying activation input tensor x.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/activation_hard_sigmoid.html","id":null,"dir":"Reference","previous_headings":"","what":"Hard sigmoid activation function. — activation_hard_sigmoid","title":"Hard sigmoid activation function. — activation_hard_sigmoid","text":"hard sigmoid activation defined : 0 x <= -3 1 x >= 3 (x/6) + 0.5 -3 < x < 3 faster, piecewise linear approximation sigmoid activation.","code":""},{"path":"https://keras3.posit.co/dev/reference/activation_hard_sigmoid.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Hard sigmoid activation function. — activation_hard_sigmoid","text":"","code":"activation_hard_sigmoid(x)"},{"path":"https://keras3.posit.co/dev/reference/activation_hard_sigmoid.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Hard sigmoid activation function. — activation_hard_sigmoid","text":"x Input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/activation_hard_sigmoid.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Hard sigmoid activation function. — activation_hard_sigmoid","text":"tensor, result applying activation input tensor x.","code":""},{"path":"https://keras3.posit.co/dev/reference/activation_hard_sigmoid.html","id":"reference","dir":"Reference","previous_headings":"","what":"Reference","title":"Hard sigmoid activation function. — activation_hard_sigmoid","text":"Wikipedia \"Hard sigmoid\"","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/activation_hard_silu.html","id":null,"dir":"Reference","previous_headings":"","what":"Hard SiLU activation function, also known as Hard Swish. — activation_hard_silu","title":"Hard SiLU activation function, also known as Hard Swish. — activation_hard_silu","text":"defined : 0 x < -3 x x > 3 x * (x + 3) / 6 -3 <= x <= 3 faster, piecewise linear approximation silu activation.","code":""},{"path":"https://keras3.posit.co/dev/reference/activation_hard_silu.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Hard SiLU activation function, also known as Hard Swish. — activation_hard_silu","text":"","code":"activation_hard_silu(x)  activation_hard_swish(x)"},{"path":"https://keras3.posit.co/dev/reference/activation_hard_silu.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Hard SiLU activation function, also known as Hard Swish. — activation_hard_silu","text":"x Input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/activation_hard_silu.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Hard SiLU activation function, also known as Hard Swish. — activation_hard_silu","text":"tensor, result applying activation input tensor x.","code":""},{"path":"https://keras3.posit.co/dev/reference/activation_hard_silu.html","id":"reference","dir":"Reference","previous_headings":"","what":"Reference","title":"Hard SiLU activation function, also known as Hard Swish. — activation_hard_silu","text":"Howard, 2019","code":""},{"path":"https://keras3.posit.co/dev/reference/activation_hard_tanh.html","id":null,"dir":"Reference","previous_headings":"","what":"HardTanh activation function. — activation_hard_tanh","title":"HardTanh activation function. — activation_hard_tanh","text":"defined : hard_tanh(x) = -1 x < -1, hard_tanh(x) = x -1 <= x <= 1, hard_tanh(x) = 1 x > 1.","code":""},{"path":"https://keras3.posit.co/dev/reference/activation_hard_tanh.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"HardTanh activation function. — activation_hard_tanh","text":"","code":"activation_hard_tanh(x)"},{"path":"https://keras3.posit.co/dev/reference/activation_hard_tanh.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"HardTanh activation function. — activation_hard_tanh","text":"x Input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/activation_hard_tanh.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"HardTanh activation function. — activation_hard_tanh","text":"tensor, result applying activation input tensor x.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/activation_leaky_relu.html","id":null,"dir":"Reference","previous_headings":"","what":"Leaky relu activation function. — activation_leaky_relu","title":"Leaky relu activation function. — activation_leaky_relu","text":"Leaky relu activation function.","code":""},{"path":"https://keras3.posit.co/dev/reference/activation_leaky_relu.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Leaky relu activation function. — activation_leaky_relu","text":"","code":"activation_leaky_relu(x, negative_slope = 0.2)"},{"path":"https://keras3.posit.co/dev/reference/activation_leaky_relu.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Leaky relu activation function. — activation_leaky_relu","text":"x Input tensor. negative_slope float controls slope values lower threshold.","code":""},{"path":"https://keras3.posit.co/dev/reference/activation_leaky_relu.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Leaky relu activation function. — activation_leaky_relu","text":"tensor, result applying activation input tensor x.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/activation_linear.html","id":null,"dir":"Reference","previous_headings":"","what":"Linear activation function (pass-through). — activation_linear","title":"Linear activation function (pass-through). — activation_linear","text":"\"linear\" activation identity function: returns input, unmodified.","code":""},{"path":"https://keras3.posit.co/dev/reference/activation_linear.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Linear activation function (pass-through). — activation_linear","text":"","code":"activation_linear(x)"},{"path":"https://keras3.posit.co/dev/reference/activation_linear.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Linear activation function (pass-through). — activation_linear","text":"x Input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/activation_linear.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Linear activation function (pass-through). — activation_linear","text":"tensor, result applying activation input tensor x.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/activation_log_sigmoid.html","id":null,"dir":"Reference","previous_headings":"","what":"Logarithm of the sigmoid activation function. — activation_log_sigmoid","title":"Logarithm of the sigmoid activation function. — activation_log_sigmoid","text":"defined f(x) = log(1 / (1 + exp(-x))).","code":""},{"path":"https://keras3.posit.co/dev/reference/activation_log_sigmoid.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Logarithm of the sigmoid activation function. — activation_log_sigmoid","text":"","code":"activation_log_sigmoid(x)"},{"path":"https://keras3.posit.co/dev/reference/activation_log_sigmoid.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Logarithm of the sigmoid activation function. — activation_log_sigmoid","text":"x Input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/activation_log_sigmoid.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Logarithm of the sigmoid activation function. — activation_log_sigmoid","text":"tensor, result applying activation input tensor x.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/activation_log_softmax.html","id":null,"dir":"Reference","previous_headings":"","what":"Log-Softmax activation function. — activation_log_softmax","title":"Log-Softmax activation function. — activation_log_softmax","text":"input vector handled independently. axis argument sets axis input function applied along.","code":""},{"path":"https://keras3.posit.co/dev/reference/activation_log_softmax.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Log-Softmax activation function. — activation_log_softmax","text":"","code":"activation_log_softmax(x, axis = -1L)"},{"path":"https://keras3.posit.co/dev/reference/activation_log_softmax.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Log-Softmax activation function. — activation_log_softmax","text":"x Input tensor. axis Integer, axis along softmax applied.","code":""},{"path":"https://keras3.posit.co/dev/reference/activation_log_softmax.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Log-Softmax activation function. — activation_log_softmax","text":"tensor, result applying activation input tensor x.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/activation_mish.html","id":null,"dir":"Reference","previous_headings":"","what":"Mish activation function. — activation_mish","title":"Mish activation function. — activation_mish","text":"defined : mish(x) = x * tanh(softplus(x)) softplus defined : softplus(x) = log(exp(x) + 1)","code":""},{"path":"https://keras3.posit.co/dev/reference/activation_mish.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Mish activation function. — activation_mish","text":"","code":"activation_mish(x)"},{"path":"https://keras3.posit.co/dev/reference/activation_mish.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Mish activation function. — activation_mish","text":"x Input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/activation_mish.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Mish activation function. — activation_mish","text":"tensor, result applying activation input tensor x.","code":""},{"path":"https://keras3.posit.co/dev/reference/activation_mish.html","id":"reference","dir":"Reference","previous_headings":"","what":"Reference","title":"Mish activation function. — activation_mish","text":"Misra, 2019","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/activation_relu.html","id":null,"dir":"Reference","previous_headings":"","what":"Applies the rectified linear unit activation function. — activation_relu","title":"Applies the rectified linear unit activation function. — activation_relu","text":"default values, returns standard ReLU activation: max(x, 0), element-wise maximum 0 input tensor. Modifying default parameters allows use non-zero thresholds, change max value activation, use non-zero multiple input values threshold.","code":""},{"path":"https://keras3.posit.co/dev/reference/activation_relu.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Applies the rectified linear unit activation function. — activation_relu","text":"","code":"activation_relu(x, negative_slope = 0, max_value = NULL, threshold = 0)"},{"path":"https://keras3.posit.co/dev/reference/activation_relu.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Applies the rectified linear unit activation function. — activation_relu","text":"x Input tensor. negative_slope numeric controls slope values lower threshold. max_value numeric sets saturation threshold (largest value function return). threshold numeric giving threshold value activation function values damped set zero.","code":""},{"path":"https://keras3.posit.co/dev/reference/activation_relu.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Applies the rectified linear unit activation function. — activation_relu","text":"tensor shape dtype input x.","code":""},{"path":"https://keras3.posit.co/dev/reference/activation_relu.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Applies the rectified linear unit activation function. — activation_relu","text":"","code":"x <- c(-10, -5, 0, 5, 10) activation_relu(x) ## tf.Tensor([ 0.  0.  0.  5. 10.], shape=(5), dtype=float32) activation_relu(x, negative_slope = 0.5) ## tf.Tensor([-5.  -2.5  0.   5.  10. ], shape=(5), dtype=float32) activation_relu(x, max_value = 5) ## tf.Tensor([0. 0. 0. 5. 5.], shape=(5), dtype=float32) activation_relu(x, threshold = 5) ## tf.Tensor([-0. -0.  0.  0. 10.], shape=(5), dtype=float32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/activation_relu6.html","id":null,"dir":"Reference","previous_headings":"","what":"Relu6 activation function. — activation_relu6","title":"Relu6 activation function. — activation_relu6","text":"ReLU function, truncated maximum value 6.","code":""},{"path":"https://keras3.posit.co/dev/reference/activation_relu6.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Relu6 activation function. — activation_relu6","text":"","code":"activation_relu6(x)"},{"path":"https://keras3.posit.co/dev/reference/activation_relu6.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Relu6 activation function. — activation_relu6","text":"x Input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/activation_relu6.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Relu6 activation function. — activation_relu6","text":"tensor, result applying activation input tensor x.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/activation_selu.html","id":null,"dir":"Reference","previous_headings":"","what":"Scaled Exponential Linear Unit (SELU). — activation_selu","title":"Scaled Exponential Linear Unit (SELU). — activation_selu","text":"Scaled Exponential Linear Unit (SELU) activation function defined : scale * x x > 0 scale * alpha * (exp(x) - 1) x < 0 alpha scale pre-defined constants (alpha = 1.67326324 scale = 1.05070098). Basically, SELU activation function multiplies scale (> 1) output activation_elu function ensure slope larger one positive inputs. values alpha scale chosen mean variance inputs preserved two consecutive layers long weights initialized correctly (see initializer_lecun_normal()) number input units \"large enough\" (see reference paper information).","code":""},{"path":"https://keras3.posit.co/dev/reference/activation_selu.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Scaled Exponential Linear Unit (SELU). — activation_selu","text":"","code":"activation_selu(x)"},{"path":"https://keras3.posit.co/dev/reference/activation_selu.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Scaled Exponential Linear Unit (SELU). — activation_selu","text":"x Input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/activation_selu.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Scaled Exponential Linear Unit (SELU). — activation_selu","text":"tensor, result applying activation input tensor x.","code":""},{"path":"https://keras3.posit.co/dev/reference/activation_selu.html","id":"notes","dir":"Reference","previous_headings":"","what":"Notes","title":"Scaled Exponential Linear Unit (SELU). — activation_selu","text":"used together initializer_lecun_normal(). used together dropout variant layer_alpha_dropout() (legacy, depracated).","code":""},{"path":"https://keras3.posit.co/dev/reference/activation_selu.html","id":"reference","dir":"Reference","previous_headings":"","what":"Reference","title":"Scaled Exponential Linear Unit (SELU). — activation_selu","text":"Klambauer et al., 2017","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/activation_sigmoid.html","id":null,"dir":"Reference","previous_headings":"","what":"Sigmoid activation function. — activation_sigmoid","title":"Sigmoid activation function. — activation_sigmoid","text":"defined : sigmoid(x) = 1 / (1 + exp(-x)). small values (<-5), sigmoid returns value close zero, large values (>5) result function gets close 1. Sigmoid equivalent 2-element softmax, second element assumed zero. sigmoid function always returns value 0 1.","code":""},{"path":"https://keras3.posit.co/dev/reference/activation_sigmoid.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Sigmoid activation function. — activation_sigmoid","text":"","code":"activation_sigmoid(x)"},{"path":"https://keras3.posit.co/dev/reference/activation_sigmoid.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Sigmoid activation function. — activation_sigmoid","text":"x Input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/activation_sigmoid.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Sigmoid activation function. — activation_sigmoid","text":"tensor, result applying activation input tensor x.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/activation_silu.html","id":null,"dir":"Reference","previous_headings":"","what":"Swish (or Silu) activation function. — activation_silu","title":"Swish (or Silu) activation function. — activation_silu","text":"defined : swish(x) = x * sigmoid(x). Swish (Silu) activation function smooth, non-monotonic function unbounded bounded .","code":""},{"path":"https://keras3.posit.co/dev/reference/activation_silu.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Swish (or Silu) activation function. — activation_silu","text":"","code":"activation_silu(x)"},{"path":"https://keras3.posit.co/dev/reference/activation_silu.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Swish (or Silu) activation function. — activation_silu","text":"x Input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/activation_silu.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Swish (or Silu) activation function. — activation_silu","text":"tensor, result applying activation input tensor x.","code":""},{"path":"https://keras3.posit.co/dev/reference/activation_silu.html","id":"reference","dir":"Reference","previous_headings":"","what":"Reference","title":"Swish (or Silu) activation function. — activation_silu","text":"Ramachandran et al., 2017","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/activation_soft_shrink.html","id":null,"dir":"Reference","previous_headings":"","what":"Soft Shrink activation function. — activation_soft_shrink","title":"Soft Shrink activation function. — activation_soft_shrink","text":"defined : soft_shrink(x) = x - threshold x > threshold, soft_shrink(x) = x + threshold x < -threshold, soft_shrink(x) = 0 otherwise.","code":""},{"path":"https://keras3.posit.co/dev/reference/activation_soft_shrink.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Soft Shrink activation function. — activation_soft_shrink","text":"","code":"activation_soft_shrink(x, threshold = 0.5)"},{"path":"https://keras3.posit.co/dev/reference/activation_soft_shrink.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Soft Shrink activation function. — activation_soft_shrink","text":"x Input tensor. threshold Threshold value. Defaults 0.5.","code":""},{"path":"https://keras3.posit.co/dev/reference/activation_soft_shrink.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Soft Shrink activation function. — activation_soft_shrink","text":"tensor, result applying activation input tensor x.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/activation_softmax.html","id":null,"dir":"Reference","previous_headings":"","what":"Softmax converts a vector of values to a probability distribution. — activation_softmax","title":"Softmax converts a vector of values to a probability distribution. — activation_softmax","text":"elements output vector range [0, 1] sum 1. input vector handled independently. axis argument sets axis input function applied along. Softmax often used activation last layer classification network result interpreted probability distribution. softmax vector x computed exp(x) / sum(exp(x)). input values log-odds resulting probability.","code":""},{"path":"https://keras3.posit.co/dev/reference/activation_softmax.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Softmax converts a vector of values to a probability distribution. — activation_softmax","text":"","code":"activation_softmax(x, axis = -1L)"},{"path":"https://keras3.posit.co/dev/reference/activation_softmax.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Softmax converts a vector of values to a probability distribution. — activation_softmax","text":"x Input tensor. axis Integer, axis along softmax applied.","code":""},{"path":"https://keras3.posit.co/dev/reference/activation_softmax.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Softmax converts a vector of values to a probability distribution. — activation_softmax","text":"tensor, result applying activation input tensor x.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/activation_softplus.html","id":null,"dir":"Reference","previous_headings":"","what":"Softplus activation function. — activation_softplus","title":"Softplus activation function. — activation_softplus","text":"defined : softplus(x) = log(exp(x) + 1).","code":""},{"path":"https://keras3.posit.co/dev/reference/activation_softplus.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Softplus activation function. — activation_softplus","text":"","code":"activation_softplus(x)"},{"path":"https://keras3.posit.co/dev/reference/activation_softplus.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Softplus activation function. — activation_softplus","text":"x Input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/activation_softplus.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Softplus activation function. — activation_softplus","text":"tensor, result applying activation input tensor x.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/activation_softsign.html","id":null,"dir":"Reference","previous_headings":"","what":"Softsign activation function. — activation_softsign","title":"Softsign activation function. — activation_softsign","text":"Softsign defined : softsign(x) = x / (abs(x) + 1).","code":""},{"path":"https://keras3.posit.co/dev/reference/activation_softsign.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Softsign activation function. — activation_softsign","text":"","code":"activation_softsign(x)"},{"path":"https://keras3.posit.co/dev/reference/activation_softsign.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Softsign activation function. — activation_softsign","text":"x Input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/activation_softsign.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Softsign activation function. — activation_softsign","text":"tensor, result applying activation input tensor x.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/activation_sparse_plus.html","id":null,"dir":"Reference","previous_headings":"","what":"SparsePlus activation function. — activation_sparse_plus","title":"SparsePlus activation function. — activation_sparse_plus","text":"SparsePlus defined : sparse_plus(x) = 0 x <= -1. sparse_plus(x) = (1/4) * (x + 1)^2 -1 < x < 1. sparse_plus(x) = x x >= 1.","code":""},{"path":"https://keras3.posit.co/dev/reference/activation_sparse_plus.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"SparsePlus activation function. — activation_sparse_plus","text":"","code":"activation_sparse_plus(x)"},{"path":"https://keras3.posit.co/dev/reference/activation_sparse_plus.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"SparsePlus activation function. — activation_sparse_plus","text":"x Input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/activation_sparse_plus.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"SparsePlus activation function. — activation_sparse_plus","text":"tensor, result applying activation input tensor x.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/activation_sparsemax.html","id":null,"dir":"Reference","previous_headings":"","what":"Sparsemax activation function. — activation_sparsemax","title":"Sparsemax activation function. — activation_sparsemax","text":"batch , class j, sparsemax activation function defined : sparsemax(x)[, j] = max(x[, j] - (x[, :]), 0).","code":""},{"path":"https://keras3.posit.co/dev/reference/activation_sparsemax.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Sparsemax activation function. — activation_sparsemax","text":"","code":"activation_sparsemax(x, axis = -1L)"},{"path":"https://keras3.posit.co/dev/reference/activation_sparsemax.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Sparsemax activation function. — activation_sparsemax","text":"x Input tensor. axis int, axis along sparsemax operation applied. (1-based)","code":""},{"path":"https://keras3.posit.co/dev/reference/activation_sparsemax.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Sparsemax activation function. — activation_sparsemax","text":"tensor, output sparsemax transformation. type shape x.","code":""},{"path":"https://keras3.posit.co/dev/reference/activation_sparsemax.html","id":"reference","dir":"Reference","previous_headings":"","what":"Reference","title":"Sparsemax activation function. — activation_sparsemax","text":"Martins et.al., 2016","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/activation_squareplus.html","id":null,"dir":"Reference","previous_headings":"","what":"Squareplus activation function. — activation_squareplus","title":"Squareplus activation function. — activation_squareplus","text":"Squareplus activation function defined : f(x) = (x + sqrt(x^2 + b)) / 2 b smoothness parameter.","code":""},{"path":"https://keras3.posit.co/dev/reference/activation_squareplus.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Squareplus activation function. — activation_squareplus","text":"","code":"activation_squareplus(x, b = 4L)"},{"path":"https://keras3.posit.co/dev/reference/activation_squareplus.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Squareplus activation function. — activation_squareplus","text":"x Input tensor. b Smoothness parameter. Defaults 4.","code":""},{"path":"https://keras3.posit.co/dev/reference/activation_squareplus.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Squareplus activation function. — activation_squareplus","text":"tensor, result applying activation input tensor x.","code":""},{"path":"https://keras3.posit.co/dev/reference/activation_squareplus.html","id":"reference","dir":"Reference","previous_headings":"","what":"Reference","title":"Squareplus activation function. — activation_squareplus","text":"Ramachandran et al., 2021","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/activation_tanh.html","id":null,"dir":"Reference","previous_headings":"","what":"Hyperbolic tangent activation function. — activation_tanh","title":"Hyperbolic tangent activation function. — activation_tanh","text":"defined : tanh(x) = sinh(x) / cosh(x), .e. tanh(x) = ((exp(x) - exp(-x)) / (exp(x) + exp(-x))).","code":""},{"path":"https://keras3.posit.co/dev/reference/activation_tanh.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Hyperbolic tangent activation function. — activation_tanh","text":"","code":"activation_tanh(x)"},{"path":"https://keras3.posit.co/dev/reference/activation_tanh.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Hyperbolic tangent activation function. — activation_tanh","text":"x Input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/activation_tanh.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Hyperbolic tangent activation function. — activation_tanh","text":"tensor, result applying activation input tensor x.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/activation_tanh_shrink.html","id":null,"dir":"Reference","previous_headings":"","what":"Tanh shrink activation function. — activation_tanh_shrink","title":"Tanh shrink activation function. — activation_tanh_shrink","text":"defined : f(x) = x - tanh(x).","code":""},{"path":"https://keras3.posit.co/dev/reference/activation_tanh_shrink.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Tanh shrink activation function. — activation_tanh_shrink","text":"","code":"activation_tanh_shrink(x)"},{"path":"https://keras3.posit.co/dev/reference/activation_tanh_shrink.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Tanh shrink activation function. — activation_tanh_shrink","text":"x Input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/activation_tanh_shrink.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Tanh shrink activation function. — activation_tanh_shrink","text":"tensor, result applying activation input tensor x.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/activation_threshold.html","id":null,"dir":"Reference","previous_headings":"","what":"Threshold activation function. — activation_threshold","title":"Threshold activation function. — activation_threshold","text":"defined : threshold(x) = x x > threshold, threshold(x) = default_value otherwise.","code":""},{"path":"https://keras3.posit.co/dev/reference/activation_threshold.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Threshold activation function. — activation_threshold","text":"","code":"activation_threshold(x, threshold, default_value)"},{"path":"https://keras3.posit.co/dev/reference/activation_threshold.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Threshold activation function. — activation_threshold","text":"x Input tensor. threshold value decides retain replace x. default_value Value assign x <= threshold.","code":""},{"path":"https://keras3.posit.co/dev/reference/activation_threshold.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Threshold activation function. — activation_threshold","text":"tensor, result applying activation input tensor x.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/active_property.html","id":null,"dir":"Reference","previous_headings":"","what":"Create an active property class method — active_property","title":"Create an active property class method — active_property","text":"Create active property class method","code":""},{"path":"https://keras3.posit.co/dev/reference/active_property.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Create an active property class method — active_property","text":"","code":"active_property(fn)"},{"path":"https://keras3.posit.co/dev/reference/active_property.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Create an active property class method — active_property","text":"fn R function","code":""},{"path":"https://keras3.posit.co/dev/reference/active_property.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Create an active property class method — active_property","text":"fn, additional R attribute cause fn converted active property converted method custom subclass.","code":""},{"path":"https://keras3.posit.co/dev/reference/active_property.html","id":"example","dir":"Reference","previous_headings":"","what":"Example","title":"Create an active property class method — active_property","text":"","code":"layer_foo <- Model(\"Foo\", ...,   metrics = active_property(function() {     list(self$d_loss_metric,          self$g_loss_metric)   }))"},{"path":"https://keras3.posit.co/dev/reference/adapt.html","id":null,"dir":"Reference","previous_headings":"","what":"Fits the state of the preprocessing layer to the data being passed — adapt","title":"Fits the state of the preprocessing layer to the data being passed — adapt","text":"Fits state preprocessing layer data passed","code":""},{"path":"https://keras3.posit.co/dev/reference/adapt.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Fits the state of the preprocessing layer to the data being passed — adapt","text":"","code":"adapt(object, data, ..., batch_size = NULL, steps = NULL)"},{"path":"https://keras3.posit.co/dev/reference/adapt.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Fits the state of the preprocessing layer to the data being passed — adapt","text":"object Preprocessing layer object data data train . can passed either tf.data.Dataset R array. ... Used forwards backwards compatibility. Passed underlying method. batch_size Integer NULL. Number asamples per state update. unspecified, batch_size default 32. specify batch_size data form TF Dataset generator (since generate batches). steps Integer NULL. Total number steps (batches samples) training input tensors TensorFlow data tensors, default NULL equal number samples dataset divided batch size, 1 determined. x tf.data.Dataset, steps NULL, epoch run input dataset exhausted. passing infinitely repeating dataset, must specify steps argument. argument supported array inputs.","code":""},{"path":"https://keras3.posit.co/dev/reference/adapt.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Fits the state of the preprocessing layer to the data being passed — adapt","text":"Returns object, invisibly.","code":""},{"path":"https://keras3.posit.co/dev/reference/adapt.html","id":"details","dir":"Reference","previous_headings":"","what":"Details","title":"Fits the state of the preprocessing layer to the data being passed — adapt","text":"calling adapt layer, preprocessing layer's state update training. order make preprocessing layers efficient distribution context, kept constant respect compiled tf.Graphs call layer. affect layer use adapting layer , adapt layer multiple times need take care re-compile compiled functions follows: adding preprocessing layer keras model, need call compile(model) subsequent call adapt(). calling preprocessing layer inside tfdatasets::dataset_map(), call dataset_map() input Dataset adapt(). using tensorflow::tf_function() directly calls preprocessing layer, need call tf_function() callable subsequent call adapt(). keras_model() example multiple adapts:         tfdatasets example multiple adapts:","code":"layer <- layer_normalization(axis = NULL) adapt(layer, c(0, 2)) model <- keras_model_sequential() |> layer() predict(model, c(0, 1, 2), verbose = FALSE) # [1] -1  0  1 ## [1] -1  0  1 adapt(layer, c(-1, 1)) compile(model)  # This is needed to re-compile model.predict! predict(model, c(0, 1, 2), verbose = FALSE) # [1] 0 1 2 ## [1] 0 1 2 layer <- layer_normalization(axis = NULL) adapt(layer, c(0, 2)) input_ds <- tfdatasets::range_dataset(0, 3) normalized_ds <- input_ds |>   tfdatasets::dataset_map(layer) str(tfdatasets::iterate(normalized_ds)) ## List of 3 ##  $ :<tf.Tensor: shape=(1), dtype=float32, numpy=array([-1.], dtype=float32)> ##  $ :<tf.Tensor: shape=(1), dtype=float32, numpy=array([0.], dtype=float32)> ##  $ :<tf.Tensor: shape=(1), dtype=float32, numpy=array([1.], dtype=float32)> adapt(layer, c(-1, 1)) normalized_ds <- input_ds |>   tfdatasets::dataset_map(layer) # Re-map over the input dataset.  normalized_ds |>   tfdatasets::as_array_iterator() |>   tfdatasets::iterate(simplify = FALSE) |>   str() ## List of 3 ##  $ : num [1(1d)] 0 ##  $ : num [1(1d)] 1 ##  $ : num [1(1d)] 2"},{"path":"https://keras3.posit.co/dev/reference/application_convnext_base.html","id":null,"dir":"Reference","previous_headings":"","what":"Instantiates the ConvNeXtBase architecture. — application_convnext_base","title":"Instantiates the ConvNeXtBase architecture. — application_convnext_base","text":"Instantiates ConvNeXtBase architecture.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_convnext_base.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Instantiates the ConvNeXtBase architecture. — application_convnext_base","text":"","code":"application_convnext_base(   include_top = TRUE,   include_preprocessing = TRUE,   weights = \"imagenet\",   input_tensor = NULL,   input_shape = NULL,   pooling = NULL,   classes = 1000L,   classifier_activation = \"softmax\",   name = \"convnext_base\" )"},{"path":"https://keras3.posit.co/dev/reference/application_convnext_base.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Instantiates the ConvNeXtBase architecture. — application_convnext_base","text":"include_top Whether include fully-connected layer top network. Defaults TRUE. include_preprocessing Boolean, whether include preprocessing layer bottom network. weights One NULL (random initialization), \"imagenet\" (pre-training ImageNet-1k), path weights file loaded. Defaults \"imagenet\". input_tensor Optional Keras tensor (.e. output keras_input()) use image input model. input_shape Optional shape tuple, specified include_top FALSE. exactly 3 inputs channels. pooling Optional pooling mode feature extraction include_top FALSE. Defaults NULL. NULL means output model 4D tensor output last convolutional layer. avg means global average pooling applied output last convolutional layer, thus output model 2D tensor. max means global max pooling applied. classes Optional number classes classify images , specified include_top TRUE, weights argument specified. Defaults 1000 (number ImageNet classes). classifier_activation str callable. activation function use \"top\" layer. Ignored unless include_top=TRUE. Set classifier_activation=NULL return logits \"top\" layer. Defaults \"softmax\". loading pretrained weights, classifier_activation can NULL \"softmax\". name name model (string).","code":""},{"path":"https://keras3.posit.co/dev/reference/application_convnext_base.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Instantiates the ConvNeXtBase architecture. — application_convnext_base","text":"model instance.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_convnext_base.html","id":"references","dir":"Reference","previous_headings":"","what":"References","title":"Instantiates the ConvNeXtBase architecture. — application_convnext_base","text":"ConvNet 2020s (CVPR 2022) image classification use cases, see page detailed examples. transfer learning use cases, make sure read guide transfer learning & fine-tuning. base, large, xlarge models first pre-trained ImageNet-21k dataset fine-tuned ImageNet-1k dataset. pre-trained parameters models assembled official repository. get sense parameters converted Keras compatible parameters, please refer repository.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_convnext_base.html","id":"note","dir":"Reference","previous_headings":"","what":"Note","title":"Instantiates the ConvNeXtBase architecture. — application_convnext_base","text":"Keras Application expects specific kind input preprocessing. ConvNeXt, preprocessing included model using Normalization layer.  ConvNeXt models expect inputs float uint8 tensors pixels values [0-255] range. calling summary() method instantiating ConvNeXt model, prefer setting expand_nested argument summary() TRUE better investigate instantiated model.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/application_convnext_large.html","id":null,"dir":"Reference","previous_headings":"","what":"Instantiates the ConvNeXtLarge architecture. — application_convnext_large","title":"Instantiates the ConvNeXtLarge architecture. — application_convnext_large","text":"Instantiates ConvNeXtLarge architecture.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_convnext_large.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Instantiates the ConvNeXtLarge architecture. — application_convnext_large","text":"","code":"application_convnext_large(   include_top = TRUE,   include_preprocessing = TRUE,   weights = \"imagenet\",   input_tensor = NULL,   input_shape = NULL,   pooling = NULL,   classes = 1000L,   classifier_activation = \"softmax\",   name = \"convnext_large\" )"},{"path":"https://keras3.posit.co/dev/reference/application_convnext_large.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Instantiates the ConvNeXtLarge architecture. — application_convnext_large","text":"include_top Whether include fully-connected layer top network. Defaults TRUE. include_preprocessing Boolean, whether include preprocessing layer bottom network. weights One NULL (random initialization), \"imagenet\" (pre-training ImageNet-1k), path weights file loaded. Defaults \"imagenet\". input_tensor Optional Keras tensor (.e. output keras_input()) use image input model. input_shape Optional shape tuple, specified include_top FALSE. exactly 3 inputs channels. pooling Optional pooling mode feature extraction include_top FALSE. Defaults NULL. NULL means output model 4D tensor output last convolutional layer. avg means global average pooling applied output last convolutional layer, thus output model 2D tensor. max means global max pooling applied. classes Optional number classes classify images , specified include_top TRUE, weights argument specified. Defaults 1000 (number ImageNet classes). classifier_activation str callable. activation function use \"top\" layer. Ignored unless include_top=TRUE. Set classifier_activation=NULL return logits \"top\" layer. Defaults \"softmax\". loading pretrained weights, classifier_activation can NULL \"softmax\". name name model (string).","code":""},{"path":"https://keras3.posit.co/dev/reference/application_convnext_large.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Instantiates the ConvNeXtLarge architecture. — application_convnext_large","text":"model instance.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_convnext_large.html","id":"references","dir":"Reference","previous_headings":"","what":"References","title":"Instantiates the ConvNeXtLarge architecture. — application_convnext_large","text":"ConvNet 2020s (CVPR 2022) image classification use cases, see page detailed examples. transfer learning use cases, make sure read guide transfer learning & fine-tuning. base, large, xlarge models first pre-trained ImageNet-21k dataset fine-tuned ImageNet-1k dataset. pre-trained parameters models assembled official repository. get sense parameters converted Keras compatible parameters, please refer repository.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_convnext_large.html","id":"note","dir":"Reference","previous_headings":"","what":"Note","title":"Instantiates the ConvNeXtLarge architecture. — application_convnext_large","text":"Keras Application expects specific kind input preprocessing. ConvNeXt, preprocessing included model using Normalization layer.  ConvNeXt models expect inputs float uint8 tensors pixels values [0-255] range. calling summary() method instantiating ConvNeXt model, prefer setting expand_nested argument summary() TRUE better investigate instantiated model.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/application_convnext_small.html","id":null,"dir":"Reference","previous_headings":"","what":"Instantiates the ConvNeXtSmall architecture. — application_convnext_small","title":"Instantiates the ConvNeXtSmall architecture. — application_convnext_small","text":"Instantiates ConvNeXtSmall architecture.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_convnext_small.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Instantiates the ConvNeXtSmall architecture. — application_convnext_small","text":"","code":"application_convnext_small(   include_top = TRUE,   include_preprocessing = TRUE,   weights = \"imagenet\",   input_tensor = NULL,   input_shape = NULL,   pooling = NULL,   classes = 1000L,   classifier_activation = \"softmax\",   name = \"convnext_small\" )"},{"path":"https://keras3.posit.co/dev/reference/application_convnext_small.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Instantiates the ConvNeXtSmall architecture. — application_convnext_small","text":"include_top Whether include fully-connected layer top network. Defaults TRUE. include_preprocessing Boolean, whether include preprocessing layer bottom network. weights One NULL (random initialization), \"imagenet\" (pre-training ImageNet-1k), path weights file loaded. Defaults \"imagenet\". input_tensor Optional Keras tensor (.e. output keras_input()) use image input model. input_shape Optional shape tuple, specified include_top FALSE. exactly 3 inputs channels. pooling Optional pooling mode feature extraction include_top FALSE. Defaults NULL. NULL means output model 4D tensor output last convolutional layer. avg means global average pooling applied output last convolutional layer, thus output model 2D tensor. max means global max pooling applied. classes Optional number classes classify images , specified include_top TRUE, weights argument specified. Defaults 1000 (number ImageNet classes). classifier_activation str callable. activation function use \"top\" layer. Ignored unless include_top=TRUE. Set classifier_activation=NULL return logits \"top\" layer. Defaults \"softmax\". loading pretrained weights, classifier_activation can NULL \"softmax\". name name model (string).","code":""},{"path":"https://keras3.posit.co/dev/reference/application_convnext_small.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Instantiates the ConvNeXtSmall architecture. — application_convnext_small","text":"model instance.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_convnext_small.html","id":"references","dir":"Reference","previous_headings":"","what":"References","title":"Instantiates the ConvNeXtSmall architecture. — application_convnext_small","text":"ConvNet 2020s (CVPR 2022) image classification use cases, see page detailed examples. transfer learning use cases, make sure read guide transfer learning & fine-tuning. base, large, xlarge models first pre-trained ImageNet-21k dataset fine-tuned ImageNet-1k dataset. pre-trained parameters models assembled official repository. get sense parameters converted Keras compatible parameters, please refer repository.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_convnext_small.html","id":"note","dir":"Reference","previous_headings":"","what":"Note","title":"Instantiates the ConvNeXtSmall architecture. — application_convnext_small","text":"Keras Application expects specific kind input preprocessing. ConvNeXt, preprocessing included model using Normalization layer.  ConvNeXt models expect inputs float uint8 tensors pixels values [0-255] range. calling summary() method instantiating ConvNeXt model, prefer setting expand_nested argument summary() TRUE better investigate instantiated model.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/application_convnext_tiny.html","id":null,"dir":"Reference","previous_headings":"","what":"Instantiates the ConvNeXtTiny architecture. — application_convnext_tiny","title":"Instantiates the ConvNeXtTiny architecture. — application_convnext_tiny","text":"Instantiates ConvNeXtTiny architecture.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_convnext_tiny.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Instantiates the ConvNeXtTiny architecture. — application_convnext_tiny","text":"","code":"application_convnext_tiny(   include_top = TRUE,   include_preprocessing = TRUE,   weights = \"imagenet\",   input_tensor = NULL,   input_shape = NULL,   pooling = NULL,   classes = 1000L,   classifier_activation = \"softmax\",   name = \"convnext_tiny\" )"},{"path":"https://keras3.posit.co/dev/reference/application_convnext_tiny.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Instantiates the ConvNeXtTiny architecture. — application_convnext_tiny","text":"include_top Whether include fully-connected layer top network. Defaults TRUE. include_preprocessing Boolean, whether include preprocessing layer bottom network. weights One NULL (random initialization), \"imagenet\" (pre-training ImageNet-1k), path weights file loaded. Defaults \"imagenet\". input_tensor Optional Keras tensor (.e. output keras_input()) use image input model. input_shape Optional shape tuple, specified include_top FALSE. exactly 3 inputs channels. pooling Optional pooling mode feature extraction include_top FALSE. Defaults NULL. NULL means output model 4D tensor output last convolutional layer. avg means global average pooling applied output last convolutional layer, thus output model 2D tensor. max means global max pooling applied. classes Optional number classes classify images , specified include_top TRUE, weights argument specified. Defaults 1000 (number ImageNet classes). classifier_activation str callable. activation function use \"top\" layer. Ignored unless include_top=TRUE. Set classifier_activation=NULL return logits \"top\" layer. Defaults \"softmax\". loading pretrained weights, classifier_activation can NULL \"softmax\". name name model (string).","code":""},{"path":"https://keras3.posit.co/dev/reference/application_convnext_tiny.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Instantiates the ConvNeXtTiny architecture. — application_convnext_tiny","text":"model instance.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_convnext_tiny.html","id":"references","dir":"Reference","previous_headings":"","what":"References","title":"Instantiates the ConvNeXtTiny architecture. — application_convnext_tiny","text":"ConvNet 2020s (CVPR 2022) image classification use cases, see page detailed examples. transfer learning use cases, make sure read guide transfer learning & fine-tuning. base, large, xlarge models first pre-trained ImageNet-21k dataset fine-tuned ImageNet-1k dataset. pre-trained parameters models assembled official repository. get sense parameters converted Keras compatible parameters, please refer repository.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_convnext_tiny.html","id":"note","dir":"Reference","previous_headings":"","what":"Note","title":"Instantiates the ConvNeXtTiny architecture. — application_convnext_tiny","text":"Keras Application expects specific kind input preprocessing. ConvNeXt, preprocessing included model using Normalization layer.  ConvNeXt models expect inputs float uint8 tensors pixels values [0-255] range. calling summary() method instantiating ConvNeXt model, prefer setting expand_nested argument summary() TRUE better investigate instantiated model.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/application_convnext_xlarge.html","id":null,"dir":"Reference","previous_headings":"","what":"Instantiates the ConvNeXtXLarge architecture. — application_convnext_xlarge","title":"Instantiates the ConvNeXtXLarge architecture. — application_convnext_xlarge","text":"Instantiates ConvNeXtXLarge architecture.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_convnext_xlarge.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Instantiates the ConvNeXtXLarge architecture. — application_convnext_xlarge","text":"","code":"application_convnext_xlarge(   include_top = TRUE,   include_preprocessing = TRUE,   weights = \"imagenet\",   input_tensor = NULL,   input_shape = NULL,   pooling = NULL,   classes = 1000L,   classifier_activation = \"softmax\",   name = \"convnext_xlarge\" )"},{"path":"https://keras3.posit.co/dev/reference/application_convnext_xlarge.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Instantiates the ConvNeXtXLarge architecture. — application_convnext_xlarge","text":"include_top Whether include fully-connected layer top network. Defaults TRUE. include_preprocessing Boolean, whether include preprocessing layer bottom network. weights One NULL (random initialization), \"imagenet\" (pre-training ImageNet-1k), path weights file loaded. Defaults \"imagenet\". input_tensor Optional Keras tensor (.e. output keras_input()) use image input model. input_shape Optional shape tuple, specified include_top FALSE. exactly 3 inputs channels. pooling Optional pooling mode feature extraction include_top FALSE. Defaults NULL. NULL means output model 4D tensor output last convolutional layer. avg means global average pooling applied output last convolutional layer, thus output model 2D tensor. max means global max pooling applied. classes Optional number classes classify images , specified include_top TRUE, weights argument specified. Defaults 1000 (number ImageNet classes). classifier_activation str callable. activation function use \"top\" layer. Ignored unless include_top=TRUE. Set classifier_activation=NULL return logits \"top\" layer. Defaults \"softmax\". loading pretrained weights, classifier_activation can NULL \"softmax\". name name model (string).","code":""},{"path":"https://keras3.posit.co/dev/reference/application_convnext_xlarge.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Instantiates the ConvNeXtXLarge architecture. — application_convnext_xlarge","text":"model instance.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_convnext_xlarge.html","id":"references","dir":"Reference","previous_headings":"","what":"References","title":"Instantiates the ConvNeXtXLarge architecture. — application_convnext_xlarge","text":"ConvNet 2020s (CVPR 2022) image classification use cases, see page detailed examples. transfer learning use cases, make sure read guide transfer learning & fine-tuning. base, large, xlarge models first pre-trained ImageNet-21k dataset fine-tuned ImageNet-1k dataset. pre-trained parameters models assembled official repository. get sense parameters converted Keras compatible parameters, please refer repository.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_convnext_xlarge.html","id":"note","dir":"Reference","previous_headings":"","what":"Note","title":"Instantiates the ConvNeXtXLarge architecture. — application_convnext_xlarge","text":"Keras Application expects specific kind input preprocessing. ConvNeXt, preprocessing included model using Normalization layer.  ConvNeXt models expect inputs float uint8 tensors pixels values [0-255] range. calling summary() method instantiating ConvNeXt model, prefer setting expand_nested argument summary() TRUE better investigate instantiated model.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/application_densenet121.html","id":null,"dir":"Reference","previous_headings":"","what":"Instantiates the Densenet121 architecture. — application_densenet121","title":"Instantiates the Densenet121 architecture. — application_densenet121","text":"Instantiates Densenet121 architecture.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_densenet121.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Instantiates the Densenet121 architecture. — application_densenet121","text":"","code":"application_densenet121(   include_top = TRUE,   weights = \"imagenet\",   input_tensor = NULL,   input_shape = NULL,   pooling = NULL,   classes = 1000L,   classifier_activation = \"softmax\",   name = \"densenet121\" )"},{"path":"https://keras3.posit.co/dev/reference/application_densenet121.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Instantiates the Densenet121 architecture. — application_densenet121","text":"include_top whether include fully-connected layer top network. weights one NULL (random initialization), \"imagenet\" (pre-training ImageNet), path weights file loaded. input_tensor optional Keras tensor (.e. output keras_input()) use image input model. input_shape optional shape tuple, specified include_top FALSE (otherwise input shape (224, 224, 3) ('channels_last' data format) (3, 224, 224) ('channels_first' data format). exactly 3 inputs channels, width height smaller 32. E.g. (200, 200, 3) one valid value. pooling Optional pooling mode feature extraction include_top FALSE. NULL means output model 4D tensor output last convolutional block. avg means global average pooling applied output last convolutional block, thus output model 2D tensor. max means global max pooling applied. classes optional number classes classify images , specified include_top TRUE, weights argument specified. Defaults 1000. classifier_activation str callable. activation function use \"top\" layer. Ignored unless include_top=TRUE. Set classifier_activation=NULL return logits \"top\" layer. loading pretrained weights, classifier_activation can NULL \"softmax\". name name model (string).","code":""},{"path":"https://keras3.posit.co/dev/reference/application_densenet121.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Instantiates the Densenet121 architecture. — application_densenet121","text":"Keras model instance.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_densenet121.html","id":"reference","dir":"Reference","previous_headings":"","what":"Reference","title":"Instantiates the Densenet121 architecture. — application_densenet121","text":"Densely Connected Convolutional Networks (CVPR 2017) Optionally loads weights pre-trained ImageNet. Note data format convention used model one specified Keras config ~/.keras/keras.json.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_densenet121.html","id":"note","dir":"Reference","previous_headings":"","what":"Note","title":"Instantiates the Densenet121 architecture. — application_densenet121","text":"Keras Application expects specific kind input preprocessing. DenseNet, call application_preprocess_inputs() inputs passing model.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/application_densenet169.html","id":null,"dir":"Reference","previous_headings":"","what":"Instantiates the Densenet169 architecture. — application_densenet169","title":"Instantiates the Densenet169 architecture. — application_densenet169","text":"Instantiates Densenet169 architecture.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_densenet169.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Instantiates the Densenet169 architecture. — application_densenet169","text":"","code":"application_densenet169(   include_top = TRUE,   weights = \"imagenet\",   input_tensor = NULL,   input_shape = NULL,   pooling = NULL,   classes = 1000L,   classifier_activation = \"softmax\",   name = \"densenet169\" )"},{"path":"https://keras3.posit.co/dev/reference/application_densenet169.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Instantiates the Densenet169 architecture. — application_densenet169","text":"include_top whether include fully-connected layer top network. weights one NULL (random initialization), \"imagenet\" (pre-training ImageNet), path weights file loaded. input_tensor optional Keras tensor (.e. output keras_input()) use image input model. input_shape optional shape tuple, specified include_top FALSE (otherwise input shape (224, 224, 3) ('channels_last' data format) (3, 224, 224) ('channels_first' data format). exactly 3 inputs channels, width height smaller 32. E.g. (200, 200, 3) one valid value. pooling Optional pooling mode feature extraction include_top FALSE. NULL means output model 4D tensor output last convolutional block. avg means global average pooling applied output last convolutional block, thus output model 2D tensor. max means global max pooling applied. classes optional number classes classify images , specified include_top TRUE, weights argument specified. Defaults 1000. classifier_activation str callable. activation function use \"top\" layer. Ignored unless include_top=TRUE. Set classifier_activation=NULL return logits \"top\" layer. loading pretrained weights, classifier_activation can NULL \"softmax\". name name model (string).","code":""},{"path":"https://keras3.posit.co/dev/reference/application_densenet169.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Instantiates the Densenet169 architecture. — application_densenet169","text":"Keras model instance.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_densenet169.html","id":"reference","dir":"Reference","previous_headings":"","what":"Reference","title":"Instantiates the Densenet169 architecture. — application_densenet169","text":"Densely Connected Convolutional Networks (CVPR 2017) Optionally loads weights pre-trained ImageNet. Note data format convention used model one specified Keras config ~/.keras/keras.json.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_densenet169.html","id":"note","dir":"Reference","previous_headings":"","what":"Note","title":"Instantiates the Densenet169 architecture. — application_densenet169","text":"Keras Application expects specific kind input preprocessing. DenseNet, call application_preprocess_inputs() inputs passing model.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/application_densenet201.html","id":null,"dir":"Reference","previous_headings":"","what":"Instantiates the Densenet201 architecture. — application_densenet201","title":"Instantiates the Densenet201 architecture. — application_densenet201","text":"Instantiates Densenet201 architecture.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_densenet201.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Instantiates the Densenet201 architecture. — application_densenet201","text":"","code":"application_densenet201(   include_top = TRUE,   weights = \"imagenet\",   input_tensor = NULL,   input_shape = NULL,   pooling = NULL,   classes = 1000L,   classifier_activation = \"softmax\",   name = \"densenet201\" )"},{"path":"https://keras3.posit.co/dev/reference/application_densenet201.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Instantiates the Densenet201 architecture. — application_densenet201","text":"include_top whether include fully-connected layer top network. weights one NULL (random initialization), \"imagenet\" (pre-training ImageNet), path weights file loaded. input_tensor optional Keras tensor (.e. output keras_input()) use image input model. input_shape optional shape tuple, specified include_top FALSE (otherwise input shape (224, 224, 3) ('channels_last' data format) (3, 224, 224) ('channels_first' data format). exactly 3 inputs channels, width height smaller 32. E.g. (200, 200, 3) one valid value. pooling Optional pooling mode feature extraction include_top FALSE. NULL means output model 4D tensor output last convolutional block. avg means global average pooling applied output last convolutional block, thus output model 2D tensor. max means global max pooling applied. classes optional number classes classify images , specified include_top TRUE, weights argument specified. Defaults 1000. classifier_activation str callable. activation function use \"top\" layer. Ignored unless include_top=TRUE. Set classifier_activation=NULL return logits \"top\" layer. loading pretrained weights, classifier_activation can NULL \"softmax\". name name model (string).","code":""},{"path":"https://keras3.posit.co/dev/reference/application_densenet201.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Instantiates the Densenet201 architecture. — application_densenet201","text":"Keras model instance.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_densenet201.html","id":"reference","dir":"Reference","previous_headings":"","what":"Reference","title":"Instantiates the Densenet201 architecture. — application_densenet201","text":"Densely Connected Convolutional Networks (CVPR 2017) Optionally loads weights pre-trained ImageNet. Note data format convention used model one specified Keras config ~/.keras/keras.json.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_densenet201.html","id":"note","dir":"Reference","previous_headings":"","what":"Note","title":"Instantiates the Densenet201 architecture. — application_densenet201","text":"Keras Application expects specific kind input preprocessing. DenseNet, call application_preprocess_inputs() inputs passing model.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_b0.html","id":null,"dir":"Reference","previous_headings":"","what":"Instantiates the EfficientNetB0 architecture. — application_efficientnet_b0","title":"Instantiates the EfficientNetB0 architecture. — application_efficientnet_b0","text":"Instantiates EfficientNetB0 architecture.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_b0.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Instantiates the EfficientNetB0 architecture. — application_efficientnet_b0","text":"","code":"application_efficientnet_b0(   include_top = TRUE,   weights = \"imagenet\",   input_tensor = NULL,   input_shape = NULL,   pooling = NULL,   classes = 1000L,   classifier_activation = \"softmax\",   name = \"efficientnetb0\",   ... )"},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_b0.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Instantiates the EfficientNetB0 architecture. — application_efficientnet_b0","text":"include_top Whether include fully-connected layer top network. Defaults TRUE. weights One NULL (random initialization), \"imagenet\" (pre-training ImageNet), path weights file loaded. Defaults \"imagenet\". input_tensor Optional Keras tensor (.e. output keras_input()) use image input model. input_shape Optional shape tuple, specified include_top FALSE. exactly 3 inputs channels. pooling Optional pooling mode feature extraction include_top FALSE. Defaults NULL. NULL means output model 4D tensor output last convolutional layer. avg means global average pooling applied output last convolutional layer, thus output model 2D tensor. max means global max pooling applied. classes Optional number classes classify images , specified include_top TRUE, weights argument specified. 1000 many ImageNet classes . Defaults 1000. classifier_activation str callable. activation function use \"top\" layer. Ignored unless include_top=TRUE. Set classifier_activation=NULL return logits \"top\" layer. Defaults 'softmax'. loading pretrained weights, classifier_activation can NULL \"softmax\". name name model (string). ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_b0.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Instantiates the EfficientNetB0 architecture. — application_efficientnet_b0","text":"model instance.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_b0.html","id":"reference","dir":"Reference","previous_headings":"","what":"Reference","title":"Instantiates the EfficientNetB0 architecture. — application_efficientnet_b0","text":"EfficientNet: Rethinking Model Scaling Convolutional Neural Networks (ICML 2019) function returns Keras image classification model, optionally loaded weights pre-trained ImageNet. image classification use cases, see page detailed examples. transfer learning use cases, make sure read guide transfer learning & fine-tuning.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_b0.html","id":"note","dir":"Reference","previous_headings":"","what":"Note","title":"Instantiates the EfficientNetB0 architecture. — application_efficientnet_b0","text":"Keras Application expects specific kind input preprocessing. EfficientNet, input preprocessing included part model (Rescaling layer), thus application_preprocess_inputs() actually pass-function. EfficientNet models expect inputs float tensors pixels values [0-255] range.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_b1.html","id":null,"dir":"Reference","previous_headings":"","what":"Instantiates the EfficientNetB1 architecture. — application_efficientnet_b1","title":"Instantiates the EfficientNetB1 architecture. — application_efficientnet_b1","text":"Instantiates EfficientNetB1 architecture.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_b1.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Instantiates the EfficientNetB1 architecture. — application_efficientnet_b1","text":"","code":"application_efficientnet_b1(   include_top = TRUE,   weights = \"imagenet\",   input_tensor = NULL,   input_shape = NULL,   pooling = NULL,   classes = 1000L,   classifier_activation = \"softmax\",   name = \"efficientnetb1\",   ... )"},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_b1.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Instantiates the EfficientNetB1 architecture. — application_efficientnet_b1","text":"include_top Whether include fully-connected layer top network. Defaults TRUE. weights One NULL (random initialization), \"imagenet\" (pre-training ImageNet), path weights file loaded. Defaults \"imagenet\". input_tensor Optional Keras tensor (.e. output keras_input()) use image input model. input_shape Optional shape tuple, specified include_top FALSE. exactly 3 inputs channels. pooling Optional pooling mode feature extraction include_top FALSE. Defaults NULL. NULL means output model 4D tensor output last convolutional layer. avg means global average pooling applied output last convolutional layer, thus output model 2D tensor. max means global max pooling applied. classes Optional number classes classify images , specified include_top TRUE, weights argument specified. 1000 many ImageNet classes . Defaults 1000. classifier_activation str callable. activation function use \"top\" layer. Ignored unless include_top=TRUE. Set classifier_activation=NULL return logits \"top\" layer. Defaults 'softmax'. loading pretrained weights, classifier_activation can NULL \"softmax\". name name model (string). ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_b1.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Instantiates the EfficientNetB1 architecture. — application_efficientnet_b1","text":"model instance.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_b1.html","id":"reference","dir":"Reference","previous_headings":"","what":"Reference","title":"Instantiates the EfficientNetB1 architecture. — application_efficientnet_b1","text":"EfficientNet: Rethinking Model Scaling Convolutional Neural Networks (ICML 2019) function returns Keras image classification model, optionally loaded weights pre-trained ImageNet. image classification use cases, see page detailed examples. transfer learning use cases, make sure read guide transfer learning & fine-tuning.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_b1.html","id":"note","dir":"Reference","previous_headings":"","what":"Note","title":"Instantiates the EfficientNetB1 architecture. — application_efficientnet_b1","text":"Keras Application expects specific kind input preprocessing. EfficientNet, input preprocessing included part model (Rescaling layer), thus application_preprocess_inputs() actually pass-function. EfficientNet models expect inputs float tensors pixels values [0-255] range.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_b2.html","id":null,"dir":"Reference","previous_headings":"","what":"Instantiates the EfficientNetB2 architecture. — application_efficientnet_b2","title":"Instantiates the EfficientNetB2 architecture. — application_efficientnet_b2","text":"Instantiates EfficientNetB2 architecture.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_b2.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Instantiates the EfficientNetB2 architecture. — application_efficientnet_b2","text":"","code":"application_efficientnet_b2(   include_top = TRUE,   weights = \"imagenet\",   input_tensor = NULL,   input_shape = NULL,   pooling = NULL,   classes = 1000L,   classifier_activation = \"softmax\",   name = \"efficientnetb2\",   ... )"},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_b2.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Instantiates the EfficientNetB2 architecture. — application_efficientnet_b2","text":"include_top Whether include fully-connected layer top network. Defaults TRUE. weights One NULL (random initialization), \"imagenet\" (pre-training ImageNet), path weights file loaded. Defaults \"imagenet\". input_tensor Optional Keras tensor (.e. output keras_input()) use image input model. input_shape Optional shape tuple, specified include_top FALSE. exactly 3 inputs channels. pooling Optional pooling mode feature extraction include_top FALSE. Defaults NULL. NULL means output model 4D tensor output last convolutional layer. avg means global average pooling applied output last convolutional layer, thus output model 2D tensor. max means global max pooling applied. classes Optional number classes classify images , specified include_top TRUE, weights argument specified. 1000 many ImageNet classes . Defaults 1000. classifier_activation str callable. activation function use \"top\" layer. Ignored unless include_top=TRUE. Set classifier_activation=NULL return logits \"top\" layer. Defaults 'softmax'. loading pretrained weights, classifier_activation can NULL \"softmax\". name name model (string). ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_b2.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Instantiates the EfficientNetB2 architecture. — application_efficientnet_b2","text":"model instance.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_b2.html","id":"reference","dir":"Reference","previous_headings":"","what":"Reference","title":"Instantiates the EfficientNetB2 architecture. — application_efficientnet_b2","text":"EfficientNet: Rethinking Model Scaling Convolutional Neural Networks (ICML 2019) function returns Keras image classification model, optionally loaded weights pre-trained ImageNet. image classification use cases, see page detailed examples. transfer learning use cases, make sure read guide transfer learning & fine-tuning.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_b2.html","id":"note","dir":"Reference","previous_headings":"","what":"Note","title":"Instantiates the EfficientNetB2 architecture. — application_efficientnet_b2","text":"Keras Application expects specific kind input preprocessing. EfficientNet, input preprocessing included part model (Rescaling layer), thus application_preprocess_inputs() actually pass-function. EfficientNet models expect inputs float tensors pixels values [0-255] range.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_b3.html","id":null,"dir":"Reference","previous_headings":"","what":"Instantiates the EfficientNetB3 architecture. — application_efficientnet_b3","title":"Instantiates the EfficientNetB3 architecture. — application_efficientnet_b3","text":"Instantiates EfficientNetB3 architecture.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_b3.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Instantiates the EfficientNetB3 architecture. — application_efficientnet_b3","text":"","code":"application_efficientnet_b3(   include_top = TRUE,   weights = \"imagenet\",   input_tensor = NULL,   input_shape = NULL,   pooling = NULL,   classes = 1000L,   classifier_activation = \"softmax\",   name = \"efficientnetb3\",   ... )"},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_b3.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Instantiates the EfficientNetB3 architecture. — application_efficientnet_b3","text":"include_top Whether include fully-connected layer top network. Defaults TRUE. weights One NULL (random initialization), \"imagenet\" (pre-training ImageNet), path weights file loaded. Defaults \"imagenet\". input_tensor Optional Keras tensor (.e. output keras_input()) use image input model. input_shape Optional shape tuple, specified include_top FALSE. exactly 3 inputs channels. pooling Optional pooling mode feature extraction include_top FALSE. Defaults NULL. NULL means output model 4D tensor output last convolutional layer. avg means global average pooling applied output last convolutional layer, thus output model 2D tensor. max means global max pooling applied. classes Optional number classes classify images , specified include_top TRUE, weights argument specified. 1000 many ImageNet classes . Defaults 1000. classifier_activation str callable. activation function use \"top\" layer. Ignored unless include_top=TRUE. Set classifier_activation=NULL return logits \"top\" layer. Defaults 'softmax'. loading pretrained weights, classifier_activation can NULL \"softmax\". name name model (string). ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_b3.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Instantiates the EfficientNetB3 architecture. — application_efficientnet_b3","text":"model instance.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_b3.html","id":"reference","dir":"Reference","previous_headings":"","what":"Reference","title":"Instantiates the EfficientNetB3 architecture. — application_efficientnet_b3","text":"EfficientNet: Rethinking Model Scaling Convolutional Neural Networks (ICML 2019) function returns Keras image classification model, optionally loaded weights pre-trained ImageNet. image classification use cases, see page detailed examples. transfer learning use cases, make sure read guide transfer learning & fine-tuning.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_b3.html","id":"note","dir":"Reference","previous_headings":"","what":"Note","title":"Instantiates the EfficientNetB3 architecture. — application_efficientnet_b3","text":"Keras Application expects specific kind input preprocessing. EfficientNet, input preprocessing included part model (Rescaling layer), thus application_preprocess_inputs() actually pass-function. EfficientNet models expect inputs float tensors pixels values [0-255] range.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_b4.html","id":null,"dir":"Reference","previous_headings":"","what":"Instantiates the EfficientNetB4 architecture. — application_efficientnet_b4","title":"Instantiates the EfficientNetB4 architecture. — application_efficientnet_b4","text":"Instantiates EfficientNetB4 architecture.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_b4.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Instantiates the EfficientNetB4 architecture. — application_efficientnet_b4","text":"","code":"application_efficientnet_b4(   include_top = TRUE,   weights = \"imagenet\",   input_tensor = NULL,   input_shape = NULL,   pooling = NULL,   classes = 1000L,   classifier_activation = \"softmax\",   name = \"efficientnetb4\",   ... )"},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_b4.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Instantiates the EfficientNetB4 architecture. — application_efficientnet_b4","text":"include_top Whether include fully-connected layer top network. Defaults TRUE. weights One NULL (random initialization), \"imagenet\" (pre-training ImageNet), path weights file loaded. Defaults \"imagenet\". input_tensor Optional Keras tensor (.e. output keras_input()) use image input model. input_shape Optional shape tuple, specified include_top FALSE. exactly 3 inputs channels. pooling Optional pooling mode feature extraction include_top FALSE. Defaults NULL. NULL means output model 4D tensor output last convolutional layer. avg means global average pooling applied output last convolutional layer, thus output model 2D tensor. max means global max pooling applied. classes Optional number classes classify images , specified include_top TRUE, weights argument specified. 1000 many ImageNet classes . Defaults 1000. classifier_activation str callable. activation function use \"top\" layer. Ignored unless include_top=TRUE. Set classifier_activation=NULL return logits \"top\" layer. Defaults 'softmax'. loading pretrained weights, classifier_activation can NULL \"softmax\". name name model (string). ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_b4.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Instantiates the EfficientNetB4 architecture. — application_efficientnet_b4","text":"model instance.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_b4.html","id":"reference","dir":"Reference","previous_headings":"","what":"Reference","title":"Instantiates the EfficientNetB4 architecture. — application_efficientnet_b4","text":"EfficientNet: Rethinking Model Scaling Convolutional Neural Networks (ICML 2019) function returns Keras image classification model, optionally loaded weights pre-trained ImageNet. image classification use cases, see page detailed examples. transfer learning use cases, make sure read guide transfer learning & fine-tuning.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_b4.html","id":"note","dir":"Reference","previous_headings":"","what":"Note","title":"Instantiates the EfficientNetB4 architecture. — application_efficientnet_b4","text":"Keras Application expects specific kind input preprocessing. EfficientNet, input preprocessing included part model (Rescaling layer), thus application_preprocess_inputs() actually pass-function. EfficientNet models expect inputs float tensors pixels values [0-255] range.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_b5.html","id":null,"dir":"Reference","previous_headings":"","what":"Instantiates the EfficientNetB5 architecture. — application_efficientnet_b5","title":"Instantiates the EfficientNetB5 architecture. — application_efficientnet_b5","text":"Instantiates EfficientNetB5 architecture.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_b5.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Instantiates the EfficientNetB5 architecture. — application_efficientnet_b5","text":"","code":"application_efficientnet_b5(   include_top = TRUE,   weights = \"imagenet\",   input_tensor = NULL,   input_shape = NULL,   pooling = NULL,   classes = 1000L,   classifier_activation = \"softmax\",   name = \"efficientnetb5\",   ... )"},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_b5.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Instantiates the EfficientNetB5 architecture. — application_efficientnet_b5","text":"include_top Whether include fully-connected layer top network. Defaults TRUE. weights One NULL (random initialization), \"imagenet\" (pre-training ImageNet), path weights file loaded. Defaults \"imagenet\". input_tensor Optional Keras tensor (.e. output keras_input()) use image input model. input_shape Optional shape tuple, specified include_top FALSE. exactly 3 inputs channels. pooling Optional pooling mode feature extraction include_top FALSE. Defaults NULL. NULL means output model 4D tensor output last convolutional layer. avg means global average pooling applied output last convolutional layer, thus output model 2D tensor. max means global max pooling applied. classes Optional number classes classify images , specified include_top TRUE, weights argument specified. 1000 many ImageNet classes . Defaults 1000. classifier_activation str callable. activation function use \"top\" layer. Ignored unless include_top=TRUE. Set classifier_activation=NULL return logits \"top\" layer. Defaults 'softmax'. loading pretrained weights, classifier_activation can NULL \"softmax\". name name model (string). ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_b5.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Instantiates the EfficientNetB5 architecture. — application_efficientnet_b5","text":"model instance.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_b5.html","id":"reference","dir":"Reference","previous_headings":"","what":"Reference","title":"Instantiates the EfficientNetB5 architecture. — application_efficientnet_b5","text":"EfficientNet: Rethinking Model Scaling Convolutional Neural Networks (ICML 2019) function returns Keras image classification model, optionally loaded weights pre-trained ImageNet. image classification use cases, see page detailed examples. transfer learning use cases, make sure read guide transfer learning & fine-tuning.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_b5.html","id":"note","dir":"Reference","previous_headings":"","what":"Note","title":"Instantiates the EfficientNetB5 architecture. — application_efficientnet_b5","text":"Keras Application expects specific kind input preprocessing. EfficientNet, input preprocessing included part model (Rescaling layer), thus application_preprocess_inputs() actually pass-function. EfficientNet models expect inputs float tensors pixels values [0-255] range.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_b6.html","id":null,"dir":"Reference","previous_headings":"","what":"Instantiates the EfficientNetB6 architecture. — application_efficientnet_b6","title":"Instantiates the EfficientNetB6 architecture. — application_efficientnet_b6","text":"Instantiates EfficientNetB6 architecture.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_b6.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Instantiates the EfficientNetB6 architecture. — application_efficientnet_b6","text":"","code":"application_efficientnet_b6(   include_top = TRUE,   weights = \"imagenet\",   input_tensor = NULL,   input_shape = NULL,   pooling = NULL,   classes = 1000L,   classifier_activation = \"softmax\",   name = \"efficientnetb6\",   ... )"},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_b6.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Instantiates the EfficientNetB6 architecture. — application_efficientnet_b6","text":"include_top Whether include fully-connected layer top network. Defaults TRUE. weights One NULL (random initialization), \"imagenet\" (pre-training ImageNet), path weights file loaded. Defaults \"imagenet\". input_tensor Optional Keras tensor (.e. output keras_input()) use image input model. input_shape Optional shape tuple, specified include_top FALSE. exactly 3 inputs channels. pooling Optional pooling mode feature extraction include_top FALSE. Defaults NULL. NULL means output model 4D tensor output last convolutional layer. avg means global average pooling applied output last convolutional layer, thus output model 2D tensor. max means global max pooling applied. classes Optional number classes classify images , specified include_top TRUE, weights argument specified. 1000 many ImageNet classes . Defaults 1000. classifier_activation str callable. activation function use \"top\" layer. Ignored unless include_top=TRUE. Set classifier_activation=NULL return logits \"top\" layer. Defaults 'softmax'. loading pretrained weights, classifier_activation can NULL \"softmax\". name name model (string). ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_b6.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Instantiates the EfficientNetB6 architecture. — application_efficientnet_b6","text":"model instance.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_b6.html","id":"reference","dir":"Reference","previous_headings":"","what":"Reference","title":"Instantiates the EfficientNetB6 architecture. — application_efficientnet_b6","text":"EfficientNet: Rethinking Model Scaling Convolutional Neural Networks (ICML 2019) function returns Keras image classification model, optionally loaded weights pre-trained ImageNet. image classification use cases, see page detailed examples. transfer learning use cases, make sure read guide transfer learning & fine-tuning.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_b6.html","id":"note","dir":"Reference","previous_headings":"","what":"Note","title":"Instantiates the EfficientNetB6 architecture. — application_efficientnet_b6","text":"Keras Application expects specific kind input preprocessing. EfficientNet, input preprocessing included part model (Rescaling layer), thus application_preprocess_inputs() actually pass-function. EfficientNet models expect inputs float tensors pixels values [0-255] range.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_b7.html","id":null,"dir":"Reference","previous_headings":"","what":"Instantiates the EfficientNetB7 architecture. — application_efficientnet_b7","title":"Instantiates the EfficientNetB7 architecture. — application_efficientnet_b7","text":"Instantiates EfficientNetB7 architecture.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_b7.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Instantiates the EfficientNetB7 architecture. — application_efficientnet_b7","text":"","code":"application_efficientnet_b7(   include_top = TRUE,   weights = \"imagenet\",   input_tensor = NULL,   input_shape = NULL,   pooling = NULL,   classes = 1000L,   classifier_activation = \"softmax\",   name = \"efficientnetb7\",   ... )"},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_b7.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Instantiates the EfficientNetB7 architecture. — application_efficientnet_b7","text":"include_top Whether include fully-connected layer top network. Defaults TRUE. weights One NULL (random initialization), \"imagenet\" (pre-training ImageNet), path weights file loaded. Defaults \"imagenet\". input_tensor Optional Keras tensor (.e. output keras_input()) use image input model. input_shape Optional shape tuple, specified include_top FALSE. exactly 3 inputs channels. pooling Optional pooling mode feature extraction include_top FALSE. Defaults NULL. NULL means output model 4D tensor output last convolutional layer. avg means global average pooling applied output last convolutional layer, thus output model 2D tensor. max means global max pooling applied. classes Optional number classes classify images , specified include_top TRUE, weights argument specified. 1000 many ImageNet classes . Defaults 1000. classifier_activation str callable. activation function use \"top\" layer. Ignored unless include_top=TRUE. Set classifier_activation=NULL return logits \"top\" layer. Defaults 'softmax'. loading pretrained weights, classifier_activation can NULL \"softmax\". name name model (string). ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_b7.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Instantiates the EfficientNetB7 architecture. — application_efficientnet_b7","text":"model instance.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_b7.html","id":"reference","dir":"Reference","previous_headings":"","what":"Reference","title":"Instantiates the EfficientNetB7 architecture. — application_efficientnet_b7","text":"EfficientNet: Rethinking Model Scaling Convolutional Neural Networks (ICML 2019) function returns Keras image classification model, optionally loaded weights pre-trained ImageNet. image classification use cases, see page detailed examples. transfer learning use cases, make sure read guide transfer learning & fine-tuning.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_b7.html","id":"note","dir":"Reference","previous_headings":"","what":"Note","title":"Instantiates the EfficientNetB7 architecture. — application_efficientnet_b7","text":"Keras Application expects specific kind input preprocessing. EfficientNet, input preprocessing included part model (Rescaling layer), thus application_preprocess_inputs() actually pass-function. EfficientNet models expect inputs float tensors pixels values [0-255] range.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_v2b0.html","id":null,"dir":"Reference","previous_headings":"","what":"Instantiates the EfficientNetV2B0 architecture. — application_efficientnet_v2b0","title":"Instantiates the EfficientNetV2B0 architecture. — application_efficientnet_v2b0","text":"Instantiates EfficientNetV2B0 architecture.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_v2b0.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Instantiates the EfficientNetV2B0 architecture. — application_efficientnet_v2b0","text":"","code":"application_efficientnet_v2b0(   include_top = TRUE,   weights = \"imagenet\",   input_tensor = NULL,   input_shape = NULL,   pooling = NULL,   classes = 1000L,   classifier_activation = \"softmax\",   include_preprocessing = TRUE,   name = \"efficientnetv2-b0\" )"},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_v2b0.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Instantiates the EfficientNetV2B0 architecture. — application_efficientnet_v2b0","text":"include_top Boolean, whether include fully-connected layer top network. Defaults TRUE. weights One NULL (random initialization), \"imagenet\" (pre-training ImageNet), path weights file loaded. Defaults \"imagenet\". input_tensor Optional Keras tensor (.e. output keras_input()) use image input model. input_shape Optional shape tuple, specified include_top FALSE. exactly 3 inputs channels. pooling Optional pooling mode feature extraction include_top FALSE. Defaults NULL. NULL means output model 4D tensor output last convolutional layer. \"avg\" means global average pooling applied output last convolutional layer, thus output model 2D tensor. \"max\" means global max pooling applied. classes Optional number classes classify images , specified include_top TRUE, weights argument specified. Defaults 1000 (number ImageNet classes). classifier_activation string callable. activation function use \"top\" layer. Ignored unless include_top=TRUE. Set classifier_activation=NULL return logits \"top\" layer. Defaults \"softmax\". loading pretrained weights, classifier_activation can NULL \"softmax\". include_preprocessing Boolean, whether include preprocessing layer bottom network. name name model (string).","code":""},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_v2b0.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Instantiates the EfficientNetV2B0 architecture. — application_efficientnet_v2b0","text":"model instance.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_v2b0.html","id":"reference","dir":"Reference","previous_headings":"","what":"Reference","title":"Instantiates the EfficientNetV2B0 architecture. — application_efficientnet_v2b0","text":"EfficientNetV2: Smaller Models Faster Training (ICML 2021) function returns Keras image classification model, optionally loaded weights pre-trained ImageNet. image classification use cases, see page detailed examples. transfer learning use cases, make sure read guide transfer learning & fine-tuning.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_v2b0.html","id":"note","dir":"Reference","previous_headings":"","what":"Note","title":"Instantiates the EfficientNetV2B0 architecture. — application_efficientnet_v2b0","text":"Keras Application expects specific kind input preprocessing. EfficientNetV2, default input preprocessing included part model (Rescaling layer), thus application_preprocess_inputs() actually pass-function. use case, EfficientNetV2 models expect inputs float tensors pixels values [0, 255] range. time, preprocessing part model (.e. Rescaling layer) can disabled setting include_preprocessing argument FALSE. preprocessing disabled EfficientNetV2 models expect inputs float tensors pixels values [-1, 1] range.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_v2b1.html","id":null,"dir":"Reference","previous_headings":"","what":"Instantiates the EfficientNetV2B1 architecture. — application_efficientnet_v2b1","title":"Instantiates the EfficientNetV2B1 architecture. — application_efficientnet_v2b1","text":"Instantiates EfficientNetV2B1 architecture.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_v2b1.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Instantiates the EfficientNetV2B1 architecture. — application_efficientnet_v2b1","text":"","code":"application_efficientnet_v2b1(   include_top = TRUE,   weights = \"imagenet\",   input_tensor = NULL,   input_shape = NULL,   pooling = NULL,   classes = 1000L,   classifier_activation = \"softmax\",   include_preprocessing = TRUE,   name = \"efficientnetv2-b1\" )"},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_v2b1.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Instantiates the EfficientNetV2B1 architecture. — application_efficientnet_v2b1","text":"include_top Boolean, whether include fully-connected layer top network. Defaults TRUE. weights One NULL (random initialization), \"imagenet\" (pre-training ImageNet), path weights file loaded. Defaults \"imagenet\". input_tensor Optional Keras tensor (.e. output keras_input()) use image input model. input_shape Optional shape tuple, specified include_top FALSE. exactly 3 inputs channels. pooling Optional pooling mode feature extraction include_top FALSE. Defaults NULL. NULL means output model 4D tensor output last convolutional layer. \"avg\" means global average pooling applied output last convolutional layer, thus output model 2D tensor. \"max\" means global max pooling applied. classes Optional number classes classify images , specified include_top TRUE, weights argument specified. Defaults 1000 (number ImageNet classes). classifier_activation string callable. activation function use \"top\" layer. Ignored unless include_top=TRUE. Set classifier_activation=NULL return logits \"top\" layer. Defaults \"softmax\". loading pretrained weights, classifier_activation can NULL \"softmax\". include_preprocessing Boolean, whether include preprocessing layer bottom network. name name model (string).","code":""},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_v2b1.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Instantiates the EfficientNetV2B1 architecture. — application_efficientnet_v2b1","text":"model instance.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_v2b1.html","id":"reference","dir":"Reference","previous_headings":"","what":"Reference","title":"Instantiates the EfficientNetV2B1 architecture. — application_efficientnet_v2b1","text":"EfficientNetV2: Smaller Models Faster Training (ICML 2021) function returns Keras image classification model, optionally loaded weights pre-trained ImageNet. image classification use cases, see page detailed examples. transfer learning use cases, make sure read guide transfer learning & fine-tuning.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_v2b1.html","id":"note","dir":"Reference","previous_headings":"","what":"Note","title":"Instantiates the EfficientNetV2B1 architecture. — application_efficientnet_v2b1","text":"Keras Application expects specific kind input preprocessing. EfficientNetV2, default input preprocessing included part model (Rescaling layer), thus application_preprocess_inputs() actually pass-function. use case, EfficientNetV2 models expect inputs float tensors pixels values [0, 255] range. time, preprocessing part model (.e. Rescaling layer) can disabled setting include_preprocessing argument FALSE. preprocessing disabled EfficientNetV2 models expect inputs float tensors pixels values [-1, 1] range.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_v2b2.html","id":null,"dir":"Reference","previous_headings":"","what":"Instantiates the EfficientNetV2B2 architecture. — application_efficientnet_v2b2","title":"Instantiates the EfficientNetV2B2 architecture. — application_efficientnet_v2b2","text":"Instantiates EfficientNetV2B2 architecture.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_v2b2.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Instantiates the EfficientNetV2B2 architecture. — application_efficientnet_v2b2","text":"","code":"application_efficientnet_v2b2(   include_top = TRUE,   weights = \"imagenet\",   input_tensor = NULL,   input_shape = NULL,   pooling = NULL,   classes = 1000L,   classifier_activation = \"softmax\",   include_preprocessing = TRUE,   name = \"efficientnetv2-b2\" )"},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_v2b2.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Instantiates the EfficientNetV2B2 architecture. — application_efficientnet_v2b2","text":"include_top Boolean, whether include fully-connected layer top network. Defaults TRUE. weights One NULL (random initialization), \"imagenet\" (pre-training ImageNet), path weights file loaded. Defaults \"imagenet\". input_tensor Optional Keras tensor (.e. output keras_input()) use image input model. input_shape Optional shape tuple, specified include_top FALSE. exactly 3 inputs channels. pooling Optional pooling mode feature extraction include_top FALSE. Defaults NULL. NULL means output model 4D tensor output last convolutional layer. \"avg\" means global average pooling applied output last convolutional layer, thus output model 2D tensor. \"max\" means global max pooling applied. classes Optional number classes classify images , specified include_top TRUE, weights argument specified. Defaults 1000 (number ImageNet classes). classifier_activation string callable. activation function use \"top\" layer. Ignored unless include_top=TRUE. Set classifier_activation=NULL return logits \"top\" layer. Defaults \"softmax\". loading pretrained weights, classifier_activation can NULL \"softmax\". include_preprocessing Boolean, whether include preprocessing layer bottom network. name name model (string).","code":""},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_v2b2.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Instantiates the EfficientNetV2B2 architecture. — application_efficientnet_v2b2","text":"model instance.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_v2b2.html","id":"reference","dir":"Reference","previous_headings":"","what":"Reference","title":"Instantiates the EfficientNetV2B2 architecture. — application_efficientnet_v2b2","text":"EfficientNetV2: Smaller Models Faster Training (ICML 2021) function returns Keras image classification model, optionally loaded weights pre-trained ImageNet. image classification use cases, see page detailed examples. transfer learning use cases, make sure read guide transfer learning & fine-tuning.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_v2b2.html","id":"note","dir":"Reference","previous_headings":"","what":"Note","title":"Instantiates the EfficientNetV2B2 architecture. — application_efficientnet_v2b2","text":"Keras Application expects specific kind input preprocessing. EfficientNetV2, default input preprocessing included part model (Rescaling layer), thus application_preprocess_inputs() actually pass-function. use case, EfficientNetV2 models expect inputs float tensors pixels values [0, 255] range. time, preprocessing part model (.e. Rescaling layer) can disabled setting include_preprocessing argument FALSE. preprocessing disabled EfficientNetV2 models expect inputs float tensors pixels values [-1, 1] range.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_v2b3.html","id":null,"dir":"Reference","previous_headings":"","what":"Instantiates the EfficientNetV2B3 architecture. — application_efficientnet_v2b3","title":"Instantiates the EfficientNetV2B3 architecture. — application_efficientnet_v2b3","text":"Instantiates EfficientNetV2B3 architecture.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_v2b3.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Instantiates the EfficientNetV2B3 architecture. — application_efficientnet_v2b3","text":"","code":"application_efficientnet_v2b3(   include_top = TRUE,   weights = \"imagenet\",   input_tensor = NULL,   input_shape = NULL,   pooling = NULL,   classes = 1000L,   classifier_activation = \"softmax\",   include_preprocessing = TRUE,   name = \"efficientnetv2-b3\" )"},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_v2b3.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Instantiates the EfficientNetV2B3 architecture. — application_efficientnet_v2b3","text":"include_top Boolean, whether include fully-connected layer top network. Defaults TRUE. weights One NULL (random initialization), \"imagenet\" (pre-training ImageNet), path weights file loaded. Defaults \"imagenet\". input_tensor Optional Keras tensor (.e. output keras_input()) use image input model. input_shape Optional shape tuple, specified include_top FALSE. exactly 3 inputs channels. pooling Optional pooling mode feature extraction include_top FALSE. Defaults NULL. NULL means output model 4D tensor output last convolutional layer. \"avg\" means global average pooling applied output last convolutional layer, thus output model 2D tensor. \"max\" means global max pooling applied. classes Optional number classes classify images , specified include_top TRUE, weights argument specified. Defaults 1000 (number ImageNet classes). classifier_activation string callable. activation function use \"top\" layer. Ignored unless include_top=TRUE. Set classifier_activation=NULL return logits \"top\" layer. Defaults \"softmax\". loading pretrained weights, classifier_activation can NULL \"softmax\". include_preprocessing Boolean, whether include preprocessing layer bottom network. name name model (string).","code":""},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_v2b3.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Instantiates the EfficientNetV2B3 architecture. — application_efficientnet_v2b3","text":"model instance.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_v2b3.html","id":"reference","dir":"Reference","previous_headings":"","what":"Reference","title":"Instantiates the EfficientNetV2B3 architecture. — application_efficientnet_v2b3","text":"EfficientNetV2: Smaller Models Faster Training (ICML 2021) function returns Keras image classification model, optionally loaded weights pre-trained ImageNet. image classification use cases, see page detailed examples. transfer learning use cases, make sure read guide transfer learning & fine-tuning.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_v2b3.html","id":"note","dir":"Reference","previous_headings":"","what":"Note","title":"Instantiates the EfficientNetV2B3 architecture. — application_efficientnet_v2b3","text":"Keras Application expects specific kind input preprocessing. EfficientNetV2, default input preprocessing included part model (Rescaling layer), thus application_preprocess_inputs() actually pass-function. use case, EfficientNetV2 models expect inputs float tensors pixels values [0, 255] range. time, preprocessing part model (.e. Rescaling layer) can disabled setting include_preprocessing argument FALSE. preprocessing disabled EfficientNetV2 models expect inputs float tensors pixels values [-1, 1] range.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_v2l.html","id":null,"dir":"Reference","previous_headings":"","what":"Instantiates the EfficientNetV2L architecture. — application_efficientnet_v2l","title":"Instantiates the EfficientNetV2L architecture. — application_efficientnet_v2l","text":"Instantiates EfficientNetV2L architecture.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_v2l.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Instantiates the EfficientNetV2L architecture. — application_efficientnet_v2l","text":"","code":"application_efficientnet_v2l(   include_top = TRUE,   weights = \"imagenet\",   input_tensor = NULL,   input_shape = NULL,   pooling = NULL,   classes = 1000L,   classifier_activation = \"softmax\",   include_preprocessing = TRUE,   name = \"efficientnetv2-l\" )"},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_v2l.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Instantiates the EfficientNetV2L architecture. — application_efficientnet_v2l","text":"include_top Boolean, whether include fully-connected layer top network. Defaults TRUE. weights One NULL (random initialization), \"imagenet\" (pre-training ImageNet), path weights file loaded. Defaults \"imagenet\". input_tensor Optional Keras tensor (.e. output keras_input()) use image input model. input_shape Optional shape tuple, specified include_top FALSE. exactly 3 inputs channels. pooling Optional pooling mode feature extraction include_top FALSE. Defaults NULL. NULL means output model 4D tensor output last convolutional layer. \"avg\" means global average pooling applied output last convolutional layer, thus output model 2D tensor. \"max\" means global max pooling applied. classes Optional number classes classify images , specified include_top TRUE, weights argument specified. Defaults 1000 (number ImageNet classes). classifier_activation string callable. activation function use \"top\" layer. Ignored unless include_top=TRUE. Set classifier_activation=NULL return logits \"top\" layer. Defaults \"softmax\". loading pretrained weights, classifier_activation can NULL \"softmax\". include_preprocessing Boolean, whether include preprocessing layer bottom network. name name model (string).","code":""},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_v2l.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Instantiates the EfficientNetV2L architecture. — application_efficientnet_v2l","text":"model instance.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_v2l.html","id":"reference","dir":"Reference","previous_headings":"","what":"Reference","title":"Instantiates the EfficientNetV2L architecture. — application_efficientnet_v2l","text":"EfficientNetV2: Smaller Models Faster Training (ICML 2021) function returns Keras image classification model, optionally loaded weights pre-trained ImageNet. image classification use cases, see page detailed examples. transfer learning use cases, make sure read guide transfer learning & fine-tuning.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_v2l.html","id":"note","dir":"Reference","previous_headings":"","what":"Note","title":"Instantiates the EfficientNetV2L architecture. — application_efficientnet_v2l","text":"Keras Application expects specific kind input preprocessing. EfficientNetV2, default input preprocessing included part model (Rescaling layer), thus application_preprocess_inputs() actually pass-function. use case, EfficientNetV2 models expect inputs float tensors pixels values [0, 255] range. time, preprocessing part model (.e. Rescaling layer) can disabled setting include_preprocessing argument FALSE. preprocessing disabled EfficientNetV2 models expect inputs float tensors pixels values [-1, 1] range.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_v2m.html","id":null,"dir":"Reference","previous_headings":"","what":"Instantiates the EfficientNetV2M architecture. — application_efficientnet_v2m","title":"Instantiates the EfficientNetV2M architecture. — application_efficientnet_v2m","text":"Instantiates EfficientNetV2M architecture.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_v2m.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Instantiates the EfficientNetV2M architecture. — application_efficientnet_v2m","text":"","code":"application_efficientnet_v2m(   include_top = TRUE,   weights = \"imagenet\",   input_tensor = NULL,   input_shape = NULL,   pooling = NULL,   classes = 1000L,   classifier_activation = \"softmax\",   include_preprocessing = TRUE,   name = \"efficientnetv2-m\" )"},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_v2m.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Instantiates the EfficientNetV2M architecture. — application_efficientnet_v2m","text":"include_top Boolean, whether include fully-connected layer top network. Defaults TRUE. weights One NULL (random initialization), \"imagenet\" (pre-training ImageNet), path weights file loaded. Defaults \"imagenet\". input_tensor Optional Keras tensor (.e. output keras_input()) use image input model. input_shape Optional shape tuple, specified include_top FALSE. exactly 3 inputs channels. pooling Optional pooling mode feature extraction include_top FALSE. Defaults NULL. NULL means output model 4D tensor output last convolutional layer. \"avg\" means global average pooling applied output last convolutional layer, thus output model 2D tensor. \"max\" means global max pooling applied. classes Optional number classes classify images , specified include_top TRUE, weights argument specified. Defaults 1000 (number ImageNet classes). classifier_activation string callable. activation function use \"top\" layer. Ignored unless include_top=TRUE. Set classifier_activation=NULL return logits \"top\" layer. Defaults \"softmax\". loading pretrained weights, classifier_activation can NULL \"softmax\". include_preprocessing Boolean, whether include preprocessing layer bottom network. name name model (string).","code":""},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_v2m.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Instantiates the EfficientNetV2M architecture. — application_efficientnet_v2m","text":"model instance.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_v2m.html","id":"reference","dir":"Reference","previous_headings":"","what":"Reference","title":"Instantiates the EfficientNetV2M architecture. — application_efficientnet_v2m","text":"EfficientNetV2: Smaller Models Faster Training (ICML 2021) function returns Keras image classification model, optionally loaded weights pre-trained ImageNet. image classification use cases, see page detailed examples. transfer learning use cases, make sure read guide transfer learning & fine-tuning.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_v2m.html","id":"note","dir":"Reference","previous_headings":"","what":"Note","title":"Instantiates the EfficientNetV2M architecture. — application_efficientnet_v2m","text":"Keras Application expects specific kind input preprocessing. EfficientNetV2, default input preprocessing included part model (Rescaling layer), thus application_preprocess_inputs() actually pass-function. use case, EfficientNetV2 models expect inputs float tensors pixels values [0, 255] range. time, preprocessing part model (.e. Rescaling layer) can disabled setting include_preprocessing argument FALSE. preprocessing disabled EfficientNetV2 models expect inputs float tensors pixels values [-1, 1] range.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_v2s.html","id":null,"dir":"Reference","previous_headings":"","what":"Instantiates the EfficientNetV2S architecture. — application_efficientnet_v2s","title":"Instantiates the EfficientNetV2S architecture. — application_efficientnet_v2s","text":"Instantiates EfficientNetV2S architecture.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_v2s.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Instantiates the EfficientNetV2S architecture. — application_efficientnet_v2s","text":"","code":"application_efficientnet_v2s(   include_top = TRUE,   weights = \"imagenet\",   input_tensor = NULL,   input_shape = NULL,   pooling = NULL,   classes = 1000L,   classifier_activation = \"softmax\",   include_preprocessing = TRUE,   name = \"efficientnetv2-s\" )"},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_v2s.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Instantiates the EfficientNetV2S architecture. — application_efficientnet_v2s","text":"include_top Boolean, whether include fully-connected layer top network. Defaults TRUE. weights One NULL (random initialization), \"imagenet\" (pre-training ImageNet), path weights file loaded. Defaults \"imagenet\". input_tensor Optional Keras tensor (.e. output keras_input()) use image input model. input_shape Optional shape tuple, specified include_top FALSE. exactly 3 inputs channels. pooling Optional pooling mode feature extraction include_top FALSE. Defaults NULL. NULL means output model 4D tensor output last convolutional layer. \"avg\" means global average pooling applied output last convolutional layer, thus output model 2D tensor. \"max\" means global max pooling applied. classes Optional number classes classify images , specified include_top TRUE, weights argument specified. Defaults 1000 (number ImageNet classes). classifier_activation string callable. activation function use \"top\" layer. Ignored unless include_top=TRUE. Set classifier_activation=NULL return logits \"top\" layer. Defaults \"softmax\". loading pretrained weights, classifier_activation can NULL \"softmax\". include_preprocessing Boolean, whether include preprocessing layer bottom network. name name model (string).","code":""},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_v2s.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Instantiates the EfficientNetV2S architecture. — application_efficientnet_v2s","text":"model instance.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_v2s.html","id":"reference","dir":"Reference","previous_headings":"","what":"Reference","title":"Instantiates the EfficientNetV2S architecture. — application_efficientnet_v2s","text":"EfficientNetV2: Smaller Models Faster Training (ICML 2021) function returns Keras image classification model, optionally loaded weights pre-trained ImageNet. image classification use cases, see page detailed examples. transfer learning use cases, make sure read guide transfer learning & fine-tuning.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_v2s.html","id":"note","dir":"Reference","previous_headings":"","what":"Note","title":"Instantiates the EfficientNetV2S architecture. — application_efficientnet_v2s","text":"Keras Application expects specific kind input preprocessing. EfficientNetV2, default input preprocessing included part model (Rescaling layer), thus application_preprocess_inputs() actually pass-function. use case, EfficientNetV2 models expect inputs float tensors pixels values [0, 255] range. time, preprocessing part model (.e. Rescaling layer) can disabled setting include_preprocessing argument FALSE. preprocessing disabled EfficientNetV2 models expect inputs float tensors pixels values [-1, 1] range.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/application_inception_resnet_v2.html","id":null,"dir":"Reference","previous_headings":"","what":"Instantiates the Inception-ResNet v2 architecture. — application_inception_resnet_v2","title":"Instantiates the Inception-ResNet v2 architecture. — application_inception_resnet_v2","text":"Instantiates Inception-ResNet v2 architecture.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_inception_resnet_v2.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Instantiates the Inception-ResNet v2 architecture. — application_inception_resnet_v2","text":"","code":"application_inception_resnet_v2(   include_top = TRUE,   weights = \"imagenet\",   input_tensor = NULL,   input_shape = NULL,   pooling = NULL,   classes = 1000L,   classifier_activation = \"softmax\",   name = \"inception_resnet_v2\" )"},{"path":"https://keras3.posit.co/dev/reference/application_inception_resnet_v2.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Instantiates the Inception-ResNet v2 architecture. — application_inception_resnet_v2","text":"include_top whether include fully-connected layer top network. weights one NULL (random initialization), \"imagenet\" (pre-training ImageNet), path weights file loaded. input_tensor optional Keras tensor (.e. output keras_input()) use image input model. input_shape optional shape tuple, specified include_top FALSE (otherwise input shape (299, 299, 3) ('channels_last' data format) (3, 299, 299) ('channels_first' data format). exactly 3 inputs channels, width height smaller 75. E.g. (150, 150, 3) one valid value. pooling Optional pooling mode feature extraction include_top FALSE. NULL means output model 4D tensor output last convolutional block. 'avg' means global average pooling applied output last convolutional block, thus output model 2D tensor. 'max' means global max pooling applied. classes optional number classes classify images , specified include_top TRUE, weights argument specified. classifier_activation str callable. activation function use \"top\" layer. Ignored unless include_top=TRUE. Set classifier_activation=NULL return logits \"top\" layer. loading pretrained weights, classifier_activation can NULL \"softmax\". name name model (string).","code":""},{"path":"https://keras3.posit.co/dev/reference/application_inception_resnet_v2.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Instantiates the Inception-ResNet v2 architecture. — application_inception_resnet_v2","text":"model instance.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_inception_resnet_v2.html","id":"reference","dir":"Reference","previous_headings":"","what":"Reference","title":"Instantiates the Inception-ResNet v2 architecture. — application_inception_resnet_v2","text":"Inception-v4, Inception-ResNet Impact Residual Connections Learning (AAAI 2017) function returns Keras image classification model, optionally loaded weights pre-trained ImageNet. image classification use cases, see page detailed examples. transfer learning use cases, make sure read guide transfer learning & fine-tuning.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_inception_resnet_v2.html","id":"note","dir":"Reference","previous_headings":"","what":"Note","title":"Instantiates the Inception-ResNet v2 architecture. — application_inception_resnet_v2","text":"Keras Application expects specific kind input preprocessing. InceptionResNetV2, call application_preprocess_inputs() inputs passing model. application_preprocess_inputs() scale input pixels -1 1.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/application_inception_v3.html","id":null,"dir":"Reference","previous_headings":"","what":"Instantiates the Inception v3 architecture. — application_inception_v3","title":"Instantiates the Inception v3 architecture. — application_inception_v3","text":"Instantiates Inception v3 architecture.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_inception_v3.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Instantiates the Inception v3 architecture. — application_inception_v3","text":"","code":"application_inception_v3(   include_top = TRUE,   weights = \"imagenet\",   input_tensor = NULL,   input_shape = NULL,   pooling = NULL,   classes = 1000L,   classifier_activation = \"softmax\",   name = \"inception_v3\" )"},{"path":"https://keras3.posit.co/dev/reference/application_inception_v3.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Instantiates the Inception v3 architecture. — application_inception_v3","text":"include_top Boolean, whether include fully-connected layer top, last layer network. Defaults TRUE. weights One NULL (random initialization), imagenet (pre-training ImageNet), path weights file loaded. Defaults \"imagenet\". input_tensor Optional Keras tensor (.e. output keras_input()) use image input model. input_tensor useful sharing inputs multiple different networks. Defaults NULL. input_shape Optional shape tuple, specified include_top FALSE (otherwise input shape (299, 299, 3) (channels_last data format) (3, 299, 299) (channels_first data format). exactly 3 inputs channels, width height smaller 75. E.g. (150, 150, 3) one valid value. input_shape ignored input_tensor provided. pooling Optional pooling mode feature extraction include_top FALSE. NULL (default) means output model 4D tensor output last convolutional block. avg means global average pooling applied output last convolutional block, thus output model 2D tensor. max means global max pooling applied. classes optional number classes classify images , specified include_top TRUE, weights argument specified. Defaults 1000. classifier_activation str callable. activation function use \"top\" layer. Ignored unless include_top=TRUE. Set classifier_activation=NULL return logits \"top\" layer. loading pretrained weights, classifier_activation can NULL \"softmax\". name name model (string).","code":""},{"path":"https://keras3.posit.co/dev/reference/application_inception_v3.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Instantiates the Inception v3 architecture. — application_inception_v3","text":"model instance.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_inception_v3.html","id":"reference","dir":"Reference","previous_headings":"","what":"Reference","title":"Instantiates the Inception v3 architecture. — application_inception_v3","text":"Rethinking Inception Architecture Computer Vision (CVPR 2016) function returns Keras image classification model, optionally loaded weights pre-trained ImageNet. image classification use cases, see page detailed examples. transfer learning use cases, make sure read guide transfer learning & fine-tuning.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_inception_v3.html","id":"note","dir":"Reference","previous_headings":"","what":"Note","title":"Instantiates the Inception v3 architecture. — application_inception_v3","text":"Keras Application expects specific kind input preprocessing. InceptionV3, call application_preprocess_inputs() inputs passing model. application_preprocess_inputs() scale input pixels -1 1.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/application_mobilenet.html","id":null,"dir":"Reference","previous_headings":"","what":"Instantiates the MobileNet architecture. — application_mobilenet","title":"Instantiates the MobileNet architecture. — application_mobilenet","text":"Instantiates MobileNet architecture.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_mobilenet.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Instantiates the MobileNet architecture. — application_mobilenet","text":"","code":"application_mobilenet(   input_shape = NULL,   alpha = 1,   depth_multiplier = 1L,   dropout = 0.001,   include_top = TRUE,   weights = \"imagenet\",   input_tensor = NULL,   pooling = NULL,   classes = 1000L,   classifier_activation = \"softmax\",   name = NULL )"},{"path":"https://keras3.posit.co/dev/reference/application_mobilenet.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Instantiates the MobileNet architecture. — application_mobilenet","text":"input_shape Optional shape tuple, specified include_top FALSE (otherwise input shape (224, 224, 3) (\"channels_last\" data format) (3, 224, 224) (\"channels_first\" data format). exactly 3 inputs channels, width height smaller 32. E.g. (200, 200, 3) one valid value. Defaults NULL. input_shape ignored input_tensor provided. alpha Controls width network. known width multiplier MobileNet paper. alpha < 1.0, proportionally decreases number filters layer. alpha > 1.0, proportionally increases number filters layer. alpha == 1, default number filters paper used layer. Defaults 1.0. depth_multiplier Depth multiplier depthwise convolution. called resolution multiplier MobileNet paper. Defaults 1.0. dropout Dropout rate. Defaults 0.001. include_top Boolean, whether include fully-connected layer top network. Defaults TRUE. weights One NULL (random initialization), \"imagenet\" (pre-training ImageNet), path weights file loaded. Defaults \"imagenet\". input_tensor Optional Keras tensor (.e. output keras_input()) use image input model. input_tensor useful sharing inputs multiple different networks. Defaults NULL. pooling Optional pooling mode feature extraction include_top FALSE. NULL (default) means output model 4D tensor output last convolutional block. avg means global average pooling applied output last convolutional block, thus output model 2D tensor. max means global max pooling applied. classes Optional number classes classify images , specified include_top TRUE, weights argument specified. Defaults 1000. classifier_activation str callable. activation function use \"top\" layer. Ignored unless include_top=TRUE. Set classifier_activation=NULL return logits \"top\" layer. loading pretrained weights, classifier_activation can NULL \"softmax\". name name model (string).","code":""},{"path":"https://keras3.posit.co/dev/reference/application_mobilenet.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Instantiates the MobileNet architecture. — application_mobilenet","text":"model instance.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_mobilenet.html","id":"reference","dir":"Reference","previous_headings":"","what":"Reference","title":"Instantiates the MobileNet architecture. — application_mobilenet","text":"MobileNets: Efficient Convolutional Neural Networks Mobile Vision Applications function returns Keras image classification model, optionally loaded weights pre-trained ImageNet. image classification use cases, see page detailed examples. transfer learning use cases, make sure read guide transfer learning & fine-tuning.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_mobilenet.html","id":"note","dir":"Reference","previous_headings":"","what":"Note","title":"Instantiates the MobileNet architecture. — application_mobilenet","text":"Keras Application expects specific kind input preprocessing. MobileNet, call application_preprocess_inputs() inputs passing model. application_preprocess_inputs() scale input pixels -1 1.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/application_mobilenet_v2.html","id":null,"dir":"Reference","previous_headings":"","what":"Instantiates the MobileNetV2 architecture. — application_mobilenet_v2","title":"Instantiates the MobileNetV2 architecture. — application_mobilenet_v2","text":"MobileNetV2 similar original MobileNet, except uses inverted residual blocks bottlenecking features. drastically lower parameter count original MobileNet. MobileNets support input size greater 32 x 32, larger image sizes offering better performance.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_mobilenet_v2.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Instantiates the MobileNetV2 architecture. — application_mobilenet_v2","text":"","code":"application_mobilenet_v2(   input_shape = NULL,   alpha = 1,   include_top = TRUE,   weights = \"imagenet\",   input_tensor = NULL,   pooling = NULL,   classes = 1000L,   classifier_activation = \"softmax\",   name = NULL )"},{"path":"https://keras3.posit.co/dev/reference/application_mobilenet_v2.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Instantiates the MobileNetV2 architecture. — application_mobilenet_v2","text":"input_shape Optional shape tuple, specified include_top FALSE (otherwise input shape (224, 224, 3) (\"channels_last\" data format) (3, 224, 224) (\"channels_first\" data format). exactly 3 inputs channels, width height smaller 32. E.g. (200, 200, 3) one valid value. Defaults NULL. input_shape ignored input_tensor provided. alpha Controls width network. known width multiplier MobileNet paper. alpha < 1.0, proportionally decreases number filters layer. alpha > 1.0, proportionally increases number filters layer. alpha == 1, default number filters paper used layer. Defaults 1.0. include_top Boolean, whether include fully-connected layer top network. Defaults TRUE. weights One NULL (random initialization), \"imagenet\" (pre-training ImageNet), path weights file loaded. Defaults \"imagenet\". input_tensor Optional Keras tensor (.e. output keras_input()) use image input model. input_tensor useful sharing inputs multiple different networks. Defaults NULL. pooling Optional pooling mode feature extraction include_top FALSE. NULL (default) means output model 4D tensor output last convolutional block. avg means global average pooling applied output last convolutional block, thus output model 2D tensor. max means global max pooling applied. classes Optional number classes classify images , specified include_top TRUE, weights argument specified. Defaults 1000. classifier_activation str callable. activation function use \"top\" layer. Ignored unless include_top=TRUE. Set classifier_activation=NULL return logits \"top\" layer. loading pretrained weights, classifier_activation can NULL \"softmax\". name name model (string).","code":""},{"path":"https://keras3.posit.co/dev/reference/application_mobilenet_v2.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Instantiates the MobileNetV2 architecture. — application_mobilenet_v2","text":"model instance.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_mobilenet_v2.html","id":"reference","dir":"Reference","previous_headings":"","what":"Reference","title":"Instantiates the MobileNetV2 architecture. — application_mobilenet_v2","text":"MobileNetV2: Inverted Residuals Linear Bottlenecks (CVPR 2018) function returns Keras image classification model, optionally loaded weights pre-trained ImageNet. image classification use cases, see page detailed examples. transfer learning use cases, make sure read guide transfer learning & fine-tuning.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_mobilenet_v2.html","id":"note","dir":"Reference","previous_headings":"","what":"Note","title":"Instantiates the MobileNetV2 architecture. — application_mobilenet_v2","text":"Keras Application expects specific kind input preprocessing. MobileNetV2, call application_preprocess_inputs() inputs passing model. application_preprocess_inputs() scale input pixels -1 1.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/application_mobilenet_v3_large.html","id":null,"dir":"Reference","previous_headings":"","what":"Instantiates the MobileNetV3Large architecture. — application_mobilenet_v3_large","title":"Instantiates the MobileNetV3Large architecture. — application_mobilenet_v3_large","text":"Instantiates MobileNetV3Large architecture.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_mobilenet_v3_large.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Instantiates the MobileNetV3Large architecture. — application_mobilenet_v3_large","text":"","code":"application_mobilenet_v3_large(   input_shape = NULL,   alpha = 1,   minimalistic = FALSE,   include_top = TRUE,   weights = \"imagenet\",   input_tensor = NULL,   classes = 1000L,   pooling = NULL,   dropout_rate = 0.2,   classifier_activation = \"softmax\",   include_preprocessing = TRUE,   name = \"MobileNetV3Large\" )"},{"path":"https://keras3.posit.co/dev/reference/application_mobilenet_v3_large.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Instantiates the MobileNetV3Large architecture. — application_mobilenet_v3_large","text":"input_shape Optional shape tuple, specified like use model input image resolution (224, 224, 3). exactly 3 inputs channels. can also omit option like infer input_shape input_tensor. choose include input_tensor input_shape input_shape used match, shapes match throw error. E.g. (160, 160, 3) one valid value. alpha controls width network. known depth multiplier MobileNetV3 paper, name kept consistency MobileNetV1 Keras. alpha < 1.0, proportionally decreases number filters layer. alpha > 1.0, proportionally increases number filters layer. alpha == 1, default number filters paper used layer. minimalistic addition large small models module also contains -called minimalistic models, models per-layer dimensions characteristic MobilenetV3 however, utilize advanced blocks (squeeze--excite units, hard-swish, 5x5 convolutions). models less efficient CPU, much performant GPU/DSP. include_top Boolean, whether include fully-connected layer top network. Defaults TRUE. weights String, one NULL (random initialization), \"imagenet\" (pre-training ImageNet), path weights file loaded. input_tensor Optional Keras tensor (.e. output keras_input()) use image input model. classes Integer, optional number classes classify images , specified include_top TRUE, weights argument specified. pooling String, optional pooling mode feature extraction include_top FALSE. NULL means output model 4D tensor output last convolutional block. avg means global average pooling applied output last convolutional block, thus output model 2D tensor. max means global max pooling applied. dropout_rate fraction input units drop last layer. classifier_activation str callable. activation function use \"top\" layer. Ignored unless include_top=TRUE. Set classifier_activation=NULL return logits \"top\" layer. loading pretrained weights, classifier_activation can NULL \"softmax\". include_preprocessing Boolean, whether include preprocessing layer (Rescaling) bottom network. Defaults TRUE. name name model (string).","code":""},{"path":"https://keras3.posit.co/dev/reference/application_mobilenet_v3_large.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Instantiates the MobileNetV3Large architecture. — application_mobilenet_v3_large","text":"model instance.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_mobilenet_v3_large.html","id":"reference","dir":"Reference","previous_headings":"","what":"Reference","title":"Instantiates the MobileNetV3Large architecture. — application_mobilenet_v3_large","text":"Searching MobileNetV3 (ICCV 2019)","code":""},{"path":"https://keras3.posit.co/dev/reference/application_mobilenet_v3_large.html","id":"the-following-table-describes-the-performance-of-mobilenets-v-","dir":"Reference","previous_headings":"","what":"The following table describes the performance of MobileNets v3:","title":"Instantiates the MobileNetV3Large architecture. — application_mobilenet_v3_large","text":"MACs stands Multiply Adds image classification use cases, see page detailed examples. transfer learning use cases, make sure read guide transfer learning & fine-tuning.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_mobilenet_v3_large.html","id":"note","dir":"Reference","previous_headings":"","what":"Note","title":"Instantiates the MobileNetV3Large architecture. — application_mobilenet_v3_large","text":"Keras Application expects specific kind input preprocessing. MobileNetV3, default input preprocessing included part model (Rescaling layer), thus application_preprocess_inputs() actually pass-function. use case, MobileNetV3 models expect inputs float tensors pixels values [0-255] range. time, preprocessing part model (.e. Rescaling layer) can disabled setting include_preprocessing argument FALSE. preprocessing disabled MobileNetV3 models expect inputs float tensors pixels values [-1, 1] range.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_mobilenet_v3_large.html","id":"call-arguments","dir":"Reference","previous_headings":"","what":"Call Arguments","title":"Instantiates the MobileNetV3Large architecture. — application_mobilenet_v3_large","text":"inputs: floating point numpy.array backend-native tensor, 4D 3 color channels, values range [0, 255] include_preprocessing TRUE range [-1, 1] otherwise.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/application_mobilenet_v3_small.html","id":null,"dir":"Reference","previous_headings":"","what":"Instantiates the MobileNetV3Small architecture. — application_mobilenet_v3_small","title":"Instantiates the MobileNetV3Small architecture. — application_mobilenet_v3_small","text":"Instantiates MobileNetV3Small architecture.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_mobilenet_v3_small.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Instantiates the MobileNetV3Small architecture. — application_mobilenet_v3_small","text":"","code":"application_mobilenet_v3_small(   input_shape = NULL,   alpha = 1,   minimalistic = FALSE,   include_top = TRUE,   weights = \"imagenet\",   input_tensor = NULL,   classes = 1000L,   pooling = NULL,   dropout_rate = 0.2,   classifier_activation = \"softmax\",   include_preprocessing = TRUE,   name = \"MobileNetV3Small\" )"},{"path":"https://keras3.posit.co/dev/reference/application_mobilenet_v3_small.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Instantiates the MobileNetV3Small architecture. — application_mobilenet_v3_small","text":"input_shape Optional shape tuple, specified like use model input image resolution (224, 224, 3). exactly 3 inputs channels. can also omit option like infer input_shape input_tensor. choose include input_tensor input_shape input_shape used match, shapes match throw error. E.g. (160, 160, 3) one valid value. alpha controls width network. known depth multiplier MobileNetV3 paper, name kept consistency MobileNetV1 Keras. alpha < 1.0, proportionally decreases number filters layer. alpha > 1.0, proportionally increases number filters layer. alpha == 1, default number filters paper used layer. minimalistic addition large small models module also contains -called minimalistic models, models per-layer dimensions characteristic MobilenetV3 however, utilize advanced blocks (squeeze--excite units, hard-swish, 5x5 convolutions). models less efficient CPU, much performant GPU/DSP. include_top Boolean, whether include fully-connected layer top network. Defaults TRUE. weights String, one NULL (random initialization), \"imagenet\" (pre-training ImageNet), path weights file loaded. input_tensor Optional Keras tensor (.e. output keras_input()) use image input model. classes Integer, optional number classes classify images , specified include_top TRUE, weights argument specified. pooling String, optional pooling mode feature extraction include_top FALSE. NULL means output model 4D tensor output last convolutional block. avg means global average pooling applied output last convolutional block, thus output model 2D tensor. max means global max pooling applied. dropout_rate fraction input units drop last layer. classifier_activation str callable. activation function use \"top\" layer. Ignored unless include_top=TRUE. Set classifier_activation=NULL return logits \"top\" layer. loading pretrained weights, classifier_activation can NULL \"softmax\". include_preprocessing Boolean, whether include preprocessing layer (Rescaling) bottom network. Defaults TRUE. name name model (string).","code":""},{"path":"https://keras3.posit.co/dev/reference/application_mobilenet_v3_small.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Instantiates the MobileNetV3Small architecture. — application_mobilenet_v3_small","text":"model instance.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_mobilenet_v3_small.html","id":"reference","dir":"Reference","previous_headings":"","what":"Reference","title":"Instantiates the MobileNetV3Small architecture. — application_mobilenet_v3_small","text":"Searching MobileNetV3 (ICCV 2019)","code":""},{"path":"https://keras3.posit.co/dev/reference/application_mobilenet_v3_small.html","id":"the-following-table-describes-the-performance-of-mobilenets-v-","dir":"Reference","previous_headings":"","what":"The following table describes the performance of MobileNets v3:","title":"Instantiates the MobileNetV3Small architecture. — application_mobilenet_v3_small","text":"MACs stands Multiply Adds image classification use cases, see page detailed examples. transfer learning use cases, make sure read guide transfer learning & fine-tuning.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_mobilenet_v3_small.html","id":"note","dir":"Reference","previous_headings":"","what":"Note","title":"Instantiates the MobileNetV3Small architecture. — application_mobilenet_v3_small","text":"Keras Application expects specific kind input preprocessing. MobileNetV3, default input preprocessing included part model (Rescaling layer), thus application_preprocess_inputs() actually pass-function. use case, MobileNetV3 models expect inputs float tensors pixels values [0-255] range. time, preprocessing part model (.e. Rescaling layer) can disabled setting include_preprocessing argument FALSE. preprocessing disabled MobileNetV3 models expect inputs float tensors pixels values [-1, 1] range.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_mobilenet_v3_small.html","id":"call-arguments","dir":"Reference","previous_headings":"","what":"Call Arguments","title":"Instantiates the MobileNetV3Small architecture. — application_mobilenet_v3_small","text":"inputs: floating point numpy.array backend-native tensor, 4D 3 color channels, values range [0, 255] include_preprocessing TRUE range [-1, 1] otherwise.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/application_nasnet_large.html","id":null,"dir":"Reference","previous_headings":"","what":"Instantiates a NASNet model in ImageNet mode. — application_nasnet_large","title":"Instantiates a NASNet model in ImageNet mode. — application_nasnet_large","text":"Instantiates NASNet model ImageNet mode.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_nasnet_large.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Instantiates a NASNet model in ImageNet mode. — application_nasnet_large","text":"","code":"application_nasnet_large(   input_shape = NULL,   include_top = TRUE,   weights = \"imagenet\",   input_tensor = NULL,   pooling = NULL,   classes = 1000L,   classifier_activation = \"softmax\",   name = \"nasnet_large\" )"},{"path":"https://keras3.posit.co/dev/reference/application_nasnet_large.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Instantiates a NASNet model in ImageNet mode. — application_nasnet_large","text":"input_shape Optional shape tuple, specified include_top FALSE (otherwise input shape (331, 331, 3) NASNetLarge. exactly 3 inputs channels, width height smaller 32. E.g. (224, 224, 3) one valid value. include_top Whether include fully-connected layer top network. weights NULL (random initialization) imagenet (ImageNet weights).  loading imagenet weights, input_shape (331, 331, 3) input_tensor Optional Keras tensor (.e. output keras_input()) use image input model. pooling Optional pooling mode feature extraction include_top FALSE. NULL means output model 4D tensor output last convolutional layer. avg means global average pooling applied output last convolutional layer, thus output model 2D tensor. max means global max pooling applied. classes Optional number classes classify images , specified include_top TRUE, weights argument specified. classifier_activation str callable. activation function use \"top\" layer. Ignored unless include_top=TRUE. Set classifier_activation=NULL return logits \"top\" layer.  loading pretrained weights, classifier_activation can NULL \"softmax\". name name model (string).","code":""},{"path":"https://keras3.posit.co/dev/reference/application_nasnet_large.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Instantiates a NASNet model in ImageNet mode. — application_nasnet_large","text":"Keras model instance.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_nasnet_large.html","id":"reference","dir":"Reference","previous_headings":"","what":"Reference","title":"Instantiates a NASNet model in ImageNet mode. — application_nasnet_large","text":"Learning Transferable Architectures Scalable Image Recognition (CVPR 2018) Optionally loads weights pre-trained ImageNet. Note data format convention used model one specified Keras config ~/.keras/keras.json.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_nasnet_large.html","id":"note","dir":"Reference","previous_headings":"","what":"Note","title":"Instantiates a NASNet model in ImageNet mode. — application_nasnet_large","text":"Keras Application expects specific kind input preprocessing. NASNet, call application_preprocess_inputs() inputs passing model.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/application_nasnet_mobile.html","id":null,"dir":"Reference","previous_headings":"","what":"Instantiates a Mobile NASNet model in ImageNet mode. — application_nasnet_mobile","title":"Instantiates a Mobile NASNet model in ImageNet mode. — application_nasnet_mobile","text":"Instantiates Mobile NASNet model ImageNet mode.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_nasnet_mobile.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Instantiates a Mobile NASNet model in ImageNet mode. — application_nasnet_mobile","text":"","code":"application_nasnet_mobile(   input_shape = NULL,   include_top = TRUE,   weights = \"imagenet\",   input_tensor = NULL,   pooling = NULL,   classes = 1000L,   classifier_activation = \"softmax\",   name = \"nasnet_mobile\" )"},{"path":"https://keras3.posit.co/dev/reference/application_nasnet_mobile.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Instantiates a Mobile NASNet model in ImageNet mode. — application_nasnet_mobile","text":"input_shape Optional shape tuple, specified include_top FALSE (otherwise input shape (224, 224, 3) NASNetMobile exactly 3 inputs channels, width height smaller 32. E.g. (224, 224, 3) one valid value. include_top Whether include fully-connected layer top network. weights NULL (random initialization) imagenet (ImageNet weights). loading imagenet weights, input_shape (224, 224, 3) input_tensor Optional Keras tensor (.e. output keras_input()) use image input model. pooling Optional pooling mode feature extraction include_top FALSE. NULL means output model 4D tensor output last convolutional layer. avg means global average pooling applied output last convolutional layer, thus output model 2D tensor. max means global max pooling applied. classes Optional number classes classify images , specified include_top TRUE, weights argument specified. classifier_activation str callable. activation function use \"top\" layer. Ignored unless include_top=TRUE. Set classifier_activation=NULL return logits \"top\" layer.  loading pretrained weights, classifier_activation can NULL \"softmax\". name name model (string).","code":""},{"path":"https://keras3.posit.co/dev/reference/application_nasnet_mobile.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Instantiates a Mobile NASNet model in ImageNet mode. — application_nasnet_mobile","text":"Keras model instance.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_nasnet_mobile.html","id":"reference","dir":"Reference","previous_headings":"","what":"Reference","title":"Instantiates a Mobile NASNet model in ImageNet mode. — application_nasnet_mobile","text":"Learning Transferable Architectures Scalable Image Recognition (CVPR 2018) Optionally loads weights pre-trained ImageNet. Note data format convention used model one specified Keras config ~/.keras/keras.json.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_nasnet_mobile.html","id":"note","dir":"Reference","previous_headings":"","what":"Note","title":"Instantiates a Mobile NASNet model in ImageNet mode. — application_nasnet_mobile","text":"Keras Application expects specific kind input preprocessing. NASNet, call application_preprocess_inputs() inputs passing model.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/application_resnet101.html","id":null,"dir":"Reference","previous_headings":"","what":"Instantiates the ResNet101 architecture. — application_resnet101","title":"Instantiates the ResNet101 architecture. — application_resnet101","text":"Instantiates ResNet101 architecture.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_resnet101.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Instantiates the ResNet101 architecture. — application_resnet101","text":"","code":"application_resnet101(   include_top = TRUE,   weights = \"imagenet\",   input_tensor = NULL,   input_shape = NULL,   pooling = NULL,   classes = 1000L,   classifier_activation = \"softmax\",   name = \"resnet101\" )"},{"path":"https://keras3.posit.co/dev/reference/application_resnet101.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Instantiates the ResNet101 architecture. — application_resnet101","text":"include_top whether include fully-connected layer top network. weights one NULL (random initialization), \"imagenet\" (pre-training ImageNet), path weights file loaded. input_tensor optional Keras tensor (.e. output keras_input()) use image input model. input_shape optional shape tuple, specified include_top FALSE (otherwise input shape (224, 224, 3) (\"channels_last\" data format) (3, 224, 224) (\"channels_first\" data format). exactly 3 inputs channels, width height smaller 32. E.g. (200, 200, 3) one valid value. pooling Optional pooling mode feature extraction include_top FALSE. NULL means output model 4D tensor output last convolutional block. avg means global average pooling applied output last convolutional block, thus output model 2D tensor. max means global max pooling applied. classes optional number classes classify images , specified include_top TRUE, weights argument specified. classifier_activation str callable. activation function use \"top\" layer. Ignored unless include_top=TRUE. Set classifier_activation=NULL return logits \"top\" layer. loading pretrained weights, classifier_activation can NULL \"softmax\". name name model (string).","code":""},{"path":"https://keras3.posit.co/dev/reference/application_resnet101.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Instantiates the ResNet101 architecture. — application_resnet101","text":"Model instance.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_resnet101.html","id":"reference","dir":"Reference","previous_headings":"","what":"Reference","title":"Instantiates the ResNet101 architecture. — application_resnet101","text":"Deep Residual Learning Image Recognition (CVPR 2015) image classification use cases, see page detailed examples. transfer learning use cases, make sure read guide transfer learning & fine-tuning.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_resnet101.html","id":"note","dir":"Reference","previous_headings":"","what":"Note","title":"Instantiates the ResNet101 architecture. — application_resnet101","text":"Keras Application expects specific kind input preprocessing. ResNet, call application_preprocess_inputs() inputs passing model. application_preprocess_inputs() convert input images RGB BGR, zero-center color channel respect ImageNet dataset, without scaling.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/application_resnet101_v2.html","id":null,"dir":"Reference","previous_headings":"","what":"Instantiates the ResNet101V2 architecture. — application_resnet101_v2","title":"Instantiates the ResNet101V2 architecture. — application_resnet101_v2","text":"Instantiates ResNet101V2 architecture.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_resnet101_v2.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Instantiates the ResNet101V2 architecture. — application_resnet101_v2","text":"","code":"application_resnet101_v2(   include_top = TRUE,   weights = \"imagenet\",   input_tensor = NULL,   input_shape = NULL,   pooling = NULL,   classes = 1000L,   classifier_activation = \"softmax\",   name = \"resnet101v2\" )"},{"path":"https://keras3.posit.co/dev/reference/application_resnet101_v2.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Instantiates the ResNet101V2 architecture. — application_resnet101_v2","text":"include_top whether include fully-connected layer top network. weights one NULL (random initialization), \"imagenet\" (pre-training ImageNet), path weights file loaded. input_tensor optional Keras tensor (.e. output keras_input()) use image input model. input_shape optional shape tuple, specified include_top FALSE (otherwise input shape (224, 224, 3) (\"channels_last\" data format) (3, 224, 224) (\"channels_first\" data format). exactly 3 inputs channels, width height smaller 32. E.g. (200, 200, 3) one valid value. pooling Optional pooling mode feature extraction include_top FALSE. NULL means output model 4D tensor output last convolutional block. avg means global average pooling applied output last convolutional block, thus output model 2D tensor. max means global max pooling applied. classes optional number classes classify images , specified include_top TRUE, weights argument specified. classifier_activation str callable. activation function use \"top\" layer. Ignored unless include_top=TRUE. Set classifier_activation=NULL return logits \"top\" layer. loading pretrained weights, classifier_activation can NULL \"softmax\". name name model (string).","code":""},{"path":"https://keras3.posit.co/dev/reference/application_resnet101_v2.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Instantiates the ResNet101V2 architecture. — application_resnet101_v2","text":"Model instance.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_resnet101_v2.html","id":"reference","dir":"Reference","previous_headings":"","what":"Reference","title":"Instantiates the ResNet101V2 architecture. — application_resnet101_v2","text":"Identity Mappings Deep Residual Networks (CVPR 2016) image classification use cases, see page detailed examples. transfer learning use cases, make sure read guide transfer learning & fine-tuning.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_resnet101_v2.html","id":"note","dir":"Reference","previous_headings":"","what":"Note","title":"Instantiates the ResNet101V2 architecture. — application_resnet101_v2","text":"Keras Application expects specific kind input preprocessing. ResNet, call application_preprocess_inputs() inputs passing model. application_preprocess_inputs() scale input pixels -1 1.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/application_resnet152.html","id":null,"dir":"Reference","previous_headings":"","what":"Instantiates the ResNet152 architecture. — application_resnet152","title":"Instantiates the ResNet152 architecture. — application_resnet152","text":"Instantiates ResNet152 architecture.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_resnet152.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Instantiates the ResNet152 architecture. — application_resnet152","text":"","code":"application_resnet152(   include_top = TRUE,   weights = \"imagenet\",   input_tensor = NULL,   input_shape = NULL,   pooling = NULL,   classes = 1000L,   classifier_activation = \"softmax\",   name = \"resnet152\" )"},{"path":"https://keras3.posit.co/dev/reference/application_resnet152.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Instantiates the ResNet152 architecture. — application_resnet152","text":"include_top whether include fully-connected layer top network. weights one NULL (random initialization), \"imagenet\" (pre-training ImageNet), path weights file loaded. input_tensor optional Keras tensor (.e. output keras_input()) use image input model. input_shape optional shape tuple, specified include_top FALSE (otherwise input shape (224, 224, 3) (\"channels_last\" data format) (3, 224, 224) (\"channels_first\" data format). exactly 3 inputs channels, width height smaller 32. E.g. (200, 200, 3) one valid value. pooling Optional pooling mode feature extraction include_top FALSE. NULL means output model 4D tensor output last convolutional block. avg means global average pooling applied output last convolutional block, thus output model 2D tensor. max means global max pooling applied. classes optional number classes classify images , specified include_top TRUE, weights argument specified. classifier_activation str callable. activation function use \"top\" layer. Ignored unless include_top=TRUE. Set classifier_activation=NULL return logits \"top\" layer. loading pretrained weights, classifier_activation can NULL \"softmax\". name name model (string).","code":""},{"path":"https://keras3.posit.co/dev/reference/application_resnet152.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Instantiates the ResNet152 architecture. — application_resnet152","text":"Model instance.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_resnet152.html","id":"reference","dir":"Reference","previous_headings":"","what":"Reference","title":"Instantiates the ResNet152 architecture. — application_resnet152","text":"Deep Residual Learning Image Recognition (CVPR 2015) image classification use cases, see page detailed examples. transfer learning use cases, make sure read guide transfer learning & fine-tuning.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_resnet152.html","id":"note","dir":"Reference","previous_headings":"","what":"Note","title":"Instantiates the ResNet152 architecture. — application_resnet152","text":"Keras Application expects specific kind input preprocessing. ResNet, call application_preprocess_inputs() inputs passing model. application_preprocess_inputs() convert input images RGB BGR, zero-center color channel respect ImageNet dataset, without scaling.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/application_resnet152_v2.html","id":null,"dir":"Reference","previous_headings":"","what":"Instantiates the ResNet152V2 architecture. — application_resnet152_v2","title":"Instantiates the ResNet152V2 architecture. — application_resnet152_v2","text":"Instantiates ResNet152V2 architecture.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_resnet152_v2.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Instantiates the ResNet152V2 architecture. — application_resnet152_v2","text":"","code":"application_resnet152_v2(   include_top = TRUE,   weights = \"imagenet\",   input_tensor = NULL,   input_shape = NULL,   pooling = NULL,   classes = 1000L,   classifier_activation = \"softmax\",   name = \"resnet152v2\" )"},{"path":"https://keras3.posit.co/dev/reference/application_resnet152_v2.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Instantiates the ResNet152V2 architecture. — application_resnet152_v2","text":"include_top whether include fully-connected layer top network. weights one NULL (random initialization), \"imagenet\" (pre-training ImageNet), path weights file loaded. input_tensor optional Keras tensor (.e. output keras_input()) use image input model. input_shape optional shape tuple, specified include_top FALSE (otherwise input shape (224, 224, 3) (\"channels_last\" data format) (3, 224, 224) (\"channels_first\" data format). exactly 3 inputs channels, width height smaller 32. E.g. (200, 200, 3) one valid value. pooling Optional pooling mode feature extraction include_top FALSE. NULL means output model 4D tensor output last convolutional block. avg means global average pooling applied output last convolutional block, thus output model 2D tensor. max means global max pooling applied. classes optional number classes classify images , specified include_top TRUE, weights argument specified. classifier_activation str callable. activation function use \"top\" layer. Ignored unless include_top=TRUE. Set classifier_activation=NULL return logits \"top\" layer. loading pretrained weights, classifier_activation can NULL \"softmax\". name name model (string).","code":""},{"path":"https://keras3.posit.co/dev/reference/application_resnet152_v2.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Instantiates the ResNet152V2 architecture. — application_resnet152_v2","text":"Model instance.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_resnet152_v2.html","id":"reference","dir":"Reference","previous_headings":"","what":"Reference","title":"Instantiates the ResNet152V2 architecture. — application_resnet152_v2","text":"Identity Mappings Deep Residual Networks (CVPR 2016) image classification use cases, see page detailed examples. transfer learning use cases, make sure read guide transfer learning & fine-tuning.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_resnet152_v2.html","id":"note","dir":"Reference","previous_headings":"","what":"Note","title":"Instantiates the ResNet152V2 architecture. — application_resnet152_v2","text":"Keras Application expects specific kind input preprocessing. ResNet, call application_preprocess_inputs() inputs passing model. application_preprocess_inputs() scale input pixels -1 1.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/application_resnet50.html","id":null,"dir":"Reference","previous_headings":"","what":"Instantiates the ResNet50 architecture. — application_resnet50","title":"Instantiates the ResNet50 architecture. — application_resnet50","text":"Instantiates ResNet50 architecture.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_resnet50.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Instantiates the ResNet50 architecture. — application_resnet50","text":"","code":"application_resnet50(   include_top = TRUE,   weights = \"imagenet\",   input_tensor = NULL,   input_shape = NULL,   pooling = NULL,   classes = 1000L,   classifier_activation = \"softmax\",   name = \"resnet50\" )"},{"path":"https://keras3.posit.co/dev/reference/application_resnet50.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Instantiates the ResNet50 architecture. — application_resnet50","text":"include_top whether include fully-connected layer top network. weights one NULL (random initialization), \"imagenet\" (pre-training ImageNet), path weights file loaded. input_tensor optional Keras tensor (.e. output keras_input()) use image input model. input_shape optional shape tuple, specified include_top FALSE (otherwise input shape (224, 224, 3) (\"channels_last\" data format) (3, 224, 224) (\"channels_first\" data format). exactly 3 inputs channels, width height smaller 32. E.g. (200, 200, 3) one valid value. pooling Optional pooling mode feature extraction include_top FALSE. NULL means output model 4D tensor output last convolutional block. avg means global average pooling applied output last convolutional block, thus output model 2D tensor. max means global max pooling applied. classes optional number classes classify images , specified include_top TRUE, weights argument specified. Defaults 1000. classifier_activation str callable. activation function use \"top\" layer. Ignored unless include_top=TRUE. Set classifier_activation=NULL return logits \"top\" layer. loading pretrained weights, classifier_activation can NULL \"softmax\". name name model (string).","code":""},{"path":"https://keras3.posit.co/dev/reference/application_resnet50.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Instantiates the ResNet50 architecture. — application_resnet50","text":"Model instance.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_resnet50.html","id":"reference","dir":"Reference","previous_headings":"","what":"Reference","title":"Instantiates the ResNet50 architecture. — application_resnet50","text":"Deep Residual Learning Image Recognition (CVPR 2015) image classification use cases, see page detailed examples. transfer learning use cases, make sure read guide transfer learning & fine-tuning.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_resnet50.html","id":"note","dir":"Reference","previous_headings":"","what":"Note","title":"Instantiates the ResNet50 architecture. — application_resnet50","text":"Keras Application expects specific kind input preprocessing. ResNet, call application_preprocess_inputs() inputs passing model. application_preprocess_inputs() convert input images RGB BGR, zero-center color channel respect ImageNet dataset, without scaling.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/application_resnet50_v2.html","id":null,"dir":"Reference","previous_headings":"","what":"Instantiates the ResNet50V2 architecture. — application_resnet50_v2","title":"Instantiates the ResNet50V2 architecture. — application_resnet50_v2","text":"Instantiates ResNet50V2 architecture.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_resnet50_v2.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Instantiates the ResNet50V2 architecture. — application_resnet50_v2","text":"","code":"application_resnet50_v2(   include_top = TRUE,   weights = \"imagenet\",   input_tensor = NULL,   input_shape = NULL,   pooling = NULL,   classes = 1000L,   classifier_activation = \"softmax\",   name = \"resnet50v2\" )"},{"path":"https://keras3.posit.co/dev/reference/application_resnet50_v2.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Instantiates the ResNet50V2 architecture. — application_resnet50_v2","text":"include_top whether include fully-connected layer top network. weights one NULL (random initialization), \"imagenet\" (pre-training ImageNet), path weights file loaded. input_tensor optional Keras tensor (.e. output keras_input()) use image input model. input_shape optional shape tuple, specified include_top FALSE (otherwise input shape (224, 224, 3) (\"channels_last\" data format) (3, 224, 224) (\"channels_first\" data format). exactly 3 inputs channels, width height smaller 32. E.g. (200, 200, 3) one valid value. pooling Optional pooling mode feature extraction include_top FALSE. NULL means output model 4D tensor output last convolutional block. avg means global average pooling applied output last convolutional block, thus output model 2D tensor. max means global max pooling applied. classes optional number classes classify images , specified include_top TRUE, weights argument specified. classifier_activation str callable. activation function use \"top\" layer. Ignored unless include_top=TRUE. Set classifier_activation=NULL return logits \"top\" layer. loading pretrained weights, classifier_activation can NULL \"softmax\". name name model (string).","code":""},{"path":"https://keras3.posit.co/dev/reference/application_resnet50_v2.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Instantiates the ResNet50V2 architecture. — application_resnet50_v2","text":"Model instance.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_resnet50_v2.html","id":"reference","dir":"Reference","previous_headings":"","what":"Reference","title":"Instantiates the ResNet50V2 architecture. — application_resnet50_v2","text":"Identity Mappings Deep Residual Networks (CVPR 2016) image classification use cases, see page detailed examples. transfer learning use cases, make sure read guide transfer learning & fine-tuning.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_resnet50_v2.html","id":"note","dir":"Reference","previous_headings":"","what":"Note","title":"Instantiates the ResNet50V2 architecture. — application_resnet50_v2","text":"Keras Application expects specific kind input preprocessing. ResNet, call application_preprocess_inputs() inputs passing model. application_preprocess_inputs() scale input pixels -1 1.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/application_vgg16.html","id":null,"dir":"Reference","previous_headings":"","what":"Instantiates the VGG16 model. — application_vgg16","title":"Instantiates the VGG16 model. — application_vgg16","text":"Instantiates VGG16 model.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_vgg16.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Instantiates the VGG16 model. — application_vgg16","text":"","code":"application_vgg16(   include_top = TRUE,   weights = \"imagenet\",   input_tensor = NULL,   input_shape = NULL,   pooling = NULL,   classes = 1000L,   classifier_activation = \"softmax\",   name = \"vgg16\" )"},{"path":"https://keras3.posit.co/dev/reference/application_vgg16.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Instantiates the VGG16 model. — application_vgg16","text":"include_top whether include 3 fully-connected layers top network. weights one NULL (random initialization), \"imagenet\" (pre-training ImageNet), path weights file loaded. input_tensor optional Keras tensor (.e. output keras_input()) use image input model. input_shape optional shape tuple, specified include_top FALSE (otherwise input shape (224, 224, 3) (channels_last data format) (3, 224, 224) (\"channels_first\" data format). exactly 3 input channels, width height smaller 32. E.g. (200, 200, 3) one valid value. pooling Optional pooling mode feature extraction include_top FALSE. NULL means output model 4D tensor output last convolutional block. avg means global average pooling applied output last convolutional block, thus output model 2D tensor. max means global max pooling applied. classes optional number classes classify images , specified include_top TRUE, weights argument specified. classifier_activation str callable. activation function use \"top\" layer. Ignored unless include_top=TRUE. Set classifier_activation=NULL return logits \"top\" layer.  loading pretrained weights, classifier_activation can NULL \"softmax\". name name model (string).","code":""},{"path":"https://keras3.posit.co/dev/reference/application_vgg16.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Instantiates the VGG16 model. — application_vgg16","text":"Model instance.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_vgg16.html","id":"reference","dir":"Reference","previous_headings":"","what":"Reference","title":"Instantiates the VGG16 model. — application_vgg16","text":"Deep Convolutional Networks Large-Scale Image Recognition (ICLR 2015) image classification use cases, see page detailed examples. transfer learning use cases, make sure read guide transfer learning & fine-tuning. default input size model 224x224.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_vgg16.html","id":"note","dir":"Reference","previous_headings":"","what":"Note","title":"Instantiates the VGG16 model. — application_vgg16","text":"Keras Application expects specific kind input preprocessing. VGG16, call application_preprocess_inputs() inputs passing model. application_preprocess_inputs() convert input images RGB BGR, zero-center color channel respect ImageNet dataset, without scaling.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/application_vgg19.html","id":null,"dir":"Reference","previous_headings":"","what":"Instantiates the VGG19 model. — application_vgg19","title":"Instantiates the VGG19 model. — application_vgg19","text":"Instantiates VGG19 model.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_vgg19.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Instantiates the VGG19 model. — application_vgg19","text":"","code":"application_vgg19(   include_top = TRUE,   weights = \"imagenet\",   input_tensor = NULL,   input_shape = NULL,   pooling = NULL,   classes = 1000L,   classifier_activation = \"softmax\",   name = \"vgg19\" )"},{"path":"https://keras3.posit.co/dev/reference/application_vgg19.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Instantiates the VGG19 model. — application_vgg19","text":"include_top whether include 3 fully-connected layers top network. weights one NULL (random initialization), \"imagenet\" (pre-training ImageNet), path weights file loaded. input_tensor optional Keras tensor (.e. output keras_input()) use image input model. input_shape optional shape tuple, specified include_top FALSE (otherwise input shape (224, 224, 3) (channels_last data format) (3, 224, 224) (\"channels_first\" data format). exactly 3 input channels, width height smaller 32. E.g. (200, 200, 3) one valid value. pooling Optional pooling mode feature extraction include_top FALSE. NULL means output model 4D tensor output last convolutional block. avg means global average pooling applied output last convolutional block, thus output model 2D tensor. max means global max pooling applied. classes optional number classes classify images , specified include_top TRUE, weights argument specified. classifier_activation str callable. activation function use \"top\" layer. Ignored unless include_top=TRUE. Set classifier_activation=NULL return logits \"top\" layer.  loading pretrained weights, classifier_activation can NULL \"softmax\". name name model (string).","code":""},{"path":"https://keras3.posit.co/dev/reference/application_vgg19.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Instantiates the VGG19 model. — application_vgg19","text":"model instance.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_vgg19.html","id":"reference","dir":"Reference","previous_headings":"","what":"Reference","title":"Instantiates the VGG19 model. — application_vgg19","text":"Deep Convolutional Networks Large-Scale Image Recognition (ICLR 2015) image classification use cases, see page detailed examples. transfer learning use cases, make sure read guide transfer learning & fine-tuning. default input size model 224x224.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_vgg19.html","id":"note","dir":"Reference","previous_headings":"","what":"Note","title":"Instantiates the VGG19 model. — application_vgg19","text":"Keras Application expects specific kind input preprocessing. VGG19, call application_preprocess_inputs() inputs passing model. application_preprocess_inputs() convert input images RGB BGR, zero-center color channel respect ImageNet dataset, without scaling.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/application_xception.html","id":null,"dir":"Reference","previous_headings":"","what":"Instantiates the Xception architecture. — application_xception","title":"Instantiates the Xception architecture. — application_xception","text":"Instantiates Xception architecture.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_xception.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Instantiates the Xception architecture. — application_xception","text":"","code":"application_xception(   include_top = TRUE,   weights = \"imagenet\",   input_tensor = NULL,   input_shape = NULL,   pooling = NULL,   classes = 1000L,   classifier_activation = \"softmax\",   name = \"xception\" )"},{"path":"https://keras3.posit.co/dev/reference/application_xception.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Instantiates the Xception architecture. — application_xception","text":"include_top whether include 3 fully-connected layers top network. weights one NULL (random initialization), \"imagenet\" (pre-training ImageNet), path weights file loaded. input_tensor optional Keras tensor (.e. output keras_input()) use image input model. input_shape optional shape tuple, specified include_top FALSE (otherwise input shape (299, 299, 3). exactly 3 inputs channels, width height smaller 71. E.g. (150, 150, 3) one valid value. pooling Optional pooling mode feature extraction include_top FALSE. NULL means output model 4D tensor output last convolutional block. avg means global average pooling applied output last convolutional block, thus output model 2D tensor. max means global max pooling applied. classes optional number classes classify images , specified include_top TRUE, weights argument specified. classifier_activation str callable. activation function use \"top\" layer. Ignored unless include_top=TRUE. Set classifier_activation=NULL return logits \"top\" layer.  loading pretrained weights, classifier_activation can NULL \"softmax\". name name model (string).","code":""},{"path":"https://keras3.posit.co/dev/reference/application_xception.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Instantiates the Xception architecture. — application_xception","text":"model instance.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_xception.html","id":"reference","dir":"Reference","previous_headings":"","what":"Reference","title":"Instantiates the Xception architecture. — application_xception","text":"Xception: Deep Learning Depthwise Separable Convolutions (CVPR 2017) image classification use cases, see page detailed examples. transfer learning use cases, make sure read guide transfer learning & fine-tuning. default input image size model 299x299.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_xception.html","id":"note","dir":"Reference","previous_headings":"","what":"Note","title":"Instantiates the Xception architecture. — application_xception","text":"Keras Application expects specific kind input preprocessing. Xception, call application_preprocess_inputs() inputs passing model. application_preprocess_inputs() scale input pixels -1 1.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/audio_dataset_from_directory.html","id":null,"dir":"Reference","previous_headings":"","what":"Generates a tf.data.Dataset from audio files in a directory. — audio_dataset_from_directory","title":"Generates a tf.data.Dataset from audio files in a directory. — audio_dataset_from_directory","text":"directory structure :   calling audio_dataset_from_directory(main_directory, labels = 'inferred') return tf.data.Dataset yields batches audio files subdirectories class_a class_b, together labels 0 1 (0 corresponding class_a 1 corresponding class_b). .wav files supported time.","code":"main_directory/ ...class_a/ ......a_audio_1.wav ......a_audio_2.wav ...class_b/ ......b_audio_1.wav ......b_audio_2.wav"},{"path":"https://keras3.posit.co/dev/reference/audio_dataset_from_directory.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Generates a tf.data.Dataset from audio files in a directory. — audio_dataset_from_directory","text":"","code":"audio_dataset_from_directory(   directory,   labels = \"inferred\",   label_mode = \"int\",   class_names = NULL,   batch_size = 32L,   sampling_rate = NULL,   output_sequence_length = NULL,   ragged = FALSE,   shuffle = TRUE,   seed = NULL,   validation_split = NULL,   subset = NULL,   follow_links = FALSE,   verbose = TRUE )"},{"path":"https://keras3.posit.co/dev/reference/audio_dataset_from_directory.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Generates a tf.data.Dataset from audio files in a directory. — audio_dataset_from_directory","text":"directory Directory data located. labels \"inferred\", contain subdirectories, containing audio files class. Otherwise, directory structure ignored. labels Either \"inferred\" (labels generated directory structure), NULL (labels), list/tuple integer labels size number audio files found directory. Labels sorted according alphanumeric order audio file paths (obtained via os.walk(directory) Python). label_mode String describing encoding labels. Options : \"int\": means labels encoded integers (e.g. sparse_categorical_crossentropy loss). \"categorical\" means labels encoded categorical vector (e.g. categorical_crossentropy loss) \"binary\" means labels (can 2) encoded float32 scalars values 0 1 (e.g. binary_crossentropy). NULL (labels). class_names valid \"labels\" \"inferred\". explicit list class names (must match names subdirectories). Used control order classes (otherwise alphanumerical order used). batch_size Size batches data. Default: 32. NULL, data batched (dataset yield individual samples). sampling_rate Audio sampling rate (samples per second). output_sequence_length Maximum length audio sequence. Audio files longer truncated output_sequence_length. set NULL, sequences batch padded length longest sequence batch. ragged Whether return Ragged dataset (sequence length). Defaults FALSE. shuffle Whether shuffle data. Defaults TRUE. set FALSE, sorts data alphanumeric order. seed Optional random seed shuffling transformations. validation_split Optional float 0 1, fraction data reserve validation. subset Subset data return. One \"training\", \"validation\" \"\". used validation_split set. follow_links Whether visits subdirectories pointed symlinks. Defaults FALSE. verbose Whether display number information classes number files found. Defaults TRUE.","code":""},{"path":"https://keras3.posit.co/dev/reference/audio_dataset_from_directory.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Generates a tf.data.Dataset from audio files in a directory. — audio_dataset_from_directory","text":"tf.data.Dataset object. label_mode NULL, yields string tensors shape (batch_size,), containing contents batch audio files. Otherwise, yields tuple (audio, labels), audio shape (batch_size, sequence_length, num_channels) labels follows format described . Rules regarding labels format: label_mode int, labels int32 tensor shape (batch_size,). label_mode binary, labels float32 tensor 1s 0s shape (batch_size, 1). label_mode categorical, labels float32 tensor shape (batch_size, num_classes), representing one-hot encoding class index.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/back-compat.html","id":null,"dir":"Reference","previous_headings":"","what":"Backward compatibility — application_nasnetlarge","title":"Backward compatibility — application_nasnetlarge","text":"Backward compatibility Backward compatibility","code":""},{"path":"https://keras3.posit.co/dev/reference/back-compat.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Backward compatibility — application_nasnetlarge","text":"","code":"application_nasnetlarge(...)  application_nasnetmobile(...)"},{"path":"https://keras3.posit.co/dev/reference/back-compat.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Backward compatibility — application_nasnetlarge","text":"... passed application_nasnet_large() application_nasnet_mobile().","code":""},{"path":"https://keras3.posit.co/dev/reference/bidirectional.html","id":null,"dir":"Reference","previous_headings":"","what":"layer_bidirectional — bidirectional","title":"layer_bidirectional — bidirectional","text":"bidirectional() alias layer_bidirectional(). See ?layer_bidirectional() full documentation.","code":""},{"path":"https://keras3.posit.co/dev/reference/bidirectional.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"layer_bidirectional — bidirectional","text":"","code":"bidirectional(   object,   layer,   merge_mode = \"concat\",   weights = NULL,   backward_layer = NULL,   ... )"},{"path":"https://keras3.posit.co/dev/reference/bidirectional.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"layer_bidirectional — bidirectional","text":"object Object compose layer . tensor, array, sequential model. layer RNN instance, layer_lstm() layer_gru(). also Layer() instance meets following criteria: sequence-processing layer (accepts 3D+ inputs). go_backwards, return_sequences return_state attribute (semantics RNN class). input_spec attribute. Implement serialization via get_config() from_config(). Note recommended way create new RNN layers write custom RNN cell use layer_rnn(), instead subclassing Layer() directly. return_sequences TRUE, output masked timestep zero regardless layer's original zero_output_for_mask value. merge_mode Mode outputs forward backward RNNs combined. One {\"sum\", \"mul\", \"concat\", \"ave\", NULL}. NULL, outputs combined, returned list. Defaults \"concat\". weights see description backward_layer Optional RNN, Layer() instance used handle backwards input processing. backward_layer provided, layer instance passed layer argument used generate backward layer automatically. Note provided backward_layer layer properties matching layer argument, particular values stateful, return_states, return_sequences, etc. addition, backward_layer layer different go_backwards argument values. ValueError raised requirements met. ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/bidirectional.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"layer_bidirectional — bidirectional","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/callback_backup_and_restore.html","id":null,"dir":"Reference","previous_headings":"","what":"Callback to back up and restore the training state. — callback_backup_and_restore","title":"Callback to back up and restore the training state. — callback_backup_and_restore","text":"callback_backup_and_restore() callback intended recover training interruption happened middle fit execution, backing training states temporary checkpoint file, end epoch. backup overwrites previously written checkpoint file, given time one checkpoint file backup/restoring purpose. training restarts completion, training state (includes model weights epoch number) restored recently saved state beginning new fit run. completion fit run, temporary checkpoint file deleted. Note user responsible bring jobs back interruption. callback important backup restore mechanism fault tolerance purpose, model restored previous checkpoint expected one used back . user changes arguments passed compile fit, checkpoint saved fault tolerance can become invalid.","code":""},{"path":"https://keras3.posit.co/dev/reference/callback_backup_and_restore.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Callback to back up and restore the training state. — callback_backup_and_restore","text":"","code":"callback_backup_and_restore(   backup_dir,   save_freq = \"epoch\",   double_checkpoint = FALSE,   delete_checkpoint = TRUE )"},{"path":"https://keras3.posit.co/dev/reference/callback_backup_and_restore.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Callback to back up and restore the training state. — callback_backup_and_restore","text":"backup_dir String, path directory store data needed restore model. directory reused elsewhere store files, e.g. backup_and_restore callback another training run, another callback (e.g. callback_model_checkpoint) training run. save_freq \"epoch\", integer, FALSE. set \"epoch\", callback saves checkpoint end epoch. set integer, callback saves checkpoint every save_freq batches. Set save_freq = FALSE using preemption checkpointing (.e. save_before_preemption = TRUE). double_checkpoint Boolean. enabled, BackupAndRestore callback save 2 last training states (current previous). interruption current state loaded due IO error (e.g. file corrupted) try restore previous one. behaviour consume twice space disk, increase fault tolerance. Defaults FALSE. delete_checkpoint Boolean. backup_and_restore callback works saving checkpoint back training state. delete_checkpoint = TRUE, checkpoint deleted training finished. Use FALSE like keep checkpoint future usage. Defaults TRUE.","code":""},{"path":"https://keras3.posit.co/dev/reference/callback_backup_and_restore.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Callback to back up and restore the training state. — callback_backup_and_restore","text":"Callback instance can passed fit.keras.src.models.model.Model().","code":""},{"path":"https://keras3.posit.co/dev/reference/callback_backup_and_restore.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Callback to back up and restore the training state. — callback_backup_and_restore","text":"","code":"callback_interrupting <- new_callback_class(   \"InterruptingCallback\",   on_epoch_begin = function(epoch, logs = NULL) {     if (epoch == 4) {       stop('Interrupting!')     }   } )  backup_dir <- tempfile() callback <- callback_backup_and_restore(backup_dir = backup_dir) model <- keras_model_sequential() %>%   layer_dense(10) model %>% compile(optimizer = optimizer_sgd(), loss = 'mse')  # ensure model is built (i.e., weights are initialized) for # callback_backup_and_restore() model(op_ones(c(5, 20))) |> invisible()  tryCatch({   model %>% fit(x = op_ones(c(5, 20)),                 y = op_zeros(5),                 epochs = 10, batch_size = 1,                 callbacks = list(callback, callback_interrupting()),                 verbose = 0) }, python.builtin.RuntimeError = function(e) message(\"Interrupted!\")) ## Interrupted! model$history$epoch ## [1] 0 1 2 # model$history %>% keras3:::to_keras_training_history() %>% as.data.frame() %>% print()  history <- model %>% fit(x = op_ones(c(5, 20)),                          y = op_zeros(5),                          epochs = 10, batch_size = 1,                          callbacks = list(callback),                          verbose = 0)  # Only 6 more epochs are run, since first training got interrupted at # zero-indexed epoch 4, second training will continue from 4 to 9. nrow(as.data.frame(history)) ## [1] 10"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/callback_csv_logger.html","id":null,"dir":"Reference","previous_headings":"","what":"Callback that streams epoch results to a CSV file. — callback_csv_logger","title":"Callback that streams epoch results to a CSV file. — callback_csv_logger","text":"Supports values can represented string, including 1D iterables atomic vectors.","code":""},{"path":"https://keras3.posit.co/dev/reference/callback_csv_logger.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Callback that streams epoch results to a CSV file. — callback_csv_logger","text":"","code":"callback_csv_logger(filename, separator = \",\", append = FALSE)"},{"path":"https://keras3.posit.co/dev/reference/callback_csv_logger.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Callback that streams epoch results to a CSV file. — callback_csv_logger","text":"filename Filename CSV file, e.g. 'run/log.csv'. separator String used separate elements CSV file. append Boolean. TRUE: append file exists (useful continuing training). FALSE: overwrite existing file.","code":""},{"path":"https://keras3.posit.co/dev/reference/callback_csv_logger.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Callback that streams epoch results to a CSV file. — callback_csv_logger","text":"Callback instance can passed fit.keras.src.models.model.Model().","code":""},{"path":"https://keras3.posit.co/dev/reference/callback_csv_logger.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Callback that streams epoch results to a CSV file. — callback_csv_logger","text":"","code":"csv_logger <- callback_csv_logger('training.log') model %>% fit(X_train, Y_train, callbacks = list(csv_logger))"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/callback_early_stopping.html","id":null,"dir":"Reference","previous_headings":"","what":"Stop training when a monitored metric has stopped improving. — callback_early_stopping","title":"Stop training when a monitored metric has stopped improving. — callback_early_stopping","text":"Assuming goal training minimize loss. , metric monitored 'loss', mode 'min'. model$fit() training loop check end every epoch whether loss longer decreasing, considering min_delta patience applicable. found longer decreasing, model$stop_training marked TRUE training terminates. quantity monitored needs available logs list. make , pass loss metrics model$compile().","code":""},{"path":"https://keras3.posit.co/dev/reference/callback_early_stopping.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Stop training when a monitored metric has stopped improving. — callback_early_stopping","text":"","code":"callback_early_stopping(   monitor = \"val_loss\",   min_delta = 0L,   patience = 0L,   verbose = 0L,   mode = \"auto\",   baseline = NULL,   restore_best_weights = FALSE,   start_from_epoch = 0L )"},{"path":"https://keras3.posit.co/dev/reference/callback_early_stopping.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Stop training when a monitored metric has stopped improving. — callback_early_stopping","text":"monitor Quantity monitored. Defaults \"val_loss\". min_delta Minimum change monitored quantity qualify improvement, .e. absolute change less min_delta, count improvement. Defaults 0. patience Number epochs improvement training stopped. Defaults 0. verbose Verbosity mode, 0 1. Mode 0 silent, mode 1 displays messages callback takes action. Defaults 0. mode One {\"auto\", \"min\", \"max\"}. min mode, training stop quantity monitored stopped decreasing; \"max\" mode stop quantity monitored stopped increasing; \"auto\" mode, direction automatically inferred name monitored quantity. Defaults \"auto\". baseline Baseline value monitored quantity. NULL, training stop model show improvement baseline. Defaults NULL. restore_best_weights Whether restore model weights epoch best value monitored quantity. FALSE, model weights obtained last step training used. epoch restored regardless performance relative baseline. epoch improves baseline, training run patience epochs restore weights best epoch set. Defaults FALSE. start_from_epoch Number epochs wait starting monitor improvement. allows warm-period improvement expected thus training stopped. Defaults 0.","code":""},{"path":"https://keras3.posit.co/dev/reference/callback_early_stopping.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Stop training when a monitored metric has stopped improving. — callback_early_stopping","text":"Callback instance can passed fit.keras.src.models.model.Model().","code":""},{"path":"https://keras3.posit.co/dev/reference/callback_early_stopping.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Stop training when a monitored metric has stopped improving. — callback_early_stopping","text":"","code":"callback <- callback_early_stopping(monitor = 'loss',                                    patience = 3) # This callback will stop the training when there is no improvement in # the loss for three consecutive epochs. model <- keras_model_sequential() %>%   layer_dense(10) model %>% compile(optimizer = optimizer_sgd(), loss = 'mse') history <- model %>% fit(x = op_ones(c(5, 20)),                          y = op_zeros(5),                          epochs = 10, batch_size = 1,                          callbacks = list(callback),                          verbose = 0) nrow(as.data.frame(history))  # Only 4 epochs are run. ## [1] 10"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/callback_lambda.html","id":null,"dir":"Reference","previous_headings":"","what":"Callback for creating simple, custom callbacks on-the-fly. — callback_lambda","title":"Callback for creating simple, custom callbacks on-the-fly. — callback_lambda","text":"callback constructed anonymous functions called appropriate time (Model.{fit | evaluate | predict}). Note callbacks expects positional arguments, : on_epoch_begin on_epoch_end expect two positional arguments: epoch, logs on_train_begin on_train_end expect one positional argument: logs on_train_batch_begin on_train_batch_end expect two positional arguments: batch, logs See Callback class definition full list functions expected arguments.","code":""},{"path":"https://keras3.posit.co/dev/reference/callback_lambda.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Callback for creating simple, custom callbacks on-the-fly. — callback_lambda","text":"","code":"callback_lambda(   on_epoch_begin = NULL,   on_epoch_end = NULL,   on_train_begin = NULL,   on_train_end = NULL,   on_train_batch_begin = NULL,   on_train_batch_end = NULL,   ... )"},{"path":"https://keras3.posit.co/dev/reference/callback_lambda.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Callback for creating simple, custom callbacks on-the-fly. — callback_lambda","text":"on_epoch_begin called beginning every epoch. on_epoch_end called end every epoch. on_train_begin called beginning model training. on_train_end called end model training. on_train_batch_begin called beginning every train batch. on_train_batch_end called end every train batch. ... function Callback() want override passing function_name = function. example, callback_lambda(.., on_train_end = train_end_fn). custom function needs arguments ones defined Callback().","code":""},{"path":"https://keras3.posit.co/dev/reference/callback_lambda.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Callback for creating simple, custom callbacks on-the-fly. — callback_lambda","text":"Callback instance can passed fit.keras.src.models.model.Model().","code":""},{"path":"https://keras3.posit.co/dev/reference/callback_lambda.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Callback for creating simple, custom callbacks on-the-fly. — callback_lambda","text":"","code":"# Print the batch number at the beginning of every batch. batch_print_callback <- callback_lambda(   on_train_batch_begin = function(batch, logs) {     print(batch)   } )  # Stream the epoch loss to a file in new-line delimited JSON format # (one valid JSON object per line) json_log <- file('loss_log.json', open = 'wt') json_logging_callback <- callback_lambda(   on_epoch_end = function(epoch, logs) {     jsonlite::write_json(       list(epoch = epoch, loss = logs$loss),       json_log,       append = TRUE     )   },   on_train_end = function(logs) {     close(json_log)   } )  # Terminate some processes after having finished model training. processes <- ... cleanup_callback <- callback_lambda(   on_train_end = function(logs) {     for (p in processes) {       if (is_alive(p)) {         terminate(p)       }     }   } )  model %>% fit(   ...,   callbacks = list(     batch_print_callback,     json_logging_callback,     cleanup_callback   ) )"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/callback_learning_rate_scheduler.html","id":null,"dir":"Reference","previous_headings":"","what":"Learning rate scheduler. — callback_learning_rate_scheduler","title":"Learning rate scheduler. — callback_learning_rate_scheduler","text":"beginning every epoch, callback gets updated learning rate value schedule function provided, current epoch current learning rate, applies updated learning rate optimizer.","code":""},{"path":"https://keras3.posit.co/dev/reference/callback_learning_rate_scheduler.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Learning rate scheduler. — callback_learning_rate_scheduler","text":"","code":"callback_learning_rate_scheduler(schedule, verbose = 0L)"},{"path":"https://keras3.posit.co/dev/reference/callback_learning_rate_scheduler.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Learning rate scheduler. — callback_learning_rate_scheduler","text":"schedule function takes epoch index (integer, indexed 0) current learning rate (float) inputs returns new learning rate output (float). verbose Integer. 0: quiet, 1: log update messages.","code":""},{"path":"https://keras3.posit.co/dev/reference/callback_learning_rate_scheduler.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Learning rate scheduler. — callback_learning_rate_scheduler","text":"Callback instance can passed fit.keras.src.models.model.Model().","code":""},{"path":"https://keras3.posit.co/dev/reference/callback_learning_rate_scheduler.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Learning rate scheduler. — callback_learning_rate_scheduler","text":"","code":"# This function keeps the initial learning rate steady for the first ten epochs # and decreases it exponentially after that. scheduler <- function(epoch, lr) {   if (epoch < 10)     return(lr)   else     return(lr * exp(-0.1)) }  model <- keras_model_sequential() |> layer_dense(units = 10) model |> compile(optimizer = optimizer_sgd(), loss = 'mse') model$optimizer$learning_rate |> as.array() |> round(5) ## [1] 0.01 callback <- callback_learning_rate_scheduler(schedule = scheduler) history <- model |> fit(x = array(runif(100), c(5, 20)),                         y = array(0, c(5, 1)),                         epochs = 15, callbacks = list(callback), verbose = 0) model$optimizer$learning_rate |> as.array() |> round(5) ## [1] 0.00607"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/callback_model_checkpoint.html","id":null,"dir":"Reference","previous_headings":"","what":"Callback to save the Keras model or model weights at some frequency. — callback_model_checkpoint","title":"Callback to save the Keras model or model weights at some frequency. — callback_model_checkpoint","text":"callback_model_checkpoint() used conjunction training using model |> fit() save model weights (checkpoint file) interval, model weights can loaded later continue training state saved. options callback provides include: Whether keep model achieved \"best performance\" far, whether save model end every epoch regardless performance. Definition \"best\"; quantity monitor whether maximized minimized. frequency save . Currently, callback supports saving end every epoch, fixed number training batches. Whether weights saved, whole model saved.","code":""},{"path":"https://keras3.posit.co/dev/reference/callback_model_checkpoint.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Callback to save the Keras model or model weights at some frequency. — callback_model_checkpoint","text":"","code":"callback_model_checkpoint(   filepath,   monitor = \"val_loss\",   verbose = 0L,   save_best_only = FALSE,   save_weights_only = FALSE,   mode = \"auto\",   save_freq = \"epoch\",   initial_value_threshold = NULL )"},{"path":"https://keras3.posit.co/dev/reference/callback_model_checkpoint.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Callback to save the Keras model or model weights at some frequency. — callback_model_checkpoint","text":"filepath string, path save model file. filepath can contain named formatting options, filled value epoch keys logs (passed on_epoch_end). filepath name needs end \".weights.h5\" save_weights_only = TRUE end \".keras\" \".h5\" checkpoint saving whole model (default). example: filepath \"{epoch:02d}-{val_loss:.2f}.keras\" \"{epoch:02d}-{val_loss:.2f}.weights.h5\", model checkpoints saved epoch number validation loss filename. directory filepath reused callbacks avoid conflicts. monitor metric name monitor. Typically metrics set model |> compile() method. Note: Prefix name \"val_\" monitor validation metrics. Use \"loss\" \"val_loss\" monitor model's total loss. specify metrics strings, like \"accuracy\", pass string (without \"val_\" prefix). pass Metric objects (created one metric_*()), monitor set metric$name. sure metric names can check contents history$metrics list returned history <- model |> fit() Multi-output models set additional prefixes metric names. verbose Verbosity mode, 0 1. Mode 0 silent, mode 1 displays messages callback takes action. save_best_only save_best_only = TRUE, saves model considered \"best\" latest best model according quantity monitored overwritten. filepath contain formatting options like {epoch} filepath overwritten new better model. save_weights_only TRUE, model's weights saved (model |> save_model_weights(filepath)), else full model saved (model |> save_model(filepath)). mode one {\"auto\", \"min\", \"max\"}. save_best_only = TRUE, decision overwrite current save file made based either maximization minimization monitored quantity. val_acc, \"max\", val_loss \"min\", etc. \"auto\" mode, mode set \"max\" quantities monitored \"acc\" start \"fmeasure\" set \"min\" rest quantities. save_freq \"epoch\" integer. using \"epoch\", callback saves model epoch. using integer, callback saves model end many batches. Model compiled steps_per_execution = N, saving criteria checked every Nth batch. Note saving aligned epochs, monitored metric may potentially less reliable (reflect little 1 batch, since metrics get reset every epoch). Defaults \"epoch\". initial_value_threshold Floating point initial \"best\" value metric monitored. applies save_best_value = TRUE. overwrites model weights already saved performance current model better value.","code":""},{"path":"https://keras3.posit.co/dev/reference/callback_model_checkpoint.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Callback to save the Keras model or model weights at some frequency. — callback_model_checkpoint","text":"Callback instance can passed fit.keras.src.models.model.Model().","code":""},{"path":"https://keras3.posit.co/dev/reference/callback_model_checkpoint.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Callback to save the Keras model or model weights at some frequency. — callback_model_checkpoint","text":"","code":"model <- keras_model_sequential(input_shape = c(10)) |>   layer_dense(1, activation = \"sigmoid\") |>   compile(loss = \"binary_crossentropy\", optimizer = \"adam\",           metrics = c('accuracy'))  EPOCHS <- 10 checkpoint_filepath <- tempfile('checkpoint-model-', fileext = \".keras\") model_checkpoint_callback <- callback_model_checkpoint(   filepath = checkpoint_filepath,   monitor = 'val_accuracy',   mode = 'max',   save_best_only = TRUE )  # Model is saved at the end of every epoch, if it's the best seen so far. model |> fit(x = random_uniform(c(2, 10)), y = op_ones(2, 1),              epochs = EPOCHS, validation_split = .5, verbose = 0,              callbacks = list(model_checkpoint_callback))  # The model (that are considered the best) can be loaded as - load_model(checkpoint_filepath) ## Model: \"sequential\" ## +---------------------------------+------------------------+---------------+ ## | Layer (type)                    | Output Shape           |       Param # | ## +=================================+========================+===============+ ## | dense (Dense)                   | (None, 1)              |            11 | ## +---------------------------------+------------------------+---------------+ ##  Total params: 35 (144.00 B) ##  Trainable params: 11 (44.00 B) ##  Non-trainable params: 0 (0.00 B) ##  Optimizer params: 24 (100.00 B) # Alternatively, one could checkpoint just the model weights as - checkpoint_filepath <- tempfile('checkpoint-', fileext = \".weights.h5\") model_checkpoint_callback <- callback_model_checkpoint(   filepath = checkpoint_filepath,   save_weights_only = TRUE,   monitor = 'val_accuracy',   mode = 'max',   save_best_only = TRUE )  # Model weights are saved at the end of every epoch, if it's the best seen # so far. # same as above model |> fit(x = random_uniform(c(2, 10)), y = op_ones(2, 1),              epochs = EPOCHS, validation_split = .5, verbose = 0,              callbacks = list(model_checkpoint_callback))  # The model weights (that are considered the best) can be loaded model |> load_model_weights(checkpoint_filepath)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/callback_reduce_lr_on_plateau.html","id":null,"dir":"Reference","previous_headings":"","what":"Reduce learning rate when a metric has stopped improving. — callback_reduce_lr_on_plateau","title":"Reduce learning rate when a metric has stopped improving. — callback_reduce_lr_on_plateau","text":"Models often benefit reducing learning rate factor 2-10 learning stagnates. callback monitors quantity improvement seen 'patience' number epochs, learning rate reduced.","code":""},{"path":"https://keras3.posit.co/dev/reference/callback_reduce_lr_on_plateau.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Reduce learning rate when a metric has stopped improving. — callback_reduce_lr_on_plateau","text":"","code":"callback_reduce_lr_on_plateau(   monitor = \"val_loss\",   factor = 0.1,   patience = 10L,   verbose = 0L,   mode = \"auto\",   min_delta = 1e-04,   cooldown = 0L,   min_lr = 0,   ... )"},{"path":"https://keras3.posit.co/dev/reference/callback_reduce_lr_on_plateau.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Reduce learning rate when a metric has stopped improving. — callback_reduce_lr_on_plateau","text":"monitor String. Quantity monitored. factor Float. Factor learning rate reduced. new_lr = lr * factor. patience Integer. Number epochs improvement learning rate reduced. verbose Integer. 0: quiet, 1: update messages. mode String. One {'auto', 'min', 'max'}. 'min' mode, learning rate reduced quantity monitored stopped decreasing; 'max' mode reduced quantity monitored stopped increasing; 'auto' mode, direction automatically inferred name monitored quantity. min_delta Float. Threshold measuring new optimum, focus significant changes. cooldown Integer. Number epochs wait resuming normal operation learning rate reduced. min_lr Float. Lower bound learning rate. ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/callback_reduce_lr_on_plateau.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Reduce learning rate when a metric has stopped improving. — callback_reduce_lr_on_plateau","text":"Callback instance can passed fit.keras.src.models.model.Model().","code":""},{"path":"https://keras3.posit.co/dev/reference/callback_reduce_lr_on_plateau.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Reduce learning rate when a metric has stopped improving. — callback_reduce_lr_on_plateau","text":"","code":"reduce_lr <- callback_reduce_lr_on_plateau(monitor = 'val_loss', factor = 0.2,                                            patience = 5, min_lr = 0.001) model %>% fit(x_train, y_train, callbacks = list(reduce_lr))"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/callback_remote_monitor.html","id":null,"dir":"Reference","previous_headings":"","what":"Callback used to stream events to a server. — callback_remote_monitor","title":"Callback used to stream events to a server. — callback_remote_monitor","text":"Requires requests library. Events sent root + '/publish/epoch/end/' default. Calls HTTP POST, data argument JSON-encoded named list event data. send_as_json = TRUE, content type request \"application/json\". Otherwise serialized JSON sent within form.","code":""},{"path":"https://keras3.posit.co/dev/reference/callback_remote_monitor.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Callback used to stream events to a server. — callback_remote_monitor","text":"","code":"callback_remote_monitor(   root = \"http://localhost:9000\",   path = \"/publish/epoch/end/\",   field = \"data\",   headers = NULL,   send_as_json = FALSE )"},{"path":"https://keras3.posit.co/dev/reference/callback_remote_monitor.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Callback used to stream events to a server. — callback_remote_monitor","text":"root String; root url target server. path String; path relative root events sent. field String; JSON field data stored. field used payload sent within form (.e. send_as_json = FALSE). headers Named list; optional custom HTTP headers. send_as_json Boolean; whether request sent \"application/json\".","code":""},{"path":"https://keras3.posit.co/dev/reference/callback_remote_monitor.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Callback used to stream events to a server. — callback_remote_monitor","text":"Callback instance can passed fit.keras.src.models.model.Model().","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/callback_swap_ema_weights.html","id":null,"dir":"Reference","previous_headings":"","what":"Swaps model weights and EMA weights before and after evaluation. — callback_swap_ema_weights","title":"Swaps model weights and EMA weights before and after evaluation. — callback_swap_ema_weights","text":"callbacks replaces model's weight values values optimizer's EMA weights (exponential moving average past model weights values, implementing \"Polyak averaging\") model evaluation, restores previous weights evaluation. SwapEMAWeights callback used conjunction optimizer sets use_ema = TRUE. Note weights swapped -place order save memory. behavior undefined modify EMA weights model weights callbacks.","code":""},{"path":"https://keras3.posit.co/dev/reference/callback_swap_ema_weights.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Swaps model weights and EMA weights before and after evaluation. — callback_swap_ema_weights","text":"","code":"callback_swap_ema_weights(swap_on_epoch = FALSE)"},{"path":"https://keras3.posit.co/dev/reference/callback_swap_ema_weights.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Swaps model weights and EMA weights before and after evaluation. — callback_swap_ema_weights","text":"swap_on_epoch Whether perform swapping on_epoch_begin() on_epoch_end(). useful want use EMA weights callbacks callback_model_checkpoint(). Defaults FALSE.","code":""},{"path":"https://keras3.posit.co/dev/reference/callback_swap_ema_weights.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Swaps model weights and EMA weights before and after evaluation. — callback_swap_ema_weights","text":"Callback instance can passed fit.keras.src.models.model.Model().","code":""},{"path":"https://keras3.posit.co/dev/reference/callback_swap_ema_weights.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Swaps model weights and EMA weights before and after evaluation. — callback_swap_ema_weights","text":"","code":"# Remember to set `use_ema=TRUE` in the optimizer optimizer <- optimizer_sgd(use_ema = TRUE) model |> compile(optimizer = optimizer, loss = ..., metrics = ...)  # Metrics will be computed with EMA weights model |> fit(X_train, Y_train,              callbacks = c(callback_swap_ema_weights()))  # If you want to save model checkpoint with EMA weights, you can set # `swap_on_epoch=TRUE` and place ModelCheckpoint after SwapEMAWeights. model |> fit(   X_train, Y_train,   callbacks = c(     callback_swap_ema_weights(swap_on_epoch = TRUE),     callback_model_checkpoint(...)   ) )"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/callback_tensorboard.html","id":null,"dir":"Reference","previous_headings":"","what":"Enable visualizations for TensorBoard. — callback_tensorboard","title":"Enable visualizations for TensorBoard. — callback_tensorboard","text":"TensorBoard visualization tool provided TensorFlow. TensorFlow installation required use callback. callback logs events TensorBoard, including: Metrics summary plots Training graph visualization Weight histograms Sampled profiling used model |> evaluate() regular validation addition epoch summaries, summary records evaluation metrics vs model$optimizer$iterations written. metric names prepended evaluation, model$optimizer$iterations step visualized TensorBoard. installed TensorFlow pip reticulate::py_install(), able launch TensorBoard command line:   R tensorflow::tensorboard(). can find information TensorBoard .","code":"tensorboard --logdir=path_to_your_logs"},{"path":"https://keras3.posit.co/dev/reference/callback_tensorboard.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Enable visualizations for TensorBoard. — callback_tensorboard","text":"","code":"callback_tensorboard(   log_dir = \"logs\",   histogram_freq = 0L,   write_graph = TRUE,   write_images = FALSE,   write_steps_per_second = FALSE,   update_freq = \"epoch\",   profile_batch = 0L,   embeddings_freq = 0L,   embeddings_metadata = NULL )"},{"path":"https://keras3.posit.co/dev/reference/callback_tensorboard.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Enable visualizations for TensorBoard. — callback_tensorboard","text":"log_dir path directory save log files parsed TensorBoard. e.g., log_dir = file.path(working_dir, 'logs'). directory reused callbacks. histogram_freq frequency (epochs) compute weight histograms layers model. set 0, histograms computed. Validation data (split) must specified histogram visualizations. write_graph (supported time) Whether visualize graph TensorBoard. Note log file can become quite large write_graph set TRUE. write_images whether write model weights visualize image TensorBoard. write_steps_per_second whether log training steps per second TensorBoard. supports epoch batch frequency logging. update_freq \"batch\" \"epoch\" integer. using \"epoch\", writes losses metrics TensorBoard every epoch. using integer, say 1000, metrics losses (including custom ones added Model.compile) logged TensorBoard every 1000 batches. \"batch\" synonym 1, meaning written every batch. Note however writing frequently TensorBoard can slow training, especially used distribution strategies incur additional synchronization overhead. Batch-level summary writing also available via train_step override. Please see TensorBoard Scalars tutorial details. profile_batch Profile batch(es) sample compute characteristics. profile_batch must non-negative integer tuple integers. pair positive integers signify range batches profile. default, profiling disabled. embeddings_freq frequency (epochs) embedding layers visualized. set 0, embeddings visualized. embeddings_metadata Named list maps embedding layer names filename file save metadata embedding layer. case metadata file used embedding layers, single filename can passed.","code":""},{"path":"https://keras3.posit.co/dev/reference/callback_tensorboard.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Enable visualizations for TensorBoard. — callback_tensorboard","text":"Callback instance can passed fit.keras.src.models.model.Model().","code":""},{"path":"https://keras3.posit.co/dev/reference/callback_tensorboard.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Enable visualizations for TensorBoard. — callback_tensorboard","text":"Custom batch-level summaries subclassed Model:   Custom batch-level summaries Functional API Model:   Profiling:","code":"tensorboard_callback <- callback_tensorboard(log_dir = \"./logs\") model %>% fit(x_train, y_train, epochs = 2, callbacks = list(tensorboard_callback)) # Then run the tensorboard command to view the visualizations. MyModel <- new_model_class(\"MyModel\",   initialize = function() {     self$dense <- layer_dense(units = 10)   },   call = function(x) {     outputs <- x |> self$dense()     tf$summary$histogram('outputs', outputs)     outputs   } )  model <- MyModel() model |> compile(optimizer = 'sgd', loss = 'mse')  # Make sure to set `update_freq = N` to log a batch-level summary every N # batches. In addition to any `tf$summary` contained in `model$call()`, # metrics added in `model |>compile` will be logged every N batches. tb_callback <- callback_tensorboard(log_dir = './logs', update_freq = 1) model |> fit(x_train, y_train, callbacks = list(tb_callback)) my_summary <- function(x) {   tf$summary$histogram('x', x)   x }  inputs <- layer_input(10) outputs <- inputs |>   layer_dense(10) |>   layer_lambda(my_summary)  model <- keras_model(inputs, outputs) model |> compile(optimizer = 'sgd', loss = 'mse')  # Make sure to set `update_freq = N` to log a batch-level summary every N # batches. In addition to any `tf.summary` contained in `Model.call`, # metrics added in `Model.compile` will be logged every N batches. tb_callback <- callback_tensorboard(log_dir = './logs', update_freq = 1) model |> fit(x_train, y_train, callbacks = list(tb_callback)) # Profile a single batch, e.g. the 5th batch. tensorboard_callback <- callback_tensorboard(   log_dir = './logs', profile_batch = 5) model |> fit(x_train, y_train, epochs = 2,              callbacks = list(tensorboard_callback))  # Profile a range of batches, e.g. from 10 to 20. tensorboard_callback <- callback_tensorboard(   log_dir = './logs', profile_batch = c(10, 20)) model |> fit(x_train, y_train, epochs = 2,              callbacks = list(tensorboard_callback))"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/callback_terminate_on_nan.html","id":null,"dir":"Reference","previous_headings":"","what":"Callback that terminates training when a NaN loss is encountered. — callback_terminate_on_nan","title":"Callback that terminates training when a NaN loss is encountered. — callback_terminate_on_nan","text":"Callback terminates training NaN loss encountered.","code":""},{"path":"https://keras3.posit.co/dev/reference/callback_terminate_on_nan.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Callback that terminates training when a NaN loss is encountered. — callback_terminate_on_nan","text":"","code":"callback_terminate_on_nan()"},{"path":"https://keras3.posit.co/dev/reference/callback_terminate_on_nan.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Callback that terminates training when a NaN loss is encountered. — callback_terminate_on_nan","text":"Callback instance can passed fit.keras.src.models.model.Model().","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/clear_session.html","id":null,"dir":"Reference","previous_headings":"","what":"Resets all state generated by Keras. — clear_session","title":"Resets all state generated by Keras. — clear_session","text":"Keras manages global state, uses implement Functional model-building API uniquify autogenerated layer names. creating many models loop, global state consume increasing amount memory time, may want clear . Calling clear_session() releases global state: helps avoid clutter old models layers, especially memory limited. Example 1: calling clear_session() creating models loop   Example 2: resetting layer name generation counter","code":"for (i in 1:100) {   # Without `clear_session()`, each iteration of this loop will   # slightly increase the size of the global state managed by Keras   model <- keras_model_sequential()   for (j in 1:10) {     model <- model |> layer_dense(units = 10)   } }  for (i in 1:100) {   # With `clear_session()` called at the beginning,   # Keras starts with a blank state at each iteration   # and memory consumption is constant over time.   clear_session()   model <- keras_model_sequential()   for (j in 1:10) {     model <- model |> layer_dense(units = 10)   } } layers <- lapply(1:10, \\(i) layer_dense(units = 10))  new_layer <- layer_dense(units = 10) print(new_layer$name) ## [1] \"dense_10\" clear_session() new_layer <- layer_dense(units = 10) print(new_layer$name) ## [1] \"dense\""},{"path":"https://keras3.posit.co/dev/reference/clear_session.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Resets all state generated by Keras. — clear_session","text":"","code":"clear_session(free_memory = TRUE)"},{"path":"https://keras3.posit.co/dev/reference/clear_session.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Resets all state generated by Keras. — clear_session","text":"free_memory Whether call Python garbage collection. usually good practice call make sure memory used deleted objects immediately freed. However, may take seconds execute, using clear_session() short loop, may want skip .","code":""},{"path":"https://keras3.posit.co/dev/reference/clear_session.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Resets all state generated by Keras. — clear_session","text":"NULL, invisibly, called side effects.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/clone_model.html","id":null,"dir":"Reference","previous_headings":"","what":"Clone a Functional or Sequential Model instance. — clone_model","title":"Clone a Functional or Sequential Model instance. — clone_model","text":"Model cloning similar calling model new inputs, except creates new layers (thus new weights) instead sharing weights existing layers. Note clone_model() preserve uniqueness shared objects within model (e.g. single variable attached two distinct layers restored two separate variables).","code":""},{"path":"https://keras3.posit.co/dev/reference/clone_model.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Clone a Functional or Sequential Model instance. — clone_model","text":"","code":"clone_model(   model,   input_tensors = NULL,   clone_function = NULL,   call_function = NULL,   recursive = FALSE,   ... )"},{"path":"https://keras3.posit.co/dev/reference/clone_model.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Clone a Functional or Sequential Model instance. — clone_model","text":"model Instance Model (Functional model Sequential model). input_tensors Optional list input tensors build model upon. provided, new keras_input() objects created. clone_function Callable signature function(layer) used clone layer target model (except Input instances). takes argument layer instance cloned, returns corresponding layer instance used model copy. unspecified, callable defaults following serialization/deserialization function: function(layer) layer$`__class__`$from_config(layer$get_config()). passing custom callable, can customize copy model, e.g. wrapping certain layers interest (might want replace LSTM instances equivalent Bidirectional(LSTM(...)) instances, example). Defaults NULL. call_function Callable signature function(layer, ...) used call cloned layer set inputs. takes layer instance, call arguments, returns call outputs. unspecified, callable defaults regular call() method: function(layer, ...) .call(layer, list(...)). passing custom callable, can insert new layers given layer. recursive Note, argument can used Functional models. Boolean. Whether recursively clone Sequential Functional models encountered original Sequential/Functional model. FALSE, inner models cloned calling clone_function(). TRUE, inner models cloned calling clone_model() clone_function, call_function, recursive arguments. Note case, call_function propagated Sequential model (since applicable Sequential models). ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/clone_model.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Clone a Functional or Sequential Model instance. — clone_model","text":"instance Model reproducing behavior original model, top new inputs tensors, using newly instantiated weights. cloned model may behave differently original model custom clone_function call_function modifies layer layer call.","code":""},{"path":"https://keras3.posit.co/dev/reference/clone_model.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Clone a Functional or Sequential Model instance. — clone_model","text":"Using clone_function make model deterministic setting random seed everywhere:   Using call_function add Dropout layer Dense layer (without recreating new layers):     Note subclassed models cloned default, since internal layer structure known. achieve equivalent functionality clone_model case subclassed model, simply make sure model class implements get_config() (optionally from_config()), call:   case subclassed model, using custom clone_function.","code":"# Create a test Sequential model. model <- keras_model_sequential(input_shape = c(728)) |>   layer_dense(32, activation = 'relu') |>   layer_dense(1, activation = 'sigmoid')  # Create a copy of the test model (with freshly initialized weights). new_model <- clone_model(model) clone_function <- function(layer) {   config <- layer$get_config()   if (\"seed\" %in% names(config))     config$seed <- 1337L   layer$`__class__`$from_config(config) }  new_model <- clone_model(model, clone_function = clone_function) call_function <- function(layer, ...) {   out <- layer(...)   if (inherits(layer, keras$layers$Dense))     out <- out |> layer_dropout(0.5)   out }  inputs <- keras_input(c(728)) outputs <- inputs |>   layer_dense(32, activation = 'relu') |>   layer_dense(1, activation = 'sigmoid') model <- keras_model(inputs, outputs)  new_model <- clone_model(   model,   clone_function = function(x) x, # Reuse the same layers.   call_function = call_function, ) new_model ## Model: \"functional_4\" ## +-----------------------------------+--------------------------+---------------+ ## | Layer (type)                      | Output Shape             |       Param # | ## +===================================+==========================+===============+ ## | keras_tensor_8 (InputLayer)       | (None, 728)              |             0 | ## +-----------------------------------+--------------------------+---------------+ ## | dense_2 (Dense)                   | (None, 32)               |        23,328 | ## +-----------------------------------+--------------------------+---------------+ ## | dropout (Dropout)                 | (None, 32)               |             0 | ## +-----------------------------------+--------------------------+---------------+ ## | dense_3 (Dense)                   | (None, 1)                |            33 | ## +-----------------------------------+--------------------------+---------------+ ## | dropout_1 (Dropout)               | (None, 1)                |             0 | ## +-----------------------------------+--------------------------+---------------+ ##  Total params: 23,361 (91.25 KB) ##  Trainable params: 23,361 (91.25 KB) ##  Non-trainable params: 0 (0.00 B) new_model <- model$`__class__`$from_config(model$get_config())"},{"path":"https://keras3.posit.co/dev/reference/compile.keras.src.models.model.Model.html","id":null,"dir":"Reference","previous_headings":"","what":"Configure a model for training. — compile.keras.src.models.model.Model","title":"Configure a model for training. — compile.keras.src.models.model.Model","text":"Configure model training.","code":""},{"path":"https://keras3.posit.co/dev/reference/compile.keras.src.models.model.Model.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Configure a model for training. — compile.keras.src.models.model.Model","text":"","code":"# S3 method for class 'keras.src.models.model.Model' compile(   object,   optimizer = \"rmsprop\",   loss = NULL,   metrics = NULL,   ...,   loss_weights = NULL,   weighted_metrics = NULL,   run_eagerly = FALSE,   steps_per_execution = 1L,   jit_compile = \"auto\",   auto_scale_loss = TRUE )"},{"path":"https://keras3.posit.co/dev/reference/compile.keras.src.models.model.Model.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Configure a model for training. — compile.keras.src.models.model.Model","text":"object Keras model object optimizer String (name optimizer) optimizer instance. See optimizer_* family. loss Loss function. May : string (name builtin loss function), custom function, Loss instance (returned loss_* family functions). loss function callable signature loss = fn(y_true, y_pred), y_true ground truth values, y_pred model's predictions. y_true shape (batch_size, d1, .. dN) (except case sparse loss functions sparse categorical crossentropy expects integer arrays shape (batch_size, d1, .. dN-1)). y_pred shape (batch_size, d1, .. dN). loss function return float tensor. metrics List metrics evaluated model training testing. can : string (name built-function), function, optionally \"name\" attribute Metric() instance. See metric_* family functions. Typically use metrics = c('accuracy'). function callable signature result = fn(y_true, y_pred). specify different metrics different outputs multi-output model, also pass named list, metrics = list(= 'accuracy', b = c('accuracy', 'mse')). can also pass list specify metric list metrics output, metrics = list(c('accuracy'), c('accuracy', 'mse')) metrics = list('accuracy', c('accuracy', 'mse')). pass strings 'accuracy' 'acc', convert one metric_binary_accuracy(), metric_categorical_accuracy(), metric_sparse_categorical_accuracy() based shapes targets model output. similar conversion done strings \"crossentropy\" \"ce\" well. metrics passed evaluated without sample weighting; like sample weighting apply, can specify metrics via weighted_metrics argument instead. providing anonymous R function, can customize printed name training assigning attr(<fn>, \"name\") <- \"my_custom_metric_name\", calling custom_metric(\"my_custom_metric_name\", <fn>) ... Additional arguments passed compile() model method. loss_weights Optional list (named unnamed) specifying scalar coefficients (R numerics) weight loss contributions different model outputs. loss value minimized model weighted sum individual losses, weighted loss_weights coefficients.  unnamed list, expected 1:1 mapping model's outputs. named list, expected map output names (strings) scalar coefficients. weighted_metrics List metrics evaluated weighted sample_weight class_weight training testing. run_eagerly Bool. TRUE, model's forward pass never compiled. recommended leave FALSE training (best performance), set TRUE debugging. steps_per_execution Int. number batches run single compiled function call. Running multiple batches inside single compiled function call can greatly improve performance TPUs small models large R/Python overhead. , one full epoch run execution. number larger size epoch passed, execution truncated size epoch. Note steps_per_execution set N, Callback$on_batch_begin Callback$on_batch_end methods called every N batches (.e. /compiled function execution). supported PyTorch backend. jit_compile Bool \"auto\". Whether use XLA compilation compiling model. jax tensorflow backends, jit_compile=\"auto\" enables XLA compilation model supports , disabled otherwise. torch backend, \"auto\" default eager execution jit_compile=True run torch.compile \"inductor\" backend. auto_scale_loss Bool. TRUE model dtype policy \"mixed_float16\", passed optimizer automatically wrapped LossScaleOptimizer, dynamically scale loss prevent underflow.","code":""},{"path":"https://keras3.posit.co/dev/reference/compile.keras.src.models.model.Model.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Configure a model for training. — compile.keras.src.models.model.Model","text":"called primarily side effect modifying object -place. first argument object also returned, invisibly, enable usage pipe.","code":""},{"path":"https://keras3.posit.co/dev/reference/compile.keras.src.models.model.Model.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Configure a model for training. — compile.keras.src.models.model.Model","text":"","code":"model |> compile(   optimizer = optimizer_adam(learning_rate = 1e-3),   loss = loss_binary_crossentropy(),   metrics = c(metric_binary_accuracy(),               metric_false_negatives()) )"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/config_backend.html","id":null,"dir":"Reference","previous_headings":"","what":"Publicly accessible method for determining the current backend. — config_backend","title":"Publicly accessible method for determining the current backend. — config_backend","text":"Publicly accessible method determining current backend.","code":""},{"path":"https://keras3.posit.co/dev/reference/config_backend.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Publicly accessible method for determining the current backend. — config_backend","text":"","code":"config_backend()"},{"path":"https://keras3.posit.co/dev/reference/config_backend.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Publicly accessible method for determining the current backend. — config_backend","text":"String, name backend Keras currently using. One \"tensorflow\", \"torch\", \"jax\".","code":""},{"path":"https://keras3.posit.co/dev/reference/config_backend.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Publicly accessible method for determining the current backend. — config_backend","text":"","code":"config_backend() ## [1] \"tensorflow\""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/config_disable_flash_attention.html","id":null,"dir":"Reference","previous_headings":"","what":"Disable flash attention. — config_disable_flash_attention","title":"Disable flash attention. — config_disable_flash_attention","text":"Flash attention offers performance optimization attention layers, making especially useful large language models (LLMs) benefit faster memory-efficient attention computations. disabled, supported layers like MultiHeadAttention use flash attention faster computations.","code":""},{"path":"https://keras3.posit.co/dev/reference/config_disable_flash_attention.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Disable flash attention. — config_disable_flash_attention","text":"","code":"config_disable_flash_attention()"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/config_disable_interactive_logging.html","id":null,"dir":"Reference","previous_headings":"","what":"Turn off interactive logging. — config_disable_interactive_logging","title":"Turn off interactive logging. — config_disable_interactive_logging","text":"interactive logging disabled, Keras sends logs absl.logging. best option using Keras non-interactive way, running training inference job server.","code":""},{"path":"https://keras3.posit.co/dev/reference/config_disable_interactive_logging.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Turn off interactive logging. — config_disable_interactive_logging","text":"","code":"config_disable_interactive_logging()"},{"path":"https://keras3.posit.co/dev/reference/config_disable_interactive_logging.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Turn off interactive logging. — config_disable_interactive_logging","text":"return value, called side effects.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/config_disable_traceback_filtering.html","id":null,"dir":"Reference","previous_headings":"","what":"Turn off traceback filtering. — config_disable_traceback_filtering","title":"Turn off traceback filtering. — config_disable_traceback_filtering","text":"Raw Keras tracebacks (also known stack traces) involve many internal frames, can challenging read , actionable end users. default, Keras filters internal frames exceptions raises, keep traceback short, readable, focused actionable (code). See also config_enable_traceback_filtering() config_is_traceback_filtering_enabled(). previously disabled traceback filtering via config_disable_traceback_filtering(), can re-enable via config_enable_traceback_filtering().","code":""},{"path":"https://keras3.posit.co/dev/reference/config_disable_traceback_filtering.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Turn off traceback filtering. — config_disable_traceback_filtering","text":"","code":"config_disable_traceback_filtering()"},{"path":"https://keras3.posit.co/dev/reference/config_disable_traceback_filtering.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Turn off traceback filtering. — config_disable_traceback_filtering","text":"return value, called side effects.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/config_dtype_policy.html","id":null,"dir":"Reference","previous_headings":"","what":"Returns the current default dtype policy object. — config_dtype_policy","title":"Returns the current default dtype policy object. — config_dtype_policy","text":"Returns current default dtype policy object.","code":""},{"path":"https://keras3.posit.co/dev/reference/config_dtype_policy.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Returns the current default dtype policy object. — config_dtype_policy","text":"","code":"config_dtype_policy()"},{"path":"https://keras3.posit.co/dev/reference/config_dtype_policy.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Returns the current default dtype policy object. — config_dtype_policy","text":"DTypePolicy object.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/config_enable_flash_attention.html","id":null,"dir":"Reference","previous_headings":"","what":"Enable flash attention. — config_enable_flash_attention","title":"Enable flash attention. — config_enable_flash_attention","text":"Flash attention offers performance optimization attention layers, making especially useful large language models (LLMs) benefit faster memory-efficient attention computations. enabled, supported layers like layer_multi_head_attention attempt use flash attention faster computations. default, feature enabled. Note enabling flash attention guarantee always used. Typically, inputs must float16 bfloat16 dtype, input layout requirements may vary depending backend.","code":""},{"path":"https://keras3.posit.co/dev/reference/config_enable_flash_attention.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Enable flash attention. — config_enable_flash_attention","text":"","code":"config_enable_flash_attention()"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/config_enable_interactive_logging.html","id":null,"dir":"Reference","previous_headings":"","what":"Turn on interactive logging. — config_enable_interactive_logging","title":"Turn on interactive logging. — config_enable_interactive_logging","text":"interactive logging enabled, Keras displays logs via stdout. provides best experience using Keras interactive environment shell notebook.","code":""},{"path":"https://keras3.posit.co/dev/reference/config_enable_interactive_logging.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Turn on interactive logging. — config_enable_interactive_logging","text":"","code":"config_enable_interactive_logging()"},{"path":"https://keras3.posit.co/dev/reference/config_enable_interactive_logging.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Turn on interactive logging. — config_enable_interactive_logging","text":"return value, called side effects.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/config_enable_traceback_filtering.html","id":null,"dir":"Reference","previous_headings":"","what":"Turn on traceback filtering. — config_enable_traceback_filtering","title":"Turn on traceback filtering. — config_enable_traceback_filtering","text":"Raw Keras tracebacks (also known stack traces) involve many internal frames, can challenging read , actionable end users. default, Keras filters internal frames exceptions raises, keep traceback short, readable, focused actionable (code). See also config_disable_traceback_filtering() config_is_traceback_filtering_enabled(). previously disabled traceback filtering via config_disable_traceback_filtering(), can re-enable via config_enable_traceback_filtering().","code":""},{"path":"https://keras3.posit.co/dev/reference/config_enable_traceback_filtering.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Turn on traceback filtering. — config_enable_traceback_filtering","text":"","code":"config_enable_traceback_filtering()"},{"path":"https://keras3.posit.co/dev/reference/config_enable_traceback_filtering.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Turn on traceback filtering. — config_enable_traceback_filtering","text":"return value, called side effects.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/config_enable_unsafe_deserialization.html","id":null,"dir":"Reference","previous_headings":"","what":"Disables safe mode globally, allowing deserialization of lambdas. — config_enable_unsafe_deserialization","title":"Disables safe mode globally, allowing deserialization of lambdas. — config_enable_unsafe_deserialization","text":"Disables safe mode globally, allowing deserialization lambdas.","code":""},{"path":"https://keras3.posit.co/dev/reference/config_enable_unsafe_deserialization.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Disables safe mode globally, allowing deserialization of lambdas. — config_enable_unsafe_deserialization","text":"","code":"config_enable_unsafe_deserialization()"},{"path":"https://keras3.posit.co/dev/reference/config_enable_unsafe_deserialization.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Disables safe mode globally, allowing deserialization of lambdas. — config_enable_unsafe_deserialization","text":"return value, called side effects.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/config_epsilon.html","id":null,"dir":"Reference","previous_headings":"","what":"Return the value of the fuzz factor used in numeric expressions. — config_epsilon","title":"Return the value of the fuzz factor used in numeric expressions. — config_epsilon","text":"Return value fuzz factor used numeric expressions.","code":""},{"path":"https://keras3.posit.co/dev/reference/config_epsilon.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Return the value of the fuzz factor used in numeric expressions. — config_epsilon","text":"","code":"config_epsilon()"},{"path":"https://keras3.posit.co/dev/reference/config_epsilon.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Return the value of the fuzz factor used in numeric expressions. — config_epsilon","text":"float.","code":""},{"path":"https://keras3.posit.co/dev/reference/config_epsilon.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Return the value of the fuzz factor used in numeric expressions. — config_epsilon","text":"","code":"config_epsilon() ## [1] 1e-07"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/config_floatx.html","id":null,"dir":"Reference","previous_headings":"","what":"Return the default float type, as a string. — config_floatx","title":"Return the default float type, as a string. — config_floatx","text":"E.g. 'bfloat16' 'float16', 'float32', 'float64'.","code":""},{"path":"https://keras3.posit.co/dev/reference/config_floatx.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Return the default float type, as a string. — config_floatx","text":"","code":"config_floatx()"},{"path":"https://keras3.posit.co/dev/reference/config_floatx.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Return the default float type, as a string. — config_floatx","text":"String, current default float type.","code":""},{"path":"https://keras3.posit.co/dev/reference/config_floatx.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Return the default float type, as a string. — config_floatx","text":"","code":"keras3::config_floatx() ## [1] \"float32\""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/config_image_data_format.html","id":null,"dir":"Reference","previous_headings":"","what":"Return the default image data format convention. — config_image_data_format","title":"Return the default image data format convention. — config_image_data_format","text":"Return default image data format convention.","code":""},{"path":"https://keras3.posit.co/dev/reference/config_image_data_format.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Return the default image data format convention. — config_image_data_format","text":"","code":"config_image_data_format()"},{"path":"https://keras3.posit.co/dev/reference/config_image_data_format.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Return the default image data format convention. — config_image_data_format","text":"string, either 'channels_first' 'channels_last'.","code":""},{"path":"https://keras3.posit.co/dev/reference/config_image_data_format.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Return the default image data format convention. — config_image_data_format","text":"","code":"config_image_data_format() ## [1] \"channels_last\""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/config_is_flash_attention_enabled.html","id":null,"dir":"Reference","previous_headings":"","what":"Checks whether flash attention is globally enabled in Keras. — config_is_flash_attention_enabled","title":"Checks whether flash attention is globally enabled in Keras. — config_is_flash_attention_enabled","text":"Flash attention performance-optimized method computing attention large models, transformers, allowing faster memory-efficient operations. function checks global Keras configuration determine flash attention enabled compatible layers (e.g., MultiHeadAttention). Note enabling flash attention guarantee always used. Typically, inputs must float16 bfloat16 dtype, input layout requirements may vary depending backend.","code":""},{"path":"https://keras3.posit.co/dev/reference/config_is_flash_attention_enabled.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Checks whether flash attention is globally enabled in Keras. — config_is_flash_attention_enabled","text":"","code":"config_is_flash_attention_enabled()"},{"path":"https://keras3.posit.co/dev/reference/config_is_flash_attention_enabled.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Checks whether flash attention is globally enabled in Keras. — config_is_flash_attention_enabled","text":"FALSE disabled; otherwise, indicates enabled.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/config_is_interactive_logging_enabled.html","id":null,"dir":"Reference","previous_headings":"","what":"Check if interactive logging is enabled. — config_is_interactive_logging_enabled","title":"Check if interactive logging is enabled. — config_is_interactive_logging_enabled","text":"switch writing logs stdout absl.logging, may use config_enable_interactive_logging() config_disable_interactive_logging().","code":""},{"path":"https://keras3.posit.co/dev/reference/config_is_interactive_logging_enabled.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Check if interactive logging is enabled. — config_is_interactive_logging_enabled","text":"","code":"config_is_interactive_logging_enabled()"},{"path":"https://keras3.posit.co/dev/reference/config_is_interactive_logging_enabled.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Check if interactive logging is enabled. — config_is_interactive_logging_enabled","text":"Boolean, TRUE interactive logging enabled, FALSE otherwise.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/config_is_traceback_filtering_enabled.html","id":null,"dir":"Reference","previous_headings":"","what":"Check if traceback filtering is enabled. — config_is_traceback_filtering_enabled","title":"Check if traceback filtering is enabled. — config_is_traceback_filtering_enabled","text":"Raw Keras tracebacks (also known stack traces) involve many internal frames, can challenging read , actionable end users. default, Keras filters internal frames exceptions raises, keep traceback short, readable, focused actionable (code). See also config_enable_traceback_filtering() config_disable_traceback_filtering(). previously disabled traceback filtering via config_disable_traceback_filtering(), can re-enable via config_enable_traceback_filtering().","code":""},{"path":"https://keras3.posit.co/dev/reference/config_is_traceback_filtering_enabled.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Check if traceback filtering is enabled. — config_is_traceback_filtering_enabled","text":"","code":"config_is_traceback_filtering_enabled()"},{"path":"https://keras3.posit.co/dev/reference/config_is_traceback_filtering_enabled.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Check if traceback filtering is enabled. — config_is_traceback_filtering_enabled","text":"Boolean, TRUE traceback filtering enabled, FALSE otherwise.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/config_set_backend.html","id":null,"dir":"Reference","previous_headings":"","what":"Reload the backend (and the Keras package). — config_set_backend","title":"Reload the backend (and the Keras package). — config_set_backend","text":"Reload backend (Keras package).","code":""},{"path":"https://keras3.posit.co/dev/reference/config_set_backend.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Reload the backend (and the Keras package). — config_set_backend","text":"","code":"config_set_backend(backend)"},{"path":"https://keras3.posit.co/dev/reference/config_set_backend.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Reload the backend (and the Keras package). — config_set_backend","text":"backend String","code":""},{"path":"https://keras3.posit.co/dev/reference/config_set_backend.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Reload the backend (and the Keras package). — config_set_backend","text":"Nothing, function called side effect.","code":""},{"path":"https://keras3.posit.co/dev/reference/config_set_backend.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Reload the backend (and the Keras package). — config_set_backend","text":"","code":"config_set_backend(\"jax\")"},{"path":"https://keras3.posit.co/dev/reference/config_set_backend.html","id":"warning","dir":"Reference","previous_headings":"","what":"WARNING","title":"Reload the backend (and the Keras package). — config_set_backend","text":"Using function dangerous done carefully. Changing backend convert type already-instantiated objects. Thus, layers / tensors / etc. already created longer usable without errors. strongly recommended keep around Keras-originated objects instances created calling config_set_backend(). includes function class instance uses Keras functionality. code needs re-executed calling config_set_backend().","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/config_set_dtype_policy.html","id":null,"dir":"Reference","previous_headings":"","what":"Sets the default dtype policy globally. — config_set_dtype_policy","title":"Sets the default dtype policy globally. — config_set_dtype_policy","text":"Sets default dtype policy globally.","code":""},{"path":"https://keras3.posit.co/dev/reference/config_set_dtype_policy.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Sets the default dtype policy globally. — config_set_dtype_policy","text":"","code":"config_set_dtype_policy(policy)"},{"path":"https://keras3.posit.co/dev/reference/config_set_dtype_policy.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Sets the default dtype policy globally. — config_set_dtype_policy","text":"policy string DTypePolicy object.","code":""},{"path":"https://keras3.posit.co/dev/reference/config_set_dtype_policy.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Sets the default dtype policy globally. — config_set_dtype_policy","text":"return value, called side effects.","code":""},{"path":"https://keras3.posit.co/dev/reference/config_set_dtype_policy.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Sets the default dtype policy globally. — config_set_dtype_policy","text":"","code":"config_set_dtype_policy(\"mixed_float16\")"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/config_set_epsilon.html","id":null,"dir":"Reference","previous_headings":"","what":"Set the value of the fuzz factor used in numeric expressions. — config_set_epsilon","title":"Set the value of the fuzz factor used in numeric expressions. — config_set_epsilon","text":"Set value fuzz factor used numeric expressions.","code":""},{"path":"https://keras3.posit.co/dev/reference/config_set_epsilon.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Set the value of the fuzz factor used in numeric expressions. — config_set_epsilon","text":"","code":"config_set_epsilon(value)"},{"path":"https://keras3.posit.co/dev/reference/config_set_epsilon.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Set the value of the fuzz factor used in numeric expressions. — config_set_epsilon","text":"value float. New value epsilon.","code":""},{"path":"https://keras3.posit.co/dev/reference/config_set_epsilon.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Set the value of the fuzz factor used in numeric expressions. — config_set_epsilon","text":"return value, called side effects.","code":""},{"path":"https://keras3.posit.co/dev/reference/config_set_epsilon.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Set the value of the fuzz factor used in numeric expressions. — config_set_epsilon","text":"","code":"config_epsilon() ## [1] 1e-07 config_set_epsilon(1e-5) config_epsilon() ## [1] 1e-05 # Set it back to the default value. config_set_epsilon(1e-7)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/config_set_floatx.html","id":null,"dir":"Reference","previous_headings":"","what":"Set the default float dtype. — config_set_floatx","title":"Set the default float dtype. — config_set_floatx","text":"Set default float dtype.","code":""},{"path":"https://keras3.posit.co/dev/reference/config_set_floatx.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Set the default float dtype. — config_set_floatx","text":"","code":"config_set_floatx(value)"},{"path":"https://keras3.posit.co/dev/reference/config_set_floatx.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Set the default float dtype. — config_set_floatx","text":"value String; 'bfloat16', 'float16', 'float32', 'float64'.","code":""},{"path":"https://keras3.posit.co/dev/reference/config_set_floatx.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Set the default float dtype. — config_set_floatx","text":"return value, called side effects.","code":""},{"path":"https://keras3.posit.co/dev/reference/config_set_floatx.html","id":"note","dir":"Reference","previous_headings":"","what":"Note","title":"Set the default float dtype. — config_set_floatx","text":"recommended set \"float16\" training, likely cause numeric stability issues. Instead, mixed precision, leverages mix float16 float32. can configured calling keras3::keras$mixed_precision$set_dtype_policy('mixed_float16').","code":""},{"path":"https://keras3.posit.co/dev/reference/config_set_floatx.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Set the default float dtype. — config_set_floatx","text":"","code":"config_floatx() ## [1] \"float32\" config_set_floatx('float64') config_floatx() ## [1] \"float64\" # Set it back to float32 config_set_floatx('float32')"},{"path":"https://keras3.posit.co/dev/reference/config_set_floatx.html","id":"raises","dir":"Reference","previous_headings":"","what":"Raises","title":"Set the default float dtype. — config_set_floatx","text":"ValueError: case invalid value.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/config_set_image_data_format.html","id":null,"dir":"Reference","previous_headings":"","what":"Set the value of the image data format convention. — config_set_image_data_format","title":"Set the value of the image data format convention. — config_set_image_data_format","text":"Set value image data format convention.","code":""},{"path":"https://keras3.posit.co/dev/reference/config_set_image_data_format.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Set the value of the image data format convention. — config_set_image_data_format","text":"","code":"config_set_image_data_format(data_format)"},{"path":"https://keras3.posit.co/dev/reference/config_set_image_data_format.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Set the value of the image data format convention. — config_set_image_data_format","text":"data_format string. 'channels_first' 'channels_last'.","code":""},{"path":"https://keras3.posit.co/dev/reference/config_set_image_data_format.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Set the value of the image data format convention. — config_set_image_data_format","text":"return value, called side effects.","code":""},{"path":"https://keras3.posit.co/dev/reference/config_set_image_data_format.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Set the value of the image data format convention. — config_set_image_data_format","text":"","code":"config_image_data_format() ## [1] \"channels_last\" # 'channels_last' keras3::config_set_image_data_format('channels_first') config_image_data_format() ## [1] \"channels_first\" # Set it back to `'channels_last'` keras3::config_set_image_data_format('channels_last')"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/constraint_maxnorm.html","id":null,"dir":"Reference","previous_headings":"","what":"MaxNorm weight constraint. — constraint_maxnorm","title":"MaxNorm weight constraint. — constraint_maxnorm","text":"Constrains weights incident hidden unit norm less equal desired value.","code":""},{"path":"https://keras3.posit.co/dev/reference/constraint_maxnorm.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"MaxNorm weight constraint. — constraint_maxnorm","text":"","code":"constraint_maxnorm(max_value = 2L, axis = 1L)"},{"path":"https://keras3.posit.co/dev/reference/constraint_maxnorm.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"MaxNorm weight constraint. — constraint_maxnorm","text":"max_value maximum norm value incoming weights. axis integer, axis along calculate weight norms. instance, Dense layer weight matrix shape (input_dim, output_dim), set axis 0 constrain weight vector length (input_dim,). Conv2D layer data_format = \"channels_last\", weight tensor shape (rows, cols, input_depth, output_depth), set axis [0, 1, 2] constrain weights filter tensor size (rows, cols, input_depth).","code":""},{"path":"https://keras3.posit.co/dev/reference/constraint_maxnorm.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"MaxNorm weight constraint. — constraint_maxnorm","text":"Constraint instance, callable can passed layer constructors used directly calling tensors.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/constraint_minmaxnorm.html","id":null,"dir":"Reference","previous_headings":"","what":"MinMaxNorm weight constraint. — constraint_minmaxnorm","title":"MinMaxNorm weight constraint. — constraint_minmaxnorm","text":"Constrains weights incident hidden unit norm lower bound upper bound.","code":""},{"path":"https://keras3.posit.co/dev/reference/constraint_minmaxnorm.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"MinMaxNorm weight constraint. — constraint_minmaxnorm","text":"","code":"constraint_minmaxnorm(min_value = 0, max_value = 1, rate = 1, axis = 1L)"},{"path":"https://keras3.posit.co/dev/reference/constraint_minmaxnorm.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"MinMaxNorm weight constraint. — constraint_minmaxnorm","text":"min_value minimum norm incoming weights. max_value maximum norm incoming weights. rate rate enforcing constraint: weights rescaled yield op_clip? (1 - rate) * norm + rate * op_clip(norm, min_value, max_value). Effectively, means rate = 1.0 stands strict enforcement constraint, rate<1.0 means weights rescaled step slowly move towards value inside desired interval. axis integer, axis along calculate weight norms. instance, Dense layer weight matrix shape (input_dim, output_dim), set axis 0 constrain weight vector length (input_dim,). Conv2D layer data_format = \"channels_last\", weight tensor shape (rows, cols, input_depth, output_depth), set axis [0, 1, 2] constrain weights filter tensor size (rows, cols, input_depth).","code":""},{"path":"https://keras3.posit.co/dev/reference/constraint_minmaxnorm.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"MinMaxNorm weight constraint. — constraint_minmaxnorm","text":"Constraint instance, callable can passed layer constructors used directly calling tensors.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/constraint_nonneg.html","id":null,"dir":"Reference","previous_headings":"","what":"Constrains the weights to be non-negative. — constraint_nonneg","title":"Constrains the weights to be non-negative. — constraint_nonneg","text":"Constrains weights non-negative.","code":""},{"path":"https://keras3.posit.co/dev/reference/constraint_nonneg.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Constrains the weights to be non-negative. — constraint_nonneg","text":"","code":"constraint_nonneg()"},{"path":"https://keras3.posit.co/dev/reference/constraint_nonneg.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Constrains the weights to be non-negative. — constraint_nonneg","text":"Constraint instance, callable can passed layer constructors used directly calling tensors.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/constraint_unitnorm.html","id":null,"dir":"Reference","previous_headings":"","what":"Constrains the weights incident to each hidden unit to have unit norm. — constraint_unitnorm","title":"Constrains the weights incident to each hidden unit to have unit norm. — constraint_unitnorm","text":"Constrains weights incident hidden unit unit norm.","code":""},{"path":"https://keras3.posit.co/dev/reference/constraint_unitnorm.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Constrains the weights incident to each hidden unit to have unit norm. — constraint_unitnorm","text":"","code":"constraint_unitnorm(axis = 1L)"},{"path":"https://keras3.posit.co/dev/reference/constraint_unitnorm.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Constrains the weights incident to each hidden unit to have unit norm. — constraint_unitnorm","text":"axis integer, axis along calculate weight norms. instance, Dense layer weight matrix shape (input_dim, output_dim), set axis 0 constrain weight vector length (input_dim,). Conv2D layer data_format = \"channels_last\", weight tensor shape (rows, cols, input_depth, output_depth), set axis [0, 1, 2] constrain weights filter tensor size (rows, cols, input_depth).","code":""},{"path":"https://keras3.posit.co/dev/reference/constraint_unitnorm.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Constrains the weights incident to each hidden unit to have unit norm. — constraint_unitnorm","text":"Constraint instance, callable can passed layer constructors used directly calling tensors.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/count_params.html","id":null,"dir":"Reference","previous_headings":"","what":"Count the total number of scalars composing the weights. — count_params","title":"Count the total number of scalars composing the weights. — count_params","text":"Count total number scalars composing weights.","code":""},{"path":"https://keras3.posit.co/dev/reference/count_params.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Count the total number of scalars composing the weights. — count_params","text":"","code":"count_params(object)"},{"path":"https://keras3.posit.co/dev/reference/count_params.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Count the total number of scalars composing the weights. — count_params","text":"object Layer model object","code":""},{"path":"https://keras3.posit.co/dev/reference/count_params.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Count the total number of scalars composing the weights. — count_params","text":"integer count","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/custom_metric.html","id":null,"dir":"Reference","previous_headings":"","what":"Custom metric function — custom_metric","title":"Custom metric function — custom_metric","text":"Custom metric function","code":""},{"path":"https://keras3.posit.co/dev/reference/custom_metric.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Custom metric function — custom_metric","text":"","code":"custom_metric(name, metric_fn)"},{"path":"https://keras3.posit.co/dev/reference/custom_metric.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Custom metric function — custom_metric","text":"name name used show training progress output metric_fn R function signature function(y_true, y_pred) accepts tensors.","code":""},{"path":"https://keras3.posit.co/dev/reference/custom_metric.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Custom metric function — custom_metric","text":"callable function __name__ attribute.","code":""},{"path":"https://keras3.posit.co/dev/reference/custom_metric.html","id":"details","dir":"Reference","previous_headings":"","what":"Details","title":"Custom metric function — custom_metric","text":"can provide arbitrary R function custom metric. Note y_true y_pred parameters tensors, computations use op_* tensor functions. Use custom_metric() function define custom metric. Note name ('mean_pred') provided custom metric function: name used within training progress output. want save load model custom metrics, also call register_keras_serializable(), specify metric call load_model(). example: load_model(\"my_model.keras\", c('mean_pred' = metric_mean_pred)). Alternatively, can wrap code call with_custom_object_scope() allow refer metric name just like built keras metrics. Alternative ways supplying custom metrics: custom_metric(): Arbitrary R function. metric_mean_wrapper(): Wrap arbitrary R function Metric instance. Create custom Metric() subclass.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/dataset_boston_housing.html","id":null,"dir":"Reference","previous_headings":"","what":"Boston housing price regression dataset — dataset_boston_housing","title":"Boston housing price regression dataset — dataset_boston_housing","text":"Dataset taken StatLib library maintained Carnegie Mellon University.","code":""},{"path":"https://keras3.posit.co/dev/reference/dataset_boston_housing.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Boston housing price regression dataset — dataset_boston_housing","text":"","code":"dataset_boston_housing(   path = \"boston_housing.npz\",   test_split = 0.2,   seed = 113L )"},{"path":"https://keras3.posit.co/dev/reference/dataset_boston_housing.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Boston housing price regression dataset — dataset_boston_housing","text":"path Path cache dataset locally (relative ~/.keras/datasets). test_split fraction data reserve test set. seed Random seed shuffling data computing test split.","code":""},{"path":"https://keras3.posit.co/dev/reference/dataset_boston_housing.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Boston housing price regression dataset — dataset_boston_housing","text":"Lists training test data: train$x, train$y, test$x, test$y. Samples contain 13 attributes houses different locations around Boston suburbs late 1970s. Targets median values houses location (k$).","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/dataset_california_housing.html","id":null,"dir":"Reference","previous_headings":"","what":"Loads the California Housing dataset. — dataset_california_housing","title":"Loads the California Housing dataset. — dataset_california_housing","text":"dataset obtained StatLib repository. continuous regression dataset 20,640 samples 8 features . target variable scalar: median house value California districts, dollars. 8 input features following: MedInc: median income block group HouseAge: median house age block group AveRooms: average number rooms per household AveBedrms: average number bedrooms per household Population: block group population AveOccup: average number household members Latitude: block group latitude Longitude: block group longitude dataset derived 1990 U.S. census, using one row per census block group. block group smallest geographical unit U.S. Census Bureau publishes sample data (block group typically population 600 3,000 people). household group people residing within home. Since average number rooms bedrooms dataset provided per household, columns may take surprisingly large values block groups households many empty houses, vacation resorts.","code":""},{"path":"https://keras3.posit.co/dev/reference/dataset_california_housing.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Loads the California Housing dataset. — dataset_california_housing","text":"","code":"dataset_california_housing(   version = \"large\",   path = \"california_housing.npz\",   test_split = 0.2,   seed = 113L )"},{"path":"https://keras3.posit.co/dev/reference/dataset_california_housing.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Loads the California Housing dataset. — dataset_california_housing","text":"version \"small\" \"large\". small version contains 600 samples, large version contains 20,640 samples. purpose small version serve approximate replacement deprecated boston_housing dataset. path path cache dataset locally (relative Sys.getenv(\"KERAS_HOME\")). test_split fraction data reserve test set. seed Random seed shuffling data computing test split.","code":""},{"path":"https://keras3.posit.co/dev/reference/dataset_california_housing.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Loads the California Housing dataset. — dataset_california_housing","text":"Nested list arrays: (x_train, y_train), (x_test, y_test). x_train, x_test: arrays shape (num_samples, 8) containing either training samples (x_train), test samples (y_train). y_train, y_test: arrays shape (num_samples) containing target scalars. targets float scalars typically 25,000 500,000 represent home prices dollars.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/dataset_cifar10.html","id":null,"dir":"Reference","previous_headings":"","what":"CIFAR10 small image classification — dataset_cifar10","title":"CIFAR10 small image classification — dataset_cifar10","text":"Dataset 50,000 32x32 color training images, labeled 10 categories, 10,000 test images.","code":""},{"path":"https://keras3.posit.co/dev/reference/dataset_cifar10.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"CIFAR10 small image classification — dataset_cifar10","text":"","code":"dataset_cifar10()"},{"path":"https://keras3.posit.co/dev/reference/dataset_cifar10.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"CIFAR10 small image classification — dataset_cifar10","text":"Lists training test data: train$x, train$y, test$x, test$y. x data array RGB image data shape (num_samples, 3, 32, 32). y data array category labels (integers range 0-9) shape (num_samples).","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/dataset_cifar100.html","id":null,"dir":"Reference","previous_headings":"","what":"CIFAR100 small image classification — dataset_cifar100","title":"CIFAR100 small image classification — dataset_cifar100","text":"Dataset 50,000 32x32 color training images, labeled 100 categories, 10,000 test images.","code":""},{"path":"https://keras3.posit.co/dev/reference/dataset_cifar100.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"CIFAR100 small image classification — dataset_cifar100","text":"","code":"dataset_cifar100(label_mode = c(\"fine\", \"coarse\"))"},{"path":"https://keras3.posit.co/dev/reference/dataset_cifar100.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"CIFAR100 small image classification — dataset_cifar100","text":"label_mode one \"fine\", \"coarse\".","code":""},{"path":"https://keras3.posit.co/dev/reference/dataset_cifar100.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"CIFAR100 small image classification — dataset_cifar100","text":"Lists training test data: train$x, train$y, test$x, test$y. x data array RGB image data shape (num_samples, 3, 32, 32). y data array category labels shape (num_samples).","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/dataset_fashion_mnist.html","id":null,"dir":"Reference","previous_headings":"","what":"Fashion-MNIST database of fashion articles — dataset_fashion_mnist","title":"Fashion-MNIST database of fashion articles — dataset_fashion_mnist","text":"Dataset 60,000 28x28 grayscale images 10 fashion article classes, along test set 10,000 images. dataset can used drop-replacement MNIST. class labels encoded integers 0-9 correspond T-shirt/top, Trouser, Pullover, Dress, Coat, Sandal, Shirt,","code":""},{"path":"https://keras3.posit.co/dev/reference/dataset_fashion_mnist.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Fashion-MNIST database of fashion articles — dataset_fashion_mnist","text":"","code":"dataset_fashion_mnist()"},{"path":"https://keras3.posit.co/dev/reference/dataset_fashion_mnist.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Fashion-MNIST database of fashion articles — dataset_fashion_mnist","text":"Lists training test data: train$x, train$y, test$x, test$y, x array grayscale image data shape (num_samples, 28, 28) y array article labels (integers range 0-9) shape (num_samples).","code":""},{"path":"https://keras3.posit.co/dev/reference/dataset_fashion_mnist.html","id":"details","dir":"Reference","previous_headings":"","what":"Details","title":"Fashion-MNIST database of fashion articles — dataset_fashion_mnist","text":"Dataset 60,000 28x28 grayscale images 10 fashion categories, along test set 10,000 images. dataset can used drop-replacement MNIST. class labels : 0 - T-shirt/top 1 - Trouser 2 - Pullover 3 - Dress 4 - Coat 5 - Sandal 6 - Shirt 7 - Sneaker 8 - Bag 9 - Ankle boot","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/dataset_imdb.html","id":null,"dir":"Reference","previous_headings":"","what":"IMDB Movie reviews sentiment classification — dataset_imdb","title":"IMDB Movie reviews sentiment classification — dataset_imdb","text":"Dataset 25,000 movies reviews IMDB, labeled sentiment (positive/negative). Reviews preprocessed, review encoded sequence word indexes (integers). convenience, words indexed overall frequency dataset, instance integer \"3\" encodes 3rd frequent word data. allows quick filtering operations : \"consider top 10,000 common words, eliminate top 20 common words\".","code":""},{"path":"https://keras3.posit.co/dev/reference/dataset_imdb.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"IMDB Movie reviews sentiment classification — dataset_imdb","text":"","code":"dataset_imdb(   path = \"imdb.npz\",   num_words = NULL,   skip_top = 0L,   maxlen = NULL,   seed = 113L,   start_char = 1L,   oov_char = 2L,   index_from = 3L )  dataset_imdb_word_index(path = \"imdb_word_index.json\")"},{"path":"https://keras3.posit.co/dev/reference/dataset_imdb.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"IMDB Movie reviews sentiment classification — dataset_imdb","text":"path cache data (relative ~/.keras/dataset). num_words Max number words include. Words ranked often occur (training set) frequent words kept skip_top Skip top N frequently occuring words (may informative). maxlen sequences longer filtered . seed random seed sample shuffling. start_char start sequence marked character. Set 1 0 usually padding character. oov_char Words cut num_words skip_top limit replaced character. index_from Index actual words index higher.","code":""},{"path":"https://keras3.posit.co/dev/reference/dataset_imdb.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"IMDB Movie reviews sentiment classification — dataset_imdb","text":"Lists training test data: train$x, train$y, test$x, test$y. x data includes integer sequences. num_words argument specific, maximum possible index value num_words-1. maxlen argument specified, largest possible sequence length maxlen. y data includes set integer labels (0 1). dataset_imdb_word_index() function returns list names words values integer.","code":""},{"path":"https://keras3.posit.co/dev/reference/dataset_imdb.html","id":"details","dir":"Reference","previous_headings":"","what":"Details","title":"IMDB Movie reviews sentiment classification — dataset_imdb","text":"convention, \"0\" stand specific word, instead used encode unknown word.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/dataset_mnist.html","id":null,"dir":"Reference","previous_headings":"","what":"MNIST database of handwritten digits — dataset_mnist","title":"MNIST database of handwritten digits — dataset_mnist","text":"Dataset 60,000 28x28 grayscale images 10 digits, along test set 10,000 images.","code":""},{"path":"https://keras3.posit.co/dev/reference/dataset_mnist.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"MNIST database of handwritten digits — dataset_mnist","text":"","code":"dataset_mnist(path = \"mnist.npz\")"},{"path":"https://keras3.posit.co/dev/reference/dataset_mnist.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"MNIST database of handwritten digits — dataset_mnist","text":"path Path cache dataset locally (relative ~/.keras/datasets).","code":""},{"path":"https://keras3.posit.co/dev/reference/dataset_mnist.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"MNIST database of handwritten digits — dataset_mnist","text":"Lists training test data: train$x, train$y, test$x, test$y, x array grayscale image data shape (num_samples, 28, 28) y array digit labels (integers range 0-9) shape (num_samples).","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/dataset_reuters.html","id":null,"dir":"Reference","previous_headings":"","what":"Reuters newswire topics classification — dataset_reuters","title":"Reuters newswire topics classification — dataset_reuters","text":"Dataset 11,228 newswires Reuters, labeled 46 topics. dataset_imdb() , wire encoded sequence word indexes (conventions).","code":""},{"path":"https://keras3.posit.co/dev/reference/dataset_reuters.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Reuters newswire topics classification — dataset_reuters","text":"","code":"dataset_reuters(   path = \"reuters.npz\",   num_words = NULL,   skip_top = 0L,   maxlen = NULL,   test_split = 0.2,   seed = 113L,   start_char = 1L,   oov_char = 2L,   index_from = 3L )  dataset_reuters_word_index(path = \"reuters_word_index.pkl\")"},{"path":"https://keras3.posit.co/dev/reference/dataset_reuters.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Reuters newswire topics classification — dataset_reuters","text":"path cache data (relative ~/.keras/dataset). num_words Max number words include. Words ranked often occur (training set) frequent words kept skip_top Skip top N frequently occuring words (may informative). maxlen Truncate sequences length. test_split Fraction dataset used test data. seed Random seed sample shuffling. start_char start sequence marked character. Set 1 0 usually padding character. oov_char words cut num_words skip_top limit replaced character. index_from index actual words index higher.","code":""},{"path":"https://keras3.posit.co/dev/reference/dataset_reuters.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Reuters newswire topics classification — dataset_reuters","text":"Lists training test data: train$x, train$y, test$x, test$y format dataset_imdb(). dataset_reuters_word_index() function returns list names words values integer. e.g. word_index[[\"giraffe\"]] might return 1234.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/deserialize_keras_object.html","id":null,"dir":"Reference","previous_headings":"","what":"Retrieve the object by deserializing the config dict. — deserialize_keras_object","title":"Retrieve the object by deserializing the config dict. — deserialize_keras_object","text":"config dict Python dictionary consists set key-value pairs, represents Keras object, Optimizer, Layer, Metrics, etc. saving loading library uses following keys record information Keras object: class_name: String. name class, exactly defined source code, \"LossesContainer\". config: Named List. Library-defined user-defined key-value pairs store configuration object, obtained object$get_config(). module: String. path python module. Built-Keras classes expect prefix keras. registered_name: String. key class registered via register_keras_serializable(package, name) API. key format '{package}>{name}', package name arguments passed register_keras_serializable(). name provided, uses class name. registered_name successfully resolves class (registered), class_name config values config dict used. registered_name used non-built-classes. example, following config list represents built-Adam optimizer relevant config:     class exported Keras namespace, library tracks module class_name. example:   following config represents user-customized MeanSquaredError loss:","code":"config <- list(   class_name = \"Adam\",   config = list(     amsgrad = FALSE,     beta_1 = 0.8999999761581421,     beta_2 = 0.9990000128746033,     epsilon = 1e-07,     learning_rate = 0.0010000000474974513,     name = \"Adam\"   ),   module = \"keras.optimizers\",   registered_name = NULL ) # Returns an `Adam` instance identical to the original one. deserialize_keras_object(config) ## <keras.src.optimizers.adam.Adam object at 0x0> config <- list(   class_name = \"MetricsList\",   config =  list(     ...   ),   module = \"keras.trainers.compile_utils\",   registered_name = \"MetricsList\" )  # Returns a `MetricsList` instance identical to the original one. deserialize_keras_object(config) # define a custom object loss_modified_mse <- Loss(   \"ModifiedMeanSquaredError\",   inherit = loss_mean_squared_error)  # register the custom object register_keras_serializable(loss_modified_mse)  # confirm object is registered get_custom_objects() ## $`keras3>ModifiedMeanSquaredError` ## <class '<r-namespace:keras3>.ModifiedMeanSquaredError'> ##  signature: ( ##    reduction='sum_over_batch_size', ##    name='mean_squared_error', ##    dtype=None ## ) get_registered_name(loss_modified_mse) ## [1] \"keras3>ModifiedMeanSquaredError\" # now custom object instances can be serialized full_config <- serialize_keras_object(loss_modified_mse())  # the `config` arguments will be passed to loss_modified_mse() str(full_config) ## List of 4 ##  $ module         : chr \"<r-namespace:keras3>\" ##  $ class_name     : chr \"ModifiedMeanSquaredError\" ##  $ config         :List of 2 ##   ..$ name     : chr \"mean_squared_error\" ##   ..$ reduction: chr \"sum_over_batch_size\" ##  $ registered_name: chr \"keras3>ModifiedMeanSquaredError\" # and custom object instances can be deserialized deserialize_keras_object(full_config) ## <LossFunctionWrapper(<function mean_squared_error at 0x0>, kwargs={})> ##  signature: (y_true, y_pred, sample_weight=None) # Returns the `ModifiedMeanSquaredError` object"},{"path":"https://keras3.posit.co/dev/reference/deserialize_keras_object.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Retrieve the object by deserializing the config dict. — deserialize_keras_object","text":"","code":"deserialize_keras_object(config, custom_objects = NULL, safe_mode = TRUE, ...)"},{"path":"https://keras3.posit.co/dev/reference/deserialize_keras_object.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Retrieve the object by deserializing the config dict. — deserialize_keras_object","text":"config Named list describing object. custom_objects Named list containing mapping custom object names corresponding classes functions. safe_mode Boolean, whether disallow unsafe lambda deserialization. safe_mode=FALSE, loading object potential trigger arbitrary code execution. argument applicable Keras v3 model format. Defaults TRUE. ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/deserialize_keras_object.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Retrieve the object by deserializing the config dict. — deserialize_keras_object","text":"object described config dictionary.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/evaluate.keras.src.models.model.Model.html","id":null,"dir":"Reference","previous_headings":"","what":"Evaluate a Keras Model — evaluate.keras.src.models.model.Model","title":"Evaluate a Keras Model — evaluate.keras.src.models.model.Model","text":"functions returns loss value metrics values model test mode. Computation done batches (see batch_size arg.)","code":""},{"path":"https://keras3.posit.co/dev/reference/evaluate.keras.src.models.model.Model.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Evaluate a Keras Model — evaluate.keras.src.models.model.Model","text":"","code":"# S3 method for class 'keras.src.models.model.Model' evaluate(   object,   x = NULL,   y = NULL,   ...,   batch_size = NULL,   verbose = getOption(\"keras.verbose\", default = \"auto\"),   sample_weight = NULL,   steps = NULL,   callbacks = NULL )"},{"path":"https://keras3.posit.co/dev/reference/evaluate.keras.src.models.model.Model.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Evaluate a Keras Model — evaluate.keras.src.models.model.Model","text":"object Keras model object x Input data. can : R array (array-like), list arrays (case model multiple inputs). backend-native tensor, list tensors (case model multiple inputs). named list mapping input names corresponding array/tensors, model named inputs. tf.data.Dataset. return tuple either (inputs, targets) (inputs, targets, sample_weights). generator returning (inputs, targets) (inputs, targets, sample_weights). y Target data. Like input data x, either R array(s) backend-native tensor(s). x tf.data.Dataset generator function, y specified (since targets obtained iterator/dataset). ... forward/backward compatability. batch_size Integer NULL. Number samples per batch computation. unspecified, batch_size default 32. specify batch_size data form tf dataset generator (since generate batches). verbose \"auto\", 0, 1, 2. Verbosity mode. 0 = silent, 1 = progress bar, 2 = single line. \"auto\" becomes 1 cases, 2 knitr render running distributed training server. Note progress bar particularly useful logged file, verbose=2 recommended running interactively (e.g. production environment). Defaults \"auto\". sample_weight Optional array tensor weights training samples, used weighting loss function (training ). can either pass flat (1D) array tensor length input samples (1:1 mapping weights samples), case temporal data, can pass 2D array tensor shape (samples, sequence_length) apply different weight every timestep every sample. argument supported x tf.data.Dataset, Python generator function. Instead, provide sample_weights third element x. Note sample weighting apply metrics specified via metrics argument compile(). apply sample weighting metrics, can specify via weighted_metrics compile() instead. steps Integer NULL. Total number steps (batches samples) declaring evaluation round finished. Ignored default value NULL. x tf.data.Dataset steps NULL, evaluation run dataset exhausted. case infinitely repeating dataset, run indefinitely. callbacks List Callback instances. List callbacks apply evaluation.","code":""},{"path":"https://keras3.posit.co/dev/reference/evaluate.keras.src.models.model.Model.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Evaluate a Keras Model — evaluate.keras.src.models.model.Model","text":"Scalar test loss (model single output metrics) list scalars (model multiple outputs /metrics). attribute model$metrics_names give display labels scalar outputs.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/export_savedmodel.keras.src.models.model.Model.html","id":null,"dir":"Reference","previous_headings":"","what":"Export the model as an artifact for inference. — export_savedmodel.keras.src.models.model.Model","title":"Export the model as an artifact for inference. — export_savedmodel.keras.src.models.model.Model","text":"(e.g. via TF-Serving). Note: can currently used TensorFlow JAX backends. method lets export model lightweight SavedModel artifact contains model's forward pass (call() method) can served via e.g. TF-Serving. forward pass registered name serve() (see example ). original code model (including custom layers may used) longer necessary reload artifact – entirely standalone. Note: feature currently supported TensorFlow, JAX Torch backends. Note: aware exported artifact may contain information local file system using format=\"onnx\", verbose=TRUE Torch backend.","code":""},{"path":"https://keras3.posit.co/dev/reference/export_savedmodel.keras.src.models.model.Model.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Export the model as an artifact for inference. — export_savedmodel.keras.src.models.model.Model","text":"","code":"# S3 method for class 'keras.src.models.model.Model' export_savedmodel(   object,   export_dir_base,   ...,   format = \"tf_saved_model\",   verbose = NULL,   input_signature = NULL )"},{"path":"https://keras3.posit.co/dev/reference/export_savedmodel.keras.src.models.model.Model.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Export the model as an artifact for inference. — export_savedmodel.keras.src.models.model.Model","text":"object keras model. export_dir_base string, file path save artifact. ... Additional keyword arguments: Specific JAX backend format=\"tf_saved_model\": is_static: Optional bool. Indicates whether fn static. Set FALSE fn involves state updates (e.g., RNG seeds counters). jax2tf_kwargs: Optional dict. Arguments jax2tf.convert. See documentation jax2tf.convert. native_serialization polymorphic_shapes provided, automatically computed. format string. export format. Supported values: \"tf_saved_model\" \"onnx\".  Defaults \"tf_saved_model\". verbose Bool. Whether print variables exported model. Defaults NULL, uses default value set different backends formats. input_signature Optional. Specifies shape dtype model inputs. Can structure keras.InputSpec, tf.TensorSpec, backend.KerasTensor, backend tensor. provided, automatically computed. Defaults NULL.","code":""},{"path":"https://keras3.posit.co/dev/reference/export_savedmodel.keras.src.models.model.Model.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Export the model as an artifact for inference. — export_savedmodel.keras.src.models.model.Model","text":"called primarily side effect exporting object. first argument, object also returned, invisibly, enable usage pipe.","code":""},{"path":"https://keras3.posit.co/dev/reference/export_savedmodel.keras.src.models.model.Model.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Export the model as an artifact for inference. — export_savedmodel.keras.src.models.model.Model","text":"export ONNX inference.","code":"# Create the artifact model |> tensorflow::export_savedmodel(\"path/to/location\")  # Later, in a different process/environment... library(tensorflow) reloaded_artifact <- tf$saved_model$load(\"path/to/location\") predictions <- reloaded_artifact$serve(input_data)  # see tfdeploy::serve_savedmodel() for serving a model over a local web api. # Export the model as a ONNX artifact model |> export_savedmodel(\"path/to/location\", format = \"onnx\")  # Load the artifact in a different process/environment onnxruntime <- reticulate::import(\"onnxruntime\") ort_session <- onnxruntime$InferenceSession(\"path/to/location\") input_data <- list(....) names(input_data) <- sapply(ort_session$get_inputs(), `[[`, \"name\") predictions <- ort_session$run(NULL, input_data)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/fit.keras.src.models.model.Model.html","id":null,"dir":"Reference","previous_headings":"","what":"Train a model for a fixed number of epochs (dataset iterations). — fit.keras.src.models.model.Model","title":"Train a model for a fixed number of epochs (dataset iterations). — fit.keras.src.models.model.Model","text":"Train model fixed number epochs (dataset iterations).","code":""},{"path":"https://keras3.posit.co/dev/reference/fit.keras.src.models.model.Model.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Train a model for a fixed number of epochs (dataset iterations). — fit.keras.src.models.model.Model","text":"","code":"# S3 method for class 'keras.src.models.model.Model' fit(   object,   x = NULL,   y = NULL,   ...,   batch_size = NULL,   epochs = 1L,   callbacks = NULL,   validation_split = 0,   validation_data = NULL,   shuffle = TRUE,   class_weight = NULL,   sample_weight = NULL,   initial_epoch = 1L,   steps_per_epoch = NULL,   validation_steps = NULL,   validation_batch_size = NULL,   validation_freq = 1L,   verbose = getOption(\"keras.verbose\", default = \"auto\"),   view_metrics = getOption(\"keras.view_metrics\", default = \"auto\") )"},{"path":"https://keras3.posit.co/dev/reference/fit.keras.src.models.model.Model.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Train a model for a fixed number of epochs (dataset iterations). — fit.keras.src.models.model.Model","text":"object Keras model object x Input data. can : array (array-like), list arrays (case model multiple inputs). backend-native tensor, list tensors (case model multiple inputs). named list mapping input names corresponding array/tensors, model named inputs. tf.data.Dataset. return tuple either (inputs, targets) (inputs, targets, sample_weights). generator returning (inputs, targets) (inputs, targets, sample_weights). y Target data. Like input data x, can either array(s) backend-native tensor(s). x TF Dataset generator, y specified (since targets obtained x). ... Additional arguments passed model fit() method. batch_size Integer NULL. Number samples per gradient update. unspecified, batch_size default 32. specify batch_size data form TF Datasets generators, (since generate batches). epochs Integer. Number epochs train model. epoch iteration entire x y data provided (unless steps_per_epoch flag set something NULL). Note conjunction initial_epoch, epochs understood \"final epoch\". model trained number iterations given epochs, merely epoch index epochs reached. callbacks List Callback() instances. List callbacks apply training. See callback_*. validation_split Fraction training data used validation data. model set apart fraction training data, train , evaluate loss model metrics data end epoch. validation data selected last samples x y data provided, shuffling. argument supported x y made arrays tensors. validation_data validation_split provided, validation_data override validation_split. validation_data Data evaluate loss model metrics end epoch. model trained data. Thus, note fact validation loss data provided using validation_split validation_data affected regularization layers like noise dropout. validation_data override validation_split. can : tuple (x_val, y_val) arrays tensors. tuple (x_val, y_val, val_sample_weights) arrays. generator returning (inputs, targets) (inputs, targets, sample_weights). shuffle Boolean, whether shuffle training data epoch. argument ignored x generator TF Dataset. class_weight Optional named list mapping class indices (integers, 0-based) weight (float) value, used weighting loss function (training ). can useful tell model \"pay attention\" samples -represented class. class_weight specified targets rank 2 greater, either y must one-hot encoded, explicit final dimension 1 must included sparse class labels. sample_weight Optional array tensor weights training samples, used weighting loss function (training ). can either pass flat (1D) array tensor length input samples (1:1 mapping weights samples), case temporal data, can pass 2D array tensor shape (samples, sequence_length) apply different weight every timestep every sample. argument supported x tf.data.Dataset, Python generator function. Instead, provide sample_weights third element x. Note sample weighting apply metrics specified via metrics argument compile(). apply sample weighting metrics, can specify via weighted_metrics compile() instead. initial_epoch Integer. Epoch start training (useful resuming previous training run). steps_per_epoch Integer NULL. Total number steps (batches samples) declaring one epoch finished starting next epoch. training input tensors arrays, default NULL means value used number samples dataset divided batch size, 1 determined. x tf.data.Dataset Python generator function, epoch run input dataset exhausted. passing infinitely repeating dataset, must specify steps_per_epoch argument, otherwise training run indefinitely. validation_steps relevant validation_data provided. Total number steps (batches samples) draw stopping performing validation end every epoch. validation_steps NULL, validation run validation_data dataset exhausted. case infinitely repeating dataset, run indefinitely. validation_steps specified part dataset consumed, evaluation start beginning dataset epoch. ensures validation samples used every time. validation_batch_size Integer NULL. Number samples per validation batch. unspecified, default batch_size. specify validation_batch_size data form TF Datasets generator instances (since generate batches). validation_freq relevant validation data provided. Specifies many training epochs run new validation run performed, e.g. validation_freq=2 runs validation every 2 epochs. verbose \"auto\", 0, 1, 2. Verbosity mode. 0 = silent, 1 = progress bar, 2 = one line per epoch. \"auto\" becomes 1 cases, 2 knitr render running distributed training server. Note progress bar particularly useful logged file, verbose=2 recommended running interactively (e.g., production environment). Defaults \"auto\". view_metrics View realtime plot training metrics (epoch). default (\"auto\") display plot running within RStudio, metrics specified model compile(), epochs > 1 verbose > 0. Set global options(keras.view_metrics = ) option establish different default.","code":""},{"path":"https://keras3.posit.co/dev/reference/fit.keras.src.models.model.Model.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Train a model for a fixed number of epochs (dataset iterations). — fit.keras.src.models.model.Model","text":"keras_training_history object, named list: list(params = <params>, metrics = <metrics>\"), S3 methods print(), plot(), .data.frame(). metrics field record training loss values metrics values successive epochs, well validation loss values validation metrics values (applicable).","code":""},{"path":"https://keras3.posit.co/dev/reference/fit.keras.src.models.model.Model.html","id":"details","dir":"Reference","previous_headings":"","what":"Details","title":"Train a model for a fixed number of epochs (dataset iterations). — fit.keras.src.models.model.Model","text":"Unpacking behavior iterator-like inputs: common pattern pass iterator like object tf.data.Dataset generator fit(), fact yield features (x) optionally targets (y) sample weights (sample_weight). Keras requires output iterator-likes unambiguous. iterator return tuple() length 1, 2, 3, optional second third elements used y sample_weight respectively. type provided wrapped length-one tuple(), effectively treating everything x. yielding named lists, still adhere top-level tuple structure, e.g. tuple(list(x0 = x0, x = x1), y). Keras attempt separate features, targets, weights keys single dict.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/freeze_weights.html","id":null,"dir":"Reference","previous_headings":"","what":"Freeze and unfreeze weights — freeze_weights","title":"Freeze and unfreeze weights — freeze_weights","text":"Freeze weights model layer longer trainable.","code":""},{"path":"https://keras3.posit.co/dev/reference/freeze_weights.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Freeze and unfreeze weights — freeze_weights","text":"","code":"freeze_weights(object, from = NULL, to = NULL, which = NULL)  unfreeze_weights(object, from = NULL, to = NULL, which = NULL)"},{"path":"https://keras3.posit.co/dev/reference/freeze_weights.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Freeze and unfreeze weights — freeze_weights","text":"object Keras model layer object Layer instance, layer name, layer index within model Layer instance, layer name, layer index within model layer names, integer positions, layers, logical vector (length(object$layers)), function returning logical vector.","code":""},{"path":"https://keras3.posit.co/dev/reference/freeze_weights.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Freeze and unfreeze weights — freeze_weights","text":"input object frozen weights returned, invisibly. Note, object modified place, return value provided make usage pipe convenient.","code":""},{"path":"https://keras3.posit.co/dev/reference/freeze_weights.html","id":"note","dir":"Reference","previous_headings":"","what":"Note","title":"Freeze and unfreeze weights — freeze_weights","text":"layer arguments inclusive. applied model, freeze unfreeze global operation layers model (.e. layers within specified range set opposite value, e.g. unfrozen call freeze). Models must compiled weights frozen unfrozen.","code":""},{"path":"https://keras3.posit.co/dev/reference/freeze_weights.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Freeze and unfreeze weights — freeze_weights","text":"","code":"# instantiate a VGG16 model conv_base <- application_vgg16(   weights = \"imagenet\",   include_top = FALSE,   input_shape = c(150, 150, 3) )  # freeze it's weights freeze_weights(conv_base)  # Note the \"Trainable\" column conv_base ## Model: \"vgg16\" ## +-----------------------------+-----------------------+------------+-------+ ## | Layer (type)                | Output Shape          |    Param # | Trai… | ## +=============================+=======================+============+=======+ ## | input_layer (InputLayer)    | (None, 150, 150, 3)   |          0 |   -   | ## +-----------------------------+-----------------------+------------+-------+ ## | block1_conv1 (Conv2D)       | (None, 150, 150, 64)  |      1,792 |   N   | ## +-----------------------------+-----------------------+------------+-------+ ## | block1_conv2 (Conv2D)       | (None, 150, 150, 64)  |     36,928 |   N   | ## +-----------------------------+-----------------------+------------+-------+ ## | block1_pool (MaxPooling2D)  | (None, 75, 75, 64)    |          0 |   -   | ## +-----------------------------+-----------------------+------------+-------+ ## | block2_conv1 (Conv2D)       | (None, 75, 75, 128)   |     73,856 |   N   | ## +-----------------------------+-----------------------+------------+-------+ ## | block2_conv2 (Conv2D)       | (None, 75, 75, 128)   |    147,584 |   N   | ## +-----------------------------+-----------------------+------------+-------+ ## | block2_pool (MaxPooling2D)  | (None, 37, 37, 128)   |          0 |   -   | ## +-----------------------------+-----------------------+------------+-------+ ## | block3_conv1 (Conv2D)       | (None, 37, 37, 256)   |    295,168 |   N   | ## +-----------------------------+-----------------------+------------+-------+ ## | block3_conv2 (Conv2D)       | (None, 37, 37, 256)   |    590,080 |   N   | ## +-----------------------------+-----------------------+------------+-------+ ## | block3_conv3 (Conv2D)       | (None, 37, 37, 256)   |    590,080 |   N   | ## +-----------------------------+-----------------------+------------+-------+ ## | block3_pool (MaxPooling2D)  | (None, 18, 18, 256)   |          0 |   -   | ## +-----------------------------+-----------------------+------------+-------+ ## | block4_conv1 (Conv2D)       | (None, 18, 18, 512)   |  1,180,160 |   N   | ## +-----------------------------+-----------------------+------------+-------+ ## | block4_conv2 (Conv2D)       | (None, 18, 18, 512)   |  2,359,808 |   N   | ## +-----------------------------+-----------------------+------------+-------+ ## | block4_conv3 (Conv2D)       | (None, 18, 18, 512)   |  2,359,808 |   N   | ## +-----------------------------+-----------------------+------------+-------+ ## | block4_pool (MaxPooling2D)  | (None, 9, 9, 512)     |          0 |   -   | ## +-----------------------------+-----------------------+------------+-------+ ## | block5_conv1 (Conv2D)       | (None, 9, 9, 512)     |  2,359,808 |   N   | ## +-----------------------------+-----------------------+------------+-------+ ## | block5_conv2 (Conv2D)       | (None, 9, 9, 512)     |  2,359,808 |   N   | ## +-----------------------------+-----------------------+------------+-------+ ## | block5_conv3 (Conv2D)       | (None, 9, 9, 512)     |  2,359,808 |   N   | ## +-----------------------------+-----------------------+------------+-------+ ## | block5_pool (MaxPooling2D)  | (None, 4, 4, 512)     |          0 |   -   | ## +-----------------------------+-----------------------+------------+-------+ ##  Total params: 14,714,688 (56.13 MB) ##  Trainable params: 0 (0.00 B) ##  Non-trainable params: 14,714,688 (56.13 MB) # create a composite model that includes the base + more layers model <- keras_model_sequential(input_batch_shape = shape(conv_base$input)) |>   conv_base() |>   layer_flatten() |>   layer_dense(units = 256, activation = \"relu\") |>   layer_dense(units = 1, activation = \"sigmoid\")  # compile model |> compile(   loss = \"binary_crossentropy\",   optimizer = optimizer_rmsprop(learning_rate = 2e-5),   metrics = c(\"accuracy\") )  model ## Model: \"sequential\" ## +-----------------------------+-----------------------+------------+-------+ ## | Layer (type)                | Output Shape          |    Param # | Trai… | ## +=============================+=======================+============+=======+ ## | vgg16 (Functional)          | (None, 4, 4, 512)     | 14,714,688 |   N   | ## +-----------------------------+-----------------------+------------+-------+ ## | flatten (Flatten)           | (None, 8192)          |          0 |   -   | ## +-----------------------------+-----------------------+------------+-------+ ## | dense (Dense)               | (None, 256)           |  2,097,408 |   Y   | ## +-----------------------------+-----------------------+------------+-------+ ## | dense_1 (Dense)             | (None, 1)             |        257 |   Y   | ## +-----------------------------+-----------------------+------------+-------+ ##  Total params: 16,812,353 (64.13 MB) ##  Trainable params: 2,097,665 (8.00 MB) ##  Non-trainable params: 14,714,688 (56.13 MB) print(model, expand_nested = TRUE) ## Model: \"sequential\" ## +-----------------------------+-----------------------+------------+-------+ ## | Layer (type)                | Output Shape          |    Param # | Trai… | ## +=============================+=======================+============+=======+ ## | vgg16 (Functional)          | (None, 4, 4, 512)     | 14,714,688 |   N   | ## +-----------------------------+-----------------------+------------+-------+ ## |    > input_layer            | (None, 150, 150, 3)   |          0 |   -   | ## | (InputLayer)                |                       |            |       | ## +-----------------------------+-----------------------+------------+-------+ ## |    > block1_conv1 (Conv2D)  | (None, 150, 150, 64)  |      1,792 |   N   | ## +-----------------------------+-----------------------+------------+-------+ ## |    > block1_conv2 (Conv2D)  | (None, 150, 150, 64)  |     36,928 |   N   | ## +-----------------------------+-----------------------+------------+-------+ ## |    > block1_pool            | (None, 75, 75, 64)    |          0 |   -   | ## | (MaxPooling2D)              |                       |            |       | ## +-----------------------------+-----------------------+------------+-------+ ## |    > block2_conv1 (Conv2D)  | (None, 75, 75, 128)   |     73,856 |   N   | ## +-----------------------------+-----------------------+------------+-------+ ## |    > block2_conv2 (Conv2D)  | (None, 75, 75, 128)   |    147,584 |   N   | ## +-----------------------------+-----------------------+------------+-------+ ## |    > block2_pool            | (None, 37, 37, 128)   |          0 |   -   | ## | (MaxPooling2D)              |                       |            |       | ## +-----------------------------+-----------------------+------------+-------+ ## |    > block3_conv1 (Conv2D)  | (None, 37, 37, 256)   |    295,168 |   N   | ## +-----------------------------+-----------------------+------------+-------+ ## |    > block3_conv2 (Conv2D)  | (None, 37, 37, 256)   |    590,080 |   N   | ## +-----------------------------+-----------------------+------------+-------+ ## |    > block3_conv3 (Conv2D)  | (None, 37, 37, 256)   |    590,080 |   N   | ## +-----------------------------+-----------------------+------------+-------+ ## |    > block3_pool            | (None, 18, 18, 256)   |          0 |   -   | ## | (MaxPooling2D)              |                       |            |       | ## +-----------------------------+-----------------------+------------+-------+ ## |    > block4_conv1 (Conv2D)  | (None, 18, 18, 512)   |  1,180,160 |   N   | ## +-----------------------------+-----------------------+------------+-------+ ## |    > block4_conv2 (Conv2D)  | (None, 18, 18, 512)   |  2,359,808 |   N   | ## +-----------------------------+-----------------------+------------+-------+ ## |    > block4_conv3 (Conv2D)  | (None, 18, 18, 512)   |  2,359,808 |   N   | ## +-----------------------------+-----------------------+------------+-------+ ## |    > block4_pool            | (None, 9, 9, 512)     |          0 |   -   | ## | (MaxPooling2D)              |                       |            |       | ## +-----------------------------+-----------------------+------------+-------+ ## |    > block5_conv1 (Conv2D)  | (None, 9, 9, 512)     |  2,359,808 |   N   | ## +-----------------------------+-----------------------+------------+-------+ ## |    > block5_conv2 (Conv2D)  | (None, 9, 9, 512)     |  2,359,808 |   N   | ## +-----------------------------+-----------------------+------------+-------+ ## |    > block5_conv3 (Conv2D)  | (None, 9, 9, 512)     |  2,359,808 |   N   | ## +-----------------------------+-----------------------+------------+-------+ ## |    > block5_pool            | (None, 4, 4, 512)     |          0 |   -   | ## | (MaxPooling2D)              |                       |            |       | ## +-----------------------------+-----------------------+------------+-------+ ## | flatten (Flatten)           | (None, 8192)          |          0 |   -   | ## +-----------------------------+-----------------------+------------+-------+ ## | dense (Dense)               | (None, 256)           |  2,097,408 |   Y   | ## +-----------------------------+-----------------------+------------+-------+ ## | dense_1 (Dense)             | (None, 1)             |        257 |   Y   | ## +-----------------------------+-----------------------+------------+-------+ ##  Total params: 16,812,353 (64.13 MB) ##  Trainable params: 2,097,665 (8.00 MB) ##  Non-trainable params: 14,714,688 (56.13 MB) # unfreeze weights from \"block5_conv1\" on unfreeze_weights(conv_base, from = \"block5_conv1\")  # compile again since we froze or unfroze weights model |> compile(   loss = \"binary_crossentropy\",   optimizer = optimizer_rmsprop(learning_rate = 2e-5),   metrics = c(\"accuracy\") )  conv_base ## Model: \"vgg16\" ## +-----------------------------+-----------------------+------------+-------+ ## | Layer (type)                | Output Shape          |    Param # | Trai… | ## +=============================+=======================+============+=======+ ## | input_layer (InputLayer)    | (None, 150, 150, 3)   |          0 |   -   | ## +-----------------------------+-----------------------+------------+-------+ ## | block1_conv1 (Conv2D)       | (None, 150, 150, 64)  |      1,792 |   N   | ## +-----------------------------+-----------------------+------------+-------+ ## | block1_conv2 (Conv2D)       | (None, 150, 150, 64)  |     36,928 |   N   | ## +-----------------------------+-----------------------+------------+-------+ ## | block1_pool (MaxPooling2D)  | (None, 75, 75, 64)    |          0 |   -   | ## +-----------------------------+-----------------------+------------+-------+ ## | block2_conv1 (Conv2D)       | (None, 75, 75, 128)   |     73,856 |   N   | ## +-----------------------------+-----------------------+------------+-------+ ## | block2_conv2 (Conv2D)       | (None, 75, 75, 128)   |    147,584 |   N   | ## +-----------------------------+-----------------------+------------+-------+ ## | block2_pool (MaxPooling2D)  | (None, 37, 37, 128)   |          0 |   -   | ## +-----------------------------+-----------------------+------------+-------+ ## | block3_conv1 (Conv2D)       | (None, 37, 37, 256)   |    295,168 |   N   | ## +-----------------------------+-----------------------+------------+-------+ ## | block3_conv2 (Conv2D)       | (None, 37, 37, 256)   |    590,080 |   N   | ## +-----------------------------+-----------------------+------------+-------+ ## | block3_conv3 (Conv2D)       | (None, 37, 37, 256)   |    590,080 |   N   | ## +-----------------------------+-----------------------+------------+-------+ ## | block3_pool (MaxPooling2D)  | (None, 18, 18, 256)   |          0 |   -   | ## +-----------------------------+-----------------------+------------+-------+ ## | block4_conv1 (Conv2D)       | (None, 18, 18, 512)   |  1,180,160 |   N   | ## +-----------------------------+-----------------------+------------+-------+ ## | block4_conv2 (Conv2D)       | (None, 18, 18, 512)   |  2,359,808 |   N   | ## +-----------------------------+-----------------------+------------+-------+ ## | block4_conv3 (Conv2D)       | (None, 18, 18, 512)   |  2,359,808 |   N   | ## +-----------------------------+-----------------------+------------+-------+ ## | block4_pool (MaxPooling2D)  | (None, 9, 9, 512)     |          0 |   -   | ## +-----------------------------+-----------------------+------------+-------+ ## | block5_conv1 (Conv2D)       | (None, 9, 9, 512)     |  2,359,808 |   Y   | ## +-----------------------------+-----------------------+------------+-------+ ## | block5_conv2 (Conv2D)       | (None, 9, 9, 512)     |  2,359,808 |   Y   | ## +-----------------------------+-----------------------+------------+-------+ ## | block5_conv3 (Conv2D)       | (None, 9, 9, 512)     |  2,359,808 |   Y   | ## +-----------------------------+-----------------------+------------+-------+ ## | block5_pool (MaxPooling2D)  | (None, 4, 4, 512)     |          0 |   -   | ## +-----------------------------+-----------------------+------------+-------+ ##  Total params: 14,714,688 (56.13 MB) ##  Trainable params: 7,079,424 (27.01 MB) ##  Non-trainable params: 7,635,264 (29.13 MB) print(model, expand_nested = TRUE) ## Model: \"sequential\" ## +-----------------------------+-----------------------+------------+-------+ ## | Layer (type)                | Output Shape          |    Param # | Trai… | ## +=============================+=======================+============+=======+ ## | vgg16 (Functional)          | (None, 4, 4, 512)     | 14,714,688 |   Y   | ## +-----------------------------+-----------------------+------------+-------+ ## |    > input_layer            | (None, 150, 150, 3)   |          0 |   -   | ## | (InputLayer)                |                       |            |       | ## +-----------------------------+-----------------------+------------+-------+ ## |    > block1_conv1 (Conv2D)  | (None, 150, 150, 64)  |      1,792 |   N   | ## +-----------------------------+-----------------------+------------+-------+ ## |    > block1_conv2 (Conv2D)  | (None, 150, 150, 64)  |     36,928 |   N   | ## +-----------------------------+-----------------------+------------+-------+ ## |    > block1_pool            | (None, 75, 75, 64)    |          0 |   -   | ## | (MaxPooling2D)              |                       |            |       | ## +-----------------------------+-----------------------+------------+-------+ ## |    > block2_conv1 (Conv2D)  | (None, 75, 75, 128)   |     73,856 |   N   | ## +-----------------------------+-----------------------+------------+-------+ ## |    > block2_conv2 (Conv2D)  | (None, 75, 75, 128)   |    147,584 |   N   | ## +-----------------------------+-----------------------+------------+-------+ ## |    > block2_pool            | (None, 37, 37, 128)   |          0 |   -   | ## | (MaxPooling2D)              |                       |            |       | ## +-----------------------------+-----------------------+------------+-------+ ## |    > block3_conv1 (Conv2D)  | (None, 37, 37, 256)   |    295,168 |   N   | ## +-----------------------------+-----------------------+------------+-------+ ## |    > block3_conv2 (Conv2D)  | (None, 37, 37, 256)   |    590,080 |   N   | ## +-----------------------------+-----------------------+------------+-------+ ## |    > block3_conv3 (Conv2D)  | (None, 37, 37, 256)   |    590,080 |   N   | ## +-----------------------------+-----------------------+------------+-------+ ## |    > block3_pool            | (None, 18, 18, 256)   |          0 |   -   | ## | (MaxPooling2D)              |                       |            |       | ## +-----------------------------+-----------------------+------------+-------+ ## |    > block4_conv1 (Conv2D)  | (None, 18, 18, 512)   |  1,180,160 |   N   | ## +-----------------------------+-----------------------+------------+-------+ ## |    > block4_conv2 (Conv2D)  | (None, 18, 18, 512)   |  2,359,808 |   N   | ## +-----------------------------+-----------------------+------------+-------+ ## |    > block4_conv3 (Conv2D)  | (None, 18, 18, 512)   |  2,359,808 |   N   | ## +-----------------------------+-----------------------+------------+-------+ ## |    > block4_pool            | (None, 9, 9, 512)     |          0 |   -   | ## | (MaxPooling2D)              |                       |            |       | ## +-----------------------------+-----------------------+------------+-------+ ## |    > block5_conv1 (Conv2D)  | (None, 9, 9, 512)     |  2,359,808 |   Y   | ## +-----------------------------+-----------------------+------------+-------+ ## |    > block5_conv2 (Conv2D)  | (None, 9, 9, 512)     |  2,359,808 |   Y   | ## +-----------------------------+-----------------------+------------+-------+ ## |    > block5_conv3 (Conv2D)  | (None, 9, 9, 512)     |  2,359,808 |   Y   | ## +-----------------------------+-----------------------+------------+-------+ ## |    > block5_pool            | (None, 4, 4, 512)     |          0 |   -   | ## | (MaxPooling2D)              |                       |            |       | ## +-----------------------------+-----------------------+------------+-------+ ## | flatten (Flatten)           | (None, 8192)          |          0 |   -   | ## +-----------------------------+-----------------------+------------+-------+ ## | dense (Dense)               | (None, 256)           |  2,097,408 |   Y   | ## +-----------------------------+-----------------------+------------+-------+ ## | dense_1 (Dense)             | (None, 1)             |        257 |   Y   | ## +-----------------------------+-----------------------+------------+-------+ ##  Total params: 16,812,353 (64.13 MB) ##  Trainable params: 9,177,089 (35.01 MB) ##  Non-trainable params: 7,635,264 (29.13 MB) # freeze only the last 5 layers freeze_weights(conv_base, from = -5) conv_base ## Model: \"vgg16\" ## +-----------------------------+-----------------------+------------+-------+ ## | Layer (type)                | Output Shape          |    Param # | Trai… | ## +=============================+=======================+============+=======+ ## | input_layer (InputLayer)    | (None, 150, 150, 3)   |          0 |   -   | ## +-----------------------------+-----------------------+------------+-------+ ## | block1_conv1 (Conv2D)       | (None, 150, 150, 64)  |      1,792 |   Y   | ## +-----------------------------+-----------------------+------------+-------+ ## | block1_conv2 (Conv2D)       | (None, 150, 150, 64)  |     36,928 |   Y   | ## +-----------------------------+-----------------------+------------+-------+ ## | block1_pool (MaxPooling2D)  | (None, 75, 75, 64)    |          0 |   -   | ## +-----------------------------+-----------------------+------------+-------+ ## | block2_conv1 (Conv2D)       | (None, 75, 75, 128)   |     73,856 |   Y   | ## +-----------------------------+-----------------------+------------+-------+ ## | block2_conv2 (Conv2D)       | (None, 75, 75, 128)   |    147,584 |   Y   | ## +-----------------------------+-----------------------+------------+-------+ ## | block2_pool (MaxPooling2D)  | (None, 37, 37, 128)   |          0 |   -   | ## +-----------------------------+-----------------------+------------+-------+ ## | block3_conv1 (Conv2D)       | (None, 37, 37, 256)   |    295,168 |   Y   | ## +-----------------------------+-----------------------+------------+-------+ ## | block3_conv2 (Conv2D)       | (None, 37, 37, 256)   |    590,080 |   Y   | ## +-----------------------------+-----------------------+------------+-------+ ## | block3_conv3 (Conv2D)       | (None, 37, 37, 256)   |    590,080 |   Y   | ## +-----------------------------+-----------------------+------------+-------+ ## | block3_pool (MaxPooling2D)  | (None, 18, 18, 256)   |          0 |   -   | ## +-----------------------------+-----------------------+------------+-------+ ## | block4_conv1 (Conv2D)       | (None, 18, 18, 512)   |  1,180,160 |   Y   | ## +-----------------------------+-----------------------+------------+-------+ ## | block4_conv2 (Conv2D)       | (None, 18, 18, 512)   |  2,359,808 |   Y   | ## +-----------------------------+-----------------------+------------+-------+ ## | block4_conv3 (Conv2D)       | (None, 18, 18, 512)   |  2,359,808 |   Y   | ## +-----------------------------+-----------------------+------------+-------+ ## | block4_pool (MaxPooling2D)  | (None, 9, 9, 512)     |          0 |   -   | ## +-----------------------------+-----------------------+------------+-------+ ## | block5_conv1 (Conv2D)       | (None, 9, 9, 512)     |  2,359,808 |   N   | ## +-----------------------------+-----------------------+------------+-------+ ## | block5_conv2 (Conv2D)       | (None, 9, 9, 512)     |  2,359,808 |   N   | ## +-----------------------------+-----------------------+------------+-------+ ## | block5_conv3 (Conv2D)       | (None, 9, 9, 512)     |  2,359,808 |   N   | ## +-----------------------------+-----------------------+------------+-------+ ## | block5_pool (MaxPooling2D)  | (None, 4, 4, 512)     |          0 |   -   | ## +-----------------------------+-----------------------+------------+-------+ ##  Total params: 14,714,688 (56.13 MB) ##  Trainable params: 7,635,264 (29.13 MB) ##  Non-trainable params: 7,079,424 (27.01 MB) # freeze only the last 5 layers, a different way unfreeze_weights(conv_base, to = -6) conv_base ## Model: \"vgg16\" ## +-----------------------------+-----------------------+------------+-------+ ## | Layer (type)                | Output Shape          |    Param # | Trai… | ## +=============================+=======================+============+=======+ ## | input_layer (InputLayer)    | (None, 150, 150, 3)   |          0 |   -   | ## +-----------------------------+-----------------------+------------+-------+ ## | block1_conv1 (Conv2D)       | (None, 150, 150, 64)  |      1,792 |   Y   | ## +-----------------------------+-----------------------+------------+-------+ ## | block1_conv2 (Conv2D)       | (None, 150, 150, 64)  |     36,928 |   Y   | ## +-----------------------------+-----------------------+------------+-------+ ## | block1_pool (MaxPooling2D)  | (None, 75, 75, 64)    |          0 |   -   | ## +-----------------------------+-----------------------+------------+-------+ ## | block2_conv1 (Conv2D)       | (None, 75, 75, 128)   |     73,856 |   Y   | ## +-----------------------------+-----------------------+------------+-------+ ## | block2_conv2 (Conv2D)       | (None, 75, 75, 128)   |    147,584 |   Y   | ## +-----------------------------+-----------------------+------------+-------+ ## | block2_pool (MaxPooling2D)  | (None, 37, 37, 128)   |          0 |   -   | ## +-----------------------------+-----------------------+------------+-------+ ## | block3_conv1 (Conv2D)       | (None, 37, 37, 256)   |    295,168 |   Y   | ## +-----------------------------+-----------------------+------------+-------+ ## | block3_conv2 (Conv2D)       | (None, 37, 37, 256)   |    590,080 |   Y   | ## +-----------------------------+-----------------------+------------+-------+ ## | block3_conv3 (Conv2D)       | (None, 37, 37, 256)   |    590,080 |   Y   | ## +-----------------------------+-----------------------+------------+-------+ ## | block3_pool (MaxPooling2D)  | (None, 18, 18, 256)   |          0 |   -   | ## +-----------------------------+-----------------------+------------+-------+ ## | block4_conv1 (Conv2D)       | (None, 18, 18, 512)   |  1,180,160 |   Y   | ## +-----------------------------+-----------------------+------------+-------+ ## | block4_conv2 (Conv2D)       | (None, 18, 18, 512)   |  2,359,808 |   Y   | ## +-----------------------------+-----------------------+------------+-------+ ## | block4_conv3 (Conv2D)       | (None, 18, 18, 512)   |  2,359,808 |   Y   | ## +-----------------------------+-----------------------+------------+-------+ ## | block4_pool (MaxPooling2D)  | (None, 9, 9, 512)     |          0 |   -   | ## +-----------------------------+-----------------------+------------+-------+ ## | block5_conv1 (Conv2D)       | (None, 9, 9, 512)     |  2,359,808 |   N   | ## +-----------------------------+-----------------------+------------+-------+ ## | block5_conv2 (Conv2D)       | (None, 9, 9, 512)     |  2,359,808 |   N   | ## +-----------------------------+-----------------------+------------+-------+ ## | block5_conv3 (Conv2D)       | (None, 9, 9, 512)     |  2,359,808 |   N   | ## +-----------------------------+-----------------------+------------+-------+ ## | block5_pool (MaxPooling2D)  | (None, 4, 4, 512)     |          0 |   -   | ## +-----------------------------+-----------------------+------------+-------+ ##  Total params: 14,714,688 (56.13 MB) ##  Trainable params: 7,635,264 (29.13 MB) ##  Non-trainable params: 7,079,424 (27.01 MB) # Freeze only layers of a certain type, e.g, BatchNorm layers batch_norm_layer_class_name <- class(layer_batch_normalization())[1] is_batch_norm_layer <- function(x) inherits(x, batch_norm_layer_class_name)  model <- application_efficientnet_b0() freeze_weights(model, which = is_batch_norm_layer) # print(model)  # equivalent to: for(layer in model$layers) {   if(is_batch_norm_layer(layer))     layer$trainable <- FALSE   else     layer$trainable <- TRUE }"},{"path":"https://keras3.posit.co/dev/reference/get_config.html","id":null,"dir":"Reference","previous_headings":"","what":"Layer/Model configuration — get_config","title":"Layer/Model configuration — get_config","text":"layer config object returned get_config() contains configuration layer model. layer model can reinstantiated later (without trained weights) configuration using from_config(). config include connectivity information, class name (handled externally).","code":""},{"path":"https://keras3.posit.co/dev/reference/get_config.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Layer/Model configuration — get_config","text":"","code":"get_config(object)  from_config(config, custom_objects = NULL)"},{"path":"https://keras3.posit.co/dev/reference/get_config.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Layer/Model configuration — get_config","text":"object Layer model object config Object layer model configuration custom_objects list custom objects needed instantiate layer, e.g., custom layers defined new_layer_class() similar.","code":""},{"path":"https://keras3.posit.co/dev/reference/get_config.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Layer/Model configuration — get_config","text":"get_config() returns object configuration, from_config() returns re-instantiation object.","code":""},{"path":"https://keras3.posit.co/dev/reference/get_config.html","id":"note","dir":"Reference","previous_headings":"","what":"Note","title":"Layer/Model configuration — get_config","text":"Objects returned get_config() serializable via RDS. want save restore model across sessions, can use save_model_config() (model configuration , weights) save_model() save model configuration weights filesystem.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/get_custom_objects.html","id":null,"dir":"Reference","previous_headings":"","what":"Get/set the currently registered custom objects. — get_custom_objects","title":"Get/set the currently registered custom objects. — get_custom_objects","text":"Custom objects set using custom_object_scope() added global list custom objects, appear returned list.","code":""},{"path":"https://keras3.posit.co/dev/reference/get_custom_objects.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Get/set the currently registered custom objects. — get_custom_objects","text":"","code":"get_custom_objects()  set_custom_objects(objects = named_list(), clear = TRUE)"},{"path":"https://keras3.posit.co/dev/reference/get_custom_objects.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Get/set the currently registered custom objects. — get_custom_objects","text":"objects named list custom objects, returned get_custom_objects() set_custom_objects(). clear bool, whether clear custom object registry populating objects.","code":""},{"path":"https://keras3.posit.co/dev/reference/get_custom_objects.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Get/set the currently registered custom objects. — get_custom_objects","text":"R named list mapping registered names registered objects. set_custom_objects() returns registry values updating, invisibly.","code":""},{"path":"https://keras3.posit.co/dev/reference/get_custom_objects.html","id":"note","dir":"Reference","previous_headings":"","what":"Note","title":"Get/set the currently registered custom objects. — get_custom_objects","text":"register_keras_serializable() preferred set_custom_objects() registering new objects.","code":""},{"path":"https://keras3.posit.co/dev/reference/get_custom_objects.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Get/set the currently registered custom objects. — get_custom_objects","text":"can use set_custom_objects() restore previous registry state.","code":"get_custom_objects() # within a function, if you want to temporarily modify the registry, function() {   orig_objects <- set_custom_objects(clear = TRUE)   on.exit(set_custom_objects(orig_objects))    ## temporarily modify the global registry   # register_keras_serializable(....)   # ....  <do work>   # on.exit(), the previous registry state is restored. }"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/get_file.html","id":null,"dir":"Reference","previous_headings":"","what":"Downloads a file from a URL if it not already in the cache. — get_file","title":"Downloads a file from a URL if it not already in the cache. — get_file","text":"default file url origin downloaded cache_dir ~/.keras, placed cache_subdir datasets, given filename fname. final location file example.txt therefore ~/.keras/datasets/example.txt. Files .tar, .tar.gz, .tar.bz, .zip formats can also extracted. Passing hash verify file download. command line programs shasum sha256sum can compute hash.","code":""},{"path":"https://keras3.posit.co/dev/reference/get_file.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Downloads a file from a URL if it not already in the cache. — get_file","text":"","code":"get_file(   fname = NULL,   origin = NULL,   ...,   file_hash = NULL,   cache_subdir = \"datasets\",   hash_algorithm = \"auto\",   extract = FALSE,   archive_format = \"auto\",   cache_dir = NULL,   force_download = FALSE )"},{"path":"https://keras3.posit.co/dev/reference/get_file.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Downloads a file from a URL if it not already in the cache. — get_file","text":"fname target single file, desired local name file. NULL, name file origin used. downloading extracting directory archive, provided fname used extraction directory name (extension). origin Original URL file. ... forward/backward compatability. file_hash expected hash string file download. sha256 md5 hash algorithms supported. cache_subdir Subdirectory Keras cache dir file saved. absolute path, e.g. \"/path//folder\" specified, file saved location. hash_algorithm Select hash algorithm verify file. options \"md5', \"sha256', \"auto'. default 'auto' detects hash algorithm use. extract TRUE, extracts archive. applicable compressed archive files like tar zip. archive_format Archive format try extracting file. Options \"auto', \"tar', \"zip', NULL. \"tar\" includes tar, tar.gz, tar.bz files. default \"auto\" corresponds c(\"tar\", \"zip\"). NULL empty list return matches found. cache_dir Location store cached files, NULL defaults Sys.getenv(\"KERAS_HOME\", \"~/.keras/\"). force_download TRUE, file always re-downloaded regardless cache state.","code":""},{"path":"https://keras3.posit.co/dev/reference/get_file.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Downloads a file from a URL if it not already in the cache. — get_file","text":"Path downloaded file. ** Warning malicious downloads ** Downloading something Internet carries risk. NEVER download file/archive trust source. recommend specify file_hash argument (hash source file known) make sure file getting one expect.","code":""},{"path":"https://keras3.posit.co/dev/reference/get_file.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Downloads a file from a URL if it not already in the cache. — get_file","text":"","code":"path_to_downloaded_file <- get_file(     \"flower_photos\",     origin = \"https://storage.googleapis.com/download.tensorflow.org/example_images/flower_photos.tgz\",     extract = TRUE )"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/get_layer.html","id":null,"dir":"Reference","previous_headings":"","what":"Retrieves a layer based on either its name (unique) or index. — get_layer","title":"Retrieves a layer based on either its name (unique) or index. — get_layer","text":"Indices based order horizontal graph traversal (bottom-) 1-based. name index provided, index take precedence.","code":""},{"path":"https://keras3.posit.co/dev/reference/get_layer.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Retrieves a layer based on either its name (unique) or index. — get_layer","text":"","code":"get_layer(object, name = NULL, index = NULL)"},{"path":"https://keras3.posit.co/dev/reference/get_layer.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Retrieves a layer based on either its name (unique) or index. — get_layer","text":"object Keras model object name String, name layer. index Integer, index layer (1-based). Also valid negative values, count end model.","code":""},{"path":"https://keras3.posit.co/dev/reference/get_layer.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Retrieves a layer based on either its name (unique) or index. — get_layer","text":"layer instance.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/get_registered_name.html","id":null,"dir":"Reference","previous_headings":"","what":"Returns the name registered to an object within the Keras framework. — get_registered_name","title":"Returns the name registered to an object within the Keras framework. — get_registered_name","text":"function part Keras serialization deserialization framework. maps objects string names associated objects serialization/deserialization.","code":""},{"path":"https://keras3.posit.co/dev/reference/get_registered_name.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Returns the name registered to an object within the Keras framework. — get_registered_name","text":"","code":"get_registered_name(obj)"},{"path":"https://keras3.posit.co/dev/reference/get_registered_name.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Returns the name registered to an object within the Keras framework. — get_registered_name","text":"obj object look .","code":""},{"path":"https://keras3.posit.co/dev/reference/get_registered_name.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Returns the name registered to an object within the Keras framework. — get_registered_name","text":"name associated object, default name object registered.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/get_registered_object.html","id":null,"dir":"Reference","previous_headings":"","what":"Returns the class associated with name if it is registered with Keras. — get_registered_object","title":"Returns the class associated with name if it is registered with Keras. — get_registered_object","text":"function part Keras serialization deserialization framework. maps strings objects associated serialization/deserialization.","code":""},{"path":"https://keras3.posit.co/dev/reference/get_registered_object.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Returns the class associated with name if it is registered with Keras. — get_registered_object","text":"","code":"get_registered_object(name, custom_objects = NULL, module_objects = NULL)"},{"path":"https://keras3.posit.co/dev/reference/get_registered_object.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Returns the class associated with name if it is registered with Keras. — get_registered_object","text":"name name look . custom_objects named list custom objects look name . Generally, custom_objects provided user. module_objects named list custom objects look name . Generally, module_objects provided midlevel library implementers.","code":""},{"path":"https://keras3.posit.co/dev/reference/get_registered_object.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Returns the class associated with name if it is registered with Keras. — get_registered_object","text":"instantiable class associated name, NULL class exists.","code":""},{"path":"https://keras3.posit.co/dev/reference/get_registered_object.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Returns the class associated with name if it is registered with Keras. — get_registered_object","text":"","code":"from_config <- function(cls, config, custom_objects = NULL) {   if ('my_custom_object_name' \\%in\\% names(config)) {     config$hidden_cls <- get_registered_object(       config$my_custom_object_name,       custom_objects = custom_objects)   } }"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/get_source_inputs.html","id":null,"dir":"Reference","previous_headings":"","what":"Returns the list of input tensors necessary to compute tensor. — get_source_inputs","title":"Returns the list of input tensors necessary to compute tensor. — get_source_inputs","text":"Output always list tensors (potentially 1 element).","code":""},{"path":"https://keras3.posit.co/dev/reference/get_source_inputs.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Returns the list of input tensors necessary to compute tensor. — get_source_inputs","text":"","code":"get_source_inputs(tensor)"},{"path":"https://keras3.posit.co/dev/reference/get_source_inputs.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Returns the list of input tensors necessary to compute tensor. — get_source_inputs","text":"tensor tensor start .","code":""},{"path":"https://keras3.posit.co/dev/reference/get_source_inputs.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Returns the list of input tensors necessary to compute tensor. — get_source_inputs","text":"List input tensors.","code":""},{"path":"https://keras3.posit.co/dev/reference/get_source_inputs.html","id":"example","dir":"Reference","previous_headings":"","what":"Example","title":"Returns the list of input tensors necessary to compute tensor. — get_source_inputs","text":"","code":"input <- keras_input(c(3)) output <- input |> layer_dense(4) |> op_multiply(5) reticulate::py_id(get_source_inputs(output)[[1]]) == reticulate::py_id(input) ## [1] TRUE"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/get_state_tree.html","id":null,"dir":"Reference","previous_headings":"","what":"Retrieves tree-like structure of model variables. — get_state_tree","title":"Retrieves tree-like structure of model variables. — get_state_tree","text":"method allows retrieval different model variables (trainable, non-trainable, optimizer, metrics). variables returned nested dictionary format, keys correspond variable names values nested representations variables.","code":""},{"path":"https://keras3.posit.co/dev/reference/get_state_tree.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Retrieves tree-like structure of model variables. — get_state_tree","text":"","code":"get_state_tree(object, value_format = \"backend_tensor\")"},{"path":"https://keras3.posit.co/dev/reference/get_state_tree.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Retrieves tree-like structure of model variables. — get_state_tree","text":"object Keras Model. value_format One \"backend_tensor\", \"numpy_array\", \"array\". kind array return leaves nested state tree.","code":""},{"path":"https://keras3.posit.co/dev/reference/get_state_tree.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Retrieves tree-like structure of model variables. — get_state_tree","text":"named list containing nested representations requested variables. names variable names, values corresponding nested named lists.","code":""},{"path":"https://keras3.posit.co/dev/reference/get_state_tree.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Retrieves tree-like structure of model variables. — get_state_tree","text":"state_tree list returned looks like:   example:","code":"model <- keras_model_sequential(name = \"my_sequential\",                                 input_shape = c(1),                                 input_name = \"my_input\") |>   layer_dense(1, activation = \"sigmoid\", name = \"my_dense\")  model |> compile(optimizer=\"adam\", loss=\"mse\", metrics=c(\"mae\")) model |> fit(matrix(1), matrix(1), verbose = 0) state_tree <- model |> get_state_tree() list(   metrics_variables = list(     loss = list(       count = ...,       total = ...     ),     mean_absolute_error = list(       count = ...,       total = ...     )   ),   trainable_variables = list(     my_sequential = list(       my_dense = list(         bias = ...,         kernel = ...       )     )   ),   non_trainable_variables = list(),   optimizer_variables = list(     adam = list(       iteration = ...,       learning_rate = ...,       my_sequential_my_dense_bias_momentum = ...,       my_sequential_my_dense_bias_velocity = ...,       my_sequential_my_dense_kernel_momentum = ...,       my_sequential_my_dense_kernel_velocity = ...     )   ) ) str(state_tree) ## List of 4 ##  $ trainable_variables    :List of 1 ##   ..$ my_sequential:List of 1 ##   .. ..$ my_dense:List of 2 ##   .. .. ..$ kernel:<tf.Variable 'my_sequential/my_dense/kernel:0' shape=(1, 1) dtype=float32, numpy=array([[-0.8338491]], dtype=float32)> ##   .. .. ..$ bias  :<tf.Variable 'my_sequential/my_dense/bias:0' shape=(1) dtype=float32, numpy=array([0.00099998], dtype=float32)> ##  $ non_trainable_variables: Named list() ##  $ optimizer_variables    :List of 1 ##   ..$ adam:List of 6 ##   .. ..$ iteration                             :<tf.Variable 'adam/iteration:0' shape=() dtype=int64, numpy=1> ##   .. ..$ learning_rate                         :<tf.Variable 'adam/learning_rate:0' shape=() dtype=float32, numpy=0.0010000000474974513> ##   .. ..$ my_sequential_my_dense_kernel_momentum:<tf.Variable 'adam/my_sequential_my_dense_kernel_momentum:0' shape=(1, 1) dtype=float32, numpy=array([[-0.02943518]], dtype=float32)> ##   .. ..$ my_sequential_my_dense_kernel_velocity:<tf.Variable 'adam/my_sequential_my_dense_kernel_velocity:0' shape=(1, 1) dtype=float32, numpy=array([[8.664299e-05]], dtype=float32)> ##   .. ..$ my_sequential_my_dense_bias_momentum  :<tf.Variable 'adam/my_sequential_my_dense_bias_momentum:0' shape=(1) dtype=float32, numpy=array([-0.02943518], dtype=float32)> ##   .. ..$ my_sequential_my_dense_bias_velocity  :<tf.Variable 'adam/my_sequential_my_dense_bias_velocity:0' shape=(1) dtype=float32, numpy=array([8.664299e-05], dtype=float32)> ##  $ metrics_variables      :List of 2 ##   ..$ loss               :List of 2 ##   .. ..$ total:<tf.Variable 'loss/total:0' shape=() dtype=float32, numpy=0.4863377809524536> ##   .. ..$ count:<tf.Variable 'loss/count:0' shape=() dtype=float32, numpy=1.0> ##   ..$ mean_absolute_error:List of 2 ##   .. ..$ total:<tf.Variable 'mean_absolute_error/total:0' shape=() dtype=float32, numpy=0.6973792314529419> ##   .. ..$ count:<tf.Variable 'mean_absolute_error_1/count:0' shape=() dtype=float32, numpy=1.0>"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/get_weights.html","id":null,"dir":"Reference","previous_headings":"","what":"Layer/Model weights as R arrays — get_weights","title":"Layer/Model weights as R arrays — get_weights","text":"Layer/Model weights R arrays","code":""},{"path":"https://keras3.posit.co/dev/reference/get_weights.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Layer/Model weights as R arrays — get_weights","text":"","code":"get_weights(object, trainable = NA)  set_weights(object, weights)"},{"path":"https://keras3.posit.co/dev/reference/get_weights.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Layer/Model weights as R arrays — get_weights","text":"object Layer model object trainable NA (default), weights returned. TRUE, weights trainable variables returned. FALSE, weights non-trainable variables returned. weights Weights R array","code":""},{"path":"https://keras3.posit.co/dev/reference/get_weights.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Layer/Model weights as R arrays — get_weights","text":"list R arrays.","code":""},{"path":"https://keras3.posit.co/dev/reference/get_weights.html","id":"note","dir":"Reference","previous_headings":"","what":"Note","title":"Layer/Model weights as R arrays — get_weights","text":"can access Layer/Model KerasVariables (also backend-native tensors like tf.Variable) object$weights, object$trainable_weights, object$non_trainable_weights","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/grapes-py_class-grapes.html","id":null,"dir":"Reference","previous_headings":"","what":"Make a python class constructor — %py_class%","title":"Make a python class constructor — %py_class%","text":"Make python class constructor","code":""},{"path":"https://keras3.posit.co/dev/reference/grapes-py_class-grapes.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Make a python class constructor — %py_class%","text":"","code":"spec %py_class% body"},{"path":"https://keras3.posit.co/dev/reference/grapes-py_class-grapes.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Make a python class constructor — %py_class%","text":"spec bare symbol MyClassName, call MyClassName(SuperClass) body expression can evaluated construct class methods.","code":""},{"path":"https://keras3.posit.co/dev/reference/grapes-py_class-grapes.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Make a python class constructor — %py_class%","text":"python class constructor, invisibly. Note, constructor also assigned parent frame.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/grapes-py_class-grapes.html","id":"ref-examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Make a python class constructor — %py_class%","text":"","code":"if (FALSE) { # \\dontrun{ MyClass %py_class% {   initialize <- function(x) {     print(\"Hi from MyClass$initialize()!\")     self$x <- x   }   my_method <- function() {     self$x   } }  my_class_instance <- MyClass(42) my_class_instance$my_method()  MyClass2(MyClass) %py_class% {   \"This will be a __doc__ string for MyClass2\"    initialize <- function(...) {     \"This will be the __doc__ string for the MyClass2.__init__() method\"     print(\"Hi from MyClass2$initialize()!\")     super$initialize(...)   } }  my_class_instance2 <- MyClass2(42) my_class_instance2$my_method()  reticulate::py_help(MyClass2) # see the __doc__ strings and more!  # In addition to `self`, there is also `private` available. # This is an R environment unique to each class instance, where you can # store objects that you don't want converted to Python, but still want # available from methods. You can also assign methods to private, and # `self` and `private` will be available in private methods.  MyClass %py_class% {    initialize <- function(x) {     print(\"Hi from MyClass$initialize()!\")     private$y <- paste(\"A Private field:\", x)   }    get_private_field <- function() {     private$y   }    private$a_private_method <- function() {     cat(\"a_private_method() was called.\\n\")     cat(\"private$y is \", sQuote(private$y), \"\\n\")   }    call_private_method <- function()     private$a_private_method()    # equivalent of @property decorator in python   an_active_property %<-active% function(x = NULL) {     if(!is.null(x)) {       cat(\"`an_active_property` was assigned\", x, \"\\n\")       return(x)     } else {       cat(\"`an_active_property` was accessed\\n\")       return(42)     }   } }  inst1 <- MyClass(1) inst2 <- MyClass(2) inst1$get_private_field() inst2$get_private_field() inst1$call_private_method() inst2$call_private_method() inst1$an_active_property inst1$an_active_property <- 11 } # }"},{"path":"https://keras3.posit.co/dev/reference/grapes-set-active-grapes.html","id":null,"dir":"Reference","previous_headings":"","what":"Make an Active Binding — %<-active%","title":"Make an Active Binding — %<-active%","text":"Make Active Binding","code":""},{"path":"https://keras3.posit.co/dev/reference/grapes-set-active-grapes.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Make an Active Binding — %<-active%","text":"","code":"sym %<-active% value"},{"path":"https://keras3.posit.co/dev/reference/grapes-set-active-grapes.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Make an Active Binding — %<-active%","text":"sym symbol bind value function call value sym accessed.","code":""},{"path":"https://keras3.posit.co/dev/reference/grapes-set-active-grapes.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Make an Active Binding — %<-active%","text":"value, invisibly","code":""},{"path":"https://keras3.posit.co/dev/reference/grapes-set-active-grapes.html","id":"details","dir":"Reference","previous_headings":"","what":"Details","title":"Make an Active Binding — %<-active%","text":"Active bindings defined %py_class% converted @property decorated methods.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/grapes-set-active-grapes.html","id":"ref-examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Make an Active Binding — %<-active%","text":"","code":"set.seed(1234) x %<-active% function(value) {   message(\"Evaluating function of active binding\")   if(missing(value))     runif(1)   else    message(\"Received: \", value) } x #> Evaluating function of active binding #> [1] 0.1137034 x #> Evaluating function of active binding #> [1] 0.6222994 x <- \"foo\" #> Evaluating function of active binding #> Received: foo x <- \"foo\" #> Evaluating function of active binding #> Received: foo x #> Evaluating function of active binding #> [1] 0.6092747 rm(x) # cleanup"},{"path":"https://keras3.posit.co/dev/reference/image_array_save.html","id":null,"dir":"Reference","previous_headings":"","what":"Saves an image stored as an array to a path or file object. — image_array_save","title":"Saves an image stored as an array to a path or file object. — image_array_save","text":"Saves image stored array path file object.","code":""},{"path":"https://keras3.posit.co/dev/reference/image_array_save.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Saves an image stored as an array to a path or file object. — image_array_save","text":"","code":"image_array_save(   x,   path,   data_format = NULL,   file_format = NULL,   scale = TRUE,   ... )"},{"path":"https://keras3.posit.co/dev/reference/image_array_save.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Saves an image stored as an array to a path or file object. — image_array_save","text":"x array. path Path file object. data_format Image data format, either \"channels_first\" \"channels_last\". file_format Optional file format override. omitted, format use determined filename extension. file object used instead filename, parameter always used. scale Whether rescale image values within [0, 255]. ... Additional keyword arguments passed PIL.Image.save().","code":""},{"path":"https://keras3.posit.co/dev/reference/image_array_save.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Saves an image stored as an array to a path or file object. — image_array_save","text":"Called primarily side effects. input x returned, invisibly, enable usage pipe.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/image_dataset_from_directory.html","id":null,"dir":"Reference","previous_headings":"","what":"Generates a tf.data.Dataset from image files in a directory. — image_dataset_from_directory","title":"Generates a tf.data.Dataset from image files in a directory. — image_dataset_from_directory","text":"directory structure :   calling image_dataset_from_directory(main_directory, labels = 'inferred') return tf.data.Dataset yields batches images subdirectories class_a class_b, together labels 0 1 (0 corresponding class_a 1 corresponding class_b). Supported image formats: .jpeg, .jpg, .png, .bmp, .gif. Animated gifs truncated first frame.","code":"main_directory/ ...class_a/ ......a_image_1.jpg ......a_image_2.jpg ...class_b/ ......b_image_1.jpg ......b_image_2.jpg"},{"path":"https://keras3.posit.co/dev/reference/image_dataset_from_directory.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Generates a tf.data.Dataset from image files in a directory. — image_dataset_from_directory","text":"","code":"image_dataset_from_directory(   directory,   labels = \"inferred\",   label_mode = \"int\",   class_names = NULL,   color_mode = \"rgb\",   batch_size = 32L,   image_size = c(256L, 256L),   shuffle = TRUE,   seed = NULL,   validation_split = NULL,   subset = NULL,   interpolation = \"bilinear\",   follow_links = FALSE,   crop_to_aspect_ratio = FALSE,   pad_to_aspect_ratio = FALSE,   data_format = NULL,   verbose = TRUE )"},{"path":"https://keras3.posit.co/dev/reference/image_dataset_from_directory.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Generates a tf.data.Dataset from image files in a directory. — image_dataset_from_directory","text":"directory Directory data located. labels \"inferred\", contain subdirectories, containing images class. Otherwise, directory structure ignored. labels Either \"inferred\" (labels generated directory structure), NULL (labels), list/tuple integer labels size number image files found directory. Labels sorted according alphanumeric order image file paths (obtained via os.walk(directory) Python). label_mode String describing encoding labels. Options : \"int\": means labels encoded integers (e.g. sparse_categorical_crossentropy loss). \"categorical\" means labels encoded categorical vector (e.g. categorical_crossentropy loss). \"binary\" means labels (can 2) encoded float32 scalars values 0 1 (e.g. binary_crossentropy). NULL (labels). class_names valid labels \"inferred\". explicit list class names (must match names subdirectories). Used control order classes (otherwise alphanumerical order used). color_mode One \"grayscale\", \"rgb\", \"rgba\". Whether images converted 1, 3, 4 channels. Defaults \"rgb\". batch_size Size batches data. Defaults 32. NULL, data batched (dataset yield individual samples). image_size Size resize images read disk, specified (height, width). Since pipeline processes batches images must size, must provided. Defaults (256, 256). shuffle Whether shuffle data. Defaults TRUE. set FALSE, sorts data alphanumeric order. seed Optional random seed shuffling transformations. validation_split Optional float 0 1, fraction data reserve validation. subset Subset data return. One \"training\", \"validation\", \"\". used validation_split set. subset = \"\", utility returns tuple two datasets (training validation datasets respectively). interpolation String, interpolation method used resizing images. Supports \"bilinear\", \"nearest\", \"bicubic\", \"area\", \"lanczos3\", \"lanczos5\", \"gaussian\", \"mitchellcubic\". Defaults \"bilinear\". follow_links Whether visit subdirectories pointed symlinks. Defaults FALSE. crop_to_aspect_ratio TRUE, resize images without aspect ratio distortion. original aspect ratio differs target aspect ratio, output image cropped return largest possible window image (size image_size) matches target aspect ratio. default (crop_to_aspect_ratio = FALSE), aspect ratio may preserved. pad_to_aspect_ratio TRUE, resize images without aspect ratio distortion. original aspect ratio differs target aspect ratio, output image padded return largest possible window image (size image_size) matches target aspect ratio. default (pad_to_aspect_ratio=FALSE), aspect ratio may preserved. data_format NULL uses config_image_data_format() otherwise either 'channel_last' 'channel_first'. verbose Whether display number information classes number files found. Defaults TRUE.","code":""},{"path":"https://keras3.posit.co/dev/reference/image_dataset_from_directory.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Generates a tf.data.Dataset from image files in a directory. — image_dataset_from_directory","text":"tf.data.Dataset object. label_mode NULL, yields float32 tensors shape (batch_size, image_size[1], image_size[2], num_channels), encoding images (see rules regarding num_channels). Otherwise, yields tuple (images, labels), images shape (batch_size, image_size[1], image_size[2], num_channels), labels follows format described . Rules regarding labels format: label_mode \"int\", labels int32 tensor shape (batch_size,). label_mode \"binary\", labels float32 tensor 1s 0s shape (batch_size, 1). label_mode \"categorical\", labels float32 tensor shape (batch_size, num_classes), representing one-hot encoding class index. Rules regarding number channels yielded images: color_mode \"grayscale\", 1 channel image tensors. color_mode \"rgb\", 3 channels image tensors. color_mode \"rgba\", 4 channels image tensors.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/image_from_array.html","id":null,"dir":"Reference","previous_headings":"","what":"Converts a 3D array to a PIL Image instance. — image_from_array","title":"Converts a 3D array to a PIL Image instance. — image_from_array","text":"Converts 3D array PIL Image instance.","code":""},{"path":"https://keras3.posit.co/dev/reference/image_from_array.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Converts a 3D array to a PIL Image instance. — image_from_array","text":"","code":"image_from_array(x, data_format = NULL, scale = TRUE, dtype = NULL)"},{"path":"https://keras3.posit.co/dev/reference/image_from_array.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Converts a 3D array to a PIL Image instance. — image_from_array","text":"x Input data, form can converted array. data_format Image data format, can either \"channels_first\" \"channels_last\". Defaults NULL, case global setting config_image_data_format() used (unless changed , defaults \"channels_last\"). scale Whether rescale image minimum maximum values 0 255 respectively. Defaults TRUE. dtype Dtype use. NULL means global setting config_floatx() used (unless changed , defaults \"float32\"). Defaults NULL.","code":""},{"path":"https://keras3.posit.co/dev/reference/image_from_array.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Converts a 3D array to a PIL Image instance. — image_from_array","text":"PIL Image instance.","code":""},{"path":"https://keras3.posit.co/dev/reference/image_from_array.html","id":"example","dir":"Reference","previous_headings":"","what":"Example","title":"Converts a 3D array to a PIL Image instance. — image_from_array","text":"","code":"img <- array(runif(30000), dim = c(100, 100, 3)) pil_img <- image_from_array(img) pil_img ## <PIL.Image.Image image mode=RGB size=100x100 at 0x0>"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/image_load.html","id":null,"dir":"Reference","previous_headings":"","what":"Loads an image into PIL format. — image_load","title":"Loads an image into PIL format. — image_load","text":"Loads image PIL format.","code":""},{"path":"https://keras3.posit.co/dev/reference/image_load.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Loads an image into PIL format. — image_load","text":"","code":"image_load(   path,   color_mode = \"rgb\",   target_size = NULL,   interpolation = \"nearest\",   keep_aspect_ratio = FALSE )"},{"path":"https://keras3.posit.co/dev/reference/image_load.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Loads an image into PIL format. — image_load","text":"path Path image file. color_mode One \"grayscale\", \"rgb\", \"rgba\". Default: \"rgb\". desired image format. target_size Either NULL (default original size) tuple ints (img_height, img_width). interpolation Interpolation method used resample image target size different loaded image. Supported methods \"nearest\", \"bilinear\", \"bicubic\". PIL version 1.1.3 newer installed, \"lanczos\" also supported. PIL version 3.4.0 newer installed, \"box\" \"hamming\" also supported. default, \"nearest\" used. keep_aspect_ratio Boolean, whether resize images target size without aspect ratio distortion. image cropped center target aspect ratio resizing.","code":""},{"path":"https://keras3.posit.co/dev/reference/image_load.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Loads an image into PIL format. — image_load","text":"PIL Image instance.","code":""},{"path":"https://keras3.posit.co/dev/reference/image_load.html","id":"example","dir":"Reference","previous_headings":"","what":"Example","title":"Loads an image into PIL format. — image_load","text":"","code":"image_path <- get_file(origin = \"https://www.r-project.org/logo/Rlogo.png\") (image <- image_load(image_path)) ## <PIL.Image.Image image mode=RGB size=724x561 at 0x0> input_arr <- image_to_array(image) str(input_arr) ##  num [1:561, 1:724, 1:3] 0 0 0 0 0 0 0 0 0 0 ... input_arr %<>% array_reshape(dim = c(1, dim(input_arr))) # Convert single image to a batch. model |> predict(input_arr)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/image_smart_resize.html","id":null,"dir":"Reference","previous_headings":"","what":"Resize images to a target size without aspect ratio distortion. — image_smart_resize","title":"Resize images to a target size without aspect ratio distortion. — image_smart_resize","text":"Image datasets typically yield images different size. However, images need batched can processed Keras layers. batched, images need share height width. simply , TF (JAX equivalent):   However, , distort aspect ratio images, since general aspect ratio size. fine many cases, always (e.g. image generation models can problem). Note passing argument preserve_aspect_ratio = TRUE tf$image$resize() preserve aspect ratio, cost longer respecting provided target size. calls :   output images actually (200, 200), distorted. Instead, parts image fit within target size get cropped . resizing process : Take largest centered crop image aspect ratio target size. instance, size = c(200, 200) input image size (340, 500), take crop (340, 340) centered along width. Resize cropped image target size. example , resize (340, 340) crop (200, 200).","code":"size <- c(200, 200) ds <- ds$map(\\(img) tf$image$resize(img, size)) size <- c(200, 200) ds <- ds$map(\\(img) image_smart_resize(img, size))"},{"path":"https://keras3.posit.co/dev/reference/image_smart_resize.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Resize images to a target size without aspect ratio distortion. — image_smart_resize","text":"","code":"image_smart_resize(   x,   size,   interpolation = \"bilinear\",   data_format = \"channels_last\",   backend_module = NULL )"},{"path":"https://keras3.posit.co/dev/reference/image_smart_resize.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Resize images to a target size without aspect ratio distortion. — image_smart_resize","text":"x Input image batch images (tensor array). Must format (height, width, channels) (batch_size, height, width, channels). size Tuple (height, width) integer. Target size. interpolation String, interpolation use resizing. Supports \"bilinear\", \"nearest\", \"bicubic\", \"lanczos3\", \"lanczos5\". Defaults 'bilinear'. data_format \"channels_last\" \"channels_first\". backend_module Backend module use (different default backend).","code":""},{"path":"https://keras3.posit.co/dev/reference/image_smart_resize.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Resize images to a target size without aspect ratio distortion. — image_smart_resize","text":"Array shape (size[1], size[2], channels). input image array, output array, backend-native tensor, output backend-native tensor.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/image_to_array.html","id":null,"dir":"Reference","previous_headings":"","what":"Converts a PIL Image instance to a matrix. — image_to_array","title":"Converts a PIL Image instance to a matrix. — image_to_array","text":"Converts PIL Image instance matrix.","code":""},{"path":"https://keras3.posit.co/dev/reference/image_to_array.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Converts a PIL Image instance to a matrix. — image_to_array","text":"","code":"image_to_array(img, data_format = NULL, dtype = NULL)"},{"path":"https://keras3.posit.co/dev/reference/image_to_array.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Converts a PIL Image instance to a matrix. — image_to_array","text":"img Input PIL Image instance. data_format Image data format, can either \"channels_first\" \"channels_last\". Defaults NULL, case global setting config_image_data_format() used (unless changed , defaults \"channels_last\"). dtype Dtype use. NULL \"double\" return R double array. \"integer\" returns R integer array. values (e.g., \"float32\") return NumPy array.","code":""},{"path":"https://keras3.posit.co/dev/reference/image_to_array.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Converts a PIL Image instance to a matrix. — image_to_array","text":"3D array.","code":""},{"path":"https://keras3.posit.co/dev/reference/image_to_array.html","id":"example","dir":"Reference","previous_headings":"","what":"Example","title":"Converts a PIL Image instance to a matrix. — image_to_array","text":"","code":"image_path <- get_file(origin = \"https://www.r-project.org/logo/Rlogo.png\") (img <- image_load(image_path)) ## <PIL.Image.Image image mode=RGB size=724x561 at 0x0> array <- image_to_array(img) str(array) ##  num [1:561, 1:724, 1:3] 0 0 0 0 0 0 0 0 0 0 ..."},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/imagenet_decode_predictions.html","id":null,"dir":"Reference","previous_headings":"","what":"Decodes the prediction of an ImageNet model. — imagenet_decode_predictions","title":"Decodes the prediction of an ImageNet model. — imagenet_decode_predictions","text":"Decodes prediction ImageNet model.","code":""},{"path":"https://keras3.posit.co/dev/reference/imagenet_decode_predictions.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Decodes the prediction of an ImageNet model. — imagenet_decode_predictions","text":"","code":"imagenet_decode_predictions(preds, top = 5)"},{"path":"https://keras3.posit.co/dev/reference/imagenet_decode_predictions.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Decodes the prediction of an ImageNet model. — imagenet_decode_predictions","text":"preds Tensor encoding batch predictions. top integer, many top-guesses return.","code":""},{"path":"https://keras3.posit.co/dev/reference/imagenet_decode_predictions.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Decodes the prediction of an ImageNet model. — imagenet_decode_predictions","text":"List data frames variables class_name, class_description, score (one data frame per sample batch input).","code":""},{"path":"https://keras3.posit.co/dev/reference/imagenet_preprocess_input.html","id":null,"dir":"Reference","previous_headings":"","what":"Preprocesses a tensor or array encoding a batch of images. — imagenet_preprocess_input","title":"Preprocesses a tensor or array encoding a batch of images. — imagenet_preprocess_input","text":"Preprocesses tensor array encoding batch images.","code":""},{"path":"https://keras3.posit.co/dev/reference/imagenet_preprocess_input.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Preprocesses a tensor or array encoding a batch of images. — imagenet_preprocess_input","text":"","code":"imagenet_preprocess_input(x, data_format = NULL, mode = \"caffe\")"},{"path":"https://keras3.posit.co/dev/reference/imagenet_preprocess_input.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Preprocesses a tensor or array encoding a batch of images. — imagenet_preprocess_input","text":"x Input Numpy symbolic tensor, 3D 4D. data_format Data format image tensor/array. mode One \"caffe\", \"tf\", \"torch\" caffe: convert images RGB BGR, zero-center color channel respect ImageNet dataset, without scaling. tf: scale pixels -1 1, sample-wise. torch: scale pixels 0 1 normalize channel respect ImageNet dataset.","code":""},{"path":"https://keras3.posit.co/dev/reference/imagenet_preprocess_input.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Preprocesses a tensor or array encoding a batch of images. — imagenet_preprocess_input","text":"Preprocessed tensor array.","code":""},{"path":"https://keras3.posit.co/dev/reference/initializer_constant.html","id":null,"dir":"Reference","previous_headings":"","what":"Initializer that generates tensors with constant values. — initializer_constant","title":"Initializer that generates tensors with constant values. — initializer_constant","text":"scalar values allowed. constant value provided must convertible dtype requested calling initializer.","code":""},{"path":"https://keras3.posit.co/dev/reference/initializer_constant.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Initializer that generates tensors with constant values. — initializer_constant","text":"","code":"initializer_constant(value = 0)"},{"path":"https://keras3.posit.co/dev/reference/initializer_constant.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Initializer that generates tensors with constant values. — initializer_constant","text":"value numeric scalar.","code":""},{"path":"https://keras3.posit.co/dev/reference/initializer_constant.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Initializer that generates tensors with constant values. — initializer_constant","text":"Initializer instance can passed layer variable constructors, called directly shape return Tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/initializer_constant.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Initializer that generates tensors with constant values. — initializer_constant","text":"","code":"# Standalone usage: initializer <- initializer_constant(10) values <- initializer(shape = c(2, 2)) # Usage in a Keras layer: initializer <- initializer_constant(10) layer <- layer_dense(units = 3, kernel_initializer = initializer)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/initializer_glorot_normal.html","id":null,"dir":"Reference","previous_headings":"","what":"The Glorot normal initializer, also called Xavier normal initializer. — initializer_glorot_normal","title":"The Glorot normal initializer, also called Xavier normal initializer. — initializer_glorot_normal","text":"Draws samples truncated normal distribution centered 0 stddev = sqrt(2 / (fan_in + fan_out)) fan_in number input units weight tensor fan_out number output units weight tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/initializer_glorot_normal.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"The Glorot normal initializer, also called Xavier normal initializer. — initializer_glorot_normal","text":"","code":"initializer_glorot_normal(seed = NULL)"},{"path":"https://keras3.posit.co/dev/reference/initializer_glorot_normal.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"The Glorot normal initializer, also called Xavier normal initializer. — initializer_glorot_normal","text":"seed integer instance random_seed_generator(). Used make behavior initializer deterministic. Note initializer seeded integer NULL (unseeded) produce random values across multiple calls. get different random values across multiple calls, use seed instance random_seed_generator().","code":""},{"path":"https://keras3.posit.co/dev/reference/initializer_glorot_normal.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"The Glorot normal initializer, also called Xavier normal initializer. — initializer_glorot_normal","text":"Initializer instance can passed layer variable constructors, called directly shape return Tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/initializer_glorot_normal.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"The Glorot normal initializer, also called Xavier normal initializer. — initializer_glorot_normal","text":"","code":"# Standalone usage: initializer <- initializer_glorot_normal() values <- initializer(shape = c(2, 2)) # Usage in a Keras layer: initializer <- initializer_glorot_normal() layer <- layer_dense(units = 3, kernel_initializer = initializer)"},{"path":"https://keras3.posit.co/dev/reference/initializer_glorot_normal.html","id":"reference","dir":"Reference","previous_headings":"","what":"Reference","title":"The Glorot normal initializer, also called Xavier normal initializer. — initializer_glorot_normal","text":"Glorot et al., 2010","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/initializer_glorot_uniform.html","id":null,"dir":"Reference","previous_headings":"","what":"The Glorot uniform initializer, also called Xavier uniform initializer. — initializer_glorot_uniform","title":"The Glorot uniform initializer, also called Xavier uniform initializer. — initializer_glorot_uniform","text":"Draws samples uniform distribution within [-limit, limit], limit = sqrt(6 / (fan_in + fan_out)) (fan_in number input units weight tensor fan_out number output units).","code":""},{"path":"https://keras3.posit.co/dev/reference/initializer_glorot_uniform.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"The Glorot uniform initializer, also called Xavier uniform initializer. — initializer_glorot_uniform","text":"","code":"initializer_glorot_uniform(seed = NULL)"},{"path":"https://keras3.posit.co/dev/reference/initializer_glorot_uniform.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"The Glorot uniform initializer, also called Xavier uniform initializer. — initializer_glorot_uniform","text":"seed integer instance random_seed_generator(). Used make behavior initializer deterministic. Note initializer seeded integer NULL (unseeded) produce random values across multiple calls. get different random values across multiple calls, use seed instance random_seed_generator().","code":""},{"path":"https://keras3.posit.co/dev/reference/initializer_glorot_uniform.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"The Glorot uniform initializer, also called Xavier uniform initializer. — initializer_glorot_uniform","text":"Initializer instance can passed layer variable constructors, called directly shape return Tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/initializer_glorot_uniform.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"The Glorot uniform initializer, also called Xavier uniform initializer. — initializer_glorot_uniform","text":"","code":"# Standalone usage: initializer <- initializer_glorot_uniform() values <- initializer(shape = c(2, 2)) # Usage in a Keras layer: initializer <- initializer_glorot_uniform() layer <- layer_dense(units = 3, kernel_initializer = initializer)"},{"path":"https://keras3.posit.co/dev/reference/initializer_glorot_uniform.html","id":"reference","dir":"Reference","previous_headings":"","what":"Reference","title":"The Glorot uniform initializer, also called Xavier uniform initializer. — initializer_glorot_uniform","text":"Glorot et al., 2010","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/initializer_he_normal.html","id":null,"dir":"Reference","previous_headings":"","what":"He normal initializer. — initializer_he_normal","title":"He normal initializer. — initializer_he_normal","text":"draws samples truncated normal distribution centered 0 stddev = sqrt(2 / fan_in) fan_in number input units weight tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/initializer_he_normal.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"He normal initializer. — initializer_he_normal","text":"","code":"initializer_he_normal(seed = NULL)"},{"path":"https://keras3.posit.co/dev/reference/initializer_he_normal.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"He normal initializer. — initializer_he_normal","text":"seed integer instance random_seed_generator(). Used make behavior initializer deterministic. Note initializer seeded integer NULL (unseeded) produce random values across multiple calls. get different random values across multiple calls, use seed instance random_seed_generator().","code":""},{"path":"https://keras3.posit.co/dev/reference/initializer_he_normal.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"He normal initializer. — initializer_he_normal","text":"Initializer instance can passed layer variable constructors, called directly shape return Tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/initializer_he_normal.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"He normal initializer. — initializer_he_normal","text":"","code":"# Standalone usage: initializer <- initializer_he_normal() values <- initializer(shape = c(2, 2)) # Usage in a Keras layer: initializer <- initializer_he_normal() layer <- layer_dense(units = 3, kernel_initializer = initializer)"},{"path":"https://keras3.posit.co/dev/reference/initializer_he_normal.html","id":"reference","dir":"Reference","previous_headings":"","what":"Reference","title":"He normal initializer. — initializer_he_normal","text":"et al., 2015","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/initializer_he_uniform.html","id":null,"dir":"Reference","previous_headings":"","what":"He uniform variance scaling initializer. — initializer_he_uniform","title":"He uniform variance scaling initializer. — initializer_he_uniform","text":"Draws samples uniform distribution within [-limit, limit], limit = sqrt(6 / fan_in) (fan_in number input units weight tensor).","code":""},{"path":"https://keras3.posit.co/dev/reference/initializer_he_uniform.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"He uniform variance scaling initializer. — initializer_he_uniform","text":"","code":"initializer_he_uniform(seed = NULL)"},{"path":"https://keras3.posit.co/dev/reference/initializer_he_uniform.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"He uniform variance scaling initializer. — initializer_he_uniform","text":"seed integer instance random_seed_generator(). Used make behavior initializer deterministic. Note initializer seeded integer NULL (unseeded) produce random values across multiple calls. get different random values across multiple calls, use seed instance random_seed_generator().","code":""},{"path":"https://keras3.posit.co/dev/reference/initializer_he_uniform.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"He uniform variance scaling initializer. — initializer_he_uniform","text":"Initializer instance can passed layer variable constructors, called directly shape return Tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/initializer_he_uniform.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"He uniform variance scaling initializer. — initializer_he_uniform","text":"","code":"# Standalone usage: initializer <- initializer_he_uniform() values <- initializer(shape = c(2, 2)) # Usage in a Keras layer: initializer <- initializer_he_uniform() layer <- layer_dense(units = 3, kernel_initializer = initializer)"},{"path":"https://keras3.posit.co/dev/reference/initializer_he_uniform.html","id":"reference","dir":"Reference","previous_headings":"","what":"Reference","title":"He uniform variance scaling initializer. — initializer_he_uniform","text":"et al., 2015","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/initializer_identity.html","id":null,"dir":"Reference","previous_headings":"","what":"Initializer that generates the identity matrix. — initializer_identity","title":"Initializer that generates the identity matrix. — initializer_identity","text":"usable generating 2D matrices.","code":""},{"path":"https://keras3.posit.co/dev/reference/initializer_identity.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Initializer that generates the identity matrix. — initializer_identity","text":"","code":"initializer_identity(gain = 1)"},{"path":"https://keras3.posit.co/dev/reference/initializer_identity.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Initializer that generates the identity matrix. — initializer_identity","text":"gain Multiplicative factor apply identity matrix.","code":""},{"path":"https://keras3.posit.co/dev/reference/initializer_identity.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Initializer that generates the identity matrix. — initializer_identity","text":"Initializer instance can passed layer variable constructors, called directly shape return Tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/initializer_identity.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Initializer that generates the identity matrix. — initializer_identity","text":"","code":"# Standalone usage: initializer <- initializer_identity() values <- initializer(shape = c(2, 2)) # Usage in a Keras layer: initializer <- initializer_identity() layer <- layer_dense(units = 3, kernel_initializer = initializer)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/initializer_lecun_normal.html","id":null,"dir":"Reference","previous_headings":"","what":"Lecun normal initializer. — initializer_lecun_normal","title":"Lecun normal initializer. — initializer_lecun_normal","text":"Initializers allow pre-specify initialization strategy, encoded Initializer object, without knowing shape dtype variable initialized. Draws samples truncated normal distribution centered 0 stddev = sqrt(1 / fan_in) fan_in number input units weight tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/initializer_lecun_normal.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Lecun normal initializer. — initializer_lecun_normal","text":"","code":"initializer_lecun_normal(seed = NULL)"},{"path":"https://keras3.posit.co/dev/reference/initializer_lecun_normal.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Lecun normal initializer. — initializer_lecun_normal","text":"seed integer instance random_seed_generator(). Used make behavior initializer deterministic. Note initializer seeded integer NULL (unseeded) produce random values across multiple calls. get different random values across multiple calls, use seed instance random_seed_generator().","code":""},{"path":"https://keras3.posit.co/dev/reference/initializer_lecun_normal.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Lecun normal initializer. — initializer_lecun_normal","text":"Initializer instance can passed layer variable constructors, called directly shape return Tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/initializer_lecun_normal.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Lecun normal initializer. — initializer_lecun_normal","text":"","code":"# Standalone usage: initializer <- initializer_lecun_normal() values <- initializer(shape = c(2, 2)) # Usage in a Keras layer: initializer <- initializer_lecun_normal() layer <- layer_dense(units = 3, kernel_initializer = initializer)"},{"path":"https://keras3.posit.co/dev/reference/initializer_lecun_normal.html","id":"reference","dir":"Reference","previous_headings":"","what":"Reference","title":"Lecun normal initializer. — initializer_lecun_normal","text":"Klambauer et al., 2017","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/initializer_lecun_uniform.html","id":null,"dir":"Reference","previous_headings":"","what":"Lecun uniform initializer. — initializer_lecun_uniform","title":"Lecun uniform initializer. — initializer_lecun_uniform","text":"Draws samples uniform distribution within [-limit, limit], limit = sqrt(3 / fan_in) (fan_in number input units weight tensor).","code":""},{"path":"https://keras3.posit.co/dev/reference/initializer_lecun_uniform.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Lecun uniform initializer. — initializer_lecun_uniform","text":"","code":"initializer_lecun_uniform(seed = NULL)"},{"path":"https://keras3.posit.co/dev/reference/initializer_lecun_uniform.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Lecun uniform initializer. — initializer_lecun_uniform","text":"seed integer instance random_seed_generator(). Used make behavior initializer deterministic. Note initializer seeded integer NULL (unseeded) produce random values across multiple calls. get different random values across multiple calls, use seed instance random_seed_generator().","code":""},{"path":"https://keras3.posit.co/dev/reference/initializer_lecun_uniform.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Lecun uniform initializer. — initializer_lecun_uniform","text":"Initializer instance can passed layer variable constructors, called directly shape return Tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/initializer_lecun_uniform.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Lecun uniform initializer. — initializer_lecun_uniform","text":"","code":"# Standalone usage: initializer <- initializer_lecun_uniform() values <- initializer(shape = c(2, 2)) # Usage in a Keras layer: initializer <- initializer_lecun_uniform() layer <- layer_dense(units = 3, kernel_initializer = initializer)"},{"path":"https://keras3.posit.co/dev/reference/initializer_lecun_uniform.html","id":"reference","dir":"Reference","previous_headings":"","what":"Reference","title":"Lecun uniform initializer. — initializer_lecun_uniform","text":"Klambauer et al., 2017","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/initializer_ones.html","id":null,"dir":"Reference","previous_headings":"","what":"Initializer that generates tensors initialized to 1. — initializer_ones","title":"Initializer that generates tensors initialized to 1. — initializer_ones","text":"Also available via shortcut function ones.","code":""},{"path":"https://keras3.posit.co/dev/reference/initializer_ones.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Initializer that generates tensors initialized to 1. — initializer_ones","text":"","code":"initializer_ones()"},{"path":"https://keras3.posit.co/dev/reference/initializer_ones.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Initializer that generates tensors initialized to 1. — initializer_ones","text":"Initializer instance can passed layer variable constructors, called directly shape return Tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/initializer_ones.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Initializer that generates tensors initialized to 1. — initializer_ones","text":"","code":"# Standalone usage: initializer <- initializer_ones() values <- initializer(shape = c(2, 2)) # Usage in a Keras layer: initializer <- initializer_ones() layer <- layer_dense(units = 3, kernel_initializer = initializer)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/initializer_orthogonal.html","id":null,"dir":"Reference","previous_headings":"","what":"Initializer that generates an orthogonal matrix. — initializer_orthogonal","title":"Initializer that generates an orthogonal matrix. — initializer_orthogonal","text":"shape tensor initialize two-dimensional, initialized orthogonal matrix obtained QR decomposition matrix random numbers drawn normal distribution. matrix fewer rows columns output orthogonal rows. Otherwise, output orthogonal columns. shape tensor initialize two-dimensional, matrix shape (shape[1] * ... * shape[n - 1], shape[n]) initialized, n length shape vector. matrix subsequently reshaped give tensor desired shape.","code":""},{"path":"https://keras3.posit.co/dev/reference/initializer_orthogonal.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Initializer that generates an orthogonal matrix. — initializer_orthogonal","text":"","code":"initializer_orthogonal(gain = 1, seed = NULL)"},{"path":"https://keras3.posit.co/dev/reference/initializer_orthogonal.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Initializer that generates an orthogonal matrix. — initializer_orthogonal","text":"gain Multiplicative factor apply orthogonal matrix. seed integer. Used make behavior initializer deterministic.","code":""},{"path":"https://keras3.posit.co/dev/reference/initializer_orthogonal.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Initializer that generates an orthogonal matrix. — initializer_orthogonal","text":"Initializer instance can passed layer variable constructors, called directly shape return Tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/initializer_orthogonal.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Initializer that generates an orthogonal matrix. — initializer_orthogonal","text":"","code":"# Standalone usage: initializer <- initializer_orthogonal() values <- initializer(shape = c(2, 2)) # Usage in a Keras layer: initializer <- initializer_orthogonal() layer <- layer_dense(units = 3, kernel_initializer = initializer)"},{"path":"https://keras3.posit.co/dev/reference/initializer_orthogonal.html","id":"reference","dir":"Reference","previous_headings":"","what":"Reference","title":"Initializer that generates an orthogonal matrix. — initializer_orthogonal","text":"Saxe et al., 2014","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/initializer_random_normal.html","id":null,"dir":"Reference","previous_headings":"","what":"Random normal initializer. — initializer_random_normal","title":"Random normal initializer. — initializer_random_normal","text":"Draws samples normal distribution given parameters.","code":""},{"path":"https://keras3.posit.co/dev/reference/initializer_random_normal.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Random normal initializer. — initializer_random_normal","text":"","code":"initializer_random_normal(mean = 0, stddev = 0.05, seed = NULL)"},{"path":"https://keras3.posit.co/dev/reference/initializer_random_normal.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Random normal initializer. — initializer_random_normal","text":"mean numeric scalar. Mean random values generate. stddev numeric scalar. Standard deviation random values generate. seed integer instance random_seed_generator(). Used make behavior initializer deterministic. Note initializer seeded integer NULL (unseeded) produce random values across multiple calls. get different random values across multiple calls, use seed instance random_seed_generator().","code":""},{"path":"https://keras3.posit.co/dev/reference/initializer_random_normal.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Random normal initializer. — initializer_random_normal","text":"Initializer instance can passed layer variable constructors, called directly shape return Tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/initializer_random_normal.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Random normal initializer. — initializer_random_normal","text":"","code":"# Standalone usage: initializer <- initializer_random_normal(mean = 0.0, stddev = 1.0) values <- initializer(shape = c(2, 2)) # Usage in a Keras layer: initializer <- initializer_random_normal(mean = 0.0, stddev = 1.0) layer <- layer_dense(units = 3, kernel_initializer = initializer)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/initializer_random_uniform.html","id":null,"dir":"Reference","previous_headings":"","what":"Random uniform initializer. — initializer_random_uniform","title":"Random uniform initializer. — initializer_random_uniform","text":"Draws samples uniform distribution given parameters.","code":""},{"path":"https://keras3.posit.co/dev/reference/initializer_random_uniform.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Random uniform initializer. — initializer_random_uniform","text":"","code":"initializer_random_uniform(minval = -0.05, maxval = 0.05, seed = NULL)"},{"path":"https://keras3.posit.co/dev/reference/initializer_random_uniform.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Random uniform initializer. — initializer_random_uniform","text":"minval numeric scalar scalar keras tensor. Lower bound range random values generate (inclusive). maxval numeric scalar scalar keras tensor. Upper bound range random values generate (exclusive). seed integer instance random_seed_generator(). Used make behavior initializer deterministic. Note initializer seeded integer NULL (unseeded) produce random values across multiple calls. get different random values across multiple calls, use seed instance random_seed_generator().","code":""},{"path":"https://keras3.posit.co/dev/reference/initializer_random_uniform.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Random uniform initializer. — initializer_random_uniform","text":"Initializer instance can passed layer variable constructors, called directly shape return Tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/initializer_random_uniform.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Random uniform initializer. — initializer_random_uniform","text":"","code":"# Standalone usage: initializer <- initializer_random_uniform(minval = 0.0, maxval = 1.0) values <- initializer(shape = c(2, 2)) # Usage in a Keras layer: initializer <- initializer_random_uniform(minval = 0.0, maxval = 1.0) layer <- layer_dense(units = 3, kernel_initializer = initializer)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/initializer_stft.html","id":null,"dir":"Reference","previous_headings":"","what":"Initializer of Conv kernels for Short-term Fourier Transformation (STFT). — initializer_stft","title":"Initializer of Conv kernels for Short-term Fourier Transformation (STFT). — initializer_stft","text":"Since formula involves complex numbers, class compute either real imaginary components final output. Additionally, initializer supports windowing functions across time dimension commonly used STFT. Windowing functions Python module scipy.signal.windows supported, including common hann hamming windowing functions. layer supports periodic windows scaling-based normalization. primarily intended use STFTSpectrogram layer.","code":""},{"path":"https://keras3.posit.co/dev/reference/initializer_stft.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Initializer of Conv kernels for Short-term Fourier Transformation (STFT). — initializer_stft","text":"","code":"initializer_stft(   side = \"real\",   window = \"hann\",   scaling = \"density\",   periodic = FALSE )"},{"path":"https://keras3.posit.co/dev/reference/initializer_stft.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Initializer of Conv kernels for Short-term Fourier Transformation (STFT). — initializer_stft","text":"side String, \"real\" \"imag\" deciding kernel compute real side imaginary side output. Defaults \"real\". window String name windowing function scipy.signal.windows module, array_like window values, NULL windowing. scaling String, \"density\" \"spectrum\" scaling window normalization, either L2 L1 normalization. NULL scaling. periodic Boolean, True, window function treated periodic. Defaults FALSE.","code":""},{"path":"https://keras3.posit.co/dev/reference/initializer_stft.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Initializer of Conv kernels for Short-term Fourier Transformation (STFT). — initializer_stft","text":"Initializer instance can passed layer variable constructors, called directly shape return Tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/initializer_stft.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Initializer of Conv kernels for Short-term Fourier Transformation (STFT). — initializer_stft","text":"","code":"# Standalone usage: initializer <- initializer_stft(\"real\", \"hann\", \"density\", FALSE) values <- initializer(shape = c(128, 1, 513))"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/initializer_truncated_normal.html","id":null,"dir":"Reference","previous_headings":"","what":"Initializer that generates a truncated normal distribution. — initializer_truncated_normal","title":"Initializer that generates a truncated normal distribution. — initializer_truncated_normal","text":"values generated similar values RandomNormal initializer, except values two standard deviations mean discarded re-drawn.","code":""},{"path":"https://keras3.posit.co/dev/reference/initializer_truncated_normal.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Initializer that generates a truncated normal distribution. — initializer_truncated_normal","text":"","code":"initializer_truncated_normal(mean = 0, stddev = 0.05, seed = NULL)"},{"path":"https://keras3.posit.co/dev/reference/initializer_truncated_normal.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Initializer that generates a truncated normal distribution. — initializer_truncated_normal","text":"mean numeric scalar. Mean random values generate. stddev numeric scalar. Standard deviation random values generate. seed integer instance random_seed_generator(). Used make behavior initializer deterministic. Note initializer seeded integer NULL (unseeded) produce random values across multiple calls. get different random values across multiple calls, use seed instance random_seed_generator().","code":""},{"path":"https://keras3.posit.co/dev/reference/initializer_truncated_normal.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Initializer that generates a truncated normal distribution. — initializer_truncated_normal","text":"Initializer instance can passed layer variable constructors, called directly shape return Tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/initializer_truncated_normal.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Initializer that generates a truncated normal distribution. — initializer_truncated_normal","text":"","code":"# Standalone usage: initializer <- initializer_truncated_normal(mean = 0, stddev = 1) values <- initializer(shape = c(2, 2)) # Usage in a Keras layer: initializer <- initializer_truncated_normal(mean = 0, stddev = 1) layer <- layer_dense(units = 3, kernel_initializer = initializer)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/initializer_variance_scaling.html","id":null,"dir":"Reference","previous_headings":"","what":"Initializer that adapts its scale to the shape of its input tensors. — initializer_variance_scaling","title":"Initializer that adapts its scale to the shape of its input tensors. — initializer_variance_scaling","text":"distribution = \"truncated_normal\" \"untruncated_normal\", samples drawn truncated/untruncated normal distribution mean zero standard deviation (truncation, used) stddev = sqrt(scale / n), n : number input units weight tensor, mode = \"fan_in\" number output units, mode = \"fan_out\" average numbers input output units, mode = \"fan_avg\" distribution = \"uniform\", samples drawn uniform distribution within [-limit, limit], limit = sqrt(3 * scale / n).","code":""},{"path":"https://keras3.posit.co/dev/reference/initializer_variance_scaling.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Initializer that adapts its scale to the shape of its input tensors. — initializer_variance_scaling","text":"","code":"initializer_variance_scaling(   scale = 1,   mode = \"fan_in\",   distribution = \"truncated_normal\",   seed = NULL )"},{"path":"https://keras3.posit.co/dev/reference/initializer_variance_scaling.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Initializer that adapts its scale to the shape of its input tensors. — initializer_variance_scaling","text":"scale Scaling factor (positive float). mode One \"fan_in\", \"fan_out\", \"fan_avg\". distribution Random distribution use. One \"truncated_normal\", \"untruncated_normal\", \"uniform\". seed integer instance random_seed_generator(). Used make behavior initializer deterministic. Note initializer seeded integer NULL (unseeded) produce random values across multiple calls. get different random values across multiple calls, use seed instance random_seed_generator().","code":""},{"path":"https://keras3.posit.co/dev/reference/initializer_variance_scaling.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Initializer that adapts its scale to the shape of its input tensors. — initializer_variance_scaling","text":"Initializer instance can passed layer variable constructors, called directly shape return Tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/initializer_variance_scaling.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Initializer that adapts its scale to the shape of its input tensors. — initializer_variance_scaling","text":"","code":"# Standalone usage: initializer <- initializer_variance_scaling(scale = 0.1, mode = 'fan_in',                                             distribution = 'uniform') values <- initializer(shape = c(2, 2)) # Usage in a Keras layer: initializer <- initializer_variance_scaling(scale = 0.1, mode = 'fan_in',                                             distribution = 'uniform') layer <- layer_dense(units = 3, kernel_initializer = initializer)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/initializer_zeros.html","id":null,"dir":"Reference","previous_headings":"","what":"Initializer that generates tensors initialized to 0. — initializer_zeros","title":"Initializer that generates tensors initialized to 0. — initializer_zeros","text":"Initializer generates tensors initialized 0.","code":""},{"path":"https://keras3.posit.co/dev/reference/initializer_zeros.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Initializer that generates tensors initialized to 0. — initializer_zeros","text":"","code":"initializer_zeros()"},{"path":"https://keras3.posit.co/dev/reference/initializer_zeros.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Initializer that generates tensors initialized to 0. — initializer_zeros","text":"Initializer instance can passed layer variable constructors, called directly shape return Tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/initializer_zeros.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Initializer that generates tensors initialized to 0. — initializer_zeros","text":"","code":"# Standalone usage: initializer <- initializer_zeros() values <- initializer(shape = c(2, 2)) # Usage in a Keras layer: initializer <- initializer_zeros() layer <- layer_dense(units = 3, kernel_initializer = initializer)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/install_keras.html","id":null,"dir":"Reference","previous_headings":"","what":"Install Keras — install_keras","title":"Install Keras — install_keras","text":"function install Keras along selected backend, including Python dependencies.","code":""},{"path":"https://keras3.posit.co/dev/reference/install_keras.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Install Keras — install_keras","text":"","code":"install_keras(   envname = \"r-keras\",   ...,   extra_packages = c(\"scipy\", \"pandas\", \"Pillow\", \"pydot\", \"ipython\",     \"tensorflow_datasets\"),   python_version = \">=3.9,<=3.11\",   backend = c(\"tensorflow\", \"jax\"),   gpu = NA,   restart_session = TRUE )"},{"path":"https://keras3.posit.co/dev/reference/install_keras.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Install Keras — install_keras","text":"envname Name path Python virtual environment ... reserved future compatibility. extra_packages Additional Python packages install alongside Keras python_version Passed reticulate::virtualenv_starter() backend backend(s) install. Accepted values include \"tensorflow\", \"jax\" \"torch\" gpu whether install GPU capable version backend. restart_session Whether restart R session installing (note occur within RStudio).","code":""},{"path":"https://keras3.posit.co/dev/reference/install_keras.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Install Keras — install_keras","text":"return value, called side effects.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/keras.html","id":null,"dir":"Reference","previous_headings":"","what":"Main Keras module — keras","title":"Main Keras module — keras","text":"keras module object equivalent reticulate::import(\"keras\") provided mainly convenience.","code":""},{"path":"https://keras3.posit.co/dev/reference/keras.html","id":"format","dir":"Reference","previous_headings":"","what":"Format","title":"Main Keras module — keras","text":"object class python.builtin.module","code":""},{"path":"https://keras3.posit.co/dev/reference/keras.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Main Keras module — keras","text":"keras Python module","code":""},{"path":"https://keras3.posit.co/dev/reference/keras3-package.html","id":null,"dir":"Reference","previous_headings":"","what":"keras3: R Interface to 'Keras' — keras3-package","title":"keras3: R Interface to 'Keras' — keras3-package","text":"Interface 'Keras' https://keras.io, high-level neural networks API. 'Keras' developed focus enabling fast experimentation, supports convolution based networks recurrent networks (well combinations two), runs seamlessly CPU GPU devices. Keras high-level neural networks API, developed focus enabling fast experimentation. Keras following key features:","code":""},{"path":"https://keras3.posit.co/dev/reference/keras3-package.html","id":"details","dir":"Reference","previous_headings":"","what":"Details","title":"keras3: R Interface to 'Keras' — keras3-package","text":"Allows code run CPU GPU, seamlessly. User-friendly API makes easy quickly prototype deep learning models. Built-support convolutional networks (computer vision), recurrent networks (sequence processing), combination . Supports arbitrary network architectures: multi-input multi-output models, layer sharing, model sharing, etc. means Keras appropriate building essentially deep learning model, memory network neural Turing machine. capable running top multiple back-ends including TensorFlow, Jax, PyTorch. See package website https://keras3.posit.co complete documentation.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/keras3-package.html","id":"author","dir":"Reference","previous_headings":"","what":"Author","title":"keras3: R Interface to 'Keras' — keras3-package","text":"Maintainer: Tomasz Kalinowski tomasz@posit.co [copyright holder] Authors: JJ Allaire [copyright holder] François Chollet [copyright holder] contributors: Daniel Falbel daniel@posit.co [contributor, copyright holder] Posit Software, PBC [copyright holder, funder] Google [copyright holder, funder] Yuan Tang terrytangyuan@gmail.com (ORCID) [contributor, copyright holder] Wouter Van Der Bijl [contributor, copyright holder] Martin Studer [contributor, copyright holder] Sigrid Keydana [contributor]","code":""},{"path":"https://keras3.posit.co/dev/reference/keras_input.html","id":null,"dir":"Reference","previous_headings":"","what":"Create a Keras tensor (Functional API input). — keras_input","title":"Create a Keras tensor (Functional API input). — keras_input","text":"Keras tensor symbolic tensor-like object, augment certain attributes allow us build Keras model just knowing inputs outputs model. instance, , b c Keras tensors, becomes possible : model <- keras_model(input = c(, b), output = c)","code":""},{"path":"https://keras3.posit.co/dev/reference/keras_input.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Create a Keras tensor (Functional API input). — keras_input","text":"","code":"keras_input(   shape = NULL,   batch_size = NULL,   dtype = NULL,   sparse = NULL,   ragged = NULL,   batch_shape = NULL,   name = NULL,   tensor = NULL,   optional = FALSE )"},{"path":"https://keras3.posit.co/dev/reference/keras_input.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Create a Keras tensor (Functional API input). — keras_input","text":"shape shape list (list integers NULL objects), including batch size. instance, shape = c(32) indicates expected input batches 32-dimensional vectors. Elements list can NULL NA; NULL/NA elements represent dimensions shape known may vary (e.g. sequence length). batch_size Optional static batch size (integer). dtype data type expected input, string (e.g. \"float32\", \"int32\"...) sparse boolean specifying whether expected input sparse tensors. Note , sparse FALSE, sparse tensors can still passed input - densified default value 0. feature supported TensorFlow backend. Defaults FALSE. ragged boolean specifying whether expected input ragged tensors. Note , ragged FALSE, ragged tensors can still passed input - densified default value 0. feature supported TensorFlow backend. Defaults FALSE. batch_shape Optional shape list (list integers NULL objects), including batch size. name Optional name string layer. unique model (reuse name twice). autogenerated provided. tensor Optional existing tensor wrap Input layer. set, layer use tensor rather creating new placeholder tensor. optional Boolean, whether input optional . optional input can accept NULL values.","code":""},{"path":"https://keras3.posit.co/dev/reference/keras_input.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Create a Keras tensor (Functional API input). — keras_input","text":"Keras tensor, can passed inputs argument (keras_model()).","code":""},{"path":"https://keras3.posit.co/dev/reference/keras_input.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Create a Keras tensor (Functional API input). — keras_input","text":"","code":"# This is a logistic regression in Keras input <- layer_input(shape=c(32)) output <- input |> layer_dense(16, activation='softmax') model <- keras_model(input, output)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/keras_model.html","id":null,"dir":"Reference","previous_headings":"","what":"Keras Model (Functional API) — keras_model","title":"Keras Model (Functional API) — keras_model","text":"model directed acyclic graph layers.","code":""},{"path":"https://keras3.posit.co/dev/reference/keras_model.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Keras Model (Functional API) — keras_model","text":"","code":"keras_model(inputs = NULL, outputs = NULL, ...)"},{"path":"https://keras3.posit.co/dev/reference/keras_model.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Keras Model (Functional API) — keras_model","text":"inputs Input tensor(s) (keras_input()) outputs Output tensors (calling layers inputs) ... additional arguments","code":""},{"path":"https://keras3.posit.co/dev/reference/keras_model.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Keras Model (Functional API) — keras_model","text":"Model instance.","code":""},{"path":"https://keras3.posit.co/dev/reference/keras_model.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Keras Model (Functional API) — keras_model","text":"","code":"library(keras3)  # input tensor inputs <- keras_input(shape = c(784))  # outputs compose input + dense layers predictions <- inputs |>   layer_dense(units = 64, activation = 'relu') |>   layer_dense(units = 64, activation = 'relu') |>   layer_dense(units = 10, activation = 'softmax')  # create and compile model model <- keras_model(inputs = inputs, outputs = predictions) model |> compile(   optimizer = 'rmsprop',   loss = 'categorical_crossentropy',   metrics = c('accuracy') )"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/keras_model_sequential.html","id":null,"dir":"Reference","previous_headings":"","what":"Keras Model composed of a linear stack of layers — keras_model_sequential","title":"Keras Model composed of a linear stack of layers — keras_model_sequential","text":"Keras Model composed linear stack layers","code":""},{"path":"https://keras3.posit.co/dev/reference/keras_model_sequential.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Keras Model composed of a linear stack of layers — keras_model_sequential","text":"","code":"keras_model_sequential(   input_shape = NULL,   name = NULL,   ...,   input_dtype = NULL,   input_batch_size = NULL,   input_sparse = NULL,   input_ragged = NULL,   input_batch_shape = NULL,   input_name = NULL,   input_tensor = NULL,   input_optional = FALSE,   trainable = TRUE,   layers = list() )"},{"path":"https://keras3.posit.co/dev/reference/keras_model_sequential.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Keras Model composed of a linear stack of layers — keras_model_sequential","text":"input_shape shape integer vector, including batch size. instance, shape=c(32) indicates expected input batches 32-dimensional vectors. Elements shape can NA; NA elements represent dimensions shape known may vary (e.g. sequence length). name Name model ... additional arguments passed keras.layers.InputLayer. input_dtype data type expected input, string (e.g. \"float32\", \"int32\"...) input_batch_size Optional static batch size (integer). input_sparse boolean specifying whether expected input sparse tensors. Note , sparse FALSE, sparse tensors can still passed input - densified default value 0. feature supported TensorFlow backend. Defaults FALSE. input_ragged bool. input_batch_shape optional way specify batch_size input_shape one argument. input_name Optional name string input layer. unique model (reuse name twice). autogenerated provided. input_tensor Optional existing tensor wrap InputLayer. set, layer use tensor rather creating new placeholder tensor. input_optional Boolean, whether input optional . optional input can accept NULL values. trainable Boolean, whether model's variables trainable. can also change trainable status model/layer freeze_weights() unfreeze_weights(). layers List layers add model.","code":""},{"path":"https://keras3.posit.co/dev/reference/keras_model_sequential.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Keras Model composed of a linear stack of layers — keras_model_sequential","text":"Sequential model instance.","code":""},{"path":"https://keras3.posit.co/dev/reference/keras_model_sequential.html","id":"note","dir":"Reference","previous_headings":"","what":"Note","title":"Keras Model composed of a linear stack of layers — keras_model_sequential","text":"input_shape omitted, model layer shapes, including final model output shape, known model built, either calling model input tensor/array like model(input), (possibly via fit()/evaluate()/predict()), explicitly calling model$build(input_shape).","code":""},{"path":"https://keras3.posit.co/dev/reference/keras_model_sequential.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Keras Model composed of a linear stack of layers — keras_model_sequential","text":"","code":"model <- keras_model_sequential(input_shape = c(784)) model |>   layer_dense(units = 32) |>   layer_activation('relu') |>   layer_dense(units = 10) |>   layer_activation('softmax')  model |> compile(   optimizer = 'rmsprop',   loss = 'categorical_crossentropy',   metrics = c('accuracy') )  model ## Model: \"sequential\" ## +---------------------------------+------------------------+---------------+ ## | Layer (type)                    | Output Shape           |       Param # | ## +=================================+========================+===============+ ## | dense (Dense)                   | (None, 32)             |        25,120 | ## +---------------------------------+------------------------+---------------+ ## | activation (Activation)         | (None, 32)             |             0 | ## +---------------------------------+------------------------+---------------+ ## | dense_1 (Dense)                 | (None, 10)             |           330 | ## +---------------------------------+------------------------+---------------+ ## | activation_1 (Activation)       | (None, 10)             |             0 | ## +---------------------------------+------------------------+---------------+ ##  Total params: 25,450 (99.41 KB) ##  Trainable params: 25,450 (99.41 KB) ##  Non-trainable params: 0 (0.00 B)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/keras_variable.html","id":null,"dir":"Reference","previous_headings":"","what":"Represents a backend-agnostic variable in Keras. — keras_variable","title":"Represents a backend-agnostic variable in Keras. — keras_variable","text":"Variable acts container state. holds tensor value can updated. JAX backend, variables used implement \"functionalization\", pattern lifting stateful operations piece computation turn stateless function.","code":""},{"path":"https://keras3.posit.co/dev/reference/keras_variable.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Represents a backend-agnostic variable in Keras. — keras_variable","text":"","code":"keras_variable(   initializer,   shape = NULL,   dtype = NULL,   trainable = TRUE,   autocast = TRUE,   aggregation = \"none\",   name = NULL )"},{"path":"https://keras3.posit.co/dev/reference/keras_variable.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Represents a backend-agnostic variable in Keras. — keras_variable","text":"initializer Initial value callable initialization. callable used, take arguments shape dtype. shape Optional. Tuple variable's shape. Required initializer callable. dtype Optional. Data type variable. Defaults global float dtype type (\"float32\" never configured). trainable Optional. Boolean indicating variable trainable. Defaults TRUE. autocast Optional. Boolean indicating whether variable supports autocasting. TRUE, layer may first convert variable compute data type accessed. Defaults TRUE. aggregation Optional string, one NULL, \"none\", \"mean\", \"sum\" \"only_first_replica\" specifying distributed variable aggregated. serves semantic annotation, taken account downstream backends users. Defaults \"none\". name Optional. unique name variable. Automatically generated set.","code":""},{"path":"https://keras3.posit.co/dev/reference/keras_variable.html","id":"attributes","dir":"Reference","previous_headings":"","what":"Attributes","title":"Represents a backend-agnostic variable in Keras. — keras_variable","text":"shape: shape variable (tuple integers). ndim: number dimensions variable (integer). dtype: data type variable (string). trainable: Whether variable trainable (boolean). autocast: Whether variable supports autocasting (boolean). aggregation: distributed variable aggregated (string). value: current value variable (NumPy array tensor). name: name variable (string). path: path variable within Keras model layer (string).","code":""},{"path":"https://keras3.posit.co/dev/reference/keras_variable.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Represents a backend-agnostic variable in Keras. — keras_variable","text":"Initializing Variable NumPy array:   Using Keras initializer create Variable:           Marking Variable non-trainable:","code":"initial_array <- array(1, c(3, 3)) variable_from_array <- keras_variable(initializer = initial_array) variable_from_initializer <- keras_variable(   initializer = initializer_ones(),   shape = c(3, 3),   dtype = \"float32\" ) new_value <- array(0, c(3, 3)) variable_from_array$assign(new_value) ## tf.Tensor( ## [[0. 0. 0.] ##  [0. 0. 0.] ##  [0. 0. 0.]], shape=(3, 3), dtype=float64) # To modify a subset of values value <- variable_from_array$value value@r[1,] <- 99 invisible(variable_from_array$assign(value)) variable_from_array ## <Variable path=variable, shape=(3, 3), dtype=float64, value=[[99. 99. 99.] ##  [ 0.  0.  0.] ##  [ 0.  0.  0.]]> non_trainable_variable <- keras_variable(   initializer = array(1, c(3, 3)),   dtype = \"float32\",   trainable = FALSE )"},{"path":"https://keras3.posit.co/dev/reference/layer_activation.html","id":null,"dir":"Reference","previous_headings":"","what":"Applies an activation function to an output. — layer_activation","title":"Applies an activation function to an output. — layer_activation","text":"Applies activation function output.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_activation.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Applies an activation function to an output. — layer_activation","text":"","code":"layer_activation(object, activation, ...)"},{"path":"https://keras3.posit.co/dev/reference/layer_activation.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Applies an activation function to an output. — layer_activation","text":"object Object compose layer . tensor, array, sequential model. activation Activation function. callable, name activation keras3::activation_* namespace. ... Base layer keyword arguments, name dtype.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_activation.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Applies an activation function to an output. — layer_activation","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_activation.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Applies an activation function to an output. — layer_activation","text":"","code":"x <- array(c(-3, -1, 0, 2)) layer <- layer_activation(activation = 'relu') layer(x) ## tf.Tensor([0. 0. 0. 2.], shape=(4), dtype=float32) layer <- layer_activation(activation = activation_relu) layer(x) ## tf.Tensor([0. 0. 0. 2.], shape=(4), dtype=float32) layer <- layer_activation(activation = op_relu) layer(x) ## tf.Tensor([0. 0. 0. 2.], shape=(4), dtype=float32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_activation_elu.html","id":null,"dir":"Reference","previous_headings":"","what":"Applies an Exponential Linear Unit function to an output. — layer_activation_elu","title":"Applies an Exponential Linear Unit function to an output. — layer_activation_elu","text":"Formula:","code":"f(x) = alpha * (exp(x) - 1.) for x < 0 f(x) = x for x >= 0"},{"path":"https://keras3.posit.co/dev/reference/layer_activation_elu.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Applies an Exponential Linear Unit function to an output. — layer_activation_elu","text":"","code":"layer_activation_elu(object, alpha = 1, ...)"},{"path":"https://keras3.posit.co/dev/reference/layer_activation_elu.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Applies an Exponential Linear Unit function to an output. — layer_activation_elu","text":"object Object compose layer . tensor, array, sequential model. alpha float, slope negative section. Defaults 1.0. ... Base layer keyword arguments, name dtype.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_activation_elu.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Applies an Exponential Linear Unit function to an output. — layer_activation_elu","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_activation_leaky_relu.html","id":null,"dir":"Reference","previous_headings":"","what":"Leaky version of a Rectified Linear Unit activation layer. — layer_activation_leaky_relu","title":"Leaky version of a Rectified Linear Unit activation layer. — layer_activation_leaky_relu","text":"layer allows small gradient unit active. Formula:","code":"f <- function(x) ifelse(x >= 0, x, alpha * x)"},{"path":"https://keras3.posit.co/dev/reference/layer_activation_leaky_relu.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Leaky version of a Rectified Linear Unit activation layer. — layer_activation_leaky_relu","text":"","code":"layer_activation_leaky_relu(object, negative_slope = 0.3, ...)"},{"path":"https://keras3.posit.co/dev/reference/layer_activation_leaky_relu.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Leaky version of a Rectified Linear Unit activation layer. — layer_activation_leaky_relu","text":"object Object compose layer . tensor, array, sequential model. negative_slope Float >= 0.0. Negative slope coefficient. Defaults 0.3. ... Base layer keyword arguments, name dtype.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_activation_leaky_relu.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Leaky version of a Rectified Linear Unit activation layer. — layer_activation_leaky_relu","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_activation_leaky_relu.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Leaky version of a Rectified Linear Unit activation layer. — layer_activation_leaky_relu","text":"","code":"leaky_relu_layer <- layer_activation_leaky_relu(negative_slope=0.5) input <- array(c(-10, -5, 0.0, 5, 10)) result <- leaky_relu_layer(input) as.array(result) ## [1] -5.0 -2.5  0.0  5.0 10.0"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_activation_parametric_relu.html","id":null,"dir":"Reference","previous_headings":"","what":"Parametric Rectified Linear Unit activation layer. — layer_activation_parametric_relu","title":"Parametric Rectified Linear Unit activation layer. — layer_activation_parametric_relu","text":"Formula:   alpha learned array shape x.","code":"f <- function(x) ifelse(x >= 0, x, alpha * x)"},{"path":"https://keras3.posit.co/dev/reference/layer_activation_parametric_relu.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Parametric Rectified Linear Unit activation layer. — layer_activation_parametric_relu","text":"","code":"layer_activation_parametric_relu(   object,   alpha_initializer = \"Zeros\",   alpha_regularizer = NULL,   alpha_constraint = NULL,   shared_axes = NULL,   ... )"},{"path":"https://keras3.posit.co/dev/reference/layer_activation_parametric_relu.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Parametric Rectified Linear Unit activation layer. — layer_activation_parametric_relu","text":"object Object compose layer . tensor, array, sequential model. alpha_initializer Initializer function weights. alpha_regularizer Regularizer weights. alpha_constraint Constraint weights. shared_axes axes along share learnable parameters activation function. example, incoming feature maps 2D convolution output shape (batch, height, width, channels), wish share parameters across space filter one set parameters, set shared_axes=[1, 2]. ... Base layer keyword arguments, name dtype.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_activation_parametric_relu.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Parametric Rectified Linear Unit activation layer. — layer_activation_parametric_relu","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_activation_relu.html","id":null,"dir":"Reference","previous_headings":"","what":"Rectified Linear Unit activation function layer. — layer_activation_relu","title":"Rectified Linear Unit activation function layer. — layer_activation_relu","text":"Formula:","code":"f <- function(x, max_value = Inf, negative_slope = 0, threshold = 0) {  x <- max(x,0)  if (x >= max_value)    max_value  else if (threshold <= x && x < max_value)    x  else    negative_slope * (x - threshold) }"},{"path":"https://keras3.posit.co/dev/reference/layer_activation_relu.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Rectified Linear Unit activation function layer. — layer_activation_relu","text":"","code":"layer_activation_relu(   object,   max_value = NULL,   negative_slope = 0,   threshold = 0,   ... )"},{"path":"https://keras3.posit.co/dev/reference/layer_activation_relu.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Rectified Linear Unit activation function layer. — layer_activation_relu","text":"object Object compose layer . tensor, array, sequential model. max_value Float >= 0. Maximum activation value. NULL means unlimited. Defaults NULL. negative_slope Float >= 0. Negative slope coefficient. Defaults 0.0. threshold Float >= 0. Threshold value thresholded activation. Defaults 0.0. ... Base layer keyword arguments, name dtype.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_activation_relu.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Rectified Linear Unit activation function layer. — layer_activation_relu","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_activation_relu.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Rectified Linear Unit activation function layer. — layer_activation_relu","text":"","code":"relu_layer <- layer_activation_relu(max_value = 10,                                     negative_slope = 0.5,                                     threshold = 0) input <- array(c(-10, -5, 0.0, 5, 10)) result <- relu_layer(input) as.array(result) ## [1] -5.0 -2.5  0.0  5.0 10.0"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_activation_softmax.html","id":null,"dir":"Reference","previous_headings":"","what":"Softmax activation layer. — layer_activation_softmax","title":"Softmax activation layer. — layer_activation_softmax","text":"Formula:","code":"exp_x = exp(x - max(x)) f(x) = exp_x / sum(exp_x)"},{"path":"https://keras3.posit.co/dev/reference/layer_activation_softmax.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Softmax activation layer. — layer_activation_softmax","text":"","code":"layer_activation_softmax(object, axis = -1L, ...)"},{"path":"https://keras3.posit.co/dev/reference/layer_activation_softmax.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Softmax activation layer. — layer_activation_softmax","text":"object Object compose layer . tensor, array, sequential model. axis Integer, list Integers, axis along softmax normalization applied. ... Base layer keyword arguments, name dtype.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_activation_softmax.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Softmax activation layer. — layer_activation_softmax","text":"Softmaxed output shape inputs.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_activation_softmax.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Softmax activation layer. — layer_activation_softmax","text":"","code":"softmax_layer <- layer_activation_softmax() input <- op_array(c(1, 2, 1)) softmax_layer(input) ## tf.Tensor([0.21194157 0.5761169  0.21194157], shape=(3), dtype=float32)"},{"path":"https://keras3.posit.co/dev/reference/layer_activation_softmax.html","id":"call-arguments","dir":"Reference","previous_headings":"","what":"Call Arguments","title":"Softmax activation layer. — layer_activation_softmax","text":"inputs: inputs (logits) softmax layer. mask: boolean mask shape inputs. mask specifies 1 keep 0 mask. Defaults NULL.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_activity_regularization.html","id":null,"dir":"Reference","previous_headings":"","what":"Layer that applies an update to the cost function based input activity. — layer_activity_regularization","title":"Layer that applies an update to the cost function based input activity. — layer_activity_regularization","text":"Layer applies update cost function based input activity.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_activity_regularization.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Layer that applies an update to the cost function based input activity. — layer_activity_regularization","text":"","code":"layer_activity_regularization(object, l1 = 0, l2 = 0, ...)"},{"path":"https://keras3.posit.co/dev/reference/layer_activity_regularization.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Layer that applies an update to the cost function based input activity. — layer_activity_regularization","text":"object Object compose layer . tensor, array, sequential model. l1 L1 regularization factor (positive float). l2 L2 regularization factor (positive float). ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_activity_regularization.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Layer that applies an update to the cost function based input activity. — layer_activity_regularization","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_activity_regularization.html","id":"input-shape","dir":"Reference","previous_headings":"","what":"Input Shape","title":"Layer that applies an update to the cost function based input activity. — layer_activity_regularization","text":"Arbitrary. Use keyword argument input_shape (tuple integers, include samples axis) using layer first layer model.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_activity_regularization.html","id":"output-shape","dir":"Reference","previous_headings":"","what":"Output Shape","title":"Layer that applies an update to the cost function based input activity. — layer_activity_regularization","text":"shape input.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_add.html","id":null,"dir":"Reference","previous_headings":"","what":"Performs elementwise addition operation. — layer_add","title":"Performs elementwise addition operation. — layer_add","text":"takes input list tensors, shape, returns single tensor (also shape).","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_add.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Performs elementwise addition operation. — layer_add","text":"","code":"layer_add(inputs, ...)"},{"path":"https://keras3.posit.co/dev/reference/layer_add.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Performs elementwise addition operation. — layer_add","text":"inputs layers combine ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_add.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Performs elementwise addition operation. — layer_add","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_add.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Performs elementwise addition operation. — layer_add","text":"Usage Keras model:","code":"input_shape <- c(1, 2, 3) x1 <- op_ones(input_shape) x2 <- op_ones(input_shape) layer_add(x1, x2) ## tf.Tensor( ## [[[2. 2. 2.] ##   [2. 2. 2.]]], shape=(1, 2, 3), dtype=float32) input1 <- layer_input(shape = c(16)) x1 <- input1 |> layer_dense(8, activation = 'relu')  input2 <- layer_input(shape = c(32)) x2 <- input2 |> layer_dense(8, activation = 'relu')  # equivalent to `added = layer_add([x1, x2))` added <- layer_add(x1, x2) output <- added |> layer_dense(4)  model <- keras_model(inputs = c(input1, input2), outputs = output)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_additive_attention.html","id":null,"dir":"Reference","previous_headings":"","what":"Additive attention layer, a.k.a. Bahdanau-style attention. — layer_additive_attention","title":"Additive attention layer, a.k.a. Bahdanau-style attention. — layer_additive_attention","text":"Inputs list 2 3 elements: query tensor shape (batch_size, Tq, dim). value tensor shape (batch_size, Tv, dim). optional key tensor shape (batch_size, Tv, dim). none supplied, value used key. calculation follows steps: Calculate attention scores using query key shape (batch_size, Tq, Tv) non-linear sum scores = reduce_sum(tanh(query + key), axis=-1). Use scores calculate softmax distribution shape (batch_size, Tq, Tv). Use softmax distribution create linear combination value shape (batch_size, Tq, dim).","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_additive_attention.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Additive attention layer, a.k.a. Bahdanau-style attention. — layer_additive_attention","text":"","code":"layer_additive_attention(object, use_scale = TRUE, dropout = 0, ...)"},{"path":"https://keras3.posit.co/dev/reference/layer_additive_attention.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Additive attention layer, a.k.a. Bahdanau-style attention. — layer_additive_attention","text":"object Object compose layer . tensor, array, sequential model. use_scale TRUE, create scalar variable scale attention scores. dropout Float 0 1. Fraction units drop attention scores. Defaults 0.0. ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_additive_attention.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Additive attention layer, a.k.a. Bahdanau-style attention. — layer_additive_attention","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_additive_attention.html","id":"call-arguments","dir":"Reference","previous_headings":"","what":"Call Arguments","title":"Additive attention layer, a.k.a. Bahdanau-style attention. — layer_additive_attention","text":"inputs: List following tensors: query: Query tensor shape (batch_size, Tq, dim). value: Value tensor shape (batch_size, Tv, dim). key: Optional key tensor shape (batch_size, Tv, dim). given, use value key value, common case. mask: List following tensors: query_mask: boolean mask tensor shape (batch_size, Tq). given, output zero positions mask==FALSE. value_mask: boolean mask tensor shape (batch_size, Tv). given, apply mask values positions mask==FALSE contribute result. return_attention_scores: bool, TRUE, returns attention scores (masking softmax) additional output argument. training: Python boolean indicating whether layer behave training mode (adding dropout) inference mode (dropout). use_causal_mask: Boolean. Set TRUE decoder self-attention. Adds mask position attend positions j > . prevents flow information future towards past. Defaults FALSE.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_additive_attention.html","id":"output","dir":"Reference","previous_headings":"","what":"Output","title":"Additive attention layer, a.k.a. Bahdanau-style attention. — layer_additive_attention","text":"Attention outputs shape (batch_size, Tq, dim). (Optional) Attention scores masking softmax shape (batch_size, Tq, Tv).","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_alpha_dropout.html","id":null,"dir":"Reference","previous_headings":"","what":"Applies Alpha Dropout to the input. — layer_alpha_dropout","title":"Applies Alpha Dropout to the input. — layer_alpha_dropout","text":"Alpha Dropout Dropout keeps mean variance inputs original values, order ensure self-normalizing property even dropout. Alpha Dropout fits well Scaled Exponential Linear Units (SELU) randomly setting activations negative saturation value.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_alpha_dropout.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Applies Alpha Dropout to the input. — layer_alpha_dropout","text":"","code":"layer_alpha_dropout(object, rate, noise_shape = NULL, seed = NULL, ...)"},{"path":"https://keras3.posit.co/dev/reference/layer_alpha_dropout.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Applies Alpha Dropout to the input. — layer_alpha_dropout","text":"object Object compose layer . tensor, array, sequential model. rate Float 0 1. multiplicative noise standard deviation sqrt(rate / (1 - rate)). noise_shape 1D integer tensor representing shape binary alpha dropout mask multiplied input. instance, inputs shape (batch_size, timesteps, features) want alpha dropout mask timesteps, can use noise_shape = (batch_size, 1, features). seed integer use random seed. ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_alpha_dropout.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Applies Alpha Dropout to the input. — layer_alpha_dropout","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_alpha_dropout.html","id":"call-arguments","dir":"Reference","previous_headings":"","what":"Call Arguments","title":"Applies Alpha Dropout to the input. — layer_alpha_dropout","text":"inputs: Input tensor (rank). training: R boolean indicating whether layer behave training mode (adding alpha dropout) inference mode (nothing).","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_attention.html","id":null,"dir":"Reference","previous_headings":"","what":"Dot-product attention layer, a.k.a. Luong-style attention. — layer_attention","title":"Dot-product attention layer, a.k.a. Luong-style attention. — layer_attention","text":"Inputs list 2 3 elements: query tensor shape (batch_size, Tq, dim). value tensor shape (batch_size, Tv, dim). optional key tensor shape (batch_size, Tv, dim). none supplied, value used key. calculation follows steps: Calculate attention scores using query key shape (batch_size, Tq, Tv). Use scores calculate softmax distribution shape (batch_size, Tq, Tv). Use softmax distribution create linear combination value shape (batch_size, Tq, dim).","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_attention.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Dot-product attention layer, a.k.a. Luong-style attention. — layer_attention","text":"","code":"layer_attention(   object,   use_scale = FALSE,   score_mode = \"dot\",   dropout = 0,   seed = NULL,   ... )"},{"path":"https://keras3.posit.co/dev/reference/layer_attention.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Dot-product attention layer, a.k.a. Luong-style attention. — layer_attention","text":"object Object compose layer . tensor, array, sequential model. use_scale TRUE, create scalar variable scale attention scores. score_mode Function use compute attention scores, one {\"dot\", \"concat\"}. \"dot\" refers dot product query key vectors. \"concat\" refers hyperbolic tangent concatenation query key vectors. dropout Float 0 1. Fraction units drop attention scores. Defaults 0.0. seed integer use random seed case dropout. ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_attention.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Dot-product attention layer, a.k.a. Luong-style attention. — layer_attention","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_attention.html","id":"call-arguments","dir":"Reference","previous_headings":"","what":"Call Arguments","title":"Dot-product attention layer, a.k.a. Luong-style attention. — layer_attention","text":"inputs: List following tensors: query: Query tensor shape (batch_size, Tq, dim). value: Value tensor shape (batch_size, Tv, dim). key: Optional key tensor shape (batch_size, Tv, dim). given, use value key value, common case. mask: List following tensors: query_mask: boolean mask tensor shape (batch_size, Tq). given, output zero positions mask==FALSE. value_mask: boolean mask tensor shape (batch_size, Tv). given, apply mask values positions mask==FALSE contribute result. return_attention_scores: bool, TRUE, returns attention scores (masking softmax) additional output argument. training: Python boolean indicating whether layer behave training mode (adding dropout) inference mode (dropout). use_causal_mask: Boolean. Set TRUE decoder self-attention. Adds mask position attend positions j > . prevents flow information future towards past. Defaults FALSE.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_attention.html","id":"output","dir":"Reference","previous_headings":"","what":"Output","title":"Dot-product attention layer, a.k.a. Luong-style attention. — layer_attention","text":"Attention outputs shape (batch_size, Tq, dim). (Optional) Attention scores masking softmax shape (batch_size, Tq, Tv).","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_aug_mix.html","id":null,"dir":"Reference","previous_headings":"","what":"Performs the AugMix data augmentation technique. — layer_aug_mix","title":"Performs the AugMix data augmentation technique. — layer_aug_mix","text":"AugMix aims produce images variety preserving image semantics local statistics. augmentation process, augmentation applied across images batch num_chains different ways, chain consisting chain_depth augmentations.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_aug_mix.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Performs the AugMix data augmentation technique. — layer_aug_mix","text":"","code":"layer_aug_mix(   object,   value_range = c(0L, 255L),   num_chains = 3L,   chain_depth = 3L,   factor = 0.3,   alpha = 1,   all_ops = TRUE,   interpolation = \"bilinear\",   seed = NULL,   data_format = NULL,   ... )"},{"path":"https://keras3.posit.co/dev/reference/layer_aug_mix.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Performs the AugMix data augmentation technique. — layer_aug_mix","text":"object Object compose layer . tensor, array, sequential model. value_range range values incoming images . Represented two number tuple written (low, high). typically either (0, 1) (0, 255) depending preprocessing pipeline set . num_chains integer representing number different chains mixed, defaults 3. chain_depth integer representing maximum number transformations applied chain. actual number transformations chain sampled randomly range [0, chain_depth]. Defaults 3. factor strength augmentation normalized value 0 1. Default 0.3. alpha float value used probability coefficients Beta Dirichlet distributions, defaults 1.0. all_ops Use operations (including random_brightness, random_color_degeneration, random_contrast random_sharpness). Default True. interpolation interpolation method use resizing operations. Options include \"nearest\", \"bilinear\". Default \"bilinear\". seed Integer. Used create random seed. data_format string, either \"channels_last\" \"channels_first\". ordering dimensions inputs. \"channels_last\" corresponds inputs shape (batch, height, width, channels) \"channels_first\" corresponds inputs shape (batch, channels, height, width). defaults image_data_format value found Keras config file ~/.keras/keras.json. never set , \"channels_last\". ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_aug_mix.html","id":"references","dir":"Reference","previous_headings":"","what":"References","title":"Performs the AugMix data augmentation technique. — layer_aug_mix","text":"AugMix paper Official Code","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_auto_contrast.html","id":null,"dir":"Reference","previous_headings":"","what":"Performs the auto-contrast operation on an image. — layer_auto_contrast","title":"Performs the auto-contrast operation on an image. — layer_auto_contrast","text":"Auto contrast stretches values image across entire available value_range. makes differences pixels obvious. example image values [0, 1] range [0, 255], auto contrast change 1 values 255. layer active training inference time.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_auto_contrast.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Performs the auto-contrast operation on an image. — layer_auto_contrast","text":"","code":"layer_auto_contrast(object, value_range = tuple(0L, 255L), ...)"},{"path":"https://keras3.posit.co/dev/reference/layer_auto_contrast.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Performs the auto-contrast operation on an image. — layer_auto_contrast","text":"object Object compose layer . tensor, array, sequential model. value_range Range values incoming images . Represented two number tuple written (low, high). typically either (0, 1) (0, 255) depending preprocessing pipeline set . Defaults (0, 255). ... forward/backward compatability.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_average.html","id":null,"dir":"Reference","previous_headings":"","what":"Averages a list of inputs element-wise.. — layer_average","title":"Averages a list of inputs element-wise.. — layer_average","text":"takes input list tensors, shape, returns single tensor (also shape).","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_average.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Averages a list of inputs element-wise.. — layer_average","text":"","code":"layer_average(inputs, ...)"},{"path":"https://keras3.posit.co/dev/reference/layer_average.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Averages a list of inputs element-wise.. — layer_average","text":"inputs layers combine ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_average.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Averages a list of inputs element-wise.. — layer_average","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_average.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Averages a list of inputs element-wise.. — layer_average","text":"Usage Keras model:","code":"input_shape <- c(1, 2, 3) x1 <- op_ones(input_shape) x2 <- op_zeros(input_shape) layer_average(x1, x2) ## tf.Tensor( ## [[[0.5 0.5 0.5] ##   [0.5 0.5 0.5]]], shape=(1, 2, 3), dtype=float32) input1 <- layer_input(shape = c(16)) x1 <- input1 |> layer_dense(8, activation = 'relu')  input2 <- layer_input(shape = c(32)) x2 <- input2 |> layer_dense(8, activation = 'relu')  added <- layer_average(x1, x2) output <- added |> layer_dense(4)  model <- keras_model(inputs = c(input1, input2), outputs = output)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_average_pooling_1d.html","id":null,"dir":"Reference","previous_headings":"","what":"Average pooling for temporal data. — layer_average_pooling_1d","title":"Average pooling for temporal data. — layer_average_pooling_1d","text":"Downsamples input representation taking average value window defined pool_size. window shifted strides.  resulting output using \"valid\" padding option shape : output_shape = (input_shape - pool_size + 1) / strides) resulting output shape using \"\" padding option : output_shape = input_shape / strides","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_average_pooling_1d.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Average pooling for temporal data. — layer_average_pooling_1d","text":"","code":"layer_average_pooling_1d(   object,   pool_size,   strides = NULL,   padding = \"valid\",   data_format = NULL,   name = NULL,   ... )"},{"path":"https://keras3.posit.co/dev/reference/layer_average_pooling_1d.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Average pooling for temporal data. — layer_average_pooling_1d","text":"object Object compose layer . tensor, array, sequential model. pool_size int, size max pooling window. strides int NULL. Specifies much pooling window moves pooling step. NULL, default pool_size. padding string, either \"valid\" \"\" (case-insensitive). \"valid\" means padding. \"\" results padding evenly left/right /input output height/width dimension input. data_format string, either \"channels_last\" \"channels_first\". ordering dimensions inputs. \"channels_last\" corresponds inputs shape (batch, steps, features) \"channels_first\" corresponds inputs shape (batch, features, steps). defaults image_data_format value found Keras config file ~/.keras/keras.json. never set , \"channels_last\". name String, name object ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_average_pooling_1d.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Average pooling for temporal data. — layer_average_pooling_1d","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_average_pooling_1d.html","id":"input-shape","dir":"Reference","previous_headings":"","what":"Input Shape","title":"Average pooling for temporal data. — layer_average_pooling_1d","text":"data_format=\"channels_last\": 3D tensor shape (batch_size, steps, features). data_format=\"channels_first\": 3D tensor shape (batch_size, features, steps).","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_average_pooling_1d.html","id":"output-shape","dir":"Reference","previous_headings":"","what":"Output Shape","title":"Average pooling for temporal data. — layer_average_pooling_1d","text":"data_format=\"channels_last\": 3D tensor shape (batch_size, downsampled_steps, features). data_format=\"channels_first\": 3D tensor shape (batch_size, features, downsampled_steps).","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_average_pooling_1d.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Average pooling for temporal data. — layer_average_pooling_1d","text":"strides=1 padding=\"valid\":     strides=2 padding=\"valid\":     strides=1 padding=\"\":","code":"x <- op_array(c(1., 2., 3., 4., 5.)) |> op_reshape(c(1, 5, 1)) output <- x |>   layer_average_pooling_1d(pool_size = 2,                            strides = 1,                            padding = \"valid\") output ## tf.Tensor( ## [[[1.5] ##   [2.5] ##   [3.5] ##   [4.5]]], shape=(1, 4, 1), dtype=float32) x <- op_array(c(1., 2., 3., 4., 5.)) |> op_reshape(c(1, 5, 1)) output <- x |>   layer_average_pooling_1d(pool_size = 2,                            strides = 2,                            padding = \"valid\") output ## tf.Tensor( ## [[[1.5] ##   [3.5]]], shape=(1, 2, 1), dtype=float32) x <- op_array(c(1., 2., 3., 4., 5.)) |> op_reshape(c(1, 5, 1)) output <- x |>   layer_average_pooling_1d(pool_size = 2,                            strides = 1,                            padding = \"same\") output ## tf.Tensor( ## [[[1.5] ##   [2.5] ##   [3.5] ##   [4.5] ##   [5. ]]], shape=(1, 5, 1), dtype=float32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_average_pooling_2d.html","id":null,"dir":"Reference","previous_headings":"","what":"Average pooling operation for 2D spatial data. — layer_average_pooling_2d","title":"Average pooling operation for 2D spatial data. — layer_average_pooling_2d","text":"Downsamples input along spatial dimensions (height width) taking average value input window (size defined pool_size) channel input. window shifted strides along dimension. resulting output using \"valid\" padding option spatial shape (number rows columns) : output_shape = math.floor((input_shape - pool_size) / strides) + 1 (input_shape >= pool_size) resulting output shape using \"\" padding option : output_shape = math.floor((input_shape - 1) / strides) + 1","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_average_pooling_2d.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Average pooling operation for 2D spatial data. — layer_average_pooling_2d","text":"","code":"layer_average_pooling_2d(   object,   pool_size,   strides = NULL,   padding = \"valid\",   data_format = NULL,   name = NULL,   ... )"},{"path":"https://keras3.posit.co/dev/reference/layer_average_pooling_2d.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Average pooling operation for 2D spatial data. — layer_average_pooling_2d","text":"object Object compose layer . tensor, array, sequential model. pool_size int list 2 integers, factors downscale (dim1, dim2). one integer specified, window length used dimensions. strides int list 2 integers, NULL. Strides values. NULL, default pool_size. one int specified, stride size used dimensions. padding string, either \"valid\" \"\" (case-insensitive). \"valid\" means padding. \"\" results padding evenly left/right /input output height/width dimension input. data_format string, either \"channels_last\" \"channels_first\". ordering dimensions inputs. \"channels_last\" corresponds inputs shape (batch, height, width, channels) \"channels_first\" corresponds inputs shape (batch, channels, height, width). defaults image_data_format value found Keras config file ~/.keras/keras.json. never set , \"channels_last\". name String, name object ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_average_pooling_2d.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Average pooling operation for 2D spatial data. — layer_average_pooling_2d","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_average_pooling_2d.html","id":"input-shape","dir":"Reference","previous_headings":"","what":"Input Shape","title":"Average pooling operation for 2D spatial data. — layer_average_pooling_2d","text":"data_format=\"channels_last\": 4D tensor shape (batch_size, height, width, channels). data_format=\"channels_first\": 4D tensor shape (batch_size, channels, height, width).","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_average_pooling_2d.html","id":"output-shape","dir":"Reference","previous_headings":"","what":"Output Shape","title":"Average pooling operation for 2D spatial data. — layer_average_pooling_2d","text":"data_format=\"channels_last\": 4D tensor shape (batch_size, pooled_height, pooled_width, channels). data_format=\"channels_first\": 4D tensor shape (batch_size, channels, pooled_height, pooled_width).","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_average_pooling_2d.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Average pooling operation for 2D spatial data. — layer_average_pooling_2d","text":"strides=(1, 1) padding=\"valid\":     strides=(2, 2) padding=\"valid\":     stride=(1, 1) padding=\"\":","code":"x <- op_array(1:9, \"float32\") |> op_reshape(c(1, 3, 3, 1)) output <- x |>   layer_average_pooling_2d(pool_size = c(2, 2),                            strides = c(1, 1),                            padding = \"valid\") output ## tf.Tensor( ## [[[[3.] ##    [4.]] ## ##   [[6.] ##    [7.]]]], shape=(1, 2, 2, 1), dtype=float32) x <- op_array(1:12, \"float32\") |> op_reshape(c(1, 3, 4, 1)) output <- x |>   layer_average_pooling_2d(pool_size = c(2, 2),                            strides = c(2, 2),                            padding = \"valid\") output ## tf.Tensor( ## [[[[3.5] ##    [5.5]]]], shape=(1, 1, 2, 1), dtype=float32) x <- op_array(1:9, \"float32\") |> op_reshape(c(1, 3, 3, 1)) output <- x |>   layer_average_pooling_2d(pool_size = c(2, 2),                            strides = c(1, 1),                            padding = \"same\") output ## tf.Tensor( ## [[[[3. ] ##    [4. ] ##    [4.5]] ## ##   [[6. ] ##    [7. ] ##    [7.5]] ## ##   [[7.5] ##    [8.5] ##    [9. ]]]], shape=(1, 3, 3, 1), dtype=float32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_average_pooling_3d.html","id":null,"dir":"Reference","previous_headings":"","what":"Average pooling operation for 3D data (spatial or spatio-temporal). — layer_average_pooling_3d","title":"Average pooling operation for 3D data (spatial or spatio-temporal). — layer_average_pooling_3d","text":"Downsamples input along spatial dimensions (depth, height, width) taking average value input window (size defined pool_size) channel input. window shifted strides along dimension.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_average_pooling_3d.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Average pooling operation for 3D data (spatial or spatio-temporal). — layer_average_pooling_3d","text":"","code":"layer_average_pooling_3d(   object,   pool_size,   strides = NULL,   padding = \"valid\",   data_format = NULL,   name = NULL,   ... )"},{"path":"https://keras3.posit.co/dev/reference/layer_average_pooling_3d.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Average pooling operation for 3D data (spatial or spatio-temporal). — layer_average_pooling_3d","text":"object Object compose layer . tensor, array, sequential model. pool_size int list 3 integers, factors downscale (dim1, dim2, dim3). one integer specified, window length used dimensions. strides int list 3 integers, NULL. Strides values. NULL, default pool_size. one int specified, stride size used dimensions. padding string, either \"valid\" \"\" (case-insensitive). \"valid\" means padding. \"\" results padding evenly left/right /input output height/width dimension input. data_format string, either \"channels_last\" \"channels_first\". ordering dimensions inputs. \"channels_last\" corresponds inputs shape (batch, spatial_dim1, spatial_dim2, spatial_dim3, channels) \"channels_first\" corresponds inputs shape (batch, channels, spatial_dim1, spatial_dim2, spatial_dim3). defaults image_data_format value found Keras config file ~/.keras/keras.json. never set , \"channels_last\". name String, name object ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_average_pooling_3d.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Average pooling operation for 3D data (spatial or spatio-temporal). — layer_average_pooling_3d","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_average_pooling_3d.html","id":"input-shape","dir":"Reference","previous_headings":"","what":"Input Shape","title":"Average pooling operation for 3D data (spatial or spatio-temporal). — layer_average_pooling_3d","text":"data_format=\"channels_last\": 5D tensor shape: (batch_size, spatial_dim1, spatial_dim2, spatial_dim3, channels) data_format=\"channels_first\": 5D tensor shape: (batch_size, channels, spatial_dim1, spatial_dim2, spatial_dim3)","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_average_pooling_3d.html","id":"output-shape","dir":"Reference","previous_headings":"","what":"Output Shape","title":"Average pooling operation for 3D data (spatial or spatio-temporal). — layer_average_pooling_3d","text":"data_format=\"channels_last\": 5D tensor shape: (batch_size, pooled_dim1, pooled_dim2, pooled_dim3, channels) data_format=\"channels_first\": 5D tensor shape: (batch_size, channels, pooled_dim1, pooled_dim2, pooled_dim3)","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_average_pooling_3d.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Average pooling operation for 3D data (spatial or spatio-temporal). — layer_average_pooling_3d","text":"","code":"depth <- height <- width <- 30 channels <- 3  inputs <- layer_input(shape = c(depth, height, width, channels)) outputs <- inputs |> layer_average_pooling_3d(pool_size = 3) outputs # Shape: (batch_size, 10, 10, 10, 3) ## <KerasTensor shape=(None, 10, 10, 10, 3), dtype=float32, sparse=False, ragged=False, name=keras_tensor_1>"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_batch_normalization.html","id":null,"dir":"Reference","previous_headings":"","what":"Layer that normalizes its inputs. — layer_batch_normalization","title":"Layer that normalizes its inputs. — layer_batch_normalization","text":"Batch normalization applies transformation maintains mean output close 0 output standard deviation close 1. Importantly, batch normalization works differently training inference. training (.e. using fit() calling layer/model argument training = TRUE), layer normalizes output using mean standard deviation current batch inputs. say, channel normalized, layer returns gamma * (batch - mean(batch)) / sqrt(var(batch) + epsilon) + beta, : epsilon small constant (configurable part constructor arguments) gamma learned scaling factor (initialized 1), can disabled passing scale = FALSE constructor. beta learned offset factor (initialized 0), can disabled passing center = FALSE constructor. inference (.e. using evaluate() predict() calling layer/model argument training = FALSE (default), layer normalizes output using moving average mean standard deviation batches seen training. say, returns gamma * (batch - self$moving_mean) / sqrt(self$moving_var+epsilon) + beta. self$moving_mean self$moving_var non-trainable variables updated time layer called training mode, : moving_mean = moving_mean * momentum + mean(batch) * (1 - momentum) moving_var = moving_var * momentum + var(batch) * (1 - momentum) , layer normalize inputs inference trained data similar statistics inference data. setting layer$trainable <- FALSE BatchNormalization layer: meaning setting layer$trainable <- FALSE freeze layer, .e. internal state change training: trainable weights updated fit() train_on_batch(), state updates run. Usually, necessarily mean layer run inference mode (normally controlled training argument can passed calling layer). \"Frozen state\" \"inference mode\" two separate concepts. However, case BatchNormalization layer, setting trainable <- FALSE layer means layer subsequently run inference mode (meaning use moving mean moving variance normalize current batch, rather using mean variance current batch). Note : Setting trainable model containing layers recursively set trainable value inner layers. value trainable attribute changed calling compile() model, new value take effect model compile() called .","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_batch_normalization.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Layer that normalizes its inputs. — layer_batch_normalization","text":"","code":"layer_batch_normalization(   object,   axis = -1L,   momentum = 0.99,   epsilon = 0.001,   center = TRUE,   scale = TRUE,   beta_initializer = \"zeros\",   gamma_initializer = \"ones\",   moving_mean_initializer = \"zeros\",   moving_variance_initializer = \"ones\",   beta_regularizer = NULL,   gamma_regularizer = NULL,   beta_constraint = NULL,   gamma_constraint = NULL,   synchronized = FALSE,   ... )"},{"path":"https://keras3.posit.co/dev/reference/layer_batch_normalization.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Layer that normalizes its inputs. — layer_batch_normalization","text":"object Object compose layer . tensor, array, sequential model. axis Integer, axis normalized (typically features axis). instance, Conv2D layer data_format = \"channels_first\", use axis = 2. momentum Momentum moving average. epsilon Small float added variance avoid dividing zero. center TRUE, add offset beta normalized tensor. FALSE, beta ignored. scale TRUE, multiply gamma. FALSE, gamma used. next layer linear can disabled since scaling done next layer. beta_initializer Initializer beta weight. gamma_initializer Initializer gamma weight. moving_mean_initializer Initializer moving mean. moving_variance_initializer Initializer moving variance. beta_regularizer Optional regularizer beta weight. gamma_regularizer Optional regularizer gamma weight. beta_constraint Optional constraint beta weight. gamma_constraint Optional constraint gamma weight. synchronized applicable TensorFlow backend. TRUE, synchronizes global batch statistics (mean variance) layer across devices training step distributed training strategy. FALSE, replica uses local batch statistics. ... Base layer keyword arguments (e.g. name dtype).","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_batch_normalization.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Layer that normalizes its inputs. — layer_batch_normalization","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_batch_normalization.html","id":"call-arguments","dir":"Reference","previous_headings":"","what":"Call Arguments","title":"Layer that normalizes its inputs. — layer_batch_normalization","text":"inputs: Input tensor (rank). training: R boolean indicating whether layer behave training mode inference mode. training = TRUE: layer normalize inputs using mean variance current batch inputs. training = FALSE: layer normalize inputs using mean variance moving statistics, learned training. mask: Binary tensor shape broadcastable inputs tensor, TRUE values indicating positions mean variance computed. Masked elements current inputs taken account mean variance computation training. prior unmasked element values taken account momentum expires.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_batch_normalization.html","id":"reference","dir":"Reference","previous_headings":"","what":"Reference","title":"Layer that normalizes its inputs. — layer_batch_normalization","text":"Ioffe Szegedy, 2015.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_bidirectional.html","id":null,"dir":"Reference","previous_headings":"","what":"Bidirectional wrapper for RNNs. — layer_bidirectional","title":"Bidirectional wrapper for RNNs. — layer_bidirectional","text":"Bidirectional wrapper RNNs.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_bidirectional.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Bidirectional wrapper for RNNs. — layer_bidirectional","text":"","code":"layer_bidirectional(   object,   layer,   merge_mode = \"concat\",   weights = NULL,   backward_layer = NULL,   ... )"},{"path":"https://keras3.posit.co/dev/reference/layer_bidirectional.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Bidirectional wrapper for RNNs. — layer_bidirectional","text":"object Object compose layer . tensor, array, sequential model. layer RNN instance, layer_lstm() layer_gru(). also Layer() instance meets following criteria: sequence-processing layer (accepts 3D+ inputs). go_backwards, return_sequences return_state attribute (semantics RNN class). input_spec attribute. Implement serialization via get_config() from_config(). Note recommended way create new RNN layers write custom RNN cell use layer_rnn(), instead subclassing Layer() directly. return_sequences TRUE, output masked timestep zero regardless layer's original zero_output_for_mask value. merge_mode Mode outputs forward backward RNNs combined. One {\"sum\", \"mul\", \"concat\", \"ave\", NULL}. NULL, outputs combined, returned list. Defaults \"concat\". weights see description backward_layer Optional RNN, Layer() instance used handle backwards input processing. backward_layer provided, layer instance passed layer argument used generate backward layer automatically. Note provided backward_layer layer properties matching layer argument, particular values stateful, return_states, return_sequences, etc. addition, backward_layer layer different go_backwards argument values. ValueError raised requirements met. ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_bidirectional.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Bidirectional wrapper for RNNs. — layer_bidirectional","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_bidirectional.html","id":"call-arguments","dir":"Reference","previous_headings":"","what":"Call Arguments","title":"Bidirectional wrapper for RNNs. — layer_bidirectional","text":"call arguments layer wrapped RNN layer. Beware passing initial_state argument call layer, first half list elements initial_state list passed forward RNN call last half list elements passed backward RNN call.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_bidirectional.html","id":"note","dir":"Reference","previous_headings":"","what":"Note","title":"Bidirectional wrapper for RNNs. — layer_bidirectional","text":"instantiating Bidirectional layer existing RNN layer instance reuse weights state RNN layer instance – Bidirectional layer freshly initialized weights.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_bidirectional.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Bidirectional wrapper for RNNs. — layer_bidirectional","text":"","code":"model <- keras_model_sequential(input_shape = c(5, 10)) %>%   layer_bidirectional(layer_lstm(units = 10, return_sequences = TRUE)) %>%   layer_bidirectional(layer_lstm(units = 10)) %>%   layer_dense(5, activation = \"softmax\")  model %>% compile(loss = \"categorical_crossentropy\",                   optimizer = \"rmsprop\")  # With custom backward layer forward_layer <- layer_lstm(units = 10, return_sequences = TRUE) backward_layer <- layer_lstm(units = 10, activation = \"relu\",                              return_sequences = TRUE, go_backwards = TRUE)  model <- keras_model_sequential(input_shape = c(5, 10)) %>%   bidirectional(forward_layer, backward_layer = backward_layer) %>%   layer_dense(5, activation = \"softmax\")  model %>% compile(loss = \"categorical_crossentropy\",                   optimizer = \"rmsprop\")"},{"path":"https://keras3.posit.co/dev/reference/layer_bidirectional.html","id":"states","dir":"Reference","previous_headings":"","what":"States","title":"Bidirectional wrapper for RNNs. — layer_bidirectional","text":"Bidirectional layer instance property states, can access layer$states. can also reset states using reset_state()","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_category_encoding.html","id":null,"dir":"Reference","previous_headings":"","what":"A preprocessing layer which encodes integer features. — layer_category_encoding","title":"A preprocessing layer which encodes integer features. — layer_category_encoding","text":"layer provides options condensing data categorical encoding total number tokens known advance. accepts integer values inputs, outputs dense sparse representation inputs. integer inputs total number tokens known, use layer_integer_lookup() instead. Note: layer safe use inside tf.data pipeline (independently backend using).","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_category_encoding.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"A preprocessing layer which encodes integer features. — layer_category_encoding","text":"","code":"layer_category_encoding(   object,   num_tokens = NULL,   output_mode = \"multi_hot\",   sparse = FALSE,   ... )"},{"path":"https://keras3.posit.co/dev/reference/layer_category_encoding.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"A preprocessing layer which encodes integer features. — layer_category_encoding","text":"object Object compose layer . tensor, array, sequential model. num_tokens total number tokens layer support. inputs layer must integers range 0 <= value < num_tokens, error thrown. output_mode Specification output layer. Values can \"one_hot\", \"multi_hot\" \"count\", configuring layer follows: - \"one_hot\": Encodes individual element input array num_tokens size, containing 1 element index. last dimension size 1, encode dimension. last dimension size 1, append new dimension encoded output. - \"multi_hot\": Encodes sample input single array num_tokens size, containing 1 vocabulary term present sample. Treats last dimension sample dimension, input shape (..., sample_length), output shape (..., num_tokens). - \"count\": Like \"multi_hot\", int array contains count number times token index appeared sample. output modes, currently output rank 2 supported. Defaults \"multi_hot\". sparse Whether return sparse tensor; backends support sparse tensors. ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_category_encoding.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"A preprocessing layer which encodes integer features. — layer_category_encoding","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_category_encoding.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"A preprocessing layer which encodes integer features. — layer_category_encoding","text":"One-hot encoding data     Multi-hot encoding data     Using weighted inputs \"count\" mode","code":"layer <- layer_category_encoding(num_tokens = 4, output_mode = \"one_hot\") x <- op_array(c(3, 2, 0, 1), \"int32\") layer(x) ## tf.Tensor( ## [[0. 0. 0. 1.] ##  [0. 0. 1. 0.] ##  [1. 0. 0. 0.] ##  [0. 1. 0. 0.]], shape=(4, 4), dtype=float32) layer <- layer_category_encoding(num_tokens = 4, output_mode = \"multi_hot\") x <- op_array(rbind(c(0, 1),                    c(0, 0),                    c(1, 2),                    c(3, 1)), \"int32\") layer(x) ## tf.Tensor( ## [[1. 1. 0. 0.] ##  [1. 0. 0. 0.] ##  [0. 1. 1. 0.] ##  [0. 1. 0. 1.]], shape=(4, 4), dtype=float32) layer <- layer_category_encoding(num_tokens = 4, output_mode = \"count\") count_weights <- op_array(rbind(c(.1, .2),                                c(.1, .1),                                c(.2, .3),                                c(.4, .2))) x <- op_array(rbind(c(0, 1),                    c(0, 0),                    c(1, 2),                    c(3, 1)), \"int32\") layer(x, count_weights = count_weights) #   array([[01, 02, 0. , 0. ], #          [02, 0. , 0. , 0. ], #          [0. , 02, 03, 0. ], #          [0. , 02, 0. , 04]]>"},{"path":"https://keras3.posit.co/dev/reference/layer_category_encoding.html","id":"call-arguments","dir":"Reference","previous_headings":"","what":"Call Arguments","title":"A preprocessing layer which encodes integer features. — layer_category_encoding","text":"inputs: 1D 2D tensor integer inputs. count_weights: tensor shape inputs indicating weight sample value summing count mode. used \"multi_hot\" \"one_hot\" modes.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_center_crop.html","id":null,"dir":"Reference","previous_headings":"","what":"A preprocessing layer which crops images. — layer_center_crop","title":"A preprocessing layer which crops images. — layer_center_crop","text":"layers crops central portion images target size. image smaller target size, resized cropped return largest possible window image matches target aspect ratio. Input pixel values can range (e.g. [0., 1.) [0, 255]).","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_center_crop.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"A preprocessing layer which crops images. — layer_center_crop","text":"","code":"layer_center_crop(object, height, width, data_format = NULL, ...)"},{"path":"https://keras3.posit.co/dev/reference/layer_center_crop.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"A preprocessing layer which crops images. — layer_center_crop","text":"object Object compose layer . tensor, array, sequential model. height Integer, height output shape. width Integer, width output shape. data_format string, either \"channels_last\" \"channels_first\". ordering dimensions inputs. \"channels_last\" corresponds inputs shape (batch, height, width, channels) \"channels_first\" corresponds inputs shape (batch, channels, height, width). defaults image_data_format value found Keras config file ~/.keras/keras.json. never set , \"channels_last\". ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_center_crop.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"A preprocessing layer which crops images. — layer_center_crop","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_center_crop.html","id":"input-shape","dir":"Reference","previous_headings":"","what":"Input Shape","title":"A preprocessing layer which crops images. — layer_center_crop","text":"3D (unbatched) 4D (batched) tensor shape: (..., height, width, channels), \"channels_last\" format, (..., channels, height, width), \"channels_first\" format.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_center_crop.html","id":"output-shape","dir":"Reference","previous_headings":"","what":"Output Shape","title":"A preprocessing layer which crops images. — layer_center_crop","text":"3D (unbatched) 4D (batched) tensor shape: (..., target_height, target_width, channels), (..., channels, target_height, target_width), \"channels_first\" format. input height/width even target height/width odd (inversely), input image left-padded 1 pixel. Note: layer safe use inside tf.data pipeline (independently backend using).","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_concatenate.html","id":null,"dir":"Reference","previous_headings":"","what":"Concatenates a list of inputs. — layer_concatenate","title":"Concatenates a list of inputs. — layer_concatenate","text":"takes input list tensors, shape except concatenation axis, returns single tensor concatenation inputs.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_concatenate.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Concatenates a list of inputs. — layer_concatenate","text":"","code":"layer_concatenate(inputs, ..., axis = -1L)"},{"path":"https://keras3.posit.co/dev/reference/layer_concatenate.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Concatenates a list of inputs. — layer_concatenate","text":"inputs layers combine ... Standard layer keyword arguments. axis Axis along concatenate.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_concatenate.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Concatenates a list of inputs. — layer_concatenate","text":"tensor, concatenation inputs alongside axis axis.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_concatenate.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Concatenates a list of inputs. — layer_concatenate","text":"Usage Keras model:","code":"x <- op_arange(20) |> op_reshape(c(2, 2, 5)) y <- op_arange(21, 40) |> op_reshape(c(2, 2, 5)) layer_concatenate(x, y, axis = 2) ## tf.Tensor( ## [[[ 1.  2.  3.  4.  5.] ##   [ 6.  7.  8.  9. 10.] ##   [21. 22. 23. 24. 25.] ##   [26. 27. 28. 29. 30.]] ## ##  [[11. 12. 13. 14. 15.] ##   [16. 17. 18. 19. 20.] ##   [31. 32. 33. 34. 35.] ##   [36. 37. 38. 39. 40.]]], shape=(2, 4, 5), dtype=float32) x1 <- op_arange(10)     |> op_reshape(c(5, 2)) |> layer_dense(8) x2 <- op_arange(11, 20) |> op_reshape(c(5, 2)) |> layer_dense(8) y <- layer_concatenate(x1, x2)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_conv_1d.html","id":null,"dir":"Reference","previous_headings":"","what":"1D convolution layer (e.g. temporal convolution). — layer_conv_1d","title":"1D convolution layer (e.g. temporal convolution). — layer_conv_1d","text":"layer creates convolution kernel convolved layer input single spatial (temporal) dimension produce tensor outputs. use_bias TRUE, bias vector created added outputs. Finally, activation NULL, applied outputs well.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_conv_1d.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"1D convolution layer (e.g. temporal convolution). — layer_conv_1d","text":"","code":"layer_conv_1d(   object,   filters,   kernel_size,   strides = 1L,   padding = \"valid\",   data_format = NULL,   dilation_rate = 1L,   groups = 1L,   activation = NULL,   use_bias = TRUE,   kernel_initializer = \"glorot_uniform\",   bias_initializer = \"zeros\",   kernel_regularizer = NULL,   bias_regularizer = NULL,   activity_regularizer = NULL,   kernel_constraint = NULL,   bias_constraint = NULL,   ... )"},{"path":"https://keras3.posit.co/dev/reference/layer_conv_1d.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"1D convolution layer (e.g. temporal convolution). — layer_conv_1d","text":"object Object compose layer . tensor, array, sequential model. filters int, dimension output space (number filters convolution). kernel_size int list 1 integer, specifying size convolution window. strides int list 1 integer, specifying stride length convolution. strides > 1 incompatible dilation_rate > 1. padding string, \"valid\", \"\" \"causal\"(case-insensitive). \"valid\" means padding. \"\" results padding evenly left/right /input. padding=\"\" strides=1, output size input. \"causal\" results causal (dilated) convolutions, e.g. output[t] depend ontail(input, t+1). Useful modeling temporal data model violate temporal order. See WaveNet: Generative Model Raw Audio, section2.1. data_format string, either \"channels_last\" \"channels_first\". ordering dimensions inputs. \"channels_last\" corresponds inputs shape (batch, steps, features) \"channels_first\" corresponds inputs shape (batch, features, steps). defaults image_data_format value found Keras config file ~/.keras/keras.json. never set , \"channels_last\". dilation_rate int list 1 integers, specifying dilation rate use dilated convolution. groups positive int specifying number groups input split along channel axis. group convolved separately filters // groups filters. output concatenation groups results along channel axis. Input channels filters must divisible groups. activation Activation function. NULL, activation applied. use_bias bool, TRUE, bias added output. kernel_initializer Initializer convolution kernel. NULL, default initializer (\"glorot_uniform\") used. bias_initializer Initializer bias vector. NULL, default initializer (\"zeros\") used. kernel_regularizer Optional regularizer convolution kernel. bias_regularizer Optional regularizer bias vector. activity_regularizer Optional regularizer function output. kernel_constraint Optional projection function applied kernel updated Optimizer (e.g. used implement norm constraints value constraints layer weights). function must take input unprojected variable must return projected variable (must shape). Constraints safe use asynchronous distributed training. bias_constraint Optional projection function applied bias updated Optimizer. ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_conv_1d.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"1D convolution layer (e.g. temporal convolution). — layer_conv_1d","text":"3D tensor representing activation(conv1d(inputs, kernel) + bias).","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_conv_1d.html","id":"input-shape","dir":"Reference","previous_headings":"","what":"Input Shape","title":"1D convolution layer (e.g. temporal convolution). — layer_conv_1d","text":"data_format=\"channels_last\": 3D tensor shape: (batch_shape, steps, channels) data_format=\"channels_first\": 3D tensor shape: (batch_shape, channels, steps)","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_conv_1d.html","id":"output-shape","dir":"Reference","previous_headings":"","what":"Output Shape","title":"1D convolution layer (e.g. temporal convolution). — layer_conv_1d","text":"data_format=\"channels_last\": 3D tensor shape: (batch_shape, new_steps, filters) data_format=\"channels_first\": 3D tensor shape: (batch_shape, filters, new_steps)","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_conv_1d.html","id":"raises","dir":"Reference","previous_headings":"","what":"Raises","title":"1D convolution layer (e.g. temporal convolution). — layer_conv_1d","text":"ValueError: strides > 1 dilation_rate > 1.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_conv_1d.html","id":"example","dir":"Reference","previous_headings":"","what":"Example","title":"1D convolution layer (e.g. temporal convolution). — layer_conv_1d","text":"","code":"# The inputs are 128-length vectors with 10 timesteps, and the # batch size is 4. x <- random_uniform(c(4, 10, 128)) y <- x |> layer_conv_1d(32, 3, activation='relu') shape(y) ## shape(4, 8, 32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_conv_1d_transpose.html","id":null,"dir":"Reference","previous_headings":"","what":"1D transposed convolution layer. — layer_conv_1d_transpose","title":"1D transposed convolution layer. — layer_conv_1d_transpose","text":"need transposed convolutions generally arise desire use transformation going opposite direction normal convolution, .e., something shape output convolution something shape input maintaining connectivity pattern compatible said convolution.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_conv_1d_transpose.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"1D transposed convolution layer. — layer_conv_1d_transpose","text":"","code":"layer_conv_1d_transpose(   object,   filters,   kernel_size,   strides = 1L,   padding = \"valid\",   data_format = NULL,   dilation_rate = 1L,   activation = NULL,   use_bias = TRUE,   kernel_initializer = \"glorot_uniform\",   bias_initializer = \"zeros\",   kernel_regularizer = NULL,   bias_regularizer = NULL,   activity_regularizer = NULL,   kernel_constraint = NULL,   bias_constraint = NULL,   ... )"},{"path":"https://keras3.posit.co/dev/reference/layer_conv_1d_transpose.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"1D transposed convolution layer. — layer_conv_1d_transpose","text":"object Object compose layer . tensor, array, sequential model. filters int, dimension output space (number filters transpose convolution). kernel_size int list 1 integer, specifying size transposed convolution window. strides int list 1 integer, specifying stride length transposed convolution. strides > 1 incompatible dilation_rate > 1. padding string, either \"valid\" \"\" (case-insensitive). \"valid\" means padding. \"\" results padding evenly left/right /input output height/width dimension input. data_format string, either \"channels_last\" \"channels_first\". ordering dimensions inputs. \"channels_last\" corresponds inputs shape (batch, steps, features) \"channels_first\" corresponds inputs shape (batch, features, steps). defaults image_data_format value found Keras config file ~/.keras/keras.json. never set , \"channels_last\". dilation_rate int list 1 integers, specifying dilation rate use dilated transposed convolution. activation Activation function. NULL, activation applied. use_bias bool, TRUE, bias added output. kernel_initializer Initializer convolution kernel. NULL, default initializer (\"glorot_uniform\") used. bias_initializer Initializer bias vector. NULL, default initializer (\"zeros\") used. kernel_regularizer Optional regularizer convolution kernel. bias_regularizer Optional regularizer bias vector. activity_regularizer Optional regularizer function output. kernel_constraint Optional projection function applied kernel updated Optimizer (e.g. used implement norm constraints value constraints layer weights). function must take input unprojected variable must return projected variable (must shape). Constraints safe use asynchronous distributed training. bias_constraint Optional projection function applied bias updated Optimizer. ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_conv_1d_transpose.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"1D transposed convolution layer. — layer_conv_1d_transpose","text":"3D tensor representing activation(conv1d_transpose(inputs, kernel) + bias).","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_conv_1d_transpose.html","id":"input-shape","dir":"Reference","previous_headings":"","what":"Input Shape","title":"1D transposed convolution layer. — layer_conv_1d_transpose","text":"data_format=\"channels_last\": 3D tensor shape: (batch_shape, steps, channels) data_format=\"channels_first\": 3D tensor shape: (batch_shape, channels, steps)","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_conv_1d_transpose.html","id":"output-shape","dir":"Reference","previous_headings":"","what":"Output Shape","title":"1D transposed convolution layer. — layer_conv_1d_transpose","text":"data_format=\"channels_last\": 3D tensor shape: (batch_shape, new_steps, filters) data_format=\"channels_first\": 3D tensor shape: (batch_shape, filters, new_steps)","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_conv_1d_transpose.html","id":"raises","dir":"Reference","previous_headings":"","what":"Raises","title":"1D transposed convolution layer. — layer_conv_1d_transpose","text":"ValueError: strides > 1 dilation_rate > 1.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_conv_1d_transpose.html","id":"references","dir":"Reference","previous_headings":"","what":"References","title":"1D transposed convolution layer. — layer_conv_1d_transpose","text":"guide convolution arithmetic deep learning Deconvolutional Networks","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_conv_1d_transpose.html","id":"example","dir":"Reference","previous_headings":"","what":"Example","title":"1D transposed convolution layer. — layer_conv_1d_transpose","text":"","code":"x <- random_uniform(c(4, 10, 128)) y <- x |> layer_conv_1d_transpose(32, 3, 2, activation='relu') shape(y) ## shape(4, 21, 32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_conv_2d.html","id":null,"dir":"Reference","previous_headings":"","what":"2D convolution layer. — layer_conv_2d","title":"2D convolution layer. — layer_conv_2d","text":"layer creates convolution kernel convolved layer input 2D spatial (temporal) dimension (height width) produce tensor outputs. use_bias TRUE, bias vector created added outputs. Finally, activation NULL, applied outputs well. Note numerical precision: general Keras operation execution results identical across backends 1e-7 precision float32, Conv2D operations may show larger variations. Due large number element-wise multiplications additions convolution operations, especially large inputs kernel sizes, accumulated floating-point differences can exceed 1e-7 threshold. variations particularly noticeable using different backends (e.g., TensorFlow vs JAX) different hardware.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_conv_2d.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"2D convolution layer. — layer_conv_2d","text":"","code":"layer_conv_2d(   object,   filters,   kernel_size,   strides = list(1L, 1L),   padding = \"valid\",   data_format = NULL,   dilation_rate = list(1L, 1L),   groups = 1L,   activation = NULL,   use_bias = TRUE,   kernel_initializer = \"glorot_uniform\",   bias_initializer = \"zeros\",   kernel_regularizer = NULL,   bias_regularizer = NULL,   activity_regularizer = NULL,   kernel_constraint = NULL,   bias_constraint = NULL,   ... )"},{"path":"https://keras3.posit.co/dev/reference/layer_conv_2d.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"2D convolution layer. — layer_conv_2d","text":"object Object compose layer . tensor, array, sequential model. filters int, dimension output space (number filters convolution). kernel_size int list 2 integer, specifying size convolution window. strides int list 2 integer, specifying stride length convolution. strides > 1 incompatible dilation_rate > 1. padding string, either \"valid\" \"\" (case-insensitive). \"valid\" means padding. \"\" results padding evenly left/right /input. padding=\"\" strides=1, output size input. data_format string, either \"channels_last\" \"channels_first\". ordering dimensions inputs. \"channels_last\" corresponds inputs shape (batch_size, height, width, channels) \"channels_first\" corresponds inputs shape (batch_size, channels, height, width). defaults image_data_format value found Keras config file ~/.keras/keras.json. never set , \"channels_last\". dilation_rate int list 2 integers, specifying dilation rate use dilated convolution. groups positive int specifying number groups input split along channel axis. group convolved separately filters // groups filters. output concatenation groups results along channel axis. Input channels filters must divisible groups. activation Activation function. NULL, activation applied. use_bias bool, TRUE, bias added output. kernel_initializer Initializer convolution kernel. NULL, default initializer (\"glorot_uniform\") used. bias_initializer Initializer bias vector. NULL, default initializer (\"zeros\") used. kernel_regularizer Optional regularizer convolution kernel. bias_regularizer Optional regularizer bias vector. activity_regularizer Optional regularizer function output. kernel_constraint Optional projection function applied kernel updated Optimizer (e.g. used implement norm constraints value constraints layer weights). function must take input unprojected variable must return projected variable (must shape). Constraints safe use asynchronous distributed training. bias_constraint Optional projection function applied bias updated Optimizer. ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_conv_2d.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"2D convolution layer. — layer_conv_2d","text":"4D tensor representing activation(conv2d(inputs, kernel) + bias).","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_conv_2d.html","id":"input-shape","dir":"Reference","previous_headings":"","what":"Input Shape","title":"2D convolution layer. — layer_conv_2d","text":"data_format=\"channels_last\": 4D tensor shape: (batch_size, height, width, channels) data_format=\"channels_first\": 4D tensor shape: (batch_size, channels, height, width)","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_conv_2d.html","id":"output-shape","dir":"Reference","previous_headings":"","what":"Output Shape","title":"2D convolution layer. — layer_conv_2d","text":"data_format=\"channels_last\": 4D tensor shape: (batch_size, new_height, new_width, filters) data_format=\"channels_first\": 4D tensor shape: (batch_size, filters, new_height, new_width)","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_conv_2d.html","id":"raises","dir":"Reference","previous_headings":"","what":"Raises","title":"2D convolution layer. — layer_conv_2d","text":"ValueError: strides > 1 dilation_rate > 1.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_conv_2d.html","id":"example","dir":"Reference","previous_headings":"","what":"Example","title":"2D convolution layer. — layer_conv_2d","text":"","code":"x <- random_uniform(c(4, 10, 10, 128)) y <- x |> layer_conv_2d(32, 3, activation='relu') shape(y) ## shape(4, 8, 8, 32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_conv_2d_transpose.html","id":null,"dir":"Reference","previous_headings":"","what":"2D transposed convolution layer. — layer_conv_2d_transpose","title":"2D transposed convolution layer. — layer_conv_2d_transpose","text":"need transposed convolutions generally arise desire use transformation going opposite direction normal convolution, .e., something shape output convolution something shape input maintaining connectivity pattern compatible said convolution.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_conv_2d_transpose.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"2D transposed convolution layer. — layer_conv_2d_transpose","text":"","code":"layer_conv_2d_transpose(   object,   filters,   kernel_size,   strides = list(1L, 1L),   padding = \"valid\",   data_format = NULL,   dilation_rate = list(1L, 1L),   activation = NULL,   use_bias = TRUE,   kernel_initializer = \"glorot_uniform\",   bias_initializer = \"zeros\",   kernel_regularizer = NULL,   bias_regularizer = NULL,   activity_regularizer = NULL,   kernel_constraint = NULL,   bias_constraint = NULL,   ... )"},{"path":"https://keras3.posit.co/dev/reference/layer_conv_2d_transpose.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"2D transposed convolution layer. — layer_conv_2d_transpose","text":"object Object compose layer . tensor, array, sequential model. filters int, dimension output space (number filters transposed convolution). kernel_size int list 1 integer, specifying size transposed convolution window. strides int list 1 integer, specifying stride length transposed convolution. strides > 1 incompatible dilation_rate > 1. padding string, either \"valid\" \"\" (case-insensitive). \"valid\" means padding. \"\" results padding evenly left/right /input. padding=\"\" strides=1, output size input. data_format string, either \"channels_last\" \"channels_first\". ordering dimensions inputs. \"channels_last\" corresponds inputs shape (batch_size, height, width, channels) \"channels_first\" corresponds inputs shape (batch_size, channels, height, width). defaults image_data_format value found Keras config file ~/.keras/keras.json. never set , \"channels_last\". dilation_rate int list 1 integers, specifying dilation rate use dilated transposed convolution. activation Activation function. NULL, activation applied. use_bias bool, TRUE, bias added output. kernel_initializer Initializer convolution kernel. NULL, default initializer (\"glorot_uniform\") used. bias_initializer Initializer bias vector. NULL, default initializer (\"zeros\") used. kernel_regularizer Optional regularizer convolution kernel. bias_regularizer Optional regularizer bias vector. activity_regularizer Optional regularizer function output. kernel_constraint Optional projection function applied kernel updated Optimizer (e.g. used implement norm constraints value constraints layer weights). function must take input unprojected variable must return projected variable (must shape). Constraints safe use asynchronous distributed training. bias_constraint Optional projection function applied bias updated Optimizer. ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_conv_2d_transpose.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"2D transposed convolution layer. — layer_conv_2d_transpose","text":"4D tensor representing activation(conv2d_transpose(inputs, kernel) + bias).","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_conv_2d_transpose.html","id":"input-shape","dir":"Reference","previous_headings":"","what":"Input Shape","title":"2D transposed convolution layer. — layer_conv_2d_transpose","text":"data_format=\"channels_last\": 4D tensor shape: (batch_size, height, width, channels) data_format=\"channels_first\": 4D tensor shape: (batch_size, channels, height, width)","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_conv_2d_transpose.html","id":"output-shape","dir":"Reference","previous_headings":"","what":"Output Shape","title":"2D transposed convolution layer. — layer_conv_2d_transpose","text":"data_format=\"channels_last\": 4D tensor shape: (batch_size, new_height, new_width, filters) data_format=\"channels_first\": 4D tensor shape: (batch_size, filters, new_height, new_width)","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_conv_2d_transpose.html","id":"raises","dir":"Reference","previous_headings":"","what":"Raises","title":"2D transposed convolution layer. — layer_conv_2d_transpose","text":"ValueError: strides > 1 dilation_rate > 1.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_conv_2d_transpose.html","id":"references","dir":"Reference","previous_headings":"","what":"References","title":"2D transposed convolution layer. — layer_conv_2d_transpose","text":"guide convolution arithmetic deep learning Deconvolutional Networks","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_conv_2d_transpose.html","id":"example","dir":"Reference","previous_headings":"","what":"Example","title":"2D transposed convolution layer. — layer_conv_2d_transpose","text":"","code":"x <- random_uniform(c(4, 10, 8, 128)) y <- x |> layer_conv_2d_transpose(32, 2, 2, activation='relu') shape(y) ## shape(4, 20, 16, 32) # (4, 20, 16, 32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_conv_3d.html","id":null,"dir":"Reference","previous_headings":"","what":"3D convolution layer. — layer_conv_3d","title":"3D convolution layer. — layer_conv_3d","text":"layer creates convolution kernel convolved layer input 3D spatial (temporal) dimension (width,height depth) produce tensor outputs. use_bias TRUE, bias vector created added outputs. Finally, activation NULL, applied outputs well.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_conv_3d.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"3D convolution layer. — layer_conv_3d","text":"","code":"layer_conv_3d(   object,   filters,   kernel_size,   strides = list(1L, 1L, 1L),   padding = \"valid\",   data_format = NULL,   dilation_rate = list(1L, 1L, 1L),   groups = 1L,   activation = NULL,   use_bias = TRUE,   kernel_initializer = \"glorot_uniform\",   bias_initializer = \"zeros\",   kernel_regularizer = NULL,   bias_regularizer = NULL,   activity_regularizer = NULL,   kernel_constraint = NULL,   bias_constraint = NULL,   ... )"},{"path":"https://keras3.posit.co/dev/reference/layer_conv_3d.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"3D convolution layer. — layer_conv_3d","text":"object Object compose layer . tensor, array, sequential model. filters int, dimension output space (number filters convolution). kernel_size int list 3 integer, specifying size convolution window. strides int list 3 integer, specifying stride length convolution. strides > 1 incompatible dilation_rate > 1. padding string, either \"valid\" \"\" (case-insensitive). \"valid\" means padding. \"\" results padding evenly left/right /input. padding=\"\" strides=1, output size input. data_format string, either \"channels_last\" \"channels_first\". ordering dimensions inputs. \"channels_last\" corresponds inputs shape (batch_size, spatial_dim1, spatial_dim2, spatial_dim3, channels) \"channels_first\" corresponds inputs shape (batch_size, channels, spatial_dim1, spatial_dim2, spatial_dim3). defaults image_data_format value found Keras config file ~/.keras/keras.json. never set , \"channels_last\". dilation_rate int list 3 integers, specifying dilation rate use dilated convolution. groups positive int specifying number groups input split along channel axis. group convolved separately filters %/% groups filters. output concatenation groups results along channel axis. Input channels filters must divisible groups. activation Activation function. NULL, activation applied. use_bias bool, TRUE, bias added output. kernel_initializer Initializer convolution kernel. NULL, default initializer (\"glorot_uniform\") used. bias_initializer Initializer bias vector. NULL, default initializer (\"zeros\") used. kernel_regularizer Optional regularizer convolution kernel. bias_regularizer Optional regularizer bias vector. activity_regularizer Optional regularizer function output. kernel_constraint Optional projection function applied kernel updated Optimizer (e.g. used implement norm constraints value constraints layer weights). function must take input unprojected variable must return projected variable (must shape). Constraints safe use asynchronous distributed training. bias_constraint Optional projection function applied bias updated Optimizer. ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_conv_3d.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"3D convolution layer. — layer_conv_3d","text":"5D tensor representing activation(conv3d(inputs, kernel) + bias).","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_conv_3d.html","id":"input-shape","dir":"Reference","previous_headings":"","what":"Input Shape","title":"3D convolution layer. — layer_conv_3d","text":"data_format=\"channels_last\": 5D tensor shape: (batch_size, spatial_dim1, spatial_dim2, spatial_dim3, channels) data_format=\"channels_first\": 5D tensor shape: (batch_size, channels, spatial_dim1, spatial_dim2, spatial_dim3)","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_conv_3d.html","id":"output-shape","dir":"Reference","previous_headings":"","what":"Output Shape","title":"3D convolution layer. — layer_conv_3d","text":"data_format=\"channels_last\": 5D tensor shape: (batch_size, new_spatial_dim1, new_spatial_dim2, new_spatial_dim3, filters) data_format=\"channels_first\": 5D tensor shape: (batch_size, filters, new_spatial_dim1, new_spatial_dim2, new_spatial_dim3)","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_conv_3d.html","id":"raises","dir":"Reference","previous_headings":"","what":"Raises","title":"3D convolution layer. — layer_conv_3d","text":"ValueError: strides > 1 dilation_rate > 1.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_conv_3d.html","id":"example","dir":"Reference","previous_headings":"","what":"Example","title":"3D convolution layer. — layer_conv_3d","text":"","code":"x <- random_uniform(c(4, 10, 10, 10, 128)) y <- x |> layer_conv_3d(32, 3, activation = 'relu') shape(y) ## shape(4, 8, 8, 8, 32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_conv_3d_transpose.html","id":null,"dir":"Reference","previous_headings":"","what":"3D transposed convolution layer. — layer_conv_3d_transpose","title":"3D transposed convolution layer. — layer_conv_3d_transpose","text":"need transposed convolutions generally arise desire use transformation going opposite direction normal convolution, .e., something shape output convolution something shape input maintaining connectivity pattern compatible said convolution.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_conv_3d_transpose.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"3D transposed convolution layer. — layer_conv_3d_transpose","text":"","code":"layer_conv_3d_transpose(   object,   filters,   kernel_size,   strides = list(1L, 1L, 1L),   padding = \"valid\",   data_format = NULL,   dilation_rate = list(1L, 1L, 1L),   activation = NULL,   use_bias = TRUE,   kernel_initializer = \"glorot_uniform\",   bias_initializer = \"zeros\",   kernel_regularizer = NULL,   bias_regularizer = NULL,   activity_regularizer = NULL,   kernel_constraint = NULL,   bias_constraint = NULL,   ... )"},{"path":"https://keras3.posit.co/dev/reference/layer_conv_3d_transpose.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"3D transposed convolution layer. — layer_conv_3d_transpose","text":"object Object compose layer . tensor, array, sequential model. filters int, dimension output space (number filters transposed convolution). kernel_size int list 1 integer, specifying size transposed convolution window. strides int list 1 integer, specifying stride length transposed convolution. strides > 1 incompatible dilation_rate > 1. padding string, either \"valid\" \"\" (case-insensitive). \"valid\" means padding. \"\" results padding evenly left/right /input. padding=\"\" strides=1, output size input. data_format string, either \"channels_last\" \"channels_first\". ordering dimensions inputs. \"channels_last\" corresponds inputs shape (batch_size, spatial_dim1, spatial_dim2, spatial_dim3, channels) \"channels_first\" corresponds inputs shape (batch_size, channels, spatial_dim1, spatial_dim2, spatial_dim3). defaults image_data_format value found Keras config file ~/.keras/keras.json. never set , \"channels_last\". dilation_rate int list 1 integers, specifying dilation rate use dilated transposed convolution. activation Activation function. NULL, activation applied. use_bias bool, TRUE, bias added output. kernel_initializer Initializer convolution kernel. NULL, default initializer (\"glorot_uniform\") used. bias_initializer Initializer bias vector. NULL, default initializer (\"zeros\") used. kernel_regularizer Optional regularizer convolution kernel. bias_regularizer Optional regularizer bias vector. activity_regularizer Optional regularizer function output. kernel_constraint Optional projection function applied kernel updated Optimizer (e.g. used implement norm constraints value constraints layer weights). function must take input unprojected variable must return projected variable (must shape). Constraints safe use asynchronous distributed training. bias_constraint Optional projection function applied bias updated Optimizer. ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_conv_3d_transpose.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"3D transposed convolution layer. — layer_conv_3d_transpose","text":"5D tensor representing activation(conv3d(inputs, kernel) + bias).","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_conv_3d_transpose.html","id":"input-shape","dir":"Reference","previous_headings":"","what":"Input Shape","title":"3D transposed convolution layer. — layer_conv_3d_transpose","text":"data_format=\"channels_last\": 5D tensor shape: (batch_size, spatial_dim1, spatial_dim2, spatial_dim3, channels) data_format=\"channels_first\": 5D tensor shape: (batch_size, channels, spatial_dim1, spatial_dim2, spatial_dim3)","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_conv_3d_transpose.html","id":"output-shape","dir":"Reference","previous_headings":"","what":"Output Shape","title":"3D transposed convolution layer. — layer_conv_3d_transpose","text":"data_format=\"channels_last\": 5D tensor shape: (batch_size, new_spatial_dim1, new_spatial_dim2, new_spatial_dim3, filters) data_format=\"channels_first\": 5D tensor shape: (batch_size, filters, new_spatial_dim1, new_spatial_dim2, new_spatial_dim3)","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_conv_3d_transpose.html","id":"raises","dir":"Reference","previous_headings":"","what":"Raises","title":"3D transposed convolution layer. — layer_conv_3d_transpose","text":"ValueError: strides > 1 dilation_rate > 1.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_conv_3d_transpose.html","id":"references","dir":"Reference","previous_headings":"","what":"References","title":"3D transposed convolution layer. — layer_conv_3d_transpose","text":"guide convolution arithmetic deep learning Deconvolutional Networks","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_conv_3d_transpose.html","id":"example","dir":"Reference","previous_headings":"","what":"Example","title":"3D transposed convolution layer. — layer_conv_3d_transpose","text":"","code":"x <- random_uniform(c(4, 10, 8, 12, 128)) y <- x |> layer_conv_3d_transpose(32, 2, 2, activation = 'relu') shape(y) ## shape(4, 20, 16, 24, 32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_conv_lstm_1d.html","id":null,"dir":"Reference","previous_headings":"","what":"1D Convolutional LSTM. — layer_conv_lstm_1d","title":"1D Convolutional LSTM. — layer_conv_lstm_1d","text":"Similar LSTM layer, input transformations recurrent transformations convolutional.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_conv_lstm_1d.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"1D Convolutional LSTM. — layer_conv_lstm_1d","text":"","code":"layer_conv_lstm_1d(   object,   filters,   kernel_size,   strides = 1L,   padding = \"valid\",   data_format = NULL,   dilation_rate = 1L,   activation = \"tanh\",   recurrent_activation = \"sigmoid\",   use_bias = TRUE,   kernel_initializer = \"glorot_uniform\",   recurrent_initializer = \"orthogonal\",   bias_initializer = \"zeros\",   unit_forget_bias = TRUE,   kernel_regularizer = NULL,   recurrent_regularizer = NULL,   bias_regularizer = NULL,   activity_regularizer = NULL,   kernel_constraint = NULL,   recurrent_constraint = NULL,   bias_constraint = NULL,   dropout = 0,   recurrent_dropout = 0,   seed = NULL,   return_sequences = FALSE,   return_state = FALSE,   go_backwards = FALSE,   stateful = FALSE,   ...,   unroll = NULL )"},{"path":"https://keras3.posit.co/dev/reference/layer_conv_lstm_1d.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"1D Convolutional LSTM. — layer_conv_lstm_1d","text":"object Object compose layer . tensor, array, sequential model. filters int, dimension output space (number filters convolution). kernel_size int tuple/list 1 integer, specifying size convolution window. strides int tuple/list 1 integer, specifying stride length convolution. strides > 1 incompatible dilation_rate > 1. padding string, \"valid\" \"\" (case-insensitive). \"valid\" means padding. \"\" results padding evenly left/right /input output height/width dimension input. data_format string, either \"channels_last\" \"channels_first\". ordering dimensions inputs. \"channels_last\" corresponds inputs shape (batch, steps, features) \"channels_first\" corresponds inputs shape (batch, features, steps). defaults image_data_format value found Keras config file ~/.keras/keras.json. never set , \"channels_last\". dilation_rate int tuple/list 1 integers, specifying dilation rate use dilated convolution. activation Activation function use. default hyperbolic tangent activation function applied (tanh(x)). recurrent_activation Activation function use recurrent step. use_bias Boolean, whether layer uses bias vector. kernel_initializer Initializer kernel weights matrix, used linear transformation inputs. recurrent_initializer Initializer recurrent_kernel weights matrix, used linear transformation recurrent state. bias_initializer Initializer bias vector. unit_forget_bias Boolean. TRUE, add 1 bias forget gate initialization. Use combination bias_initializer=\"zeros\". recommended Jozefowicz et al., 2015 kernel_regularizer Regularizer function applied kernel weights matrix. recurrent_regularizer Regularizer function applied recurrent_kernel weights matrix. bias_regularizer Regularizer function applied bias vector. activity_regularizer Regularizer function applied . kernel_constraint Constraint function applied kernel weights matrix. recurrent_constraint Constraint function applied recurrent_kernel weights matrix. bias_constraint Constraint function applied bias vector. dropout Float 0 1. Fraction units drop linear transformation inputs. recurrent_dropout Float 0 1. Fraction units drop linear transformation recurrent state. seed Random seed dropout. return_sequences Boolean. Whether return last output output sequence, full sequence. Default: FALSE. return_state Boolean. Whether return last state addition output. Default: FALSE. go_backwards Boolean (default: FALSE). TRUE, process input sequence backwards return reversed sequence. stateful Boolean (default FALSE). TRUE, last state sample index batch used initial state sample index following batch. ... forward/backward compatability. unroll Boolean (default: FALSE). TRUE, network unrolled, else symbolic loop used. Unrolling can speed-RNN, although tends memory-intensive. Unrolling suitable short sequences.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_conv_lstm_1d.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"1D Convolutional LSTM. — layer_conv_lstm_1d","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_conv_lstm_1d.html","id":"call-arguments","dir":"Reference","previous_headings":"","what":"Call Arguments","title":"1D Convolutional LSTM. — layer_conv_lstm_1d","text":"inputs: 4D tensor. initial_state: List initial state tensors passed first call cell. mask: Binary tensor shape (samples, timesteps) indicating whether given timestep masked. training: Python boolean indicating whether layer behave training mode inference mode. relevant dropout recurrent_dropout set.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_conv_lstm_1d.html","id":"input-shape","dir":"Reference","previous_headings":"","what":"Input Shape","title":"1D Convolutional LSTM. — layer_conv_lstm_1d","text":"data_format=\"channels_first\": 4D tensor shape: (samples, time, channels, rows) data_format=\"channels_last\": 4D tensor shape: (samples, time, rows, channels)","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_conv_lstm_1d.html","id":"output-shape","dir":"Reference","previous_headings":"","what":"Output Shape","title":"1D Convolutional LSTM. — layer_conv_lstm_1d","text":"return_state: list tensors. first tensor output. remaining tensors last states, 3D tensor shape: (samples, filters, new_rows) data_format='channels_first' shape: (samples, new_rows, filters) data_format='channels_last'. rows values might changed due padding. return_sequences: 4D tensor shape: (samples, timesteps, filters, new_rows) data_format='channels_first' shape: (samples, timesteps, new_rows, filters) data_format='channels_last'. Else, 3D tensor shape: (samples, filters, new_rows) data_format='channels_first' shape: (samples, new_rows, filters) data_format='channels_last'.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_conv_lstm_1d.html","id":"references","dir":"Reference","previous_headings":"","what":"References","title":"1D Convolutional LSTM. — layer_conv_lstm_1d","text":"Shi et al., 2015 (current implementation include feedback loop cells output).","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_conv_lstm_2d.html","id":null,"dir":"Reference","previous_headings":"","what":"2D Convolutional LSTM. — layer_conv_lstm_2d","title":"2D Convolutional LSTM. — layer_conv_lstm_2d","text":"Similar LSTM layer, input transformations recurrent transformations convolutional.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_conv_lstm_2d.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"2D Convolutional LSTM. — layer_conv_lstm_2d","text":"","code":"layer_conv_lstm_2d(   object,   filters,   kernel_size,   strides = 1L,   padding = \"valid\",   data_format = NULL,   dilation_rate = 1L,   activation = \"tanh\",   recurrent_activation = \"sigmoid\",   use_bias = TRUE,   kernel_initializer = \"glorot_uniform\",   recurrent_initializer = \"orthogonal\",   bias_initializer = \"zeros\",   unit_forget_bias = TRUE,   kernel_regularizer = NULL,   recurrent_regularizer = NULL,   bias_regularizer = NULL,   activity_regularizer = NULL,   kernel_constraint = NULL,   recurrent_constraint = NULL,   bias_constraint = NULL,   dropout = 0,   recurrent_dropout = 0,   seed = NULL,   return_sequences = FALSE,   return_state = FALSE,   go_backwards = FALSE,   stateful = FALSE,   ...,   unroll = NULL )"},{"path":"https://keras3.posit.co/dev/reference/layer_conv_lstm_2d.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"2D Convolutional LSTM. — layer_conv_lstm_2d","text":"object Object compose layer . tensor, array, sequential model. filters int, dimension output space (number filters convolution). kernel_size int tuple/list 2 integers, specifying size convolution window. strides int tuple/list 2 integers, specifying stride length convolution. strides > 1 incompatible dilation_rate > 1. padding string, \"valid\" \"\" (case-insensitive). \"valid\" means padding. \"\" results padding evenly left/right /input output height/width dimension input. data_format string, either \"channels_last\" \"channels_first\". ordering dimensions inputs. \"channels_last\" corresponds inputs shape (batch, steps, features) \"channels_first\" corresponds inputs shape (batch, features, steps). defaults image_data_format value found Keras config file ~/.keras/keras.json. never set , \"channels_last\". dilation_rate int tuple/list 2 integers, specifying dilation rate use dilated convolution. activation Activation function use. default hyperbolic tangent activation function applied (tanh(x)). recurrent_activation Activation function use recurrent step. use_bias Boolean, whether layer uses bias vector. kernel_initializer Initializer kernel weights matrix, used linear transformation inputs. recurrent_initializer Initializer recurrent_kernel weights matrix, used linear transformation recurrent state. bias_initializer Initializer bias vector. unit_forget_bias Boolean. TRUE, add 1 bias forget gate initialization. Use combination bias_initializer=\"zeros\". recommended Jozefowicz et al., 2015 kernel_regularizer Regularizer function applied kernel weights matrix. recurrent_regularizer Regularizer function applied recurrent_kernel weights matrix. bias_regularizer Regularizer function applied bias vector. activity_regularizer Regularizer function applied . kernel_constraint Constraint function applied kernel weights matrix. recurrent_constraint Constraint function applied recurrent_kernel weights matrix. bias_constraint Constraint function applied bias vector. dropout Float 0 1. Fraction units drop linear transformation inputs. recurrent_dropout Float 0 1. Fraction units drop linear transformation recurrent state. seed Random seed dropout. return_sequences Boolean. Whether return last output output sequence, full sequence. Default: FALSE. return_state Boolean. Whether return last state addition output. Default: FALSE. go_backwards Boolean (default: FALSE). TRUE, process input sequence backwards return reversed sequence. stateful Boolean (default FALSE). TRUE, last state sample index batch used initial state sample index following batch. ... forward/backward compatability. unroll Boolean (default: FALSE). TRUE, network unrolled, else symbolic loop used. Unrolling can speed-RNN, although tends memory-intensive. Unrolling suitable short sequences.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_conv_lstm_2d.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"2D Convolutional LSTM. — layer_conv_lstm_2d","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_conv_lstm_2d.html","id":"call-arguments","dir":"Reference","previous_headings":"","what":"Call Arguments","title":"2D Convolutional LSTM. — layer_conv_lstm_2d","text":"inputs: 5D tensor. mask: Binary tensor shape (samples, timesteps) indicating whether given timestep masked. training: Python boolean indicating whether layer behave training mode inference mode. relevant dropout recurrent_dropout set. initial_state: List initial state tensors passed first call cell.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_conv_lstm_2d.html","id":"input-shape","dir":"Reference","previous_headings":"","what":"Input Shape","title":"2D Convolutional LSTM. — layer_conv_lstm_2d","text":"data_format='channels_first': 5D tensor shape: (samples, time, channels, rows, cols) data_format='channels_last': 5D tensor shape: (samples, time, rows, cols, channels)","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_conv_lstm_2d.html","id":"output-shape","dir":"Reference","previous_headings":"","what":"Output Shape","title":"2D Convolutional LSTM. — layer_conv_lstm_2d","text":"return_state: list tensors. first tensor output. remaining tensors last states, 4D tensor shape: (samples, filters, new_rows, new_cols) data_format='channels_first' shape: (samples, new_rows, new_cols, filters) data_format='channels_last'. rows cols values might changed due padding. return_sequences: 5D tensor shape: (samples, timesteps, filters, new_rows, new_cols) data_format='channels_first' shape: (samples, timesteps, new_rows, new_cols, filters) data_format='channels_last'. Else, 4D tensor shape: (samples, filters, new_rows, new_cols) data_format='channels_first' shape: (samples, new_rows, new_cols, filters) data_format='channels_last'.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_conv_lstm_2d.html","id":"references","dir":"Reference","previous_headings":"","what":"References","title":"2D Convolutional LSTM. — layer_conv_lstm_2d","text":"Shi et al., 2015 (current implementation include feedback loop cells output).","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_conv_lstm_3d.html","id":null,"dir":"Reference","previous_headings":"","what":"3D Convolutional LSTM. — layer_conv_lstm_3d","title":"3D Convolutional LSTM. — layer_conv_lstm_3d","text":"Similar LSTM layer, input transformations recurrent transformations convolutional.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_conv_lstm_3d.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"3D Convolutional LSTM. — layer_conv_lstm_3d","text":"","code":"layer_conv_lstm_3d(   object,   filters,   kernel_size,   strides = 1L,   padding = \"valid\",   data_format = NULL,   dilation_rate = 1L,   activation = \"tanh\",   recurrent_activation = \"sigmoid\",   use_bias = TRUE,   kernel_initializer = \"glorot_uniform\",   recurrent_initializer = \"orthogonal\",   bias_initializer = \"zeros\",   unit_forget_bias = TRUE,   kernel_regularizer = NULL,   recurrent_regularizer = NULL,   bias_regularizer = NULL,   activity_regularizer = NULL,   kernel_constraint = NULL,   recurrent_constraint = NULL,   bias_constraint = NULL,   dropout = 0,   recurrent_dropout = 0,   seed = NULL,   return_sequences = FALSE,   return_state = FALSE,   go_backwards = FALSE,   stateful = FALSE,   ...,   unroll = NULL )"},{"path":"https://keras3.posit.co/dev/reference/layer_conv_lstm_3d.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"3D Convolutional LSTM. — layer_conv_lstm_3d","text":"object Object compose layer . tensor, array, sequential model. filters int, dimension output space (number filters convolution). kernel_size int tuple/list 3 integers, specifying size convolution window. strides int tuple/list 3 integers, specifying stride length convolution. strides > 1 incompatible dilation_rate > 1. padding string, \"valid\" \"\" (case-insensitive). \"valid\" means padding. \"\" results padding evenly left/right /input output height/width dimension input. data_format string, either \"channels_last\" \"channels_first\". ordering dimensions inputs. \"channels_last\" corresponds inputs shape (batch, steps, features) \"channels_first\" corresponds inputs shape (batch, features, steps). defaults image_data_format value found Keras config file ~/.keras/keras.json. never set , \"channels_last\". dilation_rate int tuple/list 3 integers, specifying dilation rate use dilated convolution. activation Activation function use. default hyperbolic tangent activation function applied (tanh(x)). recurrent_activation Activation function use recurrent step. use_bias Boolean, whether layer uses bias vector. kernel_initializer Initializer kernel weights matrix, used linear transformation inputs. recurrent_initializer Initializer recurrent_kernel weights matrix, used linear transformation recurrent state. bias_initializer Initializer bias vector. unit_forget_bias Boolean. TRUE, add 1 bias forget gate initialization. Use combination bias_initializer=\"zeros\". recommended Jozefowicz et al., 2015 kernel_regularizer Regularizer function applied kernel weights matrix. recurrent_regularizer Regularizer function applied recurrent_kernel weights matrix. bias_regularizer Regularizer function applied bias vector. activity_regularizer Regularizer function applied . kernel_constraint Constraint function applied kernel weights matrix. recurrent_constraint Constraint function applied recurrent_kernel weights matrix. bias_constraint Constraint function applied bias vector. dropout Float 0 1. Fraction units drop linear transformation inputs. recurrent_dropout Float 0 1. Fraction units drop linear transformation recurrent state. seed Random seed dropout. return_sequences Boolean. Whether return last output output sequence, full sequence. Default: FALSE. return_state Boolean. Whether return last state addition output. Default: FALSE. go_backwards Boolean (default: FALSE). TRUE, process input sequence backwards return reversed sequence. stateful Boolean (default FALSE). TRUE, last state sample index batch used initial state sample index following batch. ... forward/backward compatability. unroll Boolean (default: FALSE). TRUE, network unrolled, else symbolic loop used. Unrolling can speed-RNN, although tends memory-intensive. Unrolling suitable short sequences.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_conv_lstm_3d.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"3D Convolutional LSTM. — layer_conv_lstm_3d","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_conv_lstm_3d.html","id":"call-arguments","dir":"Reference","previous_headings":"","what":"Call Arguments","title":"3D Convolutional LSTM. — layer_conv_lstm_3d","text":"inputs: 6D tensor. mask: Binary tensor shape (samples, timesteps) indicating whether given timestep masked. training: Python boolean indicating whether layer behave training mode inference mode. relevant dropout recurrent_dropout set. initial_state: List initial state tensors passed first call cell.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_conv_lstm_3d.html","id":"input-shape","dir":"Reference","previous_headings":"","what":"Input Shape","title":"3D Convolutional LSTM. — layer_conv_lstm_3d","text":"data_format='channels_first': 5D tensor shape: (samples, time, channels, *spatial_dims) data_format='channels_last': 5D tensor shape: (samples, time, *spatial_dims, channels)","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_conv_lstm_3d.html","id":"output-shape","dir":"Reference","previous_headings":"","what":"Output Shape","title":"3D Convolutional LSTM. — layer_conv_lstm_3d","text":"return_state: list tensors. first tensor output. remaining tensors last states, 4D tensor shape: (samples, filters, *spatial_dims) data_format='channels_first' shape: (samples, *spatial_dims, filters) data_format='channels_last'. return_sequences: 5D tensor shape: (samples, timesteps, filters, *spatial_dims) data_format='channels_first' shape: (samples, timesteps, *spatial_dims, filters) data_format='channels_last'. Else, 4D tensor shape: (samples, filters, *spatial_dims) data_format='channels_first' shape: (samples, *spatial_dims, filters) data_format='channels_last'.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_conv_lstm_3d.html","id":"references","dir":"Reference","previous_headings":"","what":"References","title":"3D Convolutional LSTM. — layer_conv_lstm_3d","text":"Shi et al., 2015 (current implementation include feedback loop cells output).","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_cropping_1d.html","id":null,"dir":"Reference","previous_headings":"","what":"Cropping layer for 1D input (e.g. temporal sequence). — layer_cropping_1d","title":"Cropping layer for 1D input (e.g. temporal sequence). — layer_cropping_1d","text":"crops along time dimension (axis 2).","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_cropping_1d.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Cropping layer for 1D input (e.g. temporal sequence). — layer_cropping_1d","text":"","code":"layer_cropping_1d(object, cropping = list(1L, 1L), ...)"},{"path":"https://keras3.posit.co/dev/reference/layer_cropping_1d.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Cropping layer for 1D input (e.g. temporal sequence). — layer_cropping_1d","text":"object Object compose layer . tensor, array, sequential model. cropping Int, list int (length 2). int: many units trimmed beginning end cropping dimension (axis 1). list 2 ints: many units trimmed beginning end cropping dimension ((left_crop, right_crop)). ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_cropping_1d.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Cropping layer for 1D input (e.g. temporal sequence). — layer_cropping_1d","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_cropping_1d.html","id":"example","dir":"Reference","previous_headings":"","what":"Example","title":"Cropping layer for 1D input (e.g. temporal sequence). — layer_cropping_1d","text":"","code":"input_shape <- c(2, 3, 2) x <- op_arange(prod(input_shape)) |> op_reshape(input_shape) x ## tf.Tensor( ## [[[ 1.  2.] ##   [ 3.  4.] ##   [ 5.  6.]] ## ##  [[ 7.  8.] ##   [ 9. 10.] ##   [11. 12.]]], shape=(2, 3, 2), dtype=float32) y <- x |> layer_cropping_1d(cropping = 1) y ## tf.Tensor( ## [[[ 3.  4.]] ## ##  [[ 9. 10.]]], shape=(2, 1, 2), dtype=float32)"},{"path":"https://keras3.posit.co/dev/reference/layer_cropping_1d.html","id":"input-shape","dir":"Reference","previous_headings":"","what":"Input Shape","title":"Cropping layer for 1D input (e.g. temporal sequence). — layer_cropping_1d","text":"3D tensor shape (batch_size, axis_to_crop, features)","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_cropping_1d.html","id":"output-shape","dir":"Reference","previous_headings":"","what":"Output Shape","title":"Cropping layer for 1D input (e.g. temporal sequence). — layer_cropping_1d","text":"3D tensor shape (batch_size, cropped_axis, features)","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_cropping_2d.html","id":null,"dir":"Reference","previous_headings":"","what":"Cropping layer for 2D input (e.g. picture). — layer_cropping_2d","title":"Cropping layer for 2D input (e.g. picture). — layer_cropping_2d","text":"crops along spatial dimensions, .e. height width.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_cropping_2d.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Cropping layer for 2D input (e.g. picture). — layer_cropping_2d","text":"","code":"layer_cropping_2d(   object,   cropping = list(list(0L, 0L), list(0L, 0L)),   data_format = NULL,   ... )"},{"path":"https://keras3.posit.co/dev/reference/layer_cropping_2d.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Cropping layer for 2D input (e.g. picture). — layer_cropping_2d","text":"object Object compose layer . tensor, array, sequential model. cropping Int, list 2 ints, list 2 lists 2 ints. int: symmetric cropping applied height width. list 2 ints: interpreted two different symmetric cropping values height width: (symmetric_height_crop, symmetric_width_crop). list 2 lists 2 ints: interpreted ((top_crop, bottom_crop), (left_crop, right_crop)). data_format string, one \"channels_last\" (default) \"channels_first\". ordering dimensions inputs. \"channels_last\" corresponds inputs shape (batch_size, height, width, channels) \"channels_first\" corresponds inputs shape (batch_size, channels, height, width). unspecified, uses image_data_format value found Keras config file ~/.keras/keras.json (exists). Defaults \"channels_last\". ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_cropping_2d.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Cropping layer for 2D input (e.g. picture). — layer_cropping_2d","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_cropping_2d.html","id":"example","dir":"Reference","previous_headings":"","what":"Example","title":"Cropping layer for 2D input (e.g. picture). — layer_cropping_2d","text":"","code":"input_shape <- c(2, 28, 28, 3) x <- op_arange(prod(input_shape), dtype ='int32') |> op_reshape(input_shape) y <- x |> layer_cropping_2d(cropping=list(c(2, 2), c(4, 4))) shape(y) ## shape(2, 24, 20, 3)"},{"path":"https://keras3.posit.co/dev/reference/layer_cropping_2d.html","id":"input-shape","dir":"Reference","previous_headings":"","what":"Input Shape","title":"Cropping layer for 2D input (e.g. picture). — layer_cropping_2d","text":"4D tensor shape: data_format \"channels_last\": (batch_size, height, width, channels) data_format \"channels_first\": (batch_size, channels, height, width)","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_cropping_2d.html","id":"output-shape","dir":"Reference","previous_headings":"","what":"Output Shape","title":"Cropping layer for 2D input (e.g. picture). — layer_cropping_2d","text":"4D tensor shape: data_format \"channels_last\": (batch_size, cropped_height, cropped_width, channels) data_format \"channels_first\": (batch_size, channels, cropped_height, cropped_width)","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_cropping_3d.html","id":null,"dir":"Reference","previous_headings":"","what":"Cropping layer for 3D data (e.g. spatial or spatio-temporal). — layer_cropping_3d","title":"Cropping layer for 3D data (e.g. spatial or spatio-temporal). — layer_cropping_3d","text":"Cropping layer 3D data (e.g. spatial spatio-temporal).","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_cropping_3d.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Cropping layer for 3D data (e.g. spatial or spatio-temporal). — layer_cropping_3d","text":"","code":"layer_cropping_3d(   object,   cropping = list(list(1L, 1L), list(1L, 1L), list(1L, 1L)),   data_format = NULL,   ... )"},{"path":"https://keras3.posit.co/dev/reference/layer_cropping_3d.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Cropping layer for 3D data (e.g. spatial or spatio-temporal). — layer_cropping_3d","text":"object Object compose layer . tensor, array, sequential model. cropping Int, list 3 ints, list 3 lists 2 ints. int: symmetric cropping applied depth, height, width. list 3 ints: interpreted three different symmetric cropping values depth, height, width: (symmetric_dim1_crop, symmetric_dim2_crop, symmetric_dim3_crop). list 3 lists 2 ints: interpreted ((left_dim1_crop, right_dim1_crop), (left_dim2_crop, right_dim2_crop), (left_dim3_crop, right_dim3_crop)). data_format string, one \"channels_last\" (default) \"channels_first\". ordering dimensions inputs. \"channels_last\" corresponds inputs shape (batch_size, spatial_dim1, spatial_dim2, spatial_dim3, channels) \"channels_first\" corresponds inputs shape (batch_size, channels, spatial_dim1, spatial_dim2, spatial_dim3). unspecified, uses image_data_format value found Keras config file ~/.keras/keras.json (exists). Defaults \"channels_last\". ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_cropping_3d.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Cropping layer for 3D data (e.g. spatial or spatio-temporal). — layer_cropping_3d","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_cropping_3d.html","id":"example","dir":"Reference","previous_headings":"","what":"Example","title":"Cropping layer for 3D data (e.g. spatial or spatio-temporal). — layer_cropping_3d","text":"","code":"input_shape <- c(2, 28, 28, 10, 3) x <- input_shape %>% { op_reshape(seq(prod(.)), .) } y <- x |> layer_cropping_3d(cropping = c(2, 4, 2)) shape(y) ## shape(2, 24, 20, 6, 3)"},{"path":"https://keras3.posit.co/dev/reference/layer_cropping_3d.html","id":"input-shape","dir":"Reference","previous_headings":"","what":"Input Shape","title":"Cropping layer for 3D data (e.g. spatial or spatio-temporal). — layer_cropping_3d","text":"5D tensor shape: data_format \"channels_last\": (batch_size, first_axis_to_crop, second_axis_to_crop, third_axis_to_crop, channels) data_format \"channels_first\": (batch_size, channels, first_axis_to_crop, second_axis_to_crop, third_axis_to_crop)","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_cropping_3d.html","id":"output-shape","dir":"Reference","previous_headings":"","what":"Output Shape","title":"Cropping layer for 3D data (e.g. spatial or spatio-temporal). — layer_cropping_3d","text":"5D tensor shape: data_format \"channels_last\": (batch_size, first_cropped_axis, second_cropped_axis, third_cropped_axis, channels) data_format \"channels_first\": (batch_size, channels, first_cropped_axis, second_cropped_axis, third_cropped_axis)","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_cut_mix.html","id":null,"dir":"Reference","previous_headings":"","what":"CutMix data augmentation technique. — layer_cut_mix","title":"CutMix data augmentation technique. — layer_cut_mix","text":"CutMix data augmentation method patches cut pasted two images dataset, labels also mixed proportionally area patches.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_cut_mix.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"CutMix data augmentation technique. — layer_cut_mix","text":"","code":"layer_cut_mix(object, factor = 1, seed = NULL, data_format = NULL, ...)"},{"path":"https://keras3.posit.co/dev/reference/layer_cut_mix.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"CutMix data augmentation technique. — layer_cut_mix","text":"object Object compose layer . tensor, array, sequential model. factor single float tuple two floats 0 1. tuple numbers passed, factor sampled two values. single float passed, value 0 passed float sampled. values define range mixing weight sampled. higher factor increases variability patch sizes, leading diverse larger mixed patches. Defaults 1. seed Integer. Used create random seed. data_format string, either \"channels_last\" \"channels_first\". ordering dimensions inputs. \"channels_last\" corresponds inputs shape (batch, height, width, channels) \"channels_first\" corresponds inputs shape (batch, channels, height, width). defaults image_data_format value found Keras config file ~/.keras/keras.json. never set , \"channels_last\". ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_cut_mix.html","id":"references","dir":"Reference","previous_headings":"","what":"References","title":"CutMix data augmentation technique. — layer_cut_mix","text":"CutMix paper.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_dense.html","id":null,"dir":"Reference","previous_headings":"","what":"Just your regular densely-connected NN layer. — layer_dense","title":"Just your regular densely-connected NN layer. — layer_dense","text":"Dense implements operation: output = activation(dot(input, kernel) + bias) activation element-wise activation function passed activation argument, kernel weights matrix created layer, bias bias vector created layer (applicable use_bias TRUE).","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_dense.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Just your regular densely-connected NN layer. — layer_dense","text":"","code":"layer_dense(   object,   units,   activation = NULL,   use_bias = TRUE,   kernel_initializer = \"glorot_uniform\",   bias_initializer = \"zeros\",   kernel_regularizer = NULL,   bias_regularizer = NULL,   activity_regularizer = NULL,   kernel_constraint = NULL,   bias_constraint = NULL,   lora_rank = NULL,   ... )"},{"path":"https://keras3.posit.co/dev/reference/layer_dense.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Just your regular densely-connected NN layer. — layer_dense","text":"object Object compose layer . tensor, array, sequential model. units Positive integer, dimensionality output space. activation Activation function use. specify anything, activation applied (ie. \"linear\" activation: (x) = x). use_bias Boolean, whether layer uses bias vector. kernel_initializer Initializer kernel weights matrix. bias_initializer Initializer bias vector. kernel_regularizer Regularizer function applied kernel weights matrix. bias_regularizer Regularizer function applied bias vector. activity_regularizer Regularizer function applied output layer (\"activation\"). kernel_constraint Constraint function applied kernel weights matrix. bias_constraint Constraint function applied bias vector. lora_rank Optional integer. set, layer's forward pass implement LoRA (Low-Rank Adaptation) provided rank. LoRA sets layer's kernel non-trainable replaces delta original kernel, obtained via multiplying two lower-rank trainable matrices. can useful reduce computation cost fine-tuning large dense layers. can also enable LoRA existing Dense layer calling layer$enable_lora(rank). ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_dense.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Just your regular densely-connected NN layer. — layer_dense","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_dense.html","id":"note","dir":"Reference","previous_headings":"","what":"Note","title":"Just your regular densely-connected NN layer. — layer_dense","text":"input layer rank greater 2, Dense computes dot product inputs kernel along last axis inputs axis 0 kernel (using tf.tensordot). example, input dimensions (batch_size, d0, d1), create kernel shape (d1, units), kernel operates along axis 2 input, every sub-tensor shape (1, 1, d1) (batch_size * d0 sub-tensors). output case shape (batch_size, d0, units).","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_dense.html","id":"input-shape","dir":"Reference","previous_headings":"","what":"Input Shape","title":"Just your regular densely-connected NN layer. — layer_dense","text":"N-D tensor shape: (batch_size, ..., input_dim). common situation 2D input shape (batch_size, input_dim).","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_dense.html","id":"output-shape","dir":"Reference","previous_headings":"","what":"Output Shape","title":"Just your regular densely-connected NN layer. — layer_dense","text":"N-D tensor shape: (batch_size, ..., units). instance, 2D input shape (batch_size, input_dim), output shape (batch_size, units).","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_dense.html","id":"methods","dir":"Reference","previous_headings":"","what":"Methods","title":"Just your regular densely-connected NN layer. — layer_dense","text":"","code":"enable_lora(   rank,   a_initializer = 'he_uniform',   b_initializer = 'zeros' ) quantize(mode, type_check = TRUE)"},{"path":"https://keras3.posit.co/dev/reference/layer_dense.html","id":"readonly-properties-","dir":"Reference","previous_headings":"","what":"Readonly properties:","title":"Just your regular densely-connected NN layer. — layer_dense","text":"kernel","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_depthwise_conv_1d.html","id":null,"dir":"Reference","previous_headings":"","what":"1D depthwise convolution layer. — layer_depthwise_conv_1d","title":"1D depthwise convolution layer. — layer_depthwise_conv_1d","text":"Depthwise convolution type convolution input channel convolved different kernel (called depthwise kernel). can understand depthwise convolution first step depthwise separable convolution. implemented via following steps: Split input individual channels. Convolve channel individual depthwise kernel depth_multiplier output channels. Concatenate convolved outputs along channels axis. Unlike regular 1D convolution, depthwise convolution mix information across different input channels. depth_multiplier argument determines many filters applied one input channel. , controls amount output channels generated per input channel depthwise step.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_depthwise_conv_1d.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"1D depthwise convolution layer. — layer_depthwise_conv_1d","text":"","code":"layer_depthwise_conv_1d(   object,   kernel_size,   strides = 1L,   padding = \"valid\",   depth_multiplier = 1L,   data_format = NULL,   dilation_rate = 1L,   activation = NULL,   use_bias = TRUE,   depthwise_initializer = \"glorot_uniform\",   bias_initializer = \"zeros\",   depthwise_regularizer = NULL,   bias_regularizer = NULL,   activity_regularizer = NULL,   depthwise_constraint = NULL,   bias_constraint = NULL,   ... )"},{"path":"https://keras3.posit.co/dev/reference/layer_depthwise_conv_1d.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"1D depthwise convolution layer. — layer_depthwise_conv_1d","text":"object Object compose layer . tensor, array, sequential model. kernel_size int list 1 integer, specifying size depthwise convolution window. strides int list 1 integer, specifying stride length convolution. strides > 1 incompatible dilation_rate > 1. padding string, either \"valid\" \"\" (case-insensitive). \"valid\" means padding. \"\" results padding evenly left/right /input. padding=\"\" strides=1, output size input. depth_multiplier number depthwise convolution output channels input channel. total number depthwise convolution output channels equal input_channel * depth_multiplier. data_format string, either \"channels_last\" \"channels_first\". ordering dimensions inputs. \"channels_last\" corresponds inputs shape (batch, steps, features) \"channels_first\" corresponds inputs shape (batch, features, steps). defaults image_data_format value found Keras config file ~/.keras/keras.json. never set , \"channels_last\". dilation_rate int list 1 integers, specifying dilation rate use dilated convolution. activation Activation function. NULL, activation applied. use_bias bool, TRUE, bias added output. depthwise_initializer Initializer convolution kernel. NULL, default initializer (\"glorot_uniform\") used. bias_initializer Initializer bias vector. NULL, default initializer (\"zeros\") used. depthwise_regularizer Optional regularizer convolution kernel. bias_regularizer Optional regularizer bias vector. activity_regularizer Optional regularizer function output. depthwise_constraint Optional projection function applied kernel updated Optimizer (e.g. used implement norm constraints value constraints layer weights). function must take input unprojected variable must return projected variable (must shape). Constraints safe use asynchronous distributed training. bias_constraint Optional projection function applied bias updated Optimizer. ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_depthwise_conv_1d.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"1D depthwise convolution layer. — layer_depthwise_conv_1d","text":"3D tensor representing activation(depthwise_conv1d(inputs, kernel) + bias).","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_depthwise_conv_1d.html","id":"input-shape","dir":"Reference","previous_headings":"","what":"Input Shape","title":"1D depthwise convolution layer. — layer_depthwise_conv_1d","text":"data_format=\"channels_last\": 3D tensor shape: (batch_shape, steps, channels) data_format=\"channels_first\": 3D tensor shape: (batch_shape, channels, steps)","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_depthwise_conv_1d.html","id":"output-shape","dir":"Reference","previous_headings":"","what":"Output Shape","title":"1D depthwise convolution layer. — layer_depthwise_conv_1d","text":"data_format=\"channels_last\": 3D tensor shape: (batch_shape, new_steps, channels * depth_multiplier) data_format=\"channels_first\": 3D tensor shape: (batch_shape, channels * depth_multiplier, new_steps)","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_depthwise_conv_1d.html","id":"raises","dir":"Reference","previous_headings":"","what":"Raises","title":"1D depthwise convolution layer. — layer_depthwise_conv_1d","text":"ValueError: strides > 1 dilation_rate > 1.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_depthwise_conv_1d.html","id":"example","dir":"Reference","previous_headings":"","what":"Example","title":"1D depthwise convolution layer. — layer_depthwise_conv_1d","text":"","code":"x <- random_uniform(c(4, 10, 12)) y <- x |> layer_depthwise_conv_1d(   kernel_size = 3,   depth_multiplier = 3,   activation = 'relu' ) shape(y) ## shape(4, 8, 36)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_depthwise_conv_2d.html","id":null,"dir":"Reference","previous_headings":"","what":"2D depthwise convolution layer. — layer_depthwise_conv_2d","title":"2D depthwise convolution layer. — layer_depthwise_conv_2d","text":"Depthwise convolution type convolution input channel convolved different kernel (called depthwise kernel). can understand depthwise convolution first step depthwise separable convolution. implemented via following steps: Split input individual channels. Convolve channel individual depthwise kernel depth_multiplier output channels. Concatenate convolved outputs along channels axis. Unlike regular 2D convolution, depthwise convolution mix information across different input channels. depth_multiplier argument determines many filters applied one input channel. , controls amount output channels generated per input channel depthwise step.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_depthwise_conv_2d.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"2D depthwise convolution layer. — layer_depthwise_conv_2d","text":"","code":"layer_depthwise_conv_2d(   object,   kernel_size,   strides = list(1L, 1L),   padding = \"valid\",   depth_multiplier = 1L,   data_format = NULL,   dilation_rate = list(1L, 1L),   activation = NULL,   use_bias = TRUE,   depthwise_initializer = \"glorot_uniform\",   bias_initializer = \"zeros\",   depthwise_regularizer = NULL,   bias_regularizer = NULL,   activity_regularizer = NULL,   depthwise_constraint = NULL,   bias_constraint = NULL,   ... )"},{"path":"https://keras3.posit.co/dev/reference/layer_depthwise_conv_2d.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"2D depthwise convolution layer. — layer_depthwise_conv_2d","text":"object Object compose layer . tensor, array, sequential model. kernel_size int list 2 integer, specifying size depthwise convolution window. strides int list 2 integer, specifying stride length depthwise convolution. strides > 1 incompatible dilation_rate > 1. padding string, either \"valid\" \"\" (case-insensitive). \"valid\" means padding. \"\" results padding evenly left/right /input. padding=\"\" strides=1, output size input. depth_multiplier number depthwise convolution output channels input channel. total number depthwise convolution output channels equal input_channel * depth_multiplier. data_format string, either \"channels_last\" \"channels_first\". ordering dimensions inputs. \"channels_last\" corresponds inputs shape (batch, height, width, channels) \"channels_first\" corresponds inputs shape (batch, channels, height, width). defaults image_data_format value found Keras config file ~/.keras/keras.json. never set , \"channels_last\". dilation_rate int list 2 integers, specifying dilation rate use dilated convolution. activation Activation function. NULL, activation applied. use_bias bool, TRUE, bias added output. depthwise_initializer Initializer convolution kernel. NULL, default initializer (\"glorot_uniform\") used. bias_initializer Initializer bias vector. NULL, default initializer (\"zeros\") used. depthwise_regularizer Optional regularizer convolution kernel. bias_regularizer Optional regularizer bias vector. activity_regularizer Optional regularizer function output. depthwise_constraint Optional projection function applied kernel updated Optimizer (e.g. used implement norm constraints value constraints layer weights). function must take input unprojected variable must return projected variable (must shape). Constraints safe use asynchronous distributed training. bias_constraint Optional projection function applied bias updated Optimizer. ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_depthwise_conv_2d.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"2D depthwise convolution layer. — layer_depthwise_conv_2d","text":"4D tensor representing activation(depthwise_conv2d(inputs, kernel) + bias).","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_depthwise_conv_2d.html","id":"input-shape","dir":"Reference","previous_headings":"","what":"Input Shape","title":"2D depthwise convolution layer. — layer_depthwise_conv_2d","text":"data_format=\"channels_last\": 4D tensor shape: (batch_size, height, width, channels) data_format=\"channels_first\": 4D tensor shape: (batch_size, channels, height, width)","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_depthwise_conv_2d.html","id":"output-shape","dir":"Reference","previous_headings":"","what":"Output Shape","title":"2D depthwise convolution layer. — layer_depthwise_conv_2d","text":"data_format=\"channels_last\": 4D tensor shape: (batch_size, new_height, new_width, channels * depth_multiplier) data_format=\"channels_first\": 4D tensor shape: (batch_size, channels * depth_multiplier, new_height, new_width)","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_depthwise_conv_2d.html","id":"raises","dir":"Reference","previous_headings":"","what":"Raises","title":"2D depthwise convolution layer. — layer_depthwise_conv_2d","text":"ValueError: strides > 1 dilation_rate > 1.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_depthwise_conv_2d.html","id":"example","dir":"Reference","previous_headings":"","what":"Example","title":"2D depthwise convolution layer. — layer_depthwise_conv_2d","text":"","code":"x <- random_uniform(c(4, 10, 10, 12)) y <- x |> layer_depthwise_conv_2d(kernel_size = 3, activation = 'relu') shape(y) ## shape(4, 8, 8, 12)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_discretization.html","id":null,"dir":"Reference","previous_headings":"","what":"A preprocessing layer which buckets continuous features by ranges. — layer_discretization","title":"A preprocessing layer which buckets continuous features by ranges. — layer_discretization","text":"layer place element input data one several contiguous ranges output integer index indicating range element placed . Note: layer safe use inside tf.data pipeline (independently backend using).","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_discretization.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"A preprocessing layer which buckets continuous features by ranges. — layer_discretization","text":"","code":"layer_discretization(   object,   bin_boundaries = NULL,   num_bins = NULL,   epsilon = 0.01,   output_mode = \"int\",   sparse = FALSE,   dtype = NULL,   name = NULL )"},{"path":"https://keras3.posit.co/dev/reference/layer_discretization.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"A preprocessing layer which buckets continuous features by ranges. — layer_discretization","text":"object Object compose layer . tensor, array, sequential model. bin_boundaries list bin boundaries. leftmost rightmost bins always extend -Inf Inf, bin_boundaries = c(0, 1, 2) generates bins (-Inf, 0), [0, 1), [1, 2), [2, +Inf). option set, adapt() called. num_bins integer number bins compute. option set, bin_boundaries set adapt() called learn bin boundaries. epsilon Error tolerance, typically small fraction close zero (e.g. 0.01). Higher values epsilon increase quantile approximation, hence result unequal buckets, improve performance resource consumption. output_mode Specification output layer. Values can \"int\", \"one_hot\", \"multi_hot\", \"count\" configuring layer follows: \"int\": Return discretized bin indices directly. \"one_hot\": Encodes individual element input array size num_bins, containing 1 input's bin index. last dimension size 1, encode dimension.  last dimension size 1, append new dimension encoded output. \"multi_hot\": Encodes sample input single array size num_bins, containing 1 bin index index present sample. Treats last dimension sample dimension, input shape (..., sample_length), output shape (..., num_tokens). \"count\": \"multi_hot\", int array contains count number times bin index appeared sample. Defaults \"int\". sparse Boolean. applicable \"one_hot\", \"multi_hot\", \"count\" output modes. supported TensorFlow backend. TRUE, returns SparseTensor instead dense Tensor. Defaults FALSE. dtype datatype (e.g., \"float32\"). name String, name object","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_discretization.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"A preprocessing layer which buckets continuous features by ranges. — layer_discretization","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_discretization.html","id":"input-shape","dir":"Reference","previous_headings":"","what":"Input Shape","title":"A preprocessing layer which buckets continuous features by ranges. — layer_discretization","text":"array dimension 2 higher.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_discretization.html","id":"output-shape","dir":"Reference","previous_headings":"","what":"Output Shape","title":"A preprocessing layer which buckets continuous features by ranges. — layer_discretization","text":"input shape.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_discretization.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"A preprocessing layer which buckets continuous features by ranges. — layer_discretization","text":"Discretize float values based provided buckets.     Discretize float values based number buckets compute.","code":"input <- op_array(rbind(c(-1.5, 1, 3.4, 0.5),                        c(0, 3, 1.3, 0),                        c(-.5, 0, .5, 1),                        c(1.5, 2, 2.5, 3))) output <- input |> layer_discretization(bin_boundaries = c(0, 1, 2)) output ## tf.Tensor( ## [[0 2 3 1] ##  [1 3 2 1] ##  [0 1 1 2] ##  [2 3 3 3]], shape=(4, 4), dtype=int64) layer <- layer_discretization(num_bins = 4, epsilon = 0.01) layer |> adapt(input) layer(input) ## tf.Tensor( ## [[0 2 3 1] ##  [1 3 2 1] ##  [0 1 1 2] ##  [2 3 3 3]], shape=(4, 4), dtype=int64)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_dot.html","id":null,"dir":"Reference","previous_headings":"","what":"Computes element-wise dot product of two tensors. — layer_dot","title":"Computes element-wise dot product of two tensors. — layer_dot","text":"takes list inputs size 2, axes corresponding input along dot product performed. say x y two input tensors shapes (2, 3, 5) (2, 10, 3). batch dimension size inputs, axes correspond dimensions size corresponding inputs. e.g. axes = c(1, 2), dot product x, y result tensor shape (2, 5, 10)","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_dot.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes element-wise dot product of two tensors. — layer_dot","text":"","code":"layer_dot(inputs, ..., axes, normalize = FALSE)"},{"path":"https://keras3.posit.co/dev/reference/layer_dot.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Computes element-wise dot product of two tensors. — layer_dot","text":"inputs layers combine ... Standard layer keyword arguments. axes Integer list integers, axis axes along take dot product. list, two integers corresponding desired axis first input desired axis second input, respectively. Note size two selected axes must match. normalize Whether L2-normalize samples along dot product axis taking dot product. set TRUE, output dot product cosine proximity two samples.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_dot.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Computes element-wise dot product of two tensors. — layer_dot","text":"tensor, dot product samples inputs.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_dot.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Computes element-wise dot product of two tensors. — layer_dot","text":"Usage Keras model:","code":"x <- op_reshape(0:9,   c(1, 5, 2)) y <- op_reshape(10:19, c(1, 2, 5)) layer_dot(x, y, axes=c(2, 3)) ## tf.Tensor( ## [[[260 360] ##   [320 445]]], shape=(1, 2, 2), dtype=int32) x1 <- op_reshape(0:9, c(5, 2)) |> layer_dense(8) x2 <- op_reshape(10:19, c(5, 2)) |> layer_dense(8) shape(x1) ## shape(5, 8) shape(x2) ## shape(5, 8) y <- layer_dot(x1, x2, axes=2) shape(y) ## shape(5, 1)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_dropout.html","id":null,"dir":"Reference","previous_headings":"","what":"Applies dropout to the input. — layer_dropout","title":"Applies dropout to the input. — layer_dropout","text":"Dropout layer randomly sets input units 0 frequency rate step training time, helps prevent overfitting. Inputs set 0 scaled 1 / (1 - rate) sum inputs unchanged. Note Dropout layer applies training set TRUE call(), values dropped inference. using model.fit, training appropriately set TRUE automatically. contexts, can set argument explicitly TRUE calling layer. (contrast setting trainable=FALSE Dropout layer. trainable affect layer's behavior, Dropout variables/weights can frozen training.)","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_dropout.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Applies dropout to the input. — layer_dropout","text":"","code":"layer_dropout(object, rate, noise_shape = NULL, seed = NULL, ...)"},{"path":"https://keras3.posit.co/dev/reference/layer_dropout.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Applies dropout to the input. — layer_dropout","text":"object Object compose layer . tensor, array, sequential model. rate Float 0 1. Fraction input units drop. noise_shape 1D integer tensor representing shape binary dropout mask multiplied input. instance, inputs shape (batch_size, timesteps, features) want dropout mask timesteps, can use noise_shape=(batch_size, 1, features). seed R integer use random seed. ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_dropout.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Applies dropout to the input. — layer_dropout","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_dropout.html","id":"call-arguments","dir":"Reference","previous_headings":"","what":"Call Arguments","title":"Applies dropout to the input. — layer_dropout","text":"inputs: Input tensor (rank). training: Python boolean indicating whether layer behave training mode (adding dropout) inference mode (nothing).","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_einsum_dense.html","id":null,"dir":"Reference","previous_headings":"","what":"A layer that uses einsum as the backing computation. — layer_einsum_dense","title":"A layer that uses einsum as the backing computation. — layer_einsum_dense","text":"layer can perform einsum calculations arbitrary dimensionality.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_einsum_dense.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"A layer that uses einsum as the backing computation. — layer_einsum_dense","text":"","code":"layer_einsum_dense(   object,   equation,   output_shape,   activation = NULL,   bias_axes = NULL,   kernel_initializer = \"glorot_uniform\",   bias_initializer = \"zeros\",   kernel_regularizer = NULL,   bias_regularizer = NULL,   kernel_constraint = NULL,   bias_constraint = NULL,   lora_rank = NULL,   ... )"},{"path":"https://keras3.posit.co/dev/reference/layer_einsum_dense.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"A layer that uses einsum as the backing computation. — layer_einsum_dense","text":"object Object compose layer . tensor, array, sequential model. equation equation describing einsum perform. equation must valid einsum string form ab,bc->ac, ...ab,bc->...ac, ab...,bc->ac... 'ab', 'bc', 'ac' can valid einsum axis expression sequence. output_shape expected shape output tensor (excluding batch dimension dimensions represented ellipses). can specify NA NULL dimension unknown can inferred input shape. activation Activation function use. specify anything, activation applied (, \"linear\" activation: (x) = x). bias_axes string containing output dimension(s) apply bias . character bias_axes string correspond character output portion equation string. kernel_initializer Initializer kernel weights matrix. bias_initializer Initializer bias vector. kernel_regularizer Regularizer function applied kernel weights matrix. bias_regularizer Regularizer function applied bias vector. kernel_constraint Constraint function applied kernel weights matrix. bias_constraint Constraint function applied bias vector. lora_rank Optional integer. set, layer's forward pass implement LoRA (Low-Rank Adaptation) provided rank. LoRA sets layer's kernel non-trainable replaces delta original kernel, obtained via multiplying two lower-rank trainable matrices (factorization happens last dimension). can useful reduce computation cost fine-tuning large dense layers. can also enable LoRA existing EinsumDense layer calling layer$enable_lora(rank). ... Base layer keyword arguments, name dtype.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_einsum_dense.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"A layer that uses einsum as the backing computation. — layer_einsum_dense","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_einsum_dense.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"A layer that uses einsum as the backing computation. — layer_einsum_dense","text":"Biased dense layer einsums example shows instantiate standard Keras dense layer using einsum operations. example equivalent layer_Dense(64, use_bias=TRUE).     Applying dense layer sequence example shows instantiate layer applies dense operation every element sequence. , output_shape two values (since two non-batch dimensions output); first dimension output_shape NA, sequence dimension b unknown shape.     Applying dense layer sequence using ellipses example shows instantiate layer applies dense operation every element sequence, uses ellipsis notation instead specifying batch sequence dimensions. using ellipsis notation specified one axis, output_shape arg single value. instantiated way, layer can handle number sequence dimensions - including case sequence dimension exists.","code":"input <- layer_input(shape = c(32)) output <- input |>   layer_einsum_dense(\"ab,bc->ac\",                      output_shape = 64,                      bias_axes = \"c\") output # shape(NA, 64) ## <KerasTensor shape=(None, 64), dtype=float32, sparse=False, ragged=False, name=keras_tensor_1> input <- layer_input(shape = c(32, 128)) output <- input |>   layer_einsum_dense(\"abc,cd->abd\",                      output_shape = c(NA, 64),                      bias_axes = \"d\") output  # shape(NA, 32, 64) ## <KerasTensor shape=(None, None, 64), dtype=float32, sparse=False, ragged=False, name=keras_tensor_3> input <- layer_input(shape = c(32, 128)) output <- input |>   layer_einsum_dense(\"...x,xy->...y\",                      output_shape = 64,                      bias_axes = \"y\")  output  # shape(NA, 32, 64) ## <KerasTensor shape=(None, 32, 64), dtype=float32, sparse=False, ragged=False, name=keras_tensor_5>"},{"path":"https://keras3.posit.co/dev/reference/layer_einsum_dense.html","id":"methods","dir":"Reference","previous_headings":"","what":"Methods","title":"A layer that uses einsum as the backing computation. — layer_einsum_dense","text":"","code":"enable_lora(   rank,   a_initializer = 'he_uniform',   b_initializer = 'zeros' ) quantize(mode, type_check = TRUE)"},{"path":"https://keras3.posit.co/dev/reference/layer_einsum_dense.html","id":"readonly-properties-","dir":"Reference","previous_headings":"","what":"Readonly properties:","title":"A layer that uses einsum as the backing computation. — layer_einsum_dense","text":"kernel","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_embedding.html","id":null,"dir":"Reference","previous_headings":"","what":"Turns nonnegative integers (indexes) into dense vectors of fixed size. — layer_embedding","title":"Turns nonnegative integers (indexes) into dense vectors of fixed size. — layer_embedding","text":"e.g. rbind(4L, 20L) \\(\\rightarrow\\) rbind(c(0.25, 0.1), c(0.6, -0.2)) layer can used nonnegative integer inputs fixed range.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_embedding.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Turns nonnegative integers (indexes) into dense vectors of fixed size. — layer_embedding","text":"","code":"layer_embedding(   object,   input_dim,   output_dim,   embeddings_initializer = \"uniform\",   embeddings_regularizer = NULL,   embeddings_constraint = NULL,   mask_zero = FALSE,   weights = NULL,   lora_rank = NULL,   ... )"},{"path":"https://keras3.posit.co/dev/reference/layer_embedding.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Turns nonnegative integers (indexes) into dense vectors of fixed size. — layer_embedding","text":"object Object compose layer . tensor, array, sequential model. input_dim Integer. Size vocabulary, .e. maximum integer index + 1. output_dim Integer. Dimension dense embedding. embeddings_initializer Initializer embeddings matrix (see keras3::initializer_*). embeddings_regularizer Regularizer function applied embeddings matrix (see keras3::regularizer_*). embeddings_constraint Constraint function applied embeddings matrix (see keras3::constraint_*). mask_zero Boolean, whether input value 0 special \"padding\" value masked . useful using recurrent layers may take variable length input. TRUE, subsequent layers model need support masking exception raised. mask_zero set TRUE, consequence, index 0 used vocabulary (input_dim equal size vocabulary + 1). weights Optional floating-point matrix size (input_dim, output_dim). initial embeddings values use. lora_rank Optional integer. set, layer's forward pass implement LoRA (Low-Rank Adaptation) provided rank. LoRA sets layer's embeddings matrix non-trainable replaces delta original matrix, obtained via multiplying two lower-rank trainable matrices. can useful reduce computation cost fine-tuning large embedding layers. can also enable LoRA existing Embedding layer instance calling layer$enable_lora(rank). ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_embedding.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Turns nonnegative integers (indexes) into dense vectors of fixed size. — layer_embedding","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_embedding.html","id":"example","dir":"Reference","previous_headings":"","what":"Example","title":"Turns nonnegative integers (indexes) into dense vectors of fixed size. — layer_embedding","text":"","code":"model <- keras_model_sequential() |>   layer_embedding(1000, 64)  # The model will take as input an integer matrix of size (batch,input_length), # and the largest integer (i.e. word index) in the input # should be no larger than 999 (vocabulary size). # Now model$output_shape is (NA, 10, 64), where `NA` is the batch # dimension.  input_array <- random_integer(shape = c(32, 10), minval = 0, maxval = 1000) model |> compile('rmsprop', 'mse') output_array <- model |> predict(input_array, verbose = 0) dim(output_array)    # (32, 10, 64) ## [1] 32 10 64"},{"path":"https://keras3.posit.co/dev/reference/layer_embedding.html","id":"input-shape","dir":"Reference","previous_headings":"","what":"Input Shape","title":"Turns nonnegative integers (indexes) into dense vectors of fixed size. — layer_embedding","text":"2D tensor shape: (batch_size, input_length).","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_embedding.html","id":"output-shape","dir":"Reference","previous_headings":"","what":"Output Shape","title":"Turns nonnegative integers (indexes) into dense vectors of fixed size. — layer_embedding","text":"3D tensor shape: (batch_size, input_length, output_dim).","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_embedding.html","id":"methods","dir":"Reference","previous_headings":"","what":"Methods","title":"Turns nonnegative integers (indexes) into dense vectors of fixed size. — layer_embedding","text":"","code":"enable_lora(   rank,   a_initializer = 'he_uniform',   b_initializer = 'zeros' ) quantize(mode, type_check = TRUE) quantized_build(input_shape, mode) quantized_call(...)"},{"path":"https://keras3.posit.co/dev/reference/layer_embedding.html","id":"readonly-properties-","dir":"Reference","previous_headings":"","what":"Readonly properties:","title":"Turns nonnegative integers (indexes) into dense vectors of fixed size. — layer_embedding","text":"embeddings","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_equalization.html","id":null,"dir":"Reference","previous_headings":"","what":"Preprocessing layer for histogram equalization on image channels. — layer_equalization","title":"Preprocessing layer for histogram equalization on image channels. — layer_equalization","text":"Histogram equalization technique adjust image intensities enhance contrast effectively spreading frequent intensity values. layer applies equalization channel-wise basis, can improve visibility details images. layer works grayscale color images, performing equalization independently color channel. inference time, equalization consistently applied. Note: layer safe use inside tf.data pipeline (independently backend using).","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_equalization.html","id":"input-shape","dir":"Reference","previous_headings":"","what":"Input Shape","title":"Preprocessing layer for histogram equalization on image channels. — layer_equalization","text":"3D (unbatched) 4D (batched) tensor shape: (..., height, width, channels), \"channels_last\" format, (..., channels, height, width), \"channels_first\" format.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_equalization.html","id":"output-shape","dir":"Reference","previous_headings":"","what":"Output Shape","title":"Preprocessing layer for histogram equalization on image channels. — layer_equalization","text":"3D (unbatched) 4D (batched) tensor shape: (..., target_height, target_width, channels), (..., channels, target_height, target_width), \"channels_first\" format.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_equalization.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Preprocessing layer for histogram equalization on image channels. — layer_equalization","text":"","code":"layer_equalization(   object,   value_range = list(0L, 255L),   bins = 256L,   data_format = NULL,   ... )"},{"path":"https://keras3.posit.co/dev/reference/layer_equalization.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Preprocessing layer for histogram equalization on image channels. — layer_equalization","text":"object Object compose layer . tensor, array, sequential model. value_range Optional list/tuple 2 floats specifying lower upper limits input data values. Defaults [0, 255]. input image scaled, use appropriate range (e.g., [0.0, 1.0]). equalization scaled range, output values clipped accordingly. bins Integer specifying number histogram bins use equalization. Defaults 256, suitable 8-bit images. Larger values can provide granular intensity redistribution. data_format String, one \"channels_last\" (default) \"channels_first\". ordering dimensions inputs. \"channels_last\" corresponds inputs shape (batch, height, width, channels) \"channels_first\" corresponds inputs shape (batch, channels, height, width). ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_equalization.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Preprocessing layer for histogram equalization on image channels. — layer_equalization","text":"","code":"# Create an equalization layer for standard 8-bit images equalizer <- layer_equalization()  # An image with uneven intensity distribution image <- np_array(...) # your input image  # Apply histogram equalization equalized_image <- equalizer(image)  # For images with custom value range custom_equalizer <- layer_equalization(     value_range=c(0.0, 1.0),  # for normalized images     bins=128  # fewer bins for more subtle equalization ) custom_equalized <- custom_equalizer(normalized_image)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_feature_space.html","id":null,"dir":"Reference","previous_headings":"","what":"One-stop utility for preprocessing and encoding structured data. — layer_feature_space","title":"One-stop utility for preprocessing and encoding structured data. — layer_feature_space","text":"Available feature types: Note features can referred string name, e.g. \"integer_categorical\". using string name, default argument values used.","code":"# Plain float values. feature_float(name = NULL)  # Float values to be preprocessed via featurewise standardization # (i.e. via a `layer_normalization()` layer). feature_float_normalized(name = NULL)  # Float values to be preprocessed via linear rescaling # (i.e. via a `layer_rescaling` layer). feature_float_rescaled(scale = 1., offset = 0., name = NULL)  # Float values to be discretized. By default, the discrete # representation will then be one-hot encoded. feature_float_discretized(   num_bins,   bin_boundaries = NULL,   output_mode = \"one_hot\",   name = NULL )  # Integer values to be indexed. By default, the discrete # representation will then be one-hot encoded. feature_integer_categorical(   max_tokens = NULL,   num_oov_indices = 1,   output_mode = \"one_hot\",   name = NULL )  # String values to be indexed. By default, the discrete # representation will then be one-hot encoded. feature_string_categorical(   max_tokens = NULL,   num_oov_indices = 1,   output_mode = \"one_hot\",   name = NULL )  # Integer values to be hashed into a fixed number of bins. # By default, the discrete representation will then be one-hot encoded. feature_integer_hashed(num_bins, output_mode = \"one_hot\", name = NULL)  # String values to be hashed into a fixed number of bins. # By default, the discrete representation will then be one-hot encoded. feature_string_hashed(num_bins, output_mode = \"one_hot\", name = NULL)"},{"path":"https://keras3.posit.co/dev/reference/layer_feature_space.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"One-stop utility for preprocessing and encoding structured data. — layer_feature_space","text":"","code":"layer_feature_space(   object,   features,   output_mode = \"concat\",   crosses = NULL,   crossing_dim = 32L,   hashing_dim = 32L,   num_discretization_bins = 32L,   name = NULL,   feature_names = NULL )  feature_cross(feature_names, crossing_dim, output_mode = \"one_hot\")  feature_custom(dtype, preprocessor, output_mode)  feature_float(name = NULL)  feature_float_rescaled(scale = 1, offset = 0, name = NULL)  feature_float_normalized(name = NULL)  feature_float_discretized(   num_bins,   bin_boundaries = NULL,   output_mode = \"one_hot\",   name = NULL )  feature_integer_categorical(   max_tokens = NULL,   num_oov_indices = 1,   output_mode = \"one_hot\",   name = NULL )  feature_string_categorical(   max_tokens = NULL,   num_oov_indices = 1,   output_mode = \"one_hot\",   name = NULL )  feature_string_hashed(num_bins, output_mode = \"one_hot\", name = NULL)  feature_integer_hashed(num_bins, output_mode = \"one_hot\", name = NULL)"},{"path":"https://keras3.posit.co/dev/reference/layer_feature_space.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"One-stop utility for preprocessing and encoding structured data. — layer_feature_space","text":"object see description features see description output_mode string. layer_feature_space(), one \"concat\" \"dict\". concat mode, features get concatenated together single vector. dict mode, FeatureSpace returns named list individually encoded features (names input list names). feature_* functions, one : \"int\" \"one_hot\" \"float\". crosses List features crossed together, e.g. crosses=list(c(\"feature_1\", \"feature_2\")). features \"crossed\" hashing combined value fixed-length vector. crossing_dim Default vector size hashing crossed features. Defaults 32. hashing_dim Default vector size hashing features type \"integer_hashed\" \"string_hashed\". Defaults 32. num_discretization_bins Default number bins used discretizing features type \"float_discretized\". Defaults 32. name String, name object feature_names Named list mapping names features type specification, e.g. list(my_feature = \"integer_categorical\") list(my_feature = feature_integer_categorical()). complete list supported types, see \"Available feature types\" paragraph . dtype string, output dtype feature. E.g., \"float32\". preprocessor callable. scale, offset Passed layer_rescaling() num_bins, bin_boundaries Passed layer_discretization() max_tokens, num_oov_indices Passed layer_integer_lookup() feature_integer_categorical() layer_string_lookup() feature_string_categorical().","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_feature_space.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"One-stop utility for preprocessing and encoding structured data. — layer_feature_space","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_feature_space.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"One-stop utility for preprocessing and encoding structured data. — layer_feature_space","text":"Basic usage named list input data:   Basic usage tf.data:   Basic usage Keras Functional API:           Customizing feature feature cross:   Returning dict (named list) integer-encoded features:   Specifying Keras preprocessing layer:   Retrieving underlying Keras preprocessing layers:   Saving reloading FeatureSpace:","code":"raw_data <- list(   float_values = c(0.0, 0.1, 0.2, 0.3),   string_values = c(\"zero\", \"one\", \"two\", \"three\"),   int_values = as.integer(c(0, 1, 2, 3)) )  dataset <- tfdatasets::tensor_slices_dataset(raw_data)  feature_space <- layer_feature_space(   features = list(     float_values = \"float_normalized\",     string_values = \"string_categorical\",     int_values = \"integer_categorical\"   ),   crosses = list(c(\"string_values\", \"int_values\")),   output_mode = \"concat\" )  # Before you start using the feature_space(), # you must `adapt()` it on some data. feature_space |> adapt(dataset)  # You can call the feature_space() on a named list of # data (batched or unbatched). output_vector <- feature_space(raw_data) library(tfdatasets) # Unlabeled data preprocessed_ds <- unlabeled_dataset |>   dataset_map(feature_space)  # Labeled data preprocessed_ds <- labeled_dataset |>   dataset_map(function(x, y) tuple(feature_space(x), y)) # Retrieve a named list of Keras layer_input() objects (inputs <- feature_space$get_inputs()) ## $float_values ## <KerasTensor shape=(None, 1), dtype=float32, sparse=False, ragged=False, name=float_values> ## ## $string_values ## <KerasTensor shape=(None, 1), dtype=string, sparse=False, ragged=False, name=string_values> ## ## $int_values ## <KerasTensor shape=(None, 1), dtype=int32, sparse=False, ragged=False, name=int_values> # Retrieve the corresponding encoded Keras tensors (encoded_features <- feature_space$get_encoded_features()) ## <KerasTensor shape=(None, 43), dtype=float32, sparse=False, ragged=False, name=keras_tensor_7> # Build a Functional model outputs <- encoded_features |> layer_dense(1, activation = \"sigmoid\") model <- keras_model(inputs, outputs) feature_space <- layer_feature_space(   features = list(     float_values = feature_float_normalized(),     string_values = feature_string_categorical(max_tokens = 10),     int_values = feature_integer_categorical(max_tokens = 10)   ),   crosses = list(     feature_cross(c(\"string_values\", \"int_values\"), crossing_dim = 32)   ),   output_mode = \"concat\" ) feature_space <- layer_feature_space(   features = list(     \"string_values\" = feature_string_categorical(output_mode = \"int\"),     \"int_values\" = feature_integer_categorical(output_mode = \"int\")   ),   crosses = list(     feature_cross(       feature_names = c(\"string_values\", \"int_values\"),       crossing_dim = 32,       output_mode = \"int\"     )   ),   output_mode = \"dict\" ) # Let's say that one of the features is a short text paragraph that # we want to encode as a vector (one vector per paragraph) via TF-IDF. data <- list(text = c(\"1st string\", \"2nd string\", \"3rd string\"))  # There's a Keras layer for this: layer_text_vectorization() custom_layer <- layer_text_vectorization(output_mode = \"tf_idf\")  # We can use feature_custom() to create a custom feature # that will use our preprocessing layer. feature_space <- layer_feature_space(   features = list(     text = feature_custom(preprocessor = custom_layer,                           dtype = \"string\",                           output_mode = \"float\"     )   ),   output_mode = \"concat\" ) feature_space |> adapt(tfdatasets::tensor_slices_dataset(data)) output_vector <- feature_space(data) # The preprocessing layer of each feature is available in `$preprocessors`. preprocessing_layer <- feature_space$preprocessors$feature1  # The crossing layer of each feature cross is available in `$crossers`. # It's an instance of layer_hashed_crossing() crossing_layer <- feature_space$crossers[[\"feature1_X_feature2\"]] feature_space$save(\"featurespace.keras\") reloaded_feature_space <- keras$models$load_model(\"featurespace.keras\")"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_flatten.html","id":null,"dir":"Reference","previous_headings":"","what":"Flattens the input. Does not affect the batch size. — layer_flatten","title":"Flattens the input. Does not affect the batch size. — layer_flatten","text":"Flattens input. affect batch size.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_flatten.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Flattens the input. Does not affect the batch size. — layer_flatten","text":"","code":"layer_flatten(object, data_format = NULL, ...)"},{"path":"https://keras3.posit.co/dev/reference/layer_flatten.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Flattens the input. Does not affect the batch size. — layer_flatten","text":"object Object compose layer . tensor, array, sequential model. data_format string, one \"channels_last\" (default) \"channels_first\". ordering dimensions inputs. \"channels_last\" corresponds inputs shape (batch, ..., channels) \"channels_first\" corresponds inputs shape (batch, channels, ...). unspecified, uses image_data_format value found Keras config file ~/.keras/keras.json (exists). Defaults \"channels_last\". ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_flatten.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Flattens the input. Does not affect the batch size. — layer_flatten","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_flatten.html","id":"note","dir":"Reference","previous_headings":"","what":"Note","title":"Flattens the input. Does not affect the batch size. — layer_flatten","text":"inputs shaped (batch) without feature axis, flattening adds extra channel dimension output shape (batch, 1).","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_flatten.html","id":"example","dir":"Reference","previous_headings":"","what":"Example","title":"Flattens the input. Does not affect the batch size. — layer_flatten","text":"","code":"x <- layer_input(shape=c(10, 64)) y <- x |> layer_flatten() shape(y) ## shape(NA, 640)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_flax_module_wrapper.html","id":null,"dir":"Reference","previous_headings":"","what":"Keras Layer that wraps a Flax module. — layer_flax_module_wrapper","title":"Keras Layer that wraps a Flax module. — layer_flax_module_wrapper","text":"layer enables use Flax components form flax.linen.Module instances within Keras using JAX backend Keras. module method use forward pass can specified via method argument __call__ default. method must take following arguments exact names: self method bound module, case default __call__, module otherwise pass module. inputs: inputs model, JAX array PyTree arrays. training (optional): argument specifying training mode inference mode, TRUE passed training mode. FlaxLayer handles non-trainable state model required RNGs automatically. Note mutable parameter flax.linen.Module.apply() set DenyList([\"params\"]), therefore making assumption variables outside \"params\" collection non-trainable weights. example shows create FlaxLayer Flax Module default __call__ method training argument:   example shows wrap module method conform required signature. allows multiple input arguments training argument different name values. additionally shows use function bound module.","code":"# keras3::use_backend(\"jax\") # py_install(\"flax\", \"r-keras\")  if(config_backend() == \"jax\" &&    reticulate::py_module_available(\"flax\")) {  flax <- import(\"flax\")  MyFlaxModule(flax$linen$Module) %py_class% {   `__call__` <- flax$linen$compact(\\(self, inputs) {     inputs |>       (flax$linen$Conv(features = 32L, kernel_size = tuple(3L, 3L)))() |>       flax$linen$relu() |>       flax$linen$avg_pool(window_shape = tuple(2L, 2L),                           strides = tuple(2L, 2L)) |>       # flatten all except batch_size axis       (\\(x) x$reshape(tuple(x$shape[[1]], -1L)))() |>       (flax$linen$Dense(features = 200L))() |>       flax$linen$relu() |>       (flax$linen$Dense(features = 10L))() |>       flax$linen$softmax()   }) }  # typical usage: input <- keras_input(c(28, 28, 3)) output <- input |>   layer_flax_module_wrapper(MyFlaxModule())  model <- keras_model(input, output)  # to instantiate the layer before composing: flax_module <- MyFlaxModule() keras_layer <- layer_flax_module_wrapper(module = flax_module)  input <- keras_input(c(28, 28, 3)) output <- input |>   keras_layer()  model <- keras_model(input, output)  } flax <- import(\"flax\")  MyFlaxModule(flax$linen$Module) \\%py_class\\% {   forward <-     flax$linen$compact(\\(self, inputs1, input2, deterministic) {       # do work ....       outputs # return     }) }  my_flax_module_wrapper <- function(module, inputs, training) {   c(input1, input2) \\%<-\\% inputs   module$forward(input1, input2,!training) }  flax_module <- MyFlaxModule() keras_layer <- layer_flax_module_wrapper(module = flax_module,                                          method = my_flax_module_wrapper)"},{"path":"https://keras3.posit.co/dev/reference/layer_flax_module_wrapper.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Keras Layer that wraps a Flax module. — layer_flax_module_wrapper","text":"","code":"layer_flax_module_wrapper(object, module, method = NULL, variables = NULL, ...)"},{"path":"https://keras3.posit.co/dev/reference/layer_flax_module_wrapper.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Keras Layer that wraps a Flax module. — layer_flax_module_wrapper","text":"object Object compose layer . tensor, array, sequential model. module instance flax.linen.Module subclass. method method call model. generally method Module. provided, __call__ method used. method can also function defined Module, case must take Module first argument. used Module.init Module.apply. Details documented method argument flax.linen.Module.apply(). variables dict (named R list) containing variables module format returned flax.linen.Module.init(). contain \"params\" key , applicable, keys collections variables non-trainable state. allows passing trained parameters learned non-trainable state controlling initialization. NULL passed, module's init function called build time initialize variables model. ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_flax_module_wrapper.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Keras Layer that wraps a Flax module. — layer_flax_module_wrapper","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_gaussian_dropout.html","id":null,"dir":"Reference","previous_headings":"","what":"Apply multiplicative 1-centered Gaussian noise. — layer_gaussian_dropout","title":"Apply multiplicative 1-centered Gaussian noise. — layer_gaussian_dropout","text":"regularization layer, active training time.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_gaussian_dropout.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Apply multiplicative 1-centered Gaussian noise. — layer_gaussian_dropout","text":"","code":"layer_gaussian_dropout(object, rate, seed = NULL, ...)"},{"path":"https://keras3.posit.co/dev/reference/layer_gaussian_dropout.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Apply multiplicative 1-centered Gaussian noise. — layer_gaussian_dropout","text":"object Object compose layer . tensor, array, sequential model. rate Float, drop probability (Dropout). multiplicative noise standard deviation sqrt(rate / (1 - rate)). seed Integer, optional random seed enable deterministic behavior. ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_gaussian_dropout.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Apply multiplicative 1-centered Gaussian noise. — layer_gaussian_dropout","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_gaussian_dropout.html","id":"call-arguments","dir":"Reference","previous_headings":"","what":"Call Arguments","title":"Apply multiplicative 1-centered Gaussian noise. — layer_gaussian_dropout","text":"inputs: Input tensor (rank). training: Python boolean indicating whether layer behave training mode (adding dropout) inference mode (nothing).","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_gaussian_noise.html","id":null,"dir":"Reference","previous_headings":"","what":"Apply additive zero-centered Gaussian noise. — layer_gaussian_noise","title":"Apply additive zero-centered Gaussian noise. — layer_gaussian_noise","text":"useful mitigate overfitting (see form random data augmentation). Gaussian Noise (GS) natural choice corruption process real valued inputs. regularization layer, active training time.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_gaussian_noise.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Apply additive zero-centered Gaussian noise. — layer_gaussian_noise","text":"","code":"layer_gaussian_noise(object, stddev, seed = NULL, ...)"},{"path":"https://keras3.posit.co/dev/reference/layer_gaussian_noise.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Apply additive zero-centered Gaussian noise. — layer_gaussian_noise","text":"object Object compose layer . tensor, array, sequential model. stddev Float, standard deviation noise distribution. seed Integer, optional random seed enable deterministic behavior. ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_gaussian_noise.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Apply additive zero-centered Gaussian noise. — layer_gaussian_noise","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_gaussian_noise.html","id":"call-arguments","dir":"Reference","previous_headings":"","what":"Call Arguments","title":"Apply additive zero-centered Gaussian noise. — layer_gaussian_noise","text":"inputs: Input tensor (rank). training: Python boolean indicating whether layer behave training mode (adding noise) inference mode (nothing).","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_global_average_pooling_1d.html","id":null,"dir":"Reference","previous_headings":"","what":"Global average pooling operation for temporal data. — layer_global_average_pooling_1d","title":"Global average pooling operation for temporal data. — layer_global_average_pooling_1d","text":"Global average pooling operation temporal data.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_global_average_pooling_1d.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Global average pooling operation for temporal data. — layer_global_average_pooling_1d","text":"","code":"layer_global_average_pooling_1d(   object,   data_format = NULL,   keepdims = FALSE,   ... )"},{"path":"https://keras3.posit.co/dev/reference/layer_global_average_pooling_1d.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Global average pooling operation for temporal data. — layer_global_average_pooling_1d","text":"object Object compose layer . tensor, array, sequential model. data_format string, either \"channels_last\" \"channels_first\". ordering dimensions inputs. \"channels_last\" corresponds inputs shape (batch, steps, features) \"channels_first\" corresponds inputs shape (batch, features, steps). defaults image_data_format value found Keras config file ~/.keras/keras.json. never set , \"channels_last\". keepdims boolean, whether keep temporal dimension . keepdims FALSE (default), rank tensor reduced spatial dimensions. keepdims TRUE, temporal dimension retained length 1. behavior tf$reduce_mean() op_mean(). ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_global_average_pooling_1d.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Global average pooling operation for temporal data. — layer_global_average_pooling_1d","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_global_average_pooling_1d.html","id":"call-arguments","dir":"Reference","previous_headings":"","what":"Call Arguments","title":"Global average pooling operation for temporal data. — layer_global_average_pooling_1d","text":"inputs: 3D tensor. mask: Binary tensor shape (batch_size, steps) indicating whether given step masked (excluded average).","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_global_average_pooling_1d.html","id":"input-shape","dir":"Reference","previous_headings":"","what":"Input Shape","title":"Global average pooling operation for temporal data. — layer_global_average_pooling_1d","text":"data_format='channels_last': 3D tensor shape: (batch_size, steps, features) data_format='channels_first': 3D tensor shape: (batch_size, features, steps)","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_global_average_pooling_1d.html","id":"output-shape","dir":"Reference","previous_headings":"","what":"Output Shape","title":"Global average pooling operation for temporal data. — layer_global_average_pooling_1d","text":"keepdims=FALSE: 2D tensor shape (batch_size, features). keepdims=TRUE: data_format=\"channels_last\": 3D tensor shape (batch_size, 1, features) data_format=\"channels_first\": 3D tensor shape (batch_size, features, 1)","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_global_average_pooling_1d.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Global average pooling operation for temporal data. — layer_global_average_pooling_1d","text":"","code":"x <- random_uniform(c(2, 3, 4)) y <- x |> layer_global_average_pooling_1d() shape(y) ## shape(2, 4)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_global_average_pooling_2d.html","id":null,"dir":"Reference","previous_headings":"","what":"Global average pooling operation for 2D data. — layer_global_average_pooling_2d","title":"Global average pooling operation for 2D data. — layer_global_average_pooling_2d","text":"Global average pooling operation 2D data.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_global_average_pooling_2d.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Global average pooling operation for 2D data. — layer_global_average_pooling_2d","text":"","code":"layer_global_average_pooling_2d(   object,   data_format = NULL,   keepdims = FALSE,   ... )"},{"path":"https://keras3.posit.co/dev/reference/layer_global_average_pooling_2d.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Global average pooling operation for 2D data. — layer_global_average_pooling_2d","text":"object Object compose layer . tensor, array, sequential model. data_format string, either \"channels_last\" \"channels_first\". ordering dimensions inputs. \"channels_last\" corresponds inputs shape (batch, height, width, channels) \"channels_first\" corresponds inputs shape (batch, features, height, weight). defaults image_data_format value found Keras config file ~/.keras/keras.json. never set , \"channels_last\". keepdims boolean, whether keep temporal dimension . keepdims FALSE (default), rank tensor reduced spatial dimensions. keepdims TRUE, spatial dimension retained length 1. behavior tf$reduce_mean() op_mean(). ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_global_average_pooling_2d.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Global average pooling operation for 2D data. — layer_global_average_pooling_2d","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_global_average_pooling_2d.html","id":"input-shape","dir":"Reference","previous_headings":"","what":"Input Shape","title":"Global average pooling operation for 2D data. — layer_global_average_pooling_2d","text":"data_format='channels_last': 4D tensor shape: (batch_size, height, width, channels) data_format='channels_first': 4D tensor shape: (batch_size, channels, height, width)","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_global_average_pooling_2d.html","id":"output-shape","dir":"Reference","previous_headings":"","what":"Output Shape","title":"Global average pooling operation for 2D data. — layer_global_average_pooling_2d","text":"keepdims=FALSE: 2D tensor shape (batch_size, channels). keepdims=TRUE: data_format=\"channels_last\": 4D tensor shape (batch_size, 1, 1, channels) data_format=\"channels_first\": 4D tensor shape (batch_size, channels, 1, 1)","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_global_average_pooling_2d.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Global average pooling operation for 2D data. — layer_global_average_pooling_2d","text":"","code":"x <- random_uniform(c(2, 4, 5, 3)) y <- x |> layer_global_average_pooling_2d() shape(y) ## shape(2, 3)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_global_average_pooling_3d.html","id":null,"dir":"Reference","previous_headings":"","what":"Global average pooling operation for 3D data. — layer_global_average_pooling_3d","title":"Global average pooling operation for 3D data. — layer_global_average_pooling_3d","text":"Global average pooling operation 3D data.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_global_average_pooling_3d.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Global average pooling operation for 3D data. — layer_global_average_pooling_3d","text":"","code":"layer_global_average_pooling_3d(   object,   data_format = NULL,   keepdims = FALSE,   ... )"},{"path":"https://keras3.posit.co/dev/reference/layer_global_average_pooling_3d.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Global average pooling operation for 3D data. — layer_global_average_pooling_3d","text":"object Object compose layer . tensor, array, sequential model. data_format string, either \"channels_last\" \"channels_first\". ordering dimensions inputs. \"channels_last\" corresponds inputs shape (batch, spatial_dim1, spatial_dim2, spatial_dim3, channels) \"channels_first\" corresponds inputs shape (batch, channels, spatial_dim1, spatial_dim2, spatial_dim3). defaults image_data_format value found Keras config file ~/.keras/keras.json. never set , \"channels_last\". keepdims boolean, whether keep temporal dimension . keepdims FALSE (default), rank tensor reduced spatial dimensions. keepdims TRUE, spatial dimension retained length 1. behavior tf$reduce_mean() op_mean(). ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_global_average_pooling_3d.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Global average pooling operation for 3D data. — layer_global_average_pooling_3d","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_global_average_pooling_3d.html","id":"input-shape","dir":"Reference","previous_headings":"","what":"Input Shape","title":"Global average pooling operation for 3D data. — layer_global_average_pooling_3d","text":"data_format='channels_last': 5D tensor shape: (batch_size, spatial_dim1, spatial_dim2, spatial_dim3, channels) data_format='channels_first': 5D tensor shape: (batch_size, channels, spatial_dim1, spatial_dim2, spatial_dim3)","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_global_average_pooling_3d.html","id":"output-shape","dir":"Reference","previous_headings":"","what":"Output Shape","title":"Global average pooling operation for 3D data. — layer_global_average_pooling_3d","text":"keepdims=FALSE: 2D tensor shape (batch_size, channels). keepdims=TRUE: data_format=\"channels_last\": 5D tensor shape (batch_size, 1, 1, 1, channels) data_format=\"channels_first\": 5D tensor shape (batch_size, channels, 1, 1, 1)","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_global_average_pooling_3d.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Global average pooling operation for 3D data. — layer_global_average_pooling_3d","text":"","code":"x <- random_uniform(c(2, 4, 5, 4, 3)) y <- x |> layer_global_average_pooling_3d() shape(y) ## shape(2, 3)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_global_max_pooling_1d.html","id":null,"dir":"Reference","previous_headings":"","what":"Global max pooling operation for temporal data. — layer_global_max_pooling_1d","title":"Global max pooling operation for temporal data. — layer_global_max_pooling_1d","text":"Global max pooling operation temporal data.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_global_max_pooling_1d.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Global max pooling operation for temporal data. — layer_global_max_pooling_1d","text":"","code":"layer_global_max_pooling_1d(object, data_format = NULL, keepdims = FALSE, ...)"},{"path":"https://keras3.posit.co/dev/reference/layer_global_max_pooling_1d.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Global max pooling operation for temporal data. — layer_global_max_pooling_1d","text":"object Object compose layer . tensor, array, sequential model. data_format string, either \"channels_last\" \"channels_first\". ordering dimensions inputs. \"channels_last\" corresponds inputs shape (batch, steps, features) \"channels_first\" corresponds inputs shape (batch, features, steps). defaults image_data_format value found Keras config file ~/.keras/keras.json. never set , \"channels_last\". keepdims boolean, whether keep temporal dimension . keepdims FALSE (default), rank tensor reduced spatial dimensions. keepdims TRUE, temporal dimension retained length 1. behavior tf$reduce_mean() op_mean(). ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_global_max_pooling_1d.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Global max pooling operation for temporal data. — layer_global_max_pooling_1d","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_global_max_pooling_1d.html","id":"input-shape","dir":"Reference","previous_headings":"","what":"Input Shape","title":"Global max pooling operation for temporal data. — layer_global_max_pooling_1d","text":"data_format='channels_last': 3D tensor shape: (batch_size, steps, features) data_format='channels_first': 3D tensor shape: (batch_size, features, steps)","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_global_max_pooling_1d.html","id":"output-shape","dir":"Reference","previous_headings":"","what":"Output Shape","title":"Global max pooling operation for temporal data. — layer_global_max_pooling_1d","text":"keepdims=FALSE: 2D tensor shape (batch_size, features). keepdims=TRUE: data_format=\"channels_last\": 3D tensor shape (batch_size, 1, features) data_format=\"channels_first\": 3D tensor shape (batch_size, features, 1)","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_global_max_pooling_1d.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Global max pooling operation for temporal data. — layer_global_max_pooling_1d","text":"","code":"x <- random_uniform(c(2, 3, 4)) y <- x |> layer_global_max_pooling_1d() shape(y) ## shape(2, 4)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_global_max_pooling_2d.html","id":null,"dir":"Reference","previous_headings":"","what":"Global max pooling operation for 2D data. — layer_global_max_pooling_2d","title":"Global max pooling operation for 2D data. — layer_global_max_pooling_2d","text":"Global max pooling operation 2D data.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_global_max_pooling_2d.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Global max pooling operation for 2D data. — layer_global_max_pooling_2d","text":"","code":"layer_global_max_pooling_2d(object, data_format = NULL, keepdims = FALSE, ...)"},{"path":"https://keras3.posit.co/dev/reference/layer_global_max_pooling_2d.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Global max pooling operation for 2D data. — layer_global_max_pooling_2d","text":"object Object compose layer . tensor, array, sequential model. data_format string, either \"channels_last\" \"channels_first\". ordering dimensions inputs. \"channels_last\" corresponds inputs shape (batch, height, width, channels) \"channels_first\" corresponds inputs shape (batch, features, height, weight). defaults image_data_format value found Keras config file ~/.keras/keras.json. never set , \"channels_last\". keepdims boolean, whether keep temporal dimension . keepdims FALSE (default), rank tensor reduced spatial dimensions. keepdims TRUE, spatial dimension retained length 1. behavior tf$reduce_mean() op_mean(). ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_global_max_pooling_2d.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Global max pooling operation for 2D data. — layer_global_max_pooling_2d","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_global_max_pooling_2d.html","id":"input-shape","dir":"Reference","previous_headings":"","what":"Input Shape","title":"Global max pooling operation for 2D data. — layer_global_max_pooling_2d","text":"data_format='channels_last': 4D tensor shape: (batch_size, height, width, channels) data_format='channels_first': 4D tensor shape: (batch_size, channels, height, width)","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_global_max_pooling_2d.html","id":"output-shape","dir":"Reference","previous_headings":"","what":"Output Shape","title":"Global max pooling operation for 2D data. — layer_global_max_pooling_2d","text":"keepdims=FALSE: 2D tensor shape (batch_size, channels). keepdims=TRUE: data_format=\"channels_last\": 4D tensor shape (batch_size, 1, 1, channels) data_format=\"channels_first\": 4D tensor shape (batch_size, channels, 1, 1)","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_global_max_pooling_2d.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Global max pooling operation for 2D data. — layer_global_max_pooling_2d","text":"","code":"x <- random_uniform(c(2, 4, 5, 3)) y <- x |> layer_global_max_pooling_2d() shape(y) ## shape(2, 3)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_global_max_pooling_3d.html","id":null,"dir":"Reference","previous_headings":"","what":"Global max pooling operation for 3D data. — layer_global_max_pooling_3d","title":"Global max pooling operation for 3D data. — layer_global_max_pooling_3d","text":"Global max pooling operation 3D data.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_global_max_pooling_3d.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Global max pooling operation for 3D data. — layer_global_max_pooling_3d","text":"","code":"layer_global_max_pooling_3d(object, data_format = NULL, keepdims = FALSE, ...)"},{"path":"https://keras3.posit.co/dev/reference/layer_global_max_pooling_3d.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Global max pooling operation for 3D data. — layer_global_max_pooling_3d","text":"object Object compose layer . tensor, array, sequential model. data_format string, either \"channels_last\" \"channels_first\". ordering dimensions inputs. \"channels_last\" corresponds inputs shape (batch, spatial_dim1, spatial_dim2, spatial_dim3, channels) \"channels_first\" corresponds inputs shape (batch, channels, spatial_dim1, spatial_dim2, spatial_dim3). defaults image_data_format value found Keras config file ~/.keras/keras.json. never set , \"channels_last\". keepdims boolean, whether keep temporal dimension . keepdims FALSE (default), rank tensor reduced spatial dimensions. keepdims TRUE, spatial dimension retained length 1. behavior tf$reduce_mean() op_mean(). ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_global_max_pooling_3d.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Global max pooling operation for 3D data. — layer_global_max_pooling_3d","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_global_max_pooling_3d.html","id":"input-shape","dir":"Reference","previous_headings":"","what":"Input Shape","title":"Global max pooling operation for 3D data. — layer_global_max_pooling_3d","text":"data_format='channels_last': 5D tensor shape: (batch_size, spatial_dim1, spatial_dim2, spatial_dim3, channels) data_format='channels_first': 5D tensor shape: (batch_size, channels, spatial_dim1, spatial_dim2, spatial_dim3)","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_global_max_pooling_3d.html","id":"output-shape","dir":"Reference","previous_headings":"","what":"Output Shape","title":"Global max pooling operation for 3D data. — layer_global_max_pooling_3d","text":"keepdims=FALSE: 2D tensor shape (batch_size, channels). keepdims=TRUE: data_format=\"channels_last\": 5D tensor shape (batch_size, 1, 1, 1, channels) data_format=\"channels_first\": 5D tensor shape (batch_size, channels, 1, 1, 1)","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_global_max_pooling_3d.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Global max pooling operation for 3D data. — layer_global_max_pooling_3d","text":"","code":"x <- random_uniform(c(2, 4, 5, 4, 3)) y <- x |> layer_global_max_pooling_3d() shape(y) ## shape(2, 3)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_group_normalization.html","id":null,"dir":"Reference","previous_headings":"","what":"Group normalization layer. — layer_group_normalization","title":"Group normalization layer. — layer_group_normalization","text":"Group Normalization divides channels groups computes within group mean variance normalization. Empirically, accuracy stable batch norm wide range small batch sizes, learning rate adjusted linearly batch sizes. Relation Layer Normalization: number groups set 1, operation becomes nearly identical Layer Normalization (see Layer Normalization docs details). Relation Instance Normalization: number groups set input dimension (number groups equal number channels), operation becomes identical Instance Normalization. can achieve via groups=-1.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_group_normalization.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Group normalization layer. — layer_group_normalization","text":"","code":"layer_group_normalization(   object,   groups = 32L,   axis = -1L,   epsilon = 0.001,   center = TRUE,   scale = TRUE,   beta_initializer = \"zeros\",   gamma_initializer = \"ones\",   beta_regularizer = NULL,   gamma_regularizer = NULL,   beta_constraint = NULL,   gamma_constraint = NULL,   ... )"},{"path":"https://keras3.posit.co/dev/reference/layer_group_normalization.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Group normalization layer. — layer_group_normalization","text":"object Object compose layer . tensor, array, sequential model. groups Integer, number groups Group Normalization. Can range [1, N] N input dimension. input dimension must divisible number groups. Defaults 32. axis Integer List/Tuple. axis axes normalize across. Typically, features axis/axes. left-axes typically batch axis/axes. -1 last dimension input. Defaults -1. epsilon Small float added variance avoid dividing zero. Defaults 1e-3. center TRUE, add offset beta normalized tensor. FALSE, beta ignored. Defaults TRUE. scale TRUE, multiply gamma. FALSE, gamma used. next layer linear (also e.g. relu), can disabled since scaling done next layer. Defaults TRUE. beta_initializer Initializer beta weight. Defaults zeros. gamma_initializer Initializer gamma weight. Defaults ones. beta_regularizer Optional regularizer beta weight. NULL default. gamma_regularizer Optional regularizer gamma weight. NULL default. beta_constraint Optional constraint beta weight. NULL default. gamma_constraint Optional constraint gamma weight. NULL default. ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_group_normalization.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Group normalization layer. — layer_group_normalization","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_group_normalization.html","id":"input-shape","dir":"Reference","previous_headings":"","what":"Input Shape","title":"Group normalization layer. — layer_group_normalization","text":"Arbitrary. Use keyword argument input_shape (tuple integers, include samples axis) using layer first layer model.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_group_normalization.html","id":"output-shape","dir":"Reference","previous_headings":"","what":"Output Shape","title":"Group normalization layer. — layer_group_normalization","text":"shape input. **kwargs: Base layer keyword arguments (e.g. name dtype).","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_group_normalization.html","id":"reference","dir":"Reference","previous_headings":"","what":"Reference","title":"Group normalization layer. — layer_group_normalization","text":"Yuxin Wu & Kaiming , 2018","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_group_query_attention.html","id":null,"dir":"Reference","previous_headings":"","what":"Grouped Query Attention layer. — layer_group_query_attention","title":"Grouped Query Attention layer. — layer_group_query_attention","text":"implementation grouped-query attention introduced Ainslie et al., 2023. num_key_value_heads denotes number groups, setting num_key_value_heads 1 equivalent multi-query attention, num_key_value_heads equal num_query_heads equivalent multi-head attention. layer first projects query, key, value tensors. , key value repeated match number heads query. , query scaled dot-producted key tensors. softmaxed obtain attention probabilities. value tensors interpolated probabilities concatenated back single tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_group_query_attention.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Grouped Query Attention layer. — layer_group_query_attention","text":"","code":"layer_group_query_attention(   object,   head_dim,   num_query_heads,   num_key_value_heads,   dropout = 0,   use_bias = TRUE,   flash_attention = NULL,   kernel_initializer = \"glorot_uniform\",   bias_initializer = \"zeros\",   kernel_regularizer = NULL,   bias_regularizer = NULL,   activity_regularizer = NULL,   kernel_constraint = NULL,   bias_constraint = NULL,   seed = NULL,   ... )"},{"path":"https://keras3.posit.co/dev/reference/layer_group_query_attention.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Grouped Query Attention layer. — layer_group_query_attention","text":"object Object compose layer . tensor, array, sequential model. head_dim Size attention head. num_query_heads Number query attention heads. num_key_value_heads Number key value attention heads. dropout Dropout probability. use_bias Boolean, whether dense layers use bias vectors/matrices. flash_attention NULL, layer attempts use flash attention faster memory-efficient attention computations possible. behavior can configured using config_enable_flash_attention() config_disable_flash_attention(). kernel_initializer Initializer dense layer kernels. bias_initializer Initializer dense layer biases. kernel_regularizer Regularizer dense layer kernels. bias_regularizer Regularizer dense layer biases. activity_regularizer Regularizer dense layer activity. kernel_constraint Constraint dense layer kernels. bias_constraint Constraint dense layer kernels. seed Optional integer seed dropout layer. ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_group_query_attention.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Grouped Query Attention layer. — layer_group_query_attention","text":"attention_output: Result computation, shape (batch_dim, target_seq_len, feature_dim), target_seq_len target sequence length feature_dim query input last dim. attention_scores: (Optional) attention coefficients shape (batch_dim, num_query_heads, target_seq_len, source_seq_len).","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_group_query_attention.html","id":"call-arguments","dir":"Reference","previous_headings":"","what":"Call Arguments","title":"Grouped Query Attention layer. — layer_group_query_attention","text":"query: Query tensor shape (batch_dim, target_seq_len, feature_dim), batch_dim batch size, target_seq_len length target sequence, feature_dim dimension feature. value: Value tensor shape (batch_dim, source_seq_len, feature_dim), batch_dim batch size, source_seq_len length source sequence, feature_dim dimension feature. key: Optional key tensor shape (batch_dim, source_seq_len, feature_dim). given, use value key value, common case. attention_mask: boolean mask shape (batch_dim, target_seq_len, source_seq_len), prevents attention certain positions. boolean mask specifies query elements can attend key elements, 1 indicates attention 0 indicates attention. Broadcasting can happen missing batch dimensions head dimension. return_attention_scores: boolean indicate whether output (attention_output, attention_scores) TRUE, attention_output FALSE. Defaults FALSE. training: Python boolean indicating whether layer behave training mode (adding dropout) inference mode (dropout). go either using training mode parent layer/model FALSE (inference) parent layer. use_causal_mask: boolean indicate whether apply causal mask prevent tokens attending future tokens (e.g., used decoder Transformer).","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_gru.html","id":null,"dir":"Reference","previous_headings":"","what":"Gated Recurrent Unit - Cho et al. 2014. — layer_gru","title":"Gated Recurrent Unit - Cho et al. 2014. — layer_gru","text":"Based available runtime hardware constraints, layer choose different implementations (cuDNN-based backend-native) maximize performance. GPU available arguments layer meet requirement cuDNN kernel (see details), layer use fast cuDNN implementation using TensorFlow backend. requirements use cuDNN implementation : activation == tanh recurrent_activation == sigmoid recurrent_dropout == 0 unroll FALSE use_bias TRUE reset_after TRUE Inputs, use masking, strictly right-padded. Eager execution enabled outermost context. two variants GRU implementation. default one based v3 reset gate applied hidden state matrix multiplication. one based original order reversed. second variant compatible CuDNNGRU (GPU-) allows inference CPU. Thus separate biases kernel recurrent_kernel. use variant, set reset_after=TRUE recurrent_activation='sigmoid'. example:","code":"inputs <- random_uniform(c(32, 10, 8)) outputs <- inputs |> layer_gru(4) shape(outputs) ## shape(32, 4) # (32, 4) gru <- layer_gru(, 4, return_sequences = TRUE, return_state = TRUE) c(whole_sequence_output, final_state) %<-% gru(inputs) shape(whole_sequence_output) ## shape(32, 10, 4) shape(final_state) ## shape(32, 4)"},{"path":"https://keras3.posit.co/dev/reference/layer_gru.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Gated Recurrent Unit - Cho et al. 2014. — layer_gru","text":"","code":"layer_gru(   object,   units,   activation = \"tanh\",   recurrent_activation = \"sigmoid\",   use_bias = TRUE,   kernel_initializer = \"glorot_uniform\",   recurrent_initializer = \"orthogonal\",   bias_initializer = \"zeros\",   kernel_regularizer = NULL,   recurrent_regularizer = NULL,   bias_regularizer = NULL,   activity_regularizer = NULL,   kernel_constraint = NULL,   recurrent_constraint = NULL,   bias_constraint = NULL,   dropout = 0,   recurrent_dropout = 0,   seed = NULL,   return_sequences = FALSE,   return_state = FALSE,   go_backwards = FALSE,   stateful = FALSE,   unroll = FALSE,   reset_after = TRUE,   use_cudnn = \"auto\",   ... )"},{"path":"https://keras3.posit.co/dev/reference/layer_gru.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Gated Recurrent Unit - Cho et al. 2014. — layer_gru","text":"object Object compose layer . tensor, array, sequential model. units Positive integer, dimensionality output space. activation Activation function use. Default: hyperbolic tangent (tanh). pass NULL, activation applied (ie. \"linear\" activation: (x) = x). recurrent_activation Activation function use recurrent step. Default: sigmoid (sigmoid). pass NULL, activation applied (ie. \"linear\" activation: (x) = x). use_bias Boolean, (default TRUE), whether layer use bias vector. kernel_initializer Initializer kernel weights matrix, used linear transformation inputs. Default: \"glorot_uniform\". recurrent_initializer Initializer recurrent_kernel weights matrix, used linear transformation recurrent state. Default: \"orthogonal\". bias_initializer Initializer bias vector. Default: \"zeros\". kernel_regularizer Regularizer function applied kernel weights matrix. Default: NULL. recurrent_regularizer Regularizer function applied recurrent_kernel weights matrix. Default: NULL. bias_regularizer Regularizer function applied bias vector. Default: NULL. activity_regularizer Regularizer function applied output layer (\"activation\"). Default: NULL. kernel_constraint Constraint function applied kernel weights matrix. Default: NULL. recurrent_constraint Constraint function applied recurrent_kernel weights matrix. Default: NULL. bias_constraint Constraint function applied bias vector. Default: NULL. dropout Float 0 1. Fraction units drop linear transformation inputs. Default: 0. recurrent_dropout Float 0 1. Fraction units drop linear transformation recurrent state. Default: 0. seed Random seed dropout. return_sequences Boolean. Whether return last output output sequence, full sequence. Default: FALSE. return_state Boolean. Whether return last state addition output. Default: FALSE. go_backwards Boolean (default FALSE). TRUE, process input sequence backwards return reversed sequence. stateful Boolean (default: FALSE). TRUE, last state sample index batch used initial state sample index following batch. unroll Boolean (default: FALSE). TRUE, network unrolled, else symbolic loop used. Unrolling can speed-RNN, although tends memory-intensive. Unrolling suitable short sequences. reset_after GRU convention (whether apply reset gate matrix multiplication). FALSE \"\", TRUE \"\" (default cuDNN compatible). use_cudnn Whether use cuDNN-backed implementation. \"auto\" attempt use cuDNN feasible, fallback default implementation . ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_gru.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Gated Recurrent Unit - Cho et al. 2014. — layer_gru","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_gru.html","id":"call-arguments","dir":"Reference","previous_headings":"","what":"Call Arguments","title":"Gated Recurrent Unit - Cho et al. 2014. — layer_gru","text":"inputs: 3D tensor, shape (batch, timesteps, feature). mask: Binary tensor shape (samples, timesteps) indicating whether given timestep masked  (optional). individual TRUE entry indicates corresponding timestep utilized, FALSE entry indicates corresponding timestep ignored. Defaults NULL. training: Python boolean indicating whether layer behave training mode inference mode. argument passed cell calling . relevant dropout recurrent_dropout used  (optional). Defaults NULL. initial_state: List initial state tensors passed first call cell (optional, NULL causes creation zero-filled initial state tensors). Defaults NULL.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_hashed_crossing.html","id":null,"dir":"Reference","previous_headings":"","what":"A preprocessing layer which crosses features using the ","title":"A preprocessing layer which crosses features using the ","text":"layer performs crosses categorical features using \"hashing trick\". Conceptually, transformation can thought : hash(concatenate(features)) %% num_bins. layer currently performs crosses scalar inputs batches scalar inputs. Valid input shapes (batch_size, 1), (batch_size) (). Note: layer wraps tf.keras.layers.HashedCrossing. used part compiled computation graph model backend TensorFlow. can however used backend running eagerly. can also always used part input preprocessing pipeline backend (outside model ), recommend use layer. Note: layer safe use inside tfdatasets pipeline (independently backend using).","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_hashed_crossing.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"A preprocessing layer which crosses features using the ","text":"","code":"layer_hashed_crossing(   object,   num_bins,   output_mode = \"int\",   sparse = FALSE,   name = NULL,   dtype = NULL,   ... )"},{"path":"https://keras3.posit.co/dev/reference/layer_hashed_crossing.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"A preprocessing layer which crosses features using the ","text":"object Object compose layer . tensor, array, sequential model. num_bins Number hash bins. output_mode Specification output layer. Values can \"int\", \"one_hot\" configuring layer follows: \"int\": Return integer bin indices directly. \"one_hot\": Encodes individual element input array size num_bins, containing 1 input's bin index. Defaults \"int\". sparse Boolean. applicable \"one_hot\" mode valid using TensorFlow backend. TRUE, returns SparseTensor instead dense Tensor. Defaults FALSE. name String, name object dtype datatype (e.g., \"float32\"). ... Keyword arguments construct layer.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_hashed_crossing.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"A preprocessing layer which crosses features using the ","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_hashed_crossing.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"A preprocessing layer which crosses features using the ","text":"Crossing two scalar features.     Crossing one-hotting two scalar features.","code":"feat1 <- c('A', 'B', 'A', 'B', 'A') |> as.array() feat2 <- c(101, 101, 101, 102, 102) |> as.integer() |> as.array() layer <- layer_hashed_crossing(num_bins = 5) layer(list(feat1, feat2)) ## tf.Tensor([1 4 1 1 3], shape=(5), dtype=int64) layer <- layer_hashed_crossing(num_bins = 5, output_mode = 'one_hot') layer(list(feat1, feat2)) ## tf.Tensor( ## [[0. 1. 0. 0. 0.] ##  [0. 0. 0. 0. 1.] ##  [0. 1. 0. 0. 0.] ##  [0. 1. 0. 0. 0.] ##  [0. 0. 0. 1. 0.]], shape=(5, 5), dtype=float32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_hashing.html","id":null,"dir":"Reference","previous_headings":"","what":"A preprocessing layer which hashes and bins categorical features. — layer_hashing","title":"A preprocessing layer which hashes and bins categorical features. — layer_hashing","text":"layer transforms categorical inputs hashed output. element-wise converts ints strings ints fixed range. stable hash function uses tensorflow::ops::Fingerprint produce output consistently across platforms. layer uses FarmHash64 default, provides consistent hashed output across different platforms stable across invocations, regardless device context, mixing input bits thoroughly. want obfuscate hashed output, can also pass random salt argument constructor. case, layer use SipHash64 hash function, salt value serving additional input hash function. Note: layer internally uses TensorFlow. used part compiled computation graph model backend TensorFlow. can however used backend running eagerly. can also always used part input preprocessing pipeline backend (outside model ), recommend use layer. Note: layer safe use inside tf.data pipeline (independently backend using). Example (FarmHash64)     Example (FarmHash64) mask value     Example (SipHash64)     Example (Siphash64 single integer, salt=[133, 133])","code":"layer <- layer_hashing(num_bins = 3) inp <- c('A', 'B', 'C', 'D', 'E') |> array(dim = c(5, 1)) layer(inp) ## tf.Tensor( ## [[1] ##  [0] ##  [1] ##  [1] ##  [2]], shape=(5, 1), dtype=int64) layer <- layer_hashing(num_bins=3, mask_value='') inp <- c('A', 'B', '', 'C', 'D') |> array(dim = c(5, 1)) layer(inp) ## tf.Tensor( ## [[1] ##  [1] ##  [0] ##  [2] ##  [2]], shape=(5, 1), dtype=int64) layer <- layer_hashing(num_bins=3, salt=c(133, 137)) inp <- c('A', 'B', 'C', 'D', 'E') |> array(dim = c(5, 1)) layer(inp) ## tf.Tensor( ## [[1] ##  [2] ##  [1] ##  [0] ##  [2]], shape=(5, 1), dtype=int64) layer <- layer_hashing(num_bins=3, salt=133) inp <- c('A', 'B', 'C', 'D', 'E') |> array(dim = c(5, 1)) layer(inp) ## tf.Tensor( ## [[0] ##  [0] ##  [2] ##  [1] ##  [0]], shape=(5, 1), dtype=int64)"},{"path":"https://keras3.posit.co/dev/reference/layer_hashing.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"A preprocessing layer which hashes and bins categorical features. — layer_hashing","text":"","code":"layer_hashing(   object,   num_bins,   mask_value = NULL,   salt = NULL,   output_mode = \"int\",   sparse = FALSE,   ... )"},{"path":"https://keras3.posit.co/dev/reference/layer_hashing.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"A preprocessing layer which hashes and bins categorical features. — layer_hashing","text":"object Object compose layer . tensor, array, sequential model. num_bins Number hash bins. Note includes mask_value bin, effective number bins (num_bins - 1) mask_value set. mask_value value represents masked inputs, mapped index 0. NULL means mask term added hashing start index 0. Defaults NULL. salt single unsigned integer NULL. passed, hash function used SipHash64, values used additional input (known \"salt\" cryptography). non-zero. NULL, uses FarmHash64 hash function. also supports list 2 unsigned integer numbers, see reference paper details. Defaults NULL. output_mode Specification output layer. Values can \"int\", \"one_hot\", \"multi_hot\", \"count\" configuring layer follows: \"int\": Return integer bin indices directly. \"one_hot\": Encodes individual element input array size num_bins, containing 1 input's bin index. last dimension size 1, encode dimension. last dimension size 1, append new dimension encoded output. \"multi_hot\": Encodes sample input single array size num_bins, containing 1 bin index index present sample. Treats last dimension sample dimension, input shape (..., sample_length), output shape (..., num_tokens). \"count\": \"multi_hot\", int array contains count number times bin index appeared sample. Defaults \"int\". sparse Boolean. applicable \"one_hot\", \"multi_hot\", \"count\" output modes. supported TensorFlow backend. TRUE, returns SparseTensor instead dense Tensor. Defaults FALSE. ... Keyword arguments construct layer.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_hashing.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"A preprocessing layer which hashes and bins categorical features. — layer_hashing","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_hashing.html","id":"input-shape","dir":"Reference","previous_headings":"","what":"Input Shape","title":"A preprocessing layer which hashes and bins categorical features. — layer_hashing","text":"single string, list strings, int32 int64 tensor shape (batch_size, ...,).","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_hashing.html","id":"output-shape","dir":"Reference","previous_headings":"","what":"Output Shape","title":"A preprocessing layer which hashes and bins categorical features. — layer_hashing","text":"int32 tensor shape (batch_size, ...).","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_hashing.html","id":"reference","dir":"Reference","previous_headings":"","what":"Reference","title":"A preprocessing layer which hashes and bins categorical features. — layer_hashing","text":"SipHash salt","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_identity.html","id":null,"dir":"Reference","previous_headings":"","what":"Identity layer. — layer_identity","title":"Identity layer. — layer_identity","text":"layer used placeholder operation performed. layer just returns inputs argument output.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_identity.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Identity layer. — layer_identity","text":"","code":"layer_identity(object, ...)"},{"path":"https://keras3.posit.co/dev/reference/layer_identity.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Identity layer. — layer_identity","text":"object Object compose layer . tensor, array, sequential model. ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_identity.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Identity layer. — layer_identity","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_input.html","id":null,"dir":"Reference","previous_headings":"","what":"keras_input — layer_input","title":"keras_input — layer_input","text":"layer_input() alias keras_input(). See ?keras_input() full documentation.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_input.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"keras_input — layer_input","text":"","code":"layer_input(   shape = NULL,   batch_size = NULL,   dtype = NULL,   sparse = NULL,   ragged = NULL,   batch_shape = NULL,   name = NULL,   tensor = NULL,   optional = FALSE )"},{"path":"https://keras3.posit.co/dev/reference/layer_input.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"keras_input — layer_input","text":"shape shape list (list integers NULL objects), including batch size. instance, shape = c(32) indicates expected input batches 32-dimensional vectors. Elements list can NULL NA; NULL/NA elements represent dimensions shape known may vary (e.g. sequence length). batch_size Optional static batch size (integer). dtype data type expected input, string (e.g. \"float32\", \"int32\"...) sparse boolean specifying whether expected input sparse tensors. Note , sparse FALSE, sparse tensors can still passed input - densified default value 0. feature supported TensorFlow backend. Defaults FALSE. ragged boolean specifying whether expected input ragged tensors. Note , ragged FALSE, ragged tensors can still passed input - densified default value 0. feature supported TensorFlow backend. Defaults FALSE. batch_shape Optional shape list (list integers NULL objects), including batch size. name Optional name string layer. unique model (reuse name twice). autogenerated provided. tensor Optional existing tensor wrap Input layer. set, layer use tensor rather creating new placeholder tensor. optional Boolean, whether input optional . optional input can accept NULL values.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_input.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"keras_input — layer_input","text":"Keras tensor, can passed inputs argument (keras_model()).","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_integer_lookup.html","id":null,"dir":"Reference","previous_headings":"","what":"A preprocessing layer that maps integers to (possibly encoded) indices. — layer_integer_lookup","title":"A preprocessing layer that maps integers to (possibly encoded) indices. — layer_integer_lookup","text":"layer maps set arbitrary integer input tokens indexed integer output via table-based vocabulary lookup. layer's output indices contiguously arranged maximum vocab size, even input tokens non-continguous unbounded. layer supports multiple options encoding output via output_mode, optional support --vocabulary (OOV) tokens masking. vocabulary layer must either supplied construction learned via adapt(). adapt(), layer analyze data set, determine frequency individual integer tokens, create vocabulary . vocabulary capped size, frequent tokens used create vocabulary others treated OOV. two possible output modes layer.  output_mode \"int\", input integers converted index vocabulary (integer).  output_mode \"multi_hot\", \"count\", \"tf_idf\", input integers encoded array dimension corresponds element vocabulary. vocabulary can optionally contain mask token well OOV token (can optionally occupy multiple indices vocabulary, set num_oov_indices). position tokens vocabulary fixed. output_mode \"int\", vocabulary begin mask token index 0, followed OOV indices, followed rest vocabulary. output_mode \"multi_hot\", \"count\", \"tf_idf\" vocabulary begin OOV indices instances mask token dropped. Note: layer uses TensorFlow internally. used part compiled computation graph model backend TensorFlow. can however used backend running eagerly. can also always used part input preprocessing pipeline backend (outside model ), recommend use layer. Note: layer safe use inside tf.data pipeline (independently backend using). Note: working layer outputs directly (e.g., passing outputs another layer, using lower-level operations like [ op_*): returned indices 0-based. However, default settings, first (0) index OOV token, returned indices offset 1 may appear 1-based.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_integer_lookup.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"A preprocessing layer that maps integers to (possibly encoded) indices. — layer_integer_lookup","text":"","code":"layer_integer_lookup(   object,   max_tokens = NULL,   num_oov_indices = 1L,   mask_token = NULL,   oov_token = -1L,   vocabulary = NULL,   vocabulary_dtype = \"int64\",   idf_weights = NULL,   invert = FALSE,   output_mode = \"int\",   sparse = FALSE,   pad_to_max_tokens = FALSE,   name = NULL,   ... )"},{"path":"https://keras3.posit.co/dev/reference/layer_integer_lookup.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"A preprocessing layer that maps integers to (possibly encoded) indices. — layer_integer_lookup","text":"object Object compose layer . tensor, array, sequential model. max_tokens Maximum size vocabulary layer. specified adapting vocabulary setting pad_to_max_tokens=TRUE. NULL, cap size vocabulary. Note size includes OOV mask tokens. Defaults NULL. num_oov_indices number --vocabulary tokens use. value 1, OOV inputs modulated determine OOV value. value 0, OOV inputs cause error calling layer. Defaults 1. mask_token integer token represents masked inputs. output_mode \"int\", token included vocabulary mapped index 0. output modes, token appear vocabulary instances mask token input dropped. set NULL, mask term added. Defaults NULL. oov_token used invert TRUE. token return OOV indices. Defaults -1. vocabulary Optional. Either array integers string path text file. passing array, can pass list, list, 1D NumPy array, 1D tensor containing integer vocabulary terms. passing file path, file contain one line per term vocabulary. argument set, need adapt() layer. vocabulary_dtype dtype vocabulary terms, example \"int64\" \"int32\". Defaults \"int64\". idf_weights valid output_mode \"tf_idf\". list, list, 1D NumPy array, 1D tensor length vocabulary, containing floating point inverse document frequency weights, multiplied per sample term counts final TF-IDF weight. vocabulary argument set, output_mode \"tf_idf\", argument must supplied. invert valid output_mode \"int\". TRUE, layer map indices vocabulary items instead mapping vocabulary items indices. Defaults FALSE. output_mode Specification output layer. Values can \"int\", \"one_hot\", \"multi_hot\", \"count\", \"tf_idf\" configuring layer follows: \"int\": Return vocabulary indices input tokens. \"one_hot\": Encodes individual element input array size vocabulary, containing 1 element index. last dimension size 1, encode dimension. last dimension size 1, append new dimension encoded output. \"multi_hot\": Encodes sample input single array size vocabulary, containing 1 vocabulary term present sample. Treats last dimension sample dimension, input shape (..., sample_length), output shape (..., num_tokens). \"count\": \"multi_hot\", int array contains count number times token index appeared sample. \"tf_idf\": \"multi_hot\", TF-IDF algorithm applied find value token slot. \"int\" output, shape input output supported. output modes, currently output rank 2 supported. Defaults \"int\". sparse Boolean. applicable \"multi_hot\", \"count\", \"tf_idf\" output modes. supported TensorFlow backend. TRUE, returns SparseTensor instead dense Tensor. Defaults FALSE. pad_to_max_tokens applicable output_mode \"multi_hot\", \"count\", \"tf_idf\". TRUE, output feature axis padded max_tokens even number unique tokens vocabulary less max_tokens, resulting tensor shape (batch_size, max_tokens) regardless vocabulary size. Defaults FALSE. name String, name object ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_integer_lookup.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"A preprocessing layer that maps integers to (possibly encoded) indices. — layer_integer_lookup","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_integer_lookup.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"A preprocessing layer that maps integers to (possibly encoded) indices. — layer_integer_lookup","text":"Creating lookup layer known vocabulary example creates lookup layer pre-existing vocabulary.     Creating lookup layer adapted vocabulary example creates lookup layer generates vocabulary analyzing dataset.     Note OOV token -1 added vocabulary. remaining tokens sorted frequency (42, 2 occurrences, first) inverse sort order.     Lookups multiple OOV indices example demonstrates use lookup layer multiple OOV indices.  layer created one OOV index, OOV tokens hashed number OOV buckets, distributing OOV tokens deterministic fashion across set.     Note output OOV token 37 1, output OOV token 1000 0. -vocab terms output index increased 1 earlier examples (12 maps 2, etc) order make space extra OOV token. One-hot output Configure layer output_mode='one_hot'. Note first num_oov_indices dimensions ont_hot encoding represent OOV values.     Multi-hot output Configure layer output_mode = 'multi_hot'. Note first num_oov_indices dimensions multi_hot encoding represent OOV tokens     Token count output Configure layer output_mode='count'. multi_hot output, first num_oov_indices dimensions output represent OOV tokens.     TF-IDF output Configure layer output_mode='tf_idf'. multi_hot output, first num_oov_indices dimensions output represent OOV tokens. token bin output token_count * idf_weight, idf weights inverse document frequency weights per token. provided along vocabulary. Note idf_weight OOV tokens default average idf weights passed .     specify idf weights oov tokens, need pass entire vocabulary including leading oov token.     adapting layer \"tf_idf\" mode, input sample considered document, IDF weight per token calculated : log(1 + num_documents / (1 + token_document_count)). Inverse lookup example demonstrates map indices tokens using layer. (can also use adapt() inverse = TRUE, simplicity pass vocab example.)     Note first index correspond oov token default. Forward inverse lookup pairs example demonstrates use vocabulary standard lookup layer create inverse lookup layer.     example, input token 1000 resulted output -1, since 1000 vocabulary - got represented OOV, OOV tokens returned -1 inverse layer. Also, note inverse work, must already set forward layer vocabulary either directly via adapt() calling get_vocabulary().","code":"vocab <- c(12, 36, 1138, 42) |> as.integer() data <- op_array(rbind(c(12, 1138, 42),                        c(42, 1000, 36)))  # Note OOV tokens out <- data |> layer_integer_lookup(vocabulary = vocab) out ## tf.Tensor( ## [[1 3 4] ##  [4 0 2]], shape=(2, 3), dtype=int64) data <- op_array(rbind(c(12, 1138, 42),                        c(42, 1000, 36)))  # Note OOV tokens layer <- layer_integer_lookup() layer |> adapt(data) layer |> get_vocabulary() |> str() ## List of 6 ##  $ : int -1 ##  $ : num 42 ##  $ : num 1138 ##  $ : num 1000 ##  $ : num 36 ##  $ : num 12 layer(data) ## tf.Tensor( ## [[5 2 1] ##  [1 3 4]], shape=(2, 3), dtype=int64) vocab <- c(12, 36, 1138, 42) |> as.integer() data <- op_array(rbind(c(12, 1138, 42),                        c(37, 1000, 36)))  # Note OOV tokens out <- data |>   layer_integer_lookup(vocabulary = vocab,                        num_oov_indices = 2) out ## tf.Tensor( ## [[2 4 5] ##  [1 0 3]], shape=(2, 3), dtype=int64) vocab <- c(12, 36, 1138, 42) |> as.integer() data <- op_array(c(12, 36, 1138, 42, 7), 'int32')  # Note OOV tokens layer <- layer_integer_lookup(vocabulary = vocab,                               output_mode = 'one_hot') layer(data) ## tf.Tensor( ## [[0 1 0 0 0] ##  [0 0 1 0 0] ##  [0 0 0 1 0] ##  [0 0 0 0 1] ##  [1 0 0 0 0]], shape=(5, 5), dtype=int64) vocab <- c(12, 36, 1138, 42) |> as.integer() data <- op_array(rbind(c(12, 1138, 42, 42),                       c(42,    7, 36,  7)), \"int64\")  # Note OOV tokens layer <- layer_integer_lookup(vocabulary = vocab,                               output_mode = 'multi_hot') layer(data) ## tf.Tensor( ## [[0 1 0 1 1] ##  [1 0 1 0 1]], shape=(2, 5), dtype=int64) vocab <- c(12, 36, 1138, 42) |> as.integer() data <- rbind(c(12, 1138, 42, 42),               c(42,    7, 36,  7)) |> op_array(\"int64\") layer <- layer_integer_lookup(vocabulary = vocab,                               output_mode = 'count') layer(data) ## tf.Tensor( ## [[0 1 0 1 2] ##  [2 0 1 0 1]], shape=(2, 5), dtype=int64) vocab <- c(12, 36, 1138, 42) |> as.integer() idf_weights <- c(0.25, 0.75, 0.6, 0.4) data <- rbind(c(12, 1138, 42, 42),               c(42,    7, 36,  7)) |> op_array(\"int64\") layer <- layer_integer_lookup(output_mode = 'tf_idf',                               vocabulary = vocab,                               idf_weights = idf_weights) layer(data) ## tf.Tensor( ## [[0.   0.25 0.   0.6  0.8 ] ##  [1.   0.   0.75 0.   0.4 ]], shape=(2, 5), dtype=float32) vocab <- c(-1, 12, 36, 1138, 42) |> as.integer() idf_weights <- c(0.9, 0.25, 0.75, 0.6, 0.4) data <- rbind(c(12, 1138, 42, 42),               c(42,    7, 36,  7)) |> op_array(\"int64\") layer <- layer_integer_lookup(output_mode = 'tf_idf',                               vocabulary = vocab,                               idf_weights = idf_weights) layer(data) ## tf.Tensor( ## [[0.   0.25 0.   0.6  0.8 ] ##  [1.8  0.   0.75 0.   0.4 ]], shape=(2, 5), dtype=float32) vocab <- c(12, 36, 1138, 42) |> as.integer() data <- op_array(c(1, 3, 4,                    4, 0, 2)) |> op_reshape(c(2,-1)) |> op_cast(\"int32\") layer <- layer_integer_lookup(vocabulary = vocab, invert = TRUE) layer(data) ## tf.Tensor( ## [[  12 1138   42] ##  [  42   -1   36]], shape=(2, 3), dtype=int64) vocab <- c(12, 36, 1138, 42) |> as.integer() data <- op_array(rbind(c(12, 1138, 42), c(42, 1000, 36)), \"int32\") layer <- layer_integer_lookup(vocabulary = vocab) i_layer <- layer_integer_lookup(vocabulary = get_vocabulary(layer),                                 invert = TRUE) int_data <- layer(data) i_layer(int_data) ## tf.Tensor( ## [[  12 1138   42] ##  [  42   -1   36]], shape=(2, 3), dtype=int64)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_jax_model_wrapper.html","id":null,"dir":"Reference","previous_headings":"","what":"Keras Layer that wraps a JAX model. — layer_jax_model_wrapper","title":"Keras Layer that wraps a JAX model. — layer_jax_model_wrapper","text":"layer enables use JAX components within Keras using JAX backend Keras.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_jax_model_wrapper.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Keras Layer that wraps a JAX model. — layer_jax_model_wrapper","text":"","code":"layer_jax_model_wrapper(   object,   call_fn,   init_fn = NULL,   params = NULL,   state = NULL,   seed = NULL,   ...,   dtype = NULL )"},{"path":"https://keras3.posit.co/dev/reference/layer_jax_model_wrapper.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Keras Layer that wraps a JAX model. — layer_jax_model_wrapper","text":"object Object compose layer . tensor, array, sequential model. call_fn function call model. See description list arguments takes outputs returns. init_fn function call initialize model. See description list arguments takes outputs returns. NULL, params /state must provided. params PyTree containing model trainable parameters. allows passing trained parameters controlling initialization. params state NULL, init_fn() called build time initialize trainable parameters model. state PyTree containing model non-trainable state. allows passing learned state controlling initialization. params state NULL, call_fn() takes state argument, init_fn() called build time initialize non-trainable state model. seed Seed random number generator. Optional. ... forward/backward compatability. dtype dtype layer's computations weights. Can also keras.DTypePolicy. Optional. Defaults default policy.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_jax_model_wrapper.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Keras Layer that wraps a JAX model. — layer_jax_model_wrapper","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_jax_model_wrapper.html","id":"model-function","dir":"Reference","previous_headings":"","what":"Model function","title":"Keras Layer that wraps a JAX model. — layer_jax_model_wrapper","text":"layer accepts JAX models form function, call_fn(), must take following arguments exact names: params: trainable parameters model. state (optional): non-trainable state model. Can omitted model non-trainable state. rng (optional): jax.random.PRNGKey instance. Can omitted model need RNGs, neither training inference. inputs: inputs model, JAX array PyTree arrays. training (optional): argument specifying training mode inference mode, TRUE passed training mode. Can omitted model behaves training mode inference mode. inputs argument mandatory. Inputs model must provided via single argument. JAX model takes multiple inputs separate arguments, must combined single structure, instance tuple() dict().","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_jax_model_wrapper.html","id":"model-weights-initialization","dir":"Reference","previous_headings":"","what":"Model weights initialization","title":"Keras Layer that wraps a JAX model. — layer_jax_model_wrapper","text":"initialization params state model can handled layer, case init_fn() argument must provided. allows model initialized dynamically right shape. Alternatively, shape known, params argument optionally state argument can used create already initialized model. init_fn() function, provided, must take following arguments exact names: rng: jax.random.PRNGKey instance. inputs: JAX array PyTree arrays placeholder values provide shape inputs. training (optional): argument specifying training mode inference mode. True always passed init_fn. Can omitted regardless whether call_fn training argument.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_jax_model_wrapper.html","id":"models-with-non-trainable-state","dir":"Reference","previous_headings":"","what":"Models with non-trainable state","title":"Keras Layer that wraps a JAX model. — layer_jax_model_wrapper","text":"JAX models non-trainable state: call_fn() must state argument call_fn() must return tuple() containing outputs model new non-trainable state model init_fn() must return tuple() containing initial trainable params model initial non-trainable state model. code shows possible combination call_fn() init_fn() signatures model non-trainable state. example, model training argument rng argument call_fn().","code":"stateful_call <- function(params, state, rng, inputs, training) {   outputs <- ....   new_state <- ....   tuple(outputs, new_state) }  stateful_init <- function(rng, inputs) {   initial_params <- ....   initial_state <- ....   tuple(initial_params, initial_state) }"},{"path":"https://keras3.posit.co/dev/reference/layer_jax_model_wrapper.html","id":"models-without-non-trainable-state","dir":"Reference","previous_headings":"","what":"Models without non-trainable state","title":"Keras Layer that wraps a JAX model. — layer_jax_model_wrapper","text":"JAX models non-trainable state: call_fn() must state argument call_fn() must return outputs model init_fn() must return initial trainable params model. code shows possible combination call_fn() init_fn() signatures model without non-trainable state. example, model training argument rng argument call_fn().","code":"stateful_call <- function(pparams, inputs) {   outputs <- ....   outputs }  stateful_init <- function(rng, inputs) {   initial_params <- ....   initial_params }"},{"path":"https://keras3.posit.co/dev/reference/layer_jax_model_wrapper.html","id":"conforming-to-the-required-signature","dir":"Reference","previous_headings":"","what":"Conforming to the required signature","title":"Keras Layer that wraps a JAX model. — layer_jax_model_wrapper","text":"model different signature one required JaxLayer, one can easily write wrapper method adapt arguments. example shows model multiple inputs separate arguments, expects multiple RNGs dict, deterministic argument opposite meaning training. conform, inputs combined single structure using tuple, RNG split used populate expected dict, Boolean flag negated:","code":"jax <- import(\"jax\") my_model_fn <- function(params, rngs, input1, input2, deterministic) {   ....   if (!deterministic) {     dropout_rng <- rngs$dropout     keep <- jax$random$bernoulli(dropout_rng, dropout_rate, x$shape)     x <- jax$numpy$where(keep, x / dropout_rate, 0)     ....   }   ....   return(outputs) }  my_model_wrapper_fn <- function(params, rng, inputs, training) {   c(input1, input2) %<-% inputs   c(rng1, rng2) %<-% jax$random$split(rng)   rngs <-  list(dropout = rng1, preprocessing = rng2)   deterministic <-  !training   my_model_fn(params, rngs, input1, input2, deterministic) }  keras_layer <- layer_jax_model_wrapper(call_fn = my_model_wrapper_fn,                                        params = initial_params)"},{"path":"https://keras3.posit.co/dev/reference/layer_jax_model_wrapper.html","id":"usage-with-haiku-modules","dir":"Reference","previous_headings":"","what":"Usage with Haiku modules","title":"Keras Layer that wraps a JAX model. — layer_jax_model_wrapper","text":"JaxLayer enables use Haiku components form haiku.Module. achieved transforming module per Haiku pattern passing module.apply call_fn parameter module.init init_fn parameter needed. model non-trainable state, transformed haiku.transform_with_state. model non-trainable state, transformed haiku.transform. Additionally, optionally, module use RNGs \"apply\", can transformed haiku.without_apply_rng. following example shows create JaxLayer Haiku module uses random number generators via hk.next_rng_key() takes training positional argument:","code":"# reticulate::py_install(\"haiku\", \"r-keras\") hk <- import(\"haiku\") MyHaikuModule(hk$Module) \\%py_class\\% {    `__call__` <- \\(self, x, training) {     x <- hk$Conv2D(32L, tuple(3L, 3L))(x)     x <- jax$nn$relu(x)     x <- hk$AvgPool(tuple(1L, 2L, 2L, 1L),                     tuple(1L, 2L, 2L, 1L), \"VALID\")(x)     x <- hk$Flatten()(x)     x <- hk$Linear(200L)(x)     if (training)       x <- hk$dropout(rng = hk$next_rng_key(), rate = 0.3, x = x)     x <- jax$nn$relu(x)     x <- hk$Linear(10L)(x)     x <- jax$nn$softmax(x)     x   }  }  my_haiku_module_fn <- function(inputs, training) {   module <- MyHaikuModule()   module(inputs, training) }  transformed_module <- hk$transform(my_haiku_module_fn)  keras_layer <-   layer_jax_model_wrapper(call_fn = transformed_module$apply,                           init_fn = transformed_module$init)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_lambda.html","id":null,"dir":"Reference","previous_headings":"","what":"Wraps arbitrary expressions as a Layer object. — layer_lambda","title":"Wraps arbitrary expressions as a Layer object. — layer_lambda","text":"layer_lambda() layer exists arbitrary expressions can used Layer constructing Sequential Functional API models. Lambda layers best suited simple operations quick experimentation. advanced use cases, prefer writing new subclasses Layer using new_layer_class().","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_lambda.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Wraps arbitrary expressions as a Layer object. — layer_lambda","text":"","code":"layer_lambda(   object,   f,   output_shape = NULL,   mask = NULL,   arguments = NULL,   ... )"},{"path":"https://keras3.posit.co/dev/reference/layer_lambda.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Wraps arbitrary expressions as a Layer object. — layer_lambda","text":"object Object compose layer . tensor, array, sequential model. f function evaluated. Takes input tensor first argument. output_shape Expected output shape function. argument can usually inferred explicitly provided. Can list function. list, specifies first dimension onward; sample dimension assumed either input: output_shape = c(input_shape[1], output_shape) , input NULL sample dimension also NULL: output_shape = c(NA, output_shape). function, specifies entire shape function input shape: output_shape = f(input_shape). mask Either NULL (indicating masking) callable signature compute_mask layer method, tensor returned output mask regardless input . arguments Optional named list arguments passed function. ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_lambda.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Wraps arbitrary expressions as a Layer object. — layer_lambda","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_lambda.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Wraps arbitrary expressions as a Layer object. — layer_lambda","text":"","code":"# add a x -> x^2 layer model <- keras_model_sequential() model |> layer_lambda(\\(x) x^2)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_layer_normalization.html","id":null,"dir":"Reference","previous_headings":"","what":"Layer normalization layer (Ba et al., 2016). — layer_layer_normalization","title":"Layer normalization layer (Ba et al., 2016). — layer_layer_normalization","text":"Normalize activations previous layer given example batch independently, rather across batch like Batch Normalization. .e. applies transformation maintains mean activation within example close 0 activation standard deviation close 1. scale center enabled, layer scale normalized outputs broadcasting trainable variable gamma, center outputs broadcasting trainable variable beta. gamma default ones tensor beta default zeros tensor, centering scaling -ops training begun. , scaling centering enabled normalization equations follows: Let intermediate activations mini-batch inputs. sample x batch inputs, compute mean variance sample, normalize value sample (including small factor epsilon numerical stability), finally, transform normalized output gamma beta, learned parameters:   gamma beta span axes inputs specified axis, part inputs' shape must fully defined. example:         Note implementations layer normalization may choose define gamma beta separate set axes axes normalized across. example, Group Normalization (Wu et al. 2018) group size 1 corresponds layer_layer_normalization() normalizes across height, width, channel gamma beta span channel dimension. , layer_layer_normalization() implementation match layer_group_normalization() layer group size set 1.","code":"outputs <- inputs |> apply(1, function(x) {   x_normalized <- (x - mean(x)) /                   sqrt(var(x) + epsilon)   x_normalized * gamma + beta }) layer <- layer_layer_normalization(axis = c(2, 3, 4))  layer(op_ones(c(5, 20, 30, 40))) |> invisible() # build() shape(layer$beta) ## shape(20, 30, 40) shape(layer$gamma) ## shape(20, 30, 40)"},{"path":"https://keras3.posit.co/dev/reference/layer_layer_normalization.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Layer normalization layer (Ba et al., 2016). — layer_layer_normalization","text":"","code":"layer_layer_normalization(   object,   axis = -1L,   epsilon = 0.001,   center = TRUE,   scale = TRUE,   rms_scaling = FALSE,   beta_initializer = \"zeros\",   gamma_initializer = \"ones\",   beta_regularizer = NULL,   gamma_regularizer = NULL,   beta_constraint = NULL,   gamma_constraint = NULL,   ... )"},{"path":"https://keras3.posit.co/dev/reference/layer_layer_normalization.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Layer normalization layer (Ba et al., 2016). — layer_layer_normalization","text":"object Object compose layer . tensor, array, sequential model. axis Integer list. axis axes normalize across. Typically, features axis/axes. left-axes typically batch axis/axes. -1 last dimension input. Defaults -1. epsilon Small float added variance avoid dividing zero. Defaults 1e-3. center TRUE, add offset beta normalized tensor. FALSE, beta ignored. Defaults TRUE. scale TRUE, multiply gamma. FALSE, gamma used. next layer linear (also e.g. layer_activation_relu()), can disabled since scaling done next layer. Defaults TRUE. rms_scaling TRUE, center scale ignored, inputs scaled gamma inverse square root square inputs. approximate faster approach avoids ever computing mean input. Note equivalent computation layer_rms_normalization layer performs. beta_initializer Initializer beta weight. Defaults zeros. gamma_initializer Initializer gamma weight. Defaults ones. beta_regularizer Optional regularizer beta weight. NULL default. gamma_regularizer Optional regularizer gamma weight. NULL default. beta_constraint Optional constraint beta weight. NULL default. gamma_constraint Optional constraint gamma weight. NULL default. ... Base layer keyword arguments (e.g. name dtype).","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_layer_normalization.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Layer normalization layer (Ba et al., 2016). — layer_layer_normalization","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_layer_normalization.html","id":"reference","dir":"Reference","previous_headings":"","what":"Reference","title":"Layer normalization layer (Ba et al., 2016). — layer_layer_normalization","text":"Lei Ba et al., 2016.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_lstm.html","id":null,"dir":"Reference","previous_headings":"","what":"Long Short-Term Memory layer - Hochreiter 1997. — layer_lstm","title":"Long Short-Term Memory layer - Hochreiter 1997. — layer_lstm","text":"Based available runtime hardware constraints, layer choose different implementations (cuDNN-based backend-native) maximize performance. GPU available arguments layer meet requirement cuDNN kernel (see details), layer use fast cuDNN implementation using TensorFlow backend. requirements use cuDNN implementation : activation == tanh recurrent_activation == sigmoid recurrent_dropout == 0 unroll FALSE use_bias TRUE Inputs, use masking, strictly right-padded. Eager execution enabled outermost context. example:","code":"input <- random_uniform(c(32, 10, 8)) output <- input |> layer_lstm(4) shape(output) ## shape(32, 4) lstm <- layer_lstm(units = 4, return_sequences = TRUE, return_state = TRUE) c(whole_seq_output, final_memory_state, final_carry_state) %<-% lstm(input) shape(whole_seq_output) ## shape(32, 10, 4) shape(final_memory_state) ## shape(32, 4) shape(final_carry_state) ## shape(32, 4)"},{"path":"https://keras3.posit.co/dev/reference/layer_lstm.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Long Short-Term Memory layer - Hochreiter 1997. — layer_lstm","text":"","code":"layer_lstm(   object,   units,   activation = \"tanh\",   recurrent_activation = \"sigmoid\",   use_bias = TRUE,   kernel_initializer = \"glorot_uniform\",   recurrent_initializer = \"orthogonal\",   bias_initializer = \"zeros\",   unit_forget_bias = TRUE,   kernel_regularizer = NULL,   recurrent_regularizer = NULL,   bias_regularizer = NULL,   activity_regularizer = NULL,   kernel_constraint = NULL,   recurrent_constraint = NULL,   bias_constraint = NULL,   dropout = 0,   recurrent_dropout = 0,   seed = NULL,   return_sequences = FALSE,   return_state = FALSE,   go_backwards = FALSE,   stateful = FALSE,   unroll = FALSE,   use_cudnn = \"auto\",   ... )"},{"path":"https://keras3.posit.co/dev/reference/layer_lstm.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Long Short-Term Memory layer - Hochreiter 1997. — layer_lstm","text":"object Object compose layer . tensor, array, sequential model. units Positive integer, dimensionality output space. activation Activation function use. Default: hyperbolic tangent (tanh). pass NULL, activation applied (ie. \"linear\" activation: (x) = x). recurrent_activation Activation function use recurrent step. Default: sigmoid (sigmoid). pass NULL, activation applied (ie. \"linear\" activation: (x) = x). use_bias Boolean, (default TRUE), whether layer use bias vector. kernel_initializer Initializer kernel weights matrix, used linear transformation inputs. Default: \"glorot_uniform\". recurrent_initializer Initializer recurrent_kernel weights matrix, used linear transformation recurrent state. Default: \"orthogonal\". bias_initializer Initializer bias vector. Default: \"zeros\". unit_forget_bias Boolean (default TRUE). TRUE, add 1 bias forget gate initialization. Setting TRUE also force bias_initializer=\"zeros\". recommended Jozefowicz et al. kernel_regularizer Regularizer function applied kernel weights matrix. Default: NULL. recurrent_regularizer Regularizer function applied recurrent_kernel weights matrix. Default: NULL. bias_regularizer Regularizer function applied bias vector. Default: NULL. activity_regularizer Regularizer function applied output layer (\"activation\"). Default: NULL. kernel_constraint Constraint function applied kernel weights matrix. Default: NULL. recurrent_constraint Constraint function applied recurrent_kernel weights matrix. Default: NULL. bias_constraint Constraint function applied bias vector. Default: NULL. dropout Float 0 1. Fraction units drop linear transformation inputs. Default: 0. recurrent_dropout Float 0 1. Fraction units drop linear transformation recurrent state. Default: 0. seed Random seed dropout. return_sequences Boolean. Whether return last output output sequence, full sequence. Default: FALSE. return_state Boolean. Whether return last state addition output. Default: FALSE. go_backwards Boolean (default: FALSE). TRUE, process input sequence backwards return reversed sequence. stateful Boolean (default: FALSE). TRUE, last state sample index batch used initial state sample index following batch. unroll Boolean (default FALSE). TRUE, network unrolled, else symbolic loop used. Unrolling can speed-RNN, although tends memory-intensive. Unrolling suitable short sequences. use_cudnn Whether use cuDNN-backed implementation. \"auto\" attempt use cuDNN feasible, fallback default implementation . ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_lstm.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Long Short-Term Memory layer - Hochreiter 1997. — layer_lstm","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_lstm.html","id":"call-arguments","dir":"Reference","previous_headings":"","what":"Call Arguments","title":"Long Short-Term Memory layer - Hochreiter 1997. — layer_lstm","text":"inputs: 3D tensor, shape (batch, timesteps, feature). mask: Binary tensor shape (samples, timesteps) indicating whether given timestep masked  (optional). individual TRUE entry indicates corresponding timestep utilized, FALSE entry indicates corresponding timestep ignored. Defaults NULL. training: Boolean indicating whether layer behave training mode inference mode. argument passed cell calling . relevant dropout recurrent_dropout used  (optional). Defaults NULL. initial_state: List initial state tensors passed first call cell (optional, NULL causes creation zero-filled initial state tensors). Defaults NULL.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_masking.html","id":null,"dir":"Reference","previous_headings":"","what":"Masks a sequence by using a mask value to skip timesteps. — layer_masking","title":"Masks a sequence by using a mask value to skip timesteps. — layer_masking","text":"timestep input tensor (second dimension tensor), values input tensor timestep equal mask_value, timestep masked (skipped) downstream layers (long support masking). downstream layer support masking yet receives input mask, exception raised.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_masking.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Masks a sequence by using a mask value to skip timesteps. — layer_masking","text":"","code":"layer_masking(object, mask_value = 0, ...)"},{"path":"https://keras3.posit.co/dev/reference/layer_masking.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Masks a sequence by using a mask value to skip timesteps. — layer_masking","text":"object Object compose layer . tensor, array, sequential model. mask_value see description ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_masking.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Masks a sequence by using a mask value to skip timesteps. — layer_masking","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_masking.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Masks a sequence by using a mask value to skip timesteps. — layer_masking","text":"Consider array x shape c(samples, timesteps, features), fed LSTM layer. want mask timestep #3 #5 lack data timesteps. can: Set x[, 3, ] <- 0. x[, 5, ] <- 0. Insert layer_masking() layer mask_value = 0. LSTM layer:","code":"c(samples, timesteps, features) %<-% c(32, 10, 8) inputs <- c(samples, timesteps, features) %>% { array(runif(prod(.)), dim = .) } inputs[, 3, ] <- 0 inputs[, 5, ] <- 0  model <- keras_model_sequential() %>%   layer_masking(mask_value = 0) %>%   layer_lstm(32)  output <- model(inputs) # The time step 3 and 5 will be skipped from LSTM calculation."},{"path":"https://keras3.posit.co/dev/reference/layer_masking.html","id":"note","dir":"Reference","previous_headings":"","what":"Note","title":"Masks a sequence by using a mask value to skip timesteps. — layer_masking","text":"Keras masking convention, masked timestep denoted mask value FALSE, non-masked (.e. usable) timestep denoted mask value TRUE.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_max_num_bounding_boxes.html","id":null,"dir":"Reference","previous_headings":"","what":"Ensure the maximum number of bounding boxes. — layer_max_num_bounding_boxes","title":"Ensure the maximum number of bounding boxes. — layer_max_num_bounding_boxes","text":"Ensure maximum number bounding boxes.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_max_num_bounding_boxes.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Ensure the maximum number of bounding boxes. — layer_max_num_bounding_boxes","text":"","code":"layer_max_num_bounding_boxes(object, max_number, fill_value = -1L, ...)"},{"path":"https://keras3.posit.co/dev/reference/layer_max_num_bounding_boxes.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Ensure the maximum number of bounding boxes. — layer_max_num_bounding_boxes","text":"object Object compose layer . tensor, array, sequential model. max_number Desired output number bounding boxes. fill_value fill value boxes labels bounding_boxes. Defaults -1. ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_max_num_bounding_boxes.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Ensure the maximum number of bounding boxes. — layer_max_num_bounding_boxes","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_max_pooling_1d.html","id":null,"dir":"Reference","previous_headings":"","what":"Max pooling operation for 1D temporal data. — layer_max_pooling_1d","title":"Max pooling operation for 1D temporal data. — layer_max_pooling_1d","text":"Downsamples input representation taking maximum value spatial window size pool_size. window shifted strides. resulting output using \"valid\" padding option shape : output_shape = (input_shape - pool_size + 1) / strides). resulting output shape using \"\" padding option : output_shape = input_shape / strides","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_max_pooling_1d.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Max pooling operation for 1D temporal data. — layer_max_pooling_1d","text":"","code":"layer_max_pooling_1d(   object,   pool_size = 2L,   strides = NULL,   padding = \"valid\",   data_format = NULL,   name = NULL,   ... )"},{"path":"https://keras3.posit.co/dev/reference/layer_max_pooling_1d.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Max pooling operation for 1D temporal data. — layer_max_pooling_1d","text":"object Object compose layer . tensor, array, sequential model. pool_size int, size max pooling window. strides int NULL. Specifies much pooling window moves pooling step. NULL, default pool_size. padding string, either \"valid\" \"\" (case-insensitive). \"valid\" means padding. \"\" results padding evenly left/right /input output height/width dimension input. data_format string, either \"channels_last\" \"channels_first\". ordering dimensions inputs. \"channels_last\" corresponds inputs shape (batch, steps, features) \"channels_first\" corresponds inputs shape (batch, features, steps). defaults image_data_format value found Keras config file ~/.keras/keras.json. never set , \"channels_last\". name String, name object ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_max_pooling_1d.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Max pooling operation for 1D temporal data. — layer_max_pooling_1d","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_max_pooling_1d.html","id":"input-shape","dir":"Reference","previous_headings":"","what":"Input Shape","title":"Max pooling operation for 1D temporal data. — layer_max_pooling_1d","text":"data_format=\"channels_last\": 3D tensor shape (batch_size, steps, features). data_format=\"channels_first\": 3D tensor shape (batch_size, features, steps).","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_max_pooling_1d.html","id":"output-shape","dir":"Reference","previous_headings":"","what":"Output Shape","title":"Max pooling operation for 1D temporal data. — layer_max_pooling_1d","text":"data_format=\"channels_last\": 3D tensor shape (batch_size, downsampled_steps, features). data_format=\"channels_first\": 3D tensor shape (batch_size, features, downsampled_steps).","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_max_pooling_1d.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Max pooling operation for 1D temporal data. — layer_max_pooling_1d","text":"strides=1 padding=\"valid\":     strides=2 padding=\"valid\":     strides=1 padding=\"\":","code":"x <- op_reshape(c(1, 2, 3, 4, 5),                c(1, 5, 1)) max_pool_1d <- layer_max_pooling_1d(pool_size = 2,                                     strides = 1,                                     padding = \"valid\") max_pool_1d(x) ## tf.Tensor( ## [[[2.] ##   [3.] ##   [4.] ##   [5.]]], shape=(1, 4, 1), dtype=float32) x <- op_reshape(c(1, 2, 3, 4, 5),                c(1, 5, 1)) max_pool_1d <- layer_max_pooling_1d(pool_size = 2,                                     strides = 2,                                     padding = \"valid\") max_pool_1d(x) ## tf.Tensor( ## [[[2.] ##   [4.]]], shape=(1, 2, 1), dtype=float32) x <- op_reshape(c(1, 2, 3, 4, 5),                c(1, 5, 1)) max_pool_1d <- layer_max_pooling_1d(pool_size = 2,                                     strides = 1,                                     padding = \"same\") max_pool_1d(x) ## tf.Tensor( ## [[[2.] ##   [3.] ##   [4.] ##   [5.] ##   [5.]]], shape=(1, 5, 1), dtype=float32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_max_pooling_2d.html","id":null,"dir":"Reference","previous_headings":"","what":"Max pooling operation for 2D spatial data. — layer_max_pooling_2d","title":"Max pooling operation for 2D spatial data. — layer_max_pooling_2d","text":"Downsamples input along spatial dimensions (height width) taking maximum value input window (size defined pool_size) channel input. window shifted strides along dimension. resulting output using \"valid\" padding option spatial shape (number rows columns) : output_shape = floor((input_shape - pool_size) / strides) + 1 (input_shape >= pool_size) resulting output shape using \"\" padding option : output_shape = floor((input_shape - 1) / strides) + 1","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_max_pooling_2d.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Max pooling operation for 2D spatial data. — layer_max_pooling_2d","text":"","code":"layer_max_pooling_2d(   object,   pool_size = list(2L, 2L),   strides = NULL,   padding = \"valid\",   data_format = NULL,   name = NULL,   ... )"},{"path":"https://keras3.posit.co/dev/reference/layer_max_pooling_2d.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Max pooling operation for 2D spatial data. — layer_max_pooling_2d","text":"object Object compose layer . tensor, array, sequential model. pool_size int list 2 integers, factors downscale (dim1, dim2). one integer specified, window length used dimensions. strides int list 2 integers, NULL. Strides values. NULL, default pool_size. one int specified, stride size used dimensions. padding string, either \"valid\" \"\" (case-insensitive). \"valid\" means padding. \"\" results padding evenly left/right /input output height/width dimension input. data_format string, either \"channels_last\" \"channels_first\". ordering dimensions inputs. \"channels_last\" corresponds inputs shape (batch, height, width, channels) \"channels_first\" corresponds inputs shape (batch, channels, height, width). defaults image_data_format value found Keras config file ~/.keras/keras.json. never set , \"channels_last\". name String, name object ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_max_pooling_2d.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Max pooling operation for 2D spatial data. — layer_max_pooling_2d","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_max_pooling_2d.html","id":"input-shape","dir":"Reference","previous_headings":"","what":"Input Shape","title":"Max pooling operation for 2D spatial data. — layer_max_pooling_2d","text":"data_format=\"channels_last\": 4D tensor shape (batch_size, height, width, channels). data_format=\"channels_first\": 4D tensor shape (batch_size, channels, height, width).","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_max_pooling_2d.html","id":"output-shape","dir":"Reference","previous_headings":"","what":"Output Shape","title":"Max pooling operation for 2D spatial data. — layer_max_pooling_2d","text":"data_format=\"channels_last\": 4D tensor shape (batch_size, pooled_height, pooled_width, channels). data_format=\"channels_first\": 4D tensor shape (batch_size, channels, pooled_height, pooled_width).","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_max_pooling_2d.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Max pooling operation for 2D spatial data. — layer_max_pooling_2d","text":"strides = (1, 1) padding = \"valid\":     strides = c(2, 2) padding = \"valid\":     stride = (1, 1) padding = \"\":","code":"x <- rbind(c(1., 2., 3.),            c(4., 5., 6.),            c(7., 8., 9.)) |> op_reshape(c(1, 3, 3, 1)) max_pool_2d <- layer_max_pooling_2d(pool_size = c(2, 2),                                     strides = c(1, 1),                                     padding = \"valid\") max_pool_2d(x) ## tf.Tensor( ## [[[[5.] ##    [6.]] ## ##   [[8.] ##    [9.]]]], shape=(1, 2, 2, 1), dtype=float32) x <- rbind(c(1., 2., 3., 4.),            c(5., 6., 7., 8.),            c(9., 10., 11., 12.)) |> op_reshape(c(1, 3, 4, 1)) max_pool_2d <- layer_max_pooling_2d(pool_size = c(2, 2),                                     strides = c(2, 2),                                     padding = \"valid\") max_pool_2d(x) ## tf.Tensor( ## [[[[6.] ##    [8.]]]], shape=(1, 1, 2, 1), dtype=float32) x <- rbind(c(1., 2., 3.),            c(4., 5., 6.),            c(7., 8., 9.)) |> op_reshape(c(1, 3, 3, 1)) max_pool_2d <- layer_max_pooling_2d(pool_size = c(2, 2),                                     strides = c(1, 1),                                     padding = \"same\") max_pool_2d(x) ## tf.Tensor( ## [[[[5.] ##    [6.] ##    [6.]] ## ##   [[8.] ##    [9.] ##    [9.]] ## ##   [[8.] ##    [9.] ##    [9.]]]], shape=(1, 3, 3, 1), dtype=float32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_max_pooling_3d.html","id":null,"dir":"Reference","previous_headings":"","what":"Max pooling operation for 3D data (spatial or spatio-temporal). — layer_max_pooling_3d","title":"Max pooling operation for 3D data (spatial or spatio-temporal). — layer_max_pooling_3d","text":"Downsamples input along spatial dimensions (depth, height, width) taking maximum value input window (size defined pool_size) channel input. window shifted strides along dimension.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_max_pooling_3d.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Max pooling operation for 3D data (spatial or spatio-temporal). — layer_max_pooling_3d","text":"","code":"layer_max_pooling_3d(   object,   pool_size = list(2L, 2L, 2L),   strides = NULL,   padding = \"valid\",   data_format = NULL,   name = NULL,   ... )"},{"path":"https://keras3.posit.co/dev/reference/layer_max_pooling_3d.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Max pooling operation for 3D data (spatial or spatio-temporal). — layer_max_pooling_3d","text":"object Object compose layer . tensor, array, sequential model. pool_size int list 3 integers, factors downscale (dim1, dim2, dim3). one integer specified, window length used dimensions. strides int list 3 integers, NULL. Strides values. NULL, default pool_size. one int specified, stride size used dimensions. padding string, either \"valid\" \"\" (case-insensitive). \"valid\" means padding. \"\" results padding evenly left/right /input output height/width dimension input. data_format string, either \"channels_last\" \"channels_first\". ordering dimensions inputs. \"channels_last\" corresponds inputs shape (batch, spatial_dim1, spatial_dim2, spatial_dim3, channels) \"channels_first\" corresponds inputs shape (batch, channels, spatial_dim1, spatial_dim2, spatial_dim3). defaults image_data_format value found Keras config file ~/.keras/keras.json. never set , \"channels_last\". name String, name object ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_max_pooling_3d.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Max pooling operation for 3D data (spatial or spatio-temporal). — layer_max_pooling_3d","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_max_pooling_3d.html","id":"input-shape","dir":"Reference","previous_headings":"","what":"Input Shape","title":"Max pooling operation for 3D data (spatial or spatio-temporal). — layer_max_pooling_3d","text":"data_format=\"channels_last\": 5D tensor shape: (batch_size, spatial_dim1, spatial_dim2, spatial_dim3, channels) data_format=\"channels_first\": 5D tensor shape: (batch_size, channels, spatial_dim1, spatial_dim2, spatial_dim3)","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_max_pooling_3d.html","id":"output-shape","dir":"Reference","previous_headings":"","what":"Output Shape","title":"Max pooling operation for 3D data (spatial or spatio-temporal). — layer_max_pooling_3d","text":"data_format=\"channels_last\": 5D tensor shape: (batch_size, pooled_dim1, pooled_dim2, pooled_dim3, channels) data_format=\"channels_first\": 5D tensor shape: (batch_size, channels, pooled_dim1, pooled_dim2, pooled_dim3)","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_max_pooling_3d.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Max pooling operation for 3D data (spatial or spatio-temporal). — layer_max_pooling_3d","text":"","code":"depth <- 30 height <- 30 width <- 30 channels <- 3  inputs <- layer_input(shape=c(depth, height, width, channels)) layer <- layer_max_pooling_3d(pool_size=3) outputs <- inputs |> layer() outputs ## <KerasTensor shape=(None, 10, 10, 10, 3), dtype=float32, sparse=False, ragged=False, name=keras_tensor_1>"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_maximum.html","id":null,"dir":"Reference","previous_headings":"","what":"Computes element-wise maximum on a list of inputs. — layer_maximum","title":"Computes element-wise maximum on a list of inputs. — layer_maximum","text":"takes input list tensors, shape, returns single tensor (also shape).","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_maximum.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes element-wise maximum on a list of inputs. — layer_maximum","text":"","code":"layer_maximum(inputs, ...)"},{"path":"https://keras3.posit.co/dev/reference/layer_maximum.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Computes element-wise maximum on a list of inputs. — layer_maximum","text":"inputs layers combine ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_maximum.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Computes element-wise maximum on a list of inputs. — layer_maximum","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_maximum.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Computes element-wise maximum on a list of inputs. — layer_maximum","text":"Usage Keras model:","code":"input_shape <- c(2, 3, 4) x1 <- random_uniform(input_shape) x2 <- random_uniform(input_shape) y <- layer_maximum(x1, x2) input1 <- layer_input(shape = c(16)) x1 <- input1 |> layer_dense(8, activation = 'relu') input2 <- layer_input(shape = c(32)) x2 <- input2 |> layer_dense(8, activation = 'relu') # equivalent to `y <- layer_maximum(x1, x2)` y <- layer_maximum(x1, x2) out <- y |> layer_dense(4) model <- keras_model(inputs = c(input1, input2), outputs = out)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_mel_spectrogram.html","id":null,"dir":"Reference","previous_headings":"","what":"A preprocessing layer to convert raw audio signals to Mel spectrograms. — layer_mel_spectrogram","title":"A preprocessing layer to convert raw audio signals to Mel spectrograms. — layer_mel_spectrogram","text":"layer takes float32/float64 single batched audio signal inputs computes Mel spectrogram using Short-Time Fourier Transform Mel scaling. input 1D (unbatched) 2D (batched) tensor representing audio signals. output 2D 3D tensor representing Mel spectrograms. spectrogram image-like representation shows frequency spectrum signal time. uses x-axis represent time, y-axis represent frequency, pixel represent intensity. Mel spectrograms special type spectrogram use mel scale, approximates humans perceive sound. commonly used speech music processing tasks like speech recognition, speaker identification, music genre classification.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_mel_spectrogram.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"A preprocessing layer to convert raw audio signals to Mel spectrograms. — layer_mel_spectrogram","text":"","code":"layer_mel_spectrogram(   object,   fft_length = 2048L,   sequence_stride = 512L,   sequence_length = NULL,   window = \"hann\",   sampling_rate = 16000L,   num_mel_bins = 128L,   min_freq = 20,   max_freq = NULL,   power_to_db = TRUE,   top_db = 80,   mag_exp = 2,   min_power = 1e-10,   ref_power = 1,   ... )"},{"path":"https://keras3.posit.co/dev/reference/layer_mel_spectrogram.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"A preprocessing layer to convert raw audio signals to Mel spectrograms. — layer_mel_spectrogram","text":"object Object compose layer . tensor, array, sequential model. fft_length Integer, size FFT window. sequence_stride Integer, number samples successive STFT columns. sequence_length Integer, size window used applying window audio frame. NULL, defaults fft_length. window String, name window function use. Available values \"hann\" \"hamming\". window tensor, used directly window length must sequence_length. window NULL, windowing used. Defaults \"hann\". sampling_rate Integer, sample rate input signal. num_mel_bins Integer, number mel bins generate. min_freq Float, minimum frequency mel bins. max_freq Float, maximum frequency mel bins. NULL, defaults sampling_rate / 2. power_to_db TRUE, convert power spectrogram decibels. top_db Float, minimum negative cut-max(10 * log10(S)) - top_db. mag_exp Float, exponent magnitude spectrogram. 1 magnitude, 2 power, etc. Default 2. min_power Float, minimum value power ref_power. ref_power Float, power scaled relative 10 * log10(S / ref_power). ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_mel_spectrogram.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"A preprocessing layer to convert raw audio signals to Mel spectrograms. — layer_mel_spectrogram","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_mel_spectrogram.html","id":"references","dir":"Reference","previous_headings":"","what":"References","title":"A preprocessing layer to convert raw audio signals to Mel spectrograms. — layer_mel_spectrogram","text":"Spectrogram, Mel scale.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_mel_spectrogram.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"A preprocessing layer to convert raw audio signals to Mel spectrograms. — layer_mel_spectrogram","text":"Unbatched audio signal   Batched audio signal","code":"layer <- layer_mel_spectrogram(   num_mel_bins = 64,   sampling_rate = 8000,   sequence_stride = 256,   fft_length = 2048 ) layer(random_uniform(shape = c(16000))) |> shape() layer <- layer_mel_spectrogram(   num_mel_bins = 80,   sampling_rate = 8000,   sequence_stride = 128,   fft_length = 2048 ) layer(random_uniform(shape = c(2, 16000))) |> shape()"},{"path":"https://keras3.posit.co/dev/reference/layer_mel_spectrogram.html","id":"input-shape","dir":"Reference","previous_headings":"","what":"Input Shape","title":"A preprocessing layer to convert raw audio signals to Mel spectrograms. — layer_mel_spectrogram","text":"1D (unbatched) 2D (batched) tensor shape:(..., samples).","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_mel_spectrogram.html","id":"output-shape","dir":"Reference","previous_headings":"","what":"Output Shape","title":"A preprocessing layer to convert raw audio signals to Mel spectrograms. — layer_mel_spectrogram","text":"2D (unbatched) 3D (batched) tensor shape:(..., num_mel_bins, time).","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_minimum.html","id":null,"dir":"Reference","previous_headings":"","what":"Computes elementwise minimum on a list of inputs. — layer_minimum","title":"Computes elementwise minimum on a list of inputs. — layer_minimum","text":"takes input list tensors, shape, returns single tensor (also shape).","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_minimum.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes elementwise minimum on a list of inputs. — layer_minimum","text":"","code":"layer_minimum(inputs, ...)"},{"path":"https://keras3.posit.co/dev/reference/layer_minimum.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Computes elementwise minimum on a list of inputs. — layer_minimum","text":"inputs layers combine ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_minimum.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Computes elementwise minimum on a list of inputs. — layer_minimum","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_minimum.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Computes elementwise minimum on a list of inputs. — layer_minimum","text":"Usage Keras model:","code":"input_shape <- c(2, 3, 4) x1 <- random_uniform(input_shape) x2 <- random_uniform(input_shape) y <- layer_minimum(x1, x2) input1 <- layer_input(shape = c(16)) x1 <- input1 |> layer_dense(8, activation = 'relu') input2 <- layer_input(shape = c(32)) x2 <- input2 |> layer_dense(8, activation = 'relu') # equivalent to `y <- layer_minimum(x1, x2)` y <- layer_minimum(x1, x2) out <- y |> layer_dense(4) model <- keras_model(inputs = c(input1, input2), outputs = out)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_mix_up.html","id":null,"dir":"Reference","previous_headings":"","what":"MixUp implements the MixUp data augmentation technique. — layer_mix_up","title":"MixUp implements the MixUp data augmentation technique. — layer_mix_up","text":"MixUp implements MixUp data augmentation technique.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_mix_up.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"MixUp implements the MixUp data augmentation technique. — layer_mix_up","text":"","code":"layer_mix_up(object, alpha = 0.2, data_format = NULL, seed = NULL, ...)"},{"path":"https://keras3.posit.co/dev/reference/layer_mix_up.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"MixUp implements the MixUp data augmentation technique. — layer_mix_up","text":"object Object compose layer . tensor, array, sequential model. alpha Float 0 1. Controls blending strength. Smaller values mean less mixing, larger values allow  blending images. Defaults 0.2, recommended ImageNet1k classification. data_format String, one \"channels_last\" (default) \"channels_first\". ordering dimensions inputs. \"channels_last\" corresponds inputs shape (batch, height, width, channels) \"channels_first\" corresponds inputs shape (batch, channels, height, width). seed Integer. Used create random seed. ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_mix_up.html","id":"references","dir":"Reference","previous_headings":"","what":"References","title":"MixUp implements the MixUp data augmentation technique. — layer_mix_up","text":"MixUp paper. MixUp Object Detection paper.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_mix_up.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"MixUp implements the MixUp data augmentation technique. — layer_mix_up","text":"","code":"c(c(images, labels), .) %<-% dataset_cifar10() c(images, labels) %<-% list(images[1:8,,,], labels[1:8,]) labels <- labels |> op_one_hot(10) |> op_cast(\"float32\") mix_up <- layer_mix_up(alpha=0.2) output <- mix_up(list(images = images, labels = labels))"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_multi_head_attention.html","id":null,"dir":"Reference","previous_headings":"","what":"Multi Head Attention layer. — layer_multi_head_attention","title":"Multi Head Attention layer. — layer_multi_head_attention","text":"implementation multi-headed attention described paper \"Attention Need\" Vaswani et al., 2017. query, key, value , self-attention. timestep query attends corresponding sequence key, returns fixed-width vector. layer first projects query, key value. (effectively) list tensors length num_attention_heads, corresponding shapes (batch_size, <query dimensions>, key_dim), (batch_size, <key/value dimensions>, key_dim), (batch_size, <key/value dimensions>, value_dim). , query key tensors dot-producted scaled. softmaxed obtain attention probabilities. value tensors interpolated probabilities, concatenated back single tensor. Finally, result tensor last dimension value_dim can take linear projection return.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_multi_head_attention.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Multi Head Attention layer. — layer_multi_head_attention","text":"","code":"layer_multi_head_attention(   inputs,   num_heads,   key_dim,   value_dim = NULL,   dropout = 0,   use_bias = TRUE,   output_shape = NULL,   attention_axes = NULL,   flash_attention = NULL,   kernel_initializer = \"glorot_uniform\",   bias_initializer = \"zeros\",   kernel_regularizer = NULL,   bias_regularizer = NULL,   activity_regularizer = NULL,   kernel_constraint = NULL,   bias_constraint = NULL,   seed = NULL,   ... )"},{"path":"https://keras3.posit.co/dev/reference/layer_multi_head_attention.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Multi Head Attention layer. — layer_multi_head_attention","text":"inputs see description num_heads Number attention heads. key_dim Size attention head query key. value_dim Size attention head value. dropout Dropout probability. use_bias Boolean, whether dense layers use bias vectors/matrices. output_shape expected shape output tensor, besides batch sequence dims. specified, projects back query feature dim (query input's last dimension). attention_axes axes attention applied. NULL means attention axes, batch, heads, features. flash_attention NULL, layer attempts use flash attention faster memory-efficient attention computations possible. behavior can configured using config_enable_flash_attention() config_disable_flash_attention(). kernel_initializer Initializer dense layer kernels. bias_initializer Initializer dense layer biases. kernel_regularizer Regularizer dense layer kernels. bias_regularizer Regularizer dense layer biases. activity_regularizer Regularizer dense layer activity. kernel_constraint Constraint dense layer kernels. bias_constraint Constraint dense layer kernels. seed Optional integer seed dropout layer. ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_multi_head_attention.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Multi Head Attention layer. — layer_multi_head_attention","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_multi_head_attention.html","id":"call-arguments","dir":"Reference","previous_headings":"","what":"Call Arguments","title":"Multi Head Attention layer. — layer_multi_head_attention","text":"query: Query tensor shape (B, T, dim), B batch size, T target sequence length, dim feature dimension. value: Value tensor shape (B, S, dim), B batch size, S source sequence length, dim feature dimension. key: Optional key tensor shape (B, S, dim). given, use value key value, common case. attention_mask: boolean mask shape (B, T, S), prevents attention certain positions. boolean mask specifies query elements can attend key elements, 1 indicates attention 0 indicates attention. Broadcasting can happen missing batch dimensions head dimension. return_attention_scores: boolean indicate whether output (attention_output, attention_scores) TRUE, attention_output FALSE. Defaults FALSE. training: Python boolean indicating whether layer behave training mode (adding dropout) inference mode (dropout). go either using training mode parent layer/model, FALSE (inference) parent layer. use_causal_mask: boolean indicate whether apply causal mask prevent tokens attending future tokens (e.g., used decoder Transformer).","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_multi_head_attention.html","id":"call-return","dir":"Reference","previous_headings":"","what":"Call return","title":"Multi Head Attention layer. — layer_multi_head_attention","text":"attention_output: result computation, shape (B, T, E), T target sequence shapes E query input last dimension output_shape NULL. Otherwise, multi-head outputs projected shape specified output_shape. attention_scores: (Optional) multi-head attention coefficients attention axes.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_multi_head_attention.html","id":"properties","dir":"Reference","previous_headings":"","what":"Properties","title":"Multi Head Attention layer. — layer_multi_head_attention","text":"MultiHeadAttention Layer instance following additional read-properties: attention_axes dropout key_dense key_dim num_heads output_dense output_shape query_dense use_bias value_dense value_dim","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_multiply.html","id":null,"dir":"Reference","previous_headings":"","what":"Performs elementwise multiplication. — layer_multiply","title":"Performs elementwise multiplication. — layer_multiply","text":"takes input list tensors, shape, returns single tensor (also shape).","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_multiply.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Performs elementwise multiplication. — layer_multiply","text":"","code":"layer_multiply(inputs, ...)"},{"path":"https://keras3.posit.co/dev/reference/layer_multiply.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Performs elementwise multiplication. — layer_multiply","text":"inputs layers combine ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_multiply.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Performs elementwise multiplication. — layer_multiply","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_multiply.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Performs elementwise multiplication. — layer_multiply","text":"Usage Keras model:","code":"input_shape <- c(2, 3, 4) x1 <- random_uniform(input_shape) x2 <- random_uniform(input_shape) y <- layer_multiply(x1, x2) input1 <- layer_input(shape = c(16)) x1 <- input1 |> layer_dense(8, activation = 'relu') input2 <- layer_input(shape = c(32)) x2 <- input2 |> layer_dense(8, activation = 'relu') # equivalent to `y <- layer_multiply(x1, x2)` y <- layer_multiply(x1, x2) out <- y |> layer_dense(4) model <- keras_model(inputs = c(input1, input2), outputs = out)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_normalization.html","id":null,"dir":"Reference","previous_headings":"","what":"A preprocessing layer that normalizes continuous features. — layer_normalization","title":"A preprocessing layer that normalizes continuous features. — layer_normalization","text":"layer shift scale inputs distribution centered around 0 standard deviation 1. accomplishes precomputing mean variance data, calling (input - mean) / sqrt(var) runtime. mean variance values layer must either supplied construction learned via adapt(). adapt() compute mean variance data store layer's weights. adapt() called fit(), evaluate(), predict().","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_normalization.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"A preprocessing layer that normalizes continuous features. — layer_normalization","text":"","code":"layer_normalization(   object,   axis = -1L,   mean = NULL,   variance = NULL,   invert = FALSE,   ... )"},{"path":"https://keras3.posit.co/dev/reference/layer_normalization.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"A preprocessing layer that normalizes continuous features. — layer_normalization","text":"object Object compose layer . tensor, array, sequential model. axis Integer, list integers, NULL. axis axes separate mean variance index shape. example, shape (NULL, 5) axis=1, layer track 5 separate mean variance values last axis. axis set NULL, layer normalize elements input scalar mean variance. -1, last axis input assumed feature dimension normalized per index. Note specific case batched scalar inputs axis batch axis, default normalize index batch separately. case, consider passing axis=NULL. Defaults -1. mean mean value(s) use normalization. passed value(s) broadcast shape kept axes ; value(s) broadcast, error raised layer's build() method called. variance variance value(s) use normalization. passed value(s) broadcast shape kept axes ; value(s) broadcast, error raised layer's build() method called. invert TRUE, layer apply inverse transformation inputs: turn normalized input back original form. ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_normalization.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"A preprocessing layer that normalizes continuous features. — layer_normalization","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_normalization.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"A preprocessing layer that normalizes continuous features. — layer_normalization","text":"Calculate global mean variance analyzing dataset adapt().     Calculate mean variance index last axis.     Pass mean variance directly.     Use layer de-normalize inputs (adapting layer).","code":"adapt_data <- op_array(c(1., 2., 3., 4., 5.), dtype='float32') input_data <- op_array(c(1., 2., 3.), dtype='float32') layer <- layer_normalization(axis = NULL) layer %>% adapt(adapt_data) layer(input_data) ## tf.Tensor([-1.4142135  -0.70710677  0.        ], shape=(3), dtype=float32) adapt_data <- op_array(rbind(c(0., 7., 4.),                        c(2., 9., 6.),                        c(0., 7., 4.),                        c(2., 9., 6.)), dtype='float32') input_data <- op_array(matrix(c(0., 7., 4.), nrow = 1), dtype='float32') layer <- layer_normalization(axis=-1) layer %>% adapt(adapt_data) layer(input_data) ## tf.Tensor([[-1. -1. -1.]], shape=(1, 3), dtype=float32) input_data <- op_array(rbind(1, 2, 3), dtype='float32') layer <- layer_normalization(mean=3., variance=2.) layer(input_data) ## tf.Tensor( ## [[-1.4142135 ] ##  [-0.70710677] ##  [ 0.        ]], shape=(3, 1), dtype=float32) adapt_data <- op_array(rbind(c(0., 7., 4.),                        c(2., 9., 6.),                        c(0., 7., 4.),                        c(2., 9., 6.)), dtype='float32') input_data <- op_array(c(1., 2., 3.), dtype='float32') layer <- layer_normalization(axis=-1, invert=TRUE) layer %>% adapt(adapt_data) layer(input_data) ## tf.Tensor([[ 2. 10.  8.]], shape=(1, 3), dtype=float32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_permute.html","id":null,"dir":"Reference","previous_headings":"","what":"Permutes the dimensions of the input according to a given pattern. — layer_permute","title":"Permutes the dimensions of the input according to a given pattern. — layer_permute","text":"Useful e.g. connecting RNNs convnets.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_permute.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Permutes the dimensions of the input according to a given pattern. — layer_permute","text":"","code":"layer_permute(object, dims, ...)"},{"path":"https://keras3.posit.co/dev/reference/layer_permute.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Permutes the dimensions of the input according to a given pattern. — layer_permute","text":"object Object compose layer . tensor, array, sequential model. dims List integers. Permutation pattern include batch dimension. Indexing starts 1. instance, (1, 3, 2) permutes second third dimensions input. ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_permute.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Permutes the dimensions of the input according to a given pattern. — layer_permute","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_permute.html","id":"input-shape","dir":"Reference","previous_headings":"","what":"Input Shape","title":"Permutes the dimensions of the input according to a given pattern. — layer_permute","text":"Arbitrary.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_permute.html","id":"output-shape","dir":"Reference","previous_headings":"","what":"Output Shape","title":"Permutes the dimensions of the input according to a given pattern. — layer_permute","text":"input shape, dimensions re-ordered according specified pattern.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_permute.html","id":"example","dir":"Reference","previous_headings":"","what":"Example","title":"Permutes the dimensions of the input according to a given pattern. — layer_permute","text":"","code":"x <- layer_input(shape=c(10, 64)) y <- layer_permute(x, c(2, 1)) shape(y) ## shape(NA, 64, 10)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_pipeline.html","id":null,"dir":"Reference","previous_headings":"","what":"Applies a series of layers to an input. — layer_pipeline","title":"Applies a series of layers to an input. — layer_pipeline","text":"class useful build preprocessing pipeline, particular image data augmentation pipeline. Compared Sequential model, Pipeline features important differences: Model, just plain layer. layers pipeline compatible tf.data, pipeline also remain tf.data compatible. say, pipeline attempt convert inputs backend-native tensors tf.data context (unlike Sequential model).","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_pipeline.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Applies a series of layers to an input. — layer_pipeline","text":"","code":"layer_pipeline(layers, name = NULL)"},{"path":"https://keras3.posit.co/dev/reference/layer_pipeline.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Applies a series of layers to an input. — layer_pipeline","text":"layers list layers. name String, name object","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_pipeline.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Applies a series of layers to an input. — layer_pipeline","text":"","code":"preprocessing_pipeline <- layer_pipeline(c(   layer_auto_contrast(, ),   layer_random_zoom(, 0.2),   layer_random_rotation(, 0.2) ))  # `ds` is a tf.data.Dataset of images ds <- tfdatasets::tensor_slices_dataset(1:100) |>   tfdatasets::dataset_map(\\(.x) {     random_normal(c(28, 28))   }) |>   tfdatasets::dataset_batch(32)   #|>   # tfdatasets::dataset_take(4) |>   # iterate() |> str()  preprocessed_ds <- ds |>   tfdatasets::dataset_map(preprocessing_pipeline, num_parallel_calls = 4)"},{"path":"https://keras3.posit.co/dev/reference/layer_rand_augment.html","id":null,"dir":"Reference","previous_headings":"","what":"RandAugment performs the Rand Augment operation on input images. — layer_rand_augment","title":"RandAugment performs the Rand Augment operation on input images. — layer_rand_augment","text":"layer can thought --one image augmentation layer. policy implemented layer benchmarked extensively effective wide variety datasets.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_rand_augment.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"RandAugment performs the Rand Augment operation on input images. — layer_rand_augment","text":"","code":"layer_rand_augment(   object,   value_range = list(0L, 255L),   num_ops = 2L,   factor = 0.5,   interpolation = \"bilinear\",   seed = NULL,   data_format = NULL,   ... )"},{"path":"https://keras3.posit.co/dev/reference/layer_rand_augment.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"RandAugment performs the Rand Augment operation on input images. — layer_rand_augment","text":"object Object compose layer . tensor, array, sequential model. value_range range values input image can take. Default (0, 255). Typically, (0, 1) normalized images (0, 255) raw images. num_ops number augmentation operations apply sequentially image. Default 2. factor strength augmentation normalized value 0 1. Default 0.5. interpolation interpolation method use resizing operations. Options include nearest, bilinear. Default bilinear. seed Integer. Used create random seed. data_format String, one \"channels_last\" (default) \"channels_first\". ordering dimensions inputs. \"channels_last\" corresponds inputs shape (batch, height, width, channels) \"channels_first\" corresponds inputs shape (batch, channels, height, width). ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_rand_augment.html","id":"references","dir":"Reference","previous_headings":"","what":"References","title":"RandAugment performs the Rand Augment operation on input images. — layer_rand_augment","text":"RandAugment","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_random_brightness.html","id":null,"dir":"Reference","previous_headings":"","what":"A preprocessing layer which randomly adjusts brightness during training. — layer_random_brightness","title":"A preprocessing layer which randomly adjusts brightness during training. — layer_random_brightness","text":"layer randomly increase/reduce brightness input RGB images. inference time, output identical input. Call layer training=TRUE adjust brightness input. Note: layer safe use inside tf.data pipeline (independently backend using).","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_random_brightness.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"A preprocessing layer which randomly adjusts brightness during training. — layer_random_brightness","text":"","code":"layer_random_brightness(   object,   factor,   value_range = list(0L, 255L),   seed = NULL,   ... )"},{"path":"https://keras3.posit.co/dev/reference/layer_random_brightness.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"A preprocessing layer which randomly adjusts brightness during training. — layer_random_brightness","text":"object Object compose layer . tensor, array, sequential model. factor Float list 2 floats -1.0 1.0. factor used determine lower bound upper bound brightness adjustment. float value chosen randomly limits. -1.0 chosen, output image black, 1.0 chosen, image fully white. one float provided, eg, 0.2, -0.2 used lower bound 0.2 used upper bound. value_range Optional list 2 floats lower upper limit values input data. make change, use c(0.0, 1.0), e.g., image input scaled layer. Defaults c(0.0, 255.0). brightness adjustment scaled range, output values clipped range. seed optional integer, fixed RNG behavior. ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_random_brightness.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"A preprocessing layer which randomly adjusts brightness during training. — layer_random_brightness","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_random_brightness.html","id":"inputs","dir":"Reference","previous_headings":"","what":"Inputs","title":"A preprocessing layer which randomly adjusts brightness during training. — layer_random_brightness","text":"3D (HWC) 4D (NHWC) tensor, float int dtype. Input pixel values can range (e.g. [0., 1.) [0, 255])","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_random_brightness.html","id":"output","dir":"Reference","previous_headings":"","what":"Output","title":"A preprocessing layer which randomly adjusts brightness during training. — layer_random_brightness","text":"3D (HWC) 4D (NHWC) tensor brightness adjusted based factor. default, layer output floats. output value clipped range [0, 255], valid range RGB colors, rescaled based value_range needed.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_random_brightness.html","id":"example","dir":"Reference","previous_headings":"","what":"Example","title":"A preprocessing layer which randomly adjusts brightness during training. — layer_random_brightness","text":"","code":"random_bright <- layer_random_brightness(factor=0.2, seed = 1)  # An image with shape [2, 2, 3] image <- array(1:12, dim=c(2, 2, 3))  # Assume we randomly select the factor to be 0.1, then it will apply # 0.1 * 255 to all the channel output <- random_bright(image, training=TRUE) output ## tf.Tensor( ## [[[39 43 47] ##   [41 45 49]] ## ##  [[40 44 48] ##   [42 46 50]]], shape=(2, 2, 3), dtype=int32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_random_color_degeneration.html","id":null,"dir":"Reference","previous_headings":"","what":"Randomly performs the color degeneration operation on given images. — layer_random_color_degeneration","title":"Randomly performs the color degeneration operation on given images. — layer_random_color_degeneration","text":"sharpness operation first converts image gray scale, back color. takes weighted average original image degenerated image. makes colors appear dull.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_random_color_degeneration.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Randomly performs the color degeneration operation on given images. — layer_random_color_degeneration","text":"","code":"layer_random_color_degeneration(   object,   factor,   value_range = list(0L, 255L),   data_format = NULL,   seed = NULL,   ... )"},{"path":"https://keras3.posit.co/dev/reference/layer_random_color_degeneration.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Randomly performs the color degeneration operation on given images. — layer_random_color_degeneration","text":"object Object compose layer . tensor, array, sequential model. factor tuple two floats single float. factor controls extent image sharpness impacted. factor=0.0 makes layer perform -op operation, value 1.0 uses degenerated result entirely. Values 0 1 result linear interpolation original image sharpened image. Values 0.0 1.0. tuple used, factor sampled two values every image augmented. single float used, value 0.0 passed float sampled. order ensure value always , please pass tuple two identical floats: (0.5, 0.5). value_range range values input image can take. Default (0, 255). Typically, (0, 1) normalized images (0, 255) raw images. data_format String, one \"channels_last\" (default) \"channels_first\". ordering dimensions inputs. \"channels_last\" corresponds inputs shape (batch, height, width, channels) \"channels_first\" corresponds inputs shape (batch, channels, height, width). seed Integer. Used create random seed. ... forward/backward compatability.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_random_color_jitter.html","id":null,"dir":"Reference","previous_headings":"","what":"Randomly apply brightness, contrast, saturation — layer_random_color_jitter","title":"Randomly apply brightness, contrast, saturation — layer_random_color_jitter","text":"hue image processing operation sequentially randomly input.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_random_color_jitter.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Randomly apply brightness, contrast, saturation — layer_random_color_jitter","text":"","code":"layer_random_color_jitter(   object,   value_range = list(0L, 255L),   brightness_factor = NULL,   contrast_factor = NULL,   saturation_factor = NULL,   hue_factor = NULL,   seed = NULL,   data_format = NULL,   ... )"},{"path":"https://keras3.posit.co/dev/reference/layer_random_color_jitter.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Randomly apply brightness, contrast, saturation — layer_random_color_jitter","text":"object Object compose layer . tensor, array, sequential model. value_range range values incoming images . Represented two number tuple written [low, high]. typically either [0, 1] [0, 255] depending preprocessing pipeline set . brightness_factor Float list/tuple 2 floats -1.0 1.0. factor used determine lower bound upper bound brightness adjustment. float value chosen randomly limits. -1.0 chosen, output image black, 1.0 chosen, image fully white. one float provided, eg, 0.2, -0.2 used lower bound 0.2 used upper bound. contrast_factor positive float represented fraction value, tuple size 2 representing lower upper bound. represented single float, lower = upper. contrast factor randomly picked [1.0 - lower, 1.0 + upper]. pixel x channel, output (x - mean) * factor + mean mean mean value channel. saturation_factor tuple two floats single float. factor controls extent image saturation impacted. factor=0.5 makes layer perform -op operation. factor=0.0 makes image fully grayscale. factor=1.0 makes image fully saturated. Values 0.0 1.0. tuple used, factor sampled two values every image augmented. single float used, value 0.0 passed float sampled. ensure value always , pass tuple two identical floats: (0.5, 0.5). hue_factor single float tuple two floats. factor controls extent image hue impacted. factor=0.0 makes layer perform -op operation, value 1.0 performs aggressive contrast adjustment available. tuple used, factor sampled two values every image augmented. single float used, value 0.0 passed float sampled. order ensure value always , please pass tuple two identical floats: (0.5, 0.5). seed Integer. Used create random seed. data_format String, one \"channels_last\" (default) \"channels_first\". ordering dimensions inputs. \"channels_last\" corresponds inputs shape (batch, height, width, channels) \"channels_first\" corresponds inputs shape (batch, channels, height, width). ... forward/backward compatability.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_random_contrast.html","id":null,"dir":"Reference","previous_headings":"","what":"A preprocessing layer which randomly adjusts contrast during training. — layer_random_contrast","title":"A preprocessing layer which randomly adjusts contrast during training. — layer_random_contrast","text":"layer randomly adjust contrast image images random factor. Contrast adjusted independently channel image training. channel, layer computes mean image pixels channel adjusts component x pixel (x - mean) * contrast_factor + mean. Input pixel values can range (e.g. [0., 1.) [0, 255]) integer floating point dtype. default, layer output floats. Note: layer safe use inside tf.data pipeline (independently backend using).","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_random_contrast.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"A preprocessing layer which randomly adjusts contrast during training. — layer_random_contrast","text":"","code":"layer_random_contrast(   object,   factor,   value_range = c(0L, 255L),   seed = NULL,   ... )"},{"path":"https://keras3.posit.co/dev/reference/layer_random_contrast.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"A preprocessing layer which randomly adjusts contrast during training. — layer_random_contrast","text":"object Object compose layer . tensor, array, sequential model. factor positive float represented fraction value, tuple size 2 representing lower upper bound. represented single float, lower = upper. contrast factor randomly picked [1.0 - lower, 1.0 + upper]. pixel x channel, output (x - mean) * factor + mean mean mean value channel. value_range range values incoming images . Represented two-number tuple written tuple(low, high). typically either [0, 1] [0, 255] depending preprocessing pipeline set . seed Integer. Used create random seed. ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_random_contrast.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"A preprocessing layer which randomly adjusts contrast during training. — layer_random_contrast","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_random_contrast.html","id":"input-shape","dir":"Reference","previous_headings":"","what":"Input Shape","title":"A preprocessing layer which randomly adjusts contrast during training. — layer_random_contrast","text":"3D (unbatched) 4D (batched) tensor shape: (..., height, width, channels), \"channels_last\" format.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_random_contrast.html","id":"output-shape","dir":"Reference","previous_headings":"","what":"Output Shape","title":"A preprocessing layer which randomly adjusts contrast during training. — layer_random_contrast","text":"3D (unbatched) 4D (batched) tensor shape: (..., height, width, channels), \"channels_last\" format.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_random_crop.html","id":null,"dir":"Reference","previous_headings":"","what":"A preprocessing layer which randomly crops images during training. — layer_random_crop","title":"A preprocessing layer which randomly crops images during training. — layer_random_crop","text":"training, layer randomly choose location crop images target size. layer crop images batch cropping location. inference time, training input image smaller target size, input resized cropped return largest possible window image matches target aspect ratio. need apply random cropping inference time, set training TRUE calling layer. Input pixel values can range (e.g. [0., 1.) [0, 255]) integer floating point dtype. default, layer output floats. Note: layer safe use inside tf.data pipeline (independently backend using).","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_random_crop.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"A preprocessing layer which randomly crops images during training. — layer_random_crop","text":"","code":"layer_random_crop(   object,   height,   width,   seed = NULL,   data_format = NULL,   name = NULL,   ... )"},{"path":"https://keras3.posit.co/dev/reference/layer_random_crop.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"A preprocessing layer which randomly crops images during training. — layer_random_crop","text":"object Object compose layer . tensor, array, sequential model. height Integer, height output shape. width Integer, width output shape. seed Integer. Used create random seed. data_format string, either \"channels_last\" \"channels_first\". ordering dimensions inputs. \"channels_last\" corresponds inputs shape (batch, height, width, channels) \"channels_first\" corresponds inputs shape (batch, channels, height, width). defaults image_data_format value found Keras config file ~/.keras/keras.json. never set , \"channels_last\". name String, name object ... Base layer keyword arguments, name dtype.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_random_crop.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"A preprocessing layer which randomly crops images during training. — layer_random_crop","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_random_crop.html","id":"input-shape","dir":"Reference","previous_headings":"","what":"Input Shape","title":"A preprocessing layer which randomly crops images during training. — layer_random_crop","text":"3D (unbatched) 4D (batched) tensor shape: (..., height, width, channels), \"channels_last\" format.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_random_crop.html","id":"output-shape","dir":"Reference","previous_headings":"","what":"Output Shape","title":"A preprocessing layer which randomly crops images during training. — layer_random_crop","text":"3D (unbatched) 4D (batched) tensor shape: (..., target_height, target_width, channels).","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_random_erasing.html","id":null,"dir":"Reference","previous_headings":"","what":"Random Erasing data augmentation technique. — layer_random_erasing","title":"Random Erasing data augmentation technique. — layer_random_erasing","text":"Random Erasing data augmentation method random patches image erased (replaced constant value noise) training improve generalization.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_random_erasing.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Random Erasing data augmentation technique. — layer_random_erasing","text":"","code":"layer_random_erasing(   object,   factor = 1,   scale = list(0.02, 0.33),   fill_value = NULL,   value_range = list(0L, 255L),   seed = NULL,   data_format = NULL,   ... )"},{"path":"https://keras3.posit.co/dev/reference/layer_random_erasing.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Random Erasing data augmentation technique. — layer_random_erasing","text":"object Object compose layer . tensor, array, sequential model. factor single float tuple two floats. factor controls probability applying transformation. factor=0.0 ensures erasing applied. factor=1.0 means erasing always applied. tuple (min, max) provided, probability value sampled min max image. single float provided, probability sampled 0.0 given float. Default 1.0. scale tuple two floats representing aspect ratio range erased patch. defines width--height ratio patch erased. can help control rw shape erased region. Default (0.02, 0.33). fill_value value fill erased region . can set constant value None sample random value normal distribution. Default None. value_range range values incoming images . Represented two-number tuple written [low, high]. typically either [0, 1] [0, 255] depending preprocessing pipeline set . seed Integer. Used create random seed. data_format string, either \"channels_last\" \"channels_first\". ordering dimensions inputs. \"channels_last\" corresponds inputs shape (batch, height, width, channels) \"channels_first\" corresponds inputs shape (batch, channels, height, width). defaults image_data_format value found Keras config file ~/.keras/keras.json. never set , \"channels_last\". ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_random_erasing.html","id":"references","dir":"Reference","previous_headings":"","what":"References","title":"Random Erasing data augmentation technique. — layer_random_erasing","text":"Random Erasing paper.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_random_flip.html","id":null,"dir":"Reference","previous_headings":"","what":"A preprocessing layer which randomly flips images during training. — layer_random_flip","title":"A preprocessing layer which randomly flips images during training. — layer_random_flip","text":"layer flip images horizontally vertically based mode attribute. inference time, output identical input. Call layer training=TRUE flip input. Input pixel values can range (e.g. [0., 1.) [0, 255]) integer floating point dtype. default, layer output floats. Note: layer safe use inside tf.data pipeline (independently backend using).","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_random_flip.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"A preprocessing layer which randomly flips images during training. — layer_random_flip","text":"","code":"layer_random_flip(   object,   mode = \"horizontal_and_vertical\",   seed = NULL,   data_format = NULL,   ... )"},{"path":"https://keras3.posit.co/dev/reference/layer_random_flip.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"A preprocessing layer which randomly flips images during training. — layer_random_flip","text":"object Object compose layer . tensor, array, sequential model. mode String indicating flip mode use. Can \"horizontal\", \"vertical\", \"horizontal_and_vertical\". \"horizontal\" left-right flip \"vertical\" top-bottom flip. Defaults \"horizontal_and_vertical\" seed Integer. Used create random seed. data_format string, either \"channels_last\" \"channels_first\". ordering dimensions inputs. \"channels_last\" corresponds inputs shape (batch, height, width, channels) \"channels_first\" corresponds inputs shape (batch, channels, height, width). defaults image_data_format value found Keras config file ~/.keras/keras.json. never set , \"channels_last\". ... Base layer keyword arguments, name dtype.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_random_flip.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"A preprocessing layer which randomly flips images during training. — layer_random_flip","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_random_flip.html","id":"input-shape","dir":"Reference","previous_headings":"","what":"Input Shape","title":"A preprocessing layer which randomly flips images during training. — layer_random_flip","text":"3D (unbatched) 4D (batched) tensor shape: (..., height, width, channels), \"channels_last\" format.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_random_flip.html","id":"output-shape","dir":"Reference","previous_headings":"","what":"Output Shape","title":"A preprocessing layer which randomly flips images during training. — layer_random_flip","text":"3D (unbatched) 4D (batched) tensor shape: (..., height, width, channels), \"channels_last\" format.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_random_gaussian_blur.html","id":null,"dir":"Reference","previous_headings":"","what":"Applies random Gaussian blur to images for data augmentation. — layer_random_gaussian_blur","title":"Applies random Gaussian blur to images for data augmentation. — layer_random_gaussian_blur","text":"layer performs Gaussian blur operation input images randomly selected degree blurring, controlled factor sigma arguments.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_random_gaussian_blur.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Applies random Gaussian blur to images for data augmentation. — layer_random_gaussian_blur","text":"","code":"layer_random_gaussian_blur(   object,   factor = 1,   kernel_size = 3L,   sigma = 1,   value_range = list(0L, 255L),   data_format = NULL,   seed = NULL,   ... )"},{"path":"https://keras3.posit.co/dev/reference/layer_random_gaussian_blur.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Applies random Gaussian blur to images for data augmentation. — layer_random_gaussian_blur","text":"object Object compose layer . tensor, array, sequential model. factor single float tuple two floats. factor controls extent image hue impacted. factor=0.0 makes layer perform -op operation, value 1.0 performs aggressive blurring available. tuple used, factor sampled two values every image augmented. single float used, value 0.0 passed float sampled. Default 1.0. kernel_size Integer. Size Gaussian kernel used blurring. Must odd integer. Default 3. sigma Float tuple two floats. Standard deviation Gaussian kernel. Controls intensity blur. tuple provided, value sampled two image. Default 1.0. value_range range values incoming images . Represented two-number tuple written [low, high]. typically either [0, 1] [0, 255] depending preprocessing pipeline set . data_format string, either \"channels_last\" \"channels_first\". ordering dimensions inputs. \"channels_last\" corresponds inputs shape (batch, height, width, channels) \"channels_first\" corresponds inputs shape (batch, channels, height, width). defaults image_data_format value found Keras config file ~/.keras/keras.json. never set , \"channels_last\". seed Integer. Used create random seed. ... forward/backward compatability.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_random_grayscale.html","id":null,"dir":"Reference","previous_headings":"","what":"Preprocessing layer for random conversion of RGB images to grayscale. — layer_random_grayscale","title":"Preprocessing layer for random conversion of RGB images to grayscale. — layer_random_grayscale","text":"layer randomly converts input images grayscale specified factor. applied, maintains original number channels sets channels grayscale value. can useful data augmentation training models robust color variations. conversion preserves perceived luminance original color image using standard RGB grayscale conversion coefficients. Images selected conversion remain unchanged. Note: layer safe use inside tf.data pipeline (independently backend using).","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_random_grayscale.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Preprocessing layer for random conversion of RGB images to grayscale. — layer_random_grayscale","text":"","code":"layer_random_grayscale(   object,   factor = 0.5,   data_format = NULL,   seed = NULL,   ... )"},{"path":"https://keras3.posit.co/dev/reference/layer_random_grayscale.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Preprocessing layer for random conversion of RGB images to grayscale. — layer_random_grayscale","text":"object Object compose layer . tensor, array, sequential model. factor Float 0 1, specifying factor converting image grayscale. Defaults 0.5. value 1.0 means images converted, 0.0 means images converted. data_format String, one \"channels_last\" (default) \"channels_first\". ordering dimensions inputs. \"channels_last\" corresponds inputs shape (batch, height, width, channels) \"channels_first\" corresponds inputs shape (batch, channels, height, width). seed Initial seed random number generator ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_random_grayscale.html","id":"input-shape","dir":"Reference","previous_headings":"","what":"Input Shape","title":"Preprocessing layer for random conversion of RGB images to grayscale. — layer_random_grayscale","text":"3D (unbatched) 4D (batched) tensor shape: (..., height, width, channels), \"channels_last\" format, (..., channels, height, width), \"channels_first\" format.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_random_grayscale.html","id":"output-shape","dir":"Reference","previous_headings":"","what":"Output Shape","title":"Preprocessing layer for random conversion of RGB images to grayscale. — layer_random_grayscale","text":"input shape. output maintains number channels input, even grayscale-converted images channels value.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_random_hue.html","id":null,"dir":"Reference","previous_headings":"","what":"Randomly adjusts the hue on given images. — layer_random_hue","title":"Randomly adjusts the hue on given images. — layer_random_hue","text":"layer randomly increase/reduce hue input RGB images. image hue adjusted converting image(s) HSV rotating hue channel (H) delta. image converted back RGB.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_random_hue.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Randomly adjusts the hue on given images. — layer_random_hue","text":"","code":"layer_random_hue(   object,   factor,   value_range = list(0L, 255L),   data_format = NULL,   seed = NULL,   ... )"},{"path":"https://keras3.posit.co/dev/reference/layer_random_hue.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Randomly adjusts the hue on given images. — layer_random_hue","text":"object Object compose layer . tensor, array, sequential model. factor single float tuple two floats. factor controls extent image hue impacted. factor=0.0 makes layer perform -op operation, value 1.0 performs aggressive contrast adjustment available. tuple used, factor sampled two values every image augmented. single float used, value 0.0 passed float sampled. order ensure value always , please pass tuple two identical floats: (0.5, 0.5). value_range range values incoming images . Represented two-number tuple written [low, high]. typically either [0, 1] [0, 255] depending preprocessing pipeline set . data_format String, one \"channels_last\" (default) \"channels_first\". ordering dimensions inputs. \"channels_last\" corresponds inputs shape (batch, height, width, channels) \"channels_first\" corresponds inputs shape (batch, channels, height, width). seed Integer. Used create random seed. ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_random_hue.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Randomly adjusts the hue on given images. — layer_random_hue","text":"","code":"c(c(images, labels), .) %<-% dataset_cifar10() random_hue <- layer_random_hue(factor=0.5, value_range=c(0, 1)) images <- op_cast(images[1:8,,,], \"float32\") augmented_images_batch = random_hue(images)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_random_invert.html","id":null,"dir":"Reference","previous_headings":"","what":"Preprocessing layer for random inversion of image colors. — layer_random_invert","title":"Preprocessing layer for random inversion of image colors. — layer_random_invert","text":"layer randomly inverts colors input images specified probability range. applied, image chance colors inverted, pixel values transformed complementary values. Images selected inversion remain unchanged.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_random_invert.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Preprocessing layer for random inversion of image colors. — layer_random_invert","text":"","code":"layer_random_invert(   object,   factor = 1,   value_range = list(0L, 255L),   seed = NULL,   data_format = NULL,   ... )"},{"path":"https://keras3.posit.co/dev/reference/layer_random_invert.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Preprocessing layer for random inversion of image colors. — layer_random_invert","text":"object Object compose layer . tensor, array, sequential model. factor single float tuple two floats. factor controls probability inverting image colors. tuple provided, value sampled two values image, factor[0] minimum factor[1] maximum probability. single float provided, value 0.0 provided float sampled. Defaults (0, 1). value_range tuple list two elements. first value represents lower bound values passed images, second represents upper bound. Images passed layer values within value_range. Defaults (0, 255). seed Integer. Used create random seed. data_format string, either \"channels_last\" \"channels_first\". ordering dimensions inputs. \"channels_last\" corresponds inputs shape (batch, height, width, channels) \"channels_first\" corresponds inputs shape (batch, channels, height, width). defaults image_data_format value found Keras config file ~/.keras/keras.json. never set , \"channels_last\". ... forward/backward compatability.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_random_perspective.html","id":null,"dir":"Reference","previous_headings":"","what":"A preprocessing layer that applies random perspective transformations. — layer_random_perspective","title":"A preprocessing layer that applies random perspective transformations. — layer_random_perspective","text":"layer distorts perspective input images shifting corner points, simulating 3D-like transformation. amount distortion controlled factor scale parameters.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_random_perspective.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"A preprocessing layer that applies random perspective transformations. — layer_random_perspective","text":"","code":"layer_random_perspective(   object,   factor = 1,   scale = 1,   interpolation = \"bilinear\",   fill_value = 0,   seed = NULL,   data_format = NULL,   ... )"},{"path":"https://keras3.posit.co/dev/reference/layer_random_perspective.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"A preprocessing layer that applies random perspective transformations. — layer_random_perspective","text":"object Object compose layer . tensor, array, sequential model. factor float tuple two floats. Represents probability applying perspective transformation image batch. factor=0.0 ensures transformation applied. factor=1.0 means transformation always applied. tuple (min, max) provided, probability randomly sampled min max image. single float given, probability sampled 0.0 provided float. Default 1.0. scale float defining relative amount perspective shift. Determines much image corners displaced, affecting intensity perspective effect. interpolation Interpolation mode. Supported values: \"nearest\", \"bilinear\". fill_value float represents value filled outside boundaries fill_mode=\"constant\". seed Integer. Used create random seed. data_format string, either \"channels_last\" \"channels_first\". ordering dimensions inputs. \"channels_last\" corresponds inputs shape (batch, height, width, channels) \"channels_first\" corresponds inputs shape (batch, channels, height, width). defaults image_data_format value found Keras config file ~/.keras/keras.json. never set , \"channels_last\". ... forward/backward compatability.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_random_posterization.html","id":null,"dir":"Reference","previous_headings":"","what":"Reduces the number of bits for each color channel. — layer_random_posterization","title":"Reduces the number of bits for each color channel. — layer_random_posterization","text":"Reduces number bits color channel.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_random_posterization.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Reduces the number of bits for each color channel. — layer_random_posterization","text":"","code":"layer_random_posterization(   object,   factor,   value_range = list(0L, 255L),   data_format = NULL,   seed = NULL,   ... )"},{"path":"https://keras3.posit.co/dev/reference/layer_random_posterization.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Reduces the number of bits for each color channel. — layer_random_posterization","text":"object Object compose layer . tensor, array, sequential model. factor integer, number bits keep channel. Must value 1-8. value_range tuple list two elements. first value represents lower bound values passed images, second represents upper bound. Images passed layer values within value_range. Defaults (0, 255). data_format String, one \"channels_last\" (default) \"channels_first\". ordering dimensions inputs. \"channels_last\" corresponds inputs shape (batch, height, width, channels) \"channels_first\" corresponds inputs shape (batch, channels, height, width). seed Initial seed random number generator ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_random_posterization.html","id":"references","dir":"Reference","previous_headings":"","what":"References","title":"Reduces the number of bits for each color channel. — layer_random_posterization","text":"AutoAugment: Learning Augmentation Policies Data RandAugment: Practical automated data augmentation reduced search space","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_random_rotation.html","id":null,"dir":"Reference","previous_headings":"","what":"A preprocessing layer which randomly rotates images during training. — layer_random_rotation","title":"A preprocessing layer which randomly rotates images during training. — layer_random_rotation","text":"layer apply random rotations image, filling empty space according fill_mode. default, random rotations applied training. inference time, layer nothing. need apply random rotations inference time, pass training = TRUE calling layer. Input pixel values can range (e.g. [0., 1.) [0, 255]) integer floating point dtype. default, layer output floats. Note: layer safe use inside tf.data pipeline (independently backend using).","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_random_rotation.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"A preprocessing layer which randomly rotates images during training. — layer_random_rotation","text":"","code":"layer_random_rotation(   object,   factor,   fill_mode = \"reflect\",   interpolation = \"bilinear\",   seed = NULL,   fill_value = 0,   data_format = NULL,   ... )"},{"path":"https://keras3.posit.co/dev/reference/layer_random_rotation.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"A preprocessing layer which randomly rotates images during training. — layer_random_rotation","text":"object Object compose layer . tensor, array, sequential model. factor float represented fraction 2 Pi, tuple size 2 representing lower upper bound rotating clockwise counter-clockwise. positive values means rotating counter clock-wise, negative value means clock-wise. represented single float, value used upper lower bound. instance, factor=c(-0.2, 0.3) results output rotation random amount range [-20% * 360, 30% * 360]. factor=0.2 results output rotating random amount range [-20% * 360, 20% * 360]. fill_mode Points outside boundaries input filled according given mode (one {\"constant\", \"reflect\", \"wrap\", \"nearest\"}). reflect: (d c b | b c d | d c b ) input extended reflecting edge last pixel. constant: (k k k k | b c d | k k k k) input extended filling values beyond edge constant value k = 0. wrap: (b c d | b c d | b c d) input extended wrapping around opposite edge. nearest: (| b c d | d d d d) input extended nearest pixel. interpolation Interpolation mode. Supported values: \"nearest\", \"bilinear\". seed Integer. Used create random seed. fill_value float represents value filled outside boundaries fill_mode=\"constant\". data_format string, either \"channels_last\" \"channels_first\". ordering dimensions inputs. \"channels_last\" corresponds inputs shape (batch, height, width, channels) \"channels_first\" corresponds inputs shape (batch, channels, height, width). defaults image_data_format value found Keras config file ~/.keras/keras.json. never set , \"channels_last\". ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_random_rotation.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"A preprocessing layer which randomly rotates images during training. — layer_random_rotation","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_random_rotation.html","id":"input-shape","dir":"Reference","previous_headings":"","what":"Input Shape","title":"A preprocessing layer which randomly rotates images during training. — layer_random_rotation","text":"3D (unbatched) 4D (batched) tensor shape: (..., height, width, channels), \"channels_last\" format","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_random_rotation.html","id":"output-shape","dir":"Reference","previous_headings":"","what":"Output Shape","title":"A preprocessing layer which randomly rotates images during training. — layer_random_rotation","text":"3D (unbatched) 4D (batched) tensor shape: (..., height, width, channels), \"channels_last\" format","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_random_saturation.html","id":null,"dir":"Reference","previous_headings":"","what":"Randomly adjusts the saturation on given images. — layer_random_saturation","title":"Randomly adjusts the saturation on given images. — layer_random_saturation","text":"layer randomly increase/reduce saturation input RGB images.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_random_saturation.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Randomly adjusts the saturation on given images. — layer_random_saturation","text":"","code":"layer_random_saturation(   object,   factor,   value_range = list(0L, 255L),   data_format = NULL,   seed = NULL,   ... )"},{"path":"https://keras3.posit.co/dev/reference/layer_random_saturation.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Randomly adjusts the saturation on given images. — layer_random_saturation","text":"object Object compose layer . tensor, array, sequential model. factor tuple two floats single float. factor controls extent image saturation impacted. factor=0.5 makes layer perform -op operation. factor=0.0 makes image fully grayscale. factor=1.0 makes image fully saturated. Values 0.0 1.0. tuple used, factor sampled two values every image augmented. single float used, value 0.0 passed float sampled. ensure value always , pass tuple two identical floats: (0.5, 0.5). value_range range values incoming images . Represented two-number tuple written [low, high]. typically either [0, 1] [0, 255] depending preprocessing pipeline set . data_format String, one \"channels_last\" (default) \"channels_first\". ordering dimensions inputs. \"channels_last\" corresponds inputs shape (batch, height, width, channels) \"channels_first\" corresponds inputs shape (batch, channels, height, width). seed Integer. Used create random seed. ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_random_saturation.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Randomly adjusts the saturation on given images. — layer_random_saturation","text":"","code":"c(c(images, labels), .) %<-% dataset_cifar10() images <- images[1:8, , , ] |> op_cast(\"float32\") random_saturation <- layer_random_saturation(factor = 0.2) augmented_images <- random_saturation(images)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_random_sharpness.html","id":null,"dir":"Reference","previous_headings":"","what":"Randomly performs the sharpness operation on given images. — layer_random_sharpness","title":"Randomly performs the sharpness operation on given images. — layer_random_sharpness","text":"sharpness operation first performs blur, blends original image processed image. operation adjusts clarity edges image, ranging blurred enhanced sharpness.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_random_sharpness.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Randomly performs the sharpness operation on given images. — layer_random_sharpness","text":"","code":"layer_random_sharpness(   object,   factor,   value_range = list(0L, 255L),   data_format = NULL,   seed = NULL,   ... )"},{"path":"https://keras3.posit.co/dev/reference/layer_random_sharpness.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Randomly performs the sharpness operation on given images. — layer_random_sharpness","text":"object Object compose layer . tensor, array, sequential model. factor tuple two floats single float. factor controls extent image sharpness impacted. factor=0.0 results fully blurred image, factor=0.5 applies operation (preserving original image), factor=1.0 enhances sharpness beyond original. Values 0.0 1.0. tuple used, factor sampled two values every image augmented. single float used, value 0.0 passed float sampled. ensure value always , pass tuple two identical floats: (0.5, 0.5). value_range range values incoming images . Represented two-number tuple written [low, high]. typically either [0, 1] [0, 255] depending preprocessing pipeline set . data_format String, one \"channels_last\" (default) \"channels_first\". ordering dimensions inputs. \"channels_last\" corresponds inputs shape (batch, height, width, channels) \"channels_first\" corresponds inputs shape (batch, channels, height, width). seed Integer. Used create random seed. ... forward/backward compatability.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_random_shear.html","id":null,"dir":"Reference","previous_headings":"","what":"A preprocessing layer that randomly applies shear transformations — layer_random_shear","title":"A preprocessing layer that randomly applies shear transformations — layer_random_shear","text":"images. layer shears input images along x-axis /y-axis randomly selected factor within specified range. shear transformation applied image independently batch. Empty regions created transformation filled according fill_mode fill_value parameters.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_random_shear.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"A preprocessing layer that randomly applies shear transformations — layer_random_shear","text":"","code":"layer_random_shear(   object,   x_factor = 0,   y_factor = 0,   interpolation = \"bilinear\",   fill_mode = \"reflect\",   fill_value = 0,   data_format = NULL,   seed = NULL,   ... )"},{"path":"https://keras3.posit.co/dev/reference/layer_random_shear.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"A preprocessing layer that randomly applies shear transformations — layer_random_shear","text":"object Object compose layer . tensor, array, sequential model. x_factor tuple two floats. augmented image, value sampled provided range. float passed, range interpreted (0, x_factor). Values represent percentage image shear . example, 0.3 shears pixels 30% way across image. provided values positive. y_factor tuple two floats. augmented image, value sampled provided range. float passed, range interpreted (0, y_factor). Values represent percentage image shear . example, 0.3 shears pixels 30% way across image. provided values positive. interpolation Interpolation mode. Supported values: \"nearest\", \"bilinear\". fill_mode Points outside boundaries input filled according given mode. Available methods \"constant\", \"nearest\", \"wrap\" \"reflect\". Defaults \"reflect\". \"reflect\": (d c b | b c d | d c b ) input extended reflecting edge last pixel. \"constant\": (k k k k | b c d | k k k k) input extended filling values beyond edge constant value k specified fill_value. \"wrap\": (b c d | b c d | b c d) input extended wrapping around opposite edge. \"nearest\": (| b c d | d d d d) input extended nearest pixel. Note using torch backend, \"reflect\" redirected \"mirror\" (c d c b | b c d | c b b) torch support \"reflect\". Note torch backend support \"wrap\". fill_value float representing value filled outside boundaries fill_mode=\"constant\". data_format String, one \"channels_last\" (default) \"channels_first\". ordering dimensions inputs. \"channels_last\" corresponds inputs shape (batch, height, width, channels) \"channels_first\" corresponds inputs shape (batch, channels, height, width). seed Integer. Used create random seed. ... forward/backward compatability.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_random_translation.html","id":null,"dir":"Reference","previous_headings":"","what":"A preprocessing layer which randomly translates images during training. — layer_random_translation","title":"A preprocessing layer which randomly translates images during training. — layer_random_translation","text":"layer apply random translations image training, filling empty space according fill_mode. Input pixel values can range (e.g. [0., 1.) [0, 255]) integer floating point dtype. default, layer output floats.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_random_translation.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"A preprocessing layer which randomly translates images during training. — layer_random_translation","text":"","code":"layer_random_translation(   object,   height_factor,   width_factor,   fill_mode = \"reflect\",   interpolation = \"bilinear\",   seed = NULL,   fill_value = 0,   data_format = NULL,   ... )"},{"path":"https://keras3.posit.co/dev/reference/layer_random_translation.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"A preprocessing layer which randomly translates images during training. — layer_random_translation","text":"object Object compose layer . tensor, array, sequential model. height_factor float represented fraction value, tuple size 2 representing lower upper bound shifting vertically. negative value means shifting image , positive value means shifting image . represented single positive float, value used upper lower bound. instance, height_factor=(-0.2, 0.3) results output shifted random amount range [-20%, +30%]. height_factor=0.2 results output height shifted random amount range [-20%, +20%]. width_factor float represented fraction value, tuple size 2 representing lower upper bound shifting horizontally. negative value means shifting image left, positive value means shifting image right. represented single positive float, value used upper lower bound. instance, width_factor=(-0.2, 0.3) results output shifted left 20%, shifted right 30%. width_factor=0.2 results output height shifted left right 20%. fill_mode Points outside boundaries input filled according given mode. Available methods \"constant\", \"nearest\", \"wrap\" \"reflect\". Defaults \"reflect\". \"reflect\": (d c b | b c d | d c b ) input extended reflecting edge last pixel. \"constant\": (k k k k | b c d | k k k k) input extended filling values beyond edge constant value k specified fill_value. \"wrap\": (b c d | b c d | b c d) input extended wrapping around opposite edge. \"nearest\": (| b c d | d d d d) input extended nearest pixel. Note using torch backend, \"reflect\" redirected \"mirror\" (c d c b | b c d | c b b) torch support \"reflect\". Note torch backend support \"wrap\". interpolation Interpolation mode. Supported values: \"nearest\", \"bilinear\". seed Integer. Used create random seed. fill_value float represents value filled outside boundaries fill_mode=\"constant\". data_format string, either \"channels_last\" \"channels_first\". ordering dimensions inputs. \"channels_last\" corresponds inputs shape (batch, height, width, channels) \"channels_first\" corresponds inputs shape (batch, channels, height, width). defaults image_data_format value found Keras config file ~/.keras/keras.json. never set , \"channels_last\". ... Base layer keyword arguments, name dtype.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_random_translation.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"A preprocessing layer which randomly translates images during training. — layer_random_translation","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_random_translation.html","id":"input-shape","dir":"Reference","previous_headings":"","what":"Input Shape","title":"A preprocessing layer which randomly translates images during training. — layer_random_translation","text":"3D (unbatched) 4D (batched) tensor shape: (..., height, width, channels), \"channels_last\" format, (..., channels, height, width), \"channels_first\" format.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_random_translation.html","id":"output-shape","dir":"Reference","previous_headings":"","what":"Output Shape","title":"A preprocessing layer which randomly translates images during training. — layer_random_translation","text":"3D (unbatched) 4D (batched) tensor shape: (..., target_height, target_width, channels), (..., channels, target_height, target_width), \"channels_first\" format. Note: layer safe use inside tf.data pipeline (independently backend using).","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_random_zoom.html","id":null,"dir":"Reference","previous_headings":"","what":"A preprocessing layer which randomly zooms images during training. — layer_random_zoom","title":"A preprocessing layer which randomly zooms images during training. — layer_random_zoom","text":"layer randomly zoom axis image independently, filling empty space according fill_mode. Input pixel values can range (e.g. [0., 1.) [0, 255]) integer floating point dtype. default, layer output floats.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_random_zoom.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"A preprocessing layer which randomly zooms images during training. — layer_random_zoom","text":"","code":"layer_random_zoom(   object,   height_factor,   width_factor = NULL,   fill_mode = \"reflect\",   interpolation = \"bilinear\",   seed = NULL,   fill_value = 0,   data_format = NULL,   ... )"},{"path":"https://keras3.posit.co/dev/reference/layer_random_zoom.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"A preprocessing layer which randomly zooms images during training. — layer_random_zoom","text":"object Object compose layer . tensor, array, sequential model. height_factor float represented fraction value, list size 2 representing lower upper bound zooming vertically. represented single float, value used upper lower bound. positive value means zooming , negative value means zooming . instance, height_factor=c(0.2, 0.3) result output zoomed random amount range [+20%, +30%]. height_factor=c(-0.3, -0.2) result output zoomed random amount range [+20%, +30%]. width_factor float represented fraction value, list size 2 representing lower upper bound zooming horizontally. represented single float, value used upper lower bound. instance, width_factor=c(0.2, 0.3) result output zooming 20% 30%. width_factor=c(-0.3, -0.2) result output zooming 20% 30%. NULL means .e., zooming vertical horizontal directions preserving aspect ratio. Defaults NULL. fill_mode Points outside boundaries input filled according given mode. Available methods \"constant\", \"nearest\", \"wrap\" \"reflect\". Defaults \"reflect\". \"reflect\": (d c b | b c d | d c b ) input extended reflecting edge last pixel. \"constant\": (k k k k | b c d | k k k k) input extended filling values beyond edge constant value k specified fill_value. \"wrap\": (b c d | b c d | b c d) input extended wrapping around opposite edge. \"nearest\": (| b c d | d d d d) input extended nearest pixel. Note using torch backend, \"reflect\" redirected \"mirror\" (c d c b | b c d | c b b) torch support \"reflect\". Note torch backend support \"wrap\". interpolation Interpolation mode. Supported values: \"nearest\", \"bilinear\". seed Integer. Used create random seed. fill_value float represents value filled outside boundaries fill_mode=\"constant\". data_format string, either \"channels_last\" \"channels_first\". ordering dimensions inputs. \"channels_last\" corresponds inputs shape (batch, height, width, channels) \"channels_first\" corresponds inputs shape (batch, channels, height, width). defaults image_data_format value found Keras config file ~/.keras/keras.json. never set , \"channels_last\". ... Base layer keyword arguments, name dtype.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_random_zoom.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"A preprocessing layer which randomly zooms images during training. — layer_random_zoom","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_random_zoom.html","id":"input-shape","dir":"Reference","previous_headings":"","what":"Input Shape","title":"A preprocessing layer which randomly zooms images during training. — layer_random_zoom","text":"3D (unbatched) 4D (batched) tensor shape: (..., height, width, channels), \"channels_last\" format, (..., channels, height, width), \"channels_first\" format.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_random_zoom.html","id":"output-shape","dir":"Reference","previous_headings":"","what":"Output Shape","title":"A preprocessing layer which randomly zooms images during training. — layer_random_zoom","text":"3D (unbatched) 4D (batched) tensor shape: (..., target_height, target_width, channels), (..., channels, target_height, target_width), \"channels_first\" format. Note: layer safe use inside tf.data pipeline (independently backend using).","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_random_zoom.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"A preprocessing layer which randomly zooms images during training. — layer_random_zoom","text":"","code":"input_img <- random_uniform(c(32, 224, 224, 3)) layer <- layer_random_zoom(height_factor = .5, width_factor = .2) out_img <- layer(input_img)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_repeat_vector.html","id":null,"dir":"Reference","previous_headings":"","what":"Repeats the input n times. — layer_repeat_vector","title":"Repeats the input n times. — layer_repeat_vector","text":"Repeats input n times.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_repeat_vector.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Repeats the input n times. — layer_repeat_vector","text":"","code":"layer_repeat_vector(object, n, ...)"},{"path":"https://keras3.posit.co/dev/reference/layer_repeat_vector.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Repeats the input n times. — layer_repeat_vector","text":"object Object compose layer . tensor, array, sequential model. n Integer, repetition factor. ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_repeat_vector.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Repeats the input n times. — layer_repeat_vector","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_repeat_vector.html","id":"example","dir":"Reference","previous_headings":"","what":"Example","title":"Repeats the input n times. — layer_repeat_vector","text":"","code":"x <- layer_input(shape = 32) y <- layer_repeat_vector(x, n = 3) shape(y) ## shape(NA, 3, 32)"},{"path":"https://keras3.posit.co/dev/reference/layer_repeat_vector.html","id":"input-shape","dir":"Reference","previous_headings":"","what":"Input Shape","title":"Repeats the input n times. — layer_repeat_vector","text":"2D tensor shape (batch_size, features).","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_repeat_vector.html","id":"output-shape","dir":"Reference","previous_headings":"","what":"Output Shape","title":"Repeats the input n times. — layer_repeat_vector","text":"3D tensor shape (batch_size, n, features).","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_rescaling.html","id":null,"dir":"Reference","previous_headings":"","what":"A preprocessing layer which rescales input values to a new range. — layer_rescaling","title":"A preprocessing layer which rescales input values to a new range. — layer_rescaling","text":"layer rescales every value input (often image) multiplying scale adding offset. instance: rescale input [0, 255] range [0, 1] range, pass scale=1./255. rescale input [0, 255] range [-1, 1] range, pass scale=1./127.5, offset=-1. rescaling applied training inference. Inputs can integer floating point dtype, default layer output floats. Note: layer safe use inside tf.data pipeline (independently backend using).","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_rescaling.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"A preprocessing layer which rescales input values to a new range. — layer_rescaling","text":"","code":"layer_rescaling(object, scale, offset = 0, ...)"},{"path":"https://keras3.posit.co/dev/reference/layer_rescaling.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"A preprocessing layer which rescales input values to a new range. — layer_rescaling","text":"object Object compose layer . tensor, array, sequential model. scale Float, scale apply inputs. offset Float, offset apply inputs. ... Base layer keyword arguments, name dtype.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_rescaling.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"A preprocessing layer which rescales input values to a new range. — layer_rescaling","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_reshape.html","id":null,"dir":"Reference","previous_headings":"","what":"Layer that reshapes inputs into the given shape. — layer_reshape","title":"Layer that reshapes inputs into the given shape. — layer_reshape","text":"Layer reshapes inputs given shape.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_reshape.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Layer that reshapes inputs into the given shape. — layer_reshape","text":"","code":"layer_reshape(object, target_shape, ...)"},{"path":"https://keras3.posit.co/dev/reference/layer_reshape.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Layer that reshapes inputs into the given shape. — layer_reshape","text":"object Object compose layer . tensor, array, sequential model. target_shape Target shape. List integers, include samples dimension (batch size). ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_reshape.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Layer that reshapes inputs into the given shape. — layer_reshape","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_reshape.html","id":"input-shape","dir":"Reference","previous_headings":"","what":"Input Shape","title":"Layer that reshapes inputs into the given shape. — layer_reshape","text":"Arbitrary, although dimensions input shape must known/fixed. Use keyword argument input_shape (list integers, include samples/batch size axis) using layer first layer model.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_reshape.html","id":"output-shape","dir":"Reference","previous_headings":"","what":"Output Shape","title":"Layer that reshapes inputs into the given shape. — layer_reshape","text":"(batch_size, *target_shape)","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_reshape.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Layer that reshapes inputs into the given shape. — layer_reshape","text":"","code":"x <- layer_input(shape = 12) y <- layer_reshape(x, c(3, 4)) shape(y) ## shape(NA, 3, 4) # also supports shape inference using `-1` as dimension y <- layer_reshape(x, c(-1, 2, 2)) shape(y) ## shape(NA, 3, 2, 2)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_resizing.html","id":null,"dir":"Reference","previous_headings":"","what":"A preprocessing layer which resizes images. — layer_resizing","title":"A preprocessing layer which resizes images. — layer_resizing","text":"layer resizes image input target height width. input 4D (batched) 3D (unbatched) tensor \"channels_last\" format. Input pixel values can range (e.g. [0., 1.) [0, 255]).","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_resizing.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"A preprocessing layer which resizes images. — layer_resizing","text":"","code":"layer_resizing(   object,   height,   width,   interpolation = \"bilinear\",   crop_to_aspect_ratio = FALSE,   pad_to_aspect_ratio = FALSE,   fill_mode = \"constant\",   fill_value = 0,   data_format = NULL,   ...,   antialias = FALSE )"},{"path":"https://keras3.posit.co/dev/reference/layer_resizing.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"A preprocessing layer which resizes images. — layer_resizing","text":"object Object compose layer . tensor, array, sequential model. height Integer, height output shape. width Integer, width output shape. interpolation String, interpolation method. Supports \"bilinear\", \"nearest\", \"bicubic\", \"lanczos3\", \"lanczos5\". Defaults \"bilinear\". crop_to_aspect_ratio TRUE, resize images without aspect ratio distortion. original aspect ratio differs target aspect ratio, output image cropped return largest possible window image (size (height, width)) matches target aspect ratio. default (crop_to_aspect_ratio=FALSE), aspect ratio may preserved. pad_to_aspect_ratio TRUE, pad images without aspect ratio distortion. original aspect ratio differs target aspect ratio, output image evenly padded short side. fill_mode using pad_to_aspect_ratio=TRUE, padded areas filled according given mode. \"constant\" supported time (fill constant value, equal fill_value). fill_value Float. Padding value use pad_to_aspect_ratio=TRUE. data_format string, either \"channels_last\" \"channels_first\". ordering dimensions inputs. \"channels_last\" corresponds inputs shape (batch, height, width, channels) \"channels_first\" corresponds inputs shape (batch, channels, height, width). defaults image_data_format value found Keras config file ~/.keras/keras.json. never set , \"channels_last\". ... Base layer keyword arguments, name dtype. antialias bool","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_resizing.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"A preprocessing layer which resizes images. — layer_resizing","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_resizing.html","id":"input-shape","dir":"Reference","previous_headings":"","what":"Input Shape","title":"A preprocessing layer which resizes images. — layer_resizing","text":"3D (unbatched) 4D (batched) tensor shape: (..., height, width, channels), \"channels_last\" format, (..., channels, height, width), \"channels_first\" format.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_resizing.html","id":"output-shape","dir":"Reference","previous_headings":"","what":"Output Shape","title":"A preprocessing layer which resizes images. — layer_resizing","text":"3D (unbatched) 4D (batched) tensor shape: (..., target_height, target_width, channels), (..., channels, target_height, target_width), \"channels_first\" format. Note: layer safe use inside tf.data pipeline (independently backend using).","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_rms_normalization.html","id":null,"dir":"Reference","previous_headings":"","what":"Root Mean Square (RMS) Normalization layer. — layer_rms_normalization","title":"Root Mean Square (RMS) Normalization layer. — layer_rms_normalization","text":"layer normalizes input tensor based RMS value. Keras layer performs operation described Root Mean Square Layer Normalization Biao Zhang et al. scale enabled, layer scale normalized outputs via learnable scaling factor. , scaling enabled, normalization equations follows: Let intermediate activations mini-batch inputs.   example:","code":"rms_normalization(x) = x * rsqrt(mean(square(x))) * scale layer <- layer_rms_normalization() layer$build(shape(5, 20, 30, 10)) op_shape(layer$scale$shape) ## shape(1)  op_shape(layer(op_array(runif(10)))) ## shape(10)"},{"path":"https://keras3.posit.co/dev/reference/layer_rms_normalization.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Root Mean Square (RMS) Normalization layer. — layer_rms_normalization","text":"","code":"layer_rms_normalization(object, axis = -1L, epsilon = 1e-06, ...)"},{"path":"https://keras3.posit.co/dev/reference/layer_rms_normalization.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Root Mean Square (RMS) Normalization layer. — layer_rms_normalization","text":"object Object compose layer . tensor, array, sequential model. axis int. axis perform normalization. epsilon float. small number add avoid division zero. ... forward/backward compatability.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_rnn.html","id":null,"dir":"Reference","previous_headings":"","what":"Base class for recurrent layers — layer_rnn","title":"Base class for recurrent layers — layer_rnn","text":"Base class recurrent layers","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_rnn.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Base class for recurrent layers — layer_rnn","text":"","code":"layer_rnn(   object,   cell,   return_sequences = FALSE,   return_state = FALSE,   go_backwards = FALSE,   stateful = FALSE,   unroll = FALSE,   zero_output_for_mask = FALSE,   ... )"},{"path":"https://keras3.posit.co/dev/reference/layer_rnn.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Base class for recurrent layers — layer_rnn","text":"object Object compose layer . tensor, array, sequential model. cell RNN cell instance list RNN cell instances. RNN cell class : call(input_at_t, states_at_t) method, returning (output_at_t, states_at_t_plus_1). call method cell can also take optional argument constants, see section \"Note passing external constants\" . state_size attribute. can single integer (single state) case size recurrent state. can also list integers (one size per state). output_size attribute, single integer. get_initial_state(batch_size=NULL) method creates tensor meant fed call() initial state, user specify initial state via means. returned initial state shape (batch_size, cell.state_size). cell might choose create tensor full zeros, values based cell's implementation. inputs input tensor RNN layer, shape (batch_size, timesteps, features). method implemented cell, RNN layer create zero filled tensor shape (batch_size, cell$state_size). case cell list RNN cell instances, cells stacked top RNN, resulting efficient stacked RNN. return_sequences Boolean (default FALSE). Whether return last output output sequence, full sequence. return_state Boolean (default FALSE). Whether return last state addition output. go_backwards Boolean (default FALSE). TRUE, process input sequence backwards return reversed sequence. stateful Boolean (default FALSE). TRUE, last state sample index batch used initial state sample index following batch. unroll Boolean (default FALSE). TRUE, network unrolled, else symbolic loop used. Unrolling can speed-RNN, although tends memory-intensive. Unrolling suitable short sequences. zero_output_for_mask Boolean (default FALSE). Whether output use zeros masked timesteps. Note field used return_sequences TRUE mask provided. can useful want reuse raw output sequence RNN without interference masked timesteps, e.g., merging bidirectional RNNs. ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_rnn.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Base class for recurrent layers — layer_rnn","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_rnn.html","id":"call-arguments","dir":"Reference","previous_headings":"","what":"Call Arguments","title":"Base class for recurrent layers — layer_rnn","text":"sequences: 3-D tensor shape (batch_size, timesteps, features). initial_state: List initial state tensors passed first call cell. mask: Binary tensor shape [batch_size, timesteps] indicating whether given timestep masked. individual TRUE entry indicates corresponding timestep utilized, FALSE entry indicates corresponding timestep ignored. training: Python boolean indicating whether layer behave training mode inference mode. argument passed cell calling . use cells use dropout.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_rnn.html","id":"input-shape","dir":"Reference","previous_headings":"","what":"Input Shape","title":"Base class for recurrent layers — layer_rnn","text":"3-D tensor shape (batch_size, timesteps, features).","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_rnn.html","id":"output-shape","dir":"Reference","previous_headings":"","what":"Output Shape","title":"Base class for recurrent layers — layer_rnn","text":"return_state: list tensors. first tensor output. remaining tensors last states, shape (batch_size, state_size), state_size high dimension tensor shape. return_sequences: 3D tensor shape (batch_size, timesteps, output_size).","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_rnn.html","id":"masking-","dir":"Reference","previous_headings":"","what":"Masking:","title":"Base class for recurrent layers — layer_rnn","text":"layer supports masking input data variable number timesteps. introduce masks data, use layer_embedding() layer mask_zero parameter set TRUE. Note using statefulness RNNs: can set RNN layers 'stateful', means states computed samples one batch reused initial states samples next batch. assumes one--one mapping samples different successive batches. enable statefulness: Specify stateful=TRUE layer constructor. Specify fixed batch size model, passing batch_size=... layer_input() layer(s) model. Remember also specify batch_size=... calling fit(), otherwise use generator-like data source like tf.data.Dataset. Specify shuffle=FALSE calling fit(), since batches expected temporally ordered. reset states model, call reset_state() either specific layer, entire model. Note specifying initial state RNNs: can specify initial state RNN layers symbolically passing named argument initial_state layer reset_state(). value initial_state tensor list tensors representing initial state RNN layer.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_rnn.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Base class for recurrent layers — layer_rnn","text":"First, define RNN Cell, layer subclass.   use cell RNN layer:","code":"rnn_cell_minimal <- Layer(   \"MinimalRNNCell\",    initialize = function(units, ...) {     super$initialize(...)     self$units <- as.integer(units)     self$state_size <- as.integer(units)   },    build = function(input_shape) {     self$kernel <- self$add_weight(       shape = shape(tail(input_shape, 1), self$units),       initializer = 'uniform',       name = 'kernel'     )     self$recurrent_kernel <- self$add_weight(       shape = shape(self$units, self$units),       initializer = 'uniform',       name = 'recurrent_kernel'     )     self$built <- TRUE   },    call = function(inputs, states) {     prev_output <- states[[1]]     h <- op_matmul(inputs, self$kernel)     output <- h + op_matmul(prev_output, self$recurrent_kernel)     list(output, list(output))   } ) cell <- rnn_cell_minimal(units = 32) x <- layer_input(shape = shape(NULL, 5)) layer <- layer_rnn(cell = cell) y <- layer(x) cells <- list(rnn_cell_minimal(units = 32), rnn_cell_minimal(units = 4)) x <- layer_input(shape = shape(NULL, 5)) layer <- layer_rnn(cell = cells) y <- layer(x)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_separable_conv_1d.html","id":null,"dir":"Reference","previous_headings":"","what":"1D separable convolution layer. — layer_separable_conv_1d","title":"1D separable convolution layer. — layer_separable_conv_1d","text":"layer performs depthwise convolution acts separately channels, followed pointwise convolution mixes channels. use_bias TRUE bias initializer provided, adds bias vector output. optionally applies activation function produce final output.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_separable_conv_1d.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"1D separable convolution layer. — layer_separable_conv_1d","text":"","code":"layer_separable_conv_1d(   object,   filters,   kernel_size,   strides = 1L,   padding = \"valid\",   data_format = NULL,   dilation_rate = 1L,   depth_multiplier = 1L,   activation = NULL,   use_bias = TRUE,   depthwise_initializer = \"glorot_uniform\",   pointwise_initializer = \"glorot_uniform\",   bias_initializer = \"zeros\",   depthwise_regularizer = NULL,   pointwise_regularizer = NULL,   bias_regularizer = NULL,   activity_regularizer = NULL,   depthwise_constraint = NULL,   pointwise_constraint = NULL,   bias_constraint = NULL,   ... )"},{"path":"https://keras3.posit.co/dev/reference/layer_separable_conv_1d.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"1D separable convolution layer. — layer_separable_conv_1d","text":"object Object compose layer . tensor, array, sequential model. filters int, dimensionality output space (.e. number filters pointwise convolution). kernel_size int list 1 integers, specifying size depthwise convolution window. strides int list 1 integers, specifying stride length depthwise convolution. one int specified, stride size used dimensions. strides > 1 incompatible dilation_rate > 1. padding string, either \"valid\" \"\" (case-insensitive). \"valid\" means padding. \"\" results padding evenly left/right /input. padding=\"\" strides=1, output size input. data_format string, either \"channels_last\" \"channels_first\". ordering dimensions inputs. \"channels_last\" corresponds inputs shape (batch, steps, features) \"channels_first\" corresponds inputs shape (batch, features, steps). defaults image_data_format value found Keras config file ~/.keras/keras.json. never set , \"channels_last\". dilation_rate int list 1 integers, specifying dilation rate use dilated convolution. one int specified, dilation rate used dimensions. depth_multiplier number depthwise convolution output channels input channel. total number depthwise convolution output channels equal input_channel * depth_multiplier. activation Activation function. NULL, activation applied. use_bias bool, TRUE, bias added output. depthwise_initializer initializer depthwise convolution kernel. NULL, default initializer (\"glorot_uniform\") used. pointwise_initializer initializer pointwise convolution kernel. NULL, default initializer (\"glorot_uniform\") used. bias_initializer initializer bias vector. NULL, default initializer ('\"zeros\"') used. depthwise_regularizer Optional regularizer depthwise convolution kernel. pointwise_regularizer Optional regularizer pointwise convolution kernel. bias_regularizer Optional regularizer bias vector. activity_regularizer Optional regularizer function output. depthwise_constraint Optional projection function applied depthwise kernel updated Optimizer (e.g. used norm constraints value constraints layer weights). function must take input unprojected variable must return projected variable (must shape). pointwise_constraint Optional projection function applied pointwise kernel updated Optimizer. bias_constraint Optional projection function applied bias updated Optimizer. ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_separable_conv_1d.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"1D separable convolution layer. — layer_separable_conv_1d","text":"3D tensor representing activation(separable_conv1d(inputs, kernel) + bias).","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_separable_conv_1d.html","id":"input-shape","dir":"Reference","previous_headings":"","what":"Input Shape","title":"1D separable convolution layer. — layer_separable_conv_1d","text":"data_format=\"channels_last\": 3D tensor shape: (batch_shape, steps, channels) data_format=\"channels_first\": 3D tensor shape: (batch_shape, channels, steps)","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_separable_conv_1d.html","id":"output-shape","dir":"Reference","previous_headings":"","what":"Output Shape","title":"1D separable convolution layer. — layer_separable_conv_1d","text":"data_format=\"channels_last\": 3D tensor shape: (batch_shape, new_steps, filters) data_format=\"channels_first\": 3D tensor shape: (batch_shape, filters, new_steps)","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_separable_conv_1d.html","id":"example","dir":"Reference","previous_headings":"","what":"Example","title":"1D separable convolution layer. — layer_separable_conv_1d","text":"","code":"x <- random_uniform(c(4, 10, 12)) y <- layer_separable_conv_1d(x, 3, 2, 2, activation='relu') shape(y) ## shape(4, 5, 3)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_separable_conv_2d.html","id":null,"dir":"Reference","previous_headings":"","what":"2D separable convolution layer. — layer_separable_conv_2d","title":"2D separable convolution layer. — layer_separable_conv_2d","text":"layer performs depthwise convolution acts separately channels, followed pointwise convolution mixes channels. use_bias TRUE bias initializer provided, adds bias vector output. optionally applies activation function produce final output.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_separable_conv_2d.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"2D separable convolution layer. — layer_separable_conv_2d","text":"","code":"layer_separable_conv_2d(   object,   filters,   kernel_size,   strides = list(1L, 1L),   padding = \"valid\",   data_format = NULL,   dilation_rate = list(1L, 1L),   depth_multiplier = 1L,   activation = NULL,   use_bias = TRUE,   depthwise_initializer = \"glorot_uniform\",   pointwise_initializer = \"glorot_uniform\",   bias_initializer = \"zeros\",   depthwise_regularizer = NULL,   pointwise_regularizer = NULL,   bias_regularizer = NULL,   activity_regularizer = NULL,   depthwise_constraint = NULL,   pointwise_constraint = NULL,   bias_constraint = NULL,   ... )"},{"path":"https://keras3.posit.co/dev/reference/layer_separable_conv_2d.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"2D separable convolution layer. — layer_separable_conv_2d","text":"object Object compose layer . tensor, array, sequential model. filters int, dimensionality output space (.e. number filters pointwise convolution). kernel_size int list 2 integers, specifying size depthwise convolution window. strides int list 2 integers, specifying stride length depthwise convolution. one int specified, stride size used dimensions. strides > 1 incompatible dilation_rate > 1. padding string, either \"valid\" \"\" (case-insensitive). \"valid\" means padding. \"\" results padding evenly left/right /input. padding=\"\" strides=1, output size input. data_format string, either \"channels_last\" \"channels_first\". ordering dimensions inputs. \"channels_last\" corresponds inputs shape (batch, height, width, channels) \"channels_first\" corresponds inputs shape (batch, channels, height, width). defaults image_data_format value found Keras config file ~/.keras/keras.json. never set , \"channels_last\". dilation_rate int list 2 integers, specifying dilation rate use dilated convolution. one int specified, dilation rate used dimensions. depth_multiplier number depthwise convolution output channels input channel. total number depthwise convolution output channels equal input_channel * depth_multiplier. activation Activation function. NULL, activation applied. use_bias bool, TRUE, bias added output. depthwise_initializer initializer depthwise convolution kernel. NULL, default initializer (\"glorot_uniform\") used. pointwise_initializer initializer pointwise convolution kernel. NULL, default initializer (\"glorot_uniform\") used. bias_initializer initializer bias vector. NULL, default initializer ('\"zeros\"') used. depthwise_regularizer Optional regularizer depthwise convolution kernel. pointwise_regularizer Optional regularizer pointwise convolution kernel. bias_regularizer Optional regularizer bias vector. activity_regularizer Optional regularizer function output. depthwise_constraint Optional projection function applied depthwise kernel updated Optimizer (e.g. used norm constraints value constraints layer weights). function must take input unprojected variable must return projected variable (must shape). pointwise_constraint Optional projection function applied pointwise kernel updated Optimizer. bias_constraint Optional projection function applied bias updated Optimizer. ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_separable_conv_2d.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"2D separable convolution layer. — layer_separable_conv_2d","text":"4D tensor representing activation(separable_conv2d(inputs, kernel) + bias).","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_separable_conv_2d.html","id":"input-shape","dir":"Reference","previous_headings":"","what":"Input Shape","title":"2D separable convolution layer. — layer_separable_conv_2d","text":"data_format=\"channels_last\": 4D tensor shape: (batch_size, height, width, channels) data_format=\"channels_first\": 4D tensor shape: (batch_size, channels, height, width)","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_separable_conv_2d.html","id":"output-shape","dir":"Reference","previous_headings":"","what":"Output Shape","title":"2D separable convolution layer. — layer_separable_conv_2d","text":"data_format=\"channels_last\": 4D tensor shape: (batch_size, new_height, new_width, filters) data_format=\"channels_first\": 4D tensor shape: (batch_size, filters, new_height, new_width)","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_separable_conv_2d.html","id":"example","dir":"Reference","previous_headings":"","what":"Example","title":"2D separable convolution layer. — layer_separable_conv_2d","text":"","code":"x <- random_uniform(c(4, 10, 10, 12)) y <- layer_separable_conv_2d(x, 3, c(4, 3), 2, activation='relu') shape(y) ## shape(4, 4, 4, 3)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_simple_rnn.html","id":null,"dir":"Reference","previous_headings":"","what":"Fully-connected RNN where the output is to be fed back as the new input. — layer_simple_rnn","title":"Fully-connected RNN where the output is to be fed back as the new input. — layer_simple_rnn","text":"Fully-connected RNN output fed back new input.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_simple_rnn.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Fully-connected RNN where the output is to be fed back as the new input. — layer_simple_rnn","text":"","code":"layer_simple_rnn(   object,   units,   activation = \"tanh\",   use_bias = TRUE,   kernel_initializer = \"glorot_uniform\",   recurrent_initializer = \"orthogonal\",   bias_initializer = \"zeros\",   kernel_regularizer = NULL,   recurrent_regularizer = NULL,   bias_regularizer = NULL,   activity_regularizer = NULL,   kernel_constraint = NULL,   recurrent_constraint = NULL,   bias_constraint = NULL,   dropout = 0,   recurrent_dropout = 0,   return_sequences = FALSE,   return_state = FALSE,   go_backwards = FALSE,   stateful = FALSE,   unroll = FALSE,   seed = NULL,   ... )"},{"path":"https://keras3.posit.co/dev/reference/layer_simple_rnn.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Fully-connected RNN where the output is to be fed back as the new input. — layer_simple_rnn","text":"object Object compose layer . tensor, array, sequential model. units Positive integer, dimensionality output space. activation Activation function use. Default: hyperbolic tangent (tanh). pass NULL, activation applied (ie. \"linear\" activation: (x) = x). use_bias Boolean, (default TRUE), whether layer uses bias vector. kernel_initializer Initializer kernel weights matrix, used linear transformation inputs. Default: \"glorot_uniform\". recurrent_initializer Initializer recurrent_kernel weights matrix, used linear transformation recurrent state.  Default: \"orthogonal\". bias_initializer Initializer bias vector. Default: \"zeros\". kernel_regularizer Regularizer function applied kernel weights matrix. Default: NULL. recurrent_regularizer Regularizer function applied recurrent_kernel weights matrix. Default: NULL. bias_regularizer Regularizer function applied bias vector. Default: NULL. activity_regularizer Regularizer function applied output layer (\"activation\"). Default: NULL. kernel_constraint Constraint function applied kernel weights matrix. Default: NULL. recurrent_constraint Constraint function applied recurrent_kernel weights matrix.  Default: NULL. bias_constraint Constraint function applied bias vector. Default: NULL. dropout Float 0 1. Fraction units drop linear transformation inputs. Default: 0. recurrent_dropout Float 0 1. Fraction units drop linear transformation recurrent state. Default: 0. return_sequences Boolean. Whether return last output output sequence, full sequence. Default: FALSE. return_state Boolean. Whether return last state addition output. Default: FALSE. go_backwards Boolean (default: FALSE). TRUE, process input sequence backwards return reversed sequence. stateful Boolean (default: FALSE). TRUE, last state sample index batch used initial state sample index following batch. unroll Boolean (default: FALSE). TRUE, network unrolled, else symbolic loop used. Unrolling can speed-RNN, although tends memory-intensive. Unrolling suitable short sequences. seed Initial seed random number generator ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_simple_rnn.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Fully-connected RNN where the output is to be fed back as the new input. — layer_simple_rnn","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_simple_rnn.html","id":"call-arguments","dir":"Reference","previous_headings":"","what":"Call Arguments","title":"Fully-connected RNN where the output is to be fed back as the new input. — layer_simple_rnn","text":"sequence: 3D tensor, shape [batch, timesteps, feature]. mask: Binary tensor shape [batch, timesteps] indicating whether given timestep masked. individual TRUE entry indicates corresponding timestep utilized, FALSE entry indicates corresponding timestep ignored. training: Python boolean indicating whether layer behave training mode inference mode. argument passed cell calling . relevant dropout recurrent_dropout used. initial_state: List initial state tensors passed first call cell.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_simple_rnn.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Fully-connected RNN where the output is to be fed back as the new input. — layer_simple_rnn","text":"","code":"inputs <- random_uniform(c(32, 10, 8)) simple_rnn <- layer_simple_rnn(units = 4) output <- simple_rnn(inputs)  # The output has shape `(32, 4)`. simple_rnn <- layer_simple_rnn(     units = 4, return_sequences=TRUE, return_state=TRUE ) # whole_sequence_output has shape `(32, 10, 4)`. # final_state has shape `(32, 4)`. c(whole_sequence_output, final_state) %<-% simple_rnn(inputs)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_solarization.html","id":null,"dir":"Reference","previous_headings":"","what":"Applies (max_value - pixel + min_value) for each pixel in the image. — layer_solarization","title":"Applies (max_value - pixel + min_value) for each pixel in the image. — layer_solarization","text":"created without threshold parameter, layer performs solarization values. created specified threshold layer augments pixels threshold value.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_solarization.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Applies (max_value - pixel + min_value) for each pixel in the image. — layer_solarization","text":"","code":"layer_solarization(   object,   addition_factor = 0,   threshold_factor = 0,   value_range = tuple(0L, 255L),   seed = NULL,   ... )"},{"path":"https://keras3.posit.co/dev/reference/layer_solarization.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Applies (max_value - pixel + min_value) for each pixel in the image. — layer_solarization","text":"object Object compose layer . tensor, array, sequential model. addition_factor (Optional)  tuple two floats single float, 0 1. augmented image value sampled provided range. float passed, range interpreted (0, addition_factor). specified, value (times value range input images, e.g. 255), added pixel solarization thresholding. Defaults 0.0. threshold_factor (Optional)  tuple two floats single float. augmented image value sampled provided range. float passed, range interpreted (0, threshold_factor). specified, pixel values threshold solarized. value_range tuple list two elements. first value represents lower bound values input images, second represents upper bound. Images passed layer values within value_range. Typical values pass (0, 255) (RGB image) (0., 1.) (scaled image). seed Integer. Used create random seed. ... Base layer keyword arguments, name dtype.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_solarization.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Applies (max_value - pixel + min_value) for each pixel in the image. — layer_solarization","text":"","code":"c(c(images, labels), .) %<-% dataset_cifar10() str(images) ##  int [1:50000, 1:32, 1:32, 1:3] 59 154 255 28 170 159 164 28 134 125 ... str(images[1, 1, 1, ]) ##  int [1:3] 59 62 63 # Note that images are Tensor with values in the range [0, 255] solarization <- layer_solarization(value_range = c(0, 255)) images <- solarization(images) |> as.array() str(images[1, 1, 1, ]) ##  num [1:3] 196 193 192"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_spatial_dropout_1d.html","id":null,"dir":"Reference","previous_headings":"","what":"Spatial 1D version of Dropout. — layer_spatial_dropout_1d","title":"Spatial 1D version of Dropout. — layer_spatial_dropout_1d","text":"layer performs function Dropout, however, drops entire 1D feature maps instead individual elements. adjacent frames within feature maps strongly correlated (normally case early convolution layers) regular dropout regularize activations otherwise just result effective learning rate decrease. case, SpatialDropout1D help promote independence feature maps used instead.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_spatial_dropout_1d.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Spatial 1D version of Dropout. — layer_spatial_dropout_1d","text":"","code":"layer_spatial_dropout_1d(object, rate, seed = NULL, name = NULL, dtype = NULL)"},{"path":"https://keras3.posit.co/dev/reference/layer_spatial_dropout_1d.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Spatial 1D version of Dropout. — layer_spatial_dropout_1d","text":"object Object compose layer . tensor, array, sequential model. rate Float 0 1. Fraction input units drop. seed Initial seed random number generator name String, name object dtype datatype (e.g., \"float32\").","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_spatial_dropout_1d.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Spatial 1D version of Dropout. — layer_spatial_dropout_1d","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_spatial_dropout_1d.html","id":"call-arguments","dir":"Reference","previous_headings":"","what":"Call Arguments","title":"Spatial 1D version of Dropout. — layer_spatial_dropout_1d","text":"inputs: 3D tensor. training: Python boolean indicating whether layer behave training mode (applying dropout) inference mode (pass-).","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_spatial_dropout_1d.html","id":"input-shape","dir":"Reference","previous_headings":"","what":"Input Shape","title":"Spatial 1D version of Dropout. — layer_spatial_dropout_1d","text":"3D tensor shape: (samples, timesteps, channels)","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_spatial_dropout_1d.html","id":"output-shape","dir":"Reference","previous_headings":"","what":"Output Shape","title":"Spatial 1D version of Dropout. — layer_spatial_dropout_1d","text":"input.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_spatial_dropout_1d.html","id":"reference","dir":"Reference","previous_headings":"","what":"Reference","title":"Spatial 1D version of Dropout. — layer_spatial_dropout_1d","text":"Tompson et al., 2014","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_spatial_dropout_2d.html","id":null,"dir":"Reference","previous_headings":"","what":"Spatial 2D version of Dropout. — layer_spatial_dropout_2d","title":"Spatial 2D version of Dropout. — layer_spatial_dropout_2d","text":"version performs function Dropout, however, drops entire 2D feature maps instead individual elements. adjacent pixels within feature maps strongly correlated (normally case early convolution layers) regular dropout regularize activations otherwise just result effective learning rate decrease. case, SpatialDropout2D help promote independence feature maps used instead.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_spatial_dropout_2d.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Spatial 2D version of Dropout. — layer_spatial_dropout_2d","text":"","code":"layer_spatial_dropout_2d(   object,   rate,   data_format = NULL,   seed = NULL,   name = NULL,   dtype = NULL )"},{"path":"https://keras3.posit.co/dev/reference/layer_spatial_dropout_2d.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Spatial 2D version of Dropout. — layer_spatial_dropout_2d","text":"object Object compose layer . tensor, array, sequential model. rate Float 0 1. Fraction input units drop. data_format \"channels_first\" \"channels_last\". \"channels_first\" mode, channels dimension (depth) index 1, \"channels_last\" mode index 3. defaults image_data_format value found Keras config file ~/.keras/keras.json. never set , \"channels_last\". seed Initial seed random number generator name String, name object dtype datatype (e.g., \"float32\").","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_spatial_dropout_2d.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Spatial 2D version of Dropout. — layer_spatial_dropout_2d","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_spatial_dropout_2d.html","id":"call-arguments","dir":"Reference","previous_headings":"","what":"Call Arguments","title":"Spatial 2D version of Dropout. — layer_spatial_dropout_2d","text":"inputs: 4D tensor. training: Python boolean indicating whether layer behave training mode (applying dropout) inference mode (pass-).","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_spatial_dropout_2d.html","id":"input-shape","dir":"Reference","previous_headings":"","what":"Input Shape","title":"Spatial 2D version of Dropout. — layer_spatial_dropout_2d","text":"4D tensor shape: (samples, channels, rows, cols) data_format='channels_first' 4D tensor shape: (samples, rows, cols, channels) data_format='channels_last'.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_spatial_dropout_2d.html","id":"output-shape","dir":"Reference","previous_headings":"","what":"Output Shape","title":"Spatial 2D version of Dropout. — layer_spatial_dropout_2d","text":"input.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_spatial_dropout_2d.html","id":"reference","dir":"Reference","previous_headings":"","what":"Reference","title":"Spatial 2D version of Dropout. — layer_spatial_dropout_2d","text":"Tompson et al., 2014","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_spatial_dropout_3d.html","id":null,"dir":"Reference","previous_headings":"","what":"Spatial 3D version of Dropout. — layer_spatial_dropout_3d","title":"Spatial 3D version of Dropout. — layer_spatial_dropout_3d","text":"version performs function Dropout, however, drops entire 3D feature maps instead individual elements. adjacent voxels within feature maps strongly correlated (normally case early convolution layers) regular dropout regularize activations otherwise just result effective learning rate decrease. case, SpatialDropout3D help promote independence feature maps used instead.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_spatial_dropout_3d.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Spatial 3D version of Dropout. — layer_spatial_dropout_3d","text":"","code":"layer_spatial_dropout_3d(   object,   rate,   data_format = NULL,   seed = NULL,   name = NULL,   dtype = NULL )"},{"path":"https://keras3.posit.co/dev/reference/layer_spatial_dropout_3d.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Spatial 3D version of Dropout. — layer_spatial_dropout_3d","text":"object Object compose layer . tensor, array, sequential model. rate Float 0 1. Fraction input units drop. data_format \"channels_first\" \"channels_last\". \"channels_first\" mode, channels dimension (depth) index 1, \"channels_last\" mode index 4. defaults image_data_format value found Keras config file ~/.keras/keras.json. never set , \"channels_last\". seed Initial seed random number generator name String, name object dtype datatype (e.g., \"float32\").","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_spatial_dropout_3d.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Spatial 3D version of Dropout. — layer_spatial_dropout_3d","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_spatial_dropout_3d.html","id":"call-arguments","dir":"Reference","previous_headings":"","what":"Call Arguments","title":"Spatial 3D version of Dropout. — layer_spatial_dropout_3d","text":"inputs: 5D tensor. training: Python boolean indicating whether layer behave training mode (applying dropout) inference mode (pass-).","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_spatial_dropout_3d.html","id":"input-shape","dir":"Reference","previous_headings":"","what":"Input Shape","title":"Spatial 3D version of Dropout. — layer_spatial_dropout_3d","text":"5D tensor shape: (samples, channels, dim1, dim2, dim3) data_format='channels_first' 5D tensor shape: (samples, dim1, dim2, dim3, channels) data_format='channels_last'.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_spatial_dropout_3d.html","id":"output-shape","dir":"Reference","previous_headings":"","what":"Output Shape","title":"Spatial 3D version of Dropout. — layer_spatial_dropout_3d","text":"input.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_spatial_dropout_3d.html","id":"reference","dir":"Reference","previous_headings":"","what":"Reference","title":"Spatial 3D version of Dropout. — layer_spatial_dropout_3d","text":"Tompson et al., 2014","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_spectral_normalization.html","id":null,"dir":"Reference","previous_headings":"","what":"Performs spectral normalization on the weights of a target layer. — layer_spectral_normalization","title":"Performs spectral normalization on the weights of a target layer. — layer_spectral_normalization","text":"wrapper controls Lipschitz constant weights layer constraining spectral norm, can stabilize training GANs.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_spectral_normalization.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Performs spectral normalization on the weights of a target layer. — layer_spectral_normalization","text":"","code":"layer_spectral_normalization(object, layer, power_iterations = 1L, ...)"},{"path":"https://keras3.posit.co/dev/reference/layer_spectral_normalization.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Performs spectral normalization on the weights of a target layer. — layer_spectral_normalization","text":"object Object compose layer . tensor, array, sequential model. layer Layer instance either kernel (e.g. layer_conv_2d, layer_dense...) embeddings attribute (layer_embedding layer). power_iterations int, number iterations normalization. ... Base wrapper keyword arguments.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_spectral_normalization.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Performs spectral normalization on the weights of a target layer. — layer_spectral_normalization","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_spectral_normalization.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Performs spectral normalization on the weights of a target layer. — layer_spectral_normalization","text":"Wrap layer_conv_2d:     Wrap layer_dense:","code":"x <- random_uniform(c(1, 10, 10, 1)) conv2d <- layer_spectral_normalization(   layer = layer_conv_2d(filters = 2, kernel_size = 2) ) y <- conv2d(x) shape(y) ## shape(1, 9, 9, 2) x <- random_uniform(c(1, 10, 10, 1)) dense <- layer_spectral_normalization(layer = layer_dense(units = 10)) y <- dense(x) shape(y) ## shape(1, 10, 10, 10)"},{"path":"https://keras3.posit.co/dev/reference/layer_spectral_normalization.html","id":"reference","dir":"Reference","previous_headings":"","what":"Reference","title":"Performs spectral normalization on the weights of a target layer. — layer_spectral_normalization","text":"Spectral Normalization GAN.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_stft_spectrogram.html","id":null,"dir":"Reference","previous_headings":"","what":"Layer to compute the Short-Time Fourier Transform (STFT) on a 1D signal. — layer_stft_spectrogram","title":"Layer to compute the Short-Time Fourier Transform (STFT) on a 1D signal. — layer_stft_spectrogram","text":"layer computes Spectrograms input signal produce spectrogram. layers layer computes Spectrograms based Short-Time Fourier Transform (STFT) utilizing convolution kernels, allows parallelization GPUs trainable kernels fine-tuning support. layer allows different modes output (e.g., log-scaled magnitude, phase, power spectral density, etc.) provides flexibility windowing, padding, scaling options STFT calculation.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_stft_spectrogram.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Layer to compute the Short-Time Fourier Transform (STFT) on a 1D signal. — layer_stft_spectrogram","text":"","code":"layer_stft_spectrogram(   object,   mode = \"log\",   frame_length = 256L,   frame_step = NULL,   fft_length = NULL,   window = \"hann\",   periodic = FALSE,   scaling = \"density\",   padding = \"valid\",   expand_dims = FALSE,   data_format = NULL,   ... )"},{"path":"https://keras3.posit.co/dev/reference/layer_stft_spectrogram.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Layer to compute the Short-Time Fourier Transform (STFT) on a 1D signal. — layer_stft_spectrogram","text":"object Object compose layer . tensor, array, sequential model. mode String, output type spectrogram. Can one \"log\", \"magnitude\", \"psd\", \"real\", \"imag\", \"angle\", \"stft\". Defaults \"log\". frame_length Integer, length frame (window) STFT samples. Defaults 256. frame_step Integer, step size (hop length) consecutive frames. provided, defaults half frame_length. Defaults frame_length %/% 2. fft_length Integer, size frequency bins used Fast-Fourier Transform (FFT) apply frame. greater equal frame_length.  Recommended power two. Defaults smallest power two greater equal frame_length. window (String array_like), windowing function apply frame. Can \"hann\" (default), \"hamming\", custom window provided array_like. periodic Boolean, TRUE, window function treated periodic. Defaults FALSE. scaling String, type scaling applied window. Can \"density\", \"spectrum\", None. Default \"density\". padding String, padding strategy. Can \"valid\" \"\". Defaults \"valid\". expand_dims Boolean, TRUE, expand output spectrograms two dimensions compatible image models. Defaults FALSE. data_format String, either \"channels_last\" \"channels_first\". ordering dimensions inputs. \"channels_last\" corresponds inputs shape (batch, height, width, channels) \"channels_first\" corresponds inputs shape (batch, channels, height, weight). Defaults \"channels_last\". ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_stft_spectrogram.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Layer to compute the Short-Time Fourier Transform (STFT) on a 1D signal. — layer_stft_spectrogram","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_stft_spectrogram.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Layer to compute the Short-Time Fourier Transform (STFT) on a 1D signal. — layer_stft_spectrogram","text":"Apply non-trainable preprocessing layer 3 audio tracks 1 channel, 10 seconds sampled 16 kHz.     Apply trainable processing layer 3 stereo audio tracks 2 channels, 10 seconds sampled 16 kHz. initialized non-trainable layer, can trained jointly within model.     Similar last example, add extra dimension output image used image models. apply signal 3 input channels output image tensor, hence directly applicable image model.","code":"layer <- layer_stft_spectrogram(   mode = 'log',   frame_length = 256,   frame_step = 128, # 50% overlap   fft_length = 512,   window = \"hann\",   padding = \"valid\",   trainable = FALSE # non-trainable, preprocessing only) ) random_uniform(shape=c(3, 160000, 1)) |> layer() |> op_shape() ## shape(3, 1249, 257) layer <- layer_stft_spectrogram(   mode = 'log',   frame_length = 256,   frame_step = 128,   # 50% overlap   fft_length = 512,   window = \"hamming\", # hamming windowing function   padding = \"same\",   # padding to preserve the time dimension   trainable = TRUE,   # trainable, this is the default in keras ) random_uniform(shape=c(3, 160000, 2)) |> layer() |> op_shape() ## shape(3, 1250, 514) layer <- layer_stft_spectrogram(   mode = 'log',   frame_length = 256,   frame_step = 128,   fft_length = 512,   padding = \"same\",   expand_dims = TRUE  # this adds the extra dimension ) random_uniform(shape=c(3, 160000, 3)) |> layer() |> op_shape() ## shape(3, 1250, 257, 3)"},{"path":"https://keras3.posit.co/dev/reference/layer_stft_spectrogram.html","id":"raises","dir":"Reference","previous_headings":"","what":"Raises","title":"Layer to compute the Short-Time Fourier Transform (STFT) on a 1D signal. — layer_stft_spectrogram","text":"ValueError: invalid value provided \"mode\", \"scaling\", \"padding\", input arguments. TypeError: input data type one \"float16\", \"float32\", \"float64\".","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_stft_spectrogram.html","id":"input-shape","dir":"Reference","previous_headings":"","what":"Input Shape","title":"Layer to compute the Short-Time Fourier Transform (STFT) on a 1D signal. — layer_stft_spectrogram","text":"3D tensor shape (batch_size, time_length, input_channels), data_format==\"channels_last\", shape (batch_size, input_channels, time_length) data_format==\"channels_first\", time_length length input signal, input_channels number input channels. kernels applied channel independently.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_stft_spectrogram.html","id":"output-shape","dir":"Reference","previous_headings":"","what":"Output Shape","title":"Layer to compute the Short-Time Fourier Transform (STFT) on a 1D signal. — layer_stft_spectrogram","text":"data_format==\"channels_first\" && !expand_dims, 3D tensor: (batch_size, input_channels * freq_channels, new_time_length) data_format==\"channels_last\" && !expand_dims, 3D tensor: (batch_size, new_time_length, input_channels * freq_channels) data_format==\"channels_first\" && expand_dims, 4D tensor: (batch_size, input_channels, new_time_length, freq_channels) data_format==\"channels_last\" && expand_dims, 4D tensor: (batch_size, new_time_length, freq_channels, input_channels) new_time_length depends padding, freq_channels number FFT bins (fft_length %/% 2 + 1).","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_string_lookup.html","id":null,"dir":"Reference","previous_headings":"","what":"A preprocessing layer that maps strings to (possibly encoded) indices. — layer_string_lookup","title":"A preprocessing layer that maps strings to (possibly encoded) indices. — layer_string_lookup","text":"layer translates set arbitrary strings integer output via table-based vocabulary lookup. layer perform splitting transformation input strings. layer can split tokenize natural language, see layer_text_vectorization. vocabulary layer must either supplied construction learned via adapt(). adapt(), layer analyze data set, determine frequency individual strings tokens, create vocabulary . vocabulary capped size, frequent tokens used create vocabulary others treated --vocabulary (OOV). two possible output modes layer. output_mode \"int\", input strings converted (0-based) index vocabulary (integer). output_mode \"multi_hot\", \"count\", \"tf_idf\", input strings encoded array dimension corresponds element vocabulary. vocabulary can optionally contain mask token well OOV token (can optionally occupy multiple indices vocabulary, set num_oov_indices). position tokens vocabulary fixed. output_mode \"int\", vocabulary begin mask token (set), followed OOV indices, followed rest vocabulary. output_mode \"multi_hot\", \"count\", \"tf_idf\" vocabulary begin OOV indices instances mask token dropped. Note: layer uses TensorFlow internally. used part compiled computation graph model backend TensorFlow. can however used backend running eagerly. can also always used part input preprocessing pipeline backend (outside model ), recommend use layer. Note: layer safe use inside tf.data pipeline (independently backend using). Note: working layer outputs directly (e.g., passing outputs another layer, using lower-level operations like [ op_*): returned indices 0-based. However, default settings, first (0) index OOV token, returned indices offset 1 may appear 1-based.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_string_lookup.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"A preprocessing layer that maps strings to (possibly encoded) indices. — layer_string_lookup","text":"","code":"layer_string_lookup(   object,   max_tokens = NULL,   num_oov_indices = 1L,   mask_token = NULL,   oov_token = \"[UNK]\",   vocabulary = NULL,   idf_weights = NULL,   invert = FALSE,   output_mode = \"int\",   pad_to_max_tokens = FALSE,   sparse = FALSE,   encoding = \"utf-8\",   name = NULL,   ...,   vocabulary_dtype = NULL )"},{"path":"https://keras3.posit.co/dev/reference/layer_string_lookup.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"A preprocessing layer that maps strings to (possibly encoded) indices. — layer_string_lookup","text":"object Object compose layer . tensor, array, sequential model. max_tokens Maximum size vocabulary layer. specified adapting vocabulary setting pad_to_max_tokens=TRUE. NULL, cap size vocabulary. Note size includes OOV mask tokens. Defaults NULL. num_oov_indices number --vocabulary tokens use. value 1, OOV inputs modulated determine OOV value. value 0, OOV inputs cause error calling layer. Defaults 1. mask_token token represents masked inputs. output_mode \"int\", token included vocabulary mapped index 0. output modes, token appear vocabulary instances mask token input dropped. set NULL, mask term added. Defaults NULL. oov_token used invert TRUE. token return OOV indices. Defaults \"[UNK]\". vocabulary Optional. Either array integers string path text file. passing array, can pass list, list, 1D NumPy array, 1D tensor containing integer vocabulary terms. passing file path, file contain one line per term vocabulary. argument set, need adapt() layer. idf_weights valid output_mode \"tf_idf\". list, list, 1D NumPy array, 1D tensor length vocabulary, containing floating point inverse document frequency weights, multiplied per sample term counts final TF-IDF weight. vocabulary argument set, output_mode \"tf_idf\", argument must supplied. invert valid output_mode \"int\". TRUE, layer map indices vocabulary items instead mapping vocabulary items indices. Defaults FALSE. output_mode Specification output layer. Values can \"int\", \"one_hot\", \"multi_hot\", \"count\", \"tf_idf\" configuring layer follows: \"int\": Return vocabulary indices input tokens. \"one_hot\": Encodes individual element input array size vocabulary, containing 1 element index. last dimension size 1, encode dimension. last dimension size 1, append new dimension encoded output. \"multi_hot\": Encodes sample input single array size vocabulary, containing 1 vocabulary term present sample. Treats last dimension sample dimension, input shape (..., sample_length), output shape (..., num_tokens). \"count\": \"multi_hot\", int array contains count number times token index appeared sample. \"tf_idf\": \"multi_hot\", TF-IDF algorithm applied find value token slot. \"int\" output, shape input output supported. output modes, currently output rank 2 supported. Defaults \"int\". pad_to_max_tokens applicable output_mode \"multi_hot\", \"count\", \"tf_idf\". TRUE, output feature axis padded max_tokens even number unique tokens vocabulary less max_tokens, resulting tensor shape (batch_size, max_tokens) regardless vocabulary size. Defaults FALSE. sparse Boolean. applicable \"multi_hot\", \"count\", \"tf_idf\" output modes. supported TensorFlow backend. TRUE, returns SparseTensor instead dense Tensor. Defaults FALSE. encoding Optional. text encoding use interpret input strings. Defaults \"utf-8\". name String, name object ... forward/backward compatability. vocabulary_dtype dtype vocabulary terms, example \"int64\" \"int32\". Defaults \"int64\".","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_string_lookup.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"A preprocessing layer that maps strings to (possibly encoded) indices. — layer_string_lookup","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_string_lookup.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"A preprocessing layer that maps strings to (possibly encoded) indices. — layer_string_lookup","text":"Creating lookup layer known vocabulary example creates lookup layer pre-existing vocabulary.     Creating lookup layer adapted vocabulary example creates lookup layer generates vocabulary analyzing dataset.     Note OOV token \"[UNK]\" added vocabulary. remaining tokens sorted frequency (\"d\", 2 occurrences, first) inverse sort order.     Lookups multiple OOV indices example demonstrates use lookup layer multiple OOV indices.  layer created one OOV index, OOV values hashed number OOV buckets, distributing OOV values deterministic fashion across set.     Note output OOV value 'm' 0, output OOV value \"z\" 1. -vocab terms output index increased 1 earlier examples (maps 2, etc) order make space extra OOV value. One-hot output Configure layer output_mode='one_hot'. Note first num_oov_indices dimensions ont_hot encoding represent OOV values.     Multi-hot output Configure layer output_mode='multi_hot'. Note first num_oov_indices dimensions multi_hot encoding represent OOV values.     Token count output Configure layer output_mode='count'. multi_hot output, first num_oov_indices dimensions output represent OOV values.     TF-IDF output Configure layer output_mode=\"tf_idf\". multi_hot output, first num_oov_indices dimensions output represent OOV values. token bin output token_count * idf_weight, idf weights inverse document frequency weights per token. provided along vocabulary. Note idf_weight OOV values default average idf weights passed .     specify idf weights oov values, need pass entire vocabulary including leading oov token.     adapting layer \"tf_idf\" mode, input sample considered document, IDF weight per token calculated log(1 + num_documents / (1 + token_document_count)). Inverse lookup example demonstrates map indices strings using layer. (can also use adapt() inverse=TRUE, simplicity pass vocab example.)     Note first index correspond oov token default. Forward inverse lookup pairs example demonstrates use vocabulary standard lookup layer create inverse lookup layer.     example, input value \"z\" resulted output \"[UNK]\", since 1000 vocabulary - got represented OOV, OOV values returned \"[UNK]\" inverse layer. Also, note inverse work, must already set forward layer vocabulary either directly via adapt() calling get_vocabulary().","code":"vocab <- c(\"a\", \"b\", \"c\", \"d\") data <- rbind(c(\"a\", \"c\", \"d\"), c(\"d\", \"z\", \"b\")) layer <- layer_string_lookup(vocabulary=vocab) layer(data) ## tf.Tensor( ## [[1 3 4] ##  [4 0 2]], shape=(2, 3), dtype=int64) data <- rbind(c(\"a\", \"c\", \"d\"), c(\"d\", \"z\", \"b\")) layer <- layer_string_lookup() layer %>% adapt(data) get_vocabulary(layer) ## [1] \"[UNK]\" \"d\"     \"z\"     \"c\"     \"b\"     \"a\" data <- rbind(c(\"a\", \"c\", \"d\"), c(\"d\", \"z\", \"b\")) layer <- layer_string_lookup() layer %>% adapt(data) layer(data) ## tf.Tensor( ## [[5 3 1] ##  [1 2 4]], shape=(2, 3), dtype=int64) vocab <- c(\"a\", \"b\", \"c\", \"d\") data <- rbind(c(\"a\", \"c\", \"d\"), c(\"m\", \"z\", \"b\")) layer <- layer_string_lookup(vocabulary = vocab, num_oov_indices = 2) layer(data) ## tf.Tensor( ## [[2 4 5] ##  [0 1 3]], shape=(2, 3), dtype=int64) vocab <- c(\"a\", \"b\", \"c\", \"d\") data <- c(\"a\", \"b\", \"c\", \"d\", \"z\") layer <- layer_string_lookup(vocabulary = vocab, output_mode = 'one_hot') layer(data) ## tf.Tensor( ## [[0 1 0 0 0] ##  [0 0 1 0 0] ##  [0 0 0 1 0] ##  [0 0 0 0 1] ##  [1 0 0 0 0]], shape=(5, 5), dtype=int64) vocab <- c(\"a\", \"b\", \"c\", \"d\") data <- rbind(c(\"a\", \"c\", \"d\", \"d\"), c(\"d\", \"z\", \"b\", \"z\")) layer <- layer_string_lookup(vocabulary = vocab, output_mode = 'multi_hot') layer(data) ## tf.Tensor( ## [[0 1 0 1 1] ##  [1 0 1 0 1]], shape=(2, 5), dtype=int64) vocab <- c(\"a\", \"b\", \"c\", \"d\") data <- rbind(c(\"a\", \"c\", \"d\", \"d\"), c(\"d\", \"z\", \"b\", \"z\")) layer <- layer_string_lookup(vocabulary = vocab, output_mode = 'count') layer(data) ## tf.Tensor( ## [[0 1 0 1 2] ##  [2 0 1 0 1]], shape=(2, 5), dtype=int64) vocab <- c(\"a\", \"b\", \"c\", \"d\") idf_weights <- c(0.25, 0.75, 0.6, 0.4) data <- rbind(c(\"a\", \"c\", \"d\", \"d\"), c(\"d\", \"z\", \"b\", \"z\")) layer <- layer_string_lookup(output_mode = \"tf_idf\") layer %>% set_vocabulary(vocab, idf_weights=idf_weights) layer(data) ## tf.Tensor( ## [[0.   0.25 0.   0.6  0.8 ] ##  [1.   0.   0.75 0.   0.4 ]], shape=(2, 5), dtype=float32) vocab <- c(\"[UNK]\", \"a\", \"b\", \"c\", \"d\") idf_weights <- c(0.9, 0.25, 0.75, 0.6, 0.4) data <- rbind(c(\"a\", \"c\", \"d\", \"d\"), c(\"d\", \"z\", \"b\", \"z\")) layer <- layer_string_lookup(output_mode = \"tf_idf\") layer %>% set_vocabulary(vocab, idf_weights=idf_weights) layer(data) ## tf.Tensor( ## [[0.   0.25 0.   0.6  0.8 ] ##  [1.8  0.   0.75 0.   0.4 ]], shape=(2, 5), dtype=float32) vocab <- c(\"a\", \"b\", \"c\", \"d\") data <- rbind(c(1, 3, 4), c(4, 0, 2)) layer <- layer_string_lookup(vocabulary = vocab, invert = TRUE) layer(data) ## tf.Tensor( ## [[b'a' b'c' b'd'] ##  [b'd' b'[UNK]' b'b']], shape=(2, 3), dtype=string) vocab <- c(\"a\", \"b\", \"c\", \"d\") data <- rbind(c(\"a\", \"c\", \"d\"), c(\"d\", \"z\", \"b\")) layer <- layer_string_lookup(vocabulary = vocab) i_layer <- layer_string_lookup(vocabulary = vocab, invert = TRUE) int_data <- layer(data) i_layer(int_data) ## tf.Tensor( ## [[b'a' b'c' b'd'] ##  [b'd' b'[UNK]' b'b']], shape=(2, 3), dtype=string)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_subtract.html","id":null,"dir":"Reference","previous_headings":"","what":"Performs elementwise subtraction. — layer_subtract","title":"Performs elementwise subtraction. — layer_subtract","text":"takes input list tensors size 2 shape, returns single tensor (inputs[0] - inputs[1)) shape.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_subtract.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Performs elementwise subtraction. — layer_subtract","text":"","code":"layer_subtract(inputs, ...)"},{"path":"https://keras3.posit.co/dev/reference/layer_subtract.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Performs elementwise subtraction. — layer_subtract","text":"inputs layers combine ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_subtract.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Performs elementwise subtraction. — layer_subtract","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_subtract.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Performs elementwise subtraction. — layer_subtract","text":"Usage Keras model:","code":"input_shape <- c(2, 3, 4) x1 <- random_uniform(input_shape) x2 <- random_uniform(input_shape) y <- layer_subtract(list(x1, x2)) input1 <- layer_input(shape = 16) x1 <- layer_dense(input1, units = 8, activation = 'relu') input2 <- layer_input(shape = 32) x2 <- layer_dense(input2, units = 8, activation = 'relu') subtracted <- layer_subtract(list(x1, x2)) out <- layer_dense(subtracted, units = 4) model <- keras_model(inputs = list(input1, input2), outputs = out)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_text_vectorization.html","id":null,"dir":"Reference","previous_headings":"","what":"A preprocessing layer which maps text features to integer sequences. — layer_text_vectorization","title":"A preprocessing layer which maps text features to integer sequences. — layer_text_vectorization","text":"layer basic options managing text Keras model. transforms batch strings (one example = one string) either list token indices (one example = 1D tensor integer token indices) dense representation (one example = 1D tensor float values representing data example's tokens). layer meant handle natural language inputs. handle simple string inputs (categorical strings pre-tokenized strings) see layer_string_lookup(). vocabulary layer must either supplied construction learned via adapt(). layer adapted, analyze dataset, determine frequency individual string values, create vocabulary . vocabulary can unlimited size capped, depending configuration options layer; unique values input maximum vocabulary size, frequent terms used create vocabulary. processing example contains following steps: Standardize example (usually lowercasing + punctuation stripping) Split example substrings (usually words) Recombine substrings tokens (usually ngrams) Index tokens (associate unique int value token) Transform example using index, either vector ints dense float vector. notes passing callables customize splitting normalization layer: callable can passed Layer, want serialize object pass functions registered Keras serializables (see register_keras_serializable() details). using custom callable standardize, data received callable exactly passed layer. callable return tensor shape input. using custom callable split, data received callable 1st dimension squeezed - instead list(\"string split\", \"another string split\"), Callable see c(\"string split\", \"another string split\"). callable return tf.Tensor dtype string first dimension containing split tokens - example, see something like list(c(\"string\", \"\", \"split\"), c(\"another\", \"string\", \"\", \"split\")). Note: layer uses TensorFlow internally. used part compiled computation graph model backend TensorFlow. can however used backend running eagerly. can also always used part input preprocessing pipeline backend (outside model ), recommend use layer. Note: layer safe use inside tf.data pipeline (independently backend using).","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_text_vectorization.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"A preprocessing layer which maps text features to integer sequences. — layer_text_vectorization","text":"","code":"layer_text_vectorization(   object,   max_tokens = NULL,   standardize = \"lower_and_strip_punctuation\",   split = \"whitespace\",   ngrams = NULL,   output_mode = \"int\",   output_sequence_length = NULL,   pad_to_max_tokens = FALSE,   vocabulary = NULL,   idf_weights = NULL,   sparse = FALSE,   ragged = FALSE,   encoding = \"utf-8\",   name = NULL,   ... )  get_vocabulary(object, include_special_tokens = TRUE)  set_vocabulary(object, vocabulary, idf_weights = NULL, ...)"},{"path":"https://keras3.posit.co/dev/reference/layer_text_vectorization.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"A preprocessing layer which maps text features to integer sequences. — layer_text_vectorization","text":"object Object compose layer . tensor, array, sequential model. max_tokens Maximum size vocabulary layer. specified adapting vocabulary setting pad_to_max_tokens=TRUE. Note vocabulary contains 1 OOV token, effective number tokens (max_tokens - 1 - (1 output_mode == \"int\" else 0)). standardize Optional specification standardization apply input text. Values can : NULL: standardization. \"lower_and_strip_punctuation\": Text lowercased punctuation removed. \"lower\": Text lowercased. \"strip_punctuation\": punctuation removed. Callable: Inputs passed callable function, standardized returned. split Optional specification splitting input text. Values can : NULL: splitting. \"whitespace\": Split whitespace. \"character\": Split unicode character. Callable: Standardized inputs passed callable function, split returned. ngrams Optional specification ngrams create possibly-split input text. Values can NULL, integer list integers; passing integer create ngrams integer, passing list integers create ngrams specified values list. Passing NULL means ngrams created. output_mode Optional specification output layer. Values can \"int\", \"multi_hot\", \"count\" \"tf_idf\", configuring layer follows: \"int\": Outputs integer indices, one integer index per split string token. output_mode == \"int\", 0 reserved masked locations; reduces vocab size max_tokens - 2 instead max_tokens - 1. \"multi_hot\": Outputs single int array per batch, either vocab_size max_tokens size, containing 1s elements token mapped index exists least batch item. \"count\": Like \"multi_hot\", int array contains count number times token index appeared batch item. \"tf_idf\": Like \"multi_hot\", TF-IDF algorithm applied find value token slot. \"int\" output, shape input output supported. output modes, currently rank 1 inputs (rank 2 outputs splitting) supported. output_sequence_length valid INT mode. set, output time dimension padded truncated exactly output_sequence_length values, resulting tensor shape (batch_size, output_sequence_length) regardless many tokens resulted splitting step. Defaults NULL. ragged TRUE output_sequence_length may still truncate output. pad_to_max_tokens valid  \"multi_hot\", \"count\", \"tf_idf\" modes. TRUE, output feature axis padded max_tokens even number unique tokens vocabulary less max_tokens, resulting tensor shape (batch_size, max_tokens) regardless vocabulary size. Defaults FALSE. vocabulary Optional. Either array strings string path text file. passing array, can pass list, list, 1D NumPy array, 1D tensor containing string vocabulary terms. passing file path, file contain one line per term vocabulary. argument set, need adapt() layer. idf_weights R vector, 1D numpy array, 1D tensor inverse document frequency weights equal length vocabulary. Must set output_mode \"tf_idf\". set otherwise. sparse Boolean. applicable \"multi_hot\", \"count\", \"tf_idf\" output modes. supported TensorFlow backend. TRUE, returns SparseTensor instead dense Tensor. Defaults FALSE. ragged Boolean. applicable \"int\" output mode. supported TensorFlow backend. TRUE, returns RaggedTensor instead dense Tensor, sequence may different length string splitting. Defaults FALSE. encoding Optional. text encoding use interpret input strings. Defaults \"utf-8\". name String, name object ... forward/backward compatability. include_special_tokens TRUE, returned vocabulary include padding OOV tokens, term's index vocabulary equal term's index calling layer. FALSE, returned vocabulary include padding OOV tokens.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_text_vectorization.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"A preprocessing layer which maps text features to integer sequences. — layer_text_vectorization","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_text_vectorization.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"A preprocessing layer which maps text features to integer sequences. — layer_text_vectorization","text":"example instantiates TextVectorization layer lowercases text, splits whitespace, strips punctuation, outputs integer vocab indices.         example instantiates TextVectorization layer passing list vocabulary terms layer's initialize() method.","code":"max_tokens <- 5000  # Maximum vocab size. max_len <- 4  # Sequence length to pad the outputs to. # Create the layer. vectorize_layer <- layer_text_vectorization(     max_tokens = max_tokens,     output_mode = 'int',     output_sequence_length = max_len) # Now that the vocab layer has been created, call `adapt` on the # list of strings to create the vocabulary. vectorize_layer %>% adapt(c(\"foo bar\", \"bar baz\", \"baz bada boom\")) # Now, the layer can map strings to integers -- you can use an # embedding layer to map these integers to learned embeddings. input_data <- rbind(\"foo qux bar\", \"qux baz\") vectorize_layer(input_data) ## tf.Tensor( ## [[4 1 3 0] ##  [1 2 0 0]], shape=(2, 4), dtype=int64) vocab_data <- c(\"earth\", \"wind\", \"and\", \"fire\") max_len <- 4  # Sequence length to pad the outputs to. # Create the layer, passing the vocab directly. You can also pass the # vocabulary arg a path to a file containing one vocabulary word per # line. vectorize_layer <- layer_text_vectorization(     max_tokens = max_tokens,     output_mode = 'int',     output_sequence_length = max_len,     vocabulary = vocab_data) # Because we've passed the vocabulary directly, we don't need to adapt # the layer - the vocabulary is already set. The vocabulary contains the # padding token ('') and OOV token ('[UNK]') # as well as the passed tokens. vectorize_layer %>% get_vocabulary() ## [1] \"\"      \"[UNK]\" \"earth\" \"wind\"  \"and\"   \"fire\" # ['', '[UNK]', 'earth', 'wind', 'and', 'fire']"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_tfsm.html","id":null,"dir":"Reference","previous_headings":"","what":"Reload a Keras model/layer that was saved via export_savedmodel(). — layer_tfsm","title":"Reload a Keras model/layer that was saved via export_savedmodel(). — layer_tfsm","text":"Reload Keras model/layer saved via export_savedmodel().","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_tfsm.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Reload a Keras model/layer that was saved via export_savedmodel(). — layer_tfsm","text":"","code":"layer_tfsm(   object,   filepath,   call_endpoint = \"serve\",   call_training_endpoint = NULL,   trainable = TRUE,   name = NULL,   dtype = NULL )"},{"path":"https://keras3.posit.co/dev/reference/layer_tfsm.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Reload a Keras model/layer that was saved via export_savedmodel(). — layer_tfsm","text":"object Object compose layer . tensor, array, sequential model. filepath string, path SavedModel. call_endpoint Name endpoint use call() method reloaded layer. SavedModel created via export_savedmodel(), default endpoint name 'serve'. cases may named 'serving_default'. call_training_endpoint see description trainable see description name String, name object dtype datatype (e.g., \"float32\").","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_tfsm.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Reload a Keras model/layer that was saved via export_savedmodel(). — layer_tfsm","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_tfsm.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Reload a Keras model/layer that was saved via export_savedmodel(). — layer_tfsm","text":"reloaded object can used like regular Keras layer, supports training/fine-tuning trainable weights. Note reloaded object retains none internal structure custom methods original object – brand new layer created around saved function. Limitations: call endpoints single inputs tensor argument (may optionally named list/list tensors) supported. endpoints multiple separate input tensor arguments, consider subclassing layer_tfsm implementing call() method custom signature. need training-time behavior differ inference-time behavior (.e. need reloaded object support training=TRUE argument __call__()), make sure training-time call function saved standalone endpoint artifact, provide name layer_tfsm via call_training_endpoint argument.","code":"model <- keras_model_sequential(input_shape = c(784)) |> layer_dense(10) model |> export_savedmodel(\"path/to/artifact\") ## Saved artifact at 'path/to/artifact'. The following endpoints are available: ## ## * Endpoint 'serve' ##   args_0 (POSITIONAL_ONLY): TensorSpec(shape=(None, 784), dtype=tf.float32, name='keras_tensor') ## Output Type: ##   TensorSpec(shape=(None, 10), dtype=tf.float32, name=None) ## Captures: ##   133047893322704: TensorSpec(shape=(), dtype=tf.resource, name=None) ##   133047893325008: TensorSpec(shape=(), dtype=tf.resource, name=None) reloaded_layer <- layer_tfsm(filepath = \"path/to/artifact\") input <- random_normal(c(2, 784)) output <- reloaded_layer(input) stopifnot(all.equal(as.array(output), as.array(model(input))))"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_time_distributed.html","id":null,"dir":"Reference","previous_headings":"","what":"This wrapper allows to apply a layer to every temporal slice of an input. — layer_time_distributed","title":"This wrapper allows to apply a layer to every temporal slice of an input. — layer_time_distributed","text":"Every input least 3D, dimension index one first input considered temporal dimension. Consider batch 32 video samples, sample 128x128 RGB image channels_last data format, across 10 timesteps. batch input shape (32, 10, 128, 128, 3). can use TimeDistributed apply Conv2D layer 10 timesteps, independently:     layer_time_distributed applies instance layer_conv2d timestamps, set weights used timestamp.","code":"inputs <- keras_input(shape = c(10, 128, 128, 3), batch_size = 32) conv_2d_layer <- layer_conv_2d(filters = 64, kernel_size = c(3, 3)) outputs <- layer_time_distributed(inputs, layer = conv_2d_layer) shape(outputs) ## shape(32, 10, 126, 126, 64)"},{"path":"https://keras3.posit.co/dev/reference/layer_time_distributed.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"This wrapper allows to apply a layer to every temporal slice of an input. — layer_time_distributed","text":"","code":"layer_time_distributed(object, layer, ...)"},{"path":"https://keras3.posit.co/dev/reference/layer_time_distributed.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"This wrapper allows to apply a layer to every temporal slice of an input. — layer_time_distributed","text":"object Object compose layer . tensor, array, sequential model. layer Layer instance. ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_time_distributed.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"This wrapper allows to apply a layer to every temporal slice of an input. — layer_time_distributed","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_time_distributed.html","id":"call-arguments","dir":"Reference","previous_headings":"","what":"Call Arguments","title":"This wrapper allows to apply a layer to every temporal slice of an input. — layer_time_distributed","text":"inputs: Input tensor shape (batch, time, ...) nested tensors, shape (batch, time, ...). training: Boolean indicating whether layer behave training mode inference mode. argument passed wrapped layer (layer supports argument). mask: Binary tensor shape (samples, timesteps) indicating whether given timestep masked. argument passed wrapped layer (layer supports argument).","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_torch_module_wrapper.html","id":null,"dir":"Reference","previous_headings":"","what":"Torch module wrapper layer. — layer_torch_module_wrapper","title":"Torch module wrapper layer. — layer_torch_module_wrapper","text":"layer_torch_module_wrapper wrapper class can turn torch.nn.Module Keras layer, particular making parameters trackable Keras. layer_torch_module_wrapper() compatible PyTorch backend used TensorFlow JAX backends.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_torch_module_wrapper.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Torch module wrapper layer. — layer_torch_module_wrapper","text":"","code":"layer_torch_module_wrapper(object, module, name = NULL, ...)"},{"path":"https://keras3.posit.co/dev/reference/layer_torch_module_wrapper.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Torch module wrapper layer. — layer_torch_module_wrapper","text":"object Object compose layer . tensor, array, sequential model. module torch.nn.Module instance. LazyModule instance, parameters must initialized passing instance layer_torch_module_wrapper (e.g. calling ). name name layer (string). ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_torch_module_wrapper.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Torch module wrapper layer. — layer_torch_module_wrapper","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_torch_module_wrapper.html","id":"example","dir":"Reference","previous_headings":"","what":"Example","title":"Torch module wrapper layer. — layer_torch_module_wrapper","text":"example layer_torch_module_wrapper() can used vanilla PyTorch modules.","code":"# reticulate::py_install( #   packages = c(\"torch\", \"torchvision\", \"torchaudio\"), #   envname = \"r-keras\", #   pip_options = c(\"--index-url https://download.pytorch.org/whl/cpu\") # ) library(keras3) use_backend(\"torch\") torch <- reticulate::import(\"torch\") nn <- reticulate::import(\"torch.nn\") nnf <- reticulate::import(\"torch.nn.functional\")  Classifier(keras$Model) \\%py_class\\% {   initialize <- function(...) {     super$initialize(...)      self$conv1 <- layer_torch_module_wrapper(module = nn$Conv2d(       in_channels = 1L,       out_channels = 32L,       kernel_size = tuple(3L, 3L)     ))     self$conv2 <- layer_torch_module_wrapper(module = nn$Conv2d(       in_channels = 32L,       out_channels = 64L,       kernel_size = tuple(3L, 3L)     ))     self$pool <- nn$MaxPool2d(kernel_size = tuple(2L, 2L))     self$flatten <- nn$Flatten()     self$dropout <- nn$Dropout(p = 0.5)     self$fc <-       layer_torch_module_wrapper(module = nn$Linear(1600L, 10L))   }    call <- function(inputs) {     x <- nnf$relu(self$conv1(inputs))     x <- self$pool(x)     x <- nnf$relu(self$conv2(x))     x <- self$pool(x)     x <- self$flatten(x)     x <- self$dropout(x)     x <- self$fc(x)     nnf$softmax(x, dim = 1L)   } } model <- Classifier() model$build(shape(1, 28, 28)) cat(\"Output shape:\", format(shape(model(torch$ones(1L, 1L, 28L, 28L)))))  model |> compile(loss = \"sparse_categorical_crossentropy\",                  optimizer = \"adam\",                  metrics = \"accuracy\") model |> fit(train_loader, epochs = 5)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_unit_normalization.html","id":null,"dir":"Reference","previous_headings":"","what":"Unit normalization layer. — layer_unit_normalization","title":"Unit normalization layer. — layer_unit_normalization","text":"Normalize batch inputs input batch L2 norm equal 1 (across axes specified axis).","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_unit_normalization.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Unit normalization layer. — layer_unit_normalization","text":"","code":"layer_unit_normalization(object, axis = -1L, ...)"},{"path":"https://keras3.posit.co/dev/reference/layer_unit_normalization.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Unit normalization layer. — layer_unit_normalization","text":"object Object compose layer . tensor, array, sequential model. axis Integer list. axis axes normalize across. Typically, features axis axes. left-axes typically batch axis axes. -1 last dimension input. Defaults -1. ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_unit_normalization.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Unit normalization layer. — layer_unit_normalization","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_unit_normalization.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Unit normalization layer. — layer_unit_normalization","text":"","code":"data <- 1:6 |> op_array(\"float32\") |> op_reshape(c(2, 3)) normalized_data <- layer_unit_normalization(data) op_sum(normalized_data[1, ]^2) ## tf.Tensor(0.9999999, shape=(), dtype=float32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_upsampling_1d.html","id":null,"dir":"Reference","previous_headings":"","what":"Upsampling layer for 1D inputs. — layer_upsampling_1d","title":"Upsampling layer for 1D inputs. — layer_upsampling_1d","text":"Repeats temporal step size times along time axis.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_upsampling_1d.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Upsampling layer for 1D inputs. — layer_upsampling_1d","text":"","code":"layer_upsampling_1d(object, size = 2L, ...)"},{"path":"https://keras3.posit.co/dev/reference/layer_upsampling_1d.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Upsampling layer for 1D inputs. — layer_upsampling_1d","text":"object Object compose layer . tensor, array, sequential model. size Integer. Upsampling factor. ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_upsampling_1d.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Upsampling layer for 1D inputs. — layer_upsampling_1d","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_upsampling_1d.html","id":"example","dir":"Reference","previous_headings":"","what":"Example","title":"Upsampling layer for 1D inputs. — layer_upsampling_1d","text":"","code":"input_shape <- c(2, 2, 3) x <- seq_len(prod(input_shape)) %>% op_reshape(input_shape) x ## tf.Tensor( ## [[[ 1  2  3] ##   [ 4  5  6]] ## ##  [[ 7  8  9] ##   [10 11 12]]], shape=(2, 2, 3), dtype=int32) y <- layer_upsampling_1d(x, size = 2) y ## tf.Tensor( ## [[[ 1  2  3] ##   [ 1  2  3] ##   [ 4  5  6] ##   [ 4  5  6]] ## ##  [[ 7  8  9] ##   [ 7  8  9] ##   [10 11 12] ##   [10 11 12]]], shape=(2, 4, 3), dtype=int32)"},{"path":"https://keras3.posit.co/dev/reference/layer_upsampling_1d.html","id":"input-shape","dir":"Reference","previous_headings":"","what":"Input Shape","title":"Upsampling layer for 1D inputs. — layer_upsampling_1d","text":"3D tensor shape: (batch_size, steps, features).","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_upsampling_1d.html","id":"output-shape","dir":"Reference","previous_headings":"","what":"Output Shape","title":"Upsampling layer for 1D inputs. — layer_upsampling_1d","text":"3D tensor shape: (batch_size, upsampled_steps, features).","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_upsampling_2d.html","id":null,"dir":"Reference","previous_headings":"","what":"Upsampling layer for 2D inputs. — layer_upsampling_2d","title":"Upsampling layer for 2D inputs. — layer_upsampling_2d","text":"implementation uses interpolative resizing, given resize method (specified interpolation argument). Use interpolation=nearest repeat rows columns data.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_upsampling_2d.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Upsampling layer for 2D inputs. — layer_upsampling_2d","text":"","code":"layer_upsampling_2d(   object,   size = list(2L, 2L),   data_format = NULL,   interpolation = \"nearest\",   ... )"},{"path":"https://keras3.posit.co/dev/reference/layer_upsampling_2d.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Upsampling layer for 2D inputs. — layer_upsampling_2d","text":"object Object compose layer . tensor, array, sequential model. size Int, list 2 integers. upsampling factors rows columns. data_format string, one \"channels_last\" (default) \"channels_first\". ordering dimensions inputs. \"channels_last\" corresponds inputs shape (batch_size, height, width, channels) \"channels_first\" corresponds inputs shape (batch_size, channels, height, width). unspecified, uses image_data_format value found Keras config file ~/.keras/keras.json (exists) else \"channels_last\". Defaults \"channels_last\". interpolation string, one \"bicubic\", \"bilinear\", \"lanczos3\", \"lanczos5\", \"nearest\". ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_upsampling_2d.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Upsampling layer for 2D inputs. — layer_upsampling_2d","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_upsampling_2d.html","id":"example","dir":"Reference","previous_headings":"","what":"Example","title":"Upsampling layer for 2D inputs. — layer_upsampling_2d","text":"","code":"input_shape <- c(2, 2, 1, 3) x <- op_reshape(seq_len(prod(input_shape)), input_shape) print(x) ## tf.Tensor( ## [[[[ 1  2  3]] ## ##   [[ 4  5  6]]] ## ## ##  [[[ 7  8  9]] ## ##   [[10 11 12]]]], shape=(2, 2, 1, 3), dtype=int32) y <- layer_upsampling_2d(x, size = c(1, 2)) print(y) ## tf.Tensor( ## [[[[ 1  2  3] ##    [ 1  2  3]] ## ##   [[ 4  5  6] ##    [ 4  5  6]]] ## ## ##  [[[ 7  8  9] ##    [ 7  8  9]] ## ##   [[10 11 12] ##    [10 11 12]]]], shape=(2, 2, 2, 3), dtype=int32)"},{"path":"https://keras3.posit.co/dev/reference/layer_upsampling_2d.html","id":"input-shape","dir":"Reference","previous_headings":"","what":"Input Shape","title":"Upsampling layer for 2D inputs. — layer_upsampling_2d","text":"4D tensor shape: data_format \"channels_last\": (batch_size, rows, cols, channels) data_format \"channels_first\": (batch_size, channels, rows, cols)","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_upsampling_2d.html","id":"output-shape","dir":"Reference","previous_headings":"","what":"Output Shape","title":"Upsampling layer for 2D inputs. — layer_upsampling_2d","text":"4D tensor shape: data_format \"channels_last\": (batch_size, upsampled_rows, upsampled_cols, channels) data_format \"channels_first\": (batch_size, channels, upsampled_rows, upsampled_cols)","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_upsampling_3d.html","id":null,"dir":"Reference","previous_headings":"","what":"Upsampling layer for 3D inputs. — layer_upsampling_3d","title":"Upsampling layer for 3D inputs. — layer_upsampling_3d","text":"Repeats 1st, 2nd 3rd dimensions data size[0], size[1] size[2] respectively.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_upsampling_3d.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Upsampling layer for 3D inputs. — layer_upsampling_3d","text":"","code":"layer_upsampling_3d(object, size = list(2L, 2L, 2L), data_format = NULL, ...)"},{"path":"https://keras3.posit.co/dev/reference/layer_upsampling_3d.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Upsampling layer for 3D inputs. — layer_upsampling_3d","text":"object Object compose layer . tensor, array, sequential model. size Int, list 3 integers. upsampling factors dim1, dim2 dim3. data_format string, one \"channels_last\" (default) \"channels_first\". ordering dimensions inputs. \"channels_last\" corresponds inputs shape (batch_size, spatial_dim1, spatial_dim2, spatial_dim3, channels) \"channels_first\" corresponds inputs shape (batch_size, channels, spatial_dim1, spatial_dim2, spatial_dim3). unspecified, uses image_data_format value found Keras config file ~/.keras/keras.json (exists) else \"channels_last\". Defaults \"channels_last\". ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_upsampling_3d.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Upsampling layer for 3D inputs. — layer_upsampling_3d","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_upsampling_3d.html","id":"example","dir":"Reference","previous_headings":"","what":"Example","title":"Upsampling layer for 3D inputs. — layer_upsampling_3d","text":"","code":"input_shape <- c(2, 1, 2, 1, 3) x <- array(1, dim = input_shape) y <- layer_upsampling_3d(x, size = c(2, 2, 2)) shape(y) ## shape(2, 2, 4, 2, 3)"},{"path":"https://keras3.posit.co/dev/reference/layer_upsampling_3d.html","id":"input-shape","dir":"Reference","previous_headings":"","what":"Input Shape","title":"Upsampling layer for 3D inputs. — layer_upsampling_3d","text":"5D tensor shape: data_format \"channels_last\": (batch_size, dim1, dim2, dim3, channels) data_format \"channels_first\": (batch_size, channels, dim1, dim2, dim3)","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_upsampling_3d.html","id":"output-shape","dir":"Reference","previous_headings":"","what":"Output Shape","title":"Upsampling layer for 3D inputs. — layer_upsampling_3d","text":"5D tensor shape: data_format \"channels_last\": (batch_size, upsampled_dim1, upsampled_dim2, upsampled_dim3, channels) data_format \"channels_first\": (batch_size, channels, upsampled_dim1, upsampled_dim2, upsampled_dim3)","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_zero_padding_1d.html","id":null,"dir":"Reference","previous_headings":"","what":"Zero-padding layer for 1D input (e.g. temporal sequence). — layer_zero_padding_1d","title":"Zero-padding layer for 1D input (e.g. temporal sequence). — layer_zero_padding_1d","text":"Zero-padding layer 1D input (e.g. temporal sequence).","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_zero_padding_1d.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Zero-padding layer for 1D input (e.g. temporal sequence). — layer_zero_padding_1d","text":"","code":"layer_zero_padding_1d(object, padding = 1L, data_format = NULL, ...)"},{"path":"https://keras3.posit.co/dev/reference/layer_zero_padding_1d.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Zero-padding layer for 1D input (e.g. temporal sequence). — layer_zero_padding_1d","text":"object Object compose layer . tensor, array, sequential model. padding Int, list int (length 2). int: many zeros add beginning end padding dimension (axis 2). list 2 ints: many zeros add beginning end padding dimension ((left_pad, right_pad)). data_format string, one \"channels_last\" (default) \"channels_first\". ordering dimensions inputs. \"channels_last\" corresponds inputs shape (batch_size, axis_to_pad, channels) \"channels_first\" corresponds inputs shape (batch_size, channels, axis_to_pad). unspecified, uses image_data_format value found Keras config file ~/.keras/keras.json (exists). Defaults \"channels_last\". ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_zero_padding_1d.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Zero-padding layer for 1D input (e.g. temporal sequence). — layer_zero_padding_1d","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_zero_padding_1d.html","id":"example","dir":"Reference","previous_headings":"","what":"Example","title":"Zero-padding layer for 1D input (e.g. temporal sequence). — layer_zero_padding_1d","text":"","code":"input_shape <- c(2, 2, 3) x <- op_reshape(seq_len(prod(input_shape)), input_shape) x ## tf.Tensor( ## [[[ 1  2  3] ##   [ 4  5  6]] ## ##  [[ 7  8  9] ##   [10 11 12]]], shape=(2, 2, 3), dtype=int32) y <- layer_zero_padding_1d(x, padding = 2) y ## tf.Tensor( ## [[[ 0  0  0] ##   [ 0  0  0] ##   [ 1  2  3] ##   [ 4  5  6] ##   [ 0  0  0] ##   [ 0  0  0]] ## ##  [[ 0  0  0] ##   [ 0  0  0] ##   [ 7  8  9] ##   [10 11 12] ##   [ 0  0  0] ##   [ 0  0  0]]], shape=(2, 6, 3), dtype=int32)"},{"path":"https://keras3.posit.co/dev/reference/layer_zero_padding_1d.html","id":"input-shape","dir":"Reference","previous_headings":"","what":"Input Shape","title":"Zero-padding layer for 1D input (e.g. temporal sequence). — layer_zero_padding_1d","text":"3D tensor shape: data_format \"channels_last\": (batch_size, axis_to_pad, features) data_format \"channels_first\": (batch_size, features, axis_to_pad)","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_zero_padding_1d.html","id":"output-shape","dir":"Reference","previous_headings":"","what":"Output Shape","title":"Zero-padding layer for 1D input (e.g. temporal sequence). — layer_zero_padding_1d","text":"3D tensor shape: data_format \"channels_last\": (batch_size, padded_axis, features) data_format \"channels_first\": (batch_size, features, padded_axis)","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_zero_padding_2d.html","id":null,"dir":"Reference","previous_headings":"","what":"Zero-padding layer for 2D input (e.g. picture). — layer_zero_padding_2d","title":"Zero-padding layer for 2D input (e.g. picture). — layer_zero_padding_2d","text":"layer can add rows columns zeros top, bottom, left right side image tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_zero_padding_2d.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Zero-padding layer for 2D input (e.g. picture). — layer_zero_padding_2d","text":"","code":"layer_zero_padding_2d(object, padding = list(1L, 1L), data_format = NULL, ...)"},{"path":"https://keras3.posit.co/dev/reference/layer_zero_padding_2d.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Zero-padding layer for 2D input (e.g. picture). — layer_zero_padding_2d","text":"object Object compose layer . tensor, array, sequential model. padding Int, list 2 ints, list 2 lists 2 ints. int: symmetric padding applied height width. list 2 ints: interpreted two different symmetric padding values height width: (symmetric_height_pad, symmetric_width_pad). list 2 lists 2 ints: interpreted ((top_pad, bottom_pad), (left_pad, right_pad)). data_format string, one \"channels_last\" (default) \"channels_first\". ordering dimensions inputs. \"channels_last\" corresponds inputs shape (batch_size, height, width, channels) \"channels_first\" corresponds inputs shape (batch_size, channels, height, width). unspecified, uses image_data_format value found Keras config file ~/.keras/keras.json (exists). Defaults \"channels_last\". ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_zero_padding_2d.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Zero-padding layer for 2D input (e.g. picture). — layer_zero_padding_2d","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_zero_padding_2d.html","id":"example","dir":"Reference","previous_headings":"","what":"Example","title":"Zero-padding layer for 2D input (e.g. picture). — layer_zero_padding_2d","text":"","code":"input_shape <- c(1, 1, 2, 2) x <- op_reshape(seq_len(prod(input_shape)), input_shape) x ## tf.Tensor( ## [[[[1 2] ##    [3 4]]]], shape=(1, 1, 2, 2), dtype=int32) y <- layer_zero_padding_2d(x, padding = 1) y ## tf.Tensor( ## [[[[0 0] ##    [0 0] ##    [0 0] ##    [0 0]] ## ##   [[0 0] ##    [1 2] ##    [3 4] ##    [0 0]] ## ##   [[0 0] ##    [0 0] ##    [0 0] ##    [0 0]]]], shape=(1, 3, 4, 2), dtype=int32)"},{"path":"https://keras3.posit.co/dev/reference/layer_zero_padding_2d.html","id":"input-shape","dir":"Reference","previous_headings":"","what":"Input Shape","title":"Zero-padding layer for 2D input (e.g. picture). — layer_zero_padding_2d","text":"4D tensor shape: data_format \"channels_last\": (batch_size, height, width, channels) data_format \"channels_first\": (batch_size, channels, height, width)","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_zero_padding_2d.html","id":"output-shape","dir":"Reference","previous_headings":"","what":"Output Shape","title":"Zero-padding layer for 2D input (e.g. picture). — layer_zero_padding_2d","text":"4D tensor shape: data_format \"channels_last\": (batch_size, padded_height, padded_width, channels) data_format \"channels_first\": (batch_size, channels, padded_height, padded_width)","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_zero_padding_3d.html","id":null,"dir":"Reference","previous_headings":"","what":"Zero-padding layer for 3D data (spatial or spatio-temporal). — layer_zero_padding_3d","title":"Zero-padding layer for 3D data (spatial or spatio-temporal). — layer_zero_padding_3d","text":"Zero-padding layer 3D data (spatial spatio-temporal).","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_zero_padding_3d.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Zero-padding layer for 3D data (spatial or spatio-temporal). — layer_zero_padding_3d","text":"","code":"layer_zero_padding_3d(   object,   padding = list(list(1L, 1L), list(1L, 1L), list(1L, 1L)),   data_format = NULL,   ... )"},{"path":"https://keras3.posit.co/dev/reference/layer_zero_padding_3d.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Zero-padding layer for 3D data (spatial or spatio-temporal). — layer_zero_padding_3d","text":"object Object compose layer . tensor, array, sequential model. padding Int, list 3 ints, list 3 lists 2 ints. int: symmetric padding applied depth, height, width. list 3 ints: interpreted three different symmetric padding values depth, height, width: (symmetric_dim1_pad, symmetric_dim2_pad, symmetric_dim3_pad). list 3 lists 2 ints: interpreted ((left_dim1_pad, right_dim1_pad), (left_dim2_pad, right_dim2_pad), (left_dim3_pad, right_dim3_pad)). data_format string, one \"channels_last\" (default) \"channels_first\". ordering dimensions inputs. \"channels_last\" corresponds inputs shape (batch_size, spatial_dim1, spatial_dim2, spatial_dim3, channels) \"channels_first\" corresponds inputs shape (batch_size, channels, spatial_dim1, spatial_dim2, spatial_dim3). unspecified, uses image_data_format value found Keras config file ~/.keras/keras.json (exists). Defaults \"channels_last\". ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_zero_padding_3d.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Zero-padding layer for 3D data (spatial or spatio-temporal). — layer_zero_padding_3d","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_zero_padding_3d.html","id":"example","dir":"Reference","previous_headings":"","what":"Example","title":"Zero-padding layer for 3D data (spatial or spatio-temporal). — layer_zero_padding_3d","text":"","code":"input_shape <- c(1, 1, 2, 2, 3) x <- op_reshape(seq_len(prod(input_shape)), input_shape) x ## tf.Tensor( ## [[[[[ 1  2  3] ##     [ 4  5  6]] ## ##    [[ 7  8  9] ##     [10 11 12]]]]], shape=(1, 1, 2, 2, 3), dtype=int32) y <- layer_zero_padding_3d(x, padding = 2) shape(y) ## shape(1, 5, 6, 6, 3)"},{"path":"https://keras3.posit.co/dev/reference/layer_zero_padding_3d.html","id":"input-shape","dir":"Reference","previous_headings":"","what":"Input Shape","title":"Zero-padding layer for 3D data (spatial or spatio-temporal). — layer_zero_padding_3d","text":"5D tensor shape: data_format \"channels_last\": (batch_size, first_axis_to_pad, second_axis_to_pad, third_axis_to_pad, depth) data_format \"channels_first\": (batch_size, depth, first_axis_to_pad, second_axis_to_pad, third_axis_to_pad)","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_zero_padding_3d.html","id":"output-shape","dir":"Reference","previous_headings":"","what":"Output Shape","title":"Zero-padding layer for 3D data (spatial or spatio-temporal). — layer_zero_padding_3d","text":"5D tensor shape: data_format \"channels_last\": (batch_size, first_padded_axis, second_padded_axis, third_axis_to_pad, depth) data_format \"channels_first\": (batch_size, depth, first_padded_axis, second_padded_axis, third_axis_to_pad)","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/learning_rate_schedule_cosine_decay.html","id":null,"dir":"Reference","previous_headings":"","what":"A LearningRateSchedule that uses a cosine decay with optional warmup. — learning_rate_schedule_cosine_decay","title":"A LearningRateSchedule that uses a cosine decay with optional warmup. — learning_rate_schedule_cosine_decay","text":"See Loshchilov & Hutter, ICLR2016, SGDR: Stochastic Gradient Descent Warm Restarts. idea linear warmup learning rate, see Goyal et al.. begin training model, often want initial increase learning rate followed decay. warmup_target int, schedule applies linear increase per optimizer step learning rate initial_learning_rate warmup_target duration warmup_steps. Afterwards, applies cosine decay function taking learning rate warmup_target alpha duration decay_steps. warmup_target NULL skip warmup decay take learning rate initial_learning_rate alpha. requires step value  compute learning rate. can just pass backend variable increment training step. schedule 1-arg callable produces warmup followed decayed learning rate passed current optimizer step. can useful changing learning rate value across different invocations optimizer functions. warmup computed :   decay computed :   Example usage without warmup:   Example usage warmup:   can pass schedule directly optimizer learning rate. learning rate schedule also serializable deserializable using keras$optimizers$schedules$serialize keras$optimizers$schedules$deserialize.","code":"warmup_learning_rate <- function(step) {   completed_fraction <- step / warmup_steps   total_delta <- target_warmup - initial_learning_rate   completed_fraction * total_delta } if (is.null(warmup_target)) {   initial_decay_lr <- initial_learning_rate } else {   initial_decay_lr <- warmup_target }  decayed_learning_rate <- function(step) {   step <- min(step, decay_steps)   cosine_decay <- 0.5 * (1 + cos(pi * step / decay_steps))   decayed <- (1 - alpha) * cosine_decay + alpha   initial_decay_lr * decayed } decay_steps <- 1000 initial_learning_rate <- 0.1 lr_decayed_fn <- learning_rate_schedule_cosine_decay(     initial_learning_rate, decay_steps) decay_steps <- 1000 initial_learning_rate <- 0 warmup_steps <- 1000 target_learning_rate <- 0.1 lr_warmup_decayed_fn <- learning_rate_schedule_cosine_decay(     initial_learning_rate, decay_steps, warmup_target = target_learning_rate,     warmup_steps = warmup_steps )"},{"path":"https://keras3.posit.co/dev/reference/learning_rate_schedule_cosine_decay.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"A LearningRateSchedule that uses a cosine decay with optional warmup. — learning_rate_schedule_cosine_decay","text":"","code":"learning_rate_schedule_cosine_decay(   initial_learning_rate,   decay_steps,   alpha = 0,   name = \"CosineDecay\",   warmup_target = NULL,   warmup_steps = 0L )"},{"path":"https://keras3.posit.co/dev/reference/learning_rate_schedule_cosine_decay.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"A LearningRateSchedule that uses a cosine decay with optional warmup. — learning_rate_schedule_cosine_decay","text":"initial_learning_rate float. initial learning rate. decay_steps int. Number steps decay . alpha float. Minimum learning rate value decay fraction initial_learning_rate. name String. Optional name operation.  Defaults \"CosineDecay\". warmup_target float. target learning rate warmup phase. cast initial_learning_rate datatype. Setting NULL skip warmup begins decay phase initial_learning_rate. Otherwise scheduler warmup initial_learning_rate warmup_target. warmup_steps int. Number steps warmup .","code":""},{"path":"https://keras3.posit.co/dev/reference/learning_rate_schedule_cosine_decay.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"A LearningRateSchedule that uses a cosine decay with optional warmup. — learning_rate_schedule_cosine_decay","text":"1-arg callable learning rate schedule takes current optimizer step outputs decayed learning rate, scalar tensor type initial_learning_rate.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/learning_rate_schedule_cosine_decay_restarts.html","id":null,"dir":"Reference","previous_headings":"","what":"A LearningRateSchedule that uses a cosine decay schedule with restarts. — learning_rate_schedule_cosine_decay_restarts","title":"A LearningRateSchedule that uses a cosine decay schedule with restarts. — learning_rate_schedule_cosine_decay_restarts","text":"See Loshchilov & Hutter, ICLR2016, SGDR: Stochastic Gradient Descent Warm Restarts. training model, often useful lower learning rate training progresses. schedule applies cosine decay function restarts optimizer step, given provided initial learning rate. requires step value compute decayed learning rate. can just pass backend variable increment training step. schedule 1-arg callable produces decayed learning rate passed current optimizer step. can useful changing learning rate value across different invocations optimizer functions. learning rate multiplier first decays 1 alpha first_decay_steps steps. , warm restart performed. new warm restart runs t_mul times steps m_mul times initial learning rate new learning rate.","code":""},{"path":"https://keras3.posit.co/dev/reference/learning_rate_schedule_cosine_decay_restarts.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"A LearningRateSchedule that uses a cosine decay schedule with restarts. — learning_rate_schedule_cosine_decay_restarts","text":"","code":"learning_rate_schedule_cosine_decay_restarts(   initial_learning_rate,   first_decay_steps,   t_mul = 2,   m_mul = 1,   alpha = 0,   name = \"SGDRDecay\" )"},{"path":"https://keras3.posit.co/dev/reference/learning_rate_schedule_cosine_decay_restarts.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"A LearningRateSchedule that uses a cosine decay schedule with restarts. — learning_rate_schedule_cosine_decay_restarts","text":"initial_learning_rate float. initial learning rate. first_decay_steps integer. Number steps decay . t_mul float. Used derive number iterations -th period. m_mul float. Used derive initial learning rate -th period. alpha float. Minimum learning rate value fraction initial_learning_rate. name String. Optional name operation. Defaults \"SGDRDecay\".","code":""},{"path":"https://keras3.posit.co/dev/reference/learning_rate_schedule_cosine_decay_restarts.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"A LearningRateSchedule that uses a cosine decay schedule with restarts. — learning_rate_schedule_cosine_decay_restarts","text":"1-arg callable learning rate schedule takes current optimizer step outputs decayed learning rate, scalar tensor type initial_learning_rate.","code":""},{"path":"https://keras3.posit.co/dev/reference/learning_rate_schedule_cosine_decay_restarts.html","id":"example","dir":"Reference","previous_headings":"","what":"Example","title":"A LearningRateSchedule that uses a cosine decay schedule with restarts. — learning_rate_schedule_cosine_decay_restarts","text":"can pass schedule directly optimizer learning rate. learning rate schedule also serializable deserializable using keras$optimizers$schedules$serialize keras$optimizers$schedules$deserialize.","code":"first_decay_steps <- 1000 lr_decayed_fn <- learning_rate_schedule_cosine_decay_restarts(         0.001,         first_decay_steps)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/learning_rate_schedule_exponential_decay.html","id":null,"dir":"Reference","previous_headings":"","what":"A LearningRateSchedule that uses an exponential decay schedule. — learning_rate_schedule_exponential_decay","title":"A LearningRateSchedule that uses an exponential decay schedule. — learning_rate_schedule_exponential_decay","text":"training model, often useful lower learning rate training progresses. schedule applies exponential decay function optimizer step, given provided initial learning rate. schedule 1-arg callable produces decayed learning rate passed current optimizer step. can useful changing learning rate value across different invocations optimizer functions. computed :   argument staircase TRUE, step / decay_steps integer division decayed learning rate follows staircase function. can pass schedule directly optimizer learning rate.","code":"decayed_learning_rate <- function(step) {   initial_learning_rate * decay_rate ^ (step / decay_steps) }"},{"path":"https://keras3.posit.co/dev/reference/learning_rate_schedule_exponential_decay.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"A LearningRateSchedule that uses an exponential decay schedule. — learning_rate_schedule_exponential_decay","text":"","code":"learning_rate_schedule_exponential_decay(   initial_learning_rate,   decay_steps,   decay_rate,   staircase = FALSE,   name = \"ExponentialDecay\" )"},{"path":"https://keras3.posit.co/dev/reference/learning_rate_schedule_exponential_decay.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"A LearningRateSchedule that uses an exponential decay schedule. — learning_rate_schedule_exponential_decay","text":"initial_learning_rate float. initial learning rate. decay_steps integer. Must positive. See decay computation . decay_rate float. decay rate. staircase Boolean.  TRUE decay learning rate discrete intervals. name String.  Optional name operation.  Defaults \"ExponentialDecay\".","code":""},{"path":"https://keras3.posit.co/dev/reference/learning_rate_schedule_exponential_decay.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"A LearningRateSchedule that uses an exponential decay schedule. — learning_rate_schedule_exponential_decay","text":"1-arg callable learning rate schedule takes current optimizer step outputs decayed learning rate, scalar tensor type initial_learning_rate.","code":""},{"path":"https://keras3.posit.co/dev/reference/learning_rate_schedule_exponential_decay.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"A LearningRateSchedule that uses an exponential decay schedule. — learning_rate_schedule_exponential_decay","text":"fitting Keras model, decay every 100000 steps base 0.96:   learning rate schedule also serializable deserializable using keras$optimizers$schedules$serialize keras$optimizers$schedules$deserialize.","code":"initial_learning_rate <- 0.1 lr_schedule <- learning_rate_schedule_exponential_decay(     initial_learning_rate,     decay_steps=100000,     decay_rate=0.96,     staircase=TRUE)  model %>% compile(   optimizer = optimizer_sgd(learning_rate = lr_schedule),   loss = 'sparse_categorical_crossentropy',   metrics = c('accuracy'))  model %>% fit(data, labels, epochs=5)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/learning_rate_schedule_inverse_time_decay.html","id":null,"dir":"Reference","previous_headings":"","what":"A LearningRateSchedule that uses an inverse time decay schedule. — learning_rate_schedule_inverse_time_decay","title":"A LearningRateSchedule that uses an inverse time decay schedule. — learning_rate_schedule_inverse_time_decay","text":"training model, often useful lower learning rate training progresses. schedule applies inverse decay function optimizer step, given provided initial learning rate. requires step value compute decayed learning rate. can just pass backend variable increment training step. schedule 1-arg callable produces decayed learning rate passed current optimizer step. can useful changing learning rate value across different invocations optimizer functions. computed :   , staircase TRUE, :   can pass schedule directly optimizer_* learning rate.","code":"decayed_learning_rate <- function(step) {   initial_learning_rate / (1 + decay_rate * step / decay_step) } decayed_learning_rate <- function(step) {   initial_learning_rate /            (1 + decay_rate * floor(step / decay_step)) }"},{"path":"https://keras3.posit.co/dev/reference/learning_rate_schedule_inverse_time_decay.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"A LearningRateSchedule that uses an inverse time decay schedule. — learning_rate_schedule_inverse_time_decay","text":"","code":"learning_rate_schedule_inverse_time_decay(   initial_learning_rate,   decay_steps,   decay_rate,   staircase = FALSE,   name = \"InverseTimeDecay\" )"},{"path":"https://keras3.posit.co/dev/reference/learning_rate_schedule_inverse_time_decay.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"A LearningRateSchedule that uses an inverse time decay schedule. — learning_rate_schedule_inverse_time_decay","text":"initial_learning_rate float. initial learning rate. decay_steps often apply decay. decay_rate number.  decay rate. staircase Whether apply decay discrete staircase, o pposed continuous, fashion. name String.  Optional name operation.  Defaults \"InverseTimeDecay\".","code":""},{"path":"https://keras3.posit.co/dev/reference/learning_rate_schedule_inverse_time_decay.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"A LearningRateSchedule that uses an inverse time decay schedule. — learning_rate_schedule_inverse_time_decay","text":"1-arg callable learning rate schedule takes current optimizer step outputs decayed learning rate, scalar tensor type initial_learning_rate.","code":""},{"path":"https://keras3.posit.co/dev/reference/learning_rate_schedule_inverse_time_decay.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"A LearningRateSchedule that uses an inverse time decay schedule. — learning_rate_schedule_inverse_time_decay","text":"Fit Keras model decaying 1/t rate 0.5:","code":"... initial_learning_rate <- 0.1 decay_steps <- 1.0 decay_rate <- 0.5 learning_rate_fn <- learning_rate_schedule_inverse_time_decay(     initial_learning_rate, decay_steps, decay_rate)  model %>% compile(   optimizer = optimizer_sgd(learning_rate=learning_rate_fn),   loss = 'sparse_categorical_crossentropy',   metrics = 'accuracy') )  model %>% fit(data, labels, epochs=5)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/learning_rate_schedule_piecewise_constant_decay.html","id":null,"dir":"Reference","previous_headings":"","what":"A LearningRateSchedule that uses a piecewise constant decay schedule. — learning_rate_schedule_piecewise_constant_decay","title":"A LearningRateSchedule that uses a piecewise constant decay schedule. — learning_rate_schedule_piecewise_constant_decay","text":"function returns 1-arg callable compute piecewise constant passed current optimizer step. can useful changing learning rate value across different invocations optimizer functions.","code":""},{"path":"https://keras3.posit.co/dev/reference/learning_rate_schedule_piecewise_constant_decay.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"A LearningRateSchedule that uses a piecewise constant decay schedule. — learning_rate_schedule_piecewise_constant_decay","text":"","code":"learning_rate_schedule_piecewise_constant_decay(   boundaries,   values,   name = \"PiecewiseConstant\" )"},{"path":"https://keras3.posit.co/dev/reference/learning_rate_schedule_piecewise_constant_decay.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"A LearningRateSchedule that uses a piecewise constant decay schedule. — learning_rate_schedule_piecewise_constant_decay","text":"boundaries list Python numbers strictly increasing entries, elements type optimizer step. values list Python numbers specifies values intervals defined boundaries. one element boundaries, elements type. name string. Optional name operation. Defaults \"PiecewiseConstant\".","code":""},{"path":"https://keras3.posit.co/dev/reference/learning_rate_schedule_piecewise_constant_decay.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"A LearningRateSchedule that uses a piecewise constant decay schedule. — learning_rate_schedule_piecewise_constant_decay","text":"1-arg callable learning rate schedule takes current optimizer step outputs decayed learning rate, scalar tensor type boundary tensors. output 1-arg function takes step values[0] step <= boundaries[0], values[1] step > boundaries[0] step <= boundaries[1], ..., values[-1] step > boundaries[-1].","code":""},{"path":"https://keras3.posit.co/dev/reference/learning_rate_schedule_piecewise_constant_decay.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"A LearningRateSchedule that uses a piecewise constant decay schedule. — learning_rate_schedule_piecewise_constant_decay","text":"use learning rate 1.0 first 100001 steps, 0.5 next 10000 steps, 0.1 additional steps.   can pass schedule directly optimizer learning rate. learning rate schedule also serializable deserializable using keras$optimizers$schedules$serialize keras$optimizers$schedules$deserialize.","code":"step <- 0 boundaries <- c(100000, 110000) values <- c(1.0, 0.5, 0.1) learning_rate_fn <- learning_rate_schedule_piecewise_constant_decay(   boundaries, values)  # Later, whenever we perform an optimization step, we pass in the step. learning_rate <- learning_rate_fn(step)"},{"path":"https://keras3.posit.co/dev/reference/learning_rate_schedule_piecewise_constant_decay.html","id":"raises","dir":"Reference","previous_headings":"","what":"Raises","title":"A LearningRateSchedule that uses a piecewise constant decay schedule. — learning_rate_schedule_piecewise_constant_decay","text":"ValueError: number elements boundaries values lists match.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/learning_rate_schedule_polynomial_decay.html","id":null,"dir":"Reference","previous_headings":"","what":"A LearningRateSchedule that uses a polynomial decay schedule. — learning_rate_schedule_polynomial_decay","title":"A LearningRateSchedule that uses a polynomial decay schedule. — learning_rate_schedule_polynomial_decay","text":"commonly observed monotonically decreasing learning rate, whose degree change carefully chosen, results better performing model. schedule applies polynomial decay function optimizer step, given provided initial_learning_rate, reach end_learning_rate given decay_steps. requires step value compute decayed learning rate. can just pass backend variable increment training step. schedule 1-arg callable produces decayed learning rate passed current optimizer step. can useful changing learning rate value across different invocations optimizer functions. computed :   cycle TRUE multiple decay_steps used, first one bigger step.   can pass schedule directly Optimizer learning rate.","code":"decayed_learning_rate <- function(step) {   step = min(step, decay_steps)   ((initial_learning_rate - end_learning_rate) *     (1 - step / decay_steps) ^ (power)) +     end_learning_rate } decayed_learning_rate <- function(step) {   decay_steps = decay_steps * ceil(step / decay_steps)   ((initial_learning_rate - end_learning_rate) *       (1 - step / decay_steps) ^ (power)) +     end_learning_rate }"},{"path":"https://keras3.posit.co/dev/reference/learning_rate_schedule_polynomial_decay.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"A LearningRateSchedule that uses a polynomial decay schedule. — learning_rate_schedule_polynomial_decay","text":"","code":"learning_rate_schedule_polynomial_decay(   initial_learning_rate,   decay_steps,   end_learning_rate = 1e-04,   power = 1,   cycle = FALSE,   name = \"PolynomialDecay\" )"},{"path":"https://keras3.posit.co/dev/reference/learning_rate_schedule_polynomial_decay.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"A LearningRateSchedule that uses a polynomial decay schedule. — learning_rate_schedule_polynomial_decay","text":"initial_learning_rate float. initial learning rate. decay_steps integer. Must positive. See decay computation . end_learning_rate float. minimal end learning rate. power float. power polynomial. Defaults 1.0. cycle boolean, whether cycle beyond decay_steps. name String.  Optional name operation. Defaults \"PolynomialDecay\".","code":""},{"path":"https://keras3.posit.co/dev/reference/learning_rate_schedule_polynomial_decay.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"A LearningRateSchedule that uses a polynomial decay schedule. — learning_rate_schedule_polynomial_decay","text":"1-arg callable learning rate schedule takes current optimizer step outputs decayed learning rate, scalar tensor type initial_learning_rate.","code":""},{"path":"https://keras3.posit.co/dev/reference/learning_rate_schedule_polynomial_decay.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"A LearningRateSchedule that uses a polynomial decay schedule. — learning_rate_schedule_polynomial_decay","text":"Fit model decaying 0.1 0.01 10000 steps using sqrt (.e. power=0.5):   learning rate schedule also serializable deserializable using keras$optimizers$schedules$serialize keras$optimizers$schedules$deserialize.","code":"... starter_learning_rate <- 0.1 end_learning_rate <- 0.01 decay_steps <- 10000 learning_rate_fn <- learning_rate_schedule_polynomial_decay(     starter_learning_rate,     decay_steps,     end_learning_rate,     power=0.5)  model %>% compile(   optimizer = optimizer_sgd(learning_rate=learning_rate_fn),   loss = 'sparse_categorical_crossentropy',   metrics = 'accuracy' )  model %>% fit(data, labels, epochs=5)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/load_model.html","id":null,"dir":"Reference","previous_headings":"","what":"Loads a model saved via save_model(). — load_model","title":"Loads a model saved via save_model(). — load_model","text":"Loads model saved via save_model().","code":""},{"path":"https://keras3.posit.co/dev/reference/load_model.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Loads a model saved via save_model(). — load_model","text":"","code":"load_model(model, custom_objects = NULL, compile = TRUE, safe_mode = TRUE)"},{"path":"https://keras3.posit.co/dev/reference/load_model.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Loads a model saved via save_model(). — load_model","text":"model string, path saved model file, raw vector, returned save_model(filepath = NULL) custom_objects Optional named list mapping names custom classes functions considered deserialization. compile Boolean, whether compile model loading. safe_mode Boolean, whether disallow unsafe lambda deserialization. safe_mode=FALSE, loading object potential trigger arbitrary code execution. argument applicable Keras v3 model format. Defaults TRUE.","code":""},{"path":"https://keras3.posit.co/dev/reference/load_model.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Loads a model saved via save_model(). — load_model","text":"Keras model instance. original model compiled, argument compile = TRUE set, returned model compiled. Otherwise, model left uncompiled.","code":""},{"path":"https://keras3.posit.co/dev/reference/load_model.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Loads a model saved via save_model(). — load_model","text":"Note model variables may different name values (var$name property, e.g. \"dense_1/kernel:0\") reloaded. recommended use layer attributes access specific variables, e.g. model |> get_layer(\"dense_1\") |> _$kernel.","code":"model <- keras_model_sequential(input_shape = c(3)) |>   layer_dense(5) |>   layer_activation_softmax()  model |> save_model(\"model.keras\") loaded_model <- load_model(\"model.keras\") x <- random_uniform(c(10, 3)) stopifnot(all.equal(   model |> predict(x),   loaded_model |> predict(x) ))"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/load_model_weights.html","id":null,"dir":"Reference","previous_headings":"","what":"Load weights from a file saved via save_model_weights(). — load_model_weights","title":"Load weights from a file saved via save_model_weights(). — load_model_weights","text":"Weights loaded based network's topology. means architecture weights saved. Note layers weights taken account topological ordering, adding removing layers fine long weights. Partial weight loading modified model, instance adding new layer (weights) changing shape weights layer, can choose ignore errors continue loading setting skip_mismatch=TRUE. case layer mismatching weights skipped. warning displayed skipped layer.","code":""},{"path":"https://keras3.posit.co/dev/reference/load_model_weights.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Load weights from a file saved via save_model_weights(). — load_model_weights","text":"","code":"load_model_weights(model, filepath, skip_mismatch = FALSE, ...)"},{"path":"https://keras3.posit.co/dev/reference/load_model_weights.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Load weights from a file saved via save_model_weights(). — load_model_weights","text":"model keras model. filepath String, path weights file load. can either .weights.h5 file legacy .h5 weights file. skip_mismatch Boolean, whether skip loading layers mismatch number weights, mismatch shape weights. ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/load_model_weights.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Load weights from a file saved via save_model_weights(). — load_model_weights","text":"called primarily side effects. model returned, invisibly, enable usage pipe.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/loss_binary_crossentropy.html","id":null,"dir":"Reference","previous_headings":"","what":"Computes the cross-entropy loss between true labels and predicted labels. — loss_binary_crossentropy","title":"Computes the cross-entropy loss between true labels and predicted labels. — loss_binary_crossentropy","text":"Use cross-entropy loss binary (0 1) classification applications. loss function requires following inputs: y_true (true label): either 0 1. y_pred (predicted value): model's prediction, .e, single floating-point value either represents logit, (.e, value [-inf, inf] from_logits=TRUE) probability (.e, value [0., 1.] from_logits=FALSE).","code":""},{"path":"https://keras3.posit.co/dev/reference/loss_binary_crossentropy.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes the cross-entropy loss between true labels and predicted labels. — loss_binary_crossentropy","text":"","code":"loss_binary_crossentropy(   y_true,   y_pred,   from_logits = FALSE,   label_smoothing = 0,   axis = -1L,   ...,   reduction = \"sum_over_batch_size\",   name = \"binary_crossentropy\",   dtype = NULL )"},{"path":"https://keras3.posit.co/dev/reference/loss_binary_crossentropy.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Computes the cross-entropy loss between true labels and predicted labels. — loss_binary_crossentropy","text":"y_true Ground truth values. shape = [batch_size, d0, .. dN]. y_pred predicted values. shape = [batch_size, d0, .. dN]. from_logits Whether interpret y_pred tensor logit values. default, assume y_pred probabilities (.e., values [0, 1)). label_smoothing Float range [0, 1]. 0, smoothing occurs. > 0, compute loss predicted labels smoothed version true labels, smoothing squeezes labels towards 0.5. Larger values label_smoothing correspond heavier smoothing. axis axis along compute crossentropy (features axis). Defaults -1. ... forward/backward compatability. reduction Type reduction apply loss. almost cases \"sum_over_batch_size\". Supported options \"sum\", \"sum_over_batch_size\", \"mean\", \"mean_with_sample_weight\" NULL. \"sum\" sums loss, \"sum_over_batch_size\" \"mean\" sum loss divide sample size, \"mean_with_sample_weight\" sums loss divides sum sample weights. \"none\" NULL perform aggregation. Defaults \"sum_over_batch_size\". name Optional name loss instance. dtype dtype loss's computations. Defaults NULL, means using config_floatx(). config_floatx() \"float32\" unless set different value (via config_set_floatx()). keras$DTypePolicy provided, compute_dtype utilized.","code":""},{"path":"https://keras3.posit.co/dev/reference/loss_binary_crossentropy.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Computes the cross-entropy loss between true labels and predicted labels. — loss_binary_crossentropy","text":"Binary crossentropy loss value. shape = [batch_size, d0, .. dN-1].","code":""},{"path":"https://keras3.posit.co/dev/reference/loss_binary_crossentropy.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Computes the cross-entropy loss between true labels and predicted labels. — loss_binary_crossentropy","text":"Recommended Usage: (set from_logits=TRUE) compile() API:   standalone function:                     Default Usage: (set from_logits=FALSE)","code":"y_true <- rbind(c(0, 1), c(0, 0)) y_pred <- rbind(c(0.6, 0.4), c(0.4, 0.6)) loss <- loss_binary_crossentropy(y_true, y_pred) loss ## tf.Tensor([0.91629076 0.7135582 ], shape=(2), dtype=float32) model %>% compile(     loss = loss_binary_crossentropy(from_logits=TRUE),     ... ) # Example 1: (batch_size = 1, number of samples = 4) y_true <- op_array(c(0, 1, 0, 0)) y_pred <- op_array(c(-18.6, 0.51, 2.94, -12.8)) bce <- loss_binary_crossentropy(from_logits = TRUE) bce(y_true, y_pred) ## tf.Tensor(0.865458, shape=(), dtype=float32) # Example 2: (batch_size = 2, number of samples = 4) y_true <- rbind(c(0, 1), c(0, 0)) y_pred <- rbind(c(-18.6, 0.51), c(2.94, -12.8)) # Using default 'auto'/'sum_over_batch_size' reduction type. bce <- loss_binary_crossentropy(from_logits = TRUE) bce(y_true, y_pred) ## tf.Tensor(0.865458, shape=(), dtype=float32) # Using 'sample_weight' attribute bce(y_true, y_pred, sample_weight = c(0.8, 0.2)) ## tf.Tensor(0.2436386, shape=(), dtype=float32) # 0.243 # Using 'sum' reduction` type. bce <- loss_binary_crossentropy(from_logits = TRUE, reduction = \"sum\") bce(y_true, y_pred) ## tf.Tensor(1.730916, shape=(), dtype=float32) # Using 'none' reduction type. bce <- loss_binary_crossentropy(from_logits = TRUE, reduction = NULL) bce(y_true, y_pred) ## tf.Tensor([0.23515666 1.4957594 ], shape=(2), dtype=float32) # Make the following updates to the above \"Recommended Usage\" section # 1. Set `from_logits=FALSE` loss_binary_crossentropy() # OR ...('from_logits=FALSE') ## <LossFunctionWrapper(<function binary_crossentropy at 0x0>, kwargs={'from_logits': False, 'label_smoothing': 0.0, 'axis': -1})> ##  signature: (y_true, y_pred, sample_weight=None) # 2. Update `y_pred` to use probabilities instead of logits y_pred <- c(0.6, 0.3, 0.2, 0.8) # OR [[0.6, 0.3], [0.2, 0.8]]"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/loss_binary_focal_crossentropy.html","id":null,"dir":"Reference","previous_headings":"","what":"Computes focal cross-entropy loss between true labels and predictions. — loss_binary_focal_crossentropy","title":"Computes focal cross-entropy loss between true labels and predictions. — loss_binary_focal_crossentropy","text":"According Lin et al., 2018, helps apply focal factor -weight easy examples focus hard examples. default, focal tensor computed follows: focal_factor = (1 - output)^gamma class 1 focal_factor = output^gamma class 0 gamma focusing parameter. gamma = 0, focal effect binary crossentropy loss. apply_class_balancing == TRUE, function also takes account weight balancing factor binary classes 0 1 follows: weight = alpha class 1 (target == 1) weight = 1 - alpha class 0 alpha float range [0, 1]. Binary cross-entropy loss often used binary (0 1) classification tasks. loss function requires following inputs: y_true (true label): either 0 1. y_pred (predicted value): model's prediction, .e, single floating-point value either represents logit, (.e, value [-inf, inf] from_logits=TRUE) probability (.e, value [0., 1.] from_logits=FALSE). According Lin et al., 2018, helps apply \"focal factor\" -weight easy examples focus hard examples. default, focal tensor computed follows: focal_factor = (1 - output) ** gamma class 1 focal_factor = output ** gamma class 0 gamma focusing parameter. gamma=0, function equivalent binary crossentropy loss.","code":""},{"path":"https://keras3.posit.co/dev/reference/loss_binary_focal_crossentropy.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes focal cross-entropy loss between true labels and predictions. — loss_binary_focal_crossentropy","text":"","code":"loss_binary_focal_crossentropy(   y_true,   y_pred,   apply_class_balancing = FALSE,   alpha = 0.25,   gamma = 2,   from_logits = FALSE,   label_smoothing = 0,   axis = -1L,   ...,   reduction = \"sum_over_batch_size\",   name = \"binary_focal_crossentropy\",   dtype = NULL )"},{"path":"https://keras3.posit.co/dev/reference/loss_binary_focal_crossentropy.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Computes focal cross-entropy loss between true labels and predictions. — loss_binary_focal_crossentropy","text":"y_true Ground truth values, shape (batch_size, d0, .. dN). y_pred predicted values, shape (batch_size, d0, .. dN). apply_class_balancing bool, whether apply weight balancing binary classes 0 1. alpha weight balancing factor class 1, default 0.25 mentioned reference Lin et al., 2018.  weight class 0 1.0 - alpha. gamma focusing parameter used compute focal factor, default 2.0 mentioned reference Lin et al., 2018. from_logits Whether interpret y_pred tensor logit values. default, assume y_pred probabilities (.e., values [0, 1]). label_smoothing Float [0, 1]. 0, smoothing occurs. > 0, compute loss predicted labels smoothed version true labels, smoothing squeezes labels towards 0.5. Larger values label_smoothing correspond heavier smoothing. axis axis along compute crossentropy (features axis). Defaults -1. ... forward/backward compatability. reduction Type reduction apply loss. almost cases \"sum_over_batch_size\". Supported options \"sum\", \"sum_over_batch_size\", \"mean\", \"mean_with_sample_weight\" NULL. \"sum\" sums loss, \"sum_over_batch_size\" \"mean\" sum loss divide sample size, \"mean_with_sample_weight\" sums loss divides sum sample weights. \"none\" NULL perform aggregation. Defaults \"sum_over_batch_size\". name Optional name loss instance. dtype dtype loss's computations. Defaults NULL, means using config_floatx(). config_floatx() \"float32\" unless set different value (via config_set_floatx()). keras$DTypePolicy provided, compute_dtype utilized.","code":""},{"path":"https://keras3.posit.co/dev/reference/loss_binary_focal_crossentropy.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Computes focal cross-entropy loss between true labels and predictions. — loss_binary_focal_crossentropy","text":"Binary focal crossentropy loss value shape = [batch_size, d0, .. dN-1].","code":""},{"path":"https://keras3.posit.co/dev/reference/loss_binary_focal_crossentropy.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Computes focal cross-entropy loss between true labels and predictions. — loss_binary_focal_crossentropy","text":"compile() API:   standalone function:","code":"y_true <- rbind(c(0, 1), c(0, 0)) y_pred <- rbind(c(0.6, 0.4), c(0.4, 0.6)) loss <- loss_binary_focal_crossentropy(y_true, y_pred, gamma = 2) loss ## tf.Tensor([0.32986468 0.20579839], shape=(2), dtype=float32) model %>% compile(     loss = loss_binary_focal_crossentropy(         gamma = 2.0, from_logits = TRUE),     ... ) # Example 1: (batch_size = 1, number of samples = 4) y_true <- op_array(c(0, 1, 0, 0)) y_pred <- op_array(c(-18.6, 0.51, 2.94, -12.8)) loss <- loss_binary_focal_crossentropy(gamma = 2, from_logits = TRUE) loss(y_true, y_pred) ## tf.Tensor(0.6912122, shape=(), dtype=float32) # Apply class weight loss <- loss_binary_focal_crossentropy(   apply_class_balancing = TRUE, gamma = 2, from_logits = TRUE) loss(y_true, y_pred) ## tf.Tensor(0.5101333, shape=(), dtype=float32) # Example 2: (batch_size = 2, number of samples = 4) y_true <- rbind(c(0, 1), c(0, 0)) y_pred <- rbind(c(-18.6, 0.51), c(2.94, -12.8)) # Using default 'auto'/'sum_over_batch_size' reduction type. loss <- loss_binary_focal_crossentropy(     gamma = 3, from_logits = TRUE) loss(y_true, y_pred) ## tf.Tensor(0.6469951, shape=(), dtype=float32) # Apply class weight loss <- loss_binary_focal_crossentropy(      apply_class_balancing = TRUE, gamma = 3, from_logits = TRUE) loss(y_true, y_pred) ## tf.Tensor(0.48214132, shape=(), dtype=float32) # Using 'sample_weight' attribute with focal effect loss <- loss_binary_focal_crossentropy(     gamma = 3, from_logits = TRUE) loss(y_true, y_pred, sample_weight = c(0.8, 0.2)) ## tf.Tensor(0.13312504, shape=(), dtype=float32) # Apply class weight loss <- loss_binary_focal_crossentropy(      apply_class_balancing = TRUE, gamma = 3, from_logits = TRUE) loss(y_true, y_pred, sample_weight = c(0.8, 0.2)) ## tf.Tensor(0.09735977, shape=(), dtype=float32) # Using 'sum' reduction` type. loss <- loss_binary_focal_crossentropy(     gamma = 4, from_logits = TRUE,     reduction = \"sum\") loss(y_true, y_pred) ## tf.Tensor(1.2218808, shape=(), dtype=float32) # Apply class weight loss <- loss_binary_focal_crossentropy(     apply_class_balancing = TRUE, gamma = 4, from_logits = TRUE,     reduction = \"sum\") loss(y_true, y_pred) ## tf.Tensor(0.9140807, shape=(), dtype=float32) # Using 'none' reduction type. loss <- loss_binary_focal_crossentropy(     gamma = 5, from_logits = TRUE,     reduction = NULL) loss(y_true, y_pred) ## tf.Tensor([0.00174837 1.1561027 ], shape=(2), dtype=float32) # Apply class weight loss <- loss_binary_focal_crossentropy(     apply_class_balancing = TRUE, gamma = 5, from_logits = TRUE,     reduction = NULL) loss(y_true, y_pred) ## tf.Tensor([4.3709317e-04 8.6707699e-01], shape=(2), dtype=float32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/loss_categorical_crossentropy.html","id":null,"dir":"Reference","previous_headings":"","what":"Computes the crossentropy loss between the labels and predictions. — loss_categorical_crossentropy","title":"Computes the crossentropy loss between the labels and predictions. — loss_categorical_crossentropy","text":"Use crossentropy loss function two label classes. expect labels provided one_hot representation. want provide labels integers, please use SparseCategoricalCrossentropy loss. num_classes floating point values per feature, .e., shape y_pred y_true [batch_size, num_classes].","code":""},{"path":"https://keras3.posit.co/dev/reference/loss_categorical_crossentropy.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes the crossentropy loss between the labels and predictions. — loss_categorical_crossentropy","text":"","code":"loss_categorical_crossentropy(   y_true,   y_pred,   from_logits = FALSE,   label_smoothing = 0,   axis = -1L,   ...,   reduction = \"sum_over_batch_size\",   name = \"categorical_crossentropy\",   dtype = NULL )"},{"path":"https://keras3.posit.co/dev/reference/loss_categorical_crossentropy.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Computes the crossentropy loss between the labels and predictions. — loss_categorical_crossentropy","text":"y_true Tensor one-hot true targets. y_pred Tensor predicted targets. from_logits Whether y_pred expected logits tensor. default, assume y_pred encodes probability distribution. label_smoothing Float [0, 1]. > 0, label values smoothed, meaning confidence label values relaxed. example, 0.1, use 0.1 / num_classes non-target labels 0.9 + 0.1 / num_classes target labels. axis axis along compute crossentropy (features axis). Defaults -1. ... forward/backward compatability. reduction Type reduction apply loss. almost cases \"sum_over_batch_size\". Supported options \"sum\", \"sum_over_batch_size\", \"mean\", \"mean_with_sample_weight\" NULL. \"sum\" sums loss, \"sum_over_batch_size\" \"mean\" sum loss divide sample size, \"mean_with_sample_weight\" sums loss divides sum sample weights. \"none\" NULL perform aggregation. Defaults \"sum_over_batch_size\". name Optional name loss instance. dtype dtype loss's computations. Defaults NULL, means using config_floatx(). config_floatx() \"float32\" unless set different value (via config_set_floatx()). keras$DTypePolicy provided, compute_dtype utilized.","code":""},{"path":"https://keras3.posit.co/dev/reference/loss_categorical_crossentropy.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Computes the crossentropy loss between the labels and predictions. — loss_categorical_crossentropy","text":"Categorical crossentropy loss value.","code":""},{"path":"https://keras3.posit.co/dev/reference/loss_categorical_crossentropy.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Computes the crossentropy loss between the labels and predictions. — loss_categorical_crossentropy","text":"Standalone usage:                 Usage compile() API:","code":"y_true <- rbind(c(0, 1, 0), c(0, 0, 1)) y_pred <- rbind(c(0.05, 0.95, 0), c(0.1, 0.8, 0.1)) loss <- loss_categorical_crossentropy(y_true, y_pred) loss ## tf.Tensor([0.05129331 2.3025851 ], shape=(2), dtype=float32) y_true <- rbind(c(0, 1, 0), c(0, 0, 1)) y_pred <- rbind(c(0.05, 0.95, 0), c(0.1, 0.8, 0.1)) # Using 'auto'/'sum_over_batch_size' reduction type. cce <- loss_categorical_crossentropy() cce(y_true, y_pred) ## tf.Tensor(1.1769392, shape=(), dtype=float32) # Calling with 'sample_weight'. cce(y_true, y_pred, sample_weight = op_array(c(0.3, 0.7))) ## tf.Tensor(0.8135988, shape=(), dtype=float32) # Using 'sum' reduction type. cce <- loss_categorical_crossentropy(reduction = \"sum\") cce(y_true, y_pred) ## tf.Tensor(2.3538785, shape=(), dtype=float32) # Using 'none' reduction type. cce <- loss_categorical_crossentropy(reduction = NULL) cce(y_true, y_pred) ## tf.Tensor([0.05129331 2.3025851 ], shape=(2), dtype=float32) model %>% compile(optimizer = 'sgd',               loss=loss_categorical_crossentropy())"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/loss_categorical_focal_crossentropy.html","id":null,"dir":"Reference","previous_headings":"","what":"Computes the alpha balanced focal crossentropy loss. — loss_categorical_focal_crossentropy","title":"Computes the alpha balanced focal crossentropy loss. — loss_categorical_focal_crossentropy","text":"Use crossentropy loss function two label classes want handle class imbalance without using class_weights. expect labels provided one_hot representation. According Lin et al., 2018, helps apply focal factor -weight easy examples focus hard examples. general formula focal loss (FL) follows: FL(p_t) = (1 - p_t)^gamma * log(p_t) p_t defined follows: p_t = output y_true == 1, else 1 - output (1 - p_t)^gamma modulating_factor, gamma focusing parameter. gamma = 0, focal effect cross entropy. gamma reduces importance given simple examples smooth manner. authors use alpha-balanced variant focal loss (FL) paper: FL(p_t) = -alpha * (1 - p_t)^gamma * log(p_t) alpha weight factor classes. alpha = 1, loss able handle class imbalance properly classes weight. can constant list constants. alpha list, must length number classes. formula can generalized : FL(p_t) = alpha * (1 - p_t)^gamma * CrossEntropy(y_true, y_pred) minus comes CrossEntropy(y_true, y_pred) (CE). Extending multi-class case straightforward: FL(p_t) = alpha * (1 - p_t) ** gamma * CategoricalCE(y_true, y_pred) snippet , num_classes floating pointing values per example. shape y_pred y_true (batch_size, num_classes).","code":""},{"path":"https://keras3.posit.co/dev/reference/loss_categorical_focal_crossentropy.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes the alpha balanced focal crossentropy loss. — loss_categorical_focal_crossentropy","text":"","code":"loss_categorical_focal_crossentropy(   y_true,   y_pred,   alpha = 0.25,   gamma = 2,   from_logits = FALSE,   label_smoothing = 0,   axis = -1L,   ...,   reduction = \"sum_over_batch_size\",   name = \"categorical_focal_crossentropy\",   dtype = NULL )"},{"path":"https://keras3.posit.co/dev/reference/loss_categorical_focal_crossentropy.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Computes the alpha balanced focal crossentropy loss. — loss_categorical_focal_crossentropy","text":"y_true Tensor one-hot true targets. y_pred Tensor predicted targets. alpha weight balancing factor classes, default 0.25 mentioned reference. can list floats scalar. multi-class case, alpha may set inverse class frequency using compute_class_weight sklearn.utils. gamma focusing parameter, default 2.0 mentioned reference. helps gradually reduce importance given simple examples smooth manner. gamma = 0, focal effect categorical crossentropy. from_logits Whether output expected logits tensor. default, consider output encodes probability distribution. label_smoothing Float [0, 1]. > 0, label values smoothed, meaning confidence label values relaxed. example, 0.1, use 0.1 / num_classes non-target labels 0.9 + 0.1 / num_classes target labels. axis axis along compute crossentropy (features axis). Defaults -1. ... forward/backward compatability. reduction Type reduction apply loss. almost cases \"sum_over_batch_size\". Supported options \"sum\", \"sum_over_batch_size\", \"mean\", \"mean_with_sample_weight\" NULL. \"sum\" sums loss, \"sum_over_batch_size\" \"mean\" sum loss divide sample size, \"mean_with_sample_weight\" sums loss divides sum sample weights. \"none\" NULL perform aggregation. Defaults \"sum_over_batch_size\". name Optional name loss instance. dtype dtype loss's computations. Defaults NULL, means using config_floatx(). config_floatx() \"float32\" unless set different value (via config_set_floatx()). keras$DTypePolicy provided, compute_dtype utilized.","code":""},{"path":"https://keras3.posit.co/dev/reference/loss_categorical_focal_crossentropy.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Computes the alpha balanced focal crossentropy loss. — loss_categorical_focal_crossentropy","text":"Categorical focal crossentropy loss value.","code":""},{"path":"https://keras3.posit.co/dev/reference/loss_categorical_focal_crossentropy.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Computes the alpha balanced focal crossentropy loss. — loss_categorical_focal_crossentropy","text":"Standalone usage:                 Usage compile() API:","code":"y_true <- rbind(c(0, 1, 0), c(0, 0, 1)) y_pred <- rbind(c(0.05, 0.95, 0), c(0.1, 0.8, 0.1)) loss <- loss_categorical_focal_crossentropy(y_true, y_pred) loss ## tf.Tensor([3.2058331e-05 4.6627346e-01], shape=(2), dtype=float32) y_true <- rbind(c(0, 1, 0), c(0, 0, 1)) y_pred <- rbind(c(0.05, 0.95, 0), c(0.1, 0.8, 0.1)) # Using 'auto'/'sum_over_batch_size' reduction type. cce <- loss_categorical_focal_crossentropy() cce(y_true, y_pred) ## tf.Tensor(0.23315276, shape=(), dtype=float32) # Calling with 'sample_weight'. cce(y_true, y_pred, sample_weight = op_array(c(0.3, 0.7))) ## tf.Tensor(0.16320053, shape=(), dtype=float32) # Using 'sum' reduction type. cce <- loss_categorical_focal_crossentropy(reduction = \"sum\") cce(y_true, y_pred) ## tf.Tensor(0.46630552, shape=(), dtype=float32) # Using 'none' reduction type. cce <- loss_categorical_focal_crossentropy(reduction = NULL) cce(y_true, y_pred) ## tf.Tensor([3.2058331e-05 4.6627346e-01], shape=(2), dtype=float32) model %>% compile(   optimizer = 'adam',   loss = loss_categorical_focal_crossentropy())"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/loss_categorical_hinge.html","id":null,"dir":"Reference","previous_headings":"","what":"Computes the categorical hinge loss between y_true & y_pred. — loss_categorical_hinge","title":"Computes the categorical hinge loss between y_true & y_pred. — loss_categorical_hinge","text":"Formula:   neg=maximum((1-y_true)*y_pred) pos=sum(y_true*y_pred)","code":"loss <- maximum(neg - pos + 1, 0)"},{"path":"https://keras3.posit.co/dev/reference/loss_categorical_hinge.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes the categorical hinge loss between y_true & y_pred. — loss_categorical_hinge","text":"","code":"loss_categorical_hinge(   y_true,   y_pred,   ...,   reduction = \"sum_over_batch_size\",   name = \"categorical_hinge\",   dtype = NULL )"},{"path":"https://keras3.posit.co/dev/reference/loss_categorical_hinge.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Computes the categorical hinge loss between y_true & y_pred. — loss_categorical_hinge","text":"y_true ground truth values. y_true values expected either {-1, +1} {0, 1} (.e. one-hot-encoded tensor) shape <- [batch_size, d0, .. dN]. y_pred predicted values shape = [batch_size, d0, .. dN]. ... forward/backward compatability. reduction Type reduction apply loss. almost cases \"sum_over_batch_size\". Supported options \"sum\", \"sum_over_batch_size\", \"mean\", \"mean_with_sample_weight\" NULL. \"sum\" sums loss, \"sum_over_batch_size\" \"mean\" sum loss divide sample size, \"mean_with_sample_weight\" sums loss divides sum sample weights. \"none\" NULL perform aggregation. Defaults \"sum_over_batch_size\". name Optional name loss instance. dtype dtype loss's computations. Defaults NULL, means using config_floatx(). config_floatx() \"float32\" unless set different value (via config_set_floatx()). keras$DTypePolicy provided, compute_dtype utilized.","code":""},{"path":"https://keras3.posit.co/dev/reference/loss_categorical_hinge.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Computes the categorical hinge loss between y_true & y_pred. — loss_categorical_hinge","text":"Categorical hinge loss values shape = [batch_size, d0, .. dN-1].","code":""},{"path":"https://keras3.posit.co/dev/reference/loss_categorical_hinge.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Computes the categorical hinge loss between y_true & y_pred. — loss_categorical_hinge","text":"","code":"y_true <- rbind(c(0, 1), c(0, 0)) y_pred <- rbind(c(0.6, 0.4), c(0.4, 0.6)) loss <- loss_categorical_hinge(y_true, y_pred)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/loss_circle.html","id":null,"dir":"Reference","previous_headings":"","what":"Computes Circle Loss between integer labels and L2-normalized embeddings. — loss_circle","title":"Computes Circle Loss between integer labels and L2-normalized embeddings. — loss_circle","text":"designed minimize within-class distances maximize -class distances L2 normalized embedding space. metric learning loss designed minimize within-class distance maximize -class distance flexible manner dynamically adjusting penalty strength based optimization status similarity score. use Circle Loss effectively, model output embeddings without activation function (Dense layer activation=NULL) followed UnitNormalization layer ensure unit-norm embeddings.","code":""},{"path":"https://keras3.posit.co/dev/reference/loss_circle.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes Circle Loss between integer labels and L2-normalized embeddings. — loss_circle","text":"","code":"loss_circle(   y_true,   y_pred,   ref_labels = NULL,   ref_embeddings = NULL,   remove_diagonal = TRUE,   gamma = 80L,   margin = 0.4,   ...,   reduction = \"sum_over_batch_size\",   name = \"circle\",   dtype = NULL )"},{"path":"https://keras3.posit.co/dev/reference/loss_circle.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Computes Circle Loss between integer labels and L2-normalized embeddings. — loss_circle","text":"y_true Tensor ground truth labels integer format. y_pred Tensor predicted L2 normalized embeddings. ref_labels Optional integer tensor labels reference embeddings. NULL, defaults y_true. ref_embeddings Optional tensor L2 normalized reference embeddings. NULL, defaults y_pred. remove_diagonal Boolean, whether remove self-similarities positive mask. Defaults TRUE. gamma Scaling factor determines largest scale similarity score. Defaults 80. margin relaxation factor, distance, negatives weighted positives weighted. Similarly, distance negatives weighted positive weighted. Defaults 0.4. ... forward/backward compatability. reduction Type reduction apply loss. almost cases \"sum_over_batch_size\". Supported options \"sum\", \"sum_over_batch_size\", \"mean\", \"mean_with_sample_weight\" NULL. \"sum\" sums loss, \"sum_over_batch_size\" \"mean\" sum loss divide sample size, \"mean_with_sample_weight\" sums loss divides sum sample weights. \"none\" NULL perform aggregation. Defaults \"sum_over_batch_size\". name Optional name loss instance. dtype dtype loss's computations. Defaults NULL, means using config_floatx(). config_floatx() \"float32\" unless set different value (via config_set_floatx()). keras.DTypePolicy provided, compute_dtype utilized.","code":""},{"path":"https://keras3.posit.co/dev/reference/loss_circle.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Computes Circle Loss between integer labels and L2-normalized embeddings. — loss_circle","text":"Circle loss value.","code":""},{"path":"https://keras3.posit.co/dev/reference/loss_circle.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Computes Circle Loss between integer labels and L2-normalized embeddings. — loss_circle","text":"Usage compile() API:","code":"model <- keras_model_sequential(input_shape = c(224, 224, 3)) |>   layer_conv_2d(16, c(3, 3), activation = 'relu') |>   layer_flatten() |>   layer_dense(64, activation = NULL) |>   # No activation   layer_unit_normalization()  # L2 normalization  model |>   compile(optimizer = \"adam\", loss = loss_circle())"},{"path":"https://keras3.posit.co/dev/reference/loss_circle.html","id":"reference","dir":"Reference","previous_headings":"","what":"Reference","title":"Computes Circle Loss between integer labels and L2-normalized embeddings. — loss_circle","text":"Yifan Sun et al., 2020","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/loss_cosine_similarity.html","id":null,"dir":"Reference","previous_headings":"","what":"Computes the cosine similarity between y_true & y_pred. — loss_cosine_similarity","title":"Computes the cosine similarity between y_true & y_pred. — loss_cosine_similarity","text":"Formula:   Note number -1 1. negative number -1 0, 0 indicates orthogonality values closer -1 indicate greater similarity. makes usable loss function setting try maximize proximity predictions targets. either y_true y_pred zero vector, cosine similarity 0 regardless proximity predictions targets.","code":"loss <- -sum(l2_norm(y_true) * l2_norm(y_pred))"},{"path":"https://keras3.posit.co/dev/reference/loss_cosine_similarity.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes the cosine similarity between y_true & y_pred. — loss_cosine_similarity","text":"","code":"loss_cosine_similarity(   y_true,   y_pred,   axis = -1L,   ...,   reduction = \"sum_over_batch_size\",   name = \"cosine_similarity\",   dtype = NULL )"},{"path":"https://keras3.posit.co/dev/reference/loss_cosine_similarity.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Computes the cosine similarity between y_true & y_pred. — loss_cosine_similarity","text":"y_true Tensor true targets. y_pred Tensor predicted targets. axis axis along cosine similarity computed (features axis). Defaults -1. ... forward/backward compatability. reduction Type reduction apply loss. almost cases \"sum_over_batch_size\". Supported options \"sum\", \"sum_over_batch_size\", \"mean\", \"mean_with_sample_weight\" NULL. \"sum\" sums loss, \"sum_over_batch_size\" \"mean\" sum loss divide sample size, \"mean_with_sample_weight\" sums loss divides sum sample weights. \"none\" NULL perform aggregation. Defaults \"sum_over_batch_size\". name Optional name loss instance. dtype dtype loss's computations. Defaults NULL, means using config_floatx(). config_floatx() \"float32\" unless set different value (via config_set_floatx()). keras$DTypePolicy provided, compute_dtype utilized.","code":""},{"path":"https://keras3.posit.co/dev/reference/loss_cosine_similarity.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Computes the cosine similarity between y_true & y_pred. — loss_cosine_similarity","text":"Cosine similarity tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/loss_cosine_similarity.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Computes the cosine similarity between y_true & y_pred. — loss_cosine_similarity","text":"","code":"y_true <- rbind(c(0., 1.), c(1., 1.), c(1., 1.)) y_pred <- rbind(c(1., 0.), c(1., 1.), c(-1., -1.)) loss <- loss_cosine_similarity(y_true, y_pred, axis=-1) loss ## tf.Tensor([-0.         -0.99999994  0.99999994], shape=(3), dtype=float32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/loss_ctc.html","id":null,"dir":"Reference","previous_headings":"","what":"CTC (Connectionist Temporal Classification) loss. — loss_ctc","title":"CTC (Connectionist Temporal Classification) loss. — loss_ctc","text":"CTC (Connectionist Temporal Classification) loss.","code":""},{"path":"https://keras3.posit.co/dev/reference/loss_ctc.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"CTC (Connectionist Temporal Classification) loss. — loss_ctc","text":"","code":"loss_ctc(   y_true,   y_pred,   ...,   reduction = \"sum_over_batch_size\",   name = \"ctc\",   dtype = NULL )"},{"path":"https://keras3.posit.co/dev/reference/loss_ctc.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"CTC (Connectionist Temporal Classification) loss. — loss_ctc","text":"y_true tensor shape (batch_size, target_max_length) containing true labels integer format. 0 always represents blank/mask index used classes. y_pred tensor shape (batch_size, output_max_length, num_classes) containing logits (output model). normalized via softmax. ... forward/backward compatability. reduction Type reduction apply loss. almost cases \"sum_over_batch_size\". Supported options \"sum\", \"sum_over_batch_size\", \"mean\", \"mean_with_sample_weight\" NULL. \"sum\" sums loss, \"sum_over_batch_size\" \"mean\" sum loss divide sample size, \"mean_with_sample_weight\" sums loss divides sum sample weights. \"none\" NULL perform aggregation. Defaults \"sum_over_batch_size\". name Optional name loss instance. dtype dtype loss's computations. Defaults NULL, means using config_floatx(). config_floatx() \"float32\" unless set different value (via config_set_floatx()). keras$DTypePolicy provided, compute_dtype utilized.","code":""},{"path":"https://keras3.posit.co/dev/reference/loss_ctc.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"CTC (Connectionist Temporal Classification) loss. — loss_ctc","text":"CTC loss value.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/loss_dice.html","id":null,"dir":"Reference","previous_headings":"","what":"Computes the Dice loss value between y_true and y_pred. — loss_dice","title":"Computes the Dice loss value between y_true and y_pred. — loss_dice","text":"Formula:   Formula:","code":"loss = 1 - (2 * sum(y_true * y_pred)) / (sum(y_true) + sum(y_pred)) loss = 1 - (2 * sum(y_true * y_pred)) / (sum(y_true) + sum(y_pred))"},{"path":"https://keras3.posit.co/dev/reference/loss_dice.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes the Dice loss value between y_true and y_pred. — loss_dice","text":"","code":"loss_dice(   y_true,   y_pred,   ...,   reduction = \"sum_over_batch_size\",   name = \"dice\",   axis = NULL,   dtype = NULL )"},{"path":"https://keras3.posit.co/dev/reference/loss_dice.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Computes the Dice loss value between y_true and y_pred. — loss_dice","text":"y_true Tensor true targets. y_pred Tensor predicted targets. ... forward/backward compatability. reduction Type reduction apply loss. almost cases \"sum_over_batch_size\". Supported options \"sum\", \"sum_over_batch_size\", \"mean\", \"mean_with_sample_weight\" NULL. \"sum\" sums loss, \"sum_over_batch_size\" \"mean\" sum loss divide sample size, \"mean_with_sample_weight\" sums loss divides sum sample weights. \"none\" NULL perform aggregation. Defaults \"sum_over_batch_size\". name String, name object axis List dimensions loss calculated. Defaults NULL. dtype dtype loss's computations. Defaults NULL, means using config_floatx(). config_floatx() \"float32\" unless set different value (via config_set_floatx()). keras$DTypePolicy provided, compute_dtype utilized.","code":""},{"path":"https://keras3.posit.co/dev/reference/loss_dice.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Computes the Dice loss value between y_true and y_pred. — loss_dice","text":"y_true y_pred provided, Dice loss value. Otherwise, Loss() instance.","code":""},{"path":"https://keras3.posit.co/dev/reference/loss_dice.html","id":"example","dir":"Reference","previous_headings":"","what":"Example","title":"Computes the Dice loss value between y_true and y_pred. — loss_dice","text":"","code":"y_true <- array(c(1, 1, 0, 0,                   1, 1, 0, 0), dim = c(2, 2, 2, 1)) y_pred <- array(c(0, 0.4, 0,   0,                   1,   0, 1, 0.9), dim = c(2, 2, 2, 1))  axis <- c(2, 3, 4) loss <- loss_dice(y_true, y_pred, axis = axis) stopifnot(shape(loss) == shape(2)) loss ## tf.Tensor([0.5        0.75757575], shape=(2), dtype=float32) loss = loss_dice(y_true, y_pred) stopifnot(shape(loss) == shape()) loss ## tf.Tensor(0.6164384, shape=(), dtype=float32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/loss_hinge.html","id":null,"dir":"Reference","previous_headings":"","what":"Computes the hinge loss between y_true & y_pred. — loss_hinge","title":"Computes the hinge loss between y_true & y_pred. — loss_hinge","text":"Formula:   y_true values expected -1 1. binary (0 1) labels provided convert -1 1.","code":"loss <- mean(maximum(1 - y_true * y_pred, 0), axis=-1)"},{"path":"https://keras3.posit.co/dev/reference/loss_hinge.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes the hinge loss between y_true & y_pred. — loss_hinge","text":"","code":"loss_hinge(   y_true,   y_pred,   ...,   reduction = \"sum_over_batch_size\",   name = \"hinge\",   dtype = NULL )"},{"path":"https://keras3.posit.co/dev/reference/loss_hinge.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Computes the hinge loss between y_true & y_pred. — loss_hinge","text":"y_true ground truth values. y_true values expected -1 1. binary (0 1) labels provided converted -1 1 shape = [batch_size, d0, .. dN]. y_pred predicted values shape = [batch_size, d0, .. dN]. ... forward/backward compatability. reduction Type reduction apply loss. almost cases \"sum_over_batch_size\". Supported options \"sum\", \"sum_over_batch_size\", \"mean\", \"mean_with_sample_weight\" NULL. \"sum\" sums loss, \"sum_over_batch_size\" \"mean\" sum loss divide sample size, \"mean_with_sample_weight\" sums loss divides sum sample weights. \"none\" NULL perform aggregation. Defaults \"sum_over_batch_size\". name Optional name loss instance. dtype dtype loss's computations. Defaults NULL, means using config_floatx(). config_floatx() \"float32\" unless set different value (via config_set_floatx()). keras$DTypePolicy provided, compute_dtype utilized.","code":""},{"path":"https://keras3.posit.co/dev/reference/loss_hinge.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Computes the hinge loss between y_true & y_pred. — loss_hinge","text":"Hinge loss values shape = [batch_size, d0, .. dN-1].","code":""},{"path":"https://keras3.posit.co/dev/reference/loss_hinge.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Computes the hinge loss between y_true & y_pred. — loss_hinge","text":"","code":"y_true <- array(sample(c(-1,1), 6, replace = TRUE), dim = c(2, 3)) y_pred <- random_uniform(c(2, 3)) loss <- loss_hinge(y_true, y_pred) loss ## tf.Tensor([1.0610152  0.93285507], shape=(2), dtype=float32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/loss_huber.html","id":null,"dir":"Reference","previous_headings":"","what":"Computes the Huber loss between y_true & y_pred. — loss_huber","title":"Computes the Huber loss between y_true & y_pred. — loss_huber","text":"Formula:   See: Huber loss.","code":"for (x in error) {   if (abs(x) <= delta){     loss <- c(loss, (0.5 * x^2))   } else if (abs(x) > delta) {     loss <- c(loss, (delta * abs(x) - 0.5 * delta^2))   } } loss <- mean(loss)"},{"path":"https://keras3.posit.co/dev/reference/loss_huber.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes the Huber loss between y_true & y_pred. — loss_huber","text":"","code":"loss_huber(   y_true,   y_pred,   delta = 1,   ...,   reduction = \"sum_over_batch_size\",   name = \"huber_loss\",   dtype = NULL )"},{"path":"https://keras3.posit.co/dev/reference/loss_huber.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Computes the Huber loss between y_true & y_pred. — loss_huber","text":"y_true tensor true targets. y_pred tensor predicted targets. delta float, point Huber loss function changes quadratic linear. Defaults 1.0. ... forward/backward compatability. reduction Type reduction apply loss. Options \"sum\", \"sum_over_batch_size\" NULL. Defaults \"sum_over_batch_size\". name Optional name instance. dtype dtype loss's computations. Defaults NULL, means using config_floatx(). config_floatx() \"float32\" unless set different value (via config_set_floatx()). keras$DTypePolicy provided, compute_dtype utilized.","code":""},{"path":"https://keras3.posit.co/dev/reference/loss_huber.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Computes the Huber loss between y_true & y_pred. — loss_huber","text":"Tensor one scalar loss entry per sample.","code":""},{"path":"https://keras3.posit.co/dev/reference/loss_huber.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Computes the Huber loss between y_true & y_pred. — loss_huber","text":"","code":"y_true <- rbind(c(0, 1), c(0, 0)) y_pred <- rbind(c(0.6, 0.4), c(0.4, 0.6)) loss <- loss_huber(y_true, y_pred)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/loss_kl_divergence.html","id":null,"dir":"Reference","previous_headings":"","what":"Computes Kullback-Leibler divergence loss between y_true & y_pred. — loss_kl_divergence","title":"Computes Kullback-Leibler divergence loss between y_true & y_pred. — loss_kl_divergence","text":"Formula:   y_true y_pred expected probability distributions, values 0 1. get clipped [0, 1] range.","code":"loss <- y_true * log(y_true / y_pred)"},{"path":"https://keras3.posit.co/dev/reference/loss_kl_divergence.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes Kullback-Leibler divergence loss between y_true & y_pred. — loss_kl_divergence","text":"","code":"loss_kl_divergence(   y_true,   y_pred,   ...,   reduction = \"sum_over_batch_size\",   name = \"kl_divergence\",   dtype = NULL )"},{"path":"https://keras3.posit.co/dev/reference/loss_kl_divergence.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Computes Kullback-Leibler divergence loss between y_true & y_pred. — loss_kl_divergence","text":"y_true Tensor true targets. y_pred Tensor predicted targets. ... forward/backward compatability. reduction Type reduction apply loss. almost cases \"sum_over_batch_size\". Supported options \"sum\", \"sum_over_batch_size\", \"mean\", \"mean_with_sample_weight\" NULL. \"sum\" sums loss, \"sum_over_batch_size\" \"mean\" sum loss divide sample size, \"mean_with_sample_weight\" sums loss divides sum sample weights. \"none\" NULL perform aggregation. Defaults \"sum_over_batch_size\". name Optional name loss instance. dtype dtype loss's computations. Defaults NULL, means using config_floatx(). config_floatx() \"float32\" unless set different value (via config_set_floatx()). keras$DTypePolicy provided, compute_dtype utilized.","code":""},{"path":"https://keras3.posit.co/dev/reference/loss_kl_divergence.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Computes Kullback-Leibler divergence loss between y_true & y_pred. — loss_kl_divergence","text":"KL Divergence loss values shape = [batch_size, d0, .. dN-1].","code":""},{"path":"https://keras3.posit.co/dev/reference/loss_kl_divergence.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Computes Kullback-Leibler divergence loss between y_true & y_pred. — loss_kl_divergence","text":"","code":"y_true <- random_uniform(c(2, 3), 0, 2) y_pred <- random_uniform(c(2,3)) loss <- loss_kl_divergence(y_true, y_pred) loss ## tf.Tensor([ 2.4290292 -0.6284211], shape=(2), dtype=float32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/loss_log_cosh.html","id":null,"dir":"Reference","previous_headings":"","what":"Computes the logarithm of the hyperbolic cosine of the prediction error. — loss_log_cosh","title":"Computes the logarithm of the hyperbolic cosine of the prediction error. — loss_log_cosh","text":"Formula:   Note log(cosh(x)) approximately equal (x ** 2) / 2 small x abs(x) - log(2) large x. means 'logcosh' works mostly like mean squared error, strongly affected occasional wildly incorrect prediction.","code":"loss <- mean(log(cosh(y_pred - y_true)), axis=-1)"},{"path":"https://keras3.posit.co/dev/reference/loss_log_cosh.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes the logarithm of the hyperbolic cosine of the prediction error. — loss_log_cosh","text":"","code":"loss_log_cosh(   y_true,   y_pred,   ...,   reduction = \"sum_over_batch_size\",   name = \"log_cosh\",   dtype = NULL )"},{"path":"https://keras3.posit.co/dev/reference/loss_log_cosh.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Computes the logarithm of the hyperbolic cosine of the prediction error. — loss_log_cosh","text":"y_true Ground truth values shape = [batch_size, d0, .. dN]. y_pred predicted values shape = [batch_size, d0, .. dN]. ... forward/backward compatability. reduction Type reduction apply loss. Options \"sum\", \"sum_over_batch_size\" NULL. Defaults \"sum_over_batch_size\". name Optional name instance. dtype dtype loss's computations. Defaults NULL, means using config_floatx(). config_floatx() \"float32\" unless set different value (via config_set_floatx()). keras$DTypePolicy provided, compute_dtype utilized.","code":""},{"path":"https://keras3.posit.co/dev/reference/loss_log_cosh.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Computes the logarithm of the hyperbolic cosine of the prediction error. — loss_log_cosh","text":"Logcosh error values shape = [batch_size, d0, .. dN-1].","code":""},{"path":"https://keras3.posit.co/dev/reference/loss_log_cosh.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Computes the logarithm of the hyperbolic cosine of the prediction error. — loss_log_cosh","text":"","code":"y_true <- rbind(c(0., 1.), c(0., 0.)) y_pred <- rbind(c(1., 1.), c(0., 0.)) loss <- loss_log_cosh(y_true, y_pred) # 0.108"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/loss_mean_absolute_error.html","id":null,"dir":"Reference","previous_headings":"","what":"Computes the mean of absolute difference between labels and predictions. — loss_mean_absolute_error","title":"Computes the mean of absolute difference between labels and predictions. — loss_mean_absolute_error","text":"Formula:","code":"loss <- mean(abs(y_true - y_pred))"},{"path":"https://keras3.posit.co/dev/reference/loss_mean_absolute_error.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes the mean of absolute difference between labels and predictions. — loss_mean_absolute_error","text":"","code":"loss_mean_absolute_error(   y_true,   y_pred,   ...,   reduction = \"sum_over_batch_size\",   name = \"mean_absolute_error\",   dtype = NULL )"},{"path":"https://keras3.posit.co/dev/reference/loss_mean_absolute_error.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Computes the mean of absolute difference between labels and predictions. — loss_mean_absolute_error","text":"y_true Ground truth values shape = [batch_size, d0, .. dN]. y_pred predicted values shape = [batch_size, d0, .. dN]. ... forward/backward compatability. reduction Type reduction apply loss. almost cases \"sum_over_batch_size\". Supported options \"sum\", \"sum_over_batch_size\", \"mean\", \"mean_with_sample_weight\" NULL. \"sum\" sums loss, \"sum_over_batch_size\" \"mean\" sum loss divide sample size, \"mean_with_sample_weight\" sums loss divides sum sample weights. \"none\" NULL perform aggregation. Defaults \"sum_over_batch_size\". name Optional name loss instance. dtype dtype loss's computations. Defaults NULL, means using config_floatx(). config_floatx() \"float32\" unless set different value (via config_set_floatx()). keras$DTypePolicy provided, compute_dtype utilized.","code":""},{"path":"https://keras3.posit.co/dev/reference/loss_mean_absolute_error.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Computes the mean of absolute difference between labels and predictions. — loss_mean_absolute_error","text":"Mean absolute error values shape = [batch_size, d0, .. dN-1].","code":""},{"path":"https://keras3.posit.co/dev/reference/loss_mean_absolute_error.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Computes the mean of absolute difference between labels and predictions. — loss_mean_absolute_error","text":"","code":"y_true <- random_uniform(c(2, 3), 0, 2) y_pred <- random_uniform(c(2, 3)) loss <- loss_mean_absolute_error(y_true, y_pred)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/loss_mean_absolute_percentage_error.html","id":null,"dir":"Reference","previous_headings":"","what":"Computes the mean absolute percentage error between y_true and y_pred. — loss_mean_absolute_percentage_error","title":"Computes the mean absolute percentage error between y_true and y_pred. — loss_mean_absolute_percentage_error","text":"Formula:   Division zero prevented dividing max(y_true, epsilon) epsilon = config_epsilon() (default 1e-7).","code":"loss <- 100 * op_mean(op_abs((y_true - y_pred) / y_true),                       axis=-1)"},{"path":"https://keras3.posit.co/dev/reference/loss_mean_absolute_percentage_error.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes the mean absolute percentage error between y_true and y_pred. — loss_mean_absolute_percentage_error","text":"","code":"loss_mean_absolute_percentage_error(   y_true,   y_pred,   ...,   reduction = \"sum_over_batch_size\",   name = \"mean_absolute_percentage_error\",   dtype = NULL )"},{"path":"https://keras3.posit.co/dev/reference/loss_mean_absolute_percentage_error.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Computes the mean absolute percentage error between y_true and y_pred. — loss_mean_absolute_percentage_error","text":"y_true Ground truth values shape = [batch_size, d0, .. dN]. y_pred predicted values shape = [batch_size, d0, .. dN]. ... forward/backward compatability. reduction Type reduction apply loss. almost cases \"sum_over_batch_size\". Supported options \"sum\", \"sum_over_batch_size\", \"mean\", \"mean_with_sample_weight\" NULL. \"sum\" sums loss, \"sum_over_batch_size\" \"mean\" sum loss divide sample size, \"mean_with_sample_weight\" sums loss divides sum sample weights. \"none\" NULL perform aggregation. Defaults \"sum_over_batch_size\". name Optional name loss instance. dtype dtype loss's computations. Defaults NULL, means using config_floatx(). config_floatx() \"float32\" unless set different value (via config_set_floatx()). keras$DTypePolicy provided, compute_dtype utilized.","code":""},{"path":"https://keras3.posit.co/dev/reference/loss_mean_absolute_percentage_error.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Computes the mean absolute percentage error between y_true and y_pred. — loss_mean_absolute_percentage_error","text":"Mean absolute percentage error values shape = [batch_size, d0, ..dN-1].","code":""},{"path":"https://keras3.posit.co/dev/reference/loss_mean_absolute_percentage_error.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Computes the mean absolute percentage error between y_true and y_pred. — loss_mean_absolute_percentage_error","text":"","code":"y_true <- random_uniform(c(2, 3)) y_pred <- random_uniform(c(2, 3)) loss <- loss_mean_absolute_percentage_error(y_true, y_pred)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/loss_mean_squared_error.html","id":null,"dir":"Reference","previous_headings":"","what":"Computes the mean of squares of errors between labels and predictions. — loss_mean_squared_error","title":"Computes the mean of squares of errors between labels and predictions. — loss_mean_squared_error","text":"Formula:","code":"loss <- mean(square(y_true - y_pred))"},{"path":"https://keras3.posit.co/dev/reference/loss_mean_squared_error.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes the mean of squares of errors between labels and predictions. — loss_mean_squared_error","text":"","code":"loss_mean_squared_error(   y_true,   y_pred,   ...,   reduction = \"sum_over_batch_size\",   name = \"mean_squared_error\",   dtype = NULL )"},{"path":"https://keras3.posit.co/dev/reference/loss_mean_squared_error.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Computes the mean of squares of errors between labels and predictions. — loss_mean_squared_error","text":"y_true Ground truth values shape = [batch_size, d0, .. dN]. y_pred predicted values shape = [batch_size, d0, .. dN]. ... forward/backward compatability. reduction Type reduction apply loss. almost cases \"sum_over_batch_size\". Supported options \"sum\", \"sum_over_batch_size\", \"mean\", \"mean_with_sample_weight\" NULL. \"sum\" sums loss, \"sum_over_batch_size\" \"mean\" sum loss divide sample size, \"mean_with_sample_weight\" sums loss divides sum sample weights. \"none\" NULL perform aggregation. Defaults \"sum_over_batch_size\". name Optional name loss instance. dtype dtype loss's computations. Defaults NULL, means using config_floatx(). config_floatx() \"float32\" unless set different value (via config_set_floatx()). keras$DTypePolicy provided, compute_dtype utilized.","code":""},{"path":"https://keras3.posit.co/dev/reference/loss_mean_squared_error.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Computes the mean of squares of errors between labels and predictions. — loss_mean_squared_error","text":"Mean squared error values shape = [batch_size, d0, .. dN-1].","code":""},{"path":"https://keras3.posit.co/dev/reference/loss_mean_squared_error.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Computes the mean of squares of errors between labels and predictions. — loss_mean_squared_error","text":"","code":"y_true <- random_uniform(c(2, 3), 0, 2) y_pred <- random_uniform(c(2, 3)) loss <- loss_mean_squared_error(y_true, y_pred)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/loss_mean_squared_logarithmic_error.html","id":null,"dir":"Reference","previous_headings":"","what":"Computes the mean squared logarithmic error between y_true and y_pred. — loss_mean_squared_logarithmic_error","title":"Computes the mean squared logarithmic error between y_true and y_pred. — loss_mean_squared_logarithmic_error","text":"Note y_pred y_true less equal 0. Negative values 0 values replaced config_epsilon() (default 1e-7). Formula:","code":"loss <- mean(square(log(y_true + 1) - log(y_pred + 1)))"},{"path":"https://keras3.posit.co/dev/reference/loss_mean_squared_logarithmic_error.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes the mean squared logarithmic error between y_true and y_pred. — loss_mean_squared_logarithmic_error","text":"","code":"loss_mean_squared_logarithmic_error(   y_true,   y_pred,   ...,   reduction = \"sum_over_batch_size\",   name = \"mean_squared_logarithmic_error\",   dtype = NULL )"},{"path":"https://keras3.posit.co/dev/reference/loss_mean_squared_logarithmic_error.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Computes the mean squared logarithmic error between y_true and y_pred. — loss_mean_squared_logarithmic_error","text":"y_true Ground truth values shape = [batch_size, d0, .. dN]. y_pred predicted values shape = [batch_size, d0, .. dN]. ... forward/backward compatability. reduction Type reduction apply loss. almost cases \"sum_over_batch_size\". Supported options \"sum\", \"sum_over_batch_size\", \"mean\", \"mean_with_sample_weight\" NULL. \"sum\" sums loss, \"sum_over_batch_size\" \"mean\" sum loss divide sample size, \"mean_with_sample_weight\" sums loss divides sum sample weights. \"none\" NULL perform aggregation. Defaults \"sum_over_batch_size\". name Optional name loss instance. dtype dtype loss's computations. Defaults NULL, means using config_floatx(). config_floatx() \"float32\" unless set different value (via config_set_floatx()). keras$DTypePolicy provided, compute_dtype utilized.","code":""},{"path":"https://keras3.posit.co/dev/reference/loss_mean_squared_logarithmic_error.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Computes the mean squared logarithmic error between y_true and y_pred. — loss_mean_squared_logarithmic_error","text":"Mean squared logarithmic error values shape = [batch_size, d0, .. dN-1].","code":""},{"path":"https://keras3.posit.co/dev/reference/loss_mean_squared_logarithmic_error.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Computes the mean squared logarithmic error between y_true and y_pred. — loss_mean_squared_logarithmic_error","text":"","code":"y_true <- random_uniform(c(2, 3), 0, 2) y_pred <- random_uniform(c(2, 3)) loss <- loss_mean_squared_logarithmic_error(y_true, y_pred)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/loss_poisson.html","id":null,"dir":"Reference","previous_headings":"","what":"Computes the Poisson loss between y_true & y_pred. — loss_poisson","title":"Computes the Poisson loss between y_true & y_pred. — loss_poisson","text":"Formula:","code":"loss <- y_pred - y_true * log(y_pred)"},{"path":"https://keras3.posit.co/dev/reference/loss_poisson.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes the Poisson loss between y_true & y_pred. — loss_poisson","text":"","code":"loss_poisson(   y_true,   y_pred,   ...,   reduction = \"sum_over_batch_size\",   name = \"poisson\",   dtype = NULL )"},{"path":"https://keras3.posit.co/dev/reference/loss_poisson.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Computes the Poisson loss between y_true & y_pred. — loss_poisson","text":"y_true Ground truth values. shape = [batch_size, d0, .. dN]. y_pred predicted values. shape = [batch_size, d0, .. dN]. ... forward/backward compatability. reduction Type reduction apply loss. almost cases \"sum_over_batch_size\". Supported options \"sum\", \"sum_over_batch_size\", \"mean\", \"mean_with_sample_weight\" NULL. \"sum\" sums loss, \"sum_over_batch_size\" \"mean\" sum loss divide sample size, \"mean_with_sample_weight\" sums loss divides sum sample weights. \"none\" NULL perform aggregation. Defaults \"sum_over_batch_size\". name Optional name loss instance. dtype dtype loss's computations. Defaults NULL, means using config_floatx(). config_floatx() \"float32\" unless set different value (via config_set_floatx()). keras$DTypePolicy provided, compute_dtype utilized.","code":""},{"path":"https://keras3.posit.co/dev/reference/loss_poisson.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Computes the Poisson loss between y_true & y_pred. — loss_poisson","text":"Poisson loss values shape = [batch_size, d0, .. dN-1].","code":""},{"path":"https://keras3.posit.co/dev/reference/loss_poisson.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Computes the Poisson loss between y_true & y_pred. — loss_poisson","text":"","code":"y_true <- random_uniform(c(2, 3), 0, 2) y_pred <- random_uniform(c(2, 3)) loss <- loss_poisson(y_true, y_pred) loss ## tf.Tensor([1.6422468  0.81166863], shape=(2), dtype=float32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/loss_sparse_categorical_crossentropy.html","id":null,"dir":"Reference","previous_headings":"","what":"Computes the crossentropy loss between the labels and predictions. — loss_sparse_categorical_crossentropy","title":"Computes the crossentropy loss between the labels and predictions. — loss_sparse_categorical_crossentropy","text":"Use crossentropy loss function two label classes.  expect labels provided integers. want provide labels using one-hot representation, please use CategoricalCrossentropy loss.  # classes floating point values per feature y_pred single floating point value per feature y_true. snippet , single floating point value per example y_true num_classes floating pointing values per example y_pred. shape y_true [batch_size] shape y_pred [batch_size, num_classes].","code":""},{"path":"https://keras3.posit.co/dev/reference/loss_sparse_categorical_crossentropy.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes the crossentropy loss between the labels and predictions. — loss_sparse_categorical_crossentropy","text":"","code":"loss_sparse_categorical_crossentropy(   y_true,   y_pred,   from_logits = FALSE,   ignore_class = NULL,   axis = -1L,   ...,   reduction = \"sum_over_batch_size\",   name = \"sparse_categorical_crossentropy\",   dtype = NULL )"},{"path":"https://keras3.posit.co/dev/reference/loss_sparse_categorical_crossentropy.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Computes the crossentropy loss between the labels and predictions. — loss_sparse_categorical_crossentropy","text":"y_true Ground truth values. y_pred predicted values. from_logits Whether y_pred expected logits tensor. default, assume y_pred encodes probability distribution. ignore_class Optional integer. ID class ignored loss computation. useful, example, segmentation problems featuring \"void\" class (commonly -1 255) segmentation maps. default (ignore_class=NULL), classes considered. axis Defaults -1. dimension along entropy computed. ... forward/backward compatability. reduction Type reduction apply loss. almost cases \"sum_over_batch_size\". Supported options \"sum\", \"sum_over_batch_size\", \"mean\", \"mean_with_sample_weight\" NULL. \"sum\" sums loss, \"sum_over_batch_size\" \"mean\" sum loss divide sample size, \"mean_with_sample_weight\" sums loss divides sum sample weights. \"none\" NULL perform aggregation. Defaults \"sum_over_batch_size\". name Optional name loss instance. dtype dtype loss's computations. Defaults NULL, means using config_floatx(). config_floatx() \"float32\" unless set different value (via config_set_floatx()). keras$DTypePolicy provided, compute_dtype utilized.","code":""},{"path":"https://keras3.posit.co/dev/reference/loss_sparse_categorical_crossentropy.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Computes the crossentropy loss between the labels and predictions. — loss_sparse_categorical_crossentropy","text":"Sparse categorical crossentropy loss value.","code":""},{"path":"https://keras3.posit.co/dev/reference/loss_sparse_categorical_crossentropy.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Computes the crossentropy loss between the labels and predictions. — loss_sparse_categorical_crossentropy","text":"Usage compile() API:","code":"y_true <- c(1, 2) y_pred <- rbind(c(0.05, 0.95, 0), c(0.1, 0.8, 0.1)) loss <- loss_sparse_categorical_crossentropy(y_true, y_pred) loss ## tf.Tensor([0.05129344 2.3025851 ], shape=(2), dtype=float32) y_true <- c(1, 2) y_pred <- rbind(c(0.05, 0.95, 0), c(0.1, 0.8, 0.1)) # Using 'auto'/'sum_over_batch_size' reduction type. scce <- loss_sparse_categorical_crossentropy() scce(op_array(y_true), op_array(y_pred)) ## tf.Tensor(1.1769392, shape=(), dtype=float32) # 1.177 # Calling with 'sample_weight'. scce(op_array(y_true), op_array(y_pred), sample_weight = op_array(c(0.3, 0.7))) ## tf.Tensor(0.8135988, shape=(), dtype=float32) # Using 'sum' reduction type. scce <- loss_sparse_categorical_crossentropy(reduction=\"sum\") scce(op_array(y_true), op_array(y_pred)) ## tf.Tensor(2.3538785, shape=(), dtype=float32) # 2.354 # Using 'none' reduction type. scce <- loss_sparse_categorical_crossentropy(reduction=NULL) scce(op_array(y_true), op_array(y_pred)) ## tf.Tensor([0.05129344 2.3025851 ], shape=(2), dtype=float32) # array([0.0513, 2.303], dtype=float32) model %>% compile(optimizer = 'sgd',                   loss = loss_sparse_categorical_crossentropy())"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/loss_squared_hinge.html","id":null,"dir":"Reference","previous_headings":"","what":"Computes the squared hinge loss between y_true & y_pred. — loss_squared_hinge","title":"Computes the squared hinge loss between y_true & y_pred. — loss_squared_hinge","text":"Formula:   y_true values expected -1 1. binary (0 1) labels provided convert -1 1.","code":"loss <- square(maximum(1 - y_true * y_pred, 0))"},{"path":"https://keras3.posit.co/dev/reference/loss_squared_hinge.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes the squared hinge loss between y_true & y_pred. — loss_squared_hinge","text":"","code":"loss_squared_hinge(   y_true,   y_pred,   ...,   reduction = \"sum_over_batch_size\",   name = \"squared_hinge\",   dtype = NULL )"},{"path":"https://keras3.posit.co/dev/reference/loss_squared_hinge.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Computes the squared hinge loss between y_true & y_pred. — loss_squared_hinge","text":"y_true ground truth values. y_true values expected -1 1. binary (0 1) labels provided convert -1 1 shape = [batch_size, d0, .. dN]. y_pred predicted values shape = [batch_size, d0, .. dN]. ... forward/backward compatability. reduction Type reduction apply loss. almost cases \"sum_over_batch_size\". Supported options \"sum\", \"sum_over_batch_size\", \"mean\", \"mean_with_sample_weight\" NULL. \"sum\" sums loss, \"sum_over_batch_size\" \"mean\" sum loss divide sample size, \"mean_with_sample_weight\" sums loss divides sum sample weights. \"none\" NULL perform aggregation. Defaults \"sum_over_batch_size\". name Optional name loss instance. dtype dtype loss's computations. Defaults NULL, means using config_floatx(). config_floatx() \"float32\" unless set different value (via config_set_floatx()). keras$DTypePolicy provided, compute_dtype utilized.","code":""},{"path":"https://keras3.posit.co/dev/reference/loss_squared_hinge.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Computes the squared hinge loss between y_true & y_pred. — loss_squared_hinge","text":"Squared hinge loss values shape = [batch_size, d0, .. dN-1].","code":""},{"path":"https://keras3.posit.co/dev/reference/loss_squared_hinge.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Computes the squared hinge loss between y_true & y_pred. — loss_squared_hinge","text":"","code":"y_true <- array(sample(c(-1,1), 6, replace = TRUE), dim = c(2, 3)) y_pred <- random_uniform(c(2, 3)) loss <- loss_squared_hinge(y_true, y_pred)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/loss_tversky.html","id":null,"dir":"Reference","previous_headings":"","what":"Computes the Tversky loss value between y_true and y_pred. — loss_tversky","title":"Computes the Tversky loss value between y_true and y_pred. — loss_tversky","text":"loss function weighted alpha beta coefficients penalize false positives false negatives. alpha=0.5 beta=0.5, loss value becomes equivalent Dice Loss. loss function weighted alpha beta coefficients penalize false positives false negatives. alpha=0.5 beta=0.5, loss value becomes equivalent Dice Loss.","code":""},{"path":"https://keras3.posit.co/dev/reference/loss_tversky.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes the Tversky loss value between y_true and y_pred. — loss_tversky","text":"","code":"loss_tversky(   y_true,   y_pred,   ...,   alpha = 0.5,   beta = 0.5,   reduction = \"sum_over_batch_size\",   name = \"tversky\",   axis = NULL,   dtype = NULL )"},{"path":"https://keras3.posit.co/dev/reference/loss_tversky.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Computes the Tversky loss value between y_true and y_pred. — loss_tversky","text":"y_true tensor true targets. y_pred tensor predicted targets. ... forward/backward compatability. alpha coefficient controlling incidence false positives. Defaults 0.5. beta coefficient controlling incidence false negatives. Defaults 0.5. reduction Type reduction apply loss. almost cases \"sum_over_batch_size\". Supported options \"sum\", \"sum_over_batch_size\", \"mean\", \"mean_with_sample_weight\" NULL. \"sum\" sums loss, \"sum_over_batch_size\" \"mean\" sum loss divide sample size, \"mean_with_sample_weight\" sums loss divides sum sample weights. \"none\" NULL perform aggregation. Defaults \"sum_over_batch_size\". name Optional name loss instance. (string) axis Axis (1-based) dtype dtype loss's computations. Defaults NULL, means using config_floatx(). config_floatx() \"float32\" unless set different value (via config_set_floatx()). keras$DTypePolicy provided, compute_dtype utilized.","code":""},{"path":"https://keras3.posit.co/dev/reference/loss_tversky.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Computes the Tversky loss value between y_true and y_pred. — loss_tversky","text":"Tversky loss value.","code":""},{"path":"https://keras3.posit.co/dev/reference/loss_tversky.html","id":"reference","dir":"Reference","previous_headings":"","what":"Reference","title":"Computes the Tversky loss value between y_true and y_pred. — loss_tversky","text":"Salehi et al., 2017","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/mark_active.html","id":null,"dir":"Reference","previous_headings":"","what":"active_property — mark_active","title":"active_property — mark_active","text":"mark_active() alias active_property(). See ?active_property() full documentation.","code":""},{"path":"https://keras3.posit.co/dev/reference/mark_active.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"active_property — mark_active","text":"","code":"mark_active(fn)"},{"path":"https://keras3.posit.co/dev/reference/mark_active.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"active_property — mark_active","text":"fn R function","code":""},{"path":"https://keras3.posit.co/dev/reference/mark_active.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"active_property — mark_active","text":"fn, additional R attribute cause fn converted active property converted method custom subclass.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_auc.html","id":null,"dir":"Reference","previous_headings":"","what":"Approximates the AUC (Area under the curve) of the ROC or PR curves. — metric_auc","title":"Approximates the AUC (Area under the curve) of the ROC or PR curves. — metric_auc","text":"AUC (Area curve) ROC (Receiver operating characteristic; default) PR (Precision Recall) curves quality measures binary classifiers. Unlike accuracy, like cross-entropy losses, ROC-AUC PR-AUC evaluate operational points model. class approximates AUCs using Riemann sum. metric accumulation phrase, predictions accumulated within predefined buckets value. AUC computed interpolating per-bucket averages. buckets define evaluated operational points. metric creates four local variables, true_positives, true_negatives, false_positives false_negatives used compute AUC.  discretize AUC curve, linearly spaced set thresholds used compute pairs recall precision values. area ROC-curve therefore computed using height recall values false positive rate, area PR-curve computed using height precision values recall. value ultimately returned auc, idempotent operation computes area discretized curve precision versus recall values (computed using aforementioned variables). num_thresholds variable controls degree discretization larger numbers thresholds closely approximating true AUC. quality approximation may vary dramatically depending num_thresholds. thresholds parameter can used manually specify thresholds split predictions evenly. best approximation real AUC, predictions distributed approximately uniformly range [0, 1] (from_logits=FALSE). quality AUC approximation may poor case. Setting summation_method 'minoring' 'majoring' can help quantify error approximation providing lower upper bound estimate AUC. sample_weight NULL, weights default 1. Use sample_weight 0 mask values.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_auc.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Approximates the AUC (Area under the curve) of the ROC or PR curves. — metric_auc","text":"","code":"metric_auc(   ...,   num_thresholds = 200L,   curve = \"ROC\",   summation_method = \"interpolation\",   name = NULL,   dtype = NULL,   thresholds = NULL,   multi_label = FALSE,   num_labels = NULL,   label_weights = NULL,   from_logits = FALSE )"},{"path":"https://keras3.posit.co/dev/reference/metric_auc.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Approximates the AUC (Area under the curve) of the ROC or PR curves. — metric_auc","text":"... forward/backward compatability. num_thresholds (Optional) number thresholds use discretizing roc curve. Values must > 1. Defaults 200. curve (Optional) Specifies name curve computed, 'ROC' (default) 'PR' Precision-Recall-curve. summation_method (Optional) Specifies Riemann summation method used. 'interpolation' (default) applies mid-point summation scheme ROC.  PR-AUC, interpolates (true/false) positives ratio precision (see Davis & Goadrich 2006 details); 'minoring' applies left summation increasing intervals right summation decreasing intervals; 'majoring' opposite. name (Optional) string name metric instance. dtype (Optional) data type metric result. thresholds (Optional) list floating point values use thresholds discretizing curve. set, num_thresholds parameter ignored. Values [0, 1]. Endpoint thresholds equal {-epsilon, 1+epsilon} small positive epsilon value automatically included correctly handle predictions equal exactly 0 1. multi_label boolean indicating whether multilabel data treated , wherein AUC computed separately label averaged across labels, (FALSE) data flattened single label AUC computation. latter case, multilabel data passed AUC, label-prediction pair treated individual data point. set `FALSE“ multi-class data. num_labels (Optional) number labels, used multi_label TRUE. num_labels specified, state variables get created first call update_state. label_weights (Optional) list, array, tensor non-negative weights used compute AUCs multilabel data. multi_label TRUE, weights applied individual label AUCs averaged produce multi-label AUC. FALSE, used weight individual label predictions computing confusion matrix flattened data. Note unlike class_weights class_weights weights example depending value label, whereas label_weights depends index label flattening; therefore label_weights used multi-class data. from_logits boolean indicating whether predictions (y_pred update_state) probabilities sigmoid logits. rule thumb, using keras loss, from_logits constructor argument loss match AUC from_logits constructor argument.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_auc.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Approximates the AUC (Area under the curve) of the ROC or PR curves. — metric_auc","text":"Metric instance returned. Metric instance can passed directly compile(metrics = ), used standalone object. See ?Metric example usage.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_auc.html","id":"usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Approximates the AUC (Area under the curve) of the ROC or PR curves. — metric_auc","text":"Standalone usage:         Usage compile() API:","code":"m <- metric_auc(num_thresholds = 3) m$update_state(c(0,   0,   1,   1),                c(0, 0.5, 0.3, 0.9)) # threshold values are [0 - 1e-7, 0.5, 1 + 1e-7] # tp = [2, 1, 0], fp = [2, 0, 0], fn = [0, 1, 2], tn = [0, 2, 2] # tp_rate = recall = [1, 0.5, 0], fp_rate = [1, 0, 0] # auc = ((((1 + 0.5) / 2) * (1 - 0)) + (((0.5 + 0) / 2) * (0 - 0))) #     = 0.75 m$result() ## tf.Tensor(0.75, shape=(), dtype=float32) m$reset_state() m$update_state(c(0,   0,   1,   1),                c(0, 0.5, 0.3, 0.9),                sample_weight=c(1, 0, 0, 1)) m$result() ## tf.Tensor(1.0, shape=(), dtype=float32) # Reports the AUC of a model outputting a probability. model |> compile(   optimizer = 'sgd',   loss = loss_binary_crossentropy(),   metrics = list(metric_auc()) )  # Reports the AUC of a model outputting a logit. model |> compile(   optimizer = 'sgd',   loss = loss_binary_crossentropy(from_logits = TRUE),   metrics = list(metric_auc(from_logits = TRUE)) )"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/metric_binary_accuracy.html","id":null,"dir":"Reference","previous_headings":"","what":"Calculates how often predictions match binary labels. — metric_binary_accuracy","title":"Calculates how often predictions match binary labels. — metric_binary_accuracy","text":"metric creates two local variables, total count used compute frequency y_pred matches y_true. frequency ultimately returned binary accuracy: idempotent operation simply divides total count. sample_weight NULL, weights default 1. Use sample_weight 0 mask values.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_binary_accuracy.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Calculates how often predictions match binary labels. — metric_binary_accuracy","text":"","code":"metric_binary_accuracy(   y_true,   y_pred,   threshold = 0.5,   ...,   name = \"binary_accuracy\",   dtype = NULL )"},{"path":"https://keras3.posit.co/dev/reference/metric_binary_accuracy.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Calculates how often predictions match binary labels. — metric_binary_accuracy","text":"y_true Tensor true targets. y_pred Tensor predicted targets. threshold (Optional) Float representing threshold deciding whether prediction values 1 0. ... forward/backward compatability. name (Optional) string name metric instance. dtype (Optional) data type metric result.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_binary_accuracy.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Calculates how often predictions match binary labels. — metric_binary_accuracy","text":"y_true y_pred missing, Metric instance returned. Metric instance can passed directly compile(metrics = ), used standalone object. See ?Metric example usage. y_true y_pred provided, tensor computed value returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_binary_accuracy.html","id":"usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Calculates how often predictions match binary labels. — metric_binary_accuracy","text":"Standalone usage:             Usage compile() API:","code":"m <- metric_binary_accuracy() m$update_state(rbind(1, 1, 0, 0), rbind(0.98, 1, 0, 0.6)) m$result() ## tf.Tensor(0.75, shape=(), dtype=float32) # 0.75 m$reset_state() m$update_state(rbind(1, 1, 0, 0), rbind(0.98, 1, 0, 0.6),                sample_weight = c(1, 0, 0, 1)) m$result() ## tf.Tensor(0.5, shape=(), dtype=float32) # 0.5 model %>% compile(optimizer='sgd',                   loss='binary_crossentropy',                   metrics=list(metric_binary_accuracy()))"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/metric_binary_crossentropy.html","id":null,"dir":"Reference","previous_headings":"","what":"Computes the crossentropy metric between the labels and predictions. — metric_binary_crossentropy","title":"Computes the crossentropy metric between the labels and predictions. — metric_binary_crossentropy","text":"crossentropy metric class used two label classes (0 1).","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_binary_crossentropy.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes the crossentropy metric between the labels and predictions. — metric_binary_crossentropy","text":"","code":"metric_binary_crossentropy(   y_true,   y_pred,   from_logits = FALSE,   label_smoothing = 0,   axis = -1L,   ...,   name = \"binary_crossentropy\",   dtype = NULL )"},{"path":"https://keras3.posit.co/dev/reference/metric_binary_crossentropy.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Computes the crossentropy metric between the labels and predictions. — metric_binary_crossentropy","text":"y_true Ground truth values. shape = [batch_size, d0, .. dN]. y_pred predicted values. shape = [batch_size, d0, .. dN]. from_logits (Optional) Whether output expected logits tensor. default, consider output encodes probability distribution. label_smoothing (Optional) Float [0, 1]. > 0, label values smoothed, meaning confidence label values relaxed. e.g. label_smoothing=0.2 means use value 0.1 label \"0\" 0.9 label \"1\". axis axis along mean computed. Defaults -1. ... forward/backward compatability. name (Optional) string name metric instance. dtype (Optional) data type metric result.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_binary_crossentropy.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Computes the crossentropy metric between the labels and predictions. — metric_binary_crossentropy","text":"y_true y_pred missing, Metric instance returned. Metric instance can passed directly compile(metrics = ), used standalone object. See ?Metric example usage. y_true y_pred provided, tensor computed value returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_binary_crossentropy.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Computes the crossentropy metric between the labels and predictions. — metric_binary_crossentropy","text":"Standalone usage:         Usage compile() API:","code":"m <- metric_binary_crossentropy() m$update_state(rbind(c(0, 1), c(0, 0)), rbind(c(0.6, 0.4), c(0.4, 0.6))) m$result() ## tf.Tensor(0.8149245, shape=(), dtype=float32) m$reset_state() m$update_state(rbind(c(0, 1), c(0, 0)), rbind(c(0.6, 0.4), c(0.4, 0.6)),                sample_weight=c(1, 0)) m$result() ## tf.Tensor(0.91629076, shape=(), dtype=float32) model %>% compile(     optimizer = 'sgd',     loss = 'mse',     metrics = list(metric_binary_crossentropy()))"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/metric_binary_focal_crossentropy.html","id":null,"dir":"Reference","previous_headings":"","what":"Computes the binary focal crossentropy loss. — metric_binary_focal_crossentropy","title":"Computes the binary focal crossentropy loss. — metric_binary_focal_crossentropy","text":"According Lin et al., 2018, helps apply focal factor -weight easy examples focus hard examples. default, focal tensor computed follows: focal_factor = (1 - output)^gamma class 1 focal_factor = output^gamma class 0 gamma focusing parameter. gamma = 0, focal effect binary crossentropy loss. apply_class_balancing == TRUE, function also takes account weight balancing factor binary classes 0 1 follows: weight = alpha class 1 (target == 1) weight = 1 - alpha class 0 alpha float range [0, 1].","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_binary_focal_crossentropy.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes the binary focal crossentropy loss. — metric_binary_focal_crossentropy","text":"","code":"metric_binary_focal_crossentropy(   y_true,   y_pred,   apply_class_balancing = FALSE,   alpha = 0.25,   gamma = 2,   from_logits = FALSE,   label_smoothing = 0,   axis = -1L )"},{"path":"https://keras3.posit.co/dev/reference/metric_binary_focal_crossentropy.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Computes the binary focal crossentropy loss. — metric_binary_focal_crossentropy","text":"y_true Ground truth values, shape (batch_size, d0, .. dN). y_pred predicted values, shape (batch_size, d0, .. dN). apply_class_balancing bool, whether apply weight balancing binary classes 0 1. alpha weight balancing factor class 1, default 0.25 mentioned reference. weight class 0 1.0 - alpha. gamma focusing parameter, default 2.0 mentioned reference. from_logits Whether y_pred expected logits tensor. default, assume y_pred encodes probability distribution. label_smoothing Float [0, 1]. > 0 smooth labels squeezing towards 0.5, , using 1. - 0.5 * label_smoothing target class 0.5 * label_smoothing non-target class. axis axis along mean computed. Defaults -1.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_binary_focal_crossentropy.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Computes the binary focal crossentropy loss. — metric_binary_focal_crossentropy","text":"Binary focal crossentropy loss value shape = [batch_size, d0, .. dN-1].","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_binary_focal_crossentropy.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Computes the binary focal crossentropy loss. — metric_binary_focal_crossentropy","text":"","code":"y_true <- rbind(c(0, 1), c(0, 0)) y_pred <- rbind(c(0.6, 0.4), c(0.4, 0.6)) loss <- loss_binary_focal_crossentropy(y_true, y_pred, gamma=2) loss ## tf.Tensor([0.32986468 0.20579839], shape=(2), dtype=float32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/metric_binary_iou.html","id":null,"dir":"Reference","previous_headings":"","what":"Computes the Intersection-Over-Union metric for class 0 and/or 1. — metric_binary_iou","title":"Computes the Intersection-Over-Union metric for class 0 and/or 1. — metric_binary_iou","text":"Formula:   Intersection--Union common evaluation metric semantic image segmentation. compute IoUs, predictions accumulated confusion matrix, weighted sample_weight metric calculated . sample_weight NULL, weights default 1. Use sample_weight 0 mask values. class can used compute IoUs binary classification task predictions provided logits. First threshold applied predicted values threshold converted class 0 threshold converted class 1. IoUs classes 0 1 computed, mean IoUs classes specified target_class_ids returned.","code":"iou <- true_positives / (true_positives + false_positives + false_negatives)"},{"path":"https://keras3.posit.co/dev/reference/metric_binary_iou.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes the Intersection-Over-Union metric for class 0 and/or 1. — metric_binary_iou","text":"","code":"metric_binary_iou(   ...,   target_class_ids = list(0L, 1L),   threshold = 0.5,   name = NULL,   dtype = NULL )"},{"path":"https://keras3.posit.co/dev/reference/metric_binary_iou.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Computes the Intersection-Over-Union metric for class 0 and/or 1. — metric_binary_iou","text":"... forward/backward compatability. target_class_ids list list target class ids metric returned. Options 0, 1, c(0, 1). 0 (1), IoU metric class 0 (class 1, respectively) returned. c(0, 1), mean IoUs two classes returned. threshold threshold applies prediction logits convert either predicted class 0 logit threshold predicted class 1 logit threshold. name (Optional) string name metric instance. dtype (Optional) data type metric result.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_binary_iou.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Computes the Intersection-Over-Union metric for class 0 and/or 1. — metric_binary_iou","text":"Metric instance returned. Metric instance can passed directly compile(metrics = ), used standalone object. See ?Metric example usage.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_binary_iou.html","id":"note","dir":"Reference","previous_headings":"","what":"Note","title":"Computes the Intersection-Over-Union metric for class 0 and/or 1. — metric_binary_iou","text":"threshold=0, metric behavior IoU.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_binary_iou.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Computes the Intersection-Over-Union metric for class 0 and/or 1. — metric_binary_iou","text":"Standalone usage:                 Usage compile() API:","code":"m <- metric_binary_iou(target_class_ids=c(0L, 1L), threshold = 0.3) m$update_state(c(0, 1, 0, 1), c(0.1, 0.2, 0.4, 0.7)) ## tf.Tensor( ## [[1 1] ##  [1 1]], shape=(2, 2), dtype=int64) m$result() ## tf.Tensor(0.33333334, shape=(), dtype=float32) m$reset_state() m$update_state(c(0, 1, 0, 1), c(0.1, 0.2, 0.4, 0.7),                sample_weight = 10 * c(0.2, 0.3, 0.4, 0.1)) ## tf.Tensor( ## [[2 4] ##  [3 1]], shape=(2, 2), dtype=int64) m$result() ## tf.Tensor(0.1736111, shape=(), dtype=float32) model %>% compile(     optimizer = 'sgd',     loss = 'mse',     metrics = list(metric_binary_iou(         target_class_ids = 0L,         threshold = 0.5     )) )"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/metric_categorical_accuracy.html","id":null,"dir":"Reference","previous_headings":"","what":"Calculates how often predictions match one-hot labels. — metric_categorical_accuracy","title":"Calculates how often predictions match one-hot labels. — metric_categorical_accuracy","text":"can provide logits classes y_pred, since argmax logits probabilities . metric creates two local variables, total count used compute frequency y_pred matches y_true. frequency ultimately returned categorical accuracy: idempotent operation simply divides total count. y_pred y_true passed vectors probabilities, rather labels. necessary, use op_one_hot expand y_true vector. sample_weight NULL, weights default 1. Use sample_weight 0 mask values.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_categorical_accuracy.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Calculates how often predictions match one-hot labels. — metric_categorical_accuracy","text":"","code":"metric_categorical_accuracy(   y_true,   y_pred,   ...,   name = \"categorical_accuracy\",   dtype = NULL )"},{"path":"https://keras3.posit.co/dev/reference/metric_categorical_accuracy.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Calculates how often predictions match one-hot labels. — metric_categorical_accuracy","text":"y_true Tensor true targets. y_pred Tensor predicted targets. ... forward/backward compatability. name (Optional) string name metric instance. dtype (Optional) data type metric result.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_categorical_accuracy.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Calculates how often predictions match one-hot labels. — metric_categorical_accuracy","text":"y_true y_pred missing, Metric instance returned. Metric instance can passed directly compile(metrics = ), used standalone object. See ?Metric example usage. y_true y_pred provided, tensor computed value returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_categorical_accuracy.html","id":"usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Calculates how often predictions match one-hot labels. — metric_categorical_accuracy","text":"Standalone usage:           Usage compile() API:","code":"m <- metric_categorical_accuracy() m$update_state(rbind(c(0, 0, 1), c(0, 1, 0)), rbind(c(0.1, 0.9, 0.8),                 c(0.05, 0.95, 0))) m$result() ## tf.Tensor(0.5, shape=(), dtype=float32) m$reset_state() m$update_state(rbind(c(0, 0, 1), c(0, 1, 0)), rbind(c(0.1, 0.9, 0.8),                c(0.05, 0.95, 0)),                sample_weight = c(0.7, 0.3)) m$result() ## tf.Tensor(0.3, shape=(), dtype=float32) # 0.3 model %>% compile(optimizer = 'sgd',                   loss = 'categorical_crossentropy',                   metrics = list(metric_categorical_accuracy()))"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/metric_categorical_crossentropy.html","id":null,"dir":"Reference","previous_headings":"","what":"Computes the crossentropy metric between the labels and predictions. — metric_categorical_crossentropy","title":"Computes the crossentropy metric between the labels and predictions. — metric_categorical_crossentropy","text":"crossentropy metric class used multiple label classes (2 ). assumes labels one-hot encoded, e.g., labels values c(2, 0, 1), y_true rbind(c([0, 0, 1), c(1, 0, 0), c(0, 1, 0)).","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_categorical_crossentropy.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes the crossentropy metric between the labels and predictions. — metric_categorical_crossentropy","text":"","code":"metric_categorical_crossentropy(   y_true,   y_pred,   from_logits = FALSE,   label_smoothing = 0,   axis = -1L,   ...,   name = \"categorical_crossentropy\",   dtype = NULL )"},{"path":"https://keras3.posit.co/dev/reference/metric_categorical_crossentropy.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Computes the crossentropy metric between the labels and predictions. — metric_categorical_crossentropy","text":"y_true Tensor one-hot true targets. y_pred Tensor predicted targets. from_logits (Optional) Whether output expected logits tensor. default, consider output encodes probability distribution. label_smoothing (Optional) Float [0, 1]. > 0, label values smoothed, meaning confidence label values relaxed. e.g. label_smoothing=0.2 means use value 0.1 label \"0\" 0.9 label \"1\". axis (Optional) Defaults -1. dimension along entropy computed. ... forward/backward compatability. name (Optional) string name metric instance. dtype (Optional) data type metric result.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_categorical_crossentropy.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Computes the crossentropy metric between the labels and predictions. — metric_categorical_crossentropy","text":"y_true y_pred missing, Metric instance returned. Metric instance can passed directly compile(metrics = ), used standalone object. See ?Metric example usage. y_true y_pred provided, tensor computed value returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_categorical_crossentropy.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Computes the crossentropy metric between the labels and predictions. — metric_categorical_crossentropy","text":"Standalone usage:           Usage compile() API:","code":"# EPSILON = 1e-7, y = y_true, y` = y_pred # y` = clip_op_clip_by_value(output, EPSILON, 1. - EPSILON) # y` = rbind(c(0.05, 0.95, EPSILON), c(0.1, 0.8, 0.1)) # xent = -sum(y * log(y'), axis = -1) #      = -((log 0.95), (log 0.1)) #      = [0.051, 2.302] # Reduced xent = (0.051 + 2.302) / 2  m <- metric_categorical_crossentropy() m$update_state(rbind(c(0, 1, 0), c(0, 0, 1)),                rbind(c(0.05, 0.95, 0), c(0.1, 0.8, 0.1))) m$result() ## tf.Tensor(1.1769392, shape=(), dtype=float32) # 1.1769392 m$reset_state() m$update_state(rbind(c(0, 1, 0), c(0, 0, 1)),                rbind(c(0.05, 0.95, 0), c(0.1, 0.8, 0.1)),                sample_weight = c(0.3, 0.7)) m$result() ## tf.Tensor(1.6271976, shape=(), dtype=float32) model %>% compile(   optimizer = 'sgd',   loss = 'mse',   metrics = list(metric_categorical_crossentropy()))"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/metric_categorical_focal_crossentropy.html","id":null,"dir":"Reference","previous_headings":"","what":"Computes the categorical focal crossentropy loss. — metric_categorical_focal_crossentropy","title":"Computes the categorical focal crossentropy loss. — metric_categorical_focal_crossentropy","text":"Computes categorical focal crossentropy loss.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_categorical_focal_crossentropy.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes the categorical focal crossentropy loss. — metric_categorical_focal_crossentropy","text":"","code":"metric_categorical_focal_crossentropy(   y_true,   y_pred,   alpha = 0.25,   gamma = 2,   from_logits = FALSE,   label_smoothing = 0,   axis = -1L )"},{"path":"https://keras3.posit.co/dev/reference/metric_categorical_focal_crossentropy.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Computes the categorical focal crossentropy loss. — metric_categorical_focal_crossentropy","text":"y_true Tensor one-hot true targets. y_pred Tensor predicted targets. alpha weight balancing factor classes, default 0.25 mentioned reference. can list floats scalar. multi-class case, alpha may set inverse class frequency using compute_class_weight sklearn.utils. gamma focusing parameter, default 2.0 mentioned reference. helps gradually reduce importance given simple examples smooth manner. gamma = 0, focal effect categorical crossentropy. from_logits Whether y_pred expected logits tensor. default, assume y_pred encodes probability distribution. label_smoothing Float [0, 1]. > 0 smooth labels. example, 0.1, use 0.1 / num_classes non-target labels 0.9 + 0.1 / num_classes target labels. axis Defaults -1. dimension along entropy computed.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_categorical_focal_crossentropy.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Computes the categorical focal crossentropy loss. — metric_categorical_focal_crossentropy","text":"Categorical focal crossentropy loss value.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_categorical_focal_crossentropy.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Computes the categorical focal crossentropy loss. — metric_categorical_focal_crossentropy","text":"","code":"y_true <- rbind(c(0, 1, 0), c(0, 0, 1)) y_pred <- rbind(c(0.05, 0.9, 0.05), c(0.1, 0.85, 0.05)) loss <- loss_categorical_focal_crossentropy(y_true, y_pred) loss ## tf.Tensor([2.6340148e-04 6.7591208e-01], shape=(2), dtype=float32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/metric_categorical_hinge.html","id":null,"dir":"Reference","previous_headings":"","what":"Computes the categorical hinge metric between y_true and y_pred. — metric_categorical_hinge","title":"Computes the categorical hinge metric between y_true and y_pred. — metric_categorical_hinge","text":"Formula:   neg=maximum((1-y_true)*y_pred) pos=sum(y_true*y_pred)","code":"loss <- maximum(neg - pos + 1, 0)"},{"path":"https://keras3.posit.co/dev/reference/metric_categorical_hinge.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes the categorical hinge metric between y_true and y_pred. — metric_categorical_hinge","text":"","code":"metric_categorical_hinge(   y_true,   y_pred,   ...,   name = \"categorical_hinge\",   dtype = NULL )"},{"path":"https://keras3.posit.co/dev/reference/metric_categorical_hinge.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Computes the categorical hinge metric between y_true and y_pred. — metric_categorical_hinge","text":"y_true ground truth values. y_true values expected either {-1, +1} {0, 1} (.e. one-hot-encoded tensor) shape = [batch_size, d0, .. dN]. y_pred predicted values shape = [batch_size, d0, .. dN]. ... forward/backward compatability. name (Optional) string name metric instance. dtype (Optional) data type metric result.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_categorical_hinge.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Computes the categorical hinge metric between y_true and y_pred. — metric_categorical_hinge","text":"Categorical hinge loss values shape = [batch_size, d0, .. dN-1].","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_categorical_hinge.html","id":"usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes the categorical hinge metric between y_true and y_pred. — metric_categorical_hinge","text":"Standalone usage:","code":"m <- metric_categorical_hinge() m$update_state(rbind(c(0, 1), c(0, 0)), rbind(c(0.6, 0.4), c(0.4, 0.6))) m$result() ## tf.Tensor(1.4000001, shape=(), dtype=float32) m$reset_state() m$update_state(rbind(c(0, 1), c(0, 0)), rbind(c(0.6, 0.4), c(0.4, 0.6)),                sample_weight = c(1, 0)) m$result() ## tf.Tensor(1.2, shape=(), dtype=float32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/metric_concordance_correlation.html","id":null,"dir":"Reference","previous_headings":"","what":"Calculates the Concordance Correlation Coefficient (CCC). — metric_concordance_correlation","title":"Calculates the Concordance Correlation Coefficient (CCC). — metric_concordance_correlation","text":"Formula:   CCC evaluates agreement true values (y_true) predicted values (y_pred) considering precision accuracy. coefficient ranges -1 1, value 1 indicates perfect agreement. metric useful regression tasks important assess well predictions match true values, taking account correlation proximity 45-degree line perfect concordance.","code":"loss <- mean(   2 * (y_true - mean(y_true)) * (y_pred - mean(y_pred)) /     (var(y_true) + var(y_pred) + (mean(y_true) - mean(y_pred))^2) )"},{"path":"https://keras3.posit.co/dev/reference/metric_concordance_correlation.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Calculates the Concordance Correlation Coefficient (CCC). — metric_concordance_correlation","text":"","code":"metric_concordance_correlation(   y_true,   y_pred,   axis = -1L,   ...,   name = \"concordance_correlation\",   dtype = NULL )"},{"path":"https://keras3.posit.co/dev/reference/metric_concordance_correlation.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Calculates the Concordance Correlation Coefficient (CCC). — metric_concordance_correlation","text":"y_true Tensor true targets. y_pred Tensor predicted targets. axis (Optional) integer tuple integers axis/axes along compute metric. Defaults -1. ... forward/backward compatability. name (Optional) string name metric instance. dtype (Optional) data type metric result.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_concordance_correlation.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Calculates the Concordance Correlation Coefficient (CCC). — metric_concordance_correlation","text":"Usage compile() API:","code":"ccc <- metric_concordance_correlation(axis=-1) y_true <- rbind(c(0, 1, 0.5),                 c(1, 1, 0.2)) y_pred <- rbind(c(0.1, 0.9, 0.5),                 c(1, 0.9, 0.2)) ccc$update_state(y_true, y_pred) ccc$result() ## tf.Tensor(0.9816317, shape=(), dtype=float32) model |> compile(   optimizer = 'sgd',   loss = 'mean_squared_error',   metrics = c(metric_concordance_correlation()) )"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/metric_cosine_similarity.html","id":null,"dir":"Reference","previous_headings":"","what":"Computes the cosine similarity between the labels and predictions. — metric_cosine_similarity","title":"Computes the cosine similarity between the labels and predictions. — metric_cosine_similarity","text":"Formula:   See: Cosine Similarity. metric keeps average cosine similarity predictions labels stream data.","code":"loss <- sum(l2_norm(y_true) * l2_norm(y_pred))"},{"path":"https://keras3.posit.co/dev/reference/metric_cosine_similarity.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes the cosine similarity between the labels and predictions. — metric_cosine_similarity","text":"","code":"metric_cosine_similarity(   ...,   name = \"cosine_similarity\",   dtype = NULL,   axis = -1L )"},{"path":"https://keras3.posit.co/dev/reference/metric_cosine_similarity.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Computes the cosine similarity between the labels and predictions. — metric_cosine_similarity","text":"... forward/backward compatability. name (Optional) string name metric instance. dtype (Optional) data type metric result. axis (Optional) Defaults -1. dimension along cosine similarity computed.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_cosine_similarity.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Computes the cosine similarity between the labels and predictions. — metric_cosine_similarity","text":"Metric instance returned. Metric instance can passed directly compile(metrics = ), used standalone object. See ?Metric example usage.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_cosine_similarity.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Computes the cosine similarity between the labels and predictions. — metric_cosine_similarity","text":"Standalone usage:         Usage compile() API:","code":"m <- metric_cosine_similarity(axis=2) m$update_state(rbind(c(0., 1.), c(1., 1.)), rbind(c(1., 0.), c(1., 1.))) m$result() ## tf.Tensor(0.5, shape=(), dtype=float32) m$reset_state() m$update_state(rbind(c(0., 1.), c(1., 1.)), rbind(c(1., 0.), c(1., 1.)),                sample_weight = c(0.3, 0.7)) m$result() ## tf.Tensor(0.7, shape=(), dtype=float32) model %>% compile(   optimizer = 'sgd',   loss = 'mse',   metrics = list(metric_cosine_similarity(axis=2)))"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/metric_f1_score.html","id":null,"dir":"Reference","previous_headings":"","what":"Computes F-1 Score. — metric_f1_score","title":"Computes F-1 Score. — metric_f1_score","text":"Formula:   harmonic mean precision recall. output range [0, 1]. works multi-class multi-label classification.","code":"f1_score <- 2 * (precision * recall) / (precision + recall)"},{"path":"https://keras3.posit.co/dev/reference/metric_f1_score.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes F-1 Score. — metric_f1_score","text":"","code":"metric_f1_score(   ...,   average = NULL,   threshold = NULL,   name = \"f1_score\",   dtype = NULL )"},{"path":"https://keras3.posit.co/dev/reference/metric_f1_score.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Computes F-1 Score. — metric_f1_score","text":"... forward/backward compatability. average Type averaging performed data. Acceptable values NULL, \"micro\", \"macro\" \"weighted\". Defaults NULL. NULL, averaging performed result() return score class. \"micro\", compute metrics globally counting total true positives, false negatives false positives. \"macro\", compute metrics label, return unweighted mean. take label imbalance account. \"weighted\", compute metrics label, return average weighted support (number true instances label). alters \"macro\" account label imbalance. can result score precision recall. threshold Elements y_pred greater threshold converted 1, rest 0. threshold NULL, argmax y_pred converted 1, rest 0. name Optional. String name metric instance. dtype Optional. Data type metric result.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_f1_score.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Computes F-1 Score. — metric_f1_score","text":"Metric instance returned. Metric instance can passed directly compile(metrics = ), used standalone object. See ?Metric example usage.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_f1_score.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Computes F-1 Score. — metric_f1_score","text":"","code":"metric <- metric_f1_score(threshold = 0.5) y_true <- rbind(c(1, 1, 1),                 c(1, 0, 0),                 c(1, 1, 0)) y_pred <- rbind(c(0.2, 0.6, 0.7),                 c(0.2, 0.6, 0.6),                 c(0.6, 0.8, 0.0)) metric$update_state(y_true, y_pred) result <- metric$result() result ## tf.Tensor([0.49999997 0.79999995 0.66666657], shape=(3), dtype=float32)"},{"path":"https://keras3.posit.co/dev/reference/metric_f1_score.html","id":"returns","dir":"Reference","previous_headings":"","what":"Returns","title":"Computes F-1 Score. — metric_f1_score","text":"F-1 Score: float.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/metric_false_negatives.html","id":null,"dir":"Reference","previous_headings":"","what":"Calculates the number of false negatives. — metric_false_negatives","title":"Calculates the number of false negatives. — metric_false_negatives","text":"sample_weight given, calculates sum weights false negatives. metric creates one local variable, accumulator used keep track number false negatives. sample_weight NULL, weights default 1. Use sample_weight 0 mask values.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_false_negatives.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Calculates the number of false negatives. — metric_false_negatives","text":"","code":"metric_false_negatives(..., thresholds = NULL, name = NULL, dtype = NULL)"},{"path":"https://keras3.posit.co/dev/reference/metric_false_negatives.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Calculates the number of false negatives. — metric_false_negatives","text":"... forward/backward compatability. thresholds (Optional) Defaults 0.5. float value, Python list float threshold values [0, 1]. threshold compared prediction values determine truth value predictions (.e., threshold TRUE, FALSE). used loss function sets from_logits=TRUE (.e. sigmoid applied predictions), thresholds set 0. One metric value generated threshold value. name (Optional) string name metric instance. dtype (Optional) data type metric result.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_false_negatives.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Calculates the number of false negatives. — metric_false_negatives","text":"Metric instance returned. Metric instance can passed directly compile(metrics = ), used standalone object. See ?Metric example usage.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_false_negatives.html","id":"usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Calculates the number of false negatives. — metric_false_negatives","text":"Standalone usage:","code":"m <- metric_false_negatives() m$update_state(c(0, 1, 1, 1), c(0, 1, 0, 0)) m$result() ## tf.Tensor(2.0, shape=(), dtype=float32) m$reset_state() m$update_state(c(0, 1, 1, 1), c(0, 1, 0, 0), sample_weight=c(0, 0, 1, 0)) m$result() ## tf.Tensor(1.0, shape=(), dtype=float32) # 1.0"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/metric_false_positives.html","id":null,"dir":"Reference","previous_headings":"","what":"Calculates the number of false positives. — metric_false_positives","title":"Calculates the number of false positives. — metric_false_positives","text":"sample_weight given, calculates sum weights false positives. metric creates one local variable, accumulator used keep track number false positives. sample_weight NULL, weights default 1. Use sample_weight 0 mask values.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_false_positives.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Calculates the number of false positives. — metric_false_positives","text":"","code":"metric_false_positives(..., thresholds = NULL, name = NULL, dtype = NULL)"},{"path":"https://keras3.posit.co/dev/reference/metric_false_positives.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Calculates the number of false positives. — metric_false_positives","text":"... forward/backward compatability. thresholds (Optional) Defaults 0.5. float value, Python list float threshold values [0, 1]. threshold compared prediction values determine truth value predictions (.e., threshold TRUE, FALSE). used loss function sets from_logits=TRUE (.e. sigmoid applied predictions), thresholds set 0. One metric value generated threshold value. name (Optional) string name metric instance. dtype (Optional) data type metric result.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_false_positives.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Calculates the number of false positives. — metric_false_positives","text":"Metric instance returned. Metric instance can passed directly compile(metrics = ), used standalone object. See ?Metric example usage.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_false_positives.html","id":"usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Calculates the number of false positives. — metric_false_positives","text":"Standalone usage:","code":"m <- metric_false_positives() m$update_state(c(0, 1, 0, 0), c(0, 0, 1, 1)) m$result() ## tf.Tensor(2.0, shape=(), dtype=float32) m$reset_state() m$update_state(c(0, 1, 0, 0), c(0, 0, 1, 1), sample_weight = c(0, 0, 1, 0)) m$result() ## tf.Tensor(1.0, shape=(), dtype=float32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/metric_fbeta_score.html","id":null,"dir":"Reference","previous_headings":"","what":"Computes F-Beta score. — metric_fbeta_score","title":"Computes F-Beta score. — metric_fbeta_score","text":"Formula:   weighted harmonic mean precision recall. output range [0, 1]. works multi-class multi-label classification.","code":"b2 <- beta^2 f_beta_score <- (1 + b2) * (precision * recall) / (precision * b2 + recall)"},{"path":"https://keras3.posit.co/dev/reference/metric_fbeta_score.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes F-Beta score. — metric_fbeta_score","text":"","code":"metric_fbeta_score(   ...,   average = NULL,   beta = 1,   threshold = NULL,   name = \"fbeta_score\",   dtype = NULL )"},{"path":"https://keras3.posit.co/dev/reference/metric_fbeta_score.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Computes F-Beta score. — metric_fbeta_score","text":"... forward/backward compatability. average Type averaging performed across per-class results multi-class case. Acceptable values NULL, \"micro\", \"macro\" \"weighted\". Defaults NULL. NULL, averaging performed result() return score class. \"micro\", compute metrics globally counting total true positives, false negatives false positives. \"macro\", compute metrics label, return unweighted mean. take label imbalance account. \"weighted\", compute metrics label, return average weighted support (number true instances label). alters \"macro\" account label imbalance. can result score precision recall. beta Determines weight given recall harmonic mean precision recall (see pseudocode equation ). Defaults 1. threshold Elements y_pred greater threshold converted 1, rest 0. threshold NULL, argmax y_pred converted 1, rest 0. name Optional. String name metric instance. dtype Optional. Data type metric result.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_fbeta_score.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Computes F-Beta score. — metric_fbeta_score","text":"Metric instance returned. Metric instance can passed directly compile(metrics = ), used standalone object. See ?Metric example usage.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_fbeta_score.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Computes F-Beta score. — metric_fbeta_score","text":"","code":"metric <- metric_fbeta_score(beta = 2.0, threshold = 0.5) y_true <- rbind(c(1, 1, 1),                 c(1, 0, 0),                 c(1, 1, 0)) y_pred <- rbind(c(0.2, 0.6, 0.7),                 c(0.2, 0.6, 0.6),                 c(0.6, 0.8, 0.0)) metric$update_state(y_true, y_pred) metric$result() ## tf.Tensor([0.3846154  0.90909094 0.8333332 ], shape=(3), dtype=float32)"},{"path":"https://keras3.posit.co/dev/reference/metric_fbeta_score.html","id":"returns","dir":"Reference","previous_headings":"","what":"Returns","title":"Computes F-Beta score. — metric_fbeta_score","text":"F-Beta Score: float.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/metric_hinge.html","id":null,"dir":"Reference","previous_headings":"","what":"Computes the hinge metric between y_true and y_pred. — metric_hinge","title":"Computes the hinge metric between y_true and y_pred. — metric_hinge","text":"Formula:   y_true values expected -1 1. binary (0 1) labels provided convert -1 1.","code":"loss <- mean(maximum(1 - y_true * y_pred, 0), axis=-1)"},{"path":"https://keras3.posit.co/dev/reference/metric_hinge.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes the hinge metric between y_true and y_pred. — metric_hinge","text":"","code":"metric_hinge(y_true, y_pred, ..., name = \"hinge\", dtype = NULL)"},{"path":"https://keras3.posit.co/dev/reference/metric_hinge.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Computes the hinge metric between y_true and y_pred. — metric_hinge","text":"y_true ground truth values. y_true values expected -1 1. binary (0 1) labels provided converted -1 1 shape = [batch_size, d0, .. dN]. y_pred predicted values shape = [batch_size, d0, .. dN]. ... forward/backward compatability. name (Optional) string name metric instance. dtype (Optional) data type metric result.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_hinge.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Computes the hinge metric between y_true and y_pred. — metric_hinge","text":"y_true y_pred missing, Metric instance returned. Metric instance can passed directly compile(metrics = ), used standalone object. See ?Metric example usage. y_true y_pred provided, tensor computed value returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_hinge.html","id":"usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes the hinge metric between y_true and y_pred. — metric_hinge","text":"Standalone usage:","code":"m <- metric_hinge() m$update_state(rbind(c(0, 1), c(0, 0)), rbind(c(0.6, 0.4), c(0.4, 0.6))) m$result() ## tf.Tensor(1.3, shape=(), dtype=float32) m$reset_state() m$update_state(rbind(c(0, 1), c(0, 0)), rbind(c(0.6, 0.4), c(0.4, 0.6)),                sample_weight = c(1, 0)) m$result() ## tf.Tensor(1.1, shape=(), dtype=float32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/metric_huber.html","id":null,"dir":"Reference","previous_headings":"","what":"Computes Huber loss value. — metric_huber","title":"Computes Huber loss value. — metric_huber","text":"Formula:   See: Huber loss.","code":"for (x in error) {   if (abs(x) <= delta){     loss <- c(loss, (0.5 * x^2))   } else if (abs(x) > delta) {     loss <- c(loss, (delta * abs(x) - 0.5 * delta^2))   } } loss <- mean(loss)"},{"path":"https://keras3.posit.co/dev/reference/metric_huber.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes Huber loss value. — metric_huber","text":"","code":"metric_huber(y_true, y_pred, delta = 1)"},{"path":"https://keras3.posit.co/dev/reference/metric_huber.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Computes Huber loss value. — metric_huber","text":"y_true tensor true targets. y_pred tensor predicted targets. delta float, point Huber loss function changes quadratic linear. Defaults 1.0.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_huber.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Computes Huber loss value. — metric_huber","text":"Tensor one scalar loss entry per sample.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_huber.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Computes Huber loss value. — metric_huber","text":"","code":"y_true <- rbind(c(0, 1), c(0, 0)) y_pred <- rbind(c(0.6, 0.4), c(0.4, 0.6)) loss <- loss_huber(y_true, y_pred)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/metric_iou.html","id":null,"dir":"Reference","previous_headings":"","what":"Computes the Intersection-Over-Union metric for specific target classes. — metric_iou","title":"Computes the Intersection-Over-Union metric for specific target classes. — metric_iou","text":"Formula:   Intersection--Union common evaluation metric semantic image segmentation. compute IoUs, predictions accumulated confusion matrix, weighted sample_weight metric calculated . sample_weight NULL, weights default 1. Use sample_weight 0 mask values. Note, class first computes IoUs individual classes, returns mean IoUs classes specified target_class_ids. target_class_ids one id value, IoU specific class returned.","code":"iou <- true_positives / (true_positives + false_positives + false_negatives)"},{"path":"https://keras3.posit.co/dev/reference/metric_iou.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes the Intersection-Over-Union metric for specific target classes. — metric_iou","text":"","code":"metric_iou(   ...,   num_classes,   target_class_ids,   name = NULL,   dtype = NULL,   ignore_class = NULL,   sparse_y_true = TRUE,   sparse_y_pred = TRUE,   axis = -1L )"},{"path":"https://keras3.posit.co/dev/reference/metric_iou.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Computes the Intersection-Over-Union metric for specific target classes. — metric_iou","text":"... forward/backward compatability. num_classes possible number labels prediction task can . target_class_ids list target class ids metric returned. compute IoU specific class, list single id value provided. name (Optional) string name metric instance. dtype (Optional) data type metric result. ignore_class Optional integer. ID class ignored metric computation. useful, example, segmentation problems featuring \"void\" class (commonly -1 255) segmentation maps. default (ignore_class=NULL), classes considered. sparse_y_true Whether labels encoded using integers dense floating point vectors. FALSE, argmax function used determine sample's likely associated label. sparse_y_pred Whether predictions encoded using integers dense floating point vectors. FALSE, argmax function used determine sample's likely associated label. axis (Optional) -1 dimension containing logits. Defaults -1.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_iou.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Computes the Intersection-Over-Union metric for specific target classes. — metric_iou","text":"Metric instance returned. Metric instance can passed directly compile(metrics = ), used standalone object. See ?Metric example usage.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_iou.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Computes the Intersection-Over-Union metric for specific target classes. — metric_iou","text":"Standalone usage:                 Usage compile() API:","code":"m <- metric_iou(num_classes = 2L, target_class_ids = list(0L)) m$update_state(c(0, 0, 1, 1), c(0, 1, 0, 1)) ## tf.Tensor( ## [[1 1] ##  [1 1]], shape=(2, 2), dtype=int64) m$result() ## tf.Tensor(0.3333333, shape=(), dtype=float32) m$reset_state() m$update_state(c(0, 0, 1, 1), c(0, 1, 0, 1),                sample_weight = 10 * c(0.3, 0.3, 0.3, 0.1)) ## tf.Tensor( ## [[3 3] ##  [3 1]], shape=(2, 2), dtype=int64) m$result() ## tf.Tensor(0.3333333, shape=(), dtype=float32) model %>% compile(   optimizer = 'sgd',   loss = 'mse',   metrics = list(metric_iou(num_classes = 2L, target_class_ids = list(0L))))"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/metric_kl_divergence.html","id":null,"dir":"Reference","previous_headings":"","what":"Computes Kullback-Leibler divergence metric between y_true and — metric_kl_divergence","title":"Computes Kullback-Leibler divergence metric between y_true and — metric_kl_divergence","text":"Formula:   y_true y_pred expected probability distributions, values 0 1. get clipped [0, 1] range.","code":"loss <- y_true * log(y_true / y_pred)"},{"path":"https://keras3.posit.co/dev/reference/metric_kl_divergence.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes Kullback-Leibler divergence metric between y_true and — metric_kl_divergence","text":"","code":"metric_kl_divergence(y_true, y_pred, ..., name = \"kl_divergence\", dtype = NULL)"},{"path":"https://keras3.posit.co/dev/reference/metric_kl_divergence.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Computes Kullback-Leibler divergence metric between y_true and — metric_kl_divergence","text":"y_true Tensor true targets. y_pred Tensor predicted targets. ... forward/backward compatability. name (Optional) string name metric instance. dtype (Optional) data type metric result.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_kl_divergence.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Computes Kullback-Leibler divergence metric between y_true and — metric_kl_divergence","text":"y_true y_pred missing, Metric instance returned. Metric instance can passed directly compile(metrics = ), used standalone object. See ?Metric example usage. y_true y_pred provided, tensor computed value returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_kl_divergence.html","id":"usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes Kullback-Leibler divergence metric between y_true and — metric_kl_divergence","text":"Standalone usage:         Usage compile() API:","code":"m <- metric_kl_divergence() m$update_state(rbind(c(0, 1), c(0, 0)), rbind(c(0.6, 0.4), c(0.4, 0.6))) m$result() ## tf.Tensor(0.45814303, shape=(), dtype=float32) m$reset_state() m$update_state(rbind(c(0, 1), c(0, 0)), rbind(c(0.6, 0.4), c(0.4, 0.6)),                sample_weight = c(1, 0)) m$result() ## tf.Tensor(0.91628915, shape=(), dtype=float32) model %>% compile(optimizer = 'sgd',                   loss = 'mse',                   metrics = list(metric_kl_divergence()))"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/metric_log_cosh.html","id":null,"dir":"Reference","previous_headings":"","what":"Logarithm of the hyperbolic cosine of the prediction error. — metric_log_cosh","title":"Logarithm of the hyperbolic cosine of the prediction error. — metric_log_cosh","text":"Formula:   Note log(cosh(x)) approximately equal (x ** 2) / 2 small x abs(x) - log(2) large x. means 'logcosh' works mostly like mean squared error, strongly affected occasional wildly incorrect prediction.","code":"loss <- mean(log(cosh(y_pred - y_true)), axis=-1)"},{"path":"https://keras3.posit.co/dev/reference/metric_log_cosh.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Logarithm of the hyperbolic cosine of the prediction error. — metric_log_cosh","text":"","code":"metric_log_cosh(y_true, y_pred)"},{"path":"https://keras3.posit.co/dev/reference/metric_log_cosh.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Logarithm of the hyperbolic cosine of the prediction error. — metric_log_cosh","text":"y_true Ground truth values shape = [batch_size, d0, .. dN]. y_pred predicted values shape = [batch_size, d0, .. dN].","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_log_cosh.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Logarithm of the hyperbolic cosine of the prediction error. — metric_log_cosh","text":"Logcosh error values shape = [batch_size, d0, .. dN-1].","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_log_cosh.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Logarithm of the hyperbolic cosine of the prediction error. — metric_log_cosh","text":"","code":"y_true <- rbind(c(0., 1.), c(0., 0.)) y_pred <- rbind(c(1., 1.), c(0., 0.)) loss <- metric_log_cosh(y_true, y_pred) loss ## tf.Tensor([0.2168904 0.       ], shape=(2), dtype=float32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/metric_log_cosh_error.html","id":null,"dir":"Reference","previous_headings":"","what":"Computes the logarithm of the hyperbolic cosine of the prediction error. — metric_log_cosh_error","title":"Computes the logarithm of the hyperbolic cosine of the prediction error. — metric_log_cosh_error","text":"Formula:","code":"error <- y_pred - y_true logcosh <- mean(log((exp(error) + exp(-error))/2), axis=-1)"},{"path":"https://keras3.posit.co/dev/reference/metric_log_cosh_error.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes the logarithm of the hyperbolic cosine of the prediction error. — metric_log_cosh_error","text":"","code":"metric_log_cosh_error(..., name = \"logcosh\", dtype = NULL)"},{"path":"https://keras3.posit.co/dev/reference/metric_log_cosh_error.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Computes the logarithm of the hyperbolic cosine of the prediction error. — metric_log_cosh_error","text":"... forward/backward compatability. name (Optional) string name metric instance. dtype (Optional) data type metric result.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_log_cosh_error.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Computes the logarithm of the hyperbolic cosine of the prediction error. — metric_log_cosh_error","text":"Metric instance returned. Metric instance can passed directly compile(metrics = ), used standalone object. See ?Metric example usage.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_log_cosh_error.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Computes the logarithm of the hyperbolic cosine of the prediction error. — metric_log_cosh_error","text":"Standalone usage:         Usage compile() API:","code":"m <- metric_log_cosh_error() m$update_state(rbind(c(0, 1), c(0, 0)), rbind(c(1, 1), c(0, 0))) m$result() ## tf.Tensor(0.108445205, shape=(), dtype=float32) m$reset_state() m$update_state(rbind(c(0, 1), c(0, 0)), rbind(c(1, 1), c(0, 0)),                sample_weight = c(1, 0)) m$result() ## tf.Tensor(0.21689041, shape=(), dtype=float32) model %>% compile(optimizer = 'sgd',                   loss = 'mse',                   metrics = list(metric_log_cosh_error()))"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/metric_mean.html","id":null,"dir":"Reference","previous_headings":"","what":"Compute the (weighted) mean of the given values. — metric_mean","title":"Compute the (weighted) mean of the given values. — metric_mean","text":"example, values c(1, 3, 5, 7) mean 4. sample_weight specified c(1, 1, 0, 0) mean 2. metric creates two variables, total count. mean value returned simply total divided count.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_mean.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Compute the (weighted) mean of the given values. — metric_mean","text":"","code":"metric_mean(..., name = \"mean\", dtype = NULL)"},{"path":"https://keras3.posit.co/dev/reference/metric_mean.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Compute the (weighted) mean of the given values. — metric_mean","text":"... forward/backward compatability. name (Optional) string name metric instance. dtype (Optional) data type metric result.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_mean.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Compute the (weighted) mean of the given values. — metric_mean","text":"Metric instance returned. Metric instance can passed directly compile(metrics = ), used standalone object. See ?Metric example usage.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_mean.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Compute the (weighted) mean of the given values. — metric_mean","text":"","code":"m <- metric_mean() m$update_state(c(1, 3, 5, 7)) m$result() ## tf.Tensor(4.0, shape=(), dtype=float32) # calling a metric directly is equivalent to calling # m$update_state(); m$result() m <- metric_mean() m(c(1, 3, 5, 7)) ## tf.Tensor(4.0, shape=(), dtype=float32) m$reset_state() m$update_state(c(1, 3, 5, 7), sample_weight = c(1, 1, 0, 0)) m$result() ## tf.Tensor(2.0, shape=(), dtype=float32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/metric_mean_absolute_error.html","id":null,"dir":"Reference","previous_headings":"","what":"Computes the mean absolute error between the labels and predictions. — metric_mean_absolute_error","title":"Computes the mean absolute error between the labels and predictions. — metric_mean_absolute_error","text":"Formula:","code":"loss <- mean(abs(y_true - y_pred))"},{"path":"https://keras3.posit.co/dev/reference/metric_mean_absolute_error.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes the mean absolute error between the labels and predictions. — metric_mean_absolute_error","text":"","code":"metric_mean_absolute_error(   y_true,   y_pred,   ...,   name = \"mean_absolute_error\",   dtype = NULL )"},{"path":"https://keras3.posit.co/dev/reference/metric_mean_absolute_error.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Computes the mean absolute error between the labels and predictions. — metric_mean_absolute_error","text":"y_true Ground truth values shape = [batch_size, d0, .. dN]. y_pred predicted values shape = [batch_size, d0, .. dN]. ... forward/backward compatability. name (Optional) string name metric instance. dtype (Optional) data type metric result.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_mean_absolute_error.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Computes the mean absolute error between the labels and predictions. — metric_mean_absolute_error","text":"y_true y_pred missing, Metric instance returned. Metric instance can passed directly compile(metrics = ), used standalone object. See ?Metric example usage. y_true y_pred provided, tensor computed value returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_mean_absolute_error.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Computes the mean absolute error between the labels and predictions. — metric_mean_absolute_error","text":"Standalone usage:         Usage compile() API:","code":"m <- metric_mean_absolute_error() m$update_state(rbind(c(0, 1), c(0, 0)), rbind(c(1, 1), c(0, 0))) m$result() ## tf.Tensor(0.25, shape=(), dtype=float32) m$reset_state() m$update_state(rbind(c(0, 1), c(0, 0)), rbind(c(1, 1), c(0, 0)),                sample_weight = c(1, 0)) m$result() ## tf.Tensor(0.5, shape=(), dtype=float32) model %>% compile(     optimizer = 'sgd',     loss = 'mse',     metrics = list(metric_mean_absolute_error()))"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/metric_mean_absolute_percentage_error.html","id":null,"dir":"Reference","previous_headings":"","what":"Computes mean absolute percentage error between y_true and y_pred. — metric_mean_absolute_percentage_error","title":"Computes mean absolute percentage error between y_true and y_pred. — metric_mean_absolute_percentage_error","text":"Formula:   Division zero prevented dividing maximum(y_true, epsilon) epsilon = keras$backend$epsilon() (default 1e-7).","code":"loss <- 100 * mean(abs((y_true - y_pred) / y_true), axis=-1)"},{"path":"https://keras3.posit.co/dev/reference/metric_mean_absolute_percentage_error.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes mean absolute percentage error between y_true and y_pred. — metric_mean_absolute_percentage_error","text":"","code":"metric_mean_absolute_percentage_error(   y_true,   y_pred,   ...,   name = \"mean_absolute_percentage_error\",   dtype = NULL )"},{"path":"https://keras3.posit.co/dev/reference/metric_mean_absolute_percentage_error.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Computes mean absolute percentage error between y_true and y_pred. — metric_mean_absolute_percentage_error","text":"y_true Ground truth values shape = [batch_size, d0, .. dN]. y_pred predicted values shape = [batch_size, d0, .. dN]. ... forward/backward compatability. name (Optional) string name metric instance. dtype (Optional) data type metric result.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_mean_absolute_percentage_error.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Computes mean absolute percentage error between y_true and y_pred. — metric_mean_absolute_percentage_error","text":"y_true y_pred missing, Metric instance returned. Metric instance can passed directly compile(metrics = ), used standalone object. See ?Metric example usage. y_true y_pred provided, tensor computed value returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_mean_absolute_percentage_error.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Computes mean absolute percentage error between y_true and y_pred. — metric_mean_absolute_percentage_error","text":"Standalone usage:         Usage compile() API:","code":"m <- metric_mean_absolute_percentage_error() m$update_state(rbind(c(0, 1), c(0, 0)), rbind(c(1, 1), c(0, 0))) m$result() ## tf.Tensor(250000000.0, shape=(), dtype=float32) m$reset_state() m$update_state(rbind(c(0, 1), c(0, 0)), rbind(c(1, 1), c(0, 0)),                sample_weight = c(1, 0)) m$result() ## tf.Tensor(500000000.0, shape=(), dtype=float32) model %>% compile(     optimizer = 'sgd',     loss = 'mse',     metrics = list(metric_mean_absolute_percentage_error()))"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/metric_mean_iou.html","id":null,"dir":"Reference","previous_headings":"","what":"Computes the mean Intersection-Over-Union metric. — metric_mean_iou","title":"Computes the mean Intersection-Over-Union metric. — metric_mean_iou","text":"Formula:   Intersection--Union common evaluation metric semantic image segmentation. compute IoUs, predictions accumulated confusion matrix, weighted sample_weight metric calculated . sample_weight NULL, weights default 1. Use sample_weight 0 mask values. Note class first computes IoUs individual classes, returns mean values.","code":"iou <- true_positives / (true_positives + false_positives + false_negatives)"},{"path":"https://keras3.posit.co/dev/reference/metric_mean_iou.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes the mean Intersection-Over-Union metric. — metric_mean_iou","text":"","code":"metric_mean_iou(   ...,   num_classes,   name = NULL,   dtype = NULL,   ignore_class = NULL,   sparse_y_true = TRUE,   sparse_y_pred = TRUE,   axis = -1L )"},{"path":"https://keras3.posit.co/dev/reference/metric_mean_iou.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Computes the mean Intersection-Over-Union metric. — metric_mean_iou","text":"... forward/backward compatability. num_classes possible number labels prediction task can . value must provided, since confusion matrix dimension = [num_classes, num_classes] allocated. name (Optional) string name metric instance. dtype (Optional) data type metric result. ignore_class Optional integer. ID class ignored metric computation. useful, example, segmentation problems featuring \"void\" class (commonly -1 255) segmentation maps. default (ignore_class=NULL), classes considered. sparse_y_true Whether labels encoded using integers dense floating point vectors. FALSE, argmax function used determine sample's likely associated label. sparse_y_pred Whether predictions encoded using integers dense floating point vectors. FALSE, argmax function used determine sample's likely associated label. axis (Optional) dimension containing logits. Defaults -1.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_mean_iou.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Computes the mean Intersection-Over-Union metric. — metric_mean_iou","text":"Metric instance returned. Metric instance can passed directly compile(metrics = ), used standalone object. See ?Metric example usage.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_mean_iou.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Computes the mean Intersection-Over-Union metric. — metric_mean_iou","text":"Standalone usage:                 Usage compile() API:","code":"# cm = [[1, 1], #       [1, 1]] # sum_row = [2, 2], sum_col = [2, 2], true_positives = [1, 1] # iou = true_positives / (sum_row + sum_col - true_positives)) # result = (1 / (2 + 2 - 1) + 1 / (2 + 2 - 1)) / 2 = 0.33 m <- metric_mean_iou(num_classes = 2) m$update_state(c(0, 0, 1, 1), c(0, 1, 0, 1)) ## tf.Tensor( ## [[1 1] ##  [1 1]], shape=(2, 2), dtype=int64) m$result() ## tf.Tensor(0.33333334, shape=(), dtype=float32) m$reset_state() m$update_state(c(0, 0, 1, 1), c(0, 1, 0, 1),                sample_weight= 10 * c(0.3, 0.3, 0.3, 0.1)) ## tf.Tensor( ## [[3 3] ##  [3 1]], shape=(2, 2), dtype=int64) m$result() ## tf.Tensor(0.23809525, shape=(), dtype=float32) model %>% compile(   optimizer = 'sgd',   loss = 'mse',   metrics = list(metric_mean_iou(num_classes=2)))"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/metric_mean_squared_error.html","id":null,"dir":"Reference","previous_headings":"","what":"Computes the mean squared error between y_true and y_pred. — metric_mean_squared_error","title":"Computes the mean squared error between y_true and y_pred. — metric_mean_squared_error","text":"Formula:","code":"loss <- mean(square(y_true - y_pred))"},{"path":"https://keras3.posit.co/dev/reference/metric_mean_squared_error.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes the mean squared error between y_true and y_pred. — metric_mean_squared_error","text":"","code":"metric_mean_squared_error(   y_true,   y_pred,   ...,   name = \"mean_squared_error\",   dtype = NULL )"},{"path":"https://keras3.posit.co/dev/reference/metric_mean_squared_error.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Computes the mean squared error between y_true and y_pred. — metric_mean_squared_error","text":"y_true Ground truth values shape = [batch_size, d0, .. dN]. y_pred predicted values shape = [batch_size, d0, .. dN]. ... forward/backward compatability. name (Optional) string name metric instance. dtype (Optional) data type metric result.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_mean_squared_error.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Computes the mean squared error between y_true and y_pred. — metric_mean_squared_error","text":"y_true y_pred missing, Metric instance returned. Metric instance can passed directly compile(metrics = ), used standalone object. See ?Metric example usage. y_true y_pred provided, tensor computed value returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_mean_squared_error.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Computes the mean squared error between y_true and y_pred. — metric_mean_squared_error","text":"","code":"m <- metric_mean_squared_error() m$update_state(rbind(c(0, 1), c(0, 0)), rbind(c(1, 1), c(0, 0))) m$result() ## tf.Tensor(0.25, shape=(), dtype=float32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/metric_mean_squared_logarithmic_error.html","id":null,"dir":"Reference","previous_headings":"","what":"Computes mean squared logarithmic error between y_true and y_pred. — metric_mean_squared_logarithmic_error","title":"Computes mean squared logarithmic error between y_true and y_pred. — metric_mean_squared_logarithmic_error","text":"Formula:   Note y_pred y_true less equal 0. Negative values 0 values replaced keras$backend$epsilon() (default 1e-7).","code":"loss <- mean(square(log(y_true + 1) - log(y_pred + 1)), axis=-1)"},{"path":"https://keras3.posit.co/dev/reference/metric_mean_squared_logarithmic_error.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes mean squared logarithmic error between y_true and y_pred. — metric_mean_squared_logarithmic_error","text":"","code":"metric_mean_squared_logarithmic_error(   y_true,   y_pred,   ...,   name = \"mean_squared_logarithmic_error\",   dtype = NULL )"},{"path":"https://keras3.posit.co/dev/reference/metric_mean_squared_logarithmic_error.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Computes mean squared logarithmic error between y_true and y_pred. — metric_mean_squared_logarithmic_error","text":"y_true Ground truth values shape = [batch_size, d0, .. dN]. y_pred predicted values shape = [batch_size, d0, .. dN]. ... forward/backward compatability. name (Optional) string name metric instance. dtype (Optional) data type metric result.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_mean_squared_logarithmic_error.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Computes mean squared logarithmic error between y_true and y_pred. — metric_mean_squared_logarithmic_error","text":"y_true y_pred missing, Metric instance returned. Metric instance can passed directly compile(metrics = ), used standalone object. See ?Metric example usage. y_true y_pred provided, tensor computed value returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_mean_squared_logarithmic_error.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Computes mean squared logarithmic error between y_true and y_pred. — metric_mean_squared_logarithmic_error","text":"Standalone usage:         Usage compile() API:","code":"m <- metric_mean_squared_logarithmic_error() m$update_state(rbind(c(0, 1), c(0, 0)), rbind(c(1, 1), c(0, 0))) m$result() ## tf.Tensor(0.12011322, shape=(), dtype=float32) m$reset_state() m$update_state(rbind(c(0, 1), c(0, 0)), rbind(c(1, 1), c(0, 0)),                sample_weight = c(1, 0)) m$result() ## tf.Tensor(0.24022643, shape=(), dtype=float32) model %>% compile(   optimizer = 'sgd',   loss = 'mse',   metrics = list(metric_mean_squared_logarithmic_error()))"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/metric_mean_wrapper.html","id":null,"dir":"Reference","previous_headings":"","what":"Wrap a stateless metric function with the Mean metric. — metric_mean_wrapper","title":"Wrap a stateless metric function with the Mean metric. — metric_mean_wrapper","text":"use class quickly build mean metric function. function needs signature fn(y_true, y_pred) return per-sample loss array. metric_mean_wrapper$result() return average metric value across samples seen far. example:","code":"mse <- function(y_true, y_pred) {   (y_true - y_pred)^2 }  mse_metric <- metric_mean_wrapper(fn = mse) mse_metric$update_state(c(0, 1), c(1, 1)) mse_metric$result() ## tf.Tensor(0.5, shape=(), dtype=float32)"},{"path":"https://keras3.posit.co/dev/reference/metric_mean_wrapper.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Wrap a stateless metric function with the Mean metric. — metric_mean_wrapper","text":"","code":"metric_mean_wrapper(..., fn, name = NULL, dtype = NULL)"},{"path":"https://keras3.posit.co/dev/reference/metric_mean_wrapper.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Wrap a stateless metric function with the Mean metric. — metric_mean_wrapper","text":"... Keyword arguments pass fn. fn metric function wrap, signature fn(y_true, y_pred). name (Optional) string name metric instance. dtype (Optional) data type metric result.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_mean_wrapper.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Wrap a stateless metric function with the Mean metric. — metric_mean_wrapper","text":"Metric instance returned. Metric instance can passed directly compile(metrics = ), used standalone object. See ?Metric example usage.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/metric_one_hot_iou.html","id":null,"dir":"Reference","previous_headings":"","what":"Computes the Intersection-Over-Union metric for one-hot encoded labels. — metric_one_hot_iou","title":"Computes the Intersection-Over-Union metric for one-hot encoded labels. — metric_one_hot_iou","text":"Formula:   Intersection--Union common evaluation metric semantic image segmentation. compute IoUs, predictions accumulated confusion matrix, weighted sample_weight metric calculated . sample_weight NULL, weights default 1. Use sample_weight 0 mask values. class can used compute IoU multi-class classification tasks labels one-hot encoded (last axis one dimension per class). Note predictions also shape. compute IoU, first labels predictions converted back integer format taking argmax class axis. computation steps base IoU class apply. Note, one channel labels predictions, class class IoU. case, use IoU instead. Also, make sure num_classes equal number classes data, avoid \"labels bound\" error confusion matrix computed.","code":"iou <- true_positives / (true_positives + false_positives + false_negatives)"},{"path":"https://keras3.posit.co/dev/reference/metric_one_hot_iou.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes the Intersection-Over-Union metric for one-hot encoded labels. — metric_one_hot_iou","text":"","code":"metric_one_hot_iou(   ...,   num_classes,   target_class_ids,   name = NULL,   dtype = NULL,   ignore_class = NULL,   sparse_y_pred = FALSE,   axis = -1L )"},{"path":"https://keras3.posit.co/dev/reference/metric_one_hot_iou.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Computes the Intersection-Over-Union metric for one-hot encoded labels. — metric_one_hot_iou","text":"... forward/backward compatability. num_classes possible number labels prediction task can . target_class_ids list list target class ids metric returned. compute IoU specific class, list (list) single id value provided. name (Optional) string name metric instance. dtype (Optional) data type metric result. ignore_class Optional integer. ID class ignored metric computation. useful, example, segmentation problems featuring \"void\" class (commonly -1 255) segmentation maps. default (ignore_class=NULL), classes considered. sparse_y_pred Whether predictions encoded using integers dense floating point vectors. FALSE, argmax function used determine sample's likely associated label. axis (Optional) dimension containing logits. Defaults -1.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_one_hot_iou.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Computes the Intersection-Over-Union metric for one-hot encoded labels. — metric_one_hot_iou","text":"Metric instance returned. Metric instance can passed directly compile(metrics = ), used standalone object. See ?Metric example usage.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_one_hot_iou.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Computes the Intersection-Over-Union metric for one-hot encoded labels. — metric_one_hot_iou","text":"Standalone usage:         Usage compile() API:","code":"y_true <- rbind(c(0, 0, 1),                 c(1, 0, 0),                 c(0, 1, 0),                 c(1, 0, 0)) y_pred <- rbind(c(0.2, 0.3, 0.5),                 c(0.1, 0.2, 0.7),                 c(0.5, 0.3, 0.1),                 c(0.1, 0.4, 0.5)) sample_weight <- c(0.1, 0.2, 0.3, 0.4)  m <- metric_one_hot_iou(num_classes = 3, target_class_ids = c(0, 2)) m$update_state(y_true = y_true,                y_pred = y_pred,                sample_weight = 10*sample_weight) ## tf.Tensor( ## [[0 0 6] ##  [3 0 0] ##  [0 0 1]], shape=(3, 3), dtype=int64) m$result() ## tf.Tensor(0.071428575, shape=(), dtype=float32) model %>% compile(   optimizer = 'sgd',   loss = 'mse',   metrics = list(metric_one_hot_iou(     num_classes = 3L,     target_class_id = list(1L)   )) )"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/metric_one_hot_mean_iou.html","id":null,"dir":"Reference","previous_headings":"","what":"Computes mean Intersection-Over-Union metric for one-hot encoded labels. — metric_one_hot_mean_iou","title":"Computes mean Intersection-Over-Union metric for one-hot encoded labels. — metric_one_hot_mean_iou","text":"Formula:   Intersection--Union common evaluation metric semantic image segmentation. compute IoUs, predictions accumulated confusion matrix, weighted sample_weight metric calculated . sample_weight NULL, weights default 1. Use sample_weight 0 mask values. class can used compute mean IoU multi-class classification tasks labels one-hot encoded (last axis one dimension per class). Note predictions also shape. compute mean IoU, first labels predictions converted back integer format taking argmax class axis. computation steps base MeanIoU class apply. Note, one channel labels predictions, class class metric_mean_iou. case, use metric_mean_iou instead. Also, make sure num_classes equal number classes data, avoid \"labels bound\" error confusion matrix computed.","code":"iou <- true_positives / (true_positives + false_positives + false_negatives)"},{"path":"https://keras3.posit.co/dev/reference/metric_one_hot_mean_iou.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes mean Intersection-Over-Union metric for one-hot encoded labels. — metric_one_hot_mean_iou","text":"","code":"metric_one_hot_mean_iou(   ...,   num_classes,   name = NULL,   dtype = NULL,   ignore_class = NULL,   sparse_y_pred = FALSE,   axis = -1L )"},{"path":"https://keras3.posit.co/dev/reference/metric_one_hot_mean_iou.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Computes mean Intersection-Over-Union metric for one-hot encoded labels. — metric_one_hot_mean_iou","text":"... forward/backward compatability. num_classes possible number labels prediction task can . name (Optional) string name metric instance. dtype (Optional) data type metric result. ignore_class Optional integer. ID class ignored metric computation. useful, example, segmentation problems featuring \"void\" class (commonly -1 255) segmentation maps. default (ignore_class=NULL), classes considered. sparse_y_pred Whether predictions encoded using natural numbers probability distribution vectors. FALSE, argmax function used determine sample's likely associated label. axis (Optional) dimension containing logits. Defaults -1.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_one_hot_mean_iou.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Computes mean Intersection-Over-Union metric for one-hot encoded labels. — metric_one_hot_mean_iou","text":"Metric instance returned. Metric instance can passed directly compile(metrics = ), used standalone object. See ?Metric example usage.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_one_hot_mean_iou.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Computes mean Intersection-Over-Union metric for one-hot encoded labels. — metric_one_hot_mean_iou","text":"Standalone usage:         Usage compile() API:","code":"y_true <- rbind(c(0, 0, 1), c(1, 0, 0), c(0, 1, 0), c(1, 0, 0)) y_pred <- rbind(c(0.2, 0.3, 0.5), c(0.1, 0.2, 0.7), c(0.5, 0.3, 0.1),                 c(0.1, 0.4, 0.5)) sample_weight <- 10 * c(0.1, 0.2, 0.3, 0.4) m <- metric_one_hot_mean_iou(num_classes = 3L) m$update_state(     y_true = y_true, y_pred = y_pred, sample_weight = sample_weight) ## tf.Tensor( ## [[0 0 6] ##  [3 0 0] ##  [0 0 1]], shape=(3, 3), dtype=int64) m$result() ## tf.Tensor(0.04761905, shape=(), dtype=float32) model %>% compile(     optimizer = 'sgd',     loss = 'mse',     metrics = list(metric_one_hot_mean_iou(num_classes = 3L)))"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/metric_pearson_correlation.html","id":null,"dir":"Reference","previous_headings":"","what":"Calculates the Pearson Correlation Coefficient (PCC). — metric_pearson_correlation","title":"Calculates the Pearson Correlation Coefficient (PCC). — metric_pearson_correlation","text":"Formula:   PCC measures linear relationship true values (y_true) predicted values (y_pred). coefficient ranges -1 1, value 1 implies perfect positive linear correlation, 0 indicates linear correlation, -1 indicates perfect negative linear correlation. metric widely used regression tasks strength linear relationship predictions true labels important evaluation criterion.","code":"loss = mean(l2norm(y_true - mean(y_true) * l2norm(y_pred - mean(y_pred)))"},{"path":"https://keras3.posit.co/dev/reference/metric_pearson_correlation.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Calculates the Pearson Correlation Coefficient (PCC). — metric_pearson_correlation","text":"","code":"metric_pearson_correlation(   y_true,   y_pred,   axis = -1L,   ...,   name = \"pearson_correlation\",   dtype = NULL )"},{"path":"https://keras3.posit.co/dev/reference/metric_pearson_correlation.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Calculates the Pearson Correlation Coefficient (PCC). — metric_pearson_correlation","text":"y_true Tensor true targets. y_pred Tensor predicted targets. axis (Optional) integer tuple integers axis/axes along compute metric. Defaults -1. ... forward/backward compatability. name (Optional) string name metric instance. dtype (Optional) data type metric result.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_pearson_correlation.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Calculates the Pearson Correlation Coefficient (PCC). — metric_pearson_correlation","text":"Usage compile() API:","code":"pcc <- metric_pearson_correlation(axis = -1) y_true <- rbind(c(0, 1, 0.5),                 c(1, 1, 0.2)) y_pred <- rbind(c(0.1, 0.9, 0.5),                 c(1, 0.9, 0.2)) pcc$update_state(y_true, y_pred) pcc$result() ## tf.Tensor(0.99669963, shape=(), dtype=float32) # equivalent operation using R's stats::cor() mean(sapply(1:nrow(y_true), function(i) {   cor(y_true[i, ], y_pred[i, ]) })) ## [1] 0.9966996 model |> compile(   optimizer = 'sgd',   loss = 'mean_squared_error',   metrics = c(keras.metrics.PearsonCorrelation()) )"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/metric_poisson.html","id":null,"dir":"Reference","previous_headings":"","what":"Computes the Poisson metric between y_true and y_pred. — metric_poisson","title":"Computes the Poisson metric between y_true and y_pred. — metric_poisson","text":"Formula:","code":"metric <- y_pred - y_true * log(y_pred)"},{"path":"https://keras3.posit.co/dev/reference/metric_poisson.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes the Poisson metric between y_true and y_pred. — metric_poisson","text":"","code":"metric_poisson(y_true, y_pred, ..., name = \"poisson\", dtype = NULL)"},{"path":"https://keras3.posit.co/dev/reference/metric_poisson.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Computes the Poisson metric between y_true and y_pred. — metric_poisson","text":"y_true Ground truth values. shape = [batch_size, d0, .. dN]. y_pred predicted values. shape = [batch_size, d0, .. dN]. ... forward/backward compatability. name (Optional) string name metric instance. dtype (Optional) data type metric result.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_poisson.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Computes the Poisson metric between y_true and y_pred. — metric_poisson","text":"y_true y_pred missing, Metric instance returned. Metric instance can passed directly compile(metrics = ), used standalone object. See ?Metric example usage. y_true y_pred provided, tensor computed value returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_poisson.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Computes the Poisson metric between y_true and y_pred. — metric_poisson","text":"Standalone usage:         Usage compile() API:","code":"m <- metric_poisson() m$update_state(rbind(c(0, 1), c(0, 0)), rbind(c(1, 1), c(0, 0))) m$result() ## tf.Tensor(0.49999997, shape=(), dtype=float32) m$reset_state() m$update_state(rbind(c(0, 1), c(0, 0)), rbind(c(1, 1), c(0, 0)),                sample_weight = c(1, 0)) m$result() ## tf.Tensor(0.99999994, shape=(), dtype=float32) model %>% compile(   optimizer = 'sgd',   loss = 'mse',   metrics = list(metric_poisson()) )"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/metric_precision.html","id":null,"dir":"Reference","previous_headings":"","what":"Computes the precision of the predictions with respect to the labels. — metric_precision","title":"Computes the precision of the predictions with respect to the labels. — metric_precision","text":"metric creates two local variables, true_positives false_positives used compute precision. value ultimately returned precision, idempotent operation simply divides true_positives sum true_positives false_positives. sample_weight NULL, weights default 1. Use sample_weight 0 mask values. top_k set, calculate precision often average class among top-k classes highest predicted values batch entry correct can found label entry. class_id specified, calculate precision considering entries batch class_id threshold /top-k highest predictions, computing fraction class_id indeed correct label.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_precision.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes the precision of the predictions with respect to the labels. — metric_precision","text":"","code":"metric_precision(   ...,   thresholds = NULL,   top_k = NULL,   class_id = NULL,   name = NULL,   dtype = NULL )"},{"path":"https://keras3.posit.co/dev/reference/metric_precision.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Computes the precision of the predictions with respect to the labels. — metric_precision","text":"... forward/backward compatability. thresholds (Optional) float value, Python list float threshold values [0, 1]. threshold compared prediction values determine truth value predictions (.e., threshold TRUE, FALSE). used loss function sets from_logits=TRUE (.e. sigmoid applied predictions), thresholds set 0. One metric value generated threshold value. neither thresholds top_k set, default calculate precision thresholds=0.5. top_k (Optional) Unset default. int value specifying top-k predictions consider calculating precision. class_id (Optional) Integer class ID want binary metrics. must half-open interval [0, num_classes), num_classes last dimension predictions. name (Optional) string name metric instance. dtype (Optional) data type metric result.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_precision.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Computes the precision of the predictions with respect to the labels. — metric_precision","text":"Metric instance returned. Metric instance can passed directly compile(metrics = ), used standalone object. See ?Metric example usage.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_precision.html","id":"usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes the precision of the predictions with respect to the labels. — metric_precision","text":"Standalone usage:                 Usage compile() API:   Usage loss from_logits=TRUE:","code":"m <- metric_precision() m$update_state(c(0, 1, 1, 1),                c(1, 0, 1, 1)) m$result() |> as.double() |> signif() ## [1] 0.666667 m$reset_state() m$update_state(c(0, 1, 1, 1),                c(1, 0, 1, 1),                sample_weight = c(0, 0, 1, 0)) m$result() |> as.double() |> signif() ## [1] 1 # With top_k=2, it will calculate precision over y_true[1:2] # and y_pred[1:2] m <- metric_precision(top_k = 2) m$update_state(c(0, 0, 1, 1), c(1, 1, 1, 1)) m$result() ## tf.Tensor(0.0, shape=(), dtype=float32) # With top_k=4, it will calculate precision over y_true[1:4] # and y_pred[1:4] m <- metric_precision(top_k = 4) m$update_state(c(0, 0, 1, 1), c(1, 1, 1, 1)) m$result() ## tf.Tensor(0.5, shape=(), dtype=float32) model |> compile(   optimizer = 'sgd',   loss = 'binary_crossentropy',   metrics = list(metric_precision()) ) model |> compile(   optimizer = 'adam',   loss = loss_binary_crossentropy(from_logits = TRUE),   metrics = list(metric_precision(thresholds = 0)) )"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/metric_precision_at_recall.html","id":null,"dir":"Reference","previous_headings":"","what":"Computes best precision where recall is >= specified value. — metric_precision_at_recall","title":"Computes best precision where recall is >= specified value. — metric_precision_at_recall","text":"metric creates four local variables, true_positives, true_negatives, false_positives false_negatives used compute precision given recall. threshold given recall value computed used evaluate corresponding precision. sample_weight NULL, weights default 1. Use sample_weight 0 mask values. class_id specified, calculate precision considering entries batch class_id threshold predictions, computing fraction class_id indeed correct label.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_precision_at_recall.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes best precision where recall is >= specified value. — metric_precision_at_recall","text":"","code":"metric_precision_at_recall(   ...,   recall,   num_thresholds = 200L,   class_id = NULL,   name = NULL,   dtype = NULL )"},{"path":"https://keras3.posit.co/dev/reference/metric_precision_at_recall.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Computes best precision where recall is >= specified value. — metric_precision_at_recall","text":"... forward/backward compatability. recall scalar value range [0, 1]. num_thresholds (Optional) Defaults 200. number thresholds use matching given recall. class_id (Optional) Integer class ID want binary metrics. must half-open interval [0, num_classes), num_classes last dimension predictions. name (Optional) string name metric instance. dtype (Optional) data type metric result.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_precision_at_recall.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Computes best precision where recall is >= specified value. — metric_precision_at_recall","text":"Metric instance returned. Metric instance can passed directly compile(metrics = ), used standalone object. See ?Metric example usage.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_precision_at_recall.html","id":"usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes best precision where recall is >= specified value. — metric_precision_at_recall","text":"Standalone usage:         Usage compile() API:","code":"m <- metric_precision_at_recall(recall = 0.5) m$update_state(c(0,   0,   0,   1,   1),                c(0, 0.3, 0.8, 0.3, 0.8)) m$result() ## tf.Tensor(0.5, shape=(), dtype=float32) m$reset_state() m$update_state(c(0,   0,   0,   1,   1),                c(0, 0.3, 0.8, 0.3, 0.8),                sample_weight = c(2, 2, 2, 1, 1)) m$result() ## tf.Tensor(0.33333334, shape=(), dtype=float32) model |> compile(   optimizer = 'sgd',   loss = 'binary_crossentropy',   metrics = list(metric_precision_at_recall(recall = 0.8)) )"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/metric_r2_score.html","id":null,"dir":"Reference","previous_headings":"","what":"Computes R2 score. — metric_r2_score","title":"Computes R2 score. — metric_r2_score","text":"Formula:   also called coefficient determination. indicates close fitted regression line ground-truth data. highest score possible 1.0. indicates predictors perfectly accounts variation target. score 0.0 indicates predictors account variation target. can also negative model worse random. metric can also compute \"Adjusted R2\" score.","code":"sum_squares_residuals <- sum((y_true - y_pred) ** 2) sum_squares <- sum((y_true - mean(y_true)) ** 2) R2 <- 1 - sum_squares_residuals / sum_squares"},{"path":"https://keras3.posit.co/dev/reference/metric_r2_score.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes R2 score. — metric_r2_score","text":"","code":"metric_r2_score(   ...,   class_aggregation = \"uniform_average\",   num_regressors = 0L,   name = \"r2_score\",   dtype = NULL )"},{"path":"https://keras3.posit.co/dev/reference/metric_r2_score.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Computes R2 score. — metric_r2_score","text":"... forward/backward compatability. class_aggregation Specifies aggregate scores corresponding different output classes (target dimensions), .e. different dimensions last axis predictions. Equivalent multioutput argument Scikit-Learn. one NULL (aggregation), \"uniform_average\", \"variance_weighted_average\". num_regressors Number independent regressors used (\"Adjusted R2\" score). 0 standard R2 score. Defaults 0. name Optional. string name metric instance. dtype Optional. data type metric result.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_r2_score.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Computes R2 score. — metric_r2_score","text":"Metric instance returned. Metric instance can passed directly compile(metrics = ), used standalone object. See ?Metric example usage.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_r2_score.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Computes R2 score. — metric_r2_score","text":"","code":"y_true <- rbind(1, 4, 3) y_pred <- rbind(2, 4, 4) metric <- metric_r2_score() metric$update_state(y_true, y_pred) metric$result() ## tf.Tensor(0.57142854, shape=(), dtype=float32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/metric_recall.html","id":null,"dir":"Reference","previous_headings":"","what":"Computes the recall of the predictions with respect to the labels. — metric_recall","title":"Computes the recall of the predictions with respect to the labels. — metric_recall","text":"metric creates two local variables, true_positives false_negatives, used compute recall. value ultimately returned recall, idempotent operation simply divides true_positives sum true_positives false_negatives. sample_weight NULL, weights default 1. Use sample_weight 0 mask values. top_k set, recall computed often average class among labels batch entry top-k predictions. class_id specified, calculate recall considering entries batch class_id label, computing fraction class_id threshold /top-k predictions.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_recall.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes the recall of the predictions with respect to the labels. — metric_recall","text":"","code":"metric_recall(   ...,   thresholds = NULL,   top_k = NULL,   class_id = NULL,   name = NULL,   dtype = NULL )"},{"path":"https://keras3.posit.co/dev/reference/metric_recall.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Computes the recall of the predictions with respect to the labels. — metric_recall","text":"... forward/backward compatability. thresholds (Optional) float value, Python list float threshold values [0, 1]. threshold compared prediction values determine truth value predictions (.e., threshold TRUE, FALSE). used loss function sets from_logits=TRUE (.e. sigmoid applied predictions), thresholds set 0. One metric value generated threshold value. neither thresholds top_k set, default calculate recall thresholds=0.5. top_k (Optional) Unset default. int value specifying top-k predictions consider calculating recall. class_id (Optional) Integer class ID want binary metrics. must half-open interval [0, num_classes), num_classes last dimension predictions. name (Optional) string name metric instance. dtype (Optional) data type metric result.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_recall.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Computes the recall of the predictions with respect to the labels. — metric_recall","text":"Metric instance returned. Metric instance can passed directly compile(metrics = ), used standalone object. See ?Metric example usage.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_recall.html","id":"usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes the recall of the predictions with respect to the labels. — metric_recall","text":"Standalone usage:         Usage compile() API:   Usage loss from_logits=TRUE:","code":"m <- metric_recall() m$update_state(c(0, 1, 1, 1),                c(1, 0, 1, 1)) m$result() ## tf.Tensor(0.6666667, shape=(), dtype=float32) m$reset_state() m$update_state(c(0, 1, 1, 1),                c(1, 0, 1, 1),                sample_weight = c(0, 0, 1, 0)) m$result() ## tf.Tensor(1.0, shape=(), dtype=float32) model |> compile(   optimizer = 'sgd',   loss = 'binary_crossentropy',   metrics = list(metric_recall()) ) model |> compile(   optimizer = 'adam',   loss = loss_binary_crossentropy(from_logits = TRUE),   metrics = list(metric_recall(thresholds = 0)) )"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/metric_recall_at_precision.html","id":null,"dir":"Reference","previous_headings":"","what":"Computes best recall where precision is >= specified value. — metric_recall_at_precision","title":"Computes best recall where precision is >= specified value. — metric_recall_at_precision","text":"given score-label-distribution required precision might achievable, case 0.0 returned recall. metric creates four local variables, true_positives, true_negatives, false_positives false_negatives used compute recall given precision. threshold given precision value computed used evaluate corresponding recall. sample_weight NULL, weights default 1. Use sample_weight 0 mask values. class_id specified, calculate precision considering entries batch class_id threshold predictions, computing fraction class_id indeed correct label.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_recall_at_precision.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes best recall where precision is >= specified value. — metric_recall_at_precision","text":"","code":"metric_recall_at_precision(   ...,   precision,   num_thresholds = 200L,   class_id = NULL,   name = NULL,   dtype = NULL )"},{"path":"https://keras3.posit.co/dev/reference/metric_recall_at_precision.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Computes best recall where precision is >= specified value. — metric_recall_at_precision","text":"... forward/backward compatability. precision scalar value range [0, 1]. num_thresholds (Optional) Defaults 200. number thresholds use matching given precision. class_id (Optional) Integer class ID want binary metrics. must half-open interval [0, num_classes), num_classes last dimension predictions. name (Optional) string name metric instance. dtype (Optional) data type metric result.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_recall_at_precision.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Computes best recall where precision is >= specified value. — metric_recall_at_precision","text":"Metric instance returned. Metric instance can passed directly compile(metrics = ), used standalone object. See ?Metric example usage.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_recall_at_precision.html","id":"usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes best recall where precision is >= specified value. — metric_recall_at_precision","text":"Standalone usage:         Usage compile() API:","code":"m <- metric_recall_at_precision(precision = 0.8) m$update_state(c(0,   0,   1,   1),                c(0, 0.5, 0.3, 0.9)) m$result() ## tf.Tensor(0.5, shape=(), dtype=float32) m$reset_state() m$update_state(c(0,   0,   1,   1),                c(0, 0.5, 0.3, 0.9),                sample_weight = c(1, 0, 0, 1)) m$result() ## tf.Tensor(1.0, shape=(), dtype=float32) model |> compile(   optimizer = 'sgd',   loss = 'binary_crossentropy',   metrics = list(metric_recall_at_precision(precision = 0.8)) )"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/metric_root_mean_squared_error.html","id":null,"dir":"Reference","previous_headings":"","what":"Computes root mean squared error metric between y_true and y_pred. — metric_root_mean_squared_error","title":"Computes root mean squared error metric between y_true and y_pred. — metric_root_mean_squared_error","text":"Formula:","code":"loss <- sqrt(mean((y_pred - y_true) ^ 2))"},{"path":"https://keras3.posit.co/dev/reference/metric_root_mean_squared_error.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes root mean squared error metric between y_true and y_pred. — metric_root_mean_squared_error","text":"","code":"metric_root_mean_squared_error(   ...,   name = \"root_mean_squared_error\",   dtype = NULL )"},{"path":"https://keras3.posit.co/dev/reference/metric_root_mean_squared_error.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Computes root mean squared error metric between y_true and y_pred. — metric_root_mean_squared_error","text":"... forward/backward compatability. name (Optional) string name metric instance. dtype (Optional) data type metric result.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_root_mean_squared_error.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Computes root mean squared error metric between y_true and y_pred. — metric_root_mean_squared_error","text":"Metric instance returned. Metric instance can passed directly compile(metrics = ), used standalone object. See ?Metric example usage.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_root_mean_squared_error.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Computes root mean squared error metric between y_true and y_pred. — metric_root_mean_squared_error","text":"Standalone usage:         Usage compile() API:","code":"m <- metric_root_mean_squared_error() m$update_state(rbind(c(0, 1), c(0, 0)), rbind(c(1, 1), c(0, 0))) m$result() ## tf.Tensor(0.5, shape=(), dtype=float32) m$reset_state() m$update_state(rbind(c(0, 1), c(0, 0)), rbind(c(1, 1), c(0, 0)),                sample_weight = c(1, 0)) m$result() ## tf.Tensor(0.70710677, shape=(), dtype=float32) model %>% compile(   optimizer = 'sgd',   loss = 'mse',   metrics = list(metric_root_mean_squared_error()))"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/metric_sensitivity_at_specificity.html","id":null,"dir":"Reference","previous_headings":"","what":"Computes best sensitivity where specificity is >= specified value. — metric_sensitivity_at_specificity","title":"Computes best sensitivity where specificity is >= specified value. — metric_sensitivity_at_specificity","text":"Sensitivity measures proportion actual positives correctly identified (tp / (tp + fn)). Specificity measures proportion actual negatives correctly identified (tn / (tn + fp)). metric creates four local variables, true_positives, true_negatives, false_positives false_negatives used compute sensitivity given specificity. threshold given specificity value computed used evaluate corresponding sensitivity. sample_weight NULL, weights default 1. Use sample_weight 0 mask values. class_id specified, calculate precision considering entries batch class_id threshold predictions, computing fraction class_id indeed correct label. additional information specificity sensitivity, see following.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_sensitivity_at_specificity.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes best sensitivity where specificity is >= specified value. — metric_sensitivity_at_specificity","text":"","code":"metric_sensitivity_at_specificity(   ...,   specificity,   num_thresholds = 200L,   class_id = NULL,   name = NULL,   dtype = NULL )"},{"path":"https://keras3.posit.co/dev/reference/metric_sensitivity_at_specificity.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Computes best sensitivity where specificity is >= specified value. — metric_sensitivity_at_specificity","text":"... forward/backward compatability. specificity scalar value range [0, 1]. num_thresholds (Optional) Defaults 200. number thresholds use matching given specificity. class_id (Optional) Integer class ID want binary metrics. must half-open interval [0, num_classes), num_classes last dimension predictions. name (Optional) string name metric instance. dtype (Optional) data type metric result.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_sensitivity_at_specificity.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Computes best sensitivity where specificity is >= specified value. — metric_sensitivity_at_specificity","text":"Metric instance returned. Metric instance can passed directly compile(metrics = ), used standalone object. See ?Metric example usage.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_sensitivity_at_specificity.html","id":"usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes best sensitivity where specificity is >= specified value. — metric_sensitivity_at_specificity","text":"Standalone usage:         Usage compile() API:","code":"m <- metric_sensitivity_at_specificity(specificity = 0.5) m$update_state(c(0,   0,   0,   1,   1),                c(0, 0.3, 0.8, 0.3, 0.8)) m$result() ## tf.Tensor(0.5, shape=(), dtype=float32) m$reset_state() m$update_state(c(0,   0,   0,   1,   1),                c(0, 0.3, 0.8, 0.3, 0.8),                sample_weight = c(1, 1, 2, 2, 1)) m$result() ## tf.Tensor(0.33333334, shape=(), dtype=float32) model |> compile(   optimizer = 'sgd',   loss = 'binary_crossentropy',   metrics = list(metric_sensitivity_at_specificity()) )"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/metric_sparse_categorical_accuracy.html","id":null,"dir":"Reference","previous_headings":"","what":"Calculates how often predictions match integer labels. — metric_sparse_categorical_accuracy","title":"Calculates how often predictions match integer labels. — metric_sparse_categorical_accuracy","text":"can provide logits classes y_pred, since argmax logits probabilities . metric creates two local variables, total count used compute frequency y_pred matches y_true. frequency ultimately returned sparse categorical accuracy: idempotent operation simply divides total count. sample_weight NULL, weights default 1. Use sample_weight 0 mask values.","code":"acc <- sample_weight %*% (y_true == which.max(y_pred))"},{"path":"https://keras3.posit.co/dev/reference/metric_sparse_categorical_accuracy.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Calculates how often predictions match integer labels. — metric_sparse_categorical_accuracy","text":"","code":"metric_sparse_categorical_accuracy(   y_true,   y_pred,   ...,   name = \"sparse_categorical_accuracy\",   dtype = NULL )"},{"path":"https://keras3.posit.co/dev/reference/metric_sparse_categorical_accuracy.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Calculates how often predictions match integer labels. — metric_sparse_categorical_accuracy","text":"y_true Tensor true targets. y_pred Tensor predicted targets. ... forward/backward compatability. name (Optional) string name metric instance. dtype (Optional) data type metric result.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_sparse_categorical_accuracy.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Calculates how often predictions match integer labels. — metric_sparse_categorical_accuracy","text":"y_true y_pred missing, Metric instance returned. Metric instance can passed directly compile(metrics = ), used standalone object. See ?Metric example usage. y_true y_pred provided, tensor computed value returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_sparse_categorical_accuracy.html","id":"usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Calculates how often predictions match integer labels. — metric_sparse_categorical_accuracy","text":"Standalone usage:         Usage compile() API:","code":"m <- metric_sparse_categorical_accuracy() m$update_state(rbind(2L, 1L), rbind(c(0.1, 0.6, 0.3), c(0.05, 0.95, 0))) m$result() ## tf.Tensor(0.5, shape=(), dtype=float32) m$reset_state() m$update_state(rbind(2L, 1L), rbind(c(0.1, 0.6, 0.3), c(0.05, 0.95, 0)),                sample_weight = c(0.7, 0.3)) m$result() ## tf.Tensor(0.3, shape=(), dtype=float32) model |> compile(optimizer = 'sgd',                  loss = 'sparse_categorical_crossentropy',                  metrics = list(metric_sparse_categorical_accuracy()))"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/metric_sparse_categorical_crossentropy.html","id":null,"dir":"Reference","previous_headings":"","what":"Computes the crossentropy metric between the labels and predictions. — metric_sparse_categorical_crossentropy","title":"Computes the crossentropy metric between the labels and predictions. — metric_sparse_categorical_crossentropy","text":"Use crossentropy metric two label classes. expects labels provided integers. want provide labels one-hot encoded, please use metric_categorical_crossentropy() metric instead. num_classes floating point values per feature y_pred single floating point value per feature y_true.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_sparse_categorical_crossentropy.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes the crossentropy metric between the labels and predictions. — metric_sparse_categorical_crossentropy","text":"","code":"metric_sparse_categorical_crossentropy(   y_true,   y_pred,   from_logits = FALSE,   ignore_class = NULL,   axis = -1L,   ...,   name = \"sparse_categorical_crossentropy\",   dtype = NULL )"},{"path":"https://keras3.posit.co/dev/reference/metric_sparse_categorical_crossentropy.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Computes the crossentropy metric between the labels and predictions. — metric_sparse_categorical_crossentropy","text":"y_true Ground truth values. y_pred predicted values. from_logits (Optional) Whether output expected logits tensor. default, consider output encodes probability distribution. ignore_class Optional integer. ID class ignored loss computation. useful, example, segmentation problems featuring \"void\" class (commonly -1 255) segmentation maps. default (ignore_class=NULL), classes considered. axis (Optional) Defaults -1. dimension along entropy computed. ... forward/backward compatability. name (Optional) string name metric instance. dtype (Optional) data type metric result.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_sparse_categorical_crossentropy.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Computes the crossentropy metric between the labels and predictions. — metric_sparse_categorical_crossentropy","text":"y_true y_pred missing, Metric instance returned. Metric instance can passed directly compile(metrics = ), used standalone object. See ?Metric example usage. y_true y_pred provided, tensor computed value returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_sparse_categorical_crossentropy.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Computes the crossentropy metric between the labels and predictions. — metric_sparse_categorical_crossentropy","text":"Standalone usage:         Usage compile() API:","code":"m <- metric_sparse_categorical_crossentropy() m$update_state(array(c(1, 2)),                rbind(c(0.05, 0.95, 0), c(0.1, 0.8, 0.1))) m$result() ## tf.Tensor(1.1769392, shape=(), dtype=float32) m$reset_state() m$update_state(array(c(1, 2)),                rbind(c(0.05, 0.95, 0), c(0.1, 0.8, 0.1)),                sample_weight = c(0.3, 0.7)) m$result() ## tf.Tensor(1.6271976, shape=(), dtype=float32) model |> compile(   optimizer = 'sgd',   loss = 'mse',   metrics = list(metric_sparse_categorical_crossentropy()) )"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/metric_sparse_top_k_categorical_accuracy.html","id":null,"dir":"Reference","previous_headings":"","what":"Computes how often integer targets are in the top K predictions. — metric_sparse_top_k_categorical_accuracy","title":"Computes how often integer targets are in the top K predictions. — metric_sparse_top_k_categorical_accuracy","text":"Computes often integer targets top K predictions. default, arguments expected update_state() : y_true: tensor shape (batch_size) representing indices true categories. y_pred: tensor shape (batch_size, num_categories) containing scores sample possible categories. from_sorted_ids=TRUE, arguments expected update_state : y_true: tensor shape (batch_size) representing indices IDs true categories. y_pred: tensor shape (batch_size, N) containing indices IDs top N categories sorted order highest score lowest score. N must greater equal k. from_sorted_ids=TRUE option can efficient set categories large model optimized way retrieve top ones either without scoring without maintaining scores possible categories.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_sparse_top_k_categorical_accuracy.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes how often integer targets are in the top K predictions. — metric_sparse_top_k_categorical_accuracy","text":"","code":"metric_sparse_top_k_categorical_accuracy(   y_true,   y_pred,   k = 5L,   ...,   name = \"sparse_top_k_categorical_accuracy\",   dtype = NULL,   from_sorted_ids = FALSE )"},{"path":"https://keras3.posit.co/dev/reference/metric_sparse_top_k_categorical_accuracy.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Computes how often integer targets are in the top K predictions. — metric_sparse_top_k_categorical_accuracy","text":"y_true Tensor true targets. y_pred Tensor predicted targets. k (Optional) Number top elements look computing accuracy. Defaults 5. ... forward/backward compatability. name (Optional) string name metric instance. dtype (Optional) data type metric result. from_sorted_ids (Optional) FALSE, default, tensor passed y_pred contains unsorted scores possible categories. TRUE, y_pred contains indices IDs top categories.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_sparse_top_k_categorical_accuracy.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Computes how often integer targets are in the top K predictions. — metric_sparse_top_k_categorical_accuracy","text":"y_true y_pred missing, Metric instance returned. Metric instance can passed directly compile(metrics = ), used standalone object. See ?Metric example usage. y_true y_pred provided, tensor computed value returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_sparse_top_k_categorical_accuracy.html","id":"usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes how often integer targets are in the top K predictions. — metric_sparse_top_k_categorical_accuracy","text":"Standalone usage:             Usage compile() API:","code":"m <- metric_sparse_top_k_categorical_accuracy(k = 1L) m$update_state(   rbind(2, 1),   op_array(rbind(c(0.1, 0.9, 0.8), c(0.05, 0.95, 0)), dtype = \"float32\") ) m$result() ## tf.Tensor(0.5, shape=(), dtype=float32) m$reset_state() m$update_state(   rbind(2, 1),   op_array(rbind(c(0.1, 0.9, 0.8), c(0.05, 0.95, 0)), dtype = \"float32\"),   sample_weight = c(0.7, 0.3) ) m$result() ## tf.Tensor(0.3, shape=(), dtype=float32) m <- metric_sparse_top_k_categorical_accuracy(k = 1, from_sorted_ids = TRUE) m$update_state(array(c(2, 1)), rbind(c(1, 0, 3),                                      c(1, 2, 3))) m$result() ## tf.Tensor(0.5, shape=(), dtype=float32) model %>% compile(optimizer = 'sgd',                   loss = 'sparse_categorical_crossentropy',                   metrics = list(metric_sparse_top_k_categorical_accuracy()))"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/metric_specificity_at_sensitivity.html","id":null,"dir":"Reference","previous_headings":"","what":"Computes best specificity where sensitivity is >= specified value. — metric_specificity_at_sensitivity","title":"Computes best specificity where sensitivity is >= specified value. — metric_specificity_at_sensitivity","text":"Sensitivity measures proportion actual positives correctly identified (tp / (tp + fn)). Specificity measures proportion actual negatives correctly identified (tn / (tn + fp)). metric creates four local variables, true_positives, true_negatives, false_positives false_negatives used compute specificity given sensitivity. threshold given sensitivity value computed used evaluate corresponding specificity. sample_weight NULL, weights default 1. Use sample_weight 0 mask values. class_id specified, calculate precision considering entries batch class_id threshold predictions, computing fraction class_id indeed correct label. additional information specificity sensitivity, see following.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_specificity_at_sensitivity.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes best specificity where sensitivity is >= specified value. — metric_specificity_at_sensitivity","text":"","code":"metric_specificity_at_sensitivity(   ...,   sensitivity,   num_thresholds = 200L,   class_id = NULL,   name = NULL,   dtype = NULL )"},{"path":"https://keras3.posit.co/dev/reference/metric_specificity_at_sensitivity.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Computes best specificity where sensitivity is >= specified value. — metric_specificity_at_sensitivity","text":"... forward/backward compatability. sensitivity scalar value range [0, 1]. num_thresholds (Optional) Defaults 200. number thresholds use matching given sensitivity. class_id (Optional) Integer class ID want binary metrics. must half-open interval [0, num_classes), num_classes last dimension predictions. name (Optional) string name metric instance. dtype (Optional) data type metric result.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_specificity_at_sensitivity.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Computes best specificity where sensitivity is >= specified value. — metric_specificity_at_sensitivity","text":"Metric instance returned. Metric instance can passed directly compile(metrics = ), used standalone object. See ?Metric example usage.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_specificity_at_sensitivity.html","id":"usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes best specificity where sensitivity is >= specified value. — metric_specificity_at_sensitivity","text":"Standalone usage:         Usage compile() API:","code":"m <- metric_specificity_at_sensitivity(sensitivity = 0.5) m$update_state(c(0,   0,   0,   1,   1),                c(0, 0.3, 0.8, 0.3, 0.8)) m$result() ## tf.Tensor(0.6666667, shape=(), dtype=float32) m$reset_state() m$update_state(c(0,   0,   0,   1,   1),                c(0, 0.3, 0.8, 0.3, 0.8),                sample_weight = c(1, 1, 2, 2, 2)) m$result() ## tf.Tensor(0.5, shape=(), dtype=float32) model |> compile(   optimizer = 'sgd',   loss = 'binary_crossentropy',   metrics = list(metric_sensitivity_at_specificity()) )"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/metric_squared_hinge.html","id":null,"dir":"Reference","previous_headings":"","what":"Computes the hinge metric between y_true and y_pred. — metric_squared_hinge","title":"Computes the hinge metric between y_true and y_pred. — metric_squared_hinge","text":"Formula:   y_true values expected -1 1. binary (0 1) labels provided convert -1 1.","code":"loss <- mean(square(maximum(1 - y_true * y_pred, 0)))"},{"path":"https://keras3.posit.co/dev/reference/metric_squared_hinge.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes the hinge metric between y_true and y_pred. — metric_squared_hinge","text":"","code":"metric_squared_hinge(y_true, y_pred, ..., name = \"squared_hinge\", dtype = NULL)"},{"path":"https://keras3.posit.co/dev/reference/metric_squared_hinge.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Computes the hinge metric between y_true and y_pred. — metric_squared_hinge","text":"y_true ground truth values. y_true values expected -1 1. binary (0 1) labels provided convert -1 1 shape = [batch_size, d0, .. dN]. y_pred predicted values shape = [batch_size, d0, .. dN]. ... forward/backward compatability. name (Optional) string name metric instance. dtype (Optional) data type metric result.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_squared_hinge.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Computes the hinge metric between y_true and y_pred. — metric_squared_hinge","text":"y_true y_pred missing, Metric instance returned. Metric instance can passed directly compile(metrics = ), used standalone object. See ?Metric example usage. y_true y_pred provided, tensor computed value returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_squared_hinge.html","id":"usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes the hinge metric between y_true and y_pred. — metric_squared_hinge","text":"Standalone usage:","code":"m <- metric_squared_hinge() m$update_state(rbind(c(0, 1), c(0, 0)), rbind(c(0.6, 0.4), c(0.4, 0.6))) m$result() ## tf.Tensor(1.86, shape=(), dtype=float32) m$reset_state() m$update_state(rbind(c(0, 1), c(0, 0)), rbind(c(0.6, 0.4), c(0.4, 0.6)),                sample_weight = c(1, 0)) m$result() ## tf.Tensor(1.46, shape=(), dtype=float32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/metric_sum.html","id":null,"dir":"Reference","previous_headings":"","what":"Compute the (weighted) sum of the given values. — metric_sum","title":"Compute the (weighted) sum of the given values. — metric_sum","text":"example, values [1, 3, 5, 7] sum 16. sample_weight specified [1, 1, 0, 0] sum 4. metric creates one variable, total. ultimately returned sum value.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_sum.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Compute the (weighted) sum of the given values. — metric_sum","text":"","code":"metric_sum(..., name = \"sum\", dtype = NULL)"},{"path":"https://keras3.posit.co/dev/reference/metric_sum.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Compute the (weighted) sum of the given values. — metric_sum","text":"... forward/backward compatability. name (Optional) string name metric instance. dtype (Optional) data type metric result.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_sum.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Compute the (weighted) sum of the given values. — metric_sum","text":"Metric instance returned. Metric instance can passed directly compile(metrics = ), used standalone object. See ?Metric example usage.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_sum.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Compute the (weighted) sum of the given values. — metric_sum","text":"","code":"m <- metric_sum() m$update_state(c(1, 3, 5, 7)) m$result() ## tf.Tensor(16.0, shape=(), dtype=float32) m <- metric_sum() m$update_state(c(1, 3, 5, 7), sample_weight = c(1, 1, 0, 0)) m$result() ## tf.Tensor(4.0, shape=(), dtype=float32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/metric_top_k_categorical_accuracy.html","id":null,"dir":"Reference","previous_headings":"","what":"Computes how often targets are in the top K predictions. — metric_top_k_categorical_accuracy","title":"Computes how often targets are in the top K predictions. — metric_top_k_categorical_accuracy","text":"Computes often targets top K predictions.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_top_k_categorical_accuracy.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes how often targets are in the top K predictions. — metric_top_k_categorical_accuracy","text":"","code":"metric_top_k_categorical_accuracy(   y_true,   y_pred,   k = 5L,   ...,   name = \"top_k_categorical_accuracy\",   dtype = NULL )"},{"path":"https://keras3.posit.co/dev/reference/metric_top_k_categorical_accuracy.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Computes how often targets are in the top K predictions. — metric_top_k_categorical_accuracy","text":"y_true Tensor true targets. y_pred Tensor predicted targets. k (Optional) Number top elements look computing accuracy. Defaults 5. ... forward/backward compatability. name (Optional) string name metric instance. dtype (Optional) data type metric result.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_top_k_categorical_accuracy.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Computes how often targets are in the top K predictions. — metric_top_k_categorical_accuracy","text":"y_true y_pred missing, Metric instance returned. Metric instance can passed directly compile(metrics = ), used standalone object. See ?Metric example usage. y_true y_pred provided, tensor computed value returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_top_k_categorical_accuracy.html","id":"usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes how often targets are in the top K predictions. — metric_top_k_categorical_accuracy","text":"Standalone usage:         Usage compile() API:","code":"m <- metric_top_k_categorical_accuracy(k = 1) m$update_state(   rbind(c(0, 0, 1), c(0, 1, 0)),   op_array(rbind(c(0.1, 0.9, 0.8), c(0.05, 0.95, 0)), dtype = \"float32\") ) m$result() ## tf.Tensor(0.5, shape=(), dtype=float32) m$reset_state() m$update_state(   rbind(c(0, 0, 1), c(0, 1, 0)),   op_array(rbind(c(0.1, 0.9, 0.8), c(0.05, 0.95, 0)), dtype = \"float32\"),   sample_weight = c(0.7, 0.3)) m$result() ## tf.Tensor(0.3, shape=(), dtype=float32) model |> compile(optimizer = 'sgd',                  loss = 'categorical_crossentropy',                  metrics = list(metric_top_k_categorical_accuracy()))"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/metric_true_negatives.html","id":null,"dir":"Reference","previous_headings":"","what":"Calculates the number of true negatives. — metric_true_negatives","title":"Calculates the number of true negatives. — metric_true_negatives","text":"sample_weight given, calculates sum weights true negatives. metric creates one local variable, accumulator used keep track number true negatives. sample_weight NULL, weights default 1. Use sample_weight 0 mask values.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_true_negatives.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Calculates the number of true negatives. — metric_true_negatives","text":"","code":"metric_true_negatives(..., thresholds = NULL, name = NULL, dtype = NULL)"},{"path":"https://keras3.posit.co/dev/reference/metric_true_negatives.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Calculates the number of true negatives. — metric_true_negatives","text":"... forward/backward compatability. thresholds (Optional) Defaults 0.5. float value, Python list float threshold values [0, 1]. threshold compared prediction values determine truth value predictions (.e., threshold TRUE, FALSE). used loss function sets from_logits=TRUE (.e. sigmoid applied predictions), thresholds set 0. One metric value generated threshold value. name (Optional) string name metric instance. dtype (Optional) data type metric result.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_true_negatives.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Calculates the number of true negatives. — metric_true_negatives","text":"Metric instance returned. Metric instance can passed directly compile(metrics = ), used standalone object. See ?Metric example usage.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_true_negatives.html","id":"usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Calculates the number of true negatives. — metric_true_negatives","text":"Standalone usage:","code":"m <- metric_true_negatives() m$update_state(c(0, 1, 0, 0), c(1, 1, 0, 0)) m$result() ## tf.Tensor(2.0, shape=(), dtype=float32) m$reset_state() m$update_state(c(0, 1, 0, 0), c(1, 1, 0, 0), sample_weight = c(0, 0, 1, 0)) m$result() ## tf.Tensor(1.0, shape=(), dtype=float32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/metric_true_positives.html","id":null,"dir":"Reference","previous_headings":"","what":"Calculates the number of true positives. — metric_true_positives","title":"Calculates the number of true positives. — metric_true_positives","text":"sample_weight given, calculates sum weights true positives. metric creates one local variable, true_positives used keep track number true positives. sample_weight NULL, weights default 1. Use sample_weight 0 mask values.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_true_positives.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Calculates the number of true positives. — metric_true_positives","text":"","code":"metric_true_positives(..., thresholds = NULL, name = NULL, dtype = NULL)"},{"path":"https://keras3.posit.co/dev/reference/metric_true_positives.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Calculates the number of true positives. — metric_true_positives","text":"... forward/backward compatability. thresholds (Optional) Defaults 0.5. float value, Python list float threshold values [0, 1]. threshold compared prediction values determine truth value predictions (.e., threshold TRUE, FALSE). used loss function sets from_logits=TRUE (.e. sigmoid applied predictions), thresholds set 0. One metric value generated threshold value. name (Optional) string name metric instance. dtype (Optional) data type metric result.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_true_positives.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Calculates the number of true positives. — metric_true_positives","text":"Metric instance returned. Metric instance can passed directly compile(metrics = ), used standalone object. See ?Metric example usage.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_true_positives.html","id":"usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Calculates the number of true positives. — metric_true_positives","text":"Standalone usage:","code":"m <- metric_true_positives() m$update_state(c(0, 1, 1, 1), c(1, 0, 1, 1)) m$result() ## tf.Tensor(2.0, shape=(), dtype=float32) m$reset_state() m$update_state(c(0, 1, 1, 1), c(1, 0, 1, 1), sample_weight = c(0, 0, 1, 0)) m$result() ## tf.Tensor(1.0, shape=(), dtype=float32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/multi-assign.html","id":null,"dir":"Reference","previous_headings":"","what":"Assign values to names — %<-%","title":"Assign values to names — %<-%","text":"See %<-% details.","code":""},{"path":"https://keras3.posit.co/dev/reference/multi-assign.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Assign values to names — %<-%","text":"","code":"x %<-% value"},{"path":"https://keras3.posit.co/dev/reference/multi-assign.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Assign values to names — %<-%","text":"right-hand-side argument, value, invisibly. called primarily side-effect assigning symbols current frame.","code":""},{"path":"https://keras3.posit.co/dev/reference/new_callback_class.html","id":null,"dir":"Reference","previous_headings":"","what":"Callback — new_callback_class","title":"Callback — new_callback_class","text":"new_callback_class() alias Callback(). See ?Callback() full documentation.","code":""},{"path":"https://keras3.posit.co/dev/reference/new_callback_class.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Callback — new_callback_class","text":"","code":"new_callback_class(   classname,   on_epoch_begin = NULL,   on_epoch_end = NULL,   on_train_begin = NULL,   on_train_end = NULL,   on_train_batch_begin = NULL,   on_train_batch_end = NULL,   on_test_begin = NULL,   on_test_end = NULL,   on_test_batch_begin = NULL,   on_test_batch_end = NULL,   on_predict_begin = NULL,   on_predict_end = NULL,   on_predict_batch_begin = NULL,   on_predict_batch_end = NULL,   ...,   public = list(),   private = list(),   inherit = NULL,   parent_env = parent.frame() )"},{"path":"https://keras3.posit.co/dev/reference/new_callback_class.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Callback — new_callback_class","text":"classname String, name custom class. (Conventionally, CamelCase). on_epoch_begin   Called start epoch. Subclasses override actions run. function called TRAIN mode. Args: epoch: Integer, index epoch. logs: Named List. Currently data passed argument method may change future. on_epoch_end   Called end epoch. Subclasses override actions run. function called TRAIN mode. Args: epoch: Integer, index epoch. logs: Named List, metric results training epoch, validation epoch validation performed. Validation result keys prefixed val_. training epoch, values Model's metrics returned. Example: list(loss = 0.2, accuracy = 0.7). on_train_begin   Called beginning training. Subclasses override actions run. Args: logs: Named list. Currently data passed argument method may change future. on_train_end   Called end training. Subclasses override actions run. Args: logs: Named list. Currently output last call on_epoch_end() passed argument method may change future. on_train_batch_begin   Called beginning training batch fit() methods. Subclasses override actions run. Note steps_per_execution argument compile Model set N, method called every N batches. Args: batch: Integer, index batch within current epoch. logs: Named list. Currently data passed argument method may change future. on_train_batch_end   Called end training batch fit() methods. Subclasses override actions run. Note steps_per_execution argument compile Model set N, method called every N batches. Args: batch: Integer, index batch within current epoch. logs: Named list. Aggregated metric results batch. on_test_begin   Called beginning evaluation validation. Subclasses override actions run. Args: logs: Named list. Currently data passed argument method may change future. on_test_end   Called end evaluation validation. Subclasses override actions run. Args: logs: Named list. Currently output last call on_test_batch_end() passed argument method may change future. on_test_batch_begin   Called beginning batch evaluate() methods. Also called beginning validation batch fit() methods, validation data provided. Subclasses override actions run. Note steps_per_execution argument compile() Model set N, method called every N batches. Args: batch: Integer, index batch within current epoch. logs: Named list. Currently data passed argument method may change future. on_test_batch_end   Called end batch evaluate() methods. Also called end validation batch fit() methods, validation data provided. Subclasses override actions run. Note steps_per_execution argument compile() Model set N, method called every N batches. Args: batch: Integer, index batch within current epoch. logs: Named list. Aggregated metric results batch. on_predict_begin   Called beginning prediction. Subclasses override actions run. Args: logs: Named list. Currently data passed argument method may change future. on_predict_end   Called end prediction. Subclasses override actions run. Args: logs: Named list. Currently data passed argument method may change future. on_predict_batch_begin   Called beginning batch predict() methods. Subclasses override actions run. Note steps_per_execution argument compile() Model set N, method called every N batches. Args: batch: Integer, index batch within current epoch. logs: Named list. Currently data passed argument method may change future. on_predict_batch_end   Called end batch predict() methods. Subclasses override actions run. Note steps_per_execution argument compile Model set N, method called every N batches. Args: batch: Integer, index batch within current epoch. logs: Named list. Aggregated metric results batch. ..., public Additional methods public members custom class. private Named list R objects (typically, functions) include instance private environments. private methods symbols scope public methods (See section \"Symbols Scope\"). instance private environment. objects private invisible Keras framework Python runtime. inherit custom class subclass. default, base keras class. parent_env R environment class methods grandparent.","code":"\\(epoch, logs = NULL) \\(epoch, logs = NULL) \\(logs = NULL) \\(logs = NULL) \\(batch, logs = NULL) \\(batch, logs=NULL) \\(logs = NULL) \\(logs = NULL) \\(batch, logs = NULL) \\(batch, logs = NULL) \\(logs = NULL) \\(logs = NULL) \\(batch, logs = NULL) \\(batch, logs = NULL)"},{"path":"https://keras3.posit.co/dev/reference/new_callback_class.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Callback — new_callback_class","text":"function returns custom Callback instances, similar builtin callback functions.","code":""},{"path":"https://keras3.posit.co/dev/reference/new_layer_class.html","id":null,"dir":"Reference","previous_headings":"","what":"Layer — new_layer_class","title":"Layer — new_layer_class","text":"new_layer_class() alias Layer(). See ?Layer() full documentation.","code":""},{"path":"https://keras3.posit.co/dev/reference/new_layer_class.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Layer — new_layer_class","text":"","code":"new_layer_class(   classname,   initialize = NULL,   call = NULL,   build = NULL,   get_config = NULL,   ...,   public = list(),   private = list(),   inherit = NULL,   parent_env = parent.frame() )"},{"path":"https://keras3.posit.co/dev/reference/new_layer_class.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Layer — new_layer_class","text":"classname String, name custom class. (Conventionally, CamelCase). initialize, call, build, get_config Recommended methods implement. See description details sections. ..., public Additional methods public members custom class. private Named list R objects (typically, functions) include instance private environments. private methods symbols scope public methods (See section \"Symbols Scope\"). instance private environment. objects private invisible Keras framework Python runtime. inherit custom class subclass. default, base keras class. parent_env R environment class methods grandparent.","code":""},{"path":"https://keras3.posit.co/dev/reference/new_layer_class.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Layer — new_layer_class","text":"composing layer constructor, similar behavior layer functions like layer_dense(). first argument returned function object, enabling initialize()ing call() layer one step composing layer pipe, like   initialize() layer instance call() , pass missing NULL value object, pass arguments initialize() name.","code":"layer_foo <- Layer(\"Foo\", ....) output <- inputs |> layer_foo() layer <- layer_dense(units = 2, activation = \"relu\") layer <- layer_dense(NULL, 2, activation = \"relu\") layer <- layer_dense(, 2, activation = \"relu\")  # then you can call() the layer in a separate step outputs <- inputs |> layer()"},{"path":"https://keras3.posit.co/dev/reference/new_learning_rate_schedule_class.html","id":null,"dir":"Reference","previous_headings":"","what":"LearningRateSchedule — new_learning_rate_schedule_class","title":"LearningRateSchedule — new_learning_rate_schedule_class","text":"new_learning_rate_schedule_class() alias LearningRateSchedule(). See ?LearningRateSchedule() full documentation.","code":""},{"path":"https://keras3.posit.co/dev/reference/new_learning_rate_schedule_class.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"LearningRateSchedule — new_learning_rate_schedule_class","text":"","code":"new_learning_rate_schedule_class(   classname,   call = NULL,   initialize = NULL,   get_config = NULL,   ...,   public = list(),   private = list(),   inherit = NULL,   parent_env = parent.frame() )"},{"path":"https://keras3.posit.co/dev/reference/new_learning_rate_schedule_class.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"LearningRateSchedule — new_learning_rate_schedule_class","text":"classname String, name custom class. (Conventionally, CamelCase). call, initialize, get_config Recommended methods implement. See description details sections. ..., public Additional methods public members custom class. private Named list R objects (typically, functions) include instance private environments. private methods symbols scope public methods (See section \"Symbols Scope\"). instance private environment. objects private invisible Keras framework Python runtime. inherit custom class subclass. default, base keras class. parent_env R environment class methods grandparent.","code":""},{"path":"https://keras3.posit.co/dev/reference/new_learning_rate_schedule_class.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"LearningRateSchedule — new_learning_rate_schedule_class","text":"function returns LearningRateSchedule instances, similar built-learning_rate_schedule_* family functions.","code":""},{"path":"https://keras3.posit.co/dev/reference/new_loss_class.html","id":null,"dir":"Reference","previous_headings":"","what":"Loss — new_loss_class","title":"Loss — new_loss_class","text":"new_loss_class() alias Loss(). See ?Loss() full documentation.","code":""},{"path":"https://keras3.posit.co/dev/reference/new_loss_class.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Loss — new_loss_class","text":"","code":"new_loss_class(   classname,   call = NULL,   ...,   public = list(),   private = list(),   inherit = NULL,   parent_env = parent.frame() )"},{"path":"https://keras3.posit.co/dev/reference/new_loss_class.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Loss — new_loss_class","text":"classname String, name custom class. (Conventionally, CamelCase). call   Method implemented subclasses: Function contains logic loss calculation using y_true, y_pred. ..., public Additional methods public members custom class. private Named list R objects (typically, functions) include instance private environments. private methods symbols scope public methods (See section \"Symbols Scope\"). instance private environment. objects private invisible Keras framework Python runtime. inherit custom class subclass. default, base keras class. parent_env R environment class methods grandparent.","code":"function(y_true, y_pred)"},{"path":"https://keras3.posit.co/dev/reference/new_loss_class.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Loss — new_loss_class","text":"function returns Loss instances, similar builtin loss functions.","code":""},{"path":"https://keras3.posit.co/dev/reference/new_metric_class.html","id":null,"dir":"Reference","previous_headings":"","what":"Metric — new_metric_class","title":"Metric — new_metric_class","text":"new_metric_class() alias Metric(). See ?Metric() full documentation.","code":""},{"path":"https://keras3.posit.co/dev/reference/new_metric_class.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Metric — new_metric_class","text":"","code":"new_metric_class(   classname,   initialize = NULL,   update_state = NULL,   result = NULL,   ...,   public = list(),   private = list(),   inherit = NULL,   parent_env = parent.frame() )"},{"path":"https://keras3.posit.co/dev/reference/new_metric_class.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Metric — new_metric_class","text":"classname String, name custom class. (Conventionally, CamelCase). initialize, update_state, result Recommended methods implement. See description section. ..., public Additional methods public members custom class. private Named list R objects (typically, functions) include instance private environments. private methods symbols scope public methods (See section \"Symbols Scope\"). instance private environment. objects private invisible Keras framework Python runtime. inherit custom class subclass. default, base keras class. parent_env R environment class methods grandparent.","code":""},{"path":"https://keras3.posit.co/dev/reference/new_metric_class.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Metric — new_metric_class","text":"function returns Metric instances, similar builtin metric functions.","code":""},{"path":"https://keras3.posit.co/dev/reference/new_model_class.html","id":null,"dir":"Reference","previous_headings":"","what":"Model — new_model_class","title":"Model — new_model_class","text":"new_model_class() alias Model(). See ?Model() full documentation.","code":""},{"path":"https://keras3.posit.co/dev/reference/new_model_class.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Model — new_model_class","text":"","code":"new_model_class(   classname,   initialize = NULL,   call = NULL,   train_step = NULL,   predict_step = NULL,   test_step = NULL,   compute_loss = NULL,   compute_metrics = NULL,   ...,   public = list(),   private = list(),   inherit = NULL,   parent_env = parent.frame() )"},{"path":"https://keras3.posit.co/dev/reference/new_model_class.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Model — new_model_class","text":"classname String, name custom class. (Conventionally, CamelCase). initialize, call, train_step, predict_step, test_step, compute_loss, compute_metrics Optional methods can overridden. ..., public Additional methods public members custom class. private Named list R objects (typically, functions) include instance private environments. private methods symbols scope public methods (See section \"Symbols Scope\"). instance private environment. objects private invisible Keras framework Python runtime. inherit custom class subclass. default, base keras class. parent_env R environment class methods grandparent.","code":""},{"path":"https://keras3.posit.co/dev/reference/new_model_class.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Model — new_model_class","text":"model constructor function, can call create instance new model type.","code":""},{"path":"https://keras3.posit.co/dev/reference/newaxis.html","id":null,"dir":"Reference","previous_headings":"","what":"New axis — newaxis","title":"New axis — newaxis","text":"alias NULL. meant used [ tensors, expand dimensions tensor","code":""},{"path":"https://keras3.posit.co/dev/reference/newaxis.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"New axis — newaxis","text":"","code":"newaxis"},{"path":"https://keras3.posit.co/dev/reference/newaxis.html","id":"format","dir":"Reference","previous_headings":"","what":"Format","title":"New axis — newaxis","text":"object class NULL length 0.","code":""},{"path":"https://keras3.posit.co/dev/reference/newaxis.html","id":"details","dir":"Reference","previous_headings":"","what":"Details","title":"New axis — newaxis","text":"","code":"x <- op_convert_to_tensor(1:10)  op_shape(x) op_shape(x[]) op_shape(x[newaxis]) op_shape(x@py[newaxis]) op_shape(x@r[newaxis])  op_shape(x[newaxis, .., newaxis]) op_shape(x@py[newaxis, .., newaxis]) op_shape(x@r[newaxis, .., newaxis])"},{"path":"https://keras3.posit.co/dev/reference/normalize.html","id":null,"dir":"Reference","previous_headings":"","what":"Normalizes an array. — normalize","title":"Normalizes an array. — normalize","text":"input R array, R array returned. backend tensor, backend tensor returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/normalize.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Normalizes an array. — normalize","text":"","code":"normalize(x, axis = -1L, order = 2L)"},{"path":"https://keras3.posit.co/dev/reference/normalize.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Normalizes an array. — normalize","text":"x Array normalize. axis axis along normalize. order Normalization order (e.g. order=2 L2 norm).","code":""},{"path":"https://keras3.posit.co/dev/reference/normalize.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Normalizes an array. — normalize","text":"normalized copy array.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_abs.html","id":null,"dir":"Reference","previous_headings":"","what":"Compute the absolute value element-wise. — op_abs","title":"Compute the absolute value element-wise. — op_abs","text":"Compute absolute value element-wise.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_abs.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Compute the absolute value element-wise. — op_abs","text":"","code":"op_abs(x)"},{"path":"https://keras3.posit.co/dev/reference/op_abs.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Compute the absolute value element-wise. — op_abs","text":"x Input tensor","code":""},{"path":"https://keras3.posit.co/dev/reference/op_abs.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Compute the absolute value element-wise. — op_abs","text":"array containing absolute value element x.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_abs.html","id":"example","dir":"Reference","previous_headings":"","what":"Example","title":"Compute the absolute value element-wise. — op_abs","text":"","code":"x <- op_convert_to_tensor(c(-1.2, 1.2)) op_abs(x) ## tf.Tensor([1.2 1.2], shape=(2), dtype=float64)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_add.html","id":null,"dir":"Reference","previous_headings":"","what":"Add arguments element-wise. — op_add","title":"Add arguments element-wise. — op_add","text":"Add arguments element-wise.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_add.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Add arguments element-wise. — op_add","text":"","code":"op_add(x1, x2)"},{"path":"https://keras3.posit.co/dev/reference/op_add.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Add arguments element-wise. — op_add","text":"x1 First input tensor. x2 Second input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_add.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Add arguments element-wise. — op_add","text":"tensor containing element-wise sum x1 x2.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_add.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Add arguments element-wise. — op_add","text":"op_add also broadcasts shapes:     Note function automatically called using R operator + tensors.","code":"x1 <- op_convert_to_tensor(c(1, 4)) x2 <- op_convert_to_tensor(c(5, 6)) op_add(x1, x2) ## tf.Tensor([ 6. 10.], shape=(2), dtype=float64) # alias for x1 + x2 x1 + x2 ## tf.Tensor([ 6. 10.], shape=(2), dtype=float64) x1 <- op_convert_to_tensor(array(c(5, 5, 4, 6), dim =c(2, 2))) x2 <- op_convert_to_tensor(c(5, 6)) op_add(x1, x2) ## tf.Tensor( ## [[10. 10.] ##  [10. 12.]], shape=(2, 2), dtype=float64) x <- op_ones(c(3)) op_add(x, x) ## tf.Tensor([2. 2. 2.], shape=(3), dtype=float32) x + x ## tf.Tensor([2. 2. 2.], shape=(3), dtype=float32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_all.html","id":null,"dir":"Reference","previous_headings":"","what":"Test whether all array elements along a given axis evaluate to TRUE. — op_all","title":"Test whether all array elements along a given axis evaluate to TRUE. — op_all","text":"Test whether array elements along given axis evaluate TRUE.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_all.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Test whether all array elements along a given axis evaluate to TRUE. — op_all","text":"","code":"op_all(x, axis = NULL, keepdims = FALSE)"},{"path":"https://keras3.posit.co/dev/reference/op_all.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Test whether all array elements along a given axis evaluate to TRUE. — op_all","text":"x Input tensor. axis integer tuple integers represent axis along logical reduction performed. default (axis = NULL) perform logical dimensions input array. axis may negative, case counts last first axis. keepdims TRUE, axes reduced left result dimensions size one. option, result broadcast correctly input array. Defaults FALSE.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_all.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Test whether all array elements along a given axis evaluate to TRUE. — op_all","text":"tensor containing logical reduction axis.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_all.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Test whether all array elements along a given axis evaluate to TRUE. — op_all","text":"keepdims = TRUE outputs tensor dimensions reduced one.","code":"x <- op_convert_to_tensor(c(TRUE, FALSE)) op_all(x) ## tf.Tensor(False, shape=(), dtype=bool) (x <- op_convert_to_tensor(array(c(TRUE, FALSE, TRUE, TRUE, TRUE, TRUE), dim = c(3, 2)))) ## tf.Tensor( ## [[ True  True] ##  [False  True] ##  [ True  True]], shape=(3, 2), dtype=bool) op_all(x, axis = 1) ## tf.Tensor([False  True], shape=(2), dtype=bool) op_all(x, keepdims = TRUE) ## tf.Tensor([[False]], shape=(1, 1), dtype=bool)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_any.html","id":null,"dir":"Reference","previous_headings":"","what":"Test whether any array element along a given axis evaluates to TRUE. — op_any","title":"Test whether any array element along a given axis evaluates to TRUE. — op_any","text":"Test whether array element along given axis evaluates TRUE.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_any.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Test whether any array element along a given axis evaluates to TRUE. — op_any","text":"","code":"op_any(x, axis = NULL, keepdims = FALSE)"},{"path":"https://keras3.posit.co/dev/reference/op_any.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Test whether any array element along a given axis evaluates to TRUE. — op_any","text":"x Input tensor. axis integer tuple integers represent axis along logical reduction performed. default (axis = NULL) perform logical dimensions input array. axis may negative, case counts last first axis. keepdims TRUE, axes reduced left result dimensions size one. option, result broadcast correctly input array. Defaults FALSE.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_any.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Test whether any array element along a given axis evaluates to TRUE. — op_any","text":"tensor containing logical reduction axis.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_any.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Test whether any array element along a given axis evaluates to TRUE. — op_any","text":"keepdims = TRUE outputs tensor dimensions reduced one.","code":"x <- op_array(c(TRUE, FALSE)) op_any(x) ## tf.Tensor(True, shape=(), dtype=bool) (x <- op_reshape(c(FALSE, FALSE, FALSE,                    TRUE, FALSE, FALSE),                  c(2, 3))) ## tf.Tensor( ## [[False False False] ##  [ True False False]], shape=(2, 3), dtype=bool) op_any(x, axis = 1) ## tf.Tensor([ True False False], shape=(3), dtype=bool) op_any(x, axis = 2) ## tf.Tensor([False  True], shape=(2), dtype=bool) op_any(x, axis = -1) ## tf.Tensor([False  True], shape=(2), dtype=bool) op_any(x, keepdims = TRUE) ## tf.Tensor([[ True]], shape=(1, 1), dtype=bool) op_any(x, 1, keepdims = TRUE) ## tf.Tensor([[ True False False]], shape=(1, 3), dtype=bool)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_append.html","id":null,"dir":"Reference","previous_headings":"","what":"Append tensor x2 to the end of tensor x1. — op_append","title":"Append tensor x2 to the end of tensor x1. — op_append","text":"Append tensor x2 end tensor x1.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_append.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Append tensor x2 to the end of tensor x1. — op_append","text":"","code":"op_append(x1, x2, axis = NULL)"},{"path":"https://keras3.posit.co/dev/reference/op_append.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Append tensor x2 to the end of tensor x1. — op_append","text":"x1 First input tensor. x2 Second input tensor. axis Axis along tensor x2 appended tensor x1. NULL, tensors flattened use.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_append.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Append tensor x2 to the end of tensor x1. — op_append","text":"tensor values x2 appended x1.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_append.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Append tensor x2 to the end of tensor x1. — op_append","text":"axis specified, x1 x2 must compatible shapes.","code":"x1 <- op_convert_to_tensor(c(1, 2, 3)) x2 <- op_convert_to_tensor(rbind(c(4, 5, 6), c(7, 8, 9))) op_append(x1, x2) ## tf.Tensor([1. 2. 3. 4. 5. 6. 7. 8. 9.], shape=(9), dtype=float64) x1 <- op_convert_to_tensor(rbind(c(1, 2, 3), c(4, 5, 6))) x2 <- op_convert_to_tensor(rbind(c(7, 8, 9))) op_append(x1, x2, axis = 1) ## tf.Tensor( ## [[1. 2. 3.] ##  [4. 5. 6.] ##  [7. 8. 9.]], shape=(3, 3), dtype=float64) x3 <- op_convert_to_tensor(c(7, 8, 9)) try(op_append(x1, x3, axis = 1)) ## Error in py_call_impl(callable, call_args$unnamed, call_args$named) : ##   tensorflow.python.framework.errors_impl.InvalidArgumentError: {{function_node __wrapped__ConcatV2_N_2_device_/job:localhost/replica:0/task:0/device:CPU:0}} ConcatOp : Ranks of all input tensors should match: shape[0] = [2,3] vs. shape[1] = [3] [Op:ConcatV2] name: concat"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_arange.html","id":null,"dir":"Reference","previous_headings":"","what":"Return evenly spaced values within a given interval. — op_arange","title":"Return evenly spaced values within a given interval. — op_arange","text":"arange can called varying number positional arguments: arange(end): Values generated within half-open interval [1, end] (words, interval including start end). arange(start, end): Values generated within interval [start, end]. arange(start, end, step): Values generated within interval [start, end], spacing values given step.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_arange.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Return evenly spaced values within a given interval. — op_arange","text":"","code":"op_arange(start, end, step = 1L, dtype = NULL, include_end = TRUE)"},{"path":"https://keras3.posit.co/dev/reference/op_arange.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Return evenly spaced values within a given interval. — op_arange","text":"start Integer real, representing start interval. interval includes value. end Integer real, representing end interval. include_end FALSE, interval include value, except cases step integer floating point round-affects length . Defaults NULL. step Integer real, represent spacing values. output , distance two adjacent values, [+1] - []. default step size 1. step specified position argument, start must also given. dtype type output array. dtype given, infer data type input arguments. include_end TRUE FALSE. FALSE, end included output sequence.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_arange.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Return evenly spaced values within a given interval. — op_arange","text":"Tensor evenly spaced values. floating point arguments, length result ceiling((end - start)/step). floating point overflow, rule may result last element greater end.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_arange.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Return evenly spaced values within a given interval. — op_arange","text":"","code":"op_arange(3L) ## tf.Tensor([1 2 3], shape=(3), dtype=int32) op_arange(3) # float ## tf.Tensor([1. 2. 3.], shape=(3), dtype=float32) op_arange(3, dtype = 'int32') #int ## tf.Tensor([1 2 3], shape=(3), dtype=int32) op_arange(3L, 7L) ## tf.Tensor([3 4 5 6 7], shape=(5), dtype=int32) op_arange(3L, 7L, 2L) ## tf.Tensor([3 5 7], shape=(3), dtype=int32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_arccos.html","id":null,"dir":"Reference","previous_headings":"","what":"Trigonometric inverse cosine, element-wise. — op_arccos","title":"Trigonometric inverse cosine, element-wise. — op_arccos","text":"inverse cos , y = cos(x), x = arccos(y).","code":""},{"path":"https://keras3.posit.co/dev/reference/op_arccos.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Trigonometric inverse cosine, element-wise. — op_arccos","text":"","code":"op_arccos(x)"},{"path":"https://keras3.posit.co/dev/reference/op_arccos.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Trigonometric inverse cosine, element-wise. — op_arccos","text":"x Input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_arccos.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Trigonometric inverse cosine, element-wise. — op_arccos","text":"Tensor angle ray intersecting unit circle given x-coordinate radians [0, pi].","code":""},{"path":"https://keras3.posit.co/dev/reference/op_arccos.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Trigonometric inverse cosine, element-wise. — op_arccos","text":"","code":"x <- op_convert_to_tensor(c(1, -1)) op_arccos(x) ## tf.Tensor([0.         3.14159265], shape=(2), dtype=float64)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_arccosh.html","id":null,"dir":"Reference","previous_headings":"","what":"Inverse hyperbolic cosine, element-wise. — op_arccosh","title":"Inverse hyperbolic cosine, element-wise. — op_arccosh","text":"Inverse hyperbolic cosine, element-wise.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_arccosh.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Inverse hyperbolic cosine, element-wise. — op_arccosh","text":"","code":"op_arccosh(x)"},{"path":"https://keras3.posit.co/dev/reference/op_arccosh.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Inverse hyperbolic cosine, element-wise. — op_arccosh","text":"x Input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_arccosh.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Inverse hyperbolic cosine, element-wise. — op_arccosh","text":"Output tensor shape x.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_arccosh.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Inverse hyperbolic cosine, element-wise. — op_arccosh","text":"","code":"x <- op_convert_to_tensor(c(10, 100)) op_arccosh(x) ## tf.Tensor([2.99322285 5.29829237], shape=(2), dtype=float64)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_arcsin.html","id":null,"dir":"Reference","previous_headings":"","what":"Inverse sine, element-wise. — op_arcsin","title":"Inverse sine, element-wise. — op_arcsin","text":"Inverse sine, element-wise.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_arcsin.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Inverse sine, element-wise. — op_arcsin","text":"","code":"op_arcsin(x)"},{"path":"https://keras3.posit.co/dev/reference/op_arcsin.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Inverse sine, element-wise. — op_arcsin","text":"x Input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_arcsin.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Inverse sine, element-wise. — op_arcsin","text":"Tensor inverse sine element x, radians closed interval [-pi/2, pi/2].","code":""},{"path":"https://keras3.posit.co/dev/reference/op_arcsin.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Inverse sine, element-wise. — op_arcsin","text":"","code":"x <- op_convert_to_tensor(c(1, -1, 0)) op_arcsin(x) ## tf.Tensor([ 1.57079633 -1.57079633  0.        ], shape=(3), dtype=float64)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_arcsinh.html","id":null,"dir":"Reference","previous_headings":"","what":"Inverse hyperbolic sine, element-wise. — op_arcsinh","title":"Inverse hyperbolic sine, element-wise. — op_arcsinh","text":"Inverse hyperbolic sine, element-wise.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_arcsinh.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Inverse hyperbolic sine, element-wise. — op_arcsinh","text":"","code":"op_arcsinh(x)"},{"path":"https://keras3.posit.co/dev/reference/op_arcsinh.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Inverse hyperbolic sine, element-wise. — op_arcsinh","text":"x Input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_arcsinh.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Inverse hyperbolic sine, element-wise. — op_arcsinh","text":"Output tensor shape x.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_arcsinh.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Inverse hyperbolic sine, element-wise. — op_arcsinh","text":"","code":"x <- op_convert_to_tensor(c(1, -1, 0)) op_arcsinh(x) ## tf.Tensor([ 0.88137359 -0.88137359  0.        ], shape=(3), dtype=float64)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_arctan.html","id":null,"dir":"Reference","previous_headings":"","what":"Trigonometric inverse tangent, element-wise. — op_arctan","title":"Trigonometric inverse tangent, element-wise. — op_arctan","text":"Trigonometric inverse tangent, element-wise.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_arctan.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Trigonometric inverse tangent, element-wise. — op_arctan","text":"","code":"op_arctan(x)"},{"path":"https://keras3.posit.co/dev/reference/op_arctan.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Trigonometric inverse tangent, element-wise. — op_arctan","text":"x Input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_arctan.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Trigonometric inverse tangent, element-wise. — op_arctan","text":"Tensor inverse tangent element x, interval [-pi/2, pi/2].","code":""},{"path":"https://keras3.posit.co/dev/reference/op_arctan.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Trigonometric inverse tangent, element-wise. — op_arctan","text":"","code":"x <- op_convert_to_tensor(c(0, 1)) op_arctan(x) ## tf.Tensor([0.         0.78539816], shape=(2), dtype=float64)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_arctan2.html","id":null,"dir":"Reference","previous_headings":"","what":"Element-wise arc tangent of x1/x2 choosing the quadrant correctly. — op_arctan2","title":"Element-wise arc tangent of x1/x2 choosing the quadrant correctly. — op_arctan2","text":"quadrant (.e., branch) chosen arctan2(x1, x2) signed angle radians ray ending origin passing point (1, 0), ray ending origin passing point (x2, x1). (Note role reversal: \"y-coordinate\" first function parameter, \"x-coordinate\" second.) IEEE convention, function defined x2 = +/-0 either x1 x2 = +/-inf.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_arctan2.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Element-wise arc tangent of x1/x2 choosing the quadrant correctly. — op_arctan2","text":"","code":"op_arctan2(x1, x2)"},{"path":"https://keras3.posit.co/dev/reference/op_arctan2.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Element-wise arc tangent of x1/x2 choosing the quadrant correctly. — op_arctan2","text":"x1 First input tensor. x2 Second input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_arctan2.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Element-wise arc tangent of x1/x2 choosing the quadrant correctly. — op_arctan2","text":"Tensor angles radians, range [-pi, pi].","code":""},{"path":"https://keras3.posit.co/dev/reference/op_arctan2.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Element-wise arc tangent of x1/x2 choosing the quadrant correctly. — op_arctan2","text":"Consider four points different quadrants:     Note order parameters. arctan2 defined also x2 = 0 several points, obtaining values range [-pi, pi]:","code":"x <- op_array(c(-1, 1, 1, -1)) y <- op_array(c(-1, -1, 1, 1)) op_arctan2(y, x) * 180 / pi ## tf.Tensor([-135.  -45.   45.  135.], shape=(4), dtype=float32) op_arctan2(     op_array(c(1, -1)),     op_array(c(0, 0)) ) ## tf.Tensor([ 1.5707964 -1.5707964], shape=(2), dtype=float32) op_arctan2(     op_array(c(0, 0, Inf)),     op_array(c(+0, -0, Inf)) ) ## tf.Tensor([0.        3.1415927 0.7853982], shape=(3), dtype=float32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_arctanh.html","id":null,"dir":"Reference","previous_headings":"","what":"Inverse hyperbolic tangent, element-wise. — op_arctanh","title":"Inverse hyperbolic tangent, element-wise. — op_arctanh","text":"Inverse hyperbolic tangent, element-wise.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_arctanh.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Inverse hyperbolic tangent, element-wise. — op_arctanh","text":"","code":"op_arctanh(x)"},{"path":"https://keras3.posit.co/dev/reference/op_arctanh.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Inverse hyperbolic tangent, element-wise. — op_arctanh","text":"x Input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_arctanh.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Inverse hyperbolic tangent, element-wise. — op_arctanh","text":"Output tensor shape x.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_argmax.html","id":null,"dir":"Reference","previous_headings":"","what":"Returns the indices of the maximum values along an axis. — op_argmax","title":"Returns the indices of the maximum values along an axis. — op_argmax","text":"Returns indices maximum values along axis.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_argmax.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Returns the indices of the maximum values along an axis. — op_argmax","text":"","code":"op_argmax(x, axis = NULL, keepdims = FALSE, zero_indexed = FALSE)"},{"path":"https://keras3.posit.co/dev/reference/op_argmax.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Returns the indices of the maximum values along an axis. — op_argmax","text":"x Input tensor. axis default, index flattened tensor, otherwise along specified axis. keepdims set TRUE, axes reduced left result dimensions size one. Defaults FALSE. zero_indexed TRUE, returned indices zero-based (0 encodes first position); FALSE (default), returned indices one-based (1 encodes first position).","code":""},{"path":"https://keras3.posit.co/dev/reference/op_argmax.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Returns the indices of the maximum values along an axis. — op_argmax","text":"Tensor indices. shape x, dimension along axis removed. Note returned integer 1-based (.e., argmax first index position, returned value 1)","code":""},{"path":"https://keras3.posit.co/dev/reference/op_argmax.html","id":"note","dir":"Reference","previous_headings":"","what":"Note","title":"Returns the indices of the maximum values along an axis. — op_argmax","text":"similar R max.col(x) case 2-d array (matrix), nd-array, apply(x, axis, .max)","code":""},{"path":"https://keras3.posit.co/dev/reference/op_argmax.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Returns the indices of the maximum values along an axis. — op_argmax","text":"","code":"x <- op_array(rbind(c(11, 12, 10),                     c(12, 11, 10))) x ## tf.Tensor( ## [[11. 12. 10.] ##  [12. 11. 10.]], shape=(2, 3), dtype=float64) op_argmax(x) ## tf.Tensor(2, shape=(), dtype=int32) op_argmax(x, axis = 1) ## tf.Tensor([2 1 1], shape=(3), dtype=int32) op_argmax(x, axis = 2) ## tf.Tensor([2 1], shape=(2), dtype=int32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_argmin.html","id":null,"dir":"Reference","previous_headings":"","what":"Returns the indices of the minimum values along an axis. — op_argmin","title":"Returns the indices of the minimum values along an axis. — op_argmin","text":"Returns indices minimum values along axis.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_argmin.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Returns the indices of the minimum values along an axis. — op_argmin","text":"","code":"op_argmin(x, axis = NULL, keepdims = FALSE, zero_indexed = FALSE)"},{"path":"https://keras3.posit.co/dev/reference/op_argmin.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Returns the indices of the minimum values along an axis. — op_argmin","text":"x Input tensor. axis default, index flattened tensor, otherwise along specified axis. keepdims set TRUE, axes reduced left result dimensions size one. Defaults FALSE. zero_indexed TRUE, returned indices zero-based (0 encodes first position); FALSE (default), returned indices one-based (1 encodes first position).","code":""},{"path":"https://keras3.posit.co/dev/reference/op_argmin.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Returns the indices of the minimum values along an axis. — op_argmin","text":"Tensor indices. shape x, dimension along axis removed.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_argmin.html","id":"note","dir":"Reference","previous_headings":"","what":"Note","title":"Returns the indices of the minimum values along an axis. — op_argmin","text":"similar R expression apply(x, axis, .min), x R array.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_argmin.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Returns the indices of the minimum values along an axis. — op_argmin","text":"","code":"x <- op_arange(6L) |> op_reshape(c(2, 3)) |> op_add(10) x ## tf.Tensor( ## [[11. 12. 13.] ##  [14. 15. 16.]], shape=(2, 3), dtype=float32) op_argmin(x) ## tf.Tensor(1, shape=(), dtype=int32) op_argmin(x, axis = 1) ## tf.Tensor([1 1 1], shape=(3), dtype=int32) op_argmin(x, axis = 2) ## tf.Tensor([1 1], shape=(2), dtype=int32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_argpartition.html","id":null,"dir":"Reference","previous_headings":"","what":"Performs an indirect partition along the given axis. — op_argpartition","title":"Performs an indirect partition along the given axis. — op_argpartition","text":"returns array indices shape x index data along given axis partitioned order.","code":"x <- op_convert_to_tensor(c(9, 3, 6, 2, 8, 5, 7, 1, 10, 4)) x@r[op_argpartition(x, 3)] ## tf.Tensor([ 1.  2.  3. 10.  9.  8.  7.  6.  5.  4.], shape=(10), dtype=float64) x@r[op_argpartition(x, 5)] ## tf.Tensor([ 1.  2.  3.  4.  5. 10.  9.  8.  7.  6.], shape=(10), dtype=float64) x@r[op_argpartition(x, 7)] ## tf.Tensor([ 1.  2.  3.  4.  5.  6.  7. 10.  9.  8.], shape=(10), dtype=float64)"},{"path":"https://keras3.posit.co/dev/reference/op_argpartition.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Performs an indirect partition along the given axis. — op_argpartition","text":"","code":"op_argpartition(x, kth, axis = -1L, zero_indexed = FALSE)"},{"path":"https://keras3.posit.co/dev/reference/op_argpartition.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Performs an indirect partition along the given axis. — op_argpartition","text":"x Array sort. kth Element index partition . k-th element final sorted position smaller elements moved larger elements behind . order elements partitions undefined. provided sequence k-th partition sorted position . axis Axis along sort. default -1 (last axis). NULL, flattened array used. zero_indexed TRUE, returned indices zero-based (0 encodes first position); FALSE (default), returned indices one-based (1 encodes first position).","code":""},{"path":"https://keras3.posit.co/dev/reference/op_argpartition.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Performs an indirect partition along the given axis. — op_argpartition","text":"Array indices partition x along specified axis.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_argsort.html","id":null,"dir":"Reference","previous_headings":"","what":"Returns the indices that would sort a tensor. — op_argsort","title":"Returns the indices that would sort a tensor. — op_argsort","text":"Returns indices sort tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_argsort.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Returns the indices that would sort a tensor. — op_argsort","text":"","code":"op_argsort(x, axis = -1L, zero_indexed = FALSE)"},{"path":"https://keras3.posit.co/dev/reference/op_argsort.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Returns the indices that would sort a tensor. — op_argsort","text":"x Input tensor. axis Axis along sort. Defaults -1 (last axis). NULL, flattened tensor used. zero_indexed TRUE, returned indices zero-based (0 encodes first position); FALSE (default), returned indices one-based (1 encodes first position).","code":""},{"path":"https://keras3.posit.co/dev/reference/op_argsort.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Returns the indices that would sort a tensor. — op_argsort","text":"Tensor indices sort x along specified axis.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_argsort.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Returns the indices that would sort a tensor. — op_argsort","text":"One dimensional array:         Two-dimensional array:","code":"x <- op_array(c(3, 1, 2)) + .1 op_argsort(x) ## tf.Tensor([2 3 1], shape=(3), dtype=int32) x@r[op_argsort(x)] == op_sort(x) ## tf.Tensor([ True  True  True], shape=(3), dtype=bool) x <- op_array(rbind(c(0, 3),                     c(3, 2),                     c(4, 5))) + .1 (i <- op_argsort(x, axis = 1)) ## tf.Tensor( ## [[1 2] ##  [2 1] ##  [3 3]], shape=(3, 2), dtype=int32) x@r[i@r[, 1], ] # sort x-rows using first col of x ## tf.Tensor( ## [[0.1 3.1] ##  [3.1 2.1] ##  [4.1 5.1]], shape=(3, 2), dtype=float64) x@r[i@r[, 2], ] # sort x-rows using second col of x ## tf.Tensor( ## [[3.1 2.1] ##  [0.1 3.1] ##  [4.1 5.1]], shape=(3, 2), dtype=float64) (i <- op_argsort(x, axis = 2)) ## tf.Tensor( ## [[1 2] ##  [2 1] ##  [1 2]], shape=(3, 2), dtype=int32) x@r[, i@r[2,]] # sort x-cols using second row of x ## tf.Tensor( ## [[3.1 0.1] ##  [2.1 3.1] ##  [5.1 4.1]], shape=(3, 2), dtype=float64)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_array.html","id":null,"dir":"Reference","previous_headings":"","what":"Create a tensor. — op_array","title":"Create a tensor. — op_array","text":"Create tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_array.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Create a tensor. — op_array","text":"","code":"op_array(x, dtype = NULL)"},{"path":"https://keras3.posit.co/dev/reference/op_array.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Create a tensor. — op_array","text":"x Input tensor. dtype desired data-type tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_array.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Create a tensor. — op_array","text":"tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_array.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Create a tensor. — op_array","text":"","code":"op_array(c(1, 2, 3)) ## tf.Tensor([1. 2. 3.], shape=(3), dtype=float32) op_array(c(1, 2, 3), dtype = \"float32\") ## tf.Tensor([1. 2. 3.], shape=(3), dtype=float32) op_array(c(1, 2, 3), dtype = \"int32\") ## tf.Tensor([1 2 3], shape=(3), dtype=int32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_associative_scan.html","id":null,"dir":"Reference","previous_headings":"","what":"Performs a scan with an associative binary operation, in parallel. — op_associative_scan","title":"Performs a scan with an associative binary operation, in parallel. — op_associative_scan","text":"operation similar op_scan(), key difference op_associative_scan() parallel implementation potentially significant performance benefits, especially jit compiled. catch can used f binary associative operation (.e. must verify f(, f(b, c)) == f(f(, b), c)). introduction associative scans, refer paper: Blelloch, Guy E. 1990. Prefix Sums Applications.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_associative_scan.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Performs a scan with an associative binary operation, in parallel. — op_associative_scan","text":"","code":"op_associative_scan(f, elems, reverse = FALSE, axis = 1L)"},{"path":"https://keras3.posit.co/dev/reference/op_associative_scan.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Performs a scan with an associative binary operation, in parallel. — op_associative_scan","text":"f callable implementing associative binary operation signature  r = f(, b). Function f must associative, .e., must satisfy equation f(, f(b, c)) == f(f(, b), c). inputs result (possibly nested tree structures ) array(s) matching elems. array dimension place axis dimension. f applied elementwise axis dimension. result r shape (structure) two inputs b. elems (possibly nested tree structure ) array(s), axis dimension size num_elems. reverse boolean stating scan reversed respect axis dimension. axis integer identifying axis scan occur.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_associative_scan.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Performs a scan with an associative binary operation, in parallel. — op_associative_scan","text":"(possibly nested tree structure ) array(s) shape structure elems, k'th element axis result recursively applying f combine first k elements elems along axis. example, given elems = list(, b, c, ...), result list(, f(, b), f(f(, b), c), ...).","code":""},{"path":"https://keras3.posit.co/dev/reference/op_associative_scan.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Performs a scan with an associative binary operation, in parallel. — op_associative_scan","text":"","code":"sum_fn <- function(x, y) x + y xs <- op_arange(5L) op_associative_scan(sum_fn, xs) ## tf.Tensor([ 1  3  6 10 15], shape=(5), dtype=int32) sum_fn <- function(x, y) {   str(list(x = x, y = y))   map2(x, y, \\(.x, .y) .x + .y) }  xs <- list(op_array(1:2),            op_array(1:2),            op_array(1:2)) ys <- op_associative_scan(sum_fn, xs, axis = 1) ## List of 2 ##  $ x:List of 3 ##   ..$ :<tf.Tensor: shape=(1), dtype=int32, numpy=array([1], dtype=int32)> ##   ..$ :<tf.Tensor: shape=(1), dtype=int32, numpy=array([1], dtype=int32)> ##   ..$ :<tf.Tensor: shape=(1), dtype=int32, numpy=array([1], dtype=int32)> ##  $ y:List of 3 ##   ..$ :<tf.Tensor: shape=(1), dtype=int32, numpy=array([2], dtype=int32)> ##   ..$ :<tf.Tensor: shape=(1), dtype=int32, numpy=array([2], dtype=int32)> ##   ..$ :<tf.Tensor: shape=(1), dtype=int32, numpy=array([2], dtype=int32)> ys ## [[1]] ## tf.Tensor([1 3], shape=(2), dtype=int32) ## ## [[2]] ## tf.Tensor([1 3], shape=(2), dtype=int32) ## ## [[3]] ## tf.Tensor([1 3], shape=(2), dtype=int32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_average.html","id":null,"dir":"Reference","previous_headings":"","what":"Compute the weighted average along the specified axis. — op_average","title":"Compute the weighted average along the specified axis. — op_average","text":"Compute weighted average along specified axis.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_average.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Compute the weighted average along the specified axis. — op_average","text":"","code":"op_average(x, axis = NULL, weights = NULL)"},{"path":"https://keras3.posit.co/dev/reference/op_average.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Compute the weighted average along the specified axis. — op_average","text":"x Input tensor. axis Integer along average x. default, axis = NULL, average elements input tensor. axis negative counts last first axis. weights Tensor weights associated values x. value x contributes average according associated weight. weights array can either 1-D (case length must size along given axis) shape x. weights = NULL (default), data x assumed weight equal one. 1-D calculation : avg = sum(* weights) / sum(weights). constraint weights sum(weights) must 0.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_average.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Compute the weighted average along the specified axis. — op_average","text":"Return average along specified axis.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_average.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Compute the weighted average along the specified axis. — op_average","text":"","code":"data <- op_arange(1, 5, dtype = \"int32\") data ## tf.Tensor([1 2 3 4 5], shape=(5), dtype=int32) op_average(data) ## tf.Tensor(3.0, shape=(), dtype=float32) op_average(   op_arange(1, 11),   weights = op_arange(10, 0, -1) ) ## tf.Tensor(4.0, shape=(), dtype=float32) data <- op_arange(6) |> op_reshape(c(3, 2)) data ## tf.Tensor( ## [[1. 2.] ##  [3. 4.] ##  [5. 6.]], shape=(3, 2), dtype=float32) op_average(   data,   axis = 2,   weights = op_array(c(1/4, 3/4)) ) ## tf.Tensor([1.75 3.75 5.75], shape=(3), dtype=float32) # Error: Axis must be specified when shapes of x and weights differ. try(op_average(   data,   weights = op_array(c(1/4, 3/4)) )) ## Error in op_average(data, weights = op_array(c(1/4, 3/4))) : ##   Axis must be specified when shapes of x and weights differ."},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_average_pool.html","id":null,"dir":"Reference","previous_headings":"","what":"Average pooling operation. — op_average_pool","title":"Average pooling operation. — op_average_pool","text":"Average pooling operation.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_average_pool.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Average pooling operation. — op_average_pool","text":"","code":"op_average_pool(   inputs,   pool_size,   strides = NULL,   padding = \"valid\",   data_format = NULL )"},{"path":"https://keras3.posit.co/dev/reference/op_average_pool.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Average pooling operation. — op_average_pool","text":"inputs Tensor rank N+2. inputs shape (batch_size,) + inputs_spatial_shape + (num_channels,) data_format = \"channels_last\", (batch_size, num_channels) + inputs_spatial_shape data_format = \"channels_first\". Pooling happens spatial dimensions . pool_size int tuple/list integers size len(inputs_spatial_shape), specifying size pooling window spatial dimension input tensor. pool_size int, every spatial dimension shares pool_size. strides int tuple/list integers size len(inputs_spatial_shape). stride sliding window spatial dimension input tensor. strides int, every spatial dimension shares strides. padding string, either \"valid\" \"\". \"valid\" means padding applied, \"\" results padding evenly left/right /input output height/width dimension input strides = 1. data_format string, either \"channels_last\" \"channels_first\". data_format determines ordering dimensions inputs. data_format = \"channels_last\", inputs shape (batch_size, ..., channels) data_format = \"channels_first\", inputs shape (batch_size, channels, ...).","code":""},{"path":"https://keras3.posit.co/dev/reference/op_average_pool.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Average pooling operation. — op_average_pool","text":"tensor rank N+2, result average pooling operation.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_batch_normalization.html","id":null,"dir":"Reference","previous_headings":"","what":"Normalizes x by mean and variance. — op_batch_normalization","title":"Normalizes x by mean and variance. — op_batch_normalization","text":"op typically used batch normalization step neural network. normalizes input tensor along given axis.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_batch_normalization.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Normalizes x by mean and variance. — op_batch_normalization","text":"","code":"op_batch_normalization(   x,   mean,   variance,   axis,   offset = NULL,   scale = NULL,   epsilon = 0.001 )"},{"path":"https://keras3.posit.co/dev/reference/op_batch_normalization.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Normalizes x by mean and variance. — op_batch_normalization","text":"x Input tensor. mean mean vector length axis dimension input thensor. variance variance vector length axis dimension input tensor. axis Integer, axis normalized. offset offset vector length axis dimension input tensor. NULL, offset added normalized tensor. Defaults NULL. scale scale vector length axis dimension input tensor. NULL, normalized tensor multiplied scale. Defaults NULL. epsilon Small float added variance avoid dividing zero. Defaults 1e-3.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_batch_normalization.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Normalizes x by mean and variance. — op_batch_normalization","text":"normalized tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_batch_normalization.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Normalizes x by mean and variance. — op_batch_normalization","text":"","code":"x <- op_convert_to_tensor(rbind(c(0.1, 0.2, 0.3),                                 c(0.4, 0.5, 0.6),                                 c(0.7, 0.8, 0.9))) op_batch_normalization(   x,   mean = c(0.4, 0.5, 0.6),   variance = c(0.67, 0.67, 0.67),   axis = -1 ) ## tf.Tensor( ## [[-0.36623513 -0.36623513 -0.36623513] ##  [ 0.          0.          0.        ] ##  [ 0.36623513  0.36623513  0.36623513]], shape=(3, 3), dtype=float64)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_binary_crossentropy.html","id":null,"dir":"Reference","previous_headings":"","what":"Computes binary cross-entropy loss between target and output tensor. — op_binary_crossentropy","title":"Computes binary cross-entropy loss between target and output tensor. — op_binary_crossentropy","text":"binary cross-entropy loss commonly used binary classification tasks input sample belongs one two classes. measures dissimilarity target output probabilities logits.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_binary_crossentropy.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes binary cross-entropy loss between target and output tensor. — op_binary_crossentropy","text":"","code":"op_binary_crossentropy(target, output, from_logits = FALSE)"},{"path":"https://keras3.posit.co/dev/reference/op_binary_crossentropy.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Computes binary cross-entropy loss between target and output tensor. — op_binary_crossentropy","text":"target target tensor representing true binary labels. shape match shape output tensor. output output tensor representing predicted probabilities logits. shape match shape target tensor. from_logits (optional) Whether output tensor logits probabilities. Set TRUE output represents logits; otherwise, set FALSE output represents probabilities. Defaults FALSE.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_binary_crossentropy.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Computes binary cross-entropy loss between target and output tensor. — op_binary_crossentropy","text":"Integer tensor: computed binary cross-entropy loss target output.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_binary_crossentropy.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Computes binary cross-entropy loss between target and output tensor. — op_binary_crossentropy","text":"","code":"target <- op_array(c(0, 1, 1, 0)) output <- op_array(c(0.1, 0.9, 0.8, 0.2)) op_binary_crossentropy(target, output) ## tf.Tensor([0.10536055 0.10536055 0.22314353 0.22314353], shape=(4), dtype=float32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_bincount.html","id":null,"dir":"Reference","previous_headings":"","what":"Count the number of occurrences of each value in a tensor of integers. — op_bincount","title":"Count the number of occurrences of each value in a tensor of integers. — op_bincount","text":"Count number occurrences value tensor integers.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_bincount.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Count the number of occurrences of each value in a tensor of integers. — op_bincount","text":"","code":"op_bincount(x, weights = NULL, minlength = 0L, sparse = FALSE)"},{"path":"https://keras3.posit.co/dev/reference/op_bincount.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Count the number of occurrences of each value in a tensor of integers. — op_bincount","text":"x Input tensor. must dimension 1, must contain non-negative integer(s). weights Weight tensor. must length x. default value NULL. specified, x weighted , .e. n = x[], [n] += weight[] instead default behavior [n] += 1. minlength integer. default value 0. specified, least number bins output tensor. greater max(x) + 1, value output index higher max(x) set 0. sparse Whether return sparse tensor; backends support sparse tensors.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_bincount.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Count the number of occurrences of each value in a tensor of integers. — op_bincount","text":"1D tensor element gives number occurrence(s) index value x. length maximum max(x) + 1 minlength.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_bincount.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Count the number of occurrences of each value in a tensor of integers. — op_bincount","text":"","code":"(x <- op_array(c(1, 2, 2, 3), dtype = \"uint8\")) ## tf.Tensor([1 2 2 3], shape=(4), dtype=uint8) op_bincount(x) ## tf.Tensor([0 1 2 1], shape=(4), dtype=int32) (weights <- x / 2) ## tf.Tensor([0.5 1.  1.  1.5], shape=(4), dtype=float32) op_bincount(x, weights = weights) ## tf.Tensor([0.  0.5 2.  1.5], shape=(4), dtype=float32) minlength <- as.integer(op_max(x) + 1 + 2) # 6 op_bincount(x, minlength = minlength) ## tf.Tensor([0 1 2 1 0 0], shape=(6), dtype=int32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_bitwise_and.html","id":null,"dir":"Reference","previous_headings":"","what":"Compute the bit-wise AND of two arrays element-wise. — op_bitwise_and","title":"Compute the bit-wise AND of two arrays element-wise. — op_bitwise_and","text":"Computes bit-wise underlying binary representation integers input arrays. ufunc implements C/Python operator &.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_bitwise_and.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Compute the bit-wise AND of two arrays element-wise. — op_bitwise_and","text":"","code":"op_bitwise_and(x, y)"},{"path":"https://keras3.posit.co/dev/reference/op_bitwise_and.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Compute the bit-wise AND of two arrays element-wise. — op_bitwise_and","text":"x Input integer tensor. y Input integer tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_bitwise_and.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Compute the bit-wise AND of two arrays element-wise. — op_bitwise_and","text":"Result tensor.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_bitwise_invert.html","id":null,"dir":"Reference","previous_headings":"","what":"Compute bit-wise inversion, or bit-wise NOT, element-wise. — op_bitwise_invert","title":"Compute bit-wise inversion, or bit-wise NOT, element-wise. — op_bitwise_invert","text":"Computes bit-wise underlying binary representation integers input arrays. ufunc implements C/Python operator ~.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_bitwise_invert.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Compute bit-wise inversion, or bit-wise NOT, element-wise. — op_bitwise_invert","text":"","code":"op_bitwise_invert(x)"},{"path":"https://keras3.posit.co/dev/reference/op_bitwise_invert.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Compute bit-wise inversion, or bit-wise NOT, element-wise. — op_bitwise_invert","text":"x Input integer tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_bitwise_invert.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Compute bit-wise inversion, or bit-wise NOT, element-wise. — op_bitwise_invert","text":"Result tensor.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_bitwise_left_shift.html","id":null,"dir":"Reference","previous_headings":"","what":"Shift the bits of an integer to the left. — op_bitwise_left_shift","title":"Shift the bits of an integer to the left. — op_bitwise_left_shift","text":"Bits shifted left appending y 0s right x. Since internal representation numbers binary format, operation equivalent multiplying x 2**y.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_bitwise_left_shift.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Shift the bits of an integer to the left. — op_bitwise_left_shift","text":"","code":"op_bitwise_left_shift(x, y)"},{"path":"https://keras3.posit.co/dev/reference/op_bitwise_left_shift.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Shift the bits of an integer to the left. — op_bitwise_left_shift","text":"x Input integer tensor. y Input integer tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_bitwise_left_shift.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Shift the bits of an integer to the left. — op_bitwise_left_shift","text":"Result tensor.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_bitwise_not.html","id":null,"dir":"Reference","previous_headings":"","what":"Compute bit-wise inversion, or bit-wise NOT, element-wise. — op_bitwise_not","title":"Compute bit-wise inversion, or bit-wise NOT, element-wise. — op_bitwise_not","text":"Computes bit-wise underlying binary representation integers input arrays. ufunc implements C/Python operator ~.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_bitwise_not.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Compute bit-wise inversion, or bit-wise NOT, element-wise. — op_bitwise_not","text":"","code":"op_bitwise_not(x)"},{"path":"https://keras3.posit.co/dev/reference/op_bitwise_not.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Compute bit-wise inversion, or bit-wise NOT, element-wise. — op_bitwise_not","text":"x Input integer tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_bitwise_not.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Compute bit-wise inversion, or bit-wise NOT, element-wise. — op_bitwise_not","text":"Result tensor.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_bitwise_or.html","id":null,"dir":"Reference","previous_headings":"","what":"Compute the bit-wise OR of two arrays element-wise. — op_bitwise_or","title":"Compute the bit-wise OR of two arrays element-wise. — op_bitwise_or","text":"Computes bit-wise underlying binary representation integers input arrays. ufunc implements C/Python operator |.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_bitwise_or.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Compute the bit-wise OR of two arrays element-wise. — op_bitwise_or","text":"","code":"op_bitwise_or(x, y)"},{"path":"https://keras3.posit.co/dev/reference/op_bitwise_or.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Compute the bit-wise OR of two arrays element-wise. — op_bitwise_or","text":"x Input integer tensor. y Input integer tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_bitwise_or.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Compute the bit-wise OR of two arrays element-wise. — op_bitwise_or","text":"Result tensor.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_bitwise_right_shift.html","id":null,"dir":"Reference","previous_headings":"","what":"Shift the bits of an integer to the right. — op_bitwise_right_shift","title":"Shift the bits of an integer to the right. — op_bitwise_right_shift","text":"Bits shifted right y. internal representation numbers binary format, operation equivalent dividing x 2**y.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_bitwise_right_shift.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Shift the bits of an integer to the right. — op_bitwise_right_shift","text":"","code":"op_bitwise_right_shift(x, y)"},{"path":"https://keras3.posit.co/dev/reference/op_bitwise_right_shift.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Shift the bits of an integer to the right. — op_bitwise_right_shift","text":"x Input integer tensor. y Input integer tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_bitwise_right_shift.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Shift the bits of an integer to the right. — op_bitwise_right_shift","text":"Result tensor.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_bitwise_xor.html","id":null,"dir":"Reference","previous_headings":"","what":"Compute the bit-wise XOR of two arrays element-wise. — op_bitwise_xor","title":"Compute the bit-wise XOR of two arrays element-wise. — op_bitwise_xor","text":"Computes bit-wise XOR underlying binary representation integers input arrays. ufunc implements C/Python operator ^.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_bitwise_xor.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Compute the bit-wise XOR of two arrays element-wise. — op_bitwise_xor","text":"","code":"op_bitwise_xor(x, y)"},{"path":"https://keras3.posit.co/dev/reference/op_bitwise_xor.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Compute the bit-wise XOR of two arrays element-wise. — op_bitwise_xor","text":"x Input integer tensor. y Input integer tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_bitwise_xor.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Compute the bit-wise XOR of two arrays element-wise. — op_bitwise_xor","text":"Result tensor.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_broadcast_to.html","id":null,"dir":"Reference","previous_headings":"","what":"Broadcast a tensor to a new shape. — op_broadcast_to","title":"Broadcast a tensor to a new shape. — op_broadcast_to","text":"Broadcast tensor new shape.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_broadcast_to.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Broadcast a tensor to a new shape. — op_broadcast_to","text":"","code":"op_broadcast_to(x, shape)"},{"path":"https://keras3.posit.co/dev/reference/op_broadcast_to.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Broadcast a tensor to a new shape. — op_broadcast_to","text":"x tensor broadcast. shape shape desired tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_broadcast_to.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Broadcast a tensor to a new shape. — op_broadcast_to","text":"tensor desired shape.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_broadcast_to.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Broadcast a tensor to a new shape. — op_broadcast_to","text":"","code":"x <- op_array(c(1, 2, 3)) op_broadcast_to(x, shape = c(3, 3)) ## tf.Tensor( ## [[1. 2. 3.] ##  [1. 2. 3.] ##  [1. 2. 3.]], shape=(3, 3), dtype=float32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_cast.html","id":null,"dir":"Reference","previous_headings":"","what":"Cast a tensor to the desired dtype. — op_cast","title":"Cast a tensor to the desired dtype. — op_cast","text":"Cast tensor desired dtype.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_cast.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Cast a tensor to the desired dtype. — op_cast","text":"","code":"op_cast(x, dtype)"},{"path":"https://keras3.posit.co/dev/reference/op_cast.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Cast a tensor to the desired dtype. — op_cast","text":"x tensor variable. dtype target type.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_cast.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Cast a tensor to the desired dtype. — op_cast","text":"tensor specified dtype.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_cast.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Cast a tensor to the desired dtype. — op_cast","text":"","code":"(x <- op_arange(4)) ## tf.Tensor([1. 2. 3. 4.], shape=(4), dtype=float32) op_cast(x, dtype = \"float16\") ## tf.Tensor([1. 2. 3. 4.], shape=(4), dtype=float16)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_categorical_crossentropy.html","id":null,"dir":"Reference","previous_headings":"","what":"Computes categorical cross-entropy loss between target and output tensor. — op_categorical_crossentropy","title":"Computes categorical cross-entropy loss between target and output tensor. — op_categorical_crossentropy","text":"categorical cross-entropy loss commonly used multi-class classification tasks input sample can belong one multiple classes. measures dissimilarity target output probabilities logits.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_categorical_crossentropy.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes categorical cross-entropy loss between target and output tensor. — op_categorical_crossentropy","text":"","code":"op_categorical_crossentropy(target, output, from_logits = FALSE, axis = -1L)"},{"path":"https://keras3.posit.co/dev/reference/op_categorical_crossentropy.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Computes categorical cross-entropy loss between target and output tensor. — op_categorical_crossentropy","text":"target target tensor representing true categorical labels. shape match shape output tensor except last dimension. output output tensor representing predicted probabilities logits. shape match shape target tensor except last dimension. from_logits (optional) Whether output tensor logits probabilities. Set TRUE output represents logits; otherwise, set FALSE output represents probabilities. Defaults FALSE. axis (optional) axis along categorical cross-entropy computed. Defaults -1, corresponds last dimension tensors.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_categorical_crossentropy.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Computes categorical cross-entropy loss between target and output tensor. — op_categorical_crossentropy","text":"Integer tensor: computed categorical cross-entropy loss target output.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_categorical_crossentropy.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Computes categorical cross-entropy loss between target and output tensor. — op_categorical_crossentropy","text":"","code":"target <- op_array(rbind(c(1, 0, 0),                         c(0, 1, 0),                         c(0, 0, 1))) output <- op_array(rbind(c(0.9, 0.05, 0.05),                         c(0.1, 0.8, 0.1),                         c(0.2, 0.3, 0.5))) op_categorical_crossentropy(target, output) ## tf.Tensor([0.10536052 0.22314355 0.69314718], shape=(3), dtype=float64)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_ceil.html","id":null,"dir":"Reference","previous_headings":"","what":"Return the ceiling of the input, element-wise. — op_ceil","title":"Return the ceiling of the input, element-wise. — op_ceil","text":"ceil scalar x smallest integer , >= x.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_ceil.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Return the ceiling of the input, element-wise. — op_ceil","text":"","code":"op_ceil(x)"},{"path":"https://keras3.posit.co/dev/reference/op_ceil.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Return the ceiling of the input, element-wise. — op_ceil","text":"x Input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_ceil.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Return the ceiling of the input, element-wise. — op_ceil","text":"ceiling element x, float dtype.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_celu.html","id":null,"dir":"Reference","previous_headings":"","what":"Continuously-differentiable exponential linear unit. — op_celu","title":"Continuously-differentiable exponential linear unit. — op_celu","text":"defined : f(x) =  alpha * (exp(x / alpha) - 1) x < 0, f(x) = x x >= 0.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_celu.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Continuously-differentiable exponential linear unit. — op_celu","text":"","code":"op_celu(x, alpha = 1)"},{"path":"https://keras3.posit.co/dev/reference/op_celu.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Continuously-differentiable exponential linear unit. — op_celu","text":"x Input tensor. alpha value CELU formulation. Defaults 1.0.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_celu.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Continuously-differentiable exponential linear unit. — op_celu","text":"tensor shape x.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_celu.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Continuously-differentiable exponential linear unit. — op_celu","text":"","code":"x <- op_array(c(-1., 0., 1.)) op_celu(x) ## tf.Tensor([-0.63212055  0.          1.        ], shape=(3), dtype=float32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_cholesky.html","id":null,"dir":"Reference","previous_headings":"","what":"Computes the Cholesky decomposition of a positive semi-definite matrix. — op_cholesky","title":"Computes the Cholesky decomposition of a positive semi-definite matrix. — op_cholesky","text":"Computes Cholesky decomposition positive semi-definite matrix.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_cholesky.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes the Cholesky decomposition of a positive semi-definite matrix. — op_cholesky","text":"","code":"op_cholesky(x)"},{"path":"https://keras3.posit.co/dev/reference/op_cholesky.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Computes the Cholesky decomposition of a positive semi-definite matrix. — op_cholesky","text":"x Input tensor shape (..., M, M).","code":""},{"path":"https://keras3.posit.co/dev/reference/op_cholesky.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Computes the Cholesky decomposition of a positive semi-definite matrix. — op_cholesky","text":"tensor shape (..., M, M) representing lower triangular Cholesky factor x.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_clip.html","id":null,"dir":"Reference","previous_headings":"","what":"Clip (limit) the values in a tensor. — op_clip","title":"Clip (limit) the values in a tensor. — op_clip","text":"Given interval, values outside interval clipped interval edges. example, interval [0, 1] specified, values smaller 0 become 0, values larger 1 become 1.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_clip.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Clip (limit) the values in a tensor. — op_clip","text":"","code":"op_clip(x, x_min, x_max)"},{"path":"https://keras3.posit.co/dev/reference/op_clip.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Clip (limit) the values in a tensor. — op_clip","text":"x Input tensor. x_min Minimum value. x_max Maximum value.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_clip.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Clip (limit) the values in a tensor. — op_clip","text":"clipped tensor.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_concatenate.html","id":null,"dir":"Reference","previous_headings":"","what":"Join a sequence of tensors along an existing axis. — op_concatenate","title":"Join a sequence of tensors along an existing axis. — op_concatenate","text":"Join sequence tensors along existing axis.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_concatenate.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Join a sequence of tensors along an existing axis. — op_concatenate","text":"","code":"op_concatenate(xs, axis = 1L)"},{"path":"https://keras3.posit.co/dev/reference/op_concatenate.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Join a sequence of tensors along an existing axis. — op_concatenate","text":"xs sequence tensors concatenate. axis axis along tensors joined. Defaults 0.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_concatenate.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Join a sequence of tensors along an existing axis. — op_concatenate","text":"concatenated tensor.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_cond.html","id":null,"dir":"Reference","previous_headings":"","what":"Conditionally applies true_fn or false_fn. — op_cond","title":"Conditionally applies true_fn or false_fn. — op_cond","text":"Conditionally applies true_fn false_fn.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_cond.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Conditionally applies true_fn or false_fn. — op_cond","text":"","code":"op_cond(pred, true_fn, false_fn)"},{"path":"https://keras3.posit.co/dev/reference/op_cond.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Conditionally applies true_fn or false_fn. — op_cond","text":"pred Boolean scalar type true_fn Callable returning output pred == TRUE case. false_fn Callable returning output pred == FALSE case.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_cond.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Conditionally applies true_fn or false_fn. — op_cond","text":"output either true_fn false_fn depending pred.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_cond.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Conditionally applies true_fn or false_fn. — op_cond","text":"","code":"fn <- tensorflow::tf_function(function(x) {   op_cond(x > 0,     true_fn = \\() x + 1,     false_fn = \\() x - 1) })  fn(tensorflow::as_tensor(1)) ## tf.Tensor(2.0, shape=(), dtype=float64) fn(tensorflow::as_tensor(-1)) ## tf.Tensor(-2.0, shape=(), dtype=float64) # # Conditional side-effect (print only, no return value). file <- tempfile(fileext = \".txt\") fn <- tensorflow::tf_function(function(epochs) {   op_fori_loop(     0, epochs,     body_fun = \\(epoch, state) {       op_cond(epoch %% 20 == 0,               \\() {                 tensorflow::tf$print(                   \"epoch:\", epoch,                   output_stream = paste0(\"file://\", file))                 NULL               },               \\() {NULL})       state     },     init_val = tensorflow::as_tensor(0, \"float32\")) })  fn(tensorflow::as_tensor(100, \"float32\")) ## tf.Tensor(0.0, shape=(), dtype=float32) readLines(file) ## [1] \"epoch: 0\"  \"epoch: 20\" \"epoch: 40\" \"epoch: 60\" \"epoch: 80\" # cleanup unlink(file)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_conj.html","id":null,"dir":"Reference","previous_headings":"","what":"Returns the complex conjugate, element-wise. — op_conj","title":"Returns the complex conjugate, element-wise. — op_conj","text":"complex conjugate complex number obtained changing sign imaginary part.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_conj.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Returns the complex conjugate, element-wise. — op_conj","text":"","code":"op_conj(x)"},{"path":"https://keras3.posit.co/dev/reference/op_conj.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Returns the complex conjugate, element-wise. — op_conj","text":"x Input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_conj.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Returns the complex conjugate, element-wise. — op_conj","text":"complex conjugate element x.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_conv.html","id":null,"dir":"Reference","previous_headings":"","what":"General N-D convolution. — op_conv","title":"General N-D convolution. — op_conv","text":"ops supports 1D, 2D 3D convolution.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_conv.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"General N-D convolution. — op_conv","text":"","code":"op_conv(   inputs,   kernel,   strides = 1L,   padding = \"valid\",   data_format = NULL,   dilation_rate = 1L )"},{"path":"https://keras3.posit.co/dev/reference/op_conv.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"General N-D convolution. — op_conv","text":"inputs Tensor rank N+2. inputs shape (batch_size,) + inputs_spatial_shape + (num_channels,) data_format = \"channels_last\", (batch_size, num_channels) + inputs_spatial_shape data_format = \"channels_first\". kernel Tensor rank N+2. kernel shape (kernel_spatial_shape, num_input_channels, num_output_channels). num_input_channels match number channels inputs. strides int int tuple/list len(inputs_spatial_shape), specifying strides convolution along spatial dimension. strides int, every spatial dimension shares strides. padding string, either \"valid\" \"\". \"valid\" means padding applied, \"\" results padding evenly left/right /input output height/width dimension input strides = 1. data_format string, either \"channels_last\" \"channels_first\". data_format determines ordering dimensions inputs. data_format = \"channels_last\", inputs shape (batch_size, ..., channels) data_format = \"channels_first\", inputs shape (batch_size, channels, ...). dilation_rate int int tuple/list len(inputs_spatial_shape), specifying dilation rate use dilated convolution. dilation_rate int, every spatial dimension shares dilation_rate.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_conv.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"General N-D convolution. — op_conv","text":"tensor rank N+2, result conv operation.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_conv_transpose.html","id":null,"dir":"Reference","previous_headings":"","what":"General N-D convolution transpose. — op_conv_transpose","title":"General N-D convolution transpose. — op_conv_transpose","text":"Also known de-convolution. ops supports 1D, 2D 3D convolution.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_conv_transpose.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"General N-D convolution transpose. — op_conv_transpose","text":"","code":"op_conv_transpose(   inputs,   kernel,   strides,   padding = \"valid\",   output_padding = NULL,   data_format = NULL,   dilation_rate = 1L )"},{"path":"https://keras3.posit.co/dev/reference/op_conv_transpose.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"General N-D convolution transpose. — op_conv_transpose","text":"inputs Tensor rank N+2. inputs shape (batch_size,) + inputs_spatial_shape + (num_channels,) data_format = \"channels_last\", (batch_size, num_channels) + inputs_spatial_shape data_format = \"channels_first\". kernel Tensor rank N+2. kernel shape [kernel_spatial_shape, num_output_channels, num_input_channels], num_input_channels match number channels inputs. strides int int tuple/list len(inputs_spatial_shape), specifying strides convolution along spatial dimension. strides int, every spatial dimension shares strides. padding string, either \"valid\" \"\". \"valid\" means padding applied, \"\" results padding evenly left/right /input output height/width dimension input strides = 1. output_padding int int tuple/list len(inputs_spatial_shape), specifying amount padding along height width output tensor. Can single integer specify value spatial dimensions. amount output padding along given dimension must lower stride along dimension. set NULL (default), output shape inferred. data_format string, either \"channels_last\" \"channels_first\". data_format determines ordering dimensions inputs. data_format = \"channels_last\", inputs shape (batch_size, ..., channels) data_format = \"channels_first\", inputs shape (batch_size, channels, ...). dilation_rate int int tuple/list len(inputs_spatial_shape), specifying dilation rate use dilated convolution. dilation_rate int, every spatial dimension shares dilation_rate.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_conv_transpose.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"General N-D convolution transpose. — op_conv_transpose","text":"tensor rank N+2, result conv operation.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_convert_to_numpy.html","id":null,"dir":"Reference","previous_headings":"","what":"Convert a tensor to an R or NumPy array. — op_convert_to_numpy","title":"Convert a tensor to an R or NumPy array. — op_convert_to_numpy","text":"Convert tensor R NumPy array.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_convert_to_numpy.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Convert a tensor to an R or NumPy array. — op_convert_to_numpy","text":"","code":"op_convert_to_numpy(x)  op_convert_to_array(x)"},{"path":"https://keras3.posit.co/dev/reference/op_convert_to_numpy.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Convert a tensor to an R or NumPy array. — op_convert_to_numpy","text":"x tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_convert_to_numpy.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Convert a tensor to an R or NumPy array. — op_convert_to_numpy","text":"NumPy array R array.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_convert_to_tensor.html","id":null,"dir":"Reference","previous_headings":"","what":"Convert an array to a tensor. — op_convert_to_tensor","title":"Convert an array to a tensor. — op_convert_to_tensor","text":"Native tensors current backend left unchanged unless dtype, sparse ragged arguments set.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_convert_to_tensor.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Convert an array to a tensor. — op_convert_to_tensor","text":"","code":"op_convert_to_tensor(x, dtype = NULL, sparse = NULL, ragged = NULL)"},{"path":"https://keras3.posit.co/dev/reference/op_convert_to_tensor.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Convert an array to a tensor. — op_convert_to_tensor","text":"x R NumPy array (can nested) backend tensor. dtype target type. NULL, type x used. sparse Whether keep sparse tensors. FALSE cause sparse tensors densified. default value NULL means sparse tensors kept backend supports . ragged Whether keep ragged tensors. FALSE cause ragged tensors densified. default value NULL means ragged tensors kept backend supports .","code":""},{"path":"https://keras3.posit.co/dev/reference/op_convert_to_tensor.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Convert an array to a tensor. — op_convert_to_tensor","text":"backend tensor specified dtype sparseness.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_convert_to_tensor.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Convert an array to a tensor. — op_convert_to_tensor","text":"","code":"x <- array(c(1, 2, 3)) y <- op_convert_to_tensor(x) y ## tf.Tensor([1. 2. 3.], shape=(3), dtype=float64) op_convert_to_tensor(c(1, 3, 2, 0), \"int32\") ## tf.Tensor([1 3 2 0], shape=(4), dtype=int32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_copy.html","id":null,"dir":"Reference","previous_headings":"","what":"Returns a copy of x. — op_copy","title":"Returns a copy of x. — op_copy","text":"Returns copy x.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_copy.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Returns a copy of x. — op_copy","text":"","code":"op_copy(x)"},{"path":"https://keras3.posit.co/dev/reference/op_copy.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Returns a copy of x. — op_copy","text":"x Input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_copy.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Returns a copy of x. — op_copy","text":"copy x.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_correlate.html","id":null,"dir":"Reference","previous_headings":"","what":"Compute the cross-correlation of two 1-dimensional tensors. — op_correlate","title":"Compute the cross-correlation of two 1-dimensional tensors. — op_correlate","text":"Compute cross-correlation two 1-dimensional tensors.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_correlate.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Compute the cross-correlation of two 1-dimensional tensors. — op_correlate","text":"","code":"op_correlate(x1, x2, mode = \"valid\")"},{"path":"https://keras3.posit.co/dev/reference/op_correlate.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Compute the cross-correlation of two 1-dimensional tensors. — op_correlate","text":"x1 First 1-dimensional input tensor length M. x2 Second 1-dimensional input tensor length N. mode Either \"valid\", \"\" \"full\". default mode set \"valid\", returns output length max(M, N) - min(M, N) + 1. \"\" returns output length max(M, N). \"full\" mode returns convolution point overlap, output length N+M-1.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_correlate.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Compute the cross-correlation of two 1-dimensional tensors. — op_correlate","text":"Output tensor, cross-correlation x1 x2.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_cos.html","id":null,"dir":"Reference","previous_headings":"","what":"Cosine, element-wise. — op_cos","title":"Cosine, element-wise. — op_cos","text":"Cosine, element-wise.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_cos.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Cosine, element-wise. — op_cos","text":"","code":"op_cos(x)"},{"path":"https://keras3.posit.co/dev/reference/op_cos.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Cosine, element-wise. — op_cos","text":"x Input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_cos.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Cosine, element-wise. — op_cos","text":"corresponding cosine values.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_cosh.html","id":null,"dir":"Reference","previous_headings":"","what":"Hyperbolic cosine, element-wise. — op_cosh","title":"Hyperbolic cosine, element-wise. — op_cosh","text":"Hyperbolic cosine, element-wise.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_cosh.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Hyperbolic cosine, element-wise. — op_cosh","text":"","code":"op_cosh(x)"},{"path":"https://keras3.posit.co/dev/reference/op_cosh.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Hyperbolic cosine, element-wise. — op_cosh","text":"x Input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_cosh.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Hyperbolic cosine, element-wise. — op_cosh","text":"Output tensor shape x.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_count_nonzero.html","id":null,"dir":"Reference","previous_headings":"","what":"Counts the number of non-zero values in x along the given axis. — op_count_nonzero","title":"Counts the number of non-zero values in x along the given axis. — op_count_nonzero","text":"axis specified non-zeros tensor counted.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_count_nonzero.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Counts the number of non-zero values in x along the given axis. — op_count_nonzero","text":"","code":"op_count_nonzero(x, axis = NULL)"},{"path":"https://keras3.posit.co/dev/reference/op_count_nonzero.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Counts the number of non-zero values in x along the given axis. — op_count_nonzero","text":"x Input tensor. axis Axis tuple axes along count number non-zeros. Defaults NULL.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_count_nonzero.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Counts the number of non-zero values in x along the given axis. — op_count_nonzero","text":"integer tensor integers.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_count_nonzero.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Counts the number of non-zero values in x along the given axis. — op_count_nonzero","text":"","code":"x <- op_array(rbind(c(0, 1, 7, 0),                    c(3, 0, 2, 19))) op_count_nonzero(x) ## tf.Tensor(5, shape=(), dtype=int32) op_count_nonzero(x, axis = 1) ## tf.Tensor([1 1 2 1], shape=(4), dtype=int32) op_count_nonzero(x, axis = 2) ## tf.Tensor([2 3], shape=(2), dtype=int32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_cross.html","id":null,"dir":"Reference","previous_headings":"","what":"Returns the cross product of two (arrays of) vectors. — op_cross","title":"Returns the cross product of two (arrays of) vectors. — op_cross","text":"cross product x1 x2 R^3 vector perpendicular x1 x2. x1 x2 arrays vectors, vectors defined last axis x1 x2 default, axes can dimensions 2 3. dimension either x1 x2 2, third component input vector assumed zero cross product calculated accordingly. cases input vectors dimension 2, z-component cross product returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_cross.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Returns the cross product of two (arrays of) vectors. — op_cross","text":"","code":"op_cross(x1, x2, axisa = -1L, axisb = -1L, axisc = -1L, axis = NULL)"},{"path":"https://keras3.posit.co/dev/reference/op_cross.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Returns the cross product of two (arrays of) vectors. — op_cross","text":"x1 Components first vector(s). x2 Components second vector(s). axisa Axis x1 defines vector(s). Defaults -1. axisb Axis x2 defines vector(s). Defaults -1. axisc Axis result containing cross product vector(s). Ignored input vectors dimension 2, return scalar. default, last axis. axis defined, axis x1, x2 result defines vector(s) cross product(s). Overrides axisa, axisb axisc.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_cross.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Returns the cross product of two (arrays of) vectors. — op_cross","text":"Vector cross product(s).","code":""},{"path":"https://keras3.posit.co/dev/reference/op_cross.html","id":"note","dir":"Reference","previous_headings":"","what":"Note","title":"Returns the cross product of two (arrays of) vectors. — op_cross","text":"Torch backend support two dimensional vectors, arguments axisa, axisb axisc. Use axis instead.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_ctc_decode.html","id":null,"dir":"Reference","previous_headings":"","what":"Decodes the output of a CTC model. — op_ctc_decode","title":"Decodes the output of a CTC model. — op_ctc_decode","text":"Decodes output CTC model.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_ctc_decode.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Decodes the output of a CTC model. — op_ctc_decode","text":"","code":"op_ctc_decode(   inputs,   sequence_lengths,   strategy = \"greedy\",   beam_width = 100L,   top_paths = 1L,   merge_repeated = TRUE,   mask_index = 1L )"},{"path":"https://keras3.posit.co/dev/reference/op_ctc_decode.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Decodes the output of a CTC model. — op_ctc_decode","text":"inputs tensor shape (batch_size, max_length, num_classes) containing logits (output model). normalized via softmax. sequence_lengths tensor shape (batch_size) containing sequence lengths batch. strategy string decoding strategy. Supported values \"greedy\" \"beam_search\". beam_width integer scalar beam width used beam search. Defaults 100. top_paths integer scalar, number top paths return. Defaults 1. merge_repeated boolean scalar, whether merge repeated labels output. Defaults TRUE. mask_index integer scalar, (1-based) index mask character vocabulary. Defaults 1.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_ctc_decode.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Decodes the output of a CTC model. — op_ctc_decode","text":"list containing: tensor representing list decoded sequences. strategy=\"greedy\", shape (1, batch_size, max_length). strategy=\"beam_search\", shape (top_paths, batch_size, max_length). Note : -1 indicates blank label. strategy=\"greedy\", tensor shape (batch_size, 1) representing negative sum probability logits sequence. strategy=\"beam_search\", tensor shape (batch_size, top_paths) representing log probability sequence.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_ctc_loss.html","id":null,"dir":"Reference","previous_headings":"","what":"CTC (Connectionist Temporal Classification) loss. — op_ctc_loss","title":"CTC (Connectionist Temporal Classification) loss. — op_ctc_loss","text":"CTC (Connectionist Temporal Classification) loss.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_ctc_loss.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"CTC (Connectionist Temporal Classification) loss. — op_ctc_loss","text":"","code":"op_ctc_loss(target, output, target_length, output_length, mask_index = 1L)"},{"path":"https://keras3.posit.co/dev/reference/op_ctc_loss.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"CTC (Connectionist Temporal Classification) loss. — op_ctc_loss","text":"target tensor shape (batch_size, max_length) containing true labels integer format. output tensor shape (batch_size, max_length, num_classes) containing logits (output model). target_length tensor shape (batch_size) containing true label lengths. output_length tensor shape (batch_size) containing output lengths. mask_index index mask character vocabulary. Defaults 1 (first element vocabulary).","code":""},{"path":"https://keras3.posit.co/dev/reference/op_ctc_loss.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"CTC (Connectionist Temporal Classification) loss. — op_ctc_loss","text":"tensor, shape (batch_size), loss values.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_cumprod.html","id":null,"dir":"Reference","previous_headings":"","what":"Return the cumulative product of elements along a given axis. — op_cumprod","title":"Return the cumulative product of elements along a given axis. — op_cumprod","text":"Return cumulative product elements along given axis.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_cumprod.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Return the cumulative product of elements along a given axis. — op_cumprod","text":"","code":"op_cumprod(x, axis = NULL, dtype = NULL)"},{"path":"https://keras3.posit.co/dev/reference/op_cumprod.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Return the cumulative product of elements along a given axis. — op_cumprod","text":"x Input tensor. axis Axis along cumulative product computed. default input flattened. dtype dtype returned tensor. Defaults x$dtype.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_cumprod.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Return the cumulative product of elements along a given axis. — op_cumprod","text":"Output tensor.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_cumsum.html","id":null,"dir":"Reference","previous_headings":"","what":"Returns the cumulative sum of elements along a given axis. — op_cumsum","title":"Returns the cumulative sum of elements along a given axis. — op_cumsum","text":"Returns cumulative sum elements along given axis.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_cumsum.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Returns the cumulative sum of elements along a given axis. — op_cumsum","text":"","code":"op_cumsum(x, axis = NULL, dtype = NULL)"},{"path":"https://keras3.posit.co/dev/reference/op_cumsum.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Returns the cumulative sum of elements along a given axis. — op_cumsum","text":"x Input tensor. axis Axis along cumulative sum computed. default input flattened. dtype dtype returned tensor. Defaults x$dtype.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_cumsum.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Returns the cumulative sum of elements along a given axis. — op_cumsum","text":"Output tensor.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_custom_gradient.html","id":null,"dir":"Reference","previous_headings":"","what":"Decorator to define a function with a custom gradient. — op_custom_gradient","title":"Decorator to define a function with a custom gradient. — op_custom_gradient","text":"decorator allows fine grained control gradients sequence operations. may useful multiple reasons, including providing efficient numerically stable gradient sequence operations.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_custom_gradient.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Decorator to define a function with a custom gradient. — op_custom_gradient","text":"","code":"op_custom_gradient(f)"},{"path":"https://keras3.posit.co/dev/reference/op_custom_gradient.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Decorator to define a function with a custom gradient. — op_custom_gradient","text":"f Function f(...) returns tuple (output, grad_fn) : ... sequence unnamed arguments, tensor input nested structure tensor inputs function. output (potentially nested structure ) tensor outputs applying operations forward_fn f() .... grad_fn function signature grad_fn(..., upstream) returns list tensors size (flattened) ...: derivatives tensors output respect tensors .... upstream tensor sequence tensors holding initial value gradients tensor output.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_custom_gradient.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Decorator to define a function with a custom gradient. — op_custom_gradient","text":"function h(...) returns value f(...)[[1]] whose gradient determined f(...)[[2]].","code":""},{"path":"https://keras3.posit.co/dev/reference/op_custom_gradient.html","id":"note","dir":"Reference","previous_headings":"","what":"Note","title":"Decorator to define a function with a custom gradient. — op_custom_gradient","text":"Note grad function returns gradient computation requires ... well upstream named argument, depending backend set. JAX TensorFlow backends, requires one argument, whereas might use upstream argument case PyTorch backend. working TensorFlow/JAX backend, grad(upstream) sufficient. PyTorch, grad function requires ... well upstream, e.g. grad <- \\(..., upstream). Follow example use op_custom_gradient() way compatible backends.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_custom_gradient.html","id":"example","dir":"Reference","previous_headings":"","what":"Example","title":"Decorator to define a function with a custom gradient. — op_custom_gradient","text":"Backend-agnostic example.","code":"log1pexp <- op_custom_gradient(\\(x) {      e <- op_exp(x)      grad <- function(..., upstream = NULL) {       upstream <- upstream %||% ..1       op_multiply(upstream, 1.0 - 1.0 / op_add(1, e))     }      tuple(op_log(1 + e), grad)  })  if(config_backend() == \"tensorflow\") {   tf <- tensorflow::tf   x <- op_convert_to_tensor(100.0)   with(tf$GradientTape() %as% tape, {     tape$watch(x)     y <- log1pexp(x)   })   dy_dx <- tape$gradient(y, x)   stopifnot(as.numeric(dy_dx) == 1) }"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_depthwise_conv.html","id":null,"dir":"Reference","previous_headings":"","what":"General N-D depthwise convolution. — op_depthwise_conv","title":"General N-D depthwise convolution. — op_depthwise_conv","text":"ops supports 1D 2D depthwise convolution.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_depthwise_conv.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"General N-D depthwise convolution. — op_depthwise_conv","text":"","code":"op_depthwise_conv(   inputs,   kernel,   strides = 1L,   padding = \"valid\",   data_format = NULL,   dilation_rate = 1L )"},{"path":"https://keras3.posit.co/dev/reference/op_depthwise_conv.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"General N-D depthwise convolution. — op_depthwise_conv","text":"inputs Tensor rank N+2. inputs shape (batch_size,) + inputs_spatial_shape + (num_channels,) data_format = \"channels_last\", (batch_size, num_channels) + inputs_spatial_shape data_format = \"channels_first\". kernel Tensor rank N+2. kernel shape [kernel_spatial_shape, num_input_channels, num_channels_multiplier], num_input_channels match number channels inputs. strides int int tuple/list len(inputs_spatial_shape), specifying strides convolution along spatial dimension. strides int, every spatial dimension shares strides. padding string, either \"valid\" \"\". \"valid\" means padding applied, \"\" results padding evenly left/right /input output height/width dimension input strides = 1. data_format string, either \"channels_last\" \"channels_first\". data_format determines ordering dimensions inputs. data_format = \"channels_last\", inputs shape (batch_size, ..., channels) data_format = \"channels_first\", inputs shape (batch_size, channels, ...). dilation_rate int int tuple/list len(inputs_spatial_shape), specifying dilation rate use dilated convolution. dilation_rate int, every spatial dimension shares dilation_rate.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_depthwise_conv.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"General N-D depthwise convolution. — op_depthwise_conv","text":"tensor rank N+2, result depthwise conv operation.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_det.html","id":null,"dir":"Reference","previous_headings":"","what":"Computes the determinant of a square tensor. — op_det","title":"Computes the determinant of a square tensor. — op_det","text":"Computes determinant square tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_det.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes the determinant of a square tensor. — op_det","text":"","code":"op_det(x)"},{"path":"https://keras3.posit.co/dev/reference/op_det.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Computes the determinant of a square tensor. — op_det","text":"x Input tensor shape (..., M, M).","code":""},{"path":"https://keras3.posit.co/dev/reference/op_det.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Computes the determinant of a square tensor. — op_det","text":"tensor shape (...) representing determinant x.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_diag.html","id":null,"dir":"Reference","previous_headings":"","what":"Extract a diagonal or construct a diagonal array. — op_diag","title":"Extract a diagonal or construct a diagonal array. — op_diag","text":"Extract diagonal construct diagonal array.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_diag.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Extract a diagonal or construct a diagonal array. — op_diag","text":"","code":"op_diag(x, k = 0L)"},{"path":"https://keras3.posit.co/dev/reference/op_diag.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Extract a diagonal or construct a diagonal array. — op_diag","text":"x Input tensor. x 2-D, returns k-th diagonal x. x 1-D, return 2-D tensor x k-th diagonal. k diagonal consider. Defaults 0. Use k > 0 diagonals main diagonal, k < 0 diagonals main diagonal.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_diag.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Extract a diagonal or construct a diagonal array. — op_diag","text":"extracted diagonal constructed diagonal tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_diag.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Extract a diagonal or construct a diagonal array. — op_diag","text":"","code":"x <- op_arange(9L) |> op_reshape(c(3, 3)) x ## tf.Tensor( ## [[1 2 3] ##  [4 5 6] ##  [7 8 9]], shape=(3, 3), dtype=int32) op_diag(x) ## tf.Tensor([1 5 9], shape=(3), dtype=int32) op_diag(x, k = 1) ## tf.Tensor([2 6], shape=(2), dtype=int32) op_diag(x, k = -1) ## tf.Tensor([4 8], shape=(2), dtype=int32) op_diag(op_diag(x)) ## tf.Tensor( ## [[1 0 0] ##  [0 5 0] ##  [0 0 9]], shape=(3, 3), dtype=int32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_diagflat.html","id":null,"dir":"Reference","previous_headings":"","what":"Create a two-dimensional array with the flattened input diagonal. — op_diagflat","title":"Create a two-dimensional array with the flattened input diagonal. — op_diagflat","text":"k-th diagonal.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_diagflat.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Create a two-dimensional array with the flattened input diagonal. — op_diagflat","text":"","code":"op_diagflat(x, k = 0L)"},{"path":"https://keras3.posit.co/dev/reference/op_diagflat.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Create a two-dimensional array with the flattened input diagonal. — op_diagflat","text":"x Input tensor flattened placed diagonal. k diagonal place flattened input. Defaults 0. Use k > 0 diagonals main diagonal, k < 0 diagonals main diagonal.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_diagflat.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Create a two-dimensional array with the flattened input diagonal. — op_diagflat","text":"2-D tensor flattened input specified diagonal.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_diagonal.html","id":null,"dir":"Reference","previous_headings":"","what":"Return specified diagonals. — op_diagonal","title":"Return specified diagonals. — op_diagonal","text":"x 2-D, returns diagonal x given offset, .e., collection elements form x[, +offset]. x two dimensions, axes specified axis1 axis2 used determine 2-D sub-array whose diagonal returned. shape resulting array can determined removing axis1 axis2 appending index right equal size resulting diagonals.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_diagonal.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Return specified diagonals. — op_diagonal","text":"","code":"op_diagonal(x, offset = 0L, axis1 = 1L, axis2 = 2L)"},{"path":"https://keras3.posit.co/dev/reference/op_diagonal.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Return specified diagonals. — op_diagonal","text":"x Input tensor. offset Offset diagonal main diagonal. Can positive negative. Defaults 0 (main diagonal). axis1 Axis used first axis 2-D sub-arrays. Defaults 1 (first axis). axis2 Axis used second axis 2-D sub-arrays. Defaults 2 (second axis).","code":""},{"path":"https://keras3.posit.co/dev/reference/op_diagonal.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Return specified diagonals. — op_diagonal","text":"Tensor diagonals.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_diagonal.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Return specified diagonals. — op_diagonal","text":"","code":"x <- op_arange(4L) |> op_reshape(c(2, 2)) x ## tf.Tensor( ## [[1 2] ##  [3 4]], shape=(2, 2), dtype=int32) op_diagonal(x) ## tf.Tensor([1 4], shape=(2), dtype=int32) op_diagonal(x, offset = 1) ## tf.Tensor([2], shape=(1), dtype=int32) x <- op_array(1:8) |> op_reshape(c(2, 2, 2)) x ## tf.Tensor( ## [[[1 2] ##   [3 4]] ## ##  [[5 6] ##   [7 8]]], shape=(2, 2, 2), dtype=int32) x |> op_diagonal(0) ## tf.Tensor( ## [[1 7] ##  [2 8]], shape=(2, 2), dtype=int32) x |> op_diagonal(0, 1, 2) # same as above, the default ## tf.Tensor( ## [[1 7] ##  [2 8]], shape=(2, 2), dtype=int32) x |> op_diagonal(0, 2, 3) ## tf.Tensor( ## [[1 4] ##  [5 8]], shape=(2, 2), dtype=int32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_diff.html","id":null,"dir":"Reference","previous_headings":"","what":"Calculate the n-th discrete difference along the given axis. — op_diff","title":"Calculate the n-th discrete difference along the given axis. — op_diff","text":"first difference given [] = [+1] - [] along given axis, higher differences calculated using diff recursively.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_diff.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Calculate the n-th discrete difference along the given axis. — op_diff","text":"","code":"op_diff(a, n = 1L, axis = -1L)"},{"path":"https://keras3.posit.co/dev/reference/op_diff.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Calculate the n-th discrete difference along the given axis. — op_diff","text":"Input tensor. n number times values differenced. Defaults 1. axis Axis compute discrete difference(s) along. Defaults -1 (last axis).","code":""},{"path":"https://keras3.posit.co/dev/reference/op_diff.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Calculate the n-th discrete difference along the given axis. — op_diff","text":"Tensor diagonals.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_diff.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Calculate the n-th discrete difference along the given axis. — op_diff","text":"","code":"x <- op_array(c(1, 2, 4, 7, 0)) op_diff(x) ## tf.Tensor([ 1.  2.  3. -7.], shape=(4), dtype=float32) op_diff(x, n = 2) ## tf.Tensor([  1.   1. -10.], shape=(3), dtype=float32) x <- op_array(rbind(c(1, 3, 6, 10),                     c(0, 5, 6, 8))) op_diff(x) ## tf.Tensor( ## [[2. 3. 4.] ##  [5. 1. 2.]], shape=(2, 3), dtype=float64) op_diff(x, axis = 1) ## tf.Tensor([[-1.  2.  0. -2.]], shape=(1, 4), dtype=float64)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_digitize.html","id":null,"dir":"Reference","previous_headings":"","what":"Returns the indices of the bins to which each value in x belongs. — op_digitize","title":"Returns the indices of the bins to which each value in x belongs. — op_digitize","text":"Returns indices bins value x belongs.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_digitize.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Returns the indices of the bins to which each value in x belongs. — op_digitize","text":"","code":"op_digitize(x, bins, zero_indexed = FALSE)"},{"path":"https://keras3.posit.co/dev/reference/op_digitize.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Returns the indices of the bins to which each value in x belongs. — op_digitize","text":"x Input array binned. bins Array bins. one-dimensional monotonically increasing. zero_indexed TRUE, returned indices zero-based (0 encodes first bin); FALSE (default), returned indices one-based (1 encodes first bin).","code":""},{"path":"https://keras3.posit.co/dev/reference/op_digitize.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Returns the indices of the bins to which each value in x belongs. — op_digitize","text":"Output array indices, shape x.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_digitize.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Returns the indices of the bins to which each value in x belongs. — op_digitize","text":"","code":"x <- op_array(c(-1,    0, 1,    3, 3.5,      4.5,  1.6,    8)) bins <- array(c(    0,       3,         4.5,            7)) op_digitize(x, bins) ## tf.Tensor([1 2 2 3 3 4 2 5], shape=(8), dtype=int32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_divide.html","id":null,"dir":"Reference","previous_headings":"","what":"Divide arguments element-wise. — op_divide","title":"Divide arguments element-wise. — op_divide","text":"Note function automatically called using R operator * tensor.","code":"(x <- op_arange(4)) ## tf.Tensor([1. 2. 3. 4.], shape=(4), dtype=float32) op_divide(x, 2) ## tf.Tensor([0.5 1.  1.5 2. ], shape=(4), dtype=float32) x / 2 ## tf.Tensor([0.5 1.  1.5 2. ], shape=(4), dtype=float32)"},{"path":"https://keras3.posit.co/dev/reference/op_divide.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Divide arguments element-wise. — op_divide","text":"","code":"op_divide(x1, x2)"},{"path":"https://keras3.posit.co/dev/reference/op_divide.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Divide arguments element-wise. — op_divide","text":"x1 First input tensor. x2 Second input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_divide.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Divide arguments element-wise. — op_divide","text":"Output tensor, quotient x1/x2, element-wise.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_divide.html","id":"example","dir":"Reference","previous_headings":"","what":"Example","title":"Divide arguments element-wise. — op_divide","text":"","code":"op_divide(3, 2) ## tf.Tensor(1.5, shape=(), dtype=float32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_divide_no_nan.html","id":null,"dir":"Reference","previous_headings":"","what":"Safe element-wise division which returns 0 where the denominator is 0. — op_divide_no_nan","title":"Safe element-wise division which returns 0 where the denominator is 0. — op_divide_no_nan","text":"Safe element-wise division returns 0 denominator 0.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_divide_no_nan.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Safe element-wise division which returns 0 where the denominator is 0. — op_divide_no_nan","text":"","code":"op_divide_no_nan(x1, x2)"},{"path":"https://keras3.posit.co/dev/reference/op_divide_no_nan.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Safe element-wise division which returns 0 where the denominator is 0. — op_divide_no_nan","text":"x1 First input tensor. x2 Second input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_divide_no_nan.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Safe element-wise division which returns 0 where the denominator is 0. — op_divide_no_nan","text":"quotient x1/x2, element-wise, zero x2 zero.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_dot.html","id":null,"dir":"Reference","previous_headings":"","what":"Dot product of two tensors. — op_dot","title":"Dot product of two tensors. — op_dot","text":"x1 x2 1-D tensors, inner product vectors (without complex conjugation). x1 x2 2-D tensors, matrix multiplication. either x1 x2 0-D (scalar), equivalent x1 * x2. x1 N-D tensor x2 1-D tensor, sum product last axis x1 x2. x1 N-D tensor x2 M-D tensor (M >= 2), sum product last axis x1 second--last axis x2: dot(x1, x2)[,j,k,m] = sum([,j,:] * b[k,:,m]).","code":""},{"path":"https://keras3.posit.co/dev/reference/op_dot.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Dot product of two tensors. — op_dot","text":"","code":"op_dot(x1, x2)"},{"path":"https://keras3.posit.co/dev/reference/op_dot.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Dot product of two tensors. — op_dot","text":"x1 First argument. x2 Second argument.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_dot.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Dot product of two tensors. — op_dot","text":"Dot product x1 x2.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_dot.html","id":"note","dir":"Reference","previous_headings":"","what":"Note","title":"Dot product of two tensors. — op_dot","text":"Torch backend accept 0-D tensors arguments.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_dot_product_attention.html","id":null,"dir":"Reference","previous_headings":"","what":"Scaled dot product attention function. — op_dot_product_attention","title":"Scaled dot product attention function. — op_dot_product_attention","text":"Computes attention function Q (query), K (key), V(value): attention(Q, K, V) = softmax(Q * K / sqrt(d)) * V. define logits output Q * K probs output softmax. Throughout function, utilize following notation represent shape array: B: batch size S: length key/value T: length query N: number attention heads H: dimensions attention head K: number key/value heads G: number groups, equals N // K","code":""},{"path":"https://keras3.posit.co/dev/reference/op_dot_product_attention.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Scaled dot product attention function. — op_dot_product_attention","text":"","code":"op_dot_product_attention(   query,   key,   value,   bias = NULL,   mask = NULL,   scale = NULL,   is_causal = FALSE,   flash_attention = NULL )"},{"path":"https://keras3.posit.co/dev/reference/op_dot_product_attention.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Scaled dot product attention function. — op_dot_product_attention","text":"query query array shape (B, T, N, H). key key array shape (B, S, K, H). K equals N, multi-headed attention (MHA) performed. Otherwise, grouped query attention (GQA) performed N multiple K. multi-query attention (MQA) performed K==1 (special case GQA). value value array shape key. bias Optional bias array added logits. shape must broadcastable (B, N, T, S). mask Optional mask array used filter logits. boolean mask TRUE indicates element take part attention. additive mask, users pass bias. shape must broadcastable (B, N, T, S). scale Optional scale logits. NULL, scale set 1.0 / sqrt(H). is_causal Whether apply causal mask. flash_attention Whether use flash attention. NULL, attempt use flash attention required conditions met. Typically, inputs must float16 bfloat16 dtype input layout requirements may vary depending backend.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_dot_product_attention.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Scaled dot product attention function. — op_dot_product_attention","text":"array attention output shape query.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_dot_product_attention.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Scaled dot product attention function. — op_dot_product_attention","text":"","code":"query = random_normal(c(2, 4, 8, 16)) key = random_normal(c(2, 6, 8, 16)) value = random_normal(c(2, 6, 8, 16)) op_dot_product_attention(query, key, value) |> op_shape() ## shape(2, 4, 8, 16)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_dtype.html","id":null,"dir":"Reference","previous_headings":"","what":"Return the dtype of the tensor input as a standardized string. — op_dtype","title":"Return the dtype of the tensor input as a standardized string. — op_dtype","text":"Note due standardization, dtype compare equal backend-specific version dtype.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_dtype.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Return the dtype of the tensor input as a standardized string. — op_dtype","text":"","code":"op_dtype(x)"},{"path":"https://keras3.posit.co/dev/reference/op_dtype.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Return the dtype of the tensor input as a standardized string. — op_dtype","text":"x tensor. function try access dtype attribute input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_dtype.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Return the dtype of the tensor input as a standardized string. — op_dtype","text":"string indicating dtype input tensor, e.g. \"float32\".","code":""},{"path":"https://keras3.posit.co/dev/reference/op_dtype.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Return the dtype of the tensor input as a standardized string. — op_dtype","text":"","code":"x <- op_zeros(c(8, 12)) op_dtype(x) ## [1] \"float32\""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_eig.html","id":null,"dir":"Reference","previous_headings":"","what":"Computes the eigenvalues and eigenvectors of a square matrix. — op_eig","title":"Computes the eigenvalues and eigenvectors of a square matrix. — op_eig","text":"Computes eigenvalues eigenvectors square matrix.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_eig.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes the eigenvalues and eigenvectors of a square matrix. — op_eig","text":"","code":"op_eig(x)"},{"path":"https://keras3.posit.co/dev/reference/op_eig.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Computes the eigenvalues and eigenvectors of a square matrix. — op_eig","text":"x Input tensor shape (..., M, M).","code":""},{"path":"https://keras3.posit.co/dev/reference/op_eig.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Computes the eigenvalues and eigenvectors of a square matrix. — op_eig","text":"list two tensors: tensor shape (..., M) containing eigenvalues tensor shape (..., M, M) containing eigenvectors.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_eigh.html","id":null,"dir":"Reference","previous_headings":"","what":"Computes the eigenvalues and eigenvectors of a complex Hermitian. — op_eigh","title":"Computes the eigenvalues and eigenvectors of a complex Hermitian. — op_eigh","text":"Computes eigenvalues eigenvectors complex Hermitian.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_eigh.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes the eigenvalues and eigenvectors of a complex Hermitian. — op_eigh","text":"","code":"op_eigh(x)"},{"path":"https://keras3.posit.co/dev/reference/op_eigh.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Computes the eigenvalues and eigenvectors of a complex Hermitian. — op_eigh","text":"x Input tensor shape (..., M, M).","code":""},{"path":"https://keras3.posit.co/dev/reference/op_eigh.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Computes the eigenvalues and eigenvectors of a complex Hermitian. — op_eigh","text":"list two tensors: tensor shape (..., M) containing eigenvalues tensor shape (..., M, M) containing eigenvectors.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_einsum.html","id":null,"dir":"Reference","previous_headings":"","what":"Evaluates the Einstein summation convention on the operands. — op_einsum","title":"Evaluates the Einstein summation convention on the operands. — op_einsum","text":"Evaluates Einstein summation convention operands.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_einsum.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Evaluates the Einstein summation convention on the operands. — op_einsum","text":"","code":"op_einsum(subscripts, ...)"},{"path":"https://keras3.posit.co/dev/reference/op_einsum.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Evaluates the Einstein summation convention on the operands. — op_einsum","text":"subscripts Specifies subscripts summation comma separated list subscript labels. implicit (classical Einstein summation) calculation performed unless explicit indicator -> included well subscript labels precise output form. ... operands compute Einstein sum .","code":""},{"path":"https://keras3.posit.co/dev/reference/op_einsum.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Evaluates the Einstein summation convention on the operands. — op_einsum","text":"calculation based Einstein summation convention.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_einsum.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Evaluates the Einstein summation convention on the operands. — op_einsum","text":"Trace matrix:     Extract diagonal:     Sum axis:     higher dimensional tensors summing single axis can done ellipsis:     Compute matrix transpose reorder number axes:         Matrix vector multiplication:","code":"a <- op_arange(25) |> op_reshape(c(5, 5)) b <- op_arange(5) c <- op_arange(6) |> op_reshape(c(2, 3)) op_einsum(\"ii\", a) op_trace(a) ## tf.Tensor(65.0, shape=(), dtype=float32) ## tf.Tensor(65.0, shape=(), dtype=float32) op_einsum(\"ii -> i\", a) op_diag(a) ## tf.Tensor([ 1.  7. 13. 19. 25.], shape=(5), dtype=float32) ## tf.Tensor([ 1.  7. 13. 19. 25.], shape=(5), dtype=float32) op_einsum(\"ij -> i\", a) op_sum(a, axis = 2) ## tf.Tensor([ 15.  40.  65.  90. 115.], shape=(5), dtype=float32) ## tf.Tensor([ 15.  40.  65.  90. 115.], shape=(5), dtype=float32) op_einsum(\"...j -> ...\", a) op_sum(a, axis = -1) ## tf.Tensor([ 15.  40.  65.  90. 115.], shape=(5), dtype=float32) ## tf.Tensor([ 15.  40.  65.  90. 115.], shape=(5), dtype=float32) op_einsum(\"ji\", c) # return c unchanged ## tf.Tensor( ## [[1. 2. 3.] ##  [4. 5. 6.]], shape=(2, 3), dtype=float32) op_einsum(\"ij -> ji\", c) # transpose op_transpose(c)          # same as above ## tf.Tensor( ## [[1. 4.] ##  [2. 5.] ##  [3. 6.]], shape=(3, 2), dtype=float32) ## tf.Tensor( ## [[1. 4.] ##  [2. 5.] ##  [3. 6.]], shape=(3, 2), dtype=float32) op_einsum(\"ij, j\", a, b) op_einsum(\"...j, j\", a, b) a %*% b op_matmul(a, b) ## tf.Tensor([ 55. 130. 205. 280. 355.], shape=(5), dtype=float32) ## tf.Tensor([ 55. 130. 205. 280. 355.], shape=(5), dtype=float32) ## tf.Tensor([ 55. 130. 205. 280. 355.], shape=(5), dtype=float32) ## tf.Tensor([ 55. 130. 205. 280. 355.], shape=(5), dtype=float32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_elu.html","id":null,"dir":"Reference","previous_headings":"","what":"Exponential Linear Unit activation function. — op_elu","title":"Exponential Linear Unit activation function. — op_elu","text":"defined : f(x) =  alpha * (exp(x) - 1.) x < 0, f(x) = x x >= 0.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_elu.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Exponential Linear Unit activation function. — op_elu","text":"","code":"op_elu(x, alpha = 1)"},{"path":"https://keras3.posit.co/dev/reference/op_elu.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Exponential Linear Unit activation function. — op_elu","text":"x Input tensor. alpha scalar, slope positive section. Defaults 1.0.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_elu.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Exponential Linear Unit activation function. — op_elu","text":"tensor shape x.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_elu.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Exponential Linear Unit activation function. — op_elu","text":"","code":"x <- op_array(c(-1., 0., 1.)) op_elu(x) ## tf.Tensor([-0.63212055  0.          1.        ], shape=(3), dtype=float32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_empty.html","id":null,"dir":"Reference","previous_headings":"","what":"Return a tensor of given shape and type filled with uninitialized data. — op_empty","title":"Return a tensor of given shape and type filled with uninitialized data. — op_empty","text":"Return tensor given shape type filled uninitialized data.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_empty.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Return a tensor of given shape and type filled with uninitialized data. — op_empty","text":"","code":"op_empty(shape, dtype = NULL)"},{"path":"https://keras3.posit.co/dev/reference/op_empty.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Return a tensor of given shape and type filled with uninitialized data. — op_empty","text":"shape Shape empty tensor. dtype Desired data type empty tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_empty.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Return a tensor of given shape and type filled with uninitialized data. — op_empty","text":"empty tensor.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_equal.html","id":null,"dir":"Reference","previous_headings":"","what":"Returns (x1 == x2) element-wise. — op_equal","title":"Returns (x1 == x2) element-wise. — op_equal","text":"Note function automatically called using R operator == tensor.","code":"(x <- op_arange(4)) ## tf.Tensor([1. 2. 3. 4.], shape=(4), dtype=float32) op_equal(x, 2) ## tf.Tensor([False  True False False], shape=(4), dtype=bool) x == 2 ## tf.Tensor([False  True False False], shape=(4), dtype=bool)"},{"path":"https://keras3.posit.co/dev/reference/op_equal.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Returns (x1 == x2) element-wise. — op_equal","text":"","code":"op_equal(x1, x2)"},{"path":"https://keras3.posit.co/dev/reference/op_equal.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Returns (x1 == x2) element-wise. — op_equal","text":"x1 Tensor compare. x2 Tensor compare.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_equal.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Returns (x1 == x2) element-wise. — op_equal","text":"Output tensor, element-wise comparison x1 x2.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_erf.html","id":null,"dir":"Reference","previous_headings":"","what":"Computes the error function of x, element-wise. — op_erf","title":"Computes the error function of x, element-wise. — op_erf","text":"Computes error function x, element-wise.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_erf.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes the error function of x, element-wise. — op_erf","text":"","code":"op_erf(x)"},{"path":"https://keras3.posit.co/dev/reference/op_erf.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Computes the error function of x, element-wise. — op_erf","text":"x Input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_erf.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Computes the error function of x, element-wise. — op_erf","text":"tensor dtype x.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_erf.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Computes the error function of x, element-wise. — op_erf","text":"","code":"x <- op_array(c(-3, -2, -1, 0, 1)) op_erf(x) ## tf.Tensor([-0.99997777 -0.9953222  -0.84270084  0.          0.84270084], shape=(5), dtype=float32) # array([-0.99998 , -0.99532, -0.842701,  0.,  0.842701], dtype=float32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_erfinv.html","id":null,"dir":"Reference","previous_headings":"","what":"Computes the inverse error function of x, element-wise. — op_erfinv","title":"Computes the inverse error function of x, element-wise. — op_erfinv","text":"Computes inverse error function x, element-wise.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_erfinv.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes the inverse error function of x, element-wise. — op_erfinv","text":"","code":"op_erfinv(x)"},{"path":"https://keras3.posit.co/dev/reference/op_erfinv.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Computes the inverse error function of x, element-wise. — op_erfinv","text":"x Input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_erfinv.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Computes the inverse error function of x, element-wise. — op_erfinv","text":"tensor dtype x.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_erfinv.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Computes the inverse error function of x, element-wise. — op_erfinv","text":"","code":"x <- op_array(c(-0.5, -0.2, -0.1, 0.0, 0.3)) op_erfinv(x) ## tf.Tensor([-0.4769363  -0.17914344 -0.088856    0.          0.27246267], shape=(5), dtype=float32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_exp.html","id":null,"dir":"Reference","previous_headings":"","what":"Calculate the exponential of all elements in the input tensor. — op_exp","title":"Calculate the exponential of all elements in the input tensor. — op_exp","text":"Calculate exponential elements input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_exp.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Calculate the exponential of all elements in the input tensor. — op_exp","text":"","code":"op_exp(x)"},{"path":"https://keras3.posit.co/dev/reference/op_exp.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Calculate the exponential of all elements in the input tensor. — op_exp","text":"x Input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_exp.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Calculate the exponential of all elements in the input tensor. — op_exp","text":"Output tensor, element-wise exponential x.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_exp2.html","id":null,"dir":"Reference","previous_headings":"","what":"Calculate the base-2 exponential of all elements in the input tensor. — op_exp2","title":"Calculate the base-2 exponential of all elements in the input tensor. — op_exp2","text":"Calculate base-2 exponential elements input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_exp2.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Calculate the base-2 exponential of all elements in the input tensor. — op_exp2","text":"","code":"op_exp2(x)"},{"path":"https://keras3.posit.co/dev/reference/op_exp2.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Calculate the base-2 exponential of all elements in the input tensor. — op_exp2","text":"x Input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_exp2.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Calculate the base-2 exponential of all elements in the input tensor. — op_exp2","text":"Output tensor, element-wise base-2 exponential x.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_expand_dims.html","id":null,"dir":"Reference","previous_headings":"","what":"Expand the shape of a tensor. — op_expand_dims","title":"Expand the shape of a tensor. — op_expand_dims","text":"Insert new axis axis position expanded tensor shape.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_expand_dims.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Expand the shape of a tensor. — op_expand_dims","text":"","code":"op_expand_dims(x, axis)"},{"path":"https://keras3.posit.co/dev/reference/op_expand_dims.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Expand the shape of a tensor. — op_expand_dims","text":"x Input tensor. axis Position expanded axes new axis (axes) placed.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_expand_dims.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Expand the shape of a tensor. — op_expand_dims","text":"Output tensor number dimensions increased.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_expm1.html","id":null,"dir":"Reference","previous_headings":"","what":"Calculate exp(x) - 1 for all elements in the tensor. — op_expm1","title":"Calculate exp(x) - 1 for all elements in the tensor. — op_expm1","text":"Calculate exp(x) - 1 elements tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_expm1.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Calculate exp(x) - 1 for all elements in the tensor. — op_expm1","text":"","code":"op_expm1(x)"},{"path":"https://keras3.posit.co/dev/reference/op_expm1.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Calculate exp(x) - 1 for all elements in the tensor. — op_expm1","text":"x Input values.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_expm1.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Calculate exp(x) - 1 for all elements in the tensor. — op_expm1","text":"Output tensor, element-wise exponential minus one.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_extract_sequences.html","id":null,"dir":"Reference","previous_headings":"","what":"Expands the dimension of last axis into sequences of sequence_length. — op_extract_sequences","title":"Expands the dimension of last axis into sequences of sequence_length. — op_extract_sequences","text":"Slides window size sequence_length last axis input stride sequence_stride, replacing last axis [num_sequences, sequence_length] sequences. dimension along last axis N, number sequences can computed : num_sequences = 1 + (N - sequence_length) // sequence_stride","code":""},{"path":"https://keras3.posit.co/dev/reference/op_extract_sequences.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Expands the dimension of last axis into sequences of sequence_length. — op_extract_sequences","text":"","code":"op_extract_sequences(x, sequence_length, sequence_stride)"},{"path":"https://keras3.posit.co/dev/reference/op_extract_sequences.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Expands the dimension of last axis into sequences of sequence_length. — op_extract_sequences","text":"x Input tensor. sequence_length integer representing sequences length. sequence_stride integer representing sequences hop size.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_extract_sequences.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Expands the dimension of last axis into sequences of sequence_length. — op_extract_sequences","text":"tensor sequences shape [..., num_sequences, sequence_length].","code":""},{"path":"https://keras3.posit.co/dev/reference/op_extract_sequences.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Expands the dimension of last axis into sequences of sequence_length. — op_extract_sequences","text":"","code":"x <- op_convert_to_tensor(1:6) op_extract_sequences(x, 3, 2) ## tf.Tensor( ## [[1 2 3] ##  [3 4 5]], shape=(2, 3), dtype=int32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_eye.html","id":null,"dir":"Reference","previous_headings":"","what":"Return a 2-D tensor with ones on the diagonal and zeros elsewhere. — op_eye","title":"Return a 2-D tensor with ones on the diagonal and zeros elsewhere. — op_eye","text":"Return 2-D tensor ones diagonal zeros elsewhere.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_eye.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Return a 2-D tensor with ones on the diagonal and zeros elsewhere. — op_eye","text":"","code":"op_eye(N, M = NULL, k = 0L, dtype = NULL)"},{"path":"https://keras3.posit.co/dev/reference/op_eye.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Return a 2-D tensor with ones on the diagonal and zeros elsewhere. — op_eye","text":"N Number rows output. M Number columns output. NULL, defaults N. k Index diagonal: 0 (default) refers main diagonal, positive value refers upper diagonal, negative value lower diagonal. dtype Data type returned tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_eye.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Return a 2-D tensor with ones on the diagonal and zeros elsewhere. — op_eye","text":"Tensor ones k-th diagonal zeros elsewhere.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_fft.html","id":null,"dir":"Reference","previous_headings":"","what":"Computes the Fast Fourier Transform along last axis of input. — op_fft","title":"Computes the Fast Fourier Transform along last axis of input. — op_fft","text":"Computes Fast Fourier Transform along last axis input.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_fft.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes the Fast Fourier Transform along last axis of input. — op_fft","text":"","code":"op_fft(x)"},{"path":"https://keras3.posit.co/dev/reference/op_fft.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Computes the Fast Fourier Transform along last axis of input. — op_fft","text":"x list real imaginary parts input tensor. tensors provided floating type.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_fft.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Computes the Fast Fourier Transform along last axis of input. — op_fft","text":"list containing two tensors - real imaginary parts output tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_fft.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Computes the Fast Fourier Transform along last axis of input. — op_fft","text":"","code":"x = c(op_array(c(1., 2.)),       op_array(c(0., 1.))) op_fft(x) ## [[1]] ## tf.Tensor([ 3. -1.], shape=(2), dtype=float32) ## ## [[2]] ## tf.Tensor([ 1. -1.], shape=(2), dtype=float32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_fft2.html","id":null,"dir":"Reference","previous_headings":"","what":"Computes the 2D Fast Fourier Transform along the last two axes of input. — op_fft2","title":"Computes the 2D Fast Fourier Transform along the last two axes of input. — op_fft2","text":"Computes 2D Fast Fourier Transform along last two axes input.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_fft2.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes the 2D Fast Fourier Transform along the last two axes of input. — op_fft2","text":"","code":"op_fft2(x)"},{"path":"https://keras3.posit.co/dev/reference/op_fft2.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Computes the 2D Fast Fourier Transform along the last two axes of input. — op_fft2","text":"x list real imaginary parts input tensor. tensors provided floating type.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_fft2.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Computes the 2D Fast Fourier Transform along the last two axes of input. — op_fft2","text":"list containing two tensors - real imaginary parts output.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_fft2.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Computes the 2D Fast Fourier Transform along the last two axes of input. — op_fft2","text":"","code":"x <- c(op_array(rbind(c(1, 2),                      c(2, 1))),        op_array(rbind(c(0, 1),                      c(1, 0)))) op_fft2(x) ## [[1]] ## tf.Tensor( ## [[ 6.  0.] ##  [ 0. -2.]], shape=(2, 2), dtype=float64) ## ## [[2]] ## tf.Tensor( ## [[ 2.  0.] ##  [ 0. -2.]], shape=(2, 2), dtype=float64)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_flip.html","id":null,"dir":"Reference","previous_headings":"","what":"Reverse the order of elements in the tensor along the given axis. — op_flip","title":"Reverse the order of elements in the tensor along the given axis. — op_flip","text":"shape tensor preserved, elements reordered.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_flip.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Reverse the order of elements in the tensor along the given axis. — op_flip","text":"","code":"op_flip(x, axis = NULL)"},{"path":"https://keras3.posit.co/dev/reference/op_flip.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Reverse the order of elements in the tensor along the given axis. — op_flip","text":"x Input tensor. axis Axis axes along flip tensor. default, axis = NULL, flip axes input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_flip.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Reverse the order of elements in the tensor along the given axis. — op_flip","text":"Output tensor entries axis reversed.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_floor.html","id":null,"dir":"Reference","previous_headings":"","what":"Return the floor of the input, element-wise. — op_floor","title":"Return the floor of the input, element-wise. — op_floor","text":"floor scalar x largest integer , <= x.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_floor.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Return the floor of the input, element-wise. — op_floor","text":"","code":"op_floor(x)"},{"path":"https://keras3.posit.co/dev/reference/op_floor.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Return the floor of the input, element-wise. — op_floor","text":"x Input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_floor.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Return the floor of the input, element-wise. — op_floor","text":"Output tensor, element-wise floor x.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_floor_divide.html","id":null,"dir":"Reference","previous_headings":"","what":"Returns the largest integer smaller or equal to the division of inputs. — op_floor_divide","title":"Returns the largest integer smaller or equal to the division of inputs. — op_floor_divide","text":"Note function automatically called using R operator %/% tensor.","code":"(x <- op_arange(10)) ## tf.Tensor([ 1.  2.  3.  4.  5.  6.  7.  8.  9. 10.], shape=(10), dtype=float32) op_floor_divide(x, 2) ## tf.Tensor([0. 1. 1. 2. 2. 3. 3. 4. 4. 5.], shape=(10), dtype=float32) x %/% 2 ## tf.Tensor([0. 1. 1. 2. 2. 3. 3. 4. 4. 5.], shape=(10), dtype=float32)"},{"path":"https://keras3.posit.co/dev/reference/op_floor_divide.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Returns the largest integer smaller or equal to the division of inputs. — op_floor_divide","text":"","code":"op_floor_divide(x1, x2)"},{"path":"https://keras3.posit.co/dev/reference/op_floor_divide.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Returns the largest integer smaller or equal to the division of inputs. — op_floor_divide","text":"x1 Numerator. x2 Denominator.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_floor_divide.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Returns the largest integer smaller or equal to the division of inputs. — op_floor_divide","text":"Output tensor, y <- floor(x1/x2)","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_fori_loop.html","id":null,"dir":"Reference","previous_headings":"","what":"For loop implementation. — op_fori_loop","title":"For loop implementation. — op_fori_loop","text":"loop implementation.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_fori_loop.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"For loop implementation. — op_fori_loop","text":"","code":"op_fori_loop(lower, upper, body_fun, init_val)"},{"path":"https://keras3.posit.co/dev/reference/op_fori_loop.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"For loop implementation. — op_fori_loop","text":"lower initial value loop variable. upper upper bound loop variable. body_fun callable represents loop body. Must take two arguments: loop variable loop state. loop state updated returned function. init_val initial value loop state.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_fori_loop.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"For loop implementation. — op_fori_loop","text":"final state loop.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_fori_loop.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"For loop implementation. — op_fori_loop","text":"","code":"lower <- 0L upper <- 10L body_fun <- function(i, state) state + i init_state <- 0L final_state <- op_fori_loop(lower, upper, body_fun, init_state) final_state ## tf.Tensor(45, shape=(), dtype=int32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_full.html","id":null,"dir":"Reference","previous_headings":"","what":"Return a new tensor of given shape and type, filled with fill_value. — op_full","title":"Return a new tensor of given shape and type, filled with fill_value. — op_full","text":"Return new tensor given shape type, filled fill_value.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_full.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Return a new tensor of given shape and type, filled with fill_value. — op_full","text":"","code":"op_full(shape, fill_value, dtype = NULL)"},{"path":"https://keras3.posit.co/dev/reference/op_full.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Return a new tensor of given shape and type, filled with fill_value. — op_full","text":"shape Shape new tensor. fill_value Fill value. dtype Desired data type tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_full.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Return a new tensor of given shape and type, filled with fill_value. — op_full","text":"Output tensor.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_full_like.html","id":null,"dir":"Reference","previous_headings":"","what":"Return a full tensor with the same shape and type as the given tensor. — op_full_like","title":"Return a full tensor with the same shape and type as the given tensor. — op_full_like","text":"Return full tensor shape type given tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_full_like.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Return a full tensor with the same shape and type as the given tensor. — op_full_like","text":"","code":"op_full_like(x, fill_value, dtype = NULL)"},{"path":"https://keras3.posit.co/dev/reference/op_full_like.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Return a full tensor with the same shape and type as the given tensor. — op_full_like","text":"x Input tensor. fill_value Fill value. dtype Overrides data type result.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_full_like.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Return a full tensor with the same shape and type as the given tensor. — op_full_like","text":"Tensor fill_value shape type x.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_gelu.html","id":null,"dir":"Reference","previous_headings":"","what":"Gaussian Error Linear Unit (GELU) activation function. — op_gelu","title":"Gaussian Error Linear Unit (GELU) activation function. — op_gelu","text":"approximate TRUE, defined : f(x) = 0.5 * x * (1 + tanh(sqrt(2 / pi) * (x + 0.044715 * x^3))) approximate FALSE, defined : f(x) = x * P(X <= x) = 0.5 * x * (1 + erf(x / sqrt(2))), P(X) ~ N(0, 1).","code":""},{"path":"https://keras3.posit.co/dev/reference/op_gelu.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Gaussian Error Linear Unit (GELU) activation function. — op_gelu","text":"","code":"op_gelu(x, approximate = TRUE)"},{"path":"https://keras3.posit.co/dev/reference/op_gelu.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Gaussian Error Linear Unit (GELU) activation function. — op_gelu","text":"x Input tensor. approximate Approximate version GELU activation. Defaults TRUE.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_gelu.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Gaussian Error Linear Unit (GELU) activation function. — op_gelu","text":"tensor shape x.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_gelu.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Gaussian Error Linear Unit (GELU) activation function. — op_gelu","text":"","code":"x <- op_array(c(-1., 0., 1.)) op_gelu(x) ## tf.Tensor([-0.15880796  0.          0.841192  ], shape=(3), dtype=float32) op_gelu(x, FALSE) ## tf.Tensor([-0.15865526  0.          0.8413447 ], shape=(3), dtype=float32) x <- seq(-5, 5, .1) plot(x, op_gelu(x),      type = \"l\", #, frame.plot = FALSE,      panel.first = grid())"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_get_item.html","id":null,"dir":"Reference","previous_headings":"","what":"Return x[key]. — op_get_item","title":"Return x[key]. — op_get_item","text":"Return x[key].","code":""},{"path":"https://keras3.posit.co/dev/reference/op_get_item.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Return x[key]. — op_get_item","text":"","code":"op_get_item(x, key)"},{"path":"https://keras3.posit.co/dev/reference/op_get_item.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Return x[key]. — op_get_item","text":"x dictionary-like object key Generally, string, object __hash__ method acceptable.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_get_item.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Return x[key]. — op_get_item","text":"key.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_get_item.html","id":"note","dir":"Reference","previous_headings":"","what":"Note","title":"Return x[key]. — op_get_item","text":"Generally, calling x[[key]] x$key preferable.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_glu.html","id":null,"dir":"Reference","previous_headings":"","what":"Gated Linear Unit (GLU) activation function. — op_glu","title":"Gated Linear Unit (GLU) activation function. — op_glu","text":"defined : f(x) = * sigmoid(b) x split b along given axis.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_glu.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Gated Linear Unit (GLU) activation function. — op_glu","text":"","code":"op_glu(x, axis = -1L)"},{"path":"https://keras3.posit.co/dev/reference/op_glu.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Gated Linear Unit (GLU) activation function. — op_glu","text":"x Input tensor. axis axis along split input tensor. Defaults -1.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_glu.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Gated Linear Unit (GLU) activation function. — op_glu","text":"tensor shape half input.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_glu.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Gated Linear Unit (GLU) activation function. — op_glu","text":"","code":"x <- op_array(c(-1., 0., 1. , 1.)) op_glu(x) ## tf.Tensor([-0.7310586  0.       ], shape=(2), dtype=float32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_greater.html","id":null,"dir":"Reference","previous_headings":"","what":"Return the truth value of x1 > x2 element-wise. — op_greater","title":"Return the truth value of x1 > x2 element-wise. — op_greater","text":"Note function automatically called using R operator > tensor.","code":"(x <- op_arange(4)) ## tf.Tensor([1. 2. 3. 4.], shape=(4), dtype=float32) op_greater(x, 2) ## tf.Tensor([False False  True  True], shape=(4), dtype=bool) x > 2 ## tf.Tensor([False False  True  True], shape=(4), dtype=bool)"},{"path":"https://keras3.posit.co/dev/reference/op_greater.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Return the truth value of x1 > x2 element-wise. — op_greater","text":"","code":"op_greater(x1, x2)"},{"path":"https://keras3.posit.co/dev/reference/op_greater.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Return the truth value of x1 > x2 element-wise. — op_greater","text":"x1 First input tensor. x2 Second input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_greater.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Return the truth value of x1 > x2 element-wise. — op_greater","text":"Output tensor, element-wise comparison x1 x2.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_greater_equal.html","id":null,"dir":"Reference","previous_headings":"","what":"Return the truth value of x1 >= x2 element-wise. — op_greater_equal","title":"Return the truth value of x1 >= x2 element-wise. — op_greater_equal","text":"Note function automatically called using R operator >= tensor.","code":"(x <- op_arange(4)) ## tf.Tensor([1. 2. 3. 4.], shape=(4), dtype=float32) op_greater_equal(x, 2) ## tf.Tensor([False  True  True  True], shape=(4), dtype=bool) x >= 2 ## tf.Tensor([False  True  True  True], shape=(4), dtype=bool)"},{"path":"https://keras3.posit.co/dev/reference/op_greater_equal.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Return the truth value of x1 >= x2 element-wise. — op_greater_equal","text":"","code":"op_greater_equal(x1, x2)"},{"path":"https://keras3.posit.co/dev/reference/op_greater_equal.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Return the truth value of x1 >= x2 element-wise. — op_greater_equal","text":"x1 First input tensor. x2 Second input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_greater_equal.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Return the truth value of x1 >= x2 element-wise. — op_greater_equal","text":"Output tensor, element-wise comparison x1 x2.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_hard_shrink.html","id":null,"dir":"Reference","previous_headings":"","what":"Hard Shrink activation function. — op_hard_shrink","title":"Hard Shrink activation function. — op_hard_shrink","text":"Hard Shrink function thresholding operation defined : f(x) = x |x| > threshold, f(x) = 0 otherwise.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_hard_shrink.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Hard Shrink activation function. — op_hard_shrink","text":"","code":"op_hard_shrink(x, threshold = 0.5)"},{"path":"https://keras3.posit.co/dev/reference/op_hard_shrink.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Hard Shrink activation function. — op_hard_shrink","text":"x Input tensor. threshold Threshold value. Defaults 0.5.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_hard_shrink.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Hard Shrink activation function. — op_hard_shrink","text":"tensor shape x.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_hard_shrink.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Hard Shrink activation function. — op_hard_shrink","text":"","code":"x <- op_array(c(-0.5, 0., 1.)) op_hard_shrink(x) ## tf.Tensor([0. 0. 1.], shape=(3), dtype=float32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_hard_sigmoid.html","id":null,"dir":"Reference","previous_headings":"","what":"Hard sigmoid activation function. — op_hard_sigmoid","title":"Hard sigmoid activation function. — op_hard_sigmoid","text":"defined : 0 x < -2.5, 1 x > 2.5, (0.2 * x) + 0.5 -2.5 <= x <= 2.5.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_hard_sigmoid.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Hard sigmoid activation function. — op_hard_sigmoid","text":"","code":"op_hard_sigmoid(x)"},{"path":"https://keras3.posit.co/dev/reference/op_hard_sigmoid.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Hard sigmoid activation function. — op_hard_sigmoid","text":"x Input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_hard_sigmoid.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Hard sigmoid activation function. — op_hard_sigmoid","text":"tensor shape x.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_hard_sigmoid.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Hard sigmoid activation function. — op_hard_sigmoid","text":"","code":"x <- op_array(c(-1., 0., 1.)) op_hard_sigmoid(x) ## tf.Tensor([0.33333334 0.5        0.6666667 ], shape=(3), dtype=float32) x <- as.array(seq(-5, 5, .1)) plot(x, op_hard_sigmoid(x),      type = 'l', panel.first = grid(), frame.plot = FALSE)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_hard_silu.html","id":null,"dir":"Reference","previous_headings":"","what":"Hard SiLU activation function, also known as Hard Swish. — op_hard_silu","title":"Hard SiLU activation function, also known as Hard Swish. — op_hard_silu","text":"defined : 0 x < -3 x x > 3 x * (x + 3) / 6 -3 <= x <= 3 faster, piecewise linear approximation silu activation.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_hard_silu.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Hard SiLU activation function, also known as Hard Swish. — op_hard_silu","text":"","code":"op_hard_silu(x)  op_hard_swish(x)"},{"path":"https://keras3.posit.co/dev/reference/op_hard_silu.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Hard SiLU activation function, also known as Hard Swish. — op_hard_silu","text":"x Input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_hard_silu.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Hard SiLU activation function, also known as Hard Swish. — op_hard_silu","text":"tensor shape x.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_hard_silu.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Hard SiLU activation function, also known as Hard Swish. — op_hard_silu","text":"","code":"x <- op_convert_to_tensor(c(-3.0, -1.0, 0.0, 1.0, 3.0)) op_hard_silu(x) ## tf.Tensor([-0.         -0.33333333  0.          0.66666667  3.        ], shape=(5), dtype=float64)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_hard_tanh.html","id":null,"dir":"Reference","previous_headings":"","what":"Applies the HardTanh function element-wise. — op_hard_tanh","title":"Applies the HardTanh function element-wise. — op_hard_tanh","text":"defined : f(x) = -1 x < -1, f(x) = x -1 <= x <= 1, f(x) = 1 x > 1.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_hard_tanh.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Applies the HardTanh function element-wise. — op_hard_tanh","text":"","code":"op_hard_tanh(x)"},{"path":"https://keras3.posit.co/dev/reference/op_hard_tanh.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Applies the HardTanh function element-wise. — op_hard_tanh","text":"x Input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_hard_tanh.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Applies the HardTanh function element-wise. — op_hard_tanh","text":"Output tensor shape x values clamped -1 1.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_hard_tanh.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Applies the HardTanh function element-wise. — op_hard_tanh","text":"","code":"x <- op_array(c(-2., -1., 0., 1., 2.)) op_hard_tanh(x) ## tf.Tensor([-1. -1.  0.  1.  1.], shape=(5), dtype=float32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_histogram.html","id":null,"dir":"Reference","previous_headings":"","what":"Computes a histogram of the data tensor x. — op_histogram","title":"Computes a histogram of the data tensor x. — op_histogram","text":"Computes histogram data tensor x.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_histogram.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes a histogram of the data tensor x. — op_histogram","text":"","code":"op_histogram(x, bins = 10L, range = NULL)"},{"path":"https://keras3.posit.co/dev/reference/op_histogram.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Computes a histogram of the data tensor x. — op_histogram","text":"x Input tensor. bins integer representing number histogram bins. Defaults 10. range pair numbers representing lower upper range bins. specified, use min max x.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_histogram.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Computes a histogram of the data tensor x. — op_histogram","text":"list two tensors containing: tensor representing counts elements bin. tensor representing bin edges.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_histogram.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Computes a histogram of the data tensor x. — op_histogram","text":"","code":"input_tensor <- random_uniform(8) c(counts, edges) %<-% op_histogram(input_tensor)  counts #> tf.Tensor([2. 1. 1. 1. 0. 1. 0. 0. 0. 2.], shape=(10), dtype=float32) edges #> tf.Tensor( #> [0.01647806 0.10867032 0.20086257 0.29305482 0.38524708 0.47743934 #>  0.5696316  0.66182387 0.7540161  0.84620833 0.9384006 ], shape=(11), dtype=float32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_hstack.html","id":null,"dir":"Reference","previous_headings":"","what":"Stack tensors in sequence horizontally (column wise). — op_hstack","title":"Stack tensors in sequence horizontally (column wise). — op_hstack","text":"equivalent concatenation along first axis 1-D tensors, along second axis tensors.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_hstack.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Stack tensors in sequence horizontally (column wise). — op_hstack","text":"","code":"op_hstack(xs, ...)"},{"path":"https://keras3.posit.co/dev/reference/op_hstack.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Stack tensors in sequence horizontally (column wise). — op_hstack","text":"xs, ... list tensors.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_hstack.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Stack tensors in sequence horizontally (column wise). — op_hstack","text":"tensor formed stacking given tensors.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_identity.html","id":null,"dir":"Reference","previous_headings":"","what":"Return the identity tensor. — op_identity","title":"Return the identity tensor. — op_identity","text":"identity tensor square tensor ones main diagonal zeros elsewhere.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_identity.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Return the identity tensor. — op_identity","text":"","code":"op_identity(n, dtype = NULL)"},{"path":"https://keras3.posit.co/dev/reference/op_identity.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Return the identity tensor. — op_identity","text":"n Number rows (columns) n x n output tensor. dtype Data type output tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_identity.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Return the identity tensor. — op_identity","text":"identity tensor.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_ifft2.html","id":null,"dir":"Reference","previous_headings":"","what":"Computes the 2D Inverse Fast Fourier Transform along the last two axes of — op_ifft2","title":"Computes the 2D Inverse Fast Fourier Transform along the last two axes of — op_ifft2","text":"input.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_ifft2.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes the 2D Inverse Fast Fourier Transform along the last two axes of — op_ifft2","text":"","code":"op_ifft2(x)"},{"path":"https://keras3.posit.co/dev/reference/op_ifft2.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Computes the 2D Inverse Fast Fourier Transform along the last two axes of — op_ifft2","text":"x Tuple real imaginary parts input tensor. tensors tuple floating type.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_ifft2.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Computes the 2D Inverse Fast Fourier Transform along the last two axes of — op_ifft2","text":"tuple containing two tensors - real imaginary parts output.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_ifft2.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Computes the 2D Inverse Fast Fourier Transform along the last two axes of — op_ifft2","text":"","code":"x <- list(op_array(rbind(c(1, 2), c(2, 1))),           op_array(rbind(c(0, 1), c(1, 0)))) op_ifft2(x) ## [[1]] ## tf.Tensor( ## [[ 1.5  0. ] ##  [ 0.  -0.5]], shape=(2, 2), dtype=float64) ## ## [[2]] ## tf.Tensor( ## [[ 0.5 -0. ] ##  [-0.  -0.5]], shape=(2, 2), dtype=float64)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_imag.html","id":null,"dir":"Reference","previous_headings":"","what":"Return the imaginary part of the complex argument. — op_imag","title":"Return the imaginary part of the complex argument. — op_imag","text":"Return imaginary part complex argument.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_imag.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Return the imaginary part of the complex argument. — op_imag","text":"","code":"op_imag(x)"},{"path":"https://keras3.posit.co/dev/reference/op_imag.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Return the imaginary part of the complex argument. — op_imag","text":"x Input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_imag.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Return the imaginary part of the complex argument. — op_imag","text":"imaginary component complex argument.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_image_affine_transform.html","id":null,"dir":"Reference","previous_headings":"","what":"Applies the given transform(s) to the image(s). — op_image_affine_transform","title":"Applies the given transform(s) to the image(s). — op_image_affine_transform","text":"Applies given transform(s) image(s).","code":""},{"path":"https://keras3.posit.co/dev/reference/op_image_affine_transform.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Applies the given transform(s) to the image(s). — op_image_affine_transform","text":"","code":"op_image_affine_transform(   images,   transform,   interpolation = \"bilinear\",   fill_mode = \"constant\",   fill_value = 0L,   data_format = NULL )"},{"path":"https://keras3.posit.co/dev/reference/op_image_affine_transform.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Applies the given transform(s) to the image(s). — op_image_affine_transform","text":"images Input image batch images. Must 3D 4D. transform Projective transform matrix/matrices. vector length 8 tensor size N x 8. one row transform [a0, a1, a2, b0, b1, b2, c0, c1], maps output point (x, y) transformed input point (x', y') = ((a0 x + a1 y + a2) / k, (b0 x + b1 y + b2) / k), k = c0 x + c1 y + 1. transform inverted compared transform mapping input points output points. Note gradients backpropagated transformation parameters. Note c0 c1 effective using TensorFlow backend considered 0 using backends. interpolation Interpolation method. Available methods \"nearest\", \"bilinear\". Defaults \"bilinear\". fill_mode Points outside boundaries input filled according given mode. Available methods \"constant\", \"nearest\", \"wrap\" \"reflect\". Defaults \"constant\". \"reflect\": (d c b | b c d | d c b ) input extended reflecting edge last pixel. \"constant\": (k k k k | b c d | k k k k) input extended filling values beyond edge constant value k specified fill_value. \"wrap\": (b c d | b c d | b c d) input extended wrapping around opposite edge. \"nearest\": (| b c d | d d d d) input extended nearest pixel. fill_value Value used points outside boundaries input fill_mode = \"constant\". Defaults 0. data_format string specifying data format input tensor. can either \"channels_last\" \"channels_first\". \"channels_last\" corresponds inputs shape (batch, height, width, channels), \"channels_first\" corresponds inputs shape (batch, channels, height, width). specified, value default config_image_data_format().","code":""},{"path":"https://keras3.posit.co/dev/reference/op_image_affine_transform.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Applies the given transform(s) to the image(s). — op_image_affine_transform","text":"Applied affine transform image batch images.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_image_affine_transform.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Applies the given transform(s) to the image(s). — op_image_affine_transform","text":"","code":"x <- random_uniform(c(2, 64, 80, 3)) # batch of 2 RGB images transform <- op_array(rbind(c(1.5, 0, -20, 0, 1.5, -16, 0, 0),  # zoom                            c(1, 0, -20, 0, 1, -16, 0, 0)))  # translation)) y <- op_image_affine_transform(x, transform) shape(y) ## shape(2, 64, 80, 3) # (2, 64, 80, 3) x <- random_uniform(c(64, 80, 3)) # single RGB image transform <- op_array(c(1.0, 0.5, -20, 0.5, 1.0, -16, 0, 0))  # shear y <- op_image_affine_transform(x, transform) shape(y) ## shape(64, 80, 3) # (64, 80, 3) x <- random_uniform(c(2, 3, 64, 80)) # batch of 2 RGB images transform <- op_array(rbind(   c(1.5, 0,-20, 0, 1.5,-16, 0, 0),  # zoom   c(1, 0,-20, 0, 1,-16, 0, 0)  # translation )) y <- op_image_affine_transform(x, transform, data_format = \"channels_first\") shape(y) ## shape(2, 3, 64, 80) # (2, 3, 64, 80)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_image_crop.html","id":null,"dir":"Reference","previous_headings":"","what":"Crop images to a specified height and width. — op_image_crop","title":"Crop images to a specified height and width. — op_image_crop","text":"Crop images specified height width.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_image_crop.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Crop images to a specified height and width. — op_image_crop","text":"","code":"op_image_crop(   images,   top_cropping = NULL,   left_cropping = NULL,   bottom_cropping = NULL,   right_cropping = NULL,   target_height = NULL,   target_width = NULL,   data_format = NULL )"},{"path":"https://keras3.posit.co/dev/reference/op_image_crop.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Crop images to a specified height and width. — op_image_crop","text":"images Input image batch images. Must 3D 4D. top_cropping Number columns crop top. left_cropping Number columns crop left. bottom_cropping Number columns crop bottom. right_cropping Number columns crop right. target_height Height output images. target_width Width output images. data_format string specifying data format input tensor. can either \"channels_last\" \"channels_first\". \"channels_last\" corresponds inputs shape (batch, height, width, channels), \"channels_first\" corresponds inputs shape (batch, channels, height, width). specified, value default config_image_data_format().","code":""},{"path":"https://keras3.posit.co/dev/reference/op_image_crop.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Crop images to a specified height and width. — op_image_crop","text":"Cropped image batch images.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_image_crop.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Crop images to a specified height and width. — op_image_crop","text":"","code":"images <- op_reshape(op_arange(1, 28, dtype=\"float32\"), c(3, 3, 3)) images[, , 1] # print the first channel of the images  cropped_images <- op_image_crop(images, 0, 0, 2, 2) cropped_images[, , 1] # print the first channel of the cropped images"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_image_extract_patches.html","id":null,"dir":"Reference","previous_headings":"","what":"Extracts patches from the image(s). — op_image_extract_patches","title":"Extracts patches from the image(s). — op_image_extract_patches","text":"Extracts patches image(s).","code":""},{"path":"https://keras3.posit.co/dev/reference/op_image_extract_patches.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Extracts patches from the image(s). — op_image_extract_patches","text":"","code":"op_image_extract_patches(   images,   size,   strides = NULL,   dilation_rate = 1L,   padding = \"valid\",   data_format = \"channels_last\" )"},{"path":"https://keras3.posit.co/dev/reference/op_image_extract_patches.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Extracts patches from the image(s). — op_image_extract_patches","text":"images Input image batch images. Must 3D 4D. size Patch size int list (patch_height, patch_width) strides strides along height width. specified, NULL, defaults value size. dilation_rate input stride, specifying far two consecutive patch samples input. value 1, strides must 1. NOTE: strides > 1 supported conjunction dilation_rate > 1 padding type padding algorithm use: \"\" \"valid\". data_format string specifying data format input tensor. can either \"channels_last\" \"channels_first\". \"channels_last\" corresponds inputs shape (batch, height, width, channels), \"channels_first\" corresponds inputs shape (batch, channels, height, width). specified, value default config_image_data_format().","code":""},{"path":"https://keras3.posit.co/dev/reference/op_image_extract_patches.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Extracts patches from the image(s). — op_image_extract_patches","text":"Extracted patches 3D (batched) 4D (batched)","code":""},{"path":"https://keras3.posit.co/dev/reference/op_image_extract_patches.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Extracts patches from the image(s). — op_image_extract_patches","text":"","code":"image <- random_uniform(c(2, 20, 20, 3), dtype = \"float32\") # batch of 2 RGB images patches <- op_image_extract_patches(image, c(5, 5)) shape(patches) ## shape(2, 4, 4, 75) # (2, 4, 4, 75) image <- random_uniform(c(20, 20, 3), dtype = \"float32\") # 1 RGB image patches <- op_image_extract_patches(image, c(3, 3), c(1, 1)) shape(patches) ## shape(18, 18, 27) # (18, 18, 27)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_image_gaussian_blur.html","id":null,"dir":"Reference","previous_headings":"","what":"Applies a Gaussian blur to the image(s). — op_image_gaussian_blur","title":"Applies a Gaussian blur to the image(s). — op_image_gaussian_blur","text":"Applies Gaussian blur image(s).","code":""},{"path":"https://keras3.posit.co/dev/reference/op_image_gaussian_blur.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Applies a Gaussian blur to the image(s). — op_image_gaussian_blur","text":"","code":"op_image_gaussian_blur(   images,   kernel_size = list(3L, 3L),   sigma = list(1, 1),   data_format = NULL )"},{"path":"https://keras3.posit.co/dev/reference/op_image_gaussian_blur.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Applies a Gaussian blur to the image(s). — op_image_gaussian_blur","text":"images Input image batch images. Must 3D 4D. kernel_size tuple two integers, specifying height width Gaussian kernel. sigma tuple two floats, specifying standard deviation Gaussian kernel along height width. data_format string specifying data format input tensor. can either \"channels_last\" \"channels_first\". \"channels_last\" corresponds inputs shape (batch, height, width, channels), \"channels_first\" corresponds inputs shape (batch, channels, height, width). specified, value default keras.config.image_data_format.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_image_gaussian_blur.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Applies a Gaussian blur to the image(s). — op_image_gaussian_blur","text":"Blurred image batch images.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_image_gaussian_blur.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Applies a Gaussian blur to the image(s). — op_image_gaussian_blur","text":"","code":"x <- op_ones(c(2, 64, 80, 3))  # batch of 2 RGB images y <- op_image_gaussian_blur(x) op_shape(y) ## shape(2, 64, 80, 3) x <- op_ones(c(64, 80, 3))  # single RGB image y <- op_image_gaussian_blur(x) op_shape(y) ## shape(64, 80, 3) x <- op_ones(c(2, 3, 64, 80))  # batch of 2 RGB images, channels_first y <- op_image_gaussian_blur(x, data_format = \"channels_first\") op_shape(y) ## shape(2, 3, 64, 80)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_image_hsv_to_rgb.html","id":null,"dir":"Reference","previous_headings":"","what":"Convert HSV images to RGB. — op_image_hsv_to_rgb","title":"Convert HSV images to RGB. — op_image_hsv_to_rgb","text":"images must float dtype, output well defined values images [0, 1].","code":""},{"path":"https://keras3.posit.co/dev/reference/op_image_hsv_to_rgb.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Convert HSV images to RGB. — op_image_hsv_to_rgb","text":"","code":"op_image_hsv_to_rgb(images, data_format = NULL)"},{"path":"https://keras3.posit.co/dev/reference/op_image_hsv_to_rgb.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Convert HSV images to RGB. — op_image_hsv_to_rgb","text":"images Input image batch images. Must 3D 4D. data_format string specifying data format input tensor. can either \"channels_last\" \"channels_first\". \"channels_last\" corresponds inputs shape (batch, height, width, channels), \"channels_first\" corresponds inputs shape (batch, channels, height, width). specified, value default config_image_data_format().","code":""},{"path":"https://keras3.posit.co/dev/reference/op_image_hsv_to_rgb.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Convert HSV images to RGB. — op_image_hsv_to_rgb","text":"RGB image batch RGB images.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_image_hsv_to_rgb.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Convert HSV images to RGB. — op_image_hsv_to_rgb","text":"","code":"x <- random_uniform(c(2, 4, 4, 3)) y <- op_image_hsv_to_rgb(x) shape(y) ## shape(2, 4, 4, 3) x <- random_uniform(c(4, 4, 3)) # Single HSV image y <- op_image_hsv_to_rgb(x) shape(y) ## shape(4, 4, 3) x <- random_uniform(c(2, 3, 4, 4)) y <- op_image_hsv_to_rgb(x, data_format=\"channels_first\") shape(y) ## shape(2, 3, 4, 4)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_image_map_coordinates.html","id":null,"dir":"Reference","previous_headings":"","what":"Map the input array to new coordinates by interpolation. — op_image_map_coordinates","title":"Map the input array to new coordinates by interpolation. — op_image_map_coordinates","text":"Note interpolation near boundaries differs scipy function, fixed outstanding bug scipy/issues/2640.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_image_map_coordinates.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Map the input array to new coordinates by interpolation. — op_image_map_coordinates","text":"","code":"op_image_map_coordinates(   inputs,   coordinates,   order,   fill_mode = \"constant\",   fill_value = 0L )"},{"path":"https://keras3.posit.co/dev/reference/op_image_map_coordinates.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Map the input array to new coordinates by interpolation. — op_image_map_coordinates","text":"inputs input array. coordinates coordinates inputs evaluated. order order spline interpolation. order must 0 1. 0 indicates nearest neighbor 1 indicates linear interpolation. fill_mode Points outside boundaries inputs filled according given mode. Available methods \"constant\", \"nearest\", \"wrap\" \"mirror\" \"reflect\". Defaults \"constant\". \"constant\": (k k k k | b c d | k k k k) inputs extended filling values beyond edge constant value k specified fill_value. \"nearest\": (| b c d | d d d d) inputs extended nearest pixel. \"wrap\": (b c d | b c d | b c d) inputs extended wrapping around opposite edge. \"mirror\": (c d c b | b c d | c b b) inputs extended mirroring edge. \"reflect\": (d c b | b c d | d c b ) inputs extended reflecting edge last pixel. fill_value Value used points outside boundaries inputs fill_mode = \"constant\". Defaults 0.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_image_map_coordinates.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Map the input array to new coordinates by interpolation. — op_image_map_coordinates","text":"Output input batch inputs.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_image_pad.html","id":null,"dir":"Reference","previous_headings":"","what":"Pad images with zeros to the specified height and width. — op_image_pad","title":"Pad images with zeros to the specified height and width. — op_image_pad","text":"Pad images zeros specified height width.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_image_pad.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Pad images with zeros to the specified height and width. — op_image_pad","text":"","code":"op_image_pad(   images,   top_padding = NULL,   left_padding = NULL,   bottom_padding = NULL,   right_padding = NULL,   target_height = NULL,   target_width = NULL,   data_format = NULL )"},{"path":"https://keras3.posit.co/dev/reference/op_image_pad.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Pad images with zeros to the specified height and width. — op_image_pad","text":"images Input image batch images. Must 3D 4D. top_padding Number rows zeros add top. left_padding Number columns zeros add left. bottom_padding Number rows zeros add bottom. right_padding Number columns zeros add right. target_height Height output images. target_width Width output images. data_format string specifying data format input tensor. can either \"channels_last\" \"channels_first\". \"channels_last\" corresponds inputs shape (batch, height, width, channels), \"channels_first\" corresponds inputs shape (batch, channels, height, width). specified, value default config_image_data_format().","code":""},{"path":"https://keras3.posit.co/dev/reference/op_image_pad.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Pad images with zeros to the specified height and width. — op_image_pad","text":"Padded image batch images.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_image_pad.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Pad images with zeros to the specified height and width. — op_image_pad","text":"","code":"images <- random_uniform(c(15, 25, 3)) padded_images <- op_image_pad(     images, 2, 3, target_height = 20, target_width = 30 ) shape(padded_images) ## shape(20, 30, 3) batch_images <- random_uniform(c(2, 15, 25, 3)) padded_batch <- op_image_pad(batch_images, 2, 3,                              target_height = 20,                              target_width = 30) shape(padded_batch) ## shape(2, 20, 30, 3)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_image_perspective_transform.html","id":null,"dir":"Reference","previous_headings":"","what":"Applies a perspective transformation to the image(s). — op_image_perspective_transform","title":"Applies a perspective transformation to the image(s). — op_image_perspective_transform","text":"Applies perspective transformation image(s).","code":""},{"path":"https://keras3.posit.co/dev/reference/op_image_perspective_transform.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Applies a perspective transformation to the image(s). — op_image_perspective_transform","text":"","code":"op_image_perspective_transform(   images,   start_points,   end_points,   interpolation = \"bilinear\",   fill_value = 0L,   data_format = NULL )"},{"path":"https://keras3.posit.co/dev/reference/op_image_perspective_transform.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Applies a perspective transformation to the image(s). — op_image_perspective_transform","text":"images Input image batch images. Must 3D 4D. start_points tensor shape (N, 4, 2) (4, 2), representing source points original image define transformation. end_points tensor shape (N, 4, 2) (4, 2), representing target points output image transformation. interpolation Interpolation method. Available methods \"nearest\", \"bilinear\". Defaults \"bilinear\". fill_value Value used points outside boundaries input extrapolation needed. Defaults 0. data_format string specifying data format input tensor. can either \"channels_last\" \"channels_first\". \"channels_last\" corresponds inputs shape (batch, height, width, channels), \"channels_first\" corresponds inputs shape (batch, channels, height, width). specified, value default keras.config.image_data_format.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_image_perspective_transform.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Applies a perspective transformation to the image(s). — op_image_perspective_transform","text":"Applied perspective transform image batch images.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_image_perspective_transform.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Applies a perspective transformation to the image(s). — op_image_perspective_transform","text":"","code":"# Batch of 2 RGB images (channels_last) x <- op_ones(c(2, 64, 80, 3)) start_points <- op_stack(list(   rbind(c(0, 0), c(0, 64), c(80, 0), c(80, 64)),   rbind(c(0, 0), c(0, 64), c(80, 0), c(80, 64)) )) end_points <- op_stack(list(   rbind(c(3, 5), c(7, 64), c(76, -10), c(84, 61)),   rbind(c(8, 10), c(10, 61), c(65, 3), c(88, 43)) )) y <- op_image_perspective_transform(x, start_points, end_points) op_shape(y) ## shape(2, 64, 80, 3) # Single RGB image (channels_last) x <- op_ones(c(64, 80, 3)) start_points <- rbind(c(0, 0), c(0, 64), c(80, 0), c(80, 64)) end_points <- rbind(c(3, 5), c(7, 64), c(76, -10), c(84, 61)) y <- op_image_perspective_transform(x, start_points, end_points) op_shape(y) ## shape(64, 80, 3) # Batch of 2 RGB images (channels_first) x <- op_ones(c(2, 3, 64, 80)) start_points <- op_stack(list(   rbind(c(0, 0), c(0, 64), c(80, 0), c(80, 64)),   rbind(c(0, 0), c(0, 64), c(80, 0), c(80, 64)) )) end_points <- op_stack(list(   rbind(c(3, 5), c(7, 64), c(76, -10), c(84, 61)),   rbind(c(8, 10), c(10, 61), c(65, 3), c(88, 43)) )) y <- op_image_perspective_transform(   x, start_points, end_points,   data_format = \"channels_first\" ) op_shape(y) ## shape(2, 3, 64, 80)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_image_resize.html","id":null,"dir":"Reference","previous_headings":"","what":"Resize images to size using the specified interpolation method. — op_image_resize","title":"Resize images to size using the specified interpolation method. — op_image_resize","text":"Resize images size using specified interpolation method.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_image_resize.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Resize images to size using the specified interpolation method. — op_image_resize","text":"","code":"op_image_resize(   images,   size,   interpolation = \"bilinear\",   antialias = FALSE,   crop_to_aspect_ratio = FALSE,   pad_to_aspect_ratio = FALSE,   fill_mode = \"constant\",   fill_value = 0,   data_format = NULL )"},{"path":"https://keras3.posit.co/dev/reference/op_image_resize.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Resize images to size using the specified interpolation method. — op_image_resize","text":"images Input image batch images. Must 3D 4D. size Size output image (height, width) format. interpolation Interpolation method. Available methods \"nearest\", \"bilinear\", \"bicubic\". Defaults \"bilinear\". antialias Whether use antialiasing filter downsampling image. Defaults FALSE. crop_to_aspect_ratio TRUE, resize images without aspect ratio distortion. original aspect ratio differs target aspect ratio, output image cropped return largest possible window image (size (height, width)) matches target aspect ratio. default (crop_to_aspect_ratio=FALSE), aspect ratio may preserved. pad_to_aspect_ratio TRUE, pad images without aspect ratio distortion. original aspect ratio differs target aspect ratio, output image evenly padded short side. fill_mode using pad_to_aspect_ratio=TRUE, padded areas filled according given mode. \"constant\" supported time (fill constant value, equal fill_value). fill_value Float. Padding value use pad_to_aspect_ratio=TRUE. data_format string specifying data format input tensor. can either \"channels_last\" \"channels_first\". \"channels_last\" corresponds inputs shape (batch, height, width, channels), \"channels_first\" corresponds inputs shape (batch, channels, height, width). specified, value default config_image_data_format().","code":""},{"path":"https://keras3.posit.co/dev/reference/op_image_resize.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Resize images to size using the specified interpolation method. — op_image_resize","text":"Resized image batch images.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_image_resize.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Resize images to size using the specified interpolation method. — op_image_resize","text":"","code":"x <- random_uniform(c(2, 4, 4, 3)) # batch of 2 RGB images y <- op_image_resize(x, c(2, 2)) shape(y) ## shape(2, 2, 2, 3) x <- random_uniform(c(4, 4, 3)) # single RGB image y <- op_image_resize(x, c(2, 2)) shape(y) ## shape(2, 2, 3) x <- random_uniform(c(2, 3, 4, 4)) # batch of 2 RGB images y <- op_image_resize(x, c(2, 2), data_format = \"channels_first\") shape(y) ## shape(2, 3, 2, 2)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_image_rgb_to_grayscale.html","id":null,"dir":"Reference","previous_headings":"","what":"Convert RGB images to grayscale. — op_image_rgb_to_grayscale","title":"Convert RGB images to grayscale. — op_image_rgb_to_grayscale","text":"function converts RGB images grayscale images. supports 3D 4D tensors, last dimension represents channels.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_image_rgb_to_grayscale.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Convert RGB images to grayscale. — op_image_rgb_to_grayscale","text":"","code":"op_image_rgb_to_grayscale(images, data_format = NULL)"},{"path":"https://keras3.posit.co/dev/reference/op_image_rgb_to_grayscale.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Convert RGB images to grayscale. — op_image_rgb_to_grayscale","text":"images Input image batch images. Must 3D 4D. data_format string specifying data format input tensor. can either \"channels_last\" \"channels_first\". \"channels_last\" corresponds inputs shape (batch, height, width, channels), \"channels_first\" corresponds inputs shape (batch, channels, height, width). specified, value default config_image_data_format().","code":""},{"path":"https://keras3.posit.co/dev/reference/op_image_rgb_to_grayscale.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Convert RGB images to grayscale. — op_image_rgb_to_grayscale","text":"Grayscale image batch grayscale images.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_image_rgb_to_grayscale.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Convert RGB images to grayscale. — op_image_rgb_to_grayscale","text":"","code":"x <- random_uniform(c(2, 4, 4, 3)) y <- op_image_rgb_to_grayscale(x) shape(y) ## shape(2, 4, 4, 1) x <- random_uniform(c(4, 4, 3)) # Single RGB image y = op_image_rgb_to_grayscale(x) shape(y) ## shape(4, 4, 1) x <- random_uniform(c(2, 3, 4, 4)) y <- op_image_rgb_to_grayscale(x, data_format=\"channels_first\") shape(y) ## shape(2, 1, 4, 4)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_image_rgb_to_hsv.html","id":null,"dir":"Reference","previous_headings":"","what":"Convert RGB images to HSV. — op_image_rgb_to_hsv","title":"Convert RGB images to HSV. — op_image_rgb_to_hsv","text":"images must float dtype, output well defined values images [0, 1]. HSV values [0, 1]. hue 0 corresponds pure red, 1/3 pure green, 2/3 pure blue.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_image_rgb_to_hsv.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Convert RGB images to HSV. — op_image_rgb_to_hsv","text":"","code":"op_image_rgb_to_hsv(images, data_format = NULL)"},{"path":"https://keras3.posit.co/dev/reference/op_image_rgb_to_hsv.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Convert RGB images to HSV. — op_image_rgb_to_hsv","text":"images Input image batch images. Must 3D 4D. data_format string specifying data format input tensor. can either \"channels_last\" \"channels_first\". \"channels_last\" corresponds inputs shape (batch, height, width, channels), \"channels_first\" corresponds inputs shape (batch, channels, height, width). specified, value default config_image_data_format().","code":""},{"path":"https://keras3.posit.co/dev/reference/op_image_rgb_to_hsv.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Convert RGB images to HSV. — op_image_rgb_to_hsv","text":"HSV image batch HSV images.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_image_rgb_to_hsv.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Convert RGB images to HSV. — op_image_rgb_to_hsv","text":"","code":"x <- random_uniform(c(2, 4, 4, 3)) y <- op_image_rgb_to_hsv(x) shape(y) ## shape(2, 4, 4, 3) x <- random_uniform(c(4, 4, 3)) # Single RGB image y <- op_image_rgb_to_hsv(x) shape(y) ## shape(4, 4, 3) x <- random_uniform(c(2, 3, 4, 4)) y <- op_image_rgb_to_hsv(x, data_format = \"channels_first\") shape(y) ## shape(2, 3, 4, 4)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_in_top_k.html","id":null,"dir":"Reference","previous_headings":"","what":"Checks if the targets are in the top-k predictions. — op_in_top_k","title":"Checks if the targets are in the top-k predictions. — op_in_top_k","text":"Checks targets top-k predictions.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_in_top_k.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Checks if the targets are in the top-k predictions. — op_in_top_k","text":"","code":"op_in_top_k(targets, predictions, k)"},{"path":"https://keras3.posit.co/dev/reference/op_in_top_k.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Checks if the targets are in the top-k predictions. — op_in_top_k","text":"targets tensor true labels. predictions tensor predicted labels. k integer representing number predictions consider.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_in_top_k.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Checks if the targets are in the top-k predictions. — op_in_top_k","text":"boolean tensor shape targets, element indicates whether corresponding target top-k predictions.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_in_top_k.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Checks if the targets are in the top-k predictions. — op_in_top_k","text":"","code":"targets <- op_array(c(2, 5, 3), \"int32\") predictions <- op_array(dtype = \"float32\", rbind(   c(0.1, 0.4, 0.6, 0.9, 0.5),   c(0.1, 0.7, 0.9, 0.8, 0.3),   c(0.1, 0.6, 0.9, 0.9, 0.5) )) op_in_top_k(targets, predictions, k = 3L) ## tf.Tensor([ True False  True], shape=(3), dtype=bool)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_inner.html","id":null,"dir":"Reference","previous_headings":"","what":"Return the inner product of two tensors. — op_inner","title":"Return the inner product of two tensors. — op_inner","text":"Ordinary inner product vectors 1-D tensors (without complex conjugation), higher dimensions sum product last axes. Multidimensional arrays treated vectors flattening last axes. resulting dot product performed last axes.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_inner.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Return the inner product of two tensors. — op_inner","text":"","code":"op_inner(x1, x2)"},{"path":"https://keras3.posit.co/dev/reference/op_inner.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Return the inner product of two tensors. — op_inner","text":"x1 First input tensor. x2 Second input tensor. last dimension x1 x2 must match.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_inner.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Return the inner product of two tensors. — op_inner","text":"Output tensor. shape output determined broadcasting shapes x1 x2 removing last axes.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_inv.html","id":null,"dir":"Reference","previous_headings":"","what":"Computes the inverse of a square tensor. — op_inv","title":"Computes the inverse of a square tensor. — op_inv","text":"Computes inverse square tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_inv.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes the inverse of a square tensor. — op_inv","text":"","code":"op_inv(x)"},{"path":"https://keras3.posit.co/dev/reference/op_inv.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Computes the inverse of a square tensor. — op_inv","text":"x Input tensor shape (..., M, M).","code":""},{"path":"https://keras3.posit.co/dev/reference/op_inv.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Computes the inverse of a square tensor. — op_inv","text":"tensor shape (..., M, M) representing inverse x.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_irfft.html","id":null,"dir":"Reference","previous_headings":"","what":"Inverse real-valued Fast Fourier transform along the last axis. — op_irfft","title":"Inverse real-valued Fast Fourier transform along the last axis. — op_irfft","text":"Computes inverse 1D Discrete Fourier Transform real-valued signal inner-dimension input. inner-dimension input assumed result RFFT: fft_length / 2 + 1 unique components DFT real-valued signal. fft_length provided, computed size inner-dimension input (fft_length = 2 * (inner - 1)). FFT length used compute odd, provided since inferred properly. Along axis IRFFT computed , fft_length / 2 + 1 smaller corresponding dimension input, dimension cropped. larger, dimension padded zeros.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_irfft.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Inverse real-valued Fast Fourier transform along the last axis. — op_irfft","text":"","code":"op_irfft(x, fft_length = NULL)"},{"path":"https://keras3.posit.co/dev/reference/op_irfft.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Inverse real-valued Fast Fourier transform along the last axis. — op_irfft","text":"x List real imaginary parts input tensor. tensors list floating type. fft_length integer representing number fft length. specified, inferred length last axis x. Defaults NULL.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_irfft.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Inverse real-valued Fast Fourier transform along the last axis. — op_irfft","text":"tensor containing inverse real-valued Fast Fourier Transform along last axis x.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_irfft.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Inverse real-valued Fast Fourier transform along the last axis. — op_irfft","text":"","code":"real <- op_array(c(0, 1, 2, 3, 4)) imag <- op_array(c(0, 1, 2, 3, 4)) op_irfft(c(real, imag)) #> tf.Tensor( #> [ 2.         -2.0606601   0.5        -0.35355338  0.          0.06066012 #>  -0.5         0.35355338], shape=(8), dtype=float32) all.equal(op_irfft(op_rfft(real, 5), 5), real) #> [1] TRUE"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_is_tensor.html","id":null,"dir":"Reference","previous_headings":"","what":"Check whether the given object is a tensor. — op_is_tensor","title":"Check whether the given object is a tensor. — op_is_tensor","text":"Check whether given object tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_is_tensor.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Check whether the given object is a tensor. — op_is_tensor","text":"","code":"op_is_tensor(x)"},{"path":"https://keras3.posit.co/dev/reference/op_is_tensor.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Check whether the given object is a tensor. — op_is_tensor","text":"x variable.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_is_tensor.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Check whether the given object is a tensor. — op_is_tensor","text":"TRUE x tensor, otherwise FALSE.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_is_tensor.html","id":"note","dir":"Reference","previous_headings":"","what":"Note","title":"Check whether the given object is a tensor. — op_is_tensor","text":"checks backend specific tensors passing TensorFlow tensor return FALSE backend PyTorch JAX.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_isclose.html","id":null,"dir":"Reference","previous_headings":"","what":"Return whether two tensors are element-wise almost equal. — op_isclose","title":"Return whether two tensors are element-wise almost equal. — op_isclose","text":"Return whether two tensors element-wise almost equal.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_isclose.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Return whether two tensors are element-wise almost equal. — op_isclose","text":"","code":"op_isclose(x1, x2, rtol = 1e-05, atol = 1e-08, equal_nan = FALSE)"},{"path":"https://keras3.posit.co/dev/reference/op_isclose.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Return whether two tensors are element-wise almost equal. — op_isclose","text":"x1 First input tensor. x2 Second input tensor. rtol Relative tolerance. atol Absolute tolerance. equal_nan TRUE, element-wise NaNs considered equal.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_isclose.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Return whether two tensors are element-wise almost equal. — op_isclose","text":"Output boolean tensor.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_isfinite.html","id":null,"dir":"Reference","previous_headings":"","what":"Return whether a tensor is finite, element-wise. — op_isfinite","title":"Return whether a tensor is finite, element-wise. — op_isfinite","text":"Real values finite NaN, positive infinity, negative infinity. Complex values finite real imaginary parts finite.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_isfinite.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Return whether a tensor is finite, element-wise. — op_isfinite","text":"","code":"op_isfinite(x)"},{"path":"https://keras3.posit.co/dev/reference/op_isfinite.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Return whether a tensor is finite, element-wise. — op_isfinite","text":"x Input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_isfinite.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Return whether a tensor is finite, element-wise. — op_isfinite","text":"Output boolean tensor.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_isinf.html","id":null,"dir":"Reference","previous_headings":"","what":"Test element-wise for positive or negative infinity. — op_isinf","title":"Test element-wise for positive or negative infinity. — op_isinf","text":"Test element-wise positive negative infinity.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_isinf.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Test element-wise for positive or negative infinity. — op_isinf","text":"","code":"op_isinf(x)"},{"path":"https://keras3.posit.co/dev/reference/op_isinf.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Test element-wise for positive or negative infinity. — op_isinf","text":"x Input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_isinf.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Test element-wise for positive or negative infinity. — op_isinf","text":"Output boolean tensor.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_isnan.html","id":null,"dir":"Reference","previous_headings":"","what":"Test element-wise for NaN and return result as a boolean tensor. — op_isnan","title":"Test element-wise for NaN and return result as a boolean tensor. — op_isnan","text":"Test element-wise NaN return result boolean tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_isnan.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Test element-wise for NaN and return result as a boolean tensor. — op_isnan","text":"","code":"op_isnan(x)"},{"path":"https://keras3.posit.co/dev/reference/op_isnan.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Test element-wise for NaN and return result as a boolean tensor. — op_isnan","text":"x Input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_isnan.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Test element-wise for NaN and return result as a boolean tensor. — op_isnan","text":"Output boolean tensor.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_istft.html","id":null,"dir":"Reference","previous_headings":"","what":"Inverse Short-Time Fourier Transform along the last axis of the input. — op_istft","title":"Inverse Short-Time Fourier Transform along the last axis of the input. — op_istft","text":"reconstruct original waveform, parameters stft.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_istft.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Inverse Short-Time Fourier Transform along the last axis of the input. — op_istft","text":"","code":"op_istft(   x,   sequence_length,   sequence_stride,   fft_length,   length = NULL,   window = \"hann\",   center = TRUE )"},{"path":"https://keras3.posit.co/dev/reference/op_istft.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Inverse Short-Time Fourier Transform along the last axis of the input. — op_istft","text":"x Tuple real imaginary parts input tensor. tensors list floating type. sequence_length integer representing sequence length. sequence_stride integer representing sequence hop size. fft_length integer representing size FFT produced stft. type int32. length integer representing output clipped exactly length. specified, padding clipping take place. Defaults NULL. window string, tensor window NULL. window string, available values \"hann\" \"hamming\". window tensor, used directly window length must sequence_length. window NULL, windowing used. Defaults \"hann\". center Whether x padded sides t-th sequence centered time t * sequence_stride. Defaults TRUE.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_istft.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Inverse Short-Time Fourier Transform along the last axis of the input. — op_istft","text":"tensor containing inverse Short-Time Fourier Transform along last axis x.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_istft.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Inverse Short-Time Fourier Transform along the last axis of the input. — op_istft","text":"","code":"x <- op_convert_to_tensor(c(0, 1, 2, 3, 4)) op_istft(op_stft(x, 1, 1, 1), 1, 1, 1) ## tf.Tensor([], shape=(0), dtype=float64) # array([0.0, 1.0, 2.0, 3.0, 4.0])"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_leaky_relu.html","id":null,"dir":"Reference","previous_headings":"","what":"Leaky version of a Rectified Linear Unit activation function. — op_leaky_relu","title":"Leaky version of a Rectified Linear Unit activation function. — op_leaky_relu","text":"allows small gradient unit active, defined : f(x) = alpha * x x < 0 f(x) = x x >= 0.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_leaky_relu.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Leaky version of a Rectified Linear Unit activation function. — op_leaky_relu","text":"","code":"op_leaky_relu(x, negative_slope = 0.2)"},{"path":"https://keras3.posit.co/dev/reference/op_leaky_relu.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Leaky version of a Rectified Linear Unit activation function. — op_leaky_relu","text":"x Input tensor. negative_slope Slope activation function x < 0. Defaults 0.2.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_leaky_relu.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Leaky version of a Rectified Linear Unit activation function. — op_leaky_relu","text":"tensor shape x.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_leaky_relu.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Leaky version of a Rectified Linear Unit activation function. — op_leaky_relu","text":"","code":"x <- op_array(c(-1., 0., 1.)) op_leaky_relu(x) ## tf.Tensor([-0.2  0.   1. ], shape=(3), dtype=float32) # array([-0.2,  0. ,  1. ], shape=(3,), dtype=float64) x <- seq(-5, 5, .1) plot(x, op_leaky_relu(x),      type = 'l', panel.first = grid())"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_left_shift.html","id":null,"dir":"Reference","previous_headings":"","what":"Shift the bits of an integer to the left. — op_left_shift","title":"Shift the bits of an integer to the left. — op_left_shift","text":"Bits shifted left appending y 0s right x. Since internal representation numbers binary format, operation equivalent multiplying x 2^y.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_left_shift.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Shift the bits of an integer to the left. — op_left_shift","text":"","code":"op_left_shift(x, y)"},{"path":"https://keras3.posit.co/dev/reference/op_left_shift.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Shift the bits of an integer to the left. — op_left_shift","text":"x Input integer tensor. y Input integer tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_left_shift.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Shift the bits of an integer to the left. — op_left_shift","text":"Result tensor.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_less.html","id":null,"dir":"Reference","previous_headings":"","what":"Return the truth value of x1 < x2 element-wise. — op_less","title":"Return the truth value of x1 < x2 element-wise. — op_less","text":"Note function automatically called using R operator < tensor.","code":"(x <- op_arange(4)) ## tf.Tensor([1. 2. 3. 4.], shape=(4), dtype=float32) op_less(x, 2) ## tf.Tensor([ True False False False], shape=(4), dtype=bool) x < 2 ## tf.Tensor([ True False False False], shape=(4), dtype=bool)"},{"path":"https://keras3.posit.co/dev/reference/op_less.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Return the truth value of x1 < x2 element-wise. — op_less","text":"","code":"op_less(x1, x2)"},{"path":"https://keras3.posit.co/dev/reference/op_less.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Return the truth value of x1 < x2 element-wise. — op_less","text":"x1 First input tensor. x2 Second input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_less.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Return the truth value of x1 < x2 element-wise. — op_less","text":"Output tensor, element-wise comparison x1 x2.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_less_equal.html","id":null,"dir":"Reference","previous_headings":"","what":"Return the truth value of x1 <= x2 element-wise. — op_less_equal","title":"Return the truth value of x1 <= x2 element-wise. — op_less_equal","text":"Note function automatically called using R operator <= tensor.","code":"(x <- op_arange(4)) ## tf.Tensor([1. 2. 3. 4.], shape=(4), dtype=float32) op_less_equal(x, 2) ## tf.Tensor([ True  True False False], shape=(4), dtype=bool) x <= 2 ## tf.Tensor([ True  True False False], shape=(4), dtype=bool)"},{"path":"https://keras3.posit.co/dev/reference/op_less_equal.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Return the truth value of x1 <= x2 element-wise. — op_less_equal","text":"","code":"op_less_equal(x1, x2)"},{"path":"https://keras3.posit.co/dev/reference/op_less_equal.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Return the truth value of x1 <= x2 element-wise. — op_less_equal","text":"x1 First input tensor. x2 Second input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_less_equal.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Return the truth value of x1 <= x2 element-wise. — op_less_equal","text":"Output tensor, element-wise comparison x1 x2.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_linspace.html","id":null,"dir":"Reference","previous_headings":"","what":"Return evenly spaced numbers over a specified interval. — op_linspace","title":"Return evenly spaced numbers over a specified interval. — op_linspace","text":"Returns num evenly spaced samples, calculated interval [start, stop]. endpoint interval can optionally excluded.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_linspace.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Return evenly spaced numbers over a specified interval. — op_linspace","text":"","code":"op_linspace(   start,   stop,   num = 50L,   endpoint = TRUE,   retstep = FALSE,   dtype = NULL,   axis = 1L )"},{"path":"https://keras3.posit.co/dev/reference/op_linspace.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Return evenly spaced numbers over a specified interval. — op_linspace","text":"start starting value sequence. stop end value sequence, unless endpoint set FALSE. case, sequence consists last num + 1 evenly spaced samples, stop excluded. Note step size changes endpoint FALSE. num Number samples generate. Defaults 50. Must non-negative. endpoint TRUE, stop last sample. Otherwise, included. Defaults TRUE. retstep TRUE, return (samples, step), step spacing samples. dtype type output tensor. axis axis result store samples. Relevant start stop array-like. Defaults 1, first axis.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_linspace.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Return evenly spaced numbers over a specified interval. — op_linspace","text":"tensor evenly spaced numbers. retstep TRUE, returns (samples, step)","code":""},{"path":"https://keras3.posit.co/dev/reference/op_linspace.html","id":"note","dir":"Reference","previous_headings":"","what":"Note","title":"Return evenly spaced numbers over a specified interval. — op_linspace","text":"Torch backend support axis argument.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_log.html","id":null,"dir":"Reference","previous_headings":"","what":"Natural logarithm, element-wise. — op_log","title":"Natural logarithm, element-wise. — op_log","text":"Natural logarithm, element-wise.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_log.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Natural logarithm, element-wise. — op_log","text":"","code":"op_log(x)"},{"path":"https://keras3.posit.co/dev/reference/op_log.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Natural logarithm, element-wise. — op_log","text":"x Input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_log.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Natural logarithm, element-wise. — op_log","text":"Output tensor, element-wise natural logarithm x.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_log10.html","id":null,"dir":"Reference","previous_headings":"","what":"Return the base 10 logarithm of the input tensor, element-wise. — op_log10","title":"Return the base 10 logarithm of the input tensor, element-wise. — op_log10","text":"Return base 10 logarithm input tensor, element-wise.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_log10.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Return the base 10 logarithm of the input tensor, element-wise. — op_log10","text":"","code":"op_log10(x)"},{"path":"https://keras3.posit.co/dev/reference/op_log10.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Return the base 10 logarithm of the input tensor, element-wise. — op_log10","text":"x Input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_log10.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Return the base 10 logarithm of the input tensor, element-wise. — op_log10","text":"Output tensor, element-wise base 10 logarithm x.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_log1p.html","id":null,"dir":"Reference","previous_headings":"","what":"Returns the natural logarithm of one plus the x, element-wise. — op_log1p","title":"Returns the natural logarithm of one plus the x, element-wise. — op_log1p","text":"Calculates log(1 + x).","code":""},{"path":"https://keras3.posit.co/dev/reference/op_log1p.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Returns the natural logarithm of one plus the x, element-wise. — op_log1p","text":"","code":"op_log1p(x)"},{"path":"https://keras3.posit.co/dev/reference/op_log1p.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Returns the natural logarithm of one plus the x, element-wise. — op_log1p","text":"x Input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_log1p.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Returns the natural logarithm of one plus the x, element-wise. — op_log1p","text":"Output tensor, element-wise natural logarithm 1 + x.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_log2.html","id":null,"dir":"Reference","previous_headings":"","what":"Base-2 logarithm of x, element-wise. — op_log2","title":"Base-2 logarithm of x, element-wise. — op_log2","text":"Base-2 logarithm x, element-wise.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_log2.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Base-2 logarithm of x, element-wise. — op_log2","text":"","code":"op_log2(x)"},{"path":"https://keras3.posit.co/dev/reference/op_log2.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Base-2 logarithm of x, element-wise. — op_log2","text":"x Input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_log2.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Base-2 logarithm of x, element-wise. — op_log2","text":"Output tensor, element-wise base-2 logarithm x.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_log_sigmoid.html","id":null,"dir":"Reference","previous_headings":"","what":"Logarithm of the sigmoid activation function. — op_log_sigmoid","title":"Logarithm of the sigmoid activation function. — op_log_sigmoid","text":"defined f(x) = log(1 / (1 + exp(-x))).","code":""},{"path":"https://keras3.posit.co/dev/reference/op_log_sigmoid.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Logarithm of the sigmoid activation function. — op_log_sigmoid","text":"","code":"op_log_sigmoid(x)"},{"path":"https://keras3.posit.co/dev/reference/op_log_sigmoid.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Logarithm of the sigmoid activation function. — op_log_sigmoid","text":"x Input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_log_sigmoid.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Logarithm of the sigmoid activation function. — op_log_sigmoid","text":"tensor shape x.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_log_sigmoid.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Logarithm of the sigmoid activation function. — op_log_sigmoid","text":"","code":"x <- op_convert_to_tensor(c(-0.541391, 0.0, 0.50, 5.0)) op_log_sigmoid(x) ## tf.Tensor([-1.00004181 -0.69314718 -0.47407698 -0.00671535], shape=(4), dtype=float64)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_log_softmax.html","id":null,"dir":"Reference","previous_headings":"","what":"Log-softmax activation function. — op_log_softmax","title":"Log-softmax activation function. — op_log_softmax","text":"defined : f(x) = x - max(x) - log(sum(exp(x - max(x))))","code":""},{"path":"https://keras3.posit.co/dev/reference/op_log_softmax.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Log-softmax activation function. — op_log_softmax","text":"","code":"op_log_softmax(x, axis = -1L)"},{"path":"https://keras3.posit.co/dev/reference/op_log_softmax.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Log-softmax activation function. — op_log_softmax","text":"x Input tensor. axis Integer, axis along log-softmax applied. Defaults -1.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_log_softmax.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Log-softmax activation function. — op_log_softmax","text":"tensor shape x.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_log_softmax.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Log-softmax activation function. — op_log_softmax","text":"","code":"x <- op_array(c(-1., 0., 1.)) op_log_softmax(x) ## tf.Tensor([-2.407606   -1.4076059  -0.40760595], shape=(3), dtype=float32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_logaddexp.html","id":null,"dir":"Reference","previous_headings":"","what":"Logarithm of the sum of exponentiations of the inputs. — op_logaddexp","title":"Logarithm of the sum of exponentiations of the inputs. — op_logaddexp","text":"Calculates log(exp(x1) + exp(x2)).","code":""},{"path":"https://keras3.posit.co/dev/reference/op_logaddexp.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Logarithm of the sum of exponentiations of the inputs. — op_logaddexp","text":"","code":"op_logaddexp(x1, x2)"},{"path":"https://keras3.posit.co/dev/reference/op_logaddexp.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Logarithm of the sum of exponentiations of the inputs. — op_logaddexp","text":"x1 Input tensor. x2 Input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_logaddexp.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Logarithm of the sum of exponentiations of the inputs. — op_logaddexp","text":"Output tensor, element-wise logarithm sum exponentiations inputs.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_logdet.html","id":null,"dir":"Reference","previous_headings":"","what":"Computes log of the determinant of a hermitian positive definite matrix. — op_logdet","title":"Computes log of the determinant of a hermitian positive definite matrix. — op_logdet","text":"Computes log determinant hermitian positive definite matrix.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_logdet.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes log of the determinant of a hermitian positive definite matrix. — op_logdet","text":"","code":"op_logdet(x)"},{"path":"https://keras3.posit.co/dev/reference/op_logdet.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Computes log of the determinant of a hermitian positive definite matrix. — op_logdet","text":"x Input matrix. must 2D square.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_logdet.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Computes log of the determinant of a hermitian positive definite matrix. — op_logdet","text":"natural log determinant matrix.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_logical_and.html","id":null,"dir":"Reference","previous_headings":"","what":"Computes the element-wise logical AND of the given input tensors. — op_logical_and","title":"Computes the element-wise logical AND of the given input tensors. — op_logical_and","text":"Zeros treated FALSE non-zeros treated TRUE.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_logical_and.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes the element-wise logical AND of the given input tensors. — op_logical_and","text":"","code":"op_logical_and(x1, x2)"},{"path":"https://keras3.posit.co/dev/reference/op_logical_and.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Computes the element-wise logical AND of the given input tensors. — op_logical_and","text":"x1 Input tensor. x2 Input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_logical_and.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Computes the element-wise logical AND of the given input tensors. — op_logical_and","text":"Output tensor, element-wise logical inputs.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_logical_and.html","id":"details","dir":"Reference","previous_headings":"","what":"Details","title":"Computes the element-wise logical AND of the given input tensors. — op_logical_and","text":"Note function automatically called using R operator & tensor.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_logical_not.html","id":null,"dir":"Reference","previous_headings":"","what":"Computes the element-wise NOT of the given input tensor. — op_logical_not","title":"Computes the element-wise NOT of the given input tensor. — op_logical_not","text":"Zeros treated FALSE non-zeros treated TRUE. Note function automatically called using R operator ! tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_logical_not.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes the element-wise NOT of the given input tensor. — op_logical_not","text":"","code":"op_logical_not(x)"},{"path":"https://keras3.posit.co/dev/reference/op_logical_not.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Computes the element-wise NOT of the given input tensor. — op_logical_not","text":"x Input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_logical_not.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Computes the element-wise NOT of the given input tensor. — op_logical_not","text":"Output tensor, element-wise logical input.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_logical_or.html","id":null,"dir":"Reference","previous_headings":"","what":"Computes the element-wise logical OR of the given input tensors. — op_logical_or","title":"Computes the element-wise logical OR of the given input tensors. — op_logical_or","text":"Zeros treated FALSE non-zeros treated TRUE. Note function automatically called using R operator | tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_logical_or.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes the element-wise logical OR of the given input tensors. — op_logical_or","text":"","code":"op_logical_or(x1, x2)"},{"path":"https://keras3.posit.co/dev/reference/op_logical_or.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Computes the element-wise logical OR of the given input tensors. — op_logical_or","text":"x1 Input tensor. x2 Input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_logical_or.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Computes the element-wise logical OR of the given input tensors. — op_logical_or","text":"Output tensor, element-wise logical inputs.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_logical_xor.html","id":null,"dir":"Reference","previous_headings":"","what":"Compute the truth value of x1 XOR x2, element-wise. — op_logical_xor","title":"Compute the truth value of x1 XOR x2, element-wise. — op_logical_xor","text":"Compute truth value x1 XOR x2, element-wise.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_logical_xor.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Compute the truth value of x1 XOR x2, element-wise. — op_logical_xor","text":"","code":"op_logical_xor(x1, x2)"},{"path":"https://keras3.posit.co/dev/reference/op_logical_xor.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Compute the truth value of x1 XOR x2, element-wise. — op_logical_xor","text":"x1 First input tensor. x2 Second input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_logical_xor.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Compute the truth value of x1 XOR x2, element-wise. — op_logical_xor","text":"Output boolean tensor.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_logspace.html","id":null,"dir":"Reference","previous_headings":"","what":"Returns numbers spaced evenly on a log scale. — op_logspace","title":"Returns numbers spaced evenly on a log scale. — op_logspace","text":"linear space, sequence starts base ** start ends base ** stop (see endpoint ).","code":""},{"path":"https://keras3.posit.co/dev/reference/op_logspace.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Returns numbers spaced evenly on a log scale. — op_logspace","text":"","code":"op_logspace(   start,   stop,   num = 50L,   endpoint = TRUE,   base = 10L,   dtype = NULL,   axis = 1L )"},{"path":"https://keras3.posit.co/dev/reference/op_logspace.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Returns numbers spaced evenly on a log scale. — op_logspace","text":"start starting value sequence. stop final value sequence, unless endpoint FALSE. case, num + 1 values spaced interval log-space, last (sequence length num) returned. num Number samples generate. Defaults 50. endpoint TRUE, stop last sample. Otherwise, included. Defaults TRUE. base base log space. Defaults 10. dtype type output tensor. axis axis result store samples. Relevant start stop array-like.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_logspace.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Returns numbers spaced evenly on a log scale. — op_logspace","text":"tensor evenly spaced samples log scale.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_logspace.html","id":"note","dir":"Reference","previous_headings":"","what":"Note","title":"Returns numbers spaced evenly on a log scale. — op_logspace","text":"Torch backend support axis argument.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_logsumexp.html","id":null,"dir":"Reference","previous_headings":"","what":"Computes the logarithm of sum of exponentials of elements in a tensor. — op_logsumexp","title":"Computes the logarithm of sum of exponentials of elements in a tensor. — op_logsumexp","text":"Computes logarithm sum exponentials elements tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_logsumexp.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes the logarithm of sum of exponentials of elements in a tensor. — op_logsumexp","text":"","code":"op_logsumexp(x, axis = NULL, keepdims = FALSE)"},{"path":"https://keras3.posit.co/dev/reference/op_logsumexp.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Computes the logarithm of sum of exponentials of elements in a tensor. — op_logsumexp","text":"x Input tensor. axis integer list integers specifying axis/axes along compute sum. NULL, sum computed elements. Defaults NULL. keepdims boolean indicating whether keep dimensions input tensor computing sum. Defaults FALSE.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_logsumexp.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Computes the logarithm of sum of exponentials of elements in a tensor. — op_logsumexp","text":"tensor containing logarithm sum exponentials elements x.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_logsumexp.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Computes the logarithm of sum of exponentials of elements in a tensor. — op_logsumexp","text":"","code":"x <- op_convert_to_tensor(c(1, 2, 3)) op_logsumexp(x) ## tf.Tensor(3.4076059644443806, shape=(), dtype=float64)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_lstsq.html","id":null,"dir":"Reference","previous_headings":"","what":"Return the least-squares solution to a linear matrix equation. — op_lstsq","title":"Return the least-squares solution to a linear matrix equation. — op_lstsq","text":"Computes vector x approximately solves equation %*% x = b. equation may -, well-, -determined (.e., number linearly independent rows can less , equal , greater number linearly independent columns). square full rank, x (round-error) exact solution equation. Else, x minimizes L2 norm b - %*% x. multiple minimizing solutions, one smallest L2 norm  returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_lstsq.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Return the least-squares solution to a linear matrix equation. — op_lstsq","text":"","code":"op_lstsq(a, b, rcond = NULL)"},{"path":"https://keras3.posit.co/dev/reference/op_lstsq.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Return the least-squares solution to a linear matrix equation. — op_lstsq","text":"\"Coefficient\" matrix shape (M, N). b Ordinate \"dependent variable\" values, shape (M) (M, K). b two-dimensional, least-squares solution calculated K columns b. rcond Cut-ratio small singular values . purposes rank determination, singular values treated zero smaller rcond times largest singular value .","code":""},{"path":"https://keras3.posit.co/dev/reference/op_lstsq.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Return the least-squares solution to a linear matrix equation. — op_lstsq","text":"Tensor shape (N) (N, K) containing least-squares solutions. NOTE: output differs numpy.linalg.lstsq(). NumPy returns tuple four elements, first least-squares solutions others essentially never used. Keras returns first value. done ensure consistency across backends (achieved values) simplify API.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_lu_factor.html","id":null,"dir":"Reference","previous_headings":"","what":"Computes the lower-upper decomposition of a square matrix. — op_lu_factor","title":"Computes the lower-upper decomposition of a square matrix. — op_lu_factor","text":"Computes lower-upper decomposition square matrix.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_lu_factor.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes the lower-upper decomposition of a square matrix. — op_lu_factor","text":"","code":"op_lu_factor(x)"},{"path":"https://keras3.posit.co/dev/reference/op_lu_factor.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Computes the lower-upper decomposition of a square matrix. — op_lu_factor","text":"x tensor shape (..., M, M).","code":""},{"path":"https://keras3.posit.co/dev/reference/op_lu_factor.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Computes the lower-upper decomposition of a square matrix. — op_lu_factor","text":"tuple two tensors: tensor shape (..., M, M) containing lower upper triangular matrices tensor shape (..., M) containing pivots.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_map.html","id":null,"dir":"Reference","previous_headings":"","what":"Map a function over leading array axes. — op_map","title":"Map a function over leading array axes. — op_map","text":"Like purrr::map() base::lapply(), except inputs outputs form stacked arrays. Consider using op_vectorized_map() transform instead, unless need apply function element element reduced memory usage heterogeneous computation control flow primitives. xs array type, semantics op_map() match implementation:","code":"op_map <- function(xs, f) {   xs |>     op_unstack() |>     lapply(f) |>     op_stack() }"},{"path":"https://keras3.posit.co/dev/reference/op_map.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Map a function over leading array axes. — op_map","text":"","code":"op_map(xs, f)"},{"path":"https://keras3.posit.co/dev/reference/op_map.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Map a function over leading array axes. — op_map","text":"xs Values map along leading axis. f Callable defines function apply element-wise first axis axes xs.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_map.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Map a function over leading array axes. — op_map","text":"Mapped values.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_map.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Map a function over leading array axes. — op_map","text":"","code":"f <- function(x) x^2 xs <- op_arange(10) ys <- op_map(xs, f) ys ## tf.Tensor([  1.   4.   9.  16.  25.  36.  49.  64.  81. 100.], shape=(10), dtype=float32) f <- function(x) list(y1 = x^2, y2 = x * 10)  # Can have nested outputs ys <- op_map(xs, f) ys$y1 ## tf.Tensor([  1.   4.   9.  16.  25.  36.  49.  64.  81. 100.], shape=(10), dtype=float32) ys$y2 ## tf.Tensor([ 10.  20.  30.  40.  50.  60.  70.  80.  90. 100.], shape=(10), dtype=float32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_matmul.html","id":null,"dir":"Reference","previous_headings":"","what":"Matrix product of two tensors. — op_matmul","title":"Matrix product of two tensors. — op_matmul","text":"tensors 1-dimensional, dot product (scalar) returned. either tensor N-D, N > 2, treated stack matrices residing last two indexes broadcast accordingly. first tensor 1-D, promoted matrix prepending 1 dimensions. matrix multiplication prepended 1 removed. second tensor 1-D, promoted matrix appending 1 dimensions. matrix multiplication appended 1 removed.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_matmul.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Matrix product of two tensors. — op_matmul","text":"","code":"op_matmul(x1, x2)"},{"path":"https://keras3.posit.co/dev/reference/op_matmul.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Matrix product of two tensors. — op_matmul","text":"x1 First tensor. x2 Second tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_matmul.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Matrix product of two tensors. — op_matmul","text":"Output tensor, matrix product inputs.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_max.html","id":null,"dir":"Reference","previous_headings":"","what":"Return the maximum of a tensor or maximum along an axis. — op_max","title":"Return the maximum of a tensor or maximum along an axis. — op_max","text":"Return maximum tensor maximum along axis.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_max.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Return the maximum of a tensor or maximum along an axis. — op_max","text":"","code":"op_max(x, axis = NULL, keepdims = FALSE, initial = NULL)"},{"path":"https://keras3.posit.co/dev/reference/op_max.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Return the maximum of a tensor or maximum along an axis. — op_max","text":"x Input tensor. axis Axis axes along operate. default, flattened input used. keepdims set TRUE, axes reduced left result dimensions size one. Defaults FALSE. initial minimum value output element. Defaults NULL.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_max.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Return the maximum of a tensor or maximum along an axis. — op_max","text":"Maximum x.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_max.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Return the maximum of a tensor or maximum along an axis. — op_max","text":"","code":"(x <- op_convert_to_tensor(rbind(c(1, 3, 5), c(1, 5, 2)))) ## tf.Tensor( ## [[1. 3. 5.] ##  [1. 5. 2.]], shape=(2, 3), dtype=float64) op_max(x) ## tf.Tensor(5.0, shape=(), dtype=float64) op_max(x, axis = 1) ## tf.Tensor([1. 5. 5.], shape=(3), dtype=float64) op_max(x, axis = 1, keepdims = TRUE) ## tf.Tensor([[1. 5. 5.]], shape=(1, 3), dtype=float64)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_max_pool.html","id":null,"dir":"Reference","previous_headings":"","what":"Max pooling operation. — op_max_pool","title":"Max pooling operation. — op_max_pool","text":"Max pooling operation.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_max_pool.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Max pooling operation. — op_max_pool","text":"","code":"op_max_pool(   inputs,   pool_size,   strides = NULL,   padding = \"valid\",   data_format = NULL )"},{"path":"https://keras3.posit.co/dev/reference/op_max_pool.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Max pooling operation. — op_max_pool","text":"inputs Tensor rank N+2. inputs shape (batch_size,) + inputs_spatial_shape + (num_channels,) data_format = \"channels_last\", (batch_size, num_channels) + inputs_spatial_shape data_format = \"channels_first\". Pooling happens spatial dimensions . pool_size int tuple/list integers size len(inputs_spatial_shape), specifying size pooling window spatial dimension input tensor. pool_size int, every spatial dimension shares pool_size. strides int tuple/list integers size len(inputs_spatial_shape). stride sliding window spatial dimension input tensor. strides int, every spatial dimension shares strides. padding string, either \"valid\" \"\". \"valid\" means padding applied, \"\" results padding evenly left/right /input output height/width dimension input strides = 1. data_format string, either \"channels_last\" \"channels_first\". data_format determines ordering dimensions inputs. data_format = \"channels_last\", inputs shape (batch_size, ..., channels) data_format = \"channels_first\", inputs shape (batch_size, channels, ...).","code":""},{"path":"https://keras3.posit.co/dev/reference/op_max_pool.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Max pooling operation. — op_max_pool","text":"tensor rank N+2, result max pooling operation.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_maximum.html","id":null,"dir":"Reference","previous_headings":"","what":"Element-wise maximum of x1 and x2. — op_maximum","title":"Element-wise maximum of x1 and x2. — op_maximum","text":"Element-wise maximum x1 x2.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_maximum.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Element-wise maximum of x1 and x2. — op_maximum","text":"","code":"op_maximum(x1, x2)  op_pmax(x1, x2)"},{"path":"https://keras3.posit.co/dev/reference/op_maximum.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Element-wise maximum of x1 and x2. — op_maximum","text":"x1 First tensor. x2 Second tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_maximum.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Element-wise maximum of x1 and x2. — op_maximum","text":"Output tensor, element-wise maximum x1 x2.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_mean.html","id":null,"dir":"Reference","previous_headings":"","what":"Compute the arithmetic mean along the specified axes. — op_mean","title":"Compute the arithmetic mean along the specified axes. — op_mean","text":"Compute arithmetic mean along specified axes.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_mean.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Compute the arithmetic mean along the specified axes. — op_mean","text":"","code":"op_mean(x, axis = NULL, keepdims = FALSE)"},{"path":"https://keras3.posit.co/dev/reference/op_mean.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Compute the arithmetic mean along the specified axes. — op_mean","text":"x Input tensor. axis Axis axes along means computed. default compute mean flattened tensor. keepdims set TRUE, axes reduced left result dimensions size one.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_mean.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Compute the arithmetic mean along the specified axes. — op_mean","text":"Output tensor containing mean values.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_median.html","id":null,"dir":"Reference","previous_headings":"","what":"Compute the median along the specified axis. — op_median","title":"Compute the median along the specified axis. — op_median","text":"Compute median along specified axis.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_median.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Compute the median along the specified axis. — op_median","text":"","code":"op_median(x, axis = NULL, keepdims = FALSE)"},{"path":"https://keras3.posit.co/dev/reference/op_median.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Compute the median along the specified axis. — op_median","text":"x Input tensor. axis Axis axes along medians computed. Defaults axis = NULL compute median(s) along flattened version array. keepdims set TRUE, axes reduce left result dimensions size one.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_median.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Compute the median along the specified axis. — op_median","text":"output tensor.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_meshgrid.html","id":null,"dir":"Reference","previous_headings":"","what":"Creates grids of coordinates from coordinate vectors. — op_meshgrid","title":"Creates grids of coordinates from coordinate vectors. — op_meshgrid","text":"Given N 1-D tensors T0, T1, ..., TN-1 inputs corresponding lengths S0, S1, ..., SN-1, creates N N-dimensional tensors G0, G1, ..., GN-1 shape (S0, ..., SN-1) output Gi constructed expanding Ti result shape.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_meshgrid.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Creates grids of coordinates from coordinate vectors. — op_meshgrid","text":"","code":"op_meshgrid(..., indexing = \"xy\")"},{"path":"https://keras3.posit.co/dev/reference/op_meshgrid.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Creates grids of coordinates from coordinate vectors. — op_meshgrid","text":"... 1-D tensors representing coordinates grid. indexing \"xy\" \"ij\". \"xy\" cartesian; \"ij\" matrix indexing output. Defaults \"xy\".","code":""},{"path":"https://keras3.posit.co/dev/reference/op_meshgrid.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Creates grids of coordinates from coordinate vectors. — op_meshgrid","text":"Sequence N tensors.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_meshgrid.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Creates grids of coordinates from coordinate vectors. — op_meshgrid","text":"","code":"x <- op_array(c(1, 2, 3), \"int32\") y <- op_array(c(4, 5, 6), \"int32\") c(grid_x, grid_y) %<-% op_meshgrid(x, y, indexing = \"ij\") grid_x ## tf.Tensor( ## [[1 1 1] ##  [2 2 2] ##  [3 3 3]], shape=(3, 3), dtype=int32) # array([[1, 1, 1], #        [2, 2, 2], #        [3, 3, 3])) grid_y ## tf.Tensor( ## [[4 5 6] ##  [4 5 6] ##  [4 5 6]], shape=(3, 3), dtype=int32) # array([[4, 5, 6], #        [4, 5, 6], #        [4, 5, 6]))"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_min.html","id":null,"dir":"Reference","previous_headings":"","what":"Return the minimum of a tensor or minimum along an axis. — op_min","title":"Return the minimum of a tensor or minimum along an axis. — op_min","text":"Return minimum tensor minimum along axis.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_min.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Return the minimum of a tensor or minimum along an axis. — op_min","text":"","code":"op_min(x, axis = NULL, keepdims = FALSE, initial = NULL)"},{"path":"https://keras3.posit.co/dev/reference/op_min.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Return the minimum of a tensor or minimum along an axis. — op_min","text":"x Input tensor. axis Axis axes along operate. default, flattened input used. keepdims set TRUE, axes reduced left result dimensions size one. Defaults FALSE. initial maximum value output element. Defaults NULL.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_min.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Return the minimum of a tensor or minimum along an axis. — op_min","text":"Minimum x.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_min.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Return the minimum of a tensor or minimum along an axis. — op_min","text":"","code":"(x <- op_convert_to_tensor(rbind(c(1, 3, 5), c(1, 5, 2)))) ## tf.Tensor( ## [[1. 3. 5.] ##  [1. 5. 2.]], shape=(2, 3), dtype=float64) op_min(x) ## tf.Tensor(1.0, shape=(), dtype=float64) op_min(x, axis = 1) ## tf.Tensor([1. 3. 2.], shape=(3), dtype=float64) op_min(x, axis = 1, keepdims = TRUE) ## tf.Tensor([[1. 3. 2.]], shape=(1, 3), dtype=float64)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_minimum.html","id":null,"dir":"Reference","previous_headings":"","what":"Element-wise minimum of x1 and x2. — op_minimum","title":"Element-wise minimum of x1 and x2. — op_minimum","text":"Element-wise minimum x1 x2.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_minimum.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Element-wise minimum of x1 and x2. — op_minimum","text":"","code":"op_minimum(x1, x2)  op_pmin(x1, x2)"},{"path":"https://keras3.posit.co/dev/reference/op_minimum.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Element-wise minimum of x1 and x2. — op_minimum","text":"x1 First tensor. x2 Second tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_minimum.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Element-wise minimum of x1 and x2. — op_minimum","text":"Output tensor, element-wise minimum x1 x2.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_mod.html","id":null,"dir":"Reference","previous_headings":"","what":"Returns the element-wise remainder of division. — op_mod","title":"Returns the element-wise remainder of division. — op_mod","text":"Note function automatically called using R operator %% tensor.","code":"(x <- op_arange(10)) ## tf.Tensor([ 1.  2.  3.  4.  5.  6.  7.  8.  9. 10.], shape=(10), dtype=float32) op_mod(x, 3) ## tf.Tensor([1. 2. 0. 1. 2. 0. 1. 2. 0. 1.], shape=(10), dtype=float32) x %% 3 ## tf.Tensor([1. 2. 0. 1. 2. 0. 1. 2. 0. 1.], shape=(10), dtype=float32)"},{"path":"https://keras3.posit.co/dev/reference/op_mod.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Returns the element-wise remainder of division. — op_mod","text":"","code":"op_mod(x1, x2)"},{"path":"https://keras3.posit.co/dev/reference/op_mod.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Returns the element-wise remainder of division. — op_mod","text":"x1 First tensor. x2 Second tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_mod.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Returns the element-wise remainder of division. — op_mod","text":"Output tensor, element-wise remainder division.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_moments.html","id":null,"dir":"Reference","previous_headings":"","what":"Calculates the mean and variance of x. — op_moments","title":"Calculates the mean and variance of x. — op_moments","text":"mean variance calculated aggregating contents x across axes. x 1-D axes = c(1) just mean variance vector.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_moments.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Calculates the mean and variance of x. — op_moments","text":"","code":"op_moments(x, axes, keepdims = FALSE, synchronized = FALSE)"},{"path":"https://keras3.posit.co/dev/reference/op_moments.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Calculates the mean and variance of x. — op_moments","text":"x Input tensor. axes list axes compute mean variance. keepdims set TRUE, axes reduced left result dimensions size one. synchronized applicable TensorFlow backend. TRUE, synchronizes global batch statistics (mean variance) across devices training step distributed training strategy. FALSE, replica uses local batch statistics.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_moments.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Calculates the mean and variance of x. — op_moments","text":"list containing two tensors - mean variance.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_moments.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Calculates the mean and variance of x. — op_moments","text":"","code":"x <- op_convert_to_tensor(c(0, 1, 2, 3, 100), dtype = \"float32\") op_moments(x, axes = c(1)) ## [[1]] ## tf.Tensor(21.2, shape=(), dtype=float32) ## ## [[2]] ## tf.Tensor(1553.3601, shape=(), dtype=float32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_moveaxis.html","id":null,"dir":"Reference","previous_headings":"","what":"Move axes of a tensor to new positions. — op_moveaxis","title":"Move axes of a tensor to new positions. — op_moveaxis","text":"axes remain original order.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_moveaxis.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Move axes of a tensor to new positions. — op_moveaxis","text":"","code":"op_moveaxis(x, source, destination)"},{"path":"https://keras3.posit.co/dev/reference/op_moveaxis.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Move axes of a tensor to new positions. — op_moveaxis","text":"x Tensor whose axes reordered. source Original positions axes move. must unique. destination Destinations positions original axes. must also unique.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_moveaxis.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Move axes of a tensor to new positions. — op_moveaxis","text":"Tensor moved axes.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_multi_hot.html","id":null,"dir":"Reference","previous_headings":"","what":"Encodes integer labels as multi-hot vectors. — op_multi_hot","title":"Encodes integer labels as multi-hot vectors. — op_multi_hot","text":"function encodes integer labels multi-hot vectors, label mapped binary value resulting vector.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_multi_hot.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Encodes integer labels as multi-hot vectors. — op_multi_hot","text":"","code":"op_multi_hot(   inputs,   num_classes,   axis = -1L,   dtype = NULL,   sparse = FALSE,   ... )"},{"path":"https://keras3.posit.co/dev/reference/op_multi_hot.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Encodes integer labels as multi-hot vectors. — op_multi_hot","text":"inputs Tensor integer labels converted multi-hot vectors. num_classes Integer, total number unique classes. axis (optional) Axis along multi-hot encoding added. Defaults -1, corresponds last dimension. dtype (optional) data type resulting tensor. Default backend's float type. sparse Whether return sparse tensor; backends support sparse tensors. ... forward/backwards compatability","code":""},{"path":"https://keras3.posit.co/dev/reference/op_multi_hot.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Encodes integer labels as multi-hot vectors. — op_multi_hot","text":"Tensor: multi-hot encoded tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_multi_hot.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Encodes integer labels as multi-hot vectors. — op_multi_hot","text":"","code":"data <- op_convert_to_tensor(c(0, 4)) op_multi_hot(data, num_classes = 5) ## tf.Tensor([1. 0. 0. 0. 1.], shape=(5), dtype=float32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_multiply.html","id":null,"dir":"Reference","previous_headings":"","what":"Multiply arguments element-wise. — op_multiply","title":"Multiply arguments element-wise. — op_multiply","text":"Note function automatically called using R operator * tensor.","code":"(x <- op_arange(4)) ## tf.Tensor([1. 2. 3. 4.], shape=(4), dtype=float32) op_multiply(x, x) ## tf.Tensor([ 1.  4.  9. 16.], shape=(4), dtype=float32) x * x ## tf.Tensor([ 1.  4.  9. 16.], shape=(4), dtype=float32)"},{"path":"https://keras3.posit.co/dev/reference/op_multiply.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Multiply arguments element-wise. — op_multiply","text":"","code":"op_multiply(x1, x2)"},{"path":"https://keras3.posit.co/dev/reference/op_multiply.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Multiply arguments element-wise. — op_multiply","text":"x1 First input tensor. x2 Second input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_multiply.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Multiply arguments element-wise. — op_multiply","text":"Output tensor, element-wise product x1 x2.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_nan_to_num.html","id":null,"dir":"Reference","previous_headings":"","what":"Replace NaN with zero and infinity with large finite numbers. — op_nan_to_num","title":"Replace NaN with zero and infinity with large finite numbers. — op_nan_to_num","text":"Replace NaN zero infinity large finite numbers.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_nan_to_num.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Replace NaN with zero and infinity with large finite numbers. — op_nan_to_num","text":"","code":"op_nan_to_num(x, nan = 0, posinf = NULL, neginf = NULL)"},{"path":"https://keras3.posit.co/dev/reference/op_nan_to_num.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Replace NaN with zero and infinity with large finite numbers. — op_nan_to_num","text":"x Input data. nan Optional float int. Value replace NaN entries . posinf Optional float int. Value replace positive infinity . neginf Optional float int. Value replace negative infinity .","code":""},{"path":"https://keras3.posit.co/dev/reference/op_nan_to_num.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Replace NaN with zero and infinity with large finite numbers. — op_nan_to_num","text":"x, non-finite values replaced.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_nan_to_num.html","id":"example","dir":"Reference","previous_headings":"","what":"Example","title":"Replace NaN with zero and infinity with large finite numbers. — op_nan_to_num","text":"","code":"(x <- op_convert_to_tensor(c(1, NaN, -Inf, Inf))) ## tf.Tensor([  1.  nan -inf  inf], shape=(4), dtype=float64) op_nan_to_num(x) ## tf.Tensor([ 1.00000000e+000  0.00000000e+000 -1.79769313e+308  1.79769313e+308], shape=(4), dtype=float64) op_nan_to_num(x, nan = -1, posinf = 2, neginf = -2) ## tf.Tensor([ 1. -1. -2.  2.], shape=(4), dtype=float64)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_ndim.html","id":null,"dir":"Reference","previous_headings":"","what":"Return the number of dimensions of a tensor. — op_ndim","title":"Return the number of dimensions of a tensor. — op_ndim","text":"Return number dimensions tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_ndim.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Return the number of dimensions of a tensor. — op_ndim","text":"","code":"op_ndim(x)"},{"path":"https://keras3.posit.co/dev/reference/op_ndim.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Return the number of dimensions of a tensor. — op_ndim","text":"x Input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_ndim.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Return the number of dimensions of a tensor. — op_ndim","text":"number dimensions x.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_negative.html","id":null,"dir":"Reference","previous_headings":"","what":"Numerical negative, element-wise. — op_negative","title":"Numerical negative, element-wise. — op_negative","text":"Note function automatically called using unary R operator - tensor.","code":"(x <- op_arange(4)) ## tf.Tensor([1. 2. 3. 4.], shape=(4), dtype=float32) op_negative(x) ## tf.Tensor([-1. -2. -3. -4.], shape=(4), dtype=float32) -x ## tf.Tensor([-1. -2. -3. -4.], shape=(4), dtype=float32)"},{"path":"https://keras3.posit.co/dev/reference/op_negative.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Numerical negative, element-wise. — op_negative","text":"","code":"op_negative(x)"},{"path":"https://keras3.posit.co/dev/reference/op_negative.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Numerical negative, element-wise. — op_negative","text":"x Input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_negative.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Numerical negative, element-wise. — op_negative","text":"Output tensor, y = -x.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_nonzero.html","id":null,"dir":"Reference","previous_headings":"","what":"Return the indices of the elements that are non-zero. — op_nonzero","title":"Return the indices of the elements that are non-zero. — op_nonzero","text":"Return indices elements non-zero.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_nonzero.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Return the indices of the elements that are non-zero. — op_nonzero","text":"","code":"op_nonzero(x)"},{"path":"https://keras3.posit.co/dev/reference/op_nonzero.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Return the indices of the elements that are non-zero. — op_nonzero","text":"x Input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_nonzero.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Return the indices of the elements that are non-zero. — op_nonzero","text":"Indices elements non-zero.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_nonzero.html","id":"example","dir":"Reference","previous_headings":"","what":"Example","title":"Return the indices of the elements that are non-zero. — op_nonzero","text":"op_nonzero() indices can used <tensor>@r[","code":"(x <- op_scatter(indices = rbind(1, 5, 10), values = c(1, 2, 3), shape = c(10))) ## tf.Tensor([1. 0. 0. 0. 2. 0. 0. 0. 0. 3.], shape=(10), dtype=float64) (nz <- op_nonzero(x)) ## tf.Tensor( ## [[ 1] ##  [ 5] ##  [10]], shape=(3, 1), dtype=int32) x@r[nz] ## tf.Tensor([1. 2. 3.], shape=(3), dtype=float64) # same as `x@r[nz]` x@r[x != 0] ## tf.Tensor([1. 2. 3.], shape=(3), dtype=float64) x@r[op_cast(x, \"bool\")] ## tf.Tensor([1. 2. 3.], shape=(3), dtype=float64) # 2d example (x2 <- op_stack(c(x, op_roll(x, 1), op_roll(x, 2)))) ## tf.Tensor( ## [[1. 0. 0. 0. 2. 0. 0. 0. 0. 3.] ##  [3. 1. 0. 0. 0. 2. 0. 0. 0. 0.] ##  [0. 3. 1. 0. 0. 0. 2. 0. 0. 0.]], shape=(3, 10), dtype=float64) x2@r[op_nonzero(x2)] ## tf.Tensor([1. 2. 3. 3. 1. 2. 3. 1. 2.], shape=(9), dtype=float64) x3 <- op_stack(c(x2, x2*1.1)) x3@r[op_nonzero(x3)] ## tf.Tensor([1.  2.  3.  3.  1.  2.  3.  1.  2.  1.1 2.2 3.3 3.3 1.1 2.2 3.3 1.1 2.2], shape=(18), dtype=float64)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_norm.html","id":null,"dir":"Reference","previous_headings":"","what":"Matrix or vector norm. — op_norm","title":"Matrix or vector norm. — op_norm","text":"function able return one eight different matrix norms, one infinite number vector norms (described ), depending value ord parameter.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_norm.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Matrix or vector norm. — op_norm","text":"","code":"op_norm(x, ord = NULL, axis = NULL, keepdims = FALSE)"},{"path":"https://keras3.posit.co/dev/reference/op_norm.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Matrix or vector norm. — op_norm","text":"x Input tensor. ord Order norm (see table Notes). default NULL. axis axis integer, specifies axis x along compute vector norms. axis length 2 vector, specifies axes hold 2-D matrices, matrix norms matrices computed. keepdims set TRUE, axes reduced left result dimensions size one.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_norm.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Matrix or vector norm. — op_norm","text":"Norm matrix vector(s).","code":""},{"path":"https://keras3.posit.co/dev/reference/op_norm.html","id":"note","dir":"Reference","previous_headings":"","what":"Note","title":"Matrix or vector norm. — op_norm","text":"values ord < 1, result , strictly speaking, mathematical 'norm', may still useful various numerical purposes. following norms can calculated: matrices: ord=NULL: Frobenius norm ord=\"fro\": Frobenius norm ord=\"nuc\": nuclear norm ord=Inf: max(sum(abs(x), axis=2)) ord=-Inf: min(sum(abs(x), axis=2)) ord=0: supported ord=1: max(sum(abs(x), axis=1)) ord=-1: min(sum(abs(x), axis=1)) ord=2: 2-norm (largest sing. value) ord=-2: smallest singular value : supported vectors: ord=NULL: 2-norm ord=\"fro\": supported ord=\"nuc\": supported ord=Inf: max(abs(x)) ord=-Inf: min(abs(x)) ord=0: sum(x != 0) ord=1: ord=-1: ord=2: ord=-2: : sum(abs(x)^ord)^(1/ord)","code":""},{"path":"https://keras3.posit.co/dev/reference/op_norm.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Matrix or vector norm. — op_norm","text":"","code":"x <- op_reshape(op_arange(9, dtype=\"float32\") - 4, c(3, 3)) op_norm(x) ## tf.Tensor(8.306623, shape=(), dtype=float32) # 7.7459664"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_normalize.html","id":null,"dir":"Reference","previous_headings":"","what":"Normalizes x over the specified axis. — op_normalize","title":"Normalizes x over the specified axis. — op_normalize","text":"defined : op_normalize(x) = x / max(norm(x), epsilon).","code":""},{"path":"https://keras3.posit.co/dev/reference/op_normalize.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Normalizes x over the specified axis. — op_normalize","text":"","code":"op_normalize(x, axis = -1L, order = 2L, epsilon = NULL)"},{"path":"https://keras3.posit.co/dev/reference/op_normalize.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Normalizes x over the specified axis. — op_normalize","text":"x Input tensor. axis axis axes along perform normalization. Default -1. order exponent value norm formulation. Defaults 2. epsilon lower bound value norm. Defaults config_epsilon().","code":""},{"path":"https://keras3.posit.co/dev/reference/op_normalize.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Normalizes x over the specified axis. — op_normalize","text":"normalized array.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_normalize.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Normalizes x over the specified axis. — op_normalize","text":"","code":"x <- op_convert_to_tensor(rbind(c(1, 2, 3), c(4, 5, 6))) x_norm <- op_normalize(x) x_norm ## tf.Tensor( ## [[0.26726124 0.53452248 0.80178373] ##  [0.45584231 0.56980288 0.68376346]], shape=(2, 3), dtype=float64)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_not_equal.html","id":null,"dir":"Reference","previous_headings":"","what":"Return (x1 != x2) element-wise. — op_not_equal","title":"Return (x1 != x2) element-wise. — op_not_equal","text":"Note function automatically called using R operator != tensor.","code":"(x <- op_arange(4)) ## tf.Tensor([1. 2. 3. 4.], shape=(4), dtype=float32) op_not_equal(x, 2) ## tf.Tensor([ True False  True  True], shape=(4), dtype=bool) x != 2 ## tf.Tensor([ True False  True  True], shape=(4), dtype=bool)"},{"path":"https://keras3.posit.co/dev/reference/op_not_equal.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Return (x1 != x2) element-wise. — op_not_equal","text":"","code":"op_not_equal(x1, x2)"},{"path":"https://keras3.posit.co/dev/reference/op_not_equal.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Return (x1 != x2) element-wise. — op_not_equal","text":"x1 First input tensor. x2 Second input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_not_equal.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Return (x1 != x2) element-wise. — op_not_equal","text":"Output tensor, element-wise comparison x1 x2.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_one_hot.html","id":null,"dir":"Reference","previous_headings":"","what":"Converts integer tensor x into a one-hot tensor. — op_one_hot","title":"Converts integer tensor x into a one-hot tensor. — op_one_hot","text":"one-hot encoding representation integer value converted binary vector length equal num_classes, index corresponding integer value marked 1, indices marked 0.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_one_hot.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Converts integer tensor x into a one-hot tensor. — op_one_hot","text":"","code":"op_one_hot(   x,   num_classes,   axis = -1L,   dtype = NULL,   sparse = FALSE,   zero_indexed = FALSE )"},{"path":"https://keras3.posit.co/dev/reference/op_one_hot.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Converts integer tensor x into a one-hot tensor. — op_one_hot","text":"x Integer tensor encoded. shape can arbitrary, dtype integer. R factors coerced integer. num_classes Number classes one-hot encoding. x factor num_classes NULL missing, levels(x) used. axis Axis along encoding performed. -1 represents last axis. Defaults -1. dtype (Optional) Data type output tensor. provided, defaults default data type backend. sparse Whether return sparse tensor; backends support sparse tensors. zero_indexed TRUE, treats indices zero-based (0 encodes first position); FALSE (default), treats indices one-based (1 encodes first position).","code":""},{"path":"https://keras3.posit.co/dev/reference/op_one_hot.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Converts integer tensor x into a one-hot tensor. — op_one_hot","text":"Integer tensor: One-hot encoded tensor shape x except specified axis dimension, length num_classes. dtype output tensor determined dtype default data type backend.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_one_hot.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Converts integer tensor x into a one-hot tensor. — op_one_hot","text":"","code":"x <- op_array(c(2, 4, 3, 1), \"int32\") op_one_hot(x, num_classes = 4) ## tf.Tensor( ## [[0. 1. 0. 0.] ##  [0. 0. 0. 1.] ##  [0. 0. 1. 0.] ##  [1. 0. 0. 0.]], shape=(4, 4), dtype=float32) op_one_hot(x - 1, num_classes = 4, zero_indexed = TRUE) ## tf.Tensor( ## [[0. 1. 0. 0.] ##  [0. 0. 0. 1.] ##  [0. 0. 1. 0.] ##  [1. 0. 0. 0.]], shape=(4, 4), dtype=float32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_ones.html","id":null,"dir":"Reference","previous_headings":"","what":"Return a new tensor of given shape and type, filled with ones. — op_ones","title":"Return a new tensor of given shape and type, filled with ones. — op_ones","text":"Return new tensor given shape type, filled ones.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_ones.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Return a new tensor of given shape and type, filled with ones. — op_ones","text":"","code":"op_ones(shape, dtype = NULL)"},{"path":"https://keras3.posit.co/dev/reference/op_ones.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Return a new tensor of given shape and type, filled with ones. — op_ones","text":"shape Shape new tensor. dtype Desired data type tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_ones.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Return a new tensor of given shape and type, filled with ones. — op_ones","text":"Tensor ones given shape dtype.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_ones_like.html","id":null,"dir":"Reference","previous_headings":"","what":"Return a tensor of ones with the same shape and type of x. — op_ones_like","title":"Return a tensor of ones with the same shape and type of x. — op_ones_like","text":"Return tensor ones shape type x.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_ones_like.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Return a tensor of ones with the same shape and type of x. — op_ones_like","text":"","code":"op_ones_like(x, dtype = NULL)"},{"path":"https://keras3.posit.co/dev/reference/op_ones_like.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Return a tensor of ones with the same shape and type of x. — op_ones_like","text":"x Input tensor. dtype Overrides data type result.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_ones_like.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Return a tensor of ones with the same shape and type of x. — op_ones_like","text":"tensor ones shape type x.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_outer.html","id":null,"dir":"Reference","previous_headings":"","what":"Compute the outer product of two vectors. — op_outer","title":"Compute the outer product of two vectors. — op_outer","text":"Given two vectors x1 x2, outer product :","code":"out[i, j] = x1[i] * x2[j]"},{"path":"https://keras3.posit.co/dev/reference/op_outer.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Compute the outer product of two vectors. — op_outer","text":"","code":"op_outer(x1, x2)"},{"path":"https://keras3.posit.co/dev/reference/op_outer.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Compute the outer product of two vectors. — op_outer","text":"x1 First input tensor. x2 Second input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_outer.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Compute the outer product of two vectors. — op_outer","text":"Outer product x1 x2.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_pad.html","id":null,"dir":"Reference","previous_headings":"","what":"Pad a tensor. — op_pad","title":"Pad a tensor. — op_pad","text":"Pad tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_pad.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Pad a tensor. — op_pad","text":"","code":"op_pad(x, pad_width, mode = \"constant\", constant_values = NULL)"},{"path":"https://keras3.posit.co/dev/reference/op_pad.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Pad a tensor. — op_pad","text":"x Tensor pad. pad_width Number values padded edges axis. ((before_1, after_1), ...(before_N, after_N)) unique pad widths axis. ((, ),) yields pad axis. (pad,) int shortcut = = pad width axes. mode One \"constant\", \"edge\", \"linear_ramp\", \"maximum\", \"mean\", \"median\", \"minimum\", \"reflect\", \"symmetric\", \"wrap\", \"empty\", \"circular\". Defaults \"constant\". constant_values Value pad mode == \"constant\". Defaults 0. ValueError raised NULL mode != \"constant\".","code":""},{"path":"https://keras3.posit.co/dev/reference/op_pad.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Pad a tensor. — op_pad","text":"Padded tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_pad.html","id":"note","dir":"Reference","previous_headings":"","what":"Note","title":"Pad a tensor. — op_pad","text":"Torch backend supports modes \"constant\", \"reflect\", \"symmetric\" \"circular\". Torch backend supports \"circular\" mode. Note: Tensorflow backend supports modes \"constant\", \"reflect\" \"symmetric\".","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_polar.html","id":null,"dir":"Reference","previous_headings":"","what":"Constructs a complex tensor whose elements are Cartesian — op_polar","title":"Constructs a complex tensor whose elements are Cartesian — op_polar","text":"coordinates corresponding polar coordinates absolute value abs angle angle. operation numerically equivalent torch.polar(). equivalent scipy.lingalg.polar() performs Singular Value Decomposition. Given magnitude (abs_) angle (angle), function computes corresponding complex number form real + imaginary * 1i, : real = abs_ * cos(angle) imaginary = abs_ * sin(angle)","code":""},{"path":"https://keras3.posit.co/dev/reference/op_polar.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Constructs a complex tensor whose elements are Cartesian — op_polar","text":"","code":"op_polar(abs_, angle)"},{"path":"https://keras3.posit.co/dev/reference/op_polar.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Constructs a complex tensor whose elements are Cartesian — op_polar","text":"abs_ magnitude (absolute value) complex number. angle angle (radians) complex number.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_polar.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Constructs a complex tensor whose elements are Cartesian — op_polar","text":"complex number (array complex numbers) shape abs_ angle.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_polar.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Constructs a complex tensor whose elements are Cartesian — op_polar","text":"","code":"abs_ <- random_normal(c(1, 2)) angle <- random_normal(c(1, 2)) op_shape(op_polar(abs_, angle)) ## shape(1, 2) op_polar(abs_, angle) ## tf.Tensor([[-0.10252866-1.2460487j -0.2586355 -1.0352845j]], shape=(1, 2), dtype=complex64)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_power.html","id":null,"dir":"Reference","previous_headings":"","what":"First tensor elements raised to powers from second tensor, element-wise. — op_power","title":"First tensor elements raised to powers from second tensor, element-wise. — op_power","text":"Note function automatically called using R operator ^ tensor.","code":"(x <- op_arange(4)) ## tf.Tensor([1. 2. 3. 4.], shape=(4), dtype=float32) op_power(2, x) ## tf.Tensor([ 2.  4.  8. 16.], shape=(4), dtype=float32) 2 ^ x ## tf.Tensor([ 2.  4.  8. 16.], shape=(4), dtype=float32)"},{"path":"https://keras3.posit.co/dev/reference/op_power.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"First tensor elements raised to powers from second tensor, element-wise. — op_power","text":"","code":"op_power(x1, x2)"},{"path":"https://keras3.posit.co/dev/reference/op_power.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"First tensor elements raised to powers from second tensor, element-wise. — op_power","text":"x1 bases. x2 exponents.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_power.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"First tensor elements raised to powers from second tensor, element-wise. — op_power","text":"Output tensor, bases x1 raised exponents x2.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_prod.html","id":null,"dir":"Reference","previous_headings":"","what":"Return the product of tensor elements over a given axis. — op_prod","title":"Return the product of tensor elements over a given axis. — op_prod","text":"Return product tensor elements given axis.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_prod.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Return the product of tensor elements over a given axis. — op_prod","text":"","code":"op_prod(x, axis = NULL, keepdims = FALSE, dtype = NULL)"},{"path":"https://keras3.posit.co/dev/reference/op_prod.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Return the product of tensor elements over a given axis. — op_prod","text":"x Input tensor. axis Axis axes along product performed. default, axis = NULL, compute product elements input tensor. keepdims set TRUE, axes reduce left result dimensions size one. dtype Data type returned tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_prod.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Return the product of tensor elements over a given axis. — op_prod","text":"Product elements x given axis axes.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_psnr.html","id":null,"dir":"Reference","previous_headings":"","what":"Peak Signal-to-Noise Ratio (PSNR) function. — op_psnr","title":"Peak Signal-to-Noise Ratio (PSNR) function. — op_psnr","text":"function computes Peak Signal--Noise Ratio two signals, x1 x2. PSNR measure quality reconstructed signal. higher PSNR, closer reconstructed signal original signal. Note can become negative signal power smaller noise power.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_psnr.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Peak Signal-to-Noise Ratio (PSNR) function. — op_psnr","text":"","code":"op_psnr(x1, x2, max_val)"},{"path":"https://keras3.posit.co/dev/reference/op_psnr.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Peak Signal-to-Noise Ratio (PSNR) function. — op_psnr","text":"x1 first input signal. x2 second input signal. Must shape x1. max_val maximum possible value signals.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_psnr.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Peak Signal-to-Noise Ratio (PSNR) function. — op_psnr","text":"float: PSNR value x1 x2.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_psnr.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Peak Signal-to-Noise Ratio (PSNR) function. — op_psnr","text":"","code":"x1 <- random_normal(c(2, 4, 4, 3)) x2 <- random_normal(c(2, 4, 4, 3)) max_val <- 1.0 op_psnr(x1, x2, max_val) ## tf.Tensor(-3.5293808, shape=(), dtype=float32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_qr.html","id":null,"dir":"Reference","previous_headings":"","what":"Computes the QR decomposition of a tensor. — op_qr","title":"Computes the QR decomposition of a tensor. — op_qr","text":"Computes QR decomposition tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_qr.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes the QR decomposition of a tensor. — op_qr","text":"","code":"op_qr(x, mode = \"reduced\")"},{"path":"https://keras3.posit.co/dev/reference/op_qr.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Computes the QR decomposition of a tensor. — op_qr","text":"x Input tensor shape (..., M, N). mode string specifying mode QR decomposition. 'reduced': Returns reduced QR decomposition. (default) 'complete': Returns complete QR decomposition.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_qr.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Computes the QR decomposition of a tensor. — op_qr","text":"list containing two tensors. first tensor shape (..., M, K) orthogonal matrix q second tensor shape (..., K, N)upper triangular matrixr, K = min(M, N)`.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_qr.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Computes the QR decomposition of a tensor. — op_qr","text":"","code":"x <- op_convert_to_tensor(rbind(c(1, 2), c(3, 4), c(5, 6))) op_qr(x) ## $q ## tf.Tensor( ## [[-0.16903085  0.89708523] ##  [-0.50709255  0.27602622] ##  [-0.84515425 -0.34503278]], shape=(3, 2), dtype=float64) ## ## $r ## tf.Tensor( ## [[-5.91607978 -7.43735744] ##  [ 0.          0.82807867]], shape=(2, 2), dtype=float64) c(q, r) %<-% op_qr(x)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_quantile.html","id":null,"dir":"Reference","previous_headings":"","what":"Compute the q-th quantile(s) of the data along the specified axis. — op_quantile","title":"Compute the q-th quantile(s) of the data along the specified axis. — op_quantile","text":"Compute q-th quantile(s) data along specified axis.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_quantile.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Compute the q-th quantile(s) of the data along the specified axis. — op_quantile","text":"","code":"op_quantile(x, q, axis = NULL, method = \"linear\", keepdims = FALSE)"},{"path":"https://keras3.posit.co/dev/reference/op_quantile.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Compute the q-th quantile(s) of the data along the specified axis. — op_quantile","text":"x Input tensor. q Probability sequence probabilities quantiles compute. Values must 0 1 inclusive. axis Axis axes along quantiles computed. Defaults axis=NULL compute quantile(s) along flattened version array. method string specifies method use estimating quantile. Available methods \"linear\", \"lower\", \"higher\", \"midpoint\", \"nearest\". Defaults \"linear\". desired quantile lies two data points < j: \"linear\": + (j - ) * fraction, fraction fractional part index surrounded j. \"lower\": . \"higher\": j. \"midpoint\": (+ j) / 2 \"nearest\": j, whichever nearest. keepdims set TRUE, axes reduce left result dimensions size one.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_quantile.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Compute the q-th quantile(s) of the data along the specified axis. — op_quantile","text":"quantile(s). q single probability axis = NULL, result scalar. multiple probabilities levels given, first axis result corresponds quantiles. axes axes remain reduction x.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_ravel.html","id":null,"dir":"Reference","previous_headings":"","what":"Return a contiguous flattened tensor. — op_ravel","title":"Return a contiguous flattened tensor. — op_ravel","text":"1-D tensor, containing elements input, returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_ravel.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Return a contiguous flattened tensor. — op_ravel","text":"","code":"op_ravel(x)"},{"path":"https://keras3.posit.co/dev/reference/op_ravel.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Return a contiguous flattened tensor. — op_ravel","text":"x Input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_ravel.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Return a contiguous flattened tensor. — op_ravel","text":"Output tensor.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_real.html","id":null,"dir":"Reference","previous_headings":"","what":"Return the real part of the complex argument. — op_real","title":"Return the real part of the complex argument. — op_real","text":"Return real part complex argument.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_real.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Return the real part of the complex argument. — op_real","text":"","code":"op_real(x)"},{"path":"https://keras3.posit.co/dev/reference/op_real.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Return the real part of the complex argument. — op_real","text":"x Input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_real.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Return the real part of the complex argument. — op_real","text":"real component complex argument.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_rearrange.html","id":null,"dir":"Reference","previous_headings":"","what":"Rearranges the axes of a Keras tensor according to a specified pattern, — op_rearrange","title":"Rearranges the axes of a Keras tensor according to a specified pattern, — op_rearrange","text":"einops-style.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_rearrange.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Rearranges the axes of a Keras tensor according to a specified pattern, — op_rearrange","text":"","code":"op_rearrange(tensor, pattern, ...)"},{"path":"https://keras3.posit.co/dev/reference/op_rearrange.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Rearranges the axes of a Keras tensor according to a specified pattern, — op_rearrange","text":"tensor Input Keras tensor. pattern String describing rearrangement einops notation. ... axes_lengths, named arguments specifying lengths axes axes decomposition used.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_rearrange.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Rearranges the axes of a Keras tensor according to a specified pattern, — op_rearrange","text":"Tensor: Keras tensor rearranged axes. Follows logic : decomposition needed, reshape match decomposed dimensions. Permute known inferred axes match form output. Reshape match desired output shape. Example Usage:","code":"images <- op_ones(c(32, 30, 40, 3)) # BHWC format  # Reordering to BCHW op_rearrange(images, 'b h w c -> b c h w') |> op_shape() ## shape(32, 3, 30, 40) # \"Merge\" along first axis - concat images from a batch op_rearrange(images, 'b h w c -> (b h) w c') |> op_shape() ## shape(960, 40, 3) # \"Merge\" along second axis - concat images horizontally op_rearrange(images, 'b h w c -> h (b w) c') |> op_shape() ## shape(30, 1280, 3) # Flatten images into a CHW vector op_rearrange(images, 'b h w c -> b (c h w)') |> op_shape() ## shape(32, 3600) # Decompose H and W axes into 4 smaller patches op_rearrange(images, 'b (h1 h) (w1 w) c -> (b h1 w1) h w c', h1 = 2, w1 = 2) |> op_shape() ## shape(128, 15, 20, 3) # Space-to-depth decomposition of input axes op_rearrange(images, 'b (h h1) (w w1) c -> b h w (c h1 w1)', h1 = 2, w1 = 2) |> op_shape() ## shape(32, 15, 20, 12)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_reciprocal.html","id":null,"dir":"Reference","previous_headings":"","what":"Return the reciprocal of the argument, element-wise. — op_reciprocal","title":"Return the reciprocal of the argument, element-wise. — op_reciprocal","text":"Calculates 1/x.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_reciprocal.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Return the reciprocal of the argument, element-wise. — op_reciprocal","text":"","code":"op_reciprocal(x)"},{"path":"https://keras3.posit.co/dev/reference/op_reciprocal.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Return the reciprocal of the argument, element-wise. — op_reciprocal","text":"x Input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_reciprocal.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Return the reciprocal of the argument, element-wise. — op_reciprocal","text":"Output tensor, element-wise reciprocal x.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_relu.html","id":null,"dir":"Reference","previous_headings":"","what":"Rectified linear unit activation function. — op_relu","title":"Rectified linear unit activation function. — op_relu","text":"defined f(x) = max(0, x).","code":""},{"path":"https://keras3.posit.co/dev/reference/op_relu.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Rectified linear unit activation function. — op_relu","text":"","code":"op_relu(x)"},{"path":"https://keras3.posit.co/dev/reference/op_relu.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Rectified linear unit activation function. — op_relu","text":"x Input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_relu.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Rectified linear unit activation function. — op_relu","text":"tensor shape x.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_relu.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Rectified linear unit activation function. — op_relu","text":"","code":"x1 <- op_convert_to_tensor(c(-1, 0, 1, 0.2)) op_relu(x1) ## tf.Tensor([0.  0.  1.  0.2], shape=(4), dtype=float64) x <- seq(-10, 10, .1) plot(x, op_relu(x))"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_relu6.html","id":null,"dir":"Reference","previous_headings":"","what":"Rectified linear unit activation function with upper bound of 6. — op_relu6","title":"Rectified linear unit activation function with upper bound of 6. — op_relu6","text":"defined f(x) = op_clip(x, 0, 6).","code":""},{"path":"https://keras3.posit.co/dev/reference/op_relu6.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Rectified linear unit activation function with upper bound of 6. — op_relu6","text":"","code":"op_relu6(x)"},{"path":"https://keras3.posit.co/dev/reference/op_relu6.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Rectified linear unit activation function with upper bound of 6. — op_relu6","text":"x Input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_relu6.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Rectified linear unit activation function with upper bound of 6. — op_relu6","text":"tensor shape x.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_relu6.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Rectified linear unit activation function with upper bound of 6. — op_relu6","text":"","code":"x <- op_convert_to_tensor(c(-3, -2, 0.1, 0.2, 6, 8)) op_relu6(x) ## tf.Tensor([0.  0.  0.1 0.2 6.  6. ], shape=(6), dtype=float64) x <- seq(-10, 10, .1) plot(x, op_relu6(x))"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_repeat.html","id":null,"dir":"Reference","previous_headings":"","what":"Repeat each element of a tensor after themselves. — op_repeat","title":"Repeat each element of a tensor after themselves. — op_repeat","text":"Repeat element tensor .","code":""},{"path":"https://keras3.posit.co/dev/reference/op_repeat.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Repeat each element of a tensor after themselves. — op_repeat","text":"","code":"op_repeat(x, repeats, axis = NULL)"},{"path":"https://keras3.posit.co/dev/reference/op_repeat.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Repeat each element of a tensor after themselves. — op_repeat","text":"x Input tensor. repeats number repetitions element. axis axis along repeat values. default, use flattened input array, return flat output array.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_repeat.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Repeat each element of a tensor after themselves. — op_repeat","text":"Output tensor.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_reshape.html","id":null,"dir":"Reference","previous_headings":"","what":"Gives a new shape to a tensor without changing its data. — op_reshape","title":"Gives a new shape to a tensor without changing its data. — op_reshape","text":"Gives new shape tensor without changing data.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_reshape.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Gives a new shape to a tensor without changing its data. — op_reshape","text":"","code":"op_reshape(x, newshape)"},{"path":"https://keras3.posit.co/dev/reference/op_reshape.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Gives a new shape to a tensor without changing its data. — op_reshape","text":"x Input tensor. newshape new shape compatible original shape. One shape dimension can -1 case value inferred length array remaining dimensions.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_reshape.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Gives a new shape to a tensor without changing its data. — op_reshape","text":"reshaped tensor.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_rfft.html","id":null,"dir":"Reference","previous_headings":"","what":"Real-valued Fast Fourier Transform along the last axis of the input. — op_rfft","title":"Real-valued Fast Fourier Transform along the last axis of the input. — op_rfft","text":"Computes 1D Discrete Fourier Transform real-valued signal inner-dimension input. Since Discrete Fourier Transform real-valued signal Hermitian-symmetric, RFFT returns fft_length / 2 + 1 unique components FFT: zero-frequency term, followed fft_length / 2 positive-frequency terms. Along axis RFFT computed , fft_length smaller corresponding dimension input, dimension cropped. larger, dimension padded zeros.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_rfft.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Real-valued Fast Fourier Transform along the last axis of the input. — op_rfft","text":"","code":"op_rfft(x, fft_length = NULL)"},{"path":"https://keras3.posit.co/dev/reference/op_rfft.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Real-valued Fast Fourier Transform along the last axis of the input. — op_rfft","text":"x Input tensor. fft_length integer representing number fft length. specified, inferred length last axis x. Defaults NULL.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_rfft.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Real-valued Fast Fourier Transform along the last axis of the input. — op_rfft","text":"list containing two tensors - real imaginary parts output.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_rfft.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Real-valued Fast Fourier Transform along the last axis of the input. — op_rfft","text":"","code":"x <- op_convert_to_tensor(c(0, 1, 2, 3, 4)) op_rfft(x) ## [[1]] ## tf.Tensor([10.  -2.5 -2.5], shape=(3), dtype=float64) ## ## [[2]] ## tf.Tensor([0.         3.4409548  0.81229924], shape=(3), dtype=float64) op_rfft(x, 3) ## [[1]] ## tf.Tensor([ 3.  -1.5], shape=(2), dtype=float64) ## ## [[2]] ## tf.Tensor([0.        0.8660254], shape=(2), dtype=float64)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_right_shift.html","id":null,"dir":"Reference","previous_headings":"","what":"Shift the bits of an integer to the right. — op_right_shift","title":"Shift the bits of an integer to the right. — op_right_shift","text":"Bits shifted right y. internal representation numbers binary format, operation equivalent dividing x 2^y.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_right_shift.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Shift the bits of an integer to the right. — op_right_shift","text":"","code":"op_right_shift(x, y)"},{"path":"https://keras3.posit.co/dev/reference/op_right_shift.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Shift the bits of an integer to the right. — op_right_shift","text":"x Input integer tensor. y Input integer tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_right_shift.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Shift the bits of an integer to the right. — op_right_shift","text":"Result tensor.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_rms_normalization.html","id":null,"dir":"Reference","previous_headings":"","what":"Performs Root Mean Square (RMS) normalization on x. — op_rms_normalization","title":"Performs Root Mean Square (RMS) normalization on x. — op_rms_normalization","text":"Keras operation implements operation described Root Mean Square Layer Normalization Biao Zhang et al. operation different LayerNormalization RMS scaling. defined rms_normalization(x) = x * rsqrt(mean(square(x))) * scale","code":""},{"path":"https://keras3.posit.co/dev/reference/op_rms_normalization.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Performs Root Mean Square (RMS) normalization on x. — op_rms_normalization","text":"","code":"op_rms_normalization(x, scale = 1L, axis = -1L, epsilon = NULL)"},{"path":"https://keras3.posit.co/dev/reference/op_rms_normalization.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Performs Root Mean Square (RMS) normalization on x. — op_rms_normalization","text":"x Input tensor. scale Optional scaling factor normalization. axis axis axes along perform normalization. Default -1. epsilon lower bound value norm. Defaults config_epsilon().","code":""},{"path":"https://keras3.posit.co/dev/reference/op_rms_normalization.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Performs Root Mean Square (RMS) normalization on x. — op_rms_normalization","text":"normalized array.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_rms_normalization.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Performs Root Mean Square (RMS) normalization on x. — op_rms_normalization","text":"","code":"x <- random_uniform(c(1, 10)) x_norm <- op_rms_normalization(x, scale = 10) x_norm"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_roll.html","id":null,"dir":"Reference","previous_headings":"","what":"Roll tensor elements along a given axis. — op_roll","title":"Roll tensor elements along a given axis. — op_roll","text":"Elements roll beyond last position re-introduced first.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_roll.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Roll tensor elements along a given axis. — op_roll","text":"","code":"op_roll(x, shift, axis = NULL)"},{"path":"https://keras3.posit.co/dev/reference/op_roll.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Roll tensor elements along a given axis. — op_roll","text":"x Input tensor. shift number places elements shifted. axis axis along elements shifted. default, array flattened shifting, original shape restored.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_roll.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Roll tensor elements along a given axis. — op_roll","text":"Output tensor.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_rot90.html","id":null,"dir":"Reference","previous_headings":"","what":"Rotate an array by 90 degrees in the plane specified by axes. — op_rot90","title":"Rotate an array by 90 degrees in the plane specified by axes. — op_rot90","text":"function rotates array counterclockwise 90 degrees k times plane specified axes. Supports arrays two dimensions.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_rot90.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Rotate an array by 90 degrees in the plane specified by axes. — op_rot90","text":"","code":"op_rot90(array, k = 1L, axes = list(1L, 2L))"},{"path":"https://keras3.posit.co/dev/reference/op_rot90.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Rotate an array by 90 degrees in the plane specified by axes. — op_rot90","text":"array Input array rotate. k Number times array rotated 90 degrees. axes tuple two integers specifying plane rotation (defaults (1, 2)).","code":""},{"path":"https://keras3.posit.co/dev/reference/op_rot90.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Rotate an array by 90 degrees in the plane specified by axes. — op_rot90","text":"Rotated array.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_rot90.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Rotate an array by 90 degrees in the plane specified by axes. — op_rot90","text":"","code":"m <- 1:4 |> op_reshape(c(2, 2)) m ## tf.Tensor( ## [[1 2] ##  [3 4]], shape=(2, 2), dtype=int32) op_rot90(m) ## tf.Tensor( ## [[2 4] ##  [1 3]], shape=(2, 2), dtype=int32) m <- 1:8 |> op_reshape(c(2, 2, 2)) m ## tf.Tensor( ## [[[1 2] ##   [3 4]] ## ##  [[5 6] ##   [7 8]]], shape=(2, 2, 2), dtype=int32) op_rot90(m, k = 1, axes = c(2, 3)) ## tf.Tensor( ## [[[2 4] ##   [1 3]] ## ##  [[6 8] ##   [5 7]]], shape=(2, 2, 2), dtype=int32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_round.html","id":null,"dir":"Reference","previous_headings":"","what":"Evenly round to the given number of decimals. — op_round","title":"Evenly round to the given number of decimals. — op_round","text":"Evenly round given number decimals.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_round.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Evenly round to the given number of decimals. — op_round","text":"","code":"op_round(x, decimals = 0L)"},{"path":"https://keras3.posit.co/dev/reference/op_round.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Evenly round to the given number of decimals. — op_round","text":"x Input tensor. decimals Number decimal places round . Defaults 0.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_round.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Evenly round to the given number of decimals. — op_round","text":"Output tensor.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_rsqrt.html","id":null,"dir":"Reference","previous_headings":"","what":"Computes reciprocal of square root of x element-wise. — op_rsqrt","title":"Computes reciprocal of square root of x element-wise. — op_rsqrt","text":"Computes reciprocal square root x element-wise.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_rsqrt.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes reciprocal of square root of x element-wise. — op_rsqrt","text":"","code":"op_rsqrt(x)"},{"path":"https://keras3.posit.co/dev/reference/op_rsqrt.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Computes reciprocal of square root of x element-wise. — op_rsqrt","text":"x input tensor","code":""},{"path":"https://keras3.posit.co/dev/reference/op_rsqrt.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Computes reciprocal of square root of x element-wise. — op_rsqrt","text":"tensor dtype x.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_rsqrt.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Computes reciprocal of square root of x element-wise. — op_rsqrt","text":"","code":"x <- op_convert_to_tensor(c(1, 10, 100)) op_rsqrt(x) ## tf.Tensor([1.         0.31622777 0.1       ], shape=(3), dtype=float64) # array([1, 0.31622776, 0.1], dtype=float32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_saturate_cast.html","id":null,"dir":"Reference","previous_headings":"","what":"Performs a safe saturating cast to the desired dtype. — op_saturate_cast","title":"Performs a safe saturating cast to the desired dtype. — op_saturate_cast","text":"Saturating cast prevents data type overflow casting dtype smaller values range. E.g. op_cast(c(-1, 256), \"float32\") |> op_cast(\"uint8\") returns c(255, 0), op_cast(c(-1, 256), \"float32\") |> op_saturate_cast(\"uint8\") returns c(0, 255).","code":""},{"path":"https://keras3.posit.co/dev/reference/op_saturate_cast.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Performs a safe saturating cast to the desired dtype. — op_saturate_cast","text":"","code":"op_saturate_cast(x, dtype)"},{"path":"https://keras3.posit.co/dev/reference/op_saturate_cast.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Performs a safe saturating cast to the desired dtype. — op_saturate_cast","text":"x tensor variable. dtype target type.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_saturate_cast.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Performs a safe saturating cast to the desired dtype. — op_saturate_cast","text":"safely casted tensor specified dtype.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_saturate_cast.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Performs a safe saturating cast to the desired dtype. — op_saturate_cast","text":"Image resizing bicubic interpolation may produce values outside original range.     Casting resized image back uint8 cause overflow.     Saturate casting uint8 clip values uint8 range casting cause overflow.","code":"image2x2 <- np_array(as.integer(c(0, 1, 254, 255)), \"uint8\") |>   array_reshape(c(1, 2, 2, 1)) image4x4 <- image2x2 |>   tensorflow::tf$image$resize(shape(4, 4), method=\"bicubic\") image4x4 |> as.array() |> drop() ##           [,1]      [,2]      [,3]      [,4] ## [1,] -22.50000 -22.20462 -21.61891 -21.32353 ## [2,]  52.52605  52.82143  53.40715  53.70253 ## [3,] 201.29752 201.59288 202.17859 202.47395 ## [4,] 276.32355 276.61893 277.20465 277.50006 image4x4_casted <- op_cast(image4x4, \"uint8\") image4x4_casted |> as.array() |> drop() ##      [,1] [,2] [,3] [,4] ## [1,]  234  234  235  235 ## [2,]   52   52   53   53 ## [3,]  201  201  202  202 ## [4,]   20   20   21   21 image4x4_saturate_casted <- image4x4 |> op_saturate_cast(\"uint8\") image4x4_saturate_casted |> as.array() |> drop() ##      [,1] [,2] [,3] [,4] ## [1,]    0    0    0    0 ## [2,]   52   52   53   53 ## [3,]  201  201  202  202 ## [4,]  255  255  255  255"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_scan.html","id":null,"dir":"Reference","previous_headings":"","what":"Scan a function over leading array axes while carrying along state. — op_scan","title":"Scan a function over leading array axes while carrying along state. — op_scan","text":"type xs array type NULL, type ys array type, semantics op_scan() given roughly implementation:   loop-carried value carry (init) must hold fixed shape dtype across iterations. TensorFlow, y must match carry shape dtype. required backends.","code":"op_scan <- function(f, init, xs = NULL, length = NULL) {   xs <- xs %||% vector(\"list\", length)   if(!is.list(xs))     xs <- op_unstack(xs)   ys <- vector(\"list\", length(xs))   carry <- init   for (i in seq_along(xs)) {     c(carry, y) %<-% f(carry, xs[[i]])     ys[[i]] <- y   }   list(carry, op_stack(ys)) }"},{"path":"https://keras3.posit.co/dev/reference/op_scan.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Scan a function over leading array axes while carrying along state. — op_scan","text":"","code":"op_scan(f, init, xs = NULL, length = NULL, reverse = FALSE, unroll = 1L)"},{"path":"https://keras3.posit.co/dev/reference/op_scan.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Scan a function over leading array axes while carrying along state. — op_scan","text":"f Callable defines logic loop iteration. accepts two arguments first value loop carry second slice xs along leading axis. callable returns pair first represents new value loop carry second represents slice output. init initial loop carry value. can scalar, tensor, nested structure. must match structure first element returned f. xs Optional value scan along leading axis. can tensor nested structure. xs provided, must specify length define number loop iterations. Defaults NULL. length Optional integer specifying number loop iterations. length provided, defaults sizes leading axis arrays xs. Defaults NULL. reverse Optional boolean specifying whether run scan iteration forward reverse, equivalent reversing leading axes arrays xs ys. unroll Optional positive integer boolean specifying many scan iterations unroll within single iteration loop. integer provided, determines many unrolled loop iterations run within single rolled iteration loop. boolean provided, determine loop completely unrolled (unroll=TRUE) left completely unrolled (unroll=FALSE). Note unrolling supported JAX TensorFlow backends.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_scan.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Scan a function over leading array axes while carrying along state. — op_scan","text":"pair first element represents final loop carry value second element represents stacked outputs f scanned leading axis inputs.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_scan.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Scan a function over leading array axes while carrying along state. — op_scan","text":"","code":"sum_fn <- function(c, x) list(c + x, c + x) init <- op_array(0L) xs <- op_array(1:5) c(carry, result) %<-% op_scan(sum_fn, init, xs) carry ## tf.Tensor(15, shape=(), dtype=int32) result ## tf.Tensor([ 1  3  6 10 15], shape=(5), dtype=int32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_scatter.html","id":null,"dir":"Reference","previous_headings":"","what":"Returns a tensor of shape shape where indices are set to values. — op_scatter","title":"Returns a tensor of shape shape where indices are set to values. — op_scatter","text":"high level, operation zeros[indices] = updates returns output. equivalent :","code":"output <- op_scatter_update(op_zeros(shape), indices, values)"},{"path":"https://keras3.posit.co/dev/reference/op_scatter.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Returns a tensor of shape shape where indices are set to values. — op_scatter","text":"","code":"op_scatter(indices, values, shape)"},{"path":"https://keras3.posit.co/dev/reference/op_scatter.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Returns a tensor of shape shape where indices are set to values. — op_scatter","text":"indices tensor list specifying indices values values. values tensor, values set indices. shape Shape output tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_scatter.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Returns a tensor of shape shape where indices are set to values. — op_scatter","text":"tensor shape shape indices set values.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_scatter.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Returns a tensor of shape shape where indices are set to values. — op_scatter","text":"","code":"op_scatter(indices = rbind(1), values = 1, shape = c(5)) ## tf.Tensor([1. 0. 0. 0. 0.], shape=(5), dtype=float64) op_scatter(indices = rbind(1, 3), values = 1:2, shape = c(5)) ## tf.Tensor([1 0 2 0 0], shape=(5), dtype=int32) indices <- rbind(c(1, 2), c(2, 2)) values <- op_array(c(1, 1)) op_scatter(indices, values, shape = c(2, 2)) ## tf.Tensor( ## [[0. 1.] ##  [0. 1.]], shape=(2, 2), dtype=float32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_scatter_update.html","id":null,"dir":"Reference","previous_headings":"","what":"Update inputs via updates at scattered (sparse) indices. — op_scatter_update","title":"Update inputs via updates at scattered (sparse) indices. — op_scatter_update","text":"high level, operation inputs[indices] <- updates. Assume inputs tensor shape (D1, D2, ..., Dn), 2 main usages scatter_update. indices 2D tensor shape (num_updates, n), num_updates number updates perform, updates 1D tensor shape (num_updates). example, inputs op_zeros(c(4, 4, 4)), want update inputs[2, 3, 4] inputs[1, 2, 4] 1, can use:     2 indices 2D tensor shape (num_updates, k), num_updates number updates perform, k (k <= n) size index indices. updates n - k-D tensor shape (num_updates, shape(inputs)[-(1:k)]). example, inputs <- op_zeros(c(4, 4, 4)), want update inputs[1, 2, ] inputs[2, 3, ] [1, 1, 1, 1], indices shape (num_updates, 2) (k = 2), updates shape (num_updates, 4) (shape(inputs)[3:4] == 4). See code :","code":"inputs <- op_zeros(c(4, 4, 4)) indices <- rbind(c(2, 3, 4),                  c(1, 2, 4)) updates <- op_array(c(1, 1), \"float32\") op_scatter_update(inputs, indices, updates) ## tf.Tensor( ## [[[0. 0. 0. 0.] ##   [0. 0. 0. 1.] ##   [0. 0. 0. 0.] ##   [0. 0. 0. 0.]] ## ##  [[0. 0. 0. 0.] ##   [0. 0. 0. 0.] ##   [0. 0. 0. 1.] ##   [0. 0. 0. 0.]] ## ##  [[0. 0. 0. 0.] ##   [0. 0. 0. 0.] ##   [0. 0. 0. 0.] ##   [0. 0. 0. 0.]] ## ##  [[0. 0. 0. 0.] ##   [0. 0. 0. 0.] ##   [0. 0. 0. 0.] ##   [0. 0. 0. 0.]]], shape=(4, 4, 4), dtype=float32) inputs <- op_zeros(c(4, 4, 4)) indices <- rbind(c(2, 3),                  c(3, 4)) updates <- op_array(rbind(c(1, 1, 1, 1),                           c(1, 1, 1, 1)),                     \"float32\") op_scatter_update(inputs, indices, updates) ## tf.Tensor( ## [[[0. 0. 0. 0.] ##   [0. 0. 0. 0.] ##   [0. 0. 0. 0.] ##   [0. 0. 0. 0.]] ## ##  [[0. 0. 0. 0.] ##   [0. 0. 0. 0.] ##   [1. 1. 1. 1.] ##   [0. 0. 0. 0.]] ## ##  [[0. 0. 0. 0.] ##   [0. 0. 0. 0.] ##   [0. 0. 0. 0.] ##   [1. 1. 1. 1.]] ## ##  [[0. 0. 0. 0.] ##   [0. 0. 0. 0.] ##   [0. 0. 0. 0.] ##   [0. 0. 0. 0.]]], shape=(4, 4, 4), dtype=float32)"},{"path":"https://keras3.posit.co/dev/reference/op_scatter_update.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Update inputs via updates at scattered (sparse) indices. — op_scatter_update","text":"","code":"op_scatter_update(inputs, indices, updates)"},{"path":"https://keras3.posit.co/dev/reference/op_scatter_update.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Update inputs via updates at scattered (sparse) indices. — op_scatter_update","text":"inputs tensor, tensor updated. indices tensor list shape (N, inputs$ndim), specifying indices update. N number indices update, must equal first dimension updates. updates tensor, new values put inputs indices.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_scatter_update.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Update inputs via updates at scattered (sparse) indices. — op_scatter_update","text":"tensor, shape dtype inputs.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_searchsorted.html","id":null,"dir":"Reference","previous_headings":"","what":"Perform a binary search — op_searchsorted","title":"Perform a binary search — op_searchsorted","text":"Perform binary search, returning indices insertion values sorted_sequence maintain sorting order.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_searchsorted.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Perform a binary search — op_searchsorted","text":"","code":"op_searchsorted(sorted_sequence, values, side = \"left\", zero_indexed = FALSE)"},{"path":"https://keras3.posit.co/dev/reference/op_searchsorted.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Perform a binary search — op_searchsorted","text":"sorted_sequence 1-D input tensor, sorted along innermost dimension. values N-D tensor query insertion values. side 'left' 'right', specifying direction insert equality case (tie-breaker). zero_indexed TRUE, returned indices zero-based (0 encodes first position); FALSE (default), returned indices one-based (1 encodes first position).","code":""},{"path":"https://keras3.posit.co/dev/reference/op_searchsorted.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Perform a binary search — op_searchsorted","text":"Tensor insertion indices shape values.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_segment_max.html","id":null,"dir":"Reference","previous_headings":"","what":"Computes the max of segments in a tensor. — op_segment_max","title":"Computes the max of segments in a tensor. — op_segment_max","text":"Computes max segments tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_segment_max.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes the max of segments in a tensor. — op_segment_max","text":"","code":"op_segment_max(data, segment_ids, num_segments = NULL, sorted = FALSE)"},{"path":"https://keras3.posit.co/dev/reference/op_segment_max.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Computes the max of segments in a tensor. — op_segment_max","text":"data Input tensor. segment_ids N-D tensor containing segment indices element data. head(op_shape(data), length(op_shape(segment_ids))) match op_shape(segment_ids) num_segments integer representing total number segments. specified, inferred maximum value segment_ids. sorted boolean indicating whether segment_ids sorted. Defaults FALSE.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_segment_max.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Computes the max of segments in a tensor. — op_segment_max","text":"tensor containing max segments, element represents max corresponding segment data.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_segment_max.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Computes the max of segments in a tensor. — op_segment_max","text":"","code":"data <- op_convert_to_tensor(c(1, 2, 10, 20, 100, 200)) segment_ids <- op_array(c(1, 1, 2, 2, 3, 3), \"int32\") num_segments <- 3 op_segment_max(data, segment_ids, num_segments) ## tf.Tensor([  2.  20. 200.], shape=(3), dtype=float64) # array([2, 20, 200], dtype=int32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_segment_sum.html","id":null,"dir":"Reference","previous_headings":"","what":"Computes the sum of segments in a tensor. — op_segment_sum","title":"Computes the sum of segments in a tensor. — op_segment_sum","text":"Computes sum segments tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_segment_sum.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes the sum of segments in a tensor. — op_segment_sum","text":"","code":"op_segment_sum(data, segment_ids, num_segments = NULL, sorted = FALSE)"},{"path":"https://keras3.posit.co/dev/reference/op_segment_sum.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Computes the sum of segments in a tensor. — op_segment_sum","text":"data Input tensor. segment_ids N-D tensor containing segment indices element data. Num dims segment ids strictly smaller equal number dims data. num_segments integer representing total number segments. specified, inferred maximum value segment_ids. sorted boolean indicating whether segment_ids sorted. Defaults FALSE.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_segment_sum.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Computes the sum of segments in a tensor. — op_segment_sum","text":"tensor containing sum segments, element represents sum corresponding segment data.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_segment_sum.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Computes the sum of segments in a tensor. — op_segment_sum","text":"","code":"data <- op_array(c(1, 2, 10, 20, 100, 200)) segment_ids <- op_array(c(1, 1, 2, 2, 3, 3), \"int32\") num_segments <- 3 op_segment_sum(data, segment_ids, num_segments) ## tf.Tensor([  3.  30. 300.], shape=(3), dtype=float32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_select.html","id":null,"dir":"Reference","previous_headings":"","what":"Return elements from choicelist, based on conditions in condlist. — op_select","title":"Return elements from choicelist, based on conditions in condlist. — op_select","text":"Return elements choicelist, based conditions condlist.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_select.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Return elements from choicelist, based on conditions in condlist. — op_select","text":"","code":"op_select(condlist, choicelist, default = 0L)"},{"path":"https://keras3.posit.co/dev/reference/op_select.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Return elements from choicelist, based on conditions in condlist. — op_select","text":"condlist List boolean tensors. list conditions determine array choicelist output elements taken. multiple conditions satisfied, first one encountered condlist used. choicelist List tensors. list tensors output elements taken. list length condlist. default Optional scalar value. element inserted output conditions evaluate FALSE.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_select.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Return elements from choicelist, based on conditions in condlist. — op_select","text":"Tensor output position m m-th element tensor choicelist m-th element corresponding tensor condlist TRUE.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_select.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Return elements from choicelist, based on conditions in condlist. — op_select","text":"","code":"x <- op_arange(6L) condlist <- list(x < 3, x > 3) choicelist <- list(x, x^2) op_select(condlist, choicelist, 42) ## tf.Tensor([ 1  2 42 16 25 36], shape=(6), dtype=int32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_selu.html","id":null,"dir":"Reference","previous_headings":"","what":"Scaled Exponential Linear Unit (SELU) activation function. — op_selu","title":"Scaled Exponential Linear Unit (SELU) activation function. — op_selu","text":"defined : f(x) =  scale * alpha * (exp(x) - 1.) x < 0, f(x) = scale * x x >= 0.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_selu.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Scaled Exponential Linear Unit (SELU) activation function. — op_selu","text":"","code":"op_selu(x)"},{"path":"https://keras3.posit.co/dev/reference/op_selu.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Scaled Exponential Linear Unit (SELU) activation function. — op_selu","text":"x Input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_selu.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Scaled Exponential Linear Unit (SELU) activation function. — op_selu","text":"tensor shape x.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_selu.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Scaled Exponential Linear Unit (SELU) activation function. — op_selu","text":"","code":"x <- op_array(c(-1, 0, 1)) op_selu(x) ## tf.Tensor([-1.1113307  0.         1.050701 ], shape=(3), dtype=float32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_separable_conv.html","id":null,"dir":"Reference","previous_headings":"","what":"General N-D separable convolution. — op_separable_conv","title":"General N-D separable convolution. — op_separable_conv","text":"ops supports 1D 2D separable convolution. separable_conv depthwise conv followed pointwise conv.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_separable_conv.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"General N-D separable convolution. — op_separable_conv","text":"","code":"op_separable_conv(   inputs,   depthwise_kernel,   pointwise_kernel,   strides = 1L,   padding = \"valid\",   data_format = NULL,   dilation_rate = 1L )"},{"path":"https://keras3.posit.co/dev/reference/op_separable_conv.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"General N-D separable convolution. — op_separable_conv","text":"inputs Tensor rank N+2. inputs shape (batch_size,) + inputs_spatial_shape + (num_channels,) data_format=\"channels_last\", (batch_size, num_channels) + inputs_spatial_shape data_format=\"channels_first\". depthwise_kernel Tensor rank N+2. depthwise_kernel shape [kernel_spatial_shape, num_input_channels, num_channels_multiplier], num_input_channels match number channels inputs. pointwise_kernel Tensor rank N+2. pointwise_kernel shape (*ones_like(kernel_spatial_shape), num_input_channels * num_channels_multiplier, num_output_channels). strides int int tuple/list len(inputs_spatial_shape), specifying strides convolution along spatial dimension. strides int, every spatial dimension shares strides. padding string, either \"valid\" \"\". \"valid\" means padding applied, \"\" results padding evenly left/right /input output height/width dimension input strides=1. data_format string, either \"channels_last\" \"channels_first\". data_format determines ordering dimensions inputs. data_format=\"channels_last\", inputs shape (batch_size, ..., channels) data_format=\"channels_first\", inputs shape (batch_size, channels, ...). dilation_rate int int tuple/list len(inputs_spatial_shape), specifying dilation rate use dilated convolution. dilation_rate int, every spatial dimension shares dilation_rate.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_separable_conv.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"General N-D separable convolution. — op_separable_conv","text":"tensor rank N+2, result depthwise conv operation.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_shape.html","id":null,"dir":"Reference","previous_headings":"","what":"Gets the shape of the tensor input. — op_shape","title":"Gets the shape of the tensor input. — op_shape","text":"Gets shape tensor input.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_shape.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Gets the shape of the tensor input. — op_shape","text":"","code":"op_shape(x)"},{"path":"https://keras3.posit.co/dev/reference/op_shape.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Gets the shape of the tensor input. — op_shape","text":"x tensor. function try access shape attribute input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_shape.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Gets the shape of the tensor input. — op_shape","text":"list integers NULL values, indicating shape input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_shape.html","id":"note","dir":"Reference","previous_headings":"","what":"Note","title":"Gets the shape of the tensor input. — op_shape","text":"TensorFlow backend, x tf.Tensor dynamic shape, dimensions dynamic context compiled function tf.Tensor value instead static integer value.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_shape.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Gets the shape of the tensor input. — op_shape","text":"","code":"x <- op_zeros(c(8, 12)) op_shape(x) ## shape(8, 12)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_sigmoid.html","id":null,"dir":"Reference","previous_headings":"","what":"Sigmoid activation function. — op_sigmoid","title":"Sigmoid activation function. — op_sigmoid","text":"defined f(x) = 1 / (1 + exp(-x)).","code":""},{"path":"https://keras3.posit.co/dev/reference/op_sigmoid.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Sigmoid activation function. — op_sigmoid","text":"","code":"op_sigmoid(x)"},{"path":"https://keras3.posit.co/dev/reference/op_sigmoid.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Sigmoid activation function. — op_sigmoid","text":"x Input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_sigmoid.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Sigmoid activation function. — op_sigmoid","text":"tensor shape x.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_sigmoid.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Sigmoid activation function. — op_sigmoid","text":"","code":"x <- op_convert_to_tensor(c(-6, 1, 0, 1, 6)) op_sigmoid(x) ## tf.Tensor([0.00247262 0.73105858 0.5        0.73105858 0.99752738], shape=(5), dtype=float64)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_sign.html","id":null,"dir":"Reference","previous_headings":"","what":"Returns a tensor with the signs of the elements of x. — op_sign","title":"Returns a tensor with the signs of the elements of x. — op_sign","text":"Returns tensor signs elements x.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_sign.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Returns a tensor with the signs of the elements of x. — op_sign","text":"","code":"op_sign(x)"},{"path":"https://keras3.posit.co/dev/reference/op_sign.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Returns a tensor with the signs of the elements of x. — op_sign","text":"x Input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_sign.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Returns a tensor with the signs of the elements of x. — op_sign","text":"Output tensor shape x.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_signbit.html","id":null,"dir":"Reference","previous_headings":"","what":"Return the sign bit of the elements of x. — op_signbit","title":"Return the sign bit of the elements of x. — op_signbit","text":"output boolean tensor contains TRUE sign x negative, FALSE otherwise.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_signbit.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Return the sign bit of the elements of x. — op_signbit","text":"","code":"op_signbit(x)"},{"path":"https://keras3.posit.co/dev/reference/op_signbit.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Return the sign bit of the elements of x. — op_signbit","text":"x Input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_signbit.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Return the sign bit of the elements of x. — op_signbit","text":"Output boolean tensor shape x.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_silu.html","id":null,"dir":"Reference","previous_headings":"","what":"Sigmoid Linear Unit (SiLU) activation function, also known as Swish. — op_silu","title":"Sigmoid Linear Unit (SiLU) activation function, also known as Swish. — op_silu","text":"SiLU activation function computed sigmoid function multiplied input. defined f(x) = x * sigmoid(x).","code":""},{"path":"https://keras3.posit.co/dev/reference/op_silu.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Sigmoid Linear Unit (SiLU) activation function, also known as Swish. — op_silu","text":"","code":"op_silu(x)"},{"path":"https://keras3.posit.co/dev/reference/op_silu.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Sigmoid Linear Unit (SiLU) activation function, also known as Swish. — op_silu","text":"x Input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_silu.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Sigmoid Linear Unit (SiLU) activation function, also known as Swish. — op_silu","text":"tensor shape x.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_silu.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Sigmoid Linear Unit (SiLU) activation function, also known as Swish. — op_silu","text":"","code":"x <- op_convert_to_tensor(c(-6, 1, 0, 1, 6)) op_sigmoid(x) ## tf.Tensor([0.00247262 0.73105858 0.5        0.73105858 0.99752738], shape=(5), dtype=float64) op_silu(x) ## tf.Tensor([-0.01483574  0.73105858  0.          0.73105858  5.98516426], shape=(5), dtype=float64)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_sin.html","id":null,"dir":"Reference","previous_headings":"","what":"Trigonometric sine, element-wise. — op_sin","title":"Trigonometric sine, element-wise. — op_sin","text":"Trigonometric sine, element-wise.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_sin.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Trigonometric sine, element-wise. — op_sin","text":"","code":"op_sin(x)"},{"path":"https://keras3.posit.co/dev/reference/op_sin.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Trigonometric sine, element-wise. — op_sin","text":"x Input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_sin.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Trigonometric sine, element-wise. — op_sin","text":"Output tensor shape x.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_sinh.html","id":null,"dir":"Reference","previous_headings":"","what":"Hyperbolic sine, element-wise. — op_sinh","title":"Hyperbolic sine, element-wise. — op_sinh","text":"Hyperbolic sine, element-wise.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_sinh.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Hyperbolic sine, element-wise. — op_sinh","text":"","code":"op_sinh(x)"},{"path":"https://keras3.posit.co/dev/reference/op_sinh.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Hyperbolic sine, element-wise. — op_sinh","text":"x Input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_sinh.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Hyperbolic sine, element-wise. — op_sinh","text":"Output tensor shape x.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_size.html","id":null,"dir":"Reference","previous_headings":"","what":"Return the number of elements in a tensor. — op_size","title":"Return the number of elements in a tensor. — op_size","text":"Return number elements tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_size.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Return the number of elements in a tensor. — op_size","text":"","code":"op_size(x)"},{"path":"https://keras3.posit.co/dev/reference/op_size.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Return the number of elements in a tensor. — op_size","text":"x Input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_size.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Return the number of elements in a tensor. — op_size","text":"Number elements x.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_slice.html","id":null,"dir":"Reference","previous_headings":"","what":"Return a slice of an input tensor. — op_slice","title":"Return a slice of an input tensor. — op_slice","text":"high level, operation explicit replacement array slicing e.g. inputs[start_indices:(start_indices + shape)]. Unlike slicing via brackets, operation accept tensor start indices backends, useful indices dynamically computed via tensor operations.","code":"(inputs <- op_arange(5*5) |> op_reshape(c(5, 5))) ## tf.Tensor( ## [[ 1.  2.  3.  4.  5.] ##  [ 6.  7.  8.  9. 10.] ##  [11. 12. 13. 14. 15.] ##  [16. 17. 18. 19. 20.] ##  [21. 22. 23. 24. 25.]], shape=(5, 5), dtype=float32) start_indices <- c(3, 3) shape <- c(2, 2) op_slice(inputs, start_indices, shape) ## tf.Tensor( ## [[13. 14.] ##  [18. 19.]], shape=(2, 2), dtype=float32)"},{"path":"https://keras3.posit.co/dev/reference/op_slice.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Return a slice of an input tensor. — op_slice","text":"","code":"op_slice(inputs, start_indices, shape)"},{"path":"https://keras3.posit.co/dev/reference/op_slice.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Return a slice of an input tensor. — op_slice","text":"inputs tensor, tensor sliced. start_indices list length inputs$ndim, specifying starting indices updating. shape full shape returned slice.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_slice.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Return a slice of an input tensor. — op_slice","text":"tensor, shape dtype inputs.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_slice_update.html","id":null,"dir":"Reference","previous_headings":"","what":"Update an input by slicing in a tensor of updated values. — op_slice_update","title":"Update an input by slicing in a tensor of updated values. — op_slice_update","text":"high level, operation inputs[start_indices: start_indices + updates.shape] = updates. Assume inputs tensor shape (D1, D2, ..., Dn), start_indices must list n integers, specifying starting indices. updates must rank inputs, size dim must exceed Di - start_indices[]. example, 2D inputs inputs = op_zeros(c(5, 5)), want update intersection last 2 rows last 2 columns 1, .e., inputs[4:5, 4:5] = op_ones(c(2, 2)), can use code :","code":"inputs <- op_zeros(c(5, 5)) start_indices <- c(3, 3) updates <- op_ones(c(2, 2)) op_slice_update(inputs, start_indices, updates) ## tf.Tensor( ## [[0. 0. 0. 0. 0.] ##  [0. 0. 0. 0. 0.] ##  [0. 0. 1. 1. 0.] ##  [0. 0. 1. 1. 0.] ##  [0. 0. 0. 0. 0.]], shape=(5, 5), dtype=float32)"},{"path":"https://keras3.posit.co/dev/reference/op_slice_update.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Update an input by slicing in a tensor of updated values. — op_slice_update","text":"","code":"op_slice_update(inputs, start_indices, updates)"},{"path":"https://keras3.posit.co/dev/reference/op_slice_update.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Update an input by slicing in a tensor of updated values. — op_slice_update","text":"inputs tensor, tensor updated. start_indices list length inputs$ndim, specifying starting indices updating. updates tensor, new values put inputs indices. updates must rank inputs.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_slice_update.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Update an input by slicing in a tensor of updated values. — op_slice_update","text":"tensor, shape dtype inputs.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_slogdet.html","id":null,"dir":"Reference","previous_headings":"","what":"Compute the sign and natural logarithm of the determinant of a matrix. — op_slogdet","title":"Compute the sign and natural logarithm of the determinant of a matrix. — op_slogdet","text":"Compute sign natural logarithm determinant matrix.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_slogdet.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Compute the sign and natural logarithm of the determinant of a matrix. — op_slogdet","text":"","code":"op_slogdet(x)"},{"path":"https://keras3.posit.co/dev/reference/op_slogdet.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Compute the sign and natural logarithm of the determinant of a matrix. — op_slogdet","text":"x Input matrix. must 2D square.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_slogdet.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Compute the sign and natural logarithm of the determinant of a matrix. — op_slogdet","text":"list: (sign, logabsdet). sign number representing sign determinant. real matrix, 1, 0, -1. complex matrix, complex number absolute value 1 (.e., unit circle), else 0. logabsdet natural log absolute value determinant.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_soft_shrink.html","id":null,"dir":"Reference","previous_headings":"","what":"Soft Shrink activation function. — op_soft_shrink","title":"Soft Shrink activation function. — op_soft_shrink","text":"defined f(x) = x - threshold x > threshold, f(x) = x + threshold x < -threshold, f(x) = 0 otherwise.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_soft_shrink.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Soft Shrink activation function. — op_soft_shrink","text":"","code":"op_soft_shrink(x, threshold = 0.5)"},{"path":"https://keras3.posit.co/dev/reference/op_soft_shrink.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Soft Shrink activation function. — op_soft_shrink","text":"x Input tensor. threshold Threshold value. Defaults 0.5.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_soft_shrink.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Soft Shrink activation function. — op_soft_shrink","text":"tensor shape x.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_soft_shrink.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Soft Shrink activation function. — op_soft_shrink","text":"","code":"x <- op_array(c(-1, 0, 1)) op_soft_shrink(x) ## tf.Tensor([-0.5  0.   0.5], shape=(3), dtype=float32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_softmax.html","id":null,"dir":"Reference","previous_headings":"","what":"Softmax activation function. — op_softmax","title":"Softmax activation function. — op_softmax","text":"elements output vector lie within range (0, 1), total sum exactly 1 (excluding floating point rounding error). vector processed independently. axis argument specifies axis along function applied within input. defined : f(x) = exp(x) / sum(exp(x))","code":""},{"path":"https://keras3.posit.co/dev/reference/op_softmax.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Softmax activation function. — op_softmax","text":"","code":"op_softmax(x, axis = -1L)"},{"path":"https://keras3.posit.co/dev/reference/op_softmax.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Softmax activation function. — op_softmax","text":"x Input tensor. axis Integer, axis along softmax applied.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_softmax.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Softmax activation function. — op_softmax","text":"tensor shape x.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_softmax.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Softmax activation function. — op_softmax","text":"","code":"x <- op_array(c(-1, 0, 1)) op_softmax(x) ## tf.Tensor([0.09003057 0.24472848 0.66524094], shape=(3), dtype=float32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_softplus.html","id":null,"dir":"Reference","previous_headings":"","what":"Softplus activation function. — op_softplus","title":"Softplus activation function. — op_softplus","text":"defined f(x) = log(exp(x) + 1), log natural logarithm exp exponential function.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_softplus.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Softplus activation function. — op_softplus","text":"","code":"op_softplus(x)"},{"path":"https://keras3.posit.co/dev/reference/op_softplus.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Softplus activation function. — op_softplus","text":"x Input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_softplus.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Softplus activation function. — op_softplus","text":"tensor shape x.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_softplus.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Softplus activation function. — op_softplus","text":"","code":"x <- op_convert_to_tensor(c(-0.555, 0, 0.555)) op_softplus(x) ## tf.Tensor([0.45366606 0.69314718 1.00866606], shape=(3), dtype=float64) x <- seq(-10, 10, .1) plot(x, op_softplus(x))"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_softsign.html","id":null,"dir":"Reference","previous_headings":"","what":"Softsign activation function. — op_softsign","title":"Softsign activation function. — op_softsign","text":"defined f(x) = x / (abs(x) + 1).","code":""},{"path":"https://keras3.posit.co/dev/reference/op_softsign.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Softsign activation function. — op_softsign","text":"","code":"op_softsign(x)"},{"path":"https://keras3.posit.co/dev/reference/op_softsign.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Softsign activation function. — op_softsign","text":"x Input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_softsign.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Softsign activation function. — op_softsign","text":"tensor shape x.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_softsign.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Softsign activation function. — op_softsign","text":"","code":"x <- op_convert_to_tensor(c(-0.100, -10.0, 1.0, 0.0, 100.0)) op_softsign(x) ## tf.Tensor([-0.09090909 -0.90909091  0.5         0.          0.99009901], shape=(5), dtype=float64) x <- seq(-10, 10, .1) plot(x, op_softsign(x), ylim = c(-1, 1))"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_solve.html","id":null,"dir":"Reference","previous_headings":"","what":"Solves a linear system of equations given by a x = b. — op_solve","title":"Solves a linear system of equations given by a x = b. — op_solve","text":"Solves x equation %*% x == b.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_solve.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Solves a linear system of equations given by a x = b. — op_solve","text":"","code":"op_solve(a, b)"},{"path":"https://keras3.posit.co/dev/reference/op_solve.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Solves a linear system of equations given by a x = b. — op_solve","text":"tensor shape (..., M, M) representing coefficients matrix. b tensor shape (..., M) (..., M, N) representing right-hand side \"dependent variable\" matrix.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_solve.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Solves a linear system of equations given by a x = b. — op_solve","text":"tensor shape (..., M) (..., M, N) representing solution linear system. Returned shape identical b.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_solve.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Solves a linear system of equations given by a x = b. — op_solve","text":"","code":"a <- op_array(c(1, 2, 4, 5), dtype=\"float32\") |> op_reshape(c(2, 2)) b <- op_array(c(2, 4, 8, 10), dtype=\"float32\") |> op_reshape(c(2, 2)) op_solve(a, b) ## tf.Tensor( ## [[2. 0.] ##  [0. 2.]], shape=(2, 2), dtype=float32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_solve_triangular.html","id":null,"dir":"Reference","previous_headings":"","what":"Solves a linear system of equations given by a %*% x = b. — op_solve_triangular","title":"Solves a linear system of equations given by a %*% x = b. — op_solve_triangular","text":"Solves linear system equations given %*% x = b.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_solve_triangular.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Solves a linear system of equations given by a %*% x = b. — op_solve_triangular","text":"","code":"op_solve_triangular(a, b, lower = FALSE)"},{"path":"https://keras3.posit.co/dev/reference/op_solve_triangular.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Solves a linear system of equations given by a %*% x = b. — op_solve_triangular","text":"tensor shape (..., M, M) representing coefficients matrix. b tensor shape (..., M) (..., M, N) representing right-hand side \"dependent variable\" matrix. lower logical. Use data contained lower triangle . Default use upper triangle.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_solve_triangular.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Solves a linear system of equations given by a %*% x = b. — op_solve_triangular","text":"tensor shape (..., M) (..., M, N) representing solution linear system. Returned shape identical b.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_sort.html","id":null,"dir":"Reference","previous_headings":"","what":"Sorts the elements of x along a given axis in ascending order. — op_sort","title":"Sorts the elements of x along a given axis in ascending order. — op_sort","text":"Sorts elements x along given axis ascending order.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_sort.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Sorts the elements of x along a given axis in ascending order. — op_sort","text":"","code":"op_sort(x, axis = -1L)"},{"path":"https://keras3.posit.co/dev/reference/op_sort.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Sorts the elements of x along a given axis in ascending order. — op_sort","text":"x Input tensor. axis Axis along sort. NULL, tensor flattened sorting. Defaults -1; last axis.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_sort.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Sorts the elements of x along a given axis in ascending order. — op_sort","text":"Sorted tensor.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_sparse_categorical_crossentropy.html","id":null,"dir":"Reference","previous_headings":"","what":"Computes sparse categorical cross-entropy loss. — op_sparse_categorical_crossentropy","title":"Computes sparse categorical cross-entropy loss. — op_sparse_categorical_crossentropy","text":"sparse categorical cross-entropy loss similar categorical cross-entropy, used target tensor contains integer class labels instead one-hot encoded vectors. measures dissimilarity target output probabilities logits.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_sparse_categorical_crossentropy.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes sparse categorical cross-entropy loss. — op_sparse_categorical_crossentropy","text":"","code":"op_sparse_categorical_crossentropy(   target,   output,   from_logits = FALSE,   axis = -1L )"},{"path":"https://keras3.posit.co/dev/reference/op_sparse_categorical_crossentropy.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Computes sparse categorical cross-entropy loss. — op_sparse_categorical_crossentropy","text":"target target tensor representing true class labels integers. shape match shape output tensor except last dimension. output output tensor representing predicted probabilities logits. shape match shape target tensor except last dimension. from_logits (optional) Whether output tensor logits probabilities. Set TRUE output represents logits; otherwise, set FALSE output represents probabilities. Defaults FALSE. axis (optional) axis along sparse categorical cross-entropy computed. Defaults -1, corresponds last dimension tensors.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_sparse_categorical_crossentropy.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Computes sparse categorical cross-entropy loss. — op_sparse_categorical_crossentropy","text":"Integer tensor: computed sparse categorical cross-entropy loss target output.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_sparse_categorical_crossentropy.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Computes sparse categorical cross-entropy loss. — op_sparse_categorical_crossentropy","text":"","code":"target <- op_array(c(0, 1, 2), dtype=\"int32\") output <- op_array(rbind(c(0.9, 0.05, 0.05),                         c(0.1, 0.8,  0.1),                         c(0.2, 0.3,  0.5))) op_sparse_categorical_crossentropy(target, output) ## tf.Tensor([0.10536052 0.22314355 0.69314718], shape=(3), dtype=float64)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_sparse_plus.html","id":null,"dir":"Reference","previous_headings":"","what":"SparsePlus activation function. — op_sparse_plus","title":"SparsePlus activation function. — op_sparse_plus","text":"defined f(x) = 0 x <= -1. f(x) = (1/4) * (x + 1)^2 -1 < x < 1. f(x) = x x >= 1.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_sparse_plus.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"SparsePlus activation function. — op_sparse_plus","text":"","code":"op_sparse_plus(x)"},{"path":"https://keras3.posit.co/dev/reference/op_sparse_plus.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"SparsePlus activation function. — op_sparse_plus","text":"x Input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_sparse_plus.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"SparsePlus activation function. — op_sparse_plus","text":"tensor shape x.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_sparse_plus.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"SparsePlus activation function. — op_sparse_plus","text":"","code":"x <- op_array(c(-1.0, 0.0, 1.0)) op_sparse_plus(x) ## tf.Tensor([0.   0.25 1.  ], shape=(3), dtype=float32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_sparsemax.html","id":null,"dir":"Reference","previous_headings":"","what":"Sparsemax activation function. — op_sparsemax","title":"Sparsemax activation function. — op_sparsemax","text":"batch , class j, sparsemax activation function defined : sparsemax(x)[, j] = max(x[, j] - (x[, :]), 0).","code":""},{"path":"https://keras3.posit.co/dev/reference/op_sparsemax.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Sparsemax activation function. — op_sparsemax","text":"","code":"op_sparsemax(x, axis = -1L)"},{"path":"https://keras3.posit.co/dev/reference/op_sparsemax.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Sparsemax activation function. — op_sparsemax","text":"x Input tensor. axis int, axis along sparsemax operation applied.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_sparsemax.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Sparsemax activation function. — op_sparsemax","text":"tensor, output sparsemax transformation. type shape x.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_sparsemax.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Sparsemax activation function. — op_sparsemax","text":"","code":"x <- op_array(c(-1., 0., 1.)) op_sparsemax(x) ## tf.Tensor([0. 0. 1.], shape=(3), dtype=float32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_split.html","id":null,"dir":"Reference","previous_headings":"","what":"Split a tensor into chunks. — op_split","title":"Split a tensor into chunks. — op_split","text":"Split tensor chunks.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_split.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Split a tensor into chunks. — op_split","text":"","code":"op_split(x, indices_or_sections, axis = 1L)"},{"path":"https://keras3.posit.co/dev/reference/op_split.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Split a tensor into chunks. — op_split","text":"x Input tensor. indices_or_sections integer, N, tensor split N equal sections along axis. 1-D array sorted integers, entries indicate indices tensor split along axis. axis Axis along split. Defaults 1, first axis.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_split.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Split a tensor into chunks. — op_split","text":"list tensors.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_split.html","id":"note","dir":"Reference","previous_headings":"","what":"Note","title":"Split a tensor into chunks. — op_split","text":"split result equal division using Torch backend.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_split.html","id":"example","dir":"Reference","previous_headings":"","what":"Example","title":"Split a tensor into chunks. — op_split","text":"","code":"x <- op_arange(12)  # pass a scalar integer for n sections op_split(x, 2) ## [[1]] ## tf.Tensor([1. 2. 3. 4. 5. 6.], shape=(6), dtype=float32) ## ## [[2]] ## tf.Tensor([ 7.  8.  9. 10. 11. 12.], shape=(6), dtype=float32) op_split(x, 3) ## [[1]] ## tf.Tensor([1. 2. 3. 4.], shape=(4), dtype=float32) ## ## [[2]] ## tf.Tensor([5. 6. 7. 8.], shape=(4), dtype=float32) ## ## [[3]] ## tf.Tensor([ 9. 10. 11. 12.], shape=(4), dtype=float32) # 1-d array/tensor for indices op_split(x, array(c(3, 8))) ## [[1]] ## tf.Tensor([1. 2.], shape=(2), dtype=float32) ## ## [[2]] ## tf.Tensor([3. 4. 5. 6. 7.], shape=(5), dtype=float32) ## ## [[3]] ## tf.Tensor([ 8.  9. 10. 11. 12.], shape=(5), dtype=float32) op_split(x, array(c(3))) ## [[1]] ## tf.Tensor([1. 2.], shape=(2), dtype=float32) ## ## [[2]] ## tf.Tensor([ 3.  4.  5.  6.  7.  8.  9. 10. 11. 12.], shape=(10), dtype=float32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_sqrt.html","id":null,"dir":"Reference","previous_headings":"","what":"Return the non-negative square root of a tensor, element-wise. — op_sqrt","title":"Return the non-negative square root of a tensor, element-wise. — op_sqrt","text":"Return non-negative square root tensor, element-wise.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_sqrt.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Return the non-negative square root of a tensor, element-wise. — op_sqrt","text":"","code":"op_sqrt(x)"},{"path":"https://keras3.posit.co/dev/reference/op_sqrt.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Return the non-negative square root of a tensor, element-wise. — op_sqrt","text":"x Input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_sqrt.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Return the non-negative square root of a tensor, element-wise. — op_sqrt","text":"Output tensor, non-negative square root x.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_square.html","id":null,"dir":"Reference","previous_headings":"","what":"Return the element-wise square of the input. — op_square","title":"Return the element-wise square of the input. — op_square","text":"Return element-wise square input.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_square.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Return the element-wise square of the input. — op_square","text":"","code":"op_square(x)"},{"path":"https://keras3.posit.co/dev/reference/op_square.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Return the element-wise square of the input. — op_square","text":"x Input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_square.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Return the element-wise square of the input. — op_square","text":"Output tensor, square x.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_squareplus.html","id":null,"dir":"Reference","previous_headings":"","what":"Squareplus activation function. — op_squareplus","title":"Squareplus activation function. — op_squareplus","text":"Squareplus activation function defined : f(x) = (x + sqrt(x^2 + b)) / 2","code":""},{"path":"https://keras3.posit.co/dev/reference/op_squareplus.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Squareplus activation function. — op_squareplus","text":"","code":"op_squareplus(x, b = 4L)"},{"path":"https://keras3.posit.co/dev/reference/op_squareplus.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Squareplus activation function. — op_squareplus","text":"x Input tensor. b Smoothness parameter. Defaults 4.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_squareplus.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Squareplus activation function. — op_squareplus","text":"tensor shape x.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_squareplus.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Squareplus activation function. — op_squareplus","text":"","code":"x <- op_array(c(-1.0, 0.0, 1.0)) op_squareplus(x) ## tf.Tensor([0.618034 1.       1.618034], shape=(3), dtype=float32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_squeeze.html","id":null,"dir":"Reference","previous_headings":"","what":"Remove axes of length one from x. — op_squeeze","title":"Remove axes of length one from x. — op_squeeze","text":"Remove axes length one x.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_squeeze.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Remove axes of length one from x. — op_squeeze","text":"","code":"op_squeeze(x, axis = NULL)"},{"path":"https://keras3.posit.co/dev/reference/op_squeeze.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Remove axes of length one from x. — op_squeeze","text":"x Input tensor. axis Select subset entries length one shape.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_squeeze.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Remove axes of length one from x. — op_squeeze","text":"input tensor subset dimensions length 1 removed.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_stack.html","id":null,"dir":"Reference","previous_headings":"","what":"Join a sequence of tensors along a new axis. — op_stack","title":"Join a sequence of tensors along a new axis. — op_stack","text":"axis parameter specifies index new axis dimensions result.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_stack.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Join a sequence of tensors along a new axis. — op_stack","text":"","code":"op_stack(x, axis = 1L)"},{"path":"https://keras3.posit.co/dev/reference/op_stack.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Join a sequence of tensors along a new axis. — op_stack","text":"x sequence tensors. axis Axis along stack. Defaults 1, first axis.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_stack.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Join a sequence of tensors along a new axis. — op_stack","text":"stacked tensor.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_std.html","id":null,"dir":"Reference","previous_headings":"","what":"Compute the standard deviation along the specified axis. — op_std","title":"Compute the standard deviation along the specified axis. — op_std","text":"Compute standard deviation along specified axis.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_std.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Compute the standard deviation along the specified axis. — op_std","text":"","code":"op_std(x, axis = NULL, keepdims = FALSE)"},{"path":"https://keras3.posit.co/dev/reference/op_std.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Compute the standard deviation along the specified axis. — op_std","text":"x Input tensor. axis Axis along compute standard deviation. Default compute standard deviation flattened tensor. keepdims set TRUE, axes reduced left result dimensions size one.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_std.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Compute the standard deviation along the specified axis. — op_std","text":"Output tensor containing standard deviation values.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_stft.html","id":null,"dir":"Reference","previous_headings":"","what":"Short-Time Fourier Transform along the last axis of the input. — op_stft","title":"Short-Time Fourier Transform along the last axis of the input. — op_stft","text":"STFT computes Fourier transform short overlapping windows input. giving frequency components signal change time.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_stft.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Short-Time Fourier Transform along the last axis of the input. — op_stft","text":"","code":"op_stft(   x,   sequence_length,   sequence_stride,   fft_length,   window = \"hann\",   center = TRUE )"},{"path":"https://keras3.posit.co/dev/reference/op_stft.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Short-Time Fourier Transform along the last axis of the input. — op_stft","text":"x Input tensor. sequence_length integer representing sequence length. sequence_stride integer representing sequence hop size. fft_length integer representing size FFT apply. specified, uses smallest power 2 enclosing sequence_length. window string, tensor window NULL. window string, available values \"hann\" \"hamming\". window tensor, used directly window length must sequence_length. window NULL, windowing used. Defaults \"hann\". center Whether pad x sides t-th sequence centered time t * sequence_stride. Otherwise, t-th sequence begins time t * sequence_stride. Defaults TRUE.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_stft.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Short-Time Fourier Transform along the last axis of the input. — op_stft","text":"list containing two tensors - real imaginary parts STFT output.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_stft.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Short-Time Fourier Transform along the last axis of the input. — op_stft","text":"","code":"x <- op_array(c(0, 1, 2, 3, 4)) op_stft(x, 3, 2, 3) ## [[1]] ## tf.Tensor( ## [[ 0.  0.] ##  [ 2. -1.] ##  [ 4. -2.]], shape=(3, 2), dtype=float32) ## ## [[2]] ## tf.Tensor( ## [[ 0.         0.       ] ##  [ 0.        -1.7320508] ##  [ 0.        -3.4641016]], shape=(3, 2), dtype=float32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_stop_gradient.html","id":null,"dir":"Reference","previous_headings":"","what":"Stops gradient computation. — op_stop_gradient","title":"Stops gradient computation. — op_stop_gradient","text":"Stops gradient computation.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_stop_gradient.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Stops gradient computation. — op_stop_gradient","text":"","code":"op_stop_gradient(variable)"},{"path":"https://keras3.posit.co/dev/reference/op_stop_gradient.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Stops gradient computation. — op_stop_gradient","text":"variable tensor variable gradient computation disabled.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_stop_gradient.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Stops gradient computation. — op_stop_gradient","text":"variable gradient computation disabled.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_stop_gradient.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Stops gradient computation. — op_stop_gradient","text":"","code":"var <- op_convert_to_tensor(c(1, 2, 3), dtype=\"float32\") var <- op_stop_gradient(var)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_subset.html","id":null,"dir":"Reference","previous_headings":"","what":"Subset elements from a tensor — op_subset","title":"Subset elements from a tensor — op_subset","text":"Extract elements tensor using common R-style [ indexing idioms. function can also conveniently accessed via syntax tensor@r[...].","code":""},{"path":"https://keras3.posit.co/dev/reference/op_subset.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Subset elements from a tensor — op_subset","text":"","code":"op_subset(x, ...)  op_subset(x, ...) <- value  op_subset_set(x, ..., value)"},{"path":"https://keras3.posit.co/dev/reference/op_subset.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Subset elements from a tensor — op_subset","text":"x Input tensor. ... Indices specifying elements extract. argument ... can : integer scalar 1-d integer logical vector NULL newaxis .. symbol slice expression using : single argument supplied ..., ..1 can also : logical array shape x integer matrix ncol(..1) == op_rank(x) value new value replace selected subset .","code":""},{"path":"https://keras3.posit.co/dev/reference/op_subset.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Subset elements from a tensor — op_subset","text":"tensor containing subset elements.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_subset.html","id":"details","dir":"Reference","previous_headings":"","what":"Details","title":"Subset elements from a tensor — op_subset","text":"semantics similar R's [, differences:","code":""},{"path":"https://keras3.posit.co/dev/reference/op_subset.html","id":"differences-from-r-s-","dir":"Reference","previous_headings":"","what":"Differences from R's [:","title":"Subset elements from a tensor — op_subset","text":"Negative indices follow Python-style indexing, counting end array. NULL newaxis adds new dimension (equivalent op_expand_dims()). fewer indices dimensions (op_rank(x)) provided, missing dimensions implicitly filled. example, x matrix, x[1] returns first row. .. all_dims() expands include unspecified dimensions (see examples). Extended slicing syntax (:) supported, including: Strided steps: x@r[start:end:step] NA values start end. NA start defaults 1, NA end defaults axis size. logical array matching shape x selects elements row-wise order.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_subset.html","id":"similarities-with-r-s-","dir":"Reference","previous_headings":"","what":"Similarities with R's [:","title":"Subset elements from a tensor — op_subset","text":"Similarities R's [ (differences Python's [): Positive indices 1-based. Slices (x[start:end]) inclusive end. 1-d logical/integer arrays subset along respective axis. Multiple vectors provided different axes return intersected subsets. single integer matrix ncol() == op_rank(x) selects elements coordinates. row matrix specifies location one value, column corresponds axis tensor subsetted.  means use 2-column matrix subset matrix, 3-column matrix subset 3d array, .","code":""},{"path":"https://keras3.posit.co/dev/reference/op_subset.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Subset elements from a tensor — op_subset","text":"","code":"(x <- op_arange(5L) + 10L) ## tf.Tensor([11 12 13 14 15], shape=(5), dtype=int32) # Basic example, get first element op_subset(x, 1) ## tf.Tensor(11, shape=(), dtype=int32) # Use `@r[` syntax x@r[1]           # same as `op_subset(x, 1)` ## tf.Tensor(11, shape=(), dtype=int32) x@r[1:2]         # get the first 2 elements ## tf.Tensor([11 12], shape=(2), dtype=int32) x@r[c(1, 3)]     # first and third element ## tf.Tensor([11 13], shape=(2), dtype=int32) # Negative indices x@r[-1]          # last element ## tf.Tensor(15, shape=(), dtype=int32) x@r[-2]          # second to last element ## tf.Tensor(14, shape=(), dtype=int32) x@r[c(-1, -2)]   # last and second to last elements ## tf.Tensor([15 14], shape=(2), dtype=int32) x@r[c(-2, -1)]   # second to last and last elements ## tf.Tensor([14 15], shape=(2), dtype=int32) x@r[c(1, -1)]    # first and last elements ## tf.Tensor([11 15], shape=(2), dtype=int32) # Slices x@r[1:3]          # first 3 elements ## tf.Tensor([11 12 13], shape=(3), dtype=int32) x@r[NA:3]         # first 3 elements ## tf.Tensor([11 12 13], shape=(3), dtype=int32) x@r[1:5]          # all elements ## tf.Tensor([11 12 13 14 15], shape=(5), dtype=int32) x@r[1:-1]         # all elements ## tf.Tensor([11 12 13 14 15], shape=(5), dtype=int32) x@r[NA:NA]        # all elements ## tf.Tensor([11 12 13 14 15], shape=(5), dtype=int32) x@r[]             # all elements ## tf.Tensor([11 12 13 14 15], shape=(5), dtype=int32) x@r[1:-2]         # drop last element ## tf.Tensor([11 12 13 14], shape=(4), dtype=int32) x@r[NA:-2]        # drop last element ## tf.Tensor([11 12 13 14], shape=(4), dtype=int32) x@r[2:NA]         # drop first element ## tf.Tensor([12 13 14 15], shape=(4), dtype=int32) # 2D array examples xr <- array(1:12, c(3, 4)) x <- op_convert_to_tensor(xr)  # Basic subsetting x@r[1, ]      # first row ## tf.Tensor([ 1  4  7 10], shape=(4), dtype=int32) x@r[1]        # also first row! Missing axes are implicitly inserted ## tf.Tensor([ 1  4  7 10], shape=(4), dtype=int32) x@r[-1]       # last row ## tf.Tensor([ 3  6  9 12], shape=(4), dtype=int32) x@r[, 2]      # second column ## tf.Tensor([4 5 6], shape=(3), dtype=int32) x@r[, 2:2]    # second column, but shape preserved (like [, drop=FALSE]) ## tf.Tensor( ## [[4] ##  [5] ##  [6]], shape=(3, 1), dtype=int32) # Subsetting with a boolean array # Note: extracted elements are selected row-wise, not column-wise mask <- x >= 6 x@r[mask]             # returns a 1D tensor ## tf.Tensor([ 7 10  8 11  6  9 12], shape=(7), dtype=int32) x.r <- as.array(x) mask.r <- as.array(mask) # as.array(x)[mask] selects column-wise. Use `aperm()` to reverse search order. all(aperm(x.r)[aperm(mask.r)] == as.array(x@r[mask])) ## [1] TRUE # Subsetting with a matrix of index positions indices <- rbind(c(1, 1), c(2, 2), c(3, 3)) x@r[indices] # get diagonal elements ## tf.Tensor([1 5 9], shape=(3), dtype=int32) x.r[indices] # same as subsetting an R array ## [1] 1 5 9 # 3D array examples # Image: 4x4 pixels, 3 colors (RGB) # Tensor shape: (img_height, img_width, img_color_channels) shp <- shape(4, 4, 3) x <- op_arange(prod(shp)) |> op_reshape(shp)  # Convert to a batch of images by inserting a new axis # New shape: (batch_size, img_height, img_width, img_color_channels) x@r[newaxis, , , ] |> op_shape() ## shape(1, 4, 4, 3) x@r[newaxis] |> op_shape()  # same as above ## shape(1, 4, 4, 3) x@r[NULL] |> op_shape()     # same as above ## shape(1, 4, 4, 3) x <- x@r[newaxis] # Extract color channels x@r[, , , 1]          # red channel ## tf.Tensor( ## [[[ 1.  4.  7. 10.] ##   [13. 16. 19. 22.] ##   [25. 28. 31. 34.] ##   [37. 40. 43. 46.]]], shape=(1, 4, 4), dtype=float32) x@r[.., 1]            # red channel, same as above using .. shorthand ## tf.Tensor( ## [[[ 1.  4.  7. 10.] ##   [13. 16. 19. 22.] ##   [25. 28. 31. 34.] ##   [37. 40. 43. 46.]]], shape=(1, 4, 4), dtype=float32) x@r[.., 2]            # green channel ## tf.Tensor( ## [[[ 2.  5.  8. 11.] ##   [14. 17. 20. 23.] ##   [26. 29. 32. 35.] ##   [38. 41. 44. 47.]]], shape=(1, 4, 4), dtype=float32) x@r[.., 3]            # blue channel ## tf.Tensor( ## [[[ 3.  6.  9. 12.] ##   [15. 18. 21. 24.] ##   [27. 30. 33. 36.] ##   [39. 42. 45. 48.]]], shape=(1, 4, 4), dtype=float32) # .. expands to all unspecified axes. op_shape(x@r[]) ## shape(1, 4, 4, 3) op_shape(x@r[..]) ## shape(1, 4, 4, 3) op_shape(x@r[1, ..]) ## shape(4, 4, 3) op_shape(x@r[1, .., 1, 1]) ## shape(4) op_shape(x@r[1, 1, 1, .., 1]) ## shape() # op_subset<- uses the same semantics, but note that not all tensors # support modification. E.g., TensorFlow constant tensors cannot be modified, # while TensorFlow Variables can be.  (x <- tensorflow::tf$Variable(matrix(1, nrow = 2, ncol = 3))) ## <tf.Variable 'Variable:0' shape=(2, 3) dtype=float64, numpy= ## array([[1., 1., 1.], ##        [1., 1., 1.]])> op_subset(x, 1) <- 9 x ## <tf.Variable 'UnreadVariable' shape=(2, 3) dtype=float64, numpy= ## array([[9., 9., 9.], ##        [1., 1., 1.]])> x@r[1,1] <- 33 x ## <tf.Variable 'UnreadVariable' shape=(2, 3) dtype=float64, numpy= ## array([[33.,  9.,  9.], ##        [ 1.,  1.,  1.]])>"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_subtract.html","id":null,"dir":"Reference","previous_headings":"","what":"Subtract arguments element-wise. — op_subtract","title":"Subtract arguments element-wise. — op_subtract","text":"Note function automatically called using R operator - tensor.","code":"x <- op_ones(c(3)) op_subtract(x, x) ## tf.Tensor([0. 0. 0.], shape=(3), dtype=float32) x - x ## tf.Tensor([0. 0. 0.], shape=(3), dtype=float32)"},{"path":"https://keras3.posit.co/dev/reference/op_subtract.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Subtract arguments element-wise. — op_subtract","text":"","code":"op_subtract(x1, x2)"},{"path":"https://keras3.posit.co/dev/reference/op_subtract.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Subtract arguments element-wise. — op_subtract","text":"x1 First input tensor. x2 Second input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_subtract.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Subtract arguments element-wise. — op_subtract","text":"Output tensor, element-wise difference x1 x2.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_sum.html","id":null,"dir":"Reference","previous_headings":"","what":"Sum of a tensor over the given axes. — op_sum","title":"Sum of a tensor over the given axes. — op_sum","text":"Sum tensor given axes.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_sum.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Sum of a tensor over the given axes. — op_sum","text":"","code":"op_sum(x, axis = NULL, keepdims = FALSE)"},{"path":"https://keras3.posit.co/dev/reference/op_sum.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Sum of a tensor over the given axes. — op_sum","text":"x Input tensor. axis Axis axes along sum computed. default compute sum flattened tensor. keepdims set TRUE, axes reduced left result dimensions size one.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_sum.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Sum of a tensor over the given axes. — op_sum","text":"Output tensor containing sum.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_svd.html","id":null,"dir":"Reference","previous_headings":"","what":"Computes the singular value decomposition of a matrix. — op_svd","title":"Computes the singular value decomposition of a matrix. — op_svd","text":"Computes singular value decomposition matrix.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_svd.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes the singular value decomposition of a matrix. — op_svd","text":"","code":"op_svd(x, full_matrices = TRUE, compute_uv = TRUE)"},{"path":"https://keras3.posit.co/dev/reference/op_svd.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Computes the singular value decomposition of a matrix. — op_svd","text":"x Input tensor shape (..., M, N). full_matrices Logical compute_uv Logical","code":""},{"path":"https://keras3.posit.co/dev/reference/op_svd.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Computes the singular value decomposition of a matrix. — op_svd","text":"list three tensors: tensor shape (..., M, M) containing left singular vectors, tensor shape (..., M, N) containing singular values tensor shape (..., N, N) containing right singular vectors.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_swapaxes.html","id":null,"dir":"Reference","previous_headings":"","what":"Interchange two axes of a tensor. — op_swapaxes","title":"Interchange two axes of a tensor. — op_swapaxes","text":"Interchange two axes tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_swapaxes.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Interchange two axes of a tensor. — op_swapaxes","text":"","code":"op_swapaxes(x, axis1, axis2)"},{"path":"https://keras3.posit.co/dev/reference/op_swapaxes.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Interchange two axes of a tensor. — op_swapaxes","text":"x Input tensor. axis1 First axis. axis2 Second axis.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_swapaxes.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Interchange two axes of a tensor. — op_swapaxes","text":"tensor axes swapped.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_switch.html","id":null,"dir":"Reference","previous_headings":"","what":"Apply exactly one of the branches given by index. — op_switch","title":"Apply exactly one of the branches given by index. — op_switch","text":"index bounds, clamped within bounds. semantics switch given roughly implementation:","code":"op_switch <- function(index, branches, ...) {   index <- op_clip(index, 1, length(branches))   branches[[index]](...) }"},{"path":"https://keras3.posit.co/dev/reference/op_switch.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Apply exactly one of the branches given by index. — op_switch","text":"","code":"op_switch(index, branches, ...)"},{"path":"https://keras3.posit.co/dev/reference/op_switch.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Apply exactly one of the branches given by index. — op_switch","text":"index integer scalar indicating branch function apply (1-based). branches list functions applied based index. ... Inputs whichever branch applied.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_switch.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Apply exactly one of the branches given by index. — op_switch","text":"outputs branch(...) branch selected based index.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_switch.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Apply exactly one of the branches given by index. — op_switch","text":"","code":"add_fn <- function(x, y) x + y + 100 subtract_fn <- function(x, y) x - y - 100 x <- op_array(2) y <- op_array(0.5) branches <- list(add_fn, subtract_fn) op_switch(0, branches, x, y) # + ## tf.Tensor(102.5, shape=(), dtype=float32) op_switch(1, branches, x, y) # + ## tf.Tensor(102.5, shape=(), dtype=float32) op_switch(2, branches, x, y) # - ## tf.Tensor(-98.5, shape=(), dtype=float32) op_switch(3, branches, x, y) # - ## tf.Tensor(-98.5, shape=(), dtype=float32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_take.html","id":null,"dir":"Reference","previous_headings":"","what":"Take elements from a tensor along an axis. — op_take","title":"Take elements from a tensor along an axis. — op_take","text":"Take elements tensor along axis.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_take.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Take elements from a tensor along an axis. — op_take","text":"","code":"op_take(x, indices, axis = NULL, zero_indexed = FALSE)"},{"path":"https://keras3.posit.co/dev/reference/op_take.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Take elements from a tensor along an axis. — op_take","text":"x Source tensor. indices indices values extract. axis axis select values. default, flattened input tensor used. zero_indexed TRUE, treats indices zero-based (0 encodes first position); FALSE (default), treats indices one-based (1 encodes first position).","code":""},{"path":"https://keras3.posit.co/dev/reference/op_take.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Take elements from a tensor along an axis. — op_take","text":"corresponding tensor values.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_take_along_axis.html","id":null,"dir":"Reference","previous_headings":"","what":"Select values from x at the 1-D indices along the given axis. — op_take_along_axis","title":"Select values from x at the 1-D indices along the given axis. — op_take_along_axis","text":"Select values x 1-D indices along given axis.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_take_along_axis.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Select values from x at the 1-D indices along the given axis. — op_take_along_axis","text":"","code":"op_take_along_axis(x, indices, axis = NULL, zero_indexed = FALSE)"},{"path":"https://keras3.posit.co/dev/reference/op_take_along_axis.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Select values from x at the 1-D indices along the given axis. — op_take_along_axis","text":"x Source tensor. indices indices values extract. axis axis select values. default, flattened input tensor used. zero_indexed TRUE, treats indices zero-based (0 encodes first position); FALSE (default), treats indices one-based (1 encodes first position).","code":""},{"path":"https://keras3.posit.co/dev/reference/op_take_along_axis.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Select values from x at the 1-D indices along the given axis. — op_take_along_axis","text":"corresponding tensor values.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_tan.html","id":null,"dir":"Reference","previous_headings":"","what":"Compute tangent, element-wise. — op_tan","title":"Compute tangent, element-wise. — op_tan","text":"Compute tangent, element-wise.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_tan.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Compute tangent, element-wise. — op_tan","text":"","code":"op_tan(x)"},{"path":"https://keras3.posit.co/dev/reference/op_tan.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Compute tangent, element-wise. — op_tan","text":"x Input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_tan.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Compute tangent, element-wise. — op_tan","text":"Output tensor shape x.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_tanh.html","id":null,"dir":"Reference","previous_headings":"","what":"Hyperbolic tangent, element-wise. — op_tanh","title":"Hyperbolic tangent, element-wise. — op_tanh","text":"Hyperbolic tangent, element-wise.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_tanh.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Hyperbolic tangent, element-wise. — op_tanh","text":"","code":"op_tanh(x)"},{"path":"https://keras3.posit.co/dev/reference/op_tanh.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Hyperbolic tangent, element-wise. — op_tanh","text":"x Input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_tanh.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Hyperbolic tangent, element-wise. — op_tanh","text":"Output tensor shape x.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_tanh_shrink.html","id":null,"dir":"Reference","previous_headings":"","what":"Applies the tanh shrink function element-wise. — op_tanh_shrink","title":"Applies the tanh shrink function element-wise. — op_tanh_shrink","text":"defined : f(x) = x - tanh(x).","code":""},{"path":"https://keras3.posit.co/dev/reference/op_tanh_shrink.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Applies the tanh shrink function element-wise. — op_tanh_shrink","text":"","code":"op_tanh_shrink(x)"},{"path":"https://keras3.posit.co/dev/reference/op_tanh_shrink.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Applies the tanh shrink function element-wise. — op_tanh_shrink","text":"x Input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_tanh_shrink.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Applies the tanh shrink function element-wise. — op_tanh_shrink","text":"Output tensor shape x, element transformed according tanh shrink operation.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_tanh_shrink.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Applies the tanh shrink function element-wise. — op_tanh_shrink","text":"","code":"x <- op_array(c(-1., 0., 1.)) op_tanh_shrink(x) ## tf.Tensor([-0.23840582  0.          0.23840582], shape=(3), dtype=float32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_tensordot.html","id":null,"dir":"Reference","previous_headings":"","what":"Compute the tensor dot product along specified axes. — op_tensordot","title":"Compute the tensor dot product along specified axes. — op_tensordot","text":"Compute tensor dot product along specified axes.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_tensordot.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Compute the tensor dot product along specified axes. — op_tensordot","text":"","code":"op_tensordot(x1, x2, axes = 3L)"},{"path":"https://keras3.posit.co/dev/reference/op_tensordot.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Compute the tensor dot product along specified axes. — op_tensordot","text":"x1 First tensor. x2 Second tensor. axes integer, N, sum last N axes x1 first N axes x2 order. sizes corresponding axes must match. , list axes summed , first sequence applying x1, second x2. sequences must length.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_tensordot.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Compute the tensor dot product along specified axes. — op_tensordot","text":"tensor dot product inputs.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_threshold.html","id":null,"dir":"Reference","previous_headings":"","what":"Threshold activation function. — op_threshold","title":"Threshold activation function. — op_threshold","text":"function thresholds input x follows: f(x) = x x > threshold, f(x) = default_value otherwise.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_threshold.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Threshold activation function. — op_threshold","text":"","code":"op_threshold(x, threshold, default_value)"},{"path":"https://keras3.posit.co/dev/reference/op_threshold.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Threshold activation function. — op_threshold","text":"x Input tensor. threshold value decides retain replace x. default_value Value assign x <= threshold.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_threshold.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Threshold activation function. — op_threshold","text":"tensor shape x.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_threshold.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Threshold activation function. — op_threshold","text":"","code":"x <- op_array(c(-1.0, 0.0, 1.0, 2.0)) op_threshold(x, 1, 0) ## tf.Tensor([0. 0. 0. 2.], shape=(4), dtype=float32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_tile.html","id":null,"dir":"Reference","previous_headings":"","what":"Repeat x the number of times given by repeats. — op_tile","title":"Repeat x the number of times given by repeats. — op_tile","text":"repeats length d, result dimension max(d, x.ndim). x.ndim < d, x promoted d-dimensional prepending new axes. x.ndim > d, repeats promoted x.ndim prepending 1's .","code":""},{"path":"https://keras3.posit.co/dev/reference/op_tile.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Repeat x the number of times given by repeats. — op_tile","text":"","code":"op_tile(x, repeats)"},{"path":"https://keras3.posit.co/dev/reference/op_tile.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Repeat x the number of times given by repeats. — op_tile","text":"x Input tensor. repeats number repetitions x along axis.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_tile.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Repeat x the number of times given by repeats. — op_tile","text":"tiled output tensor.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_top_k.html","id":null,"dir":"Reference","previous_headings":"","what":"Finds the top-k values and their indices in a tensor. — op_top_k","title":"Finds the top-k values and their indices in a tensor. — op_top_k","text":"Finds top-k values indices tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_top_k.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Finds the top-k values and their indices in a tensor. — op_top_k","text":"","code":"op_top_k(x, k, sorted = TRUE, zero_indexed = FALSE)"},{"path":"https://keras3.posit.co/dev/reference/op_top_k.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Finds the top-k values and their indices in a tensor. — op_top_k","text":"x Input tensor. k integer representing number top elements retrieve. sorted boolean indicating whether sort output descending order. Defaults TRUE. zero_indexed TRUE, returned indices zero-based (0 encodes first position); FALSE (default), returned indices one-based (1 encodes first position).","code":""},{"path":"https://keras3.posit.co/dev/reference/op_top_k.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Finds the top-k values and their indices in a tensor. — op_top_k","text":"list containing two tensors. first tensor contains top-k values, second tensor contains indices top-k values input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_top_k.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Finds the top-k values and their indices in a tensor. — op_top_k","text":"","code":"x <- op_array(c(5, 2, 7, 1, 9, 3), \"int32\") op_top_k(x, k = 3) ## $values ## tf.Tensor([9 7 5], shape=(3), dtype=int32) ## ## $indices ## tf.Tensor([5 3 1], shape=(3), dtype=int32) c(values, indices) %<-% op_top_k(x, k = 3) values ## tf.Tensor([9 7 5], shape=(3), dtype=int32) indices ## tf.Tensor([5 3 1], shape=(3), dtype=int32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_trace.html","id":null,"dir":"Reference","previous_headings":"","what":"Return the sum along diagonals of the tensor. — op_trace","title":"Return the sum along diagonals of the tensor. — op_trace","text":"x 2-D, sum along diagonal given offset returned, .e., sum elements x[, +offset] . two dimensions, axes specified axis1 axis2 used determine 2-D sub-arrays whose traces returned. shape resulting tensor x axis1 axis2 removed.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_trace.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Return the sum along diagonals of the tensor. — op_trace","text":"","code":"op_trace(x, offset = 0L, axis1 = 1L, axis2 = 2L)"},{"path":"https://keras3.posit.co/dev/reference/op_trace.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Return the sum along diagonals of the tensor. — op_trace","text":"x Input tensor. offset Offset diagonal main diagonal. Can positive negative. Defaults 0. axis1 Axis used first axis 2-D sub-arrays. Defaults 1. (first axis). axis2 Axis used second axis 2-D sub-arrays. Defaults 2. (second axis).","code":""},{"path":"https://keras3.posit.co/dev/reference/op_trace.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Return the sum along diagonals of the tensor. — op_trace","text":"x 2-D, sum diagonal returned. x larger dimensions, tensor sums along diagonals returned.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_transpose.html","id":null,"dir":"Reference","previous_headings":"","what":"Returns a tensor with axes transposed. — op_transpose","title":"Returns a tensor with axes transposed. — op_transpose","text":"Returns tensor axes transposed.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_transpose.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Returns a tensor with axes transposed. — op_transpose","text":"","code":"op_transpose(x, axes = NULL)"},{"path":"https://keras3.posit.co/dev/reference/op_transpose.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Returns a tensor with axes transposed. — op_transpose","text":"x Input tensor. axes Sequence integers. Permutation dimensions x. default, order axes reversed.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_transpose.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Returns a tensor with axes transposed. — op_transpose","text":"x axes permuted.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_tri.html","id":null,"dir":"Reference","previous_headings":"","what":"Return a tensor with ones at and below a diagonal and zeros elsewhere. — op_tri","title":"Return a tensor with ones at and below a diagonal and zeros elsewhere. — op_tri","text":"Return tensor ones diagonal zeros elsewhere.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_tri.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Return a tensor with ones at and below a diagonal and zeros elsewhere. — op_tri","text":"","code":"op_tri(N, M = NULL, k = 0L, dtype = NULL)"},{"path":"https://keras3.posit.co/dev/reference/op_tri.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Return a tensor with ones at and below a diagonal and zeros elsewhere. — op_tri","text":"N Number rows tensor. M Number columns tensor. k sub-diagonal array filled. k = 0 main diagonal, k < 0 , k > 0 . default 0. dtype Data type returned tensor. default \"float32\".","code":""},{"path":"https://keras3.posit.co/dev/reference/op_tri.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Return a tensor with ones at and below a diagonal and zeros elsewhere. — op_tri","text":"Tensor lower triangle filled ones zeros elsewhere. T[, j] == 1 j <= + k, 0 otherwise.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_tril.html","id":null,"dir":"Reference","previous_headings":"","what":"Return lower triangle of a tensor. — op_tril","title":"Return lower triangle of a tensor. — op_tril","text":"tensors ndim exceeding 2, tril apply final two axes.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_tril.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Return lower triangle of a tensor. — op_tril","text":"","code":"op_tril(x, k = 0L)"},{"path":"https://keras3.posit.co/dev/reference/op_tril.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Return lower triangle of a tensor. — op_tril","text":"x Input tensor. k Diagonal zero elements. Defaults 0. main diagonal. k < 0 , k > 0 .","code":""},{"path":"https://keras3.posit.co/dev/reference/op_tril.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Return lower triangle of a tensor. — op_tril","text":"Lower triangle x, shape data type x.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_triu.html","id":null,"dir":"Reference","previous_headings":"","what":"Return upper triangle of a tensor. — op_triu","title":"Return upper triangle of a tensor. — op_triu","text":"tensors ndim exceeding 2, triu apply final two axes.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_triu.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Return upper triangle of a tensor. — op_triu","text":"","code":"op_triu(x, k = 0L)"},{"path":"https://keras3.posit.co/dev/reference/op_triu.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Return upper triangle of a tensor. — op_triu","text":"x Input tensor. k Diagonal zero elements. Defaults 0. main diagonal. k < 0 , k > 0 .","code":""},{"path":"https://keras3.posit.co/dev/reference/op_triu.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Return upper triangle of a tensor. — op_triu","text":"Upper triangle x, shape data type x.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_trunc.html","id":null,"dir":"Reference","previous_headings":"","what":"Return the truncated value of the input, element-wise. — op_trunc","title":"Return the truncated value of the input, element-wise. — op_trunc","text":"truncated value scalar x nearest integer closer zero x . short, fractional part signed number x discarded.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_trunc.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Return the truncated value of the input, element-wise. — op_trunc","text":"","code":"op_trunc(x)"},{"path":"https://keras3.posit.co/dev/reference/op_trunc.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Return the truncated value of the input, element-wise. — op_trunc","text":"x Input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_trunc.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Return the truncated value of the input, element-wise. — op_trunc","text":"truncated value element x.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_trunc.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Return the truncated value of the input, element-wise. — op_trunc","text":"","code":"x <- op_array(c(-1.7, -1.5, -0.2, 0.2, 1.5, 1.7, 2.0)) op_trunc(x) ## tf.Tensor([-1. -1. -0.  0.  1.  1.  2.], shape=(7), dtype=float32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_unravel_index.html","id":null,"dir":"Reference","previous_headings":"","what":"Convert flat indices to coordinate arrays in a given array shape. — op_unravel_index","title":"Convert flat indices to coordinate arrays in a given array shape. — op_unravel_index","text":"Convert flat indices coordinate arrays given array shape.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_unravel_index.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Convert flat indices to coordinate arrays in a given array shape. — op_unravel_index","text":"","code":"op_unravel_index(indices, shape)"},{"path":"https://keras3.posit.co/dev/reference/op_unravel_index.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Convert flat indices to coordinate arrays in a given array shape. — op_unravel_index","text":"indices integer array integers representing flat indices. shape shape array unravel .","code":""},{"path":"https://keras3.posit.co/dev/reference/op_unravel_index.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Convert flat indices to coordinate arrays in a given array shape. — op_unravel_index","text":"Tuple arrays dimension unraveled indices.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_unravel_index.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Convert flat indices to coordinate arrays in a given array shape. — op_unravel_index","text":"","code":"indices <- c(1, 5) shape <- array(c(3, 3)) op_unravel_index(indices, shape) ## [[1]] ## tf.Tensor([0 1], shape=(2), dtype=int32) ## ## [[2]] ## tf.Tensor([0 1], shape=(2), dtype=int32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_unstack.html","id":null,"dir":"Reference","previous_headings":"","what":"Unpacks the given dimension of a rank-R tensor into rank-(R-1) tensors. — op_unstack","title":"Unpacks the given dimension of a rank-R tensor into rank-(R-1) tensors. — op_unstack","text":"Unpacks given dimension rank-R tensor rank-(R-1) tensors.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_unstack.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Unpacks the given dimension of a rank-R tensor into rank-(R-1) tensors. — op_unstack","text":"","code":"op_unstack(x, num = NULL, axis = 1L)"},{"path":"https://keras3.posit.co/dev/reference/op_unstack.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Unpacks the given dimension of a rank-R tensor into rank-(R-1) tensors. — op_unstack","text":"x input tensor. num length dimension axis. Automatically inferred NULL. axis axis along unpack.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_unstack.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Unpacks the given dimension of a rank-R tensor into rank-(R-1) tensors. — op_unstack","text":"list tensors unpacked along given axis.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_unstack.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Unpacks the given dimension of a rank-R tensor into rank-(R-1) tensors. — op_unstack","text":"[3, 4))]: R:3,%204))","code":"x <- op_array(rbind(c(1, 2),                    c(3, 4))) op_unstack(x, axis=1) ## [[1]] ## tf.Tensor([1. 2.], shape=(2), dtype=float64) ## ## [[2]] ## tf.Tensor([3. 4.], shape=(2), dtype=float64) op_unstack(x, axis=2) ## [[1]] ## tf.Tensor([1. 3.], shape=(2), dtype=float64) ## ## [[2]] ## tf.Tensor([2. 4.], shape=(2), dtype=float64) all.equal(op_unstack(x),           op_unstack(x, axis = 1)) ## [1] TRUE all.equal(op_unstack(x, axis = -1),           op_unstack(x, axis = 2)) ## [1] TRUE # [array([1, 2)), array([3, 4))]"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_var.html","id":null,"dir":"Reference","previous_headings":"","what":"Compute the variance along the specified axes. — op_var","title":"Compute the variance along the specified axes. — op_var","text":"Compute variance along specified axes.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_var.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Compute the variance along the specified axes. — op_var","text":"","code":"op_var(x, axis = NULL, keepdims = FALSE)"},{"path":"https://keras3.posit.co/dev/reference/op_var.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Compute the variance along the specified axes. — op_var","text":"x Input tensor. axis Axis axes along variance computed. default compute variance flattened tensor. keepdims set TRUE, axes reduced left result dimensions size one.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_var.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Compute the variance along the specified axes. — op_var","text":"Output tensor containing variance.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_vdot.html","id":null,"dir":"Reference","previous_headings":"","what":"Return the dot product of two vectors. — op_vdot","title":"Return the dot product of two vectors. — op_vdot","text":"first argument complex, complex conjugate first argument used calculation dot product. Multidimensional tensors flattened dot product taken.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_vdot.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Return the dot product of two vectors. — op_vdot","text":"","code":"op_vdot(x1, x2)"},{"path":"https://keras3.posit.co/dev/reference/op_vdot.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Return the dot product of two vectors. — op_vdot","text":"x1 First input tensor. complex, complex conjugate taken calculation dot product. x2 Second input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_vdot.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Return the dot product of two vectors. — op_vdot","text":"Output tensor.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_vectorize.html","id":null,"dir":"Reference","previous_headings":"","what":"Turn a function into a vectorized function. — op_vectorize","title":"Turn a function into a vectorized function. — op_vectorize","text":"Turn function vectorized function.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_vectorize.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Turn a function into a vectorized function. — op_vectorize","text":"","code":"op_vectorize(func, ..., excluded = NULL, signature = NULL)"},{"path":"https://keras3.posit.co/dev/reference/op_vectorize.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Turn a function into a vectorized function. — op_vectorize","text":"func Callable single tensor argument. ... forward/backward compatability. excluded Optional set integers representing positional arguments function vectorized. passed directly func unmodified. signature Optional generalized universal function signature, e.g., \"(m,n),(n)->(m)\" vectorized matrix-vector multiplication. provided, func called (expected return) arrays shapes given size corresponding core dimensions. default, func assumed take scalar tensors input output.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_vectorize.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Turn a function into a vectorized function. — op_vectorize","text":"new function applies func every element input along axis 1 (batch axis, first axis).","code":""},{"path":"https://keras3.posit.co/dev/reference/op_vectorize.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Turn a function into a vectorized function. — op_vectorize","text":"","code":"# currently does not work w/ tensorflow backend if(config_backend() != \"tensorflow\") {    myfunc <- function(a, b) a + b    vfunc <- op_vectorize(myfunc)   y <- vfunc(c(1, 2, 3, 4), 2)   print(y)   # with Jax backend, y is:   # Array([3., 4., 5., 6.], dtype=float32) }"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_vectorized_map.html","id":null,"dir":"Reference","previous_headings":"","what":"Parallel map of function f on the first axis of tensor(s) elements. — op_vectorized_map","title":"Parallel map of function f on the first axis of tensor(s) elements. — op_vectorized_map","text":"Schematically, op_vectorized_map() maps first dimension provided tensors. elements list tensors, tensors required size first dimension, iterated together.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_vectorized_map.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Parallel map of function f on the first axis of tensor(s) elements. — op_vectorized_map","text":"","code":"op_vectorized_map(elements, f)"},{"path":"https://keras3.posit.co/dev/reference/op_vectorized_map.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Parallel map of function f on the first axis of tensor(s) elements. — op_vectorized_map","text":"elements see description f function taking either tensor, list tensors.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_vectorized_map.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Parallel map of function f on the first axis of tensor(s) elements. — op_vectorized_map","text":"tensor list tensors, result mapping f across elements.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_vectorized_map.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Parallel map of function f on the first axis of tensor(s) elements. — op_vectorized_map","text":"Note f may traced compiled. Meaning, R function may evaluated symbolic tensors using Jax TensorFlow backends, eager tensors. See output str() examples:","code":"(x <- op_arange(12L) |> op_reshape(c(3, 4))) ## tf.Tensor( ## [[ 1  2  3  4] ##  [ 5  6  7  8] ##  [ 9 10 11 12]], shape=(3, 4), dtype=int32) x |> op_vectorized_map(\\(row) {row + 10}) ## tf.Tensor( ## [[11 12 13 14] ##  [15 16 17 18] ##  [19 20 21 22]], shape=(3, 4), dtype=int32) list(x, x, x) |> op_vectorized_map(\\(rows) Reduce(`+`, rows)) ## tf.Tensor( ## [[ 3  6  9 12] ##  [15 18 21 24] ##  [27 30 33 36]], shape=(3, 4), dtype=int32) # simplest case, map f over rows of x, # where .x is 1 row of x input <- x output <- op_vectorized_map(input, function(.x) {   str(.x)   .x + 10 }) ## <tf.Tensor 'loop_body/GatherV2:0' shape=(4) dtype=int32> output ## tf.Tensor( ## [[11 12 13 14] ##  [15 16 17 18] ##  [19 20 21 22]], shape=(3, 4), dtype=int32) # map f over two tensors simultaneously. Here, # `.x` is a list of two # tensors. The return values from each call of `f(row)` are stacked to form the # final output input <- list(x, x) output <- op_vectorized_map(input, function(.x) {   str(.x)   .x[[1]] + 10 }) ## List of 2 ##  $ :<tf.Tensor 'loop_body/GatherV2:0' shape=(4) dtype=int32> ##  $ :<tf.Tensor 'loop_body/GatherV2_1:0' shape=(4) dtype=int32> output ## tf.Tensor( ## [[11 12 13 14] ##  [15 16 17 18] ##  [19 20 21 22]], shape=(3, 4), dtype=int32) # same as above, but now returning two tensors in the final output output <- op_vectorized_map(input, function(.x) {   str(.x)   c(.x1, .x2) %<-% .x   list(.x1+10, .x2+20) }) ## List of 2 ##  $ :<tf.Tensor 'loop_body/GatherV2:0' shape=(4) dtype=int32> ##  $ :<tf.Tensor 'loop_body/GatherV2_1:0' shape=(4) dtype=int32> output ## [[1]] ## tf.Tensor( ## [[11 12 13 14] ##  [15 16 17 18] ##  [19 20 21 22]], shape=(3, 4), dtype=int32) ## ## [[2]] ## tf.Tensor( ## [[21 22 23 24] ##  [25 26 27 28] ##  [29 30 31 32]], shape=(3, 4), dtype=int32) # passing named lists. # WARNING: if passing a named list, the order of elements of `.x` supplied # to `f` is not stable. Only retrieve elements by name. input <- list(name1 = x, name2 = x) output <- op_vectorized_map(input, function(.x) {   str(.x)   list(outname1 = .x$name1 + 10,        outname2 = .x$name2 + 20) }) ## List of 2 ##  $ name1:<tf.Tensor 'loop_body/GatherV2:0' shape=(4) dtype=int32> ##  $ name2:<tf.Tensor 'loop_body/GatherV2_1:0' shape=(4) dtype=int32> output ## $outname1 ## tf.Tensor( ## [[11 12 13 14] ##  [15 16 17 18] ##  [19 20 21 22]], shape=(3, 4), dtype=int32) ## ## $outname2 ## tf.Tensor( ## [[21 22 23 24] ##  [25 26 27 28] ##  [29 30 31 32]], shape=(3, 4), dtype=int32) # passing a tuple() is equivalent to passing an unnamed list() input <- tuple(x, x) output <- op_vectorized_map(input, function(.x) {   str(.x)   list(.x[[1]] + 10) }) ## List of 2 ##  $ :<tf.Tensor 'loop_body/GatherV2:0' shape=(4) dtype=int32> ##  $ :<tf.Tensor 'loop_body/GatherV2_1:0' shape=(4) dtype=int32> output ## [[1]] ## tf.Tensor( ## [[11 12 13 14] ##  [15 16 17 18] ##  [19 20 21 22]], shape=(3, 4), dtype=int32)"},{"path":"https://keras3.posit.co/dev/reference/op_vectorized_map.html","id":"debugging-f","dir":"Reference","previous_headings":"","what":"Debugging f","title":"Parallel map of function f on the first axis of tensor(s) elements. — op_vectorized_map","text":"Even eager contexts, op_vectorized_map() may trace f. case, want eagerly debug f (e.g., browser()), can swap manual (slow) implementation op_vectorized_map(). Note example debug implementation handle edge cases op_vectorized_map(), particular, f returns structure multiple tensors.","code":"op_vectorized_map_debug <- function(elements, fn) {    if (!is.list(elements)) {     # `elements` is a single tensor     batch_size <- op_shape(elements)[[1]]     out <- elements |>       op_split(batch_size) |>       lapply(fn) |>       op_stack()     return(out)   }    # `elements` is a list of tensors   batch_size <- elements[[1]] |> op_shape() |> _[[1]]   elements |>     lapply(\\(e) op_split(e, batch_size)) |>     zip_lists() |>     lapply(fn) |>     op_stack()  }"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_vstack.html","id":null,"dir":"Reference","previous_headings":"","what":"Stack tensors in sequence vertically (row wise). — op_vstack","title":"Stack tensors in sequence vertically (row wise). — op_vstack","text":"Stack tensors sequence vertically (row wise).","code":""},{"path":"https://keras3.posit.co/dev/reference/op_vstack.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Stack tensors in sequence vertically (row wise). — op_vstack","text":"","code":"op_vstack(xs, ...)"},{"path":"https://keras3.posit.co/dev/reference/op_vstack.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Stack tensors in sequence vertically (row wise). — op_vstack","text":"xs, ... Sequence tensors.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_vstack.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Stack tensors in sequence vertically (row wise). — op_vstack","text":"Tensor formed stacking given tensors.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_where.html","id":null,"dir":"Reference","previous_headings":"","what":"Return elements chosen from x1 or x2 depending on condition. — op_where","title":"Return elements chosen from x1 or x2 depending on condition. — op_where","text":"Return elements chosen x1 x2 depending condition.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_where.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Return elements chosen from x1 or x2 depending on condition. — op_where","text":"","code":"op_where(condition, x1 = NULL, x2 = NULL)"},{"path":"https://keras3.posit.co/dev/reference/op_where.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Return elements chosen from x1 or x2 depending on condition. — op_where","text":"condition TRUE, yield x1, otherwise yield x2. x1 Values choose condition TRUE. x2 Values choose condition FALSE.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_where.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Return elements chosen from x1 or x2 depending on condition. — op_where","text":"tensor elements x1 condition TRUE, elements x2 condition FALSE.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_while_loop.html","id":null,"dir":"Reference","previous_headings":"","what":"While loop implementation. — op_while_loop","title":"While loop implementation. — op_while_loop","text":"loop implementation.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_while_loop.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"While loop implementation. — op_while_loop","text":"","code":"op_while_loop(cond, body, loop_vars, maximum_iterations = NULL)"},{"path":"https://keras3.posit.co/dev/reference/op_while_loop.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"While loop implementation. — op_while_loop","text":"cond callable represents termination condition loop. Must accept loop_vars like structure argument. loop_vars tuple unnamed list, element loop_vars passed positionally callable. body callable represents loop body. Must accept loop_vars like structure argument, return update value structure. loop_vars tuple unnamed list, element loop_vars passed positionally callable. loop_vars arbitrary nested structure tensor state persist across loop iterations. maximum_iterations Optional maximum number iterations loop run. provided, cond output -ed additional condition ensuring number iterations executed greater maximum_iterations.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_while_loop.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"While loop implementation. — op_while_loop","text":"list tensors, shape dtype loop_vars.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_while_loop.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"While loop implementation. — op_while_loop","text":"","code":"i <- 0 loop_vars <- list(i)  # cond() must return a scalar bool cond <- function(i) i < 10L  # body must return same shape as loop_vars body <- function(i) list(i + 1L)  op_while_loop(cond, body, loop_vars) ## [[1]] ## tf.Tensor(10.0, shape=(), dtype=float32) x <- 0; y <- 1 cond <- \\(x, y) x < 10 body <- \\(x, y) list(x+1, y+1) op_while_loop(cond, body, list(x, y)) ## [[1]] ## tf.Tensor(10.0, shape=(), dtype=float32) ## ## [[2]] ## tf.Tensor(11.0, shape=(), dtype=float32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_zeros.html","id":null,"dir":"Reference","previous_headings":"","what":"Return a new tensor of given shape and type, filled with zeros. — op_zeros","title":"Return a new tensor of given shape and type, filled with zeros. — op_zeros","text":"Return new tensor given shape type, filled zeros.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_zeros.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Return a new tensor of given shape and type, filled with zeros. — op_zeros","text":"","code":"op_zeros(shape, dtype = NULL)"},{"path":"https://keras3.posit.co/dev/reference/op_zeros.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Return a new tensor of given shape and type, filled with zeros. — op_zeros","text":"shape Shape new tensor. dtype Desired data type tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_zeros.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Return a new tensor of given shape and type, filled with zeros. — op_zeros","text":"Tensor zeros given shape dtype.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_zeros_like.html","id":null,"dir":"Reference","previous_headings":"","what":"Return a tensor of zeros with the same shape and type as x. — op_zeros_like","title":"Return a tensor of zeros with the same shape and type as x. — op_zeros_like","text":"Return tensor zeros shape type x.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_zeros_like.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Return a tensor of zeros with the same shape and type as x. — op_zeros_like","text":"","code":"op_zeros_like(x, dtype = NULL)"},{"path":"https://keras3.posit.co/dev/reference/op_zeros_like.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Return a tensor of zeros with the same shape and type as x. — op_zeros_like","text":"x Input tensor. dtype Overrides data type result.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_zeros_like.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Return a tensor of zeros with the same shape and type as x. — op_zeros_like","text":"tensor zeros shape type x.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/optimizer_adadelta.html","id":null,"dir":"Reference","previous_headings":"","what":"Optimizer that implements the Adadelta algorithm. — optimizer_adadelta","title":"Optimizer that implements the Adadelta algorithm. — optimizer_adadelta","text":"Adadelta optimization stochastic gradient descent method based adaptive learning rate per dimension address two drawbacks: continual decay learning rates throughout training. need manually selected global learning rate. Adadelta robust extension Adagrad adapts learning rates based moving window gradient updates, instead accumulating past gradients. way, Adadelta continues learning even many updates done. Compared Adagrad, original version Adadelta set initial learning rate. version, initial learning rate can set, Keras optimizers.","code":""},{"path":"https://keras3.posit.co/dev/reference/optimizer_adadelta.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Optimizer that implements the Adadelta algorithm. — optimizer_adadelta","text":"","code":"optimizer_adadelta(   learning_rate = 0.001,   rho = 0.95,   epsilon = 1e-07,   weight_decay = NULL,   clipnorm = NULL,   clipvalue = NULL,   global_clipnorm = NULL,   use_ema = FALSE,   ema_momentum = 0.99,   ema_overwrite_frequency = NULL,   name = \"adadelta\",   ...,   loss_scale_factor = NULL,   gradient_accumulation_steps = NULL )"},{"path":"https://keras3.posit.co/dev/reference/optimizer_adadelta.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Optimizer that implements the Adadelta algorithm. — optimizer_adadelta","text":"learning_rate float, [LearningRateSchedule()] instance, callable takes arguments returns actual value use. learning rate. Defaults 0.001. Note Adadelta tends benefit higher initial learning rate values compared optimizers. match exact form original paper, use 1.0. rho floating point value. decay rate. Defaults 0.95. epsilon Small floating point value maintaining numerical stability. weight_decay Float. set, weight decay applied. clipnorm Float. set, gradient weight individually clipped norm higher value. clipvalue Float. set, gradient weight clipped higher value. global_clipnorm Float. set, gradient weights clipped global norm higher value. use_ema Boolean, defaults FALSE. TRUE, exponential moving average (EMA) applied. EMA consists computing exponential moving average weights model (weight values change training batch), periodically overwriting weights moving average. ema_momentum Float, defaults 0.99. used use_ema = TRUE. momentum use computing EMA model's weights: new_average = ema_momentum * old_average + (1 - ema_momentum) * current_variable_value. ema_overwrite_frequency Int NULL, defaults NULL. used use_ema = TRUE. Every ema_overwrite_frequency steps iterations, overwrite model variable moving average. NULL, optimizer overwrite model variables middle training, need explicitly overwrite variables end training calling optimizer$finalize_variable_values() (updates model variables -place). using built-fit() training loop, happens automatically last epoch, need anything. name String. name use momentum accumulator weights created optimizer. ... forward/backward compatability. loss_scale_factor Float NULL. float, scale factor multiplied loss computing gradients, inverse scale factor multiplied gradients updating variables. Useful preventing underflow mixed precision training. Alternately, optimizer_loss_scale() automatically set loss scale factor. gradient_accumulation_steps Int NULL. int, model optimizer variables updated every step; instead updated every gradient_accumulation_steps steps, using average value gradients since last update. known \"gradient accumulation\". can useful batch size small, order reduce gradient noise update step. EMA frequency look \"accumulated\" iterations value (optimizer steps // gradient_accumulation_steps). Learning rate schedules look \"real\" iterations value (optimizer steps).","code":""},{"path":"https://keras3.posit.co/dev/reference/optimizer_adadelta.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Optimizer that implements the Adadelta algorithm. — optimizer_adadelta","text":"Optimizer instance","code":""},{"path":"https://keras3.posit.co/dev/reference/optimizer_adadelta.html","id":"reference","dir":"Reference","previous_headings":"","what":"Reference","title":"Optimizer that implements the Adadelta algorithm. — optimizer_adadelta","text":"Zeiler, 2012","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/optimizer_adafactor.html","id":null,"dir":"Reference","previous_headings":"","what":"Optimizer that implements the Adafactor algorithm. — optimizer_adafactor","title":"Optimizer that implements the Adafactor algorithm. — optimizer_adafactor","text":"Adafactor commonly used NLP tasks, advantage taking less memory saves partial information previous gradients. default argument setup based original paper (see reference). gradients dimension > 2, Adafactor optimizer delete last 2 dimensions separately accumulator variables.","code":""},{"path":"https://keras3.posit.co/dev/reference/optimizer_adafactor.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Optimizer that implements the Adafactor algorithm. — optimizer_adafactor","text":"","code":"optimizer_adafactor(   learning_rate = 0.001,   beta_2_decay = -0.8,   epsilon_1 = 1e-30,   epsilon_2 = 0.001,   clip_threshold = 1,   relative_step = TRUE,   weight_decay = NULL,   clipnorm = NULL,   clipvalue = NULL,   global_clipnorm = NULL,   use_ema = FALSE,   ema_momentum = 0.99,   ema_overwrite_frequency = NULL,   name = \"adafactor\",   ...,   loss_scale_factor = NULL,   gradient_accumulation_steps = NULL )"},{"path":"https://keras3.posit.co/dev/reference/optimizer_adafactor.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Optimizer that implements the Adafactor algorithm. — optimizer_adafactor","text":"learning_rate float, LearningRateSchedule() instance, callable takes arguments returns actual value use. learning rate. Defaults 0.001. beta_2_decay float, defaults -0.8. decay rate beta_2. epsilon_1 float, defaults 1e-30. small offset keep denominator away 0. epsilon_2 float, defaults 1e-3. small offset avoid learning rate becoming small time. clip_threshold float, defaults 1.0. Clipping threshold. part Adafactor algorithm, independent clipnorm, clipvalue, global_clipnorm. relative_step bool, defaults TRUE. learning_rate constant relative_step=TRUE, learning rate adjusted based current iterations. default learning rate decay Adafactor. weight_decay Float. set, weight decay applied. clipnorm Float. set, gradient weight individually clipped norm higher value. clipvalue Float. set, gradient weight clipped higher value. global_clipnorm Float. set, gradient weights clipped global norm higher value. use_ema Boolean, defaults FALSE. TRUE, exponential moving average (EMA) applied. EMA consists computing exponential moving average weights model (weight values change training batch), periodically overwriting weights moving average. ema_momentum Float, defaults 0.99. used use_ema = TRUE. momentum use computing EMA model's weights: new_average = ema_momentum * old_average + (1 - ema_momentum) * current_variable_value. ema_overwrite_frequency Int NULL, defaults NULL. used use_ema=TRUE. Every ema_overwrite_frequency steps iterations, overwrite model variable moving average. NULL, optimizer overwrite model variables middle training, need explicitly overwrite variables end training calling optimizer$finalize_variable_values() (updates model variables -place). using built-fit() training loop, happens automatically last epoch, need anything. name String. name use momentum accumulator weights created optimizer. ... forward/backward compatability. loss_scale_factor Float NULL. float, scale factor multiplied loss computing gradients, inverse scale factor multiplied gradients updating variables. Useful preventing underflow mixed precision training. Alternately, optimizer_loss_scale() automatically set loss scale factor. gradient_accumulation_steps Int NULL. int, model optimizer variables updated every step; instead updated every gradient_accumulation_steps steps, using average value gradients since last update. known \"gradient accumulation\". can useful batch size small, order reduce gradient noise update step. EMA frequency look \"accumulated\" iterations value (optimizer steps // gradient_accumulation_steps). Learning rate schedules look \"real\" iterations value (optimizer steps).","code":""},{"path":"https://keras3.posit.co/dev/reference/optimizer_adafactor.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Optimizer that implements the Adafactor algorithm. — optimizer_adafactor","text":"Optimizer instance","code":""},{"path":"https://keras3.posit.co/dev/reference/optimizer_adafactor.html","id":"reference","dir":"Reference","previous_headings":"","what":"Reference","title":"Optimizer that implements the Adafactor algorithm. — optimizer_adafactor","text":"Shazeer, Noam et al., 2018.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/optimizer_adagrad.html","id":null,"dir":"Reference","previous_headings":"","what":"Optimizer that implements the Adagrad algorithm. — optimizer_adagrad","title":"Optimizer that implements the Adagrad algorithm. — optimizer_adagrad","text":"Adagrad optimizer parameter-specific learning rates, adapted relative frequently parameter gets updated training. updates parameter receives, smaller updates.","code":""},{"path":"https://keras3.posit.co/dev/reference/optimizer_adagrad.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Optimizer that implements the Adagrad algorithm. — optimizer_adagrad","text":"","code":"optimizer_adagrad(   learning_rate = 0.001,   initial_accumulator_value = 0.1,   epsilon = 1e-07,   weight_decay = NULL,   clipnorm = NULL,   clipvalue = NULL,   global_clipnorm = NULL,   use_ema = FALSE,   ema_momentum = 0.99,   ema_overwrite_frequency = NULL,   name = \"adagrad\",   ...,   loss_scale_factor = NULL,   gradient_accumulation_steps = NULL )"},{"path":"https://keras3.posit.co/dev/reference/optimizer_adagrad.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Optimizer that implements the Adagrad algorithm. — optimizer_adagrad","text":"learning_rate float, LearningRateSchedule() instance, callable takes arguments returns actual value use. learning rate. Defaults 0.001. Note Adagrad tends benefit higher initial learning rate values compared optimizers. match exact form original paper, use 1.0. initial_accumulator_value Floating point value. Starting value accumulators (per-parameter momentum values). Must non-negative. epsilon Small floating point value maintaining numerical stability. weight_decay Float. set, weight decay applied. clipnorm Float. set, gradient weight individually clipped norm higher value. clipvalue Float. set, gradient weight clipped higher value. global_clipnorm Float. set, gradient weights clipped global norm higher value. use_ema Boolean, defaults FALSE. TRUE, exponential moving average (EMA) applied. EMA consists computing exponential moving average weights model (weight values change training batch), periodically overwriting weights moving average. ema_momentum Float, defaults 0.99. used use_ema=TRUE. momentum use computing EMA model's weights: new_average = ema_momentum * old_average + (1 - ema_momentum) * current_variable_value. ema_overwrite_frequency Int NULL, defaults NULL. used use_ema=TRUE. Every ema_overwrite_frequency steps iterations, overwrite model variable moving average. NULL, optimizer overwrite model variables middle training, need explicitly overwrite variables end training calling optimizer$finalize_variable_values() (updates model variables -place). using built-fit() training loop, happens automatically last epoch, need anything. name String. name use momentum accumulator weights created optimizer. ... forward/backward compatability. loss_scale_factor Float NULL. float, scale factor multiplied loss computing gradients, inverse scale factor multiplied gradients updating variables. Useful preventing underflow mixed precision training. Alternately, optimizer_loss_scale() automatically set loss scale factor. gradient_accumulation_steps Int NULL. int, model optimizer variables updated every step; instead updated every gradient_accumulation_steps steps, using average value gradients since last update. known \"gradient accumulation\". can useful batch size small, order reduce gradient noise update step. EMA frequency look \"accumulated\" iterations value (optimizer steps // gradient_accumulation_steps). Learning rate schedules look \"real\" iterations value (optimizer steps).","code":""},{"path":"https://keras3.posit.co/dev/reference/optimizer_adagrad.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Optimizer that implements the Adagrad algorithm. — optimizer_adagrad","text":"Optimizer instance","code":""},{"path":"https://keras3.posit.co/dev/reference/optimizer_adagrad.html","id":"reference","dir":"Reference","previous_headings":"","what":"Reference","title":"Optimizer that implements the Adagrad algorithm. — optimizer_adagrad","text":"Duchi et al., 2011.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/optimizer_adam.html","id":null,"dir":"Reference","previous_headings":"","what":"Optimizer that implements the Adam algorithm. — optimizer_adam","title":"Optimizer that implements the Adam algorithm. — optimizer_adam","text":"Adam optimization stochastic gradient descent method based adaptive estimation first-order second-order moments. According Kingma et al., 2014, method \"computationally efficient, little memory requirement, invariant diagonal rescaling gradients, well suited problems large terms data/parameters\".","code":""},{"path":"https://keras3.posit.co/dev/reference/optimizer_adam.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Optimizer that implements the Adam algorithm. — optimizer_adam","text":"","code":"optimizer_adam(   learning_rate = 0.001,   beta_1 = 0.9,   beta_2 = 0.999,   epsilon = 1e-07,   amsgrad = FALSE,   weight_decay = NULL,   clipnorm = NULL,   clipvalue = NULL,   global_clipnorm = NULL,   use_ema = FALSE,   ema_momentum = 0.99,   ema_overwrite_frequency = NULL,   name = \"adam\",   ...,   loss_scale_factor = NULL,   gradient_accumulation_steps = NULL )"},{"path":"https://keras3.posit.co/dev/reference/optimizer_adam.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Optimizer that implements the Adam algorithm. — optimizer_adam","text":"learning_rate float, LearningRateSchedule() instance, callable takes arguments returns actual value use. learning rate. Defaults 0.001. beta_1 float value constant float tensor, callable takes arguments returns actual value use. exponential decay rate 1st moment estimates. Defaults 0.9. beta_2 float value constant float tensor, callable takes arguments returns actual value use. exponential decay rate 2nd moment estimates. Defaults 0.999. epsilon small constant numerical stability. epsilon \"epsilon hat\" Kingma Ba paper (formula just Section 2.1), epsilon Algorithm 1 paper. Defaults 1e-7. amsgrad Boolean. Whether apply AMSGrad variant algorithm paper \"Convergence Adam beyond\". Defaults FALSE. weight_decay Float. set, weight decay applied. clipnorm Float. set, gradient weight individually clipped norm higher value. clipvalue Float. set, gradient weight clipped higher value. global_clipnorm Float. set, gradient weights clipped global norm higher value. use_ema Boolean, defaults FALSE. TRUE, exponential moving average (EMA) applied. EMA consists computing exponential moving average weights model (weight values change training batch), periodically overwriting weights moving average. ema_momentum Float, defaults 0.99. used use_ema=TRUE. momentum use computing EMA model's weights: new_average = ema_momentum * old_average + (1 - ema_momentum) * current_variable_value. ema_overwrite_frequency Int NULL, defaults NULL. used use_ema=TRUE. Every ema_overwrite_frequency steps iterations, overwrite model variable moving average. NULL, optimizer overwrite model variables middle training, need explicitly overwrite variables end training calling optimizer$finalize_variable_values() (updates model variables -place). using built-fit() training loop, happens automatically last epoch, need anything. name String. name use momentum accumulator weights created optimizer. ... forward/backward compatability. loss_scale_factor Float NULL. float, scale factor multiplied loss computing gradients, inverse scale factor multiplied gradients updating variables. Useful preventing underflow mixed precision training. Alternately, optimizer_loss_scale() automatically set loss scale factor. gradient_accumulation_steps Int NULL. int, model optimizer variables updated every step; instead updated every gradient_accumulation_steps steps, using average value gradients since last update. known \"gradient accumulation\". can useful batch size small, order reduce gradient noise update step. EMA frequency look \"accumulated\" iterations value (optimizer steps // gradient_accumulation_steps). Learning rate schedules look \"real\" iterations value (optimizer steps).","code":""},{"path":"https://keras3.posit.co/dev/reference/optimizer_adam.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Optimizer that implements the Adam algorithm. — optimizer_adam","text":"Optimizer instance","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/optimizer_adam_w.html","id":null,"dir":"Reference","previous_headings":"","what":"Optimizer that implements the AdamW algorithm. — optimizer_adam_w","title":"Optimizer that implements the AdamW algorithm. — optimizer_adam_w","text":"AdamW optimization stochastic gradient descent method based adaptive estimation first-order second-order moments added method decay weights per techniques discussed paper, 'Decoupled Weight Decay Regularization' Loshchilov, Hutter et al., 2019. According Kingma et al., 2014, underlying Adam method \"computationally efficient, little memory requirement, invariant diagonal rescaling gradients, well suited problems large terms data/parameters\".","code":""},{"path":"https://keras3.posit.co/dev/reference/optimizer_adam_w.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Optimizer that implements the AdamW algorithm. — optimizer_adam_w","text":"","code":"optimizer_adam_w(   learning_rate = 0.001,   weight_decay = 0.004,   beta_1 = 0.9,   beta_2 = 0.999,   epsilon = 1e-07,   amsgrad = FALSE,   clipnorm = NULL,   clipvalue = NULL,   global_clipnorm = NULL,   use_ema = FALSE,   ema_momentum = 0.99,   ema_overwrite_frequency = NULL,   name = \"adamw\",   ...,   loss_scale_factor = NULL,   gradient_accumulation_steps = NULL )"},{"path":"https://keras3.posit.co/dev/reference/optimizer_adam_w.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Optimizer that implements the AdamW algorithm. — optimizer_adam_w","text":"learning_rate float, LearningRateSchedule() instance, callable takes arguments returns actual value use. learning rate. Defaults 0.001. weight_decay Float. set, weight decay applied. beta_1 float value constant float tensor, callable takes arguments returns actual value use. exponential decay rate 1st moment estimates. Defaults 0.9. beta_2 float value constant float tensor, callable takes arguments returns actual value use. exponential decay rate 2nd moment estimates. Defaults 0.999. epsilon small constant numerical stability. epsilon \"epsilon hat\" Kingma Ba paper (formula just Section 2.1), epsilon Algorithm 1 paper. Defaults 1e-7. amsgrad Boolean. Whether apply AMSGrad variant algorithm paper \"Convergence Adam beyond\". Defaults FALSE. clipnorm Float. set, gradient weight individually clipped norm higher value. clipvalue Float. set, gradient weight clipped higher value. global_clipnorm Float. set, gradient weights clipped global norm higher value. use_ema Boolean, defaults FALSE. TRUE, exponential moving average (EMA) applied. EMA consists computing exponential moving average weights model (weight values change training batch), periodically overwriting weights moving average. ema_momentum Float, defaults 0.99. used use_ema=TRUE. momentum use computing EMA model's weights: new_average = ema_momentum * old_average + (1 - ema_momentum) * current_variable_value. ema_overwrite_frequency Int NULL, defaults NULL. used use_ema=TRUE. Every ema_overwrite_frequency steps iterations, overwrite model variable moving average. NULL, optimizer overwrite model variables middle training, need explicitly overwrite variables end training calling optimizer$finalize_variable_values() (updates model variables -place). using built-fit() training loop, happens automatically last epoch, need anything. name String. name use momentum accumulator weights created optimizer. ... forward/backward compatability. loss_scale_factor Float NULL. float, scale factor multiplied loss computing gradients, inverse scale factor multiplied gradients updating variables. Useful preventing underflow mixed precision training. Alternately, optimizer_loss_scale() automatically set loss scale factor. gradient_accumulation_steps Int NULL. int, model optimizer variables updated every step; instead updated every gradient_accumulation_steps steps, using average value gradients since last update. known \"gradient accumulation\". can useful batch size small, order reduce gradient noise update step. EMA frequency look \"accumulated\" iterations value (optimizer steps // gradient_accumulation_steps). Learning rate schedules look \"real\" iterations value (optimizer steps).","code":""},{"path":"https://keras3.posit.co/dev/reference/optimizer_adam_w.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Optimizer that implements the AdamW algorithm. — optimizer_adam_w","text":"Optimizer instance","code":""},{"path":"https://keras3.posit.co/dev/reference/optimizer_adam_w.html","id":"references","dir":"Reference","previous_headings":"","what":"References","title":"Optimizer that implements the AdamW algorithm. — optimizer_adam_w","text":"Loshchilov et al., 2019 Kingma et al., 2014 adam Reddi et al., 2018 amsgrad.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/optimizer_adamax.html","id":null,"dir":"Reference","previous_headings":"","what":"Optimizer that implements the Adamax algorithm. — optimizer_adamax","title":"Optimizer that implements the Adamax algorithm. — optimizer_adamax","text":"Adamax, variant Adam based infinity norm, first-order gradient-based optimization method. Due capability adjusting learning rate based data characteristics, suited learn time-variant process, e.g., speech data dynamically changed noise conditions. Default parameters follow provided paper (see references ). Initialization:   update rule parameter w gradient g described end section 7.1 paper (see reference section):","code":"m <- 0  # Initialize initial 1st moment vector u <- 0  # Initialize the exponentially weighted infinity norm t <- 0  # Initialize timestep t <-  t + 1 m <- beta1 * m + (1 - beta) * g u <- max(beta2 * u, abs(g)) current_lr <- learning_rate / (1 - beta1 ** t) w <- w - current_lr * m / (u + epsilon)"},{"path":"https://keras3.posit.co/dev/reference/optimizer_adamax.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Optimizer that implements the Adamax algorithm. — optimizer_adamax","text":"","code":"optimizer_adamax(   learning_rate = 0.001,   beta_1 = 0.9,   beta_2 = 0.999,   epsilon = 1e-07,   weight_decay = NULL,   clipnorm = NULL,   clipvalue = NULL,   global_clipnorm = NULL,   use_ema = FALSE,   ema_momentum = 0.99,   ema_overwrite_frequency = NULL,   name = \"adamax\",   ...,   loss_scale_factor = NULL,   gradient_accumulation_steps = NULL )"},{"path":"https://keras3.posit.co/dev/reference/optimizer_adamax.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Optimizer that implements the Adamax algorithm. — optimizer_adamax","text":"learning_rate float, LearningRateSchedule() instance, callable takes arguments returns actual value use. learning rate. Defaults 0.001. beta_1 float value constant float tensor. exponential decay rate 1st moment estimates. beta_2 float value constant float tensor. exponential decay rate exponentially weighted infinity norm. epsilon small constant numerical stability. name: String. name use momentum accumulator weights created optimizer. weight_decay Float. set, weight decay applied. clipnorm Float. set, gradient weight individually clipped norm higher value. clipvalue Float. set, gradient weight clipped higher value. global_clipnorm Float. set, gradient weights clipped global norm higher value. use_ema Boolean, defaults FALSE. TRUE, exponential moving average (EMA) applied. EMA consists computing exponential moving average weights model (weight values change training batch), periodically overwriting weights moving average. ema_momentum Float, defaults 0.99. used use_ema=TRUE. momentum use computing EMA model's weights: new_average = ema_momentum * old_average + (1 - ema_momentum) * current_variable_value. ema_overwrite_frequency Int NULL, defaults NULL. used use_ema=TRUE. Every ema_overwrite_frequency steps iterations, overwrite model variable moving average. NULL, optimizer overwrite model variables middle training, need explicitly overwrite variables end training calling optimizer$finalize_variable_values() (updates model variables -place). using built-fit() training loop, happens automatically last epoch, need anything. name String, name object ... forward/backward compatability. loss_scale_factor Float NULL. float, scale factor multiplied loss computing gradients, inverse scale factor multiplied gradients updating variables. Useful preventing underflow mixed precision training. Alternately, optimizer_loss_scale() automatically set loss scale factor. gradient_accumulation_steps Int NULL. int, model optimizer variables updated every step; instead updated every gradient_accumulation_steps steps, using average value gradients since last update. known \"gradient accumulation\". can useful batch size small, order reduce gradient noise update step. EMA frequency look \"accumulated\" iterations value (optimizer steps // gradient_accumulation_steps). Learning rate schedules look \"real\" iterations value (optimizer steps).","code":""},{"path":"https://keras3.posit.co/dev/reference/optimizer_adamax.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Optimizer that implements the Adamax algorithm. — optimizer_adamax","text":"Optimizer instance","code":""},{"path":"https://keras3.posit.co/dev/reference/optimizer_adamax.html","id":"reference","dir":"Reference","previous_headings":"","what":"Reference","title":"Optimizer that implements the Adamax algorithm. — optimizer_adamax","text":"Kingma et al., 2014","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/optimizer_ftrl.html","id":null,"dir":"Reference","previous_headings":"","what":"Optimizer that implements the FTRL algorithm. — optimizer_ftrl","title":"Optimizer that implements the FTRL algorithm. — optimizer_ftrl","text":"\"Follow Regularized Leader\" (FTRL) optimization algorithm developed Google click-rate prediction early 2010s. suitable shallow models large sparse feature spaces. algorithm described McMahan et al., 2013. Keras version support online L2 regularization (L2 regularization described paper ) shrinkage-type L2 regularization (addition L2 penalty loss function). Initialization:   Update rule one variable w:   Notation: lr learning rate g gradient variable lambda_1 L1 regularization strength lambda_2 L2 regularization strength lr_power power scale n. Check documentation l2_shrinkage_regularization_strength parameter details shrinkage enabled, case gradient replaced gradient shrinkage.","code":"n <- 0 sigma <- 0 z <- 0 prev_n <- n n <- n + g^2 sigma <- (n^(-lr_power) - prev_n^(-lr_power)) / lr z <- z + g - sigma * w if (abs(z) < lambda_1) {   w <- 0 } else {   w <- (sgn(z) * lambda_1 - z) / ((beta + sqrt(n)) / alpha + lambda_2) }"},{"path":"https://keras3.posit.co/dev/reference/optimizer_ftrl.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Optimizer that implements the FTRL algorithm. — optimizer_ftrl","text":"","code":"optimizer_ftrl(   learning_rate = 0.001,   learning_rate_power = -0.5,   initial_accumulator_value = 0.1,   l1_regularization_strength = 0,   l2_regularization_strength = 0,   l2_shrinkage_regularization_strength = 0,   beta = 0,   weight_decay = NULL,   clipnorm = NULL,   clipvalue = NULL,   global_clipnorm = NULL,   use_ema = FALSE,   ema_momentum = 0.99,   ema_overwrite_frequency = NULL,   name = \"ftrl\",   ...,   loss_scale_factor = NULL,   gradient_accumulation_steps = NULL )"},{"path":"https://keras3.posit.co/dev/reference/optimizer_ftrl.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Optimizer that implements the FTRL algorithm. — optimizer_ftrl","text":"learning_rate float, LearningRateSchedule() instance, callable takes arguments returns actual value use. learning rate. Defaults 0.001. learning_rate_power float value, must less equal zero. Controls learning rate decreases training. Use zero fixed learning rate. initial_accumulator_value starting value accumulators. zero positive values allowed. l1_regularization_strength float value, must greater equal zero. Defaults 0.0. l2_regularization_strength float value, must greater equal zero. Defaults 0.0. l2_shrinkage_regularization_strength float value, must greater equal zero. differs L2 L2 stabilization penalty, whereas L2 shrinkage magnitude penalty. input sparse shrinkage happen active weights. beta float value, representing beta value paper. Defaults 0.0. weight_decay Float. set, weight decay applied. clipnorm Float. set, gradient weight individually clipped norm higher value. clipvalue Float. set, gradient weight clipped higher value. global_clipnorm Float. set, gradient weights clipped global norm higher value. use_ema Boolean, defaults FALSE. TRUE, exponential moving average (EMA) applied. EMA consists computing exponential moving average weights model (weight values change training batch), periodically overwriting weights moving average. ema_momentum Float, defaults 0.99. used use_ema=TRUE. momentum use computing EMA model's weights: new_average = ema_momentum * old_average + (1 - ema_momentum) * current_variable_value. ema_overwrite_frequency Int NULL, defaults NULL. used use_ema=TRUE. Every ema_overwrite_frequency steps iterations, overwrite model variable moving average. NULL, optimizer overwrite model variables middle training, need explicitly overwrite variables end training calling optimizer$finalize_variable_values() (updates model variables -place). using built-fit() training loop, happens automatically last epoch, need anything. name String. name use momentum accumulator weights created optimizer. ... forward/backward compatability. loss_scale_factor Float NULL. float, scale factor multiplied loss computing gradients, inverse scale factor multiplied gradients updating variables. Useful preventing underflow mixed precision training. Alternately, optimizer_loss_scale automatically set loss scale factor. gradient_accumulation_steps Int NULL. int, model optimizer variables updated every step; instead updated every gradient_accumulation_steps steps, using average value gradients since last update. known \"gradient accumulation\". can useful batch size small, order reduce gradient noise update step. EMA frequency look \"accumulated\" iterations value (optimizer steps // gradient_accumulation_steps). Learning rate schedules look \"real\" iterations value (optimizer steps).","code":""},{"path":"https://keras3.posit.co/dev/reference/optimizer_ftrl.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Optimizer that implements the FTRL algorithm. — optimizer_ftrl","text":"Optimizer instance","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/optimizer_lamb.html","id":null,"dir":"Reference","previous_headings":"","what":"Optimizer that implements the Lamb algorithm. — optimizer_lamb","title":"Optimizer that implements the Lamb algorithm. — optimizer_lamb","text":"Lamb stochastic gradient descent method uses layer-wise adaptive moments adjusts learning rate parameter based ratio norm weight norm gradient helps stabilize training process improves convergence especially large batch sizes.","code":""},{"path":"https://keras3.posit.co/dev/reference/optimizer_lamb.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Optimizer that implements the Lamb algorithm. — optimizer_lamb","text":"","code":"optimizer_lamb(   learning_rate = 0.001,   beta_1 = 0.9,   beta_2 = 0.999,   epsilon = 1e-07,   weight_decay = NULL,   clipnorm = NULL,   clipvalue = NULL,   global_clipnorm = NULL,   use_ema = FALSE,   ema_momentum = 0.99,   ema_overwrite_frequency = NULL,   loss_scale_factor = NULL,   gradient_accumulation_steps = NULL,   name = \"lamb\",   ... )"},{"path":"https://keras3.posit.co/dev/reference/optimizer_lamb.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Optimizer that implements the Lamb algorithm. — optimizer_lamb","text":"learning_rate float, LearningRateSchedule() instance, callable takes arguments returns actual value use. learning rate. Defaults 0.001. beta_1 float value constant float tensor, callable takes arguments returns actual value use. exponential decay rate 1st moment estimates. Defaults 0.9. beta_2 float value constant float tensor, callable takes arguments returns actual value use. exponential decay rate 2nd moment estimates. Defaults 0.999. epsilon small constant numerical stability. Defaults 1e-7. weight_decay Float. set, weight decay applied. clipnorm Float. set, gradient weight individually clipped norm higher value. clipvalue Float. set, gradient weight clipped higher value. global_clipnorm Float. set, gradient weights clipped global norm higher value. use_ema Boolean, defaults FALSE. TRUE, exponential moving average (EMA) applied. EMA consists computing exponential moving average weights model (weight values change training batch), periodically overwriting weights moving average. ema_momentum Float, defaults 0.99. used use_ema = TRUE. momentum use computing EMA model's weights: new_average = ema_momentum * old_average + (1 - ema_momentum) * current_variable_value. ema_overwrite_frequency Int NULL, defaults NULL. used use_ema = TRUE. Every ema_overwrite_frequency steps iterations, overwrite model variable moving average. NULL, optimizer overwrite model variables middle training, need explicitly overwrite variables end training calling optimizer$finalize_variable_values() (updates model variables -place). using built-fit() training loop, happens automatically last epoch, need anything. loss_scale_factor Float NULL. float, scale factor multiplied loss computing gradients, inverse scale factor multiplied gradients updating variables. Useful preventing underflow mixed precision training. Alternately, optimizer_loss_scale() automatically set loss scale factor. gradient_accumulation_steps Int NULL. int, model optimizer variables updated every step; instead updated every gradient_accumulation_steps steps, using average value gradients since last update. known \"gradient accumulation\". can useful batch size small, order reduce gradient noise update step. EMA frequency look \"accumulated\" iterations value (optimizer steps // gradient_accumulation_steps). Learning rate schedules look \"real\" iterations value (optimizer steps). name String. name use momentum accumulator weights created optimizer. ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/optimizer_lamb.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Optimizer that implements the Lamb algorithm. — optimizer_lamb","text":"Optimizer instance","code":""},{"path":"https://keras3.posit.co/dev/reference/optimizer_lamb.html","id":"references","dir":"Reference","previous_headings":"","what":"References","title":"Optimizer that implements the Lamb algorithm. — optimizer_lamb","text":"Yang et al.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/optimizer_lion.html","id":null,"dir":"Reference","previous_headings":"","what":"Optimizer that implements the Lion algorithm. — optimizer_lion","title":"Optimizer that implements the Lion algorithm. — optimizer_lion","text":"Lion optimizer stochastic-gradient-descent method uses sign operator control magnitude update, unlike adaptive optimizers Adam rely second-order moments. make Lion memory-efficient keeps track momentum. According authors (see reference), performance gain Adam grows batch size. update Lion produced sign operation, resulting larger norm, suitable learning rate Lion typically 3-10x smaller AdamW. weight decay Lion turn 3-10x larger AdamW maintain similar strength (lr * wd).","code":""},{"path":"https://keras3.posit.co/dev/reference/optimizer_lion.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Optimizer that implements the Lion algorithm. — optimizer_lion","text":"","code":"optimizer_lion(   learning_rate = 0.001,   beta_1 = 0.9,   beta_2 = 0.99,   weight_decay = NULL,   clipnorm = NULL,   clipvalue = NULL,   global_clipnorm = NULL,   use_ema = FALSE,   ema_momentum = 0.99,   ema_overwrite_frequency = NULL,   name = \"lion\",   ...,   loss_scale_factor = NULL,   gradient_accumulation_steps = NULL )"},{"path":"https://keras3.posit.co/dev/reference/optimizer_lion.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Optimizer that implements the Lion algorithm. — optimizer_lion","text":"learning_rate float, LearningRateSchedule() instance, callable takes arguments returns actual value use. learning rate. Defaults 0.001. beta_1 float value constant float tensor, callable takes arguments returns actual value use. rate combine current gradient 1st moment estimate. Defaults 0.9. beta_2 float value constant float tensor, callable takes arguments returns actual value use. exponential decay rate 1st moment estimate. Defaults 0.99. weight_decay Float. set, weight decay applied. clipnorm Float. set, gradient weight individually clipped norm higher value. clipvalue Float. set, gradient weight clipped higher value. global_clipnorm Float. set, gradient weights clipped global norm higher value. use_ema Boolean, defaults FALSE. TRUE, exponential moving average (EMA) applied. EMA consists computing exponential moving average weights model (weight values change training batch), periodically overwriting weights moving average. ema_momentum Float, defaults 0.99. used use_ema=TRUE. momentum use computing EMA model's weights: new_average = ema_momentum * old_average + (1 - ema_momentum) * current_variable_value. ema_overwrite_frequency Int NULL, defaults NULL. used use_ema=TRUE. Every ema_overwrite_frequency steps iterations, overwrite model variable moving average. NULL, optimizer overwrite model variables middle training, need explicitly overwrite variables end training calling optimizer$finalize_variable_values() (updates model variables -place). using built-fit() training loop, happens automatically last epoch, need anything. name String. name use momentum accumulator weights created optimizer. ... forward/backward compatability. loss_scale_factor Float NULL. float, scale factor multiplied loss computing gradients, inverse scale factor multiplied gradients updating variables. Useful preventing underflow mixed precision training. Alternately, optimizer_loss_scale() automatically set loss scale factor. gradient_accumulation_steps Int NULL. int, model optimizer variables updated every step; instead updated every gradient_accumulation_steps steps, using average value gradients since last update. known \"gradient accumulation\". can useful batch size small, order reduce gradient noise update step. EMA frequency look \"accumulated\" iterations value (optimizer steps // gradient_accumulation_steps). Learning rate schedules look \"real\" iterations value (optimizer steps).","code":""},{"path":"https://keras3.posit.co/dev/reference/optimizer_lion.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Optimizer that implements the Lion algorithm. — optimizer_lion","text":"Optimizer instance","code":""},{"path":"https://keras3.posit.co/dev/reference/optimizer_lion.html","id":"references","dir":"Reference","previous_headings":"","what":"References","title":"Optimizer that implements the Lion algorithm. — optimizer_lion","text":"Chen et al., 2023 Authors' implementation","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/optimizer_loss_scale.html","id":null,"dir":"Reference","previous_headings":"","what":"An optimizer that dynamically scales the loss to prevent underflow. — optimizer_loss_scale","title":"An optimizer that dynamically scales the loss to prevent underflow. — optimizer_loss_scale","text":"Loss scaling technique prevent numeric underflow intermediate gradients float16 used. prevent underflow, loss multiplied (\"scaled\") certain factor called \"loss scale\", causes intermediate gradients scaled loss scale well. final gradients divided (\"unscaled\") loss scale bring back original value. LossScaleOptimizer wraps another optimizer applies dynamic loss scaling . loss scale dynamically updated time follows: train step, nonfinite gradient encountered, loss scale halved, train step skipped. dynamic_growth_steps occurred since last time loss scale updated, nonfinite gradients occurred, loss scale doubled.","code":""},{"path":"https://keras3.posit.co/dev/reference/optimizer_loss_scale.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"An optimizer that dynamically scales the loss to prevent underflow. — optimizer_loss_scale","text":"","code":"optimizer_loss_scale(   inner_optimizer,   initial_scale = 32768,   dynamic_growth_steps = 2000L,   ...,   name = NULL,   weight_decay = NULL,   clipnorm = NULL,   clipvalue = NULL,   global_clipnorm = NULL,   use_ema = NULL,   ema_momentum = NULL,   ema_overwrite_frequency = NULL,   loss_scale_factor = NULL,   gradient_accumulation_steps = NULL )"},{"path":"https://keras3.posit.co/dev/reference/optimizer_loss_scale.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"An optimizer that dynamically scales the loss to prevent underflow. — optimizer_loss_scale","text":"inner_optimizer keras Optimizer instance wrap. initial_scale Float. initial loss scale. scale updated training. recommended high number, loss scale high gets lowered far quickly loss scale low gets raised. dynamic_growth_steps Int. often update scale upwards. every dynamic_growth_steps steps finite gradients, loss scale doubled. ... forward/backward compatability. name String. name use momentum accumulator weights created optimizer. weight_decay Float. set, weight decay applied. clipnorm Float. set, gradient weight individually clipped norm higher value. clipvalue Float. set, gradient weight clipped higher value. global_clipnorm Float. set, gradient weights clipped global norm higher value. use_ema Boolean, defaults FALSE. TRUE, exponential moving average (EMA) applied. EMA consists computing exponential moving average weights model (weight values change training batch), periodically overwriting weights moving average. ema_momentum Float, defaults 0.99. used use_ema=TRUE. momentum use computing EMA model's weights: new_average = ema_momentum * old_average + (1 - ema_momentum) * current_variable_value. ema_overwrite_frequency Int NULL, defaults NULL. used use_ema=TRUE. Every ema_overwrite_frequency steps iterations, overwrite model variable moving average. NULL, optimizer overwrite model variables middle training, need explicitly overwrite variables end training calling optimizer$finalize_variable_values() (updates model variables -place). using built-fit() training loop, happens automatically last epoch, need anything. loss_scale_factor Float NULL. float, scale factor multiplied loss computing gradients, inverse scale factor multiplied gradients updating variables. Useful preventing underflow mixed precision training. Alternately, optimizer_loss_scale() automatically set loss scale factor. gradient_accumulation_steps Int NULL. int, model optimizer variables updated every step; instead updated every gradient_accumulation_steps steps, using average value gradients since last update. known \"gradient accumulation\". can useful batch size small, order reduce gradient noise update step. EMA frequency look \"accumulated\" iterations value (optimizer steps // gradient_accumulation_steps). Learning rate schedules look \"real\" iterations value (optimizer steps).","code":""},{"path":"https://keras3.posit.co/dev/reference/optimizer_loss_scale.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"An optimizer that dynamically scales the loss to prevent underflow. — optimizer_loss_scale","text":"Optimizer instance","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/optimizer_nadam.html","id":null,"dir":"Reference","previous_headings":"","what":"Optimizer that implements the Nadam algorithm. — optimizer_nadam","title":"Optimizer that implements the Nadam algorithm. — optimizer_nadam","text":"Much like Adam essentially RMSprop momentum, Nadam Adam Nesterov momentum.","code":""},{"path":"https://keras3.posit.co/dev/reference/optimizer_nadam.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Optimizer that implements the Nadam algorithm. — optimizer_nadam","text":"","code":"optimizer_nadam(   learning_rate = 0.001,   beta_1 = 0.9,   beta_2 = 0.999,   epsilon = 1e-07,   weight_decay = NULL,   clipnorm = NULL,   clipvalue = NULL,   global_clipnorm = NULL,   use_ema = FALSE,   ema_momentum = 0.99,   ema_overwrite_frequency = NULL,   name = \"nadam\",   ...,   loss_scale_factor = NULL,   gradient_accumulation_steps = NULL )"},{"path":"https://keras3.posit.co/dev/reference/optimizer_nadam.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Optimizer that implements the Nadam algorithm. — optimizer_nadam","text":"learning_rate float, LearningRateSchedule() instance, callable takes arguments returns actual value use. learning rate. Defaults 0.001. beta_1 float value constant float tensor, callable takes arguments returns actual value use. exponential decay rate 1st moment estimates. Defaults 0.9. beta_2 float value constant float tensor, callable takes arguments returns actual value use. exponential decay rate 2nd moment estimates. Defaults 0.999. epsilon small constant numerical stability. epsilon \"epsilon hat\" Kingma Ba paper (formula just Section 2.1), epsilon Algorithm 1 paper. Defaults 1e-7. weight_decay Float. set, weight decay applied. clipnorm Float. set, gradient weight individually clipped norm higher value. clipvalue Float. set, gradient weight clipped higher value. global_clipnorm Float. set, gradient weights clipped global norm higher value. use_ema Boolean, defaults FALSE. TRUE, exponential moving average (EMA) applied. EMA consists computing exponential moving average weights model (weight values change training batch), periodically overwriting weights moving average. ema_momentum Float, defaults 0.99. used use_ema=TRUE. momentum use computing EMA model's weights: new_average = ema_momentum * old_average + (1 - ema_momentum) * current_variable_value. ema_overwrite_frequency Int NULL, defaults NULL. used use_ema=TRUE. Every ema_overwrite_frequency steps iterations, overwrite model variable moving average. NULL, optimizer overwrite model variables middle training, need explicitly overwrite variables end training calling optimizer$finalize_variable_values() (updates model variables -place). using built-fit() training loop, happens automatically last epoch, need anything. name String. name use momentum accumulator weights created optimizer. ... forward/backward compatability. loss_scale_factor Float NULL. float, scale factor multiplied loss computing gradients, inverse scale factor multiplied gradients updating variables. Useful preventing underflow mixed precision training. Alternately, optimizer_loss_scale() automatically set loss scale factor. gradient_accumulation_steps Int NULL. int, model optimizer variables updated every step; instead updated every gradient_accumulation_steps steps, using average value gradients since last update. known \"gradient accumulation\". can useful batch size small, order reduce gradient noise update step. EMA frequency look \"accumulated\" iterations value (optimizer steps // gradient_accumulation_steps). Learning rate schedules look \"real\" iterations value (optimizer steps).","code":""},{"path":"https://keras3.posit.co/dev/reference/optimizer_nadam.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Optimizer that implements the Nadam algorithm. — optimizer_nadam","text":"Optimizer instance","code":""},{"path":"https://keras3.posit.co/dev/reference/optimizer_nadam.html","id":"reference","dir":"Reference","previous_headings":"","what":"Reference","title":"Optimizer that implements the Nadam algorithm. — optimizer_nadam","text":"Dozat, 2015.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/optimizer_rmsprop.html","id":null,"dir":"Reference","previous_headings":"","what":"Optimizer that implements the RMSprop algorithm. — optimizer_rmsprop","title":"Optimizer that implements the RMSprop algorithm. — optimizer_rmsprop","text":"gist RMSprop : Maintain moving (discounted) average square gradients Divide gradient root average implementation RMSprop uses plain momentum, Nesterov momentum. centered version additionally maintains moving average gradients, uses average estimate variance.","code":""},{"path":"https://keras3.posit.co/dev/reference/optimizer_rmsprop.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Optimizer that implements the RMSprop algorithm. — optimizer_rmsprop","text":"","code":"optimizer_rmsprop(   learning_rate = 0.001,   rho = 0.9,   momentum = 0,   epsilon = 1e-07,   centered = FALSE,   weight_decay = NULL,   clipnorm = NULL,   clipvalue = NULL,   global_clipnorm = NULL,   use_ema = FALSE,   ema_momentum = 0.99,   ema_overwrite_frequency = NULL,   name = \"rmsprop\",   ...,   loss_scale_factor = NULL,   gradient_accumulation_steps = NULL )"},{"path":"https://keras3.posit.co/dev/reference/optimizer_rmsprop.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Optimizer that implements the RMSprop algorithm. — optimizer_rmsprop","text":"learning_rate float, learning_rate_schedule_* instance, callable takes arguments returns actual value use. learning rate. Defaults 0.001. rho float, defaults 0.9. Discounting factor old gradients. momentum float, defaults 0.0. 0.0., optimizer tracks momentum value, decay rate equals 1 - momentum. epsilon small constant numerical stability. epsilon \"epsilon hat\" Kingma Ba paper (formula just Section 2.1), epsilon Algorithm 1 paper. Defaults 1e-7. centered Boolean. TRUE, gradients normalized estimated variance gradient; FALSE, uncentered second moment. Setting TRUE may help training, slightly expensive terms computation memory. Defaults FALSE. weight_decay Float. set, weight decay applied. clipnorm Float. set, gradient weight individually clipped norm higher value. clipvalue Float. set, gradient weight clipped higher value. global_clipnorm Float. set, gradient weights clipped global norm higher value. use_ema Boolean, defaults FALSE. TRUE, exponential moving average (EMA) applied. EMA consists computing exponential moving average weights model (weight values change training batch), periodically overwriting weights moving average. ema_momentum Float, defaults 0.99. used use_ema=TRUE. momentum use computing EMA model's weights: new_average = ema_momentum * old_average + (1 - ema_momentum) * current_variable_value. ema_overwrite_frequency Int NULL, defaults NULL. used use_ema=TRUE. Every ema_overwrite_frequency steps iterations, overwrite model variable moving average. NULL, optimizer overwrite model variables middle training, need explicitly overwrite variables end training calling optimizer$finalize_variable_values() (updates model variables -place). using built-fit() training loop, happens automatically last epoch, need anything. name String. name use momentum accumulator weights created optimizer. ... forward/backward compatability. loss_scale_factor Float NULL. float, scale factor multiplied loss computing gradients, inverse scale factor multiplied gradients updating variables. Useful preventing underflow mixed precision training. Alternately, optimizer_loss_scale() automatically set loss scale factor. gradient_accumulation_steps Int NULL. int, model optimizer variables updated every step; instead updated every gradient_accumulation_steps steps, using average value gradients since last update. known \"gradient accumulation\". can useful batch size small, order reduce gradient noise update step. EMA frequency look \"accumulated\" iterations value (optimizer steps // gradient_accumulation_steps). Learning rate schedules look \"real\" iterations value (optimizer steps).","code":""},{"path":"https://keras3.posit.co/dev/reference/optimizer_rmsprop.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Optimizer that implements the RMSprop algorithm. — optimizer_rmsprop","text":"Optimizer instance","code":""},{"path":"https://keras3.posit.co/dev/reference/optimizer_rmsprop.html","id":"usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Optimizer that implements the RMSprop algorithm. — optimizer_rmsprop","text":"","code":"opt <- optimizer_rmsprop(learning_rate=0.1)"},{"path":"https://keras3.posit.co/dev/reference/optimizer_rmsprop.html","id":"reference","dir":"Reference","previous_headings":"","what":"Reference","title":"Optimizer that implements the RMSprop algorithm. — optimizer_rmsprop","text":"Hinton, 2012","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/optimizer_sgd.html","id":null,"dir":"Reference","previous_headings":"","what":"Gradient descent (with momentum) optimizer. — optimizer_sgd","title":"Gradient descent (with momentum) optimizer. — optimizer_sgd","text":"Update rule parameter w gradient g momentum 0:   Update rule momentum larger 0:   nesterov=TRUE, rule becomes:","code":"w <- w - learning_rate * g velocity <- momentum * velocity - learning_rate * g w <- w + velocity velocity <- momentum * velocity - learning_rate * g w <- w + momentum * velocity - learning_rate * g"},{"path":"https://keras3.posit.co/dev/reference/optimizer_sgd.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Gradient descent (with momentum) optimizer. — optimizer_sgd","text":"","code":"optimizer_sgd(   learning_rate = 0.01,   momentum = 0,   nesterov = FALSE,   weight_decay = NULL,   clipnorm = NULL,   clipvalue = NULL,   global_clipnorm = NULL,   use_ema = FALSE,   ema_momentum = 0.99,   ema_overwrite_frequency = NULL,   name = \"SGD\",   ...,   loss_scale_factor = NULL,   gradient_accumulation_steps = NULL )"},{"path":"https://keras3.posit.co/dev/reference/optimizer_sgd.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Gradient descent (with momentum) optimizer. — optimizer_sgd","text":"learning_rate float, learning_rate_schedule_* instance, callable takes arguments returns actual value use. learning rate. Defaults 0.01. momentum float hyperparameter >= 0 accelerates gradient descent relevant direction dampens oscillations. 0 vanilla gradient descent. Defaults 0.0. nesterov boolean. Whether apply Nesterov momentum. Defaults FALSE. weight_decay Float. set, weight decay applied. clipnorm Float. set, gradient weight individually clipped norm higher value. clipvalue Float. set, gradient weight clipped higher value. global_clipnorm Float. set, gradient weights clipped global norm higher value. use_ema Boolean, defaults FALSE. TRUE, exponential moving average (EMA) applied. EMA consists computing exponential moving average weights model (weight values change training batch), periodically overwriting weights moving average. ema_momentum Float, defaults 0.99. used use_ema=TRUE. momentum use computing EMA model's weights: new_average = ema_momentum * old_average + (1 - ema_momentum) * current_variable_value. ema_overwrite_frequency Int NULL, defaults NULL. used use_ema=TRUE. Every ema_overwrite_frequency steps iterations, overwrite model variable moving average. NULL, optimizer overwrite model variables middle training, need explicitly overwrite variables end training calling optimizer$finalize_variable_values() (updates model variables -place). using built-fit() training loop, happens automatically last epoch, need anything. name String. name use momentum accumulator weights created optimizer. ... forward/backward compatability. loss_scale_factor Float NULL. float, scale factor multiplied loss computing gradients, inverse scale factor multiplied gradients updating variables. Useful preventing underflow mixed precision training. Alternately, optimizer_loss_scale() automatically set loss scale factor. gradient_accumulation_steps Int NULL. int, model optimizer variables updated every step; instead updated every gradient_accumulation_steps steps, using average value gradients since last update. known \"gradient accumulation\". can useful batch size small, order reduce gradient noise update step. EMA frequency look \"accumulated\" iterations value (optimizer steps // gradient_accumulation_steps). Learning rate schedules look \"real\" iterations value (optimizer steps).","code":""},{"path":"https://keras3.posit.co/dev/reference/optimizer_sgd.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Gradient descent (with momentum) optimizer. — optimizer_sgd","text":"Optimizer instance","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/pad_sequences.html","id":null,"dir":"Reference","previous_headings":"","what":"Pads sequences to the same length. — pad_sequences","title":"Pads sequences to the same length. — pad_sequences","text":"function transforms list (length num_samples) sequences (lists integers) 2D NumPy array shape (num_samples, num_timesteps). num_timesteps either maxlen argument provided, length longest sequence list. Sequences shorter num_timesteps padded value num_timesteps long. Sequences longer num_timesteps truncated fit desired length. position padding truncation happens determined arguments padding truncating, respectively. Pre-padding removing values beginning sequence default.","code":"sequence <- list(c(1), c(2, 3), c(4, 5, 6)) pad_sequences(sequence) ##      [,1] [,2] [,3] ## [1,]    0    0    1 ## [2,]    0    2    3 ## [3,]    4    5    6 pad_sequences(sequence, value=-1) ##      [,1] [,2] [,3] ## [1,]   -1   -1    1 ## [2,]   -1    2    3 ## [3,]    4    5    6 pad_sequences(sequence, padding='post') ##      [,1] [,2] [,3] ## [1,]    1    0    0 ## [2,]    2    3    0 ## [3,]    4    5    6 pad_sequences(sequence, maxlen=2) ##      [,1] [,2] ## [1,]    0    1 ## [2,]    2    3 ## [3,]    5    6"},{"path":"https://keras3.posit.co/dev/reference/pad_sequences.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Pads sequences to the same length. — pad_sequences","text":"","code":"pad_sequences(   sequences,   maxlen = NULL,   dtype = \"int32\",   padding = \"pre\",   truncating = \"pre\",   value = 0 )"},{"path":"https://keras3.posit.co/dev/reference/pad_sequences.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Pads sequences to the same length. — pad_sequences","text":"sequences List sequences (sequence list integers). maxlen Optional Int, maximum length sequences. provided, sequences padded length longest individual sequence. dtype (Optional, defaults \"int32\"). Type output sequences. pad sequences variable length strings, can use object. padding String, \"pre\" \"post\" (optional, defaults \"pre\"): pad either sequence. truncating String, \"pre\" \"post\" (optional, defaults \"pre\"): remove values sequences larger maxlen, either beginning end sequences. value Float String, padding value. (Optional, defaults 0)","code":""},{"path":"https://keras3.posit.co/dev/reference/pad_sequences.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Pads sequences to the same length. — pad_sequences","text":"Array shape (len(sequences), maxlen)","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/pipe.html","id":null,"dir":"Reference","previous_headings":"","what":"Pipe operator — %>%","title":"Pipe operator — %>%","text":"See %>% details.","code":""},{"path":"https://keras3.posit.co/dev/reference/pipe.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Pipe operator — %>%","text":"","code":"lhs %>% rhs"},{"path":"https://keras3.posit.co/dev/reference/pipe.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Pipe operator — %>%","text":"commonly, result calling right hand side left hand side argument: rhs(lhs). See magritter vignette , advanced, usages.","code":""},{"path":"https://keras3.posit.co/dev/reference/plot.keras.src.models.model.Model.html","id":null,"dir":"Reference","previous_headings":"","what":"Plot a Keras model — plot.keras.src.models.model.Model","title":"Plot a Keras model — plot.keras.src.models.model.Model","text":"Plot Keras model","code":""},{"path":"https://keras3.posit.co/dev/reference/plot.keras.src.models.model.Model.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Plot a Keras model — plot.keras.src.models.model.Model","text":"","code":"# S3 method for class 'keras.src.models.model.Model' plot(   x,   show_shapes = FALSE,   show_dtype = FALSE,   show_layer_names = FALSE,   ...,   rankdir = \"TB\",   expand_nested = FALSE,   dpi = 200,   layer_range = NULL,   show_layer_activations = FALSE,   show_trainable = NA,   to_file = NULL )"},{"path":"https://keras3.posit.co/dev/reference/plot.keras.src.models.model.Model.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Plot a Keras model — plot.keras.src.models.model.Model","text":"x Keras model instance show_shapes whether display shape information. show_dtype whether display layer dtypes. show_layer_names whether display layer names. ... passed Python keras.utils.model_to_dot(). Used forward backward compatibility. rankdir string specifying format plot: 'TB' creates vertical plot; 'LR' creates horizontal plot. (argument passed PyDot) expand_nested Whether expand nested models clusters. dpi Dots per inch. Increase value image text appears excessively pixelated. layer_range list containing two character strings, starting layer name ending layer name (inclusive) indicating range layers plot generated. also accepts regex patterns instead exact name. case, start predicate first element matches layer_range[1] end predicate last element matches layer_range[2]. default NULL considers layers model. Note must pass range resultant subgraph must complete. show_layer_activations Display layer activations (layers activation property). show_trainable whether display layer trainable. to_file File name plot image. NULL (default), model drawn default graphics device. Otherwise, file saved.","code":""},{"path":"https://keras3.posit.co/dev/reference/plot.keras.src.models.model.Model.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Plot a Keras model — plot.keras.src.models.model.Model","text":"Nothing, called side effects.","code":""},{"path":"https://keras3.posit.co/dev/reference/plot.keras.src.models.model.Model.html","id":"raises","dir":"Reference","previous_headings":"","what":"Raises","title":"Plot a Keras model — plot.keras.src.models.model.Model","text":"ValueError: plot(model) called model built, unless input_shape =  argument supplied keras_model_sequential().","code":""},{"path":"https://keras3.posit.co/dev/reference/plot.keras.src.models.model.Model.html","id":"requirements","dir":"Reference","previous_headings":"","what":"Requirements","title":"Plot a Keras model — plot.keras.src.models.model.Model","text":"function requires pydot graphviz. pydot default installed install_keras(), installed keras means, can install pydot directly :   can install graphviz directly : https://graphviz.gitlab.io/download/ Linux platforms, can install graphviz via package manager. example, Ubuntu/Debian can install   conda environment, can install graphviz :","code":"reticulate::py_install(\"pydot\", pip = TRUE) sudo apt install graphviz reticulate::conda_install(packages = \"graphviz\") # Restart the R session after install."},{"path":"https://keras3.posit.co/dev/reference/plot.keras_training_history.html","id":null,"dir":"Reference","previous_headings":"","what":"Plot training history — plot.keras_training_history","title":"Plot training history — plot.keras_training_history","text":"Plots metrics recorded training.","code":""},{"path":"https://keras3.posit.co/dev/reference/plot.keras_training_history.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Plot training history — plot.keras_training_history","text":"","code":"# S3 method for class 'keras_training_history' plot(   x,   y,   metrics = NULL,   method = c(\"auto\", \"ggplot2\", \"base\"),   smooth = getOption(\"keras.plot.history.smooth\", TRUE),   theme_bw = getOption(\"keras.plot.history.theme_bw\", FALSE),   ... )"},{"path":"https://keras3.posit.co/dev/reference/plot.keras_training_history.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Plot training history — plot.keras_training_history","text":"x Training history object returned fit.keras.src.models.model.Model(). y Unused. metrics One metrics plot (e.g. c('loss', 'accuracy')). Defaults plotting captured metrics. method Method use plotting. default \"auto\" use ggplot2 available, otherwise use base graphics. smooth Whether loess smooth added plot, available ggplot2 method. number epochs smaller ten, forced false. theme_bw Use ggplot2::theme_bw() plot history black white. ... Additional parameters pass plot() method.","code":""},{"path":"https://keras3.posit.co/dev/reference/plot.keras_training_history.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Plot training history — plot.keras_training_history","text":"method == \"ggplot2\", ggplot object returned. method == \"base\", function draw graphics device return NULL, invisibly.","code":""},{"path":"https://keras3.posit.co/dev/reference/pop_layer.html","id":null,"dir":"Reference","previous_headings":"","what":"Remove the last layer in a Sequential model — pop_layer","title":"Remove the last layer in a Sequential model — pop_layer","text":"Remove last layer Sequential model","code":""},{"path":"https://keras3.posit.co/dev/reference/pop_layer.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Remove the last layer in a Sequential model — pop_layer","text":"","code":"pop_layer(object, rebuild = TRUE)"},{"path":"https://keras3.posit.co/dev/reference/pop_layer.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Remove the last layer in a Sequential model — pop_layer","text":"object Sequential keras model object rebuild bool. Whether rebuild model removing layer. Defaults TRUE.","code":""},{"path":"https://keras3.posit.co/dev/reference/pop_layer.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Remove the last layer in a Sequential model — pop_layer","text":"removed layer.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/predict.keras.src.models.model.Model.html","id":null,"dir":"Reference","previous_headings":"","what":"Generates output predictions for the input samples. — predict.keras.src.models.model.Model","title":"Generates output predictions for the input samples. — predict.keras.src.models.model.Model","text":"Generates output predictions input samples.","code":""},{"path":"https://keras3.posit.co/dev/reference/predict.keras.src.models.model.Model.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Generates output predictions for the input samples. — predict.keras.src.models.model.Model","text":"","code":"# S3 method for class 'keras.src.models.model.Model' predict(   object,   x,   ...,   batch_size = NULL,   verbose = getOption(\"keras.verbose\", default = \"auto\"),   steps = NULL,   callbacks = NULL )"},{"path":"https://keras3.posit.co/dev/reference/predict.keras.src.models.model.Model.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Generates output predictions for the input samples. — predict.keras.src.models.model.Model","text":"object Keras model object x Input samples. can : array (array-like), list arrays (case model multiple inputs). backend-native tensor, list tensors (case model multiple inputs). TF Dataset. Python generator function. ... forward/backward compatability. batch_size Integer NULL. Number samples per batch computation. unspecified, batch_size default 32. specify batch_size input data x form TF Dataset generator (since generate batches). verbose \"auto\", 0, 1, 2. Verbosity mode. 0 = silent, 1 = progress bar, 2 = one line per epoch. \"auto\" becomes 1 cases, 2 knitr render running distributed training server. Note progress bar particularly useful logged file, verbose=2 recommended running interactively (e.g., production environment). Defaults \"auto\". steps Total number steps (batches samples) draw declaring prediction round finished. steps NULL, predict() run x exhausted. case infinitely repeating dataset, predict() run indefinitely. callbacks List Callback instances. List callbacks apply prediction.","code":""},{"path":"https://keras3.posit.co/dev/reference/predict.keras.src.models.model.Model.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Generates output predictions for the input samples. — predict.keras.src.models.model.Model","text":"R array(s) predictions.","code":""},{"path":"https://keras3.posit.co/dev/reference/predict.keras.src.models.model.Model.html","id":"details","dir":"Reference","previous_headings":"","what":"Details","title":"Generates output predictions for the input samples. — predict.keras.src.models.model.Model","text":"Computation done batches. method designed batch processing large numbers inputs. intended use inside loops iterate data process small numbers inputs time. small numbers inputs fit one batch, directly call model model$call faster execution, e.g., model(x), model(x, training = FALSE) layers BatchNormalization behave differently inference.","code":""},{"path":"https://keras3.posit.co/dev/reference/predict.keras.src.models.model.Model.html","id":"note","dir":"Reference","previous_headings":"","what":"Note","title":"Generates output predictions for the input samples. — predict.keras.src.models.model.Model","text":"See FAQ entry details difference Model methods predict() call().","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/predict_on_batch.html","id":null,"dir":"Reference","previous_headings":"","what":"Returns predictions for a single batch of samples. — predict_on_batch","title":"Returns predictions for a single batch of samples. — predict_on_batch","text":"Returns predictions single batch samples.","code":""},{"path":"https://keras3.posit.co/dev/reference/predict_on_batch.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Returns predictions for a single batch of samples. — predict_on_batch","text":"","code":"predict_on_batch(object, x)"},{"path":"https://keras3.posit.co/dev/reference/predict_on_batch.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Returns predictions for a single batch of samples. — predict_on_batch","text":"object Keras model object x Input data. must array-like.","code":""},{"path":"https://keras3.posit.co/dev/reference/predict_on_batch.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Returns predictions for a single batch of samples. — predict_on_batch","text":"Array(s) predictions.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/process_utils.html","id":null,"dir":"Reference","previous_headings":"","what":"Preprocessing and postprocessing utilities — process_utils","title":"Preprocessing and postprocessing utilities — process_utils","text":"functions used preprocess postprocess inputs outputs Keras applications.","code":""},{"path":"https://keras3.posit.co/dev/reference/process_utils.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Preprocessing and postprocessing utilities — process_utils","text":"","code":"application_preprocess_inputs(model, x, ..., data_format = NULL)  application_decode_predictions(model, preds, top = 5L, ...)"},{"path":"https://keras3.posit.co/dev/reference/process_utils.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Preprocessing and postprocessing utilities — process_utils","text":"model Keras model initialized using application_ function. x batch inputs model. x missing, preprocess_input function appropriate model returned. ... Additional arguments passed preprocessing decoding function. data_format Optional data format image tensor/array. NULL means global setting config_image_data_format() used (unless changed , uses \"channels_last\"). Defaults NULL. preds batch outputs model. top number top predictions return.","code":""},{"path":"https://keras3.posit.co/dev/reference/process_utils.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Preprocessing and postprocessing utilities — process_utils","text":"list decoded predictions case application_decode_predictions(). batch preprocessed inputs case application_preprocess_inputs().","code":""},{"path":"https://keras3.posit.co/dev/reference/process_utils.html","id":"functions","dir":"Reference","previous_headings":"","what":"Functions","title":"Preprocessing and postprocessing utilities — process_utils","text":"application_preprocess_inputs(): Pre-process inputs used model application_decode_predictions(): Decode predictions model","code":""},{"path":"https://keras3.posit.co/dev/reference/process_utils.html","id":"ref-examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Preprocessing and postprocessing utilities — process_utils","text":"","code":"if (FALSE) { # \\dontrun{ model <- application_convnext_tiny()  inputs <- random_normal(c(32, 224, 224, 3)) processed_inputs <- application_preprocess_inputs(model, inputs)  preds <- random_normal(c(32, 1000)) decoded_preds <- application_decode_predictions(model, preds)  } # }"},{"path":"https://keras3.posit.co/dev/reference/quantize_weights.html","id":null,"dir":"Reference","previous_headings":"","what":"Quantize the weights of a model. — quantize_weights","title":"Quantize the weights of a model. — quantize_weights","text":"Note model must built first calling method. quantize_weights() recursively call layer$quantize(mode) layers skipped layer implement function. Currently Dense EinsumDense layers support quantization.","code":""},{"path":"https://keras3.posit.co/dev/reference/quantize_weights.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Quantize the weights of a model. — quantize_weights","text":"","code":"quantize_weights(object, mode, ...)"},{"path":"https://keras3.posit.co/dev/reference/quantize_weights.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Quantize the weights of a model. — quantize_weights","text":"object Keras Model Layer. mode mode quantization. 'int8' supported time. ... Passed object quantization method.","code":""},{"path":"https://keras3.posit.co/dev/reference/quantize_weights.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Quantize the weights of a model. — quantize_weights","text":"model, invisibly. Note just convenience usage |>, model modified -place.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/random_beta.html","id":null,"dir":"Reference","previous_headings":"","what":"Draw samples from a Beta distribution. — random_beta","title":"Draw samples from a Beta distribution. — random_beta","text":"values drawn Beta distribution parametrized alpha beta.","code":""},{"path":"https://keras3.posit.co/dev/reference/random_beta.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Draw samples from a Beta distribution. — random_beta","text":"","code":"random_beta(shape, alpha, beta, dtype = NULL, seed = NULL)"},{"path":"https://keras3.posit.co/dev/reference/random_beta.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Draw samples from a Beta distribution. — random_beta","text":"shape shape random values generate. alpha Float array floats representing first parameter alpha. Must broadcastable beta shape. beta Float array floats representing second parameter beta. Must broadcastable alpha shape. dtype Optional dtype tensor. floating point types supported. specified, config_floatx() used, defaults \"float32\" unless configured otherwise (via config_set_floatx(float_dtype)). seed Optional R integer instance random_seed_generator(). default, seed argument NULL, internal global random_seed_generator() used. seed argument can used ensure deterministic (repeatable) random number generation. Note passing integer seed value produce random values call. generate different random values repeated calls, instance random_seed_generator() must provided seed value. Remark concerning JAX backend: tracing functions JAX backend global random_seed_generator() supported. Therefore, tracing default value seed=NULL produce error, seed argument must provided.","code":""},{"path":"https://keras3.posit.co/dev/reference/random_beta.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Draw samples from a Beta distribution. — random_beta","text":"tensor random values.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/random_binomial.html","id":null,"dir":"Reference","previous_headings":"","what":"Draw samples from a Binomial distribution. — random_binomial","title":"Draw samples from a Binomial distribution. — random_binomial","text":"values drawn Binomial distribution specified trial count probability success.","code":""},{"path":"https://keras3.posit.co/dev/reference/random_binomial.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Draw samples from a Binomial distribution. — random_binomial","text":"","code":"random_binomial(shape, counts, probabilities, dtype = NULL, seed = NULL)"},{"path":"https://keras3.posit.co/dev/reference/random_binomial.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Draw samples from a Binomial distribution. — random_binomial","text":"shape shape random values generate. counts number array numbers representing number trials. must broadcastable probabilities. probabilities float array floats representing probability success individual event. must broadcastable counts. dtype Optional dtype tensor. floating point types supported. specified, config_floatx() used, defaults \"float32\" unless configured otherwise (via config_set_floatx(float_dtype)). seed Optional R integer instance random_seed_generator(). default, seed argument NULL, internal global random_seed_generator() used. seed argument can used ensure deterministic (repeatable) random number generation. Note passing integer seed value produce random values call. generate different random values repeated calls, instance random_seed_generator() must provided seed value. Remark concerning JAX backend: tracing functions JAX backend global random_seed_generator() supported. Therefore, tracing default value seed=NULL produce error, seed argument must provided.","code":""},{"path":"https://keras3.posit.co/dev/reference/random_binomial.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Draw samples from a Binomial distribution. — random_binomial","text":"tensor random values.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/random_categorical.html","id":null,"dir":"Reference","previous_headings":"","what":"Draws samples from a categorical distribution. — random_categorical","title":"Draws samples from a categorical distribution. — random_categorical","text":"function takes input logits, 2-D input tensor shape (batch_size, num_classes). row input represents categorical distribution, column index containing log-probability given class. function output 2-D tensor shape (batch_size, num_samples), row contains samples corresponding row logits. column index contains independent samples drawn input distribution.","code":"x <- matrix(c(100, .1, 99), nrow = 1) random_categorical(x, num_samples = 5, seed = 1234) ## tf.Tensor([[3 1 1 3 3]], shape=(1, 5), dtype=int32) random_categorical(x, num_samples = 5, seed = 1234,                    zero_indexed = TRUE) ## tf.Tensor([[2 0 0 2 2]], shape=(1, 5), dtype=int32) op_take(x, random_categorical(x, num_samples = 5, seed = 1234)) ## tf.Tensor([[ 99. 100. 100.  99.  99.]], shape=(1, 5), dtype=float64) op_take(x, random_categorical(x, num_samples = 5, seed = 1234,                               zero_indexed = TRUE),         zero_indexed = TRUE) ## tf.Tensor([[ 99. 100. 100.  99.  99.]], shape=(1, 5), dtype=float64)"},{"path":"https://keras3.posit.co/dev/reference/random_categorical.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Draws samples from a categorical distribution. — random_categorical","text":"","code":"random_categorical(   logits,   num_samples,   dtype = \"int32\",   seed = NULL,   zero_indexed = FALSE )"},{"path":"https://keras3.posit.co/dev/reference/random_categorical.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Draws samples from a categorical distribution. — random_categorical","text":"logits 2-D Tensor shape (batch_size, num_classes). row define categorical distribution unnormalized log-probabilities classes. num_samples Int, number independent samples draw row input. second dimension output tensor's shape. dtype Optional dtype output tensor. seed Optional R integer instance random_seed_generator(). default, seed argument NULL, internal global random_seed_generator() used. seed argument can used ensure deterministic (repeatable) random number generation. Note passing integer seed value produce random values call. generate different random values repeated calls, instance random_seed_generator() must provided seed value. Remark concerning JAX backend: tracing functions JAX backend global random_seed_generator() supported. Therefore, tracing default value seed=NULL produce error, seed argument must provided. zero_indexed TRUE, returned indices zero-based (0 encodes first position); FALSE (default), returned indices one-based (1 encodes first position).","code":""},{"path":"https://keras3.posit.co/dev/reference/random_categorical.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Draws samples from a categorical distribution. — random_categorical","text":"2-D tensor (batch_size, num_samples).","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/random_dropout.html","id":null,"dir":"Reference","previous_headings":"","what":"Randomly set some values in a tensor to 0. — random_dropout","title":"Randomly set some values in a tensor to 0. — random_dropout","text":"Randomly set portion values tensor 0.","code":""},{"path":"https://keras3.posit.co/dev/reference/random_dropout.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Randomly set some values in a tensor to 0. — random_dropout","text":"","code":"random_dropout(inputs, rate, noise_shape = NULL, seed = NULL)"},{"path":"https://keras3.posit.co/dev/reference/random_dropout.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Randomly set some values in a tensor to 0. — random_dropout","text":"inputs tensor rate numeric noise_shape shape() value seed Initial seed random number generator","code":""},{"path":"https://keras3.posit.co/dev/reference/random_dropout.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Randomly set some values in a tensor to 0. — random_dropout","text":"tensor copy inputs values set 0.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/random_gamma.html","id":null,"dir":"Reference","previous_headings":"","what":"Draw random samples from the Gamma distribution. — random_gamma","title":"Draw random samples from the Gamma distribution. — random_gamma","text":"Draw random samples Gamma distribution.","code":""},{"path":"https://keras3.posit.co/dev/reference/random_gamma.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Draw random samples from the Gamma distribution. — random_gamma","text":"","code":"random_gamma(shape, alpha, dtype = NULL, seed = NULL)"},{"path":"https://keras3.posit.co/dev/reference/random_gamma.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Draw random samples from the Gamma distribution. — random_gamma","text":"shape shape random values generate. alpha Float, parameter distribution. dtype Optional dtype tensor. floating point types supported. specified, config_floatx() used, defaults float32 unless configured otherwise (via config_set_floatx(float_dtype)). seed Optional R integer instance random_seed_generator(). default, seed argument NULL, internal global random_seed_generator() used. seed argument can used ensure deterministic (repeatable) random number generation. Note passing integer seed value produce random values call. generate different random values repeated calls, instance random_seed_generator() must provided seed value. Remark concerning JAX backend: tracing functions JAX backend global random_seed_generator() supported. Therefore, tracing default value seed=NULL produce error, seed argument must provided.","code":""},{"path":"https://keras3.posit.co/dev/reference/random_gamma.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Draw random samples from the Gamma distribution. — random_gamma","text":"tensor random values.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/random_integer.html","id":null,"dir":"Reference","previous_headings":"","what":"Draw random integers from a uniform distribution. — random_integer","title":"Draw random integers from a uniform distribution. — random_integer","text":"generated values follow uniform distribution range [minval, maxval). lower bound minval included range, upper bound maxval excluded. dtype must integer type.","code":""},{"path":"https://keras3.posit.co/dev/reference/random_integer.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Draw random integers from a uniform distribution. — random_integer","text":"","code":"random_integer(shape, minval, maxval, dtype = \"int32\", seed = NULL)"},{"path":"https://keras3.posit.co/dev/reference/random_integer.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Draw random integers from a uniform distribution. — random_integer","text":"shape shape random values generate. minval integer, lower bound range random values generate (inclusive). maxval integer, upper bound range random values generate (exclusive). dtype Optional dtype tensor. integer types supported. specified, \"int32\" used. seed Optional R integer instance random_seed_generator(). default, seed argument NULL, internal global random_seed_generator() used. seed argument can used ensure deterministic (repeatable) random number generation. Note passing integer seed value produce random values call. generate different random values repeated calls, instance random_seed_generator() must provided seed value. Remark concerning JAX backend: tracing functions JAX backend global random_seed_generator() supported. Therefore, tracing default value seed=NULL produce error, seed argument must provided.","code":""},{"path":"https://keras3.posit.co/dev/reference/random_integer.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Draw random integers from a uniform distribution. — random_integer","text":"tensor random values.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/random_normal.html","id":null,"dir":"Reference","previous_headings":"","what":"Draw random samples from a normal (Gaussian) distribution. — random_normal","title":"Draw random samples from a normal (Gaussian) distribution. — random_normal","text":"Draw random samples normal (Gaussian) distribution.","code":""},{"path":"https://keras3.posit.co/dev/reference/random_normal.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Draw random samples from a normal (Gaussian) distribution. — random_normal","text":"","code":"random_normal(shape, mean = 0, stddev = 1, dtype = NULL, seed = NULL)"},{"path":"https://keras3.posit.co/dev/reference/random_normal.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Draw random samples from a normal (Gaussian) distribution. — random_normal","text":"shape shape random values generate. mean Float, defaults 0. Mean random values generate. stddev Float, defaults 1. Standard deviation random values generate. dtype Optional dtype tensor. floating point types supported. specified, config_floatx() used, defaults float32 unless configured otherwise (via config_set_floatx(float_dtype)). seed Optional R integer instance random_seed_generator(). default, seed argument NULL, internal global random_seed_generator() used. seed argument can used ensure deterministic (repeatable) random number generation. Note passing integer seed value produce random values call. generate different random values repeated calls, instance random_seed_generator() must provided seed value. Remark concerning JAX backend: tracing functions JAX backend global random_seed_generator() supported. Therefore, tracing default value seed=NULL produce error, seed argument must provided.","code":""},{"path":"https://keras3.posit.co/dev/reference/random_normal.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Draw random samples from a normal (Gaussian) distribution. — random_normal","text":"tensor random values.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/random_seed_generator.html","id":null,"dir":"Reference","previous_headings":"","what":"Generates variable seeds upon each call to a function generating random numbers. — random_seed_generator","title":"Generates variable seeds upon each call to a function generating random numbers. — random_seed_generator","text":"Keras, random number generators (random_normal()) stateless, meaning pass integer seed (seed=42), return values repeated calls. get different values call, SeedGenerator providing state random generator used. Note random number generators default seed NULL, implies internal global SeedGenerator used. need decouple RNG global state can provide local StateGenerator either deterministic random initial state. Remark concerning JAX backen: Note use local StateGenerator seed argument required JIT compilation RNG JAX backend, use global state supported.","code":""},{"path":"https://keras3.posit.co/dev/reference/random_seed_generator.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Generates variable seeds upon each call to a function generating random numbers. — random_seed_generator","text":"","code":"random_seed_generator(seed = NULL, name = NULL, ...)"},{"path":"https://keras3.posit.co/dev/reference/random_seed_generator.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Generates variable seeds upon each call to a function generating random numbers. — random_seed_generator","text":"seed Initial seed random number generator name String, name object ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/random_seed_generator.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Generates variable seeds upon each call to a function generating random numbers. — random_seed_generator","text":"SeedGenerator instance, can passed seed =  argument random tensor generators.","code":""},{"path":"https://keras3.posit.co/dev/reference/random_seed_generator.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Generates variable seeds upon each call to a function generating random numbers. — random_seed_generator","text":"Usage layer:","code":"seed_gen <- random_seed_generator(seed = 42) values <- random_normal(shape = c(2, 3), seed = seed_gen) new_values <- random_normal(shape = c(2, 3), seed = seed_gen) layer_dropout2 <- new_layer_class(   \"dropout2\",   initialize = function(...) {     super$initialize(...)     self$seed_generator <- random_seed_generator(seed = 1337)   },   call = function(x, training = FALSE) {     if (training) {       return(random_dropout(x, rate = 0.5, seed = self$seed_generator))     }     return(x)   } )  out <- layer_dropout(rate = 0.8) out(op_ones(10), training = TRUE) ## tf.Tensor([0. 5. 5. 0. 0. 0. 0. 0. 0. 0.], shape=(10), dtype=float32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/random_shuffle.html","id":null,"dir":"Reference","previous_headings":"","what":"Shuffle the elements of a tensor uniformly at random along an axis. — random_shuffle","title":"Shuffle the elements of a tensor uniformly at random along an axis. — random_shuffle","text":"Shuffle elements tensor uniformly random along axis.","code":""},{"path":"https://keras3.posit.co/dev/reference/random_shuffle.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Shuffle the elements of a tensor uniformly at random along an axis. — random_shuffle","text":"","code":"random_shuffle(x, axis = 1L, seed = NULL)"},{"path":"https://keras3.posit.co/dev/reference/random_shuffle.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Shuffle the elements of a tensor uniformly at random along an axis. — random_shuffle","text":"x tensor shuffled. axis integer specifying axis along shuffle. Defaults 0. seed Optional R integer instance random_seed_generator(). default, seed argument NULL, internal global random_seed_generator() used. seed argument can used ensure deterministic (repeatable) random number generation. Note passing integer seed value produce random values call. generate different random values repeated calls, instance random_seed_generator() must provided seed value. Remark concerning JAX backend: tracing functions JAX backend global random_seed_generator() supported. Therefore, tracing default value seed=NULL produce error, seed argument must provided.","code":""},{"path":"https://keras3.posit.co/dev/reference/random_shuffle.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Shuffle the elements of a tensor uniformly at random along an axis. — random_shuffle","text":"tensor, copy x axis axis shuffled.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/random_truncated_normal.html","id":null,"dir":"Reference","previous_headings":"","what":"Draw samples from a truncated normal distribution. — random_truncated_normal","title":"Draw samples from a truncated normal distribution. — random_truncated_normal","text":"values drawn normal distribution specified mean standard deviation, discarding re-drawing samples two standard deviations mean.","code":""},{"path":"https://keras3.posit.co/dev/reference/random_truncated_normal.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Draw samples from a truncated normal distribution. — random_truncated_normal","text":"","code":"random_truncated_normal(shape, mean = 0, stddev = 1, dtype = NULL, seed = NULL)"},{"path":"https://keras3.posit.co/dev/reference/random_truncated_normal.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Draw samples from a truncated normal distribution. — random_truncated_normal","text":"shape shape random values generate. mean Float, defaults 0. Mean random values generate. stddev Float, defaults 1. Standard deviation random values generate. dtype Optional dtype tensor. floating point types supported. specified, config_floatx() used, defaults float32 unless configured otherwise (via config_set_floatx(float_dtype)) seed Optional R integer instance random_seed_generator(). default, seed argument NULL, internal global random_seed_generator() used. seed argument can used ensure deterministic (repeatable) random number generation. Note passing integer seed value produce random values call. generate different random values repeated calls, instance random_seed_generator() must provided seed value. Remark concerning JAX backend: tracing functions JAX backend global random_seed_generator() supported. Therefore, tracing default value seed=NULL produce error, seed argument must provided.","code":""},{"path":"https://keras3.posit.co/dev/reference/random_truncated_normal.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Draw samples from a truncated normal distribution. — random_truncated_normal","text":"tensor random values.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/random_uniform.html","id":null,"dir":"Reference","previous_headings":"","what":"Draw samples from a uniform distribution. — random_uniform","title":"Draw samples from a uniform distribution. — random_uniform","text":"generated values follow uniform distribution range [minval, maxval). lower bound minval included range, upper bound maxval excluded. dtype must floating point type, default range [0, 1).","code":""},{"path":"https://keras3.posit.co/dev/reference/random_uniform.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Draw samples from a uniform distribution. — random_uniform","text":"","code":"random_uniform(shape, minval = 0, maxval = 1, dtype = NULL, seed = NULL)"},{"path":"https://keras3.posit.co/dev/reference/random_uniform.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Draw samples from a uniform distribution. — random_uniform","text":"shape shape random values generate. minval Float, defaults 0. Lower bound range random values generate (inclusive). maxval Float, defaults 1. Upper bound range random values generate (exclusive). dtype Optional dtype tensor. floating point types supported. specified, config_floatx() used, defaults float32 unless configured otherwise (via config_set_floatx(float_dtype)) seed Optional R integer instance random_seed_generator(). default, seed argument NULL, internal global random_seed_generator() used. seed argument can used ensure deterministic (repeatable) random number generation. Note passing integer seed value produce random values call. generate different random values repeated calls, instance random_seed_generator() must provided seed value. Remark concerning JAX backend: tracing functions JAX backend global random_seed_generator() supported. Therefore, tracing default value seed=NULL produce error, seed argument must provided.","code":""},{"path":"https://keras3.posit.co/dev/reference/random_uniform.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Draw samples from a uniform distribution. — random_uniform","text":"tensor random values.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/reexports.html","id":null,"dir":"Reference","previous_headings":"","what":"Objects exported from other packages — reexports","title":"Objects exported from other packages — reexports","text":"objects imported packages. Follow links see documentation. dotty . generics compile, fit magrittr %<>% reticulate %%, array_reshape, as_iterator, iter_next, iterate, np_array, tuple, use_python, use_virtualenv tensorflow all_dims, as_tensor, evaluate, export_savedmodel, tensorboard tfruns flag_boolean, flag_integer, flag_numeric, flag_string, flags, run_dir","code":""},{"path":"https://keras3.posit.co/dev/reference/register_keras_serializable.html","id":null,"dir":"Reference","previous_headings":"","what":"Registers a custom object with the Keras serialization framework. — register_keras_serializable","title":"Registers a custom object with the Keras serialization framework. — register_keras_serializable","text":"function registers custom class function Keras custom object registry, can serialized deserialized without needing entry user-provided custom_objects argument. also injects function Keras call get object's serializable string key. Note serialized deserialized, classes must implement get_config() method. Functions requirement. object registered key 'package>name' name, defaults object name passed.","code":""},{"path":"https://keras3.posit.co/dev/reference/register_keras_serializable.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Registers a custom object with the Keras serialization framework. — register_keras_serializable","text":"","code":"register_keras_serializable(object, name = NULL, package = NULL)"},{"path":"https://keras3.posit.co/dev/reference/register_keras_serializable.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Registers a custom object with the Keras serialization framework. — register_keras_serializable","text":"object keras object. name name serialize class package. package package class belongs . used key (\"package>name\") identify class. Defaults current package name, \"Custom\" outside package.","code":""},{"path":"https://keras3.posit.co/dev/reference/register_keras_serializable.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Registers a custom object with the Keras serialization framework. — register_keras_serializable","text":"object returned invisibly, convenient piping. primarily called side effects.","code":""},{"path":"https://keras3.posit.co/dev/reference/register_keras_serializable.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Registers a custom object with the Keras serialization framework. — register_keras_serializable","text":"","code":"# Note that `'my_package'` is used as the `package` argument here, and since # the `name` argument is not provided, `'MyDense'` is used as the `name`. layer_my_dense <- Layer(\"MyDense\") register_keras_serializable(layer_my_dense, package = \"my_package\")  MyDense <- environment(layer_my_dense)$`__class__` # the python class obj stopifnot(exprs = {   get_registered_object('my_package>MyDense') == MyDense   get_registered_name(MyDense) == 'my_package>MyDense' })"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/regularizer_l1.html","id":null,"dir":"Reference","previous_headings":"","what":"A regularizer that applies a L1 regularization penalty. — regularizer_l1","title":"A regularizer that applies a L1 regularization penalty. — regularizer_l1","text":"L1 regularization penalty computed : loss = l1 * reduce_sum(abs(x)) L1 may passed layer string identifier:   case, default value used l1=0.01.","code":"dense <- layer_dense(units = 3, kernel_regularizer = 'l1')"},{"path":"https://keras3.posit.co/dev/reference/regularizer_l1.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"A regularizer that applies a L1 regularization penalty. — regularizer_l1","text":"","code":"regularizer_l1(l1 = 0.01)"},{"path":"https://keras3.posit.co/dev/reference/regularizer_l1.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"A regularizer that applies a L1 regularization penalty. — regularizer_l1","text":"l1 float, L1 regularization factor.","code":""},{"path":"https://keras3.posit.co/dev/reference/regularizer_l1.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"A regularizer that applies a L1 regularization penalty. — regularizer_l1","text":"Regularizer instance can passed layer constructors used standalone object.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/regularizer_l1_l2.html","id":null,"dir":"Reference","previous_headings":"","what":"A regularizer that applies both L1 and L2 regularization penalties. — regularizer_l1_l2","title":"A regularizer that applies both L1 and L2 regularization penalties. — regularizer_l1_l2","text":"L1 regularization penalty computed : loss = l1 * reduce_sum(abs(x)) L2 regularization penalty computed loss = l2 * reduce_sum(square(x)) L1L2 may passed layer string identifier:   case, default values used l1=0.01 l2=0.01.","code":"dense <- layer_dense(units = 3, kernel_regularizer = 'L1L2')"},{"path":"https://keras3.posit.co/dev/reference/regularizer_l1_l2.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"A regularizer that applies both L1 and L2 regularization penalties. — regularizer_l1_l2","text":"","code":"regularizer_l1_l2(l1 = 0, l2 = 0)"},{"path":"https://keras3.posit.co/dev/reference/regularizer_l1_l2.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"A regularizer that applies both L1 and L2 regularization penalties. — regularizer_l1_l2","text":"l1 float, L1 regularization factor. l2 float, L2 regularization factor.","code":""},{"path":"https://keras3.posit.co/dev/reference/regularizer_l1_l2.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"A regularizer that applies both L1 and L2 regularization penalties. — regularizer_l1_l2","text":"Regularizer instance can passed layer constructors used standalone object.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/regularizer_l2.html","id":null,"dir":"Reference","previous_headings":"","what":"A regularizer that applies a L2 regularization penalty. — regularizer_l2","title":"A regularizer that applies a L2 regularization penalty. — regularizer_l2","text":"L2 regularization penalty computed : loss = l2 * reduce_sum(square(x)) L2 may passed layer string identifier:   case, default value used l2=0.01.","code":"dense <- layer_dense(units = 3, kernel_regularizer='l2')"},{"path":"https://keras3.posit.co/dev/reference/regularizer_l2.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"A regularizer that applies a L2 regularization penalty. — regularizer_l2","text":"","code":"regularizer_l2(l2 = 0.01)"},{"path":"https://keras3.posit.co/dev/reference/regularizer_l2.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"A regularizer that applies a L2 regularization penalty. — regularizer_l2","text":"l2 float, L2 regularization factor.","code":""},{"path":"https://keras3.posit.co/dev/reference/regularizer_l2.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"A regularizer that applies a L2 regularization penalty. — regularizer_l2","text":"Regularizer instance can passed layer constructors used standalone object.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/regularizer_orthogonal.html","id":null,"dir":"Reference","previous_headings":"","what":"Regularizer that encourages input vectors to be orthogonal to each other. — regularizer_orthogonal","title":"Regularizer that encourages input vectors to be orthogonal to each other. — regularizer_orthogonal","text":"can applied either rows matrix (mode=\"rows\") columns (mode=\"columns\"). applied Dense kernel shape (input_dim, units), rows mode seek make feature vectors (.e. basis output space) orthogonal .","code":""},{"path":"https://keras3.posit.co/dev/reference/regularizer_orthogonal.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Regularizer that encourages input vectors to be orthogonal to each other. — regularizer_orthogonal","text":"","code":"regularizer_orthogonal(factor = 0.01, mode = \"rows\")"},{"path":"https://keras3.posit.co/dev/reference/regularizer_orthogonal.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Regularizer that encourages input vectors to be orthogonal to each other. — regularizer_orthogonal","text":"factor Float. regularization factor. regularization penalty proportional factor times mean dot products L2-normalized rows (mode=\"rows\", columns mode=\"columns\") inputs, excluding product row/column .  Defaults 0.01. mode String, one {\"rows\", \"columns\"}. Defaults \"rows\". rows mode, regularization effect seeks make rows input orthogonal . columns mode, seeks make columns input orthogonal .","code":""},{"path":"https://keras3.posit.co/dev/reference/regularizer_orthogonal.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Regularizer that encourages input vectors to be orthogonal to each other. — regularizer_orthogonal","text":"Regularizer instance can passed layer constructors used standalone object.","code":""},{"path":"https://keras3.posit.co/dev/reference/regularizer_orthogonal.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Regularizer that encourages input vectors to be orthogonal to each other. — regularizer_orthogonal","text":"","code":"regularizer <- regularizer_orthogonal(factor=0.01) layer <- layer_dense(units=4, kernel_regularizer=regularizer)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/reset_state.html","id":null,"dir":"Reference","previous_headings":"","what":"Reset the state for a model, layer or metric. — reset_state","title":"Reset the state for a model, layer or metric. — reset_state","text":"Reset state model, layer metric.","code":""},{"path":"https://keras3.posit.co/dev/reference/reset_state.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Reset the state for a model, layer or metric. — reset_state","text":"","code":"reset_state(object)"},{"path":"https://keras3.posit.co/dev/reference/reset_state.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Reset the state for a model, layer or metric. — reset_state","text":"object Model, Layer, Metric instance Layers resettable state (E.g., adapt()-able preprocessing layers rnn layers resettable state, layer_dense() ). Calling Layer instance without resettable-state error.","code":""},{"path":"https://keras3.posit.co/dev/reference/reset_state.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Reset the state for a model, layer or metric. — reset_state","text":"object, invisibly.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/rnn_cell_gru.html","id":null,"dir":"Reference","previous_headings":"","what":"Cell class for the GRU layer. — rnn_cell_gru","title":"Cell class for the GRU layer. — rnn_cell_gru","text":"class processes one step within whole time sequence input, whereas layer_gru() processes whole sequence.","code":""},{"path":"https://keras3.posit.co/dev/reference/rnn_cell_gru.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Cell class for the GRU layer. — rnn_cell_gru","text":"","code":"rnn_cell_gru(   units,   activation = \"tanh\",   recurrent_activation = \"sigmoid\",   use_bias = TRUE,   kernel_initializer = \"glorot_uniform\",   recurrent_initializer = \"orthogonal\",   bias_initializer = \"zeros\",   kernel_regularizer = NULL,   recurrent_regularizer = NULL,   bias_regularizer = NULL,   kernel_constraint = NULL,   recurrent_constraint = NULL,   bias_constraint = NULL,   dropout = 0,   recurrent_dropout = 0,   reset_after = TRUE,   seed = NULL,   ... )"},{"path":"https://keras3.posit.co/dev/reference/rnn_cell_gru.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Cell class for the GRU layer. — rnn_cell_gru","text":"units Positive integer, dimensionality output space. activation Activation function use. Default: hyperbolic tangent (tanh). pass NULL, activation applied (ie. \"linear\" activation: (x) = x). recurrent_activation Activation function use recurrent step. Default: sigmoid (sigmoid). pass NULL, activation applied (ie. \"linear\" activation: (x) = x). use_bias Boolean, (default TRUE), whether layer use bias vector. kernel_initializer Initializer kernel weights matrix, used linear transformation inputs. Default: \"glorot_uniform\". recurrent_initializer Initializer recurrent_kernel weights matrix, used linear transformation recurrent state. Default: \"orthogonal\". bias_initializer Initializer bias vector. Default: \"zeros\". kernel_regularizer Regularizer function applied kernel weights matrix. Default: NULL. recurrent_regularizer Regularizer function applied recurrent_kernel weights matrix. Default: NULL. bias_regularizer Regularizer function applied bias vector. Default: NULL. kernel_constraint Constraint function applied kernel weights matrix. Default: NULL. recurrent_constraint Constraint function applied recurrent_kernel weights matrix. Default: NULL. bias_constraint Constraint function applied bias vector. Default: NULL. dropout Float 0 1. Fraction units drop linear transformation inputs. Default: 0. recurrent_dropout Float 0 1. Fraction units drop linear transformation recurrent state. Default: 0. reset_after GRU convention (whether apply reset gate matrix multiplication). FALSE = \"\", TRUE = \"\" (default cuDNN compatible). seed Random seed dropout. ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/rnn_cell_gru.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Cell class for the GRU layer. — rnn_cell_gru","text":"Layer instance, intended used layer_rnn().","code":""},{"path":"https://keras3.posit.co/dev/reference/rnn_cell_gru.html","id":"call-arguments","dir":"Reference","previous_headings":"","what":"Call Arguments","title":"Cell class for the GRU layer. — rnn_cell_gru","text":"inputs: 2D tensor, shape (batch, features). states: 2D tensor shape (batch, units), state previous time step. training: Python boolean indicating whether layer behave training mode inference mode. relevant dropout recurrent_dropout used.","code":""},{"path":"https://keras3.posit.co/dev/reference/rnn_cell_gru.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Cell class for the GRU layer. — rnn_cell_gru","text":"","code":"inputs <- random_uniform(c(32, 10, 8)) outputs <- inputs |> layer_rnn(rnn_cell_gru(4)) shape(outputs) ## shape(32, 4) rnn <- layer_rnn(    cell = rnn_cell_gru(4),    return_sequences=TRUE,    return_state=TRUE) c(whole_sequence_output, final_state) %<-% rnn(inputs) shape(whole_sequence_output) ## shape(32, 10, 4) shape(final_state) ## shape(32, 4)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/rnn_cell_lstm.html","id":null,"dir":"Reference","previous_headings":"","what":"Cell class for the LSTM layer. — rnn_cell_lstm","title":"Cell class for the LSTM layer. — rnn_cell_lstm","text":"class processes one step within whole time sequence input, whereas layer_lstm() processes whole sequence.","code":""},{"path":"https://keras3.posit.co/dev/reference/rnn_cell_lstm.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Cell class for the LSTM layer. — rnn_cell_lstm","text":"","code":"rnn_cell_lstm(   units,   activation = \"tanh\",   recurrent_activation = \"sigmoid\",   use_bias = TRUE,   kernel_initializer = \"glorot_uniform\",   recurrent_initializer = \"orthogonal\",   bias_initializer = \"zeros\",   unit_forget_bias = TRUE,   kernel_regularizer = NULL,   recurrent_regularizer = NULL,   bias_regularizer = NULL,   kernel_constraint = NULL,   recurrent_constraint = NULL,   bias_constraint = NULL,   dropout = 0,   recurrent_dropout = 0,   seed = NULL,   ... )"},{"path":"https://keras3.posit.co/dev/reference/rnn_cell_lstm.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Cell class for the LSTM layer. — rnn_cell_lstm","text":"units Positive integer, dimensionality output space. activation Activation function use. Default: hyperbolic tangent (tanh). pass NULL, activation applied (ie. \"linear\" activation: (x) = x). recurrent_activation Activation function use recurrent step. Default: sigmoid (sigmoid). pass NULL, activation applied (ie. \"linear\" activation: (x) = x). use_bias Boolean, (default TRUE), whether layer use bias vector. kernel_initializer Initializer kernel weights matrix, used linear transformation inputs. Default: \"glorot_uniform\". recurrent_initializer Initializer recurrent_kernel weights matrix, used linear transformation recurrent state. Default: \"orthogonal\". bias_initializer Initializer bias vector. Default: \"zeros\". unit_forget_bias Boolean (default TRUE). TRUE, add 1 bias forget gate initialization. Setting TRUE also force bias_initializer=\"zeros\". recommended Jozefowicz et al. kernel_regularizer Regularizer function applied kernel weights matrix. Default: NULL. recurrent_regularizer Regularizer function applied recurrent_kernel weights matrix. Default: NULL. bias_regularizer Regularizer function applied bias vector. Default: NULL. kernel_constraint Constraint function applied kernel weights matrix. Default: NULL. recurrent_constraint Constraint function applied recurrent_kernel weights matrix. Default: NULL. bias_constraint Constraint function applied bias vector. Default: NULL. dropout Float 0 1. Fraction units drop linear transformation inputs. Default: 0. recurrent_dropout Float 0 1. Fraction units drop linear transformation recurrent state. Default: 0. seed Random seed dropout. ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/rnn_cell_lstm.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Cell class for the LSTM layer. — rnn_cell_lstm","text":"Layer instance, intended used layer_rnn().","code":""},{"path":"https://keras3.posit.co/dev/reference/rnn_cell_lstm.html","id":"call-arguments","dir":"Reference","previous_headings":"","what":"Call Arguments","title":"Cell class for the LSTM layer. — rnn_cell_lstm","text":"inputs: 2D tensor, shape (batch, features). states: 2D tensor shape (batch, units), state previous time step. training: Boolean indicating whether layer behave training mode inference mode. relevant dropout recurrent_dropout used.","code":""},{"path":"https://keras3.posit.co/dev/reference/rnn_cell_lstm.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Cell class for the LSTM layer. — rnn_cell_lstm","text":"","code":"inputs <- random_uniform(c(32, 10, 8)) output <- inputs |>   layer_rnn(cell = rnn_cell_lstm(4)) shape(output) ## shape(32, 4) rnn <- layer_rnn(cell = rnn_cell_lstm(4),                  return_sequences = T,                  return_state = T) c(whole_sequence_output, ...final_state) %<-% rnn(inputs) str(whole_sequence_output) ## <tf.Tensor: shape=(32, 10, 4), dtype=float32, numpy=…> str(final_state) ## List of 2 ##  $ :<tf.Tensor: shape=(32, 4), dtype=float32, numpy=…> ##  $ :<tf.Tensor: shape=(32, 4), dtype=float32, numpy=…>"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/rnn_cell_simple.html","id":null,"dir":"Reference","previous_headings":"","what":"Cell class for SimpleRNN. — rnn_cell_simple","title":"Cell class for SimpleRNN. — rnn_cell_simple","text":"class processes one step within whole time sequence input, whereas layer_simple_rnn() processes whole sequence.","code":""},{"path":"https://keras3.posit.co/dev/reference/rnn_cell_simple.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Cell class for SimpleRNN. — rnn_cell_simple","text":"","code":"rnn_cell_simple(   units,   activation = \"tanh\",   use_bias = TRUE,   kernel_initializer = \"glorot_uniform\",   recurrent_initializer = \"orthogonal\",   bias_initializer = \"zeros\",   kernel_regularizer = NULL,   recurrent_regularizer = NULL,   bias_regularizer = NULL,   kernel_constraint = NULL,   recurrent_constraint = NULL,   bias_constraint = NULL,   dropout = 0,   recurrent_dropout = 0,   seed = NULL,   ... )"},{"path":"https://keras3.posit.co/dev/reference/rnn_cell_simple.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Cell class for SimpleRNN. — rnn_cell_simple","text":"units Positive integer, dimensionality output space. activation Activation function use. Default: hyperbolic tangent (tanh). pass NULL, activation applied (ie. \"linear\" activation: (x) = x). use_bias Boolean, (default TRUE), whether layer use bias vector. kernel_initializer Initializer kernel weights matrix, used linear transformation inputs. Default: \"glorot_uniform\". recurrent_initializer Initializer recurrent_kernel weights matrix, used linear transformation recurrent state. Default: \"orthogonal\". bias_initializer Initializer bias vector. Default: \"zeros\". kernel_regularizer Regularizer function applied kernel weights matrix. Default: NULL. recurrent_regularizer Regularizer function applied recurrent_kernel weights matrix. Default: NULL. bias_regularizer Regularizer function applied bias vector. Default: NULL. kernel_constraint Constraint function applied kernel weights matrix. Default: NULL. recurrent_constraint Constraint function applied recurrent_kernel weights matrix. Default: NULL. bias_constraint Constraint function applied bias vector. Default: NULL. dropout Float 0 1. Fraction units drop linear transformation inputs. Default: 0. recurrent_dropout Float 0 1. Fraction units drop linear transformation recurrent state. Default: 0. seed Random seed dropout. ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/rnn_cell_simple.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Cell class for SimpleRNN. — rnn_cell_simple","text":"Layer instance, intended used layer_rnn().","code":""},{"path":"https://keras3.posit.co/dev/reference/rnn_cell_simple.html","id":"call-arguments","dir":"Reference","previous_headings":"","what":"Call Arguments","title":"Cell class for SimpleRNN. — rnn_cell_simple","text":"sequence: 2D tensor, shape (batch, features). states: 2D tensor shape (batch, units), state previous time step. training: Python boolean indicating whether layer behave training mode inference mode. relevant dropout recurrent_dropout used.","code":""},{"path":"https://keras3.posit.co/dev/reference/rnn_cell_simple.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Cell class for SimpleRNN. — rnn_cell_simple","text":"","code":"inputs <- random_uniform(c(32, 10, 8)) rnn <- layer_rnn(cell = rnn_cell_simple(units = 4)) output <- rnn(inputs)  # The output has shape `(32, 4)`. rnn <- layer_rnn(     cell = rnn_cell_simple(units = 4),     return_sequences=TRUE,     return_state=TRUE ) # whole_sequence_output has shape `(32, 10, 4)`. # final_state has shape `(32, 4)`. c(whole_sequence_output, final_state) %<-% rnn(inputs)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/rnn_cells_stack.html","id":null,"dir":"Reference","previous_headings":"","what":"Wrapper allowing a stack of RNN cells to behave as a single cell. — rnn_cells_stack","title":"Wrapper allowing a stack of RNN cells to behave as a single cell. — rnn_cells_stack","text":"Used implement efficient stacked RNNs.","code":""},{"path":"https://keras3.posit.co/dev/reference/rnn_cells_stack.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Wrapper allowing a stack of RNN cells to behave as a single cell. — rnn_cells_stack","text":"","code":"rnn_cells_stack(cells, ...)"},{"path":"https://keras3.posit.co/dev/reference/rnn_cells_stack.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Wrapper allowing a stack of RNN cells to behave as a single cell. — rnn_cells_stack","text":"cells List RNN cell instances. ... Unnamed arguments treated additional cells. Named arguments passed underlying layer.","code":""},{"path":"https://keras3.posit.co/dev/reference/rnn_cells_stack.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Wrapper allowing a stack of RNN cells to behave as a single cell. — rnn_cells_stack","text":"Layer instance, intended used layer_rnn().","code":""},{"path":"https://keras3.posit.co/dev/reference/rnn_cells_stack.html","id":"example","dir":"Reference","previous_headings":"","what":"Example","title":"Wrapper allowing a stack of RNN cells to behave as a single cell. — rnn_cells_stack","text":"","code":"batch_size <- 3 sentence_length <- 5 num_features <- 2 new_shape <- c(batch_size, sentence_length, num_features) x <- array(1:30, dim = new_shape)  rnn_cells <- lapply(1:2, function(x) rnn_cell_lstm(units = 128)) stacked_lstm <- rnn_cells_stack(rnn_cells) lstm_layer <- layer_rnn(cell = stacked_lstm)  result <- lstm_layer(x) str(result) ## <tf.Tensor: shape=(3, 128), dtype=float32, numpy=…>"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/save_model.html","id":null,"dir":"Reference","previous_headings":"","what":"Saves a model as a .keras file. — save_model","title":"Saves a model as a .keras file. — save_model","text":"Saves model .keras file.","code":""},{"path":"https://keras3.posit.co/dev/reference/save_model.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Saves a model as a .keras file. — save_model","text":"","code":"save_model(model, filepath = NULL, overwrite = FALSE, zipped = NULL, ...)"},{"path":"https://keras3.posit.co/dev/reference/save_model.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Saves a model as a .keras file. — save_model","text":"model keras model. filepath string, Path save model. Must end .keras. overwrite Whether overwrite existing model target location, instead ask user via interactive prompt. zipped Whether save model zipped .keras archive (default saving locally), unzipped directory (default saving Hugging Face Hub). ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/save_model.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Saves a model as a .keras file. — save_model","text":"filepath provided, function called primarily side effects, model returned invisibly. filepath provided NULL, serialized model returned R raw vector.","code":""},{"path":"https://keras3.posit.co/dev/reference/save_model.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Saves a model as a .keras file. — save_model","text":"saved .keras file zip archive contains: model's configuration (architecture) model's weights model's optimizer's state () Thus models can reinstantiated exact state.","code":"model <- keras_model_sequential(input_shape = c(3)) |>   layer_dense(5) |>   layer_activation_softmax()  model |> save_model(\"model.keras\") loaded_model <- load_model(\"model.keras\") x <- random_uniform(c(10, 3)) stopifnot(all.equal(   model |> predict(x),   loaded_model |> predict(x) )) zip::zip_list(\"model.keras\")[, \"filename\"] ## [1] \"metadata.json\"    \"config.json\"      \"model.weights.h5\""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/save_model_config.html","id":null,"dir":"Reference","previous_headings":"","what":"Save and load model configuration as JSON — save_model_config","title":"Save and load model configuration as JSON — save_model_config","text":"Save re-load models configurations JSON. Note representation include weights, architecture.","code":""},{"path":"https://keras3.posit.co/dev/reference/save_model_config.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Save and load model configuration as JSON — save_model_config","text":"","code":"save_model_config(model, filepath = NULL, overwrite = FALSE)  load_model_config(filepath, custom_objects = NULL)"},{"path":"https://keras3.posit.co/dev/reference/save_model_config.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Save and load model configuration as JSON — save_model_config","text":"model Model object save filepath path json file model config. overwrite Whether overwrite existing model configuration json filepath, instead ask user via interactive prompt. custom_objects Optional named list mapping names custom classes functions considered deserialization.","code":""},{"path":"https://keras3.posit.co/dev/reference/save_model_config.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Save and load model configuration as JSON — save_model_config","text":"called primarily side effects. model returned, invisibly, enable usage pipe.","code":""},{"path":"https://keras3.posit.co/dev/reference/save_model_config.html","id":"details","dir":"Reference","previous_headings":"","what":"Details","title":"Save and load model configuration as JSON — save_model_config","text":"Note: save_model_config() serializes model JSON using serialize_keras_object(), get_config(). serialize_keras_object() returns superset get_config(), additional information needed create class object needed restore model. See example extract get_config() value saved model.","code":""},{"path":"https://keras3.posit.co/dev/reference/save_model_config.html","id":"example","dir":"Reference","previous_headings":"","what":"Example","title":"Save and load model configuration as JSON — save_model_config","text":"","code":"model <- keras_model_sequential(input_shape = 10) |> layer_dense(10) file <- tempfile(\"model-config-\", fileext = \".json\") save_model_config(model, file)  # load a new model instance with the same architecture but different weights model2 <- load_model_config(file)  stopifnot(exprs = {   all.equal(get_config(model), get_config(model2))    # To extract the `get_config()` value from a saved model config:   all.equal(       get_config(model),       structure(jsonlite::read_json(file)$config,                 \"__class__\" = keras_model_sequential()$`__class__`)   ) })"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/save_model_weights.html","id":null,"dir":"Reference","previous_headings":"","what":"Saves all layer weights to a .weights.h5 file. — save_model_weights","title":"Saves all layer weights to a .weights.h5 file. — save_model_weights","text":"Saves layer weights .weights.h5 file.","code":""},{"path":"https://keras3.posit.co/dev/reference/save_model_weights.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Saves all layer weights to a .weights.h5 file. — save_model_weights","text":"","code":"save_model_weights(model, filepath, overwrite = FALSE)"},{"path":"https://keras3.posit.co/dev/reference/save_model_weights.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Saves all layer weights to a .weights.h5 file. — save_model_weights","text":"model keras Model object filepath string. Path save model. Must end .weights.h5. overwrite Whether overwrite existing model target location, instead ask user via interactive prompt.","code":""},{"path":"https://keras3.posit.co/dev/reference/save_model_weights.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Saves all layer weights to a .weights.h5 file. — save_model_weights","text":"called primarily side effects. model returned, invisibly, enable usage pipe.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/serialize_keras_object.html","id":null,"dir":"Reference","previous_headings":"","what":"Retrieve the full config by serializing the Keras object. — serialize_keras_object","title":"Retrieve the full config by serializing the Keras object. — serialize_keras_object","text":"serialize_keras_object() serializes Keras object named list represents object, reciprocal function deserialize_keras_object(). See deserialize_keras_object() information full config format.","code":""},{"path":"https://keras3.posit.co/dev/reference/serialize_keras_object.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Retrieve the full config by serializing the Keras object. — serialize_keras_object","text":"","code":"serialize_keras_object(obj)"},{"path":"https://keras3.posit.co/dev/reference/serialize_keras_object.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Retrieve the full config by serializing the Keras object. — serialize_keras_object","text":"obj Keras object serialize.","code":""},{"path":"https://keras3.posit.co/dev/reference/serialize_keras_object.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Retrieve the full config by serializing the Keras object. — serialize_keras_object","text":"named list represents object config. config expected contain simple types , can saved json. object can deserialized config via deserialize_keras_object().","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/set_random_seed.html","id":null,"dir":"Reference","previous_headings":"","what":"Sets all random seeds (Python, NumPy, and backend framework, e.g. TF). — set_random_seed","title":"Sets all random seeds (Python, NumPy, and backend framework, e.g. TF). — set_random_seed","text":"can use utility make almost Keras program fully deterministic. limitations apply cases network communications involved (e.g. parameter server distribution), creates additional sources randomness, certain non-deterministic cuDNN ops involved. sets: R session seed: set.seed() Python session seed: import random; random.seed(seed) Python NumPy seed: import numpy; numpy.random.seed(seed) TensorFlow seed: tf$random$set_seed(seed) (TF installed) Torch seed: import(\"torch\")$manual_seed(seed) (backend torch) disables Python hash randomization. Note TensorFlow seed set even using TensorFlow backend framework, since many workflows leverage tf$data pipelines (feature random shuffling). Likewise many workflows might leverage NumPy APIs.","code":""},{"path":"https://keras3.posit.co/dev/reference/set_random_seed.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Sets all random seeds (Python, NumPy, and backend framework, e.g. TF). — set_random_seed","text":"","code":"set_random_seed(seed)"},{"path":"https://keras3.posit.co/dev/reference/set_random_seed.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Sets all random seeds (Python, NumPy, and backend framework, e.g. TF). — set_random_seed","text":"seed Integer, random seed use.","code":""},{"path":"https://keras3.posit.co/dev/reference/set_random_seed.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Sets all random seeds (Python, NumPy, and backend framework, e.g. TF). — set_random_seed","text":"return value, called side effects.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/set_state_tree.html","id":null,"dir":"Reference","previous_headings":"","what":"Assigns values to variables of the model. — set_state_tree","title":"Assigns values to variables of the model. — set_state_tree","text":"method takes dictionary (named list) nested variable values, represents state tree model, assigns corresponding variables model. keys (list names) represent variable names (e.g., 'trainable_variables', 'optimizer_variables'), values nested dictionaries containing variable paths corresponding values.","code":""},{"path":"https://keras3.posit.co/dev/reference/set_state_tree.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Assigns values to variables of the model. — set_state_tree","text":"","code":"set_state_tree(object, state_tree)"},{"path":"https://keras3.posit.co/dev/reference/set_state_tree.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Assigns values to variables of the model. — set_state_tree","text":"object keras model. state_tree dictionary representing state tree model. keys variable names, values nested dictionaries representing variable paths values.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/shape.html","id":null,"dir":"Reference","previous_headings":"","what":"Tensor shape utility — shape","title":"Tensor shape utility — shape","text":"function can used get create tensor shape.","code":""},{"path":"https://keras3.posit.co/dev/reference/shape.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Tensor shape utility — shape","text":"","code":"shape(...)  # S3 method for class 'keras_shape' format(x, ..., prefix = TRUE)  # S3 method for class 'keras_shape' print(x, ...)  # S3 method for class 'keras_shape' x[...]  # S3 method for class 'keras_shape' as.integer(x, ...)  # S3 method for class 'keras_shape' Summary(..., na.rm = FALSE)  # S3 method for class 'keras_shape' as.list(x, ...)  # S3 method for class 'keras_shape' x == y  # S3 method for class 'keras_shape' x != y"},{"path":"https://keras3.posit.co/dev/reference/shape.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Tensor shape utility — shape","text":"... shape specification. Numerics, NULL tensors valid. NULL, NA, -1L can used specify unspecified dim size. Tensors dispatched op_shape() extract tensor shape. Values wrapped () used asis (see examples). objects coerced via .integer(). x, y keras_shape object. prefix Whether format shape object prefix. Defaults \"shape\". na.rm passed Summary group generics like prod(). Unknown axes treated NA.","code":""},{"path":"https://keras3.posit.co/dev/reference/shape.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Tensor shape utility — shape","text":"list \"keras_shape\" class attribute. element list either ) NULL, b) R integer c) scalar integer tensor (e.g., supplied TF tensor unspecified dimension function traced).","code":""},{"path":"https://keras3.posit.co/dev/reference/shape.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Tensor shape utility — shape","text":"3 ways specify unknown dimension     functions take 'shape' argument also coerce shape()     can also use shape() get shape tensor (excepting scalar integer tensors).             Combine expand shapes         Scalar integer tensors treated axis values. commonly encountered tracing function graph mode, axis size might unknown.     useful pattern unpack shape() %<-%, like :                     want resolve shape tensor can potentially scalar integer, can wrap tensor (), use op_shape().","code":"shape(1, 2, 3) ## shape(1, 2, 3) shape(NA,   2, 3) shape(NULL, 2, 3) shape(-1,   2, 3) ## shape(NA, 2, 3) ## shape(NA, 2, 3) ## shape(NA, 2, 3) layer_input(c(1, 2, 3)) layer_input(shape(1, 2, 3)) ## <KerasTensor shape=(None, 1, 2, 3), dtype=float32, sparse=False, ragged=False, name=keras_tensor> ## <KerasTensor shape=(None, 1, 2, 3), dtype=float32, sparse=False, ragged=False, name=keras_tensor_1> symbolic_tensor <- layer_input(shape(1, 2, 3)) shape(symbolic_tensor) ## shape(NA, 1, 2, 3) eager_tensor <- op_ones(c(1,2,3)) shape(eager_tensor) ## shape(1, 2, 3) op_shape(eager_tensor) ## shape(1, 2, 3) shape(symbolic_tensor, 4) ## shape(NA, 1, 2, 3, 4) shape(5, symbolic_tensor, 4) ## shape(5, NA, 1, 2, 3, 4) tfn <- tensorflow::tf_function(function(x) {   print(op_shape(x))   x }, input_signature = list(tensorflow::tf$TensorSpec(shape(1, NA, 3)))) invisible(tfn(op_ones(shape(1, 2, 3)))) ## shape(1, Tensor(\"strided_slice:0\", shape=(), dtype=int32), 3) c(batch_size, seq_len, channels) %<-% shape(x)  # `%<-%` also has support for skipping values # during unpacking with `.` and `...`. For example, # To retrieve just the first and/or last dim: c(batch_size, ...) %<-% shape(x) c(batch_size, ., .) %<-% shape(x) c(..., channels) %<-% shape(x) c(batch_size, ..., channels) %<-% shape(x) c(batch_size, ., channels) %<-% shape(x) echo_print <- function(x) {   message(\"> \", deparse(substitute(x)));   if(!is.null(x)) print(x) } tfn <- tensorflow::tf_function(function(x) {   c(axis1, axis2, axis3) %<-% shape(x)   echo_print(str(list(axis1 = axis1, axis2 = axis2, axis3 = axis3)))    echo_print(shape(axis1))               # use axis1 tensor as axis value   echo_print(shape(axis1, axis2, axis3)) # use axis1 tensor as axis value    # use shape() to compose a new shape, e.g., in multihead attention   n_heads <- 4   echo_print(shape(axis1, axis2, n_heads, axis3/n_heads))    x }, input_signature = list(tensorflow::tf$TensorSpec(shape(NA, 4, 16)))) invisible(tfn(op_ones(shape(2, 4, 16)))) ## > str(list(axis1 = axis1, axis2 = axis2, axis3 = axis3)) ## List of 3 ##  $ axis1:<tf.Tensor 'strided_slice:0' shape=() dtype=int32> ##  $ axis2: int 4 ##  $ axis3: int 16 ## > shape(axis1) ## shape(Tensor(\"strided_slice:0\", shape=(), dtype=int32)) ## > shape(axis1, axis2, axis3) ## shape(Tensor(\"strided_slice:0\", shape=(), dtype=int32), 4, 16) ## > shape(axis1, axis2, n_heads, axis3/n_heads) ## shape(Tensor(\"strided_slice:0\", shape=(), dtype=int32), 4, 4, 4) (x <- op_convert_to_tensor(2L)) ## tf.Tensor(2, shape=(), dtype=int32) # by default, shape() treats scalar integer tensors as axis values shape(x) ## shape(tf.Tensor(2, shape=(), dtype=int32)) # to access the shape of a scalar integer, # call `op_shape()`, or protect with `I()` op_shape(x) ## shape() shape(I(x)) ## shape()"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/split_dataset.html","id":null,"dir":"Reference","previous_headings":"","what":"Splits a dataset into a left half and a right half (e.g. train / test). — split_dataset","title":"Splits a dataset into a left half and a right half (e.g. train / test). — split_dataset","text":"Splits dataset left half right half (e.g. train / test).","code":""},{"path":"https://keras3.posit.co/dev/reference/split_dataset.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Splits a dataset into a left half and a right half (e.g. train / test). — split_dataset","text":"","code":"split_dataset(   dataset,   left_size = NULL,   right_size = NULL,   shuffle = FALSE,   seed = NULL )"},{"path":"https://keras3.posit.co/dev/reference/split_dataset.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Splits a dataset into a left half and a right half (e.g. train / test). — split_dataset","text":"dataset tf$data$Dataset, torch$utils$data$Dataset object, list arrays length. left_size float (range [0, 1]), signifies fraction data pack left dataset. integer, signifies number samples pack left dataset. NULL, defaults complement right_size. Defaults NULL. right_size float (range [0, 1]), signifies fraction data pack right dataset. integer, signifies number samples pack right dataset. NULL, defaults complement left_size. Defaults NULL. shuffle Boolean, whether shuffle data splitting . seed random seed shuffling.","code":""},{"path":"https://keras3.posit.co/dev/reference/split_dataset.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Splits a dataset into a left half and a right half (e.g. train / test). — split_dataset","text":"list two tf$data$Dataset objects: left right splits.","code":""},{"path":"https://keras3.posit.co/dev/reference/split_dataset.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Splits a dataset into a left half and a right half (e.g. train / test). — split_dataset","text":"","code":"data <- random_uniform(c(1000, 4)) c(left_ds, right_ds) %<-% split_dataset(list(data$numpy()), left_size = 0.8) left_ds$cardinality() ## tf.Tensor(800, shape=(), dtype=int64) right_ds$cardinality() ## tf.Tensor(200, shape=(), dtype=int64)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/summary.keras.src.models.model.Model.html","id":null,"dir":"Reference","previous_headings":"","what":"Print a summary of a Keras Model — summary.keras.src.models.model.Model","title":"Print a summary of a Keras Model — summary.keras.src.models.model.Model","text":"Print summary Keras Model","code":""},{"path":"https://keras3.posit.co/dev/reference/summary.keras.src.models.model.Model.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Print a summary of a Keras Model — summary.keras.src.models.model.Model","text":"","code":"# S3 method for class 'keras.src.models.model.Model' summary(object, ...)  # S3 method for class 'keras.src.models.model.Model' format(   x,   line_length = min(getOption(\"width\"), 180),   positions = NULL,   expand_nested = FALSE,   show_trainable = NA,   ...,   layer_range = NULL,   compact = TRUE )  # S3 method for class 'keras.src.models.model.Model' print(x, ...)"},{"path":"https://keras3.posit.co/dev/reference/summary.keras.src.models.model.Model.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Print a summary of a Keras Model — summary.keras.src.models.model.Model","text":"object, x Keras model instance ... summary() print(), passed format(). format(), passed model$summary(). line_length Total length printed lines positions Relative absolute positions log elements line. provided, defaults c(0.33, 0.55, 0.67, 1.0). expand_nested Whether expand nested models. provided, defaults FALSE. show_trainable Whether show layer trainable. provided, defaults FALSE. layer_range list, tuple, vector 2 strings, starting layer name ending layer name (inclusive) indicating range layers printed summary. also accepts regex patterns instead exact name. case, start predicate first element matches layer_range[[1]] end predicate last element matches layer_range[[1]]. default NULL considers layers model. compact Whether remove white-space lines model summary. (Default TRUE)","code":""},{"path":"https://keras3.posit.co/dev/reference/summary.keras.src.models.model.Model.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Print a summary of a Keras Model — summary.keras.src.models.model.Model","text":"format() returns length 1 character vector. print() returns model object invisibly. summary() returns output format() invisibly printing .","code":""},{"path":"https://keras3.posit.co/dev/reference/summary.keras.src.models.model.Model.html","id":"enabling-color-output-in-knitr-rmarkdown-quarto-","dir":"Reference","previous_headings":"","what":"Enabling color output in Knitr (RMarkdown, Quarto)","title":"Print a summary of a Keras Model — summary.keras.src.models.model.Model","text":"order enable color output quarto rmarkdown document html output format (include revealjs presentations), need following setup chunk:","code":"```{r setup, include = FALSE}  options(cli.num_colors = 256)  fansi::set_knit_hooks(knitr::knit_hooks)  options(width = 75) # adjust as needed for format  ```"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/test_on_batch.html","id":null,"dir":"Reference","previous_headings":"","what":"Test the model on a single batch of samples. — test_on_batch","title":"Test the model on a single batch of samples. — test_on_batch","text":"Test model single batch samples.","code":""},{"path":"https://keras3.posit.co/dev/reference/test_on_batch.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Test the model on a single batch of samples. — test_on_batch","text":"","code":"test_on_batch(object, x, y = NULL, sample_weight = NULL, ...)"},{"path":"https://keras3.posit.co/dev/reference/test_on_batch.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Test the model on a single batch of samples. — test_on_batch","text":"object Keras model object x Input data. Must array-like. y Target data. Must array-like. sample_weight Optional array length x, containing weights apply model's loss sample. case temporal data, can pass 2D array shape (samples, sequence_length), apply different weight every timestep every sample. ... forward/backward compatability","code":""},{"path":"https://keras3.posit.co/dev/reference/test_on_batch.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Test the model on a single batch of samples. — test_on_batch","text":"scalar loss value (metrics), named list loss metric values (metrics).","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/text_dataset_from_directory.html","id":null,"dir":"Reference","previous_headings":"","what":"Generates a tf.data.Dataset from text files in a directory. — text_dataset_from_directory","title":"Generates a tf.data.Dataset from text files in a directory. — text_dataset_from_directory","text":"directory structure :   calling text_dataset_from_directory(main_directory, labels='inferred') return tf.data.Dataset yields batches texts subdirectories class_a class_b, together labels 0 1 (0 corresponding class_a 1 corresponding class_b). .txt files supported time.","code":"main_directory/ ...class_a/ ......a_text_1.txt ......a_text_2.txt ...class_b/ ......b_text_1.txt ......b_text_2.txt"},{"path":"https://keras3.posit.co/dev/reference/text_dataset_from_directory.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Generates a tf.data.Dataset from text files in a directory. — text_dataset_from_directory","text":"","code":"text_dataset_from_directory(   directory,   labels = \"inferred\",   label_mode = \"int\",   class_names = NULL,   batch_size = 32L,   max_length = NULL,   shuffle = TRUE,   seed = NULL,   validation_split = NULL,   subset = NULL,   follow_links = FALSE,   verbose = TRUE )"},{"path":"https://keras3.posit.co/dev/reference/text_dataset_from_directory.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Generates a tf.data.Dataset from text files in a directory. — text_dataset_from_directory","text":"directory Directory data located. labels \"inferred\", contain subdirectories, containing text files class. Otherwise, directory structure ignored. labels Either \"inferred\" (labels generated directory structure), NULL (labels), list/tuple integer labels size number text files found directory. Labels sorted according alphanumeric order text file paths (obtained via os.walk(directory) Python). label_mode String describing encoding labels. Options : \"int\": means labels encoded integers (e.g. sparse_categorical_crossentropy loss). \"categorical\" means labels encoded categorical vector (e.g. categorical_crossentropy loss). \"binary\" means labels (can 2) encoded float32 scalars values 0 1 (e.g. binary_crossentropy). NULL (labels). class_names valid \"labels\" \"inferred\". explicit list class names (must match names subdirectories). Used control order classes (otherwise alphanumerical order used). batch_size Size batches data. NULL, data batched (dataset yield individual samples). Defaults 32. max_length Maximum size text string. Texts longer truncated max_length. shuffle Whether shuffle data. set FALSE, sorts data alphanumeric order. Defaults TRUE. seed Optional random seed shuffling transformations. validation_split Optional float 0 1, fraction data reserve validation. subset Subset data return. One \"training\", \"validation\" \"\". used validation_split set. subset=\"\", utility returns tuple two datasets (training validation datasets respectively). follow_links Whether visits subdirectories pointed symlinks. Defaults FALSE. verbose Whether display number information classes number files found. Defaults TRUE.","code":""},{"path":"https://keras3.posit.co/dev/reference/text_dataset_from_directory.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Generates a tf.data.Dataset from text files in a directory. — text_dataset_from_directory","text":"tf.data.Dataset object. label_mode NULL, yields string tensors shape (batch_size,), containing contents batch text files. Otherwise, yields tuple (texts, labels), texts shape (batch_size,) labels follows format described . Rules regarding labels format: label_mode int, labels int32 tensor shape (batch_size,). label_mode binary, labels float32 tensor 1s 0s shape (batch_size, 1). label_mode categorical, labels float32 tensor shape (batch_size, num_classes), representing one-hot encoding class index.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/time_distributed.html","id":null,"dir":"Reference","previous_headings":"","what":"layer_time_distributed — time_distributed","title":"layer_time_distributed — time_distributed","text":"time_distributed() alias layer_time_distributed(). See ?layer_time_distributed() full documentation.","code":""},{"path":"https://keras3.posit.co/dev/reference/time_distributed.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"layer_time_distributed — time_distributed","text":"","code":"time_distributed(object, layer, ...)"},{"path":"https://keras3.posit.co/dev/reference/time_distributed.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"layer_time_distributed — time_distributed","text":"object Object compose layer . tensor, array, sequential model. layer Layer instance. ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/time_distributed.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"layer_time_distributed — time_distributed","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/timeseries_dataset_from_array.html","id":null,"dir":"Reference","previous_headings":"","what":"Creates a dataset of sliding windows over a timeseries provided as array. — timeseries_dataset_from_array","title":"Creates a dataset of sliding windows over a timeseries provided as array. — timeseries_dataset_from_array","text":"function takes sequence data-points gathered equal intervals, along time series parameters length sequences/windows, spacing two sequence/windows, etc., produce batches timeseries inputs targets.","code":""},{"path":"https://keras3.posit.co/dev/reference/timeseries_dataset_from_array.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Creates a dataset of sliding windows over a timeseries provided as array. — timeseries_dataset_from_array","text":"","code":"timeseries_dataset_from_array(   data,   targets,   sequence_length,   sequence_stride = 1L,   sampling_rate = 1L,   batch_size = 128L,   shuffle = FALSE,   seed = NULL,   start_index = NULL,   end_index = NULL )"},{"path":"https://keras3.posit.co/dev/reference/timeseries_dataset_from_array.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Creates a dataset of sliding windows over a timeseries provided as array. — timeseries_dataset_from_array","text":"data array eager tensor containing consecutive data points (timesteps). first dimension expected time dimension. targets Targets corresponding timesteps data. targets[] target corresponding window starts index (see example 2 ). Pass NULL target data (case dataset yield input data). sequence_length Length output sequences (number timesteps). sequence_stride Period successive output sequences. stride s, output samples start index data[], data[+ s], data[+ 2 * s], etc. sampling_rate Period successive individual timesteps within sequences. rate r, timesteps data[], data[+ r], ... data[+ sequence_length] used creating sample sequence. batch_size Number timeseries samples batch (except maybe last one). NULL, data batched (dataset yield individual samples). shuffle Whether shuffle output samples, instead draw chronological order. seed Optional int; random seed shuffling. start_index Optional int; data points earlier (exclusive) start_index used output sequences. useful reserve part data test validation. end_index Optional int; data points later (exclusive) end_index used output sequences. useful reserve part data test validation.","code":""},{"path":"https://keras3.posit.co/dev/reference/timeseries_dataset_from_array.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Creates a dataset of sliding windows over a timeseries provided as array. — timeseries_dataset_from_array","text":"tf$data$Dataset instance. targets passed, dataset yields list (batch_of_sequences, batch_of_targets). , dataset yields batch_of_sequences. Example 1: Consider indices [0, 1, ... 98]. sequence_length=10,  sampling_rate=2, sequence_stride=3, shuffle=FALSE, dataset yield batches sequences composed following indices:   case last 2 data points discarded since full sequence can generated include (next sequence started index 81, thus last step gone 98). Example 2: Temporal regression. Consider array data scalar values, shape (steps,). generate dataset uses past 10 timesteps predict next timestep, use:     Example 3: Temporal regression many--many architectures. Consider two arrays scalar values X Y, shape (100,). resulting dataset consist samples 20 timestamps . samples overlap. generate dataset uses current timestamp predict corresponding target timestep, use:","code":"First sequence:  [0  2  4  6  8 10 12 14 16 18] Second sequence: [3  5  7  9 11 13 15 17 19 21] Third sequence:  [6  8 10 12 14 16 18 20 22 24] ... Last sequence:   [78 80 82 84 86 88 90 92 94 96] data <- op_array(1:20) input_data <- data[1:10] targets <- data[11:20] dataset <- timeseries_dataset_from_array(   input_data, targets, sequence_length=10) iter <- reticulate::as_iterator(dataset) reticulate::iter_next(iter) ## [[1]] ## tf.Tensor([[ 1  2  3  4  5  6  7  8  9 10]], shape=(1, 10), dtype=int32) ## ## [[2]] ## tf.Tensor([11], shape=(1), dtype=int32) X <- op_array(1:100) Y <- X*2  sample_length <- 20 input_dataset <- timeseries_dataset_from_array(     X, NULL, sequence_length=sample_length, sequence_stride=sample_length) target_dataset <- timeseries_dataset_from_array(     Y, NULL, sequence_length=sample_length, sequence_stride=sample_length)   inputs <- reticulate::as_iterator(input_dataset) %>% reticulate::iter_next() targets <- reticulate::as_iterator(target_dataset) %>% reticulate::iter_next()"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/to_categorical.html","id":null,"dir":"Reference","previous_headings":"","what":"Converts a class vector (integers) to binary class matrix. — to_categorical","title":"Converts a class vector (integers) to binary class matrix. — to_categorical","text":"E.g. use loss_categorical_crossentropy().","code":""},{"path":"https://keras3.posit.co/dev/reference/to_categorical.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Converts a class vector (integers) to binary class matrix. — to_categorical","text":"","code":"to_categorical(x, num_classes = NULL)"},{"path":"https://keras3.posit.co/dev/reference/to_categorical.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Converts a class vector (integers) to binary class matrix. — to_categorical","text":"x Array-like class values converted matrix (integers 0 num_classes - 1). R factors coerced integer offset 0-based, .e., .integer(x) - 1L. num_classes Total number classes. NULL, inferred max(x) + 1. Defaults NULL.","code":""},{"path":"https://keras3.posit.co/dev/reference/to_categorical.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Converts a class vector (integers) to binary class matrix. — to_categorical","text":"binary matrix representation input R array. class axis placed last.","code":""},{"path":"https://keras3.posit.co/dev/reference/to_categorical.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Converts a class vector (integers) to binary class matrix. — to_categorical","text":"","code":"a <- to_categorical(c(0, 1, 2, 3), num_classes=4) print(a) ##      [,1] [,2] [,3] [,4] ## [1,]    1    0    0    0 ## [2,]    0    1    0    0 ## [3,]    0    0    1    0 ## [4,]    0    0    0    1 b <- array(c(.9, .04, .03, .03,               .3, .45, .15, .13,               .04, .01, .94, .05,               .12, .21, .5, .17),               dim = c(4, 4)) loss <- op_categorical_crossentropy(a, b) loss ## tf.Tensor([0.41284522 0.45601739 0.54430155 0.80437282], shape=(4), dtype=float64) loss <- op_categorical_crossentropy(a, a) loss ## tf.Tensor([1.00000005e-07 1.00000005e-07 1.00000005e-07 1.00000005e-07], shape=(4), dtype=float64)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/train_on_batch.html","id":null,"dir":"Reference","previous_headings":"","what":"Runs a single gradient update on a single batch of data. — train_on_batch","title":"Runs a single gradient update on a single batch of data. — train_on_batch","text":"Runs single gradient update single batch data.","code":""},{"path":"https://keras3.posit.co/dev/reference/train_on_batch.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Runs a single gradient update on a single batch of data. — train_on_batch","text":"","code":"train_on_batch(object, x, y = NULL, sample_weight = NULL, class_weight = NULL)"},{"path":"https://keras3.posit.co/dev/reference/train_on_batch.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Runs a single gradient update on a single batch of data. — train_on_batch","text":"object Keras model object x Input data. Must array-like. y Target data. Must array-like. sample_weight Optional array length x, containing weights apply model's loss sample. case temporal data, can pass 2D array shape (samples, sequence_length), apply different weight every timestep every sample. class_weight Optional named list mapping class indices (integers, 0-based) weight (float) apply model's loss samples class training. can useful tell model \"pay attention\" samples -represented class. class_weight specified targets rank 2 greater, either y must one-hot encoded, explicit final dimension 1 must included sparse class labels.","code":""},{"path":"https://keras3.posit.co/dev/reference/train_on_batch.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Runs a single gradient update on a single batch of data. — train_on_batch","text":"scalar loss value (metrics), named list loss metric values (metrics). property model$metrics_names give display labels scalar outputs.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/use_backend.html","id":null,"dir":"Reference","previous_headings":"","what":"Configure a Keras backend — use_backend","title":"Configure a Keras backend — use_backend","text":"Configure Keras backend","code":""},{"path":"https://keras3.posit.co/dev/reference/use_backend.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Configure a Keras backend — use_backend","text":"","code":"use_backend(backend, gpu = NA)"},{"path":"https://keras3.posit.co/dev/reference/use_backend.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Configure a Keras backend — use_backend","text":"backend string, can \"tensorflow\", \"jax\", \"numpy\", \"torch\". gpu bool, whether use GPU. NA (default), attempt detect GPU availability Linux. M-series Macs, defaults FALSE TensorFlow TRUE JAX. Windows, defaults FALSE.","code":""},{"path":"https://keras3.posit.co/dev/reference/use_backend.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Configure a Keras backend — use_backend","text":"Called primarily side effects. Returns provided backend, invisibly.","code":""},{"path":"https://keras3.posit.co/dev/reference/use_backend.html","id":"details","dir":"Reference","previous_headings":"","what":"Details","title":"Configure a Keras backend — use_backend","text":"functions allow configuring backend keras use. Note one backend can configured time. function called library(keras3) calling functions within package (see example). experimental support changing backend keras initialized. using config_set_backend().","code":"library(keras3) use_backend(\"tensorflow\")"},{"path":"https://keras3.posit.co/dev/reference/with_custom_object_scope.html","id":null,"dir":"Reference","previous_headings":"","what":"Provide a scope with mappings of names to custom objects — with_custom_object_scope","title":"Provide a scope with mappings of names to custom objects — with_custom_object_scope","text":"Provide scope mappings names custom objects","code":""},{"path":"https://keras3.posit.co/dev/reference/with_custom_object_scope.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Provide a scope with mappings of names to custom objects — with_custom_object_scope","text":"","code":"with_custom_object_scope(objects, expr)"},{"path":"https://keras3.posit.co/dev/reference/with_custom_object_scope.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Provide a scope with mappings of names to custom objects — with_custom_object_scope","text":"objects Named list objects expr Expression evaluate","code":""},{"path":"https://keras3.posit.co/dev/reference/with_custom_object_scope.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Provide a scope with mappings of names to custom objects — with_custom_object_scope","text":"result evaluating expr within custom object scope.","code":""},{"path":"https://keras3.posit.co/dev/reference/with_custom_object_scope.html","id":"details","dir":"Reference","previous_headings":"","what":"Details","title":"Provide a scope with mappings of names to custom objects — with_custom_object_scope","text":"many elements Keras models can customized user objects (e.g. losses, metrics, regularizers, etc.). loading saved models use functions typically need explicitly map names user objects via custom_objects parameter. with_custom_object_scope() function provides alternative lets create named alias user object applies entire block code, automatically recognized loading saved models.","code":""},{"path":"https://keras3.posit.co/dev/reference/with_custom_object_scope.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Provide a scope with mappings of names to custom objects — with_custom_object_scope","text":"","code":"# define custom metric metric_top_3_categorical_accuracy <-   custom_metric(\"top_3_categorical_accuracy\", function(y_true, y_pred) {     metric_top_k_categorical_accuracy(y_true, y_pred, k = 3)   })  with_custom_object_scope(c(top_k_acc = sparse_top_k_cat_acc), {    # ...define model...    # compile model (refer to \"top_k_acc\" by name)   model |> compile(     loss = \"binary_crossentropy\",     optimizer = optimizer_nadam(),     metrics = c(\"top_k_acc\")   )    # save the model   model |> save_model(\"my_model.keras\")    # loading the model within the custom object scope doesn't   # require explicitly providing the custom_object   reloaded_model <- load_model(\"my_model.keras\") })"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/zip_lists.html","id":null,"dir":"Reference","previous_headings":"","what":"Zip lists — zip_lists","title":"Zip lists — zip_lists","text":"conceptually similar zip() Python, R functions purrr::transpose() data.table::transpose() (albeit, accepting elements ... instead single list), one crucial difference: provided objects named, matching done names, positions.","code":""},{"path":"https://keras3.posit.co/dev/reference/zip_lists.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Zip lists — zip_lists","text":"","code":"zip_lists(...)"},{"path":"https://keras3.posit.co/dev/reference/zip_lists.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Zip lists — zip_lists","text":"... R lists atomic vectors, optionally named.","code":""},{"path":"https://keras3.posit.co/dev/reference/zip_lists.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Zip lists — zip_lists","text":"inverted list","code":""},{"path":"https://keras3.posit.co/dev/reference/zip_lists.html","id":"details","dir":"Reference","previous_headings":"","what":"Details","title":"Zip lists — zip_lists","text":"arguments supplied must length. positional matching required, arguments provided must unnamed. matching names, arguments must set names, can different orders.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/zip_lists.html","id":"ref-examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Zip lists — zip_lists","text":"","code":"gradients <- list(\"grad_for_wt_1\", \"grad_for_wt_2\", \"grad_for_wt_3\") weights <- list(\"weight_1\", \"weight_2\", \"weight_3\") str(zip_lists(gradients, weights)) #> List of 3 #>  $ :List of 2 #>   ..$ : chr \"grad_for_wt_1\" #>   ..$ : chr \"weight_1\" #>  $ :List of 2 #>   ..$ : chr \"grad_for_wt_2\" #>   ..$ : chr \"weight_2\" #>  $ :List of 2 #>   ..$ : chr \"grad_for_wt_3\" #>   ..$ : chr \"weight_3\" str(zip_lists(gradient = gradients, weight = weights)) #> List of 3 #>  $ :List of 2 #>   ..$ gradient: chr \"grad_for_wt_1\" #>   ..$ weight  : chr \"weight_1\" #>  $ :List of 2 #>   ..$ gradient: chr \"grad_for_wt_2\" #>   ..$ weight  : chr \"weight_2\" #>  $ :List of 2 #>   ..$ gradient: chr \"grad_for_wt_3\" #>   ..$ weight  : chr \"weight_3\"  names(gradients) <- names(weights) <- paste0(\"layer_\", 1:3) str(zip_lists(gradients, weights[c(3, 1, 2)])) #> List of 3 #>  $ layer_1:List of 2 #>   ..$ : chr \"grad_for_wt_1\" #>   ..$ : chr \"weight_1\" #>  $ layer_2:List of 2 #>   ..$ : chr \"grad_for_wt_2\" #>   ..$ : chr \"weight_2\" #>  $ layer_3:List of 2 #>   ..$ : chr \"grad_for_wt_3\" #>   ..$ : chr \"weight_3\"  names(gradients) <- paste0(\"gradient_\", 1:3) try(zip_lists(gradients, weights)) # error, names don't match #> Error in zip_lists(gradients, weights) :  #>   All names of arguments provided to `zip_lists()` must match. Call `unname()` on each argument if you want positional matching # call unname directly for positional matching str(zip_lists(unname(gradients), unname(weights))) #> List of 3 #>  $ :List of 2 #>   ..$ : chr \"grad_for_wt_1\" #>   ..$ : chr \"weight_1\" #>  $ :List of 2 #>   ..$ : chr \"grad_for_wt_2\" #>   ..$ : chr \"weight_2\" #>  $ :List of 2 #>   ..$ : chr \"grad_for_wt_3\" #>   ..$ : chr \"weight_3\""},{"path":"https://keras3.posit.co/dev/news/index.html","id":"keras3-development-version","dir":"Changelog","previous_headings":"","what":"keras3 (development version)","title":"keras3 (development version)","text":"New op_subset() x@r[...] methods enable tensor subsetting using R’s [ semantics idioms. New subset assignment methods implemented tensors: op_subset(x, ...) <- value x@r[...] <- value Breaking changes: operations prefixed op_ now return 1-based indices default. following functions return consume indices changed: op_argmax(), op_argmin(), op_top_k(), op_argpartition(), op_searchsorted(), op_argsort(), op_digitize(), op_nonzero(), op_split(), op_trace(), op_swapaxes(), op_ctc_decode(), op_ctc_loss(), op_one_hot(), op_arange() op_arange() now matches semantics base::seq(). default starts, includes end value, automatically infers step direction. op_one_hot() now infers num_classes supplied factor. op_hstack() op_vstack() now accept arguments passed via .... application_decode_predictions() now returns processed data frame default decoder function predictions missing. application_preprocess_inputs() returns preprocessor function inputs missing. Various new examples added documentation, including op_scatter(), op_switch(), op_nonzero(). New x@py[...] accessor introduced Python-style 0-based indexing tensors. New Summary group generic method keras_shape, enabling usage like prod(shape(3, 4)) KERAS_HOME now set tools::R_user_dir(\"keras3\", \"cache\") ~/.keras exist KERAS_HOME unset. new op_convert_to_array() convert tensor R array. Added compatibility Keras v3.9.2. New operations added: op_rot90() op_rearrange() (Einops-style) op_signbit() op_polar() op_image_perspective_transform() op_image_gaussian_blur() New layers introduced: layer_rms_normalization() layer_aug_mix() layer_cut_mix() layer_random_invert() layer_random_erasing() layer_random_gaussian_blur() layer_random_perspective() layer_resizing() gains antialias argument. keras_input(), keras_model_sequential(), op_convert_to_tensor() gain ragged argument. layer$pop_layer() gains rebuild argument now returns removed layer. New rematerialized_call() method added Layer objects. Documentation improvements minor fixes. Fixed issue op_shape() sometimes return TensorFlow TensorShape Fixes metric_iou(), op_top_k(), op_eye() called R atomic doubles","code":""},{"path":"https://keras3.posit.co/dev/news/index.html","id":"keras3-130","dir":"Changelog","previous_headings":"","what":"keras3 1.3.0","title":"keras3 1.3.0","text":"CRAN release: 2025-03-03 Keras now uses reticulate::py_require() resolve Python dependencies. Calling install_keras() longer required (still supported). use_backend() gains gpu argument, specify GPU-capable set dependencies resolved py_require(). progress bar fit(), evaluate() predict() now defaults presenting testthat tests. dotty::. now reexported. %*% now dispatches op_matmul() tensorflow tensors, relaxed shape constraints compared tf$matmul(). Fixed issue calling Metric Loss object unnamed arguments error.","code":""},{"path":"https://keras3.posit.co/dev/news/index.html","id":"added-compatibility-with-keras-v380-user-facing-changes-1-3-0","dir":"Changelog","previous_headings":"","what":"Added compatibility with Keras v3.8.0. User-facing changes:","title":"keras3 1.3.0","text":"activation_sparse_plus() activation_sparsemax() activation_threshold() layer_equalization() layer_mix_up() layer_rand_augment() layer_random_color_degeneration() layer_random_color_jitter() layer_random_grayscale() layer_random_hue() layer_random_posterization() layer_random_saturation() layer_random_sharpness() layer_random_shear() op_diagflat() op_sparse_plus() op_sparsemax() op_threshold() op_unravel_index() Add argument axis tversky loss New: ONNX model export export_savedmodel() Doc improvements bug fixes. JAX specific changes: Add support JAX named scope TensorFlow specific changes: Make random_shuffle() XLA compilable","code":""},{"path":[]},{"path":[]},{"path":"https://keras3.posit.co/dev/news/index.html","id":"activations-1-3-0","dir":"Changelog","previous_headings":"Added compatibility with Keras v3.7.0. User-facing changes: > New functions","what":"Activations","title":"keras3 1.3.0","text":"activation_celu() activation_glu() activation_hard_shrink() activation_hard_tanh() activation_log_sigmoid() activation_soft_shrink() activation_squareplus() activation_tanh_shrink()","code":""},{"path":"https://keras3.posit.co/dev/news/index.html","id":"configuration-1-3-0","dir":"Changelog","previous_headings":"Added compatibility with Keras v3.7.0. User-facing changes: > New functions","what":"Configuration","title":"keras3 1.3.0","text":"config_disable_flash_attention() config_enable_flash_attention() config_is_flash_attention_enabled()","code":""},{"path":"https://keras3.posit.co/dev/news/index.html","id":"layers-and-initializers-1-3-0","dir":"Changelog","previous_headings":"Added compatibility with Keras v3.7.0. User-facing changes: > New functions","what":"Layers and Initializers","title":"keras3 1.3.0","text":"initializer_stft() layer_max_num_bounding_boxes() layer_stft_spectrogram()","code":""},{"path":"https://keras3.posit.co/dev/news/index.html","id":"losses-and-metrics-1-3-0","dir":"Changelog","previous_headings":"Added compatibility with Keras v3.7.0. User-facing changes: > New functions","what":"Losses and Metrics","title":"keras3 1.3.0","text":"loss_circle() metric_concordance_correlation() metric_pearson_correlation()","code":""},{"path":"https://keras3.posit.co/dev/news/index.html","id":"operations-1-3-0","dir":"Changelog","previous_headings":"Added compatibility with Keras v3.7.0. User-facing changes: > New functions","what":"Operations","title":"keras3 1.3.0","text":"op_celu() op_exp2() op_glu() op_hard_shrink() op_hard_tanh() op_ifft2() op_inner() op_soft_shrink() op_squareplus() op_tanh_shrink()","code":""},{"path":"https://keras3.posit.co/dev/news/index.html","id":"new-arguments-1-3-0","dir":"Changelog","previous_headings":"Added compatibility with Keras v3.7.0. User-facing changes: > New functions","what":"New arguments","title":"keras3 1.3.0","text":"callback_backup_and_restore(): Added double_checkpoint argument save fallback checkpoint callback_tensorboard(): Added support profile_batch argument layer_group_query_attention(): Added flash_attention seed arguments layer_multi_head_attention(): Added flash_attention argument metric_sparse_top_k_categorical_accuracy(): Added from_sorted_ids argument","code":""},{"path":"https://keras3.posit.co/dev/news/index.html","id":"performance-improvements-1-3-0","dir":"Changelog","previous_headings":"Added compatibility with Keras v3.7.0. User-facing changes:","what":"Performance improvements","title":"keras3 1.3.0","text":"Added native Flash Attention support GPU (via cuDNN) TPU (via Pallas kernel) JAX backend Added opt-native Flash Attention support GPU PyTorch backend Enabled additional kernel fusion via bias_add TensorFlow backend Added support Intel XPU devices PyTorch backend install_keras() changes: GPU available, default now install CPU build TensorFlow GPU build JAX. use GPU current session, call use_backend(\"jax\").","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/news/index.html","id":"breaking-changes-1-3-0","dir":"Changelog","previous_headings":"Added compatibility with Keras v3.6.0. User-facing changes:","what":"Breaking changes:","title":"keras3 1.3.0","text":"using get_file() extract = TRUE untar = TRUE, return value now path extracted directory, rather path archive.","code":""},{"path":"https://keras3.posit.co/dev/news/index.html","id":"other-changes-and-additions-1-3-0","dir":"Changelog","previous_headings":"Added compatibility with Keras v3.6.0. User-facing changes:","what":"Other changes and additions:","title":"keras3 1.3.0","text":"Logging now asynchronous fit(), evaluate(), predict(). enables 100% compact stacking train_step calls accelerators (e.g. running small models TPU). using custom callbacks rely on_batch_end, disable async logging. can re-enable adding self$async_safe <- TRUE callbacks. Note TensorBoard callback considered async-safe default. Default callbacks like progress bar async-safe. New bitwise operations: op_bitwise_and() op_bitwise_invert() op_bitwise_left_shift() op_bitwise_not() op_bitwise_or() op_bitwise_right_shift() op_bitwise_xor() New math operations: op_logdet() op_trunc() op_histogram() New neural network operation: op_dot_product_attention() New image preprocessing layers: layer_auto_contrast() layer_solarization() New Model functions get_state_tree() set_state_tree(), retrieving model variables, including trainable, non-trainable, optimizer variables, metric variables. New layer_pipeline() composing sequence layers. class useful building preprocessing pipeline. Compared keras_model_sequential(), layer_pipeline() key differences: ’s Model, just plain layer. layers pipeline compatible tf.data, pipeline also remain tf.data compatible, regardless backend use. New argument: export_savedmodel(verbose = ) New argument: op_normalize(epsilon = ) Various documentation improvements bug fixes.","code":""},{"path":"https://keras3.posit.co/dev/news/index.html","id":"keras3-120","dir":"Changelog","previous_headings":"","what":"keras3 1.2.0","title":"keras3 1.2.0","text":"CRAN release: 2024-09-05 Added compatibility Keras v3.5.0. User facing changes: op_associative_scan() op_searchsorted() optimizer_lamb() keras$DTypePolicy instances can now supplied dtype argument losses, metrics, layers. Add integration Hugging Face Hub. can now save models Hugging Face Hub directly save_model() load .keras models directly Hugging Face Hub load_model(). Added compatibility NumPy 2.0. Improved keras$distribution API support large models. Bug fixes performance improvements. Add data_format argument layer_zero_padding_1d() layer. Miscellaneous documentation improvements. Bug fixes performance improvements.","code":""},{"path":"https://keras3.posit.co/dev/news/index.html","id":"keras3-110","dir":"Changelog","previous_headings":"","what":"keras3 1.1.0","title":"keras3 1.1.0","text":"CRAN release: 2024-07-17 Fixed issue GPUs found running Windows WSL Linux. (reported #1456, fixed #1459) keras_shape objects (returned keras3::shape()) gain == != methods. Fixed warning tfruns::training_run() unable log optimizer learning rate. Added compatibility Keras v3.4.1 (R user facing changes). Added compatibility Keras v3.4.0. User facing changes: op_argpartition() op_map() op_scan() op_switch() op_dtype() op_lstsq() op_image_hsv_to_rgb() op_image_rgb_to_hsv() Added support arbitrary, deeply nested input/output structures Functional models (e.g. lists lists lists inputs outputs…) keras_input() gains optional argument. keras_model_sequential() gains input_optional argument. Add support float8 inference Dense EinsumDense layers. Enable layer_feature_space() used tfdatasets pipeline even backend isn’t TensorFlow. layer_string_lookup() can now take tf$SparseTensor() input. layer_string_lookup() returns \"int64\" dtype default modes now. Layer() instances gain attributes path quantization_mode. Metric()$variables now recursive. Add training argument Model$compute_loss(). split_dataset() now supports nested structures dataset. applications gain name argument, accept custom name. layer_multi_head_attention() gains seed argument. losses gain dtype argument. loss_dice() gains axis argument. op_ctc_decode(), new default mask_index = 0 op_image_* functions now use default data_format value config_image_data_format() op_isclose() gains arguments rtol, atol, equal_nan. save_model() gains argument zipped. Bugs fixes performance improvements.","code":""},{"path":"https://keras3.posit.co/dev/news/index.html","id":"keras3-100","dir":"Changelog","previous_headings":"","what":"keras3 1.0.0","title":"keras3 1.0.0","text":"CRAN release: 2024-05-21 Chains layer_* calls |> now instantiate layers order %>% pipe chains: left-hand-side first (#1440). iterate(), iter_next() as_iterator() now reexported reticulate. User facing changes upstream Keras v3.3.3: new functions: op_slogdet(), op_psnr() clone_model() gains new args: call_function, recursive Updated example usage. op_ctc_decode() strategy argument new default: \"greedy\". Updated docs. loss_ctc() default name fixed, changed \"ctc\" User facing changes upstream Keras v3.3.2: new function: op_ctc_decode() new function: op_eigh() new function: op_select() new function: op_vectorize() new function: op_image_rgb_to_grayscale() new function: loss_tversky() new args: layer_resizing(pad_to_aspect_ratio, fill_mode, fill_value) new arg: layer_embedding(weights) providing initial weights matrix new args: op_nan_to_num(nan, posinf, neginf) new args: op_image_resize(crop_to_aspect_ratio, pad_to_aspect_ratio, fill_mode, fill_value) new args: op_argmax(keepdims) op_argmin(keepdims) new arg: clear_session(free_memory) clearing without invoking garbage collector. metric_kl_divergence() loss_kl_divergence() clip inputs (y_true y_pred) [0, 1] range. new Layer() attributes: metrics, dtype_policy Added initial support float8 training layer_conv_*d() layers now support LoRa op_digitize() now supports sparse tensors. Models layers now return owned metrics recursively. Add pickling support Keras models. (e.g., via reticulate::py_save_object()) Note pickling recommended, prefer using Keras saving APIs.","code":""},{"path":"https://keras3.posit.co/dev/news/index.html","id":"keras3-020","dir":"Changelog","previous_headings":"","what":"keras3 0.2.0","title":"keras3 0.2.0","text":"CRAN release: 2024-04-18 New functions: quantize_weights(): quantize model layer weights -place. Currently, Dense, EinsumDense, Embedding layers supported (enough cover majority transformers today) layer_mel_spectrogram() layer_flax_module_wrapper() layer_jax_model_wrapper() loss_dice() random_beta() random_binomial() config_set_backend(): change backend Keras initialized. config_dtype_policy() config_set_dtype_policy() New Ops op_custom_gradient() op_batch_normalization() op_image_crop() op_divide_no_nan() op_normalize() op_correlate() ` New family linear algebra ops op_cholesky() op_det() op_eig() op_inv() op_lu_factor() op_norm() op_erfinv() op_solve_triangular() op_svd() audio_dataset_from_directory(), image_dataset_from_directory() text_dataset_from_directory() gain verbose argument (default TRUE) image_dataset_from_directory() gains pad_to_aspect_ratio argument (default FALSE) to_categorical(), op_one_hot(), fit() can now accept R factors, offset 0-based (reported #1055). op_convert_to_numpy() now returns unconverted NumPy arrays. op_array() op_convert_to_tensor() longer error casting R doubles integer types. export_savedmodel() now works Jax backend. Metric()$add_variable() method gains arg: aggregration. Layer()$add_weight() method gains args: autocast, regularizer, aggregation. op_bincount(), op_multi_hot(), op_one_hot(), layer_category_encoding() now support sparse tensors. op_custom_gradient() now supports PyTorch backend layer_lstm() layer_gru() gain arg use_cudnn, default 'auto'. Fixed issue application_preprocess_inputs() error supplied R array input. Doc improvements.","code":""},{"path":"https://keras3.posit.co/dev/news/index.html","id":"keras3-010","dir":"Changelog","previous_headings":"","what":"keras3 0.1.0","title":"keras3 0.1.0","text":"CRAN release: 2024-02-17 package rebuilt Keras 3.0. Refer https://blogs.rstudio.com/ai/posts/2024-05-21-keras3/ overview https://keras3.posit.co current --date documentation.","code":""}]
+[{"path":"https://keras3.posit.co/dev/articles/custom_train_step_in_tensorflow.html","id":"introduction","dir":"Articles","previous_headings":"","what":"Introduction","title":"Customizing what happens in `fit()` with TensorFlow","text":"’re supervised learning, can use fit() everything works smoothly. need take control every little detail, can write training loop entirely scratch. need custom training algorithm, still want benefit convenient features fit(), callbacks, built-distribution support, step fusing? core principle Keras progressive disclosure complexity. always able get lower-level workflows gradual way. shouldn’t fall cliff high-level functionality doesn’t exactly match use case. able gain control small details retaining commensurate amount high-level convenience. need customize fit() , override training step function Model class. function called fit() every batch data. able call fit() usual – running learning algorithm. Note pattern prevent building models Functional API. can whether ’re building Sequential models, Functional API models, subclassed models. Let’s see works.","code":""},{"path":"https://keras3.posit.co/dev/articles/custom_train_step_in_tensorflow.html","id":"setup","dir":"Articles","previous_headings":"","what":"Setup","title":"Customizing what happens in `fit()` with TensorFlow","text":"","code":"library(reticulate) library(tensorflow, exclude = c(\"set_random_seed\", \"shape\")) library(keras3)"},{"path":"https://keras3.posit.co/dev/articles/custom_train_step_in_tensorflow.html","id":"a-first-simple-example","dir":"Articles","previous_headings":"","what":"A first simple example","title":"Customizing what happens in `fit()` with TensorFlow","text":"Let’s start simple example: create new class subclasses Model. just override method train_step(self, data). return dictionary mapping metric names (including loss) current value. input argument data gets passed fit training data: pass arrays, calling fit(x, y, ...), data list (x, y) pass tf_dataset, calling fit(dataset, ...), data gets yielded dataset batch. body train_step() method, implement regular training update, similar already familiar . Importantly, compute loss via self.compute_loss(), wraps loss(es) function(s) passed compile(). Similarly, call metric$update_state(y, y_pred) metrics self$metrics, update state metrics passed compile(), query results self$metrics end retrieve current value. Let’s try :","code":"CustomModel <- new_model_class(   \"CustomModel\",   train_step = function(data) {     c(x, y = NULL, sample_weight = NULL) %<-% data      with(tf$GradientTape() %as% tape, {       y_pred <- self(x, training = TRUE)       loss <- self$compute_loss(y = y, y_pred = y_pred,                                 sample_weight = sample_weight)     })      # Compute gradients     trainable_vars <- self$trainable_variables     gradients <- tape$gradient(loss, trainable_vars)      # Update weights     self$optimizer$apply(gradients, trainable_vars)      # Update metrics (includes the metric that tracks the loss)     for (metric in self$metrics) {       if (metric$name == \"loss\")         metric$update_state(loss)       else         metric$update_state(y, y_pred)     }      # Return a dict mapping metric names to current value     metrics <- lapply(self$metrics, function(m) m$result())     metrics <- setNames(metrics, sapply(self$metrics, function(m) m$name))     metrics   } ) # Construct and compile an instance of CustomModel inputs <- keras_input(shape = 32) outputs <- layer_dense(inputs, 1) model <- CustomModel(inputs, outputs) model |> compile(optimizer = \"adam\", loss = \"mse\", metrics = \"mae\")  # Just use `fit` as usual x <- random_normal(c(1000, 32)) y <- random_normal(c(1000, 1)) model |> fit(x, y, epochs = 3) ## Epoch 1/3 ## 32/32 - 1s - 20ms/step - mae: 1.4339 - loss: 3.2271 ## Epoch 2/3 ## 32/32 - 0s - 3ms/step - mae: 1.3605 - loss: 2.9034 ## Epoch 3/3 ## 32/32 - 0s - 2ms/step - mae: 1.2960 - loss: 2.6272"},{"path":"https://keras3.posit.co/dev/articles/custom_train_step_in_tensorflow.html","id":"going-lower-level","dir":"Articles","previous_headings":"","what":"Going lower-level","title":"Customizing what happens in `fit()` with TensorFlow","text":"Naturally, just skip passing loss function compile(), instead everything manually train_step. Likewise metrics. ’s lower-level example, uses compile() configure optimizer: start creating Metric instances track loss MAE score (__init__()). implement custom train_step() updates state metrics (calling update_state() ), query (via result()) return current average value, displayed progress bar pass callback. Note need call reset_states() metrics epoch! Otherwise calling result() return average since start training, whereas usually work per-epoch averages. Thankfully, framework can us: just list metric want reset metrics property model. model call reset_states() object listed beginning fit() epoch beginning call evaluate().","code":"CustomModel <- new_model_class(   \"CustomModel\",   initialize = function(...) {     super$initialize(...)     self$loss_tracker <- metric_mean(name = \"loss\")     self$mae_metric <- metric_mean_absolute_error(name = \"mae\")     self$loss_fn <- loss_mean_squared_error()   },   train_step = function(data) {     c(x, y = NULL, sample_weight = NULL) %<-% data      with(tf$GradientTape() %as% tape, {       y_pred <- self(x, training = TRUE)       loss <- self$loss_fn(y, y_pred, sample_weight = sample_weight)     })      # Compute gradients     trainable_vars <- self$trainable_variables     gradients <- tape$gradient(loss, trainable_vars)      # Update weights     self$optimizer$apply(gradients, trainable_vars)      # Compute our own metrics     self$loss_tracker$update_state(loss)     self$mae_metric$update_state(y, y_pred)      # Return a dict mapping metric names to current value     list(       loss = self$loss_tracker$result(),       mae = self$mae_metric$result()     )   },   metrics = mark_active(function() {     # We list our `Metric` objects here so that `reset_states()` can be     # called automatically at the start of each epoch     # or at the start of `evaluate()`.     list(self$loss_tracker, self$mae_metric)   }) )   # Construct and compile an instance of CustomModel inputs <- keras_input(shape = 32) outputs <- layer_dense(inputs, 1) model <- CustomModel(inputs, outputs)  # We don't pass a loss or metrics here. model |> compile(optimizer = \"adam\")  # Just use `fit` as usual x <- random_normal(c(1000, 32)) y <- random_normal(c(1000, 1)) model |> fit(x, y, epochs = 3) ## Epoch 1/3 ## 32/32 - 1s - 18ms/step - loss: 2.5170 - mae: 1.2923 ## Epoch 2/3 ## 32/32 - 0s - 2ms/step - loss: 2.2689 - mae: 1.2241 ## Epoch 3/3 ## 32/32 - 0s - 2ms/step - loss: 2.0578 - mae: 1.1633"},{"path":"https://keras3.posit.co/dev/articles/custom_train_step_in_tensorflow.html","id":"supporting-sample_weight-class_weight","dir":"Articles","previous_headings":"","what":"Supporting sample_weight & class_weight","title":"Customizing what happens in `fit()` with TensorFlow","text":"may noticed first basic example didn’t make mention sample weighting. want support fit() arguments sample_weight class_weight, ’d simply following: Unpack sample_weight data argument Pass compute_loss & update_state (course, also just apply manually don’t rely compile() losses & metrics) ’s .","code":"CustomModel <- new_model_class(   \"CustomModel\",   train_step = function(data) {     c(x, y = NULL, sample_weight = NULL) %<-% data      with(tf$GradientTape() %as% tape, {       y_pred <- self(x, training = TRUE)       loss <- self$compute_loss(y = y, y_pred = y_pred,                                 sample_weight = sample_weight)     })      # Compute gradients     trainable_vars <- self$trainable_variables     gradients <- tape$gradient(loss, trainable_vars)      # Update weights     self$optimizer$apply_gradients(zip_lists(gradients, trainable_vars))      # Update metrics (includes the metric that tracks the loss)     for (metric in self$metrics) {       if (metric$name == \"loss\") {         metric$update_state(loss)       } else {         metric$update_state(y, y_pred, sample_weight = sample_weight)       }     }      # Return a dict mapping metric names to current value     metrics <- lapply(self$metrics, function(m) m$result())     metrics <- setNames(metrics, sapply(self$metrics, function(m) m$name))     metrics   } )   # Construct and compile an instance of CustomModel inputs <- keras_input(shape = 32) outputs <- layer_dense(inputs, units = 1) model <- CustomModel(inputs, outputs) model |> compile(optimizer = \"adam\", loss = \"mse\", metrics = \"mae\")  # You can now use sample_weight argument x <- random_normal(c(1000, 32)) y <- random_normal(c(1000, 1)) sw <- random_normal(c(1000, 1)) model |> fit(x, y, sample_weight = sw, epochs = 3) ## Epoch 1/3 ## 32/32 - 1s - 20ms/step - mae: 1.3434 - loss: 0.1681 ## Epoch 2/3 ## 32/32 - 0s - 2ms/step - mae: 1.3364 - loss: 0.1394 ## Epoch 3/3 ## 32/32 - 0s - 2ms/step - mae: 1.3286 - loss: 0.1148"},{"path":"https://keras3.posit.co/dev/articles/custom_train_step_in_tensorflow.html","id":"providing-your-own-evaluation-step","dir":"Articles","previous_headings":"","what":"Providing your own evaluation step","title":"Customizing what happens in `fit()` with TensorFlow","text":"want calls model.evaluate()? override test_step exactly way. ’s looks like:","code":"CustomModel <- new_model_class(   \"CustomModel\",   test_step = function(data) {     # Unpack the data     c(x, y = NULL, sw = NULL) %<-% data     # Compute predictions     y_pred = self(x, training = FALSE)     # Updates the metrics tracking the loss     self$compute_loss(y = y, y_pred = y_pred, sample_weight = sw)     # Update the metrics.     for (metric in self$metrics) {       if (metric$name != \"loss\") {         metric$update_state(y, y_pred, sample_weight = sw)       }     }     # Return a dict mapping metric names to current value.     # Note that it will include the loss (tracked in self.metrics).     metrics <- lapply(self$metrics, function(m) m$result())     metrics <- setNames(metrics, sapply(self$metrics, function(m) m$name))     metrics   } )  # Construct an instance of CustomModel inputs <- keras_input(shape = 32) outputs <- layer_dense(inputs, 1) model <- CustomModel(inputs, outputs) model |> compile(loss = \"mse\", metrics = \"mae\")  # Evaluate with our custom test_step x <- random_normal(c(1000, 32)) y <- random_normal(c(1000, 1)) model |> evaluate(x, y) ## 32/32 - 0s - 10ms/step - mae: 1.3871 - loss: 0.0000e+00 ## $loss ## tf.Tensor(0.0, shape=(), dtype=float32) ## ## $compile_metrics ## $compile_metrics$mae ## tf.Tensor(1.3871489, shape=(), dtype=float32)"},{"path":"https://keras3.posit.co/dev/articles/custom_train_step_in_tensorflow.html","id":"wrapping-up-an-end-to-end-gan-example","dir":"Articles","previous_headings":"","what":"Wrapping up: an end-to-end GAN example","title":"Customizing what happens in `fit()` with TensorFlow","text":"Let’s walk end--end example leverages everything just learned. Let’s consider: generator network meant generate 28x28x1 images. discriminator network meant classify 28x28x1 images two classes (“fake” “real”). One optimizer . loss function train discriminator. ’s feature-complete GAN class, overriding compile() use signature, implementing entire GAN algorithm 17 lines train_step: Let’s test-drive : ideas behind deep learning simple, implementation painful?","code":"# Create the discriminator discriminator <-   keras_model_sequential(name = \"discriminator\", input_shape = c(28, 28, 1)) |>    layer_conv_2d(filters = 64, kernel_size = c(3, 3),                 strides = c(2, 2),  padding = \"same\") |>   layer_activation_leaky_relu(negative_slope = 0.2) |>    layer_conv_2d(filters = 128, kernel_size = c(3, 3),                 strides = c(2, 2), padding = \"same\") |>   layer_activation_leaky_relu(negative_slope = 0.2) |>    layer_global_max_pooling_2d() |>   layer_dense(units = 1)   # Create the generator latent_dim <- 128 generator <-   keras_model_sequential(name = \"generator\", input_shape = latent_dim) |>    layer_dense(7 * 7 * 128) |>   layer_activation_leaky_relu(negative_slope = 0.2) |>    layer_reshape(target_shape = c(7, 7, 128)) |>    layer_conv_2d_transpose(filters = 128, kernel_size = c(4, 4),                           strides = c(2, 2), padding = \"same\") |>   layer_activation_leaky_relu(negative_slope = 0.2) |>    layer_conv_2d_transpose(filters = 128, kernel_size = c(4, 4),                           strides = c(2, 2), padding = \"same\") |>   layer_activation_leaky_relu(negative_slope = 0.2) |>    layer_conv_2d(filters = 1, kernel_size = c(7, 7), padding = \"same\",                 activation = \"sigmoid\") GAN <- Model(   classname = \"GAN\",    initialize = function(discriminator, generator, latent_dim, ...) {     super$initialize(...)     self$discriminator <- discriminator     self$generator <- generator     self$latent_dim <- as.integer(latent_dim)     self$d_loss_tracker <- metric_mean(name = \"d_loss\")     self$g_loss_tracker <- metric_mean(name = \"g_loss\")   },    compile = function(d_optimizer, g_optimizer, loss_fn, ...) {     super$compile(...)     self$d_optimizer <- d_optimizer     self$g_optimizer <- g_optimizer     self$loss_fn <- loss_fn   },   metrics = active_property(function() {     list(self$d_loss_tracker, self$g_loss_tracker)   }),    train_step = function(real_images) {      # Sample random points in the latent space     batch_size <- shape(real_images)[[1]]     random_latent_vectors <-       tf$random$normal(shape(batch_size, self$latent_dim))      # Decode them to fake images     generated_images <- self$generator(random_latent_vectors)      # Combine them with real images     combined_images <- op_concatenate(list(generated_images,                                            real_images))      # Assemble labels discriminating real from fake images     labels <- op_concatenate(list(op_ones(c(batch_size, 1)),                                   op_zeros(c(batch_size, 1))))      # Add random noise to the labels - important trick!     labels %<>% `+`(tf$random$uniform(shape(.), maxval = 0.05))      # Train the discriminator     with(tf$GradientTape() %as% tape, {       predictions <- self$discriminator(combined_images)       d_loss <- self$loss_fn(labels, predictions)     })     grads <- tape$gradient(d_loss, self$discriminator$trainable_weights)     self$d_optimizer$apply_gradients(       zip_lists(grads, self$discriminator$trainable_weights))      # Sample random points in the latent space     random_latent_vectors <-       tf$random$normal(shape(batch_size, self$latent_dim))      # Assemble labels that say \"all real images\"     misleading_labels <- op_zeros(c(batch_size, 1))      # Train the generator (note that we should *not* update the weights     # of the discriminator)!     with(tf$GradientTape() %as% tape, {       predictions <- self$discriminator(self$generator(random_latent_vectors))       g_loss <- self$loss_fn(misleading_labels, predictions)     })     grads <- tape$gradient(g_loss, self$generator$trainable_weights)     self$g_optimizer$apply_gradients(       zip_lists(grads, self$generator$trainable_weights))      list(d_loss = d_loss, g_loss = g_loss)   } ) batch_size <- 64 c(c(x_train, .), c(x_test, .)) %<-% dataset_mnist() all_digits <- op_concatenate(list(x_train, x_test)) all_digits <- op_reshape(all_digits, c(-1, 28, 28, 1)) dataset <- all_digits |>   tfdatasets::tensor_slices_dataset() |>   tfdatasets::dataset_map(\\(x) op_cast(x, \"float32\") / 255) |>   tfdatasets::dataset_shuffle(buffer_size = 1024) |>   tfdatasets::dataset_batch(batch_size = batch_size)  gan <- GAN(discriminator = discriminator,            generator = generator,            latent_dim = latent_dim)  gan |> compile(   d_optimizer = optimizer_adam(learning_rate = 0.0003),   g_optimizer = optimizer_adam(learning_rate = 0.0003),   loss_fn = loss_binary_crossentropy(from_logits = TRUE) )  # To limit the execution time, we only train on 100 batches. You can train on # the entire dataset. You will need about 20 epochs to get nice results. gan |> fit(   tfdatasets::dataset_take(dataset, 100),   epochs = 1 ) ## 100/100 - 5s - 51ms/step - d_loss: 0.0000e+00 - g_loss: 0.0000e+00"},{"path":"https://keras3.posit.co/dev/articles/distributed_training_with_tensorflow.html","id":"introduction","dir":"Articles","previous_headings":"","what":"Introduction","title":"Multi-GPU distributed training with TensorFlow","text":"generally two ways distribute computation across multiple devices: Data parallelism, single model gets replicated multiple devices multiple machines. processes different batches data, merge results. exist many variants setup, differ different model replicas merge results, whether stay sync every batch whether loosely coupled, etc. Model parallelism, different parts single model run different devices, processing single batch data together. works best models naturally-parallel architecture, models feature multiple branches. guide focuses data parallelism, particular synchronous data parallelism, different replicas model stay sync batch process. Synchronicity keeps model convergence behavior identical see single-device training. Specifically, guide teaches use tf.distribute API train Keras models multiple GPUs, minimal changes code, multiple GPUs (typically 2 16) installed single machine (single host, multi-device training). common setup researchers small-scale industry workflows.","code":""},{"path":"https://keras3.posit.co/dev/articles/distributed_training_with_tensorflow.html","id":"setup","dir":"Articles","previous_headings":"","what":"Setup","title":"Multi-GPU distributed training with TensorFlow","text":"","code":"library(keras3) library(tensorflow, exclude = c(\"shape\", \"set_random_seed\")) library(tfdatasets, exclude = \"shape\")"},{"path":"https://keras3.posit.co/dev/articles/distributed_training_with_tensorflow.html","id":"single-host-multi-device-synchronous-training","dir":"Articles","previous_headings":"","what":"Single-host, multi-device synchronous training","title":"Multi-GPU distributed training with TensorFlow","text":"setup, one machine several GPUs (typically 2 16). device run copy model (called replica). simplicity, follows, ’ll assume ’re dealing 8 GPUs, loss generality. works step training: current batch data (called global batch) split 8 different sub-batches (called local batches). instance, global batch 512 samples, 8 local batches 64 samples. 8 replicas independently processes local batch: run forward pass, backward pass, outputting gradient weights respect loss model local batch. weight updates originating local gradients efficiently merged across 8 replicas. done end every step, replicas always stay sync. practice, process synchronously updating weights model replicas handled level individual weight variable. done mirrored variable object. use single-host, multi-device synchronous training Keras model, use tf$distribute$MirroredStrategy API. ’s works: Instantiate MirroredStrategy, optionally configuring specific devices want use (default strategy use GPUs available). Use strategy object open scope, within scope, create Keras objects need contain variables. Typically, means creating & compiling model inside distribution scope. cases, first call fit() may also create variables, ’s good idea put fit() call scope well. Train model via fit() usual. Importantly, recommend use tf.data.Dataset objects load data multi-device distributed workflow. Schematically, looks like : ’s simple end--end runnable example:","code":"# Create a MirroredStrategy. strategy <- tf$distribute$MirroredStrategy() cat(sprintf('Number of devices: %d\\n', strategy$num_replicas_in_sync))  # Open a strategy scope. with(startegy$scope(), {   # Everything that creates variables should be under the strategy scope.   # In general this is only model construction & `compile()`.   model <- Model(...)   model |> compile(...)    # Train the model on all available devices.   model |> fit(train_dataset, validation_data=val_dataset, ...)    # Test the model on all available devices.   model |> evaluate(test_dataset) }) get_compiled_model <- function() {   inputs <- keras_input(shape = 784)   outputs <- inputs |>     layer_dense(units = 256, activation = \"relu\") |>     layer_dense(units = 256, activation = \"relu\") |>     layer_dense(units = 10)   model <- keras_model(inputs, outputs)   model |> compile(     optimizer = optimizer_adam(),     loss = loss_sparse_categorical_crossentropy(from_logits = TRUE),     metrics = list(metric_sparse_categorical_accuracy())   )   model }  get_dataset <- function(batch_size = 64) {    c(c(x_train, y_train), c(x_test, y_test)) %<-% dataset_mnist()   x_train <- array_reshape(x_train, c(-1, 784)) / 255   x_test <- array_reshape(x_test, c(-1, 784)) / 255    # Reserve 10,000 samples for validation.   val_i <- sample.int(nrow(x_train), 10000)   x_val <- x_train[val_i,]   y_val <- y_train[val_i]   x_train = x_train[-val_i,]   y_train = y_train[-val_i]    y_train <- array_reshape(y_train, c(-1, 1))   y_val <- array_reshape(y_val, c(-1, 1))   y_test <- array_reshape(y_test, c(-1, 1))    # Prepare the training dataset.   train_dataset <- list(x_train, y_train) |>     lapply(np_array, \"float32\") |>     tensor_slices_dataset() |>     dataset_batch(batch_size)    # Prepare the validation dataset.   val_dataset <- list(x_val, y_val) |>     lapply(np_array, \"float32\") |>     tensor_slices_dataset() |>     dataset_batch(batch_size)    # Prepare the test dataset.   test_dataset <- list(x_test, y_test) |>     lapply(np_array, \"float32\") |>     tensor_slices_dataset() |>     dataset_batch(batch_size)    list(train_dataset, val_dataset, test_dataset) }  # Create a MirroredStrategy. strategy <- tf$distribute$MirroredStrategy() cat(sprintf('Number of devices: %d\\n', strategy$num_replicas_in_sync)) ## Number of devices: 2 # Open a strategy scope. with(strategy$scope(), {   # Everything that creates variables should be under the strategy scope.   # In general this is only model construction & `compile()`.   model <- get_compiled_model()    c(train_dataset, val_dataset, test_dataset) %<-% get_dataset()    # Train the model on all available devices.   model |> fit(train_dataset, epochs = 2, validation_data = val_dataset)    # Test the model on all available devices.   model |> evaluate(test_dataset)  }) ## Epoch 1/2 ## 782/782 - 6s - 7ms/step - loss: 0.2616 - sparse_categorical_accuracy: 0.9232 - val_loss: 0.1510 - val_sparse_categorical_accuracy: 0.9524 ## Epoch 2/2 ## 782/782 - 4s - 5ms/step - loss: 0.1044 - sparse_categorical_accuracy: 0.9682 - val_loss: 0.1004 - val_sparse_categorical_accuracy: 0.9675 ## 157/157 - 0s - 3ms/step - loss: 0.1004 - sparse_categorical_accuracy: 0.9683 ## $loss ## [1] 0.1004203 ## ## $sparse_categorical_accuracy ## [1] 0.9683"},{"path":"https://keras3.posit.co/dev/articles/distributed_training_with_tensorflow.html","id":"using-callbacks-to-ensure-fault-tolerance","dir":"Articles","previous_headings":"","what":"Using callbacks to ensure fault tolerance","title":"Multi-GPU distributed training with TensorFlow","text":"using distributed training, always make sure strategy recover failure (fault tolerance). simplest way handle pass ModelCheckpoint callback fit(), save model regular intervals (e.g. every 100 batches every epoch). can restart training saved model. ’s simple example:","code":"# Prepare a directory to store all the checkpoints. checkpoint_dir <- \"./ckpt\" if (!dir.exists(checkpoint_dir)) {   dir.create(checkpoint_dir) }  make_or_restore_model <- function() {   # Either restore the latest model, or create a fresh one   # if there is no checkpoint available.   checkpoints <- list.files(checkpoint_dir, pattern = \"ckpt-.*\\\\.keras\",                             full.names = TRUE)    if (length(checkpoints) > 0) {     checkpoint_epochs <- as.integer(sub(\"ckpt-([0-9]+)\\\\.keras\", \"\\\\1\",                                         basename(checkpoints)))     latest_checkpoint <- checkpoints[which.max(checkpoint_epochs)]     load_model(latest_checkpoint)   } else {     get_compiled_model()   } }    run_training <- function(epochs = 1) {   # Create a MirroredStrategy.   strategy <- tf$distribute$MirroredStrategy()    # Open a strategy scope and create/restore the model   with(strategy$scope(), {     model <- make_or_restore_model()      callbacks <- list(       # This callback saves a SavedModel every epoch       # We include the current epoch in the folder name.       callback_model_checkpoint(         filepath = paste0(checkpoint_dir, \"/ckpt-{epoch}.keras\"),         save_freq = \"epoch\"       ))      model |> fit(       train_dataset,       epochs = epochs,       callbacks = callbacks,       validation_data = val_dataset,       verbose = 2     )   }) }  # Running the first time creates the model run_training(epochs = 1) ## 782/782 - 5s - 6ms/step - loss: 0.2575 - sparse_categorical_accuracy: 0.9237 - val_loss: 0.1428 - val_sparse_categorical_accuracy: 0.9537 # Calling the same function again will resume from where we left off run_training(epochs = 1) ## 782/782 - 5s - 6ms/step - loss: 0.1011 - sparse_categorical_accuracy: 0.9693 - val_loss: 0.1185 - val_sparse_categorical_accuracy: 0.9602"},{"path":"https://keras3.posit.co/dev/articles/distributed_training_with_tensorflow.html","id":"tfdata-performance-tips","dir":"Articles","previous_headings":"","what":"tf$data performance tips","title":"Multi-GPU distributed training with TensorFlow","text":"distributed training, efficiency load data can often become critical. tips make sure tf$data pipelines run fast possible. Note dataset batching creating dataset, make sure batched global batch size. instance, 8 GPUs capable running batch 64 samples, call use global batch size 512. Calling dataset_cache() call dataset_cache() dataset, data cached running first iteration data. Every subsequent iteration use cached data. cache can memory (default) local file specify. can improve performance : data expected change iteration iteration reading data remote distributed filesystem reading data local disk, data fit memory workflow significantly IO-bound (e.g. reading & decoding image files). Calling dataset_prefetch(buffer_size) almost always call dataset_prefetch(buffer_size) creating dataset. means data pipeline run asynchronously model, new samples preprocessed stored buffer current batch samples used train model. next batch prefetched GPU memory time current batch . ’s !","code":""},{"path":"https://keras3.posit.co/dev/articles/distribution.html","id":"introduction","dir":"Articles","previous_headings":"","what":"Introduction","title":"Distributed training with Keras 3","text":"Keras distribution API new interface designed facilitate distributed deep learning across variety backends like JAX, TensorFlow PyTorch. powerful API introduces suite tools enabling data model parallelism, allowing efficient scaling deep learning models multiple accelerators hosts. Whether leveraging power GPUs TPUs, API provides streamlined approach initializing distributed environments, defining device meshes, orchestrating layout tensors across computational resources. classes like DataParallel ModelParallel, abstracts complexity involved parallel computation, making easier developers accelerate machine learning workflows.","code":""},{"path":"https://keras3.posit.co/dev/articles/distribution.html","id":"how-it-works","dir":"Articles","previous_headings":"","what":"How it works","title":"Distributed training with Keras 3","text":"Keras distribution API provides global programming model allows developers compose applications operate tensors global context (working single device) automatically managing distribution across many devices. API leverages underlying framework (e.g. JAX) distribute program tensors according sharding directives procedure called single program, multiple data (SPMD) expansion. decoupling application sharding directives, API enables running application single device, multiple devices, even multiple clients, preserving global semantics.","code":""},{"path":"https://keras3.posit.co/dev/articles/distribution.html","id":"setup","dir":"Articles","previous_headings":"","what":"Setup","title":"Distributed training with Keras 3","text":"","code":"# This guide assumes there are 8 GPUs available for testing. If you don't have # 8 gpus available locally, you can set the following envvar to # make xla initialize the CPU as 8 devices, to enable local testing Sys.setenv(\"CUDA_VISIBLE_DEVICES\" = \"\") Sys.setenv(\"XLA_FLAGS\" = \"--xla_force_host_platform_device_count=8\") library(keras3)  # The distribution API is only implemented for the JAX backend for now. use_backend(\"jax\", FALSE) jax <- reticulate::import(\"jax\")  library(tfdatasets, exclude = \"shape\") # For dataset input."},{"path":"https://keras3.posit.co/dev/articles/distribution.html","id":"devicemesh-and-tensorlayout","dir":"Articles","previous_headings":"","what":"DeviceMesh and TensorLayout","title":"Distributed training with Keras 3","text":"keras$distribution$DeviceMesh class Keras distribution API represents cluster computational devices configured distributed computation. aligns similar concepts jax.sharding.Mesh tf.dtensor.Mesh, ’s used map physical devices logical mesh structure. TensorLayout class specifies tensors distributed across DeviceMesh, detailing sharding tensors along specified axes correspond names axes DeviceMesh. can find detailed concept explainers TensorFlow DTensor guide.","code":"# Retrieve the local available gpu devices. devices <- jax$devices() # \"gpu\" str(devices) ## List of 8 ##  $ :TFRT_CPU_0 ##  $ :TFRT_CPU_1 ##  $ :TFRT_CPU_2 ##  $ :TFRT_CPU_3 ##  $ :TFRT_CPU_4 ##  $ :TFRT_CPU_5 ##  $ :TFRT_CPU_6 ##  $ :TFRT_CPU_7 # Define a 2x4 device mesh with data and model parallel axes mesh <- keras$distribution$DeviceMesh(   shape = shape(2, 4),   axis_names = list(\"data\", \"model\"),   devices = devices )  # A 2D layout, which describes how a tensor is distributed across the # mesh. The layout can be visualized as a 2D grid with \"model\" as rows and # \"data\" as columns, and it is a [4, 2] grid when it mapped to the physical # devices on the mesh. layout_2d <- keras$distribution$TensorLayout(   axes = c(\"model\", \"data\"),   device_mesh = mesh )  # A 4D layout which could be used for data parallelism of an image input. replicated_layout_4d <- keras$distribution$TensorLayout(   axes = list(\"data\", NULL, NULL, NULL),   device_mesh = mesh )"},{"path":"https://keras3.posit.co/dev/articles/distribution.html","id":"distribution","dir":"Articles","previous_headings":"","what":"Distribution","title":"Distributed training with Keras 3","text":"Distribution class Keras serves foundational abstract class designed developing custom distribution strategies. encapsulates core logic needed distribute model’s variables, input data, intermediate computations across device mesh. end user, won’t interact directly class, subclasses like DataParallel ModelParallel.","code":""},{"path":"https://keras3.posit.co/dev/articles/distribution.html","id":"dataparallel","dir":"Articles","previous_headings":"","what":"DataParallel","title":"Distributed training with Keras 3","text":"DataParallel class Keras distribution API designed data parallelism strategy distributed training, model weights replicated across devices DeviceMesh, device processes portion input data. sample usage class.","code":"# Create DataParallel with list of devices. # As a shortcut, the devices can be skipped, # and Keras will detect all local available devices. # E.g. data_parallel <- DataParallel() data_parallel <- keras$distribution$DataParallel(devices = devices)  # Or you can choose to create DataParallel with a 1D `DeviceMesh`. mesh_1d <- keras$distribution$DeviceMesh(   shape = shape(8),   axis_names = list(\"data\"),   devices = devices ) data_parallel <- keras$distribution$DataParallel(device_mesh = mesh_1d)  inputs <- random_normal(c(128, 28, 28, 1)) labels <- random_normal(c(128, 10)) dataset <- tensor_slices_dataset(c(inputs, labels)) |>   dataset_batch(16)  # Set the global distribution. keras$distribution$set_distribution(data_parallel)  # Note that all the model weights from here on are replicated to # all the devices of the `DeviceMesh`. This includes the RNG # state, optimizer states, metrics, etc. The dataset fed into `model |> fit()` or # `model |> evaluate()` will be split evenly on the batch dimension, and sent to # all the devices. You don't have to do any manual aggregation of losses, # since all the computation happens in a global context. inputs <- keras_input(shape = c(28, 28, 1)) outputs <- inputs |>   layer_flatten() |>   layer_dense(units = 200, use_bias = FALSE, activation = \"relu\") |>   layer_dropout(0.4) |>   layer_dense(units = 10, activation = \"softmax\")  model <- keras_model(inputs = inputs, outputs = outputs)  model |> compile(loss = \"mse\") model |> fit(dataset, epochs = 3) ## Epoch 1/3 ## 8/8 - 0s - 42ms/step - loss: 1.1575 ## Epoch 2/3 ## 8/8 - 0s - 4ms/step - loss: 1.0557 ## Epoch 3/3 ## 8/8 - 0s - 4ms/step - loss: 1.0179 model |> evaluate(dataset) ## 8/8 - 0s - 9ms/step - loss: 0.9707 ## $loss ## [1] 0.9707004"},{"path":"https://keras3.posit.co/dev/articles/distribution.html","id":"modelparallel-and-layoutmap","dir":"Articles","previous_headings":"","what":"ModelParallel and LayoutMap","title":"Distributed training with Keras 3","text":"ModelParallel mostly useful model weights large fit single accelerator. setting allows spit model weights activation tensors across devices DeviceMesh, enable horizontal scaling large models. Unlike DataParallel model weights fully replicated, weights layout ModelParallel usually need customization best performances. introduce LayoutMap let specify TensorLayout weights intermediate tensors global perspective. LayoutMap dict-like object maps string TensorLayout instances. behaves differently normal dict string key treated regex retrieving value. class allows define naming schema TensorLayout retrieve corresponding TensorLayout instance. Typically, key used query variable$path attribute, identifier variable. shortcut, list axis names also allowed inserting value, converted TensorLayout. LayoutMap can also optionally contain DeviceMesh populate TensorLayout$device_mesh set. retrieving layout key, isn’t exact match, existing keys layout map treated regex matched input key . multiple matches, ValueError raised. matches found, NULL returned. can visualize individual weights sharded also easy change mesh structure tune computation data parallel model parallel. can adjusting shape mesh. changes needed code.","code":"mesh_2d <- keras$distribution$DeviceMesh(   shape = shape(2, 4),   axis_names = c(\"data\", \"model\"),   devices = devices ) layout_map  <- keras$distribution$LayoutMap(mesh_2d)  # The rule below means that for any weights that match with d1/kernel, it # will be sharded with model dimensions (4 devices), same for the d1/bias. # All other weights will be fully replicated. layout_map[\"d1/kernel\"] <- tuple(NULL, \"model\") layout_map[\"d1/bias\"] <- tuple(\"model\")  # You can also set the layout for the layer output like layout_map[\"d2/output\"] <- tuple(\"data\", NULL)  model_parallel <- keras$distribution$ModelParallel(   layout_map = layout_map, batch_dim_name = \"data\" )  keras$distribution$set_distribution(model_parallel)  inputs <- layer_input(shape = c(28, 28, 1)) outputs <- inputs |>   layer_flatten() |>   layer_dense(units = 200, use_bias = FALSE,               activation = \"relu\", name = \"d1\") |>   layer_dropout(0.4) |>   layer_dense(units = 10,               activation = \"softmax\",               name = \"d2\")  model <- keras_model(inputs = inputs, outputs = outputs) d1 <- get_layer(model, \"d1\") d1$kernel$value |> jax$debug$visualize_array_sharding() ## ┌───────┬───────┬───────┬───────┐ ## │       │       │       │       │ ## │       │       │       │       │ ## │       │       │       │       │ ## │       │       │       │       │ ## │CPU 0,4│CPU 1,5│CPU 2,6│CPU 3,7│ ## │       │       │       │       │ ## │       │       │       │       │ ## │       │       │       │       │ ## │       │       │       │       │ ## └───────┴───────┴───────┴───────┘ d2 <- get_layer(model, \"d2\") d2$kernel$value |> jax$debug$visualize_array_sharding() ## ┌───────────────────┐ ## │                   │ ## │                   │ ## │                   │ ## │                   │ ## │CPU 0,1,2,3,4,5,6,7│ ## │                   │ ## │                   │ ## │                   │ ## │                   │ ## └───────────────────┘ d2$bias$value |> jax$debug$visualize_array_sharding() ## ┌───────────────────┐ ## │CPU 0,1,2,3,4,5,6,7│ ## └───────────────────┘ x_batch <- dataset |>   as_iterator() |> iter_next() |>   _[[1]] |> op_convert_to_tensor()  output_array <- model(x_batch) output_array |> jax$debug$visualize_array_sharding() ## ┌─────────────┐ ## │             │ ## │ CPU 0,1,2,3 │ ## │             │ ## │             │ ## ├─────────────┤ ## │             │ ## │ CPU 4,5,6,7 │ ## │             │ ## │             │ ## └─────────────┘ # The data will be sharded across the \"data\" dimension of the method, which # has 2 devices. model |> compile(loss = \"mse\") model |> fit(dataset, epochs = 3) ## Epoch 1/3 ## 8/8 - 0s - 48ms/step - loss: 1.1454 ## Epoch 2/3 ## 8/8 - 0s - 2ms/step - loss: 1.0527 ## Epoch 3/3 ## 8/8 - 0s - 2ms/step - loss: 1.0115 model |> evaluate(dataset) ## 8/8 - 0s - 10ms/step - loss: 0.9674 ## $loss ## [1] 0.9673891 full_data_parallel_mesh <- keras$distribution$DeviceMesh(   shape = shape(8, 1),   axis_names = list(\"data\", \"model\"),   devices = devices ) more_data_parallel_mesh <- keras$distribution$DeviceMesh(   shape = shape(4, 2),   axis_names = list(\"data\", \"model\"),   devices = devices ) more_model_parallel_mesh <- keras$distribution$DeviceMesh(   shape = shape(2, 4),   axis_names = list(\"data\", \"model\"),   devices = devices ) full_model_parallel_mesh <- keras$distribution$DeviceMesh(   shape = shape(1, 8),   axis_names = list(\"data\", \"model\"),   devices = devices )"},{"path":"https://keras3.posit.co/dev/articles/distribution.html","id":"further-reading","dir":"Articles","previous_headings":"ModelParallel and LayoutMap","what":"Further reading","title":"Distributed training with Keras 3","text":"JAX Distributed arrays automatic parallelization JAX sharding module TensorFlow Distributed training DTensors TensorFlow DTensor concepts Using DTensors tf.keras","code":""},{"path":"https://keras3.posit.co/dev/articles/examples/nlp/neural_machine_translation_with_transformer.html","id":"introduction","dir":"Articles > Examples > Nlp","previous_headings":"","what":"Introduction","title":"English-to-Spanish translation with a sequence-to-sequence Transformer","text":"example, ’ll build sequence--sequence Transformer model, ’ll train English--Spanish machine translation task. ’ll learn : Vectorize text using layer_text_vectorization(). Implement layer_transformer_encoder(), layer_transformer_decoder(), layer_positional_embedding(). Prepare data training sequence--sequence model. Use trained model generate translations never-seen-input sentences (sequence--sequence inference). code featured adapted book Deep Learning R, Second Edition (chapter 11: Deep learning text). present example fairly barebones, detailed explanations building block works, well theory behind Transformers, recommend reading book.","code":""},{"path":"https://keras3.posit.co/dev/articles/examples/nlp/neural_machine_translation_with_transformer.html","id":"setup","dir":"Articles > Examples > Nlp","previous_headings":"","what":"Setup","title":"English-to-Spanish translation with a sequence-to-sequence Transformer","text":"","code":"library(glue)  library(tensorflow, exclude = c(\"set_random_seed\", \"shape\")) library(keras3)"},{"path":"https://keras3.posit.co/dev/articles/examples/nlp/neural_machine_translation_with_transformer.html","id":"downloading-the-data","dir":"Articles > Examples > Nlp","previous_headings":"","what":"Downloading the data","title":"English-to-Spanish translation with a sequence-to-sequence Transformer","text":"’ll working English--Spanish translation dataset provided Anki. Let’s download :","code":"zip_path <-   \"http://storage.googleapis.com/download.tensorflow.org/data/spa-eng.zip\" |>   get_file(origin = _, extract = TRUE)  text_path <- fs::path(zip_path, \"spa-eng/spa.txt\")"},{"path":"https://keras3.posit.co/dev/articles/examples/nlp/neural_machine_translation_with_transformer.html","id":"parsing-the-data","dir":"Articles > Examples > Nlp","previous_headings":"","what":"Parsing the data","title":"English-to-Spanish translation with a sequence-to-sequence Transformer","text":"line contains English sentence corresponding Spanish sentence. English sentence source sequence Spanish one target sequence. prepend token \"[start]\" append token \"[end]\" Spanish sentence. ’s sentence pairs look like: Now, let’s split sentence pairs training set, validation set, test set.","code":"text_file <- \"spa-eng/spa.txt\" text_pairs <- text_file %>%   readr::read_tsv(col_names = c(\"english\", \"spanish\"),                   col_types = c(\"cc\")) %>%   within(spanish %<>% paste(\"[start]\", ., \"[end]\")) df <- text_pairs[sample(nrow(text_pairs), 5), ] glue::glue_data(df, r\"(   english: {english}   spanish: {spanish} )\") |> cat(sep = \"\\n\\n\") ## english: I'm staying in Italy. ## spanish: [start] Me estoy quedando en Italia. [end] ## ## english: What's so strange about that? ## spanish: [start] ¿Qué es tan extraño acerca de eso? [end] ## ## english: All of the buses are full. ## spanish: [start] Todos los bondis están llenos. [end] ## ## english: Is this where your mother works? ## spanish: [start] ¿Es aquí donde trabaja tu madre? [end] ## ## english: Take precautions. ## spanish: [start] Ten cuidado. [end] random.shuffle(text_pairs) num_val_samples = int(0.15 * len(text_pairs)) num_train_samples = len(text_pairs) - 2 * num_val_samples train_pairs = text_pairs[:num_train_samples] val_pairs = text_pairs[num_train_samples : num_train_samples + num_val_samples] test_pairs = text_pairs[num_train_samples + num_val_samples :]  print(f\"{len(text_pairs)} total pairs\") print(f\"{len(train_pairs)} training pairs\") print(f\"{len(val_pairs)} validation pairs\") print(f\"{len(test_pairs)} test pairs\") num_test_samples <- num_val_samples <-   round(0.15 * nrow(text_pairs)) num_train_samples <- nrow(text_pairs) - num_val_samples - num_test_samples  pair_group <- sample(c(   rep(\"train\", num_train_samples),   rep(\"test\", num_test_samples),   rep(\"val\", num_val_samples) ))  train_pairs <- text_pairs[pair_group == \"train\", ] test_pairs <- text_pairs[pair_group == \"test\", ] val_pairs <- text_pairs[pair_group == \"val\", ] glue(r\"(   {nrow(text_pairs)} total pairs   {nrow(train_pairs)} training pairs   {nrow(val_pairs)} validation pairs   {nrow(test_pairs)} test pairs )\", .transformer = function(text, envir) {   val <- eval(str2lang(text), envir)   prettyNum(val, big.mark = \",\") }) ## 118,493 total pairs ## 82,945 training pairs ## 17,774 validation pairs ## 17,774 test pairs"},{"path":"https://keras3.posit.co/dev/articles/examples/nlp/neural_machine_translation_with_transformer.html","id":"vectorizing-the-text-data","dir":"Articles > Examples > Nlp","previous_headings":"","what":"Vectorizing the text data","title":"English-to-Spanish translation with a sequence-to-sequence Transformer","text":"’ll use two instances layer_text_vectorization() vectorize text data (one English one Spanish), say, turn original strings integer sequences integer represents index word vocabulary. English layer use default string standardization (strip punctuation characters) splitting scheme (split whitespace), Spanish layer use custom standardization, add character \"¿\" set punctuation characters stripped. Note: production-grade machine translation model, recommend stripping punctuation characters either language. Instead, recommend turning punctuation character token, achieve providing custom split function layer_text_vectorization(). Next, ’ll format datasets. training step, model seek predict target words N+1 (beyond) using source sentence target words 1 N. , training dataset yield tuple (inputs, targets), : inputs dictionary (named list) keys (names) encoder_inputs decoder_inputs. encoder_inputs vectorized source sentence decoder_inputs target sentence “far”, say, words 0 N used predict word N+1 (beyond) target sentence. target target sentence offset one step: provides next words target sentence – model try predict. Let’s take quick look sequence shapes (batches 64 pairs, sequences 20 steps long):","code":"punctuation_regex <- \"[¡¿]|[^[:^punct:][\\\\]]\" # the regex explained: Match ¡, or ¿, or any punctuation character except ] # # [:^punct:]: is a negated POSIX character class. # [:punct:] matches any punctuation character, so [:^punct:] matches any # character that is not a punctuation character. # [^...] negates the whole character class # So [^[:^punct:]] would matche any character that is a punctuation character. # Putting this all together, [^[:^punct:][\\\\]] matches any # punctuation character except the ] character.  custom_standardization <- function(input_string) {   input_string %>%     tf$strings$lower() %>%     tf$strings$regex_replace(punctuation_regex, \"\") }  input_string <- as_tensor(\"[start] ¡corre! [end]\") custom_standardization(input_string) ## tf.Tensor(b'[start] corre [end]', shape=(), dtype=string) vocab_size <- 15000 sequence_length <- 20  # rename to eng_vectorization eng_vectorization <- layer_text_vectorization(   max_tokens = vocab_size,   output_mode = \"int\",   output_sequence_length = sequence_length )  spa_vectorization <- layer_text_vectorization(   max_tokens = vocab_size,   output_mode = \"int\",   output_sequence_length = sequence_length + 1,   standardize = custom_standardization )  adapt(eng_vectorization, train_pairs$english) adapt(spa_vectorization, train_pairs$spanish) format_pair <- function(pair) {   eng <- pair$english |> eng_vectorization()   spa <- pair$spanish |> spa_vectorization()    spa_feature <- spa@r[NA:-2]                                                   # <1>   spa_target <- spa@r[2:NA]                                                     # <2>    features <- list(encoder_inputs = eng, decoder_inputs = spa_feature)   labels <- spa_target   sample_weight <- labels != 0    tuple(features, labels, sample_weight) }  batch_size <- 64  library(tfdatasets, exclude = \"shape\") make_dataset <- function(pairs) {   tensor_slices_dataset(pairs) |>     dataset_map(format_pair, num_parallel_calls = 4) |>     dataset_cache() |>     dataset_shuffle(2048) |>     dataset_batch(batch_size) |>     dataset_prefetch(16) }  train_ds <- make_dataset(train_pairs) val_ds <- make_dataset(val_pairs) c(inputs, targets, weights) %<-% iter_next(as_iterator(train_ds)) str(inputs) ## List of 2 ##  $ encoder_inputs:<tf.Tensor: shape=(64, 20), dtype=int64, numpy=…> ##  $ decoder_inputs:<tf.Tensor: shape=(64, 20), dtype=int64, numpy=…> str(targets) ## <tf.Tensor: shape=(64, 20), dtype=int64, numpy=…>"},{"path":"https://keras3.posit.co/dev/articles/examples/nlp/neural_machine_translation_with_transformer.html","id":"building-the-model","dir":"Articles > Examples > Nlp","previous_headings":"","what":"Building the model","title":"English-to-Spanish translation with a sequence-to-sequence Transformer","text":"sequence--sequence Transformer consists TransformerEncoder TransformerDecoder chained together. make model aware word order, also use PositionalEmbedding layer. source sequence pass TransformerEncoder, produce new representation . new representation passed TransformerDecoder, together target sequence far (target words 1 N). TransformerDecoder seek predict next words target sequence (N+1 beyond). key detail makes possible causal masking (see method get_causal_attention_mask() TransformerDecoder). TransformerDecoder sees entire sequences , thus must make sure uses information target tokens 0 N predicting token N+1 (otherwise, use information future, result model used inference time). Next, assemble end--end model.","code":"layer_transformer_encoder <- Layer(   classname = \"TransformerEncoder\",   initialize = function(embed_dim, dense_dim, num_heads, ...) {     super$initialize(...)     self$embed_dim <- embed_dim     self$dense_dim <- dense_dim     self$num_heads <- num_heads     self$attention <-       layer_multi_head_attention(num_heads = num_heads,                                  key_dim = embed_dim)      self$dense_proj <- keras_model_sequential() %>%       layer_dense(dense_dim, activation = \"relu\") %>%       layer_dense(embed_dim)      self$layernorm_1 <- layer_layer_normalization()     self$layernorm_2 <- layer_layer_normalization()     self$supports_masking <- TRUE   },    call = function(inputs, mask = NULL) {     if (!is.null(mask))       mask <- mask[, NULL, ] |> op_cast(\"int32\")      inputs %>%       { self$attention(., ., attention_mask = mask) + . } %>%       self$layernorm_1() %>%       { self$dense_proj(.) + . } %>%       self$layernorm_2()   },    get_config = function() {     config <- super$get_config()     for(name in c(\"embed_dim\", \"num_heads\", \"dense_dim\"))       config[[name]] <- self[[name]]     config   } )  layer_transformer_decoder <- Layer(   classname = \"TransformerDecoder\",    initialize = function(embed_dim, latent_dim, num_heads, ...) {     super$initialize(...)     self$embed_dim <- embed_dim     self$latent_dim <- latent_dim     self$num_heads <- num_heads     self$attention_1 <- layer_multi_head_attention(num_heads = num_heads,                                                    key_dim = embed_dim)     self$attention_2 <- layer_multi_head_attention(num_heads = num_heads,                                                    key_dim = embed_dim)     self$dense_proj <- keras_model_sequential() %>%       layer_dense(latent_dim, activation = \"relu\") %>%       layer_dense(embed_dim)      self$layernorm_1 <- layer_layer_normalization()     self$layernorm_2 <- layer_layer_normalization()     self$layernorm_3 <- layer_layer_normalization()     self$supports_masking <- TRUE   },    get_config = function() {     config <- super$get_config()     for (name in c(\"embed_dim\", \"num_heads\", \"latent_dim\"))       config[[name]] <- self[[name]]     config   },     get_causal_attention_mask = function(inputs) {     c(batch_size, sequence_length, encoding_length) %<-% op_shape(inputs)      x <- op_arange(0L, sequence_length, include_end = FALSE)     i <- x[, NULL]     j <- x[NULL, ]     mask <- op_cast(i >= j, \"int32\")      repeats <- op_stack(c(batch_size, 1L, 1L))     op_tile(mask[NULL, , ], repeats)   },    call = function(inputs, mask = NULL) {     c(inputs_seq, encoder_outputs) %<-% inputs     causal_mask <- self$get_causal_attention_mask(inputs_seq)      if (is.null(mask)) {       inputs_padding_mask <- NULL       encoder_outputs_padding_mask <- NULL     } else {       c(inputs_padding_mask, encoder_outputs_padding_mask) %<-% mask     }      attention_output_1 <- self$attention_1(       query = inputs_seq,       value = inputs_seq,       key = inputs_seq,       attention_mask = causal_mask,       query_mask = inputs_padding_mask     )     out_1 <- self$layernorm_1(inputs_seq + attention_output_1)      attention_output_2 <- self$attention_2(       query = out_1,       value = encoder_outputs,       key = encoder_outputs,       query_mask = inputs_padding_mask,       key_mask = encoder_outputs_padding_mask     )     out_2 <- self$layernorm_2(out_1 + attention_output_2)      self$layernorm_3(out_2 + self$dense_proj(out_2))   } )  layer_positional_embedding <- Layer(   classname = \"PositionalEmbedding\",    initialize = function(sequence_length, vocab_size, embed_dim, ...) {     super$initialize(...)     self$token_embeddings <- layer_embedding(       input_dim = vocab_size, output_dim = embed_dim     )     self$position_embeddings <- layer_embedding(       input_dim = sequence_length, output_dim = embed_dim     )     self$sequence_length <- sequence_length     self$vocab_size <- vocab_size     self$embed_dim <- embed_dim   },    call = function(inputs) {     c(., len) %<-% op_shape(inputs) # (batch_size, seq_len)     positions <- op_arange(0, len, dtype = \"int32\", include_end = FALSE)     embedded_tokens <- self$token_embeddings(inputs)     embedded_positions <- self$position_embeddings(positions)     embedded_tokens + embedded_positions   },    compute_mask = function(inputs, mask = NULL) {     inputs != 0L   },    get_config = function() {     config <- super$get_config()     for(name in c(\"sequence_length\", \"vocab_size\", \"embed_dim\"))       config[[name]] <- self[[name]]     config   } ) embed_dim <- 256 latent_dim <- 2048 num_heads <- 8  encoder_inputs <- layer_input(shape(NA), dtype = \"int64\",                               name = \"encoder_inputs\") encoder_outputs <- encoder_inputs %>%   layer_positional_embedding(sequence_length, vocab_size, embed_dim) %>%   layer_transformer_encoder(embed_dim, latent_dim, num_heads)  encoder <- keras_model(encoder_inputs, encoder_outputs)  decoder_inputs <-  layer_input(shape(NA), dtype = \"int64\",                                name = \"decoder_inputs\") encoded_seq_inputs <- layer_input(shape(NA, embed_dim),                                   name = \"decoder_state_inputs\")  transformer_decoder <- layer_transformer_decoder(NULL,   embed_dim, latent_dim, num_heads)  decoder_outputs <- decoder_inputs %>%   layer_positional_embedding(sequence_length, vocab_size, embed_dim) %>%   { transformer_decoder(list(., encoded_seq_inputs)) } %>%   layer_dropout(0.5) %>%   layer_dense(vocab_size, activation=\"softmax\")  decoder <- keras_model(inputs = list(decoder_inputs, encoded_seq_inputs),                        outputs = decoder_outputs)  decoder_outputs <- decoder(list(decoder_inputs, encoder_outputs))  transformer <- keras_model(   inputs = list(encoder_inputs = encoder_inputs, decoder_inputs = decoder_inputs),   outputs = decoder_outputs,   name = \"transformer\" )"},{"path":"https://keras3.posit.co/dev/articles/examples/nlp/neural_machine_translation_with_transformer.html","id":"training-our-model","dir":"Articles > Examples > Nlp","previous_headings":"","what":"Training our model","title":"English-to-Spanish translation with a sequence-to-sequence Transformer","text":"’ll use accuracy quick way monitor training progress validation data. Note machine translation typically uses BLEU scores well metrics, rather accuracy. train 1 epoch, get model actually converge train least 30 epochs.","code":"epochs <- 1  # This should be at least 30 for convergence  transformer ## Model: \"transformer\" ## ┏━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━┓ ## ┃ Layer (type)        ┃ Output Shape      ┃    Param # ┃ Connected to      ┃ ## ┡━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━┩ ## │ encoder_inputs      │ (None, None)      │          0 │ -                 │ ## │ (InputLayer)        │                   │            │                   │ ## ├─────────────────────┼───────────────────┼────────────┼───────────────────┤ ## │ positional_embeddi… │ (None, None, 256) │  3,845,120 │ encoder_inputs[0… │ ## │ (PositionalEmbeddi… │                   │            │                   │ ## ├─────────────────────┼───────────────────┼────────────┼───────────────────┤ ## │ not_equal           │ (None, None)      │          0 │ encoder_inputs[0… │ ## │ (NotEqual)          │                   │            │                   │ ## ├─────────────────────┼───────────────────┼────────────┼───────────────────┤ ## │ decoder_inputs      │ (None, None)      │          0 │ -                 │ ## │ (InputLayer)        │                   │            │                   │ ## ├─────────────────────┼───────────────────┼────────────┼───────────────────┤ ## │ transformer_encoder │ (None, None, 256) │  3,155,456 │ positional_embed… │ ## │ (TransformerEncode… │                   │            │ not_equal[0][0]   │ ## ├─────────────────────┼───────────────────┼────────────┼───────────────────┤ ## │ functional_3        │ (None, None,      │ 12,959,640 │ decoder_inputs[0… │ ## │ (Functional)        │ 15000)            │            │ transformer_enco… │ ## │                     │                   │            │ not_equal[0][0]   │ ## └─────────────────────┴───────────────────┴────────────┴───────────────────┘ ##  Total params: 19,960,216 (76.14 MB) ##  Trainable params: 19,960,216 (76.14 MB) ##  Non-trainable params: 0 (0.00 B) transformer |> compile(   \"rmsprop\",   loss = loss_sparse_categorical_crossentropy(ignore_class = 0),   metrics = \"accuracy\" )  transformer |> fit(train_ds, epochs = epochs,                    validation_data = val_ds) ## 1297/1297 - 55s - 42ms/step - accuracy: 0.1592 - loss: 3.8833 - val_accuracy: 0.1691 - val_loss: 3.1841"},{"path":"https://keras3.posit.co/dev/articles/examples/nlp/neural_machine_translation_with_transformer.html","id":"decoding-test-sentences","dir":"Articles > Examples > Nlp","previous_headings":"","what":"Decoding test sentences","title":"English-to-Spanish translation with a sequence-to-sequence Transformer","text":"Finally, let’s demonstrate translate brand new English sentences. simply feed model vectorized English sentence well target token \"[start]\", repeatedly generated next token, hit token \"[end]\". 30 epochs, get results :","code":"spa_vocab <- spa_vectorization |> get_vocabulary() max_decoded_sentence_length <- 20 tf_decode_sequence <- tf_function(function(input_sentence) {   withr::local_options(tensorflow.extract.style = \"python\")    tokenized_input_sentence <- input_sentence %>%     as_tensor(shape = c(1, 1)) %>%     eng_vectorization()   spa_vocab <- as_tensor(spa_vocab)   decoded_sentence <- as_tensor(\"[start]\", shape = c(1, 1))    for (i in tf$range(as.integer(max_decoded_sentence_length))) {      tokenized_target_sentence <-       spa_vectorization(decoded_sentence)[,NA:-1]      next_token_predictions <-       transformer(list(         encoder_inputs = tokenized_input_sentence,         decoder_inputs = tokenized_target_sentence       ))      sampled_token_index <- tf$argmax(next_token_predictions[0, i, ])     sampled_token <- spa_vocab[sampled_token_index]     decoded_sentence <-       tf$strings$join(c(decoded_sentence, sampled_token),                       separator = \" \")      if (sampled_token == \"[end]\")       break   }    decoded_sentence  })  for (i in seq(20)) {      c(input_sentence, correct_translation) %<-%       test_pairs[sample.int(nrow(test_pairs), 1), ]     cat(\"-\\n\")     cat(\"English:\", input_sentence, \"\\n\")     cat(\"Correct Translation:\", tolower(correct_translation), \"\\n\")     cat(\"  Model Translation:\", input_sentence %>% as_tensor() %>%           tf_decode_sequence() %>% as.character(), \"\\n\") } English: I'm sure everything will be fine. Correct Translation: [start] estoy segura de que todo irá bien. [end]   Model Translation: [start] estoy seguro de que todo va bien [end]"},{"path":"https://keras3.posit.co/dev/articles/examples/nlp/text_classification_from_scratch.html","id":"introduction","dir":"Articles > Examples > Nlp","previous_headings":"","what":"Introduction","title":"Text classification from scratch","text":"example shows text classification starting raw text (set text files disk). demonstrate workflow IMDB sentiment classification dataset (unprocessed version). use layer_text_vectorization() word splitting & indexing.","code":""},{"path":"https://keras3.posit.co/dev/articles/examples/nlp/text_classification_from_scratch.html","id":"setup","dir":"Articles > Examples > Nlp","previous_headings":"","what":"Setup","title":"Text classification from scratch","text":"","code":"library(tensorflow, exclude = c(\"shape\", \"set_random_seed\")) library(tfdatasets, exclude = \"shape\") library(keras3)"},{"path":"https://keras3.posit.co/dev/articles/examples/nlp/text_classification_from_scratch.html","id":"load-the-data-imdb-movie-review-sentiment-classification","dir":"Articles > Examples > Nlp","previous_headings":"","what":"Load the data: IMDB movie review sentiment classification","title":"Text classification from scratch","text":"Let’s download data inspect structure. aclImdb folder contains train test subfolder: aclImdb/train/pos aclImdb/train/neg folders contain text files, represents one review (either positive negative): interested pos neg subfolders, let’s delete subfolder text files : can use utility text_dataset_from_directory() generate labeled tf_dataset object set text files disk filed class-specific folders. Let’s use generate training, validation, test datasets. validation training datasets generated two subsets train directory, 20% samples going validation dataset 80% going training dataset. validation dataset addition test dataset useful tuning hyperparameters, model architecture, test dataset used. putting model real world however, retrained using available training data (without creating validation dataset), performance maximized. using validation_split subset arguments, make sure either specify random seed, pass shuffle=FALSE, validation & training splits get overlap. Let’s preview samples:","code":"if (!dir.exists(\"datasets/aclImdb\")) {   dir.create(\"datasets\")   download.file(     \"https://ai.stanford.edu/~amaas/data/sentiment/aclImdb_v1.tar.gz\",     \"datasets/aclImdb_v1.tar.gz\"   )   untar(\"datasets/aclImdb_v1.tar.gz\", exdir = \"datasets\")   unlink(\"datasets/aclImdb/train/unsup\", recursive = TRUE) } head(list.files(\"datasets/aclImdb/test\")) ## [1] \"labeledBow.feat\" \"neg\"             \"pos\"             \"urls_neg.txt\" ## [5] \"urls_pos.txt\" head(list.files(\"datasets/aclImdb/train\")) ## [1] \"labeledBow.feat\" \"neg\"             \"pos\"             \"unsupBow.feat\" ## [5] \"urls_neg.txt\"    \"urls_pos.txt\" writeLines(strwrap(readLines(\"datasets/aclImdb/train/pos/4229_10.txt\"))) ## Don't waste time reading my review. Go out and see this ## astonishingly good episode, which may very well be the best Columbo ## ever written! Ruth Gordon is perfectly cast as the scheming yet ## charming mystery writer who murders her son-in-law to avenge his ## murder of her daughter. Columbo is his usual rumpled, befuddled and ## far-cleverer-than-he-seems self, and this particular installment ## features fantastic chemistry between Gordon and Falk. Ironically, ## this was not written by heralded creators Levinson or Link yet is ## possibly the densest, most thoroughly original and twist-laden ## Columbo plot ever. Utterly satisfying in nearly every department ## and overflowing with droll and witty dialogue and thinking. Truly ## unexpected and inventive climax tops all. 10/10...seek this one out ## on Netflix! unlink(\"datasets/aclImdb/train/unsup\", recursive = TRUE) batch_size <- 32  raw_train_ds <- text_dataset_from_directory(   \"datasets/aclImdb/train\",   batch_size = batch_size,   validation_split = 0.2,   subset = \"training\",   seed = 1337 ) ## Found 25000 files belonging to 2 classes. ## Using 20000 files for training. raw_val_ds <- text_dataset_from_directory(   \"datasets/aclImdb/train\",   batch_size = batch_size,   validation_split = 0.2,   subset = \"validation\",   seed = 1337 ) ## Found 25000 files belonging to 2 classes. ## Using 5000 files for validation. raw_test_ds <- text_dataset_from_directory(   \"datasets/aclImdb/test\",   batch_size = batch_size ) ## Found 25000 files belonging to 2 classes. cat(\"Number of batches in raw_train_ds:\", length(raw_train_ds), \"\\n\") ## Number of batches in raw_train_ds: 625 cat(\"Number of batches in raw_val_ds:\", length(raw_val_ds), \"\\n\") ## Number of batches in raw_val_ds: 157 cat(\"Number of batches in raw_test_ds:\", length(raw_test_ds), \"\\n\") ## Number of batches in raw_test_ds: 782 # It's important to take a look at your raw data to ensure your normalization # and tokenization will work as expected. We can do that by taking a few # examples from the training set and looking at them. # This is one of the places where eager execution shines: # we can just evaluate these tensors using .numpy() # instead of needing to evaluate them in a Session/Graph context. batch <- iter_next(as_iterator(raw_train_ds)) str(batch) ## List of 2 ##  $ :<tf.Tensor: shape=(32), dtype=string, numpy=…> ##  $ :<tf.Tensor: shape=(32), dtype=int32, numpy=…> c(text_batch, label_batch) %<-% batch for (i in 1:3) {   print(text_batch[i])   print(label_batch[i]) } ## tf.Tensor(b\"I have read the novel Reaper of Ben Mezrich a fews years ago and last night I accidentally came to see this adaption.<br /><br />Although it's been years since I read the story the first time, the differences between the novel and the movie are humongous. Very important elements, which made the whole thing plausible are just written out or changed to bad.<br /><br />If the plot sounds interesting to you: go and get the novel. Its much, much, much better.<br /><br />Still 4 out of 10 since it was hard to stop watching because of the great basic plot by Ben Mezrich.\", shape=(), dtype=string) ## tf.Tensor(0, shape=(), dtype=int32) ## tf.Tensor(b'After seeing all the Jesse James, Quantrill, jayhawkers,etc films in the fifties, it is quite a thrill to see this film with a new perspective by director Ang Lee. The scene of the attack of Lawrence, Kansas is awesome. The romantic relationship between Jewel and Toby Mcguire turns out to be one of the best parts and Jonathan Rhys-Meyers is outstanding as the bad guy. All the time this film makes you feel the horror of war, and the desperate situation of the main characters who do not know if they are going to survive the next hours. Definitely worth seeing.', shape=(), dtype=string) ## tf.Tensor(1, shape=(), dtype=int32) ## tf.Tensor(b'AG was an excellent presentation of drama, suspense and thriller that is so rare to American TV. Sheriff Lucas gave many a viewer the willies. We rooted for Caleb as he strove to resist the overtures of Sheriff Lucas. We became engrossed and fearful upon learning of the unthinkable connection between these two characters. The manipulations which weekly gave cause to fear what Lucas would do next were truly surprising. This show lived up to the \"Gothic\" moniker in ways American entertainment has so seldom attempted, much less mastered. The suits definitely made a big mistake in not supporting this show. This show puts shame to the current glut of \"reality\" shows- which are so less than satisfying viewing.The call for a DVD box set is well based. This show is quality viewing for a discerning market hungry for quality viewing. A public that is tiring of over-saturation of mind-numbing reality fare will welcome this gem of real storytelling. Bring on the DVD box set!!', shape=(), dtype=string) ## tf.Tensor(1, shape=(), dtype=int32)"},{"path":"https://keras3.posit.co/dev/articles/examples/nlp/text_classification_from_scratch.html","id":"prepare-the-data","dir":"Articles > Examples > Nlp","previous_headings":"","what":"Prepare the data","title":"Text classification from scratch","text":"particular, remove <br /> tags.","code":"# Having looked at our data above, we see that the raw text contains HTML break # tags of the form '<br />'. These tags will not be removed by the default # standardizer (which doesn't strip HTML). Because of this, we will need to # create a custom standardization function. custom_standardization_fn <- function(string_tensor) {   string_tensor |>     tf$strings$lower() |> # convert to all lowercase     tf$strings$regex_replace(\"<br />\", \" \") |> # remove '<br />' HTML tag     tf$strings$regex_replace(\"[[:punct:]]\", \"\") # remove punctuation }   # Model constants. max_features <- 20000 embedding_dim <- 128 sequence_length <- 500  # Now that we have our custom standardization, we can instantiate our text # vectorization layer. We are using this layer to normalize, split, and map # strings to integers, so we set our 'output_mode' to 'int'. # Note that we're using the default split function, # and the custom standardization defined above. # We also set an explicit maximum sequence length, since the CNNs later in our # model won't support ragged sequences. vectorize_layer <- layer_text_vectorization(   standardize = custom_standardization_fn,   max_tokens = max_features,   output_mode = \"int\",   output_sequence_length = sequence_length, )  # Now that the vectorize_layer has been created, call `adapt` on a text-only # dataset to create the vocabulary. You don't have to batch, but for very large # datasets this means you're not keeping spare copies of the dataset in memory.  # Let's make a text-only dataset (no labels): text_ds <- raw_train_ds |>   dataset_map(\\(x, y) x) # Let's call `adapt`: vectorize_layer |> adapt(text_ds)"},{"path":"https://keras3.posit.co/dev/articles/examples/nlp/text_classification_from_scratch.html","id":"two-options-to-vectorize-the-data","dir":"Articles > Examples > Nlp","previous_headings":"","what":"Two options to vectorize the data","title":"Text classification from scratch","text":"2 ways can use text vectorization layer: Option 1: Make part model, obtain model processes raw strings, like : Option 2: Apply text dataset obtain dataset word indices, feed model expects integer sequences inputs. important difference two option 2 enables asynchronous CPU processing buffering data training GPU. ’re training model GPU, probably want go option get best performance. . export model production, ’d ship model accepts raw strings input, like code snippet option 1 . can done training. last section.","code":"text_input <- keras_input(shape = c(1L), dtype = \"string\", name = 'text') x <- text_input |>   vectorize_layer() |>   layer_embedding(max_features + 1, embedding_dim) vectorize_text <- function(text, label) {   text <- text |>     op_expand_dims(-1) |>     vectorize_layer()   list(text, label) }  # Vectorize the data. train_ds <- raw_train_ds |> dataset_map(vectorize_text) val_ds   <- raw_val_ds   |> dataset_map(vectorize_text) test_ds  <- raw_test_ds  |> dataset_map(vectorize_text)  # Do async prefetching / buffering of the data for best performance on GPU. train_ds <- train_ds |>   dataset_cache() |>   dataset_prefetch(buffer_size = 10) val_ds <- val_ds |>   dataset_cache() |>   dataset_prefetch(buffer_size = 10) test_ds <- test_ds |>   dataset_cache() |>   dataset_prefetch(buffer_size = 10)"},{"path":"https://keras3.posit.co/dev/articles/examples/nlp/text_classification_from_scratch.html","id":"build-a-model","dir":"Articles > Examples > Nlp","previous_headings":"","what":"Build a model","title":"Text classification from scratch","text":"choose simple 1D convnet starting Embedding layer.","code":"# A integer input for vocab indices. inputs <- keras_input(shape = c(NA), dtype = \"int64\")  predictions <- inputs |>   # Next, we add a layer to map those vocab indices into a space of dimensionality   # 'embedding_dim'.   layer_embedding(max_features, embedding_dim) |>   layer_dropout(0.5) |>   # Conv1D + global max pooling   layer_conv_1d(128, 7, padding = \"valid\", activation = \"relu\", strides = 3) |>   layer_conv_1d(128, 7, padding = \"valid\", activation = \"relu\", strides = 3) |>   layer_global_max_pooling_1d() |>   # We add a vanilla hidden layer:   layer_dense(128, activation = \"relu\") |>   layer_dropout(0.5) |>   # We project onto a single unit output layer, and squash it with a sigmoid:   layer_dense(1, activation = \"sigmoid\", name = \"predictions\")  model <- keras_model(inputs, predictions)  summary(model) ## Model: \"functional\" ## ┏━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━┓ ## ┃ Layer (type)                    ┃ Output Shape           ┃       Param # ┃ ## ┡━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━┩ ## │ input_layer (InputLayer)        │ (None, None)           │             0 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ embedding_1 (Embedding)         │ (None, None, 128)      │     2,560,000 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ dropout (Dropout)               │ (None, None, 128)      │             0 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ conv1d (Conv1D)                 │ (None, None, 128)      │       114,816 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ conv1d_1 (Conv1D)               │ (None, None, 128)      │       114,816 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ global_max_pooling1d            │ (None, 128)            │             0 │ ## │ (GlobalMaxPooling1D)            │                        │               │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ dense (Dense)                   │ (None, 128)            │        16,512 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ dropout_1 (Dropout)             │ (None, 128)            │             0 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ predictions (Dense)             │ (None, 1)              │           129 │ ## └─────────────────────────────────┴────────────────────────┴───────────────┘ ##  Total params: 2,806,273 (10.71 MB) ##  Trainable params: 2,806,273 (10.71 MB) ##  Non-trainable params: 0 (0.00 B) # Compile the model with binary crossentropy loss and an adam optimizer. model |> compile(loss = \"binary_crossentropy\",                  optimizer = \"adam\",                  metrics = \"accuracy\")"},{"path":"https://keras3.posit.co/dev/articles/examples/nlp/text_classification_from_scratch.html","id":"train-the-model","dir":"Articles > Examples > Nlp","previous_headings":"","what":"Train the model","title":"Text classification from scratch","text":"","code":"epochs <- 3  # Fit the model using the train and test datasets. model |> fit(train_ds, validation_data = val_ds, epochs = epochs) ## Epoch 1/3 ## 625/625 - 5s - 8ms/step - accuracy: 0.6964 - loss: 0.5261 - val_accuracy: 0.8668 - val_loss: 0.3185 ## Epoch 2/3 ## 625/625 - 2s - 3ms/step - accuracy: 0.9074 - loss: 0.2365 - val_accuracy: 0.8754 - val_loss: 0.3245 ## Epoch 3/3 ## 625/625 - 2s - 3ms/step - accuracy: 0.9551 - loss: 0.1220 - val_accuracy: 0.8284 - val_loss: 0.4862"},{"path":"https://keras3.posit.co/dev/articles/examples/nlp/text_classification_from_scratch.html","id":"evaluate-the-model-on-the-test-set","dir":"Articles > Examples > Nlp","previous_headings":"","what":"Evaluate the model on the test set","title":"Text classification from scratch","text":"","code":"model |> evaluate(test_ds) ## 782/782 - 1s - 2ms/step - accuracy: 0.8115 - loss: 0.5299 ## $accuracy ## [1] 0.81152 ## ## $loss ## [1] 0.5299004"},{"path":"https://keras3.posit.co/dev/articles/examples/nlp/text_classification_from_scratch.html","id":"make-an-end-to-end-model","dir":"Articles > Examples > Nlp","previous_headings":"","what":"Make an end-to-end model","title":"Text classification from scratch","text":"want obtain model capable processing raw strings, can simply create new model (using weights just trained):","code":"# A string input inputs <- keras_input(shape = c(1), dtype = \"string\") # Turn strings into vocab indices indices <- vectorize_layer(inputs) # Turn vocab indices into predictions outputs <- model(indices)  # Our end to end model end_to_end_model <- keras_model(inputs, outputs) end_to_end_model |> compile(   loss = \"binary_crossentropy\",   optimizer = \"adam\",   metrics = c(\"accuracy\") )  # Test it with `raw_test_ds`, which yields raw strings end_to_end_model |> evaluate(raw_test_ds) ## 782/782 - 3s - 4ms/step - accuracy: 0.8115 - loss: 0.5299 ## $accuracy ## [1] 0.81152 ## ## $loss ## [1] 0.5299006"},{"path":"https://keras3.posit.co/dev/articles/examples/structured_data/imbalanced_classification.html","id":"introduction","dir":"Articles > Examples > Structured_data","previous_headings":"","what":"Introduction","title":"Imbalanced classification: credit card fraud detection","text":"example looks Kaggle Credit Card Fraud Detection dataset demonstrate train classification model data highly imbalanced classes. can download data clicking “Download” link, ’re setup kaggle API key \"~/.kaggle/kagle.json\", can run following:","code":"reticulate::py_install(\"kaggle\", pip = TRUE) reticulate::py_available(TRUE) # ensure 'kaggle' is on the PATH system(\"kaggle datasets download -d mlg-ulb/creditcardfraud\") zip::unzip(\"creditcardfraud.zip\", files = \"creditcard.csv\")"},{"path":"https://keras3.posit.co/dev/articles/examples/structured_data/imbalanced_classification.html","id":"first-load-the-data","dir":"Articles > Examples > Structured_data","previous_headings":"","what":"First, load the data","title":"Imbalanced classification: credit card fraud detection","text":"","code":"library(readr) df <- read_csv(\"creditcard.csv\", col_types = cols(   Class = col_integer(),   .default = col_double() )) tibble::glimpse(df) ## Rows: 284,807 ## Columns: 31 ## $ Time   <dbl> 0, 0, 1, 1, 2, 2, 4, 7, 7, 9, 10, 10, 10, 11, 12, 12, 12, 1… ## $ V1     <dbl> -1.3598071, 1.1918571, -1.3583541, -0.9662717, -1.1582331, … ## $ V2     <dbl> -0.07278117, 0.26615071, -1.34016307, -0.18522601, 0.877736… ## $ V3     <dbl> 2.53634674, 0.16648011, 1.77320934, 1.79299334, 1.54871785,… ## $ V4     <dbl> 1.37815522, 0.44815408, 0.37977959, -0.86329128, 0.40303393… ## $ V5     <dbl> -0.33832077, 0.06001765, -0.50319813, -0.01030888, -0.40719… ## $ V6     <dbl> 0.46238778, -0.08236081, 1.80049938, 1.24720317, 0.09592146… ## $ V7     <dbl> 0.239598554, -0.078802983, 0.791460956, 0.237608940, 0.5929… ## $ V8     <dbl> 0.098697901, 0.085101655, 0.247675787, 0.377435875, -0.2705… ## $ V9     <dbl> 0.3637870, -0.2554251, -1.5146543, -1.3870241, 0.8177393, -… ## $ V10    <dbl> 0.09079417, -0.16697441, 0.20764287, -0.05495192, 0.7530744… ## $ V11    <dbl> -0.55159953, 1.61272666, 0.62450146, -0.22648726, -0.822842… ## $ V12    <dbl> -0.61780086, 1.06523531, 0.06608369, 0.17822823, 0.53819555… ## $ V13    <dbl> -0.99138985, 0.48909502, 0.71729273, 0.50775687, 1.34585159… ## $ V14    <dbl> -0.31116935, -0.14377230, -0.16594592, -0.28792375, -1.1196… ## $ V15    <dbl> 1.468176972, 0.635558093, 2.345864949, -0.631418118, 0.1751… ## $ V16    <dbl> -0.47040053, 0.46391704, -2.89008319, -1.05964725, -0.45144… ## $ V17    <dbl> 0.207971242, -0.114804663, 1.109969379, -0.684092786, -0.23… ## $ V18    <dbl> 0.02579058, -0.18336127, -0.12135931, 1.96577500, -0.038194… ## $ V19    <dbl> 0.40399296, -0.14578304, -2.26185710, -1.23262197, 0.803486… ## $ V20    <dbl> 0.25141210, -0.06908314, 0.52497973, -0.20803778, 0.4085423… ## $ V21    <dbl> -0.018306778, -0.225775248, 0.247998153, -0.108300452, -0.0… ## $ V22    <dbl> 0.277837576, -0.638671953, 0.771679402, 0.005273597, 0.7982… ## $ V23    <dbl> -0.110473910, 0.101288021, 0.909412262, -0.190320519, -0.13… ## $ V24    <dbl> 0.06692807, -0.33984648, -0.68928096, -1.17557533, 0.141266… ## $ V25    <dbl> 0.12853936, 0.16717040, -0.32764183, 0.64737603, -0.2060095… ## $ V26    <dbl> -0.18911484, 0.12589453, -0.13909657, -0.22192884, 0.502292… ## $ V27    <dbl> 0.133558377, -0.008983099, -0.055352794, 0.062722849, 0.219… ## $ V28    <dbl> -0.021053053, 0.014724169, -0.059751841, 0.061457629, 0.215… ## $ Amount <dbl> 149.62, 2.69, 378.66, 123.50, 69.99, 3.67, 4.99, 40.80, 93.… ## $ Class  <int> 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,…"},{"path":"https://keras3.posit.co/dev/articles/examples/structured_data/imbalanced_classification.html","id":"prepare-a-validation-set","dir":"Articles > Examples > Structured_data","previous_headings":"","what":"Prepare a validation set","title":"Imbalanced classification: credit card fraud detection","text":"","code":"val_idx <- nrow(df) %>% sample.int(., round( . * 0.2)) val_df <- df[val_idx, ] train_df <- df[-val_idx, ]  cat(\"Number of training samples:\", nrow(train_df), \"\\n\") ## Number of training samples: 227846 cat(\"Number of validation samples:\", nrow(val_df), \"\\n\") ## Number of validation samples: 56961"},{"path":"https://keras3.posit.co/dev/articles/examples/structured_data/imbalanced_classification.html","id":"analyze-class-imbalance-in-the-targets","dir":"Articles > Examples > Structured_data","previous_headings":"","what":"Analyze class imbalance in the targets","title":"Imbalanced classification: credit card fraud detection","text":"","code":"counts <- table(train_df$Class) counts ## ##      0      1 ## 227464    382 cat(sprintf(\"Number of positive samples in training data: %i (%.2f%% of total)\",             counts[\"1\"], 100 * counts[\"1\"] / sum(counts))) ## Number of positive samples in training data: 382 (0.17% of total) weight_for_0 = 1 / counts[\"0\"] weight_for_1 = 1 / counts[\"1\"]"},{"path":"https://keras3.posit.co/dev/articles/examples/structured_data/imbalanced_classification.html","id":"normalize-the-data-using-training-set-statistics","dir":"Articles > Examples > Structured_data","previous_headings":"","what":"Normalize the data using training set statistics","title":"Imbalanced classification: credit card fraud detection","text":"","code":"feature_names <- colnames(train_df) %>% setdiff(\"Class\")  train_features <- as.matrix(train_df[feature_names]) train_targets <- as.matrix(train_df$Class)  val_features <- as.matrix(val_df[feature_names]) val_targets <- as.matrix(val_df$Class)  train_features %<>% scale() val_features %<>% scale(center = attr(train_features, \"scaled:center\"),                         scale = attr(train_features, \"scaled:scale\"))"},{"path":"https://keras3.posit.co/dev/articles/examples/structured_data/imbalanced_classification.html","id":"build-a-binary-classification-model","dir":"Articles > Examples > Structured_data","previous_headings":"","what":"Build a binary classification model","title":"Imbalanced classification: credit card fraud detection","text":"","code":"model <-   keras_model_sequential(input_shape = ncol(train_features)) |>   layer_dense(256, activation = \"relu\") |>   layer_dense(256, activation = \"relu\") |>   layer_dropout(0.3) |>   layer_dense(256, activation = \"relu\") |>   layer_dropout(0.3) |>   layer_dense(1, activation = \"sigmoid\")  model ## Model: \"sequential\" ## ┏━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━┓ ## ┃ Layer (type)                    ┃ Output Shape           ┃       Param # ┃ ## ┡━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━┩ ## │ dense (Dense)                   │ (None, 256)            │         7,936 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ dense_1 (Dense)                 │ (None, 256)            │        65,792 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ dropout (Dropout)               │ (None, 256)            │             0 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ dense_2 (Dense)                 │ (None, 256)            │        65,792 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ dropout_1 (Dropout)             │ (None, 256)            │             0 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ dense_3 (Dense)                 │ (None, 1)              │           257 │ ## └─────────────────────────────────┴────────────────────────┴───────────────┘ ##  Total params: 139,777 (546.00 KB) ##  Trainable params: 139,777 (546.00 KB) ##  Non-trainable params: 0 (0.00 B)"},{"path":"https://keras3.posit.co/dev/articles/examples/structured_data/imbalanced_classification.html","id":"train-the-model-with-class_weight-argument","dir":"Articles > Examples > Structured_data","previous_headings":"","what":"Train the model with class_weight argument","title":"Imbalanced classification: credit card fraud detection","text":"","code":"metrics <- list(   metric_false_negatives(name = \"fn\"),   metric_false_positives(name = \"fp\"),   metric_true_negatives(name = \"tn\"),   metric_true_positives(name = \"tp\"),   metric_precision(name = \"precision\"),   metric_recall(name = \"recall\") ) model |> compile(   optimizer = optimizer_adam(1e-2),   loss = \"binary_crossentropy\",   metrics = metrics ) callbacks <- list(   callback_model_checkpoint(\"fraud_model_at_epoch_{epoch}.keras\") )  class_weight <- list(\"0\" = weight_for_0,                      \"1\" = weight_for_1)  model |> fit(   train_features, train_targets,   validation_data = list(val_features, val_targets),   class_weight = class_weight,   batch_size = 2048,   epochs = 30,   callbacks = callbacks,   verbose = 2 ) ## Epoch 1/30 ## 112/112 - 4s - 37ms/step - fn: 36.0000 - fp: 29717.0000 - loss: 2.3629e-06 - precision: 0.0115 - recall: 0.9058 - tn: 197747.0000 - tp: 346.0000 - val_fn: 12.0000 - val_fp: 1039.0000 - val_loss: 0.0984 - val_precision: 0.0862 - val_recall: 0.8909 - val_tn: 55812.0000 - val_tp: 98.0000 ## Epoch 2/30 ## 112/112 - 1s - 9ms/step - fn: 30.0000 - fp: 6694.0000 - loss: 1.3508e-06 - precision: 0.0500 - recall: 0.9215 - tn: 220770.0000 - tp: 352.0000 - val_fn: 8.0000 - val_fp: 2418.0000 - val_loss: 0.1390 - val_precision: 0.0405 - val_recall: 0.9273 - val_tn: 54433.0000 - val_tp: 102.0000 ## Epoch 3/30 ## 112/112 - 0s - 2ms/step - fn: 27.0000 - fp: 9781.0000 - loss: 1.3090e-06 - precision: 0.0350 - recall: 0.9293 - tn: 217683.0000 - tp: 355.0000 - val_fn: 10.0000 - val_fp: 984.0000 - val_loss: 0.1063 - val_precision: 0.0923 - val_recall: 0.9091 - val_tn: 55867.0000 - val_tp: 100.0000 ## Epoch 4/30 ## 112/112 - 0s - 2ms/step - fn: 32.0000 - fp: 6894.0000 - loss: 1.1099e-06 - precision: 0.0483 - recall: 0.9162 - tn: 220570.0000 - tp: 350.0000 - val_fn: 9.0000 - val_fp: 1557.0000 - val_loss: 0.0857 - val_precision: 0.0609 - val_recall: 0.9182 - val_tn: 55294.0000 - val_tp: 101.0000 ## Epoch 5/30 ## 112/112 - 0s - 2ms/step - fn: 23.0000 - fp: 8892.0000 - loss: 9.2992e-07 - precision: 0.0388 - recall: 0.9398 - tn: 218572.0000 - tp: 359.0000 - val_fn: 7.0000 - val_fp: 1492.0000 - val_loss: 0.0665 - val_precision: 0.0646 - val_recall: 0.9364 - val_tn: 55359.0000 - val_tp: 103.0000 ## Epoch 6/30 ## 112/112 - 0s - 2ms/step - fn: 18.0000 - fp: 8209.0000 - loss: 7.7549e-07 - precision: 0.0425 - recall: 0.9529 - tn: 219255.0000 - tp: 364.0000 - val_fn: 7.0000 - val_fp: 2749.0000 - val_loss: 0.1010 - val_precision: 0.0361 - val_recall: 0.9364 - val_tn: 54102.0000 - val_tp: 103.0000 ## Epoch 7/30 ## 112/112 - 0s - 2ms/step - fn: 15.0000 - fp: 7189.0000 - loss: 6.9032e-07 - precision: 0.0486 - recall: 0.9607 - tn: 220275.0000 - tp: 367.0000 - val_fn: 12.0000 - val_fp: 934.0000 - val_loss: 0.0539 - val_precision: 0.0950 - val_recall: 0.8909 - val_tn: 55917.0000 - val_tp: 98.0000 ## Epoch 8/30 ## 112/112 - 0s - 2ms/step - fn: 18.0000 - fp: 7759.0000 - loss: 8.2432e-07 - precision: 0.0448 - recall: 0.9529 - tn: 219705.0000 - tp: 364.0000 - val_fn: 10.0000 - val_fp: 1481.0000 - val_loss: 0.0777 - val_precision: 0.0633 - val_recall: 0.9091 - val_tn: 55370.0000 - val_tp: 100.0000 ## Epoch 9/30 ## 112/112 - 0s - 2ms/step - fn: 16.0000 - fp: 7565.0000 - loss: 7.2956e-07 - precision: 0.0461 - recall: 0.9581 - tn: 219899.0000 - tp: 366.0000 - val_fn: 8.0000 - val_fp: 2075.0000 - val_loss: 0.0718 - val_precision: 0.0469 - val_recall: 0.9273 - val_tn: 54776.0000 - val_tp: 102.0000 ## Epoch 10/30 ## 112/112 - 0s - 2ms/step - fn: 10.0000 - fp: 7374.0000 - loss: 5.9522e-07 - precision: 0.0480 - recall: 0.9738 - tn: 220090.0000 - tp: 372.0000 - val_fn: 11.0000 - val_fp: 995.0000 - val_loss: 0.0425 - val_precision: 0.0905 - val_recall: 0.9000 - val_tn: 55856.0000 - val_tp: 99.0000 ## Epoch 11/30 ## 112/112 - 0s - 2ms/step - fn: 10.0000 - fp: 6586.0000 - loss: 6.2194e-07 - precision: 0.0535 - recall: 0.9738 - tn: 220878.0000 - tp: 372.0000 - val_fn: 9.0000 - val_fp: 5097.0000 - val_loss: 0.2103 - val_precision: 0.0194 - val_recall: 0.9182 - val_tn: 51754.0000 - val_tp: 101.0000 ## Epoch 12/30 ## 112/112 - 0s - 2ms/step - fn: 6.0000 - fp: 8285.0000 - loss: 6.0991e-07 - precision: 0.0434 - recall: 0.9843 - tn: 219179.0000 - tp: 376.0000 - val_fn: 9.0000 - val_fp: 1926.0000 - val_loss: 0.0781 - val_precision: 0.0498 - val_recall: 0.9182 - val_tn: 54925.0000 - val_tp: 101.0000 ## Epoch 13/30 ## 112/112 - 0s - 2ms/step - fn: 7.0000 - fp: 6864.0000 - loss: 5.7087e-07 - precision: 0.0518 - recall: 0.9817 - tn: 220600.0000 - tp: 375.0000 - val_fn: 8.0000 - val_fp: 1900.0000 - val_loss: 0.0837 - val_precision: 0.0509 - val_recall: 0.9273 - val_tn: 54951.0000 - val_tp: 102.0000 ## Epoch 14/30 ## 112/112 - 0s - 2ms/step - fn: 7.0000 - fp: 7663.0000 - loss: 5.7664e-07 - precision: 0.0467 - recall: 0.9817 - tn: 219801.0000 - tp: 375.0000 - val_fn: 8.0000 - val_fp: 1540.0000 - val_loss: 0.0656 - val_precision: 0.0621 - val_recall: 0.9273 - val_tn: 55311.0000 - val_tp: 102.0000 ## Epoch 15/30 ## 112/112 - 0s - 2ms/step - fn: 6.0000 - fp: 4607.0000 - loss: 3.9004e-07 - precision: 0.0755 - recall: 0.9843 - tn: 222857.0000 - tp: 376.0000 - val_fn: 11.0000 - val_fp: 2546.0000 - val_loss: 0.0994 - val_precision: 0.0374 - val_recall: 0.9000 - val_tn: 54305.0000 - val_tp: 99.0000 ## Epoch 16/30 ## 112/112 - 0s - 2ms/step - fn: 6.0000 - fp: 6195.0000 - loss: 5.1274e-07 - precision: 0.0572 - recall: 0.9843 - tn: 221269.0000 - tp: 376.0000 - val_fn: 9.0000 - val_fp: 1652.0000 - val_loss: 0.0693 - val_precision: 0.0576 - val_recall: 0.9182 - val_tn: 55199.0000 - val_tp: 101.0000 ## Epoch 17/30 ## 112/112 - 0s - 2ms/step - fn: 6.0000 - fp: 5367.0000 - loss: 4.6364e-07 - precision: 0.0655 - recall: 0.9843 - tn: 222097.0000 - tp: 376.0000 - val_fn: 10.0000 - val_fp: 1378.0000 - val_loss: 0.0575 - val_precision: 0.0677 - val_recall: 0.9091 - val_tn: 55473.0000 - val_tp: 100.0000 ## Epoch 18/30 ## 112/112 - 0s - 2ms/step - fn: 2.0000 - fp: 3696.0000 - loss: 2.9266e-07 - precision: 0.0932 - recall: 0.9948 - tn: 223768.0000 - tp: 380.0000 - val_fn: 11.0000 - val_fp: 824.0000 - val_loss: 0.0375 - val_precision: 0.1073 - val_recall: 0.9000 - val_tn: 56027.0000 - val_tp: 99.0000 ## Epoch 19/30 ## 112/112 - 0s - 2ms/step - fn: 3.0000 - fp: 3638.0000 - loss: 3.6945e-07 - precision: 0.0943 - recall: 0.9921 - tn: 223826.0000 - tp: 379.0000 - val_fn: 4.0000 - val_fp: 4639.0000 - val_loss: 0.5475 - val_precision: 0.0223 - val_recall: 0.9636 - val_tn: 52212.0000 - val_tp: 106.0000 ## Epoch 20/30 ## 112/112 - 0s - 2ms/step - fn: 26.0000 - fp: 11743.0000 - loss: 4.2168e-06 - precision: 0.0294 - recall: 0.9319 - tn: 215721.0000 - tp: 356.0000 - val_fn: 104.0000 - val_fp: 395.0000 - val_loss: 0.0507 - val_precision: 0.0150 - val_recall: 0.0545 - val_tn: 56456.0000 - val_tp: 6.0000 ## Epoch 21/30 ## 112/112 - 0s - 2ms/step - fn: 41.0000 - fp: 8860.0000 - loss: 5.8042e-06 - precision: 0.0371 - recall: 0.8927 - tn: 218604.0000 - tp: 341.0000 - val_fn: 10.0000 - val_fp: 2791.0000 - val_loss: 0.2183 - val_precision: 0.0346 - val_recall: 0.9091 - val_tn: 54060.0000 - val_tp: 100.0000 ## Epoch 22/30 ## 112/112 - 0s - 2ms/step - fn: 15.0000 - fp: 9031.0000 - loss: 1.0670e-06 - precision: 0.0391 - recall: 0.9607 - tn: 218433.0000 - tp: 367.0000 - val_fn: 10.0000 - val_fp: 1275.0000 - val_loss: 0.0764 - val_precision: 0.0727 - val_recall: 0.9091 - val_tn: 55576.0000 - val_tp: 100.0000 ## Epoch 23/30 ## 112/112 - 0s - 2ms/step - fn: 10.0000 - fp: 5141.0000 - loss: 6.9193e-07 - precision: 0.0675 - recall: 0.9738 - tn: 222323.0000 - tp: 372.0000 - val_fn: 11.0000 - val_fp: 1872.0000 - val_loss: 0.0824 - val_precision: 0.0502 - val_recall: 0.9000 - val_tn: 54979.0000 - val_tp: 99.0000 ## Epoch 24/30 ## 112/112 - 0s - 2ms/step - fn: 2.0000 - fp: 4232.0000 - loss: 4.2135e-07 - precision: 0.0824 - recall: 0.9948 - tn: 223232.0000 - tp: 380.0000 - val_fn: 11.0000 - val_fp: 976.0000 - val_loss: 0.0488 - val_precision: 0.0921 - val_recall: 0.9000 - val_tn: 55875.0000 - val_tp: 99.0000 ## Epoch 25/30 ## 112/112 - 0s - 2ms/step - fn: 3.0000 - fp: 4149.0000 - loss: 3.7232e-07 - precision: 0.0837 - recall: 0.9921 - tn: 223315.0000 - tp: 379.0000 - val_fn: 13.0000 - val_fp: 359.0000 - val_loss: 0.0229 - val_precision: 0.2127 - val_recall: 0.8818 - val_tn: 56492.0000 - val_tp: 97.0000 ## Epoch 26/30 ## 112/112 - 0s - 2ms/step - fn: 3.0000 - fp: 3757.0000 - loss: 3.5368e-07 - precision: 0.0916 - recall: 0.9921 - tn: 223707.0000 - tp: 379.0000 - val_fn: 12.0000 - val_fp: 1111.0000 - val_loss: 0.0569 - val_precision: 0.0811 - val_recall: 0.8909 - val_tn: 55740.0000 - val_tp: 98.0000 ## Epoch 27/30 ## 112/112 - 0s - 2ms/step - fn: 2.0000 - fp: 3220.0000 - loss: 2.5512e-07 - precision: 0.1056 - recall: 0.9948 - tn: 224244.0000 - tp: 380.0000 - val_fn: 11.0000 - val_fp: 839.0000 - val_loss: 0.0396 - val_precision: 0.1055 - val_recall: 0.9000 - val_tn: 56012.0000 - val_tp: 99.0000 ## Epoch 28/30 ## 112/112 - 0s - 2ms/step - fn: 0.0000e+00 - fp: 2287.0000 - loss: 1.8121e-07 - precision: 0.1431 - recall: 1.0000 - tn: 225177.0000 - tp: 382.0000 - val_fn: 12.0000 - val_fp: 425.0000 - val_loss: 0.0272 - val_precision: 0.1874 - val_recall: 0.8909 - val_tn: 56426.0000 - val_tp: 98.0000 ## Epoch 29/30 ## 112/112 - 0s - 2ms/step - fn: 0.0000e+00 - fp: 1435.0000 - loss: 1.3034e-07 - precision: 0.2102 - recall: 1.0000 - tn: 226029.0000 - tp: 382.0000 - val_fn: 13.0000 - val_fp: 321.0000 - val_loss: 0.0250 - val_precision: 0.2321 - val_recall: 0.8818 - val_tn: 56530.0000 - val_tp: 97.0000 ## Epoch 30/30 ## 112/112 - 0s - 2ms/step - fn: 0.0000e+00 - fp: 1365.0000 - loss: 1.2132e-07 - precision: 0.2187 - recall: 1.0000 - tn: 226099.0000 - tp: 382.0000 - val_fn: 12.0000 - val_fp: 241.0000 - val_loss: 0.0183 - val_precision: 0.2891 - val_recall: 0.8909 - val_tn: 56610.0000 - val_tp: 98.0000 val_pred <- model %>%   predict(val_features) %>%   { as.integer(. > 0.5) } ## 1781/1781 - 1s - 427us/step pred_correct <- val_df$Class == val_pred cat(sprintf(\"Validation accuracy: %.2f\", mean(pred_correct))) ## Validation accuracy: 1.00 fraudulent <- val_df$Class == 1  n_fraudulent_detected <- sum(fraudulent & pred_correct) n_fraudulent_missed <- sum(fraudulent & !pred_correct) n_legitimate_flagged <- sum(!fraudulent & !pred_correct)"},{"path":"https://keras3.posit.co/dev/articles/examples/structured_data/imbalanced_classification.html","id":"conclusions","dir":"Articles > Examples > Structured_data","previous_headings":"","what":"Conclusions","title":"Imbalanced classification: credit card fraud detection","text":"end training, 56,961 validation transactions, : Correctly identifying 98 fraudulent Missing 12 fraudulent transactions cost incorrectly flagging 241 legitimate transactions real world, one put even higher weight class 1, reflect False Negatives costly False Positives. Next time credit card gets declined online purchase – .","code":""},{"path":"https://keras3.posit.co/dev/articles/examples/structured_data/structured_data_classification_with_feature_space.html","id":"introduction","dir":"Articles > Examples > Structured_data","previous_headings":"","what":"Introduction","title":"Structured data classification with FeatureSpace","text":"example demonstrates structured data classification (also known tabular data classification), starting raw CSV file. data includes numerical features, integer categorical features, string categorical features. use utility layer_feature_space() index, preprocess, encode features. code adapted example Structured data classification scratch. previous example managed low-level feature preprocessing encoding Keras preprocessing layers, example delegate everything layer_feature_space(), making workflow extremely quick easy.","code":""},{"path":"https://keras3.posit.co/dev/articles/examples/structured_data/structured_data_classification_with_feature_space.html","id":"the-dataset","dir":"Articles > Examples > Structured_data","previous_headings":"Introduction","what":"The dataset","title":"Structured data classification with FeatureSpace","text":"dataset provided Cleveland Clinic Foundation Heart Disease. ’s CSV file 303 rows. row contains information patient (sample), column describes attribute patient (feature). use features predict whether patient heart disease (binary classification). ’s description feature:","code":""},{"path":"https://keras3.posit.co/dev/articles/examples/structured_data/structured_data_classification_with_feature_space.html","id":"setup","dir":"Articles > Examples > Structured_data","previous_headings":"","what":"Setup","title":"Structured data classification with FeatureSpace","text":"","code":"library(readr) library(dplyr, warn.conflicts = FALSE) library(keras3) library(tensorflow, exclude = c(\"shape\", \"set_random_seed\")) library(tfdatasets, exclude = \"shape\")  conflicted::conflicts_prefer(   keras3::shape(),   keras3::set_random_seed(),   dplyr::filter(),   .quiet = TRUE )  use_backend(\"tensorflow\")"},{"path":"https://keras3.posit.co/dev/articles/examples/structured_data/structured_data_classification_with_feature_space.html","id":"preparing-the-data","dir":"Articles > Examples > Structured_data","previous_headings":"","what":"Preparing the data","title":"Structured data classification with FeatureSpace","text":"Let’s download data load Pandas dataframe: dataset includes 303 samples 14 columns per sample (13 features, plus target label) ’s preview samples: last column, “target”, indicates whether patient heart disease (1) (0). Let’s split data training validation set: Let’s generate tf_dataset objects dataframe: tf_dataset yields tuple (input, target) input dictionary (named list) features target value 0 1: Let’s batch datasets:","code":"file_url <-   \"http://storage.googleapis.com/download.tensorflow.org/data/heart.csv\" df <- read_csv(file_url, col_types = cols(   oldpeak = col_double(),   thal = col_character(),   .default = col_integer() ))  # the dataset has two malformed rows, filter them out df <- df |> filter(!thal %in% c(\"1\", \"2\")) glimpse(df) ## Rows: 301 ## Columns: 14 ## $ age      <int> 63, 67, 67, 37, 41, 56, 62, 57, 63, 53, 57, 56, 56, 44, 5… ## $ sex      <int> 1, 1, 1, 1, 0, 1, 0, 0, 1, 1, 1, 0, 1, 1, 1, 1, 1, 1, 0, … ## $ cp       <int> 1, 4, 4, 3, 2, 2, 4, 4, 4, 4, 4, 2, 3, 2, 3, 3, 2, 4, 3, … ## $ trestbps <int> 145, 160, 120, 130, 130, 120, 140, 120, 130, 140, 140, 14… ## $ chol     <int> 233, 286, 229, 250, 204, 236, 268, 354, 254, 203, 192, 29… ## $ fbs      <int> 1, 0, 0, 0, 0, 0, 0, 0, 0, 1, 0, 0, 1, 0, 1, 0, 0, 0, 0, … ## $ restecg  <int> 2, 2, 2, 0, 2, 0, 2, 0, 2, 2, 0, 2, 2, 0, 0, 0, 0, 0, 0, … ## $ thalach  <int> 150, 108, 129, 187, 172, 178, 160, 163, 147, 155, 148, 15… ## $ exang    <int> 0, 1, 1, 0, 0, 0, 0, 1, 0, 1, 0, 0, 1, 0, 0, 0, 0, 0, 0, … ## $ oldpeak  <dbl> 2.3, 1.5, 2.6, 3.5, 1.4, 0.8, 3.6, 0.6, 1.4, 3.1, 0.4, 1.… ## $ slope    <int> 3, 2, 2, 3, 1, 1, 3, 1, 2, 3, 2, 2, 2, 1, 1, 1, 3, 1, 1, … ## $ ca       <int> 0, 3, 2, 0, 0, 0, 2, 0, 1, 0, 0, 0, 1, 0, 0, 0, 0, 0, 0, … ## $ thal     <chr> \"fixed\", \"normal\", \"reversible\", \"normal\", \"normal\", \"nor… ## $ target   <int> 0, 1, 0, 0, 0, 0, 1, 0, 1, 0, 0, 0, 1, 0, 0, 0, 0, 0, 0, … df ## # A tibble: 301 × 14 ##      age   sex    cp trestbps  chol   fbs restecg thalach exang oldpeak ##    <int> <int> <int>    <int> <int> <int>   <int>   <int> <int>   <dbl> ##  1    63     1     1      145   233     1       2     150     0     2.3 ##  2    67     1     4      160   286     0       2     108     1     1.5 ##  3    67     1     4      120   229     0       2     129     1     2.6 ##  4    37     1     3      130   250     0       0     187     0     3.5 ##  5    41     0     2      130   204     0       2     172     0     1.4 ##  6    56     1     2      120   236     0       0     178     0     0.8 ##  7    62     0     4      140   268     0       2     160     0     3.6 ##  8    57     0     4      120   354     0       0     163     1     0.6 ##  9    63     1     4      130   254     0       2     147     0     1.4 ## 10    53     1     4      140   203     1       2     155     1     3.1 ## # ℹ 291 more rows ## # ℹ 4 more variables: slope <int>, ca <int>, thal <chr>, target <int> val_idx <- nrow(df) %>% sample.int(., . * 0.2) val_df <- df[val_idx, ] train_df <- df[-val_idx, ]  cat(sprintf(   \"Using %d samples for training and %d for validation\",   nrow(train_df), nrow(val_df) )) ## Using 241 samples for training and 60 for validation dataframe_to_dataset <- function(df) {   labels <- df |> pull(target) |> as.integer()   inputs <- df |> select(-target) |> as.list()    ds <- tensor_slices_dataset(list(inputs, labels)) |>     dataset_shuffle(nrow(df))    ds }  train_ds <- dataframe_to_dataset(train_df) val_ds <- dataframe_to_dataset(val_df) c(x, y) %<-% iter_next(as_iterator(train_ds)) cat(\"Input: \"); str(x) cat(\"Target: \"); str(y) ## Input: List of 13 ##  $ age     :<tf.Tensor: shape=(), dtype=int32, numpy=68> ##  $ sex     :<tf.Tensor: shape=(), dtype=int32, numpy=1> ##  $ cp      :<tf.Tensor: shape=(), dtype=int32, numpy=4> ##  $ trestbps:<tf.Tensor: shape=(), dtype=int32, numpy=144> ##  $ chol    :<tf.Tensor: shape=(), dtype=int32, numpy=193> ##  $ fbs     :<tf.Tensor: shape=(), dtype=int32, numpy=1> ##  $ restecg :<tf.Tensor: shape=(), dtype=int32, numpy=0> ##  $ thalach :<tf.Tensor: shape=(), dtype=int32, numpy=141> ##  $ exang   :<tf.Tensor: shape=(), dtype=int32, numpy=0> ##  $ oldpeak :<tf.Tensor: shape=(), dtype=float32, numpy=3.4000000953674316> ##  $ slope   :<tf.Tensor: shape=(), dtype=int32, numpy=2> ##  $ ca      :<tf.Tensor: shape=(), dtype=int32, numpy=2> ##  $ thal    :<tf.Tensor: shape=(), dtype=string, numpy=b'reversible'> ## Target: <tf.Tensor: shape=(), dtype=int32, numpy=1> train_ds <- train_ds |> dataset_batch(32) val_ds <- val_ds |> dataset_batch(32)"},{"path":"https://keras3.posit.co/dev/articles/examples/structured_data/structured_data_classification_with_feature_space.html","id":"configuring-a-featurespace","dir":"Articles > Examples > Structured_data","previous_headings":"","what":"Configuring a FeatureSpace","title":"Structured data classification with FeatureSpace","text":"configure feature preprocessed, instantiate layer_feature_space(), pass dictionary (named list unique names) maps name features string describes feature type. “integer categorical” features \"FBS\", one “string categorical” feature (\"thal\"), numerical features, ’d like normalize – except \"age\", ’d like discretize number bins. also use crosses argument capture feature interactions categorical features, say, create additional features represent value co-occurrences categorical features. can compute feature crosses like arbitrary sets categorical features – just tuples two features. resulting co-occurences hashed fixed-sized vector, don’t need worry whether co-occurence space large.","code":"feature_space <- layer_feature_space(   features = list(     # Categorical features encoded as integers     sex = \"integer_categorical\",     cp = \"integer_categorical\",     fbs = \"integer_categorical\",     restecg = \"integer_categorical\",     exang = \"integer_categorical\",     ca = \"integer_categorical\",     # Categorical feature encoded as string     thal = \"string_categorical\",     # Numerical features to discretize     age = \"float_discretized\",     # Numerical features to normalize     trestbps = \"float_normalized\",     chol = \"float_normalized\",     thalach = \"float_normalized\",     oldpeak = \"float_normalized\",     slope = \"float_normalized\"   ),   # We create additional features by hashing   # value co-occurrences for the   # following groups of categorical features.   crosses = list(c(\"sex\", \"age\"), c(\"thal\", \"ca\")),   # The hashing space for these co-occurrences   # wil be 32-dimensional.   crossing_dim = 32,   # Our utility will one-hot encode all categorical   # features and concat all features into a single   # vector (one vector per sample).   output_mode = \"concat\" )"},{"path":"https://keras3.posit.co/dev/articles/examples/structured_data/structured_data_classification_with_feature_space.html","id":"further-customizing-a-featurespace","dir":"Articles > Examples > Structured_data","previous_headings":"","what":"Further customizing a FeatureSpace","title":"Structured data classification with FeatureSpace","text":"Specifying feature type via string name quick easy, sometimes may want configure preprocessing feature. instance, case, categorical features don’t large set possible values – ’s handful values per feature (e.g. 1 0 feature \"FBS\"), possible values represented training set. result, don’t need reserve index represent “vocabulary” values features – default behavior. , just specify num_oov_indices=0 features tell feature preprocessor skip “vocabulary” indexing. customizations access include specifying number bins discretizing features type \"float_discretized\", dimensionality hashing space feature crossing.","code":"feature_space <- layer_feature_space(   features = list(     # Categorical features encoded as integers     sex       = feature_integer_categorical(num_oov_indices = 0),     cp        = feature_integer_categorical(num_oov_indices = 0),     fbs       = feature_integer_categorical(num_oov_indices = 0),     restecg   = feature_integer_categorical(num_oov_indices = 0),     exang     = feature_integer_categorical(num_oov_indices = 0),     ca        = feature_integer_categorical(num_oov_indices = 0),     # Categorical feature encoded as string     thal      = feature_string_categorical(num_oov_indices = 0),     # Numerical features to discretize     age       = feature_float_discretized(num_bins = 30),     # Numerical features to normalize     trestbps  = feature_float_normalized(),     chol      = feature_float_normalized(),     thalach   = feature_float_normalized(),     oldpeak   = feature_float_normalized(),     slope     = feature_float_normalized()   ),   # Specify feature cross with a custom crossing dim.   crosses = list(     feature_cross(       feature_names = c(\"sex\", \"age\"),       crossing_dim = 64     ),     feature_cross(       feature_names = c(\"thal\", \"ca\"),       crossing_dim = 16     )   ),   output_mode = \"concat\" )"},{"path":"https://keras3.posit.co/dev/articles/examples/structured_data/structured_data_classification_with_feature_space.html","id":"adapt-the-featurespace-to-the-training-data","dir":"Articles > Examples > Structured_data","previous_headings":"","what":"Adapt the FeatureSpace to the training data","title":"Structured data classification with FeatureSpace","text":"start using FeatureSpace build model, adapt training data. adapt(), FeatureSpace : Index set possible values categorical features. Compute mean variance numerical features normalize. Compute value boundaries different bins numerical features discretize. Note adapt() called tf_dataset yields dicts (named lists) feature values – labels. point, FeatureSpace can called dict raw feature values, return single concatenate vector sample, combining encoded features feature crosses.","code":"train_ds_with_no_labels <- train_ds |> dataset_map(\\(x, y) x) feature_space |> adapt(train_ds_with_no_labels) c(x, y) %<-% iter_next(as_iterator(train_ds)) preprocessed_x <- feature_space(x) preprocessed_x ## tf.Tensor( ## [[0. 0. 0. ... 0. 0. 0.] ##  [0. 0. 0. ... 0. 0. 0.] ##  [0. 0. 0. ... 0. 0. 0.] ##  ... ##  [0. 0. 0. ... 0. 0. 0.] ##  [0. 0. 0. ... 0. 0. 0.] ##  [0. 0. 0. ... 0. 0. 0.]], shape=(32, 136), dtype=float32)"},{"path":"https://keras3.posit.co/dev/articles/examples/structured_data/structured_data_classification_with_feature_space.html","id":"two-ways-to-manage-preprocessing-as-part-of-the-tf-data-pipeline-or-in-the-model-itself","dir":"Articles > Examples > Structured_data","previous_headings":"","what":"Two ways to manage preprocessing: as part of the tf.data pipeline, or in the model itself","title":"Structured data classification with FeatureSpace","text":"two ways can leverage FeatureSpace:","code":""},{"path":"https://keras3.posit.co/dev/articles/examples/structured_data/structured_data_classification_with_feature_space.html","id":"asynchronous-preprocessing-in-tf-data","dir":"Articles > Examples > Structured_data","previous_headings":"Two ways to manage preprocessing: as part of the tf.data pipeline, or in the model itself","what":"Asynchronous preprocessing in tf.data","title":"Structured data classification with FeatureSpace","text":"can make part data pipeline, model. enables asynchronous parallel preprocessing data CPU hits model. ’re training GPU TPU, want speed preprocessing. Usually, always right thing training.","code":""},{"path":"https://keras3.posit.co/dev/articles/examples/structured_data/structured_data_classification_with_feature_space.html","id":"synchronous-preprocessing-in-the-model","dir":"Articles > Examples > Structured_data","previous_headings":"Two ways to manage preprocessing: as part of the tf.data pipeline, or in the model itself","what":"Synchronous preprocessing in the model","title":"Structured data classification with FeatureSpace","text":"can make part model. means model expect dicts raw feature values, preprocessing batch done synchronously (blocking manner) rest forward pass. want end--end model can process raw feature values – keep mind model able run CPU, since types feature preprocessing (e.g. string preprocessing) GPU TPU compatible. GPU / TPU performance-sensitive settings. general, want -model preprocessing inference CPU. case, apply FeatureSpace tf.data pipeline training, inference end--end model includes FeatureSpace. Let’s create training validation dataset preprocessed batches:","code":"preprocessed_train_ds <- train_ds |>   dataset_map(\\(x, y) list(feature_space(x), y),               num_parallel_calls = tf$data$AUTOTUNE) |>   dataset_prefetch(tf$data$AUTOTUNE)  preprocessed_val_ds <- val_ds |>   dataset_map(\\(x, y) list(feature_space(x), y),               num_parallel_calls = tf$data$AUTOTUNE) |>   dataset_prefetch(tf$data$AUTOTUNE)"},{"path":"https://keras3.posit.co/dev/articles/examples/structured_data/structured_data_classification_with_feature_space.html","id":"build-a-model","dir":"Articles > Examples > Structured_data","previous_headings":"","what":"Build a model","title":"Structured data classification with FeatureSpace","text":"Time build model – rather two models: training model expects preprocessed features (one sample = one vector) inference model expects raw features (one sample = dict raw feature values)","code":"dict_inputs <- feature_space$get_inputs() encoded_features <- feature_space$get_encoded_features()  predictions <- encoded_features |>   layer_dense(32, activation=\"relu\") |>   layer_dropout(0.5) |>   layer_dense(1, activation=\"sigmoid\")  training_model <- keras_model(inputs = encoded_features,                               outputs = predictions) training_model |> compile(optimizer = \"adam\",                           loss = \"binary_crossentropy\",                           metrics = \"accuracy\")  inference_model <- keras_model(inputs = dict_inputs,                                outputs = predictions)"},{"path":"https://keras3.posit.co/dev/articles/examples/structured_data/structured_data_classification_with_feature_space.html","id":"train-the-model","dir":"Articles > Examples > Structured_data","previous_headings":"","what":"Train the model","title":"Structured data classification with FeatureSpace","text":"Let’s train model 20 epochs. Note feature preprocessing happening part tf.data pipeline, part model. quickly get 80% validation accuracy.","code":"training_model |> fit(   preprocessed_train_ds,   epochs = 20,   validation_data = preprocessed_val_ds,   verbose = 2 ) ## Epoch 1/20 ## 8/8 - 2s - 262ms/step - accuracy: 0.4108 - loss: 0.7398 - val_accuracy: 0.3667 - val_loss: 0.7218 ## Epoch 2/20 ## 8/8 - 0s - 26ms/step - accuracy: 0.5934 - loss: 0.6798 - val_accuracy: 0.5333 - val_loss: 0.6670 ## Epoch 3/20 ## 8/8 - 0s - 26ms/step - accuracy: 0.6473 - loss: 0.6384 - val_accuracy: 0.6333 - val_loss: 0.6207 ## Epoch 4/20 ## 8/8 - 0s - 26ms/step - accuracy: 0.7012 - loss: 0.5829 - val_accuracy: 0.7167 - val_loss: 0.5800 ## Epoch 5/20 ## 8/8 - 0s - 27ms/step - accuracy: 0.6846 - loss: 0.5780 - val_accuracy: 0.7500 - val_loss: 0.5445 ## Epoch 6/20 ## 8/8 - 0s - 26ms/step - accuracy: 0.7718 - loss: 0.5595 - val_accuracy: 0.7667 - val_loss: 0.5134 ## Epoch 7/20 ## 8/8 - 0s - 26ms/step - accuracy: 0.7801 - loss: 0.5166 - val_accuracy: 0.8167 - val_loss: 0.4868 ## Epoch 8/20 ## 8/8 - 0s - 26ms/step - accuracy: 0.7718 - loss: 0.5053 - val_accuracy: 0.8333 - val_loss: 0.4626 ## Epoch 9/20 ## 8/8 - 0s - 28ms/step - accuracy: 0.8091 - loss: 0.4774 - val_accuracy: 0.8333 - val_loss: 0.4401 ## Epoch 10/20 ## 8/8 - 0s - 27ms/step - accuracy: 0.8050 - loss: 0.4682 - val_accuracy: 0.8500 - val_loss: 0.4197 ## Epoch 11/20 ## 8/8 - 0s - 26ms/step - accuracy: 0.8133 - loss: 0.4541 - val_accuracy: 0.8667 - val_loss: 0.4011 ## Epoch 12/20 ## 8/8 - 0s - 27ms/step - accuracy: 0.8216 - loss: 0.4360 - val_accuracy: 0.8667 - val_loss: 0.3843 ## Epoch 13/20 ## 8/8 - 0s - 26ms/step - accuracy: 0.8174 - loss: 0.4081 - val_accuracy: 0.8667 - val_loss: 0.3684 ## Epoch 14/20 ## 8/8 - 0s - 27ms/step - accuracy: 0.8423 - loss: 0.4034 - val_accuracy: 0.8667 - val_loss: 0.3531 ## Epoch 15/20 ## 8/8 - 0s - 26ms/step - accuracy: 0.8174 - loss: 0.3956 - val_accuracy: 0.8833 - val_loss: 0.3397 ## Epoch 16/20 ## 8/8 - 0s - 27ms/step - accuracy: 0.8423 - loss: 0.3952 - val_accuracy: 0.8833 - val_loss: 0.3274 ## Epoch 17/20 ## 8/8 - 0s - 27ms/step - accuracy: 0.8548 - loss: 0.3445 - val_accuracy: 0.8833 - val_loss: 0.3169 ## Epoch 18/20 ## 8/8 - 0s - 27ms/step - accuracy: 0.8382 - loss: 0.3609 - val_accuracy: 0.8833 - val_loss: 0.3080 ## Epoch 19/20 ## 8/8 - 0s - 26ms/step - accuracy: 0.8714 - loss: 0.3412 - val_accuracy: 0.8833 - val_loss: 0.2986 ## Epoch 20/20 ## 8/8 - 0s - 26ms/step - accuracy: 0.8506 - loss: 0.3541 - val_accuracy: 0.9000 - val_loss: 0.2905"},{"path":"https://keras3.posit.co/dev/articles/examples/structured_data/structured_data_classification_with_feature_space.html","id":"inference-on-new-data-with-the-end-to-end-model","dir":"Articles > Examples > Structured_data","previous_headings":"","what":"Inference on new data with the end-to-end model","title":"Structured data classification with FeatureSpace","text":"Now, can use inference model (includes FeatureSpace) make predictions based dicts raw features values, follows:","code":"sample <- list(   age = 60,   sex = 1,   cp = 1,   trestbps = 145,   chol = 233,   fbs = 1,   restecg = 2,   thalach = 150,   exang = 0,   oldpeak = 2.3,   slope = 3,   ca = 0,   thal = \"fixed\" )  input_dict <- lapply(sample, \\(x) op_convert_to_tensor(array(x))) predictions <- inference_model |> predict(input_dict) ## 1/1 - 0s - 298ms/step glue::glue(r\"---(   This particular patient had a {(100 * predictions) |> signif(3)}% probability   of having a heart disease, as evaluated by our model. )---\") ## This particular patient had a 48.6% probability ## of having a heart disease, as evaluated by our model."},{"path":"https://keras3.posit.co/dev/articles/examples/timeseries/timeseries_anomaly_detection.html","id":"introduction","dir":"Articles > Examples > Timeseries","previous_headings":"","what":"Introduction","title":"Timeseries anomaly detection using an Autoencoder","text":"script demonstrates can use reconstruction convolutional autoencoder model detect anomalies timeseries data.","code":""},{"path":"https://keras3.posit.co/dev/articles/examples/timeseries/timeseries_anomaly_detection.html","id":"setup","dir":"Articles > Examples > Timeseries","previous_headings":"","what":"Setup","title":"Timeseries anomaly detection using an Autoencoder","text":"","code":"library(dplyr, warn.conflicts = FALSE) library(ggplot2) theme_set(theme_minimal())  library(listarrays) library(tfdatasets, exclude = c(\"shape\")) library(keras3)"},{"path":"https://keras3.posit.co/dev/articles/examples/timeseries/timeseries_anomaly_detection.html","id":"load-the-data","dir":"Articles > Examples > Timeseries","previous_headings":"","what":"Load the data","title":"Timeseries anomaly detection using an Autoencoder","text":"use Numenta Anomaly Benchmark(NAB) dataset. provides artificial timeseries data containing labeled anomalous periods behavior. Data ordered, timestamped, single-valued metrics. use art_daily_small_noise.csv file training art_daily_jumpsup.csv file testing. simplicity dataset allows us demonstrate anomaly detection.","code":"get_data <- function(url_suffix) {   url_root <- \"https://raw.githubusercontent.com/numenta/NAB/master/data/\"   url <- paste0(url_root, url_suffix)   file <- get_file(origin = url) # cache file locally   # parse csv; 2 columns with types: datetime (T), double (d)   readr::read_csv(file, col_types = \"Td\") }  df_small_noise   <- get_data(\"artificialNoAnomaly/art_daily_small_noise.csv\") df_daily_jumpsup <- get_data(\"artificialWithAnomaly/art_daily_jumpsup.csv\")"},{"path":"https://keras3.posit.co/dev/articles/examples/timeseries/timeseries_anomaly_detection.html","id":"quick-look-at-the-data","dir":"Articles > Examples > Timeseries","previous_headings":"","what":"Quick look at the data","title":"Timeseries anomaly detection using an Autoencoder","text":"","code":"df_small_noise ## # A tibble: 4,032 × 2 ##    timestamp           value ##    <dttm>              <dbl> ##  1 2014-04-01 00:00:00  18.3 ##  2 2014-04-01 00:05:00  22.0 ##  3 2014-04-01 00:10:00  18.6 ##  4 2014-04-01 00:15:00  22.0 ##  5 2014-04-01 00:20:00  21.9 ##  6 2014-04-01 00:25:00  21.2 ##  7 2014-04-01 00:30:00  20.6 ##  8 2014-04-01 00:35:00  20.3 ##  9 2014-04-01 00:40:00  21.5 ## 10 2014-04-01 00:45:00  19.2 ## # ℹ 4,022 more rows df_daily_jumpsup ## # A tibble: 4,032 × 2 ##    timestamp           value ##    <dttm>              <dbl> ##  1 2014-04-01 00:00:00  19.8 ##  2 2014-04-01 00:05:00  20.5 ##  3 2014-04-01 00:10:00  20.0 ##  4 2014-04-01 00:15:00  21.5 ##  5 2014-04-01 00:20:00  20.2 ##  6 2014-04-01 00:25:00  19.9 ##  7 2014-04-01 00:30:00  21.7 ##  8 2014-04-01 00:35:00  20.9 ##  9 2014-04-01 00:40:00  18.4 ## 10 2014-04-01 00:45:00  18.7 ## # ℹ 4,022 more rows"},{"path":[]},{"path":"https://keras3.posit.co/dev/articles/examples/timeseries/timeseries_anomaly_detection.html","id":"timeseries-data-without-anomalies","dir":"Articles > Examples > Timeseries","previous_headings":"Visualize the data","what":"Timeseries data without anomalies","title":"Timeseries anomaly detection using an Autoencoder","text":"use following data training.","code":"plot_ts <- function(df) {   ggplot(df, aes(x = timestamp, y = value)) + geom_line() +     scale_x_datetime(date_breaks = \"1 day\", date_labels = \"%b-%d\") }  plot_ts(df_small_noise) + ggtitle(\"Without Anomaly\")"},{"path":"https://keras3.posit.co/dev/articles/examples/timeseries/timeseries_anomaly_detection.html","id":"timeseries-data-with-anomalies","dir":"Articles > Examples > Timeseries","previous_headings":"Visualize the data","what":"Timeseries data with anomalies","title":"Timeseries anomaly detection using an Autoencoder","text":"use following data testing see sudden jump data detected anomaly.","code":"plot_ts(df_daily_jumpsup) + ggtitle(\"With Anomaly\")"},{"path":"https://keras3.posit.co/dev/articles/examples/timeseries/timeseries_anomaly_detection.html","id":"prepare-training-data","dir":"Articles > Examples > Timeseries","previous_headings":"","what":"Prepare training data","title":"Timeseries anomaly detection using an Autoencoder","text":"Get data values training timeseries data file normalize value data. value every 5 mins 14 days. 24 * 60 / 5 = 288 timesteps per day 288 * 14 = 4032 data points total","code":"df_train <- df_small_noise |>   mutate(value = (value - mean(value)) / sd(value))  cat(\"Number of training samples:\", nrow(df_train), \"\\n\") ## Number of training samples: 4032"},{"path":"https://keras3.posit.co/dev/articles/examples/timeseries/timeseries_anomaly_detection.html","id":"create-sequences","dir":"Articles > Examples > Timeseries","previous_headings":"Prepare training data","what":"Create sequences","title":"Timeseries anomaly detection using an Autoencoder","text":"Create sequences combining TIME_STEPS contiguous data values training data.","code":"TIME_STEPS <- 288  as_dataset <- function(df) {   x <- as.matrix(df$value)   ds <- timeseries_dataset_from_array(x, NULL, sequence_length = TIME_STEPS)   # Because the dataset is small, cast TF Dataset to an R array for convenience.   ds |> as_array_iterator() |> iterate() |> bind_on_rows() }  x_train <- as_dataset(df_train) writeLines(sprintf(\"Training input shape: (%s)\", toString(dim(x_train)))) ## Training input shape: (3745, 288, 1)"},{"path":"https://keras3.posit.co/dev/articles/examples/timeseries/timeseries_anomaly_detection.html","id":"build-a-model","dir":"Articles > Examples > Timeseries","previous_headings":"","what":"Build a model","title":"Timeseries anomaly detection using an Autoencoder","text":"build convolutional reconstruction autoencoder model. model take input shape (batch_size, sequence_length, num_features) return output shape. case, sequence_length 288 num_features 1.","code":"model <- keras_model_sequential(input_shape = c(TIME_STEPS, 1)) |>   layer_conv_1d(     filters = 32, kernel_size = 7, padding = \"same\",     strides = 2, activation = \"relu\"   ) |>   layer_dropout(rate = 0.2) |>   layer_conv_1d(     filters = 16, kernel_size = 7, padding = \"same\",     strides = 2, activation = \"relu\"   ) |>   layer_conv_1d_transpose(     filters = 16, kernel_size = 7, padding = \"same\",     strides = 2, activation = \"relu\"   ) |>   layer_dropout(rate = 0.2) |>   layer_conv_1d_transpose(     filters = 32, kernel_size = 7, padding = \"same\",     strides = 2, activation = \"relu\"   ) |>   layer_conv_1d_transpose(filters = 1, kernel_size = 7, padding = \"same\")  model |> compile(optimizer=optimizer_adam(learning_rate=0.001), loss=\"mse\") model ## Model: \"sequential\" ## ┏━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━┓ ## ┃ Layer (type)                    ┃ Output Shape           ┃       Param # ┃ ## ┡━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━┩ ## │ conv1d (Conv1D)                 │ (None, 144, 32)        │           256 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ dropout (Dropout)               │ (None, 144, 32)        │             0 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ conv1d_1 (Conv1D)               │ (None, 72, 16)         │         3,600 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ conv1d_transpose                │ (None, 144, 16)        │         1,808 │ ## │ (Conv1DTranspose)               │                        │               │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ dropout_1 (Dropout)             │ (None, 144, 16)        │             0 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ conv1d_transpose_1              │ (None, 288, 32)        │         3,616 │ ## │ (Conv1DTranspose)               │                        │               │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ conv1d_transpose_2              │ (None, 288, 1)         │           225 │ ## │ (Conv1DTranspose)               │                        │               │ ## └─────────────────────────────────┴────────────────────────┴───────────────┘ ##  Total params: 9,505 (37.13 KB) ##  Trainable params: 9,505 (37.13 KB) ##  Non-trainable params: 0 (0.00 B)"},{"path":"https://keras3.posit.co/dev/articles/examples/timeseries/timeseries_anomaly_detection.html","id":"train-the-model","dir":"Articles > Examples > Timeseries","previous_headings":"","what":"Train the model","title":"Timeseries anomaly detection using an Autoencoder","text":"Please note using x_train input target since reconstruction model. Let’s plot training validation loss see training went.","code":"history = model |> fit(   x_train, x_train,   epochs = 50,   validation_split = 0.1,   callbacks = c(     callback_early_stopping(       monitor = \"val_loss\", patience = 5, mode = \"min\"     )   ) ) ## Epoch 1/50 ## 106/106 - 4s - 39ms/step - loss: 0.1863 - val_loss: 0.0316 ## Epoch 2/50 ## 106/106 - 0s - 2ms/step - loss: 0.0333 - val_loss: 0.0233 ## Epoch 3/50 ## 106/106 - 0s - 2ms/step - loss: 0.0248 - val_loss: 0.0201 ## Epoch 4/50 ## 106/106 - 0s - 2ms/step - loss: 0.0209 - val_loss: 0.0186 ## Epoch 5/50 ## 106/106 - 0s - 2ms/step - loss: 0.0179 - val_loss: 0.0146 ## Epoch 6/50 ## 106/106 - 0s - 2ms/step - loss: 0.0150 - val_loss: 0.0112 ## Epoch 7/50 ## 106/106 - 0s - 2ms/step - loss: 0.0127 - val_loss: 0.0094 ## Epoch 8/50 ## 106/106 - 0s - 2ms/step - loss: 0.0109 - val_loss: 0.0088 ## Epoch 9/50 ## 106/106 - 0s - 2ms/step - loss: 0.0096 - val_loss: 0.0083 ## Epoch 10/50 ## 106/106 - 0s - 2ms/step - loss: 0.0086 - val_loss: 0.0069 ## Epoch 11/50 ## 106/106 - 0s - 2ms/step - loss: 0.0078 - val_loss: 0.0061 ## Epoch 12/50 ## 106/106 - 0s - 2ms/step - loss: 0.0073 - val_loss: 0.0057 ## Epoch 13/50 ## 106/106 - 0s - 2ms/step - loss: 0.0068 - val_loss: 0.0053 ## Epoch 14/50 ## 106/106 - 0s - 2ms/step - loss: 0.0064 - val_loss: 0.0050 ## Epoch 15/50 ## 106/106 - 0s - 2ms/step - loss: 0.0061 - val_loss: 0.0046 ## Epoch 16/50 ## 106/106 - 0s - 2ms/step - loss: 0.0058 - val_loss: 0.0040 ## Epoch 17/50 ## 106/106 - 0s - 2ms/step - loss: 0.0056 - val_loss: 0.0041 ## Epoch 18/50 ## 106/106 - 0s - 2ms/step - loss: 0.0053 - val_loss: 0.0035 ## Epoch 19/50 ## 106/106 - 0s - 2ms/step - loss: 0.0050 - val_loss: 0.0037 ## Epoch 20/50 ## 106/106 - 0s - 2ms/step - loss: 0.0048 - val_loss: 0.0032 ## Epoch 21/50 ## 106/106 - 0s - 2ms/step - loss: 0.0046 - val_loss: 0.0031 ## Epoch 22/50 ## 106/106 - 0s - 2ms/step - loss: 0.0044 - val_loss: 0.0032 ## Epoch 23/50 ## 106/106 - 0s - 2ms/step - loss: 0.0042 - val_loss: 0.0033 ## Epoch 24/50 ## 106/106 - 0s - 2ms/step - loss: 0.0041 - val_loss: 0.0031 ## Epoch 25/50 ## 106/106 - 0s - 2ms/step - loss: 0.0039 - val_loss: 0.0033 ## Epoch 26/50 ## 106/106 - 0s - 2ms/step - loss: 0.0037 - val_loss: 0.0030 ## Epoch 27/50 ## 106/106 - 0s - 2ms/step - loss: 0.0035 - val_loss: 0.0032 ## Epoch 28/50 ## 106/106 - 0s - 2ms/step - loss: 0.0034 - val_loss: 0.0025 ## Epoch 29/50 ## 106/106 - 0s - 2ms/step - loss: 0.0033 - val_loss: 0.0028 ## Epoch 30/50 ## 106/106 - 0s - 2ms/step - loss: 0.0032 - val_loss: 0.0030 ## Epoch 31/50 ## 106/106 - 0s - 2ms/step - loss: 0.0031 - val_loss: 0.0024 ## Epoch 32/50 ## 106/106 - 0s - 2ms/step - loss: 0.0030 - val_loss: 0.0028 ## Epoch 33/50 ## 106/106 - 0s - 2ms/step - loss: 0.0029 - val_loss: 0.0027 ## Epoch 34/50 ## 106/106 - 0s - 2ms/step - loss: 0.0028 - val_loss: 0.0022 ## Epoch 35/50 ## 106/106 - 0s - 2ms/step - loss: 0.0027 - val_loss: 0.0024 ## Epoch 36/50 ## 106/106 - 0s - 2ms/step - loss: 0.0026 - val_loss: 0.0025 ## Epoch 37/50 ## 106/106 - 0s - 2ms/step - loss: 0.0026 - val_loss: 0.0027 ## Epoch 38/50 ## 106/106 - 0s - 2ms/step - loss: 0.0025 - val_loss: 0.0024 ## Epoch 39/50 ## 106/106 - 0s - 2ms/step - loss: 0.0025 - val_loss: 0.0024 plot(history)"},{"path":"https://keras3.posit.co/dev/articles/examples/timeseries/timeseries_anomaly_detection.html","id":"detecting-anomalies","dir":"Articles > Examples > Timeseries","previous_headings":"","what":"Detecting anomalies","title":"Timeseries anomaly detection using an Autoencoder","text":"detect anomalies determining well model can reconstruct input data. Find MAE loss training samples. Find max MAE loss value. worst model performed trying reconstruct sample. make threshold anomaly detection. reconstruction loss sample greater threshold value can infer model seeing pattern isn’t familiar . label sample anomaly.","code":"# Get train MAE loss. x_train_pred <- model |> predict(x_train) ## 118/118 - 0s - 4ms/step train_mae_loss <- apply(abs(x_train_pred - x_train), 1, mean)  hist(train_mae_loss, breaks = 50) # Get reconstruction loss threshold. threshold <- max(train_mae_loss) cat(\"Reconstruction error threshold: \", threshold, \"\\n\") ## Reconstruction error threshold:  0.03994837"},{"path":"https://keras3.posit.co/dev/articles/examples/timeseries/timeseries_anomaly_detection.html","id":"compare-recontruction","dir":"Articles > Examples > Timeseries","previous_headings":"Detecting anomalies","what":"Compare recontruction","title":"Timeseries anomaly detection using an Autoencoder","text":"Just fun, let’s see model recontructed first sample. 288 timesteps day 1 training dataset.","code":"# Checking how the first sequence is learnt plot(NULL, NULL, ylab = 'Value',      xlim = c(0, TIME_STEPS),      ylim = range(c(x_train[1,,], x_train_pred[1,,]))) lines(x_train[1,,]) lines(x_train_pred[1,,], col = 'red') legend(\"topleft\", lty = 1,        legend = c(\"actual\", \"predicted\"),        col = c(\"black\", \"red\"))"},{"path":"https://keras3.posit.co/dev/articles/examples/timeseries/timeseries_anomaly_detection.html","id":"prepare-test-data","dir":"Articles > Examples > Timeseries","previous_headings":"Detecting anomalies","what":"Prepare test data","title":"Timeseries anomaly detection using an Autoencoder","text":"","code":"df_test <- df_daily_jumpsup |>   mutate(value =            (value - mean(df_small_noise$value)) /              sd(df_small_noise$value))  df_test |> head() plot_ts(df_test) # Create sequences from test values. x_test <- as_dataset(df_test)  # Get test MAE loss. x_test_pred <- model |> predict(x_test) test_mae_loss <- apply(abs(x_test_pred - x_test), 1, mean)  hist(test_mae_loss, breaks = 50, xlab = \"test MAE loss\", ylab = \"No of samples\") # Detect all the samples which are anomalies. anomalies <- test_mae_loss > threshold cat(\"Number of anomaly samples:\", sum(anomalies), \"\\n\") cat(\"Indices of anomaly samples:\", which(anomalies), \"\\n\", fill = TRUE) ## # A tibble: 6 × 2 ##   timestamp            value ##   <dttm>               <dbl> ## 1 2014-04-01 00:00:00 -0.808 ## 2 2014-04-01 00:05:00 -0.781 ## 3 2014-04-01 00:10:00 -0.801 ## 4 2014-04-01 00:15:00 -0.746 ## 5 2014-04-01 00:20:00 -0.792 ## 6 2014-04-01 00:25:00 -0.802 ## 118/118 - 0s - 808us/step ## Number of anomaly samples: 483 ## Indices of anomaly samples: 216 217 218 219 220 221 396 507 793 794 795 797 ## 799 861 974 1659 1944 1945 1946 1947 2016 2020 2036 2044 2046 2052 2053 ## 2054 2056 2057 2058 2060 2061 2062 2064 2065 2066 2068 2069 2070 2072 2073 ## 2074 2076 2077 2078 2080 2081 2082 2084 2085 2086 2088 2089 2090 2092 2093 ## 2094 2096 2097 2098 2100 2101 2102 2104 2105 2108 2112 2113 2116 2117 2120 ## 2121 2124 2126 2141 2521 2522 2523 2525 2546 2698 2700 2702 2703 2704 2705 ## 2706 2707 2708 2709 2710 2711 2712 2713 2714 2715 2716 2717 2718 2719 2720 ## 2721 2722 2723 2724 2725 2726 2727 2728 2729 2730 2731 2732 2733 2734 2735 ## 2736 2737 2738 2739 2740 2741 2742 2743 2744 2745 2746 2747 2748 2749 2750 ## 2751 2752 2753 2754 2755 2756 2757 2758 2759 2760 2761 2762 2763 2764 2765 ## 2766 2767 2768 2769 2770 2771 2772 2773 2774 2775 2776 2777 2778 2779 2780 ## 2781 2782 2783 2784 2785 2786 2787 2788 2789 2790 2791 2792 2793 2794 2795 ## 2796 2797 2798 2799 2800 2801 2802 2803 2804 2805 2806 2807 2808 2809 2810 ## 2811 2812 2813 2814 2815 2816 2817 2818 2819 2820 2821 2822 2823 2824 2825 ## 2826 2827 2828 2829 2830 2831 2832 2833 2834 2835 2836 2837 2838 2839 2840 ## 2841 2842 2843 2844 2845 2846 2847 2848 2849 2850 2851 2852 2853 2854 2855 ## 2856 2857 2858 2859 2860 2861 2862 2863 2864 2865 2866 2867 2868 2869 2870 ## 2871 2872 2873 2874 2875 2876 2877 2878 2879 2880 2881 2882 2883 2884 2885 ## 2886 2887 2888 2889 2890 2891 2892 2893 2894 2895 2896 2897 2898 2899 2900 ## 2901 2902 2903 2904 2905 2906 2907 2908 2909 2910 2911 2912 2913 2914 2915 ## 2916 2917 2918 2919 2920 2921 2922 2923 2924 2925 2926 2927 2928 2929 2930 ## 2931 2932 2933 2934 2935 2936 2937 2938 2939 2940 2941 2942 2943 2944 2945 ## 2946 2947 2948 2949 2950 2951 2952 2953 2954 2955 2956 2957 2958 2959 2960 ## 2961 2962 2963 2964 2965 2966 2967 2968 2969 2970 2971 2972 2973 2974 2975 ## 2976 2977 2978 2979 2980 2981 2982 2983 2984 2985 2986 2987 2988 2989 2990 ## 2991 2992 2993 2994 2995 2996 2997 2998 2999 3000 3001 3002 3003 3004 3005 ## 3006 3007 3008 3009 3010 3011 3012 3013 3014 3015 3016 3017 3018 3019 3020 ## 3021 3022 3023 3024 3025 3026 3027 3028 3029 3030 3031 3032 3033 3034 3035 ## 3036 3037 3038 3039 3040 3041 3042 3043 3044 3045 3046 3047 3048 3049 3050 ## 3051 3052 3053 3054 3055 3056 3057 3058 3059 3060 3061 3062 3063 3064 3065 ## 3066 3067 3068 3069 3070 3071 3072 3073 3074 3075 3076 3077 3078 3079 3080 ## 3081 3082 3083 3084 3085 3086 3087 3088 3089 3090 3091 3092 3093 3094 3095 ## 3096 3097 3098 3099 3100 3101"},{"path":"https://keras3.posit.co/dev/articles/examples/timeseries/timeseries_anomaly_detection.html","id":"plot-anomalies","dir":"Articles > Examples > Timeseries","previous_headings":"","what":"Plot anomalies","title":"Timeseries anomaly detection using an Autoencoder","text":"now know samples data anomalies. , find corresponding timestamps original test data. using following method : Let’s say time_steps = 3 10 training values. x_train look like : 0, 1, 2 1, 2, 3 2, 3, 4 3, 4, 5 4, 5, 6 5, 6, 7 6, 7, 8 7, 8, 9 except initial final time_steps-1 data values, appear time_steps number samples. , know samples [(3, 4, 5), (4, 5, 6), (5, 6, 7)] anomalies, can say data point 5 anomaly. Let’s overlay anomalies original test data plot.","code":"is_anomaly <- test_mae_loss > threshold is_anomaly <- is_anomaly &   zoo::rollsum(is_anomaly, TIME_STEPS,                align = \"right\", na.pad = TRUE) >= TIME_STEPS  with(df_test, {   plot(value ~ timestamp, type = 'l', xaxt = 'n', las = 2)   axis.POSIXct(1, at = seq(timestamp[1], tail(timestamp, 1), by = \"days\"),                format = \"%b-%d\") })  with(df_test[which(is_anomaly),], {   points(value ~ timestamp, col = \"red\") })"},{"path":"https://keras3.posit.co/dev/articles/examples/timeseries/timeseries_classification_from_scratch.html","id":"introduction","dir":"Articles > Examples > Timeseries","previous_headings":"","what":"Introduction","title":"Timeseries classification from scratch","text":"example shows timeseries classification scratch, starting raw CSV timeseries files disk. demonstrate workflow FordA dataset UCR/UEA archive.","code":""},{"path":"https://keras3.posit.co/dev/articles/examples/timeseries/timeseries_classification_from_scratch.html","id":"setup","dir":"Articles > Examples > Timeseries","previous_headings":"","what":"Setup","title":"Timeseries classification from scratch","text":"","code":"library(keras3) use_backend(\"jax\")"},{"path":[]},{"path":"https://keras3.posit.co/dev/articles/examples/timeseries/timeseries_classification_from_scratch.html","id":"dataset-description","dir":"Articles > Examples > Timeseries","previous_headings":"Load the data: the FordA dataset","what":"Dataset description","title":"Timeseries classification from scratch","text":"dataset using called FordA. data comes UCR archive. dataset contains 3601 training instances another 1320 testing instances. timeseries corresponds measurement engine noise captured motor sensor. task, goal automatically detect presence specific issue engine. problem balanced binary classification task. full description dataset can found .","code":""},{"path":"https://keras3.posit.co/dev/articles/examples/timeseries/timeseries_classification_from_scratch.html","id":"read-the-tsv-data","dir":"Articles > Examples > Timeseries","previous_headings":"Load the data: the FordA dataset","what":"Read the TSV data","title":"Timeseries classification from scratch","text":"use FordA_TRAIN file training FordA_TEST file testing. simplicity dataset allows us demonstrate effectively use ConvNets timeseries classification. file, first column corresponds label.","code":"get_data <- function(path) {   if(path |> startsWith(\"https://\"))     path <- get_file(origin = path)  # cache file locally    data <- readr::read_tsv(     path, col_names = FALSE,     # Each row is: one integer (the label),     # followed by 500 doubles (the timeseries)     col_types = paste0(\"i\", strrep(\"d\", 500))   )    y <- as.matrix(data[[1]])   x <- as.matrix(data[,-1])   dimnames(x) <- dimnames(y) <- NULL    list(x, y) }  root_url <- \"https://raw.githubusercontent.com/hfawaz/cd-diagram/master/FordA/\" c(x_train, y_train) %<-% get_data(paste0(root_url, \"FordA_TRAIN.tsv\")) c(x_test, y_test) %<-% get_data(paste0(root_url, \"FordA_TEST.tsv\"))  str(keras3:::named_list(   x_train, y_train,   x_test, y_test )) ## List of 4 ##  $ x_train: num [1:3601, 1:500] -0.797 0.805 0.728 -0.234 -0.171 ... ##  $ y_train: int [1:3601, 1] -1 1 -1 -1 -1 1 1 1 1 1 ... ##  $ x_test : num [1:1320, 1:500] -0.14 0.334 0.717 1.24 -1.159 ... ##  $ y_test : int [1:1320, 1] -1 -1 -1 1 -1 1 -1 -1 1 1 ..."},{"path":"https://keras3.posit.co/dev/articles/examples/timeseries/timeseries_classification_from_scratch.html","id":"visualize-the-data","dir":"Articles > Examples > Timeseries","previous_headings":"","what":"Visualize the data","title":"Timeseries classification from scratch","text":"visualize one timeseries example class dataset.","code":"plot(NULL, main = \"Timeseries Data\",      xlab = \"Timepoints\",  ylab = \"Values\",      xlim = c(1, ncol(x_test)),      ylim = range(x_test)) grid() lines(x_test[match(-1, y_test), ], col = \"blue\") lines(x_test[match( 1, y_test), ], col = \"red\") legend(\"topright\", legend=c(\"label -1\", \"label 1\"), col=c(\"blue\", \"red\"), lty=1)"},{"path":"https://keras3.posit.co/dev/articles/examples/timeseries/timeseries_classification_from_scratch.html","id":"standardize-the-data","dir":"Articles > Examples > Timeseries","previous_headings":"","what":"Standardize the data","title":"Timeseries classification from scratch","text":"timeseries already single length (500). However, values usually various ranges. ideal neural network; general seek make input values normalized. specific dataset, data already z-normalized: timeseries sample mean equal zero standard deviation equal one. type normalization common timeseries classification problems, see Bagnall et al. (2016). Note timeseries data used univariate, meaning one channel per timeseries example. therefore transform timeseries multivariate one one channel using simple reshaping via numpy. allow us construct model easily applicable multivariate time series. Finally, order use sparse_categorical_crossentropy, count number classes beforehand. Now shuffle training set using validation_split option later training. Standardize labels positive integers. expected labels 0 1.","code":"dim(x_train) <- c(dim(x_train), 1) dim(x_test) <- c(dim(x_test), 1) num_classes <- length(unique(y_train)) c(x_train, y_train) %<-% listarrays::shuffle_rows(x_train, y_train) # idx <- sample.int(nrow(x_train)) # x_train %<>% .[idx,, ,drop = FALSE] # y_train %<>% .[idx,  ,drop = FALSE] y_train[y_train == -1L] <- 0L y_test[y_test == -1L] <- 0L"},{"path":"https://keras3.posit.co/dev/articles/examples/timeseries/timeseries_classification_from_scratch.html","id":"build-a-model","dir":"Articles > Examples > Timeseries","previous_headings":"","what":"Build a model","title":"Timeseries classification from scratch","text":"build Fully Convolutional Neural Network originally proposed paper. implementation based TF 2 version provided . following hyperparameters (kernel_size, filters, usage BatchNorm) found via random search using KerasTuner. plot chunk unnamed-chunk-9","code":"make_model <- function(input_shape) {   inputs <- keras_input(input_shape)    outputs <- inputs |>     # conv1     layer_conv_1d(filters = 64, kernel_size = 3, padding = \"same\") |>     layer_batch_normalization() |>     layer_activation_relu() |>     # conv2     layer_conv_1d(filters = 64, kernel_size = 3, padding = \"same\") |>     layer_batch_normalization() |>     layer_activation_relu() |>     # conv3     layer_conv_1d(filters = 64, kernel_size = 3, padding = \"same\") |>     layer_batch_normalization() |>     layer_activation_relu() |>     # pooling     layer_global_average_pooling_1d() |>     # final output     layer_dense(num_classes, activation = \"softmax\")    keras_model(inputs, outputs) }  model <- make_model(input_shape = dim(x_train)[-1]) model ## Model: \"functional\" ## ┏━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━┳━━━━━━━┓ ## ┃ Layer (type)                ┃ Output Shape          ┃    Param # ┃ Trai… ┃ ## ┡━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━╇━━━━━━━┩ ## │ input_layer (InputLayer)    │ (None, 500, 1)        │          0 │   -   │ ## ├─────────────────────────────┼───────────────────────┼────────────┼───────┤ ## │ conv1d (Conv1D)             │ (None, 500, 64)       │        256 │   Y   │ ## ├─────────────────────────────┼───────────────────────┼────────────┼───────┤ ## │ batch_normalization         │ (None, 500, 64)       │        256 │   Y   │ ## │ (BatchNormalization)        │                       │            │       │ ## ├─────────────────────────────┼───────────────────────┼────────────┼───────┤ ## │ re_lu (ReLU)                │ (None, 500, 64)       │          0 │   -   │ ## ├─────────────────────────────┼───────────────────────┼────────────┼───────┤ ## │ conv1d_1 (Conv1D)           │ (None, 500, 64)       │     12,352 │   Y   │ ## ├─────────────────────────────┼───────────────────────┼────────────┼───────┤ ## │ batch_normalization_1       │ (None, 500, 64)       │        256 │   Y   │ ## │ (BatchNormalization)        │                       │            │       │ ## ├─────────────────────────────┼───────────────────────┼────────────┼───────┤ ## │ re_lu_1 (ReLU)              │ (None, 500, 64)       │          0 │   -   │ ## ├─────────────────────────────┼───────────────────────┼────────────┼───────┤ ## │ conv1d_2 (Conv1D)           │ (None, 500, 64)       │     12,352 │   Y   │ ## ├─────────────────────────────┼───────────────────────┼────────────┼───────┤ ## │ batch_normalization_2       │ (None, 500, 64)       │        256 │   Y   │ ## │ (BatchNormalization)        │                       │            │       │ ## ├─────────────────────────────┼───────────────────────┼────────────┼───────┤ ## │ re_lu_2 (ReLU)              │ (None, 500, 64)       │          0 │   -   │ ## ├─────────────────────────────┼───────────────────────┼────────────┼───────┤ ## │ global_average_pooling1d    │ (None, 64)            │          0 │   -   │ ## │ (GlobalAveragePooling1D)    │                       │            │       │ ## ├─────────────────────────────┼───────────────────────┼────────────┼───────┤ ## │ dense (Dense)               │ (None, 2)             │        130 │   Y   │ ## └─────────────────────────────┴───────────────────────┴────────────┴───────┘ ##  Total params: 25,858 (101.01 KB) ##  Trainable params: 25,474 (99.51 KB) ##  Non-trainable params: 384 (1.50 KB) plot(model, show_shapes = TRUE)"},{"path":"https://keras3.posit.co/dev/articles/examples/timeseries/timeseries_classification_from_scratch.html","id":"train-the-model","dir":"Articles > Examples > Timeseries","previous_headings":"","what":"Train the model","title":"Timeseries classification from scratch","text":"","code":"epochs <- 500 batch_size <- 32  callbacks <- c(   callback_model_checkpoint(     \"best_model.keras\", save_best_only = TRUE,     monitor = \"val_loss\"   ),   callback_reduce_lr_on_plateau(     monitor = \"val_loss\", factor = 0.5,     patience = 20, min_lr = 0.0001   ),   callback_early_stopping(     monitor = \"val_loss\", patience = 50,     verbose = 1   ) )   model |> compile(   optimizer = \"adam\",   loss = \"sparse_categorical_crossentropy\",   metrics = \"sparse_categorical_accuracy\" )  history <- model |> fit(   x_train, y_train,   batch_size = batch_size,   epochs = epochs,   callbacks = callbacks,   validation_split = 0.2 ) ## Epoch 1/500 ## 90/90 - 3s - 32ms/step - loss: 0.5310 - sparse_categorical_accuracy: 0.7191 - val_loss: 0.7907 - val_sparse_categorical_accuracy: 0.4896 - learning_rate: 1.0000e-03 ## Epoch 2/500 ## 90/90 - 1s - 11ms/step - loss: 0.4799 - sparse_categorical_accuracy: 0.7590 - val_loss: 0.9127 - val_sparse_categorical_accuracy: 0.4896 - learning_rate: 1.0000e-03 ## Epoch 3/500 ## 90/90 - 0s - 2ms/step - loss: 0.4590 - sparse_categorical_accuracy: 0.7757 - val_loss: 0.8196 - val_sparse_categorical_accuracy: 0.4896 - learning_rate: 1.0000e-03 ## Epoch 4/500 ## 90/90 - 0s - 2ms/step - loss: 0.4088 - sparse_categorical_accuracy: 0.7983 - val_loss: 0.6425 - val_sparse_categorical_accuracy: 0.5534 - learning_rate: 1.0000e-03 ## Epoch 5/500 ## 90/90 - 0s - 2ms/step - loss: 0.4141 - sparse_categorical_accuracy: 0.7889 - val_loss: 0.5519 - val_sparse_categorical_accuracy: 0.6588 - learning_rate: 1.0000e-03 ## Epoch 6/500 ## 90/90 - 0s - 2ms/step - loss: 0.3929 - sparse_categorical_accuracy: 0.8163 - val_loss: 0.4225 - val_sparse_categorical_accuracy: 0.8294 - learning_rate: 1.0000e-03 ## Epoch 7/500 ## 90/90 - 0s - 1ms/step - loss: 0.3870 - sparse_categorical_accuracy: 0.8146 - val_loss: 0.6112 - val_sparse_categorical_accuracy: 0.6949 - learning_rate: 1.0000e-03 ## Epoch 8/500 ## 90/90 - 0s - 1ms/step - loss: 0.3717 - sparse_categorical_accuracy: 0.8247 - val_loss: 0.4257 - val_sparse_categorical_accuracy: 0.7559 - learning_rate: 1.0000e-03 ## Epoch 9/500 ## 90/90 - 0s - 2ms/step - loss: 0.3698 - sparse_categorical_accuracy: 0.8306 - val_loss: 0.3653 - val_sparse_categorical_accuracy: 0.8350 - learning_rate: 1.0000e-03 ## Epoch 10/500 ## 90/90 - 0s - 1ms/step - loss: 0.3634 - sparse_categorical_accuracy: 0.8260 - val_loss: 0.4390 - val_sparse_categorical_accuracy: 0.7836 - learning_rate: 1.0000e-03 ## Epoch 11/500 ## 90/90 - 0s - 1ms/step - loss: 0.3676 - sparse_categorical_accuracy: 0.8281 - val_loss: 0.4205 - val_sparse_categorical_accuracy: 0.7822 - learning_rate: 1.0000e-03 ## Epoch 12/500 ## 90/90 - 0s - 1ms/step - loss: 0.3467 - sparse_categorical_accuracy: 0.8406 - val_loss: 0.3697 - val_sparse_categorical_accuracy: 0.8460 - learning_rate: 1.0000e-03 ## Epoch 13/500 ## 90/90 - 0s - 2ms/step - loss: 0.3310 - sparse_categorical_accuracy: 0.8594 - val_loss: 0.3341 - val_sparse_categorical_accuracy: 0.8530 - learning_rate: 1.0000e-03 ## Epoch 14/500 ## 90/90 - 0s - 1ms/step - loss: 0.3350 - sparse_categorical_accuracy: 0.8514 - val_loss: 0.5723 - val_sparse_categorical_accuracy: 0.7198 - learning_rate: 1.0000e-03 ## Epoch 15/500 ## 90/90 - 0s - 1ms/step - loss: 0.3295 - sparse_categorical_accuracy: 0.8497 - val_loss: 1.0273 - val_sparse_categorical_accuracy: 0.5465 - learning_rate: 1.0000e-03 ## Epoch 16/500 ## 90/90 - 0s - 2ms/step - loss: 0.3244 - sparse_categorical_accuracy: 0.8587 - val_loss: 0.3219 - val_sparse_categorical_accuracy: 0.8544 - learning_rate: 1.0000e-03 ## Epoch 17/500 ## 90/90 - 0s - 1ms/step - loss: 0.3082 - sparse_categorical_accuracy: 0.8646 - val_loss: 0.4187 - val_sparse_categorical_accuracy: 0.7906 - learning_rate: 1.0000e-03 ## Epoch 18/500 ## 90/90 - 0s - 1ms/step - loss: 0.3022 - sparse_categorical_accuracy: 0.8715 - val_loss: 0.6302 - val_sparse_categorical_accuracy: 0.6824 - learning_rate: 1.0000e-03 ## Epoch 19/500 ## 90/90 - 0s - 1ms/step - loss: 0.2996 - sparse_categorical_accuracy: 0.8764 - val_loss: 0.3537 - val_sparse_categorical_accuracy: 0.8447 - learning_rate: 1.0000e-03 ## Epoch 20/500 ## 90/90 - 0s - 1ms/step - loss: 0.3053 - sparse_categorical_accuracy: 0.8715 - val_loss: 0.4925 - val_sparse_categorical_accuracy: 0.7309 - learning_rate: 1.0000e-03 ## Epoch 21/500 ## 90/90 - 0s - 1ms/step - loss: 0.2827 - sparse_categorical_accuracy: 0.8826 - val_loss: 0.3893 - val_sparse_categorical_accuracy: 0.7809 - learning_rate: 1.0000e-03 ## Epoch 22/500 ## 90/90 - 0s - 2ms/step - loss: 0.2892 - sparse_categorical_accuracy: 0.8795 - val_loss: 0.3128 - val_sparse_categorical_accuracy: 0.8863 - learning_rate: 1.0000e-03 ## Epoch 23/500 ## 90/90 - 0s - 2ms/step - loss: 0.2718 - sparse_categorical_accuracy: 0.8889 - val_loss: 0.3083 - val_sparse_categorical_accuracy: 0.8571 - learning_rate: 1.0000e-03 ## Epoch 24/500 ## 90/90 - 0s - 1ms/step - loss: 0.2691 - sparse_categorical_accuracy: 0.8934 - val_loss: 0.5609 - val_sparse_categorical_accuracy: 0.7420 - learning_rate: 1.0000e-03 ## Epoch 25/500 ## 90/90 - 0s - 1ms/step - loss: 0.2635 - sparse_categorical_accuracy: 0.8948 - val_loss: 0.5264 - val_sparse_categorical_accuracy: 0.7573 - learning_rate: 1.0000e-03 ## Epoch 26/500 ## 90/90 - 0s - 1ms/step - loss: 0.3109 - sparse_categorical_accuracy: 0.8667 - val_loss: 0.3934 - val_sparse_categorical_accuracy: 0.7878 - learning_rate: 1.0000e-03 ## Epoch 27/500 ## 90/90 - 0s - 2ms/step - loss: 0.2599 - sparse_categorical_accuracy: 0.8969 - val_loss: 0.2974 - val_sparse_categorical_accuracy: 0.8558 - learning_rate: 1.0000e-03 ## Epoch 28/500 ## 90/90 - 0s - 1ms/step - loss: 0.2805 - sparse_categorical_accuracy: 0.8823 - val_loss: 0.4026 - val_sparse_categorical_accuracy: 0.8183 - learning_rate: 1.0000e-03 ## Epoch 29/500 ## 90/90 - 0s - 2ms/step - loss: 0.2517 - sparse_categorical_accuracy: 0.8948 - val_loss: 0.2699 - val_sparse_categorical_accuracy: 0.8849 - learning_rate: 1.0000e-03 ## Epoch 30/500 ## 90/90 - 0s - 1ms/step - loss: 0.2581 - sparse_categorical_accuracy: 0.8920 - val_loss: 0.2743 - val_sparse_categorical_accuracy: 0.8752 - learning_rate: 1.0000e-03 ## Epoch 31/500 ## 90/90 - 0s - 1ms/step - loss: 0.2550 - sparse_categorical_accuracy: 0.8986 - val_loss: 0.2936 - val_sparse_categorical_accuracy: 0.8655 - learning_rate: 1.0000e-03 ## Epoch 32/500 ## 90/90 - 0s - 2ms/step - loss: 0.2528 - sparse_categorical_accuracy: 0.8934 - val_loss: 0.2528 - val_sparse_categorical_accuracy: 0.8904 - learning_rate: 1.0000e-03 ## Epoch 33/500 ## 90/90 - 0s - 1ms/step - loss: 0.2390 - sparse_categorical_accuracy: 0.9062 - val_loss: 0.2961 - val_sparse_categorical_accuracy: 0.8571 - learning_rate: 1.0000e-03 ## Epoch 34/500 ## 90/90 - 0s - 1ms/step - loss: 0.2327 - sparse_categorical_accuracy: 0.9049 - val_loss: 0.3039 - val_sparse_categorical_accuracy: 0.8558 - learning_rate: 1.0000e-03 ## Epoch 35/500 ## 90/90 - 0s - 1ms/step - loss: 0.2447 - sparse_categorical_accuracy: 0.9021 - val_loss: 0.3123 - val_sparse_categorical_accuracy: 0.8710 - learning_rate: 1.0000e-03 ## Epoch 36/500 ## 90/90 - 0s - 1ms/step - loss: 0.2340 - sparse_categorical_accuracy: 0.9069 - val_loss: 0.2912 - val_sparse_categorical_accuracy: 0.8821 - learning_rate: 1.0000e-03 ## Epoch 37/500 ## 90/90 - 0s - 1ms/step - loss: 0.2352 - sparse_categorical_accuracy: 0.8997 - val_loss: 0.3983 - val_sparse_categorical_accuracy: 0.8017 - learning_rate: 1.0000e-03 ## Epoch 38/500 ## 90/90 - 0s - 1ms/step - loss: 0.2294 - sparse_categorical_accuracy: 0.9139 - val_loss: 0.3342 - val_sparse_categorical_accuracy: 0.8447 - learning_rate: 1.0000e-03 ## Epoch 39/500 ## 90/90 - 0s - 1ms/step - loss: 0.2268 - sparse_categorical_accuracy: 0.9135 - val_loss: 0.4730 - val_sparse_categorical_accuracy: 0.7975 - learning_rate: 1.0000e-03 ## Epoch 40/500 ## 90/90 - 0s - 1ms/step - loss: 0.2202 - sparse_categorical_accuracy: 0.9111 - val_loss: 0.3007 - val_sparse_categorical_accuracy: 0.8738 - learning_rate: 1.0000e-03 ## Epoch 41/500 ## 90/90 - 0s - 1ms/step - loss: 0.2064 - sparse_categorical_accuracy: 0.9233 - val_loss: 0.2924 - val_sparse_categorical_accuracy: 0.8849 - learning_rate: 1.0000e-03 ## Epoch 42/500 ## 90/90 - 0s - 1ms/step - loss: 0.1966 - sparse_categorical_accuracy: 0.9316 - val_loss: 0.8303 - val_sparse_categorical_accuracy: 0.6935 - learning_rate: 1.0000e-03 ## Epoch 43/500 ## 90/90 - 0s - 2ms/step - loss: 0.2074 - sparse_categorical_accuracy: 0.9212 - val_loss: 0.2334 - val_sparse_categorical_accuracy: 0.9043 - learning_rate: 1.0000e-03 ## Epoch 44/500 ## 90/90 - 0s - 1ms/step - loss: 0.1961 - sparse_categorical_accuracy: 0.9313 - val_loss: 0.2343 - val_sparse_categorical_accuracy: 0.9182 - learning_rate: 1.0000e-03 ## Epoch 45/500 ## 90/90 - 0s - 1ms/step - loss: 0.1923 - sparse_categorical_accuracy: 0.9302 - val_loss: 1.1167 - val_sparse_categorical_accuracy: 0.6755 - learning_rate: 1.0000e-03 ## Epoch 46/500 ## 90/90 - 0s - 1ms/step - loss: 0.1823 - sparse_categorical_accuracy: 0.9347 - val_loss: 0.8034 - val_sparse_categorical_accuracy: 0.6935 - learning_rate: 1.0000e-03 ## Epoch 47/500 ## 90/90 - 0s - 1ms/step - loss: 0.1635 - sparse_categorical_accuracy: 0.9444 - val_loss: 0.7104 - val_sparse_categorical_accuracy: 0.7115 - learning_rate: 1.0000e-03 ## Epoch 48/500 ## 90/90 - 0s - 2ms/step - loss: 0.1587 - sparse_categorical_accuracy: 0.9434 - val_loss: 0.2001 - val_sparse_categorical_accuracy: 0.9196 - learning_rate: 1.0000e-03 ## Epoch 49/500 ## 90/90 - 0s - 1ms/step - loss: 0.1492 - sparse_categorical_accuracy: 0.9483 - val_loss: 0.3399 - val_sparse_categorical_accuracy: 0.8294 - learning_rate: 1.0000e-03 ## Epoch 50/500 ## 90/90 - 0s - 1ms/step - loss: 0.1738 - sparse_categorical_accuracy: 0.9382 - val_loss: 0.2680 - val_sparse_categorical_accuracy: 0.8890 - learning_rate: 1.0000e-03 ## Epoch 51/500 ## 90/90 - 0s - 1ms/step - loss: 0.1530 - sparse_categorical_accuracy: 0.9486 - val_loss: 0.2891 - val_sparse_categorical_accuracy: 0.8544 - learning_rate: 1.0000e-03 ## Epoch 52/500 ## 90/90 - 0s - 1ms/step - loss: 0.1320 - sparse_categorical_accuracy: 0.9538 - val_loss: 1.7182 - val_sparse_categorical_accuracy: 0.7032 - learning_rate: 1.0000e-03 ## Epoch 53/500 ## 90/90 - 0s - 1ms/step - loss: 0.1277 - sparse_categorical_accuracy: 0.9559 - val_loss: 1.4269 - val_sparse_categorical_accuracy: 0.7254 - learning_rate: 1.0000e-03 ## Epoch 54/500 ## 90/90 - 0s - 1ms/step - loss: 0.1249 - sparse_categorical_accuracy: 0.9611 - val_loss: 0.7962 - val_sparse_categorical_accuracy: 0.7365 - learning_rate: 1.0000e-03 ## Epoch 55/500 ## 90/90 - 0s - 1ms/step - loss: 0.1214 - sparse_categorical_accuracy: 0.9611 - val_loss: 0.2158 - val_sparse_categorical_accuracy: 0.9071 - learning_rate: 1.0000e-03 ## Epoch 56/500 ## 90/90 - 0s - 1ms/step - loss: 0.1258 - sparse_categorical_accuracy: 0.9604 - val_loss: 0.6349 - val_sparse_categorical_accuracy: 0.7642 - learning_rate: 1.0000e-03 ## Epoch 57/500 ## 90/90 - 0s - 1ms/step - loss: 0.1268 - sparse_categorical_accuracy: 0.9604 - val_loss: 0.2129 - val_sparse_categorical_accuracy: 0.9071 - learning_rate: 1.0000e-03 ## Epoch 58/500 ## 90/90 - 0s - 1ms/step - loss: 0.1088 - sparse_categorical_accuracy: 0.9639 - val_loss: 0.7215 - val_sparse_categorical_accuracy: 0.6657 - learning_rate: 1.0000e-03 ## Epoch 59/500 ## 90/90 - 0s - 2ms/step - loss: 0.1219 - sparse_categorical_accuracy: 0.9569 - val_loss: 0.1600 - val_sparse_categorical_accuracy: 0.9320 - learning_rate: 1.0000e-03 ## Epoch 60/500 ## 90/90 - 0s - 1ms/step - loss: 0.1190 - sparse_categorical_accuracy: 0.9642 - val_loss: 0.3411 - val_sparse_categorical_accuracy: 0.8558 - learning_rate: 1.0000e-03 ## Epoch 61/500 ## 90/90 - 0s - 1ms/step - loss: 0.1177 - sparse_categorical_accuracy: 0.9653 - val_loss: 0.3984 - val_sparse_categorical_accuracy: 0.8502 - learning_rate: 1.0000e-03 ## Epoch 62/500 ## 90/90 - 0s - 1ms/step - loss: 0.1309 - sparse_categorical_accuracy: 0.9542 - val_loss: 0.3462 - val_sparse_categorical_accuracy: 0.8211 - learning_rate: 1.0000e-03 ## Epoch 63/500 ## 90/90 - 0s - 1ms/step - loss: 0.1291 - sparse_categorical_accuracy: 0.9514 - val_loss: 0.2303 - val_sparse_categorical_accuracy: 0.8960 - learning_rate: 1.0000e-03 ## Epoch 64/500 ## 90/90 - 0s - 1ms/step - loss: 0.1200 - sparse_categorical_accuracy: 0.9628 - val_loss: 0.1735 - val_sparse_categorical_accuracy: 0.9223 - learning_rate: 1.0000e-03 ## Epoch 65/500 ## 90/90 - 0s - 1ms/step - loss: 0.1138 - sparse_categorical_accuracy: 0.9625 - val_loss: 0.5628 - val_sparse_categorical_accuracy: 0.7920 - learning_rate: 1.0000e-03 ## Epoch 66/500 ## 90/90 - 0s - 2ms/step - loss: 0.1065 - sparse_categorical_accuracy: 0.9667 - val_loss: 0.1540 - val_sparse_categorical_accuracy: 0.9417 - learning_rate: 1.0000e-03 ## Epoch 67/500 ## 90/90 - 0s - 1ms/step - loss: 0.1156 - sparse_categorical_accuracy: 0.9618 - val_loss: 0.2319 - val_sparse_categorical_accuracy: 0.9043 - learning_rate: 1.0000e-03 ## Epoch 68/500 ## 90/90 - 0s - 1ms/step - loss: 0.1100 - sparse_categorical_accuracy: 0.9597 - val_loss: 0.8668 - val_sparse_categorical_accuracy: 0.7406 - learning_rate: 1.0000e-03 ## Epoch 69/500 ## 90/90 - 0s - 2ms/step - loss: 0.1165 - sparse_categorical_accuracy: 0.9618 - val_loss: 0.1460 - val_sparse_categorical_accuracy: 0.9515 - learning_rate: 1.0000e-03 ## Epoch 70/500 ## 90/90 - 0s - 1ms/step - loss: 0.1178 - sparse_categorical_accuracy: 0.9608 - val_loss: 0.8988 - val_sparse_categorical_accuracy: 0.7046 - learning_rate: 1.0000e-03 ## Epoch 71/500 ## 90/90 - 0s - 1ms/step - loss: 0.1068 - sparse_categorical_accuracy: 0.9642 - val_loss: 0.3645 - val_sparse_categorical_accuracy: 0.8585 - learning_rate: 1.0000e-03 ## Epoch 72/500 ## 90/90 - 0s - 1ms/step - loss: 0.1106 - sparse_categorical_accuracy: 0.9653 - val_loss: 0.2813 - val_sparse_categorical_accuracy: 0.8932 - learning_rate: 1.0000e-03 ## Epoch 73/500 ## 90/90 - 0s - 1ms/step - loss: 0.1117 - sparse_categorical_accuracy: 0.9632 - val_loss: 1.8641 - val_sparse_categorical_accuracy: 0.6741 - learning_rate: 1.0000e-03 ## Epoch 74/500 ## 90/90 - 0s - 1ms/step - loss: 0.1058 - sparse_categorical_accuracy: 0.9625 - val_loss: 1.4979 - val_sparse_categorical_accuracy: 0.6893 - learning_rate: 1.0000e-03 ## Epoch 75/500 ## 90/90 - 0s - 1ms/step - loss: 0.1223 - sparse_categorical_accuracy: 0.9594 - val_loss: 0.1508 - val_sparse_categorical_accuracy: 0.9473 - learning_rate: 1.0000e-03 ## Epoch 76/500 ## 90/90 - 0s - 2ms/step - loss: 0.1140 - sparse_categorical_accuracy: 0.9601 - val_loss: 0.1400 - val_sparse_categorical_accuracy: 0.9320 - learning_rate: 1.0000e-03 ## Epoch 77/500 ## 90/90 - 0s - 1ms/step - loss: 0.1050 - sparse_categorical_accuracy: 0.9639 - val_loss: 0.2169 - val_sparse_categorical_accuracy: 0.9126 - learning_rate: 1.0000e-03 ## Epoch 78/500 ## 90/90 - 0s - 1ms/step - loss: 0.0986 - sparse_categorical_accuracy: 0.9712 - val_loss: 0.3240 - val_sparse_categorical_accuracy: 0.8724 - learning_rate: 1.0000e-03 ## Epoch 79/500 ## 90/90 - 0s - 1ms/step - loss: 0.1081 - sparse_categorical_accuracy: 0.9632 - val_loss: 0.1525 - val_sparse_categorical_accuracy: 0.9362 - learning_rate: 1.0000e-03 ## Epoch 80/500 ## 90/90 - 0s - 1ms/step - loss: 0.0995 - sparse_categorical_accuracy: 0.9667 - val_loss: 0.3123 - val_sparse_categorical_accuracy: 0.8641 - learning_rate: 1.0000e-03 ## Epoch 81/500 ## 90/90 - 0s - 1ms/step - loss: 0.1058 - sparse_categorical_accuracy: 0.9632 - val_loss: 0.2851 - val_sparse_categorical_accuracy: 0.8779 - learning_rate: 1.0000e-03 ## Epoch 82/500 ## 90/90 - 0s - 1ms/step - loss: 0.0958 - sparse_categorical_accuracy: 0.9681 - val_loss: 0.1417 - val_sparse_categorical_accuracy: 0.9362 - learning_rate: 1.0000e-03 ## Epoch 83/500 ## 90/90 - 0s - 1ms/step - loss: 0.1005 - sparse_categorical_accuracy: 0.9660 - val_loss: 0.7325 - val_sparse_categorical_accuracy: 0.7157 - learning_rate: 1.0000e-03 ## Epoch 84/500 ## 90/90 - 0s - 2ms/step - loss: 0.0966 - sparse_categorical_accuracy: 0.9684 - val_loss: 0.1322 - val_sparse_categorical_accuracy: 0.9404 - learning_rate: 1.0000e-03 ## Epoch 85/500 ## 90/90 - 0s - 2ms/step - loss: 0.1018 - sparse_categorical_accuracy: 0.9667 - val_loss: 0.1202 - val_sparse_categorical_accuracy: 0.9473 - learning_rate: 1.0000e-03 ## Epoch 86/500 ## 90/90 - 0s - 1ms/step - loss: 0.0995 - sparse_categorical_accuracy: 0.9649 - val_loss: 0.2956 - val_sparse_categorical_accuracy: 0.8766 - learning_rate: 1.0000e-03 ## Epoch 87/500 ## 90/90 - 0s - 1ms/step - loss: 0.1068 - sparse_categorical_accuracy: 0.9632 - val_loss: 0.3050 - val_sparse_categorical_accuracy: 0.8710 - learning_rate: 1.0000e-03 ## Epoch 88/500 ## 90/90 - 0s - 1ms/step - loss: 0.1002 - sparse_categorical_accuracy: 0.9674 - val_loss: 0.1675 - val_sparse_categorical_accuracy: 0.9196 - learning_rate: 1.0000e-03 ## Epoch 89/500 ## 90/90 - 0s - 1ms/step - loss: 0.1042 - sparse_categorical_accuracy: 0.9608 - val_loss: 0.7157 - val_sparse_categorical_accuracy: 0.7115 - learning_rate: 1.0000e-03 ## Epoch 90/500 ## 90/90 - 0s - 1ms/step - loss: 0.0998 - sparse_categorical_accuracy: 0.9694 - val_loss: 0.1576 - val_sparse_categorical_accuracy: 0.9251 - learning_rate: 1.0000e-03 ## Epoch 91/500 ## 90/90 - 0s - 1ms/step - loss: 0.1187 - sparse_categorical_accuracy: 0.9583 - val_loss: 0.4125 - val_sparse_categorical_accuracy: 0.8031 - learning_rate: 1.0000e-03 ## Epoch 92/500 ## 90/90 - 0s - 1ms/step - loss: 0.0973 - sparse_categorical_accuracy: 0.9688 - val_loss: 0.3084 - val_sparse_categorical_accuracy: 0.8696 - learning_rate: 1.0000e-03 ## Epoch 93/500 ## 90/90 - 0s - 1ms/step - loss: 0.0972 - sparse_categorical_accuracy: 0.9677 - val_loss: 0.1446 - val_sparse_categorical_accuracy: 0.9334 - learning_rate: 1.0000e-03 ## Epoch 94/500 ## 90/90 - 0s - 1ms/step - loss: 0.0965 - sparse_categorical_accuracy: 0.9632 - val_loss: 0.1265 - val_sparse_categorical_accuracy: 0.9584 - learning_rate: 1.0000e-03 ## Epoch 95/500 ## 90/90 - 0s - 2ms/step - loss: 0.0927 - sparse_categorical_accuracy: 0.9667 - val_loss: 0.1507 - val_sparse_categorical_accuracy: 0.9376 - learning_rate: 1.0000e-03 ## Epoch 96/500 ## 90/90 - 0s - 1ms/step - loss: 0.1050 - sparse_categorical_accuracy: 0.9618 - val_loss: 0.5005 - val_sparse_categorical_accuracy: 0.8086 - learning_rate: 1.0000e-03 ## Epoch 97/500 ## 90/90 - 0s - 1ms/step - loss: 0.1134 - sparse_categorical_accuracy: 0.9597 - val_loss: 0.1732 - val_sparse_categorical_accuracy: 0.9390 - learning_rate: 1.0000e-03 ## Epoch 98/500 ## 90/90 - 0s - 1ms/step - loss: 0.0946 - sparse_categorical_accuracy: 0.9698 - val_loss: 1.8677 - val_sparse_categorical_accuracy: 0.6949 - learning_rate: 1.0000e-03 ## Epoch 99/500 ## 90/90 - 0s - 1ms/step - loss: 0.1018 - sparse_categorical_accuracy: 0.9632 - val_loss: 1.1308 - val_sparse_categorical_accuracy: 0.6782 - learning_rate: 1.0000e-03 ## Epoch 100/500 ## 90/90 - 0s - 1ms/step - loss: 0.0990 - sparse_categorical_accuracy: 0.9674 - val_loss: 0.3846 - val_sparse_categorical_accuracy: 0.8558 - learning_rate: 1.0000e-03 ## Epoch 101/500 ## 90/90 - 0s - 1ms/step - loss: 0.0891 - sparse_categorical_accuracy: 0.9705 - val_loss: 0.1480 - val_sparse_categorical_accuracy: 0.9362 - learning_rate: 1.0000e-03 ## Epoch 102/500 ## 90/90 - 0s - 1ms/step - loss: 0.1003 - sparse_categorical_accuracy: 0.9677 - val_loss: 0.1713 - val_sparse_categorical_accuracy: 0.9390 - learning_rate: 1.0000e-03 ## Epoch 103/500 ## 90/90 - 0s - 1ms/step - loss: 0.0910 - sparse_categorical_accuracy: 0.9722 - val_loss: 0.1415 - val_sparse_categorical_accuracy: 0.9542 - learning_rate: 1.0000e-03 ## Epoch 104/500 ## 90/90 - 0s - 1ms/step - loss: 0.0998 - sparse_categorical_accuracy: 0.9622 - val_loss: 0.5677 - val_sparse_categorical_accuracy: 0.7753 - learning_rate: 1.0000e-03 ## Epoch 105/500 ## 90/90 - 0s - 1ms/step - loss: 0.1037 - sparse_categorical_accuracy: 0.9649 - val_loss: 0.9369 - val_sparse_categorical_accuracy: 0.6241 - learning_rate: 1.0000e-03 ## Epoch 106/500 ## 90/90 - 1s - 10ms/step - loss: 0.0850 - sparse_categorical_accuracy: 0.9708 - val_loss: 0.1388 - val_sparse_categorical_accuracy: 0.9487 - learning_rate: 5.0000e-04 ## Epoch 107/500 ## 90/90 - 0s - 1ms/step - loss: 0.0833 - sparse_categorical_accuracy: 0.9743 - val_loss: 0.1413 - val_sparse_categorical_accuracy: 0.9376 - learning_rate: 5.0000e-04 ## Epoch 108/500 ## 90/90 - 0s - 1ms/step - loss: 0.0775 - sparse_categorical_accuracy: 0.9767 - val_loss: 0.1223 - val_sparse_categorical_accuracy: 0.9445 - learning_rate: 5.0000e-04 ## Epoch 109/500 ## 90/90 - 0s - 1ms/step - loss: 0.0837 - sparse_categorical_accuracy: 0.9753 - val_loss: 0.1275 - val_sparse_categorical_accuracy: 0.9528 - learning_rate: 5.0000e-04 ## Epoch 110/500 ## 90/90 - 0s - 1ms/step - loss: 0.0859 - sparse_categorical_accuracy: 0.9705 - val_loss: 0.1209 - val_sparse_categorical_accuracy: 0.9487 - learning_rate: 5.0000e-04 ## Epoch 111/500 ## 90/90 - 0s - 2ms/step - loss: 0.0780 - sparse_categorical_accuracy: 0.9757 - val_loss: 0.1142 - val_sparse_categorical_accuracy: 0.9584 - learning_rate: 5.0000e-04 ## Epoch 112/500 ## 90/90 - 0s - 1ms/step - loss: 0.0801 - sparse_categorical_accuracy: 0.9736 - val_loss: 0.1316 - val_sparse_categorical_accuracy: 0.9459 - learning_rate: 5.0000e-04 ## Epoch 113/500 ## 90/90 - 0s - 1ms/step - loss: 0.0740 - sparse_categorical_accuracy: 0.9753 - val_loss: 0.3385 - val_sparse_categorical_accuracy: 0.8544 - learning_rate: 5.0000e-04 ## Epoch 114/500 ## 90/90 - 0s - 1ms/step - loss: 0.0907 - sparse_categorical_accuracy: 0.9722 - val_loss: 0.1222 - val_sparse_categorical_accuracy: 0.9612 - learning_rate: 5.0000e-04 ## Epoch 115/500 ## 90/90 - 0s - 2ms/step - loss: 0.0767 - sparse_categorical_accuracy: 0.9733 - val_loss: 0.1115 - val_sparse_categorical_accuracy: 0.9584 - learning_rate: 5.0000e-04 ## Epoch 116/500 ## 90/90 - 0s - 1ms/step - loss: 0.0786 - sparse_categorical_accuracy: 0.9750 - val_loss: 0.1667 - val_sparse_categorical_accuracy: 0.9431 - learning_rate: 5.0000e-04 ## Epoch 117/500 ## 90/90 - 0s - 1ms/step - loss: 0.0760 - sparse_categorical_accuracy: 0.9750 - val_loss: 0.1519 - val_sparse_categorical_accuracy: 0.9376 - learning_rate: 5.0000e-04 ## Epoch 118/500 ## 90/90 - 0s - 1ms/step - loss: 0.0797 - sparse_categorical_accuracy: 0.9722 - val_loss: 0.1555 - val_sparse_categorical_accuracy: 0.9459 - learning_rate: 5.0000e-04 ## Epoch 119/500 ## 90/90 - 0s - 1ms/step - loss: 0.0793 - sparse_categorical_accuracy: 0.9729 - val_loss: 0.1252 - val_sparse_categorical_accuracy: 0.9598 - learning_rate: 5.0000e-04 ## Epoch 120/500 ## 90/90 - 0s - 1ms/step - loss: 0.0803 - sparse_categorical_accuracy: 0.9747 - val_loss: 0.2298 - val_sparse_categorical_accuracy: 0.9015 - learning_rate: 5.0000e-04 ## Epoch 121/500 ## 90/90 - 0s - 1ms/step - loss: 0.0793 - sparse_categorical_accuracy: 0.9736 - val_loss: 0.1140 - val_sparse_categorical_accuracy: 0.9528 - learning_rate: 5.0000e-04 ## Epoch 122/500 ## 90/90 - 0s - 1ms/step - loss: 0.0795 - sparse_categorical_accuracy: 0.9740 - val_loss: 0.1154 - val_sparse_categorical_accuracy: 0.9542 - learning_rate: 5.0000e-04 ## Epoch 123/500 ## 90/90 - 0s - 1ms/step - loss: 0.0768 - sparse_categorical_accuracy: 0.9740 - val_loss: 0.3421 - val_sparse_categorical_accuracy: 0.8724 - learning_rate: 5.0000e-04 ## Epoch 124/500 ## 90/90 - 0s - 1ms/step - loss: 0.0848 - sparse_categorical_accuracy: 0.9694 - val_loss: 0.1126 - val_sparse_categorical_accuracy: 0.9598 - learning_rate: 5.0000e-04 ## Epoch 125/500 ## 90/90 - 0s - 1ms/step - loss: 0.0800 - sparse_categorical_accuracy: 0.9722 - val_loss: 0.1821 - val_sparse_categorical_accuracy: 0.9293 - learning_rate: 5.0000e-04 ## Epoch 126/500 ## 90/90 - 0s - 2ms/step - loss: 0.0787 - sparse_categorical_accuracy: 0.9753 - val_loss: 0.1105 - val_sparse_categorical_accuracy: 0.9570 - learning_rate: 5.0000e-04 ## Epoch 127/500 ## 90/90 - 0s - 2ms/step - loss: 0.0799 - sparse_categorical_accuracy: 0.9760 - val_loss: 0.1091 - val_sparse_categorical_accuracy: 0.9584 - learning_rate: 5.0000e-04 ## Epoch 128/500 ## 90/90 - 0s - 1ms/step - loss: 0.0811 - sparse_categorical_accuracy: 0.9736 - val_loss: 0.1287 - val_sparse_categorical_accuracy: 0.9515 - learning_rate: 5.0000e-04 ## Epoch 129/500 ## 90/90 - 0s - 1ms/step - loss: 0.0797 - sparse_categorical_accuracy: 0.9736 - val_loss: 0.1143 - val_sparse_categorical_accuracy: 0.9584 - learning_rate: 5.0000e-04 ## Epoch 130/500 ## 90/90 - 0s - 1ms/step - loss: 0.0850 - sparse_categorical_accuracy: 0.9750 - val_loss: 0.1581 - val_sparse_categorical_accuracy: 0.9376 - learning_rate: 5.0000e-04 ## Epoch 131/500 ## 90/90 - 0s - 1ms/step - loss: 0.0707 - sparse_categorical_accuracy: 0.9792 - val_loss: 0.1129 - val_sparse_categorical_accuracy: 0.9639 - learning_rate: 5.0000e-04 ## Epoch 132/500 ## 90/90 - 0s - 1ms/step - loss: 0.0725 - sparse_categorical_accuracy: 0.9740 - val_loss: 0.1425 - val_sparse_categorical_accuracy: 0.9501 - learning_rate: 5.0000e-04 ## Epoch 133/500 ## 90/90 - 0s - 2ms/step - loss: 0.0705 - sparse_categorical_accuracy: 0.9764 - val_loss: 0.1098 - val_sparse_categorical_accuracy: 0.9612 - learning_rate: 5.0000e-04 ## Epoch 134/500 ## 90/90 - 0s - 1ms/step - loss: 0.0767 - sparse_categorical_accuracy: 0.9736 - val_loss: 0.1884 - val_sparse_categorical_accuracy: 0.9376 - learning_rate: 5.0000e-04 ## Epoch 135/500 ## 90/90 - 0s - 1ms/step - loss: 0.0775 - sparse_categorical_accuracy: 0.9743 - val_loss: 0.2899 - val_sparse_categorical_accuracy: 0.8988 - learning_rate: 5.0000e-04 ## Epoch 136/500 ## 90/90 - 0s - 1ms/step - loss: 0.0807 - sparse_categorical_accuracy: 0.9701 - val_loss: 0.1152 - val_sparse_categorical_accuracy: 0.9584 - learning_rate: 5.0000e-04 ## Epoch 137/500 ## 90/90 - 0s - 1ms/step - loss: 0.0839 - sparse_categorical_accuracy: 0.9750 - val_loss: 0.3530 - val_sparse_categorical_accuracy: 0.8793 - learning_rate: 5.0000e-04 ## Epoch 138/500 ## 90/90 - 0s - 1ms/step - loss: 0.0770 - sparse_categorical_accuracy: 0.9743 - val_loss: 0.1267 - val_sparse_categorical_accuracy: 0.9473 - learning_rate: 5.0000e-04 ## Epoch 139/500 ## 90/90 - 0s - 2ms/step - loss: 0.0845 - sparse_categorical_accuracy: 0.9719 - val_loss: 0.1043 - val_sparse_categorical_accuracy: 0.9626 - learning_rate: 5.0000e-04 ## Epoch 140/500 ## 90/90 - 0s - 1ms/step - loss: 0.0701 - sparse_categorical_accuracy: 0.9799 - val_loss: 0.1351 - val_sparse_categorical_accuracy: 0.9556 - learning_rate: 5.0000e-04 ## Epoch 141/500 ## 90/90 - 0s - 1ms/step - loss: 0.0712 - sparse_categorical_accuracy: 0.9764 - val_loss: 0.1121 - val_sparse_categorical_accuracy: 0.9612 - learning_rate: 5.0000e-04 ## Epoch 142/500 ## 90/90 - 0s - 1ms/step - loss: 0.0692 - sparse_categorical_accuracy: 0.9778 - val_loss: 0.1139 - val_sparse_categorical_accuracy: 0.9639 - learning_rate: 5.0000e-04 ## Epoch 143/500 ## 90/90 - 0s - 1ms/step - loss: 0.0781 - sparse_categorical_accuracy: 0.9736 - val_loss: 0.1202 - val_sparse_categorical_accuracy: 0.9515 - learning_rate: 5.0000e-04 ## Epoch 144/500 ## 90/90 - 0s - 1ms/step - loss: 0.0699 - sparse_categorical_accuracy: 0.9781 - val_loss: 0.1131 - val_sparse_categorical_accuracy: 0.9501 - learning_rate: 5.0000e-04 ## Epoch 145/500 ## 90/90 - 0s - 1ms/step - loss: 0.0712 - sparse_categorical_accuracy: 0.9771 - val_loss: 0.1097 - val_sparse_categorical_accuracy: 0.9542 - learning_rate: 5.0000e-04 ## Epoch 146/500 ## 90/90 - 0s - 1ms/step - loss: 0.0730 - sparse_categorical_accuracy: 0.9736 - val_loss: 0.1421 - val_sparse_categorical_accuracy: 0.9445 - learning_rate: 5.0000e-04 ## Epoch 147/500 ## 90/90 - 0s - 1ms/step - loss: 0.0757 - sparse_categorical_accuracy: 0.9740 - val_loss: 0.3533 - val_sparse_categorical_accuracy: 0.8752 - learning_rate: 5.0000e-04 ## Epoch 148/500 ## 90/90 - 0s - 1ms/step - loss: 0.0818 - sparse_categorical_accuracy: 0.9701 - val_loss: 0.7787 - val_sparse_categorical_accuracy: 0.7836 - learning_rate: 5.0000e-04 ## Epoch 149/500 ## 90/90 - 0s - 1ms/step - loss: 0.0717 - sparse_categorical_accuracy: 0.9771 - val_loss: 0.1214 - val_sparse_categorical_accuracy: 0.9515 - learning_rate: 5.0000e-04 ## Epoch 150/500 ## 90/90 - 0s - 1ms/step - loss: 0.0736 - sparse_categorical_accuracy: 0.9757 - val_loss: 0.1445 - val_sparse_categorical_accuracy: 0.9362 - learning_rate: 5.0000e-04 ## Epoch 151/500 ## 90/90 - 0s - 1ms/step - loss: 0.0725 - sparse_categorical_accuracy: 0.9729 - val_loss: 0.1175 - val_sparse_categorical_accuracy: 0.9528 - learning_rate: 5.0000e-04 ## Epoch 152/500 ## 90/90 - 0s - 1ms/step - loss: 0.0726 - sparse_categorical_accuracy: 0.9750 - val_loss: 0.1329 - val_sparse_categorical_accuracy: 0.9515 - learning_rate: 5.0000e-04 ## Epoch 153/500 ## 90/90 - 0s - 1ms/step - loss: 0.0727 - sparse_categorical_accuracy: 0.9757 - val_loss: 0.1806 - val_sparse_categorical_accuracy: 0.9293 - learning_rate: 5.0000e-04 ## Epoch 154/500 ## 90/90 - 0s - 1ms/step - loss: 0.0684 - sparse_categorical_accuracy: 0.9778 - val_loss: 0.3892 - val_sparse_categorical_accuracy: 0.8682 - learning_rate: 5.0000e-04 ## Epoch 155/500 ## 90/90 - 0s - 1ms/step - loss: 0.0734 - sparse_categorical_accuracy: 0.9753 - val_loss: 0.1122 - val_sparse_categorical_accuracy: 0.9542 - learning_rate: 5.0000e-04 ## Epoch 156/500 ## 90/90 - 0s - 1ms/step - loss: 0.0758 - sparse_categorical_accuracy: 0.9774 - val_loss: 0.1077 - val_sparse_categorical_accuracy: 0.9653 - learning_rate: 5.0000e-04 ## Epoch 157/500 ## 90/90 - 0s - 1ms/step - loss: 0.0734 - sparse_categorical_accuracy: 0.9747 - val_loss: 0.1594 - val_sparse_categorical_accuracy: 0.9376 - learning_rate: 5.0000e-04 ## Epoch 158/500 ## 90/90 - 0s - 1ms/step - loss: 0.0753 - sparse_categorical_accuracy: 0.9750 - val_loss: 0.1992 - val_sparse_categorical_accuracy: 0.9168 - learning_rate: 5.0000e-04 ## Epoch 159/500 ## 90/90 - 0s - 1ms/step - loss: 0.0740 - sparse_categorical_accuracy: 0.9740 - val_loss: 0.1483 - val_sparse_categorical_accuracy: 0.9362 - learning_rate: 5.0000e-04 ## Epoch 160/500 ## 90/90 - 0s - 2ms/step - loss: 0.0677 - sparse_categorical_accuracy: 0.9778 - val_loss: 0.1041 - val_sparse_categorical_accuracy: 0.9653 - learning_rate: 2.5000e-04 ## Epoch 161/500 ## 90/90 - 0s - 1ms/step - loss: 0.0686 - sparse_categorical_accuracy: 0.9778 - val_loss: 0.1327 - val_sparse_categorical_accuracy: 0.9515 - learning_rate: 2.5000e-04 ## Epoch 162/500 ## 90/90 - 0s - 1ms/step - loss: 0.0634 - sparse_categorical_accuracy: 0.9792 - val_loss: 0.1059 - val_sparse_categorical_accuracy: 0.9667 - learning_rate: 2.5000e-04 ## Epoch 163/500 ## 90/90 - 0s - 1ms/step - loss: 0.0595 - sparse_categorical_accuracy: 0.9792 - val_loss: 0.1236 - val_sparse_categorical_accuracy: 0.9612 - learning_rate: 2.5000e-04 ## Epoch 164/500 ## 90/90 - 0s - 1ms/step - loss: 0.0637 - sparse_categorical_accuracy: 0.9806 - val_loss: 0.1185 - val_sparse_categorical_accuracy: 0.9626 - learning_rate: 2.5000e-04 ## Epoch 165/500 ## 90/90 - 0s - 1ms/step - loss: 0.0607 - sparse_categorical_accuracy: 0.9816 - val_loss: 0.1480 - val_sparse_categorical_accuracy: 0.9473 - learning_rate: 2.5000e-04 ## Epoch 166/500 ## 90/90 - 0s - 1ms/step - loss: 0.0594 - sparse_categorical_accuracy: 0.9799 - val_loss: 0.1044 - val_sparse_categorical_accuracy: 0.9653 - learning_rate: 2.5000e-04 ## Epoch 167/500 ## 90/90 - 0s - 1ms/step - loss: 0.0629 - sparse_categorical_accuracy: 0.9788 - val_loss: 0.1343 - val_sparse_categorical_accuracy: 0.9445 - learning_rate: 2.5000e-04 ## Epoch 168/500 ## 90/90 - 0s - 1ms/step - loss: 0.0740 - sparse_categorical_accuracy: 0.9750 - val_loss: 0.1081 - val_sparse_categorical_accuracy: 0.9667 - learning_rate: 2.5000e-04 ## Epoch 169/500 ## 90/90 - 0s - 1ms/step - loss: 0.0649 - sparse_categorical_accuracy: 0.9799 - val_loss: 0.1136 - val_sparse_categorical_accuracy: 0.9681 - learning_rate: 2.5000e-04 ## Epoch 170/500 ## 90/90 - 0s - 2ms/step - loss: 0.0649 - sparse_categorical_accuracy: 0.9823 - val_loss: 0.1037 - val_sparse_categorical_accuracy: 0.9695 - learning_rate: 2.5000e-04 ## Epoch 171/500 ## 90/90 - 0s - 1ms/step - loss: 0.0609 - sparse_categorical_accuracy: 0.9799 - val_loss: 0.1239 - val_sparse_categorical_accuracy: 0.9653 - learning_rate: 2.5000e-04 ## Epoch 172/500 ## 90/90 - 0s - 1ms/step - loss: 0.0636 - sparse_categorical_accuracy: 0.9795 - val_loss: 0.1037 - val_sparse_categorical_accuracy: 0.9528 - learning_rate: 2.5000e-04 ## Epoch 173/500 ## 90/90 - 0s - 1ms/step - loss: 0.0608 - sparse_categorical_accuracy: 0.9816 - val_loss: 0.1296 - val_sparse_categorical_accuracy: 0.9626 - learning_rate: 2.5000e-04 ## Epoch 174/500 ## 90/90 - 0s - 2ms/step - loss: 0.0624 - sparse_categorical_accuracy: 0.9792 - val_loss: 0.0981 - val_sparse_categorical_accuracy: 0.9639 - learning_rate: 2.5000e-04 ## Epoch 175/500 ## 90/90 - 0s - 1ms/step - loss: 0.0666 - sparse_categorical_accuracy: 0.9785 - val_loss: 0.1135 - val_sparse_categorical_accuracy: 0.9487 - learning_rate: 2.5000e-04 ## Epoch 176/500 ## 90/90 - 0s - 1ms/step - loss: 0.0625 - sparse_categorical_accuracy: 0.9771 - val_loss: 0.1024 - val_sparse_categorical_accuracy: 0.9584 - learning_rate: 2.5000e-04 ## Epoch 177/500 ## 90/90 - 0s - 1ms/step - loss: 0.0616 - sparse_categorical_accuracy: 0.9806 - val_loss: 0.1036 - val_sparse_categorical_accuracy: 0.9570 - learning_rate: 2.5000e-04 ## Epoch 178/500 ## 90/90 - 0s - 1ms/step - loss: 0.0590 - sparse_categorical_accuracy: 0.9823 - val_loss: 0.2086 - val_sparse_categorical_accuracy: 0.9154 - learning_rate: 2.5000e-04 ## Epoch 179/500 ## 90/90 - 0s - 1ms/step - loss: 0.0602 - sparse_categorical_accuracy: 0.9788 - val_loss: 0.1181 - val_sparse_categorical_accuracy: 0.9653 - learning_rate: 2.5000e-04 ## Epoch 180/500 ## 90/90 - 0s - 1ms/step - loss: 0.0615 - sparse_categorical_accuracy: 0.9816 - val_loss: 0.1587 - val_sparse_categorical_accuracy: 0.9459 - learning_rate: 2.5000e-04 ## Epoch 181/500 ## 90/90 - 0s - 1ms/step - loss: 0.0647 - sparse_categorical_accuracy: 0.9813 - val_loss: 0.1836 - val_sparse_categorical_accuracy: 0.9362 - learning_rate: 2.5000e-04 ## Epoch 182/500 ## 90/90 - 0s - 1ms/step - loss: 0.0614 - sparse_categorical_accuracy: 0.9813 - val_loss: 0.2156 - val_sparse_categorical_accuracy: 0.9223 - learning_rate: 2.5000e-04 ## Epoch 183/500 ## 90/90 - 0s - 1ms/step - loss: 0.0637 - sparse_categorical_accuracy: 0.9778 - val_loss: 0.1028 - val_sparse_categorical_accuracy: 0.9584 - learning_rate: 2.5000e-04 ## Epoch 184/500 ## 90/90 - 0s - 1ms/step - loss: 0.0572 - sparse_categorical_accuracy: 0.9823 - val_loss: 0.1347 - val_sparse_categorical_accuracy: 0.9431 - learning_rate: 2.5000e-04 ## Epoch 185/500 ## 90/90 - 0s - 1ms/step - loss: 0.0593 - sparse_categorical_accuracy: 0.9826 - val_loss: 0.1402 - val_sparse_categorical_accuracy: 0.9459 - learning_rate: 2.5000e-04 ## Epoch 186/500 ## 90/90 - 0s - 1ms/step - loss: 0.0659 - sparse_categorical_accuracy: 0.9774 - val_loss: 0.1873 - val_sparse_categorical_accuracy: 0.9293 - learning_rate: 2.5000e-04 ## Epoch 187/500 ## 90/90 - 0s - 1ms/step - loss: 0.0606 - sparse_categorical_accuracy: 0.9799 - val_loss: 0.1210 - val_sparse_categorical_accuracy: 0.9570 - learning_rate: 2.5000e-04 ## Epoch 188/500 ## 90/90 - 0s - 1ms/step - loss: 0.0610 - sparse_categorical_accuracy: 0.9813 - val_loss: 0.1250 - val_sparse_categorical_accuracy: 0.9473 - learning_rate: 2.5000e-04 ## Epoch 189/500 ## 90/90 - 0s - 1ms/step - loss: 0.0570 - sparse_categorical_accuracy: 0.9830 - val_loss: 0.1227 - val_sparse_categorical_accuracy: 0.9667 - learning_rate: 2.5000e-04 ## Epoch 190/500 ## 90/90 - 0s - 1ms/step - loss: 0.0678 - sparse_categorical_accuracy: 0.9771 - val_loss: 0.1367 - val_sparse_categorical_accuracy: 0.9584 - learning_rate: 2.5000e-04 ## Epoch 191/500 ## 90/90 - 0s - 1ms/step - loss: 0.0580 - sparse_categorical_accuracy: 0.9802 - val_loss: 0.1005 - val_sparse_categorical_accuracy: 0.9653 - learning_rate: 2.5000e-04 ## Epoch 192/500 ## 90/90 - 0s - 1ms/step - loss: 0.0651 - sparse_categorical_accuracy: 0.9778 - val_loss: 0.1170 - val_sparse_categorical_accuracy: 0.9639 - learning_rate: 2.5000e-04 ## Epoch 193/500 ## 90/90 - 0s - 1ms/step - loss: 0.0593 - sparse_categorical_accuracy: 0.9823 - val_loss: 0.1235 - val_sparse_categorical_accuracy: 0.9473 - learning_rate: 2.5000e-04 ## Epoch 194/500 ## 90/90 - 0s - 1ms/step - loss: 0.0588 - sparse_categorical_accuracy: 0.9792 - val_loss: 0.1089 - val_sparse_categorical_accuracy: 0.9570 - learning_rate: 2.5000e-04 ## Epoch 195/500 ## 90/90 - 0s - 2ms/step - loss: 0.0564 - sparse_categorical_accuracy: 0.9833 - val_loss: 0.0979 - val_sparse_categorical_accuracy: 0.9681 - learning_rate: 1.2500e-04 ## Epoch 196/500 ## 90/90 - 0s - 1ms/step - loss: 0.0584 - sparse_categorical_accuracy: 0.9788 - val_loss: 0.1169 - val_sparse_categorical_accuracy: 0.9584 - learning_rate: 1.2500e-04 ## Epoch 197/500 ## 90/90 - 0s - 1ms/step - loss: 0.0553 - sparse_categorical_accuracy: 0.9802 - val_loss: 0.1380 - val_sparse_categorical_accuracy: 0.9556 - learning_rate: 1.2500e-04 ## Epoch 198/500 ## 90/90 - 0s - 1ms/step - loss: 0.0541 - sparse_categorical_accuracy: 0.9868 - val_loss: 0.1261 - val_sparse_categorical_accuracy: 0.9556 - learning_rate: 1.2500e-04 ## Epoch 199/500 ## 90/90 - 0s - 1ms/step - loss: 0.0575 - sparse_categorical_accuracy: 0.9806 - val_loss: 0.1010 - val_sparse_categorical_accuracy: 0.9681 - learning_rate: 1.2500e-04 ## Epoch 200/500 ## 90/90 - 0s - 1ms/step - loss: 0.0568 - sparse_categorical_accuracy: 0.9816 - val_loss: 0.0996 - val_sparse_categorical_accuracy: 0.9612 - learning_rate: 1.2500e-04 ## Epoch 201/500 ## 90/90 - 0s - 1ms/step - loss: 0.0547 - sparse_categorical_accuracy: 0.9819 - val_loss: 0.1096 - val_sparse_categorical_accuracy: 0.9653 - learning_rate: 1.2500e-04 ## Epoch 202/500 ## 90/90 - 0s - 1ms/step - loss: 0.0547 - sparse_categorical_accuracy: 0.9847 - val_loss: 0.1019 - val_sparse_categorical_accuracy: 0.9556 - learning_rate: 1.2500e-04 ## Epoch 203/500 ## 90/90 - 0s - 1ms/step - loss: 0.0562 - sparse_categorical_accuracy: 0.9816 - val_loss: 0.0985 - val_sparse_categorical_accuracy: 0.9639 - learning_rate: 1.2500e-04 ## Epoch 204/500 ## 90/90 - 0s - 2ms/step - loss: 0.0519 - sparse_categorical_accuracy: 0.9830 - val_loss: 0.0978 - val_sparse_categorical_accuracy: 0.9681 - learning_rate: 1.2500e-04 ## Epoch 205/500 ## 90/90 - 0s - 1ms/step - loss: 0.0558 - sparse_categorical_accuracy: 0.9802 - val_loss: 0.1035 - val_sparse_categorical_accuracy: 0.9667 - learning_rate: 1.2500e-04 ## Epoch 206/500 ## 90/90 - 0s - 1ms/step - loss: 0.0551 - sparse_categorical_accuracy: 0.9816 - val_loss: 0.1006 - val_sparse_categorical_accuracy: 0.9709 - learning_rate: 1.2500e-04 ## Epoch 207/500 ## 90/90 - 0s - 1ms/step - loss: 0.0593 - sparse_categorical_accuracy: 0.9816 - val_loss: 0.1023 - val_sparse_categorical_accuracy: 0.9556 - learning_rate: 1.2500e-04 ## Epoch 208/500 ## 90/90 - 0s - 2ms/step - loss: 0.0556 - sparse_categorical_accuracy: 0.9816 - val_loss: 0.1314 - val_sparse_categorical_accuracy: 0.9626 - learning_rate: 1.2500e-04 ## Epoch 209/500 ## 90/90 - 0s - 1ms/step - loss: 0.0539 - sparse_categorical_accuracy: 0.9809 - val_loss: 0.1529 - val_sparse_categorical_accuracy: 0.9487 - learning_rate: 1.2500e-04 ## Epoch 210/500 ## 90/90 - 0s - 1ms/step - loss: 0.0566 - sparse_categorical_accuracy: 0.9816 - val_loss: 0.0993 - val_sparse_categorical_accuracy: 0.9653 - learning_rate: 1.2500e-04 ## Epoch 211/500 ## 90/90 - 0s - 1ms/step - loss: 0.0552 - sparse_categorical_accuracy: 0.9837 - val_loss: 0.0984 - val_sparse_categorical_accuracy: 0.9639 - learning_rate: 1.2500e-04 ## Epoch 212/500 ## 90/90 - 0s - 1ms/step - loss: 0.0548 - sparse_categorical_accuracy: 0.9826 - val_loss: 0.0999 - val_sparse_categorical_accuracy: 0.9584 - learning_rate: 1.2500e-04 ## Epoch 213/500 ## 90/90 - 0s - 2ms/step - loss: 0.0573 - sparse_categorical_accuracy: 0.9840 - val_loss: 0.0968 - val_sparse_categorical_accuracy: 0.9667 - learning_rate: 1.2500e-04 ## Epoch 214/500 ## 90/90 - 0s - 1ms/step - loss: 0.0568 - sparse_categorical_accuracy: 0.9830 - val_loss: 0.1000 - val_sparse_categorical_accuracy: 0.9639 - learning_rate: 1.2500e-04 ## Epoch 215/500 ## 90/90 - 0s - 1ms/step - loss: 0.0534 - sparse_categorical_accuracy: 0.9816 - val_loss: 0.1031 - val_sparse_categorical_accuracy: 0.9667 - learning_rate: 1.2500e-04 ## Epoch 216/500 ## 90/90 - 0s - 1ms/step - loss: 0.0531 - sparse_categorical_accuracy: 0.9847 - val_loss: 0.1042 - val_sparse_categorical_accuracy: 0.9681 - learning_rate: 1.2500e-04 ## Epoch 217/500 ## 90/90 - 0s - 1ms/step - loss: 0.0532 - sparse_categorical_accuracy: 0.9826 - val_loss: 0.1096 - val_sparse_categorical_accuracy: 0.9695 - learning_rate: 1.2500e-04 ## Epoch 218/500 ## 90/90 - 0s - 1ms/step - loss: 0.0577 - sparse_categorical_accuracy: 0.9809 - val_loss: 0.1009 - val_sparse_categorical_accuracy: 0.9612 - learning_rate: 1.2500e-04 ## Epoch 219/500 ## 90/90 - 0s - 1ms/step - loss: 0.0539 - sparse_categorical_accuracy: 0.9823 - val_loss: 0.1072 - val_sparse_categorical_accuracy: 0.9695 - learning_rate: 1.2500e-04 ## Epoch 220/500 ## 90/90 - 0s - 1ms/step - loss: 0.0529 - sparse_categorical_accuracy: 0.9851 - val_loss: 0.1005 - val_sparse_categorical_accuracy: 0.9695 - learning_rate: 1.2500e-04 ## Epoch 221/500 ## 90/90 - 0s - 1ms/step - loss: 0.0554 - sparse_categorical_accuracy: 0.9799 - val_loss: 0.1036 - val_sparse_categorical_accuracy: 0.9695 - learning_rate: 1.2500e-04 ## Epoch 222/500 ## 90/90 - 0s - 1ms/step - loss: 0.0562 - sparse_categorical_accuracy: 0.9809 - val_loss: 0.1037 - val_sparse_categorical_accuracy: 0.9736 - learning_rate: 1.2500e-04 ## Epoch 223/500 ## 90/90 - 0s - 1ms/step - loss: 0.0544 - sparse_categorical_accuracy: 0.9823 - val_loss: 0.1005 - val_sparse_categorical_accuracy: 0.9736 - learning_rate: 1.2500e-04 ## Epoch 224/500 ## 90/90 - 0s - 1ms/step - loss: 0.0509 - sparse_categorical_accuracy: 0.9847 - val_loss: 0.0969 - val_sparse_categorical_accuracy: 0.9626 - learning_rate: 1.2500e-04 ## Epoch 225/500 ## 90/90 - 0s - 1ms/step - loss: 0.0569 - sparse_categorical_accuracy: 0.9823 - val_loss: 0.1261 - val_sparse_categorical_accuracy: 0.9612 - learning_rate: 1.2500e-04 ## Epoch 226/500 ## 90/90 - 0s - 1ms/step - loss: 0.0519 - sparse_categorical_accuracy: 0.9826 - val_loss: 0.1019 - val_sparse_categorical_accuracy: 0.9570 - learning_rate: 1.2500e-04 ## Epoch 227/500 ## 90/90 - 0s - 1ms/step - loss: 0.0563 - sparse_categorical_accuracy: 0.9819 - val_loss: 0.1037 - val_sparse_categorical_accuracy: 0.9584 - learning_rate: 1.2500e-04 ## Epoch 228/500 ## 90/90 - 0s - 1ms/step - loss: 0.0518 - sparse_categorical_accuracy: 0.9844 - val_loss: 0.1127 - val_sparse_categorical_accuracy: 0.9515 - learning_rate: 1.2500e-04 ## Epoch 229/500 ## 90/90 - 0s - 1ms/step - loss: 0.0521 - sparse_categorical_accuracy: 0.9865 - val_loss: 0.0973 - val_sparse_categorical_accuracy: 0.9639 - learning_rate: 1.2500e-04 ## Epoch 230/500 ## 90/90 - 0s - 1ms/step - loss: 0.0520 - sparse_categorical_accuracy: 0.9847 - val_loss: 0.0999 - val_sparse_categorical_accuracy: 0.9626 - learning_rate: 1.2500e-04 ## Epoch 231/500 ## 90/90 - 0s - 2ms/step - loss: 0.0569 - sparse_categorical_accuracy: 0.9813 - val_loss: 0.0958 - val_sparse_categorical_accuracy: 0.9626 - learning_rate: 1.2500e-04 ## Epoch 232/500 ## 90/90 - 0s - 1ms/step - loss: 0.0530 - sparse_categorical_accuracy: 0.9851 - val_loss: 0.1024 - val_sparse_categorical_accuracy: 0.9681 - learning_rate: 1.2500e-04 ## Epoch 233/500 ## 90/90 - 0s - 1ms/step - loss: 0.0523 - sparse_categorical_accuracy: 0.9844 - val_loss: 0.1019 - val_sparse_categorical_accuracy: 0.9709 - learning_rate: 1.2500e-04 ## Epoch 234/500 ## 90/90 - 0s - 1ms/step - loss: 0.0539 - sparse_categorical_accuracy: 0.9837 - val_loss: 0.1595 - val_sparse_categorical_accuracy: 0.9459 - learning_rate: 1.2500e-04 ## Epoch 235/500 ## 90/90 - 0s - 1ms/step - loss: 0.0543 - sparse_categorical_accuracy: 0.9802 - val_loss: 0.1028 - val_sparse_categorical_accuracy: 0.9653 - learning_rate: 1.2500e-04 ## Epoch 236/500 ## 90/90 - 0s - 1ms/step - loss: 0.0540 - sparse_categorical_accuracy: 0.9826 - val_loss: 0.1068 - val_sparse_categorical_accuracy: 0.9653 - learning_rate: 1.2500e-04 ## Epoch 237/500 ## 90/90 - 0s - 1ms/step - loss: 0.0543 - sparse_categorical_accuracy: 0.9809 - val_loss: 0.1528 - val_sparse_categorical_accuracy: 0.9459 - learning_rate: 1.2500e-04 ## Epoch 238/500 ## 90/90 - 0s - 1ms/step - loss: 0.0527 - sparse_categorical_accuracy: 0.9861 - val_loss: 0.0977 - val_sparse_categorical_accuracy: 0.9695 - learning_rate: 1.2500e-04 ## Epoch 239/500 ## 90/90 - 0s - 1ms/step - loss: 0.0537 - sparse_categorical_accuracy: 0.9813 - val_loss: 0.0980 - val_sparse_categorical_accuracy: 0.9598 - learning_rate: 1.2500e-04 ## Epoch 240/500 ## 90/90 - 0s - 1ms/step - loss: 0.0512 - sparse_categorical_accuracy: 0.9854 - val_loss: 0.1054 - val_sparse_categorical_accuracy: 0.9709 - learning_rate: 1.2500e-04 ## Epoch 241/500 ## 90/90 - 0s - 1ms/step - loss: 0.0541 - sparse_categorical_accuracy: 0.9837 - val_loss: 0.1325 - val_sparse_categorical_accuracy: 0.9487 - learning_rate: 1.2500e-04 ## Epoch 242/500 ## 90/90 - 0s - 1ms/step - loss: 0.0535 - sparse_categorical_accuracy: 0.9826 - val_loss: 0.1028 - val_sparse_categorical_accuracy: 0.9681 - learning_rate: 1.2500e-04 ## Epoch 243/500 ## 90/90 - 0s - 1ms/step - loss: 0.0555 - sparse_categorical_accuracy: 0.9823 - val_loss: 0.0991 - val_sparse_categorical_accuracy: 0.9626 - learning_rate: 1.2500e-04 ## Epoch 244/500 ## 90/90 - 0s - 1ms/step - loss: 0.0572 - sparse_categorical_accuracy: 0.9795 - val_loss: 0.0986 - val_sparse_categorical_accuracy: 0.9639 - learning_rate: 1.2500e-04 ## Epoch 245/500 ## 90/90 - 0s - 1ms/step - loss: 0.0520 - sparse_categorical_accuracy: 0.9833 - val_loss: 0.0980 - val_sparse_categorical_accuracy: 0.9626 - learning_rate: 1.2500e-04 ## Epoch 246/500 ## 90/90 - 0s - 1ms/step - loss: 0.0554 - sparse_categorical_accuracy: 0.9809 - val_loss: 0.0978 - val_sparse_categorical_accuracy: 0.9639 - learning_rate: 1.2500e-04 ## Epoch 247/500 ## 90/90 - 0s - 1ms/step - loss: 0.0498 - sparse_categorical_accuracy: 0.9837 - val_loss: 0.1015 - val_sparse_categorical_accuracy: 0.9681 - learning_rate: 1.2500e-04 ## Epoch 248/500 ## 90/90 - 0s - 1ms/step - loss: 0.0493 - sparse_categorical_accuracy: 0.9840 - val_loss: 0.0994 - val_sparse_categorical_accuracy: 0.9612 - learning_rate: 1.2500e-04 ## Epoch 249/500 ## 90/90 - 0s - 1ms/step - loss: 0.0543 - sparse_categorical_accuracy: 0.9847 - val_loss: 0.1054 - val_sparse_categorical_accuracy: 0.9695 - learning_rate: 1.2500e-04 ## Epoch 250/500 ## 90/90 - 0s - 1ms/step - loss: 0.0550 - sparse_categorical_accuracy: 0.9802 - val_loss: 0.0989 - val_sparse_categorical_accuracy: 0.9626 - learning_rate: 1.2500e-04 ## Epoch 251/500 ## 90/90 - 0s - 1ms/step - loss: 0.0539 - sparse_categorical_accuracy: 0.9819 - val_loss: 0.1018 - val_sparse_categorical_accuracy: 0.9667 - learning_rate: 1.2500e-04 ## Epoch 252/500 ## 90/90 - 0s - 1ms/step - loss: 0.0542 - sparse_categorical_accuracy: 0.9830 - val_loss: 0.0975 - val_sparse_categorical_accuracy: 0.9667 - learning_rate: 1.0000e-04 ## Epoch 253/500 ## 90/90 - 0s - 1ms/step - loss: 0.0538 - sparse_categorical_accuracy: 0.9830 - val_loss: 0.1065 - val_sparse_categorical_accuracy: 0.9723 - learning_rate: 1.0000e-04 ## Epoch 254/500 ## 90/90 - 0s - 1ms/step - loss: 0.0542 - sparse_categorical_accuracy: 0.9823 - val_loss: 0.1145 - val_sparse_categorical_accuracy: 0.9528 - learning_rate: 1.0000e-04 ## Epoch 255/500 ## 90/90 - 0s - 1ms/step - loss: 0.0506 - sparse_categorical_accuracy: 0.9819 - val_loss: 0.0976 - val_sparse_categorical_accuracy: 0.9681 - learning_rate: 1.0000e-04 ## Epoch 256/500 ## 90/90 - 0s - 1ms/step - loss: 0.0521 - sparse_categorical_accuracy: 0.9847 - val_loss: 0.1117 - val_sparse_categorical_accuracy: 0.9626 - learning_rate: 1.0000e-04 ## Epoch 257/500 ## 90/90 - 0s - 1ms/step - loss: 0.0504 - sparse_categorical_accuracy: 0.9854 - val_loss: 0.1201 - val_sparse_categorical_accuracy: 0.9626 - learning_rate: 1.0000e-04 ## Epoch 258/500 ## 90/90 - 0s - 1ms/step - loss: 0.0498 - sparse_categorical_accuracy: 0.9851 - val_loss: 0.1039 - val_sparse_categorical_accuracy: 0.9584 - learning_rate: 1.0000e-04 ## Epoch 259/500 ## 90/90 - 0s - 1ms/step - loss: 0.0479 - sparse_categorical_accuracy: 0.9868 - val_loss: 0.1055 - val_sparse_categorical_accuracy: 0.9695 - learning_rate: 1.0000e-04 ## Epoch 260/500 ## 90/90 - 0s - 1ms/step - loss: 0.0499 - sparse_categorical_accuracy: 0.9840 - val_loss: 0.1046 - val_sparse_categorical_accuracy: 0.9556 - learning_rate: 1.0000e-04 ## Epoch 261/500 ## 90/90 - 0s - 1ms/step - loss: 0.0529 - sparse_categorical_accuracy: 0.9823 - val_loss: 0.1022 - val_sparse_categorical_accuracy: 0.9667 - learning_rate: 1.0000e-04 ## Epoch 262/500 ## 90/90 - 0s - 1ms/step - loss: 0.0492 - sparse_categorical_accuracy: 0.9840 - val_loss: 0.1063 - val_sparse_categorical_accuracy: 0.9681 - learning_rate: 1.0000e-04 ## Epoch 263/500 ## 90/90 - 0s - 1ms/step - loss: 0.0500 - sparse_categorical_accuracy: 0.9833 - val_loss: 0.1006 - val_sparse_categorical_accuracy: 0.9626 - learning_rate: 1.0000e-04 ## Epoch 264/500 ## 90/90 - 0s - 1ms/step - loss: 0.0533 - sparse_categorical_accuracy: 0.9823 - val_loss: 0.1088 - val_sparse_categorical_accuracy: 0.9709 - learning_rate: 1.0000e-04 ## Epoch 265/500 ## 90/90 - 0s - 1ms/step - loss: 0.0549 - sparse_categorical_accuracy: 0.9802 - val_loss: 0.1010 - val_sparse_categorical_accuracy: 0.9709 - learning_rate: 1.0000e-04 ## Epoch 266/500 ## 90/90 - 0s - 1ms/step - loss: 0.0512 - sparse_categorical_accuracy: 0.9830 - val_loss: 0.1016 - val_sparse_categorical_accuracy: 0.9570 - learning_rate: 1.0000e-04 ## Epoch 267/500 ## 90/90 - 0s - 1ms/step - loss: 0.0496 - sparse_categorical_accuracy: 0.9847 - val_loss: 0.1012 - val_sparse_categorical_accuracy: 0.9570 - learning_rate: 1.0000e-04 ## Epoch 268/500 ## 90/90 - 0s - 1ms/step - loss: 0.0477 - sparse_categorical_accuracy: 0.9872 - val_loss: 0.1016 - val_sparse_categorical_accuracy: 0.9570 - learning_rate: 1.0000e-04 ## Epoch 269/500 ## 90/90 - 0s - 1ms/step - loss: 0.0523 - sparse_categorical_accuracy: 0.9861 - val_loss: 0.1012 - val_sparse_categorical_accuracy: 0.9723 - learning_rate: 1.0000e-04 ## Epoch 270/500 ## 90/90 - 0s - 1ms/step - loss: 0.0474 - sparse_categorical_accuracy: 0.9889 - val_loss: 0.0975 - val_sparse_categorical_accuracy: 0.9653 - learning_rate: 1.0000e-04 ## Epoch 271/500 ## 90/90 - 0s - 1ms/step - loss: 0.0489 - sparse_categorical_accuracy: 0.9840 - val_loss: 0.1055 - val_sparse_categorical_accuracy: 0.9695 - learning_rate: 1.0000e-04 ## Epoch 272/500 ## 90/90 - 0s - 1ms/step - loss: 0.0503 - sparse_categorical_accuracy: 0.9826 - val_loss: 0.1050 - val_sparse_categorical_accuracy: 0.9709 - learning_rate: 1.0000e-04 ## Epoch 273/500 ## 90/90 - 0s - 1ms/step - loss: 0.0514 - sparse_categorical_accuracy: 0.9844 - val_loss: 0.0980 - val_sparse_categorical_accuracy: 0.9695 - learning_rate: 1.0000e-04 ## Epoch 274/500 ## 90/90 - 0s - 1ms/step - loss: 0.0479 - sparse_categorical_accuracy: 0.9854 - val_loss: 0.1049 - val_sparse_categorical_accuracy: 0.9639 - learning_rate: 1.0000e-04 ## Epoch 275/500 ## 90/90 - 0s - 1ms/step - loss: 0.0467 - sparse_categorical_accuracy: 0.9840 - val_loss: 0.1473 - val_sparse_categorical_accuracy: 0.9556 - learning_rate: 1.0000e-04 ## Epoch 276/500 ## 90/90 - 0s - 1ms/step - loss: 0.0504 - sparse_categorical_accuracy: 0.9819 - val_loss: 0.1065 - val_sparse_categorical_accuracy: 0.9695 - learning_rate: 1.0000e-04 ## Epoch 277/500 ## 90/90 - 0s - 1ms/step - loss: 0.0505 - sparse_categorical_accuracy: 0.9847 - val_loss: 0.1187 - val_sparse_categorical_accuracy: 0.9612 - learning_rate: 1.0000e-04 ## Epoch 278/500 ## 90/90 - 0s - 1ms/step - loss: 0.0498 - sparse_categorical_accuracy: 0.9840 - val_loss: 0.1267 - val_sparse_categorical_accuracy: 0.9598 - learning_rate: 1.0000e-04 ## Epoch 279/500 ## 90/90 - 0s - 1ms/step - loss: 0.0517 - sparse_categorical_accuracy: 0.9830 - val_loss: 0.1037 - val_sparse_categorical_accuracy: 0.9653 - learning_rate: 1.0000e-04 ## Epoch 280/500 ## 90/90 - 0s - 1ms/step - loss: 0.0501 - sparse_categorical_accuracy: 0.9830 - val_loss: 0.0985 - val_sparse_categorical_accuracy: 0.9667 - learning_rate: 1.0000e-04 ## Epoch 281/500 ## 90/90 - 0s - 1ms/step - loss: 0.0468 - sparse_categorical_accuracy: 0.9865 - val_loss: 0.1119 - val_sparse_categorical_accuracy: 0.9653 - learning_rate: 1.0000e-04 ## Epoch 281: early stopping"},{"path":"https://keras3.posit.co/dev/articles/examples/timeseries/timeseries_classification_from_scratch.html","id":"evaluate-model-on-test-data","dir":"Articles > Examples > Timeseries","previous_headings":"","what":"Evaluate model on test data","title":"Timeseries classification from scratch","text":"","code":"model <- load_model(\"best_model.keras\")  results <- model |> evaluate(x_test, y_test) ## 42/42 - 1s - 16ms/step - loss: 0.0938 - sparse_categorical_accuracy: 0.9697 str(results) ## List of 2 ##  $ loss                       : num 0.0938 ##  $ sparse_categorical_accuracy: num 0.97 cat(   \"Test accuracy: \", results$sparse_categorical_accuracy, \"\\n\",   \"Test loss: \", results$loss, \"\\n\",   sep = \"\" ) ## Test accuracy: 0.969697 ## Test loss: 0.09383929"},{"path":"https://keras3.posit.co/dev/articles/examples/timeseries/timeseries_classification_from_scratch.html","id":"plot-the-models-training-history","dir":"Articles > Examples > Timeseries","previous_headings":"","what":"Plot the model’s training history","title":"Timeseries classification from scratch","text":"Plot just training validation accuracy: can see training accuracy reaches almost 0.95 100 epochs. However, observing validation accuracy can see network still needs training reaches almost 0.97 validation training accuracy 200 epochs. Beyond 200th epoch, continue training, validation accuracy start decreasing training accuracy continue increasing: model starts overfitting.","code":"plot(history) plot(history, metric = \"sparse_categorical_accuracy\") +   # scale x axis to actual number of epochs run before early stopping   ggplot2::xlim(0, length(history$metrics$loss))"},{"path":"https://keras3.posit.co/dev/articles/examples/vision/autoencoder.html","id":"introduction","dir":"Articles > Examples > Vision","previous_headings":"","what":"Introduction","title":"Convolutional autoencoder for image denoising","text":"example demonstrates implement deep convolutional autoencoder image denoising, mapping noisy digits images MNIST dataset clean digits images. implementation based original blog post titled Building Autoencoders Keras François Chollet.","code":""},{"path":"https://keras3.posit.co/dev/articles/examples/vision/autoencoder.html","id":"setup","dir":"Articles > Examples > Vision","previous_headings":"","what":"Setup","title":"Convolutional autoencoder for image denoising","text":"","code":"library(keras3)  # Normalizes the supplied array and reshapes it. preprocess <- function(array) {   array_reshape(array/255, c(dim(array)[1], 28, 28, 1)) }  # Adds random noise to each image in the supplied array. noise <- function(array) {   noise_factor <- 0.4   noisy_array <- array + noise_factor * random_normal(dim(array))   op_clip(noisy_array, 0.0, 1.0) }  display <- function(array1, array2) {   n <- 2   indices <- sample.int(dim(array1)[1], n)   images1 <- as.array(array1)[indices, , , ]   images2 <- as.array(array2)[indices, , , ]    par(mfrow = c(2, n), mar = c(0, 0, 0, 0))   for (i in seq_len(n)) {     plot(as.raster(images1[i, , ]))     plot(as.raster(images2[i, , ]))   } }"},{"path":"https://keras3.posit.co/dev/articles/examples/vision/autoencoder.html","id":"prepare-the-data","dir":"Articles > Examples > Vision","previous_headings":"","what":"Prepare the data","title":"Convolutional autoencoder for image denoising","text":"","code":"# Since we only need images from the dataset to encode and decode, we # won't use the labels. c(c(train_data, .), c(test_data, .)) %<-% dataset_mnist()  # Normalize and reshape the data train_data <- preprocess(train_data) test_data <- preprocess(test_data)  # Create a copy of the data with added noise noisy_train_data <- noise(train_data) noisy_test_data <- noise(test_data)  # Display the train data and a version of it with added noise display(train_data, noisy_train_data)"},{"path":"https://keras3.posit.co/dev/articles/examples/vision/autoencoder.html","id":"build-the-autoencoder","dir":"Articles > Examples > Vision","previous_headings":"","what":"Build the autoencoder","title":"Convolutional autoencoder for image denoising","text":"going use Functional API build convolutional autoencoder. Now can train autoencoder using train_data input data target. Notice setting validation data using format. Let’s predict test dataset display original image together prediction autoencoder. Notice predictions pretty close original images, although quite . Now know autoencoder works, let’s retrain using noisy data input clean data target. want autoencoder learn denoise images. Let’s now predict noisy data display results autoencoder. Notice autoencoder amazing job removing noise input images.","code":"input <- keras_input(shape = c(28, 28, 1))  # Encoder enc <- input |>   layer_conv_2d(filters = 32, kernel_size = c(3, 3),                 activation = \"relu\", padding = \"same\") |>   layer_max_pooling_2d(pool_size = c(2, 2), padding = \"same\") |>   layer_conv_2d(filters = 32, kernel_size = c(3, 3),                 activation = \"relu\", padding = \"same\") |>   layer_max_pooling_2d(pool_size = c(2, 2), padding = \"same\")  # Decoder dec <- enc |>   layer_conv_2d_transpose(filters = 32, kernel_size = c(3, 3), strides = 2,                           activation = \"relu\", padding = \"same\") |>   layer_conv_2d_transpose(filters = 32, kernel_size = c(3, 3), strides = 2,                           activation = \"relu\", padding = \"same\") |>   layer_conv_2d(filters = 1, kernel_size = c(3, 3),                 activation = \"sigmoid\", padding = \"same\")  # Autoencoder autoencoder <- keras_model(input, dec) autoencoder |> compile(optimizer = \"adam\", loss = \"binary_crossentropy\") autoencoder |> summary() ## Model: \"functional\" ## ┏━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━┓ ## ┃ Layer (type)                    ┃ Output Shape           ┃       Param # ┃ ## ┡━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━┩ ## │ input_layer (InputLayer)        │ (None, 28, 28, 1)      │             0 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ conv2d (Conv2D)                 │ (None, 28, 28, 32)     │           320 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ max_pooling2d (MaxPooling2D)    │ (None, 14, 14, 32)     │             0 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ conv2d_1 (Conv2D)               │ (None, 14, 14, 32)     │         9,248 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ max_pooling2d_1 (MaxPooling2D)  │ (None, 7, 7, 32)       │             0 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ conv2d_transpose                │ (None, 14, 14, 32)     │         9,248 │ ## │ (Conv2DTranspose)               │                        │               │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ conv2d_transpose_1              │ (None, 28, 28, 32)     │         9,248 │ ## │ (Conv2DTranspose)               │                        │               │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ conv2d_2 (Conv2D)               │ (None, 28, 28, 1)      │           289 │ ## └─────────────────────────────────┴────────────────────────┴───────────────┘ ##  Total params: 28,353 (110.75 KB) ##  Trainable params: 28,353 (110.75 KB) ##  Non-trainable params: 0 (0.00 B) autoencoder |> fit(   x = train_data,   y = train_data,   epochs = 50,   batch_size = 128,   shuffle = TRUE,   validation_data = list(test_data, test_data), ) ## Epoch 1/50 ## 469/469 - 6s - 12ms/step - loss: 0.1371 - val_loss: 0.0727 ## Epoch 2/50 ## 469/469 - 2s - 3ms/step - loss: 0.0712 - val_loss: 0.0692 ## Epoch 3/50 ## 469/469 - 2s - 3ms/step - loss: 0.0690 - val_loss: 0.0678 ## Epoch 4/50 ## 469/469 - 2s - 3ms/step - loss: 0.0679 - val_loss: 0.0671 ## Epoch 5/50 ## 469/469 - 2s - 3ms/step - loss: 0.0672 - val_loss: 0.0665 ## Epoch 6/50 ## 469/469 - 2s - 3ms/step - loss: 0.0666 - val_loss: 0.0660 ## Epoch 7/50 ## 469/469 - 2s - 3ms/step - loss: 0.0662 - val_loss: 0.0656 ## Epoch 8/50 ## 469/469 - 2s - 3ms/step - loss: 0.0659 - val_loss: 0.0652 ## Epoch 9/50 ## 469/469 - 2s - 3ms/step - loss: 0.0656 - val_loss: 0.0650 ## Epoch 10/50 ## 469/469 - 1s - 3ms/step - loss: 0.0653 - val_loss: 0.0648 ## Epoch 11/50 ## 469/469 - 2s - 3ms/step - loss: 0.0651 - val_loss: 0.0646 ## Epoch 12/50 ## 469/469 - 1s - 3ms/step - loss: 0.0649 - val_loss: 0.0644 ## Epoch 13/50 ## 469/469 - 2s - 3ms/step - loss: 0.0648 - val_loss: 0.0643 ## Epoch 14/50 ## 469/469 - 1s - 3ms/step - loss: 0.0646 - val_loss: 0.0641 ## Epoch 15/50 ## 469/469 - 1s - 3ms/step - loss: 0.0645 - val_loss: 0.0640 ## Epoch 16/50 ## 469/469 - 1s - 3ms/step - loss: 0.0644 - val_loss: 0.0639 ## Epoch 17/50 ## 469/469 - 2s - 3ms/step - loss: 0.0642 - val_loss: 0.0638 ## Epoch 18/50 ## 469/469 - 2s - 3ms/step - loss: 0.0641 - val_loss: 0.0637 ## Epoch 19/50 ## 469/469 - 1s - 3ms/step - loss: 0.0640 - val_loss: 0.0636 ## Epoch 20/50 ## 469/469 - 2s - 3ms/step - loss: 0.0639 - val_loss: 0.0635 ## Epoch 21/50 ## 469/469 - 2s - 3ms/step - loss: 0.0639 - val_loss: 0.0635 ## Epoch 22/50 ## 469/469 - 1s - 3ms/step - loss: 0.0638 - val_loss: 0.0634 ## Epoch 23/50 ## 469/469 - 2s - 3ms/step - loss: 0.0637 - val_loss: 0.0634 ## Epoch 24/50 ## 469/469 - 2s - 3ms/step - loss: 0.0637 - val_loss: 0.0633 ## Epoch 25/50 ## 469/469 - 2s - 3ms/step - loss: 0.0636 - val_loss: 0.0632 ## Epoch 26/50 ## 469/469 - 2s - 3ms/step - loss: 0.0636 - val_loss: 0.0632 ## Epoch 27/50 ## 469/469 - 2s - 3ms/step - loss: 0.0635 - val_loss: 0.0631 ## Epoch 28/50 ## 469/469 - 2s - 3ms/step - loss: 0.0635 - val_loss: 0.0631 ## Epoch 29/50 ## 469/469 - 2s - 3ms/step - loss: 0.0634 - val_loss: 0.0631 ## Epoch 30/50 ## 469/469 - 1s - 3ms/step - loss: 0.0634 - val_loss: 0.0630 ## Epoch 31/50 ## 469/469 - 2s - 3ms/step - loss: 0.0633 - val_loss: 0.0630 ## Epoch 32/50 ## 469/469 - 2s - 3ms/step - loss: 0.0633 - val_loss: 0.0630 ## Epoch 33/50 ## 469/469 - 2s - 3ms/step - loss: 0.0633 - val_loss: 0.0629 ## Epoch 34/50 ## 469/469 - 2s - 3ms/step - loss: 0.0632 - val_loss: 0.0629 ## Epoch 35/50 ## 469/469 - 1s - 3ms/step - loss: 0.0632 - val_loss: 0.0628 ## Epoch 36/50 ## 469/469 - 2s - 3ms/step - loss: 0.0632 - val_loss: 0.0628 ## Epoch 37/50 ## 469/469 - 1s - 3ms/step - loss: 0.0631 - val_loss: 0.0628 ## Epoch 38/50 ## 469/469 - 2s - 3ms/step - loss: 0.0631 - val_loss: 0.0628 ## Epoch 39/50 ## 469/469 - 1s - 3ms/step - loss: 0.0631 - val_loss: 0.0627 ## Epoch 40/50 ## 469/469 - 2s - 3ms/step - loss: 0.0630 - val_loss: 0.0627 ## Epoch 41/50 ## 469/469 - 2s - 3ms/step - loss: 0.0630 - val_loss: 0.0627 ## Epoch 42/50 ## 469/469 - 2s - 3ms/step - loss: 0.0630 - val_loss: 0.0626 ## Epoch 43/50 ## 469/469 - 2s - 3ms/step - loss: 0.0630 - val_loss: 0.0626 ## Epoch 44/50 ## 469/469 - 2s - 3ms/step - loss: 0.0629 - val_loss: 0.0626 ## Epoch 45/50 ## 469/469 - 2s - 3ms/step - loss: 0.0629 - val_loss: 0.0626 ## Epoch 46/50 ## 469/469 - 2s - 3ms/step - loss: 0.0629 - val_loss: 0.0625 ## Epoch 47/50 ## 469/469 - 2s - 3ms/step - loss: 0.0629 - val_loss: 0.0625 ## Epoch 48/50 ## 469/469 - 1s - 3ms/step - loss: 0.0628 - val_loss: 0.0625 ## Epoch 49/50 ## 469/469 - 2s - 3ms/step - loss: 0.0628 - val_loss: 0.0625 ## Epoch 50/50 ## 469/469 - 2s - 3ms/step - loss: 0.0628 - val_loss: 0.0625 predictions <- autoencoder |> predict(test_data) ## 313/313 - 1s - 2ms/step display(test_data, predictions) autoencoder |> fit(   x = noisy_train_data,   y = train_data,   epochs = 100,   batch_size = 128,   shuffle = TRUE,   validation_data = list(noisy_test_data, test_data), ) ## Epoch 1/100 ## 469/469 - 2s - 3ms/step - loss: 0.1006 - val_loss: 0.0939 ## Epoch 2/100 ## 469/469 - 2s - 3ms/step - loss: 0.0936 - val_loss: 0.0919 ## Epoch 3/100 ## 469/469 - 2s - 3ms/step - loss: 0.0921 - val_loss: 0.0908 ## Epoch 4/100 ## 469/469 - 2s - 3ms/step - loss: 0.0912 - val_loss: 0.0902 ## Epoch 5/100 ## 469/469 - 2s - 3ms/step - loss: 0.0905 - val_loss: 0.0896 ## Epoch 6/100 ## 469/469 - 2s - 3ms/step - loss: 0.0900 - val_loss: 0.0891 ## Epoch 7/100 ## 469/469 - 2s - 3ms/step - loss: 0.0895 - val_loss: 0.0887 ## Epoch 8/100 ## 469/469 - 2s - 3ms/step - loss: 0.0891 - val_loss: 0.0883 ## Epoch 9/100 ## 469/469 - 2s - 3ms/step - loss: 0.0887 - val_loss: 0.0880 ## Epoch 10/100 ## 469/469 - 2s - 3ms/step - loss: 0.0884 - val_loss: 0.0878 ## Epoch 11/100 ## 469/469 - 2s - 3ms/step - loss: 0.0882 - val_loss: 0.0875 ## Epoch 12/100 ## 469/469 - 2s - 3ms/step - loss: 0.0879 - val_loss: 0.0873 ## Epoch 13/100 ## 469/469 - 2s - 3ms/step - loss: 0.0877 - val_loss: 0.0871 ## Epoch 14/100 ## 469/469 - 2s - 3ms/step - loss: 0.0875 - val_loss: 0.0869 ## Epoch 15/100 ## 469/469 - 2s - 3ms/step - loss: 0.0873 - val_loss: 0.0867 ## Epoch 16/100 ## 469/469 - 2s - 3ms/step - loss: 0.0872 - val_loss: 0.0866 ## Epoch 17/100 ## 469/469 - 2s - 3ms/step - loss: 0.0870 - val_loss: 0.0864 ## Epoch 18/100 ## 469/469 - 2s - 3ms/step - loss: 0.0869 - val_loss: 0.0863 ## Epoch 19/100 ## 469/469 - 2s - 3ms/step - loss: 0.0868 - val_loss: 0.0862 ## Epoch 20/100 ## 469/469 - 2s - 3ms/step - loss: 0.0866 - val_loss: 0.0861 ## Epoch 21/100 ## 469/469 - 2s - 3ms/step - loss: 0.0865 - val_loss: 0.0860 ## Epoch 22/100 ## 469/469 - 2s - 3ms/step - loss: 0.0864 - val_loss: 0.0859 ## Epoch 23/100 ## 469/469 - 2s - 3ms/step - loss: 0.0863 - val_loss: 0.0858 ## Epoch 24/100 ## 469/469 - 1s - 3ms/step - loss: 0.0862 - val_loss: 0.0857 ## Epoch 25/100 ## 469/469 - 2s - 3ms/step - loss: 0.0862 - val_loss: 0.0856 ## Epoch 26/100 ## 469/469 - 2s - 3ms/step - loss: 0.0861 - val_loss: 0.0856 ## Epoch 27/100 ## 469/469 - 2s - 3ms/step - loss: 0.0860 - val_loss: 0.0855 ## Epoch 28/100 ## 469/469 - 2s - 3ms/step - loss: 0.0859 - val_loss: 0.0855 ## Epoch 29/100 ## 469/469 - 2s - 3ms/step - loss: 0.0859 - val_loss: 0.0854 ## Epoch 30/100 ## 469/469 - 2s - 3ms/step - loss: 0.0858 - val_loss: 0.0853 ## Epoch 31/100 ## 469/469 - 2s - 3ms/step - loss: 0.0858 - val_loss: 0.0853 ## Epoch 32/100 ## 469/469 - 2s - 3ms/step - loss: 0.0857 - val_loss: 0.0853 ## Epoch 33/100 ## 469/469 - 2s - 3ms/step - loss: 0.0857 - val_loss: 0.0852 ## Epoch 34/100 ## 469/469 - 2s - 3ms/step - loss: 0.0856 - val_loss: 0.0852 ## Epoch 35/100 ## 469/469 - 2s - 3ms/step - loss: 0.0856 - val_loss: 0.0851 ## Epoch 36/100 ## 469/469 - 2s - 3ms/step - loss: 0.0855 - val_loss: 0.0851 ## Epoch 37/100 ## 469/469 - 2s - 3ms/step - loss: 0.0855 - val_loss: 0.0851 ## Epoch 38/100 ## 469/469 - 2s - 3ms/step - loss: 0.0855 - val_loss: 0.0850 ## Epoch 39/100 ## 469/469 - 2s - 3ms/step - loss: 0.0854 - val_loss: 0.0850 ## Epoch 40/100 ## 469/469 - 2s - 3ms/step - loss: 0.0854 - val_loss: 0.0850 ## Epoch 41/100 ## 469/469 - 2s - 3ms/step - loss: 0.0854 - val_loss: 0.0850 ## Epoch 42/100 ## 469/469 - 2s - 3ms/step - loss: 0.0853 - val_loss: 0.0849 ## Epoch 43/100 ## 469/469 - 2s - 3ms/step - loss: 0.0853 - val_loss: 0.0849 ## Epoch 44/100 ## 469/469 - 2s - 3ms/step - loss: 0.0853 - val_loss: 0.0849 ## Epoch 45/100 ## 469/469 - 2s - 3ms/step - loss: 0.0853 - val_loss: 0.0849 ## Epoch 46/100 ## 469/469 - 2s - 3ms/step - loss: 0.0852 - val_loss: 0.0848 ## Epoch 47/100 ## 469/469 - 2s - 3ms/step - loss: 0.0852 - val_loss: 0.0848 ## Epoch 48/100 ## 469/469 - 2s - 3ms/step - loss: 0.0852 - val_loss: 0.0848 ## Epoch 49/100 ## 469/469 - 2s - 3ms/step - loss: 0.0852 - val_loss: 0.0848 ## Epoch 50/100 ## 469/469 - 2s - 3ms/step - loss: 0.0851 - val_loss: 0.0848 ## Epoch 51/100 ## 469/469 - 2s - 3ms/step - loss: 0.0851 - val_loss: 0.0847 ## Epoch 52/100 ## 469/469 - 2s - 3ms/step - loss: 0.0851 - val_loss: 0.0847 ## Epoch 53/100 ## 469/469 - 2s - 3ms/step - loss: 0.0851 - val_loss: 0.0847 ## Epoch 54/100 ## 469/469 - 2s - 3ms/step - loss: 0.0850 - val_loss: 0.0847 ## Epoch 55/100 ## 469/469 - 2s - 3ms/step - loss: 0.0850 - val_loss: 0.0847 ## Epoch 56/100 ## 469/469 - 2s - 3ms/step - loss: 0.0850 - val_loss: 0.0846 ## Epoch 57/100 ## 469/469 - 2s - 3ms/step - loss: 0.0850 - val_loss: 0.0846 ## Epoch 58/100 ## 469/469 - 2s - 3ms/step - loss: 0.0850 - val_loss: 0.0846 ## Epoch 59/100 ## 469/469 - 2s - 3ms/step - loss: 0.0849 - val_loss: 0.0846 ## Epoch 60/100 ## 469/469 - 2s - 3ms/step - loss: 0.0849 - val_loss: 0.0846 ## Epoch 61/100 ## 469/469 - 2s - 3ms/step - loss: 0.0849 - val_loss: 0.0846 ## Epoch 62/100 ## 469/469 - 2s - 3ms/step - loss: 0.0849 - val_loss: 0.0846 ## Epoch 63/100 ## 469/469 - 1s - 3ms/step - loss: 0.0849 - val_loss: 0.0845 ## Epoch 64/100 ## 469/469 - 2s - 3ms/step - loss: 0.0849 - val_loss: 0.0845 ## Epoch 65/100 ## 469/469 - 2s - 3ms/step - loss: 0.0848 - val_loss: 0.0845 ## Epoch 66/100 ## 469/469 - 2s - 3ms/step - loss: 0.0848 - val_loss: 0.0845 ## Epoch 67/100 ## 469/469 - 2s - 3ms/step - loss: 0.0848 - val_loss: 0.0845 ## Epoch 68/100 ## 469/469 - 2s - 3ms/step - loss: 0.0848 - val_loss: 0.0845 ## Epoch 69/100 ## 469/469 - 2s - 3ms/step - loss: 0.0848 - val_loss: 0.0845 ## Epoch 70/100 ## 469/469 - 2s - 3ms/step - loss: 0.0848 - val_loss: 0.0845 ## Epoch 71/100 ## 469/469 - 2s - 3ms/step - loss: 0.0848 - val_loss: 0.0844 ## Epoch 72/100 ## 469/469 - 2s - 3ms/step - loss: 0.0847 - val_loss: 0.0844 ## Epoch 73/100 ## 469/469 - 2s - 3ms/step - loss: 0.0847 - val_loss: 0.0844 ## Epoch 74/100 ## 469/469 - 2s - 3ms/step - loss: 0.0847 - val_loss: 0.0844 ## Epoch 75/100 ## 469/469 - 1s - 3ms/step - loss: 0.0847 - val_loss: 0.0844 ## Epoch 76/100 ## 469/469 - 2s - 3ms/step - loss: 0.0847 - val_loss: 0.0844 ## Epoch 77/100 ## 469/469 - 2s - 3ms/step - loss: 0.0847 - val_loss: 0.0844 ## Epoch 78/100 ## 469/469 - 2s - 3ms/step - loss: 0.0847 - val_loss: 0.0844 ## Epoch 79/100 ## 469/469 - 2s - 3ms/step - loss: 0.0847 - val_loss: 0.0844 ## Epoch 80/100 ## 469/469 - 2s - 3ms/step - loss: 0.0847 - val_loss: 0.0844 ## Epoch 81/100 ## 469/469 - 2s - 3ms/step - loss: 0.0846 - val_loss: 0.0844 ## Epoch 82/100 ## 469/469 - 2s - 3ms/step - loss: 0.0846 - val_loss: 0.0844 ## Epoch 83/100 ## 469/469 - 2s - 3ms/step - loss: 0.0846 - val_loss: 0.0844 ## Epoch 84/100 ## 469/469 - 2s - 3ms/step - loss: 0.0846 - val_loss: 0.0843 ## Epoch 85/100 ## 469/469 - 2s - 3ms/step - loss: 0.0846 - val_loss: 0.0843 ## Epoch 86/100 ## 469/469 - 2s - 3ms/step - loss: 0.0846 - val_loss: 0.0843 ## Epoch 87/100 ## 469/469 - 1s - 3ms/step - loss: 0.0846 - val_loss: 0.0843 ## Epoch 88/100 ## 469/469 - 2s - 3ms/step - loss: 0.0846 - val_loss: 0.0843 ## Epoch 89/100 ## 469/469 - 2s - 3ms/step - loss: 0.0846 - val_loss: 0.0843 ## Epoch 90/100 ## 469/469 - 2s - 3ms/step - loss: 0.0846 - val_loss: 0.0843 ## Epoch 91/100 ## 469/469 - 2s - 3ms/step - loss: 0.0845 - val_loss: 0.0843 ## Epoch 92/100 ## 469/469 - 2s - 3ms/step - loss: 0.0845 - val_loss: 0.0843 ## Epoch 93/100 ## 469/469 - 2s - 3ms/step - loss: 0.0845 - val_loss: 0.0843 ## Epoch 94/100 ## 469/469 - 2s - 3ms/step - loss: 0.0845 - val_loss: 0.0843 ## Epoch 95/100 ## 469/469 - 2s - 3ms/step - loss: 0.0845 - val_loss: 0.0843 ## Epoch 96/100 ## 469/469 - 1s - 3ms/step - loss: 0.0845 - val_loss: 0.0842 ## Epoch 97/100 ## 469/469 - 2s - 3ms/step - loss: 0.0845 - val_loss: 0.0842 ## Epoch 98/100 ## 469/469 - 2s - 3ms/step - loss: 0.0845 - val_loss: 0.0842 ## Epoch 99/100 ## 469/469 - 2s - 3ms/step - loss: 0.0845 - val_loss: 0.0842 ## Epoch 100/100 ## 469/469 - 1s - 3ms/step - loss: 0.0845 - val_loss: 0.0842 predictions <- autoencoder |> predict(noisy_test_data) ## 313/313 - 0s - 775us/step display(noisy_test_data, predictions)"},{"path":"https://keras3.posit.co/dev/articles/examples/vision/mnist_convnet.html","id":"setup","dir":"Articles > Examples > Vision","previous_headings":"","what":"Setup","title":"Simple MNIST convnet","text":"","code":"library(keras3)"},{"path":"https://keras3.posit.co/dev/articles/examples/vision/mnist_convnet.html","id":"prepare-the-data","dir":"Articles > Examples > Vision","previous_headings":"","what":"Prepare the data","title":"Simple MNIST convnet","text":"","code":"# Model / data parameters num_classes <- 10 input_shape <- c(28, 28, 1)  # Load the data and split it between train and test sets c(c(x_train, y_train), c(x_test, y_test)) %<-% dataset_mnist()  # Scale images to the [0, 1] range x_train <- x_train / 255 x_test <- x_test / 255 # Make sure images have shape (28, 28, 1) x_train <- op_expand_dims(x_train, -1) x_test <- op_expand_dims(x_test, -1)   dim(x_train) ## [1] 60000    28    28     1 dim(x_test) ## [1] 10000    28    28     1 # convert class vectors to binary class matrices y_train <- to_categorical(y_train, num_classes) y_test <- to_categorical(y_test, num_classes)"},{"path":"https://keras3.posit.co/dev/articles/examples/vision/mnist_convnet.html","id":"build-the-model","dir":"Articles > Examples > Vision","previous_headings":"","what":"Build the model","title":"Simple MNIST convnet","text":"","code":"model <- keras_model_sequential(input_shape = input_shape) model |>   layer_conv_2d(filters = 32, kernel_size = c(3, 3), activation = \"relu\") |>   layer_max_pooling_2d(pool_size = c(2, 2)) |>   layer_conv_2d(filters = 64, kernel_size = c(3, 3), activation = \"relu\") |>   layer_max_pooling_2d(pool_size = c(2, 2)) |>   layer_flatten() |>   layer_dropout(rate = 0.5) |>   layer_dense(units = num_classes, activation = \"softmax\")  summary(model) ## Model: \"sequential\" ## ┏━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━┓ ## ┃ Layer (type)                    ┃ Output Shape           ┃       Param # ┃ ## ┡━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━┩ ## │ conv2d (Conv2D)                 │ (None, 26, 26, 32)     │           320 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ max_pooling2d (MaxPooling2D)    │ (None, 13, 13, 32)     │             0 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ conv2d_1 (Conv2D)               │ (None, 11, 11, 64)     │        18,496 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ max_pooling2d_1 (MaxPooling2D)  │ (None, 5, 5, 64)       │             0 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ flatten (Flatten)               │ (None, 1600)           │             0 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ dropout (Dropout)               │ (None, 1600)           │             0 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ dense (Dense)                   │ (None, 10)             │        16,010 │ ## └─────────────────────────────────┴────────────────────────┴───────────────┘ ##  Total params: 34,826 (136.04 KB) ##  Trainable params: 34,826 (136.04 KB) ##  Non-trainable params: 0 (0.00 B)"},{"path":"https://keras3.posit.co/dev/articles/examples/vision/mnist_convnet.html","id":"train-the-model","dir":"Articles > Examples > Vision","previous_headings":"","what":"Train the model","title":"Simple MNIST convnet","text":"","code":"batch_size <- 128 epochs <- 15  model |> compile(   loss = \"categorical_crossentropy\",   optimizer = \"adam\",   metrics = \"accuracy\" )  model |> fit(   x_train, y_train,   batch_size = batch_size,   epochs = epochs,   validation_split = 0.1 ) ## Epoch 1/15 ## 422/422 - 5s - 11ms/step - accuracy: 0.8896 - loss: 0.3635 - val_accuracy: 0.9783 - val_loss: 0.0791 ## Epoch 2/15 ## 422/422 - 1s - 2ms/step - accuracy: 0.9666 - loss: 0.1111 - val_accuracy: 0.9850 - val_loss: 0.0548 ## Epoch 3/15 ## 422/422 - 1s - 2ms/step - accuracy: 0.9744 - loss: 0.0824 - val_accuracy: 0.9883 - val_loss: 0.0441 ## Epoch 4/15 ## 422/422 - 1s - 2ms/step - accuracy: 0.9786 - loss: 0.0694 - val_accuracy: 0.9895 - val_loss: 0.0401 ## Epoch 5/15 ## 422/422 - 1s - 2ms/step - accuracy: 0.9804 - loss: 0.0625 - val_accuracy: 0.9903 - val_loss: 0.0353 ## Epoch 6/15 ## 422/422 - 1s - 2ms/step - accuracy: 0.9824 - loss: 0.0557 - val_accuracy: 0.9907 - val_loss: 0.0332 ## Epoch 7/15 ## 422/422 - 1s - 2ms/step - accuracy: 0.9838 - loss: 0.0500 - val_accuracy: 0.9917 - val_loss: 0.0312 ## Epoch 8/15 ## 422/422 - 1s - 2ms/step - accuracy: 0.9850 - loss: 0.0480 - val_accuracy: 0.9918 - val_loss: 0.0304 ## Epoch 9/15 ## 422/422 - 1s - 2ms/step - accuracy: 0.9860 - loss: 0.0445 - val_accuracy: 0.9918 - val_loss: 0.0303 ## Epoch 10/15 ## 422/422 - 1s - 2ms/step - accuracy: 0.9861 - loss: 0.0442 - val_accuracy: 0.9912 - val_loss: 0.0293 ## Epoch 11/15 ## 422/422 - 1s - 2ms/step - accuracy: 0.9876 - loss: 0.0390 - val_accuracy: 0.9915 - val_loss: 0.0307 ## Epoch 12/15 ## 422/422 - 1s - 2ms/step - accuracy: 0.9876 - loss: 0.0374 - val_accuracy: 0.9925 - val_loss: 0.0288 ## Epoch 13/15 ## 422/422 - 1s - 2ms/step - accuracy: 0.9887 - loss: 0.0346 - val_accuracy: 0.9920 - val_loss: 0.0291 ## Epoch 14/15 ## 422/422 - 1s - 2ms/step - accuracy: 0.9887 - loss: 0.0343 - val_accuracy: 0.9923 - val_loss: 0.0276 ## Epoch 15/15 ## 422/422 - 1s - 2ms/step - accuracy: 0.9897 - loss: 0.0323 - val_accuracy: 0.9922 - val_loss: 0.0286"},{"path":"https://keras3.posit.co/dev/articles/examples/vision/mnist_convnet.html","id":"evaluate-the-trained-model","dir":"Articles > Examples > Vision","previous_headings":"","what":"Evaluate the trained model","title":"Simple MNIST convnet","text":"","code":"score <- model |> evaluate(x_test, y_test, verbose = 0) score ## $accuracy ## [1] 0.9915 ## ## $loss ## [1] 0.02471612"},{"path":"https://keras3.posit.co/dev/articles/examples/vision/mnist_siamese_graph.html","id":"introduction","dir":"Articles > Examples > Vision","previous_headings":"","what":"Introduction","title":"Train a Siamese MLP on pairs of digits from the MNIST dataset.","text":"Siamese Networks neural networks share weights two sister networks, producing embedding vectors respective inputs. supervised similarity learning, networks trained maximize contrast (distance) embeddings inputs different classes, minimizing distance embeddings similar classes, resulting embedding spaces reflect class segmentation training inputs. implementation loosely follows Hadsell-et-al.’06 [1] (see paper mode details) euclidean distance replaced subtraction layer one fully-connect (FC) layer. [1] “Dimensionality Reduction Learning Invariant Mapping” https://yann.lecun.com/exdb/publis/pdf/hadsell-chopra-lecun-06.pdf Gets 98.11% test accuracy 20 epochs. 3 seconds per epoch AMD Ryzen 7 PRO 4750U (CPU)","code":"library(keras3) contrastive_loss <- function(y_true, y_pred) {     # Contrastive loss from Hadsell-et-al.'06     # https://yann.lecun.com/exdb/publis/pdf/hadsell-chopra-lecun-06.pdf     margin = 1     margin_square = op_square(op_maximum(margin - (y_pred), 0))     op_mean((1 - y_true) * op_square(y_pred) + (y_true) * margin_square) }"},{"path":"https://keras3.posit.co/dev/articles/examples/vision/mnist_siamese_graph.html","id":"create-pairs-of-images","dir":"Articles > Examples > Vision","previous_headings":"","what":"Create pairs of images","title":"Train a Siamese MLP on pairs of digits from the MNIST dataset.","text":"train model differentiate digits different classes. example, digit 0 needs differentiated rest digits (1 9), digit 1 - 0 2 9, . carry , select N random images class (example, digit 0) pair N random images another class B (example, digit 1). , can repeat process classes digits (digit 9). paired digit 0 digits, can repeat process remaining classes rest digits (1 9).","code":"create_pairs <- function(x, y) {     # Positive and negative pair creation.     # Alternates between positive and negative pairs.     digit_indices <- tapply(1:length(y), y, list)     y1 <- y     y2 <- sapply(y, function(a) sample(0:9,1,prob=0.1+0.8*(0:9==a)))     idx1 <- 1:nrow(x)     idx2 <- sapply(as.character(y2), function(a) sample(digit_indices[[a]],1))     is_same  <- 1*(y1==y2)     list(pair1 = x[idx1,], pair2 = x[idx2,], y = is_same) }  compute_accuracy <- function(predictions, labels) {     # Compute classification accuracy with a fixed threshold on distances.     mean(labels[predictions > 0.5]) } # the data, shuffled and split between train and test sets mnist   <- dataset_mnist() x_train <- mnist$train$x y_train <- mnist$train$y x_test  <- mnist$test$x y_test  <- mnist$test$y x_train <- array_reshape(x_train, c(nrow(x_train), 784)) x_test  <- array_reshape(x_test, c(nrow(x_test), 784)) x_train <- x_train / 255 x_test  <- x_test / 255  # create training+test positive and negative pairs tr <- create_pairs(x_train, y_train) te <- create_pairs(x_test,  y_test)  names(tr) ## [1] \"pair1\" \"pair2\" \"y\""},{"path":"https://keras3.posit.co/dev/articles/examples/vision/mnist_siamese_graph.html","id":"network-definition","dir":"Articles > Examples > Vision","previous_headings":"","what":"Network definition","title":"Train a Siamese MLP on pairs of digits from the MNIST dataset.","text":"","code":"# input layers input_dim = 784 input_1 <- layer_input(shape = c(input_dim)) input_2 <- layer_input(shape = c(input_dim))  # definition of the base network that will be shared base_network <- keras_model_sequential() %>%     layer_dense(units = 128, activation = 'relu') %>%     layer_dropout(rate = 0.1) %>%     layer_dense(units = 128, activation = 'relu') %>%     layer_dropout(rate = 0.1) %>%     layer_dense(units = 128, activation = 'relu')  # because we re-use the same instance `base_network`, the weights of # the network will be shared across the two branches branch_1 <- base_network(input_1) branch_2 <- base_network(input_2)  # merging layer out <- layer_subtract(list(branch_1, branch_2)) %>%     layer_dropout(rate = 0.1) %>%     layer_dense(units = 16, activation = 'relu') %>%     layer_dense(1, activation = \"sigmoid\")  # create and compile model model <- keras_model(list(input_1, input_2), out)"},{"path":"https://keras3.posit.co/dev/articles/examples/vision/mnist_siamese_graph.html","id":"train","dir":"Articles > Examples > Vision","previous_headings":"","what":"Train","title":"Train a Siamese MLP on pairs of digits from the MNIST dataset.","text":"","code":"model %>% compile(     optimizer = \"rmsprop\",     #loss = \"binary_crossentropy\",     loss = contrastive_loss,     metrics = metric_binary_accuracy )  history <- model %>% fit(     list(tr$pair1, tr$pair2), tr$y,     batch_size = 128,     epochs = 20,     validation_data = list(         list(te$pair1, te$pair2),         te$y     ) ) ## Epoch 1/20 ## 469/469 - 6s - 13ms/step - binary_accuracy: 0.7569 - loss: 0.1640 - val_binary_accuracy: 0.8798 - val_loss: 0.0974 ## Epoch 2/20 ## 469/469 - 1s - 2ms/step - binary_accuracy: 0.8899 - loss: 0.0866 - val_binary_accuracy: 0.9227 - val_loss: 0.0626 ## Epoch 3/20 ## 469/469 - 1s - 2ms/step - binary_accuracy: 0.9233 - loss: 0.0603 - val_binary_accuracy: 0.9398 - val_loss: 0.0497 ## Epoch 4/20 ## 469/469 - 1s - 2ms/step - binary_accuracy: 0.9423 - loss: 0.0460 - val_binary_accuracy: 0.9480 - val_loss: 0.0399 ## Epoch 5/20 ## 469/469 - 1s - 2ms/step - binary_accuracy: 0.9561 - loss: 0.0356 - val_binary_accuracy: 0.9578 - val_loss: 0.0330 ## Epoch 6/20 ## 469/469 - 1s - 2ms/step - binary_accuracy: 0.9637 - loss: 0.0295 - val_binary_accuracy: 0.9613 - val_loss: 0.0313 ## Epoch 7/20 ## 469/469 - 1s - 2ms/step - binary_accuracy: 0.9686 - loss: 0.0252 - val_binary_accuracy: 0.9645 - val_loss: 0.0275 ## Epoch 8/20 ## 469/469 - 1s - 2ms/step - binary_accuracy: 0.9730 - loss: 0.0217 - val_binary_accuracy: 0.9672 - val_loss: 0.0264 ## Epoch 9/20 ## 469/469 - 1s - 2ms/step - binary_accuracy: 0.9768 - loss: 0.0190 - val_binary_accuracy: 0.9686 - val_loss: 0.0246 ## Epoch 10/20 ## 469/469 - 1s - 2ms/step - binary_accuracy: 0.9793 - loss: 0.0169 - val_binary_accuracy: 0.9701 - val_loss: 0.0242 ## Epoch 11/20 ## 469/469 - 1s - 2ms/step - binary_accuracy: 0.9805 - loss: 0.0158 - val_binary_accuracy: 0.9719 - val_loss: 0.0233 ## Epoch 12/20 ## 469/469 - 1s - 2ms/step - binary_accuracy: 0.9822 - loss: 0.0148 - val_binary_accuracy: 0.9718 - val_loss: 0.0230 ## Epoch 13/20 ## 469/469 - 1s - 2ms/step - binary_accuracy: 0.9833 - loss: 0.0135 - val_binary_accuracy: 0.9702 - val_loss: 0.0234 ## Epoch 14/20 ## 469/469 - 1s - 2ms/step - binary_accuracy: 0.9846 - loss: 0.0129 - val_binary_accuracy: 0.9730 - val_loss: 0.0220 ## Epoch 15/20 ## 469/469 - 1s - 2ms/step - binary_accuracy: 0.9857 - loss: 0.0117 - val_binary_accuracy: 0.9759 - val_loss: 0.0193 ## Epoch 16/20 ## 469/469 - 1s - 2ms/step - binary_accuracy: 0.9868 - loss: 0.0111 - val_binary_accuracy: 0.9749 - val_loss: 0.0217 ## Epoch 17/20 ## 469/469 - 1s - 2ms/step - binary_accuracy: 0.9872 - loss: 0.0106 - val_binary_accuracy: 0.9741 - val_loss: 0.0211 ## Epoch 18/20 ## 469/469 - 1s - 2ms/step - binary_accuracy: 0.9880 - loss: 0.0099 - val_binary_accuracy: 0.9746 - val_loss: 0.0201 ## Epoch 19/20 ## 469/469 - 1s - 2ms/step - binary_accuracy: 0.9879 - loss: 0.0100 - val_binary_accuracy: 0.9756 - val_loss: 0.0207 ## Epoch 20/20 ## 469/469 - 1s - 2ms/step - binary_accuracy: 0.9888 - loss: 0.0091 - val_binary_accuracy: 0.9776 - val_loss: 0.0184 plot(history)"},{"path":"https://keras3.posit.co/dev/articles/examples/vision/mnist_siamese_graph.html","id":"evaluate","dir":"Articles > Examples > Vision","previous_headings":"","what":"Evaluate","title":"Train a Siamese MLP on pairs of digits from the MNIST dataset.","text":"","code":"# compute final accuracy on training and test sets  tr_pred <- predict(model, list(tr$pair1, tr$pair2))[,1] ## 1875/1875 - 2s - 1ms/step tr_acc  <- compute_accuracy(tr_pred, tr$y) te_pred <- predict(model, list(te$pair1, te$pair2))[,1] ## 313/313 - 1s - 2ms/step te_acc  <- compute_accuracy(te_pred, te$y)  sprintf('* Accuracy on training set: %0.2f%%', (100 * tr_acc)) ## [1] \"* Accuracy on training set: 99.67%\" sprintf('* Accuracy on test set: %0.2f%%', (100 * te_acc)) ## [1] \"* Accuracy on test set: 98.25%\""},{"path":"https://keras3.posit.co/dev/articles/examples/vision/mnist_siamese_graph.html","id":"plots","dir":"Articles > Examples > Vision","previous_headings":"","what":"Plots","title":"Train a Siamese MLP on pairs of digits from the MNIST dataset.","text":"","code":"par(mfrow=c(1,1)) vioplot::vioplot( te_pred ~ te$y ) i=3 visualizePair <- function(i) {     image(rbind(matrix( te$pair1[i,],28,28)[,28:1], matrix( te$pair2[i,],28,28)[,28:1]))     title(paste(\"true:\", te$y[i],\"|  pred:\", round(te_pred[i],5))) } par(mfrow=c(3,3)) lapply(1:9, visualizePair) ## [[1]] ## NULL ## ## [[2]] ## NULL ## ## [[3]] ## NULL ## ## [[4]] ## NULL ## ## [[5]] ## NULL ## ## [[6]] ## NULL ## ## [[7]] ## NULL ## ## [[8]] ## NULL ## ## [[9]] ## NULL"},{"path":"https://keras3.posit.co/dev/articles/examples/vision/oxford_pets_image_segmentation.html","id":"download-the-data","dir":"Articles > Examples > Vision","previous_headings":"","what":"Download the data","title":"Image segmentation with a U-Net-like architecture","text":"","code":"options(timeout = 5000) download.file(   \"https://www.robots.ox.ac.uk/~vgg/data/pets/data/images.tar.gz\",   \"datasets/images.tar.gz\" ) download.file(   \"https://www.robots.ox.ac.uk/~vgg/data/pets/data/annotations.tar.gz\",   \"datasets/annotations.tar.gz\" )  untar(\"datasets/images.tar.gz\", exdir = \"datasets\") untar(\"datasets/annotations.tar.gz\", exdir = \"datasets\")"},{"path":"https://keras3.posit.co/dev/articles/examples/vision/oxford_pets_image_segmentation.html","id":"prepare-paths-of-input-images-and-target-segmentation-masks","dir":"Articles > Examples > Vision","previous_headings":"","what":"Prepare paths of input images and target segmentation masks","title":"Image segmentation with a U-Net-like architecture","text":"","code":"library(keras3) input_dir <- \"datasets/images/\" target_dir <- \"datasets/annotations/trimaps/\" img_size <- c(160, 160) num_classes <- 3 batch_size <- 32  input_img_paths <- fs::dir_ls(input_dir, glob = \"*.jpg\") |> sort() target_img_paths <- fs::dir_ls(target_dir, glob = \"*.png\") |> sort()  cat(\"Number of samples:\", length(input_img_paths), \"\\n\") ## Number of samples: 7390 for (i in 1:10) {   cat(input_img_paths[i], \"|\", target_img_paths[i], \"\\n\") } ## datasets/images/Abyssinian_1.jpg | datasets/annotations/trimaps/Abyssinian_1.png ## datasets/images/Abyssinian_10.jpg | datasets/annotations/trimaps/Abyssinian_10.png ## datasets/images/Abyssinian_100.jpg | datasets/annotations/trimaps/Abyssinian_100.png ## datasets/images/Abyssinian_101.jpg | datasets/annotations/trimaps/Abyssinian_101.png ## datasets/images/Abyssinian_102.jpg | datasets/annotations/trimaps/Abyssinian_102.png ## datasets/images/Abyssinian_103.jpg | datasets/annotations/trimaps/Abyssinian_103.png ## datasets/images/Abyssinian_104.jpg | datasets/annotations/trimaps/Abyssinian_104.png ## datasets/images/Abyssinian_105.jpg | datasets/annotations/trimaps/Abyssinian_105.png ## datasets/images/Abyssinian_106.jpg | datasets/annotations/trimaps/Abyssinian_106.png ## datasets/images/Abyssinian_107.jpg | datasets/annotations/trimaps/Abyssinian_107.png"},{"path":"https://keras3.posit.co/dev/articles/examples/vision/oxford_pets_image_segmentation.html","id":"what-does-one-input-image-and-corresponding-segmentation-mask-look-like","dir":"Articles > Examples > Vision","previous_headings":"","what":"What does one input image and corresponding segmentation mask look like?","title":"Image segmentation with a U-Net-like architecture","text":"","code":"# Display input image #10 input_img_paths[10] |>   jpeg::readJPEG() |>   as.raster() |>   plot() target_img_paths[10] |>   png::readPNG() |>   magrittr::multiply_by(255)|>   as.raster(max = 3) |>   plot()"},{"path":"https://keras3.posit.co/dev/articles/examples/vision/oxford_pets_image_segmentation.html","id":"prepare-dataset-to-load-vectorize-batches-of-data","dir":"Articles > Examples > Vision","previous_headings":"","what":"Prepare dataset to load & vectorize batches of data","title":"Image segmentation with a U-Net-like architecture","text":"","code":"library(tensorflow, exclude = c(\"shape\", \"set_random_seed\")) library(tfdatasets, exclude = \"shape\")   # Returns a tf_dataset get_dataset <- function(batch_size, img_size, input_img_paths, target_img_paths,                         max_dataset_len = NULL) {    img_size <- as.integer(img_size)    load_img_masks <- function(input_img_path, target_img_path) {     input_img <- input_img_path |>       tf$io$read_file() |>       tf$io$decode_jpeg(channels = 3) |>       tf$image$resize(img_size) |>       tf$image$convert_image_dtype(\"float32\")      target_img <- target_img_path |>       tf$io$read_file() |>       tf$io$decode_png(channels = 1) |>       tf$image$resize(img_size, method = \"nearest\") |>       tf$image$convert_image_dtype(\"uint8\")      # Ground truth labels are 1, 2, 3. Subtract one to make them 0, 1, 2:     target_img <- target_img - 1L      list(input_img, target_img)   }    if (!is.null(max_dataset_len)) {     input_img_paths <- input_img_paths[1:max_dataset_len]     target_img_paths <- target_img_paths[1:max_dataset_len]   }    list(input_img_paths, target_img_paths) |>     tensor_slices_dataset() |>     dataset_map(load_img_masks, num_parallel_calls = tf$data$AUTOTUNE)|>     dataset_batch(batch_size) }"},{"path":"https://keras3.posit.co/dev/articles/examples/vision/oxford_pets_image_segmentation.html","id":"prepare-u-net-xception-style-model","dir":"Articles > Examples > Vision","previous_headings":"","what":"Prepare U-Net Xception-style model","title":"Image segmentation with a U-Net-like architecture","text":"","code":"get_model <- function(img_size, num_classes) {    inputs <- keras_input(shape = c(img_size, 3))    ### [First half of the network: downsampling inputs] ###    # Entry block   x <- inputs |>     layer_conv_2d(filters = 32, kernel_size = 3, strides = 2, padding = \"same\") |>     layer_batch_normalization() |>     layer_activation(\"relu\")    previous_block_activation <- x  # Set aside residual    for (filters in c(64, 128, 256)) {     x <- x |>       layer_activation(\"relu\") |>       layer_separable_conv_2d(filters = filters, kernel_size = 3, padding = \"same\") |>       layer_batch_normalization() |>        layer_activation(\"relu\") |>       layer_separable_conv_2d(filters = filters, kernel_size = 3, padding = \"same\") |>       layer_batch_normalization() |>        layer_max_pooling_2d(pool_size = 3, strides = 2, padding = \"same\")      residual <- previous_block_activation |>       layer_conv_2d(filters = filters, kernel_size = 1, strides = 2, padding = \"same\")      x <- layer_add(x, residual)  # Add back residual     previous_block_activation <- x  # Set aside next residual   }    ### [Second half of the network: upsampling inputs] ###    for (filters in c(256, 128, 64, 32)) {     x <- x |>       layer_activation(\"relu\") |>       layer_conv_2d_transpose(filters = filters, kernel_size = 3, padding = \"same\") |>       layer_batch_normalization() |>        layer_activation(\"relu\") |>       layer_conv_2d_transpose(filters = filters, kernel_size = 3, padding = \"same\") |>       layer_batch_normalization() |>        layer_upsampling_2d(size = 2)      # Project residual     residual <- previous_block_activation |>       layer_upsampling_2d(size = 2) |>       layer_conv_2d(filters = filters, kernel_size = 1, padding = \"same\")      x <- layer_add(x, residual)     # Add back residual     previous_block_activation <- x  # Set aside next residual   }    # Add a per-pixel classification layer   outputs <- x |>     layer_conv_2d(num_classes, 3, activation = \"softmax\", padding = \"same\")    # Define the model   keras_model(inputs, outputs) }  # Build model model <- get_model(img_size, num_classes) summary(model) ## Model: \"functional\" ## ┏━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━┳━━━━━━━━━━━━━━━━┳━━━━━━━┓ ## ┃ Layer (type)      ┃ Output Shape    ┃   Param # ┃ Connected to   ┃ Trai… ┃ ## ┡━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━╇━━━━━━━━━━━━━━━━╇━━━━━━━┩ ## │ input_layer       │ (None, 160,     │         0 │ -              │   -   │ ## │ (InputLayer)      │ 160, 3)         │           │                │       │ ## ├───────────────────┼─────────────────┼───────────┼────────────────┼───────┤ ## │ conv2d (Conv2D)   │ (None, 80, 80,  │       896 │ input_layer[0… │   Y   │ ## │                   │ 32)             │           │                │       │ ## ├───────────────────┼─────────────────┼───────────┼────────────────┼───────┤ ## │ batch_normalizat… │ (None, 80, 80,  │       128 │ conv2d[0][0]   │   Y   │ ## │ (BatchNormalizat… │ 32)             │           │                │       │ ## ├───────────────────┼─────────────────┼───────────┼────────────────┼───────┤ ## │ activation        │ (None, 80, 80,  │         0 │ batch_normali… │   -   │ ## │ (Activation)      │ 32)             │           │                │       │ ## ├───────────────────┼─────────────────┼───────────┼────────────────┼───────┤ ## │ activation_1      │ (None, 80, 80,  │         0 │ activation[0]… │   -   │ ## │ (Activation)      │ 32)             │           │                │       │ ## ├───────────────────┼─────────────────┼───────────┼────────────────┼───────┤ ## │ separable_conv2d  │ (None, 80, 80,  │     2,400 │ activation_1[… │   Y   │ ## │ (SeparableConv2D) │ 64)             │           │                │       │ ## ├───────────────────┼─────────────────┼───────────┼────────────────┼───────┤ ## │ batch_normalizat… │ (None, 80, 80,  │       256 │ separable_con… │   Y   │ ## │ (BatchNormalizat… │ 64)             │           │                │       │ ## ├───────────────────┼─────────────────┼───────────┼────────────────┼───────┤ ## │ activation_2      │ (None, 80, 80,  │         0 │ batch_normali… │   -   │ ## │ (Activation)      │ 64)             │           │                │       │ ## ├───────────────────┼─────────────────┼───────────┼────────────────┼───────┤ ## │ separable_conv2d… │ (None, 80, 80,  │     4,736 │ activation_2[… │   Y   │ ## │ (SeparableConv2D) │ 64)             │           │                │       │ ## ├───────────────────┼─────────────────┼───────────┼────────────────┼───────┤ ## │ batch_normalizat… │ (None, 80, 80,  │       256 │ separable_con… │   Y   │ ## │ (BatchNormalizat… │ 64)             │           │                │       │ ## ├───────────────────┼─────────────────┼───────────┼────────────────┼───────┤ ## │ max_pooling2d     │ (None, 40, 40,  │         0 │ batch_normali… │   -   │ ## │ (MaxPooling2D)    │ 64)             │           │                │       │ ## ├───────────────────┼─────────────────┼───────────┼────────────────┼───────┤ ## │ conv2d_1 (Conv2D) │ (None, 40, 40,  │     2,112 │ activation[0]… │   Y   │ ## │                   │ 64)             │           │                │       │ ## ├───────────────────┼─────────────────┼───────────┼────────────────┼───────┤ ## │ add (Add)         │ (None, 40, 40,  │         0 │ max_pooling2d… │   -   │ ## │                   │ 64)             │           │ conv2d_1[0][0] │       │ ## ├───────────────────┼─────────────────┼───────────┼────────────────┼───────┤ ## │ activation_3      │ (None, 40, 40,  │         0 │ add[0][0]      │   -   │ ## │ (Activation)      │ 64)             │           │                │       │ ## ├───────────────────┼─────────────────┼───────────┼────────────────┼───────┤ ## │ separable_conv2d… │ (None, 40, 40,  │     8,896 │ activation_3[… │   Y   │ ## │ (SeparableConv2D) │ 128)            │           │                │       │ ## ├───────────────────┼─────────────────┼───────────┼────────────────┼───────┤ ## │ batch_normalizat… │ (None, 40, 40,  │       512 │ separable_con… │   Y   │ ## │ (BatchNormalizat… │ 128)            │           │                │       │ ## ├───────────────────┼─────────────────┼───────────┼────────────────┼───────┤ ## │ activation_4      │ (None, 40, 40,  │         0 │ batch_normali… │   -   │ ## │ (Activation)      │ 128)            │           │                │       │ ## ├───────────────────┼─────────────────┼───────────┼────────────────┼───────┤ ## │ separable_conv2d… │ (None, 40, 40,  │    17,664 │ activation_4[… │   Y   │ ## │ (SeparableConv2D) │ 128)            │           │                │       │ ## ├───────────────────┼─────────────────┼───────────┼────────────────┼───────┤ ## │ batch_normalizat… │ (None, 40, 40,  │       512 │ separable_con… │   Y   │ ## │ (BatchNormalizat… │ 128)            │           │                │       │ ## ├───────────────────┼─────────────────┼───────────┼────────────────┼───────┤ ## │ max_pooling2d_1   │ (None, 20, 20,  │         0 │ batch_normali… │   -   │ ## │ (MaxPooling2D)    │ 128)            │           │                │       │ ## ├───────────────────┼─────────────────┼───────────┼────────────────┼───────┤ ## │ conv2d_2 (Conv2D) │ (None, 20, 20,  │     8,320 │ add[0][0]      │   Y   │ ## │                   │ 128)            │           │                │       │ ## ├───────────────────┼─────────────────┼───────────┼────────────────┼───────┤ ## │ add_1 (Add)       │ (None, 20, 20,  │         0 │ max_pooling2d… │   -   │ ## │                   │ 128)            │           │ conv2d_2[0][0] │       │ ## ├───────────────────┼─────────────────┼───────────┼────────────────┼───────┤ ## │ activation_5      │ (None, 20, 20,  │         0 │ add_1[0][0]    │   -   │ ## │ (Activation)      │ 128)            │           │                │       │ ## ├───────────────────┼─────────────────┼───────────┼────────────────┼───────┤ ## │ separable_conv2d… │ (None, 20, 20,  │    34,176 │ activation_5[… │   Y   │ ## │ (SeparableConv2D) │ 256)            │           │                │       │ ## ├───────────────────┼─────────────────┼───────────┼────────────────┼───────┤ ## │ batch_normalizat… │ (None, 20, 20,  │     1,024 │ separable_con… │   Y   │ ## │ (BatchNormalizat… │ 256)            │           │                │       │ ## ├───────────────────┼─────────────────┼───────────┼────────────────┼───────┤ ## │ activation_6      │ (None, 20, 20,  │         0 │ batch_normali… │   -   │ ## │ (Activation)      │ 256)            │           │                │       │ ## ├───────────────────┼─────────────────┼───────────┼────────────────┼───────┤ ## │ separable_conv2d… │ (None, 20, 20,  │    68,096 │ activation_6[… │   Y   │ ## │ (SeparableConv2D) │ 256)            │           │                │       │ ## ├───────────────────┼─────────────────┼───────────┼────────────────┼───────┤ ## │ batch_normalizat… │ (None, 20, 20,  │     1,024 │ separable_con… │   Y   │ ## │ (BatchNormalizat… │ 256)            │           │                │       │ ## ├───────────────────┼─────────────────┼───────────┼────────────────┼───────┤ ## │ max_pooling2d_2   │ (None, 10, 10,  │         0 │ batch_normali… │   -   │ ## │ (MaxPooling2D)    │ 256)            │           │                │       │ ## ├───────────────────┼─────────────────┼───────────┼────────────────┼───────┤ ## │ conv2d_3 (Conv2D) │ (None, 10, 10,  │    33,024 │ add_1[0][0]    │   Y   │ ## │                   │ 256)            │           │                │       │ ## ├───────────────────┼─────────────────┼───────────┼────────────────┼───────┤ ## │ add_2 (Add)       │ (None, 10, 10,  │         0 │ max_pooling2d… │   -   │ ## │                   │ 256)            │           │ conv2d_3[0][0] │       │ ## ├───────────────────┼─────────────────┼───────────┼────────────────┼───────┤ ## │ activation_7      │ (None, 10, 10,  │         0 │ add_2[0][0]    │   -   │ ## │ (Activation)      │ 256)            │           │                │       │ ## ├───────────────────┼─────────────────┼───────────┼────────────────┼───────┤ ## │ conv2d_transpose  │ (None, 10, 10,  │   590,080 │ activation_7[… │   Y   │ ## │ (Conv2DTranspose) │ 256)            │           │                │       │ ## ├───────────────────┼─────────────────┼───────────┼────────────────┼───────┤ ## │ batch_normalizat… │ (None, 10, 10,  │     1,024 │ conv2d_transp… │   Y   │ ## │ (BatchNormalizat… │ 256)            │           │                │       │ ## ├───────────────────┼─────────────────┼───────────┼────────────────┼───────┤ ## │ activation_8      │ (None, 10, 10,  │         0 │ batch_normali… │   -   │ ## │ (Activation)      │ 256)            │           │                │       │ ## ├───────────────────┼─────────────────┼───────────┼────────────────┼───────┤ ## │ conv2d_transpose… │ (None, 10, 10,  │   590,080 │ activation_8[… │   Y   │ ## │ (Conv2DTranspose) │ 256)            │           │                │       │ ## ├───────────────────┼─────────────────┼───────────┼────────────────┼───────┤ ## │ batch_normalizat… │ (None, 10, 10,  │     1,024 │ conv2d_transp… │   Y   │ ## │ (BatchNormalizat… │ 256)            │           │                │       │ ## ├───────────────────┼─────────────────┼───────────┼────────────────┼───────┤ ## │ up_sampling2d_1   │ (None, 20, 20,  │         0 │ add_2[0][0]    │   -   │ ## │ (UpSampling2D)    │ 256)            │           │                │       │ ## ├───────────────────┼─────────────────┼───────────┼────────────────┼───────┤ ## │ up_sampling2d     │ (None, 20, 20,  │         0 │ batch_normali… │   -   │ ## │ (UpSampling2D)    │ 256)            │           │                │       │ ## ├───────────────────┼─────────────────┼───────────┼────────────────┼───────┤ ## │ conv2d_4 (Conv2D) │ (None, 20, 20,  │    65,792 │ up_sampling2d… │   Y   │ ## │                   │ 256)            │           │                │       │ ## ├───────────────────┼─────────────────┼───────────┼────────────────┼───────┤ ## │ add_3 (Add)       │ (None, 20, 20,  │         0 │ up_sampling2d… │   -   │ ## │                   │ 256)            │           │ conv2d_4[0][0] │       │ ## ├───────────────────┼─────────────────┼───────────┼────────────────┼───────┤ ## │ activation_9      │ (None, 20, 20,  │         0 │ add_3[0][0]    │   -   │ ## │ (Activation)      │ 256)            │           │                │       │ ## ├───────────────────┼─────────────────┼───────────┼────────────────┼───────┤ ## │ conv2d_transpose… │ (None, 20, 20,  │   295,040 │ activation_9[… │   Y   │ ## │ (Conv2DTranspose) │ 128)            │           │                │       │ ## ├───────────────────┼─────────────────┼───────────┼────────────────┼───────┤ ## │ batch_normalizat… │ (None, 20, 20,  │       512 │ conv2d_transp… │   Y   │ ## │ (BatchNormalizat… │ 128)            │           │                │       │ ## ├───────────────────┼─────────────────┼───────────┼────────────────┼───────┤ ## │ activation_10     │ (None, 20, 20,  │         0 │ batch_normali… │   -   │ ## │ (Activation)      │ 128)            │           │                │       │ ## ├───────────────────┼─────────────────┼───────────┼────────────────┼───────┤ ## │ conv2d_transpose… │ (None, 20, 20,  │   147,584 │ activation_10… │   Y   │ ## │ (Conv2DTranspose) │ 128)            │           │                │       │ ## ├───────────────────┼─────────────────┼───────────┼────────────────┼───────┤ ## │ batch_normalizat… │ (None, 20, 20,  │       512 │ conv2d_transp… │   Y   │ ## │ (BatchNormalizat… │ 128)            │           │                │       │ ## ├───────────────────┼─────────────────┼───────────┼────────────────┼───────┤ ## │ up_sampling2d_3   │ (None, 40, 40,  │         0 │ add_3[0][0]    │   -   │ ## │ (UpSampling2D)    │ 256)            │           │                │       │ ## ├───────────────────┼─────────────────┼───────────┼────────────────┼───────┤ ## │ up_sampling2d_2   │ (None, 40, 40,  │         0 │ batch_normali… │   -   │ ## │ (UpSampling2D)    │ 128)            │           │                │       │ ## ├───────────────────┼─────────────────┼───────────┼────────────────┼───────┤ ## │ conv2d_5 (Conv2D) │ (None, 40, 40,  │    32,896 │ up_sampling2d… │   Y   │ ## │                   │ 128)            │           │                │       │ ## ├───────────────────┼─────────────────┼───────────┼────────────────┼───────┤ ## │ add_4 (Add)       │ (None, 40, 40,  │         0 │ up_sampling2d… │   -   │ ## │                   │ 128)            │           │ conv2d_5[0][0] │       │ ## ├───────────────────┼─────────────────┼───────────┼────────────────┼───────┤ ## │ activation_11     │ (None, 40, 40,  │         0 │ add_4[0][0]    │   -   │ ## │ (Activation)      │ 128)            │           │                │       │ ## ├───────────────────┼─────────────────┼───────────┼────────────────┼───────┤ ## │ conv2d_transpose… │ (None, 40, 40,  │    73,792 │ activation_11… │   Y   │ ## │ (Conv2DTranspose) │ 64)             │           │                │       │ ## ├───────────────────┼─────────────────┼───────────┼────────────────┼───────┤ ## │ batch_normalizat… │ (None, 40, 40,  │       256 │ conv2d_transp… │   Y   │ ## │ (BatchNormalizat… │ 64)             │           │                │       │ ## ├───────────────────┼─────────────────┼───────────┼────────────────┼───────┤ ## │ activation_12     │ (None, 40, 40,  │         0 │ batch_normali… │   -   │ ## │ (Activation)      │ 64)             │           │                │       │ ## ├───────────────────┼─────────────────┼───────────┼────────────────┼───────┤ ## │ conv2d_transpose… │ (None, 40, 40,  │    36,928 │ activation_12… │   Y   │ ## │ (Conv2DTranspose) │ 64)             │           │                │       │ ## ├───────────────────┼─────────────────┼───────────┼────────────────┼───────┤ ## │ batch_normalizat… │ (None, 40, 40,  │       256 │ conv2d_transp… │   Y   │ ## │ (BatchNormalizat… │ 64)             │           │                │       │ ## ├───────────────────┼─────────────────┼───────────┼────────────────┼───────┤ ## │ up_sampling2d_5   │ (None, 80, 80,  │         0 │ add_4[0][0]    │   -   │ ## │ (UpSampling2D)    │ 128)            │           │                │       │ ## ├───────────────────┼─────────────────┼───────────┼────────────────┼───────┤ ## │ up_sampling2d_4   │ (None, 80, 80,  │         0 │ batch_normali… │   -   │ ## │ (UpSampling2D)    │ 64)             │           │                │       │ ## ├───────────────────┼─────────────────┼───────────┼────────────────┼───────┤ ## │ conv2d_6 (Conv2D) │ (None, 80, 80,  │     8,256 │ up_sampling2d… │   Y   │ ## │                   │ 64)             │           │                │       │ ## ├───────────────────┼─────────────────┼───────────┼────────────────┼───────┤ ## │ add_5 (Add)       │ (None, 80, 80,  │         0 │ up_sampling2d… │   -   │ ## │                   │ 64)             │           │ conv2d_6[0][0] │       │ ## ├───────────────────┼─────────────────┼───────────┼────────────────┼───────┤ ## │ activation_13     │ (None, 80, 80,  │         0 │ add_5[0][0]    │   -   │ ## │ (Activation)      │ 64)             │           │                │       │ ## ├───────────────────┼─────────────────┼───────────┼────────────────┼───────┤ ## │ conv2d_transpose… │ (None, 80, 80,  │    18,464 │ activation_13… │   Y   │ ## │ (Conv2DTranspose) │ 32)             │           │                │       │ ## ├───────────────────┼─────────────────┼───────────┼────────────────┼───────┤ ## │ batch_normalizat… │ (None, 80, 80,  │       128 │ conv2d_transp… │   Y   │ ## │ (BatchNormalizat… │ 32)             │           │                │       │ ## ├───────────────────┼─────────────────┼───────────┼────────────────┼───────┤ ## │ activation_14     │ (None, 80, 80,  │         0 │ batch_normali… │   -   │ ## │ (Activation)      │ 32)             │           │                │       │ ## ├───────────────────┼─────────────────┼───────────┼────────────────┼───────┤ ## │ conv2d_transpose… │ (None, 80, 80,  │     9,248 │ activation_14… │   Y   │ ## │ (Conv2DTranspose) │ 32)             │           │                │       │ ## ├───────────────────┼─────────────────┼───────────┼────────────────┼───────┤ ## │ batch_normalizat… │ (None, 80, 80,  │       128 │ conv2d_transp… │   Y   │ ## │ (BatchNormalizat… │ 32)             │           │                │       │ ## ├───────────────────┼─────────────────┼───────────┼────────────────┼───────┤ ## │ up_sampling2d_7   │ (None, 160,     │         0 │ add_5[0][0]    │   -   │ ## │ (UpSampling2D)    │ 160, 64)        │           │                │       │ ## ├───────────────────┼─────────────────┼───────────┼────────────────┼───────┤ ## │ up_sampling2d_6   │ (None, 160,     │         0 │ batch_normali… │   -   │ ## │ (UpSampling2D)    │ 160, 32)        │           │                │       │ ## ├───────────────────┼─────────────────┼───────────┼────────────────┼───────┤ ## │ conv2d_7 (Conv2D) │ (None, 160,     │     2,080 │ up_sampling2d… │   Y   │ ## │                   │ 160, 32)        │           │                │       │ ## ├───────────────────┼─────────────────┼───────────┼────────────────┼───────┤ ## │ add_6 (Add)       │ (None, 160,     │         0 │ up_sampling2d… │   -   │ ## │                   │ 160, 32)        │           │ conv2d_7[0][0] │       │ ## ├───────────────────┼─────────────────┼───────────┼────────────────┼───────┤ ## │ conv2d_8 (Conv2D) │ (None, 160,     │       867 │ add_6[0][0]    │   Y   │ ## │                   │ 160, 3)         │           │                │       │ ## └───────────────────┴─────────────────┴───────────┴────────────────┴───────┘ ##  Total params: 2,058,979 (7.85 MB) ##  Trainable params: 2,055,203 (7.84 MB) ##  Non-trainable params: 3,776 (14.75 KB)"},{"path":"https://keras3.posit.co/dev/articles/examples/vision/oxford_pets_image_segmentation.html","id":"set-aside-a-validation-split","dir":"Articles > Examples > Vision","previous_headings":"","what":"Set aside a validation split","title":"Image segmentation with a U-Net-like architecture","text":"","code":"# Split our img paths into a training and a validation set val_samples <- 1000 val_samples <- sample.int(length(input_img_paths), val_samples)  train_input_img_paths <- input_img_paths[-val_samples] train_target_img_paths <- target_img_paths[-val_samples]  val_input_img_paths <- input_img_paths[val_samples] val_target_img_paths <- target_img_paths[val_samples]  # Instantiate dataset for each split # Limit input files in `max_dataset_len` for faster epoch training time. # Remove the `max_dataset_len` arg when running with full dataset. train_dataset <- get_dataset(   batch_size,   img_size,   train_input_img_paths,   train_target_img_paths,   max_dataset_len = 1000 ) valid_dataset <- get_dataset(   batch_size, img_size, val_input_img_paths, val_target_img_paths )"},{"path":"https://keras3.posit.co/dev/articles/examples/vision/oxford_pets_image_segmentation.html","id":"train-the-model","dir":"Articles > Examples > Vision","previous_headings":"","what":"Train the model","title":"Image segmentation with a U-Net-like architecture","text":"","code":"# Configure the model for training. # We use the \"sparse\" version of categorical_crossentropy # because our target data is integers. model |> compile(   optimizer = optimizer_adam(1e-4),   loss = \"sparse_categorical_crossentropy\" )  callbacks <- list(   callback_model_checkpoint(     \"models/oxford_segmentation.keras\", save_best_only = TRUE   ) )  # Train the model, doing validation at the end of each epoch. epochs <- 50 model |> fit(     train_dataset,     epochs=epochs,     validation_data=valid_dataset,     callbacks=callbacks,     verbose=2 ) ## Epoch 1/50 ## 32/32 - 49s - 2s/step - loss: 1.4283 - val_loss: 1.5505 ## Epoch 2/50 ## 32/32 - 2s - 73ms/step - loss: 0.9223 - val_loss: 1.9896 ## Epoch 3/50 ## 32/32 - 2s - 72ms/step - loss: 0.7764 - val_loss: 2.5134 ## Epoch 4/50 ## 32/32 - 2s - 73ms/step - loss: 0.7200 - val_loss: 3.0166 ## Epoch 5/50 ## 32/32 - 2s - 74ms/step - loss: 0.6848 - val_loss: 3.2910 ## Epoch 6/50 ## 32/32 - 2s - 71ms/step - loss: 0.6557 - val_loss: 3.4529 ## Epoch 7/50 ## 32/32 - 2s - 74ms/step - loss: 0.6303 - val_loss: 3.5613 ## Epoch 8/50 ## 32/32 - 2s - 71ms/step - loss: 0.6083 - val_loss: 3.6520 ## Epoch 9/50 ## 32/32 - 2s - 68ms/step - loss: 0.5894 - val_loss: 3.7289 ## Epoch 10/50 ## 32/32 - 2s - 76ms/step - loss: 0.5726 - val_loss: 3.7945 ## Epoch 11/50 ## 32/32 - 2s - 75ms/step - loss: 0.5566 - val_loss: 3.8264 ## Epoch 12/50 ## 32/32 - 2s - 73ms/step - loss: 0.5406 - val_loss: 3.8009 ## Epoch 13/50 ## 32/32 - 2s - 73ms/step - loss: 0.5240 - val_loss: 3.7206 ## Epoch 14/50 ## 32/32 - 2s - 76ms/step - loss: 0.5060 - val_loss: 3.5849 ## Epoch 15/50 ## 32/32 - 2s - 76ms/step - loss: 0.4858 - val_loss: 3.4372 ## Epoch 16/50 ## 32/32 - 2s - 73ms/step - loss: 0.4636 - val_loss: 3.2371 ## Epoch 17/50 ## 32/32 - 2s - 78ms/step - loss: 0.4395 - val_loss: 3.0008 ## Epoch 18/50 ## 32/32 - 2s - 75ms/step - loss: 0.4140 - val_loss: 2.6982 ## Epoch 19/50 ## 32/32 - 2s - 72ms/step - loss: 0.3878 - val_loss: 2.3533 ## Epoch 20/50 ## 32/32 - 2s - 74ms/step - loss: 0.3623 - val_loss: 1.9573 ## Epoch 21/50 ## 32/32 - 2s - 73ms/step - loss: 0.3391 - val_loss: 1.6047 ## Epoch 22/50 ## 32/32 - 3s - 83ms/step - loss: 0.3202 - val_loss: 1.3140 ## Epoch 23/50 ## 32/32 - 2s - 77ms/step - loss: 0.3083 - val_loss: 1.0783 ## Epoch 24/50 ## 32/32 - 3s - 80ms/step - loss: 0.3097 - val_loss: 1.0180 ## Epoch 25/50 ## 32/32 - 2s - 77ms/step - loss: 0.3446 - val_loss: 0.9350 ## Epoch 26/50 ## 32/32 - 2s - 70ms/step - loss: 0.3603 - val_loss: 0.9866 ## Epoch 27/50 ## 32/32 - 3s - 82ms/step - loss: 0.3268 - val_loss: 0.8704 ## Epoch 28/50 ## 32/32 - 2s - 72ms/step - loss: 0.2880 - val_loss: 1.0167 ## Epoch 29/50 ## 32/32 - 2s - 75ms/step - loss: 0.2735 - val_loss: 1.1785 ## Epoch 30/50 ## 32/32 - 2s - 75ms/step - loss: 0.2686 - val_loss: 1.1817 ## Epoch 31/50 ## 32/32 - 2s - 74ms/step - loss: 0.2718 - val_loss: 1.1267 ## Epoch 32/50 ## 32/32 - 2s - 72ms/step - loss: 0.2867 - val_loss: 1.1271 ## Epoch 33/50 ## 32/32 - 2s - 75ms/step - loss: 0.3123 - val_loss: 1.2541 ## Epoch 34/50 ## 32/32 - 2s - 75ms/step - loss: 0.2997 - val_loss: 1.2340 ## Epoch 35/50 ## 32/32 - 2s - 74ms/step - loss: 0.2851 - val_loss: 1.0755 ## Epoch 36/50 ## 32/32 - 2s - 73ms/step - loss: 0.2781 - val_loss: 1.0022 ## Epoch 37/50 ## 32/32 - 2s - 75ms/step - loss: 0.2755 - val_loss: 1.5445 ## Epoch 38/50 ## 32/32 - 2s - 76ms/step - loss: 0.2629 - val_loss: 1.1026 ## Epoch 39/50 ## 32/32 - 2s - 75ms/step - loss: 0.2428 - val_loss: 1.2205 ## Epoch 40/50 ## 32/32 - 2s - 74ms/step - loss: 0.2409 - val_loss: 1.2005 ## Epoch 41/50 ## 32/32 - 2s - 74ms/step - loss: 0.2453 - val_loss: 1.2381 ## Epoch 42/50 ## 32/32 - 2s - 73ms/step - loss: 0.2363 - val_loss: 1.5562 ## Epoch 43/50 ## 32/32 - 2s - 75ms/step - loss: 0.2337 - val_loss: 1.3185 ## Epoch 44/50 ## 32/32 - 2s - 77ms/step - loss: 0.2222 - val_loss: 1.1867 ## Epoch 45/50 ## 32/32 - 2s - 77ms/step - loss: 0.2151 - val_loss: 1.0701 ## Epoch 46/50 ## 32/32 - 2s - 75ms/step - loss: 0.2099 - val_loss: 1.1109 ## Epoch 47/50 ## 32/32 - 2s - 74ms/step - loss: 0.2062 - val_loss: 1.0900 ## Epoch 48/50 ## 32/32 - 2s - 74ms/step - loss: 0.2038 - val_loss: 1.1194 ## Epoch 49/50 ## 32/32 - 2s - 71ms/step - loss: 0.1964 - val_loss: 1.0576 ## Epoch 50/50 ## 32/32 - 2s - 73ms/step - loss: 0.1871 - val_loss: 1.0851"},{"path":"https://keras3.posit.co/dev/articles/examples/vision/oxford_pets_image_segmentation.html","id":"visualize-predictions","dir":"Articles > Examples > Vision","previous_headings":"","what":"Visualize predictions","title":"Image segmentation with a U-Net-like architecture","text":"","code":"model <- load_model(\"models/oxford_segmentation.keras\") # Generate predictions for all images in the validation set val_dataset <- get_dataset(   batch_size, img_size, val_input_img_paths, val_target_img_paths ) val_preds <- predict(model, val_dataset) ## 32/32 - 4s - 117ms/step display_mask <- function(i) {   # Quick utility to display a model's prediction.   mask <- val_preds[i,,,] %>%     apply(c(1,2), which.max) %>%     array_reshape(dim = c(img_size, 1))   mask <- abind::abind(mask, mask, mask, along = 3)   plot(as.raster(mask, max = 3)) }  # Display results for validation image #10 i <- 10  par(mfrow = c(1, 3)) # Display input image input_img_paths[i] |>   jpeg::readJPEG() |>   as.raster() |>   plot()  # Display ground-truth target mask target_img_paths[i] |>   png::readPNG() |>   magrittr::multiply_by(255)|>   as.raster(max = 3) |>   plot()  # Display mask predicted by our model display_mask(i)  # Note that the model only sees inputs at 150x150."},{"path":"https://keras3.posit.co/dev/articles/functional_api.html","id":"setup","dir":"Articles","previous_headings":"","what":"Setup","title":"The Functional API","text":"","code":"library(keras3)"},{"path":"https://keras3.posit.co/dev/articles/functional_api.html","id":"introduction","dir":"Articles","previous_headings":"","what":"Introduction","title":"The Functional API","text":"Keras functional API way create models flexible sequential API. functional API can handle models non-linear topology, shared layers, even multiple inputs outputs. main idea deep learning model usually directed acyclic graph (DAG) layers. functional API way build graphs layers. Consider following model: basic graph three layers. build model using functional API, start creating input node: shape data set 784-dimensional vector. batch size always omitted since shape sample specified. , example, image input shape (32, 32, 3), use: inputs returned contains information shape dtype input data feed model. ’s shape: ’s dtype: create new node graph layers calling layer inputs object: “layer call” action like drawing arrow “inputs” layer created. ’re “passing” inputs dense layer, get x output. Let’s add layers graph layers: point, can create Model specifying inputs outputs graph layers: Let’s check model summary looks like: can also plot model graph:  , optionally, display input output shapes layer plotted graph:  figure code almost identical. code version, connection arrows replaced call operation. “graph layers” intuitive mental image deep learning model, functional API way create models closely mirrors .","code":"(input: 784-dimensional vectors)        ↧ [Dense (64 units, relu activation)]        ↧ [Dense (64 units, relu activation)]        ↧ [Dense (10 units, softmax activation)]        ↧ (output: logits of a probability distribution over 10 classes) inputs <- keras_input(shape = c(784)) # Just for demonstration purposes. img_inputs <- keras_input(shape = c(32, 32, 3)) shape(inputs) ## shape(NA, 784) inputs$dtype ## [1] \"float32\" dense <- layer_dense(units = 64, activation=\"relu\") x <- dense(inputs) outputs <- x |>   layer_dense(units = 64, activation = \"relu\") |>   layer_dense(units = 10) model <- keras_model(inputs = inputs, outputs = outputs, name = \"mnist_model\") summary(model) ## Model: \"mnist_model\" ## ┏━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━┓ ## ┃ Layer (type)                    ┃ Output Shape           ┃       Param # ┃ ## ┡━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━┩ ## │ input_layer (InputLayer)        │ (None, 784)            │             0 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ dense (Dense)                   │ (None, 64)             │        50,240 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ dense_1 (Dense)                 │ (None, 64)             │         4,160 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ dense_2 (Dense)                 │ (None, 10)             │           650 │ ## └─────────────────────────────────┴────────────────────────┴───────────────┘ ##  Total params: 55,050 (215.04 KB) ##  Trainable params: 55,050 (215.04 KB) ##  Non-trainable params: 0 (0.00 B) plot(model) plot(model, show_shapes = TRUE)"},{"path":"https://keras3.posit.co/dev/articles/functional_api.html","id":"training-evaluation-and-inference","dir":"Articles","previous_headings":"","what":"Training, evaluation, and inference","title":"The Functional API","text":"Training, evaluation, inference work exactly way models built using functional API Sequential models. Model class offers built-training loop (fit() method) built-evaluation loop (evaluate() method). Note can easily customize loops implement training routines beyond supervised learning (e.g. GANs). , load MNIST image data, reshape vectors, fit model data (monitoring performance validation split), evaluate model test data: reading, see training evaluation guide.","code":"c(c(x_train, y_train), c(x_test, y_test)) %<-% dataset_mnist()  x_train <- array_reshape(x_train, c(60000, 784)) / 255 x_test <- array_reshape(x_test, c(10000, 784)) / 255  model |> compile(   loss = loss_sparse_categorical_crossentropy(from_logits = TRUE),   optimizer = optimizer_rmsprop(),   metrics = \"accuracy\" )  history <- model |> fit(     x_train, y_train, batch_size = 64, epochs = 2, validation_split = 0.2 ) ## Epoch 1/2 ## 750/750 - 2s - 3ms/step - accuracy: 0.8980 - loss: 0.3540 - val_accuracy: 0.9444 - val_loss: 0.1898 ## Epoch 2/2 ## 750/750 - 1s - 1ms/step - accuracy: 0.9512 - loss: 0.1633 - val_accuracy: 0.9605 - val_loss: 0.1387 test_scores <- model |> evaluate(x_test, y_test, verbose=2) ## 313/313 - 1s - 2ms/step - accuracy: 0.9598 - loss: 0.1322 cat(\"Test loss:\", test_scores$loss, \"\\n\") cat(\"Test accuracy:\", test_scores$accuracy, \"\\n\") ## Test loss: 0.1321879 ## Test accuracy: 0.9598"},{"path":"https://keras3.posit.co/dev/articles/functional_api.html","id":"save-and-serialize","dir":"Articles","previous_headings":"","what":"Save and serialize","title":"The Functional API","text":"Saving model serialization work way models built using functional API Sequential models. standard way save functional model call model.save() save entire model single file. can later recreate model file, even code built model longer available. saved file includes : - model architecture - model weight values (learned training) - model training config, (passed compile()) - optimizer state, (restart training left ) details, read model serialization & saving guide.","code":"model |> save_model(\"my_model.keras\") rm(model) # Recreate the exact same model purely from the file: model <- load_model(\"my_model.keras\")"},{"path":"https://keras3.posit.co/dev/articles/functional_api.html","id":"use-the-same-graph-of-layers-to-define-multiple-models","dir":"Articles","previous_headings":"","what":"Use the same graph of layers to define multiple models","title":"The Functional API","text":"functional API, models created specifying inputs outputs graph layers. means single graph layers can used generate multiple models. example , use stack layers instantiate two models: encoder model turns image inputs 16-dimensional vectors, end--end autoencoder model training. , decoding architecture strictly symmetrical encoding architecture, output shape input shape (28, 28, 1). reverse conv_2d layer conv_2d_transpose layer, reverse max_pooling_2d layer upsampling_2d layer.","code":"encoder_input <- keras_input(shape = c(28, 28, 1), name=\"img\") encoder_output <- encoder_input |>   layer_conv_2d(16, 3, activation = \"relu\") |>   layer_conv_2d(32, 3, activation = \"relu\") |>   layer_max_pooling_2d(3) |>   layer_conv_2d(32, 3, activation = \"relu\") |>   layer_conv_2d(16, 3, activation = \"relu\") |>   layer_global_max_pooling_2d()  encoder <- keras_model(encoder_input, encoder_output, name=\"encoder\") summary(encoder) ## Model: \"encoder\" ## ┏━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━┓ ## ┃ Layer (type)                    ┃ Output Shape           ┃       Param # ┃ ## ┡━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━┩ ## │ img (InputLayer)                │ (None, 28, 28, 1)      │             0 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ conv2d (Conv2D)                 │ (None, 26, 26, 16)     │           160 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ conv2d_1 (Conv2D)               │ (None, 24, 24, 32)     │         4,640 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ max_pooling2d (MaxPooling2D)    │ (None, 8, 8, 32)       │             0 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ conv2d_2 (Conv2D)               │ (None, 6, 6, 32)       │         9,248 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ conv2d_3 (Conv2D)               │ (None, 4, 4, 16)       │         4,624 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ global_max_pooling2d            │ (None, 16)             │             0 │ ## │ (GlobalMaxPooling2D)            │                        │               │ ## └─────────────────────────────────┴────────────────────────┴───────────────┘ ##  Total params: 18,672 (72.94 KB) ##  Trainable params: 18,672 (72.94 KB) ##  Non-trainable params: 0 (0.00 B) decoder_output <- encoder_output |>   layer_reshape(c(4, 4, 1)) |>   layer_conv_2d_transpose(16, 3, activation = \"relu\") |>   layer_conv_2d_transpose(32, 3, activation = \"relu\") |>   layer_upsampling_2d(3) |>   layer_conv_2d_transpose(16, 3, activation = \"relu\") |>   layer_conv_2d_transpose(1, 3, activation = \"relu\")  autoencoder <- keras_model(encoder_input, decoder_output, name=\"autoencoder\") summary(autoencoder) ## Model: \"autoencoder\" ## ┏━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━┓ ## ┃ Layer (type)                    ┃ Output Shape           ┃       Param # ┃ ## ┡━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━┩ ## │ img (InputLayer)                │ (None, 28, 28, 1)      │             0 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ conv2d (Conv2D)                 │ (None, 26, 26, 16)     │           160 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ conv2d_1 (Conv2D)               │ (None, 24, 24, 32)     │         4,640 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ max_pooling2d (MaxPooling2D)    │ (None, 8, 8, 32)       │             0 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ conv2d_2 (Conv2D)               │ (None, 6, 6, 32)       │         9,248 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ conv2d_3 (Conv2D)               │ (None, 4, 4, 16)       │         4,624 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ global_max_pooling2d            │ (None, 16)             │             0 │ ## │ (GlobalMaxPooling2D)            │                        │               │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ reshape (Reshape)               │ (None, 4, 4, 1)        │             0 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ conv2d_transpose                │ (None, 6, 6, 16)       │           160 │ ## │ (Conv2DTranspose)               │                        │               │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ conv2d_transpose_1              │ (None, 8, 8, 32)       │         4,640 │ ## │ (Conv2DTranspose)               │                        │               │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ up_sampling2d (UpSampling2D)    │ (None, 24, 24, 32)     │             0 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ conv2d_transpose_2              │ (None, 26, 26, 16)     │         4,624 │ ## │ (Conv2DTranspose)               │                        │               │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ conv2d_transpose_3              │ (None, 28, 28, 1)      │           145 │ ## │ (Conv2DTranspose)               │                        │               │ ## └─────────────────────────────────┴────────────────────────┴───────────────┘ ##  Total params: 28,241 (110.32 KB) ##  Trainable params: 28,241 (110.32 KB) ##  Non-trainable params: 0 (0.00 B)"},{"path":"https://keras3.posit.co/dev/articles/functional_api.html","id":"all-models-are-callable-just-like-layers","dir":"Articles","previous_headings":"","what":"All models are callable, just like layers","title":"The Functional API","text":"can treat model layer invoking Input output another layer. calling model aren’t just reusing architecture model, ’re also reusing weights. see action, ’s different take autoencoder example creates encoder model, decoder model, chains two calls obtain autoencoder model: can see, model can nested: model can contain sub-models (since model just like layer). common use case model nesting ensembling. example, ’s ensemble set models single model averages predictions:","code":"encoder_input <- keras_input(shape = c(28, 28, 1), name=\"img\") encoder_output <- encoder_input |>   layer_conv_2d(16, 3, activation = \"relu\") |>   layer_conv_2d(32, 3, activation = \"relu\") |>   layer_max_pooling_2d(3) |>   layer_conv_2d(32, 3, activation = \"relu\") |>   layer_conv_2d(16, 3, activation = \"relu\") |>   layer_global_max_pooling_2d()  encoder <- keras_model(encoder_input, encoder_output, name=\"encoder\") summary(encoder) ## Model: \"encoder\" ## ┏━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━┓ ## ┃ Layer (type)                    ┃ Output Shape           ┃       Param # ┃ ## ┡━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━┩ ## │ img (InputLayer)                │ (None, 28, 28, 1)      │             0 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ conv2d_4 (Conv2D)               │ (None, 26, 26, 16)     │           160 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ conv2d_5 (Conv2D)               │ (None, 24, 24, 32)     │         4,640 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ max_pooling2d_1 (MaxPooling2D)  │ (None, 8, 8, 32)       │             0 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ conv2d_6 (Conv2D)               │ (None, 6, 6, 32)       │         9,248 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ conv2d_7 (Conv2D)               │ (None, 4, 4, 16)       │         4,624 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ global_max_pooling2d_1          │ (None, 16)             │             0 │ ## │ (GlobalMaxPooling2D)            │                        │               │ ## └─────────────────────────────────┴────────────────────────┴───────────────┘ ##  Total params: 18,672 (72.94 KB) ##  Trainable params: 18,672 (72.94 KB) ##  Non-trainable params: 0 (0.00 B) decoder_input <- keras_input(shape = c(16), name = \"encoded_img\") decoder_output <- decoder_input |>   layer_reshape(c(4, 4, 1)) |>   layer_conv_2d_transpose(16, 3, activation = \"relu\") |>   layer_conv_2d_transpose(32, 3, activation = \"relu\") |>   layer_upsampling_2d(3) |>   layer_conv_2d_transpose(16, 3, activation = \"relu\") |>   layer_conv_2d_transpose(1, 3, activation = \"relu\")  decoder <- keras_model(decoder_input, decoder_output, name = \"decoder\") summary(decoder) ## Model: \"decoder\" ## ┏━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━┓ ## ┃ Layer (type)                    ┃ Output Shape           ┃       Param # ┃ ## ┡━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━┩ ## │ encoded_img (InputLayer)        │ (None, 16)             │             0 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ reshape_1 (Reshape)             │ (None, 4, 4, 1)        │             0 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ conv2d_transpose_4              │ (None, 6, 6, 16)       │           160 │ ## │ (Conv2DTranspose)               │                        │               │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ conv2d_transpose_5              │ (None, 8, 8, 32)       │         4,640 │ ## │ (Conv2DTranspose)               │                        │               │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ up_sampling2d_1 (UpSampling2D)  │ (None, 24, 24, 32)     │             0 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ conv2d_transpose_6              │ (None, 26, 26, 16)     │         4,624 │ ## │ (Conv2DTranspose)               │                        │               │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ conv2d_transpose_7              │ (None, 28, 28, 1)      │           145 │ ## │ (Conv2DTranspose)               │                        │               │ ## └─────────────────────────────────┴────────────────────────┴───────────────┘ ##  Total params: 9,569 (37.38 KB) ##  Trainable params: 9,569 (37.38 KB) ##  Non-trainable params: 0 (0.00 B) autoencoder_input <- keras_input(shape = c(28, 28, 1), name = \"img\") encoded_img <- encoder(autoencoder_input) decoded_img <- decoder(encoded_img) autoencoder <- keras_model(autoencoder_input, decoded_img,                            name = \"autoencoder\") summary(autoencoder) ## Model: \"autoencoder\" ## ┏━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━┓ ## ┃ Layer (type)                    ┃ Output Shape           ┃       Param # ┃ ## ┡━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━┩ ## │ img (InputLayer)                │ (None, 28, 28, 1)      │             0 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ encoder (Functional)            │ (None, 16)             │        18,672 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ decoder (Functional)            │ (None, 28, 28, 1)      │         9,569 │ ## └─────────────────────────────────┴────────────────────────┴───────────────┘ ##  Total params: 28,241 (110.32 KB) ##  Trainable params: 28,241 (110.32 KB) ##  Non-trainable params: 0 (0.00 B) get_model <- function() {   inputs <- keras_input(shape = 128)   outputs <- inputs |> layer_dense(1)   keras_model(inputs, outputs) }  model1 <- get_model() model2 <- get_model() model3 <- get_model()  inputs <- keras_input(shape = 128) y1 <- model1(inputs) y2 <- model2(inputs) y3 <- model3(inputs) outputs <- layer_average(list(y1, y2, y3)) ensemble_model <- keras_model(inputs = inputs, outputs = outputs)"},{"path":[]},{"path":"https://keras3.posit.co/dev/articles/functional_api.html","id":"models-with-multiple-inputs-and-outputs","dir":"Articles","previous_headings":"Manipulate complex graph topologies","what":"Models with multiple inputs and outputs","title":"The Functional API","text":"functional API makes easy manipulate multiple inputs outputs. handled Sequential API. example, ’re building system ranking customer issue tickets priority routing correct department, model three inputs: title ticket (text input), text body ticket (text input), tags added user (categorical input) model two outputs: priority score 0 1 (scalar sigmoid output), department handle ticket (softmax output set departments). can build model lines functional API: Now plot model:  compiling model, can assign different losses output. can even assign different weights loss – modulate contribution total training loss. Since output layers different names, also specify losses loss weights corresponding layer names: Train model passing lists NumPy arrays inputs targets: calling fit Dataset object, yield either list lists like list(list(title_data, body_data, tags_data), list(priority_targets, dept_targets)) list named lists like list(list(title = title_data, body = body_data, tags = tags_data), list(priority = priority_targets, department = dept_targets)). detailed explanation, refer training evaluation guide.","code":"num_tags <- 12  # Number of unique issue tags num_words <- 10000  # Size of vocabulary obtained when preprocessing text data num_departments <- 4  # Number of departments for predictions  title_input <- # Variable-length sequence of ints   keras_input(shape(NA), name = \"title\") body_input <-  # Variable-length sequence of ints   keras_input(shape(NA), name = \"body\") tags_input <-  # Binary vectors of size `num_tags`   keras_input(shape = num_tags, name = \"tags\")  # Embed each word in the title into a 64-dimensional vector title_features <- layer_embedding(title_input, num_words, 64) # Embed each word in the text into a 64-dimensional vector body_features <- layer_embedding(body_input, num_words, 64)  # Reduce sequence of embedded words in the title # into a single 128-dimensional vector title_features <- layer_lstm(title_features, 128) # Reduce sequence of embedded words in the body # into a single 32-dimensional vector body_features <- layer_lstm(body_features, 32)  # Merge all available features into a single large vector via concatenation x <- layer_concatenate(title_features, body_features, tags_input)  # Stick a logistic regression for priority prediction on top of the features priority_pred <- layer_dense(x, 1, name = \"priority\")  # Stick a department classifier on top of the features department_pred <- layer_dense(x, num_departments, name = \"department\")  # Instantiate an end-to-end model predicting both priority and department model <- keras_model(   inputs = list(title_input, body_input, tags_input),   outputs = list(priority = priority_pred, department = department_pred) ) plot(model, show_shapes = TRUE) model |> compile(   optimizer = optimizer_rmsprop(1e-3),   loss = list(     loss_binary_crossentropy(from_logits = TRUE),     loss_categorical_crossentropy(from_logits = TRUE)   ),   loss_weights = c(1.0, 0.2) ) model |> compile(   optimizer = optimizer_rmsprop(1e-3),   loss = list(     priority = loss_binary_crossentropy(from_logits = TRUE),     department = loss_categorical_crossentropy(from_logits = TRUE)   ),   loss_weights = list(priority = 1.0, department = 0.2) ) # Dummy input data title_data <- random_integer(c(1280, 12), 0, num_words) body_data <- random_integer(c(1280, 100), 0, num_words) tags_data <- random_integer(c(1280, num_tags), 0, 2)  # Dummy target data priority_targets <- random_normal(c(1280, 1)) dept_targets <- random_integer(c(1280, num_departments), 0, 2)  model |> fit(   list(title = title_data, body = body_data, tags = tags_data),   list(priority = priority_targets, department = dept_targets),   epochs = 2,   batch_size = 32 ) ## Epoch 1/2 ## 40/40 - 3s - 64ms/step - department_loss: 2.8465 - loss: 0.7669 - priority_loss: 0.1976 ## Epoch 2/2 ## 40/40 - 0s - 5ms/step - department_loss: 2.8554 - loss: 0.7538 - priority_loss: 0.1828"},{"path":"https://keras3.posit.co/dev/articles/functional_api.html","id":"a-toy-resnet-model","dir":"Articles","previous_headings":"Manipulate complex graph topologies","what":"A toy ResNet model","title":"The Functional API","text":"addition models multiple inputs outputs, functional API makes easy manipulate non-linear connectivity topologies – models layers connected sequentially, Sequential API handle. common use case residual connections. Let’s build toy ResNet model CIFAR10 demonstrate : Plot model:  Now train model:","code":"inputs <- keras_input(shape = c(32, 32, 3), name = \"img\") block_1_output <- inputs |>   layer_conv_2d(32, kernel_size = 3, activation = \"relu\") |>   layer_conv_2d(64, kernel_size = 3, activation = \"relu\") |>   layer_max_pooling_2d(pool_size = 3)  block_2_output <- block_1_output |>   layer_conv_2d(32, kernel_size = 3, activation = \"relu\", padding = \"same\") |>   layer_conv_2d(64, kernel_size = 3, activation = \"relu\", padding = \"same\") |>   layer_add(block_1_output)  block_3_output <- block_2_output |>   layer_conv_2d(64, kernel_size = 3, activation = \"relu\", padding = \"same\") |>   layer_conv_2d(64, kernel_size = 3, activation = \"relu\", padding = \"same\") |>   layer_add(block_2_output)  outputs <- block_3_output |>   layer_conv_2d(64, 3, activation = \"relu\") |>   layer_global_average_pooling_2d() |>   layer_dense(256, activation = \"relu\") |>   layer_dropout(0.5) |>   layer_dense(10)  model <- keras_model(inputs, outputs, name = \"toy_resnet\") summary(model) ## Model: \"toy_resnet\" ## ┏━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━┓ ## ┃ Layer (type)        ┃ Output Shape      ┃    Param # ┃ Connected to      ┃ ## ┡━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━┩ ## │ img (InputLayer)    │ (None, 32, 32, 3) │          0 │ -                 │ ## ├─────────────────────┼───────────────────┼────────────┼───────────────────┤ ## │ conv2d_8 (Conv2D)   │ (None, 30, 30,    │        896 │ img[0][0]         │ ## │                     │ 32)               │            │                   │ ## ├─────────────────────┼───────────────────┼────────────┼───────────────────┤ ## │ conv2d_9 (Conv2D)   │ (None, 28, 28,    │     18,496 │ conv2d_8[0][0]    │ ## │                     │ 64)               │            │                   │ ## ├─────────────────────┼───────────────────┼────────────┼───────────────────┤ ## │ max_pooling2d_2     │ (None, 9, 9, 64)  │          0 │ conv2d_9[0][0]    │ ## │ (MaxPooling2D)      │                   │            │                   │ ## ├─────────────────────┼───────────────────┼────────────┼───────────────────┤ ## │ conv2d_10 (Conv2D)  │ (None, 9, 9, 32)  │     18,464 │ max_pooling2d_2[… │ ## ├─────────────────────┼───────────────────┼────────────┼───────────────────┤ ## │ conv2d_11 (Conv2D)  │ (None, 9, 9, 64)  │     18,496 │ conv2d_10[0][0]   │ ## ├─────────────────────┼───────────────────┼────────────┼───────────────────┤ ## │ add (Add)           │ (None, 9, 9, 64)  │          0 │ conv2d_11[0][0],  │ ## │                     │                   │            │ max_pooling2d_2[… │ ## ├─────────────────────┼───────────────────┼────────────┼───────────────────┤ ## │ conv2d_12 (Conv2D)  │ (None, 9, 9, 64)  │     36,928 │ add[0][0]         │ ## ├─────────────────────┼───────────────────┼────────────┼───────────────────┤ ## │ conv2d_13 (Conv2D)  │ (None, 9, 9, 64)  │     36,928 │ conv2d_12[0][0]   │ ## ├─────────────────────┼───────────────────┼────────────┼───────────────────┤ ## │ add_1 (Add)         │ (None, 9, 9, 64)  │          0 │ conv2d_13[0][0],  │ ## │                     │                   │            │ add[0][0]         │ ## ├─────────────────────┼───────────────────┼────────────┼───────────────────┤ ## │ conv2d_14 (Conv2D)  │ (None, 7, 7, 64)  │     36,928 │ add_1[0][0]       │ ## ├─────────────────────┼───────────────────┼────────────┼───────────────────┤ ## │ global_average_poo… │ (None, 64)        │          0 │ conv2d_14[0][0]   │ ## │ (GlobalAveragePool… │                   │            │                   │ ## ├─────────────────────┼───────────────────┼────────────┼───────────────────┤ ## │ dense_6 (Dense)     │ (None, 256)       │     16,640 │ global_average_p… │ ## ├─────────────────────┼───────────────────┼────────────┼───────────────────┤ ## │ dropout (Dropout)   │ (None, 256)       │          0 │ dense_6[0][0]     │ ## ├─────────────────────┼───────────────────┼────────────┼───────────────────┤ ## │ dense_7 (Dense)     │ (None, 10)        │      2,570 │ dropout[0][0]     │ ## └─────────────────────┴───────────────────┴────────────┴───────────────────┘ ##  Total params: 186,346 (727.91 KB) ##  Trainable params: 186,346 (727.91 KB) ##  Non-trainable params: 0 (0.00 B) plot(model, show_shapes = TRUE) c(c(x_train, y_train), c(x_test, y_test)) %<-% dataset_cifar10()  x_train <- x_train / 255.0 x_test <- x_test / 255.0  model |> compile(   optimizer = optimizer_rmsprop(1e-3),   loss = loss_sparse_categorical_crossentropy(from_logits = TRUE),   metrics = \"acc\" ) # We restrict the data to the first 1000 samples so as to limit the # guide render time. # Try to train on the entire dataset until convergence! model |> fit(   x_train[1:1000, , , ],   y_train[1:1000, ],   batch_size = 64,   epochs = 1,   validation_split = 0.2 ) ## 13/13 - 6s - 430ms/step - acc: 0.1250 - loss: 2.2998 - val_acc: 0.1250 - val_loss: 2.2939"},{"path":"https://keras3.posit.co/dev/articles/functional_api.html","id":"shared-layers","dir":"Articles","previous_headings":"","what":"Shared layers","title":"The Functional API","text":"Another good use functional API models use shared layers. Shared layers layer instances reused multiple times model – learn features correspond multiple paths graph--layers. Shared layers often used encode inputs similar spaces (say, two different pieces text feature similar vocabulary). enable sharing information across different inputs, make possible train model less data. given word seen one inputs, benefit processing inputs pass shared layer. share layer functional API, call layer instance multiple times. instance, ’s Embedding layer shared across two different text inputs:","code":"# Embedding for 1000 unique words mapped to 128-dimensional vectors shared_embedding <- layer_embedding(input_dim = 1000, output_dim = 128)  # Variable-length sequence of integers text_input_a <- keras_input(shape = shape(NA), dtype=\"int32\")  # Variable-length sequence of integers text_input_b <- keras_input(shape = shape(NA), dtype=\"int32\")  # Reuse the same layer to encode both inputs encoded_input_a <- shared_embedding(text_input_a) encoded_input_b <- shared_embedding(text_input_b)"},{"path":"https://keras3.posit.co/dev/articles/functional_api.html","id":"extract-and-reuse-nodes-in-the-graph-of-layers","dir":"Articles","previous_headings":"","what":"Extract and reuse nodes in the graph of layers","title":"The Functional API","text":"graph layers manipulating static data structure, can accessed inspected. able plot functional models images. also means can access activations intermediate layers (“nodes” graph) reuse elsewhere – useful something like feature extraction. Let’s look example. VGG19 model weights pretrained ImageNet: intermediate activations model, obtained querying graph data structure: Use features create new feature-extraction model returns values intermediate layer activations: comes handy tasks like neural style transfer, among things.","code":"vgg19 <- application_vgg19() features_list <- lapply(vgg19$layers, function(x) x$output) feat_extraction_model <- keras_model(inputs = vgg19$input,                                      outputs = features_list)  img <- random_normal(c(1, 224, 224, 3)) extracted_features <- feat_extraction_model(img)"},{"path":"https://keras3.posit.co/dev/articles/functional_api.html","id":"extend-the-api-using-custom-layers","dir":"Articles","previous_headings":"","what":"Extend the API using custom layers","title":"The Functional API","text":"keras includes wide range built-layers, example: Convolutional layers: conv_1d, conv_2d, conv_3d, conv_2d_transpose Pooling layers: max_pooling_1d, max_pooling_2d, max_pooling_3d, average_pooling_3d RNN layers: gru, lstm, conv_lstm_2d batch_normalization, dropout, embedding, etc. don’t find need, ’s easy extend API creating layers. layers subclass Layer class implement: call method, specifies computation done layer. build method, creates weights layer (just style convention since can create weights initialize, well). learn creating layers scratch, read custom layers models guide. following basic implementation layer_dense(): serialization support custom layer, define get_config() method returns constructor arguments layer instance: Optionally, implement class method from_config(cls, config) used recreating layer instance given config dictionary. default implementation from_config :","code":"custom_dense <- Layer(   classname = \"CustomDense\",   initialize = function(units = 32) {     super$initialize()     self$units <- as.integer(units)   },   build = function(input_shape) {     self$w <- self$add_weight(       shape = shape(input_shape[[2]], self$units),       initializer = \"random_normal\",       trainable = TRUE,     )     self$b <- self$add_weight(       shape = shape(self$units),             initializer=\"random_normal\",       trainable = TRUE         )   },   call = function(inputs) {     op_matmul(inputs, self$w) + self$b   } )  inputs <- keras_input(c(4)) outputs <- custom_dense(inputs, 10)  model <- keras_model(inputs, outputs) custom_dense <- Layer(   classname = \"CustomDense\",    initialize = function(units = 32, ...) {     super$initialize()     self$units <- as.integer(units)   },    build = function(input_shape) {     self$w <- self$add_weight(       shape = shape(input_shape[[2]], self$units),       initializer = \"random_normal\",       trainable = TRUE,     )     self$b <- self$add_weight(       shape = shape(self$units),             initializer=\"random_normal\",       trainable = TRUE         )   },    call = function(inputs) {     op_matmul(inputs, self$w) + self$b   },    get_config = function() {     list(units = self$units)   } )  inputs <- keras_input(c(4)) outputs <- custom_dense(inputs, 10)  model <- keras_model(inputs, outputs) config <- get_config(model)  new_model <-   from_config(config, custom_objects = list(CustomDense = custom_dense)) from_config <- function(cls, config) {   do.call(cls, config) }"},{"path":"https://keras3.posit.co/dev/articles/functional_api.html","id":"when-to-use-the-functional-api","dir":"Articles","previous_headings":"","what":"When to use the functional API","title":"The Functional API","text":"use Keras functional API create new model, just subclass Model class directly? general, functional API higher-level, easier safer, number features subclassed models support. However, model subclassing provides greater flexibility building models easily expressible directed acyclic graphs layers. example, implement Tree-RNN functional API subclass Model directly. -depth look differences functional API model subclassing, read Symbolic Imperative APIs TensorFlow 2.0?.","code":""},{"path":"https://keras3.posit.co/dev/articles/functional_api.html","id":"functional-api-strengths","dir":"Articles","previous_headings":"When to use the functional API","what":"Functional API strengths:","title":"The Functional API","text":"following properties also true Sequential models (also data structures), true subclassed models (R Python (byte)code, data structures).","code":""},{"path":"https://keras3.posit.co/dev/articles/functional_api.html","id":"less-verbose","dir":"Articles","previous_headings":"When to use the functional API > Functional API strengths:","what":"Less verbose","title":"The Functional API","text":"super$initialize(...), call = function(...), self$..., etc. Compare: subclassed version:","code":"inputs <- keras_input(shape = shape(32)) outputs <- inputs |>   layer_dense(64, activation = \"relu\") |>   layer_dense(10) mlp <- keras_model(inputs, outputs) MLP <- Model(   classname = \"MLP\",   initialize = function(...) {     super$initialize(...)     self$dense_1 <- layer_dense(units = 64, activation = \"relu\")     self$dense_2 <- layer_dense(units = 10)   },   call = function(inputs) {     inputs |>       self$dense_1() |>       self$dense_2()   } )  # Instantiate the model. mlp <- MLP() # Necessary to create the model's state. # The model doesn't have a state until it's called at least once. out <- mlp(op_zeros(c(1, 32)))"},{"path":"https://keras3.posit.co/dev/articles/functional_api.html","id":"model-validation-while-defining-its-connectivity-graph","dir":"Articles","previous_headings":"When to use the functional API > Functional API strengths:","what":"Model validation while defining its connectivity graph","title":"The Functional API","text":"functional API, input specification (shape dtype) created advance (using Input). Every time call layer, layer checks specification passed matches assumptions, raise helpful error message . guarantees model can build functional API run. debugging – convergence-related debugging – happens statically model construction execution time. similar type checking compiler.","code":""},{"path":"https://keras3.posit.co/dev/articles/functional_api.html","id":"a-functional-model-is-plottable-and-inspectable","dir":"Articles","previous_headings":"When to use the functional API > Functional API strengths:","what":"A functional model is plottable and inspectable","title":"The Functional API","text":"can plot model graph, can easily access intermediate nodes graph. example, extract reuse activations intermediate layers (seen previous example):","code":"features_list <- lapply(vgg19$layers, function(x) x$output) feat_extraction_model <- keras_model(inputs = vgg19$input,                                      outputs = features_list)"},{"path":"https://keras3.posit.co/dev/articles/functional_api.html","id":"a-functional-model-can-be-serialized-or-cloned","dir":"Articles","previous_headings":"When to use the functional API > Functional API strengths:","what":"A functional model can be serialized or cloned","title":"The Functional API","text":"functional model data structure rather piece code, safely serializable can saved single file allows recreate exact model without access original code. See serialization & saving guide. serialize subclassed model, necessary implementer specify get_config() from_config() method model level.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/articles/functional_api.html","id":"it-does-not-support-dynamic-architectures","dir":"Articles","previous_headings":"When to use the functional API > Functional API weakness:","what":"It does not support dynamic architectures","title":"The Functional API","text":"functional API treats models DAGs layers. true deep learning architectures, – example, recursive networks Tree RNNs follow assumption implemented functional API.","code":""},{"path":"https://keras3.posit.co/dev/articles/functional_api.html","id":"mix-and-match-api-styles","dir":"Articles","previous_headings":"","what":"Mix-and-match API styles","title":"The Functional API","text":"Choosing functional API Model subclassing isn’t binary decision restricts one category models. models keras API can interact , whether ’re Sequential models, functional models, subclassed models written scratch. can always use functional model Sequential model part subclassed model layer: can use subclassed layer model functional API long implements call method follows one following patterns: call(inputs, ...) – inputs tensor nested structure tensors (e.g. list tensors), ... non-tensor arguments (non-inputs). call(inputs, training = NULL, ...) – training boolean indicating whether layer behave training mode inference mode. call(inputs, mask = NULL, ...) – mask boolean mask tensor (useful RNNs, instance). call(inputs, training = NULL, mask = NULL, ...) – course, can masking training-specific behavior time. Additionally, implement get_config() method custom Layer model, functional models create still serializable cloneable. ’s quick example custom RNN, written scratch, used functional model:","code":"units <- 32 timesteps <- 10 input_dim <- 5  # Define a Functional model inputs <- keras_input(shape(NA, units)) outputs <- inputs |>   layer_global_average_pooling_1d() |>   layer_dense(units = 1)  model <- keras_model(inputs, outputs)  layer_custom_rnn <- Layer(   classname = \"CustomRNN\",   initialize = function(...) {     super$initialize(...)     self$units <- units     self$projection_1 <- layer_dense(units = units, activation = \"tanh\")     self$projection_2 <- layer_dense(units = units, activation = \"tanh\")     self$classifier <- model   },   call = function(inputs, ...) {     outputs <- list()     state <- op_zeros(c(shape(inputs)[[1]], self$units))     for (t in 1:(shape(inputs)[[2]])) {       x <- inputs[, t, ]       h <- self$projection_1(x)       y <- h + self$projection_2(state)       state <- y       outputs[[t]] <- y     }     features <- op_stack(outputs, axis = 2)     self$classifier(features)   } )  rnn <- layer_custom_rnn() out <- rnn(op_zeros(c(1, timesteps, input_dim))) units <- 32 timesteps <- 10 input_dim <- 5 batch_size <- 16  layer_custom_rnn <- Layer(   \"custom_rnn\",   initialize = function(...) {     super$initialize(...)     self$units <- units     self$projection_1 <- layer_dense(units = units, activation = \"tanh\")     self$projection_2 <- layer_dense(units = units, activation = \"tanh\")     self$classifier <- layer_dense(units = 1)   },   call = function(inputs, ...) {     outputs <- list()     state <- op_zeros(c(shape(inputs)[[1]], self$units))     for (t in 1:(shape(inputs)[[2]])) {       x <- inputs[, t, ]       h <- self$projection_1(x)       y <- h + self$projection_2(state)       state <- y       outputs[[t]] <- y     }     features <- op_stack(outputs, axis = 2)     self$classifier(features)   } )  # Note that you specify a static batch size for the inputs with the `batch_shape` # arg, because the inner computation of `layer_custom_rnn()` requires a static batch size # (when you create the `state` zeros tensor). inputs <- keras_input(batch_shape = shape(batch_size, timesteps, input_dim)) outputs <- inputs |>   layer_conv_1d(filters = 32, kernel_size = 3) |>   layer_custom_rnn()  model <- keras_model(inputs, outputs) out <- model(op_zeros(c(1, 10, 5)))"},{"path":"https://keras3.posit.co/dev/articles/getting_started.html","id":"overview","dir":"Articles","previous_headings":"","what":"Overview","title":"Getting Started with Keras","text":"Keras high-level neural networks API developed focus enabling fast experimentation. able go idea result least possible delay key good research. Keras following key features: Allows code run CPU GPU, seamlessly. User-friendly API makes easy quickly prototype deep learning models. Built-support convolutional networks (computer vision), recurrent networks (sequence processing), combination . Supports arbitrary network architectures: multi-input multi-output models, layer sharing, model sharing, etc. means Keras appropriate building essentially deep learning model, memory network neural Turing machine. website provides documentation R interface Keras. See main Keras website https://keras.io additional information project.","code":""},{"path":"https://keras3.posit.co/dev/articles/getting_started.html","id":"installation","dir":"Articles","previous_headings":"","what":"Installation","title":"Getting Started with Keras","text":"First, install keras R package: install development version : Keras R interface requires backend engine installed. TensorFlow default. provide default installation Keras TensorFlow GPU capable, GPU available. want customized installation, e.g. see documentation install_keras() installation section.","code":"install.packages(\"keras3\") remotes::install_github(\"rstudio/keras3\") keras3::install_keras(backend = \"tensorflow\")"},{"path":"https://keras3.posit.co/dev/articles/getting_started.html","id":"mnist-example","dir":"Articles","previous_headings":"","what":"MNIST Example","title":"Getting Started with Keras","text":"can learn basics Keras walking simple example: recognizing handwritten digits MNIST dataset. MNIST consists 28 x 28 grayscale images handwritten digits like :  dataset also includes labels image, telling us digit . example, labels images 5, 0, 4, 1.","code":""},{"path":"https://keras3.posit.co/dev/articles/getting_started.html","id":"preparing-the-data","dir":"Articles","previous_headings":"MNIST Example","what":"Preparing the Data","title":"Getting Started with Keras","text":"MNIST dataset included Keras can accessed using dataset_mnist() function. load dataset create variables test training data: x data 3-d array (images, width, height) grayscale values. prepare data training convert 3-d arrays matrices reshaping width height single dimension (28x28 images flattened length 784 vectors). , convert grayscale values integers ranging 0 255 floating point values ranging 0 1: Note use array_reshape() function rather dim<-() function reshape array. data re-interpreted using row-major semantics (opposed R’s default column-major semantics), turn compatible way numerical libraries called Keras interpret array dimensions. y data integer vector values ranging 0 9. prepare data training one-hot encode vectors binary class matrices using Keras to_categorical() function:","code":"library(keras3) mnist <- dataset_mnist() x_train <- mnist$train$x y_train <- mnist$train$y x_test <- mnist$test$x y_test <- mnist$test$y # reshape x_train <- array_reshape(x_train, c(nrow(x_train), 784)) x_test <- array_reshape(x_test, c(nrow(x_test), 784)) # rescale x_train <- x_train / 255 x_test <- x_test / 255 y_train <- to_categorical(y_train, 10) y_test <- to_categorical(y_test, 10)"},{"path":"https://keras3.posit.co/dev/articles/getting_started.html","id":"defining-the-model","dir":"Articles","previous_headings":"MNIST Example","what":"Defining the Model","title":"Getting Started with Keras","text":"core data structure Keras model, way organize layers. simplest type model Sequential model, linear stack layers. begin creating sequential model adding layers using pipe (|>) operator: input_shape argument first layer specifies shape input data (length 784 numeric vector representing grayscale image). final layer outputs length 10 numeric vector (probabilities digit) using softmax activation function. Use summary() function print details model: Next, compile model appropriate loss function, optimizer, metrics:","code":"model <- keras_model_sequential(input_shape = c(784)) model |>   layer_dense(units = 256, activation = 'relu') |>   layer_dropout(rate = 0.4) |>   layer_dense(units = 128, activation = 'relu') |>   layer_dropout(rate = 0.3) |>   layer_dense(units = 10, activation = 'softmax') summary(model) ## Model: \"sequential\" ## ┏━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━┓ ## ┃ Layer (type)                    ┃ Output Shape           ┃       Param # ┃ ## ┡━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━┩ ## │ dense (Dense)                   │ (None, 256)            │       200,960 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ dropout (Dropout)               │ (None, 256)            │             0 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ dense_1 (Dense)                 │ (None, 128)            │        32,896 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ dropout_1 (Dropout)             │ (None, 128)            │             0 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ dense_2 (Dense)                 │ (None, 10)             │         1,290 │ ## └─────────────────────────────────┴────────────────────────┴───────────────┘ ##  Total params: 235,146 (918.54 KB) ##  Trainable params: 235,146 (918.54 KB) ##  Non-trainable params: 0 (0.00 B) plot(model) model |> compile(   loss = 'categorical_crossentropy',   optimizer = optimizer_rmsprop(),   metrics = c('accuracy') )"},{"path":"https://keras3.posit.co/dev/articles/getting_started.html","id":"training-and-evaluation","dir":"Articles","previous_headings":"MNIST Example","what":"Training and Evaluation","title":"Getting Started with Keras","text":"Use fit() function train model 30 epochs using batches 128 images: history object returned fit() includes loss accuracy metrics can plot: Evaluate model’s performance test data: Generate predictions new data: Keras provides vocabulary building deep learning models simple, elegant, intuitive. Building question answering system, image classification model, neural Turing machine, model just straightforward.","code":"history <- model |> fit(   x_train, y_train,   epochs = 30, batch_size = 128,   validation_split = 0.2 ) plot(history) model |> evaluate(x_test, y_test) ## 313/313 - 1s - 2ms/step - accuracy: 0.9810 - loss: 0.0842 ## $accuracy ## [1] 0.981 ## ## $loss ## [1] 0.08419506 probs <- model |> predict(x_test) ## 313/313 - 0s - 1ms/step max.col(probs) - 1L ##   [1] 7 2 1 0 4 1 4 9 5 9 0 6 9 0 1 5 9 7 3 4 9 6 6 5 4 0 7 4 0 1 3 1 3 4 7 ##  [36] 2 7 1 2 1 1 7 4 2 3 5 1 2 4 4 6 3 5 5 6 0 4 1 9 5 7 8 9 3 7 4 6 4 3 0 ##  [71] 7 0 2 9 1 7 3 2 9 7 7 6 2 7 8 4 7 3 6 1 3 6 9 3 1 4 1 7 6 9 ##  [ reached 'max' / getOption(\"max.print\") -- omitted 9900 entries ]"},{"path":"https://keras3.posit.co/dev/articles/getting_started.html","id":"deep-learning-with-r-book","dir":"Articles","previous_headings":"MNIST Example","what":"Deep Learning with R Book","title":"Getting Started with Keras","text":"want comprehensive introduction Keras concepts practice deep learning, recommend Deep Learning R, 2nd Edition book Manning. book collaboration François Chollet, creator (Python) Keras, J.J. Allaire, wrote original R interface Keras, Tomasz Kalinowski, maintainer R interface Keras. book presumes significant knowledge machine learning deep learning, goes way basic theory advanced practical applications, using R interface Keras.","code":""},{"path":"https://keras3.posit.co/dev/articles/getting_started.html","id":"why-this-name-keras","dir":"Articles","previous_headings":"","what":"Why this name, Keras?","title":"Getting Started with Keras","text":"Keras (κέρας) means horn Greek. reference literary image ancient Greek Latin literature, first found Odyssey, dream spirits (Oneiroi, singular Oneiros) divided deceive men false visions, arrive Earth gate ivory, announce future come pass, arrive gate horn. ’s play words κέρας (horn) / κραίνω (fulfill), ἐλέφας (ivory) / ἐλεφαίρομαι (deceive). Keras initially developed part research effort project ONEIROS (Open-ended Neuro-Electronic Intelligent Robot Operating System). “Oneiroi beyond unravelling –can sure tale tell? men look comes pass. Two gates give passage fleeting Oneiroi; one made horn, one ivory. Oneiroi pass sawn ivory deceitful, bearing message fulfilled; come polished horn truth behind , accomplished men see .” Homer, Odyssey 19. 562 ff (Shewring translation).","code":""},{"path":"https://keras3.posit.co/dev/articles/intro_to_keras_for_engineers.html","id":"introduction","dir":"Articles","previous_headings":"","what":"Introduction","title":"Introduction to Keras for engineers","text":"Keras 3 deep learning framework works TensorFlow, JAX, PyTorch interchangeably. notebook walk key Keras 3 workflows. Let’s start installing Keras 3:","code":"install.packages(\"keras3\") keras3::install_keras()"},{"path":"https://keras3.posit.co/dev/articles/intro_to_keras_for_engineers.html","id":"setup","dir":"Articles","previous_headings":"","what":"Setup","title":"Introduction to Keras for engineers","text":"’re going using tensorflow backend – can edit string \"jax\" \"torch\" hit “Restart runtime”, whole notebook run just ! entire guide backend-agnostic.","code":"library(tensorflow, exclude = c(\"shape\", \"set_random_seed\")) library(keras3)  # Note that you must configure the backend # before calling any other keras functions. # The backend cannot be changed once the # package is imported. use_backend(\"tensorflow\")"},{"path":"https://keras3.posit.co/dev/articles/intro_to_keras_for_engineers.html","id":"a-first-example-a-mnist-convnet","dir":"Articles","previous_headings":"","what":"A first example: A MNIST convnet","title":"Introduction to Keras for engineers","text":"Let’s start Hello World ML: training convnet classify MNIST digits. ’s data: ’s model. Different model-building options Keras offers include: Sequential API (use ) Functional API (typical) Writing models via subclassing (advanced use cases) ’s model summary: use compile() method specify optimizer, loss function, metrics monitor. Note JAX TensorFlow backends, XLA compilation turned default. Let’s train evaluate model. ’ll set aside validation split 15% data training monitor generalization unseen data. training, saving model end epoch. can also save model latest state like : reload like : Next, can query predictions class probabilities predict(): ’s basics!","code":"# Load the data and split it between train and test sets c(c(x_train, y_train), c(x_test, y_test)) %<-% keras3::dataset_mnist()  # Scale images to the [0, 1] range x_train <- x_train / 255 x_test <- x_test / 255  # Make sure images have shape (28, 28, 1) x_train <- array_reshape(x_train, c(-1, 28, 28, 1)) x_test <- array_reshape(x_test, c(-1, 28, 28, 1))  dim(x_train) ## [1] 60000    28    28     1 dim(x_test) ## [1] 10000    28    28     1 # Model parameters num_classes <- 10 input_shape <- c(28, 28, 1)  model <- keras_model_sequential(input_shape = input_shape) model |>   layer_conv_2d(filters = 64, kernel_size = c(3, 3), activation = \"relu\") |>   layer_conv_2d(filters = 64, kernel_size = c(3, 3), activation = \"relu\") |>   layer_max_pooling_2d(pool_size = c(2, 2)) |>   layer_conv_2d(filters = 128, kernel_size = c(3, 3), activation = \"relu\") |>   layer_conv_2d(filters = 128, kernel_size = c(3, 3), activation = \"relu\") |>   layer_global_average_pooling_2d() |>   layer_dropout(rate = 0.5) |>   layer_dense(units = num_classes, activation = \"softmax\") summary(model) ## Model: \"sequential\" ## ┏━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━┓ ## ┃ Layer (type)                    ┃ Output Shape           ┃       Param # ┃ ## ┡━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━┩ ## │ conv2d (Conv2D)                 │ (None, 26, 26, 64)     │           640 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ conv2d_1 (Conv2D)               │ (None, 24, 24, 64)     │        36,928 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ max_pooling2d (MaxPooling2D)    │ (None, 12, 12, 64)     │             0 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ conv2d_2 (Conv2D)               │ (None, 10, 10, 128)    │        73,856 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ conv2d_3 (Conv2D)               │ (None, 8, 8, 128)      │       147,584 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ global_average_pooling2d        │ (None, 128)            │             0 │ ## │ (GlobalAveragePooling2D)        │                        │               │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ dropout (Dropout)               │ (None, 128)            │             0 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ dense (Dense)                   │ (None, 10)             │         1,290 │ ## └─────────────────────────────────┴────────────────────────┴───────────────┘ ##  Total params: 260,298 (1016.79 KB) ##  Trainable params: 260,298 (1016.79 KB) ##  Non-trainable params: 0 (0.00 B) model |> compile(   optimizer = \"adam\",   loss = \"sparse_categorical_crossentropy\",   metrics = list(     metric_sparse_categorical_accuracy(name = \"acc\")   ) ) batch_size <- 128 epochs <- 10  callbacks <- list(   callback_model_checkpoint(filepath=\"model_at_epoch_{epoch}.keras\"),   callback_early_stopping(monitor=\"val_loss\", patience=2) )  model |> fit(   x_train, y_train,   batch_size = batch_size,   epochs = epochs,   validation_split = 0.15,   callbacks = callbacks ) ## Epoch 1/10 ## 399/399 - 7s - 17ms/step - acc: 0.7464 - loss: 0.7476 - val_acc: 0.9627 - val_loss: 0.1272 ## Epoch 2/10 ## 399/399 - 2s - 5ms/step - acc: 0.9376 - loss: 0.2072 - val_acc: 0.9771 - val_loss: 0.0776 ## Epoch 3/10 ## 399/399 - 2s - 5ms/step - acc: 0.9571 - loss: 0.1475 - val_acc: 0.9816 - val_loss: 0.0633 ## Epoch 4/10 ## 399/399 - 2s - 5ms/step - acc: 0.9652 - loss: 0.1175 - val_acc: 0.9863 - val_loss: 0.0488 ## Epoch 5/10 ## 399/399 - 2s - 5ms/step - acc: 0.9713 - loss: 0.1004 - val_acc: 0.9870 - val_loss: 0.0469 ## Epoch 6/10 ## 399/399 - 2s - 5ms/step - acc: 0.9754 - loss: 0.0854 - val_acc: 0.9879 - val_loss: 0.0407 ## Epoch 7/10 ## 399/399 - 2s - 5ms/step - acc: 0.9767 - loss: 0.0788 - val_acc: 0.9899 - val_loss: 0.0378 ## Epoch 8/10 ## 399/399 - 2s - 5ms/step - acc: 0.9795 - loss: 0.0688 - val_acc: 0.9867 - val_loss: 0.0438 ## Epoch 9/10 ## 399/399 - 2s - 5ms/step - acc: 0.9799 - loss: 0.0666 - val_acc: 0.9893 - val_loss: 0.0396 score <- model |> evaluate(x_test, y_test, verbose = 0) save_model(model, \"final_model.keras\", overwrite=TRUE) model <- load_model(\"final_model.keras\") predictions <- model |> predict(x_test) ## 313/313 - 1s - 2ms/step dim(predictions) ## [1] 10000    10"},{"path":"https://keras3.posit.co/dev/articles/intro_to_keras_for_engineers.html","id":"writing-cross-framework-custom-components","dir":"Articles","previous_headings":"","what":"Writing cross-framework custom components","title":"Introduction to Keras for engineers","text":"Keras enables write custom Layers, Models, Metrics, Losses, Optimizers work across TensorFlow, JAX, PyTorch codebase. Let’s take look custom layers first. op_ namespace contains: implementation NumPy API, e.g. op_stack op_matmul. set neural network specific ops absent NumPy, op_conv op_binary_crossentropy. Let’s make custom Dense layer works backends: Next, let’s make custom Dropout layer relies random_* namespace: Next, let’s write custom subclassed model uses two custom layers: Let’s compile fit :","code":"layer_my_dense <- Layer(   classname = \"MyDense\",   initialize = function(units, activation = NULL, name = NULL, ...) {     super$initialize(name = name, ...)     self$units <- units     self$activation <- activation   },   build = function(input_shape) {     input_dim <- tail(input_shape, 1)     self$w <- self$add_weight(       shape = shape(input_dim, self$units),       initializer = initializer_glorot_normal(),       name = \"kernel\",       trainable = TRUE     )     self$b <- self$add_weight(       shape = shape(self$units),       initializer = initializer_zeros(),       name = \"bias\",       trainable = TRUE     )   },   call = function(inputs) {     # Use Keras ops to create backend-agnostic layers/metrics/etc.     x <- op_matmul(inputs, self$w) + self$b     if (!is.null(self$activation))       x <- self$activation(x)     x   } ) layer_my_dropout <- Layer(   \"MyDropout\",   initialize = function(rate, name = NULL, seed = NULL, ...) {     super$initialize(name = name)     self$rate <- rate     # Use seed_generator for managing RNG state.     # It is a state element and its seed variable is     # tracked as part of `layer$variables`.     self$seed_generator <- random_seed_generator(seed)   },   call = function(inputs) {     # Use `keras3::random_*` for random ops.     random_dropout(inputs, self$rate, seed = self$seed_generator)   } ) MyModel <- Model(   \"MyModel\",   initialize = function(num_classes, ...) {     super$initialize(...)     self$conv_base <-       keras_model_sequential() |>       layer_conv_2d(64, kernel_size = c(3, 3), activation = \"relu\") |>       layer_conv_2d(64, kernel_size = c(3, 3), activation = \"relu\") |>       layer_max_pooling_2d(pool_size = c(2, 2)) |>       layer_conv_2d(128, kernel_size = c(3, 3), activation = \"relu\") |>       layer_conv_2d(128, kernel_size = c(3, 3), activation = \"relu\") |>       layer_global_average_pooling_2d()      self$dp <- layer_my_dropout(rate = 0.5)     self$dense <- layer_my_dense(units = num_classes,                                  activation = activation_softmax)   },   call = function(inputs) {     inputs |>       self$conv_base() |>       self$dp() |>       self$dense()   } ) model <- MyModel(num_classes = 10) model |> compile(   loss = loss_sparse_categorical_crossentropy(),   optimizer = optimizer_adam(learning_rate = 1e-3),   metrics = list(     metric_sparse_categorical_accuracy(name = \"acc\")   ) )  model |> fit(   x_train, y_train,   batch_size = batch_size,   epochs = 1, # For speed   validation_split = 0.15 ) ## 399/399 - 6s - 15ms/step - acc: 0.7355 - loss: 0.7721 - val_acc: 0.9287 - val_loss: 0.2384"},{"path":"https://keras3.posit.co/dev/articles/intro_to_keras_for_engineers.html","id":"training-models-on-arbitrary-data-sources","dir":"Articles","previous_headings":"","what":"Training models on arbitrary data sources","title":"Introduction to Keras for engineers","text":"Keras models can trained evaluated wide variety data sources, independently backend ’re using. includes: Arrays Dataframes TensorFlow tf_dataset objects PyTorch DataLoader objects Keras PyDataset objects work whether ’re using TensorFlow, JAX, PyTorch Keras backend. Let’s try tf_dataset:","code":"library(tfdatasets, exclude = \"shape\")  train_dataset <- list(x_train, y_train) |>   tensor_slices_dataset() |>   dataset_batch(batch_size) |>   dataset_prefetch(buffer_size = tf$data$AUTOTUNE)  test_dataset <- list(x_test, y_test) |>   tensor_slices_dataset() |>   dataset_batch(batch_size) |>   dataset_prefetch(buffer_size = tf$data$AUTOTUNE)  model <- MyModel(num_classes = 10) model |> compile(   loss = loss_sparse_categorical_crossentropy(),   optimizer = optimizer_adam(learning_rate = 1e-3),   metrics = list(     metric_sparse_categorical_accuracy(name = \"acc\")   ) )  model |> fit(train_dataset, epochs = 1, validation_data = test_dataset) ## 469/469 - 7s - 15ms/step - acc: 0.7502 - loss: 0.7455 - val_acc: 0.9107 - val_loss: 0.2982"},{"path":"https://keras3.posit.co/dev/articles/intro_to_keras_for_engineers.html","id":"further-reading","dir":"Articles","previous_headings":"","what":"Further reading","title":"Introduction to Keras for engineers","text":"concludes short overview new multi-backend capabilities Keras 3. Next, can learn :","code":""},{"path":"https://keras3.posit.co/dev/articles/intro_to_keras_for_engineers.html","id":"how-to-customize-what-happens-in-fit","dir":"Articles","previous_headings":"Further reading","what":"How to customize what happens in fit()","title":"Introduction to Keras for engineers","text":"Want implement non-standard training algorithm still want benefit power usability fit()? ’s easy customize fit() support arbitrary use cases: Customizing happens fit() TensorFlow","code":""},{"path":"https://keras3.posit.co/dev/articles/intro_to_keras_for_engineers.html","id":"how-to-write-custom-training-loops","dir":"Articles","previous_headings":"","what":"How to write custom training loops","title":"Introduction to Keras for engineers","text":"Writing training loop scratch TensorFlow","code":""},{"path":"https://keras3.posit.co/dev/articles/intro_to_keras_for_engineers.html","id":"how-to-distribute-training","dir":"Articles","previous_headings":"","what":"How to distribute training","title":"Introduction to Keras for engineers","text":"Guide distributed training TensorFlow Enjoy library! 🚀","code":""},{"path":"https://keras3.posit.co/dev/articles/intro_to_keras_for_researchers.html","id":"setup","dir":"Articles","previous_headings":"","what":"Setup","title":"Introduction to Keras for Researchers","text":"","code":"library(keras3) library(tensorflow)"},{"path":"https://keras3.posit.co/dev/articles/intro_to_keras_for_researchers.html","id":"introduction","dir":"Articles","previous_headings":"","what":"Introduction","title":"Introduction to Keras for Researchers","text":"machine learning researcher? publish NeurIPS push state---art CV NLP? guide serve first introduction core Keras & TensorFlow API concepts. guide, learn : Tensors, variables, gradients TensorFlow Creating layers subclassing [Layer] class Writing low-level training loops Tracking losses created layers via add_loss() method Tracking metrics low-level training loop Speeding execution compiled [tensorflow::tf_function()] Executing layers training inference mode Keras Functional API also see Keras API action two end--end research examples: Variational Autoencoder, Hypernetwork.","code":""},{"path":"https://keras3.posit.co/dev/articles/intro_to_keras_for_researchers.html","id":"tensors","dir":"Articles","previous_headings":"","what":"Tensors","title":"Introduction to Keras for Researchers","text":"TensorFlow infrastructure layer differentiable programming. heart, ’s framework manipulating N-dimensional arrays (tensors), much like NumPy. However, three key differences NumPy TensorFlow: TensorFlow can leverage hardware accelerators GPUs TPUs. TensorFlow can automatically compute gradient arbitrary differentiable tensor expressions. TensorFlow computation can distributed large numbers devices single machine, large number machines (potentially multiple devices ). Let’s take look object core TensorFlow: Tensor. ’s constant tensor: can get value R array calling .array(): features attributes dtype shape: common way create constant tensors via tf$ones tf$zeros: can also create random constant tensors:","code":"x <- tf$constant(rbind(c(5, 2), c(1, 3))) print(x) ## tf.Tensor( ## [[5. 2.] ##  [1. 3.]], shape=(2, 2), dtype=float64) as.array(x) ##      [,1] [,2] ## [1,]    5    2 ## [2,]    1    3 x$dtype ## tf.float64 x$shape ## TensorShape([2, 2]) tf$ones(shape = shape(2, 1)) ## tf.Tensor( ## [[1.] ##  [1.]], shape=(2, 1), dtype=float32) tf$zeros(shape = shape(2, 1)) ## tf.Tensor( ## [[0.] ##  [0.]], shape=(2, 1), dtype=float32) x <- random_normal(shape = c(2, 2), mean = 0.0, stddev = 1.0) x <- random_uniform(shape = c(2, 2), minval = 0, maxval = 10)"},{"path":"https://keras3.posit.co/dev/articles/intro_to_keras_for_researchers.html","id":"variables","dir":"Articles","previous_headings":"","what":"Variables","title":"Introduction to Keras for Researchers","text":"Variables special tensors used store mutable state (weights neural network). create Variable using initial value: update value Variable using methods $assign(value), $assign_add(increment), $assign_sub(decrement):","code":"initial_value <- random_normal(shape=c(2, 2)) a <- tf$Variable(initial_value) print(a) ## <tf.Variable 'Variable:0' shape=(2, 2) dtype=float32, numpy= ## array([[ 0.9057419 ,  0.7916686 ], ##        [ 0.28754202, -0.5408822 ]], dtype=float32)> new_value <- random_normal(shape=c(2, 2)) a$assign(new_value) ## <tf.Variable 'UnreadVariable' shape=(2, 2) dtype=float32, numpy= ## array([[-0.3405368 , -2.1463926 ], ##        [ 1.2602988 ,  0.12241419]], dtype=float32)> added_value <- random_normal(shape=c(2, 2)) a$assign_add(added_value) ## <tf.Variable 'UnreadVariable' shape=(2, 2) dtype=float32, numpy= ## array([[ 0.04820395, -2.6854615 ], ##        [ 0.23246336,  1.4535258 ]], dtype=float32)>"},{"path":"https://keras3.posit.co/dev/articles/intro_to_keras_for_researchers.html","id":"doing-math-in-tensorflow","dir":"Articles","previous_headings":"","what":"Doing math in TensorFlow","title":"Introduction to Keras for Researchers","text":"’ve used NumPy, math TensorFlow look familiar. main difference TensorFlow code can run GPU TPU.","code":"a <- random_normal(shape=c(2, 2)) b <- random_normal(shape=c(2, 2))  c <- a + b d <- tf$square(c) e <- tf$exp(d)"},{"path":"https://keras3.posit.co/dev/articles/intro_to_keras_for_researchers.html","id":"gradients","dir":"Articles","previous_headings":"","what":"Gradients","title":"Introduction to Keras for Researchers","text":"’s another big difference R: can automatically retrieve gradient differentiable expression. Just open GradientTape, start “watching” tensor via tape$watch(), compose differentiable expression using tensor input: default, variables watched automatically, don’t need manually watch : Note can compute higher-order derivatives nesting tapes:","code":"a <- random_normal(shape=c(2, 2)) b <- random_normal(shape=c(2, 2))  with(tf$GradientTape() %as% tape, {   tape$watch(a)  # Start recording the history of operations applied to `a`   c <- tf$sqrt(tf$square(a) + tf$square(b))  # Do some math using `a`   # What's the gradient of `c` with respect to `a`?   dc_da <- tape$gradient(c, a)   print(dc_da) }) ## tf.Tensor( ## [[ 0.9969011  -0.7707146 ] ##  [ 0.23378514  0.96255165]], shape=(2, 2), dtype=float32) a <- tf$Variable(a)  with(tf$GradientTape() %as% tape, {   c <- tf$sqrt(tf$square(a) + tf$square(b))   dc_da <- tape$gradient(c, a)   print(dc_da) }) ## tf.Tensor( ## [[ 0.9969011  -0.7707146 ] ##  [ 0.23378514  0.96255165]], shape=(2, 2), dtype=float32) with(tf$GradientTape() %as% outer_tape, {   with(tf$GradientTape() %as% tape, {     c <- tf$sqrt(tf$square(a) + tf$square(b))     dc_da <- tape$gradient(c, a)   })   d2c_da2 <- outer_tape$gradient(dc_da, a)   print(d2c_da2) }) ## tf.Tensor( ## [[3.3447742e-03 7.1282005e-01] ##  [5.7464113e+00 5.5013180e-02]], shape=(2, 2), dtype=float32)"},{"path":"https://keras3.posit.co/dev/articles/intro_to_keras_for_researchers.html","id":"keras-layers","dir":"Articles","previous_headings":"","what":"Keras layers","title":"Introduction to Keras for Researchers","text":"TensorFlow infrastructure layer differentiable programming, dealing tensors, variables, gradients, Keras user interface deep learning, dealing layers, models, optimizers, loss functions, metrics, . Keras serves high-level API TensorFlow: Keras makes TensorFlow simple productive. Layer class fundamental abstraction Keras. Layer encapsulates state (weights) computation (defined call method). simple layer looks like . self$add_weight() method gives shortcut creating weights: use Layer instance much like R function: weight variables (created initialize) automatically tracked weights property: many built-layers available, Dense Conv2D LSTM fancier ones like Conv3DTranspose ConvLSTM2D. smart reusing built-functionality.","code":"Linear <- new_layer_class(   \"Linear\",   initialize = function(units = 32, input_dim = 32) {     super$initialize()     self$w <- self$add_weight(       shape = shape(input_dim, units),       initializer = \"random_normal\",       trainable = TRUE     )     self$b <- self$add_weight(       shape = shape(units),       initializer = \"zeros\",       trainable = TRUE     )   },   call = function(inputs) {     tf$matmul(inputs, self$w) + self$b   } ) # Instantiate our layer. linear_layer <- Linear(units=4, input_dim=2)  # The layer can be treated as a function. # Here we call it on some data. y <- linear_layer(tf$ones(shape(2, 2))) linear_layer$weights ## [[1]] ## <KerasVariable shape=(2, 4), dtype=float32, path=linear/variable> ## ## [[2]] ## <KerasVariable shape=(4), dtype=float32, path=linear/variable_1>"},{"path":"https://keras3.posit.co/dev/articles/intro_to_keras_for_researchers.html","id":"layer-weight-creation-in-buildinput_shape","dir":"Articles","previous_headings":"","what":"Layer weight creation in build(input_shape)","title":"Introduction to Keras for Researchers","text":"’s often good idea defer weight creation build() method, don’t need specify input dim/shape layer construction time:","code":"Linear <- new_layer_class(   \"Linear\",   initialize = function(units = 32) {     super$initialize()     self$units <- units   },   build = function(input_shape) {     self$w <- self$add_weight(       shape = shape(input_shape[-1], self$units),       initializer = \"random_normal\",       trainable = TRUE     )     self$b <- self$add_weight(       shape = shape(self$units),       initializer = \"zeros\",       trainable = TRUE     )   },   call = function(inputs) {     tf$matmul(inputs, self$w) + self$b   } )  # Instantiate our layer. linear_layer <- Linear(units = 4)  # This will also call `build(input_shape)` and create the weights. y <- linear_layer(tf$ones(shape(2, 2)))"},{"path":"https://keras3.posit.co/dev/articles/intro_to_keras_for_researchers.html","id":"layer-gradients","dir":"Articles","previous_headings":"","what":"Layer gradients","title":"Introduction to Keras for Researchers","text":"can automatically retrieve gradients weights layer calling inside GradientTape. Using gradients, can update weights layer, either manually, using optimizer object. course, can modify gradients using , need .","code":"# Prepare a dataset. c(c(x_train, y_train), .) %<-% dataset_mnist()  x_train <- array_reshape(x_train, c(60000, 784)) / 255  dataset <- tfdatasets::tensor_slices_dataset(list(x_train, y_train)) %>%   tfdatasets::dataset_shuffle(buffer_size=1024) %>%   tfdatasets::dataset_batch(64)  # Instantiate our linear layer (defined above) with 10 units. linear_layer <- Linear(units = 10)  # Instantiate a logistic loss function that expects integer targets. loss_fn <- loss_sparse_categorical_crossentropy(from_logits=TRUE)  # Instantiate an optimizer. optimizer <- optimizer_sgd(learning_rate=1e-3)  # Iterate over the batches of the dataset. coro::loop(for(data in dataset) {   # Open a GradientTape.   with(tf$GradientTape() %as% tape, {     # Forward pass.     logits <- linear_layer(data[[1]])      # Loss value for this batch.     loss_value <- loss_fn(data[[2]], logits)   })    # Get gradients of the loss wrt the weights.   gradients <- tape$gradient(loss_value, linear_layer$trainable_weights)    # Update the weights of our linear layer.   optimizer$apply_gradients(zip_lists(gradients, linear_layer$trainable_weights)) }) loss_value ## tf.Tensor(1.2819729, shape=(), dtype=float32)"},{"path":"https://keras3.posit.co/dev/articles/intro_to_keras_for_researchers.html","id":"trainable-and-non-trainable-weights","dir":"Articles","previous_headings":"","what":"Trainable and non-trainable weights","title":"Introduction to Keras for Researchers","text":"Weights created layers can either trainable non-trainable. ’re exposed trainable_weights non_trainable_weights respectively. ’s layer non-trainable weight:","code":"ComputeSum <- new_layer_class(   \"ComputeSum\",   initialize = function(input_dim) {     super$initialize()     # Create a non-trainable weight.     self$total <- self$add_weight(       initializer = \"zeros\",       shape = shape(input_dim),       trainable = FALSE     )   },   call = function(inputs) {     self$total$assign_add(tf$reduce_sum(inputs, axis=0L))     self$total   } )  my_sum <- ComputeSum(input_dim = 2) x <- tf$ones(shape(2, 2))  as.array(my_sum(x)) ## [1] 2 2 as.array(my_sum(x)) ## [1] 4 4 my_sum$trainable_weights ## list()"},{"path":"https://keras3.posit.co/dev/articles/intro_to_keras_for_researchers.html","id":"layers-that-own-layers","dir":"Articles","previous_headings":"","what":"Layers that own layers","title":"Introduction to Keras for Researchers","text":"Layers can recursively nested create bigger computation blocks. layer track weights sublayers (trainable non-trainable). Note manually-created MLP equivalent following built-option:","code":"# Let's reuse the Linear class # with a `build` method that we defined above.  MLP <- new_layer_class(   \"MLP\",   initialize = function() {     super$initialize()     self$linear_1 <- Linear(units = 32)     self$linear_2 <- Linear(units = 32)     self$linear_3 <- Linear(units = 10)   },   call = function(inputs) {     x <- self$linear_1(inputs)     x <- tf$nn$relu(x)     x <- self$linear_2(x)     x <- tf$nn$relu(x)     return(self$linear_3(x))   } )  mlp <- MLP()  # The first call to the `mlp` object will create the weights. y <- mlp(tf$ones(shape=shape(3, 64)))  # Weights are recursively tracked. length(mlp$weights) ## [1] 6 mlp <- keras_model_sequential() %>%   layer_dense(units = 32, activation = \"relu\") %>%   layer_dense(units = 32, activation = \"relu\") %>%   layer_dense(units = 10)"},{"path":"https://keras3.posit.co/dev/articles/intro_to_keras_for_researchers.html","id":"tracking-losses-created-by-layers","dir":"Articles","previous_headings":"","what":"Tracking losses created by layers","title":"Introduction to Keras for Researchers","text":"Layers can create losses forward pass via add_loss() method. especially useful regularization losses. losses created sublayers recursively tracked parent layers. ’s layer creates activity regularization loss: model incorporating layer track regularization loss: losses cleared top-level layer start forward pass – don’t accumulate. layer.losses always contains losses created last forward pass. typically use losses summing computing gradients writing training loop.","code":"# A layer that creates an activity sparsity regularization loss ActivityRegularization <- new_layer_class(   \"ActivityRegularization\",   initialize = function(rate=1e-2) {     super$initialize()     self$rate <- rate   },   call = function(inputs) {     self$add_loss(self$rate * tf$reduce_sum(tf$abs(inputs)))     inputs   } ) # Let's use the loss layer in a MLP block. SparseMLP <- new_layer_class(   \"SparseMLP\",   initialize = function() {     super$initialize()     self$linear_1 <- Linear(units = 32)     self$reg <- ActivityRegularization(rate = 1e-2)     self$linear_3 <- Linear(units = 10)   },   call = function(inputs) {     x <- self$linear_1(inputs)     x <- tf$nn$relu(x)     x <- self$reg(x)     return(self$linear_3(x))   } )  mlp <- SparseMLP() y <- mlp(tf$ones(shape(10, 10)))  mlp$losses  # List containing one float32 scalar ## [[1]] ## tf.Tensor(0.18065463, shape=(), dtype=float32) # Losses correspond to the *last* forward pass. mlp <- SparseMLP() mlp(tf$ones(shape(10, 10))) ## tf.Tensor( ## [[ 0.0388482  -0.03920118  0.01624808 -0.01361975 -0.01354899  0.07107338 ##   -0.01077365  0.05688906 -0.02838149 -0.04084621] ##  [ 0.0388482  -0.03920118  0.01624808 -0.01361975 -0.01354899  0.07107338 ##   -0.01077365  0.05688906 -0.02838149 -0.04084621] ##  [ 0.0388482  -0.03920118  0.01624808 -0.01361975 -0.01354899  0.07107338 ##   -0.01077365  0.05688906 -0.02838149 -0.04084621] ##  [ 0.0388482  -0.03920118  0.01624808 -0.01361975 -0.01354899  0.07107338 ##   -0.01077365  0.05688906 -0.02838149 -0.04084621] ##  [ 0.0388482  -0.03920118  0.01624808 -0.01361975 -0.01354899  0.07107338 ##   -0.01077365  0.05688906 -0.02838149 -0.04084621] ##  [ 0.0388482  -0.03920118  0.01624808 -0.01361975 -0.01354899  0.07107338 ##   -0.01077365  0.05688906 -0.02838149 -0.04084621] ##  [ 0.0388482  -0.03920118  0.01624808 -0.01361975 -0.01354899  0.07107338 ##   -0.01077365  0.05688906 -0.02838149 -0.04084621] ##  [ 0.0388482  -0.03920118  0.01624808 -0.01361975 -0.01354899  0.07107338 ##   -0.01077365  0.05688906 -0.02838149 -0.04084621] ##  [ 0.0388482  -0.03920118  0.01624808 -0.01361975 -0.01354899  0.07107338 ##   -0.01077365  0.05688906 -0.02838149 -0.04084621] ##  [ 0.0388482  -0.03920118  0.01624808 -0.01361975 -0.01354899  0.07107338 ##   -0.01077365  0.05688906 -0.02838149 -0.04084621]], shape=(10, 10), dtype=float32) length(mlp$losses) ## [1] 1 mlp(tf$ones(shape(10, 10))) ## tf.Tensor( ## [[ 0.0388482  -0.03920118  0.01624808 -0.01361975 -0.01354899  0.07107338 ##   -0.01077365  0.05688906 -0.02838149 -0.04084621] ##  [ 0.0388482  -0.03920118  0.01624808 -0.01361975 -0.01354899  0.07107338 ##   -0.01077365  0.05688906 -0.02838149 -0.04084621] ##  [ 0.0388482  -0.03920118  0.01624808 -0.01361975 -0.01354899  0.07107338 ##   -0.01077365  0.05688906 -0.02838149 -0.04084621] ##  [ 0.0388482  -0.03920118  0.01624808 -0.01361975 -0.01354899  0.07107338 ##   -0.01077365  0.05688906 -0.02838149 -0.04084621] ##  [ 0.0388482  -0.03920118  0.01624808 -0.01361975 -0.01354899  0.07107338 ##   -0.01077365  0.05688906 -0.02838149 -0.04084621] ##  [ 0.0388482  -0.03920118  0.01624808 -0.01361975 -0.01354899  0.07107338 ##   -0.01077365  0.05688906 -0.02838149 -0.04084621] ##  [ 0.0388482  -0.03920118  0.01624808 -0.01361975 -0.01354899  0.07107338 ##   -0.01077365  0.05688906 -0.02838149 -0.04084621] ##  [ 0.0388482  -0.03920118  0.01624808 -0.01361975 -0.01354899  0.07107338 ##   -0.01077365  0.05688906 -0.02838149 -0.04084621] ##  [ 0.0388482  -0.03920118  0.01624808 -0.01361975 -0.01354899  0.07107338 ##   -0.01077365  0.05688906 -0.02838149 -0.04084621] ##  [ 0.0388482  -0.03920118  0.01624808 -0.01361975 -0.01354899  0.07107338 ##   -0.01077365  0.05688906 -0.02838149 -0.04084621]], shape=(10, 10), dtype=float32) length(mlp$losses)  # No accumulation. ## [1] 1 # Let's demonstrate how to use these losses in a training loop.  # Prepare a dataset. c(c(x_train, y_train), .) %<-% dataset_mnist() x_train <- array_reshape(x_train, c(60000, 784)) / 255  dataset <- tfdatasets::tensor_slices_dataset(list(x_train, y_train)) %>%   tfdatasets::dataset_shuffle(buffer_size=1024) %>%   tfdatasets::dataset_batch(64)  # A new MLP. mlp <- SparseMLP()  # Loss and optimizer. loss_fn <- loss_sparse_categorical_crossentropy(from_logits=TRUE) optimizer <- optimizer_sgd(learning_rate=1e-3)  coro::loop(for(data in dataset) {   x <- data[[1]]   y <- data[[2]]   with(tf$GradientTape() %as% tape, {     # Forward pass.     logits <- mlp(x)      # External loss value for this batch.     loss <- loss_fn(y, logits)      # Add the losses created during the forward pass.     loss <- loss + Reduce(`+`, mlp$losses)      # Get gradients of the loss wrt the weights.     gradients <- tape$gradient(loss, mlp$trainable_weights)      # Update the weights of our linear layer.     optimizer$apply_gradients(zip_lists(gradients, mlp$trainable_weights))   }) })"},{"path":"https://keras3.posit.co/dev/articles/intro_to_keras_for_researchers.html","id":"keeping-track-of-training-metrics","dir":"Articles","previous_headings":"","what":"Keeping track of training metrics","title":"Introduction to Keras for Researchers","text":"Keras offers broad range built-metrics, like metric_auc metric_precision_at_recall. ’s also easy create metrics lines code. use metric custom training loop, : Instantiate metric object, e.g. metric = metric_auc() Call metric$udpate_state(targets, predictions) method batch data Query result via metric$result() Reset metric’s state end epoch start evaluation via metric$reset_state() ’s simple example: can also define metrics subclassing keras.metrics.Metric. need override three functions called : Override update_state() update statistic values. Override result() return metric value. Override reset_state() reset metric initial state. example implement F1-score metric (support sample weighting). Let’s test-drive :","code":"# Instantiate a metric object accuracy <- metric_sparse_categorical_accuracy()  # Prepare our layer, loss, and optimizer. model <- keras_model_sequential() %>%   layer_dense(units = 32, activation = \"relu\") %>%   layer_dense(units = 32, activation = \"relu\") %>%   layer_dense(units = 10) loss_fn <- loss_sparse_categorical_crossentropy(from_logits = TRUE) optimizer <- optimizer_adam(learning_rate=1e-3)  for (epoch in seq_len(2)) {   coro::loop(for (data in dataset) {     x <- data[[1]]     y <- data[[2]]     with(tf$GradientTape() %as% tape, {       # Forward pass.       logits <- model(x)        # External loss value for this batch.       loss_value <- loss_fn(y, logits)     })      # Update the state of the `accuracy` metric.     accuracy$update_state(y, logits)      # Update the weights of the model to minimize the loss value.     gradients <- tape$gradient(loss_value, model$trainable_weights)     optimizer$apply_gradients(zip_lists(gradients, model$trainable_weights))    })   cat(\"Epoch:\", epoch, \"Accuracy:\", as.numeric(accuracy$result()), \"\\n\")   accuracy$reset_state() } ## Epoch: 1 Accuracy: 0.8757833 ## Epoch: 2 Accuracy: 0.93915 F1Score <- new_metric_class(   \"F1Score\",   initialize = function(self, name=\"f1_score\", dtype=\"float32\", threshold=0.5, ...) {     super$initialize(name=name, dtype=dtype, ...)     self$threshold <- threshold     self$true_positives <- self$add_weight(       name=\"tp\", dtype=dtype, initializer=\"zeros\"     )     self$false_positives <- self$add_weight(       name=\"fp\", dtype=dtype, initializer=\"zeros\"     )     self$false_negatives <- self$add_weight(       name=\"fn\", dtype=dtype, initializer=\"zeros\"     )   },   update_state = function(y_true, y_pred, sample_weight=NULL) {     y_pred <- tf$math$greater_equal(y_pred, self$threshold)     y_true <- tf$cast(y_true, tf$bool)     y_pred <- tf$cast(y_pred, tf$bool)      true_positives <- tf$cast(y_true & y_pred, self$dtype)     false_positives <- tf$cast((!y_true) & y_pred, self$dtype)     false_negatives <- tf$cast(y_true & (!y_pred), self$dtype)      if (!is.null(sample_weight)) {       sample_weight <- tf$cast(sample_weight, self$dtype)       true_positives <- true_positives * sample_weight       false_positives <- false_positives * sample_weight       false_negatives <- false_negatives * sample_weight     }      self$true_positives$assign_add(tf$reduce_sum(true_positives))     self$false_positives$assign_add(tf$reduce_sum(false_positives))     self$false_negatives$assign_add(tf$reduce_sum(false_negatives))   },    result = function() {     precision <- self$true_positives / (self$true_positives + self$false_positives)     recall <- self$true_positives / (self$true_positives + self$false_negatives)     f1_score <- 2 * precision * recall / (precision + recall)     f1_score   },    reset_state = function() {     self$true_positives$assign(0)     self$false_positives$assign(0)     self$false_negatives$assign(0)   } ) m <- F1Score() m$update_state(c(0, 1, 0, 0), c(0.3, 0.5, 0.8, 0.9)) cat(\"Intermediate result:\", as.numeric(m$result()), \"\\n\") ## Intermediate result: 0.5 m$update_state(c(1, 1, 1, 1), c(0.1, 0.7, 0.6, 0.0)) cat(\"Final result:\", as.numeric(m$result()), \"\\n\") ## Final result: 0.6"},{"path":"https://keras3.posit.co/dev/articles/intro_to_keras_for_researchers.html","id":"compiled-functions","dir":"Articles","previous_headings":"","what":"Compiled functions","title":"Introduction to Keras for Researchers","text":"Running eagerly great debugging, get better performance compiling computation static graphs. Static graphs researcher’s best friends. can compile function wrapping tf.function decorator.","code":"# Prepare our layer, loss, and optimizer. model <- keras_model_sequential() %>%   layer_dense(units = 32, activation = \"relu\") %>%   layer_dense(units = 32, activation = \"relu\") %>%   layer_dense(units = 10) loss_fn <- loss_sparse_categorical_crossentropy(from_logits = TRUE) optimizer <- optimizer_adam(learning_rate=1e-3)  # Create a training step function. train_on_batch <- tf_function(function(x, y) {   with(tf$GradientTape() %as% tape, {     # Forward pass.     logits <- model(x)     # External loss value for this batch.     loss_value <- loss_fn(y, logits)   })   # Update the weights of the model to minimize the loss value.   gradients <- tape$gradient(loss_value, model$trainable_weights)   optimizer$apply_gradients(zip_lists(gradients, model$trainable_weights))   loss_value })   # Prepare a dataset. c(c(x_train, y_train), .) %<-% dataset_mnist() x_train <- array_reshape(x_train, c(60000, 784)) / 255  dataset <- tfdatasets::tensor_slices_dataset(list(x_train, y_train)) %>%   tfdatasets::dataset_shuffle(buffer_size=1024) %>%   tfdatasets::dataset_batch(64)  i <- 0 coro::loop(for (data in dataset) {   i <- i + 1   x <- data[[1]]   y <- data[[2]]   loss <- train_on_batch(x, y)   if (i %% 100 == 0)     cat(\"Loss:\", as.numeric(loss), \"\\n\") }) ## Loss: 0.551749 ## Loss: 0.2131135 ## Loss: 0.2765952 ## Loss: 0.1296219 ## Loss: 0.2657076 ## Loss: 0.2683381 ## Loss: 0.1570166 ## Loss: 0.3139241 ## Loss: 0.08981849"},{"path":"https://keras3.posit.co/dev/articles/intro_to_keras_for_researchers.html","id":"training-mode-inference-mode","dir":"Articles","previous_headings":"","what":"Training mode & inference mode","title":"Introduction to Keras for Researchers","text":"layers, particular BatchNormalization layer Dropout layer, different behaviors training inference. layers, standard practice expose training (boolean) argument call method. exposing argument call, enable built-training evaluation loops (e.g. fit) correctly use layer training inference modes.","code":"Dropout <- new_layer_class(   \"Dropout\",   initialize = function(rate) {     super$initialize()     self$rate <- rate   },   call = function(inputs, training = NULL) {     if (!is.null(training) && training) {       return(tf$nn$dropout(inputs, rate = self$rate))     }     inputs   } )  MLPWithDropout <- new_layer_class(   \"MLPWithDropout\",   initialize = function() {     super$initialize()     self$linear_1 <- Linear(units = 32)     self$dropout <- Dropout(rate = 0.5)     self$linear_3 <- Linear(units = 10)   },   call = function(inputs, training = NULL) {     x <- self$linear_1(inputs)     x <- tf$nn$relu(x)     x <- self$dropout(x, training = training)     self$linear_3(x)   } )  mlp <- MLPWithDropout() y_train <- mlp(tf$ones(shape(2, 2)), training=TRUE) y_test <- mlp(tf$ones(shape(2, 2)), training=FALSE)"},{"path":"https://keras3.posit.co/dev/articles/intro_to_keras_for_researchers.html","id":"the-functional-api-for-model-building","dir":"Articles","previous_headings":"","what":"The Functional API for model-building","title":"Introduction to Keras for Researchers","text":"build deep learning models, don’t use object-oriented programming time. layers ’ve seen far can also composed functionally, like (call “Functional API”): Functional API tends concise subclassing, provides advantages (generally advantages functional, typed languages provide untyped OO development). However, can used define DAGs layers – recursive networks defined Layer subclasses instead. Learn Functional API . research workflows, may often find mix--matching OO models Functional models. Note Model class also features built-training & evaluation loops: fit(), predict() evaluate() (configured via compile() method). built-functions give access following built-training infrastructure features: Callbacks. can leverage built-callbacks early-stopping, model checkpointing, monitoring training TensorBoard. can also implement custom callbacks needed. Distributed training. can easily scale training multiple GPUs, TPU, even multiple machines tf.distribute API – changes code. Step fusing. steps_per_execution argument Model.compile(), can process multiple batches single tf.function call, greatly improves device utilization TPUs. won’t go details, provide simple code example . leverages built-training infrastructure implement MNIST example . can always subclass Model class (works exactly like subclassing Layer) want leverage built-training loops OO models. Just override Model$train_step() customize happens fit() retaining support built-infrastructure features outlined – callbacks, zero-code distribution support, step fusing support. may also override test_step() customize happens evaluate(), override predict_step() customize happens predict(). information, please refer guide.","code":"# We use an `Input` object to describe the shape and dtype of the inputs. # This is the deep learning equivalent of *declaring a type*. # The shape argument is per-sample; it does not include the batch size. # The functional API focused on defining per-sample transformations. # The model we create will automatically batch the per-sample transformations, # so that it can be called on batches of data. inputs <- layer_input(shape = 16, dtype = \"float32\")  # We call layers on these \"type\" objects # and they return updated types (new shapes/dtypes). outputs <- inputs %>%   Linear(units = 32) %>% # We are reusing the Linear layer we defined earlier.   Dropout(rate = 0.5) %>% # We are reusing the Dropout layer we defined earlier.   Linear(units = 10)  # A functional `Model` can be defined by specifying inputs and outputs. # A model is itself a layer like any other. model <- keras_model(inputs, outputs)  # A functional model already has weights, before being called on any data. # That's because we defined its input shape in advance (in `Input`). length(model$weights) ## [1] 4 # Let's call our model on some data, for fun. y <- model(tf$ones(shape(2, 16))) y$shape ## TensorShape([2, 10]) # You can pass a `training` argument in `__call__` # (it will get passed down to the Dropout layer). y <- model(tf$ones(shape(2, 16)), training=TRUE) inputs <- layer_input(shape = 784, dtype=\"float32\") outputs <- inputs %>%   layer_dense(units = 32, activation = \"relu\") %>%   layer_dense(units = 32, activation = \"relu\") %>%   layer_dense(units = 10) model <- keras_model(inputs, outputs)  # Specify the loss, optimizer, and metrics with `compile()`. model %>% compile(     loss = loss_sparse_categorical_crossentropy(from_logits=TRUE),     optimizer=optimizer_adam(learning_rate=1e-3),     metrics=list(metric_sparse_categorical_accuracy()), )  # Train the model with the dataset for 2 epochs. model %>% fit(dataset, epochs=2) ## Epoch 1/2 ## 938/938 - 4s - 4ms/step - loss: 0.3958 - sparse_categorical_accuracy: 0.8866 ## Epoch 2/2 ## 938/938 - 1s - 960us/step - loss: 0.1888 - sparse_categorical_accuracy: 0.9443 predictions <- model %>% predict(dataset) ## 938/938 - 1s - 1ms/step model %>% evaluate(dataset) ## 938/938 - 1s - 1ms/step - loss: 0.1763 - sparse_categorical_accuracy: 0.9454 ## $loss ## [1] 0.1763445 ## ## $sparse_categorical_accuracy ## [1] 0.9454167 CustomModel <- new_model_class(   \"CustomModel\",   initialize = function(...) {     super$initialize(...)     self$loss_tracker <- metric_mean(name=\"loss\")     self$accuracy <- metric_sparse_categorical_accuracy()     self$loss_fn <- loss_sparse_categorical_crossentropy(from_logits=TRUE)     self$optimizer <- optimizer_adam(learning_rate=1e-3)   },   train_step = function(data) {     c(x, y = NULL, sample_weight = NULL) %<-% data     with(tf$GradientTape() %as% tape, {       y_pred <- self(x, training=TRUE)       loss <- self$loss_fn(y = y, y_pred = y_pred, sample_weight=sample_weight)     })     gradients <- tape$gradient(loss, self$trainable_variables)     self$optimizer$apply_gradients(       zip_lists(gradients, self$trainable_variables)     )      # Update metrics (includes the metric that tracks the loss)     self$loss_tracker$update_state(loss)     self$accuracy$update_state(y, y_pred, sample_weight=sample_weight)     # Return a list mapping metric names to current value     list(       loss = self$loss_tracker$result(),       accuracy = self$accuracy$result()     )   },   metrics = mark_active(function() {     list(self$loss_tracker, self$accuracy)   }) )  inputs <- layer_input(shape = 784, dtype=\"float32\") outputs <- inputs %>%   layer_dense(units = 32, activation = \"relu\") %>%   layer_dense(units = 32, activation = \"relu\") %>%   layer_dense(units = 10) model <- CustomModel(inputs, outputs) model %>% compile() model %>% fit(dataset, epochs=2) ## Epoch 1/2 ## 938/938 - 2s - 2ms/step - loss: 0.3869 - sparse_categorical_accuracy: 0.8924 ## Epoch 2/2 ## 938/938 - 1s - 1ms/step - loss: 0.2163 - sparse_categorical_accuracy: 0.9370"},{"path":"https://keras3.posit.co/dev/articles/intro_to_keras_for_researchers.html","id":"end-to-end-experiment-example-1-variational-autoencoders-","dir":"Articles","previous_headings":"","what":"End-to-end experiment example 1: variational autoencoders.","title":"Introduction to Keras for Researchers","text":"things ’ve learned far: Layer encapsulates state (created __init__ build) computation (defined call). Layers can recursively nested create new, bigger computation blocks. can easily write highly hackable training loops opening GradientTape, calling model inside tape’s scope, retrieving gradients applying via optimizer. can speed training loops using @tf.function decorator. Layers can create track losses (typically regularization losses) via self.add_loss(). Let’s put things together end--end example: ’re going implement Variational AutoEncoder (VAE). ’ll train MNIST digits. VAE subclass Layer, built nested composition layers subclass Layer. feature regularization loss (KL divergence). model definition. First, Encoder class, uses Sampling layer map MNIST digit latent-space triplet (z_mean, z_log_var, z). Next, Decoder class, maps probabilistic latent space coordinates back MNIST digit. Finally, VariationalAutoEncoder composes together encoder decoder, creates KL divergence regularization loss via add_loss(). Now, let’s write training loop. training step decorated @tf.function compile super fast graph function. can see, building training type model Keras quick painless.","code":"Sampling <- new_layer_class(   \"Sampling\",   call = function(inputs) {     c(z_mean, z_log_var) %<-% inputs     batch <- op_shape(z_mean)[[1]]     dim <- op_shape(z_mean)[[2]]     epsilon <- random_normal(shape = c(batch, dim))     z_mean + op_exp(0.5 * z_log_var) * epsilon   } )  Encoder <- new_layer_class(   \"Encoder\",   initialize = function(latent_dim = 32, intermediate_dim = 64, ...) {     super$initialize(...)     self$dense_proj <- layer_dense(units = intermediate_dim, activation = \"relu\")     self$dense_mean <- layer_dense(units = latent_dim)     self$dense_log_var <- layer_dense(units = latent_dim)     self$sampling <- Sampling()   },   call = function(inputs) {     x <- self$dense_proj(inputs)     z_mean <- self$dense_mean(x)     z_log_var <- self$dense_log_var(x)     z <- self$sampling(list(z_mean, z_log_var))     list(z_mean, z_log_var, z)   } ) Decoder <- new_layer_class(   \"Decoder\",   initialize = function(original_dim, intermediate_dim = 64, ...) {     super$initialize(...)     self$dense_proj <- layer_dense(units = intermediate_dim, activation = \"relu\")     self$dense_output <- layer_dense(units = original_dim, activation = \"sigmoid\")   },   call = function(inputs) {     x <- self$dense_proj(inputs)     self$dense_output(x)   } ) VariationalAutoEncoder <- new_model_class(   \"VariationalAutoEncoder\",   initialize = function(original_dim,         intermediate_dim=64,         latent_dim=32,         name=\"autoencoder\", ...) {     super$initialize(name = name, ...)     self$original_dim <- original_dim     self$encoder <- Encoder(       latent_dim = latent_dim,       intermediate_dim = intermediate_dim     )     self$decoder <- Decoder(       original_dim = original_dim,       intermediate_dim = intermediate_dim     )   },   call = function(inputs) {     c(z_mean, z_log_var, z) %<-% self$encoder(inputs)     reconstructed <- self$decoder(z)     # Add KL divergence regularization loss.     kl_loss <- -0.5 * op_mean(       z_log_var - op_square(z_mean) - op_exp(z_log_var) + 1     )     self$add_loss(kl_loss)     reconstructed   } ) # Our model. vae <- VariationalAutoEncoder(   original_dim = 784,   intermediate_dim = 64,   latent_dim = 32 )  # Loss and optimizer. loss_fn <- loss_mean_squared_error() optimizer = optimizer_adam(learning_rate=1e-3)  # Prepare a dataset. c(c(x_train, .), .) %<-% dataset_mnist() x_train <- array_reshape(x_train, c(60000, 784)) / 255  dataset <- tfdatasets::tensor_slices_dataset(x_train) %>%   tfdatasets::dataset_shuffle(buffer_size=1024) %>%   tfdatasets::dataset_batch(32)   training_step <- tf_function(function(x) {   with(tf$GradientTape() %as% tape, {     reconstructed <- vae(x)  # Compute input reconstruction.     # Compute loss.     loss <- loss_fn(x, reconstructed)     loss <- loss + op_sum(vae$losses)  # Add KLD term.   })   # Update the weights of the VAE.   grads <- tape$gradient(loss, vae$trainable_weights)   optimizer$apply_gradients(zip_lists(grads, vae$trainable_weights))   loss })  losses <- c()  # Keep track of the losses over time. coro::loop(for(data in dataset) {   loss <- training_step(data)    # Logging.   losses[length(losses) + 1] <- as.numeric(loss)   if (length(losses) %% 100 == 0) {     cat(\"Step:\", length(losses), \"Loss:\", mean(losses), \"\\n\")   }   # Stop after 1000 steps.   # Training the model to convergence is left   # as an exercise to the reader.   if (length(losses) >= 1000) {     break   } }) ## Step: 100 Loss: 0.1270978 ## Step: 200 Loss: 0.1003238 ## Step: 300 Loss: 0.09001128 ## Step: 400 Loss: 0.08493649 ## Step: 500 Loss: 0.08171404 ## Step: 600 Loss: 0.07926706 ## Step: 700 Loss: 0.07790599 ## Step: 800 Loss: 0.07670419 ## Step: 900 Loss: 0.07570736 ## Step: 1000 Loss: 0.07476593"},{"path":"https://keras3.posit.co/dev/articles/intro_to_keras_for_researchers.html","id":"end-to-end-experiment-example-2-hypernetworks-","dir":"Articles","previous_headings":"","what":"End-to-end experiment example 2: hypernetworks.","title":"Introduction to Keras for Researchers","text":"Let’s take look another kind research experiment: hypernetworks. idea use small deep neural network (hypernetwork) generate weights larger network (main network). Let’s implement really trivial hypernetwork: ’ll use small 2-layer network generate weights larger 3-layer network. training loop. batch data: use hypernetwork generate array weight coefficients, weights_pred reshape coefficients kernel & bias tensors main_network run forward pass main_network compute actual MNIST predictions run backprop weights hypernetwork minimize final classification loss Implementing arbitrary research ideas Keras straightforward highly productive. Imagine trying 25 ideas per day (20 minutes per experiment average)! Keras designed go idea results fast possible, believe key great research. hope enjoyed quick introduction. Let us know build Keras!","code":"input_dim <- 784 classes <- 10  # This is the main network we'll actually use to predict labels. inputs <- layer_input(shape = input_dim) dense1 <- layer_dense(units = 64, activation = \"relu\") dense1$built <- TRUE  dense2 <- layer_dense(units = classes) dense2$built <- TRUE  outputs <- inputs %>% dense1() %>% dense2() main_network <- keras_model(inputs, outputs)  # This is the number of weight coefficients to generate. Each layer in the # main network requires output_dim * input_dim + output_dim coefficients. num_weights_to_generate <- (classes * 64 + classes) + (64 * input_dim + 64)  # This is the hypernetwork that generates the weights of the `main_network` above. hypernetwork <- keras_model_sequential() %>%   layer_dense(units=16, activation=\"relu\") %>%   layer_dense(units=num_weights_to_generate, activation=\"sigmoid\") # Loss and optimizer. loss_fn <- loss_sparse_categorical_crossentropy(from_logits = TRUE) optimizer <- optimizer_adam(learning_rate=1e-4)  # Prepare a dataset. c(c(x_train, y_train), .) %<-% dataset_mnist() x_train <- array_reshape(x_train, c(60000, 784)) / 255  dataset <- tfdatasets::tensor_slices_dataset(list(x_train, y_train)) %>%   tfdatasets::dataset_shuffle(buffer_size=1024) %>%   # We'll use a batch size of 1 for this experiment.   tfdatasets::dataset_batch(1)  train_step <- function(x, y) {   with(tf$GradientTape() %as% tape, {     weights_pred <- hypernetwork(x)      # Reshape them to the expected shapes for w and b for the outer model.     # Layer 1 kernel.     start_index <- 1     w1_shape <- c(input_dim, 64)     w1_coeffs <- weights_pred[, start_index:(start_index + prod(w1_shape) - 1)]     w1 <- tf$reshape(w1_coeffs, as.integer(w1_shape))     start_index <- start_index + prod(w1_shape)      # Layer 1 bias.     b1_shape <- c(64)     b1_coeffs <- weights_pred[, start_index:(start_index + prod(b1_shape) - 1)]     b1 <- tf$reshape(b1_coeffs, as.integer(b1_shape))     start_index <- start_index + prod(b1_shape)      # Layer 2 kernel.     w2_shape <- c(64, classes)     w2_coeffs <- weights_pred[, start_index:(start_index + prod(w2_shape) - 1)]     w2 <- tf$reshape(w2_coeffs, as.integer(w2_shape))     start_index <- start_index + prod(w2_shape)      # Layer 2 bias.     b2_shape <- c(classes)     b2_coeffs <- weights_pred[, start_index:(start_index + prod(b2_shape) - 1)]     b2 <- tf$reshape(b2_coeffs, as.integer(b2_shape))     start_index <- start_index + prod(b2_shape)      # Set the weight predictions as the weight variables on the outer model.     dense1$kernel <- w1     dense1$bias <- b1     dense2$kernel <- w2     dense2$bias <- b2      # Inference on the outer model.     preds <- main_network(x)     loss <- loss_fn(y, preds)   })    grads <- tape$gradient(loss, hypernetwork$trainable_weights)   optimizer$apply_gradients(zip_lists(grads, hypernetwork$trainable_weights))   loss }  losses <- c()  # Keep track of the losses over time. coro::loop(for (data in dataset) {   x <- data[[1]]   y <- data[[2]]   loss <- train_step(x, y)    # Logging.   losses[length(losses) + 1] <- as.numeric(loss)   if (length(losses) %% 100 == 0) {     cat(\"Step:\", length(losses), \"Loss:\", mean(losses), \"\\n\")   }   # Stop after 1000 steps.   # Training the model to convergence is left   # as an exercise to the reader.   if (length(losses) >= 1000) {     break   } }) ## Step: 100 Loss: 2.536778 ## Step: 200 Loss: 2.236472 ## Step: 300 Loss: 2.119417 ## Step: 400 Loss: 2.040341 ## Step: 500 Loss: 1.949125 ## Step: 600 Loss: 1.859384 ## Step: 700 Loss: 1.845726 ## Step: 800 Loss: 1.820594 ## Step: 900 Loss: 1.771334 ## Step: 1000 Loss: 1.730648"},{"path":"https://keras3.posit.co/dev/articles/making_new_layers_and_models_via_subclassing.html","id":"introduction","dir":"Articles","previous_headings":"","what":"Introduction","title":"Making new layers and models via subclassing","text":"guide cover everything need know build subclassed layers models. particular, ’ll learn following features: Layer class add_weight() method Trainable non-trainable weights build() method Making sure layers can used backend add_loss() method training argument call() mask argument call() Making sure layers can serialized Let’s dive .","code":""},{"path":"https://keras3.posit.co/dev/articles/making_new_layers_and_models_via_subclassing.html","id":"setup","dir":"Articles","previous_headings":"","what":"Setup","title":"Making new layers and models via subclassing","text":"","code":"library(keras3) library(tensorflow, exclude = c(\"set_random_seed\", \"shape\")) library(tfdatasets, exclude = \"shape\")"},{"path":"https://keras3.posit.co/dev/articles/making_new_layers_and_models_via_subclassing.html","id":"the-layer-class-the-combination-of-state-weights-and-some-computation","dir":"Articles","previous_headings":"","what":"The Layer class: the combination of state (weights) and some computation","title":"Making new layers and models via subclassing","text":"One central abstractions Keras Layer class. layer encapsulates state (layer’s “weights”) transformation inputs outputs (“call”, layer’s forward pass). ’s densely-connected layer. two state variables: variables w b. use layer calling tensor input(s), much like R function. Note weights w b automatically tracked layer upon set layer attributes:","code":"layer_linear <- Layer(\"Linear\",    initialize = function(units = 32, input_dim = 32, ...) {     super$initialize(...)     self$w <- self$add_weight(       shape = shape(input_dim, units),       initializer = \"random_normal\",       trainable = TRUE     )     self$b <- self$add_weight(       shape = shape(units),       initializer = \"zeros\",       trainable = TRUE     )   },    call = function(inputs) {     op_matmul(inputs, self$w) + self$b   } ) x <- op_ones(c(2, 2)) linear_layer <- layer_linear(units = 4, input_dim = 2) y <- linear_layer(x) print(y) ## tf.Tensor( ## [[0.02153057 0.15450525 0.0205495  0.04493225] ##  [0.02153057 0.15450525 0.0205495  0.04493225]], shape=(2, 4), dtype=float32) linear_layer$weights ## [[1]] ## <Variable path=linear/variable, shape=(2, 4), dtype=float32, value=[[-0.06251299  0.05335509  0.01485647 -0.00985784] ##  [ 0.08404355  0.10115016  0.00569303  0.05479009]]> ## ## [[2]] ## <Variable path=linear/variable_1, shape=(4), dtype=float32, value=[0. 0. 0. 0.]>"},{"path":"https://keras3.posit.co/dev/articles/making_new_layers_and_models_via_subclassing.html","id":"layers-can-have-non-trainable-weights","dir":"Articles","previous_headings":"","what":"Layers can have non-trainable weights","title":"Making new layers and models via subclassing","text":"Besides trainable weights, can add non-trainable weights layer well. weights meant taken account backpropagation, training layer. ’s add use non-trainable weight: ’s part layer$weights, gets categorized non-trainable weight:","code":"layer_compute_sum <- Layer(   \"ComputeSum\",   initialize = function(input_dim) {     super$initialize()     self$total <- self$add_weight(       initializer = \"zeros\",       shape = shape(input_dim),       trainable = FALSE     )   },   call = function(inputs) {     self$total$assign_add(op_sum(inputs, axis = 1))     self$total   } )  x <- op_ones(c(2, 2)) my_sum <- layer_compute_sum(input_dim = 2) y <- my_sum(x) print(as.array(y)) ## [1] 2 2 y <- my_sum(x) print(as.array(y)) ## [1] 4 4 cat(\"weights:\", length(my_sum$weights)) ## weights: 1 cat(\"non-trainable weights:\", length(my_sum$non_trainable_weights)) ## non-trainable weights: 1 # It's not included in the trainable weights: cat(\"trainable_weights:\", length(my_sum$trainable_weights)) ## trainable_weights: 0"},{"path":"https://keras3.posit.co/dev/articles/making_new_layers_and_models_via_subclassing.html","id":"best-practice-deferring-weight-creation-until-the-shape-of-the-inputs-is-known","dir":"Articles","previous_headings":"","what":"Best practice: deferring weight creation until the shape of the inputs is known","title":"Making new layers and models via subclassing","text":"Linear layer took input_dim argument used compute shape weights w b initialize(): many cases, may know advance size inputs, like lazily create weights value becomes known, time instantiating layer. Keras API, recommend creating layer weights build(self, inputs_shape) method layer. Like : call() method layer automatically run build first time called. now layer ’s lazy thus easier use: Implementing build() separately shown nicely separates creating weights using weights every call.","code":"layer_linear <- Layer(\"Linear\",    initialize = function(units = 32, input_dim = 32, ...) {     super$initialize(...)     self$w <- self$add_weight(       shape = shape(input_dim, units),       initializer = \"random_normal\",       trainable = TRUE     )     self$b <- self$add_weight(       shape = shape(units),       initializer = \"zeros\",       trainable = TRUE     )   },    call = function(inputs) {     op_matmul(inputs, self$w) + self$b   } ) layer_linear <- Layer(   \"Linear\",   initialize = function(units = 32, ...) {     self$units <- as.integer(units)     super$initialize(...)   },   build = function(input_shape) {     self$w <- self$add_weight(       shape = shape(tail(input_shape, 1), self$units),       initializer = \"random_normal\",       trainable = TRUE     )     self$b <- self$add_weight(       shape = shape(self$units),       initializer = \"zeros\",       trainable = TRUE     )   },   call = function(inputs) {     op_matmul(inputs, self$w) + self$b   } ) # At instantiation, we don't know on what inputs this is going to get called linear_layer <- layer_linear(units = 32)  # The layer's weights are created dynamically the first time the layer is called y <- linear_layer(x)"},{"path":"https://keras3.posit.co/dev/articles/making_new_layers_and_models_via_subclassing.html","id":"layers-are-recursively-composable","dir":"Articles","previous_headings":"","what":"Layers are recursively composable","title":"Making new layers and models via subclassing","text":"assign Layer instance attribute another Layer, outer layer start tracking weights created inner layer. recommend creating sublayers initialize() method leave first call() trigger building weights.","code":"MLPBlock <- Layer(   \"MLPBlock\",   initialize = function() {     super$initialize()     self$linear_1 <- layer_linear(units = 32)     self$linear_2 <- layer_linear(units = 32)     self$linear_3 <- layer_linear(units = 1)   },   call = function(inputs) {     inputs |>       self$linear_1() |>       activation_relu() |>       self$linear_2() |>       activation_relu() |>       self$linear_3()   } )  mlp <- MLPBlock() # The first call to the `mlp` will create the weights y <- mlp(op_ones(shape = c(3, 64)))  cat(\"weights:\", length(mlp$weights), \"\\n\") ## weights: 6 cat(\"trainable weights:\", length(mlp$trainable_weights), \"\\n\") ## trainable weights: 6"},{"path":"https://keras3.posit.co/dev/articles/making_new_layers_and_models_via_subclassing.html","id":"backend-agnostic-layers-and-backend-specific-layers","dir":"Articles","previous_headings":"","what":"Backend-agnostic layers and backend-specific layers","title":"Making new layers and models via subclassing","text":"long layer uses APIs ops namespace (ie. using functions starting op_), (Keras namespaces activations_*, random_*, layer_*), can used backend – TensorFlow, JAX, PyTorch. layers ’ve seen far guide work Keras backends. ops namespace gives access : NumPy API, e.g. op_matmul, op_sum, op_reshape, op_stack, etc. Neural networks-specific APIs op_softmax, op_conv, op_binary_crossentropy, op_relu, etc. can also use backend-native APIs layers (tf$nn functions), , layer usable backend question. instance, write following JAX-specific layer using jax$numpy: equivalent TensorFlow-specific layer: equivalent PyTorch-specific layer: cross-backend compatibility tremendously useful property, strongly recommend seek always make layers backend-agnostic leveraging Keras APIs.","code":"# keras3::install_keras(backend = c(\"jax\")) jax <- reticulate::import(\"jax\")  Linear <- new_layer_class(   ...   call = function(inputs) {     jax$numpy$matmul(inputs, self$w) + self$b   } ) library(tensorflow)  Linear <- new_layer_class(   ...   call = function(inputs) {     tf$matmul(inputs, self$w) + self$b   } ) torch <- reticulate::import(\"torch\")  Linear <- new_layer_class(   ...   call = function(inputs) {     torch$matmul(inputs, self$w) + self$b   } )"},{"path":"https://keras3.posit.co/dev/articles/making_new_layers_and_models_via_subclassing.html","id":"the-add_loss-method","dir":"Articles","previous_headings":"","what":"The add_loss() method","title":"Making new layers and models via subclassing","text":"writing call() method layer, can create loss tensors want use later, writing training loop. doable calling self$add_loss(value): losses (including created inner layer) can retrieved via layer$losses. property reset start every call top-level layer, layer$losses always contains loss values created last forward pass. addition, loss property also contains regularization losses created weights inner layer: losses meant taken account writing custom training loops. also work seamlessly fit() (get automatically summed added main loss, ):","code":"# A layer that creates an activity regularization loss layer_activity_regularization <- Layer(   \"ActivityRegularizationLayer\",   initialize = function(rate = 1e-2) {     self$rate <- as.numeric(rate)     super$initialize()   },   call = function(inputs) {     self$add_loss(self$rate * op_mean(inputs))     inputs   } ) layer_outer <- Layer(   \"OuterLayer\",   initialize = function() {     super$initialize()     self$activity_reg <- layer_activity_regularization(rate = 1e-2)   },   call = function(inputs) {     self$activity_reg(inputs)     inputs   } )  layer <- layer_outer() # No losses yet since the layer has never been called cat(\"losses:\", length(layer$losses), \"\\n\") ## losses: 0 x <- layer(op_zeros(c(1, 1))) # We created one loss value cat(\"losses:\", length(layer$losses), \"\\n\") ## losses: 1 # `layer$losses` gets reset at the start of each call x <- layer(op_zeros(c(1, 1))) # This is the loss created during the call above cat(\"losses:\", length(layer$losses), \"\\n\") ## losses: 1 layer_outer_with_kernel_regularizer <- Layer(   \"OuterLayerWithKernelRegularizer\",   initialize = function() {     super$initialize()     self$dense <- layer_dense(units = 32,                               kernel_regularizer = regularizer_l2(1e-3))   },   call = function(inputs) {     self$dense(inputs)   } )  layer <- layer_outer_with_kernel_regularizer() x <- layer(op_zeros(c(1, 1)))  # This is `1e-3 * sum(layer$dense$kernel ** 2)`, # created by the `kernel_regularizer` above. print(layer$losses) ## [[1]] ## tf.Tensor(0.002025157, shape=(), dtype=float32) inputs <- keras_input(shape = 3) outputs <- inputs |> layer_activity_regularization() model <- keras_model(inputs, outputs)  # If there is a loss passed in `compile`, the regularization # losses get added to it model |> compile(optimizer = \"adam\", loss = \"mse\") model |> fit(random_normal(c(2, 3)), random_normal(c(2, 3)), epochs = 1) ## 1/1 - 0s - 131ms/step - loss: 1.9081 # It's also possible not to pass any loss in `compile`, # since the model already has a loss to minimize, via the `add_loss` # call during the forward pass! model |> compile(optimizer = \"adam\") model |> fit(random_normal(c(2, 3)), random_normal(c(2, 3)), epochs = 1) ## 1/1 - 0s - 106ms/step - loss: 1.6613"},{"path":"https://keras3.posit.co/dev/articles/making_new_layers_and_models_via_subclassing.html","id":"you-can-optionally-enable-serialization-on-your-layers","dir":"Articles","previous_headings":"","what":"You can optionally enable serialization on your layers","title":"Making new layers and models via subclassing","text":"need custom layers serializable part Functional model, can optionally implement get_config() method: Note initialize() method base Layer class takes keyword arguments, particular name dtype. ’s good practice pass arguments parent class initialize() include layer config: need flexibility deserializing layer config, can also override from_config() class method. base implementation from_config(): learn serialization saving, see complete guide saving serializing models.","code":"layer_linear <- Layer(   \"Linear\",   initialize = function(units = 32) {     self$units <- as.integer(units)     super$initialize()   },   build = function(input_shape) {     self$w <- self$add_weight(       shape = shape(tail(input_shape, 1), self$units),       initializer = \"random_normal\",       trainable = TRUE     )     self$b <- self$add_weight(       shape = shape(self$units),       initializer = \"zeros\",       trainable = TRUE     )   },   call = function(inputs) {     op_matmul(inputs, self$w) + self$b   },   get_config = function() {     list(units = self$units)   } )  # Now you can recreate the layer from its config: layer <- layer_linear(units = 64) config <- get_config(layer) str(config) ## List of 1 ##  $ units: int 64 ##  - attr(*, \"__class__\")=<class '<r-globalenv>.Linear'> new_layer <- from_config(config) Linear <- new_layer_class(   \"Linear\",   initialize = function(units = 32, ...) {     self$units <- as.integer(units)     super$initialize(...)   },   build = function(input_shape) {     self$w <- self$add_weight(       shape = shape(tail(input_shape, 1), self$units),       initializer = \"random_normal\",       trainable = TRUE     )     self$b <- self$add_weight(       shape = shape(self$units),       initializer = \"zeros\",       trainable = TRUE     )   },   call = function(inputs) {     op_matmul(inputs, self$w) + self$b   },   get_config = function() {     list(units = self$units)   } )  layer <- Linear(units = 64) config <- get_config(layer) str(config) ## List of 1 ##  $ units: int 64 ##  - attr(*, \"__class__\")=<class '<r-globalenv>.Linear'> new_layer <- from_config(config) Layer(   ...,   from_config = function(config) {     # calling `__class__`() creates a new instance and calls initialize()     do.call(`__class__`, config)   } )"},{"path":"https://keras3.posit.co/dev/articles/making_new_layers_and_models_via_subclassing.html","id":"privileged-training-argument-in-the-call-method","dir":"Articles","previous_headings":"","what":"Privileged training argument in the call() method","title":"Making new layers and models via subclassing","text":"layers, particular BatchNormalization layer Dropout layer, different behaviors training inference. layers, standard practice expose training (boolean) argument call() method. exposing argument call(), enable built-training evaluation loops (e.g. fit()) correctly use layer training inference.","code":"layer_custom_dropout <- Layer(   \"CustomDropout\",   initialize = function(rate, ...) {     super$initialize(...)     self$rate <- rate     self$seed_generator <- random_seed_generator(1337)   },   call = function(inputs, training = NULL) {     if (isTRUE(training))       return(random_dropout(inputs, rate = self$rate,                             seed = self.seed_generator))     inputs   } )"},{"path":"https://keras3.posit.co/dev/articles/making_new_layers_and_models_via_subclassing.html","id":"privileged-mask-argument-in-the-call-method","dir":"Articles","previous_headings":"","what":"Privileged mask argument in the call() method","title":"Making new layers and models via subclassing","text":"privileged argument supported call() mask argument. find Keras RNN layers. mask boolean tensor (one boolean value per timestep input) used skip certain input timesteps processing timeseries data. Keras automatically pass correct mask argument call() layers support , mask generated prior layer. Mask-generating layers Embedding layer configured mask_zero = TRUE, Masking layer.","code":""},{"path":"https://keras3.posit.co/dev/articles/making_new_layers_and_models_via_subclassing.html","id":"the-model-class","dir":"Articles","previous_headings":"","what":"The Model class","title":"Making new layers and models via subclassing","text":"general, use Layer class define inner computation blocks, use Model class define outer model – object train. instance, ResNet50 model, several ResNet blocks subclassing Layer, single Model encompassing entire ResNet50 network. Model class API Layer, following differences: exposes built-training, evaluation, prediction loops (fit(), evaluate(), predict()). exposes list inner layers, via model$layers property. exposes saving serialization APIs (save(), save_weights()…) Effectively, Layer class corresponds refer literature “layer” (“convolution layer” “recurrent layer”) “block” (“ResNet block” “Inception block”). Meanwhile, Model class corresponds referred literature “model” (“deep learning model”) “network” (“deep neural network”). ’re wondering, “use Layer class Model class?”, ask : need call fit() ? need call save() ? , go Model. (either class just block bigger system, writing training & saving code ), use Layer. instance, take mini-resnet example , use build Model train fit(), save save_weights():","code":"ResNet <- Model(   \"ResNet\",   initialize = function(num_classes = 1000, ...) {     super$initialize(...)     self$block_1 <- layer_resnet_block()     self$block_2 <- layer_resnet_block()     self$global_pool <- layer_global_average_pooling_2d()     self$classifier <- layer_dense(num_classes)   },   call = function(inputs) {     inputs |>       self$block_1() |>       self$block_2() |>       self$global_pool() |>       self$classifier()   } )  resnet <- ResNet() dataset <- ... resnet |> fit(dataset, epochs=10) resnet |> save_model(\"filepath.keras\")"},{"path":"https://keras3.posit.co/dev/articles/making_new_layers_and_models_via_subclassing.html","id":"putting-it-all-together-an-end-to-end-example","dir":"Articles","previous_headings":"","what":"Putting it all together: an end-to-end example","title":"Making new layers and models via subclassing","text":"’s ’ve learned far: Layer encapsulate state (created initialize() build()) computation (defined call()). Layers can recursively nested create new, bigger computation blocks. Layers backend-agnostic long use Keras APIs. can use backend-native APIs (jax$numpy, torch$nn tf$nn), layer usable specific backend. Layers can create track losses (typically regularization losses) via add_loss(). outer container, thing want train, Model. Model just like Layer, added training serialization utilities. Let’s put things together end--end example: ’re going implement Variational AutoEncoder (VAE) backend-agnostic fashion – runs TensorFlow, JAX, PyTorch. ’ll train MNIST digits. VAE subclass Model, built nested composition layers subclass Layer. feature regularization loss (KL divergence). Let’s train MNIST using fit() API:","code":"layer_sampling <- Layer(   \"Sampling\",   initialize = function(...) {     super$initialize(...)     self$seed_generator <- random_seed_generator(1337)   },   call = function(inputs) {     c(z_mean, z_log_var) %<-% inputs     batch <- op_shape(z_mean)[[1]]     dim <- op_shape(z_mean)[[2]]     epsilon <- random_normal(shape = c(batch, dim),                              seed=self$seed_generator)     z_mean + op_exp(0.5 * z_log_var) * epsilon   } )  # Maps MNIST digits to a triplet (z_mean, z_log_var, z). layer_encoder <- Layer(   \"Encoder\",   initialize = function(latent_dim = 32, intermediate_dim = 64, ...) {     super$initialize(...)     self$dense_proj <-       layer_dense(units = intermediate_dim,  activation = \"relu\")     self$dense_mean <- layer_dense(units = latent_dim)     self$dense_log_var <- layer_dense(units = latent_dim)     self$sampling <- layer_sampling()   },   call = function(inputs) {     x <- self$dense_proj(inputs)     z_mean <- self$dense_mean(x)     z_log_var <- self$dense_log_var(x)     z <- self$sampling(list(z_mean, z_log_var))     list(z_mean, z_log_var, z)   } )  # Converts z, the encoded digit vector, back into a readable digit. layer_decoder <- Layer(   \"Decoder\",   initialize = function(original_dim, intermediate_dim = 64, ...) {     super$initialize(...)     self$dense_proj <-       layer_dense(units = intermediate_dim, activation = \"relu\")     self$dense_output <-       layer_dense(units = original_dim, activation = \"sigmoid\")   },   call = function(inputs) {     x <- self$dense_proj(inputs)     self$dense_output(x)   } )  # Combines the encoder and decoder into an end-to-end model for training. VariationalAutoEncoder <- Model(   \"VariationalAutoEncoder\",    initialize = function(original_dim, intermediate_dim = 64, latent_dim = 32,                         name = \"autoencoder\", ...) {     super$initialize(name = name, ...)     self$original_dim <- original_dim     self$encoder <- layer_encoder(latent_dim = latent_dim,                             intermediate_dim = intermediate_dim)     self$decoder <- layer_decoder(original_dim = original_dim,                             intermediate_dim = intermediate_dim)   },    call = function(inputs) {     c(z_mean, z_log_var, z) %<-% self$encoder(inputs)     reconstructed <- self$decoder(z)     # Add KL divergence regularization loss.     kl_loss <- -0.5 * op_mean(z_log_var - op_square(z_mean) - op_exp(z_log_var) + 1)     self$add_loss(kl_loss)     reconstructed   } ) c(c(x_train, .), .) %<-% dataset_mnist() x_train <- x_train |>   op_reshape(c(60000, 784)) |>   op_cast(\"float32\") |>   op_divide(255)  original_dim <- 784 vae <- VariationalAutoEncoder(   original_dim = 784,   intermediate_dim = 64,   latent_dim = 32 )  optimizer <- optimizer_adam(learning_rate = 1e-3) vae |> compile(optimizer, loss = loss_mean_squared_error())  vae |> fit(x_train, x_train, epochs = 2, batch_size = 64) ## Epoch 1/2 ## 938/938 - 4s - 4ms/step - loss: 0.0748 ## Epoch 2/2 ## 938/938 - 1s - 1ms/step - loss: 0.0676"},{"path":"https://keras3.posit.co/dev/articles/sequential_model.html","id":"setup","dir":"Articles","previous_headings":"","what":"Setup","title":"The Sequential model","text":"","code":"library(keras3)"},{"path":"https://keras3.posit.co/dev/articles/sequential_model.html","id":"when-to-use-a-sequential-model","dir":"Articles","previous_headings":"","what":"When to use a Sequential model","title":"The Sequential model","text":"Sequential model appropriate plain stack layers layer exactly one input tensor one output tensor. Schematically, following Sequential model: equivalent function: Sequential model appropriate : model multiple inputs multiple outputs layers multiple inputs multiple outputs need layer sharing want non-linear topology (e.g. residual connection, multi-branch model)","code":"model <- keras_model_sequential() |>   layer_dense(units = 2, activation = \"relu\", name = \"layer1\") |>   layer_dense(units = 3, activation = \"relu\", name = \"layer2\") |>   layer_dense(units = 4, name = \"layer3\")  # Call model on a test input x <- op_ones(c(3, 3)) y <- model(x) # Create 3 layers layer1 <- layer_dense(units = 2, activation=\"relu\", name=\"layer1\") layer2 <- layer_dense(units = 3, activation=\"relu\", name=\"layer2\") layer3 <- layer_dense(units = 4, name=\"layer3\")  # Call layers on a test input x <- op_ones(c(3, 3)) y <- x |> layer1() |> layer2() |> layer3()"},{"path":"https://keras3.posit.co/dev/articles/sequential_model.html","id":"creating-a-sequential-model","dir":"Articles","previous_headings":"","what":"Creating a Sequential model","title":"The Sequential model","text":"can create Sequential model piping layers keras_model_sequential() object: passing list layers keras_model_sequential(): layers accessible via layers attribute: can also create Sequential model incrementally: Note ’s also corresponding pop_layer() method remove layers: Sequential model behaves much like stack layers. Also note Sequential constructor accepts name argument, just like layer model Keras. useful annotate TensorBoard graphs semantically meaningful names.","code":"model <- keras_model_sequential() |>   layer_dense(units = 2, activation = \"relu\") |>   layer_dense(units = 3, activation = \"relu\") |>   layer_dense(units = 4) model <- keras_model_sequential(layers = list(   layer_dense(units = 2, activation = \"relu\"),   layer_dense(units = 3, activation = \"relu\"),   layer_dense(units = 4) )) model$layers ## [[1]] ## <Dense name=dense_3, built=False> ##  signature: (*args, **kwargs) ## ## [[2]] ## <Dense name=dense_4, built=False> ##  signature: (*args, **kwargs) ## ## [[3]] ## <Dense name=dense_5, built=False> ##  signature: (*args, **kwargs) model <- keras_model_sequential() model |> layer_dense(units = 2, activation=\"relu\") model |> layer_dense(units = 3, activation=\"relu\") model |> layer_dense(units = 4) model |> pop_layer() ## <Dense name=dense_8, built=False> ##  signature: (*args, **kwargs) length(model$layers)  # 2 ## [1] 2 model <- keras_model_sequential(name = \"my_sequential\") model |> layer_dense(units = 2, activation=\"relu\", name = \"layer1\") model |> layer_dense(units = 3, activation=\"relu\", name = \"layer2\") model |> layer_dense(units = 4, name = \"layer3\")"},{"path":"https://keras3.posit.co/dev/articles/sequential_model.html","id":"specifying-the-input-shape-in-advance","dir":"Articles","previous_headings":"","what":"Specifying the input shape in advance","title":"The Sequential model","text":"Generally, layers Keras need know shape inputs order able create weights. create layer like , initially, weights: creates weights first time called input, since shape weights depends shape inputs: Naturally, also applies Sequential models. instantiate Sequential model without input shape, isn’t “built”: weights (calling model$weights results error stating just ). weights created model first sees input data: model “built”, can call summary() method display contents: However, can useful building Sequential model incrementally able display summary model far, including current output shape. case, start model passing input_shape argument model, knows input shape start: Models built predefined input shape like always weights (even seeing data) always defined output shape. general, ’s recommended best practice always specify input shape Sequential model advance know .","code":"layer <- layer_dense(units = 3) layer$weights  # Empty ## list() # Call layer on a test input x <- op_ones(c(1, 4)) y <- layer(x) layer$weights  # Now it has weights, of shape (4, 3) and (3,) ## [[1]] ## <Variable path=dense_9/kernel, shape=(4, 3), dtype=float32, value=[[ 0.48581433  0.78749573  0.61015   ] ##  [ 0.7962619   0.7261175  -0.8046875 ] ##  [-0.6189915   0.37973273  0.50559556] ##  [-0.5455791  -0.60714126  0.19791973]]> ## ## [[2]] ## <Variable path=dense_9/bias, shape=(3), dtype=float32, value=[0. 0. 0.]> model <- keras_model_sequential() |>   layer_dense(units = 2, activation = \"relu\") |>   layer_dense(units = 3, activation = \"relu\") |>   layer_dense(units = 4) # No weights at this stage!  # At this point, you can't do this: # model$weights   # Call the model on a test input x <- op_ones(c(1, 4)) y <- model(x) length(model$weights) ## [1] 6 summary(model) ## Model: \"sequential_4\" ## ┏━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━┓ ## ┃ Layer (type)                    ┃ Output Shape           ┃       Param # ┃ ## ┡━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━┩ ## │ dense_10 (Dense)                │ (1, 2)                 │            10 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ dense_11 (Dense)                │ (1, 3)                 │             9 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ dense_12 (Dense)                │ (1, 4)                 │            16 │ ## └─────────────────────────────────┴────────────────────────┴───────────────┘ ##  Total params: 35 (140.00 B) ##  Trainable params: 35 (140.00 B) ##  Non-trainable params: 0 (0.00 B) model <- keras_model_sequential(input_shape = 4) |>   layer_dense(units = 2, activation = \"relu\") summary(model) ## Model: \"sequential_5\" ## ┏━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━┓ ## ┃ Layer (type)                    ┃ Output Shape           ┃       Param # ┃ ## ┡━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━┩ ## │ dense_13 (Dense)                │ (None, 2)              │            10 │ ## └─────────────────────────────────┴────────────────────────┴───────────────┘ ##  Total params: 10 (40.00 B) ##  Trainable params: 10 (40.00 B) ##  Non-trainable params: 0 (0.00 B) model$layers ## [[1]] ## <Dense name=dense_13, built=True> ##  signature: (*args, **kwargs)"},{"path":"https://keras3.posit.co/dev/articles/sequential_model.html","id":"a-common-debugging-workflow-add-layers-summary","dir":"Articles","previous_headings":"","what":"A common debugging workflow: add layers + summary()","title":"The Sequential model","text":"building new Sequential architecture, ’s useful incrementally stack layers |> frequently print model summaries. instance, enables monitor stack Conv2D MaxPooling2D layers downsampling image feature maps: practical, right? Note |> equivalent calling model$add(), modifies model -place, don’t need reassign model symbol step.","code":"model <- keras_model_sequential(input_shape = c(250, 250, 3)) |>   layer_conv_2d(filters = 32, kernel_size = 5, strides = 2, activation = \"relu\") |>   layer_conv_2d(filters = 32, kernel_size = 3, activation = \"relu\") |>   layer_max_pooling_2d(pool_size = c(3, 3))  # Can you guess what the current output shape is at this point? Probably not. # Let's just print it: summary(model) ## Model: \"sequential_6\" ## ┏━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━┓ ## ┃ Layer (type)                    ┃ Output Shape           ┃       Param # ┃ ## ┡━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━┩ ## │ conv2d (Conv2D)                 │ (None, 123, 123, 32)   │         2,432 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ conv2d_1 (Conv2D)               │ (None, 121, 121, 32)   │         9,248 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ max_pooling2d (MaxPooling2D)    │ (None, 40, 40, 32)     │             0 │ ## └─────────────────────────────────┴────────────────────────┴───────────────┘ ##  Total params: 11,680 (45.62 KB) ##  Trainable params: 11,680 (45.62 KB) ##  Non-trainable params: 0 (0.00 B) # The answer was: (40, 40, 32), so we can keep downsampling...  model |>   layer_conv_2d(filters = 32, kernel_size = 3, activation = \"relu\") |>   layer_conv_2d(filters = 32, kernel_size = 3, activation = \"relu\") |>   layer_max_pooling_2d(pool_size = 3) |>   layer_conv_2d(filters = 32, kernel_size = 3, activation = \"relu\") |>   layer_conv_2d(filters = 32, kernel_size = 3, activation = \"relu\") |>   layer_max_pooling_2d(pool_size = 2)  # And now? summary(model) ## Model: \"sequential_6\" ## ┏━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━┓ ## ┃ Layer (type)                    ┃ Output Shape           ┃       Param # ┃ ## ┡━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━┩ ## │ conv2d (Conv2D)                 │ (None, 123, 123, 32)   │         2,432 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ conv2d_1 (Conv2D)               │ (None, 121, 121, 32)   │         9,248 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ max_pooling2d (MaxPooling2D)    │ (None, 40, 40, 32)     │             0 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ conv2d_2 (Conv2D)               │ (None, 38, 38, 32)     │         9,248 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ conv2d_3 (Conv2D)               │ (None, 36, 36, 32)     │         9,248 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ max_pooling2d_1 (MaxPooling2D)  │ (None, 12, 12, 32)     │             0 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ conv2d_4 (Conv2D)               │ (None, 10, 10, 32)     │         9,248 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ conv2d_5 (Conv2D)               │ (None, 8, 8, 32)       │         9,248 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ max_pooling2d_2 (MaxPooling2D)  │ (None, 4, 4, 32)       │             0 │ ## └─────────────────────────────────┴────────────────────────┴───────────────┘ ##  Total params: 48,672 (190.12 KB) ##  Trainable params: 48,672 (190.12 KB) ##  Non-trainable params: 0 (0.00 B) # Now that we have 4x4 feature maps, time to apply global max pooling. model |>   layer_global_max_pooling_2d()  # Finally, we add a classification layer. model |>   layer_dense(units = 10, activation = \"softmax\")"},{"path":"https://keras3.posit.co/dev/articles/sequential_model.html","id":"what-to-do-once-you-have-a-model","dir":"Articles","previous_headings":"","what":"What to do once you have a model","title":"The Sequential model","text":"model architecture ready, want : Train model, evaluate , run inference. See guide training & evaluation built-loops Save model disk restore . See guide serialization & saving.","code":""},{"path":"https://keras3.posit.co/dev/articles/sequential_model.html","id":"feature-extraction-with-a-sequential-model","dir":"Articles","previous_headings":"","what":"Feature extraction with a Sequential model","title":"The Sequential model","text":"Sequential model built, behaves like Functional API model. means every layer input output attribute. attributes can used neat things, like quickly creating model extracts outputs intermediate layers Sequential model: ’s similar example extract features one layer:","code":"initial_model <- keras_model_sequential(input_shape = c(250, 250, 3)) |>   layer_conv_2d(filters = 32, kernel_size = 5, strides = 2, activation = \"relu\") |>   layer_conv_2d(filters = 32, kernel_size = 3, activation = \"relu\") |>   layer_conv_2d(filters = 32, kernel_size = 3, activation = \"relu\")   feature_extractor <- keras_model(     inputs = initial_model$inputs,     outputs = lapply(initial_model$layers, function(x) x$output), )  # Call feature extractor on test input. x <- op_ones(c(1, 250, 250, 3)) features <- feature_extractor(x) initial_model <-   keras_model_sequential(input_shape = c(250, 250, 3)) |>   layer_conv_2d(filters = 32, kernel_size = 5, strides = 2,                 activation = \"relu\") |>   layer_conv_2d(filters = 32, kernel_size = 3, activation = \"relu\",                 name = \"my_intermediate_layer\") |>   layer_conv_2d(filters = 32, kernel_size = 3, activation = \"relu\")  feature_extractor <- keras_model(   inputs = initial_model$inputs,   outputs = get_layer(initial_model, \"my_intermediate_layer\")$output, )  # Call feature extractor on test input. x <- op_ones(c(1, 250, 250, 3)) features <- feature_extractor(x)"},{"path":"https://keras3.posit.co/dev/articles/sequential_model.html","id":"transfer-learning-with-a-sequential-model","dir":"Articles","previous_headings":"","what":"Transfer learning with a Sequential model","title":"The Sequential model","text":"Transfer learning consists freezing bottom layers model training top layers. aren’t familiar , make sure read guide transfer learning. two common transfer learning blueprint involving Sequential models. First, let’s say Sequential model, want freeze layers except last one. case, can call freeze_weights(). Alternatively, can iterate model$layers set layer$trainable <- FALSE layer, except last one. Like : Another common blueprint use Sequential model stack pre-trained model freshly initialized classification layers. Like : transfer learning, probably find frequently using two patterns. ’s need know Sequential models! find building models Keras, see: Guide Functional API Guide making new Layers & Models via subclassing","code":"model <- keras_model_sequential(input_shape = 784) |>   layer_dense(units = 32, activation = \"relu\") |>   layer_dense(units = 32, activation = \"relu\") |>   layer_dense(units = 32, activation = \"relu\") |>   layer_dense(units = 10) # Presumably you would want to first load pre-trained weights. model |> load_model_weights(...) # Freeze all layers except the last one. model |> freeze_weights(from = 1, to = -2) model # note the \"Trainable\" column now visible in the summary table ## Model: \"sequential_9\" ## ┏━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━┳━━━━━━━┓ ## ┃ Layer (type)                ┃ Output Shape          ┃    Param # ┃ Trai… ┃ ## ┡━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━╇━━━━━━━┩ ## │ dense_15 (Dense)            │ (None, 32)            │     25,120 │   N   │ ## ├─────────────────────────────┼───────────────────────┼────────────┼───────┤ ## │ dense_16 (Dense)            │ (None, 32)            │      1,056 │   N   │ ## ├─────────────────────────────┼───────────────────────┼────────────┼───────┤ ## │ dense_17 (Dense)            │ (None, 32)            │      1,056 │   N   │ ## ├─────────────────────────────┼───────────────────────┼────────────┼───────┤ ## │ dense_18 (Dense)            │ (None, 10)            │        330 │   Y   │ ## └─────────────────────────────┴───────────────────────┴────────────┴───────┘ ##  Total params: 27,562 (107.66 KB) ##  Trainable params: 330 (1.29 KB) ##  Non-trainable params: 27,232 (106.38 KB) # Another way to freeze all layers except the last one. for (layer in model$layers[-length(model$layers)]) {   layer$trainable <- FALSE }  # Recompile and train (this will only update the weights of the last layer). model |> compile(...) model |> fit(...) # Load a convolutional base with pre-trained weights base_model <- application_xception(weights = 'imagenet',                                    include_top = FALSE,                                    pooling = 'avg')  # Freeze the base model freeze_weights(base_model)  # Use a Sequential model to add a trainable classifier on top model <- keras_model_sequential() |>   base_model() |>   layer_dense(1000) # Compile & train model |> compile(...) model |> fit(...)"},{"path":"https://keras3.posit.co/dev/articles/serialization_and_saving.html","id":"introduction","dir":"Articles","previous_headings":"","what":"Introduction","title":"Save, serialize, and export models","text":"Keras model consists multiple components: architecture, configuration, specifies layers model contain, ’re connected. set weights values (“state model”). optimizer (defined compiling model). set losses metrics (defined compiling model). Keras API saves pieces together unified format, marked .keras extension. zip archive consisting following: JSON-based configuration file (config.json): Records model, layer, trackables’ configuration. H5-based state file, model.weights.h5 (whole model), directory keys layers weights. metadata file JSON, storing things current Keras version. Let’s take look works.","code":""},{"path":"https://keras3.posit.co/dev/articles/serialization_and_saving.html","id":"how-to-save-and-load-a-model","dir":"Articles","previous_headings":"","what":"How to save and load a model","title":"Save, serialize, and export models","text":"10 seconds read guide, ’s need know. Saving Keras model: Loading model back: Now, let’s look details.","code":"# Get model (Sequential, Functional Model, or Model subclass) model <- ...  # The filename needs to end with the .keras extension model |> save_model('path/to/location.keras') model <- load_model('path/to/location.keras')"},{"path":"https://keras3.posit.co/dev/articles/serialization_and_saving.html","id":"setup","dir":"Articles","previous_headings":"","what":"Setup","title":"Save, serialize, and export models","text":"","code":"library(keras3)"},{"path":"https://keras3.posit.co/dev/articles/serialization_and_saving.html","id":"saving","dir":"Articles","previous_headings":"","what":"Saving","title":"Save, serialize, and export models","text":"section saving entire model single file. file include: model’s architecture/config model’s weight values (learned training) model’s compilation information (compile() called) optimizer state, (enables restart training left)","code":""},{"path":"https://keras3.posit.co/dev/articles/serialization_and_saving.html","id":"apis","dir":"Articles","previous_headings":"Saving","what":"APIs","title":"Save, serialize, and export models","text":"can save model save_model(). can load back load_model(). supported format Keras 3 “Keras v3” format, uses .keras extension. Example:","code":"get_model <- function() {   # Create a simple model.   inputs <- keras_input(shape(32))   outputs <- inputs |> layer_dense(1)   model <-  keras_model(inputs, outputs)   model |> compile(optimizer = optimizer_adam(), loss = \"mean_squared_error\")   model }  model <- get_model()  # Train the model. test_input <- random_uniform(c(128, 32)) test_target <- random_uniform(c(128, 1)) model |> fit(test_input, test_target)  # Calling `save('my_model.keras')` creates a zip archive `my_model.keras`. model |> save_model(\"my_model.keras\")  # It can be used to reconstruct the model identically. reconstructed_model <- load_model(\"my_model.keras\")  # Let's check: stopifnot(all.equal(   model |> predict(test_input),   reconstructed_model |> predict(test_input) ))"},{"path":"https://keras3.posit.co/dev/articles/serialization_and_saving.html","id":"custom-objects","dir":"Articles","previous_headings":"Saving","what":"Custom objects","title":"Save, serialize, and export models","text":"section covers basic workflows handling custom layers, functions, models Keras saving reloading. saving model includes custom objects, subclassed Layer, must define get_config() method object class. arguments passed constructor (initialize() method) custom object aren’t simple objects (anything types like ints, strings, etc.), must also explicitly deserialize arguments from_config() class method. Like : Please see Defining config methods section details examples. saved .keras file lightweight store Python code custom objects. Therefore, reload model, load_model requires access definition custom objects used one following methods: Registering custom objects (preferred), Passing custom objects directly loading, Using custom object scope examples workflow:","code":"layer_custom <- Layer(   \"CustomLayer\",   initialize = function(sublayer, ...) {     super$initialize(...)     self$sublayer <- sublayer   },   call = function(x) {     self$sublayer(x)   },   get_config = function() {     base_config <- super$get_config()     config <- list(       sublayer = serialize_keras_object(self$sublayer)     )     c(base_config, config)   },   from_config = function(cls, config) {     sublayer_config <- config$sublayer     sublayer <- deserialize_keras_object(sublayer_config)     cls(sublayer, !!!config)   } )"},{"path":"https://keras3.posit.co/dev/articles/serialization_and_saving.html","id":"registering-custom-objects-preferred","dir":"Articles","previous_headings":"Saving > Custom objects","what":"Registering custom objects (preferred)","title":"Save, serialize, and export models","text":"preferred method, custom object registration greatly simplifies saving loading code. Calling register_keras_serializable() custom object registers object globally master list, allowing Keras recognize object loading model. Let’s create custom model involving custom layer custom activation function demonstrate . Example:","code":"# Clear all previously registered custom objects set_custom_objects(clear = TRUE) ## named list() layer_custom <- Layer(   \"CustomLayer\",   initialize = function(self, factor) {     super$initialize()     self$factor = factor   },    call = function(self, x) {     x * self$factor   },    get_config = function(self) {     list(factor = self$factor)   } )  # Upon registration, you can optionally specify a package or a name. # If left blank, the package defaults to \"Custom\" and the name defaults to # the class name. register_keras_serializable(layer_custom, package = \"MyLayers\")  custom_fn <- keras3:::py_func2(function(x) x^2, name = \"custom_fn\", convert = TRUE)  register_keras_serializable(custom_fn, name=\"custom_fn\", package=\"my_package\")   # Create the model. get_model <- function() {   inputs <- keras_input(shape(4))   mid <- inputs |> layer_custom(0.5)   outputs <- mid |> layer_dense(1, activation = custom_fn)   model <- keras_model(inputs, outputs)   model |> compile(optimizer = \"rmsprop\", loss = \"mean_squared_error\")   model }   # Train the model. train_model <- function(model) {   input <- random_uniform(c(4, 4))   target <- random_uniform(c(4, 1))   model |> fit(input, target, verbose = FALSE, epochs = 1)   model }  test_input <- random_uniform(c(4, 4)) test_target <- random_uniform(c(4, 1))  model <- get_model() |> train_model() model |> save_model(\"custom_model.keras\", overwrite = TRUE)  # Now, we can simply load without worrying about our custom objects. reconstructed_model <- load_model(\"custom_model.keras\")  # Let's check: stopifnot(all.equal(   model |> predict(test_input, verbose = FALSE),   reconstructed_model |> predict(test_input, verbose = FALSE) ))"},{"path":"https://keras3.posit.co/dev/articles/serialization_and_saving.html","id":"passing-custom-objects-to-load_model","dir":"Articles","previous_headings":"Saving > Custom objects","what":"Passing custom objects to load_model()","title":"Save, serialize, and export models","text":"","code":"model <- get_model() |> train_model()  # Calling `save_model('my_model.keras')` creates a zip archive `my_model.keras`. model |> save_model(\"custom_model.keras\", overwrite = TRUE)  # Upon loading, pass a named list containing the custom objects used in the # `custom_objects` argument of `load_model()`. reconstructed_model <-  load_model(  \"custom_model.keras\",   custom_objects = list(CustomLayer = layer_custom,                         custom_fn = custom_fn), )  # Let's check: stopifnot(all.equal(   model |> predict(test_input, verbose = FALSE),   reconstructed_model |> predict(test_input, verbose = FALSE) ))"},{"path":"https://keras3.posit.co/dev/articles/serialization_and_saving.html","id":"using-a-custom-object-scope","dir":"Articles","previous_headings":"Saving > Custom objects","what":"Using a custom object scope","title":"Save, serialize, and export models","text":"code within custom object scope able recognize custom objects passed scope argument. Therefore, loading model within scope allow loading custom objects. Example:","code":"model <- get_model() |> train_model() model |> save_model(\"custom_model.keras\", overwrite = TRUE)  # Pass the custom objects dictionary to a custom object scope and place # the `keras.models.load_model()` call within the scope. custom_objects <- list(CustomLayer = layer_custom, custom_fn = custom_fn)  with_custom_object_scope(custom_objects, {   reconstructed_model <- load_model(\"custom_model.keras\") })  # Let's check: stopifnot(all.equal(   model |> predict(test_input, verbose = FALSE),   reconstructed_model |> predict(test_input, verbose = FALSE) ))"},{"path":"https://keras3.posit.co/dev/articles/serialization_and_saving.html","id":"model-serialization","dir":"Articles","previous_headings":"Saving","what":"Model serialization","title":"Save, serialize, and export models","text":"section saving model’s configuration, without state. model’s configuration (architecture) specifies layers model contains, layers connected. configuration model, model can created freshly initialized state (weights compilation information).","code":""},{"path":"https://keras3.posit.co/dev/articles/serialization_and_saving.html","id":"apis-1","dir":"Articles","previous_headings":"Saving > Model serialization","what":"APIs","title":"Save, serialize, and export models","text":"following serialization APIs available: clone_model(model): make (randomly initialized) copy model. get_config() cls.from_config(): retrieve configuration layer model, recreate model instance config, respectively. keras.models.model_to_json() keras.models.model_from_json(): similar, JSON strings. keras.saving.serialize_keras_object(): retrieve configuration arbitrary Keras object. keras.saving.deserialize_keras_object(): recreate object instance configuration.","code":""},{"path":"https://keras3.posit.co/dev/articles/serialization_and_saving.html","id":"in-memory-model-cloning","dir":"Articles","previous_headings":"Saving > Model serialization","what":"In-memory model cloning","title":"Save, serialize, and export models","text":"can -memory cloning model via clone_model(). equivalent getting config recreating model config (preserve compilation information layer weights values). Example:","code":"new_model <- clone_model(model)"},{"path":"https://keras3.posit.co/dev/articles/serialization_and_saving.html","id":"get_config-and-from_config","dir":"Articles","previous_headings":"Saving > Model serialization","what":"get_config() and from_config()","title":"Save, serialize, and export models","text":"Calling get_config(model) get_config(layer) return named list containing configuration model layer, respectively. define get_config() contain arguments needed initialize() method model layer. loading time, from_config(config) method call initialize() arguments reconstruct model layer. Layer example: Now let’s reconstruct layer using from_config() method: Sequential model example: Functional model example:","code":"layer <- layer_dense(, 3, activation=\"relu\") layer_config <- get_config(layer) str(layer_config) ## List of 12 ##  $ name              : chr \"dense_4\" ##  $ trainable         : logi TRUE ##  $ dtype             :List of 4 ##   ..$ module         : chr \"keras\" ##   ..$ class_name     : chr \"DTypePolicy\" ##   ..$ config         :List of 1 ##   .. ..$ name: chr \"float32\" ##   ..$ registered_name: NULL ##  $ units             : int 3 ##  $ activation        : chr \"relu\" ##  $ use_bias          : logi TRUE ##  $ kernel_initializer:List of 4 ##   ..$ module         : chr \"keras.initializers\" ##   ..$ class_name     : chr \"GlorotUniform\" ##   ..$ config         :List of 1 ##   .. ..$ seed: NULL ##   ..$ registered_name: NULL ##  $ bias_initializer  :List of 4 ##   ..$ module         : chr \"keras.initializers\" ##   ..$ class_name     : chr \"Zeros\" ##   ..$ config         : Named list() ##   ..$ registered_name: NULL ##  $ kernel_regularizer: NULL ##  $ bias_regularizer  : NULL ##  $ kernel_constraint : NULL ##  $ bias_constraint   : NULL ##  - attr(*, \"__class__\")=<class 'keras.src.layers.core.dense.Dense'> new_layer <- from_config(layer_config) model <- keras_model_sequential(input_shape = c(32)) |>   layer_dense(1) config <- get_config(model) new_model <- from_config(config) inputs <- keras_input(c(32)) outputs <- inputs |> layer_dense(1) model <- keras_model(inputs, outputs) config <- get_config(model) new_model <- from_config(config)"},{"path":"https://keras3.posit.co/dev/articles/serialization_and_saving.html","id":"save_model_config-and-load_model_config","dir":"Articles","previous_headings":"Saving > Model serialization","what":"save_model_config() and load_model_config()","title":"Save, serialize, and export models","text":"similar get_config / from_config, except turns model JSON file, can loaded without original model class. also specific models, isn’t meant layers. Example:","code":"model <- keras_model_sequential(input_shape = c(32)) |>   layer_dense(1) save_model_config(model, \"model_config.json\") new_model <- load_model_config(\"model_config.json\") unlink(\"model_config.json\")"},{"path":"https://keras3.posit.co/dev/articles/serialization_and_saving.html","id":"arbitrary-object-serialization-and-deserialization","dir":"Articles","previous_headings":"Saving > Model serialization","what":"Arbitrary object serialization and deserialization","title":"Save, serialize, and export models","text":"serialize_keras_object() deserialize_keras_object() APIs general-purpose APIs can used serialize deserialize Keras object custom object. foundation saving model architecture behind serialize()/deserialize() calls keras. Example: Note serialization format containing necessary information proper reconstruction: module containing name Keras module identifying module object comes class_name containing name object’s class. config information needed reconstruct object registered_name custom objects. See . Now can reconstruct regularizer.","code":"my_reg <- regularizer_l1(0.005) config <- serialize_keras_object(my_reg) str(config) ## List of 4 ##  $ module         : chr \"keras.regularizers\" ##  $ class_name     : chr \"L1\" ##  $ config         :List of 1 ##   ..$ l1: num 0.005 ##  $ registered_name: NULL new_reg <- deserialize_keras_object(config) new_reg ## <keras.src.regularizers.regularizers.L1 object> ##  signature: (x)"},{"path":"https://keras3.posit.co/dev/articles/serialization_and_saving.html","id":"model-weights-saving","dir":"Articles","previous_headings":"Saving","what":"Model weights saving","title":"Save, serialize, and export models","text":"can choose save & load model’s weights. can useful : need model inference: case won’t need restart training, don’t need compilation information optimizer state. transfer learning: case training new model reusing state prior model, don’t need compilation information prior model.","code":""},{"path":"https://keras3.posit.co/dev/articles/serialization_and_saving.html","id":"apis-for-in-memory-weight-transfer","dir":"Articles","previous_headings":"Saving > Model weights saving","what":"APIs for in-memory weight transfer","title":"Save, serialize, and export models","text":"Weights can copied different objects using get_weights() set_weights(): get_weights(<layer>): Returns list arrays weight values. set_weights(<layer>weights): Sets model/layer weights values provided (arrays). Examples: Transferring weights one layer another, memory Transferring weights one model another model compatible architecture, memory case stateless layers stateless layers change order number weights, models can compatible architectures even extra/missing stateless layers.","code":"create_layer <- function() {   layer <- layer_dense(, 64, activation = \"relu\", name = \"dense_2\")   layer$build(shape(NA, 784))   layer }  layer_1 <- create_layer() layer_2 <- create_layer()  # Copy weights from layer 1 to layer 2 layer_2 |> set_weights(get_weights(layer_1)) # Create a simple functional model inputs <- keras_input(shape=c(784), name=\"digits\") outputs <- inputs |>   layer_dense(64, activation = \"relu\", name = \"dense_1\") |>   layer_dense(64, activation = \"relu\", name = \"dense_2\") |>   layer_dense(10, name = \"predictions\") functional_model <- keras_model(inputs = inputs, outputs = outputs,                                name = \"3_layer_mlp\")  # Define a subclassed model with the same architecture SubclassedModel <- new_model_class(   \"SubclassedModel\",   initialize = function(output_dim, name = NULL) {     super$initialize(name = name)     self$output_dim <- output_dim |> as.integer()     self$dense_1 <- layer_dense(, 64, activation = \"relu\",                                 name = \"dense_1\")     self$dense_2 <- layer_dense(, 64, activation = \"relu\",                                 name = \"dense_2\")     self$dense_3 <- layer_dense(, self$output_dim,                                 name = \"predictions\")   },    call = function(inputs) {     inputs |>       self$dense_1() |>       self$dense_2() |>       self$dense_3()   },    get_config = function(self) {     list(output_dim = self$output_dim,          name = self$name)   } )   subclassed_model <- SubclassedModel(10) # Call the subclassed model once to create the weights. subclassed_model(op_ones(c(1, 784))) |> invisible()  # Copy weights from functional_model to subclassed_model. set_weights(subclassed_model, get_weights(functional_model))  stopifnot(all.equal(   get_weights(functional_model),   get_weights(subclassed_model) )) input <- keras_input(shape = c(784), name = \"digits\") output <- input |>   layer_dense(64, activation = \"relu\", name = \"dense_1\") |>   layer_dense(64, activation = \"relu\", name = \"dense_2\") |>   layer_dense(10, name = \"predictions\") functional_model <- keras_model(inputs, outputs,                                 name = \"3_layer_mlp\")  input <- keras_input(shape = c(784), name = \"digits\") output <- input |>   layer_dense(64, activation = \"relu\", name = \"dense_1\") |>   layer_dense(64, activation = \"relu\", name = \"dense_2\") |>   # Add a dropout layer, which does not contain any weights.   layer_dropout(0.5) |>   layer_dense(10, name = \"predictions\")  functional_model_with_dropout <-   keras_model(input, output, name = \"3_layer_mlp\")  set_weights(functional_model_with_dropout,             get_weights(functional_model))"},{"path":"https://keras3.posit.co/dev/articles/serialization_and_saving.html","id":"apis-for-saving-weights-to-disk-loading-them-back","dir":"Articles","previous_headings":"Saving > Model weights saving","what":"APIs for saving weights to disk & loading them back","title":"Save, serialize, and export models","text":"Weights can saved disk calling save_model_weights(filepath). filename end .weights.h5. Example: Note using freeze_weights() may result different output get_weights(layer) ordering model contains nested layers.","code":"sequential_model = keras_model_sequential(input_shape = c(784),                                           input_name = \"digits\") |>   layer_dense(64, activation = \"relu\", name = \"dense_1\") |>   layer_dense(64, activation = \"relu\", name = \"dense_2\") |>   layer_dense(10, name = \"predictions\") sequential_model |> save_model_weights(\"my_model.weights.h5\") sequential_model |> load_model_weights(\"my_model.weights.h5\")"},{"path":"https://keras3.posit.co/dev/articles/serialization_and_saving.html","id":"transfer-learning-example","dir":"Articles","previous_headings":"Saving > Model weights saving > APIs for saving weights to disk & loading them back","what":"Transfer learning example","title":"Save, serialize, and export models","text":"loading pretrained weights weights file, recommended load weights original checkpointed model, extract desired weights/layers new model. Example:","code":"create_functional_model <- function() {   inputs <- keras_input(shape = c(784), name = \"digits\")   outputs <- inputs |>     layer_dense(64, activation = \"relu\", name = \"dense_1\") |>     layer_dense(64, activation = \"relu\", name = \"dense_2\") |>     layer_dense(10, name = \"predictions\")   keras_model(inputs, outputs, name = \"3_layer_mlp\") }  functional_model <- create_functional_model() functional_model |> save_model_weights(\"pretrained.weights.h5\")  # In a separate program: pretrained_model <- create_functional_model() pretrained_model |> load_model_weights(\"pretrained.weights.h5\")  # Create a new model by extracting layers from the original model: extracted_layers <- pretrained_model$layers |> head(-1) model <- keras_model_sequential(layers = extracted_layers) |>   layer_dense(5, name = \"dense_3\") summary(model) ## Model: \"sequential_4\" ## ┏━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━┓ ## ┃ Layer (type)                    ┃ Output Shape           ┃       Param # ┃ ## ┡━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━┩ ## │ dense_1 (Dense)                 │ (None, 64)             │        50,240 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ dense_2 (Dense)                 │ (None, 64)             │         4,160 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ dense_3 (Dense)                 │ (None, 5)              │           325 │ ## └─────────────────────────────────┴────────────────────────┴───────────────┘ ##  Total params: 54,725 (213.77 KB) ##  Trainable params: 54,725 (213.77 KB) ##  Non-trainable params: 0 (0.00 B)"},{"path":[]},{"path":"https://keras3.posit.co/dev/articles/serialization_and_saving.html","id":"defining-the-config-methods","dir":"Articles","previous_headings":"Saving > Appendix: Handling custom objects","what":"Defining the config methods","title":"Save, serialize, and export models","text":"Specifications: get_config() return JSON-serializable named list order compatible Keras architecture model-saving APIs. from_config(config) (class method) return new layer model object created config. default implementation returns .call(cls, config). NOTE: constructor arguments already serializable, e.g. strings ints, non-custom Keras objects, overriding from_config() necessary. However, complex objects layers models passed initialize(), deserialization must handled explicitly either initialize overriding from_config() method. Example: Note overriding from_config unnecessary MyDense hidden_units, kernel_initializer, kernel_regularizer ints, strings, built-Keras object, respectively. means default from_config implementation cls(!!!config) work intended. complex objects, layers models passed initialize(), example, must explicitly deserialize objects. Let’s take look example model from_config override necessary. Example:","code":"layer_my_dense <- register_keras_serializable(   package = \"MyLayers\", name = \"KernelMult\",   object = Layer(     \"MyDense\",     initialize = function(units,                           ...,                           kernel_regularizer = NULL,                           kernel_initializer = NULL,                           nested_model = NULL) {       super$initialize(...)       self$hidden_units <- units       self$kernel_regularizer <- kernel_regularizer       self$kernel_initializer <- kernel_initializer       self$nested_model <- nested_model     },     get_config = function() {       config <- super$get_config()       # Update the config with the custom layer's parameters       config <- modifyList(config, list(         units = self$hidden_units,         kernel_regularizer = self$kernel_regularizer,         kernel_initializer = self$kernel_initializer,         nested_model = self$nested_model       ))       config     },     build = function(input_shape) {       input_units <- tail(input_shape, 1)       self$kernel <- self$add_weight(         name = \"kernel\",         shape = shape(input_units, self$hidden_units),         regularizer = self$kernel_regularizer,         initializer = self$kernel_initializer,       )     },     call = function(inputs) {       op_matmul(inputs, self$kernel)     }   ) )   layer <- layer_my_dense(units = 16,                         kernel_regularizer = \"l1\",                         kernel_initializer = \"ones\") layer3 <- layer_my_dense(units = 64, nested_model = layer)  config <- serialize_keras_object(layer3) str(config) ## List of 4 ##  $ module         : chr \"<r-globalenv>\" ##  $ class_name     : chr \"MyDense\" ##  $ config         :List of 5 ##   ..$ name        : chr \"my_dense_1\" ##   ..$ trainable   : logi TRUE ##   ..$ dtype       :List of 4 ##   .. ..$ module         : chr \"keras\" ##   .. ..$ class_name     : chr \"DTypePolicy\" ##   .. ..$ config         :List of 1 ##   .. .. ..$ name: chr \"float32\" ##   .. ..$ registered_name: NULL ##   ..$ units       : num 64 ##   ..$ nested_model:List of 4 ##   .. ..$ module         : chr \"<r-globalenv>\" ##   .. ..$ class_name     : chr \"MyDense\" ##   .. ..$ config         :List of 6 ##   .. .. ..$ name              : chr \"my_dense\" ##   .. .. ..$ trainable         : logi TRUE ##   .. .. ..$ dtype             :List of 4 ##   .. .. .. ..$ module         : chr \"keras\" ##   .. .. .. ..$ class_name     : chr \"DTypePolicy\" ##   .. .. .. ..$ config         :List of 1 ##   .. .. .. .. ..$ name: chr \"float32\" ##   .. .. .. ..$ registered_name: NULL ##   .. .. ..$ units             : num 16 ##   .. .. ..$ kernel_regularizer: chr \"l1\" ##   .. .. ..$ kernel_initializer: chr \"ones\" ##   .. ..$ registered_name: chr \"MyLayers>KernelMult\" ##  $ registered_name: chr \"MyLayers>KernelMult\" new_layer <- deserialize_keras_object(config) new_layer ## <MyDense name=my_dense_1, built=False> ##  signature: (*args, **kwargs) `%||%` <- \\(x, y) if(is.null(x)) y else x layer_custom_model <- register_keras_serializable(   package = \"ComplexModels\",   object = Layer(     \"CustomModel\",     initialize = function(first_layer, second_layer = NULL, ...) {       super$initialize(...)       self$first_layer <- first_layer       self$second_layer <- second_layer %||% layer_dense(, 8)     },      get_config = function() {       config <- super$get_config()       config <- modifyList(config, list(         first_layer = self$first_layer,         second_layer = self$second_layer       ))       config     },      from_config = function(config) {       config$first_layer %<>% deserialize_keras_object()       config$second_layer %<>% deserialize_keras_object()       # note that the class is available in methods under the classname symbol,       # (`CustomModel` for this class), and also under the symbol `__class__`       cls(!!!config)       # CustomModel(!!!config)     },     call = function(self, inputs) {       inputs |>         self$first_layer() |>         self$second_layer()     }   ) )  # Let's make our first layer the custom layer from the previous example (MyDense) inputs <- keras_input(c(32)) outputs <-  inputs |> layer_custom_model(first_layer=layer) model <- keras_model(inputs, outputs)  config <- get_config(model) new_model <- from_config(config)"},{"path":"https://keras3.posit.co/dev/articles/serialization_and_saving.html","id":"how-custom-objects-are-serialized","dir":"Articles","previous_headings":"Saving > Appendix: Handling custom objects","what":"How custom objects are serialized","title":"Save, serialize, and export models","text":"serialization format special key custom objects registered via register_keras_serializable(). registered_name key allows easy retrieval loading/deserialization time also allowing users add custom naming. Let’s take look config serializing custom layer MyDense defined . Example: shown, registered_name key contains lookup information Keras master list, including package MyLayers custom name KernelMult gave calling register_keras_serializables(). Take look custom class definition/registration . Note class_name key contains original name class, allowing proper re-initialization from_config. Additionally, note module key NULL since custom object.","code":"layer <- layer_my_dense(   units = 16,   kernel_regularizer = regularizer_l1_l2(l1 = 1e-5, l2 = 1e-4),   kernel_initializer = \"ones\", ) config <- serialize_keras_object(layer) str(config) ## List of 4 ##  $ module         : chr \"<r-globalenv>\" ##  $ class_name     : chr \"MyDense\" ##  $ config         :List of 6 ##   ..$ name              : chr \"my_dense_2\" ##   ..$ trainable         : logi TRUE ##   ..$ dtype             :List of 4 ##   .. ..$ module         : chr \"keras\" ##   .. ..$ class_name     : chr \"DTypePolicy\" ##   .. ..$ config         :List of 1 ##   .. .. ..$ name: chr \"float32\" ##   .. ..$ registered_name: NULL ##   ..$ units             : num 16 ##   ..$ kernel_regularizer:List of 4 ##   .. ..$ module         : chr \"keras.regularizers\" ##   .. ..$ class_name     : chr \"L1L2\" ##   .. ..$ config         :List of 2 ##   .. .. ..$ l1: num 1e-05 ##   .. .. ..$ l2: num 1e-04 ##   .. ..$ registered_name: NULL ##   ..$ kernel_initializer: chr \"ones\" ##  $ registered_name: chr \"MyLayers>KernelMult\""},{"path":"https://keras3.posit.co/dev/articles/training_with_built_in_methods.html","id":"setup","dir":"Articles","previous_headings":"","what":"Setup","title":"Training & evaluation with the built-in methods","text":"","code":"library(keras3)"},{"path":"https://keras3.posit.co/dev/articles/training_with_built_in_methods.html","id":"introduction","dir":"Articles","previous_headings":"","what":"Introduction","title":"Training & evaluation with the built-in methods","text":"guide covers training, evaluation, prediction (inference) models using built-APIs training & validation (fit(), evaluate() predict()). interested leveraging fit() specifying training step function, see Customizing happens fit() guide. interested writing training & evaluation loops scratch, see guide Writing training loop scratch. general, whether using built-loops writing , model training & evaluation works strictly way across every kind Keras model – Sequential models, models built Functional API, models written scratch via model subclassing.","code":""},{"path":"https://keras3.posit.co/dev/articles/training_with_built_in_methods.html","id":"api-overview-a-first-end-to-end-example","dir":"Articles","previous_headings":"","what":"API overview: a first end-to-end example","title":"Training & evaluation with the built-in methods","text":"passing data built-training loops model, either use: Arrays (data small fits memory) tf_dataset objects PyTorch DataLoader instances next paragraphs, ’ll use MNIST dataset NumPy arrays, order demonstrate use optimizers, losses, metrics. Afterwards, ’ll take close look options. Let’s consider following model (, build Functional API, Sequential model subclassed model well): ’s typical end--end workflow looks like, consisting : Training Validation holdout set generated original training data Evaluation test data ’ll use MNIST data example. specify training configuration (optimizer, loss, metrics): call fit(), train model slicing data “batches” size batch_size, repeatedly iterating entire dataset given number epochs. returned history object holds record loss values metric values training: evaluate model test data via evaluate(): Now, let’s review piece workflow detail.","code":"inputs <- keras_input(shape = 784, name=\"digits\") outputs <- inputs |>   layer_dense(units = 64, activation = \"relu\", name = \"dense_1\") |>   layer_dense(units = 64, activation = \"relu\", name = \"dense_2\") |>   layer_dense(units = 10, activation = \"softmax\", name = \"predictions\") model <- keras_model(inputs = inputs, outputs = outputs) summary(model) ## Model: \"functional\" ## ┏━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━┓ ## ┃ Layer (type)                    ┃ Output Shape           ┃       Param # ┃ ## ┡━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━┩ ## │ digits (InputLayer)             │ (None, 784)            │             0 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ dense_1 (Dense)                 │ (None, 64)             │        50,240 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ dense_2 (Dense)                 │ (None, 64)             │         4,160 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ predictions (Dense)             │ (None, 10)             │           650 │ ## └─────────────────────────────────┴────────────────────────┴───────────────┘ ##  Total params: 55,050 (215.04 KB) ##  Trainable params: 55,050 (215.04 KB) ##  Non-trainable params: 0 (0.00 B) c(c(x_train, y_train), c(x_test, y_test)) %<-% dataset_mnist()  # Preprocess the data (these are NumPy arrays) x_train <- array_reshape(x_train, c(60000, 784)) / 255 x_test <- array_reshape(x_test, c(10000, 784)) / 255  # Reserve 10,000 samples for validation x_val <- x_train[1:10000,] y_val <- y_train[1:10000] x_train <- x_train[-c(1:10000),] y_train <- y_train[-c(1:10000)] model |> compile(   # Optimizer   optimizer = optimizer_rmsprop(),   # Loss function to minimize   loss = loss_sparse_categorical_crossentropy(),   # List of metrics to monitor   metrics = list(metric_sparse_categorical_accuracy()) ) history <- model |> fit(   x_train, y_train,   batch_size = 64,   epochs = 2,   # We pass some validation for   # monitoring validation loss and metrics   # at the end of each epoch   validation_data = list(x_val, y_val) ) ## Epoch 1/2 ## 782/782 - 3s - 3ms/step - loss: 0.3410 - sparse_categorical_accuracy: 0.9034 - val_loss: 0.1855 - val_sparse_categorical_accuracy: 0.9460 ## Epoch 2/2 ## 782/782 - 1s - 1ms/step - loss: 0.1590 - sparse_categorical_accuracy: 0.9538 - val_loss: 0.1323 - val_sparse_categorical_accuracy: 0.9621 history ## ## Final epoch (plot to see history): ##                            loss: 0.159 ##     sparse_categorical_accuracy: 0.9538 ##                        val_loss: 0.1323 ## val_sparse_categorical_accuracy: 0.9621 # Evaluate the model on the test data using `evaluate` results <- model |> evaluate(x_test, y_test, batch_size=128) ## 79/79 - 0s - 4ms/step - loss: 0.1273 - sparse_categorical_accuracy: 0.9621 str(results) ## List of 2 ##  $ loss                       : num 0.127 ##  $ sparse_categorical_accuracy: num 0.962 # Generate predictions (probabilities -- the output of the last layer) # on new data using `predict` predictions <- model |> predict(x_test[1:2,]) ## 1/1 - 0s - 172ms/step dim(predictions) ## [1]  2 10"},{"path":"https://keras3.posit.co/dev/articles/training_with_built_in_methods.html","id":"the-compile-method-specifying-a-loss-metrics-and-an-optimizer","dir":"Articles","previous_headings":"","what":"The compile() method: specifying a loss, metrics, and an optimizer","title":"Training & evaluation with the built-in methods","text":"train model fit(), need specify loss function, optimizer, optionally, metrics monitor. pass model arguments compile() method: metrics argument list – model can number metrics. model multiple outputs, can specify different losses metrics output, can modulate contribution output total loss model. find details Passing data multi-input, multi-output models section. Note ’re satisfied default settings, many cases optimizer, loss, metrics can specified via string identifiers shortcut: later reuse, let’s put model definition compile step functions; call several times across different examples guide.","code":"model |> compile(   optimizer = optimizer_rmsprop(learning_rate = 1e-3),   loss = loss_sparse_categorical_crossentropy(),   metrics = list(metric_sparse_categorical_accuracy()) ) model |> compile(   optimizer = \"rmsprop\",   loss = \"sparse_categorical_crossentropy\",   metrics = c(\"sparse_categorical_accuracy\") ) get_uncompiled_model <- function() {   inputs <- keras_input(shape = 784, name = \"digits\")   outputs <- inputs |>     layer_dense(units = 64, activation = \"relu\", name = \"dense_1\") |>     layer_dense(units = 64, activation = \"relu\", name = \"dense_2\") |>     layer_dense(units = 10, activation = \"softmax\", name = \"predictions\")   keras_model(inputs = inputs, outputs = outputs) }  get_compiled_model <- function() {   model <- get_uncompiled_model()   model |> compile(     optimizer = \"rmsprop\",     loss = \"sparse_categorical_crossentropy\",     metrics = c(\"sparse_categorical_accuracy\")   )   model }"},{"path":"https://keras3.posit.co/dev/articles/training_with_built_in_methods.html","id":"many-built-in-optimizers-losses-and-metrics-are-available","dir":"Articles","previous_headings":"The compile() method: specifying a loss, metrics, and an optimizer","what":"Many built-in optimizers, losses, and metrics are available","title":"Training & evaluation with the built-in methods","text":"general, won’t create losses, metrics, optimizers scratch, need likely already part Keras API: Optimizers: [optimizer_sgd()] (without momentum) [optimizer_rmsprop()] [optimizer_adam()] etc. Losses: [loss_mean_squared_error()] [loss_kl_divergence()] [loss_cosine_similarity()] etc. Metrics: [metric_auc()] [metric_precision()] [metric_recall()] etc.","code":""},{"path":"https://keras3.posit.co/dev/articles/training_with_built_in_methods.html","id":"custom-losses","dir":"Articles","previous_headings":"The compile() method: specifying a loss, metrics, and an optimizer","what":"Custom losses","title":"Training & evaluation with the built-in methods","text":"need create custom loss, Keras provides three ways . first method involves creating function accepts inputs y_true y_pred. following example shows loss function computes mean squared error real data predictions: need loss function takes parameters beside y_true y_pred, can subclass Keras base Loss class using [Loss()] implement following two methods: initialize(): accept parameters pass call loss function call(y_true, y_pred): use targets (y_true) model predictions (y_pred) compute model’s loss Let’s say want use mean squared error, added term de-incentivize prediction values far 0.5 (assume categorical targets one-hot encoded take values 0 1). creates incentive model confident, may help reduce overfitting (won’t know works try!). ’s :","code":"custom_mean_squared_error <- function(y_true, y_pred) {   op_mean(op_square(y_true - y_pred), axis = -1) }  model <- get_uncompiled_model() model |> compile(optimizer = \"adam\",                  loss = custom_mean_squared_error)  # We need to one-hot encode the labels to use MSE y_train_one_hot <- op_one_hot(y_train, num_classes = 10) model |> fit(x_train, y_train_one_hot, batch_size = 64, epochs = 2) ## Epoch 1/2 ## 782/782 - 2s - 2ms/step - loss: 0.0157 ## Epoch 2/2 ## 782/782 - 1s - 1ms/step - loss: 0.0074 loss_custom_mse <- Loss(   classname = \"CustomMSE\",   initialize = function(regularization_factor = 0.1, name = \"custom_mse\") {     super$initialize(name = name)     self$regularization_factor <- regularization_factor   },   call = function(y_true, y_pred) {     mse <- op_mean(op_square(y_true - y_pred), axis = -1)     reg <- op_mean(op_square(0.5 - y_pred), axis = -1)     mse + reg * self$regularization_factor   } )  model <- get_uncompiled_model() model |> compile(optimizer=\"adam\", loss = loss_custom_mse())  y_train_one_hot <- op_one_hot(y_train, num_classes=10) model |> fit(x_train, y_train_one_hot, batch_size=64, epochs=1) ## 782/782 - 2s - 2ms/step - loss: 0.0388"},{"path":"https://keras3.posit.co/dev/articles/training_with_built_in_methods.html","id":"custom-metrics","dir":"Articles","previous_headings":"The compile() method: specifying a loss, metrics, and an optimizer","what":"Custom metrics","title":"Training & evaluation with the built-in methods","text":"need metric isn’t part API, can easily create custom metrics subclassing Keras base Metric class using [Metric()]. need implement 4 methods: initialize(), create state variables metric. update_state(y_true, y_pred, sample_weight = NULL), uses targets y_true model predictions y_pred update state variables. result(), uses state variables compute final results. reset_state(), reinitializes state metric. State update results computation kept separate (update_state() result(), respectively) cases, results computation might expensive done periodically. ’s simple example showing implement CategoricalTruePositives metric counts many samples correctly classified belonging given class:","code":"metric_categorical_true_positives <- Metric(   \"CategoricalTruePositives\",    initialize = function(name = \"categorical_true_positives\", ...) {     super$initialize(name = name, ...)     self$true_positives <- self$add_variable(shape = shape(),                                              name = \"ctp\",                                              initializer = \"zeros\")   },    update_state = function(y_true, y_pred, sample_weight = NULL) {     y_pred <- op_argmax(y_pred, axis = 2) |> op_reshape(c(-1, 1))     values <- op_cast(y_true, \"int32\") == op_cast(y_pred, \"int32\")     values <- op_cast(values, \"float32\")     if (!is.null(sample_weight)) {       sample_weight <- op_cast(sample_weight, \"float32\")       values <- op_multiply(values, sample_weight)     }     self$true_positives$assign_add(op_sum(values))   },    result = function() {     self$true_positives$value   },    reset_state = function() {     self$true_positives$assign(0.0)   } )  model <- get_uncompiled_model() model |> compile(   optimizer = optimizer_rmsprop(learning_rate = 1e-3),   loss = loss_sparse_categorical_crossentropy(),   metrics = c(metric_categorical_true_positives()) ) history <- model |> fit(x_train, y_train, batch_size = 64, epochs = 3) ## Epoch 1/3 ## 782/782 - 2s - 2ms/step - categorical_true_positives: 285017.0000 - loss: 0.3444 ## Epoch 2/3 ## 782/782 - 1s - 1ms/step - categorical_true_positives: 284268.0000 - loss: 0.1658 ## Epoch 3/3 ## 782/782 - 1s - 973us/step - categorical_true_positives: 284173.0000 - loss: 0.1207"},{"path":"https://keras3.posit.co/dev/articles/training_with_built_in_methods.html","id":"handling-losses-and-metrics-that-dont-fit-the-standard-signature","dir":"Articles","previous_headings":"The compile() method: specifying a loss, metrics, and an optimizer","what":"Handling losses and metrics that don’t fit the standard signature","title":"Training & evaluation with the built-in methods","text":"overwhelming majority losses metrics can computed y_true y_pred, y_pred output model – . instance, regularization loss may require activation layer (targets case), activation may model output. cases, can call self$add_loss(loss_value) inside call method custom layer. Losses added way get added “main” loss training (one passed compile()). ’s simple example adds activity regularization (note activity regularization built-Keras layers – layer just sake providing concrete example): Note pass losses via add_loss(), becomes possible call compile() without loss function, since model already loss minimize. Consider following LogisticEndpoint layer: takes inputs targets & logits, tracks crossentropy loss via add_loss(). can use model two inputs (input data & targets), compiled without loss argument, like : information training multi-input models, see section Passing data multi-input, multi-output models.","code":"layer_custom_activity_regularizer <- Layer(   \"ActivityRegularization\",   call = function(inputs) {     self$add_loss(op_sum(inputs) * 0.1)     inputs  # Pass-through layer.   } )  inputs <- keras_input(shape = 784, name = \"digits\") outputs <- inputs |>   layer_dense(units = 32, activation = \"relu\", name = \"dense_1\") |>   layer_custom_activity_regularizer() |>   layer_dense(units = 64, activation = \"relu\", name = \"dense_2\") |>   layer_dense(units = 10, name = \"predictions\")  model <- keras_model(inputs = inputs, outputs = outputs) model |> compile(optimizer = optimizer_rmsprop(learning_rate = 1e-3),                  loss = loss_sparse_categorical_crossentropy(from_logits = TRUE))  # The displayed loss will be much higher than before # due to the regularization component. model |> fit(x_train, y_train, batch_size = 64, epochs = 1) ## 782/782 - 2s - 2ms/step - loss: 2.3721 layer_logistic_endpoint <- Layer(   \"LogisticEndpoint\",   initialize = function(name = NULL) {     super$initialize(name = name)     self$loss_fn <- loss_binary_crossentropy(from_logits = TRUE)   },   call = function(targets, logits, sample_weights = NULL) {     # Compute the training-time loss value and add it     # to the layer using `self.add_loss()`.     loss <- self$loss_fn(targets, logits, sample_weights)     self$add_loss(loss)      # Return the inference-time prediction tensor (for `predict()`).     op_softmax(logits)   } ) inputs <- keras_input(shape = 3, name = \"inputs\") targets <- keras_input(shape = 10, name = \"targets\")  logits <- inputs |> layer_dense(10) predictions <- layer_logistic_endpoint(name = \"predictions\")(targets, logits)  model <- keras_model(inputs = list(inputs, targets),                      outputs = predictions) model |> compile(optimizer = \"adam\")  # No loss argument!  data <- list(   inputs = random_normal(c(3, 3)),   targets = random_normal(c(3, 10)) ) model |> fit(data, epochs = 1) ## 1/1 - 0s - 439ms/step - loss: 0.9638"},{"path":"https://keras3.posit.co/dev/articles/training_with_built_in_methods.html","id":"automatically-setting-apart-a-validation-holdout-set","dir":"Articles","previous_headings":"The compile() method: specifying a loss, metrics, and an optimizer","what":"Automatically setting apart a validation holdout set","title":"Training & evaluation with the built-in methods","text":"first end--end example saw, used validation_data argument pass list arrays list(x_val, y_val) model evaluating validation loss validation metrics end epoch. ’s another option: argument validation_split allows automatically reserve part training data validation. argument value represents fraction data reserved validation, set number higher 0 lower 1. instance, validation_split = 0.2 means “use 20% data validation”, validation_split = 0.6 means “use 60% data validation”. way validation computed taking last x% samples arrays received fit() call, shuffling. Note can use validation_split training NumPy data.","code":"model <- get_compiled_model() model |> fit(x_train, y_train,              batch_size = 64,              validation_split = 0.2, epochs = 1) ## 625/625 - 2s - 3ms/step - loss: 0.3817 - sparse_categorical_accuracy: 0.8919 - val_loss: 0.1953 - val_sparse_categorical_accuracy: 0.9431"},{"path":"https://keras3.posit.co/dev/articles/training_with_built_in_methods.html","id":"training-evaluation-using-tf-dataset-objects","dir":"Articles","previous_headings":"","what":"Training & evaluation using TF Dataset objects","title":"Training & evaluation with the built-in methods","text":"past paragraphs, ’ve seen handle losses, metrics, optimizers, ’ve seen use validation_data validation_split arguments fit(), data passed arrays. Another option use iterator-like, tf.data.Dataset, PyTorch DataLoader, R generator function. Let’s take look former. tfdatasets R package containes set utilities loading preprocessing data way ’s fast scalable. complete guide creating Datasets, see tf.data documentation. can use tf.data train Keras models regardless backend ’re using – whether ’s JAX, PyTorch, TensorFlow. can pass Dataset instance directly methods fit(), evaluate(), predict(): Note Dataset reset end epoch, can reused next epoch. want run training specific number batches Dataset, can pass steps_per_epoch argument, specifies many training steps model run using Dataset moving next epoch. can also pass Dataset instance validation_data argument fit(): end epoch, model iterate validation dataset compute validation loss validation metrics. want run validation specific number batches dataset, can pass validation_steps argument, specifies many validation steps model run validation dataset interrupting validation moving next epoch: Note validation dataset reset use (always evaluating samples epoch epoch). argument validation_split (generating holdout set training data) supported training Dataset objects, since feature requires ability index samples datasets, possible general Dataset API.","code":"library(tfdatasets, exclude = \"shape\") model <- get_compiled_model()  # First, let's create a training Dataset instance. # For the sake of our example, we'll use the same MNIST data as before. train_dataset <- tensor_slices_dataset(list(x_train, y_train))  # Shuffle and slice the dataset. train_dataset <- train_dataset |>   dataset_shuffle(buffer_size=1024) |>   dataset_batch(64)  # Now we get a test dataset. test_dataset <-   tensor_slices_dataset(list(x_test, y_test)) |>   dataset_batch(64)  # Since the dataset already takes care of batching, # we don't pass a `batch_size` argument. model |> fit(train_dataset, epochs = 3) ## Epoch 1/3 ## 782/782 - 2s - 2ms/step - loss: 0.3365 - sparse_categorical_accuracy: 0.9041 ## Epoch 2/3 ## 782/782 - 1s - 1ms/step - loss: 0.1606 - sparse_categorical_accuracy: 0.9521 ## Epoch 3/3 ## 782/782 - 1s - 1ms/step - loss: 0.1178 - sparse_categorical_accuracy: 0.9646 # You can also evaluate or predict on a dataset. result <- model |> evaluate(test_dataset) ## 157/157 - 1s - 3ms/step - loss: 0.1213 - sparse_categorical_accuracy: 0.9596 result ## $loss ## [1] 0.1213005 ## ## $sparse_categorical_accuracy ## [1] 0.9596 model <- get_compiled_model()  # Prepare the training dataset train_dataset <- tensor_slices_dataset(list(x_train, y_train)) train_dataset <- train_dataset |>   dataset_shuffle(buffer_size = 1024) |>   dataset_batch(64)  # Only use the 100 batches per epoch (that's 64 * 100 samples) model |> fit(train_dataset, epochs = 3, steps_per_epoch = 100) ## Epoch 1/3 ## 100/100 - 1s - 8ms/step - loss: 0.8017 - sparse_categorical_accuracy: 0.7806 ## Epoch 2/3 ## 100/100 - 0s - 1ms/step - loss: 0.3661 - sparse_categorical_accuracy: 0.9006 ## Epoch 3/3 ## 100/100 - 0s - 960us/step - loss: 0.3009 - sparse_categorical_accuracy: 0.9106 model <- get_compiled_model()  # Prepare the training dataset train_dataset <- tensor_slices_dataset(list(x_train, y_train)) train_dataset <- train_dataset |>   dataset_shuffle(buffer_size=1024) |>   dataset_batch(64)  # Prepare the validation dataset val_dataset <- tensor_slices_dataset(list(x_val, y_val)) val_dataset <- val_dataset |> dataset_batch(64)  model |> fit(train_dataset, epochs = 1, validation_data = val_dataset) ## 782/782 - 2s - 3ms/step - loss: 0.3428 - sparse_categorical_accuracy: 0.9022 - val_loss: 0.2337 - val_sparse_categorical_accuracy: 0.9291 model  <- get_compiled_model()  # Prepare the training dataset train_dataset <- tensor_slices_dataset(list(x_train, y_train)) train_dataset <- train_dataset |>   dataset_shuffle(buffer_size = 1024) |>   dataset_batch(64)  # Prepare the validation dataset val_dataset <- tensor_slices_dataset(list(x_val, y_val)) val_dataset <- val_dataset |> dataset_batch(64)  model %>% fit(   train_dataset,   epochs = 1,   # Only run validation using the first 10 batches of the dataset   # using the `validation_steps` argument   validation_data = val_dataset,   validation_steps = 10, ) ## 782/782 - 2s - 3ms/step - loss: 0.3391 - sparse_categorical_accuracy: 0.9035 - val_loss: 0.1997 - val_sparse_categorical_accuracy: 0.9391"},{"path":"https://keras3.posit.co/dev/articles/training_with_built_in_methods.html","id":"using-sample-weighting-and-class-weighting","dir":"Articles","previous_headings":"","what":"Using sample weighting and class weighting","title":"Training & evaluation with the built-in methods","text":"default settings weight sample decided frequency dataset. two methods weight data, independent sample frequency: Class weights Sample weights","code":""},{"path":"https://keras3.posit.co/dev/articles/training_with_built_in_methods.html","id":"class-weights","dir":"Articles","previous_headings":"Using sample weighting and class weighting","what":"Class weights","title":"Training & evaluation with the built-in methods","text":"set passing named list class_weight argument fit(). list maps class indices weight used samples belonging class. can used balance classes without resampling, train model gives importance particular class. instance, class “0” half represented class “1” data, use model |> fit(..., class_weight = c(\"0\" = 1, \"1\" = 0.5)). ’s R example use class weights sample weights give importance correct classification class #5 (digit “5” MNIST dataset).","code":"class_weight <- c(     \"0\" = 1.0,     \"1\" = 1.0,     \"2\" = 1.0,     \"3\" = 1.0,     \"4\" = 1.0,     # Set weight \"2\" for class \"5\",     # making this class 2x more important     \"5\" = 2.0,     \"6\" = 1.0,     \"7\" = 1.0,     \"8\" = 1.0,     \"9\" = 1.0 )  model <- get_compiled_model() model |> fit(x_train, y_train,              class_weight = class_weight,              batch_size = 64, epochs = 1) ## 782/782 - 2s - 2ms/step - loss: 0.3713 - sparse_categorical_accuracy: 0.9018"},{"path":"https://keras3.posit.co/dev/articles/training_with_built_in_methods.html","id":"sample-weights","dir":"Articles","previous_headings":"Using sample weighting and class weighting","what":"Sample weights","title":"Training & evaluation with the built-in methods","text":"fine grained control, building classifier, can use sample_weights. training R arrays: Pass sample_weight argument fit(). training tf_dataset sort iterator: yield (input_batch, label_batch, sample_weight_batch) tuples. “sample weights” array array numbers specify much weight sample batch computing total loss. commonly used imbalanced classification problems (idea give weight rarely-seen classes). weights used ones zeros, array can used mask loss function (entirely discarding contribution certain samples total loss). ’s matching Dataset example:","code":"sample_weight <- rep(1.0, length(y_train)) sample_weight[y_train == 5] <- 2.0  model <- get_compiled_model() model |> fit(   x_train, y_train,   sample_weight = sample_weight,   batch_size = 64, epochs = 1 ) ## 782/782 - 2s - 2ms/step - loss: 0.3740 - sparse_categorical_accuracy: 0.9015 sample_weight <- rep(1.0, length(y_train)) sample_weight[y_train == 5] <- 2.0  # Create a Dataset that includes sample weights # (3rd element in the return tuple). train_dataset <- tensor_slices_dataset(list(     x_train, y_train, sample_weight ))  # Shuffle and slice the dataset. train_dataset <- train_dataset |>   dataset_shuffle(buffer_size = 1024) |>   dataset_batch(64)  model <- get_compiled_model() model |> fit(train_dataset, epochs = 1) ## 782/782 - 2s - 2ms/step - loss: 0.3654 - sparse_categorical_accuracy: 0.9057"},{"path":"https://keras3.posit.co/dev/articles/training_with_built_in_methods.html","id":"passing-data-to-multi-input-multi-output-models","dir":"Articles","previous_headings":"","what":"Passing data to multi-input, multi-output models","title":"Training & evaluation with the built-in methods","text":"previous examples, considering model single input (tensor shape (764)) single output (prediction tensor shape (10)). models multiple inputs outputs? Consider following model, image input shape (32, 32, 3) (’s (height, width, channels)) time series input shape (NA, 10) (’s (timesteps, features)). model two outputs computed combination inputs: “score” (shape (1)) probability distribution five classes (shape (5)). Let’s plot model, can clearly see ’re (note shapes shown plot batch shapes, rather per-sample shapes). plot chunk unnamed-chunk-26 compilation time, can specify different losses different outputs, passing loss functions list: passed single loss function model, loss function applied every output (appropriate ). Likewise metrics: Since gave names output layers, also specify per-output losses metrics via named list: recommend use names 2 outputs. ’s possible give different weights different output-specific losses (instance, one might wish privilege “score” loss example, giving 2x importance class loss), using loss_weights argument: also choose compute loss certain outputs, outputs meant prediction training: Passing data multi-input multi-output model fit() works similar way specifying loss function compile: can pass lists arrays (1:1 mapping outputs received loss function) dicts mapping output names arrays. ’s Dataset use case: similarly R arrays, Dataset return tuple named lists (dicts).","code":"image_input <- keras_input(c(32, 32, 3), name = \"img_input\") timeseries_input <- keras_input(c(NA, 10), name = \"ts_input\")  x1 <- image_input |>   layer_conv_2d(filters = 3, kernel_size = c(3, 3)) |>   layer_global_max_pooling_2d()  x2 <- timeseries_input |>   layer_conv_1d(filters = 3, kernel_size = 3) |>   layer_global_max_pooling_1d()  x <- layer_concatenate(x1, x2)  score_output <- layer_dense(x, 1, name = \"score_output\") class_output <- layer_dense(x, 5, name = \"class_output\")  model <- keras_model(   inputs = list(image_input, timeseries_input),   outputs = list(score_output, class_output) ) plot(model, show_shapes = TRUE) model |> compile(   optimizer = optimizer_rmsprop(1e-3),   loss = list(     loss_mean_squared_error(),     loss_categorical_crossentropy()   ) ) model |> compile(   optimizer = optimizer_rmsprop(1e-3),   loss = list(     loss_mean_squared_error(),     loss_categorical_crossentropy()   ),   metrics = list(     list(       metric_mean_absolute_error(),       metric_mean_absolute_percentage_error()     ),     list(metric_categorical_accuracy())   ) ) model |> compile(   optimizer = optimizer_rmsprop(1e-3),   loss = list(     score_output = loss_mean_squared_error(),     class_output = loss_categorical_crossentropy()   ),   metrics = list(     score_output = list(       metric_mean_absolute_error(),       metric_mean_absolute_percentage_error()     ),     class_output = list(metric_categorical_accuracy())   ) ) model |> compile(   optimizer = optimizer_rmsprop(1e-3),   loss = list(     score_output = loss_mean_squared_error(),     class_output = loss_categorical_crossentropy()   ),   metrics = list(     score_output = list(       metric_mean_absolute_error(),       metric_mean_absolute_percentage_error()     ),     class_output = list(metric_categorical_accuracy())   ),   loss_weights = list(score_output = 2.0, class_output = 1.0) ) # loss list, positional version model |> compile(   optimizer = optimizer_rmsprop(1e-3),   loss = list(NULL, loss_categorical_crossentropy()) )  # Or loss list, named version model |> compile(   optimizer = optimizer_rmsprop(1e-3),   loss = list(class_output = loss_categorical_crossentropy()) ) model |> compile(   optimizer = optimizer_rmsprop(1e-3),   loss = list(     loss_mean_squared_error(),     loss_binary_crossentropy()   ) )  # Generate dummy data img_data <- random_normal(c(100, 32, 32, 3)) ts_data <- random_normal(c(100, 20, 10)) score_targets <- random_normal(c(100, 1)) class_targets <- random_normal(c(100, 5))  # Fit on unnamed lists (positional matching) model |> fit(     list(img_data, ts_data),     list(score_targets, class_targets),     batch_size=32,     epochs=1 ) ## 4/4 - 2s - 540ms/step - class_output_categorical_accuracy: 0.2300 - class_output_loss: 1.8162 - loss: 3.8764 - score_output_loss: 1.6820 - score_output_mean_absolute_error: 0.8826 - score_output_mean_absolute_percentage_error: 1826.5676 # Alternatively, fit on named lists (names matching) model <- keras_model(   inputs = list(image_input, timeseries_input),   outputs = list(score_output = score_output,                  class_output = class_output) ) |> compile(   optimizer = optimizer_rmsprop(1e-3),   loss = list(     loss_mean_squared_error(),     loss_binary_crossentropy()   ) )  model |> fit(   list(img_input = img_data, ts_input = ts_data),   list(score_output = score_targets, class_output = class_targets),   batch_size = 32,   epochs = 1 ) ## 4/4 - 2s - 392ms/step - class_output_loss: 4.1489 - loss: 4.5443 - score_output_loss: -4.1125e-01 train_dataset <- tensor_slices_dataset(list(   list(img_input = img_data, ts_input = ts_data),   list(score_output = score_targets, class_output = class_targets) )) train_dataset <- train_dataset |>   dataset_shuffle(buffer_size = 1024) |>   dataset_batch(64)  model |> fit(train_dataset, epochs = 1) ## 2/2 - 1s - 706ms/step - class_output_loss: 3.8547 - loss: 3.5725 - score_output_loss: -1.5209e-01"},{"path":"https://keras3.posit.co/dev/articles/training_with_built_in_methods.html","id":"using-callbacks","dir":"Articles","previous_headings":"","what":"Using callbacks","title":"Training & evaluation with the built-in methods","text":"Callbacks Keras objects called different points training (start epoch, end batch, end epoch, etc.). can used implement certain behaviors, : validation different points training (beyond built-per-epoch validation) Checkpointing model regular intervals exceeds certain accuracy threshold Changing learning rate model training seems plateauing fine-tuning top layers training seems plateauing Sending email instant message notifications training ends certain performance threshold exceeded Etc. Callbacks can passed list call fit():","code":"model <- get_compiled_model()  callbacks <- list(   callback_early_stopping(     # Stop training when `val_loss` is no longer improving     monitor = \"val_loss\",     # \"no longer improving\" being defined as \"no better than 1e-2 less\"     min_delta = 1e-2,     # \"no longer improving\" being further defined as \"for at least 2 epochs\"     patience = 2,     verbose = 1   ) ) model |> fit(   x_train,   y_train,   epochs = 20,   batch_size = 64,   callbacks = callbacks,   validation_split = 0.2, ) ## Epoch 1/20 ## 625/625 - 2s - 3ms/step - loss: 0.3695 - sparse_categorical_accuracy: 0.8961 - val_loss: 0.1873 - val_sparse_categorical_accuracy: 0.9469 ## Epoch 2/20 ## 625/625 - 1s - 1ms/step - loss: 0.1752 - sparse_categorical_accuracy: 0.9488 - val_loss: 0.1405 - val_sparse_categorical_accuracy: 0.9576 ## Epoch 3/20 ## 625/625 - 1s - 1ms/step - loss: 0.1277 - sparse_categorical_accuracy: 0.9626 - val_loss: 0.1217 - val_sparse_categorical_accuracy: 0.9644 ## Epoch 4/20 ## 625/625 - 1s - 1ms/step - loss: 0.1009 - sparse_categorical_accuracy: 0.9702 - val_loss: 0.1138 - val_sparse_categorical_accuracy: 0.9667 ## Epoch 5/20 ## 625/625 - 1s - 1ms/step - loss: 0.0823 - sparse_categorical_accuracy: 0.9758 - val_loss: 0.1086 - val_sparse_categorical_accuracy: 0.9679 ## Epoch 6/20 ## 625/625 - 1s - 1ms/step - loss: 0.0682 - sparse_categorical_accuracy: 0.9800 - val_loss: 0.1084 - val_sparse_categorical_accuracy: 0.9685 ## Epoch 7/20 ## 625/625 - 1s - 1ms/step - loss: 0.0573 - sparse_categorical_accuracy: 0.9834 - val_loss: 0.1100 - val_sparse_categorical_accuracy: 0.9698 ## Epoch 8/20 ## 625/625 - 1s - 1ms/step - loss: 0.0485 - sparse_categorical_accuracy: 0.9868 - val_loss: 0.1113 - val_sparse_categorical_accuracy: 0.9702 ## Epoch 8: early stopping"},{"path":"https://keras3.posit.co/dev/articles/training_with_built_in_methods.html","id":"many-built-in-callbacks-are-available","dir":"Articles","previous_headings":"Using callbacks","what":"Many built-in callbacks are available","title":"Training & evaluation with the built-in methods","text":"many built-callbacks already available Keras, : callback_model_checkpoint(): Periodically save model. callback_early_stopping(): Stop training training longer improving validation metrics. callback_tensorboard(): periodically write model logs can visualized TensorBoard (details section “Visualization”). callback_csv_logger(): streams loss metrics data CSV file. etc. See callbacks documentation complete list.","code":""},{"path":"https://keras3.posit.co/dev/articles/training_with_built_in_methods.html","id":"writing-your-own-callback","dir":"Articles","previous_headings":"Using callbacks","what":"Writing your own callback","title":"Training & evaluation with the built-in methods","text":"can create custom callback subclassing base [Callback()] class. callback access associated model class property self$model. Make sure read complete guide writing custom callbacks. ’s simple example saving list per-batch loss values training:","code":"callback_loss_history <- Callback(   classname = \"LossHistory\",   initialize = function(file = \"per_training_batch_losses.txt\", ...) {     super$initialize(...)     private$file <- file   },   on_train_begin = function(logs = NULL) {     private$per_batch_losses <- fastmap::faststack()   },   on_train_batch_begin = function(batch, logs = NULL) {     private$per_batch_losses$push(logs$loss)   },   on_train_end = function(logs = NULL) {     per_batch_losses <- private$per_batch_losses$as_list() |> as.numeric()     write(per_batch_losses, private$file)   } )"},{"path":"https://keras3.posit.co/dev/articles/training_with_built_in_methods.html","id":"checkpointing-models","dir":"Articles","previous_headings":"","what":"Checkpointing models","title":"Training & evaluation with the built-in methods","text":"’re training model relatively large datasets, ’s crucial save checkpoints model frequent intervals. easiest way achieve [callback_model_checkpoint()]: ModelCheckpoint callback can used implement fault-tolerance: ability restart training last saved state model case training gets randomly interrupted. ’s basic example: call also write callback saving restoring models. complete guide serialization saving, see guide saving serializing Models.","code":"model <- get_compiled_model()  callbacks <- list(   callback_model_checkpoint(     # Path where to save the model     # The two parameters below mean that we will overwrite     # the current checkpoint if and only if     # the `val_loss` score has improved.     # The saved model name will include the current epoch.     filepath = \"mymodel_{epoch}.keras\",     save_best_only = TRUE,     # Only save a model if `val_loss` has improved.     monitor = \"val_loss\",     verbose = 1   ) ) model |> fit(   x_train, y_train,   epochs = 2, batch_size = 64,   callbacks = callbacks,   validation_split = 0.2 ) ## Epoch 1/2 ## ## Epoch 1: val_loss improved from None to 0.19488, saving model to mymodel_1.keras ## 625/625 - 2s - 3ms/step - loss: 0.3793 - sparse_categorical_accuracy: 0.8936 - val_loss: 0.1949 - val_sparse_categorical_accuracy: 0.9447 ## Epoch 2/2 ## ## Epoch 2: val_loss improved from 0.19488 to 0.14489, saving model to mymodel_2.keras ## 625/625 - 1s - 1ms/step - loss: 0.1777 - sparse_categorical_accuracy: 0.9475 - val_loss: 0.1449 - val_sparse_categorical_accuracy: 0.9590 # Prepare a directory to store all the checkpoints. checkpoint_dir <- \"./ckpt\" fs::dir_create(checkpoint_dir)  make_or_restore_model <- function() {   # Either restore the latest (best) model, or create a fresh one   # if there is no checkpoint available.   checkpoints <- Sys.glob(file.path(checkpoint_dir, \"model-loss=*.keras\"))    if (length(checkpoints) > 0) {     checkpoint_losses <- sub(\"^model-loss=([0-9.]+)\\\\.keras$\", \"\\\\1\",                              basename(checkpoints)) |> as.numeric()     best_checkpoint <- checkpoints[which.min(checkpoint_losses)]     load_model(best_checkpoint)   } else {     get_compiled_model()   } }  model <- make_or_restore_model() callbacks <- list(   # This callback saves the model every 100 batches.   # We include the training loss in the saved model name.   callback_model_checkpoint(     filepath = file.path(checkpoint_dir, \"model-loss={loss:.2f}.keras\"),     save_freq = 100   ) ) model |> fit(x_train, y_train, epochs = 1, callbacks = callbacks) ## 1563/1563 - 3s - 2ms/step - loss: 0.2932 - sparse_categorical_accuracy: 0.9145"},{"path":"https://keras3.posit.co/dev/articles/training_with_built_in_methods.html","id":"using-learning-rate-schedules","dir":"Articles","previous_headings":"","what":"Using learning rate schedules","title":"Training & evaluation with the built-in methods","text":"common pattern training deep learning models gradually reduce learning training progresses. generally known “learning rate decay”. learning decay schedule static (fixed advance, function current epoch current batch index), dynamic (responding current behavior model, particular validation loss).","code":""},{"path":"https://keras3.posit.co/dev/articles/training_with_built_in_methods.html","id":"passing-a-schedule-to-an-optimizer","dir":"Articles","previous_headings":"Using learning rate schedules","what":"Passing a schedule to an optimizer","title":"Training & evaluation with the built-in methods","text":"can easily use static learning rate decay schedule passing schedule object learning_rate argument optimizer: Several built-schedules available: ExponentialDecay, PiecewiseConstantDecay, PolynomialDecay, InverseTimeDecay.","code":"initial_learning_rate <- 0.1 lr_schedule <- learning_rate_schedule_exponential_decay(     initial_learning_rate, decay_steps=100000, decay_rate=0.96,     staircase=TRUE )  optimizer <- optimizer_rmsprop(learning_rate = lr_schedule)"},{"path":"https://keras3.posit.co/dev/articles/training_with_built_in_methods.html","id":"using-callbacks-to-implement-a-dynamic-learning-rate-schedule","dir":"Articles","previous_headings":"Using learning rate schedules","what":"Using callbacks to implement a dynamic learning rate schedule","title":"Training & evaluation with the built-in methods","text":"dynamic learning rate schedule (instance, decreasing learning rate validation loss longer improving) achieved schedule objects, since optimizer access validation metrics. However, callbacks access metrics, including validation metrics! can thus achieve pattern using callback modifies current learning rate optimizer. fact, even built-[callback_reduce_lr_on_plateau()].","code":""},{"path":"https://keras3.posit.co/dev/articles/training_with_built_in_methods.html","id":"visualizing-loss-and-metrics-during-training-with-tensorboard","dir":"Articles","previous_headings":"","what":"Visualizing loss and metrics during training with TensorBoard","title":"Training & evaluation with the built-in methods","text":"best way keep eye model training use TensorBoard – browser-based application can run locally provides : Live plots loss metrics training evaluation (optionally) Visualizations histograms layer activations (optionally) 3D visualizations embedding spaces learned layer_embedding() installed TensorFlow pip, able launch TensorBoard command line: R using:","code":"tensorboard --logdir=/full_path_to_your_logs tensorflow::tensorboard(logdir = \"/full_path_to_your_logs\")"},{"path":"https://keras3.posit.co/dev/articles/training_with_built_in_methods.html","id":"using-the-tensorboard-callback","dir":"Articles","previous_headings":"Visualizing loss and metrics during training with TensorBoard","what":"Using the TensorBoard callback","title":"Training & evaluation with the built-in methods","text":"easiest way use TensorBoard Keras model fit() method [callback_tensorboard()]. simplest case, just specify want callback write logs, ’re good go: information, see callback_tensorboard().","code":"tb_callback <- callback_tensorboard(   log_dir = \"/full_path_to_your_logs\",   histogram_freq = 0, # How often to log histogram visualizations   embeddings_freq = 0, # How often to log embedding visualizations   update_freq = \"epoch\", # How often to write logs (default: once per epoch) )"},{"path":"https://keras3.posit.co/dev/articles/transfer_learning.html","id":"setup","dir":"Articles","previous_headings":"","what":"Setup","title":"Transfer learning & fine-tuning","text":"","code":"library(keras3)"},{"path":"https://keras3.posit.co/dev/articles/transfer_learning.html","id":"introduction","dir":"Articles","previous_headings":"","what":"Introduction","title":"Transfer learning & fine-tuning","text":"Transfer learning consists taking features learned one problem, leveraging new, similar problem. instance, features model learned identify raccoon may useful kick-start model meant identify tanukis. Transfer learning usually done tasks dataset little data train full-scale model scratch. common incarnation transfer learning context deep learning following workflow: Take layers previously trained model. Freeze , avoid destroying information contain future training rounds. Add new, trainable layers top frozen layers. learn turn old features predictions new dataset. Train new layers dataset. last, optional step, fine-tuning, consists unfreezing entire model obtained (part ), re-training new data low learning rate. can potentially achieve meaningful improvements, incrementally adapting pretrained features new data. First, go Keras trainable API detail, underlies transfer learning & fine-tuning workflows. , ’ll demonstrate typical workflow taking model pretrained ImageNet dataset, retraining Kaggle “cats vs dogs” classification dataset. adapted Deep Learning R 2016 blog post “building powerful image classification models using little data”.","code":""},{"path":"https://keras3.posit.co/dev/articles/transfer_learning.html","id":"freezing-layers-understanding-the-trainable-attribute","dir":"Articles","previous_headings":"","what":"Freezing layers: understanding the trainable attribute","title":"Transfer learning & fine-tuning","text":"Layers & models three weight attributes: weights list weights variables layer. trainable_weights list meant updated (via gradient descent) minimize loss training. non_trainable_weights list aren’t meant trained. Typically updated model forward pass. Example: Dense layer 2 trainable weights (kernel & bias) general, weights trainable weights. built-layer non-trainable weights [layer_batch_normalization()]. uses non-trainable weights keep track mean variance inputs training. learn use non-trainable weights custom layers, see guide writing new layers scratch. Example: BatchNormalization layer 2 trainable weights 2 non-trainable weights Layers & models also feature boolean attribute trainable. value can changed. Setting layer$trainable FALSE moves layer’s weights trainable non-trainable. called “freezing” layer: state frozen layer won’t updated training (either training fit() training custom loop relies trainable_weights apply gradient updates). Example: setting trainable False trainable weight becomes non-trainable, value longer updated training. confuse layer$trainable attribute argument training layer$call() (controls whether layer run forward pass inference mode training mode). information, see Keras FAQ.","code":"layer <- layer_dense(units = 3) layer$build(shape(NULL, 4))  # Create the weights  length(layer$weights) ## [1] 2 length(layer$trainable_weights) ## [1] 2 length(layer$non_trainable_weights) ## [1] 0 layer <- layer_batch_normalization() layer$build(shape(NA, 4))  # Create the weights  length(layer$weights) ## [1] 4 length(layer$trainable_weights) ## [1] 2 length(layer$non_trainable_weights) ## [1] 2 layer <- layer_dense(units = 3) layer$build(shape(NULL, 4))  # Create the weights layer$trainable <- FALSE  # Freeze the layer  length(layer$weights) ## [1] 2 length(layer$trainable_weights) ## [1] 0 length(layer$non_trainable_weights) ## [1] 2 # Make a model with 2 layers layer1 <- layer_dense(units = 3, activation = \"relu\") layer2 <- layer_dense(units = 3, activation = \"sigmoid\") model <- keras_model_sequential(input_shape = 3) |>   layer1() |>   layer2()  # Freeze the first layer layer1$trainable <- FALSE  # Keep a copy of the weights of layer1 for later reference # (get_weights() returns a list of R arrays, #  layer$weights returns a list of KerasVariables) initial_layer1_weights_values <- get_weights(layer1)  # Train the model model |> compile(optimizer = \"adam\", loss = \"mse\") model |> fit(random_normal(c(2, 3)), random_normal(c(2, 3)), epochs = 1) ## 1/1 - 0s - 436ms/step - loss: 2.8362 # Check that the weights of layer1 have not changed during training final_layer1_weights_values <- get_weights(layer1)  all.equal(initial_layer1_weights_values,           final_layer1_weights_values) ## [1] TRUE"},{"path":"https://keras3.posit.co/dev/articles/transfer_learning.html","id":"recursive-setting-of-the-trainable-attribute","dir":"Articles","previous_headings":"","what":"Recursive setting of the trainable attribute","title":"Transfer learning & fine-tuning","text":"set $trainable <- FALSE model layer sublayers, children layers become non-trainable well. Example:","code":"inner_model <- keras_model_sequential(input_shape = 3) |>   layer_dense(units = 3, activation = \"relu\") |>   layer_dense(units = 3, activation = \"relu\")  model <- keras_model_sequential(input_shape = 3) |>   inner_model() |>   layer_dense(units = 3, activation = \"sigmoid\")  model$trainable <- FALSE  # Freeze the outer model  inner_model$trainable  # All layers in `model` are now frozen ## [1] FALSE inner_model$layers[[1]]$trainable  # `trainable` is propagated recursively ## [1] FALSE"},{"path":"https://keras3.posit.co/dev/articles/transfer_learning.html","id":"the-typical-transfer-learning-workflow","dir":"Articles","previous_headings":"","what":"The typical transfer-learning workflow","title":"Transfer learning & fine-tuning","text":"leads us typical transfer learning workflow can implemented Keras: Instantiate base model load pre-trained weights . Freeze layers base model setting trainable <- FALSE. Create new model top output one (several) layers base model. Train new model new dataset. Note alternative, lightweight workflow also : Instantiate base model load pre-trained weights . Run new dataset record output one (several) layers base model. called feature extraction. Use output input data new, smaller model. key advantage second workflow run base model data, rather per epoch training. ’s lot faster & cheaper. issue second workflow, though, doesn’t allow dynamically modify input data new model training, required data augmentation, instance. Transfer learning typically used tasks new dataset little data train full-scale model scratch, scenarios data augmentation important. follows, focus first workflow. ’s first workflow looks like Keras: First, instantiate base model pre-trained weights. , freeze base model. Create new model top. Train model new data.","code":"base_model <- application_xception(   weights = 'imagenet', # Load weights pre-trained on ImageNet.   input_shape = c(150, 150, 3),   include_top = FALSE  # Do not include the ImageNet classifier at the top. ) base_model$trainable <- FALSE inputs <- keras_input(shape = c(150, 150, 3)) # We make sure that the base_model is running in inference mode here, # by passing `training <- FALSE`. This is important for fine-tuning, as you will # learn in a few paragraphs. outputs <- inputs |>   base_model(training = FALSE) |>   # Convert features of shape `base_model$output_shape[-1]` to vectors   layer_global_average_pooling_2d() |>   # A Dense classifier with a single unit (binary classification)   layer_dense(1)  model <- keras_model(inputs, outputs) model |> compile(   optimizer = optimizer_adam(),   loss = loss_binary_crossentropy(from_logits = TRUE),   metrics = list(metric_binary_accuracy()) ) model |> fit(new_dataset, epochs = 20,              callbacks = ..., validation_data = ...)"},{"path":"https://keras3.posit.co/dev/articles/transfer_learning.html","id":"fine-tuning","dir":"Articles","previous_headings":"","what":"Fine-tuning","title":"Transfer learning & fine-tuning","text":"model converged new data, can try unfreeze part base model retrain whole model end--end low learning rate. optional last step can potentially give incremental improvements. also potentially lead quick overfitting – keep mind. critical step model frozen layers trained convergence. mix randomly-initialized trainable layers trainable layers hold pre-trained features, randomly-initialized layers cause large gradient updates training, destroy pre-trained features. ’s also critical use low learning rate stage, training much larger model first round training, dataset typically small. result, risk overfitting quickly apply large weight updates. , want readapt pretrained weights incremental way. implement fine-tuning whole base model: Important note compile() trainable Calling compile() model meant “freeze” behavior model. implies trainable attribute values time model compiled preserved throughout lifetime model, compile called . Hence, change trainable value, make sure call compile() model changes taken account. Important notes BatchNormalization layer Many image models contain BatchNormalization layers. layer special case every imaginable count. things keep mind. BatchNormalization contains 2 non-trainable weights get updated training. variables tracking mean variance inputs. set bn_layer$trainable <- FALSE, BatchNormalization layer run inference mode, update mean & variance statistics. case layers general, weight trainability & inference/training modes two orthogonal concepts. two tied case BatchNormalization layer. unfreeze model contains BatchNormalization layers order fine-tuning, keep BatchNormalization layers inference mode passing training = FALSE calling base model. Otherwise updates applied non-trainable weights suddenly destroy model learned. ’ll see pattern action end--end example end guide.","code":"# Unfreeze the base model base_model$trainable <- TRUE  # It's important to recompile your model after you make any changes # to the `trainable` attribute of any inner layer, so that your changes # are take into account model |> compile(   optimizer = optimizer_adam(1e-5), # Very low learning rate   loss = loss_binary_crossentropy(from_logits = TRUE),   metrics = c(metric_binary_accuracy()) )  # Train end-to-end. Be careful to stop before you overfit! model |> fit(new_dataset, epochs = 10,              callbacks = ..., validation_data = ...)"},{"path":"https://keras3.posit.co/dev/articles/transfer_learning.html","id":"an-end-to-end-example-fine-tuning-an-image-classification-model-on-a-cats-vs--dogs-dataset","dir":"Articles","previous_headings":"","what":"An end-to-end example: fine-tuning an image classification model on a cats vs. dogs dataset","title":"Transfer learning & fine-tuning","text":"solidify concepts, let’s walk concrete end--end transfer learning & fine-tuning example. load Xception model, pre-trained ImageNet, use Kaggle “cats vs. dogs” classification dataset.","code":""},{"path":"https://keras3.posit.co/dev/articles/transfer_learning.html","id":"getting-the-data","dir":"Articles","previous_headings":"An end-to-end example: fine-tuning an image classification model on a cats vs. dogs dataset","what":"Getting the data","title":"Transfer learning & fine-tuning","text":"First, let’s fetch cats vs. dogs dataset using TFDS. dataset, ’ll probably want use utility [image_dataset_from_directory()] generate similar labeled dataset objects set images disk filed class-specific folders. Transfer learning useful working small datasets. keep dataset small, use 40% original training data (25,000 images) training, 10% validation, 10% testing. first 9 images training dataset – can see, ’re different sizes. can also see label 1 “dog” label 0 “cat”.","code":"reticulate::py_require(\"tensorflow-datasets\") tfds <- reticulate::import(\"tensorflow_datasets\")  c(train_ds, validation_ds, test_ds) %<-% tfds$load(   \"cats_vs_dogs\",   # Reserve 10% for validation and 10% for test   split = c(\"train[:40%]\", \"train[40%:50%]\", \"train[50%:60%]\"),   as_supervised = TRUE  # Include labels )  length(train_ds) ## [1] 9305 library(tfdatasets, exclude = \"shape\")  par(mfrow = c(3, 3), mar = c(1,0,1.5,0)) train_ds |>   dataset_take(9) |>   as_array_iterator() |>   iterate(function(batch) {     c(image, label) %<-% batch     plot(as.raster(image, max = 255L))     title(sprintf(       \"label: %s   size: %s\",       label, paste(dim(image), collapse = \" x \")))   })"},{"path":"https://keras3.posit.co/dev/articles/transfer_learning.html","id":"standardizing-the-data","dir":"Articles","previous_headings":"An end-to-end example: fine-tuning an image classification model on a cats vs. dogs dataset","what":"Standardizing the data","title":"Transfer learning & fine-tuning","text":"raw images variety sizes. addition, pixel consists 3 integer values 0 255 (RGB level values). isn’t great fit feeding neural network. need 2 things: Standardize fixed image size. pick 150x150. Normalize pixel values -1 1. ’ll using Normalization layer part model . general, ’s good practice develop models take raw data input, opposed models take already-preprocessed data. reason , model expects preprocessed data, time export model use elsewhere (web browser, mobile app), ’ll need reimplement exact preprocessing pipeline. gets tricky quickly. least possible amount preprocessing hitting model. , ’ll image resizing data pipeline (deep neural network can process contiguous batches data), ’ll input value scaling part model, create . Let’s resize images 150x150:","code":"resize_fn <- layer_resizing(width = 150, height = 150) resize_pair <- function(x, y) list(resize_fn(x), y)  train_ds <- train_ds |> dataset_map(resize_pair) validation_ds <- validation_ds |> dataset_map(resize_pair) test_ds <- test_ds |> dataset_map(resize_pair)"},{"path":"https://keras3.posit.co/dev/articles/transfer_learning.html","id":"using-random-data-augmentation","dir":"Articles","previous_headings":"An end-to-end example: fine-tuning an image classification model on a cats vs. dogs dataset","what":"Using random data augmentation","title":"Transfer learning & fine-tuning","text":"don’t large image dataset, ’s good practice artificially introduce sample diversity applying random yet realistic transformations training images, random horizontal flipping small random rotations. helps expose model different aspects training data slowing overfitting. Let’s batch data use prefetching optimize loading speed. Let’s visualize first image first batch looks like various random transformations:","code":"data_augmentation <- keras_model_sequential() |>   layer_random_flip(\"horizontal\") |>   layer_random_rotation(.1)  train_ds <- train_ds %>%   dataset_map(function(x, y) list(data_augmentation(x), y)) library(tensorflow, exclude = c(\"shape\", \"set_random_seed\")) batch_size <- 64  train_ds <- train_ds |>   dataset_batch(batch_size) |>   dataset_prefetch()  validation_ds <- validation_ds |>   dataset_batch(batch_size) |>   dataset_prefetch()  test_ds <- test_ds |>   dataset_batch(batch_size) |>   dataset_prefetch() batch <- train_ds |>   dataset_take(1) |>   as_iterator() |>   iter_next()  c(images, labels) %<-% batch first_image <- images[1, all_dims(), drop = TRUE] augmented_image <- data_augmentation(first_image, training = TRUE)  plot_image <- function(image, main = deparse1(substitute(image))) {   image |>     as.array() |> # convert from tensor to R array     as.raster(max = 255) |>     plot()    if(!is.null(main))     title(main) }  par(mfrow = c(2, 2), mar = c(1, 1, 1.5, 1)) plot_image(first_image) plot_image(augmented_image) plot_image(data_augmentation(first_image, training = TRUE), \"augmented 2\") plot_image(data_augmentation(first_image, training = TRUE), \"augmented 3\")"},{"path":"https://keras3.posit.co/dev/articles/transfer_learning.html","id":"build-a-model","dir":"Articles","previous_headings":"","what":"Build a model","title":"Transfer learning & fine-tuning","text":"Now let’s built model follows blueprint ’ve explained earlier. Note : add Rescaling layer scale input values (initially [0, 255] range) [-1, 1] range. add Dropout layer classification layer, regularization. make sure pass training=FALSE calling base model, runs inference mode, batchnorm statistics don’t get updated even unfreeze base model fine-tuning.","code":"base_model <- application_xception(   weights = \"imagenet\", # Load weights pre-trained on ImageNet.   input_shape = c(150, 150, 3),   include_top = FALSE, # Do not include the ImageNet classifier at the top. )  # Freeze the base_model base_model$trainable <- FALSE  # Create new model on top inputs <- keras_input(shape = c(150, 150, 3))  # Pre-trained Xception weights requires that input be scaled # from (0, 255) to a range of (-1., +1.), the rescaling layer # outputs: `(inputs * scale) + offset` scale_layer <- layer_rescaling(scale = 1 / 127.5, offset = -1) x <- scale_layer(inputs)  # The base model contains batchnorm layers. We want to keep them in inference mode # when we unfreeze the base model for fine-tuning, so we make sure that the # base_model is running in inference mode here. outputs <- x |>   base_model(training = FALSE) |>   layer_global_average_pooling_2d() |>   layer_dropout(0.2) |>   layer_dense(1)  model <- keras_model(inputs, outputs)  summary(model, show_trainable = TRUE) ## Model: \"functional_8\" ## ┏━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━┳━━━━━━━┓ ## ┃ Layer (type)                ┃ Output Shape          ┃    Param # ┃ Trai… ┃ ## ┡━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━╇━━━━━━━┩ ## │ input_layer_7 (InputLayer)  │ (None, 150, 150, 3)   │          0 │   -   │ ## ├─────────────────────────────┼───────────────────────┼────────────┼───────┤ ## │ rescaling (Rescaling)       │ (None, 150, 150, 3)   │          0 │   -   │ ## ├─────────────────────────────┼───────────────────────┼────────────┼───────┤ ## │ xception (Functional)       │ (None, 5, 5, 2048)    │ 20,861,480 │   N   │ ## ├─────────────────────────────┼───────────────────────┼────────────┼───────┤ ## │ global_average_pooling2d_1  │ (None, 2048)          │          0 │   -   │ ## │ (GlobalAveragePooling2D)    │                       │            │       │ ## ├─────────────────────────────┼───────────────────────┼────────────┼───────┤ ## │ dropout (Dropout)           │ (None, 2048)          │          0 │   -   │ ## ├─────────────────────────────┼───────────────────────┼────────────┼───────┤ ## │ dense_8 (Dense)             │ (None, 1)             │      2,049 │   Y   │ ## └─────────────────────────────┴───────────────────────┴────────────┴───────┘ ##  Total params: 20,863,529 (79.59 MB) ##  Trainable params: 2,049 (8.00 KB) ##  Non-trainable params: 20,861,480 (79.58 MB)"},{"path":"https://keras3.posit.co/dev/articles/transfer_learning.html","id":"train-the-top-layer","dir":"Articles","previous_headings":"","what":"Train the top layer","title":"Transfer learning & fine-tuning","text":"","code":"model |> compile(   optimizer = optimizer_adam(),   loss = loss_binary_crossentropy(from_logits = TRUE),   metrics = list(metric_binary_accuracy()) )  epochs <- 1 model |> fit(train_ds, epochs = epochs, validation_data = validation_ds) ## 146/146 - 57s - 389ms/step - binary_accuracy: 0.9183 - loss: 0.1887 - val_binary_accuracy: 0.9669 - val_loss: 0.0926"},{"path":"https://keras3.posit.co/dev/articles/transfer_learning.html","id":"do-a-round-of-fine-tuning-of-the-entire-model","dir":"Articles","previous_headings":"","what":"Do a round of fine-tuning of the entire model","title":"Transfer learning & fine-tuning","text":"Finally, let’s unfreeze base model train entire model end--end low learning rate. Importantly, although base model becomes trainable, still running inference mode since passed training=FALSE calling built model. means batch normalization layers inside won’t update batch statistics. , wreck havoc representations learned model far. 10 epochs, fine-tuning gains us nice improvement . Let’s evaluate model test dataset:","code":"# Unfreeze the base_model. Note that it keeps running in inference mode # since we passed `training=FALSE` when calling it. This means that # the batchnorm layers will not update their batch statistics. # This prevents the batchnorm layers from undoing all the training # we've done so far. base_model$trainable <- TRUE summary(model, show_trainable = TRUE) ## Model: \"functional_8\" ## ┏━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━┳━━━━━━━┓ ## ┃ Layer (type)                ┃ Output Shape          ┃    Param # ┃ Trai… ┃ ## ┡━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━╇━━━━━━━┩ ## │ input_layer_7 (InputLayer)  │ (None, 150, 150, 3)   │          0 │   -   │ ## ├─────────────────────────────┼───────────────────────┼────────────┼───────┤ ## │ rescaling (Rescaling)       │ (None, 150, 150, 3)   │          0 │   -   │ ## ├─────────────────────────────┼───────────────────────┼────────────┼───────┤ ## │ xception (Functional)       │ (None, 5, 5, 2048)    │ 20,861,480 │   Y   │ ## ├─────────────────────────────┼───────────────────────┼────────────┼───────┤ ## │ global_average_pooling2d_1  │ (None, 2048)          │          0 │   -   │ ## │ (GlobalAveragePooling2D)    │                       │            │       │ ## ├─────────────────────────────┼───────────────────────┼────────────┼───────┤ ## │ dropout (Dropout)           │ (None, 2048)          │          0 │   -   │ ## ├─────────────────────────────┼───────────────────────┼────────────┼───────┤ ## │ dense_8 (Dense)             │ (None, 1)             │      2,049 │   Y   │ ## └─────────────────────────────┴───────────────────────┴────────────┴───────┘ ##  Total params: 20,867,629 (79.60 MB) ##  Trainable params: 20,809,001 (79.38 MB) ##  Non-trainable params: 54,528 (213.00 KB) ##  Optimizer params: 4,100 (16.02 KB) model |> compile(   optimizer = optimizer_adam(1e-5), # Low learning rate   loss = loss_binary_crossentropy(from_logits = TRUE),   metrics = list(metric_binary_accuracy()) )  epochs <- 1 model |> fit(train_ds, epochs = epochs, validation_data = validation_ds) ## 146/146 - 86s - 592ms/step - binary_accuracy: 0.8660 - loss: 0.3212 - val_binary_accuracy: 0.9652 - val_loss: 0.1022 model |> evaluate(test_ds) ## 37/37 - 2s - 43ms/step - binary_accuracy: 0.9540 - loss: 0.1102 ## $binary_accuracy ## [1] 0.9539983 ## ## $loss ## [1] 0.1102476"},{"path":"https://keras3.posit.co/dev/articles/understanding_masking_and_padding.html","id":"setup","dir":"Articles","previous_headings":"","what":"Setup","title":"Understanding masking & padding","text":"","code":"library(keras3)"},{"path":"https://keras3.posit.co/dev/articles/understanding_masking_and_padding.html","id":"introduction","dir":"Articles","previous_headings":"","what":"Introduction","title":"Understanding masking & padding","text":"Masking way tell sequence-processing layers certain timesteps input missing, thus skipped processing data. Padding special form masking masked steps start end sequence. Padding comes need encode sequence data contiguous batches: order make sequences batch fit given standard length, necessary pad truncate sequences. Let’s take close look.","code":""},{"path":"https://keras3.posit.co/dev/articles/understanding_masking_and_padding.html","id":"padding-sequence-data","dir":"Articles","previous_headings":"","what":"Padding sequence data","title":"Understanding masking & padding","text":"processing sequence data, common individual samples different lengths. Consider following example (text tokenized words): vocabulary lookup, data might vectorized integers, e.g.: data nested list individual samples length 3, 5, 6, respectively. Since input data deep learning model must single tensor (shape e.g. (batch_size, 6, vocab_size) case), samples shorter longest item need padded placeholder value (alternatively, one might also truncate long samples padding short samples). Keras provides utility function truncate pad Python lists common length: pad_sequences.","code":"data <- list(   c(\"Hello\", \"world\", \"!\"),   c(\"How\", \"are\", \"you\", \"doing\", \"today\"),   c(\"The\", \"weather\", \"will\", \"be\", \"nice\", \"tomorrow\") ) data <- list(   c(71, 1331, 4231),   c(73, 8, 3215, 55, 927),   c(83, 91, 1, 645, 1253, 927) ) raw_inputs <- list(   c(711, 632, 71),   c(73, 8, 3215, 55, 927),   c(83, 91, 1, 645, 1253, 927) )  # By default, this will pad using 0s; it is configurable via the # \"value\" parameter. # Note that you could use \"pre\" padding (at the beginning) or # \"post\" padding (at the end). # We recommend using \"post\" padding when working with RNN layers # (in order to be able to use the # CuDNN implementation of the layers). padded_inputs <- pad_sequences(raw_inputs, padding=\"post\") padded_inputs ##      [,1] [,2] [,3] [,4] [,5] [,6] ## [1,]  711  632   71    0    0    0 ## [2,]   73    8 3215   55  927    0 ## [3,]   83   91    1  645 1253  927"},{"path":"https://keras3.posit.co/dev/articles/understanding_masking_and_padding.html","id":"masking","dir":"Articles","previous_headings":"","what":"Masking","title":"Understanding masking & padding","text":"Now samples uniform length, model must informed part data actually padding ignored. mechanism masking. three ways introduce input masks Keras models: Add layer_masking layer. Configure layer_embedding layer mask_zero=TRUE. Pass mask argument manually calling layers support argument (e.g. RNN layers).","code":""},{"path":"https://keras3.posit.co/dev/articles/understanding_masking_and_padding.html","id":"mask-generating-layers-embedding-and-masking","dir":"Articles","previous_headings":"","what":"Mask-generating layers: Embedding and Masking","title":"Understanding masking & padding","text":"hood, layers create mask tensor (2D tensor shape (batch, sequence_length)), attach tensor output returned Masking Embedding layer. can see printed result, mask 2D boolean tensor shape (batch_size, sequence_length), individual FALSE entry indicates corresponding timestep ignored processing.","code":"embedding <- layer_embedding(input_dim=5000, output_dim=16, mask_zero=TRUE) masked_output <- embedding(padded_inputs)  masked_output$`_keras_mask` ## tf.Tensor( ## [[ True  True  True False False False] ##  [ True  True  True  True  True False] ##  [ True  True  True  True  True  True]], shape=(3, 6), dtype=bool) masking_layer <- layer_masking() # Simulate the embedding lookup by expanding the 2D input to 3D, # with embedding dimension of 10. unmasked_embedding <- op_cast(     op_tile(op_expand_dims(padded_inputs, axis=-1), c(1L, 1L, 10L)),     dtype=\"float32\" )  masked_embedding <- masking_layer(unmasked_embedding) masked_embedding$`_keras_mask` ## tf.Tensor( ## [[ True  True  True False False False] ##  [ True  True  True  True  True False] ##  [ True  True  True  True  True  True]], shape=(3, 6), dtype=bool)"},{"path":"https://keras3.posit.co/dev/articles/understanding_masking_and_padding.html","id":"mask-propagation-in-the-functional-api-and-sequential-api","dir":"Articles","previous_headings":"","what":"Mask propagation in the Functional API and Sequential API","title":"Understanding masking & padding","text":"using Functional API Sequential API, mask generated Embedding Masking layer propagated network layer capable using (example, RNN layers). Keras automatically fetch mask corresponding input pass layer knows use . instance, following Sequential model, LSTM layer automatically receive mask, means ignore padded values: also case following Functional API model:","code":"model <- keras_model_sequential() %>%   layer_embedding(input_dim=5000, output_dim=16, mask_zero=TRUE) %>%   layer_lstm(units=32) inputs <- keras_input(shape = shape(NULL), dtype=\"int32\") outputs <- inputs %>%   layer_embedding(input_dim=5000, output_dim=16, mask_zero=TRUE) %>%   layer_lstm(units=32)  model <- keras_model(inputs, outputs)"},{"path":"https://keras3.posit.co/dev/articles/understanding_masking_and_padding.html","id":"passing-mask-tensors-directly-to-layers","dir":"Articles","previous_headings":"","what":"Passing mask tensors directly to layers","title":"Understanding masking & padding","text":"Layers can handle masks (LSTM layer) mask argument call method. Meanwhile, layers produce mask (e.g. Embedding) expose compute_mask(input, previous_mask) method can call. Thus, can pass output compute_mask() method mask-producing layer call method mask-consuming layer, like :","code":"MyLayer <- new_layer_class(   \"MyLayer\",   initialize = function(...) {     super$initialize(...)     self$embedding <- layer_embedding(       input_dim=5000, output_dim=16, mask_zero=TRUE     )     self$lstm <- layer_lstm(units=32)   },   call = function(inputs) {     inputs %>%       self$embedding() %>%       # Note that you could also prepare a `mask` tensor manually.       # It only needs to be a boolean tensor       # with the right shape, i.e. (batch_size, timesteps).       self$lstm(mask=self$embedding$compute_mask(inputs))   } )  layer <- MyLayer() x <- random_integer(c(32, 10), 0, 100) layer(x) ## tf.Tensor( ## [[ 0.00130048 -0.00113367 -0.00715671 ... -0.00107615 -0.00162071 ##    0.00135018] ##  [-0.004185    0.00726349  0.00520932 ...  0.00119117  0.00230441 ##    0.00174123] ##  [-0.00537032 -0.00164898 -0.00238435 ... -0.00154158 -0.0038603 ##   -0.00105811] ##  ... ##  [ 0.00622133 -0.00905907 -0.00599518 ...  0.00025823 -0.00142478 ##   -0.00125036] ##  [-0.00523904  0.00336683 -0.00299453 ...  0.00876719  0.00172074 ##    0.00903089] ##  [-0.00393721  0.00058538  0.00503809 ... -0.00203075  0.00325885 ##   -0.00299755]], shape=(32, 32), dtype=float32)"},{"path":"https://keras3.posit.co/dev/articles/understanding_masking_and_padding.html","id":"supporting-masking-in-your-custom-layers","dir":"Articles","previous_headings":"","what":"Supporting masking in your custom layers","title":"Understanding masking & padding","text":"Sometimes, may need write layers generate mask (like Embedding), layers need modify current mask. instance, layer produces tensor different time dimension input, Concatenate layer concatenates time dimension, need modify current mask downstream layers able properly take masked timesteps account. , layer implement layer.compute_mask() method, produces new mask given input current mask. example TemporalSplit layer needs modify current mask. another example CustomEmbedding layer capable generating mask input values: Note: details format limitations related masking, see serialization guide.","code":"TemporalSplit <- new_layer_class(   \"TemporalSplit\",   call = function(inputs) {     # Expect the input to be 3D and mask to be 2D, split the input tensor into 2     # subtensors along the time axis (axis 1).     op_split(inputs, 2, axis=2)   },   compute_mask = function(inputs, mask = NULL) {     # Also split the mask into 2 if it presents.     if (!is.null(mask)) {       op_split(mask, 2, axis=2)     } else {       NULL     }   } )  c(first_half, second_half) %<-% TemporalSplit(masked_embedding) first_half$`_keras_mask` ## tf.Tensor( ## [[ True  True  True] ##  [ True  True  True] ##  [ True  True  True]], shape=(3, 3), dtype=bool) second_half$`_keras_mask` ## tf.Tensor( ## [[False False False] ##  [ True  True False] ##  [ True  True  True]], shape=(3, 3), dtype=bool) CustomEmbedding <- new_layer_class(   \"CustomEmbedding\",   initialize = function(input_dim, output_dim, mask_zero=FALSE, ...) {     super$initialize(...)     self$input_dim <- as.integer(input_dim)     self$output_dim <- as.integer(output_dim)     self$mask_zero <- mask_zero   },   build = function(input_shape) {     self$embeddings <- self$add_weight(       shape=c(self$input_dim, self$output_dim),       initializer=\"random_normal\",       dtype=\"float32\"     )   },   call = function(inputs) {     inputs <- op_cast(inputs, \"int32\")     op_take(self$embeddings, inputs)   },   compute_mask = function(inputs, mask=NULL) {     if (!self$mask_zero) {       NULL     } else {       op_not_equal(inputs, 0)     }   } )  layer <- CustomEmbedding(input_dim = 10, output_dim = 32, mask_zero=TRUE) x <- random_integer(c(3, 10), 0, 9)  y <- layer(x) mask <- layer$compute_mask(x)  mask ## tf.Tensor( ## [[ True  True  True  True  True  True  True  True  True  True] ##  [ True  True  True  True  True  True  True  True False  True] ##  [ True  True  True  True False  True  True  True  True  True]], shape=(3, 10), dtype=bool)"},{"path":"https://keras3.posit.co/dev/articles/understanding_masking_and_padding.html","id":"opting-in-to-mask-propagation-on-compatible-layers","dir":"Articles","previous_headings":"","what":"Opting-in to mask propagation on compatible layers","title":"Understanding masking & padding","text":"layers don’t modify time dimension, don’t need modify current mask. However, may still want able propagate current mask, unchanged, next layer. opt-behavior. default, custom layer destroy current mask (since framework way tell whether propagating mask safe ). custom layer modify time dimension, want able propagate current input mask, set self.supports_masking = True layer constructor. case, default behavior compute_mask() just pass current mask . ’s example layer whitelisted mask propagation: can now use custom layer -mask-generating layer (like Embedding) mask-consuming layer (like LSTM), pass mask along reaches mask-consuming layer.","code":"MyActivation <- new_layer_class(   \"MyActivation\",   initialize = function(...) {     super$initialize(...)     self$supports_masking <- TRUE   },   call = function(inputs) {     op_relu(inputs)   } ) inputs <- keras_input(shape = shape(NULL), dtype=\"int32\") outputs <- inputs %>%   layer_embedding(input_dim=5000, output_dim=16, mask_zero=TRUE) %>%   MyActivation() %>%   layer_lstm(units=32)  model <- keras_model(inputs, outputs) y <- model(random_integer(c(32, 100), 0, 5000))"},{"path":"https://keras3.posit.co/dev/articles/understanding_masking_and_padding.html","id":"writing-layers-that-need-mask-information","dir":"Articles","previous_headings":"","what":"Writing layers that need mask information","title":"Understanding masking & padding","text":"layers mask consumers: accept mask argument call use determine whether skip certain time steps. write layer, can simply add mask=None argument call signature. mask associated inputs passed layer whenever available. ’s simple example : layer computes softmax time dimension (axis 1) input sequence, discarding masked timesteps.","code":"TemporalSoftmax <- new_layer_class(   \"TemporalSoftmax\",   initialize = function(...) {     super$initialize(...)     self$supports_masking <- TRUE   },   call = function(inputs, mask=NULL) {     if (is.null(mask)) {       stop(\"`TemporalSoftmax` layer requires a previous layer to support masking.\")     }     broadcast_float_mask <- op_expand_dims(op_cast(mask, \"float32\"), -1)     inputs_exp <- op_exp(inputs) * broadcast_float_mask     inputs_sum <- op_sum(inputs_exp * broadcast_float_mask, axis=-1, keepdims=TRUE)     inputs_exp / inputs_sum   } )  inputs <- keras_input(shape = shape(NULL), dtype=\"int32\") outputs <- inputs %>%   layer_embedding(input_dim=10, output_dim=32, mask_zero=TRUE) %>%   layer_dense(1) %>%   TemporalSoftmax()  model <- keras_model(inputs, outputs) y <- model(random_integer(c(32, 100), 0, 10))"},{"path":"https://keras3.posit.co/dev/articles/understanding_masking_and_padding.html","id":"summary","dir":"Articles","previous_headings":"","what":"Summary","title":"Understanding masking & padding","text":"need know padding & masking Keras. recap: “Masking” layers able know skip / ignore certain timesteps sequence inputs. layers mask-generators: Embedding can generate mask input values (mask_zero=TRUE), can Masking layer. layers mask-consumers: expose mask argument call method. case RNN layers. Functional API Sequential API, mask information propagated automatically. using layers standalone way, can pass mask arguments layers manually. can easily write layers modify current mask, generate new mask, consume mask associated inputs.","code":""},{"path":"https://keras3.posit.co/dev/articles/writing_a_custom_training_loop_in_tensorflow.html","id":"setup","dir":"Articles","previous_headings":"","what":"Setup","title":"Writing a training loop from scratch in TensorFlow","text":"","code":"library(tensorflow, exclude = c(\"shape\", \"set_random_seed\")) library(tfdatasets, exclude = \"shape\") library(keras3)  # This guide can only be run with the TensorFlow backend. use_backend(\"tensorflow\")"},{"path":"https://keras3.posit.co/dev/articles/writing_a_custom_training_loop_in_tensorflow.html","id":"introduction","dir":"Articles","previous_headings":"","what":"Introduction","title":"Writing a training loop from scratch in TensorFlow","text":"Keras provides default training evaluation loops, fit() evaluate(). usage covered guide Training & evaluation built-methods. want customize learning algorithm model still leveraging convenience fit() (instance, train GAN using fit()), can subclass Model class implement train_step() method, called repeatedly fit(). covered guide Customizing happens fit(). Now, want low-level control training & evaluation, write training & evaluation loops scratch. guide .","code":""},{"path":"https://keras3.posit.co/dev/articles/writing_a_custom_training_loop_in_tensorflow.html","id":"using-the-gradienttape-a-first-end-to-end-example","dir":"Articles","previous_headings":"","what":"Using the GradientTape: a first end-to-end example","title":"Writing a training loop from scratch in TensorFlow","text":"Calling model inside GradientTape scope enables retrieve gradients trainable weights layer respect loss value. Using optimizer instance, can use gradients update variables (can retrieve using model$trainable_weights). Let’s consider simple MNIST model: Let’s train using mini-batch gradient custom training loop. First, ’re going need optimizer, loss function, dataset: ’s training loop: open loop iterates epochs epoch, iterate dataset, batches batch, open GradientTape() scope Inside scope, call model (forward pass) compute loss Outside scope, retrieve gradients weights model regard loss Finally, use optimizer update weights model based gradients","code":"get_model <- function() {   inputs <- keras_input(shape = 784, name = \"digits\")   outputs <- inputs |>     layer_dense(units = 64, activation = \"relu\") |>     layer_dense(units = 64, activation = \"relu\") |>     layer_dense(units = 10, name = \"predictions\")   keras_model(inputs = inputs, outputs = outputs) }  model <- get_model() # Instantiate an optimizer. optimizer <- optimizer_adam(learning_rate = 1e-3) # Instantiate a loss function. loss_fn <- loss_sparse_categorical_crossentropy(from_logits = TRUE)  # Prepare the training dataset. batch_size <- 64 c(c(x_train, y_train), c(x_test, y_test)) %<-% dataset_mnist() x_train <- array_reshape(x_train, c(-1, 784)) x_test <- array_reshape(x_test, c(-1, 784))  # Reserve 10,000 samples for validation. val_i <- sample.int(nrow(x_train), 10000) x_val <- x_train[val_i,] y_val <- y_train[val_i] x_train %<>% .[-val_i,] y_train %<>% .[-val_i]  # Prepare the training dataset. train_dataset <- list(x_train, y_train) |>   tensor_slices_dataset() |>   dataset_shuffle(buffer_size = 1024) |>   dataset_batch(batch_size)  # Prepare the validation dataset. val_dataset <- list(x_val, y_val) |>   tensor_slices_dataset() |>   dataset_batch(batch_size) epochs <- 3 for (epoch in seq_len(epochs)) {   cat(\"Start of epoch \", epoch, \"\\n\")    # Iterate over the batches of the dataset.   step <- 0   iterator <- as_iterator(train_dataset)   while (!is.null(batch <- iter_next(iterator))) {     c(x_batch_train, y_batch_train) %<-% batch     step <- step + 1     # Open a GradientTape to record the operations run     # during the forward pass, which enables auto-differentiation.     with(tf$GradientTape() %as% tape, {       # Run the forward pass of the layer.       # The operations that the layer applies       # to its inputs are going to be recorded       # on the GradientTape.       logits <- model(x_batch_train, training = TRUE) # Logits for this minibatch        # Compute the loss value for this minibatch.       loss_value <- loss_fn(y_batch_train, logits)     })      # Use the gradient tape to automatically retrieve     # the gradients of the trainable variables with respect to the loss.     gradients <- tape$gradient(loss_value, model$trainable_weights)      # Run one step of gradient descent by updating     # the value of the variables to minimize the loss.     optimizer$apply(gradients, model$trainable_weights)      # Log every 200 batches.     if (step %% 200 == 0) {       cat(sprintf(\"Training loss (for one batch) at step %d: %.4f\\n\",                   step, loss_value))       cat(sprintf(\"Seen so far: %d samples\\n\", (step * batch_size)))     }   } } ## Start of epoch  1 ## Training loss (for one batch) at step 200: 1.1675 ## Seen so far: 12800 samples ## Training loss (for one batch) at step 400: 1.6450 ## Seen so far: 25600 samples ## Training loss (for one batch) at step 600: 0.7421 ## Seen so far: 38400 samples ## Start of epoch  2 ## Training loss (for one batch) at step 200: 0.7254 ## Seen so far: 12800 samples ## Training loss (for one batch) at step 400: 0.2476 ## Seen so far: 25600 samples ## Training loss (for one batch) at step 600: 0.3578 ## Seen so far: 38400 samples ## Start of epoch  3 ## Training loss (for one batch) at step 200: 0.2974 ## Seen so far: 12800 samples ## Training loss (for one batch) at step 400: 0.4441 ## Seen so far: 25600 samples ## Training loss (for one batch) at step 600: 0.1461 ## Seen so far: 38400 samples"},{"path":"https://keras3.posit.co/dev/articles/writing_a_custom_training_loop_in_tensorflow.html","id":"low-level-handling-of-metrics","dir":"Articles","previous_headings":"","what":"Low-level handling of metrics","title":"Writing a training loop from scratch in TensorFlow","text":"Let’s add metrics monitoring basic loop. can readily reuse built-metrics (custom ones wrote) training loops written scratch. ’s flow: Instantiate metric start loop Call metric$update_state() batch Call metric$result() need display current value metric Call metric$reset_state() need clear state metric (typically end epoch) Let’s use knowledge compute SparseCategoricalAccuracy validation data end epoch: ’s training & evaluation loop:","code":"# Get a fresh model model <- get_model()  # Instantiate an optimizer to train the model. optimizer <- optimizer_adam(learning_rate = 1e-3) # Instantiate a loss function. loss_fn <- loss_sparse_categorical_crossentropy(from_logits = TRUE)  # Prepare the metrics. train_acc_metric <- metric_sparse_categorical_accuracy() val_acc_metric <- metric_sparse_categorical_accuracy() epochs <- 2 time <- Sys.time() for (epoch in seq_len(epochs)) {   cat(\"Start of epoch \", epoch, \"\\n\")    # Iterate over the batches of the dataset.   step <- 0   train_dataset_iterator <- as_iterator(train_dataset)   while (!is.null(train_batch <- iter_next(train_dataset_iterator))) {     c(x_batch_train, y_batch_train) %<-% train_batch     step <- step + 1      with(tf$GradientTape() %as% tape, {       logits <- model(x_batch_train, training = TRUE)       loss_value <- loss_fn(y_batch_train, logits)     })     gradients <- tape$gradient(loss_value, model$trainable_weights)     optimizer$apply(gradients, model$trainable_weights)      # Update training metric.     train_acc_metric$update_state(y_batch_train, logits)     if (step %% 200 == 0) {       cat(sprintf(         \"Training loss (for one batch) at step %d: %.4f\\n\", step, loss_value))       cat(sprintf(\"Seen so far: %d samples \\n\", (step * batch_size)))     }   }    # Display metrics at the end of each epoch.   train_acc <- train_acc_metric$result()   cat(sprintf(\"Training acc over epoch: %.4f\\n\", train_acc))    # Reset training metrics at the end of each epoch   train_acc_metric$reset_state()    # Run a validation loop at the end of each epoch.   iterate(val_dataset, function(val_batch) {     c(x_batch_val, y_batch_val) %<-% val_batch     val_logits <- model(x_batch_val, training = FALSE)     val_acc_metric$update_state(y_batch_val, val_logits)   })   val_acc <- val_acc_metric$result()   val_acc_metric$reset_state()   cat(sprintf(\"Validation acc: %.4f\\n\", val_acc)) } ## Start of epoch  1 ## Training loss (for one batch) at step 200: 1.9564 ## Seen so far: 12800 samples ## Training loss (for one batch) at step 400: 1.3562 ## Seen so far: 25600 samples ## Training loss (for one batch) at step 600: 0.4850 ## Seen so far: 38400 samples ## Training acc over epoch: 0.7836 ## Validation acc: 0.8681 ## Start of epoch  2 ## Training loss (for one batch) at step 200: 0.7679 ## Seen so far: 12800 samples ## Training loss (for one batch) at step 400: 0.5674 ## Seen so far: 25600 samples ## Training loss (for one batch) at step 600: 0.4705 ## Seen so far: 38400 samples ## Training acc over epoch: 0.8838 ## Validation acc: 0.9049 Sys.time() - time ## Time difference of 39.32149 secs"},{"path":"https://keras3.posit.co/dev/articles/writing_a_custom_training_loop_in_tensorflow.html","id":"speeding-up-your-training-step-with-tf_function","dir":"Articles","previous_headings":"","what":"Speeding-up your training step with tf_function()","title":"Writing a training loop from scratch in TensorFlow","text":"default runtime TensorFlow 2 eager execution. , training loop executes eagerly. great debugging, graph compilation definite performance advantage. Describing computation static graph enables framework apply global performance optimizations. impossible framework constrained greedily execute one operation another, knowledge comes next. can compile static graph function takes tensors input. Just add @tf.function decorator , like : Let’s evaluation step: Now, let’s re-run training loop compiled training step: Much faster, isn’t ?","code":"train_step <- tf_function(function(x, y) {   with(tf$GradientTape() %as% tape, {     logits <- model(x, training = TRUE)     loss_value <- loss_fn(y, logits)   })   gradients <- tape$gradient(loss_value, model$trainable_weights)   optimizer$apply(gradients, model$trainable_weights)   train_acc_metric$update_state(y, logits)   invisible(NULL) # return nothing }) test_step <- tf_function(function(x, y) {   val_logits <- model(x, training=FALSE)   val_acc_metric$update_state(y, val_logits)   invisible(NULL) # return nothing }) epochs <- 2 time <- Sys.time() for (epoch in seq_len(epochs)) {   cat(\"Start of epoch \", epoch, \"\\n\")    # Iterate over the batches of the dataset.   step <- 0   while (!is.null(batch <- iter_next(iterator))) {     c(x_batch_train, y_batch_train) %<-% batch     step <- step + 1     train_step(x_batch_train, y_batch_train)      if (step %% 200 == 0) {       cat(sprintf(         \"Training loss (for one batch) at step %d: %.4f\\n\", step, loss_value       ))       cat(sprintf(\"Seen so far: %d samples \\n\", (step * batch_size)))     }   }    # Display metrics at the end of each epoch.   train_acc <- train_acc_metric$result()   cat(sprintf(\"Training acc over epoch: %.4f\\n\", train_acc))     # Reset training metrics at the end of each epoch   train_acc_metric$reset_state()    # Run a validation loop at the end of each epoch.    iterate(val_dataset, function(val_batch) {     c(x_batch_val, y_batch_val) %<-% val_batch     test_step(x_batch_val, y_batch_val)   })    val_acc <- val_acc_metric$result()   val_acc_metric$reset_state()   cat(sprintf(\"Validation acc: %.4f\\n\", val_acc)) } ## Start of epoch  1 ## Training acc over epoch: 0.0000 ## Validation acc: 0.9049 ## Start of epoch  2 ## Training acc over epoch: 0.0000 ## Validation acc: 0.9049 Sys.time() - time ## Time difference of 0.4121809 secs"},{"path":"https://keras3.posit.co/dev/articles/writing_a_custom_training_loop_in_tensorflow.html","id":"low-level-handling-of-losses-tracked-by-the-model","dir":"Articles","previous_headings":"","what":"Low-level handling of losses tracked by the model","title":"Writing a training loop from scratch in TensorFlow","text":"Layers models recursively track losses created forward pass layers call self$add_loss(value). resulting list scalar loss values available via property model$losses end forward pass. want using loss components, sum add main loss training step. Consider layer, creates activity regularization loss: Let’s build really simple model uses : ’s training step look like now:","code":"layer_activity_regularization <- Layer(   \"ActivityRegularizationLayer\",   call = function(inputs) {     self$add_loss(0.1 * op_mean(inputs))     inputs   } ) inputs <- keras_input(shape = 784, name=\"digits\") outputs <- inputs |>   layer_dense(units = 64, activation = \"relu\") |>   layer_activity_regularization() |>   layer_dense(units = 64, activation = \"relu\") |>   layer_dense(units = 10, name = \"predictions\") model <- keras_model(inputs = inputs, outputs = outputs) train_step <- tf_function(function(x, y) {   with(tf$GradientTape() %as% tape, {     logits <- model(x, training = TRUE)     loss_value <- Reduce(`+`, c(loss_fn(y, logits),                                 model$losses))   })   gradients <- tape$gradient(loss_value, model$trainable_weights)   optimizer$apply(gradients, model$trainable_weights)   train_acc_metric$update_state(y, logits)   invisible(NULL) })"},{"path":"https://keras3.posit.co/dev/articles/writing_a_custom_training_loop_in_tensorflow.html","id":"summary","dir":"Articles","previous_headings":"","what":"Summary","title":"Writing a training loop from scratch in TensorFlow","text":"Now know everything know using built-training loops writing scratch. conclude, ’s simple end--end example ties together everything ’ve learned guide: DCGAN trained MNIST digits.","code":""},{"path":"https://keras3.posit.co/dev/articles/writing_a_custom_training_loop_in_tensorflow.html","id":"end-to-end-example-a-gan-training-loop-from-scratch","dir":"Articles","previous_headings":"","what":"End-to-end example: a GAN training loop from scratch","title":"Writing a training loop from scratch in TensorFlow","text":"may familiar Generative Adversarial Networks (GANs). GANs can generate new images look almost real, learning latent distribution training dataset images (“latent space” images). GAN made two parts: “generator” model maps points latent space points image space, “discriminator” model, classifier can tell difference real images (training dataset) fake images (output generator network). GAN training loop looks like : Train discriminator. Sample batch random points latent space. Turn points fake images via “generator” model. Get batch real images combine generated images. Train “discriminator” model classify generated vs. real images. Train generator. Sample random points latent space. Turn points fake images via “generator” network. Get batch real images combine generated images. Train “generator” model “fool” discriminator classify fake images real. much detailed overview GANs works, see Deep Learning Python. Let’s implement training loop. First, create discriminator meant classify fake vs real digits: let’s create generator network, turns latent vectors outputs shape (28, 28, 1) (representing MNIST digits): ’s key bit: training loop. can see quite straightforward. training step function takes 17 lines. Let’s train GAN, repeatedly calling train_step batches images. Since discriminator generator convnets, ’re going want run code GPU. ’s ! ’ll get nice-looking fake MNIST digits just ~30s training GPU.","code":"# Create the discriminator discriminator <-   keras_model_sequential(name = \"discriminator\",                          input_shape = c(28, 28, 1)) |>   layer_conv_2d(filters = 64, kernel_size = c(3, 3),                 strides = c(2, 2), padding = \"same\") |>   layer_activation_leaky_relu(negative_slope = 0.2) |>   layer_conv_2d(filters = 128, kernel_size = c(3, 3),                 strides = c(2, 2), padding = \"same\") |>   layer_activation_leaky_relu(negative_slope = 0.2) |>   layer_global_max_pooling_2d() |>   layer_dense(units = 1)  summary(discriminator) ## Model: \"discriminator\" ## ┏━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━┓ ## ┃ Layer (type)                    ┃ Output Shape           ┃       Param # ┃ ## ┡━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━┩ ## │ conv2d (Conv2D)                 │ (None, 14, 14, 64)     │           640 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ leaky_re_lu (LeakyReLU)         │ (None, 14, 14, 64)     │             0 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ conv2d_1 (Conv2D)               │ (None, 7, 7, 128)      │        73,856 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ leaky_re_lu_1 (LeakyReLU)       │ (None, 7, 7, 128)      │             0 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ global_max_pooling2d            │ (None, 128)            │             0 │ ## │ (GlobalMaxPooling2D)            │                        │               │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ dense_6 (Dense)                 │ (None, 1)              │           129 │ ## └─────────────────────────────────┴────────────────────────┴───────────────┘ ##  Total params: 74,625 (291.50 KB) ##  Trainable params: 74,625 (291.50 KB) ##  Non-trainable params: 0 (0.00 B) latent_dim <- 128L  generator <-   keras_model_sequential(name = \"generator\",                          input_shape = latent_dim) |>   layer_dense(7 * 7 * 128) |>   layer_activation_leaky_relu(negative_slope = 0.2) |>   layer_reshape(target_shape = c(7, 7, 128)) |>   layer_conv_2d_transpose(filters = 128, kernel_size = c(4, 4),                           strides = c(2, 2), padding = \"same\") |>   layer_activation_leaky_relu(negative_slope = 0.2) |>   layer_conv_2d_transpose(filters = 128, kernel_size = c(4, 4),                           strides = c(2, 2), padding = \"same\") |>   layer_activation_leaky_relu(negative_slope = 0.2) |>   layer_conv_2d(filters = 1, kernel_size = c(7, 7), padding = \"same\",                 activation = \"sigmoid\")  summary(generator) ## Model: \"generator\" ## ┏━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━┓ ## ┃ Layer (type)                    ┃ Output Shape           ┃       Param # ┃ ## ┡━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━┩ ## │ dense_7 (Dense)                 │ (None, 6272)           │       809,088 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ leaky_re_lu_2 (LeakyReLU)       │ (None, 6272)           │             0 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ reshape (Reshape)               │ (None, 7, 7, 128)      │             0 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ conv2d_transpose                │ (None, 14, 14, 128)    │       262,272 │ ## │ (Conv2DTranspose)               │                        │               │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ leaky_re_lu_3 (LeakyReLU)       │ (None, 14, 14, 128)    │             0 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ conv2d_transpose_1              │ (None, 28, 28, 128)    │       262,272 │ ## │ (Conv2DTranspose)               │                        │               │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ leaky_re_lu_4 (LeakyReLU)       │ (None, 28, 28, 128)    │             0 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ conv2d_2 (Conv2D)               │ (None, 28, 28, 1)      │         6,273 │ ## └─────────────────────────────────┴────────────────────────┴───────────────┘ ##  Total params: 1,339,905 (5.11 MB) ##  Trainable params: 1,339,905 (5.11 MB) ##  Non-trainable params: 0 (0.00 B) # Instantiate one optimizer for the discriminator and another for the generator. d_optimizer <- optimizer_adam(learning_rate = 0.0003) g_optimizer <- optimizer_adam(learning_rate = 0.0004)  # Instantiate a loss function. loss_fn <- loss_binary_crossentropy(from_logits = TRUE)  train_step <- tf_function(function(real_images) {   # Sample random points in the latent space   c(batch_size, ...) %<-% op_shape(real_images)   random_latent_vectors <-     tf$random$normal(shape(batch_size, latent_dim))    # Decode them to fake images   generated_images <- generator(random_latent_vectors)    # Combine them with real images   combined_images <- tf$concat(list(generated_images, real_images),                                axis = 0L)    # Assemble labels discriminating real from fake images   labels <- tf$concat(list(tf$ones(shape(batch_size, 1)),                            tf$zeros(shape(batch_size, 1))),                       axis = 0L)    # Add random noise to the labels - important trick!   labels %<>% `+`(tf$random$uniform(tf$shape(.), maxval = 0.05))    # Train the discriminator   with(tf$GradientTape() %as% tape, {     predictions <- discriminator(combined_images)     d_loss <- loss_fn(labels, predictions)   })   grads <- tape$gradient(d_loss, discriminator$trainable_weights)   d_optimizer$apply(grads, discriminator$trainable_weights)    # Sample random points in the latent space   random_latent_vectors <-     tf$random$normal(shape(batch_size, latent_dim))    # Assemble labels that say \"all real images\"   misleading_labels <- tf$zeros(shape(batch_size, 1))    # Train the generator (note that we should *not* update the weights   # of the discriminator)!   with(tf$GradientTape() %as% tape, {     predictions <- discriminator(generator(random_latent_vectors))     g_loss <- loss_fn(misleading_labels, predictions)   })    grads <- tape$gradient(g_loss, generator$trainable_weights)   g_optimizer$apply(grads, generator$trainable_weights)    list(d_loss, g_loss, generated_images) }) # Prepare the dataset. We use both the training & test MNIST digits. batch_size <- 64 c(c(x_train, .), c(x_test, .)) %<-% dataset_mnist() all_digits <- op_concatenate(list(x_train, x_test)) all_digits <- op_reshape(all_digits, c(-1, 28, 28, 1)) dataset <- all_digits |>   tensor_slices_dataset() |>   dataset_map(\\(x) op_cast(x, \"float32\") / 255) |>   dataset_shuffle(buffer_size = 1024) |>   dataset_batch(batch_size = batch_size)  epochs <- 1 # In practice you need at least 20 epochs to generate nice digits. save_dir <- \"./\"  for (epoch in seq_len(epochs)) {   cat(\"Start epoch: \", epoch, \"\\n\")   step <- 0   train_iterator <- as_iterator(dataset)   while (!is.null(real_images <- iter_next(train_iterator))) {     step <- step + 1     # Train the discriminator & generator on one batch of real images.     c(d_loss, g_loss, generated_images) %<-% train_step(real_images)      # Logging.     if (step %% 200 == 0) {       # Print metrics       cat(sprintf(\"discriminator loss at step %d: %.2f\\n\", step, d_loss))       cat(sprintf(\"adversarial loss at step %d: %.2f\\n\", step, g_loss))     }      # To limit execution time we stop after 10 steps.     # Remove the lines below to actually train the model!     if (step > 10)       break   } } ## Start epoch:  1"},{"path":"https://keras3.posit.co/dev/articles/writing_your_own_callbacks.html","id":"introduction","dir":"Articles","previous_headings":"","what":"Introduction","title":"Writing your own callbacks","text":"callback powerful tool customize behavior Keras model training, evaluation, inference. Examples include keras.callbacks.TensorBoard visualize training progress results TensorBoard, keras.callbacks.ModelCheckpoint periodically save model training. guide, learn Keras callback , can , can build . provide demos simple callback applications get started.","code":""},{"path":"https://keras3.posit.co/dev/articles/writing_your_own_callbacks.html","id":"setup","dir":"Articles","previous_headings":"","what":"Setup","title":"Writing your own callbacks","text":"","code":"library(keras3)"},{"path":"https://keras3.posit.co/dev/articles/writing_your_own_callbacks.html","id":"keras-callbacks-overview","dir":"Articles","previous_headings":"","what":"Keras callbacks overview","title":"Writing your own callbacks","text":"callbacks subclass keras.callbacks.Callback class, override set methods called various stages training, testing, predicting. Callbacks useful get view internal states statistics model training. can pass list callbacks (keyword argument callbacks) following model methods: fit() evaluate() predict()","code":""},{"path":[]},{"path":[]},{"path":"https://keras3.posit.co/dev/articles/writing_your_own_callbacks.html","id":"on_traintestpredict_beginlogs-null","dir":"Articles","previous_headings":"An overview of callback methods > Global methods","what":"on_(train|test|predict)_begin(logs = NULL)","title":"Writing your own callbacks","text":"Called beginning fit/evaluate/predict.","code":""},{"path":"https://keras3.posit.co/dev/articles/writing_your_own_callbacks.html","id":"on_traintestpredict_endlogs-null","dir":"Articles","previous_headings":"An overview of callback methods > Global methods","what":"on_(train|test|predict)_end(logs = NULL)","title":"Writing your own callbacks","text":"Called end fit/evaluate/predict.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/articles/writing_your_own_callbacks.html","id":"on_traintestpredict_batch_beginbatch-logs-null","dir":"Articles","previous_headings":"An overview of callback methods > Batch-level methods for training/testing/predicting","what":"on_(train|test|predict)_batch_begin(batch, logs = NULL)","title":"Writing your own callbacks","text":"Called right processing batch training/testing/predicting.","code":""},{"path":"https://keras3.posit.co/dev/articles/writing_your_own_callbacks.html","id":"on_traintestpredict_batch_endbatch-logs-null","dir":"Articles","previous_headings":"An overview of callback methods > Batch-level methods for training/testing/predicting","what":"on_(train|test|predict)_batch_end(batch, logs = NULL)","title":"Writing your own callbacks","text":"Called end training/testing/predicting batch. Within method, logs named list containing metrics results.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/articles/writing_your_own_callbacks.html","id":"on_epoch_beginepoch-logs-null","dir":"Articles","previous_headings":"An overview of callback methods > Epoch-level methods (training only)","what":"on_epoch_begin(epoch, logs = NULL)","title":"Writing your own callbacks","text":"Called beginning epoch training.","code":""},{"path":"https://keras3.posit.co/dev/articles/writing_your_own_callbacks.html","id":"on_epoch_endepoch-logs-null","dir":"Articles","previous_headings":"An overview of callback methods > Epoch-level methods (training only)","what":"on_epoch_end(epoch, logs = NULL)","title":"Writing your own callbacks","text":"Called end epoch training.","code":""},{"path":"https://keras3.posit.co/dev/articles/writing_your_own_callbacks.html","id":"a-basic-example","dir":"Articles","previous_headings":"","what":"A basic example","title":"Writing your own callbacks","text":"Let’s take look concrete example. get started, let’s import tensorflow define simple Sequential Keras model: , load MNIST data training testing Keras datasets API: Now, define simple custom callback logs: fit/evaluate/predict starts & ends epoch starts & ends training batch starts & ends evaluation (test) batch starts & ends inference (prediction) batch starts & ends Let’s try :","code":"# Define the Keras model to add callbacks to get_model <- function() {   model <- keras_model_sequential()   model |> layer_dense(units = 1)   model |> compile(     optimizer = optimizer_rmsprop(learning_rate = 0.1),     loss = \"mean_squared_error\",     metrics = \"mean_absolute_error\"   )   model } # Load example MNIST data and pre-process it mnist <- dataset_mnist()  flatten_and_rescale <- function(x) {   x <- array_reshape(x, c(-1, 784))   x <- x / 255   x }  mnist$train$x <- flatten_and_rescale(mnist$train$x) mnist$test$x  <- flatten_and_rescale(mnist$test$x)  # limit to 1000 samples n <- 1000 mnist$train$x <- mnist$train$x[1:n,] mnist$train$y <- mnist$train$y[1:n] mnist$test$x  <- mnist$test$x[1:n,] mnist$test$y  <- mnist$test$y[1:n] show <- function(msg, logs) {   cat(glue::glue(msg, .envir = parent.frame()),       \"got logs: \", sep = \"; \")   str(logs); cat(\"\\n\") }  callback_custom <- Callback(   \"CustomCallback\",   on_train_begin         = \\(logs = NULL)        show(\"Starting training\", logs),   on_epoch_begin         = \\(epoch, logs = NULL) show(\"Start epoch {epoch} of training\", logs),   on_train_batch_begin   = \\(batch, logs = NULL) show(\"...Training: start of batch {batch}\", logs),   on_train_batch_end     = \\(batch, logs = NULL) show(\"...Training: end of batch {batch}\",  logs),   on_epoch_end           = \\(epoch, logs = NULL) show(\"End epoch {epoch} of training\", logs),   on_train_end           = \\(logs = NULL)        show(\"Stop training\", logs),     on_test_begin          = \\(logs = NULL)        show(\"Start testing\", logs),   on_test_batch_begin    = \\(batch, logs = NULL) show(\"...Evaluating: start of batch {batch}\", logs),   on_test_batch_end      = \\(batch, logs = NULL) show(\"...Evaluating: end of batch {batch}\", logs),   on_test_end            = \\(logs = NULL)        show(\"Stop testing\", logs),    on_predict_begin       = \\(logs = NULL)        show(\"Start predicting\", logs),   on_predict_end         = \\(logs = NULL)        show(\"Stop predicting\", logs),   on_predict_batch_begin = \\(batch, logs = NULL) show(\"...Predicting: start of batch {batch}\", logs),   on_predict_batch_end   = \\(batch, logs = NULL) show(\"...Predicting: end of batch {batch}\", logs), ) model <- get_model() model |> fit(   mnist$train$x, mnist$train$y,   batch_size = 128,   epochs = 2,   verbose = 0,   validation_split = 0.5,   callbacks = list(callback_custom()) ) ## Starting training; got logs:  Named list() ## ## Start epoch 1 of training; got logs:  Named list() ## ## ...Training: start of batch 1; got logs:  Named list() ## ## ...Training: end of batch 1; got logs: List of 2 ##  $ loss               : num 25.9 ##  $ mean_absolute_error: num 4.19 ## ## ...Training: start of batch 2; got logs:  Named list() ## ## ...Training: end of batch 2; got logs: List of 2 ##  $ loss               : num 433 ##  $ mean_absolute_error: num 15.5 ## ## ...Training: start of batch 3; got logs:  Named list() ## ## ...Training: end of batch 3; got logs: List of 2 ##  $ loss               : num 297 ##  $ mean_absolute_error: num 11.8 ## ## ...Training: start of batch 4; got logs:  Named list() ## ## ...Training: end of batch 4; got logs: List of 2 ##  $ loss               : num 231 ##  $ mean_absolute_error: num 9.68 ## ## Start testing; got logs:  Named list() ## ## ...Evaluating: start of batch 1; got logs:  Named list() ## ## ...Evaluating: end of batch 1; got logs: List of 2 ##  $ loss               : num 8.1 ##  $ mean_absolute_error: num 2.3 ## ## ...Evaluating: start of batch 2; got logs:  Named list() ## ## ...Evaluating: end of batch 2; got logs: List of 2 ##  $ loss               : num 7.58 ##  $ mean_absolute_error: num 2.23 ## ## ...Evaluating: start of batch 3; got logs:  Named list() ## ## ...Evaluating: end of batch 3; got logs: List of 2 ##  $ loss               : num 7.38 ##  $ mean_absolute_error: num 2.21 ## ## ...Evaluating: start of batch 4; got logs:  Named list() ## ## ...Evaluating: end of batch 4; got logs: List of 2 ##  $ loss               : num 7.3 ##  $ mean_absolute_error: num 2.21 ## ## Stop testing; got logs: List of 2 ##  $ loss               : num 7.3 ##  $ mean_absolute_error: num 2.21 ## ## End epoch 1 of training; got logs: List of 4 ##  $ loss                   : num 231 ##  $ mean_absolute_error    : num 9.68 ##  $ val_loss               : num 7.3 ##  $ val_mean_absolute_error: num 2.21 ## ## Start epoch 2 of training; got logs:  Named list() ## ## ...Training: start of batch 1; got logs:  Named list() ## ## ...Training: end of batch 1; got logs: List of 2 ##  $ loss               : num 7.44 ##  $ mean_absolute_error: num 2.27 ## ## ...Training: start of batch 2; got logs:  Named list() ## ## ...Training: end of batch 2; got logs: List of 2 ##  $ loss               : num 6.81 ##  $ mean_absolute_error: num 2.16 ## ## ...Training: start of batch 3; got logs:  Named list() ## ## ...Training: end of batch 3; got logs: List of 2 ##  $ loss               : num 6.12 ##  $ mean_absolute_error: num 2.06 ## ## ...Training: start of batch 4; got logs:  Named list() ## ## ...Training: end of batch 4; got logs: List of 2 ##  $ loss               : num 6.08 ##  $ mean_absolute_error: num 2.04 ## ## Start testing; got logs:  Named list() ## ## ...Evaluating: start of batch 1; got logs:  Named list() ## ## ...Evaluating: end of batch 1; got logs: List of 2 ##  $ loss               : num 5.54 ##  $ mean_absolute_error: num 1.92 ## ## ...Evaluating: start of batch 2; got logs:  Named list() ## ## ...Evaluating: end of batch 2; got logs: List of 2 ##  $ loss               : num 5.31 ##  $ mean_absolute_error: num 1.87 ## ## ...Evaluating: start of batch 3; got logs:  Named list() ## ## ...Evaluating: end of batch 3; got logs: List of 2 ##  $ loss               : num 5.11 ##  $ mean_absolute_error: num 1.8 ## ## ...Evaluating: start of batch 4; got logs:  Named list() ## ## ...Evaluating: end of batch 4; got logs: List of 2 ##  $ loss               : num 5.15 ##  $ mean_absolute_error: num 1.82 ## ## Stop testing; got logs: List of 2 ##  $ loss               : num 5.15 ##  $ mean_absolute_error: num 1.82 ## ## End epoch 2 of training; got logs: List of 4 ##  $ loss                   : num 6.08 ##  $ mean_absolute_error    : num 2.04 ##  $ val_loss               : num 5.15 ##  $ val_mean_absolute_error: num 1.82 ## ## Stop training; got logs: List of 4 ##  $ loss                   : num 6.08 ##  $ mean_absolute_error    : num 2.04 ##  $ val_loss               : num 5.15 ##  $ val_mean_absolute_error: num 1.82 res <- model |> evaluate(   mnist$test$x, mnist$test$y,   batch_size = 128, verbose = 0,   callbacks = list(callback_custom()) ) ## Start testing; got logs:  Named list() ## ## ...Evaluating: start of batch 1; got logs:  Named list() ## ## ...Evaluating: end of batch 1; got logs: List of 2 ##  $ loss               : num 5.2 ##  $ mean_absolute_error: num 1.84 ## ## ...Evaluating: start of batch 2; got logs:  Named list() ## ## ...Evaluating: end of batch 2; got logs: List of 2 ##  $ loss               : num 4.62 ##  $ mean_absolute_error: num 1.73 ## ## ...Evaluating: start of batch 3; got logs:  Named list() ## ## ...Evaluating: end of batch 3; got logs: List of 2 ##  $ loss               : num 4.61 ##  $ mean_absolute_error: num 1.74 ## ## ...Evaluating: start of batch 4; got logs:  Named list() ## ## ...Evaluating: end of batch 4; got logs: List of 2 ##  $ loss               : num 4.65 ##  $ mean_absolute_error: num 1.75 ## ## ...Evaluating: start of batch 5; got logs:  Named list() ## ## ...Evaluating: end of batch 5; got logs: List of 2 ##  $ loss               : num 4.84 ##  $ mean_absolute_error: num 1.77 ## ## ...Evaluating: start of batch 6; got logs:  Named list() ## ## ...Evaluating: end of batch 6; got logs: List of 2 ##  $ loss               : num 4.76 ##  $ mean_absolute_error: num 1.76 ## ## ...Evaluating: start of batch 7; got logs:  Named list() ## ## ...Evaluating: end of batch 7; got logs: List of 2 ##  $ loss               : num 4.74 ##  $ mean_absolute_error: num 1.76 ## ## ...Evaluating: start of batch 8; got logs:  Named list() ## ## ...Evaluating: end of batch 8; got logs: List of 2 ##  $ loss               : num 4.67 ##  $ mean_absolute_error: num 1.75 ## ## Stop testing; got logs: List of 2 ##  $ loss               : num 4.67 ##  $ mean_absolute_error: num 1.75 res <- model |> predict(   mnist$test$x,   batch_size = 128, verbose = 0,   callbacks = list(callback_custom()) ) ## Start predicting; got logs:  Named list() ## ## ...Predicting: start of batch 1; got logs:  Named list() ## ## ...Predicting: end of batch 1; got logs: List of 1 ##  $ outputs:<tf.Tensor: shape=(128, 1), dtype=float32, numpy=…> ## ## ...Predicting: start of batch 2; got logs:  Named list() ## ## ...Predicting: end of batch 2; got logs: List of 1 ##  $ outputs:<tf.Tensor: shape=(128, 1), dtype=float32, numpy=…> ## ## ...Predicting: start of batch 3; got logs:  Named list() ## ## ...Predicting: end of batch 3; got logs: List of 1 ##  $ outputs:<tf.Tensor: shape=(128, 1), dtype=float32, numpy=…> ## ## ...Predicting: start of batch 4; got logs:  Named list() ## ## ...Predicting: end of batch 4; got logs: List of 1 ##  $ outputs:<tf.Tensor: shape=(128, 1), dtype=float32, numpy=…> ## ## ...Predicting: start of batch 5; got logs:  Named list() ## ## ...Predicting: end of batch 5; got logs: List of 1 ##  $ outputs:<tf.Tensor: shape=(128, 1), dtype=float32, numpy=…> ## ## ...Predicting: start of batch 6; got logs:  Named list() ## ## ...Predicting: end of batch 6; got logs: List of 1 ##  $ outputs:<tf.Tensor: shape=(128, 1), dtype=float32, numpy=…> ## ## ...Predicting: start of batch 7; got logs:  Named list() ## ## ...Predicting: end of batch 7; got logs: List of 1 ##  $ outputs:<tf.Tensor: shape=(128, 1), dtype=float32, numpy=…> ## ## ...Predicting: start of batch 8; got logs:  Named list() ## ## ...Predicting: end of batch 8; got logs: List of 1 ##  $ outputs:<tf.Tensor: shape=(104, 1), dtype=float32, numpy=…> ## ## Stop predicting; got logs:  Named list()"},{"path":"https://keras3.posit.co/dev/articles/writing_your_own_callbacks.html","id":"usage-of-logs-list","dir":"Articles","previous_headings":"A basic example","what":"Usage of logs list","title":"Writing your own callbacks","text":"logs named list contains loss value, metrics end batch epoch. Example includes loss mean absolute error. information callbacks, can check Keras callback API documentation.","code":"callback_print_loss_and_mae <- Callback(   \"LossAndErrorPrintingCallback\",    on_train_batch_end = function(batch, logs = NULL)     cat(sprintf(\"Up to batch %i, the average loss is %7.2f.\\n\",                 batch,  logs$loss)),    on_test_batch_end = function(batch, logs = NULL)     cat(sprintf(\"Up to batch %i, the average loss is %7.2f.\\n\",                 batch, logs$loss)),    on_epoch_end = function(epoch, logs = NULL)     cat(sprintf(       \"The average loss for epoch %2i is %9.2f and mean absolute error is %7.2f.\\n\",       epoch, logs$loss, logs$mean_absolute_error     )) )   model <- get_model() model |> fit(   mnist$train$x, mnist$train$y,   epochs = 2, verbose = 0, batch_size = 128,   callbacks = list(callback_print_loss_and_mae()) ) ## Up to batch 1, the average loss is   25.12. ## Up to batch 2, the average loss is  398.92. ## Up to batch 3, the average loss is  274.04. ## Up to batch 4, the average loss is  208.32. ## Up to batch 5, the average loss is  168.15. ## Up to batch 6, the average loss is  141.31. ## Up to batch 7, the average loss is  122.19. ## Up to batch 8, the average loss is  110.05. ## The average loss for epoch  1 is    110.05 and mean absolute error is    5.79. ## Up to batch 1, the average loss is    4.71. ## Up to batch 2, the average loss is    4.74. ## Up to batch 3, the average loss is    4.81. ## Up to batch 4, the average loss is    5.07. ## Up to batch 5, the average loss is    5.08. ## Up to batch 6, the average loss is    5.09. ## Up to batch 7, the average loss is    5.19. ## Up to batch 8, the average loss is    5.51. ## The average loss for epoch  2 is      5.51 and mean absolute error is    1.90. res = model |> evaluate(   mnist$test$x, mnist$test$y,   verbose = 0, batch_size = 128,   callbacks = list(callback_print_loss_and_mae()) ) ## Up to batch 1, the average loss is   15.86. ## Up to batch 2, the average loss is   16.13. ## Up to batch 3, the average loss is   16.02. ## Up to batch 4, the average loss is   16.11. ## Up to batch 5, the average loss is   16.23. ## Up to batch 6, the average loss is   16.68. ## Up to batch 7, the average loss is   16.61. ## Up to batch 8, the average loss is   16.54."},{"path":"https://keras3.posit.co/dev/articles/writing_your_own_callbacks.html","id":"usage-of-selfmodel-attribute","dir":"Articles","previous_headings":"","what":"Usage of self$model attribute","title":"Writing your own callbacks","text":"addition receiving log information one methods called, callbacks access model associated current round training/evaluation/inference: self$model. things can self$model callback: Set self$model$stop_training <- TRUE immediately interrupt training. Mutate hyperparameters optimizer (available self$model$optimizer), self$model$optimizer$learning_rate. Save model period intervals. Record output model |> predict() test samples end epoch, use sanity check training. Extract visualizations intermediate features end epoch, monitor model learning time. etc. Let’s see action couple examples.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/articles/writing_your_own_callbacks.html","id":"early-stopping-at-minimum-loss","dir":"Articles","previous_headings":"Examples of Keras callback applications","what":"Early stopping at minimum loss","title":"Writing your own callbacks","text":"first example shows creation Callback stops training minimum loss reached, setting attribute self$model$stop_training (boolean). Optionally, can provide argument patience specify many epochs wait stopping reached local minimum. callback_early_stopping() provides complete general implementation.","code":"callback_early_stopping_at_min_loss <- Callback(   \"EarlyStoppingAtMinLoss\",   `__doc__` =     \"Stop training when the loss is at its min, i.e. the loss stops decreasing.      Arguments:         patience: Number of epochs to wait after min has been hit. After this         number of no improvement, training stops.     \",    initialize = function(patience = 0) {     super$initialize()     self$patience <- patience     # best_weights to store the weights at which the minimum loss occurs.     self$best_weights <- NULL   },    on_train_begin = function(logs = NULL) {     # The number of epoch it has waited when loss is no longer minimum.     self$wait <- 0     # The epoch the training stops at.     self$stopped_epoch <- 0     # Initialize the best as infinity.     self$best <- Inf   },    on_epoch_end = function(epoch, logs = NULL) {     current <- logs$loss     if (current < self$best) {       self$best <- current       self$wait <- 0L       # Record the best weights if current results is better (less).       self$best_weights <- get_weights(self$model)     } else {       add(self$wait) <- 1L       if (self$wait >= self$patience) {         self$stopped_epoch <- epoch         self$model$stop_training <- TRUE         cat(\"Restoring model weights from the end of the best epoch.\\n\")         model$set_weights(self$best_weights)       }     }   },    on_train_end = function(logs = NULL)     if (self$stopped_epoch > 0)       cat(sprintf(\"Epoch %05d: early stopping\\n\", self$stopped_epoch + 1)) ) `add<-` <- `+`   model <- get_model() model |> fit(   mnist$train$x,   mnist$train$y,   epochs = 30,   batch_size = 64,   verbose = 0,   callbacks = list(callback_print_loss_and_mae(),                    callback_early_stopping_at_min_loss()) ) ## Up to batch 1, the average loss is   30.54. ## Up to batch 2, the average loss is  513.27. ## Up to batch 3, the average loss is  352.60. ## Up to batch 4, the average loss is  266.37. ## Up to batch 5, the average loss is  214.68. ## Up to batch 6, the average loss is  179.97. ## Up to batch 7, the average loss is  155.06. ## Up to batch 8, the average loss is  136.59. ## Up to batch 9, the average loss is  121.96. ## Up to batch 10, the average loss is  110.28. ## Up to batch 11, the average loss is  100.72. ## Up to batch 12, the average loss is   92.71. ## Up to batch 13, the average loss is   85.95. ## Up to batch 14, the average loss is   80.21. ## Up to batch 15, the average loss is   75.17. ## Up to batch 16, the average loss is   72.48. ## The average loss for epoch  1 is     72.48 and mean absolute error is    4.08. ## Up to batch 1, the average loss is    7.98. ## Up to batch 2, the average loss is    9.92. ## Up to batch 3, the average loss is   12.88. ## Up to batch 4, the average loss is   16.61. ## Up to batch 5, the average loss is   20.49. ## Up to batch 6, the average loss is   26.14. ## Up to batch 7, the average loss is   30.44. ## Up to batch 8, the average loss is   33.76. ## Up to batch 9, the average loss is   36.32. ## Up to batch 10, the average loss is   35.26. ## Up to batch 11, the average loss is   34.22. ## Up to batch 12, the average loss is   33.53. ## Up to batch 13, the average loss is   32.84. ## Up to batch 14, the average loss is   31.80. ## Up to batch 15, the average loss is   31.39. ## Up to batch 16, the average loss is   31.45. ## The average loss for epoch  2 is     31.45 and mean absolute error is    4.82. ## Up to batch 1, the average loss is   39.60. ## Up to batch 2, the average loss is   41.95. ## Up to batch 3, the average loss is   41.29. ## Up to batch 4, the average loss is   36.77. ## Up to batch 5, the average loss is   32.08. ## Up to batch 6, the average loss is   28.17. ## Up to batch 7, the average loss is   25.33. ## Up to batch 8, the average loss is   23.56. ## Up to batch 9, the average loss is   22.28. ## Up to batch 10, the average loss is   21.22. ## Up to batch 11, the average loss is   20.87. ## Up to batch 12, the average loss is   22.25. ## Up to batch 13, the average loss is   25.08. ## Up to batch 14, the average loss is   27.87. ## Up to batch 15, the average loss is   31.72. ## Up to batch 16, the average loss is   33.21. ## The average loss for epoch  3 is     33.21 and mean absolute error is    4.79. ## Restoring model weights from the end of the best epoch. ## Epoch 00004: early stopping"},{"path":"https://keras3.posit.co/dev/articles/writing_your_own_callbacks.html","id":"learning-rate-scheduling","dir":"Articles","previous_headings":"Examples of Keras callback applications","what":"Learning rate scheduling","title":"Writing your own callbacks","text":"example, show custom Callback can used dynamically change learning rate optimizer course training. See keras$callbacks$LearningRateScheduler general implementations (RStudio, press F1 cursor LearningRateScheduler browser open page).","code":"callback_custom_learning_rate_scheduler <- Callback(   \"CustomLearningRateScheduler\",   `__doc__` =   \"Learning rate scheduler which sets the learning rate according to schedule.      Arguments:         schedule: a function that takes an epoch index             (integer, indexed from 0) and current learning rate             as inputs and returns a new learning rate as output (float).     \",    initialize = function(schedule) {     super$initialize()     self$schedule <- schedule   },    on_epoch_begin = function(epoch, logs = NULL) {     ## When in doubt about what types of objects are in scope (e.g., self$model)     ## use a debugger to interact with the actual objects at the console!     # browser()      if (!\"learning_rate\" %in% names(self$model$optimizer))       stop('Optimizer must have a \"learning_rate\" attribute.')      # # Get the current learning rate from model's optimizer.     # use as.numeric() to convert the keras variablea to an R numeric     lr <- as.numeric(self$model$optimizer$learning_rate)     # # Call schedule function to get the scheduled learning rate.     scheduled_lr <- self$schedule(epoch, lr)     # # Set the value back to the optimizer before this epoch starts     optimizer <- self$model$optimizer     optimizer$learning_rate <- scheduled_lr     cat(sprintf(\"\\nEpoch %03d: Learning rate is %6.4f.\\n\", epoch, scheduled_lr))   } )  LR_SCHEDULE <- tibble::tribble(   ~start_epoch, ~learning_rate,              0,            0.1,              3,           0.05,              6,           0.01,              9,          0.005,             12,          0.001,   )  last <- function(x) x[length(x)] lr_schedule <- function(epoch, learning_rate) {   \"Helper function to retrieve the scheduled learning rate based on epoch.\"   with(LR_SCHEDULE, learning_rate[last(which(epoch >= start_epoch))]) }  model <- get_model() model |> fit(   mnist$train$x,   mnist$train$y,   epochs = 14,   batch_size = 64,   verbose = 0,   callbacks = list(     callback_print_loss_and_mae(),     callback_custom_learning_rate_scheduler(lr_schedule)   ) ) ## ## Epoch 001: Learning rate is 0.1000. ## Up to batch 1, the average loss is   29.36. ## Up to batch 2, the average loss is  513.95. ## Up to batch 3, the average loss is  352.70. ## Up to batch 4, the average loss is  266.46. ## Up to batch 5, the average loss is  214.73. ## Up to batch 6, the average loss is  180.00. ## Up to batch 7, the average loss is  155.05. ## Up to batch 8, the average loss is  136.64. ## Up to batch 9, the average loss is  121.97. ## Up to batch 10, the average loss is  110.30. ## Up to batch 11, the average loss is  100.76. ## Up to batch 12, the average loss is   92.74. ## Up to batch 13, the average loss is   85.95. ## Up to batch 14, the average loss is   80.18. ## Up to batch 15, the average loss is   75.11. ## Up to batch 16, the average loss is   72.38. ## The average loss for epoch  1 is     72.38 and mean absolute error is    4.04. ## ## Epoch 002: Learning rate is 0.1000. ## Up to batch 1, the average loss is    6.95. ## Up to batch 2, the average loss is    8.71. ## Up to batch 3, the average loss is   11.42. ## Up to batch 4, the average loss is   15.15. ## Up to batch 5, the average loss is   19.28. ## Up to batch 6, the average loss is   25.54. ## Up to batch 7, the average loss is   30.38. ## Up to batch 8, the average loss is   33.95. ## Up to batch 9, the average loss is   36.58. ## Up to batch 10, the average loss is   35.46. ## Up to batch 11, the average loss is   34.34. ## Up to batch 12, the average loss is   33.51. ## Up to batch 13, the average loss is   32.67. ## Up to batch 14, the average loss is   31.54. ## Up to batch 15, the average loss is   31.05. ## Up to batch 16, the average loss is   31.09. ## The average loss for epoch  2 is     31.09 and mean absolute error is    4.77. ## ## Epoch 003: Learning rate is 0.0500. ## Up to batch 1, the average loss is   40.40. ## Up to batch 2, the average loss is   22.33. ## Up to batch 3, the average loss is   16.18. ## Up to batch 4, the average loss is   13.09. ## Up to batch 5, the average loss is   11.48. ## Up to batch 6, the average loss is   10.21. ## Up to batch 7, the average loss is    9.22. ## Up to batch 8, the average loss is    8.70. ## Up to batch 9, the average loss is    8.16. ## Up to batch 10, the average loss is    7.80. ## Up to batch 11, the average loss is    7.50. ## Up to batch 12, the average loss is    7.17. ## Up to batch 13, the average loss is    6.89. ## Up to batch 14, the average loss is    6.70. ## Up to batch 15, the average loss is    6.52. ## Up to batch 16, the average loss is    6.54. ## The average loss for epoch  3 is      6.54 and mean absolute error is    1.93. ## ## Epoch 004: Learning rate is 0.0500. ## Up to batch 1, the average loss is    8.74. ## Up to batch 2, the average loss is    8.34. ## Up to batch 3, the average loss is    9.09. ## Up to batch 4, the average loss is    9.72. ## Up to batch 5, the average loss is   10.48. ## Up to batch 6, the average loss is   11.69. ## Up to batch 7, the average loss is   11.83. ## Up to batch 8, the average loss is   11.56. ## Up to batch 9, the average loss is   11.24. ## Up to batch 10, the average loss is   10.84. ## Up to batch 11, the average loss is   10.66. ## Up to batch 12, the average loss is   10.44. ## Up to batch 13, the average loss is   10.21. ## Up to batch 14, the average loss is   10.06. ## Up to batch 15, the average loss is   10.00. ## Up to batch 16, the average loss is   10.20. ## The average loss for epoch  4 is     10.20 and mean absolute error is    2.71. ## ## Epoch 005: Learning rate is 0.0500. ## Up to batch 1, the average loss is   17.26. ## Up to batch 2, the average loss is   14.09. ## Up to batch 3, the average loss is   12.67. ## Up to batch 4, the average loss is   11.44. ## Up to batch 5, the average loss is   10.54. ## Up to batch 6, the average loss is   10.10. ## Up to batch 7, the average loss is    9.53. ## Up to batch 8, the average loss is    9.17. ## Up to batch 9, the average loss is    8.78. ## Up to batch 10, the average loss is    8.49. ## Up to batch 11, the average loss is    8.50. ## Up to batch 12, the average loss is    8.59. ## Up to batch 13, the average loss is    8.68. ## Up to batch 14, the average loss is    8.86. ## Up to batch 15, the average loss is    9.17. ## Up to batch 16, the average loss is    9.53. ## The average loss for epoch  5 is      9.53 and mean absolute error is    2.58. ## ## Epoch 006: Learning rate is 0.0100. ## Up to batch 1, the average loss is   17.04. ## Up to batch 2, the average loss is   14.85. ## Up to batch 3, the average loss is   11.53. ## Up to batch 4, the average loss is    9.65. ## Up to batch 5, the average loss is    8.44. ## Up to batch 6, the average loss is    7.50. ## Up to batch 7, the average loss is    6.74. ## Up to batch 8, the average loss is    6.56. ## Up to batch 9, the average loss is    6.18. ## Up to batch 10, the average loss is    5.87. ## Up to batch 11, the average loss is    5.63. ## Up to batch 12, the average loss is    5.45. ## Up to batch 13, the average loss is    5.23. ## Up to batch 14, the average loss is    5.12. ## Up to batch 15, the average loss is    4.96. ## Up to batch 16, the average loss is    4.91. ## The average loss for epoch  6 is      4.91 and mean absolute error is    1.67. ## ## Epoch 007: Learning rate is 0.0100. ## Up to batch 1, the average loss is    3.65. ## Up to batch 2, the average loss is    3.04. ## Up to batch 3, the average loss is    2.88. ## Up to batch 4, the average loss is    2.85. ## Up to batch 5, the average loss is    2.88. ## Up to batch 6, the average loss is    2.81. ## Up to batch 7, the average loss is    2.70. ## Up to batch 8, the average loss is    2.96. ## Up to batch 9, the average loss is    2.96. ## Up to batch 10, the average loss is    2.93. ## Up to batch 11, the average loss is    2.95. ## Up to batch 12, the average loss is    2.98. ## Up to batch 13, the average loss is    2.97. ## Up to batch 14, the average loss is    3.01. ## Up to batch 15, the average loss is    3.00. ## Up to batch 16, the average loss is    3.05. ## The average loss for epoch  7 is      3.05 and mean absolute error is    1.34. ## ## Epoch 008: Learning rate is 0.0100. ## Up to batch 1, the average loss is    3.69. ## Up to batch 2, the average loss is    3.21. ## Up to batch 3, the average loss is    3.00. ## Up to batch 4, the average loss is    2.91. ## Up to batch 5, the average loss is    2.94. ## Up to batch 6, the average loss is    2.85. ## Up to batch 7, the average loss is    2.72. ## Up to batch 8, the average loss is    2.95. ## Up to batch 9, the average loss is    2.97. ## Up to batch 10, the average loss is    2.93. ## Up to batch 11, the average loss is    2.96. ## Up to batch 12, the average loss is    2.98. ## Up to batch 13, the average loss is    2.99. ## Up to batch 14, the average loss is    3.05. ## Up to batch 15, the average loss is    3.08. ## Up to batch 16, the average loss is    3.14. ## The average loss for epoch  8 is      3.14 and mean absolute error is    1.36. ## ## Epoch 009: Learning rate is 0.0050. ## Up to batch 1, the average loss is    3.71. ## Up to batch 2, the average loss is    2.93. ## Up to batch 3, the average loss is    2.76. ## Up to batch 4, the average loss is    2.70. ## Up to batch 5, the average loss is    2.76. ## Up to batch 6, the average loss is    2.69. ## Up to batch 7, the average loss is    2.57. ## Up to batch 8, the average loss is    2.79. ## Up to batch 9, the average loss is    2.80. ## Up to batch 10, the average loss is    2.77. ## Up to batch 11, the average loss is    2.79. ## Up to batch 12, the average loss is    2.80. ## Up to batch 13, the average loss is    2.78. ## Up to batch 14, the average loss is    2.81. ## Up to batch 15, the average loss is    2.80. ## Up to batch 16, the average loss is    2.83. ## The average loss for epoch  9 is      2.83 and mean absolute error is    1.28. ## ## Epoch 010: Learning rate is 0.0050. ## Up to batch 1, the average loss is    3.02. ## Up to batch 2, the average loss is    2.69. ## Up to batch 3, the average loss is    2.58. ## Up to batch 4, the average loss is    2.57. ## Up to batch 5, the average loss is    2.65. ## Up to batch 6, the average loss is    2.60. ## Up to batch 7, the average loss is    2.48. ## Up to batch 8, the average loss is    2.72. ## Up to batch 9, the average loss is    2.74. ## Up to batch 10, the average loss is    2.71. ## Up to batch 11, the average loss is    2.74. ## Up to batch 12, the average loss is    2.75. ## Up to batch 13, the average loss is    2.74. ## Up to batch 14, the average loss is    2.77. ## Up to batch 15, the average loss is    2.77. ## Up to batch 16, the average loss is    2.80. ## The average loss for epoch 10 is      2.80 and mean absolute error is    1.27. ## ## Epoch 011: Learning rate is 0.0050. ## Up to batch 1, the average loss is    3.01. ## Up to batch 2, the average loss is    2.69. ## Up to batch 3, the average loss is    2.58. ## Up to batch 4, the average loss is    2.56. ## Up to batch 5, the average loss is    2.63. ## Up to batch 6, the average loss is    2.58. ## Up to batch 7, the average loss is    2.47. ## Up to batch 8, the average loss is    2.70. ## Up to batch 9, the average loss is    2.72. ## Up to batch 10, the average loss is    2.69. ## Up to batch 11, the average loss is    2.71. ## Up to batch 12, the average loss is    2.72. ## Up to batch 13, the average loss is    2.71. ## Up to batch 14, the average loss is    2.75. ## Up to batch 15, the average loss is    2.74. ## Up to batch 16, the average loss is    2.77. ## The average loss for epoch 11 is      2.77 and mean absolute error is    1.27. ## ## Epoch 012: Learning rate is 0.0010. ## Up to batch 1, the average loss is    2.96. ## Up to batch 2, the average loss is    2.53. ## Up to batch 3, the average loss is    2.47. ## Up to batch 4, the average loss is    2.46. ## Up to batch 5, the average loss is    2.54. ## Up to batch 6, the average loss is    2.48. ## Up to batch 7, the average loss is    2.39. ## Up to batch 8, the average loss is    2.60. ## Up to batch 9, the average loss is    2.62. ## Up to batch 10, the average loss is    2.59. ## Up to batch 11, the average loss is    2.61. ## Up to batch 12, the average loss is    2.62. ## Up to batch 13, the average loss is    2.60. ## Up to batch 14, the average loss is    2.64. ## Up to batch 15, the average loss is    2.62. ## Up to batch 16, the average loss is    2.64. ## The average loss for epoch 12 is      2.64 and mean absolute error is    1.24. ## ## Epoch 013: Learning rate is 0.0010. ## Up to batch 1, the average loss is    2.82. ## Up to batch 2, the average loss is    2.46. ## Up to batch 3, the average loss is    2.42. ## Up to batch 4, the average loss is    2.42. ## Up to batch 5, the average loss is    2.50. ## Up to batch 6, the average loss is    2.45. ## Up to batch 7, the average loss is    2.36. ## Up to batch 8, the average loss is    2.57. ## Up to batch 9, the average loss is    2.59. ## Up to batch 10, the average loss is    2.57. ## Up to batch 11, the average loss is    2.59. ## Up to batch 12, the average loss is    2.60. ## Up to batch 13, the average loss is    2.59. ## Up to batch 14, the average loss is    2.62. ## Up to batch 15, the average loss is    2.61. ## Up to batch 16, the average loss is    2.63. ## The average loss for epoch 13 is      2.63 and mean absolute error is    1.23. ## ## Epoch 014: Learning rate is 0.0010. ## Up to batch 1, the average loss is    2.79. ## Up to batch 2, the average loss is    2.44. ## Up to batch 3, the average loss is    2.40. ## Up to batch 4, the average loss is    2.41. ## Up to batch 5, the average loss is    2.49. ## Up to batch 6, the average loss is    2.44. ## Up to batch 7, the average loss is    2.34. ## Up to batch 8, the average loss is    2.56. ## Up to batch 9, the average loss is    2.58. ## Up to batch 10, the average loss is    2.56. ## Up to batch 11, the average loss is    2.58. ## Up to batch 12, the average loss is    2.59. ## Up to batch 13, the average loss is    2.58. ## Up to batch 14, the average loss is    2.61. ## Up to batch 15, the average loss is    2.60. ## Up to batch 16, the average loss is    2.62. ## The average loss for epoch 14 is      2.62 and mean absolute error is    1.23."},{"path":"https://keras3.posit.co/dev/articles/writing_your_own_callbacks.html","id":"built-in-keras-callbacks","dir":"Articles","previous_headings":"Examples of Keras callback applications","what":"Built-in Keras callbacks","title":"Writing your own callbacks","text":"sure check existing Keras callbacks reading API docs. Applications include logging CSV, saving model, visualizing metrics TensorBoard, lot !","code":""},{"path":"https://keras3.posit.co/dev/authors.html","id":null,"dir":"","previous_headings":"","what":"Authors","title":"Authors and Citation","text":"Tomasz Kalinowski. Author, copyright holder, maintainer. Daniel Falbel. Contributor, copyright holder. JJ Allaire. Author, copyright holder. François Chollet. Author, copyright holder. Posit Software, PBC. Copyright holder, funder. Google. Copyright holder, funder. Yuan Tang. Contributor, copyright holder. Wouter Van Der Bijl. Contributor, copyright holder. Martin Studer. Contributor, copyright holder. Sigrid Keydana. Contributor.","code":""},{"path":"https://keras3.posit.co/dev/authors.html","id":"citation","dir":"","previous_headings":"","what":"Citation","title":"Authors and Citation","text":"Kalinowski T, Allaire J, Chollet F (2025). keras3: R Interface 'Keras'. R package version 1.4.0.9000, https://keras3.posit.co/.","code":"@Manual{,   title = {keras3: R Interface to 'Keras'},   author = {Tomasz Kalinowski and JJ Allaire and François Chollet},   year = {2025},   note = {R package version 1.4.0.9000},   url = {https://keras3.posit.co/}, }"},{"path":"https://keras3.posit.co/dev/index.html","id":"r-interface-to-keras","dir":"","previous_headings":"","what":"R Interface to Keras","title":"R Interface to Keras","text":"Keras high-level neural networks API developed focus enabling fast experimentation. able go idea result least possible delay key good research. Keras following key features: Allows code run CPU GPU, seamlessly. User-friendly API makes easy quickly prototype deep learning models. Built-support convolutional networks (computer vision), recurrent networks (sequence processing), combination . Supports arbitrary network architectures: multi-input multi-output models, layer sharing, model sharing, etc. means Keras appropriate building essentially deep learning model, memory network neural Turing machine. See package website https://keras3.posit.co complete documentation.","code":""},{"path":"https://keras3.posit.co/dev/reference/Callback.html","id":null,"dir":"Reference","previous_headings":"","what":"Define a custom Callback class — Callback","title":"Define a custom Callback class — Callback","text":"Callbacks can passed keras methods fit(), evaluate(), predict() order hook various stages model training, evaluation, inference lifecycle. create custom callback, call Callback() override method associated stage interest.","code":""},{"path":"https://keras3.posit.co/dev/reference/Callback.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Define a custom Callback class — Callback","text":"","code":"Callback(   classname,   on_epoch_begin = NULL,   on_epoch_end = NULL,   on_train_begin = NULL,   on_train_end = NULL,   on_train_batch_begin = NULL,   on_train_batch_end = NULL,   on_test_begin = NULL,   on_test_end = NULL,   on_test_batch_begin = NULL,   on_test_batch_end = NULL,   on_predict_begin = NULL,   on_predict_end = NULL,   on_predict_batch_begin = NULL,   on_predict_batch_end = NULL,   ...,   public = list(),   private = list(),   inherit = NULL,   parent_env = parent.frame() )"},{"path":"https://keras3.posit.co/dev/reference/Callback.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Define a custom Callback class — Callback","text":"classname String, name custom class. (Conventionally, CamelCase). on_epoch_begin   Called start epoch. Subclasses override actions run. function called TRAIN mode. Args: epoch: Integer, index epoch. logs: Named List. Currently data passed argument method may change future. on_epoch_end   Called end epoch. Subclasses override actions run. function called TRAIN mode. Args: epoch: Integer, index epoch. logs: Named List, metric results training epoch, validation epoch validation performed. Validation result keys prefixed val_. training epoch, values Model's metrics returned. Example: list(loss = 0.2, accuracy = 0.7). on_train_begin   Called beginning training. Subclasses override actions run. Args: logs: Named list. Currently data passed argument method may change future. on_train_end   Called end training. Subclasses override actions run. Args: logs: Named list. Currently output last call on_epoch_end() passed argument method may change future. on_train_batch_begin   Called beginning training batch fit() methods. Subclasses override actions run. Note steps_per_execution argument compile Model set N, method called every N batches. Args: batch: Integer, index batch within current epoch. logs: Named list. Currently data passed argument method may change future. on_train_batch_end   Called end training batch fit() methods. Subclasses override actions run. Note steps_per_execution argument compile Model set N, method called every N batches. Args: batch: Integer, index batch within current epoch. logs: Named list. Aggregated metric results batch. on_test_begin   Called beginning evaluation validation. Subclasses override actions run. Args: logs: Named list. Currently data passed argument method may change future. on_test_end   Called end evaluation validation. Subclasses override actions run. Args: logs: Named list. Currently output last call on_test_batch_end() passed argument method may change future. on_test_batch_begin   Called beginning batch evaluate() methods. Also called beginning validation batch fit() methods, validation data provided. Subclasses override actions run. Note steps_per_execution argument compile() Model set N, method called every N batches. Args: batch: Integer, index batch within current epoch. logs: Named list. Currently data passed argument method may change future. on_test_batch_end   Called end batch evaluate() methods. Also called end validation batch fit() methods, validation data provided. Subclasses override actions run. Note steps_per_execution argument compile() Model set N, method called every N batches. Args: batch: Integer, index batch within current epoch. logs: Named list. Aggregated metric results batch. on_predict_begin   Called beginning prediction. Subclasses override actions run. Args: logs: Named list. Currently data passed argument method may change future. on_predict_end   Called end prediction. Subclasses override actions run. Args: logs: Named list. Currently data passed argument method may change future. on_predict_batch_begin   Called beginning batch predict() methods. Subclasses override actions run. Note steps_per_execution argument compile() Model set N, method called every N batches. Args: batch: Integer, index batch within current epoch. logs: Named list. Currently data passed argument method may change future. on_predict_batch_end   Called end batch predict() methods. Subclasses override actions run. Note steps_per_execution argument compile Model set N, method called every N batches. Args: batch: Integer, index batch within current epoch. logs: Named list. Aggregated metric results batch. ..., public Additional methods public members custom class. private Named list R objects (typically, functions) include instance private environments. private methods symbols scope public methods (See section \"Symbols Scope\"). instance private environment. objects private invisible Keras framework Python runtime. inherit custom class subclass. default, base keras class. parent_env R environment class methods grandparent.","code":"\\(epoch, logs = NULL) \\(epoch, logs = NULL) \\(logs = NULL) \\(logs = NULL) \\(batch, logs = NULL) \\(batch, logs=NULL) \\(logs = NULL) \\(logs = NULL) \\(batch, logs = NULL) \\(batch, logs = NULL) \\(logs = NULL) \\(logs = NULL) \\(batch, logs = NULL) \\(batch, logs = NULL)"},{"path":"https://keras3.posit.co/dev/reference/Callback.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Define a custom Callback class — Callback","text":"function returns custom Callback instances, similar builtin callback functions.","code":""},{"path":"https://keras3.posit.co/dev/reference/Callback.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Define a custom Callback class — Callback","text":"R function custom methods (public private) following symbols scope: self: Layer instance. super: Layer superclass. private: R environment specific class instance. objects defined invisible Keras framework. __class__ current class type object. also available alias symbol, value supplied Layer(classname = )","code":"training_finished <- FALSE callback_mark_finished <- Callback(\"MarkFinished\",   on_train_end = function(logs = NULL) {     training_finished <<- TRUE   } )  model <- keras_model_sequential(input_shape = c(1)) |>   layer_dense(1) model |> compile(loss = 'mean_squared_error') model |> fit(op_ones(c(1, 1)), op_ones(c(1, 1)),              callbacks = callback_mark_finished()) stopifnot(isTRUE(training_finished))"},{"path":"https://keras3.posit.co/dev/reference/Callback.html","id":"attributes-accessible-via-self-","dir":"Reference","previous_headings":"","what":"Attributes (accessible via self$)","title":"Define a custom Callback class — Callback","text":"params: Named list, Training parameters (e.g. verbosity, batch size, number epochs, ...). model: Instance Model. Reference model trained. logs named list callback methods take argument contain keys quantities relevant current batch epoch (see method-specific docstrings).","code":""},{"path":"https://keras3.posit.co/dev/reference/Callback.html","id":"symbols-in-scope","dir":"Reference","previous_headings":"","what":"Symbols in scope","title":"Define a custom Callback class — Callback","text":"R function custom methods (public private) following symbols scope: self: custom class instance. super: custom class superclass. private: R environment specific class instance. objects assigned invisible Keras framework. __class__ .symbol(classname): custom class type object.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/Constraint.html","id":null,"dir":"Reference","previous_headings":"","what":"Define a custom Constraint class — Constraint","title":"Define a custom Constraint class — Constraint","text":"Base class weight constraints. Constraint() instance works like stateless function. Users subclass Constraint class override call() method, takes single weight parameter return projected version parameter (e.g. normalized clipped). Constraints can used various Keras layers via kernel_constraint bias_constraint arguments. simple example non-negative weight constraint:     Usage layer:","code":"constraint_nonnegative <- Constraint(\"NonNegative\",   call = function(w) {     w * op_cast(w >= 0, dtype = w$dtype)   } ) weight <- op_convert_to_tensor(c(-1, 1)) constraint_nonnegative()(weight) ## tf.Tensor([-0.  1.], shape=(2), dtype=float64) layer_dense(units = 4, kernel_constraint = constraint_nonnegative()) ## <Dense name=dense, built=False> ##  signature: (*args, **kwargs)"},{"path":"https://keras3.posit.co/dev/reference/Constraint.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Define a custom Constraint class — Constraint","text":"","code":"Constraint(   classname,   call = NULL,   get_config = NULL,   ...,   public = list(),   private = list(),   inherit = NULL,   parent_env = parent.frame() )"},{"path":"https://keras3.posit.co/dev/reference/Constraint.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Define a custom Constraint class — Constraint","text":"classname String, name custom class. (Conventionally, CamelCase). call   Applies constraint input weight variable. default, inputs weight variable modified. Users override method implement projection function. Args: w: Input weight variable. Returns: Projected variable (default, returns unmodified inputs). get_config   Function returns named list object config. constraint config named list (JSON-serializable) can used reinstantiate object (via .call(<constraint_class>, <config>)). ..., public Additional methods public members custom class. private Named list R objects (typically, functions) include instance private environments. private methods symbols scope public methods (See section \"Symbols Scope\"). instance private environment. objects private invisible Keras framework Python runtime. inherit custom class subclass. default, base keras class. parent_env R environment class methods grandparent.","code":"\\(w) \\()"},{"path":"https://keras3.posit.co/dev/reference/Constraint.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Define a custom Constraint class — Constraint","text":"function returns Constraint instances, similar builtin constraint functions like constraint_maxnorm().","code":""},{"path":"https://keras3.posit.co/dev/reference/Constraint.html","id":"symbols-in-scope","dir":"Reference","previous_headings":"","what":"Symbols in scope","title":"Define a custom Constraint class — Constraint","text":"R function custom methods (public private) following symbols scope: self: custom class instance. super: custom class superclass. private: R environment specific class instance. objects assigned invisible Keras framework. __class__ .symbol(classname): custom class type object.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/Layer.html","id":null,"dir":"Reference","previous_headings":"","what":"Define a custom Layer class. — Layer","title":"Define a custom Layer class. — Layer","text":"layer callable object takes input one tensors outputs one tensors. involves computation, defined call() method, state (weight variables). State can created: initialize(), instance via self$add_weight(); optional build() method, invoked first call() layer, supplies shape(s) input(s), may known initialization time. Layers recursively composable: assign Layer instance attribute another Layer, outer layer start tracking weights created inner layer. Nested layers instantiated initialize() method build() method. Users just instantiate layer treat callable.","code":""},{"path":"https://keras3.posit.co/dev/reference/Layer.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Define a custom Layer class. — Layer","text":"","code":"Layer(   classname,   initialize = NULL,   call = NULL,   build = NULL,   get_config = NULL,   ...,   public = list(),   private = list(),   inherit = NULL,   parent_env = parent.frame() )"},{"path":"https://keras3.posit.co/dev/reference/Layer.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Define a custom Layer class. — Layer","text":"classname String, name custom class. (Conventionally, CamelCase). initialize, call, build, get_config Recommended methods implement. See description details sections. ..., public Additional methods public members custom class. private Named list R objects (typically, functions) include instance private environments. private methods symbols scope public methods (See section \"Symbols Scope\"). instance private environment. objects private invisible Keras framework Python runtime. inherit custom class subclass. default, base keras class. parent_env R environment class methods grandparent.","code":""},{"path":"https://keras3.posit.co/dev/reference/Layer.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Define a custom Layer class. — Layer","text":"composing layer constructor, similar behavior layer functions like layer_dense(). first argument returned function object, enabling initialize()ing call() layer one step composing layer pipe, like   initialize() layer instance call() , pass missing NULL value object, pass arguments initialize() name.","code":"layer_foo <- Layer(\"Foo\", ....) output <- inputs |> layer_foo() layer <- layer_dense(units = 2, activation = \"relu\") layer <- layer_dense(NULL, 2, activation = \"relu\") layer <- layer_dense(, 2, activation = \"relu\")  # then you can call() the layer in a separate step outputs <- inputs |> layer()"},{"path":"https://keras3.posit.co/dev/reference/Layer.html","id":"symbols-in-scope","dir":"Reference","previous_headings":"","what":"Symbols in scope","title":"Define a custom Layer class. — Layer","text":"R function custom methods (public private) following symbols scope: self: custom class instance. super: custom class superclass. private: R environment specific class instance. objects assigned invisible Keras framework. __class__ .symbol(classname): custom class type object.","code":""},{"path":"https://keras3.posit.co/dev/reference/Layer.html","id":"attributes","dir":"Reference","previous_headings":"","what":"Attributes","title":"Define a custom Layer class. — Layer","text":"name: name layer (string). dtype: Dtype layer's weights. Alias layer$variable_dtype. variable_dtype: Dtype layer's weights. compute_dtype: dtype layer's computations. Layers automatically cast inputs dtype, causes computations output also dtype. mixed precision used keras$mixed_precision$DTypePolicy, different variable_dtype. trainable_weights: List variables included backprop. non_trainable_weights: List variables included backprop. weights: concatenation lists trainable_weights non_trainable_weights (order). trainable: Whether layer trained (boolean), .e. whether potentially-trainable weights returned part layer$trainable_weights. input_spec: Optional (list ) InputSpec object(s) specifying constraints inputs can accepted layer. recommend custom Layers implement following methods: initialize(): Defines custom layer attributes, creates layer weights depend input shapes, using add_weight(), state. build(input_shape): method can used create weights depend shape(s) input(s), using add_weight(), state. Calling call() automatically build layer (built yet) calling build(). call(...): Method called making sure build() called. call() performs logic applying layer input arguments. Two reserved arguments can optionally use call() : training (boolean, whether call inference mode training mode). mask (boolean tensor encoding masked timesteps input, used e.g. RNN layers). typical signature method call(inputs), user optionally add training mask layer need . get_config(): Returns named list containing configuration used initialize layer. list names differ arguments initialize(), override from_config() well. method used saving layer model contains layer.","code":""},{"path":"https://keras3.posit.co/dev/reference/Layer.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Define a custom Layer class. — Layer","text":"basic example: layer two variables, w b, returns y <- (w %*% x) + b. shows implement build() call(). Variables set attributes layer tracked weights layers (layer$weights).   Besides trainable weights, updated via backpropagation training, layers can also non-trainable weights. weights meant updated manually call(). example layer computes running sum inputs:","code":"layer_simple_dense <- Layer(   \"SimpleDense\",   initialize = function(units = 32) {     super$initialize()     self$units <- units   },    # Create the state of the layer (weights)   build = function(input_shape) {     self$kernel <- self$add_weight(       shape = shape(tail(input_shape, 1), self$units),       initializer = \"glorot_uniform\",       trainable = TRUE,       name = \"kernel\"     )     self$bias = self$add_weight(       shape = shape(self$units),       initializer = \"zeros\",       trainable = TRUE,       name = \"bias\"     )   },    # Defines the computation   call = function(self, inputs) {     op_matmul(inputs, self$kernel) + self$bias   } )  # Instantiates the layer. # Supply missing `object` arg to skip invoking `call()` and instead return # the Layer instance linear_layer <- layer_simple_dense(, 4)  # This will call `build(input_shape)` and create the weights, # and then invoke `call()`. y <- linear_layer(op_ones(c(2, 2))) stopifnot(length(linear_layer$weights) == 2)  # These weights are trainable, so they're listed in `trainable_weights`: stopifnot(length(linear_layer$trainable_weights) == 2) layer_compute_sum <- Layer(   classname = \"ComputeSum\",    initialize = function(input_dim) {     super$initialize()      # Create a non-trainable weight.     self$total <- self$add_weight(       shape = shape(),       initializer = \"zeros\",       trainable = FALSE,       name = \"total\"     )   },    call = function(inputs) {     self$total$assign(self$total + op_sum(inputs))     self$total   } )  my_sum <- layer_compute_sum(, 2) x <- op_ones(c(2, 2)) y <- my_sum(x)  stopifnot(exprs = {   all.equal(my_sum$weights,               list(my_sum$total))   all.equal(my_sum$non_trainable_weights, list(my_sum$total))   all.equal(my_sum$trainable_weights,     list()) })"},{"path":"https://keras3.posit.co/dev/reference/Layer.html","id":"methods-available","dir":"Reference","previous_headings":"","what":"Methods available","title":"Define a custom Layer class. — Layer","text":"Initialize self. method typically called custom initialize() method. Example:   Args: trainable: Boolean, whether layer's variables trainable. name: String name layer. dtype: dtype layer's computations weights. Can also keras$DTypePolicy, allows computation weight dtype differ. Defaults NULL. NULL means use config_dtype_policy(), \"float32\" policy unless set different value (via config_set_dtype_policy()).    Can called inside call() method add scalar loss. Example:         Add weight variable layer. Alias add_weight().    Add weight variable layer. Args: shape: shape variable (defined shape()) Must fully-defined (NA/NULL/-1 entries). Defaults () (scalar) unspecified. initializer: Initializer object use populate initial variable value, string name built-initializer (e.g. \"random_normal\"). unspecified, defaults \"glorot_uniform\" floating-point variables \"zeros\" types (e.g. int, bool). dtype: Dtype variable create, e.g. \"float32\". unspecified, defaults layer's variable dtype (defaults \"float32\" unspecified). trainable: Boolean, whether variable trainable via backprop whether updates managed manually. Defaults TRUE. autocast: Boolean, whether autocast layers variables accessing . Defaults TRUE. regularizer: Regularizer object call apply penalty weight. penalties summed loss function optimization. Defaults NULL. constraint: Constraint object call variable optimizer update, string name built-constraint. Defaults NULL. aggregation: Optional string, one NULL, \"none\", \"mean\", \"sum\" \"only_first_replica\". Annotates variable type multi-replica aggregation used variable writing custom data parallel training loops. Defaults \"none\". overwrite_with_gradient: Boolean, whether overwrite variable computed gradient. Useful float8 training. Defaults FALSE. name: String name variable. Useful debugging purposes. Returns: backend tensor, wrapped KerasVariable class. KerasVariable class Methods: assign(value) assign_add(value) assign_sub(value) numpy() (calling .array(<variable>) preferred) Properties/Attributes: value dtype ndim shape (calling shape(<variable>) preferred) trainable       Builds layer's states supplied config (named list args). default, method calls .call(build, config$input_shape) method, creates weights based layer's input shape supplied config. config contains information needed load layer's state, override method. Args: config: Named list containing input shape associated layer.    See description                Count total number scalars composing weights. Returns: integer count.    Returns named list layer's input shape. method returns config (named list) can used build_from_config(config) create states (e.g. Variables Lookup tables) needed layer. default, config contains input shape layer built . writing custom layer creates state unusual way, override method make sure state already created Keras attempts load value upon model loading. Returns: named list containing input shape associated layer.    Returns config object. object config named list (serializable) containing information needed re-instantiate . config expected serializable JSON, expected consist (potentially complex, nested) structure names lists consisting simple objects like strings, ints.    Return values layer$weights list R NumPy arrays.    Currently, Dense, EinsumDense Embedding layers support -place quantization via quantize() method. Example:            Enable rematerialization dynamically layer's call method. Args: layer_call: original call method layer. ...: additional args Returns: rematerialized version layer's call method.    Loads state layer. can override method take full control state layer loaded upon calling load_model(). Args: store: Named list state model loaded.    Saves state layer. can override method take full control state layer saved upon calling save_model(). Args: store: Named list state model saved.    Sets values weights list R NumPy arrays.    Call layer without side effects. Args: trainable_variables: List trainable variables model. non_trainable_variables: List non-trainable variables model. ...: Positional named arguments passed call(). return_losses: TRUE, stateless_call() return list losses created call() part return values. Returns: unnamed list. default, returns list(outputs, non_trainable_variables). return_losses = TRUE, returns list(outputs, non_trainable_variables, losses). Note: non_trainable_variables include non-trainable weights BatchNormalization statistics, also RNG seed state (random operations part layer, dropout), Metric state (metrics attached layer). elements state layer. Example:         Creates layer config. class method, meaning, R function self symbol (class instance) scope. Use __class__ classname symbol provided Layer() constructed) resolve class definition. default implementation :   method reverse get_config(), capable instantiating layer config named list. handle layer connectivity (handled Network), weights (handled set_weights()). Args: config: named list, typically output get_config(). Returns: layer instance.","code":"initialize(...,            activity_regularizer = NULL,            trainable = TRUE,            dtype = NULL,            autocast = TRUE,            name = NULL) layer_my_layer <- Layer(\"MyLayer\",   initialize = function(units, ..., dtype = NULL, name = NULL) {     super$initialize(..., dtype = dtype, name = name)     # .... finish initializing `self` instance   } ) add_loss(loss) Layer(\"MyLayer\",   ...   call = function(x) {     self$add_loss(op_sum(x))     x   } add_metric(...) add_variable(...) add_weight(shape = NULL,            initializer = NULL,            dtype = NULL,            trainable = TRUE,            autocast = TRUE,            regularizer = NULL,            constraint = NULL,            aggregation = 'none',            overwrite_with_gradient = FALSE,            name = NULL) build(input_shape) build_from_config(config) call(...) compute_mask(inputs, previous_mask) compute_output_shape(...) compute_output_spec(...) symbolic_call(...) count_params() get_build_config() get_config() get_weights() quantize(mode, type_check = TRUE) model$quantize(\"int8\") # quantize model in-place model |> predict(data) # faster inference quantized_build(input_shape, mode) quantized_call(...) rematerialized_call(layer_call, ...) load_own_variables(store) save_own_variables(store) set_weights(weights) stateless_call(trainable_variables, non_trainable_variables,                ..., return_losses = FALSE) model <- ... data <- ... trainable_variables <- model$trainable_variables non_trainable_variables <- model$non_trainable_variables # Call the model with zero side effects c(outputs, non_trainable_variables) %<-% model$stateless_call(     trainable_variables,     non_trainable_variables,     data ) # Attach the updated state to the model # (until you do this, the model is still in its pre-call state). purrr::walk2(   model$non_trainable_variables, non_trainable_variables,   \\(variable, value) variable$assign(value)) symbolic_call(...) from_config(config) from_config = function(config) {   do.call(`__class__`, config) }"},{"path":"https://keras3.posit.co/dev/reference/Layer.html","id":"readonly-properties-","dir":"Reference","previous_headings":"","what":"Readonly properties:","title":"Define a custom Layer class. — Layer","text":"compute_dtype dtype computations performed layer. dtype Alias layer$variable_dtype. input_dtype dtype layer inputs converted . losses List scalar losses add_loss(), regularizers sublayers. metrics List metrics. metrics_variables List metric variables. non_trainable_variables List non-trainable layer state. extends layer$non_trainable_weights include state used layer including state metrics SeedGenerators. non_trainable_weights List non-trainable weight variables layer. weights updated optimizer training. Unlike, layer$non_trainable_variables excludes metric state random seeds. trainable_variables List trainable layer state. equivalent layer$trainable_weights. trainable_weights List trainable weight variables layer. weights get updated optimizer training. path path layer. layer built yet, NULL. quantization_mode quantization mode layer, NULL quantized. variable_dtype dtype state (weights) layer. variables List layer state, including random seeds. extends layer$weights include state used layer including SeedGenerators. Note metrics variables included , use metrics_variables visit metric variables. weights List weight variables layer. Unlike, layer$variables excludes metric state random seeds. input Retrieves input tensor(s) symbolic operation. returns tensor(s) corresponding first time operation called. Returns: Input tensor list input tensors. output Retrieves output tensor(s) layer. returns tensor(s) corresponding first time operation called. Returns: Output tensor list output tensors.","code":""},{"path":"https://keras3.posit.co/dev/reference/Layer.html","id":"data-descriptors-attributes-","dir":"Reference","previous_headings":"","what":"Data descriptors (Attributes):","title":"Define a custom Layer class. — Layer","text":"dtype_policy input_spec supports_masking Whether layer supports computing mask using compute_mask. trainable Settable boolean, whether layer trainable .","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/LearningRateSchedule.html","id":null,"dir":"Reference","previous_headings":"","what":"Define a custom LearningRateSchedule class — LearningRateSchedule","title":"Define a custom LearningRateSchedule class — LearningRateSchedule","text":"Subclass Keras LearningRateSchedule base class. can use learning rate schedule modulate learning rate optimizer changes time. Several built-learning rate schedules available, learning_rate_schedule_exponential_decay() learning_rate_schedule_piecewise_constant_decay():   LearningRateSchedule() instance can passed learning_rate argument optimizer. implement schedule object, implement call method, takes step argument (scalar integer backend tensor, current training step count). Note step 0-based (.e., first step 0). Like Keras object, can also optionally make object serializable implementing get_config() from_config() methods.","code":"lr_schedule <- learning_rate_schedule_exponential_decay(   initial_learning_rate = 1e-2,   decay_steps = 10000,   decay_rate = 0.9 ) optimizer <- optimizer_sgd(learning_rate = lr_schedule)"},{"path":"https://keras3.posit.co/dev/reference/LearningRateSchedule.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Define a custom LearningRateSchedule class — LearningRateSchedule","text":"","code":"LearningRateSchedule(   classname,   call = NULL,   initialize = NULL,   get_config = NULL,   ...,   public = list(),   private = list(),   inherit = NULL,   parent_env = parent.frame() )"},{"path":"https://keras3.posit.co/dev/reference/LearningRateSchedule.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Define a custom LearningRateSchedule class — LearningRateSchedule","text":"classname String, name custom class. (Conventionally, CamelCase). call, initialize, get_config Recommended methods implement. See description details sections. ..., public Additional methods public members custom class. private Named list R objects (typically, functions) include instance private environments. private methods symbols scope public methods (See section \"Symbols Scope\"). instance private environment. objects private invisible Keras framework Python runtime. inherit custom class subclass. default, base keras class. parent_env R environment class methods grandparent.","code":""},{"path":"https://keras3.posit.co/dev/reference/LearningRateSchedule.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Define a custom LearningRateSchedule class — LearningRateSchedule","text":"function returns LearningRateSchedule instances, similar built-learning_rate_schedule_* family functions.","code":""},{"path":"https://keras3.posit.co/dev/reference/LearningRateSchedule.html","id":"example","dir":"Reference","previous_headings":"","what":"Example","title":"Define a custom LearningRateSchedule class — LearningRateSchedule","text":"","code":"my_custom_learning_rate_schedule <- LearningRateSchedule(   classname = \"MyLRSchedule\",    initialize = function(initial_learning_rate) {     self$initial_learning_rate <- initial_learning_rate   },    call = function(step) {     # note that `step` is a tensor     # and call() will be traced via tf_function() or similar.      str(step) # <KerasVariable shape=(), dtype=int64, path=SGD/iteration>      # print 'step' every 1000 steps     op_cond((step %% 1000) == 0,             \\() {tensorflow::tf$print(step); NULL},             \\() {NULL})     self$initial_learning_rate / (step + 1)   } )  optimizer <- optimizer_sgd(   learning_rate = my_custom_learning_rate_schedule(0.1) )  # You can also call schedule instances directly # (e.g., for interactive testing, or if implementing a custom optimizer) schedule <- my_custom_learning_rate_schedule(0.1) step <- keras$Variable(initializer = op_ones,                        shape = shape(),                        dtype = \"int64\") schedule(step) ## <Variable path=variable, shape=(), dtype=int64, value=1> ## tf.Tensor(0.0, shape=(), dtype=float64)"},{"path":"https://keras3.posit.co/dev/reference/LearningRateSchedule.html","id":"methods-available-","dir":"Reference","previous_headings":"","what":"Methods available:","title":"Define a custom LearningRateSchedule class — LearningRateSchedule","text":"","code":"get_config()"},{"path":"https://keras3.posit.co/dev/reference/LearningRateSchedule.html","id":"symbols-in-scope","dir":"Reference","previous_headings":"","what":"Symbols in scope","title":"Define a custom LearningRateSchedule class — LearningRateSchedule","text":"R function custom methods (public private) following symbols scope: self: custom class instance. super: custom class superclass. private: R environment specific class instance. objects assigned invisible Keras framework. __class__ .symbol(classname): custom class type object.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/Loss.html","id":null,"dir":"Reference","previous_headings":"","what":"Subclass the base Loss class — Loss","title":"Subclass the base Loss class — Loss","text":"Use define custom loss class. Note, cases need subclass Loss define custom loss: can also pass bare R function, named R function defined custom_metric(), loss function compile().","code":""},{"path":"https://keras3.posit.co/dev/reference/Loss.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Subclass the base Loss class — Loss","text":"","code":"Loss(   classname,   call = NULL,   ...,   public = list(),   private = list(),   inherit = NULL,   parent_env = parent.frame() )"},{"path":"https://keras3.posit.co/dev/reference/Loss.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Subclass the base Loss class — Loss","text":"classname String, name custom class. (Conventionally, CamelCase). call   Method implemented subclasses: Function contains logic loss calculation using y_true, y_pred. ..., public Additional methods public members custom class. private Named list R objects (typically, functions) include instance private environments. private methods symbols scope public methods (See section \"Symbols Scope\"). instance private environment. objects private invisible Keras framework Python runtime. inherit custom class subclass. default, base keras class. parent_env R environment class methods grandparent.","code":"function(y_true, y_pred)"},{"path":"https://keras3.posit.co/dev/reference/Loss.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Subclass the base Loss class — Loss","text":"function returns Loss instances, similar builtin loss functions.","code":""},{"path":"https://keras3.posit.co/dev/reference/Loss.html","id":"details","dir":"Reference","previous_headings":"","what":"Details","title":"Subclass the base Loss class — Loss","text":"Example subclass implementation:","code":"loss_custom_mse <- Loss(   classname = \"CustomMeanSquaredError\",   call = function(y_true, y_pred) {     op_mean(op_square(y_pred - y_true), axis = -1)   } )  # Usage in compile() model <- keras_model_sequential(input_shape = 10) |> layer_dense(10) model |> compile(loss = loss_custom_mse())  # Standalone usage mse <- loss_custom_mse(name = \"my_custom_mse_instance\")  y_true <- op_arange(20) |> op_reshape(c(4, 5)) y_pred <- op_arange(20) |> op_reshape(c(4, 5)) * 2 (loss <- mse(y_true, y_pred)) ## tf.Tensor(143.5, shape=(), dtype=float32) loss2 <- (y_pred - y_true)^2 |>   op_mean(axis = -1) |>   op_mean()  stopifnot(all.equal(as.array(loss), as.array(loss2)))  sample_weight <-array(c(.25, .25, 1, 1)) (weighted_loss <- mse(y_true, y_pred, sample_weight = sample_weight)) ## tf.Tensor(129.0625, shape=(), dtype=float32) weighted_loss2 <- (y_true - y_pred)^2 |>   op_mean(axis = -1) |>   op_multiply(sample_weight) |>   op_mean()  stopifnot(all.equal(as.array(weighted_loss),                     as.array(weighted_loss2)))"},{"path":"https://keras3.posit.co/dev/reference/Loss.html","id":"methods-defined-by-base-loss-class-","dir":"Reference","previous_headings":"","what":"Methods defined by base Loss class:","title":"Subclass the base Loss class — Loss","text":"Args: name: Optional name loss instance. reduction: Type reduction apply loss. almost cases \"sum_over_batch_size\". Supported options \"sum\", \"sum_over_batch_size\", \"mean\", \"mean_with_sample_weight\" NULL. \"sum\" sums loss, \"sum_over_batch_size\" \"mean\" sum loss divide sample size, \"mean_with_sample_weight\" sums loss divides sum sample weights. \"none\" NULL perform aggregation. Defaults \"sum_over_batch_size\". dtype: dtype loss's computations. Defaults NULL, means using config_floatx(). config_floatx() \"float32\" unless set different value (via config_set_floatx()). keras$DTypePolicy provided, compute_dtype utilized.    Call loss instance function, optionally sample_weight.","code":"initialize(name=NULL, reduction=\"sum_over_batch_size\", dtype=NULL) __call__(y_true, y_pred, sample_weight=NULL) get_config()"},{"path":"https://keras3.posit.co/dev/reference/Loss.html","id":"readonly-properties-","dir":"Reference","previous_headings":"","what":"Readonly properties:","title":"Subclass the base Loss class — Loss","text":"","code":"dtype"},{"path":"https://keras3.posit.co/dev/reference/Loss.html","id":"symbols-in-scope","dir":"Reference","previous_headings":"","what":"Symbols in scope","title":"Subclass the base Loss class — Loss","text":"R function custom methods (public private) following symbols scope: self: custom class instance. super: custom class superclass. private: R environment specific class instance. objects assigned invisible Keras framework. __class__ .symbol(classname): custom class type object.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/Metric.html","id":null,"dir":"Reference","previous_headings":"","what":"Subclass the base Metric class — Metric","title":"Subclass the base Metric class — Metric","text":"Metric object encapsulates metric logic state can used track model performance training. returned family metric functions start prefix metric_*, well returned custom metrics defined Metric().","code":""},{"path":"https://keras3.posit.co/dev/reference/Metric.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Subclass the base Metric class — Metric","text":"","code":"Metric(   classname,   initialize = NULL,   update_state = NULL,   result = NULL,   ...,   public = list(),   private = list(),   inherit = NULL,   parent_env = parent.frame() )"},{"path":"https://keras3.posit.co/dev/reference/Metric.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Subclass the base Metric class — Metric","text":"classname String, name custom class. (Conventionally, CamelCase). initialize, update_state, result Recommended methods implement. See description section. ..., public Additional methods public members custom class. private Named list R objects (typically, functions) include instance private environments. private methods symbols scope public methods (See section \"Symbols Scope\"). instance private environment. objects private invisible Keras framework Python runtime. inherit custom class subclass. default, base keras class. parent_env R environment class methods grandparent.","code":""},{"path":"https://keras3.posit.co/dev/reference/Metric.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Subclass the base Metric class — Metric","text":"function returns Metric instances, similar builtin metric functions.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/Metric.html","id":"usage-with-compile-","dir":"Reference","previous_headings":"","what":"Usage with compile():","title":"Subclass the base Metric class — Metric","text":"","code":"model |> compile(   optimizer = 'sgd',   loss = 'mse',   metrics = c(metric_SOME_METRIC(), metric_SOME_OTHER_METRIC()) )"},{"path":"https://keras3.posit.co/dev/reference/Metric.html","id":"standalone-usage-","dir":"Reference","previous_headings":"","what":"Standalone usage:","title":"Subclass the base Metric class — Metric","text":"","code":"m <- metric_SOME_METRIC() for (e in seq(epochs)) {   for (i in seq(train_steps)) {     c(y_true, y_pred, sample_weight = NULL) %<-% ...     m$update_state(y_true, y_pred, sample_weight)   }   cat('Final epoch result: ', as.numeric(m$result()), \"\\n\")   m$reset_state() }"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/Metric.html","id":"usage-with-compile--1","dir":"Reference","previous_headings":"","what":"Usage with compile():","title":"Subclass the base Metric class — Metric","text":"implemented subclasses (custom metrics): initialize(): state variables created method calling self$add_variable() like: self$var <- self$add_variable(...). update_state(): Updates state variables like: self$var$assign(...). result(): Computes returns scalar value named list scalar values metric state variables. Example subclass implementation:","code":"model <- keras_model_sequential() model |>   layer_dense(64, activation = \"relu\") |>   layer_dense(64, activation = \"relu\") |>   layer_dense(10, activation = \"softmax\") model |>   compile(optimizer = optimizer_rmsprop(0.01),           loss = loss_categorical_crossentropy(),           metrics = metric_categorical_accuracy())  data <- random_uniform(c(1000, 32)) labels <- random_uniform(c(1000, 10))  model |> fit(data, labels, verbose = 0) metric_binary_true_positives <- Metric(   classname = \"BinaryTruePositives\",    initialize = function(name = 'binary_true_positives', ...) {     super$initialize(name = name, ...)     self$true_positives <-       self$add_weight(shape = shape(),                       initializer = 'zeros',                       name = 'true_positives')   },    update_state = function(y_true, y_pred, sample_weight = NULL) {     y_true <- op_cast(y_true, \"bool\")     y_pred <- op_cast(y_pred, \"bool\")      values <- y_true & y_pred # `&` calls op_logical_and()     values <- op_cast(values, self$dtype)     if (!is.null(sample_weight)) {       sample_weight <- op_cast(sample_weight, self$dtype)       sample_weight <- op_broadcast_to(sample_weight, shape(values))       values <- values * sample_weight # `*` calls op_multiply()     }     self$true_positives$assign(self$true_positives + op_sum(values))   },    result = function() {     self$true_positives   } ) model <- keras_model_sequential(input_shape = 32) |> layer_dense(10) model |> compile(loss = loss_binary_crossentropy(),                  metrics = list(metric_binary_true_positives())) model |> fit(data, labels, verbose = 0)"},{"path":"https://keras3.posit.co/dev/reference/Metric.html","id":"methods-defined-by-the-base-metric-class-","dir":"Reference","previous_headings":"","what":"Methods defined by the base Metric class:","title":"Subclass the base Metric class — Metric","text":"Calling metric instance self like m(...) equivalent calling:      Initialize self. Args: name: Optional name metric instance. dtype: dtype metric's computations. Defaults NULL, means using config_floatx(). config_floatx() \"float32\" unless set different value (via config_set_floatx()). keras$DTypePolicy provided, compute_dtype utilized.          Return serializable config metric.    Reset metric state variables. function called epochs/steps, metric evaluated training.    Compute current metric value. Returns: scalar tensor, named list scalar tensors.             Accumulate statistics metric.","code":"__call__(...) function(...) {   m$update_state(...)   m$result() } initialize(dtype=NULL, name=NULL) add_variable(shape, initializer, dtype=NULL, aggregation = 'sum', name=NULL) add_weight(shape=shape(), initializer=NULL, dtype=NULL, name=NULL) get_config() reset_state() result() stateless_result(metric_variables) stateless_reset_state() stateless_update_state(metric_variables, ...) update_state(...)"},{"path":"https://keras3.posit.co/dev/reference/Metric.html","id":"readonly-properties","dir":"Reference","previous_headings":"","what":"Readonly properties","title":"Subclass the base Metric class — Metric","text":"dtype variables","code":""},{"path":"https://keras3.posit.co/dev/reference/Metric.html","id":"symbols-in-scope","dir":"Reference","previous_headings":"","what":"Symbols in scope","title":"Subclass the base Metric class — Metric","text":"R function custom methods (public private) following symbols scope: self: custom class instance. super: custom class superclass. private: R environment specific class instance. objects assigned invisible Keras framework. __class__ .symbol(classname): custom class type object.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/Model.html","id":null,"dir":"Reference","previous_headings":"","what":"Subclass the base Keras Model Class — Model","title":"Subclass the base Keras Model Class — Model","text":"advanced use cases need subclass base Model type, e.g., want override train_step() method. just want create define keras model, prefer keras_model() keras_model_sequential(). just want encapsulate custom logic state, need customize training behavior (besides calling self$add_loss() call() method), prefer Layer().","code":""},{"path":"https://keras3.posit.co/dev/reference/Model.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Subclass the base Keras Model Class — Model","text":"","code":"Model(   classname,   initialize = NULL,   call = NULL,   train_step = NULL,   predict_step = NULL,   test_step = NULL,   compute_loss = NULL,   compute_metrics = NULL,   ...,   public = list(),   private = list(),   inherit = NULL,   parent_env = parent.frame() )"},{"path":"https://keras3.posit.co/dev/reference/Model.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Subclass the base Keras Model Class — Model","text":"classname String, name custom class. (Conventionally, CamelCase). initialize, call, train_step, predict_step, test_step, compute_loss, compute_metrics Optional methods can overridden. ..., public Additional methods public members custom class. private Named list R objects (typically, functions) include instance private environments. private methods symbols scope public methods (See section \"Symbols Scope\"). instance private environment. objects private invisible Keras framework Python runtime. inherit custom class subclass. default, base keras class. parent_env R environment class methods grandparent.","code":""},{"path":"https://keras3.posit.co/dev/reference/Model.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Subclass the base Keras Model Class — Model","text":"model constructor function, can call create instance new model type.","code":""},{"path":"https://keras3.posit.co/dev/reference/Model.html","id":"symbols-in-scope","dir":"Reference","previous_headings":"","what":"Symbols in scope","title":"Subclass the base Keras Model Class — Model","text":"R function custom methods (public private) following symbols scope: self: custom class instance. super: custom class superclass. private: R environment specific class instance. objects assigned invisible Keras framework. __class__ .symbol(classname): custom class type object.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/activation_celu.html","id":null,"dir":"Reference","previous_headings":"","what":"Continuously Differentiable Exponential Linear Unit. — activation_celu","title":"Continuously Differentiable Exponential Linear Unit. — activation_celu","text":"CeLU activation function defined : celu(x) = alpha * (exp(x / alpha) - 1) x < 0,celu(x) = x x >= 0. alpha scaling parameter controls activation's shape.","code":""},{"path":"https://keras3.posit.co/dev/reference/activation_celu.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Continuously Differentiable Exponential Linear Unit. — activation_celu","text":"","code":"activation_celu(x, alpha = 1)"},{"path":"https://keras3.posit.co/dev/reference/activation_celu.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Continuously Differentiable Exponential Linear Unit. — activation_celu","text":"x Input tensor. alpha  value CeLU formulation. Defaults 1.0.","code":""},{"path":"https://keras3.posit.co/dev/reference/activation_celu.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Continuously Differentiable Exponential Linear Unit. — activation_celu","text":"tensor, result applying activation input tensor x.","code":""},{"path":"https://keras3.posit.co/dev/reference/activation_celu.html","id":"reference","dir":"Reference","previous_headings":"","what":"Reference","title":"Continuously Differentiable Exponential Linear Unit. — activation_celu","text":"Barron, J. T., 2017","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/activation_elu.html","id":null,"dir":"Reference","previous_headings":"","what":"Exponential Linear Unit. — activation_elu","title":"Exponential Linear Unit. — activation_elu","text":"exponential linear unit (ELU) alpha > 0 defined : x x > 0 alpha * exp(x) - 1 x < 0 ELUs negative values pushes mean activations closer zero. Mean activations closer zero enable faster learning bring gradient closer natural gradient. ELUs saturate negative value argument gets smaller. Saturation means small derivative decreases variation information propagated next layer.","code":""},{"path":"https://keras3.posit.co/dev/reference/activation_elu.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Exponential Linear Unit. — activation_elu","text":"","code":"activation_elu(x, alpha = 1)"},{"path":"https://keras3.posit.co/dev/reference/activation_elu.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Exponential Linear Unit. — activation_elu","text":"x Input tensor. alpha scalar, slope positive section. Defaults 1.0.","code":""},{"path":"https://keras3.posit.co/dev/reference/activation_elu.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Exponential Linear Unit. — activation_elu","text":"tensor, result applying activation input tensor x.","code":""},{"path":"https://keras3.posit.co/dev/reference/activation_elu.html","id":"reference","dir":"Reference","previous_headings":"","what":"Reference","title":"Exponential Linear Unit. — activation_elu","text":"Clevert et al., 2016","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/activation_exponential.html","id":null,"dir":"Reference","previous_headings":"","what":"Exponential activation function. — activation_exponential","title":"Exponential activation function. — activation_exponential","text":"Exponential activation function.","code":""},{"path":"https://keras3.posit.co/dev/reference/activation_exponential.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Exponential activation function. — activation_exponential","text":"","code":"activation_exponential(x)"},{"path":"https://keras3.posit.co/dev/reference/activation_exponential.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Exponential activation function. — activation_exponential","text":"x Input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/activation_exponential.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Exponential activation function. — activation_exponential","text":"tensor, result applying activation input tensor x.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/activation_gelu.html","id":null,"dir":"Reference","previous_headings":"","what":"Gaussian error linear unit (GELU) activation function. — activation_gelu","title":"Gaussian error linear unit (GELU) activation function. — activation_gelu","text":"Gaussian error linear unit (GELU) defined : gelu(x) = x * P(X <= x) P(X) ~ N(0, 1), .e. gelu(x) = 0.5 * x * (1 + erf(x / sqrt(2))). GELU weights inputs value, rather gating inputs sign ReLU.","code":""},{"path":"https://keras3.posit.co/dev/reference/activation_gelu.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Gaussian error linear unit (GELU) activation function. — activation_gelu","text":"","code":"activation_gelu(x, approximate = FALSE)"},{"path":"https://keras3.posit.co/dev/reference/activation_gelu.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Gaussian error linear unit (GELU) activation function. — activation_gelu","text":"x Input tensor. approximate bool, whether enable approximation.","code":""},{"path":"https://keras3.posit.co/dev/reference/activation_gelu.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Gaussian error linear unit (GELU) activation function. — activation_gelu","text":"tensor, result applying activation input tensor x.","code":""},{"path":"https://keras3.posit.co/dev/reference/activation_gelu.html","id":"reference","dir":"Reference","previous_headings":"","what":"Reference","title":"Gaussian error linear unit (GELU) activation function. — activation_gelu","text":"Hendrycks et al., 2016","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/activation_glu.html","id":null,"dir":"Reference","previous_headings":"","what":"Gated Linear Unit (GLU) activation function. — activation_glu","title":"Gated Linear Unit (GLU) activation function. — activation_glu","text":"GLU activation function defined : glu(x) = * sigmoid(b), x split two equal parts b along given axis.","code":""},{"path":"https://keras3.posit.co/dev/reference/activation_glu.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Gated Linear Unit (GLU) activation function. — activation_glu","text":"","code":"activation_glu(x, axis = -1L)"},{"path":"https://keras3.posit.co/dev/reference/activation_glu.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Gated Linear Unit (GLU) activation function. — activation_glu","text":"x Input tensor. axis axis along split input tensor. Defaults -1.","code":""},{"path":"https://keras3.posit.co/dev/reference/activation_glu.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Gated Linear Unit (GLU) activation function. — activation_glu","text":"tensor, result applying activation input tensor x.","code":""},{"path":"https://keras3.posit.co/dev/reference/activation_glu.html","id":"reference","dir":"Reference","previous_headings":"","what":"Reference","title":"Gated Linear Unit (GLU) activation function. — activation_glu","text":"Dauphin et al., 2017","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/activation_hard_shrink.html","id":null,"dir":"Reference","previous_headings":"","what":"Hard Shrink activation function. — activation_hard_shrink","title":"Hard Shrink activation function. — activation_hard_shrink","text":"defined : hard_shrink(x) = x |x| > threshold, hard_shrink(x) = 0 otherwise.","code":""},{"path":"https://keras3.posit.co/dev/reference/activation_hard_shrink.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Hard Shrink activation function. — activation_hard_shrink","text":"","code":"activation_hard_shrink(x, threshold = 0.5)"},{"path":"https://keras3.posit.co/dev/reference/activation_hard_shrink.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Hard Shrink activation function. — activation_hard_shrink","text":"x Input tensor. threshold Threshold value. Defaults 0.5.","code":""},{"path":"https://keras3.posit.co/dev/reference/activation_hard_shrink.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Hard Shrink activation function. — activation_hard_shrink","text":"tensor, result applying activation input tensor x.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/activation_hard_sigmoid.html","id":null,"dir":"Reference","previous_headings":"","what":"Hard sigmoid activation function. — activation_hard_sigmoid","title":"Hard sigmoid activation function. — activation_hard_sigmoid","text":"hard sigmoid activation defined : 0 x <= -3 1 x >= 3 (x/6) + 0.5 -3 < x < 3 faster, piecewise linear approximation sigmoid activation.","code":""},{"path":"https://keras3.posit.co/dev/reference/activation_hard_sigmoid.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Hard sigmoid activation function. — activation_hard_sigmoid","text":"","code":"activation_hard_sigmoid(x)"},{"path":"https://keras3.posit.co/dev/reference/activation_hard_sigmoid.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Hard sigmoid activation function. — activation_hard_sigmoid","text":"x Input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/activation_hard_sigmoid.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Hard sigmoid activation function. — activation_hard_sigmoid","text":"tensor, result applying activation input tensor x.","code":""},{"path":"https://keras3.posit.co/dev/reference/activation_hard_sigmoid.html","id":"reference","dir":"Reference","previous_headings":"","what":"Reference","title":"Hard sigmoid activation function. — activation_hard_sigmoid","text":"Wikipedia \"Hard sigmoid\"","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/activation_hard_silu.html","id":null,"dir":"Reference","previous_headings":"","what":"Hard SiLU activation function, also known as Hard Swish. — activation_hard_silu","title":"Hard SiLU activation function, also known as Hard Swish. — activation_hard_silu","text":"defined : 0 x < -3 x x > 3 x * (x + 3) / 6 -3 <= x <= 3 faster, piecewise linear approximation silu activation.","code":""},{"path":"https://keras3.posit.co/dev/reference/activation_hard_silu.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Hard SiLU activation function, also known as Hard Swish. — activation_hard_silu","text":"","code":"activation_hard_silu(x)  activation_hard_swish(x)"},{"path":"https://keras3.posit.co/dev/reference/activation_hard_silu.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Hard SiLU activation function, also known as Hard Swish. — activation_hard_silu","text":"x Input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/activation_hard_silu.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Hard SiLU activation function, also known as Hard Swish. — activation_hard_silu","text":"tensor, result applying activation input tensor x.","code":""},{"path":"https://keras3.posit.co/dev/reference/activation_hard_silu.html","id":"reference","dir":"Reference","previous_headings":"","what":"Reference","title":"Hard SiLU activation function, also known as Hard Swish. — activation_hard_silu","text":"Howard, 2019","code":""},{"path":"https://keras3.posit.co/dev/reference/activation_hard_tanh.html","id":null,"dir":"Reference","previous_headings":"","what":"HardTanh activation function. — activation_hard_tanh","title":"HardTanh activation function. — activation_hard_tanh","text":"defined : hard_tanh(x) = -1 x < -1, hard_tanh(x) = x -1 <= x <= 1, hard_tanh(x) = 1 x > 1.","code":""},{"path":"https://keras3.posit.co/dev/reference/activation_hard_tanh.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"HardTanh activation function. — activation_hard_tanh","text":"","code":"activation_hard_tanh(x)"},{"path":"https://keras3.posit.co/dev/reference/activation_hard_tanh.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"HardTanh activation function. — activation_hard_tanh","text":"x Input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/activation_hard_tanh.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"HardTanh activation function. — activation_hard_tanh","text":"tensor, result applying activation input tensor x.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/activation_leaky_relu.html","id":null,"dir":"Reference","previous_headings":"","what":"Leaky relu activation function. — activation_leaky_relu","title":"Leaky relu activation function. — activation_leaky_relu","text":"Leaky relu activation function.","code":""},{"path":"https://keras3.posit.co/dev/reference/activation_leaky_relu.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Leaky relu activation function. — activation_leaky_relu","text":"","code":"activation_leaky_relu(x, negative_slope = 0.2)"},{"path":"https://keras3.posit.co/dev/reference/activation_leaky_relu.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Leaky relu activation function. — activation_leaky_relu","text":"x Input tensor. negative_slope float controls slope values lower threshold.","code":""},{"path":"https://keras3.posit.co/dev/reference/activation_leaky_relu.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Leaky relu activation function. — activation_leaky_relu","text":"tensor, result applying activation input tensor x.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/activation_linear.html","id":null,"dir":"Reference","previous_headings":"","what":"Linear activation function (pass-through). — activation_linear","title":"Linear activation function (pass-through). — activation_linear","text":"\"linear\" activation identity function: returns input, unmodified.","code":""},{"path":"https://keras3.posit.co/dev/reference/activation_linear.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Linear activation function (pass-through). — activation_linear","text":"","code":"activation_linear(x)"},{"path":"https://keras3.posit.co/dev/reference/activation_linear.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Linear activation function (pass-through). — activation_linear","text":"x Input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/activation_linear.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Linear activation function (pass-through). — activation_linear","text":"tensor, result applying activation input tensor x.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/activation_log_sigmoid.html","id":null,"dir":"Reference","previous_headings":"","what":"Logarithm of the sigmoid activation function. — activation_log_sigmoid","title":"Logarithm of the sigmoid activation function. — activation_log_sigmoid","text":"defined f(x) = log(1 / (1 + exp(-x))).","code":""},{"path":"https://keras3.posit.co/dev/reference/activation_log_sigmoid.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Logarithm of the sigmoid activation function. — activation_log_sigmoid","text":"","code":"activation_log_sigmoid(x)"},{"path":"https://keras3.posit.co/dev/reference/activation_log_sigmoid.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Logarithm of the sigmoid activation function. — activation_log_sigmoid","text":"x Input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/activation_log_sigmoid.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Logarithm of the sigmoid activation function. — activation_log_sigmoid","text":"tensor, result applying activation input tensor x.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/activation_log_softmax.html","id":null,"dir":"Reference","previous_headings":"","what":"Log-Softmax activation function. — activation_log_softmax","title":"Log-Softmax activation function. — activation_log_softmax","text":"input vector handled independently. axis argument sets axis input function applied along.","code":""},{"path":"https://keras3.posit.co/dev/reference/activation_log_softmax.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Log-Softmax activation function. — activation_log_softmax","text":"","code":"activation_log_softmax(x, axis = -1L)"},{"path":"https://keras3.posit.co/dev/reference/activation_log_softmax.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Log-Softmax activation function. — activation_log_softmax","text":"x Input tensor. axis Integer, axis along softmax applied.","code":""},{"path":"https://keras3.posit.co/dev/reference/activation_log_softmax.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Log-Softmax activation function. — activation_log_softmax","text":"tensor, result applying activation input tensor x.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/activation_mish.html","id":null,"dir":"Reference","previous_headings":"","what":"Mish activation function. — activation_mish","title":"Mish activation function. — activation_mish","text":"defined : mish(x) = x * tanh(softplus(x)) softplus defined : softplus(x) = log(exp(x) + 1)","code":""},{"path":"https://keras3.posit.co/dev/reference/activation_mish.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Mish activation function. — activation_mish","text":"","code":"activation_mish(x)"},{"path":"https://keras3.posit.co/dev/reference/activation_mish.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Mish activation function. — activation_mish","text":"x Input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/activation_mish.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Mish activation function. — activation_mish","text":"tensor, result applying activation input tensor x.","code":""},{"path":"https://keras3.posit.co/dev/reference/activation_mish.html","id":"reference","dir":"Reference","previous_headings":"","what":"Reference","title":"Mish activation function. — activation_mish","text":"Misra, 2019","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/activation_relu.html","id":null,"dir":"Reference","previous_headings":"","what":"Applies the rectified linear unit activation function. — activation_relu","title":"Applies the rectified linear unit activation function. — activation_relu","text":"default values, returns standard ReLU activation: max(x, 0), element-wise maximum 0 input tensor. Modifying default parameters allows use non-zero thresholds, change max value activation, use non-zero multiple input values threshold.","code":""},{"path":"https://keras3.posit.co/dev/reference/activation_relu.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Applies the rectified linear unit activation function. — activation_relu","text":"","code":"activation_relu(x, negative_slope = 0, max_value = NULL, threshold = 0)"},{"path":"https://keras3.posit.co/dev/reference/activation_relu.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Applies the rectified linear unit activation function. — activation_relu","text":"x Input tensor. negative_slope numeric controls slope values lower threshold. max_value numeric sets saturation threshold (largest value function return). threshold numeric giving threshold value activation function values damped set zero.","code":""},{"path":"https://keras3.posit.co/dev/reference/activation_relu.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Applies the rectified linear unit activation function. — activation_relu","text":"tensor shape dtype input x.","code":""},{"path":"https://keras3.posit.co/dev/reference/activation_relu.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Applies the rectified linear unit activation function. — activation_relu","text":"","code":"x <- c(-10, -5, 0, 5, 10) activation_relu(x) ## tf.Tensor([ 0.  0.  0.  5. 10.], shape=(5), dtype=float32) activation_relu(x, negative_slope = 0.5) ## tf.Tensor([-5.  -2.5  0.   5.  10. ], shape=(5), dtype=float32) activation_relu(x, max_value = 5) ## tf.Tensor([0. 0. 0. 5. 5.], shape=(5), dtype=float32) activation_relu(x, threshold = 5) ## tf.Tensor([-0. -0.  0.  0. 10.], shape=(5), dtype=float32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/activation_relu6.html","id":null,"dir":"Reference","previous_headings":"","what":"Relu6 activation function. — activation_relu6","title":"Relu6 activation function. — activation_relu6","text":"ReLU function, truncated maximum value 6.","code":""},{"path":"https://keras3.posit.co/dev/reference/activation_relu6.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Relu6 activation function. — activation_relu6","text":"","code":"activation_relu6(x)"},{"path":"https://keras3.posit.co/dev/reference/activation_relu6.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Relu6 activation function. — activation_relu6","text":"x Input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/activation_relu6.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Relu6 activation function. — activation_relu6","text":"tensor, result applying activation input tensor x.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/activation_selu.html","id":null,"dir":"Reference","previous_headings":"","what":"Scaled Exponential Linear Unit (SELU). — activation_selu","title":"Scaled Exponential Linear Unit (SELU). — activation_selu","text":"Scaled Exponential Linear Unit (SELU) activation function defined : scale * x x > 0 scale * alpha * (exp(x) - 1) x < 0 alpha scale pre-defined constants (alpha = 1.67326324 scale = 1.05070098). Basically, SELU activation function multiplies scale (> 1) output activation_elu function ensure slope larger one positive inputs. values alpha scale chosen mean variance inputs preserved two consecutive layers long weights initialized correctly (see initializer_lecun_normal()) number input units \"large enough\" (see reference paper information).","code":""},{"path":"https://keras3.posit.co/dev/reference/activation_selu.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Scaled Exponential Linear Unit (SELU). — activation_selu","text":"","code":"activation_selu(x)"},{"path":"https://keras3.posit.co/dev/reference/activation_selu.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Scaled Exponential Linear Unit (SELU). — activation_selu","text":"x Input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/activation_selu.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Scaled Exponential Linear Unit (SELU). — activation_selu","text":"tensor, result applying activation input tensor x.","code":""},{"path":"https://keras3.posit.co/dev/reference/activation_selu.html","id":"notes","dir":"Reference","previous_headings":"","what":"Notes","title":"Scaled Exponential Linear Unit (SELU). — activation_selu","text":"used together initializer_lecun_normal(). used together dropout variant layer_alpha_dropout() (legacy, depracated).","code":""},{"path":"https://keras3.posit.co/dev/reference/activation_selu.html","id":"reference","dir":"Reference","previous_headings":"","what":"Reference","title":"Scaled Exponential Linear Unit (SELU). — activation_selu","text":"Klambauer et al., 2017","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/activation_sigmoid.html","id":null,"dir":"Reference","previous_headings":"","what":"Sigmoid activation function. — activation_sigmoid","title":"Sigmoid activation function. — activation_sigmoid","text":"defined : sigmoid(x) = 1 / (1 + exp(-x)). small values (<-5), sigmoid returns value close zero, large values (>5) result function gets close 1. Sigmoid equivalent 2-element softmax, second element assumed zero. sigmoid function always returns value 0 1.","code":""},{"path":"https://keras3.posit.co/dev/reference/activation_sigmoid.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Sigmoid activation function. — activation_sigmoid","text":"","code":"activation_sigmoid(x)"},{"path":"https://keras3.posit.co/dev/reference/activation_sigmoid.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Sigmoid activation function. — activation_sigmoid","text":"x Input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/activation_sigmoid.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Sigmoid activation function. — activation_sigmoid","text":"tensor, result applying activation input tensor x.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/activation_silu.html","id":null,"dir":"Reference","previous_headings":"","what":"Swish (or Silu) activation function. — activation_silu","title":"Swish (or Silu) activation function. — activation_silu","text":"defined : swish(x) = x * sigmoid(x). Swish (Silu) activation function smooth, non-monotonic function unbounded bounded .","code":""},{"path":"https://keras3.posit.co/dev/reference/activation_silu.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Swish (or Silu) activation function. — activation_silu","text":"","code":"activation_silu(x)"},{"path":"https://keras3.posit.co/dev/reference/activation_silu.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Swish (or Silu) activation function. — activation_silu","text":"x Input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/activation_silu.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Swish (or Silu) activation function. — activation_silu","text":"tensor, result applying activation input tensor x.","code":""},{"path":"https://keras3.posit.co/dev/reference/activation_silu.html","id":"reference","dir":"Reference","previous_headings":"","what":"Reference","title":"Swish (or Silu) activation function. — activation_silu","text":"Ramachandran et al., 2017","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/activation_soft_shrink.html","id":null,"dir":"Reference","previous_headings":"","what":"Soft Shrink activation function. — activation_soft_shrink","title":"Soft Shrink activation function. — activation_soft_shrink","text":"defined : soft_shrink(x) = x - threshold x > threshold, soft_shrink(x) = x + threshold x < -threshold, soft_shrink(x) = 0 otherwise.","code":""},{"path":"https://keras3.posit.co/dev/reference/activation_soft_shrink.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Soft Shrink activation function. — activation_soft_shrink","text":"","code":"activation_soft_shrink(x, threshold = 0.5)"},{"path":"https://keras3.posit.co/dev/reference/activation_soft_shrink.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Soft Shrink activation function. — activation_soft_shrink","text":"x Input tensor. threshold Threshold value. Defaults 0.5.","code":""},{"path":"https://keras3.posit.co/dev/reference/activation_soft_shrink.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Soft Shrink activation function. — activation_soft_shrink","text":"tensor, result applying activation input tensor x.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/activation_softmax.html","id":null,"dir":"Reference","previous_headings":"","what":"Softmax converts a vector of values to a probability distribution. — activation_softmax","title":"Softmax converts a vector of values to a probability distribution. — activation_softmax","text":"elements output vector range [0, 1] sum 1. input vector handled independently. axis argument sets axis input function applied along. Softmax often used activation last layer classification network result interpreted probability distribution. softmax vector x computed exp(x) / sum(exp(x)). input values log-odds resulting probability.","code":""},{"path":"https://keras3.posit.co/dev/reference/activation_softmax.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Softmax converts a vector of values to a probability distribution. — activation_softmax","text":"","code":"activation_softmax(x, axis = -1L)"},{"path":"https://keras3.posit.co/dev/reference/activation_softmax.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Softmax converts a vector of values to a probability distribution. — activation_softmax","text":"x Input tensor. axis Integer, axis along softmax applied.","code":""},{"path":"https://keras3.posit.co/dev/reference/activation_softmax.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Softmax converts a vector of values to a probability distribution. — activation_softmax","text":"tensor, result applying activation input tensor x.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/activation_softplus.html","id":null,"dir":"Reference","previous_headings":"","what":"Softplus activation function. — activation_softplus","title":"Softplus activation function. — activation_softplus","text":"defined : softplus(x) = log(exp(x) + 1).","code":""},{"path":"https://keras3.posit.co/dev/reference/activation_softplus.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Softplus activation function. — activation_softplus","text":"","code":"activation_softplus(x)"},{"path":"https://keras3.posit.co/dev/reference/activation_softplus.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Softplus activation function. — activation_softplus","text":"x Input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/activation_softplus.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Softplus activation function. — activation_softplus","text":"tensor, result applying activation input tensor x.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/activation_softsign.html","id":null,"dir":"Reference","previous_headings":"","what":"Softsign activation function. — activation_softsign","title":"Softsign activation function. — activation_softsign","text":"Softsign defined : softsign(x) = x / (abs(x) + 1).","code":""},{"path":"https://keras3.posit.co/dev/reference/activation_softsign.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Softsign activation function. — activation_softsign","text":"","code":"activation_softsign(x)"},{"path":"https://keras3.posit.co/dev/reference/activation_softsign.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Softsign activation function. — activation_softsign","text":"x Input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/activation_softsign.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Softsign activation function. — activation_softsign","text":"tensor, result applying activation input tensor x.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/activation_sparse_plus.html","id":null,"dir":"Reference","previous_headings":"","what":"SparsePlus activation function. — activation_sparse_plus","title":"SparsePlus activation function. — activation_sparse_plus","text":"SparsePlus defined : sparse_plus(x) = 0 x <= -1. sparse_plus(x) = (1/4) * (x + 1)^2 -1 < x < 1. sparse_plus(x) = x x >= 1.","code":""},{"path":"https://keras3.posit.co/dev/reference/activation_sparse_plus.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"SparsePlus activation function. — activation_sparse_plus","text":"","code":"activation_sparse_plus(x)"},{"path":"https://keras3.posit.co/dev/reference/activation_sparse_plus.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"SparsePlus activation function. — activation_sparse_plus","text":"x Input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/activation_sparse_plus.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"SparsePlus activation function. — activation_sparse_plus","text":"tensor, result applying activation input tensor x.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/activation_sparse_sigmoid.html","id":null,"dir":"Reference","previous_headings":"","what":"Sparse sigmoid activation function. — activation_sparse_sigmoid","title":"Sparse sigmoid activation function. — activation_sparse_sigmoid","text":"defined f(x) = 0 x <= -1, f(x) = 0.5 * (x + 1) -1 < x < 1, f(x) = 1 x >= 1.","code":""},{"path":"https://keras3.posit.co/dev/reference/activation_sparse_sigmoid.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Sparse sigmoid activation function. — activation_sparse_sigmoid","text":"","code":"activation_sparse_sigmoid(x)"},{"path":"https://keras3.posit.co/dev/reference/activation_sparse_sigmoid.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Sparse sigmoid activation function. — activation_sparse_sigmoid","text":"x Input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/activation_sparse_sigmoid.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Sparse sigmoid activation function. — activation_sparse_sigmoid","text":"tensor, result applying activation input tensor x.","code":""},{"path":"https://keras3.posit.co/dev/reference/activation_sparse_sigmoid.html","id":"reference","dir":"Reference","previous_headings":"","what":"Reference","title":"Sparse sigmoid activation function. — activation_sparse_sigmoid","text":"M. Blondel, . F. T. Martins, V. Niculae, 2019","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/activation_sparsemax.html","id":null,"dir":"Reference","previous_headings":"","what":"Sparsemax activation function. — activation_sparsemax","title":"Sparsemax activation function. — activation_sparsemax","text":"batch , class j, sparsemax activation function defined : sparsemax(x)[, j] = max(x[, j] - (x[, :]), 0).","code":""},{"path":"https://keras3.posit.co/dev/reference/activation_sparsemax.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Sparsemax activation function. — activation_sparsemax","text":"","code":"activation_sparsemax(x, axis = -1L)"},{"path":"https://keras3.posit.co/dev/reference/activation_sparsemax.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Sparsemax activation function. — activation_sparsemax","text":"x Input tensor. axis int, axis along sparsemax operation applied. (1-based)","code":""},{"path":"https://keras3.posit.co/dev/reference/activation_sparsemax.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Sparsemax activation function. — activation_sparsemax","text":"tensor, output sparsemax transformation. type shape x.","code":""},{"path":"https://keras3.posit.co/dev/reference/activation_sparsemax.html","id":"reference","dir":"Reference","previous_headings":"","what":"Reference","title":"Sparsemax activation function. — activation_sparsemax","text":"Martins et.al., 2016","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/activation_squareplus.html","id":null,"dir":"Reference","previous_headings":"","what":"Squareplus activation function. — activation_squareplus","title":"Squareplus activation function. — activation_squareplus","text":"Squareplus activation function defined : f(x) = (x + sqrt(x^2 + b)) / 2 b smoothness parameter.","code":""},{"path":"https://keras3.posit.co/dev/reference/activation_squareplus.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Squareplus activation function. — activation_squareplus","text":"","code":"activation_squareplus(x, b = 4L)"},{"path":"https://keras3.posit.co/dev/reference/activation_squareplus.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Squareplus activation function. — activation_squareplus","text":"x Input tensor. b Smoothness parameter. Defaults 4.","code":""},{"path":"https://keras3.posit.co/dev/reference/activation_squareplus.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Squareplus activation function. — activation_squareplus","text":"tensor, result applying activation input tensor x.","code":""},{"path":"https://keras3.posit.co/dev/reference/activation_squareplus.html","id":"reference","dir":"Reference","previous_headings":"","what":"Reference","title":"Squareplus activation function. — activation_squareplus","text":"Ramachandran et al., 2021","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/activation_tanh.html","id":null,"dir":"Reference","previous_headings":"","what":"Hyperbolic tangent activation function. — activation_tanh","title":"Hyperbolic tangent activation function. — activation_tanh","text":"defined : tanh(x) = sinh(x) / cosh(x), .e. tanh(x) = ((exp(x) - exp(-x)) / (exp(x) + exp(-x))).","code":""},{"path":"https://keras3.posit.co/dev/reference/activation_tanh.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Hyperbolic tangent activation function. — activation_tanh","text":"","code":"activation_tanh(x)"},{"path":"https://keras3.posit.co/dev/reference/activation_tanh.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Hyperbolic tangent activation function. — activation_tanh","text":"x Input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/activation_tanh.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Hyperbolic tangent activation function. — activation_tanh","text":"tensor, result applying activation input tensor x.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/activation_tanh_shrink.html","id":null,"dir":"Reference","previous_headings":"","what":"Tanh shrink activation function. — activation_tanh_shrink","title":"Tanh shrink activation function. — activation_tanh_shrink","text":"defined : f(x) = x - tanh(x).","code":""},{"path":"https://keras3.posit.co/dev/reference/activation_tanh_shrink.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Tanh shrink activation function. — activation_tanh_shrink","text":"","code":"activation_tanh_shrink(x)"},{"path":"https://keras3.posit.co/dev/reference/activation_tanh_shrink.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Tanh shrink activation function. — activation_tanh_shrink","text":"x Input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/activation_tanh_shrink.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Tanh shrink activation function. — activation_tanh_shrink","text":"tensor, result applying activation input tensor x.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/activation_threshold.html","id":null,"dir":"Reference","previous_headings":"","what":"Threshold activation function. — activation_threshold","title":"Threshold activation function. — activation_threshold","text":"defined : threshold(x) = x x > threshold, threshold(x) = default_value otherwise.","code":""},{"path":"https://keras3.posit.co/dev/reference/activation_threshold.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Threshold activation function. — activation_threshold","text":"","code":"activation_threshold(x, threshold, default_value)"},{"path":"https://keras3.posit.co/dev/reference/activation_threshold.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Threshold activation function. — activation_threshold","text":"x Input tensor. threshold value decides retain replace x. default_value Value assign x <= threshold.","code":""},{"path":"https://keras3.posit.co/dev/reference/activation_threshold.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Threshold activation function. — activation_threshold","text":"tensor, result applying activation input tensor x.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/active_property.html","id":null,"dir":"Reference","previous_headings":"","what":"Create an active property class method — active_property","title":"Create an active property class method — active_property","text":"Create active property class method","code":""},{"path":"https://keras3.posit.co/dev/reference/active_property.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Create an active property class method — active_property","text":"","code":"active_property(fn)"},{"path":"https://keras3.posit.co/dev/reference/active_property.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Create an active property class method — active_property","text":"fn R function","code":""},{"path":"https://keras3.posit.co/dev/reference/active_property.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Create an active property class method — active_property","text":"fn, additional R attribute cause fn converted active property converted method custom subclass.","code":""},{"path":"https://keras3.posit.co/dev/reference/active_property.html","id":"example","dir":"Reference","previous_headings":"","what":"Example","title":"Create an active property class method — active_property","text":"","code":"layer_foo <- Model(\"Foo\", ...,   metrics = active_property(function() {     list(self$d_loss_metric,          self$g_loss_metric)   }))"},{"path":"https://keras3.posit.co/dev/reference/adapt.html","id":null,"dir":"Reference","previous_headings":"","what":"Fits the state of the preprocessing layer to the data being passed — adapt","title":"Fits the state of the preprocessing layer to the data being passed — adapt","text":"Fits state preprocessing layer data passed","code":""},{"path":"https://keras3.posit.co/dev/reference/adapt.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Fits the state of the preprocessing layer to the data being passed — adapt","text":"","code":"adapt(object, data, ..., batch_size = NULL, steps = NULL)"},{"path":"https://keras3.posit.co/dev/reference/adapt.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Fits the state of the preprocessing layer to the data being passed — adapt","text":"object Preprocessing layer object data data train . can passed either tf.data.Dataset R array. ... Used forwards backwards compatibility. Passed underlying method. batch_size Integer NULL. Number asamples per state update. unspecified, batch_size default 32. specify batch_size data form TF Dataset generator (since generate batches). steps Integer NULL. Total number steps (batches samples) training input tensors TensorFlow data tensors, default NULL equal number samples dataset divided batch size, 1 determined. x tf.data.Dataset, steps NULL, epoch run input dataset exhausted. passing infinitely repeating dataset, must specify steps argument. argument supported array inputs.","code":""},{"path":"https://keras3.posit.co/dev/reference/adapt.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Fits the state of the preprocessing layer to the data being passed — adapt","text":"Returns object, invisibly.","code":""},{"path":"https://keras3.posit.co/dev/reference/adapt.html","id":"details","dir":"Reference","previous_headings":"","what":"Details","title":"Fits the state of the preprocessing layer to the data being passed — adapt","text":"calling adapt layer, preprocessing layer's state update training. order make preprocessing layers efficient distribution context, kept constant respect compiled tf.Graphs call layer. affect layer use adapting layer , adapt layer multiple times need take care re-compile compiled functions follows: adding preprocessing layer keras model, need call compile(model) subsequent call adapt(). calling preprocessing layer inside tfdatasets::dataset_map(), call dataset_map() input Dataset adapt(). using tensorflow::tf_function() directly calls preprocessing layer, need call tf_function() callable subsequent call adapt(). keras_model() example multiple adapts:         tfdatasets example multiple adapts:","code":"layer <- layer_normalization(axis = NULL) adapt(layer, c(0, 2)) model <- keras_model_sequential() |> layer() predict(model, c(0, 1, 2), verbose = FALSE) # [1] -1  0  1 ## [1] -1  0  1 adapt(layer, c(-1, 1)) compile(model)  # This is needed to re-compile model.predict! predict(model, c(0, 1, 2), verbose = FALSE) # [1] 0 1 2 ## [1] 0 1 2 layer <- layer_normalization(axis = NULL) adapt(layer, c(0, 2)) input_ds <- tfdatasets::range_dataset(0, 3) normalized_ds <- input_ds |>   tfdatasets::dataset_map(layer) str(tfdatasets::iterate(normalized_ds)) ## List of 3 ##  $ :<tf.Tensor: shape=(1), dtype=float32, numpy=array([-1.], dtype=float32)> ##  $ :<tf.Tensor: shape=(1), dtype=float32, numpy=array([0.], dtype=float32)> ##  $ :<tf.Tensor: shape=(1), dtype=float32, numpy=array([1.], dtype=float32)> adapt(layer, c(-1, 1)) normalized_ds <- input_ds |>   tfdatasets::dataset_map(layer) # Re-map over the input dataset.  normalized_ds |>   tfdatasets::as_array_iterator() |>   tfdatasets::iterate(simplify = FALSE) |>   str() ## List of 3 ##  $ : num [1(1d)] 0 ##  $ : num [1(1d)] 1 ##  $ : num [1(1d)] 2"},{"path":"https://keras3.posit.co/dev/reference/application_convnext_base.html","id":null,"dir":"Reference","previous_headings":"","what":"Instantiates the ConvNeXtBase architecture. — application_convnext_base","title":"Instantiates the ConvNeXtBase architecture. — application_convnext_base","text":"Instantiates ConvNeXtBase architecture.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_convnext_base.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Instantiates the ConvNeXtBase architecture. — application_convnext_base","text":"","code":"application_convnext_base(   include_top = TRUE,   include_preprocessing = TRUE,   weights = \"imagenet\",   input_tensor = NULL,   input_shape = NULL,   pooling = NULL,   classes = 1000L,   classifier_activation = \"softmax\",   name = \"convnext_base\" )"},{"path":"https://keras3.posit.co/dev/reference/application_convnext_base.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Instantiates the ConvNeXtBase architecture. — application_convnext_base","text":"include_top Whether include fully-connected layer top network. Defaults TRUE. include_preprocessing Boolean, whether include preprocessing layer bottom network. weights One NULL (random initialization), \"imagenet\" (pre-training ImageNet-1k), path weights file loaded. Defaults \"imagenet\". input_tensor Optional Keras tensor (.e. output keras_input()) use image input model. input_shape Optional shape tuple, specified include_top FALSE. exactly 3 inputs channels. pooling Optional pooling mode feature extraction include_top FALSE. Defaults NULL. NULL means output model 4D tensor output last convolutional layer. avg means global average pooling applied output last convolutional layer, thus output model 2D tensor. max means global max pooling applied. classes Optional number classes classify images , specified include_top TRUE, weights argument specified. Defaults 1000 (number ImageNet classes). classifier_activation str callable. activation function use \"top\" layer. Ignored unless include_top=TRUE. Set classifier_activation=NULL return logits \"top\" layer. Defaults \"softmax\". loading pretrained weights, classifier_activation can NULL \"softmax\". name name model (string).","code":""},{"path":"https://keras3.posit.co/dev/reference/application_convnext_base.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Instantiates the ConvNeXtBase architecture. — application_convnext_base","text":"model instance.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_convnext_base.html","id":"references","dir":"Reference","previous_headings":"","what":"References","title":"Instantiates the ConvNeXtBase architecture. — application_convnext_base","text":"ConvNet 2020s (CVPR 2022) image classification use cases, see page detailed examples. transfer learning use cases, make sure read guide transfer learning & fine-tuning. base, large, xlarge models first pre-trained ImageNet-21k dataset fine-tuned ImageNet-1k dataset. pre-trained parameters models assembled official repository. get sense parameters converted Keras compatible parameters, please refer repository.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_convnext_base.html","id":"note","dir":"Reference","previous_headings":"","what":"Note","title":"Instantiates the ConvNeXtBase architecture. — application_convnext_base","text":"Keras Application expects specific kind input preprocessing. ConvNeXt, preprocessing included model using Normalization layer.  ConvNeXt models expect inputs float uint8 tensors pixels values [0-255] range. calling summary() method instantiating ConvNeXt model, prefer setting expand_nested argument summary() TRUE better investigate instantiated model.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/application_convnext_large.html","id":null,"dir":"Reference","previous_headings":"","what":"Instantiates the ConvNeXtLarge architecture. — application_convnext_large","title":"Instantiates the ConvNeXtLarge architecture. — application_convnext_large","text":"Instantiates ConvNeXtLarge architecture.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_convnext_large.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Instantiates the ConvNeXtLarge architecture. — application_convnext_large","text":"","code":"application_convnext_large(   include_top = TRUE,   include_preprocessing = TRUE,   weights = \"imagenet\",   input_tensor = NULL,   input_shape = NULL,   pooling = NULL,   classes = 1000L,   classifier_activation = \"softmax\",   name = \"convnext_large\" )"},{"path":"https://keras3.posit.co/dev/reference/application_convnext_large.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Instantiates the ConvNeXtLarge architecture. — application_convnext_large","text":"include_top Whether include fully-connected layer top network. Defaults TRUE. include_preprocessing Boolean, whether include preprocessing layer bottom network. weights One NULL (random initialization), \"imagenet\" (pre-training ImageNet-1k), path weights file loaded. Defaults \"imagenet\". input_tensor Optional Keras tensor (.e. output keras_input()) use image input model. input_shape Optional shape tuple, specified include_top FALSE. exactly 3 inputs channels. pooling Optional pooling mode feature extraction include_top FALSE. Defaults NULL. NULL means output model 4D tensor output last convolutional layer. avg means global average pooling applied output last convolutional layer, thus output model 2D tensor. max means global max pooling applied. classes Optional number classes classify images , specified include_top TRUE, weights argument specified. Defaults 1000 (number ImageNet classes). classifier_activation str callable. activation function use \"top\" layer. Ignored unless include_top=TRUE. Set classifier_activation=NULL return logits \"top\" layer. Defaults \"softmax\". loading pretrained weights, classifier_activation can NULL \"softmax\". name name model (string).","code":""},{"path":"https://keras3.posit.co/dev/reference/application_convnext_large.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Instantiates the ConvNeXtLarge architecture. — application_convnext_large","text":"model instance.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_convnext_large.html","id":"references","dir":"Reference","previous_headings":"","what":"References","title":"Instantiates the ConvNeXtLarge architecture. — application_convnext_large","text":"ConvNet 2020s (CVPR 2022) image classification use cases, see page detailed examples. transfer learning use cases, make sure read guide transfer learning & fine-tuning. base, large, xlarge models first pre-trained ImageNet-21k dataset fine-tuned ImageNet-1k dataset. pre-trained parameters models assembled official repository. get sense parameters converted Keras compatible parameters, please refer repository.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_convnext_large.html","id":"note","dir":"Reference","previous_headings":"","what":"Note","title":"Instantiates the ConvNeXtLarge architecture. — application_convnext_large","text":"Keras Application expects specific kind input preprocessing. ConvNeXt, preprocessing included model using Normalization layer.  ConvNeXt models expect inputs float uint8 tensors pixels values [0-255] range. calling summary() method instantiating ConvNeXt model, prefer setting expand_nested argument summary() TRUE better investigate instantiated model.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/application_convnext_small.html","id":null,"dir":"Reference","previous_headings":"","what":"Instantiates the ConvNeXtSmall architecture. — application_convnext_small","title":"Instantiates the ConvNeXtSmall architecture. — application_convnext_small","text":"Instantiates ConvNeXtSmall architecture.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_convnext_small.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Instantiates the ConvNeXtSmall architecture. — application_convnext_small","text":"","code":"application_convnext_small(   include_top = TRUE,   include_preprocessing = TRUE,   weights = \"imagenet\",   input_tensor = NULL,   input_shape = NULL,   pooling = NULL,   classes = 1000L,   classifier_activation = \"softmax\",   name = \"convnext_small\" )"},{"path":"https://keras3.posit.co/dev/reference/application_convnext_small.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Instantiates the ConvNeXtSmall architecture. — application_convnext_small","text":"include_top Whether include fully-connected layer top network. Defaults TRUE. include_preprocessing Boolean, whether include preprocessing layer bottom network. weights One NULL (random initialization), \"imagenet\" (pre-training ImageNet-1k), path weights file loaded. Defaults \"imagenet\". input_tensor Optional Keras tensor (.e. output keras_input()) use image input model. input_shape Optional shape tuple, specified include_top FALSE. exactly 3 inputs channels. pooling Optional pooling mode feature extraction include_top FALSE. Defaults NULL. NULL means output model 4D tensor output last convolutional layer. avg means global average pooling applied output last convolutional layer, thus output model 2D tensor. max means global max pooling applied. classes Optional number classes classify images , specified include_top TRUE, weights argument specified. Defaults 1000 (number ImageNet classes). classifier_activation str callable. activation function use \"top\" layer. Ignored unless include_top=TRUE. Set classifier_activation=NULL return logits \"top\" layer. Defaults \"softmax\". loading pretrained weights, classifier_activation can NULL \"softmax\". name name model (string).","code":""},{"path":"https://keras3.posit.co/dev/reference/application_convnext_small.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Instantiates the ConvNeXtSmall architecture. — application_convnext_small","text":"model instance.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_convnext_small.html","id":"references","dir":"Reference","previous_headings":"","what":"References","title":"Instantiates the ConvNeXtSmall architecture. — application_convnext_small","text":"ConvNet 2020s (CVPR 2022) image classification use cases, see page detailed examples. transfer learning use cases, make sure read guide transfer learning & fine-tuning. base, large, xlarge models first pre-trained ImageNet-21k dataset fine-tuned ImageNet-1k dataset. pre-trained parameters models assembled official repository. get sense parameters converted Keras compatible parameters, please refer repository.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_convnext_small.html","id":"note","dir":"Reference","previous_headings":"","what":"Note","title":"Instantiates the ConvNeXtSmall architecture. — application_convnext_small","text":"Keras Application expects specific kind input preprocessing. ConvNeXt, preprocessing included model using Normalization layer.  ConvNeXt models expect inputs float uint8 tensors pixels values [0-255] range. calling summary() method instantiating ConvNeXt model, prefer setting expand_nested argument summary() TRUE better investigate instantiated model.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/application_convnext_tiny.html","id":null,"dir":"Reference","previous_headings":"","what":"Instantiates the ConvNeXtTiny architecture. — application_convnext_tiny","title":"Instantiates the ConvNeXtTiny architecture. — application_convnext_tiny","text":"Instantiates ConvNeXtTiny architecture.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_convnext_tiny.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Instantiates the ConvNeXtTiny architecture. — application_convnext_tiny","text":"","code":"application_convnext_tiny(   include_top = TRUE,   include_preprocessing = TRUE,   weights = \"imagenet\",   input_tensor = NULL,   input_shape = NULL,   pooling = NULL,   classes = 1000L,   classifier_activation = \"softmax\",   name = \"convnext_tiny\" )"},{"path":"https://keras3.posit.co/dev/reference/application_convnext_tiny.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Instantiates the ConvNeXtTiny architecture. — application_convnext_tiny","text":"include_top Whether include fully-connected layer top network. Defaults TRUE. include_preprocessing Boolean, whether include preprocessing layer bottom network. weights One NULL (random initialization), \"imagenet\" (pre-training ImageNet-1k), path weights file loaded. Defaults \"imagenet\". input_tensor Optional Keras tensor (.e. output keras_input()) use image input model. input_shape Optional shape tuple, specified include_top FALSE. exactly 3 inputs channels. pooling Optional pooling mode feature extraction include_top FALSE. Defaults NULL. NULL means output model 4D tensor output last convolutional layer. avg means global average pooling applied output last convolutional layer, thus output model 2D tensor. max means global max pooling applied. classes Optional number classes classify images , specified include_top TRUE, weights argument specified. Defaults 1000 (number ImageNet classes). classifier_activation str callable. activation function use \"top\" layer. Ignored unless include_top=TRUE. Set classifier_activation=NULL return logits \"top\" layer. Defaults \"softmax\". loading pretrained weights, classifier_activation can NULL \"softmax\". name name model (string).","code":""},{"path":"https://keras3.posit.co/dev/reference/application_convnext_tiny.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Instantiates the ConvNeXtTiny architecture. — application_convnext_tiny","text":"model instance.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_convnext_tiny.html","id":"references","dir":"Reference","previous_headings":"","what":"References","title":"Instantiates the ConvNeXtTiny architecture. — application_convnext_tiny","text":"ConvNet 2020s (CVPR 2022) image classification use cases, see page detailed examples. transfer learning use cases, make sure read guide transfer learning & fine-tuning. base, large, xlarge models first pre-trained ImageNet-21k dataset fine-tuned ImageNet-1k dataset. pre-trained parameters models assembled official repository. get sense parameters converted Keras compatible parameters, please refer repository.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_convnext_tiny.html","id":"note","dir":"Reference","previous_headings":"","what":"Note","title":"Instantiates the ConvNeXtTiny architecture. — application_convnext_tiny","text":"Keras Application expects specific kind input preprocessing. ConvNeXt, preprocessing included model using Normalization layer.  ConvNeXt models expect inputs float uint8 tensors pixels values [0-255] range. calling summary() method instantiating ConvNeXt model, prefer setting expand_nested argument summary() TRUE better investigate instantiated model.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/application_convnext_xlarge.html","id":null,"dir":"Reference","previous_headings":"","what":"Instantiates the ConvNeXtXLarge architecture. — application_convnext_xlarge","title":"Instantiates the ConvNeXtXLarge architecture. — application_convnext_xlarge","text":"Instantiates ConvNeXtXLarge architecture.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_convnext_xlarge.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Instantiates the ConvNeXtXLarge architecture. — application_convnext_xlarge","text":"","code":"application_convnext_xlarge(   include_top = TRUE,   include_preprocessing = TRUE,   weights = \"imagenet\",   input_tensor = NULL,   input_shape = NULL,   pooling = NULL,   classes = 1000L,   classifier_activation = \"softmax\",   name = \"convnext_xlarge\" )"},{"path":"https://keras3.posit.co/dev/reference/application_convnext_xlarge.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Instantiates the ConvNeXtXLarge architecture. — application_convnext_xlarge","text":"include_top Whether include fully-connected layer top network. Defaults TRUE. include_preprocessing Boolean, whether include preprocessing layer bottom network. weights One NULL (random initialization), \"imagenet\" (pre-training ImageNet-1k), path weights file loaded. Defaults \"imagenet\". input_tensor Optional Keras tensor (.e. output keras_input()) use image input model. input_shape Optional shape tuple, specified include_top FALSE. exactly 3 inputs channels. pooling Optional pooling mode feature extraction include_top FALSE. Defaults NULL. NULL means output model 4D tensor output last convolutional layer. avg means global average pooling applied output last convolutional layer, thus output model 2D tensor. max means global max pooling applied. classes Optional number classes classify images , specified include_top TRUE, weights argument specified. Defaults 1000 (number ImageNet classes). classifier_activation str callable. activation function use \"top\" layer. Ignored unless include_top=TRUE. Set classifier_activation=NULL return logits \"top\" layer. Defaults \"softmax\". loading pretrained weights, classifier_activation can NULL \"softmax\". name name model (string).","code":""},{"path":"https://keras3.posit.co/dev/reference/application_convnext_xlarge.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Instantiates the ConvNeXtXLarge architecture. — application_convnext_xlarge","text":"model instance.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_convnext_xlarge.html","id":"references","dir":"Reference","previous_headings":"","what":"References","title":"Instantiates the ConvNeXtXLarge architecture. — application_convnext_xlarge","text":"ConvNet 2020s (CVPR 2022) image classification use cases, see page detailed examples. transfer learning use cases, make sure read guide transfer learning & fine-tuning. base, large, xlarge models first pre-trained ImageNet-21k dataset fine-tuned ImageNet-1k dataset. pre-trained parameters models assembled official repository. get sense parameters converted Keras compatible parameters, please refer repository.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_convnext_xlarge.html","id":"note","dir":"Reference","previous_headings":"","what":"Note","title":"Instantiates the ConvNeXtXLarge architecture. — application_convnext_xlarge","text":"Keras Application expects specific kind input preprocessing. ConvNeXt, preprocessing included model using Normalization layer.  ConvNeXt models expect inputs float uint8 tensors pixels values [0-255] range. calling summary() method instantiating ConvNeXt model, prefer setting expand_nested argument summary() TRUE better investigate instantiated model.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/application_densenet121.html","id":null,"dir":"Reference","previous_headings":"","what":"Instantiates the Densenet121 architecture. — application_densenet121","title":"Instantiates the Densenet121 architecture. — application_densenet121","text":"Instantiates Densenet121 architecture.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_densenet121.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Instantiates the Densenet121 architecture. — application_densenet121","text":"","code":"application_densenet121(   include_top = TRUE,   weights = \"imagenet\",   input_tensor = NULL,   input_shape = NULL,   pooling = NULL,   classes = 1000L,   classifier_activation = \"softmax\",   name = \"densenet121\" )"},{"path":"https://keras3.posit.co/dev/reference/application_densenet121.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Instantiates the Densenet121 architecture. — application_densenet121","text":"include_top whether include fully-connected layer top network. weights one NULL (random initialization), \"imagenet\" (pre-training ImageNet), path weights file loaded. input_tensor optional Keras tensor (.e. output keras_input()) use image input model. input_shape optional shape tuple, specified include_top FALSE (otherwise input shape (224, 224, 3) ('channels_last' data format) (3, 224, 224) ('channels_first' data format). exactly 3 inputs channels, width height smaller 32. E.g. (200, 200, 3) one valid value. pooling Optional pooling mode feature extraction include_top FALSE. NULL means output model 4D tensor output last convolutional block. avg means global average pooling applied output last convolutional block, thus output model 2D tensor. max means global max pooling applied. classes optional number classes classify images , specified include_top TRUE, weights argument specified. Defaults 1000. classifier_activation str callable. activation function use \"top\" layer. Ignored unless include_top=TRUE. Set classifier_activation=NULL return logits \"top\" layer. loading pretrained weights, classifier_activation can NULL \"softmax\". name name model (string).","code":""},{"path":"https://keras3.posit.co/dev/reference/application_densenet121.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Instantiates the Densenet121 architecture. — application_densenet121","text":"Keras model instance.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_densenet121.html","id":"reference","dir":"Reference","previous_headings":"","what":"Reference","title":"Instantiates the Densenet121 architecture. — application_densenet121","text":"Densely Connected Convolutional Networks (CVPR 2017) Optionally loads weights pre-trained ImageNet. Note data format convention used model one specified Keras config ~/.keras/keras.json.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_densenet121.html","id":"note","dir":"Reference","previous_headings":"","what":"Note","title":"Instantiates the Densenet121 architecture. — application_densenet121","text":"Keras Application expects specific kind input preprocessing. DenseNet, call application_preprocess_inputs() inputs passing model.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/application_densenet169.html","id":null,"dir":"Reference","previous_headings":"","what":"Instantiates the Densenet169 architecture. — application_densenet169","title":"Instantiates the Densenet169 architecture. — application_densenet169","text":"Instantiates Densenet169 architecture.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_densenet169.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Instantiates the Densenet169 architecture. — application_densenet169","text":"","code":"application_densenet169(   include_top = TRUE,   weights = \"imagenet\",   input_tensor = NULL,   input_shape = NULL,   pooling = NULL,   classes = 1000L,   classifier_activation = \"softmax\",   name = \"densenet169\" )"},{"path":"https://keras3.posit.co/dev/reference/application_densenet169.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Instantiates the Densenet169 architecture. — application_densenet169","text":"include_top whether include fully-connected layer top network. weights one NULL (random initialization), \"imagenet\" (pre-training ImageNet), path weights file loaded. input_tensor optional Keras tensor (.e. output keras_input()) use image input model. input_shape optional shape tuple, specified include_top FALSE (otherwise input shape (224, 224, 3) ('channels_last' data format) (3, 224, 224) ('channels_first' data format). exactly 3 inputs channels, width height smaller 32. E.g. (200, 200, 3) one valid value. pooling Optional pooling mode feature extraction include_top FALSE. NULL means output model 4D tensor output last convolutional block. avg means global average pooling applied output last convolutional block, thus output model 2D tensor. max means global max pooling applied. classes optional number classes classify images , specified include_top TRUE, weights argument specified. Defaults 1000. classifier_activation str callable. activation function use \"top\" layer. Ignored unless include_top=TRUE. Set classifier_activation=NULL return logits \"top\" layer. loading pretrained weights, classifier_activation can NULL \"softmax\". name name model (string).","code":""},{"path":"https://keras3.posit.co/dev/reference/application_densenet169.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Instantiates the Densenet169 architecture. — application_densenet169","text":"Keras model instance.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_densenet169.html","id":"reference","dir":"Reference","previous_headings":"","what":"Reference","title":"Instantiates the Densenet169 architecture. — application_densenet169","text":"Densely Connected Convolutional Networks (CVPR 2017) Optionally loads weights pre-trained ImageNet. Note data format convention used model one specified Keras config ~/.keras/keras.json.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_densenet169.html","id":"note","dir":"Reference","previous_headings":"","what":"Note","title":"Instantiates the Densenet169 architecture. — application_densenet169","text":"Keras Application expects specific kind input preprocessing. DenseNet, call application_preprocess_inputs() inputs passing model.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/application_densenet201.html","id":null,"dir":"Reference","previous_headings":"","what":"Instantiates the Densenet201 architecture. — application_densenet201","title":"Instantiates the Densenet201 architecture. — application_densenet201","text":"Instantiates Densenet201 architecture.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_densenet201.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Instantiates the Densenet201 architecture. — application_densenet201","text":"","code":"application_densenet201(   include_top = TRUE,   weights = \"imagenet\",   input_tensor = NULL,   input_shape = NULL,   pooling = NULL,   classes = 1000L,   classifier_activation = \"softmax\",   name = \"densenet201\" )"},{"path":"https://keras3.posit.co/dev/reference/application_densenet201.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Instantiates the Densenet201 architecture. — application_densenet201","text":"include_top whether include fully-connected layer top network. weights one NULL (random initialization), \"imagenet\" (pre-training ImageNet), path weights file loaded. input_tensor optional Keras tensor (.e. output keras_input()) use image input model. input_shape optional shape tuple, specified include_top FALSE (otherwise input shape (224, 224, 3) ('channels_last' data format) (3, 224, 224) ('channels_first' data format). exactly 3 inputs channels, width height smaller 32. E.g. (200, 200, 3) one valid value. pooling Optional pooling mode feature extraction include_top FALSE. NULL means output model 4D tensor output last convolutional block. avg means global average pooling applied output last convolutional block, thus output model 2D tensor. max means global max pooling applied. classes optional number classes classify images , specified include_top TRUE, weights argument specified. Defaults 1000. classifier_activation str callable. activation function use \"top\" layer. Ignored unless include_top=TRUE. Set classifier_activation=NULL return logits \"top\" layer. loading pretrained weights, classifier_activation can NULL \"softmax\". name name model (string).","code":""},{"path":"https://keras3.posit.co/dev/reference/application_densenet201.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Instantiates the Densenet201 architecture. — application_densenet201","text":"Keras model instance.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_densenet201.html","id":"reference","dir":"Reference","previous_headings":"","what":"Reference","title":"Instantiates the Densenet201 architecture. — application_densenet201","text":"Densely Connected Convolutional Networks (CVPR 2017) Optionally loads weights pre-trained ImageNet. Note data format convention used model one specified Keras config ~/.keras/keras.json.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_densenet201.html","id":"note","dir":"Reference","previous_headings":"","what":"Note","title":"Instantiates the Densenet201 architecture. — application_densenet201","text":"Keras Application expects specific kind input preprocessing. DenseNet, call application_preprocess_inputs() inputs passing model.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_b0.html","id":null,"dir":"Reference","previous_headings":"","what":"Instantiates the EfficientNetB0 architecture. — application_efficientnet_b0","title":"Instantiates the EfficientNetB0 architecture. — application_efficientnet_b0","text":"Instantiates EfficientNetB0 architecture.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_b0.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Instantiates the EfficientNetB0 architecture. — application_efficientnet_b0","text":"","code":"application_efficientnet_b0(   include_top = TRUE,   weights = \"imagenet\",   input_tensor = NULL,   input_shape = NULL,   pooling = NULL,   classes = 1000L,   classifier_activation = \"softmax\",   name = \"efficientnetb0\",   ... )"},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_b0.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Instantiates the EfficientNetB0 architecture. — application_efficientnet_b0","text":"include_top Whether include fully-connected layer top network. Defaults TRUE. weights One NULL (random initialization), \"imagenet\" (pre-training ImageNet), path weights file loaded. Defaults \"imagenet\". input_tensor Optional Keras tensor (.e. output keras_input()) use image input model. input_shape Optional shape tuple, specified include_top FALSE. exactly 3 inputs channels. pooling Optional pooling mode feature extraction include_top FALSE. Defaults NULL. NULL means output model 4D tensor output last convolutional layer. avg means global average pooling applied output last convolutional layer, thus output model 2D tensor. max means global max pooling applied. classes Optional number classes classify images , specified include_top TRUE, weights argument specified. 1000 many ImageNet classes . Defaults 1000. classifier_activation str callable. activation function use \"top\" layer. Ignored unless include_top=TRUE. Set classifier_activation=NULL return logits \"top\" layer. Defaults 'softmax'. loading pretrained weights, classifier_activation can NULL \"softmax\". name name model (string). ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_b0.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Instantiates the EfficientNetB0 architecture. — application_efficientnet_b0","text":"model instance.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_b0.html","id":"reference","dir":"Reference","previous_headings":"","what":"Reference","title":"Instantiates the EfficientNetB0 architecture. — application_efficientnet_b0","text":"EfficientNet: Rethinking Model Scaling Convolutional Neural Networks (ICML 2019) function returns Keras image classification model, optionally loaded weights pre-trained ImageNet. image classification use cases, see page detailed examples. transfer learning use cases, make sure read guide transfer learning & fine-tuning.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_b0.html","id":"note","dir":"Reference","previous_headings":"","what":"Note","title":"Instantiates the EfficientNetB0 architecture. — application_efficientnet_b0","text":"Keras Application expects specific kind input preprocessing. EfficientNet, input preprocessing included part model (Rescaling layer), thus application_preprocess_inputs() actually pass-function. EfficientNet models expect inputs float tensors pixels values [0-255] range.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_b1.html","id":null,"dir":"Reference","previous_headings":"","what":"Instantiates the EfficientNetB1 architecture. — application_efficientnet_b1","title":"Instantiates the EfficientNetB1 architecture. — application_efficientnet_b1","text":"Instantiates EfficientNetB1 architecture.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_b1.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Instantiates the EfficientNetB1 architecture. — application_efficientnet_b1","text":"","code":"application_efficientnet_b1(   include_top = TRUE,   weights = \"imagenet\",   input_tensor = NULL,   input_shape = NULL,   pooling = NULL,   classes = 1000L,   classifier_activation = \"softmax\",   name = \"efficientnetb1\",   ... )"},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_b1.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Instantiates the EfficientNetB1 architecture. — application_efficientnet_b1","text":"include_top Whether include fully-connected layer top network. Defaults TRUE. weights One NULL (random initialization), \"imagenet\" (pre-training ImageNet), path weights file loaded. Defaults \"imagenet\". input_tensor Optional Keras tensor (.e. output keras_input()) use image input model. input_shape Optional shape tuple, specified include_top FALSE. exactly 3 inputs channels. pooling Optional pooling mode feature extraction include_top FALSE. Defaults NULL. NULL means output model 4D tensor output last convolutional layer. avg means global average pooling applied output last convolutional layer, thus output model 2D tensor. max means global max pooling applied. classes Optional number classes classify images , specified include_top TRUE, weights argument specified. 1000 many ImageNet classes . Defaults 1000. classifier_activation str callable. activation function use \"top\" layer. Ignored unless include_top=TRUE. Set classifier_activation=NULL return logits \"top\" layer. Defaults 'softmax'. loading pretrained weights, classifier_activation can NULL \"softmax\". name name model (string). ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_b1.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Instantiates the EfficientNetB1 architecture. — application_efficientnet_b1","text":"model instance.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_b1.html","id":"reference","dir":"Reference","previous_headings":"","what":"Reference","title":"Instantiates the EfficientNetB1 architecture. — application_efficientnet_b1","text":"EfficientNet: Rethinking Model Scaling Convolutional Neural Networks (ICML 2019) function returns Keras image classification model, optionally loaded weights pre-trained ImageNet. image classification use cases, see page detailed examples. transfer learning use cases, make sure read guide transfer learning & fine-tuning.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_b1.html","id":"note","dir":"Reference","previous_headings":"","what":"Note","title":"Instantiates the EfficientNetB1 architecture. — application_efficientnet_b1","text":"Keras Application expects specific kind input preprocessing. EfficientNet, input preprocessing included part model (Rescaling layer), thus application_preprocess_inputs() actually pass-function. EfficientNet models expect inputs float tensors pixels values [0-255] range.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_b2.html","id":null,"dir":"Reference","previous_headings":"","what":"Instantiates the EfficientNetB2 architecture. — application_efficientnet_b2","title":"Instantiates the EfficientNetB2 architecture. — application_efficientnet_b2","text":"Instantiates EfficientNetB2 architecture.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_b2.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Instantiates the EfficientNetB2 architecture. — application_efficientnet_b2","text":"","code":"application_efficientnet_b2(   include_top = TRUE,   weights = \"imagenet\",   input_tensor = NULL,   input_shape = NULL,   pooling = NULL,   classes = 1000L,   classifier_activation = \"softmax\",   name = \"efficientnetb2\",   ... )"},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_b2.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Instantiates the EfficientNetB2 architecture. — application_efficientnet_b2","text":"include_top Whether include fully-connected layer top network. Defaults TRUE. weights One NULL (random initialization), \"imagenet\" (pre-training ImageNet), path weights file loaded. Defaults \"imagenet\". input_tensor Optional Keras tensor (.e. output keras_input()) use image input model. input_shape Optional shape tuple, specified include_top FALSE. exactly 3 inputs channels. pooling Optional pooling mode feature extraction include_top FALSE. Defaults NULL. NULL means output model 4D tensor output last convolutional layer. avg means global average pooling applied output last convolutional layer, thus output model 2D tensor. max means global max pooling applied. classes Optional number classes classify images , specified include_top TRUE, weights argument specified. 1000 many ImageNet classes . Defaults 1000. classifier_activation str callable. activation function use \"top\" layer. Ignored unless include_top=TRUE. Set classifier_activation=NULL return logits \"top\" layer. Defaults 'softmax'. loading pretrained weights, classifier_activation can NULL \"softmax\". name name model (string). ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_b2.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Instantiates the EfficientNetB2 architecture. — application_efficientnet_b2","text":"model instance.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_b2.html","id":"reference","dir":"Reference","previous_headings":"","what":"Reference","title":"Instantiates the EfficientNetB2 architecture. — application_efficientnet_b2","text":"EfficientNet: Rethinking Model Scaling Convolutional Neural Networks (ICML 2019) function returns Keras image classification model, optionally loaded weights pre-trained ImageNet. image classification use cases, see page detailed examples. transfer learning use cases, make sure read guide transfer learning & fine-tuning.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_b2.html","id":"note","dir":"Reference","previous_headings":"","what":"Note","title":"Instantiates the EfficientNetB2 architecture. — application_efficientnet_b2","text":"Keras Application expects specific kind input preprocessing. EfficientNet, input preprocessing included part model (Rescaling layer), thus application_preprocess_inputs() actually pass-function. EfficientNet models expect inputs float tensors pixels values [0-255] range.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_b3.html","id":null,"dir":"Reference","previous_headings":"","what":"Instantiates the EfficientNetB3 architecture. — application_efficientnet_b3","title":"Instantiates the EfficientNetB3 architecture. — application_efficientnet_b3","text":"Instantiates EfficientNetB3 architecture.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_b3.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Instantiates the EfficientNetB3 architecture. — application_efficientnet_b3","text":"","code":"application_efficientnet_b3(   include_top = TRUE,   weights = \"imagenet\",   input_tensor = NULL,   input_shape = NULL,   pooling = NULL,   classes = 1000L,   classifier_activation = \"softmax\",   name = \"efficientnetb3\",   ... )"},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_b3.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Instantiates the EfficientNetB3 architecture. — application_efficientnet_b3","text":"include_top Whether include fully-connected layer top network. Defaults TRUE. weights One NULL (random initialization), \"imagenet\" (pre-training ImageNet), path weights file loaded. Defaults \"imagenet\". input_tensor Optional Keras tensor (.e. output keras_input()) use image input model. input_shape Optional shape tuple, specified include_top FALSE. exactly 3 inputs channels. pooling Optional pooling mode feature extraction include_top FALSE. Defaults NULL. NULL means output model 4D tensor output last convolutional layer. avg means global average pooling applied output last convolutional layer, thus output model 2D tensor. max means global max pooling applied. classes Optional number classes classify images , specified include_top TRUE, weights argument specified. 1000 many ImageNet classes . Defaults 1000. classifier_activation str callable. activation function use \"top\" layer. Ignored unless include_top=TRUE. Set classifier_activation=NULL return logits \"top\" layer. Defaults 'softmax'. loading pretrained weights, classifier_activation can NULL \"softmax\". name name model (string). ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_b3.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Instantiates the EfficientNetB3 architecture. — application_efficientnet_b3","text":"model instance.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_b3.html","id":"reference","dir":"Reference","previous_headings":"","what":"Reference","title":"Instantiates the EfficientNetB3 architecture. — application_efficientnet_b3","text":"EfficientNet: Rethinking Model Scaling Convolutional Neural Networks (ICML 2019) function returns Keras image classification model, optionally loaded weights pre-trained ImageNet. image classification use cases, see page detailed examples. transfer learning use cases, make sure read guide transfer learning & fine-tuning.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_b3.html","id":"note","dir":"Reference","previous_headings":"","what":"Note","title":"Instantiates the EfficientNetB3 architecture. — application_efficientnet_b3","text":"Keras Application expects specific kind input preprocessing. EfficientNet, input preprocessing included part model (Rescaling layer), thus application_preprocess_inputs() actually pass-function. EfficientNet models expect inputs float tensors pixels values [0-255] range.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_b4.html","id":null,"dir":"Reference","previous_headings":"","what":"Instantiates the EfficientNetB4 architecture. — application_efficientnet_b4","title":"Instantiates the EfficientNetB4 architecture. — application_efficientnet_b4","text":"Instantiates EfficientNetB4 architecture.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_b4.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Instantiates the EfficientNetB4 architecture. — application_efficientnet_b4","text":"","code":"application_efficientnet_b4(   include_top = TRUE,   weights = \"imagenet\",   input_tensor = NULL,   input_shape = NULL,   pooling = NULL,   classes = 1000L,   classifier_activation = \"softmax\",   name = \"efficientnetb4\",   ... )"},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_b4.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Instantiates the EfficientNetB4 architecture. — application_efficientnet_b4","text":"include_top Whether include fully-connected layer top network. Defaults TRUE. weights One NULL (random initialization), \"imagenet\" (pre-training ImageNet), path weights file loaded. Defaults \"imagenet\". input_tensor Optional Keras tensor (.e. output keras_input()) use image input model. input_shape Optional shape tuple, specified include_top FALSE. exactly 3 inputs channels. pooling Optional pooling mode feature extraction include_top FALSE. Defaults NULL. NULL means output model 4D tensor output last convolutional layer. avg means global average pooling applied output last convolutional layer, thus output model 2D tensor. max means global max pooling applied. classes Optional number classes classify images , specified include_top TRUE, weights argument specified. 1000 many ImageNet classes . Defaults 1000. classifier_activation str callable. activation function use \"top\" layer. Ignored unless include_top=TRUE. Set classifier_activation=NULL return logits \"top\" layer. Defaults 'softmax'. loading pretrained weights, classifier_activation can NULL \"softmax\". name name model (string). ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_b4.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Instantiates the EfficientNetB4 architecture. — application_efficientnet_b4","text":"model instance.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_b4.html","id":"reference","dir":"Reference","previous_headings":"","what":"Reference","title":"Instantiates the EfficientNetB4 architecture. — application_efficientnet_b4","text":"EfficientNet: Rethinking Model Scaling Convolutional Neural Networks (ICML 2019) function returns Keras image classification model, optionally loaded weights pre-trained ImageNet. image classification use cases, see page detailed examples. transfer learning use cases, make sure read guide transfer learning & fine-tuning.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_b4.html","id":"note","dir":"Reference","previous_headings":"","what":"Note","title":"Instantiates the EfficientNetB4 architecture. — application_efficientnet_b4","text":"Keras Application expects specific kind input preprocessing. EfficientNet, input preprocessing included part model (Rescaling layer), thus application_preprocess_inputs() actually pass-function. EfficientNet models expect inputs float tensors pixels values [0-255] range.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_b5.html","id":null,"dir":"Reference","previous_headings":"","what":"Instantiates the EfficientNetB5 architecture. — application_efficientnet_b5","title":"Instantiates the EfficientNetB5 architecture. — application_efficientnet_b5","text":"Instantiates EfficientNetB5 architecture.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_b5.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Instantiates the EfficientNetB5 architecture. — application_efficientnet_b5","text":"","code":"application_efficientnet_b5(   include_top = TRUE,   weights = \"imagenet\",   input_tensor = NULL,   input_shape = NULL,   pooling = NULL,   classes = 1000L,   classifier_activation = \"softmax\",   name = \"efficientnetb5\",   ... )"},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_b5.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Instantiates the EfficientNetB5 architecture. — application_efficientnet_b5","text":"include_top Whether include fully-connected layer top network. Defaults TRUE. weights One NULL (random initialization), \"imagenet\" (pre-training ImageNet), path weights file loaded. Defaults \"imagenet\". input_tensor Optional Keras tensor (.e. output keras_input()) use image input model. input_shape Optional shape tuple, specified include_top FALSE. exactly 3 inputs channels. pooling Optional pooling mode feature extraction include_top FALSE. Defaults NULL. NULL means output model 4D tensor output last convolutional layer. avg means global average pooling applied output last convolutional layer, thus output model 2D tensor. max means global max pooling applied. classes Optional number classes classify images , specified include_top TRUE, weights argument specified. 1000 many ImageNet classes . Defaults 1000. classifier_activation str callable. activation function use \"top\" layer. Ignored unless include_top=TRUE. Set classifier_activation=NULL return logits \"top\" layer. Defaults 'softmax'. loading pretrained weights, classifier_activation can NULL \"softmax\". name name model (string). ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_b5.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Instantiates the EfficientNetB5 architecture. — application_efficientnet_b5","text":"model instance.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_b5.html","id":"reference","dir":"Reference","previous_headings":"","what":"Reference","title":"Instantiates the EfficientNetB5 architecture. — application_efficientnet_b5","text":"EfficientNet: Rethinking Model Scaling Convolutional Neural Networks (ICML 2019) function returns Keras image classification model, optionally loaded weights pre-trained ImageNet. image classification use cases, see page detailed examples. transfer learning use cases, make sure read guide transfer learning & fine-tuning.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_b5.html","id":"note","dir":"Reference","previous_headings":"","what":"Note","title":"Instantiates the EfficientNetB5 architecture. — application_efficientnet_b5","text":"Keras Application expects specific kind input preprocessing. EfficientNet, input preprocessing included part model (Rescaling layer), thus application_preprocess_inputs() actually pass-function. EfficientNet models expect inputs float tensors pixels values [0-255] range.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_b6.html","id":null,"dir":"Reference","previous_headings":"","what":"Instantiates the EfficientNetB6 architecture. — application_efficientnet_b6","title":"Instantiates the EfficientNetB6 architecture. — application_efficientnet_b6","text":"Instantiates EfficientNetB6 architecture.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_b6.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Instantiates the EfficientNetB6 architecture. — application_efficientnet_b6","text":"","code":"application_efficientnet_b6(   include_top = TRUE,   weights = \"imagenet\",   input_tensor = NULL,   input_shape = NULL,   pooling = NULL,   classes = 1000L,   classifier_activation = \"softmax\",   name = \"efficientnetb6\",   ... )"},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_b6.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Instantiates the EfficientNetB6 architecture. — application_efficientnet_b6","text":"include_top Whether include fully-connected layer top network. Defaults TRUE. weights One NULL (random initialization), \"imagenet\" (pre-training ImageNet), path weights file loaded. Defaults \"imagenet\". input_tensor Optional Keras tensor (.e. output keras_input()) use image input model. input_shape Optional shape tuple, specified include_top FALSE. exactly 3 inputs channels. pooling Optional pooling mode feature extraction include_top FALSE. Defaults NULL. NULL means output model 4D tensor output last convolutional layer. avg means global average pooling applied output last convolutional layer, thus output model 2D tensor. max means global max pooling applied. classes Optional number classes classify images , specified include_top TRUE, weights argument specified. 1000 many ImageNet classes . Defaults 1000. classifier_activation str callable. activation function use \"top\" layer. Ignored unless include_top=TRUE. Set classifier_activation=NULL return logits \"top\" layer. Defaults 'softmax'. loading pretrained weights, classifier_activation can NULL \"softmax\". name name model (string). ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_b6.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Instantiates the EfficientNetB6 architecture. — application_efficientnet_b6","text":"model instance.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_b6.html","id":"reference","dir":"Reference","previous_headings":"","what":"Reference","title":"Instantiates the EfficientNetB6 architecture. — application_efficientnet_b6","text":"EfficientNet: Rethinking Model Scaling Convolutional Neural Networks (ICML 2019) function returns Keras image classification model, optionally loaded weights pre-trained ImageNet. image classification use cases, see page detailed examples. transfer learning use cases, make sure read guide transfer learning & fine-tuning.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_b6.html","id":"note","dir":"Reference","previous_headings":"","what":"Note","title":"Instantiates the EfficientNetB6 architecture. — application_efficientnet_b6","text":"Keras Application expects specific kind input preprocessing. EfficientNet, input preprocessing included part model (Rescaling layer), thus application_preprocess_inputs() actually pass-function. EfficientNet models expect inputs float tensors pixels values [0-255] range.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_b7.html","id":null,"dir":"Reference","previous_headings":"","what":"Instantiates the EfficientNetB7 architecture. — application_efficientnet_b7","title":"Instantiates the EfficientNetB7 architecture. — application_efficientnet_b7","text":"Instantiates EfficientNetB7 architecture.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_b7.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Instantiates the EfficientNetB7 architecture. — application_efficientnet_b7","text":"","code":"application_efficientnet_b7(   include_top = TRUE,   weights = \"imagenet\",   input_tensor = NULL,   input_shape = NULL,   pooling = NULL,   classes = 1000L,   classifier_activation = \"softmax\",   name = \"efficientnetb7\",   ... )"},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_b7.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Instantiates the EfficientNetB7 architecture. — application_efficientnet_b7","text":"include_top Whether include fully-connected layer top network. Defaults TRUE. weights One NULL (random initialization), \"imagenet\" (pre-training ImageNet), path weights file loaded. Defaults \"imagenet\". input_tensor Optional Keras tensor (.e. output keras_input()) use image input model. input_shape Optional shape tuple, specified include_top FALSE. exactly 3 inputs channels. pooling Optional pooling mode feature extraction include_top FALSE. Defaults NULL. NULL means output model 4D tensor output last convolutional layer. avg means global average pooling applied output last convolutional layer, thus output model 2D tensor. max means global max pooling applied. classes Optional number classes classify images , specified include_top TRUE, weights argument specified. 1000 many ImageNet classes . Defaults 1000. classifier_activation str callable. activation function use \"top\" layer. Ignored unless include_top=TRUE. Set classifier_activation=NULL return logits \"top\" layer. Defaults 'softmax'. loading pretrained weights, classifier_activation can NULL \"softmax\". name name model (string). ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_b7.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Instantiates the EfficientNetB7 architecture. — application_efficientnet_b7","text":"model instance.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_b7.html","id":"reference","dir":"Reference","previous_headings":"","what":"Reference","title":"Instantiates the EfficientNetB7 architecture. — application_efficientnet_b7","text":"EfficientNet: Rethinking Model Scaling Convolutional Neural Networks (ICML 2019) function returns Keras image classification model, optionally loaded weights pre-trained ImageNet. image classification use cases, see page detailed examples. transfer learning use cases, make sure read guide transfer learning & fine-tuning.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_b7.html","id":"note","dir":"Reference","previous_headings":"","what":"Note","title":"Instantiates the EfficientNetB7 architecture. — application_efficientnet_b7","text":"Keras Application expects specific kind input preprocessing. EfficientNet, input preprocessing included part model (Rescaling layer), thus application_preprocess_inputs() actually pass-function. EfficientNet models expect inputs float tensors pixels values [0-255] range.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_v2b0.html","id":null,"dir":"Reference","previous_headings":"","what":"Instantiates the EfficientNetV2B0 architecture. — application_efficientnet_v2b0","title":"Instantiates the EfficientNetV2B0 architecture. — application_efficientnet_v2b0","text":"Instantiates EfficientNetV2B0 architecture.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_v2b0.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Instantiates the EfficientNetV2B0 architecture. — application_efficientnet_v2b0","text":"","code":"application_efficientnet_v2b0(   include_top = TRUE,   weights = \"imagenet\",   input_tensor = NULL,   input_shape = NULL,   pooling = NULL,   classes = 1000L,   classifier_activation = \"softmax\",   include_preprocessing = TRUE,   name = \"efficientnetv2-b0\" )"},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_v2b0.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Instantiates the EfficientNetV2B0 architecture. — application_efficientnet_v2b0","text":"include_top Boolean, whether include fully-connected layer top network. Defaults TRUE. weights One NULL (random initialization), \"imagenet\" (pre-training ImageNet), path weights file loaded. Defaults \"imagenet\". input_tensor Optional Keras tensor (.e. output keras_input()) use image input model. input_shape Optional shape tuple, specified include_top FALSE. exactly 3 inputs channels. pooling Optional pooling mode feature extraction include_top FALSE. Defaults NULL. NULL means output model 4D tensor output last convolutional layer. \"avg\" means global average pooling applied output last convolutional layer, thus output model 2D tensor. \"max\" means global max pooling applied. classes Optional number classes classify images , specified include_top TRUE, weights argument specified. Defaults 1000 (number ImageNet classes). classifier_activation string callable. activation function use \"top\" layer. Ignored unless include_top=TRUE. Set classifier_activation=NULL return logits \"top\" layer. Defaults \"softmax\". loading pretrained weights, classifier_activation can NULL \"softmax\". include_preprocessing Boolean, whether include preprocessing layer bottom network. name name model (string).","code":""},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_v2b0.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Instantiates the EfficientNetV2B0 architecture. — application_efficientnet_v2b0","text":"model instance.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_v2b0.html","id":"reference","dir":"Reference","previous_headings":"","what":"Reference","title":"Instantiates the EfficientNetV2B0 architecture. — application_efficientnet_v2b0","text":"EfficientNetV2: Smaller Models Faster Training (ICML 2021) function returns Keras image classification model, optionally loaded weights pre-trained ImageNet. image classification use cases, see page detailed examples. transfer learning use cases, make sure read guide transfer learning & fine-tuning.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_v2b0.html","id":"note","dir":"Reference","previous_headings":"","what":"Note","title":"Instantiates the EfficientNetV2B0 architecture. — application_efficientnet_v2b0","text":"Keras Application expects specific kind input preprocessing. EfficientNetV2, default input preprocessing included part model (Rescaling layer), thus application_preprocess_inputs() actually pass-function. use case, EfficientNetV2 models expect inputs float tensors pixels values [0, 255] range. time, preprocessing part model (.e. Rescaling layer) can disabled setting include_preprocessing argument FALSE. preprocessing disabled EfficientNetV2 models expect inputs float tensors pixels values [-1, 1] range.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_v2b1.html","id":null,"dir":"Reference","previous_headings":"","what":"Instantiates the EfficientNetV2B1 architecture. — application_efficientnet_v2b1","title":"Instantiates the EfficientNetV2B1 architecture. — application_efficientnet_v2b1","text":"Instantiates EfficientNetV2B1 architecture.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_v2b1.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Instantiates the EfficientNetV2B1 architecture. — application_efficientnet_v2b1","text":"","code":"application_efficientnet_v2b1(   include_top = TRUE,   weights = \"imagenet\",   input_tensor = NULL,   input_shape = NULL,   pooling = NULL,   classes = 1000L,   classifier_activation = \"softmax\",   include_preprocessing = TRUE,   name = \"efficientnetv2-b1\" )"},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_v2b1.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Instantiates the EfficientNetV2B1 architecture. — application_efficientnet_v2b1","text":"include_top Boolean, whether include fully-connected layer top network. Defaults TRUE. weights One NULL (random initialization), \"imagenet\" (pre-training ImageNet), path weights file loaded. Defaults \"imagenet\". input_tensor Optional Keras tensor (.e. output keras_input()) use image input model. input_shape Optional shape tuple, specified include_top FALSE. exactly 3 inputs channels. pooling Optional pooling mode feature extraction include_top FALSE. Defaults NULL. NULL means output model 4D tensor output last convolutional layer. \"avg\" means global average pooling applied output last convolutional layer, thus output model 2D tensor. \"max\" means global max pooling applied. classes Optional number classes classify images , specified include_top TRUE, weights argument specified. Defaults 1000 (number ImageNet classes). classifier_activation string callable. activation function use \"top\" layer. Ignored unless include_top=TRUE. Set classifier_activation=NULL return logits \"top\" layer. Defaults \"softmax\". loading pretrained weights, classifier_activation can NULL \"softmax\". include_preprocessing Boolean, whether include preprocessing layer bottom network. name name model (string).","code":""},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_v2b1.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Instantiates the EfficientNetV2B1 architecture. — application_efficientnet_v2b1","text":"model instance.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_v2b1.html","id":"reference","dir":"Reference","previous_headings":"","what":"Reference","title":"Instantiates the EfficientNetV2B1 architecture. — application_efficientnet_v2b1","text":"EfficientNetV2: Smaller Models Faster Training (ICML 2021) function returns Keras image classification model, optionally loaded weights pre-trained ImageNet. image classification use cases, see page detailed examples. transfer learning use cases, make sure read guide transfer learning & fine-tuning.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_v2b1.html","id":"note","dir":"Reference","previous_headings":"","what":"Note","title":"Instantiates the EfficientNetV2B1 architecture. — application_efficientnet_v2b1","text":"Keras Application expects specific kind input preprocessing. EfficientNetV2, default input preprocessing included part model (Rescaling layer), thus application_preprocess_inputs() actually pass-function. use case, EfficientNetV2 models expect inputs float tensors pixels values [0, 255] range. time, preprocessing part model (.e. Rescaling layer) can disabled setting include_preprocessing argument FALSE. preprocessing disabled EfficientNetV2 models expect inputs float tensors pixels values [-1, 1] range.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_v2b2.html","id":null,"dir":"Reference","previous_headings":"","what":"Instantiates the EfficientNetV2B2 architecture. — application_efficientnet_v2b2","title":"Instantiates the EfficientNetV2B2 architecture. — application_efficientnet_v2b2","text":"Instantiates EfficientNetV2B2 architecture.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_v2b2.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Instantiates the EfficientNetV2B2 architecture. — application_efficientnet_v2b2","text":"","code":"application_efficientnet_v2b2(   include_top = TRUE,   weights = \"imagenet\",   input_tensor = NULL,   input_shape = NULL,   pooling = NULL,   classes = 1000L,   classifier_activation = \"softmax\",   include_preprocessing = TRUE,   name = \"efficientnetv2-b2\" )"},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_v2b2.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Instantiates the EfficientNetV2B2 architecture. — application_efficientnet_v2b2","text":"include_top Boolean, whether include fully-connected layer top network. Defaults TRUE. weights One NULL (random initialization), \"imagenet\" (pre-training ImageNet), path weights file loaded. Defaults \"imagenet\". input_tensor Optional Keras tensor (.e. output keras_input()) use image input model. input_shape Optional shape tuple, specified include_top FALSE. exactly 3 inputs channels. pooling Optional pooling mode feature extraction include_top FALSE. Defaults NULL. NULL means output model 4D tensor output last convolutional layer. \"avg\" means global average pooling applied output last convolutional layer, thus output model 2D tensor. \"max\" means global max pooling applied. classes Optional number classes classify images , specified include_top TRUE, weights argument specified. Defaults 1000 (number ImageNet classes). classifier_activation string callable. activation function use \"top\" layer. Ignored unless include_top=TRUE. Set classifier_activation=NULL return logits \"top\" layer. Defaults \"softmax\". loading pretrained weights, classifier_activation can NULL \"softmax\". include_preprocessing Boolean, whether include preprocessing layer bottom network. name name model (string).","code":""},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_v2b2.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Instantiates the EfficientNetV2B2 architecture. — application_efficientnet_v2b2","text":"model instance.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_v2b2.html","id":"reference","dir":"Reference","previous_headings":"","what":"Reference","title":"Instantiates the EfficientNetV2B2 architecture. — application_efficientnet_v2b2","text":"EfficientNetV2: Smaller Models Faster Training (ICML 2021) function returns Keras image classification model, optionally loaded weights pre-trained ImageNet. image classification use cases, see page detailed examples. transfer learning use cases, make sure read guide transfer learning & fine-tuning.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_v2b2.html","id":"note","dir":"Reference","previous_headings":"","what":"Note","title":"Instantiates the EfficientNetV2B2 architecture. — application_efficientnet_v2b2","text":"Keras Application expects specific kind input preprocessing. EfficientNetV2, default input preprocessing included part model (Rescaling layer), thus application_preprocess_inputs() actually pass-function. use case, EfficientNetV2 models expect inputs float tensors pixels values [0, 255] range. time, preprocessing part model (.e. Rescaling layer) can disabled setting include_preprocessing argument FALSE. preprocessing disabled EfficientNetV2 models expect inputs float tensors pixels values [-1, 1] range.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_v2b3.html","id":null,"dir":"Reference","previous_headings":"","what":"Instantiates the EfficientNetV2B3 architecture. — application_efficientnet_v2b3","title":"Instantiates the EfficientNetV2B3 architecture. — application_efficientnet_v2b3","text":"Instantiates EfficientNetV2B3 architecture.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_v2b3.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Instantiates the EfficientNetV2B3 architecture. — application_efficientnet_v2b3","text":"","code":"application_efficientnet_v2b3(   include_top = TRUE,   weights = \"imagenet\",   input_tensor = NULL,   input_shape = NULL,   pooling = NULL,   classes = 1000L,   classifier_activation = \"softmax\",   include_preprocessing = TRUE,   name = \"efficientnetv2-b3\" )"},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_v2b3.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Instantiates the EfficientNetV2B3 architecture. — application_efficientnet_v2b3","text":"include_top Boolean, whether include fully-connected layer top network. Defaults TRUE. weights One NULL (random initialization), \"imagenet\" (pre-training ImageNet), path weights file loaded. Defaults \"imagenet\". input_tensor Optional Keras tensor (.e. output keras_input()) use image input model. input_shape Optional shape tuple, specified include_top FALSE. exactly 3 inputs channels. pooling Optional pooling mode feature extraction include_top FALSE. Defaults NULL. NULL means output model 4D tensor output last convolutional layer. \"avg\" means global average pooling applied output last convolutional layer, thus output model 2D tensor. \"max\" means global max pooling applied. classes Optional number classes classify images , specified include_top TRUE, weights argument specified. Defaults 1000 (number ImageNet classes). classifier_activation string callable. activation function use \"top\" layer. Ignored unless include_top=TRUE. Set classifier_activation=NULL return logits \"top\" layer. Defaults \"softmax\". loading pretrained weights, classifier_activation can NULL \"softmax\". include_preprocessing Boolean, whether include preprocessing layer bottom network. name name model (string).","code":""},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_v2b3.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Instantiates the EfficientNetV2B3 architecture. — application_efficientnet_v2b3","text":"model instance.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_v2b3.html","id":"reference","dir":"Reference","previous_headings":"","what":"Reference","title":"Instantiates the EfficientNetV2B3 architecture. — application_efficientnet_v2b3","text":"EfficientNetV2: Smaller Models Faster Training (ICML 2021) function returns Keras image classification model, optionally loaded weights pre-trained ImageNet. image classification use cases, see page detailed examples. transfer learning use cases, make sure read guide transfer learning & fine-tuning.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_v2b3.html","id":"note","dir":"Reference","previous_headings":"","what":"Note","title":"Instantiates the EfficientNetV2B3 architecture. — application_efficientnet_v2b3","text":"Keras Application expects specific kind input preprocessing. EfficientNetV2, default input preprocessing included part model (Rescaling layer), thus application_preprocess_inputs() actually pass-function. use case, EfficientNetV2 models expect inputs float tensors pixels values [0, 255] range. time, preprocessing part model (.e. Rescaling layer) can disabled setting include_preprocessing argument FALSE. preprocessing disabled EfficientNetV2 models expect inputs float tensors pixels values [-1, 1] range.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_v2l.html","id":null,"dir":"Reference","previous_headings":"","what":"Instantiates the EfficientNetV2L architecture. — application_efficientnet_v2l","title":"Instantiates the EfficientNetV2L architecture. — application_efficientnet_v2l","text":"Instantiates EfficientNetV2L architecture.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_v2l.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Instantiates the EfficientNetV2L architecture. — application_efficientnet_v2l","text":"","code":"application_efficientnet_v2l(   include_top = TRUE,   weights = \"imagenet\",   input_tensor = NULL,   input_shape = NULL,   pooling = NULL,   classes = 1000L,   classifier_activation = \"softmax\",   include_preprocessing = TRUE,   name = \"efficientnetv2-l\" )"},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_v2l.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Instantiates the EfficientNetV2L architecture. — application_efficientnet_v2l","text":"include_top Boolean, whether include fully-connected layer top network. Defaults TRUE. weights One NULL (random initialization), \"imagenet\" (pre-training ImageNet), path weights file loaded. Defaults \"imagenet\". input_tensor Optional Keras tensor (.e. output keras_input()) use image input model. input_shape Optional shape tuple, specified include_top FALSE. exactly 3 inputs channels. pooling Optional pooling mode feature extraction include_top FALSE. Defaults NULL. NULL means output model 4D tensor output last convolutional layer. \"avg\" means global average pooling applied output last convolutional layer, thus output model 2D tensor. \"max\" means global max pooling applied. classes Optional number classes classify images , specified include_top TRUE, weights argument specified. Defaults 1000 (number ImageNet classes). classifier_activation string callable. activation function use \"top\" layer. Ignored unless include_top=TRUE. Set classifier_activation=NULL return logits \"top\" layer. Defaults \"softmax\". loading pretrained weights, classifier_activation can NULL \"softmax\". include_preprocessing Boolean, whether include preprocessing layer bottom network. name name model (string).","code":""},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_v2l.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Instantiates the EfficientNetV2L architecture. — application_efficientnet_v2l","text":"model instance.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_v2l.html","id":"reference","dir":"Reference","previous_headings":"","what":"Reference","title":"Instantiates the EfficientNetV2L architecture. — application_efficientnet_v2l","text":"EfficientNetV2: Smaller Models Faster Training (ICML 2021) function returns Keras image classification model, optionally loaded weights pre-trained ImageNet. image classification use cases, see page detailed examples. transfer learning use cases, make sure read guide transfer learning & fine-tuning.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_v2l.html","id":"note","dir":"Reference","previous_headings":"","what":"Note","title":"Instantiates the EfficientNetV2L architecture. — application_efficientnet_v2l","text":"Keras Application expects specific kind input preprocessing. EfficientNetV2, default input preprocessing included part model (Rescaling layer), thus application_preprocess_inputs() actually pass-function. use case, EfficientNetV2 models expect inputs float tensors pixels values [0, 255] range. time, preprocessing part model (.e. Rescaling layer) can disabled setting include_preprocessing argument FALSE. preprocessing disabled EfficientNetV2 models expect inputs float tensors pixels values [-1, 1] range.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_v2m.html","id":null,"dir":"Reference","previous_headings":"","what":"Instantiates the EfficientNetV2M architecture. — application_efficientnet_v2m","title":"Instantiates the EfficientNetV2M architecture. — application_efficientnet_v2m","text":"Instantiates EfficientNetV2M architecture.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_v2m.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Instantiates the EfficientNetV2M architecture. — application_efficientnet_v2m","text":"","code":"application_efficientnet_v2m(   include_top = TRUE,   weights = \"imagenet\",   input_tensor = NULL,   input_shape = NULL,   pooling = NULL,   classes = 1000L,   classifier_activation = \"softmax\",   include_preprocessing = TRUE,   name = \"efficientnetv2-m\" )"},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_v2m.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Instantiates the EfficientNetV2M architecture. — application_efficientnet_v2m","text":"include_top Boolean, whether include fully-connected layer top network. Defaults TRUE. weights One NULL (random initialization), \"imagenet\" (pre-training ImageNet), path weights file loaded. Defaults \"imagenet\". input_tensor Optional Keras tensor (.e. output keras_input()) use image input model. input_shape Optional shape tuple, specified include_top FALSE. exactly 3 inputs channels. pooling Optional pooling mode feature extraction include_top FALSE. Defaults NULL. NULL means output model 4D tensor output last convolutional layer. \"avg\" means global average pooling applied output last convolutional layer, thus output model 2D tensor. \"max\" means global max pooling applied. classes Optional number classes classify images , specified include_top TRUE, weights argument specified. Defaults 1000 (number ImageNet classes). classifier_activation string callable. activation function use \"top\" layer. Ignored unless include_top=TRUE. Set classifier_activation=NULL return logits \"top\" layer. Defaults \"softmax\". loading pretrained weights, classifier_activation can NULL \"softmax\". include_preprocessing Boolean, whether include preprocessing layer bottom network. name name model (string).","code":""},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_v2m.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Instantiates the EfficientNetV2M architecture. — application_efficientnet_v2m","text":"model instance.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_v2m.html","id":"reference","dir":"Reference","previous_headings":"","what":"Reference","title":"Instantiates the EfficientNetV2M architecture. — application_efficientnet_v2m","text":"EfficientNetV2: Smaller Models Faster Training (ICML 2021) function returns Keras image classification model, optionally loaded weights pre-trained ImageNet. image classification use cases, see page detailed examples. transfer learning use cases, make sure read guide transfer learning & fine-tuning.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_v2m.html","id":"note","dir":"Reference","previous_headings":"","what":"Note","title":"Instantiates the EfficientNetV2M architecture. — application_efficientnet_v2m","text":"Keras Application expects specific kind input preprocessing. EfficientNetV2, default input preprocessing included part model (Rescaling layer), thus application_preprocess_inputs() actually pass-function. use case, EfficientNetV2 models expect inputs float tensors pixels values [0, 255] range. time, preprocessing part model (.e. Rescaling layer) can disabled setting include_preprocessing argument FALSE. preprocessing disabled EfficientNetV2 models expect inputs float tensors pixels values [-1, 1] range.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_v2s.html","id":null,"dir":"Reference","previous_headings":"","what":"Instantiates the EfficientNetV2S architecture. — application_efficientnet_v2s","title":"Instantiates the EfficientNetV2S architecture. — application_efficientnet_v2s","text":"Instantiates EfficientNetV2S architecture.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_v2s.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Instantiates the EfficientNetV2S architecture. — application_efficientnet_v2s","text":"","code":"application_efficientnet_v2s(   include_top = TRUE,   weights = \"imagenet\",   input_tensor = NULL,   input_shape = NULL,   pooling = NULL,   classes = 1000L,   classifier_activation = \"softmax\",   include_preprocessing = TRUE,   name = \"efficientnetv2-s\" )"},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_v2s.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Instantiates the EfficientNetV2S architecture. — application_efficientnet_v2s","text":"include_top Boolean, whether include fully-connected layer top network. Defaults TRUE. weights One NULL (random initialization), \"imagenet\" (pre-training ImageNet), path weights file loaded. Defaults \"imagenet\". input_tensor Optional Keras tensor (.e. output keras_input()) use image input model. input_shape Optional shape tuple, specified include_top FALSE. exactly 3 inputs channels. pooling Optional pooling mode feature extraction include_top FALSE. Defaults NULL. NULL means output model 4D tensor output last convolutional layer. \"avg\" means global average pooling applied output last convolutional layer, thus output model 2D tensor. \"max\" means global max pooling applied. classes Optional number classes classify images , specified include_top TRUE, weights argument specified. Defaults 1000 (number ImageNet classes). classifier_activation string callable. activation function use \"top\" layer. Ignored unless include_top=TRUE. Set classifier_activation=NULL return logits \"top\" layer. Defaults \"softmax\". loading pretrained weights, classifier_activation can NULL \"softmax\". include_preprocessing Boolean, whether include preprocessing layer bottom network. name name model (string).","code":""},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_v2s.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Instantiates the EfficientNetV2S architecture. — application_efficientnet_v2s","text":"model instance.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_v2s.html","id":"reference","dir":"Reference","previous_headings":"","what":"Reference","title":"Instantiates the EfficientNetV2S architecture. — application_efficientnet_v2s","text":"EfficientNetV2: Smaller Models Faster Training (ICML 2021) function returns Keras image classification model, optionally loaded weights pre-trained ImageNet. image classification use cases, see page detailed examples. transfer learning use cases, make sure read guide transfer learning & fine-tuning.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_v2s.html","id":"note","dir":"Reference","previous_headings":"","what":"Note","title":"Instantiates the EfficientNetV2S architecture. — application_efficientnet_v2s","text":"Keras Application expects specific kind input preprocessing. EfficientNetV2, default input preprocessing included part model (Rescaling layer), thus application_preprocess_inputs() actually pass-function. use case, EfficientNetV2 models expect inputs float tensors pixels values [0, 255] range. time, preprocessing part model (.e. Rescaling layer) can disabled setting include_preprocessing argument FALSE. preprocessing disabled EfficientNetV2 models expect inputs float tensors pixels values [-1, 1] range.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/application_inception_resnet_v2.html","id":null,"dir":"Reference","previous_headings":"","what":"Instantiates the Inception-ResNet v2 architecture. — application_inception_resnet_v2","title":"Instantiates the Inception-ResNet v2 architecture. — application_inception_resnet_v2","text":"Instantiates Inception-ResNet v2 architecture.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_inception_resnet_v2.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Instantiates the Inception-ResNet v2 architecture. — application_inception_resnet_v2","text":"","code":"application_inception_resnet_v2(   include_top = TRUE,   weights = \"imagenet\",   input_tensor = NULL,   input_shape = NULL,   pooling = NULL,   classes = 1000L,   classifier_activation = \"softmax\",   name = \"inception_resnet_v2\" )"},{"path":"https://keras3.posit.co/dev/reference/application_inception_resnet_v2.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Instantiates the Inception-ResNet v2 architecture. — application_inception_resnet_v2","text":"include_top whether include fully-connected layer top network. weights one NULL (random initialization), \"imagenet\" (pre-training ImageNet), path weights file loaded. input_tensor optional Keras tensor (.e. output keras_input()) use image input model. input_shape optional shape tuple, specified include_top FALSE (otherwise input shape (299, 299, 3) ('channels_last' data format) (3, 299, 299) ('channels_first' data format). exactly 3 inputs channels, width height smaller 75. E.g. (150, 150, 3) one valid value. pooling Optional pooling mode feature extraction include_top FALSE. NULL means output model 4D tensor output last convolutional block. 'avg' means global average pooling applied output last convolutional block, thus output model 2D tensor. 'max' means global max pooling applied. classes optional number classes classify images , specified include_top TRUE, weights argument specified. classifier_activation str callable. activation function use \"top\" layer. Ignored unless include_top=TRUE. Set classifier_activation=NULL return logits \"top\" layer. loading pretrained weights, classifier_activation can NULL \"softmax\". name name model (string).","code":""},{"path":"https://keras3.posit.co/dev/reference/application_inception_resnet_v2.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Instantiates the Inception-ResNet v2 architecture. — application_inception_resnet_v2","text":"model instance.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_inception_resnet_v2.html","id":"reference","dir":"Reference","previous_headings":"","what":"Reference","title":"Instantiates the Inception-ResNet v2 architecture. — application_inception_resnet_v2","text":"Inception-v4, Inception-ResNet Impact Residual Connections Learning (AAAI 2017) function returns Keras image classification model, optionally loaded weights pre-trained ImageNet. image classification use cases, see page detailed examples. transfer learning use cases, make sure read guide transfer learning & fine-tuning.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_inception_resnet_v2.html","id":"note","dir":"Reference","previous_headings":"","what":"Note","title":"Instantiates the Inception-ResNet v2 architecture. — application_inception_resnet_v2","text":"Keras Application expects specific kind input preprocessing. InceptionResNetV2, call application_preprocess_inputs() inputs passing model. application_preprocess_inputs() scale input pixels -1 1.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/application_inception_v3.html","id":null,"dir":"Reference","previous_headings":"","what":"Instantiates the Inception v3 architecture. — application_inception_v3","title":"Instantiates the Inception v3 architecture. — application_inception_v3","text":"Instantiates Inception v3 architecture.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_inception_v3.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Instantiates the Inception v3 architecture. — application_inception_v3","text":"","code":"application_inception_v3(   include_top = TRUE,   weights = \"imagenet\",   input_tensor = NULL,   input_shape = NULL,   pooling = NULL,   classes = 1000L,   classifier_activation = \"softmax\",   name = \"inception_v3\" )"},{"path":"https://keras3.posit.co/dev/reference/application_inception_v3.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Instantiates the Inception v3 architecture. — application_inception_v3","text":"include_top Boolean, whether include fully-connected layer top, last layer network. Defaults TRUE. weights One NULL (random initialization), imagenet (pre-training ImageNet), path weights file loaded. Defaults \"imagenet\". input_tensor Optional Keras tensor (.e. output keras_input()) use image input model. input_tensor useful sharing inputs multiple different networks. Defaults NULL. input_shape Optional shape tuple, specified include_top FALSE (otherwise input shape (299, 299, 3) (channels_last data format) (3, 299, 299) (channels_first data format). exactly 3 inputs channels, width height smaller 75. E.g. (150, 150, 3) one valid value. input_shape ignored input_tensor provided. pooling Optional pooling mode feature extraction include_top FALSE. NULL (default) means output model 4D tensor output last convolutional block. avg means global average pooling applied output last convolutional block, thus output model 2D tensor. max means global max pooling applied. classes optional number classes classify images , specified include_top TRUE, weights argument specified. Defaults 1000. classifier_activation str callable. activation function use \"top\" layer. Ignored unless include_top=TRUE. Set classifier_activation=NULL return logits \"top\" layer. loading pretrained weights, classifier_activation can NULL \"softmax\". name name model (string).","code":""},{"path":"https://keras3.posit.co/dev/reference/application_inception_v3.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Instantiates the Inception v3 architecture. — application_inception_v3","text":"model instance.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_inception_v3.html","id":"reference","dir":"Reference","previous_headings":"","what":"Reference","title":"Instantiates the Inception v3 architecture. — application_inception_v3","text":"Rethinking Inception Architecture Computer Vision (CVPR 2016) function returns Keras image classification model, optionally loaded weights pre-trained ImageNet. image classification use cases, see page detailed examples. transfer learning use cases, make sure read guide transfer learning & fine-tuning.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_inception_v3.html","id":"note","dir":"Reference","previous_headings":"","what":"Note","title":"Instantiates the Inception v3 architecture. — application_inception_v3","text":"Keras Application expects specific kind input preprocessing. InceptionV3, call application_preprocess_inputs() inputs passing model. application_preprocess_inputs() scale input pixels -1 1.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/application_mobilenet.html","id":null,"dir":"Reference","previous_headings":"","what":"Instantiates the MobileNet architecture. — application_mobilenet","title":"Instantiates the MobileNet architecture. — application_mobilenet","text":"Instantiates MobileNet architecture.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_mobilenet.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Instantiates the MobileNet architecture. — application_mobilenet","text":"","code":"application_mobilenet(   input_shape = NULL,   alpha = 1,   depth_multiplier = 1L,   dropout = 0.001,   include_top = TRUE,   weights = \"imagenet\",   input_tensor = NULL,   pooling = NULL,   classes = 1000L,   classifier_activation = \"softmax\",   name = NULL )"},{"path":"https://keras3.posit.co/dev/reference/application_mobilenet.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Instantiates the MobileNet architecture. — application_mobilenet","text":"input_shape Optional shape tuple, specified include_top FALSE (otherwise input shape (224, 224, 3) (\"channels_last\" data format) (3, 224, 224) (\"channels_first\" data format). exactly 3 inputs channels, width height smaller 32. E.g. (200, 200, 3) one valid value. Defaults NULL. input_shape ignored input_tensor provided. alpha Controls width network. known width multiplier MobileNet paper. alpha < 1.0, proportionally decreases number filters layer. alpha > 1.0, proportionally increases number filters layer. alpha == 1, default number filters paper used layer. Defaults 1.0. depth_multiplier Depth multiplier depthwise convolution. called resolution multiplier MobileNet paper. Defaults 1.0. dropout Dropout rate. Defaults 0.001. include_top Boolean, whether include fully-connected layer top network. Defaults TRUE. weights One NULL (random initialization), \"imagenet\" (pre-training ImageNet), path weights file loaded. Defaults \"imagenet\". input_tensor Optional Keras tensor (.e. output keras_input()) use image input model. input_tensor useful sharing inputs multiple different networks. Defaults NULL. pooling Optional pooling mode feature extraction include_top FALSE. NULL (default) means output model 4D tensor output last convolutional block. avg means global average pooling applied output last convolutional block, thus output model 2D tensor. max means global max pooling applied. classes Optional number classes classify images , specified include_top TRUE, weights argument specified. Defaults 1000. classifier_activation str callable. activation function use \"top\" layer. Ignored unless include_top=TRUE. Set classifier_activation=NULL return logits \"top\" layer. loading pretrained weights, classifier_activation can NULL \"softmax\". name name model (string).","code":""},{"path":"https://keras3.posit.co/dev/reference/application_mobilenet.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Instantiates the MobileNet architecture. — application_mobilenet","text":"model instance.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_mobilenet.html","id":"reference","dir":"Reference","previous_headings":"","what":"Reference","title":"Instantiates the MobileNet architecture. — application_mobilenet","text":"MobileNets: Efficient Convolutional Neural Networks Mobile Vision Applications function returns Keras image classification model, optionally loaded weights pre-trained ImageNet. image classification use cases, see page detailed examples. transfer learning use cases, make sure read guide transfer learning & fine-tuning.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_mobilenet.html","id":"note","dir":"Reference","previous_headings":"","what":"Note","title":"Instantiates the MobileNet architecture. — application_mobilenet","text":"Keras Application expects specific kind input preprocessing. MobileNet, call application_preprocess_inputs() inputs passing model. application_preprocess_inputs() scale input pixels -1 1.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/application_mobilenet_v2.html","id":null,"dir":"Reference","previous_headings":"","what":"Instantiates the MobileNetV2 architecture. — application_mobilenet_v2","title":"Instantiates the MobileNetV2 architecture. — application_mobilenet_v2","text":"MobileNetV2 similar original MobileNet, except uses inverted residual blocks bottlenecking features. drastically lower parameter count original MobileNet. MobileNets support input size greater 32 x 32, larger image sizes offering better performance.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_mobilenet_v2.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Instantiates the MobileNetV2 architecture. — application_mobilenet_v2","text":"","code":"application_mobilenet_v2(   input_shape = NULL,   alpha = 1,   include_top = TRUE,   weights = \"imagenet\",   input_tensor = NULL,   pooling = NULL,   classes = 1000L,   classifier_activation = \"softmax\",   name = NULL )"},{"path":"https://keras3.posit.co/dev/reference/application_mobilenet_v2.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Instantiates the MobileNetV2 architecture. — application_mobilenet_v2","text":"input_shape Optional shape tuple, specified include_top FALSE (otherwise input shape (224, 224, 3) (\"channels_last\" data format) (3, 224, 224) (\"channels_first\" data format). exactly 3 inputs channels, width height smaller 32. E.g. (200, 200, 3) one valid value. Defaults NULL. input_shape ignored input_tensor provided. alpha Controls width network. known width multiplier MobileNet paper. alpha < 1.0, proportionally decreases number filters layer. alpha > 1.0, proportionally increases number filters layer. alpha == 1, default number filters paper used layer. Defaults 1.0. include_top Boolean, whether include fully-connected layer top network. Defaults TRUE. weights One NULL (random initialization), \"imagenet\" (pre-training ImageNet), path weights file loaded. Defaults \"imagenet\". input_tensor Optional Keras tensor (.e. output keras_input()) use image input model. input_tensor useful sharing inputs multiple different networks. Defaults NULL. pooling Optional pooling mode feature extraction include_top FALSE. NULL (default) means output model 4D tensor output last convolutional block. avg means global average pooling applied output last convolutional block, thus output model 2D tensor. max means global max pooling applied. classes Optional number classes classify images , specified include_top TRUE, weights argument specified. Defaults 1000. classifier_activation str callable. activation function use \"top\" layer. Ignored unless include_top=TRUE. Set classifier_activation=NULL return logits \"top\" layer. loading pretrained weights, classifier_activation can NULL \"softmax\". name name model (string).","code":""},{"path":"https://keras3.posit.co/dev/reference/application_mobilenet_v2.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Instantiates the MobileNetV2 architecture. — application_mobilenet_v2","text":"model instance.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_mobilenet_v2.html","id":"reference","dir":"Reference","previous_headings":"","what":"Reference","title":"Instantiates the MobileNetV2 architecture. — application_mobilenet_v2","text":"MobileNetV2: Inverted Residuals Linear Bottlenecks (CVPR 2018) function returns Keras image classification model, optionally loaded weights pre-trained ImageNet. image classification use cases, see page detailed examples. transfer learning use cases, make sure read guide transfer learning & fine-tuning.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_mobilenet_v2.html","id":"note","dir":"Reference","previous_headings":"","what":"Note","title":"Instantiates the MobileNetV2 architecture. — application_mobilenet_v2","text":"Keras Application expects specific kind input preprocessing. MobileNetV2, call application_preprocess_inputs() inputs passing model. application_preprocess_inputs() scale input pixels -1 1.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/application_mobilenet_v3_large.html","id":null,"dir":"Reference","previous_headings":"","what":"Instantiates the MobileNetV3Large architecture. — application_mobilenet_v3_large","title":"Instantiates the MobileNetV3Large architecture. — application_mobilenet_v3_large","text":"Instantiates MobileNetV3Large architecture.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_mobilenet_v3_large.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Instantiates the MobileNetV3Large architecture. — application_mobilenet_v3_large","text":"","code":"application_mobilenet_v3_large(   input_shape = NULL,   alpha = 1,   minimalistic = FALSE,   include_top = TRUE,   weights = \"imagenet\",   input_tensor = NULL,   classes = 1000L,   pooling = NULL,   dropout_rate = 0.2,   classifier_activation = \"softmax\",   include_preprocessing = TRUE,   name = \"MobileNetV3Large\" )"},{"path":"https://keras3.posit.co/dev/reference/application_mobilenet_v3_large.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Instantiates the MobileNetV3Large architecture. — application_mobilenet_v3_large","text":"input_shape Optional shape tuple, specified like use model input image resolution (224, 224, 3). exactly 3 inputs channels. can also omit option like infer input_shape input_tensor. choose include input_tensor input_shape input_shape used match, shapes match throw error. E.g. (160, 160, 3) one valid value. alpha controls width network. known depth multiplier MobileNetV3 paper, name kept consistency MobileNetV1 Keras. alpha < 1.0, proportionally decreases number filters layer. alpha > 1.0, proportionally increases number filters layer. alpha == 1, default number filters paper used layer. minimalistic addition large small models module also contains -called minimalistic models, models per-layer dimensions characteristic MobilenetV3 however, utilize advanced blocks (squeeze--excite units, hard-swish, 5x5 convolutions). models less efficient CPU, much performant GPU/DSP. include_top Boolean, whether include fully-connected layer top network. Defaults TRUE. weights String, one NULL (random initialization), \"imagenet\" (pre-training ImageNet), path weights file loaded. input_tensor Optional Keras tensor (.e. output keras_input()) use image input model. classes Integer, optional number classes classify images , specified include_top TRUE, weights argument specified. pooling String, optional pooling mode feature extraction include_top FALSE. NULL means output model 4D tensor output last convolutional block. avg means global average pooling applied output last convolutional block, thus output model 2D tensor. max means global max pooling applied. dropout_rate fraction input units drop last layer. classifier_activation str callable. activation function use \"top\" layer. Ignored unless include_top=TRUE. Set classifier_activation=NULL return logits \"top\" layer. loading pretrained weights, classifier_activation can NULL \"softmax\". include_preprocessing Boolean, whether include preprocessing layer (Rescaling) bottom network. Defaults TRUE. name name model (string).","code":""},{"path":"https://keras3.posit.co/dev/reference/application_mobilenet_v3_large.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Instantiates the MobileNetV3Large architecture. — application_mobilenet_v3_large","text":"model instance.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_mobilenet_v3_large.html","id":"reference","dir":"Reference","previous_headings":"","what":"Reference","title":"Instantiates the MobileNetV3Large architecture. — application_mobilenet_v3_large","text":"Searching MobileNetV3 (ICCV 2019)","code":""},{"path":"https://keras3.posit.co/dev/reference/application_mobilenet_v3_large.html","id":"the-following-table-describes-the-performance-of-mobilenets-v-","dir":"Reference","previous_headings":"","what":"The following table describes the performance of MobileNets v3:","title":"Instantiates the MobileNetV3Large architecture. — application_mobilenet_v3_large","text":"MACs stands Multiply Adds image classification use cases, see page detailed examples. transfer learning use cases, make sure read guide transfer learning & fine-tuning.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_mobilenet_v3_large.html","id":"note","dir":"Reference","previous_headings":"","what":"Note","title":"Instantiates the MobileNetV3Large architecture. — application_mobilenet_v3_large","text":"Keras Application expects specific kind input preprocessing. MobileNetV3, default input preprocessing included part model (Rescaling layer), thus application_preprocess_inputs() actually pass-function. use case, MobileNetV3 models expect inputs float tensors pixels values [0-255] range. time, preprocessing part model (.e. Rescaling layer) can disabled setting include_preprocessing argument FALSE. preprocessing disabled MobileNetV3 models expect inputs float tensors pixels values [-1, 1] range.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_mobilenet_v3_large.html","id":"call-arguments","dir":"Reference","previous_headings":"","what":"Call Arguments","title":"Instantiates the MobileNetV3Large architecture. — application_mobilenet_v3_large","text":"inputs: floating point numpy.array backend-native tensor, 4D 3 color channels, values range [0, 255] include_preprocessing TRUE range [-1, 1] otherwise.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/application_mobilenet_v3_small.html","id":null,"dir":"Reference","previous_headings":"","what":"Instantiates the MobileNetV3Small architecture. — application_mobilenet_v3_small","title":"Instantiates the MobileNetV3Small architecture. — application_mobilenet_v3_small","text":"Instantiates MobileNetV3Small architecture.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_mobilenet_v3_small.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Instantiates the MobileNetV3Small architecture. — application_mobilenet_v3_small","text":"","code":"application_mobilenet_v3_small(   input_shape = NULL,   alpha = 1,   minimalistic = FALSE,   include_top = TRUE,   weights = \"imagenet\",   input_tensor = NULL,   classes = 1000L,   pooling = NULL,   dropout_rate = 0.2,   classifier_activation = \"softmax\",   include_preprocessing = TRUE,   name = \"MobileNetV3Small\" )"},{"path":"https://keras3.posit.co/dev/reference/application_mobilenet_v3_small.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Instantiates the MobileNetV3Small architecture. — application_mobilenet_v3_small","text":"input_shape Optional shape tuple, specified like use model input image resolution (224, 224, 3). exactly 3 inputs channels. can also omit option like infer input_shape input_tensor. choose include input_tensor input_shape input_shape used match, shapes match throw error. E.g. (160, 160, 3) one valid value. alpha controls width network. known depth multiplier MobileNetV3 paper, name kept consistency MobileNetV1 Keras. alpha < 1.0, proportionally decreases number filters layer. alpha > 1.0, proportionally increases number filters layer. alpha == 1, default number filters paper used layer. minimalistic addition large small models module also contains -called minimalistic models, models per-layer dimensions characteristic MobilenetV3 however, utilize advanced blocks (squeeze--excite units, hard-swish, 5x5 convolutions). models less efficient CPU, much performant GPU/DSP. include_top Boolean, whether include fully-connected layer top network. Defaults TRUE. weights String, one NULL (random initialization), \"imagenet\" (pre-training ImageNet), path weights file loaded. input_tensor Optional Keras tensor (.e. output keras_input()) use image input model. classes Integer, optional number classes classify images , specified include_top TRUE, weights argument specified. pooling String, optional pooling mode feature extraction include_top FALSE. NULL means output model 4D tensor output last convolutional block. avg means global average pooling applied output last convolutional block, thus output model 2D tensor. max means global max pooling applied. dropout_rate fraction input units drop last layer. classifier_activation str callable. activation function use \"top\" layer. Ignored unless include_top=TRUE. Set classifier_activation=NULL return logits \"top\" layer. loading pretrained weights, classifier_activation can NULL \"softmax\". include_preprocessing Boolean, whether include preprocessing layer (Rescaling) bottom network. Defaults TRUE. name name model (string).","code":""},{"path":"https://keras3.posit.co/dev/reference/application_mobilenet_v3_small.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Instantiates the MobileNetV3Small architecture. — application_mobilenet_v3_small","text":"model instance.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_mobilenet_v3_small.html","id":"reference","dir":"Reference","previous_headings":"","what":"Reference","title":"Instantiates the MobileNetV3Small architecture. — application_mobilenet_v3_small","text":"Searching MobileNetV3 (ICCV 2019)","code":""},{"path":"https://keras3.posit.co/dev/reference/application_mobilenet_v3_small.html","id":"the-following-table-describes-the-performance-of-mobilenets-v-","dir":"Reference","previous_headings":"","what":"The following table describes the performance of MobileNets v3:","title":"Instantiates the MobileNetV3Small architecture. — application_mobilenet_v3_small","text":"MACs stands Multiply Adds image classification use cases, see page detailed examples. transfer learning use cases, make sure read guide transfer learning & fine-tuning.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_mobilenet_v3_small.html","id":"note","dir":"Reference","previous_headings":"","what":"Note","title":"Instantiates the MobileNetV3Small architecture. — application_mobilenet_v3_small","text":"Keras Application expects specific kind input preprocessing. MobileNetV3, default input preprocessing included part model (Rescaling layer), thus application_preprocess_inputs() actually pass-function. use case, MobileNetV3 models expect inputs float tensors pixels values [0-255] range. time, preprocessing part model (.e. Rescaling layer) can disabled setting include_preprocessing argument FALSE. preprocessing disabled MobileNetV3 models expect inputs float tensors pixels values [-1, 1] range.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_mobilenet_v3_small.html","id":"call-arguments","dir":"Reference","previous_headings":"","what":"Call Arguments","title":"Instantiates the MobileNetV3Small architecture. — application_mobilenet_v3_small","text":"inputs: floating point numpy.array backend-native tensor, 4D 3 color channels, values range [0, 255] include_preprocessing TRUE range [-1, 1] otherwise.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/application_nasnet_large.html","id":null,"dir":"Reference","previous_headings":"","what":"Instantiates a NASNet model in ImageNet mode. — application_nasnet_large","title":"Instantiates a NASNet model in ImageNet mode. — application_nasnet_large","text":"Instantiates NASNet model ImageNet mode.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_nasnet_large.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Instantiates a NASNet model in ImageNet mode. — application_nasnet_large","text":"","code":"application_nasnet_large(   input_shape = NULL,   include_top = TRUE,   weights = \"imagenet\",   input_tensor = NULL,   pooling = NULL,   classes = 1000L,   classifier_activation = \"softmax\",   name = \"nasnet_large\" )"},{"path":"https://keras3.posit.co/dev/reference/application_nasnet_large.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Instantiates a NASNet model in ImageNet mode. — application_nasnet_large","text":"input_shape Optional shape tuple, specified include_top FALSE (otherwise input shape (331, 331, 3) NASNetLarge. exactly 3 inputs channels, width height smaller 32. E.g. (224, 224, 3) one valid value. include_top Whether include fully-connected layer top network. weights NULL (random initialization) imagenet (ImageNet weights).  loading imagenet weights, input_shape (331, 331, 3) input_tensor Optional Keras tensor (.e. output keras_input()) use image input model. pooling Optional pooling mode feature extraction include_top FALSE. NULL means output model 4D tensor output last convolutional layer. avg means global average pooling applied output last convolutional layer, thus output model 2D tensor. max means global max pooling applied. classes Optional number classes classify images , specified include_top TRUE, weights argument specified. classifier_activation str callable. activation function use \"top\" layer. Ignored unless include_top=TRUE. Set classifier_activation=NULL return logits \"top\" layer.  loading pretrained weights, classifier_activation can NULL \"softmax\". name name model (string).","code":""},{"path":"https://keras3.posit.co/dev/reference/application_nasnet_large.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Instantiates a NASNet model in ImageNet mode. — application_nasnet_large","text":"Keras model instance.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_nasnet_large.html","id":"reference","dir":"Reference","previous_headings":"","what":"Reference","title":"Instantiates a NASNet model in ImageNet mode. — application_nasnet_large","text":"Learning Transferable Architectures Scalable Image Recognition (CVPR 2018) Optionally loads weights pre-trained ImageNet. Note data format convention used model one specified Keras config ~/.keras/keras.json.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_nasnet_large.html","id":"note","dir":"Reference","previous_headings":"","what":"Note","title":"Instantiates a NASNet model in ImageNet mode. — application_nasnet_large","text":"Keras Application expects specific kind input preprocessing. NASNet, call application_preprocess_inputs() inputs passing model.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/application_nasnet_mobile.html","id":null,"dir":"Reference","previous_headings":"","what":"Instantiates a Mobile NASNet model in ImageNet mode. — application_nasnet_mobile","title":"Instantiates a Mobile NASNet model in ImageNet mode. — application_nasnet_mobile","text":"Instantiates Mobile NASNet model ImageNet mode.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_nasnet_mobile.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Instantiates a Mobile NASNet model in ImageNet mode. — application_nasnet_mobile","text":"","code":"application_nasnet_mobile(   input_shape = NULL,   include_top = TRUE,   weights = \"imagenet\",   input_tensor = NULL,   pooling = NULL,   classes = 1000L,   classifier_activation = \"softmax\",   name = \"nasnet_mobile\" )"},{"path":"https://keras3.posit.co/dev/reference/application_nasnet_mobile.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Instantiates a Mobile NASNet model in ImageNet mode. — application_nasnet_mobile","text":"input_shape Optional shape tuple, specified include_top FALSE (otherwise input shape (224, 224, 3) NASNetMobile exactly 3 inputs channels, width height smaller 32. E.g. (224, 224, 3) one valid value. include_top Whether include fully-connected layer top network. weights NULL (random initialization) imagenet (ImageNet weights). loading imagenet weights, input_shape (224, 224, 3) input_tensor Optional Keras tensor (.e. output keras_input()) use image input model. pooling Optional pooling mode feature extraction include_top FALSE. NULL means output model 4D tensor output last convolutional layer. avg means global average pooling applied output last convolutional layer, thus output model 2D tensor. max means global max pooling applied. classes Optional number classes classify images , specified include_top TRUE, weights argument specified. classifier_activation str callable. activation function use \"top\" layer. Ignored unless include_top=TRUE. Set classifier_activation=NULL return logits \"top\" layer.  loading pretrained weights, classifier_activation can NULL \"softmax\". name name model (string).","code":""},{"path":"https://keras3.posit.co/dev/reference/application_nasnet_mobile.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Instantiates a Mobile NASNet model in ImageNet mode. — application_nasnet_mobile","text":"Keras model instance.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_nasnet_mobile.html","id":"reference","dir":"Reference","previous_headings":"","what":"Reference","title":"Instantiates a Mobile NASNet model in ImageNet mode. — application_nasnet_mobile","text":"Learning Transferable Architectures Scalable Image Recognition (CVPR 2018) Optionally loads weights pre-trained ImageNet. Note data format convention used model one specified Keras config ~/.keras/keras.json.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_nasnet_mobile.html","id":"note","dir":"Reference","previous_headings":"","what":"Note","title":"Instantiates a Mobile NASNet model in ImageNet mode. — application_nasnet_mobile","text":"Keras Application expects specific kind input preprocessing. NASNet, call application_preprocess_inputs() inputs passing model.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/application_resnet101.html","id":null,"dir":"Reference","previous_headings":"","what":"Instantiates the ResNet101 architecture. — application_resnet101","title":"Instantiates the ResNet101 architecture. — application_resnet101","text":"Instantiates ResNet101 architecture.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_resnet101.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Instantiates the ResNet101 architecture. — application_resnet101","text":"","code":"application_resnet101(   include_top = TRUE,   weights = \"imagenet\",   input_tensor = NULL,   input_shape = NULL,   pooling = NULL,   classes = 1000L,   classifier_activation = \"softmax\",   name = \"resnet101\" )"},{"path":"https://keras3.posit.co/dev/reference/application_resnet101.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Instantiates the ResNet101 architecture. — application_resnet101","text":"include_top whether include fully-connected layer top network. weights one NULL (random initialization), \"imagenet\" (pre-training ImageNet), path weights file loaded. input_tensor optional Keras tensor (.e. output keras_input()) use image input model. input_shape optional shape tuple, specified include_top FALSE (otherwise input shape (224, 224, 3) (\"channels_last\" data format) (3, 224, 224) (\"channels_first\" data format). exactly 3 inputs channels, width height smaller 32. E.g. (200, 200, 3) one valid value. pooling Optional pooling mode feature extraction include_top FALSE. NULL means output model 4D tensor output last convolutional block. avg means global average pooling applied output last convolutional block, thus output model 2D tensor. max means global max pooling applied. classes optional number classes classify images , specified include_top TRUE, weights argument specified. classifier_activation str callable. activation function use \"top\" layer. Ignored unless include_top=TRUE. Set classifier_activation=NULL return logits \"top\" layer. loading pretrained weights, classifier_activation can NULL \"softmax\". name name model (string).","code":""},{"path":"https://keras3.posit.co/dev/reference/application_resnet101.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Instantiates the ResNet101 architecture. — application_resnet101","text":"Model instance.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_resnet101.html","id":"reference","dir":"Reference","previous_headings":"","what":"Reference","title":"Instantiates the ResNet101 architecture. — application_resnet101","text":"Deep Residual Learning Image Recognition (CVPR 2015) image classification use cases, see page detailed examples. transfer learning use cases, make sure read guide transfer learning & fine-tuning.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_resnet101.html","id":"note","dir":"Reference","previous_headings":"","what":"Note","title":"Instantiates the ResNet101 architecture. — application_resnet101","text":"Keras Application expects specific kind input preprocessing. ResNet, call application_preprocess_inputs() inputs passing model. application_preprocess_inputs() convert input images RGB BGR, zero-center color channel respect ImageNet dataset, without scaling.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/application_resnet101_v2.html","id":null,"dir":"Reference","previous_headings":"","what":"Instantiates the ResNet101V2 architecture. — application_resnet101_v2","title":"Instantiates the ResNet101V2 architecture. — application_resnet101_v2","text":"Instantiates ResNet101V2 architecture.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_resnet101_v2.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Instantiates the ResNet101V2 architecture. — application_resnet101_v2","text":"","code":"application_resnet101_v2(   include_top = TRUE,   weights = \"imagenet\",   input_tensor = NULL,   input_shape = NULL,   pooling = NULL,   classes = 1000L,   classifier_activation = \"softmax\",   name = \"resnet101v2\" )"},{"path":"https://keras3.posit.co/dev/reference/application_resnet101_v2.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Instantiates the ResNet101V2 architecture. — application_resnet101_v2","text":"include_top whether include fully-connected layer top network. weights one NULL (random initialization), \"imagenet\" (pre-training ImageNet), path weights file loaded. input_tensor optional Keras tensor (.e. output keras_input()) use image input model. input_shape optional shape tuple, specified include_top FALSE (otherwise input shape (224, 224, 3) (\"channels_last\" data format) (3, 224, 224) (\"channels_first\" data format). exactly 3 inputs channels, width height smaller 32. E.g. (200, 200, 3) one valid value. pooling Optional pooling mode feature extraction include_top FALSE. NULL means output model 4D tensor output last convolutional block. avg means global average pooling applied output last convolutional block, thus output model 2D tensor. max means global max pooling applied. classes optional number classes classify images , specified include_top TRUE, weights argument specified. classifier_activation str callable. activation function use \"top\" layer. Ignored unless include_top=TRUE. Set classifier_activation=NULL return logits \"top\" layer. loading pretrained weights, classifier_activation can NULL \"softmax\". name name model (string).","code":""},{"path":"https://keras3.posit.co/dev/reference/application_resnet101_v2.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Instantiates the ResNet101V2 architecture. — application_resnet101_v2","text":"Model instance.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_resnet101_v2.html","id":"reference","dir":"Reference","previous_headings":"","what":"Reference","title":"Instantiates the ResNet101V2 architecture. — application_resnet101_v2","text":"Identity Mappings Deep Residual Networks (CVPR 2016) image classification use cases, see page detailed examples. transfer learning use cases, make sure read guide transfer learning & fine-tuning.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_resnet101_v2.html","id":"note","dir":"Reference","previous_headings":"","what":"Note","title":"Instantiates the ResNet101V2 architecture. — application_resnet101_v2","text":"Keras Application expects specific kind input preprocessing. ResNet, call application_preprocess_inputs() inputs passing model. application_preprocess_inputs() scale input pixels -1 1.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/application_resnet152.html","id":null,"dir":"Reference","previous_headings":"","what":"Instantiates the ResNet152 architecture. — application_resnet152","title":"Instantiates the ResNet152 architecture. — application_resnet152","text":"Instantiates ResNet152 architecture.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_resnet152.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Instantiates the ResNet152 architecture. — application_resnet152","text":"","code":"application_resnet152(   include_top = TRUE,   weights = \"imagenet\",   input_tensor = NULL,   input_shape = NULL,   pooling = NULL,   classes = 1000L,   classifier_activation = \"softmax\",   name = \"resnet152\" )"},{"path":"https://keras3.posit.co/dev/reference/application_resnet152.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Instantiates the ResNet152 architecture. — application_resnet152","text":"include_top whether include fully-connected layer top network. weights one NULL (random initialization), \"imagenet\" (pre-training ImageNet), path weights file loaded. input_tensor optional Keras tensor (.e. output keras_input()) use image input model. input_shape optional shape tuple, specified include_top FALSE (otherwise input shape (224, 224, 3) (\"channels_last\" data format) (3, 224, 224) (\"channels_first\" data format). exactly 3 inputs channels, width height smaller 32. E.g. (200, 200, 3) one valid value. pooling Optional pooling mode feature extraction include_top FALSE. NULL means output model 4D tensor output last convolutional block. avg means global average pooling applied output last convolutional block, thus output model 2D tensor. max means global max pooling applied. classes optional number classes classify images , specified include_top TRUE, weights argument specified. classifier_activation str callable. activation function use \"top\" layer. Ignored unless include_top=TRUE. Set classifier_activation=NULL return logits \"top\" layer. loading pretrained weights, classifier_activation can NULL \"softmax\". name name model (string).","code":""},{"path":"https://keras3.posit.co/dev/reference/application_resnet152.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Instantiates the ResNet152 architecture. — application_resnet152","text":"Model instance.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_resnet152.html","id":"reference","dir":"Reference","previous_headings":"","what":"Reference","title":"Instantiates the ResNet152 architecture. — application_resnet152","text":"Deep Residual Learning Image Recognition (CVPR 2015) image classification use cases, see page detailed examples. transfer learning use cases, make sure read guide transfer learning & fine-tuning.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_resnet152.html","id":"note","dir":"Reference","previous_headings":"","what":"Note","title":"Instantiates the ResNet152 architecture. — application_resnet152","text":"Keras Application expects specific kind input preprocessing. ResNet, call application_preprocess_inputs() inputs passing model. application_preprocess_inputs() convert input images RGB BGR, zero-center color channel respect ImageNet dataset, without scaling.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/application_resnet152_v2.html","id":null,"dir":"Reference","previous_headings":"","what":"Instantiates the ResNet152V2 architecture. — application_resnet152_v2","title":"Instantiates the ResNet152V2 architecture. — application_resnet152_v2","text":"Instantiates ResNet152V2 architecture.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_resnet152_v2.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Instantiates the ResNet152V2 architecture. — application_resnet152_v2","text":"","code":"application_resnet152_v2(   include_top = TRUE,   weights = \"imagenet\",   input_tensor = NULL,   input_shape = NULL,   pooling = NULL,   classes = 1000L,   classifier_activation = \"softmax\",   name = \"resnet152v2\" )"},{"path":"https://keras3.posit.co/dev/reference/application_resnet152_v2.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Instantiates the ResNet152V2 architecture. — application_resnet152_v2","text":"include_top whether include fully-connected layer top network. weights one NULL (random initialization), \"imagenet\" (pre-training ImageNet), path weights file loaded. input_tensor optional Keras tensor (.e. output keras_input()) use image input model. input_shape optional shape tuple, specified include_top FALSE (otherwise input shape (224, 224, 3) (\"channels_last\" data format) (3, 224, 224) (\"channels_first\" data format). exactly 3 inputs channels, width height smaller 32. E.g. (200, 200, 3) one valid value. pooling Optional pooling mode feature extraction include_top FALSE. NULL means output model 4D tensor output last convolutional block. avg means global average pooling applied output last convolutional block, thus output model 2D tensor. max means global max pooling applied. classes optional number classes classify images , specified include_top TRUE, weights argument specified. classifier_activation str callable. activation function use \"top\" layer. Ignored unless include_top=TRUE. Set classifier_activation=NULL return logits \"top\" layer. loading pretrained weights, classifier_activation can NULL \"softmax\". name name model (string).","code":""},{"path":"https://keras3.posit.co/dev/reference/application_resnet152_v2.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Instantiates the ResNet152V2 architecture. — application_resnet152_v2","text":"Model instance.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_resnet152_v2.html","id":"reference","dir":"Reference","previous_headings":"","what":"Reference","title":"Instantiates the ResNet152V2 architecture. — application_resnet152_v2","text":"Identity Mappings Deep Residual Networks (CVPR 2016) image classification use cases, see page detailed examples. transfer learning use cases, make sure read guide transfer learning & fine-tuning.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_resnet152_v2.html","id":"note","dir":"Reference","previous_headings":"","what":"Note","title":"Instantiates the ResNet152V2 architecture. — application_resnet152_v2","text":"Keras Application expects specific kind input preprocessing. ResNet, call application_preprocess_inputs() inputs passing model. application_preprocess_inputs() scale input pixels -1 1.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/application_resnet50.html","id":null,"dir":"Reference","previous_headings":"","what":"Instantiates the ResNet50 architecture. — application_resnet50","title":"Instantiates the ResNet50 architecture. — application_resnet50","text":"Instantiates ResNet50 architecture.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_resnet50.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Instantiates the ResNet50 architecture. — application_resnet50","text":"","code":"application_resnet50(   include_top = TRUE,   weights = \"imagenet\",   input_tensor = NULL,   input_shape = NULL,   pooling = NULL,   classes = 1000L,   classifier_activation = \"softmax\",   name = \"resnet50\" )"},{"path":"https://keras3.posit.co/dev/reference/application_resnet50.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Instantiates the ResNet50 architecture. — application_resnet50","text":"include_top whether include fully-connected layer top network. weights one NULL (random initialization), \"imagenet\" (pre-training ImageNet), path weights file loaded. input_tensor optional Keras tensor (.e. output keras_input()) use image input model. input_shape optional shape tuple, specified include_top FALSE (otherwise input shape (224, 224, 3) (\"channels_last\" data format) (3, 224, 224) (\"channels_first\" data format). exactly 3 inputs channels, width height smaller 32. E.g. (200, 200, 3) one valid value. pooling Optional pooling mode feature extraction include_top FALSE. NULL means output model 4D tensor output last convolutional block. avg means global average pooling applied output last convolutional block, thus output model 2D tensor. max means global max pooling applied. classes optional number classes classify images , specified include_top TRUE, weights argument specified. Defaults 1000. classifier_activation str callable. activation function use \"top\" layer. Ignored unless include_top=TRUE. Set classifier_activation=NULL return logits \"top\" layer. loading pretrained weights, classifier_activation can NULL \"softmax\". name name model (string).","code":""},{"path":"https://keras3.posit.co/dev/reference/application_resnet50.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Instantiates the ResNet50 architecture. — application_resnet50","text":"Model instance.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_resnet50.html","id":"reference","dir":"Reference","previous_headings":"","what":"Reference","title":"Instantiates the ResNet50 architecture. — application_resnet50","text":"Deep Residual Learning Image Recognition (CVPR 2015) image classification use cases, see page detailed examples. transfer learning use cases, make sure read guide transfer learning & fine-tuning.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_resnet50.html","id":"note","dir":"Reference","previous_headings":"","what":"Note","title":"Instantiates the ResNet50 architecture. — application_resnet50","text":"Keras Application expects specific kind input preprocessing. ResNet, call application_preprocess_inputs() inputs passing model. application_preprocess_inputs() convert input images RGB BGR, zero-center color channel respect ImageNet dataset, without scaling.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/application_resnet50_v2.html","id":null,"dir":"Reference","previous_headings":"","what":"Instantiates the ResNet50V2 architecture. — application_resnet50_v2","title":"Instantiates the ResNet50V2 architecture. — application_resnet50_v2","text":"Instantiates ResNet50V2 architecture.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_resnet50_v2.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Instantiates the ResNet50V2 architecture. — application_resnet50_v2","text":"","code":"application_resnet50_v2(   include_top = TRUE,   weights = \"imagenet\",   input_tensor = NULL,   input_shape = NULL,   pooling = NULL,   classes = 1000L,   classifier_activation = \"softmax\",   name = \"resnet50v2\" )"},{"path":"https://keras3.posit.co/dev/reference/application_resnet50_v2.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Instantiates the ResNet50V2 architecture. — application_resnet50_v2","text":"include_top whether include fully-connected layer top network. weights one NULL (random initialization), \"imagenet\" (pre-training ImageNet), path weights file loaded. input_tensor optional Keras tensor (.e. output keras_input()) use image input model. input_shape optional shape tuple, specified include_top FALSE (otherwise input shape (224, 224, 3) (\"channels_last\" data format) (3, 224, 224) (\"channels_first\" data format). exactly 3 inputs channels, width height smaller 32. E.g. (200, 200, 3) one valid value. pooling Optional pooling mode feature extraction include_top FALSE. NULL means output model 4D tensor output last convolutional block. avg means global average pooling applied output last convolutional block, thus output model 2D tensor. max means global max pooling applied. classes optional number classes classify images , specified include_top TRUE, weights argument specified. classifier_activation str callable. activation function use \"top\" layer. Ignored unless include_top=TRUE. Set classifier_activation=NULL return logits \"top\" layer. loading pretrained weights, classifier_activation can NULL \"softmax\". name name model (string).","code":""},{"path":"https://keras3.posit.co/dev/reference/application_resnet50_v2.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Instantiates the ResNet50V2 architecture. — application_resnet50_v2","text":"Model instance.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_resnet50_v2.html","id":"reference","dir":"Reference","previous_headings":"","what":"Reference","title":"Instantiates the ResNet50V2 architecture. — application_resnet50_v2","text":"Identity Mappings Deep Residual Networks (CVPR 2016) image classification use cases, see page detailed examples. transfer learning use cases, make sure read guide transfer learning & fine-tuning.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_resnet50_v2.html","id":"note","dir":"Reference","previous_headings":"","what":"Note","title":"Instantiates the ResNet50V2 architecture. — application_resnet50_v2","text":"Keras Application expects specific kind input preprocessing. ResNet, call application_preprocess_inputs() inputs passing model. application_preprocess_inputs() scale input pixels -1 1.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/application_vgg16.html","id":null,"dir":"Reference","previous_headings":"","what":"Instantiates the VGG16 model. — application_vgg16","title":"Instantiates the VGG16 model. — application_vgg16","text":"Instantiates VGG16 model.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_vgg16.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Instantiates the VGG16 model. — application_vgg16","text":"","code":"application_vgg16(   include_top = TRUE,   weights = \"imagenet\",   input_tensor = NULL,   input_shape = NULL,   pooling = NULL,   classes = 1000L,   classifier_activation = \"softmax\",   name = \"vgg16\" )"},{"path":"https://keras3.posit.co/dev/reference/application_vgg16.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Instantiates the VGG16 model. — application_vgg16","text":"include_top whether include 3 fully-connected layers top network. weights one NULL (random initialization), \"imagenet\" (pre-training ImageNet), path weights file loaded. input_tensor optional Keras tensor (.e. output keras_input()) use image input model. input_shape optional shape tuple, specified include_top FALSE (otherwise input shape (224, 224, 3) (channels_last data format) (3, 224, 224) (\"channels_first\" data format). exactly 3 input channels, width height smaller 32. E.g. (200, 200, 3) one valid value. pooling Optional pooling mode feature extraction include_top FALSE. NULL means output model 4D tensor output last convolutional block. avg means global average pooling applied output last convolutional block, thus output model 2D tensor. max means global max pooling applied. classes optional number classes classify images , specified include_top TRUE, weights argument specified. classifier_activation str callable. activation function use \"top\" layer. Ignored unless include_top=TRUE. Set classifier_activation=NULL return logits \"top\" layer.  loading pretrained weights, classifier_activation can NULL \"softmax\". name name model (string).","code":""},{"path":"https://keras3.posit.co/dev/reference/application_vgg16.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Instantiates the VGG16 model. — application_vgg16","text":"Model instance.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_vgg16.html","id":"reference","dir":"Reference","previous_headings":"","what":"Reference","title":"Instantiates the VGG16 model. — application_vgg16","text":"Deep Convolutional Networks Large-Scale Image Recognition (ICLR 2015) image classification use cases, see page detailed examples. transfer learning use cases, make sure read guide transfer learning & fine-tuning. default input size model 224x224.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_vgg16.html","id":"note","dir":"Reference","previous_headings":"","what":"Note","title":"Instantiates the VGG16 model. — application_vgg16","text":"Keras Application expects specific kind input preprocessing. VGG16, call application_preprocess_inputs() inputs passing model. application_preprocess_inputs() convert input images RGB BGR, zero-center color channel respect ImageNet dataset, without scaling.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/application_vgg19.html","id":null,"dir":"Reference","previous_headings":"","what":"Instantiates the VGG19 model. — application_vgg19","title":"Instantiates the VGG19 model. — application_vgg19","text":"Instantiates VGG19 model.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_vgg19.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Instantiates the VGG19 model. — application_vgg19","text":"","code":"application_vgg19(   include_top = TRUE,   weights = \"imagenet\",   input_tensor = NULL,   input_shape = NULL,   pooling = NULL,   classes = 1000L,   classifier_activation = \"softmax\",   name = \"vgg19\" )"},{"path":"https://keras3.posit.co/dev/reference/application_vgg19.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Instantiates the VGG19 model. — application_vgg19","text":"include_top whether include 3 fully-connected layers top network. weights one NULL (random initialization), \"imagenet\" (pre-training ImageNet), path weights file loaded. input_tensor optional Keras tensor (.e. output keras_input()) use image input model. input_shape optional shape tuple, specified include_top FALSE (otherwise input shape (224, 224, 3) (channels_last data format) (3, 224, 224) (\"channels_first\" data format). exactly 3 input channels, width height smaller 32. E.g. (200, 200, 3) one valid value. pooling Optional pooling mode feature extraction include_top FALSE. NULL means output model 4D tensor output last convolutional block. avg means global average pooling applied output last convolutional block, thus output model 2D tensor. max means global max pooling applied. classes optional number classes classify images , specified include_top TRUE, weights argument specified. classifier_activation str callable. activation function use \"top\" layer. Ignored unless include_top=TRUE. Set classifier_activation=NULL return logits \"top\" layer.  loading pretrained weights, classifier_activation can NULL \"softmax\". name name model (string).","code":""},{"path":"https://keras3.posit.co/dev/reference/application_vgg19.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Instantiates the VGG19 model. — application_vgg19","text":"model instance.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_vgg19.html","id":"reference","dir":"Reference","previous_headings":"","what":"Reference","title":"Instantiates the VGG19 model. — application_vgg19","text":"Deep Convolutional Networks Large-Scale Image Recognition (ICLR 2015) image classification use cases, see page detailed examples. transfer learning use cases, make sure read guide transfer learning & fine-tuning. default input size model 224x224.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_vgg19.html","id":"note","dir":"Reference","previous_headings":"","what":"Note","title":"Instantiates the VGG19 model. — application_vgg19","text":"Keras Application expects specific kind input preprocessing. VGG19, call application_preprocess_inputs() inputs passing model. application_preprocess_inputs() convert input images RGB BGR, zero-center color channel respect ImageNet dataset, without scaling.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/application_xception.html","id":null,"dir":"Reference","previous_headings":"","what":"Instantiates the Xception architecture. — application_xception","title":"Instantiates the Xception architecture. — application_xception","text":"Instantiates Xception architecture.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_xception.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Instantiates the Xception architecture. — application_xception","text":"","code":"application_xception(   include_top = TRUE,   weights = \"imagenet\",   input_tensor = NULL,   input_shape = NULL,   pooling = NULL,   classes = 1000L,   classifier_activation = \"softmax\",   name = \"xception\" )"},{"path":"https://keras3.posit.co/dev/reference/application_xception.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Instantiates the Xception architecture. — application_xception","text":"include_top whether include 3 fully-connected layers top network. weights one NULL (random initialization), \"imagenet\" (pre-training ImageNet), path weights file loaded. input_tensor optional Keras tensor (.e. output keras_input()) use image input model. input_shape optional shape tuple, specified include_top FALSE (otherwise input shape (299, 299, 3). exactly 3 inputs channels, width height smaller 71. E.g. (150, 150, 3) one valid value. pooling Optional pooling mode feature extraction include_top FALSE. NULL means output model 4D tensor output last convolutional block. avg means global average pooling applied output last convolutional block, thus output model 2D tensor. max means global max pooling applied. classes optional number classes classify images , specified include_top TRUE, weights argument specified. classifier_activation str callable. activation function use \"top\" layer. Ignored unless include_top=TRUE. Set classifier_activation=NULL return logits \"top\" layer.  loading pretrained weights, classifier_activation can NULL \"softmax\". name name model (string).","code":""},{"path":"https://keras3.posit.co/dev/reference/application_xception.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Instantiates the Xception architecture. — application_xception","text":"model instance.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_xception.html","id":"reference","dir":"Reference","previous_headings":"","what":"Reference","title":"Instantiates the Xception architecture. — application_xception","text":"Xception: Deep Learning Depthwise Separable Convolutions (CVPR 2017) image classification use cases, see page detailed examples. transfer learning use cases, make sure read guide transfer learning & fine-tuning. default input image size model 299x299.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_xception.html","id":"note","dir":"Reference","previous_headings":"","what":"Note","title":"Instantiates the Xception architecture. — application_xception","text":"Keras Application expects specific kind input preprocessing. Xception, call application_preprocess_inputs() inputs passing model. application_preprocess_inputs() scale input pixels -1 1.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/audio_dataset_from_directory.html","id":null,"dir":"Reference","previous_headings":"","what":"Generates a tf.data.Dataset from audio files in a directory. — audio_dataset_from_directory","title":"Generates a tf.data.Dataset from audio files in a directory. — audio_dataset_from_directory","text":"directory structure :   calling audio_dataset_from_directory(main_directory, labels = 'inferred') return tf.data.Dataset yields batches audio files subdirectories class_a class_b, together labels 0 1 (0 corresponding class_a 1 corresponding class_b). .wav files supported time.","code":"main_directory/ ...class_a/ ......a_audio_1.wav ......a_audio_2.wav ...class_b/ ......b_audio_1.wav ......b_audio_2.wav"},{"path":"https://keras3.posit.co/dev/reference/audio_dataset_from_directory.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Generates a tf.data.Dataset from audio files in a directory. — audio_dataset_from_directory","text":"","code":"audio_dataset_from_directory(   directory,   labels = \"inferred\",   label_mode = \"int\",   class_names = NULL,   batch_size = 32L,   sampling_rate = NULL,   output_sequence_length = NULL,   ragged = FALSE,   shuffle = TRUE,   seed = NULL,   validation_split = NULL,   subset = NULL,   follow_links = FALSE,   verbose = TRUE )"},{"path":"https://keras3.posit.co/dev/reference/audio_dataset_from_directory.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Generates a tf.data.Dataset from audio files in a directory. — audio_dataset_from_directory","text":"directory Directory data located. labels \"inferred\", contain subdirectories, containing audio files class. Otherwise, directory structure ignored. labels Either \"inferred\" (labels generated directory structure), NULL (labels), list/tuple integer labels size number audio files found directory. Labels sorted according alphanumeric order audio file paths (obtained via os.walk(directory) Python). label_mode String describing encoding labels. Options : \"int\": means labels encoded integers (e.g. sparse_categorical_crossentropy loss). \"categorical\" means labels encoded categorical vector (e.g. categorical_crossentropy loss) \"binary\" means labels (can 2) encoded float32 scalars values 0 1 (e.g. binary_crossentropy). NULL (labels). class_names valid \"labels\" \"inferred\". explicit list class names (must match names subdirectories). Used control order classes (otherwise alphanumerical order used). batch_size Size batches data. Default: 32. NULL, data batched (dataset yield individual samples). sampling_rate Audio sampling rate (samples per second). output_sequence_length Maximum length audio sequence. Audio files longer truncated output_sequence_length. set NULL, sequences batch padded length longest sequence batch. ragged Whether return Ragged dataset (sequence length). Defaults FALSE. shuffle Whether shuffle data. Defaults TRUE. set FALSE, sorts data alphanumeric order. seed Optional random seed shuffling transformations. validation_split Optional float 0 1, fraction data reserve validation. subset Subset data return. One \"training\", \"validation\" \"\". used validation_split set. follow_links Whether visits subdirectories pointed symlinks. Defaults FALSE. verbose Whether display number information classes number files found. Defaults TRUE.","code":""},{"path":"https://keras3.posit.co/dev/reference/audio_dataset_from_directory.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Generates a tf.data.Dataset from audio files in a directory. — audio_dataset_from_directory","text":"tf.data.Dataset object. label_mode NULL, yields string tensors shape (batch_size,), containing contents batch audio files. Otherwise, yields tuple (audio, labels), audio shape (batch_size, sequence_length, num_channels) labels follows format described . Rules regarding labels format: label_mode int, labels int32 tensor shape (batch_size,). label_mode binary, labels float32 tensor 1s 0s shape (batch_size, 1). label_mode categorical, labels float32 tensor shape (batch_size, num_classes), representing one-hot encoding class index.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/back-compat.html","id":null,"dir":"Reference","previous_headings":"","what":"Backward compatibility — application_nasnetlarge","title":"Backward compatibility — application_nasnetlarge","text":"Backward compatibility Backward compatibility","code":""},{"path":"https://keras3.posit.co/dev/reference/back-compat.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Backward compatibility — application_nasnetlarge","text":"","code":"application_nasnetlarge(...)  application_nasnetmobile(...)"},{"path":"https://keras3.posit.co/dev/reference/back-compat.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Backward compatibility — application_nasnetlarge","text":"... passed application_nasnet_large() application_nasnet_mobile().","code":""},{"path":"https://keras3.posit.co/dev/reference/bidirectional.html","id":null,"dir":"Reference","previous_headings":"","what":"layer_bidirectional — bidirectional","title":"layer_bidirectional — bidirectional","text":"bidirectional() alias layer_bidirectional(). See ?layer_bidirectional() full documentation.","code":""},{"path":"https://keras3.posit.co/dev/reference/bidirectional.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"layer_bidirectional — bidirectional","text":"","code":"bidirectional(   object,   layer,   merge_mode = \"concat\",   weights = NULL,   backward_layer = NULL,   ... )"},{"path":"https://keras3.posit.co/dev/reference/bidirectional.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"layer_bidirectional — bidirectional","text":"object Object compose layer . tensor, array, sequential model. layer RNN instance, layer_lstm() layer_gru(). also Layer() instance meets following criteria: sequence-processing layer (accepts 3D+ inputs). go_backwards, return_sequences return_state attribute (semantics RNN class). input_spec attribute. Implement serialization via get_config() from_config(). Note recommended way create new RNN layers write custom RNN cell use layer_rnn(), instead subclassing Layer() directly. return_sequences TRUE, output masked timestep zero regardless layer's original zero_output_for_mask value. merge_mode Mode outputs forward backward RNNs combined. One {\"sum\", \"mul\", \"concat\", \"ave\", NULL}. NULL, outputs combined, returned list. Defaults \"concat\". weights see description backward_layer Optional RNN, Layer() instance used handle backwards input processing. backward_layer provided, layer instance passed layer argument used generate backward layer automatically. Note provided backward_layer layer properties matching layer argument, particular values stateful, return_states, return_sequences, etc. addition, backward_layer layer different go_backwards argument values. ValueError raised requirements met. ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/bidirectional.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"layer_bidirectional — bidirectional","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/callback_backup_and_restore.html","id":null,"dir":"Reference","previous_headings":"","what":"Callback to back up and restore the training state. — callback_backup_and_restore","title":"Callback to back up and restore the training state. — callback_backup_and_restore","text":"callback_backup_and_restore() callback intended recover training interruption happened middle fit execution, backing training states temporary checkpoint file, end epoch. backup overwrites previously written checkpoint file, given time one checkpoint file backup/restoring purpose. training restarts completion, training state (includes model weights epoch number) restored recently saved state beginning new fit run. completion fit run, temporary checkpoint file deleted. Note user responsible bring jobs back interruption. callback important backup restore mechanism fault tolerance purpose, model restored previous checkpoint expected one used back . user changes arguments passed compile fit, checkpoint saved fault tolerance can become invalid.","code":""},{"path":"https://keras3.posit.co/dev/reference/callback_backup_and_restore.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Callback to back up and restore the training state. — callback_backup_and_restore","text":"","code":"callback_backup_and_restore(   backup_dir,   save_freq = \"epoch\",   double_checkpoint = FALSE,   delete_checkpoint = TRUE )"},{"path":"https://keras3.posit.co/dev/reference/callback_backup_and_restore.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Callback to back up and restore the training state. — callback_backup_and_restore","text":"backup_dir String, path directory store data needed restore model. directory reused elsewhere store files, e.g. backup_and_restore callback another training run, another callback (e.g. callback_model_checkpoint) training run. save_freq \"epoch\", integer, FALSE. set \"epoch\", callback saves checkpoint end epoch. set integer, callback saves checkpoint every save_freq batches. Set save_freq = FALSE using preemption checkpointing (.e. save_before_preemption = TRUE). double_checkpoint Boolean. enabled, BackupAndRestore callback save 2 last training states (current previous). interruption current state loaded due IO error (e.g. file corrupted) try restore previous one. behaviour consume twice space disk, increase fault tolerance. Defaults FALSE. delete_checkpoint Boolean. backup_and_restore callback works saving checkpoint back training state. delete_checkpoint = TRUE, checkpoint deleted training finished. Use FALSE like keep checkpoint future usage. Defaults TRUE.","code":""},{"path":"https://keras3.posit.co/dev/reference/callback_backup_and_restore.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Callback to back up and restore the training state. — callback_backup_and_restore","text":"Callback instance can passed fit.keras.src.models.model.Model().","code":""},{"path":"https://keras3.posit.co/dev/reference/callback_backup_and_restore.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Callback to back up and restore the training state. — callback_backup_and_restore","text":"","code":"callback_interrupting <- new_callback_class(   \"InterruptingCallback\",   on_epoch_begin = function(epoch, logs = NULL) {     if (epoch == 4) {       stop('Interrupting!')     }   } )  backup_dir <- tempfile() callback <- callback_backup_and_restore(backup_dir = backup_dir) model <- keras_model_sequential() %>%   layer_dense(10) model %>% compile(optimizer = optimizer_sgd(), loss = 'mse')  # ensure model is built (i.e., weights are initialized) for # callback_backup_and_restore() model(op_ones(c(5, 20))) |> invisible()  tryCatch({   model %>% fit(x = op_ones(c(5, 20)),                 y = op_zeros(5),                 epochs = 10, batch_size = 1,                 callbacks = list(callback, callback_interrupting()),                 verbose = 0) }, python.builtin.RuntimeError = function(e) message(\"Interrupted!\")) ## Interrupted! model$history$epoch ## [1] 0 1 2 # model$history %>% keras3:::to_keras_training_history() %>% as.data.frame() %>% print()  history <- model %>% fit(x = op_ones(c(5, 20)),                          y = op_zeros(5),                          epochs = 10, batch_size = 1,                          callbacks = list(callback),                          verbose = 0)  # Only 6 more epochs are run, since first training got interrupted at # zero-indexed epoch 4, second training will continue from 4 to 9. nrow(as.data.frame(history)) ## [1] 10"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/callback_csv_logger.html","id":null,"dir":"Reference","previous_headings":"","what":"Callback that streams epoch results to a CSV file. — callback_csv_logger","title":"Callback that streams epoch results to a CSV file. — callback_csv_logger","text":"Supports values can represented string, including 1D iterables atomic vectors.","code":""},{"path":"https://keras3.posit.co/dev/reference/callback_csv_logger.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Callback that streams epoch results to a CSV file. — callback_csv_logger","text":"","code":"callback_csv_logger(filename, separator = \",\", append = FALSE)"},{"path":"https://keras3.posit.co/dev/reference/callback_csv_logger.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Callback that streams epoch results to a CSV file. — callback_csv_logger","text":"filename Filename CSV file, e.g. 'run/log.csv'. separator String used separate elements CSV file. append Boolean. TRUE: append file exists (useful continuing training). FALSE: overwrite existing file.","code":""},{"path":"https://keras3.posit.co/dev/reference/callback_csv_logger.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Callback that streams epoch results to a CSV file. — callback_csv_logger","text":"Callback instance can passed fit.keras.src.models.model.Model().","code":""},{"path":"https://keras3.posit.co/dev/reference/callback_csv_logger.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Callback that streams epoch results to a CSV file. — callback_csv_logger","text":"","code":"csv_logger <- callback_csv_logger('training.log') model %>% fit(X_train, Y_train, callbacks = list(csv_logger))"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/callback_early_stopping.html","id":null,"dir":"Reference","previous_headings":"","what":"Stop training when a monitored metric has stopped improving. — callback_early_stopping","title":"Stop training when a monitored metric has stopped improving. — callback_early_stopping","text":"Assuming goal training minimize loss. , metric monitored 'loss', mode 'min'. model$fit() training loop check end every epoch whether loss longer decreasing, considering min_delta patience applicable. found longer decreasing, model$stop_training marked TRUE training terminates. quantity monitored needs available logs list. make , pass loss metrics model$compile().","code":""},{"path":"https://keras3.posit.co/dev/reference/callback_early_stopping.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Stop training when a monitored metric has stopped improving. — callback_early_stopping","text":"","code":"callback_early_stopping(   monitor = \"val_loss\",   min_delta = 0L,   patience = 0L,   verbose = 0L,   mode = \"auto\",   baseline = NULL,   restore_best_weights = FALSE,   start_from_epoch = 0L )"},{"path":"https://keras3.posit.co/dev/reference/callback_early_stopping.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Stop training when a monitored metric has stopped improving. — callback_early_stopping","text":"monitor Quantity monitored. Defaults \"val_loss\". min_delta Minimum change monitored quantity qualify improvement, .e. absolute change less min_delta, count improvement. Defaults 0. patience Number epochs improvement training stopped. Defaults 0. verbose Verbosity mode, 0 1. Mode 0 silent, mode 1 displays messages callback takes action. Defaults 0. mode One {\"auto\", \"min\", \"max\"}. min mode, training stop quantity monitored stopped decreasing; \"max\" mode stop quantity monitored stopped increasing; \"auto\" mode, direction automatically inferred name monitored quantity. Defaults \"auto\". baseline Baseline value monitored quantity. NULL, training stop model show improvement baseline. Defaults NULL. restore_best_weights Whether restore model weights epoch best value monitored quantity. FALSE, model weights obtained last step training used. epoch restored regardless performance relative baseline. epoch improves baseline, training run patience epochs restore weights best epoch set. Defaults FALSE. start_from_epoch Number epochs wait starting monitor improvement. allows warm-period improvement expected thus training stopped. Defaults 0.","code":""},{"path":"https://keras3.posit.co/dev/reference/callback_early_stopping.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Stop training when a monitored metric has stopped improving. — callback_early_stopping","text":"Callback instance can passed fit.keras.src.models.model.Model().","code":""},{"path":"https://keras3.posit.co/dev/reference/callback_early_stopping.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Stop training when a monitored metric has stopped improving. — callback_early_stopping","text":"","code":"callback <- callback_early_stopping(monitor = 'loss',                                    patience = 3) # This callback will stop the training when there is no improvement in # the loss for three consecutive epochs. model <- keras_model_sequential() %>%   layer_dense(10) model %>% compile(optimizer = optimizer_sgd(), loss = 'mse') history <- model %>% fit(x = op_ones(c(5, 20)),                          y = op_zeros(5),                          epochs = 10, batch_size = 1,                          callbacks = list(callback),                          verbose = 0) nrow(as.data.frame(history))  # Only 4 epochs are run. ## [1] 10"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/callback_lambda.html","id":null,"dir":"Reference","previous_headings":"","what":"Callback for creating simple, custom callbacks on-the-fly. — callback_lambda","title":"Callback for creating simple, custom callbacks on-the-fly. — callback_lambda","text":"callback constructed anonymous functions called appropriate time (Model.{fit | evaluate | predict}). Note callbacks expects positional arguments, : on_epoch_begin on_epoch_end expect two positional arguments: epoch, logs on_train_begin on_train_end expect one positional argument: logs on_train_batch_begin on_train_batch_end expect positional argument batch named argument logs See Callback class definition full list functions expected arguments.","code":""},{"path":"https://keras3.posit.co/dev/reference/callback_lambda.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Callback for creating simple, custom callbacks on-the-fly. — callback_lambda","text":"","code":"callback_lambda(   on_epoch_begin = NULL,   on_epoch_end = NULL,   on_train_begin = NULL,   on_train_end = NULL,   on_train_batch_begin = NULL,   on_train_batch_end = NULL,   ... )"},{"path":"https://keras3.posit.co/dev/reference/callback_lambda.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Callback for creating simple, custom callbacks on-the-fly. — callback_lambda","text":"on_epoch_begin called beginning every epoch. on_epoch_end called end every epoch. on_train_begin called beginning model training. on_train_end called end model training. on_train_batch_begin called beginning every train batch. on_train_batch_end called end every train batch. ... function Callback() want override passing function_name = function. example, callback_lambda(.., on_train_end = train_end_fn). custom function needs arguments ones defined Callback().","code":""},{"path":"https://keras3.posit.co/dev/reference/callback_lambda.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Callback for creating simple, custom callbacks on-the-fly. — callback_lambda","text":"Callback instance can passed fit.keras.src.models.model.Model().","code":""},{"path":"https://keras3.posit.co/dev/reference/callback_lambda.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Callback for creating simple, custom callbacks on-the-fly. — callback_lambda","text":"","code":"# Print the batch number at the beginning of every batch. batch_print_callback <- callback_lambda(   on_train_batch_begin = function(batch, logs) {     print(batch)   } )  # Stream the epoch loss to a file in new-line delimited JSON format # (one valid JSON object per line) json_log <- file('loss_log.json', open = 'wt') json_logging_callback <- callback_lambda(   on_epoch_end = function(epoch, logs) {     jsonlite::write_json(       list(epoch = epoch, loss = logs$loss),       json_log,       append = TRUE     )   },   on_train_end = function(logs) {     close(json_log)   } )  # Terminate some processes after having finished model training. processes <- ... cleanup_callback <- callback_lambda(   on_train_end = function(logs) {     for (p in processes) {       if (is_alive(p)) {         terminate(p)       }     }   } )  model %>% fit(   ...,   callbacks = list(     batch_print_callback,     json_logging_callback,     cleanup_callback   ) )"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/callback_learning_rate_scheduler.html","id":null,"dir":"Reference","previous_headings":"","what":"Learning rate scheduler. — callback_learning_rate_scheduler","title":"Learning rate scheduler. — callback_learning_rate_scheduler","text":"beginning every epoch, callback gets updated learning rate value schedule function provided, current epoch current learning rate, applies updated learning rate optimizer.","code":""},{"path":"https://keras3.posit.co/dev/reference/callback_learning_rate_scheduler.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Learning rate scheduler. — callback_learning_rate_scheduler","text":"","code":"callback_learning_rate_scheduler(schedule, verbose = 0L)"},{"path":"https://keras3.posit.co/dev/reference/callback_learning_rate_scheduler.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Learning rate scheduler. — callback_learning_rate_scheduler","text":"schedule function takes epoch index (integer, indexed 0) current learning rate (float) inputs returns new learning rate output (float). verbose Integer. 0: quiet, 1: log update messages.","code":""},{"path":"https://keras3.posit.co/dev/reference/callback_learning_rate_scheduler.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Learning rate scheduler. — callback_learning_rate_scheduler","text":"Callback instance can passed fit.keras.src.models.model.Model().","code":""},{"path":"https://keras3.posit.co/dev/reference/callback_learning_rate_scheduler.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Learning rate scheduler. — callback_learning_rate_scheduler","text":"","code":"# This function keeps the initial learning rate steady for the first ten epochs # and decreases it exponentially after that. scheduler <- function(epoch, lr) {   if (epoch < 10)     return(lr)   else     return(lr * exp(-0.1)) }  model <- keras_model_sequential() |> layer_dense(units = 10) model |> compile(optimizer = optimizer_sgd(), loss = 'mse') model$optimizer$learning_rate |> as.array() |> round(5) ## [1] 0.01 callback <- callback_learning_rate_scheduler(schedule = scheduler) history <- model |> fit(x = array(runif(100), c(5, 20)),                         y = array(0, c(5, 1)),                         epochs = 15, callbacks = list(callback), verbose = 0) model$optimizer$learning_rate |> as.array() |> round(5) ## [1] 0.00607"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/callback_model_checkpoint.html","id":null,"dir":"Reference","previous_headings":"","what":"Callback to save the Keras model or model weights at some frequency. — callback_model_checkpoint","title":"Callback to save the Keras model or model weights at some frequency. — callback_model_checkpoint","text":"callback_model_checkpoint() used conjunction training using model |> fit() save model weights (checkpoint file) interval, model weights can loaded later continue training state saved. options callback provides include: Whether keep model achieved \"best performance\" far, whether save model end every epoch regardless performance. Definition \"best\"; quantity monitor whether maximized minimized. frequency save . Currently, callback supports saving end every epoch, fixed number training batches. Whether weights saved, whole model saved.","code":""},{"path":"https://keras3.posit.co/dev/reference/callback_model_checkpoint.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Callback to save the Keras model or model weights at some frequency. — callback_model_checkpoint","text":"","code":"callback_model_checkpoint(   filepath,   monitor = \"val_loss\",   verbose = 0L,   save_best_only = FALSE,   save_weights_only = FALSE,   mode = \"auto\",   save_freq = \"epoch\",   initial_value_threshold = NULL )"},{"path":"https://keras3.posit.co/dev/reference/callback_model_checkpoint.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Callback to save the Keras model or model weights at some frequency. — callback_model_checkpoint","text":"filepath string, path save model file. filepath can contain named formatting options, filled value epoch keys logs (passed on_epoch_end). filepath name needs end \".weights.h5\" save_weights_only = TRUE end \".keras\" \".h5\" checkpoint saving whole model (default). example: filepath \"{epoch:02d}-{val_loss:.2f}.keras\" \"{epoch:02d}-{val_loss:.2f}.weights.h5\", model checkpoints saved epoch number validation loss filename. directory filepath reused callbacks avoid conflicts. monitor metric name monitor. Typically metrics set model |> compile() method. Note: Prefix name \"val_\" monitor validation metrics. Use \"loss\" \"val_loss\" monitor model's total loss. specify metrics strings, like \"accuracy\", pass string (without \"val_\" prefix). pass Metric objects (created one metric_*()), monitor set metric$name. sure metric names can check contents history$metrics list returned history <- model |> fit() Multi-output models set additional prefixes metric names. verbose Verbosity mode, 0 1. Mode 0 silent, mode 1 displays messages callback takes action. save_best_only save_best_only = TRUE, saves model considered \"best\" latest best model according quantity monitored overwritten. filepath contain formatting options like {epoch} filepath overwritten new better model. save_weights_only TRUE, model's weights saved (model |> save_model_weights(filepath)), else full model saved (model |> save_model(filepath)). mode one {\"auto\", \"min\", \"max\"}. save_best_only = TRUE, decision overwrite current save file made based either maximization minimization monitored quantity. val_acc, \"max\", val_loss \"min\", etc. \"auto\" mode, direction automatically inferred name monitored quantity. save_freq \"epoch\" integer. using \"epoch\", callback saves model epoch. using integer, callback saves model end many batches. Model compiled steps_per_execution = N, saving criteria checked every Nth batch. Note saving aligned epochs, monitored metric may potentially less reliable (reflect little 1 batch, since metrics get reset every epoch). Defaults \"epoch\". initial_value_threshold Floating point initial \"best\" value metric monitored. applies save_best_value = TRUE. overwrites model weights already saved performance current model better value.","code":""},{"path":"https://keras3.posit.co/dev/reference/callback_model_checkpoint.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Callback to save the Keras model or model weights at some frequency. — callback_model_checkpoint","text":"Callback instance can passed fit.keras.src.models.model.Model().","code":""},{"path":"https://keras3.posit.co/dev/reference/callback_model_checkpoint.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Callback to save the Keras model or model weights at some frequency. — callback_model_checkpoint","text":"","code":"model <- keras_model_sequential(input_shape = c(10)) |>   layer_dense(1, activation = \"sigmoid\") |>   compile(loss = \"binary_crossentropy\", optimizer = \"adam\",           metrics = c('accuracy'))  EPOCHS <- 10 checkpoint_filepath <- tempfile('checkpoint-model-', fileext = \".keras\") model_checkpoint_callback <- callback_model_checkpoint(   filepath = checkpoint_filepath,   monitor = 'val_accuracy',   mode = 'max',   save_best_only = TRUE )  # Model is saved at the end of every epoch, if it's the best seen so far. model |> fit(x = random_uniform(c(2, 10)), y = op_ones(2, 1),              epochs = EPOCHS, validation_split = .5, verbose = 0,              callbacks = list(model_checkpoint_callback))  # The model (that are considered the best) can be loaded as - load_model(checkpoint_filepath) ## Model: \"sequential\" ## +---------------------------------+------------------------+---------------+ ## | Layer (type)                    | Output Shape           |       Param # | ## +=================================+========================+===============+ ## | dense (Dense)                   | (None, 1)              |            11 | ## +---------------------------------+------------------------+---------------+ ##  Total params: 35 (144.00 B) ##  Trainable params: 11 (44.00 B) ##  Non-trainable params: 0 (0.00 B) ##  Optimizer params: 24 (100.00 B) # Alternatively, one could checkpoint just the model weights as - checkpoint_filepath <- tempfile('checkpoint-', fileext = \".weights.h5\") model_checkpoint_callback <- callback_model_checkpoint(   filepath = checkpoint_filepath,   save_weights_only = TRUE,   monitor = 'val_accuracy',   mode = 'max',   save_best_only = TRUE )  # Model weights are saved at the end of every epoch, if it's the best seen # so far. # same as above model |> fit(x = random_uniform(c(2, 10)), y = op_ones(2, 1),              epochs = EPOCHS, validation_split = .5, verbose = 0,              callbacks = list(model_checkpoint_callback))  # The model weights (that are considered the best) can be loaded model |> load_model_weights(checkpoint_filepath)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/callback_reduce_lr_on_plateau.html","id":null,"dir":"Reference","previous_headings":"","what":"Reduce learning rate when a metric has stopped improving. — callback_reduce_lr_on_plateau","title":"Reduce learning rate when a metric has stopped improving. — callback_reduce_lr_on_plateau","text":"Models often benefit reducing learning rate factor 2-10 learning stagnates. callback monitors quantity improvement seen 'patience' number epochs, learning rate reduced.","code":""},{"path":"https://keras3.posit.co/dev/reference/callback_reduce_lr_on_plateau.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Reduce learning rate when a metric has stopped improving. — callback_reduce_lr_on_plateau","text":"","code":"callback_reduce_lr_on_plateau(   monitor = \"val_loss\",   factor = 0.1,   patience = 10L,   verbose = 0L,   mode = \"auto\",   min_delta = 1e-04,   cooldown = 0L,   min_lr = 0,   ... )"},{"path":"https://keras3.posit.co/dev/reference/callback_reduce_lr_on_plateau.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Reduce learning rate when a metric has stopped improving. — callback_reduce_lr_on_plateau","text":"monitor String. Quantity monitored. factor Float. Factor learning rate reduced. new_lr = lr * factor. patience Integer. Number epochs improvement learning rate reduced. verbose Integer. 0: quiet, 1: update messages. mode String. One {'auto', 'min', 'max'}. 'min' mode, learning rate reduced quantity monitored stopped decreasing; 'max' mode reduced quantity monitored stopped increasing; 'auto' mode, direction automatically inferred name monitored quantity. min_delta Float. Threshold measuring new optimum, focus significant changes. cooldown Integer. Number epochs wait resuming normal operation learning rate reduced. min_lr Float. Lower bound learning rate. ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/callback_reduce_lr_on_plateau.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Reduce learning rate when a metric has stopped improving. — callback_reduce_lr_on_plateau","text":"Callback instance can passed fit.keras.src.models.model.Model().","code":""},{"path":"https://keras3.posit.co/dev/reference/callback_reduce_lr_on_plateau.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Reduce learning rate when a metric has stopped improving. — callback_reduce_lr_on_plateau","text":"","code":"reduce_lr <- callback_reduce_lr_on_plateau(monitor = 'val_loss', factor = 0.2,                                            patience = 5, min_lr = 0.001) model %>% fit(x_train, y_train, callbacks = list(reduce_lr))"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/callback_remote_monitor.html","id":null,"dir":"Reference","previous_headings":"","what":"Callback used to stream events to a server. — callback_remote_monitor","title":"Callback used to stream events to a server. — callback_remote_monitor","text":"Requires requests library. Events sent root + '/publish/epoch/end/' default. Calls HTTP POST, data argument JSON-encoded named list event data. send_as_json = TRUE, content type request \"application/json\". Otherwise serialized JSON sent within form.","code":""},{"path":"https://keras3.posit.co/dev/reference/callback_remote_monitor.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Callback used to stream events to a server. — callback_remote_monitor","text":"","code":"callback_remote_monitor(   root = \"http://localhost:9000\",   path = \"/publish/epoch/end/\",   field = \"data\",   headers = NULL,   send_as_json = FALSE )"},{"path":"https://keras3.posit.co/dev/reference/callback_remote_monitor.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Callback used to stream events to a server. — callback_remote_monitor","text":"root String; root url target server. path String; path relative root events sent. field String; JSON field data stored. field used payload sent within form (.e. send_as_json = FALSE). headers Named list; optional custom HTTP headers. send_as_json Boolean; whether request sent \"application/json\".","code":""},{"path":"https://keras3.posit.co/dev/reference/callback_remote_monitor.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Callback used to stream events to a server. — callback_remote_monitor","text":"Callback instance can passed fit.keras.src.models.model.Model().","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/callback_swap_ema_weights.html","id":null,"dir":"Reference","previous_headings":"","what":"Swaps model weights and EMA weights before and after evaluation. — callback_swap_ema_weights","title":"Swaps model weights and EMA weights before and after evaluation. — callback_swap_ema_weights","text":"callbacks replaces model's weight values values optimizer's EMA weights (exponential moving average past model weights values, implementing \"Polyak averaging\") model evaluation, restores previous weights evaluation. SwapEMAWeights callback used conjunction optimizer sets use_ema = TRUE. Note weights swapped -place order save memory. behavior undefined modify EMA weights model weights callbacks.","code":""},{"path":"https://keras3.posit.co/dev/reference/callback_swap_ema_weights.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Swaps model weights and EMA weights before and after evaluation. — callback_swap_ema_weights","text":"","code":"callback_swap_ema_weights(swap_on_epoch = FALSE)"},{"path":"https://keras3.posit.co/dev/reference/callback_swap_ema_weights.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Swaps model weights and EMA weights before and after evaluation. — callback_swap_ema_weights","text":"swap_on_epoch Whether perform swapping on_epoch_begin() on_epoch_end(). useful want use EMA weights callbacks callback_model_checkpoint(). Defaults FALSE.","code":""},{"path":"https://keras3.posit.co/dev/reference/callback_swap_ema_weights.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Swaps model weights and EMA weights before and after evaluation. — callback_swap_ema_weights","text":"Callback instance can passed fit.keras.src.models.model.Model().","code":""},{"path":"https://keras3.posit.co/dev/reference/callback_swap_ema_weights.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Swaps model weights and EMA weights before and after evaluation. — callback_swap_ema_weights","text":"","code":"# Remember to set `use_ema=TRUE` in the optimizer optimizer <- optimizer_sgd(use_ema = TRUE) model |> compile(optimizer = optimizer, loss = ..., metrics = ...)  # Metrics will be computed with EMA weights model |> fit(X_train, Y_train,              callbacks = c(callback_swap_ema_weights()))  # If you want to save model checkpoint with EMA weights, you can set # `swap_on_epoch=TRUE` and place ModelCheckpoint after SwapEMAWeights. model |> fit(   X_train, Y_train,   callbacks = c(     callback_swap_ema_weights(swap_on_epoch = TRUE),     callback_model_checkpoint(...)   ) )"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/callback_tensorboard.html","id":null,"dir":"Reference","previous_headings":"","what":"Enable visualizations for TensorBoard. — callback_tensorboard","title":"Enable visualizations for TensorBoard. — callback_tensorboard","text":"TensorBoard visualization tool provided TensorFlow. TensorFlow installation required use callback. callback logs events TensorBoard, including: Metrics summary plots Training graph visualization Weight histograms Sampled profiling used model |> evaluate() regular validation addition epoch summaries, summary records evaluation metrics vs model$optimizer$iterations written. metric names prepended evaluation, model$optimizer$iterations step visualized TensorBoard. installed TensorFlow pip reticulate::py_install(), able launch TensorBoard command line:   R tensorflow::tensorboard(). can find information TensorBoard .","code":"tensorboard --logdir=path_to_your_logs"},{"path":"https://keras3.posit.co/dev/reference/callback_tensorboard.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Enable visualizations for TensorBoard. — callback_tensorboard","text":"","code":"callback_tensorboard(   log_dir = \"logs\",   histogram_freq = 0L,   write_graph = TRUE,   write_images = FALSE,   write_steps_per_second = FALSE,   update_freq = \"epoch\",   profile_batch = 0L,   embeddings_freq = 0L,   embeddings_metadata = NULL )"},{"path":"https://keras3.posit.co/dev/reference/callback_tensorboard.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Enable visualizations for TensorBoard. — callback_tensorboard","text":"log_dir path directory save log files parsed TensorBoard. e.g., log_dir = file.path(working_dir, 'logs'). directory reused callbacks. histogram_freq frequency (epochs) compute weight histograms layers model. set 0, histograms computed. Validation data (split) must specified histogram visualizations. write_graph (supported time) Whether visualize graph TensorBoard. Note log file can become quite large write_graph set TRUE. write_images whether write model weights visualize image TensorBoard. write_steps_per_second whether log training steps per second TensorBoard. supports epoch batch frequency logging. update_freq \"batch\" \"epoch\" integer. using \"epoch\", writes losses metrics TensorBoard every epoch. using integer, say 1000, metrics losses (including custom ones added Model.compile) logged TensorBoard every 1000 batches. \"batch\" synonym 1, meaning written every batch. Note however writing frequently TensorBoard can slow training, especially used distribution strategies incur additional synchronization overhead. Batch-level summary writing also available via train_step override. Please see TensorBoard Scalars tutorial details. profile_batch Profile batch(es) sample compute characteristics. profile_batch must non-negative integer tuple integers. pair positive integers signify range batches profile. default, profiling disabled. embeddings_freq frequency (epochs) embedding layers visualized. set 0, embeddings visualized. embeddings_metadata Named list maps embedding layer names filename file save metadata embedding layer. case metadata file used embedding layers, single filename can passed.","code":""},{"path":"https://keras3.posit.co/dev/reference/callback_tensorboard.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Enable visualizations for TensorBoard. — callback_tensorboard","text":"Callback instance can passed fit.keras.src.models.model.Model().","code":""},{"path":"https://keras3.posit.co/dev/reference/callback_tensorboard.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Enable visualizations for TensorBoard. — callback_tensorboard","text":"Custom batch-level summaries subclassed Model:   Custom batch-level summaries Functional API Model:   Profiling:","code":"tensorboard_callback <- callback_tensorboard(log_dir = \"./logs\") model %>% fit(x_train, y_train, epochs = 2, callbacks = list(tensorboard_callback)) # Then run the tensorboard command to view the visualizations. MyModel <- new_model_class(\"MyModel\",   initialize = function() {     self$dense <- layer_dense(units = 10)   },   call = function(x) {     outputs <- x |> self$dense()     tf$summary$histogram('outputs', outputs)     outputs   } )  model <- MyModel() model |> compile(optimizer = 'sgd', loss = 'mse')  # Make sure to set `update_freq = N` to log a batch-level summary every N # batches. In addition to any `tf$summary` contained in `model$call()`, # metrics added in `model |>compile` will be logged every N batches. tb_callback <- callback_tensorboard(log_dir = './logs', update_freq = 1) model |> fit(x_train, y_train, callbacks = list(tb_callback)) my_summary <- function(x) {   tf$summary$histogram('x', x)   x }  inputs <- layer_input(10) outputs <- inputs |>   layer_dense(10) |>   layer_lambda(my_summary)  model <- keras_model(inputs, outputs) model |> compile(optimizer = 'sgd', loss = 'mse')  # Make sure to set `update_freq = N` to log a batch-level summary every N # batches. In addition to any `tf.summary` contained in `Model.call`, # metrics added in `Model.compile` will be logged every N batches. tb_callback <- callback_tensorboard(log_dir = './logs', update_freq = 1) model |> fit(x_train, y_train, callbacks = list(tb_callback)) # Profile a single batch, e.g. the 5th batch. tensorboard_callback <- callback_tensorboard(   log_dir = './logs', profile_batch = 5) model |> fit(x_train, y_train, epochs = 2,              callbacks = list(tensorboard_callback))  # Profile a range of batches, e.g. from 10 to 20. tensorboard_callback <- callback_tensorboard(   log_dir = './logs', profile_batch = c(10, 20)) model |> fit(x_train, y_train, epochs = 2,              callbacks = list(tensorboard_callback))"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/callback_terminate_on_nan.html","id":null,"dir":"Reference","previous_headings":"","what":"Callback that terminates training when a NaN loss is encountered. — callback_terminate_on_nan","title":"Callback that terminates training when a NaN loss is encountered. — callback_terminate_on_nan","text":"Callback terminates training NaN loss encountered.","code":""},{"path":"https://keras3.posit.co/dev/reference/callback_terminate_on_nan.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Callback that terminates training when a NaN loss is encountered. — callback_terminate_on_nan","text":"","code":"callback_terminate_on_nan()"},{"path":"https://keras3.posit.co/dev/reference/callback_terminate_on_nan.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Callback that terminates training when a NaN loss is encountered. — callback_terminate_on_nan","text":"Callback instance can passed fit.keras.src.models.model.Model().","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/clear_session.html","id":null,"dir":"Reference","previous_headings":"","what":"Resets all state generated by Keras. — clear_session","title":"Resets all state generated by Keras. — clear_session","text":"Keras manages global state, uses implement Functional model-building API uniquify autogenerated layer names. creating many models loop, global state consume increasing amount memory time, may want clear . Calling clear_session() releases global state: helps avoid clutter old models layers, especially memory limited. Example 1: calling clear_session() creating models loop   Example 2: resetting layer name generation counter","code":"for (i in 1:100) {   # Without `clear_session()`, each iteration of this loop will   # slightly increase the size of the global state managed by Keras   model <- keras_model_sequential()   for (j in 1:10) {     model <- model |> layer_dense(units = 10)   } }  for (i in 1:100) {   # With `clear_session()` called at the beginning,   # Keras starts with a blank state at each iteration   # and memory consumption is constant over time.   clear_session()   model <- keras_model_sequential()   for (j in 1:10) {     model <- model |> layer_dense(units = 10)   } } layers <- lapply(1:10, \\(i) layer_dense(units = 10))  new_layer <- layer_dense(units = 10) print(new_layer$name) ## [1] \"dense_10\" clear_session() new_layer <- layer_dense(units = 10) print(new_layer$name) ## [1] \"dense\""},{"path":"https://keras3.posit.co/dev/reference/clear_session.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Resets all state generated by Keras. — clear_session","text":"","code":"clear_session(free_memory = TRUE)"},{"path":"https://keras3.posit.co/dev/reference/clear_session.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Resets all state generated by Keras. — clear_session","text":"free_memory Whether call Python garbage collection. usually good practice call make sure memory used deleted objects immediately freed. However, may take seconds execute, using clear_session() short loop, may want skip .","code":""},{"path":"https://keras3.posit.co/dev/reference/clear_session.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Resets all state generated by Keras. — clear_session","text":"NULL, invisibly, called side effects.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/clone_model.html","id":null,"dir":"Reference","previous_headings":"","what":"Clone a Functional or Sequential Model instance. — clone_model","title":"Clone a Functional or Sequential Model instance. — clone_model","text":"Model cloning similar calling model new inputs, except creates new layers (thus new weights) instead sharing weights existing layers. Note clone_model() preserve uniqueness shared objects within model (e.g. single variable attached two distinct layers restored two separate variables).","code":""},{"path":"https://keras3.posit.co/dev/reference/clone_model.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Clone a Functional or Sequential Model instance. — clone_model","text":"","code":"clone_model(   model,   input_tensors = NULL,   clone_function = NULL,   call_function = NULL,   recursive = FALSE,   ... )"},{"path":"https://keras3.posit.co/dev/reference/clone_model.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Clone a Functional or Sequential Model instance. — clone_model","text":"model Instance Model (Functional model Sequential model). input_tensors Optional list input tensors build model upon. provided, new keras_input() objects created. clone_function Callable signature function(layer) used clone layer target model (except Input instances). takes argument layer instance cloned, returns corresponding layer instance used model copy. unspecified, callable defaults following serialization/deserialization function: function(layer) layer$`__class__`$from_config(layer$get_config()). passing custom callable, can customize copy model, e.g. wrapping certain layers interest (might want replace LSTM instances equivalent Bidirectional(LSTM(...)) instances, example). Defaults NULL. call_function Callable signature function(layer, ...) used call cloned layer set inputs. takes layer instance, call arguments, returns call outputs. unspecified, callable defaults regular call() method: function(layer, ...) .call(layer, list(...)). passing custom callable, can insert new layers given layer. recursive Note, argument can used Functional models. Boolean. Whether recursively clone Sequential Functional models encountered original Sequential/Functional model. FALSE, inner models cloned calling clone_function(). TRUE, inner models cloned calling clone_model() clone_function, call_function, recursive arguments. Note case, call_function propagated Sequential model (since applicable Sequential models). ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/clone_model.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Clone a Functional or Sequential Model instance. — clone_model","text":"instance Model reproducing behavior original model, top new inputs tensors, using newly instantiated weights. cloned model may behave differently original model custom clone_function call_function modifies layer layer call.","code":""},{"path":"https://keras3.posit.co/dev/reference/clone_model.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Clone a Functional or Sequential Model instance. — clone_model","text":"Using clone_function make model deterministic setting random seed everywhere:   Using call_function add Dropout layer Dense layer (without recreating new layers):     Note subclassed models cloned default, since internal layer structure known. achieve equivalent functionality clone_model case subclassed model, simply make sure model class implements get_config() (optionally from_config()), call:   case subclassed model, using custom clone_function.","code":"# Create a test Sequential model. model <- keras_model_sequential(input_shape = c(728)) |>   layer_dense(32, activation = 'relu') |>   layer_dense(1, activation = 'sigmoid')  # Create a copy of the test model (with freshly initialized weights). new_model <- clone_model(model) clone_function <- function(layer) {   config <- layer$get_config()   if (\"seed\" %in% names(config))     config$seed <- 1337L   layer$`__class__`$from_config(config) }  new_model <- clone_model(model, clone_function = clone_function) call_function <- function(layer, ...) {   out <- layer(...)   if (inherits(layer, keras$layers$Dense))     out <- out |> layer_dropout(0.5)   out }  inputs <- keras_input(c(728)) outputs <- inputs |>   layer_dense(32, activation = 'relu') |>   layer_dense(1, activation = 'sigmoid') model <- keras_model(inputs, outputs)  new_model <- clone_model(   model,   clone_function = function(x) x, # Reuse the same layers.   call_function = call_function, ) new_model ## Model: \"functional_4\" ## +-----------------------------------+--------------------------+---------------+ ## | Layer (type)                      | Output Shape             |       Param # | ## +===================================+==========================+===============+ ## | keras_tensor_8 (InputLayer)       | (None, 728)              |             0 | ## +-----------------------------------+--------------------------+---------------+ ## | dense_2 (Dense)                   | (None, 32)               |        23,328 | ## +-----------------------------------+--------------------------+---------------+ ## | dropout (Dropout)                 | (None, 32)               |             0 | ## +-----------------------------------+--------------------------+---------------+ ## | dense_3 (Dense)                   | (None, 1)                |            33 | ## +-----------------------------------+--------------------------+---------------+ ## | dropout_1 (Dropout)               | (None, 1)                |             0 | ## +-----------------------------------+--------------------------+---------------+ ##  Total params: 23,361 (91.25 KB) ##  Trainable params: 23,361 (91.25 KB) ##  Non-trainable params: 0 (0.00 B) new_model <- model$`__class__`$from_config(model$get_config())"},{"path":"https://keras3.posit.co/dev/reference/compile.keras.src.models.model.Model.html","id":null,"dir":"Reference","previous_headings":"","what":"Configure a model for training. — compile.keras.src.models.model.Model","title":"Configure a model for training. — compile.keras.src.models.model.Model","text":"Configure model training.","code":""},{"path":"https://keras3.posit.co/dev/reference/compile.keras.src.models.model.Model.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Configure a model for training. — compile.keras.src.models.model.Model","text":"","code":"# S3 method for class 'keras.src.models.model.Model' compile(   object,   optimizer = \"rmsprop\",   loss = NULL,   metrics = NULL,   ...,   loss_weights = NULL,   weighted_metrics = NULL,   run_eagerly = FALSE,   steps_per_execution = 1L,   jit_compile = \"auto\",   auto_scale_loss = TRUE )"},{"path":"https://keras3.posit.co/dev/reference/compile.keras.src.models.model.Model.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Configure a model for training. — compile.keras.src.models.model.Model","text":"object Keras model object optimizer String (name optimizer) optimizer instance. See optimizer_* family. loss Loss function. May : string (name builtin loss function), custom function, Loss instance (returned loss_* family functions). loss function callable signature loss = fn(y_true, y_pred), y_true ground truth values, y_pred model's predictions. y_true shape (batch_size, d1, .. dN) (except case sparse loss functions sparse categorical crossentropy expects integer arrays shape (batch_size, d1, .. dN-1)). y_pred shape (batch_size, d1, .. dN). loss function return float tensor. metrics List metrics evaluated model training testing. can : string (name built-function), function, optionally \"name\" attribute Metric() instance. See metric_* family functions. Typically use metrics = c('accuracy'). function callable signature result = fn(y_true, y_pred). specify different metrics different outputs multi-output model, also pass named list, metrics = list(= 'accuracy', b = c('accuracy', 'mse')). can also pass list specify metric list metrics output, metrics = list(c('accuracy'), c('accuracy', 'mse')) metrics = list('accuracy', c('accuracy', 'mse')). pass strings 'accuracy' 'acc', convert one metric_binary_accuracy(), metric_categorical_accuracy(), metric_sparse_categorical_accuracy() based shapes targets model output. similar conversion done strings \"crossentropy\" \"ce\" well. metrics passed evaluated without sample weighting; like sample weighting apply, can specify metrics via weighted_metrics argument instead. providing anonymous R function, can customize printed name training assigning attr(<fn>, \"name\") <- \"my_custom_metric_name\", calling custom_metric(\"my_custom_metric_name\", <fn>) ... Additional arguments passed compile() model method. loss_weights Optional list (named unnamed) specifying scalar coefficients (R numerics) weight loss contributions different model outputs. loss value minimized model weighted sum individual losses, weighted loss_weights coefficients.  unnamed list, expected 1:1 mapping model's outputs. named list, expected map output names (strings) scalar coefficients. weighted_metrics List metrics evaluated weighted sample_weight class_weight training testing. run_eagerly Bool. TRUE, model's forward pass never compiled. recommended leave FALSE training (best performance), set TRUE debugging. steps_per_execution Int. number batches run single compiled function call. Running multiple batches inside single compiled function call can greatly improve performance TPUs small models large R/Python overhead. , one full epoch run execution. number larger size epoch passed, execution truncated size epoch. Note steps_per_execution set N, Callback$on_batch_begin Callback$on_batch_end methods called every N batches (.e. /compiled function execution). supported PyTorch backend. jit_compile Bool \"auto\". Whether use XLA compilation compiling model. jax tensorflow backends, jit_compile=\"auto\" enables XLA compilation model supports , disabled otherwise. torch backend, \"auto\" default eager execution jit_compile=True run torch.compile \"inductor\" backend. auto_scale_loss Bool. TRUE model dtype policy \"mixed_float16\", passed optimizer automatically wrapped LossScaleOptimizer, dynamically scale loss prevent underflow.","code":""},{"path":"https://keras3.posit.co/dev/reference/compile.keras.src.models.model.Model.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Configure a model for training. — compile.keras.src.models.model.Model","text":"called primarily side effect modifying object -place. first argument object also returned, invisibly, enable usage pipe.","code":""},{"path":"https://keras3.posit.co/dev/reference/compile.keras.src.models.model.Model.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Configure a model for training. — compile.keras.src.models.model.Model","text":"","code":"model |> compile(   optimizer = optimizer_adam(learning_rate = 1e-3),   loss = loss_binary_crossentropy(),   metrics = c(metric_binary_accuracy(),               metric_false_negatives()) )"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/config_backend.html","id":null,"dir":"Reference","previous_headings":"","what":"Publicly accessible method for determining the current backend. — config_backend","title":"Publicly accessible method for determining the current backend. — config_backend","text":"Publicly accessible method determining current backend.","code":""},{"path":"https://keras3.posit.co/dev/reference/config_backend.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Publicly accessible method for determining the current backend. — config_backend","text":"","code":"config_backend()"},{"path":"https://keras3.posit.co/dev/reference/config_backend.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Publicly accessible method for determining the current backend. — config_backend","text":"String, name backend Keras currently using. One \"tensorflow\", \"torch\", \"jax\".","code":""},{"path":"https://keras3.posit.co/dev/reference/config_backend.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Publicly accessible method for determining the current backend. — config_backend","text":"","code":"config_backend() ## [1] \"tensorflow\""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/config_disable_flash_attention.html","id":null,"dir":"Reference","previous_headings":"","what":"Disable flash attention. — config_disable_flash_attention","title":"Disable flash attention. — config_disable_flash_attention","text":"Flash attention offers performance optimization attention layers, making especially useful large language models (LLMs) benefit faster memory-efficient attention computations. disabled, supported layers like MultiHeadAttention use flash attention faster computations.","code":""},{"path":"https://keras3.posit.co/dev/reference/config_disable_flash_attention.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Disable flash attention. — config_disable_flash_attention","text":"","code":"config_disable_flash_attention()"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/config_disable_interactive_logging.html","id":null,"dir":"Reference","previous_headings":"","what":"Turn off interactive logging. — config_disable_interactive_logging","title":"Turn off interactive logging. — config_disable_interactive_logging","text":"interactive logging disabled, Keras sends logs absl.logging. best option using Keras non-interactive way, running training inference job server.","code":""},{"path":"https://keras3.posit.co/dev/reference/config_disable_interactive_logging.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Turn off interactive logging. — config_disable_interactive_logging","text":"","code":"config_disable_interactive_logging()"},{"path":"https://keras3.posit.co/dev/reference/config_disable_interactive_logging.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Turn off interactive logging. — config_disable_interactive_logging","text":"return value, called side effects.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/config_disable_traceback_filtering.html","id":null,"dir":"Reference","previous_headings":"","what":"Turn off traceback filtering. — config_disable_traceback_filtering","title":"Turn off traceback filtering. — config_disable_traceback_filtering","text":"Raw Keras tracebacks (also known stack traces) involve many internal frames, can challenging read , actionable end users. default, Keras filters internal frames exceptions raises, keep traceback short, readable, focused actionable (code). See also config_enable_traceback_filtering() config_is_traceback_filtering_enabled(). previously disabled traceback filtering via config_disable_traceback_filtering(), can re-enable via config_enable_traceback_filtering().","code":""},{"path":"https://keras3.posit.co/dev/reference/config_disable_traceback_filtering.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Turn off traceback filtering. — config_disable_traceback_filtering","text":"","code":"config_disable_traceback_filtering()"},{"path":"https://keras3.posit.co/dev/reference/config_disable_traceback_filtering.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Turn off traceback filtering. — config_disable_traceback_filtering","text":"return value, called side effects.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/config_dtype_policy.html","id":null,"dir":"Reference","previous_headings":"","what":"Returns the current default dtype policy object. — config_dtype_policy","title":"Returns the current default dtype policy object. — config_dtype_policy","text":"Returns current default dtype policy object.","code":""},{"path":"https://keras3.posit.co/dev/reference/config_dtype_policy.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Returns the current default dtype policy object. — config_dtype_policy","text":"","code":"config_dtype_policy()"},{"path":"https://keras3.posit.co/dev/reference/config_dtype_policy.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Returns the current default dtype policy object. — config_dtype_policy","text":"DTypePolicy object.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/config_enable_flash_attention.html","id":null,"dir":"Reference","previous_headings":"","what":"Enable flash attention. — config_enable_flash_attention","title":"Enable flash attention. — config_enable_flash_attention","text":"Flash attention offers performance optimization attention layers, making especially useful large language models (LLMs) benefit faster memory-efficient attention computations. enabled, supported layers like layer_multi_head_attention attempt use flash attention faster computations. default, feature enabled. Note enabling flash attention guarantee always used. Typically, inputs must float16 bfloat16 dtype, input layout requirements may vary depending backend.","code":""},{"path":"https://keras3.posit.co/dev/reference/config_enable_flash_attention.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Enable flash attention. — config_enable_flash_attention","text":"","code":"config_enable_flash_attention()"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/config_enable_interactive_logging.html","id":null,"dir":"Reference","previous_headings":"","what":"Turn on interactive logging. — config_enable_interactive_logging","title":"Turn on interactive logging. — config_enable_interactive_logging","text":"interactive logging enabled, Keras displays logs via stdout. provides best experience using Keras interactive environment shell notebook.","code":""},{"path":"https://keras3.posit.co/dev/reference/config_enable_interactive_logging.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Turn on interactive logging. — config_enable_interactive_logging","text":"","code":"config_enable_interactive_logging()"},{"path":"https://keras3.posit.co/dev/reference/config_enable_interactive_logging.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Turn on interactive logging. — config_enable_interactive_logging","text":"return value, called side effects.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/config_enable_traceback_filtering.html","id":null,"dir":"Reference","previous_headings":"","what":"Turn on traceback filtering. — config_enable_traceback_filtering","title":"Turn on traceback filtering. — config_enable_traceback_filtering","text":"Raw Keras tracebacks (also known stack traces) involve many internal frames, can challenging read , actionable end users. default, Keras filters internal frames exceptions raises, keep traceback short, readable, focused actionable (code). See also config_disable_traceback_filtering() config_is_traceback_filtering_enabled(). previously disabled traceback filtering via config_disable_traceback_filtering(), can re-enable via config_enable_traceback_filtering().","code":""},{"path":"https://keras3.posit.co/dev/reference/config_enable_traceback_filtering.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Turn on traceback filtering. — config_enable_traceback_filtering","text":"","code":"config_enable_traceback_filtering()"},{"path":"https://keras3.posit.co/dev/reference/config_enable_traceback_filtering.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Turn on traceback filtering. — config_enable_traceback_filtering","text":"return value, called side effects.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/config_enable_unsafe_deserialization.html","id":null,"dir":"Reference","previous_headings":"","what":"Disables safe mode globally, allowing deserialization of lambdas. — config_enable_unsafe_deserialization","title":"Disables safe mode globally, allowing deserialization of lambdas. — config_enable_unsafe_deserialization","text":"Disables safe mode globally, allowing deserialization lambdas.","code":""},{"path":"https://keras3.posit.co/dev/reference/config_enable_unsafe_deserialization.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Disables safe mode globally, allowing deserialization of lambdas. — config_enable_unsafe_deserialization","text":"","code":"config_enable_unsafe_deserialization()"},{"path":"https://keras3.posit.co/dev/reference/config_enable_unsafe_deserialization.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Disables safe mode globally, allowing deserialization of lambdas. — config_enable_unsafe_deserialization","text":"return value, called side effects.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/config_epsilon.html","id":null,"dir":"Reference","previous_headings":"","what":"Return the value of the fuzz factor used in numeric expressions. — config_epsilon","title":"Return the value of the fuzz factor used in numeric expressions. — config_epsilon","text":"Return value fuzz factor used numeric expressions.","code":""},{"path":"https://keras3.posit.co/dev/reference/config_epsilon.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Return the value of the fuzz factor used in numeric expressions. — config_epsilon","text":"","code":"config_epsilon()"},{"path":"https://keras3.posit.co/dev/reference/config_epsilon.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Return the value of the fuzz factor used in numeric expressions. — config_epsilon","text":"float.","code":""},{"path":"https://keras3.posit.co/dev/reference/config_epsilon.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Return the value of the fuzz factor used in numeric expressions. — config_epsilon","text":"","code":"config_epsilon() ## [1] 1e-07"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/config_floatx.html","id":null,"dir":"Reference","previous_headings":"","what":"Return the default float type, as a string. — config_floatx","title":"Return the default float type, as a string. — config_floatx","text":"E.g. 'bfloat16' 'float16', 'float32', 'float64'.","code":""},{"path":"https://keras3.posit.co/dev/reference/config_floatx.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Return the default float type, as a string. — config_floatx","text":"","code":"config_floatx()"},{"path":"https://keras3.posit.co/dev/reference/config_floatx.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Return the default float type, as a string. — config_floatx","text":"String, current default float type.","code":""},{"path":"https://keras3.posit.co/dev/reference/config_floatx.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Return the default float type, as a string. — config_floatx","text":"","code":"keras3::config_floatx() ## [1] \"float32\""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/config_image_data_format.html","id":null,"dir":"Reference","previous_headings":"","what":"Return the default image data format convention. — config_image_data_format","title":"Return the default image data format convention. — config_image_data_format","text":"Return default image data format convention.","code":""},{"path":"https://keras3.posit.co/dev/reference/config_image_data_format.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Return the default image data format convention. — config_image_data_format","text":"","code":"config_image_data_format()"},{"path":"https://keras3.posit.co/dev/reference/config_image_data_format.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Return the default image data format convention. — config_image_data_format","text":"string, either 'channels_first' 'channels_last'.","code":""},{"path":"https://keras3.posit.co/dev/reference/config_image_data_format.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Return the default image data format convention. — config_image_data_format","text":"","code":"config_image_data_format() ## [1] \"channels_last\""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/config_is_flash_attention_enabled.html","id":null,"dir":"Reference","previous_headings":"","what":"Checks whether flash attention is globally enabled in Keras. — config_is_flash_attention_enabled","title":"Checks whether flash attention is globally enabled in Keras. — config_is_flash_attention_enabled","text":"Flash attention performance-optimized method computing attention large models, transformers, allowing faster memory-efficient operations. function checks global Keras configuration determine flash attention enabled compatible layers (e.g., MultiHeadAttention). Note enabling flash attention guarantee always used. Typically, inputs must float16 bfloat16 dtype, input layout requirements may vary depending backend.","code":""},{"path":"https://keras3.posit.co/dev/reference/config_is_flash_attention_enabled.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Checks whether flash attention is globally enabled in Keras. — config_is_flash_attention_enabled","text":"","code":"config_is_flash_attention_enabled()"},{"path":"https://keras3.posit.co/dev/reference/config_is_flash_attention_enabled.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Checks whether flash attention is globally enabled in Keras. — config_is_flash_attention_enabled","text":"FALSE disabled; otherwise, indicates enabled.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/config_is_interactive_logging_enabled.html","id":null,"dir":"Reference","previous_headings":"","what":"Check if interactive logging is enabled. — config_is_interactive_logging_enabled","title":"Check if interactive logging is enabled. — config_is_interactive_logging_enabled","text":"switch writing logs stdout absl.logging, may use config_enable_interactive_logging() config_disable_interactive_logging().","code":""},{"path":"https://keras3.posit.co/dev/reference/config_is_interactive_logging_enabled.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Check if interactive logging is enabled. — config_is_interactive_logging_enabled","text":"","code":"config_is_interactive_logging_enabled()"},{"path":"https://keras3.posit.co/dev/reference/config_is_interactive_logging_enabled.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Check if interactive logging is enabled. — config_is_interactive_logging_enabled","text":"Boolean, TRUE interactive logging enabled, FALSE otherwise.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/config_is_nnx_enabled.html","id":null,"dir":"Reference","previous_headings":"","what":"Check whether NNX-specific features are enabled on the JAX backend. — config_is_nnx_enabled","title":"Check whether NNX-specific features are enabled on the JAX backend. — config_is_nnx_enabled","text":"Check whether NNX-specific features enabled JAX backend.","code":""},{"path":"https://keras3.posit.co/dev/reference/config_is_nnx_enabled.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Check whether NNX-specific features are enabled on the JAX backend. — config_is_nnx_enabled","text":"","code":"config_is_nnx_enabled()"},{"path":"https://keras3.posit.co/dev/reference/config_is_nnx_enabled.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Check whether NNX-specific features are enabled on the JAX backend. — config_is_nnx_enabled","text":"Logical flag; TRUE NNX backend features enabled, FALSE otherwise.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/config_is_traceback_filtering_enabled.html","id":null,"dir":"Reference","previous_headings":"","what":"Check if traceback filtering is enabled. — config_is_traceback_filtering_enabled","title":"Check if traceback filtering is enabled. — config_is_traceback_filtering_enabled","text":"Raw Keras tracebacks (also known stack traces) involve many internal frames, can challenging read , actionable end users. default, Keras filters internal frames exceptions raises, keep traceback short, readable, focused actionable (code). See also config_enable_traceback_filtering() config_disable_traceback_filtering(). previously disabled traceback filtering via config_disable_traceback_filtering(), can re-enable via config_enable_traceback_filtering().","code":""},{"path":"https://keras3.posit.co/dev/reference/config_is_traceback_filtering_enabled.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Check if traceback filtering is enabled. — config_is_traceback_filtering_enabled","text":"","code":"config_is_traceback_filtering_enabled()"},{"path":"https://keras3.posit.co/dev/reference/config_is_traceback_filtering_enabled.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Check if traceback filtering is enabled. — config_is_traceback_filtering_enabled","text":"Boolean, TRUE traceback filtering enabled, FALSE otherwise.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/config_max_epochs.html","id":null,"dir":"Reference","previous_headings":"","what":"Configure the default training loop limits. — config_max_epochs","title":"Configure the default training loop limits. — config_max_epochs","text":"helpers control caps built-training loops obey running fit(), evaluate(), predict(). values can also provided via KERAS_MAX_EPOCHS KERAS_MAX_STEPS_PER_EPOCH environment variables quickly constrain run without modifying source code.","code":""},{"path":"https://keras3.posit.co/dev/reference/config_max_epochs.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Configure the default training loop limits. — config_max_epochs","text":"","code":"config_max_epochs()  config_set_max_epochs(max_epochs)  config_max_steps_per_epoch()  config_set_max_steps_per_epoch(max_steps_per_epoch)"},{"path":"https://keras3.posit.co/dev/reference/config_max_epochs.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Configure the default training loop limits. — config_max_epochs","text":"max_epochs Integer upper bound epochs processed built-training loops. Use NULL remove cap. max_steps_per_epoch Integer upper bound steps processed per epoch built-training loops. Use NULL remove cap.","code":""},{"path":"https://keras3.posit.co/dev/reference/config_max_epochs.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Configure the default training loop limits. — config_max_epochs","text":"config_max_epochs() config_max_steps_per_epoch() return current integer limits (NULL cap unset). setter variants return NULL invisibly called side effects.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/config_set_backend.html","id":null,"dir":"Reference","previous_headings":"","what":"Reload the backend (and the Keras package). — config_set_backend","title":"Reload the backend (and the Keras package). — config_set_backend","text":"Reload backend (Keras package).","code":""},{"path":"https://keras3.posit.co/dev/reference/config_set_backend.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Reload the backend (and the Keras package). — config_set_backend","text":"","code":"config_set_backend(backend)"},{"path":"https://keras3.posit.co/dev/reference/config_set_backend.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Reload the backend (and the Keras package). — config_set_backend","text":"backend String","code":""},{"path":"https://keras3.posit.co/dev/reference/config_set_backend.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Reload the backend (and the Keras package). — config_set_backend","text":"Nothing, function called side effect.","code":""},{"path":"https://keras3.posit.co/dev/reference/config_set_backend.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Reload the backend (and the Keras package). — config_set_backend","text":"","code":"config_set_backend(\"jax\")"},{"path":"https://keras3.posit.co/dev/reference/config_set_backend.html","id":"warning","dir":"Reference","previous_headings":"","what":"WARNING","title":"Reload the backend (and the Keras package). — config_set_backend","text":"Using function dangerous done carefully. Changing backend convert type already-instantiated objects. Thus, layers / tensors / etc. already created longer usable without errors. strongly recommended keep around Keras-originated objects instances created calling config_set_backend(). includes function class instance uses Keras functionality. code needs re-executed calling config_set_backend().","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/config_set_dtype_policy.html","id":null,"dir":"Reference","previous_headings":"","what":"Sets the default dtype policy globally. — config_set_dtype_policy","title":"Sets the default dtype policy globally. — config_set_dtype_policy","text":"Sets default dtype policy globally.","code":""},{"path":"https://keras3.posit.co/dev/reference/config_set_dtype_policy.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Sets the default dtype policy globally. — config_set_dtype_policy","text":"","code":"config_set_dtype_policy(policy)"},{"path":"https://keras3.posit.co/dev/reference/config_set_dtype_policy.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Sets the default dtype policy globally. — config_set_dtype_policy","text":"policy string DTypePolicy object.","code":""},{"path":"https://keras3.posit.co/dev/reference/config_set_dtype_policy.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Sets the default dtype policy globally. — config_set_dtype_policy","text":"return value, called side effects.","code":""},{"path":"https://keras3.posit.co/dev/reference/config_set_dtype_policy.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Sets the default dtype policy globally. — config_set_dtype_policy","text":"","code":"config_set_dtype_policy(\"mixed_float16\")"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/config_set_epsilon.html","id":null,"dir":"Reference","previous_headings":"","what":"Set the value of the fuzz factor used in numeric expressions. — config_set_epsilon","title":"Set the value of the fuzz factor used in numeric expressions. — config_set_epsilon","text":"Set value fuzz factor used numeric expressions.","code":""},{"path":"https://keras3.posit.co/dev/reference/config_set_epsilon.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Set the value of the fuzz factor used in numeric expressions. — config_set_epsilon","text":"","code":"config_set_epsilon(value)"},{"path":"https://keras3.posit.co/dev/reference/config_set_epsilon.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Set the value of the fuzz factor used in numeric expressions. — config_set_epsilon","text":"value float. New value epsilon.","code":""},{"path":"https://keras3.posit.co/dev/reference/config_set_epsilon.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Set the value of the fuzz factor used in numeric expressions. — config_set_epsilon","text":"return value, called side effects.","code":""},{"path":"https://keras3.posit.co/dev/reference/config_set_epsilon.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Set the value of the fuzz factor used in numeric expressions. — config_set_epsilon","text":"","code":"config_epsilon() ## [1] 1e-07 config_set_epsilon(1e-5) config_epsilon() ## [1] 1e-05 # Set it back to the default value. config_set_epsilon(1e-7)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/config_set_floatx.html","id":null,"dir":"Reference","previous_headings":"","what":"Set the default float dtype. — config_set_floatx","title":"Set the default float dtype. — config_set_floatx","text":"Set default float dtype.","code":""},{"path":"https://keras3.posit.co/dev/reference/config_set_floatx.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Set the default float dtype. — config_set_floatx","text":"","code":"config_set_floatx(value)"},{"path":"https://keras3.posit.co/dev/reference/config_set_floatx.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Set the default float dtype. — config_set_floatx","text":"value String; 'bfloat16', 'float16', 'float32', 'float64'.","code":""},{"path":"https://keras3.posit.co/dev/reference/config_set_floatx.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Set the default float dtype. — config_set_floatx","text":"return value, called side effects.","code":""},{"path":"https://keras3.posit.co/dev/reference/config_set_floatx.html","id":"note","dir":"Reference","previous_headings":"","what":"Note","title":"Set the default float dtype. — config_set_floatx","text":"recommended set \"float16\" training, likely cause numeric stability issues. Instead, mixed precision, leverages mix float16 float32. can configured calling keras3::keras$mixed_precision$set_dtype_policy('mixed_float16').","code":""},{"path":"https://keras3.posit.co/dev/reference/config_set_floatx.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Set the default float dtype. — config_set_floatx","text":"","code":"config_floatx() ## [1] \"float32\" config_set_floatx('float64') config_floatx() ## [1] \"float64\" # Set it back to float32 config_set_floatx('float32')"},{"path":"https://keras3.posit.co/dev/reference/config_set_floatx.html","id":"raises","dir":"Reference","previous_headings":"","what":"Raises","title":"Set the default float dtype. — config_set_floatx","text":"ValueError: case invalid value.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/config_set_image_data_format.html","id":null,"dir":"Reference","previous_headings":"","what":"Set the value of the image data format convention. — config_set_image_data_format","title":"Set the value of the image data format convention. — config_set_image_data_format","text":"Set value image data format convention.","code":""},{"path":"https://keras3.posit.co/dev/reference/config_set_image_data_format.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Set the value of the image data format convention. — config_set_image_data_format","text":"","code":"config_set_image_data_format(data_format)"},{"path":"https://keras3.posit.co/dev/reference/config_set_image_data_format.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Set the value of the image data format convention. — config_set_image_data_format","text":"data_format string. 'channels_first' 'channels_last'.","code":""},{"path":"https://keras3.posit.co/dev/reference/config_set_image_data_format.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Set the value of the image data format convention. — config_set_image_data_format","text":"return value, called side effects.","code":""},{"path":"https://keras3.posit.co/dev/reference/config_set_image_data_format.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Set the value of the image data format convention. — config_set_image_data_format","text":"","code":"config_image_data_format() ## [1] \"channels_last\" # 'channels_last' keras3::config_set_image_data_format('channels_first') config_image_data_format() ## [1] \"channels_first\" # Set it back to `'channels_last'` keras3::config_set_image_data_format('channels_last')"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/constraint_maxnorm.html","id":null,"dir":"Reference","previous_headings":"","what":"MaxNorm weight constraint. — constraint_maxnorm","title":"MaxNorm weight constraint. — constraint_maxnorm","text":"Constrains weights incident hidden unit norm less equal desired value.","code":""},{"path":"https://keras3.posit.co/dev/reference/constraint_maxnorm.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"MaxNorm weight constraint. — constraint_maxnorm","text":"","code":"constraint_maxnorm(max_value = 2L, axis = 1L)"},{"path":"https://keras3.posit.co/dev/reference/constraint_maxnorm.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"MaxNorm weight constraint. — constraint_maxnorm","text":"max_value maximum norm value incoming weights. axis integer, axis along calculate weight norms. instance, Dense layer weight matrix shape (input_dim, output_dim), set axis 0 constrain weight vector length (input_dim,). Conv2D layer data_format = \"channels_last\", weight tensor shape (rows, cols, input_depth, output_depth), set axis [0, 1, 2] constrain weights filter tensor size (rows, cols, input_depth).","code":""},{"path":"https://keras3.posit.co/dev/reference/constraint_maxnorm.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"MaxNorm weight constraint. — constraint_maxnorm","text":"Constraint instance, callable can passed layer constructors used directly calling tensors.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/constraint_minmaxnorm.html","id":null,"dir":"Reference","previous_headings":"","what":"MinMaxNorm weight constraint. — constraint_minmaxnorm","title":"MinMaxNorm weight constraint. — constraint_minmaxnorm","text":"Constrains weights incident hidden unit norm lower bound upper bound.","code":""},{"path":"https://keras3.posit.co/dev/reference/constraint_minmaxnorm.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"MinMaxNorm weight constraint. — constraint_minmaxnorm","text":"","code":"constraint_minmaxnorm(min_value = 0, max_value = 1, rate = 1, axis = 1L)"},{"path":"https://keras3.posit.co/dev/reference/constraint_minmaxnorm.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"MinMaxNorm weight constraint. — constraint_minmaxnorm","text":"min_value minimum norm incoming weights. max_value maximum norm incoming weights. rate rate enforcing constraint: weights rescaled yield op_clip? (1 - rate) * norm + rate * op_clip(norm, min_value, max_value). Effectively, means rate = 1.0 stands strict enforcement constraint, rate<1.0 means weights rescaled step slowly move towards value inside desired interval. axis integer, axis along calculate weight norms. instance, Dense layer weight matrix shape (input_dim, output_dim), set axis 0 constrain weight vector length (input_dim,). Conv2D layer data_format = \"channels_last\", weight tensor shape (rows, cols, input_depth, output_depth), set axis [0, 1, 2] constrain weights filter tensor size (rows, cols, input_depth).","code":""},{"path":"https://keras3.posit.co/dev/reference/constraint_minmaxnorm.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"MinMaxNorm weight constraint. — constraint_minmaxnorm","text":"Constraint instance, callable can passed layer constructors used directly calling tensors.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/constraint_nonneg.html","id":null,"dir":"Reference","previous_headings":"","what":"Constrains the weights to be non-negative. — constraint_nonneg","title":"Constrains the weights to be non-negative. — constraint_nonneg","text":"Constrains weights non-negative.","code":""},{"path":"https://keras3.posit.co/dev/reference/constraint_nonneg.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Constrains the weights to be non-negative. — constraint_nonneg","text":"","code":"constraint_nonneg()"},{"path":"https://keras3.posit.co/dev/reference/constraint_nonneg.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Constrains the weights to be non-negative. — constraint_nonneg","text":"Constraint instance, callable can passed layer constructors used directly calling tensors.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/constraint_unitnorm.html","id":null,"dir":"Reference","previous_headings":"","what":"Constrains the weights incident to each hidden unit to have unit norm. — constraint_unitnorm","title":"Constrains the weights incident to each hidden unit to have unit norm. — constraint_unitnorm","text":"Constrains weights incident hidden unit unit norm.","code":""},{"path":"https://keras3.posit.co/dev/reference/constraint_unitnorm.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Constrains the weights incident to each hidden unit to have unit norm. — constraint_unitnorm","text":"","code":"constraint_unitnorm(axis = 1L)"},{"path":"https://keras3.posit.co/dev/reference/constraint_unitnorm.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Constrains the weights incident to each hidden unit to have unit norm. — constraint_unitnorm","text":"axis integer, axis along calculate weight norms. instance, Dense layer weight matrix shape (input_dim, output_dim), set axis 0 constrain weight vector length (input_dim,). Conv2D layer data_format = \"channels_last\", weight tensor shape (rows, cols, input_depth, output_depth), set axis [0, 1, 2] constrain weights filter tensor size (rows, cols, input_depth).","code":""},{"path":"https://keras3.posit.co/dev/reference/constraint_unitnorm.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Constrains the weights incident to each hidden unit to have unit norm. — constraint_unitnorm","text":"Constraint instance, callable can passed layer constructors used directly calling tensors.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/count_params.html","id":null,"dir":"Reference","previous_headings":"","what":"Count the total number of scalars composing the weights. — count_params","title":"Count the total number of scalars composing the weights. — count_params","text":"Count total number scalars composing weights.","code":""},{"path":"https://keras3.posit.co/dev/reference/count_params.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Count the total number of scalars composing the weights. — count_params","text":"","code":"count_params(object)"},{"path":"https://keras3.posit.co/dev/reference/count_params.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Count the total number of scalars composing the weights. — count_params","text":"object Layer model object","code":""},{"path":"https://keras3.posit.co/dev/reference/count_params.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Count the total number of scalars composing the weights. — count_params","text":"integer count","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/custom_metric.html","id":null,"dir":"Reference","previous_headings":"","what":"Custom metric function — custom_metric","title":"Custom metric function — custom_metric","text":"Custom metric function","code":""},{"path":"https://keras3.posit.co/dev/reference/custom_metric.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Custom metric function — custom_metric","text":"","code":"custom_metric(name, metric_fn)"},{"path":"https://keras3.posit.co/dev/reference/custom_metric.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Custom metric function — custom_metric","text":"name name used show training progress output metric_fn R function signature function(y_true, y_pred) accepts tensors.","code":""},{"path":"https://keras3.posit.co/dev/reference/custom_metric.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Custom metric function — custom_metric","text":"callable function __name__ attribute.","code":""},{"path":"https://keras3.posit.co/dev/reference/custom_metric.html","id":"details","dir":"Reference","previous_headings":"","what":"Details","title":"Custom metric function — custom_metric","text":"can provide arbitrary R function custom metric. Note y_true y_pred parameters tensors, computations use op_* tensor functions. Use custom_metric() function define custom metric. Note name ('mean_pred') provided custom metric function: name used within training progress output. want save load model custom metrics, also call register_keras_serializable(), specify metric call load_model(). example: load_model(\"my_model.keras\", c('mean_pred' = metric_mean_pred)). Alternatively, can wrap code call with_custom_object_scope() allow refer metric name just like built keras metrics. Alternative ways supplying custom metrics: custom_metric(): Arbitrary R function. metric_mean_wrapper(): Wrap arbitrary R function Metric instance. Create custom Metric() subclass.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/dataset_boston_housing.html","id":null,"dir":"Reference","previous_headings":"","what":"Boston housing price regression dataset — dataset_boston_housing","title":"Boston housing price regression dataset — dataset_boston_housing","text":"Dataset taken StatLib library maintained Carnegie Mellon University.","code":""},{"path":"https://keras3.posit.co/dev/reference/dataset_boston_housing.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Boston housing price regression dataset — dataset_boston_housing","text":"","code":"dataset_boston_housing(   path = \"boston_housing.npz\",   test_split = 0.2,   seed = 113L )"},{"path":"https://keras3.posit.co/dev/reference/dataset_boston_housing.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Boston housing price regression dataset — dataset_boston_housing","text":"path Path cache dataset locally (relative ~/.keras/datasets). test_split fraction data reserve test set. seed Random seed shuffling data computing test split.","code":""},{"path":"https://keras3.posit.co/dev/reference/dataset_boston_housing.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Boston housing price regression dataset — dataset_boston_housing","text":"Lists training test data: train$x, train$y, test$x, test$y. Samples contain 13 attributes houses different locations around Boston suburbs late 1970s. Targets median values houses location (k$).","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/dataset_california_housing.html","id":null,"dir":"Reference","previous_headings":"","what":"Loads the California Housing dataset. — dataset_california_housing","title":"Loads the California Housing dataset. — dataset_california_housing","text":"dataset obtained StatLib repository. continuous regression dataset 20,640 samples 8 features . target variable scalar: median house value California districts, dollars. 8 input features following: MedInc: median income block group HouseAge: median house age block group AveRooms: average number rooms per household AveBedrms: average number bedrooms per household Population: block group population AveOccup: average number household members Latitude: block group latitude Longitude: block group longitude dataset derived 1990 U.S. census, using one row per census block group. block group smallest geographical unit U.S. Census Bureau publishes sample data (block group typically population 600 3,000 people). household group people residing within home. Since average number rooms bedrooms dataset provided per household, columns may take surprisingly large values block groups households many empty houses, vacation resorts.","code":""},{"path":"https://keras3.posit.co/dev/reference/dataset_california_housing.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Loads the California Housing dataset. — dataset_california_housing","text":"","code":"dataset_california_housing(   version = \"large\",   path = \"california_housing.npz\",   test_split = 0.2,   seed = 113L )"},{"path":"https://keras3.posit.co/dev/reference/dataset_california_housing.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Loads the California Housing dataset. — dataset_california_housing","text":"version \"small\" \"large\". small version contains 600 samples, large version contains 20,640 samples. purpose small version serve approximate replacement deprecated boston_housing dataset. path path cache dataset locally (relative Sys.getenv(\"KERAS_HOME\")). test_split fraction data reserve test set. seed Random seed shuffling data computing test split.","code":""},{"path":"https://keras3.posit.co/dev/reference/dataset_california_housing.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Loads the California Housing dataset. — dataset_california_housing","text":"Nested list arrays: (x_train, y_train), (x_test, y_test). x_train, x_test: arrays shape (num_samples, 8) containing either training samples (x_train), test samples (y_train). y_train, y_test: arrays shape (num_samples) containing target scalars. targets float scalars typically 25,000 500,000 represent home prices dollars.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/dataset_cifar10.html","id":null,"dir":"Reference","previous_headings":"","what":"CIFAR10 small image classification — dataset_cifar10","title":"CIFAR10 small image classification — dataset_cifar10","text":"Dataset 50,000 32x32 color training images, labeled 10 categories, 10,000 test images.","code":""},{"path":"https://keras3.posit.co/dev/reference/dataset_cifar10.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"CIFAR10 small image classification — dataset_cifar10","text":"","code":"dataset_cifar10()"},{"path":"https://keras3.posit.co/dev/reference/dataset_cifar10.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"CIFAR10 small image classification — dataset_cifar10","text":"Lists training test data: train$x, train$y, test$x, test$y. x data array RGB image data shape (num_samples, 3, 32, 32). y data array category labels (integers range 0-9) shape (num_samples).","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/dataset_cifar100.html","id":null,"dir":"Reference","previous_headings":"","what":"CIFAR100 small image classification — dataset_cifar100","title":"CIFAR100 small image classification — dataset_cifar100","text":"Dataset 50,000 32x32 color training images, labeled 100 categories, 10,000 test images.","code":""},{"path":"https://keras3.posit.co/dev/reference/dataset_cifar100.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"CIFAR100 small image classification — dataset_cifar100","text":"","code":"dataset_cifar100(label_mode = c(\"fine\", \"coarse\"))"},{"path":"https://keras3.posit.co/dev/reference/dataset_cifar100.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"CIFAR100 small image classification — dataset_cifar100","text":"label_mode one \"fine\", \"coarse\".","code":""},{"path":"https://keras3.posit.co/dev/reference/dataset_cifar100.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"CIFAR100 small image classification — dataset_cifar100","text":"Lists training test data: train$x, train$y, test$x, test$y. x data array RGB image data shape (num_samples, 3, 32, 32). y data array category labels shape (num_samples).","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/dataset_fashion_mnist.html","id":null,"dir":"Reference","previous_headings":"","what":"Fashion-MNIST database of fashion articles — dataset_fashion_mnist","title":"Fashion-MNIST database of fashion articles — dataset_fashion_mnist","text":"Dataset 60,000 28x28 grayscale images 10 fashion article classes, along test set 10,000 images. dataset can used drop-replacement MNIST. class labels encoded integers 0-9 correspond T-shirt/top, Trouser, Pullover, Dress, Coat, Sandal, Shirt,","code":""},{"path":"https://keras3.posit.co/dev/reference/dataset_fashion_mnist.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Fashion-MNIST database of fashion articles — dataset_fashion_mnist","text":"","code":"dataset_fashion_mnist()"},{"path":"https://keras3.posit.co/dev/reference/dataset_fashion_mnist.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Fashion-MNIST database of fashion articles — dataset_fashion_mnist","text":"Lists training test data: train$x, train$y, test$x, test$y, x array grayscale image data shape (num_samples, 28, 28) y array article labels (integers range 0-9) shape (num_samples).","code":""},{"path":"https://keras3.posit.co/dev/reference/dataset_fashion_mnist.html","id":"details","dir":"Reference","previous_headings":"","what":"Details","title":"Fashion-MNIST database of fashion articles — dataset_fashion_mnist","text":"Dataset 60,000 28x28 grayscale images 10 fashion categories, along test set 10,000 images. dataset can used drop-replacement MNIST. class labels : 0 - T-shirt/top 1 - Trouser 2 - Pullover 3 - Dress 4 - Coat 5 - Sandal 6 - Shirt 7 - Sneaker 8 - Bag 9 - Ankle boot","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/dataset_imdb.html","id":null,"dir":"Reference","previous_headings":"","what":"IMDB Movie reviews sentiment classification — dataset_imdb","title":"IMDB Movie reviews sentiment classification — dataset_imdb","text":"Dataset 25,000 movies reviews IMDB, labeled sentiment (positive/negative). Reviews preprocessed, review encoded sequence word indexes (integers). convenience, words indexed overall frequency dataset, instance integer \"3\" encodes 3rd frequent word data. allows quick filtering operations : \"consider top 10,000 common words, eliminate top 20 common words\".","code":""},{"path":"https://keras3.posit.co/dev/reference/dataset_imdb.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"IMDB Movie reviews sentiment classification — dataset_imdb","text":"","code":"dataset_imdb(   path = \"imdb.npz\",   num_words = NULL,   skip_top = 0L,   maxlen = NULL,   seed = 113L,   start_char = 1L,   oov_char = 2L,   index_from = 3L )  dataset_imdb_word_index(path = \"imdb_word_index.json\")"},{"path":"https://keras3.posit.co/dev/reference/dataset_imdb.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"IMDB Movie reviews sentiment classification — dataset_imdb","text":"path cache data (relative ~/.keras/dataset). num_words Max number words include. Words ranked often occur (training set) frequent words kept skip_top Skip top N frequently occuring words (may informative). maxlen sequences longer filtered . seed random seed sample shuffling. start_char start sequence marked character. Set 1 0 usually padding character. oov_char Words cut num_words skip_top limit replaced character. index_from Index actual words index higher.","code":""},{"path":"https://keras3.posit.co/dev/reference/dataset_imdb.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"IMDB Movie reviews sentiment classification — dataset_imdb","text":"Lists training test data: train$x, train$y, test$x, test$y. x data includes integer sequences. num_words argument specific, maximum possible index value num_words-1. maxlen argument specified, largest possible sequence length maxlen. y data includes set integer labels (0 1). dataset_imdb_word_index() function returns list names words values integer.","code":""},{"path":"https://keras3.posit.co/dev/reference/dataset_imdb.html","id":"details","dir":"Reference","previous_headings":"","what":"Details","title":"IMDB Movie reviews sentiment classification — dataset_imdb","text":"convention, \"0\" stand specific word, instead used encode unknown word.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/dataset_mnist.html","id":null,"dir":"Reference","previous_headings":"","what":"MNIST database of handwritten digits — dataset_mnist","title":"MNIST database of handwritten digits — dataset_mnist","text":"Dataset 60,000 28x28 grayscale images 10 digits, along test set 10,000 images.","code":""},{"path":"https://keras3.posit.co/dev/reference/dataset_mnist.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"MNIST database of handwritten digits — dataset_mnist","text":"","code":"dataset_mnist(path = \"mnist.npz\")"},{"path":"https://keras3.posit.co/dev/reference/dataset_mnist.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"MNIST database of handwritten digits — dataset_mnist","text":"path Path cache dataset locally (relative ~/.keras/datasets).","code":""},{"path":"https://keras3.posit.co/dev/reference/dataset_mnist.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"MNIST database of handwritten digits — dataset_mnist","text":"Lists training test data: train$x, train$y, test$x, test$y, x array grayscale image data shape (num_samples, 28, 28) y array digit labels (integers range 0-9) shape (num_samples).","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/dataset_reuters.html","id":null,"dir":"Reference","previous_headings":"","what":"Reuters newswire topics classification — dataset_reuters","title":"Reuters newswire topics classification — dataset_reuters","text":"Dataset 11,228 newswires Reuters, labeled 46 topics. dataset_imdb() , wire encoded sequence word indexes (conventions).","code":""},{"path":"https://keras3.posit.co/dev/reference/dataset_reuters.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Reuters newswire topics classification — dataset_reuters","text":"","code":"dataset_reuters(   path = \"reuters.npz\",   num_words = NULL,   skip_top = 0L,   maxlen = NULL,   test_split = 0.2,   seed = 113L,   start_char = 1L,   oov_char = 2L,   index_from = 3L )  dataset_reuters_word_index(path = \"reuters_word_index.pkl\")"},{"path":"https://keras3.posit.co/dev/reference/dataset_reuters.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Reuters newswire topics classification — dataset_reuters","text":"path cache data (relative ~/.keras/dataset). num_words Max number words include. Words ranked often occur (training set) frequent words kept skip_top Skip top N frequently occuring words (may informative). maxlen Truncate sequences length. test_split Fraction dataset used test data. seed Random seed sample shuffling. start_char start sequence marked character. Set 1 0 usually padding character. oov_char words cut num_words skip_top limit replaced character. index_from index actual words index higher.","code":""},{"path":"https://keras3.posit.co/dev/reference/dataset_reuters.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Reuters newswire topics classification — dataset_reuters","text":"Lists training test data: train$x, train$y, test$x, test$y format dataset_imdb(). dataset_reuters_word_index() function returns list names words values integer. e.g. word_index[[\"giraffe\"]] might return 1234.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/deserialize_keras_object.html","id":null,"dir":"Reference","previous_headings":"","what":"Retrieve the object by deserializing the config dict. — deserialize_keras_object","title":"Retrieve the object by deserializing the config dict. — deserialize_keras_object","text":"config dict Python dictionary consists set key-value pairs, represents Keras object, Optimizer, Layer, Metrics, etc. saving loading library uses following keys record information Keras object: class_name: String. name class, exactly defined source code, \"LossesContainer\". config: Named List. Library-defined user-defined key-value pairs store configuration object, obtained object$get_config(). module: String. path python module. Built-Keras classes expect prefix keras. registered_name: String. key class registered via register_keras_serializable(package, name) API. key format '{package}>{name}', package name arguments passed register_keras_serializable(). name provided, uses class name. registered_name successfully resolves class (registered), class_name config values config dict used. registered_name used non-built-classes. example, following config list represents built-Adam optimizer relevant config:     class exported Keras namespace, library tracks module class_name. example:   following config represents user-customized MeanSquaredError loss:","code":"config <- list(   class_name = \"Adam\",   config = list(     amsgrad = FALSE,     beta_1 = 0.8999999761581421,     beta_2 = 0.9990000128746033,     epsilon = 1e-07,     learning_rate = 0.0010000000474974513,     name = \"Adam\"   ),   module = \"keras.optimizers\",   registered_name = NULL ) # Returns an `Adam` instance identical to the original one. deserialize_keras_object(config) ## <keras.src.optimizers.adam.Adam object at 0x0> config <- list(   class_name = \"MetricsList\",   config =  list(     ...   ),   module = \"keras.trainers.compile_utils\",   registered_name = \"MetricsList\" )  # Returns a `MetricsList` instance identical to the original one. deserialize_keras_object(config) # define a custom object loss_modified_mse <- Loss(   \"ModifiedMeanSquaredError\",   inherit = loss_mean_squared_error)  # register the custom object register_keras_serializable(loss_modified_mse)  # confirm object is registered get_custom_objects() ## $`keras3>ModifiedMeanSquaredError` ## <class '<r-namespace:keras3>.ModifiedMeanSquaredError'> ##  signature: ( ##    reduction='sum_over_batch_size', ##    name='mean_squared_error', ##    dtype=None ## ) get_registered_name(loss_modified_mse) ## [1] \"keras3>ModifiedMeanSquaredError\" # now custom object instances can be serialized full_config <- serialize_keras_object(loss_modified_mse())  # the `config` arguments will be passed to loss_modified_mse() str(full_config) ## List of 4 ##  $ module         : chr \"<r-namespace:keras3>\" ##  $ class_name     : chr \"ModifiedMeanSquaredError\" ##  $ config         :List of 2 ##   ..$ name     : chr \"mean_squared_error\" ##   ..$ reduction: chr \"sum_over_batch_size\" ##  $ registered_name: chr \"keras3>ModifiedMeanSquaredError\" # and custom object instances can be deserialized deserialize_keras_object(full_config) ## <LossFunctionWrapper(<function mean_squared_error at 0x0>, kwargs={})> ##  signature: (y_true, y_pred, sample_weight=None) # Returns the `ModifiedMeanSquaredError` object"},{"path":"https://keras3.posit.co/dev/reference/deserialize_keras_object.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Retrieve the object by deserializing the config dict. — deserialize_keras_object","text":"","code":"deserialize_keras_object(config, custom_objects = NULL, safe_mode = TRUE, ...)"},{"path":"https://keras3.posit.co/dev/reference/deserialize_keras_object.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Retrieve the object by deserializing the config dict. — deserialize_keras_object","text":"config Named list describing object. custom_objects Named list containing mapping custom object names corresponding classes functions. safe_mode Boolean, whether disallow unsafe lambda deserialization. safe_mode=FALSE, loading object potential trigger arbitrary code execution. argument applicable Keras v3 model format. Defaults TRUE. ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/deserialize_keras_object.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Retrieve the object by deserializing the config dict. — deserialize_keras_object","text":"object described config dictionary.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/evaluate.keras.src.models.model.Model.html","id":null,"dir":"Reference","previous_headings":"","what":"Evaluate a Keras Model — evaluate.keras.src.models.model.Model","title":"Evaluate a Keras Model — evaluate.keras.src.models.model.Model","text":"functions returns loss value metrics values model test mode. Computation done batches (see batch_size arg.)","code":""},{"path":"https://keras3.posit.co/dev/reference/evaluate.keras.src.models.model.Model.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Evaluate a Keras Model — evaluate.keras.src.models.model.Model","text":"","code":"# S3 method for class 'keras.src.models.model.Model' evaluate(   object,   x = NULL,   y = NULL,   ...,   batch_size = NULL,   verbose = getOption(\"keras.verbose\", default = \"auto\"),   sample_weight = NULL,   steps = NULL,   callbacks = NULL )"},{"path":"https://keras3.posit.co/dev/reference/evaluate.keras.src.models.model.Model.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Evaluate a Keras Model — evaluate.keras.src.models.model.Model","text":"object Keras model object x Input data. can : R array (array-like), list arrays (case model multiple inputs). backend-native tensor, list tensors (case model multiple inputs). named list mapping input names corresponding array/tensors, model named inputs. tf.data.Dataset. return tuple either (inputs, targets) (inputs, targets, sample_weights). generator returning (inputs, targets) (inputs, targets, sample_weights). y Target data. Like input data x, either R array(s) backend-native tensor(s). x tf.data.Dataset generator function, y specified (since targets obtained iterator/dataset). ... forward/backward compatability. batch_size Integer NULL. Number samples per batch computation. unspecified, batch_size default 32. specify batch_size data form tf dataset generator (since generate batches). verbose \"auto\", 0, 1, 2. Verbosity mode. 0 = silent, 1 = progress bar, 2 = single line. \"auto\" becomes 1 cases, 2 knitr render running distributed training server. Note progress bar particularly useful logged file, verbose=2 recommended running interactively (e.g. production environment). Defaults \"auto\". sample_weight Optional array tensor weights training samples, used weighting loss function (training ). can either pass flat (1D) array tensor length input samples (1:1 mapping weights samples), case temporal data, can pass 2D array tensor shape (samples, sequence_length) apply different weight every timestep every sample. argument supported x tf.data.Dataset, Python generator function. Instead, provide sample_weights third element x. Note sample weighting apply metrics specified via metrics argument compile(). apply sample weighting metrics, can specify via weighted_metrics compile() instead. steps Integer NULL. Total number steps (batches samples) declaring evaluation round finished. Ignored default value NULL. x tf.data.Dataset steps NULL, evaluation run dataset exhausted. case infinitely repeating dataset, run indefinitely. callbacks List Callback instances. List callbacks apply evaluation.","code":""},{"path":"https://keras3.posit.co/dev/reference/evaluate.keras.src.models.model.Model.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Evaluate a Keras Model — evaluate.keras.src.models.model.Model","text":"Scalar test loss (model single output metrics) list scalars (model multiple outputs /metrics). attribute model$metrics_names give display labels scalar outputs.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/export_savedmodel.keras.src.models.model.Model.html","id":null,"dir":"Reference","previous_headings":"","what":"Export the model as an artifact for inference. — export_savedmodel.keras.src.models.model.Model","title":"Export the model as an artifact for inference. — export_savedmodel.keras.src.models.model.Model","text":"(e.g. via TF-Serving). Note: can currently used TensorFlow JAX backends. method lets export model lightweight SavedModel artifact contains model's forward pass (call() method) can served via e.g. TF-Serving. forward pass registered name serve() (see example ). original code model (including custom layers may used) longer necessary reload artifact – entirely standalone. Note: feature currently supported TensorFlow, JAX Torch backends. Note: aware exported artifact may contain information local file system using format=\"onnx\", verbose=TRUE Torch backend.","code":""},{"path":"https://keras3.posit.co/dev/reference/export_savedmodel.keras.src.models.model.Model.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Export the model as an artifact for inference. — export_savedmodel.keras.src.models.model.Model","text":"","code":"# S3 method for class 'keras.src.models.model.Model' export_savedmodel(   object,   export_dir_base,   ...,   format = \"tf_saved_model\",   verbose = NULL,   input_signature = NULL )"},{"path":"https://keras3.posit.co/dev/reference/export_savedmodel.keras.src.models.model.Model.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Export the model as an artifact for inference. — export_savedmodel.keras.src.models.model.Model","text":"object keras model. export_dir_base string, file path save artifact. ... Additional keyword arguments: Specific JAX backend format=\"tf_saved_model\": is_static: Optional bool. Indicates whether fn static. Set FALSE fn involves state updates (e.g., RNG seeds counters). jax2tf_kwargs: Optional dict. Arguments jax2tf.convert. See documentation jax2tf.convert. native_serialization polymorphic_shapes provided, automatically computed. format string. export format. Supported values: \"tf_saved_model\" \"onnx\".  Defaults \"tf_saved_model\". verbose Bool. Whether print variables exported model. Defaults NULL, uses default value set different backends formats. input_signature Optional. Specifies shape dtype model inputs. Can structure keras.InputSpec, tf.TensorSpec, backend.KerasTensor, backend tensor. provided, automatically computed. Defaults NULL.","code":""},{"path":"https://keras3.posit.co/dev/reference/export_savedmodel.keras.src.models.model.Model.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Export the model as an artifact for inference. — export_savedmodel.keras.src.models.model.Model","text":"called primarily side effect exporting object. first argument, object also returned, invisibly, enable usage pipe.","code":""},{"path":"https://keras3.posit.co/dev/reference/export_savedmodel.keras.src.models.model.Model.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Export the model as an artifact for inference. — export_savedmodel.keras.src.models.model.Model","text":"export ONNX inference.","code":"# Create the artifact model |> tensorflow::export_savedmodel(\"path/to/location\")  # Later, in a different process/environment... library(tensorflow) reloaded_artifact <- tf$saved_model$load(\"path/to/location\") predictions <- reloaded_artifact$serve(input_data)  # see tfdeploy::serve_savedmodel() for serving a model over a local web api. # Export the model as a ONNX artifact model |> export_savedmodel(\"path/to/location\", format = \"onnx\")  # Load the artifact in a different process/environment onnxruntime <- reticulate::import(\"onnxruntime\") ort_session <- onnxruntime$InferenceSession(\"path/to/location\") input_data <- list(....) names(input_data) <- sapply(ort_session$get_inputs(), `[[`, \"name\") predictions <- ort_session$run(NULL, input_data)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/fit.keras.src.models.model.Model.html","id":null,"dir":"Reference","previous_headings":"","what":"Train a model for a fixed number of epochs (dataset iterations). — fit.keras.src.models.model.Model","title":"Train a model for a fixed number of epochs (dataset iterations). — fit.keras.src.models.model.Model","text":"Train model fixed number epochs (dataset iterations).","code":""},{"path":"https://keras3.posit.co/dev/reference/fit.keras.src.models.model.Model.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Train a model for a fixed number of epochs (dataset iterations). — fit.keras.src.models.model.Model","text":"","code":"# S3 method for class 'keras.src.models.model.Model' fit(   object,   x = NULL,   y = NULL,   ...,   batch_size = NULL,   epochs = 1L,   callbacks = NULL,   validation_split = 0,   validation_data = NULL,   shuffle = TRUE,   class_weight = NULL,   sample_weight = NULL,   initial_epoch = 1L,   steps_per_epoch = NULL,   validation_steps = NULL,   validation_batch_size = NULL,   validation_freq = 1L,   verbose = getOption(\"keras.verbose\", default = \"auto\"),   view_metrics = getOption(\"keras.view_metrics\", default = \"auto\") )"},{"path":"https://keras3.posit.co/dev/reference/fit.keras.src.models.model.Model.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Train a model for a fixed number of epochs (dataset iterations). — fit.keras.src.models.model.Model","text":"object Keras model object x Input data. can : array (array-like), list arrays (case model multiple inputs). backend-native tensor, list tensors (case model multiple inputs). named list mapping input names corresponding array/tensors, model named inputs. tf.data.Dataset. return tuple either (inputs, targets) (inputs, targets, sample_weights). generator returning (inputs, targets) (inputs, targets, sample_weights). y Target data. Like input data x, can either array(s) backend-native tensor(s). x TF Dataset generator, y specified (since targets obtained x). ... Additional arguments passed model fit() method. batch_size Integer NULL. Number samples per gradient update. unspecified, batch_size default 32. specify batch_size data form TF Datasets generators, (since generate batches). epochs Integer. Number epochs train model. epoch iteration entire x y data provided (unless steps_per_epoch flag set something NULL). Note conjunction initial_epoch, epochs understood \"final epoch\". model trained number iterations given epochs, merely epoch index epochs reached. callbacks List Callback() instances. List callbacks apply training. See callback_*. validation_split Fraction training data used validation data. model set apart fraction training data, train , evaluate loss model metrics data end epoch. validation data selected last samples x y data provided, shuffling. argument supported x y made arrays tensors. validation_data validation_split provided, validation_data override validation_split. validation_data Data evaluate loss model metrics end epoch. model trained data. Thus, note fact validation loss data provided using validation_split validation_data affected regularization layers like noise dropout. validation_data override validation_split. can : tuple (x_val, y_val) arrays tensors. tuple (x_val, y_val, val_sample_weights) arrays. generator returning (inputs, targets) (inputs, targets, sample_weights). shuffle Boolean, whether shuffle training data epoch. argument ignored x generator TF Dataset. class_weight Optional named list mapping class indices (integers, 0-based) weight (float) value, used weighting loss function (training ). can useful tell model \"pay attention\" samples -represented class. class_weight specified targets rank 2 greater, either y must one-hot encoded, explicit final dimension 1 must included sparse class labels. sample_weight Optional array tensor weights training samples, used weighting loss function (training ). can either pass flat (1D) array tensor length input samples (1:1 mapping weights samples), case temporal data, can pass 2D array tensor shape (samples, sequence_length) apply different weight every timestep every sample. argument supported x tf.data.Dataset, Python generator function. Instead, provide sample_weights third element x. Note sample weighting apply metrics specified via metrics argument compile(). apply sample weighting metrics, can specify via weighted_metrics compile() instead. initial_epoch Integer. Epoch start training (useful resuming previous training run). steps_per_epoch Integer NULL. Total number steps (batches samples) declaring one epoch finished starting next epoch. training input tensors arrays, default NULL means value used number samples dataset divided batch size, 1 determined. x tf.data.Dataset Python generator function, epoch run input dataset exhausted. passing infinitely repeating dataset, must specify steps_per_epoch argument, otherwise training run indefinitely. validation_steps relevant validation_data provided. Total number steps (batches samples) draw stopping performing validation end every epoch. validation_steps NULL, validation run validation_data dataset exhausted. case infinitely repeating dataset, run indefinitely. validation_steps specified part dataset consumed, evaluation start beginning dataset epoch. ensures validation samples used every time. validation_batch_size Integer NULL. Number samples per validation batch. unspecified, default batch_size. specify validation_batch_size data form TF Datasets generator instances (since generate batches). validation_freq relevant validation data provided. Specifies many training epochs run new validation run performed, e.g. validation_freq=2 runs validation every 2 epochs. verbose \"auto\", 0, 1, 2. Verbosity mode. 0 = silent, 1 = progress bar, 2 = one line per epoch. \"auto\" becomes 1 cases, 2 knitr render running distributed training server. Note progress bar particularly useful logged file, verbose=2 recommended running interactively (e.g., production environment). Defaults \"auto\". view_metrics View realtime plot training metrics (epoch). default (\"auto\") display plot running within RStudio, metrics specified model compile(), epochs > 1 verbose > 0. Set global options(keras.view_metrics = ) option establish different default.","code":""},{"path":"https://keras3.posit.co/dev/reference/fit.keras.src.models.model.Model.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Train a model for a fixed number of epochs (dataset iterations). — fit.keras.src.models.model.Model","text":"keras_training_history object, named list: list(params = <params>, metrics = <metrics>\"), S3 methods print(), plot(), .data.frame(). metrics field record training loss values metrics values successive epochs, well validation loss values validation metrics values (applicable).","code":""},{"path":"https://keras3.posit.co/dev/reference/fit.keras.src.models.model.Model.html","id":"details","dir":"Reference","previous_headings":"","what":"Details","title":"Train a model for a fixed number of epochs (dataset iterations). — fit.keras.src.models.model.Model","text":"Unpacking behavior iterator-like inputs: common pattern pass iterator like object tf.data.Dataset generator fit(), fact yield features (x) optionally targets (y) sample weights (sample_weight). Keras requires output iterator-likes unambiguous. iterator return tuple() length 1, 2, 3, optional second third elements used y sample_weight respectively. type provided wrapped length-one tuple(), effectively treating everything x. yielding named lists, still adhere top-level tuple structure, e.g. tuple(list(x0 = x0, x = x1), y). Keras attempt separate features, targets, weights keys single dict.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/freeze_weights.html","id":null,"dir":"Reference","previous_headings":"","what":"Freeze and unfreeze weights — freeze_weights","title":"Freeze and unfreeze weights — freeze_weights","text":"Freeze weights model layer longer trainable.","code":""},{"path":"https://keras3.posit.co/dev/reference/freeze_weights.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Freeze and unfreeze weights — freeze_weights","text":"","code":"freeze_weights(object, from = NULL, to = NULL, which = NULL)  unfreeze_weights(object, from = NULL, to = NULL, which = NULL)"},{"path":"https://keras3.posit.co/dev/reference/freeze_weights.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Freeze and unfreeze weights — freeze_weights","text":"object Keras model layer object Layer instance, layer name, layer index within model Layer instance, layer name, layer index within model layer names, integer positions, layers, logical vector (length(object$layers)), function returning logical vector.","code":""},{"path":"https://keras3.posit.co/dev/reference/freeze_weights.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Freeze and unfreeze weights — freeze_weights","text":"input object frozen weights returned, invisibly. Note, object modified place, return value provided make usage pipe convenient.","code":""},{"path":"https://keras3.posit.co/dev/reference/freeze_weights.html","id":"note","dir":"Reference","previous_headings":"","what":"Note","title":"Freeze and unfreeze weights — freeze_weights","text":"layer arguments inclusive. applied model, freeze unfreeze global operation layers model (.e. layers within specified range set opposite value, e.g. unfrozen call freeze). Models must compiled weights frozen unfrozen.","code":""},{"path":"https://keras3.posit.co/dev/reference/freeze_weights.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Freeze and unfreeze weights — freeze_weights","text":"","code":"# instantiate a VGG16 model conv_base <- application_vgg16(   weights = \"imagenet\",   include_top = FALSE,   input_shape = c(150, 150, 3) )  # freeze it's weights freeze_weights(conv_base)  # Note the \"Trainable\" column conv_base ## Model: \"vgg16\" ## +-----------------------------+-----------------------+------------+-------+ ## | Layer (type)                | Output Shape          |    Param # | Trai… | ## +=============================+=======================+============+=======+ ## | input_layer (InputLayer)    | (None, 150, 150, 3)   |          0 |   -   | ## +-----------------------------+-----------------------+------------+-------+ ## | block1_conv1 (Conv2D)       | (None, 150, 150, 64)  |      1,792 |   N   | ## +-----------------------------+-----------------------+------------+-------+ ## | block1_conv2 (Conv2D)       | (None, 150, 150, 64)  |     36,928 |   N   | ## +-----------------------------+-----------------------+------------+-------+ ## | block1_pool (MaxPooling2D)  | (None, 75, 75, 64)    |          0 |   -   | ## +-----------------------------+-----------------------+------------+-------+ ## | block2_conv1 (Conv2D)       | (None, 75, 75, 128)   |     73,856 |   N   | ## +-----------------------------+-----------------------+------------+-------+ ## | block2_conv2 (Conv2D)       | (None, 75, 75, 128)   |    147,584 |   N   | ## +-----------------------------+-----------------------+------------+-------+ ## | block2_pool (MaxPooling2D)  | (None, 37, 37, 128)   |          0 |   -   | ## +-----------------------------+-----------------------+------------+-------+ ## | block3_conv1 (Conv2D)       | (None, 37, 37, 256)   |    295,168 |   N   | ## +-----------------------------+-----------------------+------------+-------+ ## | block3_conv2 (Conv2D)       | (None, 37, 37, 256)   |    590,080 |   N   | ## +-----------------------------+-----------------------+------------+-------+ ## | block3_conv3 (Conv2D)       | (None, 37, 37, 256)   |    590,080 |   N   | ## +-----------------------------+-----------------------+------------+-------+ ## | block3_pool (MaxPooling2D)  | (None, 18, 18, 256)   |          0 |   -   | ## +-----------------------------+-----------------------+------------+-------+ ## | block4_conv1 (Conv2D)       | (None, 18, 18, 512)   |  1,180,160 |   N   | ## +-----------------------------+-----------------------+------------+-------+ ## | block4_conv2 (Conv2D)       | (None, 18, 18, 512)   |  2,359,808 |   N   | ## +-----------------------------+-----------------------+------------+-------+ ## | block4_conv3 (Conv2D)       | (None, 18, 18, 512)   |  2,359,808 |   N   | ## +-----------------------------+-----------------------+------------+-------+ ## | block4_pool (MaxPooling2D)  | (None, 9, 9, 512)     |          0 |   -   | ## +-----------------------------+-----------------------+------------+-------+ ## | block5_conv1 (Conv2D)       | (None, 9, 9, 512)     |  2,359,808 |   N   | ## +-----------------------------+-----------------------+------------+-------+ ## | block5_conv2 (Conv2D)       | (None, 9, 9, 512)     |  2,359,808 |   N   | ## +-----------------------------+-----------------------+------------+-------+ ## | block5_conv3 (Conv2D)       | (None, 9, 9, 512)     |  2,359,808 |   N   | ## +-----------------------------+-----------------------+------------+-------+ ## | block5_pool (MaxPooling2D)  | (None, 4, 4, 512)     |          0 |   -   | ## +-----------------------------+-----------------------+------------+-------+ ##  Total params: 14,714,688 (56.13 MB) ##  Trainable params: 0 (0.00 B) ##  Non-trainable params: 14,714,688 (56.13 MB) # create a composite model that includes the base + more layers model <- keras_model_sequential(input_batch_shape = shape(conv_base$input)) |>   conv_base() |>   layer_flatten() |>   layer_dense(units = 256, activation = \"relu\") |>   layer_dense(units = 1, activation = \"sigmoid\")  # compile model |> compile(   loss = \"binary_crossentropy\",   optimizer = optimizer_rmsprop(learning_rate = 2e-5),   metrics = c(\"accuracy\") )  model ## Model: \"sequential\" ## +-----------------------------+-----------------------+------------+-------+ ## | Layer (type)                | Output Shape          |    Param # | Trai… | ## +=============================+=======================+============+=======+ ## | vgg16 (Functional)          | (None, 4, 4, 512)     | 14,714,688 |   N   | ## +-----------------------------+-----------------------+------------+-------+ ## | flatten (Flatten)           | (None, 8192)          |          0 |   -   | ## +-----------------------------+-----------------------+------------+-------+ ## | dense (Dense)               | (None, 256)           |  2,097,408 |   Y   | ## +-----------------------------+-----------------------+------------+-------+ ## | dense_1 (Dense)             | (None, 1)             |        257 |   Y   | ## +-----------------------------+-----------------------+------------+-------+ ##  Total params: 16,812,353 (64.13 MB) ##  Trainable params: 2,097,665 (8.00 MB) ##  Non-trainable params: 14,714,688 (56.13 MB) print(model, expand_nested = TRUE) ## Model: \"sequential\" ## +-----------------------------+-----------------------+------------+-------+ ## | Layer (type)                | Output Shape          |    Param # | Trai… | ## +=============================+=======================+============+=======+ ## | vgg16 (Functional)          | (None, 4, 4, 512)     | 14,714,688 |   N   | ## +-----------------------------+-----------------------+------------+-------+ ## |    > input_layer            | (None, 150, 150, 3)   |          0 |   -   | ## | (InputLayer)                |                       |            |       | ## +-----------------------------+-----------------------+------------+-------+ ## |    > block1_conv1 (Conv2D)  | (None, 150, 150, 64)  |      1,792 |   N   | ## +-----------------------------+-----------------------+------------+-------+ ## |    > block1_conv2 (Conv2D)  | (None, 150, 150, 64)  |     36,928 |   N   | ## +-----------------------------+-----------------------+------------+-------+ ## |    > block1_pool            | (None, 75, 75, 64)    |          0 |   -   | ## | (MaxPooling2D)              |                       |            |       | ## +-----------------------------+-----------------------+------------+-------+ ## |    > block2_conv1 (Conv2D)  | (None, 75, 75, 128)   |     73,856 |   N   | ## +-----------------------------+-----------------------+------------+-------+ ## |    > block2_conv2 (Conv2D)  | (None, 75, 75, 128)   |    147,584 |   N   | ## +-----------------------------+-----------------------+------------+-------+ ## |    > block2_pool            | (None, 37, 37, 128)   |          0 |   -   | ## | (MaxPooling2D)              |                       |            |       | ## +-----------------------------+-----------------------+------------+-------+ ## |    > block3_conv1 (Conv2D)  | (None, 37, 37, 256)   |    295,168 |   N   | ## +-----------------------------+-----------------------+------------+-------+ ## |    > block3_conv2 (Conv2D)  | (None, 37, 37, 256)   |    590,080 |   N   | ## +-----------------------------+-----------------------+------------+-------+ ## |    > block3_conv3 (Conv2D)  | (None, 37, 37, 256)   |    590,080 |   N   | ## +-----------------------------+-----------------------+------------+-------+ ## |    > block3_pool            | (None, 18, 18, 256)   |          0 |   -   | ## | (MaxPooling2D)              |                       |            |       | ## +-----------------------------+-----------------------+------------+-------+ ## |    > block4_conv1 (Conv2D)  | (None, 18, 18, 512)   |  1,180,160 |   N   | ## +-----------------------------+-----------------------+------------+-------+ ## |    > block4_conv2 (Conv2D)  | (None, 18, 18, 512)   |  2,359,808 |   N   | ## +-----------------------------+-----------------------+------------+-------+ ## |    > block4_conv3 (Conv2D)  | (None, 18, 18, 512)   |  2,359,808 |   N   | ## +-----------------------------+-----------------------+------------+-------+ ## |    > block4_pool            | (None, 9, 9, 512)     |          0 |   -   | ## | (MaxPooling2D)              |                       |            |       | ## +-----------------------------+-----------------------+------------+-------+ ## |    > block5_conv1 (Conv2D)  | (None, 9, 9, 512)     |  2,359,808 |   N   | ## +-----------------------------+-----------------------+------------+-------+ ## |    > block5_conv2 (Conv2D)  | (None, 9, 9, 512)     |  2,359,808 |   N   | ## +-----------------------------+-----------------------+------------+-------+ ## |    > block5_conv3 (Conv2D)  | (None, 9, 9, 512)     |  2,359,808 |   N   | ## +-----------------------------+-----------------------+------------+-------+ ## |    > block5_pool            | (None, 4, 4, 512)     |          0 |   -   | ## | (MaxPooling2D)              |                       |            |       | ## +-----------------------------+-----------------------+------------+-------+ ## | flatten (Flatten)           | (None, 8192)          |          0 |   -   | ## +-----------------------------+-----------------------+------------+-------+ ## | dense (Dense)               | (None, 256)           |  2,097,408 |   Y   | ## +-----------------------------+-----------------------+------------+-------+ ## | dense_1 (Dense)             | (None, 1)             |        257 |   Y   | ## +-----------------------------+-----------------------+------------+-------+ ##  Total params: 16,812,353 (64.13 MB) ##  Trainable params: 2,097,665 (8.00 MB) ##  Non-trainable params: 14,714,688 (56.13 MB) # unfreeze weights from \"block5_conv1\" on unfreeze_weights(conv_base, from = \"block5_conv1\")  # compile again since we froze or unfroze weights model |> compile(   loss = \"binary_crossentropy\",   optimizer = optimizer_rmsprop(learning_rate = 2e-5),   metrics = c(\"accuracy\") )  conv_base ## Model: \"vgg16\" ## +-----------------------------+-----------------------+------------+-------+ ## | Layer (type)                | Output Shape          |    Param # | Trai… | ## +=============================+=======================+============+=======+ ## | input_layer (InputLayer)    | (None, 150, 150, 3)   |          0 |   -   | ## +-----------------------------+-----------------------+------------+-------+ ## | block1_conv1 (Conv2D)       | (None, 150, 150, 64)  |      1,792 |   N   | ## +-----------------------------+-----------------------+------------+-------+ ## | block1_conv2 (Conv2D)       | (None, 150, 150, 64)  |     36,928 |   N   | ## +-----------------------------+-----------------------+------------+-------+ ## | block1_pool (MaxPooling2D)  | (None, 75, 75, 64)    |          0 |   -   | ## +-----------------------------+-----------------------+------------+-------+ ## | block2_conv1 (Conv2D)       | (None, 75, 75, 128)   |     73,856 |   N   | ## +-----------------------------+-----------------------+------------+-------+ ## | block2_conv2 (Conv2D)       | (None, 75, 75, 128)   |    147,584 |   N   | ## +-----------------------------+-----------------------+------------+-------+ ## | block2_pool (MaxPooling2D)  | (None, 37, 37, 128)   |          0 |   -   | ## +-----------------------------+-----------------------+------------+-------+ ## | block3_conv1 (Conv2D)       | (None, 37, 37, 256)   |    295,168 |   N   | ## +-----------------------------+-----------------------+------------+-------+ ## | block3_conv2 (Conv2D)       | (None, 37, 37, 256)   |    590,080 |   N   | ## +-----------------------------+-----------------------+------------+-------+ ## | block3_conv3 (Conv2D)       | (None, 37, 37, 256)   |    590,080 |   N   | ## +-----------------------------+-----------------------+------------+-------+ ## | block3_pool (MaxPooling2D)  | (None, 18, 18, 256)   |          0 |   -   | ## +-----------------------------+-----------------------+------------+-------+ ## | block4_conv1 (Conv2D)       | (None, 18, 18, 512)   |  1,180,160 |   N   | ## +-----------------------------+-----------------------+------------+-------+ ## | block4_conv2 (Conv2D)       | (None, 18, 18, 512)   |  2,359,808 |   N   | ## +-----------------------------+-----------------------+------------+-------+ ## | block4_conv3 (Conv2D)       | (None, 18, 18, 512)   |  2,359,808 |   N   | ## +-----------------------------+-----------------------+------------+-------+ ## | block4_pool (MaxPooling2D)  | (None, 9, 9, 512)     |          0 |   -   | ## +-----------------------------+-----------------------+------------+-------+ ## | block5_conv1 (Conv2D)       | (None, 9, 9, 512)     |  2,359,808 |   Y   | ## +-----------------------------+-----------------------+------------+-------+ ## | block5_conv2 (Conv2D)       | (None, 9, 9, 512)     |  2,359,808 |   Y   | ## +-----------------------------+-----------------------+------------+-------+ ## | block5_conv3 (Conv2D)       | (None, 9, 9, 512)     |  2,359,808 |   Y   | ## +-----------------------------+-----------------------+------------+-------+ ## | block5_pool (MaxPooling2D)  | (None, 4, 4, 512)     |          0 |   -   | ## +-----------------------------+-----------------------+------------+-------+ ##  Total params: 14,714,688 (56.13 MB) ##  Trainable params: 7,079,424 (27.01 MB) ##  Non-trainable params: 7,635,264 (29.13 MB) print(model, expand_nested = TRUE) ## Model: \"sequential\" ## +-----------------------------+-----------------------+------------+-------+ ## | Layer (type)                | Output Shape          |    Param # | Trai… | ## +=============================+=======================+============+=======+ ## | vgg16 (Functional)          | (None, 4, 4, 512)     | 14,714,688 |   Y   | ## +-----------------------------+-----------------------+------------+-------+ ## |    > input_layer            | (None, 150, 150, 3)   |          0 |   -   | ## | (InputLayer)                |                       |            |       | ## +-----------------------------+-----------------------+------------+-------+ ## |    > block1_conv1 (Conv2D)  | (None, 150, 150, 64)  |      1,792 |   N   | ## +-----------------------------+-----------------------+------------+-------+ ## |    > block1_conv2 (Conv2D)  | (None, 150, 150, 64)  |     36,928 |   N   | ## +-----------------------------+-----------------------+------------+-------+ ## |    > block1_pool            | (None, 75, 75, 64)    |          0 |   -   | ## | (MaxPooling2D)              |                       |            |       | ## +-----------------------------+-----------------------+------------+-------+ ## |    > block2_conv1 (Conv2D)  | (None, 75, 75, 128)   |     73,856 |   N   | ## +-----------------------------+-----------------------+------------+-------+ ## |    > block2_conv2 (Conv2D)  | (None, 75, 75, 128)   |    147,584 |   N   | ## +-----------------------------+-----------------------+------------+-------+ ## |    > block2_pool            | (None, 37, 37, 128)   |          0 |   -   | ## | (MaxPooling2D)              |                       |            |       | ## +-----------------------------+-----------------------+------------+-------+ ## |    > block3_conv1 (Conv2D)  | (None, 37, 37, 256)   |    295,168 |   N   | ## +-----------------------------+-----------------------+------------+-------+ ## |    > block3_conv2 (Conv2D)  | (None, 37, 37, 256)   |    590,080 |   N   | ## +-----------------------------+-----------------------+------------+-------+ ## |    > block3_conv3 (Conv2D)  | (None, 37, 37, 256)   |    590,080 |   N   | ## +-----------------------------+-----------------------+------------+-------+ ## |    > block3_pool            | (None, 18, 18, 256)   |          0 |   -   | ## | (MaxPooling2D)              |                       |            |       | ## +-----------------------------+-----------------------+------------+-------+ ## |    > block4_conv1 (Conv2D)  | (None, 18, 18, 512)   |  1,180,160 |   N   | ## +-----------------------------+-----------------------+------------+-------+ ## |    > block4_conv2 (Conv2D)  | (None, 18, 18, 512)   |  2,359,808 |   N   | ## +-----------------------------+-----------------------+------------+-------+ ## |    > block4_conv3 (Conv2D)  | (None, 18, 18, 512)   |  2,359,808 |   N   | ## +-----------------------------+-----------------------+------------+-------+ ## |    > block4_pool            | (None, 9, 9, 512)     |          0 |   -   | ## | (MaxPooling2D)              |                       |            |       | ## +-----------------------------+-----------------------+------------+-------+ ## |    > block5_conv1 (Conv2D)  | (None, 9, 9, 512)     |  2,359,808 |   Y   | ## +-----------------------------+-----------------------+------------+-------+ ## |    > block5_conv2 (Conv2D)  | (None, 9, 9, 512)     |  2,359,808 |   Y   | ## +-----------------------------+-----------------------+------------+-------+ ## |    > block5_conv3 (Conv2D)  | (None, 9, 9, 512)     |  2,359,808 |   Y   | ## +-----------------------------+-----------------------+------------+-------+ ## |    > block5_pool            | (None, 4, 4, 512)     |          0 |   -   | ## | (MaxPooling2D)              |                       |            |       | ## +-----------------------------+-----------------------+------------+-------+ ## | flatten (Flatten)           | (None, 8192)          |          0 |   -   | ## +-----------------------------+-----------------------+------------+-------+ ## | dense (Dense)               | (None, 256)           |  2,097,408 |   Y   | ## +-----------------------------+-----------------------+------------+-------+ ## | dense_1 (Dense)             | (None, 1)             |        257 |   Y   | ## +-----------------------------+-----------------------+------------+-------+ ##  Total params: 16,812,353 (64.13 MB) ##  Trainable params: 9,177,089 (35.01 MB) ##  Non-trainable params: 7,635,264 (29.13 MB) # freeze only the last 5 layers freeze_weights(conv_base, from = -5) conv_base ## Model: \"vgg16\" ## +-----------------------------+-----------------------+------------+-------+ ## | Layer (type)                | Output Shape          |    Param # | Trai… | ## +=============================+=======================+============+=======+ ## | input_layer (InputLayer)    | (None, 150, 150, 3)   |          0 |   -   | ## +-----------------------------+-----------------------+------------+-------+ ## | block1_conv1 (Conv2D)       | (None, 150, 150, 64)  |      1,792 |   Y   | ## +-----------------------------+-----------------------+------------+-------+ ## | block1_conv2 (Conv2D)       | (None, 150, 150, 64)  |     36,928 |   Y   | ## +-----------------------------+-----------------------+------------+-------+ ## | block1_pool (MaxPooling2D)  | (None, 75, 75, 64)    |          0 |   -   | ## +-----------------------------+-----------------------+------------+-------+ ## | block2_conv1 (Conv2D)       | (None, 75, 75, 128)   |     73,856 |   Y   | ## +-----------------------------+-----------------------+------------+-------+ ## | block2_conv2 (Conv2D)       | (None, 75, 75, 128)   |    147,584 |   Y   | ## +-----------------------------+-----------------------+------------+-------+ ## | block2_pool (MaxPooling2D)  | (None, 37, 37, 128)   |          0 |   -   | ## +-----------------------------+-----------------------+------------+-------+ ## | block3_conv1 (Conv2D)       | (None, 37, 37, 256)   |    295,168 |   Y   | ## +-----------------------------+-----------------------+------------+-------+ ## | block3_conv2 (Conv2D)       | (None, 37, 37, 256)   |    590,080 |   Y   | ## +-----------------------------+-----------------------+------------+-------+ ## | block3_conv3 (Conv2D)       | (None, 37, 37, 256)   |    590,080 |   Y   | ## +-----------------------------+-----------------------+------------+-------+ ## | block3_pool (MaxPooling2D)  | (None, 18, 18, 256)   |          0 |   -   | ## +-----------------------------+-----------------------+------------+-------+ ## | block4_conv1 (Conv2D)       | (None, 18, 18, 512)   |  1,180,160 |   Y   | ## +-----------------------------+-----------------------+------------+-------+ ## | block4_conv2 (Conv2D)       | (None, 18, 18, 512)   |  2,359,808 |   Y   | ## +-----------------------------+-----------------------+------------+-------+ ## | block4_conv3 (Conv2D)       | (None, 18, 18, 512)   |  2,359,808 |   Y   | ## +-----------------------------+-----------------------+------------+-------+ ## | block4_pool (MaxPooling2D)  | (None, 9, 9, 512)     |          0 |   -   | ## +-----------------------------+-----------------------+------------+-------+ ## | block5_conv1 (Conv2D)       | (None, 9, 9, 512)     |  2,359,808 |   N   | ## +-----------------------------+-----------------------+------------+-------+ ## | block5_conv2 (Conv2D)       | (None, 9, 9, 512)     |  2,359,808 |   N   | ## +-----------------------------+-----------------------+------------+-------+ ## | block5_conv3 (Conv2D)       | (None, 9, 9, 512)     |  2,359,808 |   N   | ## +-----------------------------+-----------------------+------------+-------+ ## | block5_pool (MaxPooling2D)  | (None, 4, 4, 512)     |          0 |   -   | ## +-----------------------------+-----------------------+------------+-------+ ##  Total params: 14,714,688 (56.13 MB) ##  Trainable params: 7,635,264 (29.13 MB) ##  Non-trainable params: 7,079,424 (27.01 MB) # freeze only the last 5 layers, a different way unfreeze_weights(conv_base, to = -6) conv_base ## Model: \"vgg16\" ## +-----------------------------+-----------------------+------------+-------+ ## | Layer (type)                | Output Shape          |    Param # | Trai… | ## +=============================+=======================+============+=======+ ## | input_layer (InputLayer)    | (None, 150, 150, 3)   |          0 |   -   | ## +-----------------------------+-----------------------+------------+-------+ ## | block1_conv1 (Conv2D)       | (None, 150, 150, 64)  |      1,792 |   Y   | ## +-----------------------------+-----------------------+------------+-------+ ## | block1_conv2 (Conv2D)       | (None, 150, 150, 64)  |     36,928 |   Y   | ## +-----------------------------+-----------------------+------------+-------+ ## | block1_pool (MaxPooling2D)  | (None, 75, 75, 64)    |          0 |   -   | ## +-----------------------------+-----------------------+------------+-------+ ## | block2_conv1 (Conv2D)       | (None, 75, 75, 128)   |     73,856 |   Y   | ## +-----------------------------+-----------------------+------------+-------+ ## | block2_conv2 (Conv2D)       | (None, 75, 75, 128)   |    147,584 |   Y   | ## +-----------------------------+-----------------------+------------+-------+ ## | block2_pool (MaxPooling2D)  | (None, 37, 37, 128)   |          0 |   -   | ## +-----------------------------+-----------------------+------------+-------+ ## | block3_conv1 (Conv2D)       | (None, 37, 37, 256)   |    295,168 |   Y   | ## +-----------------------------+-----------------------+------------+-------+ ## | block3_conv2 (Conv2D)       | (None, 37, 37, 256)   |    590,080 |   Y   | ## +-----------------------------+-----------------------+------------+-------+ ## | block3_conv3 (Conv2D)       | (None, 37, 37, 256)   |    590,080 |   Y   | ## +-----------------------------+-----------------------+------------+-------+ ## | block3_pool (MaxPooling2D)  | (None, 18, 18, 256)   |          0 |   -   | ## +-----------------------------+-----------------------+------------+-------+ ## | block4_conv1 (Conv2D)       | (None, 18, 18, 512)   |  1,180,160 |   Y   | ## +-----------------------------+-----------------------+------------+-------+ ## | block4_conv2 (Conv2D)       | (None, 18, 18, 512)   |  2,359,808 |   Y   | ## +-----------------------------+-----------------------+------------+-------+ ## | block4_conv3 (Conv2D)       | (None, 18, 18, 512)   |  2,359,808 |   Y   | ## +-----------------------------+-----------------------+------------+-------+ ## | block4_pool (MaxPooling2D)  | (None, 9, 9, 512)     |          0 |   -   | ## +-----------------------------+-----------------------+------------+-------+ ## | block5_conv1 (Conv2D)       | (None, 9, 9, 512)     |  2,359,808 |   N   | ## +-----------------------------+-----------------------+------------+-------+ ## | block5_conv2 (Conv2D)       | (None, 9, 9, 512)     |  2,359,808 |   N   | ## +-----------------------------+-----------------------+------------+-------+ ## | block5_conv3 (Conv2D)       | (None, 9, 9, 512)     |  2,359,808 |   N   | ## +-----------------------------+-----------------------+------------+-------+ ## | block5_pool (MaxPooling2D)  | (None, 4, 4, 512)     |          0 |   -   | ## +-----------------------------+-----------------------+------------+-------+ ##  Total params: 14,714,688 (56.13 MB) ##  Trainable params: 7,635,264 (29.13 MB) ##  Non-trainable params: 7,079,424 (27.01 MB) # Freeze only layers of a certain type, e.g, BatchNorm layers # batch_norm_layer_class_name <- class(layer_batch_normalization())[1] # is_batch_norm_layer <- function(x) inherits(x, batch_norm_layer_class_name) is_batch_norm_layer <- function(x) inherits(x, keras$layers$BatchNormalization)  model <- application_efficientnet_v2b0() freeze_weights(model, which = is_batch_norm_layer) # print(model)  # equivalent to: for(layer in model$layers) {   if(is_batch_norm_layer(layer))     layer$trainable <- FALSE   else     layer$trainable <- TRUE }"},{"path":"https://keras3.posit.co/dev/reference/get_config.html","id":null,"dir":"Reference","previous_headings":"","what":"Layer/Model configuration — get_config","title":"Layer/Model configuration — get_config","text":"layer config object returned get_config() contains configuration layer model. layer model can reinstantiated later (without trained weights) configuration using from_config(). config include connectivity information, class name (handled externally).","code":""},{"path":"https://keras3.posit.co/dev/reference/get_config.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Layer/Model configuration — get_config","text":"","code":"get_config(object)  from_config(config, custom_objects = NULL)"},{"path":"https://keras3.posit.co/dev/reference/get_config.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Layer/Model configuration — get_config","text":"object Layer model object config Object layer model configuration custom_objects list custom objects needed instantiate layer, e.g., custom layers defined new_layer_class() similar.","code":""},{"path":"https://keras3.posit.co/dev/reference/get_config.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Layer/Model configuration — get_config","text":"get_config() returns object configuration, from_config() returns re-instantiation object.","code":""},{"path":"https://keras3.posit.co/dev/reference/get_config.html","id":"note","dir":"Reference","previous_headings":"","what":"Note","title":"Layer/Model configuration — get_config","text":"Objects returned get_config() serializable via RDS. want save restore model across sessions, can use save_model_config() (model configuration , weights) save_model() save model configuration weights filesystem.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/get_custom_objects.html","id":null,"dir":"Reference","previous_headings":"","what":"Get/set the currently registered custom objects. — get_custom_objects","title":"Get/set the currently registered custom objects. — get_custom_objects","text":"Custom objects set using custom_object_scope() added global list custom objects, appear returned list.","code":""},{"path":"https://keras3.posit.co/dev/reference/get_custom_objects.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Get/set the currently registered custom objects. — get_custom_objects","text":"","code":"get_custom_objects()  set_custom_objects(objects = named_list(), clear = TRUE)"},{"path":"https://keras3.posit.co/dev/reference/get_custom_objects.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Get/set the currently registered custom objects. — get_custom_objects","text":"objects named list custom objects, returned get_custom_objects() set_custom_objects(). clear bool, whether clear custom object registry populating objects.","code":""},{"path":"https://keras3.posit.co/dev/reference/get_custom_objects.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Get/set the currently registered custom objects. — get_custom_objects","text":"R named list mapping registered names registered objects. set_custom_objects() returns registry values updating, invisibly.","code":""},{"path":"https://keras3.posit.co/dev/reference/get_custom_objects.html","id":"note","dir":"Reference","previous_headings":"","what":"Note","title":"Get/set the currently registered custom objects. — get_custom_objects","text":"register_keras_serializable() preferred set_custom_objects() registering new objects.","code":""},{"path":"https://keras3.posit.co/dev/reference/get_custom_objects.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Get/set the currently registered custom objects. — get_custom_objects","text":"can use set_custom_objects() restore previous registry state.","code":"get_custom_objects() # within a function, if you want to temporarily modify the registry, function() {   orig_objects <- set_custom_objects(clear = TRUE)   on.exit(set_custom_objects(orig_objects))    ## temporarily modify the global registry   # register_keras_serializable(....)   # ....  <do work>   # on.exit(), the previous registry state is restored. }"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/get_file.html","id":null,"dir":"Reference","previous_headings":"","what":"Downloads a file from a URL if it not already in the cache. — get_file","title":"Downloads a file from a URL if it not already in the cache. — get_file","text":"default file url origin downloaded cache_dir ~/.keras, placed cache_subdir datasets, given filename fname. final location file example.txt therefore ~/.keras/datasets/example.txt. Files .tar, .tar.gz, .tar.bz, .zip formats can also extracted. Passing hash verify file download. command line programs shasum sha256sum can compute hash.","code":""},{"path":"https://keras3.posit.co/dev/reference/get_file.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Downloads a file from a URL if it not already in the cache. — get_file","text":"","code":"get_file(   fname = NULL,   origin = NULL,   ...,   file_hash = NULL,   cache_subdir = \"datasets\",   hash_algorithm = \"auto\",   extract = FALSE,   archive_format = \"auto\",   cache_dir = NULL,   force_download = FALSE )"},{"path":"https://keras3.posit.co/dev/reference/get_file.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Downloads a file from a URL if it not already in the cache. — get_file","text":"fname target single file, desired local name file. NULL, name file origin used. downloading extracting directory archive, provided fname used extraction directory name (extension). origin Original URL file. ... forward/backward compatability. file_hash expected hash string file download. sha256 md5 hash algorithms supported. cache_subdir Subdirectory Keras cache dir file saved. absolute path, e.g. \"/path//folder\" specified, file saved location. hash_algorithm Select hash algorithm verify file. options \"md5', \"sha256', \"auto'. default 'auto' detects hash algorithm use. extract TRUE, extracts archive. applicable compressed archive files like tar zip. archive_format Archive format try extracting file. Options \"auto', \"tar', \"zip', NULL. \"tar\" includes tar, tar.gz, tar.bz files. default \"auto\" corresponds c(\"tar\", \"zip\"). NULL empty list return matches found. cache_dir Location store cached files, NULL defaults Sys.getenv(\"KERAS_HOME\", \"~/.keras/\"). force_download TRUE, file always re-downloaded regardless cache state.","code":""},{"path":"https://keras3.posit.co/dev/reference/get_file.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Downloads a file from a URL if it not already in the cache. — get_file","text":"Path downloaded file. ** Warning malicious downloads ** Downloading something Internet carries risk. NEVER download file/archive trust source. recommend specify file_hash argument (hash source file known) make sure file getting one expect.","code":""},{"path":"https://keras3.posit.co/dev/reference/get_file.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Downloads a file from a URL if it not already in the cache. — get_file","text":"","code":"path_to_downloaded_file <- get_file(     \"flower_photos\",     origin = \"https://storage.googleapis.com/download.tensorflow.org/example_images/flower_photos.tgz\",     extract = TRUE )"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/get_layer.html","id":null,"dir":"Reference","previous_headings":"","what":"Retrieves a layer based on either its name (unique) or index. — get_layer","title":"Retrieves a layer based on either its name (unique) or index. — get_layer","text":"Indices based order horizontal graph traversal (bottom-) 1-based. name index provided, index take precedence.","code":""},{"path":"https://keras3.posit.co/dev/reference/get_layer.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Retrieves a layer based on either its name (unique) or index. — get_layer","text":"","code":"get_layer(object, name = NULL, index = NULL)"},{"path":"https://keras3.posit.co/dev/reference/get_layer.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Retrieves a layer based on either its name (unique) or index. — get_layer","text":"object Keras model object name String, name layer. index Integer, index layer (1-based). Also valid negative values, count end model.","code":""},{"path":"https://keras3.posit.co/dev/reference/get_layer.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Retrieves a layer based on either its name (unique) or index. — get_layer","text":"layer instance.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/get_registered_name.html","id":null,"dir":"Reference","previous_headings":"","what":"Returns the name registered to an object within the Keras framework. — get_registered_name","title":"Returns the name registered to an object within the Keras framework. — get_registered_name","text":"function part Keras serialization deserialization framework. maps objects string names associated objects serialization/deserialization.","code":""},{"path":"https://keras3.posit.co/dev/reference/get_registered_name.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Returns the name registered to an object within the Keras framework. — get_registered_name","text":"","code":"get_registered_name(obj)"},{"path":"https://keras3.posit.co/dev/reference/get_registered_name.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Returns the name registered to an object within the Keras framework. — get_registered_name","text":"obj object look .","code":""},{"path":"https://keras3.posit.co/dev/reference/get_registered_name.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Returns the name registered to an object within the Keras framework. — get_registered_name","text":"name associated object, default name object registered.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/get_registered_object.html","id":null,"dir":"Reference","previous_headings":"","what":"Returns the class associated with name if it is registered with Keras. — get_registered_object","title":"Returns the class associated with name if it is registered with Keras. — get_registered_object","text":"function part Keras serialization deserialization framework. maps strings objects associated serialization/deserialization.","code":""},{"path":"https://keras3.posit.co/dev/reference/get_registered_object.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Returns the class associated with name if it is registered with Keras. — get_registered_object","text":"","code":"get_registered_object(name, custom_objects = NULL, module_objects = NULL)"},{"path":"https://keras3.posit.co/dev/reference/get_registered_object.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Returns the class associated with name if it is registered with Keras. — get_registered_object","text":"name name look . custom_objects named list custom objects look name . Generally, custom_objects provided user. module_objects named list custom objects look name . Generally, module_objects provided midlevel library implementers.","code":""},{"path":"https://keras3.posit.co/dev/reference/get_registered_object.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Returns the class associated with name if it is registered with Keras. — get_registered_object","text":"instantiable class associated name, NULL class exists.","code":""},{"path":"https://keras3.posit.co/dev/reference/get_registered_object.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Returns the class associated with name if it is registered with Keras. — get_registered_object","text":"","code":"from_config <- function(cls, config, custom_objects = NULL) {   if ('my_custom_object_name' \\%in\\% names(config)) {     config$hidden_cls <- get_registered_object(       config$my_custom_object_name,       custom_objects = custom_objects)   } }"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/get_source_inputs.html","id":null,"dir":"Reference","previous_headings":"","what":"Returns the list of input tensors necessary to compute tensor. — get_source_inputs","title":"Returns the list of input tensors necessary to compute tensor. — get_source_inputs","text":"Output always list tensors (potentially 1 element).","code":""},{"path":"https://keras3.posit.co/dev/reference/get_source_inputs.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Returns the list of input tensors necessary to compute tensor. — get_source_inputs","text":"","code":"get_source_inputs(tensor)"},{"path":"https://keras3.posit.co/dev/reference/get_source_inputs.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Returns the list of input tensors necessary to compute tensor. — get_source_inputs","text":"tensor tensor start .","code":""},{"path":"https://keras3.posit.co/dev/reference/get_source_inputs.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Returns the list of input tensors necessary to compute tensor. — get_source_inputs","text":"List input tensors.","code":""},{"path":"https://keras3.posit.co/dev/reference/get_source_inputs.html","id":"example","dir":"Reference","previous_headings":"","what":"Example","title":"Returns the list of input tensors necessary to compute tensor. — get_source_inputs","text":"","code":"input <- keras_input(c(3)) output <- input |> layer_dense(4) |> op_multiply(5) reticulate::py_id(get_source_inputs(output)[[1]]) == reticulate::py_id(input) ## [1] TRUE"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/get_state_tree.html","id":null,"dir":"Reference","previous_headings":"","what":"Retrieves tree-like structure of model variables. — get_state_tree","title":"Retrieves tree-like structure of model variables. — get_state_tree","text":"method allows retrieval different model variables (trainable, non-trainable, optimizer, metrics). variables returned nested dictionary format, keys correspond variable names values nested representations variables.","code":""},{"path":"https://keras3.posit.co/dev/reference/get_state_tree.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Retrieves tree-like structure of model variables. — get_state_tree","text":"","code":"get_state_tree(object, value_format = \"backend_tensor\")"},{"path":"https://keras3.posit.co/dev/reference/get_state_tree.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Retrieves tree-like structure of model variables. — get_state_tree","text":"object Keras Model. value_format One \"backend_tensor\", \"numpy_array\", \"array\". kind array return leaves nested state tree.","code":""},{"path":"https://keras3.posit.co/dev/reference/get_state_tree.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Retrieves tree-like structure of model variables. — get_state_tree","text":"named list containing nested representations requested variables. names variable names, values corresponding nested named lists.","code":""},{"path":"https://keras3.posit.co/dev/reference/get_state_tree.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Retrieves tree-like structure of model variables. — get_state_tree","text":"state_tree list returned looks like:   example:","code":"model <- keras_model_sequential(name = \"my_sequential\",                                 input_shape = c(1),                                 input_name = \"my_input\") |>   layer_dense(1, activation = \"sigmoid\", name = \"my_dense\")  model |> compile(optimizer=\"adam\", loss=\"mse\", metrics=c(\"mae\")) model |> fit(matrix(1), matrix(1), verbose = 0) state_tree <- model |> get_state_tree() list(   metrics_variables = list(     loss = list(       count = ...,       total = ...     ),     mean_absolute_error = list(       count = ...,       total = ...     )   ),   trainable_variables = list(     my_sequential = list(       my_dense = list(         bias = ...,         kernel = ...       )     )   ),   non_trainable_variables = list(),   optimizer_variables = list(     adam = list(       iteration = ...,       learning_rate = ...,       my_sequential_my_dense_bias_momentum = ...,       my_sequential_my_dense_bias_velocity = ...,       my_sequential_my_dense_kernel_momentum = ...,       my_sequential_my_dense_kernel_velocity = ...     )   ) ) str(state_tree) ## List of 4 ##  $ trainable_variables    :List of 1 ##   ..$ my_sequential:List of 1 ##   .. ..$ my_dense:List of 2 ##   .. .. ..$ kernel:<tf.Variable 'my_sequential/my_dense/kernel:0' shape=(1, 1) dtype=float32, numpy=array([[-0.8338491]], dtype=float32)> ##   .. .. ..$ bias  :<tf.Variable 'my_sequential/my_dense/bias:0' shape=(1) dtype=float32, numpy=array([0.00099998], dtype=float32)> ##  $ non_trainable_variables: Named list() ##  $ optimizer_variables    :List of 1 ##   ..$ adam:List of 6 ##   .. ..$ iteration                             :<tf.Variable 'adam/iteration:0' shape=() dtype=int64, numpy=1> ##   .. ..$ learning_rate                         :<tf.Variable 'adam/learning_rate:0' shape=() dtype=float32, numpy=0.0010000000474974513> ##   .. ..$ my_sequential_my_dense_kernel_momentum:<tf.Variable 'adam/my_sequential_my_dense_kernel_momentum:0' shape=(1, 1) dtype=float32, numpy=array([[-0.02943518]], dtype=float32)> ##   .. ..$ my_sequential_my_dense_kernel_velocity:<tf.Variable 'adam/my_sequential_my_dense_kernel_velocity:0' shape=(1, 1) dtype=float32, numpy=array([[8.664299e-05]], dtype=float32)> ##   .. ..$ my_sequential_my_dense_bias_momentum  :<tf.Variable 'adam/my_sequential_my_dense_bias_momentum:0' shape=(1) dtype=float32, numpy=array([-0.02943518], dtype=float32)> ##   .. ..$ my_sequential_my_dense_bias_velocity  :<tf.Variable 'adam/my_sequential_my_dense_bias_velocity:0' shape=(1) dtype=float32, numpy=array([8.664299e-05], dtype=float32)> ##  $ metrics_variables      :List of 2 ##   ..$ loss               :List of 2 ##   .. ..$ total:<tf.Variable 'loss/total:0' shape=() dtype=float32, numpy=0.4863377809524536> ##   .. ..$ count:<tf.Variable 'loss/count:0' shape=() dtype=float32, numpy=1.0> ##   ..$ mean_absolute_error:List of 2 ##   .. ..$ total:<tf.Variable 'mean_absolute_error/total:0' shape=() dtype=float32, numpy=0.6973792314529419> ##   .. ..$ count:<tf.Variable 'mean_absolute_error_1/count:0' shape=() dtype=float32, numpy=1.0>"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/get_weights.html","id":null,"dir":"Reference","previous_headings":"","what":"Layer/Model weights as R arrays — get_weights","title":"Layer/Model weights as R arrays — get_weights","text":"Layer/Model weights R arrays","code":""},{"path":"https://keras3.posit.co/dev/reference/get_weights.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Layer/Model weights as R arrays — get_weights","text":"","code":"get_weights(object, trainable = NA)  set_weights(object, weights)"},{"path":"https://keras3.posit.co/dev/reference/get_weights.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Layer/Model weights as R arrays — get_weights","text":"object Layer model object trainable NA (default), weights returned. TRUE, weights trainable variables returned. FALSE, weights non-trainable variables returned. weights Weights R array","code":""},{"path":"https://keras3.posit.co/dev/reference/get_weights.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Layer/Model weights as R arrays — get_weights","text":"list R arrays.","code":""},{"path":"https://keras3.posit.co/dev/reference/get_weights.html","id":"note","dir":"Reference","previous_headings":"","what":"Note","title":"Layer/Model weights as R arrays — get_weights","text":"can access Layer/Model KerasVariables (also backend-native tensors like tf.Variable) object$weights, object$trainable_weights, object$non_trainable_weights","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/grapes-py_class-grapes.html","id":null,"dir":"Reference","previous_headings":"","what":"Make a python class constructor — %py_class%","title":"Make a python class constructor — %py_class%","text":"Make python class constructor","code":""},{"path":"https://keras3.posit.co/dev/reference/grapes-py_class-grapes.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Make a python class constructor — %py_class%","text":"","code":"spec %py_class% body"},{"path":"https://keras3.posit.co/dev/reference/grapes-py_class-grapes.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Make a python class constructor — %py_class%","text":"spec bare symbol MyClassName, call MyClassName(SuperClass) body expression can evaluated construct class methods.","code":""},{"path":"https://keras3.posit.co/dev/reference/grapes-py_class-grapes.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Make a python class constructor — %py_class%","text":"python class constructor, invisibly. Note, constructor also assigned parent frame.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/grapes-py_class-grapes.html","id":"ref-examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Make a python class constructor — %py_class%","text":"","code":"if (FALSE) { # \\dontrun{ MyClass %py_class% {   initialize <- function(x) {     print(\"Hi from MyClass$initialize()!\")     self$x <- x   }   my_method <- function() {     self$x   } }  my_class_instance <- MyClass(42) my_class_instance$my_method()  MyClass2(MyClass) %py_class% {   \"This will be a __doc__ string for MyClass2\"    initialize <- function(...) {     \"This will be the __doc__ string for the MyClass2.__init__() method\"     print(\"Hi from MyClass2$initialize()!\")     super$initialize(...)   } }  my_class_instance2 <- MyClass2(42) my_class_instance2$my_method()  reticulate::py_help(MyClass2) # see the __doc__ strings and more!  # In addition to `self`, there is also `private` available. # This is an R environment unique to each class instance, where you can # store objects that you don't want converted to Python, but still want # available from methods. You can also assign methods to private, and # `self` and `private` will be available in private methods.  MyClass %py_class% {    initialize <- function(x) {     print(\"Hi from MyClass$initialize()!\")     private$y <- paste(\"A Private field:\", x)   }    get_private_field <- function() {     private$y   }    private$a_private_method <- function() {     cat(\"a_private_method() was called.\\n\")     cat(\"private$y is \", sQuote(private$y), \"\\n\")   }    call_private_method <- function()     private$a_private_method()    # equivalent of @property decorator in python   an_active_property %<-active% function(x = NULL) {     if(!is.null(x)) {       cat(\"`an_active_property` was assigned\", x, \"\\n\")       return(x)     } else {       cat(\"`an_active_property` was accessed\\n\")       return(42)     }   } }  inst1 <- MyClass(1) inst2 <- MyClass(2) inst1$get_private_field() inst2$get_private_field() inst1$call_private_method() inst2$call_private_method() inst1$an_active_property inst1$an_active_property <- 11 } # }"},{"path":"https://keras3.posit.co/dev/reference/grapes-set-active-grapes.html","id":null,"dir":"Reference","previous_headings":"","what":"Make an Active Binding — %<-active%","title":"Make an Active Binding — %<-active%","text":"Make Active Binding","code":""},{"path":"https://keras3.posit.co/dev/reference/grapes-set-active-grapes.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Make an Active Binding — %<-active%","text":"","code":"sym %<-active% value"},{"path":"https://keras3.posit.co/dev/reference/grapes-set-active-grapes.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Make an Active Binding — %<-active%","text":"sym symbol bind value function call value sym accessed.","code":""},{"path":"https://keras3.posit.co/dev/reference/grapes-set-active-grapes.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Make an Active Binding — %<-active%","text":"value, invisibly","code":""},{"path":"https://keras3.posit.co/dev/reference/grapes-set-active-grapes.html","id":"details","dir":"Reference","previous_headings":"","what":"Details","title":"Make an Active Binding — %<-active%","text":"Active bindings defined %py_class% converted @property decorated methods.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/grapes-set-active-grapes.html","id":"ref-examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Make an Active Binding — %<-active%","text":"","code":"set.seed(1234) x %<-active% function(value) {   message(\"Evaluating function of active binding\")   if(missing(value))     runif(1)   else    message(\"Received: \", value) } x #> Evaluating function of active binding #> [1] 0.1137034 x #> Evaluating function of active binding #> [1] 0.6222994 x <- \"foo\" #> Evaluating function of active binding #> Received: foo x <- \"foo\" #> Evaluating function of active binding #> Received: foo x #> Evaluating function of active binding #> [1] 0.6092747 rm(x) # cleanup"},{"path":"https://keras3.posit.co/dev/reference/image_array_save.html","id":null,"dir":"Reference","previous_headings":"","what":"Saves an image stored as an array to a path or file object. — image_array_save","title":"Saves an image stored as an array to a path or file object. — image_array_save","text":"Saves image stored array path file object.","code":""},{"path":"https://keras3.posit.co/dev/reference/image_array_save.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Saves an image stored as an array to a path or file object. — image_array_save","text":"","code":"image_array_save(   x,   path,   data_format = NULL,   file_format = NULL,   scale = TRUE,   ... )"},{"path":"https://keras3.posit.co/dev/reference/image_array_save.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Saves an image stored as an array to a path or file object. — image_array_save","text":"x array. path Path file object. data_format Image data format, either \"channels_first\" \"channels_last\". file_format Optional file format override. omitted, format use determined filename extension. file object used instead filename, parameter always used. scale Whether rescale image values within [0, 255]. ... Additional keyword arguments passed PIL.Image.save().","code":""},{"path":"https://keras3.posit.co/dev/reference/image_array_save.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Saves an image stored as an array to a path or file object. — image_array_save","text":"Called primarily side effects. input x returned, invisibly, enable usage pipe.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/image_dataset_from_directory.html","id":null,"dir":"Reference","previous_headings":"","what":"Generates a tf.data.Dataset from image files in a directory. — image_dataset_from_directory","title":"Generates a tf.data.Dataset from image files in a directory. — image_dataset_from_directory","text":"directory structure :   calling image_dataset_from_directory(main_directory, labels = 'inferred') return tf.data.Dataset yields batches images subdirectories class_a class_b, together labels 0 1 (0 corresponding class_a 1 corresponding class_b). Supported image formats: .jpeg, .jpg, .png, .bmp, .gif. Animated gifs truncated first frame.","code":"main_directory/ ...class_a/ ......a_image_1.jpg ......a_image_2.jpg ...class_b/ ......b_image_1.jpg ......b_image_2.jpg"},{"path":"https://keras3.posit.co/dev/reference/image_dataset_from_directory.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Generates a tf.data.Dataset from image files in a directory. — image_dataset_from_directory","text":"","code":"image_dataset_from_directory(   directory,   labels = \"inferred\",   label_mode = \"int\",   class_names = NULL,   color_mode = \"rgb\",   batch_size = 32L,   image_size = c(256L, 256L),   shuffle = TRUE,   seed = NULL,   validation_split = NULL,   subset = NULL,   interpolation = \"bilinear\",   follow_links = FALSE,   crop_to_aspect_ratio = FALSE,   pad_to_aspect_ratio = FALSE,   data_format = NULL,   verbose = TRUE )"},{"path":"https://keras3.posit.co/dev/reference/image_dataset_from_directory.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Generates a tf.data.Dataset from image files in a directory. — image_dataset_from_directory","text":"directory Directory data located. labels \"inferred\", contain subdirectories, containing images class. Otherwise, directory structure ignored. labels Either \"inferred\" (labels generated directory structure), NULL (labels), list/tuple integer labels size number image files found directory. Labels sorted according alphanumeric order image file paths (obtained via os.walk(directory) Python). label_mode String describing encoding labels. Options : \"int\": means labels encoded integers (e.g. sparse_categorical_crossentropy loss). \"categorical\" means labels encoded categorical vector (e.g. categorical_crossentropy loss). \"binary\" means labels (can 2) encoded float32 scalars values 0 1 (e.g. binary_crossentropy). NULL (labels). class_names valid labels \"inferred\". explicit list class names (must match names subdirectories). Used control order classes (otherwise alphanumerical order used). color_mode One \"grayscale\", \"rgb\", \"rgba\". Whether images converted 1, 3, 4 channels. Defaults \"rgb\". batch_size Size batches data. Defaults 32. NULL, data batched (dataset yield individual samples). image_size Size resize images read disk, specified (height, width). Since pipeline processes batches images must size, must provided. Defaults (256, 256). shuffle Whether shuffle data. Defaults TRUE. set FALSE, sorts data alphanumeric order. seed Optional random seed shuffling transformations. validation_split Optional float 0 1, fraction data reserve validation. subset Subset data return. One \"training\", \"validation\", \"\". used validation_split set. subset = \"\", utility returns tuple two datasets (training validation datasets respectively). interpolation String, interpolation method used resizing images. Supports \"bilinear\", \"nearest\", \"bicubic\", \"area\", \"lanczos3\", \"lanczos5\", \"gaussian\", \"mitchellcubic\". Defaults \"bilinear\". follow_links Whether visit subdirectories pointed symlinks. Defaults FALSE. crop_to_aspect_ratio TRUE, resize images without aspect ratio distortion. original aspect ratio differs target aspect ratio, output image cropped return largest possible window image (size image_size) matches target aspect ratio. default (crop_to_aspect_ratio = FALSE), aspect ratio may preserved. pad_to_aspect_ratio TRUE, resize images without aspect ratio distortion. original aspect ratio differs target aspect ratio, output image padded return largest possible window image (size image_size) matches target aspect ratio. default (pad_to_aspect_ratio=FALSE), aspect ratio may preserved. data_format NULL uses config_image_data_format() otherwise either 'channel_last' 'channel_first'. verbose Whether display number information classes number files found. Defaults TRUE.","code":""},{"path":"https://keras3.posit.co/dev/reference/image_dataset_from_directory.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Generates a tf.data.Dataset from image files in a directory. — image_dataset_from_directory","text":"tf.data.Dataset object. label_mode NULL, yields float32 tensors shape (batch_size, image_size[1], image_size[2], num_channels), encoding images (see rules regarding num_channels). Otherwise, yields tuple (images, labels), images shape (batch_size, image_size[1], image_size[2], num_channels), labels follows format described . Rules regarding labels format: label_mode \"int\", labels int32 tensor shape (batch_size,). label_mode \"binary\", labels float32 tensor 1s 0s shape (batch_size, 1). label_mode \"categorical\", labels float32 tensor shape (batch_size, num_classes), representing one-hot encoding class index. Rules regarding number channels yielded images: color_mode \"grayscale\", 1 channel image tensors. color_mode \"rgb\", 3 channels image tensors. color_mode \"rgba\", 4 channels image tensors.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/image_from_array.html","id":null,"dir":"Reference","previous_headings":"","what":"Converts a 3D array to a PIL Image instance. — image_from_array","title":"Converts a 3D array to a PIL Image instance. — image_from_array","text":"Converts 3D array PIL Image instance.","code":""},{"path":"https://keras3.posit.co/dev/reference/image_from_array.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Converts a 3D array to a PIL Image instance. — image_from_array","text":"","code":"image_from_array(x, data_format = NULL, scale = TRUE, dtype = NULL)"},{"path":"https://keras3.posit.co/dev/reference/image_from_array.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Converts a 3D array to a PIL Image instance. — image_from_array","text":"x Input data, form can converted array. data_format Image data format, can either \"channels_first\" \"channels_last\". Defaults NULL, case global setting config_image_data_format() used (unless changed , defaults \"channels_last\"). scale Whether rescale image minimum maximum values 0 255 respectively. Defaults TRUE. dtype Dtype use. NULL means global setting config_floatx() used (unless changed , defaults \"float32\"). Defaults NULL.","code":""},{"path":"https://keras3.posit.co/dev/reference/image_from_array.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Converts a 3D array to a PIL Image instance. — image_from_array","text":"PIL Image instance.","code":""},{"path":"https://keras3.posit.co/dev/reference/image_from_array.html","id":"example","dir":"Reference","previous_headings":"","what":"Example","title":"Converts a 3D array to a PIL Image instance. — image_from_array","text":"","code":"img <- array(runif(30000), dim = c(100, 100, 3)) pil_img <- image_from_array(img) pil_img ## <PIL.Image.Image image mode=RGB size=100x100 at 0x0>"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/image_load.html","id":null,"dir":"Reference","previous_headings":"","what":"Loads an image into PIL format. — image_load","title":"Loads an image into PIL format. — image_load","text":"Loads image PIL format.","code":""},{"path":"https://keras3.posit.co/dev/reference/image_load.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Loads an image into PIL format. — image_load","text":"","code":"image_load(   path,   color_mode = \"rgb\",   target_size = NULL,   interpolation = \"nearest\",   keep_aspect_ratio = FALSE )"},{"path":"https://keras3.posit.co/dev/reference/image_load.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Loads an image into PIL format. — image_load","text":"path Path image file. color_mode One \"grayscale\", \"rgb\", \"rgba\". Default: \"rgb\". desired image format. target_size Either NULL (default original size) tuple ints (img_height, img_width). interpolation Interpolation method used resample image target size different loaded image. Supported methods \"nearest\", \"bilinear\", \"bicubic\". PIL version 1.1.3 newer installed, \"lanczos\" also supported. PIL version 3.4.0 newer installed, \"box\" \"hamming\" also supported. default, \"nearest\" used. keep_aspect_ratio Boolean, whether resize images target size without aspect ratio distortion. image cropped center target aspect ratio resizing.","code":""},{"path":"https://keras3.posit.co/dev/reference/image_load.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Loads an image into PIL format. — image_load","text":"PIL Image instance.","code":""},{"path":"https://keras3.posit.co/dev/reference/image_load.html","id":"example","dir":"Reference","previous_headings":"","what":"Example","title":"Loads an image into PIL format. — image_load","text":"","code":"image_path <- get_file(origin = \"https://www.r-project.org/logo/Rlogo.png\") (image <- image_load(image_path)) ## <PIL.Image.Image image mode=RGB size=724x561 at 0x0> input_arr <- image_to_array(image) str(input_arr) ##  num [1:561, 1:724, 1:3] 0 0 0 0 0 0 0 0 0 0 ... input_arr %<>% array_reshape(dim = c(1, dim(input_arr))) # Convert single image to a batch. model |> predict(input_arr)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/image_smart_resize.html","id":null,"dir":"Reference","previous_headings":"","what":"Resize images to a target size without aspect ratio distortion. — image_smart_resize","title":"Resize images to a target size without aspect ratio distortion. — image_smart_resize","text":"Image datasets typically yield images different size. However, images need batched can processed Keras layers. batched, images need share height width. simply , TF (JAX equivalent):   However, , distort aspect ratio images, since general aspect ratio size. fine many cases, always (e.g. image generation models can problem). Note passing argument preserve_aspect_ratio = TRUE tf$image$resize() preserve aspect ratio, cost longer respecting provided target size. calls :   output images actually (200, 200), distorted. Instead, parts image fit within target size get cropped . resizing process : Take largest centered crop image aspect ratio target size. instance, size = c(200, 200) input image size (340, 500), take crop (340, 340) centered along width. Resize cropped image target size. example , resize (340, 340) crop (200, 200).","code":"size <- c(200, 200) ds <- ds$map(\\(img) tf$image$resize(img, size)) size <- c(200, 200) ds <- ds$map(\\(img) image_smart_resize(img, size))"},{"path":"https://keras3.posit.co/dev/reference/image_smart_resize.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Resize images to a target size without aspect ratio distortion. — image_smart_resize","text":"","code":"image_smart_resize(   x,   size,   interpolation = \"bilinear\",   data_format = \"channels_last\",   backend_module = NULL )"},{"path":"https://keras3.posit.co/dev/reference/image_smart_resize.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Resize images to a target size without aspect ratio distortion. — image_smart_resize","text":"x Input image batch images (tensor array). Must format (height, width, channels) (batch_size, height, width, channels). size Tuple (height, width) integer. Target size. interpolation String, interpolation use resizing. Supports \"bilinear\", \"nearest\", \"bicubic\", \"lanczos3\", \"lanczos5\". Defaults 'bilinear'. data_format \"channels_last\" \"channels_first\". backend_module Backend module use (different default backend).","code":""},{"path":"https://keras3.posit.co/dev/reference/image_smart_resize.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Resize images to a target size without aspect ratio distortion. — image_smart_resize","text":"Array shape (size[1], size[2], channels). input image array, output array, backend-native tensor, output backend-native tensor.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/image_to_array.html","id":null,"dir":"Reference","previous_headings":"","what":"Converts a PIL Image instance to a matrix. — image_to_array","title":"Converts a PIL Image instance to a matrix. — image_to_array","text":"Converts PIL Image instance matrix.","code":""},{"path":"https://keras3.posit.co/dev/reference/image_to_array.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Converts a PIL Image instance to a matrix. — image_to_array","text":"","code":"image_to_array(img, data_format = NULL, dtype = NULL)"},{"path":"https://keras3.posit.co/dev/reference/image_to_array.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Converts a PIL Image instance to a matrix. — image_to_array","text":"img Input PIL Image instance. data_format Image data format, can either \"channels_first\" \"channels_last\". Defaults NULL, case global setting config_image_data_format() used (unless changed , defaults \"channels_last\"). dtype Dtype use. NULL \"double\" return R double array. \"integer\" returns R integer array. values (e.g., \"float32\") return NumPy array.","code":""},{"path":"https://keras3.posit.co/dev/reference/image_to_array.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Converts a PIL Image instance to a matrix. — image_to_array","text":"3D array.","code":""},{"path":"https://keras3.posit.co/dev/reference/image_to_array.html","id":"example","dir":"Reference","previous_headings":"","what":"Example","title":"Converts a PIL Image instance to a matrix. — image_to_array","text":"","code":"image_path <- get_file(origin = \"https://www.r-project.org/logo/Rlogo.png\") (img <- image_load(image_path)) ## <PIL.Image.Image image mode=RGB size=724x561 at 0x0> array <- image_to_array(img) str(array) ##  num [1:561, 1:724, 1:3] 0 0 0 0 0 0 0 0 0 0 ..."},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/imagenet_decode_predictions.html","id":null,"dir":"Reference","previous_headings":"","what":"Decodes the prediction of an ImageNet model. — imagenet_decode_predictions","title":"Decodes the prediction of an ImageNet model. — imagenet_decode_predictions","text":"Decodes prediction ImageNet model.","code":""},{"path":"https://keras3.posit.co/dev/reference/imagenet_decode_predictions.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Decodes the prediction of an ImageNet model. — imagenet_decode_predictions","text":"","code":"imagenet_decode_predictions(preds, top = 5)"},{"path":"https://keras3.posit.co/dev/reference/imagenet_decode_predictions.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Decodes the prediction of an ImageNet model. — imagenet_decode_predictions","text":"preds Tensor encoding batch predictions. top integer, many top-guesses return.","code":""},{"path":"https://keras3.posit.co/dev/reference/imagenet_decode_predictions.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Decodes the prediction of an ImageNet model. — imagenet_decode_predictions","text":"List data frames variables class_name, class_description, score (one data frame per sample batch input).","code":""},{"path":"https://keras3.posit.co/dev/reference/imagenet_preprocess_input.html","id":null,"dir":"Reference","previous_headings":"","what":"Preprocesses a tensor or array encoding a batch of images. — imagenet_preprocess_input","title":"Preprocesses a tensor or array encoding a batch of images. — imagenet_preprocess_input","text":"Preprocesses tensor array encoding batch images.","code":""},{"path":"https://keras3.posit.co/dev/reference/imagenet_preprocess_input.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Preprocesses a tensor or array encoding a batch of images. — imagenet_preprocess_input","text":"","code":"imagenet_preprocess_input(x, data_format = NULL, mode = \"caffe\")"},{"path":"https://keras3.posit.co/dev/reference/imagenet_preprocess_input.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Preprocesses a tensor or array encoding a batch of images. — imagenet_preprocess_input","text":"x Input Numpy symbolic tensor, 3D 4D. data_format Data format image tensor/array. mode One \"caffe\", \"tf\", \"torch\" caffe: convert images RGB BGR, zero-center color channel respect ImageNet dataset, without scaling. tf: scale pixels -1 1, sample-wise. torch: scale pixels 0 1 normalize channel respect ImageNet dataset.","code":""},{"path":"https://keras3.posit.co/dev/reference/imagenet_preprocess_input.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Preprocesses a tensor or array encoding a batch of images. — imagenet_preprocess_input","text":"Preprocessed tensor array.","code":""},{"path":"https://keras3.posit.co/dev/reference/initializer_constant.html","id":null,"dir":"Reference","previous_headings":"","what":"Initializer that generates tensors with constant values. — initializer_constant","title":"Initializer that generates tensors with constant values. — initializer_constant","text":"scalar values allowed. constant value provided must convertible dtype requested calling initializer.","code":""},{"path":"https://keras3.posit.co/dev/reference/initializer_constant.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Initializer that generates tensors with constant values. — initializer_constant","text":"","code":"initializer_constant(value = 0)"},{"path":"https://keras3.posit.co/dev/reference/initializer_constant.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Initializer that generates tensors with constant values. — initializer_constant","text":"value numeric scalar.","code":""},{"path":"https://keras3.posit.co/dev/reference/initializer_constant.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Initializer that generates tensors with constant values. — initializer_constant","text":"Initializer instance can passed layer variable constructors, called directly shape return Tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/initializer_constant.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Initializer that generates tensors with constant values. — initializer_constant","text":"","code":"# Standalone usage: initializer <- initializer_constant(10) values <- initializer(shape = c(2, 2)) # Usage in a Keras layer: initializer <- initializer_constant(10) layer <- layer_dense(units = 3, kernel_initializer = initializer)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/initializer_glorot_normal.html","id":null,"dir":"Reference","previous_headings":"","what":"The Glorot normal initializer, also called Xavier normal initializer. — initializer_glorot_normal","title":"The Glorot normal initializer, also called Xavier normal initializer. — initializer_glorot_normal","text":"Draws samples truncated normal distribution centered 0 stddev = sqrt(2 / (fan_in + fan_out)) fan_in number input units weight tensor fan_out number output units weight tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/initializer_glorot_normal.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"The Glorot normal initializer, also called Xavier normal initializer. — initializer_glorot_normal","text":"","code":"initializer_glorot_normal(seed = NULL)"},{"path":"https://keras3.posit.co/dev/reference/initializer_glorot_normal.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"The Glorot normal initializer, also called Xavier normal initializer. — initializer_glorot_normal","text":"seed integer instance random_seed_generator(). Used make behavior initializer deterministic. Note initializer seeded integer NULL (unseeded) produce random values across multiple calls. get different random values across multiple calls, use seed instance random_seed_generator().","code":""},{"path":"https://keras3.posit.co/dev/reference/initializer_glorot_normal.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"The Glorot normal initializer, also called Xavier normal initializer. — initializer_glorot_normal","text":"Initializer instance can passed layer variable constructors, called directly shape return Tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/initializer_glorot_normal.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"The Glorot normal initializer, also called Xavier normal initializer. — initializer_glorot_normal","text":"","code":"# Standalone usage: initializer <- initializer_glorot_normal() values <- initializer(shape = c(2, 2)) # Usage in a Keras layer: initializer <- initializer_glorot_normal() layer <- layer_dense(units = 3, kernel_initializer = initializer)"},{"path":"https://keras3.posit.co/dev/reference/initializer_glorot_normal.html","id":"reference","dir":"Reference","previous_headings":"","what":"Reference","title":"The Glorot normal initializer, also called Xavier normal initializer. — initializer_glorot_normal","text":"Glorot et al., 2010","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/initializer_glorot_uniform.html","id":null,"dir":"Reference","previous_headings":"","what":"The Glorot uniform initializer, also called Xavier uniform initializer. — initializer_glorot_uniform","title":"The Glorot uniform initializer, also called Xavier uniform initializer. — initializer_glorot_uniform","text":"Draws samples uniform distribution within [-limit, limit], limit = sqrt(6 / (fan_in + fan_out)) (fan_in number input units weight tensor fan_out number output units).","code":""},{"path":"https://keras3.posit.co/dev/reference/initializer_glorot_uniform.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"The Glorot uniform initializer, also called Xavier uniform initializer. — initializer_glorot_uniform","text":"","code":"initializer_glorot_uniform(seed = NULL)"},{"path":"https://keras3.posit.co/dev/reference/initializer_glorot_uniform.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"The Glorot uniform initializer, also called Xavier uniform initializer. — initializer_glorot_uniform","text":"seed integer instance random_seed_generator(). Used make behavior initializer deterministic. Note initializer seeded integer NULL (unseeded) produce random values across multiple calls. get different random values across multiple calls, use seed instance random_seed_generator().","code":""},{"path":"https://keras3.posit.co/dev/reference/initializer_glorot_uniform.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"The Glorot uniform initializer, also called Xavier uniform initializer. — initializer_glorot_uniform","text":"Initializer instance can passed layer variable constructors, called directly shape return Tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/initializer_glorot_uniform.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"The Glorot uniform initializer, also called Xavier uniform initializer. — initializer_glorot_uniform","text":"","code":"# Standalone usage: initializer <- initializer_glorot_uniform() values <- initializer(shape = c(2, 2)) # Usage in a Keras layer: initializer <- initializer_glorot_uniform() layer <- layer_dense(units = 3, kernel_initializer = initializer)"},{"path":"https://keras3.posit.co/dev/reference/initializer_glorot_uniform.html","id":"reference","dir":"Reference","previous_headings":"","what":"Reference","title":"The Glorot uniform initializer, also called Xavier uniform initializer. — initializer_glorot_uniform","text":"Glorot et al., 2010","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/initializer_he_normal.html","id":null,"dir":"Reference","previous_headings":"","what":"He normal initializer. — initializer_he_normal","title":"He normal initializer. — initializer_he_normal","text":"draws samples truncated normal distribution centered 0 stddev = sqrt(2 / fan_in) fan_in number input units weight tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/initializer_he_normal.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"He normal initializer. — initializer_he_normal","text":"","code":"initializer_he_normal(seed = NULL)"},{"path":"https://keras3.posit.co/dev/reference/initializer_he_normal.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"He normal initializer. — initializer_he_normal","text":"seed integer instance random_seed_generator(). Used make behavior initializer deterministic. Note initializer seeded integer NULL (unseeded) produce random values across multiple calls. get different random values across multiple calls, use seed instance random_seed_generator().","code":""},{"path":"https://keras3.posit.co/dev/reference/initializer_he_normal.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"He normal initializer. — initializer_he_normal","text":"Initializer instance can passed layer variable constructors, called directly shape return Tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/initializer_he_normal.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"He normal initializer. — initializer_he_normal","text":"","code":"# Standalone usage: initializer <- initializer_he_normal() values <- initializer(shape = c(2, 2)) # Usage in a Keras layer: initializer <- initializer_he_normal() layer <- layer_dense(units = 3, kernel_initializer = initializer)"},{"path":"https://keras3.posit.co/dev/reference/initializer_he_normal.html","id":"reference","dir":"Reference","previous_headings":"","what":"Reference","title":"He normal initializer. — initializer_he_normal","text":"et al., 2015","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/initializer_he_uniform.html","id":null,"dir":"Reference","previous_headings":"","what":"He uniform variance scaling initializer. — initializer_he_uniform","title":"He uniform variance scaling initializer. — initializer_he_uniform","text":"Draws samples uniform distribution within [-limit, limit], limit = sqrt(6 / fan_in) (fan_in number input units weight tensor).","code":""},{"path":"https://keras3.posit.co/dev/reference/initializer_he_uniform.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"He uniform variance scaling initializer. — initializer_he_uniform","text":"","code":"initializer_he_uniform(seed = NULL)"},{"path":"https://keras3.posit.co/dev/reference/initializer_he_uniform.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"He uniform variance scaling initializer. — initializer_he_uniform","text":"seed integer instance random_seed_generator(). Used make behavior initializer deterministic. Note initializer seeded integer NULL (unseeded) produce random values across multiple calls. get different random values across multiple calls, use seed instance random_seed_generator().","code":""},{"path":"https://keras3.posit.co/dev/reference/initializer_he_uniform.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"He uniform variance scaling initializer. — initializer_he_uniform","text":"Initializer instance can passed layer variable constructors, called directly shape return Tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/initializer_he_uniform.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"He uniform variance scaling initializer. — initializer_he_uniform","text":"","code":"# Standalone usage: initializer <- initializer_he_uniform() values <- initializer(shape = c(2, 2)) # Usage in a Keras layer: initializer <- initializer_he_uniform() layer <- layer_dense(units = 3, kernel_initializer = initializer)"},{"path":"https://keras3.posit.co/dev/reference/initializer_he_uniform.html","id":"reference","dir":"Reference","previous_headings":"","what":"Reference","title":"He uniform variance scaling initializer. — initializer_he_uniform","text":"et al., 2015","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/initializer_identity.html","id":null,"dir":"Reference","previous_headings":"","what":"Initializer that generates the identity matrix. — initializer_identity","title":"Initializer that generates the identity matrix. — initializer_identity","text":"usable generating 2D matrices.","code":""},{"path":"https://keras3.posit.co/dev/reference/initializer_identity.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Initializer that generates the identity matrix. — initializer_identity","text":"","code":"initializer_identity(gain = 1)"},{"path":"https://keras3.posit.co/dev/reference/initializer_identity.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Initializer that generates the identity matrix. — initializer_identity","text":"gain Multiplicative factor apply identity matrix.","code":""},{"path":"https://keras3.posit.co/dev/reference/initializer_identity.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Initializer that generates the identity matrix. — initializer_identity","text":"Initializer instance can passed layer variable constructors, called directly shape return Tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/initializer_identity.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Initializer that generates the identity matrix. — initializer_identity","text":"","code":"# Standalone usage: initializer <- initializer_identity() values <- initializer(shape = c(2, 2)) # Usage in a Keras layer: initializer <- initializer_identity() layer <- layer_dense(units = 3, kernel_initializer = initializer)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/initializer_lecun_normal.html","id":null,"dir":"Reference","previous_headings":"","what":"Lecun normal initializer. — initializer_lecun_normal","title":"Lecun normal initializer. — initializer_lecun_normal","text":"Initializers allow pre-specify initialization strategy, encoded Initializer object, without knowing shape dtype variable initialized. Draws samples truncated normal distribution centered 0 stddev = sqrt(1 / fan_in) fan_in number input units weight tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/initializer_lecun_normal.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Lecun normal initializer. — initializer_lecun_normal","text":"","code":"initializer_lecun_normal(seed = NULL)"},{"path":"https://keras3.posit.co/dev/reference/initializer_lecun_normal.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Lecun normal initializer. — initializer_lecun_normal","text":"seed integer instance random_seed_generator(). Used make behavior initializer deterministic. Note initializer seeded integer NULL (unseeded) produce random values across multiple calls. get different random values across multiple calls, use seed instance random_seed_generator().","code":""},{"path":"https://keras3.posit.co/dev/reference/initializer_lecun_normal.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Lecun normal initializer. — initializer_lecun_normal","text":"Initializer instance can passed layer variable constructors, called directly shape return Tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/initializer_lecun_normal.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Lecun normal initializer. — initializer_lecun_normal","text":"","code":"# Standalone usage: initializer <- initializer_lecun_normal() values <- initializer(shape = c(2, 2)) # Usage in a Keras layer: initializer <- initializer_lecun_normal() layer <- layer_dense(units = 3, kernel_initializer = initializer)"},{"path":"https://keras3.posit.co/dev/reference/initializer_lecun_normal.html","id":"reference","dir":"Reference","previous_headings":"","what":"Reference","title":"Lecun normal initializer. — initializer_lecun_normal","text":"Klambauer et al., 2017","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/initializer_lecun_uniform.html","id":null,"dir":"Reference","previous_headings":"","what":"Lecun uniform initializer. — initializer_lecun_uniform","title":"Lecun uniform initializer. — initializer_lecun_uniform","text":"Draws samples uniform distribution within [-limit, limit], limit = sqrt(3 / fan_in) (fan_in number input units weight tensor).","code":""},{"path":"https://keras3.posit.co/dev/reference/initializer_lecun_uniform.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Lecun uniform initializer. — initializer_lecun_uniform","text":"","code":"initializer_lecun_uniform(seed = NULL)"},{"path":"https://keras3.posit.co/dev/reference/initializer_lecun_uniform.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Lecun uniform initializer. — initializer_lecun_uniform","text":"seed integer instance random_seed_generator(). Used make behavior initializer deterministic. Note initializer seeded integer NULL (unseeded) produce random values across multiple calls. get different random values across multiple calls, use seed instance random_seed_generator().","code":""},{"path":"https://keras3.posit.co/dev/reference/initializer_lecun_uniform.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Lecun uniform initializer. — initializer_lecun_uniform","text":"Initializer instance can passed layer variable constructors, called directly shape return Tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/initializer_lecun_uniform.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Lecun uniform initializer. — initializer_lecun_uniform","text":"","code":"# Standalone usage: initializer <- initializer_lecun_uniform() values <- initializer(shape = c(2, 2)) # Usage in a Keras layer: initializer <- initializer_lecun_uniform() layer <- layer_dense(units = 3, kernel_initializer = initializer)"},{"path":"https://keras3.posit.co/dev/reference/initializer_lecun_uniform.html","id":"reference","dir":"Reference","previous_headings":"","what":"Reference","title":"Lecun uniform initializer. — initializer_lecun_uniform","text":"Klambauer et al., 2017","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/initializer_ones.html","id":null,"dir":"Reference","previous_headings":"","what":"Initializer that generates tensors initialized to 1. — initializer_ones","title":"Initializer that generates tensors initialized to 1. — initializer_ones","text":"Also available via shortcut function ones.","code":""},{"path":"https://keras3.posit.co/dev/reference/initializer_ones.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Initializer that generates tensors initialized to 1. — initializer_ones","text":"","code":"initializer_ones()"},{"path":"https://keras3.posit.co/dev/reference/initializer_ones.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Initializer that generates tensors initialized to 1. — initializer_ones","text":"Initializer instance can passed layer variable constructors, called directly shape return Tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/initializer_ones.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Initializer that generates tensors initialized to 1. — initializer_ones","text":"","code":"# Standalone usage: initializer <- initializer_ones() values <- initializer(shape = c(2, 2)) # Usage in a Keras layer: initializer <- initializer_ones() layer <- layer_dense(units = 3, kernel_initializer = initializer)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/initializer_orthogonal.html","id":null,"dir":"Reference","previous_headings":"","what":"Initializer that generates an orthogonal matrix. — initializer_orthogonal","title":"Initializer that generates an orthogonal matrix. — initializer_orthogonal","text":"shape tensor initialize two-dimensional, initialized orthogonal matrix obtained QR decomposition matrix random numbers drawn normal distribution. matrix fewer rows columns output orthogonal rows. Otherwise, output orthogonal columns. shape tensor initialize two-dimensional, matrix shape (shape[1] * ... * shape[n - 1], shape[n]) initialized, n length shape vector. matrix subsequently reshaped give tensor desired shape.","code":""},{"path":"https://keras3.posit.co/dev/reference/initializer_orthogonal.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Initializer that generates an orthogonal matrix. — initializer_orthogonal","text":"","code":"initializer_orthogonal(gain = 1, seed = NULL)"},{"path":"https://keras3.posit.co/dev/reference/initializer_orthogonal.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Initializer that generates an orthogonal matrix. — initializer_orthogonal","text":"gain Multiplicative factor apply orthogonal matrix. seed integer. Used make behavior initializer deterministic.","code":""},{"path":"https://keras3.posit.co/dev/reference/initializer_orthogonal.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Initializer that generates an orthogonal matrix. — initializer_orthogonal","text":"Initializer instance can passed layer variable constructors, called directly shape return Tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/initializer_orthogonal.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Initializer that generates an orthogonal matrix. — initializer_orthogonal","text":"","code":"# Standalone usage: initializer <- initializer_orthogonal() values <- initializer(shape = c(2, 2)) # Usage in a Keras layer: initializer <- initializer_orthogonal() layer <- layer_dense(units = 3, kernel_initializer = initializer)"},{"path":"https://keras3.posit.co/dev/reference/initializer_orthogonal.html","id":"reference","dir":"Reference","previous_headings":"","what":"Reference","title":"Initializer that generates an orthogonal matrix. — initializer_orthogonal","text":"Saxe et al., 2014","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/initializer_random_normal.html","id":null,"dir":"Reference","previous_headings":"","what":"Random normal initializer. — initializer_random_normal","title":"Random normal initializer. — initializer_random_normal","text":"Draws samples normal distribution given parameters.","code":""},{"path":"https://keras3.posit.co/dev/reference/initializer_random_normal.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Random normal initializer. — initializer_random_normal","text":"","code":"initializer_random_normal(mean = 0, stddev = 0.05, seed = NULL)"},{"path":"https://keras3.posit.co/dev/reference/initializer_random_normal.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Random normal initializer. — initializer_random_normal","text":"mean numeric scalar. Mean random values generate. stddev numeric scalar. Standard deviation random values generate. seed integer instance random_seed_generator(). Used make behavior initializer deterministic. Note initializer seeded integer NULL (unseeded) produce random values across multiple calls. get different random values across multiple calls, use seed instance random_seed_generator().","code":""},{"path":"https://keras3.posit.co/dev/reference/initializer_random_normal.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Random normal initializer. — initializer_random_normal","text":"Initializer instance can passed layer variable constructors, called directly shape return Tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/initializer_random_normal.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Random normal initializer. — initializer_random_normal","text":"","code":"# Standalone usage: initializer <- initializer_random_normal(mean = 0.0, stddev = 1.0) values <- initializer(shape = c(2, 2)) # Usage in a Keras layer: initializer <- initializer_random_normal(mean = 0.0, stddev = 1.0) layer <- layer_dense(units = 3, kernel_initializer = initializer)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/initializer_random_uniform.html","id":null,"dir":"Reference","previous_headings":"","what":"Random uniform initializer. — initializer_random_uniform","title":"Random uniform initializer. — initializer_random_uniform","text":"Draws samples uniform distribution given parameters.","code":""},{"path":"https://keras3.posit.co/dev/reference/initializer_random_uniform.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Random uniform initializer. — initializer_random_uniform","text":"","code":"initializer_random_uniform(minval = -0.05, maxval = 0.05, seed = NULL)"},{"path":"https://keras3.posit.co/dev/reference/initializer_random_uniform.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Random uniform initializer. — initializer_random_uniform","text":"minval numeric scalar scalar keras tensor. Lower bound range random values generate (inclusive). maxval numeric scalar scalar keras tensor. Upper bound range random values generate (exclusive). seed integer instance random_seed_generator(). Used make behavior initializer deterministic. Note initializer seeded integer NULL (unseeded) produce random values across multiple calls. get different random values across multiple calls, use seed instance random_seed_generator().","code":""},{"path":"https://keras3.posit.co/dev/reference/initializer_random_uniform.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Random uniform initializer. — initializer_random_uniform","text":"Initializer instance can passed layer variable constructors, called directly shape return Tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/initializer_random_uniform.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Random uniform initializer. — initializer_random_uniform","text":"","code":"# Standalone usage: initializer <- initializer_random_uniform(minval = 0.0, maxval = 1.0) values <- initializer(shape = c(2, 2)) # Usage in a Keras layer: initializer <- initializer_random_uniform(minval = 0.0, maxval = 1.0) layer <- layer_dense(units = 3, kernel_initializer = initializer)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/initializer_stft.html","id":null,"dir":"Reference","previous_headings":"","what":"Initializer of Conv kernels for Short-term Fourier Transformation (STFT). — initializer_stft","title":"Initializer of Conv kernels for Short-term Fourier Transformation (STFT). — initializer_stft","text":"Since formula involves complex numbers, class compute either real imaginary components final output. Additionally, initializer supports windowing functions across time dimension commonly used STFT. Windowing functions Python module scipy.signal.windows supported, including common hann hamming windowing functions. layer supports periodic windows scaling-based normalization. primarily intended use STFTSpectrogram layer.","code":""},{"path":"https://keras3.posit.co/dev/reference/initializer_stft.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Initializer of Conv kernels for Short-term Fourier Transformation (STFT). — initializer_stft","text":"","code":"initializer_stft(   side = \"real\",   window = \"hann\",   scaling = \"density\",   periodic = FALSE )"},{"path":"https://keras3.posit.co/dev/reference/initializer_stft.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Initializer of Conv kernels for Short-term Fourier Transformation (STFT). — initializer_stft","text":"side String, \"real\" \"imag\" deciding kernel compute real side imaginary side output. Defaults \"real\". window String name windowing function scipy.signal.windows module, array_like window values, NULL windowing. scaling String, \"density\" \"spectrum\" scaling window normalization, either L2 L1 normalization. NULL scaling. periodic Boolean, True, window function treated periodic. Defaults FALSE.","code":""},{"path":"https://keras3.posit.co/dev/reference/initializer_stft.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Initializer of Conv kernels for Short-term Fourier Transformation (STFT). — initializer_stft","text":"Initializer instance can passed layer variable constructors, called directly shape return Tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/initializer_stft.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Initializer of Conv kernels for Short-term Fourier Transformation (STFT). — initializer_stft","text":"","code":"# Standalone usage: initializer <- initializer_stft(\"real\", \"hann\", \"density\", FALSE) values <- initializer(shape = c(128, 1, 513))"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/initializer_truncated_normal.html","id":null,"dir":"Reference","previous_headings":"","what":"Initializer that generates a truncated normal distribution. — initializer_truncated_normal","title":"Initializer that generates a truncated normal distribution. — initializer_truncated_normal","text":"values generated similar values RandomNormal initializer, except values two standard deviations mean discarded re-drawn.","code":""},{"path":"https://keras3.posit.co/dev/reference/initializer_truncated_normal.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Initializer that generates a truncated normal distribution. — initializer_truncated_normal","text":"","code":"initializer_truncated_normal(mean = 0, stddev = 0.05, seed = NULL)"},{"path":"https://keras3.posit.co/dev/reference/initializer_truncated_normal.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Initializer that generates a truncated normal distribution. — initializer_truncated_normal","text":"mean numeric scalar. Mean random values generate. stddev numeric scalar. Standard deviation random values generate. seed integer instance random_seed_generator(). Used make behavior initializer deterministic. Note initializer seeded integer NULL (unseeded) produce random values across multiple calls. get different random values across multiple calls, use seed instance random_seed_generator().","code":""},{"path":"https://keras3.posit.co/dev/reference/initializer_truncated_normal.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Initializer that generates a truncated normal distribution. — initializer_truncated_normal","text":"Initializer instance can passed layer variable constructors, called directly shape return Tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/initializer_truncated_normal.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Initializer that generates a truncated normal distribution. — initializer_truncated_normal","text":"","code":"# Standalone usage: initializer <- initializer_truncated_normal(mean = 0, stddev = 1) values <- initializer(shape = c(2, 2)) # Usage in a Keras layer: initializer <- initializer_truncated_normal(mean = 0, stddev = 1) layer <- layer_dense(units = 3, kernel_initializer = initializer)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/initializer_variance_scaling.html","id":null,"dir":"Reference","previous_headings":"","what":"Initializer that adapts its scale to the shape of its input tensors. — initializer_variance_scaling","title":"Initializer that adapts its scale to the shape of its input tensors. — initializer_variance_scaling","text":"distribution = \"truncated_normal\" \"untruncated_normal\", samples drawn truncated/untruncated normal distribution mean zero standard deviation (truncation, used) stddev = sqrt(scale / n), n : number input units weight tensor, mode = \"fan_in\" number output units, mode = \"fan_out\" average numbers input output units, mode = \"fan_avg\" distribution = \"uniform\", samples drawn uniform distribution within [-limit, limit], limit = sqrt(3 * scale / n).","code":""},{"path":"https://keras3.posit.co/dev/reference/initializer_variance_scaling.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Initializer that adapts its scale to the shape of its input tensors. — initializer_variance_scaling","text":"","code":"initializer_variance_scaling(   scale = 1,   mode = \"fan_in\",   distribution = \"truncated_normal\",   seed = NULL )"},{"path":"https://keras3.posit.co/dev/reference/initializer_variance_scaling.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Initializer that adapts its scale to the shape of its input tensors. — initializer_variance_scaling","text":"scale Scaling factor (positive float). mode One \"fan_in\", \"fan_out\", \"fan_avg\". distribution Random distribution use. One \"truncated_normal\", \"untruncated_normal\", \"uniform\". seed integer instance random_seed_generator(). Used make behavior initializer deterministic. Note initializer seeded integer NULL (unseeded) produce random values across multiple calls. get different random values across multiple calls, use seed instance random_seed_generator().","code":""},{"path":"https://keras3.posit.co/dev/reference/initializer_variance_scaling.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Initializer that adapts its scale to the shape of its input tensors. — initializer_variance_scaling","text":"Initializer instance can passed layer variable constructors, called directly shape return Tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/initializer_variance_scaling.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Initializer that adapts its scale to the shape of its input tensors. — initializer_variance_scaling","text":"","code":"# Standalone usage: initializer <- initializer_variance_scaling(scale = 0.1, mode = 'fan_in',                                             distribution = 'uniform') values <- initializer(shape = c(2, 2)) # Usage in a Keras layer: initializer <- initializer_variance_scaling(scale = 0.1, mode = 'fan_in',                                             distribution = 'uniform') layer <- layer_dense(units = 3, kernel_initializer = initializer)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/initializer_zeros.html","id":null,"dir":"Reference","previous_headings":"","what":"Initializer that generates tensors initialized to 0. — initializer_zeros","title":"Initializer that generates tensors initialized to 0. — initializer_zeros","text":"Initializer generates tensors initialized 0.","code":""},{"path":"https://keras3.posit.co/dev/reference/initializer_zeros.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Initializer that generates tensors initialized to 0. — initializer_zeros","text":"","code":"initializer_zeros()"},{"path":"https://keras3.posit.co/dev/reference/initializer_zeros.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Initializer that generates tensors initialized to 0. — initializer_zeros","text":"Initializer instance can passed layer variable constructors, called directly shape return Tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/initializer_zeros.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Initializer that generates tensors initialized to 0. — initializer_zeros","text":"","code":"# Standalone usage: initializer <- initializer_zeros() values <- initializer(shape = c(2, 2)) # Usage in a Keras layer: initializer <- initializer_zeros() layer <- layer_dense(units = 3, kernel_initializer = initializer)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/install_keras.html","id":null,"dir":"Reference","previous_headings":"","what":"Install Keras — install_keras","title":"Install Keras — install_keras","text":"function install Keras along selected backend, including Python dependencies.","code":""},{"path":"https://keras3.posit.co/dev/reference/install_keras.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Install Keras — install_keras","text":"","code":"install_keras(   envname = \"r-keras\",   ...,   extra_packages = c(\"scipy\", \"pandas\", \"Pillow\", \"pydot\", \"ipython\",     \"tensorflow_datasets\"),   python_version = \">=3.9,<=3.11\",   backend = c(\"tensorflow\", \"jax\"),   gpu = NA,   restart_session = TRUE )"},{"path":"https://keras3.posit.co/dev/reference/install_keras.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Install Keras — install_keras","text":"envname Name path Python virtual environment ... reserved future compatibility. extra_packages Additional Python packages install alongside Keras python_version Passed reticulate::virtualenv_starter() backend backend(s) install. Accepted values include \"tensorflow\", \"jax\" \"torch\" gpu whether install GPU capable version backend. restart_session Whether restart R session installing (note occur within RStudio).","code":""},{"path":"https://keras3.posit.co/dev/reference/install_keras.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Install Keras — install_keras","text":"return value, called side effects.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/keras.html","id":null,"dir":"Reference","previous_headings":"","what":"Main Keras module — keras","title":"Main Keras module — keras","text":"keras module object equivalent reticulate::import(\"keras\") provided mainly convenience.","code":""},{"path":"https://keras3.posit.co/dev/reference/keras.html","id":"format","dir":"Reference","previous_headings":"","what":"Format","title":"Main Keras module — keras","text":"object class python.builtin.module","code":""},{"path":"https://keras3.posit.co/dev/reference/keras.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Main Keras module — keras","text":"keras Python module","code":""},{"path":"https://keras3.posit.co/dev/reference/keras3-package.html","id":null,"dir":"Reference","previous_headings":"","what":"keras3: R Interface to 'Keras' — keras3-package","title":"keras3: R Interface to 'Keras' — keras3-package","text":"Interface 'Keras' https://keras.io, high-level neural networks API. 'Keras' developed focus enabling fast experimentation, supports convolution based networks recurrent networks (well combinations two), runs seamlessly CPU GPU devices. Keras high-level neural networks API, developed focus enabling fast experimentation. Keras following key features:","code":""},{"path":"https://keras3.posit.co/dev/reference/keras3-package.html","id":"details","dir":"Reference","previous_headings":"","what":"Details","title":"keras3: R Interface to 'Keras' — keras3-package","text":"Allows code run CPU GPU, seamlessly. User-friendly API makes easy quickly prototype deep learning models. Built-support convolutional networks (computer vision), recurrent networks (sequence processing), combination . Supports arbitrary network architectures: multi-input multi-output models, layer sharing, model sharing, etc. means Keras appropriate building essentially deep learning model, memory network neural Turing machine. capable running top multiple back-ends including TensorFlow, Jax, PyTorch. See package website https://keras3.posit.co complete documentation.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/keras3-package.html","id":"author","dir":"Reference","previous_headings":"","what":"Author","title":"keras3: R Interface to 'Keras' — keras3-package","text":"Maintainer: Tomasz Kalinowski tomasz@posit.co [copyright holder] Authors: JJ Allaire [copyright holder] François Chollet [copyright holder] contributors: Daniel Falbel daniel@posit.co [contributor, copyright holder] Posit Software, PBC [copyright holder, funder] Google [copyright holder, funder] Yuan Tang terrytangyuan@gmail.com (ORCID) [contributor, copyright holder] Wouter Van Der Bijl [contributor, copyright holder] Martin Studer [contributor, copyright holder] Sigrid Keydana [contributor]","code":""},{"path":"https://keras3.posit.co/dev/reference/keras_input.html","id":null,"dir":"Reference","previous_headings":"","what":"Create a Keras tensor (Functional API input). — keras_input","title":"Create a Keras tensor (Functional API input). — keras_input","text":"Keras tensor symbolic tensor-like object, augment certain attributes allow us build Keras model just knowing inputs outputs model. instance, , b c Keras tensors, becomes possible : model <- keras_model(input = c(, b), output = c)","code":""},{"path":"https://keras3.posit.co/dev/reference/keras_input.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Create a Keras tensor (Functional API input). — keras_input","text":"","code":"keras_input(   shape = NULL,   batch_size = NULL,   dtype = NULL,   sparse = NULL,   ragged = NULL,   batch_shape = NULL,   name = NULL,   tensor = NULL,   optional = FALSE )"},{"path":"https://keras3.posit.co/dev/reference/keras_input.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Create a Keras tensor (Functional API input). — keras_input","text":"shape shape list (list integers NULL objects), including batch size. instance, shape = c(32) indicates expected input batches 32-dimensional vectors. Elements list can NULL NA; NULL/NA elements represent dimensions shape known may vary (e.g. sequence length). batch_size Optional static batch size (integer). dtype data type expected input, string (e.g. \"float32\", \"int32\"...) sparse boolean specifying whether expected input sparse tensors. Note , sparse FALSE, sparse tensors can still passed input - densified default value 0. feature supported TensorFlow backend. Defaults FALSE. ragged boolean specifying whether expected input ragged tensors. Note , ragged FALSE, ragged tensors can still passed input - densified default value 0. feature supported TensorFlow backend. Defaults FALSE. batch_shape Optional shape list (list integers NULL objects), including batch size. name Optional name string layer. unique model (reuse name twice). autogenerated provided. tensor Optional existing tensor wrap Input layer. set, layer use tensor rather creating new placeholder tensor. optional Boolean, whether input optional . optional input can accept NULL values.","code":""},{"path":"https://keras3.posit.co/dev/reference/keras_input.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Create a Keras tensor (Functional API input). — keras_input","text":"Keras tensor, can passed inputs argument (keras_model()).","code":""},{"path":"https://keras3.posit.co/dev/reference/keras_input.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Create a Keras tensor (Functional API input). — keras_input","text":"","code":"# This is a logistic regression in Keras input <- layer_input(shape=c(32)) output <- input |> layer_dense(16, activation='softmax') model <- keras_model(input, output)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/keras_model.html","id":null,"dir":"Reference","previous_headings":"","what":"Keras Model (Functional API) — keras_model","title":"Keras Model (Functional API) — keras_model","text":"model directed acyclic graph layers.","code":""},{"path":"https://keras3.posit.co/dev/reference/keras_model.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Keras Model (Functional API) — keras_model","text":"","code":"keras_model(inputs = NULL, outputs = NULL, ...)"},{"path":"https://keras3.posit.co/dev/reference/keras_model.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Keras Model (Functional API) — keras_model","text":"inputs Input tensor(s) (keras_input()) outputs Output tensors (calling layers inputs) ... additional arguments","code":""},{"path":"https://keras3.posit.co/dev/reference/keras_model.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Keras Model (Functional API) — keras_model","text":"Model instance.","code":""},{"path":"https://keras3.posit.co/dev/reference/keras_model.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Keras Model (Functional API) — keras_model","text":"","code":"library(keras3)  # input tensor inputs <- keras_input(shape = c(784))  # outputs compose input + dense layers predictions <- inputs |>   layer_dense(units = 64, activation = 'relu') |>   layer_dense(units = 64, activation = 'relu') |>   layer_dense(units = 10, activation = 'softmax')  # create and compile model model <- keras_model(inputs = inputs, outputs = predictions) model |> compile(   optimizer = 'rmsprop',   loss = 'categorical_crossentropy',   metrics = c('accuracy') )"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/keras_model_sequential.html","id":null,"dir":"Reference","previous_headings":"","what":"Keras Model composed of a linear stack of layers — keras_model_sequential","title":"Keras Model composed of a linear stack of layers — keras_model_sequential","text":"Keras Model composed linear stack layers","code":""},{"path":"https://keras3.posit.co/dev/reference/keras_model_sequential.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Keras Model composed of a linear stack of layers — keras_model_sequential","text":"","code":"keras_model_sequential(   input_shape = NULL,   name = NULL,   ...,   input_dtype = NULL,   input_batch_size = NULL,   input_sparse = NULL,   input_ragged = NULL,   input_batch_shape = NULL,   input_name = NULL,   input_tensor = NULL,   input_optional = FALSE,   trainable = TRUE,   layers = list() )"},{"path":"https://keras3.posit.co/dev/reference/keras_model_sequential.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Keras Model composed of a linear stack of layers — keras_model_sequential","text":"input_shape shape integer vector, including batch size. instance, shape=c(32) indicates expected input batches 32-dimensional vectors. Elements shape can NA; NA elements represent dimensions shape known may vary (e.g. sequence length). name Name model ... additional arguments passed keras.layers.InputLayer. input_dtype data type expected input, string (e.g. \"float32\", \"int32\"...) input_batch_size Optional static batch size (integer). input_sparse boolean specifying whether expected input sparse tensors. Note , sparse FALSE, sparse tensors can still passed input - densified default value 0. feature supported TensorFlow backend. Defaults FALSE. input_ragged bool. input_batch_shape optional way specify batch_size input_shape one argument. input_name Optional name string input layer. unique model (reuse name twice). autogenerated provided. input_tensor Optional existing tensor wrap InputLayer. set, layer use tensor rather creating new placeholder tensor. input_optional Boolean, whether input optional . optional input can accept NULL values. trainable Boolean, whether model's variables trainable. can also change trainable status model/layer freeze_weights() unfreeze_weights(). layers List layers add model.","code":""},{"path":"https://keras3.posit.co/dev/reference/keras_model_sequential.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Keras Model composed of a linear stack of layers — keras_model_sequential","text":"Sequential model instance.","code":""},{"path":"https://keras3.posit.co/dev/reference/keras_model_sequential.html","id":"note","dir":"Reference","previous_headings":"","what":"Note","title":"Keras Model composed of a linear stack of layers — keras_model_sequential","text":"input_shape omitted, model layer shapes, including final model output shape, known model built, either calling model input tensor/array like model(input), (possibly via fit()/evaluate()/predict()), explicitly calling model$build(input_shape).","code":""},{"path":"https://keras3.posit.co/dev/reference/keras_model_sequential.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Keras Model composed of a linear stack of layers — keras_model_sequential","text":"","code":"model <- keras_model_sequential(input_shape = c(784)) model |>   layer_dense(units = 32) |>   layer_activation('relu') |>   layer_dense(units = 10) |>   layer_activation('softmax')  model |> compile(   optimizer = 'rmsprop',   loss = 'categorical_crossentropy',   metrics = c('accuracy') )  model ## Model: \"sequential\" ## +---------------------------------+------------------------+---------------+ ## | Layer (type)                    | Output Shape           |       Param # | ## +=================================+========================+===============+ ## | dense (Dense)                   | (None, 32)             |        25,120 | ## +---------------------------------+------------------------+---------------+ ## | activation (Activation)         | (None, 32)             |             0 | ## +---------------------------------+------------------------+---------------+ ## | dense_1 (Dense)                 | (None, 10)             |           330 | ## +---------------------------------+------------------------+---------------+ ## | activation_1 (Activation)       | (None, 10)             |             0 | ## +---------------------------------+------------------------+---------------+ ##  Total params: 25,450 (99.41 KB) ##  Trainable params: 25,450 (99.41 KB) ##  Non-trainable params: 0 (0.00 B)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/keras_variable.html","id":null,"dir":"Reference","previous_headings":"","what":"Represents a backend-agnostic variable in Keras. — keras_variable","title":"Represents a backend-agnostic variable in Keras. — keras_variable","text":"Variable acts container state. holds tensor value can updated. JAX backend, variables used implement \"functionalization\", pattern lifting stateful operations piece computation turn stateless function.","code":""},{"path":"https://keras3.posit.co/dev/reference/keras_variable.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Represents a backend-agnostic variable in Keras. — keras_variable","text":"","code":"keras_variable(   initializer,   shape = NULL,   dtype = NULL,   trainable = TRUE,   autocast = TRUE,   aggregation = \"none\",   synchronization = \"auto\",   name = NULL,   ... )"},{"path":"https://keras3.posit.co/dev/reference/keras_variable.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Represents a backend-agnostic variable in Keras. — keras_variable","text":"initializer Initial value callable initialization. callable used, take arguments shape dtype. shape Optional. Tuple variable's shape. Required initializer callable. dtype Optional. Data type variable. Defaults global float dtype type (\"float32\" never configured). trainable Optional. Boolean indicating variable trainable. Defaults TRUE. autocast Optional. Boolean indicating whether variable supports autocasting. TRUE, layer may first convert variable compute data type accessed. Defaults TRUE. aggregation Optional string, one NULL, \"none\", \"mean\", \"sum\" \"only_first_replica\" specifying distributed variable aggregated. serves semantic annotation, taken account downstream backends users. Defaults \"none\". synchronization Optional string specifying distributed values synchronized. Defaults \"auto\". name Optional. unique name variable. Automatically generated set. ... Additional backend-specific keyword arguments forwarded keras$Variable().","code":""},{"path":"https://keras3.posit.co/dev/reference/keras_variable.html","id":"attributes","dir":"Reference","previous_headings":"","what":"Attributes","title":"Represents a backend-agnostic variable in Keras. — keras_variable","text":"shape: shape variable (tuple integers). ndim: number dimensions variable (integer). dtype: data type variable (string). trainable: Whether variable trainable (boolean). autocast: Whether variable supports autocasting (boolean). aggregation: distributed variable aggregated (string). synchronization: Strategy synchronizing variable across devices (string). value: current value variable (NumPy array tensor). name: name variable (string). path: path variable within Keras model layer (string).","code":""},{"path":"https://keras3.posit.co/dev/reference/keras_variable.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Represents a backend-agnostic variable in Keras. — keras_variable","text":"Initializing Variable NumPy array:   Using Keras initializer create Variable:           Marking Variable non-trainable:","code":"initial_array <- array(1, c(3, 3)) variable_from_array <- keras_variable(initializer = initial_array) variable_from_initializer <- keras_variable(   initializer = initializer_ones(),   shape = c(3, 3),   dtype = \"float32\" ) new_value <- array(0, c(3, 3)) variable_from_array$assign(new_value) ## tf.Tensor( ## [[0. 0. 0.] ##  [0. 0. 0.] ##  [0. 0. 0.]], shape=(3, 3), dtype=float64) # To modify a subset of values value <- variable_from_array$value value@r[1,] <- 99 invisible(variable_from_array$assign(value)) variable_from_array ## <Variable path=variable, shape=(3, 3), dtype=float64, value=[[99. 99. 99.] ##  [ 0.  0.  0.] ##  [ 0.  0.  0.]]> non_trainable_variable <- keras_variable(   initializer = array(1, c(3, 3)),   dtype = \"float32\",   trainable = FALSE )"},{"path":"https://keras3.posit.co/dev/reference/layer_activation.html","id":null,"dir":"Reference","previous_headings":"","what":"Applies an activation function to an output. — layer_activation","title":"Applies an activation function to an output. — layer_activation","text":"Applies activation function output.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_activation.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Applies an activation function to an output. — layer_activation","text":"","code":"layer_activation(object, activation, ...)"},{"path":"https://keras3.posit.co/dev/reference/layer_activation.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Applies an activation function to an output. — layer_activation","text":"object Object compose layer . tensor, array, sequential model. activation Activation function. callable, name activation keras3::activation_* namespace. ... Base layer keyword arguments, name dtype.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_activation.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Applies an activation function to an output. — layer_activation","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_activation.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Applies an activation function to an output. — layer_activation","text":"","code":"x <- array(c(-3, -1, 0, 2)) layer <- layer_activation(activation = 'relu') layer(x) ## tf.Tensor([0. 0. 0. 2.], shape=(4), dtype=float32) layer <- layer_activation(activation = activation_relu) layer(x) ## tf.Tensor([0. 0. 0. 2.], shape=(4), dtype=float32) layer <- layer_activation(activation = op_relu) layer(x) ## tf.Tensor([0. 0. 0. 2.], shape=(4), dtype=float32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_activation_elu.html","id":null,"dir":"Reference","previous_headings":"","what":"Applies an Exponential Linear Unit function to an output. — layer_activation_elu","title":"Applies an Exponential Linear Unit function to an output. — layer_activation_elu","text":"Formula:","code":"f(x) = alpha * (exp(x) - 1.) for x < 0 f(x) = x for x >= 0"},{"path":"https://keras3.posit.co/dev/reference/layer_activation_elu.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Applies an Exponential Linear Unit function to an output. — layer_activation_elu","text":"","code":"layer_activation_elu(object, alpha = 1, ...)"},{"path":"https://keras3.posit.co/dev/reference/layer_activation_elu.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Applies an Exponential Linear Unit function to an output. — layer_activation_elu","text":"object Object compose layer . tensor, array, sequential model. alpha float, slope negative section. Defaults 1.0. ... Base layer keyword arguments, name dtype.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_activation_elu.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Applies an Exponential Linear Unit function to an output. — layer_activation_elu","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_activation_leaky_relu.html","id":null,"dir":"Reference","previous_headings":"","what":"Leaky version of a Rectified Linear Unit activation layer. — layer_activation_leaky_relu","title":"Leaky version of a Rectified Linear Unit activation layer. — layer_activation_leaky_relu","text":"layer allows small gradient unit active. Formula:","code":"f <- function(x) ifelse(x >= 0, x, alpha * x)"},{"path":"https://keras3.posit.co/dev/reference/layer_activation_leaky_relu.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Leaky version of a Rectified Linear Unit activation layer. — layer_activation_leaky_relu","text":"","code":"layer_activation_leaky_relu(object, negative_slope = 0.3, ...)"},{"path":"https://keras3.posit.co/dev/reference/layer_activation_leaky_relu.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Leaky version of a Rectified Linear Unit activation layer. — layer_activation_leaky_relu","text":"object Object compose layer . tensor, array, sequential model. negative_slope Float >= 0.0. Negative slope coefficient. Defaults 0.3. ... Base layer keyword arguments, name dtype.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_activation_leaky_relu.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Leaky version of a Rectified Linear Unit activation layer. — layer_activation_leaky_relu","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_activation_leaky_relu.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Leaky version of a Rectified Linear Unit activation layer. — layer_activation_leaky_relu","text":"","code":"leaky_relu_layer <- layer_activation_leaky_relu(negative_slope=0.5) input <- array(c(-10, -5, 0.0, 5, 10)) result <- leaky_relu_layer(input) as.array(result) ## [1] -5.0 -2.5  0.0  5.0 10.0"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_activation_parametric_relu.html","id":null,"dir":"Reference","previous_headings":"","what":"Parametric Rectified Linear Unit activation layer. — layer_activation_parametric_relu","title":"Parametric Rectified Linear Unit activation layer. — layer_activation_parametric_relu","text":"Formula:   alpha learned array shape x.","code":"f <- function(x) ifelse(x >= 0, x, alpha * x)"},{"path":"https://keras3.posit.co/dev/reference/layer_activation_parametric_relu.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Parametric Rectified Linear Unit activation layer. — layer_activation_parametric_relu","text":"","code":"layer_activation_parametric_relu(   object,   alpha_initializer = \"Zeros\",   alpha_regularizer = NULL,   alpha_constraint = NULL,   shared_axes = NULL,   ... )"},{"path":"https://keras3.posit.co/dev/reference/layer_activation_parametric_relu.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Parametric Rectified Linear Unit activation layer. — layer_activation_parametric_relu","text":"object Object compose layer . tensor, array, sequential model. alpha_initializer Initializer function weights. alpha_regularizer Regularizer weights. alpha_constraint Constraint weights. shared_axes axes along share learnable parameters activation function. example, incoming feature maps 2D convolution output shape (batch, height, width, channels), wish share parameters across space filter one set parameters, set shared_axes=[1, 2]. ... Base layer keyword arguments, name dtype.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_activation_parametric_relu.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Parametric Rectified Linear Unit activation layer. — layer_activation_parametric_relu","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_activation_relu.html","id":null,"dir":"Reference","previous_headings":"","what":"Rectified Linear Unit activation function layer. — layer_activation_relu","title":"Rectified Linear Unit activation function layer. — layer_activation_relu","text":"Formula:","code":"f <- function(x, max_value = Inf, negative_slope = 0, threshold = 0) {  x <- max(x,0)  if (x >= max_value)    max_value  else if (threshold <= x && x < max_value)    x  else    negative_slope * (x - threshold) }"},{"path":"https://keras3.posit.co/dev/reference/layer_activation_relu.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Rectified Linear Unit activation function layer. — layer_activation_relu","text":"","code":"layer_activation_relu(   object,   max_value = NULL,   negative_slope = 0,   threshold = 0,   ... )"},{"path":"https://keras3.posit.co/dev/reference/layer_activation_relu.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Rectified Linear Unit activation function layer. — layer_activation_relu","text":"object Object compose layer . tensor, array, sequential model. max_value Float >= 0. Maximum activation value. NULL means unlimited. Defaults NULL. negative_slope Float >= 0. Negative slope coefficient. Defaults 0.0. threshold Float >= 0. Threshold value thresholded activation. Defaults 0.0. ... Base layer keyword arguments, name dtype.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_activation_relu.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Rectified Linear Unit activation function layer. — layer_activation_relu","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_activation_relu.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Rectified Linear Unit activation function layer. — layer_activation_relu","text":"","code":"relu_layer <- layer_activation_relu(max_value = 10,                                     negative_slope = 0.5,                                     threshold = 0) input <- array(c(-10, -5, 0.0, 5, 10)) result <- relu_layer(input) as.array(result) ## [1] -5.0 -2.5  0.0  5.0 10.0"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_activation_softmax.html","id":null,"dir":"Reference","previous_headings":"","what":"Softmax activation layer. — layer_activation_softmax","title":"Softmax activation layer. — layer_activation_softmax","text":"Formula:","code":"exp_x = exp(x - max(x)) f(x) = exp_x / sum(exp_x)"},{"path":"https://keras3.posit.co/dev/reference/layer_activation_softmax.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Softmax activation layer. — layer_activation_softmax","text":"","code":"layer_activation_softmax(object, axis = -1L, ...)"},{"path":"https://keras3.posit.co/dev/reference/layer_activation_softmax.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Softmax activation layer. — layer_activation_softmax","text":"object Object compose layer . tensor, array, sequential model. axis Integer, list Integers, axis along softmax normalization applied. ... Base layer keyword arguments, name dtype.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_activation_softmax.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Softmax activation layer. — layer_activation_softmax","text":"Softmaxed output shape inputs.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_activation_softmax.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Softmax activation layer. — layer_activation_softmax","text":"","code":"softmax_layer <- layer_activation_softmax() input <- op_array(c(1, 2, 1)) softmax_layer(input) ## tf.Tensor([0.21194157 0.5761169  0.21194157], shape=(3), dtype=float32)"},{"path":"https://keras3.posit.co/dev/reference/layer_activation_softmax.html","id":"call-arguments","dir":"Reference","previous_headings":"","what":"Call Arguments","title":"Softmax activation layer. — layer_activation_softmax","text":"inputs: inputs (logits) softmax layer. mask: boolean mask shape inputs. mask specifies 1 keep 0 mask. Defaults NULL.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_activity_regularization.html","id":null,"dir":"Reference","previous_headings":"","what":"Layer that applies an update to the cost function based input activity. — layer_activity_regularization","title":"Layer that applies an update to the cost function based input activity. — layer_activity_regularization","text":"Layer applies update cost function based input activity.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_activity_regularization.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Layer that applies an update to the cost function based input activity. — layer_activity_regularization","text":"","code":"layer_activity_regularization(object, l1 = 0, l2 = 0, ...)"},{"path":"https://keras3.posit.co/dev/reference/layer_activity_regularization.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Layer that applies an update to the cost function based input activity. — layer_activity_regularization","text":"object Object compose layer . tensor, array, sequential model. l1 L1 regularization factor (positive float). l2 L2 regularization factor (positive float). ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_activity_regularization.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Layer that applies an update to the cost function based input activity. — layer_activity_regularization","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_activity_regularization.html","id":"input-shape","dir":"Reference","previous_headings":"","what":"Input Shape","title":"Layer that applies an update to the cost function based input activity. — layer_activity_regularization","text":"Arbitrary. Use keyword argument input_shape (tuple integers, include samples axis) using layer first layer model.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_activity_regularization.html","id":"output-shape","dir":"Reference","previous_headings":"","what":"Output Shape","title":"Layer that applies an update to the cost function based input activity. — layer_activity_regularization","text":"shape input.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_add.html","id":null,"dir":"Reference","previous_headings":"","what":"Performs elementwise addition operation. — layer_add","title":"Performs elementwise addition operation. — layer_add","text":"takes input list tensors, shape, returns single tensor (also shape).","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_add.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Performs elementwise addition operation. — layer_add","text":"","code":"layer_add(inputs, ...)"},{"path":"https://keras3.posit.co/dev/reference/layer_add.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Performs elementwise addition operation. — layer_add","text":"inputs layers combine ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_add.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Performs elementwise addition operation. — layer_add","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_add.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Performs elementwise addition operation. — layer_add","text":"Usage Keras model:","code":"input_shape <- c(1, 2, 3) x1 <- op_ones(input_shape) x2 <- op_ones(input_shape) layer_add(x1, x2) ## tf.Tensor( ## [[[2. 2. 2.] ##   [2. 2. 2.]]], shape=(1, 2, 3), dtype=float32) input1 <- layer_input(shape = c(16)) x1 <- input1 |> layer_dense(8, activation = 'relu')  input2 <- layer_input(shape = c(32)) x2 <- input2 |> layer_dense(8, activation = 'relu')  # equivalent to `added = layer_add([x1, x2))` added <- layer_add(x1, x2) output <- added |> layer_dense(4)  model <- keras_model(inputs = c(input1, input2), outputs = output)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_additive_attention.html","id":null,"dir":"Reference","previous_headings":"","what":"Additive attention layer, a.k.a. Bahdanau-style attention. — layer_additive_attention","title":"Additive attention layer, a.k.a. Bahdanau-style attention. — layer_additive_attention","text":"Inputs list 2 3 elements: query tensor shape (batch_size, Tq, dim). value tensor shape (batch_size, Tv, dim). optional key tensor shape (batch_size, Tv, dim). none supplied, value used key. calculation follows steps: Calculate attention scores using query key shape (batch_size, Tq, Tv) non-linear sum scores = reduce_sum(tanh(query + key), axis=-1). Use scores calculate softmax distribution shape (batch_size, Tq, Tv). Use softmax distribution create linear combination value shape (batch_size, Tq, dim).","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_additive_attention.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Additive attention layer, a.k.a. Bahdanau-style attention. — layer_additive_attention","text":"","code":"layer_additive_attention(object, use_scale = TRUE, dropout = 0, ...)"},{"path":"https://keras3.posit.co/dev/reference/layer_additive_attention.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Additive attention layer, a.k.a. Bahdanau-style attention. — layer_additive_attention","text":"object Object compose layer . tensor, array, sequential model. use_scale TRUE, create scalar variable scale attention scores. dropout Float 0 1. Fraction units drop attention scores. Defaults 0.0. ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_additive_attention.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Additive attention layer, a.k.a. Bahdanau-style attention. — layer_additive_attention","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_additive_attention.html","id":"call-arguments","dir":"Reference","previous_headings":"","what":"Call Arguments","title":"Additive attention layer, a.k.a. Bahdanau-style attention. — layer_additive_attention","text":"inputs: List following tensors: query: Query tensor shape (batch_size, Tq, dim). value: Value tensor shape (batch_size, Tv, dim). key: Optional key tensor shape (batch_size, Tv, dim). given, use value key value, common case. mask: List following tensors: query_mask: boolean mask tensor shape (batch_size, Tq). given, output zero positions mask==FALSE. value_mask: boolean mask tensor shape (batch_size, Tv). given, apply mask values positions mask==FALSE contribute result. return_attention_scores: bool, TRUE, returns attention scores (masking softmax) additional output argument. training: Python boolean indicating whether layer behave training mode (adding dropout) inference mode (dropout). use_causal_mask: Boolean. Set TRUE decoder self-attention. Adds mask position attend positions j > . prevents flow information future towards past. Defaults FALSE.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_additive_attention.html","id":"output","dir":"Reference","previous_headings":"","what":"Output","title":"Additive attention layer, a.k.a. Bahdanau-style attention. — layer_additive_attention","text":"Attention outputs shape (batch_size, Tq, dim). (Optional) Attention scores masking softmax shape (batch_size, Tq, Tv).","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_alpha_dropout.html","id":null,"dir":"Reference","previous_headings":"","what":"Applies Alpha Dropout to the input. — layer_alpha_dropout","title":"Applies Alpha Dropout to the input. — layer_alpha_dropout","text":"Alpha Dropout Dropout keeps mean variance inputs original values, order ensure self-normalizing property even dropout. Alpha Dropout fits well Scaled Exponential Linear Units (SELU) randomly setting activations negative saturation value.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_alpha_dropout.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Applies Alpha Dropout to the input. — layer_alpha_dropout","text":"","code":"layer_alpha_dropout(object, rate, noise_shape = NULL, seed = NULL, ...)"},{"path":"https://keras3.posit.co/dev/reference/layer_alpha_dropout.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Applies Alpha Dropout to the input. — layer_alpha_dropout","text":"object Object compose layer . tensor, array, sequential model. rate Float 0 1. multiplicative noise standard deviation sqrt(rate / (1 - rate)). noise_shape 1D integer tensor representing shape binary alpha dropout mask multiplied input. instance, inputs shape (batch_size, timesteps, features) want alpha dropout mask timesteps, can use noise_shape = (batch_size, 1, features). seed integer use random seed. ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_alpha_dropout.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Applies Alpha Dropout to the input. — layer_alpha_dropout","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_alpha_dropout.html","id":"call-arguments","dir":"Reference","previous_headings":"","what":"Call Arguments","title":"Applies Alpha Dropout to the input. — layer_alpha_dropout","text":"inputs: Input tensor (rank). training: R boolean indicating whether layer behave training mode (adding alpha dropout) inference mode (nothing).","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_attention.html","id":null,"dir":"Reference","previous_headings":"","what":"Dot-product attention layer, a.k.a. Luong-style attention. — layer_attention","title":"Dot-product attention layer, a.k.a. Luong-style attention. — layer_attention","text":"Inputs list 2 3 elements: query tensor shape (batch_size, Tq, dim). value tensor shape (batch_size, Tv, dim). optional key tensor shape (batch_size, Tv, dim). none supplied, value used key. calculation follows steps: Calculate attention scores using query key shape (batch_size, Tq, Tv). Use scores calculate softmax distribution shape (batch_size, Tq, Tv). Use softmax distribution create linear combination value shape (batch_size, Tq, dim).","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_attention.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Dot-product attention layer, a.k.a. Luong-style attention. — layer_attention","text":"","code":"layer_attention(   object,   use_scale = FALSE,   score_mode = \"dot\",   dropout = 0,   seed = NULL,   ... )"},{"path":"https://keras3.posit.co/dev/reference/layer_attention.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Dot-product attention layer, a.k.a. Luong-style attention. — layer_attention","text":"object Object compose layer . tensor, array, sequential model. use_scale TRUE, create scalar variable scale attention scores. score_mode Function use compute attention scores, one {\"dot\", \"concat\"}. \"dot\" refers dot product query key vectors. \"concat\" refers hyperbolic tangent concatenation query key vectors. dropout Float 0 1. Fraction units drop attention scores. Defaults 0.0. seed integer use random seed case dropout. ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_attention.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Dot-product attention layer, a.k.a. Luong-style attention. — layer_attention","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_attention.html","id":"call-arguments","dir":"Reference","previous_headings":"","what":"Call Arguments","title":"Dot-product attention layer, a.k.a. Luong-style attention. — layer_attention","text":"inputs: List following tensors: query: Query tensor shape (batch_size, Tq, dim). value: Value tensor shape (batch_size, Tv, dim). key: Optional key tensor shape (batch_size, Tv, dim). given, use value key value, common case. mask: List following tensors: query_mask: boolean mask tensor shape (batch_size, Tq). given, output zero positions mask==FALSE. value_mask: boolean mask tensor shape (batch_size, Tv). given, apply mask values positions mask==FALSE contribute result. return_attention_scores: bool, TRUE, returns attention scores (masking softmax) additional output argument. training: Python boolean indicating whether layer behave training mode (adding dropout) inference mode (dropout). use_causal_mask: Boolean. Set TRUE decoder self-attention. Adds mask position attend positions j > . prevents flow information future towards past. Defaults FALSE.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_attention.html","id":"output","dir":"Reference","previous_headings":"","what":"Output","title":"Dot-product attention layer, a.k.a. Luong-style attention. — layer_attention","text":"Attention outputs shape (batch_size, Tq, dim). (Optional) Attention scores masking softmax shape (batch_size, Tq, Tv).","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_aug_mix.html","id":null,"dir":"Reference","previous_headings":"","what":"Performs the AugMix data augmentation technique. — layer_aug_mix","title":"Performs the AugMix data augmentation technique. — layer_aug_mix","text":"AugMix aims produce images variety preserving image semantics local statistics. augmentation process, augmentation applied across images batch num_chains different ways, chain consisting chain_depth augmentations.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_aug_mix.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Performs the AugMix data augmentation technique. — layer_aug_mix","text":"","code":"layer_aug_mix(   object,   value_range = c(0L, 255L),   num_chains = 3L,   chain_depth = 3L,   factor = 0.3,   alpha = 1,   all_ops = TRUE,   interpolation = \"bilinear\",   seed = NULL,   data_format = NULL,   ... )"},{"path":"https://keras3.posit.co/dev/reference/layer_aug_mix.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Performs the AugMix data augmentation technique. — layer_aug_mix","text":"object Object compose layer . tensor, array, sequential model. value_range range values incoming images . Represented two number tuple written (low, high). typically either (0, 1) (0, 255) depending preprocessing pipeline set . num_chains integer representing number different chains mixed, defaults 3. chain_depth integer representing maximum number transformations applied chain. actual number transformations chain sampled randomly range [0, chain_depth]. Defaults 3. factor strength augmentation normalized value 0 1. Default 0.3. alpha float value used probability coefficients Beta Dirichlet distributions, defaults 1.0. all_ops Use operations (including random_brightness, random_color_degeneration, random_contrast random_sharpness). Default True. interpolation interpolation method use resizing operations. Options include \"nearest\", \"bilinear\". Default \"bilinear\". seed Integer. Used create random seed. data_format string, either \"channels_last\" \"channels_first\". ordering dimensions inputs. \"channels_last\" corresponds inputs shape (batch, height, width, channels) \"channels_first\" corresponds inputs shape (batch, channels, height, width). defaults image_data_format value found Keras config file ~/.keras/keras.json. never set , \"channels_last\". ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_aug_mix.html","id":"references","dir":"Reference","previous_headings":"","what":"References","title":"Performs the AugMix data augmentation technique. — layer_aug_mix","text":"AugMix paper Official Code","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_auto_contrast.html","id":null,"dir":"Reference","previous_headings":"","what":"Performs the auto-contrast operation on an image. — layer_auto_contrast","title":"Performs the auto-contrast operation on an image. — layer_auto_contrast","text":"Auto contrast stretches values image across entire available value_range. makes differences pixels obvious. example image values [0, 1] range [0, 255], auto contrast change 1 values 255. layer active training inference time.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_auto_contrast.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Performs the auto-contrast operation on an image. — layer_auto_contrast","text":"","code":"layer_auto_contrast(object, value_range = tuple(0L, 255L), ...)"},{"path":"https://keras3.posit.co/dev/reference/layer_auto_contrast.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Performs the auto-contrast operation on an image. — layer_auto_contrast","text":"object Object compose layer . tensor, array, sequential model. value_range Range values incoming images . Represented two number tuple written (low, high). typically either (0, 1) (0, 255) depending preprocessing pipeline set . Defaults (0, 255). ... forward/backward compatability.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_average.html","id":null,"dir":"Reference","previous_headings":"","what":"Averages a list of inputs element-wise.. — layer_average","title":"Averages a list of inputs element-wise.. — layer_average","text":"takes input list tensors, shape, returns single tensor (also shape).","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_average.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Averages a list of inputs element-wise.. — layer_average","text":"","code":"layer_average(inputs, ...)"},{"path":"https://keras3.posit.co/dev/reference/layer_average.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Averages a list of inputs element-wise.. — layer_average","text":"inputs layers combine ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_average.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Averages a list of inputs element-wise.. — layer_average","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_average.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Averages a list of inputs element-wise.. — layer_average","text":"Usage Keras model:","code":"input_shape <- c(1, 2, 3) x1 <- op_ones(input_shape) x2 <- op_zeros(input_shape) layer_average(x1, x2) ## tf.Tensor( ## [[[0.5 0.5 0.5] ##   [0.5 0.5 0.5]]], shape=(1, 2, 3), dtype=float32) input1 <- layer_input(shape = c(16)) x1 <- input1 |> layer_dense(8, activation = 'relu')  input2 <- layer_input(shape = c(32)) x2 <- input2 |> layer_dense(8, activation = 'relu')  added <- layer_average(x1, x2) output <- added |> layer_dense(4)  model <- keras_model(inputs = c(input1, input2), outputs = output)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_average_pooling_1d.html","id":null,"dir":"Reference","previous_headings":"","what":"Average pooling for temporal data. — layer_average_pooling_1d","title":"Average pooling for temporal data. — layer_average_pooling_1d","text":"Downsamples input representation taking average value window defined pool_size. window shifted strides.  resulting output using \"valid\" padding option shape : output_shape = (input_shape - pool_size + 1) / strides) resulting output shape using \"\" padding option : output_shape = input_shape / strides","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_average_pooling_1d.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Average pooling for temporal data. — layer_average_pooling_1d","text":"","code":"layer_average_pooling_1d(   object,   pool_size,   strides = NULL,   padding = \"valid\",   data_format = NULL,   name = NULL,   ... )"},{"path":"https://keras3.posit.co/dev/reference/layer_average_pooling_1d.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Average pooling for temporal data. — layer_average_pooling_1d","text":"object Object compose layer . tensor, array, sequential model. pool_size int, size max pooling window. strides int NULL. Specifies much pooling window moves pooling step. NULL, default pool_size. padding string, either \"valid\" \"\" (case-insensitive). \"valid\" means padding. \"\" results padding evenly left/right /input output height/width dimension input. data_format string, either \"channels_last\" \"channels_first\". ordering dimensions inputs. \"channels_last\" corresponds inputs shape (batch, steps, features) \"channels_first\" corresponds inputs shape (batch, features, steps). defaults image_data_format value found Keras config file ~/.keras/keras.json. never set , \"channels_last\". name String, name object ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_average_pooling_1d.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Average pooling for temporal data. — layer_average_pooling_1d","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_average_pooling_1d.html","id":"input-shape","dir":"Reference","previous_headings":"","what":"Input Shape","title":"Average pooling for temporal data. — layer_average_pooling_1d","text":"data_format=\"channels_last\": 3D tensor shape (batch_size, steps, features). data_format=\"channels_first\": 3D tensor shape (batch_size, features, steps).","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_average_pooling_1d.html","id":"output-shape","dir":"Reference","previous_headings":"","what":"Output Shape","title":"Average pooling for temporal data. — layer_average_pooling_1d","text":"data_format=\"channels_last\": 3D tensor shape (batch_size, downsampled_steps, features). data_format=\"channels_first\": 3D tensor shape (batch_size, features, downsampled_steps).","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_average_pooling_1d.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Average pooling for temporal data. — layer_average_pooling_1d","text":"strides=1 padding=\"valid\":     strides=2 padding=\"valid\":     strides=1 padding=\"\":","code":"x <- op_array(c(1., 2., 3., 4., 5.)) |> op_reshape(c(1, 5, 1)) output <- x |>   layer_average_pooling_1d(pool_size = 2,                            strides = 1,                            padding = \"valid\") output ## tf.Tensor( ## [[[1.5] ##   [2.5] ##   [3.5] ##   [4.5]]], shape=(1, 4, 1), dtype=float32) x <- op_array(c(1., 2., 3., 4., 5.)) |> op_reshape(c(1, 5, 1)) output <- x |>   layer_average_pooling_1d(pool_size = 2,                            strides = 2,                            padding = \"valid\") output ## tf.Tensor( ## [[[1.5] ##   [3.5]]], shape=(1, 2, 1), dtype=float32) x <- op_array(c(1., 2., 3., 4., 5.)) |> op_reshape(c(1, 5, 1)) output <- x |>   layer_average_pooling_1d(pool_size = 2,                            strides = 1,                            padding = \"same\") output ## tf.Tensor( ## [[[1.5] ##   [2.5] ##   [3.5] ##   [4.5] ##   [5. ]]], shape=(1, 5, 1), dtype=float32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_average_pooling_2d.html","id":null,"dir":"Reference","previous_headings":"","what":"Average pooling operation for 2D spatial data. — layer_average_pooling_2d","title":"Average pooling operation for 2D spatial data. — layer_average_pooling_2d","text":"Downsamples input along spatial dimensions (height width) taking average value input window (size defined pool_size) channel input. window shifted strides along dimension. resulting output using \"valid\" padding option spatial shape (number rows columns) : output_shape = math.floor((input_shape - pool_size) / strides) + 1 (input_shape >= pool_size) resulting output shape using \"\" padding option : output_shape = input_shape","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_average_pooling_2d.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Average pooling operation for 2D spatial data. — layer_average_pooling_2d","text":"","code":"layer_average_pooling_2d(   object,   pool_size,   strides = NULL,   padding = \"valid\",   data_format = NULL,   name = NULL,   ... )"},{"path":"https://keras3.posit.co/dev/reference/layer_average_pooling_2d.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Average pooling operation for 2D spatial data. — layer_average_pooling_2d","text":"object Object compose layer . tensor, array, sequential model. pool_size int list 2 integers, factors downscale (dim1, dim2). one integer specified, window length used dimensions. strides int list 2 integers, NULL. Strides values. NULL, default pool_size. one int specified, stride size used dimensions. padding string, either \"valid\" \"\" (case-insensitive). \"valid\" means padding. \"\" results padding evenly left/right /input output height/width dimension input. data_format string, either \"channels_last\" \"channels_first\". ordering dimensions inputs. \"channels_last\" corresponds inputs shape (batch, height, width, channels) \"channels_first\" corresponds inputs shape (batch, channels, height, width). defaults image_data_format value found Keras config file ~/.keras/keras.json. never set , \"channels_last\". name String, name object ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_average_pooling_2d.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Average pooling operation for 2D spatial data. — layer_average_pooling_2d","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_average_pooling_2d.html","id":"input-shape","dir":"Reference","previous_headings":"","what":"Input Shape","title":"Average pooling operation for 2D spatial data. — layer_average_pooling_2d","text":"data_format=\"channels_last\": 4D tensor shape (batch_size, height, width, channels). data_format=\"channels_first\": 4D tensor shape (batch_size, channels, height, width).","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_average_pooling_2d.html","id":"output-shape","dir":"Reference","previous_headings":"","what":"Output Shape","title":"Average pooling operation for 2D spatial data. — layer_average_pooling_2d","text":"data_format=\"channels_last\": 4D tensor shape (batch_size, pooled_height, pooled_width, channels). data_format=\"channels_first\": 4D tensor shape (batch_size, channels, pooled_height, pooled_width).","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_average_pooling_2d.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Average pooling operation for 2D spatial data. — layer_average_pooling_2d","text":"strides=(1, 1) padding=\"valid\":     strides=(2, 2) padding=\"valid\":     stride=(1, 1) padding=\"\":","code":"x <- op_array(1:9, \"float32\") |> op_reshape(c(1, 3, 3, 1)) output <- x |>   layer_average_pooling_2d(pool_size = c(2, 2),                            strides = c(1, 1),                            padding = \"valid\") output ## tf.Tensor( ## [[[[3.] ##    [4.]] ## ##   [[6.] ##    [7.]]]], shape=(1, 2, 2, 1), dtype=float32) x <- op_array(1:12, \"float32\") |> op_reshape(c(1, 3, 4, 1)) output <- x |>   layer_average_pooling_2d(pool_size = c(2, 2),                            strides = c(2, 2),                            padding = \"valid\") output ## tf.Tensor( ## [[[[3.5] ##    [5.5]]]], shape=(1, 1, 2, 1), dtype=float32) x <- op_array(1:9, \"float32\") |> op_reshape(c(1, 3, 3, 1)) output <- x |>   layer_average_pooling_2d(pool_size = c(2, 2),                            strides = c(1, 1),                            padding = \"same\") output ## tf.Tensor( ## [[[[3. ] ##    [4. ] ##    [4.5]] ## ##   [[6. ] ##    [7. ] ##    [7.5]] ## ##   [[7.5] ##    [8.5] ##    [9. ]]]], shape=(1, 3, 3, 1), dtype=float32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_average_pooling_3d.html","id":null,"dir":"Reference","previous_headings":"","what":"Average pooling operation for 3D data (spatial or spatio-temporal). — layer_average_pooling_3d","title":"Average pooling operation for 3D data (spatial or spatio-temporal). — layer_average_pooling_3d","text":"Downsamples input along spatial dimensions (depth, height, width) taking average value input window (size defined pool_size) channel input. window shifted strides along dimension.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_average_pooling_3d.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Average pooling operation for 3D data (spatial or spatio-temporal). — layer_average_pooling_3d","text":"","code":"layer_average_pooling_3d(   object,   pool_size,   strides = NULL,   padding = \"valid\",   data_format = NULL,   name = NULL,   ... )"},{"path":"https://keras3.posit.co/dev/reference/layer_average_pooling_3d.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Average pooling operation for 3D data (spatial or spatio-temporal). — layer_average_pooling_3d","text":"object Object compose layer . tensor, array, sequential model. pool_size int list 3 integers, factors downscale (dim1, dim2, dim3). one integer specified, window length used dimensions. strides int list 3 integers, NULL. Strides values. NULL, default pool_size. one int specified, stride size used dimensions. padding string, either \"valid\" \"\" (case-insensitive). \"valid\" means padding. \"\" results padding evenly left/right /input output height/width dimension input. data_format string, either \"channels_last\" \"channels_first\". ordering dimensions inputs. \"channels_last\" corresponds inputs shape (batch, spatial_dim1, spatial_dim2, spatial_dim3, channels) \"channels_first\" corresponds inputs shape (batch, channels, spatial_dim1, spatial_dim2, spatial_dim3). defaults image_data_format value found Keras config file ~/.keras/keras.json. never set , \"channels_last\". name String, name object ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_average_pooling_3d.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Average pooling operation for 3D data (spatial or spatio-temporal). — layer_average_pooling_3d","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_average_pooling_3d.html","id":"input-shape","dir":"Reference","previous_headings":"","what":"Input Shape","title":"Average pooling operation for 3D data (spatial or spatio-temporal). — layer_average_pooling_3d","text":"data_format=\"channels_last\": 5D tensor shape: (batch_size, spatial_dim1, spatial_dim2, spatial_dim3, channels) data_format=\"channels_first\": 5D tensor shape: (batch_size, channels, spatial_dim1, spatial_dim2, spatial_dim3)","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_average_pooling_3d.html","id":"output-shape","dir":"Reference","previous_headings":"","what":"Output Shape","title":"Average pooling operation for 3D data (spatial or spatio-temporal). — layer_average_pooling_3d","text":"data_format=\"channels_last\": 5D tensor shape: (batch_size, pooled_dim1, pooled_dim2, pooled_dim3, channels) data_format=\"channels_first\": 5D tensor shape: (batch_size, channels, pooled_dim1, pooled_dim2, pooled_dim3)","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_average_pooling_3d.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Average pooling operation for 3D data (spatial or spatio-temporal). — layer_average_pooling_3d","text":"","code":"depth <- height <- width <- 30 channels <- 3  inputs <- layer_input(shape = c(depth, height, width, channels)) outputs <- inputs |> layer_average_pooling_3d(pool_size = 3) outputs # Shape: (batch_size, 10, 10, 10, 3) ## <KerasTensor shape=(None, 10, 10, 10, 3), dtype=float32, sparse=False, ragged=False, name=keras_tensor_1>"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_batch_normalization.html","id":null,"dir":"Reference","previous_headings":"","what":"Layer that normalizes its inputs. — layer_batch_normalization","title":"Layer that normalizes its inputs. — layer_batch_normalization","text":"Batch normalization applies transformation maintains mean output close 0 output standard deviation close 1. Importantly, batch normalization works differently training inference. training (.e. using fit() calling layer/model argument training = TRUE), layer normalizes output using mean standard deviation current batch inputs. say, channel normalized, layer returns gamma * (batch - mean(batch)) / sqrt(var(batch) + epsilon) + beta, : epsilon small constant (configurable part constructor arguments) gamma learned scaling factor (initialized 1), can disabled passing scale = FALSE constructor. beta learned offset factor (initialized 0), can disabled passing center = FALSE constructor. inference (.e. using evaluate() predict() calling layer/model argument training = FALSE (default), layer normalizes output using moving average mean standard deviation batches seen training. say, returns gamma * (batch - self$moving_mean) / sqrt(self$moving_var+epsilon) + beta. self$moving_mean self$moving_var non-trainable variables updated time layer called training mode, : moving_mean = moving_mean * momentum + mean(batch) * (1 - momentum) moving_var = moving_var * momentum + var(batch) * (1 - momentum) , layer normalize inputs inference trained data similar statistics inference data. setting layer$trainable <- FALSE BatchNormalization layer: meaning setting layer$trainable <- FALSE freeze layer, .e. internal state change training: trainable weights updated fit() train_on_batch(), state updates run. Usually, necessarily mean layer run inference mode (normally controlled training argument can passed calling layer). \"Frozen state\" \"inference mode\" two separate concepts. However, case BatchNormalization layer, setting trainable <- FALSE layer means layer subsequently run inference mode (meaning use moving mean moving variance normalize current batch, rather using mean variance current batch). Note : Setting trainable model containing layers recursively set trainable value inner layers. value trainable attribute changed calling compile() model, new value take effect model compile() called .","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_batch_normalization.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Layer that normalizes its inputs. — layer_batch_normalization","text":"","code":"layer_batch_normalization(   object,   axis = -1L,   momentum = 0.99,   epsilon = 0.001,   center = TRUE,   scale = TRUE,   beta_initializer = \"zeros\",   gamma_initializer = \"ones\",   moving_mean_initializer = \"zeros\",   moving_variance_initializer = \"ones\",   beta_regularizer = NULL,   gamma_regularizer = NULL,   beta_constraint = NULL,   gamma_constraint = NULL,   synchronized = FALSE,   ... )"},{"path":"https://keras3.posit.co/dev/reference/layer_batch_normalization.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Layer that normalizes its inputs. — layer_batch_normalization","text":"object Object compose layer . tensor, array, sequential model. axis Integer, axis normalized (typically features axis). instance, Conv2D layer data_format = \"channels_first\", use axis = 2. momentum Momentum moving average. epsilon Small float added variance avoid dividing zero. center TRUE, add offset beta normalized tensor. FALSE, beta ignored. scale TRUE, multiply gamma. FALSE, gamma used. next layer linear can disabled since scaling done next layer. beta_initializer Initializer beta weight. gamma_initializer Initializer gamma weight. moving_mean_initializer Initializer moving mean. moving_variance_initializer Initializer moving variance. beta_regularizer Optional regularizer beta weight. gamma_regularizer Optional regularizer gamma weight. beta_constraint Optional constraint beta weight. gamma_constraint Optional constraint gamma weight. synchronized applicable TensorFlow backend. TRUE, synchronizes global batch statistics (mean variance) layer across devices training step distributed training strategy. FALSE, replica uses local batch statistics. ... Base layer keyword arguments (e.g. name dtype).","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_batch_normalization.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Layer that normalizes its inputs. — layer_batch_normalization","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_batch_normalization.html","id":"call-arguments","dir":"Reference","previous_headings":"","what":"Call Arguments","title":"Layer that normalizes its inputs. — layer_batch_normalization","text":"inputs: Input tensor (rank). training: R boolean indicating whether layer behave training mode inference mode. training = TRUE: layer normalize inputs using mean variance current batch inputs. training = FALSE: layer normalize inputs using mean variance moving statistics, learned training. mask: Binary tensor shape broadcastable inputs tensor, TRUE values indicating positions mean variance computed. Masked elements current inputs taken account mean variance computation training. prior unmasked element values taken account momentum expires.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_batch_normalization.html","id":"reference","dir":"Reference","previous_headings":"","what":"Reference","title":"Layer that normalizes its inputs. — layer_batch_normalization","text":"Ioffe Szegedy, 2015.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_bidirectional.html","id":null,"dir":"Reference","previous_headings":"","what":"Bidirectional wrapper for RNNs. — layer_bidirectional","title":"Bidirectional wrapper for RNNs. — layer_bidirectional","text":"Bidirectional wrapper RNNs.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_bidirectional.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Bidirectional wrapper for RNNs. — layer_bidirectional","text":"","code":"layer_bidirectional(   object,   layer,   merge_mode = \"concat\",   weights = NULL,   backward_layer = NULL,   ... )"},{"path":"https://keras3.posit.co/dev/reference/layer_bidirectional.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Bidirectional wrapper for RNNs. — layer_bidirectional","text":"object Object compose layer . tensor, array, sequential model. layer RNN instance, layer_lstm() layer_gru(). also Layer() instance meets following criteria: sequence-processing layer (accepts 3D+ inputs). go_backwards, return_sequences return_state attribute (semantics RNN class). input_spec attribute. Implement serialization via get_config() from_config(). Note recommended way create new RNN layers write custom RNN cell use layer_rnn(), instead subclassing Layer() directly. return_sequences TRUE, output masked timestep zero regardless layer's original zero_output_for_mask value. merge_mode Mode outputs forward backward RNNs combined. One {\"sum\", \"mul\", \"concat\", \"ave\", NULL}. NULL, outputs combined, returned list. Defaults \"concat\". weights see description backward_layer Optional RNN, Layer() instance used handle backwards input processing. backward_layer provided, layer instance passed layer argument used generate backward layer automatically. Note provided backward_layer layer properties matching layer argument, particular values stateful, return_states, return_sequences, etc. addition, backward_layer layer different go_backwards argument values. ValueError raised requirements met. ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_bidirectional.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Bidirectional wrapper for RNNs. — layer_bidirectional","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_bidirectional.html","id":"call-arguments","dir":"Reference","previous_headings":"","what":"Call Arguments","title":"Bidirectional wrapper for RNNs. — layer_bidirectional","text":"call arguments layer wrapped RNN layer. Beware passing initial_state argument call layer, first half list elements initial_state list passed forward RNN call last half list elements passed backward RNN call.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_bidirectional.html","id":"note","dir":"Reference","previous_headings":"","what":"Note","title":"Bidirectional wrapper for RNNs. — layer_bidirectional","text":"instantiating Bidirectional layer existing RNN layer instance reuse weights state RNN layer instance – Bidirectional layer freshly initialized weights.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_bidirectional.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Bidirectional wrapper for RNNs. — layer_bidirectional","text":"","code":"model <- keras_model_sequential(input_shape = c(5, 10)) %>%   layer_bidirectional(layer_lstm(units = 10, return_sequences = TRUE)) %>%   layer_bidirectional(layer_lstm(units = 10)) %>%   layer_dense(5, activation = \"softmax\")  model %>% compile(loss = \"categorical_crossentropy\",                   optimizer = \"rmsprop\")  # With custom backward layer forward_layer <- layer_lstm(units = 10, return_sequences = TRUE) backward_layer <- layer_lstm(units = 10, activation = \"relu\",                              return_sequences = TRUE, go_backwards = TRUE)  model <- keras_model_sequential(input_shape = c(5, 10)) %>%   bidirectional(forward_layer, backward_layer = backward_layer) %>%   layer_dense(5, activation = \"softmax\")  model %>% compile(loss = \"categorical_crossentropy\",                   optimizer = \"rmsprop\")"},{"path":"https://keras3.posit.co/dev/reference/layer_bidirectional.html","id":"states","dir":"Reference","previous_headings":"","what":"States","title":"Bidirectional wrapper for RNNs. — layer_bidirectional","text":"Bidirectional layer instance property states, can access layer$states. can also reset states using reset_state()","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_category_encoding.html","id":null,"dir":"Reference","previous_headings":"","what":"A preprocessing layer which encodes integer features. — layer_category_encoding","title":"A preprocessing layer which encodes integer features. — layer_category_encoding","text":"layer provides options condensing data categorical encoding total number tokens known advance. accepts integer values inputs, outputs dense sparse representation inputs. integer inputs total number tokens known, use layer_integer_lookup() instead. Note: layer safe use inside tf.data pipeline (independently backend using).","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_category_encoding.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"A preprocessing layer which encodes integer features. — layer_category_encoding","text":"","code":"layer_category_encoding(   object,   num_tokens = NULL,   output_mode = \"multi_hot\",   sparse = FALSE,   ... )"},{"path":"https://keras3.posit.co/dev/reference/layer_category_encoding.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"A preprocessing layer which encodes integer features. — layer_category_encoding","text":"object Object compose layer . tensor, array, sequential model. num_tokens total number tokens layer support. inputs layer must integers range 0 <= value < num_tokens, error thrown. output_mode Specification output layer. Values can \"one_hot\", \"multi_hot\" \"count\", configuring layer follows: - \"one_hot\": Encodes individual element input array num_tokens size, containing 1 element index. last dimension size 1, encode dimension. last dimension size 1, append new dimension encoded output. - \"multi_hot\": Encodes sample input single array num_tokens size, containing 1 vocabulary term present sample. Treats last dimension sample dimension, input shape (..., sample_length), output shape (..., num_tokens). - \"count\": Like \"multi_hot\", int array contains count number times token index appeared sample. output modes, currently output rank 2 supported. Defaults \"multi_hot\". sparse Whether return sparse tensor; backends support sparse tensors. ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_category_encoding.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"A preprocessing layer which encodes integer features. — layer_category_encoding","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_category_encoding.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"A preprocessing layer which encodes integer features. — layer_category_encoding","text":"One-hot encoding data     Multi-hot encoding data     Using weighted inputs \"count\" mode","code":"layer <- layer_category_encoding(num_tokens = 4, output_mode = \"one_hot\") x <- op_array(c(3, 2, 0, 1), \"int32\") layer(x) ## tf.Tensor( ## [[0. 0. 0. 1.] ##  [0. 0. 1. 0.] ##  [1. 0. 0. 0.] ##  [0. 1. 0. 0.]], shape=(4, 4), dtype=float32) layer <- layer_category_encoding(num_tokens = 4, output_mode = \"multi_hot\") x <- op_array(rbind(c(0, 1),                    c(0, 0),                    c(1, 2),                    c(3, 1)), \"int32\") layer(x) ## tf.Tensor( ## [[1. 1. 0. 0.] ##  [1. 0. 0. 0.] ##  [0. 1. 1. 0.] ##  [0. 1. 0. 1.]], shape=(4, 4), dtype=float32) layer <- layer_category_encoding(num_tokens = 4, output_mode = \"count\") count_weights <- op_array(rbind(c(.1, .2),                                c(.1, .1),                                c(.2, .3),                                c(.4, .2))) x <- op_array(rbind(c(0, 1),                    c(0, 0),                    c(1, 2),                    c(3, 1)), \"int32\") layer(x, count_weights = count_weights) #   array([[01, 02, 0. , 0. ], #          [02, 0. , 0. , 0. ], #          [0. , 02, 03, 0. ], #          [0. , 02, 0. , 04]]>"},{"path":"https://keras3.posit.co/dev/reference/layer_category_encoding.html","id":"call-arguments","dir":"Reference","previous_headings":"","what":"Call Arguments","title":"A preprocessing layer which encodes integer features. — layer_category_encoding","text":"inputs: 1D 2D tensor integer inputs. count_weights: tensor shape inputs indicating weight sample value summing count mode. used \"multi_hot\" \"one_hot\" modes.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_center_crop.html","id":null,"dir":"Reference","previous_headings":"","what":"A preprocessing layer which crops images. — layer_center_crop","title":"A preprocessing layer which crops images. — layer_center_crop","text":"layers crops central portion images target size. image smaller target size, resized cropped return largest possible window image matches target aspect ratio. Input pixel values can range (e.g. [0., 1.) [0, 255]).","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_center_crop.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"A preprocessing layer which crops images. — layer_center_crop","text":"","code":"layer_center_crop(object, height, width, data_format = NULL, ...)"},{"path":"https://keras3.posit.co/dev/reference/layer_center_crop.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"A preprocessing layer which crops images. — layer_center_crop","text":"object Object compose layer . tensor, array, sequential model. height Integer, height output shape. width Integer, width output shape. data_format string, either \"channels_last\" \"channels_first\". ordering dimensions inputs. \"channels_last\" corresponds inputs shape (batch, height, width, channels) \"channels_first\" corresponds inputs shape (batch, channels, height, width). defaults image_data_format value found Keras config file ~/.keras/keras.json. never set , \"channels_last\". ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_center_crop.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"A preprocessing layer which crops images. — layer_center_crop","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_center_crop.html","id":"input-shape","dir":"Reference","previous_headings":"","what":"Input Shape","title":"A preprocessing layer which crops images. — layer_center_crop","text":"3D (unbatched) 4D (batched) tensor shape: (..., height, width, channels), \"channels_last\" format, (..., channels, height, width), \"channels_first\" format.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_center_crop.html","id":"output-shape","dir":"Reference","previous_headings":"","what":"Output Shape","title":"A preprocessing layer which crops images. — layer_center_crop","text":"3D (unbatched) 4D (batched) tensor shape: (..., target_height, target_width, channels), (..., channels, target_height, target_width), \"channels_first\" format. input height/width even target height/width odd (inversely), input image left-padded 1 pixel. Note: layer safe use inside tf.data pipeline (independently backend using).","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_concatenate.html","id":null,"dir":"Reference","previous_headings":"","what":"Concatenates a list of inputs. — layer_concatenate","title":"Concatenates a list of inputs. — layer_concatenate","text":"takes input list tensors, shape except concatenation axis, returns single tensor concatenation inputs.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_concatenate.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Concatenates a list of inputs. — layer_concatenate","text":"","code":"layer_concatenate(inputs, ..., axis = -1L)"},{"path":"https://keras3.posit.co/dev/reference/layer_concatenate.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Concatenates a list of inputs. — layer_concatenate","text":"inputs layers combine ... Standard layer keyword arguments. axis Axis along concatenate.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_concatenate.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Concatenates a list of inputs. — layer_concatenate","text":"tensor, concatenation inputs alongside axis axis.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_concatenate.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Concatenates a list of inputs. — layer_concatenate","text":"Usage Keras model:","code":"x <- op_arange(20) |> op_reshape(c(2, 2, 5)) y <- op_arange(21, 40) |> op_reshape(c(2, 2, 5)) layer_concatenate(x, y, axis = 2) ## tf.Tensor( ## [[[ 1.  2.  3.  4.  5.] ##   [ 6.  7.  8.  9. 10.] ##   [21. 22. 23. 24. 25.] ##   [26. 27. 28. 29. 30.]] ## ##  [[11. 12. 13. 14. 15.] ##   [16. 17. 18. 19. 20.] ##   [31. 32. 33. 34. 35.] ##   [36. 37. 38. 39. 40.]]], shape=(2, 4, 5), dtype=float32) x1 <- op_arange(10)     |> op_reshape(c(5, 2)) |> layer_dense(8) x2 <- op_arange(11, 20) |> op_reshape(c(5, 2)) |> layer_dense(8) y <- layer_concatenate(x1, x2)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_conv_1d.html","id":null,"dir":"Reference","previous_headings":"","what":"1D convolution layer (e.g. temporal convolution). — layer_conv_1d","title":"1D convolution layer (e.g. temporal convolution). — layer_conv_1d","text":"layer creates convolution kernel convolved layer input single spatial (temporal) dimension produce tensor outputs. use_bias TRUE, bias vector created added outputs. Finally, activation NULL, applied outputs well.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_conv_1d.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"1D convolution layer (e.g. temporal convolution). — layer_conv_1d","text":"","code":"layer_conv_1d(   object,   filters,   kernel_size,   strides = 1L,   padding = \"valid\",   data_format = NULL,   dilation_rate = 1L,   groups = 1L,   activation = NULL,   use_bias = TRUE,   kernel_initializer = \"glorot_uniform\",   bias_initializer = \"zeros\",   kernel_regularizer = NULL,   bias_regularizer = NULL,   activity_regularizer = NULL,   kernel_constraint = NULL,   bias_constraint = NULL,   ... )"},{"path":"https://keras3.posit.co/dev/reference/layer_conv_1d.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"1D convolution layer (e.g. temporal convolution). — layer_conv_1d","text":"object Object compose layer . tensor, array, sequential model. filters int, dimension output space (number filters convolution). kernel_size int list 1 integer, specifying size convolution window. strides int list 1 integer, specifying stride length convolution. strides > 1 incompatible dilation_rate > 1. padding string, \"valid\", \"\" \"causal\"(case-insensitive). \"valid\" means padding. \"\" results padding evenly left/right /input. padding=\"\" strides=1, output size input. \"causal\" results causal (dilated) convolutions, e.g. output[t] depend ontail(input, t+1). Useful modeling temporal data model violate temporal order. See WaveNet: Generative Model Raw Audio, section2.1. data_format string, either \"channels_last\" \"channels_first\". ordering dimensions inputs. \"channels_last\" corresponds inputs shape (batch, steps, features) \"channels_first\" corresponds inputs shape (batch, features, steps). defaults image_data_format value found Keras config file ~/.keras/keras.json. never set , \"channels_last\". dilation_rate int list 1 integers, specifying dilation rate use dilated convolution. groups positive int specifying number groups input split along channel axis. group convolved separately filters // groups filters. output concatenation groups results along channel axis. Input channels filters must divisible groups. activation Activation function. NULL, activation applied. use_bias bool, TRUE, bias added output. kernel_initializer Initializer convolution kernel. NULL, default initializer (\"glorot_uniform\") used. bias_initializer Initializer bias vector. NULL, default initializer (\"zeros\") used. kernel_regularizer Optional regularizer convolution kernel. bias_regularizer Optional regularizer bias vector. activity_regularizer Optional regularizer function output. kernel_constraint Optional projection function applied kernel updated Optimizer (e.g. used implement norm constraints value constraints layer weights). function must take input unprojected variable must return projected variable (must shape). Constraints safe use asynchronous distributed training. bias_constraint Optional projection function applied bias updated Optimizer. ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_conv_1d.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"1D convolution layer (e.g. temporal convolution). — layer_conv_1d","text":"3D tensor representing activation(conv1d(inputs, kernel) + bias).","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_conv_1d.html","id":"input-shape","dir":"Reference","previous_headings":"","what":"Input Shape","title":"1D convolution layer (e.g. temporal convolution). — layer_conv_1d","text":"data_format=\"channels_last\": 3D tensor shape: (batch_shape, steps, channels) data_format=\"channels_first\": 3D tensor shape: (batch_shape, channels, steps)","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_conv_1d.html","id":"output-shape","dir":"Reference","previous_headings":"","what":"Output Shape","title":"1D convolution layer (e.g. temporal convolution). — layer_conv_1d","text":"data_format=\"channels_last\": 3D tensor shape: (batch_shape, new_steps, filters) data_format=\"channels_first\": 3D tensor shape: (batch_shape, filters, new_steps)","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_conv_1d.html","id":"raises","dir":"Reference","previous_headings":"","what":"Raises","title":"1D convolution layer (e.g. temporal convolution). — layer_conv_1d","text":"ValueError: strides > 1 dilation_rate > 1.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_conv_1d.html","id":"example","dir":"Reference","previous_headings":"","what":"Example","title":"1D convolution layer (e.g. temporal convolution). — layer_conv_1d","text":"","code":"# The inputs are 128-length vectors with 10 timesteps, and the # batch size is 4. x <- random_uniform(c(4, 10, 128)) y <- x |> layer_conv_1d(32, 3, activation='relu') shape(y) ## shape(4, 8, 32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_conv_1d_transpose.html","id":null,"dir":"Reference","previous_headings":"","what":"1D transposed convolution layer. — layer_conv_1d_transpose","title":"1D transposed convolution layer. — layer_conv_1d_transpose","text":"need transposed convolutions generally arise desire use transformation going opposite direction normal convolution, .e., something shape output convolution something shape input maintaining connectivity pattern compatible said convolution.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_conv_1d_transpose.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"1D transposed convolution layer. — layer_conv_1d_transpose","text":"","code":"layer_conv_1d_transpose(   object,   filters,   kernel_size,   strides = 1L,   padding = \"valid\",   output_padding = NULL,   data_format = NULL,   dilation_rate = 1L,   activation = NULL,   use_bias = TRUE,   kernel_initializer = \"glorot_uniform\",   bias_initializer = \"zeros\",   kernel_regularizer = NULL,   bias_regularizer = NULL,   activity_regularizer = NULL,   kernel_constraint = NULL,   bias_constraint = NULL,   ... )"},{"path":"https://keras3.posit.co/dev/reference/layer_conv_1d_transpose.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"1D transposed convolution layer. — layer_conv_1d_transpose","text":"object Object compose layer . tensor, array, sequential model. filters int, dimension output space (number filters transpose convolution). kernel_size int list 1 integer, specifying size transposed convolution window. strides int list 1 integer, specifying stride length transposed convolution. strides > 1 incompatible dilation_rate > 1. padding string, either \"valid\" \"\" (case-insensitive). \"valid\" means padding. \"\" results padding evenly left/right /input output height/width dimension input. output_padding Scalar integer. Amount padding add output length. Must less stride. NULL (default) output size inferred. data_format string, either \"channels_last\" \"channels_first\". ordering dimensions inputs. \"channels_last\" corresponds inputs shape (batch, steps, features) \"channels_first\" corresponds inputs shape (batch, features, steps). defaults image_data_format value found Keras config file ~/.keras/keras.json. never set , \"channels_last\". dilation_rate Scalar integer. Specifies dilation rate. Values 1 currently require strides = 1 rates greater 1 supported. activation Activation function. NULL, activation applied. use_bias bool, TRUE, bias added output. kernel_initializer Initializer convolution kernel. NULL, default initializer (\"glorot_uniform\") used. bias_initializer Initializer bias vector. NULL, default initializer (\"zeros\") used. kernel_regularizer Optional regularizer convolution kernel. bias_regularizer Optional regularizer bias vector. activity_regularizer Optional regularizer function output. kernel_constraint Optional projection function applied kernel updated Optimizer (e.g. used implement norm constraints value constraints layer weights). function must take input unprojected variable must return projected variable (must shape). Constraints safe use asynchronous distributed training. bias_constraint Optional projection function applied bias updated Optimizer. ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_conv_1d_transpose.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"1D transposed convolution layer. — layer_conv_1d_transpose","text":"3D tensor representing activation(conv1d_transpose(inputs, kernel) + bias).","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_conv_1d_transpose.html","id":"input-shape","dir":"Reference","previous_headings":"","what":"Input Shape","title":"1D transposed convolution layer. — layer_conv_1d_transpose","text":"data_format=\"channels_last\": 3D tensor shape: (batch_shape, steps, channels) data_format=\"channels_first\": 3D tensor shape: (batch_shape, channels, steps)","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_conv_1d_transpose.html","id":"output-shape","dir":"Reference","previous_headings":"","what":"Output Shape","title":"1D transposed convolution layer. — layer_conv_1d_transpose","text":"data_format=\"channels_last\": 3D tensor shape: (batch_shape, new_steps, filters) data_format=\"channels_first\": 3D tensor shape: (batch_shape, filters, new_steps)","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_conv_1d_transpose.html","id":"raises","dir":"Reference","previous_headings":"","what":"Raises","title":"1D transposed convolution layer. — layer_conv_1d_transpose","text":"ValueError: strides > 1 dilation_rate > 1.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_conv_1d_transpose.html","id":"references","dir":"Reference","previous_headings":"","what":"References","title":"1D transposed convolution layer. — layer_conv_1d_transpose","text":"guide convolution arithmetic deep learning Deconvolutional Networks","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_conv_1d_transpose.html","id":"example","dir":"Reference","previous_headings":"","what":"Example","title":"1D transposed convolution layer. — layer_conv_1d_transpose","text":"","code":"x <- random_uniform(c(4, 10, 128)) y <- x |> layer_conv_1d_transpose(32, 3, 2, activation='relu') shape(y) ## shape(4, 21, 32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_conv_2d.html","id":null,"dir":"Reference","previous_headings":"","what":"2D convolution layer. — layer_conv_2d","title":"2D convolution layer. — layer_conv_2d","text":"layer creates convolution kernel convolved layer input 2D spatial (temporal) dimension (height width) produce tensor outputs. use_bias TRUE, bias vector created added outputs. Finally, activation NULL, applied outputs well. Note numerical precision: general Keras operation execution results identical across backends 1e-7 precision float32, Conv2D operations may show larger variations. Due large number element-wise multiplications additions convolution operations, especially large inputs kernel sizes, accumulated floating-point differences can exceed 1e-7 threshold. variations particularly noticeable using different backends (e.g., TensorFlow vs JAX) different hardware.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_conv_2d.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"2D convolution layer. — layer_conv_2d","text":"","code":"layer_conv_2d(   object,   filters,   kernel_size,   strides = list(1L, 1L),   padding = \"valid\",   data_format = NULL,   dilation_rate = list(1L, 1L),   groups = 1L,   activation = NULL,   use_bias = TRUE,   kernel_initializer = \"glorot_uniform\",   bias_initializer = \"zeros\",   kernel_regularizer = NULL,   bias_regularizer = NULL,   activity_regularizer = NULL,   kernel_constraint = NULL,   bias_constraint = NULL,   ... )"},{"path":"https://keras3.posit.co/dev/reference/layer_conv_2d.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"2D convolution layer. — layer_conv_2d","text":"object Object compose layer . tensor, array, sequential model. filters int, dimension output space (number filters convolution). kernel_size int list 2 integer, specifying size convolution window. strides int list 2 integer, specifying stride length convolution. strides > 1 incompatible dilation_rate > 1. padding string, either \"valid\" \"\" (case-insensitive). \"valid\" means padding. \"\" results padding evenly left/right /input. padding=\"\" strides=1, output size input. data_format string, either \"channels_last\" \"channels_first\". ordering dimensions inputs. \"channels_last\" corresponds inputs shape (batch_size, height, width, channels) \"channels_first\" corresponds inputs shape (batch_size, channels, height, width). defaults image_data_format value found Keras config file ~/.keras/keras.json. never set , \"channels_last\". dilation_rate int list 2 integers, specifying dilation rate use dilated convolution. groups positive int specifying number groups input split along channel axis. group convolved separately filters // groups filters. output concatenation groups results along channel axis. Input channels filters must divisible groups. activation Activation function. NULL, activation applied. use_bias bool, TRUE, bias added output. kernel_initializer Initializer convolution kernel. NULL, default initializer (\"glorot_uniform\") used. bias_initializer Initializer bias vector. NULL, default initializer (\"zeros\") used. kernel_regularizer Optional regularizer convolution kernel. bias_regularizer Optional regularizer bias vector. activity_regularizer Optional regularizer function output. kernel_constraint Optional projection function applied kernel updated Optimizer (e.g. used implement norm constraints value constraints layer weights). function must take input unprojected variable must return projected variable (must shape). Constraints safe use asynchronous distributed training. bias_constraint Optional projection function applied bias updated Optimizer. ... forward/backward compatability. output_padding Scalar integer vector two integers. Amount padding add output depth, height, width. element must smaller corresponding stride. NULL (default) output size inferred.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_conv_2d.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"2D convolution layer. — layer_conv_2d","text":"4D tensor representing activation(conv2d(inputs, kernel) + bias).","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_conv_2d.html","id":"input-shape","dir":"Reference","previous_headings":"","what":"Input Shape","title":"2D convolution layer. — layer_conv_2d","text":"data_format=\"channels_last\": 4D tensor shape: (batch_size, height, width, channels) data_format=\"channels_first\": 4D tensor shape: (batch_size, channels, height, width)","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_conv_2d.html","id":"output-shape","dir":"Reference","previous_headings":"","what":"Output Shape","title":"2D convolution layer. — layer_conv_2d","text":"data_format=\"channels_last\": 4D tensor shape: (batch_size, new_height, new_width, filters) data_format=\"channels_first\": 4D tensor shape: (batch_size, filters, new_height, new_width)","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_conv_2d.html","id":"raises","dir":"Reference","previous_headings":"","what":"Raises","title":"2D convolution layer. — layer_conv_2d","text":"ValueError: strides > 1 dilation_rate > 1.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_conv_2d.html","id":"example","dir":"Reference","previous_headings":"","what":"Example","title":"2D convolution layer. — layer_conv_2d","text":"","code":"x <- random_uniform(c(4, 10, 10, 128)) y <- x |> layer_conv_2d(32, 3, activation='relu') shape(y) ## shape(4, 8, 8, 32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_conv_2d_transpose.html","id":null,"dir":"Reference","previous_headings":"","what":"2D transposed convolution layer. — layer_conv_2d_transpose","title":"2D transposed convolution layer. — layer_conv_2d_transpose","text":"need transposed convolutions generally arise desire use transformation going opposite direction normal convolution, .e., something shape output convolution something shape input maintaining connectivity pattern compatible said convolution.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_conv_2d_transpose.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"2D transposed convolution layer. — layer_conv_2d_transpose","text":"","code":"layer_conv_2d_transpose(   object,   filters,   kernel_size,   strides = list(1L, 1L),   padding = \"valid\",   output_padding = NULL,   data_format = NULL,   dilation_rate = list(1L, 1L),   activation = NULL,   use_bias = TRUE,   kernel_initializer = \"glorot_uniform\",   bias_initializer = \"zeros\",   kernel_regularizer = NULL,   bias_regularizer = NULL,   activity_regularizer = NULL,   kernel_constraint = NULL,   bias_constraint = NULL,   ... )"},{"path":"https://keras3.posit.co/dev/reference/layer_conv_2d_transpose.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"2D transposed convolution layer. — layer_conv_2d_transpose","text":"object Object compose layer . tensor, array, sequential model. filters int, dimension output space (number filters transposed convolution). kernel_size int list 1 integer, specifying size transposed convolution window. strides int list 1 integer, specifying stride length transposed convolution. strides > 1 incompatible dilation_rate > 1. padding string, either \"valid\" \"\" (case-insensitive). \"valid\" means padding. \"\" results padding evenly left/right /input. padding=\"\" strides=1, output size input. data_format string, either \"channels_last\" \"channels_first\". ordering dimensions inputs. \"channels_last\" corresponds inputs shape (batch_size, height, width, channels) \"channels_first\" corresponds inputs shape (batch_size, channels, height, width). defaults image_data_format value found Keras config file ~/.keras/keras.json. never set , \"channels_last\". dilation_rate Scalar integer vector 2 integers specifying dilation rate. Values 1 require strides = 1; different rates per dimension supported. activation Activation function. NULL, activation applied. use_bias bool, TRUE, bias added output. kernel_initializer Initializer convolution kernel. NULL, default initializer (\"glorot_uniform\") used. bias_initializer Initializer bias vector. NULL, default initializer (\"zeros\") used. kernel_regularizer Optional regularizer convolution kernel. bias_regularizer Optional regularizer bias vector. activity_regularizer Optional regularizer function output. kernel_constraint Optional projection function applied kernel updated Optimizer (e.g. used implement norm constraints value constraints layer weights). function must take input unprojected variable must return projected variable (must shape). Constraints safe use asynchronous distributed training. bias_constraint Optional projection function applied bias updated Optimizer. ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_conv_2d_transpose.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"2D transposed convolution layer. — layer_conv_2d_transpose","text":"4D tensor representing activation(conv2d_transpose(inputs, kernel) + bias).","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_conv_2d_transpose.html","id":"input-shape","dir":"Reference","previous_headings":"","what":"Input Shape","title":"2D transposed convolution layer. — layer_conv_2d_transpose","text":"data_format=\"channels_last\": 4D tensor shape: (batch_size, height, width, channels) data_format=\"channels_first\": 4D tensor shape: (batch_size, channels, height, width)","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_conv_2d_transpose.html","id":"output-shape","dir":"Reference","previous_headings":"","what":"Output Shape","title":"2D transposed convolution layer. — layer_conv_2d_transpose","text":"data_format=\"channels_last\": 4D tensor shape: (batch_size, new_height, new_width, filters) data_format=\"channels_first\": 4D tensor shape: (batch_size, filters, new_height, new_width)","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_conv_2d_transpose.html","id":"raises","dir":"Reference","previous_headings":"","what":"Raises","title":"2D transposed convolution layer. — layer_conv_2d_transpose","text":"ValueError: strides > 1 dilation_rate > 1.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_conv_2d_transpose.html","id":"references","dir":"Reference","previous_headings":"","what":"References","title":"2D transposed convolution layer. — layer_conv_2d_transpose","text":"guide convolution arithmetic deep learning Deconvolutional Networks","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_conv_2d_transpose.html","id":"example","dir":"Reference","previous_headings":"","what":"Example","title":"2D transposed convolution layer. — layer_conv_2d_transpose","text":"","code":"x <- random_uniform(c(4, 10, 8, 128)) y <- x |> layer_conv_2d_transpose(32, 2, 2, activation='relu') shape(y) ## shape(4, 20, 16, 32) # (4, 20, 16, 32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_conv_3d.html","id":null,"dir":"Reference","previous_headings":"","what":"3D convolution layer. — layer_conv_3d","title":"3D convolution layer. — layer_conv_3d","text":"layer creates convolution kernel convolved layer input 3D spatial (temporal) dimension (width,height depth) produce tensor outputs. use_bias TRUE, bias vector created added outputs. Finally, activation NULL, applied outputs well.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_conv_3d.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"3D convolution layer. — layer_conv_3d","text":"","code":"layer_conv_3d(   object,   filters,   kernel_size,   strides = list(1L, 1L, 1L),   padding = \"valid\",   data_format = NULL,   dilation_rate = list(1L, 1L, 1L),   groups = 1L,   activation = NULL,   use_bias = TRUE,   kernel_initializer = \"glorot_uniform\",   bias_initializer = \"zeros\",   kernel_regularizer = NULL,   bias_regularizer = NULL,   activity_regularizer = NULL,   kernel_constraint = NULL,   bias_constraint = NULL,   ... )"},{"path":"https://keras3.posit.co/dev/reference/layer_conv_3d.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"3D convolution layer. — layer_conv_3d","text":"object Object compose layer . tensor, array, sequential model. filters int, dimension output space (number filters convolution). kernel_size int list 3 integer, specifying size convolution window. strides int list 3 integer, specifying stride length convolution. strides > 1 incompatible dilation_rate > 1. padding string, either \"valid\" \"\" (case-insensitive). \"valid\" means padding. \"\" results padding evenly left/right /input. padding=\"\" strides=1, output size input. data_format string, either \"channels_last\" \"channels_first\". ordering dimensions inputs. \"channels_last\" corresponds inputs shape (batch_size, spatial_dim1, spatial_dim2, spatial_dim3, channels) \"channels_first\" corresponds inputs shape (batch_size, channels, spatial_dim1, spatial_dim2, spatial_dim3). defaults image_data_format value found Keras config file ~/.keras/keras.json. never set , \"channels_last\". dilation_rate int vector 3 ints, specifying dilation rate use dilated convolution. groups positive int specifying number groups input split along channel axis. group convolved separately filters %/% groups filters. output concatenation groups results along channel axis. Input channels filters must divisible groups. activation Activation function. NULL, activation applied. use_bias bool, TRUE, bias added output. kernel_initializer Initializer convolution kernel. NULL, default initializer (\"glorot_uniform\") used. bias_initializer Initializer bias vector. NULL, default initializer (\"zeros\") used. kernel_regularizer Optional regularizer convolution kernel. bias_regularizer Optional regularizer bias vector. activity_regularizer Optional regularizer function output. kernel_constraint Optional projection function applied kernel updated Optimizer (e.g. used implement norm constraints value constraints layer weights). function must take input unprojected variable must return projected variable (must shape). Constraints safe use asynchronous distributed training. bias_constraint Optional projection function applied bias updated Optimizer. ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_conv_3d.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"3D convolution layer. — layer_conv_3d","text":"5D tensor representing activation(conv3d(inputs, kernel) + bias).","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_conv_3d.html","id":"input-shape","dir":"Reference","previous_headings":"","what":"Input Shape","title":"3D convolution layer. — layer_conv_3d","text":"data_format=\"channels_last\": 5D tensor shape: (batch_size, spatial_dim1, spatial_dim2, spatial_dim3, channels) data_format=\"channels_first\": 5D tensor shape: (batch_size, channels, spatial_dim1, spatial_dim2, spatial_dim3)","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_conv_3d.html","id":"output-shape","dir":"Reference","previous_headings":"","what":"Output Shape","title":"3D convolution layer. — layer_conv_3d","text":"data_format=\"channels_last\": 5D tensor shape: (batch_size, new_spatial_dim1, new_spatial_dim2, new_spatial_dim3, filters) data_format=\"channels_first\": 5D tensor shape: (batch_size, filters, new_spatial_dim1, new_spatial_dim2, new_spatial_dim3)","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_conv_3d.html","id":"raises","dir":"Reference","previous_headings":"","what":"Raises","title":"3D convolution layer. — layer_conv_3d","text":"ValueError: strides > 1 dilation_rate > 1.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_conv_3d.html","id":"example","dir":"Reference","previous_headings":"","what":"Example","title":"3D convolution layer. — layer_conv_3d","text":"","code":"x <- random_uniform(c(4, 10, 10, 10, 128)) y <- x |> layer_conv_3d(32, 3, activation = 'relu') shape(y) ## shape(4, 8, 8, 8, 32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_conv_3d_transpose.html","id":null,"dir":"Reference","previous_headings":"","what":"3D transposed convolution layer. — layer_conv_3d_transpose","title":"3D transposed convolution layer. — layer_conv_3d_transpose","text":"need transposed convolutions generally arise desire use transformation going opposite direction normal convolution, .e., something shape output convolution something shape input maintaining connectivity pattern compatible said convolution.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_conv_3d_transpose.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"3D transposed convolution layer. — layer_conv_3d_transpose","text":"","code":"layer_conv_3d_transpose(   object,   filters,   kernel_size,   strides = list(1L, 1L, 1L),   padding = \"valid\",   output_padding = NULL,   data_format = NULL,   dilation_rate = list(1L, 1L, 1L),   activation = NULL,   use_bias = TRUE,   kernel_initializer = \"glorot_uniform\",   bias_initializer = \"zeros\",   kernel_regularizer = NULL,   bias_regularizer = NULL,   activity_regularizer = NULL,   kernel_constraint = NULL,   bias_constraint = NULL,   ... )"},{"path":"https://keras3.posit.co/dev/reference/layer_conv_3d_transpose.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"3D transposed convolution layer. — layer_conv_3d_transpose","text":"object Object compose layer . tensor, array, sequential model. filters int, dimension output space (number filters transposed convolution). kernel_size int list 1 integer, specifying size transposed convolution window. strides int list 1 integer, specifying stride length transposed convolution. strides > 1 incompatible dilation_rate > 1. padding string, either \"valid\" \"\" (case-insensitive). \"valid\" means padding. \"\" results padding evenly left/right /input. padding=\"\" strides=1, output size input. data_format string, either \"channels_last\" \"channels_first\". ordering dimensions inputs. \"channels_last\" corresponds inputs shape (batch_size, spatial_dim1, spatial_dim2, spatial_dim3, channels) \"channels_first\" corresponds inputs shape (batch_size, channels, spatial_dim1, spatial_dim2, spatial_dim3). defaults image_data_format value found Keras config file ~/.keras/keras.json. never set , \"channels_last\". dilation_rate Scalar integer vector 3 integers specifying dilation rate. Values 1 require strides = 1; different rates per dimension supported. activation Activation function. NULL, activation applied. use_bias bool, TRUE, bias added output. kernel_initializer Initializer convolution kernel. NULL, default initializer (\"glorot_uniform\") used. bias_initializer Initializer bias vector. NULL, default initializer (\"zeros\") used. kernel_regularizer Optional regularizer convolution kernel. bias_regularizer Optional regularizer bias vector. activity_regularizer Optional regularizer function output. kernel_constraint Optional projection function applied kernel updated Optimizer (e.g. used implement norm constraints value constraints layer weights). function must take input unprojected variable must return projected variable (must shape). Constraints safe use asynchronous distributed training. bias_constraint Optional projection function applied bias updated Optimizer. ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_conv_3d_transpose.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"3D transposed convolution layer. — layer_conv_3d_transpose","text":"5D tensor representing activation(conv3d(inputs, kernel) + bias).","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_conv_3d_transpose.html","id":"input-shape","dir":"Reference","previous_headings":"","what":"Input Shape","title":"3D transposed convolution layer. — layer_conv_3d_transpose","text":"data_format=\"channels_last\": 5D tensor shape: (batch_size, spatial_dim1, spatial_dim2, spatial_dim3, channels) data_format=\"channels_first\": 5D tensor shape: (batch_size, channels, spatial_dim1, spatial_dim2, spatial_dim3)","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_conv_3d_transpose.html","id":"output-shape","dir":"Reference","previous_headings":"","what":"Output Shape","title":"3D transposed convolution layer. — layer_conv_3d_transpose","text":"data_format=\"channels_last\": 5D tensor shape: (batch_size, new_spatial_dim1, new_spatial_dim2, new_spatial_dim3, filters) data_format=\"channels_first\": 5D tensor shape: (batch_size, filters, new_spatial_dim1, new_spatial_dim2, new_spatial_dim3)","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_conv_3d_transpose.html","id":"raises","dir":"Reference","previous_headings":"","what":"Raises","title":"3D transposed convolution layer. — layer_conv_3d_transpose","text":"ValueError: strides > 1 dilation_rate > 1.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_conv_3d_transpose.html","id":"references","dir":"Reference","previous_headings":"","what":"References","title":"3D transposed convolution layer. — layer_conv_3d_transpose","text":"guide convolution arithmetic deep learning Deconvolutional Networks","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_conv_3d_transpose.html","id":"example","dir":"Reference","previous_headings":"","what":"Example","title":"3D transposed convolution layer. — layer_conv_3d_transpose","text":"","code":"x <- random_uniform(c(4, 10, 8, 12, 128)) y <- x |> layer_conv_3d_transpose(32, 2, 2, activation = 'relu') shape(y) ## shape(4, 20, 16, 24, 32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_conv_lstm_1d.html","id":null,"dir":"Reference","previous_headings":"","what":"1D Convolutional LSTM. — layer_conv_lstm_1d","title":"1D Convolutional LSTM. — layer_conv_lstm_1d","text":"Similar LSTM layer, input transformations recurrent transformations convolutional.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_conv_lstm_1d.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"1D Convolutional LSTM. — layer_conv_lstm_1d","text":"","code":"layer_conv_lstm_1d(   object,   filters,   kernel_size,   strides = 1L,   padding = \"valid\",   data_format = NULL,   dilation_rate = 1L,   activation = \"tanh\",   recurrent_activation = \"sigmoid\",   use_bias = TRUE,   kernel_initializer = \"glorot_uniform\",   recurrent_initializer = \"orthogonal\",   bias_initializer = \"zeros\",   unit_forget_bias = TRUE,   kernel_regularizer = NULL,   recurrent_regularizer = NULL,   bias_regularizer = NULL,   activity_regularizer = NULL,   kernel_constraint = NULL,   recurrent_constraint = NULL,   bias_constraint = NULL,   dropout = 0,   recurrent_dropout = 0,   seed = NULL,   return_sequences = FALSE,   return_state = FALSE,   go_backwards = FALSE,   stateful = FALSE,   ...,   unroll = NULL )"},{"path":"https://keras3.posit.co/dev/reference/layer_conv_lstm_1d.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"1D Convolutional LSTM. — layer_conv_lstm_1d","text":"object Object compose layer . tensor, array, sequential model. filters int, dimension output space (number filters convolution). kernel_size int tuple/list 1 integer, specifying size convolution window. strides int tuple/list 1 integer, specifying stride length convolution. strides > 1 incompatible dilation_rate > 1. padding string, \"valid\" \"\" (case-insensitive). \"valid\" means padding. \"\" results padding evenly left/right /input output height/width dimension input. data_format string, either \"channels_last\" \"channels_first\". ordering dimensions inputs. \"channels_last\" corresponds inputs shape (batch, steps, features) \"channels_first\" corresponds inputs shape (batch, features, steps). defaults image_data_format value found Keras config file ~/.keras/keras.json. never set , \"channels_last\". dilation_rate int tuple/list 1 integers, specifying dilation rate use dilated convolution. activation Activation function use. default hyperbolic tangent activation function applied (tanh(x)). recurrent_activation Activation function use recurrent step. use_bias Boolean, whether layer uses bias vector. kernel_initializer Initializer kernel weights matrix, used linear transformation inputs. recurrent_initializer Initializer recurrent_kernel weights matrix, used linear transformation recurrent state. bias_initializer Initializer bias vector. unit_forget_bias Boolean. TRUE, add 1 bias forget gate initialization. Use combination bias_initializer=\"zeros\". recommended Jozefowicz et al., 2015 kernel_regularizer Regularizer function applied kernel weights matrix. recurrent_regularizer Regularizer function applied recurrent_kernel weights matrix. bias_regularizer Regularizer function applied bias vector. activity_regularizer Regularizer function applied . kernel_constraint Constraint function applied kernel weights matrix. recurrent_constraint Constraint function applied recurrent_kernel weights matrix. bias_constraint Constraint function applied bias vector. dropout Float 0 1. Fraction units drop linear transformation inputs. recurrent_dropout Float 0 1. Fraction units drop linear transformation recurrent state. seed Random seed dropout. return_sequences Boolean. Whether return last output output sequence, full sequence. Default: FALSE. return_state Boolean. Whether return last state addition output. Default: FALSE. go_backwards Boolean (default: FALSE). TRUE, process input sequence backwards return reversed sequence. stateful Boolean (default FALSE). TRUE, last state sample index batch used initial state sample index following batch. ... forward/backward compatability. unroll Boolean (default: FALSE). TRUE, network unrolled, else symbolic loop used. Unrolling can speed-RNN, although tends memory-intensive. Unrolling suitable short sequences.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_conv_lstm_1d.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"1D Convolutional LSTM. — layer_conv_lstm_1d","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_conv_lstm_1d.html","id":"call-arguments","dir":"Reference","previous_headings":"","what":"Call Arguments","title":"1D Convolutional LSTM. — layer_conv_lstm_1d","text":"inputs: 4D tensor. initial_state: List initial state tensors passed first call cell. mask: Binary tensor shape (samples, timesteps) indicating whether given timestep masked. training: Python boolean indicating whether layer behave training mode inference mode. relevant dropout recurrent_dropout set.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_conv_lstm_1d.html","id":"input-shape","dir":"Reference","previous_headings":"","what":"Input Shape","title":"1D Convolutional LSTM. — layer_conv_lstm_1d","text":"data_format=\"channels_first\": 4D tensor shape: (samples, time, channels, rows) data_format=\"channels_last\": 4D tensor shape: (samples, time, rows, channels)","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_conv_lstm_1d.html","id":"output-shape","dir":"Reference","previous_headings":"","what":"Output Shape","title":"1D Convolutional LSTM. — layer_conv_lstm_1d","text":"return_state: list tensors. first tensor output. remaining tensors last states, 3D tensor shape: (samples, filters, new_rows) data_format='channels_first' shape: (samples, new_rows, filters) data_format='channels_last'. rows values might changed due padding. return_sequences: 4D tensor shape: (samples, timesteps, filters, new_rows) data_format='channels_first' shape: (samples, timesteps, new_rows, filters) data_format='channels_last'. Else, 3D tensor shape: (samples, filters, new_rows) data_format='channels_first' shape: (samples, new_rows, filters) data_format='channels_last'.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_conv_lstm_1d.html","id":"references","dir":"Reference","previous_headings":"","what":"References","title":"1D Convolutional LSTM. — layer_conv_lstm_1d","text":"Shi et al., 2015 (current implementation include feedback loop cells output).","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_conv_lstm_2d.html","id":null,"dir":"Reference","previous_headings":"","what":"2D Convolutional LSTM. — layer_conv_lstm_2d","title":"2D Convolutional LSTM. — layer_conv_lstm_2d","text":"Similar LSTM layer, input transformations recurrent transformations convolutional.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_conv_lstm_2d.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"2D Convolutional LSTM. — layer_conv_lstm_2d","text":"","code":"layer_conv_lstm_2d(   object,   filters,   kernel_size,   strides = 1L,   padding = \"valid\",   data_format = NULL,   dilation_rate = 1L,   activation = \"tanh\",   recurrent_activation = \"sigmoid\",   use_bias = TRUE,   kernel_initializer = \"glorot_uniform\",   recurrent_initializer = \"orthogonal\",   bias_initializer = \"zeros\",   unit_forget_bias = TRUE,   kernel_regularizer = NULL,   recurrent_regularizer = NULL,   bias_regularizer = NULL,   activity_regularizer = NULL,   kernel_constraint = NULL,   recurrent_constraint = NULL,   bias_constraint = NULL,   dropout = 0,   recurrent_dropout = 0,   seed = NULL,   return_sequences = FALSE,   return_state = FALSE,   go_backwards = FALSE,   stateful = FALSE,   ...,   unroll = NULL )"},{"path":"https://keras3.posit.co/dev/reference/layer_conv_lstm_2d.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"2D Convolutional LSTM. — layer_conv_lstm_2d","text":"object Object compose layer . tensor, array, sequential model. filters int, dimension output space (number filters convolution). kernel_size int tuple/list 2 integers, specifying size convolution window. strides int tuple/list 2 integers, specifying stride length convolution. strides > 1 incompatible dilation_rate > 1. padding string, \"valid\" \"\" (case-insensitive). \"valid\" means padding. \"\" results padding evenly left/right /input output height/width dimension input. data_format string, either \"channels_last\" \"channels_first\". ordering dimensions inputs. \"channels_last\" corresponds inputs shape (batch, steps, features) \"channels_first\" corresponds inputs shape (batch, features, steps). defaults image_data_format value found Keras config file ~/.keras/keras.json. never set , \"channels_last\". dilation_rate int tuple/list 2 integers, specifying dilation rate use dilated convolution. activation Activation function use. default hyperbolic tangent activation function applied (tanh(x)). recurrent_activation Activation function use recurrent step. use_bias Boolean, whether layer uses bias vector. kernel_initializer Initializer kernel weights matrix, used linear transformation inputs. recurrent_initializer Initializer recurrent_kernel weights matrix, used linear transformation recurrent state. bias_initializer Initializer bias vector. unit_forget_bias Boolean. TRUE, add 1 bias forget gate initialization. Use combination bias_initializer=\"zeros\". recommended Jozefowicz et al., 2015 kernel_regularizer Regularizer function applied kernel weights matrix. recurrent_regularizer Regularizer function applied recurrent_kernel weights matrix. bias_regularizer Regularizer function applied bias vector. activity_regularizer Regularizer function applied . kernel_constraint Constraint function applied kernel weights matrix. recurrent_constraint Constraint function applied recurrent_kernel weights matrix. bias_constraint Constraint function applied bias vector. dropout Float 0 1. Fraction units drop linear transformation inputs. recurrent_dropout Float 0 1. Fraction units drop linear transformation recurrent state. seed Random seed dropout. return_sequences Boolean. Whether return last output output sequence, full sequence. Default: FALSE. return_state Boolean. Whether return last state addition output. Default: FALSE. go_backwards Boolean (default: FALSE). TRUE, process input sequence backwards return reversed sequence. stateful Boolean (default FALSE). TRUE, last state sample index batch used initial state sample index following batch. ... forward/backward compatability. unroll Boolean (default: FALSE). TRUE, network unrolled, else symbolic loop used. Unrolling can speed-RNN, although tends memory-intensive. Unrolling suitable short sequences.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_conv_lstm_2d.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"2D Convolutional LSTM. — layer_conv_lstm_2d","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_conv_lstm_2d.html","id":"call-arguments","dir":"Reference","previous_headings":"","what":"Call Arguments","title":"2D Convolutional LSTM. — layer_conv_lstm_2d","text":"inputs: 5D tensor. mask: Binary tensor shape (samples, timesteps) indicating whether given timestep masked. training: Python boolean indicating whether layer behave training mode inference mode. relevant dropout recurrent_dropout set. initial_state: List initial state tensors passed first call cell.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_conv_lstm_2d.html","id":"input-shape","dir":"Reference","previous_headings":"","what":"Input Shape","title":"2D Convolutional LSTM. — layer_conv_lstm_2d","text":"data_format='channels_first': 5D tensor shape: (samples, time, channels, rows, cols) data_format='channels_last': 5D tensor shape: (samples, time, rows, cols, channels)","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_conv_lstm_2d.html","id":"output-shape","dir":"Reference","previous_headings":"","what":"Output Shape","title":"2D Convolutional LSTM. — layer_conv_lstm_2d","text":"return_state: list tensors. first tensor output. remaining tensors last states, 4D tensor shape: (samples, filters, new_rows, new_cols) data_format='channels_first' shape: (samples, new_rows, new_cols, filters) data_format='channels_last'. rows cols values might changed due padding. return_sequences: 5D tensor shape: (samples, timesteps, filters, new_rows, new_cols) data_format='channels_first' shape: (samples, timesteps, new_rows, new_cols, filters) data_format='channels_last'. Else, 4D tensor shape: (samples, filters, new_rows, new_cols) data_format='channels_first' shape: (samples, new_rows, new_cols, filters) data_format='channels_last'.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_conv_lstm_2d.html","id":"references","dir":"Reference","previous_headings":"","what":"References","title":"2D Convolutional LSTM. — layer_conv_lstm_2d","text":"Shi et al., 2015 (current implementation include feedback loop cells output).","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_conv_lstm_3d.html","id":null,"dir":"Reference","previous_headings":"","what":"3D Convolutional LSTM. — layer_conv_lstm_3d","title":"3D Convolutional LSTM. — layer_conv_lstm_3d","text":"Similar LSTM layer, input transformations recurrent transformations convolutional.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_conv_lstm_3d.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"3D Convolutional LSTM. — layer_conv_lstm_3d","text":"","code":"layer_conv_lstm_3d(   object,   filters,   kernel_size,   strides = 1L,   padding = \"valid\",   data_format = NULL,   dilation_rate = 1L,   activation = \"tanh\",   recurrent_activation = \"sigmoid\",   use_bias = TRUE,   kernel_initializer = \"glorot_uniform\",   recurrent_initializer = \"orthogonal\",   bias_initializer = \"zeros\",   unit_forget_bias = TRUE,   kernel_regularizer = NULL,   recurrent_regularizer = NULL,   bias_regularizer = NULL,   activity_regularizer = NULL,   kernel_constraint = NULL,   recurrent_constraint = NULL,   bias_constraint = NULL,   dropout = 0,   recurrent_dropout = 0,   seed = NULL,   return_sequences = FALSE,   return_state = FALSE,   go_backwards = FALSE,   stateful = FALSE,   ...,   unroll = NULL )"},{"path":"https://keras3.posit.co/dev/reference/layer_conv_lstm_3d.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"3D Convolutional LSTM. — layer_conv_lstm_3d","text":"object Object compose layer . tensor, array, sequential model. filters int, dimension output space (number filters convolution). kernel_size int tuple/list 3 integers, specifying size convolution window. strides int tuple/list 3 integers, specifying stride length convolution. strides > 1 incompatible dilation_rate > 1. padding string, \"valid\" \"\" (case-insensitive). \"valid\" means padding. \"\" results padding evenly left/right /input output height/width dimension input. data_format string, either \"channels_last\" \"channels_first\". ordering dimensions inputs. \"channels_last\" corresponds inputs shape (batch, steps, features) \"channels_first\" corresponds inputs shape (batch, features, steps). defaults image_data_format value found Keras config file ~/.keras/keras.json. never set , \"channels_last\". dilation_rate int tuple/list 3 integers, specifying dilation rate use dilated convolution. activation Activation function use. default hyperbolic tangent activation function applied (tanh(x)). recurrent_activation Activation function use recurrent step. use_bias Boolean, whether layer uses bias vector. kernel_initializer Initializer kernel weights matrix, used linear transformation inputs. recurrent_initializer Initializer recurrent_kernel weights matrix, used linear transformation recurrent state. bias_initializer Initializer bias vector. unit_forget_bias Boolean. TRUE, add 1 bias forget gate initialization. Use combination bias_initializer=\"zeros\". recommended Jozefowicz et al., 2015 kernel_regularizer Regularizer function applied kernel weights matrix. recurrent_regularizer Regularizer function applied recurrent_kernel weights matrix. bias_regularizer Regularizer function applied bias vector. activity_regularizer Regularizer function applied . kernel_constraint Constraint function applied kernel weights matrix. recurrent_constraint Constraint function applied recurrent_kernel weights matrix. bias_constraint Constraint function applied bias vector. dropout Float 0 1. Fraction units drop linear transformation inputs. recurrent_dropout Float 0 1. Fraction units drop linear transformation recurrent state. seed Random seed dropout. return_sequences Boolean. Whether return last output output sequence, full sequence. Default: FALSE. return_state Boolean. Whether return last state addition output. Default: FALSE. go_backwards Boolean (default: FALSE). TRUE, process input sequence backwards return reversed sequence. stateful Boolean (default FALSE). TRUE, last state sample index batch used initial state sample index following batch. ... forward/backward compatability. unroll Boolean (default: FALSE). TRUE, network unrolled, else symbolic loop used. Unrolling can speed-RNN, although tends memory-intensive. Unrolling suitable short sequences.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_conv_lstm_3d.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"3D Convolutional LSTM. — layer_conv_lstm_3d","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_conv_lstm_3d.html","id":"call-arguments","dir":"Reference","previous_headings":"","what":"Call Arguments","title":"3D Convolutional LSTM. — layer_conv_lstm_3d","text":"inputs: 6D tensor. mask: Binary tensor shape (samples, timesteps) indicating whether given timestep masked. training: Python boolean indicating whether layer behave training mode inference mode. relevant dropout recurrent_dropout set. initial_state: List initial state tensors passed first call cell.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_conv_lstm_3d.html","id":"input-shape","dir":"Reference","previous_headings":"","what":"Input Shape","title":"3D Convolutional LSTM. — layer_conv_lstm_3d","text":"data_format='channels_first': 5D tensor shape: (samples, time, channels, *spatial_dims) data_format='channels_last': 5D tensor shape: (samples, time, *spatial_dims, channels)","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_conv_lstm_3d.html","id":"output-shape","dir":"Reference","previous_headings":"","what":"Output Shape","title":"3D Convolutional LSTM. — layer_conv_lstm_3d","text":"return_state: list tensors. first tensor output. remaining tensors last states, 4D tensor shape: (samples, filters, *spatial_dims) data_format='channels_first' shape: (samples, *spatial_dims, filters) data_format='channels_last'. return_sequences: 5D tensor shape: (samples, timesteps, filters, *spatial_dims) data_format='channels_first' shape: (samples, timesteps, *spatial_dims, filters) data_format='channels_last'. Else, 4D tensor shape: (samples, filters, *spatial_dims) data_format='channels_first' shape: (samples, *spatial_dims, filters) data_format='channels_last'.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_conv_lstm_3d.html","id":"references","dir":"Reference","previous_headings":"","what":"References","title":"3D Convolutional LSTM. — layer_conv_lstm_3d","text":"Shi et al., 2015 (current implementation include feedback loop cells output).","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_cropping_1d.html","id":null,"dir":"Reference","previous_headings":"","what":"Cropping layer for 1D input (e.g. temporal sequence). — layer_cropping_1d","title":"Cropping layer for 1D input (e.g. temporal sequence). — layer_cropping_1d","text":"crops along time dimension (axis 2).","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_cropping_1d.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Cropping layer for 1D input (e.g. temporal sequence). — layer_cropping_1d","text":"","code":"layer_cropping_1d(object, cropping = list(1L, 1L), ...)"},{"path":"https://keras3.posit.co/dev/reference/layer_cropping_1d.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Cropping layer for 1D input (e.g. temporal sequence). — layer_cropping_1d","text":"object Object compose layer . tensor, array, sequential model. cropping Int, list int (length 2). int: many units trimmed beginning end cropping dimension (axis 1). list 2 ints: many units trimmed beginning end cropping dimension ((left_crop, right_crop)). ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_cropping_1d.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Cropping layer for 1D input (e.g. temporal sequence). — layer_cropping_1d","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_cropping_1d.html","id":"example","dir":"Reference","previous_headings":"","what":"Example","title":"Cropping layer for 1D input (e.g. temporal sequence). — layer_cropping_1d","text":"","code":"input_shape <- c(2, 3, 2) x <- op_arange(prod(input_shape)) |> op_reshape(input_shape) x ## tf.Tensor( ## [[[ 1.  2.] ##   [ 3.  4.] ##   [ 5.  6.]] ## ##  [[ 7.  8.] ##   [ 9. 10.] ##   [11. 12.]]], shape=(2, 3, 2), dtype=float32) y <- x |> layer_cropping_1d(cropping = 1) y ## tf.Tensor( ## [[[ 3.  4.]] ## ##  [[ 9. 10.]]], shape=(2, 1, 2), dtype=float32)"},{"path":"https://keras3.posit.co/dev/reference/layer_cropping_1d.html","id":"input-shape","dir":"Reference","previous_headings":"","what":"Input Shape","title":"Cropping layer for 1D input (e.g. temporal sequence). — layer_cropping_1d","text":"3D tensor shape (batch_size, axis_to_crop, features)","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_cropping_1d.html","id":"output-shape","dir":"Reference","previous_headings":"","what":"Output Shape","title":"Cropping layer for 1D input (e.g. temporal sequence). — layer_cropping_1d","text":"3D tensor shape (batch_size, cropped_axis, features)","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_cropping_2d.html","id":null,"dir":"Reference","previous_headings":"","what":"Cropping layer for 2D input (e.g. picture). — layer_cropping_2d","title":"Cropping layer for 2D input (e.g. picture). — layer_cropping_2d","text":"crops along spatial dimensions, .e. height width.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_cropping_2d.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Cropping layer for 2D input (e.g. picture). — layer_cropping_2d","text":"","code":"layer_cropping_2d(   object,   cropping = list(list(0L, 0L), list(0L, 0L)),   data_format = NULL,   ... )"},{"path":"https://keras3.posit.co/dev/reference/layer_cropping_2d.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Cropping layer for 2D input (e.g. picture). — layer_cropping_2d","text":"object Object compose layer . tensor, array, sequential model. cropping Int, list 2 ints, list 2 lists 2 ints. int: symmetric cropping applied height width. list 2 ints: interpreted two different symmetric cropping values height width: (symmetric_height_crop, symmetric_width_crop). list 2 lists 2 ints: interpreted ((top_crop, bottom_crop), (left_crop, right_crop)). data_format string, one \"channels_last\" (default) \"channels_first\". ordering dimensions inputs. \"channels_last\" corresponds inputs shape (batch_size, height, width, channels) \"channels_first\" corresponds inputs shape (batch_size, channels, height, width). unspecified, uses image_data_format value found Keras config file ~/.keras/keras.json (exists). Defaults \"channels_last\". ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_cropping_2d.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Cropping layer for 2D input (e.g. picture). — layer_cropping_2d","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_cropping_2d.html","id":"example","dir":"Reference","previous_headings":"","what":"Example","title":"Cropping layer for 2D input (e.g. picture). — layer_cropping_2d","text":"","code":"input_shape <- c(2, 28, 28, 3) x <- op_arange(prod(input_shape), dtype ='int32') |> op_reshape(input_shape) y <- x |> layer_cropping_2d(cropping=list(c(2, 2), c(4, 4))) shape(y) ## shape(2, 24, 20, 3)"},{"path":"https://keras3.posit.co/dev/reference/layer_cropping_2d.html","id":"input-shape","dir":"Reference","previous_headings":"","what":"Input Shape","title":"Cropping layer for 2D input (e.g. picture). — layer_cropping_2d","text":"4D tensor shape: data_format \"channels_last\": (batch_size, height, width, channels) data_format \"channels_first\": (batch_size, channels, height, width)","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_cropping_2d.html","id":"output-shape","dir":"Reference","previous_headings":"","what":"Output Shape","title":"Cropping layer for 2D input (e.g. picture). — layer_cropping_2d","text":"4D tensor shape: data_format \"channels_last\": (batch_size, cropped_height, cropped_width, channels) data_format \"channels_first\": (batch_size, channels, cropped_height, cropped_width)","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_cropping_3d.html","id":null,"dir":"Reference","previous_headings":"","what":"Cropping layer for 3D data (e.g. spatial or spatio-temporal). — layer_cropping_3d","title":"Cropping layer for 3D data (e.g. spatial or spatio-temporal). — layer_cropping_3d","text":"Cropping layer 3D data (e.g. spatial spatio-temporal).","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_cropping_3d.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Cropping layer for 3D data (e.g. spatial or spatio-temporal). — layer_cropping_3d","text":"","code":"layer_cropping_3d(   object,   cropping = list(list(1L, 1L), list(1L, 1L), list(1L, 1L)),   data_format = NULL,   ... )"},{"path":"https://keras3.posit.co/dev/reference/layer_cropping_3d.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Cropping layer for 3D data (e.g. spatial or spatio-temporal). — layer_cropping_3d","text":"object Object compose layer . tensor, array, sequential model. cropping Int, list 3 ints, list 3 lists 2 ints. int: symmetric cropping applied depth, height, width. list 3 ints: interpreted three different symmetric cropping values depth, height, width: (symmetric_dim1_crop, symmetric_dim2_crop, symmetric_dim3_crop). list 3 lists 2 ints: interpreted ((left_dim1_crop, right_dim1_crop), (left_dim2_crop, right_dim2_crop), (left_dim3_crop, right_dim3_crop)). data_format string, one \"channels_last\" (default) \"channels_first\". ordering dimensions inputs. \"channels_last\" corresponds inputs shape (batch_size, spatial_dim1, spatial_dim2, spatial_dim3, channels) \"channels_first\" corresponds inputs shape (batch_size, channels, spatial_dim1, spatial_dim2, spatial_dim3). unspecified, uses image_data_format value found Keras config file ~/.keras/keras.json (exists). Defaults \"channels_last\". ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_cropping_3d.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Cropping layer for 3D data (e.g. spatial or spatio-temporal). — layer_cropping_3d","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_cropping_3d.html","id":"example","dir":"Reference","previous_headings":"","what":"Example","title":"Cropping layer for 3D data (e.g. spatial or spatio-temporal). — layer_cropping_3d","text":"","code":"input_shape <- c(2, 28, 28, 10, 3) x <- input_shape %>% { op_reshape(seq(prod(.)), .) } y <- x |> layer_cropping_3d(cropping = c(2, 4, 2)) shape(y) ## shape(2, 24, 20, 6, 3)"},{"path":"https://keras3.posit.co/dev/reference/layer_cropping_3d.html","id":"input-shape","dir":"Reference","previous_headings":"","what":"Input Shape","title":"Cropping layer for 3D data (e.g. spatial or spatio-temporal). — layer_cropping_3d","text":"5D tensor shape: data_format \"channels_last\": (batch_size, first_axis_to_crop, second_axis_to_crop, third_axis_to_crop, channels) data_format \"channels_first\": (batch_size, channels, first_axis_to_crop, second_axis_to_crop, third_axis_to_crop)","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_cropping_3d.html","id":"output-shape","dir":"Reference","previous_headings":"","what":"Output Shape","title":"Cropping layer for 3D data (e.g. spatial or spatio-temporal). — layer_cropping_3d","text":"5D tensor shape: data_format \"channels_last\": (batch_size, first_cropped_axis, second_cropped_axis, third_cropped_axis, channels) data_format \"channels_first\": (batch_size, channels, first_cropped_axis, second_cropped_axis, third_cropped_axis)","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_cut_mix.html","id":null,"dir":"Reference","previous_headings":"","what":"CutMix data augmentation technique. — layer_cut_mix","title":"CutMix data augmentation technique. — layer_cut_mix","text":"CutMix data augmentation method patches cut pasted two images dataset, labels also mixed proportionally area patches.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_cut_mix.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"CutMix data augmentation technique. — layer_cut_mix","text":"","code":"layer_cut_mix(object, factor = 1, seed = NULL, data_format = NULL, ...)"},{"path":"https://keras3.posit.co/dev/reference/layer_cut_mix.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"CutMix data augmentation technique. — layer_cut_mix","text":"object Object compose layer . tensor, array, sequential model. factor single float tuple two floats 0 1. tuple numbers passed, factor sampled two values. single float passed, value 0 passed float sampled. values define range mixing weight sampled. higher factor increases variability patch sizes, leading diverse larger mixed patches. Defaults 1. seed Integer. Used create random seed. data_format string, either \"channels_last\" \"channels_first\". ordering dimensions inputs. \"channels_last\" corresponds inputs shape (batch, height, width, channels) \"channels_first\" corresponds inputs shape (batch, channels, height, width). defaults image_data_format value found Keras config file ~/.keras/keras.json. never set , \"channels_last\". ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_cut_mix.html","id":"references","dir":"Reference","previous_headings":"","what":"References","title":"CutMix data augmentation technique. — layer_cut_mix","text":"CutMix paper.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_dense.html","id":null,"dir":"Reference","previous_headings":"","what":"Just your regular densely-connected NN layer. — layer_dense","title":"Just your regular densely-connected NN layer. — layer_dense","text":"Dense implements operation: output = activation(dot(input, kernel) + bias) activation element-wise activation function passed activation argument, kernel weights matrix created layer, bias bias vector created layer (applicable use_bias TRUE).","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_dense.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Just your regular densely-connected NN layer. — layer_dense","text":"","code":"layer_dense(   object,   units,   activation = NULL,   use_bias = TRUE,   kernel_initializer = \"glorot_uniform\",   bias_initializer = \"zeros\",   kernel_regularizer = NULL,   bias_regularizer = NULL,   activity_regularizer = NULL,   kernel_constraint = NULL,   bias_constraint = NULL,   lora_rank = NULL,   lora_alpha = NULL,   ... )"},{"path":"https://keras3.posit.co/dev/reference/layer_dense.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Just your regular densely-connected NN layer. — layer_dense","text":"object Object compose layer . tensor, array, sequential model. units Positive integer, dimensionality output space. activation Activation function use. specify anything, activation applied (ie. \"linear\" activation: (x) = x). use_bias Boolean, whether layer uses bias vector. kernel_initializer Initializer kernel weights matrix. bias_initializer Initializer bias vector. kernel_regularizer Regularizer function applied kernel weights matrix. bias_regularizer Regularizer function applied bias vector. activity_regularizer Regularizer function applied output layer (\"activation\"). kernel_constraint Constraint function applied kernel weights matrix. bias_constraint Constraint function applied bias vector. lora_rank Optional integer. set, layer's forward pass implement LoRA (Low-Rank Adaptation) provided rank. LoRA sets layer's kernel non-trainable replaces delta original kernel, obtained via multiplying two lower-rank trainable matrices. can useful reduce computation cost fine-tuning large dense layers. can also enable LoRA existing Dense layer calling layer$enable_lora(rank). lora_alpha Optional integer. Scales low-rank adaptation delta forward pass. delta scaled lora_alpha / lora_rank, letting tune LoRA adjustment strength independently lora_rank. ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_dense.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Just your regular densely-connected NN layer. — layer_dense","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_dense.html","id":"note","dir":"Reference","previous_headings":"","what":"Note","title":"Just your regular densely-connected NN layer. — layer_dense","text":"input layer rank greater 2, Dense computes dot product inputs kernel along last axis inputs axis 0 kernel (using tf.tensordot). example, input dimensions (batch_size, d0, d1), create kernel shape (d1, units), kernel operates along axis 2 input, every sub-tensor shape (1, 1, d1) (batch_size * d0 sub-tensors). output case shape (batch_size, d0, units).","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_dense.html","id":"input-shape","dir":"Reference","previous_headings":"","what":"Input Shape","title":"Just your regular densely-connected NN layer. — layer_dense","text":"N-D tensor shape: (batch_size, ..., input_dim). common situation 2D input shape (batch_size, input_dim).","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_dense.html","id":"output-shape","dir":"Reference","previous_headings":"","what":"Output Shape","title":"Just your regular densely-connected NN layer. — layer_dense","text":"N-D tensor shape: (batch_size, ..., units). instance, 2D input shape (batch_size, input_dim), output shape (batch_size, units).","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_dense.html","id":"methods","dir":"Reference","previous_headings":"","what":"Methods","title":"Just your regular densely-connected NN layer. — layer_dense","text":"","code":"enable_lora(   rank,   lora_alpha = NULL,   a_initializer = 'he_uniform',   b_initializer = 'zeros' ) quantize(mode, type_check = TRUE)"},{"path":"https://keras3.posit.co/dev/reference/layer_dense.html","id":"readonly-properties-","dir":"Reference","previous_headings":"","what":"Readonly properties:","title":"Just your regular densely-connected NN layer. — layer_dense","text":"kernel","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_depthwise_conv_1d.html","id":null,"dir":"Reference","previous_headings":"","what":"1D depthwise convolution layer. — layer_depthwise_conv_1d","title":"1D depthwise convolution layer. — layer_depthwise_conv_1d","text":"Depthwise convolution type convolution input channel convolved different kernel (called depthwise kernel). can understand depthwise convolution first step depthwise separable convolution. implemented via following steps: Split input individual channels. Convolve channel individual depthwise kernel depth_multiplier output channels. Concatenate convolved outputs along channels axis. Unlike regular 1D convolution, depthwise convolution mix information across different input channels. depth_multiplier argument determines many filters applied one input channel. , controls amount output channels generated per input channel depthwise step.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_depthwise_conv_1d.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"1D depthwise convolution layer. — layer_depthwise_conv_1d","text":"","code":"layer_depthwise_conv_1d(   object,   kernel_size,   strides = 1L,   padding = \"valid\",   depth_multiplier = 1L,   data_format = NULL,   dilation_rate = 1L,   activation = NULL,   use_bias = TRUE,   depthwise_initializer = \"glorot_uniform\",   bias_initializer = \"zeros\",   depthwise_regularizer = NULL,   bias_regularizer = NULL,   activity_regularizer = NULL,   depthwise_constraint = NULL,   bias_constraint = NULL,   ... )"},{"path":"https://keras3.posit.co/dev/reference/layer_depthwise_conv_1d.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"1D depthwise convolution layer. — layer_depthwise_conv_1d","text":"object Object compose layer . tensor, array, sequential model. kernel_size int list 1 integer, specifying size depthwise convolution window. strides int list 1 integer, specifying stride length convolution. strides > 1 incompatible dilation_rate > 1. padding string, either \"valid\" \"\" (case-insensitive). \"valid\" means padding. \"\" results padding evenly left/right /input. padding=\"\" strides=1, output size input. depth_multiplier number depthwise convolution output channels input channel. total number depthwise convolution output channels equal input_channel * depth_multiplier. data_format string, either \"channels_last\" \"channels_first\". ordering dimensions inputs. \"channels_last\" corresponds inputs shape (batch, steps, features) \"channels_first\" corresponds inputs shape (batch, features, steps). defaults image_data_format value found Keras config file ~/.keras/keras.json. never set , \"channels_last\". dilation_rate int list 1 integers, specifying dilation rate use dilated convolution. activation Activation function. NULL, activation applied. use_bias bool, TRUE, bias added output. depthwise_initializer Initializer convolution kernel. NULL, default initializer (\"glorot_uniform\") used. bias_initializer Initializer bias vector. NULL, default initializer (\"zeros\") used. depthwise_regularizer Optional regularizer convolution kernel. bias_regularizer Optional regularizer bias vector. activity_regularizer Optional regularizer function output. depthwise_constraint Optional projection function applied kernel updated Optimizer (e.g. used implement norm constraints value constraints layer weights). function must take input unprojected variable must return projected variable (must shape). Constraints safe use asynchronous distributed training. bias_constraint Optional projection function applied bias updated Optimizer. ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_depthwise_conv_1d.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"1D depthwise convolution layer. — layer_depthwise_conv_1d","text":"3D tensor representing activation(depthwise_conv1d(inputs, kernel) + bias).","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_depthwise_conv_1d.html","id":"input-shape","dir":"Reference","previous_headings":"","what":"Input Shape","title":"1D depthwise convolution layer. — layer_depthwise_conv_1d","text":"data_format=\"channels_last\": 3D tensor shape: (batch_shape, steps, channels) data_format=\"channels_first\": 3D tensor shape: (batch_shape, channels, steps)","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_depthwise_conv_1d.html","id":"output-shape","dir":"Reference","previous_headings":"","what":"Output Shape","title":"1D depthwise convolution layer. — layer_depthwise_conv_1d","text":"data_format=\"channels_last\": 3D tensor shape: (batch_shape, new_steps, channels * depth_multiplier) data_format=\"channels_first\": 3D tensor shape: (batch_shape, channels * depth_multiplier, new_steps)","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_depthwise_conv_1d.html","id":"raises","dir":"Reference","previous_headings":"","what":"Raises","title":"1D depthwise convolution layer. — layer_depthwise_conv_1d","text":"ValueError: strides > 1 dilation_rate > 1.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_depthwise_conv_1d.html","id":"example","dir":"Reference","previous_headings":"","what":"Example","title":"1D depthwise convolution layer. — layer_depthwise_conv_1d","text":"","code":"x <- random_uniform(c(4, 10, 12)) y <- x |> layer_depthwise_conv_1d(   kernel_size = 3,   depth_multiplier = 3,   activation = 'relu' ) shape(y) ## shape(4, 8, 36)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_depthwise_conv_2d.html","id":null,"dir":"Reference","previous_headings":"","what":"2D depthwise convolution layer. — layer_depthwise_conv_2d","title":"2D depthwise convolution layer. — layer_depthwise_conv_2d","text":"Depthwise convolution type convolution input channel convolved different kernel (called depthwise kernel). can understand depthwise convolution first step depthwise separable convolution. implemented via following steps: Split input individual channels. Convolve channel individual depthwise kernel depth_multiplier output channels. Concatenate convolved outputs along channels axis. Unlike regular 2D convolution, depthwise convolution mix information across different input channels. depth_multiplier argument determines many filters applied one input channel. , controls amount output channels generated per input channel depthwise step.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_depthwise_conv_2d.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"2D depthwise convolution layer. — layer_depthwise_conv_2d","text":"","code":"layer_depthwise_conv_2d(   object,   kernel_size,   strides = list(1L, 1L),   padding = \"valid\",   depth_multiplier = 1L,   data_format = NULL,   dilation_rate = list(1L, 1L),   activation = NULL,   use_bias = TRUE,   depthwise_initializer = \"glorot_uniform\",   bias_initializer = \"zeros\",   depthwise_regularizer = NULL,   bias_regularizer = NULL,   activity_regularizer = NULL,   depthwise_constraint = NULL,   bias_constraint = NULL,   ... )"},{"path":"https://keras3.posit.co/dev/reference/layer_depthwise_conv_2d.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"2D depthwise convolution layer. — layer_depthwise_conv_2d","text":"object Object compose layer . tensor, array, sequential model. kernel_size int list 2 integer, specifying size depthwise convolution window. strides int list 2 integer, specifying stride length depthwise convolution. strides > 1 incompatible dilation_rate > 1. padding string, either \"valid\" \"\" (case-insensitive). \"valid\" means padding. \"\" results padding evenly left/right /input. padding=\"\" strides=1, output size input. depth_multiplier number depthwise convolution output channels input channel. total number depthwise convolution output channels equal input_channel * depth_multiplier. data_format string, either \"channels_last\" \"channels_first\". ordering dimensions inputs. \"channels_last\" corresponds inputs shape (batch, height, width, channels) \"channels_first\" corresponds inputs shape (batch, channels, height, width). defaults image_data_format value found Keras config file ~/.keras/keras.json. never set , \"channels_last\". dilation_rate int list 2 integers, specifying dilation rate use dilated convolution. activation Activation function. NULL, activation applied. use_bias bool, TRUE, bias added output. depthwise_initializer Initializer convolution kernel. NULL, default initializer (\"glorot_uniform\") used. bias_initializer Initializer bias vector. NULL, default initializer (\"zeros\") used. depthwise_regularizer Optional regularizer convolution kernel. bias_regularizer Optional regularizer bias vector. activity_regularizer Optional regularizer function output. depthwise_constraint Optional projection function applied kernel updated Optimizer (e.g. used implement norm constraints value constraints layer weights). function must take input unprojected variable must return projected variable (must shape). Constraints safe use asynchronous distributed training. bias_constraint Optional projection function applied bias updated Optimizer. ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_depthwise_conv_2d.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"2D depthwise convolution layer. — layer_depthwise_conv_2d","text":"4D tensor representing activation(depthwise_conv2d(inputs, kernel) + bias).","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_depthwise_conv_2d.html","id":"input-shape","dir":"Reference","previous_headings":"","what":"Input Shape","title":"2D depthwise convolution layer. — layer_depthwise_conv_2d","text":"data_format=\"channels_last\": 4D tensor shape: (batch_size, height, width, channels) data_format=\"channels_first\": 4D tensor shape: (batch_size, channels, height, width)","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_depthwise_conv_2d.html","id":"output-shape","dir":"Reference","previous_headings":"","what":"Output Shape","title":"2D depthwise convolution layer. — layer_depthwise_conv_2d","text":"data_format=\"channels_last\": 4D tensor shape: (batch_size, new_height, new_width, channels * depth_multiplier) data_format=\"channels_first\": 4D tensor shape: (batch_size, channels * depth_multiplier, new_height, new_width)","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_depthwise_conv_2d.html","id":"raises","dir":"Reference","previous_headings":"","what":"Raises","title":"2D depthwise convolution layer. — layer_depthwise_conv_2d","text":"ValueError: strides > 1 dilation_rate > 1.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_depthwise_conv_2d.html","id":"example","dir":"Reference","previous_headings":"","what":"Example","title":"2D depthwise convolution layer. — layer_depthwise_conv_2d","text":"","code":"x <- random_uniform(c(4, 10, 10, 12)) y <- x |> layer_depthwise_conv_2d(kernel_size = 3, activation = 'relu') shape(y) ## shape(4, 8, 8, 12)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_discretization.html","id":null,"dir":"Reference","previous_headings":"","what":"A preprocessing layer which buckets continuous features by ranges. — layer_discretization","title":"A preprocessing layer which buckets continuous features by ranges. — layer_discretization","text":"layer place element input data one several contiguous ranges output integer index indicating range element placed . Note: layer safe use inside tf.data pipeline (independently backend using).","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_discretization.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"A preprocessing layer which buckets continuous features by ranges. — layer_discretization","text":"","code":"layer_discretization(   object,   bin_boundaries = NULL,   num_bins = NULL,   epsilon = 0.01,   output_mode = \"int\",   sparse = FALSE,   dtype = NULL,   name = NULL )"},{"path":"https://keras3.posit.co/dev/reference/layer_discretization.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"A preprocessing layer which buckets continuous features by ranges. — layer_discretization","text":"object Object compose layer . tensor, array, sequential model. bin_boundaries list bin boundaries. leftmost rightmost bins always extend -Inf Inf, bin_boundaries = c(0, 1, 2) generates bins (-Inf, 0), [0, 1), [1, 2), [2, +Inf). option set, adapt() called. num_bins integer number bins compute. option set, bin_boundaries set adapt() called learn bin boundaries. epsilon Error tolerance, typically small fraction close zero (e.g. 0.01). Higher values epsilon increase quantile approximation, hence result unequal buckets, improve performance resource consumption. output_mode Specification output layer. Values can \"int\", \"one_hot\", \"multi_hot\", \"count\" configuring layer follows: \"int\": Return discretized bin indices directly. \"one_hot\": Encodes individual element input array size num_bins, containing 1 input's bin index. last dimension size 1, encode dimension.  last dimension size 1, append new dimension encoded output. \"multi_hot\": Encodes sample input single array size num_bins, containing 1 bin index index present sample. Treats last dimension sample dimension, input shape (..., sample_length), output shape (..., num_tokens). \"count\": \"multi_hot\", int array contains count number times bin index appeared sample. Defaults \"int\". sparse Boolean. applicable \"one_hot\", \"multi_hot\", \"count\" output modes. supported TensorFlow backend. TRUE, returns SparseTensor instead dense Tensor. Defaults FALSE. dtype datatype (e.g., \"float32\"). name String, name object","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_discretization.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"A preprocessing layer which buckets continuous features by ranges. — layer_discretization","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_discretization.html","id":"input-shape","dir":"Reference","previous_headings":"","what":"Input Shape","title":"A preprocessing layer which buckets continuous features by ranges. — layer_discretization","text":"array dimension 2 higher.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_discretization.html","id":"output-shape","dir":"Reference","previous_headings":"","what":"Output Shape","title":"A preprocessing layer which buckets continuous features by ranges. — layer_discretization","text":"input shape.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_discretization.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"A preprocessing layer which buckets continuous features by ranges. — layer_discretization","text":"Discretize float values based provided buckets.     Discretize float values based number buckets compute.","code":"input <- op_array(rbind(c(-1.5, 1, 3.4, 0.5),                        c(0, 3, 1.3, 0),                        c(-.5, 0, .5, 1),                        c(1.5, 2, 2.5, 3))) output <- input |> layer_discretization(bin_boundaries = c(0, 1, 2)) output ## tf.Tensor( ## [[0 2 3 1] ##  [1 3 2 1] ##  [0 1 1 2] ##  [2 3 3 3]], shape=(4, 4), dtype=int64) layer <- layer_discretization(num_bins = 4, epsilon = 0.01) layer |> adapt(input) layer(input) ## tf.Tensor( ## [[0 2 3 1] ##  [1 3 2 1] ##  [0 1 1 2] ##  [2 3 3 3]], shape=(4, 4), dtype=int64)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_dot.html","id":null,"dir":"Reference","previous_headings":"","what":"Computes element-wise dot product of two tensors. — layer_dot","title":"Computes element-wise dot product of two tensors. — layer_dot","text":"takes list inputs size 2, axes corresponding input along dot product performed. say x y two input tensors shapes (2, 3, 5) (2, 10, 3). batch dimension size inputs, axes correspond dimensions size corresponding inputs. e.g. axes = c(1, 2), dot product x, y result tensor shape (2, 5, 10)","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_dot.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes element-wise dot product of two tensors. — layer_dot","text":"","code":"layer_dot(inputs, ..., axes, normalize = FALSE)"},{"path":"https://keras3.posit.co/dev/reference/layer_dot.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Computes element-wise dot product of two tensors. — layer_dot","text":"inputs layers combine ... Standard layer keyword arguments. axes Integer list integers, axis axes along take dot product. list, two integers corresponding desired axis first input desired axis second input, respectively. Note size two selected axes must match. normalize Whether L2-normalize samples along dot product axis taking dot product. set TRUE, output dot product cosine proximity two samples.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_dot.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Computes element-wise dot product of two tensors. — layer_dot","text":"tensor, dot product samples inputs.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_dot.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Computes element-wise dot product of two tensors. — layer_dot","text":"Usage Keras model:","code":"x <- op_reshape(0:9,   c(1, 5, 2)) y <- op_reshape(10:19, c(1, 2, 5)) layer_dot(x, y, axes=c(2, 3)) ## tf.Tensor( ## [[[260 360] ##   [320 445]]], shape=(1, 2, 2), dtype=int32) x1 <- op_reshape(0:9, c(5, 2)) |> layer_dense(8) x2 <- op_reshape(10:19, c(5, 2)) |> layer_dense(8) shape(x1) ## shape(5, 8) shape(x2) ## shape(5, 8) y <- layer_dot(x1, x2, axes=2) shape(y) ## shape(5, 1)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_dropout.html","id":null,"dir":"Reference","previous_headings":"","what":"Applies dropout to the input. — layer_dropout","title":"Applies dropout to the input. — layer_dropout","text":"Dropout layer randomly sets input units 0 frequency rate step training time, helps prevent overfitting. Inputs set 0 scaled 1 / (1 - rate) sum inputs unchanged. Note Dropout layer applies training set TRUE call(), values dropped inference. using model.fit, training appropriately set TRUE automatically. contexts, can set argument explicitly TRUE calling layer. (contrast setting trainable=FALSE Dropout layer. trainable affect layer's behavior, Dropout variables/weights can frozen training.)","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_dropout.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Applies dropout to the input. — layer_dropout","text":"","code":"layer_dropout(object, rate, noise_shape = NULL, seed = NULL, ...)"},{"path":"https://keras3.posit.co/dev/reference/layer_dropout.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Applies dropout to the input. — layer_dropout","text":"object Object compose layer . tensor, array, sequential model. rate Float 0 1. Fraction input units drop. noise_shape 1D integer tensor representing shape binary dropout mask multiplied input. instance, inputs shape (batch_size, timesteps, features) want dropout mask timesteps, can use noise_shape=(batch_size, 1, features). seed R integer use random seed. ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_dropout.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Applies dropout to the input. — layer_dropout","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_dropout.html","id":"call-arguments","dir":"Reference","previous_headings":"","what":"Call Arguments","title":"Applies dropout to the input. — layer_dropout","text":"inputs: Input tensor (rank). training: Python boolean indicating whether layer behave training mode (adding dropout) inference mode (nothing).","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_einsum_dense.html","id":null,"dir":"Reference","previous_headings":"","what":"A layer that uses einsum as the backing computation. — layer_einsum_dense","title":"A layer that uses einsum as the backing computation. — layer_einsum_dense","text":"layer can perform einsum calculations arbitrary dimensionality.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_einsum_dense.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"A layer that uses einsum as the backing computation. — layer_einsum_dense","text":"","code":"layer_einsum_dense(   object,   equation,   output_shape,   activation = NULL,   bias_axes = NULL,   kernel_initializer = \"glorot_uniform\",   bias_initializer = \"zeros\",   kernel_regularizer = NULL,   bias_regularizer = NULL,   kernel_constraint = NULL,   bias_constraint = NULL,   lora_rank = NULL,   lora_alpha = NULL,   ... )"},{"path":"https://keras3.posit.co/dev/reference/layer_einsum_dense.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"A layer that uses einsum as the backing computation. — layer_einsum_dense","text":"object Object compose layer . tensor, array, sequential model. equation equation describing einsum perform. equation must valid einsum string form ab,bc->ac, ...ab,bc->...ac, ab...,bc->ac... 'ab', 'bc', 'ac' can valid einsum axis expression sequence. output_shape expected shape output tensor (excluding batch dimension dimensions represented ellipses). can specify NA NULL dimension unknown can inferred input shape. activation Activation function use. specify anything, activation applied (, \"linear\" activation: (x) = x). bias_axes string containing output dimension(s) apply bias . character bias_axes string correspond character output portion equation string. kernel_initializer Initializer kernel weights matrix. bias_initializer Initializer bias vector. kernel_regularizer Regularizer function applied kernel weights matrix. bias_regularizer Regularizer function applied bias vector. kernel_constraint Constraint function applied kernel weights matrix. bias_constraint Constraint function applied bias vector. lora_rank Optional integer. set, layer's forward pass implement LoRA (Low-Rank Adaptation) provided rank. LoRA sets layer's kernel non-trainable replaces delta original kernel, obtained via multiplying two lower-rank trainable matrices (factorization happens last dimension). can useful reduce computation cost fine-tuning large dense layers. can also enable LoRA existing EinsumDense layer calling layer$enable_lora(rank). lora_alpha Optional integer. Scales low-rank adaptation delta forward pass. delta scaled lora_alpha / lora_rank, letting tune LoRA adjustment strength independently lora_rank. ... Base layer keyword arguments, name dtype.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_einsum_dense.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"A layer that uses einsum as the backing computation. — layer_einsum_dense","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_einsum_dense.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"A layer that uses einsum as the backing computation. — layer_einsum_dense","text":"Biased dense layer einsums example shows instantiate standard Keras dense layer using einsum operations. example equivalent layer_Dense(64, use_bias=TRUE).     Applying dense layer sequence example shows instantiate layer applies dense operation every element sequence. , output_shape two values (since two non-batch dimensions output); first dimension output_shape NA, sequence dimension b unknown shape.     Applying dense layer sequence using ellipses example shows instantiate layer applies dense operation every element sequence, uses ellipsis notation instead specifying batch sequence dimensions. using ellipsis notation specified one axis, output_shape arg single value. instantiated way, layer can handle number sequence dimensions - including case sequence dimension exists.","code":"input <- layer_input(shape = c(32)) output <- input |>   layer_einsum_dense(\"ab,bc->ac\",                      output_shape = 64,                      bias_axes = \"c\") output # shape(NA, 64) ## <KerasTensor shape=(None, 64), dtype=float32, sparse=False, ragged=False, name=keras_tensor_1> input <- layer_input(shape = c(32, 128)) output <- input |>   layer_einsum_dense(\"abc,cd->abd\",                      output_shape = c(NA, 64),                      bias_axes = \"d\") output  # shape(NA, 32, 64) ## <KerasTensor shape=(None, None, 64), dtype=float32, sparse=False, ragged=False, name=keras_tensor_3> input <- layer_input(shape = c(32, 128)) output <- input |>   layer_einsum_dense(\"...x,xy->...y\",                      output_shape = 64,                      bias_axes = \"y\")  output  # shape(NA, 32, 64) ## <KerasTensor shape=(None, 32, 64), dtype=float32, sparse=False, ragged=False, name=keras_tensor_5>"},{"path":"https://keras3.posit.co/dev/reference/layer_einsum_dense.html","id":"methods","dir":"Reference","previous_headings":"","what":"Methods","title":"A layer that uses einsum as the backing computation. — layer_einsum_dense","text":"","code":"enable_lora(   rank,   a_initializer = 'he_uniform',   b_initializer = 'zeros' ) quantize(mode, type_check = TRUE)"},{"path":"https://keras3.posit.co/dev/reference/layer_einsum_dense.html","id":"readonly-properties-","dir":"Reference","previous_headings":"","what":"Readonly properties:","title":"A layer that uses einsum as the backing computation. — layer_einsum_dense","text":"kernel","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_embedding.html","id":null,"dir":"Reference","previous_headings":"","what":"Turns nonnegative integers (indexes) into dense vectors of fixed size. — layer_embedding","title":"Turns nonnegative integers (indexes) into dense vectors of fixed size. — layer_embedding","text":"e.g. rbind(4L, 20L) \\(\\rightarrow\\) rbind(c(0.25, 0.1), c(0.6, -0.2)) layer can used nonnegative integer inputs fixed range.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_embedding.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Turns nonnegative integers (indexes) into dense vectors of fixed size. — layer_embedding","text":"","code":"layer_embedding(   object,   input_dim,   output_dim,   embeddings_initializer = \"uniform\",   embeddings_regularizer = NULL,   embeddings_constraint = NULL,   mask_zero = FALSE,   weights = NULL,   lora_rank = NULL,   lora_alpha = NULL,   ... )"},{"path":"https://keras3.posit.co/dev/reference/layer_embedding.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Turns nonnegative integers (indexes) into dense vectors of fixed size. — layer_embedding","text":"object Object compose layer . tensor, array, sequential model. input_dim Integer. Size vocabulary, .e. maximum integer index + 1. output_dim Integer. Dimension dense embedding. embeddings_initializer Initializer embeddings matrix (see keras3::initializer_*). embeddings_regularizer Regularizer function applied embeddings matrix (see keras3::regularizer_*). embeddings_constraint Constraint function applied embeddings matrix (see keras3::constraint_*). mask_zero Boolean, whether input value 0 special \"padding\" value masked . useful using recurrent layers may take variable length input. TRUE, subsequent layers model need support masking exception raised. mask_zero set TRUE, consequence, index 0 used vocabulary (input_dim equal size vocabulary + 1). weights Optional floating-point matrix size (input_dim, output_dim). initial embeddings values use. lora_rank Optional integer. set, layer's forward pass implement LoRA (Low-Rank Adaptation) provided rank. LoRA sets layer's embeddings matrix non-trainable replaces delta original matrix, obtained via multiplying two lower-rank trainable matrices. can useful reduce computation cost fine-tuning large embedding layers. can also enable LoRA existing Embedding layer instance calling layer$enable_lora(rank). lora_alpha Optional integer. Scales low-rank adaptation delta forward pass. delta scaled lora_alpha / lora_rank, letting tune LoRA adjustment strength independently lora_rank. ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_embedding.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Turns nonnegative integers (indexes) into dense vectors of fixed size. — layer_embedding","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_embedding.html","id":"example","dir":"Reference","previous_headings":"","what":"Example","title":"Turns nonnegative integers (indexes) into dense vectors of fixed size. — layer_embedding","text":"","code":"model <- keras_model_sequential() |>   layer_embedding(1000, 64)  # The model will take as input an integer matrix of size (batch,input_length), # and the largest integer (i.e. word index) in the input # should be no larger than 999 (vocabulary size). # Now model$output_shape is (NA, 10, 64), where `NA` is the batch # dimension.  input_array <- random_integer(shape = c(32, 10), minval = 0, maxval = 1000) model |> compile('rmsprop', 'mse') output_array <- model |> predict(input_array, verbose = 0) dim(output_array)    # (32, 10, 64) ## [1] 32 10 64"},{"path":"https://keras3.posit.co/dev/reference/layer_embedding.html","id":"input-shape","dir":"Reference","previous_headings":"","what":"Input Shape","title":"Turns nonnegative integers (indexes) into dense vectors of fixed size. — layer_embedding","text":"2D tensor shape: (batch_size, input_length).","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_embedding.html","id":"output-shape","dir":"Reference","previous_headings":"","what":"Output Shape","title":"Turns nonnegative integers (indexes) into dense vectors of fixed size. — layer_embedding","text":"3D tensor shape: (batch_size, input_length, output_dim).","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_embedding.html","id":"methods","dir":"Reference","previous_headings":"","what":"Methods","title":"Turns nonnegative integers (indexes) into dense vectors of fixed size. — layer_embedding","text":"","code":"enable_lora(   rank,   lora_alpha = NULL,   a_initializer = 'he_uniform',   b_initializer = 'zeros' ) compute_output_spec(...) quantize(mode, type_check = TRUE) quantized_build(input_shape, mode) quantized_call(...)"},{"path":"https://keras3.posit.co/dev/reference/layer_embedding.html","id":"readonly-properties-","dir":"Reference","previous_headings":"","what":"Readonly properties:","title":"Turns nonnegative integers (indexes) into dense vectors of fixed size. — layer_embedding","text":"embeddings","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_equalization.html","id":null,"dir":"Reference","previous_headings":"","what":"Preprocessing layer for histogram equalization on image channels. — layer_equalization","title":"Preprocessing layer for histogram equalization on image channels. — layer_equalization","text":"Histogram equalization technique adjust image intensities enhance contrast effectively spreading frequent intensity values. layer applies equalization channel-wise basis, can improve visibility details images. layer works grayscale color images, performing equalization independently color channel. inference time, equalization consistently applied. Note: layer safe use inside tf.data pipeline (independently backend using).","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_equalization.html","id":"input-shape","dir":"Reference","previous_headings":"","what":"Input Shape","title":"Preprocessing layer for histogram equalization on image channels. — layer_equalization","text":"3D (unbatched) 4D (batched) tensor shape: (..., height, width, channels), \"channels_last\" format, (..., channels, height, width), \"channels_first\" format.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_equalization.html","id":"output-shape","dir":"Reference","previous_headings":"","what":"Output Shape","title":"Preprocessing layer for histogram equalization on image channels. — layer_equalization","text":"3D (unbatched) 4D (batched) tensor shape: (..., target_height, target_width, channels), (..., channels, target_height, target_width), \"channels_first\" format.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_equalization.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Preprocessing layer for histogram equalization on image channels. — layer_equalization","text":"","code":"layer_equalization(   object,   value_range = list(0L, 255L),   bins = 256L,   data_format = NULL,   ... )"},{"path":"https://keras3.posit.co/dev/reference/layer_equalization.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Preprocessing layer for histogram equalization on image channels. — layer_equalization","text":"object Object compose layer . tensor, array, sequential model. value_range Optional list/tuple 2 floats specifying lower upper limits input data values. Defaults [0, 255]. input image scaled, use appropriate range (e.g., [0.0, 1.0]). equalization scaled range, output values clipped accordingly. bins Integer specifying number histogram bins use equalization. Defaults 256, suitable 8-bit images. Larger values can provide granular intensity redistribution. data_format String, one \"channels_last\" (default) \"channels_first\". ordering dimensions inputs. \"channels_last\" corresponds inputs shape (batch, height, width, channels) \"channels_first\" corresponds inputs shape (batch, channels, height, width). ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_equalization.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Preprocessing layer for histogram equalization on image channels. — layer_equalization","text":"","code":"# Create an equalization layer for standard 8-bit images equalizer <- layer_equalization()  # An image with uneven intensity distribution image <- np_array(...) # your input image  # Apply histogram equalization equalized_image <- equalizer(image)  # For images with custom value range custom_equalizer <- layer_equalization(     value_range=c(0.0, 1.0),  # for normalized images     bins=128  # fewer bins for more subtle equalization ) custom_equalized <- custom_equalizer(normalized_image)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_feature_space.html","id":null,"dir":"Reference","previous_headings":"","what":"One-stop utility for preprocessing and encoding structured data. — layer_feature_space","title":"One-stop utility for preprocessing and encoding structured data. — layer_feature_space","text":"Available feature types: Note features can referred string name, e.g. \"integer_categorical\". using string name, default argument values used.","code":"# Plain float values. feature_float(name = NULL)  # Float values to be preprocessed via featurewise standardization # (i.e. via a `layer_normalization()` layer). feature_float_normalized(name = NULL)  # Float values to be preprocessed via linear rescaling # (i.e. via a `layer_rescaling` layer). feature_float_rescaled(scale = 1., offset = 0., name = NULL)  # Float values to be discretized. By default, the discrete # representation will then be one-hot encoded. feature_float_discretized(   num_bins,   bin_boundaries = NULL,   output_mode = \"one_hot\",   name = NULL )  # Integer values to be indexed. By default, the discrete # representation will then be one-hot encoded. feature_integer_categorical(   max_tokens = NULL,   num_oov_indices = 1,   output_mode = \"one_hot\",   name = NULL )  # String values to be indexed. By default, the discrete # representation will then be one-hot encoded. feature_string_categorical(   max_tokens = NULL,   num_oov_indices = 1,   output_mode = \"one_hot\",   name = NULL )  # Integer values to be hashed into a fixed number of bins. # By default, the discrete representation will then be one-hot encoded. feature_integer_hashed(num_bins, output_mode = \"one_hot\", name = NULL)  # String values to be hashed into a fixed number of bins. # By default, the discrete representation will then be one-hot encoded. feature_string_hashed(num_bins, output_mode = \"one_hot\", name = NULL)"},{"path":"https://keras3.posit.co/dev/reference/layer_feature_space.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"One-stop utility for preprocessing and encoding structured data. — layer_feature_space","text":"","code":"layer_feature_space(   object,   features,   output_mode = \"concat\",   crosses = NULL,   crossing_dim = 32L,   hashing_dim = 32L,   num_discretization_bins = 32L,   name = NULL,   feature_names = NULL )  feature_cross(feature_names, crossing_dim, output_mode = \"one_hot\")  feature_custom(dtype, preprocessor, output_mode)  feature_float(name = NULL)  feature_float_rescaled(scale = 1, offset = 0, name = NULL)  feature_float_normalized(name = NULL)  feature_float_discretized(   num_bins,   bin_boundaries = NULL,   output_mode = \"one_hot\",   name = NULL )  feature_integer_categorical(   max_tokens = NULL,   num_oov_indices = 1,   output_mode = \"one_hot\",   name = NULL )  feature_string_categorical(   max_tokens = NULL,   num_oov_indices = 1,   output_mode = \"one_hot\",   name = NULL )  feature_string_hashed(num_bins, output_mode = \"one_hot\", name = NULL)  feature_integer_hashed(num_bins, output_mode = \"one_hot\", name = NULL)"},{"path":"https://keras3.posit.co/dev/reference/layer_feature_space.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"One-stop utility for preprocessing and encoding structured data. — layer_feature_space","text":"object see description features see description output_mode string. layer_feature_space(), one \"concat\" \"dict\". concat mode, features get concatenated together single vector. dict mode, FeatureSpace returns named list individually encoded features (names input list names). feature_* functions, one : \"int\" \"one_hot\" \"float\". crosses List features crossed together, e.g. crosses=list(c(\"feature_1\", \"feature_2\")). features \"crossed\" hashing combined value fixed-length vector. crossing_dim Default vector size hashing crossed features. Defaults 32. hashing_dim Default vector size hashing features type \"integer_hashed\" \"string_hashed\". Defaults 32. num_discretization_bins Default number bins used discretizing features type \"float_discretized\". Defaults 32. name String, name object feature_names Named list mapping names features type specification, e.g. list(my_feature = \"integer_categorical\") list(my_feature = feature_integer_categorical()). complete list supported types, see \"Available feature types\" paragraph . dtype string, output dtype feature. E.g., \"float32\". preprocessor callable. scale, offset Passed layer_rescaling() num_bins, bin_boundaries Passed layer_discretization() max_tokens, num_oov_indices Passed layer_integer_lookup() feature_integer_categorical() layer_string_lookup() feature_string_categorical().","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_feature_space.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"One-stop utility for preprocessing and encoding structured data. — layer_feature_space","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_feature_space.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"One-stop utility for preprocessing and encoding structured data. — layer_feature_space","text":"Basic usage named list input data:   Basic usage tf.data:   Basic usage Keras Functional API:           Customizing feature feature cross:   Returning dict (named list) integer-encoded features:   Specifying Keras preprocessing layer:   Retrieving underlying Keras preprocessing layers:   Saving reloading FeatureSpace:","code":"raw_data <- list(   float_values = c(0.0, 0.1, 0.2, 0.3),   string_values = c(\"zero\", \"one\", \"two\", \"three\"),   int_values = as.integer(c(0, 1, 2, 3)) )  dataset <- tfdatasets::tensor_slices_dataset(raw_data)  feature_space <- layer_feature_space(   features = list(     float_values = \"float_normalized\",     string_values = \"string_categorical\",     int_values = \"integer_categorical\"   ),   crosses = list(c(\"string_values\", \"int_values\")),   output_mode = \"concat\" )  # Before you start using the feature_space(), # you must `adapt()` it on some data. feature_space |> adapt(dataset)  # You can call the feature_space() on a named list of # data (batched or unbatched). output_vector <- feature_space(raw_data) library(tfdatasets) # Unlabeled data preprocessed_ds <- unlabeled_dataset |>   dataset_map(feature_space)  # Labeled data preprocessed_ds <- labeled_dataset |>   dataset_map(function(x, y) tuple(feature_space(x), y)) # Retrieve a named list of Keras layer_input() objects (inputs <- feature_space$get_inputs()) ## $float_values ## <KerasTensor shape=(None, 1), dtype=float32, sparse=False, ragged=False, name=float_values> ## ## $string_values ## <KerasTensor shape=(None, 1), dtype=string, sparse=False, ragged=False, name=string_values> ## ## $int_values ## <KerasTensor shape=(None, 1), dtype=int32, sparse=False, ragged=False, name=int_values> # Retrieve the corresponding encoded Keras tensors (encoded_features <- feature_space$get_encoded_features()) ## <KerasTensor shape=(None, 43), dtype=float32, sparse=False, ragged=False, name=keras_tensor_7> # Build a Functional model outputs <- encoded_features |> layer_dense(1, activation = \"sigmoid\") model <- keras_model(inputs, outputs) feature_space <- layer_feature_space(   features = list(     float_values = feature_float_normalized(),     string_values = feature_string_categorical(max_tokens = 10),     int_values = feature_integer_categorical(max_tokens = 10)   ),   crosses = list(     feature_cross(c(\"string_values\", \"int_values\"), crossing_dim = 32)   ),   output_mode = \"concat\" ) feature_space <- layer_feature_space(   features = list(     \"string_values\" = feature_string_categorical(output_mode = \"int\"),     \"int_values\" = feature_integer_categorical(output_mode = \"int\")   ),   crosses = list(     feature_cross(       feature_names = c(\"string_values\", \"int_values\"),       crossing_dim = 32,       output_mode = \"int\"     )   ),   output_mode = \"dict\" ) # Let's say that one of the features is a short text paragraph that # we want to encode as a vector (one vector per paragraph) via TF-IDF. data <- list(text = c(\"1st string\", \"2nd string\", \"3rd string\"))  # There's a Keras layer for this: layer_text_vectorization() custom_layer <- layer_text_vectorization(output_mode = \"tf_idf\")  # We can use feature_custom() to create a custom feature # that will use our preprocessing layer. feature_space <- layer_feature_space(   features = list(     text = feature_custom(preprocessor = custom_layer,                           dtype = \"string\",                           output_mode = \"float\"     )   ),   output_mode = \"concat\" ) feature_space |> adapt(tfdatasets::tensor_slices_dataset(data)) output_vector <- feature_space(data) # The preprocessing layer of each feature is available in `$preprocessors`. preprocessing_layer <- feature_space$preprocessors$feature1  # The crossing layer of each feature cross is available in `$crossers`. # It's an instance of layer_hashed_crossing() crossing_layer <- feature_space$crossers[[\"feature1_X_feature2\"]] feature_space$save(\"featurespace.keras\") reloaded_feature_space <- keras$models$load_model(\"featurespace.keras\")"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_flatten.html","id":null,"dir":"Reference","previous_headings":"","what":"Flattens the input. Does not affect the batch size. — layer_flatten","title":"Flattens the input. Does not affect the batch size. — layer_flatten","text":"Flattens input. affect batch size.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_flatten.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Flattens the input. Does not affect the batch size. — layer_flatten","text":"","code":"layer_flatten(object, data_format = NULL, ...)"},{"path":"https://keras3.posit.co/dev/reference/layer_flatten.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Flattens the input. Does not affect the batch size. — layer_flatten","text":"object Object compose layer . tensor, array, sequential model. data_format string, one \"channels_last\" (default) \"channels_first\". ordering dimensions inputs. \"channels_last\" corresponds inputs shape (batch, ..., channels) \"channels_first\" corresponds inputs shape (batch, channels, ...). unspecified, uses image_data_format value found Keras config file ~/.keras/keras.json (exists). Defaults \"channels_last\". ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_flatten.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Flattens the input. Does not affect the batch size. — layer_flatten","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_flatten.html","id":"note","dir":"Reference","previous_headings":"","what":"Note","title":"Flattens the input. Does not affect the batch size. — layer_flatten","text":"inputs shaped (batch) without feature axis, flattening adds extra channel dimension output shape (batch, 1).","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_flatten.html","id":"example","dir":"Reference","previous_headings":"","what":"Example","title":"Flattens the input. Does not affect the batch size. — layer_flatten","text":"","code":"x <- layer_input(shape=c(10, 64)) y <- x |> layer_flatten() shape(y) ## shape(NA, 640)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_flax_module_wrapper.html","id":null,"dir":"Reference","previous_headings":"","what":"Keras Layer that wraps a Flax module. — layer_flax_module_wrapper","title":"Keras Layer that wraps a Flax module. — layer_flax_module_wrapper","text":"layer enables use Flax components form flax.linen.Module instances within Keras using JAX backend Keras. module method use forward pass can specified via method argument __call__ default. method must take following arguments exact names: self method bound module, case default __call__, module otherwise pass module. inputs: inputs model, JAX array PyTree arrays. training (optional): argument specifying training mode inference mode, TRUE passed training mode. FlaxLayer handles non-trainable state model required RNGs automatically. Note mutable parameter flax.linen.Module.apply() set DenyList([\"params\"]), therefore making assumption variables outside \"params\" collection non-trainable weights. example shows create FlaxLayer Flax Module default __call__ method training argument:   example shows wrap module method conform required signature. allows multiple input arguments training argument different name values. additionally shows use function bound module.","code":"# keras3::use_backend(\"jax\") # py_install(\"flax\", \"r-keras\")  if(config_backend() == \"jax\" &&    reticulate::py_module_available(\"flax\")) {  flax <- import(\"flax\")  MyFlaxModule(flax$linen$Module) %py_class% {   `__call__` <- flax$linen$compact(\\(self, inputs) {     inputs |>       (flax$linen$Conv(features = 32L, kernel_size = tuple(3L, 3L)))() |>       flax$linen$relu() |>       flax$linen$avg_pool(window_shape = tuple(2L, 2L),                           strides = tuple(2L, 2L)) |>       # flatten all except batch_size axis       (\\(x) x$reshape(tuple(x$shape[[1]], -1L)))() |>       (flax$linen$Dense(features = 200L))() |>       flax$linen$relu() |>       (flax$linen$Dense(features = 10L))() |>       flax$linen$softmax()   }) }  # typical usage: input <- keras_input(c(28, 28, 3)) output <- input |>   layer_flax_module_wrapper(MyFlaxModule())  model <- keras_model(input, output)  # to instantiate the layer before composing: flax_module <- MyFlaxModule() keras_layer <- layer_flax_module_wrapper(module = flax_module)  input <- keras_input(c(28, 28, 3)) output <- input |>   keras_layer()  model <- keras_model(input, output)  } flax <- import(\"flax\")  MyFlaxModule(flax$linen$Module) \\%py_class\\% {   forward <-     flax$linen$compact(\\(self, inputs1, input2, deterministic) {       # do work ....       outputs # return     }) }  my_flax_module_wrapper <- function(module, inputs, training) {   c(input1, input2) \\%<-\\% inputs   module$forward(input1, input2,!training) }  flax_module <- MyFlaxModule() keras_layer <- layer_flax_module_wrapper(module = flax_module,                                          method = my_flax_module_wrapper)"},{"path":"https://keras3.posit.co/dev/reference/layer_flax_module_wrapper.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Keras Layer that wraps a Flax module. — layer_flax_module_wrapper","text":"","code":"layer_flax_module_wrapper(object, module, method = NULL, variables = NULL, ...)"},{"path":"https://keras3.posit.co/dev/reference/layer_flax_module_wrapper.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Keras Layer that wraps a Flax module. — layer_flax_module_wrapper","text":"object Object compose layer . tensor, array, sequential model. module instance flax.linen.Module subclass. method method call model. generally method Module. provided, __call__ method used. method can also function defined Module, case must take Module first argument. used Module.init Module.apply. Details documented method argument flax.linen.Module.apply(). variables dict (named R list) containing variables module format returned flax.linen.Module.init(). contain \"params\" key , applicable, keys collections variables non-trainable state. allows passing trained parameters learned non-trainable state controlling initialization. NULL passed, module's init function called build time initialize variables model. ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_flax_module_wrapper.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Keras Layer that wraps a Flax module. — layer_flax_module_wrapper","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_gaussian_dropout.html","id":null,"dir":"Reference","previous_headings":"","what":"Apply multiplicative 1-centered Gaussian noise. — layer_gaussian_dropout","title":"Apply multiplicative 1-centered Gaussian noise. — layer_gaussian_dropout","text":"regularization layer, active training time.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_gaussian_dropout.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Apply multiplicative 1-centered Gaussian noise. — layer_gaussian_dropout","text":"","code":"layer_gaussian_dropout(object, rate, seed = NULL, ...)"},{"path":"https://keras3.posit.co/dev/reference/layer_gaussian_dropout.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Apply multiplicative 1-centered Gaussian noise. — layer_gaussian_dropout","text":"object Object compose layer . tensor, array, sequential model. rate Float, drop probability (Dropout). multiplicative noise standard deviation sqrt(rate / (1 - rate)). seed Integer, optional random seed enable deterministic behavior. ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_gaussian_dropout.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Apply multiplicative 1-centered Gaussian noise. — layer_gaussian_dropout","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_gaussian_dropout.html","id":"call-arguments","dir":"Reference","previous_headings":"","what":"Call Arguments","title":"Apply multiplicative 1-centered Gaussian noise. — layer_gaussian_dropout","text":"inputs: Input tensor (rank). training: Python boolean indicating whether layer behave training mode (adding dropout) inference mode (nothing).","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_gaussian_noise.html","id":null,"dir":"Reference","previous_headings":"","what":"Apply additive zero-centered Gaussian noise. — layer_gaussian_noise","title":"Apply additive zero-centered Gaussian noise. — layer_gaussian_noise","text":"useful mitigate overfitting (see form random data augmentation). Gaussian Noise (GS) natural choice corruption process real valued inputs. regularization layer, active training time.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_gaussian_noise.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Apply additive zero-centered Gaussian noise. — layer_gaussian_noise","text":"","code":"layer_gaussian_noise(object, stddev, seed = NULL, ...)"},{"path":"https://keras3.posit.co/dev/reference/layer_gaussian_noise.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Apply additive zero-centered Gaussian noise. — layer_gaussian_noise","text":"object Object compose layer . tensor, array, sequential model. stddev Float, standard deviation noise distribution. seed Integer, optional random seed enable deterministic behavior. ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_gaussian_noise.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Apply additive zero-centered Gaussian noise. — layer_gaussian_noise","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_gaussian_noise.html","id":"call-arguments","dir":"Reference","previous_headings":"","what":"Call Arguments","title":"Apply additive zero-centered Gaussian noise. — layer_gaussian_noise","text":"inputs: Input tensor (rank). training: Python boolean indicating whether layer behave training mode (adding noise) inference mode (nothing).","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_global_average_pooling_1d.html","id":null,"dir":"Reference","previous_headings":"","what":"Global average pooling operation for temporal data. — layer_global_average_pooling_1d","title":"Global average pooling operation for temporal data. — layer_global_average_pooling_1d","text":"Global average pooling operation temporal data.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_global_average_pooling_1d.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Global average pooling operation for temporal data. — layer_global_average_pooling_1d","text":"","code":"layer_global_average_pooling_1d(   object,   data_format = NULL,   keepdims = FALSE,   ... )"},{"path":"https://keras3.posit.co/dev/reference/layer_global_average_pooling_1d.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Global average pooling operation for temporal data. — layer_global_average_pooling_1d","text":"object Object compose layer . tensor, array, sequential model. data_format string, either \"channels_last\" \"channels_first\". ordering dimensions inputs. \"channels_last\" corresponds inputs shape (batch, steps, features) \"channels_first\" corresponds inputs shape (batch, features, steps). defaults image_data_format value found Keras config file ~/.keras/keras.json. never set , \"channels_last\". keepdims boolean, whether keep temporal dimension . keepdims FALSE (default), rank tensor reduced spatial dimensions. keepdims TRUE, temporal dimension retained length 1. behavior tf$reduce_mean() op_mean(). ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_global_average_pooling_1d.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Global average pooling operation for temporal data. — layer_global_average_pooling_1d","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_global_average_pooling_1d.html","id":"call-arguments","dir":"Reference","previous_headings":"","what":"Call Arguments","title":"Global average pooling operation for temporal data. — layer_global_average_pooling_1d","text":"inputs: 3D tensor. mask: Binary tensor shape (batch_size, steps) indicating whether given step masked (excluded average).","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_global_average_pooling_1d.html","id":"input-shape","dir":"Reference","previous_headings":"","what":"Input Shape","title":"Global average pooling operation for temporal data. — layer_global_average_pooling_1d","text":"data_format='channels_last': 3D tensor shape: (batch_size, steps, features) data_format='channels_first': 3D tensor shape: (batch_size, features, steps)","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_global_average_pooling_1d.html","id":"output-shape","dir":"Reference","previous_headings":"","what":"Output Shape","title":"Global average pooling operation for temporal data. — layer_global_average_pooling_1d","text":"keepdims=FALSE: 2D tensor shape (batch_size, features). keepdims=TRUE: data_format=\"channels_last\": 3D tensor shape (batch_size, 1, features) data_format=\"channels_first\": 3D tensor shape (batch_size, features, 1)","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_global_average_pooling_1d.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Global average pooling operation for temporal data. — layer_global_average_pooling_1d","text":"","code":"x <- random_uniform(c(2, 3, 4)) y <- x |> layer_global_average_pooling_1d() shape(y) ## shape(2, 4)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_global_average_pooling_2d.html","id":null,"dir":"Reference","previous_headings":"","what":"Global average pooling operation for 2D data. — layer_global_average_pooling_2d","title":"Global average pooling operation for 2D data. — layer_global_average_pooling_2d","text":"Global average pooling operation 2D data.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_global_average_pooling_2d.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Global average pooling operation for 2D data. — layer_global_average_pooling_2d","text":"","code":"layer_global_average_pooling_2d(   object,   data_format = NULL,   keepdims = FALSE,   ... )"},{"path":"https://keras3.posit.co/dev/reference/layer_global_average_pooling_2d.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Global average pooling operation for 2D data. — layer_global_average_pooling_2d","text":"object Object compose layer . tensor, array, sequential model. data_format string, either \"channels_last\" \"channels_first\". ordering dimensions inputs. \"channels_last\" corresponds inputs shape (batch, height, width, channels) \"channels_first\" corresponds inputs shape (batch, features, height, weight). defaults image_data_format value found Keras config file ~/.keras/keras.json. never set , \"channels_last\". keepdims boolean, whether keep temporal dimension . keepdims FALSE (default), rank tensor reduced spatial dimensions. keepdims TRUE, spatial dimension retained length 1. behavior tf$reduce_mean() op_mean(). ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_global_average_pooling_2d.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Global average pooling operation for 2D data. — layer_global_average_pooling_2d","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_global_average_pooling_2d.html","id":"input-shape","dir":"Reference","previous_headings":"","what":"Input Shape","title":"Global average pooling operation for 2D data. — layer_global_average_pooling_2d","text":"data_format='channels_last': 4D tensor shape: (batch_size, height, width, channels) data_format='channels_first': 4D tensor shape: (batch_size, channels, height, width)","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_global_average_pooling_2d.html","id":"output-shape","dir":"Reference","previous_headings":"","what":"Output Shape","title":"Global average pooling operation for 2D data. — layer_global_average_pooling_2d","text":"keepdims=FALSE: 2D tensor shape (batch_size, channels). keepdims=TRUE: data_format=\"channels_last\": 4D tensor shape (batch_size, 1, 1, channels) data_format=\"channels_first\": 4D tensor shape (batch_size, channels, 1, 1)","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_global_average_pooling_2d.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Global average pooling operation for 2D data. — layer_global_average_pooling_2d","text":"","code":"x <- random_uniform(c(2, 4, 5, 3)) y <- x |> layer_global_average_pooling_2d() shape(y) ## shape(2, 3)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_global_average_pooling_3d.html","id":null,"dir":"Reference","previous_headings":"","what":"Global average pooling operation for 3D data. — layer_global_average_pooling_3d","title":"Global average pooling operation for 3D data. — layer_global_average_pooling_3d","text":"Global average pooling operation 3D data.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_global_average_pooling_3d.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Global average pooling operation for 3D data. — layer_global_average_pooling_3d","text":"","code":"layer_global_average_pooling_3d(   object,   data_format = NULL,   keepdims = FALSE,   ... )"},{"path":"https://keras3.posit.co/dev/reference/layer_global_average_pooling_3d.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Global average pooling operation for 3D data. — layer_global_average_pooling_3d","text":"object Object compose layer . tensor, array, sequential model. data_format string, either \"channels_last\" \"channels_first\". ordering dimensions inputs. \"channels_last\" corresponds inputs shape (batch, spatial_dim1, spatial_dim2, spatial_dim3, channels) \"channels_first\" corresponds inputs shape (batch, channels, spatial_dim1, spatial_dim2, spatial_dim3). defaults image_data_format value found Keras config file ~/.keras/keras.json. never set , \"channels_last\". keepdims boolean, whether keep temporal dimension . keepdims FALSE (default), rank tensor reduced spatial dimensions. keepdims TRUE, spatial dimension retained length 1. behavior tf$reduce_mean() op_mean(). ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_global_average_pooling_3d.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Global average pooling operation for 3D data. — layer_global_average_pooling_3d","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_global_average_pooling_3d.html","id":"input-shape","dir":"Reference","previous_headings":"","what":"Input Shape","title":"Global average pooling operation for 3D data. — layer_global_average_pooling_3d","text":"data_format='channels_last': 5D tensor shape: (batch_size, spatial_dim1, spatial_dim2, spatial_dim3, channels) data_format='channels_first': 5D tensor shape: (batch_size, channels, spatial_dim1, spatial_dim2, spatial_dim3)","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_global_average_pooling_3d.html","id":"output-shape","dir":"Reference","previous_headings":"","what":"Output Shape","title":"Global average pooling operation for 3D data. — layer_global_average_pooling_3d","text":"keepdims=FALSE: 2D tensor shape (batch_size, channels). keepdims=TRUE: data_format=\"channels_last\": 5D tensor shape (batch_size, 1, 1, 1, channels) data_format=\"channels_first\": 5D tensor shape (batch_size, channels, 1, 1, 1)","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_global_average_pooling_3d.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Global average pooling operation for 3D data. — layer_global_average_pooling_3d","text":"","code":"x <- random_uniform(c(2, 4, 5, 4, 3)) y <- x |> layer_global_average_pooling_3d() shape(y) ## shape(2, 3)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_global_max_pooling_1d.html","id":null,"dir":"Reference","previous_headings":"","what":"Global max pooling operation for temporal data. — layer_global_max_pooling_1d","title":"Global max pooling operation for temporal data. — layer_global_max_pooling_1d","text":"Global max pooling operation temporal data.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_global_max_pooling_1d.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Global max pooling operation for temporal data. — layer_global_max_pooling_1d","text":"","code":"layer_global_max_pooling_1d(object, data_format = NULL, keepdims = FALSE, ...)"},{"path":"https://keras3.posit.co/dev/reference/layer_global_max_pooling_1d.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Global max pooling operation for temporal data. — layer_global_max_pooling_1d","text":"object Object compose layer . tensor, array, sequential model. data_format string, either \"channels_last\" \"channels_first\". ordering dimensions inputs. \"channels_last\" corresponds inputs shape (batch, steps, features) \"channels_first\" corresponds inputs shape (batch, features, steps). defaults image_data_format value found Keras config file ~/.keras/keras.json. never set , \"channels_last\". keepdims boolean, whether keep temporal dimension . keepdims FALSE (default), rank tensor reduced spatial dimensions. keepdims TRUE, temporal dimension retained length 1. behavior tf$reduce_mean() op_mean(). ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_global_max_pooling_1d.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Global max pooling operation for temporal data. — layer_global_max_pooling_1d","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_global_max_pooling_1d.html","id":"input-shape","dir":"Reference","previous_headings":"","what":"Input Shape","title":"Global max pooling operation for temporal data. — layer_global_max_pooling_1d","text":"data_format='channels_last': 3D tensor shape: (batch_size, steps, features) data_format='channels_first': 3D tensor shape: (batch_size, features, steps)","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_global_max_pooling_1d.html","id":"output-shape","dir":"Reference","previous_headings":"","what":"Output Shape","title":"Global max pooling operation for temporal data. — layer_global_max_pooling_1d","text":"keepdims=FALSE: 2D tensor shape (batch_size, features). keepdims=TRUE: data_format=\"channels_last\": 3D tensor shape (batch_size, 1, features) data_format=\"channels_first\": 3D tensor shape (batch_size, features, 1)","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_global_max_pooling_1d.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Global max pooling operation for temporal data. — layer_global_max_pooling_1d","text":"","code":"x <- random_uniform(c(2, 3, 4)) y <- x |> layer_global_max_pooling_1d() shape(y) ## shape(2, 4)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_global_max_pooling_2d.html","id":null,"dir":"Reference","previous_headings":"","what":"Global max pooling operation for 2D data. — layer_global_max_pooling_2d","title":"Global max pooling operation for 2D data. — layer_global_max_pooling_2d","text":"Global max pooling operation 2D data.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_global_max_pooling_2d.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Global max pooling operation for 2D data. — layer_global_max_pooling_2d","text":"","code":"layer_global_max_pooling_2d(object, data_format = NULL, keepdims = FALSE, ...)"},{"path":"https://keras3.posit.co/dev/reference/layer_global_max_pooling_2d.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Global max pooling operation for 2D data. — layer_global_max_pooling_2d","text":"object Object compose layer . tensor, array, sequential model. data_format string, either \"channels_last\" \"channels_first\". ordering dimensions inputs. \"channels_last\" corresponds inputs shape (batch, height, width, channels) \"channels_first\" corresponds inputs shape (batch, features, height, weight). defaults image_data_format value found Keras config file ~/.keras/keras.json. never set , \"channels_last\". keepdims boolean, whether keep temporal dimension . keepdims FALSE (default), rank tensor reduced spatial dimensions. keepdims TRUE, spatial dimension retained length 1. behavior tf$reduce_mean() op_mean(). ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_global_max_pooling_2d.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Global max pooling operation for 2D data. — layer_global_max_pooling_2d","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_global_max_pooling_2d.html","id":"input-shape","dir":"Reference","previous_headings":"","what":"Input Shape","title":"Global max pooling operation for 2D data. — layer_global_max_pooling_2d","text":"data_format='channels_last': 4D tensor shape: (batch_size, height, width, channels) data_format='channels_first': 4D tensor shape: (batch_size, channels, height, width)","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_global_max_pooling_2d.html","id":"output-shape","dir":"Reference","previous_headings":"","what":"Output Shape","title":"Global max pooling operation for 2D data. — layer_global_max_pooling_2d","text":"keepdims=FALSE: 2D tensor shape (batch_size, channels). keepdims=TRUE: data_format=\"channels_last\": 4D tensor shape (batch_size, 1, 1, channels) data_format=\"channels_first\": 4D tensor shape (batch_size, channels, 1, 1)","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_global_max_pooling_2d.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Global max pooling operation for 2D data. — layer_global_max_pooling_2d","text":"","code":"x <- random_uniform(c(2, 4, 5, 3)) y <- x |> layer_global_max_pooling_2d() shape(y) ## shape(2, 3)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_global_max_pooling_3d.html","id":null,"dir":"Reference","previous_headings":"","what":"Global max pooling operation for 3D data. — layer_global_max_pooling_3d","title":"Global max pooling operation for 3D data. — layer_global_max_pooling_3d","text":"Global max pooling operation 3D data.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_global_max_pooling_3d.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Global max pooling operation for 3D data. — layer_global_max_pooling_3d","text":"","code":"layer_global_max_pooling_3d(object, data_format = NULL, keepdims = FALSE, ...)"},{"path":"https://keras3.posit.co/dev/reference/layer_global_max_pooling_3d.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Global max pooling operation for 3D data. — layer_global_max_pooling_3d","text":"object Object compose layer . tensor, array, sequential model. data_format string, either \"channels_last\" \"channels_first\". ordering dimensions inputs. \"channels_last\" corresponds inputs shape (batch, spatial_dim1, spatial_dim2, spatial_dim3, channels) \"channels_first\" corresponds inputs shape (batch, channels, spatial_dim1, spatial_dim2, spatial_dim3). defaults image_data_format value found Keras config file ~/.keras/keras.json. never set , \"channels_last\". keepdims boolean, whether keep temporal dimension . keepdims FALSE (default), rank tensor reduced spatial dimensions. keepdims TRUE, spatial dimension retained length 1. behavior tf$reduce_mean() op_mean(). ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_global_max_pooling_3d.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Global max pooling operation for 3D data. — layer_global_max_pooling_3d","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_global_max_pooling_3d.html","id":"input-shape","dir":"Reference","previous_headings":"","what":"Input Shape","title":"Global max pooling operation for 3D data. — layer_global_max_pooling_3d","text":"data_format='channels_last': 5D tensor shape: (batch_size, spatial_dim1, spatial_dim2, spatial_dim3, channels) data_format='channels_first': 5D tensor shape: (batch_size, channels, spatial_dim1, spatial_dim2, spatial_dim3)","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_global_max_pooling_3d.html","id":"output-shape","dir":"Reference","previous_headings":"","what":"Output Shape","title":"Global max pooling operation for 3D data. — layer_global_max_pooling_3d","text":"keepdims=FALSE: 2D tensor shape (batch_size, channels). keepdims=TRUE: data_format=\"channels_last\": 5D tensor shape (batch_size, 1, 1, 1, channels) data_format=\"channels_first\": 5D tensor shape (batch_size, channels, 1, 1, 1)","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_global_max_pooling_3d.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Global max pooling operation for 3D data. — layer_global_max_pooling_3d","text":"","code":"x <- random_uniform(c(2, 4, 5, 4, 3)) y <- x |> layer_global_max_pooling_3d() shape(y) ## shape(2, 3)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_group_normalization.html","id":null,"dir":"Reference","previous_headings":"","what":"Group normalization layer. — layer_group_normalization","title":"Group normalization layer. — layer_group_normalization","text":"Group Normalization divides channels groups computes within group mean variance normalization. Empirically, accuracy stable batch norm wide range small batch sizes, learning rate adjusted linearly batch sizes. Relation Layer Normalization: number groups set 1, operation becomes nearly identical Layer Normalization (see Layer Normalization docs details). Relation Instance Normalization: number groups set input dimension (number groups equal number channels), operation becomes identical Instance Normalization. can achieve via groups=-1.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_group_normalization.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Group normalization layer. — layer_group_normalization","text":"","code":"layer_group_normalization(   object,   groups = 32L,   axis = -1L,   epsilon = 0.001,   center = TRUE,   scale = TRUE,   beta_initializer = \"zeros\",   gamma_initializer = \"ones\",   beta_regularizer = NULL,   gamma_regularizer = NULL,   beta_constraint = NULL,   gamma_constraint = NULL,   ... )"},{"path":"https://keras3.posit.co/dev/reference/layer_group_normalization.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Group normalization layer. — layer_group_normalization","text":"object Object compose layer . tensor, array, sequential model. groups Integer, number groups Group Normalization. Can range [1, N] N input dimension. input dimension must divisible number groups. Defaults 32. axis Integer List/Tuple. axis axes normalize across. Typically, features axis/axes. left-axes typically batch axis/axes. -1 last dimension input. Defaults -1. epsilon Small float added variance avoid dividing zero. Defaults 1e-3. center TRUE, add offset beta normalized tensor. FALSE, beta ignored. Defaults TRUE. scale TRUE, multiply gamma. FALSE, gamma used. next layer linear (also e.g. relu), can disabled since scaling done next layer. Defaults TRUE. beta_initializer Initializer beta weight. Defaults zeros. gamma_initializer Initializer gamma weight. Defaults ones. beta_regularizer Optional regularizer beta weight. NULL default. gamma_regularizer Optional regularizer gamma weight. NULL default. beta_constraint Optional constraint beta weight. NULL default. gamma_constraint Optional constraint gamma weight. NULL default. ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_group_normalization.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Group normalization layer. — layer_group_normalization","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_group_normalization.html","id":"input-shape","dir":"Reference","previous_headings":"","what":"Input Shape","title":"Group normalization layer. — layer_group_normalization","text":"Arbitrary. Use keyword argument input_shape (tuple integers, include samples axis) using layer first layer model.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_group_normalization.html","id":"output-shape","dir":"Reference","previous_headings":"","what":"Output Shape","title":"Group normalization layer. — layer_group_normalization","text":"shape input. **kwargs: Base layer keyword arguments (e.g. name dtype).","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_group_normalization.html","id":"reference","dir":"Reference","previous_headings":"","what":"Reference","title":"Group normalization layer. — layer_group_normalization","text":"Yuxin Wu & Kaiming , 2018","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_group_query_attention.html","id":null,"dir":"Reference","previous_headings":"","what":"Grouped Query Attention layer. — layer_group_query_attention","title":"Grouped Query Attention layer. — layer_group_query_attention","text":"implementation grouped-query attention introduced Ainslie et al., 2023. num_key_value_heads denotes number groups, setting num_key_value_heads 1 equivalent multi-query attention, num_key_value_heads equal num_query_heads equivalent multi-head attention. layer first projects query, key, value tensors. , key value repeated match number heads query. , query scaled dot-producted key tensors. softmaxed obtain attention probabilities. value tensors interpolated probabilities concatenated back single tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_group_query_attention.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Grouped Query Attention layer. — layer_group_query_attention","text":"","code":"layer_group_query_attention(   object,   head_dim,   num_query_heads,   num_key_value_heads,   dropout = 0,   use_bias = TRUE,   flash_attention = NULL,   kernel_initializer = \"glorot_uniform\",   bias_initializer = \"zeros\",   kernel_regularizer = NULL,   bias_regularizer = NULL,   activity_regularizer = NULL,   kernel_constraint = NULL,   bias_constraint = NULL,   seed = NULL,   ... )"},{"path":"https://keras3.posit.co/dev/reference/layer_group_query_attention.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Grouped Query Attention layer. — layer_group_query_attention","text":"object Object compose layer . tensor, array, sequential model. head_dim Size attention head. num_query_heads Number query attention heads. num_key_value_heads Number key value attention heads. dropout Dropout probability. use_bias Boolean, whether dense layers use bias vectors/matrices. flash_attention NULL, layer attempts use flash attention faster memory-efficient attention computations possible. behavior can configured using config_enable_flash_attention() config_disable_flash_attention(). kernel_initializer Initializer dense layer kernels. bias_initializer Initializer dense layer biases. kernel_regularizer Regularizer dense layer kernels. bias_regularizer Regularizer dense layer biases. activity_regularizer Regularizer dense layer activity. kernel_constraint Constraint dense layer kernels. bias_constraint Constraint dense layer kernels. seed Optional integer seed dropout layer. ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_group_query_attention.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Grouped Query Attention layer. — layer_group_query_attention","text":"attention_output: Result computation, shape (batch_dim, target_seq_len, feature_dim), target_seq_len target sequence length feature_dim query input last dim. attention_scores: (Optional) attention coefficients shape (batch_dim, num_query_heads, target_seq_len, source_seq_len).","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_group_query_attention.html","id":"call-arguments","dir":"Reference","previous_headings":"","what":"Call Arguments","title":"Grouped Query Attention layer. — layer_group_query_attention","text":"query: Query tensor shape (batch_dim, target_seq_len, feature_dim), batch_dim batch size, target_seq_len length target sequence, feature_dim dimension feature. value: Value tensor shape (batch_dim, source_seq_len, feature_dim), batch_dim batch size, source_seq_len length source sequence, feature_dim dimension feature. key: Optional key tensor shape (batch_dim, source_seq_len, feature_dim). given, use value key value, common case. attention_mask: boolean mask shape (batch_dim, target_seq_len, source_seq_len), prevents attention certain positions. boolean mask specifies query elements can attend key elements, 1 indicates attention 0 indicates attention. Broadcasting can happen missing batch dimensions head dimension. return_attention_scores: boolean indicate whether output (attention_output, attention_scores) TRUE, attention_output FALSE. Defaults FALSE. training: Python boolean indicating whether layer behave training mode (adding dropout) inference mode (dropout). go either using training mode parent layer/model FALSE (inference) parent layer. use_causal_mask: boolean indicate whether apply causal mask prevent tokens attending future tokens (e.g., used decoder Transformer).","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_gru.html","id":null,"dir":"Reference","previous_headings":"","what":"Gated Recurrent Unit - Cho et al. 2014. — layer_gru","title":"Gated Recurrent Unit - Cho et al. 2014. — layer_gru","text":"Based available runtime hardware constraints, layer choose different implementations (cuDNN-based backend-native) maximize performance. GPU available arguments layer meet requirement cuDNN kernel (see details), layer use fast cuDNN implementation using TensorFlow backend. requirements use cuDNN implementation : activation == tanh recurrent_activation == sigmoid recurrent_dropout == 0 unroll FALSE use_bias TRUE reset_after TRUE Inputs, use masking, strictly right-padded. Eager execution enabled outermost context. two variants GRU implementation. default one based v3 reset gate applied hidden state matrix multiplication. one based original order reversed. second variant compatible CuDNNGRU (GPU-) allows inference CPU. Thus separate biases kernel recurrent_kernel. use variant, set reset_after=TRUE recurrent_activation='sigmoid'. example:","code":"inputs <- random_uniform(c(32, 10, 8)) outputs <- inputs |> layer_gru(4) shape(outputs) ## shape(32, 4) # (32, 4) gru <- layer_gru(, 4, return_sequences = TRUE, return_state = TRUE) c(whole_sequence_output, final_state) %<-% gru(inputs) shape(whole_sequence_output) ## shape(32, 10, 4) shape(final_state) ## shape(32, 4)"},{"path":"https://keras3.posit.co/dev/reference/layer_gru.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Gated Recurrent Unit - Cho et al. 2014. — layer_gru","text":"","code":"layer_gru(   object,   units,   activation = \"tanh\",   recurrent_activation = \"sigmoid\",   use_bias = TRUE,   kernel_initializer = \"glorot_uniform\",   recurrent_initializer = \"orthogonal\",   bias_initializer = \"zeros\",   kernel_regularizer = NULL,   recurrent_regularizer = NULL,   bias_regularizer = NULL,   activity_regularizer = NULL,   kernel_constraint = NULL,   recurrent_constraint = NULL,   bias_constraint = NULL,   dropout = 0,   recurrent_dropout = 0,   seed = NULL,   return_sequences = FALSE,   return_state = FALSE,   go_backwards = FALSE,   stateful = FALSE,   unroll = FALSE,   reset_after = TRUE,   use_cudnn = \"auto\",   ... )"},{"path":"https://keras3.posit.co/dev/reference/layer_gru.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Gated Recurrent Unit - Cho et al. 2014. — layer_gru","text":"object Object compose layer . tensor, array, sequential model. units Positive integer, dimensionality output space. activation Activation function use. Default: hyperbolic tangent (tanh). pass NULL, activation applied (ie. \"linear\" activation: (x) = x). recurrent_activation Activation function use recurrent step. Default: sigmoid (sigmoid). pass NULL, activation applied (ie. \"linear\" activation: (x) = x). use_bias Boolean, (default TRUE), whether layer use bias vector. kernel_initializer Initializer kernel weights matrix, used linear transformation inputs. Default: \"glorot_uniform\". recurrent_initializer Initializer recurrent_kernel weights matrix, used linear transformation recurrent state. Default: \"orthogonal\". bias_initializer Initializer bias vector. Default: \"zeros\". kernel_regularizer Regularizer function applied kernel weights matrix. Default: NULL. recurrent_regularizer Regularizer function applied recurrent_kernel weights matrix. Default: NULL. bias_regularizer Regularizer function applied bias vector. Default: NULL. activity_regularizer Regularizer function applied output layer (\"activation\"). Default: NULL. kernel_constraint Constraint function applied kernel weights matrix. Default: NULL. recurrent_constraint Constraint function applied recurrent_kernel weights matrix. Default: NULL. bias_constraint Constraint function applied bias vector. Default: NULL. dropout Float 0 1. Fraction units drop linear transformation inputs. Default: 0. recurrent_dropout Float 0 1. Fraction units drop linear transformation recurrent state. Default: 0. seed Random seed dropout. return_sequences Boolean. Whether return last output output sequence, full sequence. Default: FALSE. return_state Boolean. Whether return last state addition output. Default: FALSE. go_backwards Boolean (default FALSE). TRUE, process input sequence backwards return reversed sequence. stateful Boolean (default: FALSE). TRUE, last state sample index batch used initial state sample index following batch. unroll Boolean (default: FALSE). TRUE, network unrolled, else symbolic loop used. Unrolling can speed-RNN, although tends memory-intensive. Unrolling suitable short sequences. reset_after GRU convention (whether apply reset gate matrix multiplication). FALSE \"\", TRUE \"\" (default cuDNN compatible). use_cudnn Whether use cuDNN-backed implementation. \"auto\" attempt use cuDNN feasible, fallback default implementation . ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_gru.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Gated Recurrent Unit - Cho et al. 2014. — layer_gru","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_gru.html","id":"call-arguments","dir":"Reference","previous_headings":"","what":"Call Arguments","title":"Gated Recurrent Unit - Cho et al. 2014. — layer_gru","text":"inputs: 3D tensor, shape (batch, timesteps, feature). mask: Binary tensor shape (samples, timesteps) indicating whether given timestep masked  (optional). individual TRUE entry indicates corresponding timestep utilized, FALSE entry indicates corresponding timestep ignored. Defaults NULL. training: Python boolean indicating whether layer behave training mode inference mode. argument passed cell calling . relevant dropout recurrent_dropout used  (optional). Defaults NULL. initial_state: List initial state tensors passed first call cell (optional, NULL causes creation zero-filled initial state tensors). Defaults NULL.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_hashed_crossing.html","id":null,"dir":"Reference","previous_headings":"","what":"A preprocessing layer which crosses features using the ","title":"A preprocessing layer which crosses features using the ","text":"layer performs crosses categorical features using \"hashing trick\". Conceptually, transformation can thought : hash(concatenate(features)) %% num_bins. layer currently performs crosses scalar inputs batches scalar inputs. Valid input shapes (batch_size, 1), (batch_size) (). Note: layer wraps tf.keras.layers.HashedCrossing. used part compiled computation graph model backend TensorFlow. can however used backend running eagerly. can also always used part input preprocessing pipeline backend (outside model ), recommend use layer. Note: layer safe use inside tfdatasets pipeline (independently backend using).","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_hashed_crossing.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"A preprocessing layer which crosses features using the ","text":"","code":"layer_hashed_crossing(   object,   num_bins,   output_mode = \"int\",   sparse = FALSE,   name = NULL,   dtype = NULL,   ... )"},{"path":"https://keras3.posit.co/dev/reference/layer_hashed_crossing.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"A preprocessing layer which crosses features using the ","text":"object Object compose layer . tensor, array, sequential model. num_bins Number hash bins. output_mode Specification output layer. Values can \"int\", \"one_hot\" configuring layer follows: \"int\": Return integer bin indices directly. \"one_hot\": Encodes individual element input array size num_bins, containing 1 input's bin index. Defaults \"int\". sparse Boolean. applicable \"one_hot\" mode valid using TensorFlow backend. TRUE, returns SparseTensor instead dense Tensor. Defaults FALSE. name String, name object dtype datatype (e.g., \"float32\"). ... Keyword arguments construct layer.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_hashed_crossing.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"A preprocessing layer which crosses features using the ","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_hashed_crossing.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"A preprocessing layer which crosses features using the ","text":"Crossing two scalar features.     Crossing one-hotting two scalar features.","code":"feat1 <- c('A', 'B', 'A', 'B', 'A') |> as.array() feat2 <- c(101, 101, 101, 102, 102) |> as.integer() |> as.array() layer <- layer_hashed_crossing(num_bins = 5) layer(list(feat1, feat2)) ## tf.Tensor([1 4 1 1 3], shape=(5), dtype=int64) layer <- layer_hashed_crossing(num_bins = 5, output_mode = 'one_hot') layer(list(feat1, feat2)) ## tf.Tensor( ## [[0. 1. 0. 0. 0.] ##  [0. 0. 0. 0. 1.] ##  [0. 1. 0. 0. 0.] ##  [0. 1. 0. 0. 0.] ##  [0. 0. 0. 1. 0.]], shape=(5, 5), dtype=float32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_hashing.html","id":null,"dir":"Reference","previous_headings":"","what":"A preprocessing layer which hashes and bins categorical features. — layer_hashing","title":"A preprocessing layer which hashes and bins categorical features. — layer_hashing","text":"layer transforms categorical inputs hashed output. element-wise converts ints strings ints fixed range. stable hash function uses tensorflow::ops::Fingerprint produce output consistently across platforms. layer uses FarmHash64 default, provides consistent hashed output across different platforms stable across invocations, regardless device context, mixing input bits thoroughly. want obfuscate hashed output, can also pass random salt argument constructor. case, layer use SipHash64 hash function, salt value serving additional input hash function. Note: layer internally uses TensorFlow. used part compiled computation graph model backend TensorFlow. can however used backend running eagerly. can also always used part input preprocessing pipeline backend (outside model ), recommend use layer. Note: layer safe use inside tf.data pipeline (independently backend using). Example (FarmHash64)     Example (FarmHash64) mask value     Example (SipHash64)     Example (Siphash64 single integer, salt=[133, 133])","code":"layer <- layer_hashing(num_bins = 3) inp <- c('A', 'B', 'C', 'D', 'E') |> array(dim = c(5, 1)) layer(inp) ## tf.Tensor( ## [[1] ##  [0] ##  [1] ##  [1] ##  [2]], shape=(5, 1), dtype=int64) layer <- layer_hashing(num_bins=3, mask_value='') inp <- c('A', 'B', '', 'C', 'D') |> array(dim = c(5, 1)) layer(inp) ## tf.Tensor( ## [[1] ##  [1] ##  [0] ##  [2] ##  [2]], shape=(5, 1), dtype=int64) layer <- layer_hashing(num_bins=3, salt=c(133, 137)) inp <- c('A', 'B', 'C', 'D', 'E') |> array(dim = c(5, 1)) layer(inp) ## tf.Tensor( ## [[1] ##  [2] ##  [1] ##  [0] ##  [2]], shape=(5, 1), dtype=int64) layer <- layer_hashing(num_bins=3, salt=133) inp <- c('A', 'B', 'C', 'D', 'E') |> array(dim = c(5, 1)) layer(inp) ## tf.Tensor( ## [[0] ##  [0] ##  [2] ##  [1] ##  [0]], shape=(5, 1), dtype=int64)"},{"path":"https://keras3.posit.co/dev/reference/layer_hashing.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"A preprocessing layer which hashes and bins categorical features. — layer_hashing","text":"","code":"layer_hashing(   object,   num_bins,   mask_value = NULL,   salt = NULL,   output_mode = \"int\",   sparse = FALSE,   ... )"},{"path":"https://keras3.posit.co/dev/reference/layer_hashing.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"A preprocessing layer which hashes and bins categorical features. — layer_hashing","text":"object Object compose layer . tensor, array, sequential model. num_bins Number hash bins. Note includes mask_value bin, effective number bins (num_bins - 1) mask_value set. mask_value value represents masked inputs, mapped index 0. NULL means mask term added hashing start index 0. Defaults NULL. salt single unsigned integer NULL. passed, hash function used SipHash64, values used additional input (known \"salt\" cryptography). non-zero. NULL, uses FarmHash64 hash function. also supports list 2 unsigned integer numbers, see reference paper details. Defaults NULL. output_mode Specification output layer. Values can \"int\", \"one_hot\", \"multi_hot\", \"count\" configuring layer follows: \"int\": Return integer bin indices directly. \"one_hot\": Encodes individual element input array size num_bins, containing 1 input's bin index. last dimension size 1, encode dimension. last dimension size 1, append new dimension encoded output. \"multi_hot\": Encodes sample input single array size num_bins, containing 1 bin index index present sample. Treats last dimension sample dimension, input shape (..., sample_length), output shape (..., num_tokens). \"count\": \"multi_hot\", int array contains count number times bin index appeared sample. Defaults \"int\". sparse Boolean. applicable \"one_hot\", \"multi_hot\", \"count\" output modes. supported TensorFlow backend. TRUE, returns SparseTensor instead dense Tensor. Defaults FALSE. ... Keyword arguments construct layer.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_hashing.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"A preprocessing layer which hashes and bins categorical features. — layer_hashing","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_hashing.html","id":"input-shape","dir":"Reference","previous_headings":"","what":"Input Shape","title":"A preprocessing layer which hashes and bins categorical features. — layer_hashing","text":"single string, list strings, int32 int64 tensor shape (batch_size, ...,).","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_hashing.html","id":"output-shape","dir":"Reference","previous_headings":"","what":"Output Shape","title":"A preprocessing layer which hashes and bins categorical features. — layer_hashing","text":"int32 tensor shape (batch_size, ...).","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_hashing.html","id":"reference","dir":"Reference","previous_headings":"","what":"Reference","title":"A preprocessing layer which hashes and bins categorical features. — layer_hashing","text":"SipHash salt","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_identity.html","id":null,"dir":"Reference","previous_headings":"","what":"Identity layer. — layer_identity","title":"Identity layer. — layer_identity","text":"layer used placeholder operation performed. layer just returns inputs argument output.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_identity.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Identity layer. — layer_identity","text":"","code":"layer_identity(object, ...)"},{"path":"https://keras3.posit.co/dev/reference/layer_identity.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Identity layer. — layer_identity","text":"object Object compose layer . tensor, array, sequential model. ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_identity.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Identity layer. — layer_identity","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_input.html","id":null,"dir":"Reference","previous_headings":"","what":"keras_input — layer_input","title":"keras_input — layer_input","text":"layer_input() alias keras_input(). See ?keras_input() full documentation.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_input.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"keras_input — layer_input","text":"","code":"layer_input(   shape = NULL,   batch_size = NULL,   dtype = NULL,   sparse = NULL,   ragged = NULL,   batch_shape = NULL,   name = NULL,   tensor = NULL,   optional = FALSE )"},{"path":"https://keras3.posit.co/dev/reference/layer_input.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"keras_input — layer_input","text":"shape shape list (list integers NULL objects), including batch size. instance, shape = c(32) indicates expected input batches 32-dimensional vectors. Elements list can NULL NA; NULL/NA elements represent dimensions shape known may vary (e.g. sequence length). batch_size Optional static batch size (integer). dtype data type expected input, string (e.g. \"float32\", \"int32\"...) sparse boolean specifying whether expected input sparse tensors. Note , sparse FALSE, sparse tensors can still passed input - densified default value 0. feature supported TensorFlow backend. Defaults FALSE. ragged boolean specifying whether expected input ragged tensors. Note , ragged FALSE, ragged tensors can still passed input - densified default value 0. feature supported TensorFlow backend. Defaults FALSE. batch_shape Optional shape list (list integers NULL objects), including batch size. name Optional name string layer. unique model (reuse name twice). autogenerated provided. tensor Optional existing tensor wrap Input layer. set, layer use tensor rather creating new placeholder tensor. optional Boolean, whether input optional . optional input can accept NULL values.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_input.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"keras_input — layer_input","text":"Keras tensor, can passed inputs argument (keras_model()).","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_integer_lookup.html","id":null,"dir":"Reference","previous_headings":"","what":"A preprocessing layer that maps integers to (possibly encoded) indices. — layer_integer_lookup","title":"A preprocessing layer that maps integers to (possibly encoded) indices. — layer_integer_lookup","text":"layer maps set arbitrary integer input tokens indexed integer output via table-based vocabulary lookup. layer's output indices contiguously arranged maximum vocab size, even input tokens non-continguous unbounded. layer supports multiple options encoding output via output_mode, optional support --vocabulary (OOV) tokens masking. vocabulary layer must either supplied construction learned via adapt(). adapt(), layer analyze data set, determine frequency individual integer tokens, create vocabulary . vocabulary capped size, frequent tokens used create vocabulary others treated OOV. two possible output modes layer.  output_mode \"int\", input integers converted index vocabulary (integer).  output_mode \"multi_hot\", \"count\", \"tf_idf\", input integers encoded array dimension corresponds element vocabulary. vocabulary can optionally contain mask token well OOV token (can optionally occupy multiple indices vocabulary, set num_oov_indices). position tokens vocabulary fixed. output_mode \"int\", vocabulary begin mask token index 0, followed OOV indices, followed rest vocabulary. output_mode \"multi_hot\", \"count\", \"tf_idf\" vocabulary begin OOV indices instances mask token dropped. Note: layer uses TensorFlow internally. used part compiled computation graph model backend TensorFlow. can however used backend running eagerly. can also always used part input preprocessing pipeline backend (outside model ), recommend use layer. Note: layer safe use inside tf.data pipeline (independently backend using). Note: working layer outputs directly (e.g., passing outputs another layer, using lower-level operations like [ op_*): returned indices 0-based. However, default settings, first (0) index OOV token, returned indices offset 1 may appear 1-based.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_integer_lookup.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"A preprocessing layer that maps integers to (possibly encoded) indices. — layer_integer_lookup","text":"","code":"layer_integer_lookup(   object,   max_tokens = NULL,   num_oov_indices = 1L,   mask_token = NULL,   oov_token = -1L,   vocabulary = NULL,   vocabulary_dtype = \"int64\",   idf_weights = NULL,   invert = FALSE,   output_mode = \"int\",   sparse = FALSE,   pad_to_max_tokens = FALSE,   name = NULL,   ... )"},{"path":"https://keras3.posit.co/dev/reference/layer_integer_lookup.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"A preprocessing layer that maps integers to (possibly encoded) indices. — layer_integer_lookup","text":"object Object compose layer . tensor, array, sequential model. max_tokens Maximum size vocabulary layer. specified adapting vocabulary setting pad_to_max_tokens=TRUE. NULL, cap size vocabulary. Note size includes OOV mask tokens. Defaults NULL. num_oov_indices number --vocabulary tokens use. value 1, OOV inputs modulated determine OOV value. value 0, OOV inputs cause error calling layer. Defaults 1. mask_token integer token represents masked inputs. output_mode \"int\", token included vocabulary mapped index 0. output modes, token appear vocabulary instances mask token input dropped. set NULL, mask term added. Defaults NULL. oov_token used invert TRUE. token return OOV indices. Defaults -1. vocabulary Optional. Either array integers string path text file. passing array, can pass list, list, 1D NumPy array, 1D tensor containing integer vocabulary terms. passing file path, file contain one line per term vocabulary. argument set, need adapt() layer. vocabulary_dtype dtype vocabulary terms, example \"int64\" \"int32\". Defaults \"int64\". idf_weights valid output_mode \"tf_idf\". list, list, 1D NumPy array, 1D tensor length vocabulary, containing floating point inverse document frequency weights, multiplied per sample term counts final TF-IDF weight. vocabulary argument set, output_mode \"tf_idf\", argument must supplied. invert valid output_mode \"int\". TRUE, layer map indices vocabulary items instead mapping vocabulary items indices. Defaults FALSE. output_mode Specification output layer. Values can \"int\", \"one_hot\", \"multi_hot\", \"count\", \"tf_idf\" configuring layer follows: \"int\": Return vocabulary indices input tokens. \"one_hot\": Encodes individual element input array size vocabulary, containing 1 element index. last dimension size 1, encode dimension. last dimension size 1, append new dimension encoded output. \"multi_hot\": Encodes sample input single array size vocabulary, containing 1 vocabulary term present sample. Treats last dimension sample dimension, input shape (..., sample_length), output shape (..., num_tokens). \"count\": \"multi_hot\", int array contains count number times token index appeared sample. \"tf_idf\": \"multi_hot\", TF-IDF algorithm applied find value token slot. \"int\" output, shape input output supported. output modes, currently output rank 2 supported. Defaults \"int\". sparse Boolean. applicable \"multi_hot\", \"count\", \"tf_idf\" output modes. supported TensorFlow backend. TRUE, returns SparseTensor instead dense Tensor. Defaults FALSE. pad_to_max_tokens applicable output_mode \"multi_hot\", \"count\", \"tf_idf\". TRUE, output feature axis padded max_tokens even number unique tokens vocabulary less max_tokens, resulting tensor shape (batch_size, max_tokens) regardless vocabulary size. Defaults FALSE. name String, name object ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_integer_lookup.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"A preprocessing layer that maps integers to (possibly encoded) indices. — layer_integer_lookup","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_integer_lookup.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"A preprocessing layer that maps integers to (possibly encoded) indices. — layer_integer_lookup","text":"Creating lookup layer known vocabulary example creates lookup layer pre-existing vocabulary.     Creating lookup layer adapted vocabulary example creates lookup layer generates vocabulary analyzing dataset.     Note OOV token -1 added vocabulary. remaining tokens sorted frequency (42, 2 occurrences, first) inverse sort order.     Lookups multiple OOV indices example demonstrates use lookup layer multiple OOV indices.  layer created one OOV index, OOV tokens hashed number OOV buckets, distributing OOV tokens deterministic fashion across set.     Note output OOV token 37 1, output OOV token 1000 0. -vocab terms output index increased 1 earlier examples (12 maps 2, etc) order make space extra OOV token. One-hot output Configure layer output_mode='one_hot'. Note first num_oov_indices dimensions ont_hot encoding represent OOV values.     Multi-hot output Configure layer output_mode = 'multi_hot'. Note first num_oov_indices dimensions multi_hot encoding represent OOV tokens     Token count output Configure layer output_mode='count'. multi_hot output, first num_oov_indices dimensions output represent OOV tokens.     TF-IDF output Configure layer output_mode='tf_idf'. multi_hot output, first num_oov_indices dimensions output represent OOV tokens. token bin output token_count * idf_weight, idf weights inverse document frequency weights per token. provided along vocabulary. Note idf_weight OOV tokens default average idf weights passed .     specify idf weights oov tokens, need pass entire vocabulary including leading oov token.     adapting layer \"tf_idf\" mode, input sample considered document, IDF weight per token calculated : log(1 + num_documents / (1 + token_document_count)). Inverse lookup example demonstrates map indices tokens using layer. (can also use adapt() inverse = TRUE, simplicity pass vocab example.)     Note first index correspond oov token default. Forward inverse lookup pairs example demonstrates use vocabulary standard lookup layer create inverse lookup layer.     example, input token 1000 resulted output -1, since 1000 vocabulary - got represented OOV, OOV tokens returned -1 inverse layer. Also, note inverse work, must already set forward layer vocabulary either directly via adapt() calling get_vocabulary().","code":"vocab <- c(12, 36, 1138, 42) |> as.integer() data <- op_array(rbind(c(12, 1138, 42),                        c(42, 1000, 36)))  # Note OOV tokens out <- data |> layer_integer_lookup(vocabulary = vocab) out ## tf.Tensor( ## [[1 3 4] ##  [4 0 2]], shape=(2, 3), dtype=int64) data <- op_array(rbind(c(12, 1138, 42),                        c(42, 1000, 36)))  # Note OOV tokens layer <- layer_integer_lookup() layer |> adapt(data) layer |> get_vocabulary() |> str() ## List of 6 ##  $ : int -1 ##  $ : num 42 ##  $ : num 1138 ##  $ : num 1000 ##  $ : num 36 ##  $ : num 12 layer(data) ## tf.Tensor( ## [[5 2 1] ##  [1 3 4]], shape=(2, 3), dtype=int64) vocab <- c(12, 36, 1138, 42) |> as.integer() data <- op_array(rbind(c(12, 1138, 42),                        c(37, 1000, 36)))  # Note OOV tokens out <- data |>   layer_integer_lookup(vocabulary = vocab,                        num_oov_indices = 2) out ## tf.Tensor( ## [[2 4 5] ##  [1 0 3]], shape=(2, 3), dtype=int64) vocab <- c(12, 36, 1138, 42) |> as.integer() data <- op_array(c(12, 36, 1138, 42, 7), 'int32')  # Note OOV tokens layer <- layer_integer_lookup(vocabulary = vocab,                               output_mode = 'one_hot') layer(data) ## tf.Tensor( ## [[0 1 0 0 0] ##  [0 0 1 0 0] ##  [0 0 0 1 0] ##  [0 0 0 0 1] ##  [1 0 0 0 0]], shape=(5, 5), dtype=int64) vocab <- c(12, 36, 1138, 42) |> as.integer() data <- op_array(rbind(c(12, 1138, 42, 42),                       c(42,    7, 36,  7)), \"int64\")  # Note OOV tokens layer <- layer_integer_lookup(vocabulary = vocab,                               output_mode = 'multi_hot') layer(data) ## tf.Tensor( ## [[0 1 0 1 1] ##  [1 0 1 0 1]], shape=(2, 5), dtype=int64) vocab <- c(12, 36, 1138, 42) |> as.integer() data <- rbind(c(12, 1138, 42, 42),               c(42,    7, 36,  7)) |> op_array(\"int64\") layer <- layer_integer_lookup(vocabulary = vocab,                               output_mode = 'count') layer(data) ## tf.Tensor( ## [[0 1 0 1 2] ##  [2 0 1 0 1]], shape=(2, 5), dtype=int64) vocab <- c(12, 36, 1138, 42) |> as.integer() idf_weights <- c(0.25, 0.75, 0.6, 0.4) data <- rbind(c(12, 1138, 42, 42),               c(42,    7, 36,  7)) |> op_array(\"int64\") layer <- layer_integer_lookup(output_mode = 'tf_idf',                               vocabulary = vocab,                               idf_weights = idf_weights) layer(data) ## tf.Tensor( ## [[0.   0.25 0.   0.6  0.8 ] ##  [1.   0.   0.75 0.   0.4 ]], shape=(2, 5), dtype=float32) vocab <- c(-1, 12, 36, 1138, 42) |> as.integer() idf_weights <- c(0.9, 0.25, 0.75, 0.6, 0.4) data <- rbind(c(12, 1138, 42, 42),               c(42,    7, 36,  7)) |> op_array(\"int64\") layer <- layer_integer_lookup(output_mode = 'tf_idf',                               vocabulary = vocab,                               idf_weights = idf_weights) layer(data) ## tf.Tensor( ## [[0.   0.25 0.   0.6  0.8 ] ##  [1.8  0.   0.75 0.   0.4 ]], shape=(2, 5), dtype=float32) vocab <- c(12, 36, 1138, 42) |> as.integer() data <- op_array(c(1, 3, 4,                    4, 0, 2)) |> op_reshape(c(2,-1)) |> op_cast(\"int32\") layer <- layer_integer_lookup(vocabulary = vocab, invert = TRUE) layer(data) ## tf.Tensor( ## [[  12 1138   42] ##  [  42   -1   36]], shape=(2, 3), dtype=int64) vocab <- c(12, 36, 1138, 42) |> as.integer() data <- op_array(rbind(c(12, 1138, 42), c(42, 1000, 36)), \"int32\") layer <- layer_integer_lookup(vocabulary = vocab) i_layer <- layer_integer_lookup(vocabulary = get_vocabulary(layer),                                 invert = TRUE) int_data <- layer(data) i_layer(int_data) ## tf.Tensor( ## [[  12 1138   42] ##  [  42   -1   36]], shape=(2, 3), dtype=int64)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_jax_model_wrapper.html","id":null,"dir":"Reference","previous_headings":"","what":"Keras Layer that wraps a JAX model. — layer_jax_model_wrapper","title":"Keras Layer that wraps a JAX model. — layer_jax_model_wrapper","text":"layer enables use JAX components within Keras using JAX backend Keras.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_jax_model_wrapper.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Keras Layer that wraps a JAX model. — layer_jax_model_wrapper","text":"","code":"layer_jax_model_wrapper(   object,   call_fn,   init_fn = NULL,   params = NULL,   state = NULL,   seed = NULL,   ...,   dtype = NULL )"},{"path":"https://keras3.posit.co/dev/reference/layer_jax_model_wrapper.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Keras Layer that wraps a JAX model. — layer_jax_model_wrapper","text":"object Object compose layer . tensor, array, sequential model. call_fn function call model. See description list arguments takes outputs returns. init_fn function call initialize model. See description list arguments takes outputs returns. NULL, params /state must provided. params PyTree containing model trainable parameters. allows passing trained parameters controlling initialization. params state NULL, init_fn() called build time initialize trainable parameters model. state PyTree containing model non-trainable state. allows passing learned state controlling initialization. params state NULL, call_fn() takes state argument, init_fn() called build time initialize non-trainable state model. seed Seed random number generator. Optional. ... forward/backward compatability. dtype dtype layer's computations weights. Can also keras.DTypePolicy. Optional. Defaults default policy.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_jax_model_wrapper.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Keras Layer that wraps a JAX model. — layer_jax_model_wrapper","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_jax_model_wrapper.html","id":"model-function","dir":"Reference","previous_headings":"","what":"Model function","title":"Keras Layer that wraps a JAX model. — layer_jax_model_wrapper","text":"layer accepts JAX models form function, call_fn(), must take following arguments exact names: params: trainable parameters model. state (optional): non-trainable state model. Can omitted model non-trainable state. rng (optional): jax.random.PRNGKey instance. Can omitted model need RNGs, neither training inference. inputs: inputs model, JAX array PyTree arrays. training (optional): argument specifying training mode inference mode, TRUE passed training mode. Can omitted model behaves training mode inference mode. inputs argument mandatory. Inputs model must provided via single argument. JAX model takes multiple inputs separate arguments, must combined single structure, instance tuple() dict().","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_jax_model_wrapper.html","id":"model-weights-initialization","dir":"Reference","previous_headings":"","what":"Model weights initialization","title":"Keras Layer that wraps a JAX model. — layer_jax_model_wrapper","text":"initialization params state model can handled layer, case init_fn() argument must provided. allows model initialized dynamically right shape. Alternatively, shape known, params argument optionally state argument can used create already initialized model. init_fn() function, provided, must take following arguments exact names: rng: jax.random.PRNGKey instance. inputs: JAX array PyTree arrays placeholder values provide shape inputs. training (optional): argument specifying training mode inference mode. True always passed init_fn. Can omitted regardless whether call_fn training argument.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_jax_model_wrapper.html","id":"models-with-non-trainable-state","dir":"Reference","previous_headings":"","what":"Models with non-trainable state","title":"Keras Layer that wraps a JAX model. — layer_jax_model_wrapper","text":"JAX models non-trainable state: call_fn() must state argument call_fn() must return tuple() containing outputs model new non-trainable state model init_fn() must return tuple() containing initial trainable params model initial non-trainable state model. code shows possible combination call_fn() init_fn() signatures model non-trainable state. example, model training argument rng argument call_fn().","code":"stateful_call <- function(params, state, rng, inputs, training) {   outputs <- ....   new_state <- ....   tuple(outputs, new_state) }  stateful_init <- function(rng, inputs) {   initial_params <- ....   initial_state <- ....   tuple(initial_params, initial_state) }"},{"path":"https://keras3.posit.co/dev/reference/layer_jax_model_wrapper.html","id":"models-without-non-trainable-state","dir":"Reference","previous_headings":"","what":"Models without non-trainable state","title":"Keras Layer that wraps a JAX model. — layer_jax_model_wrapper","text":"JAX models non-trainable state: call_fn() must state argument call_fn() must return outputs model init_fn() must return initial trainable params model. code shows possible combination call_fn() init_fn() signatures model without non-trainable state. example, model training argument rng argument call_fn().","code":"stateful_call <- function(pparams, inputs) {   outputs <- ....   outputs }  stateful_init <- function(rng, inputs) {   initial_params <- ....   initial_params }"},{"path":"https://keras3.posit.co/dev/reference/layer_jax_model_wrapper.html","id":"conforming-to-the-required-signature","dir":"Reference","previous_headings":"","what":"Conforming to the required signature","title":"Keras Layer that wraps a JAX model. — layer_jax_model_wrapper","text":"model different signature one required JaxLayer, one can easily write wrapper method adapt arguments. example shows model multiple inputs separate arguments, expects multiple RNGs dict, deterministic argument opposite meaning training. conform, inputs combined single structure using tuple, RNG split used populate expected dict, Boolean flag negated:","code":"jax <- import(\"jax\") my_model_fn <- function(params, rngs, input1, input2, deterministic) {   ....   if (!deterministic) {     dropout_rng <- rngs$dropout     keep <- jax$random$bernoulli(dropout_rng, dropout_rate, x$shape)     x <- jax$numpy$where(keep, x / dropout_rate, 0)     ....   }   ....   return(outputs) }  my_model_wrapper_fn <- function(params, rng, inputs, training) {   c(input1, input2) %<-% inputs   c(rng1, rng2) %<-% jax$random$split(rng)   rngs <-  list(dropout = rng1, preprocessing = rng2)   deterministic <-  !training   my_model_fn(params, rngs, input1, input2, deterministic) }  keras_layer <- layer_jax_model_wrapper(call_fn = my_model_wrapper_fn,                                        params = initial_params)"},{"path":"https://keras3.posit.co/dev/reference/layer_jax_model_wrapper.html","id":"usage-with-haiku-modules","dir":"Reference","previous_headings":"","what":"Usage with Haiku modules","title":"Keras Layer that wraps a JAX model. — layer_jax_model_wrapper","text":"JaxLayer enables use Haiku components form haiku.Module. achieved transforming module per Haiku pattern passing module.apply call_fn parameter module.init init_fn parameter needed. model non-trainable state, transformed haiku.transform_with_state. model non-trainable state, transformed haiku.transform. Additionally, optionally, module use RNGs \"apply\", can transformed haiku.without_apply_rng. following example shows create JaxLayer Haiku module uses random number generators via hk.next_rng_key() takes training positional argument:","code":"# reticulate::py_install(\"haiku\", \"r-keras\") hk <- import(\"haiku\") MyHaikuModule(hk$Module) \\%py_class\\% {    `__call__` <- \\(self, x, training) {     x <- hk$Conv2D(32L, tuple(3L, 3L))(x)     x <- jax$nn$relu(x)     x <- hk$AvgPool(tuple(1L, 2L, 2L, 1L),                     tuple(1L, 2L, 2L, 1L), \"VALID\")(x)     x <- hk$Flatten()(x)     x <- hk$Linear(200L)(x)     if (training)       x <- hk$dropout(rng = hk$next_rng_key(), rate = 0.3, x = x)     x <- jax$nn$relu(x)     x <- hk$Linear(10L)(x)     x <- jax$nn$softmax(x)     x   }  }  my_haiku_module_fn <- function(inputs, training) {   module <- MyHaikuModule()   module(inputs, training) }  transformed_module <- hk$transform(my_haiku_module_fn)  keras_layer <-   layer_jax_model_wrapper(call_fn = transformed_module$apply,                           init_fn = transformed_module$init)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_lambda.html","id":null,"dir":"Reference","previous_headings":"","what":"Wraps arbitrary expressions as a Layer object. — layer_lambda","title":"Wraps arbitrary expressions as a Layer object. — layer_lambda","text":"layer_lambda() layer exists arbitrary expressions can used Layer constructing Sequential Functional API models. Lambda layers best suited simple operations quick experimentation. advanced use cases, prefer writing new subclasses Layer using new_layer_class().","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_lambda.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Wraps arbitrary expressions as a Layer object. — layer_lambda","text":"","code":"layer_lambda(   object,   f,   output_shape = NULL,   mask = NULL,   arguments = NULL,   ... )"},{"path":"https://keras3.posit.co/dev/reference/layer_lambda.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Wraps arbitrary expressions as a Layer object. — layer_lambda","text":"object Object compose layer . tensor, array, sequential model. f function evaluated. Takes input tensor first argument. output_shape Expected output shape function. argument can usually inferred explicitly provided. Can list function. list, specifies first dimension onward; sample dimension assumed either input: output_shape = c(input_shape[1], output_shape) , input NULL sample dimension also NULL: output_shape = c(NA, output_shape). function, specifies entire shape function input shape: output_shape = f(input_shape). mask Either NULL (indicating masking) callable signature compute_mask layer method, tensor returned output mask regardless input . arguments Optional named list arguments passed function. ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_lambda.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Wraps arbitrary expressions as a Layer object. — layer_lambda","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_lambda.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Wraps arbitrary expressions as a Layer object. — layer_lambda","text":"","code":"# add a x -> x^2 layer model <- keras_model_sequential() model |> layer_lambda(\\(x) x^2)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_layer_normalization.html","id":null,"dir":"Reference","previous_headings":"","what":"Layer normalization layer (Ba et al., 2016). — layer_layer_normalization","title":"Layer normalization layer (Ba et al., 2016). — layer_layer_normalization","text":"Normalize activations previous layer given example batch independently, rather across batch like Batch Normalization. .e. applies transformation maintains mean activation within example close 0 activation standard deviation close 1. scale center enabled, layer scale normalized outputs broadcasting trainable variable gamma, center outputs broadcasting trainable variable beta. gamma default ones tensor beta default zeros tensor, centering scaling -ops training begun. , scaling centering enabled normalization equations follows: Let intermediate activations mini-batch inputs. sample x batch inputs, compute mean variance sample, normalize value sample (including small factor epsilon numerical stability), finally, transform normalized output gamma beta, learned parameters:   gamma beta span axes inputs specified axis, part inputs' shape must fully defined. example:         Note implementations layer normalization may choose define gamma beta separate set axes axes normalized across. example, Group Normalization (Wu et al. 2018) group size 1 corresponds layer_layer_normalization() normalizes across height, width, channel gamma beta span channel dimension. , layer_layer_normalization() implementation match layer_group_normalization() layer group size set 1.","code":"outputs <- inputs |> apply(1, function(x) {   x_normalized <- (x - mean(x)) /                   sqrt(var(x) + epsilon)   x_normalized * gamma + beta }) layer <- layer_layer_normalization(axis = c(2, 3, 4))  layer(op_ones(c(5, 20, 30, 40))) |> invisible() # build() shape(layer$beta) ## shape(20, 30, 40) shape(layer$gamma) ## shape(20, 30, 40)"},{"path":"https://keras3.posit.co/dev/reference/layer_layer_normalization.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Layer normalization layer (Ba et al., 2016). — layer_layer_normalization","text":"","code":"layer_layer_normalization(   object,   axis = -1L,   epsilon = 0.001,   center = TRUE,   scale = TRUE,   beta_initializer = \"zeros\",   gamma_initializer = \"ones\",   beta_regularizer = NULL,   gamma_regularizer = NULL,   beta_constraint = NULL,   gamma_constraint = NULL,   ... )"},{"path":"https://keras3.posit.co/dev/reference/layer_layer_normalization.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Layer normalization layer (Ba et al., 2016). — layer_layer_normalization","text":"object Object compose layer . tensor, array, sequential model. axis Integer list. axis axes normalize across. Typically, features axis/axes. left-axes typically batch axis/axes. -1 last dimension input. Defaults -1. epsilon Small float added variance avoid dividing zero. Defaults 1e-3. center TRUE, add offset beta normalized tensor. FALSE, beta ignored. Defaults TRUE. scale TRUE, multiply gamma. FALSE, gamma used. next layer linear (also e.g. layer_activation_relu()), can disabled since scaling done next layer. Defaults TRUE. beta_initializer Initializer beta weight. Defaults zeros. gamma_initializer Initializer gamma weight. Defaults ones. beta_regularizer Optional regularizer beta weight. NULL default. gamma_regularizer Optional regularizer gamma weight. NULL default. beta_constraint Optional constraint beta weight. NULL default. gamma_constraint Optional constraint gamma weight. NULL default. ... Base layer keyword arguments (e.g. name dtype).","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_layer_normalization.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Layer normalization layer (Ba et al., 2016). — layer_layer_normalization","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_layer_normalization.html","id":"reference","dir":"Reference","previous_headings":"","what":"Reference","title":"Layer normalization layer (Ba et al., 2016). — layer_layer_normalization","text":"Lei Ba et al., 2016.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_lstm.html","id":null,"dir":"Reference","previous_headings":"","what":"Long Short-Term Memory layer - Hochreiter 1997. — layer_lstm","title":"Long Short-Term Memory layer - Hochreiter 1997. — layer_lstm","text":"Based available runtime hardware constraints, layer choose different implementations (cuDNN-based backend-native) maximize performance. GPU available arguments layer meet requirement cuDNN kernel (see details), layer use fast cuDNN implementation using TensorFlow backend. requirements use cuDNN implementation : activation == tanh recurrent_activation == sigmoid recurrent_dropout == 0 unroll FALSE use_bias TRUE Inputs, use masking, strictly right-padded. Eager execution enabled outermost context. example:","code":"input <- random_uniform(c(32, 10, 8)) output <- input |> layer_lstm(4) shape(output) ## shape(32, 4) lstm <- layer_lstm(units = 4, return_sequences = TRUE, return_state = TRUE) c(whole_seq_output, final_memory_state, final_carry_state) %<-% lstm(input) shape(whole_seq_output) ## shape(32, 10, 4) shape(final_memory_state) ## shape(32, 4) shape(final_carry_state) ## shape(32, 4)"},{"path":"https://keras3.posit.co/dev/reference/layer_lstm.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Long Short-Term Memory layer - Hochreiter 1997. — layer_lstm","text":"","code":"layer_lstm(   object,   units,   activation = \"tanh\",   recurrent_activation = \"sigmoid\",   use_bias = TRUE,   kernel_initializer = \"glorot_uniform\",   recurrent_initializer = \"orthogonal\",   bias_initializer = \"zeros\",   unit_forget_bias = TRUE,   kernel_regularizer = NULL,   recurrent_regularizer = NULL,   bias_regularizer = NULL,   activity_regularizer = NULL,   kernel_constraint = NULL,   recurrent_constraint = NULL,   bias_constraint = NULL,   dropout = 0,   recurrent_dropout = 0,   seed = NULL,   return_sequences = FALSE,   return_state = FALSE,   go_backwards = FALSE,   stateful = FALSE,   unroll = FALSE,   use_cudnn = \"auto\",   ... )"},{"path":"https://keras3.posit.co/dev/reference/layer_lstm.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Long Short-Term Memory layer - Hochreiter 1997. — layer_lstm","text":"object Object compose layer . tensor, array, sequential model. units Positive integer, dimensionality output space. activation Activation function use. Default: hyperbolic tangent (tanh). pass NULL, activation applied (ie. \"linear\" activation: (x) = x). recurrent_activation Activation function use recurrent step. Default: sigmoid (sigmoid). pass NULL, activation applied (ie. \"linear\" activation: (x) = x). use_bias Boolean, (default TRUE), whether layer use bias vector. kernel_initializer Initializer kernel weights matrix, used linear transformation inputs. Default: \"glorot_uniform\". recurrent_initializer Initializer recurrent_kernel weights matrix, used linear transformation recurrent state. Default: \"orthogonal\". bias_initializer Initializer bias vector. Default: \"zeros\". unit_forget_bias Boolean (default TRUE). TRUE, add 1 bias forget gate initialization. Setting TRUE also force bias_initializer=\"zeros\". recommended Jozefowicz et al. kernel_regularizer Regularizer function applied kernel weights matrix. Default: NULL. recurrent_regularizer Regularizer function applied recurrent_kernel weights matrix. Default: NULL. bias_regularizer Regularizer function applied bias vector. Default: NULL. activity_regularizer Regularizer function applied output layer (\"activation\"). Default: NULL. kernel_constraint Constraint function applied kernel weights matrix. Default: NULL. recurrent_constraint Constraint function applied recurrent_kernel weights matrix. Default: NULL. bias_constraint Constraint function applied bias vector. Default: NULL. dropout Float 0 1. Fraction units drop linear transformation inputs. Default: 0. recurrent_dropout Float 0 1. Fraction units drop linear transformation recurrent state. Default: 0. seed Random seed dropout. return_sequences Boolean. Whether return last output output sequence, full sequence. Default: FALSE. return_state Boolean. Whether return last state addition output. Default: FALSE. go_backwards Boolean (default: FALSE). TRUE, process input sequence backwards return reversed sequence. stateful Boolean (default: FALSE). TRUE, last state sample index batch used initial state sample index following batch. unroll Boolean (default FALSE). TRUE, network unrolled, else symbolic loop used. Unrolling can speed-RNN, although tends memory-intensive. Unrolling suitable short sequences. use_cudnn Whether use cuDNN-backed implementation. \"auto\" attempt use cuDNN feasible, fallback default implementation . ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_lstm.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Long Short-Term Memory layer - Hochreiter 1997. — layer_lstm","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_lstm.html","id":"call-arguments","dir":"Reference","previous_headings":"","what":"Call Arguments","title":"Long Short-Term Memory layer - Hochreiter 1997. — layer_lstm","text":"inputs: 3D tensor, shape (batch, timesteps, feature). mask: Binary tensor shape (samples, timesteps) indicating whether given timestep masked  (optional). individual TRUE entry indicates corresponding timestep utilized, FALSE entry indicates corresponding timestep ignored. Defaults NULL. training: Boolean indicating whether layer behave training mode inference mode. argument passed cell calling . relevant dropout recurrent_dropout used  (optional). Defaults NULL. initial_state: List initial state tensors passed first call cell (optional, NULL causes creation zero-filled initial state tensors). Defaults NULL.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_masking.html","id":null,"dir":"Reference","previous_headings":"","what":"Masks a sequence by using a mask value to skip timesteps. — layer_masking","title":"Masks a sequence by using a mask value to skip timesteps. — layer_masking","text":"timestep input tensor (second dimension tensor), values input tensor timestep equal mask_value, timestep masked (skipped) downstream layers (long support masking). downstream layer support masking yet receives input mask, exception raised.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_masking.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Masks a sequence by using a mask value to skip timesteps. — layer_masking","text":"","code":"layer_masking(object, mask_value = 0, ...)"},{"path":"https://keras3.posit.co/dev/reference/layer_masking.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Masks a sequence by using a mask value to skip timesteps. — layer_masking","text":"object Object compose layer . tensor, array, sequential model. mask_value see description ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_masking.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Masks a sequence by using a mask value to skip timesteps. — layer_masking","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_masking.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Masks a sequence by using a mask value to skip timesteps. — layer_masking","text":"Consider array x shape c(samples, timesteps, features), fed LSTM layer. want mask timestep #3 #5 lack data timesteps. can: Set x[, 3, ] <- 0. x[, 5, ] <- 0. Insert layer_masking() layer mask_value = 0. LSTM layer:","code":"c(samples, timesteps, features) %<-% c(32, 10, 8) inputs <- c(samples, timesteps, features) %>% { array(runif(prod(.)), dim = .) } inputs[, 3, ] <- 0 inputs[, 5, ] <- 0  model <- keras_model_sequential() %>%   layer_masking(mask_value = 0) %>%   layer_lstm(32)  output <- model(inputs) # The time step 3 and 5 will be skipped from LSTM calculation."},{"path":"https://keras3.posit.co/dev/reference/layer_masking.html","id":"note","dir":"Reference","previous_headings":"","what":"Note","title":"Masks a sequence by using a mask value to skip timesteps. — layer_masking","text":"Keras masking convention, masked timestep denoted mask value FALSE, non-masked (.e. usable) timestep denoted mask value TRUE.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_max_num_bounding_boxes.html","id":null,"dir":"Reference","previous_headings":"","what":"Ensure the maximum number of bounding boxes. — layer_max_num_bounding_boxes","title":"Ensure the maximum number of bounding boxes. — layer_max_num_bounding_boxes","text":"Ensure maximum number bounding boxes.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_max_num_bounding_boxes.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Ensure the maximum number of bounding boxes. — layer_max_num_bounding_boxes","text":"","code":"layer_max_num_bounding_boxes(object, max_number, fill_value = -1L, ...)"},{"path":"https://keras3.posit.co/dev/reference/layer_max_num_bounding_boxes.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Ensure the maximum number of bounding boxes. — layer_max_num_bounding_boxes","text":"object Object compose layer . tensor, array, sequential model. max_number Desired output number bounding boxes. fill_value fill value boxes labels bounding_boxes. Defaults -1. ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_max_num_bounding_boxes.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Ensure the maximum number of bounding boxes. — layer_max_num_bounding_boxes","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_max_pooling_1d.html","id":null,"dir":"Reference","previous_headings":"","what":"Max pooling operation for 1D temporal data. — layer_max_pooling_1d","title":"Max pooling operation for 1D temporal data. — layer_max_pooling_1d","text":"Downsamples input representation taking maximum value spatial window size pool_size. window shifted strides. resulting output using \"valid\" padding option shape : output_shape = (input_shape - pool_size + 1) / strides). resulting output shape using \"\" padding option : output_shape = input_shape / strides","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_max_pooling_1d.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Max pooling operation for 1D temporal data. — layer_max_pooling_1d","text":"","code":"layer_max_pooling_1d(   object,   pool_size = 2L,   strides = NULL,   padding = \"valid\",   data_format = NULL,   name = NULL,   ... )"},{"path":"https://keras3.posit.co/dev/reference/layer_max_pooling_1d.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Max pooling operation for 1D temporal data. — layer_max_pooling_1d","text":"object Object compose layer . tensor, array, sequential model. pool_size int, size max pooling window. strides int NULL. Specifies much pooling window moves pooling step. NULL, default pool_size. padding string, either \"valid\" \"\" (case-insensitive). \"valid\" means padding. \"\" results padding evenly left/right /input output height/width dimension input. data_format string, either \"channels_last\" \"channels_first\". ordering dimensions inputs. \"channels_last\" corresponds inputs shape (batch, steps, features) \"channels_first\" corresponds inputs shape (batch, features, steps). defaults image_data_format value found Keras config file ~/.keras/keras.json. never set , \"channels_last\". name String, name object ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_max_pooling_1d.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Max pooling operation for 1D temporal data. — layer_max_pooling_1d","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_max_pooling_1d.html","id":"input-shape","dir":"Reference","previous_headings":"","what":"Input Shape","title":"Max pooling operation for 1D temporal data. — layer_max_pooling_1d","text":"data_format=\"channels_last\": 3D tensor shape (batch_size, steps, features). data_format=\"channels_first\": 3D tensor shape (batch_size, features, steps).","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_max_pooling_1d.html","id":"output-shape","dir":"Reference","previous_headings":"","what":"Output Shape","title":"Max pooling operation for 1D temporal data. — layer_max_pooling_1d","text":"data_format=\"channels_last\": 3D tensor shape (batch_size, downsampled_steps, features). data_format=\"channels_first\": 3D tensor shape (batch_size, features, downsampled_steps).","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_max_pooling_1d.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Max pooling operation for 1D temporal data. — layer_max_pooling_1d","text":"strides=1 padding=\"valid\":     strides=2 padding=\"valid\":     strides=1 padding=\"\":","code":"x <- op_reshape(c(1, 2, 3, 4, 5),                c(1, 5, 1)) max_pool_1d <- layer_max_pooling_1d(pool_size = 2,                                     strides = 1,                                     padding = \"valid\") max_pool_1d(x) ## tf.Tensor( ## [[[2.] ##   [3.] ##   [4.] ##   [5.]]], shape=(1, 4, 1), dtype=float32) x <- op_reshape(c(1, 2, 3, 4, 5),                c(1, 5, 1)) max_pool_1d <- layer_max_pooling_1d(pool_size = 2,                                     strides = 2,                                     padding = \"valid\") max_pool_1d(x) ## tf.Tensor( ## [[[2.] ##   [4.]]], shape=(1, 2, 1), dtype=float32) x <- op_reshape(c(1, 2, 3, 4, 5),                c(1, 5, 1)) max_pool_1d <- layer_max_pooling_1d(pool_size = 2,                                     strides = 1,                                     padding = \"same\") max_pool_1d(x) ## tf.Tensor( ## [[[2.] ##   [3.] ##   [4.] ##   [5.] ##   [5.]]], shape=(1, 5, 1), dtype=float32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_max_pooling_2d.html","id":null,"dir":"Reference","previous_headings":"","what":"Max pooling operation for 2D spatial data. — layer_max_pooling_2d","title":"Max pooling operation for 2D spatial data. — layer_max_pooling_2d","text":"Downsamples input along spatial dimensions (height width) taking maximum value input window (size defined pool_size) channel input. window shifted strides along dimension. resulting output using \"valid\" padding option spatial shape (number rows columns) : output_shape = floor((input_shape - pool_size) / strides) + 1 (input_shape >= pool_size) resulting output shape using \"\" padding option : output_shape = floor((input_shape - 1) / strides) + 1","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_max_pooling_2d.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Max pooling operation for 2D spatial data. — layer_max_pooling_2d","text":"","code":"layer_max_pooling_2d(   object,   pool_size = list(2L, 2L),   strides = NULL,   padding = \"valid\",   data_format = NULL,   name = NULL,   ... )"},{"path":"https://keras3.posit.co/dev/reference/layer_max_pooling_2d.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Max pooling operation for 2D spatial data. — layer_max_pooling_2d","text":"object Object compose layer . tensor, array, sequential model. pool_size int list 2 integers, factors downscale (dim1, dim2). one integer specified, window length used dimensions. strides int list 2 integers, NULL. Strides values. NULL, default pool_size. one int specified, stride size used dimensions. padding string, either \"valid\" \"\" (case-insensitive). \"valid\" means padding. \"\" results padding evenly left/right /input output height/width dimension input. data_format string, either \"channels_last\" \"channels_first\". ordering dimensions inputs. \"channels_last\" corresponds inputs shape (batch, height, width, channels) \"channels_first\" corresponds inputs shape (batch, channels, height, width). defaults image_data_format value found Keras config file ~/.keras/keras.json. never set , \"channels_last\". name String, name object ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_max_pooling_2d.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Max pooling operation for 2D spatial data. — layer_max_pooling_2d","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_max_pooling_2d.html","id":"input-shape","dir":"Reference","previous_headings":"","what":"Input Shape","title":"Max pooling operation for 2D spatial data. — layer_max_pooling_2d","text":"data_format=\"channels_last\": 4D tensor shape (batch_size, height, width, channels). data_format=\"channels_first\": 4D tensor shape (batch_size, channels, height, width).","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_max_pooling_2d.html","id":"output-shape","dir":"Reference","previous_headings":"","what":"Output Shape","title":"Max pooling operation for 2D spatial data. — layer_max_pooling_2d","text":"data_format=\"channels_last\": 4D tensor shape (batch_size, pooled_height, pooled_width, channels). data_format=\"channels_first\": 4D tensor shape (batch_size, channels, pooled_height, pooled_width).","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_max_pooling_2d.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Max pooling operation for 2D spatial data. — layer_max_pooling_2d","text":"strides = (1, 1) padding = \"valid\":     strides = c(2, 2) padding = \"valid\":     stride = (1, 1) padding = \"\":","code":"x <- rbind(c(1., 2., 3.),            c(4., 5., 6.),            c(7., 8., 9.)) |> op_reshape(c(1, 3, 3, 1)) max_pool_2d <- layer_max_pooling_2d(pool_size = c(2, 2),                                     strides = c(1, 1),                                     padding = \"valid\") max_pool_2d(x) ## tf.Tensor( ## [[[[5.] ##    [6.]] ## ##   [[8.] ##    [9.]]]], shape=(1, 2, 2, 1), dtype=float32) x <- rbind(c(1., 2., 3., 4.),            c(5., 6., 7., 8.),            c(9., 10., 11., 12.)) |> op_reshape(c(1, 3, 4, 1)) max_pool_2d <- layer_max_pooling_2d(pool_size = c(2, 2),                                     strides = c(2, 2),                                     padding = \"valid\") max_pool_2d(x) ## tf.Tensor( ## [[[[6.] ##    [8.]]]], shape=(1, 1, 2, 1), dtype=float32) x <- rbind(c(1., 2., 3.),            c(4., 5., 6.),            c(7., 8., 9.)) |> op_reshape(c(1, 3, 3, 1)) max_pool_2d <- layer_max_pooling_2d(pool_size = c(2, 2),                                     strides = c(1, 1),                                     padding = \"same\") max_pool_2d(x) ## tf.Tensor( ## [[[[5.] ##    [6.] ##    [6.]] ## ##   [[8.] ##    [9.] ##    [9.]] ## ##   [[8.] ##    [9.] ##    [9.]]]], shape=(1, 3, 3, 1), dtype=float32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_max_pooling_3d.html","id":null,"dir":"Reference","previous_headings":"","what":"Max pooling operation for 3D data (spatial or spatio-temporal). — layer_max_pooling_3d","title":"Max pooling operation for 3D data (spatial or spatio-temporal). — layer_max_pooling_3d","text":"Downsamples input along spatial dimensions (depth, height, width) taking maximum value input window (size defined pool_size) channel input. window shifted strides along dimension.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_max_pooling_3d.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Max pooling operation for 3D data (spatial or spatio-temporal). — layer_max_pooling_3d","text":"","code":"layer_max_pooling_3d(   object,   pool_size = list(2L, 2L, 2L),   strides = NULL,   padding = \"valid\",   data_format = NULL,   name = NULL,   ... )"},{"path":"https://keras3.posit.co/dev/reference/layer_max_pooling_3d.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Max pooling operation for 3D data (spatial or spatio-temporal). — layer_max_pooling_3d","text":"object Object compose layer . tensor, array, sequential model. pool_size int list 3 integers, factors downscale (dim1, dim2, dim3). one integer specified, window length used dimensions. strides int list 3 integers, NULL. Strides values. NULL, default pool_size. one int specified, stride size used dimensions. padding string, either \"valid\" \"\" (case-insensitive). \"valid\" means padding. \"\" results padding evenly left/right /input output height/width dimension input. data_format string, either \"channels_last\" \"channels_first\". ordering dimensions inputs. \"channels_last\" corresponds inputs shape (batch, spatial_dim1, spatial_dim2, spatial_dim3, channels) \"channels_first\" corresponds inputs shape (batch, channels, spatial_dim1, spatial_dim2, spatial_dim3). defaults image_data_format value found Keras config file ~/.keras/keras.json. never set , \"channels_last\". name String, name object ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_max_pooling_3d.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Max pooling operation for 3D data (spatial or spatio-temporal). — layer_max_pooling_3d","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_max_pooling_3d.html","id":"input-shape","dir":"Reference","previous_headings":"","what":"Input Shape","title":"Max pooling operation for 3D data (spatial or spatio-temporal). — layer_max_pooling_3d","text":"data_format=\"channels_last\": 5D tensor shape: (batch_size, spatial_dim1, spatial_dim2, spatial_dim3, channels) data_format=\"channels_first\": 5D tensor shape: (batch_size, channels, spatial_dim1, spatial_dim2, spatial_dim3)","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_max_pooling_3d.html","id":"output-shape","dir":"Reference","previous_headings":"","what":"Output Shape","title":"Max pooling operation for 3D data (spatial or spatio-temporal). — layer_max_pooling_3d","text":"data_format=\"channels_last\": 5D tensor shape: (batch_size, pooled_dim1, pooled_dim2, pooled_dim3, channels) data_format=\"channels_first\": 5D tensor shape: (batch_size, channels, pooled_dim1, pooled_dim2, pooled_dim3)","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_max_pooling_3d.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Max pooling operation for 3D data (spatial or spatio-temporal). — layer_max_pooling_3d","text":"","code":"depth <- 30 height <- 30 width <- 30 channels <- 3  inputs <- layer_input(shape=c(depth, height, width, channels)) layer <- layer_max_pooling_3d(pool_size=3) outputs <- inputs |> layer() outputs ## <KerasTensor shape=(None, 10, 10, 10, 3), dtype=float32, sparse=False, ragged=False, name=keras_tensor_1>"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_maximum.html","id":null,"dir":"Reference","previous_headings":"","what":"Computes element-wise maximum on a list of inputs. — layer_maximum","title":"Computes element-wise maximum on a list of inputs. — layer_maximum","text":"takes input list tensors, shape, returns single tensor (also shape).","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_maximum.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes element-wise maximum on a list of inputs. — layer_maximum","text":"","code":"layer_maximum(inputs, ...)"},{"path":"https://keras3.posit.co/dev/reference/layer_maximum.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Computes element-wise maximum on a list of inputs. — layer_maximum","text":"inputs layers combine ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_maximum.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Computes element-wise maximum on a list of inputs. — layer_maximum","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_maximum.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Computes element-wise maximum on a list of inputs. — layer_maximum","text":"Usage Keras model:","code":"input_shape <- c(2, 3, 4) x1 <- random_uniform(input_shape) x2 <- random_uniform(input_shape) y <- layer_maximum(x1, x2) input1 <- layer_input(shape = c(16)) x1 <- input1 |> layer_dense(8, activation = 'relu') input2 <- layer_input(shape = c(32)) x2 <- input2 |> layer_dense(8, activation = 'relu') # equivalent to `y <- layer_maximum(x1, x2)` y <- layer_maximum(x1, x2) out <- y |> layer_dense(4) model <- keras_model(inputs = c(input1, input2), outputs = out)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_mel_spectrogram.html","id":null,"dir":"Reference","previous_headings":"","what":"A preprocessing layer to convert raw audio signals to Mel spectrograms. — layer_mel_spectrogram","title":"A preprocessing layer to convert raw audio signals to Mel spectrograms. — layer_mel_spectrogram","text":"layer takes float32/float64 single batched audio signal inputs computes Mel spectrogram using Short-Time Fourier Transform Mel scaling. input 1D (unbatched) 2D (batched) tensor representing audio signals. output 2D 3D tensor representing Mel spectrograms. spectrogram image-like representation shows frequency spectrum signal time. uses x-axis represent time, y-axis represent frequency, pixel represent intensity. Mel spectrograms special type spectrogram use mel scale, approximates humans perceive sound. commonly used speech music processing tasks like speech recognition, speaker identification, music genre classification.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_mel_spectrogram.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"A preprocessing layer to convert raw audio signals to Mel spectrograms. — layer_mel_spectrogram","text":"","code":"layer_mel_spectrogram(   object,   fft_length = 2048L,   sequence_stride = 512L,   sequence_length = NULL,   window = \"hann\",   sampling_rate = 16000L,   num_mel_bins = 128L,   min_freq = 20,   max_freq = NULL,   power_to_db = TRUE,   top_db = 80,   mag_exp = 2,   min_power = 1e-10,   ref_power = 1,   ... )"},{"path":"https://keras3.posit.co/dev/reference/layer_mel_spectrogram.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"A preprocessing layer to convert raw audio signals to Mel spectrograms. — layer_mel_spectrogram","text":"object Object compose layer . tensor, array, sequential model. fft_length Integer, size FFT window. sequence_stride Integer, number samples successive STFT columns. sequence_length Integer, size window used applying window audio frame. NULL, defaults fft_length. window String, name window function use. Available values \"hann\" \"hamming\". window tensor, used directly window length must sequence_length. window NULL, windowing used. Defaults \"hann\". sampling_rate Integer, sample rate input signal. num_mel_bins Integer, number mel bins generate. min_freq Float, minimum frequency mel bins. max_freq Float, maximum frequency mel bins. NULL, defaults sampling_rate / 2. power_to_db TRUE, convert power spectrogram decibels. top_db Float, minimum negative cut-max(10 * log10(S)) - top_db. mag_exp Float, exponent magnitude spectrogram. 1 magnitude, 2 power, etc. Default 2. min_power Float, minimum value power ref_power. ref_power Float, power scaled relative 10 * log10(S / ref_power). ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_mel_spectrogram.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"A preprocessing layer to convert raw audio signals to Mel spectrograms. — layer_mel_spectrogram","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_mel_spectrogram.html","id":"references","dir":"Reference","previous_headings":"","what":"References","title":"A preprocessing layer to convert raw audio signals to Mel spectrograms. — layer_mel_spectrogram","text":"Spectrogram, Mel scale.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_mel_spectrogram.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"A preprocessing layer to convert raw audio signals to Mel spectrograms. — layer_mel_spectrogram","text":"Unbatched audio signal   Batched audio signal","code":"layer <- layer_mel_spectrogram(   num_mel_bins = 64,   sampling_rate = 8000,   sequence_stride = 256,   fft_length = 2048 ) layer(random_uniform(shape = c(16000))) |> shape() layer <- layer_mel_spectrogram(   num_mel_bins = 80,   sampling_rate = 8000,   sequence_stride = 128,   fft_length = 2048 ) layer(random_uniform(shape = c(2, 16000))) |> shape()"},{"path":"https://keras3.posit.co/dev/reference/layer_mel_spectrogram.html","id":"input-shape","dir":"Reference","previous_headings":"","what":"Input Shape","title":"A preprocessing layer to convert raw audio signals to Mel spectrograms. — layer_mel_spectrogram","text":"1D (unbatched) 2D (batched) tensor shape:(..., samples).","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_mel_spectrogram.html","id":"output-shape","dir":"Reference","previous_headings":"","what":"Output Shape","title":"A preprocessing layer to convert raw audio signals to Mel spectrograms. — layer_mel_spectrogram","text":"2D (unbatched) 3D (batched) tensor shape:(..., num_mel_bins, time).","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_minimum.html","id":null,"dir":"Reference","previous_headings":"","what":"Computes elementwise minimum on a list of inputs. — layer_minimum","title":"Computes elementwise minimum on a list of inputs. — layer_minimum","text":"takes input list tensors, shape, returns single tensor (also shape).","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_minimum.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes elementwise minimum on a list of inputs. — layer_minimum","text":"","code":"layer_minimum(inputs, ...)"},{"path":"https://keras3.posit.co/dev/reference/layer_minimum.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Computes elementwise minimum on a list of inputs. — layer_minimum","text":"inputs layers combine ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_minimum.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Computes elementwise minimum on a list of inputs. — layer_minimum","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_minimum.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Computes elementwise minimum on a list of inputs. — layer_minimum","text":"Usage Keras model:","code":"input_shape <- c(2, 3, 4) x1 <- random_uniform(input_shape) x2 <- random_uniform(input_shape) y <- layer_minimum(x1, x2) input1 <- layer_input(shape = c(16)) x1 <- input1 |> layer_dense(8, activation = 'relu') input2 <- layer_input(shape = c(32)) x2 <- input2 |> layer_dense(8, activation = 'relu') # equivalent to `y <- layer_minimum(x1, x2)` y <- layer_minimum(x1, x2) out <- y |> layer_dense(4) model <- keras_model(inputs = c(input1, input2), outputs = out)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_mix_up.html","id":null,"dir":"Reference","previous_headings":"","what":"MixUp implements the MixUp data augmentation technique. — layer_mix_up","title":"MixUp implements the MixUp data augmentation technique. — layer_mix_up","text":"MixUp implements MixUp data augmentation technique.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_mix_up.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"MixUp implements the MixUp data augmentation technique. — layer_mix_up","text":"","code":"layer_mix_up(object, alpha = 0.2, data_format = NULL, seed = NULL, ...)"},{"path":"https://keras3.posit.co/dev/reference/layer_mix_up.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"MixUp implements the MixUp data augmentation technique. — layer_mix_up","text":"object Object compose layer . tensor, array, sequential model. alpha Float 0 1. Controls blending strength. Smaller values mean less mixing, larger values allow  blending images. Defaults 0.2, recommended ImageNet1k classification. data_format String, one \"channels_last\" (default) \"channels_first\". ordering dimensions inputs. \"channels_last\" corresponds inputs shape (batch, height, width, channels) \"channels_first\" corresponds inputs shape (batch, channels, height, width). seed Integer. Used create random seed. ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_mix_up.html","id":"references","dir":"Reference","previous_headings":"","what":"References","title":"MixUp implements the MixUp data augmentation technique. — layer_mix_up","text":"MixUp paper. MixUp Object Detection paper.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_mix_up.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"MixUp implements the MixUp data augmentation technique. — layer_mix_up","text":"","code":"c(c(images, labels), .) %<-% dataset_cifar10() c(images, labels) %<-% list(images[1:8,,,], labels[1:8,]) labels <- labels |> op_one_hot(10) |> op_cast(\"float32\") mix_up <- layer_mix_up(alpha=0.2) output <- mix_up(list(images = images, labels = labels))"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_multi_head_attention.html","id":null,"dir":"Reference","previous_headings":"","what":"Multi Head Attention layer. — layer_multi_head_attention","title":"Multi Head Attention layer. — layer_multi_head_attention","text":"implementation multi-headed attention described paper \"Attention Need\" Vaswani et al., 2017. query, key, value , self-attention. timestep query attends corresponding sequence key, returns fixed-width vector. layer first projects query, key value. (effectively) list tensors length num_attention_heads, corresponding shapes (batch_size, <query dimensions>, key_dim), (batch_size, <key/value dimensions>, key_dim), (batch_size, <key/value dimensions>, value_dim). , query key tensors dot-producted scaled. softmaxed obtain attention probabilities. value tensors interpolated probabilities, concatenated back single tensor. Finally, result tensor last dimension value_dim can take linear projection return.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_multi_head_attention.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Multi Head Attention layer. — layer_multi_head_attention","text":"","code":"layer_multi_head_attention(   inputs,   num_heads,   key_dim,   value_dim = NULL,   dropout = 0,   use_bias = TRUE,   output_shape = NULL,   attention_axes = NULL,   flash_attention = NULL,   kernel_initializer = \"glorot_uniform\",   bias_initializer = \"zeros\",   kernel_regularizer = NULL,   bias_regularizer = NULL,   activity_regularizer = NULL,   kernel_constraint = NULL,   bias_constraint = NULL,   seed = NULL,   ... )"},{"path":"https://keras3.posit.co/dev/reference/layer_multi_head_attention.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Multi Head Attention layer. — layer_multi_head_attention","text":"inputs see description num_heads Number attention heads. key_dim Size attention head query key. value_dim Size attention head value. dropout Dropout probability. use_bias Boolean, whether dense layers use bias vectors/matrices. output_shape expected shape output tensor, besides batch sequence dims. specified, projects back query feature dim (query input's last dimension). attention_axes axes attention applied. NULL means attention axes, batch, heads, features. flash_attention NULL, layer attempts use flash attention faster memory-efficient attention computations possible. behavior can configured using config_enable_flash_attention() config_disable_flash_attention(). kernel_initializer Initializer dense layer kernels. bias_initializer Initializer dense layer biases. kernel_regularizer Regularizer dense layer kernels. bias_regularizer Regularizer dense layer biases. activity_regularizer Regularizer dense layer activity. kernel_constraint Constraint dense layer kernels. bias_constraint Constraint dense layer kernels. seed Optional integer seed dropout layer. ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_multi_head_attention.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Multi Head Attention layer. — layer_multi_head_attention","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_multi_head_attention.html","id":"call-arguments","dir":"Reference","previous_headings":"","what":"Call Arguments","title":"Multi Head Attention layer. — layer_multi_head_attention","text":"query: Query tensor shape (B, T, dim), B batch size, T target sequence length, dim feature dimension. value: Value tensor shape (B, S, dim), B batch size, S source sequence length, dim feature dimension. key: Optional key tensor shape (B, S, dim). given, use value key value, common case. attention_mask: boolean mask shape (B, T, S), prevents attention certain positions. boolean mask specifies query elements can attend key elements, 1 indicates attention 0 indicates attention. Broadcasting can happen missing batch dimensions head dimension. return_attention_scores: boolean indicate whether output (attention_output, attention_scores) TRUE, attention_output FALSE. Defaults FALSE. training: Python boolean indicating whether layer behave training mode (adding dropout) inference mode (dropout). go either using training mode parent layer/model, FALSE (inference) parent layer. use_causal_mask: boolean indicate whether apply causal mask prevent tokens attending future tokens (e.g., used decoder Transformer).","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_multi_head_attention.html","id":"call-return","dir":"Reference","previous_headings":"","what":"Call return","title":"Multi Head Attention layer. — layer_multi_head_attention","text":"attention_output: result computation, shape (B, T, E), T target sequence shapes E query input last dimension output_shape NULL. Otherwise, multi-head outputs projected shape specified output_shape. attention_scores: (Optional) multi-head attention coefficients attention axes.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_multi_head_attention.html","id":"properties","dir":"Reference","previous_headings":"","what":"Properties","title":"Multi Head Attention layer. — layer_multi_head_attention","text":"MultiHeadAttention Layer instance following additional read-properties: attention_axes dropout key_dense key_dim num_heads output_dense output_shape query_dense use_bias value_dense value_dim","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_multiply.html","id":null,"dir":"Reference","previous_headings":"","what":"Performs elementwise multiplication. — layer_multiply","title":"Performs elementwise multiplication. — layer_multiply","text":"takes input list tensors, shape, returns single tensor (also shape).","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_multiply.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Performs elementwise multiplication. — layer_multiply","text":"","code":"layer_multiply(inputs, ...)"},{"path":"https://keras3.posit.co/dev/reference/layer_multiply.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Performs elementwise multiplication. — layer_multiply","text":"inputs layers combine ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_multiply.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Performs elementwise multiplication. — layer_multiply","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_multiply.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Performs elementwise multiplication. — layer_multiply","text":"Usage Keras model:","code":"input_shape <- c(2, 3, 4) x1 <- random_uniform(input_shape) x2 <- random_uniform(input_shape) y <- layer_multiply(x1, x2) input1 <- layer_input(shape = c(16)) x1 <- input1 |> layer_dense(8, activation = 'relu') input2 <- layer_input(shape = c(32)) x2 <- input2 |> layer_dense(8, activation = 'relu') # equivalent to `y <- layer_multiply(x1, x2)` y <- layer_multiply(x1, x2) out <- y |> layer_dense(4) model <- keras_model(inputs = c(input1, input2), outputs = out)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_normalization.html","id":null,"dir":"Reference","previous_headings":"","what":"A preprocessing layer that normalizes continuous features. — layer_normalization","title":"A preprocessing layer that normalizes continuous features. — layer_normalization","text":"layer shift scale inputs distribution centered around 0 standard deviation 1. accomplishes precomputing mean variance data, calling (input - mean) / sqrt(var) runtime. mean variance values layer must either supplied construction learned via adapt(). adapt() compute mean variance data store layer's weights. adapt() called fit(), evaluate(), predict().","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_normalization.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"A preprocessing layer that normalizes continuous features. — layer_normalization","text":"","code":"layer_normalization(   object,   axis = -1L,   mean = NULL,   variance = NULL,   invert = FALSE,   ... )"},{"path":"https://keras3.posit.co/dev/reference/layer_normalization.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"A preprocessing layer that normalizes continuous features. — layer_normalization","text":"object Object compose layer . tensor, array, sequential model. axis Integer, list integers, NULL. axis axes separate mean variance index shape. example, shape (NULL, 5) axis=1, layer track 5 separate mean variance values last axis. axis set NULL, layer normalize elements input scalar mean variance. -1, last axis input assumed feature dimension normalized per index. Note specific case batched scalar inputs axis batch axis, default normalize index batch separately. case, consider passing axis=NULL. Defaults -1. mean mean value(s) use normalization. passed value(s) broadcast shape kept axes ; value(s) broadcast, error raised layer's build() method called. variance variance value(s) use normalization. passed value(s) broadcast shape kept axes ; value(s) broadcast, error raised layer's build() method called. invert TRUE, layer apply inverse transformation inputs: turn normalized input back original form. ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_normalization.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"A preprocessing layer that normalizes continuous features. — layer_normalization","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_normalization.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"A preprocessing layer that normalizes continuous features. — layer_normalization","text":"Calculate global mean variance analyzing dataset adapt().     Calculate mean variance index last axis.     Pass mean variance directly.     Use layer de-normalize inputs (adapting layer).","code":"adapt_data <- op_array(c(1., 2., 3., 4., 5.), dtype='float32') input_data <- op_array(c(1., 2., 3.), dtype='float32') layer <- layer_normalization(axis = NULL) layer %>% adapt(adapt_data) layer(input_data) ## tf.Tensor([-1.4142135  -0.70710677  0.        ], shape=(3), dtype=float32) adapt_data <- op_array(rbind(c(0., 7., 4.),                        c(2., 9., 6.),                        c(0., 7., 4.),                        c(2., 9., 6.)), dtype='float32') input_data <- op_array(matrix(c(0., 7., 4.), nrow = 1), dtype='float32') layer <- layer_normalization(axis=-1) layer %>% adapt(adapt_data) layer(input_data) ## tf.Tensor([[-1. -1. -1.]], shape=(1, 3), dtype=float32) input_data <- op_array(rbind(1, 2, 3), dtype='float32') layer <- layer_normalization(mean=3., variance=2.) layer(input_data) ## tf.Tensor( ## [[-1.4142135 ] ##  [-0.70710677] ##  [ 0.        ]], shape=(3, 1), dtype=float32) adapt_data <- op_array(rbind(c(0., 7., 4.),                        c(2., 9., 6.),                        c(0., 7., 4.),                        c(2., 9., 6.)), dtype='float32') input_data <- op_array(c(1., 2., 3.), dtype='float32') layer <- layer_normalization(axis=-1, invert=TRUE) layer %>% adapt(adapt_data) layer(input_data) ## tf.Tensor([[ 2. 10.  8.]], shape=(1, 3), dtype=float32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_permute.html","id":null,"dir":"Reference","previous_headings":"","what":"Permutes the dimensions of the input according to a given pattern. — layer_permute","title":"Permutes the dimensions of the input according to a given pattern. — layer_permute","text":"Useful e.g. connecting RNNs convnets.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_permute.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Permutes the dimensions of the input according to a given pattern. — layer_permute","text":"","code":"layer_permute(object, dims, ...)"},{"path":"https://keras3.posit.co/dev/reference/layer_permute.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Permutes the dimensions of the input according to a given pattern. — layer_permute","text":"object Object compose layer . tensor, array, sequential model. dims List integers. Permutation pattern include batch dimension. Indexing starts 1. instance, (1, 3, 2) permutes second third dimensions input. ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_permute.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Permutes the dimensions of the input according to a given pattern. — layer_permute","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_permute.html","id":"input-shape","dir":"Reference","previous_headings":"","what":"Input Shape","title":"Permutes the dimensions of the input according to a given pattern. — layer_permute","text":"Arbitrary.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_permute.html","id":"output-shape","dir":"Reference","previous_headings":"","what":"Output Shape","title":"Permutes the dimensions of the input according to a given pattern. — layer_permute","text":"input shape, dimensions re-ordered according specified pattern.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_permute.html","id":"example","dir":"Reference","previous_headings":"","what":"Example","title":"Permutes the dimensions of the input according to a given pattern. — layer_permute","text":"","code":"x <- layer_input(shape=c(10, 64)) y <- layer_permute(x, c(2, 1)) shape(y) ## shape(NA, 64, 10)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_pipeline.html","id":null,"dir":"Reference","previous_headings":"","what":"Applies a series of layers to an input. — layer_pipeline","title":"Applies a series of layers to an input. — layer_pipeline","text":"class useful build preprocessing pipeline, particular image data augmentation pipeline. Compared Sequential model, Pipeline features important differences: Model, just plain layer. layers pipeline compatible tf.data, pipeline also remain tf.data compatible. say, pipeline attempt convert inputs backend-native tensors tf.data context (unlike Sequential model).","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_pipeline.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Applies a series of layers to an input. — layer_pipeline","text":"","code":"layer_pipeline(layers, name = NULL)"},{"path":"https://keras3.posit.co/dev/reference/layer_pipeline.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Applies a series of layers to an input. — layer_pipeline","text":"layers list layers. name String, name object","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_pipeline.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Applies a series of layers to an input. — layer_pipeline","text":"","code":"preprocessing_pipeline <- layer_pipeline(c(   layer_auto_contrast(, ),   layer_random_zoom(, 0.2),   layer_random_rotation(, 0.2) ))  # `ds` is a tf.data.Dataset of images ds <- tfdatasets::tensor_slices_dataset(1:100) |>   tfdatasets::dataset_map(\\(.x) {     random_normal(c(28, 28))   }) |>   tfdatasets::dataset_batch(32)   #|>   # tfdatasets::dataset_take(4) |>   # iterate() |> str()  preprocessed_ds <- ds |>   tfdatasets::dataset_map(preprocessing_pipeline, num_parallel_calls = 4)"},{"path":"https://keras3.posit.co/dev/reference/layer_rand_augment.html","id":null,"dir":"Reference","previous_headings":"","what":"RandAugment performs the Rand Augment operation on input images. — layer_rand_augment","title":"RandAugment performs the Rand Augment operation on input images. — layer_rand_augment","text":"layer can thought --one image augmentation layer. policy implemented layer benchmarked extensively effective wide variety datasets.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_rand_augment.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"RandAugment performs the Rand Augment operation on input images. — layer_rand_augment","text":"","code":"layer_rand_augment(   object,   value_range = list(0L, 255L),   num_ops = 2L,   factor = 0.5,   interpolation = \"bilinear\",   seed = NULL,   data_format = NULL,   ... )"},{"path":"https://keras3.posit.co/dev/reference/layer_rand_augment.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"RandAugment performs the Rand Augment operation on input images. — layer_rand_augment","text":"object Object compose layer . tensor, array, sequential model. value_range range values input image can take. Default (0, 255). Typically, (0, 1) normalized images (0, 255) raw images. num_ops number augmentation operations apply sequentially image. Default 2. factor strength augmentation normalized value 0 1. Default 0.5. interpolation interpolation method use resizing operations. Options include nearest, bilinear. Default bilinear. seed Integer. Used create random seed. data_format String, one \"channels_last\" (default) \"channels_first\". ordering dimensions inputs. \"channels_last\" corresponds inputs shape (batch, height, width, channels) \"channels_first\" corresponds inputs shape (batch, channels, height, width). ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_rand_augment.html","id":"references","dir":"Reference","previous_headings":"","what":"References","title":"RandAugment performs the Rand Augment operation on input images. — layer_rand_augment","text":"RandAugment","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_random_brightness.html","id":null,"dir":"Reference","previous_headings":"","what":"A preprocessing layer which randomly adjusts brightness during training. — layer_random_brightness","title":"A preprocessing layer which randomly adjusts brightness during training. — layer_random_brightness","text":"layer randomly increase/reduce brightness input RGB images. inference time, output identical input. Call layer training=TRUE adjust brightness input. Note: layer safe use inside tf.data pipeline (independently backend using).","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_random_brightness.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"A preprocessing layer which randomly adjusts brightness during training. — layer_random_brightness","text":"","code":"layer_random_brightness(   object,   factor,   value_range = list(0L, 255L),   seed = NULL,   ... )"},{"path":"https://keras3.posit.co/dev/reference/layer_random_brightness.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"A preprocessing layer which randomly adjusts brightness during training. — layer_random_brightness","text":"object Object compose layer . tensor, array, sequential model. factor Float list 2 floats -1.0 1.0. factor used determine lower bound upper bound brightness adjustment. float value chosen randomly limits. -1.0 chosen, output image black, 1.0 chosen, image fully white. one float provided, eg, 0.2, -0.2 used lower bound 0.2 used upper bound. value_range Optional list 2 floats lower upper limit values input data. make change, use c(0.0, 1.0), e.g., image input scaled layer. Defaults c(0.0, 255.0). brightness adjustment scaled range, output values clipped range. seed optional integer, fixed RNG behavior. ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_random_brightness.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"A preprocessing layer which randomly adjusts brightness during training. — layer_random_brightness","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_random_brightness.html","id":"inputs","dir":"Reference","previous_headings":"","what":"Inputs","title":"A preprocessing layer which randomly adjusts brightness during training. — layer_random_brightness","text":"3D (HWC) 4D (NHWC) tensor, float int dtype. Input pixel values can range (e.g. [0., 1.) [0, 255])","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_random_brightness.html","id":"output","dir":"Reference","previous_headings":"","what":"Output","title":"A preprocessing layer which randomly adjusts brightness during training. — layer_random_brightness","text":"3D (HWC) 4D (NHWC) tensor brightness adjusted based factor. default, layer output floats. output value clipped range [0, 255], valid range RGB colors, rescaled based value_range needed.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_random_brightness.html","id":"example","dir":"Reference","previous_headings":"","what":"Example","title":"A preprocessing layer which randomly adjusts brightness during training. — layer_random_brightness","text":"","code":"random_bright <- layer_random_brightness(factor=0.2, seed = 1)  # An image with shape [2, 2, 3] image <- array(1:12, dim=c(2, 2, 3))  # Assume we randomly select the factor to be 0.1, then it will apply # 0.1 * 255 to all the channel output <- random_bright(image, training=TRUE) output ## tf.Tensor( ## [[[39 43 47] ##   [41 45 49]] ## ##  [[40 44 48] ##   [42 46 50]]], shape=(2, 2, 3), dtype=int32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_random_color_degeneration.html","id":null,"dir":"Reference","previous_headings":"","what":"Randomly performs the color degeneration operation on given images. — layer_random_color_degeneration","title":"Randomly performs the color degeneration operation on given images. — layer_random_color_degeneration","text":"sharpness operation first converts image gray scale, back color. takes weighted average original image degenerated image. makes colors appear dull.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_random_color_degeneration.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Randomly performs the color degeneration operation on given images. — layer_random_color_degeneration","text":"","code":"layer_random_color_degeneration(   object,   factor,   value_range = list(0L, 255L),   data_format = NULL,   seed = NULL,   ... )"},{"path":"https://keras3.posit.co/dev/reference/layer_random_color_degeneration.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Randomly performs the color degeneration operation on given images. — layer_random_color_degeneration","text":"object Object compose layer . tensor, array, sequential model. factor tuple two floats single float. factor controls extent image sharpness impacted. factor=0.0 makes layer perform -op operation, value 1.0 uses degenerated result entirely. Values 0 1 result linear interpolation original image sharpened image. Values 0.0 1.0. tuple used, factor sampled two values every image augmented. single float used, value 0.0 passed float sampled. order ensure value always , please pass tuple two identical floats: (0.5, 0.5). value_range range values input image can take. Default (0, 255). Typically, (0, 1) normalized images (0, 255) raw images. data_format String, one \"channels_last\" (default) \"channels_first\". ordering dimensions inputs. \"channels_last\" corresponds inputs shape (batch, height, width, channels) \"channels_first\" corresponds inputs shape (batch, channels, height, width). seed Integer. Used create random seed. ... forward/backward compatability.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_random_color_jitter.html","id":null,"dir":"Reference","previous_headings":"","what":"Randomly apply brightness, contrast, saturation — layer_random_color_jitter","title":"Randomly apply brightness, contrast, saturation — layer_random_color_jitter","text":"hue image processing operation sequentially randomly input.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_random_color_jitter.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Randomly apply brightness, contrast, saturation — layer_random_color_jitter","text":"","code":"layer_random_color_jitter(   object,   value_range = list(0L, 255L),   brightness_factor = NULL,   contrast_factor = NULL,   saturation_factor = NULL,   hue_factor = NULL,   seed = NULL,   data_format = NULL,   ... )"},{"path":"https://keras3.posit.co/dev/reference/layer_random_color_jitter.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Randomly apply brightness, contrast, saturation — layer_random_color_jitter","text":"object Object compose layer . tensor, array, sequential model. value_range range values incoming images . Represented two number tuple written [low, high]. typically either [0, 1] [0, 255] depending preprocessing pipeline set . brightness_factor Float list/tuple 2 floats -1.0 1.0. factor used determine lower bound upper bound brightness adjustment. float value chosen randomly limits. -1.0 chosen, output image black, 1.0 chosen, image fully white. one float provided, eg, 0.2, -0.2 used lower bound 0.2 used upper bound. contrast_factor positive float represented fraction value, tuple size 2 representing lower upper bound. represented single float, lower = upper. contrast factor randomly picked [1.0 - lower, 1.0 + upper]. pixel x channel, output (x - mean) * factor + mean mean mean value channel. saturation_factor tuple two floats single float. factor controls extent image saturation impacted. factor=0.5 makes layer perform -op operation. factor=0.0 makes image fully grayscale. factor=1.0 makes image fully saturated. Values 0.0 1.0. tuple used, factor sampled two values every image augmented. single float used, value 0.0 passed float sampled. ensure value always , pass tuple two identical floats: (0.5, 0.5). hue_factor single float tuple two floats. factor controls extent image hue impacted. factor=0.0 makes layer perform -op operation, value 1.0 performs aggressive contrast adjustment available. tuple used, factor sampled two values every image augmented. single float used, value 0.0 passed float sampled. order ensure value always , please pass tuple two identical floats: (0.5, 0.5). seed Integer. Used create random seed. data_format String, one \"channels_last\" (default) \"channels_first\". ordering dimensions inputs. \"channels_last\" corresponds inputs shape (batch, height, width, channels) \"channels_first\" corresponds inputs shape (batch, channels, height, width). ... forward/backward compatability.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_random_contrast.html","id":null,"dir":"Reference","previous_headings":"","what":"A preprocessing layer which randomly adjusts contrast during training. — layer_random_contrast","title":"A preprocessing layer which randomly adjusts contrast during training. — layer_random_contrast","text":"layer randomly adjust contrast image images random factor. Contrast adjusted independently channel image training. channel, layer computes mean image pixels channel adjusts component x pixel (x - mean) * contrast_factor + mean. Input pixel values can range (e.g. [0., 1.) [0, 255]) integer floating point dtype. default, layer output floats. Note: layer safe use inside tf.data pipeline (independently backend using).","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_random_contrast.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"A preprocessing layer which randomly adjusts contrast during training. — layer_random_contrast","text":"","code":"layer_random_contrast(   object,   factor,   value_range = c(0L, 255L),   seed = NULL,   ... )"},{"path":"https://keras3.posit.co/dev/reference/layer_random_contrast.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"A preprocessing layer which randomly adjusts contrast during training. — layer_random_contrast","text":"object Object compose layer . tensor, array, sequential model. factor positive float represented fraction value, tuple size 2 representing lower upper bound. represented single float, lower = upper. contrast factor randomly picked [1.0 - lower, 1.0 + upper]. pixel x channel, output (x - mean) * factor + mean mean mean value channel. value_range range values incoming images . Represented two-number tuple written tuple(low, high). typically either [0, 1] [0, 255] depending preprocessing pipeline set . seed Integer. Used create random seed. ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_random_contrast.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"A preprocessing layer which randomly adjusts contrast during training. — layer_random_contrast","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_random_contrast.html","id":"input-shape","dir":"Reference","previous_headings":"","what":"Input Shape","title":"A preprocessing layer which randomly adjusts contrast during training. — layer_random_contrast","text":"3D (unbatched) 4D (batched) tensor shape: (..., height, width, channels), \"channels_last\" format.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_random_contrast.html","id":"output-shape","dir":"Reference","previous_headings":"","what":"Output Shape","title":"A preprocessing layer which randomly adjusts contrast during training. — layer_random_contrast","text":"3D (unbatched) 4D (batched) tensor shape: (..., height, width, channels), \"channels_last\" format.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_random_crop.html","id":null,"dir":"Reference","previous_headings":"","what":"A preprocessing layer which randomly crops images during training. — layer_random_crop","title":"A preprocessing layer which randomly crops images during training. — layer_random_crop","text":"training, layer randomly choose location crop images target size. layer crop images batch cropping location. inference time, training input image smaller target size, input resized cropped return largest possible window image matches target aspect ratio. need apply random cropping inference time, set training TRUE calling layer. Input pixel values can range (e.g. [0., 1.) [0, 255]) integer floating point dtype. default, layer output floats. Note: layer safe use inside tf.data pipeline (independently backend using).","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_random_crop.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"A preprocessing layer which randomly crops images during training. — layer_random_crop","text":"","code":"layer_random_crop(   object,   height,   width,   seed = NULL,   data_format = NULL,   name = NULL,   ... )"},{"path":"https://keras3.posit.co/dev/reference/layer_random_crop.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"A preprocessing layer which randomly crops images during training. — layer_random_crop","text":"object Object compose layer . tensor, array, sequential model. height Integer, height output shape. width Integer, width output shape. seed Integer. Used create random seed. data_format string, either \"channels_last\" \"channels_first\". ordering dimensions inputs. \"channels_last\" corresponds inputs shape (batch, height, width, channels) \"channels_first\" corresponds inputs shape (batch, channels, height, width). defaults image_data_format value found Keras config file ~/.keras/keras.json. never set , \"channels_last\". name String, name object ... Base layer keyword arguments, name dtype.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_random_crop.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"A preprocessing layer which randomly crops images during training. — layer_random_crop","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_random_crop.html","id":"input-shape","dir":"Reference","previous_headings":"","what":"Input Shape","title":"A preprocessing layer which randomly crops images during training. — layer_random_crop","text":"3D (unbatched) 4D (batched) tensor shape: (..., height, width, channels), \"channels_last\" format.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_random_crop.html","id":"output-shape","dir":"Reference","previous_headings":"","what":"Output Shape","title":"A preprocessing layer which randomly crops images during training. — layer_random_crop","text":"3D (unbatched) 4D (batched) tensor shape: (..., target_height, target_width, channels).","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_random_elastic_transform.html","id":null,"dir":"Reference","previous_headings":"","what":"A preprocessing layer that applies random elastic transformations. — layer_random_elastic_transform","title":"A preprocessing layer that applies random elastic transformations. — layer_random_elastic_transform","text":"layer distorts input images applying elastic deformations, simulating physically realistic transformation. magnitude distortion controlled scale parameter, factor determines probability applying transformation.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_random_elastic_transform.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"A preprocessing layer that applies random elastic transformations. — layer_random_elastic_transform","text":"","code":"layer_random_elastic_transform(   object,   factor = 1,   scale = 1,   interpolation = \"bilinear\",   fill_mode = \"reflect\",   fill_value = 0,   value_range = list(0L, 255L),   seed = NULL,   data_format = NULL,   ... )"},{"path":"https://keras3.posit.co/dev/reference/layer_random_elastic_transform.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"A preprocessing layer that applies random elastic transformations. — layer_random_elastic_transform","text":"object Object compose layer . tensor, array, sequential model. factor single float tuple two floats. factor controls probability applying transformation. factor = 0.0 ensures transformation applied. factor = 1.0 means transformation always applied. tuple (min, max) provided, probability value sampled min max image. single float provided, probability sampled 0.0 given float. Default 1.0. scale float tuple two floats defining magnitude distortion applied. tuple (min, max) provided, random scale value sampled within range. single float provided, random scale value sampled 0.0 given float. Default 1.0. interpolation Interpolation mode. Supported values: \"nearest\", \"bilinear\". fill_mode Points outside boundaries input filled according given mode. Available methods \"constant\", \"nearest\", \"wrap\" \"reflect\". Defaults \"reflect\". \"reflect\": (d c b | b c d | d c b ) input extended reflecting edge last pixel. \"constant\": (k k k k | b c d | k k k k) input extended filling values beyond edge constant value k specified fill_value. \"wrap\": (b c d | b c d | b c d) input extended wrapping around opposite edge. \"nearest\": (| b c d | d d d d) input extended nearest pixel. using torch backend, \"reflect\" redirected \"mirror\" torch support \"reflect\". torch backend also support \"wrap\". fill_value float representing value fill outside boundaries fill_mode = \"constant\". value_range range values incoming images . Represented two-number tuple written [low, high]. typically either [0, 1] [0, 255] depending preprocessing pipeline set . seed Integer. Used create random seed. data_format string, either \"channels_last\" \"channels_first\". ordering dimensions inputs. \"channels_last\" corresponds inputs shape (batch, height, width, channels) \"channels_first\" corresponds inputs shape (batch, channels, height, width). defaults image_data_format value found Keras config file ~/.keras/keras.json. never set , \"channels_last\". ... forward/backward compatability.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_random_erasing.html","id":null,"dir":"Reference","previous_headings":"","what":"Random Erasing data augmentation technique. — layer_random_erasing","title":"Random Erasing data augmentation technique. — layer_random_erasing","text":"Random Erasing data augmentation method random patches image erased (replaced constant value noise) training improve generalization.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_random_erasing.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Random Erasing data augmentation technique. — layer_random_erasing","text":"","code":"layer_random_erasing(   object,   factor = 1,   scale = list(0.02, 0.33),   fill_value = NULL,   value_range = list(0L, 255L),   seed = NULL,   data_format = NULL,   ... )"},{"path":"https://keras3.posit.co/dev/reference/layer_random_erasing.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Random Erasing data augmentation technique. — layer_random_erasing","text":"object Object compose layer . tensor, array, sequential model. factor single float tuple two floats. factor controls probability applying transformation. factor=0.0 ensures erasing applied. factor=1.0 means erasing always applied. tuple (min, max) provided, probability value sampled min max image. single float provided, probability sampled 0.0 given float. Default 1.0. scale tuple two floats representing aspect ratio range erased patch. defines width--height ratio patch erased. can help control rw shape erased region. Default (0.02, 0.33). fill_value value fill erased region . can set constant value None sample random value normal distribution. Default None. value_range range values incoming images . Represented two-number tuple written [low, high]. typically either [0, 1] [0, 255] depending preprocessing pipeline set . seed Integer. Used create random seed. data_format string, either \"channels_last\" \"channels_first\". ordering dimensions inputs. \"channels_last\" corresponds inputs shape (batch, height, width, channels) \"channels_first\" corresponds inputs shape (batch, channels, height, width). defaults image_data_format value found Keras config file ~/.keras/keras.json. never set , \"channels_last\". ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_random_erasing.html","id":"references","dir":"Reference","previous_headings":"","what":"References","title":"Random Erasing data augmentation technique. — layer_random_erasing","text":"Random Erasing paper.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_random_flip.html","id":null,"dir":"Reference","previous_headings":"","what":"A preprocessing layer which randomly flips images during training. — layer_random_flip","title":"A preprocessing layer which randomly flips images during training. — layer_random_flip","text":"layer flip images horizontally vertically based mode attribute. inference time, output identical input. Call layer training=TRUE flip input. Input pixel values can range (e.g. [0., 1.) [0, 255]) integer floating point dtype. default, layer output floats. Note: layer safe use inside tf.data pipeline (independently backend using).","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_random_flip.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"A preprocessing layer which randomly flips images during training. — layer_random_flip","text":"","code":"layer_random_flip(   object,   mode = \"horizontal_and_vertical\",   seed = NULL,   data_format = NULL,   ... )"},{"path":"https://keras3.posit.co/dev/reference/layer_random_flip.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"A preprocessing layer which randomly flips images during training. — layer_random_flip","text":"object Object compose layer . tensor, array, sequential model. mode String indicating flip mode use. Can \"horizontal\", \"vertical\", \"horizontal_and_vertical\". \"horizontal\" left-right flip \"vertical\" top-bottom flip. Defaults \"horizontal_and_vertical\" seed Integer. Used create random seed. data_format string, either \"channels_last\" \"channels_first\". ordering dimensions inputs. \"channels_last\" corresponds inputs shape (batch, height, width, channels) \"channels_first\" corresponds inputs shape (batch, channels, height, width). defaults image_data_format value found Keras config file ~/.keras/keras.json. never set , \"channels_last\". ... Base layer keyword arguments, name dtype.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_random_flip.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"A preprocessing layer which randomly flips images during training. — layer_random_flip","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_random_flip.html","id":"input-shape","dir":"Reference","previous_headings":"","what":"Input Shape","title":"A preprocessing layer which randomly flips images during training. — layer_random_flip","text":"3D (unbatched) 4D (batched) tensor shape: (..., height, width, channels), \"channels_last\" format.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_random_flip.html","id":"output-shape","dir":"Reference","previous_headings":"","what":"Output Shape","title":"A preprocessing layer which randomly flips images during training. — layer_random_flip","text":"3D (unbatched) 4D (batched) tensor shape: (..., height, width, channels), \"channels_last\" format.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_random_gaussian_blur.html","id":null,"dir":"Reference","previous_headings":"","what":"Applies random Gaussian blur to images for data augmentation. — layer_random_gaussian_blur","title":"Applies random Gaussian blur to images for data augmentation. — layer_random_gaussian_blur","text":"layer performs Gaussian blur operation input images randomly selected degree blurring, controlled factor sigma arguments.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_random_gaussian_blur.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Applies random Gaussian blur to images for data augmentation. — layer_random_gaussian_blur","text":"","code":"layer_random_gaussian_blur(   object,   factor = 1,   kernel_size = 3L,   sigma = 1,   value_range = list(0L, 255L),   data_format = NULL,   seed = NULL,   ... )"},{"path":"https://keras3.posit.co/dev/reference/layer_random_gaussian_blur.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Applies random Gaussian blur to images for data augmentation. — layer_random_gaussian_blur","text":"object Object compose layer . tensor, array, sequential model. factor single float tuple two floats. factor controls extent image hue impacted. factor=0.0 makes layer perform -op operation, value 1.0 performs aggressive blurring available. tuple used, factor sampled two values every image augmented. single float used, value 0.0 passed float sampled. Default 1.0. kernel_size Integer. Size Gaussian kernel used blurring. Must odd integer. Default 3. sigma Float tuple two floats. Standard deviation Gaussian kernel. Controls intensity blur. tuple provided, value sampled two image. Default 1.0. value_range range values incoming images . Represented two-number tuple written [low, high]. typically either [0, 1] [0, 255] depending preprocessing pipeline set . data_format string, either \"channels_last\" \"channels_first\". ordering dimensions inputs. \"channels_last\" corresponds inputs shape (batch, height, width, channels) \"channels_first\" corresponds inputs shape (batch, channels, height, width). defaults image_data_format value found Keras config file ~/.keras/keras.json. never set , \"channels_last\". seed Integer. Used create random seed. ... forward/backward compatability.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_random_grayscale.html","id":null,"dir":"Reference","previous_headings":"","what":"Preprocessing layer for random conversion of RGB images to grayscale. — layer_random_grayscale","title":"Preprocessing layer for random conversion of RGB images to grayscale. — layer_random_grayscale","text":"layer randomly converts input images grayscale specified factor. applied, maintains original number channels sets channels grayscale value. can useful data augmentation training models robust color variations. conversion preserves perceived luminance original color image using standard RGB grayscale conversion coefficients. Images selected conversion remain unchanged. Note: layer safe use inside tf.data pipeline (independently backend using).","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_random_grayscale.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Preprocessing layer for random conversion of RGB images to grayscale. — layer_random_grayscale","text":"","code":"layer_random_grayscale(   object,   factor = 0.5,   data_format = NULL,   seed = NULL,   ... )"},{"path":"https://keras3.posit.co/dev/reference/layer_random_grayscale.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Preprocessing layer for random conversion of RGB images to grayscale. — layer_random_grayscale","text":"object Object compose layer . tensor, array, sequential model. factor Float 0 1, specifying factor converting image grayscale. Defaults 0.5. value 1.0 means images converted, 0.0 means images converted. data_format String, one \"channels_last\" (default) \"channels_first\". ordering dimensions inputs. \"channels_last\" corresponds inputs shape (batch, height, width, channels) \"channels_first\" corresponds inputs shape (batch, channels, height, width). seed Initial seed random number generator ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_random_grayscale.html","id":"input-shape","dir":"Reference","previous_headings":"","what":"Input Shape","title":"Preprocessing layer for random conversion of RGB images to grayscale. — layer_random_grayscale","text":"3D (unbatched) 4D (batched) tensor shape: (..., height, width, channels), \"channels_last\" format, (..., channels, height, width), \"channels_first\" format.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_random_grayscale.html","id":"output-shape","dir":"Reference","previous_headings":"","what":"Output Shape","title":"Preprocessing layer for random conversion of RGB images to grayscale. — layer_random_grayscale","text":"input shape. output maintains number channels input, even grayscale-converted images channels value.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_random_hue.html","id":null,"dir":"Reference","previous_headings":"","what":"Randomly adjusts the hue on given images. — layer_random_hue","title":"Randomly adjusts the hue on given images. — layer_random_hue","text":"layer randomly increase/reduce hue input RGB images. image hue adjusted converting image(s) HSV rotating hue channel (H) delta. image converted back RGB.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_random_hue.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Randomly adjusts the hue on given images. — layer_random_hue","text":"","code":"layer_random_hue(   object,   factor,   value_range = list(0L, 255L),   data_format = NULL,   seed = NULL,   ... )"},{"path":"https://keras3.posit.co/dev/reference/layer_random_hue.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Randomly adjusts the hue on given images. — layer_random_hue","text":"object Object compose layer . tensor, array, sequential model. factor single float tuple two floats. factor controls extent image hue impacted. factor=0.0 makes layer perform -op operation, value 1.0 performs aggressive contrast adjustment available. tuple used, factor sampled two values every image augmented. single float used, value 0.0 passed float sampled. order ensure value always , please pass tuple two identical floats: (0.5, 0.5). value_range range values incoming images . Represented two-number tuple written [low, high]. typically either [0, 1] [0, 255] depending preprocessing pipeline set . data_format String, one \"channels_last\" (default) \"channels_first\". ordering dimensions inputs. \"channels_last\" corresponds inputs shape (batch, height, width, channels) \"channels_first\" corresponds inputs shape (batch, channels, height, width). seed Integer. Used create random seed. ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_random_hue.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Randomly adjusts the hue on given images. — layer_random_hue","text":"","code":"c(c(images, labels), .) %<-% dataset_cifar10() random_hue <- layer_random_hue(factor=0.5, value_range=c(0, 1)) images <- op_cast(images[1:8,,,], \"float32\") augmented_images_batch = random_hue(images)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_random_invert.html","id":null,"dir":"Reference","previous_headings":"","what":"Preprocessing layer for random inversion of image colors. — layer_random_invert","title":"Preprocessing layer for random inversion of image colors. — layer_random_invert","text":"layer randomly inverts colors input images specified probability range. applied, image chance colors inverted, pixel values transformed complementary values. Images selected inversion remain unchanged.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_random_invert.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Preprocessing layer for random inversion of image colors. — layer_random_invert","text":"","code":"layer_random_invert(   object,   factor = 1,   value_range = list(0L, 255L),   seed = NULL,   data_format = NULL,   ... )"},{"path":"https://keras3.posit.co/dev/reference/layer_random_invert.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Preprocessing layer for random inversion of image colors. — layer_random_invert","text":"object Object compose layer . tensor, array, sequential model. factor single float tuple two floats. factor controls probability inverting image colors. tuple provided, value sampled two values image, factor[0] minimum factor[1] maximum probability. single float provided, value 0.0 provided float sampled. Defaults (0, 1). value_range tuple list two elements. first value represents lower bound values passed images, second represents upper bound. Images passed layer values within value_range. Defaults (0, 255). seed Integer. Used create random seed. data_format string, either \"channels_last\" \"channels_first\". ordering dimensions inputs. \"channels_last\" corresponds inputs shape (batch, height, width, channels) \"channels_first\" corresponds inputs shape (batch, channels, height, width). defaults image_data_format value found Keras config file ~/.keras/keras.json. never set , \"channels_last\". ... forward/backward compatability.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_random_perspective.html","id":null,"dir":"Reference","previous_headings":"","what":"A preprocessing layer that applies random perspective transformations. — layer_random_perspective","title":"A preprocessing layer that applies random perspective transformations. — layer_random_perspective","text":"layer distorts perspective input images shifting corner points, simulating 3D-like transformation. amount distortion controlled factor scale parameters.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_random_perspective.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"A preprocessing layer that applies random perspective transformations. — layer_random_perspective","text":"","code":"layer_random_perspective(   object,   factor = 1,   scale = 1,   interpolation = \"bilinear\",   fill_value = 0,   seed = NULL,   data_format = NULL,   ... )"},{"path":"https://keras3.posit.co/dev/reference/layer_random_perspective.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"A preprocessing layer that applies random perspective transformations. — layer_random_perspective","text":"object Object compose layer . tensor, array, sequential model. factor float tuple two floats. Represents probability applying perspective transformation image batch. factor=0.0 ensures transformation applied. factor=1.0 means transformation always applied. tuple (min, max) provided, probability randomly sampled min max image. single float given, probability sampled 0.0 provided float. Default 1.0. scale float defining relative amount perspective shift. Determines much image corners displaced, affecting intensity perspective effect. interpolation Interpolation mode. Supported values: \"nearest\", \"bilinear\". fill_value float represents value filled outside boundaries fill_mode=\"constant\". seed Integer. Used create random seed. data_format string, either \"channels_last\" \"channels_first\". ordering dimensions inputs. \"channels_last\" corresponds inputs shape (batch, height, width, channels) \"channels_first\" corresponds inputs shape (batch, channels, height, width). defaults image_data_format value found Keras config file ~/.keras/keras.json. never set , \"channels_last\". ... forward/backward compatability.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_random_posterization.html","id":null,"dir":"Reference","previous_headings":"","what":"Reduces the number of bits for each color channel. — layer_random_posterization","title":"Reduces the number of bits for each color channel. — layer_random_posterization","text":"Reduces number bits color channel.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_random_posterization.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Reduces the number of bits for each color channel. — layer_random_posterization","text":"","code":"layer_random_posterization(   object,   factor,   value_range = list(0L, 255L),   data_format = NULL,   seed = NULL,   ... )"},{"path":"https://keras3.posit.co/dev/reference/layer_random_posterization.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Reduces the number of bits for each color channel. — layer_random_posterization","text":"object Object compose layer . tensor, array, sequential model. factor integer, number bits keep channel. Must value 1-8. value_range tuple list two elements. first value represents lower bound values passed images, second represents upper bound. Images passed layer values within value_range. Defaults (0, 255). data_format String, one \"channels_last\" (default) \"channels_first\". ordering dimensions inputs. \"channels_last\" corresponds inputs shape (batch, height, width, channels) \"channels_first\" corresponds inputs shape (batch, channels, height, width). seed Initial seed random number generator ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_random_posterization.html","id":"references","dir":"Reference","previous_headings":"","what":"References","title":"Reduces the number of bits for each color channel. — layer_random_posterization","text":"AutoAugment: Learning Augmentation Policies Data RandAugment: Practical automated data augmentation reduced search space","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_random_rotation.html","id":null,"dir":"Reference","previous_headings":"","what":"A preprocessing layer which randomly rotates images during training. — layer_random_rotation","title":"A preprocessing layer which randomly rotates images during training. — layer_random_rotation","text":"layer apply random rotations image, filling empty space according fill_mode. default, random rotations applied training. inference time, layer nothing. need apply random rotations inference time, pass training = TRUE calling layer. Input pixel values can range (e.g. [0., 1.) [0, 255]) integer floating point dtype. default, layer output floats. Note: layer safe use inside tf.data pipeline (independently backend using).","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_random_rotation.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"A preprocessing layer which randomly rotates images during training. — layer_random_rotation","text":"","code":"layer_random_rotation(   object,   factor,   fill_mode = \"reflect\",   interpolation = \"bilinear\",   seed = NULL,   fill_value = 0,   data_format = NULL,   ... )"},{"path":"https://keras3.posit.co/dev/reference/layer_random_rotation.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"A preprocessing layer which randomly rotates images during training. — layer_random_rotation","text":"object Object compose layer . tensor, array, sequential model. factor float represented fraction 2 Pi, tuple size 2 representing lower upper bound rotating clockwise counter-clockwise. positive values means rotating counter clock-wise, negative value means clock-wise. represented single float, value used upper lower bound. instance, factor=c(-0.2, 0.3) results output rotation random amount range [-20% * 360, 30% * 360]. factor=0.2 results output rotating random amount range [-20% * 360, 20% * 360]. fill_mode Points outside boundaries input filled according given mode (one {\"constant\", \"reflect\", \"wrap\", \"nearest\"}). reflect: (d c b | b c d | d c b ) input extended reflecting edge last pixel. constant: (k k k k | b c d | k k k k) input extended filling values beyond edge constant value k = 0. wrap: (b c d | b c d | b c d) input extended wrapping around opposite edge. nearest: (| b c d | d d d d) input extended nearest pixel. interpolation Interpolation mode. Supported values: \"nearest\", \"bilinear\". seed Integer. Used create random seed. fill_value float represents value filled outside boundaries fill_mode=\"constant\". data_format string, either \"channels_last\" \"channels_first\". ordering dimensions inputs. \"channels_last\" corresponds inputs shape (batch, height, width, channels) \"channels_first\" corresponds inputs shape (batch, channels, height, width). defaults image_data_format value found Keras config file ~/.keras/keras.json. never set , \"channels_last\". ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_random_rotation.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"A preprocessing layer which randomly rotates images during training. — layer_random_rotation","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_random_rotation.html","id":"input-shape","dir":"Reference","previous_headings":"","what":"Input Shape","title":"A preprocessing layer which randomly rotates images during training. — layer_random_rotation","text":"3D (unbatched) 4D (batched) tensor shape: (..., height, width, channels), \"channels_last\" format","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_random_rotation.html","id":"output-shape","dir":"Reference","previous_headings":"","what":"Output Shape","title":"A preprocessing layer which randomly rotates images during training. — layer_random_rotation","text":"3D (unbatched) 4D (batched) tensor shape: (..., height, width, channels), \"channels_last\" format","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_random_saturation.html","id":null,"dir":"Reference","previous_headings":"","what":"Randomly adjusts the saturation on given images. — layer_random_saturation","title":"Randomly adjusts the saturation on given images. — layer_random_saturation","text":"layer randomly increase/reduce saturation input RGB images.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_random_saturation.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Randomly adjusts the saturation on given images. — layer_random_saturation","text":"","code":"layer_random_saturation(   object,   factor,   value_range = list(0L, 255L),   data_format = NULL,   seed = NULL,   ... )"},{"path":"https://keras3.posit.co/dev/reference/layer_random_saturation.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Randomly adjusts the saturation on given images. — layer_random_saturation","text":"object Object compose layer . tensor, array, sequential model. factor tuple two floats single float. factor controls extent image saturation impacted. factor=0.5 makes layer perform -op operation. factor=0.0 makes image fully grayscale. factor=1.0 makes image fully saturated. Values 0.0 1.0. tuple used, factor sampled two values every image augmented. single float used, value 0.0 passed float sampled. ensure value always , pass tuple two identical floats: (0.5, 0.5). value_range range values incoming images . Represented two-number tuple written [low, high]. typically either [0, 1] [0, 255] depending preprocessing pipeline set . data_format String, one \"channels_last\" (default) \"channels_first\". ordering dimensions inputs. \"channels_last\" corresponds inputs shape (batch, height, width, channels) \"channels_first\" corresponds inputs shape (batch, channels, height, width). seed Integer. Used create random seed. ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_random_saturation.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Randomly adjusts the saturation on given images. — layer_random_saturation","text":"","code":"c(c(images, labels), .) %<-% dataset_cifar10() images <- images[1:8, , , ] |> op_cast(\"float32\") random_saturation <- layer_random_saturation(factor = 0.2) augmented_images <- random_saturation(images)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_random_sharpness.html","id":null,"dir":"Reference","previous_headings":"","what":"Randomly performs the sharpness operation on given images. — layer_random_sharpness","title":"Randomly performs the sharpness operation on given images. — layer_random_sharpness","text":"sharpness operation first performs blur, blends original image processed image. operation adjusts clarity edges image, ranging blurred enhanced sharpness.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_random_sharpness.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Randomly performs the sharpness operation on given images. — layer_random_sharpness","text":"","code":"layer_random_sharpness(   object,   factor,   value_range = list(0L, 255L),   data_format = NULL,   seed = NULL,   ... )"},{"path":"https://keras3.posit.co/dev/reference/layer_random_sharpness.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Randomly performs the sharpness operation on given images. — layer_random_sharpness","text":"object Object compose layer . tensor, array, sequential model. factor tuple two floats single float. factor controls extent image sharpness impacted. factor=0.0 results fully blurred image, factor=0.5 applies operation (preserving original image), factor=1.0 enhances sharpness beyond original. Values 0.0 1.0. tuple used, factor sampled two values every image augmented. single float used, value 0.0 passed float sampled. ensure value always , pass tuple two identical floats: (0.5, 0.5). value_range range values incoming images . Represented two-number tuple written [low, high]. typically either [0, 1] [0, 255] depending preprocessing pipeline set . data_format String, one \"channels_last\" (default) \"channels_first\". ordering dimensions inputs. \"channels_last\" corresponds inputs shape (batch, height, width, channels) \"channels_first\" corresponds inputs shape (batch, channels, height, width). seed Integer. Used create random seed. ... forward/backward compatability.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_random_shear.html","id":null,"dir":"Reference","previous_headings":"","what":"A preprocessing layer that randomly applies shear transformations — layer_random_shear","title":"A preprocessing layer that randomly applies shear transformations — layer_random_shear","text":"images. layer shears input images along x-axis /y-axis randomly selected factor within specified range. shear transformation applied image independently batch. Empty regions created transformation filled according fill_mode fill_value parameters.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_random_shear.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"A preprocessing layer that randomly applies shear transformations — layer_random_shear","text":"","code":"layer_random_shear(   object,   x_factor = 0,   y_factor = 0,   interpolation = \"bilinear\",   fill_mode = \"reflect\",   fill_value = 0,   data_format = NULL,   seed = NULL,   ... )"},{"path":"https://keras3.posit.co/dev/reference/layer_random_shear.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"A preprocessing layer that randomly applies shear transformations — layer_random_shear","text":"object Object compose layer . tensor, array, sequential model. x_factor tuple two floats. augmented image, value sampled provided range. float passed, range interpreted (0, x_factor). Values represent percentage image shear . example, 0.3 shears pixels 30% way across image. provided values positive. y_factor tuple two floats. augmented image, value sampled provided range. float passed, range interpreted (0, y_factor). Values represent percentage image shear . example, 0.3 shears pixels 30% way across image. provided values positive. interpolation Interpolation mode. Supported values: \"nearest\", \"bilinear\". fill_mode Points outside boundaries input filled according given mode. Available methods \"constant\", \"nearest\", \"wrap\" \"reflect\". Defaults \"reflect\". \"reflect\": (d c b | b c d | d c b ) input extended reflecting edge last pixel. \"constant\": (k k k k | b c d | k k k k) input extended filling values beyond edge constant value k specified fill_value. \"wrap\": (b c d | b c d | b c d) input extended wrapping around opposite edge. \"nearest\": (| b c d | d d d d) input extended nearest pixel. Note using torch backend, \"reflect\" redirected \"mirror\" (c d c b | b c d | c b b) torch support \"reflect\". Note torch backend support \"wrap\". fill_value float representing value filled outside boundaries fill_mode=\"constant\". data_format String, one \"channels_last\" (default) \"channels_first\". ordering dimensions inputs. \"channels_last\" corresponds inputs shape (batch, height, width, channels) \"channels_first\" corresponds inputs shape (batch, channels, height, width). seed Integer. Used create random seed. ... forward/backward compatability.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_random_translation.html","id":null,"dir":"Reference","previous_headings":"","what":"A preprocessing layer which randomly translates images during training. — layer_random_translation","title":"A preprocessing layer which randomly translates images during training. — layer_random_translation","text":"layer apply random translations image training, filling empty space according fill_mode. Input pixel values can range (e.g. [0., 1.) [0, 255]) integer floating point dtype. default, layer output floats.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_random_translation.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"A preprocessing layer which randomly translates images during training. — layer_random_translation","text":"","code":"layer_random_translation(   object,   height_factor,   width_factor,   fill_mode = \"reflect\",   interpolation = \"bilinear\",   seed = NULL,   fill_value = 0,   data_format = NULL,   ... )"},{"path":"https://keras3.posit.co/dev/reference/layer_random_translation.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"A preprocessing layer which randomly translates images during training. — layer_random_translation","text":"object Object compose layer . tensor, array, sequential model. height_factor float represented fraction value, tuple size 2 representing lower upper bound shifting vertically. negative value means shifting image , positive value means shifting image . represented single positive float, value used upper lower bound. instance, height_factor=(-0.2, 0.3) results output shifted random amount range [-20%, +30%]. height_factor=0.2 results output height shifted random amount range [-20%, +20%]. width_factor float represented fraction value, tuple size 2 representing lower upper bound shifting horizontally. negative value means shifting image left, positive value means shifting image right. represented single positive float, value used upper lower bound. instance, width_factor=(-0.2, 0.3) results output shifted left 20%, shifted right 30%. width_factor=0.2 results output height shifted left right 20%. fill_mode Points outside boundaries input filled according given mode. Available methods \"constant\", \"nearest\", \"wrap\" \"reflect\". Defaults \"reflect\". \"reflect\": (d c b | b c d | d c b ) input extended reflecting edge last pixel. \"constant\": (k k k k | b c d | k k k k) input extended filling values beyond edge constant value k specified fill_value. \"wrap\": (b c d | b c d | b c d) input extended wrapping around opposite edge. \"nearest\": (| b c d | d d d d) input extended nearest pixel. Note using torch backend, \"reflect\" redirected \"mirror\" (c d c b | b c d | c b b) torch support \"reflect\". Note torch backend support \"wrap\". interpolation Interpolation mode. Supported values: \"nearest\", \"bilinear\". seed Integer. Used create random seed. fill_value float represents value filled outside boundaries fill_mode=\"constant\". data_format string, either \"channels_last\" \"channels_first\". ordering dimensions inputs. \"channels_last\" corresponds inputs shape (batch, height, width, channels) \"channels_first\" corresponds inputs shape (batch, channels, height, width). defaults image_data_format value found Keras config file ~/.keras/keras.json. never set , \"channels_last\". ... Base layer keyword arguments, name dtype.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_random_translation.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"A preprocessing layer which randomly translates images during training. — layer_random_translation","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_random_translation.html","id":"input-shape","dir":"Reference","previous_headings":"","what":"Input Shape","title":"A preprocessing layer which randomly translates images during training. — layer_random_translation","text":"3D (unbatched) 4D (batched) tensor shape: (..., height, width, channels), \"channels_last\" format, (..., channels, height, width), \"channels_first\" format.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_random_translation.html","id":"output-shape","dir":"Reference","previous_headings":"","what":"Output Shape","title":"A preprocessing layer which randomly translates images during training. — layer_random_translation","text":"3D (unbatched) 4D (batched) tensor shape: (..., target_height, target_width, channels), (..., channels, target_height, target_width), \"channels_first\" format. Note: layer safe use inside tf.data pipeline (independently backend using).","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_random_zoom.html","id":null,"dir":"Reference","previous_headings":"","what":"A preprocessing layer which randomly zooms images during training. — layer_random_zoom","title":"A preprocessing layer which randomly zooms images during training. — layer_random_zoom","text":"layer randomly zoom axis image independently, filling empty space according fill_mode. Input pixel values can range (e.g. [0., 1.) [0, 255]) integer floating point dtype. default, layer output floats.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_random_zoom.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"A preprocessing layer which randomly zooms images during training. — layer_random_zoom","text":"","code":"layer_random_zoom(   object,   height_factor,   width_factor = NULL,   fill_mode = \"reflect\",   interpolation = \"bilinear\",   seed = NULL,   fill_value = 0,   data_format = NULL,   ... )"},{"path":"https://keras3.posit.co/dev/reference/layer_random_zoom.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"A preprocessing layer which randomly zooms images during training. — layer_random_zoom","text":"object Object compose layer . tensor, array, sequential model. height_factor float represented fraction value, list size 2 representing lower upper bound zooming vertically. represented single float, value used upper lower bound. positive value means zooming , negative value means zooming . instance, height_factor=c(0.2, 0.3) result output zoomed random amount range [+20%, +30%]. height_factor=c(-0.3, -0.2) result output zoomed random amount range [+20%, +30%]. width_factor float represented fraction value, list size 2 representing lower upper bound zooming horizontally. represented single float, value used upper lower bound. instance, width_factor=c(0.2, 0.3) result output zooming 20% 30%. width_factor=c(-0.3, -0.2) result output zooming 20% 30%. NULL means .e., zooming vertical horizontal directions preserving aspect ratio. Defaults NULL. fill_mode Points outside boundaries input filled according given mode. Available methods \"constant\", \"nearest\", \"wrap\" \"reflect\". Defaults \"reflect\". \"reflect\": (d c b | b c d | d c b ) input extended reflecting edge last pixel. \"constant\": (k k k k | b c d | k k k k) input extended filling values beyond edge constant value k specified fill_value. \"wrap\": (b c d | b c d | b c d) input extended wrapping around opposite edge. \"nearest\": (| b c d | d d d d) input extended nearest pixel. Note using torch backend, \"reflect\" redirected \"mirror\" (c d c b | b c d | c b b) torch support \"reflect\". Note torch backend support \"wrap\". interpolation Interpolation mode. Supported values: \"nearest\", \"bilinear\". seed Integer. Used create random seed. fill_value float represents value filled outside boundaries fill_mode=\"constant\". data_format string, either \"channels_last\" \"channels_first\". ordering dimensions inputs. \"channels_last\" corresponds inputs shape (batch, height, width, channels) \"channels_first\" corresponds inputs shape (batch, channels, height, width). defaults image_data_format value found Keras config file ~/.keras/keras.json. never set , \"channels_last\". ... Base layer keyword arguments, name dtype.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_random_zoom.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"A preprocessing layer which randomly zooms images during training. — layer_random_zoom","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_random_zoom.html","id":"input-shape","dir":"Reference","previous_headings":"","what":"Input Shape","title":"A preprocessing layer which randomly zooms images during training. — layer_random_zoom","text":"3D (unbatched) 4D (batched) tensor shape: (..., height, width, channels), \"channels_last\" format, (..., channels, height, width), \"channels_first\" format.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_random_zoom.html","id":"output-shape","dir":"Reference","previous_headings":"","what":"Output Shape","title":"A preprocessing layer which randomly zooms images during training. — layer_random_zoom","text":"3D (unbatched) 4D (batched) tensor shape: (..., target_height, target_width, channels), (..., channels, target_height, target_width), \"channels_first\" format. Note: layer safe use inside tf.data pipeline (independently backend using).","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_random_zoom.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"A preprocessing layer which randomly zooms images during training. — layer_random_zoom","text":"","code":"input_img <- random_uniform(c(32, 224, 224, 3)) layer <- layer_random_zoom(height_factor = .5, width_factor = .2) out_img <- layer(input_img)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_repeat_vector.html","id":null,"dir":"Reference","previous_headings":"","what":"Repeats the input n times. — layer_repeat_vector","title":"Repeats the input n times. — layer_repeat_vector","text":"Repeats input n times.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_repeat_vector.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Repeats the input n times. — layer_repeat_vector","text":"","code":"layer_repeat_vector(object, n, ...)"},{"path":"https://keras3.posit.co/dev/reference/layer_repeat_vector.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Repeats the input n times. — layer_repeat_vector","text":"object Object compose layer . tensor, array, sequential model. n Integer, repetition factor. ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_repeat_vector.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Repeats the input n times. — layer_repeat_vector","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_repeat_vector.html","id":"example","dir":"Reference","previous_headings":"","what":"Example","title":"Repeats the input n times. — layer_repeat_vector","text":"","code":"x <- layer_input(shape = 32) y <- layer_repeat_vector(x, n = 3) shape(y) ## shape(NA, 3, 32)"},{"path":"https://keras3.posit.co/dev/reference/layer_repeat_vector.html","id":"input-shape","dir":"Reference","previous_headings":"","what":"Input Shape","title":"Repeats the input n times. — layer_repeat_vector","text":"2D tensor shape (batch_size, features).","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_repeat_vector.html","id":"output-shape","dir":"Reference","previous_headings":"","what":"Output Shape","title":"Repeats the input n times. — layer_repeat_vector","text":"3D tensor shape (batch_size, n, features).","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_rescaling.html","id":null,"dir":"Reference","previous_headings":"","what":"A preprocessing layer which rescales input values to a new range. — layer_rescaling","title":"A preprocessing layer which rescales input values to a new range. — layer_rescaling","text":"layer rescales every value input (often image) multiplying scale adding offset. instance: rescale input [0, 255] range [0, 1] range, pass scale=1./255. rescale input [0, 255] range [-1, 1] range, pass scale=1./127.5, offset=-1. rescaling applied training inference. Inputs can integer floating point dtype, default layer output floats. Note: layer safe use inside tf.data pipeline (independently backend using).","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_rescaling.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"A preprocessing layer which rescales input values to a new range. — layer_rescaling","text":"","code":"layer_rescaling(object, scale, offset = 0, ...)"},{"path":"https://keras3.posit.co/dev/reference/layer_rescaling.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"A preprocessing layer which rescales input values to a new range. — layer_rescaling","text":"object Object compose layer . tensor, array, sequential model. scale Numeric scalar, vector, array. scale apply inputs. scalar, scale applied every feature channel; vector array, scaling applied per channel. offset Numeric scalar, vector, array. offset apply inputs. scalar, offset applied every feature channel; vector array, shift applied per channel. ... Base layer keyword arguments, name dtype.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_rescaling.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"A preprocessing layer which rescales input values to a new range. — layer_rescaling","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_reshape.html","id":null,"dir":"Reference","previous_headings":"","what":"Layer that reshapes inputs into the given shape. — layer_reshape","title":"Layer that reshapes inputs into the given shape. — layer_reshape","text":"Layer reshapes inputs given shape.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_reshape.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Layer that reshapes inputs into the given shape. — layer_reshape","text":"","code":"layer_reshape(object, target_shape, ...)"},{"path":"https://keras3.posit.co/dev/reference/layer_reshape.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Layer that reshapes inputs into the given shape. — layer_reshape","text":"object Object compose layer . tensor, array, sequential model. target_shape Target shape. List integers, include samples dimension (batch size). ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_reshape.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Layer that reshapes inputs into the given shape. — layer_reshape","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_reshape.html","id":"input-shape","dir":"Reference","previous_headings":"","what":"Input Shape","title":"Layer that reshapes inputs into the given shape. — layer_reshape","text":"Arbitrary, although dimensions input shape must known/fixed. Use keyword argument input_shape (list integers, include samples/batch size axis) using layer first layer model.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_reshape.html","id":"output-shape","dir":"Reference","previous_headings":"","what":"Output Shape","title":"Layer that reshapes inputs into the given shape. — layer_reshape","text":"(batch_size, *target_shape)","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_reshape.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Layer that reshapes inputs into the given shape. — layer_reshape","text":"","code":"x <- layer_input(shape = 12) y <- layer_reshape(x, c(3, 4)) shape(y) ## shape(NA, 3, 4) # also supports shape inference using `-1` as dimension y <- layer_reshape(x, c(-1, 2, 2)) shape(y) ## shape(NA, 3, 2, 2)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_resizing.html","id":null,"dir":"Reference","previous_headings":"","what":"A preprocessing layer which resizes images. — layer_resizing","title":"A preprocessing layer which resizes images. — layer_resizing","text":"layer resizes image input target height width. input 4D (batched) 3D (unbatched) tensor \"channels_last\" format. Input pixel values can range (e.g. [0., 1.) [0, 255]).","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_resizing.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"A preprocessing layer which resizes images. — layer_resizing","text":"","code":"layer_resizing(   object,   height,   width,   interpolation = \"bilinear\",   crop_to_aspect_ratio = FALSE,   pad_to_aspect_ratio = FALSE,   fill_mode = \"constant\",   fill_value = 0,   data_format = NULL,   ...,   antialias = FALSE )"},{"path":"https://keras3.posit.co/dev/reference/layer_resizing.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"A preprocessing layer which resizes images. — layer_resizing","text":"object Object compose layer . tensor, array, sequential model. height Integer, height output shape. width Integer, width output shape. interpolation String, interpolation method. Supports \"bilinear\", \"nearest\", \"bicubic\", \"lanczos3\", \"lanczos5\". Defaults \"bilinear\". crop_to_aspect_ratio TRUE, resize images without aspect ratio distortion. original aspect ratio differs target aspect ratio, output image cropped return largest possible window image (size (height, width)) matches target aspect ratio. default (crop_to_aspect_ratio=FALSE), aspect ratio may preserved. pad_to_aspect_ratio TRUE, pad images without aspect ratio distortion. original aspect ratio differs target aspect ratio, output image evenly padded short side. fill_mode using pad_to_aspect_ratio=TRUE, padded areas filled according given mode. \"constant\" supported time (fill constant value, equal fill_value). fill_value Float. Padding value use pad_to_aspect_ratio=TRUE. data_format string, either \"channels_last\" \"channels_first\". ordering dimensions inputs. \"channels_last\" corresponds inputs shape (batch, height, width, channels) \"channels_first\" corresponds inputs shape (batch, channels, height, width). defaults image_data_format value found Keras config file ~/.keras/keras.json. never set , \"channels_last\". ... Base layer keyword arguments, name dtype. antialias bool","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_resizing.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"A preprocessing layer which resizes images. — layer_resizing","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_resizing.html","id":"input-shape","dir":"Reference","previous_headings":"","what":"Input Shape","title":"A preprocessing layer which resizes images. — layer_resizing","text":"3D (unbatched) 4D (batched) tensor shape: (..., height, width, channels), \"channels_last\" format, (..., channels, height, width), \"channels_first\" format.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_resizing.html","id":"output-shape","dir":"Reference","previous_headings":"","what":"Output Shape","title":"A preprocessing layer which resizes images. — layer_resizing","text":"3D (unbatched) 4D (batched) tensor shape: (..., target_height, target_width, channels), (..., channels, target_height, target_width), \"channels_first\" format. Note: layer safe use inside tf.data pipeline (independently backend using).","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_rms_normalization.html","id":null,"dir":"Reference","previous_headings":"","what":"Root Mean Square (RMS) Normalization layer. — layer_rms_normalization","title":"Root Mean Square (RMS) Normalization layer. — layer_rms_normalization","text":"layer normalizes input tensor based RMS value. Keras layer performs operation described Root Mean Square Layer Normalization Biao Zhang et al. scale enabled, layer scale normalized outputs via learnable scaling factor. , scaling enabled, normalization equations follows: Let intermediate activations mini-batch inputs.   example:","code":"rms_normalization(x) = x * rsqrt(mean(square(x))) * scale layer <- layer_rms_normalization() layer$build(shape(5, 20, 30, 10)) op_shape(layer$scale$shape) ## shape(1)  op_shape(layer(op_array(runif(10)))) ## shape(10)"},{"path":"https://keras3.posit.co/dev/reference/layer_rms_normalization.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Root Mean Square (RMS) Normalization layer. — layer_rms_normalization","text":"","code":"layer_rms_normalization(object, axis = -1L, epsilon = 1e-06, ...)"},{"path":"https://keras3.posit.co/dev/reference/layer_rms_normalization.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Root Mean Square (RMS) Normalization layer. — layer_rms_normalization","text":"object Object compose layer . tensor, array, sequential model. axis int. axis perform normalization. epsilon float. small number add avoid division zero. ... forward/backward compatability.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_rnn.html","id":null,"dir":"Reference","previous_headings":"","what":"Base class for recurrent layers — layer_rnn","title":"Base class for recurrent layers — layer_rnn","text":"Base class recurrent layers","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_rnn.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Base class for recurrent layers — layer_rnn","text":"","code":"layer_rnn(   object,   cell,   return_sequences = FALSE,   return_state = FALSE,   go_backwards = FALSE,   stateful = FALSE,   unroll = FALSE,   zero_output_for_mask = FALSE,   ... )"},{"path":"https://keras3.posit.co/dev/reference/layer_rnn.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Base class for recurrent layers — layer_rnn","text":"object Object compose layer . tensor, array, sequential model. cell RNN cell instance list RNN cell instances. RNN cell class : call(input_at_t, states_at_t) method, returning (output_at_t, states_at_t_plus_1). call method cell can also take optional argument constants, see section \"Note passing external constants\" . state_size attribute. can single integer (single state) case size recurrent state. can also list integers (one size per state). output_size attribute, single integer. get_initial_state(batch_size=NULL) method creates tensor meant fed call() initial state, user specify initial state via means. returned initial state shape (batch_size, cell.state_size). cell might choose create tensor full zeros, values based cell's implementation. inputs input tensor RNN layer, shape (batch_size, timesteps, features). method implemented cell, RNN layer create zero filled tensor shape (batch_size, cell$state_size). case cell list RNN cell instances, cells stacked top RNN, resulting efficient stacked RNN. return_sequences Boolean (default FALSE). Whether return last output output sequence, full sequence. return_state Boolean (default FALSE). Whether return last state addition output. go_backwards Boolean (default FALSE). TRUE, process input sequence backwards return reversed sequence. stateful Boolean (default FALSE). TRUE, last state sample index batch used initial state sample index following batch. unroll Boolean (default FALSE). TRUE, network unrolled, else symbolic loop used. Unrolling can speed-RNN, although tends memory-intensive. Unrolling suitable short sequences. zero_output_for_mask Boolean (default FALSE). Whether output use zeros masked timesteps. Note field used return_sequences TRUE mask provided. can useful want reuse raw output sequence RNN without interference masked timesteps, e.g., merging bidirectional RNNs. ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_rnn.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Base class for recurrent layers — layer_rnn","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_rnn.html","id":"call-arguments","dir":"Reference","previous_headings":"","what":"Call Arguments","title":"Base class for recurrent layers — layer_rnn","text":"sequences: 3-D tensor shape (batch_size, timesteps, features). initial_state: List initial state tensors passed first call cell. mask: Binary tensor shape [batch_size, timesteps] indicating whether given timestep masked. individual TRUE entry indicates corresponding timestep utilized, FALSE entry indicates corresponding timestep ignored. training: Python boolean indicating whether layer behave training mode inference mode. argument passed cell calling . use cells use dropout.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_rnn.html","id":"input-shape","dir":"Reference","previous_headings":"","what":"Input Shape","title":"Base class for recurrent layers — layer_rnn","text":"3-D tensor shape (batch_size, timesteps, features).","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_rnn.html","id":"output-shape","dir":"Reference","previous_headings":"","what":"Output Shape","title":"Base class for recurrent layers — layer_rnn","text":"return_state: list tensors. first tensor output. remaining tensors last states, shape (batch_size, state_size), state_size high dimension tensor shape. return_sequences: 3D tensor shape (batch_size, timesteps, output_size).","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_rnn.html","id":"masking-","dir":"Reference","previous_headings":"","what":"Masking:","title":"Base class for recurrent layers — layer_rnn","text":"layer supports masking input data variable number timesteps. introduce masks data, use layer_embedding() layer mask_zero parameter set TRUE. Note using statefulness RNNs: can set RNN layers 'stateful', means states computed samples one batch reused initial states samples next batch. assumes one--one mapping samples different successive batches. enable statefulness: Specify stateful=TRUE layer constructor. Specify fixed batch size model, passing batch_size=... layer_input() layer(s) model. Remember also specify batch_size=... calling fit(), otherwise use generator-like data source like tf.data.Dataset. Specify shuffle=FALSE calling fit(), since batches expected temporally ordered. reset states model, call reset_state() either specific layer, entire model. Note specifying initial state RNNs: can specify initial state RNN layers symbolically passing named argument initial_state layer reset_state(). value initial_state tensor list tensors representing initial state RNN layer.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_rnn.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Base class for recurrent layers — layer_rnn","text":"First, define RNN Cell, layer subclass.   use cell RNN layer:","code":"rnn_cell_minimal <- Layer(   \"MinimalRNNCell\",    initialize = function(units, ...) {     super$initialize(...)     self$units <- as.integer(units)     self$state_size <- as.integer(units)   },    build = function(input_shape) {     self$kernel <- self$add_weight(       shape = shape(tail(input_shape, 1), self$units),       initializer = 'uniform',       name = 'kernel'     )     self$recurrent_kernel <- self$add_weight(       shape = shape(self$units, self$units),       initializer = 'uniform',       name = 'recurrent_kernel'     )   },    call = function(inputs, states) {     prev_output <- states[[1]]     h <- op_matmul(inputs, self$kernel)     output <- h + op_matmul(prev_output, self$recurrent_kernel)     list(output, list(output))   } ) cell <- rnn_cell_minimal(units = 32) x <- layer_input(shape = shape(NULL, 5)) layer <- layer_rnn(cell = cell) y <- layer(x) cells <- list(rnn_cell_minimal(units = 32), rnn_cell_minimal(units = 4)) x <- layer_input(shape = shape(NULL, 5)) layer <- layer_rnn(cell = cells) y <- layer(x)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_separable_conv_1d.html","id":null,"dir":"Reference","previous_headings":"","what":"1D separable convolution layer. — layer_separable_conv_1d","title":"1D separable convolution layer. — layer_separable_conv_1d","text":"layer performs depthwise convolution acts separately channels, followed pointwise convolution mixes channels. use_bias TRUE bias initializer provided, adds bias vector output. optionally applies activation function produce final output.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_separable_conv_1d.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"1D separable convolution layer. — layer_separable_conv_1d","text":"","code":"layer_separable_conv_1d(   object,   filters,   kernel_size,   strides = 1L,   padding = \"valid\",   data_format = NULL,   dilation_rate = 1L,   depth_multiplier = 1L,   activation = NULL,   use_bias = TRUE,   depthwise_initializer = \"glorot_uniform\",   pointwise_initializer = \"glorot_uniform\",   bias_initializer = \"zeros\",   depthwise_regularizer = NULL,   pointwise_regularizer = NULL,   bias_regularizer = NULL,   activity_regularizer = NULL,   depthwise_constraint = NULL,   pointwise_constraint = NULL,   bias_constraint = NULL,   ... )"},{"path":"https://keras3.posit.co/dev/reference/layer_separable_conv_1d.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"1D separable convolution layer. — layer_separable_conv_1d","text":"object Object compose layer . tensor, array, sequential model. filters int, dimensionality output space (.e. number filters pointwise convolution). kernel_size int list 1 integers, specifying size depthwise convolution window. strides int list 1 integers, specifying stride length depthwise convolution. one int specified, stride size used dimensions. strides > 1 incompatible dilation_rate > 1. padding string, either \"valid\" \"\" (case-insensitive). \"valid\" means padding. \"\" results padding evenly left/right /input. padding=\"\" strides=1, output size input. data_format string, either \"channels_last\" \"channels_first\". ordering dimensions inputs. \"channels_last\" corresponds inputs shape (batch, steps, features) \"channels_first\" corresponds inputs shape (batch, features, steps). defaults image_data_format value found Keras config file ~/.keras/keras.json. never set , \"channels_last\". dilation_rate int list 1 integers, specifying dilation rate use dilated convolution. one int specified, dilation rate used dimensions. depth_multiplier number depthwise convolution output channels input channel. total number depthwise convolution output channels equal input_channel * depth_multiplier. activation Activation function. NULL, activation applied. use_bias bool, TRUE, bias added output. depthwise_initializer initializer depthwise convolution kernel. NULL, default initializer (\"glorot_uniform\") used. pointwise_initializer initializer pointwise convolution kernel. NULL, default initializer (\"glorot_uniform\") used. bias_initializer initializer bias vector. NULL, default initializer ('\"zeros\"') used. depthwise_regularizer Optional regularizer depthwise convolution kernel. pointwise_regularizer Optional regularizer pointwise convolution kernel. bias_regularizer Optional regularizer bias vector. activity_regularizer Optional regularizer function output. depthwise_constraint Optional projection function applied depthwise kernel updated Optimizer (e.g. used norm constraints value constraints layer weights). function must take input unprojected variable must return projected variable (must shape). pointwise_constraint Optional projection function applied pointwise kernel updated Optimizer. bias_constraint Optional projection function applied bias updated Optimizer. ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_separable_conv_1d.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"1D separable convolution layer. — layer_separable_conv_1d","text":"3D tensor representing activation(separable_conv1d(inputs, kernel) + bias).","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_separable_conv_1d.html","id":"input-shape","dir":"Reference","previous_headings":"","what":"Input Shape","title":"1D separable convolution layer. — layer_separable_conv_1d","text":"data_format=\"channels_last\": 3D tensor shape: (batch_shape, steps, channels) data_format=\"channels_first\": 3D tensor shape: (batch_shape, channels, steps)","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_separable_conv_1d.html","id":"output-shape","dir":"Reference","previous_headings":"","what":"Output Shape","title":"1D separable convolution layer. — layer_separable_conv_1d","text":"data_format=\"channels_last\": 3D tensor shape: (batch_shape, new_steps, filters) data_format=\"channels_first\": 3D tensor shape: (batch_shape, filters, new_steps)","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_separable_conv_1d.html","id":"example","dir":"Reference","previous_headings":"","what":"Example","title":"1D separable convolution layer. — layer_separable_conv_1d","text":"","code":"x <- random_uniform(c(4, 10, 12)) y <- layer_separable_conv_1d(x, 3, 2, 2, activation='relu') shape(y) ## shape(4, 5, 3)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_separable_conv_2d.html","id":null,"dir":"Reference","previous_headings":"","what":"2D separable convolution layer. — layer_separable_conv_2d","title":"2D separable convolution layer. — layer_separable_conv_2d","text":"layer performs depthwise convolution acts separately channels, followed pointwise convolution mixes channels. use_bias TRUE bias initializer provided, adds bias vector output. optionally applies activation function produce final output.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_separable_conv_2d.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"2D separable convolution layer. — layer_separable_conv_2d","text":"","code":"layer_separable_conv_2d(   object,   filters,   kernel_size,   strides = list(1L, 1L),   padding = \"valid\",   data_format = NULL,   dilation_rate = list(1L, 1L),   depth_multiplier = 1L,   activation = NULL,   use_bias = TRUE,   depthwise_initializer = \"glorot_uniform\",   pointwise_initializer = \"glorot_uniform\",   bias_initializer = \"zeros\",   depthwise_regularizer = NULL,   pointwise_regularizer = NULL,   bias_regularizer = NULL,   activity_regularizer = NULL,   depthwise_constraint = NULL,   pointwise_constraint = NULL,   bias_constraint = NULL,   ... )"},{"path":"https://keras3.posit.co/dev/reference/layer_separable_conv_2d.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"2D separable convolution layer. — layer_separable_conv_2d","text":"object Object compose layer . tensor, array, sequential model. filters int, dimensionality output space (.e. number filters pointwise convolution). kernel_size int list 2 integers, specifying size depthwise convolution window. strides int list 2 integers, specifying stride length depthwise convolution. one int specified, stride size used dimensions. strides > 1 incompatible dilation_rate > 1. padding string, either \"valid\" \"\" (case-insensitive). \"valid\" means padding. \"\" results padding evenly left/right /input. padding=\"\" strides=1, output size input. data_format string, either \"channels_last\" \"channels_first\". ordering dimensions inputs. \"channels_last\" corresponds inputs shape (batch, height, width, channels) \"channels_first\" corresponds inputs shape (batch, channels, height, width). defaults image_data_format value found Keras config file ~/.keras/keras.json. never set , \"channels_last\". dilation_rate int list 2 integers, specifying dilation rate use dilated convolution. one int specified, dilation rate used dimensions. depth_multiplier number depthwise convolution output channels input channel. total number depthwise convolution output channels equal input_channel * depth_multiplier. activation Activation function. NULL, activation applied. use_bias bool, TRUE, bias added output. depthwise_initializer initializer depthwise convolution kernel. NULL, default initializer (\"glorot_uniform\") used. pointwise_initializer initializer pointwise convolution kernel. NULL, default initializer (\"glorot_uniform\") used. bias_initializer initializer bias vector. NULL, default initializer ('\"zeros\"') used. depthwise_regularizer Optional regularizer depthwise convolution kernel. pointwise_regularizer Optional regularizer pointwise convolution kernel. bias_regularizer Optional regularizer bias vector. activity_regularizer Optional regularizer function output. depthwise_constraint Optional projection function applied depthwise kernel updated Optimizer (e.g. used norm constraints value constraints layer weights). function must take input unprojected variable must return projected variable (must shape). pointwise_constraint Optional projection function applied pointwise kernel updated Optimizer. bias_constraint Optional projection function applied bias updated Optimizer. ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_separable_conv_2d.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"2D separable convolution layer. — layer_separable_conv_2d","text":"4D tensor representing activation(separable_conv2d(inputs, kernel) + bias).","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_separable_conv_2d.html","id":"input-shape","dir":"Reference","previous_headings":"","what":"Input Shape","title":"2D separable convolution layer. — layer_separable_conv_2d","text":"data_format=\"channels_last\": 4D tensor shape: (batch_size, height, width, channels) data_format=\"channels_first\": 4D tensor shape: (batch_size, channels, height, width)","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_separable_conv_2d.html","id":"output-shape","dir":"Reference","previous_headings":"","what":"Output Shape","title":"2D separable convolution layer. — layer_separable_conv_2d","text":"data_format=\"channels_last\": 4D tensor shape: (batch_size, new_height, new_width, filters) data_format=\"channels_first\": 4D tensor shape: (batch_size, filters, new_height, new_width)","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_separable_conv_2d.html","id":"example","dir":"Reference","previous_headings":"","what":"Example","title":"2D separable convolution layer. — layer_separable_conv_2d","text":"","code":"x <- random_uniform(c(4, 10, 10, 12)) y <- layer_separable_conv_2d(x, 3, c(4, 3), 2, activation='relu') shape(y) ## shape(4, 4, 4, 3)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_simple_rnn.html","id":null,"dir":"Reference","previous_headings":"","what":"Fully-connected RNN where the output is to be fed back as the new input. — layer_simple_rnn","title":"Fully-connected RNN where the output is to be fed back as the new input. — layer_simple_rnn","text":"Fully-connected RNN output fed back new input.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_simple_rnn.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Fully-connected RNN where the output is to be fed back as the new input. — layer_simple_rnn","text":"","code":"layer_simple_rnn(   object,   units,   activation = \"tanh\",   use_bias = TRUE,   kernel_initializer = \"glorot_uniform\",   recurrent_initializer = \"orthogonal\",   bias_initializer = \"zeros\",   kernel_regularizer = NULL,   recurrent_regularizer = NULL,   bias_regularizer = NULL,   activity_regularizer = NULL,   kernel_constraint = NULL,   recurrent_constraint = NULL,   bias_constraint = NULL,   dropout = 0,   recurrent_dropout = 0,   return_sequences = FALSE,   return_state = FALSE,   go_backwards = FALSE,   stateful = FALSE,   unroll = FALSE,   seed = NULL,   ... )"},{"path":"https://keras3.posit.co/dev/reference/layer_simple_rnn.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Fully-connected RNN where the output is to be fed back as the new input. — layer_simple_rnn","text":"object Object compose layer . tensor, array, sequential model. units Positive integer, dimensionality output space. activation Activation function use. Default: hyperbolic tangent (tanh). pass NULL, activation applied (ie. \"linear\" activation: (x) = x). use_bias Boolean, (default TRUE), whether layer uses bias vector. kernel_initializer Initializer kernel weights matrix, used linear transformation inputs. Default: \"glorot_uniform\". recurrent_initializer Initializer recurrent_kernel weights matrix, used linear transformation recurrent state.  Default: \"orthogonal\". bias_initializer Initializer bias vector. Default: \"zeros\". kernel_regularizer Regularizer function applied kernel weights matrix. Default: NULL. recurrent_regularizer Regularizer function applied recurrent_kernel weights matrix. Default: NULL. bias_regularizer Regularizer function applied bias vector. Default: NULL. activity_regularizer Regularizer function applied output layer (\"activation\"). Default: NULL. kernel_constraint Constraint function applied kernel weights matrix. Default: NULL. recurrent_constraint Constraint function applied recurrent_kernel weights matrix.  Default: NULL. bias_constraint Constraint function applied bias vector. Default: NULL. dropout Float 0 1. Fraction units drop linear transformation inputs. Default: 0. recurrent_dropout Float 0 1. Fraction units drop linear transformation recurrent state. Default: 0. return_sequences Boolean. Whether return last output output sequence, full sequence. Default: FALSE. return_state Boolean. Whether return last state addition output. Default: FALSE. go_backwards Boolean (default: FALSE). TRUE, process input sequence backwards return reversed sequence. stateful Boolean (default: FALSE). TRUE, last state sample index batch used initial state sample index following batch. unroll Boolean (default: FALSE). TRUE, network unrolled, else symbolic loop used. Unrolling can speed-RNN, although tends memory-intensive. Unrolling suitable short sequences. seed Initial seed random number generator ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_simple_rnn.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Fully-connected RNN where the output is to be fed back as the new input. — layer_simple_rnn","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_simple_rnn.html","id":"call-arguments","dir":"Reference","previous_headings":"","what":"Call Arguments","title":"Fully-connected RNN where the output is to be fed back as the new input. — layer_simple_rnn","text":"sequence: 3D tensor, shape [batch, timesteps, feature]. mask: Binary tensor shape [batch, timesteps] indicating whether given timestep masked. individual TRUE entry indicates corresponding timestep utilized, FALSE entry indicates corresponding timestep ignored. training: Python boolean indicating whether layer behave training mode inference mode. argument passed cell calling . relevant dropout recurrent_dropout used. initial_state: List initial state tensors passed first call cell.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_simple_rnn.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Fully-connected RNN where the output is to be fed back as the new input. — layer_simple_rnn","text":"","code":"inputs <- random_uniform(c(32, 10, 8)) simple_rnn <- layer_simple_rnn(units = 4) output <- simple_rnn(inputs)  # The output has shape `(32, 4)`. simple_rnn <- layer_simple_rnn(     units = 4, return_sequences=TRUE, return_state=TRUE ) # whole_sequence_output has shape `(32, 10, 4)`. # final_state has shape `(32, 4)`. c(whole_sequence_output, final_state) %<-% simple_rnn(inputs)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_solarization.html","id":null,"dir":"Reference","previous_headings":"","what":"Applies (max_value - pixel + min_value) for each pixel in the image. — layer_solarization","title":"Applies (max_value - pixel + min_value) for each pixel in the image. — layer_solarization","text":"created without threshold parameter, layer performs solarization values. created specified threshold layer augments pixels threshold value.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_solarization.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Applies (max_value - pixel + min_value) for each pixel in the image. — layer_solarization","text":"","code":"layer_solarization(   object,   addition_factor = 0,   threshold_factor = 0,   value_range = tuple(0L, 255L),   seed = NULL,   ... )"},{"path":"https://keras3.posit.co/dev/reference/layer_solarization.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Applies (max_value - pixel + min_value) for each pixel in the image. — layer_solarization","text":"object Object compose layer . tensor, array, sequential model. addition_factor (Optional)  tuple two floats single float, 0 1. augmented image value sampled provided range. float passed, range interpreted (0, addition_factor). specified, value (times value range input images, e.g. 255), added pixel solarization thresholding. Defaults 0.0. threshold_factor (Optional)  tuple two floats single float. augmented image value sampled provided range. float passed, range interpreted (0, threshold_factor). specified, pixel values threshold solarized. value_range tuple list two elements. first value represents lower bound values input images, second represents upper bound. Images passed layer values within value_range. Typical values pass (0, 255) (RGB image) (0., 1.) (scaled image). seed Integer. Used create random seed. ... Base layer keyword arguments, name dtype.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_solarization.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Applies (max_value - pixel + min_value) for each pixel in the image. — layer_solarization","text":"","code":"c(c(images, labels), .) %<-% dataset_cifar10() str(images) ##  int [1:50000, 1:32, 1:32, 1:3] 59 154 255 28 170 159 164 28 134 125 ... str(images[1, 1, 1, ]) ##  int [1:3] 59 62 63 # Note that images are Tensor with values in the range [0, 255] solarization <- layer_solarization(value_range = c(0, 255)) images <- solarization(images) |> as.array() str(images[1, 1, 1, ]) ##  num [1:3] 196 193 192"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_spatial_dropout_1d.html","id":null,"dir":"Reference","previous_headings":"","what":"Spatial 1D version of Dropout. — layer_spatial_dropout_1d","title":"Spatial 1D version of Dropout. — layer_spatial_dropout_1d","text":"layer performs function Dropout, however, drops entire 1D feature maps instead individual elements. adjacent frames within feature maps strongly correlated (normally case early convolution layers) regular dropout regularize activations otherwise just result effective learning rate decrease. case, SpatialDropout1D help promote independence feature maps used instead.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_spatial_dropout_1d.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Spatial 1D version of Dropout. — layer_spatial_dropout_1d","text":"","code":"layer_spatial_dropout_1d(object, rate, seed = NULL, name = NULL, dtype = NULL)"},{"path":"https://keras3.posit.co/dev/reference/layer_spatial_dropout_1d.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Spatial 1D version of Dropout. — layer_spatial_dropout_1d","text":"object Object compose layer . tensor, array, sequential model. rate Float 0 1. Fraction input units drop. seed Initial seed random number generator name String, name object dtype datatype (e.g., \"float32\").","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_spatial_dropout_1d.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Spatial 1D version of Dropout. — layer_spatial_dropout_1d","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_spatial_dropout_1d.html","id":"call-arguments","dir":"Reference","previous_headings":"","what":"Call Arguments","title":"Spatial 1D version of Dropout. — layer_spatial_dropout_1d","text":"inputs: 3D tensor. training: Python boolean indicating whether layer behave training mode (applying dropout) inference mode (pass-).","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_spatial_dropout_1d.html","id":"input-shape","dir":"Reference","previous_headings":"","what":"Input Shape","title":"Spatial 1D version of Dropout. — layer_spatial_dropout_1d","text":"3D tensor shape: (samples, timesteps, channels)","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_spatial_dropout_1d.html","id":"output-shape","dir":"Reference","previous_headings":"","what":"Output Shape","title":"Spatial 1D version of Dropout. — layer_spatial_dropout_1d","text":"input.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_spatial_dropout_1d.html","id":"reference","dir":"Reference","previous_headings":"","what":"Reference","title":"Spatial 1D version of Dropout. — layer_spatial_dropout_1d","text":"Tompson et al., 2014","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_spatial_dropout_2d.html","id":null,"dir":"Reference","previous_headings":"","what":"Spatial 2D version of Dropout. — layer_spatial_dropout_2d","title":"Spatial 2D version of Dropout. — layer_spatial_dropout_2d","text":"version performs function Dropout, however, drops entire 2D feature maps instead individual elements. adjacent pixels within feature maps strongly correlated (normally case early convolution layers) regular dropout regularize activations otherwise just result effective learning rate decrease. case, SpatialDropout2D help promote independence feature maps used instead.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_spatial_dropout_2d.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Spatial 2D version of Dropout. — layer_spatial_dropout_2d","text":"","code":"layer_spatial_dropout_2d(   object,   rate,   data_format = NULL,   seed = NULL,   name = NULL,   dtype = NULL )"},{"path":"https://keras3.posit.co/dev/reference/layer_spatial_dropout_2d.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Spatial 2D version of Dropout. — layer_spatial_dropout_2d","text":"object Object compose layer . tensor, array, sequential model. rate Float 0 1. Fraction input units drop. data_format \"channels_first\" \"channels_last\". \"channels_first\" mode, channels dimension (depth) index 1, \"channels_last\" mode index 3. defaults image_data_format value found Keras config file ~/.keras/keras.json. never set , \"channels_last\". seed Initial seed random number generator name String, name object dtype datatype (e.g., \"float32\").","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_spatial_dropout_2d.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Spatial 2D version of Dropout. — layer_spatial_dropout_2d","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_spatial_dropout_2d.html","id":"call-arguments","dir":"Reference","previous_headings":"","what":"Call Arguments","title":"Spatial 2D version of Dropout. — layer_spatial_dropout_2d","text":"inputs: 4D tensor. training: Python boolean indicating whether layer behave training mode (applying dropout) inference mode (pass-).","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_spatial_dropout_2d.html","id":"input-shape","dir":"Reference","previous_headings":"","what":"Input Shape","title":"Spatial 2D version of Dropout. — layer_spatial_dropout_2d","text":"4D tensor shape: (samples, channels, rows, cols) data_format='channels_first' 4D tensor shape: (samples, rows, cols, channels) data_format='channels_last'.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_spatial_dropout_2d.html","id":"output-shape","dir":"Reference","previous_headings":"","what":"Output Shape","title":"Spatial 2D version of Dropout. — layer_spatial_dropout_2d","text":"input.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_spatial_dropout_2d.html","id":"reference","dir":"Reference","previous_headings":"","what":"Reference","title":"Spatial 2D version of Dropout. — layer_spatial_dropout_2d","text":"Tompson et al., 2014","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_spatial_dropout_3d.html","id":null,"dir":"Reference","previous_headings":"","what":"Spatial 3D version of Dropout. — layer_spatial_dropout_3d","title":"Spatial 3D version of Dropout. — layer_spatial_dropout_3d","text":"version performs function Dropout, however, drops entire 3D feature maps instead individual elements. adjacent voxels within feature maps strongly correlated (normally case early convolution layers) regular dropout regularize activations otherwise just result effective learning rate decrease. case, SpatialDropout3D help promote independence feature maps used instead.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_spatial_dropout_3d.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Spatial 3D version of Dropout. — layer_spatial_dropout_3d","text":"","code":"layer_spatial_dropout_3d(   object,   rate,   data_format = NULL,   seed = NULL,   name = NULL,   dtype = NULL )"},{"path":"https://keras3.posit.co/dev/reference/layer_spatial_dropout_3d.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Spatial 3D version of Dropout. — layer_spatial_dropout_3d","text":"object Object compose layer . tensor, array, sequential model. rate Float 0 1. Fraction input units drop. data_format \"channels_first\" \"channels_last\". \"channels_first\" mode, channels dimension (depth) index 1, \"channels_last\" mode index 4. defaults image_data_format value found Keras config file ~/.keras/keras.json. never set , \"channels_last\". seed Initial seed random number generator name String, name object dtype datatype (e.g., \"float32\").","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_spatial_dropout_3d.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Spatial 3D version of Dropout. — layer_spatial_dropout_3d","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_spatial_dropout_3d.html","id":"call-arguments","dir":"Reference","previous_headings":"","what":"Call Arguments","title":"Spatial 3D version of Dropout. — layer_spatial_dropout_3d","text":"inputs: 5D tensor. training: Python boolean indicating whether layer behave training mode (applying dropout) inference mode (pass-).","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_spatial_dropout_3d.html","id":"input-shape","dir":"Reference","previous_headings":"","what":"Input Shape","title":"Spatial 3D version of Dropout. — layer_spatial_dropout_3d","text":"5D tensor shape: (samples, channels, dim1, dim2, dim3) data_format='channels_first' 5D tensor shape: (samples, dim1, dim2, dim3, channels) data_format='channels_last'.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_spatial_dropout_3d.html","id":"output-shape","dir":"Reference","previous_headings":"","what":"Output Shape","title":"Spatial 3D version of Dropout. — layer_spatial_dropout_3d","text":"input.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_spatial_dropout_3d.html","id":"reference","dir":"Reference","previous_headings":"","what":"Reference","title":"Spatial 3D version of Dropout. — layer_spatial_dropout_3d","text":"Tompson et al., 2014","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_spectral_normalization.html","id":null,"dir":"Reference","previous_headings":"","what":"Performs spectral normalization on the weights of a target layer. — layer_spectral_normalization","title":"Performs spectral normalization on the weights of a target layer. — layer_spectral_normalization","text":"wrapper controls Lipschitz constant weights layer constraining spectral norm, can stabilize training GANs.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_spectral_normalization.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Performs spectral normalization on the weights of a target layer. — layer_spectral_normalization","text":"","code":"layer_spectral_normalization(object, layer, power_iterations = 1L, ...)"},{"path":"https://keras3.posit.co/dev/reference/layer_spectral_normalization.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Performs spectral normalization on the weights of a target layer. — layer_spectral_normalization","text":"object Object compose layer . tensor, array, sequential model. layer Layer instance either kernel (e.g. layer_conv_2d, layer_dense...) embeddings attribute (layer_embedding layer). power_iterations int, number iterations normalization. ... Base wrapper keyword arguments.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_spectral_normalization.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Performs spectral normalization on the weights of a target layer. — layer_spectral_normalization","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_spectral_normalization.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Performs spectral normalization on the weights of a target layer. — layer_spectral_normalization","text":"Wrap layer_conv_2d:     Wrap layer_dense:","code":"x <- random_uniform(c(1, 10, 10, 1)) conv2d <- layer_spectral_normalization(   layer = layer_conv_2d(filters = 2, kernel_size = 2) ) y <- conv2d(x) shape(y) ## shape(1, 9, 9, 2) x <- random_uniform(c(1, 10, 10, 1)) dense <- layer_spectral_normalization(layer = layer_dense(units = 10)) y <- dense(x) shape(y) ## shape(1, 10, 10, 10)"},{"path":"https://keras3.posit.co/dev/reference/layer_spectral_normalization.html","id":"reference","dir":"Reference","previous_headings":"","what":"Reference","title":"Performs spectral normalization on the weights of a target layer. — layer_spectral_normalization","text":"Spectral Normalization GAN.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_stft_spectrogram.html","id":null,"dir":"Reference","previous_headings":"","what":"Layer to compute the Short-Time Fourier Transform (STFT) on a 1D signal. — layer_stft_spectrogram","title":"Layer to compute the Short-Time Fourier Transform (STFT) on a 1D signal. — layer_stft_spectrogram","text":"layer computes Spectrograms input signal produce spectrogram. layers layer computes Spectrograms based Short-Time Fourier Transform (STFT) utilizing convolution kernels, allows parallelization GPUs trainable kernels fine-tuning support. layer allows different modes output (e.g., log-scaled magnitude, phase, power spectral density, etc.) provides flexibility windowing, padding, scaling options STFT calculation.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_stft_spectrogram.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Layer to compute the Short-Time Fourier Transform (STFT) on a 1D signal. — layer_stft_spectrogram","text":"","code":"layer_stft_spectrogram(   object,   mode = \"log\",   frame_length = 256L,   frame_step = NULL,   fft_length = NULL,   window = \"hann\",   periodic = FALSE,   scaling = \"density\",   padding = \"valid\",   expand_dims = FALSE,   data_format = NULL,   ... )"},{"path":"https://keras3.posit.co/dev/reference/layer_stft_spectrogram.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Layer to compute the Short-Time Fourier Transform (STFT) on a 1D signal. — layer_stft_spectrogram","text":"object Object compose layer . tensor, array, sequential model. mode String, output type spectrogram. Can one \"log\", \"magnitude\", \"psd\", \"real\", \"imag\", \"angle\", \"stft\". Defaults \"log\". frame_length Integer, length frame (window) STFT samples. Defaults 256. frame_step Integer, step size (hop length) consecutive frames. provided, defaults half frame_length. Defaults frame_length %/% 2. fft_length Integer, size frequency bins used Fast-Fourier Transform (FFT) apply frame. greater equal frame_length.  Recommended power two. Defaults smallest power two greater equal frame_length. window (String array_like), windowing function apply frame. Can \"hann\" (default), \"hamming\", custom window provided array_like. periodic Boolean, TRUE, window function treated periodic. Defaults FALSE. scaling String, type scaling applied window. Can \"density\", \"spectrum\", None. Default \"density\". padding String, padding strategy. Can \"valid\" \"\". Defaults \"valid\". expand_dims Boolean, TRUE, expand output spectrograms two dimensions compatible image models. Defaults FALSE. data_format String, either \"channels_last\" \"channels_first\". ordering dimensions inputs. \"channels_last\" corresponds inputs shape (batch, height, width, channels) \"channels_first\" corresponds inputs shape (batch, channels, height, weight). Defaults \"channels_last\". ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_stft_spectrogram.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Layer to compute the Short-Time Fourier Transform (STFT) on a 1D signal. — layer_stft_spectrogram","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_stft_spectrogram.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Layer to compute the Short-Time Fourier Transform (STFT) on a 1D signal. — layer_stft_spectrogram","text":"Apply non-trainable preprocessing layer 3 audio tracks 1 channel, 10 seconds sampled 16 kHz.     Apply trainable processing layer 3 stereo audio tracks 2 channels, 10 seconds sampled 16 kHz. initialized non-trainable layer, can trained jointly within model.     Similar last example, add extra dimension output image used image models. apply signal 3 input channels output image tensor, hence directly applicable image model.","code":"layer <- layer_stft_spectrogram(   mode = 'log',   frame_length = 256,   frame_step = 128, # 50% overlap   fft_length = 512,   window = \"hann\",   padding = \"valid\",   trainable = FALSE # non-trainable, preprocessing only) ) random_uniform(shape=c(3, 160000, 1)) |> layer() |> op_shape() ## shape(3, 1249, 257) layer <- layer_stft_spectrogram(   mode = 'log',   frame_length = 256,   frame_step = 128,   # 50% overlap   fft_length = 512,   window = \"hamming\", # hamming windowing function   padding = \"same\",   # padding to preserve the time dimension   trainable = TRUE,   # trainable, this is the default in keras ) random_uniform(shape=c(3, 160000, 2)) |> layer() |> op_shape() ## shape(3, 1250, 514) layer <- layer_stft_spectrogram(   mode = 'log',   frame_length = 256,   frame_step = 128,   fft_length = 512,   padding = \"same\",   expand_dims = TRUE  # this adds the extra dimension ) random_uniform(shape=c(3, 160000, 3)) |> layer() |> op_shape() ## shape(3, 1250, 257, 3)"},{"path":"https://keras3.posit.co/dev/reference/layer_stft_spectrogram.html","id":"raises","dir":"Reference","previous_headings":"","what":"Raises","title":"Layer to compute the Short-Time Fourier Transform (STFT) on a 1D signal. — layer_stft_spectrogram","text":"ValueError: invalid value provided \"mode\", \"scaling\", \"padding\", input arguments. TypeError: input data type one \"float16\", \"float32\", \"float64\".","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_stft_spectrogram.html","id":"input-shape","dir":"Reference","previous_headings":"","what":"Input Shape","title":"Layer to compute the Short-Time Fourier Transform (STFT) on a 1D signal. — layer_stft_spectrogram","text":"3D tensor shape (batch_size, time_length, input_channels), data_format==\"channels_last\", shape (batch_size, input_channels, time_length) data_format==\"channels_first\", time_length length input signal, input_channels number input channels. kernels applied channel independently.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_stft_spectrogram.html","id":"output-shape","dir":"Reference","previous_headings":"","what":"Output Shape","title":"Layer to compute the Short-Time Fourier Transform (STFT) on a 1D signal. — layer_stft_spectrogram","text":"data_format==\"channels_first\" && !expand_dims, 3D tensor: (batch_size, input_channels * freq_channels, new_time_length) data_format==\"channels_last\" && !expand_dims, 3D tensor: (batch_size, new_time_length, input_channels * freq_channels) data_format==\"channels_first\" && expand_dims, 4D tensor: (batch_size, input_channels, new_time_length, freq_channels) data_format==\"channels_last\" && expand_dims, 4D tensor: (batch_size, new_time_length, freq_channels, input_channels) new_time_length depends padding, freq_channels number FFT bins (fft_length %/% 2 + 1).","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_string_lookup.html","id":null,"dir":"Reference","previous_headings":"","what":"A preprocessing layer that maps strings to (possibly encoded) indices. — layer_string_lookup","title":"A preprocessing layer that maps strings to (possibly encoded) indices. — layer_string_lookup","text":"layer translates set arbitrary strings integer output via table-based vocabulary lookup. layer perform splitting transformation input strings. layer can split tokenize natural language, see layer_text_vectorization. vocabulary layer must either supplied construction learned via adapt(). adapt(), layer analyze data set, determine frequency individual strings tokens, create vocabulary . vocabulary capped size, frequent tokens used create vocabulary others treated --vocabulary (OOV). two possible output modes layer. output_mode \"int\", input strings converted (0-based) index vocabulary (integer). output_mode \"multi_hot\", \"count\", \"tf_idf\", input strings encoded array dimension corresponds element vocabulary. vocabulary can optionally contain mask token well OOV token (can optionally occupy multiple indices vocabulary, set num_oov_indices). position tokens vocabulary fixed. output_mode \"int\", vocabulary begin mask token (set), followed OOV indices, followed rest vocabulary. output_mode \"multi_hot\", \"count\", \"tf_idf\" vocabulary begin OOV indices instances mask token dropped. Note: layer uses TensorFlow internally. used part compiled computation graph model backend TensorFlow. can however used backend running eagerly. can also always used part input preprocessing pipeline backend (outside model ), recommend use layer. Note: layer safe use inside tf.data pipeline (independently backend using). Note: working layer outputs directly (e.g., passing outputs another layer, using lower-level operations like [ op_*): returned indices 0-based. However, default settings, first (0) index OOV token, returned indices offset 1 may appear 1-based.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_string_lookup.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"A preprocessing layer that maps strings to (possibly encoded) indices. — layer_string_lookup","text":"","code":"layer_string_lookup(   object,   max_tokens = NULL,   num_oov_indices = 1L,   mask_token = NULL,   oov_token = \"[UNK]\",   vocabulary = NULL,   idf_weights = NULL,   invert = FALSE,   output_mode = \"int\",   pad_to_max_tokens = FALSE,   sparse = FALSE,   encoding = \"utf-8\",   name = NULL,   ...,   vocabulary_dtype = NULL )"},{"path":"https://keras3.posit.co/dev/reference/layer_string_lookup.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"A preprocessing layer that maps strings to (possibly encoded) indices. — layer_string_lookup","text":"object Object compose layer . tensor, array, sequential model. max_tokens Maximum size vocabulary layer. specified adapting vocabulary setting pad_to_max_tokens=TRUE. NULL, cap size vocabulary. Note size includes OOV mask tokens. Defaults NULL. num_oov_indices number --vocabulary tokens use. value 1, OOV inputs modulated determine OOV value. value 0, OOV inputs cause error calling layer. Defaults 1. mask_token token represents masked inputs. output_mode \"int\", token included vocabulary mapped index 0. output modes, token appear vocabulary instances mask token input dropped. set NULL, mask term added. Defaults NULL. oov_token used invert TRUE. token return OOV indices. Defaults \"[UNK]\". vocabulary Optional. Either array integers string path text file. passing array, can pass list, list, 1D NumPy array, 1D tensor containing integer vocabulary terms. passing file path, file contain one line per term vocabulary. argument set, need adapt() layer. idf_weights valid output_mode \"tf_idf\". list, list, 1D NumPy array, 1D tensor length vocabulary, containing floating point inverse document frequency weights, multiplied per sample term counts final TF-IDF weight. vocabulary argument set, output_mode \"tf_idf\", argument must supplied. invert valid output_mode \"int\". TRUE, layer map indices vocabulary items instead mapping vocabulary items indices. Defaults FALSE. output_mode Specification output layer. Values can \"int\", \"one_hot\", \"multi_hot\", \"count\", \"tf_idf\" configuring layer follows: \"int\": Return vocabulary indices input tokens. \"one_hot\": Encodes individual element input array size vocabulary, containing 1 element index. last dimension size 1, encode dimension. last dimension size 1, append new dimension encoded output. \"multi_hot\": Encodes sample input single array size vocabulary, containing 1 vocabulary term present sample. Treats last dimension sample dimension, input shape (..., sample_length), output shape (..., num_tokens). \"count\": \"multi_hot\", int array contains count number times token index appeared sample. \"tf_idf\": \"multi_hot\", TF-IDF algorithm applied find value token slot. \"int\" output, shape input output supported. output modes, currently output rank 2 supported. Defaults \"int\". pad_to_max_tokens applicable output_mode \"multi_hot\", \"count\", \"tf_idf\". TRUE, output feature axis padded max_tokens even number unique tokens vocabulary less max_tokens, resulting tensor shape (batch_size, max_tokens) regardless vocabulary size. Defaults FALSE. sparse Boolean. applicable \"multi_hot\", \"count\", \"tf_idf\" output modes. supported TensorFlow backend. TRUE, returns SparseTensor instead dense Tensor. Defaults FALSE. encoding Optional. text encoding use interpret input strings. Defaults \"utf-8\". name String, name object ... forward/backward compatability. vocabulary_dtype dtype vocabulary terms, example \"int64\" \"int32\". Defaults \"int64\".","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_string_lookup.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"A preprocessing layer that maps strings to (possibly encoded) indices. — layer_string_lookup","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_string_lookup.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"A preprocessing layer that maps strings to (possibly encoded) indices. — layer_string_lookup","text":"Creating lookup layer known vocabulary example creates lookup layer pre-existing vocabulary.     Creating lookup layer adapted vocabulary example creates lookup layer generates vocabulary analyzing dataset.     Note OOV token \"[UNK]\" added vocabulary. remaining tokens sorted frequency (\"d\", 2 occurrences, first) inverse sort order.     Lookups multiple OOV indices example demonstrates use lookup layer multiple OOV indices.  layer created one OOV index, OOV values hashed number OOV buckets, distributing OOV values deterministic fashion across set.     Note output OOV value 'm' 0, output OOV value \"z\" 1. -vocab terms output index increased 1 earlier examples (maps 2, etc) order make space extra OOV value. One-hot output Configure layer output_mode='one_hot'. Note first num_oov_indices dimensions ont_hot encoding represent OOV values.     Multi-hot output Configure layer output_mode='multi_hot'. Note first num_oov_indices dimensions multi_hot encoding represent OOV values.     Token count output Configure layer output_mode='count'. multi_hot output, first num_oov_indices dimensions output represent OOV values.     TF-IDF output Configure layer output_mode=\"tf_idf\". multi_hot output, first num_oov_indices dimensions output represent OOV values. token bin output token_count * idf_weight, idf weights inverse document frequency weights per token. provided along vocabulary. Note idf_weight OOV values default average idf weights passed .     specify idf weights oov values, need pass entire vocabulary including leading oov token.     adapting layer \"tf_idf\" mode, input sample considered document, IDF weight per token calculated log(1 + num_documents / (1 + token_document_count)). Inverse lookup example demonstrates map indices strings using layer. (can also use adapt() inverse=TRUE, simplicity pass vocab example.)     Note first index correspond oov token default. Forward inverse lookup pairs example demonstrates use vocabulary standard lookup layer create inverse lookup layer.     example, input value \"z\" resulted output \"[UNK]\", since 1000 vocabulary - got represented OOV, OOV values returned \"[UNK]\" inverse layer. Also, note inverse work, must already set forward layer vocabulary either directly via adapt() calling get_vocabulary().","code":"vocab <- c(\"a\", \"b\", \"c\", \"d\") data <- rbind(c(\"a\", \"c\", \"d\"), c(\"d\", \"z\", \"b\")) layer <- layer_string_lookup(vocabulary=vocab) layer(data) ## tf.Tensor( ## [[1 3 4] ##  [4 0 2]], shape=(2, 3), dtype=int64) data <- rbind(c(\"a\", \"c\", \"d\"), c(\"d\", \"z\", \"b\")) layer <- layer_string_lookup() layer %>% adapt(data) get_vocabulary(layer) ## [1] \"[UNK]\" \"d\"     \"z\"     \"c\"     \"b\"     \"a\" data <- rbind(c(\"a\", \"c\", \"d\"), c(\"d\", \"z\", \"b\")) layer <- layer_string_lookup() layer %>% adapt(data) layer(data) ## tf.Tensor( ## [[5 3 1] ##  [1 2 4]], shape=(2, 3), dtype=int64) vocab <- c(\"a\", \"b\", \"c\", \"d\") data <- rbind(c(\"a\", \"c\", \"d\"), c(\"m\", \"z\", \"b\")) layer <- layer_string_lookup(vocabulary = vocab, num_oov_indices = 2) layer(data) ## tf.Tensor( ## [[2 4 5] ##  [0 1 3]], shape=(2, 3), dtype=int64) vocab <- c(\"a\", \"b\", \"c\", \"d\") data <- c(\"a\", \"b\", \"c\", \"d\", \"z\") layer <- layer_string_lookup(vocabulary = vocab, output_mode = 'one_hot') layer(data) ## tf.Tensor( ## [[0 1 0 0 0] ##  [0 0 1 0 0] ##  [0 0 0 1 0] ##  [0 0 0 0 1] ##  [1 0 0 0 0]], shape=(5, 5), dtype=int64) vocab <- c(\"a\", \"b\", \"c\", \"d\") data <- rbind(c(\"a\", \"c\", \"d\", \"d\"), c(\"d\", \"z\", \"b\", \"z\")) layer <- layer_string_lookup(vocabulary = vocab, output_mode = 'multi_hot') layer(data) ## tf.Tensor( ## [[0 1 0 1 1] ##  [1 0 1 0 1]], shape=(2, 5), dtype=int64) vocab <- c(\"a\", \"b\", \"c\", \"d\") data <- rbind(c(\"a\", \"c\", \"d\", \"d\"), c(\"d\", \"z\", \"b\", \"z\")) layer <- layer_string_lookup(vocabulary = vocab, output_mode = 'count') layer(data) ## tf.Tensor( ## [[0 1 0 1 2] ##  [2 0 1 0 1]], shape=(2, 5), dtype=int64) vocab <- c(\"a\", \"b\", \"c\", \"d\") idf_weights <- c(0.25, 0.75, 0.6, 0.4) data <- rbind(c(\"a\", \"c\", \"d\", \"d\"), c(\"d\", \"z\", \"b\", \"z\")) layer <- layer_string_lookup(output_mode = \"tf_idf\") layer %>% set_vocabulary(vocab, idf_weights=idf_weights) layer(data) ## tf.Tensor( ## [[0.   0.25 0.   0.6  0.8 ] ##  [1.   0.   0.75 0.   0.4 ]], shape=(2, 5), dtype=float32) vocab <- c(\"[UNK]\", \"a\", \"b\", \"c\", \"d\") idf_weights <- c(0.9, 0.25, 0.75, 0.6, 0.4) data <- rbind(c(\"a\", \"c\", \"d\", \"d\"), c(\"d\", \"z\", \"b\", \"z\")) layer <- layer_string_lookup(output_mode = \"tf_idf\") layer %>% set_vocabulary(vocab, idf_weights=idf_weights) layer(data) ## tf.Tensor( ## [[0.   0.25 0.   0.6  0.8 ] ##  [1.8  0.   0.75 0.   0.4 ]], shape=(2, 5), dtype=float32) vocab <- c(\"a\", \"b\", \"c\", \"d\") data <- rbind(c(1, 3, 4), c(4, 0, 2)) layer <- layer_string_lookup(vocabulary = vocab, invert = TRUE) layer(data) ## tf.Tensor( ## [[b'a' b'c' b'd'] ##  [b'd' b'[UNK]' b'b']], shape=(2, 3), dtype=string) vocab <- c(\"a\", \"b\", \"c\", \"d\") data <- rbind(c(\"a\", \"c\", \"d\"), c(\"d\", \"z\", \"b\")) layer <- layer_string_lookup(vocabulary = vocab) i_layer <- layer_string_lookup(vocabulary = vocab, invert = TRUE) int_data <- layer(data) i_layer(int_data) ## tf.Tensor( ## [[b'a' b'c' b'd'] ##  [b'd' b'[UNK]' b'b']], shape=(2, 3), dtype=string)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_subtract.html","id":null,"dir":"Reference","previous_headings":"","what":"Performs elementwise subtraction. — layer_subtract","title":"Performs elementwise subtraction. — layer_subtract","text":"takes input list tensors size 2 shape, returns single tensor (inputs[0] - inputs[1)) shape.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_subtract.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Performs elementwise subtraction. — layer_subtract","text":"","code":"layer_subtract(inputs, ...)"},{"path":"https://keras3.posit.co/dev/reference/layer_subtract.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Performs elementwise subtraction. — layer_subtract","text":"inputs layers combine ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_subtract.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Performs elementwise subtraction. — layer_subtract","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_subtract.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Performs elementwise subtraction. — layer_subtract","text":"Usage Keras model:","code":"input_shape <- c(2, 3, 4) x1 <- random_uniform(input_shape) x2 <- random_uniform(input_shape) y <- layer_subtract(list(x1, x2)) input1 <- layer_input(shape = 16) x1 <- layer_dense(input1, units = 8, activation = 'relu') input2 <- layer_input(shape = 32) x2 <- layer_dense(input2, units = 8, activation = 'relu') subtracted <- layer_subtract(list(x1, x2)) out <- layer_dense(subtracted, units = 4) model <- keras_model(inputs = list(input1, input2), outputs = out)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_text_vectorization.html","id":null,"dir":"Reference","previous_headings":"","what":"A preprocessing layer which maps text features to integer sequences. — layer_text_vectorization","title":"A preprocessing layer which maps text features to integer sequences. — layer_text_vectorization","text":"layer basic options managing text Keras model. transforms batch strings (one example = one string) either list token indices (one example = 1D tensor integer token indices) dense representation (one example = 1D tensor float values representing data example's tokens). layer meant handle natural language inputs. handle simple string inputs (categorical strings pre-tokenized strings) see layer_string_lookup(). vocabulary layer must either supplied construction learned via adapt(). layer adapted, analyze dataset, determine frequency individual string values, create vocabulary . vocabulary can unlimited size capped, depending configuration options layer; unique values input maximum vocabulary size, frequent terms used create vocabulary. processing example contains following steps: Standardize example (usually lowercasing + punctuation stripping) Split example substrings (usually words) Recombine substrings tokens (usually ngrams) Index tokens (associate unique int value token) Transform example using index, either vector ints dense float vector. notes passing callables customize splitting normalization layer: callable can passed Layer, want serialize object pass functions registered Keras serializables (see register_keras_serializable() details). using custom callable standardize, data received callable exactly passed layer. callable return tensor shape input. using custom callable split, data received callable 1st dimension squeezed - instead list(\"string split\", \"another string split\"), Callable see c(\"string split\", \"another string split\"). callable return tf.Tensor dtype string first dimension containing split tokens - example, see something like list(c(\"string\", \"\", \"split\"), c(\"another\", \"string\", \"\", \"split\")). Note: layer uses TensorFlow internally. used part compiled computation graph model backend TensorFlow. can however used backend running eagerly. can also always used part input preprocessing pipeline backend (outside model ), recommend use layer. Note: layer safe use inside tf.data pipeline (independently backend using).","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_text_vectorization.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"A preprocessing layer which maps text features to integer sequences. — layer_text_vectorization","text":"","code":"layer_text_vectorization(   object,   max_tokens = NULL,   standardize = \"lower_and_strip_punctuation\",   split = \"whitespace\",   ngrams = NULL,   output_mode = \"int\",   output_sequence_length = NULL,   pad_to_max_tokens = FALSE,   vocabulary = NULL,   idf_weights = NULL,   sparse = FALSE,   ragged = FALSE,   encoding = \"utf-8\",   name = NULL,   ... )  get_vocabulary(object, include_special_tokens = TRUE)  set_vocabulary(object, vocabulary, idf_weights = NULL, ...)"},{"path":"https://keras3.posit.co/dev/reference/layer_text_vectorization.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"A preprocessing layer which maps text features to integer sequences. — layer_text_vectorization","text":"object Object compose layer . tensor, array, sequential model. max_tokens Maximum size vocabulary layer. specified adapting vocabulary setting pad_to_max_tokens=TRUE. Note vocabulary contains 1 OOV token, effective number tokens (max_tokens - 1 - (1 output_mode == \"int\" else 0)). standardize Optional specification standardization apply input text. Values can : NULL: standardization. \"lower_and_strip_punctuation\": Text lowercased punctuation removed. \"lower\": Text lowercased. \"strip_punctuation\": punctuation removed. Callable: Inputs passed callable function, standardized returned. split Optional specification splitting input text. Values can : NULL: splitting. \"whitespace\": Split whitespace. \"character\": Split unicode character. Callable: Standardized inputs passed callable function, split returned. ngrams Optional specification ngrams create possibly-split input text. Values can NULL, integer list integers; passing integer create ngrams integer, passing list integers create ngrams specified values list. Passing NULL means ngrams created. output_mode Optional specification output layer. Values can \"int\", \"multi_hot\", \"count\" \"tf_idf\", configuring layer follows: \"int\": Outputs integer indices, one integer index per split string token. output_mode == \"int\", 0 reserved masked locations; reduces vocab size max_tokens - 2 instead max_tokens - 1. \"multi_hot\": Outputs single int array per batch, either vocab_size max_tokens size, containing 1s elements token mapped index exists least batch item. \"count\": Like \"multi_hot\", int array contains count number times token index appeared batch item. \"tf_idf\": Like \"multi_hot\", TF-IDF algorithm applied find value token slot. \"int\" output, shape input output supported. output modes, currently rank 1 inputs (rank 2 outputs splitting) supported. output_sequence_length valid INT mode. set, output time dimension padded truncated exactly output_sequence_length values, resulting tensor shape (batch_size, output_sequence_length) regardless many tokens resulted splitting step. Defaults NULL. ragged TRUE output_sequence_length may still truncate output. pad_to_max_tokens valid  \"multi_hot\", \"count\", \"tf_idf\" modes. TRUE, output feature axis padded max_tokens even number unique tokens vocabulary less max_tokens, resulting tensor shape (batch_size, max_tokens) regardless vocabulary size. Defaults FALSE. vocabulary Optional. Either array strings string path text file. passing array, can pass list, list, 1D NumPy array, 1D tensor containing string vocabulary terms. passing file path, file contain one line per term vocabulary. argument set, need adapt() layer. idf_weights R vector, 1D numpy array, 1D tensor inverse document frequency weights equal length vocabulary. Must set output_mode \"tf_idf\". set otherwise. sparse Boolean. applicable \"multi_hot\", \"count\", \"tf_idf\" output modes. supported TensorFlow backend. TRUE, returns SparseTensor instead dense Tensor. Defaults FALSE. ragged Boolean. applicable \"int\" output mode. supported TensorFlow backend. TRUE, returns RaggedTensor instead dense Tensor, sequence may different length string splitting. Defaults FALSE. encoding Optional. text encoding use interpret input strings. Defaults \"utf-8\". name String, name object ... forward/backward compatability. include_special_tokens TRUE, returned vocabulary include padding OOV tokens, term's index vocabulary equal term's index calling layer. FALSE, returned vocabulary include padding OOV tokens.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_text_vectorization.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"A preprocessing layer which maps text features to integer sequences. — layer_text_vectorization","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_text_vectorization.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"A preprocessing layer which maps text features to integer sequences. — layer_text_vectorization","text":"example instantiates TextVectorization layer lowercases text, splits whitespace, strips punctuation, outputs integer vocab indices.         example instantiates TextVectorization layer passing list vocabulary terms layer's initialize() method.","code":"max_tokens <- 5000  # Maximum vocab size. max_len <- 4  # Sequence length to pad the outputs to. # Create the layer. vectorize_layer <- layer_text_vectorization(     max_tokens = max_tokens,     output_mode = 'int',     output_sequence_length = max_len) # Now that the vocab layer has been created, call `adapt` on the # list of strings to create the vocabulary. vectorize_layer %>% adapt(c(\"foo bar\", \"bar baz\", \"baz bada boom\")) # Now, the layer can map strings to integers -- you can use an # embedding layer to map these integers to learned embeddings. input_data <- rbind(\"foo qux bar\", \"qux baz\") vectorize_layer(input_data) ## tf.Tensor( ## [[4 1 3 0] ##  [1 2 0 0]], shape=(2, 4), dtype=int64) vocab_data <- c(\"earth\", \"wind\", \"and\", \"fire\") max_len <- 4  # Sequence length to pad the outputs to. # Create the layer, passing the vocab directly. You can also pass the # vocabulary arg a path to a file containing one vocabulary word per # line. vectorize_layer <- layer_text_vectorization(     max_tokens = max_tokens,     output_mode = 'int',     output_sequence_length = max_len,     vocabulary = vocab_data) # Because we've passed the vocabulary directly, we don't need to adapt # the layer - the vocabulary is already set. The vocabulary contains the # padding token ('') and OOV token ('[UNK]') # as well as the passed tokens. vectorize_layer %>% get_vocabulary() ## [1] \"\"      \"[UNK]\" \"earth\" \"wind\"  \"and\"   \"fire\" # ['', '[UNK]', 'earth', 'wind', 'and', 'fire']"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_tfsm.html","id":null,"dir":"Reference","previous_headings":"","what":"Reload a Keras model/layer that was saved via export_savedmodel(). — layer_tfsm","title":"Reload a Keras model/layer that was saved via export_savedmodel(). — layer_tfsm","text":"Reload Keras model/layer saved via export_savedmodel().","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_tfsm.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Reload a Keras model/layer that was saved via export_savedmodel(). — layer_tfsm","text":"","code":"layer_tfsm(   object,   filepath,   call_endpoint = \"serve\",   call_training_endpoint = NULL,   trainable = TRUE,   name = NULL,   dtype = NULL )"},{"path":"https://keras3.posit.co/dev/reference/layer_tfsm.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Reload a Keras model/layer that was saved via export_savedmodel(). — layer_tfsm","text":"object Object compose layer . tensor, array, sequential model. filepath string, path SavedModel. call_endpoint Name endpoint use call() method reloaded layer. SavedModel created via export_savedmodel(), default endpoint name 'serve'. cases may named 'serving_default'. call_training_endpoint see description trainable see description name String, name object dtype datatype (e.g., \"float32\").","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_tfsm.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Reload a Keras model/layer that was saved via export_savedmodel(). — layer_tfsm","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_tfsm.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Reload a Keras model/layer that was saved via export_savedmodel(). — layer_tfsm","text":"reloaded object can used like regular Keras layer, supports training/fine-tuning trainable weights. Note reloaded object retains none internal structure custom methods original object – brand new layer created around saved function. Limitations: call endpoints single inputs tensor argument (may optionally named list/list tensors) supported. endpoints multiple separate input tensor arguments, consider subclassing layer_tfsm implementing call() method custom signature. need training-time behavior differ inference-time behavior (.e. need reloaded object support training=TRUE argument __call__()), make sure training-time call function saved standalone endpoint artifact, provide name layer_tfsm via call_training_endpoint argument.","code":"model <- keras_model_sequential(input_shape = c(784)) |> layer_dense(10) model |> export_savedmodel(\"path/to/artifact\") ## Saved artifact at 'path/to/artifact'. The following endpoints are available: ## ## * Endpoint 'serve' ##   args_0 (POSITIONAL_ONLY): TensorSpec(shape=(None, 784), dtype=tf.float32, name='keras_tensor') ## Output Type: ##   TensorSpec(shape=(None, 10), dtype=tf.float32, name=None) ## Captures: ##   131671756274576: TensorSpec(shape=(), dtype=tf.resource, name=None) ##   131671756265936: TensorSpec(shape=(), dtype=tf.resource, name=None) reloaded_layer <- layer_tfsm(filepath = \"path/to/artifact\") input <- random_normal(c(2, 784)) output <- reloaded_layer(input) stopifnot(all.equal(as.array(output), as.array(model(input))))"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_time_distributed.html","id":null,"dir":"Reference","previous_headings":"","what":"This wrapper allows to apply a layer to every temporal slice of an input. — layer_time_distributed","title":"This wrapper allows to apply a layer to every temporal slice of an input. — layer_time_distributed","text":"Every input least 3D, dimension index one first input considered temporal dimension. Consider batch 32 video samples, sample 128x128 RGB image channels_last data format, across 10 timesteps. batch input shape (32, 10, 128, 128, 3). can use TimeDistributed apply Conv2D layer 10 timesteps, independently:     layer_time_distributed applies instance layer_conv2d timestamps, set weights used timestamp.","code":"inputs <- keras_input(shape = c(10, 128, 128, 3), batch_size = 32) conv_2d_layer <- layer_conv_2d(filters = 64, kernel_size = c(3, 3)) outputs <- layer_time_distributed(inputs, layer = conv_2d_layer) shape(outputs) ## shape(32, 10, 126, 126, 64)"},{"path":"https://keras3.posit.co/dev/reference/layer_time_distributed.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"This wrapper allows to apply a layer to every temporal slice of an input. — layer_time_distributed","text":"","code":"layer_time_distributed(object, layer, ...)"},{"path":"https://keras3.posit.co/dev/reference/layer_time_distributed.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"This wrapper allows to apply a layer to every temporal slice of an input. — layer_time_distributed","text":"object Object compose layer . tensor, array, sequential model. layer Layer instance. ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_time_distributed.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"This wrapper allows to apply a layer to every temporal slice of an input. — layer_time_distributed","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_time_distributed.html","id":"call-arguments","dir":"Reference","previous_headings":"","what":"Call Arguments","title":"This wrapper allows to apply a layer to every temporal slice of an input. — layer_time_distributed","text":"inputs: Input tensor shape (batch, time, ...) nested tensors, shape (batch, time, ...). training: Boolean indicating whether layer behave training mode inference mode. argument passed wrapped layer (layer supports argument). mask: Binary tensor shape (samples, timesteps) indicating whether given timestep masked. argument passed wrapped layer (layer supports argument).","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_torch_module_wrapper.html","id":null,"dir":"Reference","previous_headings":"","what":"Torch module wrapper layer. — layer_torch_module_wrapper","title":"Torch module wrapper layer. — layer_torch_module_wrapper","text":"layer_torch_module_wrapper wrapper class can turn torch.nn.Module Keras layer, particular making parameters trackable Keras. layer_torch_module_wrapper() compatible PyTorch backend used TensorFlow JAX backends.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_torch_module_wrapper.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Torch module wrapper layer. — layer_torch_module_wrapper","text":"","code":"layer_torch_module_wrapper(   object,   module,   output_shape = NULL,   name = NULL,   ... )"},{"path":"https://keras3.posit.co/dev/reference/layer_torch_module_wrapper.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Torch module wrapper layer. — layer_torch_module_wrapper","text":"object Object compose layer . tensor, array, sequential model. module torch.nn.Module instance. LazyModule instance, parameters must initialized passing instance layer_torch_module_wrapper (e.g. calling ). output_shape Shape output layer. Helps Keras infer shapes. name name layer (string). ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_torch_module_wrapper.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Torch module wrapper layer. — layer_torch_module_wrapper","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_torch_module_wrapper.html","id":"example","dir":"Reference","previous_headings":"","what":"Example","title":"Torch module wrapper layer. — layer_torch_module_wrapper","text":"example layer_torch_module_wrapper() can used vanilla PyTorch modules.","code":"# reticulate::py_install( #   packages = c(\"torch\", \"torchvision\", \"torchaudio\"), #   envname = \"r-keras\", #   pip_options = c(\"--index-url https://download.pytorch.org/whl/cpu\") # ) library(keras3) use_backend(\"torch\") torch <- reticulate::import(\"torch\") nn <- reticulate::import(\"torch.nn\") nnf <- reticulate::import(\"torch.nn.functional\")  Classifier(keras$Model) \\%py_class\\% {   initialize <- function(...) {     super$initialize(...)      self$conv1 <- layer_torch_module_wrapper(module = nn$Conv2d(       in_channels = 1L,       out_channels = 32L,       kernel_size = tuple(3L, 3L)     ))     self$conv2 <- layer_torch_module_wrapper(module = nn$Conv2d(       in_channels = 32L,       out_channels = 64L,       kernel_size = tuple(3L, 3L)     ))     self$pool <- nn$MaxPool2d(kernel_size = tuple(2L, 2L))     self$flatten <- nn$Flatten()     self$dropout <- nn$Dropout(p = 0.5)     self$fc <-       layer_torch_module_wrapper(module = nn$Linear(1600L, 10L))   }    call <- function(inputs) {     x <- nnf$relu(self$conv1(inputs))     x <- self$pool(x)     x <- nnf$relu(self$conv2(x))     x <- self$pool(x)     x <- self$flatten(x)     x <- self$dropout(x)     x <- self$fc(x)     nnf$softmax(x, dim = 1L)   } } model <- Classifier() model$build(shape(1, 28, 28)) cat(\"Output shape:\", format(shape(model(torch$ones(1L, 1L, 28L, 28L)))))  model |> compile(loss = \"sparse_categorical_crossentropy\",                  optimizer = \"adam\",                  metrics = \"accuracy\") model |> fit(train_loader, epochs = 5)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_unit_normalization.html","id":null,"dir":"Reference","previous_headings":"","what":"Unit normalization layer. — layer_unit_normalization","title":"Unit normalization layer. — layer_unit_normalization","text":"Normalize batch inputs input batch L2 norm equal 1 (across axes specified axis).","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_unit_normalization.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Unit normalization layer. — layer_unit_normalization","text":"","code":"layer_unit_normalization(object, axis = -1L, ...)"},{"path":"https://keras3.posit.co/dev/reference/layer_unit_normalization.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Unit normalization layer. — layer_unit_normalization","text":"object Object compose layer . tensor, array, sequential model. axis Integer list. axis axes normalize across. Typically, features axis axes. left-axes typically batch axis axes. -1 last dimension input. Defaults -1. ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_unit_normalization.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Unit normalization layer. — layer_unit_normalization","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_unit_normalization.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Unit normalization layer. — layer_unit_normalization","text":"","code":"data <- 1:6 |> op_array(\"float32\") |> op_reshape(c(2, 3)) normalized_data <- layer_unit_normalization(data) op_sum(normalized_data[1, ]^2) ## tf.Tensor(0.9999999, shape=(), dtype=float32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_upsampling_1d.html","id":null,"dir":"Reference","previous_headings":"","what":"Upsampling layer for 1D inputs. — layer_upsampling_1d","title":"Upsampling layer for 1D inputs. — layer_upsampling_1d","text":"Repeats temporal step size times along time axis.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_upsampling_1d.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Upsampling layer for 1D inputs. — layer_upsampling_1d","text":"","code":"layer_upsampling_1d(object, size = 2L, ...)"},{"path":"https://keras3.posit.co/dev/reference/layer_upsampling_1d.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Upsampling layer for 1D inputs. — layer_upsampling_1d","text":"object Object compose layer . tensor, array, sequential model. size Integer. Upsampling factor. ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_upsampling_1d.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Upsampling layer for 1D inputs. — layer_upsampling_1d","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_upsampling_1d.html","id":"example","dir":"Reference","previous_headings":"","what":"Example","title":"Upsampling layer for 1D inputs. — layer_upsampling_1d","text":"","code":"input_shape <- c(2, 2, 3) x <- seq_len(prod(input_shape)) %>% op_reshape(input_shape) x ## tf.Tensor( ## [[[ 1  2  3] ##   [ 4  5  6]] ## ##  [[ 7  8  9] ##   [10 11 12]]], shape=(2, 2, 3), dtype=int32) y <- layer_upsampling_1d(x, size = 2) y ## tf.Tensor( ## [[[ 1  2  3] ##   [ 1  2  3] ##   [ 4  5  6] ##   [ 4  5  6]] ## ##  [[ 7  8  9] ##   [ 7  8  9] ##   [10 11 12] ##   [10 11 12]]], shape=(2, 4, 3), dtype=int32)"},{"path":"https://keras3.posit.co/dev/reference/layer_upsampling_1d.html","id":"input-shape","dir":"Reference","previous_headings":"","what":"Input Shape","title":"Upsampling layer for 1D inputs. — layer_upsampling_1d","text":"3D tensor shape: (batch_size, steps, features).","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_upsampling_1d.html","id":"output-shape","dir":"Reference","previous_headings":"","what":"Output Shape","title":"Upsampling layer for 1D inputs. — layer_upsampling_1d","text":"3D tensor shape: (batch_size, upsampled_steps, features).","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_upsampling_2d.html","id":null,"dir":"Reference","previous_headings":"","what":"Upsampling layer for 2D inputs. — layer_upsampling_2d","title":"Upsampling layer for 2D inputs. — layer_upsampling_2d","text":"implementation uses interpolative resizing, given resize method (specified interpolation argument). Use interpolation=nearest repeat rows columns data.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_upsampling_2d.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Upsampling layer for 2D inputs. — layer_upsampling_2d","text":"","code":"layer_upsampling_2d(   object,   size = list(2L, 2L),   data_format = NULL,   interpolation = \"nearest\",   ... )"},{"path":"https://keras3.posit.co/dev/reference/layer_upsampling_2d.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Upsampling layer for 2D inputs. — layer_upsampling_2d","text":"object Object compose layer . tensor, array, sequential model. size Int, list 2 integers. upsampling factors rows columns. data_format string, one \"channels_last\" (default) \"channels_first\". ordering dimensions inputs. \"channels_last\" corresponds inputs shape (batch_size, height, width, channels) \"channels_first\" corresponds inputs shape (batch_size, channels, height, width). unspecified, uses image_data_format value found Keras config file ~/.keras/keras.json (exists) else \"channels_last\". Defaults \"channels_last\". interpolation string, one \"bicubic\", \"bilinear\", \"lanczos3\", \"lanczos5\", \"nearest\". ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_upsampling_2d.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Upsampling layer for 2D inputs. — layer_upsampling_2d","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_upsampling_2d.html","id":"example","dir":"Reference","previous_headings":"","what":"Example","title":"Upsampling layer for 2D inputs. — layer_upsampling_2d","text":"","code":"input_shape <- c(2, 2, 1, 3) x <- op_reshape(seq_len(prod(input_shape)), input_shape) print(x) ## tf.Tensor( ## [[[[ 1  2  3]] ## ##   [[ 4  5  6]]] ## ## ##  [[[ 7  8  9]] ## ##   [[10 11 12]]]], shape=(2, 2, 1, 3), dtype=int32) y <- layer_upsampling_2d(x, size = c(1, 2)) print(y) ## tf.Tensor( ## [[[[ 1  2  3] ##    [ 1  2  3]] ## ##   [[ 4  5  6] ##    [ 4  5  6]]] ## ## ##  [[[ 7  8  9] ##    [ 7  8  9]] ## ##   [[10 11 12] ##    [10 11 12]]]], shape=(2, 2, 2, 3), dtype=int32)"},{"path":"https://keras3.posit.co/dev/reference/layer_upsampling_2d.html","id":"input-shape","dir":"Reference","previous_headings":"","what":"Input Shape","title":"Upsampling layer for 2D inputs. — layer_upsampling_2d","text":"4D tensor shape: data_format \"channels_last\": (batch_size, rows, cols, channels) data_format \"channels_first\": (batch_size, channels, rows, cols)","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_upsampling_2d.html","id":"output-shape","dir":"Reference","previous_headings":"","what":"Output Shape","title":"Upsampling layer for 2D inputs. — layer_upsampling_2d","text":"4D tensor shape: data_format \"channels_last\": (batch_size, upsampled_rows, upsampled_cols, channels) data_format \"channels_first\": (batch_size, channels, upsampled_rows, upsampled_cols)","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_upsampling_3d.html","id":null,"dir":"Reference","previous_headings":"","what":"Upsampling layer for 3D inputs. — layer_upsampling_3d","title":"Upsampling layer for 3D inputs. — layer_upsampling_3d","text":"Repeats 1st, 2nd 3rd dimensions data size[0], size[1] size[2] respectively.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_upsampling_3d.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Upsampling layer for 3D inputs. — layer_upsampling_3d","text":"","code":"layer_upsampling_3d(object, size = list(2L, 2L, 2L), data_format = NULL, ...)"},{"path":"https://keras3.posit.co/dev/reference/layer_upsampling_3d.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Upsampling layer for 3D inputs. — layer_upsampling_3d","text":"object Object compose layer . tensor, array, sequential model. size Int, list 3 integers. upsampling factors dim1, dim2 dim3. data_format string, one \"channels_last\" (default) \"channels_first\". ordering dimensions inputs. \"channels_last\" corresponds inputs shape (batch_size, spatial_dim1, spatial_dim2, spatial_dim3, channels) \"channels_first\" corresponds inputs shape (batch_size, channels, spatial_dim1, spatial_dim2, spatial_dim3). unspecified, uses image_data_format value found Keras config file ~/.keras/keras.json (exists) else \"channels_last\". Defaults \"channels_last\". ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_upsampling_3d.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Upsampling layer for 3D inputs. — layer_upsampling_3d","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_upsampling_3d.html","id":"example","dir":"Reference","previous_headings":"","what":"Example","title":"Upsampling layer for 3D inputs. — layer_upsampling_3d","text":"","code":"input_shape <- c(2, 1, 2, 1, 3) x <- array(1, dim = input_shape) y <- layer_upsampling_3d(x, size = c(2, 2, 2)) shape(y) ## shape(2, 2, 4, 2, 3)"},{"path":"https://keras3.posit.co/dev/reference/layer_upsampling_3d.html","id":"input-shape","dir":"Reference","previous_headings":"","what":"Input Shape","title":"Upsampling layer for 3D inputs. — layer_upsampling_3d","text":"5D tensor shape: data_format \"channels_last\": (batch_size, dim1, dim2, dim3, channels) data_format \"channels_first\": (batch_size, channels, dim1, dim2, dim3)","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_upsampling_3d.html","id":"output-shape","dir":"Reference","previous_headings":"","what":"Output Shape","title":"Upsampling layer for 3D inputs. — layer_upsampling_3d","text":"5D tensor shape: data_format \"channels_last\": (batch_size, upsampled_dim1, upsampled_dim2, upsampled_dim3, channels) data_format \"channels_first\": (batch_size, channels, upsampled_dim1, upsampled_dim2, upsampled_dim3)","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_zero_padding_1d.html","id":null,"dir":"Reference","previous_headings":"","what":"Zero-padding layer for 1D input (e.g. temporal sequence). — layer_zero_padding_1d","title":"Zero-padding layer for 1D input (e.g. temporal sequence). — layer_zero_padding_1d","text":"Zero-padding layer 1D input (e.g. temporal sequence).","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_zero_padding_1d.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Zero-padding layer for 1D input (e.g. temporal sequence). — layer_zero_padding_1d","text":"","code":"layer_zero_padding_1d(object, padding = 1L, data_format = NULL, ...)"},{"path":"https://keras3.posit.co/dev/reference/layer_zero_padding_1d.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Zero-padding layer for 1D input (e.g. temporal sequence). — layer_zero_padding_1d","text":"object Object compose layer . tensor, array, sequential model. padding Int, list int (length 2). int: many zeros add beginning end padding dimension (axis 2). list 2 ints: many zeros add beginning end padding dimension ((left_pad, right_pad)). data_format string, one \"channels_last\" (default) \"channels_first\". ordering dimensions inputs. \"channels_last\" corresponds inputs shape (batch_size, axis_to_pad, channels) \"channels_first\" corresponds inputs shape (batch_size, channels, axis_to_pad). unspecified, uses image_data_format value found Keras config file ~/.keras/keras.json (exists). Defaults \"channels_last\". ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_zero_padding_1d.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Zero-padding layer for 1D input (e.g. temporal sequence). — layer_zero_padding_1d","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_zero_padding_1d.html","id":"example","dir":"Reference","previous_headings":"","what":"Example","title":"Zero-padding layer for 1D input (e.g. temporal sequence). — layer_zero_padding_1d","text":"","code":"input_shape <- c(2, 2, 3) x <- op_reshape(seq_len(prod(input_shape)), input_shape) x ## tf.Tensor( ## [[[ 1  2  3] ##   [ 4  5  6]] ## ##  [[ 7  8  9] ##   [10 11 12]]], shape=(2, 2, 3), dtype=int32) y <- layer_zero_padding_1d(x, padding = 2) y ## tf.Tensor( ## [[[ 0  0  0] ##   [ 0  0  0] ##   [ 1  2  3] ##   [ 4  5  6] ##   [ 0  0  0] ##   [ 0  0  0]] ## ##  [[ 0  0  0] ##   [ 0  0  0] ##   [ 7  8  9] ##   [10 11 12] ##   [ 0  0  0] ##   [ 0  0  0]]], shape=(2, 6, 3), dtype=int32)"},{"path":"https://keras3.posit.co/dev/reference/layer_zero_padding_1d.html","id":"input-shape","dir":"Reference","previous_headings":"","what":"Input Shape","title":"Zero-padding layer for 1D input (e.g. temporal sequence). — layer_zero_padding_1d","text":"3D tensor shape: data_format \"channels_last\": (batch_size, axis_to_pad, features) data_format \"channels_first\": (batch_size, features, axis_to_pad)","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_zero_padding_1d.html","id":"output-shape","dir":"Reference","previous_headings":"","what":"Output Shape","title":"Zero-padding layer for 1D input (e.g. temporal sequence). — layer_zero_padding_1d","text":"3D tensor shape: data_format \"channels_last\": (batch_size, padded_axis, features) data_format \"channels_first\": (batch_size, features, padded_axis)","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_zero_padding_2d.html","id":null,"dir":"Reference","previous_headings":"","what":"Zero-padding layer for 2D input (e.g. picture). — layer_zero_padding_2d","title":"Zero-padding layer for 2D input (e.g. picture). — layer_zero_padding_2d","text":"layer can add rows columns zeros top, bottom, left right side image tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_zero_padding_2d.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Zero-padding layer for 2D input (e.g. picture). — layer_zero_padding_2d","text":"","code":"layer_zero_padding_2d(object, padding = list(1L, 1L), data_format = NULL, ...)"},{"path":"https://keras3.posit.co/dev/reference/layer_zero_padding_2d.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Zero-padding layer for 2D input (e.g. picture). — layer_zero_padding_2d","text":"object Object compose layer . tensor, array, sequential model. padding Int, list 2 ints, list 2 lists 2 ints. int: symmetric padding applied height width. list 2 ints: interpreted two different symmetric padding values height width: (symmetric_height_pad, symmetric_width_pad). list 2 lists 2 ints: interpreted ((top_pad, bottom_pad), (left_pad, right_pad)). data_format string, one \"channels_last\" (default) \"channels_first\". ordering dimensions inputs. \"channels_last\" corresponds inputs shape (batch_size, height, width, channels) \"channels_first\" corresponds inputs shape (batch_size, channels, height, width). unspecified, uses image_data_format value found Keras config file ~/.keras/keras.json (exists). Defaults \"channels_last\". ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_zero_padding_2d.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Zero-padding layer for 2D input (e.g. picture). — layer_zero_padding_2d","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_zero_padding_2d.html","id":"example","dir":"Reference","previous_headings":"","what":"Example","title":"Zero-padding layer for 2D input (e.g. picture). — layer_zero_padding_2d","text":"","code":"input_shape <- c(1, 1, 2, 2) x <- op_reshape(seq_len(prod(input_shape)), input_shape) x ## tf.Tensor( ## [[[[1 2] ##    [3 4]]]], shape=(1, 1, 2, 2), dtype=int32) y <- layer_zero_padding_2d(x, padding = 1) y ## tf.Tensor( ## [[[[0 0] ##    [0 0] ##    [0 0] ##    [0 0]] ## ##   [[0 0] ##    [1 2] ##    [3 4] ##    [0 0]] ## ##   [[0 0] ##    [0 0] ##    [0 0] ##    [0 0]]]], shape=(1, 3, 4, 2), dtype=int32)"},{"path":"https://keras3.posit.co/dev/reference/layer_zero_padding_2d.html","id":"input-shape","dir":"Reference","previous_headings":"","what":"Input Shape","title":"Zero-padding layer for 2D input (e.g. picture). — layer_zero_padding_2d","text":"4D tensor shape: data_format \"channels_last\": (batch_size, height, width, channels) data_format \"channels_first\": (batch_size, channels, height, width)","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_zero_padding_2d.html","id":"output-shape","dir":"Reference","previous_headings":"","what":"Output Shape","title":"Zero-padding layer for 2D input (e.g. picture). — layer_zero_padding_2d","text":"4D tensor shape: data_format \"channels_last\": (batch_size, padded_height, padded_width, channels) data_format \"channels_first\": (batch_size, channels, padded_height, padded_width)","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_zero_padding_3d.html","id":null,"dir":"Reference","previous_headings":"","what":"Zero-padding layer for 3D data (spatial or spatio-temporal). — layer_zero_padding_3d","title":"Zero-padding layer for 3D data (spatial or spatio-temporal). — layer_zero_padding_3d","text":"Zero-padding layer 3D data (spatial spatio-temporal).","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_zero_padding_3d.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Zero-padding layer for 3D data (spatial or spatio-temporal). — layer_zero_padding_3d","text":"","code":"layer_zero_padding_3d(   object,   padding = list(list(1L, 1L), list(1L, 1L), list(1L, 1L)),   data_format = NULL,   ... )"},{"path":"https://keras3.posit.co/dev/reference/layer_zero_padding_3d.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Zero-padding layer for 3D data (spatial or spatio-temporal). — layer_zero_padding_3d","text":"object Object compose layer . tensor, array, sequential model. padding Int, list 3 ints, list 3 lists 2 ints. int: symmetric padding applied depth, height, width. list 3 ints: interpreted three different symmetric padding values depth, height, width: (symmetric_dim1_pad, symmetric_dim2_pad, symmetric_dim3_pad). list 3 lists 2 ints: interpreted ((left_dim1_pad, right_dim1_pad), (left_dim2_pad, right_dim2_pad), (left_dim3_pad, right_dim3_pad)). data_format string, one \"channels_last\" (default) \"channels_first\". ordering dimensions inputs. \"channels_last\" corresponds inputs shape (batch_size, spatial_dim1, spatial_dim2, spatial_dim3, channels) \"channels_first\" corresponds inputs shape (batch_size, channels, spatial_dim1, spatial_dim2, spatial_dim3). unspecified, uses image_data_format value found Keras config file ~/.keras/keras.json (exists). Defaults \"channels_last\". ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_zero_padding_3d.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Zero-padding layer for 3D data (spatial or spatio-temporal). — layer_zero_padding_3d","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_zero_padding_3d.html","id":"example","dir":"Reference","previous_headings":"","what":"Example","title":"Zero-padding layer for 3D data (spatial or spatio-temporal). — layer_zero_padding_3d","text":"","code":"input_shape <- c(1, 1, 2, 2, 3) x <- op_reshape(seq_len(prod(input_shape)), input_shape) x ## tf.Tensor( ## [[[[[ 1  2  3] ##     [ 4  5  6]] ## ##    [[ 7  8  9] ##     [10 11 12]]]]], shape=(1, 1, 2, 2, 3), dtype=int32) y <- layer_zero_padding_3d(x, padding = 2) shape(y) ## shape(1, 5, 6, 6, 3)"},{"path":"https://keras3.posit.co/dev/reference/layer_zero_padding_3d.html","id":"input-shape","dir":"Reference","previous_headings":"","what":"Input Shape","title":"Zero-padding layer for 3D data (spatial or spatio-temporal). — layer_zero_padding_3d","text":"5D tensor shape: data_format \"channels_last\": (batch_size, first_axis_to_pad, second_axis_to_pad, third_axis_to_pad, depth) data_format \"channels_first\": (batch_size, depth, first_axis_to_pad, second_axis_to_pad, third_axis_to_pad)","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_zero_padding_3d.html","id":"output-shape","dir":"Reference","previous_headings":"","what":"Output Shape","title":"Zero-padding layer for 3D data (spatial or spatio-temporal). — layer_zero_padding_3d","text":"5D tensor shape: data_format \"channels_last\": (batch_size, first_padded_axis, second_padded_axis, third_axis_to_pad, depth) data_format \"channels_first\": (batch_size, depth, first_padded_axis, second_padded_axis, third_axis_to_pad)","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/learning_rate_schedule_cosine_decay.html","id":null,"dir":"Reference","previous_headings":"","what":"A LearningRateSchedule that uses a cosine decay with optional warmup. — learning_rate_schedule_cosine_decay","title":"A LearningRateSchedule that uses a cosine decay with optional warmup. — learning_rate_schedule_cosine_decay","text":"See Loshchilov & Hutter, ICLR2016, SGDR: Stochastic Gradient Descent Warm Restarts. idea linear warmup learning rate, see Goyal et al.. begin training model, often want initial increase learning rate followed decay. warmup_target int, schedule applies linear increase per optimizer step learning rate initial_learning_rate warmup_target duration warmup_steps. Afterwards, applies cosine decay function taking learning rate warmup_target alpha duration decay_steps. warmup_target NULL skip warmup decay take learning rate initial_learning_rate alpha. requires step value  compute learning rate. can just pass backend variable increment training step. schedule 1-arg callable produces warmup followed decayed learning rate passed current optimizer step. can useful changing learning rate value across different invocations optimizer functions. warmup computed :   decay computed :   Example usage without warmup:   Example usage warmup:   can pass schedule directly optimizer learning rate. learning rate schedule also serializable deserializable using keras$optimizers$schedules$serialize keras$optimizers$schedules$deserialize.","code":"warmup_learning_rate <- function(step) {   completed_fraction <- step / warmup_steps   total_delta <- target_warmup - initial_learning_rate   completed_fraction * total_delta } if (is.null(warmup_target)) {   initial_decay_lr <- initial_learning_rate } else {   initial_decay_lr <- warmup_target }  decayed_learning_rate <- function(step) {   step <- min(step, decay_steps)   cosine_decay <- 0.5 * (1 + cos(pi * step / decay_steps))   decayed <- (1 - alpha) * cosine_decay + alpha   initial_decay_lr * decayed } decay_steps <- 1000 initial_learning_rate <- 0.1 lr_decayed_fn <- learning_rate_schedule_cosine_decay(     initial_learning_rate, decay_steps) decay_steps <- 1000 initial_learning_rate <- 0 warmup_steps <- 1000 target_learning_rate <- 0.1 lr_warmup_decayed_fn <- learning_rate_schedule_cosine_decay(     initial_learning_rate, decay_steps, warmup_target = target_learning_rate,     warmup_steps = warmup_steps )"},{"path":"https://keras3.posit.co/dev/reference/learning_rate_schedule_cosine_decay.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"A LearningRateSchedule that uses a cosine decay with optional warmup. — learning_rate_schedule_cosine_decay","text":"","code":"learning_rate_schedule_cosine_decay(   initial_learning_rate,   decay_steps,   alpha = 0,   name = \"CosineDecay\",   warmup_target = NULL,   warmup_steps = 0L )"},{"path":"https://keras3.posit.co/dev/reference/learning_rate_schedule_cosine_decay.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"A LearningRateSchedule that uses a cosine decay with optional warmup. — learning_rate_schedule_cosine_decay","text":"initial_learning_rate float. initial learning rate. decay_steps int. Number steps decay . alpha float. Minimum learning rate value decay fraction initial_learning_rate. name String. Optional name operation.  Defaults \"CosineDecay\". warmup_target float. target learning rate warmup phase. cast initial_learning_rate datatype. Setting NULL skip warmup begins decay phase initial_learning_rate. Otherwise scheduler warmup initial_learning_rate warmup_target. warmup_steps int. Number steps warmup .","code":""},{"path":"https://keras3.posit.co/dev/reference/learning_rate_schedule_cosine_decay.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"A LearningRateSchedule that uses a cosine decay with optional warmup. — learning_rate_schedule_cosine_decay","text":"1-arg callable learning rate schedule takes current optimizer step outputs decayed learning rate, scalar tensor type initial_learning_rate.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/learning_rate_schedule_cosine_decay_restarts.html","id":null,"dir":"Reference","previous_headings":"","what":"A LearningRateSchedule that uses a cosine decay schedule with restarts. — learning_rate_schedule_cosine_decay_restarts","title":"A LearningRateSchedule that uses a cosine decay schedule with restarts. — learning_rate_schedule_cosine_decay_restarts","text":"See Loshchilov & Hutter, ICLR2016, SGDR: Stochastic Gradient Descent Warm Restarts. training model, often useful lower learning rate training progresses. schedule applies cosine decay function restarts optimizer step, given provided initial learning rate. requires step value compute decayed learning rate. can just pass backend variable increment training step. schedule 1-arg callable produces decayed learning rate passed current optimizer step. can useful changing learning rate value across different invocations optimizer functions. learning rate multiplier first decays 1 alpha first_decay_steps steps. , warm restart performed. new warm restart runs t_mul times steps m_mul times initial learning rate new learning rate.","code":""},{"path":"https://keras3.posit.co/dev/reference/learning_rate_schedule_cosine_decay_restarts.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"A LearningRateSchedule that uses a cosine decay schedule with restarts. — learning_rate_schedule_cosine_decay_restarts","text":"","code":"learning_rate_schedule_cosine_decay_restarts(   initial_learning_rate,   first_decay_steps,   t_mul = 2,   m_mul = 1,   alpha = 0,   name = \"SGDRDecay\" )"},{"path":"https://keras3.posit.co/dev/reference/learning_rate_schedule_cosine_decay_restarts.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"A LearningRateSchedule that uses a cosine decay schedule with restarts. — learning_rate_schedule_cosine_decay_restarts","text":"initial_learning_rate float. initial learning rate. first_decay_steps integer. Number steps decay . t_mul float. Used derive number iterations -th period. m_mul float. Used derive initial learning rate -th period. alpha float. Minimum learning rate value fraction initial_learning_rate. name String. Optional name operation. Defaults \"SGDRDecay\".","code":""},{"path":"https://keras3.posit.co/dev/reference/learning_rate_schedule_cosine_decay_restarts.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"A LearningRateSchedule that uses a cosine decay schedule with restarts. — learning_rate_schedule_cosine_decay_restarts","text":"1-arg callable learning rate schedule takes current optimizer step outputs decayed learning rate, scalar tensor type initial_learning_rate.","code":""},{"path":"https://keras3.posit.co/dev/reference/learning_rate_schedule_cosine_decay_restarts.html","id":"example","dir":"Reference","previous_headings":"","what":"Example","title":"A LearningRateSchedule that uses a cosine decay schedule with restarts. — learning_rate_schedule_cosine_decay_restarts","text":"can pass schedule directly optimizer learning rate. learning rate schedule also serializable deserializable using keras$optimizers$schedules$serialize keras$optimizers$schedules$deserialize.","code":"first_decay_steps <- 1000 lr_decayed_fn <- learning_rate_schedule_cosine_decay_restarts(         0.001,         first_decay_steps)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/learning_rate_schedule_exponential_decay.html","id":null,"dir":"Reference","previous_headings":"","what":"A LearningRateSchedule that uses an exponential decay schedule. — learning_rate_schedule_exponential_decay","title":"A LearningRateSchedule that uses an exponential decay schedule. — learning_rate_schedule_exponential_decay","text":"training model, often useful lower learning rate training progresses. schedule applies exponential decay function optimizer step, given provided initial learning rate. schedule 1-arg callable produces decayed learning rate passed current optimizer step. can useful changing learning rate value across different invocations optimizer functions. computed :   argument staircase TRUE, step / decay_steps integer division decayed learning rate follows staircase function. can pass schedule directly optimizer learning rate.","code":"decayed_learning_rate <- function(step) {   initial_learning_rate * decay_rate ^ (step / decay_steps) }"},{"path":"https://keras3.posit.co/dev/reference/learning_rate_schedule_exponential_decay.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"A LearningRateSchedule that uses an exponential decay schedule. — learning_rate_schedule_exponential_decay","text":"","code":"learning_rate_schedule_exponential_decay(   initial_learning_rate,   decay_steps,   decay_rate,   staircase = FALSE,   name = \"ExponentialDecay\" )"},{"path":"https://keras3.posit.co/dev/reference/learning_rate_schedule_exponential_decay.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"A LearningRateSchedule that uses an exponential decay schedule. — learning_rate_schedule_exponential_decay","text":"initial_learning_rate float. initial learning rate. decay_steps integer. Must positive. See decay computation . decay_rate float. decay rate. staircase Boolean.  TRUE decay learning rate discrete intervals. name String.  Optional name operation.  Defaults \"ExponentialDecay\".","code":""},{"path":"https://keras3.posit.co/dev/reference/learning_rate_schedule_exponential_decay.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"A LearningRateSchedule that uses an exponential decay schedule. — learning_rate_schedule_exponential_decay","text":"1-arg callable learning rate schedule takes current optimizer step outputs decayed learning rate, scalar tensor type initial_learning_rate.","code":""},{"path":"https://keras3.posit.co/dev/reference/learning_rate_schedule_exponential_decay.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"A LearningRateSchedule that uses an exponential decay schedule. — learning_rate_schedule_exponential_decay","text":"fitting Keras model, decay every 100000 steps base 0.96:   learning rate schedule also serializable deserializable using keras$optimizers$schedules$serialize keras$optimizers$schedules$deserialize.","code":"initial_learning_rate <- 0.1 lr_schedule <- learning_rate_schedule_exponential_decay(     initial_learning_rate,     decay_steps=100000,     decay_rate=0.96,     staircase=TRUE)  model %>% compile(   optimizer = optimizer_sgd(learning_rate = lr_schedule),   loss = 'sparse_categorical_crossentropy',   metrics = c('accuracy'))  model %>% fit(data, labels, epochs=5)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/learning_rate_schedule_inverse_time_decay.html","id":null,"dir":"Reference","previous_headings":"","what":"A LearningRateSchedule that uses an inverse time decay schedule. — learning_rate_schedule_inverse_time_decay","title":"A LearningRateSchedule that uses an inverse time decay schedule. — learning_rate_schedule_inverse_time_decay","text":"training model, often useful lower learning rate training progresses. schedule applies inverse decay function optimizer step, given provided initial learning rate. requires step value compute decayed learning rate. can just pass backend variable increment training step. schedule 1-arg callable produces decayed learning rate passed current optimizer step. can useful changing learning rate value across different invocations optimizer functions. computed :   , staircase TRUE, :   can pass schedule directly optimizer_* learning rate.","code":"decayed_learning_rate <- function(step) {   initial_learning_rate / (1 + decay_rate * step / decay_step) } decayed_learning_rate <- function(step) {   initial_learning_rate /            (1 + decay_rate * floor(step / decay_step)) }"},{"path":"https://keras3.posit.co/dev/reference/learning_rate_schedule_inverse_time_decay.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"A LearningRateSchedule that uses an inverse time decay schedule. — learning_rate_schedule_inverse_time_decay","text":"","code":"learning_rate_schedule_inverse_time_decay(   initial_learning_rate,   decay_steps,   decay_rate,   staircase = FALSE,   name = \"InverseTimeDecay\" )"},{"path":"https://keras3.posit.co/dev/reference/learning_rate_schedule_inverse_time_decay.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"A LearningRateSchedule that uses an inverse time decay schedule. — learning_rate_schedule_inverse_time_decay","text":"initial_learning_rate float. initial learning rate. decay_steps often apply decay. decay_rate number.  decay rate. staircase Whether apply decay discrete staircase, o pposed continuous, fashion. name String.  Optional name operation.  Defaults \"InverseTimeDecay\".","code":""},{"path":"https://keras3.posit.co/dev/reference/learning_rate_schedule_inverse_time_decay.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"A LearningRateSchedule that uses an inverse time decay schedule. — learning_rate_schedule_inverse_time_decay","text":"1-arg callable learning rate schedule takes current optimizer step outputs decayed learning rate, scalar tensor type initial_learning_rate.","code":""},{"path":"https://keras3.posit.co/dev/reference/learning_rate_schedule_inverse_time_decay.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"A LearningRateSchedule that uses an inverse time decay schedule. — learning_rate_schedule_inverse_time_decay","text":"Fit Keras model decaying 1/t rate 0.5:","code":"... initial_learning_rate <- 0.1 decay_steps <- 1.0 decay_rate <- 0.5 learning_rate_fn <- learning_rate_schedule_inverse_time_decay(     initial_learning_rate, decay_steps, decay_rate)  model %>% compile(   optimizer = optimizer_sgd(learning_rate=learning_rate_fn),   loss = 'sparse_categorical_crossentropy',   metrics = 'accuracy') )  model %>% fit(data, labels, epochs=5)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/learning_rate_schedule_piecewise_constant_decay.html","id":null,"dir":"Reference","previous_headings":"","what":"A LearningRateSchedule that uses a piecewise constant decay schedule. — learning_rate_schedule_piecewise_constant_decay","title":"A LearningRateSchedule that uses a piecewise constant decay schedule. — learning_rate_schedule_piecewise_constant_decay","text":"function returns 1-arg callable compute piecewise constant passed current optimizer step. can useful changing learning rate value across different invocations optimizer functions.","code":""},{"path":"https://keras3.posit.co/dev/reference/learning_rate_schedule_piecewise_constant_decay.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"A LearningRateSchedule that uses a piecewise constant decay schedule. — learning_rate_schedule_piecewise_constant_decay","text":"","code":"learning_rate_schedule_piecewise_constant_decay(   boundaries,   values,   name = \"PiecewiseConstant\" )"},{"path":"https://keras3.posit.co/dev/reference/learning_rate_schedule_piecewise_constant_decay.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"A LearningRateSchedule that uses a piecewise constant decay schedule. — learning_rate_schedule_piecewise_constant_decay","text":"boundaries list Python numbers strictly increasing entries, elements type optimizer step. values list Python numbers specifies values intervals defined boundaries. one element boundaries, elements type. name string. Optional name operation. Defaults \"PiecewiseConstant\".","code":""},{"path":"https://keras3.posit.co/dev/reference/learning_rate_schedule_piecewise_constant_decay.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"A LearningRateSchedule that uses a piecewise constant decay schedule. — learning_rate_schedule_piecewise_constant_decay","text":"1-arg callable learning rate schedule takes current optimizer step outputs decayed learning rate, scalar tensor type boundary tensors. output 1-arg function takes step values[0] step <= boundaries[0], values[1] step > boundaries[0] step <= boundaries[1], ..., values[-1] step > boundaries[-1].","code":""},{"path":"https://keras3.posit.co/dev/reference/learning_rate_schedule_piecewise_constant_decay.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"A LearningRateSchedule that uses a piecewise constant decay schedule. — learning_rate_schedule_piecewise_constant_decay","text":"use learning rate 1.0 first 100001 steps, 0.5 next 10000 steps, 0.1 additional steps.   can pass schedule directly optimizer learning rate. learning rate schedule also serializable deserializable using keras$optimizers$schedules$serialize keras$optimizers$schedules$deserialize.","code":"step <- 0 boundaries <- c(100000, 110000) values <- c(1.0, 0.5, 0.1) learning_rate_fn <- learning_rate_schedule_piecewise_constant_decay(   boundaries, values)  # Later, whenever we perform an optimization step, we pass in the step. learning_rate <- learning_rate_fn(step)"},{"path":"https://keras3.posit.co/dev/reference/learning_rate_schedule_piecewise_constant_decay.html","id":"raises","dir":"Reference","previous_headings":"","what":"Raises","title":"A LearningRateSchedule that uses a piecewise constant decay schedule. — learning_rate_schedule_piecewise_constant_decay","text":"ValueError: number elements boundaries values lists match.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/learning_rate_schedule_polynomial_decay.html","id":null,"dir":"Reference","previous_headings":"","what":"A LearningRateSchedule that uses a polynomial decay schedule. — learning_rate_schedule_polynomial_decay","title":"A LearningRateSchedule that uses a polynomial decay schedule. — learning_rate_schedule_polynomial_decay","text":"commonly observed monotonically decreasing learning rate, whose degree change carefully chosen, results better performing model. schedule applies polynomial decay function optimizer step, given provided initial_learning_rate, reach end_learning_rate given decay_steps. requires step value compute decayed learning rate. can just pass backend variable increment training step. schedule 1-arg callable produces decayed learning rate passed current optimizer step. can useful changing learning rate value across different invocations optimizer functions. computed :   cycle TRUE multiple decay_steps used, first one bigger step.   can pass schedule directly Optimizer learning rate.","code":"decayed_learning_rate <- function(step) {   step = min(step, decay_steps)   ((initial_learning_rate - end_learning_rate) *     (1 - step / decay_steps) ^ (power)) +     end_learning_rate } decayed_learning_rate <- function(step) {   decay_steps = decay_steps * ceil(step / decay_steps)   ((initial_learning_rate - end_learning_rate) *       (1 - step / decay_steps) ^ (power)) +     end_learning_rate }"},{"path":"https://keras3.posit.co/dev/reference/learning_rate_schedule_polynomial_decay.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"A LearningRateSchedule that uses a polynomial decay schedule. — learning_rate_schedule_polynomial_decay","text":"","code":"learning_rate_schedule_polynomial_decay(   initial_learning_rate,   decay_steps,   end_learning_rate = 1e-04,   power = 1,   cycle = FALSE,   name = \"PolynomialDecay\" )"},{"path":"https://keras3.posit.co/dev/reference/learning_rate_schedule_polynomial_decay.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"A LearningRateSchedule that uses a polynomial decay schedule. — learning_rate_schedule_polynomial_decay","text":"initial_learning_rate float. initial learning rate. decay_steps integer. Must positive. See decay computation . end_learning_rate float. minimal end learning rate. power float. power polynomial. Defaults 1.0. cycle boolean, whether cycle beyond decay_steps. name String.  Optional name operation. Defaults \"PolynomialDecay\".","code":""},{"path":"https://keras3.posit.co/dev/reference/learning_rate_schedule_polynomial_decay.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"A LearningRateSchedule that uses a polynomial decay schedule. — learning_rate_schedule_polynomial_decay","text":"1-arg callable learning rate schedule takes current optimizer step outputs decayed learning rate, scalar tensor type initial_learning_rate.","code":""},{"path":"https://keras3.posit.co/dev/reference/learning_rate_schedule_polynomial_decay.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"A LearningRateSchedule that uses a polynomial decay schedule. — learning_rate_schedule_polynomial_decay","text":"Fit model decaying 0.1 0.01 10000 steps using sqrt (.e. power=0.5):   learning rate schedule also serializable deserializable using keras$optimizers$schedules$serialize keras$optimizers$schedules$deserialize.","code":"... starter_learning_rate <- 0.1 end_learning_rate <- 0.01 decay_steps <- 10000 learning_rate_fn <- learning_rate_schedule_polynomial_decay(     starter_learning_rate,     decay_steps,     end_learning_rate,     power=0.5)  model %>% compile(   optimizer = optimizer_sgd(learning_rate=learning_rate_fn),   loss = 'sparse_categorical_crossentropy',   metrics = 'accuracy' )  model %>% fit(data, labels, epochs=5)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/load_model.html","id":null,"dir":"Reference","previous_headings":"","what":"Loads a model saved via save_model(). — load_model","title":"Loads a model saved via save_model(). — load_model","text":"Loads model saved via save_model().","code":""},{"path":"https://keras3.posit.co/dev/reference/load_model.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Loads a model saved via save_model(). — load_model","text":"","code":"load_model(model, custom_objects = NULL, compile = TRUE, safe_mode = TRUE)"},{"path":"https://keras3.posit.co/dev/reference/load_model.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Loads a model saved via save_model(). — load_model","text":"model string, path saved model file, raw vector, returned save_model(filepath = NULL) custom_objects Optional named list mapping names custom classes functions considered deserialization. compile Boolean, whether compile model loading. safe_mode Boolean, whether disallow unsafe lambda deserialization. safe_mode=FALSE, loading object potential trigger arbitrary code execution. argument applicable Keras v3 model format. Defaults TRUE.","code":""},{"path":"https://keras3.posit.co/dev/reference/load_model.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Loads a model saved via save_model(). — load_model","text":"Keras model instance. original model compiled, argument compile = TRUE set, returned model compiled. Otherwise, model left uncompiled.","code":""},{"path":"https://keras3.posit.co/dev/reference/load_model.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Loads a model saved via save_model(). — load_model","text":"Note model variables may different name values (var$name property, e.g. \"dense_1/kernel:0\") reloaded. recommended use layer attributes access specific variables, e.g. model |> get_layer(\"dense_1\") |> _$kernel.","code":"model <- keras_model_sequential(input_shape = c(3)) |>   layer_dense(5) |>   layer_activation_softmax()  model |> save_model(\"model.keras\") loaded_model <- load_model(\"model.keras\") x <- random_uniform(c(10, 3)) stopifnot(all.equal(   model |> predict(x),   loaded_model |> predict(x) ))"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/load_model_weights.html","id":null,"dir":"Reference","previous_headings":"","what":"Load the weights from a single file or sharded files. — load_model_weights","title":"Load the weights from a single file or sharded files. — load_model_weights","text":"Weights loaded based network's topology. means architecture weights saved. Note layers weights taken account topological ordering, adding removing layers fine long weights. Sharding loading sharded weights, specify filepath ending \".weights.json\" (configuration file), corresponding shard files (*_xxxxx.weights.h5) located alongside . Partial weight loading modified model, instance adding new layer (weights) changing shape weights layer, can choose ignore errors continue loading setting skip_mismatch=TRUE. case layer mismatching weights skipped. warning displayed skipped layer.","code":""},{"path":"https://keras3.posit.co/dev/reference/load_model_weights.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Load the weights from a single file or sharded files. — load_model_weights","text":"","code":"load_model_weights(model, filepath, skip_mismatch = FALSE, ...)"},{"path":"https://keras3.posit.co/dev/reference/load_model_weights.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Load the weights from a single file or sharded files. — load_model_weights","text":"model keras model. filepath Path path-like object weights. Accepts .weights.h5, legacy .h5, sharded weights .weights.json manifest sitting alongside shard files (*_xxxxx.weights.h5). skip_mismatch Boolean, whether skip loading layers mismatch number weights, mismatch shape weights. ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/load_model_weights.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Load the weights from a single file or sharded files. — load_model_weights","text":"called primarily side effects. model returned, invisibly, enable usage pipe.","code":""},{"path":"https://keras3.posit.co/dev/reference/load_model_weights.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Load the weights from a single file or sharded files. — load_model_weights","text":"","code":"model |> load_model_weights(\"model.weights.h5\") model |> load_model_weights(\"model.weights.json\")"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/loss_binary_crossentropy.html","id":null,"dir":"Reference","previous_headings":"","what":"Computes the cross-entropy loss between true labels and predicted labels. — loss_binary_crossentropy","title":"Computes the cross-entropy loss between true labels and predicted labels. — loss_binary_crossentropy","text":"Use cross-entropy loss binary (0 1) classification applications. loss function requires following inputs: y_true (true label): either 0 1. y_pred (predicted value): model's prediction, .e, single floating-point value either represents logit, (.e, value [-inf, inf] from_logits=TRUE) probability (.e, value [0., 1.] from_logits=FALSE).","code":""},{"path":"https://keras3.posit.co/dev/reference/loss_binary_crossentropy.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes the cross-entropy loss between true labels and predicted labels. — loss_binary_crossentropy","text":"","code":"loss_binary_crossentropy(   y_true,   y_pred,   from_logits = FALSE,   label_smoothing = 0,   axis = -1L,   ...,   reduction = \"sum_over_batch_size\",   name = \"binary_crossentropy\",   dtype = NULL )"},{"path":"https://keras3.posit.co/dev/reference/loss_binary_crossentropy.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Computes the cross-entropy loss between true labels and predicted labels. — loss_binary_crossentropy","text":"y_true Ground truth values. shape = [batch_size, d0, .. dN]. y_pred predicted values. shape = [batch_size, d0, .. dN]. from_logits Whether interpret y_pred tensor logit values. default, assume y_pred probabilities (.e., values [0, 1)). label_smoothing Float range [0, 1]. 0, smoothing occurs. > 0, compute loss predicted labels smoothed version true labels, smoothing squeezes labels towards 0.5. Larger values label_smoothing correspond heavier smoothing. axis axis along compute crossentropy (features axis). Defaults -1. ... forward/backward compatability. reduction Type reduction apply loss. almost cases \"sum_over_batch_size\". Supported options \"sum\", \"sum_over_batch_size\", \"mean\", \"mean_with_sample_weight\" NULL. \"sum\" sums loss, \"sum_over_batch_size\" \"mean\" sum loss divide sample size, \"mean_with_sample_weight\" sums loss divides sum sample weights. \"none\" NULL perform aggregation. Defaults \"sum_over_batch_size\". name Optional name loss instance. dtype dtype loss's computations. Defaults NULL, means using config_floatx(). config_floatx() \"float32\" unless set different value (via config_set_floatx()). keras$DTypePolicy provided, compute_dtype utilized.","code":""},{"path":"https://keras3.posit.co/dev/reference/loss_binary_crossentropy.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Computes the cross-entropy loss between true labels and predicted labels. — loss_binary_crossentropy","text":"Binary crossentropy loss value. shape = [batch_size, d0, .. dN-1].","code":""},{"path":"https://keras3.posit.co/dev/reference/loss_binary_crossentropy.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Computes the cross-entropy loss between true labels and predicted labels. — loss_binary_crossentropy","text":"Recommended Usage: (set from_logits=TRUE) compile() API:   standalone function:                     Default Usage: (set from_logits=FALSE)","code":"y_true <- rbind(c(0, 1), c(0, 0)) y_pred <- rbind(c(0.6, 0.4), c(0.4, 0.6)) loss <- loss_binary_crossentropy(y_true, y_pred) loss ## tf.Tensor([0.91629076 0.7135582 ], shape=(2), dtype=float32) model %>% compile(     loss = loss_binary_crossentropy(from_logits=TRUE),     ... ) # Example 1: (batch_size = 1, number of samples = 4) y_true <- op_array(c(0, 1, 0, 0)) y_pred <- op_array(c(-18.6, 0.51, 2.94, -12.8)) bce <- loss_binary_crossentropy(from_logits = TRUE) bce(y_true, y_pred) ## tf.Tensor(0.865458, shape=(), dtype=float32) # Example 2: (batch_size = 2, number of samples = 4) y_true <- rbind(c(0, 1), c(0, 0)) y_pred <- rbind(c(-18.6, 0.51), c(2.94, -12.8)) # Using default 'auto'/'sum_over_batch_size' reduction type. bce <- loss_binary_crossentropy(from_logits = TRUE) bce(y_true, y_pred) ## tf.Tensor(0.865458, shape=(), dtype=float32) # Using 'sample_weight' attribute bce(y_true, y_pred, sample_weight = c(0.8, 0.2)) ## tf.Tensor(0.2436386, shape=(), dtype=float32) # 0.243 # Using 'sum' reduction` type. bce <- loss_binary_crossentropy(from_logits = TRUE, reduction = \"sum\") bce(y_true, y_pred) ## tf.Tensor(1.730916, shape=(), dtype=float32) # Using 'none' reduction type. bce <- loss_binary_crossentropy(from_logits = TRUE, reduction = NULL) bce(y_true, y_pred) ## tf.Tensor([0.23515666 1.4957594 ], shape=(2), dtype=float32) # Make the following updates to the above \"Recommended Usage\" section # 1. Set `from_logits=FALSE` loss_binary_crossentropy() # OR ...('from_logits=FALSE') ## <LossFunctionWrapper(<function binary_crossentropy at 0x0>, kwargs={'from_logits': False, 'label_smoothing': 0.0, 'axis': -1})> ##  signature: (y_true, y_pred, sample_weight=None) # 2. Update `y_pred` to use probabilities instead of logits y_pred <- c(0.6, 0.3, 0.2, 0.8) # OR [[0.6, 0.3], [0.2, 0.8]]"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/loss_binary_focal_crossentropy.html","id":null,"dir":"Reference","previous_headings":"","what":"Computes focal cross-entropy loss between true labels and predictions. — loss_binary_focal_crossentropy","title":"Computes focal cross-entropy loss between true labels and predictions. — loss_binary_focal_crossentropy","text":"According Lin et al., 2018, helps apply focal factor -weight easy examples focus hard examples. default, focal tensor computed follows: focal_factor = (1 - output)^gamma class 1 focal_factor = output^gamma class 0 gamma focusing parameter. gamma = 0, focal effect binary crossentropy loss. apply_class_balancing == TRUE, function also takes account weight balancing factor binary classes 0 1 follows: weight = alpha class 1 (target == 1) weight = 1 - alpha class 0 alpha float range [0, 1]. Binary cross-entropy loss often used binary (0 1) classification tasks. loss function requires following inputs: y_true (true label): either 0 1. y_pred (predicted value): model's prediction, .e, single floating-point value either represents logit, (.e, value [-inf, inf] from_logits=TRUE) probability (.e, value [0., 1.] from_logits=FALSE). According Lin et al., 2018, helps apply \"focal factor\" -weight easy examples focus hard examples. default, focal tensor computed follows: focal_factor = (1 - output) ** gamma class 1 focal_factor = output ** gamma class 0 gamma focusing parameter. gamma=0, function equivalent binary crossentropy loss.","code":""},{"path":"https://keras3.posit.co/dev/reference/loss_binary_focal_crossentropy.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes focal cross-entropy loss between true labels and predictions. — loss_binary_focal_crossentropy","text":"","code":"loss_binary_focal_crossentropy(   y_true,   y_pred,   apply_class_balancing = FALSE,   alpha = 0.25,   gamma = 2,   from_logits = FALSE,   label_smoothing = 0,   axis = -1L,   ...,   reduction = \"sum_over_batch_size\",   name = \"binary_focal_crossentropy\",   dtype = NULL )"},{"path":"https://keras3.posit.co/dev/reference/loss_binary_focal_crossentropy.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Computes focal cross-entropy loss between true labels and predictions. — loss_binary_focal_crossentropy","text":"y_true Ground truth values, shape (batch_size, d0, .. dN). y_pred predicted values, shape (batch_size, d0, .. dN). apply_class_balancing bool, whether apply weight balancing binary classes 0 1. alpha weight balancing factor class 1, default 0.25 mentioned reference Lin et al., 2018.  weight class 0 1.0 - alpha. gamma focusing parameter used compute focal factor, default 2.0 mentioned reference Lin et al., 2018. from_logits Whether interpret y_pred tensor logit values. default, assume y_pred probabilities (.e., values [0, 1]). label_smoothing Float [0, 1]. 0, smoothing occurs. > 0, compute loss predicted labels smoothed version true labels, smoothing squeezes labels towards 0.5. Larger values label_smoothing correspond heavier smoothing. axis axis along compute crossentropy (features axis). Defaults -1. ... forward/backward compatability. reduction Type reduction apply loss. almost cases \"sum_over_batch_size\". Supported options \"sum\", \"sum_over_batch_size\", \"mean\", \"mean_with_sample_weight\" NULL. \"sum\" sums loss, \"sum_over_batch_size\" \"mean\" sum loss divide sample size, \"mean_with_sample_weight\" sums loss divides sum sample weights. \"none\" NULL perform aggregation. Defaults \"sum_over_batch_size\". name Optional name loss instance. dtype dtype loss's computations. Defaults NULL, means using config_floatx(). config_floatx() \"float32\" unless set different value (via config_set_floatx()). keras$DTypePolicy provided, compute_dtype utilized.","code":""},{"path":"https://keras3.posit.co/dev/reference/loss_binary_focal_crossentropy.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Computes focal cross-entropy loss between true labels and predictions. — loss_binary_focal_crossentropy","text":"Binary focal crossentropy loss value shape = [batch_size, d0, .. dN-1].","code":""},{"path":"https://keras3.posit.co/dev/reference/loss_binary_focal_crossentropy.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Computes focal cross-entropy loss between true labels and predictions. — loss_binary_focal_crossentropy","text":"compile() API:   standalone function:","code":"y_true <- rbind(c(0, 1), c(0, 0)) y_pred <- rbind(c(0.6, 0.4), c(0.4, 0.6)) loss <- loss_binary_focal_crossentropy(y_true, y_pred, gamma = 2) loss ## tf.Tensor([0.32986468 0.20579839], shape=(2), dtype=float32) model %>% compile(     loss = loss_binary_focal_crossentropy(         gamma = 2.0, from_logits = TRUE),     ... ) # Example 1: (batch_size = 1, number of samples = 4) y_true <- op_array(c(0, 1, 0, 0)) y_pred <- op_array(c(-18.6, 0.51, 2.94, -12.8)) loss <- loss_binary_focal_crossentropy(gamma = 2, from_logits = TRUE) loss(y_true, y_pred) ## tf.Tensor(0.6912122, shape=(), dtype=float32) # Apply class weight loss <- loss_binary_focal_crossentropy(   apply_class_balancing = TRUE, gamma = 2, from_logits = TRUE) loss(y_true, y_pred) ## tf.Tensor(0.5101333, shape=(), dtype=float32) # Example 2: (batch_size = 2, number of samples = 4) y_true <- rbind(c(0, 1), c(0, 0)) y_pred <- rbind(c(-18.6, 0.51), c(2.94, -12.8)) # Using default 'auto'/'sum_over_batch_size' reduction type. loss <- loss_binary_focal_crossentropy(     gamma = 3, from_logits = TRUE) loss(y_true, y_pred) ## tf.Tensor(0.6469951, shape=(), dtype=float32) # Apply class weight loss <- loss_binary_focal_crossentropy(      apply_class_balancing = TRUE, gamma = 3, from_logits = TRUE) loss(y_true, y_pred) ## tf.Tensor(0.48214132, shape=(), dtype=float32) # Using 'sample_weight' attribute with focal effect loss <- loss_binary_focal_crossentropy(     gamma = 3, from_logits = TRUE) loss(y_true, y_pred, sample_weight = c(0.8, 0.2)) ## tf.Tensor(0.13312504, shape=(), dtype=float32) # Apply class weight loss <- loss_binary_focal_crossentropy(      apply_class_balancing = TRUE, gamma = 3, from_logits = TRUE) loss(y_true, y_pred, sample_weight = c(0.8, 0.2)) ## tf.Tensor(0.09735977, shape=(), dtype=float32) # Using 'sum' reduction` type. loss <- loss_binary_focal_crossentropy(     gamma = 4, from_logits = TRUE,     reduction = \"sum\") loss(y_true, y_pred) ## tf.Tensor(1.2218808, shape=(), dtype=float32) # Apply class weight loss <- loss_binary_focal_crossentropy(     apply_class_balancing = TRUE, gamma = 4, from_logits = TRUE,     reduction = \"sum\") loss(y_true, y_pred) ## tf.Tensor(0.9140807, shape=(), dtype=float32) # Using 'none' reduction type. loss <- loss_binary_focal_crossentropy(     gamma = 5, from_logits = TRUE,     reduction = NULL) loss(y_true, y_pred) ## tf.Tensor([0.00174837 1.1561027 ], shape=(2), dtype=float32) # Apply class weight loss <- loss_binary_focal_crossentropy(     apply_class_balancing = TRUE, gamma = 5, from_logits = TRUE,     reduction = NULL) loss(y_true, y_pred) ## tf.Tensor([4.3709317e-04 8.6707699e-01], shape=(2), dtype=float32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/loss_categorical_crossentropy.html","id":null,"dir":"Reference","previous_headings":"","what":"Computes the crossentropy loss between the labels and predictions. — loss_categorical_crossentropy","title":"Computes the crossentropy loss between the labels and predictions. — loss_categorical_crossentropy","text":"Use crossentropy loss function two label classes. expect labels provided one_hot representation. want provide labels integers, please use SparseCategoricalCrossentropy loss. num_classes floating point values per feature, .e., shape y_pred y_true [batch_size, num_classes].","code":""},{"path":"https://keras3.posit.co/dev/reference/loss_categorical_crossentropy.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes the crossentropy loss between the labels and predictions. — loss_categorical_crossentropy","text":"","code":"loss_categorical_crossentropy(   y_true,   y_pred,   from_logits = FALSE,   label_smoothing = 0,   axis = -1L,   ...,   reduction = \"sum_over_batch_size\",   name = \"categorical_crossentropy\",   dtype = NULL )"},{"path":"https://keras3.posit.co/dev/reference/loss_categorical_crossentropy.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Computes the crossentropy loss between the labels and predictions. — loss_categorical_crossentropy","text":"y_true Tensor one-hot true targets. y_pred Tensor predicted targets. from_logits Whether y_pred expected logits tensor. default, assume y_pred encodes probability distribution. label_smoothing Float [0, 1]. > 0, label values smoothed, meaning confidence label values relaxed. example, 0.1, use 0.1 / num_classes non-target labels 0.9 + 0.1 / num_classes target labels. axis axis along compute crossentropy (features axis). Defaults -1. ... forward/backward compatability. reduction Type reduction apply loss. almost cases \"sum_over_batch_size\". Supported options \"sum\", \"sum_over_batch_size\", \"mean\", \"mean_with_sample_weight\" NULL. \"sum\" sums loss, \"sum_over_batch_size\" \"mean\" sum loss divide sample size, \"mean_with_sample_weight\" sums loss divides sum sample weights. \"none\" NULL perform aggregation. Defaults \"sum_over_batch_size\". name Optional name loss instance. dtype dtype loss's computations. Defaults NULL, means using config_floatx(). config_floatx() \"float32\" unless set different value (via config_set_floatx()). keras$DTypePolicy provided, compute_dtype utilized.","code":""},{"path":"https://keras3.posit.co/dev/reference/loss_categorical_crossentropy.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Computes the crossentropy loss between the labels and predictions. — loss_categorical_crossentropy","text":"Categorical crossentropy loss value.","code":""},{"path":"https://keras3.posit.co/dev/reference/loss_categorical_crossentropy.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Computes the crossentropy loss between the labels and predictions. — loss_categorical_crossentropy","text":"Standalone usage:                 Usage compile() API:","code":"y_true <- rbind(c(0, 1, 0), c(0, 0, 1)) y_pred <- rbind(c(0.05, 0.95, 0), c(0.1, 0.8, 0.1)) loss <- loss_categorical_crossentropy(y_true, y_pred) loss ## tf.Tensor([0.05129331 2.3025851 ], shape=(2), dtype=float32) y_true <- rbind(c(0, 1, 0), c(0, 0, 1)) y_pred <- rbind(c(0.05, 0.95, 0), c(0.1, 0.8, 0.1)) # Using 'auto'/'sum_over_batch_size' reduction type. cce <- loss_categorical_crossentropy() cce(y_true, y_pred) ## tf.Tensor(1.1769392, shape=(), dtype=float32) # Calling with 'sample_weight'. cce(y_true, y_pred, sample_weight = op_array(c(0.3, 0.7))) ## tf.Tensor(0.8135988, shape=(), dtype=float32) # Using 'sum' reduction type. cce <- loss_categorical_crossentropy(reduction = \"sum\") cce(y_true, y_pred) ## tf.Tensor(2.3538785, shape=(), dtype=float32) # Using 'none' reduction type. cce <- loss_categorical_crossentropy(reduction = NULL) cce(y_true, y_pred) ## tf.Tensor([0.05129331 2.3025851 ], shape=(2), dtype=float32) model %>% compile(optimizer = 'sgd',               loss=loss_categorical_crossentropy())"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/loss_categorical_focal_crossentropy.html","id":null,"dir":"Reference","previous_headings":"","what":"Computes the alpha balanced focal crossentropy loss. — loss_categorical_focal_crossentropy","title":"Computes the alpha balanced focal crossentropy loss. — loss_categorical_focal_crossentropy","text":"Use crossentropy loss function two label classes want handle class imbalance without using class_weights. expect labels provided one_hot representation. According Lin et al., 2018, helps apply focal factor -weight easy examples focus hard examples. general formula focal loss (FL) follows: FL(p_t) = (1 - p_t)^gamma * log(p_t) p_t defined follows: p_t = output y_true == 1, else 1 - output (1 - p_t)^gamma modulating_factor, gamma focusing parameter. gamma = 0, focal effect cross entropy. gamma reduces importance given simple examples smooth manner. authors use alpha-balanced variant focal loss (FL) paper: FL(p_t) = -alpha * (1 - p_t)^gamma * log(p_t) alpha weight factor classes. alpha = 1, loss able handle class imbalance properly classes weight. can constant list constants. alpha list, must length number classes. formula can generalized : FL(p_t) = alpha * (1 - p_t)^gamma * CrossEntropy(y_true, y_pred) minus comes CrossEntropy(y_true, y_pred) (CE). Extending multi-class case straightforward: FL(p_t) = alpha * (1 - p_t) ** gamma * CategoricalCE(y_true, y_pred) snippet , num_classes floating pointing values per example. shape y_pred y_true (batch_size, num_classes).","code":""},{"path":"https://keras3.posit.co/dev/reference/loss_categorical_focal_crossentropy.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes the alpha balanced focal crossentropy loss. — loss_categorical_focal_crossentropy","text":"","code":"loss_categorical_focal_crossentropy(   y_true,   y_pred,   alpha = 0.25,   gamma = 2,   from_logits = FALSE,   label_smoothing = 0,   axis = -1L,   ...,   reduction = \"sum_over_batch_size\",   name = \"categorical_focal_crossentropy\",   dtype = NULL )"},{"path":"https://keras3.posit.co/dev/reference/loss_categorical_focal_crossentropy.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Computes the alpha balanced focal crossentropy loss. — loss_categorical_focal_crossentropy","text":"y_true Tensor one-hot true targets. y_pred Tensor predicted targets. alpha weight balancing factor classes, default 0.25 mentioned reference. can list floats scalar. multi-class case, alpha may set inverse class frequency using compute_class_weight sklearn.utils. gamma focusing parameter, default 2.0 mentioned reference. helps gradually reduce importance given simple examples smooth manner. gamma = 0, focal effect categorical crossentropy. from_logits Whether output expected logits tensor. default, consider output encodes probability distribution. label_smoothing Float [0, 1]. > 0, label values smoothed, meaning confidence label values relaxed. example, 0.1, use 0.1 / num_classes non-target labels 0.9 + 0.1 / num_classes target labels. axis axis along compute crossentropy (features axis). Defaults -1. ... forward/backward compatability. reduction Type reduction apply loss. almost cases \"sum_over_batch_size\". Supported options \"sum\", \"sum_over_batch_size\", \"mean\", \"mean_with_sample_weight\" NULL. \"sum\" sums loss, \"sum_over_batch_size\" \"mean\" sum loss divide sample size, \"mean_with_sample_weight\" sums loss divides sum sample weights. \"none\" NULL perform aggregation. Defaults \"sum_over_batch_size\". name Optional name loss instance. dtype dtype loss's computations. Defaults NULL, means using config_floatx(). config_floatx() \"float32\" unless set different value (via config_set_floatx()). keras$DTypePolicy provided, compute_dtype utilized.","code":""},{"path":"https://keras3.posit.co/dev/reference/loss_categorical_focal_crossentropy.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Computes the alpha balanced focal crossentropy loss. — loss_categorical_focal_crossentropy","text":"Categorical focal crossentropy loss value.","code":""},{"path":"https://keras3.posit.co/dev/reference/loss_categorical_focal_crossentropy.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Computes the alpha balanced focal crossentropy loss. — loss_categorical_focal_crossentropy","text":"Standalone usage:                 Usage compile() API:","code":"y_true <- rbind(c(0, 1, 0), c(0, 0, 1)) y_pred <- rbind(c(0.05, 0.95, 0), c(0.1, 0.8, 0.1)) loss <- loss_categorical_focal_crossentropy(y_true, y_pred) loss ## tf.Tensor([3.2058331e-05 4.6627346e-01], shape=(2), dtype=float32) y_true <- rbind(c(0, 1, 0), c(0, 0, 1)) y_pred <- rbind(c(0.05, 0.95, 0), c(0.1, 0.8, 0.1)) # Using 'auto'/'sum_over_batch_size' reduction type. cce <- loss_categorical_focal_crossentropy() cce(y_true, y_pred) ## tf.Tensor(0.23315276, shape=(), dtype=float32) # Calling with 'sample_weight'. cce(y_true, y_pred, sample_weight = op_array(c(0.3, 0.7))) ## tf.Tensor(0.16320053, shape=(), dtype=float32) # Using 'sum' reduction type. cce <- loss_categorical_focal_crossentropy(reduction = \"sum\") cce(y_true, y_pred) ## tf.Tensor(0.46630552, shape=(), dtype=float32) # Using 'none' reduction type. cce <- loss_categorical_focal_crossentropy(reduction = NULL) cce(y_true, y_pred) ## tf.Tensor([3.2058331e-05 4.6627346e-01], shape=(2), dtype=float32) model %>% compile(   optimizer = 'adam',   loss = loss_categorical_focal_crossentropy())"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/loss_categorical_generalized_cross_entropy.html","id":null,"dir":"Reference","previous_headings":"","what":"Computes the generalized cross entropy loss. — loss_categorical_generalized_cross_entropy","title":"Computes the generalized cross entropy loss. — loss_categorical_generalized_cross_entropy","text":"generalized cross entropy (GCE) loss offers robustness noisy labels interpolating categorical cross entropy (q -> 0) mean absolute error (q -> 1). true-class probability p noise parameter q, loss loss = (1 - p^q) / q.","code":""},{"path":"https://keras3.posit.co/dev/reference/loss_categorical_generalized_cross_entropy.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes the generalized cross entropy loss. — loss_categorical_generalized_cross_entropy","text":"","code":"loss_categorical_generalized_cross_entropy(   y_true,   y_pred,   q = 0.5,   ...,   reduction = \"sum_over_batch_size\",   name = \"categorical_generalized_cross_entropy\",   dtype = NULL )"},{"path":"https://keras3.posit.co/dev/reference/loss_categorical_generalized_cross_entropy.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Computes the generalized cross entropy loss. — loss_categorical_generalized_cross_entropy","text":"y_true Integer class indices shape (batch_size) (batch_size, 1). y_pred Predicted class probabilities shape (batch_size, num_classes). q Float (0, 1). Controls transition cross entropy mean absolute error. Defaults 0.5. q approaches 0: behaves like categorical cross entropy. q approaches 1: behaves like mean absolute error. ... forward/backward compatibility. reduction Type reduction apply loss. almost cases \"sum_over_batch_size\". Supported options \"sum\", \"sum_over_batch_size\", \"mean\", \"mean_with_sample_weight\" NULL. \"sum\" sums loss, \"sum_over_batch_size\" \"mean\" sum loss divide sample size, \"mean_with_sample_weight\" sums loss divides sum sample weights. \"none\" NULL perform aggregation. Defaults \"sum_over_batch_size\". name Optional name loss instance. dtype Dtype used loss computations. Defaults config_floatx() (global float type).","code":""},{"path":"https://keras3.posit.co/dev/reference/loss_categorical_generalized_cross_entropy.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Computes the generalized cross entropy loss. — loss_categorical_generalized_cross_entropy","text":"Generalized cross entropy loss value(s).","code":""},{"path":"https://keras3.posit.co/dev/reference/loss_categorical_generalized_cross_entropy.html","id":"references","dir":"Reference","previous_headings":"","what":"References","title":"Computes the generalized cross entropy loss. — loss_categorical_generalized_cross_entropy","text":"Zhang & Sabuncu (2018), \"Generalized Cross Entropy Loss Training Deep Neural Networks Noisy Labels\"","code":""},{"path":"https://keras3.posit.co/dev/reference/loss_categorical_generalized_cross_entropy.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Computes the generalized cross entropy loss. — loss_categorical_generalized_cross_entropy","text":"","code":"y_true <- c(0L, 1L, 0L, 1L) y_pred <- rbind(   c(0.7, 0.3),   c(0.2, 0.8),   c(0.6, 0.4),   c(0.4, 0.6) ) gce <- loss_categorical_generalized_cross_entropy(q = 0.7) gce(y_true, y_pred) ## tf.Tensor(0.34529287, shape=(), dtype=float32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/loss_categorical_hinge.html","id":null,"dir":"Reference","previous_headings":"","what":"Computes the categorical hinge loss between y_true & y_pred. — loss_categorical_hinge","title":"Computes the categorical hinge loss between y_true & y_pred. — loss_categorical_hinge","text":"Formula:   neg=maximum((1-y_true)*y_pred) pos=sum(y_true*y_pred)","code":"loss <- maximum(neg - pos + 1, 0)"},{"path":"https://keras3.posit.co/dev/reference/loss_categorical_hinge.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes the categorical hinge loss between y_true & y_pred. — loss_categorical_hinge","text":"","code":"loss_categorical_hinge(   y_true,   y_pred,   ...,   reduction = \"sum_over_batch_size\",   name = \"categorical_hinge\",   dtype = NULL )"},{"path":"https://keras3.posit.co/dev/reference/loss_categorical_hinge.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Computes the categorical hinge loss between y_true & y_pred. — loss_categorical_hinge","text":"y_true ground truth values. y_true values expected either {-1, +1} {0, 1} (.e. one-hot-encoded tensor) shape <- [batch_size, d0, .. dN]. y_pred predicted values shape = [batch_size, d0, .. dN]. ... forward/backward compatability. reduction Type reduction apply loss. almost cases \"sum_over_batch_size\". Supported options \"sum\", \"sum_over_batch_size\", \"mean\", \"mean_with_sample_weight\" NULL. \"sum\" sums loss, \"sum_over_batch_size\" \"mean\" sum loss divide sample size, \"mean_with_sample_weight\" sums loss divides sum sample weights. \"none\" NULL perform aggregation. Defaults \"sum_over_batch_size\". name Optional name loss instance. dtype dtype loss's computations. Defaults NULL, means using config_floatx(). config_floatx() \"float32\" unless set different value (via config_set_floatx()). keras$DTypePolicy provided, compute_dtype utilized.","code":""},{"path":"https://keras3.posit.co/dev/reference/loss_categorical_hinge.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Computes the categorical hinge loss between y_true & y_pred. — loss_categorical_hinge","text":"Categorical hinge loss values shape = [batch_size, d0, .. dN-1].","code":""},{"path":"https://keras3.posit.co/dev/reference/loss_categorical_hinge.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Computes the categorical hinge loss between y_true & y_pred. — loss_categorical_hinge","text":"","code":"y_true <- rbind(c(0, 1), c(0, 0)) y_pred <- rbind(c(0.6, 0.4), c(0.4, 0.6)) loss <- loss_categorical_hinge(y_true, y_pred)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/loss_circle.html","id":null,"dir":"Reference","previous_headings":"","what":"Computes Circle Loss between integer labels and L2-normalized embeddings. — loss_circle","title":"Computes Circle Loss between integer labels and L2-normalized embeddings. — loss_circle","text":"designed minimize within-class distances maximize -class distances L2 normalized embedding space. metric learning loss designed minimize within-class distance maximize -class distance flexible manner dynamically adjusting penalty strength based optimization status similarity score. use Circle Loss effectively, model output embeddings without activation function (Dense layer activation=NULL) followed UnitNormalization layer ensure unit-norm embeddings.","code":""},{"path":"https://keras3.posit.co/dev/reference/loss_circle.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes Circle Loss between integer labels and L2-normalized embeddings. — loss_circle","text":"","code":"loss_circle(   y_true,   y_pred,   ref_labels = NULL,   ref_embeddings = NULL,   remove_diagonal = TRUE,   gamma = 80L,   margin = 0.4,   ...,   reduction = \"sum_over_batch_size\",   name = \"circle\",   dtype = NULL )"},{"path":"https://keras3.posit.co/dev/reference/loss_circle.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Computes Circle Loss between integer labels and L2-normalized embeddings. — loss_circle","text":"y_true Tensor ground truth labels integer format. y_pred Tensor predicted L2 normalized embeddings. ref_labels Optional integer tensor labels reference embeddings. NULL, defaults y_true. ref_embeddings Optional tensor L2 normalized reference embeddings. NULL, defaults y_pred. remove_diagonal Boolean, whether remove self-similarities positive mask. Defaults TRUE. gamma Scaling factor determines largest scale similarity score. Defaults 80. margin relaxation factor, distance, negatives weighted positives weighted. Similarly, distance negatives weighted positive weighted. Defaults 0.4. ... forward/backward compatability. reduction Type reduction apply loss. almost cases \"sum_over_batch_size\". Supported options \"sum\", \"sum_over_batch_size\", \"mean\", \"mean_with_sample_weight\" NULL. \"sum\" sums loss, \"sum_over_batch_size\" \"mean\" sum loss divide sample size, \"mean_with_sample_weight\" sums loss divides sum sample weights. \"none\" NULL perform aggregation. Defaults \"sum_over_batch_size\". name Optional name loss instance. dtype dtype loss's computations. Defaults NULL, means using config_floatx(). config_floatx() \"float32\" unless set different value (via config_set_floatx()). keras.DTypePolicy provided, compute_dtype utilized.","code":""},{"path":"https://keras3.posit.co/dev/reference/loss_circle.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Computes Circle Loss between integer labels and L2-normalized embeddings. — loss_circle","text":"Circle loss value.","code":""},{"path":"https://keras3.posit.co/dev/reference/loss_circle.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Computes Circle Loss between integer labels and L2-normalized embeddings. — loss_circle","text":"Usage compile() API:","code":"model <- keras_model_sequential(input_shape = c(224, 224, 3)) |>   layer_conv_2d(16, c(3, 3), activation = 'relu') |>   layer_flatten() |>   layer_dense(64, activation = NULL) |>   # No activation   layer_unit_normalization()  # L2 normalization  model |>   compile(optimizer = \"adam\", loss = loss_circle())"},{"path":"https://keras3.posit.co/dev/reference/loss_circle.html","id":"reference","dir":"Reference","previous_headings":"","what":"Reference","title":"Computes Circle Loss between integer labels and L2-normalized embeddings. — loss_circle","text":"Yifan Sun et al., 2020","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/loss_cosine_similarity.html","id":null,"dir":"Reference","previous_headings":"","what":"Computes the cosine similarity between y_true & y_pred. — loss_cosine_similarity","title":"Computes the cosine similarity between y_true & y_pred. — loss_cosine_similarity","text":"Formula:   Note number -1 1. negative number -1 0, 0 indicates orthogonality values closer -1 indicate greater similarity. makes usable loss function setting try maximize proximity predictions targets. either y_true y_pred zero vector, cosine similarity 0 regardless proximity predictions targets.","code":"loss <- -sum(l2_norm(y_true) * l2_norm(y_pred))"},{"path":"https://keras3.posit.co/dev/reference/loss_cosine_similarity.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes the cosine similarity between y_true & y_pred. — loss_cosine_similarity","text":"","code":"loss_cosine_similarity(   y_true,   y_pred,   axis = -1L,   ...,   reduction = \"sum_over_batch_size\",   name = \"cosine_similarity\",   dtype = NULL )"},{"path":"https://keras3.posit.co/dev/reference/loss_cosine_similarity.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Computes the cosine similarity between y_true & y_pred. — loss_cosine_similarity","text":"y_true Tensor true targets. y_pred Tensor predicted targets. axis axis along cosine similarity computed (features axis). Defaults -1. ... forward/backward compatability. reduction Type reduction apply loss. almost cases \"sum_over_batch_size\". Supported options \"sum\", \"sum_over_batch_size\", \"mean\", \"mean_with_sample_weight\" NULL. \"sum\" sums loss, \"sum_over_batch_size\" \"mean\" sum loss divide sample size, \"mean_with_sample_weight\" sums loss divides sum sample weights. \"none\" NULL perform aggregation. Defaults \"sum_over_batch_size\". name Optional name loss instance. dtype dtype loss's computations. Defaults NULL, means using config_floatx(). config_floatx() \"float32\" unless set different value (via config_set_floatx()). keras$DTypePolicy provided, compute_dtype utilized.","code":""},{"path":"https://keras3.posit.co/dev/reference/loss_cosine_similarity.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Computes the cosine similarity between y_true & y_pred. — loss_cosine_similarity","text":"Cosine similarity tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/loss_cosine_similarity.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Computes the cosine similarity between y_true & y_pred. — loss_cosine_similarity","text":"","code":"y_true <- rbind(c(0., 1.), c(1., 1.), c(1., 1.)) y_pred <- rbind(c(1., 0.), c(1., 1.), c(-1., -1.)) loss <- loss_cosine_similarity(y_true, y_pred, axis=-1) loss ## tf.Tensor([-0.         -0.99999994  0.99999994], shape=(3), dtype=float32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/loss_ctc.html","id":null,"dir":"Reference","previous_headings":"","what":"CTC (Connectionist Temporal Classification) loss. — loss_ctc","title":"CTC (Connectionist Temporal Classification) loss. — loss_ctc","text":"CTC (Connectionist Temporal Classification) loss.","code":""},{"path":"https://keras3.posit.co/dev/reference/loss_ctc.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"CTC (Connectionist Temporal Classification) loss. — loss_ctc","text":"","code":"loss_ctc(   y_true,   y_pred,   ...,   reduction = \"sum_over_batch_size\",   name = \"ctc\",   dtype = NULL )"},{"path":"https://keras3.posit.co/dev/reference/loss_ctc.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"CTC (Connectionist Temporal Classification) loss. — loss_ctc","text":"y_true tensor shape (batch_size, target_max_length) containing true labels integer format. 0 always represents blank/mask index used classes. y_pred tensor shape (batch_size, output_max_length, num_classes) containing logits (output model). normalized via softmax. ... forward/backward compatability. reduction Type reduction apply loss. almost cases \"sum_over_batch_size\". Supported options \"sum\", \"sum_over_batch_size\", \"mean\", \"mean_with_sample_weight\" NULL. \"sum\" sums loss, \"sum_over_batch_size\" \"mean\" sum loss divide sample size, \"mean_with_sample_weight\" sums loss divides sum sample weights. \"none\" NULL perform aggregation. Defaults \"sum_over_batch_size\". name Optional name loss instance. dtype dtype loss's computations. Defaults NULL, means using config_floatx(). config_floatx() \"float32\" unless set different value (via config_set_floatx()). keras$DTypePolicy provided, compute_dtype utilized.","code":""},{"path":"https://keras3.posit.co/dev/reference/loss_ctc.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"CTC (Connectionist Temporal Classification) loss. — loss_ctc","text":"CTC loss value.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/loss_dice.html","id":null,"dir":"Reference","previous_headings":"","what":"Computes the Dice loss value between y_true and y_pred. — loss_dice","title":"Computes the Dice loss value between y_true and y_pred. — loss_dice","text":"Formula:   Formula:","code":"loss = 1 - (2 * sum(y_true * y_pred)) / (sum(y_true) + sum(y_pred)) loss = 1 - (2 * sum(y_true * y_pred)) / (sum(y_true) + sum(y_pred))"},{"path":"https://keras3.posit.co/dev/reference/loss_dice.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes the Dice loss value between y_true and y_pred. — loss_dice","text":"","code":"loss_dice(   y_true,   y_pred,   ...,   reduction = \"sum_over_batch_size\",   name = \"dice\",   axis = NULL,   dtype = NULL )"},{"path":"https://keras3.posit.co/dev/reference/loss_dice.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Computes the Dice loss value between y_true and y_pred. — loss_dice","text":"y_true Tensor true targets. y_pred Tensor predicted targets. ... forward/backward compatability. reduction Type reduction apply loss. almost cases \"sum_over_batch_size\". Supported options \"sum\", \"sum_over_batch_size\", \"mean\", \"mean_with_sample_weight\" NULL. \"sum\" sums loss, \"sum_over_batch_size\" \"mean\" sum loss divide sample size, \"mean_with_sample_weight\" sums loss divides sum sample weights. \"none\" NULL perform aggregation. Defaults \"sum_over_batch_size\". name String, name object axis List dimensions loss calculated. Defaults NULL. dtype dtype loss's computations. Defaults NULL, means using config_floatx(). config_floatx() \"float32\" unless set different value (via config_set_floatx()). keras$DTypePolicy provided, compute_dtype utilized.","code":""},{"path":"https://keras3.posit.co/dev/reference/loss_dice.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Computes the Dice loss value between y_true and y_pred. — loss_dice","text":"y_true y_pred provided, Dice loss value. Otherwise, Loss() instance.","code":""},{"path":"https://keras3.posit.co/dev/reference/loss_dice.html","id":"example","dir":"Reference","previous_headings":"","what":"Example","title":"Computes the Dice loss value between y_true and y_pred. — loss_dice","text":"","code":"y_true <- array(c(1, 1, 0, 0,                   1, 1, 0, 0), dim = c(2, 2, 2, 1)) y_pred <- array(c(0, 0.4, 0,   0,                   1,   0, 1, 0.9), dim = c(2, 2, 2, 1))  axis <- c(2, 3, 4) loss_fn <- loss_dice(axis = axis, reduction = NULL) loss <- loss_fn(y_true, y_pred) stopifnot(shape(loss) == shape(2)) loss ## tf.Tensor([0.5        0.75757575], shape=(2), dtype=float32) loss_fn <- loss_dice() loss <- loss_fn(y_true, y_pred) stopifnot(shape(loss) == shape()) loss ## tf.Tensor(0.6164384, shape=(), dtype=float32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/loss_hinge.html","id":null,"dir":"Reference","previous_headings":"","what":"Computes the hinge loss between y_true & y_pred. — loss_hinge","title":"Computes the hinge loss between y_true & y_pred. — loss_hinge","text":"Formula:   y_true values expected -1 1. binary (0 1) labels provided convert -1 1.","code":"loss <- mean(maximum(1 - y_true * y_pred, 0), axis=-1)"},{"path":"https://keras3.posit.co/dev/reference/loss_hinge.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes the hinge loss between y_true & y_pred. — loss_hinge","text":"","code":"loss_hinge(   y_true,   y_pred,   ...,   reduction = \"sum_over_batch_size\",   name = \"hinge\",   dtype = NULL )"},{"path":"https://keras3.posit.co/dev/reference/loss_hinge.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Computes the hinge loss between y_true & y_pred. — loss_hinge","text":"y_true ground truth values. y_true values expected -1 1. binary (0 1) labels provided converted -1 1 shape = [batch_size, d0, .. dN]. y_pred predicted values shape = [batch_size, d0, .. dN]. ... forward/backward compatability. reduction Type reduction apply loss. almost cases \"sum_over_batch_size\". Supported options \"sum\", \"sum_over_batch_size\", \"mean\", \"mean_with_sample_weight\" NULL. \"sum\" sums loss, \"sum_over_batch_size\" \"mean\" sum loss divide sample size, \"mean_with_sample_weight\" sums loss divides sum sample weights. \"none\" NULL perform aggregation. Defaults \"sum_over_batch_size\". name Optional name loss instance. dtype dtype loss's computations. Defaults NULL, means using config_floatx(). config_floatx() \"float32\" unless set different value (via config_set_floatx()). keras$DTypePolicy provided, compute_dtype utilized.","code":""},{"path":"https://keras3.posit.co/dev/reference/loss_hinge.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Computes the hinge loss between y_true & y_pred. — loss_hinge","text":"Hinge loss values shape = [batch_size, d0, .. dN-1].","code":""},{"path":"https://keras3.posit.co/dev/reference/loss_hinge.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Computes the hinge loss between y_true & y_pred. — loss_hinge","text":"","code":"y_true <- array(sample(c(-1,1), 6, replace = TRUE), dim = c(2, 3)) y_pred <- random_uniform(c(2, 3)) loss <- loss_hinge(y_true, y_pred) loss ## tf.Tensor([1.0610152  0.93285507], shape=(2), dtype=float32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/loss_huber.html","id":null,"dir":"Reference","previous_headings":"","what":"Computes the Huber loss between y_true & y_pred. — loss_huber","title":"Computes the Huber loss between y_true & y_pred. — loss_huber","text":"Formula:   See: Huber loss.","code":"for (x in error) {   if (abs(x) <= delta){     loss <- c(loss, (0.5 * x^2))   } else if (abs(x) > delta) {     loss <- c(loss, (delta * abs(x) - 0.5 * delta^2))   } } loss <- mean(loss)"},{"path":"https://keras3.posit.co/dev/reference/loss_huber.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes the Huber loss between y_true & y_pred. — loss_huber","text":"","code":"loss_huber(   y_true,   y_pred,   delta = 1,   ...,   reduction = \"sum_over_batch_size\",   name = \"huber_loss\",   dtype = NULL )"},{"path":"https://keras3.posit.co/dev/reference/loss_huber.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Computes the Huber loss between y_true & y_pred. — loss_huber","text":"y_true tensor true targets. y_pred tensor predicted targets. delta float, point Huber loss function changes quadratic linear. Defaults 1.0. ... forward/backward compatability. reduction Type reduction apply loss. Options \"sum\", \"sum_over_batch_size\" NULL. Defaults \"sum_over_batch_size\". name Optional name instance. dtype dtype loss's computations. Defaults NULL, means using config_floatx(). config_floatx() \"float32\" unless set different value (via config_set_floatx()). keras$DTypePolicy provided, compute_dtype utilized.","code":""},{"path":"https://keras3.posit.co/dev/reference/loss_huber.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Computes the Huber loss between y_true & y_pred. — loss_huber","text":"Tensor one scalar loss entry per sample.","code":""},{"path":"https://keras3.posit.co/dev/reference/loss_huber.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Computes the Huber loss between y_true & y_pred. — loss_huber","text":"","code":"y_true <- rbind(c(0, 1), c(0, 0)) y_pred <- rbind(c(0.6, 0.4), c(0.4, 0.6)) loss <- loss_huber(y_true, y_pred)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/loss_kl_divergence.html","id":null,"dir":"Reference","previous_headings":"","what":"Computes Kullback-Leibler divergence loss between y_true & y_pred. — loss_kl_divergence","title":"Computes Kullback-Leibler divergence loss between y_true & y_pred. — loss_kl_divergence","text":"Formula:   y_true y_pred expected probability distributions, values 0 1. get clipped [0, 1] range.","code":"loss <- y_true * log(y_true / y_pred)"},{"path":"https://keras3.posit.co/dev/reference/loss_kl_divergence.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes Kullback-Leibler divergence loss between y_true & y_pred. — loss_kl_divergence","text":"","code":"loss_kl_divergence(   y_true,   y_pred,   ...,   reduction = \"sum_over_batch_size\",   name = \"kl_divergence\",   dtype = NULL )"},{"path":"https://keras3.posit.co/dev/reference/loss_kl_divergence.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Computes Kullback-Leibler divergence loss between y_true & y_pred. — loss_kl_divergence","text":"y_true Tensor true targets. y_pred Tensor predicted targets. ... forward/backward compatability. reduction Type reduction apply loss. almost cases \"sum_over_batch_size\". Supported options \"sum\", \"sum_over_batch_size\", \"mean\", \"mean_with_sample_weight\" NULL. \"sum\" sums loss, \"sum_over_batch_size\" \"mean\" sum loss divide sample size, \"mean_with_sample_weight\" sums loss divides sum sample weights. \"none\" NULL perform aggregation. Defaults \"sum_over_batch_size\". name Optional name loss instance. dtype dtype loss's computations. Defaults NULL, means using config_floatx(). config_floatx() \"float32\" unless set different value (via config_set_floatx()). keras$DTypePolicy provided, compute_dtype utilized.","code":""},{"path":"https://keras3.posit.co/dev/reference/loss_kl_divergence.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Computes Kullback-Leibler divergence loss between y_true & y_pred. — loss_kl_divergence","text":"KL Divergence loss values shape = [batch_size, d0, .. dN-1].","code":""},{"path":"https://keras3.posit.co/dev/reference/loss_kl_divergence.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Computes Kullback-Leibler divergence loss between y_true & y_pred. — loss_kl_divergence","text":"","code":"y_true <- random_uniform(c(2, 3), 0, 2) y_pred <- random_uniform(c(2,3)) loss <- loss_kl_divergence(y_true, y_pred) loss ## tf.Tensor([ 2.4290292 -0.6284211], shape=(2), dtype=float32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/loss_log_cosh.html","id":null,"dir":"Reference","previous_headings":"","what":"Computes the logarithm of the hyperbolic cosine of the prediction error. — loss_log_cosh","title":"Computes the logarithm of the hyperbolic cosine of the prediction error. — loss_log_cosh","text":"Formula:   Note log(cosh(x)) approximately equal (x ** 2) / 2 small x abs(x) - log(2) large x. means 'logcosh' works mostly like mean squared error, strongly affected occasional wildly incorrect prediction.","code":"loss <- mean(log(cosh(y_pred - y_true)), axis=-1)"},{"path":"https://keras3.posit.co/dev/reference/loss_log_cosh.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes the logarithm of the hyperbolic cosine of the prediction error. — loss_log_cosh","text":"","code":"loss_log_cosh(   y_true,   y_pred,   ...,   reduction = \"sum_over_batch_size\",   name = \"log_cosh\",   dtype = NULL )"},{"path":"https://keras3.posit.co/dev/reference/loss_log_cosh.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Computes the logarithm of the hyperbolic cosine of the prediction error. — loss_log_cosh","text":"y_true Ground truth values shape = [batch_size, d0, .. dN]. y_pred predicted values shape = [batch_size, d0, .. dN]. ... forward/backward compatability. reduction Type reduction apply loss. Options \"sum\", \"sum_over_batch_size\" NULL. Defaults \"sum_over_batch_size\". name Optional name instance. dtype dtype loss's computations. Defaults NULL, means using config_floatx(). config_floatx() \"float32\" unless set different value (via config_set_floatx()). keras$DTypePolicy provided, compute_dtype utilized.","code":""},{"path":"https://keras3.posit.co/dev/reference/loss_log_cosh.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Computes the logarithm of the hyperbolic cosine of the prediction error. — loss_log_cosh","text":"Logcosh error values shape = [batch_size, d0, .. dN-1].","code":""},{"path":"https://keras3.posit.co/dev/reference/loss_log_cosh.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Computes the logarithm of the hyperbolic cosine of the prediction error. — loss_log_cosh","text":"","code":"y_true <- rbind(c(0., 1.), c(0., 0.)) y_pred <- rbind(c(1., 1.), c(0., 0.)) loss <- loss_log_cosh(y_true, y_pred) # 0.108"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/loss_mean_absolute_error.html","id":null,"dir":"Reference","previous_headings":"","what":"Computes the mean of absolute difference between labels and predictions. — loss_mean_absolute_error","title":"Computes the mean of absolute difference between labels and predictions. — loss_mean_absolute_error","text":"Formula:","code":"loss <- mean(abs(y_true - y_pred))"},{"path":"https://keras3.posit.co/dev/reference/loss_mean_absolute_error.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes the mean of absolute difference between labels and predictions. — loss_mean_absolute_error","text":"","code":"loss_mean_absolute_error(   y_true,   y_pred,   ...,   reduction = \"sum_over_batch_size\",   name = \"mean_absolute_error\",   dtype = NULL )"},{"path":"https://keras3.posit.co/dev/reference/loss_mean_absolute_error.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Computes the mean of absolute difference between labels and predictions. — loss_mean_absolute_error","text":"y_true Ground truth values shape = [batch_size, d0, .. dN]. y_pred predicted values shape = [batch_size, d0, .. dN]. ... forward/backward compatability. reduction Type reduction apply loss. almost cases \"sum_over_batch_size\". Supported options \"sum\", \"sum_over_batch_size\", \"mean\", \"mean_with_sample_weight\" NULL. \"sum\" sums loss, \"sum_over_batch_size\" \"mean\" sum loss divide sample size, \"mean_with_sample_weight\" sums loss divides sum sample weights. \"none\" NULL perform aggregation. Defaults \"sum_over_batch_size\". name Optional name loss instance. dtype dtype loss's computations. Defaults NULL, means using config_floatx(). config_floatx() \"float32\" unless set different value (via config_set_floatx()). keras$DTypePolicy provided, compute_dtype utilized.","code":""},{"path":"https://keras3.posit.co/dev/reference/loss_mean_absolute_error.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Computes the mean of absolute difference between labels and predictions. — loss_mean_absolute_error","text":"Mean absolute error values shape = [batch_size, d0, .. dN-1].","code":""},{"path":"https://keras3.posit.co/dev/reference/loss_mean_absolute_error.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Computes the mean of absolute difference between labels and predictions. — loss_mean_absolute_error","text":"","code":"y_true <- random_uniform(c(2, 3), 0, 2) y_pred <- random_uniform(c(2, 3)) loss <- loss_mean_absolute_error(y_true, y_pred)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/loss_mean_absolute_percentage_error.html","id":null,"dir":"Reference","previous_headings":"","what":"Computes the mean absolute percentage error between y_true and y_pred. — loss_mean_absolute_percentage_error","title":"Computes the mean absolute percentage error between y_true and y_pred. — loss_mean_absolute_percentage_error","text":"Formula:   Division zero prevented dividing max(y_true, epsilon) epsilon = config_epsilon() (default 1e-7).","code":"loss <- 100 * op_mean(op_abs((y_true - y_pred) / y_true),                       axis=-1)"},{"path":"https://keras3.posit.co/dev/reference/loss_mean_absolute_percentage_error.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes the mean absolute percentage error between y_true and y_pred. — loss_mean_absolute_percentage_error","text":"","code":"loss_mean_absolute_percentage_error(   y_true,   y_pred,   ...,   reduction = \"sum_over_batch_size\",   name = \"mean_absolute_percentage_error\",   dtype = NULL )"},{"path":"https://keras3.posit.co/dev/reference/loss_mean_absolute_percentage_error.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Computes the mean absolute percentage error between y_true and y_pred. — loss_mean_absolute_percentage_error","text":"y_true Ground truth values shape = [batch_size, d0, .. dN]. y_pred predicted values shape = [batch_size, d0, .. dN]. ... forward/backward compatability. reduction Type reduction apply loss. almost cases \"sum_over_batch_size\". Supported options \"sum\", \"sum_over_batch_size\", \"mean\", \"mean_with_sample_weight\" NULL. \"sum\" sums loss, \"sum_over_batch_size\" \"mean\" sum loss divide sample size, \"mean_with_sample_weight\" sums loss divides sum sample weights. \"none\" NULL perform aggregation. Defaults \"sum_over_batch_size\". name Optional name loss instance. dtype dtype loss's computations. Defaults NULL, means using config_floatx(). config_floatx() \"float32\" unless set different value (via config_set_floatx()). keras$DTypePolicy provided, compute_dtype utilized.","code":""},{"path":"https://keras3.posit.co/dev/reference/loss_mean_absolute_percentage_error.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Computes the mean absolute percentage error between y_true and y_pred. — loss_mean_absolute_percentage_error","text":"Mean absolute percentage error values shape = [batch_size, d0, ..dN-1].","code":""},{"path":"https://keras3.posit.co/dev/reference/loss_mean_absolute_percentage_error.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Computes the mean absolute percentage error between y_true and y_pred. — loss_mean_absolute_percentage_error","text":"","code":"y_true <- random_uniform(c(2, 3)) y_pred <- random_uniform(c(2, 3)) loss <- loss_mean_absolute_percentage_error(y_true, y_pred)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/loss_mean_squared_error.html","id":null,"dir":"Reference","previous_headings":"","what":"Computes the mean of squares of errors between labels and predictions. — loss_mean_squared_error","title":"Computes the mean of squares of errors between labels and predictions. — loss_mean_squared_error","text":"Formula:","code":"loss <- mean(square(y_true - y_pred))"},{"path":"https://keras3.posit.co/dev/reference/loss_mean_squared_error.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes the mean of squares of errors between labels and predictions. — loss_mean_squared_error","text":"","code":"loss_mean_squared_error(   y_true,   y_pred,   ...,   reduction = \"sum_over_batch_size\",   name = \"mean_squared_error\",   dtype = NULL )"},{"path":"https://keras3.posit.co/dev/reference/loss_mean_squared_error.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Computes the mean of squares of errors between labels and predictions. — loss_mean_squared_error","text":"y_true Ground truth values shape = [batch_size, d0, .. dN]. y_pred predicted values shape = [batch_size, d0, .. dN]. ... forward/backward compatability. reduction Type reduction apply loss. almost cases \"sum_over_batch_size\". Supported options \"sum\", \"sum_over_batch_size\", \"mean\", \"mean_with_sample_weight\" NULL. \"sum\" sums loss, \"sum_over_batch_size\" \"mean\" sum loss divide sample size, \"mean_with_sample_weight\" sums loss divides sum sample weights. \"none\" NULL perform aggregation. Defaults \"sum_over_batch_size\". name Optional name loss instance. dtype dtype loss's computations. Defaults NULL, means using config_floatx(). config_floatx() \"float32\" unless set different value (via config_set_floatx()). keras$DTypePolicy provided, compute_dtype utilized.","code":""},{"path":"https://keras3.posit.co/dev/reference/loss_mean_squared_error.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Computes the mean of squares of errors between labels and predictions. — loss_mean_squared_error","text":"Mean squared error values shape = [batch_size, d0, .. dN-1].","code":""},{"path":"https://keras3.posit.co/dev/reference/loss_mean_squared_error.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Computes the mean of squares of errors between labels and predictions. — loss_mean_squared_error","text":"","code":"y_true <- random_uniform(c(2, 3), 0, 2) y_pred <- random_uniform(c(2, 3)) loss <- loss_mean_squared_error(y_true, y_pred)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/loss_mean_squared_logarithmic_error.html","id":null,"dir":"Reference","previous_headings":"","what":"Computes the mean squared logarithmic error between y_true and y_pred. — loss_mean_squared_logarithmic_error","title":"Computes the mean squared logarithmic error between y_true and y_pred. — loss_mean_squared_logarithmic_error","text":"Note y_pred y_true less equal 0. Negative values 0 values replaced config_epsilon() (default 1e-7). Formula:","code":"loss <- mean(square(log(y_true + 1) - log(y_pred + 1)))"},{"path":"https://keras3.posit.co/dev/reference/loss_mean_squared_logarithmic_error.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes the mean squared logarithmic error between y_true and y_pred. — loss_mean_squared_logarithmic_error","text":"","code":"loss_mean_squared_logarithmic_error(   y_true,   y_pred,   ...,   reduction = \"sum_over_batch_size\",   name = \"mean_squared_logarithmic_error\",   dtype = NULL )"},{"path":"https://keras3.posit.co/dev/reference/loss_mean_squared_logarithmic_error.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Computes the mean squared logarithmic error between y_true and y_pred. — loss_mean_squared_logarithmic_error","text":"y_true Ground truth values shape = [batch_size, d0, .. dN]. y_pred predicted values shape = [batch_size, d0, .. dN]. ... forward/backward compatability. reduction Type reduction apply loss. almost cases \"sum_over_batch_size\". Supported options \"sum\", \"sum_over_batch_size\", \"mean\", \"mean_with_sample_weight\" NULL. \"sum\" sums loss, \"sum_over_batch_size\" \"mean\" sum loss divide sample size, \"mean_with_sample_weight\" sums loss divides sum sample weights. \"none\" NULL perform aggregation. Defaults \"sum_over_batch_size\". name Optional name loss instance. dtype dtype loss's computations. Defaults NULL, means using config_floatx(). config_floatx() \"float32\" unless set different value (via config_set_floatx()). keras$DTypePolicy provided, compute_dtype utilized.","code":""},{"path":"https://keras3.posit.co/dev/reference/loss_mean_squared_logarithmic_error.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Computes the mean squared logarithmic error between y_true and y_pred. — loss_mean_squared_logarithmic_error","text":"Mean squared logarithmic error values shape = [batch_size, d0, .. dN-1].","code":""},{"path":"https://keras3.posit.co/dev/reference/loss_mean_squared_logarithmic_error.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Computes the mean squared logarithmic error between y_true and y_pred. — loss_mean_squared_logarithmic_error","text":"","code":"y_true <- random_uniform(c(2, 3), 0, 2) y_pred <- random_uniform(c(2, 3)) loss <- loss_mean_squared_logarithmic_error(y_true, y_pred)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/loss_poisson.html","id":null,"dir":"Reference","previous_headings":"","what":"Computes the Poisson loss between y_true & y_pred. — loss_poisson","title":"Computes the Poisson loss between y_true & y_pred. — loss_poisson","text":"Formula:","code":"loss <- y_pred - y_true * log(y_pred)"},{"path":"https://keras3.posit.co/dev/reference/loss_poisson.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes the Poisson loss between y_true & y_pred. — loss_poisson","text":"","code":"loss_poisson(   y_true,   y_pred,   ...,   reduction = \"sum_over_batch_size\",   name = \"poisson\",   dtype = NULL )"},{"path":"https://keras3.posit.co/dev/reference/loss_poisson.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Computes the Poisson loss between y_true & y_pred. — loss_poisson","text":"y_true Ground truth values. shape = [batch_size, d0, .. dN]. y_pred predicted values. shape = [batch_size, d0, .. dN]. ... forward/backward compatability. reduction Type reduction apply loss. almost cases \"sum_over_batch_size\". Supported options \"sum\", \"sum_over_batch_size\", \"mean\", \"mean_with_sample_weight\" NULL. \"sum\" sums loss, \"sum_over_batch_size\" \"mean\" sum loss divide sample size, \"mean_with_sample_weight\" sums loss divides sum sample weights. \"none\" NULL perform aggregation. Defaults \"sum_over_batch_size\". name Optional name loss instance. dtype dtype loss's computations. Defaults NULL, means using config_floatx(). config_floatx() \"float32\" unless set different value (via config_set_floatx()). keras$DTypePolicy provided, compute_dtype utilized.","code":""},{"path":"https://keras3.posit.co/dev/reference/loss_poisson.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Computes the Poisson loss between y_true & y_pred. — loss_poisson","text":"Poisson loss values shape = [batch_size, d0, .. dN-1].","code":""},{"path":"https://keras3.posit.co/dev/reference/loss_poisson.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Computes the Poisson loss between y_true & y_pred. — loss_poisson","text":"","code":"y_true <- random_uniform(c(2, 3), 0, 2) y_pred <- random_uniform(c(2, 3)) loss <- loss_poisson(y_true, y_pred) loss ## tf.Tensor([1.6422468  0.81166863], shape=(2), dtype=float32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/loss_sparse_categorical_crossentropy.html","id":null,"dir":"Reference","previous_headings":"","what":"Computes the crossentropy loss between the labels and predictions. — loss_sparse_categorical_crossentropy","title":"Computes the crossentropy loss between the labels and predictions. — loss_sparse_categorical_crossentropy","text":"Use crossentropy loss function two label classes.  expect labels provided integers. want provide labels using one-hot representation, please use CategoricalCrossentropy loss.  # classes floating point values per feature y_pred single floating point value per feature y_true. snippet , single floating point value per example y_true num_classes floating pointing values per example y_pred. shape y_true [batch_size] shape y_pred [batch_size, num_classes].","code":""},{"path":"https://keras3.posit.co/dev/reference/loss_sparse_categorical_crossentropy.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes the crossentropy loss between the labels and predictions. — loss_sparse_categorical_crossentropy","text":"","code":"loss_sparse_categorical_crossentropy(   y_true,   y_pred,   from_logits = FALSE,   ignore_class = NULL,   axis = -1L,   ...,   reduction = \"sum_over_batch_size\",   name = \"sparse_categorical_crossentropy\",   dtype = NULL )"},{"path":"https://keras3.posit.co/dev/reference/loss_sparse_categorical_crossentropy.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Computes the crossentropy loss between the labels and predictions. — loss_sparse_categorical_crossentropy","text":"y_true Ground truth values. y_pred predicted values. from_logits Whether y_pred expected logits tensor. default, assume y_pred encodes probability distribution. ignore_class Optional integer. ID class ignored loss computation. useful, example, segmentation problems featuring \"void\" class (commonly -1 255) segmentation maps. default (ignore_class=NULL), classes considered. axis Defaults -1. dimension along entropy computed. ... forward/backward compatability. reduction Type reduction apply loss. almost cases \"sum_over_batch_size\". Supported options \"sum\", \"sum_over_batch_size\", \"mean\", \"mean_with_sample_weight\" NULL. \"sum\" sums loss, \"sum_over_batch_size\" \"mean\" sum loss divide sample size, \"mean_with_sample_weight\" sums loss divides sum sample weights. \"none\" NULL perform aggregation. Defaults \"sum_over_batch_size\". name Optional name loss instance. dtype dtype loss's computations. Defaults NULL, means using config_floatx(). config_floatx() \"float32\" unless set different value (via config_set_floatx()). keras$DTypePolicy provided, compute_dtype utilized.","code":""},{"path":"https://keras3.posit.co/dev/reference/loss_sparse_categorical_crossentropy.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Computes the crossentropy loss between the labels and predictions. — loss_sparse_categorical_crossentropy","text":"Sparse categorical crossentropy loss value.","code":""},{"path":"https://keras3.posit.co/dev/reference/loss_sparse_categorical_crossentropy.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Computes the crossentropy loss between the labels and predictions. — loss_sparse_categorical_crossentropy","text":"Usage compile() API:","code":"y_true <- op_array(c(1L, 2L)) y_pred <- op_array(rbind(c(0.05, 0.95, 0), c(0.1, 0.8, 0.1))) loss <- loss_sparse_categorical_crossentropy(y_true, y_pred) loss ## tf.Tensor([0.05129339 2.30258509], shape=(2), dtype=float64) y_true <- op_array(c(1L, 2L)) y_pred <- op_array(rbind(c(0.05, 0.95, 0), c(0.1, 0.8, 0.1))) # Using 'auto'/'sum_over_batch_size' reduction type. scce <- loss_sparse_categorical_crossentropy() scce(y_true, y_pred) ## tf.Tensor(1.1769392, shape=(), dtype=float32) # Calling with 'sample_weight'. scce(y_true, y_pred, sample_weight = op_array(c(0.3, 0.7))) ## tf.Tensor(0.8135988, shape=(), dtype=float32) # Using 'sum' reduction type. scce <- loss_sparse_categorical_crossentropy(reduction=\"sum\") scce(y_true, y_pred) ## tf.Tensor(2.3538785, shape=(), dtype=float32) # Using 'none' reduction type. scce <- loss_sparse_categorical_crossentropy(reduction=NULL) scce(y_true, y_pred) ## tf.Tensor([0.05129344 2.3025851 ], shape=(2), dtype=float32) model %>% compile(optimizer = 'sgd',                   loss = loss_sparse_categorical_crossentropy())"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/loss_squared_hinge.html","id":null,"dir":"Reference","previous_headings":"","what":"Computes the squared hinge loss between y_true & y_pred. — loss_squared_hinge","title":"Computes the squared hinge loss between y_true & y_pred. — loss_squared_hinge","text":"Formula:   y_true values expected -1 1. binary (0 1) labels provided convert -1 1.","code":"loss <- square(maximum(1 - y_true * y_pred, 0))"},{"path":"https://keras3.posit.co/dev/reference/loss_squared_hinge.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes the squared hinge loss between y_true & y_pred. — loss_squared_hinge","text":"","code":"loss_squared_hinge(   y_true,   y_pred,   ...,   reduction = \"sum_over_batch_size\",   name = \"squared_hinge\",   dtype = NULL )"},{"path":"https://keras3.posit.co/dev/reference/loss_squared_hinge.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Computes the squared hinge loss between y_true & y_pred. — loss_squared_hinge","text":"y_true ground truth values. y_true values expected -1 1. binary (0 1) labels provided convert -1 1 shape = [batch_size, d0, .. dN]. y_pred predicted values shape = [batch_size, d0, .. dN]. ... forward/backward compatability. reduction Type reduction apply loss. almost cases \"sum_over_batch_size\". Supported options \"sum\", \"sum_over_batch_size\", \"mean\", \"mean_with_sample_weight\" NULL. \"sum\" sums loss, \"sum_over_batch_size\" \"mean\" sum loss divide sample size, \"mean_with_sample_weight\" sums loss divides sum sample weights. \"none\" NULL perform aggregation. Defaults \"sum_over_batch_size\". name Optional name loss instance. dtype dtype loss's computations. Defaults NULL, means using config_floatx(). config_floatx() \"float32\" unless set different value (via config_set_floatx()). keras$DTypePolicy provided, compute_dtype utilized.","code":""},{"path":"https://keras3.posit.co/dev/reference/loss_squared_hinge.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Computes the squared hinge loss between y_true & y_pred. — loss_squared_hinge","text":"Squared hinge loss values shape = [batch_size, d0, .. dN-1].","code":""},{"path":"https://keras3.posit.co/dev/reference/loss_squared_hinge.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Computes the squared hinge loss between y_true & y_pred. — loss_squared_hinge","text":"","code":"y_true <- array(sample(c(-1,1), 6, replace = TRUE), dim = c(2, 3)) y_pred <- random_uniform(c(2, 3)) loss <- loss_squared_hinge(y_true, y_pred)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/loss_tversky.html","id":null,"dir":"Reference","previous_headings":"","what":"Computes the Tversky loss value between y_true and y_pred. — loss_tversky","title":"Computes the Tversky loss value between y_true and y_pred. — loss_tversky","text":"loss function weighted alpha beta coefficients penalize false positives false negatives. alpha=0.5 beta=0.5, loss value becomes equivalent Dice Loss. loss function weighted alpha beta coefficients penalize false positives false negatives. alpha=0.5 beta=0.5, loss value becomes equivalent Dice Loss.","code":""},{"path":"https://keras3.posit.co/dev/reference/loss_tversky.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes the Tversky loss value between y_true and y_pred. — loss_tversky","text":"","code":"loss_tversky(   y_true,   y_pred,   ...,   alpha = 0.5,   beta = 0.5,   reduction = \"sum_over_batch_size\",   name = \"tversky\",   axis = NULL,   dtype = NULL )"},{"path":"https://keras3.posit.co/dev/reference/loss_tversky.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Computes the Tversky loss value between y_true and y_pred. — loss_tversky","text":"y_true tensor true targets. y_pred tensor predicted targets. ... forward/backward compatability. alpha coefficient controlling incidence false positives. Defaults 0.5. beta coefficient controlling incidence false negatives. Defaults 0.5. reduction Type reduction apply loss. almost cases \"sum_over_batch_size\". Supported options \"sum\", \"sum_over_batch_size\", \"mean\", \"mean_with_sample_weight\" NULL. \"sum\" sums loss, \"sum_over_batch_size\" \"mean\" sum loss divide sample size, \"mean_with_sample_weight\" sums loss divides sum sample weights. \"none\" NULL perform aggregation. Defaults \"sum_over_batch_size\". name Optional name loss instance. (string) axis Axis (1-based) dtype dtype loss's computations. Defaults NULL, means using config_floatx(). config_floatx() \"float32\" unless set different value (via config_set_floatx()). keras$DTypePolicy provided, compute_dtype utilized.","code":""},{"path":"https://keras3.posit.co/dev/reference/loss_tversky.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Computes the Tversky loss value between y_true and y_pred. — loss_tversky","text":"Tversky loss value.","code":""},{"path":"https://keras3.posit.co/dev/reference/loss_tversky.html","id":"reference","dir":"Reference","previous_headings":"","what":"Reference","title":"Computes the Tversky loss value between y_true and y_pred. — loss_tversky","text":"Salehi et al., 2017","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/mark_active.html","id":null,"dir":"Reference","previous_headings":"","what":"active_property — mark_active","title":"active_property — mark_active","text":"mark_active() alias active_property(). See ?active_property() full documentation.","code":""},{"path":"https://keras3.posit.co/dev/reference/mark_active.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"active_property — mark_active","text":"","code":"mark_active(fn)"},{"path":"https://keras3.posit.co/dev/reference/mark_active.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"active_property — mark_active","text":"fn R function","code":""},{"path":"https://keras3.posit.co/dev/reference/mark_active.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"active_property — mark_active","text":"fn, additional R attribute cause fn converted active property converted method custom subclass.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_auc.html","id":null,"dir":"Reference","previous_headings":"","what":"Approximates the AUC (Area under the curve) of the ROC or PR curves. — metric_auc","title":"Approximates the AUC (Area under the curve) of the ROC or PR curves. — metric_auc","text":"AUC (Area curve) ROC (Receiver operating characteristic; default) PR (Precision Recall) curves quality measures binary classifiers. Unlike accuracy, like cross-entropy losses, ROC-AUC PR-AUC evaluate operational points model. class approximates AUCs using Riemann sum. metric accumulation phrase, predictions accumulated within predefined buckets value. AUC computed interpolating per-bucket averages. buckets define evaluated operational points. metric creates four local variables, true_positives, true_negatives, false_positives false_negatives used compute AUC.  discretize AUC curve, linearly spaced set thresholds used compute pairs recall precision values. area ROC-curve therefore computed using height recall values false positive rate, area PR-curve computed using height precision values recall. value ultimately returned auc, idempotent operation computes area discretized curve precision versus recall values (computed using aforementioned variables). num_thresholds variable controls degree discretization larger numbers thresholds closely approximating true AUC. quality approximation may vary dramatically depending num_thresholds. thresholds parameter can used manually specify thresholds split predictions evenly. best approximation real AUC, predictions distributed approximately uniformly range [0, 1] (from_logits=FALSE). quality AUC approximation may poor case. Setting summation_method 'minoring' 'majoring' can help quantify error approximation providing lower upper bound estimate AUC. sample_weight NULL, weights default 1. Use sample_weight 0 mask values.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_auc.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Approximates the AUC (Area under the curve) of the ROC or PR curves. — metric_auc","text":"","code":"metric_auc(   ...,   num_thresholds = 200L,   curve = \"ROC\",   summation_method = \"interpolation\",   name = NULL,   dtype = NULL,   thresholds = NULL,   multi_label = FALSE,   num_labels = NULL,   label_weights = NULL,   from_logits = FALSE )"},{"path":"https://keras3.posit.co/dev/reference/metric_auc.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Approximates the AUC (Area under the curve) of the ROC or PR curves. — metric_auc","text":"... forward/backward compatability. num_thresholds (Optional) number thresholds use discretizing roc curve. Values must > 1. Defaults 200. curve (Optional) Specifies name curve computed, 'ROC' (default) 'PR' Precision-Recall-curve. summation_method (Optional) Specifies Riemann summation method used. 'interpolation' (default) applies mid-point summation scheme ROC.  PR-AUC, interpolates (true/false) positives ratio precision (see Davis & Goadrich 2006 details); 'minoring' applies left summation increasing intervals right summation decreasing intervals; 'majoring' opposite. name (Optional) string name metric instance. dtype (Optional) data type metric result. thresholds (Optional) list floating point values use thresholds discretizing curve. set, num_thresholds parameter ignored. Values [0, 1]. Endpoint thresholds equal {-epsilon, 1+epsilon} small positive epsilon value automatically included correctly handle predictions equal exactly 0 1. multi_label boolean indicating whether multilabel data treated , wherein AUC computed separately label averaged across labels, (FALSE) data flattened single label AUC computation. latter case, multilabel data passed AUC, label-prediction pair treated individual data point. set `FALSE“ multi-class data. num_labels (Optional) number labels, used multi_label TRUE. num_labels specified, state variables get created first call update_state. label_weights (Optional) list, array, tensor non-negative weights used compute AUCs multilabel data. multi_label TRUE, weights applied individual label AUCs averaged produce multi-label AUC. FALSE, used weight individual label predictions computing confusion matrix flattened data. Note unlike class_weights class_weights weights example depending value label, whereas label_weights depends index label flattening; therefore label_weights used multi-class data. from_logits boolean indicating whether predictions (y_pred update_state) probabilities sigmoid logits. rule thumb, using keras loss, from_logits constructor argument loss match AUC from_logits constructor argument.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_auc.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Approximates the AUC (Area under the curve) of the ROC or PR curves. — metric_auc","text":"Metric instance returned. Metric instance can passed directly compile(metrics = ), used standalone object. See ?Metric example usage.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_auc.html","id":"usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Approximates the AUC (Area under the curve) of the ROC or PR curves. — metric_auc","text":"Standalone usage:         Usage compile() API:","code":"m <- metric_auc(num_thresholds = 3) m$update_state(c(0,   0,   1,   1),                c(0, 0.5, 0.3, 0.9)) # threshold values are [0 - 1e-7, 0.5, 1 + 1e-7] # tp = [2, 1, 0], fp = [2, 0, 0], fn = [0, 1, 2], tn = [0, 2, 2] # tp_rate = recall = [1, 0.5, 0], fp_rate = [1, 0, 0] # auc = ((((1 + 0.5) / 2) * (1 - 0)) + (((0.5 + 0) / 2) * (0 - 0))) #     = 0.75 m$result() ## tf.Tensor(0.75, shape=(), dtype=float32) m$reset_state() m$update_state(c(0,   0,   1,   1),                c(0, 0.5, 0.3, 0.9),                sample_weight=c(1, 0, 0, 1)) m$result() ## tf.Tensor(1.0, shape=(), dtype=float32) # Reports the AUC of a model outputting a probability. model |> compile(   optimizer = 'sgd',   loss = loss_binary_crossentropy(),   metrics = list(metric_auc()) )  # Reports the AUC of a model outputting a logit. model |> compile(   optimizer = 'sgd',   loss = loss_binary_crossentropy(from_logits = TRUE),   metrics = list(metric_auc(from_logits = TRUE)) )"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/metric_binary_accuracy.html","id":null,"dir":"Reference","previous_headings":"","what":"Calculates how often predictions match binary labels. — metric_binary_accuracy","title":"Calculates how often predictions match binary labels. — metric_binary_accuracy","text":"metric creates two local variables, total count used compute frequency y_pred matches y_true. frequency ultimately returned binary accuracy: idempotent operation simply divides total count. sample_weight NULL, weights default 1. Use sample_weight 0 mask values.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_binary_accuracy.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Calculates how often predictions match binary labels. — metric_binary_accuracy","text":"","code":"metric_binary_accuracy(   y_true,   y_pred,   threshold = 0.5,   ...,   name = \"binary_accuracy\",   dtype = NULL )"},{"path":"https://keras3.posit.co/dev/reference/metric_binary_accuracy.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Calculates how often predictions match binary labels. — metric_binary_accuracy","text":"y_true Tensor true targets. y_pred Tensor predicted targets. threshold (Optional) Float representing threshold deciding whether prediction values 1 0. ... forward/backward compatability. name (Optional) string name metric instance. dtype (Optional) data type metric result.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_binary_accuracy.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Calculates how often predictions match binary labels. — metric_binary_accuracy","text":"y_true y_pred missing, Metric instance returned. Metric instance can passed directly compile(metrics = ), used standalone object. See ?Metric example usage. y_true y_pred provided, tensor computed value returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_binary_accuracy.html","id":"usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Calculates how often predictions match binary labels. — metric_binary_accuracy","text":"Standalone usage:             Usage compile() API:","code":"m <- metric_binary_accuracy() m$update_state(rbind(1, 1, 0, 0), rbind(0.98, 1, 0, 0.6)) m$result() ## tf.Tensor(0.75, shape=(), dtype=float32) # 0.75 m$reset_state() m$update_state(rbind(1, 1, 0, 0), rbind(0.98, 1, 0, 0.6),                sample_weight = c(1, 0, 0, 1)) m$result() ## tf.Tensor(0.5, shape=(), dtype=float32) # 0.5 model %>% compile(optimizer='sgd',                   loss='binary_crossentropy',                   metrics=list(metric_binary_accuracy()))"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/metric_binary_crossentropy.html","id":null,"dir":"Reference","previous_headings":"","what":"Computes the crossentropy metric between the labels and predictions. — metric_binary_crossentropy","title":"Computes the crossentropy metric between the labels and predictions. — metric_binary_crossentropy","text":"crossentropy metric class used two label classes (0 1).","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_binary_crossentropy.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes the crossentropy metric between the labels and predictions. — metric_binary_crossentropy","text":"","code":"metric_binary_crossentropy(   y_true,   y_pred,   from_logits = FALSE,   label_smoothing = 0,   axis = -1L,   ...,   name = \"binary_crossentropy\",   dtype = NULL )"},{"path":"https://keras3.posit.co/dev/reference/metric_binary_crossentropy.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Computes the crossentropy metric between the labels and predictions. — metric_binary_crossentropy","text":"y_true Ground truth values. shape = [batch_size, d0, .. dN]. y_pred predicted values. shape = [batch_size, d0, .. dN]. from_logits (Optional) Whether output expected logits tensor. default, consider output encodes probability distribution. label_smoothing (Optional) Float [0, 1]. > 0, label values smoothed, meaning confidence label values relaxed. e.g. label_smoothing=0.2 means use value 0.1 label \"0\" 0.9 label \"1\". axis axis along mean computed. Defaults -1. ... forward/backward compatability. name (Optional) string name metric instance. dtype (Optional) data type metric result.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_binary_crossentropy.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Computes the crossentropy metric between the labels and predictions. — metric_binary_crossentropy","text":"y_true y_pred missing, Metric instance returned. Metric instance can passed directly compile(metrics = ), used standalone object. See ?Metric example usage. y_true y_pred provided, tensor computed value returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_binary_crossentropy.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Computes the crossentropy metric between the labels and predictions. — metric_binary_crossentropy","text":"Standalone usage:         Usage compile() API:","code":"m <- metric_binary_crossentropy() m$update_state(rbind(c(0, 1), c(0, 0)), rbind(c(0.6, 0.4), c(0.4, 0.6))) m$result() ## tf.Tensor(0.8149245, shape=(), dtype=float32) m$reset_state() m$update_state(rbind(c(0, 1), c(0, 0)), rbind(c(0.6, 0.4), c(0.4, 0.6)),                sample_weight=c(1, 0)) m$result() ## tf.Tensor(0.91629076, shape=(), dtype=float32) model %>% compile(     optimizer = 'sgd',     loss = 'mse',     metrics = list(metric_binary_crossentropy()))"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/metric_binary_focal_crossentropy.html","id":null,"dir":"Reference","previous_headings":"","what":"Computes the binary focal crossentropy loss. — metric_binary_focal_crossentropy","title":"Computes the binary focal crossentropy loss. — metric_binary_focal_crossentropy","text":"According Lin et al., 2018, helps apply focal factor -weight easy examples focus hard examples. default, focal tensor computed follows: focal_factor = (1 - output)^gamma class 1 focal_factor = output^gamma class 0 gamma focusing parameter. gamma = 0, focal effect binary crossentropy loss. apply_class_balancing == TRUE, function also takes account weight balancing factor binary classes 0 1 follows: weight = alpha class 1 (target == 1) weight = 1 - alpha class 0 alpha float range [0, 1].","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_binary_focal_crossentropy.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes the binary focal crossentropy loss. — metric_binary_focal_crossentropy","text":"","code":"metric_binary_focal_crossentropy(   y_true,   y_pred,   apply_class_balancing = FALSE,   alpha = 0.25,   gamma = 2,   from_logits = FALSE,   label_smoothing = 0,   axis = -1L )"},{"path":"https://keras3.posit.co/dev/reference/metric_binary_focal_crossentropy.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Computes the binary focal crossentropy loss. — metric_binary_focal_crossentropy","text":"y_true Ground truth values, shape (batch_size, d0, .. dN). y_pred predicted values, shape (batch_size, d0, .. dN). apply_class_balancing bool, whether apply weight balancing binary classes 0 1. alpha weight balancing factor class 1, default 0.25 mentioned reference. weight class 0 1.0 - alpha. gamma focusing parameter, default 2.0 mentioned reference. from_logits Whether y_pred expected logits tensor. default, assume y_pred encodes probability distribution. label_smoothing Float [0, 1]. > 0 smooth labels squeezing towards 0.5, , using 1. - 0.5 * label_smoothing target class 0.5 * label_smoothing non-target class. axis axis along mean computed. Defaults -1.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_binary_focal_crossentropy.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Computes the binary focal crossentropy loss. — metric_binary_focal_crossentropy","text":"Binary focal crossentropy loss value shape = [batch_size, d0, .. dN-1].","code":""},{"path":[]},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/metric_binary_iou.html","id":null,"dir":"Reference","previous_headings":"","what":"Computes the Intersection-Over-Union metric for class 0 and/or 1. — metric_binary_iou","title":"Computes the Intersection-Over-Union metric for class 0 and/or 1. — metric_binary_iou","text":"Formula:   Intersection--Union common evaluation metric semantic image segmentation. compute IoUs, predictions accumulated confusion matrix, weighted sample_weight metric calculated . sample_weight NULL, weights default 1. Use sample_weight 0 mask values. class can used compute IoUs binary classification task predictions provided logits. First threshold applied predicted values threshold converted class 0 threshold converted class 1. IoUs classes 0 1 computed, mean IoUs classes specified target_class_ids returned.","code":"iou <- true_positives / (true_positives + false_positives + false_negatives)"},{"path":"https://keras3.posit.co/dev/reference/metric_binary_iou.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes the Intersection-Over-Union metric for class 0 and/or 1. — metric_binary_iou","text":"","code":"metric_binary_iou(   ...,   target_class_ids = list(0L, 1L),   threshold = 0.5,   name = NULL,   dtype = NULL )"},{"path":"https://keras3.posit.co/dev/reference/metric_binary_iou.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Computes the Intersection-Over-Union metric for class 0 and/or 1. — metric_binary_iou","text":"... forward/backward compatability. target_class_ids list list target class ids metric returned. Options 0, 1, c(0, 1). 0 (1), IoU metric class 0 (class 1, respectively) returned. c(0, 1), mean IoUs two classes returned. threshold threshold applies prediction logits convert either predicted class 0 logit threshold predicted class 1 logit threshold. name (Optional) string name metric instance. dtype (Optional) data type metric result.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_binary_iou.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Computes the Intersection-Over-Union metric for class 0 and/or 1. — metric_binary_iou","text":"Metric instance returned. Metric instance can passed directly compile(metrics = ), used standalone object. See ?Metric example usage.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_binary_iou.html","id":"note","dir":"Reference","previous_headings":"","what":"Note","title":"Computes the Intersection-Over-Union metric for class 0 and/or 1. — metric_binary_iou","text":"threshold=0, metric behavior IoU.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_binary_iou.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Computes the Intersection-Over-Union metric for class 0 and/or 1. — metric_binary_iou","text":"Standalone usage:                 Usage compile() API:","code":"m <- metric_binary_iou(target_class_ids=c(0L, 1L), threshold = 0.3) m$update_state(c(0, 1, 0, 1), c(0.1, 0.2, 0.4, 0.7)) ## tf.Tensor( ## [[1 1] ##  [1 1]], shape=(2, 2), dtype=int64) m$result() ## tf.Tensor(0.33333334, shape=(), dtype=float32) m$reset_state() m$update_state(c(0, 1, 0, 1), c(0.1, 0.2, 0.4, 0.7),                sample_weight = 10 * c(0.2, 0.3, 0.4, 0.1)) ## tf.Tensor( ## [[2 4] ##  [3 1]], shape=(2, 2), dtype=int64) m$result() ## tf.Tensor(0.1736111, shape=(), dtype=float32) model %>% compile(     optimizer = 'sgd',     loss = 'mse',     metrics = list(metric_binary_iou(         target_class_ids = 0L,         threshold = 0.5     )) )"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/metric_categorical_accuracy.html","id":null,"dir":"Reference","previous_headings":"","what":"Calculates how often predictions match one-hot labels. — metric_categorical_accuracy","title":"Calculates how often predictions match one-hot labels. — metric_categorical_accuracy","text":"can provide logits classes y_pred, since argmax logits probabilities . metric creates two local variables, total count used compute frequency y_pred matches y_true. frequency ultimately returned categorical accuracy: idempotent operation simply divides total count. y_pred y_true passed vectors probabilities, rather labels. necessary, use op_one_hot expand y_true vector. sample_weight NULL, weights default 1. Use sample_weight 0 mask values.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_categorical_accuracy.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Calculates how often predictions match one-hot labels. — metric_categorical_accuracy","text":"","code":"metric_categorical_accuracy(   y_true,   y_pred,   ...,   name = \"categorical_accuracy\",   dtype = NULL )"},{"path":"https://keras3.posit.co/dev/reference/metric_categorical_accuracy.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Calculates how often predictions match one-hot labels. — metric_categorical_accuracy","text":"y_true Tensor true targets. y_pred Tensor predicted targets. ... forward/backward compatability. name (Optional) string name metric instance. dtype (Optional) data type metric result.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_categorical_accuracy.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Calculates how often predictions match one-hot labels. — metric_categorical_accuracy","text":"y_true y_pred missing, Metric instance returned. Metric instance can passed directly compile(metrics = ), used standalone object. See ?Metric example usage. y_true y_pred provided, tensor computed value returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_categorical_accuracy.html","id":"usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Calculates how often predictions match one-hot labels. — metric_categorical_accuracy","text":"Standalone usage:           Usage compile() API:","code":"m <- metric_categorical_accuracy() m$update_state(rbind(c(0, 0, 1), c(0, 1, 0)), rbind(c(0.1, 0.9, 0.8),                 c(0.05, 0.95, 0))) m$result() ## tf.Tensor(0.5, shape=(), dtype=float32) m$reset_state() m$update_state(rbind(c(0, 0, 1), c(0, 1, 0)), rbind(c(0.1, 0.9, 0.8),                c(0.05, 0.95, 0)),                sample_weight = c(0.7, 0.3)) m$result() ## tf.Tensor(0.3, shape=(), dtype=float32) # 0.3 model %>% compile(optimizer = 'sgd',                   loss = 'categorical_crossentropy',                   metrics = list(metric_categorical_accuracy()))"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/metric_categorical_crossentropy.html","id":null,"dir":"Reference","previous_headings":"","what":"Computes the crossentropy metric between the labels and predictions. — metric_categorical_crossentropy","title":"Computes the crossentropy metric between the labels and predictions. — metric_categorical_crossentropy","text":"crossentropy metric class used multiple label classes (2 ). assumes labels one-hot encoded, e.g., labels values c(2, 0, 1), y_true rbind(c([0, 0, 1), c(1, 0, 0), c(0, 1, 0)).","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_categorical_crossentropy.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes the crossentropy metric between the labels and predictions. — metric_categorical_crossentropy","text":"","code":"metric_categorical_crossentropy(   y_true,   y_pred,   from_logits = FALSE,   label_smoothing = 0,   axis = -1L,   ...,   name = \"categorical_crossentropy\",   dtype = NULL )"},{"path":"https://keras3.posit.co/dev/reference/metric_categorical_crossentropy.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Computes the crossentropy metric between the labels and predictions. — metric_categorical_crossentropy","text":"y_true Tensor one-hot true targets. y_pred Tensor predicted targets. from_logits (Optional) Whether output expected logits tensor. default, consider output encodes probability distribution. label_smoothing (Optional) Float [0, 1]. > 0, label values smoothed, meaning confidence label values relaxed. e.g. label_smoothing=0.2 means use value 0.1 label \"0\" 0.9 label \"1\". axis (Optional) Defaults -1. dimension along entropy computed. ... forward/backward compatability. name (Optional) string name metric instance. dtype (Optional) data type metric result.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_categorical_crossentropy.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Computes the crossentropy metric between the labels and predictions. — metric_categorical_crossentropy","text":"y_true y_pred missing, Metric instance returned. Metric instance can passed directly compile(metrics = ), used standalone object. See ?Metric example usage. y_true y_pred provided, tensor computed value returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_categorical_crossentropy.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Computes the crossentropy metric between the labels and predictions. — metric_categorical_crossentropy","text":"Standalone usage:           Usage compile() API:","code":"# EPSILON = 1e-7, y = y_true, y` = y_pred # y` = clip_op_clip_by_value(output, EPSILON, 1. - EPSILON) # y` = rbind(c(0.05, 0.95, EPSILON), c(0.1, 0.8, 0.1)) # xent = -sum(y * log(y'), axis = -1) #      = -((log 0.95), (log 0.1)) #      = [0.051, 2.302] # Reduced xent = (0.051 + 2.302) / 2  m <- metric_categorical_crossentropy() m$update_state(rbind(c(0, 1, 0), c(0, 0, 1)),                rbind(c(0.05, 0.95, 0), c(0.1, 0.8, 0.1))) m$result() ## tf.Tensor(1.1769392, shape=(), dtype=float32) # 1.1769392 m$reset_state() m$update_state(rbind(c(0, 1, 0), c(0, 0, 1)),                rbind(c(0.05, 0.95, 0), c(0.1, 0.8, 0.1)),                sample_weight = c(0.3, 0.7)) m$result() ## tf.Tensor(1.6271976, shape=(), dtype=float32) model %>% compile(   optimizer = 'sgd',   loss = 'mse',   metrics = list(metric_categorical_crossentropy()))"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/metric_categorical_focal_crossentropy.html","id":null,"dir":"Reference","previous_headings":"","what":"Computes the categorical focal crossentropy loss. — metric_categorical_focal_crossentropy","title":"Computes the categorical focal crossentropy loss. — metric_categorical_focal_crossentropy","text":"Computes categorical focal crossentropy loss.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_categorical_focal_crossentropy.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes the categorical focal crossentropy loss. — metric_categorical_focal_crossentropy","text":"","code":"metric_categorical_focal_crossentropy(   y_true,   y_pred,   alpha = 0.25,   gamma = 2,   from_logits = FALSE,   label_smoothing = 0,   axis = -1L )"},{"path":"https://keras3.posit.co/dev/reference/metric_categorical_focal_crossentropy.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Computes the categorical focal crossentropy loss. — metric_categorical_focal_crossentropy","text":"y_true Tensor one-hot true targets. y_pred Tensor predicted targets. alpha weight balancing factor classes, default 0.25 mentioned reference. can list floats scalar. multi-class case, alpha may set inverse class frequency using compute_class_weight sklearn.utils. gamma focusing parameter, default 2.0 mentioned reference. helps gradually reduce importance given simple examples smooth manner. gamma = 0, focal effect categorical crossentropy. from_logits Whether y_pred expected logits tensor. default, assume y_pred encodes probability distribution. label_smoothing Float [0, 1]. > 0 smooth labels. example, 0.1, use 0.1 / num_classes non-target labels 0.9 + 0.1 / num_classes target labels. axis Defaults -1. dimension along entropy computed.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_categorical_focal_crossentropy.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Computes the categorical focal crossentropy loss. — metric_categorical_focal_crossentropy","text":"Categorical focal crossentropy loss value.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_categorical_focal_crossentropy.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Computes the categorical focal crossentropy loss. — metric_categorical_focal_crossentropy","text":"","code":"y_true <- rbind(c(0, 1, 0), c(0, 0, 1)) y_pred <- rbind(c(0.05, 0.9, 0.05), c(0.1, 0.85, 0.05)) loss <- loss_categorical_focal_crossentropy(y_true, y_pred) loss ## tf.Tensor([2.6340148e-04 6.7591208e-01], shape=(2), dtype=float32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/metric_categorical_hinge.html","id":null,"dir":"Reference","previous_headings":"","what":"Computes the categorical hinge metric between y_true and y_pred. — metric_categorical_hinge","title":"Computes the categorical hinge metric between y_true and y_pred. — metric_categorical_hinge","text":"Formula:   neg=maximum((1-y_true)*y_pred) pos=sum(y_true*y_pred)","code":"loss <- maximum(neg - pos + 1, 0)"},{"path":"https://keras3.posit.co/dev/reference/metric_categorical_hinge.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes the categorical hinge metric between y_true and y_pred. — metric_categorical_hinge","text":"","code":"metric_categorical_hinge(   y_true,   y_pred,   ...,   name = \"categorical_hinge\",   dtype = NULL )"},{"path":"https://keras3.posit.co/dev/reference/metric_categorical_hinge.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Computes the categorical hinge metric between y_true and y_pred. — metric_categorical_hinge","text":"y_true ground truth values. y_true values expected either {-1, +1} {0, 1} (.e. one-hot-encoded tensor) shape = [batch_size, d0, .. dN]. y_pred predicted values shape = [batch_size, d0, .. dN]. ... forward/backward compatability. name (Optional) string name metric instance. dtype (Optional) data type metric result.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_categorical_hinge.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Computes the categorical hinge metric between y_true and y_pred. — metric_categorical_hinge","text":"Categorical hinge loss values shape = [batch_size, d0, .. dN-1].","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_categorical_hinge.html","id":"usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes the categorical hinge metric between y_true and y_pred. — metric_categorical_hinge","text":"Standalone usage:","code":"m <- metric_categorical_hinge() m$update_state(rbind(c(0, 1), c(0, 0)), rbind(c(0.6, 0.4), c(0.4, 0.6))) m$result() ## tf.Tensor(1.4000001, shape=(), dtype=float32) m$reset_state() m$update_state(rbind(c(0, 1), c(0, 0)), rbind(c(0.6, 0.4), c(0.4, 0.6)),                sample_weight = c(1, 0)) m$result() ## tf.Tensor(1.2, shape=(), dtype=float32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/metric_concordance_correlation.html","id":null,"dir":"Reference","previous_headings":"","what":"Calculates the Concordance Correlation Coefficient (CCC). — metric_concordance_correlation","title":"Calculates the Concordance Correlation Coefficient (CCC). — metric_concordance_correlation","text":"Formula:   CCC evaluates agreement true values (y_true) predicted values (y_pred) considering precision accuracy. coefficient ranges -1 1, value 1 indicates perfect agreement. metric useful regression tasks important assess well predictions match true values, taking account correlation proximity 45-degree line perfect concordance.","code":"loss <- mean(   2 * (y_true - mean(y_true)) * (y_pred - mean(y_pred)) /     (var(y_true) + var(y_pred) + (mean(y_true) - mean(y_pred))^2) )"},{"path":"https://keras3.posit.co/dev/reference/metric_concordance_correlation.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Calculates the Concordance Correlation Coefficient (CCC). — metric_concordance_correlation","text":"","code":"metric_concordance_correlation(   y_true,   y_pred,   axis = -1L,   ...,   name = \"concordance_correlation\",   dtype = NULL )"},{"path":"https://keras3.posit.co/dev/reference/metric_concordance_correlation.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Calculates the Concordance Correlation Coefficient (CCC). — metric_concordance_correlation","text":"y_true Tensor true targets. y_pred Tensor predicted targets. axis (Optional) integer tuple integers axis/axes along compute metric. Defaults -1. ... forward/backward compatability. name (Optional) string name metric instance. dtype (Optional) data type metric result.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_concordance_correlation.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Calculates the Concordance Correlation Coefficient (CCC). — metric_concordance_correlation","text":"Usage compile() API:","code":"ccc <- metric_concordance_correlation(axis=-1) y_true <- rbind(c(0, 1, 0.5),                 c(1, 1, 0.2)) y_pred <- rbind(c(0.1, 0.9, 0.5),                 c(1, 0.9, 0.2)) ccc$update_state(y_true, y_pred) ccc$result() ## tf.Tensor(0.9816317, shape=(), dtype=float32) model |> compile(   optimizer = 'sgd',   loss = 'mean_squared_error',   metrics = c(metric_concordance_correlation()) )"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/metric_cosine_similarity.html","id":null,"dir":"Reference","previous_headings":"","what":"Computes the cosine similarity between the labels and predictions. — metric_cosine_similarity","title":"Computes the cosine similarity between the labels and predictions. — metric_cosine_similarity","text":"Formula:   See: Cosine Similarity. metric keeps average cosine similarity predictions labels stream data.","code":"loss <- sum(l2_norm(y_true) * l2_norm(y_pred))"},{"path":"https://keras3.posit.co/dev/reference/metric_cosine_similarity.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes the cosine similarity between the labels and predictions. — metric_cosine_similarity","text":"","code":"metric_cosine_similarity(   ...,   name = \"cosine_similarity\",   dtype = NULL,   axis = -1L )"},{"path":"https://keras3.posit.co/dev/reference/metric_cosine_similarity.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Computes the cosine similarity between the labels and predictions. — metric_cosine_similarity","text":"... forward/backward compatability. name (Optional) string name metric instance. dtype (Optional) data type metric result. axis (Optional) Defaults -1. dimension along cosine similarity computed.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_cosine_similarity.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Computes the cosine similarity between the labels and predictions. — metric_cosine_similarity","text":"Metric instance returned. Metric instance can passed directly compile(metrics = ), used standalone object. See ?Metric example usage.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_cosine_similarity.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Computes the cosine similarity between the labels and predictions. — metric_cosine_similarity","text":"Standalone usage:         Usage compile() API:","code":"m <- metric_cosine_similarity(axis=2) m$update_state(rbind(c(0., 1.), c(1., 1.)), rbind(c(1., 0.), c(1., 1.))) m$result() ## tf.Tensor(0.5, shape=(), dtype=float32) m$reset_state() m$update_state(rbind(c(0., 1.), c(1., 1.)), rbind(c(1., 0.), c(1., 1.)),                sample_weight = c(0.3, 0.7)) m$result() ## tf.Tensor(0.7, shape=(), dtype=float32) model %>% compile(   optimizer = 'sgd',   loss = 'mse',   metrics = list(metric_cosine_similarity(axis=2)))"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/metric_f1_score.html","id":null,"dir":"Reference","previous_headings":"","what":"Computes F-1 Score. — metric_f1_score","title":"Computes F-1 Score. — metric_f1_score","text":"Formula:   harmonic mean precision recall. output range [0, 1]. works multi-class multi-label classification.","code":"f1_score <- 2 * (precision * recall) / (precision + recall)"},{"path":"https://keras3.posit.co/dev/reference/metric_f1_score.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes F-1 Score. — metric_f1_score","text":"","code":"metric_f1_score(   ...,   average = NULL,   threshold = NULL,   name = \"f1_score\",   dtype = NULL )"},{"path":"https://keras3.posit.co/dev/reference/metric_f1_score.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Computes F-1 Score. — metric_f1_score","text":"... forward/backward compatability. average Type averaging performed data. Acceptable values NULL, \"micro\", \"macro\" \"weighted\". Defaults NULL. NULL, averaging performed result() return score class. \"micro\", compute metrics globally counting total true positives, false negatives false positives. \"macro\", compute metrics label, return unweighted mean. take label imbalance account. \"weighted\", compute metrics label, return average weighted support (number true instances label). alters \"macro\" account label imbalance. can result score precision recall. threshold Elements y_pred greater threshold converted 1, rest 0. threshold NULL, argmax y_pred converted 1, rest 0. name Optional. String name metric instance. dtype Optional. Data type metric result.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_f1_score.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Computes F-1 Score. — metric_f1_score","text":"Metric instance returned. Metric instance can passed directly compile(metrics = ), used standalone object. See ?Metric example usage.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_f1_score.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Computes F-1 Score. — metric_f1_score","text":"","code":"metric <- metric_f1_score(threshold = 0.5) y_true <- rbind(c(1, 1, 1),                 c(1, 0, 0),                 c(1, 1, 0)) y_pred <- rbind(c(0.2, 0.6, 0.7),                 c(0.2, 0.6, 0.6),                 c(0.6, 0.8, 0.0)) metric$update_state(y_true, y_pred) result <- metric$result() result ## tf.Tensor([0.49999997 0.79999995 0.66666657], shape=(3), dtype=float32)"},{"path":"https://keras3.posit.co/dev/reference/metric_f1_score.html","id":"returns","dir":"Reference","previous_headings":"","what":"Returns","title":"Computes F-1 Score. — metric_f1_score","text":"F-1 Score: float.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/metric_false_negatives.html","id":null,"dir":"Reference","previous_headings":"","what":"Calculates the number of false negatives. — metric_false_negatives","title":"Calculates the number of false negatives. — metric_false_negatives","text":"sample_weight given, calculates sum weights false negatives. metric creates one local variable, accumulator used keep track number false negatives. sample_weight NULL, weights default 1. Use sample_weight 0 mask values.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_false_negatives.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Calculates the number of false negatives. — metric_false_negatives","text":"","code":"metric_false_negatives(..., thresholds = NULL, name = NULL, dtype = NULL)"},{"path":"https://keras3.posit.co/dev/reference/metric_false_negatives.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Calculates the number of false negatives. — metric_false_negatives","text":"... forward/backward compatability. thresholds (Optional) Defaults 0.5. float value, Python list float threshold values [0, 1]. threshold compared prediction values determine truth value predictions (.e., threshold TRUE, FALSE). used loss function sets from_logits=TRUE (.e. sigmoid applied predictions), thresholds set 0. One metric value generated threshold value. name (Optional) string name metric instance. dtype (Optional) data type metric result.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_false_negatives.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Calculates the number of false negatives. — metric_false_negatives","text":"Metric instance returned. Metric instance can passed directly compile(metrics = ), used standalone object. See ?Metric example usage.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_false_negatives.html","id":"usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Calculates the number of false negatives. — metric_false_negatives","text":"Standalone usage:","code":"m <- metric_false_negatives() m$update_state(c(0, 1, 1, 1), c(0, 1, 0, 0)) m$result() ## tf.Tensor(2.0, shape=(), dtype=float32) m$reset_state() m$update_state(c(0, 1, 1, 1), c(0, 1, 0, 0), sample_weight=c(0, 0, 1, 0)) m$result() ## tf.Tensor(1.0, shape=(), dtype=float32) # 1.0"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/metric_false_positives.html","id":null,"dir":"Reference","previous_headings":"","what":"Calculates the number of false positives. — metric_false_positives","title":"Calculates the number of false positives. — metric_false_positives","text":"sample_weight given, calculates sum weights false positives. metric creates one local variable, accumulator used keep track number false positives. sample_weight NULL, weights default 1. Use sample_weight 0 mask values.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_false_positives.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Calculates the number of false positives. — metric_false_positives","text":"","code":"metric_false_positives(..., thresholds = NULL, name = NULL, dtype = NULL)"},{"path":"https://keras3.posit.co/dev/reference/metric_false_positives.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Calculates the number of false positives. — metric_false_positives","text":"... forward/backward compatability. thresholds (Optional) Defaults 0.5. float value, Python list float threshold values [0, 1]. threshold compared prediction values determine truth value predictions (.e., threshold TRUE, FALSE). used loss function sets from_logits=TRUE (.e. sigmoid applied predictions), thresholds set 0. One metric value generated threshold value. name (Optional) string name metric instance. dtype (Optional) data type metric result.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_false_positives.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Calculates the number of false positives. — metric_false_positives","text":"Metric instance returned. Metric instance can passed directly compile(metrics = ), used standalone object. See ?Metric example usage.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_false_positives.html","id":"usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Calculates the number of false positives. — metric_false_positives","text":"Standalone usage:","code":"m <- metric_false_positives() m$update_state(c(0, 1, 0, 0), c(0, 0, 1, 1)) m$result() ## tf.Tensor(2.0, shape=(), dtype=float32) m$reset_state() m$update_state(c(0, 1, 0, 0), c(0, 0, 1, 1), sample_weight = c(0, 0, 1, 0)) m$result() ## tf.Tensor(1.0, shape=(), dtype=float32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/metric_fbeta_score.html","id":null,"dir":"Reference","previous_headings":"","what":"Computes F-Beta score. — metric_fbeta_score","title":"Computes F-Beta score. — metric_fbeta_score","text":"Formula:   weighted harmonic mean precision recall. output range [0, 1]. works multi-class multi-label classification.","code":"b2 <- beta^2 f_beta_score <- (1 + b2) * (precision * recall) / (precision * b2 + recall)"},{"path":"https://keras3.posit.co/dev/reference/metric_fbeta_score.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes F-Beta score. — metric_fbeta_score","text":"","code":"metric_fbeta_score(   ...,   average = NULL,   beta = 1,   threshold = NULL,   name = \"fbeta_score\",   dtype = NULL )"},{"path":"https://keras3.posit.co/dev/reference/metric_fbeta_score.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Computes F-Beta score. — metric_fbeta_score","text":"... forward/backward compatability. average Type averaging performed across per-class results multi-class case. Acceptable values NULL, \"micro\", \"macro\" \"weighted\". Defaults NULL. NULL, averaging performed result() return score class. \"micro\", compute metrics globally counting total true positives, false negatives false positives. \"macro\", compute metrics label, return unweighted mean. take label imbalance account. \"weighted\", compute metrics label, return average weighted support (number true instances label). alters \"macro\" account label imbalance. can result score precision recall. beta Determines weight given recall harmonic mean precision recall (see pseudocode equation ). Defaults 1. threshold Elements y_pred greater threshold converted 1, rest 0. threshold NULL, argmax y_pred converted 1, rest 0. name Optional. String name metric instance. dtype Optional. Data type metric result.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_fbeta_score.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Computes F-Beta score. — metric_fbeta_score","text":"Metric instance returned. Metric instance can passed directly compile(metrics = ), used standalone object. See ?Metric example usage.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_fbeta_score.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Computes F-Beta score. — metric_fbeta_score","text":"","code":"metric <- metric_fbeta_score(beta = 2.0, threshold = 0.5) y_true <- rbind(c(1, 1, 1),                 c(1, 0, 0),                 c(1, 1, 0)) y_pred <- rbind(c(0.2, 0.6, 0.7),                 c(0.2, 0.6, 0.6),                 c(0.6, 0.8, 0.0)) metric$update_state(y_true, y_pred) metric$result() ## tf.Tensor([0.3846154  0.90909094 0.8333332 ], shape=(3), dtype=float32)"},{"path":"https://keras3.posit.co/dev/reference/metric_fbeta_score.html","id":"returns","dir":"Reference","previous_headings":"","what":"Returns","title":"Computes F-Beta score. — metric_fbeta_score","text":"F-Beta Score: float.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/metric_hinge.html","id":null,"dir":"Reference","previous_headings":"","what":"Computes the hinge metric between y_true and y_pred. — metric_hinge","title":"Computes the hinge metric between y_true and y_pred. — metric_hinge","text":"Formula:   y_true values expected -1 1. binary (0 1) labels provided convert -1 1.","code":"loss <- mean(maximum(1 - y_true * y_pred, 0), axis=-1)"},{"path":"https://keras3.posit.co/dev/reference/metric_hinge.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes the hinge metric between y_true and y_pred. — metric_hinge","text":"","code":"metric_hinge(y_true, y_pred, ..., name = \"hinge\", dtype = NULL)"},{"path":"https://keras3.posit.co/dev/reference/metric_hinge.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Computes the hinge metric between y_true and y_pred. — metric_hinge","text":"y_true ground truth values. y_true values expected -1 1. binary (0 1) labels provided converted -1 1 shape = [batch_size, d0, .. dN]. y_pred predicted values shape = [batch_size, d0, .. dN]. ... forward/backward compatability. name (Optional) string name metric instance. dtype (Optional) data type metric result.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_hinge.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Computes the hinge metric between y_true and y_pred. — metric_hinge","text":"y_true y_pred missing, Metric instance returned. Metric instance can passed directly compile(metrics = ), used standalone object. See ?Metric example usage. y_true y_pred provided, tensor computed value returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_hinge.html","id":"usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes the hinge metric between y_true and y_pred. — metric_hinge","text":"Standalone usage:","code":"m <- metric_hinge() m$update_state(rbind(c(0, 1), c(0, 0)), rbind(c(0.6, 0.4), c(0.4, 0.6))) m$result() ## tf.Tensor(1.3, shape=(), dtype=float32) m$reset_state() m$update_state(rbind(c(0, 1), c(0, 0)), rbind(c(0.6, 0.4), c(0.4, 0.6)),                sample_weight = c(1, 0)) m$result() ## tf.Tensor(1.1, shape=(), dtype=float32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/metric_huber.html","id":null,"dir":"Reference","previous_headings":"","what":"Computes Huber loss value. — metric_huber","title":"Computes Huber loss value. — metric_huber","text":"Formula:   See: Huber loss.","code":"for (x in error) {   if (abs(x) <= delta){     loss <- c(loss, (0.5 * x^2))   } else if (abs(x) > delta) {     loss <- c(loss, (delta * abs(x) - 0.5 * delta^2))   } } loss <- mean(loss)"},{"path":"https://keras3.posit.co/dev/reference/metric_huber.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes Huber loss value. — metric_huber","text":"","code":"metric_huber(y_true, y_pred, delta = 1)"},{"path":"https://keras3.posit.co/dev/reference/metric_huber.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Computes Huber loss value. — metric_huber","text":"y_true tensor true targets. y_pred tensor predicted targets. delta float, point Huber loss function changes quadratic linear. Defaults 1.0.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_huber.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Computes Huber loss value. — metric_huber","text":"Tensor one scalar loss entry per sample.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_huber.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Computes Huber loss value. — metric_huber","text":"","code":"y_true <- rbind(c(0, 1), c(0, 0)) y_pred <- rbind(c(0.6, 0.4), c(0.4, 0.6)) loss <- loss_huber(y_true, y_pred)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/metric_iou.html","id":null,"dir":"Reference","previous_headings":"","what":"Computes the Intersection-Over-Union metric for specific target classes. — metric_iou","title":"Computes the Intersection-Over-Union metric for specific target classes. — metric_iou","text":"Formula:   Intersection--Union common evaluation metric semantic image segmentation. compute IoUs, predictions accumulated confusion matrix, weighted sample_weight metric calculated . sample_weight NULL, weights default 1. Use sample_weight 0 mask values. Note, class first computes IoUs individual classes, returns mean IoUs classes specified target_class_ids. target_class_ids one id value, IoU specific class returned.","code":"iou <- true_positives / (true_positives + false_positives + false_negatives)"},{"path":"https://keras3.posit.co/dev/reference/metric_iou.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes the Intersection-Over-Union metric for specific target classes. — metric_iou","text":"","code":"metric_iou(   ...,   num_classes,   target_class_ids,   name = NULL,   dtype = NULL,   ignore_class = NULL,   sparse_y_true = TRUE,   sparse_y_pred = TRUE,   axis = -1L )"},{"path":"https://keras3.posit.co/dev/reference/metric_iou.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Computes the Intersection-Over-Union metric for specific target classes. — metric_iou","text":"... forward/backward compatability. num_classes possible number labels prediction task can . target_class_ids list target class ids metric returned. compute IoU specific class, list single id value provided. name (Optional) string name metric instance. dtype (Optional) data type metric result. ignore_class Optional integer. ID class ignored metric computation. useful, example, segmentation problems featuring \"void\" class (commonly -1 255) segmentation maps. default (ignore_class=NULL), classes considered. sparse_y_true Whether labels encoded using integers dense floating point vectors. FALSE, argmax function used determine sample's likely associated label. sparse_y_pred Whether predictions encoded using integers dense floating point vectors. FALSE, argmax function used determine sample's likely associated label. axis (Optional) -1 dimension containing logits. Defaults -1.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_iou.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Computes the Intersection-Over-Union metric for specific target classes. — metric_iou","text":"Metric instance returned. Metric instance can passed directly compile(metrics = ), used standalone object. See ?Metric example usage.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_iou.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Computes the Intersection-Over-Union metric for specific target classes. — metric_iou","text":"Standalone usage:                 Usage compile() API:","code":"m <- metric_iou(num_classes = 2L, target_class_ids = list(0L)) m$update_state(c(0, 0, 1, 1), c(0, 1, 0, 1)) ## tf.Tensor( ## [[1 1] ##  [1 1]], shape=(2, 2), dtype=int64) m$result() ## tf.Tensor(0.3333333, shape=(), dtype=float32) m$reset_state() m$update_state(c(0, 0, 1, 1), c(0, 1, 0, 1),                sample_weight = 10 * c(0.3, 0.3, 0.3, 0.1)) ## tf.Tensor( ## [[3 3] ##  [3 1]], shape=(2, 2), dtype=int64) m$result() ## tf.Tensor(0.3333333, shape=(), dtype=float32) model %>% compile(   optimizer = 'sgd',   loss = 'mse',   metrics = list(metric_iou(num_classes = 2L, target_class_ids = list(0L))))"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/metric_kl_divergence.html","id":null,"dir":"Reference","previous_headings":"","what":"Computes Kullback-Leibler divergence metric between y_true and — metric_kl_divergence","title":"Computes Kullback-Leibler divergence metric between y_true and — metric_kl_divergence","text":"Formula:   y_true y_pred expected probability distributions, values 0 1. get clipped [0, 1] range.","code":"loss <- y_true * log(y_true / y_pred)"},{"path":"https://keras3.posit.co/dev/reference/metric_kl_divergence.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes Kullback-Leibler divergence metric between y_true and — metric_kl_divergence","text":"","code":"metric_kl_divergence(y_true, y_pred, ..., name = \"kl_divergence\", dtype = NULL)"},{"path":"https://keras3.posit.co/dev/reference/metric_kl_divergence.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Computes Kullback-Leibler divergence metric between y_true and — metric_kl_divergence","text":"y_true Tensor true targets. y_pred Tensor predicted targets. ... forward/backward compatability. name (Optional) string name metric instance. dtype (Optional) data type metric result.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_kl_divergence.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Computes Kullback-Leibler divergence metric between y_true and — metric_kl_divergence","text":"y_true y_pred missing, Metric instance returned. Metric instance can passed directly compile(metrics = ), used standalone object. See ?Metric example usage. y_true y_pred provided, tensor computed value returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_kl_divergence.html","id":"usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes Kullback-Leibler divergence metric between y_true and — metric_kl_divergence","text":"Standalone usage:         Usage compile() API:","code":"m <- metric_kl_divergence() m$update_state(rbind(c(0, 1), c(0, 0)), rbind(c(0.6, 0.4), c(0.4, 0.6))) m$result() ## tf.Tensor(0.45814303, shape=(), dtype=float32) m$reset_state() m$update_state(rbind(c(0, 1), c(0, 0)), rbind(c(0.6, 0.4), c(0.4, 0.6)),                sample_weight = c(1, 0)) m$result() ## tf.Tensor(0.91628915, shape=(), dtype=float32) model %>% compile(optimizer = 'sgd',                   loss = 'mse',                   metrics = list(metric_kl_divergence()))"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/metric_log_cosh.html","id":null,"dir":"Reference","previous_headings":"","what":"Logarithm of the hyperbolic cosine of the prediction error. — metric_log_cosh","title":"Logarithm of the hyperbolic cosine of the prediction error. — metric_log_cosh","text":"Formula:   Note log(cosh(x)) approximately equal (x ** 2) / 2 small x abs(x) - log(2) large x. means 'logcosh' works mostly like mean squared error, strongly affected occasional wildly incorrect prediction.","code":"loss <- mean(log(cosh(y_pred - y_true)), axis=-1)"},{"path":"https://keras3.posit.co/dev/reference/metric_log_cosh.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Logarithm of the hyperbolic cosine of the prediction error. — metric_log_cosh","text":"","code":"metric_log_cosh(y_true, y_pred)"},{"path":"https://keras3.posit.co/dev/reference/metric_log_cosh.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Logarithm of the hyperbolic cosine of the prediction error. — metric_log_cosh","text":"y_true Ground truth values shape = [batch_size, d0, .. dN]. y_pred predicted values shape = [batch_size, d0, .. dN].","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_log_cosh.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Logarithm of the hyperbolic cosine of the prediction error. — metric_log_cosh","text":"Logcosh error values shape = [batch_size, d0, .. dN-1].","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_log_cosh.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Logarithm of the hyperbolic cosine of the prediction error. — metric_log_cosh","text":"","code":"y_true <- rbind(c(0., 1.), c(0., 0.)) y_pred <- rbind(c(1., 1.), c(0., 0.)) loss <- metric_log_cosh(y_true, y_pred) loss ## tf.Tensor([0.2168904 0.       ], shape=(2), dtype=float32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/metric_log_cosh_error.html","id":null,"dir":"Reference","previous_headings":"","what":"Computes the logarithm of the hyperbolic cosine of the prediction error. — metric_log_cosh_error","title":"Computes the logarithm of the hyperbolic cosine of the prediction error. — metric_log_cosh_error","text":"Formula:","code":"error <- y_pred - y_true logcosh <- mean(log((exp(error) + exp(-error))/2), axis=-1)"},{"path":"https://keras3.posit.co/dev/reference/metric_log_cosh_error.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes the logarithm of the hyperbolic cosine of the prediction error. — metric_log_cosh_error","text":"","code":"metric_log_cosh_error(..., name = \"logcosh\", dtype = NULL)"},{"path":"https://keras3.posit.co/dev/reference/metric_log_cosh_error.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Computes the logarithm of the hyperbolic cosine of the prediction error. — metric_log_cosh_error","text":"... forward/backward compatability. name (Optional) string name metric instance. dtype (Optional) data type metric result.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_log_cosh_error.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Computes the logarithm of the hyperbolic cosine of the prediction error. — metric_log_cosh_error","text":"Metric instance returned. Metric instance can passed directly compile(metrics = ), used standalone object. See ?Metric example usage.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_log_cosh_error.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Computes the logarithm of the hyperbolic cosine of the prediction error. — metric_log_cosh_error","text":"Standalone usage:         Usage compile() API:","code":"m <- metric_log_cosh_error() m$update_state(rbind(c(0, 1), c(0, 0)), rbind(c(1, 1), c(0, 0))) m$result() ## tf.Tensor(0.108445205, shape=(), dtype=float32) m$reset_state() m$update_state(rbind(c(0, 1), c(0, 0)), rbind(c(1, 1), c(0, 0)),                sample_weight = c(1, 0)) m$result() ## tf.Tensor(0.21689041, shape=(), dtype=float32) model %>% compile(optimizer = 'sgd',                   loss = 'mse',                   metrics = list(metric_log_cosh_error()))"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/metric_mean.html","id":null,"dir":"Reference","previous_headings":"","what":"Compute the (weighted) mean of the given values. — metric_mean","title":"Compute the (weighted) mean of the given values. — metric_mean","text":"example, values c(1, 3, 5, 7) mean 4. sample_weight specified c(1, 1, 0, 0) mean 2. metric creates two variables, total count. mean value returned simply total divided count.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_mean.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Compute the (weighted) mean of the given values. — metric_mean","text":"","code":"metric_mean(..., name = \"mean\", dtype = NULL)"},{"path":"https://keras3.posit.co/dev/reference/metric_mean.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Compute the (weighted) mean of the given values. — metric_mean","text":"... forward/backward compatability. name (Optional) string name metric instance. dtype (Optional) data type metric result.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_mean.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Compute the (weighted) mean of the given values. — metric_mean","text":"Metric instance returned. Metric instance can passed directly compile(metrics = ), used standalone object. See ?Metric example usage.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_mean.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Compute the (weighted) mean of the given values. — metric_mean","text":"","code":"m <- metric_mean() m$update_state(c(1, 3, 5, 7)) m$result() ## tf.Tensor(4.0, shape=(), dtype=float32) # calling a metric directly is equivalent to calling # m$update_state(); m$result() m <- metric_mean() m(c(1, 3, 5, 7)) ## tf.Tensor(4.0, shape=(), dtype=float32) m$reset_state() m$update_state(c(1, 3, 5, 7), sample_weight = c(1, 1, 0, 0)) m$result() ## tf.Tensor(2.0, shape=(), dtype=float32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/metric_mean_absolute_error.html","id":null,"dir":"Reference","previous_headings":"","what":"Computes the mean absolute error between the labels and predictions. — metric_mean_absolute_error","title":"Computes the mean absolute error between the labels and predictions. — metric_mean_absolute_error","text":"Formula:","code":"loss <- mean(abs(y_true - y_pred))"},{"path":"https://keras3.posit.co/dev/reference/metric_mean_absolute_error.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes the mean absolute error between the labels and predictions. — metric_mean_absolute_error","text":"","code":"metric_mean_absolute_error(   y_true,   y_pred,   ...,   name = \"mean_absolute_error\",   dtype = NULL )"},{"path":"https://keras3.posit.co/dev/reference/metric_mean_absolute_error.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Computes the mean absolute error between the labels and predictions. — metric_mean_absolute_error","text":"y_true Ground truth values shape = [batch_size, d0, .. dN]. y_pred predicted values shape = [batch_size, d0, .. dN]. ... forward/backward compatability. name (Optional) string name metric instance. dtype (Optional) data type metric result.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_mean_absolute_error.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Computes the mean absolute error between the labels and predictions. — metric_mean_absolute_error","text":"y_true y_pred missing, Metric instance returned. Metric instance can passed directly compile(metrics = ), used standalone object. See ?Metric example usage. y_true y_pred provided, tensor computed value returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_mean_absolute_error.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Computes the mean absolute error between the labels and predictions. — metric_mean_absolute_error","text":"Standalone usage:         Usage compile() API:","code":"m <- metric_mean_absolute_error() m$update_state(rbind(c(0, 1), c(0, 0)), rbind(c(1, 1), c(0, 0))) m$result() ## tf.Tensor(0.25, shape=(), dtype=float32) m$reset_state() m$update_state(rbind(c(0, 1), c(0, 0)), rbind(c(1, 1), c(0, 0)),                sample_weight = c(1, 0)) m$result() ## tf.Tensor(0.5, shape=(), dtype=float32) model %>% compile(     optimizer = 'sgd',     loss = 'mse',     metrics = list(metric_mean_absolute_error()))"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/metric_mean_absolute_percentage_error.html","id":null,"dir":"Reference","previous_headings":"","what":"Computes mean absolute percentage error between y_true and y_pred. — metric_mean_absolute_percentage_error","title":"Computes mean absolute percentage error between y_true and y_pred. — metric_mean_absolute_percentage_error","text":"Formula:   Division zero prevented dividing maximum(y_true, epsilon) epsilon = keras$backend$epsilon() (default 1e-7).","code":"loss <- 100 * mean(abs((y_true - y_pred) / y_true), axis=-1)"},{"path":"https://keras3.posit.co/dev/reference/metric_mean_absolute_percentage_error.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes mean absolute percentage error between y_true and y_pred. — metric_mean_absolute_percentage_error","text":"","code":"metric_mean_absolute_percentage_error(   y_true,   y_pred,   ...,   name = \"mean_absolute_percentage_error\",   dtype = NULL )"},{"path":"https://keras3.posit.co/dev/reference/metric_mean_absolute_percentage_error.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Computes mean absolute percentage error between y_true and y_pred. — metric_mean_absolute_percentage_error","text":"y_true Ground truth values shape = [batch_size, d0, .. dN]. y_pred predicted values shape = [batch_size, d0, .. dN]. ... forward/backward compatability. name (Optional) string name metric instance. dtype (Optional) data type metric result.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_mean_absolute_percentage_error.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Computes mean absolute percentage error between y_true and y_pred. — metric_mean_absolute_percentage_error","text":"y_true y_pred missing, Metric instance returned. Metric instance can passed directly compile(metrics = ), used standalone object. See ?Metric example usage. y_true y_pred provided, tensor computed value returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_mean_absolute_percentage_error.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Computes mean absolute percentage error between y_true and y_pred. — metric_mean_absolute_percentage_error","text":"Standalone usage:         Usage compile() API:","code":"m <- metric_mean_absolute_percentage_error() m$update_state(rbind(c(0, 1), c(0, 0)), rbind(c(1, 1), c(0, 0))) m$result() ## tf.Tensor(2.5e+08, shape=(), dtype=float32) m$reset_state() m$update_state(rbind(c(0, 1), c(0, 0)), rbind(c(1, 1), c(0, 0)),                sample_weight = c(1, 0)) m$result() ## tf.Tensor(5e+08, shape=(), dtype=float32) model %>% compile(     optimizer = 'sgd',     loss = 'mse',     metrics = list(metric_mean_absolute_percentage_error()))"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/metric_mean_iou.html","id":null,"dir":"Reference","previous_headings":"","what":"Computes the mean Intersection-Over-Union metric. — metric_mean_iou","title":"Computes the mean Intersection-Over-Union metric. — metric_mean_iou","text":"Formula:   Intersection--Union common evaluation metric semantic image segmentation. compute IoUs, predictions accumulated confusion matrix, weighted sample_weight metric calculated . sample_weight NULL, weights default 1. Use sample_weight 0 mask values. Note class first computes IoUs individual classes, returns mean values.","code":"iou <- true_positives / (true_positives + false_positives + false_negatives)"},{"path":"https://keras3.posit.co/dev/reference/metric_mean_iou.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes the mean Intersection-Over-Union metric. — metric_mean_iou","text":"","code":"metric_mean_iou(   ...,   num_classes,   name = NULL,   dtype = NULL,   ignore_class = NULL,   sparse_y_true = TRUE,   sparse_y_pred = TRUE,   axis = -1L )"},{"path":"https://keras3.posit.co/dev/reference/metric_mean_iou.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Computes the mean Intersection-Over-Union metric. — metric_mean_iou","text":"... forward/backward compatability. num_classes possible number labels prediction task can . value must provided, since confusion matrix dimension = [num_classes, num_classes] allocated. name (Optional) string name metric instance. dtype (Optional) data type metric result. ignore_class Optional integer. ID class ignored metric computation. useful, example, segmentation problems featuring \"void\" class (commonly -1 255) segmentation maps. default (ignore_class=NULL), classes considered. sparse_y_true Whether labels encoded using integers dense floating point vectors. FALSE, argmax function used determine sample's likely associated label. sparse_y_pred Whether predictions encoded using integers dense floating point vectors. FALSE, argmax function used determine sample's likely associated label. axis (Optional) dimension containing logits. Defaults -1.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_mean_iou.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Computes the mean Intersection-Over-Union metric. — metric_mean_iou","text":"Metric instance returned. Metric instance can passed directly compile(metrics = ), used standalone object. See ?Metric example usage.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_mean_iou.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Computes the mean Intersection-Over-Union metric. — metric_mean_iou","text":"Standalone usage:                 Usage compile() API:","code":"# cm = [[1, 1], #       [1, 1]] # sum_row = [2, 2], sum_col = [2, 2], true_positives = [1, 1] # iou = true_positives / (sum_row + sum_col - true_positives)) # result = (1 / (2 + 2 - 1) + 1 / (2 + 2 - 1)) / 2 = 0.33 m <- metric_mean_iou(num_classes = 2) m$update_state(c(0, 0, 1, 1), c(0, 1, 0, 1)) ## tf.Tensor( ## [[1 1] ##  [1 1]], shape=(2, 2), dtype=int64) m$result() ## tf.Tensor(0.33333334, shape=(), dtype=float32) m$reset_state() m$update_state(c(0, 0, 1, 1), c(0, 1, 0, 1),                sample_weight= 10 * c(0.3, 0.3, 0.3, 0.1)) ## tf.Tensor( ## [[3 3] ##  [3 1]], shape=(2, 2), dtype=int64) m$result() ## tf.Tensor(0.23809525, shape=(), dtype=float32) model %>% compile(   optimizer = 'sgd',   loss = 'mse',   metrics = list(metric_mean_iou(num_classes=2)))"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/metric_mean_squared_error.html","id":null,"dir":"Reference","previous_headings":"","what":"Computes the mean squared error between y_true and y_pred. — metric_mean_squared_error","title":"Computes the mean squared error between y_true and y_pred. — metric_mean_squared_error","text":"Formula:","code":"loss <- mean(square(y_true - y_pred))"},{"path":"https://keras3.posit.co/dev/reference/metric_mean_squared_error.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes the mean squared error between y_true and y_pred. — metric_mean_squared_error","text":"","code":"metric_mean_squared_error(   y_true,   y_pred,   ...,   name = \"mean_squared_error\",   dtype = NULL )"},{"path":"https://keras3.posit.co/dev/reference/metric_mean_squared_error.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Computes the mean squared error between y_true and y_pred. — metric_mean_squared_error","text":"y_true Ground truth values shape = [batch_size, d0, .. dN]. y_pred predicted values shape = [batch_size, d0, .. dN]. ... forward/backward compatability. name (Optional) string name metric instance. dtype (Optional) data type metric result.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_mean_squared_error.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Computes the mean squared error between y_true and y_pred. — metric_mean_squared_error","text":"y_true y_pred missing, Metric instance returned. Metric instance can passed directly compile(metrics = ), used standalone object. See ?Metric example usage. y_true y_pred provided, tensor computed value returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_mean_squared_error.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Computes the mean squared error between y_true and y_pred. — metric_mean_squared_error","text":"","code":"m <- metric_mean_squared_error() m$update_state(rbind(c(0, 1), c(0, 0)), rbind(c(1, 1), c(0, 0))) m$result() ## tf.Tensor(0.25, shape=(), dtype=float32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/metric_mean_squared_logarithmic_error.html","id":null,"dir":"Reference","previous_headings":"","what":"Computes mean squared logarithmic error between y_true and y_pred. — metric_mean_squared_logarithmic_error","title":"Computes mean squared logarithmic error between y_true and y_pred. — metric_mean_squared_logarithmic_error","text":"Formula:   Note y_pred y_true less equal 0. Negative values 0 values replaced keras$backend$epsilon() (default 1e-7).","code":"loss <- mean(square(log(y_true + 1) - log(y_pred + 1)), axis=-1)"},{"path":"https://keras3.posit.co/dev/reference/metric_mean_squared_logarithmic_error.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes mean squared logarithmic error between y_true and y_pred. — metric_mean_squared_logarithmic_error","text":"","code":"metric_mean_squared_logarithmic_error(   y_true,   y_pred,   ...,   name = \"mean_squared_logarithmic_error\",   dtype = NULL )"},{"path":"https://keras3.posit.co/dev/reference/metric_mean_squared_logarithmic_error.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Computes mean squared logarithmic error between y_true and y_pred. — metric_mean_squared_logarithmic_error","text":"y_true Ground truth values shape = [batch_size, d0, .. dN]. y_pred predicted values shape = [batch_size, d0, .. dN]. ... forward/backward compatability. name (Optional) string name metric instance. dtype (Optional) data type metric result.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_mean_squared_logarithmic_error.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Computes mean squared logarithmic error between y_true and y_pred. — metric_mean_squared_logarithmic_error","text":"y_true y_pred missing, Metric instance returned. Metric instance can passed directly compile(metrics = ), used standalone object. See ?Metric example usage. y_true y_pred provided, tensor computed value returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_mean_squared_logarithmic_error.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Computes mean squared logarithmic error between y_true and y_pred. — metric_mean_squared_logarithmic_error","text":"Standalone usage:         Usage compile() API:","code":"m <- metric_mean_squared_logarithmic_error() m$update_state(rbind(c(0, 1), c(0, 0)), rbind(c(1, 1), c(0, 0))) m$result() ## tf.Tensor(0.12011322, shape=(), dtype=float32) m$reset_state() m$update_state(rbind(c(0, 1), c(0, 0)), rbind(c(1, 1), c(0, 0)),                sample_weight = c(1, 0)) m$result() ## tf.Tensor(0.24022643, shape=(), dtype=float32) model %>% compile(   optimizer = 'sgd',   loss = 'mse',   metrics = list(metric_mean_squared_logarithmic_error()))"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/metric_mean_wrapper.html","id":null,"dir":"Reference","previous_headings":"","what":"Wrap a stateless metric function with the Mean metric. — metric_mean_wrapper","title":"Wrap a stateless metric function with the Mean metric. — metric_mean_wrapper","text":"use class quickly build mean metric function. function needs signature fn(y_true, y_pred) return per-sample loss array. metric_mean_wrapper$result() return average metric value across samples seen far. example:","code":"mse <- function(y_true, y_pred) {   (y_true - y_pred)^2 }  mse_metric <- metric_mean_wrapper(fn = mse) mse_metric$update_state(c(0, 1), c(1, 1)) mse_metric$result() ## tf.Tensor(0.5, shape=(), dtype=float32)"},{"path":"https://keras3.posit.co/dev/reference/metric_mean_wrapper.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Wrap a stateless metric function with the Mean metric. — metric_mean_wrapper","text":"","code":"metric_mean_wrapper(..., fn, name = NULL, dtype = NULL)"},{"path":"https://keras3.posit.co/dev/reference/metric_mean_wrapper.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Wrap a stateless metric function with the Mean metric. — metric_mean_wrapper","text":"... Keyword arguments pass fn. fn metric function wrap, signature fn(y_true, y_pred). name (Optional) string name metric instance. dtype (Optional) data type metric result.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_mean_wrapper.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Wrap a stateless metric function with the Mean metric. — metric_mean_wrapper","text":"Metric instance returned. Metric instance can passed directly compile(metrics = ), used standalone object. See ?Metric example usage.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/metric_one_hot_iou.html","id":null,"dir":"Reference","previous_headings":"","what":"Computes the Intersection-Over-Union metric for one-hot encoded labels. — metric_one_hot_iou","title":"Computes the Intersection-Over-Union metric for one-hot encoded labels. — metric_one_hot_iou","text":"Formula:   Intersection--Union common evaluation metric semantic image segmentation. compute IoUs, predictions accumulated confusion matrix, weighted sample_weight metric calculated . sample_weight NULL, weights default 1. Use sample_weight 0 mask values. class can used compute IoU multi-class classification tasks labels one-hot encoded (last axis one dimension per class). Note predictions also shape. compute IoU, first labels predictions converted back integer format taking argmax class axis. computation steps base IoU class apply. Note, one channel labels predictions, class class IoU. case, use IoU instead. Also, make sure num_classes equal number classes data, avoid \"labels bound\" error confusion matrix computed.","code":"iou <- true_positives / (true_positives + false_positives + false_negatives)"},{"path":"https://keras3.posit.co/dev/reference/metric_one_hot_iou.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes the Intersection-Over-Union metric for one-hot encoded labels. — metric_one_hot_iou","text":"","code":"metric_one_hot_iou(   ...,   num_classes,   target_class_ids,   name = NULL,   dtype = NULL,   ignore_class = NULL,   sparse_y_pred = FALSE,   axis = -1L )"},{"path":"https://keras3.posit.co/dev/reference/metric_one_hot_iou.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Computes the Intersection-Over-Union metric for one-hot encoded labels. — metric_one_hot_iou","text":"... forward/backward compatability. num_classes possible number labels prediction task can . target_class_ids list list target class ids metric returned. compute IoU specific class, list (list) single id value provided. name (Optional) string name metric instance. dtype (Optional) data type metric result. ignore_class Optional integer. ID class ignored metric computation. useful, example, segmentation problems featuring \"void\" class (commonly -1 255) segmentation maps. default (ignore_class=NULL), classes considered. sparse_y_pred Whether predictions encoded using integers dense floating point vectors. FALSE, argmax function used determine sample's likely associated label. axis (Optional) dimension containing logits. Defaults -1.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_one_hot_iou.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Computes the Intersection-Over-Union metric for one-hot encoded labels. — metric_one_hot_iou","text":"Metric instance returned. Metric instance can passed directly compile(metrics = ), used standalone object. See ?Metric example usage.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_one_hot_iou.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Computes the Intersection-Over-Union metric for one-hot encoded labels. — metric_one_hot_iou","text":"Standalone usage:         Usage compile() API:","code":"y_true <- rbind(c(0, 0, 1),                 c(1, 0, 0),                 c(0, 1, 0),                 c(1, 0, 0)) y_pred <- rbind(c(0.2, 0.3, 0.5),                 c(0.1, 0.2, 0.7),                 c(0.5, 0.3, 0.1),                 c(0.1, 0.4, 0.5)) sample_weight <- c(0.1, 0.2, 0.3, 0.4)  m <- metric_one_hot_iou(num_classes = 3, target_class_ids = c(0, 2)) m$update_state(y_true = y_true,                y_pred = y_pred,                sample_weight = 10*sample_weight) ## tf.Tensor( ## [[0 0 6] ##  [3 0 0] ##  [0 0 1]], shape=(3, 3), dtype=int64) m$result() ## tf.Tensor(0.071428575, shape=(), dtype=float32) model %>% compile(   optimizer = 'sgd',   loss = 'mse',   metrics = list(metric_one_hot_iou(     num_classes = 3L,     target_class_id = list(1L)   )) )"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/metric_one_hot_mean_iou.html","id":null,"dir":"Reference","previous_headings":"","what":"Computes mean Intersection-Over-Union metric for one-hot encoded labels. — metric_one_hot_mean_iou","title":"Computes mean Intersection-Over-Union metric for one-hot encoded labels. — metric_one_hot_mean_iou","text":"Formula:   Intersection--Union common evaluation metric semantic image segmentation. compute IoUs, predictions accumulated confusion matrix, weighted sample_weight metric calculated . sample_weight NULL, weights default 1. Use sample_weight 0 mask values. class can used compute mean IoU multi-class classification tasks labels one-hot encoded (last axis one dimension per class). Note predictions also shape. compute mean IoU, first labels predictions converted back integer format taking argmax class axis. computation steps base MeanIoU class apply. Note, one channel labels predictions, class class metric_mean_iou. case, use metric_mean_iou instead. Also, make sure num_classes equal number classes data, avoid \"labels bound\" error confusion matrix computed.","code":"iou <- true_positives / (true_positives + false_positives + false_negatives)"},{"path":"https://keras3.posit.co/dev/reference/metric_one_hot_mean_iou.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes mean Intersection-Over-Union metric for one-hot encoded labels. — metric_one_hot_mean_iou","text":"","code":"metric_one_hot_mean_iou(   ...,   num_classes,   name = NULL,   dtype = NULL,   ignore_class = NULL,   sparse_y_pred = FALSE,   axis = -1L )"},{"path":"https://keras3.posit.co/dev/reference/metric_one_hot_mean_iou.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Computes mean Intersection-Over-Union metric for one-hot encoded labels. — metric_one_hot_mean_iou","text":"... forward/backward compatability. num_classes possible number labels prediction task can . name (Optional) string name metric instance. dtype (Optional) data type metric result. ignore_class Optional integer. ID class ignored metric computation. useful, example, segmentation problems featuring \"void\" class (commonly -1 255) segmentation maps. default (ignore_class=NULL), classes considered. sparse_y_pred Whether predictions encoded using natural numbers probability distribution vectors. FALSE, argmax function used determine sample's likely associated label. axis (Optional) dimension containing logits. Defaults -1.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_one_hot_mean_iou.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Computes mean Intersection-Over-Union metric for one-hot encoded labels. — metric_one_hot_mean_iou","text":"Metric instance returned. Metric instance can passed directly compile(metrics = ), used standalone object. See ?Metric example usage.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_one_hot_mean_iou.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Computes mean Intersection-Over-Union metric for one-hot encoded labels. — metric_one_hot_mean_iou","text":"Standalone usage:         Usage compile() API:","code":"y_true <- rbind(c(0, 0, 1), c(1, 0, 0), c(0, 1, 0), c(1, 0, 0)) y_pred <- rbind(c(0.2, 0.3, 0.5), c(0.1, 0.2, 0.7), c(0.5, 0.3, 0.1),                 c(0.1, 0.4, 0.5)) sample_weight <- 10 * c(0.1, 0.2, 0.3, 0.4) m <- metric_one_hot_mean_iou(num_classes = 3L) m$update_state(     y_true = y_true, y_pred = y_pred, sample_weight = sample_weight) ## tf.Tensor( ## [[0 0 6] ##  [3 0 0] ##  [0 0 1]], shape=(3, 3), dtype=int64) m$result() ## tf.Tensor(0.04761905, shape=(), dtype=float32) model %>% compile(     optimizer = 'sgd',     loss = 'mse',     metrics = list(metric_one_hot_mean_iou(num_classes = 3L)))"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/metric_pearson_correlation.html","id":null,"dir":"Reference","previous_headings":"","what":"Calculates the Pearson Correlation Coefficient (PCC). — metric_pearson_correlation","title":"Calculates the Pearson Correlation Coefficient (PCC). — metric_pearson_correlation","text":"Formula:   PCC measures linear relationship true values (y_true) predicted values (y_pred). coefficient ranges -1 1, value 1 implies perfect positive linear correlation, 0 indicates linear correlation, -1 indicates perfect negative linear correlation. metric widely used regression tasks strength linear relationship predictions true labels important evaluation criterion.","code":"loss = mean(l2norm(y_true - mean(y_true) * l2norm(y_pred - mean(y_pred)))"},{"path":"https://keras3.posit.co/dev/reference/metric_pearson_correlation.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Calculates the Pearson Correlation Coefficient (PCC). — metric_pearson_correlation","text":"","code":"metric_pearson_correlation(   y_true,   y_pred,   axis = -1L,   ...,   name = \"pearson_correlation\",   dtype = NULL )"},{"path":"https://keras3.posit.co/dev/reference/metric_pearson_correlation.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Calculates the Pearson Correlation Coefficient (PCC). — metric_pearson_correlation","text":"y_true Tensor true targets. y_pred Tensor predicted targets. axis (Optional) integer tuple integers axis/axes along compute metric. Defaults -1. ... forward/backward compatability. name (Optional) string name metric instance. dtype (Optional) data type metric result.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_pearson_correlation.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Calculates the Pearson Correlation Coefficient (PCC). — metric_pearson_correlation","text":"Usage compile() API:","code":"pcc <- metric_pearson_correlation(axis = -1) y_true <- rbind(c(0, 1, 0.5),                 c(1, 1, 0.2)) y_pred <- rbind(c(0.1, 0.9, 0.5),                 c(1, 0.9, 0.2)) pcc$update_state(y_true, y_pred) pcc$result() ## tf.Tensor(0.99669963, shape=(), dtype=float32) # equivalent operation using R's stats::cor() mean(sapply(1:nrow(y_true), function(i) {   cor(y_true[i, ], y_pred[i, ]) })) ## [1] 0.9966996 model |> compile(   optimizer = 'sgd',   loss = 'mean_squared_error',   metrics = c(keras.metrics.PearsonCorrelation()) )"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/metric_poisson.html","id":null,"dir":"Reference","previous_headings":"","what":"Computes the Poisson metric between y_true and y_pred. — metric_poisson","title":"Computes the Poisson metric between y_true and y_pred. — metric_poisson","text":"Formula:","code":"metric <- y_pred - y_true * log(y_pred)"},{"path":"https://keras3.posit.co/dev/reference/metric_poisson.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes the Poisson metric between y_true and y_pred. — metric_poisson","text":"","code":"metric_poisson(y_true, y_pred, ..., name = \"poisson\", dtype = NULL)"},{"path":"https://keras3.posit.co/dev/reference/metric_poisson.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Computes the Poisson metric between y_true and y_pred. — metric_poisson","text":"y_true Ground truth values. shape = [batch_size, d0, .. dN]. y_pred predicted values. shape = [batch_size, d0, .. dN]. ... forward/backward compatability. name (Optional) string name metric instance. dtype (Optional) data type metric result.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_poisson.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Computes the Poisson metric between y_true and y_pred. — metric_poisson","text":"y_true y_pred missing, Metric instance returned. Metric instance can passed directly compile(metrics = ), used standalone object. See ?Metric example usage. y_true y_pred provided, tensor computed value returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_poisson.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Computes the Poisson metric between y_true and y_pred. — metric_poisson","text":"Standalone usage:         Usage compile() API:","code":"m <- metric_poisson() m$update_state(rbind(c(0, 1), c(0, 0)), rbind(c(1, 1), c(0, 0))) m$result() ## tf.Tensor(0.49999997, shape=(), dtype=float32) m$reset_state() m$update_state(rbind(c(0, 1), c(0, 0)), rbind(c(1, 1), c(0, 0)),                sample_weight = c(1, 0)) m$result() ## tf.Tensor(0.99999994, shape=(), dtype=float32) model %>% compile(   optimizer = 'sgd',   loss = 'mse',   metrics = list(metric_poisson()) )"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/metric_precision.html","id":null,"dir":"Reference","previous_headings":"","what":"Computes the precision of the predictions with respect to the labels. — metric_precision","title":"Computes the precision of the predictions with respect to the labels. — metric_precision","text":"metric creates two local variables, true_positives false_positives used compute precision. value ultimately returned precision, idempotent operation simply divides true_positives sum true_positives false_positives. sample_weight NULL, weights default 1. Use sample_weight 0 mask values. top_k set, calculate precision often average class among top-k classes highest predicted values batch entry correct can found label entry. class_id specified, calculate precision considering entries batch class_id threshold /top-k highest predictions, computing fraction class_id indeed correct label.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_precision.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes the precision of the predictions with respect to the labels. — metric_precision","text":"","code":"metric_precision(   ...,   thresholds = NULL,   top_k = NULL,   class_id = NULL,   name = NULL,   dtype = NULL )"},{"path":"https://keras3.posit.co/dev/reference/metric_precision.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Computes the precision of the predictions with respect to the labels. — metric_precision","text":"... forward/backward compatability. thresholds (Optional) float value, Python list float threshold values [0, 1]. threshold compared prediction values determine truth value predictions (.e., threshold TRUE, FALSE). used loss function sets from_logits=TRUE (.e. sigmoid applied predictions), thresholds set 0. One metric value generated threshold value. neither thresholds top_k set, default calculate precision thresholds=0.5. top_k (Optional) Unset default. int value specifying top-k predictions consider calculating precision. class_id (Optional) Integer class ID want binary metrics. must half-open interval [0, num_classes), num_classes last dimension predictions. name (Optional) string name metric instance. dtype (Optional) data type metric result.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_precision.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Computes the precision of the predictions with respect to the labels. — metric_precision","text":"Metric instance returned. Metric instance can passed directly compile(metrics = ), used standalone object. See ?Metric example usage.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_precision.html","id":"usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes the precision of the predictions with respect to the labels. — metric_precision","text":"Standalone usage:                 Usage compile() API:   Usage loss from_logits=TRUE:","code":"m <- metric_precision() m$update_state(c(0, 1, 1, 1),                c(1, 0, 1, 1)) m$result() |> as.double() |> signif() ## [1] 0.666667 m$reset_state() m$update_state(c(0, 1, 1, 1),                c(1, 0, 1, 1),                sample_weight = c(0, 0, 1, 0)) m$result() |> as.double() |> signif() ## [1] 1 # With top_k=2, it will calculate precision over y_true[1:2] # and y_pred[1:2] m <- metric_precision(top_k = 2) m$update_state(c(0, 0, 1, 1), c(1, 1, 1, 1)) m$result() ## tf.Tensor(0.0, shape=(), dtype=float32) # With top_k=4, it will calculate precision over y_true[1:4] # and y_pred[1:4] m <- metric_precision(top_k = 4) m$update_state(c(0, 0, 1, 1), c(1, 1, 1, 1)) m$result() ## tf.Tensor(0.5, shape=(), dtype=float32) model |> compile(   optimizer = 'sgd',   loss = 'binary_crossentropy',   metrics = list(metric_precision()) ) model |> compile(   optimizer = 'adam',   loss = loss_binary_crossentropy(from_logits = TRUE),   metrics = list(metric_precision(thresholds = 0)) )"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/metric_precision_at_recall.html","id":null,"dir":"Reference","previous_headings":"","what":"Computes best precision where recall is >= specified value. — metric_precision_at_recall","title":"Computes best precision where recall is >= specified value. — metric_precision_at_recall","text":"metric creates four local variables, true_positives, true_negatives, false_positives false_negatives used compute precision given recall. threshold given recall value computed used evaluate corresponding precision. sample_weight NULL, weights default 1. Use sample_weight 0 mask values. class_id specified, calculate precision considering entries batch class_id threshold predictions, computing fraction class_id indeed correct label.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_precision_at_recall.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes best precision where recall is >= specified value. — metric_precision_at_recall","text":"","code":"metric_precision_at_recall(   ...,   recall,   num_thresholds = 200L,   class_id = NULL,   name = NULL,   dtype = NULL )"},{"path":"https://keras3.posit.co/dev/reference/metric_precision_at_recall.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Computes best precision where recall is >= specified value. — metric_precision_at_recall","text":"... forward/backward compatability. recall scalar value range [0, 1]. num_thresholds (Optional) Defaults 200. number thresholds use matching given recall. class_id (Optional) Integer class ID want binary metrics. must half-open interval [0, num_classes), num_classes last dimension predictions. name (Optional) string name metric instance. dtype (Optional) data type metric result.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_precision_at_recall.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Computes best precision where recall is >= specified value. — metric_precision_at_recall","text":"Metric instance returned. Metric instance can passed directly compile(metrics = ), used standalone object. See ?Metric example usage.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_precision_at_recall.html","id":"usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes best precision where recall is >= specified value. — metric_precision_at_recall","text":"Standalone usage:         Usage compile() API:","code":"m <- metric_precision_at_recall(recall = 0.5) m$update_state(c(0,   0,   0,   1,   1),                c(0, 0.3, 0.8, 0.3, 0.8)) m$result() ## tf.Tensor(0.5, shape=(), dtype=float32) m$reset_state() m$update_state(c(0,   0,   0,   1,   1),                c(0, 0.3, 0.8, 0.3, 0.8),                sample_weight = c(2, 2, 2, 1, 1)) m$result() ## tf.Tensor(0.33333334, shape=(), dtype=float32) model |> compile(   optimizer = 'sgd',   loss = 'binary_crossentropy',   metrics = list(metric_precision_at_recall(recall = 0.8)) )"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/metric_r2_score.html","id":null,"dir":"Reference","previous_headings":"","what":"Computes R2 score. — metric_r2_score","title":"Computes R2 score. — metric_r2_score","text":"Formula:   also called coefficient determination. indicates close fitted regression line ground-truth data. highest score possible 1.0. indicates predictors perfectly accounts variation target. score 0.0 indicates predictors account variation target. can also negative model worse random. metric can also compute \"Adjusted R2\" score.","code":"sum_squares_residuals <- sum((y_true - y_pred) ** 2) sum_squares <- sum((y_true - mean(y_true)) ** 2) R2 <- 1 - sum_squares_residuals / sum_squares"},{"path":"https://keras3.posit.co/dev/reference/metric_r2_score.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes R2 score. — metric_r2_score","text":"","code":"metric_r2_score(   ...,   class_aggregation = \"uniform_average\",   num_regressors = 0L,   name = \"r2_score\",   dtype = NULL )"},{"path":"https://keras3.posit.co/dev/reference/metric_r2_score.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Computes R2 score. — metric_r2_score","text":"... forward/backward compatability. class_aggregation Specifies aggregate scores corresponding different output classes (target dimensions), .e. different dimensions last axis predictions. Equivalent multioutput argument Scikit-Learn. one NULL (aggregation), \"uniform_average\", \"variance_weighted_average\". num_regressors Number independent regressors used (\"Adjusted R2\" score). 0 standard R2 score. Defaults 0. name Optional. string name metric instance. dtype Optional. data type metric result.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_r2_score.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Computes R2 score. — metric_r2_score","text":"Metric instance returned. Metric instance can passed directly compile(metrics = ), used standalone object. See ?Metric example usage.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_r2_score.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Computes R2 score. — metric_r2_score","text":"","code":"y_true <- rbind(1, 4, 3) y_pred <- rbind(2, 4, 4) metric <- metric_r2_score() metric$update_state(y_true, y_pred) metric$result() ## tf.Tensor(0.57142854, shape=(), dtype=float32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/metric_recall.html","id":null,"dir":"Reference","previous_headings":"","what":"Computes the recall of the predictions with respect to the labels. — metric_recall","title":"Computes the recall of the predictions with respect to the labels. — metric_recall","text":"metric creates two local variables, true_positives false_negatives, used compute recall. value ultimately returned recall, idempotent operation simply divides true_positives sum true_positives false_negatives. sample_weight NULL, weights default 1. Use sample_weight 0 mask values. top_k set, recall computed often average class among labels batch entry top-k predictions. class_id specified, calculate recall considering entries batch class_id label, computing fraction class_id threshold /top-k predictions.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_recall.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes the recall of the predictions with respect to the labels. — metric_recall","text":"","code":"metric_recall(   ...,   thresholds = NULL,   top_k = NULL,   class_id = NULL,   name = NULL,   dtype = NULL )"},{"path":"https://keras3.posit.co/dev/reference/metric_recall.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Computes the recall of the predictions with respect to the labels. — metric_recall","text":"... forward/backward compatability. thresholds (Optional) float value, Python list float threshold values [0, 1]. threshold compared prediction values determine truth value predictions (.e., threshold TRUE, FALSE). used loss function sets from_logits=TRUE (.e. sigmoid applied predictions), thresholds set 0. One metric value generated threshold value. neither thresholds top_k set, default calculate recall thresholds=0.5. top_k (Optional) Unset default. int value specifying top-k predictions consider calculating recall. class_id (Optional) Integer class ID want binary metrics. must half-open interval [0, num_classes), num_classes last dimension predictions. name (Optional) string name metric instance. dtype (Optional) data type metric result.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_recall.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Computes the recall of the predictions with respect to the labels. — metric_recall","text":"Metric instance returned. Metric instance can passed directly compile(metrics = ), used standalone object. See ?Metric example usage.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_recall.html","id":"usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes the recall of the predictions with respect to the labels. — metric_recall","text":"Standalone usage:         Usage compile() API:   Usage loss from_logits=TRUE:","code":"m <- metric_recall() m$update_state(c(0, 1, 1, 1),                c(1, 0, 1, 1)) m$result() ## tf.Tensor(0.6666667, shape=(), dtype=float32) m$reset_state() m$update_state(c(0, 1, 1, 1),                c(1, 0, 1, 1),                sample_weight = c(0, 0, 1, 0)) m$result() ## tf.Tensor(1.0, shape=(), dtype=float32) model |> compile(   optimizer = 'sgd',   loss = 'binary_crossentropy',   metrics = list(metric_recall()) ) model |> compile(   optimizer = 'adam',   loss = loss_binary_crossentropy(from_logits = TRUE),   metrics = list(metric_recall(thresholds = 0)) )"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/metric_recall_at_precision.html","id":null,"dir":"Reference","previous_headings":"","what":"Computes best recall where precision is >= specified value. — metric_recall_at_precision","title":"Computes best recall where precision is >= specified value. — metric_recall_at_precision","text":"given score-label-distribution required precision might achievable, case 0.0 returned recall. metric creates four local variables, true_positives, true_negatives, false_positives false_negatives used compute recall given precision. threshold given precision value computed used evaluate corresponding recall. sample_weight NULL, weights default 1. Use sample_weight 0 mask values. class_id specified, calculate precision considering entries batch class_id threshold predictions, computing fraction class_id indeed correct label.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_recall_at_precision.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes best recall where precision is >= specified value. — metric_recall_at_precision","text":"","code":"metric_recall_at_precision(   ...,   precision,   num_thresholds = 200L,   class_id = NULL,   name = NULL,   dtype = NULL )"},{"path":"https://keras3.posit.co/dev/reference/metric_recall_at_precision.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Computes best recall where precision is >= specified value. — metric_recall_at_precision","text":"... forward/backward compatability. precision scalar value range [0, 1]. num_thresholds (Optional) Defaults 200. number thresholds use matching given precision. class_id (Optional) Integer class ID want binary metrics. must half-open interval [0, num_classes), num_classes last dimension predictions. name (Optional) string name metric instance. dtype (Optional) data type metric result.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_recall_at_precision.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Computes best recall where precision is >= specified value. — metric_recall_at_precision","text":"Metric instance returned. Metric instance can passed directly compile(metrics = ), used standalone object. See ?Metric example usage.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_recall_at_precision.html","id":"usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes best recall where precision is >= specified value. — metric_recall_at_precision","text":"Standalone usage:         Usage compile() API:","code":"m <- metric_recall_at_precision(precision = 0.8) m$update_state(c(0,   0,   1,   1),                c(0, 0.5, 0.3, 0.9)) m$result() ## tf.Tensor(0.5, shape=(), dtype=float32) m$reset_state() m$update_state(c(0,   0,   1,   1),                c(0, 0.5, 0.3, 0.9),                sample_weight = c(1, 0, 0, 1)) m$result() ## tf.Tensor(1.0, shape=(), dtype=float32) model |> compile(   optimizer = 'sgd',   loss = 'binary_crossentropy',   metrics = list(metric_recall_at_precision(precision = 0.8)) )"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/metric_root_mean_squared_error.html","id":null,"dir":"Reference","previous_headings":"","what":"Computes root mean squared error metric between y_true and y_pred. — metric_root_mean_squared_error","title":"Computes root mean squared error metric between y_true and y_pred. — metric_root_mean_squared_error","text":"Formula:","code":"loss <- sqrt(mean((y_pred - y_true) ^ 2))"},{"path":"https://keras3.posit.co/dev/reference/metric_root_mean_squared_error.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes root mean squared error metric between y_true and y_pred. — metric_root_mean_squared_error","text":"","code":"metric_root_mean_squared_error(   ...,   name = \"root_mean_squared_error\",   dtype = NULL )"},{"path":"https://keras3.posit.co/dev/reference/metric_root_mean_squared_error.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Computes root mean squared error metric between y_true and y_pred. — metric_root_mean_squared_error","text":"... forward/backward compatability. name (Optional) string name metric instance. dtype (Optional) data type metric result.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_root_mean_squared_error.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Computes root mean squared error metric between y_true and y_pred. — metric_root_mean_squared_error","text":"Metric instance returned. Metric instance can passed directly compile(metrics = ), used standalone object. See ?Metric example usage.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_root_mean_squared_error.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Computes root mean squared error metric between y_true and y_pred. — metric_root_mean_squared_error","text":"Standalone usage:         Usage compile() API:","code":"m <- metric_root_mean_squared_error() m$update_state(rbind(c(0, 1), c(0, 0)), rbind(c(1, 1), c(0, 0))) m$result() ## tf.Tensor(0.5, shape=(), dtype=float32) m$reset_state() m$update_state(rbind(c(0, 1), c(0, 0)), rbind(c(1, 1), c(0, 0)),                sample_weight = c(1, 0)) m$result() ## tf.Tensor(0.70710677, shape=(), dtype=float32) model %>% compile(   optimizer = 'sgd',   loss = 'mse',   metrics = list(metric_root_mean_squared_error()))"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/metric_sensitivity_at_specificity.html","id":null,"dir":"Reference","previous_headings":"","what":"Computes best sensitivity where specificity is >= specified value. — metric_sensitivity_at_specificity","title":"Computes best sensitivity where specificity is >= specified value. — metric_sensitivity_at_specificity","text":"Sensitivity measures proportion actual positives correctly identified (tp / (tp + fn)). Specificity measures proportion actual negatives correctly identified (tn / (tn + fp)). metric creates four local variables, true_positives, true_negatives, false_positives false_negatives used compute sensitivity given specificity. threshold given specificity value computed used evaluate corresponding sensitivity. sample_weight NULL, weights default 1. Use sample_weight 0 mask values. class_id specified, calculate precision considering entries batch class_id threshold predictions, computing fraction class_id indeed correct label. additional information specificity sensitivity, see following.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_sensitivity_at_specificity.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes best sensitivity where specificity is >= specified value. — metric_sensitivity_at_specificity","text":"","code":"metric_sensitivity_at_specificity(   ...,   specificity,   num_thresholds = 200L,   class_id = NULL,   name = NULL,   dtype = NULL )"},{"path":"https://keras3.posit.co/dev/reference/metric_sensitivity_at_specificity.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Computes best sensitivity where specificity is >= specified value. — metric_sensitivity_at_specificity","text":"... forward/backward compatability. specificity scalar value range [0, 1]. num_thresholds (Optional) Defaults 200. number thresholds use matching given specificity. class_id (Optional) Integer class ID want binary metrics. must half-open interval [0, num_classes), num_classes last dimension predictions. name (Optional) string name metric instance. dtype (Optional) data type metric result.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_sensitivity_at_specificity.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Computes best sensitivity where specificity is >= specified value. — metric_sensitivity_at_specificity","text":"Metric instance returned. Metric instance can passed directly compile(metrics = ), used standalone object. See ?Metric example usage.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_sensitivity_at_specificity.html","id":"usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes best sensitivity where specificity is >= specified value. — metric_sensitivity_at_specificity","text":"Standalone usage:         Usage compile() API:","code":"m <- metric_sensitivity_at_specificity(specificity = 0.5) m$update_state(c(0,   0,   0,   1,   1),                c(0, 0.3, 0.8, 0.3, 0.8)) m$result() ## tf.Tensor(0.5, shape=(), dtype=float32) m$reset_state() m$update_state(c(0,   0,   0,   1,   1),                c(0, 0.3, 0.8, 0.3, 0.8),                sample_weight = c(1, 1, 2, 2, 1)) m$result() ## tf.Tensor(0.33333334, shape=(), dtype=float32) model |> compile(   optimizer = 'sgd',   loss = 'binary_crossentropy',   metrics = list(metric_sensitivity_at_specificity()) )"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/metric_sparse_categorical_accuracy.html","id":null,"dir":"Reference","previous_headings":"","what":"Calculates how often predictions match integer labels. — metric_sparse_categorical_accuracy","title":"Calculates how often predictions match integer labels. — metric_sparse_categorical_accuracy","text":"can provide logits classes y_pred, since argmax logits probabilities . metric creates two local variables, total count used compute frequency y_pred matches y_true. frequency ultimately returned sparse categorical accuracy: idempotent operation simply divides total count. sample_weight NULL, weights default 1. Use sample_weight 0 mask values.","code":"acc <- sample_weight %*% (y_true == which.max(y_pred))"},{"path":"https://keras3.posit.co/dev/reference/metric_sparse_categorical_accuracy.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Calculates how often predictions match integer labels. — metric_sparse_categorical_accuracy","text":"","code":"metric_sparse_categorical_accuracy(   y_true,   y_pred,   ...,   name = \"sparse_categorical_accuracy\",   dtype = NULL )"},{"path":"https://keras3.posit.co/dev/reference/metric_sparse_categorical_accuracy.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Calculates how often predictions match integer labels. — metric_sparse_categorical_accuracy","text":"y_true Tensor true targets. y_pred Tensor predicted targets. ... forward/backward compatability. name (Optional) string name metric instance. dtype (Optional) data type metric result.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_sparse_categorical_accuracy.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Calculates how often predictions match integer labels. — metric_sparse_categorical_accuracy","text":"y_true y_pred missing, Metric instance returned. Metric instance can passed directly compile(metrics = ), used standalone object. See ?Metric example usage. y_true y_pred provided, tensor computed value returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_sparse_categorical_accuracy.html","id":"usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Calculates how often predictions match integer labels. — metric_sparse_categorical_accuracy","text":"Standalone usage:         Usage compile() API:","code":"m <- metric_sparse_categorical_accuracy() m$update_state(rbind(2L, 1L), rbind(c(0.1, 0.6, 0.3), c(0.05, 0.95, 0))) m$result() ## tf.Tensor(0.5, shape=(), dtype=float32) m$reset_state() m$update_state(rbind(2L, 1L), rbind(c(0.1, 0.6, 0.3), c(0.05, 0.95, 0)),                sample_weight = c(0.7, 0.3)) m$result() ## tf.Tensor(0.3, shape=(), dtype=float32) model |> compile(optimizer = 'sgd',                  loss = 'sparse_categorical_crossentropy',                  metrics = list(metric_sparse_categorical_accuracy()))"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/metric_sparse_categorical_crossentropy.html","id":null,"dir":"Reference","previous_headings":"","what":"Computes the crossentropy metric between the labels and predictions. — metric_sparse_categorical_crossentropy","title":"Computes the crossentropy metric between the labels and predictions. — metric_sparse_categorical_crossentropy","text":"Use crossentropy metric two label classes. expects labels provided integers. want provide labels one-hot encoded, please use metric_categorical_crossentropy() metric instead. num_classes floating point values per feature y_pred single floating point value per feature y_true.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_sparse_categorical_crossentropy.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes the crossentropy metric between the labels and predictions. — metric_sparse_categorical_crossentropy","text":"","code":"metric_sparse_categorical_crossentropy(   y_true,   y_pred,   from_logits = FALSE,   ignore_class = NULL,   axis = -1L,   ...,   name = \"sparse_categorical_crossentropy\",   dtype = NULL )"},{"path":"https://keras3.posit.co/dev/reference/metric_sparse_categorical_crossentropy.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Computes the crossentropy metric between the labels and predictions. — metric_sparse_categorical_crossentropy","text":"y_true Ground truth values. y_pred predicted values. from_logits (Optional) Whether output expected logits tensor. default, consider output encodes probability distribution. ignore_class Optional integer. ID class ignored loss computation. useful, example, segmentation problems featuring \"void\" class (commonly -1 255) segmentation maps. default (ignore_class=NULL), classes considered. axis (Optional) Defaults -1. dimension along entropy computed. ... forward/backward compatability. name (Optional) string name metric instance. dtype (Optional) data type metric result.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_sparse_categorical_crossentropy.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Computes the crossentropy metric between the labels and predictions. — metric_sparse_categorical_crossentropy","text":"y_true y_pred missing, Metric instance returned. Metric instance can passed directly compile(metrics = ), used standalone object. See ?Metric example usage. y_true y_pred provided, tensor computed value returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_sparse_categorical_crossentropy.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Computes the crossentropy metric between the labels and predictions. — metric_sparse_categorical_crossentropy","text":"Standalone usage:         Usage compile() API:","code":"m <- metric_sparse_categorical_crossentropy() m$update_state(array(c(1, 2)),                rbind(c(0.05, 0.95, 0), c(0.1, 0.8, 0.1))) m$result() ## tf.Tensor(1.1769392, shape=(), dtype=float32) m$reset_state() m$update_state(array(c(1, 2)),                rbind(c(0.05, 0.95, 0), c(0.1, 0.8, 0.1)),                sample_weight = c(0.3, 0.7)) m$result() ## tf.Tensor(1.6271976, shape=(), dtype=float32) model |> compile(   optimizer = 'sgd',   loss = 'mse',   metrics = list(metric_sparse_categorical_crossentropy()) )"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/metric_sparse_top_k_categorical_accuracy.html","id":null,"dir":"Reference","previous_headings":"","what":"Computes how often integer targets are in the top K predictions. — metric_sparse_top_k_categorical_accuracy","title":"Computes how often integer targets are in the top K predictions. — metric_sparse_top_k_categorical_accuracy","text":"Computes often integer targets top K predictions. default, arguments expected update_state() : y_true: tensor shape (batch_size) representing indices true categories. y_pred: tensor shape (batch_size, num_categories) containing scores sample possible categories. from_sorted_ids=TRUE, arguments expected update_state : y_true: tensor shape (batch_size) representing indices IDs true categories. y_pred: tensor shape (batch_size, N) containing indices IDs top N categories sorted order highest score lowest score. N must greater equal k. from_sorted_ids=TRUE option can efficient set categories large model optimized way retrieve top ones either without scoring without maintaining scores possible categories.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_sparse_top_k_categorical_accuracy.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes how often integer targets are in the top K predictions. — metric_sparse_top_k_categorical_accuracy","text":"","code":"metric_sparse_top_k_categorical_accuracy(   y_true,   y_pred,   k = 5L,   ...,   name = \"sparse_top_k_categorical_accuracy\",   dtype = NULL,   from_sorted_ids = FALSE )"},{"path":"https://keras3.posit.co/dev/reference/metric_sparse_top_k_categorical_accuracy.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Computes how often integer targets are in the top K predictions. — metric_sparse_top_k_categorical_accuracy","text":"y_true Tensor true targets. y_pred Tensor predicted targets. k (Optional) Number top elements look computing accuracy. Defaults 5. ... forward/backward compatability. name (Optional) string name metric instance. dtype (Optional) data type metric result. from_sorted_ids (Optional) FALSE, default, tensor passed y_pred contains unsorted scores possible categories. TRUE, y_pred contains indices IDs top categories.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_sparse_top_k_categorical_accuracy.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Computes how often integer targets are in the top K predictions. — metric_sparse_top_k_categorical_accuracy","text":"y_true y_pred missing, Metric instance returned. Metric instance can passed directly compile(metrics = ), used standalone object. See ?Metric example usage. y_true y_pred provided, tensor computed value returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_sparse_top_k_categorical_accuracy.html","id":"usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes how often integer targets are in the top K predictions. — metric_sparse_top_k_categorical_accuracy","text":"Standalone usage:             Usage compile() API:","code":"m <- metric_sparse_top_k_categorical_accuracy(k = 1L) m$update_state(   rbind(2, 1),   op_array(rbind(c(0.1, 0.9, 0.8), c(0.05, 0.95, 0)), dtype = \"float32\") ) m$result() ## tf.Tensor(0.5, shape=(), dtype=float32) m$reset_state() m$update_state(   rbind(2, 1),   op_array(rbind(c(0.1, 0.9, 0.8), c(0.05, 0.95, 0)), dtype = \"float32\"),   sample_weight = c(0.7, 0.3) ) m$result() ## tf.Tensor(0.3, shape=(), dtype=float32) m <- metric_sparse_top_k_categorical_accuracy(k = 1, from_sorted_ids = TRUE) m$update_state(array(c(2, 1)), rbind(c(1, 0, 3),                                      c(1, 2, 3))) m$result() ## tf.Tensor(0.5, shape=(), dtype=float32) model %>% compile(optimizer = 'sgd',                   loss = 'sparse_categorical_crossentropy',                   metrics = list(metric_sparse_top_k_categorical_accuracy()))"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/metric_specificity_at_sensitivity.html","id":null,"dir":"Reference","previous_headings":"","what":"Computes best specificity where sensitivity is >= specified value. — metric_specificity_at_sensitivity","title":"Computes best specificity where sensitivity is >= specified value. — metric_specificity_at_sensitivity","text":"Sensitivity measures proportion actual positives correctly identified (tp / (tp + fn)). Specificity measures proportion actual negatives correctly identified (tn / (tn + fp)). metric creates four local variables, true_positives, true_negatives, false_positives false_negatives used compute specificity given sensitivity. threshold given sensitivity value computed used evaluate corresponding specificity. sample_weight NULL, weights default 1. Use sample_weight 0 mask values. class_id specified, calculate precision considering entries batch class_id threshold predictions, computing fraction class_id indeed correct label. additional information specificity sensitivity, see following.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_specificity_at_sensitivity.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes best specificity where sensitivity is >= specified value. — metric_specificity_at_sensitivity","text":"","code":"metric_specificity_at_sensitivity(   ...,   sensitivity,   num_thresholds = 200L,   class_id = NULL,   name = NULL,   dtype = NULL )"},{"path":"https://keras3.posit.co/dev/reference/metric_specificity_at_sensitivity.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Computes best specificity where sensitivity is >= specified value. — metric_specificity_at_sensitivity","text":"... forward/backward compatability. sensitivity scalar value range [0, 1]. num_thresholds (Optional) Defaults 200. number thresholds use matching given sensitivity. class_id (Optional) Integer class ID want binary metrics. must half-open interval [0, num_classes), num_classes last dimension predictions. name (Optional) string name metric instance. dtype (Optional) data type metric result.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_specificity_at_sensitivity.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Computes best specificity where sensitivity is >= specified value. — metric_specificity_at_sensitivity","text":"Metric instance returned. Metric instance can passed directly compile(metrics = ), used standalone object. See ?Metric example usage.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_specificity_at_sensitivity.html","id":"usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes best specificity where sensitivity is >= specified value. — metric_specificity_at_sensitivity","text":"Standalone usage:         Usage compile() API:","code":"m <- metric_specificity_at_sensitivity(sensitivity = 0.5) m$update_state(c(0,   0,   0,   1,   1),                c(0, 0.3, 0.8, 0.3, 0.8)) m$result() ## tf.Tensor(0.6666667, shape=(), dtype=float32) m$reset_state() m$update_state(c(0,   0,   0,   1,   1),                c(0, 0.3, 0.8, 0.3, 0.8),                sample_weight = c(1, 1, 2, 2, 2)) m$result() ## tf.Tensor(0.5, shape=(), dtype=float32) model |> compile(   optimizer = 'sgd',   loss = 'binary_crossentropy',   metrics = list(metric_sensitivity_at_specificity()) )"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/metric_squared_hinge.html","id":null,"dir":"Reference","previous_headings":"","what":"Computes the hinge metric between y_true and y_pred. — metric_squared_hinge","title":"Computes the hinge metric between y_true and y_pred. — metric_squared_hinge","text":"Formula:   y_true values expected -1 1. binary (0 1) labels provided convert -1 1.","code":"loss <- mean(square(maximum(1 - y_true * y_pred, 0)))"},{"path":"https://keras3.posit.co/dev/reference/metric_squared_hinge.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes the hinge metric between y_true and y_pred. — metric_squared_hinge","text":"","code":"metric_squared_hinge(y_true, y_pred, ..., name = \"squared_hinge\", dtype = NULL)"},{"path":"https://keras3.posit.co/dev/reference/metric_squared_hinge.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Computes the hinge metric between y_true and y_pred. — metric_squared_hinge","text":"y_true ground truth values. y_true values expected -1 1. binary (0 1) labels provided convert -1 1 shape = [batch_size, d0, .. dN]. y_pred predicted values shape = [batch_size, d0, .. dN]. ... forward/backward compatability. name (Optional) string name metric instance. dtype (Optional) data type metric result.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_squared_hinge.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Computes the hinge metric between y_true and y_pred. — metric_squared_hinge","text":"y_true y_pred missing, Metric instance returned. Metric instance can passed directly compile(metrics = ), used standalone object. See ?Metric example usage. y_true y_pred provided, tensor computed value returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_squared_hinge.html","id":"usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes the hinge metric between y_true and y_pred. — metric_squared_hinge","text":"Standalone usage:","code":"m <- metric_squared_hinge() m$update_state(rbind(c(0, 1), c(0, 0)), rbind(c(0.6, 0.4), c(0.4, 0.6))) m$result() ## tf.Tensor(1.86, shape=(), dtype=float32) m$reset_state() m$update_state(rbind(c(0, 1), c(0, 0)), rbind(c(0.6, 0.4), c(0.4, 0.6)),                sample_weight = c(1, 0)) m$result() ## tf.Tensor(1.46, shape=(), dtype=float32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/metric_sum.html","id":null,"dir":"Reference","previous_headings":"","what":"Compute the (weighted) sum of the given values. — metric_sum","title":"Compute the (weighted) sum of the given values. — metric_sum","text":"example, values [1, 3, 5, 7] sum 16. sample_weight specified [1, 1, 0, 0] sum 4. metric creates one variable, total. ultimately returned sum value.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_sum.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Compute the (weighted) sum of the given values. — metric_sum","text":"","code":"metric_sum(..., name = \"sum\", dtype = NULL)"},{"path":"https://keras3.posit.co/dev/reference/metric_sum.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Compute the (weighted) sum of the given values. — metric_sum","text":"... forward/backward compatability. name (Optional) string name metric instance. dtype (Optional) data type metric result.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_sum.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Compute the (weighted) sum of the given values. — metric_sum","text":"Metric instance returned. Metric instance can passed directly compile(metrics = ), used standalone object. See ?Metric example usage.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_sum.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Compute the (weighted) sum of the given values. — metric_sum","text":"","code":"m <- metric_sum() m$update_state(c(1, 3, 5, 7)) m$result() ## tf.Tensor(16.0, shape=(), dtype=float32) m <- metric_sum() m$update_state(c(1, 3, 5, 7), sample_weight = c(1, 1, 0, 0)) m$result() ## tf.Tensor(4.0, shape=(), dtype=float32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/metric_top_k_categorical_accuracy.html","id":null,"dir":"Reference","previous_headings":"","what":"Computes how often targets are in the top K predictions. — metric_top_k_categorical_accuracy","title":"Computes how often targets are in the top K predictions. — metric_top_k_categorical_accuracy","text":"Computes often targets top K predictions.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_top_k_categorical_accuracy.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes how often targets are in the top K predictions. — metric_top_k_categorical_accuracy","text":"","code":"metric_top_k_categorical_accuracy(   y_true,   y_pred,   k = 5L,   ...,   name = \"top_k_categorical_accuracy\",   dtype = NULL )"},{"path":"https://keras3.posit.co/dev/reference/metric_top_k_categorical_accuracy.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Computes how often targets are in the top K predictions. — metric_top_k_categorical_accuracy","text":"y_true Tensor true targets. y_pred Tensor predicted targets. k (Optional) Number top elements look computing accuracy. Defaults 5. ... forward/backward compatability. name (Optional) string name metric instance. dtype (Optional) data type metric result.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_top_k_categorical_accuracy.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Computes how often targets are in the top K predictions. — metric_top_k_categorical_accuracy","text":"y_true y_pred missing, Metric instance returned. Metric instance can passed directly compile(metrics = ), used standalone object. See ?Metric example usage. y_true y_pred provided, tensor computed value returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_top_k_categorical_accuracy.html","id":"usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes how often targets are in the top K predictions. — metric_top_k_categorical_accuracy","text":"Standalone usage:         Usage compile() API:","code":"m <- metric_top_k_categorical_accuracy(k = 1) m$update_state(   rbind(c(0, 0, 1), c(0, 1, 0)),   op_array(rbind(c(0.1, 0.9, 0.8), c(0.05, 0.95, 0)), dtype = \"float32\") ) m$result() ## tf.Tensor(0.5, shape=(), dtype=float32) m$reset_state() m$update_state(   rbind(c(0, 0, 1), c(0, 1, 0)),   op_array(rbind(c(0.1, 0.9, 0.8), c(0.05, 0.95, 0)), dtype = \"float32\"),   sample_weight = c(0.7, 0.3)) m$result() ## tf.Tensor(0.3, shape=(), dtype=float32) model |> compile(optimizer = 'sgd',                  loss = 'categorical_crossentropy',                  metrics = list(metric_top_k_categorical_accuracy()))"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/metric_true_negatives.html","id":null,"dir":"Reference","previous_headings":"","what":"Calculates the number of true negatives. — metric_true_negatives","title":"Calculates the number of true negatives. — metric_true_negatives","text":"sample_weight given, calculates sum weights true negatives. metric creates one local variable, accumulator used keep track number true negatives. sample_weight NULL, weights default 1. Use sample_weight 0 mask values.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_true_negatives.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Calculates the number of true negatives. — metric_true_negatives","text":"","code":"metric_true_negatives(..., thresholds = NULL, name = NULL, dtype = NULL)"},{"path":"https://keras3.posit.co/dev/reference/metric_true_negatives.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Calculates the number of true negatives. — metric_true_negatives","text":"... forward/backward compatability. thresholds (Optional) Defaults 0.5. float value, Python list float threshold values [0, 1]. threshold compared prediction values determine truth value predictions (.e., threshold TRUE, FALSE). used loss function sets from_logits=TRUE (.e. sigmoid applied predictions), thresholds set 0. One metric value generated threshold value. name (Optional) string name metric instance. dtype (Optional) data type metric result.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_true_negatives.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Calculates the number of true negatives. — metric_true_negatives","text":"Metric instance returned. Metric instance can passed directly compile(metrics = ), used standalone object. See ?Metric example usage.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_true_negatives.html","id":"usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Calculates the number of true negatives. — metric_true_negatives","text":"Standalone usage:","code":"m <- metric_true_negatives() m$update_state(c(0, 1, 0, 0), c(1, 1, 0, 0)) m$result() ## tf.Tensor(2.0, shape=(), dtype=float32) m$reset_state() m$update_state(c(0, 1, 0, 0), c(1, 1, 0, 0), sample_weight = c(0, 0, 1, 0)) m$result() ## tf.Tensor(1.0, shape=(), dtype=float32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/metric_true_positives.html","id":null,"dir":"Reference","previous_headings":"","what":"Calculates the number of true positives. — metric_true_positives","title":"Calculates the number of true positives. — metric_true_positives","text":"sample_weight given, calculates sum weights true positives. metric creates one local variable, true_positives used keep track number true positives. sample_weight NULL, weights default 1. Use sample_weight 0 mask values.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_true_positives.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Calculates the number of true positives. — metric_true_positives","text":"","code":"metric_true_positives(..., thresholds = NULL, name = NULL, dtype = NULL)"},{"path":"https://keras3.posit.co/dev/reference/metric_true_positives.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Calculates the number of true positives. — metric_true_positives","text":"... forward/backward compatability. thresholds (Optional) Defaults 0.5. float value, Python list float threshold values [0, 1]. threshold compared prediction values determine truth value predictions (.e., threshold TRUE, FALSE). used loss function sets from_logits=TRUE (.e. sigmoid applied predictions), thresholds set 0. One metric value generated threshold value. name (Optional) string name metric instance. dtype (Optional) data type metric result.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_true_positives.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Calculates the number of true positives. — metric_true_positives","text":"Metric instance returned. Metric instance can passed directly compile(metrics = ), used standalone object. See ?Metric example usage.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_true_positives.html","id":"usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Calculates the number of true positives. — metric_true_positives","text":"Standalone usage:","code":"m <- metric_true_positives() m$update_state(c(0, 1, 1, 1), c(1, 0, 1, 1)) m$result() ## tf.Tensor(2.0, shape=(), dtype=float32) m$reset_state() m$update_state(c(0, 1, 1, 1), c(1, 0, 1, 1), sample_weight = c(0, 0, 1, 0)) m$result() ## tf.Tensor(1.0, shape=(), dtype=float32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/multi-assign.html","id":null,"dir":"Reference","previous_headings":"","what":"Assign values to names — %<-%","title":"Assign values to names — %<-%","text":"See %<-% details.","code":""},{"path":"https://keras3.posit.co/dev/reference/multi-assign.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Assign values to names — %<-%","text":"","code":"x %<-% value"},{"path":"https://keras3.posit.co/dev/reference/multi-assign.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Assign values to names — %<-%","text":"right-hand-side argument, value, invisibly. called primarily side-effect assigning symbols current frame.","code":""},{"path":"https://keras3.posit.co/dev/reference/new_callback_class.html","id":null,"dir":"Reference","previous_headings":"","what":"Callback — new_callback_class","title":"Callback — new_callback_class","text":"new_callback_class() alias Callback(). See ?Callback() full documentation.","code":""},{"path":"https://keras3.posit.co/dev/reference/new_callback_class.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Callback — new_callback_class","text":"","code":"new_callback_class(   classname,   on_epoch_begin = NULL,   on_epoch_end = NULL,   on_train_begin = NULL,   on_train_end = NULL,   on_train_batch_begin = NULL,   on_train_batch_end = NULL,   on_test_begin = NULL,   on_test_end = NULL,   on_test_batch_begin = NULL,   on_test_batch_end = NULL,   on_predict_begin = NULL,   on_predict_end = NULL,   on_predict_batch_begin = NULL,   on_predict_batch_end = NULL,   ...,   public = list(),   private = list(),   inherit = NULL,   parent_env = parent.frame() )"},{"path":"https://keras3.posit.co/dev/reference/new_callback_class.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Callback — new_callback_class","text":"classname String, name custom class. (Conventionally, CamelCase). on_epoch_begin   Called start epoch. Subclasses override actions run. function called TRAIN mode. Args: epoch: Integer, index epoch. logs: Named List. Currently data passed argument method may change future. on_epoch_end   Called end epoch. Subclasses override actions run. function called TRAIN mode. Args: epoch: Integer, index epoch. logs: Named List, metric results training epoch, validation epoch validation performed. Validation result keys prefixed val_. training epoch, values Model's metrics returned. Example: list(loss = 0.2, accuracy = 0.7). on_train_begin   Called beginning training. Subclasses override actions run. Args: logs: Named list. Currently data passed argument method may change future. on_train_end   Called end training. Subclasses override actions run. Args: logs: Named list. Currently output last call on_epoch_end() passed argument method may change future. on_train_batch_begin   Called beginning training batch fit() methods. Subclasses override actions run. Note steps_per_execution argument compile Model set N, method called every N batches. Args: batch: Integer, index batch within current epoch. logs: Named list. Currently data passed argument method may change future. on_train_batch_end   Called end training batch fit() methods. Subclasses override actions run. Note steps_per_execution argument compile Model set N, method called every N batches. Args: batch: Integer, index batch within current epoch. logs: Named list. Aggregated metric results batch. on_test_begin   Called beginning evaluation validation. Subclasses override actions run. Args: logs: Named list. Currently data passed argument method may change future. on_test_end   Called end evaluation validation. Subclasses override actions run. Args: logs: Named list. Currently output last call on_test_batch_end() passed argument method may change future. on_test_batch_begin   Called beginning batch evaluate() methods. Also called beginning validation batch fit() methods, validation data provided. Subclasses override actions run. Note steps_per_execution argument compile() Model set N, method called every N batches. Args: batch: Integer, index batch within current epoch. logs: Named list. Currently data passed argument method may change future. on_test_batch_end   Called end batch evaluate() methods. Also called end validation batch fit() methods, validation data provided. Subclasses override actions run. Note steps_per_execution argument compile() Model set N, method called every N batches. Args: batch: Integer, index batch within current epoch. logs: Named list. Aggregated metric results batch. on_predict_begin   Called beginning prediction. Subclasses override actions run. Args: logs: Named list. Currently data passed argument method may change future. on_predict_end   Called end prediction. Subclasses override actions run. Args: logs: Named list. Currently data passed argument method may change future. on_predict_batch_begin   Called beginning batch predict() methods. Subclasses override actions run. Note steps_per_execution argument compile() Model set N, method called every N batches. Args: batch: Integer, index batch within current epoch. logs: Named list. Currently data passed argument method may change future. on_predict_batch_end   Called end batch predict() methods. Subclasses override actions run. Note steps_per_execution argument compile Model set N, method called every N batches. Args: batch: Integer, index batch within current epoch. logs: Named list. Aggregated metric results batch. ..., public Additional methods public members custom class. private Named list R objects (typically, functions) include instance private environments. private methods symbols scope public methods (See section \"Symbols Scope\"). instance private environment. objects private invisible Keras framework Python runtime. inherit custom class subclass. default, base keras class. parent_env R environment class methods grandparent.","code":"\\(epoch, logs = NULL) \\(epoch, logs = NULL) \\(logs = NULL) \\(logs = NULL) \\(batch, logs = NULL) \\(batch, logs=NULL) \\(logs = NULL) \\(logs = NULL) \\(batch, logs = NULL) \\(batch, logs = NULL) \\(logs = NULL) \\(logs = NULL) \\(batch, logs = NULL) \\(batch, logs = NULL)"},{"path":"https://keras3.posit.co/dev/reference/new_callback_class.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Callback — new_callback_class","text":"function returns custom Callback instances, similar builtin callback functions.","code":""},{"path":"https://keras3.posit.co/dev/reference/new_layer_class.html","id":null,"dir":"Reference","previous_headings":"","what":"Layer — new_layer_class","title":"Layer — new_layer_class","text":"new_layer_class() alias Layer(). See ?Layer() full documentation.","code":""},{"path":"https://keras3.posit.co/dev/reference/new_layer_class.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Layer — new_layer_class","text":"","code":"new_layer_class(   classname,   initialize = NULL,   call = NULL,   build = NULL,   get_config = NULL,   ...,   public = list(),   private = list(),   inherit = NULL,   parent_env = parent.frame() )"},{"path":"https://keras3.posit.co/dev/reference/new_layer_class.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Layer — new_layer_class","text":"classname String, name custom class. (Conventionally, CamelCase). initialize, call, build, get_config Recommended methods implement. See description details sections. ..., public Additional methods public members custom class. private Named list R objects (typically, functions) include instance private environments. private methods symbols scope public methods (See section \"Symbols Scope\"). instance private environment. objects private invisible Keras framework Python runtime. inherit custom class subclass. default, base keras class. parent_env R environment class methods grandparent.","code":""},{"path":"https://keras3.posit.co/dev/reference/new_layer_class.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Layer — new_layer_class","text":"composing layer constructor, similar behavior layer functions like layer_dense(). first argument returned function object, enabling initialize()ing call() layer one step composing layer pipe, like   initialize() layer instance call() , pass missing NULL value object, pass arguments initialize() name.","code":"layer_foo <- Layer(\"Foo\", ....) output <- inputs |> layer_foo() layer <- layer_dense(units = 2, activation = \"relu\") layer <- layer_dense(NULL, 2, activation = \"relu\") layer <- layer_dense(, 2, activation = \"relu\")  # then you can call() the layer in a separate step outputs <- inputs |> layer()"},{"path":"https://keras3.posit.co/dev/reference/new_learning_rate_schedule_class.html","id":null,"dir":"Reference","previous_headings":"","what":"LearningRateSchedule — new_learning_rate_schedule_class","title":"LearningRateSchedule — new_learning_rate_schedule_class","text":"new_learning_rate_schedule_class() alias LearningRateSchedule(). See ?LearningRateSchedule() full documentation.","code":""},{"path":"https://keras3.posit.co/dev/reference/new_learning_rate_schedule_class.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"LearningRateSchedule — new_learning_rate_schedule_class","text":"","code":"new_learning_rate_schedule_class(   classname,   call = NULL,   initialize = NULL,   get_config = NULL,   ...,   public = list(),   private = list(),   inherit = NULL,   parent_env = parent.frame() )"},{"path":"https://keras3.posit.co/dev/reference/new_learning_rate_schedule_class.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"LearningRateSchedule — new_learning_rate_schedule_class","text":"classname String, name custom class. (Conventionally, CamelCase). call, initialize, get_config Recommended methods implement. See description details sections. ..., public Additional methods public members custom class. private Named list R objects (typically, functions) include instance private environments. private methods symbols scope public methods (See section \"Symbols Scope\"). instance private environment. objects private invisible Keras framework Python runtime. inherit custom class subclass. default, base keras class. parent_env R environment class methods grandparent.","code":""},{"path":"https://keras3.posit.co/dev/reference/new_learning_rate_schedule_class.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"LearningRateSchedule — new_learning_rate_schedule_class","text":"function returns LearningRateSchedule instances, similar built-learning_rate_schedule_* family functions.","code":""},{"path":"https://keras3.posit.co/dev/reference/new_loss_class.html","id":null,"dir":"Reference","previous_headings":"","what":"Loss — new_loss_class","title":"Loss — new_loss_class","text":"new_loss_class() alias Loss(). See ?Loss() full documentation.","code":""},{"path":"https://keras3.posit.co/dev/reference/new_loss_class.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Loss — new_loss_class","text":"","code":"new_loss_class(   classname,   call = NULL,   ...,   public = list(),   private = list(),   inherit = NULL,   parent_env = parent.frame() )"},{"path":"https://keras3.posit.co/dev/reference/new_loss_class.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Loss — new_loss_class","text":"classname String, name custom class. (Conventionally, CamelCase). call   Method implemented subclasses: Function contains logic loss calculation using y_true, y_pred. ..., public Additional methods public members custom class. private Named list R objects (typically, functions) include instance private environments. private methods symbols scope public methods (See section \"Symbols Scope\"). instance private environment. objects private invisible Keras framework Python runtime. inherit custom class subclass. default, base keras class. parent_env R environment class methods grandparent.","code":"function(y_true, y_pred)"},{"path":"https://keras3.posit.co/dev/reference/new_loss_class.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Loss — new_loss_class","text":"function returns Loss instances, similar builtin loss functions.","code":""},{"path":"https://keras3.posit.co/dev/reference/new_metric_class.html","id":null,"dir":"Reference","previous_headings":"","what":"Metric — new_metric_class","title":"Metric — new_metric_class","text":"new_metric_class() alias Metric(). See ?Metric() full documentation.","code":""},{"path":"https://keras3.posit.co/dev/reference/new_metric_class.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Metric — new_metric_class","text":"","code":"new_metric_class(   classname,   initialize = NULL,   update_state = NULL,   result = NULL,   ...,   public = list(),   private = list(),   inherit = NULL,   parent_env = parent.frame() )"},{"path":"https://keras3.posit.co/dev/reference/new_metric_class.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Metric — new_metric_class","text":"classname String, name custom class. (Conventionally, CamelCase). initialize, update_state, result Recommended methods implement. See description section. ..., public Additional methods public members custom class. private Named list R objects (typically, functions) include instance private environments. private methods symbols scope public methods (See section \"Symbols Scope\"). instance private environment. objects private invisible Keras framework Python runtime. inherit custom class subclass. default, base keras class. parent_env R environment class methods grandparent.","code":""},{"path":"https://keras3.posit.co/dev/reference/new_metric_class.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Metric — new_metric_class","text":"function returns Metric instances, similar builtin metric functions.","code":""},{"path":"https://keras3.posit.co/dev/reference/new_model_class.html","id":null,"dir":"Reference","previous_headings":"","what":"Model — new_model_class","title":"Model — new_model_class","text":"new_model_class() alias Model(). See ?Model() full documentation.","code":""},{"path":"https://keras3.posit.co/dev/reference/new_model_class.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Model — new_model_class","text":"","code":"new_model_class(   classname,   initialize = NULL,   call = NULL,   train_step = NULL,   predict_step = NULL,   test_step = NULL,   compute_loss = NULL,   compute_metrics = NULL,   ...,   public = list(),   private = list(),   inherit = NULL,   parent_env = parent.frame() )"},{"path":"https://keras3.posit.co/dev/reference/new_model_class.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Model — new_model_class","text":"classname String, name custom class. (Conventionally, CamelCase). initialize, call, train_step, predict_step, test_step, compute_loss, compute_metrics Optional methods can overridden. ..., public Additional methods public members custom class. private Named list R objects (typically, functions) include instance private environments. private methods symbols scope public methods (See section \"Symbols Scope\"). instance private environment. objects private invisible Keras framework Python runtime. inherit custom class subclass. default, base keras class. parent_env R environment class methods grandparent.","code":""},{"path":"https://keras3.posit.co/dev/reference/new_model_class.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Model — new_model_class","text":"model constructor function, can call create instance new model type.","code":""},{"path":"https://keras3.posit.co/dev/reference/newaxis.html","id":null,"dir":"Reference","previous_headings":"","what":"New axis — newaxis","title":"New axis — newaxis","text":"alias NULL. meant used [ tensors, expand dimensions tensor","code":""},{"path":"https://keras3.posit.co/dev/reference/newaxis.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"New axis — newaxis","text":"","code":"newaxis"},{"path":"https://keras3.posit.co/dev/reference/newaxis.html","id":"format","dir":"Reference","previous_headings":"","what":"Format","title":"New axis — newaxis","text":"object class NULL length 0.","code":""},{"path":"https://keras3.posit.co/dev/reference/newaxis.html","id":"details","dir":"Reference","previous_headings":"","what":"Details","title":"New axis — newaxis","text":"","code":"x <- op_convert_to_tensor(1:10)  op_shape(x) op_shape(x[]) op_shape(x[newaxis]) op_shape(x@py[newaxis]) op_shape(x@r[newaxis])  op_shape(x[newaxis, .., newaxis]) op_shape(x@py[newaxis, .., newaxis]) op_shape(x@r[newaxis, .., newaxis])"},{"path":"https://keras3.posit.co/dev/reference/normalize.html","id":null,"dir":"Reference","previous_headings":"","what":"Normalizes an array. — normalize","title":"Normalizes an array. — normalize","text":"input R array, R array returned. backend tensor, backend tensor returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/normalize.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Normalizes an array. — normalize","text":"","code":"normalize(x, axis = -1L, order = 2L)"},{"path":"https://keras3.posit.co/dev/reference/normalize.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Normalizes an array. — normalize","text":"x Array normalize. axis axis along normalize. order Normalization order (e.g. order=2 L2 norm).","code":""},{"path":"https://keras3.posit.co/dev/reference/normalize.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Normalizes an array. — normalize","text":"normalized copy array.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_abs.html","id":null,"dir":"Reference","previous_headings":"","what":"Compute the absolute value element-wise. — op_abs","title":"Compute the absolute value element-wise. — op_abs","text":"Compute absolute value element-wise.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_abs.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Compute the absolute value element-wise. — op_abs","text":"","code":"op_abs(x)"},{"path":"https://keras3.posit.co/dev/reference/op_abs.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Compute the absolute value element-wise. — op_abs","text":"x Input tensor","code":""},{"path":"https://keras3.posit.co/dev/reference/op_abs.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Compute the absolute value element-wise. — op_abs","text":"array containing absolute value element x.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_abs.html","id":"example","dir":"Reference","previous_headings":"","what":"Example","title":"Compute the absolute value element-wise. — op_abs","text":"","code":"x <- op_convert_to_tensor(c(-1.2, 1.2)) op_abs(x) ## tf.Tensor([1.2 1.2], shape=(2), dtype=float64)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_add.html","id":null,"dir":"Reference","previous_headings":"","what":"Add arguments element-wise. — op_add","title":"Add arguments element-wise. — op_add","text":"Add arguments element-wise.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_add.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Add arguments element-wise. — op_add","text":"","code":"op_add(x1, x2)"},{"path":"https://keras3.posit.co/dev/reference/op_add.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Add arguments element-wise. — op_add","text":"x1 First input tensor. x2 Second input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_add.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Add arguments element-wise. — op_add","text":"tensor containing element-wise sum x1 x2.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_add.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Add arguments element-wise. — op_add","text":"op_add also broadcasts shapes:     Note function automatically called using R operator + tensors.","code":"x1 <- op_convert_to_tensor(c(1, 4)) x2 <- op_convert_to_tensor(c(5, 6)) op_add(x1, x2) ## tf.Tensor([ 6. 10.], shape=(2), dtype=float64) # alias for x1 + x2 x1 + x2 ## tf.Tensor([ 6. 10.], shape=(2), dtype=float64) x1 <- op_convert_to_tensor(array(c(5, 5, 4, 6), dim =c(2, 2))) x2 <- op_convert_to_tensor(c(5, 6)) op_add(x1, x2) ## tf.Tensor( ## [[10. 10.] ##  [10. 12.]], shape=(2, 2), dtype=float64) x <- op_ones(c(3)) op_add(x, x) ## tf.Tensor([2. 2. 2.], shape=(3), dtype=float32) x + x ## tf.Tensor([2. 2. 2.], shape=(3), dtype=float32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_all.html","id":null,"dir":"Reference","previous_headings":"","what":"Test whether all array elements along a given axis evaluate to TRUE. — op_all","title":"Test whether all array elements along a given axis evaluate to TRUE. — op_all","text":"Test whether array elements along given axis evaluate TRUE.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_all.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Test whether all array elements along a given axis evaluate to TRUE. — op_all","text":"","code":"op_all(x, axis = NULL, keepdims = FALSE)"},{"path":"https://keras3.posit.co/dev/reference/op_all.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Test whether all array elements along a given axis evaluate to TRUE. — op_all","text":"x Input tensor. axis integer tuple integers represent axis along logical reduction performed. default (axis = NULL) perform logical dimensions input array. axis may negative, case counts last first axis. keepdims TRUE, axes reduced left result dimensions size one. option, result broadcast correctly input array. Defaults FALSE.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_all.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Test whether all array elements along a given axis evaluate to TRUE. — op_all","text":"tensor containing logical reduction axis.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_all.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Test whether all array elements along a given axis evaluate to TRUE. — op_all","text":"keepdims = TRUE outputs tensor dimensions reduced one.","code":"x <- op_convert_to_tensor(c(TRUE, FALSE)) op_all(x) ## tf.Tensor(False, shape=(), dtype=bool) (x <- op_convert_to_tensor(array(c(TRUE, FALSE, TRUE, TRUE, TRUE, TRUE), dim = c(3, 2)))) ## tf.Tensor( ## [[ True  True] ##  [False  True] ##  [ True  True]], shape=(3, 2), dtype=bool) op_all(x, axis = 1) ## tf.Tensor([False  True], shape=(2), dtype=bool) op_all(x, keepdims = TRUE) ## tf.Tensor([[False]], shape=(1, 1), dtype=bool)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_angle.html","id":null,"dir":"Reference","previous_headings":"","what":"Element-wise angle of a complex tensor. — op_angle","title":"Element-wise angle of a complex tensor. — op_angle","text":"Returns phase angle (radians) element x.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_angle.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Element-wise angle of a complex tensor. — op_angle","text":"","code":"op_angle(x)"},{"path":"https://keras3.posit.co/dev/reference/op_angle.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Element-wise angle of a complex tensor. — op_angle","text":"x Input tensor. Can real complex.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_angle.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Element-wise angle of a complex tensor. — op_angle","text":"Tensor shape x, containing element-wise angles.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_angle.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Element-wise angle of a complex tensor. — op_angle","text":"","code":"x <- op_convert_to_tensor(matrix(c(1 + 3i, 2 - 5i, 4 - 3i, 3 + 2i), nrow = 2)) op_angle(x) ## tf.Tensor( ## [[ 1.24904577 -0.64350111] ##  [-1.19028995  0.5880026 ]], shape=(2, 2), dtype=float64)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_any.html","id":null,"dir":"Reference","previous_headings":"","what":"Test whether any array element along a given axis evaluates to TRUE. — op_any","title":"Test whether any array element along a given axis evaluates to TRUE. — op_any","text":"Test whether array element along given axis evaluates TRUE.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_any.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Test whether any array element along a given axis evaluates to TRUE. — op_any","text":"","code":"op_any(x, axis = NULL, keepdims = FALSE)"},{"path":"https://keras3.posit.co/dev/reference/op_any.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Test whether any array element along a given axis evaluates to TRUE. — op_any","text":"x Input tensor. axis integer tuple integers represent axis along logical reduction performed. default (axis = NULL) perform logical dimensions input array. axis may negative, case counts last first axis. keepdims TRUE, axes reduced left result dimensions size one. option, result broadcast correctly input array. Defaults FALSE.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_any.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Test whether any array element along a given axis evaluates to TRUE. — op_any","text":"tensor containing logical reduction axis.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_any.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Test whether any array element along a given axis evaluates to TRUE. — op_any","text":"keepdims = TRUE outputs tensor dimensions reduced one.","code":"x <- op_array(c(TRUE, FALSE)) op_any(x) ## tf.Tensor(True, shape=(), dtype=bool) (x <- op_reshape(c(FALSE, FALSE, FALSE,                    TRUE, FALSE, FALSE),                  c(2, 3))) ## tf.Tensor( ## [[False False False] ##  [ True False False]], shape=(2, 3), dtype=bool) op_any(x, axis = 1) ## tf.Tensor([ True False False], shape=(3), dtype=bool) op_any(x, axis = 2) ## tf.Tensor([False  True], shape=(2), dtype=bool) op_any(x, axis = -1) ## tf.Tensor([False  True], shape=(2), dtype=bool) op_any(x, keepdims = TRUE) ## tf.Tensor([[ True]], shape=(1, 1), dtype=bool) op_any(x, 1, keepdims = TRUE) ## tf.Tensor([[ True False False]], shape=(1, 3), dtype=bool)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_append.html","id":null,"dir":"Reference","previous_headings":"","what":"Append tensor x2 to the end of tensor x1. — op_append","title":"Append tensor x2 to the end of tensor x1. — op_append","text":"Append tensor x2 end tensor x1.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_append.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Append tensor x2 to the end of tensor x1. — op_append","text":"","code":"op_append(x1, x2, axis = NULL)"},{"path":"https://keras3.posit.co/dev/reference/op_append.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Append tensor x2 to the end of tensor x1. — op_append","text":"x1 First input tensor. x2 Second input tensor. axis Axis along tensor x2 appended tensor x1. NULL, tensors flattened use.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_append.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Append tensor x2 to the end of tensor x1. — op_append","text":"tensor values x2 appended x1.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_append.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Append tensor x2 to the end of tensor x1. — op_append","text":"axis specified, x1 x2 must compatible shapes.","code":"x1 <- op_convert_to_tensor(c(1, 2, 3)) x2 <- op_convert_to_tensor(rbind(c(4, 5, 6), c(7, 8, 9))) op_append(x1, x2) ## tf.Tensor([1. 2. 3. 4. 5. 6. 7. 8. 9.], shape=(9), dtype=float64) x1 <- op_convert_to_tensor(rbind(c(1, 2, 3), c(4, 5, 6))) x2 <- op_convert_to_tensor(rbind(c(7, 8, 9))) op_append(x1, x2, axis = 1) ## tf.Tensor( ## [[1. 2. 3.] ##  [4. 5. 6.] ##  [7. 8. 9.]], shape=(3, 3), dtype=float64) x3 <- op_convert_to_tensor(c(7, 8, 9)) try(op_append(x1, x3, axis = 1)) ## Error in py_call_impl(callable, call_args$unnamed, call_args$named) : ##   tensorflow.python.framework.errors_impl.InvalidArgumentError: {{function_node __wrapped__ConcatV2_N_2_device_/job:localhost/replica:0/task:0/device:CPU:0}} ConcatOp : Ranks of all input tensors should match: shape[0] = [2,3] vs. shape[1] = [3] [Op:ConcatV2] name: concat"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_arange.html","id":null,"dir":"Reference","previous_headings":"","what":"Return evenly spaced values within a given interval. — op_arange","title":"Return evenly spaced values within a given interval. — op_arange","text":"arange can called varying number positional arguments: arange(end): Values generated within half-open interval [1, end] (words, interval including start end). arange(start, end): Values generated within interval [start, end]. arange(start, end, step): Values generated within interval [start, end], spacing values given step.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_arange.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Return evenly spaced values within a given interval. — op_arange","text":"","code":"op_arange(start, end, step = 1L, dtype = NULL, include_end = TRUE)"},{"path":"https://keras3.posit.co/dev/reference/op_arange.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Return evenly spaced values within a given interval. — op_arange","text":"start Integer real, representing start interval. interval includes value. end Integer real, representing end interval. include_end FALSE, interval include value, except cases step integer floating point round-affects length . Defaults NULL. step Integer real, represent spacing values. output , distance two adjacent values, [+1] - []. default step size 1. step specified position argument, start must also given. dtype type output array. dtype given, infer data type input arguments. include_end TRUE FALSE. FALSE, end included output sequence.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_arange.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Return evenly spaced values within a given interval. — op_arange","text":"Tensor evenly spaced values. floating point arguments, length result ceiling((end - start)/step). floating point overflow, rule may result last element greater end.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_arange.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Return evenly spaced values within a given interval. — op_arange","text":"","code":"op_arange(3L) ## tf.Tensor([1 2 3], shape=(3), dtype=int32) op_arange(3) # float ## tf.Tensor([1. 2. 3.], shape=(3), dtype=float32) op_arange(3, dtype = 'int32') #int ## tf.Tensor([1 2 3], shape=(3), dtype=int32) op_arange(3L, 7L) ## tf.Tensor([3 4 5 6 7], shape=(5), dtype=int32) op_arange(3L, 7L, 2L) ## tf.Tensor([3 5 7], shape=(3), dtype=int32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_arccos.html","id":null,"dir":"Reference","previous_headings":"","what":"Trigonometric inverse cosine, element-wise. — op_arccos","title":"Trigonometric inverse cosine, element-wise. — op_arccos","text":"inverse cos , y = cos(x), x = arccos(y).","code":""},{"path":"https://keras3.posit.co/dev/reference/op_arccos.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Trigonometric inverse cosine, element-wise. — op_arccos","text":"","code":"op_arccos(x)"},{"path":"https://keras3.posit.co/dev/reference/op_arccos.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Trigonometric inverse cosine, element-wise. — op_arccos","text":"x Input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_arccos.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Trigonometric inverse cosine, element-wise. — op_arccos","text":"Tensor angle ray intersecting unit circle given x-coordinate radians [0, pi].","code":""},{"path":"https://keras3.posit.co/dev/reference/op_arccos.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Trigonometric inverse cosine, element-wise. — op_arccos","text":"","code":"x <- op_convert_to_tensor(c(1, -1)) op_arccos(x) ## tf.Tensor([0.         3.14159265], shape=(2), dtype=float64)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_arccosh.html","id":null,"dir":"Reference","previous_headings":"","what":"Inverse hyperbolic cosine, element-wise. — op_arccosh","title":"Inverse hyperbolic cosine, element-wise. — op_arccosh","text":"Inverse hyperbolic cosine, element-wise.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_arccosh.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Inverse hyperbolic cosine, element-wise. — op_arccosh","text":"","code":"op_arccosh(x)"},{"path":"https://keras3.posit.co/dev/reference/op_arccosh.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Inverse hyperbolic cosine, element-wise. — op_arccosh","text":"x Input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_arccosh.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Inverse hyperbolic cosine, element-wise. — op_arccosh","text":"Output tensor shape x.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_arccosh.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Inverse hyperbolic cosine, element-wise. — op_arccosh","text":"","code":"x <- op_convert_to_tensor(c(10, 100)) op_arccosh(x) ## tf.Tensor([2.99322285 5.29829237], shape=(2), dtype=float64)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_arcsin.html","id":null,"dir":"Reference","previous_headings":"","what":"Inverse sine, element-wise. — op_arcsin","title":"Inverse sine, element-wise. — op_arcsin","text":"Inverse sine, element-wise.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_arcsin.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Inverse sine, element-wise. — op_arcsin","text":"","code":"op_arcsin(x)"},{"path":"https://keras3.posit.co/dev/reference/op_arcsin.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Inverse sine, element-wise. — op_arcsin","text":"x Input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_arcsin.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Inverse sine, element-wise. — op_arcsin","text":"Tensor inverse sine element x, radians closed interval [-pi/2, pi/2].","code":""},{"path":"https://keras3.posit.co/dev/reference/op_arcsin.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Inverse sine, element-wise. — op_arcsin","text":"","code":"x <- op_convert_to_tensor(c(1, -1, 0)) op_arcsin(x) ## tf.Tensor([ 1.57079633 -1.57079633  0.        ], shape=(3), dtype=float64)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_arcsinh.html","id":null,"dir":"Reference","previous_headings":"","what":"Inverse hyperbolic sine, element-wise. — op_arcsinh","title":"Inverse hyperbolic sine, element-wise. — op_arcsinh","text":"Inverse hyperbolic sine, element-wise.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_arcsinh.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Inverse hyperbolic sine, element-wise. — op_arcsinh","text":"","code":"op_arcsinh(x)"},{"path":"https://keras3.posit.co/dev/reference/op_arcsinh.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Inverse hyperbolic sine, element-wise. — op_arcsinh","text":"x Input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_arcsinh.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Inverse hyperbolic sine, element-wise. — op_arcsinh","text":"Output tensor shape x.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_arcsinh.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Inverse hyperbolic sine, element-wise. — op_arcsinh","text":"","code":"x <- op_convert_to_tensor(c(1, -1, 0)) op_arcsinh(x) ## tf.Tensor([ 0.88137359 -0.88137359  0.        ], shape=(3), dtype=float64)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_arctan.html","id":null,"dir":"Reference","previous_headings":"","what":"Trigonometric inverse tangent, element-wise. — op_arctan","title":"Trigonometric inverse tangent, element-wise. — op_arctan","text":"Trigonometric inverse tangent, element-wise.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_arctan.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Trigonometric inverse tangent, element-wise. — op_arctan","text":"","code":"op_arctan(x)"},{"path":"https://keras3.posit.co/dev/reference/op_arctan.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Trigonometric inverse tangent, element-wise. — op_arctan","text":"x Input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_arctan.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Trigonometric inverse tangent, element-wise. — op_arctan","text":"Tensor inverse tangent element x, interval [-pi/2, pi/2].","code":""},{"path":"https://keras3.posit.co/dev/reference/op_arctan.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Trigonometric inverse tangent, element-wise. — op_arctan","text":"","code":"x <- op_convert_to_tensor(c(0, 1)) op_arctan(x) ## tf.Tensor([0.         0.78539816], shape=(2), dtype=float64)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_arctan2.html","id":null,"dir":"Reference","previous_headings":"","what":"Element-wise arc tangent of x1/x2 choosing the quadrant correctly. — op_arctan2","title":"Element-wise arc tangent of x1/x2 choosing the quadrant correctly. — op_arctan2","text":"quadrant (.e., branch) chosen arctan2(x1, x2) signed angle radians ray ending origin passing point (1, 0), ray ending origin passing point (x2, x1). (Note role reversal: \"y-coordinate\" first function parameter, \"x-coordinate\" second.) IEEE convention, function defined x2 = +/-0 either x1 x2 = +/-inf.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_arctan2.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Element-wise arc tangent of x1/x2 choosing the quadrant correctly. — op_arctan2","text":"","code":"op_arctan2(x1, x2)"},{"path":"https://keras3.posit.co/dev/reference/op_arctan2.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Element-wise arc tangent of x1/x2 choosing the quadrant correctly. — op_arctan2","text":"x1 First input tensor. x2 Second input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_arctan2.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Element-wise arc tangent of x1/x2 choosing the quadrant correctly. — op_arctan2","text":"Tensor angles radians, range [-pi, pi].","code":""},{"path":"https://keras3.posit.co/dev/reference/op_arctan2.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Element-wise arc tangent of x1/x2 choosing the quadrant correctly. — op_arctan2","text":"Consider four points different quadrants:     Note order parameters. arctan2 defined also x2 = 0 several points, obtaining values range [-pi, pi]:","code":"x <- op_array(c(-1, 1, 1, -1)) y <- op_array(c(-1, -1, 1, 1)) op_arctan2(y, x) * 180 / pi ## tf.Tensor([-135.  -45.   45.  135.], shape=(4), dtype=float32) op_arctan2(     op_array(c(1, -1)),     op_array(c(0, 0)) ) ## tf.Tensor([ 1.5707964 -1.5707964], shape=(2), dtype=float32) op_arctan2(     op_array(c(0, 0, Inf)),     op_array(c(+0, -0, Inf)) ) ## tf.Tensor([0.        3.1415927 0.7853982], shape=(3), dtype=float32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_arctanh.html","id":null,"dir":"Reference","previous_headings":"","what":"Inverse hyperbolic tangent, element-wise. — op_arctanh","title":"Inverse hyperbolic tangent, element-wise. — op_arctanh","text":"Inverse hyperbolic tangent, element-wise.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_arctanh.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Inverse hyperbolic tangent, element-wise. — op_arctanh","text":"","code":"op_arctanh(x)"},{"path":"https://keras3.posit.co/dev/reference/op_arctanh.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Inverse hyperbolic tangent, element-wise. — op_arctanh","text":"x Input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_arctanh.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Inverse hyperbolic tangent, element-wise. — op_arctanh","text":"Output tensor shape x.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_argmax.html","id":null,"dir":"Reference","previous_headings":"","what":"Returns the indices of the maximum values along an axis. — op_argmax","title":"Returns the indices of the maximum values along an axis. — op_argmax","text":"Returns indices maximum values along axis.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_argmax.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Returns the indices of the maximum values along an axis. — op_argmax","text":"","code":"op_argmax(x, axis = NULL, keepdims = FALSE, zero_indexed = FALSE)"},{"path":"https://keras3.posit.co/dev/reference/op_argmax.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Returns the indices of the maximum values along an axis. — op_argmax","text":"x Input tensor. axis default, index flattened tensor, otherwise along specified axis. keepdims set TRUE, axes reduced left result dimensions size one. Defaults FALSE. zero_indexed TRUE, returned indices zero-based (0 encodes first position); FALSE (default), returned indices one-based (1 encodes first position).","code":""},{"path":"https://keras3.posit.co/dev/reference/op_argmax.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Returns the indices of the maximum values along an axis. — op_argmax","text":"Tensor indices. shape x, dimension along axis removed. Note returned integer 1-based (.e., argmax first index position, returned value 1)","code":""},{"path":"https://keras3.posit.co/dev/reference/op_argmax.html","id":"note","dir":"Reference","previous_headings":"","what":"Note","title":"Returns the indices of the maximum values along an axis. — op_argmax","text":"similar R max.col(x) case 2-d array (matrix), nd-array, apply(x, axis, .max)","code":""},{"path":"https://keras3.posit.co/dev/reference/op_argmax.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Returns the indices of the maximum values along an axis. — op_argmax","text":"","code":"x <- op_array(rbind(c(11, 12, 10),                     c(12, 11, 10))) x ## tf.Tensor( ## [[11. 12. 10.] ##  [12. 11. 10.]], shape=(2, 3), dtype=float64) op_argmax(x) ## tf.Tensor(2, shape=(), dtype=int32) op_argmax(x, axis = 1) ## tf.Tensor([2 1 1], shape=(3), dtype=int32) op_argmax(x, axis = 2) ## tf.Tensor([2 1], shape=(2), dtype=int32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_argmin.html","id":null,"dir":"Reference","previous_headings":"","what":"Returns the indices of the minimum values along an axis. — op_argmin","title":"Returns the indices of the minimum values along an axis. — op_argmin","text":"Returns indices minimum values along axis.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_argmin.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Returns the indices of the minimum values along an axis. — op_argmin","text":"","code":"op_argmin(x, axis = NULL, keepdims = FALSE, zero_indexed = FALSE)"},{"path":"https://keras3.posit.co/dev/reference/op_argmin.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Returns the indices of the minimum values along an axis. — op_argmin","text":"x Input tensor. axis default, index flattened tensor, otherwise along specified axis. keepdims set TRUE, axes reduced left result dimensions size one. Defaults FALSE. zero_indexed TRUE, returned indices zero-based (0 encodes first position); FALSE (default), returned indices one-based (1 encodes first position).","code":""},{"path":"https://keras3.posit.co/dev/reference/op_argmin.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Returns the indices of the minimum values along an axis. — op_argmin","text":"Tensor indices. shape x, dimension along axis removed.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_argmin.html","id":"note","dir":"Reference","previous_headings":"","what":"Note","title":"Returns the indices of the minimum values along an axis. — op_argmin","text":"similar R expression apply(x, axis, .min), x R array.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_argmin.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Returns the indices of the minimum values along an axis. — op_argmin","text":"","code":"x <- op_arange(6L) |> op_reshape(c(2, 3)) |> op_add(10) x ## tf.Tensor( ## [[11. 12. 13.] ##  [14. 15. 16.]], shape=(2, 3), dtype=float32) op_argmin(x) ## tf.Tensor(1, shape=(), dtype=int32) op_argmin(x, axis = 1) ## tf.Tensor([1 1 1], shape=(3), dtype=int32) op_argmin(x, axis = 2) ## tf.Tensor([1 1], shape=(2), dtype=int32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_argpartition.html","id":null,"dir":"Reference","previous_headings":"","what":"Performs an indirect partition along the given axis. — op_argpartition","title":"Performs an indirect partition along the given axis. — op_argpartition","text":"returns array indices shape x index data along given axis partitioned order.","code":"x <- op_convert_to_tensor(c(9, 3, 6, 2, 8, 5, 7, 1, 10, 4)) x@r[op_argpartition(x, 3)] ## tf.Tensor([ 1.  2.  3. 10.  9.  8.  7.  6.  5.  4.], shape=(10), dtype=float64) x@r[op_argpartition(x, 5)] ## tf.Tensor([ 1.  2.  3.  4.  5. 10.  9.  8.  7.  6.], shape=(10), dtype=float64) x@r[op_argpartition(x, 7)] ## tf.Tensor([ 1.  2.  3.  4.  5.  6.  7. 10.  9.  8.], shape=(10), dtype=float64)"},{"path":"https://keras3.posit.co/dev/reference/op_argpartition.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Performs an indirect partition along the given axis. — op_argpartition","text":"","code":"op_argpartition(x, kth, axis = -1L, zero_indexed = FALSE)"},{"path":"https://keras3.posit.co/dev/reference/op_argpartition.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Performs an indirect partition along the given axis. — op_argpartition","text":"x Array sort. kth Element index partition . k-th element final sorted position smaller elements moved larger elements behind . order elements partitions undefined. provided sequence k-th partition sorted position . axis Axis along sort. default -1 (last axis). NULL, flattened array used. zero_indexed TRUE, returned indices zero-based (0 encodes first position); FALSE (default), returned indices one-based (1 encodes first position).","code":""},{"path":"https://keras3.posit.co/dev/reference/op_argpartition.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Performs an indirect partition along the given axis. — op_argpartition","text":"Array indices partition x along specified axis.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_argsort.html","id":null,"dir":"Reference","previous_headings":"","what":"Returns the indices that would sort a tensor. — op_argsort","title":"Returns the indices that would sort a tensor. — op_argsort","text":"Returns indices sort tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_argsort.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Returns the indices that would sort a tensor. — op_argsort","text":"","code":"op_argsort(x, axis = -1L, zero_indexed = FALSE)"},{"path":"https://keras3.posit.co/dev/reference/op_argsort.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Returns the indices that would sort a tensor. — op_argsort","text":"x Input tensor. axis Axis along sort. Defaults -1 (last axis). NULL, flattened tensor used. zero_indexed TRUE, returned indices zero-based (0 encodes first position); FALSE (default), returned indices one-based (1 encodes first position).","code":""},{"path":"https://keras3.posit.co/dev/reference/op_argsort.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Returns the indices that would sort a tensor. — op_argsort","text":"Tensor indices sort x along specified axis.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_argsort.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Returns the indices that would sort a tensor. — op_argsort","text":"One dimensional array:         Two-dimensional array:","code":"x <- op_array(c(3, 1, 2)) + .1 op_argsort(x) ## tf.Tensor([2 3 1], shape=(3), dtype=int32) x@r[op_argsort(x)] == op_sort(x) ## tf.Tensor([ True  True  True], shape=(3), dtype=bool) x <- op_array(rbind(c(0, 3),                     c(3, 2),                     c(4, 5))) + .1 (i <- op_argsort(x, axis = 1)) ## tf.Tensor( ## [[1 2] ##  [2 1] ##  [3 3]], shape=(3, 2), dtype=int32) x@r[i@r[, 1], ] # sort x-rows using first col of x ## tf.Tensor( ## [[0.1 3.1] ##  [3.1 2.1] ##  [4.1 5.1]], shape=(3, 2), dtype=float64) x@r[i@r[, 2], ] # sort x-rows using second col of x ## tf.Tensor( ## [[3.1 2.1] ##  [0.1 3.1] ##  [4.1 5.1]], shape=(3, 2), dtype=float64) (i <- op_argsort(x, axis = 2)) ## tf.Tensor( ## [[1 2] ##  [2 1] ##  [1 2]], shape=(3, 2), dtype=int32) x@r[, i@r[2,]] # sort x-cols using second row of x ## tf.Tensor( ## [[3.1 0.1] ##  [2.1 3.1] ##  [5.1 4.1]], shape=(3, 2), dtype=float64)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_array.html","id":null,"dir":"Reference","previous_headings":"","what":"Create a tensor. — op_array","title":"Create a tensor. — op_array","text":"Create tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_array.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Create a tensor. — op_array","text":"","code":"op_array(x, dtype = NULL)"},{"path":"https://keras3.posit.co/dev/reference/op_array.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Create a tensor. — op_array","text":"x Input tensor. dtype desired data-type tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_array.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Create a tensor. — op_array","text":"tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_array.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Create a tensor. — op_array","text":"","code":"op_array(c(1, 2, 3)) ## tf.Tensor([1. 2. 3.], shape=(3), dtype=float32) op_array(c(1, 2, 3), dtype = \"float32\") ## tf.Tensor([1. 2. 3.], shape=(3), dtype=float32) op_array(c(1, 2, 3), dtype = \"int32\") ## tf.Tensor([1 2 3], shape=(3), dtype=int32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_associative_scan.html","id":null,"dir":"Reference","previous_headings":"","what":"Performs a scan with an associative binary operation, in parallel. — op_associative_scan","title":"Performs a scan with an associative binary operation, in parallel. — op_associative_scan","text":"operation similar op_scan(), key difference op_associative_scan() parallel implementation potentially significant performance benefits, especially jit compiled. catch can used f binary associative operation (.e. must verify f(, f(b, c)) == f(f(, b), c)). introduction associative scans, refer paper: Blelloch, Guy E. 1990. Prefix Sums Applications.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_associative_scan.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Performs a scan with an associative binary operation, in parallel. — op_associative_scan","text":"","code":"op_associative_scan(f, elems, reverse = FALSE, axis = 1L)"},{"path":"https://keras3.posit.co/dev/reference/op_associative_scan.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Performs a scan with an associative binary operation, in parallel. — op_associative_scan","text":"f callable implementing associative binary operation signature  r = f(, b). Function f must associative, .e., must satisfy equation f(, f(b, c)) == f(f(, b), c). inputs result (possibly nested tree structures ) array(s) matching elems. array dimension place axis dimension. f applied elementwise axis dimension. result r shape (structure) two inputs b. elems (possibly nested tree structure ) array(s), axis dimension size num_elems. reverse boolean stating scan reversed respect axis dimension. axis integer identifying axis scan occur.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_associative_scan.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Performs a scan with an associative binary operation, in parallel. — op_associative_scan","text":"(possibly nested tree structure ) array(s) shape structure elems, k'th element axis result recursively applying f combine first k elements elems along axis. example, given elems = list(, b, c, ...), result list(, f(, b), f(f(, b), c), ...).","code":""},{"path":"https://keras3.posit.co/dev/reference/op_associative_scan.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Performs a scan with an associative binary operation, in parallel. — op_associative_scan","text":"","code":"sum_fn <- function(x, y) x + y xs <- op_arange(5L) op_associative_scan(sum_fn, xs) ## tf.Tensor([ 1  3  6 10 15], shape=(5), dtype=int32) sum_fn <- function(x, y) {   str(list(x = x, y = y))   map2(x, y, \\(.x, .y) .x + .y) }  xs <- list(op_array(1:2),            op_array(1:2),            op_array(1:2)) ys <- op_associative_scan(sum_fn, xs, axis = 1) ## List of 2 ##  $ x:List of 3 ##   ..$ :<tf.Tensor: shape=(1), dtype=int32, numpy=array([1], dtype=int32)> ##   ..$ :<tf.Tensor: shape=(1), dtype=int32, numpy=array([1], dtype=int32)> ##   ..$ :<tf.Tensor: shape=(1), dtype=int32, numpy=array([1], dtype=int32)> ##  $ y:List of 3 ##   ..$ :<tf.Tensor: shape=(1), dtype=int32, numpy=array([2], dtype=int32)> ##   ..$ :<tf.Tensor: shape=(1), dtype=int32, numpy=array([2], dtype=int32)> ##   ..$ :<tf.Tensor: shape=(1), dtype=int32, numpy=array([2], dtype=int32)> ys ## [[1]] ## tf.Tensor([1 3], shape=(2), dtype=int32) ## ## [[2]] ## tf.Tensor([1 3], shape=(2), dtype=int32) ## ## [[3]] ## tf.Tensor([1 3], shape=(2), dtype=int32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_average.html","id":null,"dir":"Reference","previous_headings":"","what":"Compute the weighted average along the specified axis. — op_average","title":"Compute the weighted average along the specified axis. — op_average","text":"Compute weighted average along specified axis.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_average.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Compute the weighted average along the specified axis. — op_average","text":"","code":"op_average(x, axis = NULL, weights = NULL)"},{"path":"https://keras3.posit.co/dev/reference/op_average.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Compute the weighted average along the specified axis. — op_average","text":"x Input tensor. axis Integer along average x. default, axis = NULL, average elements input tensor. axis negative counts last first axis. weights Tensor weights associated values x. value x contributes average according associated weight. weights array can either 1-D (case length must size along given axis) shape x. weights = NULL (default), data x assumed weight equal one. 1-D calculation : avg = sum(* weights) / sum(weights). constraint weights sum(weights) must 0.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_average.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Compute the weighted average along the specified axis. — op_average","text":"Return average along specified axis.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_average.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Compute the weighted average along the specified axis. — op_average","text":"","code":"data <- op_arange(1, 5, dtype = \"int32\") data ## tf.Tensor([1 2 3 4 5], shape=(5), dtype=int32) op_average(data) ## tf.Tensor(3.0, shape=(), dtype=float32) op_average(   op_arange(1, 11),   weights = op_arange(10, 0, -1) ) ## tf.Tensor(4.0, shape=(), dtype=float32) data <- op_arange(6) |> op_reshape(c(3, 2)) data ## tf.Tensor( ## [[1. 2.] ##  [3. 4.] ##  [5. 6.]], shape=(3, 2), dtype=float32) op_average(   data,   axis = 2,   weights = op_array(c(1/4, 3/4)) ) ## tf.Tensor([1.75 3.75 5.75], shape=(3), dtype=float32) # Error: Axis must be specified when shapes of x and weights differ. try(op_average(   data,   weights = op_array(c(1/4, 3/4)) )) ## Error in op_average(data, weights = op_array(c(1/4, 3/4))) : ##   Axis must be specified when shapes of x and weights differ."},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_average_pool.html","id":null,"dir":"Reference","previous_headings":"","what":"Average pooling operation. — op_average_pool","title":"Average pooling operation. — op_average_pool","text":"Average pooling operation.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_average_pool.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Average pooling operation. — op_average_pool","text":"","code":"op_average_pool(   inputs,   pool_size,   strides = NULL,   padding = \"valid\",   data_format = NULL )"},{"path":"https://keras3.posit.co/dev/reference/op_average_pool.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Average pooling operation. — op_average_pool","text":"inputs Tensor rank N+2. inputs shape (batch_size,) + inputs_spatial_shape + (num_channels,) data_format = \"channels_last\", (batch_size, num_channels) + inputs_spatial_shape data_format = \"channels_first\". Pooling happens spatial dimensions . pool_size int tuple/list integers size len(inputs_spatial_shape), specifying size pooling window spatial dimension input tensor. pool_size int, every spatial dimension shares pool_size. strides int tuple/list integers size len(inputs_spatial_shape). stride sliding window spatial dimension input tensor. strides int, every spatial dimension shares strides. padding string, either \"valid\" \"\". \"valid\" means padding applied, \"\" results padding evenly left/right /input output height/width dimension input strides = 1. data_format string, either \"channels_last\" \"channels_first\". data_format determines ordering dimensions inputs. data_format = \"channels_last\", inputs shape (batch_size, ..., channels) data_format = \"channels_first\", inputs shape (batch_size, channels, ...).","code":""},{"path":"https://keras3.posit.co/dev/reference/op_average_pool.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Average pooling operation. — op_average_pool","text":"tensor rank N+2, result average pooling operation.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_bartlett.html","id":null,"dir":"Reference","previous_headings":"","what":"Bartlett window function. — op_bartlett","title":"Bartlett window function. — op_bartlett","text":"Bartlett window triangular window rises falls linearly.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_bartlett.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Bartlett window function. — op_bartlett","text":"","code":"op_bartlett(x)"},{"path":"https://keras3.posit.co/dev/reference/op_bartlett.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Bartlett window function. — op_bartlett","text":"x Length window. Must positive integer.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_bartlett.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Bartlett window function. — op_bartlett","text":"1D tensor containing window values.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_bartlett.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Bartlett window function. — op_bartlett","text":"","code":"op_bartlett(5) ## tf.Tensor([0.  0.5 1.  0.5 0. ], shape=(5), dtype=float32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_batch_normalization.html","id":null,"dir":"Reference","previous_headings":"","what":"Normalizes x by mean and variance. — op_batch_normalization","title":"Normalizes x by mean and variance. — op_batch_normalization","text":"op typically used batch normalization step neural network. normalizes input tensor along given axis.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_batch_normalization.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Normalizes x by mean and variance. — op_batch_normalization","text":"","code":"op_batch_normalization(   x,   mean,   variance,   axis,   offset = NULL,   scale = NULL,   epsilon = 0.001 )"},{"path":"https://keras3.posit.co/dev/reference/op_batch_normalization.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Normalizes x by mean and variance. — op_batch_normalization","text":"x Input tensor. mean mean vector length axis dimension input thensor. variance variance vector length axis dimension input tensor. axis Integer, axis normalized. offset offset vector length axis dimension input tensor. NULL, offset added normalized tensor. Defaults NULL. scale scale vector length axis dimension input tensor. NULL, normalized tensor multiplied scale. Defaults NULL. epsilon Small float added variance avoid dividing zero. Defaults 1e-3.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_batch_normalization.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Normalizes x by mean and variance. — op_batch_normalization","text":"normalized tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_batch_normalization.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Normalizes x by mean and variance. — op_batch_normalization","text":"","code":"x <- op_convert_to_tensor(rbind(c(0.1, 0.2, 0.3),                                 c(0.4, 0.5, 0.6),                                 c(0.7, 0.8, 0.9))) op_batch_normalization(   x,   mean = c(0.4, 0.5, 0.6),   variance = c(0.67, 0.67, 0.67),   axis = -1 ) ## tf.Tensor( ## [[-0.36623513 -0.36623513 -0.36623513] ##  [ 0.          0.          0.        ] ##  [ 0.36623513  0.36623513  0.36623513]], shape=(3, 3), dtype=float64)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_binary_crossentropy.html","id":null,"dir":"Reference","previous_headings":"","what":"Computes binary cross-entropy loss between target and output tensor. — op_binary_crossentropy","title":"Computes binary cross-entropy loss between target and output tensor. — op_binary_crossentropy","text":"binary cross-entropy loss commonly used binary classification tasks input sample belongs one two classes. measures dissimilarity target output probabilities logits.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_binary_crossentropy.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes binary cross-entropy loss between target and output tensor. — op_binary_crossentropy","text":"","code":"op_binary_crossentropy(target, output, from_logits = FALSE)"},{"path":"https://keras3.posit.co/dev/reference/op_binary_crossentropy.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Computes binary cross-entropy loss between target and output tensor. — op_binary_crossentropy","text":"target target tensor representing true binary labels. shape match shape output tensor. output output tensor representing predicted probabilities logits. shape match shape target tensor. from_logits (optional) Whether output tensor logits probabilities. Set TRUE output represents logits; otherwise, set FALSE output represents probabilities. Defaults FALSE.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_binary_crossentropy.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Computes binary cross-entropy loss between target and output tensor. — op_binary_crossentropy","text":"Integer tensor: computed binary cross-entropy loss target output.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_binary_crossentropy.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Computes binary cross-entropy loss between target and output tensor. — op_binary_crossentropy","text":"","code":"target <- op_array(c(0, 1, 1, 0)) output <- op_array(c(0.1, 0.9, 0.8, 0.2)) op_binary_crossentropy(target, output) ## tf.Tensor([0.10536055 0.10536055 0.22314353 0.22314353], shape=(4), dtype=float32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_bincount.html","id":null,"dir":"Reference","previous_headings":"","what":"Count the number of occurrences of each value in a tensor of integers. — op_bincount","title":"Count the number of occurrences of each value in a tensor of integers. — op_bincount","text":"Count number occurrences value tensor integers.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_bincount.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Count the number of occurrences of each value in a tensor of integers. — op_bincount","text":"","code":"op_bincount(x, weights = NULL, minlength = 0L, sparse = FALSE)"},{"path":"https://keras3.posit.co/dev/reference/op_bincount.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Count the number of occurrences of each value in a tensor of integers. — op_bincount","text":"x Input tensor. must dimension 1, must contain non-negative integer(s). weights Weight tensor. must length x. default value NULL. specified, x weighted , .e. n = x[], [n] += weight[] instead default behavior [n] += 1. minlength integer. default value 0. specified, least number bins output tensor. greater max(x) + 1, value output index higher max(x) set 0. sparse Whether return sparse tensor; backends support sparse tensors.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_bincount.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Count the number of occurrences of each value in a tensor of integers. — op_bincount","text":"1D tensor element gives number occurrence(s) index value x. length maximum max(x) + 1 minlength.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_bincount.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Count the number of occurrences of each value in a tensor of integers. — op_bincount","text":"","code":"(x <- op_array(c(1, 2, 2, 3), dtype = \"uint8\")) ## tf.Tensor([1 2 2 3], shape=(4), dtype=uint8) op_bincount(x) ## tf.Tensor([0 1 2 1], shape=(4), dtype=int32) (weights <- x / 2) ## tf.Tensor([0.5 1.  1.  1.5], shape=(4), dtype=float32) op_bincount(x, weights = weights) ## tf.Tensor([0.  0.5 2.  1.5], shape=(4), dtype=float32) minlength <- as.integer(op_max(x) + 1 + 2) # 6 op_bincount(x, minlength = minlength) ## tf.Tensor([0 1 2 1 0 0], shape=(6), dtype=int32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_bitwise_and.html","id":null,"dir":"Reference","previous_headings":"","what":"Compute the bit-wise AND of two arrays element-wise. — op_bitwise_and","title":"Compute the bit-wise AND of two arrays element-wise. — op_bitwise_and","text":"Computes bit-wise underlying binary representation integers input arrays. ufunc implements C/Python operator &.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_bitwise_and.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Compute the bit-wise AND of two arrays element-wise. — op_bitwise_and","text":"","code":"op_bitwise_and(x, y)"},{"path":"https://keras3.posit.co/dev/reference/op_bitwise_and.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Compute the bit-wise AND of two arrays element-wise. — op_bitwise_and","text":"x Input integer tensor. y Input integer tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_bitwise_and.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Compute the bit-wise AND of two arrays element-wise. — op_bitwise_and","text":"Result tensor.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_bitwise_invert.html","id":null,"dir":"Reference","previous_headings":"","what":"Compute bit-wise inversion, or bit-wise NOT, element-wise. — op_bitwise_invert","title":"Compute bit-wise inversion, or bit-wise NOT, element-wise. — op_bitwise_invert","text":"Computes bit-wise underlying binary representation integers input arrays. ufunc implements C/Python operator ~.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_bitwise_invert.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Compute bit-wise inversion, or bit-wise NOT, element-wise. — op_bitwise_invert","text":"","code":"op_bitwise_invert(x)"},{"path":"https://keras3.posit.co/dev/reference/op_bitwise_invert.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Compute bit-wise inversion, or bit-wise NOT, element-wise. — op_bitwise_invert","text":"x Input integer tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_bitwise_invert.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Compute bit-wise inversion, or bit-wise NOT, element-wise. — op_bitwise_invert","text":"Result tensor.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_bitwise_left_shift.html","id":null,"dir":"Reference","previous_headings":"","what":"Shift the bits of an integer to the left. — op_bitwise_left_shift","title":"Shift the bits of an integer to the left. — op_bitwise_left_shift","text":"Bits shifted left appending y 0s right x. Since internal representation numbers binary format, operation equivalent multiplying x 2**y.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_bitwise_left_shift.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Shift the bits of an integer to the left. — op_bitwise_left_shift","text":"","code":"op_bitwise_left_shift(x, y)"},{"path":"https://keras3.posit.co/dev/reference/op_bitwise_left_shift.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Shift the bits of an integer to the left. — op_bitwise_left_shift","text":"x Input integer tensor. y Input integer tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_bitwise_left_shift.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Shift the bits of an integer to the left. — op_bitwise_left_shift","text":"Result tensor.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_bitwise_not.html","id":null,"dir":"Reference","previous_headings":"","what":"Compute bit-wise inversion, or bit-wise NOT, element-wise. — op_bitwise_not","title":"Compute bit-wise inversion, or bit-wise NOT, element-wise. — op_bitwise_not","text":"Computes bit-wise underlying binary representation integers input arrays. ufunc implements C/Python operator ~.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_bitwise_not.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Compute bit-wise inversion, or bit-wise NOT, element-wise. — op_bitwise_not","text":"","code":"op_bitwise_not(x)"},{"path":"https://keras3.posit.co/dev/reference/op_bitwise_not.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Compute bit-wise inversion, or bit-wise NOT, element-wise. — op_bitwise_not","text":"x Input integer tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_bitwise_not.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Compute bit-wise inversion, or bit-wise NOT, element-wise. — op_bitwise_not","text":"Result tensor.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_bitwise_or.html","id":null,"dir":"Reference","previous_headings":"","what":"Compute the bit-wise OR of two arrays element-wise. — op_bitwise_or","title":"Compute the bit-wise OR of two arrays element-wise. — op_bitwise_or","text":"Computes bit-wise underlying binary representation integers input arrays. ufunc implements C/Python operator |.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_bitwise_or.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Compute the bit-wise OR of two arrays element-wise. — op_bitwise_or","text":"","code":"op_bitwise_or(x, y)"},{"path":"https://keras3.posit.co/dev/reference/op_bitwise_or.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Compute the bit-wise OR of two arrays element-wise. — op_bitwise_or","text":"x Input integer tensor. y Input integer tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_bitwise_or.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Compute the bit-wise OR of two arrays element-wise. — op_bitwise_or","text":"Result tensor.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_bitwise_right_shift.html","id":null,"dir":"Reference","previous_headings":"","what":"Shift the bits of an integer to the right. — op_bitwise_right_shift","title":"Shift the bits of an integer to the right. — op_bitwise_right_shift","text":"Bits shifted right y. internal representation numbers binary format, operation equivalent dividing x 2**y.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_bitwise_right_shift.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Shift the bits of an integer to the right. — op_bitwise_right_shift","text":"","code":"op_bitwise_right_shift(x, y)"},{"path":"https://keras3.posit.co/dev/reference/op_bitwise_right_shift.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Shift the bits of an integer to the right. — op_bitwise_right_shift","text":"x Input integer tensor. y Input integer tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_bitwise_right_shift.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Shift the bits of an integer to the right. — op_bitwise_right_shift","text":"Result tensor.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_bitwise_xor.html","id":null,"dir":"Reference","previous_headings":"","what":"Compute the bit-wise XOR of two arrays element-wise. — op_bitwise_xor","title":"Compute the bit-wise XOR of two arrays element-wise. — op_bitwise_xor","text":"Computes bit-wise XOR underlying binary representation integers input arrays. ufunc implements C/Python operator ^.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_bitwise_xor.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Compute the bit-wise XOR of two arrays element-wise. — op_bitwise_xor","text":"","code":"op_bitwise_xor(x, y)"},{"path":"https://keras3.posit.co/dev/reference/op_bitwise_xor.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Compute the bit-wise XOR of two arrays element-wise. — op_bitwise_xor","text":"x Input integer tensor. y Input integer tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_bitwise_xor.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Compute the bit-wise XOR of two arrays element-wise. — op_bitwise_xor","text":"Result tensor.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_blackman.html","id":null,"dir":"Reference","previous_headings":"","what":"Blackman window function. — op_blackman","title":"Blackman window function. — op_blackman","text":"Blackman window taper formed using weighted cosine.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_blackman.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Blackman window function. — op_blackman","text":"","code":"op_blackman(x)"},{"path":"https://keras3.posit.co/dev/reference/op_blackman.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Blackman window function. — op_blackman","text":"x Length window. Must positive integer.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_blackman.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Blackman window function. — op_blackman","text":"1D tensor containing window values.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_blackman.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Blackman window function. — op_blackman","text":"[-1.4901161e-08  3.4000003e-01  9.9999994e-01  3.3999997e-01","code":"op_blackman(5) ## tf.Tensor( ## [-1.4901161e-08  3.4000003e-01  9.9999994e-01  3.3999997e-01 ##  -1.4901161e-08], shape=(5), dtype=float32)"},{"path":[]},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_broadcast_to.html","id":null,"dir":"Reference","previous_headings":"","what":"Broadcast a tensor to a new shape. — op_broadcast_to","title":"Broadcast a tensor to a new shape. — op_broadcast_to","text":"Broadcast tensor new shape.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_broadcast_to.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Broadcast a tensor to a new shape. — op_broadcast_to","text":"","code":"op_broadcast_to(x, shape)"},{"path":"https://keras3.posit.co/dev/reference/op_broadcast_to.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Broadcast a tensor to a new shape. — op_broadcast_to","text":"x tensor broadcast. shape shape desired tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_broadcast_to.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Broadcast a tensor to a new shape. — op_broadcast_to","text":"tensor desired shape.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_broadcast_to.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Broadcast a tensor to a new shape. — op_broadcast_to","text":"","code":"x <- op_array(c(1, 2, 3)) op_broadcast_to(x, shape = c(3, 3)) ## tf.Tensor( ## [[1. 2. 3.] ##  [1. 2. 3.] ##  [1. 2. 3.]], shape=(3, 3), dtype=float32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_cast.html","id":null,"dir":"Reference","previous_headings":"","what":"Cast a tensor to the desired dtype. — op_cast","title":"Cast a tensor to the desired dtype. — op_cast","text":"Cast tensor desired dtype.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_cast.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Cast a tensor to the desired dtype. — op_cast","text":"","code":"op_cast(x, dtype)"},{"path":"https://keras3.posit.co/dev/reference/op_cast.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Cast a tensor to the desired dtype. — op_cast","text":"x tensor variable. dtype target type.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_cast.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Cast a tensor to the desired dtype. — op_cast","text":"tensor specified dtype.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_cast.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Cast a tensor to the desired dtype. — op_cast","text":"","code":"(x <- op_arange(4)) ## tf.Tensor([1. 2. 3. 4.], shape=(4), dtype=float32) op_cast(x, dtype = \"float16\") ## tf.Tensor([1. 2. 3. 4.], shape=(4), dtype=float16)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_categorical_crossentropy.html","id":null,"dir":"Reference","previous_headings":"","what":"Computes categorical cross-entropy loss between target and output tensor. — op_categorical_crossentropy","title":"Computes categorical cross-entropy loss between target and output tensor. — op_categorical_crossentropy","text":"categorical cross-entropy loss commonly used multi-class classification tasks input sample can belong one multiple classes. measures dissimilarity target output probabilities logits.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_categorical_crossentropy.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes categorical cross-entropy loss between target and output tensor. — op_categorical_crossentropy","text":"","code":"op_categorical_crossentropy(target, output, from_logits = FALSE, axis = -1L)"},{"path":"https://keras3.posit.co/dev/reference/op_categorical_crossentropy.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Computes categorical cross-entropy loss between target and output tensor. — op_categorical_crossentropy","text":"target target tensor representing true categorical labels. shape match shape output tensor except last dimension. output output tensor representing predicted probabilities logits. shape match shape target tensor except last dimension. from_logits (optional) Whether output tensor logits probabilities. Set TRUE output represents logits; otherwise, set FALSE output represents probabilities. Defaults FALSE. axis (optional) axis along categorical cross-entropy computed. Defaults -1, corresponds last dimension tensors.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_categorical_crossentropy.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Computes categorical cross-entropy loss between target and output tensor. — op_categorical_crossentropy","text":"Integer tensor: computed categorical cross-entropy loss target output.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_categorical_crossentropy.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Computes categorical cross-entropy loss between target and output tensor. — op_categorical_crossentropy","text":"","code":"target <- op_array(rbind(c(1, 0, 0),                         c(0, 1, 0),                         c(0, 0, 1))) output <- op_array(rbind(c(0.9, 0.05, 0.05),                         c(0.1, 0.8, 0.1),                         c(0.2, 0.3, 0.5))) op_categorical_crossentropy(target, output) ## tf.Tensor([0.10536052 0.22314355 0.69314718], shape=(3), dtype=float64)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_cbrt.html","id":null,"dir":"Reference","previous_headings":"","what":"Computes the cube root of the input tensor, element-wise. — op_cbrt","title":"Computes the cube root of the input tensor, element-wise. — op_cbrt","text":"Returns real-valued cube root x, handling negative inputs real domain.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_cbrt.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes the cube root of the input tensor, element-wise. — op_cbrt","text":"","code":"op_cbrt(x)"},{"path":"https://keras3.posit.co/dev/reference/op_cbrt.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Computes the cube root of the input tensor, element-wise. — op_cbrt","text":"x Input tensor.Examples","code":"op_cbrt(c(-8, 0, 8)) ## tf.Tensor([-2.  0.  2.], shape=(3), dtype=float32)"},{"path":"https://keras3.posit.co/dev/reference/op_cbrt.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Computes the cube root of the input tensor, element-wise. — op_cbrt","text":"tensor containing cube root element x.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_ceil.html","id":null,"dir":"Reference","previous_headings":"","what":"Return the ceiling of the input, element-wise. — op_ceil","title":"Return the ceiling of the input, element-wise. — op_ceil","text":"ceil scalar x smallest integer , >= x.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_ceil.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Return the ceiling of the input, element-wise. — op_ceil","text":"","code":"op_ceil(x)"},{"path":"https://keras3.posit.co/dev/reference/op_ceil.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Return the ceiling of the input, element-wise. — op_ceil","text":"x Input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_ceil.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Return the ceiling of the input, element-wise. — op_ceil","text":"ceiling element x, float dtype.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_celu.html","id":null,"dir":"Reference","previous_headings":"","what":"Continuously-differentiable exponential linear unit. — op_celu","title":"Continuously-differentiable exponential linear unit. — op_celu","text":"defined : f(x) =  alpha * (exp(x / alpha) - 1) x < 0, f(x) = x x >= 0.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_celu.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Continuously-differentiable exponential linear unit. — op_celu","text":"","code":"op_celu(x, alpha = 1)"},{"path":"https://keras3.posit.co/dev/reference/op_celu.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Continuously-differentiable exponential linear unit. — op_celu","text":"x Input tensor. alpha value CELU formulation. Defaults 1.0.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_celu.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Continuously-differentiable exponential linear unit. — op_celu","text":"tensor shape x.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_celu.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Continuously-differentiable exponential linear unit. — op_celu","text":"","code":"x <- op_array(c(-1., 0., 1.)) op_celu(x) ## tf.Tensor([-0.63212055  0.          1.        ], shape=(3), dtype=float32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_cholesky.html","id":null,"dir":"Reference","previous_headings":"","what":"Computes the Cholesky decomposition of a positive semi-definite matrix. — op_cholesky","title":"Computes the Cholesky decomposition of a positive semi-definite matrix. — op_cholesky","text":"Computes Cholesky decomposition positive semi-definite matrix.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_cholesky.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes the Cholesky decomposition of a positive semi-definite matrix. — op_cholesky","text":"","code":"op_cholesky(x)"},{"path":"https://keras3.posit.co/dev/reference/op_cholesky.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Computes the Cholesky decomposition of a positive semi-definite matrix. — op_cholesky","text":"x Input tensor shape (..., M, M).","code":""},{"path":"https://keras3.posit.co/dev/reference/op_cholesky.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Computes the Cholesky decomposition of a positive semi-definite matrix. — op_cholesky","text":"tensor shape (..., M, M) representing lower triangular Cholesky factor x.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_clip.html","id":null,"dir":"Reference","previous_headings":"","what":"Clip (limit) the values in a tensor. — op_clip","title":"Clip (limit) the values in a tensor. — op_clip","text":"Given interval, values outside interval clipped interval edges. example, interval [0, 1] specified, values smaller 0 become 0, values larger 1 become 1.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_clip.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Clip (limit) the values in a tensor. — op_clip","text":"","code":"op_clip(x, x_min, x_max)"},{"path":"https://keras3.posit.co/dev/reference/op_clip.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Clip (limit) the values in a tensor. — op_clip","text":"x Input tensor. x_min Minimum value. x_max Maximum value.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_clip.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Clip (limit) the values in a tensor. — op_clip","text":"clipped tensor.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_concatenate.html","id":null,"dir":"Reference","previous_headings":"","what":"Join a sequence of tensors along an existing axis. — op_concatenate","title":"Join a sequence of tensors along an existing axis. — op_concatenate","text":"Join sequence tensors along existing axis.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_concatenate.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Join a sequence of tensors along an existing axis. — op_concatenate","text":"","code":"op_concatenate(xs, axis = 1L)"},{"path":"https://keras3.posit.co/dev/reference/op_concatenate.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Join a sequence of tensors along an existing axis. — op_concatenate","text":"xs sequence tensors concatenate. axis axis along tensors joined. Defaults 0.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_concatenate.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Join a sequence of tensors along an existing axis. — op_concatenate","text":"concatenated tensor.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_cond.html","id":null,"dir":"Reference","previous_headings":"","what":"Conditionally applies true_fn or false_fn. — op_cond","title":"Conditionally applies true_fn or false_fn. — op_cond","text":"Conditionally applies true_fn false_fn.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_cond.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Conditionally applies true_fn or false_fn. — op_cond","text":"","code":"op_cond(pred, true_fn, false_fn)"},{"path":"https://keras3.posit.co/dev/reference/op_cond.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Conditionally applies true_fn or false_fn. — op_cond","text":"pred Boolean scalar type true_fn Callable returning output pred == TRUE case. false_fn Callable returning output pred == FALSE case.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_cond.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Conditionally applies true_fn or false_fn. — op_cond","text":"output either true_fn false_fn depending pred.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_cond.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Conditionally applies true_fn or false_fn. — op_cond","text":"","code":"fn <- tensorflow::tf_function(function(x) {   op_cond(x > 0,     true_fn = \\() x + 1,     false_fn = \\() x - 1) })  fn(tensorflow::as_tensor(1)) ## tf.Tensor(2.0, shape=(), dtype=float64) fn(tensorflow::as_tensor(-1)) ## tf.Tensor(-2.0, shape=(), dtype=float64) # # Conditional side-effect (print only, no return value). file <- tempfile(fileext = \".txt\") fn <- tensorflow::tf_function(function(epochs) {   op_fori_loop(     0, epochs,     body_fun = \\(epoch, state) {       op_cond(epoch %% 20 == 0,               \\() {                 tensorflow::tf$print(                   \"epoch:\", epoch,                   output_stream = paste0(\"file://\", file))                 NULL               },               \\() {NULL})       state     },     init_val = tensorflow::as_tensor(0, \"float32\")) })  fn(tensorflow::as_tensor(100, \"float32\")) ## tf.Tensor(0.0, shape=(), dtype=float32) readLines(file) ## [1] \"epoch: 0\"  \"epoch: 20\" \"epoch: 40\" \"epoch: 60\" \"epoch: 80\" # cleanup unlink(file)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_conj.html","id":null,"dir":"Reference","previous_headings":"","what":"Returns the complex conjugate, element-wise. — op_conj","title":"Returns the complex conjugate, element-wise. — op_conj","text":"complex conjugate complex number obtained changing sign imaginary part.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_conj.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Returns the complex conjugate, element-wise. — op_conj","text":"","code":"op_conj(x)"},{"path":"https://keras3.posit.co/dev/reference/op_conj.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Returns the complex conjugate, element-wise. — op_conj","text":"x Input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_conj.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Returns the complex conjugate, element-wise. — op_conj","text":"complex conjugate element x.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_conv.html","id":null,"dir":"Reference","previous_headings":"","what":"General N-D convolution. — op_conv","title":"General N-D convolution. — op_conv","text":"ops supports 1D, 2D 3D convolution.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_conv.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"General N-D convolution. — op_conv","text":"","code":"op_conv(   inputs,   kernel,   strides = 1L,   padding = \"valid\",   data_format = NULL,   dilation_rate = 1L )"},{"path":"https://keras3.posit.co/dev/reference/op_conv.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"General N-D convolution. — op_conv","text":"inputs Tensor rank N+2. inputs shape (batch_size,) + inputs_spatial_shape + (num_channels,) data_format = \"channels_last\", (batch_size, num_channels) + inputs_spatial_shape data_format = \"channels_first\". kernel Tensor rank N+2. kernel shape (kernel_spatial_shape, num_input_channels, num_output_channels). num_input_channels match number channels inputs. strides int int tuple/list len(inputs_spatial_shape), specifying strides convolution along spatial dimension. strides int, every spatial dimension shares strides. Defaults 1. padding string, either \"valid\" \"\". \"valid\" means padding applied, \"\" results padding evenly left/right /input output height/width dimension input strides = 1. data_format string, either \"channels_last\" \"channels_first\". data_format determines ordering dimensions inputs. data_format = \"channels_last\", inputs shape (batch_size, ..., channels) data_format = \"channels_first\", inputs shape (batch_size, channels, ...). dilation_rate int int tuple/list len(inputs_spatial_shape), specifying dilation rate use dilated convolution. dilation_rate int, every spatial dimension shares dilation_rate.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_conv.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"General N-D convolution. — op_conv","text":"tensor rank N+2, result conv operation.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_conv_transpose.html","id":null,"dir":"Reference","previous_headings":"","what":"General N-D convolution transpose. — op_conv_transpose","title":"General N-D convolution transpose. — op_conv_transpose","text":"Also known de-convolution. ops supports 1D, 2D 3D convolution.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_conv_transpose.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"General N-D convolution transpose. — op_conv_transpose","text":"","code":"op_conv_transpose(   inputs,   kernel,   strides = 1L,   padding = \"valid\",   output_padding = NULL,   data_format = NULL,   dilation_rate = 1L )"},{"path":"https://keras3.posit.co/dev/reference/op_conv_transpose.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"General N-D convolution transpose. — op_conv_transpose","text":"inputs Tensor rank N+2. inputs shape (batch_size,) + inputs_spatial_shape + (num_channels,) data_format = \"channels_last\", (batch_size, num_channels) + inputs_spatial_shape data_format = \"channels_first\". kernel Tensor rank N+2. kernel shape [kernel_spatial_shape, num_output_channels, num_input_channels], num_input_channels match number channels inputs. strides int int tuple/list len(inputs_spatial_shape), specifying strides convolution along spatial dimension. strides int, every spatial dimension shares strides. padding string, either \"valid\" \"\". \"valid\" means padding applied, \"\" results padding evenly left/right /input output height/width dimension input strides = 1. output_padding int int tuple/list len(inputs_spatial_shape), specifying amount padding along height width output tensor. Can single integer specify value spatial dimensions. amount output padding along given dimension must lower stride along dimension. set NULL (default), output shape inferred. data_format string, either \"channels_last\" \"channels_first\". data_format determines ordering dimensions inputs. data_format = \"channels_last\", inputs shape (batch_size, ..., channels) data_format = \"channels_first\", inputs shape (batch_size, channels, ...). dilation_rate int int tuple/list len(inputs_spatial_shape), specifying dilation rate use dilated convolution. dilation_rate int, every spatial dimension shares dilation_rate.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_conv_transpose.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"General N-D convolution transpose. — op_conv_transpose","text":"tensor rank N+2, result conv operation.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_convert_to_numpy.html","id":null,"dir":"Reference","previous_headings":"","what":"Convert a tensor to an R or NumPy array. — op_convert_to_numpy","title":"Convert a tensor to an R or NumPy array. — op_convert_to_numpy","text":"Convert tensor R NumPy array.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_convert_to_numpy.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Convert a tensor to an R or NumPy array. — op_convert_to_numpy","text":"","code":"op_convert_to_numpy(x)  op_convert_to_array(x)"},{"path":"https://keras3.posit.co/dev/reference/op_convert_to_numpy.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Convert a tensor to an R or NumPy array. — op_convert_to_numpy","text":"x tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_convert_to_numpy.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Convert a tensor to an R or NumPy array. — op_convert_to_numpy","text":"NumPy array R array.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_convert_to_tensor.html","id":null,"dir":"Reference","previous_headings":"","what":"Convert an array to a tensor. — op_convert_to_tensor","title":"Convert an array to a tensor. — op_convert_to_tensor","text":"Native tensors current backend left unchanged unless dtype, sparse ragged arguments set.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_convert_to_tensor.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Convert an array to a tensor. — op_convert_to_tensor","text":"","code":"op_convert_to_tensor(x, dtype = NULL, sparse = NULL, ragged = NULL)"},{"path":"https://keras3.posit.co/dev/reference/op_convert_to_tensor.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Convert an array to a tensor. — op_convert_to_tensor","text":"x R NumPy array (can nested) backend tensor. dtype target type. NULL, type x used. sparse Whether keep sparse tensors. FALSE cause sparse tensors densified. default value NULL means sparse tensors kept backend supports . ragged Whether keep ragged tensors. FALSE cause ragged tensors densified. default value NULL means ragged tensors kept backend supports .","code":""},{"path":"https://keras3.posit.co/dev/reference/op_convert_to_tensor.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Convert an array to a tensor. — op_convert_to_tensor","text":"backend tensor specified dtype sparseness.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_convert_to_tensor.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Convert an array to a tensor. — op_convert_to_tensor","text":"","code":"x <- array(c(1, 2, 3)) y <- op_convert_to_tensor(x) y ## tf.Tensor([1. 2. 3.], shape=(3), dtype=float64) op_convert_to_tensor(c(1, 3, 2, 0), \"int32\") ## tf.Tensor([1 3 2 0], shape=(4), dtype=int32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_copy.html","id":null,"dir":"Reference","previous_headings":"","what":"Returns a copy of x. — op_copy","title":"Returns a copy of x. — op_copy","text":"Returns copy x.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_copy.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Returns a copy of x. — op_copy","text":"","code":"op_copy(x)"},{"path":"https://keras3.posit.co/dev/reference/op_copy.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Returns a copy of x. — op_copy","text":"x Input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_copy.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Returns a copy of x. — op_copy","text":"copy x.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_corrcoef.html","id":null,"dir":"Reference","previous_headings":"","what":"Compute the Pearson correlation coefficient matrix. — op_corrcoef","title":"Compute the Pearson correlation coefficient matrix. — op_corrcoef","text":"Compute Pearson correlation coefficient matrix.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_corrcoef.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Compute the Pearson correlation coefficient matrix. — op_corrcoef","text":"","code":"op_corrcoef(x)"},{"path":"https://keras3.posit.co/dev/reference/op_corrcoef.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Compute the Pearson correlation coefficient matrix. — op_corrcoef","text":"x 2D tensor shape (N, D), N number variables D number observations.Examples","code":"x <- op_array(matrix(c(1, 2, 3,                        2, 3, 4), nrow = 2, byrow = TRUE)) op_corrcoef(x) ## tf.Tensor( ## [[1. 1.] ##  [1. 1.]], shape=(2, 2), dtype=float64)"},{"path":"https://keras3.posit.co/dev/reference/op_corrcoef.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Compute the Pearson correlation coefficient matrix. — op_corrcoef","text":"tensor shape (N, N) representing correlation matrix.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_correlate.html","id":null,"dir":"Reference","previous_headings":"","what":"Compute the cross-correlation of two 1-dimensional tensors. — op_correlate","title":"Compute the cross-correlation of two 1-dimensional tensors. — op_correlate","text":"Compute cross-correlation two 1-dimensional tensors.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_correlate.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Compute the cross-correlation of two 1-dimensional tensors. — op_correlate","text":"","code":"op_correlate(x1, x2, mode = \"valid\")"},{"path":"https://keras3.posit.co/dev/reference/op_correlate.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Compute the cross-correlation of two 1-dimensional tensors. — op_correlate","text":"x1 First 1-dimensional input tensor length M. x2 Second 1-dimensional input tensor length N. mode Either \"valid\", \"\" \"full\". default mode set \"valid\", returns output length max(M, N) - min(M, N) + 1. \"\" returns output length max(M, N). \"full\" mode returns convolution point overlap, output length N+M-1.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_correlate.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Compute the cross-correlation of two 1-dimensional tensors. — op_correlate","text":"Output tensor, cross-correlation x1 x2.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_cos.html","id":null,"dir":"Reference","previous_headings":"","what":"Cosine, element-wise. — op_cos","title":"Cosine, element-wise. — op_cos","text":"Cosine, element-wise.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_cos.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Cosine, element-wise. — op_cos","text":"","code":"op_cos(x)"},{"path":"https://keras3.posit.co/dev/reference/op_cos.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Cosine, element-wise. — op_cos","text":"x Input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_cos.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Cosine, element-wise. — op_cos","text":"corresponding cosine values.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_cosh.html","id":null,"dir":"Reference","previous_headings":"","what":"Hyperbolic cosine, element-wise. — op_cosh","title":"Hyperbolic cosine, element-wise. — op_cosh","text":"Hyperbolic cosine, element-wise.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_cosh.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Hyperbolic cosine, element-wise. — op_cosh","text":"","code":"op_cosh(x)"},{"path":"https://keras3.posit.co/dev/reference/op_cosh.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Hyperbolic cosine, element-wise. — op_cosh","text":"x Input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_cosh.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Hyperbolic cosine, element-wise. — op_cosh","text":"Output tensor shape x.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_count_nonzero.html","id":null,"dir":"Reference","previous_headings":"","what":"Counts the number of non-zero values in x along the given axis. — op_count_nonzero","title":"Counts the number of non-zero values in x along the given axis. — op_count_nonzero","text":"axis specified non-zeros tensor counted.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_count_nonzero.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Counts the number of non-zero values in x along the given axis. — op_count_nonzero","text":"","code":"op_count_nonzero(x, axis = NULL)"},{"path":"https://keras3.posit.co/dev/reference/op_count_nonzero.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Counts the number of non-zero values in x along the given axis. — op_count_nonzero","text":"x Input tensor. axis Axis tuple axes along count number non-zeros. Defaults NULL.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_count_nonzero.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Counts the number of non-zero values in x along the given axis. — op_count_nonzero","text":"integer tensor integers.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_count_nonzero.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Counts the number of non-zero values in x along the given axis. — op_count_nonzero","text":"","code":"x <- op_array(rbind(c(0, 1, 7, 0),                    c(3, 0, 2, 19))) op_count_nonzero(x) ## tf.Tensor(5, shape=(), dtype=int32) op_count_nonzero(x, axis = 1) ## tf.Tensor([1 1 2 1], shape=(4), dtype=int32) op_count_nonzero(x, axis = 2) ## tf.Tensor([2 3], shape=(2), dtype=int32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_cross.html","id":null,"dir":"Reference","previous_headings":"","what":"Returns the cross product of two (arrays of) vectors. — op_cross","title":"Returns the cross product of two (arrays of) vectors. — op_cross","text":"cross product x1 x2 R^3 vector perpendicular x1 x2. x1 x2 arrays vectors, vectors defined last axis x1 x2 default, axes can dimensions 2 3. dimension either x1 x2 2, third component input vector assumed zero cross product calculated accordingly. cases input vectors dimension 2, z-component cross product returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_cross.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Returns the cross product of two (arrays of) vectors. — op_cross","text":"","code":"op_cross(x1, x2, axisa = -1L, axisb = -1L, axisc = -1L, axis = NULL)"},{"path":"https://keras3.posit.co/dev/reference/op_cross.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Returns the cross product of two (arrays of) vectors. — op_cross","text":"x1 Components first vector(s). x2 Components second vector(s). axisa Axis x1 defines vector(s). Defaults -1. axisb Axis x2 defines vector(s). Defaults -1. axisc Axis result containing cross product vector(s). Ignored input vectors dimension 2, return scalar. default, last axis. axis defined, axis x1, x2 result defines vector(s) cross product(s). Overrides axisa, axisb axisc.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_cross.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Returns the cross product of two (arrays of) vectors. — op_cross","text":"Vector cross product(s).","code":""},{"path":"https://keras3.posit.co/dev/reference/op_cross.html","id":"note","dir":"Reference","previous_headings":"","what":"Note","title":"Returns the cross product of two (arrays of) vectors. — op_cross","text":"Torch backend support two dimensional vectors, arguments axisa, axisb axisc. Use axis instead.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_ctc_decode.html","id":null,"dir":"Reference","previous_headings":"","what":"Decodes the output of a CTC model. — op_ctc_decode","title":"Decodes the output of a CTC model. — op_ctc_decode","text":"Decodes output CTC model.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_ctc_decode.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Decodes the output of a CTC model. — op_ctc_decode","text":"","code":"op_ctc_decode(   inputs,   sequence_lengths,   strategy = \"greedy\",   beam_width = 100L,   top_paths = 1L,   merge_repeated = TRUE,   mask_index = 1L )"},{"path":"https://keras3.posit.co/dev/reference/op_ctc_decode.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Decodes the output of a CTC model. — op_ctc_decode","text":"inputs tensor shape (batch_size, max_length, num_classes) containing logits (output model). normalized via softmax. sequence_lengths tensor shape (batch_size) containing sequence lengths batch. strategy string decoding strategy. Supported values \"greedy\" \"beam_search\". beam_width integer scalar beam width used beam search. Defaults 100. top_paths integer scalar, number top paths return. Defaults 1. merge_repeated boolean scalar, whether merge repeated labels output. Defaults TRUE. mask_index integer scalar, (1-based) index mask character vocabulary. Defaults 1.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_ctc_decode.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Decodes the output of a CTC model. — op_ctc_decode","text":"list containing: tensor representing list decoded sequences. strategy=\"greedy\", shape (1, batch_size, max_length). strategy=\"beam_search\", shape (top_paths, batch_size, max_length). Note : -1 indicates blank label. strategy=\"greedy\", tensor shape (batch_size, 1) representing negative sum probability logits sequence. strategy=\"beam_search\", tensor shape (batch_size, top_paths) representing log probability sequence.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_ctc_loss.html","id":null,"dir":"Reference","previous_headings":"","what":"CTC (Connectionist Temporal Classification) loss. — op_ctc_loss","title":"CTC (Connectionist Temporal Classification) loss. — op_ctc_loss","text":"CTC (Connectionist Temporal Classification) loss.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_ctc_loss.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"CTC (Connectionist Temporal Classification) loss. — op_ctc_loss","text":"","code":"op_ctc_loss(target, output, target_length, output_length, mask_index = 1L)"},{"path":"https://keras3.posit.co/dev/reference/op_ctc_loss.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"CTC (Connectionist Temporal Classification) loss. — op_ctc_loss","text":"target tensor shape (batch_size, max_length) containing true labels integer format. output tensor shape (batch_size, max_length, num_classes) containing logits (output model). target_length tensor shape (batch_size) containing true label lengths. output_length tensor shape (batch_size) containing output lengths. mask_index index mask character vocabulary. Defaults 1 (first element vocabulary).","code":""},{"path":"https://keras3.posit.co/dev/reference/op_ctc_loss.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"CTC (Connectionist Temporal Classification) loss. — op_ctc_loss","text":"tensor, shape (batch_size), loss values.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_cumprod.html","id":null,"dir":"Reference","previous_headings":"","what":"Return the cumulative product of elements along a given axis. — op_cumprod","title":"Return the cumulative product of elements along a given axis. — op_cumprod","text":"Return cumulative product elements along given axis.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_cumprod.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Return the cumulative product of elements along a given axis. — op_cumprod","text":"","code":"op_cumprod(x, axis = NULL, dtype = NULL)"},{"path":"https://keras3.posit.co/dev/reference/op_cumprod.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Return the cumulative product of elements along a given axis. — op_cumprod","text":"x Input tensor. axis Axis along cumulative product computed. default input flattened. dtype dtype returned tensor. Defaults x$dtype.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_cumprod.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Return the cumulative product of elements along a given axis. — op_cumprod","text":"Output tensor.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_cumsum.html","id":null,"dir":"Reference","previous_headings":"","what":"Returns the cumulative sum of elements along a given axis. — op_cumsum","title":"Returns the cumulative sum of elements along a given axis. — op_cumsum","text":"Returns cumulative sum elements along given axis.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_cumsum.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Returns the cumulative sum of elements along a given axis. — op_cumsum","text":"","code":"op_cumsum(x, axis = NULL, dtype = NULL)"},{"path":"https://keras3.posit.co/dev/reference/op_cumsum.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Returns the cumulative sum of elements along a given axis. — op_cumsum","text":"x Input tensor. axis Axis along cumulative sum computed. default input flattened. dtype dtype returned tensor. Defaults x$dtype.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_cumsum.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Returns the cumulative sum of elements along a given axis. — op_cumsum","text":"Output tensor.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_custom_gradient.html","id":null,"dir":"Reference","previous_headings":"","what":"Decorator to define a function with a custom gradient. — op_custom_gradient","title":"Decorator to define a function with a custom gradient. — op_custom_gradient","text":"decorator allows fine grained control gradients sequence operations. may useful multiple reasons, including providing efficient numerically stable gradient sequence operations.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_custom_gradient.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Decorator to define a function with a custom gradient. — op_custom_gradient","text":"","code":"op_custom_gradient(f)"},{"path":"https://keras3.posit.co/dev/reference/op_custom_gradient.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Decorator to define a function with a custom gradient. — op_custom_gradient","text":"f Function f(...) returns tuple (output, grad_fn) : ... sequence unnamed arguments, tensor input nested structure tensor inputs function. output (potentially nested structure ) tensor outputs applying operations forward_fn f() .... grad_fn function signature grad_fn(..., upstream) returns list tensors size (flattened) ...: derivatives tensors output respect tensors .... upstream tensor sequence tensors holding initial value gradients tensor output.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_custom_gradient.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Decorator to define a function with a custom gradient. — op_custom_gradient","text":"function h(...) returns value f(...)[[1]] whose gradient determined f(...)[[2]].","code":""},{"path":"https://keras3.posit.co/dev/reference/op_custom_gradient.html","id":"note","dir":"Reference","previous_headings":"","what":"Note","title":"Decorator to define a function with a custom gradient. — op_custom_gradient","text":"Note grad function returns gradient computation requires ... well upstream named argument, depending backend set. JAX TensorFlow backends, requires one argument, whereas might use upstream argument case PyTorch backend. working TensorFlow/JAX backend, grad(upstream) sufficient. PyTorch, grad function requires ... well upstream, e.g. grad <- \\(..., upstream). Follow example use op_custom_gradient() way compatible backends.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_custom_gradient.html","id":"example","dir":"Reference","previous_headings":"","what":"Example","title":"Decorator to define a function with a custom gradient. — op_custom_gradient","text":"Backend-agnostic example.","code":"log1pexp <- op_custom_gradient(\\(x) {      e <- op_exp(x)      grad <- function(..., upstream = NULL) {       upstream <- upstream %||% ..1       op_multiply(upstream, 1.0 - 1.0 / op_add(1, e))     }      tuple(op_log(1 + e), grad)  })  if(config_backend() == \"tensorflow\") {   tf <- tensorflow::tf   x <- op_convert_to_tensor(100.0)   with(tf$GradientTape() %as% tape, {     tape$watch(x)     y <- log1pexp(x)   })   dy_dx <- tape$gradient(y, x)   stopifnot(as.numeric(dy_dx) == 1) }"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_deg2rad.html","id":null,"dir":"Reference","previous_headings":"","what":"Convert angles from degrees to radians. — op_deg2rad","title":"Convert angles from degrees to radians. — op_deg2rad","text":"conversion defined : rad = deg * (pi / 180).","code":""},{"path":"https://keras3.posit.co/dev/reference/op_deg2rad.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Convert angles from degrees to radians. — op_deg2rad","text":"","code":"op_deg2rad(x)"},{"path":"https://keras3.posit.co/dev/reference/op_deg2rad.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Convert angles from degrees to radians. — op_deg2rad","text":"x Input tensor angles degrees.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_deg2rad.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Convert angles from degrees to radians. — op_deg2rad","text":"tensor containing angles converted radians.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_deg2rad.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Convert angles from degrees to radians. — op_deg2rad","text":"","code":"op_deg2rad(c(0, 90, 180)) ## tf.Tensor([0.        1.5707964 3.1415927], shape=(3), dtype=float32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_depthwise_conv.html","id":null,"dir":"Reference","previous_headings":"","what":"General N-D depthwise convolution. — op_depthwise_conv","title":"General N-D depthwise convolution. — op_depthwise_conv","text":"ops supports 1D 2D depthwise convolution.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_depthwise_conv.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"General N-D depthwise convolution. — op_depthwise_conv","text":"","code":"op_depthwise_conv(   inputs,   kernel,   strides = 1L,   padding = \"valid\",   data_format = NULL,   dilation_rate = 1L )"},{"path":"https://keras3.posit.co/dev/reference/op_depthwise_conv.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"General N-D depthwise convolution. — op_depthwise_conv","text":"inputs Tensor rank N+2. inputs shape (batch_size,) + inputs_spatial_shape + (num_channels,) data_format = \"channels_last\", (batch_size, num_channels) + inputs_spatial_shape data_format = \"channels_first\". kernel Tensor rank N+2. kernel shape [kernel_spatial_shape, num_input_channels, num_channels_multiplier], num_input_channels match number channels inputs. strides int int tuple/list len(inputs_spatial_shape), specifying strides convolution along spatial dimension. strides int, every spatial dimension shares strides. padding string, either \"valid\" \"\". \"valid\" means padding applied, \"\" results padding evenly left/right /input output height/width dimension input strides = 1. data_format string, either \"channels_last\" \"channels_first\". data_format determines ordering dimensions inputs. data_format = \"channels_last\", inputs shape (batch_size, ..., channels) data_format = \"channels_first\", inputs shape (batch_size, channels, ...). dilation_rate int int tuple/list len(inputs_spatial_shape), specifying dilation rate use dilated convolution. dilation_rate int, every spatial dimension shares dilation_rate.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_depthwise_conv.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"General N-D depthwise convolution. — op_depthwise_conv","text":"tensor rank N+2, result depthwise conv operation.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_det.html","id":null,"dir":"Reference","previous_headings":"","what":"Computes the determinant of a square tensor. — op_det","title":"Computes the determinant of a square tensor. — op_det","text":"Computes determinant square tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_det.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes the determinant of a square tensor. — op_det","text":"","code":"op_det(x)"},{"path":"https://keras3.posit.co/dev/reference/op_det.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Computes the determinant of a square tensor. — op_det","text":"x Input tensor shape (..., M, M).","code":""},{"path":"https://keras3.posit.co/dev/reference/op_det.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Computes the determinant of a square tensor. — op_det","text":"tensor shape (...) representing determinant x.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_diag.html","id":null,"dir":"Reference","previous_headings":"","what":"Extract a diagonal or construct a diagonal array. — op_diag","title":"Extract a diagonal or construct a diagonal array. — op_diag","text":"Extract diagonal construct diagonal array.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_diag.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Extract a diagonal or construct a diagonal array. — op_diag","text":"","code":"op_diag(x, k = 0L)"},{"path":"https://keras3.posit.co/dev/reference/op_diag.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Extract a diagonal or construct a diagonal array. — op_diag","text":"x Input tensor. x 2-D, returns k-th diagonal x. x 1-D, return 2-D tensor x k-th diagonal. k diagonal consider. Defaults 0. Use k > 0 diagonals main diagonal, k < 0 diagonals main diagonal.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_diag.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Extract a diagonal or construct a diagonal array. — op_diag","text":"extracted diagonal constructed diagonal tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_diag.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Extract a diagonal or construct a diagonal array. — op_diag","text":"","code":"x <- op_arange(9L) |> op_reshape(c(3, 3)) x ## tf.Tensor( ## [[1 2 3] ##  [4 5 6] ##  [7 8 9]], shape=(3, 3), dtype=int32) op_diag(x) ## tf.Tensor([1 5 9], shape=(3), dtype=int32) op_diag(x, k = 1) ## tf.Tensor([2 6], shape=(2), dtype=int32) op_diag(x, k = -1) ## tf.Tensor([4 8], shape=(2), dtype=int32) op_diag(op_diag(x)) ## tf.Tensor( ## [[1 0 0] ##  [0 5 0] ##  [0 0 9]], shape=(3, 3), dtype=int32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_diagflat.html","id":null,"dir":"Reference","previous_headings":"","what":"Create a two-dimensional array with the flattened input diagonal. — op_diagflat","title":"Create a two-dimensional array with the flattened input diagonal. — op_diagflat","text":"k-th diagonal.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_diagflat.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Create a two-dimensional array with the flattened input diagonal. — op_diagflat","text":"","code":"op_diagflat(x, k = 0L)"},{"path":"https://keras3.posit.co/dev/reference/op_diagflat.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Create a two-dimensional array with the flattened input diagonal. — op_diagflat","text":"x Input tensor flattened placed diagonal. k diagonal place flattened input. Defaults 0. Use k > 0 diagonals main diagonal, k < 0 diagonals main diagonal.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_diagflat.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Create a two-dimensional array with the flattened input diagonal. — op_diagflat","text":"2-D tensor flattened input specified diagonal.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_diagonal.html","id":null,"dir":"Reference","previous_headings":"","what":"Return specified diagonals. — op_diagonal","title":"Return specified diagonals. — op_diagonal","text":"x 2-D, returns diagonal x given offset, .e., collection elements form x[, +offset]. x two dimensions, axes specified axis1 axis2 used determine 2-D sub-array whose diagonal returned. shape resulting array can determined removing axis1 axis2 appending index right equal size resulting diagonals.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_diagonal.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Return specified diagonals. — op_diagonal","text":"","code":"op_diagonal(x, offset = 0L, axis1 = 1L, axis2 = 2L)"},{"path":"https://keras3.posit.co/dev/reference/op_diagonal.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Return specified diagonals. — op_diagonal","text":"x Input tensor. offset Offset diagonal main diagonal. Can positive negative. Defaults 0 (main diagonal). axis1 Axis used first axis 2-D sub-arrays. Defaults 1 (first axis). axis2 Axis used second axis 2-D sub-arrays. Defaults 2 (second axis).","code":""},{"path":"https://keras3.posit.co/dev/reference/op_diagonal.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Return specified diagonals. — op_diagonal","text":"Tensor diagonals.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_diagonal.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Return specified diagonals. — op_diagonal","text":"","code":"x <- op_arange(4L) |> op_reshape(c(2, 2)) x ## tf.Tensor( ## [[1 2] ##  [3 4]], shape=(2, 2), dtype=int32) op_diagonal(x) ## tf.Tensor([1 4], shape=(2), dtype=int32) op_diagonal(x, offset = 1) ## tf.Tensor([2], shape=(1), dtype=int32) x <- op_array(1:8) |> op_reshape(c(2, 2, 2)) x ## tf.Tensor( ## [[[1 2] ##   [3 4]] ## ##  [[5 6] ##   [7 8]]], shape=(2, 2, 2), dtype=int32) x |> op_diagonal(0) ## tf.Tensor( ## [[1 7] ##  [2 8]], shape=(2, 2), dtype=int32) x |> op_diagonal(0, 1, 2) # same as above, the default ## tf.Tensor( ## [[1 7] ##  [2 8]], shape=(2, 2), dtype=int32) x |> op_diagonal(0, 2, 3) ## tf.Tensor( ## [[1 4] ##  [5 8]], shape=(2, 2), dtype=int32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_diff.html","id":null,"dir":"Reference","previous_headings":"","what":"Calculate the n-th discrete difference along the given axis. — op_diff","title":"Calculate the n-th discrete difference along the given axis. — op_diff","text":"first difference given [] = [+1] - [] along given axis, higher differences calculated using diff recursively.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_diff.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Calculate the n-th discrete difference along the given axis. — op_diff","text":"","code":"op_diff(a, n = 1L, axis = -1L)"},{"path":"https://keras3.posit.co/dev/reference/op_diff.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Calculate the n-th discrete difference along the given axis. — op_diff","text":"Input tensor. n number times values differenced. Defaults 1. axis Axis compute discrete difference(s) along. Defaults -1 (last axis).","code":""},{"path":"https://keras3.posit.co/dev/reference/op_diff.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Calculate the n-th discrete difference along the given axis. — op_diff","text":"Tensor diagonals.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_diff.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Calculate the n-th discrete difference along the given axis. — op_diff","text":"","code":"x <- op_array(c(1, 2, 4, 7, 0)) op_diff(x) ## tf.Tensor([ 1.  2.  3. -7.], shape=(4), dtype=float32) op_diff(x, n = 2) ## tf.Tensor([  1.   1. -10.], shape=(3), dtype=float32) x <- op_array(rbind(c(1, 3, 6, 10),                     c(0, 5, 6, 8))) op_diff(x) ## tf.Tensor( ## [[2. 3. 4.] ##  [5. 1. 2.]], shape=(2, 3), dtype=float64) op_diff(x, axis = 1) ## tf.Tensor([[-1.  2.  0. -2.]], shape=(1, 4), dtype=float64)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_digitize.html","id":null,"dir":"Reference","previous_headings":"","what":"Returns the indices of the bins to which each value in x belongs. — op_digitize","title":"Returns the indices of the bins to which each value in x belongs. — op_digitize","text":"Returns indices bins value x belongs.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_digitize.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Returns the indices of the bins to which each value in x belongs. — op_digitize","text":"","code":"op_digitize(x, bins, zero_indexed = FALSE)"},{"path":"https://keras3.posit.co/dev/reference/op_digitize.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Returns the indices of the bins to which each value in x belongs. — op_digitize","text":"x Input array binned. bins Array bins. one-dimensional monotonically increasing. zero_indexed TRUE, returned indices zero-based (0 encodes first bin); FALSE (default), returned indices one-based (1 encodes first bin).","code":""},{"path":"https://keras3.posit.co/dev/reference/op_digitize.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Returns the indices of the bins to which each value in x belongs. — op_digitize","text":"Output array indices, shape x.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_digitize.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Returns the indices of the bins to which each value in x belongs. — op_digitize","text":"","code":"x <- op_array(c(-1,    0, 1,    3, 3.5,      4.5,  1.6,    8)) bins <- array(c(    0,       3,         4.5,            7)) op_digitize(x, bins) ## tf.Tensor([1 2 2 3 3 4 2 5], shape=(8), dtype=int32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_divide.html","id":null,"dir":"Reference","previous_headings":"","what":"Divide arguments element-wise. — op_divide","title":"Divide arguments element-wise. — op_divide","text":"Note function automatically called using R operator * tensor.","code":"(x <- op_arange(4)) ## tf.Tensor([1. 2. 3. 4.], shape=(4), dtype=float32) op_divide(x, 2) ## tf.Tensor([0.5 1.  1.5 2. ], shape=(4), dtype=float32) x / 2 ## tf.Tensor([0.5 1.  1.5 2. ], shape=(4), dtype=float32)"},{"path":"https://keras3.posit.co/dev/reference/op_divide.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Divide arguments element-wise. — op_divide","text":"","code":"op_divide(x1, x2)"},{"path":"https://keras3.posit.co/dev/reference/op_divide.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Divide arguments element-wise. — op_divide","text":"x1 First input tensor. x2 Second input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_divide.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Divide arguments element-wise. — op_divide","text":"Output tensor, quotient x1/x2, element-wise.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_divide.html","id":"example","dir":"Reference","previous_headings":"","what":"Example","title":"Divide arguments element-wise. — op_divide","text":"","code":"op_divide(3, 2) ## tf.Tensor(1.5, shape=(), dtype=float32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_divide_no_nan.html","id":null,"dir":"Reference","previous_headings":"","what":"Safe element-wise division which returns 0 where the denominator is 0. — op_divide_no_nan","title":"Safe element-wise division which returns 0 where the denominator is 0. — op_divide_no_nan","text":"Safe element-wise division returns 0 denominator 0.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_divide_no_nan.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Safe element-wise division which returns 0 where the denominator is 0. — op_divide_no_nan","text":"","code":"op_divide_no_nan(x1, x2)"},{"path":"https://keras3.posit.co/dev/reference/op_divide_no_nan.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Safe element-wise division which returns 0 where the denominator is 0. — op_divide_no_nan","text":"x1 First input tensor. x2 Second input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_divide_no_nan.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Safe element-wise division which returns 0 where the denominator is 0. — op_divide_no_nan","text":"quotient x1/x2, element-wise, zero x2 zero.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_dot.html","id":null,"dir":"Reference","previous_headings":"","what":"Dot product of two tensors. — op_dot","title":"Dot product of two tensors. — op_dot","text":"x1 x2 1-D tensors, inner product vectors (without complex conjugation). x1 x2 2-D tensors, matrix multiplication. either x1 x2 0-D (scalar), equivalent x1 * x2. x1 N-D tensor x2 1-D tensor, sum product last axis x1 x2. x1 N-D tensor x2 M-D tensor (M >= 2), sum product last axis x1 second--last axis x2: dot(x1, x2)[,j,k,m] = sum([,j,:] * b[k,:,m]).","code":""},{"path":"https://keras3.posit.co/dev/reference/op_dot.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Dot product of two tensors. — op_dot","text":"","code":"op_dot(x1, x2)"},{"path":"https://keras3.posit.co/dev/reference/op_dot.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Dot product of two tensors. — op_dot","text":"x1 First argument. x2 Second argument.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_dot.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Dot product of two tensors. — op_dot","text":"Dot product x1 x2.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_dot.html","id":"note","dir":"Reference","previous_headings":"","what":"Note","title":"Dot product of two tensors. — op_dot","text":"Torch backend accept 0-D tensors arguments.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_dot_product_attention.html","id":null,"dir":"Reference","previous_headings":"","what":"Scaled dot product attention function. — op_dot_product_attention","title":"Scaled dot product attention function. — op_dot_product_attention","text":"Computes attention function Q (query), K (key), V(value): attention(Q, K, V) = softmax(Q * K / sqrt(d)) * V. define logits output Q * K probs output softmax. Throughout function, utilize following notation represent shape array: B: batch size S: length key/value T: length query N: number attention heads H: dimensions attention head K: number key/value heads G: number groups, equals N // K","code":""},{"path":"https://keras3.posit.co/dev/reference/op_dot_product_attention.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Scaled dot product attention function. — op_dot_product_attention","text":"","code":"op_dot_product_attention(   query,   key,   value,   bias = NULL,   mask = NULL,   scale = NULL,   is_causal = FALSE,   flash_attention = NULL,   attn_logits_soft_cap = NULL )"},{"path":"https://keras3.posit.co/dev/reference/op_dot_product_attention.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Scaled dot product attention function. — op_dot_product_attention","text":"query query array shape (B, T, N, H). key key array shape (B, S, K, H). K equals N, multi-headed attention (MHA) performed. Otherwise, grouped query attention (GQA) performed N multiple K. multi-query attention (MQA) performed K==1 (special case GQA). value value array shape key. bias Optional bias array added logits. shape must broadcastable (B, N, T, S). mask Optional mask array used filter logits. boolean mask TRUE indicates element take part attention. additive mask, users pass bias. shape must broadcastable (B, N, T, S). scale Optional scale logits. NULL, scale set 1.0 / sqrt(H). is_causal Whether apply causal mask. flash_attention Whether use flash attention. NULL, attempt use flash attention required conditions met. Typically, inputs must float16 bfloat16 dtype input layout requirements may vary depending backend. attn_logits_soft_cap Optional numeric cap attention logits softmax. supported JAX TPU backend.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_dot_product_attention.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Scaled dot product attention function. — op_dot_product_attention","text":"array attention output shape query.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_dot_product_attention.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Scaled dot product attention function. — op_dot_product_attention","text":"","code":"query = random_normal(c(2, 4, 8, 16)) key = random_normal(c(2, 6, 8, 16)) value = random_normal(c(2, 6, 8, 16)) op_dot_product_attention(query, key, value) |> op_shape() ## shape(2, 4, 8, 16)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_dtype.html","id":null,"dir":"Reference","previous_headings":"","what":"Return the dtype of the tensor input as a standardized string. — op_dtype","title":"Return the dtype of the tensor input as a standardized string. — op_dtype","text":"Note due standardization, dtype compare equal backend-specific version dtype.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_dtype.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Return the dtype of the tensor input as a standardized string. — op_dtype","text":"","code":"op_dtype(x)"},{"path":"https://keras3.posit.co/dev/reference/op_dtype.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Return the dtype of the tensor input as a standardized string. — op_dtype","text":"x tensor. function try access dtype attribute input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_dtype.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Return the dtype of the tensor input as a standardized string. — op_dtype","text":"string indicating dtype input tensor, e.g. \"float32\".","code":""},{"path":"https://keras3.posit.co/dev/reference/op_dtype.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Return the dtype of the tensor input as a standardized string. — op_dtype","text":"","code":"x <- op_zeros(c(8, 12)) op_dtype(x) ## [1] \"float32\""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_eig.html","id":null,"dir":"Reference","previous_headings":"","what":"Computes the eigenvalues and eigenvectors of a square matrix. — op_eig","title":"Computes the eigenvalues and eigenvectors of a square matrix. — op_eig","text":"Computes eigenvalues eigenvectors square matrix.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_eig.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes the eigenvalues and eigenvectors of a square matrix. — op_eig","text":"","code":"op_eig(x)"},{"path":"https://keras3.posit.co/dev/reference/op_eig.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Computes the eigenvalues and eigenvectors of a square matrix. — op_eig","text":"x Input tensor shape (..., M, M).","code":""},{"path":"https://keras3.posit.co/dev/reference/op_eig.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Computes the eigenvalues and eigenvectors of a square matrix. — op_eig","text":"list two tensors: tensor shape (..., M) containing eigenvalues tensor shape (..., M, M) containing eigenvectors.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_eigh.html","id":null,"dir":"Reference","previous_headings":"","what":"Computes the eigenvalues and eigenvectors of a complex Hermitian. — op_eigh","title":"Computes the eigenvalues and eigenvectors of a complex Hermitian. — op_eigh","text":"Computes eigenvalues eigenvectors complex Hermitian.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_eigh.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes the eigenvalues and eigenvectors of a complex Hermitian. — op_eigh","text":"","code":"op_eigh(x)"},{"path":"https://keras3.posit.co/dev/reference/op_eigh.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Computes the eigenvalues and eigenvectors of a complex Hermitian. — op_eigh","text":"x Input tensor shape (..., M, M).","code":""},{"path":"https://keras3.posit.co/dev/reference/op_eigh.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Computes the eigenvalues and eigenvectors of a complex Hermitian. — op_eigh","text":"list two tensors: tensor shape (..., M) containing eigenvalues tensor shape (..., M, M) containing eigenvectors.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_einsum.html","id":null,"dir":"Reference","previous_headings":"","what":"Evaluates the Einstein summation convention on the operands. — op_einsum","title":"Evaluates the Einstein summation convention on the operands. — op_einsum","text":"Evaluates Einstein summation convention operands.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_einsum.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Evaluates the Einstein summation convention on the operands. — op_einsum","text":"","code":"op_einsum(subscripts, ...)"},{"path":"https://keras3.posit.co/dev/reference/op_einsum.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Evaluates the Einstein summation convention on the operands. — op_einsum","text":"subscripts Specifies subscripts summation comma separated list subscript labels. implicit (classical Einstein summation) calculation performed unless explicit indicator -> included well subscript labels precise output form. ... operands compute Einstein sum unnamed arguments. Additional named arguments forwarded underlying backend.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_einsum.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Evaluates the Einstein summation convention on the operands. — op_einsum","text":"calculation based Einstein summation convention.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_einsum.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Evaluates the Einstein summation convention on the operands. — op_einsum","text":"Trace matrix:     Extract diagonal:     Sum axis:     higher dimensional tensors summing single axis can done ellipsis:     Compute matrix transpose reorder number axes:         Matrix vector multiplication:","code":"a <- op_arange(25) |> op_reshape(c(5, 5)) b <- op_arange(5) c <- op_arange(6) |> op_reshape(c(2, 3)) op_einsum(\"ii\", a) op_trace(a) ## tf.Tensor(65.0, shape=(), dtype=float32) ## tf.Tensor(65.0, shape=(), dtype=float32) op_einsum(\"ii -> i\", a) op_diag(a) ## tf.Tensor([ 1.  7. 13. 19. 25.], shape=(5), dtype=float32) ## tf.Tensor([ 1.  7. 13. 19. 25.], shape=(5), dtype=float32) op_einsum(\"ij -> i\", a) op_sum(a, axis = 2) ## tf.Tensor([ 15.  40.  65.  90. 115.], shape=(5), dtype=float32) ## tf.Tensor([ 15.  40.  65.  90. 115.], shape=(5), dtype=float32) op_einsum(\"...j -> ...\", a) op_sum(a, axis = -1) ## tf.Tensor([ 15.  40.  65.  90. 115.], shape=(5), dtype=float32) ## tf.Tensor([ 15.  40.  65.  90. 115.], shape=(5), dtype=float32) op_einsum(\"ji\", c) # return c unchanged ## tf.Tensor( ## [[1. 2. 3.] ##  [4. 5. 6.]], shape=(2, 3), dtype=float32) op_einsum(\"ij -> ji\", c) # transpose op_transpose(c)          # same as above ## tf.Tensor( ## [[1. 4.] ##  [2. 5.] ##  [3. 6.]], shape=(3, 2), dtype=float32) ## tf.Tensor( ## [[1. 4.] ##  [2. 5.] ##  [3. 6.]], shape=(3, 2), dtype=float32) op_einsum(\"ij, j\", a, b) op_einsum(\"...j, j\", a, b) a %*% b op_matmul(a, b) ## tf.Tensor([ 55. 130. 205. 280. 355.], shape=(5), dtype=float32) ## tf.Tensor([ 55. 130. 205. 280. 355.], shape=(5), dtype=float32) ## tf.Tensor([ 55. 130. 205. 280. 355.], shape=(5), dtype=float32) ## tf.Tensor([ 55. 130. 205. 280. 355.], shape=(5), dtype=float32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_elu.html","id":null,"dir":"Reference","previous_headings":"","what":"Exponential Linear Unit activation function. — op_elu","title":"Exponential Linear Unit activation function. — op_elu","text":"defined : f(x) =  alpha * (exp(x) - 1.) x < 0, f(x) = x x >= 0.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_elu.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Exponential Linear Unit activation function. — op_elu","text":"","code":"op_elu(x, alpha = 1)"},{"path":"https://keras3.posit.co/dev/reference/op_elu.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Exponential Linear Unit activation function. — op_elu","text":"x Input tensor. alpha scalar, slope positive section. Defaults 1.0.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_elu.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Exponential Linear Unit activation function. — op_elu","text":"tensor shape x.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_elu.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Exponential Linear Unit activation function. — op_elu","text":"","code":"x <- op_array(c(-1., 0., 1.)) op_elu(x) ## tf.Tensor([-0.63212055  0.          1.        ], shape=(3), dtype=float32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_empty.html","id":null,"dir":"Reference","previous_headings":"","what":"Return a tensor of given shape and type filled with uninitialized data. — op_empty","title":"Return a tensor of given shape and type filled with uninitialized data. — op_empty","text":"Return tensor given shape type filled uninitialized data.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_empty.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Return a tensor of given shape and type filled with uninitialized data. — op_empty","text":"","code":"op_empty(shape, dtype = NULL)"},{"path":"https://keras3.posit.co/dev/reference/op_empty.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Return a tensor of given shape and type filled with uninitialized data. — op_empty","text":"shape Shape empty tensor. dtype Desired data type empty tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_empty.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Return a tensor of given shape and type filled with uninitialized data. — op_empty","text":"empty tensor.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_equal.html","id":null,"dir":"Reference","previous_headings":"","what":"Returns (x1 == x2) element-wise. — op_equal","title":"Returns (x1 == x2) element-wise. — op_equal","text":"Note function automatically called using R operator == tensor.","code":"(x <- op_arange(4)) ## tf.Tensor([1. 2. 3. 4.], shape=(4), dtype=float32) op_equal(x, 2) ## tf.Tensor([False  True False False], shape=(4), dtype=bool) x == 2 ## tf.Tensor([False  True False False], shape=(4), dtype=bool)"},{"path":"https://keras3.posit.co/dev/reference/op_equal.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Returns (x1 == x2) element-wise. — op_equal","text":"","code":"op_equal(x1, x2)"},{"path":"https://keras3.posit.co/dev/reference/op_equal.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Returns (x1 == x2) element-wise. — op_equal","text":"x1 Tensor compare. x2 Tensor compare.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_equal.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Returns (x1 == x2) element-wise. — op_equal","text":"Output tensor, element-wise comparison x1 x2.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_erf.html","id":null,"dir":"Reference","previous_headings":"","what":"Computes the error function of x, element-wise. — op_erf","title":"Computes the error function of x, element-wise. — op_erf","text":"Computes error function x, element-wise.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_erf.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes the error function of x, element-wise. — op_erf","text":"","code":"op_erf(x)"},{"path":"https://keras3.posit.co/dev/reference/op_erf.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Computes the error function of x, element-wise. — op_erf","text":"x Input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_erf.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Computes the error function of x, element-wise. — op_erf","text":"tensor dtype x.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_erf.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Computes the error function of x, element-wise. — op_erf","text":"","code":"x <- op_array(c(-3, -2, -1, 0, 1)) op_erf(x) ## tf.Tensor([-0.999978   -0.99532235 -0.84270084  0.          0.84270084], shape=(5), dtype=float32) # array([-0.99998 , -0.99532, -0.842701,  0.,  0.842701], dtype=float32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_erfinv.html","id":null,"dir":"Reference","previous_headings":"","what":"Computes the inverse error function of x, element-wise. — op_erfinv","title":"Computes the inverse error function of x, element-wise. — op_erfinv","text":"Computes inverse error function x, element-wise.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_erfinv.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes the inverse error function of x, element-wise. — op_erfinv","text":"","code":"op_erfinv(x)"},{"path":"https://keras3.posit.co/dev/reference/op_erfinv.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Computes the inverse error function of x, element-wise. — op_erfinv","text":"x Input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_erfinv.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Computes the inverse error function of x, element-wise. — op_erfinv","text":"tensor dtype x.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_erfinv.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Computes the inverse error function of x, element-wise. — op_erfinv","text":"","code":"x <- op_array(c(-0.5, -0.2, -0.1, 0.0, 0.3)) op_erfinv(x) ## tf.Tensor([-0.4769363  -0.17914344 -0.088856    0.          0.27246267], shape=(5), dtype=float32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_exp.html","id":null,"dir":"Reference","previous_headings":"","what":"Calculate the exponential of all elements in the input tensor. — op_exp","title":"Calculate the exponential of all elements in the input tensor. — op_exp","text":"Calculate exponential elements input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_exp.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Calculate the exponential of all elements in the input tensor. — op_exp","text":"","code":"op_exp(x)"},{"path":"https://keras3.posit.co/dev/reference/op_exp.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Calculate the exponential of all elements in the input tensor. — op_exp","text":"x Input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_exp.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Calculate the exponential of all elements in the input tensor. — op_exp","text":"Output tensor, element-wise exponential x.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_exp2.html","id":null,"dir":"Reference","previous_headings":"","what":"Calculate the base-2 exponential of all elements in the input tensor. — op_exp2","title":"Calculate the base-2 exponential of all elements in the input tensor. — op_exp2","text":"Calculate base-2 exponential elements input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_exp2.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Calculate the base-2 exponential of all elements in the input tensor. — op_exp2","text":"","code":"op_exp2(x)"},{"path":"https://keras3.posit.co/dev/reference/op_exp2.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Calculate the base-2 exponential of all elements in the input tensor. — op_exp2","text":"x Input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_exp2.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Calculate the base-2 exponential of all elements in the input tensor. — op_exp2","text":"Output tensor, element-wise base-2 exponential x.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_expand_dims.html","id":null,"dir":"Reference","previous_headings":"","what":"Expand the shape of a tensor. — op_expand_dims","title":"Expand the shape of a tensor. — op_expand_dims","text":"Insert new axis axis position expanded tensor shape.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_expand_dims.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Expand the shape of a tensor. — op_expand_dims","text":"","code":"op_expand_dims(x, axis)"},{"path":"https://keras3.posit.co/dev/reference/op_expand_dims.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Expand the shape of a tensor. — op_expand_dims","text":"x Input tensor. axis Position expanded axes new axis (axes) placed.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_expand_dims.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Expand the shape of a tensor. — op_expand_dims","text":"Output tensor number dimensions increased.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_expm1.html","id":null,"dir":"Reference","previous_headings":"","what":"Calculate exp(x) - 1 for all elements in the tensor. — op_expm1","title":"Calculate exp(x) - 1 for all elements in the tensor. — op_expm1","text":"Calculate exp(x) - 1 elements tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_expm1.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Calculate exp(x) - 1 for all elements in the tensor. — op_expm1","text":"","code":"op_expm1(x)"},{"path":"https://keras3.posit.co/dev/reference/op_expm1.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Calculate exp(x) - 1 for all elements in the tensor. — op_expm1","text":"x Input values.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_expm1.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Calculate exp(x) - 1 for all elements in the tensor. — op_expm1","text":"Output tensor, element-wise exponential minus one.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_extract_sequences.html","id":null,"dir":"Reference","previous_headings":"","what":"Expands the dimension of last axis into sequences of sequence_length. — op_extract_sequences","title":"Expands the dimension of last axis into sequences of sequence_length. — op_extract_sequences","text":"Slides window size sequence_length last axis input stride sequence_stride, replacing last axis [num_sequences, sequence_length] sequences. dimension along last axis N, number sequences can computed : num_sequences = 1 + (N - sequence_length) // sequence_stride","code":""},{"path":"https://keras3.posit.co/dev/reference/op_extract_sequences.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Expands the dimension of last axis into sequences of sequence_length. — op_extract_sequences","text":"","code":"op_extract_sequences(x, sequence_length, sequence_stride)"},{"path":"https://keras3.posit.co/dev/reference/op_extract_sequences.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Expands the dimension of last axis into sequences of sequence_length. — op_extract_sequences","text":"x Input tensor. sequence_length integer representing sequences length. sequence_stride integer representing sequences hop size.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_extract_sequences.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Expands the dimension of last axis into sequences of sequence_length. — op_extract_sequences","text":"tensor sequences shape [..., num_sequences, sequence_length].","code":""},{"path":"https://keras3.posit.co/dev/reference/op_extract_sequences.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Expands the dimension of last axis into sequences of sequence_length. — op_extract_sequences","text":"","code":"x <- op_convert_to_tensor(1:6) op_extract_sequences(x, 3, 2) ## tf.Tensor( ## [[1 2 3] ##  [3 4 5]], shape=(2, 3), dtype=int32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_eye.html","id":null,"dir":"Reference","previous_headings":"","what":"Return a 2-D tensor with ones on the diagonal and zeros elsewhere. — op_eye","title":"Return a 2-D tensor with ones on the diagonal and zeros elsewhere. — op_eye","text":"Return 2-D tensor ones diagonal zeros elsewhere.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_eye.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Return a 2-D tensor with ones on the diagonal and zeros elsewhere. — op_eye","text":"","code":"op_eye(N, M = NULL, k = 0L, dtype = NULL)"},{"path":"https://keras3.posit.co/dev/reference/op_eye.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Return a 2-D tensor with ones on the diagonal and zeros elsewhere. — op_eye","text":"N Number rows output. M Number columns output. NULL, defaults N. k Index diagonal: 0 (default) refers main diagonal, positive value refers upper diagonal, negative value lower diagonal. dtype Data type returned tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_eye.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Return a 2-D tensor with ones on the diagonal and zeros elsewhere. — op_eye","text":"Tensor ones k-th diagonal zeros elsewhere.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_fft.html","id":null,"dir":"Reference","previous_headings":"","what":"Computes the Fast Fourier Transform along last axis of input. — op_fft","title":"Computes the Fast Fourier Transform along last axis of input. — op_fft","text":"Computes Fast Fourier Transform along last axis input.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_fft.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes the Fast Fourier Transform along last axis of input. — op_fft","text":"","code":"op_fft(x)"},{"path":"https://keras3.posit.co/dev/reference/op_fft.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Computes the Fast Fourier Transform along last axis of input. — op_fft","text":"x list real imaginary parts input tensor. tensors provided floating type.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_fft.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Computes the Fast Fourier Transform along last axis of input. — op_fft","text":"list containing two tensors - real imaginary parts output tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_fft.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Computes the Fast Fourier Transform along last axis of input. — op_fft","text":"","code":"x = c(op_array(c(1., 2.)),       op_array(c(0., 1.))) op_fft(x) ## [[1]] ## tf.Tensor([ 3. -1.], shape=(2), dtype=float32) ## ## [[2]] ## tf.Tensor([ 1. -1.], shape=(2), dtype=float32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_fft2.html","id":null,"dir":"Reference","previous_headings":"","what":"Computes the 2D Fast Fourier Transform along the last two axes of input. — op_fft2","title":"Computes the 2D Fast Fourier Transform along the last two axes of input. — op_fft2","text":"Computes 2D Fast Fourier Transform along last two axes input.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_fft2.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes the 2D Fast Fourier Transform along the last two axes of input. — op_fft2","text":"","code":"op_fft2(x)"},{"path":"https://keras3.posit.co/dev/reference/op_fft2.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Computes the 2D Fast Fourier Transform along the last two axes of input. — op_fft2","text":"x list real imaginary parts input tensor. tensors provided floating type.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_fft2.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Computes the 2D Fast Fourier Transform along the last two axes of input. — op_fft2","text":"list containing two tensors - real imaginary parts output.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_fft2.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Computes the 2D Fast Fourier Transform along the last two axes of input. — op_fft2","text":"","code":"x <- c(op_array(rbind(c(1, 2),                      c(2, 1))),        op_array(rbind(c(0, 1),                      c(1, 0)))) op_fft2(x) ## [[1]] ## tf.Tensor( ## [[ 6.  0.] ##  [ 0. -2.]], shape=(2, 2), dtype=float64) ## ## [[2]] ## tf.Tensor( ## [[ 2.  0.] ##  [ 0. -2.]], shape=(2, 2), dtype=float64)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_flip.html","id":null,"dir":"Reference","previous_headings":"","what":"Reverse the order of elements in the tensor along the given axis. — op_flip","title":"Reverse the order of elements in the tensor along the given axis. — op_flip","text":"shape tensor preserved, elements reordered.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_flip.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Reverse the order of elements in the tensor along the given axis. — op_flip","text":"","code":"op_flip(x, axis = NULL)"},{"path":"https://keras3.posit.co/dev/reference/op_flip.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Reverse the order of elements in the tensor along the given axis. — op_flip","text":"x Input tensor. axis Axis axes along flip tensor. default, axis = NULL, flip axes input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_flip.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Reverse the order of elements in the tensor along the given axis. — op_flip","text":"Output tensor entries axis reversed.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_floor.html","id":null,"dir":"Reference","previous_headings":"","what":"Return the floor of the input, element-wise. — op_floor","title":"Return the floor of the input, element-wise. — op_floor","text":"floor scalar x largest integer , <= x.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_floor.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Return the floor of the input, element-wise. — op_floor","text":"","code":"op_floor(x)"},{"path":"https://keras3.posit.co/dev/reference/op_floor.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Return the floor of the input, element-wise. — op_floor","text":"x Input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_floor.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Return the floor of the input, element-wise. — op_floor","text":"Output tensor, element-wise floor x.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_floor_divide.html","id":null,"dir":"Reference","previous_headings":"","what":"Returns the largest integer smaller or equal to the division of inputs. — op_floor_divide","title":"Returns the largest integer smaller or equal to the division of inputs. — op_floor_divide","text":"Note function automatically called using R operator %/% tensor.","code":"(x <- op_arange(10)) ## tf.Tensor([ 1.  2.  3.  4.  5.  6.  7.  8.  9. 10.], shape=(10), dtype=float32) op_floor_divide(x, 2) ## tf.Tensor([0. 1. 1. 2. 2. 3. 3. 4. 4. 5.], shape=(10), dtype=float32) x %/% 2 ## tf.Tensor([0. 1. 1. 2. 2. 3. 3. 4. 4. 5.], shape=(10), dtype=float32)"},{"path":"https://keras3.posit.co/dev/reference/op_floor_divide.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Returns the largest integer smaller or equal to the division of inputs. — op_floor_divide","text":"","code":"op_floor_divide(x1, x2)"},{"path":"https://keras3.posit.co/dev/reference/op_floor_divide.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Returns the largest integer smaller or equal to the division of inputs. — op_floor_divide","text":"x1 Numerator. x2 Denominator.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_floor_divide.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Returns the largest integer smaller or equal to the division of inputs. — op_floor_divide","text":"Output tensor, y <- floor(x1/x2)","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_fori_loop.html","id":null,"dir":"Reference","previous_headings":"","what":"For loop implementation. — op_fori_loop","title":"For loop implementation. — op_fori_loop","text":"loop implementation.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_fori_loop.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"For loop implementation. — op_fori_loop","text":"","code":"op_fori_loop(lower, upper, body_fun, init_val)"},{"path":"https://keras3.posit.co/dev/reference/op_fori_loop.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"For loop implementation. — op_fori_loop","text":"lower initial value loop variable. upper upper bound loop variable. body_fun callable represents loop body. Must take two arguments: loop variable loop state. loop state updated returned function. init_val initial value loop state.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_fori_loop.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"For loop implementation. — op_fori_loop","text":"final state loop.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_fori_loop.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"For loop implementation. — op_fori_loop","text":"","code":"lower <- 0L upper <- 10L body_fun <- function(i, state) state + i init_state <- 0L final_state <- op_fori_loop(lower, upper, body_fun, init_state) final_state ## tf.Tensor(45, shape=(), dtype=int32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_full.html","id":null,"dir":"Reference","previous_headings":"","what":"Return a new tensor of given shape and type, filled with fill_value. — op_full","title":"Return a new tensor of given shape and type, filled with fill_value. — op_full","text":"Return new tensor given shape type, filled fill_value.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_full.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Return a new tensor of given shape and type, filled with fill_value. — op_full","text":"","code":"op_full(shape, fill_value, dtype = NULL)"},{"path":"https://keras3.posit.co/dev/reference/op_full.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Return a new tensor of given shape and type, filled with fill_value. — op_full","text":"shape Shape new tensor. fill_value Fill value. dtype Desired data type tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_full.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Return a new tensor of given shape and type, filled with fill_value. — op_full","text":"Output tensor.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_full_like.html","id":null,"dir":"Reference","previous_headings":"","what":"Return a full tensor with the same shape and type as the given tensor. — op_full_like","title":"Return a full tensor with the same shape and type as the given tensor. — op_full_like","text":"Return full tensor shape type given tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_full_like.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Return a full tensor with the same shape and type as the given tensor. — op_full_like","text":"","code":"op_full_like(x, fill_value, dtype = NULL)"},{"path":"https://keras3.posit.co/dev/reference/op_full_like.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Return a full tensor with the same shape and type as the given tensor. — op_full_like","text":"x Input tensor. fill_value Fill value. dtype Overrides data type result.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_full_like.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Return a full tensor with the same shape and type as the given tensor. — op_full_like","text":"Tensor fill_value shape type x.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_gelu.html","id":null,"dir":"Reference","previous_headings":"","what":"Gaussian Error Linear Unit (GELU) activation function. — op_gelu","title":"Gaussian Error Linear Unit (GELU) activation function. — op_gelu","text":"approximate TRUE, defined : f(x) = 0.5 * x * (1 + tanh(sqrt(2 / pi) * (x + 0.044715 * x^3))) approximate FALSE, defined : f(x) = x * P(X <= x) = 0.5 * x * (1 + erf(x / sqrt(2))), P(X) ~ N(0, 1).","code":""},{"path":"https://keras3.posit.co/dev/reference/op_gelu.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Gaussian Error Linear Unit (GELU) activation function. — op_gelu","text":"","code":"op_gelu(x, approximate = TRUE)"},{"path":"https://keras3.posit.co/dev/reference/op_gelu.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Gaussian Error Linear Unit (GELU) activation function. — op_gelu","text":"x Input tensor. approximate Approximate version GELU activation. Defaults TRUE.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_gelu.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Gaussian Error Linear Unit (GELU) activation function. — op_gelu","text":"tensor shape x.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_gelu.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Gaussian Error Linear Unit (GELU) activation function. — op_gelu","text":"","code":"x <- op_array(c(-1., 0., 1.)) op_gelu(x) ## tf.Tensor([-0.158808  0.        0.841192], shape=(3), dtype=float32) op_gelu(x, FALSE) ## tf.Tensor([-0.15865526  0.          0.8413447 ], shape=(3), dtype=float32) x <- seq(-5, 5, .1) plot(x, op_gelu(x),      type = \"l\", #, frame.plot = FALSE,      panel.first = grid())"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_get_item.html","id":null,"dir":"Reference","previous_headings":"","what":"Return x[key]. — op_get_item","title":"Return x[key]. — op_get_item","text":"Return x[key].","code":""},{"path":"https://keras3.posit.co/dev/reference/op_get_item.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Return x[key]. — op_get_item","text":"","code":"op_get_item(x, key)"},{"path":"https://keras3.posit.co/dev/reference/op_get_item.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Return x[key]. — op_get_item","text":"x dictionary-like object key Generally, string, object __hash__ method acceptable.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_get_item.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Return x[key]. — op_get_item","text":"key.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_get_item.html","id":"note","dir":"Reference","previous_headings":"","what":"Note","title":"Return x[key]. — op_get_item","text":"Generally, calling x[[key]] x$key preferable.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_glu.html","id":null,"dir":"Reference","previous_headings":"","what":"Gated Linear Unit (GLU) activation function. — op_glu","title":"Gated Linear Unit (GLU) activation function. — op_glu","text":"defined : f(x) = * sigmoid(b) x split b along given axis.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_glu.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Gated Linear Unit (GLU) activation function. — op_glu","text":"","code":"op_glu(x, axis = -1L)"},{"path":"https://keras3.posit.co/dev/reference/op_glu.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Gated Linear Unit (GLU) activation function. — op_glu","text":"x Input tensor. axis axis along split input tensor. Defaults -1.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_glu.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Gated Linear Unit (GLU) activation function. — op_glu","text":"tensor shape half input.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_glu.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Gated Linear Unit (GLU) activation function. — op_glu","text":"","code":"x <- op_array(c(-1., 0., 1. , 1.)) op_glu(x) ## tf.Tensor([-0.7310586  0.       ], shape=(2), dtype=float32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_greater.html","id":null,"dir":"Reference","previous_headings":"","what":"Return the truth value of x1 > x2 element-wise. — op_greater","title":"Return the truth value of x1 > x2 element-wise. — op_greater","text":"Note function automatically called using R operator > tensor.","code":"(x <- op_arange(4)) ## tf.Tensor([1. 2. 3. 4.], shape=(4), dtype=float32) op_greater(x, 2) ## tf.Tensor([False False  True  True], shape=(4), dtype=bool) x > 2 ## tf.Tensor([False False  True  True], shape=(4), dtype=bool)"},{"path":"https://keras3.posit.co/dev/reference/op_greater.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Return the truth value of x1 > x2 element-wise. — op_greater","text":"","code":"op_greater(x1, x2)"},{"path":"https://keras3.posit.co/dev/reference/op_greater.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Return the truth value of x1 > x2 element-wise. — op_greater","text":"x1 First input tensor. x2 Second input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_greater.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Return the truth value of x1 > x2 element-wise. — op_greater","text":"Output tensor, element-wise comparison x1 x2.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_greater_equal.html","id":null,"dir":"Reference","previous_headings":"","what":"Return the truth value of x1 >= x2 element-wise. — op_greater_equal","title":"Return the truth value of x1 >= x2 element-wise. — op_greater_equal","text":"Note function automatically called using R operator >= tensor.","code":"(x <- op_arange(4)) ## tf.Tensor([1. 2. 3. 4.], shape=(4), dtype=float32) op_greater_equal(x, 2) ## tf.Tensor([False  True  True  True], shape=(4), dtype=bool) x >= 2 ## tf.Tensor([False  True  True  True], shape=(4), dtype=bool)"},{"path":"https://keras3.posit.co/dev/reference/op_greater_equal.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Return the truth value of x1 >= x2 element-wise. — op_greater_equal","text":"","code":"op_greater_equal(x1, x2)"},{"path":"https://keras3.posit.co/dev/reference/op_greater_equal.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Return the truth value of x1 >= x2 element-wise. — op_greater_equal","text":"x1 First input tensor. x2 Second input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_greater_equal.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Return the truth value of x1 >= x2 element-wise. — op_greater_equal","text":"Output tensor, element-wise comparison x1 x2.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_hamming.html","id":null,"dir":"Reference","previous_headings":"","what":"Hamming window function. — op_hamming","title":"Hamming window function. — op_hamming","text":"Hamming window defined : w[n] = 0.54 - 0.46 * cos(2 * pi * n / (N - 1)) 0 <= n <= N - 1.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_hamming.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Hamming window function. — op_hamming","text":"","code":"op_hamming(x)"},{"path":"https://keras3.posit.co/dev/reference/op_hamming.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Hamming window function. — op_hamming","text":"x Length window. Must positive integer.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_hamming.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Hamming window function. — op_hamming","text":"1D tensor containing window values.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_hamming.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Hamming window function. — op_hamming","text":"","code":"op_hamming(5) ## tf.Tensor([0.08000001 0.54       1.         0.54       0.08000001], shape=(5), dtype=float32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_hanning.html","id":null,"dir":"Reference","previous_headings":"","what":"Hanning window function. — op_hanning","title":"Hanning window function. — op_hanning","text":"Hanning window defined : w[n] = 0.5 - 0.5 * cos(2 * pi * n / (N - 1)) 0 <= n <= N - 1.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_hanning.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Hanning window function. — op_hanning","text":"","code":"op_hanning(x)"},{"path":"https://keras3.posit.co/dev/reference/op_hanning.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Hanning window function. — op_hanning","text":"x Length window. Must positive integer.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_hanning.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Hanning window function. — op_hanning","text":"1D tensor containing window values.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_hanning.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Hanning window function. — op_hanning","text":"","code":"op_hanning(5) ## tf.Tensor([0.  0.5 1.  0.5 0. ], shape=(5), dtype=float32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_hard_shrink.html","id":null,"dir":"Reference","previous_headings":"","what":"Hard Shrink activation function. — op_hard_shrink","title":"Hard Shrink activation function. — op_hard_shrink","text":"Hard Shrink function thresholding operation defined : f(x) = x |x| > threshold, f(x) = 0 otherwise.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_hard_shrink.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Hard Shrink activation function. — op_hard_shrink","text":"","code":"op_hard_shrink(x, threshold = 0.5)"},{"path":"https://keras3.posit.co/dev/reference/op_hard_shrink.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Hard Shrink activation function. — op_hard_shrink","text":"x Input tensor. threshold Threshold value. Defaults 0.5.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_hard_shrink.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Hard Shrink activation function. — op_hard_shrink","text":"tensor shape x.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_hard_shrink.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Hard Shrink activation function. — op_hard_shrink","text":"","code":"x <- op_array(c(-0.5, 0., 1.)) op_hard_shrink(x) ## tf.Tensor([0. 0. 1.], shape=(3), dtype=float32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_hard_sigmoid.html","id":null,"dir":"Reference","previous_headings":"","what":"Hard sigmoid activation function. — op_hard_sigmoid","title":"Hard sigmoid activation function. — op_hard_sigmoid","text":"defined : 0 x < -2.5, 1 x > 2.5, (0.2 * x) + 0.5 -2.5 <= x <= 2.5.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_hard_sigmoid.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Hard sigmoid activation function. — op_hard_sigmoid","text":"","code":"op_hard_sigmoid(x)"},{"path":"https://keras3.posit.co/dev/reference/op_hard_sigmoid.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Hard sigmoid activation function. — op_hard_sigmoid","text":"x Input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_hard_sigmoid.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Hard sigmoid activation function. — op_hard_sigmoid","text":"tensor shape x.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_hard_sigmoid.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Hard sigmoid activation function. — op_hard_sigmoid","text":"","code":"x <- op_array(c(-1., 0., 1.)) op_hard_sigmoid(x) ## tf.Tensor([0.33333334 0.5        0.6666667 ], shape=(3), dtype=float32) x <- as.array(seq(-5, 5, .1)) plot(x, op_hard_sigmoid(x),      type = 'l', panel.first = grid(), frame.plot = FALSE)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_hard_silu.html","id":null,"dir":"Reference","previous_headings":"","what":"Hard SiLU activation function, also known as Hard Swish. — op_hard_silu","title":"Hard SiLU activation function, also known as Hard Swish. — op_hard_silu","text":"defined : 0 x < -3 x x > 3 x * (x + 3) / 6 -3 <= x <= 3 faster, piecewise linear approximation silu activation.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_hard_silu.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Hard SiLU activation function, also known as Hard Swish. — op_hard_silu","text":"","code":"op_hard_silu(x)  op_hard_swish(x)"},{"path":"https://keras3.posit.co/dev/reference/op_hard_silu.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Hard SiLU activation function, also known as Hard Swish. — op_hard_silu","text":"x Input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_hard_silu.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Hard SiLU activation function, also known as Hard Swish. — op_hard_silu","text":"tensor shape x.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_hard_silu.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Hard SiLU activation function, also known as Hard Swish. — op_hard_silu","text":"","code":"x <- op_convert_to_tensor(c(-3.0, -1.0, 0.0, 1.0, 3.0)) op_hard_silu(x) ## tf.Tensor([-0.         -0.33333333  0.          0.66666667  3.        ], shape=(5), dtype=float64)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_hard_tanh.html","id":null,"dir":"Reference","previous_headings":"","what":"Applies the HardTanh function element-wise. — op_hard_tanh","title":"Applies the HardTanh function element-wise. — op_hard_tanh","text":"defined : f(x) = -1 x < -1, f(x) = x -1 <= x <= 1, f(x) = 1 x > 1.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_hard_tanh.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Applies the HardTanh function element-wise. — op_hard_tanh","text":"","code":"op_hard_tanh(x)"},{"path":"https://keras3.posit.co/dev/reference/op_hard_tanh.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Applies the HardTanh function element-wise. — op_hard_tanh","text":"x Input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_hard_tanh.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Applies the HardTanh function element-wise. — op_hard_tanh","text":"Output tensor shape x values clamped -1 1.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_hard_tanh.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Applies the HardTanh function element-wise. — op_hard_tanh","text":"","code":"x <- op_array(c(-2., -1., 0., 1., 2.)) op_hard_tanh(x) ## tf.Tensor([-1. -1.  0.  1.  1.], shape=(5), dtype=float32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_heaviside.html","id":null,"dir":"Reference","previous_headings":"","what":"Heaviside step function. — op_heaviside","title":"Heaviside step function. — op_heaviside","text":"Heaviside step function defined : heaviside(x1, x2) = 0 x1 < 0, heaviside(x1, x2) = 1 x1 > 0, heaviside(x1, x2) = x2 x1 == 0.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_heaviside.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Heaviside step function. — op_heaviside","text":"","code":"op_heaviside(x1, x2)"},{"path":"https://keras3.posit.co/dev/reference/op_heaviside.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Heaviside step function. — op_heaviside","text":"x1 Tensor input. x2 Value use x1 == 0.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_heaviside.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Heaviside step function. — op_heaviside","text":"tensor broadcast x1 x2 containing 0, 1, x2.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_heaviside.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Heaviside step function. — op_heaviside","text":"","code":"x1 <- op_array(c(-2, 0, 3)) op_heaviside(x1, 0.5) ## tf.Tensor([0.  0.5 1. ], shape=(3), dtype=float32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_histogram.html","id":null,"dir":"Reference","previous_headings":"","what":"Computes a histogram of the data tensor x. — op_histogram","title":"Computes a histogram of the data tensor x. — op_histogram","text":"Computes histogram data tensor x.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_histogram.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes a histogram of the data tensor x. — op_histogram","text":"","code":"op_histogram(x, bins = 10L, range = NULL)"},{"path":"https://keras3.posit.co/dev/reference/op_histogram.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Computes a histogram of the data tensor x. — op_histogram","text":"x Input tensor. bins integer representing number histogram bins. Defaults 10. range pair numbers representing lower upper range bins. specified, use min max x.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_histogram.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Computes a histogram of the data tensor x. — op_histogram","text":"list two tensors containing: tensor representing counts elements bin. tensor representing bin edges.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_histogram.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Computes a histogram of the data tensor x. — op_histogram","text":"","code":"input_tensor <- random_uniform(8) c(counts, edges) %<-% op_histogram(input_tensor)  counts #> tf.Tensor([2. 1. 1. 1. 0. 1. 0. 0. 0. 2.], shape=(10), dtype=float32) edges #> tf.Tensor( #> [0.01647806 0.10867032 0.20086257 0.29305482 0.38524708 0.47743934 #>  0.5696316  0.66182387 0.7540161  0.84620833 0.9384006 ], shape=(11), dtype=float32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_hstack.html","id":null,"dir":"Reference","previous_headings":"","what":"Stack tensors in sequence horizontally (column wise). — op_hstack","title":"Stack tensors in sequence horizontally (column wise). — op_hstack","text":"equivalent concatenation along first axis 1-D tensors, along second axis tensors.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_hstack.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Stack tensors in sequence horizontally (column wise). — op_hstack","text":"","code":"op_hstack(xs, ...)"},{"path":"https://keras3.posit.co/dev/reference/op_hstack.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Stack tensors in sequence horizontally (column wise). — op_hstack","text":"xs, ... list tensors.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_hstack.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Stack tensors in sequence horizontally (column wise). — op_hstack","text":"tensor formed stacking given tensors.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_identity.html","id":null,"dir":"Reference","previous_headings":"","what":"Return the identity tensor. — op_identity","title":"Return the identity tensor. — op_identity","text":"identity tensor square tensor ones main diagonal zeros elsewhere.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_identity.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Return the identity tensor. — op_identity","text":"","code":"op_identity(n, dtype = NULL)"},{"path":"https://keras3.posit.co/dev/reference/op_identity.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Return the identity tensor. — op_identity","text":"n Number rows (columns) n x n output tensor. dtype Data type output tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_identity.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Return the identity tensor. — op_identity","text":"identity tensor.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_ifft2.html","id":null,"dir":"Reference","previous_headings":"","what":"Computes the 2D Inverse Fast Fourier Transform along the last two axes of — op_ifft2","title":"Computes the 2D Inverse Fast Fourier Transform along the last two axes of — op_ifft2","text":"input.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_ifft2.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes the 2D Inverse Fast Fourier Transform along the last two axes of — op_ifft2","text":"","code":"op_ifft2(x)"},{"path":"https://keras3.posit.co/dev/reference/op_ifft2.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Computes the 2D Inverse Fast Fourier Transform along the last two axes of — op_ifft2","text":"x Tuple real imaginary parts input tensor. tensors tuple floating type.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_ifft2.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Computes the 2D Inverse Fast Fourier Transform along the last two axes of — op_ifft2","text":"tuple containing two tensors - real imaginary parts output.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_ifft2.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Computes the 2D Inverse Fast Fourier Transform along the last two axes of — op_ifft2","text":"","code":"x <- list(op_array(rbind(c(1, 2), c(2, 1))),           op_array(rbind(c(0, 1), c(1, 0)))) op_ifft2(x) ## [[1]] ## tf.Tensor( ## [[ 1.5  0. ] ##  [ 0.  -0.5]], shape=(2, 2), dtype=float64) ## ## [[2]] ## tf.Tensor( ## [[ 0.5 -0. ] ##  [-0.  -0.5]], shape=(2, 2), dtype=float64)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_imag.html","id":null,"dir":"Reference","previous_headings":"","what":"Return the imaginary part of the complex argument. — op_imag","title":"Return the imaginary part of the complex argument. — op_imag","text":"Return imaginary part complex argument.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_imag.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Return the imaginary part of the complex argument. — op_imag","text":"","code":"op_imag(x)"},{"path":"https://keras3.posit.co/dev/reference/op_imag.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Return the imaginary part of the complex argument. — op_imag","text":"x Input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_imag.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Return the imaginary part of the complex argument. — op_imag","text":"imaginary component complex argument.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_image_affine_transform.html","id":null,"dir":"Reference","previous_headings":"","what":"Applies the given transform(s) to the image(s). — op_image_affine_transform","title":"Applies the given transform(s) to the image(s). — op_image_affine_transform","text":"Applies given transform(s) image(s).","code":""},{"path":"https://keras3.posit.co/dev/reference/op_image_affine_transform.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Applies the given transform(s) to the image(s). — op_image_affine_transform","text":"","code":"op_image_affine_transform(   images,   transform,   interpolation = \"bilinear\",   fill_mode = \"constant\",   fill_value = 0L,   data_format = NULL )"},{"path":"https://keras3.posit.co/dev/reference/op_image_affine_transform.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Applies the given transform(s) to the image(s). — op_image_affine_transform","text":"images Input image batch images. Must 3D 4D. transform Projective transform matrix/matrices. vector length 8 tensor size N x 8. one row transform [a0, a1, a2, b0, b1, b2, c0, c1], maps output point (x, y) transformed input point (x', y') = ((a0 x + a1 y + a2) / k, (b0 x + b1 y + b2) / k), k = c0 x + c1 y + 1. transform inverted compared transform mapping input points output points. Note gradients backpropagated transformation parameters. Note c0 c1 effective using TensorFlow backend considered 0 using backends. interpolation Interpolation method. Available methods \"nearest\", \"bilinear\". Defaults \"bilinear\". fill_mode Points outside boundaries input filled according given mode. Available methods \"constant\", \"nearest\", \"wrap\" \"reflect\". Defaults \"constant\". \"reflect\": (d c b | b c d | d c b ) input extended reflecting edge last pixel. \"constant\": (k k k k | b c d | k k k k) input extended filling values beyond edge constant value k specified fill_value. \"wrap\": (b c d | b c d | b c d) input extended wrapping around opposite edge. \"nearest\": (| b c d | d d d d) input extended nearest pixel. fill_value Value used points outside boundaries input fill_mode = \"constant\". Defaults 0. data_format string specifying data format input tensor. can either \"channels_last\" \"channels_first\". \"channels_last\" corresponds inputs shape (batch, height, width, channels), \"channels_first\" corresponds inputs shape (batch, channels, height, width). specified, value default config_image_data_format().","code":""},{"path":"https://keras3.posit.co/dev/reference/op_image_affine_transform.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Applies the given transform(s) to the image(s). — op_image_affine_transform","text":"Applied affine transform image batch images.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_image_affine_transform.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Applies the given transform(s) to the image(s). — op_image_affine_transform","text":"","code":"x <- random_uniform(c(2, 64, 80, 3)) # batch of 2 RGB images transform <- op_array(rbind(c(1.5, 0, -20, 0, 1.5, -16, 0, 0),  # zoom                            c(1, 0, -20, 0, 1, -16, 0, 0)))  # translation)) y <- op_image_affine_transform(x, transform) shape(y) ## shape(2, 64, 80, 3) # (2, 64, 80, 3) x <- random_uniform(c(64, 80, 3)) # single RGB image transform <- op_array(c(1.0, 0.5, -20, 0.5, 1.0, -16, 0, 0))  # shear y <- op_image_affine_transform(x, transform) shape(y) ## shape(64, 80, 3) # (64, 80, 3) x <- random_uniform(c(2, 3, 64, 80)) # batch of 2 RGB images transform <- op_array(rbind(   c(1.5, 0,-20, 0, 1.5,-16, 0, 0),  # zoom   c(1, 0,-20, 0, 1,-16, 0, 0)  # translation )) y <- op_image_affine_transform(x, transform, data_format = \"channels_first\") shape(y) ## shape(2, 3, 64, 80) # (2, 3, 64, 80)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_image_crop.html","id":null,"dir":"Reference","previous_headings":"","what":"Crop images to a specified height and width. — op_image_crop","title":"Crop images to a specified height and width. — op_image_crop","text":"Crop images specified height width.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_image_crop.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Crop images to a specified height and width. — op_image_crop","text":"","code":"op_image_crop(   images,   top_cropping = NULL,   left_cropping = NULL,   bottom_cropping = NULL,   right_cropping = NULL,   target_height = NULL,   target_width = NULL,   data_format = NULL )"},{"path":"https://keras3.posit.co/dev/reference/op_image_crop.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Crop images to a specified height and width. — op_image_crop","text":"images Input image batch images. Must 3D 4D. top_cropping Number columns crop top. left_cropping Number columns crop left. bottom_cropping Number columns crop bottom. right_cropping Number columns crop right. target_height Height output images. target_width Width output images. data_format string specifying data format input tensor. can either \"channels_last\" \"channels_first\". \"channels_last\" corresponds inputs shape (batch, height, width, channels), \"channels_first\" corresponds inputs shape (batch, channels, height, width). specified, value default config_image_data_format().","code":""},{"path":"https://keras3.posit.co/dev/reference/op_image_crop.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Crop images to a specified height and width. — op_image_crop","text":"Cropped image batch images.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_image_crop.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Crop images to a specified height and width. — op_image_crop","text":"","code":"images <- op_reshape(op_arange(1, 28, dtype=\"float32\"), c(3, 3, 3)) images[, , 1] # print the first channel of the images  cropped_images <- op_image_crop(images, 0, 0, 2, 2) cropped_images[, , 1] # print the first channel of the cropped images"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_image_elastic_transform.html","id":null,"dir":"Reference","previous_headings":"","what":"Applies elastic deformation to the image(s). — op_image_elastic_transform","title":"Applies elastic deformation to the image(s). — op_image_elastic_transform","text":"Apply random elastic deformation 3D 4D image tensors.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_image_elastic_transform.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Applies elastic deformation to the image(s). — op_image_elastic_transform","text":"","code":"op_image_elastic_transform(   images,   alpha = 20,   sigma = 5,   interpolation = \"bilinear\",   fill_mode = \"reflect\",   fill_value = 0,   seed = NULL,   data_format = NULL )"},{"path":"https://keras3.posit.co/dev/reference/op_image_elastic_transform.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Applies elastic deformation to the image(s). — op_image_elastic_transform","text":"images Input image batch images. Must 3D 4D. alpha Scaling factor controls intensity deformation. sigma Standard deviation Gaussian filter used smoothing displacement fields. interpolation Interpolation method. Available methods \"nearest\", \"bilinear\". Defaults \"bilinear\". fill_mode Points outside boundaries input filled according given mode. Available methods \"constant\", \"nearest\", \"wrap\" \"reflect\". Defaults \"reflect\". \"reflect\": (d c b | b c d | d c b ) input extended reflecting edge last pixel. \"constant\": (k k k k | b c d | k k k k) input extended filling values beyond edge constant value k specified fill_value. \"wrap\": (b c d | b c d | b c d) input extended wrapping around opposite edge. \"nearest\": (| b c d | d d d d) input extended nearest pixel. fill_value Value used points outside boundaries input fill_mode=\"constant\". Defaults 0. seed Optional integer seed random number generator. data_format string specifying data format input tensor. can either \"channels_last\" \"channels_first\". \"channels_last\" corresponds inputs shape (batch, height, width, channels), \"channels_first\" corresponds inputs shape (batch, channels, height, width). specified, value default keras.config.image_data_format.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_image_elastic_transform.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Applies elastic deformation to the image(s). — op_image_elastic_transform","text":"Transformed image batch images elastic deformation.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_image_elastic_transform.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Applies elastic deformation to the image(s). — op_image_elastic_transform","text":"","code":"x <- random_uniform(c(2, 64, 80, 3))  # batch of 2 RGB images y <- op_image_elastic_transform(x) op_shape(y) ## shape(2, 64, 80, 3) x <- random_uniform(c(64, 80, 3))  # single RGB image y <- op_image_elastic_transform(x) op_shape(y) ## shape(64, 80, 3) x <- random_uniform(c(2, 3, 64, 80))  # batch of 2 RGB images y <- op_image_elastic_transform(   x,   data_format = \"channels_first\",   seed = 123 ) op_shape(y) ## shape(2, 3, 64, 80)"},{"path":"https://keras3.posit.co/dev/reference/op_image_extract_patches.html","id":null,"dir":"Reference","previous_headings":"","what":"Extracts patches from the image(s). — op_image_extract_patches","title":"Extracts patches from the image(s). — op_image_extract_patches","text":"Extracts patches image(s).","code":""},{"path":"https://keras3.posit.co/dev/reference/op_image_extract_patches.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Extracts patches from the image(s). — op_image_extract_patches","text":"","code":"op_image_extract_patches(   images,   size,   strides = NULL,   dilation_rate = 1L,   padding = \"valid\",   data_format = \"channels_last\" )"},{"path":"https://keras3.posit.co/dev/reference/op_image_extract_patches.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Extracts patches from the image(s). — op_image_extract_patches","text":"images Input image batch images. Must 3D 4D. size Patch size int list (patch_height, patch_width) strides strides along height width. specified, NULL, defaults value size. dilation_rate input stride, specifying far two consecutive patch samples input. value 1, strides must 1. NOTE: strides > 1 supported conjunction dilation_rate > 1 padding type padding algorithm use: \"\" \"valid\". data_format string specifying data format input tensor. can either \"channels_last\" \"channels_first\". \"channels_last\" corresponds inputs shape (batch, height, width, channels), \"channels_first\" corresponds inputs shape (batch, channels, height, width). specified, value default config_image_data_format().","code":""},{"path":"https://keras3.posit.co/dev/reference/op_image_extract_patches.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Extracts patches from the image(s). — op_image_extract_patches","text":"Extracted patches 3D (batched) 4D (batched)","code":""},{"path":"https://keras3.posit.co/dev/reference/op_image_extract_patches.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Extracts patches from the image(s). — op_image_extract_patches","text":"","code":"image <- random_uniform(c(2, 20, 20, 3), dtype = \"float32\") # batch of 2 RGB images patches <- op_image_extract_patches(image, c(5, 5)) shape(patches) ## shape(2, 4, 4, 75) # (2, 4, 4, 75) image <- random_uniform(c(20, 20, 3), dtype = \"float32\") # 1 RGB image patches <- op_image_extract_patches(image, c(3, 3), c(1, 1)) shape(patches) ## shape(18, 18, 27) # (18, 18, 27)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_image_gaussian_blur.html","id":null,"dir":"Reference","previous_headings":"","what":"Applies a Gaussian blur to the image(s). — op_image_gaussian_blur","title":"Applies a Gaussian blur to the image(s). — op_image_gaussian_blur","text":"Applies Gaussian blur image(s).","code":""},{"path":"https://keras3.posit.co/dev/reference/op_image_gaussian_blur.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Applies a Gaussian blur to the image(s). — op_image_gaussian_blur","text":"","code":"op_image_gaussian_blur(   images,   kernel_size = list(3L, 3L),   sigma = list(1, 1),   data_format = NULL )"},{"path":"https://keras3.posit.co/dev/reference/op_image_gaussian_blur.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Applies a Gaussian blur to the image(s). — op_image_gaussian_blur","text":"images Input image batch images. Must 3D 4D. kernel_size tuple two integers, specifying height width Gaussian kernel. sigma tuple two floats, specifying standard deviation Gaussian kernel along height width. data_format string specifying data format input tensor. can either \"channels_last\" \"channels_first\". \"channels_last\" corresponds inputs shape (batch, height, width, channels), \"channels_first\" corresponds inputs shape (batch, channels, height, width). specified, value default keras.config.image_data_format.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_image_gaussian_blur.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Applies a Gaussian blur to the image(s). — op_image_gaussian_blur","text":"Blurred image batch images.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_image_gaussian_blur.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Applies a Gaussian blur to the image(s). — op_image_gaussian_blur","text":"","code":"x <- op_ones(c(2, 64, 80, 3))  # batch of 2 RGB images y <- op_image_gaussian_blur(x) op_shape(y) ## shape(2, 64, 80, 3) x <- op_ones(c(64, 80, 3))  # single RGB image y <- op_image_gaussian_blur(x) op_shape(y) ## shape(64, 80, 3) x <- op_ones(c(2, 3, 64, 80))  # batch of 2 RGB images, channels_first y <- op_image_gaussian_blur(x, data_format = \"channels_first\") op_shape(y) ## shape(2, 3, 64, 80)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_image_hsv_to_rgb.html","id":null,"dir":"Reference","previous_headings":"","what":"Convert HSV images to RGB. — op_image_hsv_to_rgb","title":"Convert HSV images to RGB. — op_image_hsv_to_rgb","text":"images must float dtype, output well defined values images [0, 1].","code":""},{"path":"https://keras3.posit.co/dev/reference/op_image_hsv_to_rgb.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Convert HSV images to RGB. — op_image_hsv_to_rgb","text":"","code":"op_image_hsv_to_rgb(images, data_format = NULL)"},{"path":"https://keras3.posit.co/dev/reference/op_image_hsv_to_rgb.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Convert HSV images to RGB. — op_image_hsv_to_rgb","text":"images Input image batch images. Must 3D 4D. data_format string specifying data format input tensor. can either \"channels_last\" \"channels_first\". \"channels_last\" corresponds inputs shape (batch, height, width, channels), \"channels_first\" corresponds inputs shape (batch, channels, height, width). specified, value default config_image_data_format().","code":""},{"path":"https://keras3.posit.co/dev/reference/op_image_hsv_to_rgb.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Convert HSV images to RGB. — op_image_hsv_to_rgb","text":"RGB image batch RGB images.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_image_hsv_to_rgb.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Convert HSV images to RGB. — op_image_hsv_to_rgb","text":"","code":"x <- random_uniform(c(2, 4, 4, 3)) y <- op_image_hsv_to_rgb(x) shape(y) ## shape(2, 4, 4, 3) x <- random_uniform(c(4, 4, 3)) # Single HSV image y <- op_image_hsv_to_rgb(x) shape(y) ## shape(4, 4, 3) x <- random_uniform(c(2, 3, 4, 4)) y <- op_image_hsv_to_rgb(x, data_format=\"channels_first\") shape(y) ## shape(2, 3, 4, 4)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_image_map_coordinates.html","id":null,"dir":"Reference","previous_headings":"","what":"Map the input array to new coordinates by interpolation. — op_image_map_coordinates","title":"Map the input array to new coordinates by interpolation. — op_image_map_coordinates","text":"Note interpolation near boundaries differs scipy function, fixed outstanding bug scipy/issues/2640.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_image_map_coordinates.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Map the input array to new coordinates by interpolation. — op_image_map_coordinates","text":"","code":"op_image_map_coordinates(   inputs,   coordinates,   order,   fill_mode = \"constant\",   fill_value = 0L )"},{"path":"https://keras3.posit.co/dev/reference/op_image_map_coordinates.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Map the input array to new coordinates by interpolation. — op_image_map_coordinates","text":"inputs input array. coordinates coordinates inputs evaluated. order order spline interpolation. order must 0 1. 0 indicates nearest neighbor 1 indicates linear interpolation. fill_mode Points outside boundaries inputs filled according given mode. Available methods \"constant\", \"nearest\", \"wrap\" \"mirror\" \"reflect\". Defaults \"constant\". \"constant\": (k k k k | b c d | k k k k) inputs extended filling values beyond edge constant value k specified fill_value. \"nearest\": (| b c d | d d d d) inputs extended nearest pixel. \"wrap\": (b c d | b c d | b c d) inputs extended wrapping around opposite edge. \"mirror\": (c d c b | b c d | c b b) inputs extended mirroring edge. \"reflect\": (d c b | b c d | d c b ) inputs extended reflecting edge last pixel. fill_value Value used points outside boundaries inputs fill_mode = \"constant\". Defaults 0.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_image_map_coordinates.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Map the input array to new coordinates by interpolation. — op_image_map_coordinates","text":"Output input batch inputs.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_image_pad.html","id":null,"dir":"Reference","previous_headings":"","what":"Pad images with zeros to the specified height and width. — op_image_pad","title":"Pad images with zeros to the specified height and width. — op_image_pad","text":"Pad images zeros specified height width.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_image_pad.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Pad images with zeros to the specified height and width. — op_image_pad","text":"","code":"op_image_pad(   images,   top_padding = NULL,   left_padding = NULL,   bottom_padding = NULL,   right_padding = NULL,   target_height = NULL,   target_width = NULL,   data_format = NULL )"},{"path":"https://keras3.posit.co/dev/reference/op_image_pad.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Pad images with zeros to the specified height and width. — op_image_pad","text":"images Input image batch images. Must 3D 4D. top_padding Number rows zeros add top. left_padding Number columns zeros add left. bottom_padding Number rows zeros add bottom. right_padding Number columns zeros add right. target_height Height output images. target_width Width output images. data_format string specifying data format input tensor. can either \"channels_last\" \"channels_first\". \"channels_last\" corresponds inputs shape (batch, height, width, channels), \"channels_first\" corresponds inputs shape (batch, channels, height, width). specified, value default config_image_data_format().","code":""},{"path":"https://keras3.posit.co/dev/reference/op_image_pad.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Pad images with zeros to the specified height and width. — op_image_pad","text":"Padded image batch images.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_image_pad.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Pad images with zeros to the specified height and width. — op_image_pad","text":"","code":"images <- random_uniform(c(15, 25, 3)) padded_images <- op_image_pad(     images, 2, 3, target_height = 20, target_width = 30 ) shape(padded_images) ## shape(20, 30, 3) batch_images <- random_uniform(c(2, 15, 25, 3)) padded_batch <- op_image_pad(batch_images, 2, 3,                              target_height = 20,                              target_width = 30) shape(padded_batch) ## shape(2, 20, 30, 3)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_image_perspective_transform.html","id":null,"dir":"Reference","previous_headings":"","what":"Applies a perspective transformation to the image(s). — op_image_perspective_transform","title":"Applies a perspective transformation to the image(s). — op_image_perspective_transform","text":"Applies perspective transformation image(s).","code":""},{"path":"https://keras3.posit.co/dev/reference/op_image_perspective_transform.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Applies a perspective transformation to the image(s). — op_image_perspective_transform","text":"","code":"op_image_perspective_transform(   images,   start_points,   end_points,   interpolation = \"bilinear\",   fill_value = 0L,   data_format = NULL )"},{"path":"https://keras3.posit.co/dev/reference/op_image_perspective_transform.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Applies a perspective transformation to the image(s). — op_image_perspective_transform","text":"images Input image batch images. Must 3D 4D. start_points tensor shape (N, 4, 2) (4, 2), representing source points original image define transformation. end_points tensor shape (N, 4, 2) (4, 2), representing target points output image transformation. interpolation Interpolation method. Available methods \"nearest\", \"bilinear\". Defaults \"bilinear\". fill_value Value used points outside boundaries input extrapolation needed. Defaults 0. data_format string specifying data format input tensor. can either \"channels_last\" \"channels_first\". \"channels_last\" corresponds inputs shape (batch, height, width, channels), \"channels_first\" corresponds inputs shape (batch, channels, height, width). specified, value default keras.config.image_data_format.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_image_perspective_transform.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Applies a perspective transformation to the image(s). — op_image_perspective_transform","text":"Applied perspective transform image batch images.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_image_perspective_transform.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Applies a perspective transformation to the image(s). — op_image_perspective_transform","text":"","code":"# Batch of 2 RGB images (channels_last) x <- op_ones(c(2, 64, 80, 3)) start_points <- op_stack(list(   rbind(c(0, 0), c(0, 64), c(80, 0), c(80, 64)),   rbind(c(0, 0), c(0, 64), c(80, 0), c(80, 64)) )) end_points <- op_stack(list(   rbind(c(3, 5), c(7, 64), c(76, -10), c(84, 61)),   rbind(c(8, 10), c(10, 61), c(65, 3), c(88, 43)) )) y <- op_image_perspective_transform(x, start_points, end_points) op_shape(y) ## shape(2, 64, 80, 3) # Single RGB image (channels_last) x <- op_ones(c(64, 80, 3)) start_points <- rbind(c(0, 0), c(0, 64), c(80, 0), c(80, 64)) end_points <- rbind(c(3, 5), c(7, 64), c(76, -10), c(84, 61)) y <- op_image_perspective_transform(x, start_points, end_points) op_shape(y) ## shape(64, 80, 3) # Batch of 2 RGB images (channels_first) x <- op_ones(c(2, 3, 64, 80)) start_points <- op_stack(list(   rbind(c(0, 0), c(0, 64), c(80, 0), c(80, 64)),   rbind(c(0, 0), c(0, 64), c(80, 0), c(80, 64)) )) end_points <- op_stack(list(   rbind(c(3, 5), c(7, 64), c(76, -10), c(84, 61)),   rbind(c(8, 10), c(10, 61), c(65, 3), c(88, 43)) )) y <- op_image_perspective_transform(   x, start_points, end_points,   data_format = \"channels_first\" ) op_shape(y) ## shape(2, 3, 64, 80)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_image_resize.html","id":null,"dir":"Reference","previous_headings":"","what":"Resize images to size using the specified interpolation method. — op_image_resize","title":"Resize images to size using the specified interpolation method. — op_image_resize","text":"Resize images size using specified interpolation method.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_image_resize.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Resize images to size using the specified interpolation method. — op_image_resize","text":"","code":"op_image_resize(   images,   size,   interpolation = \"bilinear\",   antialias = FALSE,   crop_to_aspect_ratio = FALSE,   pad_to_aspect_ratio = FALSE,   fill_mode = \"constant\",   fill_value = 0,   data_format = NULL )"},{"path":"https://keras3.posit.co/dev/reference/op_image_resize.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Resize images to size using the specified interpolation method. — op_image_resize","text":"images Input image batch images. Must 3D 4D. size Size output image (height, width) format. interpolation Interpolation method. Available methods \"nearest\", \"bilinear\", \"bicubic\". Defaults \"bilinear\". antialias Whether use antialiasing filter downsampling image. Defaults FALSE. crop_to_aspect_ratio TRUE, resize images without aspect ratio distortion. original aspect ratio differs target aspect ratio, output image cropped return largest possible window image (size (height, width)) matches target aspect ratio. default (crop_to_aspect_ratio=FALSE), aspect ratio may preserved. pad_to_aspect_ratio TRUE, pad images without aspect ratio distortion. original aspect ratio differs target aspect ratio, output image evenly padded short side. fill_mode using pad_to_aspect_ratio=TRUE, padded areas filled according given mode. \"constant\" supported time (fill constant value, equal fill_value). fill_value Float. Padding value use pad_to_aspect_ratio=TRUE. data_format string specifying data format input tensor. can either \"channels_last\" \"channels_first\". \"channels_last\" corresponds inputs shape (batch, height, width, channels), \"channels_first\" corresponds inputs shape (batch, channels, height, width). specified, value default config_image_data_format().","code":""},{"path":"https://keras3.posit.co/dev/reference/op_image_resize.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Resize images to size using the specified interpolation method. — op_image_resize","text":"Resized image batch images.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_image_resize.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Resize images to size using the specified interpolation method. — op_image_resize","text":"","code":"x <- random_uniform(c(2, 4, 4, 3)) # batch of 2 RGB images y <- op_image_resize(x, c(2, 2)) shape(y) ## shape(2, 2, 2, 3) x <- random_uniform(c(4, 4, 3)) # single RGB image y <- op_image_resize(x, c(2, 2)) shape(y) ## shape(2, 2, 3) x <- random_uniform(c(2, 3, 4, 4)) # batch of 2 RGB images y <- op_image_resize(x, c(2, 2), data_format = \"channels_first\") shape(y) ## shape(2, 3, 2, 2)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_image_rgb_to_grayscale.html","id":null,"dir":"Reference","previous_headings":"","what":"Convert RGB images to grayscale. — op_image_rgb_to_grayscale","title":"Convert RGB images to grayscale. — op_image_rgb_to_grayscale","text":"function converts RGB images grayscale images. supports 3D 4D tensors, last dimension represents channels.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_image_rgb_to_grayscale.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Convert RGB images to grayscale. — op_image_rgb_to_grayscale","text":"","code":"op_image_rgb_to_grayscale(images, data_format = NULL)"},{"path":"https://keras3.posit.co/dev/reference/op_image_rgb_to_grayscale.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Convert RGB images to grayscale. — op_image_rgb_to_grayscale","text":"images Input image batch images. Must 3D 4D. data_format string specifying data format input tensor. can either \"channels_last\" \"channels_first\". \"channels_last\" corresponds inputs shape (batch, height, width, channels), \"channels_first\" corresponds inputs shape (batch, channels, height, width). specified, value default config_image_data_format().","code":""},{"path":"https://keras3.posit.co/dev/reference/op_image_rgb_to_grayscale.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Convert RGB images to grayscale. — op_image_rgb_to_grayscale","text":"Grayscale image batch grayscale images.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_image_rgb_to_grayscale.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Convert RGB images to grayscale. — op_image_rgb_to_grayscale","text":"","code":"x <- random_uniform(c(2, 4, 4, 3)) y <- op_image_rgb_to_grayscale(x) shape(y) ## shape(2, 4, 4, 1) x <- random_uniform(c(4, 4, 3)) # Single RGB image y = op_image_rgb_to_grayscale(x) shape(y) ## shape(4, 4, 1) x <- random_uniform(c(2, 3, 4, 4)) y <- op_image_rgb_to_grayscale(x, data_format=\"channels_first\") shape(y) ## shape(2, 1, 4, 4)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_image_rgb_to_hsv.html","id":null,"dir":"Reference","previous_headings":"","what":"Convert RGB images to HSV. — op_image_rgb_to_hsv","title":"Convert RGB images to HSV. — op_image_rgb_to_hsv","text":"images must float dtype, output well defined values images [0, 1]. HSV values [0, 1]. hue 0 corresponds pure red, 1/3 pure green, 2/3 pure blue.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_image_rgb_to_hsv.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Convert RGB images to HSV. — op_image_rgb_to_hsv","text":"","code":"op_image_rgb_to_hsv(images, data_format = NULL)"},{"path":"https://keras3.posit.co/dev/reference/op_image_rgb_to_hsv.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Convert RGB images to HSV. — op_image_rgb_to_hsv","text":"images Input image batch images. Must 3D 4D. data_format string specifying data format input tensor. can either \"channels_last\" \"channels_first\". \"channels_last\" corresponds inputs shape (batch, height, width, channels), \"channels_first\" corresponds inputs shape (batch, channels, height, width). specified, value default config_image_data_format().","code":""},{"path":"https://keras3.posit.co/dev/reference/op_image_rgb_to_hsv.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Convert RGB images to HSV. — op_image_rgb_to_hsv","text":"HSV image batch HSV images.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_image_rgb_to_hsv.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Convert RGB images to HSV. — op_image_rgb_to_hsv","text":"","code":"x <- random_uniform(c(2, 4, 4, 3)) y <- op_image_rgb_to_hsv(x) shape(y) ## shape(2, 4, 4, 3) x <- random_uniform(c(4, 4, 3)) # Single RGB image y <- op_image_rgb_to_hsv(x) shape(y) ## shape(4, 4, 3) x <- random_uniform(c(2, 3, 4, 4)) y <- op_image_rgb_to_hsv(x, data_format = \"channels_first\") shape(y) ## shape(2, 3, 4, 4)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_in_top_k.html","id":null,"dir":"Reference","previous_headings":"","what":"Checks if the targets are in the top-k predictions. — op_in_top_k","title":"Checks if the targets are in the top-k predictions. — op_in_top_k","text":"Checks targets top-k predictions.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_in_top_k.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Checks if the targets are in the top-k predictions. — op_in_top_k","text":"","code":"op_in_top_k(targets, predictions, k)"},{"path":"https://keras3.posit.co/dev/reference/op_in_top_k.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Checks if the targets are in the top-k predictions. — op_in_top_k","text":"targets tensor true labels. predictions tensor predicted labels. k integer representing number predictions consider.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_in_top_k.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Checks if the targets are in the top-k predictions. — op_in_top_k","text":"boolean tensor shape targets, element indicates whether corresponding target top-k predictions.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_in_top_k.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Checks if the targets are in the top-k predictions. — op_in_top_k","text":"","code":"targets <- op_array(c(2, 5, 3), \"int32\") predictions <- op_array(dtype = \"float32\", rbind(   c(0.1, 0.4, 0.6, 0.9, 0.5),   c(0.1, 0.7, 0.9, 0.8, 0.3),   c(0.1, 0.6, 0.9, 0.9, 0.5) )) op_in_top_k(targets, predictions, k = 3L) ## tf.Tensor([ True False  True], shape=(3), dtype=bool)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_inner.html","id":null,"dir":"Reference","previous_headings":"","what":"Return the inner product of two tensors. — op_inner","title":"Return the inner product of two tensors. — op_inner","text":"Ordinary inner product vectors 1-D tensors (without complex conjugation), higher dimensions sum product last axes. Multidimensional arrays treated vectors flattening last axes. resulting dot product performed last axes.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_inner.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Return the inner product of two tensors. — op_inner","text":"","code":"op_inner(x1, x2)"},{"path":"https://keras3.posit.co/dev/reference/op_inner.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Return the inner product of two tensors. — op_inner","text":"x1 First input tensor. x2 Second input tensor. last dimension x1 x2 must match.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_inner.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Return the inner product of two tensors. — op_inner","text":"Output tensor. shape output determined broadcasting shapes x1 x2 removing last axes.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_inv.html","id":null,"dir":"Reference","previous_headings":"","what":"Computes the inverse of a square tensor. — op_inv","title":"Computes the inverse of a square tensor. — op_inv","text":"Computes inverse square tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_inv.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes the inverse of a square tensor. — op_inv","text":"","code":"op_inv(x)"},{"path":"https://keras3.posit.co/dev/reference/op_inv.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Computes the inverse of a square tensor. — op_inv","text":"x Input tensor shape (..., M, M).","code":""},{"path":"https://keras3.posit.co/dev/reference/op_inv.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Computes the inverse of a square tensor. — op_inv","text":"tensor shape (..., M, M) representing inverse x.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_irfft.html","id":null,"dir":"Reference","previous_headings":"","what":"Inverse real-valued Fast Fourier transform along the last axis. — op_irfft","title":"Inverse real-valued Fast Fourier transform along the last axis. — op_irfft","text":"Computes inverse 1D Discrete Fourier Transform real-valued signal inner-dimension input. inner-dimension input assumed result RFFT: fft_length / 2 + 1 unique components DFT real-valued signal. fft_length provided, computed size inner-dimension input (fft_length = 2 * (inner - 1)). FFT length used compute odd, provided since inferred properly. Along axis IRFFT computed , fft_length / 2 + 1 smaller corresponding dimension input, dimension cropped. larger, dimension padded zeros.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_irfft.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Inverse real-valued Fast Fourier transform along the last axis. — op_irfft","text":"","code":"op_irfft(x, fft_length = NULL)"},{"path":"https://keras3.posit.co/dev/reference/op_irfft.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Inverse real-valued Fast Fourier transform along the last axis. — op_irfft","text":"x List real imaginary parts input tensor. tensors list floating type. fft_length integer representing number fft length. specified, inferred length last axis x. Defaults NULL.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_irfft.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Inverse real-valued Fast Fourier transform along the last axis. — op_irfft","text":"tensor containing inverse real-valued Fast Fourier Transform along last axis x.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_irfft.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Inverse real-valued Fast Fourier transform along the last axis. — op_irfft","text":"","code":"real <- op_array(c(0, 1, 2, 3, 4)) imag <- op_array(c(0, 1, 2, 3, 4)) op_irfft(c(real, imag)) #> tf.Tensor( #> [ 2.         -2.0606601   0.5        -0.35355338  0.          0.06066012 #>  -0.5         0.35355338], shape=(8), dtype=float32) all.equal(op_irfft(op_rfft(real, 5), 5), real) #> [1] TRUE"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_is_tensor.html","id":null,"dir":"Reference","previous_headings":"","what":"Check whether the given object is a tensor. — op_is_tensor","title":"Check whether the given object is a tensor. — op_is_tensor","text":"Check whether given object tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_is_tensor.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Check whether the given object is a tensor. — op_is_tensor","text":"","code":"op_is_tensor(x)"},{"path":"https://keras3.posit.co/dev/reference/op_is_tensor.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Check whether the given object is a tensor. — op_is_tensor","text":"x variable.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_is_tensor.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Check whether the given object is a tensor. — op_is_tensor","text":"TRUE x tensor, otherwise FALSE.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_is_tensor.html","id":"note","dir":"Reference","previous_headings":"","what":"Note","title":"Check whether the given object is a tensor. — op_is_tensor","text":"checks backend specific tensors passing TensorFlow tensor return FALSE backend PyTorch JAX.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_isclose.html","id":null,"dir":"Reference","previous_headings":"","what":"Return whether two tensors are element-wise almost equal. — op_isclose","title":"Return whether two tensors are element-wise almost equal. — op_isclose","text":"Return whether two tensors element-wise almost equal.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_isclose.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Return whether two tensors are element-wise almost equal. — op_isclose","text":"","code":"op_isclose(x1, x2, rtol = 1e-05, atol = 1e-08, equal_nan = FALSE)"},{"path":"https://keras3.posit.co/dev/reference/op_isclose.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Return whether two tensors are element-wise almost equal. — op_isclose","text":"x1 First input tensor. x2 Second input tensor. rtol Relative tolerance. atol Absolute tolerance. equal_nan TRUE, element-wise NaNs considered equal.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_isclose.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Return whether two tensors are element-wise almost equal. — op_isclose","text":"Output boolean tensor.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_isfinite.html","id":null,"dir":"Reference","previous_headings":"","what":"Return whether a tensor is finite, element-wise. — op_isfinite","title":"Return whether a tensor is finite, element-wise. — op_isfinite","text":"Real values finite NaN, positive infinity, negative infinity. Complex values finite real imaginary parts finite.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_isfinite.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Return whether a tensor is finite, element-wise. — op_isfinite","text":"","code":"op_isfinite(x)"},{"path":"https://keras3.posit.co/dev/reference/op_isfinite.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Return whether a tensor is finite, element-wise. — op_isfinite","text":"x Input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_isfinite.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Return whether a tensor is finite, element-wise. — op_isfinite","text":"Output boolean tensor.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_isinf.html","id":null,"dir":"Reference","previous_headings":"","what":"Test element-wise for positive or negative infinity. — op_isinf","title":"Test element-wise for positive or negative infinity. — op_isinf","text":"Test element-wise positive negative infinity.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_isinf.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Test element-wise for positive or negative infinity. — op_isinf","text":"","code":"op_isinf(x)"},{"path":"https://keras3.posit.co/dev/reference/op_isinf.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Test element-wise for positive or negative infinity. — op_isinf","text":"x Input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_isinf.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Test element-wise for positive or negative infinity. — op_isinf","text":"Output boolean tensor.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_isnan.html","id":null,"dir":"Reference","previous_headings":"","what":"Test element-wise for NaN and return result as a boolean tensor. — op_isnan","title":"Test element-wise for NaN and return result as a boolean tensor. — op_isnan","text":"Test element-wise NaN return result boolean tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_isnan.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Test element-wise for NaN and return result as a boolean tensor. — op_isnan","text":"","code":"op_isnan(x)"},{"path":"https://keras3.posit.co/dev/reference/op_isnan.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Test element-wise for NaN and return result as a boolean tensor. — op_isnan","text":"x Input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_isnan.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Test element-wise for NaN and return result as a boolean tensor. — op_isnan","text":"Output boolean tensor.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_istft.html","id":null,"dir":"Reference","previous_headings":"","what":"Inverse Short-Time Fourier Transform along the last axis of the input. — op_istft","title":"Inverse Short-Time Fourier Transform along the last axis of the input. — op_istft","text":"reconstruct original waveform, parameters stft.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_istft.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Inverse Short-Time Fourier Transform along the last axis of the input. — op_istft","text":"","code":"op_istft(   x,   sequence_length,   sequence_stride,   fft_length,   length = NULL,   window = \"hann\",   center = TRUE )"},{"path":"https://keras3.posit.co/dev/reference/op_istft.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Inverse Short-Time Fourier Transform along the last axis of the input. — op_istft","text":"x Tuple real imaginary parts input tensor. tensors list floating type. sequence_length integer representing sequence length. sequence_stride integer representing sequence hop size. fft_length integer representing size FFT produced stft. type int32. length integer representing output clipped exactly length. specified, padding clipping take place. Defaults NULL. window string, tensor window NULL. window string, available values \"hann\" \"hamming\". window tensor, used directly window length must sequence_length. window NULL, windowing used. Defaults \"hann\". center Whether x padded sides t-th sequence centered time t * sequence_stride. Defaults TRUE.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_istft.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Inverse Short-Time Fourier Transform along the last axis of the input. — op_istft","text":"tensor containing inverse Short-Time Fourier Transform along last axis x.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_istft.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Inverse Short-Time Fourier Transform along the last axis of the input. — op_istft","text":"","code":"x <- op_convert_to_tensor(c(0, 1, 2, 3, 4)) op_istft(op_stft(x, 1, 1, 1), 1, 1, 1) ## tf.Tensor([], shape=(0), dtype=float64) # array([0.0, 1.0, 2.0, 3.0, 4.0])"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_kaiser.html","id":null,"dir":"Reference","previous_headings":"","what":"Kaiser window function. — op_kaiser","title":"Kaiser window function. — op_kaiser","text":"Kaiser window defined : w[n] = I0(beta * sqrt(1 - (2 * n / (N - 1) - 1)^2)) / I0(beta) I0 modified zeroth-order Bessel function first kind.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_kaiser.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Kaiser window function. — op_kaiser","text":"","code":"op_kaiser(x, beta)"},{"path":"https://keras3.posit.co/dev/reference/op_kaiser.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Kaiser window function. — op_kaiser","text":"x Length window. Must positive integer. beta Shape parameter window.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_kaiser.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Kaiser window function. — op_kaiser","text":"1D tensor containing window values.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_kaiser.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Kaiser window function. — op_kaiser","text":"","code":"op_kaiser(5, beta = 14) ## tf.Tensor([7.7268669e-06 1.6493215e-01 1.0000000e+00 1.6493215e-01 7.7268669e-06], shape=(5), dtype=float32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_layer_normalization.html","id":null,"dir":"Reference","previous_headings":"","what":"Layer normalization (Ba et al., 2016). — op_layer_normalization","title":"Layer normalization (Ba et al., 2016). — op_layer_normalization","text":"Normalizes activations x example independently centering mean 0 scaling unit variance along specified axis.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_layer_normalization.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Layer normalization (Ba et al., 2016). — op_layer_normalization","text":"","code":"op_layer_normalization(   x,   gamma = NULL,   beta = NULL,   axis = -1L,   epsilon = NULL,   ... )"},{"path":"https://keras3.posit.co/dev/reference/op_layer_normalization.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Layer normalization (Ba et al., 2016). — op_layer_normalization","text":"x Input tensor. gamma Optional scaling factor applied normalized output. beta Optional offset added normalized output. axis Axis axes along compute statistics. Defaults -1. epsilon Small constant added variance numerical stability. ... forward/backward compatibility.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_layer_normalization.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Layer normalization (Ba et al., 2016). — op_layer_normalization","text":"Tensor shape x containing normalized values.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_layer_normalization.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Layer normalization (Ba et al., 2016). — op_layer_normalization","text":"","code":"x <- op_arange(5, dtype = \"float32\") op_layer_normalization(x) ## tf.Tensor([-1.4142134 -0.7071067  0.         0.7071068  1.4142137], shape=(5), dtype=float32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_leaky_relu.html","id":null,"dir":"Reference","previous_headings":"","what":"Leaky version of a Rectified Linear Unit activation function. — op_leaky_relu","title":"Leaky version of a Rectified Linear Unit activation function. — op_leaky_relu","text":"allows small gradient unit active, defined : f(x) = alpha * x x < 0 f(x) = x x >= 0.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_leaky_relu.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Leaky version of a Rectified Linear Unit activation function. — op_leaky_relu","text":"","code":"op_leaky_relu(x, negative_slope = 0.2)"},{"path":"https://keras3.posit.co/dev/reference/op_leaky_relu.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Leaky version of a Rectified Linear Unit activation function. — op_leaky_relu","text":"x Input tensor. negative_slope Slope activation function x < 0. Defaults 0.2.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_leaky_relu.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Leaky version of a Rectified Linear Unit activation function. — op_leaky_relu","text":"tensor shape x.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_leaky_relu.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Leaky version of a Rectified Linear Unit activation function. — op_leaky_relu","text":"","code":"x <- op_array(c(-1., 0., 1.)) op_leaky_relu(x) ## tf.Tensor([-0.2  0.   1. ], shape=(3), dtype=float32) # array([-0.2,  0. ,  1. ], shape=(3,), dtype=float64) x <- seq(-5, 5, .1) plot(x, op_leaky_relu(x),      type = 'l', panel.first = grid())"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_left_shift.html","id":null,"dir":"Reference","previous_headings":"","what":"Shift the bits of an integer to the left. — op_left_shift","title":"Shift the bits of an integer to the left. — op_left_shift","text":"Bits shifted left appending y 0s right x. Since internal representation numbers binary format, operation equivalent multiplying x 2^y.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_left_shift.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Shift the bits of an integer to the left. — op_left_shift","text":"","code":"op_left_shift(x, y)"},{"path":"https://keras3.posit.co/dev/reference/op_left_shift.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Shift the bits of an integer to the left. — op_left_shift","text":"x Input integer tensor. y Input integer tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_left_shift.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Shift the bits of an integer to the left. — op_left_shift","text":"Result tensor.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_less.html","id":null,"dir":"Reference","previous_headings":"","what":"Return the truth value of x1 < x2 element-wise. — op_less","title":"Return the truth value of x1 < x2 element-wise. — op_less","text":"Note function automatically called using R operator < tensor.","code":"(x <- op_arange(4)) ## tf.Tensor([1. 2. 3. 4.], shape=(4), dtype=float32) op_less(x, 2) ## tf.Tensor([ True False False False], shape=(4), dtype=bool) x < 2 ## tf.Tensor([ True False False False], shape=(4), dtype=bool)"},{"path":"https://keras3.posit.co/dev/reference/op_less.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Return the truth value of x1 < x2 element-wise. — op_less","text":"","code":"op_less(x1, x2)"},{"path":"https://keras3.posit.co/dev/reference/op_less.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Return the truth value of x1 < x2 element-wise. — op_less","text":"x1 First input tensor. x2 Second input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_less.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Return the truth value of x1 < x2 element-wise. — op_less","text":"Output tensor, element-wise comparison x1 x2.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_less_equal.html","id":null,"dir":"Reference","previous_headings":"","what":"Return the truth value of x1 <= x2 element-wise. — op_less_equal","title":"Return the truth value of x1 <= x2 element-wise. — op_less_equal","text":"Note function automatically called using R operator <= tensor.","code":"(x <- op_arange(4)) ## tf.Tensor([1. 2. 3. 4.], shape=(4), dtype=float32) op_less_equal(x, 2) ## tf.Tensor([ True  True False False], shape=(4), dtype=bool) x <= 2 ## tf.Tensor([ True  True False False], shape=(4), dtype=bool)"},{"path":"https://keras3.posit.co/dev/reference/op_less_equal.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Return the truth value of x1 <= x2 element-wise. — op_less_equal","text":"","code":"op_less_equal(x1, x2)"},{"path":"https://keras3.posit.co/dev/reference/op_less_equal.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Return the truth value of x1 <= x2 element-wise. — op_less_equal","text":"x1 First input tensor. x2 Second input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_less_equal.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Return the truth value of x1 <= x2 element-wise. — op_less_equal","text":"Output tensor, element-wise comparison x1 x2.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_linspace.html","id":null,"dir":"Reference","previous_headings":"","what":"Return evenly spaced numbers over a specified interval. — op_linspace","title":"Return evenly spaced numbers over a specified interval. — op_linspace","text":"Returns num evenly spaced samples, calculated interval [start, stop]. endpoint interval can optionally excluded.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_linspace.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Return evenly spaced numbers over a specified interval. — op_linspace","text":"","code":"op_linspace(   start,   stop,   num = 50L,   endpoint = TRUE,   retstep = FALSE,   dtype = NULL,   axis = 1L )"},{"path":"https://keras3.posit.co/dev/reference/op_linspace.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Return evenly spaced numbers over a specified interval. — op_linspace","text":"start starting value sequence. stop end value sequence, unless endpoint set FALSE. case, sequence consists last num + 1 evenly spaced samples, stop excluded. Note step size changes endpoint FALSE. num Number samples generate. Defaults 50. Must non-negative. endpoint TRUE, stop last sample. Otherwise, included. Defaults TRUE. retstep TRUE, return (samples, step), step spacing samples. dtype type output tensor. axis axis result store samples. Relevant start stop array-like. Defaults 1, first axis.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_linspace.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Return evenly spaced numbers over a specified interval. — op_linspace","text":"tensor evenly spaced numbers. retstep TRUE, returns (samples, step)","code":""},{"path":"https://keras3.posit.co/dev/reference/op_linspace.html","id":"note","dir":"Reference","previous_headings":"","what":"Note","title":"Return evenly spaced numbers over a specified interval. — op_linspace","text":"Torch backend support axis argument.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_log.html","id":null,"dir":"Reference","previous_headings":"","what":"Natural logarithm, element-wise. — op_log","title":"Natural logarithm, element-wise. — op_log","text":"Natural logarithm, element-wise.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_log.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Natural logarithm, element-wise. — op_log","text":"","code":"op_log(x)"},{"path":"https://keras3.posit.co/dev/reference/op_log.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Natural logarithm, element-wise. — op_log","text":"x Input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_log.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Natural logarithm, element-wise. — op_log","text":"Output tensor, element-wise natural logarithm x.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_log10.html","id":null,"dir":"Reference","previous_headings":"","what":"Return the base 10 logarithm of the input tensor, element-wise. — op_log10","title":"Return the base 10 logarithm of the input tensor, element-wise. — op_log10","text":"Return base 10 logarithm input tensor, element-wise.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_log10.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Return the base 10 logarithm of the input tensor, element-wise. — op_log10","text":"","code":"op_log10(x)"},{"path":"https://keras3.posit.co/dev/reference/op_log10.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Return the base 10 logarithm of the input tensor, element-wise. — op_log10","text":"x Input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_log10.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Return the base 10 logarithm of the input tensor, element-wise. — op_log10","text":"Output tensor, element-wise base 10 logarithm x.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_log1p.html","id":null,"dir":"Reference","previous_headings":"","what":"Returns the natural logarithm of one plus the x, element-wise. — op_log1p","title":"Returns the natural logarithm of one plus the x, element-wise. — op_log1p","text":"Calculates log(1 + x).","code":""},{"path":"https://keras3.posit.co/dev/reference/op_log1p.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Returns the natural logarithm of one plus the x, element-wise. — op_log1p","text":"","code":"op_log1p(x)"},{"path":"https://keras3.posit.co/dev/reference/op_log1p.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Returns the natural logarithm of one plus the x, element-wise. — op_log1p","text":"x Input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_log1p.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Returns the natural logarithm of one plus the x, element-wise. — op_log1p","text":"Output tensor, element-wise natural logarithm 1 + x.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_log2.html","id":null,"dir":"Reference","previous_headings":"","what":"Base-2 logarithm of x, element-wise. — op_log2","title":"Base-2 logarithm of x, element-wise. — op_log2","text":"Base-2 logarithm x, element-wise.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_log2.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Base-2 logarithm of x, element-wise. — op_log2","text":"","code":"op_log2(x)"},{"path":"https://keras3.posit.co/dev/reference/op_log2.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Base-2 logarithm of x, element-wise. — op_log2","text":"x Input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_log2.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Base-2 logarithm of x, element-wise. — op_log2","text":"Output tensor, element-wise base-2 logarithm x.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_log_sigmoid.html","id":null,"dir":"Reference","previous_headings":"","what":"Logarithm of the sigmoid activation function. — op_log_sigmoid","title":"Logarithm of the sigmoid activation function. — op_log_sigmoid","text":"defined f(x) = log(1 / (1 + exp(-x))).","code":""},{"path":"https://keras3.posit.co/dev/reference/op_log_sigmoid.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Logarithm of the sigmoid activation function. — op_log_sigmoid","text":"","code":"op_log_sigmoid(x)"},{"path":"https://keras3.posit.co/dev/reference/op_log_sigmoid.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Logarithm of the sigmoid activation function. — op_log_sigmoid","text":"x Input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_log_sigmoid.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Logarithm of the sigmoid activation function. — op_log_sigmoid","text":"tensor shape x.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_log_sigmoid.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Logarithm of the sigmoid activation function. — op_log_sigmoid","text":"","code":"x <- op_convert_to_tensor(c(-0.541391, 0.0, 0.50, 5.0)) op_log_sigmoid(x) ## tf.Tensor([-1.00004181 -0.69314718 -0.47407698 -0.00671535], shape=(4), dtype=float64)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_log_softmax.html","id":null,"dir":"Reference","previous_headings":"","what":"Log-softmax activation function. — op_log_softmax","title":"Log-softmax activation function. — op_log_softmax","text":"defined : f(x) = x - max(x) - log(sum(exp(x - max(x))))","code":""},{"path":"https://keras3.posit.co/dev/reference/op_log_softmax.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Log-softmax activation function. — op_log_softmax","text":"","code":"op_log_softmax(x, axis = -1L)"},{"path":"https://keras3.posit.co/dev/reference/op_log_softmax.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Log-softmax activation function. — op_log_softmax","text":"x Input tensor. axis Integer, axis along log-softmax applied. Defaults -1.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_log_softmax.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Log-softmax activation function. — op_log_softmax","text":"tensor shape x.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_log_softmax.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Log-softmax activation function. — op_log_softmax","text":"","code":"x <- op_array(c(-1., 0., 1.)) op_log_softmax(x) ## tf.Tensor([-2.407606   -1.4076059  -0.40760595], shape=(3), dtype=float32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_logaddexp.html","id":null,"dir":"Reference","previous_headings":"","what":"Logarithm of the sum of exponentiations of the inputs. — op_logaddexp","title":"Logarithm of the sum of exponentiations of the inputs. — op_logaddexp","text":"Calculates log(exp(x1) + exp(x2)).","code":""},{"path":"https://keras3.posit.co/dev/reference/op_logaddexp.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Logarithm of the sum of exponentiations of the inputs. — op_logaddexp","text":"","code":"op_logaddexp(x1, x2)"},{"path":"https://keras3.posit.co/dev/reference/op_logaddexp.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Logarithm of the sum of exponentiations of the inputs. — op_logaddexp","text":"x1 Input tensor. x2 Input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_logaddexp.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Logarithm of the sum of exponentiations of the inputs. — op_logaddexp","text":"Output tensor, element-wise logarithm sum exponentiations inputs.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_logdet.html","id":null,"dir":"Reference","previous_headings":"","what":"Computes log of the determinant of a hermitian positive definite matrix. — op_logdet","title":"Computes log of the determinant of a hermitian positive definite matrix. — op_logdet","text":"Computes log determinant hermitian positive definite matrix.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_logdet.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes log of the determinant of a hermitian positive definite matrix. — op_logdet","text":"","code":"op_logdet(x)"},{"path":"https://keras3.posit.co/dev/reference/op_logdet.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Computes log of the determinant of a hermitian positive definite matrix. — op_logdet","text":"x Input matrix. must 2D square.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_logdet.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Computes log of the determinant of a hermitian positive definite matrix. — op_logdet","text":"natural log determinant matrix.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_logical_and.html","id":null,"dir":"Reference","previous_headings":"","what":"Computes the element-wise logical AND of the given input tensors. — op_logical_and","title":"Computes the element-wise logical AND of the given input tensors. — op_logical_and","text":"Zeros treated FALSE non-zeros treated TRUE.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_logical_and.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes the element-wise logical AND of the given input tensors. — op_logical_and","text":"","code":"op_logical_and(x1, x2)"},{"path":"https://keras3.posit.co/dev/reference/op_logical_and.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Computes the element-wise logical AND of the given input tensors. — op_logical_and","text":"x1 Input tensor. x2 Input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_logical_and.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Computes the element-wise logical AND of the given input tensors. — op_logical_and","text":"Output tensor, element-wise logical inputs.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_logical_and.html","id":"details","dir":"Reference","previous_headings":"","what":"Details","title":"Computes the element-wise logical AND of the given input tensors. — op_logical_and","text":"Note function automatically called using R operator & tensor.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_logical_not.html","id":null,"dir":"Reference","previous_headings":"","what":"Computes the element-wise NOT of the given input tensor. — op_logical_not","title":"Computes the element-wise NOT of the given input tensor. — op_logical_not","text":"Zeros treated FALSE non-zeros treated TRUE. Note function automatically called using R operator ! tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_logical_not.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes the element-wise NOT of the given input tensor. — op_logical_not","text":"","code":"op_logical_not(x)"},{"path":"https://keras3.posit.co/dev/reference/op_logical_not.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Computes the element-wise NOT of the given input tensor. — op_logical_not","text":"x Input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_logical_not.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Computes the element-wise NOT of the given input tensor. — op_logical_not","text":"Output tensor, element-wise logical input.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_logical_or.html","id":null,"dir":"Reference","previous_headings":"","what":"Computes the element-wise logical OR of the given input tensors. — op_logical_or","title":"Computes the element-wise logical OR of the given input tensors. — op_logical_or","text":"Zeros treated FALSE non-zeros treated TRUE. Note function automatically called using R operator | tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_logical_or.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes the element-wise logical OR of the given input tensors. — op_logical_or","text":"","code":"op_logical_or(x1, x2)"},{"path":"https://keras3.posit.co/dev/reference/op_logical_or.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Computes the element-wise logical OR of the given input tensors. — op_logical_or","text":"x1 Input tensor. x2 Input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_logical_or.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Computes the element-wise logical OR of the given input tensors. — op_logical_or","text":"Output tensor, element-wise logical inputs.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_logical_xor.html","id":null,"dir":"Reference","previous_headings":"","what":"Compute the truth value of x1 XOR x2, element-wise. — op_logical_xor","title":"Compute the truth value of x1 XOR x2, element-wise. — op_logical_xor","text":"Compute truth value x1 XOR x2, element-wise.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_logical_xor.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Compute the truth value of x1 XOR x2, element-wise. — op_logical_xor","text":"","code":"op_logical_xor(x1, x2)"},{"path":"https://keras3.posit.co/dev/reference/op_logical_xor.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Compute the truth value of x1 XOR x2, element-wise. — op_logical_xor","text":"x1 First input tensor. x2 Second input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_logical_xor.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Compute the truth value of x1 XOR x2, element-wise. — op_logical_xor","text":"Output boolean tensor.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_logspace.html","id":null,"dir":"Reference","previous_headings":"","what":"Returns numbers spaced evenly on a log scale. — op_logspace","title":"Returns numbers spaced evenly on a log scale. — op_logspace","text":"linear space, sequence starts base ** start ends base ** stop (see endpoint ).","code":""},{"path":"https://keras3.posit.co/dev/reference/op_logspace.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Returns numbers spaced evenly on a log scale. — op_logspace","text":"","code":"op_logspace(   start,   stop,   num = 50L,   endpoint = TRUE,   base = 10L,   dtype = NULL,   axis = 1L )"},{"path":"https://keras3.posit.co/dev/reference/op_logspace.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Returns numbers spaced evenly on a log scale. — op_logspace","text":"start starting value sequence. stop final value sequence, unless endpoint FALSE. case, num + 1 values spaced interval log-space, last (sequence length num) returned. num Number samples generate. Defaults 50. endpoint TRUE, stop last sample. Otherwise, included. Defaults TRUE. base base log space. Defaults 10. dtype type output tensor. axis axis result store samples. Relevant start stop array-like.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_logspace.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Returns numbers spaced evenly on a log scale. — op_logspace","text":"tensor evenly spaced samples log scale.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_logspace.html","id":"note","dir":"Reference","previous_headings":"","what":"Note","title":"Returns numbers spaced evenly on a log scale. — op_logspace","text":"Torch backend support axis argument.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_logsumexp.html","id":null,"dir":"Reference","previous_headings":"","what":"Computes the logarithm of sum of exponentials of elements in a tensor. — op_logsumexp","title":"Computes the logarithm of sum of exponentials of elements in a tensor. — op_logsumexp","text":"Computes logarithm sum exponentials elements tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_logsumexp.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes the logarithm of sum of exponentials of elements in a tensor. — op_logsumexp","text":"","code":"op_logsumexp(x, axis = NULL, keepdims = FALSE)"},{"path":"https://keras3.posit.co/dev/reference/op_logsumexp.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Computes the logarithm of sum of exponentials of elements in a tensor. — op_logsumexp","text":"x Input tensor. axis integer list integers specifying axis/axes along compute sum. NULL, sum computed elements. Defaults NULL. keepdims boolean indicating whether keep dimensions input tensor computing sum. Defaults FALSE.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_logsumexp.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Computes the logarithm of sum of exponentials of elements in a tensor. — op_logsumexp","text":"tensor containing logarithm sum exponentials elements x.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_logsumexp.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Computes the logarithm of sum of exponentials of elements in a tensor. — op_logsumexp","text":"","code":"x <- op_convert_to_tensor(c(1, 2, 3)) op_logsumexp(x) ## tf.Tensor(3.4076059644443806, shape=(), dtype=float64)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_lstsq.html","id":null,"dir":"Reference","previous_headings":"","what":"Return the least-squares solution to a linear matrix equation. — op_lstsq","title":"Return the least-squares solution to a linear matrix equation. — op_lstsq","text":"Computes vector x approximately solves equation %*% x = b. equation may -, well-, -determined (.e., number linearly independent rows can less , equal , greater number linearly independent columns). square full rank, x (round-error) exact solution equation. Else, x minimizes L2 norm b - %*% x. multiple minimizing solutions, one smallest L2 norm  returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_lstsq.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Return the least-squares solution to a linear matrix equation. — op_lstsq","text":"","code":"op_lstsq(a, b, rcond = NULL)"},{"path":"https://keras3.posit.co/dev/reference/op_lstsq.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Return the least-squares solution to a linear matrix equation. — op_lstsq","text":"\"Coefficient\" matrix shape (M, N). b Ordinate \"dependent variable\" values, shape (M) (M, K). b two-dimensional, least-squares solution calculated K columns b. rcond Cut-ratio small singular values . purposes rank determination, singular values treated zero smaller rcond times largest singular value .","code":""},{"path":"https://keras3.posit.co/dev/reference/op_lstsq.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Return the least-squares solution to a linear matrix equation. — op_lstsq","text":"Tensor shape (N) (N, K) containing least-squares solutions. NOTE: output differs numpy.linalg.lstsq(). NumPy returns tuple four elements, first least-squares solutions others essentially never used. Keras returns first value. done ensure consistency across backends (achieved values) simplify API.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_lu_factor.html","id":null,"dir":"Reference","previous_headings":"","what":"Computes the lower-upper decomposition of a square matrix. — op_lu_factor","title":"Computes the lower-upper decomposition of a square matrix. — op_lu_factor","text":"Computes lower-upper decomposition square matrix.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_lu_factor.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes the lower-upper decomposition of a square matrix. — op_lu_factor","text":"","code":"op_lu_factor(x)"},{"path":"https://keras3.posit.co/dev/reference/op_lu_factor.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Computes the lower-upper decomposition of a square matrix. — op_lu_factor","text":"x tensor shape (..., M, M).","code":""},{"path":"https://keras3.posit.co/dev/reference/op_lu_factor.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Computes the lower-upper decomposition of a square matrix. — op_lu_factor","text":"tuple two tensors: tensor shape (..., M, M) containing lower upper triangular matrices tensor shape (..., M) containing pivots.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_map.html","id":null,"dir":"Reference","previous_headings":"","what":"Map a function over leading array axes. — op_map","title":"Map a function over leading array axes. — op_map","text":"Like purrr::map() base::lapply(), except inputs outputs form stacked arrays. Consider using op_vectorized_map() transform instead, unless need apply function element element reduced memory usage heterogeneous computation control flow primitives. xs array type, semantics op_map() match implementation:","code":"op_map <- function(xs, f) {   xs |>     op_unstack() |>     lapply(f) |>     op_stack() }"},{"path":"https://keras3.posit.co/dev/reference/op_map.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Map a function over leading array axes. — op_map","text":"","code":"op_map(xs, f)"},{"path":"https://keras3.posit.co/dev/reference/op_map.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Map a function over leading array axes. — op_map","text":"xs Values map along leading axis. f Callable defines function apply element-wise first axis axes xs.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_map.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Map a function over leading array axes. — op_map","text":"Mapped values.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_map.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Map a function over leading array axes. — op_map","text":"","code":"f <- function(x) x^2 xs <- op_arange(10) ys <- op_map(xs, f) ys ## tf.Tensor([  1.   4.   9.  16.  25.  36.  49.  64.  81. 100.], shape=(10), dtype=float32) f <- function(x) list(y1 = x^2, y2 = x * 10)  # Can have nested outputs ys <- op_map(xs, f) ys$y1 ## tf.Tensor([  1.   4.   9.  16.  25.  36.  49.  64.  81. 100.], shape=(10), dtype=float32) ys$y2 ## tf.Tensor([ 10.  20.  30.  40.  50.  60.  70.  80.  90. 100.], shape=(10), dtype=float32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_matmul.html","id":null,"dir":"Reference","previous_headings":"","what":"Matrix product of two tensors. — op_matmul","title":"Matrix product of two tensors. — op_matmul","text":"tensors 1-dimensional, dot product (scalar) returned. either tensor N-D, N > 2, treated stack matrices residing last two indexes broadcast accordingly. first tensor 1-D, promoted matrix prepending 1 dimensions. matrix multiplication prepended 1 removed. second tensor 1-D, promoted matrix appending 1 dimensions. matrix multiplication appended 1 removed.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_matmul.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Matrix product of two tensors. — op_matmul","text":"","code":"op_matmul(x1, x2)"},{"path":"https://keras3.posit.co/dev/reference/op_matmul.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Matrix product of two tensors. — op_matmul","text":"x1 First tensor. x2 Second tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_matmul.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Matrix product of two tensors. — op_matmul","text":"Output tensor, matrix product inputs.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_max.html","id":null,"dir":"Reference","previous_headings":"","what":"Return the maximum of a tensor or maximum along an axis. — op_max","title":"Return the maximum of a tensor or maximum along an axis. — op_max","text":"Return maximum tensor maximum along axis.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_max.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Return the maximum of a tensor or maximum along an axis. — op_max","text":"","code":"op_max(x, axis = NULL, keepdims = FALSE, initial = NULL)"},{"path":"https://keras3.posit.co/dev/reference/op_max.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Return the maximum of a tensor or maximum along an axis. — op_max","text":"x Input tensor. axis Axis axes along operate. default, flattened input used. keepdims set TRUE, axes reduced left result dimensions size one. Defaults FALSE. initial minimum value output element. Defaults NULL.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_max.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Return the maximum of a tensor or maximum along an axis. — op_max","text":"Maximum x.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_max.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Return the maximum of a tensor or maximum along an axis. — op_max","text":"","code":"(x <- op_convert_to_tensor(rbind(c(1, 3, 5), c(1, 5, 2)))) ## tf.Tensor( ## [[1. 3. 5.] ##  [1. 5. 2.]], shape=(2, 3), dtype=float64) op_max(x) ## tf.Tensor(5.0, shape=(), dtype=float64) op_max(x, axis = 1) ## tf.Tensor([1. 5. 5.], shape=(3), dtype=float64) op_max(x, axis = 1, keepdims = TRUE) ## tf.Tensor([[1. 5. 5.]], shape=(1, 3), dtype=float64)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_max_pool.html","id":null,"dir":"Reference","previous_headings":"","what":"Max pooling operation. — op_max_pool","title":"Max pooling operation. — op_max_pool","text":"Max pooling operation.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_max_pool.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Max pooling operation. — op_max_pool","text":"","code":"op_max_pool(   inputs,   pool_size,   strides = NULL,   padding = \"valid\",   data_format = NULL )"},{"path":"https://keras3.posit.co/dev/reference/op_max_pool.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Max pooling operation. — op_max_pool","text":"inputs Tensor rank N+2. inputs shape (batch_size,) + inputs_spatial_shape + (num_channels,) data_format = \"channels_last\", (batch_size, num_channels) + inputs_spatial_shape data_format = \"channels_first\". Pooling happens spatial dimensions . pool_size int tuple/list integers size len(inputs_spatial_shape), specifying size pooling window spatial dimension input tensor. pool_size int, every spatial dimension shares pool_size. strides int tuple/list integers size len(inputs_spatial_shape). stride sliding window spatial dimension input tensor. strides int, every spatial dimension shares strides. padding string, either \"valid\" \"\". \"valid\" means padding applied, \"\" results padding evenly left/right /input output height/width dimension input strides = 1. data_format string, either \"channels_last\" \"channels_first\". data_format determines ordering dimensions inputs. data_format = \"channels_last\", inputs shape (batch_size, ..., channels) data_format = \"channels_first\", inputs shape (batch_size, channels, ...).","code":""},{"path":"https://keras3.posit.co/dev/reference/op_max_pool.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Max pooling operation. — op_max_pool","text":"tensor rank N+2, result max pooling operation.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_maximum.html","id":null,"dir":"Reference","previous_headings":"","what":"Element-wise maximum of x1 and x2. — op_maximum","title":"Element-wise maximum of x1 and x2. — op_maximum","text":"Element-wise maximum x1 x2.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_maximum.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Element-wise maximum of x1 and x2. — op_maximum","text":"","code":"op_maximum(x1, x2)  op_pmax(x1, x2)"},{"path":"https://keras3.posit.co/dev/reference/op_maximum.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Element-wise maximum of x1 and x2. — op_maximum","text":"x1 First tensor. x2 Second tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_maximum.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Element-wise maximum of x1 and x2. — op_maximum","text":"Output tensor, element-wise maximum x1 x2.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_mean.html","id":null,"dir":"Reference","previous_headings":"","what":"Compute the arithmetic mean along the specified axes. — op_mean","title":"Compute the arithmetic mean along the specified axes. — op_mean","text":"Compute arithmetic mean along specified axes.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_mean.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Compute the arithmetic mean along the specified axes. — op_mean","text":"","code":"op_mean(x, axis = NULL, keepdims = FALSE)"},{"path":"https://keras3.posit.co/dev/reference/op_mean.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Compute the arithmetic mean along the specified axes. — op_mean","text":"x Input tensor. axis Axis axes along means computed. default compute mean flattened tensor. keepdims set TRUE, axes reduced left result dimensions size one.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_mean.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Compute the arithmetic mean along the specified axes. — op_mean","text":"Output tensor containing mean values.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_median.html","id":null,"dir":"Reference","previous_headings":"","what":"Compute the median along the specified axis. — op_median","title":"Compute the median along the specified axis. — op_median","text":"Compute median along specified axis.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_median.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Compute the median along the specified axis. — op_median","text":"","code":"op_median(x, axis = NULL, keepdims = FALSE)"},{"path":"https://keras3.posit.co/dev/reference/op_median.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Compute the median along the specified axis. — op_median","text":"x Input tensor. axis Axis axes along medians computed. Defaults axis = NULL compute median(s) along flattened version array. keepdims set TRUE, axes reduce left result dimensions size one.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_median.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Compute the median along the specified axis. — op_median","text":"output tensor.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_meshgrid.html","id":null,"dir":"Reference","previous_headings":"","what":"Creates grids of coordinates from coordinate vectors. — op_meshgrid","title":"Creates grids of coordinates from coordinate vectors. — op_meshgrid","text":"Given N 1-D tensors T0, T1, ..., TN-1 inputs corresponding lengths S0, S1, ..., SN-1, creates N N-dimensional tensors G0, G1, ..., GN-1 shape (S0, ..., SN-1) output Gi constructed expanding Ti result shape.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_meshgrid.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Creates grids of coordinates from coordinate vectors. — op_meshgrid","text":"","code":"op_meshgrid(..., indexing = \"xy\")"},{"path":"https://keras3.posit.co/dev/reference/op_meshgrid.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Creates grids of coordinates from coordinate vectors. — op_meshgrid","text":"... 1-D tensors representing coordinates grid. indexing \"xy\" \"ij\". \"xy\" cartesian; \"ij\" matrix indexing output. Defaults \"xy\".","code":""},{"path":"https://keras3.posit.co/dev/reference/op_meshgrid.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Creates grids of coordinates from coordinate vectors. — op_meshgrid","text":"Sequence N tensors.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_meshgrid.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Creates grids of coordinates from coordinate vectors. — op_meshgrid","text":"","code":"x <- op_array(c(1, 2, 3), \"int32\") y <- op_array(c(4, 5, 6), \"int32\") c(grid_x, grid_y) %<-% op_meshgrid(x, y, indexing = \"ij\") grid_x ## tf.Tensor( ## [[1 1 1] ##  [2 2 2] ##  [3 3 3]], shape=(3, 3), dtype=int32) # array([[1, 1, 1], #        [2, 2, 2], #        [3, 3, 3])) grid_y ## tf.Tensor( ## [[4 5 6] ##  [4 5 6] ##  [4 5 6]], shape=(3, 3), dtype=int32) # array([[4, 5, 6], #        [4, 5, 6], #        [4, 5, 6]))"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_min.html","id":null,"dir":"Reference","previous_headings":"","what":"Return the minimum of a tensor or minimum along an axis. — op_min","title":"Return the minimum of a tensor or minimum along an axis. — op_min","text":"Return minimum tensor minimum along axis.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_min.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Return the minimum of a tensor or minimum along an axis. — op_min","text":"","code":"op_min(x, axis = NULL, keepdims = FALSE, initial = NULL)"},{"path":"https://keras3.posit.co/dev/reference/op_min.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Return the minimum of a tensor or minimum along an axis. — op_min","text":"x Input tensor. axis Axis axes along operate. default, flattened input used. keepdims set TRUE, axes reduced left result dimensions size one. Defaults FALSE. initial maximum value output element. Defaults NULL.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_min.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Return the minimum of a tensor or minimum along an axis. — op_min","text":"Minimum x.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_min.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Return the minimum of a tensor or minimum along an axis. — op_min","text":"","code":"(x <- op_convert_to_tensor(rbind(c(1, 3, 5), c(1, 5, 2)))) ## tf.Tensor( ## [[1. 3. 5.] ##  [1. 5. 2.]], shape=(2, 3), dtype=float64) op_min(x) ## tf.Tensor(1.0, shape=(), dtype=float64) op_min(x, axis = 1) ## tf.Tensor([1. 3. 2.], shape=(3), dtype=float64) op_min(x, axis = 1, keepdims = TRUE) ## tf.Tensor([[1. 3. 2.]], shape=(1, 3), dtype=float64)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_minimum.html","id":null,"dir":"Reference","previous_headings":"","what":"Element-wise minimum of x1 and x2. — op_minimum","title":"Element-wise minimum of x1 and x2. — op_minimum","text":"Element-wise minimum x1 x2.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_minimum.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Element-wise minimum of x1 and x2. — op_minimum","text":"","code":"op_minimum(x1, x2)  op_pmin(x1, x2)"},{"path":"https://keras3.posit.co/dev/reference/op_minimum.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Element-wise minimum of x1 and x2. — op_minimum","text":"x1 First tensor. x2 Second tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_minimum.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Element-wise minimum of x1 and x2. — op_minimum","text":"Output tensor, element-wise minimum x1 x2.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_mod.html","id":null,"dir":"Reference","previous_headings":"","what":"Returns the element-wise remainder of division. — op_mod","title":"Returns the element-wise remainder of division. — op_mod","text":"Note function automatically called using R operator %% tensor.","code":"(x <- op_arange(10)) ## tf.Tensor([ 1.  2.  3.  4.  5.  6.  7.  8.  9. 10.], shape=(10), dtype=float32) op_mod(x, 3) ## tf.Tensor([1. 2. 0. 1. 2. 0. 1. 2. 0. 1.], shape=(10), dtype=float32) x %% 3 ## tf.Tensor([1. 2. 0. 1. 2. 0. 1. 2. 0. 1.], shape=(10), dtype=float32)"},{"path":"https://keras3.posit.co/dev/reference/op_mod.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Returns the element-wise remainder of division. — op_mod","text":"","code":"op_mod(x1, x2)"},{"path":"https://keras3.posit.co/dev/reference/op_mod.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Returns the element-wise remainder of division. — op_mod","text":"x1 First tensor. x2 Second tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_mod.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Returns the element-wise remainder of division. — op_mod","text":"Output tensor, element-wise remainder division.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_moments.html","id":null,"dir":"Reference","previous_headings":"","what":"Calculates the mean and variance of x. — op_moments","title":"Calculates the mean and variance of x. — op_moments","text":"mean variance calculated aggregating contents x across axes. x 1-D axes = c(1) just mean variance vector.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_moments.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Calculates the mean and variance of x. — op_moments","text":"","code":"op_moments(x, axes, keepdims = FALSE, synchronized = FALSE)"},{"path":"https://keras3.posit.co/dev/reference/op_moments.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Calculates the mean and variance of x. — op_moments","text":"x Input tensor. axes list axes compute mean variance. keepdims set TRUE, axes reduced left result dimensions size one. synchronized applicable TensorFlow backend. TRUE, synchronizes global batch statistics (mean variance) across devices training step distributed training strategy. FALSE, replica uses local batch statistics.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_moments.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Calculates the mean and variance of x. — op_moments","text":"list containing two tensors - mean variance.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_moments.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Calculates the mean and variance of x. — op_moments","text":"","code":"x <- op_convert_to_tensor(c(0, 1, 2, 3, 100), dtype = \"float32\") op_moments(x, axes = c(1)) ## [[1]] ## tf.Tensor(21.2, shape=(), dtype=float32) ## ## [[2]] ## tf.Tensor(1553.3601, shape=(), dtype=float32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_moveaxis.html","id":null,"dir":"Reference","previous_headings":"","what":"Move axes of a tensor to new positions. — op_moveaxis","title":"Move axes of a tensor to new positions. — op_moveaxis","text":"axes remain original order.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_moveaxis.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Move axes of a tensor to new positions. — op_moveaxis","text":"","code":"op_moveaxis(x, source, destination)"},{"path":"https://keras3.posit.co/dev/reference/op_moveaxis.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Move axes of a tensor to new positions. — op_moveaxis","text":"x Tensor whose axes reordered. source Original positions axes move. must unique. destination Destinations positions original axes. must also unique.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_moveaxis.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Move axes of a tensor to new positions. — op_moveaxis","text":"Tensor moved axes.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_multi_hot.html","id":null,"dir":"Reference","previous_headings":"","what":"Encodes integer labels as multi-hot vectors. — op_multi_hot","title":"Encodes integer labels as multi-hot vectors. — op_multi_hot","text":"function encodes integer labels multi-hot vectors, label mapped binary value resulting vector.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_multi_hot.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Encodes integer labels as multi-hot vectors. — op_multi_hot","text":"","code":"op_multi_hot(   inputs,   num_classes,   axis = -1L,   dtype = NULL,   sparse = FALSE,   ... )"},{"path":"https://keras3.posit.co/dev/reference/op_multi_hot.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Encodes integer labels as multi-hot vectors. — op_multi_hot","text":"inputs Tensor integer labels converted multi-hot vectors. num_classes Integer, total number unique classes. axis (optional) Axis along multi-hot encoding added. Defaults -1, corresponds last dimension. dtype (optional) data type resulting tensor. Default backend's float type. sparse Whether return sparse tensor; backends support sparse tensors. ... forward/backwards compatability","code":""},{"path":"https://keras3.posit.co/dev/reference/op_multi_hot.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Encodes integer labels as multi-hot vectors. — op_multi_hot","text":"Tensor: multi-hot encoded tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_multi_hot.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Encodes integer labels as multi-hot vectors. — op_multi_hot","text":"","code":"data <- op_convert_to_tensor(c(0, 4)) op_multi_hot(data, num_classes = 5) ## tf.Tensor([1. 0. 0. 0. 1.], shape=(5), dtype=float32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_multiply.html","id":null,"dir":"Reference","previous_headings":"","what":"Multiply arguments element-wise. — op_multiply","title":"Multiply arguments element-wise. — op_multiply","text":"Note function automatically called using R operator * tensor.","code":"(x <- op_arange(4)) ## tf.Tensor([1. 2. 3. 4.], shape=(4), dtype=float32) op_multiply(x, x) ## tf.Tensor([ 1.  4.  9. 16.], shape=(4), dtype=float32) x * x ## tf.Tensor([ 1.  4.  9. 16.], shape=(4), dtype=float32)"},{"path":"https://keras3.posit.co/dev/reference/op_multiply.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Multiply arguments element-wise. — op_multiply","text":"","code":"op_multiply(x1, x2)"},{"path":"https://keras3.posit.co/dev/reference/op_multiply.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Multiply arguments element-wise. — op_multiply","text":"x1 First input tensor. x2 Second input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_multiply.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Multiply arguments element-wise. — op_multiply","text":"Output tensor, element-wise product x1 x2.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_nan_to_num.html","id":null,"dir":"Reference","previous_headings":"","what":"Replace NaN with zero and infinity with large finite numbers. — op_nan_to_num","title":"Replace NaN with zero and infinity with large finite numbers. — op_nan_to_num","text":"Replace NaN zero infinity large finite numbers.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_nan_to_num.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Replace NaN with zero and infinity with large finite numbers. — op_nan_to_num","text":"","code":"op_nan_to_num(x, nan = 0, posinf = NULL, neginf = NULL)"},{"path":"https://keras3.posit.co/dev/reference/op_nan_to_num.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Replace NaN with zero and infinity with large finite numbers. — op_nan_to_num","text":"x Input data. nan Optional float int. Value replace NaN entries . posinf Optional float int. Value replace positive infinity . neginf Optional float int. Value replace negative infinity .","code":""},{"path":"https://keras3.posit.co/dev/reference/op_nan_to_num.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Replace NaN with zero and infinity with large finite numbers. — op_nan_to_num","text":"x, non-finite values replaced.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_nan_to_num.html","id":"example","dir":"Reference","previous_headings":"","what":"Example","title":"Replace NaN with zero and infinity with large finite numbers. — op_nan_to_num","text":"","code":"(x <- op_convert_to_tensor(c(1, NaN, -Inf, Inf))) ## tf.Tensor([  1.  nan -inf  inf], shape=(4), dtype=float64) op_nan_to_num(x) ## tf.Tensor([ 1.00000000e+000  0.00000000e+000 -1.79769313e+308  1.79769313e+308], shape=(4), dtype=float64) op_nan_to_num(x, nan = -1, posinf = 2, neginf = -2) ## tf.Tensor([ 1. -1. -2.  2.], shape=(4), dtype=float64)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_ndim.html","id":null,"dir":"Reference","previous_headings":"","what":"Return the number of dimensions of a tensor. — op_ndim","title":"Return the number of dimensions of a tensor. — op_ndim","text":"Return number dimensions tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_ndim.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Return the number of dimensions of a tensor. — op_ndim","text":"","code":"op_ndim(x)"},{"path":"https://keras3.posit.co/dev/reference/op_ndim.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Return the number of dimensions of a tensor. — op_ndim","text":"x Input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_ndim.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Return the number of dimensions of a tensor. — op_ndim","text":"number dimensions x.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_negative.html","id":null,"dir":"Reference","previous_headings":"","what":"Numerical negative, element-wise. — op_negative","title":"Numerical negative, element-wise. — op_negative","text":"Note function automatically called using unary R operator - tensor.","code":"(x <- op_arange(4)) ## tf.Tensor([1. 2. 3. 4.], shape=(4), dtype=float32) op_negative(x) ## tf.Tensor([-1. -2. -3. -4.], shape=(4), dtype=float32) -x ## tf.Tensor([-1. -2. -3. -4.], shape=(4), dtype=float32)"},{"path":"https://keras3.posit.co/dev/reference/op_negative.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Numerical negative, element-wise. — op_negative","text":"","code":"op_negative(x)"},{"path":"https://keras3.posit.co/dev/reference/op_negative.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Numerical negative, element-wise. — op_negative","text":"x Input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_negative.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Numerical negative, element-wise. — op_negative","text":"Output tensor, y = -x.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_nonzero.html","id":null,"dir":"Reference","previous_headings":"","what":"Return the indices of the elements that are non-zero. — op_nonzero","title":"Return the indices of the elements that are non-zero. — op_nonzero","text":"Return indices elements non-zero.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_nonzero.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Return the indices of the elements that are non-zero. — op_nonzero","text":"","code":"op_nonzero(x)"},{"path":"https://keras3.posit.co/dev/reference/op_nonzero.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Return the indices of the elements that are non-zero. — op_nonzero","text":"x Input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_nonzero.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Return the indices of the elements that are non-zero. — op_nonzero","text":"Indices elements non-zero.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_nonzero.html","id":"example","dir":"Reference","previous_headings":"","what":"Example","title":"Return the indices of the elements that are non-zero. — op_nonzero","text":"op_nonzero() indices can used <tensor>@r[","code":"(x <- op_scatter(indices = rbind(1, 5, 10), values = c(1, 2, 3), shape = c(10))) ## tf.Tensor([1. 0. 0. 0. 2. 0. 0. 0. 0. 3.], shape=(10), dtype=float64) (nz <- op_nonzero(x)) ## tf.Tensor( ## [[ 1] ##  [ 5] ##  [10]], shape=(3, 1), dtype=int32) x@r[nz] ## tf.Tensor([1. 2. 3.], shape=(3), dtype=float64) # same as `x@r[nz]` x@r[x != 0] ## tf.Tensor([1. 2. 3.], shape=(3), dtype=float64) x@r[op_cast(x, \"bool\")] ## tf.Tensor([1. 2. 3.], shape=(3), dtype=float64) # 2d example (x2 <- op_stack(c(x, op_roll(x, 1), op_roll(x, 2)))) ## tf.Tensor( ## [[1. 0. 0. 0. 2. 0. 0. 0. 0. 3.] ##  [3. 1. 0. 0. 0. 2. 0. 0. 0. 0.] ##  [0. 3. 1. 0. 0. 0. 2. 0. 0. 0.]], shape=(3, 10), dtype=float64) x2@r[op_nonzero(x2)] ## tf.Tensor([1. 2. 3. 3. 1. 2. 3. 1. 2.], shape=(9), dtype=float64) x3 <- op_stack(c(x2, x2*1.1)) x3@r[op_nonzero(x3)] ## tf.Tensor([1.  2.  3.  3.  1.  2.  3.  1.  2.  1.1 2.2 3.3 3.3 1.1 2.2 3.3 1.1 2.2], shape=(18), dtype=float64)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_norm.html","id":null,"dir":"Reference","previous_headings":"","what":"Matrix or vector norm. — op_norm","title":"Matrix or vector norm. — op_norm","text":"function able return one eight different matrix norms, one infinite number vector norms (described ), depending value ord parameter.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_norm.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Matrix or vector norm. — op_norm","text":"","code":"op_norm(x, ord = NULL, axis = NULL, keepdims = FALSE)"},{"path":"https://keras3.posit.co/dev/reference/op_norm.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Matrix or vector norm. — op_norm","text":"x Input tensor. ord Order norm (see table Notes). default NULL. axis axis integer, specifies axis x along compute vector norms. axis length 2 vector, specifies axes hold 2-D matrices, matrix norms matrices computed. keepdims set TRUE, axes reduced left result dimensions size one.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_norm.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Matrix or vector norm. — op_norm","text":"Norm matrix vector(s).","code":""},{"path":"https://keras3.posit.co/dev/reference/op_norm.html","id":"note","dir":"Reference","previous_headings":"","what":"Note","title":"Matrix or vector norm. — op_norm","text":"values ord < 1, result , strictly speaking, mathematical 'norm', may still useful various numerical purposes. following norms can calculated: matrices: ord=NULL: Frobenius norm ord=\"fro\": Frobenius norm ord=\"nuc\": nuclear norm ord=Inf: max(sum(abs(x), axis=2)) ord=-Inf: min(sum(abs(x), axis=2)) ord=0: supported ord=1: max(sum(abs(x), axis=1)) ord=-1: min(sum(abs(x), axis=1)) ord=2: 2-norm (largest sing. value) ord=-2: smallest singular value : supported vectors: ord=NULL: 2-norm ord=\"fro\": supported ord=\"nuc\": supported ord=Inf: max(abs(x)) ord=-Inf: min(abs(x)) ord=0: sum(x != 0) ord=1: ord=-1: ord=2: ord=-2: : sum(abs(x)^ord)^(1/ord)","code":""},{"path":"https://keras3.posit.co/dev/reference/op_norm.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Matrix or vector norm. — op_norm","text":"","code":"x <- op_reshape(op_arange(9, dtype=\"float32\") - 4, c(3, 3)) op_norm(x) ## tf.Tensor(8.306623, shape=(), dtype=float32) # 7.7459664"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_normalize.html","id":null,"dir":"Reference","previous_headings":"","what":"Normalizes x over the specified axis. — op_normalize","title":"Normalizes x over the specified axis. — op_normalize","text":"defined : op_normalize(x) = x / max(norm(x), epsilon).","code":""},{"path":"https://keras3.posit.co/dev/reference/op_normalize.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Normalizes x over the specified axis. — op_normalize","text":"","code":"op_normalize(x, axis = -1L, order = 2L, epsilon = NULL)"},{"path":"https://keras3.posit.co/dev/reference/op_normalize.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Normalizes x over the specified axis. — op_normalize","text":"x Input tensor. axis axis axes along perform normalization. Default -1. order exponent value norm formulation. Defaults 2. epsilon lower bound value norm. Defaults config_epsilon().","code":""},{"path":"https://keras3.posit.co/dev/reference/op_normalize.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Normalizes x over the specified axis. — op_normalize","text":"normalized array.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_normalize.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Normalizes x over the specified axis. — op_normalize","text":"","code":"x <- op_convert_to_tensor(rbind(c(1, 2, 3), c(4, 5, 6))) x_norm <- op_normalize(x) x_norm ## tf.Tensor( ## [[0.26726124 0.53452248 0.80178373] ##  [0.45584231 0.56980288 0.68376346]], shape=(2, 3), dtype=float64)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_not_equal.html","id":null,"dir":"Reference","previous_headings":"","what":"Return (x1 != x2) element-wise. — op_not_equal","title":"Return (x1 != x2) element-wise. — op_not_equal","text":"Note function automatically called using R operator != tensor.","code":"(x <- op_arange(4)) ## tf.Tensor([1. 2. 3. 4.], shape=(4), dtype=float32) op_not_equal(x, 2) ## tf.Tensor([ True False  True  True], shape=(4), dtype=bool) x != 2 ## tf.Tensor([ True False  True  True], shape=(4), dtype=bool)"},{"path":"https://keras3.posit.co/dev/reference/op_not_equal.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Return (x1 != x2) element-wise. — op_not_equal","text":"","code":"op_not_equal(x1, x2)"},{"path":"https://keras3.posit.co/dev/reference/op_not_equal.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Return (x1 != x2) element-wise. — op_not_equal","text":"x1 First input tensor. x2 Second input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_not_equal.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Return (x1 != x2) element-wise. — op_not_equal","text":"Output tensor, element-wise comparison x1 x2.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_one_hot.html","id":null,"dir":"Reference","previous_headings":"","what":"Converts integer tensor x into a one-hot tensor. — op_one_hot","title":"Converts integer tensor x into a one-hot tensor. — op_one_hot","text":"one-hot encoding representation integer value converted binary vector length equal num_classes, index corresponding integer value marked 1, indices marked 0.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_one_hot.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Converts integer tensor x into a one-hot tensor. — op_one_hot","text":"","code":"op_one_hot(   x,   num_classes,   axis = -1L,   dtype = NULL,   sparse = FALSE,   zero_indexed = FALSE )"},{"path":"https://keras3.posit.co/dev/reference/op_one_hot.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Converts integer tensor x into a one-hot tensor. — op_one_hot","text":"x Integer tensor encoded. shape can arbitrary, dtype integer. R factors coerced integer. num_classes Number classes one-hot encoding. x factor num_classes NULL missing, levels(x) used. axis Axis along encoding performed. -1 represents last axis. Defaults -1. dtype (Optional) Data type output tensor. provided, defaults default data type backend. sparse Whether return sparse tensor; backends support sparse tensors. zero_indexed TRUE, treats indices zero-based (0 encodes first position); FALSE (default), treats indices one-based (1 encodes first position).","code":""},{"path":"https://keras3.posit.co/dev/reference/op_one_hot.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Converts integer tensor x into a one-hot tensor. — op_one_hot","text":"Integer tensor: One-hot encoded tensor shape x except specified axis dimension, length num_classes. dtype output tensor determined dtype default data type backend.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_one_hot.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Converts integer tensor x into a one-hot tensor. — op_one_hot","text":"","code":"x <- op_array(c(2, 4, 3, 1), \"int32\") op_one_hot(x, num_classes = 4) ## tf.Tensor( ## [[0. 1. 0. 0.] ##  [0. 0. 0. 1.] ##  [0. 0. 1. 0.] ##  [1. 0. 0. 0.]], shape=(4, 4), dtype=float32) op_one_hot(x - 1, num_classes = 4, zero_indexed = TRUE) ## tf.Tensor( ## [[0. 1. 0. 0.] ##  [0. 0. 0. 1.] ##  [0. 0. 1. 0.] ##  [1. 0. 0. 0.]], shape=(4, 4), dtype=float32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_ones.html","id":null,"dir":"Reference","previous_headings":"","what":"Return a new tensor of given shape and type, filled with ones. — op_ones","title":"Return a new tensor of given shape and type, filled with ones. — op_ones","text":"Return new tensor given shape type, filled ones.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_ones.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Return a new tensor of given shape and type, filled with ones. — op_ones","text":"","code":"op_ones(shape, dtype = NULL)"},{"path":"https://keras3.posit.co/dev/reference/op_ones.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Return a new tensor of given shape and type, filled with ones. — op_ones","text":"shape Shape new tensor. dtype Desired data type tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_ones.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Return a new tensor of given shape and type, filled with ones. — op_ones","text":"Tensor ones given shape dtype.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_ones_like.html","id":null,"dir":"Reference","previous_headings":"","what":"Return a tensor of ones with the same shape and type of x. — op_ones_like","title":"Return a tensor of ones with the same shape and type of x. — op_ones_like","text":"Return tensor ones shape type x.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_ones_like.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Return a tensor of ones with the same shape and type of x. — op_ones_like","text":"","code":"op_ones_like(x, dtype = NULL)"},{"path":"https://keras3.posit.co/dev/reference/op_ones_like.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Return a tensor of ones with the same shape and type of x. — op_ones_like","text":"x Input tensor. dtype Overrides data type result.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_ones_like.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Return a tensor of ones with the same shape and type of x. — op_ones_like","text":"tensor ones shape type x.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_outer.html","id":null,"dir":"Reference","previous_headings":"","what":"Compute the outer product of two vectors. — op_outer","title":"Compute the outer product of two vectors. — op_outer","text":"Given two vectors x1 x2, outer product :","code":"out[i, j] = x1[i] * x2[j]"},{"path":"https://keras3.posit.co/dev/reference/op_outer.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Compute the outer product of two vectors. — op_outer","text":"","code":"op_outer(x1, x2)"},{"path":"https://keras3.posit.co/dev/reference/op_outer.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Compute the outer product of two vectors. — op_outer","text":"x1 First input tensor. x2 Second input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_outer.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Compute the outer product of two vectors. — op_outer","text":"Outer product x1 x2.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_pad.html","id":null,"dir":"Reference","previous_headings":"","what":"Pad a tensor. — op_pad","title":"Pad a tensor. — op_pad","text":"Pad tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_pad.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Pad a tensor. — op_pad","text":"","code":"op_pad(x, pad_width, mode = \"constant\", constant_values = NULL)"},{"path":"https://keras3.posit.co/dev/reference/op_pad.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Pad a tensor. — op_pad","text":"x Tensor pad. pad_width Number values padded edges axis. ((before_1, after_1), ...(before_N, after_N)) unique pad widths axis. ((, ),) yields pad axis. (pad,) int shortcut = = pad width axes. mode One \"constant\", \"edge\", \"linear_ramp\", \"maximum\", \"mean\", \"median\", \"minimum\", \"reflect\", \"symmetric\", \"wrap\", \"empty\", \"circular\". Defaults \"constant\". constant_values Value pad mode == \"constant\". Defaults 0. ValueError raised NULL mode != \"constant\".","code":""},{"path":"https://keras3.posit.co/dev/reference/op_pad.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Pad a tensor. — op_pad","text":"Padded tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_pad.html","id":"note","dir":"Reference","previous_headings":"","what":"Note","title":"Pad a tensor. — op_pad","text":"Torch backend supports modes \"constant\", \"reflect\", \"symmetric\" \"circular\". Torch backend supports \"circular\" mode. Note: Tensorflow backend supports modes \"constant\", \"reflect\" \"symmetric\".","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_polar.html","id":null,"dir":"Reference","previous_headings":"","what":"Constructs a complex tensor whose elements are Cartesian — op_polar","title":"Constructs a complex tensor whose elements are Cartesian — op_polar","text":"coordinates corresponding polar coordinates absolute value abs angle angle. operation numerically equivalent torch.polar(). equivalent scipy.lingalg.polar() performs Singular Value Decomposition. Given magnitude (abs_) angle (angle), function computes corresponding complex number form real + imaginary * 1i, : real = abs_ * cos(angle) imaginary = abs_ * sin(angle)","code":""},{"path":"https://keras3.posit.co/dev/reference/op_polar.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Constructs a complex tensor whose elements are Cartesian — op_polar","text":"","code":"op_polar(abs_, angle)"},{"path":"https://keras3.posit.co/dev/reference/op_polar.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Constructs a complex tensor whose elements are Cartesian — op_polar","text":"abs_ magnitude (absolute value) complex number. angle angle (radians) complex number.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_polar.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Constructs a complex tensor whose elements are Cartesian — op_polar","text":"complex number (array complex numbers) shape abs_ angle.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_polar.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Constructs a complex tensor whose elements are Cartesian — op_polar","text":"","code":"abs_ <- random_normal(c(1, 2)) angle <- random_normal(c(1, 2)) op_shape(op_polar(abs_, angle)) ## shape(1, 2) op_polar(abs_, angle) ## tf.Tensor([[-0.10252866-1.2460487j -0.2586355 -1.0352845j]], shape=(1, 2), dtype=complex64)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_power.html","id":null,"dir":"Reference","previous_headings":"","what":"First tensor elements raised to powers from second tensor, element-wise. — op_power","title":"First tensor elements raised to powers from second tensor, element-wise. — op_power","text":"Note function automatically called using R operator ^ tensor.","code":"(x <- op_arange(4)) ## tf.Tensor([1. 2. 3. 4.], shape=(4), dtype=float32) op_power(2, x) ## tf.Tensor([ 2.  4.  8. 16.], shape=(4), dtype=float32) 2 ^ x ## tf.Tensor([ 2.  4.  8. 16.], shape=(4), dtype=float32)"},{"path":"https://keras3.posit.co/dev/reference/op_power.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"First tensor elements raised to powers from second tensor, element-wise. — op_power","text":"","code":"op_power(x1, x2)"},{"path":"https://keras3.posit.co/dev/reference/op_power.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"First tensor elements raised to powers from second tensor, element-wise. — op_power","text":"x1 bases. x2 exponents.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_power.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"First tensor elements raised to powers from second tensor, element-wise. — op_power","text":"Output tensor, bases x1 raised exponents x2.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_prod.html","id":null,"dir":"Reference","previous_headings":"","what":"Return the product of tensor elements over a given axis. — op_prod","title":"Return the product of tensor elements over a given axis. — op_prod","text":"Return product tensor elements given axis.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_prod.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Return the product of tensor elements over a given axis. — op_prod","text":"","code":"op_prod(x, axis = NULL, keepdims = FALSE, dtype = NULL)"},{"path":"https://keras3.posit.co/dev/reference/op_prod.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Return the product of tensor elements over a given axis. — op_prod","text":"x Input tensor. axis Axis axes along product performed. default, axis = NULL, compute product elements input tensor. keepdims set TRUE, axes reduce left result dimensions size one. dtype Data type returned tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_prod.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Return the product of tensor elements over a given axis. — op_prod","text":"Product elements x given axis axes.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_psnr.html","id":null,"dir":"Reference","previous_headings":"","what":"Peak Signal-to-Noise Ratio (PSNR) function. — op_psnr","title":"Peak Signal-to-Noise Ratio (PSNR) function. — op_psnr","text":"function computes Peak Signal--Noise Ratio two signals, x1 x2. PSNR measure quality reconstructed signal. higher PSNR, closer reconstructed signal original signal. Note can become negative signal power smaller noise power.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_psnr.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Peak Signal-to-Noise Ratio (PSNR) function. — op_psnr","text":"","code":"op_psnr(x1, x2, max_val)"},{"path":"https://keras3.posit.co/dev/reference/op_psnr.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Peak Signal-to-Noise Ratio (PSNR) function. — op_psnr","text":"x1 first input signal. x2 second input signal. Must shape x1. max_val maximum possible value signals.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_psnr.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Peak Signal-to-Noise Ratio (PSNR) function. — op_psnr","text":"float: PSNR value x1 x2.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_psnr.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Peak Signal-to-Noise Ratio (PSNR) function. — op_psnr","text":"","code":"x1 <- random_normal(c(2, 4, 4, 3)) x2 <- random_normal(c(2, 4, 4, 3)) max_val <- 1.0 op_psnr(x1, x2, max_val) ## tf.Tensor(-3.5293808, shape=(), dtype=float32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_qr.html","id":null,"dir":"Reference","previous_headings":"","what":"Computes the QR decomposition of a tensor. — op_qr","title":"Computes the QR decomposition of a tensor. — op_qr","text":"Computes QR decomposition tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_qr.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes the QR decomposition of a tensor. — op_qr","text":"","code":"op_qr(x, mode = \"reduced\")"},{"path":"https://keras3.posit.co/dev/reference/op_qr.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Computes the QR decomposition of a tensor. — op_qr","text":"x Input tensor shape (..., M, N). mode string specifying mode QR decomposition. 'reduced': Returns reduced QR decomposition. (default) 'complete': Returns complete QR decomposition.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_qr.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Computes the QR decomposition of a tensor. — op_qr","text":"list containing two tensors. first tensor shape (..., M, K) orthogonal matrix q second tensor shape (..., K, N)upper triangular matrixr, K = min(M, N)`.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_qr.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Computes the QR decomposition of a tensor. — op_qr","text":"","code":"x <- op_convert_to_tensor(rbind(c(1, 2), c(3, 4), c(5, 6))) op_qr(x) ## $q ## tf.Tensor( ## [[-0.16903085  0.89708523] ##  [-0.50709255  0.27602622] ##  [-0.84515425 -0.34503278]], shape=(3, 2), dtype=float64) ## ## $r ## tf.Tensor( ## [[-5.91607978 -7.43735744] ##  [ 0.          0.82807867]], shape=(2, 2), dtype=float64) c(q, r) %<-% op_qr(x)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_quantile.html","id":null,"dir":"Reference","previous_headings":"","what":"Compute the q-th quantile(s) of the data along the specified axis. — op_quantile","title":"Compute the q-th quantile(s) of the data along the specified axis. — op_quantile","text":"Compute q-th quantile(s) data along specified axis.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_quantile.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Compute the q-th quantile(s) of the data along the specified axis. — op_quantile","text":"","code":"op_quantile(x, q, axis = NULL, method = \"linear\", keepdims = FALSE)"},{"path":"https://keras3.posit.co/dev/reference/op_quantile.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Compute the q-th quantile(s) of the data along the specified axis. — op_quantile","text":"x Input tensor. q Probability sequence probabilities quantiles compute. Values must 0 1 inclusive. axis Axis axes along quantiles computed. Defaults axis=NULL compute quantile(s) along flattened version array. method string specifies method use estimating quantile. Available methods \"linear\", \"lower\", \"higher\", \"midpoint\", \"nearest\". Defaults \"linear\". desired quantile lies two data points < j: \"linear\": + (j - ) * fraction, fraction fractional part index surrounded j. \"lower\": . \"higher\": j. \"midpoint\": (+ j) / 2 \"nearest\": j, whichever nearest. keepdims set TRUE, axes reduce left result dimensions size one.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_quantile.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Compute the q-th quantile(s) of the data along the specified axis. — op_quantile","text":"quantile(s). q single probability axis = NULL, result scalar. multiple probabilities levels given, first axis result corresponds quantiles. axes axes remain reduction x.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_ravel.html","id":null,"dir":"Reference","previous_headings":"","what":"Return a contiguous flattened tensor. — op_ravel","title":"Return a contiguous flattened tensor. — op_ravel","text":"1-D tensor, containing elements input, returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_ravel.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Return a contiguous flattened tensor. — op_ravel","text":"","code":"op_ravel(x)"},{"path":"https://keras3.posit.co/dev/reference/op_ravel.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Return a contiguous flattened tensor. — op_ravel","text":"x Input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_ravel.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Return a contiguous flattened tensor. — op_ravel","text":"Output tensor.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_real.html","id":null,"dir":"Reference","previous_headings":"","what":"Return the real part of the complex argument. — op_real","title":"Return the real part of the complex argument. — op_real","text":"Return real part complex argument.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_real.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Return the real part of the complex argument. — op_real","text":"","code":"op_real(x)"},{"path":"https://keras3.posit.co/dev/reference/op_real.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Return the real part of the complex argument. — op_real","text":"x Input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_real.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Return the real part of the complex argument. — op_real","text":"real component complex argument.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_rearrange.html","id":null,"dir":"Reference","previous_headings":"","what":"Rearranges the axes of a Keras tensor according to a specified pattern, — op_rearrange","title":"Rearranges the axes of a Keras tensor according to a specified pattern, — op_rearrange","text":"einops-style.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_rearrange.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Rearranges the axes of a Keras tensor according to a specified pattern, — op_rearrange","text":"","code":"op_rearrange(tensor, pattern, ...)"},{"path":"https://keras3.posit.co/dev/reference/op_rearrange.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Rearranges the axes of a Keras tensor according to a specified pattern, — op_rearrange","text":"tensor Input Keras tensor. pattern String describing rearrangement einops notation. ... axes_lengths, named arguments specifying lengths axes axes decomposition used.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_rearrange.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Rearranges the axes of a Keras tensor according to a specified pattern, — op_rearrange","text":"Tensor: Keras tensor rearranged axes. Follows logic : decomposition needed, reshape match decomposed dimensions. Permute known inferred axes match form output. Reshape match desired output shape. Example Usage:","code":"images <- op_ones(c(32, 30, 40, 3)) # BHWC format  # Reordering to BCHW op_rearrange(images, 'b h w c -> b c h w') |> op_shape() ## shape(32, 3, 30, 40) # \"Merge\" along first axis - concat images from a batch op_rearrange(images, 'b h w c -> (b h) w c') |> op_shape() ## shape(960, 40, 3) # \"Merge\" along second axis - concat images horizontally op_rearrange(images, 'b h w c -> h (b w) c') |> op_shape() ## shape(30, 1280, 3) # Flatten images into a CHW vector op_rearrange(images, 'b h w c -> b (c h w)') |> op_shape() ## shape(32, 3600) # Decompose H and W axes into 4 smaller patches op_rearrange(images, 'b (h1 h) (w1 w) c -> (b h1 w1) h w c', h1 = 2, w1 = 2) |> op_shape() ## shape(128, 15, 20, 3) # Space-to-depth decomposition of input axes op_rearrange(images, 'b (h h1) (w w1) c -> b h w (c h1 w1)', h1 = 2, w1 = 2) |> op_shape() ## shape(32, 15, 20, 12)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_reciprocal.html","id":null,"dir":"Reference","previous_headings":"","what":"Return the reciprocal of the argument, element-wise. — op_reciprocal","title":"Return the reciprocal of the argument, element-wise. — op_reciprocal","text":"Calculates 1/x.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_reciprocal.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Return the reciprocal of the argument, element-wise. — op_reciprocal","text":"","code":"op_reciprocal(x)"},{"path":"https://keras3.posit.co/dev/reference/op_reciprocal.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Return the reciprocal of the argument, element-wise. — op_reciprocal","text":"x Input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_reciprocal.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Return the reciprocal of the argument, element-wise. — op_reciprocal","text":"Output tensor, element-wise reciprocal x.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_relu.html","id":null,"dir":"Reference","previous_headings":"","what":"Rectified linear unit activation function. — op_relu","title":"Rectified linear unit activation function. — op_relu","text":"defined f(x) = max(0, x).","code":""},{"path":"https://keras3.posit.co/dev/reference/op_relu.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Rectified linear unit activation function. — op_relu","text":"","code":"op_relu(x)"},{"path":"https://keras3.posit.co/dev/reference/op_relu.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Rectified linear unit activation function. — op_relu","text":"x Input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_relu.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Rectified linear unit activation function. — op_relu","text":"tensor shape x.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_relu.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Rectified linear unit activation function. — op_relu","text":"","code":"x1 <- op_convert_to_tensor(c(-1, 0, 1, 0.2)) op_relu(x1) ## tf.Tensor([0.  0.  1.  0.2], shape=(4), dtype=float64) x <- seq(-10, 10, .1) plot(x, op_relu(x))"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_relu6.html","id":null,"dir":"Reference","previous_headings":"","what":"Rectified linear unit activation function with upper bound of 6. — op_relu6","title":"Rectified linear unit activation function with upper bound of 6. — op_relu6","text":"defined f(x) = op_clip(x, 0, 6).","code":""},{"path":"https://keras3.posit.co/dev/reference/op_relu6.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Rectified linear unit activation function with upper bound of 6. — op_relu6","text":"","code":"op_relu6(x)"},{"path":"https://keras3.posit.co/dev/reference/op_relu6.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Rectified linear unit activation function with upper bound of 6. — op_relu6","text":"x Input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_relu6.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Rectified linear unit activation function with upper bound of 6. — op_relu6","text":"tensor shape x.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_relu6.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Rectified linear unit activation function with upper bound of 6. — op_relu6","text":"","code":"x <- op_convert_to_tensor(c(-3, -2, 0.1, 0.2, 6, 8)) op_relu6(x) ## tf.Tensor([0.  0.  0.1 0.2 6.  6. ], shape=(6), dtype=float64) x <- seq(-10, 10, .1) plot(x, op_relu6(x))"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_repeat.html","id":null,"dir":"Reference","previous_headings":"","what":"Repeat each element of a tensor after themselves. — op_repeat","title":"Repeat each element of a tensor after themselves. — op_repeat","text":"Repeat element tensor .","code":""},{"path":"https://keras3.posit.co/dev/reference/op_repeat.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Repeat each element of a tensor after themselves. — op_repeat","text":"","code":"op_repeat(x, repeats, axis = NULL)"},{"path":"https://keras3.posit.co/dev/reference/op_repeat.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Repeat each element of a tensor after themselves. — op_repeat","text":"x Input tensor. repeats number repetitions element. axis axis along repeat values. default, use flattened input array, return flat output array.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_repeat.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Repeat each element of a tensor after themselves. — op_repeat","text":"Output tensor.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_reshape.html","id":null,"dir":"Reference","previous_headings":"","what":"Gives a new shape to a tensor without changing its data. — op_reshape","title":"Gives a new shape to a tensor without changing its data. — op_reshape","text":"Gives new shape tensor without changing data.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_reshape.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Gives a new shape to a tensor without changing its data. — op_reshape","text":"","code":"op_reshape(x, newshape)"},{"path":"https://keras3.posit.co/dev/reference/op_reshape.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Gives a new shape to a tensor without changing its data. — op_reshape","text":"x Input tensor. newshape new shape compatible original shape. One shape dimension can -1 case value inferred length array remaining dimensions.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_reshape.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Gives a new shape to a tensor without changing its data. — op_reshape","text":"reshaped tensor.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_rfft.html","id":null,"dir":"Reference","previous_headings":"","what":"Real-valued Fast Fourier Transform along the last axis of the input. — op_rfft","title":"Real-valued Fast Fourier Transform along the last axis of the input. — op_rfft","text":"Computes 1D Discrete Fourier Transform real-valued signal inner-dimension input. Since Discrete Fourier Transform real-valued signal Hermitian-symmetric, RFFT returns fft_length / 2 + 1 unique components FFT: zero-frequency term, followed fft_length / 2 positive-frequency terms. Along axis RFFT computed , fft_length smaller corresponding dimension input, dimension cropped. larger, dimension padded zeros.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_rfft.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Real-valued Fast Fourier Transform along the last axis of the input. — op_rfft","text":"","code":"op_rfft(x, fft_length = NULL)"},{"path":"https://keras3.posit.co/dev/reference/op_rfft.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Real-valued Fast Fourier Transform along the last axis of the input. — op_rfft","text":"x Input tensor. fft_length integer representing number fft length. specified, inferred length last axis x. Defaults NULL.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_rfft.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Real-valued Fast Fourier Transform along the last axis of the input. — op_rfft","text":"list containing two tensors - real imaginary parts output.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_rfft.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Real-valued Fast Fourier Transform along the last axis of the input. — op_rfft","text":"","code":"x <- op_convert_to_tensor(c(0, 1, 2, 3, 4)) op_rfft(x) ## [[1]] ## tf.Tensor([10.  -2.5 -2.5], shape=(3), dtype=float64) ## ## [[2]] ## tf.Tensor([0.         3.4409548  0.81229924], shape=(3), dtype=float64) op_rfft(x, 3) ## [[1]] ## tf.Tensor([ 3.  -1.5], shape=(2), dtype=float64) ## ## [[2]] ## tf.Tensor([0.        0.8660254], shape=(2), dtype=float64)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_right_shift.html","id":null,"dir":"Reference","previous_headings":"","what":"Shift the bits of an integer to the right. — op_right_shift","title":"Shift the bits of an integer to the right. — op_right_shift","text":"Bits shifted right y. internal representation numbers binary format, operation equivalent dividing x 2^y.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_right_shift.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Shift the bits of an integer to the right. — op_right_shift","text":"","code":"op_right_shift(x, y)"},{"path":"https://keras3.posit.co/dev/reference/op_right_shift.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Shift the bits of an integer to the right. — op_right_shift","text":"x Input integer tensor. y Input integer tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_right_shift.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Shift the bits of an integer to the right. — op_right_shift","text":"Result tensor.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_rms_normalization.html","id":null,"dir":"Reference","previous_headings":"","what":"Performs Root Mean Square (RMS) normalization on x. — op_rms_normalization","title":"Performs Root Mean Square (RMS) normalization on x. — op_rms_normalization","text":"Keras operation implements operation described Root Mean Square Layer Normalization Biao Zhang et al. operation different LayerNormalization RMS scaling. defined rms_normalization(x) = x * rsqrt(mean(square(x))) * scale","code":""},{"path":"https://keras3.posit.co/dev/reference/op_rms_normalization.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Performs Root Mean Square (RMS) normalization on x. — op_rms_normalization","text":"","code":"op_rms_normalization(x, scale = NULL, axis = -1L, epsilon = NULL)"},{"path":"https://keras3.posit.co/dev/reference/op_rms_normalization.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Performs Root Mean Square (RMS) normalization on x. — op_rms_normalization","text":"x Input tensor. scale Optional scaling factor normalization. axis axis axes along perform normalization. Defaults -1. epsilon lower bound value norm. Defaults config_epsilon().","code":""},{"path":"https://keras3.posit.co/dev/reference/op_rms_normalization.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Performs Root Mean Square (RMS) normalization on x. — op_rms_normalization","text":"normalized array.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_rms_normalization.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Performs Root Mean Square (RMS) normalization on x. — op_rms_normalization","text":"[-1.1701705   0.9987454   0.27809578 -0.18452731  1.5731977   1.8934134","code":"x <- random_normal(c(1, 10)) op_rms_normalization(x) ## tf.Tensor( ## [[-1.1701705   0.9987454   0.27809578 -0.18452731  1.5731977   1.8934134 ##    0.10656697  1.0256069   0.6185074  -0.12678868]], shape=(1, 10), dtype=float32) op_rms_normalization(x, scale = 10) ## tf.Tensor( ## [[-11.701706    9.987453    2.7809577  -1.845273   15.731977   18.934134 ##     1.0656698  10.256069    6.185074   -1.2678868]], shape=(1, 10), dtype=float32)"},{"path":"https://keras3.posit.co/dev/reference/op_rms_normalization.html","id":"-r-a-","dir":"Reference","previous_headings":"","what":"0.10656697  1.0256069   0.6185074  -0.12678868]: R:-1.1701705%20%20%200.9987454%20%20%200.27809578%20-0.18452731%20%201.5731977%20%20%201.8934134%0A##%20%20%20%200.10656697%20%201.0256069%20%20%200.6185074%20%20-0.12678868","title":"Performs Root Mean Square (RMS) normalization on x. — op_rms_normalization","text":"[-11.701706    9.987453    2.7809577  -1.845273   15.731977   18.934134","code":""},{"path":[]},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_roll.html","id":null,"dir":"Reference","previous_headings":"","what":"Roll tensor elements along a given axis. — op_roll","title":"Roll tensor elements along a given axis. — op_roll","text":"Elements roll beyond last position re-introduced first.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_roll.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Roll tensor elements along a given axis. — op_roll","text":"","code":"op_roll(x, shift, axis = NULL)"},{"path":"https://keras3.posit.co/dev/reference/op_roll.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Roll tensor elements along a given axis. — op_roll","text":"x Input tensor. shift number places elements shifted. axis axis along elements shifted. default, array flattened shifting, original shape restored.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_roll.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Roll tensor elements along a given axis. — op_roll","text":"Output tensor.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_rot90.html","id":null,"dir":"Reference","previous_headings":"","what":"Rotate an array by 90 degrees in the plane specified by axes. — op_rot90","title":"Rotate an array by 90 degrees in the plane specified by axes. — op_rot90","text":"function rotates array counterclockwise 90 degrees k times plane specified axes. Supports arrays two dimensions.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_rot90.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Rotate an array by 90 degrees in the plane specified by axes. — op_rot90","text":"","code":"op_rot90(array, k = 1L, axes = list(1L, 2L))"},{"path":"https://keras3.posit.co/dev/reference/op_rot90.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Rotate an array by 90 degrees in the plane specified by axes. — op_rot90","text":"array Input array rotate. k Number times array rotated 90 degrees. axes tuple two integers specifying plane rotation (defaults (1, 2)).","code":""},{"path":"https://keras3.posit.co/dev/reference/op_rot90.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Rotate an array by 90 degrees in the plane specified by axes. — op_rot90","text":"Rotated array.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_rot90.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Rotate an array by 90 degrees in the plane specified by axes. — op_rot90","text":"","code":"m <- 1:4 |> op_reshape(c(2, 2)) m ## tf.Tensor( ## [[1 2] ##  [3 4]], shape=(2, 2), dtype=int32) op_rot90(m) ## tf.Tensor( ## [[2 4] ##  [1 3]], shape=(2, 2), dtype=int32) m <- 1:8 |> op_reshape(c(2, 2, 2)) m ## tf.Tensor( ## [[[1 2] ##   [3 4]] ## ##  [[5 6] ##   [7 8]]], shape=(2, 2, 2), dtype=int32) op_rot90(m, k = 1, axes = c(2, 3)) ## tf.Tensor( ## [[[2 4] ##   [1 3]] ## ##  [[6 8] ##   [5 7]]], shape=(2, 2, 2), dtype=int32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_round.html","id":null,"dir":"Reference","previous_headings":"","what":"Evenly round to the given number of decimals. — op_round","title":"Evenly round to the given number of decimals. — op_round","text":"Evenly round given number decimals.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_round.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Evenly round to the given number of decimals. — op_round","text":"","code":"op_round(x, decimals = 0L)"},{"path":"https://keras3.posit.co/dev/reference/op_round.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Evenly round to the given number of decimals. — op_round","text":"x Input tensor. decimals Number decimal places round . Defaults 0.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_round.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Evenly round to the given number of decimals. — op_round","text":"Output tensor.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_rsqrt.html","id":null,"dir":"Reference","previous_headings":"","what":"Computes reciprocal of square root of x element-wise. — op_rsqrt","title":"Computes reciprocal of square root of x element-wise. — op_rsqrt","text":"Computes reciprocal square root x element-wise.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_rsqrt.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes reciprocal of square root of x element-wise. — op_rsqrt","text":"","code":"op_rsqrt(x)"},{"path":"https://keras3.posit.co/dev/reference/op_rsqrt.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Computes reciprocal of square root of x element-wise. — op_rsqrt","text":"x input tensor","code":""},{"path":"https://keras3.posit.co/dev/reference/op_rsqrt.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Computes reciprocal of square root of x element-wise. — op_rsqrt","text":"tensor dtype x.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_rsqrt.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Computes reciprocal of square root of x element-wise. — op_rsqrt","text":"","code":"x <- op_convert_to_tensor(c(1, 10, 100)) op_rsqrt(x) ## tf.Tensor([1.         0.31622777 0.1       ], shape=(3), dtype=float64) # array([1, 0.31622776, 0.1], dtype=float32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_saturate_cast.html","id":null,"dir":"Reference","previous_headings":"","what":"Performs a safe saturating cast to the desired dtype. — op_saturate_cast","title":"Performs a safe saturating cast to the desired dtype. — op_saturate_cast","text":"Saturating cast prevents data type overflow casting dtype smaller values range. E.g. op_cast(c(-1, 256), \"float32\") |> op_cast(\"uint8\") returns c(255, 0), op_cast(c(-1, 256), \"float32\") |> op_saturate_cast(\"uint8\") returns c(0, 255).","code":""},{"path":"https://keras3.posit.co/dev/reference/op_saturate_cast.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Performs a safe saturating cast to the desired dtype. — op_saturate_cast","text":"","code":"op_saturate_cast(x, dtype)"},{"path":"https://keras3.posit.co/dev/reference/op_saturate_cast.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Performs a safe saturating cast to the desired dtype. — op_saturate_cast","text":"x tensor variable. dtype target type.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_saturate_cast.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Performs a safe saturating cast to the desired dtype. — op_saturate_cast","text":"safely casted tensor specified dtype.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_saturate_cast.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Performs a safe saturating cast to the desired dtype. — op_saturate_cast","text":"Image resizing bicubic interpolation may produce values outside original range.     Casting resized image back uint8 cause overflow.     Saturate casting uint8 clip values uint8 range casting cause overflow.","code":"image2x2 <- np_array(as.integer(c(0, 1, 254, 255)), \"uint8\") |>   array_reshape(c(1, 2, 2, 1)) image4x4 <- image2x2 |>   tensorflow::tf$image$resize(shape(4, 4), method=\"bicubic\") image4x4 |> as.array() |> drop() ##           [,1]      [,2]      [,3]      [,4] ## [1,] -22.50000 -22.20462 -21.61891 -21.32353 ## [2,]  52.52605  52.82143  53.40715  53.70253 ## [3,] 201.29752 201.59288 202.17859 202.47395 ## [4,] 276.32355 276.61893 277.20465 277.50006 image4x4_casted <- op_cast(image4x4, \"uint8\") image4x4_casted |> as.array() |> drop() ##      [,1] [,2] [,3] [,4] ## [1,]  234  234  235  235 ## [2,]   52   52   53   53 ## [3,]  201  201  202  202 ## [4,]   20   20   21   21 image4x4_saturate_casted <- image4x4 |> op_saturate_cast(\"uint8\") image4x4_saturate_casted |> as.array() |> drop() ##      [,1] [,2] [,3] [,4] ## [1,]    0    0    0    0 ## [2,]   52   52   53   53 ## [3,]  201  201  202  202 ## [4,]  255  255  255  255"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_scan.html","id":null,"dir":"Reference","previous_headings":"","what":"Scan a function over leading array axes while carrying along state. — op_scan","title":"Scan a function over leading array axes while carrying along state. — op_scan","text":"type xs array type NULL, type ys array type, semantics op_scan() given roughly implementation:   loop-carried value carry (init) must hold fixed shape dtype across iterations. TensorFlow, y must match carry shape dtype. required backends.","code":"op_scan <- function(f, init, xs = NULL, length = NULL) {   xs <- xs %||% vector(\"list\", length)   if(!is.list(xs))     xs <- op_unstack(xs)   ys <- vector(\"list\", length(xs))   carry <- init   for (i in seq_along(xs)) {     c(carry, y) %<-% f(carry, xs[[i]])     ys[[i]] <- y   }   list(carry, op_stack(ys)) }"},{"path":"https://keras3.posit.co/dev/reference/op_scan.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Scan a function over leading array axes while carrying along state. — op_scan","text":"","code":"op_scan(f, init, xs = NULL, length = NULL, reverse = FALSE, unroll = 1L)"},{"path":"https://keras3.posit.co/dev/reference/op_scan.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Scan a function over leading array axes while carrying along state. — op_scan","text":"f Callable defines logic loop iteration. accepts two arguments first value loop carry second slice xs along leading axis. callable returns pair first represents new value loop carry second represents slice output. init initial loop carry value. can scalar, tensor, nested structure. must match structure first element returned f. xs Optional value scan along leading axis. can tensor nested structure. xs provided, must specify length define number loop iterations. Defaults NULL. length Optional integer specifying number loop iterations. length provided, defaults sizes leading axis arrays xs. Defaults NULL. reverse Optional boolean specifying whether run scan iteration forward reverse, equivalent reversing leading axes arrays xs ys. unroll Optional positive integer boolean specifying many scan iterations unroll within single iteration loop. integer provided, determines many unrolled loop iterations run within single rolled iteration loop. boolean provided, determine loop completely unrolled (unroll=TRUE) left completely unrolled (unroll=FALSE). Note unrolling supported JAX TensorFlow backends.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_scan.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Scan a function over leading array axes while carrying along state. — op_scan","text":"pair first element represents final loop carry value second element represents stacked outputs f scanned leading axis inputs.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_scan.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Scan a function over leading array axes while carrying along state. — op_scan","text":"","code":"sum_fn <- function(c, x) list(c + x, c + x) init <- op_array(0L) xs <- op_array(1:5) c(carry, result) %<-% op_scan(sum_fn, init, xs) carry ## tf.Tensor(15, shape=(), dtype=int32) result ## tf.Tensor([ 1  3  6 10 15], shape=(5), dtype=int32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_scatter.html","id":null,"dir":"Reference","previous_headings":"","what":"Returns a tensor of shape shape where indices are set to values. — op_scatter","title":"Returns a tensor of shape shape where indices are set to values. — op_scatter","text":"high level, operation zeros[indices] = updates returns output. equivalent :","code":"output <- op_scatter_update(op_zeros(shape), indices, values)"},{"path":"https://keras3.posit.co/dev/reference/op_scatter.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Returns a tensor of shape shape where indices are set to values. — op_scatter","text":"","code":"op_scatter(indices, values, shape)"},{"path":"https://keras3.posit.co/dev/reference/op_scatter.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Returns a tensor of shape shape where indices are set to values. — op_scatter","text":"indices tensor list specifying indices values values. values tensor, values set indices. shape Shape output tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_scatter.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Returns a tensor of shape shape where indices are set to values. — op_scatter","text":"tensor shape shape indices set values.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_scatter.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Returns a tensor of shape shape where indices are set to values. — op_scatter","text":"","code":"op_scatter(indices = rbind(1), values = 1, shape = c(5)) ## tf.Tensor([1. 0. 0. 0. 0.], shape=(5), dtype=float64) op_scatter(indices = rbind(1, 3), values = 1:2, shape = c(5)) ## tf.Tensor([1 0 2 0 0], shape=(5), dtype=int32) indices <- rbind(c(1, 2), c(2, 2)) values <- op_array(c(1, 1)) op_scatter(indices, values, shape = c(2, 2)) ## tf.Tensor( ## [[0. 1.] ##  [0. 1.]], shape=(2, 2), dtype=float32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_scatter_update.html","id":null,"dir":"Reference","previous_headings":"","what":"Update inputs via updates at scattered (sparse) indices. — op_scatter_update","title":"Update inputs via updates at scattered (sparse) indices. — op_scatter_update","text":"high level, operation inputs[indices] <- updates. Assume inputs tensor shape (D1, D2, ..., Dn), 2 main usages scatter_update. indices 2D tensor shape (num_updates, n), num_updates number updates perform, updates 1D tensor shape (num_updates). example, inputs op_zeros(c(4, 4, 4)), want update inputs[2, 3, 4] inputs[1, 2, 4] 1, can use:     2 indices 2D tensor shape (num_updates, k), num_updates number updates perform, k (k <= n) size index indices. updates n - k-D tensor shape (num_updates, shape(inputs)[-(1:k)]). example, inputs <- op_zeros(c(4, 4, 4)), want update inputs[1, 2, ] inputs[2, 3, ] [1, 1, 1, 1], indices shape (num_updates, 2) (k = 2), updates shape (num_updates, 4) (shape(inputs)[3:4] == 4). See code :","code":"inputs <- op_zeros(c(4, 4, 4)) indices <- rbind(c(2, 3, 4),                  c(1, 2, 4)) updates <- op_array(c(1, 1), \"float32\") op_scatter_update(inputs, indices, updates) ## tf.Tensor( ## [[[0. 0. 0. 0.] ##   [0. 0. 0. 1.] ##   [0. 0. 0. 0.] ##   [0. 0. 0. 0.]] ## ##  [[0. 0. 0. 0.] ##   [0. 0. 0. 0.] ##   [0. 0. 0. 1.] ##   [0. 0. 0. 0.]] ## ##  [[0. 0. 0. 0.] ##   [0. 0. 0. 0.] ##   [0. 0. 0. 0.] ##   [0. 0. 0. 0.]] ## ##  [[0. 0. 0. 0.] ##   [0. 0. 0. 0.] ##   [0. 0. 0. 0.] ##   [0. 0. 0. 0.]]], shape=(4, 4, 4), dtype=float32) inputs <- op_zeros(c(4, 4, 4)) indices <- rbind(c(2, 3),                  c(3, 4)) updates <- op_array(rbind(c(1, 1, 1, 1),                           c(1, 1, 1, 1)),                     \"float32\") op_scatter_update(inputs, indices, updates) ## tf.Tensor( ## [[[0. 0. 0. 0.] ##   [0. 0. 0. 0.] ##   [0. 0. 0. 0.] ##   [0. 0. 0. 0.]] ## ##  [[0. 0. 0. 0.] ##   [0. 0. 0. 0.] ##   [1. 1. 1. 1.] ##   [0. 0. 0. 0.]] ## ##  [[0. 0. 0. 0.] ##   [0. 0. 0. 0.] ##   [0. 0. 0. 0.] ##   [1. 1. 1. 1.]] ## ##  [[0. 0. 0. 0.] ##   [0. 0. 0. 0.] ##   [0. 0. 0. 0.] ##   [0. 0. 0. 0.]]], shape=(4, 4, 4), dtype=float32)"},{"path":"https://keras3.posit.co/dev/reference/op_scatter_update.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Update inputs via updates at scattered (sparse) indices. — op_scatter_update","text":"","code":"op_scatter_update(inputs, indices, updates)"},{"path":"https://keras3.posit.co/dev/reference/op_scatter_update.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Update inputs via updates at scattered (sparse) indices. — op_scatter_update","text":"inputs tensor, tensor updated. indices tensor list shape (N, inputs$ndim), specifying indices update. N number indices update, must equal first dimension updates. updates tensor, new values put inputs indices.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_scatter_update.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Update inputs via updates at scattered (sparse) indices. — op_scatter_update","text":"tensor, shape dtype inputs.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_searchsorted.html","id":null,"dir":"Reference","previous_headings":"","what":"Perform a binary search. — op_searchsorted","title":"Perform a binary search. — op_searchsorted","text":"Return insertion indices keep values sorted order placed sorted_sequence.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_searchsorted.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Perform a binary search. — op_searchsorted","text":"","code":"op_searchsorted(sorted_sequence, values, side = \"left\", zero_indexed = FALSE)"},{"path":"https://keras3.posit.co/dev/reference/op_searchsorted.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Perform a binary search. — op_searchsorted","text":"sorted_sequence 1-D input tensor sorted along innermost dimension. values N-D tensor query insertion values. side Either \"left\" \"right\", choosing side insert ties. zero_indexed TRUE, returned indices zero-based (0 encodes first position); FALSE (default), returned indices one-based (1 encodes first position).","code":""},{"path":"https://keras3.posit.co/dev/reference/op_searchsorted.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Perform a binary search. — op_searchsorted","text":"Tensor insertion indices shape values.","code":""},{"path":[]},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_segment_max.html","id":null,"dir":"Reference","previous_headings":"","what":"Computes the max of segments in a tensor. — op_segment_max","title":"Computes the max of segments in a tensor. — op_segment_max","text":"Computes max segments tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_segment_max.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes the max of segments in a tensor. — op_segment_max","text":"","code":"op_segment_max(data, segment_ids, num_segments = NULL, sorted = FALSE)"},{"path":"https://keras3.posit.co/dev/reference/op_segment_max.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Computes the max of segments in a tensor. — op_segment_max","text":"data Input tensor. segment_ids N-D tensor containing segment indices element data. head(op_shape(data), length(op_shape(segment_ids))) match op_shape(segment_ids) num_segments integer representing total number segments. specified, inferred maximum value segment_ids. sorted boolean indicating whether segment_ids sorted. Defaults FALSE.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_segment_max.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Computes the max of segments in a tensor. — op_segment_max","text":"tensor containing max segments, element represents max corresponding segment data.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_segment_max.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Computes the max of segments in a tensor. — op_segment_max","text":"","code":"data <- op_convert_to_tensor(c(1, 2, 10, 20, 100, 200)) segment_ids <- op_array(c(1, 1, 2, 2, 3, 3), \"int32\") num_segments <- 3 op_segment_max(data, segment_ids, num_segments) ## tf.Tensor([  2.  20. 200.], shape=(3), dtype=float64) # array([2, 20, 200], dtype=int32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_segment_sum.html","id":null,"dir":"Reference","previous_headings":"","what":"Computes the sum of segments in a tensor. — op_segment_sum","title":"Computes the sum of segments in a tensor. — op_segment_sum","text":"Computes sum segments tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_segment_sum.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes the sum of segments in a tensor. — op_segment_sum","text":"","code":"op_segment_sum(data, segment_ids, num_segments = NULL, sorted = FALSE)"},{"path":"https://keras3.posit.co/dev/reference/op_segment_sum.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Computes the sum of segments in a tensor. — op_segment_sum","text":"data Input tensor. segment_ids N-D tensor containing segment indices element data. Num dims segment ids strictly smaller equal number dims data. num_segments integer representing total number segments. specified, inferred maximum value segment_ids. sorted boolean indicating whether segment_ids sorted. Defaults FALSE.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_segment_sum.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Computes the sum of segments in a tensor. — op_segment_sum","text":"tensor containing sum segments, element represents sum corresponding segment data.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_segment_sum.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Computes the sum of segments in a tensor. — op_segment_sum","text":"","code":"data <- op_array(c(1, 2, 10, 20, 100, 200)) segment_ids <- op_array(c(1, 1, 2, 2, 3, 3), \"int32\") num_segments <- 3 op_segment_sum(data, segment_ids, num_segments) ## tf.Tensor([  3.  30. 300.], shape=(3), dtype=float32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_select.html","id":null,"dir":"Reference","previous_headings":"","what":"Return elements from choicelist, based on conditions in condlist. — op_select","title":"Return elements from choicelist, based on conditions in condlist. — op_select","text":"Return elements choicelist, based conditions condlist.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_select.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Return elements from choicelist, based on conditions in condlist. — op_select","text":"","code":"op_select(condlist, choicelist, default = 0L)"},{"path":"https://keras3.posit.co/dev/reference/op_select.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Return elements from choicelist, based on conditions in condlist. — op_select","text":"condlist List boolean tensors. list conditions determine array choicelist output elements taken. multiple conditions satisfied, first one encountered condlist used. choicelist List tensors. list tensors output elements taken. list length condlist. default Optional scalar value. element inserted output conditions evaluate FALSE.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_select.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Return elements from choicelist, based on conditions in condlist. — op_select","text":"Tensor output position m m-th element tensor choicelist m-th element corresponding tensor condlist TRUE.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_select.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Return elements from choicelist, based on conditions in condlist. — op_select","text":"","code":"x <- op_arange(6L) condlist <- list(x < 3, x > 3) choicelist <- list(x, x^2) op_select(condlist, choicelist, 42) ## tf.Tensor([ 1  2 42 16 25 36], shape=(6), dtype=int32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_selu.html","id":null,"dir":"Reference","previous_headings":"","what":"Scaled Exponential Linear Unit (SELU) activation function. — op_selu","title":"Scaled Exponential Linear Unit (SELU) activation function. — op_selu","text":"defined : f(x) =  scale * alpha * (exp(x) - 1.) x < 0, f(x) = scale * x x >= 0.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_selu.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Scaled Exponential Linear Unit (SELU) activation function. — op_selu","text":"","code":"op_selu(x)"},{"path":"https://keras3.posit.co/dev/reference/op_selu.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Scaled Exponential Linear Unit (SELU) activation function. — op_selu","text":"x Input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_selu.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Scaled Exponential Linear Unit (SELU) activation function. — op_selu","text":"tensor shape x.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_selu.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Scaled Exponential Linear Unit (SELU) activation function. — op_selu","text":"","code":"x <- op_array(c(-1, 0, 1)) op_selu(x) ## tf.Tensor([-1.1113307  0.         1.050701 ], shape=(3), dtype=float32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_separable_conv.html","id":null,"dir":"Reference","previous_headings":"","what":"General N-D separable convolution. — op_separable_conv","title":"General N-D separable convolution. — op_separable_conv","text":"ops supports 1D 2D separable convolution. separable_conv depthwise conv followed pointwise conv.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_separable_conv.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"General N-D separable convolution. — op_separable_conv","text":"","code":"op_separable_conv(   inputs,   depthwise_kernel,   pointwise_kernel,   strides = 1L,   padding = \"valid\",   data_format = NULL,   dilation_rate = 1L )"},{"path":"https://keras3.posit.co/dev/reference/op_separable_conv.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"General N-D separable convolution. — op_separable_conv","text":"inputs Tensor rank N+2. inputs shape (batch_size,) + inputs_spatial_shape + (num_channels,) data_format=\"channels_last\", (batch_size, num_channels) + inputs_spatial_shape data_format=\"channels_first\". depthwise_kernel Tensor rank N+2. depthwise_kernel shape [kernel_spatial_shape, num_input_channels, num_channels_multiplier], num_input_channels match number channels inputs. pointwise_kernel Tensor rank N+2. pointwise_kernel shape (*ones_like(kernel_spatial_shape), num_input_channels * num_channels_multiplier, num_output_channels). strides int int tuple/list len(inputs_spatial_shape), specifying strides convolution along spatial dimension. strides int, every spatial dimension shares strides. padding string, either \"valid\" \"\". \"valid\" means padding applied, \"\" results padding evenly left/right /input output height/width dimension input strides=1. data_format string, either \"channels_last\" \"channels_first\". data_format determines ordering dimensions inputs. data_format=\"channels_last\", inputs shape (batch_size, ..., channels) data_format=\"channels_first\", inputs shape (batch_size, channels, ...). dilation_rate int int tuple/list len(inputs_spatial_shape), specifying dilation rate use dilated convolution. dilation_rate int, every spatial dimension shares dilation_rate.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_separable_conv.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"General N-D separable convolution. — op_separable_conv","text":"tensor rank N+2, result depthwise conv operation.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_shape.html","id":null,"dir":"Reference","previous_headings":"","what":"Gets the shape of the tensor input. — op_shape","title":"Gets the shape of the tensor input. — op_shape","text":"Gets shape tensor input.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_shape.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Gets the shape of the tensor input. — op_shape","text":"","code":"op_shape(x)"},{"path":"https://keras3.posit.co/dev/reference/op_shape.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Gets the shape of the tensor input. — op_shape","text":"x tensor. function try access shape attribute input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_shape.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Gets the shape of the tensor input. — op_shape","text":"list integers NULL values, indicating shape input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_shape.html","id":"note","dir":"Reference","previous_headings":"","what":"Note","title":"Gets the shape of the tensor input. — op_shape","text":"TensorFlow backend, x tf.Tensor dynamic shape, dimensions dynamic context compiled function tf.Tensor value instead static integer value.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_shape.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Gets the shape of the tensor input. — op_shape","text":"","code":"x <- op_zeros(c(8, 12)) op_shape(x) ## shape(8, 12)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_sigmoid.html","id":null,"dir":"Reference","previous_headings":"","what":"Sigmoid activation function. — op_sigmoid","title":"Sigmoid activation function. — op_sigmoid","text":"defined f(x) = 1 / (1 + exp(-x)).","code":""},{"path":"https://keras3.posit.co/dev/reference/op_sigmoid.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Sigmoid activation function. — op_sigmoid","text":"","code":"op_sigmoid(x)"},{"path":"https://keras3.posit.co/dev/reference/op_sigmoid.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Sigmoid activation function. — op_sigmoid","text":"x Input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_sigmoid.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Sigmoid activation function. — op_sigmoid","text":"tensor shape x.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_sigmoid.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Sigmoid activation function. — op_sigmoid","text":"","code":"x <- op_convert_to_tensor(c(-6, 1, 0, 1, 6)) op_sigmoid(x) ## tf.Tensor([0.00247262 0.73105858 0.5        0.73105858 0.99752738], shape=(5), dtype=float64)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_sign.html","id":null,"dir":"Reference","previous_headings":"","what":"Returns a tensor with the signs of the elements of x. — op_sign","title":"Returns a tensor with the signs of the elements of x. — op_sign","text":"Returns tensor signs elements x.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_sign.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Returns a tensor with the signs of the elements of x. — op_sign","text":"","code":"op_sign(x)"},{"path":"https://keras3.posit.co/dev/reference/op_sign.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Returns a tensor with the signs of the elements of x. — op_sign","text":"x Input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_sign.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Returns a tensor with the signs of the elements of x. — op_sign","text":"Output tensor shape x.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_signbit.html","id":null,"dir":"Reference","previous_headings":"","what":"Return the sign bit of the elements of x. — op_signbit","title":"Return the sign bit of the elements of x. — op_signbit","text":"output boolean tensor contains TRUE sign x negative, FALSE otherwise.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_signbit.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Return the sign bit of the elements of x. — op_signbit","text":"","code":"op_signbit(x)"},{"path":"https://keras3.posit.co/dev/reference/op_signbit.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Return the sign bit of the elements of x. — op_signbit","text":"x Input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_signbit.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Return the sign bit of the elements of x. — op_signbit","text":"Output boolean tensor shape x.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_silu.html","id":null,"dir":"Reference","previous_headings":"","what":"Sigmoid Linear Unit (SiLU) activation function, also known as Swish. — op_silu","title":"Sigmoid Linear Unit (SiLU) activation function, also known as Swish. — op_silu","text":"SiLU activation function computed sigmoid function multiplied input. defined f(x) = x * sigmoid(x).","code":""},{"path":"https://keras3.posit.co/dev/reference/op_silu.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Sigmoid Linear Unit (SiLU) activation function, also known as Swish. — op_silu","text":"","code":"op_silu(x)"},{"path":"https://keras3.posit.co/dev/reference/op_silu.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Sigmoid Linear Unit (SiLU) activation function, also known as Swish. — op_silu","text":"x Input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_silu.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Sigmoid Linear Unit (SiLU) activation function, also known as Swish. — op_silu","text":"tensor shape x.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_silu.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Sigmoid Linear Unit (SiLU) activation function, also known as Swish. — op_silu","text":"","code":"x <- op_convert_to_tensor(c(-6, 1, 0, 1, 6)) op_sigmoid(x) ## tf.Tensor([0.00247262 0.73105858 0.5        0.73105858 0.99752738], shape=(5), dtype=float64) op_silu(x) ## tf.Tensor([-0.01483574  0.73105858  0.          0.73105858  5.98516426], shape=(5), dtype=float64)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_sin.html","id":null,"dir":"Reference","previous_headings":"","what":"Trigonometric sine, element-wise. — op_sin","title":"Trigonometric sine, element-wise. — op_sin","text":"Trigonometric sine, element-wise.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_sin.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Trigonometric sine, element-wise. — op_sin","text":"","code":"op_sin(x)"},{"path":"https://keras3.posit.co/dev/reference/op_sin.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Trigonometric sine, element-wise. — op_sin","text":"x Input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_sin.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Trigonometric sine, element-wise. — op_sin","text":"Output tensor shape x.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_sinh.html","id":null,"dir":"Reference","previous_headings":"","what":"Hyperbolic sine, element-wise. — op_sinh","title":"Hyperbolic sine, element-wise. — op_sinh","text":"Hyperbolic sine, element-wise.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_sinh.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Hyperbolic sine, element-wise. — op_sinh","text":"","code":"op_sinh(x)"},{"path":"https://keras3.posit.co/dev/reference/op_sinh.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Hyperbolic sine, element-wise. — op_sinh","text":"x Input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_sinh.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Hyperbolic sine, element-wise. — op_sinh","text":"Output tensor shape x.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_size.html","id":null,"dir":"Reference","previous_headings":"","what":"Return the number of elements in a tensor. — op_size","title":"Return the number of elements in a tensor. — op_size","text":"Return number elements tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_size.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Return the number of elements in a tensor. — op_size","text":"","code":"op_size(x)"},{"path":"https://keras3.posit.co/dev/reference/op_size.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Return the number of elements in a tensor. — op_size","text":"x Input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_size.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Return the number of elements in a tensor. — op_size","text":"Number elements x.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_slice.html","id":null,"dir":"Reference","previous_headings":"","what":"Return a slice of an input tensor. — op_slice","title":"Return a slice of an input tensor. — op_slice","text":"high level, operation explicit replacement array slicing e.g. inputs[start_indices:(start_indices + shape)]. Unlike slicing via brackets, operation accept tensor start indices backends, useful indices dynamically computed via tensor operations.","code":"(inputs <- op_arange(5*5) |> op_reshape(c(5, 5))) ## tf.Tensor( ## [[ 1.  2.  3.  4.  5.] ##  [ 6.  7.  8.  9. 10.] ##  [11. 12. 13. 14. 15.] ##  [16. 17. 18. 19. 20.] ##  [21. 22. 23. 24. 25.]], shape=(5, 5), dtype=float32) start_indices <- c(3, 3) shape <- c(2, 2) op_slice(inputs, start_indices, shape) ## tf.Tensor( ## [[13. 14.] ##  [18. 19.]], shape=(2, 2), dtype=float32)"},{"path":"https://keras3.posit.co/dev/reference/op_slice.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Return a slice of an input tensor. — op_slice","text":"","code":"op_slice(inputs, start_indices, shape)"},{"path":"https://keras3.posit.co/dev/reference/op_slice.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Return a slice of an input tensor. — op_slice","text":"inputs tensor, tensor sliced. start_indices list length inputs$ndim, specifying starting indices updating. shape full shape returned slice.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_slice.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Return a slice of an input tensor. — op_slice","text":"tensor, shape dtype inputs.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_slice_update.html","id":null,"dir":"Reference","previous_headings":"","what":"Update an input by slicing in a tensor of updated values. — op_slice_update","title":"Update an input by slicing in a tensor of updated values. — op_slice_update","text":"high level, operation inputs[start_indices: start_indices + updates.shape] = updates. Assume inputs tensor shape (D1, D2, ..., Dn), start_indices must list n integers, specifying starting indices. updates must rank inputs, size dim must exceed Di - start_indices[]. example, 2D inputs inputs = op_zeros(c(5, 5)), want update intersection last 2 rows last 2 columns 1, .e., inputs[4:5, 4:5] = op_ones(c(2, 2)), can use code :","code":"inputs <- op_zeros(c(5, 5)) start_indices <- c(3, 3) updates <- op_ones(c(2, 2)) op_slice_update(inputs, start_indices, updates) ## tf.Tensor( ## [[0. 0. 0. 0. 0.] ##  [0. 0. 0. 0. 0.] ##  [0. 0. 1. 1. 0.] ##  [0. 0. 1. 1. 0.] ##  [0. 0. 0. 0. 0.]], shape=(5, 5), dtype=float32)"},{"path":"https://keras3.posit.co/dev/reference/op_slice_update.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Update an input by slicing in a tensor of updated values. — op_slice_update","text":"","code":"op_slice_update(inputs, start_indices, updates)"},{"path":"https://keras3.posit.co/dev/reference/op_slice_update.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Update an input by slicing in a tensor of updated values. — op_slice_update","text":"inputs tensor, tensor updated. start_indices list length inputs$ndim, specifying starting indices updating. updates tensor, new values put inputs indices. updates must rank inputs.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_slice_update.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Update an input by slicing in a tensor of updated values. — op_slice_update","text":"tensor, shape dtype inputs.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_slogdet.html","id":null,"dir":"Reference","previous_headings":"","what":"Compute the sign and natural logarithm of the determinant of a matrix. — op_slogdet","title":"Compute the sign and natural logarithm of the determinant of a matrix. — op_slogdet","text":"Compute sign natural logarithm determinant matrix.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_slogdet.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Compute the sign and natural logarithm of the determinant of a matrix. — op_slogdet","text":"","code":"op_slogdet(x)"},{"path":"https://keras3.posit.co/dev/reference/op_slogdet.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Compute the sign and natural logarithm of the determinant of a matrix. — op_slogdet","text":"x Input matrix. must 2D square.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_slogdet.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Compute the sign and natural logarithm of the determinant of a matrix. — op_slogdet","text":"list: (sign, logabsdet). sign number representing sign determinant. real matrix, 1, 0, -1. complex matrix, complex number absolute value 1 (.e., unit circle), else 0. logabsdet natural log absolute value determinant.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_soft_shrink.html","id":null,"dir":"Reference","previous_headings":"","what":"Soft Shrink activation function. — op_soft_shrink","title":"Soft Shrink activation function. — op_soft_shrink","text":"defined f(x) = x - threshold x > threshold, f(x) = x + threshold x < -threshold, f(x) = 0 otherwise.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_soft_shrink.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Soft Shrink activation function. — op_soft_shrink","text":"","code":"op_soft_shrink(x, threshold = 0.5)"},{"path":"https://keras3.posit.co/dev/reference/op_soft_shrink.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Soft Shrink activation function. — op_soft_shrink","text":"x Input tensor. threshold Threshold value. Defaults 0.5.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_soft_shrink.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Soft Shrink activation function. — op_soft_shrink","text":"tensor shape x.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_soft_shrink.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Soft Shrink activation function. — op_soft_shrink","text":"","code":"x <- op_array(c(-1, 0, 1)) op_soft_shrink(x) ## tf.Tensor([-0.5  0.   0.5], shape=(3), dtype=float32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_softmax.html","id":null,"dir":"Reference","previous_headings":"","what":"Softmax activation function. — op_softmax","title":"Softmax activation function. — op_softmax","text":"elements output vector lie within range (0, 1), total sum exactly 1 (excluding floating point rounding error). vector processed independently. axis argument specifies axis along function applied within input. defined : f(x) = exp(x) / sum(exp(x))","code":""},{"path":"https://keras3.posit.co/dev/reference/op_softmax.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Softmax activation function. — op_softmax","text":"","code":"op_softmax(x, axis = -1L)"},{"path":"https://keras3.posit.co/dev/reference/op_softmax.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Softmax activation function. — op_softmax","text":"x Input tensor. axis Integer, axis along softmax applied.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_softmax.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Softmax activation function. — op_softmax","text":"tensor shape x.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_softmax.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Softmax activation function. — op_softmax","text":"","code":"x <- op_array(c(-1, 0, 1)) op_softmax(x) ## tf.Tensor([0.09003057 0.24472848 0.66524094], shape=(3), dtype=float32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_softplus.html","id":null,"dir":"Reference","previous_headings":"","what":"Softplus activation function. — op_softplus","title":"Softplus activation function. — op_softplus","text":"defined f(x) = log(exp(x) + 1), log natural logarithm exp exponential function.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_softplus.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Softplus activation function. — op_softplus","text":"","code":"op_softplus(x)"},{"path":"https://keras3.posit.co/dev/reference/op_softplus.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Softplus activation function. — op_softplus","text":"x Input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_softplus.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Softplus activation function. — op_softplus","text":"tensor shape x.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_softplus.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Softplus activation function. — op_softplus","text":"","code":"x <- op_convert_to_tensor(c(-0.555, 0, 0.555)) op_softplus(x) ## tf.Tensor([0.45366606 0.69314718 1.00866606], shape=(3), dtype=float64) x <- seq(-10, 10, .1) plot(x, op_softplus(x))"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_softsign.html","id":null,"dir":"Reference","previous_headings":"","what":"Softsign activation function. — op_softsign","title":"Softsign activation function. — op_softsign","text":"defined f(x) = x / (abs(x) + 1).","code":""},{"path":"https://keras3.posit.co/dev/reference/op_softsign.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Softsign activation function. — op_softsign","text":"","code":"op_softsign(x)"},{"path":"https://keras3.posit.co/dev/reference/op_softsign.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Softsign activation function. — op_softsign","text":"x Input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_softsign.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Softsign activation function. — op_softsign","text":"tensor shape x.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_softsign.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Softsign activation function. — op_softsign","text":"","code":"x <- op_convert_to_tensor(c(-0.100, -10.0, 1.0, 0.0, 100.0)) op_softsign(x) ## tf.Tensor([-0.09090909 -0.90909091  0.5         0.          0.99009901], shape=(5), dtype=float64) x <- seq(-10, 10, .1) plot(x, op_softsign(x), ylim = c(-1, 1))"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_solve.html","id":null,"dir":"Reference","previous_headings":"","what":"Solves a linear system of equations given by a x = b. — op_solve","title":"Solves a linear system of equations given by a x = b. — op_solve","text":"Solves x equation %*% x == b.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_solve.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Solves a linear system of equations given by a x = b. — op_solve","text":"","code":"op_solve(a, b)"},{"path":"https://keras3.posit.co/dev/reference/op_solve.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Solves a linear system of equations given by a x = b. — op_solve","text":"tensor shape (..., M, M) representing coefficients matrix. b tensor shape (..., M) (..., M, N) representing right-hand side \"dependent variable\" matrix.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_solve.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Solves a linear system of equations given by a x = b. — op_solve","text":"tensor shape (..., M) (..., M, N) representing solution linear system. Returned shape identical b.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_solve.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Solves a linear system of equations given by a x = b. — op_solve","text":"","code":"a <- op_array(c(1, 2, 4, 5), dtype=\"float32\") |> op_reshape(c(2, 2)) b <- op_array(c(2, 4, 8, 10), dtype=\"float32\") |> op_reshape(c(2, 2)) op_solve(a, b) ## tf.Tensor( ## [[2. 0.] ##  [0. 2.]], shape=(2, 2), dtype=float32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_solve_triangular.html","id":null,"dir":"Reference","previous_headings":"","what":"Solves a linear system of equations given by a %*% x = b. — op_solve_triangular","title":"Solves a linear system of equations given by a %*% x = b. — op_solve_triangular","text":"Solves linear system equations given %*% x = b.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_solve_triangular.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Solves a linear system of equations given by a %*% x = b. — op_solve_triangular","text":"","code":"op_solve_triangular(a, b, lower = FALSE)"},{"path":"https://keras3.posit.co/dev/reference/op_solve_triangular.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Solves a linear system of equations given by a %*% x = b. — op_solve_triangular","text":"tensor shape (..., M, M) representing coefficients matrix. b tensor shape (..., M) (..., M, N) representing right-hand side \"dependent variable\" matrix. lower logical. Use data contained lower triangle . Default use upper triangle.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_solve_triangular.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Solves a linear system of equations given by a %*% x = b. — op_solve_triangular","text":"tensor shape (..., M) (..., M, N) representing solution linear system. Returned shape identical b.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_sort.html","id":null,"dir":"Reference","previous_headings":"","what":"Sorts the elements of x along a given axis in ascending order. — op_sort","title":"Sorts the elements of x along a given axis in ascending order. — op_sort","text":"Sorts elements x along given axis ascending order.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_sort.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Sorts the elements of x along a given axis in ascending order. — op_sort","text":"","code":"op_sort(x, axis = -1L)"},{"path":"https://keras3.posit.co/dev/reference/op_sort.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Sorts the elements of x along a given axis in ascending order. — op_sort","text":"x Input tensor. axis Axis along sort. NULL, tensor flattened sorting. Defaults -1; last axis.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_sort.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Sorts the elements of x along a given axis in ascending order. — op_sort","text":"Sorted tensor.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_sparse_categorical_crossentropy.html","id":null,"dir":"Reference","previous_headings":"","what":"Computes sparse categorical cross-entropy loss. — op_sparse_categorical_crossentropy","title":"Computes sparse categorical cross-entropy loss. — op_sparse_categorical_crossentropy","text":"sparse categorical cross-entropy loss similar categorical cross-entropy, used target tensor contains integer class labels instead one-hot encoded vectors. measures dissimilarity target output probabilities logits.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_sparse_categorical_crossentropy.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes sparse categorical cross-entropy loss. — op_sparse_categorical_crossentropy","text":"","code":"op_sparse_categorical_crossentropy(   target,   output,   from_logits = FALSE,   axis = -1L )"},{"path":"https://keras3.posit.co/dev/reference/op_sparse_categorical_crossentropy.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Computes sparse categorical cross-entropy loss. — op_sparse_categorical_crossentropy","text":"target target tensor representing true class labels integers. shape match shape output tensor except last dimension. output output tensor representing predicted probabilities logits. shape match shape target tensor except last dimension. from_logits (optional) Whether output tensor logits probabilities. Set TRUE output represents logits; otherwise, set FALSE output represents probabilities. Defaults FALSE. axis (optional) axis along sparse categorical cross-entropy computed. Defaults -1, corresponds last dimension tensors.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_sparse_categorical_crossentropy.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Computes sparse categorical cross-entropy loss. — op_sparse_categorical_crossentropy","text":"Integer tensor: computed sparse categorical cross-entropy loss target output.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_sparse_categorical_crossentropy.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Computes sparse categorical cross-entropy loss. — op_sparse_categorical_crossentropy","text":"","code":"target <- op_array(c(0, 1, 2), dtype=\"int32\") output <- op_array(rbind(c(0.9, 0.05, 0.05),                         c(0.1, 0.8,  0.1),                         c(0.2, 0.3,  0.5))) op_sparse_categorical_crossentropy(target, output) ## tf.Tensor([0.10536052 0.22314355 0.69314718], shape=(3), dtype=float64)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_sparse_plus.html","id":null,"dir":"Reference","previous_headings":"","what":"SparsePlus activation function. — op_sparse_plus","title":"SparsePlus activation function. — op_sparse_plus","text":"defined f(x) = 0 x <= -1. f(x) = (1/4) * (x + 1)^2 -1 < x < 1. f(x) = x x >= 1.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_sparse_plus.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"SparsePlus activation function. — op_sparse_plus","text":"","code":"op_sparse_plus(x)"},{"path":"https://keras3.posit.co/dev/reference/op_sparse_plus.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"SparsePlus activation function. — op_sparse_plus","text":"x Input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_sparse_plus.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"SparsePlus activation function. — op_sparse_plus","text":"tensor shape x.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_sparse_plus.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"SparsePlus activation function. — op_sparse_plus","text":"","code":"x <- op_array(c(-1.0, 0.0, 1.0)) op_sparse_plus(x) ## tf.Tensor([0.   0.25 1.  ], shape=(3), dtype=float32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_sparse_sigmoid.html","id":null,"dir":"Reference","previous_headings":"","what":"Sparse sigmoid activation function. — op_sparse_sigmoid","title":"Sparse sigmoid activation function. — op_sparse_sigmoid","text":"defined f(x) = 0 x <= -1, f(x) = 0.5 * (x + 1) -1 < x < 1, f(x) = 1 x >= 1.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_sparse_sigmoid.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Sparse sigmoid activation function. — op_sparse_sigmoid","text":"","code":"op_sparse_sigmoid(x)"},{"path":"https://keras3.posit.co/dev/reference/op_sparse_sigmoid.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Sparse sigmoid activation function. — op_sparse_sigmoid","text":"x Input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_sparse_sigmoid.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Sparse sigmoid activation function. — op_sparse_sigmoid","text":"tensor shape x.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_sparse_sigmoid.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Sparse sigmoid activation function. — op_sparse_sigmoid","text":"","code":"x <- op_array(c(-1.0, 0.0, 1.0)) op_sparse_sigmoid(x) ## tf.Tensor([0.  0.5 1. ], shape=(3), dtype=float32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_sparsemax.html","id":null,"dir":"Reference","previous_headings":"","what":"Sparsemax activation function. — op_sparsemax","title":"Sparsemax activation function. — op_sparsemax","text":"batch , class j, sparsemax activation function defined : sparsemax(x)[, j] = max(x[, j] - (x[, :]), 0).","code":""},{"path":"https://keras3.posit.co/dev/reference/op_sparsemax.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Sparsemax activation function. — op_sparsemax","text":"","code":"op_sparsemax(x, axis = -1L)"},{"path":"https://keras3.posit.co/dev/reference/op_sparsemax.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Sparsemax activation function. — op_sparsemax","text":"x Input tensor. axis int, axis along sparsemax operation applied.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_sparsemax.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Sparsemax activation function. — op_sparsemax","text":"tensor, output sparsemax transformation. type shape x.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_sparsemax.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Sparsemax activation function. — op_sparsemax","text":"","code":"x <- op_array(c(-1., 0., 1.)) op_sparsemax(x) ## tf.Tensor([0. 0. 1.], shape=(3), dtype=float32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_split.html","id":null,"dir":"Reference","previous_headings":"","what":"Split a tensor into chunks. — op_split","title":"Split a tensor into chunks. — op_split","text":"Split tensor chunks.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_split.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Split a tensor into chunks. — op_split","text":"","code":"op_split(x, indices_or_sections, axis = 1L)"},{"path":"https://keras3.posit.co/dev/reference/op_split.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Split a tensor into chunks. — op_split","text":"x Input tensor. indices_or_sections integer, N, tensor split N equal sections along axis. 1-D array sorted integers, entries indicate indices tensor split along axis. axis Axis along split. Defaults 1, first axis.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_split.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Split a tensor into chunks. — op_split","text":"list tensors.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_split.html","id":"note","dir":"Reference","previous_headings":"","what":"Note","title":"Split a tensor into chunks. — op_split","text":"split result equal division using Torch backend.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_split.html","id":"example","dir":"Reference","previous_headings":"","what":"Example","title":"Split a tensor into chunks. — op_split","text":"","code":"x <- op_arange(12)  # pass a scalar integer for n sections op_split(x, 2) ## [[1]] ## tf.Tensor([1. 2. 3. 4. 5. 6.], shape=(6), dtype=float32) ## ## [[2]] ## tf.Tensor([ 7.  8.  9. 10. 11. 12.], shape=(6), dtype=float32) op_split(x, 3) ## [[1]] ## tf.Tensor([1. 2. 3. 4.], shape=(4), dtype=float32) ## ## [[2]] ## tf.Tensor([5. 6. 7. 8.], shape=(4), dtype=float32) ## ## [[3]] ## tf.Tensor([ 9. 10. 11. 12.], shape=(4), dtype=float32) # 1-d array/tensor for indices op_split(x, array(c(3, 8))) ## [[1]] ## tf.Tensor([1. 2.], shape=(2), dtype=float32) ## ## [[2]] ## tf.Tensor([3. 4. 5. 6. 7.], shape=(5), dtype=float32) ## ## [[3]] ## tf.Tensor([ 8.  9. 10. 11. 12.], shape=(5), dtype=float32) op_split(x, array(c(3))) ## [[1]] ## tf.Tensor([1. 2.], shape=(2), dtype=float32) ## ## [[2]] ## tf.Tensor([ 3.  4.  5.  6.  7.  8.  9. 10. 11. 12.], shape=(10), dtype=float32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_sqrt.html","id":null,"dir":"Reference","previous_headings":"","what":"Return the non-negative square root of a tensor, element-wise. — op_sqrt","title":"Return the non-negative square root of a tensor, element-wise. — op_sqrt","text":"Return non-negative square root tensor, element-wise.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_sqrt.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Return the non-negative square root of a tensor, element-wise. — op_sqrt","text":"","code":"op_sqrt(x)"},{"path":"https://keras3.posit.co/dev/reference/op_sqrt.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Return the non-negative square root of a tensor, element-wise. — op_sqrt","text":"x Input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_sqrt.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Return the non-negative square root of a tensor, element-wise. — op_sqrt","text":"Output tensor, non-negative square root x.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_square.html","id":null,"dir":"Reference","previous_headings":"","what":"Return the element-wise square of the input. — op_square","title":"Return the element-wise square of the input. — op_square","text":"Return element-wise square input.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_square.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Return the element-wise square of the input. — op_square","text":"","code":"op_square(x)"},{"path":"https://keras3.posit.co/dev/reference/op_square.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Return the element-wise square of the input. — op_square","text":"x Input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_square.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Return the element-wise square of the input. — op_square","text":"Output tensor, square x.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_squareplus.html","id":null,"dir":"Reference","previous_headings":"","what":"Squareplus activation function. — op_squareplus","title":"Squareplus activation function. — op_squareplus","text":"Squareplus activation function defined : f(x) = (x + sqrt(x^2 + b)) / 2","code":""},{"path":"https://keras3.posit.co/dev/reference/op_squareplus.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Squareplus activation function. — op_squareplus","text":"","code":"op_squareplus(x, b = 4L)"},{"path":"https://keras3.posit.co/dev/reference/op_squareplus.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Squareplus activation function. — op_squareplus","text":"x Input tensor. b Smoothness parameter. Defaults 4.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_squareplus.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Squareplus activation function. — op_squareplus","text":"tensor shape x.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_squareplus.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Squareplus activation function. — op_squareplus","text":"","code":"x <- op_array(c(-1.0, 0.0, 1.0)) op_squareplus(x) ## tf.Tensor([0.618034 1.       1.618034], shape=(3), dtype=float32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_squeeze.html","id":null,"dir":"Reference","previous_headings":"","what":"Remove axes of length one from x. — op_squeeze","title":"Remove axes of length one from x. — op_squeeze","text":"Remove axes length one x.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_squeeze.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Remove axes of length one from x. — op_squeeze","text":"","code":"op_squeeze(x, axis = NULL)"},{"path":"https://keras3.posit.co/dev/reference/op_squeeze.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Remove axes of length one from x. — op_squeeze","text":"x Input tensor. axis Select subset entries length one shape.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_squeeze.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Remove axes of length one from x. — op_squeeze","text":"input tensor subset dimensions length 1 removed.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_stack.html","id":null,"dir":"Reference","previous_headings":"","what":"Join a sequence of tensors along a new axis. — op_stack","title":"Join a sequence of tensors along a new axis. — op_stack","text":"axis parameter specifies index new axis dimensions result.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_stack.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Join a sequence of tensors along a new axis. — op_stack","text":"","code":"op_stack(x, axis = 1L)"},{"path":"https://keras3.posit.co/dev/reference/op_stack.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Join a sequence of tensors along a new axis. — op_stack","text":"x sequence tensors. axis Axis along stack. Defaults 1, first axis.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_stack.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Join a sequence of tensors along a new axis. — op_stack","text":"stacked tensor.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_std.html","id":null,"dir":"Reference","previous_headings":"","what":"Compute the standard deviation along the specified axis. — op_std","title":"Compute the standard deviation along the specified axis. — op_std","text":"Compute standard deviation along specified axis.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_std.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Compute the standard deviation along the specified axis. — op_std","text":"","code":"op_std(x, axis = NULL, keepdims = FALSE)"},{"path":"https://keras3.posit.co/dev/reference/op_std.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Compute the standard deviation along the specified axis. — op_std","text":"x Input tensor. axis Axis along compute standard deviation. Default compute standard deviation flattened tensor. keepdims set TRUE, axes reduced left result dimensions size one.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_std.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Compute the standard deviation along the specified axis. — op_std","text":"Output tensor containing standard deviation values.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_stft.html","id":null,"dir":"Reference","previous_headings":"","what":"Short-Time Fourier Transform along the last axis of the input. — op_stft","title":"Short-Time Fourier Transform along the last axis of the input. — op_stft","text":"STFT computes Fourier transform short overlapping windows input. giving frequency components signal change time.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_stft.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Short-Time Fourier Transform along the last axis of the input. — op_stft","text":"","code":"op_stft(   x,   sequence_length,   sequence_stride,   fft_length,   window = \"hann\",   center = TRUE )"},{"path":"https://keras3.posit.co/dev/reference/op_stft.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Short-Time Fourier Transform along the last axis of the input. — op_stft","text":"x Input tensor. sequence_length integer representing sequence length. sequence_stride integer representing sequence hop size. fft_length integer representing size FFT apply. specified, uses smallest power 2 enclosing sequence_length. window string, tensor window NULL. window string, available values \"hann\" \"hamming\". window tensor, used directly window length must sequence_length. window NULL, windowing used. Defaults \"hann\". center Whether pad x sides t-th sequence centered time t * sequence_stride. Otherwise, t-th sequence begins time t * sequence_stride. Defaults TRUE.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_stft.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Short-Time Fourier Transform along the last axis of the input. — op_stft","text":"list containing two tensors - real imaginary parts STFT output.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_stft.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Short-Time Fourier Transform along the last axis of the input. — op_stft","text":"","code":"x <- op_array(c(0, 1, 2, 3, 4)) op_stft(x, 3, 2, 3) ## [[1]] ## tf.Tensor( ## [[ 0.  0.] ##  [ 2. -1.] ##  [ 4. -2.]], shape=(3, 2), dtype=float32) ## ## [[2]] ## tf.Tensor( ## [[ 0.         0.       ] ##  [ 0.        -1.7320508] ##  [ 0.        -3.4641016]], shape=(3, 2), dtype=float32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_stop_gradient.html","id":null,"dir":"Reference","previous_headings":"","what":"Stops gradient computation. — op_stop_gradient","title":"Stops gradient computation. — op_stop_gradient","text":"Stops gradient computation.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_stop_gradient.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Stops gradient computation. — op_stop_gradient","text":"","code":"op_stop_gradient(variable)"},{"path":"https://keras3.posit.co/dev/reference/op_stop_gradient.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Stops gradient computation. — op_stop_gradient","text":"variable tensor variable gradient computation disabled.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_stop_gradient.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Stops gradient computation. — op_stop_gradient","text":"variable gradient computation disabled.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_stop_gradient.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Stops gradient computation. — op_stop_gradient","text":"","code":"var <- op_convert_to_tensor(c(1, 2, 3), dtype=\"float32\") var <- op_stop_gradient(var)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_subset.html","id":null,"dir":"Reference","previous_headings":"","what":"Subset elements from a tensor — op_subset","title":"Subset elements from a tensor — op_subset","text":"Extract elements tensor using common R-style [ indexing idioms. function can also conveniently accessed via syntax tensor@r[...].","code":""},{"path":"https://keras3.posit.co/dev/reference/op_subset.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Subset elements from a tensor — op_subset","text":"","code":"op_subset(x, ...)  op_subset(x, ...) <- value  op_subset_set(x, ..., value)"},{"path":"https://keras3.posit.co/dev/reference/op_subset.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Subset elements from a tensor — op_subset","text":"x Input tensor. ... Indices specifying elements extract. argument ... can : integer scalar 1-d integer logical vector NULL newaxis .. symbol slice expression using : single argument supplied ..., ..1 can also : logical array shape x integer matrix ncol(..1) == op_rank(x) value new value replace selected subset .","code":""},{"path":"https://keras3.posit.co/dev/reference/op_subset.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Subset elements from a tensor — op_subset","text":"tensor containing subset elements.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_subset.html","id":"details","dir":"Reference","previous_headings":"","what":"Details","title":"Subset elements from a tensor — op_subset","text":"semantics similar R's [, differences:","code":""},{"path":"https://keras3.posit.co/dev/reference/op_subset.html","id":"differences-from-r-s-","dir":"Reference","previous_headings":"","what":"Differences from R's [:","title":"Subset elements from a tensor — op_subset","text":"Negative indices follow Python-style indexing, counting end array. NULL newaxis adds new dimension (equivalent op_expand_dims()). fewer indices dimensions (op_rank(x)) provided, missing dimensions implicitly filled. example, x matrix, x[1] returns first row. .. all_dims() expands include unspecified dimensions (see examples). Extended slicing syntax (:) supported, including: Strided steps: x@r[start:end:step] NA values start end. NA start defaults 1, NA end defaults axis size. logical array matching shape x selects elements row-wise order.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_subset.html","id":"similarities-with-r-s-","dir":"Reference","previous_headings":"","what":"Similarities with R's [:","title":"Subset elements from a tensor — op_subset","text":"Similarities R's [ (differences Python's [): Positive indices 1-based. Slices (x[start:end]) inclusive end. 1-d logical/integer arrays subset along respective axis. Multiple vectors provided different axes return intersected subsets. single integer matrix ncol() == op_rank(x) selects elements coordinates. row matrix specifies location one value, column corresponds axis tensor subsetted.  means use 2-column matrix subset matrix, 3-column matrix subset 3d array, .","code":""},{"path":"https://keras3.posit.co/dev/reference/op_subset.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Subset elements from a tensor — op_subset","text":"","code":"(x <- op_arange(5L) + 10L) ## tf.Tensor([11 12 13 14 15], shape=(5), dtype=int32) # Basic example, get first element op_subset(x, 1) ## tf.Tensor(11, shape=(), dtype=int32) # Use `@r[` syntax x@r[1]           # same as `op_subset(x, 1)` ## tf.Tensor(11, shape=(), dtype=int32) x@r[1:2]         # get the first 2 elements ## tf.Tensor([11 12], shape=(2), dtype=int32) x@r[c(1, 3)]     # first and third element ## tf.Tensor([11 13], shape=(2), dtype=int32) # Negative indices x@r[-1]          # last element ## tf.Tensor(15, shape=(), dtype=int32) x@r[-2]          # second to last element ## tf.Tensor(14, shape=(), dtype=int32) x@r[c(-1, -2)]   # last and second to last elements ## tf.Tensor([15 14], shape=(2), dtype=int32) x@r[c(-2, -1)]   # second to last and last elements ## tf.Tensor([14 15], shape=(2), dtype=int32) x@r[c(1, -1)]    # first and last elements ## tf.Tensor([11 15], shape=(2), dtype=int32) # Slices x@r[1:3]          # first 3 elements ## tf.Tensor([11 12 13], shape=(3), dtype=int32) x@r[NA:3]         # first 3 elements ## tf.Tensor([11 12 13], shape=(3), dtype=int32) x@r[1:5]          # all elements ## tf.Tensor([11 12 13 14 15], shape=(5), dtype=int32) x@r[1:-1]         # all elements ## tf.Tensor([11 12 13 14 15], shape=(5), dtype=int32) x@r[NA:NA]        # all elements ## tf.Tensor([11 12 13 14 15], shape=(5), dtype=int32) x@r[]             # all elements ## tf.Tensor([11 12 13 14 15], shape=(5), dtype=int32) x@r[1:-2]         # drop last element ## tf.Tensor([11 12 13 14], shape=(4), dtype=int32) x@r[NA:-2]        # drop last element ## tf.Tensor([11 12 13 14], shape=(4), dtype=int32) x@r[2:NA]         # drop first element ## tf.Tensor([12 13 14 15], shape=(4), dtype=int32) # 2D array examples xr <- array(1:12, c(3, 4)) x <- op_convert_to_tensor(xr)  # Basic subsetting x@r[1, ]      # first row ## tf.Tensor([ 1  4  7 10], shape=(4), dtype=int32) x@r[1]        # also first row! Missing axes are implicitly inserted ## tf.Tensor([ 1  4  7 10], shape=(4), dtype=int32) x@r[-1]       # last row ## tf.Tensor([ 3  6  9 12], shape=(4), dtype=int32) x@r[, 2]      # second column ## tf.Tensor([4 5 6], shape=(3), dtype=int32) x@r[, 2:2]    # second column, but shape preserved (like [, drop=FALSE]) ## tf.Tensor( ## [[4] ##  [5] ##  [6]], shape=(3, 1), dtype=int32) # Subsetting with a boolean array # Note: extracted elements are selected row-wise, not column-wise mask <- x >= 6 x@r[mask]             # returns a 1D tensor ## tf.Tensor([ 7 10  8 11  6  9 12], shape=(7), dtype=int32) x.r <- as.array(x) mask.r <- as.array(mask) # as.array(x)[mask] selects column-wise. Use `aperm()` to reverse search order. all(aperm(x.r)[aperm(mask.r)] == as.array(x@r[mask])) ## [1] TRUE # Subsetting with a matrix of index positions indices <- rbind(c(1, 1), c(2, 2), c(3, 3)) x@r[indices] # get diagonal elements ## tf.Tensor([1 5 9], shape=(3), dtype=int32) x.r[indices] # same as subsetting an R array ## [1] 1 5 9 # 3D array examples # Image: 4x4 pixels, 3 colors (RGB) # Tensor shape: (img_height, img_width, img_color_channels) shp <- shape(4, 4, 3) x <- op_arange(prod(shp)) |> op_reshape(shp)  # Convert to a batch of images by inserting a new axis # New shape: (batch_size, img_height, img_width, img_color_channels) x@r[newaxis, , , ] |> op_shape() ## shape(1, 4, 4, 3) x@r[newaxis] |> op_shape()  # same as above ## shape(1, 4, 4, 3) x@r[NULL] |> op_shape()     # same as above ## shape(1, 4, 4, 3) x <- x@r[newaxis] # Extract color channels x@r[, , , 1]          # red channel ## tf.Tensor( ## [[[ 1.  4.  7. 10.] ##   [13. 16. 19. 22.] ##   [25. 28. 31. 34.] ##   [37. 40. 43. 46.]]], shape=(1, 4, 4), dtype=float32) x@r[.., 1]            # red channel, same as above using .. shorthand ## tf.Tensor( ## [[[ 1.  4.  7. 10.] ##   [13. 16. 19. 22.] ##   [25. 28. 31. 34.] ##   [37. 40. 43. 46.]]], shape=(1, 4, 4), dtype=float32) x@r[.., 2]            # green channel ## tf.Tensor( ## [[[ 2.  5.  8. 11.] ##   [14. 17. 20. 23.] ##   [26. 29. 32. 35.] ##   [38. 41. 44. 47.]]], shape=(1, 4, 4), dtype=float32) x@r[.., 3]            # blue channel ## tf.Tensor( ## [[[ 3.  6.  9. 12.] ##   [15. 18. 21. 24.] ##   [27. 30. 33. 36.] ##   [39. 42. 45. 48.]]], shape=(1, 4, 4), dtype=float32) # .. expands to all unspecified axes. op_shape(x@r[]) ## shape(1, 4, 4, 3) op_shape(x@r[..]) ## shape(1, 4, 4, 3) op_shape(x@r[1, ..]) ## shape(4, 4, 3) op_shape(x@r[1, .., 1, 1]) ## shape(4) op_shape(x@r[1, 1, 1, .., 1]) ## shape() # op_subset<- uses the same semantics, but note that not all tensors # support modification. E.g., TensorFlow constant tensors cannot be modified, # while TensorFlow Variables can be.  (x <- tensorflow::tf$Variable(matrix(1, nrow = 2, ncol = 3))) ## <tf.Variable 'Variable:0' shape=(2, 3) dtype=float64, numpy= ## array([[1., 1., 1.], ##        [1., 1., 1.]])> op_subset(x, 1) <- 9 x ## <tf.Variable 'UnreadVariable' shape=(2, 3) dtype=float64, numpy= ## array([[9., 9., 9.], ##        [1., 1., 1.]])> x@r[1,1] <- 33 x ## <tf.Variable 'UnreadVariable' shape=(2, 3) dtype=float64, numpy= ## array([[33.,  9.,  9.], ##        [ 1.,  1.,  1.]])>"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_subtract.html","id":null,"dir":"Reference","previous_headings":"","what":"Subtract arguments element-wise. — op_subtract","title":"Subtract arguments element-wise. — op_subtract","text":"Note function automatically called using R operator - tensor.","code":"x <- op_ones(c(3)) op_subtract(x, x) ## tf.Tensor([0. 0. 0.], shape=(3), dtype=float32) x - x ## tf.Tensor([0. 0. 0.], shape=(3), dtype=float32)"},{"path":"https://keras3.posit.co/dev/reference/op_subtract.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Subtract arguments element-wise. — op_subtract","text":"","code":"op_subtract(x1, x2)"},{"path":"https://keras3.posit.co/dev/reference/op_subtract.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Subtract arguments element-wise. — op_subtract","text":"x1 First input tensor. x2 Second input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_subtract.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Subtract arguments element-wise. — op_subtract","text":"Output tensor, element-wise difference x1 x2.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_sum.html","id":null,"dir":"Reference","previous_headings":"","what":"Sum of a tensor over the given axes. — op_sum","title":"Sum of a tensor over the given axes. — op_sum","text":"Sum tensor given axes.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_sum.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Sum of a tensor over the given axes. — op_sum","text":"","code":"op_sum(x, axis = NULL, keepdims = FALSE)"},{"path":"https://keras3.posit.co/dev/reference/op_sum.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Sum of a tensor over the given axes. — op_sum","text":"x Input tensor. axis Axis axes along sum computed. default compute sum flattened tensor. keepdims set TRUE, axes reduced left result dimensions size one.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_sum.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Sum of a tensor over the given axes. — op_sum","text":"Output tensor containing sum.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_svd.html","id":null,"dir":"Reference","previous_headings":"","what":"Computes the singular value decomposition of a matrix. — op_svd","title":"Computes the singular value decomposition of a matrix. — op_svd","text":"Computes singular value decomposition matrix.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_svd.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes the singular value decomposition of a matrix. — op_svd","text":"","code":"op_svd(x, full_matrices = TRUE, compute_uv = TRUE)"},{"path":"https://keras3.posit.co/dev/reference/op_svd.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Computes the singular value decomposition of a matrix. — op_svd","text":"x Input tensor shape (..., M, N). full_matrices Logical compute_uv Logical","code":""},{"path":"https://keras3.posit.co/dev/reference/op_svd.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Computes the singular value decomposition of a matrix. — op_svd","text":"list three tensors: tensor shape (..., M, M) containing left singular vectors, tensor shape (..., M, N) containing singular values tensor shape (..., N, N) containing right singular vectors.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_swapaxes.html","id":null,"dir":"Reference","previous_headings":"","what":"Interchange two axes of a tensor. — op_swapaxes","title":"Interchange two axes of a tensor. — op_swapaxes","text":"Interchange two axes tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_swapaxes.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Interchange two axes of a tensor. — op_swapaxes","text":"","code":"op_swapaxes(x, axis1, axis2)"},{"path":"https://keras3.posit.co/dev/reference/op_swapaxes.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Interchange two axes of a tensor. — op_swapaxes","text":"x Input tensor. axis1 First axis. axis2 Second axis.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_swapaxes.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Interchange two axes of a tensor. — op_swapaxes","text":"tensor axes swapped.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_switch.html","id":null,"dir":"Reference","previous_headings":"","what":"Apply exactly one of the branches given by index. — op_switch","title":"Apply exactly one of the branches given by index. — op_switch","text":"index bounds, clamped within bounds. semantics switch given roughly implementation:","code":"op_switch <- function(index, branches, ...) {   index <- op_clip(index, 1, length(branches))   branches[[index]](...) }"},{"path":"https://keras3.posit.co/dev/reference/op_switch.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Apply exactly one of the branches given by index. — op_switch","text":"","code":"op_switch(index, branches, ...)"},{"path":"https://keras3.posit.co/dev/reference/op_switch.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Apply exactly one of the branches given by index. — op_switch","text":"index integer scalar indicating branch function apply (1-based). branches list functions applied based index. ... Inputs whichever branch applied.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_switch.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Apply exactly one of the branches given by index. — op_switch","text":"outputs branch(...) branch selected based index.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_switch.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Apply exactly one of the branches given by index. — op_switch","text":"","code":"add_fn <- function(x, y) x + y + 100 subtract_fn <- function(x, y) x - y - 100 x <- op_array(2) y <- op_array(0.5) branches <- list(add_fn, subtract_fn) op_switch(0, branches, x, y) # + ## tf.Tensor(102.5, shape=(), dtype=float32) op_switch(1, branches, x, y) # + ## tf.Tensor(102.5, shape=(), dtype=float32) op_switch(2, branches, x, y) # - ## tf.Tensor(-98.5, shape=(), dtype=float32) op_switch(3, branches, x, y) # - ## tf.Tensor(-98.5, shape=(), dtype=float32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_take.html","id":null,"dir":"Reference","previous_headings":"","what":"Take elements from a tensor along an axis. — op_take","title":"Take elements from a tensor along an axis. — op_take","text":"Take elements tensor along axis.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_take.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Take elements from a tensor along an axis. — op_take","text":"","code":"op_take(x, indices, axis = NULL, zero_indexed = FALSE)"},{"path":"https://keras3.posit.co/dev/reference/op_take.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Take elements from a tensor along an axis. — op_take","text":"x Source tensor. indices indices values extract. axis axis select values. default, flattened input tensor used. zero_indexed TRUE, treats indices zero-based (0 encodes first position); FALSE (default), treats indices one-based (1 encodes first position).","code":""},{"path":"https://keras3.posit.co/dev/reference/op_take.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Take elements from a tensor along an axis. — op_take","text":"corresponding tensor values.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_take_along_axis.html","id":null,"dir":"Reference","previous_headings":"","what":"Select values from x at the 1-D indices along the given axis. — op_take_along_axis","title":"Select values from x at the 1-D indices along the given axis. — op_take_along_axis","text":"Select values x 1-D indices along given axis.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_take_along_axis.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Select values from x at the 1-D indices along the given axis. — op_take_along_axis","text":"","code":"op_take_along_axis(x, indices, axis = NULL, zero_indexed = FALSE)"},{"path":"https://keras3.posit.co/dev/reference/op_take_along_axis.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Select values from x at the 1-D indices along the given axis. — op_take_along_axis","text":"x Source tensor. indices indices values extract. axis axis select values. default, flattened input tensor used. zero_indexed TRUE, treats indices zero-based (0 encodes first position); FALSE (default), treats indices one-based (1 encodes first position).","code":""},{"path":"https://keras3.posit.co/dev/reference/op_take_along_axis.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Select values from x at the 1-D indices along the given axis. — op_take_along_axis","text":"corresponding tensor values.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_tan.html","id":null,"dir":"Reference","previous_headings":"","what":"Compute tangent, element-wise. — op_tan","title":"Compute tangent, element-wise. — op_tan","text":"Compute tangent, element-wise.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_tan.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Compute tangent, element-wise. — op_tan","text":"","code":"op_tan(x)"},{"path":"https://keras3.posit.co/dev/reference/op_tan.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Compute tangent, element-wise. — op_tan","text":"x Input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_tan.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Compute tangent, element-wise. — op_tan","text":"Output tensor shape x.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_tanh.html","id":null,"dir":"Reference","previous_headings":"","what":"Hyperbolic tangent, element-wise. — op_tanh","title":"Hyperbolic tangent, element-wise. — op_tanh","text":"Hyperbolic tangent, element-wise.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_tanh.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Hyperbolic tangent, element-wise. — op_tanh","text":"","code":"op_tanh(x)"},{"path":"https://keras3.posit.co/dev/reference/op_tanh.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Hyperbolic tangent, element-wise. — op_tanh","text":"x Input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_tanh.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Hyperbolic tangent, element-wise. — op_tanh","text":"Output tensor shape x.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_tanh_shrink.html","id":null,"dir":"Reference","previous_headings":"","what":"Applies the tanh shrink function element-wise. — op_tanh_shrink","title":"Applies the tanh shrink function element-wise. — op_tanh_shrink","text":"defined : f(x) = x - tanh(x).","code":""},{"path":"https://keras3.posit.co/dev/reference/op_tanh_shrink.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Applies the tanh shrink function element-wise. — op_tanh_shrink","text":"","code":"op_tanh_shrink(x)"},{"path":"https://keras3.posit.co/dev/reference/op_tanh_shrink.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Applies the tanh shrink function element-wise. — op_tanh_shrink","text":"x Input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_tanh_shrink.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Applies the tanh shrink function element-wise. — op_tanh_shrink","text":"Output tensor shape x, element transformed according tanh shrink operation.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_tanh_shrink.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Applies the tanh shrink function element-wise. — op_tanh_shrink","text":"","code":"x <- op_array(c(-1., 0., 1.)) op_tanh_shrink(x) ## tf.Tensor([-0.23840582  0.          0.23840582], shape=(3), dtype=float32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_tensordot.html","id":null,"dir":"Reference","previous_headings":"","what":"Compute the tensor dot product along specified axes. — op_tensordot","title":"Compute the tensor dot product along specified axes. — op_tensordot","text":"Compute tensor dot product along specified axes.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_tensordot.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Compute the tensor dot product along specified axes. — op_tensordot","text":"","code":"op_tensordot(x1, x2, axes = 3L)"},{"path":"https://keras3.posit.co/dev/reference/op_tensordot.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Compute the tensor dot product along specified axes. — op_tensordot","text":"x1 First tensor. x2 Second tensor. axes integer, N, sum last N axes x1 first N axes x2 order. sizes corresponding axes must match. , list axes summed , first sequence applying x1, second x2. sequences must length.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_tensordot.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Compute the tensor dot product along specified axes. — op_tensordot","text":"tensor dot product inputs.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_threshold.html","id":null,"dir":"Reference","previous_headings":"","what":"Threshold activation function. — op_threshold","title":"Threshold activation function. — op_threshold","text":"function thresholds input x follows: f(x) = x x > threshold, f(x) = default_value otherwise.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_threshold.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Threshold activation function. — op_threshold","text":"","code":"op_threshold(x, threshold, default_value)"},{"path":"https://keras3.posit.co/dev/reference/op_threshold.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Threshold activation function. — op_threshold","text":"x Input tensor. threshold value decides retain replace x. default_value Value assign x <= threshold.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_threshold.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Threshold activation function. — op_threshold","text":"tensor shape x.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_threshold.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Threshold activation function. — op_threshold","text":"","code":"x <- op_array(c(-1.0, 0.0, 1.0, 2.0)) op_threshold(x, 1, 0) ## tf.Tensor([0. 0. 0. 2.], shape=(4), dtype=float32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_tile.html","id":null,"dir":"Reference","previous_headings":"","what":"Repeat x the number of times given by repeats. — op_tile","title":"Repeat x the number of times given by repeats. — op_tile","text":"repeats length d, result dimension max(d, x.ndim). x.ndim < d, x promoted d-dimensional prepending new axes. x.ndim > d, repeats promoted x.ndim prepending 1's .","code":""},{"path":"https://keras3.posit.co/dev/reference/op_tile.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Repeat x the number of times given by repeats. — op_tile","text":"","code":"op_tile(x, repeats)"},{"path":"https://keras3.posit.co/dev/reference/op_tile.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Repeat x the number of times given by repeats. — op_tile","text":"x Input tensor. repeats number repetitions x along axis.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_tile.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Repeat x the number of times given by repeats. — op_tile","text":"tiled output tensor.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_top_k.html","id":null,"dir":"Reference","previous_headings":"","what":"Finds the top-k values and their indices in a tensor. — op_top_k","title":"Finds the top-k values and their indices in a tensor. — op_top_k","text":"Finds top-k values indices tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_top_k.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Finds the top-k values and their indices in a tensor. — op_top_k","text":"","code":"op_top_k(x, k, sorted = TRUE, zero_indexed = FALSE)"},{"path":"https://keras3.posit.co/dev/reference/op_top_k.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Finds the top-k values and their indices in a tensor. — op_top_k","text":"x Input tensor. k integer representing number top elements retrieve. sorted boolean indicating whether sort output descending order. Defaults TRUE. zero_indexed TRUE, returned indices zero-based (0 encodes first position); FALSE (default), returned indices one-based (1 encodes first position).","code":""},{"path":"https://keras3.posit.co/dev/reference/op_top_k.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Finds the top-k values and their indices in a tensor. — op_top_k","text":"list containing two tensors. first tensor contains top-k values, second tensor contains indices top-k values input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_top_k.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Finds the top-k values and their indices in a tensor. — op_top_k","text":"","code":"x <- op_array(c(5, 2, 7, 1, 9, 3), \"int32\") op_top_k(x, k = 3) ## $values ## tf.Tensor([9 7 5], shape=(3), dtype=int32) ## ## $indices ## tf.Tensor([5 3 1], shape=(3), dtype=int32) c(values, indices) %<-% op_top_k(x, k = 3) values ## tf.Tensor([9 7 5], shape=(3), dtype=int32) indices ## tf.Tensor([5 3 1], shape=(3), dtype=int32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_trace.html","id":null,"dir":"Reference","previous_headings":"","what":"Return the sum along diagonals of the tensor. — op_trace","title":"Return the sum along diagonals of the tensor. — op_trace","text":"x 2-D, sum along diagonal given offset returned, .e., sum elements x[, +offset] . two dimensions, axes specified axis1 axis2 used determine 2-D sub-arrays whose traces returned. shape resulting tensor x axis1 axis2 removed.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_trace.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Return the sum along diagonals of the tensor. — op_trace","text":"","code":"op_trace(x, offset = 0L, axis1 = 1L, axis2 = 2L)"},{"path":"https://keras3.posit.co/dev/reference/op_trace.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Return the sum along diagonals of the tensor. — op_trace","text":"x Input tensor. offset Offset diagonal main diagonal. Can positive negative. Defaults 0. axis1 Axis used first axis 2-D sub-arrays. Defaults 1. (first axis). axis2 Axis used second axis 2-D sub-arrays. Defaults 2. (second axis).","code":""},{"path":"https://keras3.posit.co/dev/reference/op_trace.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Return the sum along diagonals of the tensor. — op_trace","text":"x 2-D, sum diagonal returned. x larger dimensions, tensor sums along diagonals returned.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_transpose.html","id":null,"dir":"Reference","previous_headings":"","what":"Returns a tensor with axes transposed. — op_transpose","title":"Returns a tensor with axes transposed. — op_transpose","text":"Returns tensor axes transposed.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_transpose.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Returns a tensor with axes transposed. — op_transpose","text":"","code":"op_transpose(x, axes = NULL)"},{"path":"https://keras3.posit.co/dev/reference/op_transpose.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Returns a tensor with axes transposed. — op_transpose","text":"x Input tensor. axes Sequence integers. Permutation dimensions x. default, order axes reversed.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_transpose.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Returns a tensor with axes transposed. — op_transpose","text":"x axes permuted.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_tri.html","id":null,"dir":"Reference","previous_headings":"","what":"Return a tensor with ones at and below a diagonal and zeros elsewhere. — op_tri","title":"Return a tensor with ones at and below a diagonal and zeros elsewhere. — op_tri","text":"Return tensor ones diagonal zeros elsewhere.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_tri.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Return a tensor with ones at and below a diagonal and zeros elsewhere. — op_tri","text":"","code":"op_tri(N, M = NULL, k = 0L, dtype = NULL)"},{"path":"https://keras3.posit.co/dev/reference/op_tri.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Return a tensor with ones at and below a diagonal and zeros elsewhere. — op_tri","text":"N Number rows tensor. M Number columns tensor. k sub-diagonal array filled. k = 0 main diagonal, k < 0 , k > 0 . default 0. dtype Data type returned tensor. default \"float32\".","code":""},{"path":"https://keras3.posit.co/dev/reference/op_tri.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Return a tensor with ones at and below a diagonal and zeros elsewhere. — op_tri","text":"Tensor lower triangle filled ones zeros elsewhere. T[, j] == 1 j <= + k, 0 otherwise.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_tril.html","id":null,"dir":"Reference","previous_headings":"","what":"Return lower triangle of a tensor. — op_tril","title":"Return lower triangle of a tensor. — op_tril","text":"tensors ndim exceeding 2, tril apply final two axes.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_tril.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Return lower triangle of a tensor. — op_tril","text":"","code":"op_tril(x, k = 0L)"},{"path":"https://keras3.posit.co/dev/reference/op_tril.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Return lower triangle of a tensor. — op_tril","text":"x Input tensor. k Diagonal zero elements. Defaults 0. main diagonal. k < 0 , k > 0 .","code":""},{"path":"https://keras3.posit.co/dev/reference/op_tril.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Return lower triangle of a tensor. — op_tril","text":"Lower triangle x, shape data type x.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_triu.html","id":null,"dir":"Reference","previous_headings":"","what":"Return upper triangle of a tensor. — op_triu","title":"Return upper triangle of a tensor. — op_triu","text":"tensors ndim exceeding 2, triu apply final two axes.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_triu.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Return upper triangle of a tensor. — op_triu","text":"","code":"op_triu(x, k = 0L)"},{"path":"https://keras3.posit.co/dev/reference/op_triu.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Return upper triangle of a tensor. — op_triu","text":"x Input tensor. k Diagonal zero elements. Defaults 0. main diagonal. k < 0 , k > 0 .","code":""},{"path":"https://keras3.posit.co/dev/reference/op_triu.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Return upper triangle of a tensor. — op_triu","text":"Upper triangle x, shape data type x.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_trunc.html","id":null,"dir":"Reference","previous_headings":"","what":"Return the truncated value of the input, element-wise. — op_trunc","title":"Return the truncated value of the input, element-wise. — op_trunc","text":"truncated value scalar x nearest integer closer zero x . short, fractional part signed number x discarded.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_trunc.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Return the truncated value of the input, element-wise. — op_trunc","text":"","code":"op_trunc(x)"},{"path":"https://keras3.posit.co/dev/reference/op_trunc.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Return the truncated value of the input, element-wise. — op_trunc","text":"x Input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_trunc.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Return the truncated value of the input, element-wise. — op_trunc","text":"truncated value element x.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_trunc.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Return the truncated value of the input, element-wise. — op_trunc","text":"","code":"x <- op_array(c(-1.7, -1.5, -0.2, 0.2, 1.5, 1.7, 2.0)) op_trunc(x) ## tf.Tensor([-1. -1. -0.  0.  1.  1.  2.], shape=(7), dtype=float32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_unravel_index.html","id":null,"dir":"Reference","previous_headings":"","what":"Convert flat indices to coordinate arrays in a given array shape. — op_unravel_index","title":"Convert flat indices to coordinate arrays in a given array shape. — op_unravel_index","text":"Convert flat indices coordinate arrays given array shape.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_unravel_index.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Convert flat indices to coordinate arrays in a given array shape. — op_unravel_index","text":"","code":"op_unravel_index(indices, shape)"},{"path":"https://keras3.posit.co/dev/reference/op_unravel_index.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Convert flat indices to coordinate arrays in a given array shape. — op_unravel_index","text":"indices integer array integers representing flat indices. shape shape array unravel .","code":""},{"path":"https://keras3.posit.co/dev/reference/op_unravel_index.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Convert flat indices to coordinate arrays in a given array shape. — op_unravel_index","text":"Tuple arrays dimension unraveled indices.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_unravel_index.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Convert flat indices to coordinate arrays in a given array shape. — op_unravel_index","text":"","code":"indices <- c(1, 5) shape <- array(c(3, 3)) op_unravel_index(indices, shape) ## [[1]] ## tf.Tensor([0 1], shape=(2), dtype=int32) ## ## [[2]] ## tf.Tensor([0 1], shape=(2), dtype=int32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_unstack.html","id":null,"dir":"Reference","previous_headings":"","what":"Unpacks the given dimension of a rank-R tensor into rank-(R-1) tensors. — op_unstack","title":"Unpacks the given dimension of a rank-R tensor into rank-(R-1) tensors. — op_unstack","text":"Unpacks given dimension rank-R tensor rank-(R-1) tensors.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_unstack.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Unpacks the given dimension of a rank-R tensor into rank-(R-1) tensors. — op_unstack","text":"","code":"op_unstack(x, num = NULL, axis = 1L)"},{"path":"https://keras3.posit.co/dev/reference/op_unstack.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Unpacks the given dimension of a rank-R tensor into rank-(R-1) tensors. — op_unstack","text":"x input tensor. num length dimension axis. Automatically inferred NULL. axis axis along unpack.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_unstack.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Unpacks the given dimension of a rank-R tensor into rank-(R-1) tensors. — op_unstack","text":"list tensors unpacked along given axis.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_unstack.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Unpacks the given dimension of a rank-R tensor into rank-(R-1) tensors. — op_unstack","text":"[3, 4))]: R:3,%204))","code":"x <- op_array(rbind(c(1, 2),                    c(3, 4))) op_unstack(x, axis=1) ## [[1]] ## tf.Tensor([1. 2.], shape=(2), dtype=float64) ## ## [[2]] ## tf.Tensor([3. 4.], shape=(2), dtype=float64) op_unstack(x, axis=2) ## [[1]] ## tf.Tensor([1. 3.], shape=(2), dtype=float64) ## ## [[2]] ## tf.Tensor([2. 4.], shape=(2), dtype=float64) all.equal(op_unstack(x),           op_unstack(x, axis = 1)) ## [1] TRUE all.equal(op_unstack(x, axis = -1),           op_unstack(x, axis = 2)) ## [1] TRUE # [array([1, 2)), array([3, 4))]"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_var.html","id":null,"dir":"Reference","previous_headings":"","what":"Compute the variance along the specified axes. — op_var","title":"Compute the variance along the specified axes. — op_var","text":"Compute variance along specified axes.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_var.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Compute the variance along the specified axes. — op_var","text":"","code":"op_var(x, axis = NULL, keepdims = FALSE)"},{"path":"https://keras3.posit.co/dev/reference/op_var.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Compute the variance along the specified axes. — op_var","text":"x Input tensor. axis Axis axes along variance computed. default compute variance flattened tensor. keepdims set TRUE, axes reduced left result dimensions size one.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_var.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Compute the variance along the specified axes. — op_var","text":"Output tensor containing variance.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_vdot.html","id":null,"dir":"Reference","previous_headings":"","what":"Return the dot product of two vectors. — op_vdot","title":"Return the dot product of two vectors. — op_vdot","text":"first argument complex, complex conjugate first argument used calculation dot product. Multidimensional tensors flattened dot product taken.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_vdot.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Return the dot product of two vectors. — op_vdot","text":"","code":"op_vdot(x1, x2)"},{"path":"https://keras3.posit.co/dev/reference/op_vdot.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Return the dot product of two vectors. — op_vdot","text":"x1 First input tensor. complex, complex conjugate taken calculation dot product. x2 Second input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_vdot.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Return the dot product of two vectors. — op_vdot","text":"Output tensor.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_vectorize.html","id":null,"dir":"Reference","previous_headings":"","what":"Turn a function into a vectorized function. — op_vectorize","title":"Turn a function into a vectorized function. — op_vectorize","text":"Turn function vectorized function.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_vectorize.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Turn a function into a vectorized function. — op_vectorize","text":"","code":"op_vectorize(func, ..., excluded = NULL, signature = NULL)"},{"path":"https://keras3.posit.co/dev/reference/op_vectorize.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Turn a function into a vectorized function. — op_vectorize","text":"func Callable single tensor argument. ... forward/backward compatability. excluded Optional set integers representing positional arguments function vectorized. passed directly func unmodified. signature Optional generalized universal function signature, e.g., \"(m,n),(n)->(m)\" vectorized matrix-vector multiplication. provided, func called (expected return) arrays shapes given size corresponding core dimensions. default, func assumed take scalar tensors input output.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_vectorize.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Turn a function into a vectorized function. — op_vectorize","text":"new function applies func every element input along axis 1 (batch axis, first axis).","code":""},{"path":"https://keras3.posit.co/dev/reference/op_vectorize.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Turn a function into a vectorized function. — op_vectorize","text":"","code":"# currently does not work w/ tensorflow backend if(config_backend() != \"tensorflow\") {    myfunc <- function(a, b) a + b    vfunc <- op_vectorize(myfunc)   y <- vfunc(c(1, 2, 3, 4), 2)   print(y)   # with Jax backend, y is:   # Array([3., 4., 5., 6.], dtype=float32) }"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_vectorized_map.html","id":null,"dir":"Reference","previous_headings":"","what":"Parallel map of function f on the first axis of tensor(s) elements. — op_vectorized_map","title":"Parallel map of function f on the first axis of tensor(s) elements. — op_vectorized_map","text":"Schematically, op_vectorized_map() maps first dimension provided tensors. elements list tensors, tensors required size first dimension, iterated together.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_vectorized_map.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Parallel map of function f on the first axis of tensor(s) elements. — op_vectorized_map","text":"","code":"op_vectorized_map(elements, f)"},{"path":"https://keras3.posit.co/dev/reference/op_vectorized_map.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Parallel map of function f on the first axis of tensor(s) elements. — op_vectorized_map","text":"elements see description f function taking either tensor, list tensors.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_vectorized_map.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Parallel map of function f on the first axis of tensor(s) elements. — op_vectorized_map","text":"tensor list tensors, result mapping f across elements.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_vectorized_map.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Parallel map of function f on the first axis of tensor(s) elements. — op_vectorized_map","text":"Note f may traced compiled. Meaning, R function may evaluated symbolic tensors using Jax TensorFlow backends, eager tensors. See output str() examples:","code":"(x <- op_arange(12L) |> op_reshape(c(3, 4))) ## tf.Tensor( ## [[ 1  2  3  4] ##  [ 5  6  7  8] ##  [ 9 10 11 12]], shape=(3, 4), dtype=int32) x |> op_vectorized_map(\\(row) {row + 10}) ## tf.Tensor( ## [[11 12 13 14] ##  [15 16 17 18] ##  [19 20 21 22]], shape=(3, 4), dtype=int32) list(x, x, x) |> op_vectorized_map(\\(rows) Reduce(`+`, rows)) ## tf.Tensor( ## [[ 3  6  9 12] ##  [15 18 21 24] ##  [27 30 33 36]], shape=(3, 4), dtype=int32) # simplest case, map f over rows of x, # where .x is 1 row of x input <- x output <- op_vectorized_map(input, function(.x) {   str(.x)   .x + 10 }) ## <tf.Tensor 'loop_body/GatherV2:0' shape=(4) dtype=int32> output ## tf.Tensor( ## [[11 12 13 14] ##  [15 16 17 18] ##  [19 20 21 22]], shape=(3, 4), dtype=int32) # map f over two tensors simultaneously. Here, # `.x` is a list of two # tensors. The return values from each call of `f(row)` are stacked to form the # final output input <- list(x, x) output <- op_vectorized_map(input, function(.x) {   str(.x)   .x[[1]] + 10 }) ## List of 2 ##  $ :<tf.Tensor 'loop_body/GatherV2:0' shape=(4) dtype=int32> ##  $ :<tf.Tensor 'loop_body/GatherV2_1:0' shape=(4) dtype=int32> output ## tf.Tensor( ## [[11 12 13 14] ##  [15 16 17 18] ##  [19 20 21 22]], shape=(3, 4), dtype=int32) # same as above, but now returning two tensors in the final output output <- op_vectorized_map(input, function(.x) {   str(.x)   c(.x1, .x2) %<-% .x   list(.x1+10, .x2+20) }) ## List of 2 ##  $ :<tf.Tensor 'loop_body/GatherV2:0' shape=(4) dtype=int32> ##  $ :<tf.Tensor 'loop_body/GatherV2_1:0' shape=(4) dtype=int32> output ## [[1]] ## tf.Tensor( ## [[11 12 13 14] ##  [15 16 17 18] ##  [19 20 21 22]], shape=(3, 4), dtype=int32) ## ## [[2]] ## tf.Tensor( ## [[21 22 23 24] ##  [25 26 27 28] ##  [29 30 31 32]], shape=(3, 4), dtype=int32) # passing named lists. # WARNING: if passing a named list, the order of elements of `.x` supplied # to `f` is not stable. Only retrieve elements by name. input <- list(name1 = x, name2 = x) output <- op_vectorized_map(input, function(.x) {   str(.x)   list(outname1 = .x$name1 + 10,        outname2 = .x$name2 + 20) }) ## List of 2 ##  $ name1:<tf.Tensor 'loop_body/GatherV2:0' shape=(4) dtype=int32> ##  $ name2:<tf.Tensor 'loop_body/GatherV2_1:0' shape=(4) dtype=int32> output ## $outname1 ## tf.Tensor( ## [[11 12 13 14] ##  [15 16 17 18] ##  [19 20 21 22]], shape=(3, 4), dtype=int32) ## ## $outname2 ## tf.Tensor( ## [[21 22 23 24] ##  [25 26 27 28] ##  [29 30 31 32]], shape=(3, 4), dtype=int32) # passing a tuple() is equivalent to passing an unnamed list() input <- tuple(x, x) output <- op_vectorized_map(input, function(.x) {   str(.x)   list(.x[[1]] + 10) }) ## List of 2 ##  $ :<tf.Tensor 'loop_body/GatherV2:0' shape=(4) dtype=int32> ##  $ :<tf.Tensor 'loop_body/GatherV2_1:0' shape=(4) dtype=int32> output ## [[1]] ## tf.Tensor( ## [[11 12 13 14] ##  [15 16 17 18] ##  [19 20 21 22]], shape=(3, 4), dtype=int32)"},{"path":"https://keras3.posit.co/dev/reference/op_vectorized_map.html","id":"debugging-f","dir":"Reference","previous_headings":"","what":"Debugging f","title":"Parallel map of function f on the first axis of tensor(s) elements. — op_vectorized_map","text":"Even eager contexts, op_vectorized_map() may trace f. case, want eagerly debug f (e.g., browser()), can swap manual (slow) implementation op_vectorized_map(). Note example debug implementation handle edge cases op_vectorized_map(), particular, f returns structure multiple tensors.","code":"op_vectorized_map_debug <- function(elements, fn) {    if (!is.list(elements)) {     # `elements` is a single tensor     batch_size <- op_shape(elements)[[1]]     out <- elements |>       op_split(batch_size) |>       lapply(fn) |>       op_stack()     return(out)   }    # `elements` is a list of tensors   batch_size <- elements[[1]] |> op_shape() |> _[[1]]   elements |>     lapply(\\(e) op_split(e, batch_size)) |>     zip_lists() |>     lapply(fn) |>     op_stack()  }"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_view_as_complex.html","id":null,"dir":"Reference","previous_headings":"","what":"Convert a real tensor with two channels into a complex tensor. — op_view_as_complex","title":"Convert a real tensor with two channels into a complex tensor. — op_view_as_complex","text":"Expects real-valued tensor whose last dimension size 2, holding real imaginary parts. Returns corresponding complex tensor last dimension removed.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_view_as_complex.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Convert a real tensor with two channels into a complex tensor. — op_view_as_complex","text":"","code":"op_view_as_complex(x)"},{"path":"https://keras3.posit.co/dev/reference/op_view_as_complex.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Convert a real tensor with two channels into a complex tensor. — op_view_as_complex","text":"x Real-valued tensor whose trailing dimension encodes complex components.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_view_as_complex.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Convert a real tensor with two channels into a complex tensor. — op_view_as_complex","text":"complex tensor shape op_shape(x)[-length(op_shape(x))].","code":""},{"path":"https://keras3.posit.co/dev/reference/op_view_as_complex.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Convert a real tensor with two channels into a complex tensor. — op_view_as_complex","text":"","code":"x <- op_array(matrix(c(1, 2, 3, 4), nrow = 2, byrow = TRUE)) op_view_as_complex(x) ## tf.Tensor([1.+2.j 3.+4.j], shape=(2), dtype=complex64)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_view_as_real.html","id":null,"dir":"Reference","previous_headings":"","what":"Convert a complex tensor into a stacked real representation. — op_view_as_real","title":"Convert a complex tensor into a stacked real representation. — op_view_as_real","text":"Produces real-valued tensor last dimension gathers real imaginary parts complex input.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_view_as_real.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Convert a complex tensor into a stacked real representation. — op_view_as_real","text":"","code":"op_view_as_real(x)"},{"path":"https://keras3.posit.co/dev/reference/op_view_as_real.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Convert a complex tensor into a stacked real representation. — op_view_as_real","text":"x Complex-valued tensor converted.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_view_as_real.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Convert a complex tensor into a stacked real representation. — op_view_as_real","text":"real tensor shape c(op_shape(x), 2) containing real imaginary parts.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_view_as_real.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Convert a complex tensor into a stacked real representation. — op_view_as_real","text":"","code":"x <- op_array(matrix(c(1, 2, 3, 4), nrow = 2, byrow = TRUE)) z <- op_view_as_complex(x) op_view_as_real(z) ## tf.Tensor( ## [[1. 2.] ##  [3. 4.]], shape=(2, 2), dtype=float32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_vstack.html","id":null,"dir":"Reference","previous_headings":"","what":"Stack tensors in sequence vertically (row wise). — op_vstack","title":"Stack tensors in sequence vertically (row wise). — op_vstack","text":"Stack tensors sequence vertically (row wise).","code":""},{"path":"https://keras3.posit.co/dev/reference/op_vstack.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Stack tensors in sequence vertically (row wise). — op_vstack","text":"","code":"op_vstack(xs, ...)"},{"path":"https://keras3.posit.co/dev/reference/op_vstack.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Stack tensors in sequence vertically (row wise). — op_vstack","text":"xs, ... Sequence tensors.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_vstack.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Stack tensors in sequence vertically (row wise). — op_vstack","text":"Tensor formed stacking given tensors.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_where.html","id":null,"dir":"Reference","previous_headings":"","what":"Return elements chosen from x1 or x2 depending on condition. — op_where","title":"Return elements chosen from x1 or x2 depending on condition. — op_where","text":"Return elements chosen x1 x2 depending condition.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_where.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Return elements chosen from x1 or x2 depending on condition. — op_where","text":"","code":"op_where(condition, x1 = NULL, x2 = NULL)"},{"path":"https://keras3.posit.co/dev/reference/op_where.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Return elements chosen from x1 or x2 depending on condition. — op_where","text":"condition TRUE, yield x1, otherwise yield x2. x1 Values choose condition TRUE. x2 Values choose condition FALSE.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_where.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Return elements chosen from x1 or x2 depending on condition. — op_where","text":"tensor elements x1 condition TRUE, elements x2 condition FALSE.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_while_loop.html","id":null,"dir":"Reference","previous_headings":"","what":"While loop implementation. — op_while_loop","title":"While loop implementation. — op_while_loop","text":"loop implementation.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_while_loop.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"While loop implementation. — op_while_loop","text":"","code":"op_while_loop(cond, body, loop_vars, maximum_iterations = NULL)"},{"path":"https://keras3.posit.co/dev/reference/op_while_loop.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"While loop implementation. — op_while_loop","text":"cond callable represents termination condition loop. Must accept loop_vars like structure argument. loop_vars tuple unnamed list, element loop_vars passed positionally callable. body callable represents loop body. Must accept loop_vars like structure argument, return update value structure. loop_vars tuple unnamed list, element loop_vars passed positionally callable. loop_vars arbitrary nested structure tensor state persist across loop iterations. maximum_iterations Optional maximum number iterations loop run. provided, cond output -ed additional condition ensuring number iterations executed greater maximum_iterations.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_while_loop.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"While loop implementation. — op_while_loop","text":"list tensors, shape dtype loop_vars.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_while_loop.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"While loop implementation. — op_while_loop","text":"","code":"i <- 0 loop_vars <- list(i)  # cond() must return a scalar bool cond <- function(i) i < 10L  # body must return same shape as loop_vars body <- function(i) list(i + 1L)  op_while_loop(cond, body, loop_vars) ## [[1]] ## tf.Tensor(10.0, shape=(), dtype=float32) x <- 0; y <- 1 cond <- \\(x, y) x < 10 body <- \\(x, y) list(x+1, y+1) op_while_loop(cond, body, list(x, y)) ## [[1]] ## tf.Tensor(10.0, shape=(), dtype=float32) ## ## [[2]] ## tf.Tensor(11.0, shape=(), dtype=float32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_zeros.html","id":null,"dir":"Reference","previous_headings":"","what":"Return a new tensor of given shape and type, filled with zeros. — op_zeros","title":"Return a new tensor of given shape and type, filled with zeros. — op_zeros","text":"Return new tensor given shape type, filled zeros.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_zeros.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Return a new tensor of given shape and type, filled with zeros. — op_zeros","text":"","code":"op_zeros(shape, dtype = NULL)"},{"path":"https://keras3.posit.co/dev/reference/op_zeros.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Return a new tensor of given shape and type, filled with zeros. — op_zeros","text":"shape Shape new tensor. dtype Desired data type tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_zeros.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Return a new tensor of given shape and type, filled with zeros. — op_zeros","text":"Tensor zeros given shape dtype.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_zeros_like.html","id":null,"dir":"Reference","previous_headings":"","what":"Return a tensor of zeros with the same shape and type as x. — op_zeros_like","title":"Return a tensor of zeros with the same shape and type as x. — op_zeros_like","text":"Return tensor zeros shape type x.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_zeros_like.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Return a tensor of zeros with the same shape and type as x. — op_zeros_like","text":"","code":"op_zeros_like(x, dtype = NULL)"},{"path":"https://keras3.posit.co/dev/reference/op_zeros_like.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Return a tensor of zeros with the same shape and type as x. — op_zeros_like","text":"x Input tensor. dtype Overrides data type result.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_zeros_like.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Return a tensor of zeros with the same shape and type as x. — op_zeros_like","text":"tensor zeros shape type x.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/optimizer_adadelta.html","id":null,"dir":"Reference","previous_headings":"","what":"Optimizer that implements the Adadelta algorithm. — optimizer_adadelta","title":"Optimizer that implements the Adadelta algorithm. — optimizer_adadelta","text":"Adadelta optimization stochastic gradient descent method based adaptive learning rate per dimension address two drawbacks: continual decay learning rates throughout training. need manually selected global learning rate. Adadelta robust extension Adagrad adapts learning rates based moving window gradient updates, instead accumulating past gradients. way, Adadelta continues learning even many updates done. Compared Adagrad, original version Adadelta set initial learning rate. version, initial learning rate can set, Keras optimizers.","code":""},{"path":"https://keras3.posit.co/dev/reference/optimizer_adadelta.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Optimizer that implements the Adadelta algorithm. — optimizer_adadelta","text":"","code":"optimizer_adadelta(   learning_rate = 0.001,   rho = 0.95,   epsilon = 1e-07,   weight_decay = NULL,   clipnorm = NULL,   clipvalue = NULL,   global_clipnorm = NULL,   use_ema = FALSE,   ema_momentum = 0.99,   ema_overwrite_frequency = NULL,   name = \"adadelta\",   ...,   loss_scale_factor = NULL,   gradient_accumulation_steps = NULL )"},{"path":"https://keras3.posit.co/dev/reference/optimizer_adadelta.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Optimizer that implements the Adadelta algorithm. — optimizer_adadelta","text":"learning_rate float, [LearningRateSchedule()] instance, callable takes arguments returns actual value use. learning rate. Defaults 0.001. Note Adadelta tends benefit higher initial learning rate values compared optimizers. match exact form original paper, use 1.0. rho floating point value. decay rate. Defaults 0.95. epsilon Small floating point value maintaining numerical stability. weight_decay Float. set, weight decay applied. clipnorm Float. set, gradient weight individually clipped norm higher value. clipvalue Float. set, gradient weight clipped higher value. global_clipnorm Float. set, gradient weights clipped global norm higher value. use_ema Boolean, defaults FALSE. TRUE, exponential moving average (EMA) applied. EMA consists computing exponential moving average weights model (weight values change training batch), periodically overwriting weights moving average. ema_momentum Float, defaults 0.99. used use_ema = TRUE. momentum use computing EMA model's weights: new_average = ema_momentum * old_average + (1 - ema_momentum) * current_variable_value. ema_overwrite_frequency Int NULL, defaults NULL. used use_ema = TRUE. Every ema_overwrite_frequency steps iterations, overwrite model variable moving average. NULL, optimizer overwrite model variables middle training, need explicitly overwrite variables end training calling optimizer$finalize_variable_values() (updates model variables -place). using built-fit() training loop, happens automatically last epoch, need anything. name String. name use momentum accumulator weights created optimizer. ... forward/backward compatability. loss_scale_factor Float NULL. float, scale factor multiplied loss computing gradients, inverse scale factor multiplied gradients updating variables. Useful preventing underflow mixed precision training. Alternately, optimizer_loss_scale() automatically set loss scale factor. gradient_accumulation_steps Int NULL. int, model optimizer variables updated every step; instead updated every gradient_accumulation_steps steps, using average value gradients since last update. known \"gradient accumulation\". can useful batch size small, order reduce gradient noise update step. EMA frequency look \"accumulated\" iterations value (optimizer steps // gradient_accumulation_steps). Learning rate schedules look \"real\" iterations value (optimizer steps).","code":""},{"path":"https://keras3.posit.co/dev/reference/optimizer_adadelta.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Optimizer that implements the Adadelta algorithm. — optimizer_adadelta","text":"Optimizer instance","code":""},{"path":"https://keras3.posit.co/dev/reference/optimizer_adadelta.html","id":"reference","dir":"Reference","previous_headings":"","what":"Reference","title":"Optimizer that implements the Adadelta algorithm. — optimizer_adadelta","text":"Zeiler, 2012","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/optimizer_adafactor.html","id":null,"dir":"Reference","previous_headings":"","what":"Optimizer that implements the Adafactor algorithm. — optimizer_adafactor","title":"Optimizer that implements the Adafactor algorithm. — optimizer_adafactor","text":"Adafactor commonly used NLP tasks, advantage taking less memory saves partial information previous gradients. default argument setup based original paper (see reference). gradients dimension > 2, Adafactor optimizer delete last 2 dimensions separately accumulator variables.","code":""},{"path":"https://keras3.posit.co/dev/reference/optimizer_adafactor.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Optimizer that implements the Adafactor algorithm. — optimizer_adafactor","text":"","code":"optimizer_adafactor(   learning_rate = 0.001,   beta_2_decay = -0.8,   epsilon_1 = 1e-30,   epsilon_2 = 0.001,   clip_threshold = 1,   relative_step = TRUE,   weight_decay = NULL,   clipnorm = NULL,   clipvalue = NULL,   global_clipnorm = NULL,   use_ema = FALSE,   ema_momentum = 0.99,   ema_overwrite_frequency = NULL,   name = \"adafactor\",   ...,   loss_scale_factor = NULL,   gradient_accumulation_steps = NULL )"},{"path":"https://keras3.posit.co/dev/reference/optimizer_adafactor.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Optimizer that implements the Adafactor algorithm. — optimizer_adafactor","text":"learning_rate float, LearningRateSchedule() instance, callable takes arguments returns actual value use. learning rate. Defaults 0.001. beta_2_decay float, defaults -0.8. decay rate beta_2. epsilon_1 float, defaults 1e-30. small offset keep denominator away 0. epsilon_2 float, defaults 1e-3. small offset avoid learning rate becoming small time. clip_threshold float, defaults 1.0. Clipping threshold. part Adafactor algorithm, independent clipnorm, clipvalue, global_clipnorm. relative_step bool, defaults TRUE. learning_rate constant relative_step=TRUE, learning rate adjusted based current iterations. default learning rate decay Adafactor. weight_decay Float. set, weight decay applied. clipnorm Float. set, gradient weight individually clipped norm higher value. clipvalue Float. set, gradient weight clipped higher value. global_clipnorm Float. set, gradient weights clipped global norm higher value. use_ema Boolean, defaults FALSE. TRUE, exponential moving average (EMA) applied. EMA consists computing exponential moving average weights model (weight values change training batch), periodically overwriting weights moving average. ema_momentum Float, defaults 0.99. used use_ema = TRUE. momentum use computing EMA model's weights: new_average = ema_momentum * old_average + (1 - ema_momentum) * current_variable_value. ema_overwrite_frequency Int NULL, defaults NULL. used use_ema=TRUE. Every ema_overwrite_frequency steps iterations, overwrite model variable moving average. NULL, optimizer overwrite model variables middle training, need explicitly overwrite variables end training calling optimizer$finalize_variable_values() (updates model variables -place). using built-fit() training loop, happens automatically last epoch, need anything. name String. name use momentum accumulator weights created optimizer. ... forward/backward compatability. loss_scale_factor Float NULL. float, scale factor multiplied loss computing gradients, inverse scale factor multiplied gradients updating variables. Useful preventing underflow mixed precision training. Alternately, optimizer_loss_scale() automatically set loss scale factor. gradient_accumulation_steps Int NULL. int, model optimizer variables updated every step; instead updated every gradient_accumulation_steps steps, using average value gradients since last update. known \"gradient accumulation\". can useful batch size small, order reduce gradient noise update step. EMA frequency look \"accumulated\" iterations value (optimizer steps // gradient_accumulation_steps). Learning rate schedules look \"real\" iterations value (optimizer steps).","code":""},{"path":"https://keras3.posit.co/dev/reference/optimizer_adafactor.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Optimizer that implements the Adafactor algorithm. — optimizer_adafactor","text":"Optimizer instance","code":""},{"path":"https://keras3.posit.co/dev/reference/optimizer_adafactor.html","id":"reference","dir":"Reference","previous_headings":"","what":"Reference","title":"Optimizer that implements the Adafactor algorithm. — optimizer_adafactor","text":"Shazeer, Noam et al., 2018.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/optimizer_adagrad.html","id":null,"dir":"Reference","previous_headings":"","what":"Optimizer that implements the Adagrad algorithm. — optimizer_adagrad","title":"Optimizer that implements the Adagrad algorithm. — optimizer_adagrad","text":"Adagrad optimizer parameter-specific learning rates, adapted relative frequently parameter gets updated training. updates parameter receives, smaller updates.","code":""},{"path":"https://keras3.posit.co/dev/reference/optimizer_adagrad.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Optimizer that implements the Adagrad algorithm. — optimizer_adagrad","text":"","code":"optimizer_adagrad(   learning_rate = 0.001,   initial_accumulator_value = 0.1,   epsilon = 1e-07,   weight_decay = NULL,   clipnorm = NULL,   clipvalue = NULL,   global_clipnorm = NULL,   use_ema = FALSE,   ema_momentum = 0.99,   ema_overwrite_frequency = NULL,   name = \"adagrad\",   ...,   loss_scale_factor = NULL,   gradient_accumulation_steps = NULL )"},{"path":"https://keras3.posit.co/dev/reference/optimizer_adagrad.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Optimizer that implements the Adagrad algorithm. — optimizer_adagrad","text":"learning_rate float, LearningRateSchedule() instance, callable takes arguments returns actual value use. learning rate. Defaults 0.001. Note Adagrad tends benefit higher initial learning rate values compared optimizers. match exact form original paper, use 1.0. initial_accumulator_value Floating point value. Starting value accumulators (per-parameter momentum values). Must non-negative. epsilon Small floating point value maintaining numerical stability. weight_decay Float. set, weight decay applied. clipnorm Float. set, gradient weight individually clipped norm higher value. clipvalue Float. set, gradient weight clipped higher value. global_clipnorm Float. set, gradient weights clipped global norm higher value. use_ema Boolean, defaults FALSE. TRUE, exponential moving average (EMA) applied. EMA consists computing exponential moving average weights model (weight values change training batch), periodically overwriting weights moving average. ema_momentum Float, defaults 0.99. used use_ema=TRUE. momentum use computing EMA model's weights: new_average = ema_momentum * old_average + (1 - ema_momentum) * current_variable_value. ema_overwrite_frequency Int NULL, defaults NULL. used use_ema=TRUE. Every ema_overwrite_frequency steps iterations, overwrite model variable moving average. NULL, optimizer overwrite model variables middle training, need explicitly overwrite variables end training calling optimizer$finalize_variable_values() (updates model variables -place). using built-fit() training loop, happens automatically last epoch, need anything. name String. name use momentum accumulator weights created optimizer. ... forward/backward compatability. loss_scale_factor Float NULL. float, scale factor multiplied loss computing gradients, inverse scale factor multiplied gradients updating variables. Useful preventing underflow mixed precision training. Alternately, optimizer_loss_scale() automatically set loss scale factor. gradient_accumulation_steps Int NULL. int, model optimizer variables updated every step; instead updated every gradient_accumulation_steps steps, using average value gradients since last update. known \"gradient accumulation\". can useful batch size small, order reduce gradient noise update step. EMA frequency look \"accumulated\" iterations value (optimizer steps // gradient_accumulation_steps). Learning rate schedules look \"real\" iterations value (optimizer steps).","code":""},{"path":"https://keras3.posit.co/dev/reference/optimizer_adagrad.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Optimizer that implements the Adagrad algorithm. — optimizer_adagrad","text":"Optimizer instance","code":""},{"path":"https://keras3.posit.co/dev/reference/optimizer_adagrad.html","id":"reference","dir":"Reference","previous_headings":"","what":"Reference","title":"Optimizer that implements the Adagrad algorithm. — optimizer_adagrad","text":"Duchi et al., 2011.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/optimizer_adam.html","id":null,"dir":"Reference","previous_headings":"","what":"Optimizer that implements the Adam algorithm. — optimizer_adam","title":"Optimizer that implements the Adam algorithm. — optimizer_adam","text":"Adam optimization stochastic gradient descent method based adaptive estimation first-order second-order moments. According Kingma et al., 2014, method \"computationally efficient, little memory requirement, invariant diagonal rescaling gradients, well suited problems large terms data/parameters\".","code":""},{"path":"https://keras3.posit.co/dev/reference/optimizer_adam.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Optimizer that implements the Adam algorithm. — optimizer_adam","text":"","code":"optimizer_adam(   learning_rate = 0.001,   beta_1 = 0.9,   beta_2 = 0.999,   epsilon = 1e-07,   amsgrad = FALSE,   weight_decay = NULL,   clipnorm = NULL,   clipvalue = NULL,   global_clipnorm = NULL,   use_ema = FALSE,   ema_momentum = 0.99,   ema_overwrite_frequency = NULL,   name = \"adam\",   ...,   loss_scale_factor = NULL,   gradient_accumulation_steps = NULL )"},{"path":"https://keras3.posit.co/dev/reference/optimizer_adam.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Optimizer that implements the Adam algorithm. — optimizer_adam","text":"learning_rate float, LearningRateSchedule() instance, callable takes arguments returns actual value use. learning rate. Defaults 0.001. beta_1 float value constant float tensor, callable takes arguments returns actual value use. exponential decay rate 1st moment estimates. Defaults 0.9. beta_2 float value constant float tensor, callable takes arguments returns actual value use. exponential decay rate 2nd moment estimates. Defaults 0.999. epsilon small constant numerical stability. epsilon \"epsilon hat\" Kingma Ba paper (formula just Section 2.1), epsilon Algorithm 1 paper. Defaults 1e-7. amsgrad Boolean. Whether apply AMSGrad variant algorithm paper \"Convergence Adam beyond\". Defaults FALSE. weight_decay Float. set, weight decay applied. clipnorm Float. set, gradient weight individually clipped norm higher value. clipvalue Float. set, gradient weight clipped higher value. global_clipnorm Float. set, gradient weights clipped global norm higher value. use_ema Boolean, defaults FALSE. TRUE, exponential moving average (EMA) applied. EMA consists computing exponential moving average weights model (weight values change training batch), periodically overwriting weights moving average. ema_momentum Float, defaults 0.99. used use_ema=TRUE. momentum use computing EMA model's weights: new_average = ema_momentum * old_average + (1 - ema_momentum) * current_variable_value. ema_overwrite_frequency Int NULL, defaults NULL. used use_ema=TRUE. Every ema_overwrite_frequency steps iterations, overwrite model variable moving average. NULL, optimizer overwrite model variables middle training, need explicitly overwrite variables end training calling optimizer$finalize_variable_values() (updates model variables -place). using built-fit() training loop, happens automatically last epoch, need anything. name String. name use momentum accumulator weights created optimizer. ... forward/backward compatability. loss_scale_factor Float NULL. float, scale factor multiplied loss computing gradients, inverse scale factor multiplied gradients updating variables. Useful preventing underflow mixed precision training. Alternately, optimizer_loss_scale() automatically set loss scale factor. gradient_accumulation_steps Int NULL. int, model optimizer variables updated every step; instead updated every gradient_accumulation_steps steps, using average value gradients since last update. known \"gradient accumulation\". can useful batch size small, order reduce gradient noise update step. EMA frequency look \"accumulated\" iterations value (optimizer steps // gradient_accumulation_steps). Learning rate schedules look \"real\" iterations value (optimizer steps).","code":""},{"path":"https://keras3.posit.co/dev/reference/optimizer_adam.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Optimizer that implements the Adam algorithm. — optimizer_adam","text":"Optimizer instance","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/optimizer_adam_w.html","id":null,"dir":"Reference","previous_headings":"","what":"Optimizer that implements the AdamW algorithm. — optimizer_adam_w","title":"Optimizer that implements the AdamW algorithm. — optimizer_adam_w","text":"AdamW optimization stochastic gradient descent method based adaptive estimation first-order second-order moments added method decay weights per techniques discussed paper, 'Decoupled Weight Decay Regularization' Loshchilov, Hutter et al., 2019. According Kingma et al., 2014, underlying Adam method \"computationally efficient, little memory requirement, invariant diagonal rescaling gradients, well suited problems large terms data/parameters\".","code":""},{"path":"https://keras3.posit.co/dev/reference/optimizer_adam_w.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Optimizer that implements the AdamW algorithm. — optimizer_adam_w","text":"","code":"optimizer_adam_w(   learning_rate = 0.001,   weight_decay = 0.004,   beta_1 = 0.9,   beta_2 = 0.999,   epsilon = 1e-07,   amsgrad = FALSE,   clipnorm = NULL,   clipvalue = NULL,   global_clipnorm = NULL,   use_ema = FALSE,   ema_momentum = 0.99,   ema_overwrite_frequency = NULL,   name = \"adamw\",   ...,   loss_scale_factor = NULL,   gradient_accumulation_steps = NULL )"},{"path":"https://keras3.posit.co/dev/reference/optimizer_adam_w.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Optimizer that implements the AdamW algorithm. — optimizer_adam_w","text":"learning_rate float, LearningRateSchedule() instance, callable takes arguments returns actual value use. learning rate. Defaults 0.001. weight_decay Float. set, weight decay applied. beta_1 float value constant float tensor, callable takes arguments returns actual value use. exponential decay rate 1st moment estimates. Defaults 0.9. beta_2 float value constant float tensor, callable takes arguments returns actual value use. exponential decay rate 2nd moment estimates. Defaults 0.999. epsilon small constant numerical stability. epsilon \"epsilon hat\" Kingma Ba paper (formula just Section 2.1), epsilon Algorithm 1 paper. Defaults 1e-7. amsgrad Boolean. Whether apply AMSGrad variant algorithm paper \"Convergence Adam beyond\". Defaults FALSE. clipnorm Float. set, gradient weight individually clipped norm higher value. clipvalue Float. set, gradient weight clipped higher value. global_clipnorm Float. set, gradient weights clipped global norm higher value. use_ema Boolean, defaults FALSE. TRUE, exponential moving average (EMA) applied. EMA consists computing exponential moving average weights model (weight values change training batch), periodically overwriting weights moving average. ema_momentum Float, defaults 0.99. used use_ema=TRUE. momentum use computing EMA model's weights: new_average = ema_momentum * old_average + (1 - ema_momentum) * current_variable_value. ema_overwrite_frequency Int NULL, defaults NULL. used use_ema=TRUE. Every ema_overwrite_frequency steps iterations, overwrite model variable moving average. NULL, optimizer overwrite model variables middle training, need explicitly overwrite variables end training calling optimizer$finalize_variable_values() (updates model variables -place). using built-fit() training loop, happens automatically last epoch, need anything. name String. name use momentum accumulator weights created optimizer. ... forward/backward compatability. loss_scale_factor Float NULL. float, scale factor multiplied loss computing gradients, inverse scale factor multiplied gradients updating variables. Useful preventing underflow mixed precision training. Alternately, optimizer_loss_scale() automatically set loss scale factor. gradient_accumulation_steps Int NULL. int, model optimizer variables updated every step; instead updated every gradient_accumulation_steps steps, using average value gradients since last update. known \"gradient accumulation\". can useful batch size small, order reduce gradient noise update step. EMA frequency look \"accumulated\" iterations value (optimizer steps // gradient_accumulation_steps). Learning rate schedules look \"real\" iterations value (optimizer steps).","code":""},{"path":"https://keras3.posit.co/dev/reference/optimizer_adam_w.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Optimizer that implements the AdamW algorithm. — optimizer_adam_w","text":"Optimizer instance","code":""},{"path":"https://keras3.posit.co/dev/reference/optimizer_adam_w.html","id":"references","dir":"Reference","previous_headings":"","what":"References","title":"Optimizer that implements the AdamW algorithm. — optimizer_adam_w","text":"Loshchilov et al., 2019 Kingma et al., 2014 adam Reddi et al., 2018 amsgrad.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/optimizer_adamax.html","id":null,"dir":"Reference","previous_headings":"","what":"Optimizer that implements the Adamax algorithm. — optimizer_adamax","title":"Optimizer that implements the Adamax algorithm. — optimizer_adamax","text":"Adamax, variant Adam based infinity norm, first-order gradient-based optimization method. Due capability adjusting learning rate based data characteristics, suited learn time-variant process, e.g., speech data dynamically changed noise conditions. Default parameters follow provided paper (see references ). Initialization:   update rule parameter w gradient g described end section 7.1 paper (see reference section):","code":"m <- 0  # Initialize initial 1st moment vector u <- 0  # Initialize the exponentially weighted infinity norm t <- 0  # Initialize timestep t <-  t + 1 m <- beta1 * m + (1 - beta) * g u <- max(beta2 * u, abs(g)) current_lr <- learning_rate / (1 - beta1 ** t) w <- w - current_lr * m / (u + epsilon)"},{"path":"https://keras3.posit.co/dev/reference/optimizer_adamax.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Optimizer that implements the Adamax algorithm. — optimizer_adamax","text":"","code":"optimizer_adamax(   learning_rate = 0.001,   beta_1 = 0.9,   beta_2 = 0.999,   epsilon = 1e-07,   weight_decay = NULL,   clipnorm = NULL,   clipvalue = NULL,   global_clipnorm = NULL,   use_ema = FALSE,   ema_momentum = 0.99,   ema_overwrite_frequency = NULL,   name = \"adamax\",   ...,   loss_scale_factor = NULL,   gradient_accumulation_steps = NULL )"},{"path":"https://keras3.posit.co/dev/reference/optimizer_adamax.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Optimizer that implements the Adamax algorithm. — optimizer_adamax","text":"learning_rate float, LearningRateSchedule() instance, callable takes arguments returns actual value use. learning rate. Defaults 0.001. beta_1 float value constant float tensor. exponential decay rate 1st moment estimates. beta_2 float value constant float tensor. exponential decay rate exponentially weighted infinity norm. epsilon small constant numerical stability. name: String. name use momentum accumulator weights created optimizer. weight_decay Float. set, weight decay applied. clipnorm Float. set, gradient weight individually clipped norm higher value. clipvalue Float. set, gradient weight clipped higher value. global_clipnorm Float. set, gradient weights clipped global norm higher value. use_ema Boolean, defaults FALSE. TRUE, exponential moving average (EMA) applied. EMA consists computing exponential moving average weights model (weight values change training batch), periodically overwriting weights moving average. ema_momentum Float, defaults 0.99. used use_ema=TRUE. momentum use computing EMA model's weights: new_average = ema_momentum * old_average + (1 - ema_momentum) * current_variable_value. ema_overwrite_frequency Int NULL, defaults NULL. used use_ema=TRUE. Every ema_overwrite_frequency steps iterations, overwrite model variable moving average. NULL, optimizer overwrite model variables middle training, need explicitly overwrite variables end training calling optimizer$finalize_variable_values() (updates model variables -place). using built-fit() training loop, happens automatically last epoch, need anything. name String, name object ... forward/backward compatability. loss_scale_factor Float NULL. float, scale factor multiplied loss computing gradients, inverse scale factor multiplied gradients updating variables. Useful preventing underflow mixed precision training. Alternately, optimizer_loss_scale() automatically set loss scale factor. gradient_accumulation_steps Int NULL. int, model optimizer variables updated every step; instead updated every gradient_accumulation_steps steps, using average value gradients since last update. known \"gradient accumulation\". can useful batch size small, order reduce gradient noise update step. EMA frequency look \"accumulated\" iterations value (optimizer steps // gradient_accumulation_steps). Learning rate schedules look \"real\" iterations value (optimizer steps).","code":""},{"path":"https://keras3.posit.co/dev/reference/optimizer_adamax.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Optimizer that implements the Adamax algorithm. — optimizer_adamax","text":"Optimizer instance","code":""},{"path":"https://keras3.posit.co/dev/reference/optimizer_adamax.html","id":"reference","dir":"Reference","previous_headings":"","what":"Reference","title":"Optimizer that implements the Adamax algorithm. — optimizer_adamax","text":"Kingma et al., 2014","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/optimizer_ftrl.html","id":null,"dir":"Reference","previous_headings":"","what":"Optimizer that implements the FTRL algorithm. — optimizer_ftrl","title":"Optimizer that implements the FTRL algorithm. — optimizer_ftrl","text":"\"Follow Regularized Leader\" (FTRL) optimization algorithm developed Google click-rate prediction early 2010s. suitable shallow models large sparse feature spaces. algorithm described McMahan et al., 2013. Keras version support online L2 regularization (L2 regularization described paper ) shrinkage-type L2 regularization (addition L2 penalty loss function). Initialization:   Update rule one variable w:   Notation: lr learning rate g gradient variable lambda_1 L1 regularization strength lambda_2 L2 regularization strength lr_power power scale n. Check documentation l2_shrinkage_regularization_strength parameter details shrinkage enabled, case gradient replaced gradient shrinkage.","code":"n <- 0 sigma <- 0 z <- 0 prev_n <- n n <- n + g^2 sigma <- (n^(-lr_power) - prev_n^(-lr_power)) / lr z <- z + g - sigma * w if (abs(z) < lambda_1) {   w <- 0 } else {   w <- (sgn(z) * lambda_1 - z) / ((beta + sqrt(n)) / alpha + lambda_2) }"},{"path":"https://keras3.posit.co/dev/reference/optimizer_ftrl.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Optimizer that implements the FTRL algorithm. — optimizer_ftrl","text":"","code":"optimizer_ftrl(   learning_rate = 0.001,   learning_rate_power = -0.5,   initial_accumulator_value = 0.1,   l1_regularization_strength = 0,   l2_regularization_strength = 0,   l2_shrinkage_regularization_strength = 0,   beta = 0,   weight_decay = NULL,   clipnorm = NULL,   clipvalue = NULL,   global_clipnorm = NULL,   use_ema = FALSE,   ema_momentum = 0.99,   ema_overwrite_frequency = NULL,   name = \"ftrl\",   ...,   loss_scale_factor = NULL,   gradient_accumulation_steps = NULL )"},{"path":"https://keras3.posit.co/dev/reference/optimizer_ftrl.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Optimizer that implements the FTRL algorithm. — optimizer_ftrl","text":"learning_rate float, LearningRateSchedule() instance, callable takes arguments returns actual value use. learning rate. Defaults 0.001. learning_rate_power float value, must less equal zero. Controls learning rate decreases training. Use zero fixed learning rate. initial_accumulator_value starting value accumulators. zero positive values allowed. l1_regularization_strength float value, must greater equal zero. Defaults 0.0. l2_regularization_strength float value, must greater equal zero. Defaults 0.0. l2_shrinkage_regularization_strength float value, must greater equal zero. differs L2 L2 stabilization penalty, whereas L2 shrinkage magnitude penalty. input sparse shrinkage happen active weights. beta float value, representing beta value paper. Defaults 0.0. weight_decay Float. set, weight decay applied. clipnorm Float. set, gradient weight individually clipped norm higher value. clipvalue Float. set, gradient weight clipped higher value. global_clipnorm Float. set, gradient weights clipped global norm higher value. use_ema Boolean, defaults FALSE. TRUE, exponential moving average (EMA) applied. EMA consists computing exponential moving average weights model (weight values change training batch), periodically overwriting weights moving average. ema_momentum Float, defaults 0.99. used use_ema=TRUE. momentum use computing EMA model's weights: new_average = ema_momentum * old_average + (1 - ema_momentum) * current_variable_value. ema_overwrite_frequency Int NULL, defaults NULL. used use_ema=TRUE. Every ema_overwrite_frequency steps iterations, overwrite model variable moving average. NULL, optimizer overwrite model variables middle training, need explicitly overwrite variables end training calling optimizer$finalize_variable_values() (updates model variables -place). using built-fit() training loop, happens automatically last epoch, need anything. name String. name use momentum accumulator weights created optimizer. ... forward/backward compatability. loss_scale_factor Float NULL. float, scale factor multiplied loss computing gradients, inverse scale factor multiplied gradients updating variables. Useful preventing underflow mixed precision training. Alternately, optimizer_loss_scale automatically set loss scale factor. gradient_accumulation_steps Int NULL. int, model optimizer variables updated every step; instead updated every gradient_accumulation_steps steps, using average value gradients since last update. known \"gradient accumulation\". can useful batch size small, order reduce gradient noise update step. EMA frequency look \"accumulated\" iterations value (optimizer steps // gradient_accumulation_steps). Learning rate schedules look \"real\" iterations value (optimizer steps).","code":""},{"path":"https://keras3.posit.co/dev/reference/optimizer_ftrl.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Optimizer that implements the FTRL algorithm. — optimizer_ftrl","text":"Optimizer instance","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/optimizer_lamb.html","id":null,"dir":"Reference","previous_headings":"","what":"Optimizer that implements the Lamb algorithm. — optimizer_lamb","title":"Optimizer that implements the Lamb algorithm. — optimizer_lamb","text":"Lamb stochastic gradient descent method uses layer-wise adaptive moments adjusts learning rate parameter based ratio norm weight norm gradient helps stabilize training process improves convergence especially large batch sizes.","code":""},{"path":"https://keras3.posit.co/dev/reference/optimizer_lamb.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Optimizer that implements the Lamb algorithm. — optimizer_lamb","text":"","code":"optimizer_lamb(   learning_rate = 0.001,   beta_1 = 0.9,   beta_2 = 0.999,   epsilon = 1e-07,   weight_decay = NULL,   clipnorm = NULL,   clipvalue = NULL,   global_clipnorm = NULL,   use_ema = FALSE,   ema_momentum = 0.99,   ema_overwrite_frequency = NULL,   loss_scale_factor = NULL,   gradient_accumulation_steps = NULL,   name = \"lamb\",   ... )"},{"path":"https://keras3.posit.co/dev/reference/optimizer_lamb.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Optimizer that implements the Lamb algorithm. — optimizer_lamb","text":"learning_rate float, LearningRateSchedule() instance, callable takes arguments returns actual value use. learning rate. Defaults 0.001. beta_1 float value constant float tensor, callable takes arguments returns actual value use. exponential decay rate 1st moment estimates. Defaults 0.9. beta_2 float value constant float tensor, callable takes arguments returns actual value use. exponential decay rate 2nd moment estimates. Defaults 0.999. epsilon small constant numerical stability. Defaults 1e-7. weight_decay Float. set, weight decay applied. clipnorm Float. set, gradient weight individually clipped norm higher value. clipvalue Float. set, gradient weight clipped higher value. global_clipnorm Float. set, gradient weights clipped global norm higher value. use_ema Boolean, defaults FALSE. TRUE, exponential moving average (EMA) applied. EMA consists computing exponential moving average weights model (weight values change training batch), periodically overwriting weights moving average. ema_momentum Float, defaults 0.99. used use_ema = TRUE. momentum use computing EMA model's weights: new_average = ema_momentum * old_average + (1 - ema_momentum) * current_variable_value. ema_overwrite_frequency Int NULL, defaults NULL. used use_ema = TRUE. Every ema_overwrite_frequency steps iterations, overwrite model variable moving average. NULL, optimizer overwrite model variables middle training, need explicitly overwrite variables end training calling optimizer$finalize_variable_values() (updates model variables -place). using built-fit() training loop, happens automatically last epoch, need anything. loss_scale_factor Float NULL. float, scale factor multiplied loss computing gradients, inverse scale factor multiplied gradients updating variables. Useful preventing underflow mixed precision training. Alternately, optimizer_loss_scale() automatically set loss scale factor. gradient_accumulation_steps Int NULL. int, model optimizer variables updated every step; instead updated every gradient_accumulation_steps steps, using average value gradients since last update. known \"gradient accumulation\". can useful batch size small, order reduce gradient noise update step. EMA frequency look \"accumulated\" iterations value (optimizer steps // gradient_accumulation_steps). Learning rate schedules look \"real\" iterations value (optimizer steps). name String. name use momentum accumulator weights created optimizer. ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/optimizer_lamb.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Optimizer that implements the Lamb algorithm. — optimizer_lamb","text":"Optimizer instance","code":""},{"path":"https://keras3.posit.co/dev/reference/optimizer_lamb.html","id":"references","dir":"Reference","previous_headings":"","what":"References","title":"Optimizer that implements the Lamb algorithm. — optimizer_lamb","text":"Yang et al.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/optimizer_lion.html","id":null,"dir":"Reference","previous_headings":"","what":"Optimizer that implements the Lion algorithm. — optimizer_lion","title":"Optimizer that implements the Lion algorithm. — optimizer_lion","text":"Lion optimizer stochastic-gradient-descent method uses sign operator control magnitude update, unlike adaptive optimizers Adam rely second-order moments. makes Lion memory-efficient keeps track momentum. According authors (see reference), performance gain Adam grows batch size. update Lion produced sign operation, resulting larger norm, suitable learning rate Lion typically 3-10x smaller AdamW. weight decay Lion turn 3-10x larger AdamW maintain similar strength (lr * wd).","code":""},{"path":"https://keras3.posit.co/dev/reference/optimizer_lion.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Optimizer that implements the Lion algorithm. — optimizer_lion","text":"","code":"optimizer_lion(   learning_rate = 0.001,   beta_1 = 0.9,   beta_2 = 0.99,   weight_decay = NULL,   clipnorm = NULL,   clipvalue = NULL,   global_clipnorm = NULL,   use_ema = FALSE,   ema_momentum = 0.99,   ema_overwrite_frequency = NULL,   name = \"lion\",   ...,   loss_scale_factor = NULL,   gradient_accumulation_steps = NULL )"},{"path":"https://keras3.posit.co/dev/reference/optimizer_lion.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Optimizer that implements the Lion algorithm. — optimizer_lion","text":"learning_rate float, LearningRateSchedule() instance, callable takes arguments returns actual value use. learning rate. Defaults 0.001. beta_1 float value constant float tensor, callable takes arguments returns actual value use. rate combine current gradient 1st moment estimate. Defaults 0.9. beta_2 float value constant float tensor, callable takes arguments returns actual value use. exponential decay rate 1st moment estimate. Defaults 0.99. weight_decay Float. set, weight decay applied. clipnorm Float. set, gradient weight individually clipped norm higher value. clipvalue Float. set, gradient weight clipped higher value. global_clipnorm Float. set, gradient weights clipped global norm higher value. use_ema Boolean, defaults FALSE. TRUE, exponential moving average (EMA) applied. EMA consists computing exponential moving average weights model (weight values change training batch), periodically overwriting weights moving average. ema_momentum Float, defaults 0.99. used use_ema=TRUE. momentum use computing EMA model's weights: new_average = ema_momentum * old_average + (1 - ema_momentum) * current_variable_value. ema_overwrite_frequency Int NULL, defaults NULL. used use_ema=TRUE. Every ema_overwrite_frequency steps iterations, overwrite model variable moving average. NULL, optimizer overwrite model variables middle training, need explicitly overwrite variables end training calling optimizer$finalize_variable_values() (updates model variables -place). using built-fit() training loop, happens automatically last epoch, need anything. name String. name use momentum accumulator weights created optimizer. ... forward/backward compatability. loss_scale_factor Float NULL. float, scale factor multiplied loss computing gradients, inverse scale factor multiplied gradients updating variables. Useful preventing underflow mixed precision training. Alternately, optimizer_loss_scale() automatically set loss scale factor. gradient_accumulation_steps Int NULL. int, model optimizer variables updated every step; instead updated every gradient_accumulation_steps steps, using average value gradients since last update. known \"gradient accumulation\". can useful batch size small, order reduce gradient noise update step. EMA frequency look \"accumulated\" iterations value (optimizer steps // gradient_accumulation_steps). Learning rate schedules look \"real\" iterations value (optimizer steps).","code":""},{"path":"https://keras3.posit.co/dev/reference/optimizer_lion.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Optimizer that implements the Lion algorithm. — optimizer_lion","text":"Optimizer instance","code":""},{"path":"https://keras3.posit.co/dev/reference/optimizer_lion.html","id":"references","dir":"Reference","previous_headings":"","what":"References","title":"Optimizer that implements the Lion algorithm. — optimizer_lion","text":"Chen et al., 2023 Authors' implementation","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/optimizer_loss_scale.html","id":null,"dir":"Reference","previous_headings":"","what":"An optimizer that dynamically scales the loss to prevent underflow. — optimizer_loss_scale","title":"An optimizer that dynamically scales the loss to prevent underflow. — optimizer_loss_scale","text":"Loss scaling technique prevent numeric underflow intermediate gradients float16 used. prevent underflow, loss multiplied (\"scaled\") certain factor called \"loss scale\", causes intermediate gradients scaled loss scale well. final gradients divided (\"unscaled\") loss scale bring back original value. LossScaleOptimizer wraps another optimizer applies dynamic loss scaling . loss scale dynamically updated time follows: train step, nonfinite gradient encountered, loss scale halved, train step skipped. dynamic_growth_steps occurred since last time loss scale updated, nonfinite gradients occurred, loss scale doubled.","code":""},{"path":"https://keras3.posit.co/dev/reference/optimizer_loss_scale.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"An optimizer that dynamically scales the loss to prevent underflow. — optimizer_loss_scale","text":"","code":"optimizer_loss_scale(   inner_optimizer,   initial_scale = 32768,   dynamic_growth_steps = 2000L,   ...,   name = NULL,   weight_decay = NULL,   clipnorm = NULL,   clipvalue = NULL,   global_clipnorm = NULL,   use_ema = NULL,   ema_momentum = NULL,   ema_overwrite_frequency = NULL,   loss_scale_factor = NULL,   gradient_accumulation_steps = NULL )"},{"path":"https://keras3.posit.co/dev/reference/optimizer_loss_scale.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"An optimizer that dynamically scales the loss to prevent underflow. — optimizer_loss_scale","text":"inner_optimizer keras Optimizer instance wrap. initial_scale Float. initial loss scale. scale updated training. recommended high number, loss scale high gets lowered far quickly loss scale low gets raised. dynamic_growth_steps Int. often update scale upwards. every dynamic_growth_steps steps finite gradients, loss scale doubled. ... forward/backward compatability. name String. name use momentum accumulator weights created optimizer. weight_decay Float. set, weight decay applied. clipnorm Float. set, gradient weight individually clipped norm higher value. clipvalue Float. set, gradient weight clipped higher value. global_clipnorm Float. set, gradient weights clipped global norm higher value. use_ema Boolean, defaults FALSE. TRUE, exponential moving average (EMA) applied. EMA consists computing exponential moving average weights model (weight values change training batch), periodically overwriting weights moving average. ema_momentum Float, defaults 0.99. used use_ema=TRUE. momentum use computing EMA model's weights: new_average = ema_momentum * old_average + (1 - ema_momentum) * current_variable_value. ema_overwrite_frequency Int NULL, defaults NULL. used use_ema=TRUE. Every ema_overwrite_frequency steps iterations, overwrite model variable moving average. NULL, optimizer overwrite model variables middle training, need explicitly overwrite variables end training calling optimizer$finalize_variable_values() (updates model variables -place). using built-fit() training loop, happens automatically last epoch, need anything. loss_scale_factor Float NULL. float, scale factor multiplied loss computing gradients, inverse scale factor multiplied gradients updating variables. Useful preventing underflow mixed precision training. Alternately, optimizer_loss_scale() automatically set loss scale factor. gradient_accumulation_steps Int NULL. int, model optimizer variables updated every step; instead updated every gradient_accumulation_steps steps, using average value gradients since last update. known \"gradient accumulation\". can useful batch size small, order reduce gradient noise update step. EMA frequency look \"accumulated\" iterations value (optimizer steps // gradient_accumulation_steps). Learning rate schedules look \"real\" iterations value (optimizer steps).","code":""},{"path":"https://keras3.posit.co/dev/reference/optimizer_loss_scale.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"An optimizer that dynamically scales the loss to prevent underflow. — optimizer_loss_scale","text":"Optimizer instance","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/optimizer_muon.html","id":null,"dir":"Reference","previous_headings":"","what":"Optimizer that implements the Muon algorithm. — optimizer_muon","title":"Optimizer that implements the Muon algorithm. — optimizer_muon","text":"Note optimizer used following layers: Embedding layer Final output fully connected layer 0,1-D variables optimized using AdamW. Muon optimizer can use Muon update step AdamW update step based following: variable 2D, 3D 4D, AdamW step used. configurable. argument exclude_embeddings (defaults TRUE) set TRUE, AdamW step used. variable name matches expression listed argument exclude_layers (list), AdamW step used. variable uses Muon step. Typically, need pass name densely-connected output layer exclude_layers, e.g. exclude_layers = \"output_dense\".","code":""},{"path":"https://keras3.posit.co/dev/reference/optimizer_muon.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Optimizer that implements the Muon algorithm. — optimizer_muon","text":"","code":"optimizer_muon(   learning_rate = 0.001,   adam_beta_1 = 0.9,   adam_beta_2 = 0.999,   epsilon = 1e-07,   weight_decay = 0.1,   clipnorm = NULL,   clipvalue = NULL,   global_clipnorm = NULL,   use_ema = FALSE,   ema_momentum = 0.99,   ema_overwrite_frequency = NULL,   loss_scale_factor = NULL,   gradient_accumulation_steps = NULL,   name = \"muon\",   exclude_layers = NULL,   exclude_embeddings = TRUE,   muon_a = 3.4445,   muon_b = -4.775,   muon_c = 2.0315,   adam_lr_ratio = 0.1,   momentum = 0.95,   ns_steps = 6L,   nesterov = TRUE,   ... )"},{"path":"https://keras3.posit.co/dev/reference/optimizer_muon.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Optimizer that implements the Muon algorithm. — optimizer_muon","text":"learning_rate float, LearningRateSchedule() instance, callable takes arguments returns actual value use. learning rate. Defaults 0.001. adam_beta_1 float value constant float tensor, callable takes arguments returns actual value use. exponential decay rate 1st moment estimates. Defaults 0.9. adam_beta_2 float value constant float tensor, ora callable takes arguments returns actual value use. exponential decay rate 2nd moment estimates. Defaults 0.999. epsilon small constant numerical stability. \"epsilon hat\" Kingma Ba paper (formula just Section 2.1), epsilon Algorithm 1 paper. used AdamW. Defaults 1e-7. name String, name object exclude_layers List strings, keywords layer names exclude. layers keywords path use AdamW. exclude_embeddings Boolean value. TRUE, embedding layers use AdamW. muon_a Float, parameter muon algorithm. recommended use default value. muon_b Float, parameter b muon algorithm. recommended use default value. muon_c Float, parameter c muon algorithm. recommended use default value. adam_lr_ratio Float, ratio learning rate using Adam main learning rate. recommended set 0.1. momentum Float, momentum used internal SGD. ns_steps Integer, number Newton-Schulz iterations run. nesterov Boolean, whether use Nesterov-style momentum. ... forward/backward compatibility.","code":""},{"path":"https://keras3.posit.co/dev/reference/optimizer_muon.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Optimizer that implements the Muon algorithm. — optimizer_muon","text":"Optimizer instance","code":""},{"path":"https://keras3.posit.co/dev/reference/optimizer_muon.html","id":"references","dir":"Reference","previous_headings":"","what":"References","title":"Optimizer that implements the Muon algorithm. — optimizer_muon","text":"Original implementation Liu et al, 2025","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/optimizer_nadam.html","id":null,"dir":"Reference","previous_headings":"","what":"Optimizer that implements the Nadam algorithm. — optimizer_nadam","title":"Optimizer that implements the Nadam algorithm. — optimizer_nadam","text":"Much like Adam essentially RMSprop momentum, Nadam Adam Nesterov momentum.","code":""},{"path":"https://keras3.posit.co/dev/reference/optimizer_nadam.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Optimizer that implements the Nadam algorithm. — optimizer_nadam","text":"","code":"optimizer_nadam(   learning_rate = 0.001,   beta_1 = 0.9,   beta_2 = 0.999,   epsilon = 1e-07,   weight_decay = NULL,   clipnorm = NULL,   clipvalue = NULL,   global_clipnorm = NULL,   use_ema = FALSE,   ema_momentum = 0.99,   ema_overwrite_frequency = NULL,   name = \"nadam\",   ...,   loss_scale_factor = NULL,   gradient_accumulation_steps = NULL )"},{"path":"https://keras3.posit.co/dev/reference/optimizer_nadam.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Optimizer that implements the Nadam algorithm. — optimizer_nadam","text":"learning_rate float, LearningRateSchedule() instance, callable takes arguments returns actual value use. learning rate. Defaults 0.001. beta_1 float value constant float tensor, callable takes arguments returns actual value use. exponential decay rate 1st moment estimates. Defaults 0.9. beta_2 float value constant float tensor, callable takes arguments returns actual value use. exponential decay rate 2nd moment estimates. Defaults 0.999. epsilon small constant numerical stability. epsilon \"epsilon hat\" Kingma Ba paper (formula just Section 2.1), epsilon Algorithm 1 paper. Defaults 1e-7. weight_decay Float. set, weight decay applied. clipnorm Float. set, gradient weight individually clipped norm higher value. clipvalue Float. set, gradient weight clipped higher value. global_clipnorm Float. set, gradient weights clipped global norm higher value. use_ema Boolean, defaults FALSE. TRUE, exponential moving average (EMA) applied. EMA consists computing exponential moving average weights model (weight values change training batch), periodically overwriting weights moving average. ema_momentum Float, defaults 0.99. used use_ema=TRUE. momentum use computing EMA model's weights: new_average = ema_momentum * old_average + (1 - ema_momentum) * current_variable_value. ema_overwrite_frequency Int NULL, defaults NULL. used use_ema=TRUE. Every ema_overwrite_frequency steps iterations, overwrite model variable moving average. NULL, optimizer overwrite model variables middle training, need explicitly overwrite variables end training calling optimizer$finalize_variable_values() (updates model variables -place). using built-fit() training loop, happens automatically last epoch, need anything. name String. name use momentum accumulator weights created optimizer. ... forward/backward compatability. loss_scale_factor Float NULL. float, scale factor multiplied loss computing gradients, inverse scale factor multiplied gradients updating variables. Useful preventing underflow mixed precision training. Alternately, optimizer_loss_scale() automatically set loss scale factor. gradient_accumulation_steps Int NULL. int, model optimizer variables updated every step; instead updated every gradient_accumulation_steps steps, using average value gradients since last update. known \"gradient accumulation\". can useful batch size small, order reduce gradient noise update step. EMA frequency look \"accumulated\" iterations value (optimizer steps // gradient_accumulation_steps). Learning rate schedules look \"real\" iterations value (optimizer steps).","code":""},{"path":"https://keras3.posit.co/dev/reference/optimizer_nadam.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Optimizer that implements the Nadam algorithm. — optimizer_nadam","text":"Optimizer instance","code":""},{"path":"https://keras3.posit.co/dev/reference/optimizer_nadam.html","id":"reference","dir":"Reference","previous_headings":"","what":"Reference","title":"Optimizer that implements the Nadam algorithm. — optimizer_nadam","text":"Dozat, 2015.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/optimizer_rmsprop.html","id":null,"dir":"Reference","previous_headings":"","what":"Optimizer that implements the RMSprop algorithm. — optimizer_rmsprop","title":"Optimizer that implements the RMSprop algorithm. — optimizer_rmsprop","text":"gist RMSprop : Maintain moving (discounted) average square gradients Divide gradient root average implementation RMSprop uses plain momentum, Nesterov momentum. centered version additionally maintains moving average gradients, uses average estimate variance.","code":""},{"path":"https://keras3.posit.co/dev/reference/optimizer_rmsprop.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Optimizer that implements the RMSprop algorithm. — optimizer_rmsprop","text":"","code":"optimizer_rmsprop(   learning_rate = 0.001,   rho = 0.9,   momentum = 0,   epsilon = 1e-07,   centered = FALSE,   weight_decay = NULL,   clipnorm = NULL,   clipvalue = NULL,   global_clipnorm = NULL,   use_ema = FALSE,   ema_momentum = 0.99,   ema_overwrite_frequency = NULL,   name = \"rmsprop\",   ...,   loss_scale_factor = NULL,   gradient_accumulation_steps = NULL )"},{"path":"https://keras3.posit.co/dev/reference/optimizer_rmsprop.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Optimizer that implements the RMSprop algorithm. — optimizer_rmsprop","text":"learning_rate float, learning_rate_schedule_* instance, callable takes arguments returns actual value use. learning rate. Defaults 0.001. rho float, defaults 0.9. Discounting factor old gradients. momentum float, defaults 0.0. 0.0., optimizer tracks momentum value, decay rate equals 1 - momentum. epsilon small constant numerical stability. epsilon \"epsilon hat\" Kingma Ba paper (formula just Section 2.1), epsilon Algorithm 1 paper. Defaults 1e-7. centered Boolean. TRUE, gradients normalized estimated variance gradient; FALSE, uncentered second moment. Setting TRUE may help training, slightly expensive terms computation memory. Defaults FALSE. weight_decay Float. set, weight decay applied. clipnorm Float. set, gradient weight individually clipped norm higher value. clipvalue Float. set, gradient weight clipped higher value. global_clipnorm Float. set, gradient weights clipped global norm higher value. use_ema Boolean, defaults FALSE. TRUE, exponential moving average (EMA) applied. EMA consists computing exponential moving average weights model (weight values change training batch), periodically overwriting weights moving average. ema_momentum Float, defaults 0.99. used use_ema=TRUE. momentum use computing EMA model's weights: new_average = ema_momentum * old_average + (1 - ema_momentum) * current_variable_value. ema_overwrite_frequency Int NULL, defaults NULL. used use_ema=TRUE. Every ema_overwrite_frequency steps iterations, overwrite model variable moving average. NULL, optimizer overwrite model variables middle training, need explicitly overwrite variables end training calling optimizer$finalize_variable_values() (updates model variables -place). using built-fit() training loop, happens automatically last epoch, need anything. name String. name use momentum accumulator weights created optimizer. ... forward/backward compatability. loss_scale_factor Float NULL. float, scale factor multiplied loss computing gradients, inverse scale factor multiplied gradients updating variables. Useful preventing underflow mixed precision training. Alternately, optimizer_loss_scale() automatically set loss scale factor. gradient_accumulation_steps Int NULL. int, model optimizer variables updated every step; instead updated every gradient_accumulation_steps steps, using average value gradients since last update. known \"gradient accumulation\". can useful batch size small, order reduce gradient noise update step. EMA frequency look \"accumulated\" iterations value (optimizer steps // gradient_accumulation_steps). Learning rate schedules look \"real\" iterations value (optimizer steps).","code":""},{"path":"https://keras3.posit.co/dev/reference/optimizer_rmsprop.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Optimizer that implements the RMSprop algorithm. — optimizer_rmsprop","text":"Optimizer instance","code":""},{"path":"https://keras3.posit.co/dev/reference/optimizer_rmsprop.html","id":"usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Optimizer that implements the RMSprop algorithm. — optimizer_rmsprop","text":"","code":"opt <- optimizer_rmsprop(learning_rate=0.1)"},{"path":"https://keras3.posit.co/dev/reference/optimizer_rmsprop.html","id":"reference","dir":"Reference","previous_headings":"","what":"Reference","title":"Optimizer that implements the RMSprop algorithm. — optimizer_rmsprop","text":"Hinton, 2012","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/optimizer_sgd.html","id":null,"dir":"Reference","previous_headings":"","what":"Gradient descent (with momentum) optimizer. — optimizer_sgd","title":"Gradient descent (with momentum) optimizer. — optimizer_sgd","text":"Update rule parameter w gradient g momentum 0:   Update rule momentum larger 0:   nesterov=TRUE, rule becomes:","code":"w <- w - learning_rate * g velocity <- momentum * velocity - learning_rate * g w <- w + velocity velocity <- momentum * velocity - learning_rate * g w <- w + momentum * velocity - learning_rate * g"},{"path":"https://keras3.posit.co/dev/reference/optimizer_sgd.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Gradient descent (with momentum) optimizer. — optimizer_sgd","text":"","code":"optimizer_sgd(   learning_rate = 0.01,   momentum = 0,   nesterov = FALSE,   weight_decay = NULL,   clipnorm = NULL,   clipvalue = NULL,   global_clipnorm = NULL,   use_ema = FALSE,   ema_momentum = 0.99,   ema_overwrite_frequency = NULL,   name = \"SGD\",   ...,   loss_scale_factor = NULL,   gradient_accumulation_steps = NULL )"},{"path":"https://keras3.posit.co/dev/reference/optimizer_sgd.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Gradient descent (with momentum) optimizer. — optimizer_sgd","text":"learning_rate float, learning_rate_schedule_* instance, callable takes arguments returns actual value use. learning rate. Defaults 0.01. momentum float hyperparameter >= 0 accelerates gradient descent relevant direction dampens oscillations. 0 vanilla gradient descent. Defaults 0.0. nesterov boolean. Whether apply Nesterov momentum. Defaults FALSE. weight_decay Float. set, weight decay applied. clipnorm Float. set, gradient weight individually clipped norm higher value. clipvalue Float. set, gradient weight clipped higher value. global_clipnorm Float. set, gradient weights clipped global norm higher value. use_ema Boolean, defaults FALSE. TRUE, exponential moving average (EMA) applied. EMA consists computing exponential moving average weights model (weight values change training batch), periodically overwriting weights moving average. ema_momentum Float, defaults 0.99. used use_ema=TRUE. momentum use computing EMA model's weights: new_average = ema_momentum * old_average + (1 - ema_momentum) * current_variable_value. ema_overwrite_frequency Int NULL, defaults NULL. used use_ema=TRUE. Every ema_overwrite_frequency steps iterations, overwrite model variable moving average. NULL, optimizer overwrite model variables middle training, need explicitly overwrite variables end training calling optimizer$finalize_variable_values() (updates model variables -place). using built-fit() training loop, happens automatically last epoch, need anything. name String. name use momentum accumulator weights created optimizer. ... forward/backward compatability. loss_scale_factor Float NULL. float, scale factor multiplied loss computing gradients, inverse scale factor multiplied gradients updating variables. Useful preventing underflow mixed precision training. Alternately, optimizer_loss_scale() automatically set loss scale factor. gradient_accumulation_steps Int NULL. int, model optimizer variables updated every step; instead updated every gradient_accumulation_steps steps, using average value gradients since last update. known \"gradient accumulation\". can useful batch size small, order reduce gradient noise update step. EMA frequency look \"accumulated\" iterations value (optimizer steps // gradient_accumulation_steps). Learning rate schedules look \"real\" iterations value (optimizer steps).","code":""},{"path":"https://keras3.posit.co/dev/reference/optimizer_sgd.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Gradient descent (with momentum) optimizer. — optimizer_sgd","text":"Optimizer instance","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/pad_sequences.html","id":null,"dir":"Reference","previous_headings":"","what":"Pads sequences to the same length. — pad_sequences","title":"Pads sequences to the same length. — pad_sequences","text":"function transforms list (length num_samples) sequences (lists integers) 2D NumPy array shape (num_samples, num_timesteps). num_timesteps either maxlen argument provided, length longest sequence list. Sequences shorter num_timesteps padded value num_timesteps long. Sequences longer num_timesteps truncated fit desired length. position padding truncation happens determined arguments padding truncating, respectively. Pre-padding removing values beginning sequence default.","code":"sequence <- list(c(1), c(2, 3), c(4, 5, 6)) pad_sequences(sequence) ##      [,1] [,2] [,3] ## [1,]    0    0    1 ## [2,]    0    2    3 ## [3,]    4    5    6 pad_sequences(sequence, value=-1) ##      [,1] [,2] [,3] ## [1,]   -1   -1    1 ## [2,]   -1    2    3 ## [3,]    4    5    6 pad_sequences(sequence, padding='post') ##      [,1] [,2] [,3] ## [1,]    1    0    0 ## [2,]    2    3    0 ## [3,]    4    5    6 pad_sequences(sequence, maxlen=2) ##      [,1] [,2] ## [1,]    0    1 ## [2,]    2    3 ## [3,]    5    6"},{"path":"https://keras3.posit.co/dev/reference/pad_sequences.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Pads sequences to the same length. — pad_sequences","text":"","code":"pad_sequences(   sequences,   maxlen = NULL,   dtype = \"int32\",   padding = \"pre\",   truncating = \"pre\",   value = 0 )"},{"path":"https://keras3.posit.co/dev/reference/pad_sequences.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Pads sequences to the same length. — pad_sequences","text":"sequences List sequences (sequence list integers). maxlen Optional Int, maximum length sequences. provided, sequences padded length longest individual sequence. dtype (Optional, defaults \"int32\"). Type output sequences. pad sequences variable length strings, can use object. padding String, \"pre\" \"post\" (optional, defaults \"pre\"): pad either sequence. truncating String, \"pre\" \"post\" (optional, defaults \"pre\"): remove values sequences larger maxlen, either beginning end sequences. value Float String, padding value. (Optional, defaults 0)","code":""},{"path":"https://keras3.posit.co/dev/reference/pad_sequences.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Pads sequences to the same length. — pad_sequences","text":"Array shape (len(sequences), maxlen)","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/pipe.html","id":null,"dir":"Reference","previous_headings":"","what":"Pipe operator — %>%","title":"Pipe operator — %>%","text":"See %>% details.","code":""},{"path":"https://keras3.posit.co/dev/reference/pipe.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Pipe operator — %>%","text":"","code":"lhs %>% rhs"},{"path":"https://keras3.posit.co/dev/reference/pipe.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Pipe operator — %>%","text":"commonly, result calling right hand side left hand side argument: rhs(lhs). See magritter vignette , advanced, usages.","code":""},{"path":"https://keras3.posit.co/dev/reference/plot.keras.src.models.model.Model.html","id":null,"dir":"Reference","previous_headings":"","what":"Plot a Keras model — plot.keras.src.models.model.Model","title":"Plot a Keras model — plot.keras.src.models.model.Model","text":"Plot Keras model","code":""},{"path":"https://keras3.posit.co/dev/reference/plot.keras.src.models.model.Model.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Plot a Keras model — plot.keras.src.models.model.Model","text":"","code":"# S3 method for class 'keras.src.models.model.Model' plot(   x,   show_shapes = FALSE,   show_dtype = FALSE,   show_layer_names = FALSE,   ...,   rankdir = \"TB\",   expand_nested = FALSE,   dpi = 200,   layer_range = NULL,   show_layer_activations = FALSE,   show_trainable = NA,   to_file = NULL )"},{"path":"https://keras3.posit.co/dev/reference/plot.keras.src.models.model.Model.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Plot a Keras model — plot.keras.src.models.model.Model","text":"x Keras model instance show_shapes whether display shape information. show_dtype whether display layer dtypes. show_layer_names whether display layer names. ... passed Python keras.utils.model_to_dot(). Used forward backward compatibility. rankdir string specifying format plot: 'TB' creates vertical plot; 'LR' creates horizontal plot. (argument passed PyDot) expand_nested Whether expand nested models clusters. dpi Dots per inch. Increase value image text appears excessively pixelated. layer_range list containing two character strings, starting layer name ending layer name (inclusive) indicating range layers plot generated. also accepts regex patterns instead exact name. case, start predicate first element matches layer_range[1] end predicate last element matches layer_range[2]. default NULL considers layers model. Note must pass range resultant subgraph must complete. show_layer_activations Display layer activations (layers activation property). show_trainable whether display layer trainable. to_file File name plot image. NULL (default), model drawn default graphics device. Otherwise, file saved.","code":""},{"path":"https://keras3.posit.co/dev/reference/plot.keras.src.models.model.Model.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Plot a Keras model — plot.keras.src.models.model.Model","text":"Nothing, called side effects.","code":""},{"path":"https://keras3.posit.co/dev/reference/plot.keras.src.models.model.Model.html","id":"raises","dir":"Reference","previous_headings":"","what":"Raises","title":"Plot a Keras model — plot.keras.src.models.model.Model","text":"ValueError: plot(model) called model built, unless input_shape =  argument supplied keras_model_sequential().","code":""},{"path":"https://keras3.posit.co/dev/reference/plot.keras.src.models.model.Model.html","id":"requirements","dir":"Reference","previous_headings":"","what":"Requirements","title":"Plot a Keras model — plot.keras.src.models.model.Model","text":"function requires pydot graphviz. pydot default installed install_keras(), installed Keras means, can install pydot directly :   can install graphviz directly : https://graphviz.gitlab.io/download/ Linux platforms, can install graphviz via package manager. example, Ubuntu/Debian can install   macOS can install graphviz using brew:   conda environment, can install graphviz :","code":"reticulate::py_install(\"pydot\", pip = TRUE) sudo apt install graphviz brew install graphviz reticulate::conda_install(packages = \"graphviz\") # Restart the R session after install."},{"path":"https://keras3.posit.co/dev/reference/plot.keras_training_history.html","id":null,"dir":"Reference","previous_headings":"","what":"Plot training history — plot.keras_training_history","title":"Plot training history — plot.keras_training_history","text":"Plots metrics recorded training.","code":""},{"path":"https://keras3.posit.co/dev/reference/plot.keras_training_history.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Plot training history — plot.keras_training_history","text":"","code":"# S3 method for class 'keras_training_history' plot(   x,   y,   metrics = NULL,   method = c(\"auto\", \"ggplot2\", \"base\"),   smooth = getOption(\"keras.plot.history.smooth\", TRUE),   theme_bw = getOption(\"keras.plot.history.theme_bw\", FALSE),   ... )"},{"path":"https://keras3.posit.co/dev/reference/plot.keras_training_history.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Plot training history — plot.keras_training_history","text":"x Training history object returned fit.keras.src.models.model.Model(). y Unused. metrics One metrics plot (e.g. c('loss', 'accuracy')). Defaults plotting captured metrics. method Method use plotting. default \"auto\" use ggplot2 available, otherwise use base graphics. smooth Whether loess smooth added plot, available ggplot2 method. number epochs smaller ten, forced false. theme_bw Use ggplot2::theme_bw() plot history black white. ... Additional parameters pass plot() method.","code":""},{"path":"https://keras3.posit.co/dev/reference/plot.keras_training_history.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Plot training history — plot.keras_training_history","text":"method == \"ggplot2\", ggplot object returned. method == \"base\", function draw graphics device return NULL, invisibly.","code":""},{"path":"https://keras3.posit.co/dev/reference/pop_layer.html","id":null,"dir":"Reference","previous_headings":"","what":"Remove the last layer in a Sequential model — pop_layer","title":"Remove the last layer in a Sequential model — pop_layer","text":"Remove last layer Sequential model","code":""},{"path":"https://keras3.posit.co/dev/reference/pop_layer.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Remove the last layer in a Sequential model — pop_layer","text":"","code":"pop_layer(object, rebuild = TRUE)"},{"path":"https://keras3.posit.co/dev/reference/pop_layer.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Remove the last layer in a Sequential model — pop_layer","text":"object Sequential keras model object rebuild bool. Whether rebuild model removing layer. Defaults TRUE.","code":""},{"path":"https://keras3.posit.co/dev/reference/pop_layer.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Remove the last layer in a Sequential model — pop_layer","text":"removed layer.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/predict.keras.src.models.model.Model.html","id":null,"dir":"Reference","previous_headings":"","what":"Generates output predictions for the input samples. — predict.keras.src.models.model.Model","title":"Generates output predictions for the input samples. — predict.keras.src.models.model.Model","text":"Generates output predictions input samples.","code":""},{"path":"https://keras3.posit.co/dev/reference/predict.keras.src.models.model.Model.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Generates output predictions for the input samples. — predict.keras.src.models.model.Model","text":"","code":"# S3 method for class 'keras.src.models.model.Model' predict(   object,   x,   ...,   batch_size = NULL,   verbose = getOption(\"keras.verbose\", default = \"auto\"),   steps = NULL,   callbacks = NULL )"},{"path":"https://keras3.posit.co/dev/reference/predict.keras.src.models.model.Model.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Generates output predictions for the input samples. — predict.keras.src.models.model.Model","text":"object Keras model object x Input samples. can : array (array-like), list arrays (case model multiple inputs). backend-native tensor, list tensors (case model multiple inputs). TF Dataset. Python generator function. ... forward/backward compatability. batch_size Integer NULL. Number samples per batch computation. unspecified, batch_size default 32. specify batch_size input data x form TF Dataset generator (since generate batches). verbose \"auto\", 0, 1, 2. Verbosity mode. 0 = silent, 1 = progress bar, 2 = one line per epoch. \"auto\" becomes 1 cases, 2 knitr render running distributed training server. Note progress bar particularly useful logged file, verbose=2 recommended running interactively (e.g., production environment). Defaults \"auto\". steps Total number steps (batches samples) draw declaring prediction round finished. steps NULL, predict() run x exhausted. case infinitely repeating dataset, predict() run indefinitely. callbacks List Callback instances. List callbacks apply prediction.","code":""},{"path":"https://keras3.posit.co/dev/reference/predict.keras.src.models.model.Model.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Generates output predictions for the input samples. — predict.keras.src.models.model.Model","text":"R array(s) predictions.","code":""},{"path":"https://keras3.posit.co/dev/reference/predict.keras.src.models.model.Model.html","id":"details","dir":"Reference","previous_headings":"","what":"Details","title":"Generates output predictions for the input samples. — predict.keras.src.models.model.Model","text":"Computation done batches. method designed batch processing large numbers inputs. intended use inside loops iterate data process small numbers inputs time. small numbers inputs fit one batch, directly call model model$call faster execution, e.g., model(x), model(x, training = FALSE) layers BatchNormalization behave differently inference.","code":""},{"path":"https://keras3.posit.co/dev/reference/predict.keras.src.models.model.Model.html","id":"note","dir":"Reference","previous_headings":"","what":"Note","title":"Generates output predictions for the input samples. — predict.keras.src.models.model.Model","text":"See FAQ entry details difference Model methods predict() call().","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/predict_on_batch.html","id":null,"dir":"Reference","previous_headings":"","what":"Returns predictions for a single batch of samples. — predict_on_batch","title":"Returns predictions for a single batch of samples. — predict_on_batch","text":"Returns predictions single batch samples.","code":""},{"path":"https://keras3.posit.co/dev/reference/predict_on_batch.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Returns predictions for a single batch of samples. — predict_on_batch","text":"","code":"predict_on_batch(object, x)"},{"path":"https://keras3.posit.co/dev/reference/predict_on_batch.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Returns predictions for a single batch of samples. — predict_on_batch","text":"object Keras model object x Input data. must array-like.","code":""},{"path":"https://keras3.posit.co/dev/reference/predict_on_batch.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Returns predictions for a single batch of samples. — predict_on_batch","text":"Array(s) predictions.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/process_utils.html","id":null,"dir":"Reference","previous_headings":"","what":"Preprocessing and postprocessing utilities — process_utils","title":"Preprocessing and postprocessing utilities — process_utils","text":"functions used preprocess postprocess inputs outputs Keras applications.","code":""},{"path":"https://keras3.posit.co/dev/reference/process_utils.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Preprocessing and postprocessing utilities — process_utils","text":"","code":"application_preprocess_inputs(model, x, ..., data_format = NULL)  application_decode_predictions(model, preds, top = 5L, ...)"},{"path":"https://keras3.posit.co/dev/reference/process_utils.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Preprocessing and postprocessing utilities — process_utils","text":"model Keras model initialized using application_ function. x batch inputs model. x missing, preprocess_input function appropriate model returned. ... Additional arguments passed preprocessing decoding function. data_format Optional data format image tensor/array. NULL means global setting config_image_data_format() used (unless changed , uses \"channels_last\"). Defaults NULL. preds batch outputs model. top number top predictions return.","code":""},{"path":"https://keras3.posit.co/dev/reference/process_utils.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Preprocessing and postprocessing utilities — process_utils","text":"list decoded predictions case application_decode_predictions(). batch preprocessed inputs case application_preprocess_inputs().","code":""},{"path":"https://keras3.posit.co/dev/reference/process_utils.html","id":"functions","dir":"Reference","previous_headings":"","what":"Functions","title":"Preprocessing and postprocessing utilities — process_utils","text":"application_preprocess_inputs(): Pre-process inputs used model application_decode_predictions(): Decode predictions model","code":""},{"path":"https://keras3.posit.co/dev/reference/process_utils.html","id":"ref-examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Preprocessing and postprocessing utilities — process_utils","text":"","code":"if (FALSE) { # \\dontrun{ model <- application_convnext_tiny()  inputs <- random_normal(c(32, 224, 224, 3)) processed_inputs <- application_preprocess_inputs(model, inputs)  preds <- random_normal(c(32, 1000)) decoded_preds <- application_decode_predictions(model, preds)  } # }"},{"path":"https://keras3.posit.co/dev/reference/quantize_weights.html","id":null,"dir":"Reference","previous_headings":"","what":"Quantize the weights of a model. — quantize_weights","title":"Quantize the weights of a model. — quantize_weights","text":"Note model must built first calling method. quantize_weights() recursively call layer$quantize(mode) layers skipped layer implement function. Currently Dense EinsumDense layers support quantization.","code":""},{"path":"https://keras3.posit.co/dev/reference/quantize_weights.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Quantize the weights of a model. — quantize_weights","text":"","code":"quantize_weights(object, mode, ...)"},{"path":"https://keras3.posit.co/dev/reference/quantize_weights.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Quantize the weights of a model. — quantize_weights","text":"object Keras Model Layer. mode mode quantization. 'int8' supported time. ... Passed object quantization method.","code":""},{"path":"https://keras3.posit.co/dev/reference/quantize_weights.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Quantize the weights of a model. — quantize_weights","text":"model, invisibly. Note just convenience usage |>, model modified -place.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/random_beta.html","id":null,"dir":"Reference","previous_headings":"","what":"Draw samples from a Beta distribution. — random_beta","title":"Draw samples from a Beta distribution. — random_beta","text":"values drawn Beta distribution parametrized alpha beta.","code":""},{"path":"https://keras3.posit.co/dev/reference/random_beta.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Draw samples from a Beta distribution. — random_beta","text":"","code":"random_beta(shape, alpha, beta, dtype = NULL, seed = NULL)"},{"path":"https://keras3.posit.co/dev/reference/random_beta.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Draw samples from a Beta distribution. — random_beta","text":"shape shape random values generate. alpha Float array floats representing first parameter alpha. Must broadcastable beta shape. beta Float array floats representing second parameter beta. Must broadcastable alpha shape. dtype Optional dtype tensor. floating point types supported. specified, config_floatx() used, defaults \"float32\" unless configured otherwise (via config_set_floatx(float_dtype)). seed Optional R integer instance random_seed_generator(). default, seed argument NULL, internal global random_seed_generator() used. seed argument can used ensure deterministic (repeatable) random number generation. Note passing integer seed value produce random values call. generate different random values repeated calls, instance random_seed_generator() must provided seed value. Remark concerning JAX backend: tracing functions JAX backend global random_seed_generator() supported. Therefore, tracing default value seed=NULL produce error, seed argument must provided.","code":""},{"path":"https://keras3.posit.co/dev/reference/random_beta.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Draw samples from a Beta distribution. — random_beta","text":"tensor random values.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/random_binomial.html","id":null,"dir":"Reference","previous_headings":"","what":"Draw samples from a Binomial distribution. — random_binomial","title":"Draw samples from a Binomial distribution. — random_binomial","text":"values drawn Binomial distribution specified trial count probability success.","code":""},{"path":"https://keras3.posit.co/dev/reference/random_binomial.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Draw samples from a Binomial distribution. — random_binomial","text":"","code":"random_binomial(shape, counts, probabilities, dtype = NULL, seed = NULL)"},{"path":"https://keras3.posit.co/dev/reference/random_binomial.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Draw samples from a Binomial distribution. — random_binomial","text":"shape shape random values generate. counts number array numbers representing number trials. must broadcastable probabilities. probabilities float array floats representing probability success individual event. must broadcastable counts. dtype Optional dtype tensor. floating point types supported. specified, config_floatx() used, defaults \"float32\" unless configured otherwise (via config_set_floatx(float_dtype)). seed Optional R integer instance random_seed_generator(). default, seed argument NULL, internal global random_seed_generator() used. seed argument can used ensure deterministic (repeatable) random number generation. Note passing integer seed value produce random values call. generate different random values repeated calls, instance random_seed_generator() must provided seed value. Remark concerning JAX backend: tracing functions JAX backend global random_seed_generator() supported. Therefore, tracing default value seed=NULL produce error, seed argument must provided.","code":""},{"path":"https://keras3.posit.co/dev/reference/random_binomial.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Draw samples from a Binomial distribution. — random_binomial","text":"tensor random values.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/random_categorical.html","id":null,"dir":"Reference","previous_headings":"","what":"Draws samples from a categorical distribution. — random_categorical","title":"Draws samples from a categorical distribution. — random_categorical","text":"function takes input logits, 2-D input tensor shape (batch_size, num_classes). row input represents categorical distribution, column index containing log-probability given class. function output 2-D tensor shape (batch_size, num_samples), row contains samples corresponding row logits. column index contains independent samples drawn input distribution.","code":"x <- matrix(c(100, .1, 99), nrow = 1) random_categorical(x, num_samples = 5, seed = 1234) ## tf.Tensor([[3 1 1 3 3]], shape=(1, 5), dtype=int32) random_categorical(x, num_samples = 5, seed = 1234,                    zero_indexed = TRUE) ## tf.Tensor([[2 0 0 2 2]], shape=(1, 5), dtype=int32) op_take(x, random_categorical(x, num_samples = 5, seed = 1234)) ## tf.Tensor([[ 99. 100. 100.  99.  99.]], shape=(1, 5), dtype=float64) op_take(x, random_categorical(x, num_samples = 5, seed = 1234,                               zero_indexed = TRUE),         zero_indexed = TRUE) ## tf.Tensor([[ 99. 100. 100.  99.  99.]], shape=(1, 5), dtype=float64)"},{"path":"https://keras3.posit.co/dev/reference/random_categorical.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Draws samples from a categorical distribution. — random_categorical","text":"","code":"random_categorical(   logits,   num_samples,   dtype = \"int32\",   seed = NULL,   zero_indexed = FALSE )"},{"path":"https://keras3.posit.co/dev/reference/random_categorical.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Draws samples from a categorical distribution. — random_categorical","text":"logits 2-D Tensor shape (batch_size, num_classes). row define categorical distribution unnormalized log-probabilities classes. num_samples Int, number independent samples draw row input. second dimension output tensor's shape. dtype Optional dtype output tensor. seed Optional R integer instance random_seed_generator(). default, seed argument NULL, internal global random_seed_generator() used. seed argument can used ensure deterministic (repeatable) random number generation. Note passing integer seed value produce random values call. generate different random values repeated calls, instance random_seed_generator() must provided seed value. Remark concerning JAX backend: tracing functions JAX backend global random_seed_generator() supported. Therefore, tracing default value seed=NULL produce error, seed argument must provided. zero_indexed TRUE, returned indices zero-based (0 encodes first position); FALSE (default), returned indices one-based (1 encodes first position).","code":""},{"path":"https://keras3.posit.co/dev/reference/random_categorical.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Draws samples from a categorical distribution. — random_categorical","text":"2-D tensor (batch_size, num_samples).","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/random_dropout.html","id":null,"dir":"Reference","previous_headings":"","what":"Randomly set some values in a tensor to 0. — random_dropout","title":"Randomly set some values in a tensor to 0. — random_dropout","text":"Randomly set portion values tensor 0.","code":""},{"path":"https://keras3.posit.co/dev/reference/random_dropout.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Randomly set some values in a tensor to 0. — random_dropout","text":"","code":"random_dropout(inputs, rate, noise_shape = NULL, seed = NULL)"},{"path":"https://keras3.posit.co/dev/reference/random_dropout.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Randomly set some values in a tensor to 0. — random_dropout","text":"inputs tensor rate numeric noise_shape shape() value seed Initial seed random number generator","code":""},{"path":"https://keras3.posit.co/dev/reference/random_dropout.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Randomly set some values in a tensor to 0. — random_dropout","text":"tensor copy inputs values set 0.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/random_gamma.html","id":null,"dir":"Reference","previous_headings":"","what":"Draw random samples from the Gamma distribution. — random_gamma","title":"Draw random samples from the Gamma distribution. — random_gamma","text":"Draw random samples Gamma distribution.","code":""},{"path":"https://keras3.posit.co/dev/reference/random_gamma.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Draw random samples from the Gamma distribution. — random_gamma","text":"","code":"random_gamma(shape, alpha, dtype = NULL, seed = NULL)"},{"path":"https://keras3.posit.co/dev/reference/random_gamma.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Draw random samples from the Gamma distribution. — random_gamma","text":"shape shape random values generate. alpha Float, parameter distribution. dtype Optional dtype tensor. floating point types supported. specified, config_floatx() used, defaults float32 unless configured otherwise (via config_set_floatx(float_dtype)). seed Optional R integer instance random_seed_generator(). default, seed argument NULL, internal global random_seed_generator() used. seed argument can used ensure deterministic (repeatable) random number generation. Note passing integer seed value produce random values call. generate different random values repeated calls, instance random_seed_generator() must provided seed value. Remark concerning JAX backend: tracing functions JAX backend global random_seed_generator() supported. Therefore, tracing default value seed=NULL produce error, seed argument must provided.","code":""},{"path":"https://keras3.posit.co/dev/reference/random_gamma.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Draw random samples from the Gamma distribution. — random_gamma","text":"tensor random values.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/random_integer.html","id":null,"dir":"Reference","previous_headings":"","what":"Draw random integers from a uniform distribution. — random_integer","title":"Draw random integers from a uniform distribution. — random_integer","text":"generated values follow uniform distribution range [minval, maxval). lower bound minval included range, upper bound maxval excluded. dtype must integer type.","code":""},{"path":"https://keras3.posit.co/dev/reference/random_integer.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Draw random integers from a uniform distribution. — random_integer","text":"","code":"random_integer(shape, minval, maxval, dtype = \"int32\", seed = NULL)"},{"path":"https://keras3.posit.co/dev/reference/random_integer.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Draw random integers from a uniform distribution. — random_integer","text":"shape shape random values generate. minval integer, lower bound range random values generate (inclusive). maxval integer, upper bound range random values generate (exclusive). dtype Optional dtype tensor. integer types supported. specified, \"int32\" used. seed Optional R integer instance random_seed_generator(). default, seed argument NULL, internal global random_seed_generator() used. seed argument can used ensure deterministic (repeatable) random number generation. Note passing integer seed value produce random values call. generate different random values repeated calls, instance random_seed_generator() must provided seed value. Remark concerning JAX backend: tracing functions JAX backend global random_seed_generator() supported. Therefore, tracing default value seed=NULL produce error, seed argument must provided.","code":""},{"path":"https://keras3.posit.co/dev/reference/random_integer.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Draw random integers from a uniform distribution. — random_integer","text":"tensor random values.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/random_normal.html","id":null,"dir":"Reference","previous_headings":"","what":"Draw random samples from a normal (Gaussian) distribution. — random_normal","title":"Draw random samples from a normal (Gaussian) distribution. — random_normal","text":"Draw random samples normal (Gaussian) distribution.","code":""},{"path":"https://keras3.posit.co/dev/reference/random_normal.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Draw random samples from a normal (Gaussian) distribution. — random_normal","text":"","code":"random_normal(shape, mean = 0, stddev = 1, dtype = NULL, seed = NULL)"},{"path":"https://keras3.posit.co/dev/reference/random_normal.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Draw random samples from a normal (Gaussian) distribution. — random_normal","text":"shape shape random values generate. mean Float, defaults 0. Mean random values generate. stddev Float, defaults 1. Standard deviation random values generate. dtype Optional dtype tensor. floating point types supported. specified, config_floatx() used, defaults float32 unless configured otherwise (via config_set_floatx(float_dtype)). seed Optional R integer instance random_seed_generator(). default, seed argument NULL, internal global random_seed_generator() used. seed argument can used ensure deterministic (repeatable) random number generation. Note passing integer seed value produce random values call. generate different random values repeated calls, instance random_seed_generator() must provided seed value. Remark concerning JAX backend: tracing functions JAX backend global random_seed_generator() supported. Therefore, tracing default value seed=NULL produce error, seed argument must provided.","code":""},{"path":"https://keras3.posit.co/dev/reference/random_normal.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Draw random samples from a normal (Gaussian) distribution. — random_normal","text":"tensor random values.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/random_seed_generator.html","id":null,"dir":"Reference","previous_headings":"","what":"Generates variable seeds upon each call to a function generating random numbers. — random_seed_generator","title":"Generates variable seeds upon each call to a function generating random numbers. — random_seed_generator","text":"Keras, random number generators (random_normal()) stateless, meaning pass integer seed (seed=42), return values repeated calls. get different values call, SeedGenerator providing state random generator used. Note random number generators default seed NULL, implies internal global SeedGenerator used. need decouple RNG global state can provide local StateGenerator either deterministic random initial state. Remark concerning JAX backen: Note use local StateGenerator seed argument required JIT compilation RNG JAX backend, use global state supported.","code":""},{"path":"https://keras3.posit.co/dev/reference/random_seed_generator.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Generates variable seeds upon each call to a function generating random numbers. — random_seed_generator","text":"","code":"random_seed_generator(seed = NULL, name = NULL, ...)"},{"path":"https://keras3.posit.co/dev/reference/random_seed_generator.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Generates variable seeds upon each call to a function generating random numbers. — random_seed_generator","text":"seed Initial seed random number generator name String, name object ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/random_seed_generator.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Generates variable seeds upon each call to a function generating random numbers. — random_seed_generator","text":"SeedGenerator instance, can passed seed =  argument random tensor generators.","code":""},{"path":"https://keras3.posit.co/dev/reference/random_seed_generator.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Generates variable seeds upon each call to a function generating random numbers. — random_seed_generator","text":"Usage layer:","code":"seed_gen <- random_seed_generator(seed = 42) values <- random_normal(shape = c(2, 3), seed = seed_gen) new_values <- random_normal(shape = c(2, 3), seed = seed_gen) layer_dropout2 <- new_layer_class(   \"dropout2\",   initialize = function(...) {     super$initialize(...)     self$seed_generator <- random_seed_generator(seed = 1337)   },   call = function(x, training = FALSE) {     if (training) {       return(random_dropout(x, rate = 0.5, seed = self$seed_generator))     }     return(x)   } )  out <- layer_dropout(rate = 0.8) out(op_ones(10), training = TRUE) ## tf.Tensor([0. 5. 5. 0. 0. 0. 0. 0. 0. 0.], shape=(10), dtype=float32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/random_shuffle.html","id":null,"dir":"Reference","previous_headings":"","what":"Shuffle the elements of a tensor uniformly at random along an axis. — random_shuffle","title":"Shuffle the elements of a tensor uniformly at random along an axis. — random_shuffle","text":"Shuffle elements tensor uniformly random along axis.","code":""},{"path":"https://keras3.posit.co/dev/reference/random_shuffle.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Shuffle the elements of a tensor uniformly at random along an axis. — random_shuffle","text":"","code":"random_shuffle(x, axis = 1L, seed = NULL)"},{"path":"https://keras3.posit.co/dev/reference/random_shuffle.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Shuffle the elements of a tensor uniformly at random along an axis. — random_shuffle","text":"x tensor shuffled. axis integer specifying axis along shuffle. Defaults 0. seed Optional R integer instance random_seed_generator(). default, seed argument NULL, internal global random_seed_generator() used. seed argument can used ensure deterministic (repeatable) random number generation. Note passing integer seed value produce random values call. generate different random values repeated calls, instance random_seed_generator() must provided seed value. Remark concerning JAX backend: tracing functions JAX backend global random_seed_generator() supported. Therefore, tracing default value seed=NULL produce error, seed argument must provided.","code":""},{"path":"https://keras3.posit.co/dev/reference/random_shuffle.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Shuffle the elements of a tensor uniformly at random along an axis. — random_shuffle","text":"tensor, copy x axis axis shuffled.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/random_truncated_normal.html","id":null,"dir":"Reference","previous_headings":"","what":"Draw samples from a truncated normal distribution. — random_truncated_normal","title":"Draw samples from a truncated normal distribution. — random_truncated_normal","text":"values drawn normal distribution specified mean standard deviation, discarding re-drawing samples two standard deviations mean.","code":""},{"path":"https://keras3.posit.co/dev/reference/random_truncated_normal.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Draw samples from a truncated normal distribution. — random_truncated_normal","text":"","code":"random_truncated_normal(shape, mean = 0, stddev = 1, dtype = NULL, seed = NULL)"},{"path":"https://keras3.posit.co/dev/reference/random_truncated_normal.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Draw samples from a truncated normal distribution. — random_truncated_normal","text":"shape shape random values generate. mean Float, defaults 0. Mean random values generate. stddev Float, defaults 1. Standard deviation random values generate. dtype Optional dtype tensor. floating point types supported. specified, config_floatx() used, defaults float32 unless configured otherwise (via config_set_floatx(float_dtype)) seed Optional R integer instance random_seed_generator(). default, seed argument NULL, internal global random_seed_generator() used. seed argument can used ensure deterministic (repeatable) random number generation. Note passing integer seed value produce random values call. generate different random values repeated calls, instance random_seed_generator() must provided seed value. Remark concerning JAX backend: tracing functions JAX backend global random_seed_generator() supported. Therefore, tracing default value seed=NULL produce error, seed argument must provided.","code":""},{"path":"https://keras3.posit.co/dev/reference/random_truncated_normal.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Draw samples from a truncated normal distribution. — random_truncated_normal","text":"tensor random values.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/random_uniform.html","id":null,"dir":"Reference","previous_headings":"","what":"Draw samples from a uniform distribution. — random_uniform","title":"Draw samples from a uniform distribution. — random_uniform","text":"generated values follow uniform distribution range [minval, maxval). lower bound minval included range, upper bound maxval excluded. dtype must floating point type, default range [0, 1).","code":""},{"path":"https://keras3.posit.co/dev/reference/random_uniform.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Draw samples from a uniform distribution. — random_uniform","text":"","code":"random_uniform(shape, minval = 0, maxval = 1, dtype = NULL, seed = NULL)"},{"path":"https://keras3.posit.co/dev/reference/random_uniform.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Draw samples from a uniform distribution. — random_uniform","text":"shape shape random values generate. minval Float, defaults 0. Lower bound range random values generate (inclusive). maxval Float, defaults 1. Upper bound range random values generate (exclusive). dtype Optional dtype tensor. floating point types supported. specified, config_floatx() used, defaults float32 unless configured otherwise (via config_set_floatx(float_dtype)) seed Optional R integer instance random_seed_generator(). default, seed argument NULL, internal global random_seed_generator() used. seed argument can used ensure deterministic (repeatable) random number generation. Note passing integer seed value produce random values call. generate different random values repeated calls, instance random_seed_generator() must provided seed value. Remark concerning JAX backend: tracing functions JAX backend global random_seed_generator() supported. Therefore, tracing default value seed=NULL produce error, seed argument must provided.","code":""},{"path":"https://keras3.posit.co/dev/reference/random_uniform.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Draw samples from a uniform distribution. — random_uniform","text":"tensor random values.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/reexports.html","id":null,"dir":"Reference","previous_headings":"","what":"Objects exported from other packages — reexports","title":"Objects exported from other packages — reexports","text":"objects imported packages. Follow links see documentation. dotty . generics compile, fit, generate magrittr %<>% reticulate %%, array_reshape, as_iterator, iter_next, iterate, np_array, tuple, use_python, use_virtualenv tensorflow all_dims, as_tensor, evaluate, export_savedmodel, tensorboard tfruns flag_boolean, flag_integer, flag_numeric, flag_string, flags, run_dir","code":""},{"path":"https://keras3.posit.co/dev/reference/register_keras_serializable.html","id":null,"dir":"Reference","previous_headings":"","what":"Registers a custom object with the Keras serialization framework. — register_keras_serializable","title":"Registers a custom object with the Keras serialization framework. — register_keras_serializable","text":"function registers custom class function Keras custom object registry, can serialized deserialized without needing entry user-provided custom_objects argument. also injects function Keras call get object's serializable string key. Note serialized deserialized, classes must implement get_config() method. Functions requirement. object registered key 'package>name' name, defaults object name passed.","code":""},{"path":"https://keras3.posit.co/dev/reference/register_keras_serializable.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Registers a custom object with the Keras serialization framework. — register_keras_serializable","text":"","code":"register_keras_serializable(object, name = NULL, package = NULL)"},{"path":"https://keras3.posit.co/dev/reference/register_keras_serializable.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Registers a custom object with the Keras serialization framework. — register_keras_serializable","text":"object keras object. name name serialize class package. package package class belongs . used key (\"package>name\") identify class. Defaults current package name, \"Custom\" outside package.","code":""},{"path":"https://keras3.posit.co/dev/reference/register_keras_serializable.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Registers a custom object with the Keras serialization framework. — register_keras_serializable","text":"object returned invisibly, convenient piping. primarily called side effects.","code":""},{"path":"https://keras3.posit.co/dev/reference/register_keras_serializable.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Registers a custom object with the Keras serialization framework. — register_keras_serializable","text":"","code":"# Note that `'my_package'` is used as the `package` argument here, and since # the `name` argument is not provided, `'MyDense'` is used as the `name`. layer_my_dense <- Layer(\"MyDense\") register_keras_serializable(layer_my_dense, package = \"my_package\")  MyDense <- environment(layer_my_dense)$`__class__` # the python class obj stopifnot(exprs = {   get_registered_object('my_package>MyDense') == MyDense   get_registered_name(MyDense) == 'my_package>MyDense' })"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/regularizer_l1.html","id":null,"dir":"Reference","previous_headings":"","what":"A regularizer that applies a L1 regularization penalty. — regularizer_l1","title":"A regularizer that applies a L1 regularization penalty. — regularizer_l1","text":"L1 regularization penalty computed : loss = l1 * reduce_sum(abs(x)) L1 may passed layer string identifier:   case, default value used l1=0.01.","code":"dense <- layer_dense(units = 3, kernel_regularizer = 'l1')"},{"path":"https://keras3.posit.co/dev/reference/regularizer_l1.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"A regularizer that applies a L1 regularization penalty. — regularizer_l1","text":"","code":"regularizer_l1(l1 = 0.01)"},{"path":"https://keras3.posit.co/dev/reference/regularizer_l1.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"A regularizer that applies a L1 regularization penalty. — regularizer_l1","text":"l1 float, L1 regularization factor.","code":""},{"path":"https://keras3.posit.co/dev/reference/regularizer_l1.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"A regularizer that applies a L1 regularization penalty. — regularizer_l1","text":"Regularizer instance can passed layer constructors used standalone object.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/regularizer_l1_l2.html","id":null,"dir":"Reference","previous_headings":"","what":"A regularizer that applies both L1 and L2 regularization penalties. — regularizer_l1_l2","title":"A regularizer that applies both L1 and L2 regularization penalties. — regularizer_l1_l2","text":"L1 regularization penalty computed : loss = l1 * reduce_sum(abs(x)) L2 regularization penalty computed loss = l2 * reduce_sum(square(x)) L1L2 may passed layer string identifier:   case, default values used l1=0.01 l2=0.01.","code":"dense <- layer_dense(units = 3, kernel_regularizer = 'L1L2')"},{"path":"https://keras3.posit.co/dev/reference/regularizer_l1_l2.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"A regularizer that applies both L1 and L2 regularization penalties. — regularizer_l1_l2","text":"","code":"regularizer_l1_l2(l1 = 0, l2 = 0)"},{"path":"https://keras3.posit.co/dev/reference/regularizer_l1_l2.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"A regularizer that applies both L1 and L2 regularization penalties. — regularizer_l1_l2","text":"l1 float, L1 regularization factor. l2 float, L2 regularization factor.","code":""},{"path":"https://keras3.posit.co/dev/reference/regularizer_l1_l2.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"A regularizer that applies both L1 and L2 regularization penalties. — regularizer_l1_l2","text":"Regularizer instance can passed layer constructors used standalone object.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/regularizer_l2.html","id":null,"dir":"Reference","previous_headings":"","what":"A regularizer that applies a L2 regularization penalty. — regularizer_l2","title":"A regularizer that applies a L2 regularization penalty. — regularizer_l2","text":"L2 regularization penalty computed : loss = l2 * reduce_sum(square(x)) L2 may passed layer string identifier:   case, default value used l2=0.01.","code":"dense <- layer_dense(units = 3, kernel_regularizer='l2')"},{"path":"https://keras3.posit.co/dev/reference/regularizer_l2.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"A regularizer that applies a L2 regularization penalty. — regularizer_l2","text":"","code":"regularizer_l2(l2 = 0.01)"},{"path":"https://keras3.posit.co/dev/reference/regularizer_l2.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"A regularizer that applies a L2 regularization penalty. — regularizer_l2","text":"l2 float, L2 regularization factor.","code":""},{"path":"https://keras3.posit.co/dev/reference/regularizer_l2.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"A regularizer that applies a L2 regularization penalty. — regularizer_l2","text":"Regularizer instance can passed layer constructors used standalone object.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/regularizer_orthogonal.html","id":null,"dir":"Reference","previous_headings":"","what":"Regularizer that encourages input vectors to be orthogonal to each other. — regularizer_orthogonal","title":"Regularizer that encourages input vectors to be orthogonal to each other. — regularizer_orthogonal","text":"can applied either rows matrix (mode=\"rows\") columns (mode=\"columns\"). applied Dense kernel shape (input_dim, units), rows mode seek make feature vectors (.e. basis output space) orthogonal .","code":""},{"path":"https://keras3.posit.co/dev/reference/regularizer_orthogonal.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Regularizer that encourages input vectors to be orthogonal to each other. — regularizer_orthogonal","text":"","code":"regularizer_orthogonal(factor = 0.01, mode = \"rows\")"},{"path":"https://keras3.posit.co/dev/reference/regularizer_orthogonal.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Regularizer that encourages input vectors to be orthogonal to each other. — regularizer_orthogonal","text":"factor Float. regularization factor. regularization penalty proportional factor times mean dot products L2-normalized rows (mode=\"rows\", columns mode=\"columns\") inputs, excluding product row/column .  Defaults 0.01. mode String, one {\"rows\", \"columns\"}. Defaults \"rows\". rows mode, regularization effect seeks make rows input orthogonal . columns mode, seeks make columns input orthogonal .","code":""},{"path":"https://keras3.posit.co/dev/reference/regularizer_orthogonal.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Regularizer that encourages input vectors to be orthogonal to each other. — regularizer_orthogonal","text":"Regularizer instance can passed layer constructors used standalone object.","code":""},{"path":"https://keras3.posit.co/dev/reference/regularizer_orthogonal.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Regularizer that encourages input vectors to be orthogonal to each other. — regularizer_orthogonal","text":"","code":"regularizer <- regularizer_orthogonal(factor=0.01) layer <- layer_dense(units=4, kernel_regularizer=regularizer)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/reset_state.html","id":null,"dir":"Reference","previous_headings":"","what":"Reset the state for a model, layer or metric. — reset_state","title":"Reset the state for a model, layer or metric. — reset_state","text":"Reset state model, layer metric.","code":""},{"path":"https://keras3.posit.co/dev/reference/reset_state.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Reset the state for a model, layer or metric. — reset_state","text":"","code":"reset_state(object)"},{"path":"https://keras3.posit.co/dev/reference/reset_state.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Reset the state for a model, layer or metric. — reset_state","text":"object Model, Layer, Metric instance Layers resettable state (E.g., adapt()-able preprocessing layers rnn layers resettable state, layer_dense() ). Calling Layer instance without resettable-state error.","code":""},{"path":"https://keras3.posit.co/dev/reference/reset_state.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Reset the state for a model, layer or metric. — reset_state","text":"object, invisibly.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/rnn_cell_gru.html","id":null,"dir":"Reference","previous_headings":"","what":"Cell class for the GRU layer. — rnn_cell_gru","title":"Cell class for the GRU layer. — rnn_cell_gru","text":"class processes one step within whole time sequence input, whereas layer_gru() processes whole sequence.","code":""},{"path":"https://keras3.posit.co/dev/reference/rnn_cell_gru.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Cell class for the GRU layer. — rnn_cell_gru","text":"","code":"rnn_cell_gru(   units,   activation = \"tanh\",   recurrent_activation = \"sigmoid\",   use_bias = TRUE,   kernel_initializer = \"glorot_uniform\",   recurrent_initializer = \"orthogonal\",   bias_initializer = \"zeros\",   kernel_regularizer = NULL,   recurrent_regularizer = NULL,   bias_regularizer = NULL,   kernel_constraint = NULL,   recurrent_constraint = NULL,   bias_constraint = NULL,   dropout = 0,   recurrent_dropout = 0,   reset_after = TRUE,   seed = NULL,   ... )"},{"path":"https://keras3.posit.co/dev/reference/rnn_cell_gru.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Cell class for the GRU layer. — rnn_cell_gru","text":"units Positive integer, dimensionality output space. activation Activation function use. Default: hyperbolic tangent (tanh). pass NULL, activation applied (ie. \"linear\" activation: (x) = x). recurrent_activation Activation function use recurrent step. Default: sigmoid (sigmoid). pass NULL, activation applied (ie. \"linear\" activation: (x) = x). use_bias Boolean, (default TRUE), whether layer use bias vector. kernel_initializer Initializer kernel weights matrix, used linear transformation inputs. Default: \"glorot_uniform\". recurrent_initializer Initializer recurrent_kernel weights matrix, used linear transformation recurrent state. Default: \"orthogonal\". bias_initializer Initializer bias vector. Default: \"zeros\". kernel_regularizer Regularizer function applied kernel weights matrix. Default: NULL. recurrent_regularizer Regularizer function applied recurrent_kernel weights matrix. Default: NULL. bias_regularizer Regularizer function applied bias vector. Default: NULL. kernel_constraint Constraint function applied kernel weights matrix. Default: NULL. recurrent_constraint Constraint function applied recurrent_kernel weights matrix. Default: NULL. bias_constraint Constraint function applied bias vector. Default: NULL. dropout Float 0 1. Fraction units drop linear transformation inputs. Default: 0. recurrent_dropout Float 0 1. Fraction units drop linear transformation recurrent state. Default: 0. reset_after GRU convention (whether apply reset gate matrix multiplication). FALSE = \"\", TRUE = \"\" (default cuDNN compatible). seed Random seed dropout. ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/rnn_cell_gru.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Cell class for the GRU layer. — rnn_cell_gru","text":"Layer instance, intended used layer_rnn().","code":""},{"path":"https://keras3.posit.co/dev/reference/rnn_cell_gru.html","id":"call-arguments","dir":"Reference","previous_headings":"","what":"Call Arguments","title":"Cell class for the GRU layer. — rnn_cell_gru","text":"inputs: 2D tensor, shape (batch, features). states: 2D tensor shape (batch, units), state previous time step. training: Python boolean indicating whether layer behave training mode inference mode. relevant dropout recurrent_dropout used.","code":""},{"path":"https://keras3.posit.co/dev/reference/rnn_cell_gru.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Cell class for the GRU layer. — rnn_cell_gru","text":"","code":"inputs <- random_uniform(c(32, 10, 8)) outputs <- inputs |> layer_rnn(rnn_cell_gru(4)) shape(outputs) ## shape(32, 4) rnn <- layer_rnn(    cell = rnn_cell_gru(4),    return_sequences=TRUE,    return_state=TRUE) c(whole_sequence_output, final_state) %<-% rnn(inputs) shape(whole_sequence_output) ## shape(32, 10, 4) shape(final_state) ## shape(32, 4)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/rnn_cell_lstm.html","id":null,"dir":"Reference","previous_headings":"","what":"Cell class for the LSTM layer. — rnn_cell_lstm","title":"Cell class for the LSTM layer. — rnn_cell_lstm","text":"class processes one step within whole time sequence input, whereas layer_lstm() processes whole sequence.","code":""},{"path":"https://keras3.posit.co/dev/reference/rnn_cell_lstm.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Cell class for the LSTM layer. — rnn_cell_lstm","text":"","code":"rnn_cell_lstm(   units,   activation = \"tanh\",   recurrent_activation = \"sigmoid\",   use_bias = TRUE,   kernel_initializer = \"glorot_uniform\",   recurrent_initializer = \"orthogonal\",   bias_initializer = \"zeros\",   unit_forget_bias = TRUE,   kernel_regularizer = NULL,   recurrent_regularizer = NULL,   bias_regularizer = NULL,   kernel_constraint = NULL,   recurrent_constraint = NULL,   bias_constraint = NULL,   dropout = 0,   recurrent_dropout = 0,   seed = NULL,   ... )"},{"path":"https://keras3.posit.co/dev/reference/rnn_cell_lstm.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Cell class for the LSTM layer. — rnn_cell_lstm","text":"units Positive integer, dimensionality output space. activation Activation function use. Default: hyperbolic tangent (tanh). pass NULL, activation applied (ie. \"linear\" activation: (x) = x). recurrent_activation Activation function use recurrent step. Default: sigmoid (sigmoid). pass NULL, activation applied (ie. \"linear\" activation: (x) = x). use_bias Boolean, (default TRUE), whether layer use bias vector. kernel_initializer Initializer kernel weights matrix, used linear transformation inputs. Default: \"glorot_uniform\". recurrent_initializer Initializer recurrent_kernel weights matrix, used linear transformation recurrent state. Default: \"orthogonal\". bias_initializer Initializer bias vector. Default: \"zeros\". unit_forget_bias Boolean (default TRUE). TRUE, add 1 bias forget gate initialization. Setting TRUE also force bias_initializer=\"zeros\". recommended Jozefowicz et al. kernel_regularizer Regularizer function applied kernel weights matrix. Default: NULL. recurrent_regularizer Regularizer function applied recurrent_kernel weights matrix. Default: NULL. bias_regularizer Regularizer function applied bias vector. Default: NULL. kernel_constraint Constraint function applied kernel weights matrix. Default: NULL. recurrent_constraint Constraint function applied recurrent_kernel weights matrix. Default: NULL. bias_constraint Constraint function applied bias vector. Default: NULL. dropout Float 0 1. Fraction units drop linear transformation inputs. Default: 0. recurrent_dropout Float 0 1. Fraction units drop linear transformation recurrent state. Default: 0. seed Random seed dropout. ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/rnn_cell_lstm.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Cell class for the LSTM layer. — rnn_cell_lstm","text":"Layer instance, intended used layer_rnn().","code":""},{"path":"https://keras3.posit.co/dev/reference/rnn_cell_lstm.html","id":"call-arguments","dir":"Reference","previous_headings":"","what":"Call Arguments","title":"Cell class for the LSTM layer. — rnn_cell_lstm","text":"inputs: 2D tensor, shape (batch, features). states: 2D tensor shape (batch, units), state previous time step. training: Boolean indicating whether layer behave training mode inference mode. relevant dropout recurrent_dropout used.","code":""},{"path":"https://keras3.posit.co/dev/reference/rnn_cell_lstm.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Cell class for the LSTM layer. — rnn_cell_lstm","text":"","code":"inputs <- random_uniform(c(32, 10, 8)) output <- inputs |>   layer_rnn(cell = rnn_cell_lstm(4)) shape(output) ## shape(32, 4) rnn <- layer_rnn(cell = rnn_cell_lstm(4),                  return_sequences = T,                  return_state = T) c(whole_sequence_output, ..final_state) %<-% rnn(inputs) str(whole_sequence_output) ## <tf.Tensor: shape=(32, 10, 4), dtype=float32, numpy=…> str(final_state) ## List of 2 ##  $ :<tf.Tensor: shape=(32, 4), dtype=float32, numpy=…> ##  $ :<tf.Tensor: shape=(32, 4), dtype=float32, numpy=…>"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/rnn_cell_simple.html","id":null,"dir":"Reference","previous_headings":"","what":"Cell class for SimpleRNN. — rnn_cell_simple","title":"Cell class for SimpleRNN. — rnn_cell_simple","text":"class processes one step within whole time sequence input, whereas layer_simple_rnn() processes whole sequence.","code":""},{"path":"https://keras3.posit.co/dev/reference/rnn_cell_simple.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Cell class for SimpleRNN. — rnn_cell_simple","text":"","code":"rnn_cell_simple(   units,   activation = \"tanh\",   use_bias = TRUE,   kernel_initializer = \"glorot_uniform\",   recurrent_initializer = \"orthogonal\",   bias_initializer = \"zeros\",   kernel_regularizer = NULL,   recurrent_regularizer = NULL,   bias_regularizer = NULL,   kernel_constraint = NULL,   recurrent_constraint = NULL,   bias_constraint = NULL,   dropout = 0,   recurrent_dropout = 0,   seed = NULL,   ... )"},{"path":"https://keras3.posit.co/dev/reference/rnn_cell_simple.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Cell class for SimpleRNN. — rnn_cell_simple","text":"units Positive integer, dimensionality output space. activation Activation function use. Default: hyperbolic tangent (tanh). pass NULL, activation applied (ie. \"linear\" activation: (x) = x). use_bias Boolean, (default TRUE), whether layer use bias vector. kernel_initializer Initializer kernel weights matrix, used linear transformation inputs. Default: \"glorot_uniform\". recurrent_initializer Initializer recurrent_kernel weights matrix, used linear transformation recurrent state. Default: \"orthogonal\". bias_initializer Initializer bias vector. Default: \"zeros\". kernel_regularizer Regularizer function applied kernel weights matrix. Default: NULL. recurrent_regularizer Regularizer function applied recurrent_kernel weights matrix. Default: NULL. bias_regularizer Regularizer function applied bias vector. Default: NULL. kernel_constraint Constraint function applied kernel weights matrix. Default: NULL. recurrent_constraint Constraint function applied recurrent_kernel weights matrix. Default: NULL. bias_constraint Constraint function applied bias vector. Default: NULL. dropout Float 0 1. Fraction units drop linear transformation inputs. Default: 0. recurrent_dropout Float 0 1. Fraction units drop linear transformation recurrent state. Default: 0. seed Random seed dropout. ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/rnn_cell_simple.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Cell class for SimpleRNN. — rnn_cell_simple","text":"Layer instance, intended used layer_rnn().","code":""},{"path":"https://keras3.posit.co/dev/reference/rnn_cell_simple.html","id":"call-arguments","dir":"Reference","previous_headings":"","what":"Call Arguments","title":"Cell class for SimpleRNN. — rnn_cell_simple","text":"sequence: 2D tensor, shape (batch, features). states: 2D tensor shape (batch, units), state previous time step. training: Python boolean indicating whether layer behave training mode inference mode. relevant dropout recurrent_dropout used.","code":""},{"path":"https://keras3.posit.co/dev/reference/rnn_cell_simple.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Cell class for SimpleRNN. — rnn_cell_simple","text":"","code":"inputs <- random_uniform(c(32, 10, 8)) rnn <- layer_rnn(cell = rnn_cell_simple(units = 4)) output <- rnn(inputs)  # The output has shape `(32, 4)`. rnn <- layer_rnn(     cell = rnn_cell_simple(units = 4),     return_sequences=TRUE,     return_state=TRUE ) # whole_sequence_output has shape `(32, 10, 4)`. # final_state has shape `(32, 4)`. c(whole_sequence_output, final_state) %<-% rnn(inputs)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/rnn_cells_stack.html","id":null,"dir":"Reference","previous_headings":"","what":"Wrapper allowing a stack of RNN cells to behave as a single cell. — rnn_cells_stack","title":"Wrapper allowing a stack of RNN cells to behave as a single cell. — rnn_cells_stack","text":"Used implement efficient stacked RNNs.","code":""},{"path":"https://keras3.posit.co/dev/reference/rnn_cells_stack.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Wrapper allowing a stack of RNN cells to behave as a single cell. — rnn_cells_stack","text":"","code":"rnn_cells_stack(cells, ...)"},{"path":"https://keras3.posit.co/dev/reference/rnn_cells_stack.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Wrapper allowing a stack of RNN cells to behave as a single cell. — rnn_cells_stack","text":"cells List RNN cell instances. ... Unnamed arguments treated additional cells. Named arguments passed underlying layer.","code":""},{"path":"https://keras3.posit.co/dev/reference/rnn_cells_stack.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Wrapper allowing a stack of RNN cells to behave as a single cell. — rnn_cells_stack","text":"Layer instance, intended used layer_rnn().","code":""},{"path":"https://keras3.posit.co/dev/reference/rnn_cells_stack.html","id":"example","dir":"Reference","previous_headings":"","what":"Example","title":"Wrapper allowing a stack of RNN cells to behave as a single cell. — rnn_cells_stack","text":"","code":"batch_size <- 3 sentence_length <- 5 num_features <- 2 new_shape <- c(batch_size, sentence_length, num_features) x <- array(1:30, dim = new_shape)  rnn_cells <- lapply(1:2, function(x) rnn_cell_lstm(units = 128)) stacked_lstm <- rnn_cells_stack(rnn_cells) lstm_layer <- layer_rnn(cell = stacked_lstm)  result <- lstm_layer(x) str(result) ## <tf.Tensor: shape=(3, 128), dtype=float32, numpy=…>"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/save_model.html","id":null,"dir":"Reference","previous_headings":"","what":"Saves a model as a .keras file. — save_model","title":"Saves a model as a .keras file. — save_model","text":"Saves model .keras file.","code":""},{"path":"https://keras3.posit.co/dev/reference/save_model.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Saves a model as a .keras file. — save_model","text":"","code":"save_model(model, filepath = NULL, overwrite = FALSE, zipped = NULL, ...)"},{"path":"https://keras3.posit.co/dev/reference/save_model.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Saves a model as a .keras file. — save_model","text":"model keras model. filepath string, Path save model. Must end .keras. overwrite Whether overwrite existing model target location, instead ask user via interactive prompt. zipped Whether save model zipped .keras archive (default saving locally), unzipped directory (default saving Hugging Face Hub). ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/save_model.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Saves a model as a .keras file. — save_model","text":"filepath provided, function called primarily side effects, model returned invisibly. filepath provided NULL, serialized model returned R raw vector.","code":""},{"path":"https://keras3.posit.co/dev/reference/save_model.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Saves a model as a .keras file. — save_model","text":"saved .keras file zip archive contains: model's configuration (architecture) model's weights model's optimizer's state () Thus models can reinstantiated exact state.","code":"model <- keras_model_sequential(input_shape = c(3)) |>   layer_dense(5) |>   layer_activation_softmax()  model |> save_model(\"model.keras\") loaded_model <- load_model(\"model.keras\") x <- random_uniform(c(10, 3)) stopifnot(all.equal(   model |> predict(x),   loaded_model |> predict(x) )) zip::zip_list(\"model.keras\")[, \"filename\"] ## [1] \"metadata.json\"    \"config.json\"      \"model.weights.h5\""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/save_model_config.html","id":null,"dir":"Reference","previous_headings":"","what":"Save and load model configuration as JSON — save_model_config","title":"Save and load model configuration as JSON — save_model_config","text":"Save re-load models configurations JSON. Note representation include weights, architecture.","code":""},{"path":"https://keras3.posit.co/dev/reference/save_model_config.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Save and load model configuration as JSON — save_model_config","text":"","code":"save_model_config(model, filepath = NULL, overwrite = FALSE)  load_model_config(filepath, custom_objects = NULL)"},{"path":"https://keras3.posit.co/dev/reference/save_model_config.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Save and load model configuration as JSON — save_model_config","text":"model Model object save filepath path json file model config. overwrite Whether overwrite existing model configuration json filepath, instead ask user via interactive prompt. custom_objects Optional named list mapping names custom classes functions considered deserialization.","code":""},{"path":"https://keras3.posit.co/dev/reference/save_model_config.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Save and load model configuration as JSON — save_model_config","text":"called primarily side effects. model returned, invisibly, enable usage pipe.","code":""},{"path":"https://keras3.posit.co/dev/reference/save_model_config.html","id":"details","dir":"Reference","previous_headings":"","what":"Details","title":"Save and load model configuration as JSON — save_model_config","text":"Note: save_model_config() serializes model JSON using serialize_keras_object(), get_config(). serialize_keras_object() returns superset get_config(), additional information needed create class object needed restore model. See example extract get_config() value saved model.","code":""},{"path":"https://keras3.posit.co/dev/reference/save_model_config.html","id":"example","dir":"Reference","previous_headings":"","what":"Example","title":"Save and load model configuration as JSON — save_model_config","text":"","code":"model <- keras_model_sequential(input_shape = 10) |> layer_dense(10) file <- tempfile(\"model-config-\", fileext = \".json\") save_model_config(model, file)  # load a new model instance with the same architecture but different weights model2 <- load_model_config(file)  stopifnot(exprs = {   all.equal(get_config(model), get_config(model2))    # To extract the `get_config()` value from a saved model config:   all.equal(       get_config(model),       structure(jsonlite::read_json(file)$config,                 \"__class__\" = keras_model_sequential()$`__class__`)   ) })"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/save_model_weights.html","id":null,"dir":"Reference","previous_headings":"","what":"Saves all weights to a single file or sharded files. — save_model_weights","title":"Saves all weights to a single file or sharded files. — save_model_weights","text":"default, weights saved single .weights.h5 file. Enable sharding via max_shard_size split weights across multiple files (GB) produce .weights.json manifest tracks shard metadata. saved sharded files contain: *.weights.json: configuration file containing metadata weight_map entries. *_xxxxxx.weights.h5: weight shards limited max_shard_size.","code":"model <-   keras_model_sequential(input_shape = 2) |>   layer_dense(4)  path_h5 <- tempfile(fileext = \".weights.h5\") path_json <- tempfile(fileext = \".weights.json\")  model |> save_model_weights(path_h5) model |> save_model_weights(path_json, max_shard_size = 0.01)  model |> load_model_weights(path_h5) model |> load_model_weights(path_json)"},{"path":"https://keras3.posit.co/dev/reference/save_model_weights.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Saves all weights to a single file or sharded files. — save_model_weights","text":"","code":"save_model_weights(model, filepath, overwrite = FALSE, max_shard_size = NULL)"},{"path":"https://keras3.posit.co/dev/reference/save_model_weights.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Saves all weights to a single file or sharded files. — save_model_weights","text":"model keras Model object. filepath Path weights saved. Accepts .weights.h5, sharding enabled, .weights.json manifest path. .weights.h5 provided sharding, filename overridden end .weights.json. overwrite Whether overwrite existing weights target location, instead ask user via interactive prompt. max_shard_size Numeric size GB sharded file. Use NULL disable sharding.","code":""},{"path":"https://keras3.posit.co/dev/reference/save_model_weights.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Saves all weights to a single file or sharded files. — save_model_weights","text":"called primarily side effects. model returned, invisibly, enable usage pipe.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/serialize_keras_object.html","id":null,"dir":"Reference","previous_headings":"","what":"Retrieve the full config by serializing the Keras object. — serialize_keras_object","title":"Retrieve the full config by serializing the Keras object. — serialize_keras_object","text":"serialize_keras_object() serializes Keras object named list represents object, reciprocal function deserialize_keras_object(). See deserialize_keras_object() information full config format.","code":""},{"path":"https://keras3.posit.co/dev/reference/serialize_keras_object.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Retrieve the full config by serializing the Keras object. — serialize_keras_object","text":"","code":"serialize_keras_object(obj)"},{"path":"https://keras3.posit.co/dev/reference/serialize_keras_object.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Retrieve the full config by serializing the Keras object. — serialize_keras_object","text":"obj Keras object serialize.","code":""},{"path":"https://keras3.posit.co/dev/reference/serialize_keras_object.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Retrieve the full config by serializing the Keras object. — serialize_keras_object","text":"named list represents object config. config expected contain simple types , can saved json. object can deserialized config via deserialize_keras_object().","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/set_random_seed.html","id":null,"dir":"Reference","previous_headings":"","what":"Sets all random seeds (Python, NumPy, and backend framework, e.g. TF). — set_random_seed","title":"Sets all random seeds (Python, NumPy, and backend framework, e.g. TF). — set_random_seed","text":"can use utility make almost Keras program fully deterministic. limitations apply cases network communications involved (e.g. parameter server distribution), creates additional sources randomness, certain non-deterministic cuDNN ops involved. sets: R session seed: set.seed() Python session seed: import random; random.seed(seed) Python NumPy seed: import numpy; numpy.random.seed(seed) TensorFlow seed: tf$random$set_seed(seed) (TF installed) Torch seed: import(\"torch\")$manual_seed(seed) (backend torch) disables Python hash randomization. Note TensorFlow seed set even using TensorFlow backend framework, since many workflows leverage tf$data pipelines (feature random shuffling). Likewise many workflows might leverage NumPy APIs.","code":""},{"path":"https://keras3.posit.co/dev/reference/set_random_seed.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Sets all random seeds (Python, NumPy, and backend framework, e.g. TF). — set_random_seed","text":"","code":"set_random_seed(seed)"},{"path":"https://keras3.posit.co/dev/reference/set_random_seed.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Sets all random seeds (Python, NumPy, and backend framework, e.g. TF). — set_random_seed","text":"seed Integer, random seed use.","code":""},{"path":"https://keras3.posit.co/dev/reference/set_random_seed.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Sets all random seeds (Python, NumPy, and backend framework, e.g. TF). — set_random_seed","text":"return value, called side effects.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/set_state_tree.html","id":null,"dir":"Reference","previous_headings":"","what":"Assigns values to variables of the model. — set_state_tree","title":"Assigns values to variables of the model. — set_state_tree","text":"method takes dictionary (named list) nested variable values, represents state tree model, assigns corresponding variables model. keys (list names) represent variable names (e.g., 'trainable_variables', 'optimizer_variables'), values nested dictionaries containing variable paths corresponding values.","code":""},{"path":"https://keras3.posit.co/dev/reference/set_state_tree.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Assigns values to variables of the model. — set_state_tree","text":"","code":"set_state_tree(object, state_tree)"},{"path":"https://keras3.posit.co/dev/reference/set_state_tree.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Assigns values to variables of the model. — set_state_tree","text":"object keras model. state_tree dictionary representing state tree model. keys variable names, values nested dictionaries representing variable paths values.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/shape.html","id":null,"dir":"Reference","previous_headings":"","what":"Tensor shape utility — shape","title":"Tensor shape utility — shape","text":"function can used get create tensor shape.","code":""},{"path":"https://keras3.posit.co/dev/reference/shape.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Tensor shape utility — shape","text":"","code":"shape(...)  # S3 method for class 'keras_shape' format(x, ..., prefix = TRUE)  # S3 method for class 'keras_shape' print(x, ...)  # S3 method for class 'keras_shape' x[...]  # S3 method for class 'keras_shape' as.integer(x, ...)  # S3 method for class 'keras_shape' Summary(..., na.rm = FALSE)  # S3 method for class 'keras_shape' as.list(x, ...)  # S3 method for class 'keras_shape' x == y  # S3 method for class 'keras_shape' x != y"},{"path":"https://keras3.posit.co/dev/reference/shape.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Tensor shape utility — shape","text":"... shape specification. Numerics, NULL tensors valid. NULL, NA, -1L can used specify unspecified dim size. Tensors dispatched op_shape() extract tensor shape. Values wrapped () used asis (see examples). objects coerced via .integer(). x, y keras_shape object. prefix Whether format shape object prefix. Defaults \"shape\". na.rm passed Summary group generics like prod(). Unknown axes treated NA.","code":""},{"path":"https://keras3.posit.co/dev/reference/shape.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Tensor shape utility — shape","text":"list \"keras_shape\" class attribute. element list either ) NULL, b) R integer c) scalar integer tensor (e.g., supplied TF tensor unspecified dimension function traced).","code":""},{"path":"https://keras3.posit.co/dev/reference/shape.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Tensor shape utility — shape","text":"3 ways specify unknown dimension     functions take 'shape' argument also coerce shape()     can also use shape() get shape tensor (excepting scalar integer tensors).             Combine expand shapes         Scalar integer tensors treated axis values. commonly encountered tracing function graph mode, axis size might unknown.     useful pattern unpack shape() %<-%, like :                     want resolve shape tensor can potentially scalar integer, can wrap tensor (), use op_shape().","code":"shape(1, 2, 3) ## shape(1, 2, 3) shape(NA,   2, 3) shape(NULL, 2, 3) shape(-1,   2, 3) ## shape(NA, 2, 3) ## shape(NA, 2, 3) ## shape(NA, 2, 3) layer_input(c(1, 2, 3)) layer_input(shape(1, 2, 3)) ## <KerasTensor shape=(None, 1, 2, 3), dtype=float32, sparse=False, ragged=False, name=keras_tensor> ## <KerasTensor shape=(None, 1, 2, 3), dtype=float32, sparse=False, ragged=False, name=keras_tensor_1> symbolic_tensor <- layer_input(shape(1, 2, 3)) shape(symbolic_tensor) ## shape(NA, 1, 2, 3) eager_tensor <- op_ones(c(1,2,3)) shape(eager_tensor) ## shape(1, 2, 3) op_shape(eager_tensor) ## shape(1, 2, 3) shape(symbolic_tensor, 4) ## shape(NA, 1, 2, 3, 4) shape(5, symbolic_tensor, 4) ## shape(5, NA, 1, 2, 3, 4) tfn <- tensorflow::tf_function(function(x) {   print(op_shape(x))   x }, input_signature = list(tensorflow::tf$TensorSpec(shape(1, NA, 3)))) invisible(tfn(op_ones(shape(1, 2, 3)))) ## shape(1, Tensor(\"strided_slice:0\", shape=(), dtype=int32), 3) c(batch_size, seq_len, channels) %<-% shape(x)  # `%<-%` also has support for skipping values # during unpacking with `.` and `...`. For example, # To retrieve just the first and/or last dim: c(batch_size, ...) %<-% shape(x) c(batch_size, ., .) %<-% shape(x) c(..., channels) %<-% shape(x) c(batch_size, ..., channels) %<-% shape(x) c(batch_size, ., channels) %<-% shape(x) echo_print <- function(x) {   message(\"> \", deparse(substitute(x)));   if(!is.null(x)) print(x) } tfn <- tensorflow::tf_function(function(x) {   c(axis1, axis2, axis3) %<-% shape(x)   echo_print(str(list(axis1 = axis1, axis2 = axis2, axis3 = axis3)))    echo_print(shape(axis1))               # use axis1 tensor as axis value   echo_print(shape(axis1, axis2, axis3)) # use axis1 tensor as axis value    # use shape() to compose a new shape, e.g., in multihead attention   n_heads <- 4   echo_print(shape(axis1, axis2, n_heads, axis3/n_heads))    x }, input_signature = list(tensorflow::tf$TensorSpec(shape(NA, 4, 16)))) invisible(tfn(op_ones(shape(2, 4, 16)))) ## > str(list(axis1 = axis1, axis2 = axis2, axis3 = axis3)) ## List of 3 ##  $ axis1:<tf.Tensor 'strided_slice:0' shape=() dtype=int32> ##  $ axis2: int 4 ##  $ axis3: int 16 ## > shape(axis1) ## shape(Tensor(\"strided_slice:0\", shape=(), dtype=int32)) ## > shape(axis1, axis2, axis3) ## shape(Tensor(\"strided_slice:0\", shape=(), dtype=int32), 4, 16) ## > shape(axis1, axis2, n_heads, axis3/n_heads) ## shape(Tensor(\"strided_slice:0\", shape=(), dtype=int32), 4, 4, 4) (x <- op_convert_to_tensor(2L)) ## tf.Tensor(2, shape=(), dtype=int32) # by default, shape() treats scalar integer tensors as axis values shape(x) ## shape(tf.Tensor(2, shape=(), dtype=int32)) # to access the shape of a scalar integer, # call `op_shape()`, or protect with `I()` op_shape(x) ## shape() shape(I(x)) ## shape()"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/split_dataset.html","id":null,"dir":"Reference","previous_headings":"","what":"Splits a dataset into a left half and a right half (e.g. train / test). — split_dataset","title":"Splits a dataset into a left half and a right half (e.g. train / test). — split_dataset","text":"Splits dataset left half right half (e.g. train / test).","code":""},{"path":"https://keras3.posit.co/dev/reference/split_dataset.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Splits a dataset into a left half and a right half (e.g. train / test). — split_dataset","text":"","code":"split_dataset(   dataset,   left_size = NULL,   right_size = NULL,   shuffle = FALSE,   seed = NULL )"},{"path":"https://keras3.posit.co/dev/reference/split_dataset.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Splits a dataset into a left half and a right half (e.g. train / test). — split_dataset","text":"dataset tf$data$Dataset, torch$utils$data$Dataset object, list arrays length. left_size float (range [0, 1]), signifies fraction data pack left dataset. integer, signifies number samples pack left dataset. NULL, defaults complement right_size. Defaults NULL. right_size float (range [0, 1]), signifies fraction data pack right dataset. integer, signifies number samples pack right dataset. NULL, defaults complement left_size. Defaults NULL. shuffle Boolean, whether shuffle data splitting . seed random seed shuffling.","code":""},{"path":"https://keras3.posit.co/dev/reference/split_dataset.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Splits a dataset into a left half and a right half (e.g. train / test). — split_dataset","text":"list two tf$data$Dataset objects: left right splits.","code":""},{"path":"https://keras3.posit.co/dev/reference/split_dataset.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Splits a dataset into a left half and a right half (e.g. train / test). — split_dataset","text":"","code":"data <- random_uniform(c(1000, 4)) c(left_ds, right_ds) %<-% split_dataset(list(data$numpy()), left_size = 0.8) left_ds$cardinality() ## tf.Tensor(800, shape=(), dtype=int64) right_ds$cardinality() ## tf.Tensor(200, shape=(), dtype=int64)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/summary.keras.src.models.model.Model.html","id":null,"dir":"Reference","previous_headings":"","what":"Print a summary of a Keras Model — summary.keras.src.models.model.Model","title":"Print a summary of a Keras Model — summary.keras.src.models.model.Model","text":"Print summary Keras Model","code":""},{"path":"https://keras3.posit.co/dev/reference/summary.keras.src.models.model.Model.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Print a summary of a Keras Model — summary.keras.src.models.model.Model","text":"","code":"# S3 method for class 'keras.src.models.model.Model' summary(object, ...)  # S3 method for class 'keras.src.models.model.Model' format(   x,   line_length = min(getOption(\"width\"), 180),   positions = NULL,   expand_nested = FALSE,   show_trainable = NA,   ...,   layer_range = NULL,   compact = TRUE )  # S3 method for class 'keras.src.models.model.Model' print(x, ...)"},{"path":"https://keras3.posit.co/dev/reference/summary.keras.src.models.model.Model.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Print a summary of a Keras Model — summary.keras.src.models.model.Model","text":"object, x Keras model instance ... summary() print(), passed format(). format(), passed model$summary(). line_length Total length printed lines positions Relative absolute positions log elements line. provided, defaults c(0.33, 0.55, 0.67, 1.0). expand_nested Whether expand nested models. provided, defaults FALSE. show_trainable Whether show layer trainable. provided, defaults FALSE. layer_range list, tuple, vector 2 strings, starting layer name ending layer name (inclusive) indicating range layers printed summary. also accepts regex patterns instead exact name. case, start predicate first element matches layer_range[[1]] end predicate last element matches layer_range[[1]]. default NULL considers layers model. compact Whether remove white-space lines model summary. (Default TRUE)","code":""},{"path":"https://keras3.posit.co/dev/reference/summary.keras.src.models.model.Model.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Print a summary of a Keras Model — summary.keras.src.models.model.Model","text":"format() returns length 1 character vector. print() returns model object invisibly. summary() returns output format() invisibly printing .","code":""},{"path":"https://keras3.posit.co/dev/reference/summary.keras.src.models.model.Model.html","id":"enabling-color-output-in-knitr-rmarkdown-quarto-","dir":"Reference","previous_headings":"","what":"Enabling color output in Knitr (RMarkdown, Quarto)","title":"Print a summary of a Keras Model — summary.keras.src.models.model.Model","text":"order enable color output quarto rmarkdown document html output format (include revealjs presentations), need following setup chunk:","code":"```{r setup, include = FALSE}  options(cli.num_colors = 256)  fansi::set_knit_hooks(knitr::knit_hooks)  options(width = 75) # adjust as needed for format  ```"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/test_on_batch.html","id":null,"dir":"Reference","previous_headings":"","what":"Test the model on a single batch of samples. — test_on_batch","title":"Test the model on a single batch of samples. — test_on_batch","text":"Test model single batch samples.","code":""},{"path":"https://keras3.posit.co/dev/reference/test_on_batch.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Test the model on a single batch of samples. — test_on_batch","text":"","code":"test_on_batch(object, x, y = NULL, sample_weight = NULL, ...)"},{"path":"https://keras3.posit.co/dev/reference/test_on_batch.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Test the model on a single batch of samples. — test_on_batch","text":"object Keras model object x Input data. Must array-like. y Target data. Must array-like. sample_weight Optional array length x, containing weights apply model's loss sample. case temporal data, can pass 2D array shape (samples, sequence_length), apply different weight every timestep every sample. ... forward/backward compatability","code":""},{"path":"https://keras3.posit.co/dev/reference/test_on_batch.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Test the model on a single batch of samples. — test_on_batch","text":"scalar loss value (metrics), named list loss metric values (metrics).","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/text_dataset_from_directory.html","id":null,"dir":"Reference","previous_headings":"","what":"Generates a tf.data.Dataset from text files in a directory. — text_dataset_from_directory","title":"Generates a tf.data.Dataset from text files in a directory. — text_dataset_from_directory","text":"directory structure :   calling text_dataset_from_directory(main_directory, labels='inferred') return tf.data.Dataset yields batches texts subdirectories class_a class_b, together labels 0 1 (0 corresponding class_a 1 corresponding class_b). .txt files supported time.","code":"main_directory/ ...class_a/ ......a_text_1.txt ......a_text_2.txt ...class_b/ ......b_text_1.txt ......b_text_2.txt"},{"path":"https://keras3.posit.co/dev/reference/text_dataset_from_directory.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Generates a tf.data.Dataset from text files in a directory. — text_dataset_from_directory","text":"","code":"text_dataset_from_directory(   directory,   labels = \"inferred\",   label_mode = \"int\",   class_names = NULL,   batch_size = 32L,   max_length = NULL,   shuffle = TRUE,   seed = NULL,   validation_split = NULL,   subset = NULL,   follow_links = FALSE,   verbose = TRUE )"},{"path":"https://keras3.posit.co/dev/reference/text_dataset_from_directory.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Generates a tf.data.Dataset from text files in a directory. — text_dataset_from_directory","text":"directory Directory data located. labels \"inferred\", contain subdirectories, containing text files class. Otherwise, directory structure ignored. labels Either \"inferred\" (labels generated directory structure), NULL (labels), list/tuple integer labels size number text files found directory. Labels sorted according alphanumeric order text file paths (obtained via os.walk(directory) Python). label_mode String describing encoding labels. Options : \"int\": means labels encoded integers (e.g. sparse_categorical_crossentropy loss). \"categorical\" means labels encoded categorical vector (e.g. categorical_crossentropy loss). \"binary\" means labels (can 2) encoded float32 scalars values 0 1 (e.g. binary_crossentropy). NULL (labels). class_names valid \"labels\" \"inferred\". explicit list class names (must match names subdirectories). Used control order classes (otherwise alphanumerical order used). batch_size Size batches data. NULL, data batched (dataset yield individual samples). Defaults 32. max_length Maximum size text string. Texts longer truncated max_length. shuffle Whether shuffle data. set FALSE, sorts data alphanumeric order. Defaults TRUE. seed Optional random seed shuffling transformations. validation_split Optional float 0 1, fraction data reserve validation. subset Subset data return. One \"training\", \"validation\" \"\". used validation_split set. subset=\"\", utility returns tuple two datasets (training validation datasets respectively). follow_links Whether visits subdirectories pointed symlinks. Defaults FALSE. verbose Whether display number information classes number files found. Defaults TRUE.","code":""},{"path":"https://keras3.posit.co/dev/reference/text_dataset_from_directory.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Generates a tf.data.Dataset from text files in a directory. — text_dataset_from_directory","text":"tf.data.Dataset object. label_mode NULL, yields string tensors shape (batch_size,), containing contents batch text files. Otherwise, yields tuple (texts, labels), texts shape (batch_size,) labels follows format described . Rules regarding labels format: label_mode int, labels int32 tensor shape (batch_size,). label_mode binary, labels float32 tensor 1s 0s shape (batch_size, 1). label_mode categorical, labels float32 tensor shape (batch_size, num_classes), representing one-hot encoding class index.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/time_distributed.html","id":null,"dir":"Reference","previous_headings":"","what":"layer_time_distributed — time_distributed","title":"layer_time_distributed — time_distributed","text":"time_distributed() alias layer_time_distributed(). See ?layer_time_distributed() full documentation.","code":""},{"path":"https://keras3.posit.co/dev/reference/time_distributed.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"layer_time_distributed — time_distributed","text":"","code":"time_distributed(object, layer, ...)"},{"path":"https://keras3.posit.co/dev/reference/time_distributed.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"layer_time_distributed — time_distributed","text":"object Object compose layer . tensor, array, sequential model. layer Layer instance. ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/time_distributed.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"layer_time_distributed — time_distributed","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/timeseries_dataset_from_array.html","id":null,"dir":"Reference","previous_headings":"","what":"Creates a dataset of sliding windows over a timeseries provided as array. — timeseries_dataset_from_array","title":"Creates a dataset of sliding windows over a timeseries provided as array. — timeseries_dataset_from_array","text":"function takes sequence data-points gathered equal intervals, along time series parameters length sequences/windows, spacing two sequence/windows, etc., produce batches timeseries inputs targets.","code":""},{"path":"https://keras3.posit.co/dev/reference/timeseries_dataset_from_array.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Creates a dataset of sliding windows over a timeseries provided as array. — timeseries_dataset_from_array","text":"","code":"timeseries_dataset_from_array(   data,   targets,   sequence_length,   sequence_stride = 1L,   sampling_rate = 1L,   batch_size = 128L,   shuffle = FALSE,   seed = NULL,   start_index = NULL,   end_index = NULL )"},{"path":"https://keras3.posit.co/dev/reference/timeseries_dataset_from_array.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Creates a dataset of sliding windows over a timeseries provided as array. — timeseries_dataset_from_array","text":"data array eager tensor containing consecutive data points (timesteps). first dimension expected time dimension. targets Targets corresponding timesteps data. targets[] target corresponding window starts index (see example 2 ). Pass NULL target data (case dataset yield input data). sequence_length Length output sequences (number timesteps). sequence_stride Period successive output sequences. stride s, output samples start index data[], data[+ s], data[+ 2 * s], etc. sampling_rate Period successive individual timesteps within sequences. rate r, timesteps data[], data[+ r], ... data[+ sequence_length] used creating sample sequence. batch_size Number timeseries samples batch (except maybe last one). NULL, data batched (dataset yield individual samples). shuffle Whether shuffle output samples, instead draw chronological order. seed Optional int; random seed shuffling. start_index Optional int; data points earlier (exclusive) start_index used output sequences. useful reserve part data test validation. end_index Optional int; data points later (exclusive) end_index used output sequences. useful reserve part data test validation.","code":""},{"path":"https://keras3.posit.co/dev/reference/timeseries_dataset_from_array.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Creates a dataset of sliding windows over a timeseries provided as array. — timeseries_dataset_from_array","text":"tf$data$Dataset instance. targets passed, dataset yields list (batch_of_sequences, batch_of_targets). , dataset yields batch_of_sequences. Example 1: Consider indices [0, 1, ... 98]. sequence_length=10,  sampling_rate=2, sequence_stride=3, shuffle=FALSE, dataset yield batches sequences composed following indices:   case last 2 data points discarded since full sequence can generated include (next sequence started index 81, thus last step gone 98). Example 2: Temporal regression. Consider array data scalar values, shape (steps,). generate dataset uses past 10 timesteps predict next timestep, use:     Example 3: Temporal regression many--many architectures. Consider two arrays scalar values X Y, shape (100,). resulting dataset consist samples 20 timestamps . samples overlap. generate dataset uses current timestamp predict corresponding target timestep, use:","code":"First sequence:  [0  2  4  6  8 10 12 14 16 18] Second sequence: [3  5  7  9 11 13 15 17 19 21] Third sequence:  [6  8 10 12 14 16 18 20 22 24] ... Last sequence:   [78 80 82 84 86 88 90 92 94 96] data <- op_array(1:20) input_data <- data[1:10] targets <- data[11:20] dataset <- timeseries_dataset_from_array(   input_data, targets, sequence_length=10) iter <- reticulate::as_iterator(dataset) reticulate::iter_next(iter) ## [[1]] ## tf.Tensor([[ 1  2  3  4  5  6  7  8  9 10]], shape=(1, 10), dtype=int32) ## ## [[2]] ## tf.Tensor([11], shape=(1), dtype=int32) X <- op_array(1:100) Y <- X*2  sample_length <- 20 input_dataset <- timeseries_dataset_from_array(     X, NULL, sequence_length=sample_length, sequence_stride=sample_length) target_dataset <- timeseries_dataset_from_array(     Y, NULL, sequence_length=sample_length, sequence_stride=sample_length)   inputs <- reticulate::as_iterator(input_dataset) %>% reticulate::iter_next() targets <- reticulate::as_iterator(target_dataset) %>% reticulate::iter_next()"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/to_categorical.html","id":null,"dir":"Reference","previous_headings":"","what":"Converts a class vector (integers) to binary class matrix. — to_categorical","title":"Converts a class vector (integers) to binary class matrix. — to_categorical","text":"E.g. use loss_categorical_crossentropy().","code":""},{"path":"https://keras3.posit.co/dev/reference/to_categorical.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Converts a class vector (integers) to binary class matrix. — to_categorical","text":"","code":"to_categorical(x, num_classes = NULL)"},{"path":"https://keras3.posit.co/dev/reference/to_categorical.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Converts a class vector (integers) to binary class matrix. — to_categorical","text":"x Array-like class values converted matrix (integers 0 num_classes - 1). R factors coerced integer offset 0-based, .e., .integer(x) - 1L. num_classes Total number classes. NULL, inferred max(x) + 1. Defaults NULL.","code":""},{"path":"https://keras3.posit.co/dev/reference/to_categorical.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Converts a class vector (integers) to binary class matrix. — to_categorical","text":"binary matrix representation input R array. class axis placed last.","code":""},{"path":"https://keras3.posit.co/dev/reference/to_categorical.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Converts a class vector (integers) to binary class matrix. — to_categorical","text":"","code":"a <- to_categorical(c(0, 1, 2, 3), num_classes=4) print(a) ##      [,1] [,2] [,3] [,4] ## [1,]    1    0    0    0 ## [2,]    0    1    0    0 ## [3,]    0    0    1    0 ## [4,]    0    0    0    1 b <- array(c(.9, .04, .03, .03,               .3, .45, .15, .13,               .04, .01, .94, .05,               .12, .21, .5, .17),               dim = c(4, 4)) loss <- op_categorical_crossentropy(a, b) loss ## tf.Tensor([0.41284522 0.45601739 0.54430155 0.80437282], shape=(4), dtype=float64) loss <- op_categorical_crossentropy(a, a) loss ## tf.Tensor([1.00000005e-07 1.00000005e-07 1.00000005e-07 1.00000005e-07], shape=(4), dtype=float64)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/train_on_batch.html","id":null,"dir":"Reference","previous_headings":"","what":"Runs a single gradient update on a single batch of data. — train_on_batch","title":"Runs a single gradient update on a single batch of data. — train_on_batch","text":"Runs single gradient update single batch data.","code":""},{"path":"https://keras3.posit.co/dev/reference/train_on_batch.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Runs a single gradient update on a single batch of data. — train_on_batch","text":"","code":"train_on_batch(object, x, y = NULL, sample_weight = NULL, class_weight = NULL)"},{"path":"https://keras3.posit.co/dev/reference/train_on_batch.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Runs a single gradient update on a single batch of data. — train_on_batch","text":"object Keras model object x Input data. Must array-like. y Target data. Must array-like. sample_weight Optional array length x, containing weights apply model's loss sample. case temporal data, can pass 2D array shape (samples, sequence_length), apply different weight every timestep every sample. class_weight Optional named list mapping class indices (integers, 0-based) weight (float) apply model's loss samples class training. can useful tell model \"pay attention\" samples -represented class. class_weight specified targets rank 2 greater, either y must one-hot encoded, explicit final dimension 1 must included sparse class labels.","code":""},{"path":"https://keras3.posit.co/dev/reference/train_on_batch.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Runs a single gradient update on a single batch of data. — train_on_batch","text":"scalar loss value (metrics), named list loss metric values (metrics). property model$metrics_names give display labels scalar outputs.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/use_backend.html","id":null,"dir":"Reference","previous_headings":"","what":"Configure a Keras backend — use_backend","title":"Configure a Keras backend — use_backend","text":"Configure Keras backend","code":""},{"path":"https://keras3.posit.co/dev/reference/use_backend.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Configure a Keras backend — use_backend","text":"","code":"use_backend(backend, gpu = NA)"},{"path":"https://keras3.posit.co/dev/reference/use_backend.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Configure a Keras backend — use_backend","text":"backend string, can \"tensorflow\", \"jax\", \"numpy\", \"torch\". gpu bool, whether use GPU. NA (default), attempt detect GPU availability Linux. M-series Macs, defaults FALSE TensorFlow TRUE JAX. Windows, defaults FALSE.","code":""},{"path":"https://keras3.posit.co/dev/reference/use_backend.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Configure a Keras backend — use_backend","text":"Called primarily side effects. Returns provided backend, invisibly.","code":""},{"path":"https://keras3.posit.co/dev/reference/use_backend.html","id":"details","dir":"Reference","previous_headings":"","what":"Details","title":"Configure a Keras backend — use_backend","text":"functions allow configuring backend keras use. Note one backend can configured time. function called library(keras3) calling functions within package (see example). experimental support changing backend keras initialized. using config_set_backend().","code":"library(keras3) use_backend(\"tensorflow\")"},{"path":"https://keras3.posit.co/dev/reference/with_custom_object_scope.html","id":null,"dir":"Reference","previous_headings":"","what":"Provide a scope with mappings of names to custom objects — with_custom_object_scope","title":"Provide a scope with mappings of names to custom objects — with_custom_object_scope","text":"Provide scope mappings names custom objects","code":""},{"path":"https://keras3.posit.co/dev/reference/with_custom_object_scope.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Provide a scope with mappings of names to custom objects — with_custom_object_scope","text":"","code":"with_custom_object_scope(objects, expr)"},{"path":"https://keras3.posit.co/dev/reference/with_custom_object_scope.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Provide a scope with mappings of names to custom objects — with_custom_object_scope","text":"objects Named list objects expr Expression evaluate","code":""},{"path":"https://keras3.posit.co/dev/reference/with_custom_object_scope.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Provide a scope with mappings of names to custom objects — with_custom_object_scope","text":"result evaluating expr within custom object scope.","code":""},{"path":"https://keras3.posit.co/dev/reference/with_custom_object_scope.html","id":"details","dir":"Reference","previous_headings":"","what":"Details","title":"Provide a scope with mappings of names to custom objects — with_custom_object_scope","text":"many elements Keras models can customized user objects (e.g. losses, metrics, regularizers, etc.). loading saved models use functions typically need explicitly map names user objects via custom_objects parameter. with_custom_object_scope() function provides alternative lets create named alias user object applies entire block code, automatically recognized loading saved models.","code":""},{"path":"https://keras3.posit.co/dev/reference/with_custom_object_scope.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Provide a scope with mappings of names to custom objects — with_custom_object_scope","text":"","code":"# define custom metric metric_top_3_categorical_accuracy <-   custom_metric(\"top_3_categorical_accuracy\", function(y_true, y_pred) {     metric_top_k_categorical_accuracy(y_true, y_pred, k = 3)   })  with_custom_object_scope(c(top_k_acc = sparse_top_k_cat_acc), {    # ...define model...    # compile model (refer to \"top_k_acc\" by name)   model |> compile(     loss = \"binary_crossentropy\",     optimizer = optimizer_nadam(),     metrics = c(\"top_k_acc\")   )    # save the model   model |> save_model(\"my_model.keras\")    # loading the model within the custom object scope doesn't   # require explicitly providing the custom_object   reloaded_model <- load_model(\"my_model.keras\") })"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/zip_lists.html","id":null,"dir":"Reference","previous_headings":"","what":"Zip lists — zip_lists","title":"Zip lists — zip_lists","text":"conceptually similar zip() Python, R functions purrr::transpose() data.table::transpose() (albeit, accepting elements ... instead single list), one crucial difference: provided objects named, matching done names, positions.","code":""},{"path":"https://keras3.posit.co/dev/reference/zip_lists.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Zip lists — zip_lists","text":"","code":"zip_lists(...)"},{"path":"https://keras3.posit.co/dev/reference/zip_lists.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Zip lists — zip_lists","text":"... R lists atomic vectors, optionally named.","code":""},{"path":"https://keras3.posit.co/dev/reference/zip_lists.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Zip lists — zip_lists","text":"inverted list","code":""},{"path":"https://keras3.posit.co/dev/reference/zip_lists.html","id":"details","dir":"Reference","previous_headings":"","what":"Details","title":"Zip lists — zip_lists","text":"arguments supplied must length. positional matching required, arguments provided must unnamed. matching names, arguments must set names, can different orders.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/zip_lists.html","id":"ref-examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Zip lists — zip_lists","text":"","code":"gradients <- list(\"grad_for_wt_1\", \"grad_for_wt_2\", \"grad_for_wt_3\") weights <- list(\"weight_1\", \"weight_2\", \"weight_3\") str(zip_lists(gradients, weights)) #> List of 3 #>  $ :List of 2 #>   ..$ : chr \"grad_for_wt_1\" #>   ..$ : chr \"weight_1\" #>  $ :List of 2 #>   ..$ : chr \"grad_for_wt_2\" #>   ..$ : chr \"weight_2\" #>  $ :List of 2 #>   ..$ : chr \"grad_for_wt_3\" #>   ..$ : chr \"weight_3\" str(zip_lists(gradient = gradients, weight = weights)) #> List of 3 #>  $ :List of 2 #>   ..$ gradient: chr \"grad_for_wt_1\" #>   ..$ weight  : chr \"weight_1\" #>  $ :List of 2 #>   ..$ gradient: chr \"grad_for_wt_2\" #>   ..$ weight  : chr \"weight_2\" #>  $ :List of 2 #>   ..$ gradient: chr \"grad_for_wt_3\" #>   ..$ weight  : chr \"weight_3\"  names(gradients) <- names(weights) <- paste0(\"layer_\", 1:3) str(zip_lists(gradients, weights[c(3, 1, 2)])) #> List of 3 #>  $ layer_1:List of 2 #>   ..$ : chr \"grad_for_wt_1\" #>   ..$ : chr \"weight_1\" #>  $ layer_2:List of 2 #>   ..$ : chr \"grad_for_wt_2\" #>   ..$ : chr \"weight_2\" #>  $ layer_3:List of 2 #>   ..$ : chr \"grad_for_wt_3\" #>   ..$ : chr \"weight_3\"  names(gradients) <- paste0(\"gradient_\", 1:3) try(zip_lists(gradients, weights)) # error, names don't match #> Error in zip_lists(gradients, weights) :  #>   All names of arguments provided to `zip_lists()` must match. Call `unname()` on each argument if you want positional matching # call unname directly for positional matching str(zip_lists(unname(gradients), unname(weights))) #> List of 3 #>  $ :List of 2 #>   ..$ : chr \"grad_for_wt_1\" #>   ..$ : chr \"weight_1\" #>  $ :List of 2 #>   ..$ : chr \"grad_for_wt_2\" #>   ..$ : chr \"weight_2\" #>  $ :List of 2 #>   ..$ : chr \"grad_for_wt_3\" #>   ..$ : chr \"weight_3\""},{"path":"https://keras3.posit.co/dev/news/index.html","id":"keras3-development-version","dir":"Changelog","previous_headings":"","what":"keras3 (development version)","title":"keras3 (development version)","text":"Added S3 methods JAX array: str, .array, .double, .integer, .numeric. Added str S3 method Keras Variables. layer_reshape() can now accept -1 sentinel automatically calculated axis size. Updated dependencies declared use_backend(\"jax\", gpu=TRUE) compatability keras-hub. Added training loop configuration helpers: config_max_epochs(), config_set_max_epochs(), config_max_steps_per_epoch(), config_set_max_steps_per_epoch(). caps can also set via KERAS_MAX_EPOCHS KERAS_MAX_STEPS_PER_EPOCH environment variables. Added config_is_nnx_enabled() check whether JAX NNX features enabled. LoRA-enabled layers (layer_dense(), layer_embedding(), layer_einsum_dense()) gain lora_alpha argument scale adaptation delta independently chosen rank. keras_variable() now accepts synchronization argument distributed strategies. Layer$add_weight() gains overwrite_with_gradient option layers now provide symbolic_call() method. Transposed convolution utilities now follow latest Keras API: op_conv_transpose() defaults strides = 1 layer_conv_*_transpose() layers expose output_padding precise shape control. layer_torch_module_wrapper() gains output_shape argument help Keras infer shapes wrapping PyTorch modules. save_model_weights() adds max_shard_size argument split large weight files manageable shards. Added elastic deformation utilities images: layer_random_elastic_transform() lower-level op_image_elastic_transform(). Added loss_categorical_generalized_cross_entropy() training noisy labels. Added Muon optimizer via optimizer_muon(). Added complex-valued helpers: S3 Arg() methods tensors, op_angle(), conversions op_view_as_real() / op_view_as_complex(). Added signal window operations: op_bartlett(), op_blackman(), op_hamming(), op_hanning(), op_kaiser(). Expanded numeric operations op_layer_normalization(), op_cbrt(), op_corrcoef(), op_deg2rad(), op_heaviside(), new op_sparse_sigmoid() plus matching activation_sparse_sigmoid(), attn_logits_soft_cap argument op_dot_product_attention(). layer_layer_normalization() removes rms_scaling argument.","code":""},{"path":"https://keras3.posit.co/dev/news/index.html","id":"keras3-140","dir":"Changelog","previous_headings":"","what":"keras3 1.4.0","title":"keras3 1.4.0","text":"CRAN release: 2025-05-04 New op_subset() x@r[...] methods enable tensor subsetting using R’s [ semantics idioms. New subset assignment methods implemented tensors: op_subset(x, ...) <- value x@r[...] <- value Breaking changes: operations prefixed op_ now return 1-based indices default. following functions return consume indices changed: op_argmax(), op_argmin(), op_top_k(), op_argpartition(), op_searchsorted(), op_argsort(), op_digitize(), op_nonzero(), op_split(), op_trace(), op_swapaxes(), op_ctc_decode(), op_ctc_loss(), op_one_hot(), op_arange() op_arange() now matches semantics base::seq(). default starts, includes end value, automatically infers step direction. op_one_hot() now infers num_classes supplied factor. op_hstack() op_vstack() now accept arguments passed via .... application_decode_predictions() now returns processed data frame default decoder function predictions missing. application_preprocess_inputs() returns preprocessor function inputs missing. Various new examples added documentation, including op_scatter(), op_switch(), op_nonzero(). New x@py[...] accessor introduced Python-style 0-based indexing tensors. New Summary group generic method keras_shape, enabling usage like prod(shape(3, 4)) KERAS_HOME now set tools::R_user_dir(\"keras3\", \"cache\") ~/.keras exist KERAS_HOME unset. new op_convert_to_array() convert tensor R array. Added compatibility Keras v3.9.2. New operations added: op_rot90() op_rearrange() (Einops-style) op_signbit() op_polar() op_image_perspective_transform() op_image_gaussian_blur() New layers introduced: layer_rms_normalization() layer_aug_mix() layer_cut_mix() layer_random_invert() layer_random_erasing() layer_random_gaussian_blur() layer_random_perspective() layer_resizing() gains antialias argument. keras_input(), keras_model_sequential(), op_convert_to_tensor() gain ragged argument. layer$pop_layer() gains rebuild argument now returns removed layer. New rematerialized_call() method added Layer objects. Documentation improvements minor fixes. Fixed issue op_shape() sometimes return TensorFlow TensorShape Fixes metric_iou(), op_top_k(), op_eye() called R atomic doubles","code":""},{"path":"https://keras3.posit.co/dev/news/index.html","id":"keras3-130","dir":"Changelog","previous_headings":"","what":"keras3 1.3.0","title":"keras3 1.3.0","text":"CRAN release: 2025-03-03 Keras now uses reticulate::py_require() resolve Python dependencies. Calling install_keras() longer required (still supported). use_backend() gains gpu argument, specify GPU-capable set dependencies resolved py_require(). progress bar fit(), evaluate() predict() now defaults presenting testthat tests. dotty::. now reexported. %*% now dispatches op_matmul() tensorflow tensors, relaxed shape constraints compared tf$matmul(). Fixed issue calling Metric Loss object unnamed arguments error.","code":""},{"path":"https://keras3.posit.co/dev/news/index.html","id":"added-compatibility-with-keras-v380-user-facing-changes-1-3-0","dir":"Changelog","previous_headings":"","what":"Added compatibility with Keras v3.8.0. User-facing changes:","title":"keras3 1.3.0","text":"activation_sparse_plus() activation_sparsemax() activation_threshold() layer_equalization() layer_mix_up() layer_rand_augment() layer_random_color_degeneration() layer_random_color_jitter() layer_random_grayscale() layer_random_hue() layer_random_posterization() layer_random_saturation() layer_random_sharpness() layer_random_shear() op_diagflat() op_sparse_plus() op_sparsemax() op_threshold() op_unravel_index() Add argument axis tversky loss New: ONNX model export export_savedmodel() Doc improvements bug fixes. JAX specific changes: Add support JAX named scope TensorFlow specific changes: Make random_shuffle() XLA compilable","code":""},{"path":[]},{"path":[]},{"path":"https://keras3.posit.co/dev/news/index.html","id":"activations-1-3-0","dir":"Changelog","previous_headings":"Added compatibility with Keras v3.7.0. User-facing changes: > New functions","what":"Activations","title":"keras3 1.3.0","text":"activation_celu() activation_glu() activation_hard_shrink() activation_hard_tanh() activation_log_sigmoid() activation_soft_shrink() activation_squareplus() activation_tanh_shrink()","code":""},{"path":"https://keras3.posit.co/dev/news/index.html","id":"configuration-1-3-0","dir":"Changelog","previous_headings":"Added compatibility with Keras v3.7.0. User-facing changes: > New functions","what":"Configuration","title":"keras3 1.3.0","text":"config_disable_flash_attention() config_enable_flash_attention() config_is_flash_attention_enabled()","code":""},{"path":"https://keras3.posit.co/dev/news/index.html","id":"layers-and-initializers-1-3-0","dir":"Changelog","previous_headings":"Added compatibility with Keras v3.7.0. User-facing changes: > New functions","what":"Layers and Initializers","title":"keras3 1.3.0","text":"initializer_stft() layer_max_num_bounding_boxes() layer_stft_spectrogram()","code":""},{"path":"https://keras3.posit.co/dev/news/index.html","id":"losses-and-metrics-1-3-0","dir":"Changelog","previous_headings":"Added compatibility with Keras v3.7.0. User-facing changes: > New functions","what":"Losses and Metrics","title":"keras3 1.3.0","text":"loss_circle() metric_concordance_correlation() metric_pearson_correlation()","code":""},{"path":"https://keras3.posit.co/dev/news/index.html","id":"operations-1-3-0","dir":"Changelog","previous_headings":"Added compatibility with Keras v3.7.0. User-facing changes: > New functions","what":"Operations","title":"keras3 1.3.0","text":"op_celu() op_exp2() op_glu() op_hard_shrink() op_hard_tanh() op_ifft2() op_inner() op_soft_shrink() op_squareplus() op_tanh_shrink()","code":""},{"path":"https://keras3.posit.co/dev/news/index.html","id":"new-arguments-1-3-0","dir":"Changelog","previous_headings":"Added compatibility with Keras v3.7.0. User-facing changes: > New functions","what":"New arguments","title":"keras3 1.3.0","text":"callback_backup_and_restore(): Added double_checkpoint argument save fallback checkpoint callback_tensorboard(): Added support profile_batch argument layer_group_query_attention(): Added flash_attention seed arguments layer_multi_head_attention(): Added flash_attention argument metric_sparse_top_k_categorical_accuracy(): Added from_sorted_ids argument","code":""},{"path":"https://keras3.posit.co/dev/news/index.html","id":"performance-improvements-1-3-0","dir":"Changelog","previous_headings":"Added compatibility with Keras v3.7.0. User-facing changes:","what":"Performance improvements","title":"keras3 1.3.0","text":"Added native Flash Attention support GPU (via cuDNN) TPU (via Pallas kernel) JAX backend Added opt-native Flash Attention support GPU PyTorch backend Enabled additional kernel fusion via bias_add TensorFlow backend Added support Intel XPU devices PyTorch backend install_keras() changes: GPU available, default now install CPU build TensorFlow GPU build JAX. use GPU current session, call use_backend(\"jax\").","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/news/index.html","id":"breaking-changes-1-3-0","dir":"Changelog","previous_headings":"Added compatibility with Keras v3.6.0. User-facing changes:","what":"Breaking changes:","title":"keras3 1.3.0","text":"using get_file() extract = TRUE untar = TRUE, return value now path extracted directory, rather path archive.","code":""},{"path":"https://keras3.posit.co/dev/news/index.html","id":"other-changes-and-additions-1-3-0","dir":"Changelog","previous_headings":"Added compatibility with Keras v3.6.0. User-facing changes:","what":"Other changes and additions:","title":"keras3 1.3.0","text":"Logging now asynchronous fit(), evaluate(), predict(). enables 100% compact stacking train_step calls accelerators (e.g. running small models TPU). using custom callbacks rely on_batch_end, disable async logging. can re-enable adding self$async_safe <- TRUE callbacks. Note TensorBoard callback considered async-safe default. Default callbacks like progress bar async-safe. New bitwise operations: op_bitwise_and() op_bitwise_invert() op_bitwise_left_shift() op_bitwise_not() op_bitwise_or() op_bitwise_right_shift() op_bitwise_xor() New math operations: op_logdet() op_trunc() op_histogram() New neural network operation: op_dot_product_attention() New image preprocessing layers: layer_auto_contrast() layer_solarization() New Model functions get_state_tree() set_state_tree(), retrieving model variables, including trainable, non-trainable, optimizer variables, metric variables. New layer_pipeline() composing sequence layers. class useful building preprocessing pipeline. Compared keras_model_sequential(), layer_pipeline() key differences: ’s Model, just plain layer. layers pipeline compatible tf.data, pipeline also remain tf.data compatible, regardless backend use. New argument: export_savedmodel(verbose = ) New argument: op_normalize(epsilon = ) Various documentation improvements bug fixes.","code":""},{"path":"https://keras3.posit.co/dev/news/index.html","id":"keras3-120","dir":"Changelog","previous_headings":"","what":"keras3 1.2.0","title":"keras3 1.2.0","text":"CRAN release: 2024-09-05 Added compatibility Keras v3.5.0. User facing changes: op_associative_scan() op_searchsorted() optimizer_lamb() keras$DTypePolicy instances can now supplied dtype argument losses, metrics, layers. Add integration Hugging Face Hub. can now save models Hugging Face Hub directly save_model() load .keras models directly Hugging Face Hub load_model(). Added compatibility NumPy 2.0. Improved keras$distribution API support large models. Bug fixes performance improvements. Add data_format argument layer_zero_padding_1d() layer. Miscellaneous documentation improvements. Bug fixes performance improvements.","code":""},{"path":"https://keras3.posit.co/dev/news/index.html","id":"keras3-110","dir":"Changelog","previous_headings":"","what":"keras3 1.1.0","title":"keras3 1.1.0","text":"CRAN release: 2024-07-17 Fixed issue GPUs found running Windows WSL Linux. (reported #1456, fixed #1459) keras_shape objects (returned keras3::shape()) gain == != methods. Fixed warning tfruns::training_run() unable log optimizer learning rate. Added compatibility Keras v3.4.1 (R user facing changes). Added compatibility Keras v3.4.0. User facing changes: op_argpartition() op_map() op_scan() op_switch() op_dtype() op_lstsq() op_image_hsv_to_rgb() op_image_rgb_to_hsv() Added support arbitrary, deeply nested input/output structures Functional models (e.g. lists lists lists inputs outputs…) keras_input() gains optional argument. keras_model_sequential() gains input_optional argument. Add support float8 inference Dense EinsumDense layers. Enable layer_feature_space() used tfdatasets pipeline even backend isn’t TensorFlow. layer_string_lookup() can now take tf$SparseTensor() input. layer_string_lookup() returns \"int64\" dtype default modes now. Layer() instances gain attributes path quantization_mode. Metric()$variables now recursive. Add training argument Model$compute_loss(). split_dataset() now supports nested structures dataset. applications gain name argument, accept custom name. layer_multi_head_attention() gains seed argument. losses gain dtype argument. loss_dice() gains axis argument. op_ctc_decode(), new default mask_index = 0 op_image_* functions now use default data_format value config_image_data_format() op_isclose() gains arguments rtol, atol, equal_nan. save_model() gains argument zipped. Bugs fixes performance improvements.","code":""},{"path":"https://keras3.posit.co/dev/news/index.html","id":"keras3-100","dir":"Changelog","previous_headings":"","what":"keras3 1.0.0","title":"keras3 1.0.0","text":"CRAN release: 2024-05-21 Chains layer_* calls |> now instantiate layers order %>% pipe chains: left-hand-side first (#1440). iterate(), iter_next() as_iterator() now reexported reticulate. User facing changes upstream Keras v3.3.3: new functions: op_slogdet(), op_psnr() clone_model() gains new args: call_function, recursive Updated example usage. op_ctc_decode() strategy argument new default: \"greedy\". Updated docs. loss_ctc() default name fixed, changed \"ctc\" User facing changes upstream Keras v3.3.2: new function: op_ctc_decode() new function: op_eigh() new function: op_select() new function: op_vectorize() new function: op_image_rgb_to_grayscale() new function: loss_tversky() new args: layer_resizing(pad_to_aspect_ratio, fill_mode, fill_value) new arg: layer_embedding(weights) providing initial weights matrix new args: op_nan_to_num(nan, posinf, neginf) new args: op_image_resize(crop_to_aspect_ratio, pad_to_aspect_ratio, fill_mode, fill_value) new args: op_argmax(keepdims) op_argmin(keepdims) new arg: clear_session(free_memory) clearing without invoking garbage collector. metric_kl_divergence() loss_kl_divergence() clip inputs (y_true y_pred) [0, 1] range. new Layer() attributes: metrics, dtype_policy Added initial support float8 training layer_conv_*d() layers now support LoRa op_digitize() now supports sparse tensors. Models layers now return owned metrics recursively. Add pickling support Keras models. (e.g., via reticulate::py_save_object()) Note pickling recommended, prefer using Keras saving APIs.","code":""},{"path":"https://keras3.posit.co/dev/news/index.html","id":"keras3-020","dir":"Changelog","previous_headings":"","what":"keras3 0.2.0","title":"keras3 0.2.0","text":"CRAN release: 2024-04-18 New functions: quantize_weights(): quantize model layer weights -place. Currently, Dense, EinsumDense, Embedding layers supported (enough cover majority transformers today) layer_mel_spectrogram() layer_flax_module_wrapper() layer_jax_model_wrapper() loss_dice() random_beta() random_binomial() config_set_backend(): change backend Keras initialized. config_dtype_policy() config_set_dtype_policy() New Ops op_custom_gradient() op_batch_normalization() op_image_crop() op_divide_no_nan() op_normalize() op_correlate() ` New family linear algebra ops op_cholesky() op_det() op_eig() op_inv() op_lu_factor() op_norm() op_erfinv() op_solve_triangular() op_svd() audio_dataset_from_directory(), image_dataset_from_directory() text_dataset_from_directory() gain verbose argument (default TRUE) image_dataset_from_directory() gains pad_to_aspect_ratio argument (default FALSE) to_categorical(), op_one_hot(), fit() can now accept R factors, offset 0-based (reported #1055). op_convert_to_numpy() now returns unconverted NumPy arrays. op_array() op_convert_to_tensor() longer error casting R doubles integer types. export_savedmodel() now works Jax backend. Metric()$add_variable() method gains arg: aggregration. Layer()$add_weight() method gains args: autocast, regularizer, aggregation. op_bincount(), op_multi_hot(), op_one_hot(), layer_category_encoding() now support sparse tensors. op_custom_gradient() now supports PyTorch backend layer_lstm() layer_gru() gain arg use_cudnn, default 'auto'. Fixed issue application_preprocess_inputs() error supplied R array input. Doc improvements.","code":""},{"path":"https://keras3.posit.co/dev/news/index.html","id":"keras3-010","dir":"Changelog","previous_headings":"","what":"keras3 0.1.0","title":"keras3 0.1.0","text":"CRAN release: 2024-02-17 package rebuilt Keras 3.0. Refer https://blogs.rstudio.com/ai/posts/2024-05-21-keras3/ overview https://keras3.posit.co current --date documentation.","code":""}]
diff --git a/docs/dev/sitemap.xml b/docs/dev/sitemap.xml
index a36eaa348..22cb169ca 100644
--- a/docs/dev/sitemap.xml
+++ b/docs/dev/sitemap.xml
@@ -61,6 +61,7 @@
 <url><loc>https://keras3.posit.co/dev/reference/activation_softplus.html</loc></url>
 <url><loc>https://keras3.posit.co/dev/reference/activation_softsign.html</loc></url>
 <url><loc>https://keras3.posit.co/dev/reference/activation_sparse_plus.html</loc></url>
+<url><loc>https://keras3.posit.co/dev/reference/activation_sparse_sigmoid.html</loc></url>
 <url><loc>https://keras3.posit.co/dev/reference/activation_sparsemax.html</loc></url>
 <url><loc>https://keras3.posit.co/dev/reference/activation_squareplus.html</loc></url>
 <url><loc>https://keras3.posit.co/dev/reference/activation_tanh.html</loc></url>
@@ -139,7 +140,9 @@
 <url><loc>https://keras3.posit.co/dev/reference/config_image_data_format.html</loc></url>
 <url><loc>https://keras3.posit.co/dev/reference/config_is_flash_attention_enabled.html</loc></url>
 <url><loc>https://keras3.posit.co/dev/reference/config_is_interactive_logging_enabled.html</loc></url>
+<url><loc>https://keras3.posit.co/dev/reference/config_is_nnx_enabled.html</loc></url>
 <url><loc>https://keras3.posit.co/dev/reference/config_is_traceback_filtering_enabled.html</loc></url>
+<url><loc>https://keras3.posit.co/dev/reference/config_max_epochs.html</loc></url>
 <url><loc>https://keras3.posit.co/dev/reference/config_set_backend.html</loc></url>
 <url><loc>https://keras3.posit.co/dev/reference/config_set_dtype_policy.html</loc></url>
 <url><loc>https://keras3.posit.co/dev/reference/config_set_epsilon.html</loc></url>
@@ -294,6 +297,7 @@
 <url><loc>https://keras3.posit.co/dev/reference/layer_random_color_jitter.html</loc></url>
 <url><loc>https://keras3.posit.co/dev/reference/layer_random_contrast.html</loc></url>
 <url><loc>https://keras3.posit.co/dev/reference/layer_random_crop.html</loc></url>
+<url><loc>https://keras3.posit.co/dev/reference/layer_random_elastic_transform.html</loc></url>
 <url><loc>https://keras3.posit.co/dev/reference/layer_random_erasing.html</loc></url>
 <url><loc>https://keras3.posit.co/dev/reference/layer_random_flip.html</loc></url>
 <url><loc>https://keras3.posit.co/dev/reference/layer_random_gaussian_blur.html</loc></url>
@@ -348,6 +352,7 @@
 <url><loc>https://keras3.posit.co/dev/reference/loss_binary_focal_crossentropy.html</loc></url>
 <url><loc>https://keras3.posit.co/dev/reference/loss_categorical_crossentropy.html</loc></url>
 <url><loc>https://keras3.posit.co/dev/reference/loss_categorical_focal_crossentropy.html</loc></url>
+<url><loc>https://keras3.posit.co/dev/reference/loss_categorical_generalized_cross_entropy.html</loc></url>
 <url><loc>https://keras3.posit.co/dev/reference/loss_categorical_hinge.html</loc></url>
 <url><loc>https://keras3.posit.co/dev/reference/loss_circle.html</loc></url>
 <url><loc>https://keras3.posit.co/dev/reference/loss_cosine_similarity.html</loc></url>
@@ -426,6 +431,7 @@
 <url><loc>https://keras3.posit.co/dev/reference/op_abs.html</loc></url>
 <url><loc>https://keras3.posit.co/dev/reference/op_add.html</loc></url>
 <url><loc>https://keras3.posit.co/dev/reference/op_all.html</loc></url>
+<url><loc>https://keras3.posit.co/dev/reference/op_angle.html</loc></url>
 <url><loc>https://keras3.posit.co/dev/reference/op_any.html</loc></url>
 <url><loc>https://keras3.posit.co/dev/reference/op_append.html</loc></url>
 <url><loc>https://keras3.posit.co/dev/reference/op_arange.html</loc></url>
@@ -444,6 +450,7 @@
 <url><loc>https://keras3.posit.co/dev/reference/op_associative_scan.html</loc></url>
 <url><loc>https://keras3.posit.co/dev/reference/op_average.html</loc></url>
 <url><loc>https://keras3.posit.co/dev/reference/op_average_pool.html</loc></url>
+<url><loc>https://keras3.posit.co/dev/reference/op_bartlett.html</loc></url>
 <url><loc>https://keras3.posit.co/dev/reference/op_batch_normalization.html</loc></url>
 <url><loc>https://keras3.posit.co/dev/reference/op_binary_crossentropy.html</loc></url>
 <url><loc>https://keras3.posit.co/dev/reference/op_bincount.html</loc></url>
@@ -454,9 +461,11 @@
 <url><loc>https://keras3.posit.co/dev/reference/op_bitwise_or.html</loc></url>
 <url><loc>https://keras3.posit.co/dev/reference/op_bitwise_right_shift.html</loc></url>
 <url><loc>https://keras3.posit.co/dev/reference/op_bitwise_xor.html</loc></url>
+<url><loc>https://keras3.posit.co/dev/reference/op_blackman.html</loc></url>
 <url><loc>https://keras3.posit.co/dev/reference/op_broadcast_to.html</loc></url>
 <url><loc>https://keras3.posit.co/dev/reference/op_cast.html</loc></url>
 <url><loc>https://keras3.posit.co/dev/reference/op_categorical_crossentropy.html</loc></url>
+<url><loc>https://keras3.posit.co/dev/reference/op_cbrt.html</loc></url>
 <url><loc>https://keras3.posit.co/dev/reference/op_ceil.html</loc></url>
 <url><loc>https://keras3.posit.co/dev/reference/op_celu.html</loc></url>
 <url><loc>https://keras3.posit.co/dev/reference/op_cholesky.html</loc></url>
@@ -469,6 +478,7 @@
 <url><loc>https://keras3.posit.co/dev/reference/op_convert_to_numpy.html</loc></url>
 <url><loc>https://keras3.posit.co/dev/reference/op_convert_to_tensor.html</loc></url>
 <url><loc>https://keras3.posit.co/dev/reference/op_copy.html</loc></url>
+<url><loc>https://keras3.posit.co/dev/reference/op_corrcoef.html</loc></url>
 <url><loc>https://keras3.posit.co/dev/reference/op_correlate.html</loc></url>
 <url><loc>https://keras3.posit.co/dev/reference/op_cos.html</loc></url>
 <url><loc>https://keras3.posit.co/dev/reference/op_cosh.html</loc></url>
@@ -479,6 +489,7 @@
 <url><loc>https://keras3.posit.co/dev/reference/op_cumprod.html</loc></url>
 <url><loc>https://keras3.posit.co/dev/reference/op_cumsum.html</loc></url>
 <url><loc>https://keras3.posit.co/dev/reference/op_custom_gradient.html</loc></url>
+<url><loc>https://keras3.posit.co/dev/reference/op_deg2rad.html</loc></url>
 <url><loc>https://keras3.posit.co/dev/reference/op_depthwise_conv.html</loc></url>
 <url><loc>https://keras3.posit.co/dev/reference/op_det.html</loc></url>
 <url><loc>https://keras3.posit.co/dev/reference/op_diag.html</loc></url>
@@ -518,10 +529,13 @@
 <url><loc>https://keras3.posit.co/dev/reference/op_glu.html</loc></url>
 <url><loc>https://keras3.posit.co/dev/reference/op_greater.html</loc></url>
 <url><loc>https://keras3.posit.co/dev/reference/op_greater_equal.html</loc></url>
+<url><loc>https://keras3.posit.co/dev/reference/op_hamming.html</loc></url>
+<url><loc>https://keras3.posit.co/dev/reference/op_hanning.html</loc></url>
 <url><loc>https://keras3.posit.co/dev/reference/op_hard_shrink.html</loc></url>
 <url><loc>https://keras3.posit.co/dev/reference/op_hard_sigmoid.html</loc></url>
 <url><loc>https://keras3.posit.co/dev/reference/op_hard_silu.html</loc></url>
 <url><loc>https://keras3.posit.co/dev/reference/op_hard_tanh.html</loc></url>
+<url><loc>https://keras3.posit.co/dev/reference/op_heaviside.html</loc></url>
 <url><loc>https://keras3.posit.co/dev/reference/op_histogram.html</loc></url>
 <url><loc>https://keras3.posit.co/dev/reference/op_hstack.html</loc></url>
 <url><loc>https://keras3.posit.co/dev/reference/op_identity.html</loc></url>
@@ -529,6 +543,7 @@
 <url><loc>https://keras3.posit.co/dev/reference/op_imag.html</loc></url>
 <url><loc>https://keras3.posit.co/dev/reference/op_image_affine_transform.html</loc></url>
 <url><loc>https://keras3.posit.co/dev/reference/op_image_crop.html</loc></url>
+<url><loc>https://keras3.posit.co/dev/reference/op_image_elastic_transform.html</loc></url>
 <url><loc>https://keras3.posit.co/dev/reference/op_image_extract_patches.html</loc></url>
 <url><loc>https://keras3.posit.co/dev/reference/op_image_gaussian_blur.html</loc></url>
 <url><loc>https://keras3.posit.co/dev/reference/op_image_hsv_to_rgb.html</loc></url>
@@ -548,6 +563,8 @@
 <url><loc>https://keras3.posit.co/dev/reference/op_isinf.html</loc></url>
 <url><loc>https://keras3.posit.co/dev/reference/op_isnan.html</loc></url>
 <url><loc>https://keras3.posit.co/dev/reference/op_istft.html</loc></url>
+<url><loc>https://keras3.posit.co/dev/reference/op_kaiser.html</loc></url>
+<url><loc>https://keras3.posit.co/dev/reference/op_layer_normalization.html</loc></url>
 <url><loc>https://keras3.posit.co/dev/reference/op_leaky_relu.html</loc></url>
 <url><loc>https://keras3.posit.co/dev/reference/op_left_shift.html</loc></url>
 <url><loc>https://keras3.posit.co/dev/reference/op_less.html</loc></url>
@@ -647,6 +664,7 @@
 <url><loc>https://keras3.posit.co/dev/reference/op_sort.html</loc></url>
 <url><loc>https://keras3.posit.co/dev/reference/op_sparse_categorical_crossentropy.html</loc></url>
 <url><loc>https://keras3.posit.co/dev/reference/op_sparse_plus.html</loc></url>
+<url><loc>https://keras3.posit.co/dev/reference/op_sparse_sigmoid.html</loc></url>
 <url><loc>https://keras3.posit.co/dev/reference/op_sparsemax.html</loc></url>
 <url><loc>https://keras3.posit.co/dev/reference/op_split.html</loc></url>
 <url><loc>https://keras3.posit.co/dev/reference/op_sqrt.html</loc></url>
@@ -684,6 +702,8 @@
 <url><loc>https://keras3.posit.co/dev/reference/op_vdot.html</loc></url>
 <url><loc>https://keras3.posit.co/dev/reference/op_vectorize.html</loc></url>
 <url><loc>https://keras3.posit.co/dev/reference/op_vectorized_map.html</loc></url>
+<url><loc>https://keras3.posit.co/dev/reference/op_view_as_complex.html</loc></url>
+<url><loc>https://keras3.posit.co/dev/reference/op_view_as_real.html</loc></url>
 <url><loc>https://keras3.posit.co/dev/reference/op_vstack.html</loc></url>
 <url><loc>https://keras3.posit.co/dev/reference/op_where.html</loc></url>
 <url><loc>https://keras3.posit.co/dev/reference/op_while_loop.html</loc></url>
@@ -699,6 +719,7 @@
 <url><loc>https://keras3.posit.co/dev/reference/optimizer_lamb.html</loc></url>
 <url><loc>https://keras3.posit.co/dev/reference/optimizer_lion.html</loc></url>
 <url><loc>https://keras3.posit.co/dev/reference/optimizer_loss_scale.html</loc></url>
+<url><loc>https://keras3.posit.co/dev/reference/optimizer_muon.html</loc></url>
 <url><loc>https://keras3.posit.co/dev/reference/optimizer_nadam.html</loc></url>
 <url><loc>https://keras3.posit.co/dev/reference/optimizer_rmsprop.html</loc></url>
 <url><loc>https://keras3.posit.co/dev/reference/optimizer_sgd.html</loc></url>
diff --git a/tools/make-website.R b/tools/make-website.R
index 93db16611..c435607fc 100755
--- a/tools/make-website.R
+++ b/tools/make-website.R
@@ -1,7 +1,7 @@
 #!/usr/bin/env Rscript
 
 
-pkgdown::build_site()
+pkgdown::build_site(devel = FALSE)
 
 
 envir::import_from(magrittr, `%<>%`)

From 8aeb96af6e5951b3156ce244ce4c89d4280bc15e Mon Sep 17 00:00:00 2001
From: Tomasz Kalinowski <kalinowskit@gmail.com>
Date: Sat, 27 Sep 2025 14:26:49 -0400
Subject: [PATCH 78/81] R CMD check fixes

---
 R/layers-convolutional.R | 16 +++++++----
 R/optimizers.R           | 57 +++++++++++++++++++++++++++++++++++++++-
 2 files changed, 67 insertions(+), 6 deletions(-)

diff --git a/R/layers-convolutional.R b/R/layers-convolutional.R
index f419e51f0..e83eab6a1 100644
--- a/R/layers-convolutional.R
+++ b/R/layers-convolutional.R
@@ -348,11 +348,6 @@ function (object, filters, kernel_size, strides = 1L, padding = "valid",
 #' the left/right or up/down of the input. When `padding="same"` and
 #' `strides=1`, the output has the same size as the input.
 #'
-#' @param output_padding
-#' Scalar integer or vector of two integers. Amount of padding to add to the output
-#' depth, height, and width. Each element must be smaller than the corresponding
-#' stride. When `NULL` (default) the output size is inferred.
-#'
 #' @param data_format
 #' string, either `"channels_last"` or `"channels_first"`.
 #' The ordering of the dimensions in the inputs. `"channels_last"`
@@ -500,6 +495,11 @@ function (object, filters, kernel_size, strides = list(1L, 1L),
 #' the left/right or up/down of the input. When `padding="same"` and
 #' `strides=1`, the output has the same size as the input.
 #'
+#' @param output_padding
+#' Scalar integer or vector of two integers. Amount of padding to add to the
+#' height and width of the output tensor. Each element must be smaller than the
+#' corresponding stride. When `NULL` (default) the output size is inferred.
+#'
 #' @param data_format
 #' string, either `"channels_last"` or `"channels_first"`.
 #' The ordering of the dimensions in the inputs. `"channels_last"`
@@ -791,6 +791,12 @@ function (object, filters, kernel_size, strides = list(1L, 1L,
 #' the left/right or up/down of the input. When `padding="same"` and
 #' `strides=1`, the output has the same size as the input.
 #'
+#' @param output_padding
+#' Scalar integer or vector of three integers. Amount of padding to add to the
+#' depth, height, and width of the output tensor. Each element must be smaller
+#' than the corresponding stride. When `NULL` (default) the output size is
+#' inferred.
+#'
 #' @param data_format
 #' string, either `"channels_last"` or `"channels_first"`.
 #' The ordering of the dimensions in the inputs. `"channels_last"`
diff --git a/R/optimizers.R b/R/optimizers.R
index a3ddccb16..d3e48ea0f 100644
--- a/R/optimizers.R
+++ b/R/optimizers.R
@@ -1277,7 +1277,7 @@ function (inner_optimizer, initial_scale = 32768, dynamic_growth_steps = 2000L,
 #'
 #' 1. Embedding layer
 #' 2. Final output fully connected layer
-#' 3. Any {0,1}-D variables
+#' 3. Any 0- or 1-D variables
 #'
 #' These should all be optimized using AdamW.
 #'
@@ -1326,6 +1326,61 @@ function (inner_optimizer, initial_scale = 32768, dynamic_growth_steps = 2000L,
 #' not the epsilon in Algorithm 1 of the paper.
 #' It is used as in AdamW. Defaults to `1e-7`.
 #'
+#' @param weight_decay
+#' Float. If set, weight decay is applied.
+#'
+#' @param clipnorm
+#' Float. If set, the gradient of each weight is individually
+#' clipped so that its norm is no higher than this value.
+#'
+#' @param clipvalue
+#' Float. If set, the gradient of each weight is clipped to be
+#' no higher than this value.
+#'
+#' @param global_clipnorm
+#' Float. If set, the gradient of all weights is clipped
+#' so that their global norm is no higher than this value.
+#'
+#' @param use_ema
+#' Boolean, defaults to `FALSE`.
+#' If `TRUE`, exponential moving average (EMA) is applied. EMA consists of
+#' computing an exponential moving average of the weights of the model (as
+#' the weight values change after each training batch), and periodically
+#' overwriting the weights with their moving average.
+#'
+#' @param ema_momentum
+#' Float, defaults to `0.99`. Only used if `use_ema = TRUE`.
+#' This is the momentum to use when computing the EMA of the model's
+#' weights: `new_average = ema_momentum * old_average + (1 - ema_momentum) *
+#' current_variable_value`.
+#'
+#' @param ema_overwrite_frequency
+#' Int or `NULL`, defaults to `NULL`. Only used if `use_ema = TRUE`. Every
+#' `ema_overwrite_frequency` steps of iterations, we overwrite the model
+#' variable by its moving average. If `NULL`, the optimizer does not overwrite
+#' model variables in the middle of training, and you need to explicitly
+#' overwrite the variables at the end of training by calling
+#' `optimizer$finalize_variable_values()` (which updates the model variables
+#' in-place). When using the built-in `fit()` training loop, this happens
+#' automatically after the last epoch, and you don't need to do anything.
+#'
+#' @param loss_scale_factor
+#' Float or `NULL`. If a float, the scale factor will be multiplied the loss
+#' before computing gradients, and the inverse of the scale factor will be
+#' multiplied by the gradients before updating variables. Useful for
+#' preventing underflow during mixed precision training. Alternately,
+#' [`optimizer_loss_scale()`] will automatically set a loss scale factor.
+#'
+#' @param gradient_accumulation_steps
+#' Int or `NULL`. If an int, model and optimizer variables will not be updated
+#' at every step; instead they will be updated every `gradient_accumulation_steps`
+#' steps, using the average value of the gradients since the last update. This
+#' is known as "gradient accumulation". This can be useful when your batch size is
+#' very small, in order to reduce gradient noise at each update step. EMA
+#' frequency will look at "accumulated" iterations value (optimizer steps //
+#' gradient_accumulation_steps). Learning rate schedules will look at "real"
+#' iterations value (optimizer steps).
+#'
 #' @param exclude_layers
 #' List of strings, keywords of layer names to exclude.
 #' All layers with keywords in their path will use AdamW.

From 37fa8fd77a6de3ba9a258d78c763d27f6945b453 Mon Sep 17 00:00:00 2001
From: Tomasz Kalinowski <kalinowskit@gmail.com>
Date: Sat, 27 Sep 2025 17:01:40 -0400
Subject: [PATCH 79/81] redocument

---
 man/layer_conv_2d.Rd           |  4 ---
 man/layer_conv_2d_transpose.Rd |  4 +++
 man/layer_conv_3d_transpose.Rd |  5 ++++
 man/layer_tfsm.Rd              |  4 +--
 man/optimizer_muon.Rd          | 47 +++++++++++++++++++++++++++++++++-
 5 files changed, 57 insertions(+), 7 deletions(-)

diff --git a/man/layer_conv_2d.Rd b/man/layer_conv_2d.Rd
index 8e4e39d1c..09ad7e2f5 100644
--- a/man/layer_conv_2d.Rd
+++ b/man/layer_conv_2d.Rd
@@ -89,10 +89,6 @@ are not safe to use when doing asynchronous distributed training.}
 bias after being updated by an \code{Optimizer}.}
 
 \item{...}{For forward/backward compatability.}
-
-\item{output_padding}{Scalar integer or vector of two integers. Amount of padding to add to the output
-depth, height, and width. Each element must be smaller than the corresponding
-stride. When \code{NULL} (default) the output size is inferred.}
 }
 \value{
 A 4D tensor representing \code{activation(conv2d(inputs, kernel) + bias)}.
diff --git a/man/layer_conv_2d_transpose.Rd b/man/layer_conv_2d_transpose.Rd
index f0663933b..48146077d 100644
--- a/man/layer_conv_2d_transpose.Rd
+++ b/man/layer_conv_2d_transpose.Rd
@@ -43,6 +43,10 @@ of the transposed convolution. \code{strides > 1} is incompatible with
 the left/right or up/down of the input. When \code{padding="same"} and
 \code{strides=1}, the output has the same size as the input.}
 
+\item{output_padding}{Scalar integer or vector of two integers. Amount of padding to add to the
+height and width of the output tensor. Each element must be smaller than the
+corresponding stride. When \code{NULL} (default) the output size is inferred.}
+
 \item{data_format}{string, either \code{"channels_last"} or \code{"channels_first"}.
 The ordering of the dimensions in the inputs. \code{"channels_last"}
 corresponds to inputs with shape
diff --git a/man/layer_conv_3d_transpose.Rd b/man/layer_conv_3d_transpose.Rd
index 79e3cf54f..47aa451f7 100644
--- a/man/layer_conv_3d_transpose.Rd
+++ b/man/layer_conv_3d_transpose.Rd
@@ -43,6 +43,11 @@ of the transposed convolution. \code{strides > 1} is incompatible with
 the left/right or up/down of the input. When \code{padding="same"} and
 \code{strides=1}, the output has the same size as the input.}
 
+\item{output_padding}{Scalar integer or vector of three integers. Amount of padding to add to the
+depth, height, and width of the output tensor. Each element must be smaller
+than the corresponding stride. When \code{NULL} (default) the output size is
+inferred.}
+
 \item{data_format}{string, either \code{"channels_last"} or \code{"channels_first"}.
 The ordering of the dimensions in the inputs. \code{"channels_last"}
 corresponds to inputs with shape
diff --git a/man/layer_tfsm.Rd b/man/layer_tfsm.Rd
index 058847963..e195f7ca4 100644
--- a/man/layer_tfsm.Rd
+++ b/man/layer_tfsm.Rd
@@ -59,8 +59,8 @@ model |> export_savedmodel("path/to/artifact")
 ## Output Type:
 ##   TensorSpec(shape=(None, 10), dtype=tf.float32, name=None)
 ## Captures:
-##   131671756274576: TensorSpec(shape=(), dtype=tf.resource, name=None)
-##   131671756265936: TensorSpec(shape=(), dtype=tf.resource, name=None)
+##   131539756464656: TensorSpec(shape=(), dtype=tf.resource, name=None)
+##   131539756464464: TensorSpec(shape=(), dtype=tf.resource, name=None)
 
 }\if{html}{\out{</div>}}
 
diff --git a/man/optimizer_muon.Rd b/man/optimizer_muon.Rd
index 31fcab94f..9005597d6 100644
--- a/man/optimizer_muon.Rd
+++ b/man/optimizer_muon.Rd
@@ -53,6 +53,51 @@ The exponential decay rate for the 2nd moment estimates. Defaults to
 not the epsilon in Algorithm 1 of the paper.
 It is used as in AdamW. Defaults to \code{1e-7}.}
 
+\item{weight_decay}{Float. If set, weight decay is applied.}
+
+\item{clipnorm}{Float. If set, the gradient of each weight is individually
+clipped so that its norm is no higher than this value.}
+
+\item{clipvalue}{Float. If set, the gradient of each weight is clipped to be
+no higher than this value.}
+
+\item{global_clipnorm}{Float. If set, the gradient of all weights is clipped
+so that their global norm is no higher than this value.}
+
+\item{use_ema}{Boolean, defaults to \code{FALSE}.
+If \code{TRUE}, exponential moving average (EMA) is applied. EMA consists of
+computing an exponential moving average of the weights of the model (as
+the weight values change after each training batch), and periodically
+overwriting the weights with their moving average.}
+
+\item{ema_momentum}{Float, defaults to \code{0.99}. Only used if \code{use_ema = TRUE}.
+This is the momentum to use when computing the EMA of the model's
+weights: \code{new_average = ema_momentum * old_average + (1 - ema_momentum) * current_variable_value}.}
+
+\item{ema_overwrite_frequency}{Int or \code{NULL}, defaults to \code{NULL}. Only used if \code{use_ema = TRUE}. Every
+\code{ema_overwrite_frequency} steps of iterations, we overwrite the model
+variable by its moving average. If \code{NULL}, the optimizer does not overwrite
+model variables in the middle of training, and you need to explicitly
+overwrite the variables at the end of training by calling
+\code{optimizer$finalize_variable_values()} (which updates the model variables
+in-place). When using the built-in \code{fit()} training loop, this happens
+automatically after the last epoch, and you don't need to do anything.}
+
+\item{loss_scale_factor}{Float or \code{NULL}. If a float, the scale factor will be multiplied the loss
+before computing gradients, and the inverse of the scale factor will be
+multiplied by the gradients before updating variables. Useful for
+preventing underflow during mixed precision training. Alternately,
+\code{\link[=optimizer_loss_scale]{optimizer_loss_scale()}} will automatically set a loss scale factor.}
+
+\item{gradient_accumulation_steps}{Int or \code{NULL}. If an int, model and optimizer variables will not be updated
+at every step; instead they will be updated every \code{gradient_accumulation_steps}
+steps, using the average value of the gradients since the last update. This
+is known as "gradient accumulation". This can be useful when your batch size is
+very small, in order to reduce gradient noise at each update step. EMA
+frequency will look at "accumulated" iterations value (optimizer steps //
+gradient_accumulation_steps). Learning rate schedules will look at "real"
+iterations value (optimizer steps).}
+
 \item{name}{String, name for the object}
 
 \item{exclude_layers}{List of strings, keywords of layer names to exclude.
@@ -90,7 +135,7 @@ Note that this optimizer should not be used in the following layers:
 \enumerate{
 \item Embedding layer
 \item Final output fully connected layer
-\item Any {0,1}-D variables
+\item Any 0- or 1-D variables
 }
 
 These should all be optimized using AdamW.

From 8dab59b34a2ca3c7a3cc90b863b5b9a9ae315f7e Mon Sep 17 00:00:00 2001
From: Tomasz Kalinowski <kalinowskit@gmail.com>
Date: Sun, 28 Sep 2025 13:53:48 -0400
Subject: [PATCH 80/81] roxygen warnings fixes

---
 R/metrics.R                             |  7 ++---
 R/ops-numpy.R                           | 34 +++++++++++--------------
 R/ops.R                                 |  3 +--
 R/py-classes.R                          |  6 ++---
 man/grapes-py_class-grapes.Rd           |  2 +-
 man/grapes-set-active-grapes.Rd         |  2 +-
 man/layer_tfsm.Rd                       |  4 +--
 man/metric_binary_focal_crossentropy.Rd | 22 ++++++++++++----
 man/op_blackman.Rd                      | 15 -----------
 man/op_cbrt.Rd                          | 18 +++++++------
 man/op_corrcoef.Rd                      | 20 ++++++++-------
 man/op_searchsorted.Rd                  | 14 ++++++++--
 man/roxygen/meta.R                      |  2 +-
 tools/make-man.R                        | 21 +++++++++++----
 14 files changed, 94 insertions(+), 76 deletions(-)

diff --git a/R/metrics.R b/R/metrics.R
index 1697ed363..42405adbc 100644
--- a/R/metrics.R
+++ b/R/metrics.R
@@ -27,11 +27,12 @@
 #' focal_loss
 #'
 #' # Compare with binary crossentropy.
-#' # Binary focal crossentropy emphasises harder examples, yielding a larger
+#' # Binary focal crossentropy emphasizes harder examples, yielding a larger
 #' # relative loss where the model struggles.
 #' bce_loss <- loss_binary_crossentropy(y_true, y_pred)
-#' cbind(focal_loss, bce_loss,
-#'       ratio = focal_loss / bce_loss)
+#' cbind(focal_loss = as.array(focal_loss),
+#'       bce_loss = as.array(bce_loss),
+#'       ratio = as.array(focal_loss / bce_loss))
 #' ```
 #'
 #' @returns
diff --git a/R/ops-numpy.R b/R/ops-numpy.R
index bcc5d581c..c326e866a 100644
--- a/R/ops-numpy.R
+++ b/R/ops-numpy.R
@@ -73,12 +73,6 @@ function (x, kth, axis = -1L, zero_indexed = FALSE)
 
 #' Compute the Pearson correlation coefficient matrix.
 #'
-#' @returns
-#' A tensor of shape `(N, N)` representing the correlation matrix.
-#'
-#' @param x
-#' A 2D tensor of shape `(N, D)`, where `N` is the number of variables
-#' and `D` is the number of observations.
 #'
 #' # Examples
 #' ```{r}
@@ -87,6 +81,13 @@ function (x, kth, axis = -1L, zero_indexed = FALSE)
 #' op_corrcoef(x)
 #' ```
 #'
+#' @param x
+#' A 2D tensor of shape `(N, D)`, where `N` is the number of variables
+#' and `D` is the number of observations.
+#'
+#' @returns
+#' A tensor of shape `(N, N)` representing the correlation matrix.
+#'
 #' @export
 #' @family numpy ops
 #' @family ops
@@ -102,17 +103,17 @@ ops$corrcoef(x)
 #' Returns the real-valued cube root of `x`, handling negative inputs in the
 #' real domain.
 #'
-#' @returns
-#' A tensor containing the cube root of each element in `x`.
-#'
-#' @param x
-#' Input tensor.
-#'
 #' # Examples
 #' ```{r}
 #' op_cbrt(c(-8, 0, 8))
 #' ```
 #'
+#' @param x
+#' Input tensor.
+#'
+#' @returns
+#' A tensor containing the cube root of each element in `x`.
+#'
 #' @export
 #' @family numpy ops
 #' @family ops
@@ -178,17 +179,12 @@ ops$bartlett(as_integer(x))
 #' @description
 #' The Blackman window is a taper formed by using a weighted cosine.
 #'
-#' # Examples
-#' ```{r}
-#' op_blackman(5)
-#' ```
+#' @param x
+#' Length of the window. Must be a positive integer.
 #'
 #' @returns
 #' A 1D tensor containing the window values.
 #'
-#' @param x
-#' Length of the window. Must be a positive integer.
-#'
 #' @export
 #' @family numpy ops
 #' @family ops
diff --git a/R/ops.R b/R/ops.R
index 1b459c836..3365c6f7a 100644
--- a/R/ops.R
+++ b/R/ops.R
@@ -398,8 +398,7 @@ function (inputs, indices, updates)
 #' @tether keras.ops.searchsorted
 op_searchsorted <-
 function (sorted_sequence, values, side = "left", zero_indexed = FALSE) {
-
-  result <- ops$searchsorted(as_array(sorted_sequence), as_array(values), side)
+  result <- ops$searchsorted(as_py_array(sorted_sequence), as_py_array(values), side)
   if (zero_indexed) result else result + 1L
 }
 
diff --git a/R/py-classes.R b/R/py-classes.R
index d55a4dd93..27680f036 100644
--- a/R/py-classes.R
+++ b/R/py-classes.R
@@ -493,7 +493,7 @@ custom_fn <- function(name, fn) {
 #' @export
 #' @aliases py_class
 #' @keywords internal
-#' @seealso [`%<-active%()`]
+#' @seealso \code{\link{\%<-active\%}}
 #'
 #' @examples
 #' \dontrun{
@@ -754,10 +754,10 @@ function(x) {
 #' @returns `value`, invisibly
 #' @export
 #'
-#' @details Active bindings defined in a [`%py_class%`] are converted to
+#' @details Active bindings defined in a \code{\link{\%py_class\%}} are converted to
 #'   `@property` decorated methods.
 #'
-#' @seealso [`makeActiveBinding()`]
+#' @seealso [`makeActiveBinding()`], \code{\link{\%py_class\%}}
 #' @keywords internal
 #'
 #' @examples
diff --git a/man/grapes-py_class-grapes.Rd b/man/grapes-py_class-grapes.Rd
index 7c1af55c7..189dfe803 100644
--- a/man/grapes-py_class-grapes.Rd
+++ b/man/grapes-py_class-grapes.Rd
@@ -98,6 +98,6 @@ inst1$an_active_property <- 11
 }
 }
 \seealso{
-\code{\link[=\%<-active\%]{\%<-active\%()}}
+\code{\link{\%<-active\%}}
 }
 \keyword{internal}
diff --git a/man/grapes-set-active-grapes.Rd b/man/grapes-set-active-grapes.Rd
index 242aa3038..3bb6b0471 100644
--- a/man/grapes-set-active-grapes.Rd
+++ b/man/grapes-set-active-grapes.Rd
@@ -38,6 +38,6 @@ x
 rm(x) # cleanup
 }
 \seealso{
-\code{\link[=makeActiveBinding]{makeActiveBinding()}}
+\code{\link[=makeActiveBinding]{makeActiveBinding()}}, \code{\link{\%py_class\%}}
 }
 \keyword{internal}
diff --git a/man/layer_tfsm.Rd b/man/layer_tfsm.Rd
index e195f7ca4..2abe0b1e2 100644
--- a/man/layer_tfsm.Rd
+++ b/man/layer_tfsm.Rd
@@ -59,8 +59,8 @@ model |> export_savedmodel("path/to/artifact")
 ## Output Type:
 ##   TensorSpec(shape=(None, 10), dtype=tf.float32, name=None)
 ## Captures:
-##   131539756464656: TensorSpec(shape=(), dtype=tf.resource, name=None)
-##   131539756464464: TensorSpec(shape=(), dtype=tf.resource, name=None)
+##   140648635143312: TensorSpec(shape=(), dtype=tf.resource, name=None)
+##   140648304538064: TensorSpec(shape=(), dtype=tf.resource, name=None)
 
 }\if{html}{\out{</div>}}
 
diff --git a/man/metric_binary_focal_crossentropy.Rd b/man/metric_binary_focal_crossentropy.Rd
index b4a486291..c1958381e 100644
--- a/man/metric_binary_focal_crossentropy.Rd
+++ b/man/metric_binary_focal_crossentropy.Rd
@@ -61,17 +61,29 @@ weight balancing factor for the binary classes 0 and 1 as follows:
 where \code{alpha} is a float in the range of \verb{[0, 1]}.
 }
 \section{Examples}{
-\if{html}{\out{<div class="sourceCode {r}">}}\preformatted{y_true <- rbind(c(0, 1), c(0, 0))
+\if{html}{\out{<div class="sourceCode r">}}\preformatted{y_true <- rbind(c(0, 1), c(0, 0))
 y_pred <- rbind(c(0.6, 0.4), c(0.4, 0.6))
 focal_loss <- loss_binary_focal_crossentropy(y_true, y_pred, gamma = 2)
 focal_loss
+}\if{html}{\out{</div>}}
+
+\if{html}{\out{<div class="sourceCode">}}\preformatted{## tf.Tensor([0.32986468 0.20579839], shape=(2), dtype=float32)
+
+}\if{html}{\out{</div>}}
 
-# Compare with binary crossentropy.
-# Binary focal crossentropy emphasises harder examples, yielding a larger
+\if{html}{\out{<div class="sourceCode r">}}\preformatted{# Compare with binary crossentropy.
+# Binary focal crossentropy emphasizes harder examples, yielding a larger
 # relative loss where the model struggles.
 bce_loss <- loss_binary_crossentropy(y_true, y_pred)
-cbind(focal_loss, bce_loss,
-      ratio = focal_loss / bce_loss)
+cbind(focal_loss = as.array(focal_loss),
+      bce_loss = as.array(bce_loss),
+      ratio = as.array(focal_loss / bce_loss))
+}\if{html}{\out{</div>}}
+
+\if{html}{\out{<div class="sourceCode">}}\preformatted{##      focal_loss  bce_loss     ratio
+## [1,]  0.3298647 0.9162908 0.3600000
+## [2,]  0.2057984 0.7135582 0.2884115
+
 }\if{html}{\out{</div>}}
 }
 
diff --git a/man/op_blackman.Rd b/man/op_blackman.Rd
index c8a253f3f..32af58d6a 100644
--- a/man/op_blackman.Rd
+++ b/man/op_blackman.Rd
@@ -15,21 +15,6 @@ A 1D tensor containing the window values.
 \description{
 The Blackman window is a taper formed by using a weighted cosine.
 }
-\section{Examples}{
-\if{html}{\out{<div class="sourceCode r">}}\preformatted{op_blackman(5)
-}\if{html}{\out{</div>}}
-
-\if{html}{\out{<div class="sourceCode">}}\preformatted{## tf.Tensor(
-## [-1.4901161e-08  3.4000003e-01  9.9999994e-01  3.3999997e-01
-##  -1.4901161e-08], shape=(5), dtype=float32)
-
-}\if{html}{\out{</div>}}
-
-[-1.4901161e-08  3.4000003e-01  9.9999994e-01  3.3999997e-01
-\subsection{-1.4901161e-08]: R:-1.4901161e-08\%20\%203.4000003e-01\%20\%209.9999994e-01\%20\%203.3999997e-01\%0A##\%20\%20-1.4901161e-08}{
-}
-}
-
 \seealso{
 Other numpy ops: \cr
 \code{\link{op_abs}()} \cr
diff --git a/man/op_cbrt.Rd b/man/op_cbrt.Rd
index a023a3679..9e38d8210 100644
--- a/man/op_cbrt.Rd
+++ b/man/op_cbrt.Rd
@@ -7,14 +7,7 @@
 op_cbrt(x)
 }
 \arguments{
-\item{x}{Input tensor.Examples
-
-\if{html}{\out{<div class="sourceCode r">}}\preformatted{op_cbrt(c(-8, 0, 8))
-}\if{html}{\out{</div>}}
-
-\if{html}{\out{<div class="sourceCode">}}\preformatted{## tf.Tensor([-2.  0.  2.], shape=(3), dtype=float32)
-
-}\if{html}{\out{</div>}}}
+\item{x}{Input tensor.}
 }
 \value{
 A tensor containing the cube root of each element in \code{x}.
@@ -23,6 +16,15 @@ A tensor containing the cube root of each element in \code{x}.
 Returns the real-valued cube root of \code{x}, handling negative inputs in the
 real domain.
 }
+\section{Examples}{
+\if{html}{\out{<div class="sourceCode r">}}\preformatted{op_cbrt(c(-8, 0, 8))
+}\if{html}{\out{</div>}}
+
+\if{html}{\out{<div class="sourceCode">}}\preformatted{## tf.Tensor([-2.  0.  2.], shape=(3), dtype=float32)
+
+}\if{html}{\out{</div>}}
+}
+
 \seealso{
 Other numpy ops: \cr
 \code{\link{op_abs}()} \cr
diff --git a/man/op_corrcoef.Rd b/man/op_corrcoef.Rd
index f7c7f4326..33992b897 100644
--- a/man/op_corrcoef.Rd
+++ b/man/op_corrcoef.Rd
@@ -8,8 +8,15 @@ op_corrcoef(x)
 }
 \arguments{
 \item{x}{A 2D tensor of shape \verb{(N, D)}, where \code{N} is the number of variables
-and \code{D} is the number of observations.Examples
-
+and \code{D} is the number of observations.}
+}
+\value{
+A tensor of shape \verb{(N, N)} representing the correlation matrix.
+}
+\description{
+Compute the Pearson correlation coefficient matrix.
+}
+\section{Examples}{
 \if{html}{\out{<div class="sourceCode r">}}\preformatted{x <- op_array(matrix(c(1, 2, 3,
                        2, 3, 4), nrow = 2, byrow = TRUE))
 op_corrcoef(x)
@@ -19,14 +26,9 @@ op_corrcoef(x)
 ## [[1. 1.]
 ##  [1. 1.]], shape=(2, 2), dtype=float64)
 
-}\if{html}{\out{</div>}}}
-}
-\value{
-A tensor of shape \verb{(N, N)} representing the correlation matrix.
-}
-\description{
-Compute the Pearson correlation coefficient matrix.
+}\if{html}{\out{</div>}}
 }
+
 \seealso{
 Other numpy ops: \cr
 \code{\link{op_abs}()} \cr
diff --git a/man/op_searchsorted.Rd b/man/op_searchsorted.Rd
index e92933767..719dbc3b7 100644
--- a/man/op_searchsorted.Rd
+++ b/man/op_searchsorted.Rd
@@ -25,10 +25,20 @@ Return insertion indices that keep \code{values} in sorted order when placed
 into \code{sorted_sequence}.
 }
 \section{Examples}{
-\if{html}{\out{<div class="sourceCode {r}">}}\preformatted{sorted <- op_array(c(-1, 0, 2, 4))
+\if{html}{\out{<div class="sourceCode r">}}\preformatted{sorted <- op_array(c(-1, 0, 2, 4))
 queries <- c(-2, 0, 3, 7)
 op_searchsorted(sorted, queries)
-op_searchsorted(sorted, queries, side = "right", zero_indexed = TRUE)
+}\if{html}{\out{</div>}}
+
+\if{html}{\out{<div class="sourceCode">}}\preformatted{## tf.Tensor([1 2 4 5], shape=(4), dtype=int32)
+
+}\if{html}{\out{</div>}}
+
+\if{html}{\out{<div class="sourceCode r">}}\preformatted{op_searchsorted(sorted, queries, side = "right", zero_indexed = TRUE)
+}\if{html}{\out{</div>}}
+
+\if{html}{\out{<div class="sourceCode">}}\preformatted{## tf.Tensor([0 2 3 4], shape=(4), dtype=int32)
+
 }\if{html}{\out{</div>}}
 }
 
diff --git a/man/roxygen/meta.R b/man/roxygen/meta.R
index bbbed92f7..9b0fb06f7 100644
--- a/man/roxygen/meta.R
+++ b/man/roxygen/meta.R
@@ -176,7 +176,7 @@ list(
   knitr_chunk_options = list(
     comment = "##",
     collapse = FALSE,
-    eval = !interactive(), # for faster interactive workflow
+    eval = TRUE, # FALSE for faster interactive workflow
     keras.roxy.post_process_output = TRUE
   )
 )
diff --git a/tools/make-man.R b/tools/make-man.R
index c09fe6077..0e315eb20 100755
--- a/tools/make-man.R
+++ b/tools/make-man.R
@@ -1,12 +1,25 @@
 #!/usr/bin/env Rscript
-Sys.setenv("KERAS_BACKEND" = "tensorflow-cpu")
+writeLines(format(Sys.getpid()), "pidfile")
+Sys.setenv(
+  "KERAS_BACKEND" = "tensorflow-cpu"
+  # "KERAS_BACKEND_CONFIGURED" = "yes"
+)
+# options(warn = 2, error = browser)
+# keras3:::op_blackman(5)
 devtools::document()
 
 # reticulate::py_require(c(
 #   "keras",
 #   if (Sys.info()[["sysname"]] == "Linux") "tensorflow-cpu" else "tensorflow"
 # ))
-if(!"source:tools/utils.R" %in% search()) envir::attach_source("tools/utils.R")
+# if(!"source:tools/utils.R" %in% search()) envir::attach_source("tools/utils.R")
+
+library(stringr)
+envir::import_from(magrittr, `%<>%`)
+envir::import_from(purrr, walk)
+str_prefix <- function(x, prefix, ...) str_c(prefix, x, ...)
+
+
 
 # rd_file <- "man/activation_elu.Rd"
 itemize_family <- function(rd_file) {
@@ -46,6 +59,4 @@ Sys.glob("man/*.Rd") %>%
   walk(cr_family)
   # walk(itemize_family)
 
-
-remotes::install_local(force = TRUE)
-
+system2("Rscript", "-e 'remotes::install_local(force = TRUE)'")

From 8790244d8a64dd338974835561bc572bd7037f2e Mon Sep 17 00:00:00 2001
From: Tomasz Kalinowski <kalinowskit@gmail.com>
Date: Sun, 28 Sep 2025 15:11:46 -0400
Subject: [PATCH 81/81] more roxygen fixes

---
 R/ops-nn.R                                    |  2 +-
 docs/dev/pkgdown.yml                          |  2 +-
 .../dev/reference/grapes-py_class-grapes.html |  2 +-
 .../reference/grapes-set-active-grapes.html   |  2 +-
 docs/dev/reference/layer_conv_2d.html         |  6 --
 .../reference/layer_conv_2d_transpose.html    |  6 ++
 .../reference/layer_conv_3d_transpose.html    |  7 ++
 docs/dev/reference/layer_tfsm.html            |  4 +-
 .../metric_binary_focal_crossentropy.html     | 18 ++++-
 docs/dev/reference/op_blackman.html           | 14 ----
 docs/dev/reference/op_cbrt.html               | 11 +--
 docs/dev/reference/op_corrcoef.html           | 19 ++---
 docs/dev/reference/op_rms_normalization.html  | 25 +------
 docs/dev/reference/op_searchsorted.html       |  9 ++-
 docs/dev/reference/optimizer_muon.html        | 69 ++++++++++++++++++-
 docs/dev/search.json                          |  2 +-
 man/layer_tfsm.Rd                             |  4 +-
 man/op_rms_normalization.Rd                   | 25 +------
 tools/make-man.R                              | 11 +--
 19 files changed, 137 insertions(+), 101 deletions(-)

diff --git a/R/ops-nn.R b/R/ops-nn.R
index 3b9542f2f..936b0ef5f 100644
--- a/R/ops-nn.R
+++ b/R/ops-nn.R
@@ -717,7 +717,7 @@ op_polar <-
 #'
 #' # Examples
 #'
-#' ```{r}
+#' ```{r, eval = FALSE}
 #' x <- random_normal(c(1, 10))
 #' op_rms_normalization(x)
 #' op_rms_normalization(x, scale = 10)
diff --git a/docs/dev/pkgdown.yml b/docs/dev/pkgdown.yml
index 8643bbbd9..cae7a8183 100644
--- a/docs/dev/pkgdown.yml
+++ b/docs/dev/pkgdown.yml
@@ -28,7 +28,7 @@ articles:
   understanding_masking_and_padding: understanding_masking_and_padding.html
   writing_a_custom_training_loop_in_tensorflow: writing_a_custom_training_loop_in_tensorflow.html
   writing_your_own_callbacks: writing_your_own_callbacks.html
-last_built: 2025-09-27T14:52Z
+last_built: 2025-09-28T18:59Z
 urls:
   reference: https://keras3.posit.co/reference
   article: https://keras3.posit.co/articles
diff --git a/docs/dev/reference/grapes-py_class-grapes.html b/docs/dev/reference/grapes-py_class-grapes.html
index 97c19ed5c..e273dc5b3 100644
--- a/docs/dev/reference/grapes-py_class-grapes.html
+++ b/docs/dev/reference/grapes-py_class-grapes.html
@@ -82,7 +82,7 @@ <h2 id="value">Value<a class="anchor" aria-label="anchor" href="#value"></a></h2
     </div>
     <div class="section level2">
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
-    <div class="dont-index"><p><code><a href="grapes-set-active-grapes.html">%&lt;-active%()</a></code></p></div>
+    <div class="dont-index"><p><code>%&lt;-active%</code></p></div>
     </div>
 
     <div class="section level2">
diff --git a/docs/dev/reference/grapes-set-active-grapes.html b/docs/dev/reference/grapes-set-active-grapes.html
index df391ba6c..6c6f26a4b 100644
--- a/docs/dev/reference/grapes-set-active-grapes.html
+++ b/docs/dev/reference/grapes-set-active-grapes.html
@@ -85,7 +85,7 @@ <h2 id="details">Details<a class="anchor" aria-label="anchor" href="#details"></
     </div>
     <div class="section level2">
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
-    <div class="dont-index"><p><code><a href="https://rdrr.io/r/base/bindenv.html" class="external-link">makeActiveBinding()</a></code></p></div>
+    <div class="dont-index"><p><code><a href="https://rdrr.io/r/base/bindenv.html" class="external-link">makeActiveBinding()</a></code>, <code><a href="grapes-py_class-grapes.html">%py_class%</a></code></p></div>
     </div>
 
     <div class="section level2">
diff --git a/docs/dev/reference/layer_conv_2d.html b/docs/dev/reference/layer_conv_2d.html
index ff4140118..57d51368e 100644
--- a/docs/dev/reference/layer_conv_2d.html
+++ b/docs/dev/reference/layer_conv_2d.html
@@ -220,12 +220,6 @@ <h2 id="arguments">Arguments<a class="anchor" aria-label="anchor" href="#argumen
 <dt id="arg--">...<a class="anchor" aria-label="anchor" href="#arg--"></a></dt>
 <dd><p>For forward/backward compatability.</p></dd>
 
-
-<dt id="arg-output-padding">output_padding<a class="anchor" aria-label="anchor" href="#arg-output-padding"></a></dt>
-<dd><p>Scalar integer or vector of two integers. Amount of padding to add to the output
-depth, height, and width. Each element must be smaller than the corresponding
-stride. When <code>NULL</code> (default) the output size is inferred.</p></dd>
-
 </dl></div>
     <div class="section level2">
     <h2 id="value">Value<a class="anchor" aria-label="anchor" href="#value"></a></h2>
diff --git a/docs/dev/reference/layer_conv_2d_transpose.html b/docs/dev/reference/layer_conv_2d_transpose.html
index e3e1c0dc0..1294aa42b 100644
--- a/docs/dev/reference/layer_conv_2d_transpose.html
+++ b/docs/dev/reference/layer_conv_2d_transpose.html
@@ -124,6 +124,12 @@ <h2 id="arguments">Arguments<a class="anchor" aria-label="anchor" href="#argumen
 <code>strides=1</code>, the output has the same size as the input.</p></dd>
 
 
+<dt id="arg-output-padding">output_padding<a class="anchor" aria-label="anchor" href="#arg-output-padding"></a></dt>
+<dd><p>Scalar integer or vector of two integers. Amount of padding to add to the
+height and width of the output tensor. Each element must be smaller than the
+corresponding stride. When <code>NULL</code> (default) the output size is inferred.</p></dd>
+
+
 <dt id="arg-data-format">data_format<a class="anchor" aria-label="anchor" href="#arg-data-format"></a></dt>
 <dd><p>string, either <code>"channels_last"</code> or <code>"channels_first"</code>.
 The ordering of the dimensions in the inputs. <code>"channels_last"</code>
diff --git a/docs/dev/reference/layer_conv_3d_transpose.html b/docs/dev/reference/layer_conv_3d_transpose.html
index 2d865af0a..353680d95 100644
--- a/docs/dev/reference/layer_conv_3d_transpose.html
+++ b/docs/dev/reference/layer_conv_3d_transpose.html
@@ -124,6 +124,13 @@ <h2 id="arguments">Arguments<a class="anchor" aria-label="anchor" href="#argumen
 <code>strides=1</code>, the output has the same size as the input.</p></dd>
 
 
+<dt id="arg-output-padding">output_padding<a class="anchor" aria-label="anchor" href="#arg-output-padding"></a></dt>
+<dd><p>Scalar integer or vector of three integers. Amount of padding to add to the
+depth, height, and width of the output tensor. Each element must be smaller
+than the corresponding stride. When <code>NULL</code> (default) the output size is
+inferred.</p></dd>
+
+
 <dt id="arg-data-format">data_format<a class="anchor" aria-label="anchor" href="#arg-data-format"></a></dt>
 <dd><p>string, either <code>"channels_last"</code> or <code>"channels_first"</code>.
 The ordering of the dimensions in the inputs. <code>"channels_last"</code>
diff --git a/docs/dev/reference/layer_tfsm.html b/docs/dev/reference/layer_tfsm.html
index c89aaf65c..b987309f5 100644
--- a/docs/dev/reference/layer_tfsm.html
+++ b/docs/dev/reference/layer_tfsm.html
@@ -126,8 +126,8 @@ <h2 id="examples">Examples<a class="anchor" aria-label="anchor" href="#examples"
 <span id="cb1-5"><a href="#cb1-5" tabindex="-1"></a><span class="do">## Output Type:</span></span>
 <span id="cb1-6"><a href="#cb1-6" tabindex="-1"></a><span class="do">##   TensorSpec(shape=(None, 10), dtype=tf.float32, name=None)</span></span>
 <span id="cb1-7"><a href="#cb1-7" tabindex="-1"></a><span class="do">## Captures:</span></span>
-<span id="cb1-8"><a href="#cb1-8" tabindex="-1"></a><span class="do">##   131671756274576: TensorSpec(shape=(), dtype=tf.resource, name=None)</span></span>
-<span id="cb1-9"><a href="#cb1-9" tabindex="-1"></a><span class="do">##   131671756265936: TensorSpec(shape=(), dtype=tf.resource, name=None)</span></span>
+<span id="cb1-8"><a href="#cb1-8" tabindex="-1"></a><span class="do">##   133630129211152: TensorSpec(shape=(), dtype=tf.resource, name=None)</span></span>
+<span id="cb1-9"><a href="#cb1-9" tabindex="-1"></a><span class="do">##   133630129217872: TensorSpec(shape=(), dtype=tf.resource, name=None)</span></span>
 <span id="cb1-10"><a href="#cb1-10" tabindex="-1"></a></span></code></pre><p></p></div>
 <p></p><div class="sourceCode r"><pre><code><span><span class="va">reloaded_layer</span> <span class="op">&lt;-</span> <span class="fu"><a href="../reference/layer_tfsm.html">layer_tfsm</a></span><span class="op">(</span>filepath <span class="op">=</span> <span class="st">"path/to/artifact"</span><span class="op">)</span></span>
 <span><span class="va">input</span> <span class="op">&lt;-</span> <span class="fu"><a href="../reference/random_normal.html">random_normal</a></span><span class="op">(</span><span class="fu"><a href="https://rdrr.io/r/base/c.html" class="external-link">c</a></span><span class="op">(</span><span class="fl">2</span>, <span class="fl">784</span><span class="op">)</span><span class="op">)</span></span>
diff --git a/docs/dev/reference/metric_binary_focal_crossentropy.html b/docs/dev/reference/metric_binary_focal_crossentropy.html
index e4a85eb5b..c3252b422 100644
--- a/docs/dev/reference/metric_binary_focal_crossentropy.html
+++ b/docs/dev/reference/metric_binary_focal_crossentropy.html
@@ -154,7 +154,23 @@ <h2 id="value">Value<a class="anchor" aria-label="anchor" href="#value"></a></h2
     </div>
     <div class="section level2">
     <h2 id="examples">Examples<a class="anchor" aria-label="anchor" href="#examples"></a></h2>
-    <p></p><div class="sourceCode {r}"><pre><code></code></pre><p></p></div>
+    <p></p><div class="sourceCode r"><pre><code><span><span class="va">y_true</span> <span class="op">&lt;-</span> <span class="fu"><a href="https://rdrr.io/r/base/cbind.html" class="external-link">rbind</a></span><span class="op">(</span><span class="fu"><a href="https://rdrr.io/r/base/c.html" class="external-link">c</a></span><span class="op">(</span><span class="fl">0</span>, <span class="fl">1</span><span class="op">)</span>, <span class="fu"><a href="https://rdrr.io/r/base/c.html" class="external-link">c</a></span><span class="op">(</span><span class="fl">0</span>, <span class="fl">0</span><span class="op">)</span><span class="op">)</span></span>
+<span><span class="va">y_pred</span> <span class="op">&lt;-</span> <span class="fu"><a href="https://rdrr.io/r/base/cbind.html" class="external-link">rbind</a></span><span class="op">(</span><span class="fu"><a href="https://rdrr.io/r/base/c.html" class="external-link">c</a></span><span class="op">(</span><span class="fl">0.6</span>, <span class="fl">0.4</span><span class="op">)</span>, <span class="fu"><a href="https://rdrr.io/r/base/c.html" class="external-link">c</a></span><span class="op">(</span><span class="fl">0.4</span>, <span class="fl">0.6</span><span class="op">)</span><span class="op">)</span></span>
+<span><span class="va">focal_loss</span> <span class="op">&lt;-</span> <span class="fu"><a href="../reference/loss_binary_focal_crossentropy.html">loss_binary_focal_crossentropy</a></span><span class="op">(</span><span class="va">y_true</span>, <span class="va">y_pred</span>, gamma <span class="op">=</span> <span class="fl">2</span><span class="op">)</span></span>
+<span><span class="va">focal_loss</span></span></code></pre><p></p></div>
+<p></p><div class="sourceCode"><pre><code><span id="cb1-1"><a href="#cb1-1" tabindex="-1"></a><span class="do">## tf.Tensor([0.32986468 0.20579839], shape=(2), dtype=float32)</span></span>
+<span id="cb1-2"><a href="#cb1-2" tabindex="-1"></a></span></code></pre><p></p></div>
+<p></p><div class="sourceCode r"><pre><code><span><span class="co"># Compare with binary crossentropy.</span></span>
+<span><span class="co"># Binary focal crossentropy emphasizes harder examples, yielding a larger</span></span>
+<span><span class="co"># relative loss where the model struggles.</span></span>
+<span><span class="va">bce_loss</span> <span class="op">&lt;-</span> <span class="fu"><a href="../reference/loss_binary_crossentropy.html">loss_binary_crossentropy</a></span><span class="op">(</span><span class="va">y_true</span>, <span class="va">y_pred</span><span class="op">)</span></span>
+<span><span class="fu"><a href="https://rdrr.io/r/base/cbind.html" class="external-link">cbind</a></span><span class="op">(</span>focal_loss <span class="op">=</span> <span class="fu"><a href="https://rdrr.io/r/base/array.html" class="external-link">as.array</a></span><span class="op">(</span><span class="va">focal_loss</span><span class="op">)</span>,</span>
+<span>      bce_loss <span class="op">=</span> <span class="fu"><a href="https://rdrr.io/r/base/array.html" class="external-link">as.array</a></span><span class="op">(</span><span class="va">bce_loss</span><span class="op">)</span>,</span>
+<span>      ratio <span class="op">=</span> <span class="fu"><a href="https://rdrr.io/r/base/array.html" class="external-link">as.array</a></span><span class="op">(</span><span class="va">focal_loss</span> <span class="op">/</span> <span class="va">bce_loss</span><span class="op">)</span><span class="op">)</span></span></code></pre><p></p></div>
+<p></p><div class="sourceCode"><pre><code><span id="cb1-1"><a href="#cb1-1" tabindex="-1"></a><span class="do">##      focal_loss  bce_loss     ratio</span></span>
+<span id="cb1-2"><a href="#cb1-2" tabindex="-1"></a><span class="do">## [1,]  0.3298647 0.9162908 0.3600000</span></span>
+<span id="cb1-3"><a href="#cb1-3" tabindex="-1"></a><span class="do">## [2,]  0.2057984 0.7135582 0.2884115</span></span>
+<span id="cb1-4"><a href="#cb1-4" tabindex="-1"></a></span></code></pre><p></p></div>
     </div>
     <div class="section level2">
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
diff --git a/docs/dev/reference/op_blackman.html b/docs/dev/reference/op_blackman.html
index cac2fc0df..82682e1ed 100644
--- a/docs/dev/reference/op_blackman.html
+++ b/docs/dev/reference/op_blackman.html
@@ -73,20 +73,6 @@ <h2 id="arguments">Arguments<a class="anchor" aria-label="anchor" href="#argumen
     <div class="section level2">
     <h2 id="value">Value<a class="anchor" aria-label="anchor" href="#value"></a></h2>
     <p>A 1D tensor containing the window values.</p>
-    </div>
-    <div class="section level2">
-    <h2 id="examples">Examples<a class="anchor" aria-label="anchor" href="#examples"></a></h2>
-    <p></p><div class="sourceCode r"><pre><code><span><span class="fu"><a href="../reference/op_blackman.html">op_blackman</a></span><span class="op">(</span><span class="fl">5</span><span class="op">)</span></span></code></pre><p></p></div>
-<p></p><div class="sourceCode"><pre><code><span id="cb1-1"><a href="#cb1-1" tabindex="-1"></a><span class="do">## tf.Tensor(</span></span>
-<span id="cb1-2"><a href="#cb1-2" tabindex="-1"></a><span class="do">## [-1.4901161e-08  3.4000003e-01  9.9999994e-01  3.3999997e-01</span></span>
-<span id="cb1-3"><a href="#cb1-3" tabindex="-1"></a><span class="do">##  -1.4901161e-08], shape=(5), dtype=float32)</span></span>
-<span id="cb1-4"><a href="#cb1-4" tabindex="-1"></a></span></code></pre><p></p></div>
-<p>[-1.4901161e-08  3.4000003e-01  9.9999994e-01  3.3999997e-01</p><div class="section">
-<h3 id="-e-r-e-e-e-e-a-e-">-1.4901161e-08]: R:-1.4901161e-08%20%203.4000003e-01%20%209.9999994e-01%20%203.3999997e-01%0A##%20%20-1.4901161e-08<a class="anchor" aria-label="anchor" href="#-e-r-e-e-e-e-a-e-"></a></h3>
-
-
-</div>
-
     </div>
     <div class="section level2">
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
diff --git a/docs/dev/reference/op_cbrt.html b/docs/dev/reference/op_cbrt.html
index b6051b021..e47392e47 100644
--- a/docs/dev/reference/op_cbrt.html
+++ b/docs/dev/reference/op_cbrt.html
@@ -70,15 +70,18 @@ <h2 id="arguments">Arguments<a class="anchor" aria-label="anchor" href="#argumen
 
 
 <dl><dt id="arg-x">x<a class="anchor" aria-label="anchor" href="#arg-x"></a></dt>
-<dd><p>Input tensor.Examples</p>
-<p></p><div class="sourceCode r"><pre><code><span><span class="fu"><a href="../reference/op_cbrt.html">op_cbrt</a></span><span class="op">(</span><span class="fu"><a href="https://rdrr.io/r/base/c.html" class="external-link">c</a></span><span class="op">(</span><span class="op">-</span><span class="fl">8</span>, <span class="fl">0</span>, <span class="fl">8</span><span class="op">)</span><span class="op">)</span></span></code></pre><p></p></div>
-<p></p><div class="sourceCode"><pre><code><span id="cb1-1"><a href="#cb1-1" tabindex="-1"></a><span class="do">## tf.Tensor([-2.  0.  2.], shape=(3), dtype=float32)</span></span>
-<span id="cb1-2"><a href="#cb1-2" tabindex="-1"></a></span></code></pre><p></p></div></dd>
+<dd><p>Input tensor.</p></dd>
 
 </dl></div>
     <div class="section level2">
     <h2 id="value">Value<a class="anchor" aria-label="anchor" href="#value"></a></h2>
     <p>A tensor containing the cube root of each element in <code>x</code>.</p>
+    </div>
+    <div class="section level2">
+    <h2 id="examples">Examples<a class="anchor" aria-label="anchor" href="#examples"></a></h2>
+    <p></p><div class="sourceCode r"><pre><code><span><span class="fu"><a href="../reference/op_cbrt.html">op_cbrt</a></span><span class="op">(</span><span class="fu"><a href="https://rdrr.io/r/base/c.html" class="external-link">c</a></span><span class="op">(</span><span class="op">-</span><span class="fl">8</span>, <span class="fl">0</span>, <span class="fl">8</span><span class="op">)</span><span class="op">)</span></span></code></pre><p></p></div>
+<p></p><div class="sourceCode"><pre><code><span id="cb1-1"><a href="#cb1-1" tabindex="-1"></a><span class="do">## tf.Tensor([-2.  0.  2.], shape=(3), dtype=float32)</span></span>
+<span id="cb1-2"><a href="#cb1-2" tabindex="-1"></a></span></code></pre><p></p></div>
     </div>
     <div class="section level2">
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
diff --git a/docs/dev/reference/op_corrcoef.html b/docs/dev/reference/op_corrcoef.html
index 63aab19bb..e026b4bf4 100644
--- a/docs/dev/reference/op_corrcoef.html
+++ b/docs/dev/reference/op_corrcoef.html
@@ -68,19 +68,22 @@ <h2 id="arguments">Arguments<a class="anchor" aria-label="anchor" href="#argumen
 
 <dl><dt id="arg-x">x<a class="anchor" aria-label="anchor" href="#arg-x"></a></dt>
 <dd><p>A 2D tensor of shape <code>(N, D)</code>, where <code>N</code> is the number of variables
-and <code>D</code> is the number of observations.Examples</p>
-<p></p><div class="sourceCode r"><pre><code><span><span class="va">x</span> <span class="op">&lt;-</span> <span class="fu"><a href="../reference/op_array.html">op_array</a></span><span class="op">(</span><span class="fu"><a href="https://rdrr.io/r/base/matrix.html" class="external-link">matrix</a></span><span class="op">(</span><span class="fu"><a href="https://rdrr.io/r/base/c.html" class="external-link">c</a></span><span class="op">(</span><span class="fl">1</span>, <span class="fl">2</span>, <span class="fl">3</span>,</span>
-<span>                       <span class="fl">2</span>, <span class="fl">3</span>, <span class="fl">4</span><span class="op">)</span>, nrow <span class="op">=</span> <span class="fl">2</span>, byrow <span class="op">=</span> <span class="cn">TRUE</span><span class="op">)</span><span class="op">)</span></span>
-<span><span class="fu"><a href="../reference/op_corrcoef.html">op_corrcoef</a></span><span class="op">(</span><span class="va">x</span><span class="op">)</span></span></code></pre><p></p></div>
-<p></p><div class="sourceCode"><pre><code><span id="cb1-1"><a href="#cb1-1" tabindex="-1"></a><span class="do">## tf.Tensor(</span></span>
-<span id="cb1-2"><a href="#cb1-2" tabindex="-1"></a><span class="do">## [[1. 1.]</span></span>
-<span id="cb1-3"><a href="#cb1-3" tabindex="-1"></a><span class="do">##  [1. 1.]], shape=(2, 2), dtype=float64)</span></span>
-<span id="cb1-4"><a href="#cb1-4" tabindex="-1"></a></span></code></pre><p></p></div></dd>
+and <code>D</code> is the number of observations.</p></dd>
 
 </dl></div>
     <div class="section level2">
     <h2 id="value">Value<a class="anchor" aria-label="anchor" href="#value"></a></h2>
     <p>A tensor of shape <code>(N, N)</code> representing the correlation matrix.</p>
+    </div>
+    <div class="section level2">
+    <h2 id="examples">Examples<a class="anchor" aria-label="anchor" href="#examples"></a></h2>
+    <p></p><div class="sourceCode r"><pre><code><span><span class="va">x</span> <span class="op">&lt;-</span> <span class="fu"><a href="../reference/op_array.html">op_array</a></span><span class="op">(</span><span class="fu"><a href="https://rdrr.io/r/base/matrix.html" class="external-link">matrix</a></span><span class="op">(</span><span class="fu"><a href="https://rdrr.io/r/base/c.html" class="external-link">c</a></span><span class="op">(</span><span class="fl">1</span>, <span class="fl">2</span>, <span class="fl">3</span>,</span>
+<span>                       <span class="fl">2</span>, <span class="fl">3</span>, <span class="fl">4</span><span class="op">)</span>, nrow <span class="op">=</span> <span class="fl">2</span>, byrow <span class="op">=</span> <span class="cn">TRUE</span><span class="op">)</span><span class="op">)</span></span>
+<span><span class="fu"><a href="../reference/op_corrcoef.html">op_corrcoef</a></span><span class="op">(</span><span class="va">x</span><span class="op">)</span></span></code></pre><p></p></div>
+<p></p><div class="sourceCode"><pre><code><span id="cb1-1"><a href="#cb1-1" tabindex="-1"></a><span class="do">## tf.Tensor(</span></span>
+<span id="cb1-2"><a href="#cb1-2" tabindex="-1"></a><span class="do">## [[1. 1.]</span></span>
+<span id="cb1-3"><a href="#cb1-3" tabindex="-1"></a><span class="do">##  [1. 1.]], shape=(2, 2), dtype=float64)</span></span>
+<span id="cb1-4"><a href="#cb1-4" tabindex="-1"></a></span></code></pre><p></p></div>
     </div>
     <div class="section level2">
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
diff --git a/docs/dev/reference/op_rms_normalization.html b/docs/dev/reference/op_rms_normalization.html
index ccd51a1c6..45bebb7bf 100644
--- a/docs/dev/reference/op_rms_normalization.html
+++ b/docs/dev/reference/op_rms_normalization.html
@@ -102,29 +102,8 @@ <h2 id="value">Value<a class="anchor" aria-label="anchor" href="#value"></a></h2
     <div class="section level2">
     <h2 id="examples">Examples<a class="anchor" aria-label="anchor" href="#examples"></a></h2>
     <p></p><div class="sourceCode r"><pre><code><span><span class="va">x</span> <span class="op">&lt;-</span> <span class="fu"><a href="../reference/random_normal.html">random_normal</a></span><span class="op">(</span><span class="fu"><a href="https://rdrr.io/r/base/c.html" class="external-link">c</a></span><span class="op">(</span><span class="fl">1</span>, <span class="fl">10</span><span class="op">)</span><span class="op">)</span></span>
-<span><span class="fu"><a href="../reference/op_rms_normalization.html">op_rms_normalization</a></span><span class="op">(</span><span class="va">x</span><span class="op">)</span></span></code></pre><p></p></div>
-<p></p><div class="sourceCode"><pre><code><span id="cb1-1"><a href="#cb1-1" tabindex="-1"></a><span class="do">## tf.Tensor(</span></span>
-<span id="cb1-2"><a href="#cb1-2" tabindex="-1"></a><span class="do">## [[-1.1701705   0.9987454   0.27809578 -0.18452731  1.5731977   1.8934134</span></span>
-<span id="cb1-3"><a href="#cb1-3" tabindex="-1"></a><span class="do">##    0.10656697  1.0256069   0.6185074  -0.12678868]], shape=(1, 10), dtype=float32)</span></span>
-<span id="cb1-4"><a href="#cb1-4" tabindex="-1"></a></span></code></pre><p></p></div>
-<p></p><div class="sourceCode r"><pre><code><span><span class="fu"><a href="../reference/op_rms_normalization.html">op_rms_normalization</a></span><span class="op">(</span><span class="va">x</span>, scale <span class="op">=</span> <span class="fl">10</span><span class="op">)</span></span></code></pre><p></p></div>
-<p></p><div class="sourceCode"><pre><code><span id="cb1-1"><a href="#cb1-1" tabindex="-1"></a><span class="do">## tf.Tensor(</span></span>
-<span id="cb1-2"><a href="#cb1-2" tabindex="-1"></a><span class="do">## [[-11.701706    9.987453    2.7809577  -1.845273   15.731977   18.934134</span></span>
-<span id="cb1-3"><a href="#cb1-3" tabindex="-1"></a><span class="do">##     1.0656698  10.256069    6.185074   -1.2678868]], shape=(1, 10), dtype=float32)</span></span>
-<span id="cb1-4"><a href="#cb1-4" tabindex="-1"></a></span></code></pre><p></p></div>
-<p>[-1.1701705   0.9987454   0.27809578 -0.18452731  1.5731977   1.8934134</p><div class="section">
-<h3 id="-r-a-">0.10656697  1.0256069   0.6185074  -0.12678868]: R:-1.1701705%20%20%200.9987454%20%20%200.27809578%20-0.18452731%20%201.5731977%20%20%201.8934134%0A##%20%20%20%200.10656697%20%201.0256069%20%20%200.6185074%20%20-0.12678868<a class="anchor" aria-label="anchor" href="#-r-a-"></a></h3>
-
-
-<p>[-11.701706    9.987453    2.7809577  -1.845273   15.731977   18.934134</p>
-</div>
-
-<div class="section">
-<h3 id="-r-a--1">1.0656698  10.256069    6.185074   -1.2678868]: R:-11.701706%20%20%20%209.987453%20%20%20%202.7809577%20%20-1.845273%20%20%2015.731977%20%20%2018.934134%0A##%20%20%20%20%201.0656698%20%2010.256069%20%20%20%206.185074%20%20%20-1.2678868<a class="anchor" aria-label="anchor" href="#-r-a--1"></a></h3>
-
-
-</div>
-
+<span><span class="fu"><a href="../reference/op_rms_normalization.html">op_rms_normalization</a></span><span class="op">(</span><span class="va">x</span><span class="op">)</span></span>
+<span><span class="fu"><a href="../reference/op_rms_normalization.html">op_rms_normalization</a></span><span class="op">(</span><span class="va">x</span>, scale <span class="op">=</span> <span class="fl">10</span><span class="op">)</span></span></code></pre><p></p></div>
     </div>
     <div class="section level2">
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
diff --git a/docs/dev/reference/op_searchsorted.html b/docs/dev/reference/op_searchsorted.html
index 96ac61b3a..344c8fd7d 100644
--- a/docs/dev/reference/op_searchsorted.html
+++ b/docs/dev/reference/op_searchsorted.html
@@ -93,7 +93,14 @@ <h2 id="value">Value<a class="anchor" aria-label="anchor" href="#value"></a></h2
     </div>
     <div class="section level2">
     <h2 id="examples">Examples<a class="anchor" aria-label="anchor" href="#examples"></a></h2>
-    <p></p><div class="sourceCode {r}"><pre><code></code></pre><p></p></div>
+    <p></p><div class="sourceCode r"><pre><code><span><span class="va">sorted</span> <span class="op">&lt;-</span> <span class="fu"><a href="../reference/op_array.html">op_array</a></span><span class="op">(</span><span class="fu"><a href="https://rdrr.io/r/base/c.html" class="external-link">c</a></span><span class="op">(</span><span class="op">-</span><span class="fl">1</span>, <span class="fl">0</span>, <span class="fl">2</span>, <span class="fl">4</span><span class="op">)</span><span class="op">)</span></span>
+<span><span class="va">queries</span> <span class="op">&lt;-</span> <span class="fu"><a href="https://rdrr.io/r/base/c.html" class="external-link">c</a></span><span class="op">(</span><span class="op">-</span><span class="fl">2</span>, <span class="fl">0</span>, <span class="fl">3</span>, <span class="fl">7</span><span class="op">)</span></span>
+<span><span class="fu"><a href="../reference/op_searchsorted.html">op_searchsorted</a></span><span class="op">(</span><span class="va">sorted</span>, <span class="va">queries</span><span class="op">)</span></span></code></pre><p></p></div>
+<p></p><div class="sourceCode"><pre><code><span id="cb1-1"><a href="#cb1-1" tabindex="-1"></a><span class="do">## tf.Tensor([1 2 4 5], shape=(4), dtype=int32)</span></span>
+<span id="cb1-2"><a href="#cb1-2" tabindex="-1"></a></span></code></pre><p></p></div>
+<p></p><div class="sourceCode r"><pre><code><span><span class="fu"><a href="../reference/op_searchsorted.html">op_searchsorted</a></span><span class="op">(</span><span class="va">sorted</span>, <span class="va">queries</span>, side <span class="op">=</span> <span class="st">"right"</span>, zero_indexed <span class="op">=</span> <span class="cn">TRUE</span><span class="op">)</span></span></code></pre><p></p></div>
+<p></p><div class="sourceCode"><pre><code><span id="cb1-1"><a href="#cb1-1" tabindex="-1"></a><span class="do">## tf.Tensor([0 2 3 4], shape=(4), dtype=int32)</span></span>
+<span id="cb1-2"><a href="#cb1-2" tabindex="-1"></a></span></code></pre><p></p></div>
     </div>
     <div class="section level2">
     <h2 id="see-also">See also<a class="anchor" aria-label="anchor" href="#see-also"></a></h2>
diff --git a/docs/dev/reference/optimizer_muon.html b/docs/dev/reference/optimizer_muon.html
index 4cb945b28..bd1aa8c57 100644
--- a/docs/dev/reference/optimizer_muon.html
+++ b/docs/dev/reference/optimizer_muon.html
@@ -2,7 +2,7 @@
 <!-- Generated by pkgdown: do not edit by hand --><html lang="en"><head><meta http-equiv="Content-Type" content="text/html; charset=UTF-8"><meta charset="utf-8"><meta http-equiv="X-UA-Compatible" content="IE=edge"><meta name="viewport" content="width=device-width, initial-scale=1, shrink-to-fit=no"><title>Optimizer that implements the Muon algorithm. — optimizer_muon • keras3</title><!-- favicons --><link rel="icon" type="image/png" sizes="96x96" href="../favicon-96x96.png"><link rel="icon" type="”image/svg+xml”" href="../favicon.svg"><link rel="apple-touch-icon" sizes="180x180" href="../apple-touch-icon.png"><link rel="icon" sizes="any" href="../favicon.ico"><link rel="manifest" href="../site.webmanifest"><script src="../deps/jquery-3.6.0/jquery-3.6.0.min.js"></script><meta name="viewport" content="width=device-width, initial-scale=1, shrink-to-fit=no"><link href="../deps/bootstrap-5.3.1/bootstrap.min.css" rel="stylesheet"><script src="../deps/bootstrap-5.3.1/bootstrap.bundle.min.js"></script><link href="../deps/font-awesome-6.5.2/css/all.min.css" rel="stylesheet"><link href="../deps/font-awesome-6.5.2/css/v4-shims.min.css" rel="stylesheet"><script src="../deps/headroom-0.11.0/headroom.min.js"></script><script src="../deps/headroom-0.11.0/jQuery.headroom.min.js"></script><script src="../deps/bootstrap-toc-1.0.1/bootstrap-toc.min.js"></script><script src="../deps/clipboard.js-2.0.11/clipboard.min.js"></script><script src="../deps/search-1.0.0/autocomplete.jquery.min.js"></script><script src="../deps/search-1.0.0/fuse.min.js"></script><script src="../deps/search-1.0.0/mark.min.js"></script><!-- pkgdown --><script src="../pkgdown.js"></script><link href="../extra.css" rel="stylesheet"><meta property="og:title" content="Optimizer that implements the Muon algorithm. — optimizer_muon"><meta name="description" content="Note that this optimizer should not be used in the following layers:
 Embedding layer
 Final output fully connected layer
-Any 0,1-D variables
+Any 0- or 1-D variables
 
 
 These should all be optimized using AdamW.
@@ -23,7 +23,7 @@
 exclude_layers = &quot;output_dense&quot;."><meta property="og:description" content="Note that this optimizer should not be used in the following layers:
 Embedding layer
 Final output fully connected layer
-Any 0,1-D variables
+Any 0- or 1-D variables
 
 
 These should all be optimized using AdamW.
@@ -98,7 +98,7 @@ <h1>Optimizer that implements the Muon algorithm.</h1>
     <div class="ref-description section level2">
     <p>Note that this optimizer should not be used in the following layers:</p><ol><li><p>Embedding layer</p></li>
 <li><p>Final output fully connected layer</p></li>
-<li><p>Any 0,1-D variables</p></li>
+<li><p>Any 0- or 1-D variables</p></li>
 </ol><p>These should all be optimized using AdamW.</p>
 <p>The Muon optimizer can use both the Muon update step or the
 AdamW update step based on the following:</p><ul><li><p>For any variable that isn't 2D, 3D or 4D, the AdamW step
@@ -177,6 +177,69 @@ <h2 id="arguments">Arguments<a class="anchor" aria-label="anchor" href="#argumen
 It is used as in AdamW. Defaults to <code>1e-7</code>.</p></dd>
 
 
+<dt id="arg-weight-decay">weight_decay<a class="anchor" aria-label="anchor" href="#arg-weight-decay"></a></dt>
+<dd><p>Float. If set, weight decay is applied.</p></dd>
+
+
+<dt id="arg-clipnorm">clipnorm<a class="anchor" aria-label="anchor" href="#arg-clipnorm"></a></dt>
+<dd><p>Float. If set, the gradient of each weight is individually
+clipped so that its norm is no higher than this value.</p></dd>
+
+
+<dt id="arg-clipvalue">clipvalue<a class="anchor" aria-label="anchor" href="#arg-clipvalue"></a></dt>
+<dd><p>Float. If set, the gradient of each weight is clipped to be
+no higher than this value.</p></dd>
+
+
+<dt id="arg-global-clipnorm">global_clipnorm<a class="anchor" aria-label="anchor" href="#arg-global-clipnorm"></a></dt>
+<dd><p>Float. If set, the gradient of all weights is clipped
+so that their global norm is no higher than this value.</p></dd>
+
+
+<dt id="arg-use-ema">use_ema<a class="anchor" aria-label="anchor" href="#arg-use-ema"></a></dt>
+<dd><p>Boolean, defaults to <code>FALSE</code>.
+If <code>TRUE</code>, exponential moving average (EMA) is applied. EMA consists of
+computing an exponential moving average of the weights of the model (as
+the weight values change after each training batch), and periodically
+overwriting the weights with their moving average.</p></dd>
+
+
+<dt id="arg-ema-momentum">ema_momentum<a class="anchor" aria-label="anchor" href="#arg-ema-momentum"></a></dt>
+<dd><p>Float, defaults to <code>0.99</code>. Only used if <code>use_ema = TRUE</code>.
+This is the momentum to use when computing the EMA of the model's
+weights: <code>new_average = ema_momentum * old_average + (1 - ema_momentum) * current_variable_value</code>.</p></dd>
+
+
+<dt id="arg-ema-overwrite-frequency">ema_overwrite_frequency<a class="anchor" aria-label="anchor" href="#arg-ema-overwrite-frequency"></a></dt>
+<dd><p>Int or <code>NULL</code>, defaults to <code>NULL</code>. Only used if <code>use_ema = TRUE</code>. Every
+<code>ema_overwrite_frequency</code> steps of iterations, we overwrite the model
+variable by its moving average. If <code>NULL</code>, the optimizer does not overwrite
+model variables in the middle of training, and you need to explicitly
+overwrite the variables at the end of training by calling
+<code>optimizer$finalize_variable_values()</code> (which updates the model variables
+in-place). When using the built-in <code><a href="https://generics.r-lib.org/reference/fit.html" class="external-link">fit()</a></code> training loop, this happens
+automatically after the last epoch, and you don't need to do anything.</p></dd>
+
+
+<dt id="arg-loss-scale-factor">loss_scale_factor<a class="anchor" aria-label="anchor" href="#arg-loss-scale-factor"></a></dt>
+<dd><p>Float or <code>NULL</code>. If a float, the scale factor will be multiplied the loss
+before computing gradients, and the inverse of the scale factor will be
+multiplied by the gradients before updating variables. Useful for
+preventing underflow during mixed precision training. Alternately,
+<code><a href="optimizer_loss_scale.html">optimizer_loss_scale()</a></code> will automatically set a loss scale factor.</p></dd>
+
+
+<dt id="arg-gradient-accumulation-steps">gradient_accumulation_steps<a class="anchor" aria-label="anchor" href="#arg-gradient-accumulation-steps"></a></dt>
+<dd><p>Int or <code>NULL</code>. If an int, model and optimizer variables will not be updated
+at every step; instead they will be updated every <code>gradient_accumulation_steps</code>
+steps, using the average value of the gradients since the last update. This
+is known as "gradient accumulation". This can be useful when your batch size is
+very small, in order to reduce gradient noise at each update step. EMA
+frequency will look at "accumulated" iterations value (optimizer steps //
+gradient_accumulation_steps). Learning rate schedules will look at "real"
+iterations value (optimizer steps).</p></dd>
+
+
 <dt id="arg-name">name<a class="anchor" aria-label="anchor" href="#arg-name"></a></dt>
 <dd><p>String, name for the object</p></dd>
 
diff --git a/docs/dev/search.json b/docs/dev/search.json
index 1cc1e4d61..27e78064f 100644
--- a/docs/dev/search.json
+++ b/docs/dev/search.json
@@ -1 +1 @@
-[{"path":"https://keras3.posit.co/dev/articles/custom_train_step_in_tensorflow.html","id":"introduction","dir":"Articles","previous_headings":"","what":"Introduction","title":"Customizing what happens in `fit()` with TensorFlow","text":"’re supervised learning, can use fit() everything works smoothly. need take control every little detail, can write training loop entirely scratch. need custom training algorithm, still want benefit convenient features fit(), callbacks, built-distribution support, step fusing? core principle Keras progressive disclosure complexity. always able get lower-level workflows gradual way. shouldn’t fall cliff high-level functionality doesn’t exactly match use case. able gain control small details retaining commensurate amount high-level convenience. need customize fit() , override training step function Model class. function called fit() every batch data. able call fit() usual – running learning algorithm. Note pattern prevent building models Functional API. can whether ’re building Sequential models, Functional API models, subclassed models. Let’s see works.","code":""},{"path":"https://keras3.posit.co/dev/articles/custom_train_step_in_tensorflow.html","id":"setup","dir":"Articles","previous_headings":"","what":"Setup","title":"Customizing what happens in `fit()` with TensorFlow","text":"","code":"library(reticulate) library(tensorflow, exclude = c(\"set_random_seed\", \"shape\")) library(keras3)"},{"path":"https://keras3.posit.co/dev/articles/custom_train_step_in_tensorflow.html","id":"a-first-simple-example","dir":"Articles","previous_headings":"","what":"A first simple example","title":"Customizing what happens in `fit()` with TensorFlow","text":"Let’s start simple example: create new class subclasses Model. just override method train_step(self, data). return dictionary mapping metric names (including loss) current value. input argument data gets passed fit training data: pass arrays, calling fit(x, y, ...), data list (x, y) pass tf_dataset, calling fit(dataset, ...), data gets yielded dataset batch. body train_step() method, implement regular training update, similar already familiar . Importantly, compute loss via self.compute_loss(), wraps loss(es) function(s) passed compile(). Similarly, call metric$update_state(y, y_pred) metrics self$metrics, update state metrics passed compile(), query results self$metrics end retrieve current value. Let’s try :","code":"CustomModel <- new_model_class(   \"CustomModel\",   train_step = function(data) {     c(x, y = NULL, sample_weight = NULL) %<-% data      with(tf$GradientTape() %as% tape, {       y_pred <- self(x, training = TRUE)       loss <- self$compute_loss(y = y, y_pred = y_pred,                                 sample_weight = sample_weight)     })      # Compute gradients     trainable_vars <- self$trainable_variables     gradients <- tape$gradient(loss, trainable_vars)      # Update weights     self$optimizer$apply(gradients, trainable_vars)      # Update metrics (includes the metric that tracks the loss)     for (metric in self$metrics) {       if (metric$name == \"loss\")         metric$update_state(loss)       else         metric$update_state(y, y_pred)     }      # Return a dict mapping metric names to current value     metrics <- lapply(self$metrics, function(m) m$result())     metrics <- setNames(metrics, sapply(self$metrics, function(m) m$name))     metrics   } ) # Construct and compile an instance of CustomModel inputs <- keras_input(shape = 32) outputs <- layer_dense(inputs, 1) model <- CustomModel(inputs, outputs) model |> compile(optimizer = \"adam\", loss = \"mse\", metrics = \"mae\")  # Just use `fit` as usual x <- random_normal(c(1000, 32)) y <- random_normal(c(1000, 1)) model |> fit(x, y, epochs = 3) ## Epoch 1/3 ## 32/32 - 1s - 20ms/step - mae: 1.4339 - loss: 3.2271 ## Epoch 2/3 ## 32/32 - 0s - 3ms/step - mae: 1.3605 - loss: 2.9034 ## Epoch 3/3 ## 32/32 - 0s - 2ms/step - mae: 1.2960 - loss: 2.6272"},{"path":"https://keras3.posit.co/dev/articles/custom_train_step_in_tensorflow.html","id":"going-lower-level","dir":"Articles","previous_headings":"","what":"Going lower-level","title":"Customizing what happens in `fit()` with TensorFlow","text":"Naturally, just skip passing loss function compile(), instead everything manually train_step. Likewise metrics. ’s lower-level example, uses compile() configure optimizer: start creating Metric instances track loss MAE score (__init__()). implement custom train_step() updates state metrics (calling update_state() ), query (via result()) return current average value, displayed progress bar pass callback. Note need call reset_states() metrics epoch! Otherwise calling result() return average since start training, whereas usually work per-epoch averages. Thankfully, framework can us: just list metric want reset metrics property model. model call reset_states() object listed beginning fit() epoch beginning call evaluate().","code":"CustomModel <- new_model_class(   \"CustomModel\",   initialize = function(...) {     super$initialize(...)     self$loss_tracker <- metric_mean(name = \"loss\")     self$mae_metric <- metric_mean_absolute_error(name = \"mae\")     self$loss_fn <- loss_mean_squared_error()   },   train_step = function(data) {     c(x, y = NULL, sample_weight = NULL) %<-% data      with(tf$GradientTape() %as% tape, {       y_pred <- self(x, training = TRUE)       loss <- self$loss_fn(y, y_pred, sample_weight = sample_weight)     })      # Compute gradients     trainable_vars <- self$trainable_variables     gradients <- tape$gradient(loss, trainable_vars)      # Update weights     self$optimizer$apply(gradients, trainable_vars)      # Compute our own metrics     self$loss_tracker$update_state(loss)     self$mae_metric$update_state(y, y_pred)      # Return a dict mapping metric names to current value     list(       loss = self$loss_tracker$result(),       mae = self$mae_metric$result()     )   },   metrics = mark_active(function() {     # We list our `Metric` objects here so that `reset_states()` can be     # called automatically at the start of each epoch     # or at the start of `evaluate()`.     list(self$loss_tracker, self$mae_metric)   }) )   # Construct and compile an instance of CustomModel inputs <- keras_input(shape = 32) outputs <- layer_dense(inputs, 1) model <- CustomModel(inputs, outputs)  # We don't pass a loss or metrics here. model |> compile(optimizer = \"adam\")  # Just use `fit` as usual x <- random_normal(c(1000, 32)) y <- random_normal(c(1000, 1)) model |> fit(x, y, epochs = 3) ## Epoch 1/3 ## 32/32 - 1s - 18ms/step - loss: 2.5170 - mae: 1.2923 ## Epoch 2/3 ## 32/32 - 0s - 2ms/step - loss: 2.2689 - mae: 1.2241 ## Epoch 3/3 ## 32/32 - 0s - 2ms/step - loss: 2.0578 - mae: 1.1633"},{"path":"https://keras3.posit.co/dev/articles/custom_train_step_in_tensorflow.html","id":"supporting-sample_weight-class_weight","dir":"Articles","previous_headings":"","what":"Supporting sample_weight & class_weight","title":"Customizing what happens in `fit()` with TensorFlow","text":"may noticed first basic example didn’t make mention sample weighting. want support fit() arguments sample_weight class_weight, ’d simply following: Unpack sample_weight data argument Pass compute_loss & update_state (course, also just apply manually don’t rely compile() losses & metrics) ’s .","code":"CustomModel <- new_model_class(   \"CustomModel\",   train_step = function(data) {     c(x, y = NULL, sample_weight = NULL) %<-% data      with(tf$GradientTape() %as% tape, {       y_pred <- self(x, training = TRUE)       loss <- self$compute_loss(y = y, y_pred = y_pred,                                 sample_weight = sample_weight)     })      # Compute gradients     trainable_vars <- self$trainable_variables     gradients <- tape$gradient(loss, trainable_vars)      # Update weights     self$optimizer$apply_gradients(zip_lists(gradients, trainable_vars))      # Update metrics (includes the metric that tracks the loss)     for (metric in self$metrics) {       if (metric$name == \"loss\") {         metric$update_state(loss)       } else {         metric$update_state(y, y_pred, sample_weight = sample_weight)       }     }      # Return a dict mapping metric names to current value     metrics <- lapply(self$metrics, function(m) m$result())     metrics <- setNames(metrics, sapply(self$metrics, function(m) m$name))     metrics   } )   # Construct and compile an instance of CustomModel inputs <- keras_input(shape = 32) outputs <- layer_dense(inputs, units = 1) model <- CustomModel(inputs, outputs) model |> compile(optimizer = \"adam\", loss = \"mse\", metrics = \"mae\")  # You can now use sample_weight argument x <- random_normal(c(1000, 32)) y <- random_normal(c(1000, 1)) sw <- random_normal(c(1000, 1)) model |> fit(x, y, sample_weight = sw, epochs = 3) ## Epoch 1/3 ## 32/32 - 1s - 20ms/step - mae: 1.3434 - loss: 0.1681 ## Epoch 2/3 ## 32/32 - 0s - 2ms/step - mae: 1.3364 - loss: 0.1394 ## Epoch 3/3 ## 32/32 - 0s - 2ms/step - mae: 1.3286 - loss: 0.1148"},{"path":"https://keras3.posit.co/dev/articles/custom_train_step_in_tensorflow.html","id":"providing-your-own-evaluation-step","dir":"Articles","previous_headings":"","what":"Providing your own evaluation step","title":"Customizing what happens in `fit()` with TensorFlow","text":"want calls model.evaluate()? override test_step exactly way. ’s looks like:","code":"CustomModel <- new_model_class(   \"CustomModel\",   test_step = function(data) {     # Unpack the data     c(x, y = NULL, sw = NULL) %<-% data     # Compute predictions     y_pred = self(x, training = FALSE)     # Updates the metrics tracking the loss     self$compute_loss(y = y, y_pred = y_pred, sample_weight = sw)     # Update the metrics.     for (metric in self$metrics) {       if (metric$name != \"loss\") {         metric$update_state(y, y_pred, sample_weight = sw)       }     }     # Return a dict mapping metric names to current value.     # Note that it will include the loss (tracked in self.metrics).     metrics <- lapply(self$metrics, function(m) m$result())     metrics <- setNames(metrics, sapply(self$metrics, function(m) m$name))     metrics   } )  # Construct an instance of CustomModel inputs <- keras_input(shape = 32) outputs <- layer_dense(inputs, 1) model <- CustomModel(inputs, outputs) model |> compile(loss = \"mse\", metrics = \"mae\")  # Evaluate with our custom test_step x <- random_normal(c(1000, 32)) y <- random_normal(c(1000, 1)) model |> evaluate(x, y) ## 32/32 - 0s - 10ms/step - mae: 1.3871 - loss: 0.0000e+00 ## $loss ## tf.Tensor(0.0, shape=(), dtype=float32) ## ## $compile_metrics ## $compile_metrics$mae ## tf.Tensor(1.3871489, shape=(), dtype=float32)"},{"path":"https://keras3.posit.co/dev/articles/custom_train_step_in_tensorflow.html","id":"wrapping-up-an-end-to-end-gan-example","dir":"Articles","previous_headings":"","what":"Wrapping up: an end-to-end GAN example","title":"Customizing what happens in `fit()` with TensorFlow","text":"Let’s walk end--end example leverages everything just learned. Let’s consider: generator network meant generate 28x28x1 images. discriminator network meant classify 28x28x1 images two classes (“fake” “real”). One optimizer . loss function train discriminator. ’s feature-complete GAN class, overriding compile() use signature, implementing entire GAN algorithm 17 lines train_step: Let’s test-drive : ideas behind deep learning simple, implementation painful?","code":"# Create the discriminator discriminator <-   keras_model_sequential(name = \"discriminator\", input_shape = c(28, 28, 1)) |>    layer_conv_2d(filters = 64, kernel_size = c(3, 3),                 strides = c(2, 2),  padding = \"same\") |>   layer_activation_leaky_relu(negative_slope = 0.2) |>    layer_conv_2d(filters = 128, kernel_size = c(3, 3),                 strides = c(2, 2), padding = \"same\") |>   layer_activation_leaky_relu(negative_slope = 0.2) |>    layer_global_max_pooling_2d() |>   layer_dense(units = 1)   # Create the generator latent_dim <- 128 generator <-   keras_model_sequential(name = \"generator\", input_shape = latent_dim) |>    layer_dense(7 * 7 * 128) |>   layer_activation_leaky_relu(negative_slope = 0.2) |>    layer_reshape(target_shape = c(7, 7, 128)) |>    layer_conv_2d_transpose(filters = 128, kernel_size = c(4, 4),                           strides = c(2, 2), padding = \"same\") |>   layer_activation_leaky_relu(negative_slope = 0.2) |>    layer_conv_2d_transpose(filters = 128, kernel_size = c(4, 4),                           strides = c(2, 2), padding = \"same\") |>   layer_activation_leaky_relu(negative_slope = 0.2) |>    layer_conv_2d(filters = 1, kernel_size = c(7, 7), padding = \"same\",                 activation = \"sigmoid\") GAN <- Model(   classname = \"GAN\",    initialize = function(discriminator, generator, latent_dim, ...) {     super$initialize(...)     self$discriminator <- discriminator     self$generator <- generator     self$latent_dim <- as.integer(latent_dim)     self$d_loss_tracker <- metric_mean(name = \"d_loss\")     self$g_loss_tracker <- metric_mean(name = \"g_loss\")   },    compile = function(d_optimizer, g_optimizer, loss_fn, ...) {     super$compile(...)     self$d_optimizer <- d_optimizer     self$g_optimizer <- g_optimizer     self$loss_fn <- loss_fn   },   metrics = active_property(function() {     list(self$d_loss_tracker, self$g_loss_tracker)   }),    train_step = function(real_images) {      # Sample random points in the latent space     batch_size <- shape(real_images)[[1]]     random_latent_vectors <-       tf$random$normal(shape(batch_size, self$latent_dim))      # Decode them to fake images     generated_images <- self$generator(random_latent_vectors)      # Combine them with real images     combined_images <- op_concatenate(list(generated_images,                                            real_images))      # Assemble labels discriminating real from fake images     labels <- op_concatenate(list(op_ones(c(batch_size, 1)),                                   op_zeros(c(batch_size, 1))))      # Add random noise to the labels - important trick!     labels %<>% `+`(tf$random$uniform(shape(.), maxval = 0.05))      # Train the discriminator     with(tf$GradientTape() %as% tape, {       predictions <- self$discriminator(combined_images)       d_loss <- self$loss_fn(labels, predictions)     })     grads <- tape$gradient(d_loss, self$discriminator$trainable_weights)     self$d_optimizer$apply_gradients(       zip_lists(grads, self$discriminator$trainable_weights))      # Sample random points in the latent space     random_latent_vectors <-       tf$random$normal(shape(batch_size, self$latent_dim))      # Assemble labels that say \"all real images\"     misleading_labels <- op_zeros(c(batch_size, 1))      # Train the generator (note that we should *not* update the weights     # of the discriminator)!     with(tf$GradientTape() %as% tape, {       predictions <- self$discriminator(self$generator(random_latent_vectors))       g_loss <- self$loss_fn(misleading_labels, predictions)     })     grads <- tape$gradient(g_loss, self$generator$trainable_weights)     self$g_optimizer$apply_gradients(       zip_lists(grads, self$generator$trainable_weights))      list(d_loss = d_loss, g_loss = g_loss)   } ) batch_size <- 64 c(c(x_train, .), c(x_test, .)) %<-% dataset_mnist() all_digits <- op_concatenate(list(x_train, x_test)) all_digits <- op_reshape(all_digits, c(-1, 28, 28, 1)) dataset <- all_digits |>   tfdatasets::tensor_slices_dataset() |>   tfdatasets::dataset_map(\\(x) op_cast(x, \"float32\") / 255) |>   tfdatasets::dataset_shuffle(buffer_size = 1024) |>   tfdatasets::dataset_batch(batch_size = batch_size)  gan <- GAN(discriminator = discriminator,            generator = generator,            latent_dim = latent_dim)  gan |> compile(   d_optimizer = optimizer_adam(learning_rate = 0.0003),   g_optimizer = optimizer_adam(learning_rate = 0.0003),   loss_fn = loss_binary_crossentropy(from_logits = TRUE) )  # To limit the execution time, we only train on 100 batches. You can train on # the entire dataset. You will need about 20 epochs to get nice results. gan |> fit(   tfdatasets::dataset_take(dataset, 100),   epochs = 1 ) ## 100/100 - 5s - 51ms/step - d_loss: 0.0000e+00 - g_loss: 0.0000e+00"},{"path":"https://keras3.posit.co/dev/articles/distributed_training_with_tensorflow.html","id":"introduction","dir":"Articles","previous_headings":"","what":"Introduction","title":"Multi-GPU distributed training with TensorFlow","text":"generally two ways distribute computation across multiple devices: Data parallelism, single model gets replicated multiple devices multiple machines. processes different batches data, merge results. exist many variants setup, differ different model replicas merge results, whether stay sync every batch whether loosely coupled, etc. Model parallelism, different parts single model run different devices, processing single batch data together. works best models naturally-parallel architecture, models feature multiple branches. guide focuses data parallelism, particular synchronous data parallelism, different replicas model stay sync batch process. Synchronicity keeps model convergence behavior identical see single-device training. Specifically, guide teaches use tf.distribute API train Keras models multiple GPUs, minimal changes code, multiple GPUs (typically 2 16) installed single machine (single host, multi-device training). common setup researchers small-scale industry workflows.","code":""},{"path":"https://keras3.posit.co/dev/articles/distributed_training_with_tensorflow.html","id":"setup","dir":"Articles","previous_headings":"","what":"Setup","title":"Multi-GPU distributed training with TensorFlow","text":"","code":"library(keras3) library(tensorflow, exclude = c(\"shape\", \"set_random_seed\")) library(tfdatasets, exclude = \"shape\")"},{"path":"https://keras3.posit.co/dev/articles/distributed_training_with_tensorflow.html","id":"single-host-multi-device-synchronous-training","dir":"Articles","previous_headings":"","what":"Single-host, multi-device synchronous training","title":"Multi-GPU distributed training with TensorFlow","text":"setup, one machine several GPUs (typically 2 16). device run copy model (called replica). simplicity, follows, ’ll assume ’re dealing 8 GPUs, loss generality. works step training: current batch data (called global batch) split 8 different sub-batches (called local batches). instance, global batch 512 samples, 8 local batches 64 samples. 8 replicas independently processes local batch: run forward pass, backward pass, outputting gradient weights respect loss model local batch. weight updates originating local gradients efficiently merged across 8 replicas. done end every step, replicas always stay sync. practice, process synchronously updating weights model replicas handled level individual weight variable. done mirrored variable object. use single-host, multi-device synchronous training Keras model, use tf$distribute$MirroredStrategy API. ’s works: Instantiate MirroredStrategy, optionally configuring specific devices want use (default strategy use GPUs available). Use strategy object open scope, within scope, create Keras objects need contain variables. Typically, means creating & compiling model inside distribution scope. cases, first call fit() may also create variables, ’s good idea put fit() call scope well. Train model via fit() usual. Importantly, recommend use tf.data.Dataset objects load data multi-device distributed workflow. Schematically, looks like : ’s simple end--end runnable example:","code":"# Create a MirroredStrategy. strategy <- tf$distribute$MirroredStrategy() cat(sprintf('Number of devices: %d\\n', strategy$num_replicas_in_sync))  # Open a strategy scope. with(startegy$scope(), {   # Everything that creates variables should be under the strategy scope.   # In general this is only model construction & `compile()`.   model <- Model(...)   model |> compile(...)    # Train the model on all available devices.   model |> fit(train_dataset, validation_data=val_dataset, ...)    # Test the model on all available devices.   model |> evaluate(test_dataset) }) get_compiled_model <- function() {   inputs <- keras_input(shape = 784)   outputs <- inputs |>     layer_dense(units = 256, activation = \"relu\") |>     layer_dense(units = 256, activation = \"relu\") |>     layer_dense(units = 10)   model <- keras_model(inputs, outputs)   model |> compile(     optimizer = optimizer_adam(),     loss = loss_sparse_categorical_crossentropy(from_logits = TRUE),     metrics = list(metric_sparse_categorical_accuracy())   )   model }  get_dataset <- function(batch_size = 64) {    c(c(x_train, y_train), c(x_test, y_test)) %<-% dataset_mnist()   x_train <- array_reshape(x_train, c(-1, 784)) / 255   x_test <- array_reshape(x_test, c(-1, 784)) / 255    # Reserve 10,000 samples for validation.   val_i <- sample.int(nrow(x_train), 10000)   x_val <- x_train[val_i,]   y_val <- y_train[val_i]   x_train = x_train[-val_i,]   y_train = y_train[-val_i]    y_train <- array_reshape(y_train, c(-1, 1))   y_val <- array_reshape(y_val, c(-1, 1))   y_test <- array_reshape(y_test, c(-1, 1))    # Prepare the training dataset.   train_dataset <- list(x_train, y_train) |>     lapply(np_array, \"float32\") |>     tensor_slices_dataset() |>     dataset_batch(batch_size)    # Prepare the validation dataset.   val_dataset <- list(x_val, y_val) |>     lapply(np_array, \"float32\") |>     tensor_slices_dataset() |>     dataset_batch(batch_size)    # Prepare the test dataset.   test_dataset <- list(x_test, y_test) |>     lapply(np_array, \"float32\") |>     tensor_slices_dataset() |>     dataset_batch(batch_size)    list(train_dataset, val_dataset, test_dataset) }  # Create a MirroredStrategy. strategy <- tf$distribute$MirroredStrategy() cat(sprintf('Number of devices: %d\\n', strategy$num_replicas_in_sync)) ## Number of devices: 2 # Open a strategy scope. with(strategy$scope(), {   # Everything that creates variables should be under the strategy scope.   # In general this is only model construction & `compile()`.   model <- get_compiled_model()    c(train_dataset, val_dataset, test_dataset) %<-% get_dataset()    # Train the model on all available devices.   model |> fit(train_dataset, epochs = 2, validation_data = val_dataset)    # Test the model on all available devices.   model |> evaluate(test_dataset)  }) ## Epoch 1/2 ## 782/782 - 6s - 7ms/step - loss: 0.2616 - sparse_categorical_accuracy: 0.9232 - val_loss: 0.1510 - val_sparse_categorical_accuracy: 0.9524 ## Epoch 2/2 ## 782/782 - 4s - 5ms/step - loss: 0.1044 - sparse_categorical_accuracy: 0.9682 - val_loss: 0.1004 - val_sparse_categorical_accuracy: 0.9675 ## 157/157 - 0s - 3ms/step - loss: 0.1004 - sparse_categorical_accuracy: 0.9683 ## $loss ## [1] 0.1004203 ## ## $sparse_categorical_accuracy ## [1] 0.9683"},{"path":"https://keras3.posit.co/dev/articles/distributed_training_with_tensorflow.html","id":"using-callbacks-to-ensure-fault-tolerance","dir":"Articles","previous_headings":"","what":"Using callbacks to ensure fault tolerance","title":"Multi-GPU distributed training with TensorFlow","text":"using distributed training, always make sure strategy recover failure (fault tolerance). simplest way handle pass ModelCheckpoint callback fit(), save model regular intervals (e.g. every 100 batches every epoch). can restart training saved model. ’s simple example:","code":"# Prepare a directory to store all the checkpoints. checkpoint_dir <- \"./ckpt\" if (!dir.exists(checkpoint_dir)) {   dir.create(checkpoint_dir) }  make_or_restore_model <- function() {   # Either restore the latest model, or create a fresh one   # if there is no checkpoint available.   checkpoints <- list.files(checkpoint_dir, pattern = \"ckpt-.*\\\\.keras\",                             full.names = TRUE)    if (length(checkpoints) > 0) {     checkpoint_epochs <- as.integer(sub(\"ckpt-([0-9]+)\\\\.keras\", \"\\\\1\",                                         basename(checkpoints)))     latest_checkpoint <- checkpoints[which.max(checkpoint_epochs)]     load_model(latest_checkpoint)   } else {     get_compiled_model()   } }    run_training <- function(epochs = 1) {   # Create a MirroredStrategy.   strategy <- tf$distribute$MirroredStrategy()    # Open a strategy scope and create/restore the model   with(strategy$scope(), {     model <- make_or_restore_model()      callbacks <- list(       # This callback saves a SavedModel every epoch       # We include the current epoch in the folder name.       callback_model_checkpoint(         filepath = paste0(checkpoint_dir, \"/ckpt-{epoch}.keras\"),         save_freq = \"epoch\"       ))      model |> fit(       train_dataset,       epochs = epochs,       callbacks = callbacks,       validation_data = val_dataset,       verbose = 2     )   }) }  # Running the first time creates the model run_training(epochs = 1) ## 782/782 - 5s - 6ms/step - loss: 0.2575 - sparse_categorical_accuracy: 0.9237 - val_loss: 0.1428 - val_sparse_categorical_accuracy: 0.9537 # Calling the same function again will resume from where we left off run_training(epochs = 1) ## 782/782 - 5s - 6ms/step - loss: 0.1011 - sparse_categorical_accuracy: 0.9693 - val_loss: 0.1185 - val_sparse_categorical_accuracy: 0.9602"},{"path":"https://keras3.posit.co/dev/articles/distributed_training_with_tensorflow.html","id":"tfdata-performance-tips","dir":"Articles","previous_headings":"","what":"tf$data performance tips","title":"Multi-GPU distributed training with TensorFlow","text":"distributed training, efficiency load data can often become critical. tips make sure tf$data pipelines run fast possible. Note dataset batching creating dataset, make sure batched global batch size. instance, 8 GPUs capable running batch 64 samples, call use global batch size 512. Calling dataset_cache() call dataset_cache() dataset, data cached running first iteration data. Every subsequent iteration use cached data. cache can memory (default) local file specify. can improve performance : data expected change iteration iteration reading data remote distributed filesystem reading data local disk, data fit memory workflow significantly IO-bound (e.g. reading & decoding image files). Calling dataset_prefetch(buffer_size) almost always call dataset_prefetch(buffer_size) creating dataset. means data pipeline run asynchronously model, new samples preprocessed stored buffer current batch samples used train model. next batch prefetched GPU memory time current batch . ’s !","code":""},{"path":"https://keras3.posit.co/dev/articles/distribution.html","id":"introduction","dir":"Articles","previous_headings":"","what":"Introduction","title":"Distributed training with Keras 3","text":"Keras distribution API new interface designed facilitate distributed deep learning across variety backends like JAX, TensorFlow PyTorch. powerful API introduces suite tools enabling data model parallelism, allowing efficient scaling deep learning models multiple accelerators hosts. Whether leveraging power GPUs TPUs, API provides streamlined approach initializing distributed environments, defining device meshes, orchestrating layout tensors across computational resources. classes like DataParallel ModelParallel, abstracts complexity involved parallel computation, making easier developers accelerate machine learning workflows.","code":""},{"path":"https://keras3.posit.co/dev/articles/distribution.html","id":"how-it-works","dir":"Articles","previous_headings":"","what":"How it works","title":"Distributed training with Keras 3","text":"Keras distribution API provides global programming model allows developers compose applications operate tensors global context (working single device) automatically managing distribution across many devices. API leverages underlying framework (e.g. JAX) distribute program tensors according sharding directives procedure called single program, multiple data (SPMD) expansion. decoupling application sharding directives, API enables running application single device, multiple devices, even multiple clients, preserving global semantics.","code":""},{"path":"https://keras3.posit.co/dev/articles/distribution.html","id":"setup","dir":"Articles","previous_headings":"","what":"Setup","title":"Distributed training with Keras 3","text":"","code":"# This guide assumes there are 8 GPUs available for testing. If you don't have # 8 gpus available locally, you can set the following envvar to # make xla initialize the CPU as 8 devices, to enable local testing Sys.setenv(\"CUDA_VISIBLE_DEVICES\" = \"\") Sys.setenv(\"XLA_FLAGS\" = \"--xla_force_host_platform_device_count=8\") library(keras3)  # The distribution API is only implemented for the JAX backend for now. use_backend(\"jax\", FALSE) jax <- reticulate::import(\"jax\")  library(tfdatasets, exclude = \"shape\") # For dataset input."},{"path":"https://keras3.posit.co/dev/articles/distribution.html","id":"devicemesh-and-tensorlayout","dir":"Articles","previous_headings":"","what":"DeviceMesh and TensorLayout","title":"Distributed training with Keras 3","text":"keras$distribution$DeviceMesh class Keras distribution API represents cluster computational devices configured distributed computation. aligns similar concepts jax.sharding.Mesh tf.dtensor.Mesh, ’s used map physical devices logical mesh structure. TensorLayout class specifies tensors distributed across DeviceMesh, detailing sharding tensors along specified axes correspond names axes DeviceMesh. can find detailed concept explainers TensorFlow DTensor guide.","code":"# Retrieve the local available gpu devices. devices <- jax$devices() # \"gpu\" str(devices) ## List of 8 ##  $ :TFRT_CPU_0 ##  $ :TFRT_CPU_1 ##  $ :TFRT_CPU_2 ##  $ :TFRT_CPU_3 ##  $ :TFRT_CPU_4 ##  $ :TFRT_CPU_5 ##  $ :TFRT_CPU_6 ##  $ :TFRT_CPU_7 # Define a 2x4 device mesh with data and model parallel axes mesh <- keras$distribution$DeviceMesh(   shape = shape(2, 4),   axis_names = list(\"data\", \"model\"),   devices = devices )  # A 2D layout, which describes how a tensor is distributed across the # mesh. The layout can be visualized as a 2D grid with \"model\" as rows and # \"data\" as columns, and it is a [4, 2] grid when it mapped to the physical # devices on the mesh. layout_2d <- keras$distribution$TensorLayout(   axes = c(\"model\", \"data\"),   device_mesh = mesh )  # A 4D layout which could be used for data parallelism of an image input. replicated_layout_4d <- keras$distribution$TensorLayout(   axes = list(\"data\", NULL, NULL, NULL),   device_mesh = mesh )"},{"path":"https://keras3.posit.co/dev/articles/distribution.html","id":"distribution","dir":"Articles","previous_headings":"","what":"Distribution","title":"Distributed training with Keras 3","text":"Distribution class Keras serves foundational abstract class designed developing custom distribution strategies. encapsulates core logic needed distribute model’s variables, input data, intermediate computations across device mesh. end user, won’t interact directly class, subclasses like DataParallel ModelParallel.","code":""},{"path":"https://keras3.posit.co/dev/articles/distribution.html","id":"dataparallel","dir":"Articles","previous_headings":"","what":"DataParallel","title":"Distributed training with Keras 3","text":"DataParallel class Keras distribution API designed data parallelism strategy distributed training, model weights replicated across devices DeviceMesh, device processes portion input data. sample usage class.","code":"# Create DataParallel with list of devices. # As a shortcut, the devices can be skipped, # and Keras will detect all local available devices. # E.g. data_parallel <- DataParallel() data_parallel <- keras$distribution$DataParallel(devices = devices)  # Or you can choose to create DataParallel with a 1D `DeviceMesh`. mesh_1d <- keras$distribution$DeviceMesh(   shape = shape(8),   axis_names = list(\"data\"),   devices = devices ) data_parallel <- keras$distribution$DataParallel(device_mesh = mesh_1d)  inputs <- random_normal(c(128, 28, 28, 1)) labels <- random_normal(c(128, 10)) dataset <- tensor_slices_dataset(c(inputs, labels)) |>   dataset_batch(16)  # Set the global distribution. keras$distribution$set_distribution(data_parallel)  # Note that all the model weights from here on are replicated to # all the devices of the `DeviceMesh`. This includes the RNG # state, optimizer states, metrics, etc. The dataset fed into `model |> fit()` or # `model |> evaluate()` will be split evenly on the batch dimension, and sent to # all the devices. You don't have to do any manual aggregation of losses, # since all the computation happens in a global context. inputs <- keras_input(shape = c(28, 28, 1)) outputs <- inputs |>   layer_flatten() |>   layer_dense(units = 200, use_bias = FALSE, activation = \"relu\") |>   layer_dropout(0.4) |>   layer_dense(units = 10, activation = \"softmax\")  model <- keras_model(inputs = inputs, outputs = outputs)  model |> compile(loss = \"mse\") model |> fit(dataset, epochs = 3) ## Epoch 1/3 ## 8/8 - 0s - 42ms/step - loss: 1.1575 ## Epoch 2/3 ## 8/8 - 0s - 4ms/step - loss: 1.0557 ## Epoch 3/3 ## 8/8 - 0s - 4ms/step - loss: 1.0179 model |> evaluate(dataset) ## 8/8 - 0s - 9ms/step - loss: 0.9707 ## $loss ## [1] 0.9707004"},{"path":"https://keras3.posit.co/dev/articles/distribution.html","id":"modelparallel-and-layoutmap","dir":"Articles","previous_headings":"","what":"ModelParallel and LayoutMap","title":"Distributed training with Keras 3","text":"ModelParallel mostly useful model weights large fit single accelerator. setting allows spit model weights activation tensors across devices DeviceMesh, enable horizontal scaling large models. Unlike DataParallel model weights fully replicated, weights layout ModelParallel usually need customization best performances. introduce LayoutMap let specify TensorLayout weights intermediate tensors global perspective. LayoutMap dict-like object maps string TensorLayout instances. behaves differently normal dict string key treated regex retrieving value. class allows define naming schema TensorLayout retrieve corresponding TensorLayout instance. Typically, key used query variable$path attribute, identifier variable. shortcut, list axis names also allowed inserting value, converted TensorLayout. LayoutMap can also optionally contain DeviceMesh populate TensorLayout$device_mesh set. retrieving layout key, isn’t exact match, existing keys layout map treated regex matched input key . multiple matches, ValueError raised. matches found, NULL returned. can visualize individual weights sharded also easy change mesh structure tune computation data parallel model parallel. can adjusting shape mesh. changes needed code.","code":"mesh_2d <- keras$distribution$DeviceMesh(   shape = shape(2, 4),   axis_names = c(\"data\", \"model\"),   devices = devices ) layout_map  <- keras$distribution$LayoutMap(mesh_2d)  # The rule below means that for any weights that match with d1/kernel, it # will be sharded with model dimensions (4 devices), same for the d1/bias. # All other weights will be fully replicated. layout_map[\"d1/kernel\"] <- tuple(NULL, \"model\") layout_map[\"d1/bias\"] <- tuple(\"model\")  # You can also set the layout for the layer output like layout_map[\"d2/output\"] <- tuple(\"data\", NULL)  model_parallel <- keras$distribution$ModelParallel(   layout_map = layout_map, batch_dim_name = \"data\" )  keras$distribution$set_distribution(model_parallel)  inputs <- layer_input(shape = c(28, 28, 1)) outputs <- inputs |>   layer_flatten() |>   layer_dense(units = 200, use_bias = FALSE,               activation = \"relu\", name = \"d1\") |>   layer_dropout(0.4) |>   layer_dense(units = 10,               activation = \"softmax\",               name = \"d2\")  model <- keras_model(inputs = inputs, outputs = outputs) d1 <- get_layer(model, \"d1\") d1$kernel$value |> jax$debug$visualize_array_sharding() ## ┌───────┬───────┬───────┬───────┐ ## │       │       │       │       │ ## │       │       │       │       │ ## │       │       │       │       │ ## │       │       │       │       │ ## │CPU 0,4│CPU 1,5│CPU 2,6│CPU 3,7│ ## │       │       │       │       │ ## │       │       │       │       │ ## │       │       │       │       │ ## │       │       │       │       │ ## └───────┴───────┴───────┴───────┘ d2 <- get_layer(model, \"d2\") d2$kernel$value |> jax$debug$visualize_array_sharding() ## ┌───────────────────┐ ## │                   │ ## │                   │ ## │                   │ ## │                   │ ## │CPU 0,1,2,3,4,5,6,7│ ## │                   │ ## │                   │ ## │                   │ ## │                   │ ## └───────────────────┘ d2$bias$value |> jax$debug$visualize_array_sharding() ## ┌───────────────────┐ ## │CPU 0,1,2,3,4,5,6,7│ ## └───────────────────┘ x_batch <- dataset |>   as_iterator() |> iter_next() |>   _[[1]] |> op_convert_to_tensor()  output_array <- model(x_batch) output_array |> jax$debug$visualize_array_sharding() ## ┌─────────────┐ ## │             │ ## │ CPU 0,1,2,3 │ ## │             │ ## │             │ ## ├─────────────┤ ## │             │ ## │ CPU 4,5,6,7 │ ## │             │ ## │             │ ## └─────────────┘ # The data will be sharded across the \"data\" dimension of the method, which # has 2 devices. model |> compile(loss = \"mse\") model |> fit(dataset, epochs = 3) ## Epoch 1/3 ## 8/8 - 0s - 48ms/step - loss: 1.1454 ## Epoch 2/3 ## 8/8 - 0s - 2ms/step - loss: 1.0527 ## Epoch 3/3 ## 8/8 - 0s - 2ms/step - loss: 1.0115 model |> evaluate(dataset) ## 8/8 - 0s - 10ms/step - loss: 0.9674 ## $loss ## [1] 0.9673891 full_data_parallel_mesh <- keras$distribution$DeviceMesh(   shape = shape(8, 1),   axis_names = list(\"data\", \"model\"),   devices = devices ) more_data_parallel_mesh <- keras$distribution$DeviceMesh(   shape = shape(4, 2),   axis_names = list(\"data\", \"model\"),   devices = devices ) more_model_parallel_mesh <- keras$distribution$DeviceMesh(   shape = shape(2, 4),   axis_names = list(\"data\", \"model\"),   devices = devices ) full_model_parallel_mesh <- keras$distribution$DeviceMesh(   shape = shape(1, 8),   axis_names = list(\"data\", \"model\"),   devices = devices )"},{"path":"https://keras3.posit.co/dev/articles/distribution.html","id":"further-reading","dir":"Articles","previous_headings":"ModelParallel and LayoutMap","what":"Further reading","title":"Distributed training with Keras 3","text":"JAX Distributed arrays automatic parallelization JAX sharding module TensorFlow Distributed training DTensors TensorFlow DTensor concepts Using DTensors tf.keras","code":""},{"path":"https://keras3.posit.co/dev/articles/examples/nlp/neural_machine_translation_with_transformer.html","id":"introduction","dir":"Articles > Examples > Nlp","previous_headings":"","what":"Introduction","title":"English-to-Spanish translation with a sequence-to-sequence Transformer","text":"example, ’ll build sequence--sequence Transformer model, ’ll train English--Spanish machine translation task. ’ll learn : Vectorize text using layer_text_vectorization(). Implement layer_transformer_encoder(), layer_transformer_decoder(), layer_positional_embedding(). Prepare data training sequence--sequence model. Use trained model generate translations never-seen-input sentences (sequence--sequence inference). code featured adapted book Deep Learning R, Second Edition (chapter 11: Deep learning text). present example fairly barebones, detailed explanations building block works, well theory behind Transformers, recommend reading book.","code":""},{"path":"https://keras3.posit.co/dev/articles/examples/nlp/neural_machine_translation_with_transformer.html","id":"setup","dir":"Articles > Examples > Nlp","previous_headings":"","what":"Setup","title":"English-to-Spanish translation with a sequence-to-sequence Transformer","text":"","code":"library(glue)  library(tensorflow, exclude = c(\"set_random_seed\", \"shape\")) library(keras3)"},{"path":"https://keras3.posit.co/dev/articles/examples/nlp/neural_machine_translation_with_transformer.html","id":"downloading-the-data","dir":"Articles > Examples > Nlp","previous_headings":"","what":"Downloading the data","title":"English-to-Spanish translation with a sequence-to-sequence Transformer","text":"’ll working English--Spanish translation dataset provided Anki. Let’s download :","code":"zip_path <-   \"http://storage.googleapis.com/download.tensorflow.org/data/spa-eng.zip\" |>   get_file(origin = _, extract = TRUE)  text_path <- fs::path(zip_path, \"spa-eng/spa.txt\")"},{"path":"https://keras3.posit.co/dev/articles/examples/nlp/neural_machine_translation_with_transformer.html","id":"parsing-the-data","dir":"Articles > Examples > Nlp","previous_headings":"","what":"Parsing the data","title":"English-to-Spanish translation with a sequence-to-sequence Transformer","text":"line contains English sentence corresponding Spanish sentence. English sentence source sequence Spanish one target sequence. prepend token \"[start]\" append token \"[end]\" Spanish sentence. ’s sentence pairs look like: Now, let’s split sentence pairs training set, validation set, test set.","code":"text_file <- \"spa-eng/spa.txt\" text_pairs <- text_file %>%   readr::read_tsv(col_names = c(\"english\", \"spanish\"),                   col_types = c(\"cc\")) %>%   within(spanish %<>% paste(\"[start]\", ., \"[end]\")) df <- text_pairs[sample(nrow(text_pairs), 5), ] glue::glue_data(df, r\"(   english: {english}   spanish: {spanish} )\") |> cat(sep = \"\\n\\n\") ## english: I'm staying in Italy. ## spanish: [start] Me estoy quedando en Italia. [end] ## ## english: What's so strange about that? ## spanish: [start] ¿Qué es tan extraño acerca de eso? [end] ## ## english: All of the buses are full. ## spanish: [start] Todos los bondis están llenos. [end] ## ## english: Is this where your mother works? ## spanish: [start] ¿Es aquí donde trabaja tu madre? [end] ## ## english: Take precautions. ## spanish: [start] Ten cuidado. [end] random.shuffle(text_pairs) num_val_samples = int(0.15 * len(text_pairs)) num_train_samples = len(text_pairs) - 2 * num_val_samples train_pairs = text_pairs[:num_train_samples] val_pairs = text_pairs[num_train_samples : num_train_samples + num_val_samples] test_pairs = text_pairs[num_train_samples + num_val_samples :]  print(f\"{len(text_pairs)} total pairs\") print(f\"{len(train_pairs)} training pairs\") print(f\"{len(val_pairs)} validation pairs\") print(f\"{len(test_pairs)} test pairs\") num_test_samples <- num_val_samples <-   round(0.15 * nrow(text_pairs)) num_train_samples <- nrow(text_pairs) - num_val_samples - num_test_samples  pair_group <- sample(c(   rep(\"train\", num_train_samples),   rep(\"test\", num_test_samples),   rep(\"val\", num_val_samples) ))  train_pairs <- text_pairs[pair_group == \"train\", ] test_pairs <- text_pairs[pair_group == \"test\", ] val_pairs <- text_pairs[pair_group == \"val\", ] glue(r\"(   {nrow(text_pairs)} total pairs   {nrow(train_pairs)} training pairs   {nrow(val_pairs)} validation pairs   {nrow(test_pairs)} test pairs )\", .transformer = function(text, envir) {   val <- eval(str2lang(text), envir)   prettyNum(val, big.mark = \",\") }) ## 118,493 total pairs ## 82,945 training pairs ## 17,774 validation pairs ## 17,774 test pairs"},{"path":"https://keras3.posit.co/dev/articles/examples/nlp/neural_machine_translation_with_transformer.html","id":"vectorizing-the-text-data","dir":"Articles > Examples > Nlp","previous_headings":"","what":"Vectorizing the text data","title":"English-to-Spanish translation with a sequence-to-sequence Transformer","text":"’ll use two instances layer_text_vectorization() vectorize text data (one English one Spanish), say, turn original strings integer sequences integer represents index word vocabulary. English layer use default string standardization (strip punctuation characters) splitting scheme (split whitespace), Spanish layer use custom standardization, add character \"¿\" set punctuation characters stripped. Note: production-grade machine translation model, recommend stripping punctuation characters either language. Instead, recommend turning punctuation character token, achieve providing custom split function layer_text_vectorization(). Next, ’ll format datasets. training step, model seek predict target words N+1 (beyond) using source sentence target words 1 N. , training dataset yield tuple (inputs, targets), : inputs dictionary (named list) keys (names) encoder_inputs decoder_inputs. encoder_inputs vectorized source sentence decoder_inputs target sentence “far”, say, words 0 N used predict word N+1 (beyond) target sentence. target target sentence offset one step: provides next words target sentence – model try predict. Let’s take quick look sequence shapes (batches 64 pairs, sequences 20 steps long):","code":"punctuation_regex <- \"[¡¿]|[^[:^punct:][\\\\]]\" # the regex explained: Match ¡, or ¿, or any punctuation character except ] # # [:^punct:]: is a negated POSIX character class. # [:punct:] matches any punctuation character, so [:^punct:] matches any # character that is not a punctuation character. # [^...] negates the whole character class # So [^[:^punct:]] would matche any character that is a punctuation character. # Putting this all together, [^[:^punct:][\\\\]] matches any # punctuation character except the ] character.  custom_standardization <- function(input_string) {   input_string %>%     tf$strings$lower() %>%     tf$strings$regex_replace(punctuation_regex, \"\") }  input_string <- as_tensor(\"[start] ¡corre! [end]\") custom_standardization(input_string) ## tf.Tensor(b'[start] corre [end]', shape=(), dtype=string) vocab_size <- 15000 sequence_length <- 20  # rename to eng_vectorization eng_vectorization <- layer_text_vectorization(   max_tokens = vocab_size,   output_mode = \"int\",   output_sequence_length = sequence_length )  spa_vectorization <- layer_text_vectorization(   max_tokens = vocab_size,   output_mode = \"int\",   output_sequence_length = sequence_length + 1,   standardize = custom_standardization )  adapt(eng_vectorization, train_pairs$english) adapt(spa_vectorization, train_pairs$spanish) format_pair <- function(pair) {   eng <- pair$english |> eng_vectorization()   spa <- pair$spanish |> spa_vectorization()    spa_feature <- spa@r[NA:-2]                                                   # <1>   spa_target <- spa@r[2:NA]                                                     # <2>    features <- list(encoder_inputs = eng, decoder_inputs = spa_feature)   labels <- spa_target   sample_weight <- labels != 0    tuple(features, labels, sample_weight) }  batch_size <- 64  library(tfdatasets, exclude = \"shape\") make_dataset <- function(pairs) {   tensor_slices_dataset(pairs) |>     dataset_map(format_pair, num_parallel_calls = 4) |>     dataset_cache() |>     dataset_shuffle(2048) |>     dataset_batch(batch_size) |>     dataset_prefetch(16) }  train_ds <- make_dataset(train_pairs) val_ds <- make_dataset(val_pairs) c(inputs, targets, weights) %<-% iter_next(as_iterator(train_ds)) str(inputs) ## List of 2 ##  $ encoder_inputs:<tf.Tensor: shape=(64, 20), dtype=int64, numpy=…> ##  $ decoder_inputs:<tf.Tensor: shape=(64, 20), dtype=int64, numpy=…> str(targets) ## <tf.Tensor: shape=(64, 20), dtype=int64, numpy=…>"},{"path":"https://keras3.posit.co/dev/articles/examples/nlp/neural_machine_translation_with_transformer.html","id":"building-the-model","dir":"Articles > Examples > Nlp","previous_headings":"","what":"Building the model","title":"English-to-Spanish translation with a sequence-to-sequence Transformer","text":"sequence--sequence Transformer consists TransformerEncoder TransformerDecoder chained together. make model aware word order, also use PositionalEmbedding layer. source sequence pass TransformerEncoder, produce new representation . new representation passed TransformerDecoder, together target sequence far (target words 1 N). TransformerDecoder seek predict next words target sequence (N+1 beyond). key detail makes possible causal masking (see method get_causal_attention_mask() TransformerDecoder). TransformerDecoder sees entire sequences , thus must make sure uses information target tokens 0 N predicting token N+1 (otherwise, use information future, result model used inference time). Next, assemble end--end model.","code":"layer_transformer_encoder <- Layer(   classname = \"TransformerEncoder\",   initialize = function(embed_dim, dense_dim, num_heads, ...) {     super$initialize(...)     self$embed_dim <- embed_dim     self$dense_dim <- dense_dim     self$num_heads <- num_heads     self$attention <-       layer_multi_head_attention(num_heads = num_heads,                                  key_dim = embed_dim)      self$dense_proj <- keras_model_sequential() %>%       layer_dense(dense_dim, activation = \"relu\") %>%       layer_dense(embed_dim)      self$layernorm_1 <- layer_layer_normalization()     self$layernorm_2 <- layer_layer_normalization()     self$supports_masking <- TRUE   },    call = function(inputs, mask = NULL) {     if (!is.null(mask))       mask <- mask[, NULL, ] |> op_cast(\"int32\")      inputs %>%       { self$attention(., ., attention_mask = mask) + . } %>%       self$layernorm_1() %>%       { self$dense_proj(.) + . } %>%       self$layernorm_2()   },    get_config = function() {     config <- super$get_config()     for(name in c(\"embed_dim\", \"num_heads\", \"dense_dim\"))       config[[name]] <- self[[name]]     config   } )  layer_transformer_decoder <- Layer(   classname = \"TransformerDecoder\",    initialize = function(embed_dim, latent_dim, num_heads, ...) {     super$initialize(...)     self$embed_dim <- embed_dim     self$latent_dim <- latent_dim     self$num_heads <- num_heads     self$attention_1 <- layer_multi_head_attention(num_heads = num_heads,                                                    key_dim = embed_dim)     self$attention_2 <- layer_multi_head_attention(num_heads = num_heads,                                                    key_dim = embed_dim)     self$dense_proj <- keras_model_sequential() %>%       layer_dense(latent_dim, activation = \"relu\") %>%       layer_dense(embed_dim)      self$layernorm_1 <- layer_layer_normalization()     self$layernorm_2 <- layer_layer_normalization()     self$layernorm_3 <- layer_layer_normalization()     self$supports_masking <- TRUE   },    get_config = function() {     config <- super$get_config()     for (name in c(\"embed_dim\", \"num_heads\", \"latent_dim\"))       config[[name]] <- self[[name]]     config   },     get_causal_attention_mask = function(inputs) {     c(batch_size, sequence_length, encoding_length) %<-% op_shape(inputs)      x <- op_arange(0L, sequence_length, include_end = FALSE)     i <- x[, NULL]     j <- x[NULL, ]     mask <- op_cast(i >= j, \"int32\")      repeats <- op_stack(c(batch_size, 1L, 1L))     op_tile(mask[NULL, , ], repeats)   },    call = function(inputs, mask = NULL) {     c(inputs_seq, encoder_outputs) %<-% inputs     causal_mask <- self$get_causal_attention_mask(inputs_seq)      if (is.null(mask)) {       inputs_padding_mask <- NULL       encoder_outputs_padding_mask <- NULL     } else {       c(inputs_padding_mask, encoder_outputs_padding_mask) %<-% mask     }      attention_output_1 <- self$attention_1(       query = inputs_seq,       value = inputs_seq,       key = inputs_seq,       attention_mask = causal_mask,       query_mask = inputs_padding_mask     )     out_1 <- self$layernorm_1(inputs_seq + attention_output_1)      attention_output_2 <- self$attention_2(       query = out_1,       value = encoder_outputs,       key = encoder_outputs,       query_mask = inputs_padding_mask,       key_mask = encoder_outputs_padding_mask     )     out_2 <- self$layernorm_2(out_1 + attention_output_2)      self$layernorm_3(out_2 + self$dense_proj(out_2))   } )  layer_positional_embedding <- Layer(   classname = \"PositionalEmbedding\",    initialize = function(sequence_length, vocab_size, embed_dim, ...) {     super$initialize(...)     self$token_embeddings <- layer_embedding(       input_dim = vocab_size, output_dim = embed_dim     )     self$position_embeddings <- layer_embedding(       input_dim = sequence_length, output_dim = embed_dim     )     self$sequence_length <- sequence_length     self$vocab_size <- vocab_size     self$embed_dim <- embed_dim   },    call = function(inputs) {     c(., len) %<-% op_shape(inputs) # (batch_size, seq_len)     positions <- op_arange(0, len, dtype = \"int32\", include_end = FALSE)     embedded_tokens <- self$token_embeddings(inputs)     embedded_positions <- self$position_embeddings(positions)     embedded_tokens + embedded_positions   },    compute_mask = function(inputs, mask = NULL) {     inputs != 0L   },    get_config = function() {     config <- super$get_config()     for(name in c(\"sequence_length\", \"vocab_size\", \"embed_dim\"))       config[[name]] <- self[[name]]     config   } ) embed_dim <- 256 latent_dim <- 2048 num_heads <- 8  encoder_inputs <- layer_input(shape(NA), dtype = \"int64\",                               name = \"encoder_inputs\") encoder_outputs <- encoder_inputs %>%   layer_positional_embedding(sequence_length, vocab_size, embed_dim) %>%   layer_transformer_encoder(embed_dim, latent_dim, num_heads)  encoder <- keras_model(encoder_inputs, encoder_outputs)  decoder_inputs <-  layer_input(shape(NA), dtype = \"int64\",                                name = \"decoder_inputs\") encoded_seq_inputs <- layer_input(shape(NA, embed_dim),                                   name = \"decoder_state_inputs\")  transformer_decoder <- layer_transformer_decoder(NULL,   embed_dim, latent_dim, num_heads)  decoder_outputs <- decoder_inputs %>%   layer_positional_embedding(sequence_length, vocab_size, embed_dim) %>%   { transformer_decoder(list(., encoded_seq_inputs)) } %>%   layer_dropout(0.5) %>%   layer_dense(vocab_size, activation=\"softmax\")  decoder <- keras_model(inputs = list(decoder_inputs, encoded_seq_inputs),                        outputs = decoder_outputs)  decoder_outputs <- decoder(list(decoder_inputs, encoder_outputs))  transformer <- keras_model(   inputs = list(encoder_inputs = encoder_inputs, decoder_inputs = decoder_inputs),   outputs = decoder_outputs,   name = \"transformer\" )"},{"path":"https://keras3.posit.co/dev/articles/examples/nlp/neural_machine_translation_with_transformer.html","id":"training-our-model","dir":"Articles > Examples > Nlp","previous_headings":"","what":"Training our model","title":"English-to-Spanish translation with a sequence-to-sequence Transformer","text":"’ll use accuracy quick way monitor training progress validation data. Note machine translation typically uses BLEU scores well metrics, rather accuracy. train 1 epoch, get model actually converge train least 30 epochs.","code":"epochs <- 1  # This should be at least 30 for convergence  transformer ## Model: \"transformer\" ## ┏━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━┓ ## ┃ Layer (type)        ┃ Output Shape      ┃    Param # ┃ Connected to      ┃ ## ┡━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━┩ ## │ encoder_inputs      │ (None, None)      │          0 │ -                 │ ## │ (InputLayer)        │                   │            │                   │ ## ├─────────────────────┼───────────────────┼────────────┼───────────────────┤ ## │ positional_embeddi… │ (None, None, 256) │  3,845,120 │ encoder_inputs[0… │ ## │ (PositionalEmbeddi… │                   │            │                   │ ## ├─────────────────────┼───────────────────┼────────────┼───────────────────┤ ## │ not_equal           │ (None, None)      │          0 │ encoder_inputs[0… │ ## │ (NotEqual)          │                   │            │                   │ ## ├─────────────────────┼───────────────────┼────────────┼───────────────────┤ ## │ decoder_inputs      │ (None, None)      │          0 │ -                 │ ## │ (InputLayer)        │                   │            │                   │ ## ├─────────────────────┼───────────────────┼────────────┼───────────────────┤ ## │ transformer_encoder │ (None, None, 256) │  3,155,456 │ positional_embed… │ ## │ (TransformerEncode… │                   │            │ not_equal[0][0]   │ ## ├─────────────────────┼───────────────────┼────────────┼───────────────────┤ ## │ functional_3        │ (None, None,      │ 12,959,640 │ decoder_inputs[0… │ ## │ (Functional)        │ 15000)            │            │ transformer_enco… │ ## │                     │                   │            │ not_equal[0][0]   │ ## └─────────────────────┴───────────────────┴────────────┴───────────────────┘ ##  Total params: 19,960,216 (76.14 MB) ##  Trainable params: 19,960,216 (76.14 MB) ##  Non-trainable params: 0 (0.00 B) transformer |> compile(   \"rmsprop\",   loss = loss_sparse_categorical_crossentropy(ignore_class = 0),   metrics = \"accuracy\" )  transformer |> fit(train_ds, epochs = epochs,                    validation_data = val_ds) ## 1297/1297 - 55s - 42ms/step - accuracy: 0.1592 - loss: 3.8833 - val_accuracy: 0.1691 - val_loss: 3.1841"},{"path":"https://keras3.posit.co/dev/articles/examples/nlp/neural_machine_translation_with_transformer.html","id":"decoding-test-sentences","dir":"Articles > Examples > Nlp","previous_headings":"","what":"Decoding test sentences","title":"English-to-Spanish translation with a sequence-to-sequence Transformer","text":"Finally, let’s demonstrate translate brand new English sentences. simply feed model vectorized English sentence well target token \"[start]\", repeatedly generated next token, hit token \"[end]\". 30 epochs, get results :","code":"spa_vocab <- spa_vectorization |> get_vocabulary() max_decoded_sentence_length <- 20 tf_decode_sequence <- tf_function(function(input_sentence) {   withr::local_options(tensorflow.extract.style = \"python\")    tokenized_input_sentence <- input_sentence %>%     as_tensor(shape = c(1, 1)) %>%     eng_vectorization()   spa_vocab <- as_tensor(spa_vocab)   decoded_sentence <- as_tensor(\"[start]\", shape = c(1, 1))    for (i in tf$range(as.integer(max_decoded_sentence_length))) {      tokenized_target_sentence <-       spa_vectorization(decoded_sentence)[,NA:-1]      next_token_predictions <-       transformer(list(         encoder_inputs = tokenized_input_sentence,         decoder_inputs = tokenized_target_sentence       ))      sampled_token_index <- tf$argmax(next_token_predictions[0, i, ])     sampled_token <- spa_vocab[sampled_token_index]     decoded_sentence <-       tf$strings$join(c(decoded_sentence, sampled_token),                       separator = \" \")      if (sampled_token == \"[end]\")       break   }    decoded_sentence  })  for (i in seq(20)) {      c(input_sentence, correct_translation) %<-%       test_pairs[sample.int(nrow(test_pairs), 1), ]     cat(\"-\\n\")     cat(\"English:\", input_sentence, \"\\n\")     cat(\"Correct Translation:\", tolower(correct_translation), \"\\n\")     cat(\"  Model Translation:\", input_sentence %>% as_tensor() %>%           tf_decode_sequence() %>% as.character(), \"\\n\") } English: I'm sure everything will be fine. Correct Translation: [start] estoy segura de que todo irá bien. [end]   Model Translation: [start] estoy seguro de que todo va bien [end]"},{"path":"https://keras3.posit.co/dev/articles/examples/nlp/text_classification_from_scratch.html","id":"introduction","dir":"Articles > Examples > Nlp","previous_headings":"","what":"Introduction","title":"Text classification from scratch","text":"example shows text classification starting raw text (set text files disk). demonstrate workflow IMDB sentiment classification dataset (unprocessed version). use layer_text_vectorization() word splitting & indexing.","code":""},{"path":"https://keras3.posit.co/dev/articles/examples/nlp/text_classification_from_scratch.html","id":"setup","dir":"Articles > Examples > Nlp","previous_headings":"","what":"Setup","title":"Text classification from scratch","text":"","code":"library(tensorflow, exclude = c(\"shape\", \"set_random_seed\")) library(tfdatasets, exclude = \"shape\") library(keras3)"},{"path":"https://keras3.posit.co/dev/articles/examples/nlp/text_classification_from_scratch.html","id":"load-the-data-imdb-movie-review-sentiment-classification","dir":"Articles > Examples > Nlp","previous_headings":"","what":"Load the data: IMDB movie review sentiment classification","title":"Text classification from scratch","text":"Let’s download data inspect structure. aclImdb folder contains train test subfolder: aclImdb/train/pos aclImdb/train/neg folders contain text files, represents one review (either positive negative): interested pos neg subfolders, let’s delete subfolder text files : can use utility text_dataset_from_directory() generate labeled tf_dataset object set text files disk filed class-specific folders. Let’s use generate training, validation, test datasets. validation training datasets generated two subsets train directory, 20% samples going validation dataset 80% going training dataset. validation dataset addition test dataset useful tuning hyperparameters, model architecture, test dataset used. putting model real world however, retrained using available training data (without creating validation dataset), performance maximized. using validation_split subset arguments, make sure either specify random seed, pass shuffle=FALSE, validation & training splits get overlap. Let’s preview samples:","code":"if (!dir.exists(\"datasets/aclImdb\")) {   dir.create(\"datasets\")   download.file(     \"https://ai.stanford.edu/~amaas/data/sentiment/aclImdb_v1.tar.gz\",     \"datasets/aclImdb_v1.tar.gz\"   )   untar(\"datasets/aclImdb_v1.tar.gz\", exdir = \"datasets\")   unlink(\"datasets/aclImdb/train/unsup\", recursive = TRUE) } head(list.files(\"datasets/aclImdb/test\")) ## [1] \"labeledBow.feat\" \"neg\"             \"pos\"             \"urls_neg.txt\" ## [5] \"urls_pos.txt\" head(list.files(\"datasets/aclImdb/train\")) ## [1] \"labeledBow.feat\" \"neg\"             \"pos\"             \"unsupBow.feat\" ## [5] \"urls_neg.txt\"    \"urls_pos.txt\" writeLines(strwrap(readLines(\"datasets/aclImdb/train/pos/4229_10.txt\"))) ## Don't waste time reading my review. Go out and see this ## astonishingly good episode, which may very well be the best Columbo ## ever written! Ruth Gordon is perfectly cast as the scheming yet ## charming mystery writer who murders her son-in-law to avenge his ## murder of her daughter. Columbo is his usual rumpled, befuddled and ## far-cleverer-than-he-seems self, and this particular installment ## features fantastic chemistry between Gordon and Falk. Ironically, ## this was not written by heralded creators Levinson or Link yet is ## possibly the densest, most thoroughly original and twist-laden ## Columbo plot ever. Utterly satisfying in nearly every department ## and overflowing with droll and witty dialogue and thinking. Truly ## unexpected and inventive climax tops all. 10/10...seek this one out ## on Netflix! unlink(\"datasets/aclImdb/train/unsup\", recursive = TRUE) batch_size <- 32  raw_train_ds <- text_dataset_from_directory(   \"datasets/aclImdb/train\",   batch_size = batch_size,   validation_split = 0.2,   subset = \"training\",   seed = 1337 ) ## Found 25000 files belonging to 2 classes. ## Using 20000 files for training. raw_val_ds <- text_dataset_from_directory(   \"datasets/aclImdb/train\",   batch_size = batch_size,   validation_split = 0.2,   subset = \"validation\",   seed = 1337 ) ## Found 25000 files belonging to 2 classes. ## Using 5000 files for validation. raw_test_ds <- text_dataset_from_directory(   \"datasets/aclImdb/test\",   batch_size = batch_size ) ## Found 25000 files belonging to 2 classes. cat(\"Number of batches in raw_train_ds:\", length(raw_train_ds), \"\\n\") ## Number of batches in raw_train_ds: 625 cat(\"Number of batches in raw_val_ds:\", length(raw_val_ds), \"\\n\") ## Number of batches in raw_val_ds: 157 cat(\"Number of batches in raw_test_ds:\", length(raw_test_ds), \"\\n\") ## Number of batches in raw_test_ds: 782 # It's important to take a look at your raw data to ensure your normalization # and tokenization will work as expected. We can do that by taking a few # examples from the training set and looking at them. # This is one of the places where eager execution shines: # we can just evaluate these tensors using .numpy() # instead of needing to evaluate them in a Session/Graph context. batch <- iter_next(as_iterator(raw_train_ds)) str(batch) ## List of 2 ##  $ :<tf.Tensor: shape=(32), dtype=string, numpy=…> ##  $ :<tf.Tensor: shape=(32), dtype=int32, numpy=…> c(text_batch, label_batch) %<-% batch for (i in 1:3) {   print(text_batch[i])   print(label_batch[i]) } ## tf.Tensor(b\"I have read the novel Reaper of Ben Mezrich a fews years ago and last night I accidentally came to see this adaption.<br /><br />Although it's been years since I read the story the first time, the differences between the novel and the movie are humongous. Very important elements, which made the whole thing plausible are just written out or changed to bad.<br /><br />If the plot sounds interesting to you: go and get the novel. Its much, much, much better.<br /><br />Still 4 out of 10 since it was hard to stop watching because of the great basic plot by Ben Mezrich.\", shape=(), dtype=string) ## tf.Tensor(0, shape=(), dtype=int32) ## tf.Tensor(b'After seeing all the Jesse James, Quantrill, jayhawkers,etc films in the fifties, it is quite a thrill to see this film with a new perspective by director Ang Lee. The scene of the attack of Lawrence, Kansas is awesome. The romantic relationship between Jewel and Toby Mcguire turns out to be one of the best parts and Jonathan Rhys-Meyers is outstanding as the bad guy. All the time this film makes you feel the horror of war, and the desperate situation of the main characters who do not know if they are going to survive the next hours. Definitely worth seeing.', shape=(), dtype=string) ## tf.Tensor(1, shape=(), dtype=int32) ## tf.Tensor(b'AG was an excellent presentation of drama, suspense and thriller that is so rare to American TV. Sheriff Lucas gave many a viewer the willies. We rooted for Caleb as he strove to resist the overtures of Sheriff Lucas. We became engrossed and fearful upon learning of the unthinkable connection between these two characters. The manipulations which weekly gave cause to fear what Lucas would do next were truly surprising. This show lived up to the \"Gothic\" moniker in ways American entertainment has so seldom attempted, much less mastered. The suits definitely made a big mistake in not supporting this show. This show puts shame to the current glut of \"reality\" shows- which are so less than satisfying viewing.The call for a DVD box set is well based. This show is quality viewing for a discerning market hungry for quality viewing. A public that is tiring of over-saturation of mind-numbing reality fare will welcome this gem of real storytelling. Bring on the DVD box set!!', shape=(), dtype=string) ## tf.Tensor(1, shape=(), dtype=int32)"},{"path":"https://keras3.posit.co/dev/articles/examples/nlp/text_classification_from_scratch.html","id":"prepare-the-data","dir":"Articles > Examples > Nlp","previous_headings":"","what":"Prepare the data","title":"Text classification from scratch","text":"particular, remove <br /> tags.","code":"# Having looked at our data above, we see that the raw text contains HTML break # tags of the form '<br />'. These tags will not be removed by the default # standardizer (which doesn't strip HTML). Because of this, we will need to # create a custom standardization function. custom_standardization_fn <- function(string_tensor) {   string_tensor |>     tf$strings$lower() |> # convert to all lowercase     tf$strings$regex_replace(\"<br />\", \" \") |> # remove '<br />' HTML tag     tf$strings$regex_replace(\"[[:punct:]]\", \"\") # remove punctuation }   # Model constants. max_features <- 20000 embedding_dim <- 128 sequence_length <- 500  # Now that we have our custom standardization, we can instantiate our text # vectorization layer. We are using this layer to normalize, split, and map # strings to integers, so we set our 'output_mode' to 'int'. # Note that we're using the default split function, # and the custom standardization defined above. # We also set an explicit maximum sequence length, since the CNNs later in our # model won't support ragged sequences. vectorize_layer <- layer_text_vectorization(   standardize = custom_standardization_fn,   max_tokens = max_features,   output_mode = \"int\",   output_sequence_length = sequence_length, )  # Now that the vectorize_layer has been created, call `adapt` on a text-only # dataset to create the vocabulary. You don't have to batch, but for very large # datasets this means you're not keeping spare copies of the dataset in memory.  # Let's make a text-only dataset (no labels): text_ds <- raw_train_ds |>   dataset_map(\\(x, y) x) # Let's call `adapt`: vectorize_layer |> adapt(text_ds)"},{"path":"https://keras3.posit.co/dev/articles/examples/nlp/text_classification_from_scratch.html","id":"two-options-to-vectorize-the-data","dir":"Articles > Examples > Nlp","previous_headings":"","what":"Two options to vectorize the data","title":"Text classification from scratch","text":"2 ways can use text vectorization layer: Option 1: Make part model, obtain model processes raw strings, like : Option 2: Apply text dataset obtain dataset word indices, feed model expects integer sequences inputs. important difference two option 2 enables asynchronous CPU processing buffering data training GPU. ’re training model GPU, probably want go option get best performance. . export model production, ’d ship model accepts raw strings input, like code snippet option 1 . can done training. last section.","code":"text_input <- keras_input(shape = c(1L), dtype = \"string\", name = 'text') x <- text_input |>   vectorize_layer() |>   layer_embedding(max_features + 1, embedding_dim) vectorize_text <- function(text, label) {   text <- text |>     op_expand_dims(-1) |>     vectorize_layer()   list(text, label) }  # Vectorize the data. train_ds <- raw_train_ds |> dataset_map(vectorize_text) val_ds   <- raw_val_ds   |> dataset_map(vectorize_text) test_ds  <- raw_test_ds  |> dataset_map(vectorize_text)  # Do async prefetching / buffering of the data for best performance on GPU. train_ds <- train_ds |>   dataset_cache() |>   dataset_prefetch(buffer_size = 10) val_ds <- val_ds |>   dataset_cache() |>   dataset_prefetch(buffer_size = 10) test_ds <- test_ds |>   dataset_cache() |>   dataset_prefetch(buffer_size = 10)"},{"path":"https://keras3.posit.co/dev/articles/examples/nlp/text_classification_from_scratch.html","id":"build-a-model","dir":"Articles > Examples > Nlp","previous_headings":"","what":"Build a model","title":"Text classification from scratch","text":"choose simple 1D convnet starting Embedding layer.","code":"# A integer input for vocab indices. inputs <- keras_input(shape = c(NA), dtype = \"int64\")  predictions <- inputs |>   # Next, we add a layer to map those vocab indices into a space of dimensionality   # 'embedding_dim'.   layer_embedding(max_features, embedding_dim) |>   layer_dropout(0.5) |>   # Conv1D + global max pooling   layer_conv_1d(128, 7, padding = \"valid\", activation = \"relu\", strides = 3) |>   layer_conv_1d(128, 7, padding = \"valid\", activation = \"relu\", strides = 3) |>   layer_global_max_pooling_1d() |>   # We add a vanilla hidden layer:   layer_dense(128, activation = \"relu\") |>   layer_dropout(0.5) |>   # We project onto a single unit output layer, and squash it with a sigmoid:   layer_dense(1, activation = \"sigmoid\", name = \"predictions\")  model <- keras_model(inputs, predictions)  summary(model) ## Model: \"functional\" ## ┏━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━┓ ## ┃ Layer (type)                    ┃ Output Shape           ┃       Param # ┃ ## ┡━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━┩ ## │ input_layer (InputLayer)        │ (None, None)           │             0 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ embedding_1 (Embedding)         │ (None, None, 128)      │     2,560,000 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ dropout (Dropout)               │ (None, None, 128)      │             0 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ conv1d (Conv1D)                 │ (None, None, 128)      │       114,816 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ conv1d_1 (Conv1D)               │ (None, None, 128)      │       114,816 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ global_max_pooling1d            │ (None, 128)            │             0 │ ## │ (GlobalMaxPooling1D)            │                        │               │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ dense (Dense)                   │ (None, 128)            │        16,512 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ dropout_1 (Dropout)             │ (None, 128)            │             0 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ predictions (Dense)             │ (None, 1)              │           129 │ ## └─────────────────────────────────┴────────────────────────┴───────────────┘ ##  Total params: 2,806,273 (10.71 MB) ##  Trainable params: 2,806,273 (10.71 MB) ##  Non-trainable params: 0 (0.00 B) # Compile the model with binary crossentropy loss and an adam optimizer. model |> compile(loss = \"binary_crossentropy\",                  optimizer = \"adam\",                  metrics = \"accuracy\")"},{"path":"https://keras3.posit.co/dev/articles/examples/nlp/text_classification_from_scratch.html","id":"train-the-model","dir":"Articles > Examples > Nlp","previous_headings":"","what":"Train the model","title":"Text classification from scratch","text":"","code":"epochs <- 3  # Fit the model using the train and test datasets. model |> fit(train_ds, validation_data = val_ds, epochs = epochs) ## Epoch 1/3 ## 625/625 - 5s - 8ms/step - accuracy: 0.6964 - loss: 0.5261 - val_accuracy: 0.8668 - val_loss: 0.3185 ## Epoch 2/3 ## 625/625 - 2s - 3ms/step - accuracy: 0.9074 - loss: 0.2365 - val_accuracy: 0.8754 - val_loss: 0.3245 ## Epoch 3/3 ## 625/625 - 2s - 3ms/step - accuracy: 0.9551 - loss: 0.1220 - val_accuracy: 0.8284 - val_loss: 0.4862"},{"path":"https://keras3.posit.co/dev/articles/examples/nlp/text_classification_from_scratch.html","id":"evaluate-the-model-on-the-test-set","dir":"Articles > Examples > Nlp","previous_headings":"","what":"Evaluate the model on the test set","title":"Text classification from scratch","text":"","code":"model |> evaluate(test_ds) ## 782/782 - 1s - 2ms/step - accuracy: 0.8115 - loss: 0.5299 ## $accuracy ## [1] 0.81152 ## ## $loss ## [1] 0.5299004"},{"path":"https://keras3.posit.co/dev/articles/examples/nlp/text_classification_from_scratch.html","id":"make-an-end-to-end-model","dir":"Articles > Examples > Nlp","previous_headings":"","what":"Make an end-to-end model","title":"Text classification from scratch","text":"want obtain model capable processing raw strings, can simply create new model (using weights just trained):","code":"# A string input inputs <- keras_input(shape = c(1), dtype = \"string\") # Turn strings into vocab indices indices <- vectorize_layer(inputs) # Turn vocab indices into predictions outputs <- model(indices)  # Our end to end model end_to_end_model <- keras_model(inputs, outputs) end_to_end_model |> compile(   loss = \"binary_crossentropy\",   optimizer = \"adam\",   metrics = c(\"accuracy\") )  # Test it with `raw_test_ds`, which yields raw strings end_to_end_model |> evaluate(raw_test_ds) ## 782/782 - 3s - 4ms/step - accuracy: 0.8115 - loss: 0.5299 ## $accuracy ## [1] 0.81152 ## ## $loss ## [1] 0.5299006"},{"path":"https://keras3.posit.co/dev/articles/examples/structured_data/imbalanced_classification.html","id":"introduction","dir":"Articles > Examples > Structured_data","previous_headings":"","what":"Introduction","title":"Imbalanced classification: credit card fraud detection","text":"example looks Kaggle Credit Card Fraud Detection dataset demonstrate train classification model data highly imbalanced classes. can download data clicking “Download” link, ’re setup kaggle API key \"~/.kaggle/kagle.json\", can run following:","code":"reticulate::py_install(\"kaggle\", pip = TRUE) reticulate::py_available(TRUE) # ensure 'kaggle' is on the PATH system(\"kaggle datasets download -d mlg-ulb/creditcardfraud\") zip::unzip(\"creditcardfraud.zip\", files = \"creditcard.csv\")"},{"path":"https://keras3.posit.co/dev/articles/examples/structured_data/imbalanced_classification.html","id":"first-load-the-data","dir":"Articles > Examples > Structured_data","previous_headings":"","what":"First, load the data","title":"Imbalanced classification: credit card fraud detection","text":"","code":"library(readr) df <- read_csv(\"creditcard.csv\", col_types = cols(   Class = col_integer(),   .default = col_double() )) tibble::glimpse(df) ## Rows: 284,807 ## Columns: 31 ## $ Time   <dbl> 0, 0, 1, 1, 2, 2, 4, 7, 7, 9, 10, 10, 10, 11, 12, 12, 12, 1… ## $ V1     <dbl> -1.3598071, 1.1918571, -1.3583541, -0.9662717, -1.1582331, … ## $ V2     <dbl> -0.07278117, 0.26615071, -1.34016307, -0.18522601, 0.877736… ## $ V3     <dbl> 2.53634674, 0.16648011, 1.77320934, 1.79299334, 1.54871785,… ## $ V4     <dbl> 1.37815522, 0.44815408, 0.37977959, -0.86329128, 0.40303393… ## $ V5     <dbl> -0.33832077, 0.06001765, -0.50319813, -0.01030888, -0.40719… ## $ V6     <dbl> 0.46238778, -0.08236081, 1.80049938, 1.24720317, 0.09592146… ## $ V7     <dbl> 0.239598554, -0.078802983, 0.791460956, 0.237608940, 0.5929… ## $ V8     <dbl> 0.098697901, 0.085101655, 0.247675787, 0.377435875, -0.2705… ## $ V9     <dbl> 0.3637870, -0.2554251, -1.5146543, -1.3870241, 0.8177393, -… ## $ V10    <dbl> 0.09079417, -0.16697441, 0.20764287, -0.05495192, 0.7530744… ## $ V11    <dbl> -0.55159953, 1.61272666, 0.62450146, -0.22648726, -0.822842… ## $ V12    <dbl> -0.61780086, 1.06523531, 0.06608369, 0.17822823, 0.53819555… ## $ V13    <dbl> -0.99138985, 0.48909502, 0.71729273, 0.50775687, 1.34585159… ## $ V14    <dbl> -0.31116935, -0.14377230, -0.16594592, -0.28792375, -1.1196… ## $ V15    <dbl> 1.468176972, 0.635558093, 2.345864949, -0.631418118, 0.1751… ## $ V16    <dbl> -0.47040053, 0.46391704, -2.89008319, -1.05964725, -0.45144… ## $ V17    <dbl> 0.207971242, -0.114804663, 1.109969379, -0.684092786, -0.23… ## $ V18    <dbl> 0.02579058, -0.18336127, -0.12135931, 1.96577500, -0.038194… ## $ V19    <dbl> 0.40399296, -0.14578304, -2.26185710, -1.23262197, 0.803486… ## $ V20    <dbl> 0.25141210, -0.06908314, 0.52497973, -0.20803778, 0.4085423… ## $ V21    <dbl> -0.018306778, -0.225775248, 0.247998153, -0.108300452, -0.0… ## $ V22    <dbl> 0.277837576, -0.638671953, 0.771679402, 0.005273597, 0.7982… ## $ V23    <dbl> -0.110473910, 0.101288021, 0.909412262, -0.190320519, -0.13… ## $ V24    <dbl> 0.06692807, -0.33984648, -0.68928096, -1.17557533, 0.141266… ## $ V25    <dbl> 0.12853936, 0.16717040, -0.32764183, 0.64737603, -0.2060095… ## $ V26    <dbl> -0.18911484, 0.12589453, -0.13909657, -0.22192884, 0.502292… ## $ V27    <dbl> 0.133558377, -0.008983099, -0.055352794, 0.062722849, 0.219… ## $ V28    <dbl> -0.021053053, 0.014724169, -0.059751841, 0.061457629, 0.215… ## $ Amount <dbl> 149.62, 2.69, 378.66, 123.50, 69.99, 3.67, 4.99, 40.80, 93.… ## $ Class  <int> 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,…"},{"path":"https://keras3.posit.co/dev/articles/examples/structured_data/imbalanced_classification.html","id":"prepare-a-validation-set","dir":"Articles > Examples > Structured_data","previous_headings":"","what":"Prepare a validation set","title":"Imbalanced classification: credit card fraud detection","text":"","code":"val_idx <- nrow(df) %>% sample.int(., round( . * 0.2)) val_df <- df[val_idx, ] train_df <- df[-val_idx, ]  cat(\"Number of training samples:\", nrow(train_df), \"\\n\") ## Number of training samples: 227846 cat(\"Number of validation samples:\", nrow(val_df), \"\\n\") ## Number of validation samples: 56961"},{"path":"https://keras3.posit.co/dev/articles/examples/structured_data/imbalanced_classification.html","id":"analyze-class-imbalance-in-the-targets","dir":"Articles > Examples > Structured_data","previous_headings":"","what":"Analyze class imbalance in the targets","title":"Imbalanced classification: credit card fraud detection","text":"","code":"counts <- table(train_df$Class) counts ## ##      0      1 ## 227464    382 cat(sprintf(\"Number of positive samples in training data: %i (%.2f%% of total)\",             counts[\"1\"], 100 * counts[\"1\"] / sum(counts))) ## Number of positive samples in training data: 382 (0.17% of total) weight_for_0 = 1 / counts[\"0\"] weight_for_1 = 1 / counts[\"1\"]"},{"path":"https://keras3.posit.co/dev/articles/examples/structured_data/imbalanced_classification.html","id":"normalize-the-data-using-training-set-statistics","dir":"Articles > Examples > Structured_data","previous_headings":"","what":"Normalize the data using training set statistics","title":"Imbalanced classification: credit card fraud detection","text":"","code":"feature_names <- colnames(train_df) %>% setdiff(\"Class\")  train_features <- as.matrix(train_df[feature_names]) train_targets <- as.matrix(train_df$Class)  val_features <- as.matrix(val_df[feature_names]) val_targets <- as.matrix(val_df$Class)  train_features %<>% scale() val_features %<>% scale(center = attr(train_features, \"scaled:center\"),                         scale = attr(train_features, \"scaled:scale\"))"},{"path":"https://keras3.posit.co/dev/articles/examples/structured_data/imbalanced_classification.html","id":"build-a-binary-classification-model","dir":"Articles > Examples > Structured_data","previous_headings":"","what":"Build a binary classification model","title":"Imbalanced classification: credit card fraud detection","text":"","code":"model <-   keras_model_sequential(input_shape = ncol(train_features)) |>   layer_dense(256, activation = \"relu\") |>   layer_dense(256, activation = \"relu\") |>   layer_dropout(0.3) |>   layer_dense(256, activation = \"relu\") |>   layer_dropout(0.3) |>   layer_dense(1, activation = \"sigmoid\")  model ## Model: \"sequential\" ## ┏━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━┓ ## ┃ Layer (type)                    ┃ Output Shape           ┃       Param # ┃ ## ┡━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━┩ ## │ dense (Dense)                   │ (None, 256)            │         7,936 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ dense_1 (Dense)                 │ (None, 256)            │        65,792 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ dropout (Dropout)               │ (None, 256)            │             0 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ dense_2 (Dense)                 │ (None, 256)            │        65,792 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ dropout_1 (Dropout)             │ (None, 256)            │             0 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ dense_3 (Dense)                 │ (None, 1)              │           257 │ ## └─────────────────────────────────┴────────────────────────┴───────────────┘ ##  Total params: 139,777 (546.00 KB) ##  Trainable params: 139,777 (546.00 KB) ##  Non-trainable params: 0 (0.00 B)"},{"path":"https://keras3.posit.co/dev/articles/examples/structured_data/imbalanced_classification.html","id":"train-the-model-with-class_weight-argument","dir":"Articles > Examples > Structured_data","previous_headings":"","what":"Train the model with class_weight argument","title":"Imbalanced classification: credit card fraud detection","text":"","code":"metrics <- list(   metric_false_negatives(name = \"fn\"),   metric_false_positives(name = \"fp\"),   metric_true_negatives(name = \"tn\"),   metric_true_positives(name = \"tp\"),   metric_precision(name = \"precision\"),   metric_recall(name = \"recall\") ) model |> compile(   optimizer = optimizer_adam(1e-2),   loss = \"binary_crossentropy\",   metrics = metrics ) callbacks <- list(   callback_model_checkpoint(\"fraud_model_at_epoch_{epoch}.keras\") )  class_weight <- list(\"0\" = weight_for_0,                      \"1\" = weight_for_1)  model |> fit(   train_features, train_targets,   validation_data = list(val_features, val_targets),   class_weight = class_weight,   batch_size = 2048,   epochs = 30,   callbacks = callbacks,   verbose = 2 ) ## Epoch 1/30 ## 112/112 - 4s - 37ms/step - fn: 36.0000 - fp: 29717.0000 - loss: 2.3629e-06 - precision: 0.0115 - recall: 0.9058 - tn: 197747.0000 - tp: 346.0000 - val_fn: 12.0000 - val_fp: 1039.0000 - val_loss: 0.0984 - val_precision: 0.0862 - val_recall: 0.8909 - val_tn: 55812.0000 - val_tp: 98.0000 ## Epoch 2/30 ## 112/112 - 1s - 9ms/step - fn: 30.0000 - fp: 6694.0000 - loss: 1.3508e-06 - precision: 0.0500 - recall: 0.9215 - tn: 220770.0000 - tp: 352.0000 - val_fn: 8.0000 - val_fp: 2418.0000 - val_loss: 0.1390 - val_precision: 0.0405 - val_recall: 0.9273 - val_tn: 54433.0000 - val_tp: 102.0000 ## Epoch 3/30 ## 112/112 - 0s - 2ms/step - fn: 27.0000 - fp: 9781.0000 - loss: 1.3090e-06 - precision: 0.0350 - recall: 0.9293 - tn: 217683.0000 - tp: 355.0000 - val_fn: 10.0000 - val_fp: 984.0000 - val_loss: 0.1063 - val_precision: 0.0923 - val_recall: 0.9091 - val_tn: 55867.0000 - val_tp: 100.0000 ## Epoch 4/30 ## 112/112 - 0s - 2ms/step - fn: 32.0000 - fp: 6894.0000 - loss: 1.1099e-06 - precision: 0.0483 - recall: 0.9162 - tn: 220570.0000 - tp: 350.0000 - val_fn: 9.0000 - val_fp: 1557.0000 - val_loss: 0.0857 - val_precision: 0.0609 - val_recall: 0.9182 - val_tn: 55294.0000 - val_tp: 101.0000 ## Epoch 5/30 ## 112/112 - 0s - 2ms/step - fn: 23.0000 - fp: 8892.0000 - loss: 9.2992e-07 - precision: 0.0388 - recall: 0.9398 - tn: 218572.0000 - tp: 359.0000 - val_fn: 7.0000 - val_fp: 1492.0000 - val_loss: 0.0665 - val_precision: 0.0646 - val_recall: 0.9364 - val_tn: 55359.0000 - val_tp: 103.0000 ## Epoch 6/30 ## 112/112 - 0s - 2ms/step - fn: 18.0000 - fp: 8209.0000 - loss: 7.7549e-07 - precision: 0.0425 - recall: 0.9529 - tn: 219255.0000 - tp: 364.0000 - val_fn: 7.0000 - val_fp: 2749.0000 - val_loss: 0.1010 - val_precision: 0.0361 - val_recall: 0.9364 - val_tn: 54102.0000 - val_tp: 103.0000 ## Epoch 7/30 ## 112/112 - 0s - 2ms/step - fn: 15.0000 - fp: 7189.0000 - loss: 6.9032e-07 - precision: 0.0486 - recall: 0.9607 - tn: 220275.0000 - tp: 367.0000 - val_fn: 12.0000 - val_fp: 934.0000 - val_loss: 0.0539 - val_precision: 0.0950 - val_recall: 0.8909 - val_tn: 55917.0000 - val_tp: 98.0000 ## Epoch 8/30 ## 112/112 - 0s - 2ms/step - fn: 18.0000 - fp: 7759.0000 - loss: 8.2432e-07 - precision: 0.0448 - recall: 0.9529 - tn: 219705.0000 - tp: 364.0000 - val_fn: 10.0000 - val_fp: 1481.0000 - val_loss: 0.0777 - val_precision: 0.0633 - val_recall: 0.9091 - val_tn: 55370.0000 - val_tp: 100.0000 ## Epoch 9/30 ## 112/112 - 0s - 2ms/step - fn: 16.0000 - fp: 7565.0000 - loss: 7.2956e-07 - precision: 0.0461 - recall: 0.9581 - tn: 219899.0000 - tp: 366.0000 - val_fn: 8.0000 - val_fp: 2075.0000 - val_loss: 0.0718 - val_precision: 0.0469 - val_recall: 0.9273 - val_tn: 54776.0000 - val_tp: 102.0000 ## Epoch 10/30 ## 112/112 - 0s - 2ms/step - fn: 10.0000 - fp: 7374.0000 - loss: 5.9522e-07 - precision: 0.0480 - recall: 0.9738 - tn: 220090.0000 - tp: 372.0000 - val_fn: 11.0000 - val_fp: 995.0000 - val_loss: 0.0425 - val_precision: 0.0905 - val_recall: 0.9000 - val_tn: 55856.0000 - val_tp: 99.0000 ## Epoch 11/30 ## 112/112 - 0s - 2ms/step - fn: 10.0000 - fp: 6586.0000 - loss: 6.2194e-07 - precision: 0.0535 - recall: 0.9738 - tn: 220878.0000 - tp: 372.0000 - val_fn: 9.0000 - val_fp: 5097.0000 - val_loss: 0.2103 - val_precision: 0.0194 - val_recall: 0.9182 - val_tn: 51754.0000 - val_tp: 101.0000 ## Epoch 12/30 ## 112/112 - 0s - 2ms/step - fn: 6.0000 - fp: 8285.0000 - loss: 6.0991e-07 - precision: 0.0434 - recall: 0.9843 - tn: 219179.0000 - tp: 376.0000 - val_fn: 9.0000 - val_fp: 1926.0000 - val_loss: 0.0781 - val_precision: 0.0498 - val_recall: 0.9182 - val_tn: 54925.0000 - val_tp: 101.0000 ## Epoch 13/30 ## 112/112 - 0s - 2ms/step - fn: 7.0000 - fp: 6864.0000 - loss: 5.7087e-07 - precision: 0.0518 - recall: 0.9817 - tn: 220600.0000 - tp: 375.0000 - val_fn: 8.0000 - val_fp: 1900.0000 - val_loss: 0.0837 - val_precision: 0.0509 - val_recall: 0.9273 - val_tn: 54951.0000 - val_tp: 102.0000 ## Epoch 14/30 ## 112/112 - 0s - 2ms/step - fn: 7.0000 - fp: 7663.0000 - loss: 5.7664e-07 - precision: 0.0467 - recall: 0.9817 - tn: 219801.0000 - tp: 375.0000 - val_fn: 8.0000 - val_fp: 1540.0000 - val_loss: 0.0656 - val_precision: 0.0621 - val_recall: 0.9273 - val_tn: 55311.0000 - val_tp: 102.0000 ## Epoch 15/30 ## 112/112 - 0s - 2ms/step - fn: 6.0000 - fp: 4607.0000 - loss: 3.9004e-07 - precision: 0.0755 - recall: 0.9843 - tn: 222857.0000 - tp: 376.0000 - val_fn: 11.0000 - val_fp: 2546.0000 - val_loss: 0.0994 - val_precision: 0.0374 - val_recall: 0.9000 - val_tn: 54305.0000 - val_tp: 99.0000 ## Epoch 16/30 ## 112/112 - 0s - 2ms/step - fn: 6.0000 - fp: 6195.0000 - loss: 5.1274e-07 - precision: 0.0572 - recall: 0.9843 - tn: 221269.0000 - tp: 376.0000 - val_fn: 9.0000 - val_fp: 1652.0000 - val_loss: 0.0693 - val_precision: 0.0576 - val_recall: 0.9182 - val_tn: 55199.0000 - val_tp: 101.0000 ## Epoch 17/30 ## 112/112 - 0s - 2ms/step - fn: 6.0000 - fp: 5367.0000 - loss: 4.6364e-07 - precision: 0.0655 - recall: 0.9843 - tn: 222097.0000 - tp: 376.0000 - val_fn: 10.0000 - val_fp: 1378.0000 - val_loss: 0.0575 - val_precision: 0.0677 - val_recall: 0.9091 - val_tn: 55473.0000 - val_tp: 100.0000 ## Epoch 18/30 ## 112/112 - 0s - 2ms/step - fn: 2.0000 - fp: 3696.0000 - loss: 2.9266e-07 - precision: 0.0932 - recall: 0.9948 - tn: 223768.0000 - tp: 380.0000 - val_fn: 11.0000 - val_fp: 824.0000 - val_loss: 0.0375 - val_precision: 0.1073 - val_recall: 0.9000 - val_tn: 56027.0000 - val_tp: 99.0000 ## Epoch 19/30 ## 112/112 - 0s - 2ms/step - fn: 3.0000 - fp: 3638.0000 - loss: 3.6945e-07 - precision: 0.0943 - recall: 0.9921 - tn: 223826.0000 - tp: 379.0000 - val_fn: 4.0000 - val_fp: 4639.0000 - val_loss: 0.5475 - val_precision: 0.0223 - val_recall: 0.9636 - val_tn: 52212.0000 - val_tp: 106.0000 ## Epoch 20/30 ## 112/112 - 0s - 2ms/step - fn: 26.0000 - fp: 11743.0000 - loss: 4.2168e-06 - precision: 0.0294 - recall: 0.9319 - tn: 215721.0000 - tp: 356.0000 - val_fn: 104.0000 - val_fp: 395.0000 - val_loss: 0.0507 - val_precision: 0.0150 - val_recall: 0.0545 - val_tn: 56456.0000 - val_tp: 6.0000 ## Epoch 21/30 ## 112/112 - 0s - 2ms/step - fn: 41.0000 - fp: 8860.0000 - loss: 5.8042e-06 - precision: 0.0371 - recall: 0.8927 - tn: 218604.0000 - tp: 341.0000 - val_fn: 10.0000 - val_fp: 2791.0000 - val_loss: 0.2183 - val_precision: 0.0346 - val_recall: 0.9091 - val_tn: 54060.0000 - val_tp: 100.0000 ## Epoch 22/30 ## 112/112 - 0s - 2ms/step - fn: 15.0000 - fp: 9031.0000 - loss: 1.0670e-06 - precision: 0.0391 - recall: 0.9607 - tn: 218433.0000 - tp: 367.0000 - val_fn: 10.0000 - val_fp: 1275.0000 - val_loss: 0.0764 - val_precision: 0.0727 - val_recall: 0.9091 - val_tn: 55576.0000 - val_tp: 100.0000 ## Epoch 23/30 ## 112/112 - 0s - 2ms/step - fn: 10.0000 - fp: 5141.0000 - loss: 6.9193e-07 - precision: 0.0675 - recall: 0.9738 - tn: 222323.0000 - tp: 372.0000 - val_fn: 11.0000 - val_fp: 1872.0000 - val_loss: 0.0824 - val_precision: 0.0502 - val_recall: 0.9000 - val_tn: 54979.0000 - val_tp: 99.0000 ## Epoch 24/30 ## 112/112 - 0s - 2ms/step - fn: 2.0000 - fp: 4232.0000 - loss: 4.2135e-07 - precision: 0.0824 - recall: 0.9948 - tn: 223232.0000 - tp: 380.0000 - val_fn: 11.0000 - val_fp: 976.0000 - val_loss: 0.0488 - val_precision: 0.0921 - val_recall: 0.9000 - val_tn: 55875.0000 - val_tp: 99.0000 ## Epoch 25/30 ## 112/112 - 0s - 2ms/step - fn: 3.0000 - fp: 4149.0000 - loss: 3.7232e-07 - precision: 0.0837 - recall: 0.9921 - tn: 223315.0000 - tp: 379.0000 - val_fn: 13.0000 - val_fp: 359.0000 - val_loss: 0.0229 - val_precision: 0.2127 - val_recall: 0.8818 - val_tn: 56492.0000 - val_tp: 97.0000 ## Epoch 26/30 ## 112/112 - 0s - 2ms/step - fn: 3.0000 - fp: 3757.0000 - loss: 3.5368e-07 - precision: 0.0916 - recall: 0.9921 - tn: 223707.0000 - tp: 379.0000 - val_fn: 12.0000 - val_fp: 1111.0000 - val_loss: 0.0569 - val_precision: 0.0811 - val_recall: 0.8909 - val_tn: 55740.0000 - val_tp: 98.0000 ## Epoch 27/30 ## 112/112 - 0s - 2ms/step - fn: 2.0000 - fp: 3220.0000 - loss: 2.5512e-07 - precision: 0.1056 - recall: 0.9948 - tn: 224244.0000 - tp: 380.0000 - val_fn: 11.0000 - val_fp: 839.0000 - val_loss: 0.0396 - val_precision: 0.1055 - val_recall: 0.9000 - val_tn: 56012.0000 - val_tp: 99.0000 ## Epoch 28/30 ## 112/112 - 0s - 2ms/step - fn: 0.0000e+00 - fp: 2287.0000 - loss: 1.8121e-07 - precision: 0.1431 - recall: 1.0000 - tn: 225177.0000 - tp: 382.0000 - val_fn: 12.0000 - val_fp: 425.0000 - val_loss: 0.0272 - val_precision: 0.1874 - val_recall: 0.8909 - val_tn: 56426.0000 - val_tp: 98.0000 ## Epoch 29/30 ## 112/112 - 0s - 2ms/step - fn: 0.0000e+00 - fp: 1435.0000 - loss: 1.3034e-07 - precision: 0.2102 - recall: 1.0000 - tn: 226029.0000 - tp: 382.0000 - val_fn: 13.0000 - val_fp: 321.0000 - val_loss: 0.0250 - val_precision: 0.2321 - val_recall: 0.8818 - val_tn: 56530.0000 - val_tp: 97.0000 ## Epoch 30/30 ## 112/112 - 0s - 2ms/step - fn: 0.0000e+00 - fp: 1365.0000 - loss: 1.2132e-07 - precision: 0.2187 - recall: 1.0000 - tn: 226099.0000 - tp: 382.0000 - val_fn: 12.0000 - val_fp: 241.0000 - val_loss: 0.0183 - val_precision: 0.2891 - val_recall: 0.8909 - val_tn: 56610.0000 - val_tp: 98.0000 val_pred <- model %>%   predict(val_features) %>%   { as.integer(. > 0.5) } ## 1781/1781 - 1s - 427us/step pred_correct <- val_df$Class == val_pred cat(sprintf(\"Validation accuracy: %.2f\", mean(pred_correct))) ## Validation accuracy: 1.00 fraudulent <- val_df$Class == 1  n_fraudulent_detected <- sum(fraudulent & pred_correct) n_fraudulent_missed <- sum(fraudulent & !pred_correct) n_legitimate_flagged <- sum(!fraudulent & !pred_correct)"},{"path":"https://keras3.posit.co/dev/articles/examples/structured_data/imbalanced_classification.html","id":"conclusions","dir":"Articles > Examples > Structured_data","previous_headings":"","what":"Conclusions","title":"Imbalanced classification: credit card fraud detection","text":"end training, 56,961 validation transactions, : Correctly identifying 98 fraudulent Missing 12 fraudulent transactions cost incorrectly flagging 241 legitimate transactions real world, one put even higher weight class 1, reflect False Negatives costly False Positives. Next time credit card gets declined online purchase – .","code":""},{"path":"https://keras3.posit.co/dev/articles/examples/structured_data/structured_data_classification_with_feature_space.html","id":"introduction","dir":"Articles > Examples > Structured_data","previous_headings":"","what":"Introduction","title":"Structured data classification with FeatureSpace","text":"example demonstrates structured data classification (also known tabular data classification), starting raw CSV file. data includes numerical features, integer categorical features, string categorical features. use utility layer_feature_space() index, preprocess, encode features. code adapted example Structured data classification scratch. previous example managed low-level feature preprocessing encoding Keras preprocessing layers, example delegate everything layer_feature_space(), making workflow extremely quick easy.","code":""},{"path":"https://keras3.posit.co/dev/articles/examples/structured_data/structured_data_classification_with_feature_space.html","id":"the-dataset","dir":"Articles > Examples > Structured_data","previous_headings":"Introduction","what":"The dataset","title":"Structured data classification with FeatureSpace","text":"dataset provided Cleveland Clinic Foundation Heart Disease. ’s CSV file 303 rows. row contains information patient (sample), column describes attribute patient (feature). use features predict whether patient heart disease (binary classification). ’s description feature:","code":""},{"path":"https://keras3.posit.co/dev/articles/examples/structured_data/structured_data_classification_with_feature_space.html","id":"setup","dir":"Articles > Examples > Structured_data","previous_headings":"","what":"Setup","title":"Structured data classification with FeatureSpace","text":"","code":"library(readr) library(dplyr, warn.conflicts = FALSE) library(keras3) library(tensorflow, exclude = c(\"shape\", \"set_random_seed\")) library(tfdatasets, exclude = \"shape\")  conflicted::conflicts_prefer(   keras3::shape(),   keras3::set_random_seed(),   dplyr::filter(),   .quiet = TRUE )  use_backend(\"tensorflow\")"},{"path":"https://keras3.posit.co/dev/articles/examples/structured_data/structured_data_classification_with_feature_space.html","id":"preparing-the-data","dir":"Articles > Examples > Structured_data","previous_headings":"","what":"Preparing the data","title":"Structured data classification with FeatureSpace","text":"Let’s download data load Pandas dataframe: dataset includes 303 samples 14 columns per sample (13 features, plus target label) ’s preview samples: last column, “target”, indicates whether patient heart disease (1) (0). Let’s split data training validation set: Let’s generate tf_dataset objects dataframe: tf_dataset yields tuple (input, target) input dictionary (named list) features target value 0 1: Let’s batch datasets:","code":"file_url <-   \"http://storage.googleapis.com/download.tensorflow.org/data/heart.csv\" df <- read_csv(file_url, col_types = cols(   oldpeak = col_double(),   thal = col_character(),   .default = col_integer() ))  # the dataset has two malformed rows, filter them out df <- df |> filter(!thal %in% c(\"1\", \"2\")) glimpse(df) ## Rows: 301 ## Columns: 14 ## $ age      <int> 63, 67, 67, 37, 41, 56, 62, 57, 63, 53, 57, 56, 56, 44, 5… ## $ sex      <int> 1, 1, 1, 1, 0, 1, 0, 0, 1, 1, 1, 0, 1, 1, 1, 1, 1, 1, 0, … ## $ cp       <int> 1, 4, 4, 3, 2, 2, 4, 4, 4, 4, 4, 2, 3, 2, 3, 3, 2, 4, 3, … ## $ trestbps <int> 145, 160, 120, 130, 130, 120, 140, 120, 130, 140, 140, 14… ## $ chol     <int> 233, 286, 229, 250, 204, 236, 268, 354, 254, 203, 192, 29… ## $ fbs      <int> 1, 0, 0, 0, 0, 0, 0, 0, 0, 1, 0, 0, 1, 0, 1, 0, 0, 0, 0, … ## $ restecg  <int> 2, 2, 2, 0, 2, 0, 2, 0, 2, 2, 0, 2, 2, 0, 0, 0, 0, 0, 0, … ## $ thalach  <int> 150, 108, 129, 187, 172, 178, 160, 163, 147, 155, 148, 15… ## $ exang    <int> 0, 1, 1, 0, 0, 0, 0, 1, 0, 1, 0, 0, 1, 0, 0, 0, 0, 0, 0, … ## $ oldpeak  <dbl> 2.3, 1.5, 2.6, 3.5, 1.4, 0.8, 3.6, 0.6, 1.4, 3.1, 0.4, 1.… ## $ slope    <int> 3, 2, 2, 3, 1, 1, 3, 1, 2, 3, 2, 2, 2, 1, 1, 1, 3, 1, 1, … ## $ ca       <int> 0, 3, 2, 0, 0, 0, 2, 0, 1, 0, 0, 0, 1, 0, 0, 0, 0, 0, 0, … ## $ thal     <chr> \"fixed\", \"normal\", \"reversible\", \"normal\", \"normal\", \"nor… ## $ target   <int> 0, 1, 0, 0, 0, 0, 1, 0, 1, 0, 0, 0, 1, 0, 0, 0, 0, 0, 0, … df ## # A tibble: 301 × 14 ##      age   sex    cp trestbps  chol   fbs restecg thalach exang oldpeak ##    <int> <int> <int>    <int> <int> <int>   <int>   <int> <int>   <dbl> ##  1    63     1     1      145   233     1       2     150     0     2.3 ##  2    67     1     4      160   286     0       2     108     1     1.5 ##  3    67     1     4      120   229     0       2     129     1     2.6 ##  4    37     1     3      130   250     0       0     187     0     3.5 ##  5    41     0     2      130   204     0       2     172     0     1.4 ##  6    56     1     2      120   236     0       0     178     0     0.8 ##  7    62     0     4      140   268     0       2     160     0     3.6 ##  8    57     0     4      120   354     0       0     163     1     0.6 ##  9    63     1     4      130   254     0       2     147     0     1.4 ## 10    53     1     4      140   203     1       2     155     1     3.1 ## # ℹ 291 more rows ## # ℹ 4 more variables: slope <int>, ca <int>, thal <chr>, target <int> val_idx <- nrow(df) %>% sample.int(., . * 0.2) val_df <- df[val_idx, ] train_df <- df[-val_idx, ]  cat(sprintf(   \"Using %d samples for training and %d for validation\",   nrow(train_df), nrow(val_df) )) ## Using 241 samples for training and 60 for validation dataframe_to_dataset <- function(df) {   labels <- df |> pull(target) |> as.integer()   inputs <- df |> select(-target) |> as.list()    ds <- tensor_slices_dataset(list(inputs, labels)) |>     dataset_shuffle(nrow(df))    ds }  train_ds <- dataframe_to_dataset(train_df) val_ds <- dataframe_to_dataset(val_df) c(x, y) %<-% iter_next(as_iterator(train_ds)) cat(\"Input: \"); str(x) cat(\"Target: \"); str(y) ## Input: List of 13 ##  $ age     :<tf.Tensor: shape=(), dtype=int32, numpy=68> ##  $ sex     :<tf.Tensor: shape=(), dtype=int32, numpy=1> ##  $ cp      :<tf.Tensor: shape=(), dtype=int32, numpy=4> ##  $ trestbps:<tf.Tensor: shape=(), dtype=int32, numpy=144> ##  $ chol    :<tf.Tensor: shape=(), dtype=int32, numpy=193> ##  $ fbs     :<tf.Tensor: shape=(), dtype=int32, numpy=1> ##  $ restecg :<tf.Tensor: shape=(), dtype=int32, numpy=0> ##  $ thalach :<tf.Tensor: shape=(), dtype=int32, numpy=141> ##  $ exang   :<tf.Tensor: shape=(), dtype=int32, numpy=0> ##  $ oldpeak :<tf.Tensor: shape=(), dtype=float32, numpy=3.4000000953674316> ##  $ slope   :<tf.Tensor: shape=(), dtype=int32, numpy=2> ##  $ ca      :<tf.Tensor: shape=(), dtype=int32, numpy=2> ##  $ thal    :<tf.Tensor: shape=(), dtype=string, numpy=b'reversible'> ## Target: <tf.Tensor: shape=(), dtype=int32, numpy=1> train_ds <- train_ds |> dataset_batch(32) val_ds <- val_ds |> dataset_batch(32)"},{"path":"https://keras3.posit.co/dev/articles/examples/structured_data/structured_data_classification_with_feature_space.html","id":"configuring-a-featurespace","dir":"Articles > Examples > Structured_data","previous_headings":"","what":"Configuring a FeatureSpace","title":"Structured data classification with FeatureSpace","text":"configure feature preprocessed, instantiate layer_feature_space(), pass dictionary (named list unique names) maps name features string describes feature type. “integer categorical” features \"FBS\", one “string categorical” feature (\"thal\"), numerical features, ’d like normalize – except \"age\", ’d like discretize number bins. also use crosses argument capture feature interactions categorical features, say, create additional features represent value co-occurrences categorical features. can compute feature crosses like arbitrary sets categorical features – just tuples two features. resulting co-occurences hashed fixed-sized vector, don’t need worry whether co-occurence space large.","code":"feature_space <- layer_feature_space(   features = list(     # Categorical features encoded as integers     sex = \"integer_categorical\",     cp = \"integer_categorical\",     fbs = \"integer_categorical\",     restecg = \"integer_categorical\",     exang = \"integer_categorical\",     ca = \"integer_categorical\",     # Categorical feature encoded as string     thal = \"string_categorical\",     # Numerical features to discretize     age = \"float_discretized\",     # Numerical features to normalize     trestbps = \"float_normalized\",     chol = \"float_normalized\",     thalach = \"float_normalized\",     oldpeak = \"float_normalized\",     slope = \"float_normalized\"   ),   # We create additional features by hashing   # value co-occurrences for the   # following groups of categorical features.   crosses = list(c(\"sex\", \"age\"), c(\"thal\", \"ca\")),   # The hashing space for these co-occurrences   # wil be 32-dimensional.   crossing_dim = 32,   # Our utility will one-hot encode all categorical   # features and concat all features into a single   # vector (one vector per sample).   output_mode = \"concat\" )"},{"path":"https://keras3.posit.co/dev/articles/examples/structured_data/structured_data_classification_with_feature_space.html","id":"further-customizing-a-featurespace","dir":"Articles > Examples > Structured_data","previous_headings":"","what":"Further customizing a FeatureSpace","title":"Structured data classification with FeatureSpace","text":"Specifying feature type via string name quick easy, sometimes may want configure preprocessing feature. instance, case, categorical features don’t large set possible values – ’s handful values per feature (e.g. 1 0 feature \"FBS\"), possible values represented training set. result, don’t need reserve index represent “vocabulary” values features – default behavior. , just specify num_oov_indices=0 features tell feature preprocessor skip “vocabulary” indexing. customizations access include specifying number bins discretizing features type \"float_discretized\", dimensionality hashing space feature crossing.","code":"feature_space <- layer_feature_space(   features = list(     # Categorical features encoded as integers     sex       = feature_integer_categorical(num_oov_indices = 0),     cp        = feature_integer_categorical(num_oov_indices = 0),     fbs       = feature_integer_categorical(num_oov_indices = 0),     restecg   = feature_integer_categorical(num_oov_indices = 0),     exang     = feature_integer_categorical(num_oov_indices = 0),     ca        = feature_integer_categorical(num_oov_indices = 0),     # Categorical feature encoded as string     thal      = feature_string_categorical(num_oov_indices = 0),     # Numerical features to discretize     age       = feature_float_discretized(num_bins = 30),     # Numerical features to normalize     trestbps  = feature_float_normalized(),     chol      = feature_float_normalized(),     thalach   = feature_float_normalized(),     oldpeak   = feature_float_normalized(),     slope     = feature_float_normalized()   ),   # Specify feature cross with a custom crossing dim.   crosses = list(     feature_cross(       feature_names = c(\"sex\", \"age\"),       crossing_dim = 64     ),     feature_cross(       feature_names = c(\"thal\", \"ca\"),       crossing_dim = 16     )   ),   output_mode = \"concat\" )"},{"path":"https://keras3.posit.co/dev/articles/examples/structured_data/structured_data_classification_with_feature_space.html","id":"adapt-the-featurespace-to-the-training-data","dir":"Articles > Examples > Structured_data","previous_headings":"","what":"Adapt the FeatureSpace to the training data","title":"Structured data classification with FeatureSpace","text":"start using FeatureSpace build model, adapt training data. adapt(), FeatureSpace : Index set possible values categorical features. Compute mean variance numerical features normalize. Compute value boundaries different bins numerical features discretize. Note adapt() called tf_dataset yields dicts (named lists) feature values – labels. point, FeatureSpace can called dict raw feature values, return single concatenate vector sample, combining encoded features feature crosses.","code":"train_ds_with_no_labels <- train_ds |> dataset_map(\\(x, y) x) feature_space |> adapt(train_ds_with_no_labels) c(x, y) %<-% iter_next(as_iterator(train_ds)) preprocessed_x <- feature_space(x) preprocessed_x ## tf.Tensor( ## [[0. 0. 0. ... 0. 0. 0.] ##  [0. 0. 0. ... 0. 0. 0.] ##  [0. 0. 0. ... 0. 0. 0.] ##  ... ##  [0. 0. 0. ... 0. 0. 0.] ##  [0. 0. 0. ... 0. 0. 0.] ##  [0. 0. 0. ... 0. 0. 0.]], shape=(32, 136), dtype=float32)"},{"path":"https://keras3.posit.co/dev/articles/examples/structured_data/structured_data_classification_with_feature_space.html","id":"two-ways-to-manage-preprocessing-as-part-of-the-tf-data-pipeline-or-in-the-model-itself","dir":"Articles > Examples > Structured_data","previous_headings":"","what":"Two ways to manage preprocessing: as part of the tf.data pipeline, or in the model itself","title":"Structured data classification with FeatureSpace","text":"two ways can leverage FeatureSpace:","code":""},{"path":"https://keras3.posit.co/dev/articles/examples/structured_data/structured_data_classification_with_feature_space.html","id":"asynchronous-preprocessing-in-tf-data","dir":"Articles > Examples > Structured_data","previous_headings":"Two ways to manage preprocessing: as part of the tf.data pipeline, or in the model itself","what":"Asynchronous preprocessing in tf.data","title":"Structured data classification with FeatureSpace","text":"can make part data pipeline, model. enables asynchronous parallel preprocessing data CPU hits model. ’re training GPU TPU, want speed preprocessing. Usually, always right thing training.","code":""},{"path":"https://keras3.posit.co/dev/articles/examples/structured_data/structured_data_classification_with_feature_space.html","id":"synchronous-preprocessing-in-the-model","dir":"Articles > Examples > Structured_data","previous_headings":"Two ways to manage preprocessing: as part of the tf.data pipeline, or in the model itself","what":"Synchronous preprocessing in the model","title":"Structured data classification with FeatureSpace","text":"can make part model. means model expect dicts raw feature values, preprocessing batch done synchronously (blocking manner) rest forward pass. want end--end model can process raw feature values – keep mind model able run CPU, since types feature preprocessing (e.g. string preprocessing) GPU TPU compatible. GPU / TPU performance-sensitive settings. general, want -model preprocessing inference CPU. case, apply FeatureSpace tf.data pipeline training, inference end--end model includes FeatureSpace. Let’s create training validation dataset preprocessed batches:","code":"preprocessed_train_ds <- train_ds |>   dataset_map(\\(x, y) list(feature_space(x), y),               num_parallel_calls = tf$data$AUTOTUNE) |>   dataset_prefetch(tf$data$AUTOTUNE)  preprocessed_val_ds <- val_ds |>   dataset_map(\\(x, y) list(feature_space(x), y),               num_parallel_calls = tf$data$AUTOTUNE) |>   dataset_prefetch(tf$data$AUTOTUNE)"},{"path":"https://keras3.posit.co/dev/articles/examples/structured_data/structured_data_classification_with_feature_space.html","id":"build-a-model","dir":"Articles > Examples > Structured_data","previous_headings":"","what":"Build a model","title":"Structured data classification with FeatureSpace","text":"Time build model – rather two models: training model expects preprocessed features (one sample = one vector) inference model expects raw features (one sample = dict raw feature values)","code":"dict_inputs <- feature_space$get_inputs() encoded_features <- feature_space$get_encoded_features()  predictions <- encoded_features |>   layer_dense(32, activation=\"relu\") |>   layer_dropout(0.5) |>   layer_dense(1, activation=\"sigmoid\")  training_model <- keras_model(inputs = encoded_features,                               outputs = predictions) training_model |> compile(optimizer = \"adam\",                           loss = \"binary_crossentropy\",                           metrics = \"accuracy\")  inference_model <- keras_model(inputs = dict_inputs,                                outputs = predictions)"},{"path":"https://keras3.posit.co/dev/articles/examples/structured_data/structured_data_classification_with_feature_space.html","id":"train-the-model","dir":"Articles > Examples > Structured_data","previous_headings":"","what":"Train the model","title":"Structured data classification with FeatureSpace","text":"Let’s train model 20 epochs. Note feature preprocessing happening part tf.data pipeline, part model. quickly get 80% validation accuracy.","code":"training_model |> fit(   preprocessed_train_ds,   epochs = 20,   validation_data = preprocessed_val_ds,   verbose = 2 ) ## Epoch 1/20 ## 8/8 - 2s - 262ms/step - accuracy: 0.4108 - loss: 0.7398 - val_accuracy: 0.3667 - val_loss: 0.7218 ## Epoch 2/20 ## 8/8 - 0s - 26ms/step - accuracy: 0.5934 - loss: 0.6798 - val_accuracy: 0.5333 - val_loss: 0.6670 ## Epoch 3/20 ## 8/8 - 0s - 26ms/step - accuracy: 0.6473 - loss: 0.6384 - val_accuracy: 0.6333 - val_loss: 0.6207 ## Epoch 4/20 ## 8/8 - 0s - 26ms/step - accuracy: 0.7012 - loss: 0.5829 - val_accuracy: 0.7167 - val_loss: 0.5800 ## Epoch 5/20 ## 8/8 - 0s - 27ms/step - accuracy: 0.6846 - loss: 0.5780 - val_accuracy: 0.7500 - val_loss: 0.5445 ## Epoch 6/20 ## 8/8 - 0s - 26ms/step - accuracy: 0.7718 - loss: 0.5595 - val_accuracy: 0.7667 - val_loss: 0.5134 ## Epoch 7/20 ## 8/8 - 0s - 26ms/step - accuracy: 0.7801 - loss: 0.5166 - val_accuracy: 0.8167 - val_loss: 0.4868 ## Epoch 8/20 ## 8/8 - 0s - 26ms/step - accuracy: 0.7718 - loss: 0.5053 - val_accuracy: 0.8333 - val_loss: 0.4626 ## Epoch 9/20 ## 8/8 - 0s - 28ms/step - accuracy: 0.8091 - loss: 0.4774 - val_accuracy: 0.8333 - val_loss: 0.4401 ## Epoch 10/20 ## 8/8 - 0s - 27ms/step - accuracy: 0.8050 - loss: 0.4682 - val_accuracy: 0.8500 - val_loss: 0.4197 ## Epoch 11/20 ## 8/8 - 0s - 26ms/step - accuracy: 0.8133 - loss: 0.4541 - val_accuracy: 0.8667 - val_loss: 0.4011 ## Epoch 12/20 ## 8/8 - 0s - 27ms/step - accuracy: 0.8216 - loss: 0.4360 - val_accuracy: 0.8667 - val_loss: 0.3843 ## Epoch 13/20 ## 8/8 - 0s - 26ms/step - accuracy: 0.8174 - loss: 0.4081 - val_accuracy: 0.8667 - val_loss: 0.3684 ## Epoch 14/20 ## 8/8 - 0s - 27ms/step - accuracy: 0.8423 - loss: 0.4034 - val_accuracy: 0.8667 - val_loss: 0.3531 ## Epoch 15/20 ## 8/8 - 0s - 26ms/step - accuracy: 0.8174 - loss: 0.3956 - val_accuracy: 0.8833 - val_loss: 0.3397 ## Epoch 16/20 ## 8/8 - 0s - 27ms/step - accuracy: 0.8423 - loss: 0.3952 - val_accuracy: 0.8833 - val_loss: 0.3274 ## Epoch 17/20 ## 8/8 - 0s - 27ms/step - accuracy: 0.8548 - loss: 0.3445 - val_accuracy: 0.8833 - val_loss: 0.3169 ## Epoch 18/20 ## 8/8 - 0s - 27ms/step - accuracy: 0.8382 - loss: 0.3609 - val_accuracy: 0.8833 - val_loss: 0.3080 ## Epoch 19/20 ## 8/8 - 0s - 26ms/step - accuracy: 0.8714 - loss: 0.3412 - val_accuracy: 0.8833 - val_loss: 0.2986 ## Epoch 20/20 ## 8/8 - 0s - 26ms/step - accuracy: 0.8506 - loss: 0.3541 - val_accuracy: 0.9000 - val_loss: 0.2905"},{"path":"https://keras3.posit.co/dev/articles/examples/structured_data/structured_data_classification_with_feature_space.html","id":"inference-on-new-data-with-the-end-to-end-model","dir":"Articles > Examples > Structured_data","previous_headings":"","what":"Inference on new data with the end-to-end model","title":"Structured data classification with FeatureSpace","text":"Now, can use inference model (includes FeatureSpace) make predictions based dicts raw features values, follows:","code":"sample <- list(   age = 60,   sex = 1,   cp = 1,   trestbps = 145,   chol = 233,   fbs = 1,   restecg = 2,   thalach = 150,   exang = 0,   oldpeak = 2.3,   slope = 3,   ca = 0,   thal = \"fixed\" )  input_dict <- lapply(sample, \\(x) op_convert_to_tensor(array(x))) predictions <- inference_model |> predict(input_dict) ## 1/1 - 0s - 298ms/step glue::glue(r\"---(   This particular patient had a {(100 * predictions) |> signif(3)}% probability   of having a heart disease, as evaluated by our model. )---\") ## This particular patient had a 48.6% probability ## of having a heart disease, as evaluated by our model."},{"path":"https://keras3.posit.co/dev/articles/examples/timeseries/timeseries_anomaly_detection.html","id":"introduction","dir":"Articles > Examples > Timeseries","previous_headings":"","what":"Introduction","title":"Timeseries anomaly detection using an Autoencoder","text":"script demonstrates can use reconstruction convolutional autoencoder model detect anomalies timeseries data.","code":""},{"path":"https://keras3.posit.co/dev/articles/examples/timeseries/timeseries_anomaly_detection.html","id":"setup","dir":"Articles > Examples > Timeseries","previous_headings":"","what":"Setup","title":"Timeseries anomaly detection using an Autoencoder","text":"","code":"library(dplyr, warn.conflicts = FALSE) library(ggplot2) theme_set(theme_minimal())  library(listarrays) library(tfdatasets, exclude = c(\"shape\")) library(keras3)"},{"path":"https://keras3.posit.co/dev/articles/examples/timeseries/timeseries_anomaly_detection.html","id":"load-the-data","dir":"Articles > Examples > Timeseries","previous_headings":"","what":"Load the data","title":"Timeseries anomaly detection using an Autoencoder","text":"use Numenta Anomaly Benchmark(NAB) dataset. provides artificial timeseries data containing labeled anomalous periods behavior. Data ordered, timestamped, single-valued metrics. use art_daily_small_noise.csv file training art_daily_jumpsup.csv file testing. simplicity dataset allows us demonstrate anomaly detection.","code":"get_data <- function(url_suffix) {   url_root <- \"https://raw.githubusercontent.com/numenta/NAB/master/data/\"   url <- paste0(url_root, url_suffix)   file <- get_file(origin = url) # cache file locally   # parse csv; 2 columns with types: datetime (T), double (d)   readr::read_csv(file, col_types = \"Td\") }  df_small_noise   <- get_data(\"artificialNoAnomaly/art_daily_small_noise.csv\") df_daily_jumpsup <- get_data(\"artificialWithAnomaly/art_daily_jumpsup.csv\")"},{"path":"https://keras3.posit.co/dev/articles/examples/timeseries/timeseries_anomaly_detection.html","id":"quick-look-at-the-data","dir":"Articles > Examples > Timeseries","previous_headings":"","what":"Quick look at the data","title":"Timeseries anomaly detection using an Autoencoder","text":"","code":"df_small_noise ## # A tibble: 4,032 × 2 ##    timestamp           value ##    <dttm>              <dbl> ##  1 2014-04-01 00:00:00  18.3 ##  2 2014-04-01 00:05:00  22.0 ##  3 2014-04-01 00:10:00  18.6 ##  4 2014-04-01 00:15:00  22.0 ##  5 2014-04-01 00:20:00  21.9 ##  6 2014-04-01 00:25:00  21.2 ##  7 2014-04-01 00:30:00  20.6 ##  8 2014-04-01 00:35:00  20.3 ##  9 2014-04-01 00:40:00  21.5 ## 10 2014-04-01 00:45:00  19.2 ## # ℹ 4,022 more rows df_daily_jumpsup ## # A tibble: 4,032 × 2 ##    timestamp           value ##    <dttm>              <dbl> ##  1 2014-04-01 00:00:00  19.8 ##  2 2014-04-01 00:05:00  20.5 ##  3 2014-04-01 00:10:00  20.0 ##  4 2014-04-01 00:15:00  21.5 ##  5 2014-04-01 00:20:00  20.2 ##  6 2014-04-01 00:25:00  19.9 ##  7 2014-04-01 00:30:00  21.7 ##  8 2014-04-01 00:35:00  20.9 ##  9 2014-04-01 00:40:00  18.4 ## 10 2014-04-01 00:45:00  18.7 ## # ℹ 4,022 more rows"},{"path":[]},{"path":"https://keras3.posit.co/dev/articles/examples/timeseries/timeseries_anomaly_detection.html","id":"timeseries-data-without-anomalies","dir":"Articles > Examples > Timeseries","previous_headings":"Visualize the data","what":"Timeseries data without anomalies","title":"Timeseries anomaly detection using an Autoencoder","text":"use following data training.","code":"plot_ts <- function(df) {   ggplot(df, aes(x = timestamp, y = value)) + geom_line() +     scale_x_datetime(date_breaks = \"1 day\", date_labels = \"%b-%d\") }  plot_ts(df_small_noise) + ggtitle(\"Without Anomaly\")"},{"path":"https://keras3.posit.co/dev/articles/examples/timeseries/timeseries_anomaly_detection.html","id":"timeseries-data-with-anomalies","dir":"Articles > Examples > Timeseries","previous_headings":"Visualize the data","what":"Timeseries data with anomalies","title":"Timeseries anomaly detection using an Autoencoder","text":"use following data testing see sudden jump data detected anomaly.","code":"plot_ts(df_daily_jumpsup) + ggtitle(\"With Anomaly\")"},{"path":"https://keras3.posit.co/dev/articles/examples/timeseries/timeseries_anomaly_detection.html","id":"prepare-training-data","dir":"Articles > Examples > Timeseries","previous_headings":"","what":"Prepare training data","title":"Timeseries anomaly detection using an Autoencoder","text":"Get data values training timeseries data file normalize value data. value every 5 mins 14 days. 24 * 60 / 5 = 288 timesteps per day 288 * 14 = 4032 data points total","code":"df_train <- df_small_noise |>   mutate(value = (value - mean(value)) / sd(value))  cat(\"Number of training samples:\", nrow(df_train), \"\\n\") ## Number of training samples: 4032"},{"path":"https://keras3.posit.co/dev/articles/examples/timeseries/timeseries_anomaly_detection.html","id":"create-sequences","dir":"Articles > Examples > Timeseries","previous_headings":"Prepare training data","what":"Create sequences","title":"Timeseries anomaly detection using an Autoencoder","text":"Create sequences combining TIME_STEPS contiguous data values training data.","code":"TIME_STEPS <- 288  as_dataset <- function(df) {   x <- as.matrix(df$value)   ds <- timeseries_dataset_from_array(x, NULL, sequence_length = TIME_STEPS)   # Because the dataset is small, cast TF Dataset to an R array for convenience.   ds |> as_array_iterator() |> iterate() |> bind_on_rows() }  x_train <- as_dataset(df_train) writeLines(sprintf(\"Training input shape: (%s)\", toString(dim(x_train)))) ## Training input shape: (3745, 288, 1)"},{"path":"https://keras3.posit.co/dev/articles/examples/timeseries/timeseries_anomaly_detection.html","id":"build-a-model","dir":"Articles > Examples > Timeseries","previous_headings":"","what":"Build a model","title":"Timeseries anomaly detection using an Autoencoder","text":"build convolutional reconstruction autoencoder model. model take input shape (batch_size, sequence_length, num_features) return output shape. case, sequence_length 288 num_features 1.","code":"model <- keras_model_sequential(input_shape = c(TIME_STEPS, 1)) |>   layer_conv_1d(     filters = 32, kernel_size = 7, padding = \"same\",     strides = 2, activation = \"relu\"   ) |>   layer_dropout(rate = 0.2) |>   layer_conv_1d(     filters = 16, kernel_size = 7, padding = \"same\",     strides = 2, activation = \"relu\"   ) |>   layer_conv_1d_transpose(     filters = 16, kernel_size = 7, padding = \"same\",     strides = 2, activation = \"relu\"   ) |>   layer_dropout(rate = 0.2) |>   layer_conv_1d_transpose(     filters = 32, kernel_size = 7, padding = \"same\",     strides = 2, activation = \"relu\"   ) |>   layer_conv_1d_transpose(filters = 1, kernel_size = 7, padding = \"same\")  model |> compile(optimizer=optimizer_adam(learning_rate=0.001), loss=\"mse\") model ## Model: \"sequential\" ## ┏━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━┓ ## ┃ Layer (type)                    ┃ Output Shape           ┃       Param # ┃ ## ┡━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━┩ ## │ conv1d (Conv1D)                 │ (None, 144, 32)        │           256 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ dropout (Dropout)               │ (None, 144, 32)        │             0 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ conv1d_1 (Conv1D)               │ (None, 72, 16)         │         3,600 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ conv1d_transpose                │ (None, 144, 16)        │         1,808 │ ## │ (Conv1DTranspose)               │                        │               │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ dropout_1 (Dropout)             │ (None, 144, 16)        │             0 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ conv1d_transpose_1              │ (None, 288, 32)        │         3,616 │ ## │ (Conv1DTranspose)               │                        │               │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ conv1d_transpose_2              │ (None, 288, 1)         │           225 │ ## │ (Conv1DTranspose)               │                        │               │ ## └─────────────────────────────────┴────────────────────────┴───────────────┘ ##  Total params: 9,505 (37.13 KB) ##  Trainable params: 9,505 (37.13 KB) ##  Non-trainable params: 0 (0.00 B)"},{"path":"https://keras3.posit.co/dev/articles/examples/timeseries/timeseries_anomaly_detection.html","id":"train-the-model","dir":"Articles > Examples > Timeseries","previous_headings":"","what":"Train the model","title":"Timeseries anomaly detection using an Autoencoder","text":"Please note using x_train input target since reconstruction model. Let’s plot training validation loss see training went.","code":"history = model |> fit(   x_train, x_train,   epochs = 50,   validation_split = 0.1,   callbacks = c(     callback_early_stopping(       monitor = \"val_loss\", patience = 5, mode = \"min\"     )   ) ) ## Epoch 1/50 ## 106/106 - 4s - 39ms/step - loss: 0.1863 - val_loss: 0.0316 ## Epoch 2/50 ## 106/106 - 0s - 2ms/step - loss: 0.0333 - val_loss: 0.0233 ## Epoch 3/50 ## 106/106 - 0s - 2ms/step - loss: 0.0248 - val_loss: 0.0201 ## Epoch 4/50 ## 106/106 - 0s - 2ms/step - loss: 0.0209 - val_loss: 0.0186 ## Epoch 5/50 ## 106/106 - 0s - 2ms/step - loss: 0.0179 - val_loss: 0.0146 ## Epoch 6/50 ## 106/106 - 0s - 2ms/step - loss: 0.0150 - val_loss: 0.0112 ## Epoch 7/50 ## 106/106 - 0s - 2ms/step - loss: 0.0127 - val_loss: 0.0094 ## Epoch 8/50 ## 106/106 - 0s - 2ms/step - loss: 0.0109 - val_loss: 0.0088 ## Epoch 9/50 ## 106/106 - 0s - 2ms/step - loss: 0.0096 - val_loss: 0.0083 ## Epoch 10/50 ## 106/106 - 0s - 2ms/step - loss: 0.0086 - val_loss: 0.0069 ## Epoch 11/50 ## 106/106 - 0s - 2ms/step - loss: 0.0078 - val_loss: 0.0061 ## Epoch 12/50 ## 106/106 - 0s - 2ms/step - loss: 0.0073 - val_loss: 0.0057 ## Epoch 13/50 ## 106/106 - 0s - 2ms/step - loss: 0.0068 - val_loss: 0.0053 ## Epoch 14/50 ## 106/106 - 0s - 2ms/step - loss: 0.0064 - val_loss: 0.0050 ## Epoch 15/50 ## 106/106 - 0s - 2ms/step - loss: 0.0061 - val_loss: 0.0046 ## Epoch 16/50 ## 106/106 - 0s - 2ms/step - loss: 0.0058 - val_loss: 0.0040 ## Epoch 17/50 ## 106/106 - 0s - 2ms/step - loss: 0.0056 - val_loss: 0.0041 ## Epoch 18/50 ## 106/106 - 0s - 2ms/step - loss: 0.0053 - val_loss: 0.0035 ## Epoch 19/50 ## 106/106 - 0s - 2ms/step - loss: 0.0050 - val_loss: 0.0037 ## Epoch 20/50 ## 106/106 - 0s - 2ms/step - loss: 0.0048 - val_loss: 0.0032 ## Epoch 21/50 ## 106/106 - 0s - 2ms/step - loss: 0.0046 - val_loss: 0.0031 ## Epoch 22/50 ## 106/106 - 0s - 2ms/step - loss: 0.0044 - val_loss: 0.0032 ## Epoch 23/50 ## 106/106 - 0s - 2ms/step - loss: 0.0042 - val_loss: 0.0033 ## Epoch 24/50 ## 106/106 - 0s - 2ms/step - loss: 0.0041 - val_loss: 0.0031 ## Epoch 25/50 ## 106/106 - 0s - 2ms/step - loss: 0.0039 - val_loss: 0.0033 ## Epoch 26/50 ## 106/106 - 0s - 2ms/step - loss: 0.0037 - val_loss: 0.0030 ## Epoch 27/50 ## 106/106 - 0s - 2ms/step - loss: 0.0035 - val_loss: 0.0032 ## Epoch 28/50 ## 106/106 - 0s - 2ms/step - loss: 0.0034 - val_loss: 0.0025 ## Epoch 29/50 ## 106/106 - 0s - 2ms/step - loss: 0.0033 - val_loss: 0.0028 ## Epoch 30/50 ## 106/106 - 0s - 2ms/step - loss: 0.0032 - val_loss: 0.0030 ## Epoch 31/50 ## 106/106 - 0s - 2ms/step - loss: 0.0031 - val_loss: 0.0024 ## Epoch 32/50 ## 106/106 - 0s - 2ms/step - loss: 0.0030 - val_loss: 0.0028 ## Epoch 33/50 ## 106/106 - 0s - 2ms/step - loss: 0.0029 - val_loss: 0.0027 ## Epoch 34/50 ## 106/106 - 0s - 2ms/step - loss: 0.0028 - val_loss: 0.0022 ## Epoch 35/50 ## 106/106 - 0s - 2ms/step - loss: 0.0027 - val_loss: 0.0024 ## Epoch 36/50 ## 106/106 - 0s - 2ms/step - loss: 0.0026 - val_loss: 0.0025 ## Epoch 37/50 ## 106/106 - 0s - 2ms/step - loss: 0.0026 - val_loss: 0.0027 ## Epoch 38/50 ## 106/106 - 0s - 2ms/step - loss: 0.0025 - val_loss: 0.0024 ## Epoch 39/50 ## 106/106 - 0s - 2ms/step - loss: 0.0025 - val_loss: 0.0024 plot(history)"},{"path":"https://keras3.posit.co/dev/articles/examples/timeseries/timeseries_anomaly_detection.html","id":"detecting-anomalies","dir":"Articles > Examples > Timeseries","previous_headings":"","what":"Detecting anomalies","title":"Timeseries anomaly detection using an Autoencoder","text":"detect anomalies determining well model can reconstruct input data. Find MAE loss training samples. Find max MAE loss value. worst model performed trying reconstruct sample. make threshold anomaly detection. reconstruction loss sample greater threshold value can infer model seeing pattern isn’t familiar . label sample anomaly.","code":"# Get train MAE loss. x_train_pred <- model |> predict(x_train) ## 118/118 - 0s - 4ms/step train_mae_loss <- apply(abs(x_train_pred - x_train), 1, mean)  hist(train_mae_loss, breaks = 50) # Get reconstruction loss threshold. threshold <- max(train_mae_loss) cat(\"Reconstruction error threshold: \", threshold, \"\\n\") ## Reconstruction error threshold:  0.03994837"},{"path":"https://keras3.posit.co/dev/articles/examples/timeseries/timeseries_anomaly_detection.html","id":"compare-recontruction","dir":"Articles > Examples > Timeseries","previous_headings":"Detecting anomalies","what":"Compare recontruction","title":"Timeseries anomaly detection using an Autoencoder","text":"Just fun, let’s see model recontructed first sample. 288 timesteps day 1 training dataset.","code":"# Checking how the first sequence is learnt plot(NULL, NULL, ylab = 'Value',      xlim = c(0, TIME_STEPS),      ylim = range(c(x_train[1,,], x_train_pred[1,,]))) lines(x_train[1,,]) lines(x_train_pred[1,,], col = 'red') legend(\"topleft\", lty = 1,        legend = c(\"actual\", \"predicted\"),        col = c(\"black\", \"red\"))"},{"path":"https://keras3.posit.co/dev/articles/examples/timeseries/timeseries_anomaly_detection.html","id":"prepare-test-data","dir":"Articles > Examples > Timeseries","previous_headings":"Detecting anomalies","what":"Prepare test data","title":"Timeseries anomaly detection using an Autoencoder","text":"","code":"df_test <- df_daily_jumpsup |>   mutate(value =            (value - mean(df_small_noise$value)) /              sd(df_small_noise$value))  df_test |> head() plot_ts(df_test) # Create sequences from test values. x_test <- as_dataset(df_test)  # Get test MAE loss. x_test_pred <- model |> predict(x_test) test_mae_loss <- apply(abs(x_test_pred - x_test), 1, mean)  hist(test_mae_loss, breaks = 50, xlab = \"test MAE loss\", ylab = \"No of samples\") # Detect all the samples which are anomalies. anomalies <- test_mae_loss > threshold cat(\"Number of anomaly samples:\", sum(anomalies), \"\\n\") cat(\"Indices of anomaly samples:\", which(anomalies), \"\\n\", fill = TRUE) ## # A tibble: 6 × 2 ##   timestamp            value ##   <dttm>               <dbl> ## 1 2014-04-01 00:00:00 -0.808 ## 2 2014-04-01 00:05:00 -0.781 ## 3 2014-04-01 00:10:00 -0.801 ## 4 2014-04-01 00:15:00 -0.746 ## 5 2014-04-01 00:20:00 -0.792 ## 6 2014-04-01 00:25:00 -0.802 ## 118/118 - 0s - 808us/step ## Number of anomaly samples: 483 ## Indices of anomaly samples: 216 217 218 219 220 221 396 507 793 794 795 797 ## 799 861 974 1659 1944 1945 1946 1947 2016 2020 2036 2044 2046 2052 2053 ## 2054 2056 2057 2058 2060 2061 2062 2064 2065 2066 2068 2069 2070 2072 2073 ## 2074 2076 2077 2078 2080 2081 2082 2084 2085 2086 2088 2089 2090 2092 2093 ## 2094 2096 2097 2098 2100 2101 2102 2104 2105 2108 2112 2113 2116 2117 2120 ## 2121 2124 2126 2141 2521 2522 2523 2525 2546 2698 2700 2702 2703 2704 2705 ## 2706 2707 2708 2709 2710 2711 2712 2713 2714 2715 2716 2717 2718 2719 2720 ## 2721 2722 2723 2724 2725 2726 2727 2728 2729 2730 2731 2732 2733 2734 2735 ## 2736 2737 2738 2739 2740 2741 2742 2743 2744 2745 2746 2747 2748 2749 2750 ## 2751 2752 2753 2754 2755 2756 2757 2758 2759 2760 2761 2762 2763 2764 2765 ## 2766 2767 2768 2769 2770 2771 2772 2773 2774 2775 2776 2777 2778 2779 2780 ## 2781 2782 2783 2784 2785 2786 2787 2788 2789 2790 2791 2792 2793 2794 2795 ## 2796 2797 2798 2799 2800 2801 2802 2803 2804 2805 2806 2807 2808 2809 2810 ## 2811 2812 2813 2814 2815 2816 2817 2818 2819 2820 2821 2822 2823 2824 2825 ## 2826 2827 2828 2829 2830 2831 2832 2833 2834 2835 2836 2837 2838 2839 2840 ## 2841 2842 2843 2844 2845 2846 2847 2848 2849 2850 2851 2852 2853 2854 2855 ## 2856 2857 2858 2859 2860 2861 2862 2863 2864 2865 2866 2867 2868 2869 2870 ## 2871 2872 2873 2874 2875 2876 2877 2878 2879 2880 2881 2882 2883 2884 2885 ## 2886 2887 2888 2889 2890 2891 2892 2893 2894 2895 2896 2897 2898 2899 2900 ## 2901 2902 2903 2904 2905 2906 2907 2908 2909 2910 2911 2912 2913 2914 2915 ## 2916 2917 2918 2919 2920 2921 2922 2923 2924 2925 2926 2927 2928 2929 2930 ## 2931 2932 2933 2934 2935 2936 2937 2938 2939 2940 2941 2942 2943 2944 2945 ## 2946 2947 2948 2949 2950 2951 2952 2953 2954 2955 2956 2957 2958 2959 2960 ## 2961 2962 2963 2964 2965 2966 2967 2968 2969 2970 2971 2972 2973 2974 2975 ## 2976 2977 2978 2979 2980 2981 2982 2983 2984 2985 2986 2987 2988 2989 2990 ## 2991 2992 2993 2994 2995 2996 2997 2998 2999 3000 3001 3002 3003 3004 3005 ## 3006 3007 3008 3009 3010 3011 3012 3013 3014 3015 3016 3017 3018 3019 3020 ## 3021 3022 3023 3024 3025 3026 3027 3028 3029 3030 3031 3032 3033 3034 3035 ## 3036 3037 3038 3039 3040 3041 3042 3043 3044 3045 3046 3047 3048 3049 3050 ## 3051 3052 3053 3054 3055 3056 3057 3058 3059 3060 3061 3062 3063 3064 3065 ## 3066 3067 3068 3069 3070 3071 3072 3073 3074 3075 3076 3077 3078 3079 3080 ## 3081 3082 3083 3084 3085 3086 3087 3088 3089 3090 3091 3092 3093 3094 3095 ## 3096 3097 3098 3099 3100 3101"},{"path":"https://keras3.posit.co/dev/articles/examples/timeseries/timeseries_anomaly_detection.html","id":"plot-anomalies","dir":"Articles > Examples > Timeseries","previous_headings":"","what":"Plot anomalies","title":"Timeseries anomaly detection using an Autoencoder","text":"now know samples data anomalies. , find corresponding timestamps original test data. using following method : Let’s say time_steps = 3 10 training values. x_train look like : 0, 1, 2 1, 2, 3 2, 3, 4 3, 4, 5 4, 5, 6 5, 6, 7 6, 7, 8 7, 8, 9 except initial final time_steps-1 data values, appear time_steps number samples. , know samples [(3, 4, 5), (4, 5, 6), (5, 6, 7)] anomalies, can say data point 5 anomaly. Let’s overlay anomalies original test data plot.","code":"is_anomaly <- test_mae_loss > threshold is_anomaly <- is_anomaly &   zoo::rollsum(is_anomaly, TIME_STEPS,                align = \"right\", na.pad = TRUE) >= TIME_STEPS  with(df_test, {   plot(value ~ timestamp, type = 'l', xaxt = 'n', las = 2)   axis.POSIXct(1, at = seq(timestamp[1], tail(timestamp, 1), by = \"days\"),                format = \"%b-%d\") })  with(df_test[which(is_anomaly),], {   points(value ~ timestamp, col = \"red\") })"},{"path":"https://keras3.posit.co/dev/articles/examples/timeseries/timeseries_classification_from_scratch.html","id":"introduction","dir":"Articles > Examples > Timeseries","previous_headings":"","what":"Introduction","title":"Timeseries classification from scratch","text":"example shows timeseries classification scratch, starting raw CSV timeseries files disk. demonstrate workflow FordA dataset UCR/UEA archive.","code":""},{"path":"https://keras3.posit.co/dev/articles/examples/timeseries/timeseries_classification_from_scratch.html","id":"setup","dir":"Articles > Examples > Timeseries","previous_headings":"","what":"Setup","title":"Timeseries classification from scratch","text":"","code":"library(keras3) use_backend(\"jax\")"},{"path":[]},{"path":"https://keras3.posit.co/dev/articles/examples/timeseries/timeseries_classification_from_scratch.html","id":"dataset-description","dir":"Articles > Examples > Timeseries","previous_headings":"Load the data: the FordA dataset","what":"Dataset description","title":"Timeseries classification from scratch","text":"dataset using called FordA. data comes UCR archive. dataset contains 3601 training instances another 1320 testing instances. timeseries corresponds measurement engine noise captured motor sensor. task, goal automatically detect presence specific issue engine. problem balanced binary classification task. full description dataset can found .","code":""},{"path":"https://keras3.posit.co/dev/articles/examples/timeseries/timeseries_classification_from_scratch.html","id":"read-the-tsv-data","dir":"Articles > Examples > Timeseries","previous_headings":"Load the data: the FordA dataset","what":"Read the TSV data","title":"Timeseries classification from scratch","text":"use FordA_TRAIN file training FordA_TEST file testing. simplicity dataset allows us demonstrate effectively use ConvNets timeseries classification. file, first column corresponds label.","code":"get_data <- function(path) {   if(path |> startsWith(\"https://\"))     path <- get_file(origin = path)  # cache file locally    data <- readr::read_tsv(     path, col_names = FALSE,     # Each row is: one integer (the label),     # followed by 500 doubles (the timeseries)     col_types = paste0(\"i\", strrep(\"d\", 500))   )    y <- as.matrix(data[[1]])   x <- as.matrix(data[,-1])   dimnames(x) <- dimnames(y) <- NULL    list(x, y) }  root_url <- \"https://raw.githubusercontent.com/hfawaz/cd-diagram/master/FordA/\" c(x_train, y_train) %<-% get_data(paste0(root_url, \"FordA_TRAIN.tsv\")) c(x_test, y_test) %<-% get_data(paste0(root_url, \"FordA_TEST.tsv\"))  str(keras3:::named_list(   x_train, y_train,   x_test, y_test )) ## List of 4 ##  $ x_train: num [1:3601, 1:500] -0.797 0.805 0.728 -0.234 -0.171 ... ##  $ y_train: int [1:3601, 1] -1 1 -1 -1 -1 1 1 1 1 1 ... ##  $ x_test : num [1:1320, 1:500] -0.14 0.334 0.717 1.24 -1.159 ... ##  $ y_test : int [1:1320, 1] -1 -1 -1 1 -1 1 -1 -1 1 1 ..."},{"path":"https://keras3.posit.co/dev/articles/examples/timeseries/timeseries_classification_from_scratch.html","id":"visualize-the-data","dir":"Articles > Examples > Timeseries","previous_headings":"","what":"Visualize the data","title":"Timeseries classification from scratch","text":"visualize one timeseries example class dataset.","code":"plot(NULL, main = \"Timeseries Data\",      xlab = \"Timepoints\",  ylab = \"Values\",      xlim = c(1, ncol(x_test)),      ylim = range(x_test)) grid() lines(x_test[match(-1, y_test), ], col = \"blue\") lines(x_test[match( 1, y_test), ], col = \"red\") legend(\"topright\", legend=c(\"label -1\", \"label 1\"), col=c(\"blue\", \"red\"), lty=1)"},{"path":"https://keras3.posit.co/dev/articles/examples/timeseries/timeseries_classification_from_scratch.html","id":"standardize-the-data","dir":"Articles > Examples > Timeseries","previous_headings":"","what":"Standardize the data","title":"Timeseries classification from scratch","text":"timeseries already single length (500). However, values usually various ranges. ideal neural network; general seek make input values normalized. specific dataset, data already z-normalized: timeseries sample mean equal zero standard deviation equal one. type normalization common timeseries classification problems, see Bagnall et al. (2016). Note timeseries data used univariate, meaning one channel per timeseries example. therefore transform timeseries multivariate one one channel using simple reshaping via numpy. allow us construct model easily applicable multivariate time series. Finally, order use sparse_categorical_crossentropy, count number classes beforehand. Now shuffle training set using validation_split option later training. Standardize labels positive integers. expected labels 0 1.","code":"dim(x_train) <- c(dim(x_train), 1) dim(x_test) <- c(dim(x_test), 1) num_classes <- length(unique(y_train)) c(x_train, y_train) %<-% listarrays::shuffle_rows(x_train, y_train) # idx <- sample.int(nrow(x_train)) # x_train %<>% .[idx,, ,drop = FALSE] # y_train %<>% .[idx,  ,drop = FALSE] y_train[y_train == -1L] <- 0L y_test[y_test == -1L] <- 0L"},{"path":"https://keras3.posit.co/dev/articles/examples/timeseries/timeseries_classification_from_scratch.html","id":"build-a-model","dir":"Articles > Examples > Timeseries","previous_headings":"","what":"Build a model","title":"Timeseries classification from scratch","text":"build Fully Convolutional Neural Network originally proposed paper. implementation based TF 2 version provided . following hyperparameters (kernel_size, filters, usage BatchNorm) found via random search using KerasTuner. plot chunk unnamed-chunk-9","code":"make_model <- function(input_shape) {   inputs <- keras_input(input_shape)    outputs <- inputs |>     # conv1     layer_conv_1d(filters = 64, kernel_size = 3, padding = \"same\") |>     layer_batch_normalization() |>     layer_activation_relu() |>     # conv2     layer_conv_1d(filters = 64, kernel_size = 3, padding = \"same\") |>     layer_batch_normalization() |>     layer_activation_relu() |>     # conv3     layer_conv_1d(filters = 64, kernel_size = 3, padding = \"same\") |>     layer_batch_normalization() |>     layer_activation_relu() |>     # pooling     layer_global_average_pooling_1d() |>     # final output     layer_dense(num_classes, activation = \"softmax\")    keras_model(inputs, outputs) }  model <- make_model(input_shape = dim(x_train)[-1]) model ## Model: \"functional\" ## ┏━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━┳━━━━━━━┓ ## ┃ Layer (type)                ┃ Output Shape          ┃    Param # ┃ Trai… ┃ ## ┡━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━╇━━━━━━━┩ ## │ input_layer (InputLayer)    │ (None, 500, 1)        │          0 │   -   │ ## ├─────────────────────────────┼───────────────────────┼────────────┼───────┤ ## │ conv1d (Conv1D)             │ (None, 500, 64)       │        256 │   Y   │ ## ├─────────────────────────────┼───────────────────────┼────────────┼───────┤ ## │ batch_normalization         │ (None, 500, 64)       │        256 │   Y   │ ## │ (BatchNormalization)        │                       │            │       │ ## ├─────────────────────────────┼───────────────────────┼────────────┼───────┤ ## │ re_lu (ReLU)                │ (None, 500, 64)       │          0 │   -   │ ## ├─────────────────────────────┼───────────────────────┼────────────┼───────┤ ## │ conv1d_1 (Conv1D)           │ (None, 500, 64)       │     12,352 │   Y   │ ## ├─────────────────────────────┼───────────────────────┼────────────┼───────┤ ## │ batch_normalization_1       │ (None, 500, 64)       │        256 │   Y   │ ## │ (BatchNormalization)        │                       │            │       │ ## ├─────────────────────────────┼───────────────────────┼────────────┼───────┤ ## │ re_lu_1 (ReLU)              │ (None, 500, 64)       │          0 │   -   │ ## ├─────────────────────────────┼───────────────────────┼────────────┼───────┤ ## │ conv1d_2 (Conv1D)           │ (None, 500, 64)       │     12,352 │   Y   │ ## ├─────────────────────────────┼───────────────────────┼────────────┼───────┤ ## │ batch_normalization_2       │ (None, 500, 64)       │        256 │   Y   │ ## │ (BatchNormalization)        │                       │            │       │ ## ├─────────────────────────────┼───────────────────────┼────────────┼───────┤ ## │ re_lu_2 (ReLU)              │ (None, 500, 64)       │          0 │   -   │ ## ├─────────────────────────────┼───────────────────────┼────────────┼───────┤ ## │ global_average_pooling1d    │ (None, 64)            │          0 │   -   │ ## │ (GlobalAveragePooling1D)    │                       │            │       │ ## ├─────────────────────────────┼───────────────────────┼────────────┼───────┤ ## │ dense (Dense)               │ (None, 2)             │        130 │   Y   │ ## └─────────────────────────────┴───────────────────────┴────────────┴───────┘ ##  Total params: 25,858 (101.01 KB) ##  Trainable params: 25,474 (99.51 KB) ##  Non-trainable params: 384 (1.50 KB) plot(model, show_shapes = TRUE)"},{"path":"https://keras3.posit.co/dev/articles/examples/timeseries/timeseries_classification_from_scratch.html","id":"train-the-model","dir":"Articles > Examples > Timeseries","previous_headings":"","what":"Train the model","title":"Timeseries classification from scratch","text":"","code":"epochs <- 500 batch_size <- 32  callbacks <- c(   callback_model_checkpoint(     \"best_model.keras\", save_best_only = TRUE,     monitor = \"val_loss\"   ),   callback_reduce_lr_on_plateau(     monitor = \"val_loss\", factor = 0.5,     patience = 20, min_lr = 0.0001   ),   callback_early_stopping(     monitor = \"val_loss\", patience = 50,     verbose = 1   ) )   model |> compile(   optimizer = \"adam\",   loss = \"sparse_categorical_crossentropy\",   metrics = \"sparse_categorical_accuracy\" )  history <- model |> fit(   x_train, y_train,   batch_size = batch_size,   epochs = epochs,   callbacks = callbacks,   validation_split = 0.2 ) ## Epoch 1/500 ## 90/90 - 3s - 32ms/step - loss: 0.5310 - sparse_categorical_accuracy: 0.7191 - val_loss: 0.7907 - val_sparse_categorical_accuracy: 0.4896 - learning_rate: 1.0000e-03 ## Epoch 2/500 ## 90/90 - 1s - 11ms/step - loss: 0.4799 - sparse_categorical_accuracy: 0.7590 - val_loss: 0.9127 - val_sparse_categorical_accuracy: 0.4896 - learning_rate: 1.0000e-03 ## Epoch 3/500 ## 90/90 - 0s - 2ms/step - loss: 0.4590 - sparse_categorical_accuracy: 0.7757 - val_loss: 0.8196 - val_sparse_categorical_accuracy: 0.4896 - learning_rate: 1.0000e-03 ## Epoch 4/500 ## 90/90 - 0s - 2ms/step - loss: 0.4088 - sparse_categorical_accuracy: 0.7983 - val_loss: 0.6425 - val_sparse_categorical_accuracy: 0.5534 - learning_rate: 1.0000e-03 ## Epoch 5/500 ## 90/90 - 0s - 2ms/step - loss: 0.4141 - sparse_categorical_accuracy: 0.7889 - val_loss: 0.5519 - val_sparse_categorical_accuracy: 0.6588 - learning_rate: 1.0000e-03 ## Epoch 6/500 ## 90/90 - 0s - 2ms/step - loss: 0.3929 - sparse_categorical_accuracy: 0.8163 - val_loss: 0.4225 - val_sparse_categorical_accuracy: 0.8294 - learning_rate: 1.0000e-03 ## Epoch 7/500 ## 90/90 - 0s - 1ms/step - loss: 0.3870 - sparse_categorical_accuracy: 0.8146 - val_loss: 0.6112 - val_sparse_categorical_accuracy: 0.6949 - learning_rate: 1.0000e-03 ## Epoch 8/500 ## 90/90 - 0s - 1ms/step - loss: 0.3717 - sparse_categorical_accuracy: 0.8247 - val_loss: 0.4257 - val_sparse_categorical_accuracy: 0.7559 - learning_rate: 1.0000e-03 ## Epoch 9/500 ## 90/90 - 0s - 2ms/step - loss: 0.3698 - sparse_categorical_accuracy: 0.8306 - val_loss: 0.3653 - val_sparse_categorical_accuracy: 0.8350 - learning_rate: 1.0000e-03 ## Epoch 10/500 ## 90/90 - 0s - 1ms/step - loss: 0.3634 - sparse_categorical_accuracy: 0.8260 - val_loss: 0.4390 - val_sparse_categorical_accuracy: 0.7836 - learning_rate: 1.0000e-03 ## Epoch 11/500 ## 90/90 - 0s - 1ms/step - loss: 0.3676 - sparse_categorical_accuracy: 0.8281 - val_loss: 0.4205 - val_sparse_categorical_accuracy: 0.7822 - learning_rate: 1.0000e-03 ## Epoch 12/500 ## 90/90 - 0s - 1ms/step - loss: 0.3467 - sparse_categorical_accuracy: 0.8406 - val_loss: 0.3697 - val_sparse_categorical_accuracy: 0.8460 - learning_rate: 1.0000e-03 ## Epoch 13/500 ## 90/90 - 0s - 2ms/step - loss: 0.3310 - sparse_categorical_accuracy: 0.8594 - val_loss: 0.3341 - val_sparse_categorical_accuracy: 0.8530 - learning_rate: 1.0000e-03 ## Epoch 14/500 ## 90/90 - 0s - 1ms/step - loss: 0.3350 - sparse_categorical_accuracy: 0.8514 - val_loss: 0.5723 - val_sparse_categorical_accuracy: 0.7198 - learning_rate: 1.0000e-03 ## Epoch 15/500 ## 90/90 - 0s - 1ms/step - loss: 0.3295 - sparse_categorical_accuracy: 0.8497 - val_loss: 1.0273 - val_sparse_categorical_accuracy: 0.5465 - learning_rate: 1.0000e-03 ## Epoch 16/500 ## 90/90 - 0s - 2ms/step - loss: 0.3244 - sparse_categorical_accuracy: 0.8587 - val_loss: 0.3219 - val_sparse_categorical_accuracy: 0.8544 - learning_rate: 1.0000e-03 ## Epoch 17/500 ## 90/90 - 0s - 1ms/step - loss: 0.3082 - sparse_categorical_accuracy: 0.8646 - val_loss: 0.4187 - val_sparse_categorical_accuracy: 0.7906 - learning_rate: 1.0000e-03 ## Epoch 18/500 ## 90/90 - 0s - 1ms/step - loss: 0.3022 - sparse_categorical_accuracy: 0.8715 - val_loss: 0.6302 - val_sparse_categorical_accuracy: 0.6824 - learning_rate: 1.0000e-03 ## Epoch 19/500 ## 90/90 - 0s - 1ms/step - loss: 0.2996 - sparse_categorical_accuracy: 0.8764 - val_loss: 0.3537 - val_sparse_categorical_accuracy: 0.8447 - learning_rate: 1.0000e-03 ## Epoch 20/500 ## 90/90 - 0s - 1ms/step - loss: 0.3053 - sparse_categorical_accuracy: 0.8715 - val_loss: 0.4925 - val_sparse_categorical_accuracy: 0.7309 - learning_rate: 1.0000e-03 ## Epoch 21/500 ## 90/90 - 0s - 1ms/step - loss: 0.2827 - sparse_categorical_accuracy: 0.8826 - val_loss: 0.3893 - val_sparse_categorical_accuracy: 0.7809 - learning_rate: 1.0000e-03 ## Epoch 22/500 ## 90/90 - 0s - 2ms/step - loss: 0.2892 - sparse_categorical_accuracy: 0.8795 - val_loss: 0.3128 - val_sparse_categorical_accuracy: 0.8863 - learning_rate: 1.0000e-03 ## Epoch 23/500 ## 90/90 - 0s - 2ms/step - loss: 0.2718 - sparse_categorical_accuracy: 0.8889 - val_loss: 0.3083 - val_sparse_categorical_accuracy: 0.8571 - learning_rate: 1.0000e-03 ## Epoch 24/500 ## 90/90 - 0s - 1ms/step - loss: 0.2691 - sparse_categorical_accuracy: 0.8934 - val_loss: 0.5609 - val_sparse_categorical_accuracy: 0.7420 - learning_rate: 1.0000e-03 ## Epoch 25/500 ## 90/90 - 0s - 1ms/step - loss: 0.2635 - sparse_categorical_accuracy: 0.8948 - val_loss: 0.5264 - val_sparse_categorical_accuracy: 0.7573 - learning_rate: 1.0000e-03 ## Epoch 26/500 ## 90/90 - 0s - 1ms/step - loss: 0.3109 - sparse_categorical_accuracy: 0.8667 - val_loss: 0.3934 - val_sparse_categorical_accuracy: 0.7878 - learning_rate: 1.0000e-03 ## Epoch 27/500 ## 90/90 - 0s - 2ms/step - loss: 0.2599 - sparse_categorical_accuracy: 0.8969 - val_loss: 0.2974 - val_sparse_categorical_accuracy: 0.8558 - learning_rate: 1.0000e-03 ## Epoch 28/500 ## 90/90 - 0s - 1ms/step - loss: 0.2805 - sparse_categorical_accuracy: 0.8823 - val_loss: 0.4026 - val_sparse_categorical_accuracy: 0.8183 - learning_rate: 1.0000e-03 ## Epoch 29/500 ## 90/90 - 0s - 2ms/step - loss: 0.2517 - sparse_categorical_accuracy: 0.8948 - val_loss: 0.2699 - val_sparse_categorical_accuracy: 0.8849 - learning_rate: 1.0000e-03 ## Epoch 30/500 ## 90/90 - 0s - 1ms/step - loss: 0.2581 - sparse_categorical_accuracy: 0.8920 - val_loss: 0.2743 - val_sparse_categorical_accuracy: 0.8752 - learning_rate: 1.0000e-03 ## Epoch 31/500 ## 90/90 - 0s - 1ms/step - loss: 0.2550 - sparse_categorical_accuracy: 0.8986 - val_loss: 0.2936 - val_sparse_categorical_accuracy: 0.8655 - learning_rate: 1.0000e-03 ## Epoch 32/500 ## 90/90 - 0s - 2ms/step - loss: 0.2528 - sparse_categorical_accuracy: 0.8934 - val_loss: 0.2528 - val_sparse_categorical_accuracy: 0.8904 - learning_rate: 1.0000e-03 ## Epoch 33/500 ## 90/90 - 0s - 1ms/step - loss: 0.2390 - sparse_categorical_accuracy: 0.9062 - val_loss: 0.2961 - val_sparse_categorical_accuracy: 0.8571 - learning_rate: 1.0000e-03 ## Epoch 34/500 ## 90/90 - 0s - 1ms/step - loss: 0.2327 - sparse_categorical_accuracy: 0.9049 - val_loss: 0.3039 - val_sparse_categorical_accuracy: 0.8558 - learning_rate: 1.0000e-03 ## Epoch 35/500 ## 90/90 - 0s - 1ms/step - loss: 0.2447 - sparse_categorical_accuracy: 0.9021 - val_loss: 0.3123 - val_sparse_categorical_accuracy: 0.8710 - learning_rate: 1.0000e-03 ## Epoch 36/500 ## 90/90 - 0s - 1ms/step - loss: 0.2340 - sparse_categorical_accuracy: 0.9069 - val_loss: 0.2912 - val_sparse_categorical_accuracy: 0.8821 - learning_rate: 1.0000e-03 ## Epoch 37/500 ## 90/90 - 0s - 1ms/step - loss: 0.2352 - sparse_categorical_accuracy: 0.8997 - val_loss: 0.3983 - val_sparse_categorical_accuracy: 0.8017 - learning_rate: 1.0000e-03 ## Epoch 38/500 ## 90/90 - 0s - 1ms/step - loss: 0.2294 - sparse_categorical_accuracy: 0.9139 - val_loss: 0.3342 - val_sparse_categorical_accuracy: 0.8447 - learning_rate: 1.0000e-03 ## Epoch 39/500 ## 90/90 - 0s - 1ms/step - loss: 0.2268 - sparse_categorical_accuracy: 0.9135 - val_loss: 0.4730 - val_sparse_categorical_accuracy: 0.7975 - learning_rate: 1.0000e-03 ## Epoch 40/500 ## 90/90 - 0s - 1ms/step - loss: 0.2202 - sparse_categorical_accuracy: 0.9111 - val_loss: 0.3007 - val_sparse_categorical_accuracy: 0.8738 - learning_rate: 1.0000e-03 ## Epoch 41/500 ## 90/90 - 0s - 1ms/step - loss: 0.2064 - sparse_categorical_accuracy: 0.9233 - val_loss: 0.2924 - val_sparse_categorical_accuracy: 0.8849 - learning_rate: 1.0000e-03 ## Epoch 42/500 ## 90/90 - 0s - 1ms/step - loss: 0.1966 - sparse_categorical_accuracy: 0.9316 - val_loss: 0.8303 - val_sparse_categorical_accuracy: 0.6935 - learning_rate: 1.0000e-03 ## Epoch 43/500 ## 90/90 - 0s - 2ms/step - loss: 0.2074 - sparse_categorical_accuracy: 0.9212 - val_loss: 0.2334 - val_sparse_categorical_accuracy: 0.9043 - learning_rate: 1.0000e-03 ## Epoch 44/500 ## 90/90 - 0s - 1ms/step - loss: 0.1961 - sparse_categorical_accuracy: 0.9313 - val_loss: 0.2343 - val_sparse_categorical_accuracy: 0.9182 - learning_rate: 1.0000e-03 ## Epoch 45/500 ## 90/90 - 0s - 1ms/step - loss: 0.1923 - sparse_categorical_accuracy: 0.9302 - val_loss: 1.1167 - val_sparse_categorical_accuracy: 0.6755 - learning_rate: 1.0000e-03 ## Epoch 46/500 ## 90/90 - 0s - 1ms/step - loss: 0.1823 - sparse_categorical_accuracy: 0.9347 - val_loss: 0.8034 - val_sparse_categorical_accuracy: 0.6935 - learning_rate: 1.0000e-03 ## Epoch 47/500 ## 90/90 - 0s - 1ms/step - loss: 0.1635 - sparse_categorical_accuracy: 0.9444 - val_loss: 0.7104 - val_sparse_categorical_accuracy: 0.7115 - learning_rate: 1.0000e-03 ## Epoch 48/500 ## 90/90 - 0s - 2ms/step - loss: 0.1587 - sparse_categorical_accuracy: 0.9434 - val_loss: 0.2001 - val_sparse_categorical_accuracy: 0.9196 - learning_rate: 1.0000e-03 ## Epoch 49/500 ## 90/90 - 0s - 1ms/step - loss: 0.1492 - sparse_categorical_accuracy: 0.9483 - val_loss: 0.3399 - val_sparse_categorical_accuracy: 0.8294 - learning_rate: 1.0000e-03 ## Epoch 50/500 ## 90/90 - 0s - 1ms/step - loss: 0.1738 - sparse_categorical_accuracy: 0.9382 - val_loss: 0.2680 - val_sparse_categorical_accuracy: 0.8890 - learning_rate: 1.0000e-03 ## Epoch 51/500 ## 90/90 - 0s - 1ms/step - loss: 0.1530 - sparse_categorical_accuracy: 0.9486 - val_loss: 0.2891 - val_sparse_categorical_accuracy: 0.8544 - learning_rate: 1.0000e-03 ## Epoch 52/500 ## 90/90 - 0s - 1ms/step - loss: 0.1320 - sparse_categorical_accuracy: 0.9538 - val_loss: 1.7182 - val_sparse_categorical_accuracy: 0.7032 - learning_rate: 1.0000e-03 ## Epoch 53/500 ## 90/90 - 0s - 1ms/step - loss: 0.1277 - sparse_categorical_accuracy: 0.9559 - val_loss: 1.4269 - val_sparse_categorical_accuracy: 0.7254 - learning_rate: 1.0000e-03 ## Epoch 54/500 ## 90/90 - 0s - 1ms/step - loss: 0.1249 - sparse_categorical_accuracy: 0.9611 - val_loss: 0.7962 - val_sparse_categorical_accuracy: 0.7365 - learning_rate: 1.0000e-03 ## Epoch 55/500 ## 90/90 - 0s - 1ms/step - loss: 0.1214 - sparse_categorical_accuracy: 0.9611 - val_loss: 0.2158 - val_sparse_categorical_accuracy: 0.9071 - learning_rate: 1.0000e-03 ## Epoch 56/500 ## 90/90 - 0s - 1ms/step - loss: 0.1258 - sparse_categorical_accuracy: 0.9604 - val_loss: 0.6349 - val_sparse_categorical_accuracy: 0.7642 - learning_rate: 1.0000e-03 ## Epoch 57/500 ## 90/90 - 0s - 1ms/step - loss: 0.1268 - sparse_categorical_accuracy: 0.9604 - val_loss: 0.2129 - val_sparse_categorical_accuracy: 0.9071 - learning_rate: 1.0000e-03 ## Epoch 58/500 ## 90/90 - 0s - 1ms/step - loss: 0.1088 - sparse_categorical_accuracy: 0.9639 - val_loss: 0.7215 - val_sparse_categorical_accuracy: 0.6657 - learning_rate: 1.0000e-03 ## Epoch 59/500 ## 90/90 - 0s - 2ms/step - loss: 0.1219 - sparse_categorical_accuracy: 0.9569 - val_loss: 0.1600 - val_sparse_categorical_accuracy: 0.9320 - learning_rate: 1.0000e-03 ## Epoch 60/500 ## 90/90 - 0s - 1ms/step - loss: 0.1190 - sparse_categorical_accuracy: 0.9642 - val_loss: 0.3411 - val_sparse_categorical_accuracy: 0.8558 - learning_rate: 1.0000e-03 ## Epoch 61/500 ## 90/90 - 0s - 1ms/step - loss: 0.1177 - sparse_categorical_accuracy: 0.9653 - val_loss: 0.3984 - val_sparse_categorical_accuracy: 0.8502 - learning_rate: 1.0000e-03 ## Epoch 62/500 ## 90/90 - 0s - 1ms/step - loss: 0.1309 - sparse_categorical_accuracy: 0.9542 - val_loss: 0.3462 - val_sparse_categorical_accuracy: 0.8211 - learning_rate: 1.0000e-03 ## Epoch 63/500 ## 90/90 - 0s - 1ms/step - loss: 0.1291 - sparse_categorical_accuracy: 0.9514 - val_loss: 0.2303 - val_sparse_categorical_accuracy: 0.8960 - learning_rate: 1.0000e-03 ## Epoch 64/500 ## 90/90 - 0s - 1ms/step - loss: 0.1200 - sparse_categorical_accuracy: 0.9628 - val_loss: 0.1735 - val_sparse_categorical_accuracy: 0.9223 - learning_rate: 1.0000e-03 ## Epoch 65/500 ## 90/90 - 0s - 1ms/step - loss: 0.1138 - sparse_categorical_accuracy: 0.9625 - val_loss: 0.5628 - val_sparse_categorical_accuracy: 0.7920 - learning_rate: 1.0000e-03 ## Epoch 66/500 ## 90/90 - 0s - 2ms/step - loss: 0.1065 - sparse_categorical_accuracy: 0.9667 - val_loss: 0.1540 - val_sparse_categorical_accuracy: 0.9417 - learning_rate: 1.0000e-03 ## Epoch 67/500 ## 90/90 - 0s - 1ms/step - loss: 0.1156 - sparse_categorical_accuracy: 0.9618 - val_loss: 0.2319 - val_sparse_categorical_accuracy: 0.9043 - learning_rate: 1.0000e-03 ## Epoch 68/500 ## 90/90 - 0s - 1ms/step - loss: 0.1100 - sparse_categorical_accuracy: 0.9597 - val_loss: 0.8668 - val_sparse_categorical_accuracy: 0.7406 - learning_rate: 1.0000e-03 ## Epoch 69/500 ## 90/90 - 0s - 2ms/step - loss: 0.1165 - sparse_categorical_accuracy: 0.9618 - val_loss: 0.1460 - val_sparse_categorical_accuracy: 0.9515 - learning_rate: 1.0000e-03 ## Epoch 70/500 ## 90/90 - 0s - 1ms/step - loss: 0.1178 - sparse_categorical_accuracy: 0.9608 - val_loss: 0.8988 - val_sparse_categorical_accuracy: 0.7046 - learning_rate: 1.0000e-03 ## Epoch 71/500 ## 90/90 - 0s - 1ms/step - loss: 0.1068 - sparse_categorical_accuracy: 0.9642 - val_loss: 0.3645 - val_sparse_categorical_accuracy: 0.8585 - learning_rate: 1.0000e-03 ## Epoch 72/500 ## 90/90 - 0s - 1ms/step - loss: 0.1106 - sparse_categorical_accuracy: 0.9653 - val_loss: 0.2813 - val_sparse_categorical_accuracy: 0.8932 - learning_rate: 1.0000e-03 ## Epoch 73/500 ## 90/90 - 0s - 1ms/step - loss: 0.1117 - sparse_categorical_accuracy: 0.9632 - val_loss: 1.8641 - val_sparse_categorical_accuracy: 0.6741 - learning_rate: 1.0000e-03 ## Epoch 74/500 ## 90/90 - 0s - 1ms/step - loss: 0.1058 - sparse_categorical_accuracy: 0.9625 - val_loss: 1.4979 - val_sparse_categorical_accuracy: 0.6893 - learning_rate: 1.0000e-03 ## Epoch 75/500 ## 90/90 - 0s - 1ms/step - loss: 0.1223 - sparse_categorical_accuracy: 0.9594 - val_loss: 0.1508 - val_sparse_categorical_accuracy: 0.9473 - learning_rate: 1.0000e-03 ## Epoch 76/500 ## 90/90 - 0s - 2ms/step - loss: 0.1140 - sparse_categorical_accuracy: 0.9601 - val_loss: 0.1400 - val_sparse_categorical_accuracy: 0.9320 - learning_rate: 1.0000e-03 ## Epoch 77/500 ## 90/90 - 0s - 1ms/step - loss: 0.1050 - sparse_categorical_accuracy: 0.9639 - val_loss: 0.2169 - val_sparse_categorical_accuracy: 0.9126 - learning_rate: 1.0000e-03 ## Epoch 78/500 ## 90/90 - 0s - 1ms/step - loss: 0.0986 - sparse_categorical_accuracy: 0.9712 - val_loss: 0.3240 - val_sparse_categorical_accuracy: 0.8724 - learning_rate: 1.0000e-03 ## Epoch 79/500 ## 90/90 - 0s - 1ms/step - loss: 0.1081 - sparse_categorical_accuracy: 0.9632 - val_loss: 0.1525 - val_sparse_categorical_accuracy: 0.9362 - learning_rate: 1.0000e-03 ## Epoch 80/500 ## 90/90 - 0s - 1ms/step - loss: 0.0995 - sparse_categorical_accuracy: 0.9667 - val_loss: 0.3123 - val_sparse_categorical_accuracy: 0.8641 - learning_rate: 1.0000e-03 ## Epoch 81/500 ## 90/90 - 0s - 1ms/step - loss: 0.1058 - sparse_categorical_accuracy: 0.9632 - val_loss: 0.2851 - val_sparse_categorical_accuracy: 0.8779 - learning_rate: 1.0000e-03 ## Epoch 82/500 ## 90/90 - 0s - 1ms/step - loss: 0.0958 - sparse_categorical_accuracy: 0.9681 - val_loss: 0.1417 - val_sparse_categorical_accuracy: 0.9362 - learning_rate: 1.0000e-03 ## Epoch 83/500 ## 90/90 - 0s - 1ms/step - loss: 0.1005 - sparse_categorical_accuracy: 0.9660 - val_loss: 0.7325 - val_sparse_categorical_accuracy: 0.7157 - learning_rate: 1.0000e-03 ## Epoch 84/500 ## 90/90 - 0s - 2ms/step - loss: 0.0966 - sparse_categorical_accuracy: 0.9684 - val_loss: 0.1322 - val_sparse_categorical_accuracy: 0.9404 - learning_rate: 1.0000e-03 ## Epoch 85/500 ## 90/90 - 0s - 2ms/step - loss: 0.1018 - sparse_categorical_accuracy: 0.9667 - val_loss: 0.1202 - val_sparse_categorical_accuracy: 0.9473 - learning_rate: 1.0000e-03 ## Epoch 86/500 ## 90/90 - 0s - 1ms/step - loss: 0.0995 - sparse_categorical_accuracy: 0.9649 - val_loss: 0.2956 - val_sparse_categorical_accuracy: 0.8766 - learning_rate: 1.0000e-03 ## Epoch 87/500 ## 90/90 - 0s - 1ms/step - loss: 0.1068 - sparse_categorical_accuracy: 0.9632 - val_loss: 0.3050 - val_sparse_categorical_accuracy: 0.8710 - learning_rate: 1.0000e-03 ## Epoch 88/500 ## 90/90 - 0s - 1ms/step - loss: 0.1002 - sparse_categorical_accuracy: 0.9674 - val_loss: 0.1675 - val_sparse_categorical_accuracy: 0.9196 - learning_rate: 1.0000e-03 ## Epoch 89/500 ## 90/90 - 0s - 1ms/step - loss: 0.1042 - sparse_categorical_accuracy: 0.9608 - val_loss: 0.7157 - val_sparse_categorical_accuracy: 0.7115 - learning_rate: 1.0000e-03 ## Epoch 90/500 ## 90/90 - 0s - 1ms/step - loss: 0.0998 - sparse_categorical_accuracy: 0.9694 - val_loss: 0.1576 - val_sparse_categorical_accuracy: 0.9251 - learning_rate: 1.0000e-03 ## Epoch 91/500 ## 90/90 - 0s - 1ms/step - loss: 0.1187 - sparse_categorical_accuracy: 0.9583 - val_loss: 0.4125 - val_sparse_categorical_accuracy: 0.8031 - learning_rate: 1.0000e-03 ## Epoch 92/500 ## 90/90 - 0s - 1ms/step - loss: 0.0973 - sparse_categorical_accuracy: 0.9688 - val_loss: 0.3084 - val_sparse_categorical_accuracy: 0.8696 - learning_rate: 1.0000e-03 ## Epoch 93/500 ## 90/90 - 0s - 1ms/step - loss: 0.0972 - sparse_categorical_accuracy: 0.9677 - val_loss: 0.1446 - val_sparse_categorical_accuracy: 0.9334 - learning_rate: 1.0000e-03 ## Epoch 94/500 ## 90/90 - 0s - 1ms/step - loss: 0.0965 - sparse_categorical_accuracy: 0.9632 - val_loss: 0.1265 - val_sparse_categorical_accuracy: 0.9584 - learning_rate: 1.0000e-03 ## Epoch 95/500 ## 90/90 - 0s - 2ms/step - loss: 0.0927 - sparse_categorical_accuracy: 0.9667 - val_loss: 0.1507 - val_sparse_categorical_accuracy: 0.9376 - learning_rate: 1.0000e-03 ## Epoch 96/500 ## 90/90 - 0s - 1ms/step - loss: 0.1050 - sparse_categorical_accuracy: 0.9618 - val_loss: 0.5005 - val_sparse_categorical_accuracy: 0.8086 - learning_rate: 1.0000e-03 ## Epoch 97/500 ## 90/90 - 0s - 1ms/step - loss: 0.1134 - sparse_categorical_accuracy: 0.9597 - val_loss: 0.1732 - val_sparse_categorical_accuracy: 0.9390 - learning_rate: 1.0000e-03 ## Epoch 98/500 ## 90/90 - 0s - 1ms/step - loss: 0.0946 - sparse_categorical_accuracy: 0.9698 - val_loss: 1.8677 - val_sparse_categorical_accuracy: 0.6949 - learning_rate: 1.0000e-03 ## Epoch 99/500 ## 90/90 - 0s - 1ms/step - loss: 0.1018 - sparse_categorical_accuracy: 0.9632 - val_loss: 1.1308 - val_sparse_categorical_accuracy: 0.6782 - learning_rate: 1.0000e-03 ## Epoch 100/500 ## 90/90 - 0s - 1ms/step - loss: 0.0990 - sparse_categorical_accuracy: 0.9674 - val_loss: 0.3846 - val_sparse_categorical_accuracy: 0.8558 - learning_rate: 1.0000e-03 ## Epoch 101/500 ## 90/90 - 0s - 1ms/step - loss: 0.0891 - sparse_categorical_accuracy: 0.9705 - val_loss: 0.1480 - val_sparse_categorical_accuracy: 0.9362 - learning_rate: 1.0000e-03 ## Epoch 102/500 ## 90/90 - 0s - 1ms/step - loss: 0.1003 - sparse_categorical_accuracy: 0.9677 - val_loss: 0.1713 - val_sparse_categorical_accuracy: 0.9390 - learning_rate: 1.0000e-03 ## Epoch 103/500 ## 90/90 - 0s - 1ms/step - loss: 0.0910 - sparse_categorical_accuracy: 0.9722 - val_loss: 0.1415 - val_sparse_categorical_accuracy: 0.9542 - learning_rate: 1.0000e-03 ## Epoch 104/500 ## 90/90 - 0s - 1ms/step - loss: 0.0998 - sparse_categorical_accuracy: 0.9622 - val_loss: 0.5677 - val_sparse_categorical_accuracy: 0.7753 - learning_rate: 1.0000e-03 ## Epoch 105/500 ## 90/90 - 0s - 1ms/step - loss: 0.1037 - sparse_categorical_accuracy: 0.9649 - val_loss: 0.9369 - val_sparse_categorical_accuracy: 0.6241 - learning_rate: 1.0000e-03 ## Epoch 106/500 ## 90/90 - 1s - 10ms/step - loss: 0.0850 - sparse_categorical_accuracy: 0.9708 - val_loss: 0.1388 - val_sparse_categorical_accuracy: 0.9487 - learning_rate: 5.0000e-04 ## Epoch 107/500 ## 90/90 - 0s - 1ms/step - loss: 0.0833 - sparse_categorical_accuracy: 0.9743 - val_loss: 0.1413 - val_sparse_categorical_accuracy: 0.9376 - learning_rate: 5.0000e-04 ## Epoch 108/500 ## 90/90 - 0s - 1ms/step - loss: 0.0775 - sparse_categorical_accuracy: 0.9767 - val_loss: 0.1223 - val_sparse_categorical_accuracy: 0.9445 - learning_rate: 5.0000e-04 ## Epoch 109/500 ## 90/90 - 0s - 1ms/step - loss: 0.0837 - sparse_categorical_accuracy: 0.9753 - val_loss: 0.1275 - val_sparse_categorical_accuracy: 0.9528 - learning_rate: 5.0000e-04 ## Epoch 110/500 ## 90/90 - 0s - 1ms/step - loss: 0.0859 - sparse_categorical_accuracy: 0.9705 - val_loss: 0.1209 - val_sparse_categorical_accuracy: 0.9487 - learning_rate: 5.0000e-04 ## Epoch 111/500 ## 90/90 - 0s - 2ms/step - loss: 0.0780 - sparse_categorical_accuracy: 0.9757 - val_loss: 0.1142 - val_sparse_categorical_accuracy: 0.9584 - learning_rate: 5.0000e-04 ## Epoch 112/500 ## 90/90 - 0s - 1ms/step - loss: 0.0801 - sparse_categorical_accuracy: 0.9736 - val_loss: 0.1316 - val_sparse_categorical_accuracy: 0.9459 - learning_rate: 5.0000e-04 ## Epoch 113/500 ## 90/90 - 0s - 1ms/step - loss: 0.0740 - sparse_categorical_accuracy: 0.9753 - val_loss: 0.3385 - val_sparse_categorical_accuracy: 0.8544 - learning_rate: 5.0000e-04 ## Epoch 114/500 ## 90/90 - 0s - 1ms/step - loss: 0.0907 - sparse_categorical_accuracy: 0.9722 - val_loss: 0.1222 - val_sparse_categorical_accuracy: 0.9612 - learning_rate: 5.0000e-04 ## Epoch 115/500 ## 90/90 - 0s - 2ms/step - loss: 0.0767 - sparse_categorical_accuracy: 0.9733 - val_loss: 0.1115 - val_sparse_categorical_accuracy: 0.9584 - learning_rate: 5.0000e-04 ## Epoch 116/500 ## 90/90 - 0s - 1ms/step - loss: 0.0786 - sparse_categorical_accuracy: 0.9750 - val_loss: 0.1667 - val_sparse_categorical_accuracy: 0.9431 - learning_rate: 5.0000e-04 ## Epoch 117/500 ## 90/90 - 0s - 1ms/step - loss: 0.0760 - sparse_categorical_accuracy: 0.9750 - val_loss: 0.1519 - val_sparse_categorical_accuracy: 0.9376 - learning_rate: 5.0000e-04 ## Epoch 118/500 ## 90/90 - 0s - 1ms/step - loss: 0.0797 - sparse_categorical_accuracy: 0.9722 - val_loss: 0.1555 - val_sparse_categorical_accuracy: 0.9459 - learning_rate: 5.0000e-04 ## Epoch 119/500 ## 90/90 - 0s - 1ms/step - loss: 0.0793 - sparse_categorical_accuracy: 0.9729 - val_loss: 0.1252 - val_sparse_categorical_accuracy: 0.9598 - learning_rate: 5.0000e-04 ## Epoch 120/500 ## 90/90 - 0s - 1ms/step - loss: 0.0803 - sparse_categorical_accuracy: 0.9747 - val_loss: 0.2298 - val_sparse_categorical_accuracy: 0.9015 - learning_rate: 5.0000e-04 ## Epoch 121/500 ## 90/90 - 0s - 1ms/step - loss: 0.0793 - sparse_categorical_accuracy: 0.9736 - val_loss: 0.1140 - val_sparse_categorical_accuracy: 0.9528 - learning_rate: 5.0000e-04 ## Epoch 122/500 ## 90/90 - 0s - 1ms/step - loss: 0.0795 - sparse_categorical_accuracy: 0.9740 - val_loss: 0.1154 - val_sparse_categorical_accuracy: 0.9542 - learning_rate: 5.0000e-04 ## Epoch 123/500 ## 90/90 - 0s - 1ms/step - loss: 0.0768 - sparse_categorical_accuracy: 0.9740 - val_loss: 0.3421 - val_sparse_categorical_accuracy: 0.8724 - learning_rate: 5.0000e-04 ## Epoch 124/500 ## 90/90 - 0s - 1ms/step - loss: 0.0848 - sparse_categorical_accuracy: 0.9694 - val_loss: 0.1126 - val_sparse_categorical_accuracy: 0.9598 - learning_rate: 5.0000e-04 ## Epoch 125/500 ## 90/90 - 0s - 1ms/step - loss: 0.0800 - sparse_categorical_accuracy: 0.9722 - val_loss: 0.1821 - val_sparse_categorical_accuracy: 0.9293 - learning_rate: 5.0000e-04 ## Epoch 126/500 ## 90/90 - 0s - 2ms/step - loss: 0.0787 - sparse_categorical_accuracy: 0.9753 - val_loss: 0.1105 - val_sparse_categorical_accuracy: 0.9570 - learning_rate: 5.0000e-04 ## Epoch 127/500 ## 90/90 - 0s - 2ms/step - loss: 0.0799 - sparse_categorical_accuracy: 0.9760 - val_loss: 0.1091 - val_sparse_categorical_accuracy: 0.9584 - learning_rate: 5.0000e-04 ## Epoch 128/500 ## 90/90 - 0s - 1ms/step - loss: 0.0811 - sparse_categorical_accuracy: 0.9736 - val_loss: 0.1287 - val_sparse_categorical_accuracy: 0.9515 - learning_rate: 5.0000e-04 ## Epoch 129/500 ## 90/90 - 0s - 1ms/step - loss: 0.0797 - sparse_categorical_accuracy: 0.9736 - val_loss: 0.1143 - val_sparse_categorical_accuracy: 0.9584 - learning_rate: 5.0000e-04 ## Epoch 130/500 ## 90/90 - 0s - 1ms/step - loss: 0.0850 - sparse_categorical_accuracy: 0.9750 - val_loss: 0.1581 - val_sparse_categorical_accuracy: 0.9376 - learning_rate: 5.0000e-04 ## Epoch 131/500 ## 90/90 - 0s - 1ms/step - loss: 0.0707 - sparse_categorical_accuracy: 0.9792 - val_loss: 0.1129 - val_sparse_categorical_accuracy: 0.9639 - learning_rate: 5.0000e-04 ## Epoch 132/500 ## 90/90 - 0s - 1ms/step - loss: 0.0725 - sparse_categorical_accuracy: 0.9740 - val_loss: 0.1425 - val_sparse_categorical_accuracy: 0.9501 - learning_rate: 5.0000e-04 ## Epoch 133/500 ## 90/90 - 0s - 2ms/step - loss: 0.0705 - sparse_categorical_accuracy: 0.9764 - val_loss: 0.1098 - val_sparse_categorical_accuracy: 0.9612 - learning_rate: 5.0000e-04 ## Epoch 134/500 ## 90/90 - 0s - 1ms/step - loss: 0.0767 - sparse_categorical_accuracy: 0.9736 - val_loss: 0.1884 - val_sparse_categorical_accuracy: 0.9376 - learning_rate: 5.0000e-04 ## Epoch 135/500 ## 90/90 - 0s - 1ms/step - loss: 0.0775 - sparse_categorical_accuracy: 0.9743 - val_loss: 0.2899 - val_sparse_categorical_accuracy: 0.8988 - learning_rate: 5.0000e-04 ## Epoch 136/500 ## 90/90 - 0s - 1ms/step - loss: 0.0807 - sparse_categorical_accuracy: 0.9701 - val_loss: 0.1152 - val_sparse_categorical_accuracy: 0.9584 - learning_rate: 5.0000e-04 ## Epoch 137/500 ## 90/90 - 0s - 1ms/step - loss: 0.0839 - sparse_categorical_accuracy: 0.9750 - val_loss: 0.3530 - val_sparse_categorical_accuracy: 0.8793 - learning_rate: 5.0000e-04 ## Epoch 138/500 ## 90/90 - 0s - 1ms/step - loss: 0.0770 - sparse_categorical_accuracy: 0.9743 - val_loss: 0.1267 - val_sparse_categorical_accuracy: 0.9473 - learning_rate: 5.0000e-04 ## Epoch 139/500 ## 90/90 - 0s - 2ms/step - loss: 0.0845 - sparse_categorical_accuracy: 0.9719 - val_loss: 0.1043 - val_sparse_categorical_accuracy: 0.9626 - learning_rate: 5.0000e-04 ## Epoch 140/500 ## 90/90 - 0s - 1ms/step - loss: 0.0701 - sparse_categorical_accuracy: 0.9799 - val_loss: 0.1351 - val_sparse_categorical_accuracy: 0.9556 - learning_rate: 5.0000e-04 ## Epoch 141/500 ## 90/90 - 0s - 1ms/step - loss: 0.0712 - sparse_categorical_accuracy: 0.9764 - val_loss: 0.1121 - val_sparse_categorical_accuracy: 0.9612 - learning_rate: 5.0000e-04 ## Epoch 142/500 ## 90/90 - 0s - 1ms/step - loss: 0.0692 - sparse_categorical_accuracy: 0.9778 - val_loss: 0.1139 - val_sparse_categorical_accuracy: 0.9639 - learning_rate: 5.0000e-04 ## Epoch 143/500 ## 90/90 - 0s - 1ms/step - loss: 0.0781 - sparse_categorical_accuracy: 0.9736 - val_loss: 0.1202 - val_sparse_categorical_accuracy: 0.9515 - learning_rate: 5.0000e-04 ## Epoch 144/500 ## 90/90 - 0s - 1ms/step - loss: 0.0699 - sparse_categorical_accuracy: 0.9781 - val_loss: 0.1131 - val_sparse_categorical_accuracy: 0.9501 - learning_rate: 5.0000e-04 ## Epoch 145/500 ## 90/90 - 0s - 1ms/step - loss: 0.0712 - sparse_categorical_accuracy: 0.9771 - val_loss: 0.1097 - val_sparse_categorical_accuracy: 0.9542 - learning_rate: 5.0000e-04 ## Epoch 146/500 ## 90/90 - 0s - 1ms/step - loss: 0.0730 - sparse_categorical_accuracy: 0.9736 - val_loss: 0.1421 - val_sparse_categorical_accuracy: 0.9445 - learning_rate: 5.0000e-04 ## Epoch 147/500 ## 90/90 - 0s - 1ms/step - loss: 0.0757 - sparse_categorical_accuracy: 0.9740 - val_loss: 0.3533 - val_sparse_categorical_accuracy: 0.8752 - learning_rate: 5.0000e-04 ## Epoch 148/500 ## 90/90 - 0s - 1ms/step - loss: 0.0818 - sparse_categorical_accuracy: 0.9701 - val_loss: 0.7787 - val_sparse_categorical_accuracy: 0.7836 - learning_rate: 5.0000e-04 ## Epoch 149/500 ## 90/90 - 0s - 1ms/step - loss: 0.0717 - sparse_categorical_accuracy: 0.9771 - val_loss: 0.1214 - val_sparse_categorical_accuracy: 0.9515 - learning_rate: 5.0000e-04 ## Epoch 150/500 ## 90/90 - 0s - 1ms/step - loss: 0.0736 - sparse_categorical_accuracy: 0.9757 - val_loss: 0.1445 - val_sparse_categorical_accuracy: 0.9362 - learning_rate: 5.0000e-04 ## Epoch 151/500 ## 90/90 - 0s - 1ms/step - loss: 0.0725 - sparse_categorical_accuracy: 0.9729 - val_loss: 0.1175 - val_sparse_categorical_accuracy: 0.9528 - learning_rate: 5.0000e-04 ## Epoch 152/500 ## 90/90 - 0s - 1ms/step - loss: 0.0726 - sparse_categorical_accuracy: 0.9750 - val_loss: 0.1329 - val_sparse_categorical_accuracy: 0.9515 - learning_rate: 5.0000e-04 ## Epoch 153/500 ## 90/90 - 0s - 1ms/step - loss: 0.0727 - sparse_categorical_accuracy: 0.9757 - val_loss: 0.1806 - val_sparse_categorical_accuracy: 0.9293 - learning_rate: 5.0000e-04 ## Epoch 154/500 ## 90/90 - 0s - 1ms/step - loss: 0.0684 - sparse_categorical_accuracy: 0.9778 - val_loss: 0.3892 - val_sparse_categorical_accuracy: 0.8682 - learning_rate: 5.0000e-04 ## Epoch 155/500 ## 90/90 - 0s - 1ms/step - loss: 0.0734 - sparse_categorical_accuracy: 0.9753 - val_loss: 0.1122 - val_sparse_categorical_accuracy: 0.9542 - learning_rate: 5.0000e-04 ## Epoch 156/500 ## 90/90 - 0s - 1ms/step - loss: 0.0758 - sparse_categorical_accuracy: 0.9774 - val_loss: 0.1077 - val_sparse_categorical_accuracy: 0.9653 - learning_rate: 5.0000e-04 ## Epoch 157/500 ## 90/90 - 0s - 1ms/step - loss: 0.0734 - sparse_categorical_accuracy: 0.9747 - val_loss: 0.1594 - val_sparse_categorical_accuracy: 0.9376 - learning_rate: 5.0000e-04 ## Epoch 158/500 ## 90/90 - 0s - 1ms/step - loss: 0.0753 - sparse_categorical_accuracy: 0.9750 - val_loss: 0.1992 - val_sparse_categorical_accuracy: 0.9168 - learning_rate: 5.0000e-04 ## Epoch 159/500 ## 90/90 - 0s - 1ms/step - loss: 0.0740 - sparse_categorical_accuracy: 0.9740 - val_loss: 0.1483 - val_sparse_categorical_accuracy: 0.9362 - learning_rate: 5.0000e-04 ## Epoch 160/500 ## 90/90 - 0s - 2ms/step - loss: 0.0677 - sparse_categorical_accuracy: 0.9778 - val_loss: 0.1041 - val_sparse_categorical_accuracy: 0.9653 - learning_rate: 2.5000e-04 ## Epoch 161/500 ## 90/90 - 0s - 1ms/step - loss: 0.0686 - sparse_categorical_accuracy: 0.9778 - val_loss: 0.1327 - val_sparse_categorical_accuracy: 0.9515 - learning_rate: 2.5000e-04 ## Epoch 162/500 ## 90/90 - 0s - 1ms/step - loss: 0.0634 - sparse_categorical_accuracy: 0.9792 - val_loss: 0.1059 - val_sparse_categorical_accuracy: 0.9667 - learning_rate: 2.5000e-04 ## Epoch 163/500 ## 90/90 - 0s - 1ms/step - loss: 0.0595 - sparse_categorical_accuracy: 0.9792 - val_loss: 0.1236 - val_sparse_categorical_accuracy: 0.9612 - learning_rate: 2.5000e-04 ## Epoch 164/500 ## 90/90 - 0s - 1ms/step - loss: 0.0637 - sparse_categorical_accuracy: 0.9806 - val_loss: 0.1185 - val_sparse_categorical_accuracy: 0.9626 - learning_rate: 2.5000e-04 ## Epoch 165/500 ## 90/90 - 0s - 1ms/step - loss: 0.0607 - sparse_categorical_accuracy: 0.9816 - val_loss: 0.1480 - val_sparse_categorical_accuracy: 0.9473 - learning_rate: 2.5000e-04 ## Epoch 166/500 ## 90/90 - 0s - 1ms/step - loss: 0.0594 - sparse_categorical_accuracy: 0.9799 - val_loss: 0.1044 - val_sparse_categorical_accuracy: 0.9653 - learning_rate: 2.5000e-04 ## Epoch 167/500 ## 90/90 - 0s - 1ms/step - loss: 0.0629 - sparse_categorical_accuracy: 0.9788 - val_loss: 0.1343 - val_sparse_categorical_accuracy: 0.9445 - learning_rate: 2.5000e-04 ## Epoch 168/500 ## 90/90 - 0s - 1ms/step - loss: 0.0740 - sparse_categorical_accuracy: 0.9750 - val_loss: 0.1081 - val_sparse_categorical_accuracy: 0.9667 - learning_rate: 2.5000e-04 ## Epoch 169/500 ## 90/90 - 0s - 1ms/step - loss: 0.0649 - sparse_categorical_accuracy: 0.9799 - val_loss: 0.1136 - val_sparse_categorical_accuracy: 0.9681 - learning_rate: 2.5000e-04 ## Epoch 170/500 ## 90/90 - 0s - 2ms/step - loss: 0.0649 - sparse_categorical_accuracy: 0.9823 - val_loss: 0.1037 - val_sparse_categorical_accuracy: 0.9695 - learning_rate: 2.5000e-04 ## Epoch 171/500 ## 90/90 - 0s - 1ms/step - loss: 0.0609 - sparse_categorical_accuracy: 0.9799 - val_loss: 0.1239 - val_sparse_categorical_accuracy: 0.9653 - learning_rate: 2.5000e-04 ## Epoch 172/500 ## 90/90 - 0s - 1ms/step - loss: 0.0636 - sparse_categorical_accuracy: 0.9795 - val_loss: 0.1037 - val_sparse_categorical_accuracy: 0.9528 - learning_rate: 2.5000e-04 ## Epoch 173/500 ## 90/90 - 0s - 1ms/step - loss: 0.0608 - sparse_categorical_accuracy: 0.9816 - val_loss: 0.1296 - val_sparse_categorical_accuracy: 0.9626 - learning_rate: 2.5000e-04 ## Epoch 174/500 ## 90/90 - 0s - 2ms/step - loss: 0.0624 - sparse_categorical_accuracy: 0.9792 - val_loss: 0.0981 - val_sparse_categorical_accuracy: 0.9639 - learning_rate: 2.5000e-04 ## Epoch 175/500 ## 90/90 - 0s - 1ms/step - loss: 0.0666 - sparse_categorical_accuracy: 0.9785 - val_loss: 0.1135 - val_sparse_categorical_accuracy: 0.9487 - learning_rate: 2.5000e-04 ## Epoch 176/500 ## 90/90 - 0s - 1ms/step - loss: 0.0625 - sparse_categorical_accuracy: 0.9771 - val_loss: 0.1024 - val_sparse_categorical_accuracy: 0.9584 - learning_rate: 2.5000e-04 ## Epoch 177/500 ## 90/90 - 0s - 1ms/step - loss: 0.0616 - sparse_categorical_accuracy: 0.9806 - val_loss: 0.1036 - val_sparse_categorical_accuracy: 0.9570 - learning_rate: 2.5000e-04 ## Epoch 178/500 ## 90/90 - 0s - 1ms/step - loss: 0.0590 - sparse_categorical_accuracy: 0.9823 - val_loss: 0.2086 - val_sparse_categorical_accuracy: 0.9154 - learning_rate: 2.5000e-04 ## Epoch 179/500 ## 90/90 - 0s - 1ms/step - loss: 0.0602 - sparse_categorical_accuracy: 0.9788 - val_loss: 0.1181 - val_sparse_categorical_accuracy: 0.9653 - learning_rate: 2.5000e-04 ## Epoch 180/500 ## 90/90 - 0s - 1ms/step - loss: 0.0615 - sparse_categorical_accuracy: 0.9816 - val_loss: 0.1587 - val_sparse_categorical_accuracy: 0.9459 - learning_rate: 2.5000e-04 ## Epoch 181/500 ## 90/90 - 0s - 1ms/step - loss: 0.0647 - sparse_categorical_accuracy: 0.9813 - val_loss: 0.1836 - val_sparse_categorical_accuracy: 0.9362 - learning_rate: 2.5000e-04 ## Epoch 182/500 ## 90/90 - 0s - 1ms/step - loss: 0.0614 - sparse_categorical_accuracy: 0.9813 - val_loss: 0.2156 - val_sparse_categorical_accuracy: 0.9223 - learning_rate: 2.5000e-04 ## Epoch 183/500 ## 90/90 - 0s - 1ms/step - loss: 0.0637 - sparse_categorical_accuracy: 0.9778 - val_loss: 0.1028 - val_sparse_categorical_accuracy: 0.9584 - learning_rate: 2.5000e-04 ## Epoch 184/500 ## 90/90 - 0s - 1ms/step - loss: 0.0572 - sparse_categorical_accuracy: 0.9823 - val_loss: 0.1347 - val_sparse_categorical_accuracy: 0.9431 - learning_rate: 2.5000e-04 ## Epoch 185/500 ## 90/90 - 0s - 1ms/step - loss: 0.0593 - sparse_categorical_accuracy: 0.9826 - val_loss: 0.1402 - val_sparse_categorical_accuracy: 0.9459 - learning_rate: 2.5000e-04 ## Epoch 186/500 ## 90/90 - 0s - 1ms/step - loss: 0.0659 - sparse_categorical_accuracy: 0.9774 - val_loss: 0.1873 - val_sparse_categorical_accuracy: 0.9293 - learning_rate: 2.5000e-04 ## Epoch 187/500 ## 90/90 - 0s - 1ms/step - loss: 0.0606 - sparse_categorical_accuracy: 0.9799 - val_loss: 0.1210 - val_sparse_categorical_accuracy: 0.9570 - learning_rate: 2.5000e-04 ## Epoch 188/500 ## 90/90 - 0s - 1ms/step - loss: 0.0610 - sparse_categorical_accuracy: 0.9813 - val_loss: 0.1250 - val_sparse_categorical_accuracy: 0.9473 - learning_rate: 2.5000e-04 ## Epoch 189/500 ## 90/90 - 0s - 1ms/step - loss: 0.0570 - sparse_categorical_accuracy: 0.9830 - val_loss: 0.1227 - val_sparse_categorical_accuracy: 0.9667 - learning_rate: 2.5000e-04 ## Epoch 190/500 ## 90/90 - 0s - 1ms/step - loss: 0.0678 - sparse_categorical_accuracy: 0.9771 - val_loss: 0.1367 - val_sparse_categorical_accuracy: 0.9584 - learning_rate: 2.5000e-04 ## Epoch 191/500 ## 90/90 - 0s - 1ms/step - loss: 0.0580 - sparse_categorical_accuracy: 0.9802 - val_loss: 0.1005 - val_sparse_categorical_accuracy: 0.9653 - learning_rate: 2.5000e-04 ## Epoch 192/500 ## 90/90 - 0s - 1ms/step - loss: 0.0651 - sparse_categorical_accuracy: 0.9778 - val_loss: 0.1170 - val_sparse_categorical_accuracy: 0.9639 - learning_rate: 2.5000e-04 ## Epoch 193/500 ## 90/90 - 0s - 1ms/step - loss: 0.0593 - sparse_categorical_accuracy: 0.9823 - val_loss: 0.1235 - val_sparse_categorical_accuracy: 0.9473 - learning_rate: 2.5000e-04 ## Epoch 194/500 ## 90/90 - 0s - 1ms/step - loss: 0.0588 - sparse_categorical_accuracy: 0.9792 - val_loss: 0.1089 - val_sparse_categorical_accuracy: 0.9570 - learning_rate: 2.5000e-04 ## Epoch 195/500 ## 90/90 - 0s - 2ms/step - loss: 0.0564 - sparse_categorical_accuracy: 0.9833 - val_loss: 0.0979 - val_sparse_categorical_accuracy: 0.9681 - learning_rate: 1.2500e-04 ## Epoch 196/500 ## 90/90 - 0s - 1ms/step - loss: 0.0584 - sparse_categorical_accuracy: 0.9788 - val_loss: 0.1169 - val_sparse_categorical_accuracy: 0.9584 - learning_rate: 1.2500e-04 ## Epoch 197/500 ## 90/90 - 0s - 1ms/step - loss: 0.0553 - sparse_categorical_accuracy: 0.9802 - val_loss: 0.1380 - val_sparse_categorical_accuracy: 0.9556 - learning_rate: 1.2500e-04 ## Epoch 198/500 ## 90/90 - 0s - 1ms/step - loss: 0.0541 - sparse_categorical_accuracy: 0.9868 - val_loss: 0.1261 - val_sparse_categorical_accuracy: 0.9556 - learning_rate: 1.2500e-04 ## Epoch 199/500 ## 90/90 - 0s - 1ms/step - loss: 0.0575 - sparse_categorical_accuracy: 0.9806 - val_loss: 0.1010 - val_sparse_categorical_accuracy: 0.9681 - learning_rate: 1.2500e-04 ## Epoch 200/500 ## 90/90 - 0s - 1ms/step - loss: 0.0568 - sparse_categorical_accuracy: 0.9816 - val_loss: 0.0996 - val_sparse_categorical_accuracy: 0.9612 - learning_rate: 1.2500e-04 ## Epoch 201/500 ## 90/90 - 0s - 1ms/step - loss: 0.0547 - sparse_categorical_accuracy: 0.9819 - val_loss: 0.1096 - val_sparse_categorical_accuracy: 0.9653 - learning_rate: 1.2500e-04 ## Epoch 202/500 ## 90/90 - 0s - 1ms/step - loss: 0.0547 - sparse_categorical_accuracy: 0.9847 - val_loss: 0.1019 - val_sparse_categorical_accuracy: 0.9556 - learning_rate: 1.2500e-04 ## Epoch 203/500 ## 90/90 - 0s - 1ms/step - loss: 0.0562 - sparse_categorical_accuracy: 0.9816 - val_loss: 0.0985 - val_sparse_categorical_accuracy: 0.9639 - learning_rate: 1.2500e-04 ## Epoch 204/500 ## 90/90 - 0s - 2ms/step - loss: 0.0519 - sparse_categorical_accuracy: 0.9830 - val_loss: 0.0978 - val_sparse_categorical_accuracy: 0.9681 - learning_rate: 1.2500e-04 ## Epoch 205/500 ## 90/90 - 0s - 1ms/step - loss: 0.0558 - sparse_categorical_accuracy: 0.9802 - val_loss: 0.1035 - val_sparse_categorical_accuracy: 0.9667 - learning_rate: 1.2500e-04 ## Epoch 206/500 ## 90/90 - 0s - 1ms/step - loss: 0.0551 - sparse_categorical_accuracy: 0.9816 - val_loss: 0.1006 - val_sparse_categorical_accuracy: 0.9709 - learning_rate: 1.2500e-04 ## Epoch 207/500 ## 90/90 - 0s - 1ms/step - loss: 0.0593 - sparse_categorical_accuracy: 0.9816 - val_loss: 0.1023 - val_sparse_categorical_accuracy: 0.9556 - learning_rate: 1.2500e-04 ## Epoch 208/500 ## 90/90 - 0s - 2ms/step - loss: 0.0556 - sparse_categorical_accuracy: 0.9816 - val_loss: 0.1314 - val_sparse_categorical_accuracy: 0.9626 - learning_rate: 1.2500e-04 ## Epoch 209/500 ## 90/90 - 0s - 1ms/step - loss: 0.0539 - sparse_categorical_accuracy: 0.9809 - val_loss: 0.1529 - val_sparse_categorical_accuracy: 0.9487 - learning_rate: 1.2500e-04 ## Epoch 210/500 ## 90/90 - 0s - 1ms/step - loss: 0.0566 - sparse_categorical_accuracy: 0.9816 - val_loss: 0.0993 - val_sparse_categorical_accuracy: 0.9653 - learning_rate: 1.2500e-04 ## Epoch 211/500 ## 90/90 - 0s - 1ms/step - loss: 0.0552 - sparse_categorical_accuracy: 0.9837 - val_loss: 0.0984 - val_sparse_categorical_accuracy: 0.9639 - learning_rate: 1.2500e-04 ## Epoch 212/500 ## 90/90 - 0s - 1ms/step - loss: 0.0548 - sparse_categorical_accuracy: 0.9826 - val_loss: 0.0999 - val_sparse_categorical_accuracy: 0.9584 - learning_rate: 1.2500e-04 ## Epoch 213/500 ## 90/90 - 0s - 2ms/step - loss: 0.0573 - sparse_categorical_accuracy: 0.9840 - val_loss: 0.0968 - val_sparse_categorical_accuracy: 0.9667 - learning_rate: 1.2500e-04 ## Epoch 214/500 ## 90/90 - 0s - 1ms/step - loss: 0.0568 - sparse_categorical_accuracy: 0.9830 - val_loss: 0.1000 - val_sparse_categorical_accuracy: 0.9639 - learning_rate: 1.2500e-04 ## Epoch 215/500 ## 90/90 - 0s - 1ms/step - loss: 0.0534 - sparse_categorical_accuracy: 0.9816 - val_loss: 0.1031 - val_sparse_categorical_accuracy: 0.9667 - learning_rate: 1.2500e-04 ## Epoch 216/500 ## 90/90 - 0s - 1ms/step - loss: 0.0531 - sparse_categorical_accuracy: 0.9847 - val_loss: 0.1042 - val_sparse_categorical_accuracy: 0.9681 - learning_rate: 1.2500e-04 ## Epoch 217/500 ## 90/90 - 0s - 1ms/step - loss: 0.0532 - sparse_categorical_accuracy: 0.9826 - val_loss: 0.1096 - val_sparse_categorical_accuracy: 0.9695 - learning_rate: 1.2500e-04 ## Epoch 218/500 ## 90/90 - 0s - 1ms/step - loss: 0.0577 - sparse_categorical_accuracy: 0.9809 - val_loss: 0.1009 - val_sparse_categorical_accuracy: 0.9612 - learning_rate: 1.2500e-04 ## Epoch 219/500 ## 90/90 - 0s - 1ms/step - loss: 0.0539 - sparse_categorical_accuracy: 0.9823 - val_loss: 0.1072 - val_sparse_categorical_accuracy: 0.9695 - learning_rate: 1.2500e-04 ## Epoch 220/500 ## 90/90 - 0s - 1ms/step - loss: 0.0529 - sparse_categorical_accuracy: 0.9851 - val_loss: 0.1005 - val_sparse_categorical_accuracy: 0.9695 - learning_rate: 1.2500e-04 ## Epoch 221/500 ## 90/90 - 0s - 1ms/step - loss: 0.0554 - sparse_categorical_accuracy: 0.9799 - val_loss: 0.1036 - val_sparse_categorical_accuracy: 0.9695 - learning_rate: 1.2500e-04 ## Epoch 222/500 ## 90/90 - 0s - 1ms/step - loss: 0.0562 - sparse_categorical_accuracy: 0.9809 - val_loss: 0.1037 - val_sparse_categorical_accuracy: 0.9736 - learning_rate: 1.2500e-04 ## Epoch 223/500 ## 90/90 - 0s - 1ms/step - loss: 0.0544 - sparse_categorical_accuracy: 0.9823 - val_loss: 0.1005 - val_sparse_categorical_accuracy: 0.9736 - learning_rate: 1.2500e-04 ## Epoch 224/500 ## 90/90 - 0s - 1ms/step - loss: 0.0509 - sparse_categorical_accuracy: 0.9847 - val_loss: 0.0969 - val_sparse_categorical_accuracy: 0.9626 - learning_rate: 1.2500e-04 ## Epoch 225/500 ## 90/90 - 0s - 1ms/step - loss: 0.0569 - sparse_categorical_accuracy: 0.9823 - val_loss: 0.1261 - val_sparse_categorical_accuracy: 0.9612 - learning_rate: 1.2500e-04 ## Epoch 226/500 ## 90/90 - 0s - 1ms/step - loss: 0.0519 - sparse_categorical_accuracy: 0.9826 - val_loss: 0.1019 - val_sparse_categorical_accuracy: 0.9570 - learning_rate: 1.2500e-04 ## Epoch 227/500 ## 90/90 - 0s - 1ms/step - loss: 0.0563 - sparse_categorical_accuracy: 0.9819 - val_loss: 0.1037 - val_sparse_categorical_accuracy: 0.9584 - learning_rate: 1.2500e-04 ## Epoch 228/500 ## 90/90 - 0s - 1ms/step - loss: 0.0518 - sparse_categorical_accuracy: 0.9844 - val_loss: 0.1127 - val_sparse_categorical_accuracy: 0.9515 - learning_rate: 1.2500e-04 ## Epoch 229/500 ## 90/90 - 0s - 1ms/step - loss: 0.0521 - sparse_categorical_accuracy: 0.9865 - val_loss: 0.0973 - val_sparse_categorical_accuracy: 0.9639 - learning_rate: 1.2500e-04 ## Epoch 230/500 ## 90/90 - 0s - 1ms/step - loss: 0.0520 - sparse_categorical_accuracy: 0.9847 - val_loss: 0.0999 - val_sparse_categorical_accuracy: 0.9626 - learning_rate: 1.2500e-04 ## Epoch 231/500 ## 90/90 - 0s - 2ms/step - loss: 0.0569 - sparse_categorical_accuracy: 0.9813 - val_loss: 0.0958 - val_sparse_categorical_accuracy: 0.9626 - learning_rate: 1.2500e-04 ## Epoch 232/500 ## 90/90 - 0s - 1ms/step - loss: 0.0530 - sparse_categorical_accuracy: 0.9851 - val_loss: 0.1024 - val_sparse_categorical_accuracy: 0.9681 - learning_rate: 1.2500e-04 ## Epoch 233/500 ## 90/90 - 0s - 1ms/step - loss: 0.0523 - sparse_categorical_accuracy: 0.9844 - val_loss: 0.1019 - val_sparse_categorical_accuracy: 0.9709 - learning_rate: 1.2500e-04 ## Epoch 234/500 ## 90/90 - 0s - 1ms/step - loss: 0.0539 - sparse_categorical_accuracy: 0.9837 - val_loss: 0.1595 - val_sparse_categorical_accuracy: 0.9459 - learning_rate: 1.2500e-04 ## Epoch 235/500 ## 90/90 - 0s - 1ms/step - loss: 0.0543 - sparse_categorical_accuracy: 0.9802 - val_loss: 0.1028 - val_sparse_categorical_accuracy: 0.9653 - learning_rate: 1.2500e-04 ## Epoch 236/500 ## 90/90 - 0s - 1ms/step - loss: 0.0540 - sparse_categorical_accuracy: 0.9826 - val_loss: 0.1068 - val_sparse_categorical_accuracy: 0.9653 - learning_rate: 1.2500e-04 ## Epoch 237/500 ## 90/90 - 0s - 1ms/step - loss: 0.0543 - sparse_categorical_accuracy: 0.9809 - val_loss: 0.1528 - val_sparse_categorical_accuracy: 0.9459 - learning_rate: 1.2500e-04 ## Epoch 238/500 ## 90/90 - 0s - 1ms/step - loss: 0.0527 - sparse_categorical_accuracy: 0.9861 - val_loss: 0.0977 - val_sparse_categorical_accuracy: 0.9695 - learning_rate: 1.2500e-04 ## Epoch 239/500 ## 90/90 - 0s - 1ms/step - loss: 0.0537 - sparse_categorical_accuracy: 0.9813 - val_loss: 0.0980 - val_sparse_categorical_accuracy: 0.9598 - learning_rate: 1.2500e-04 ## Epoch 240/500 ## 90/90 - 0s - 1ms/step - loss: 0.0512 - sparse_categorical_accuracy: 0.9854 - val_loss: 0.1054 - val_sparse_categorical_accuracy: 0.9709 - learning_rate: 1.2500e-04 ## Epoch 241/500 ## 90/90 - 0s - 1ms/step - loss: 0.0541 - sparse_categorical_accuracy: 0.9837 - val_loss: 0.1325 - val_sparse_categorical_accuracy: 0.9487 - learning_rate: 1.2500e-04 ## Epoch 242/500 ## 90/90 - 0s - 1ms/step - loss: 0.0535 - sparse_categorical_accuracy: 0.9826 - val_loss: 0.1028 - val_sparse_categorical_accuracy: 0.9681 - learning_rate: 1.2500e-04 ## Epoch 243/500 ## 90/90 - 0s - 1ms/step - loss: 0.0555 - sparse_categorical_accuracy: 0.9823 - val_loss: 0.0991 - val_sparse_categorical_accuracy: 0.9626 - learning_rate: 1.2500e-04 ## Epoch 244/500 ## 90/90 - 0s - 1ms/step - loss: 0.0572 - sparse_categorical_accuracy: 0.9795 - val_loss: 0.0986 - val_sparse_categorical_accuracy: 0.9639 - learning_rate: 1.2500e-04 ## Epoch 245/500 ## 90/90 - 0s - 1ms/step - loss: 0.0520 - sparse_categorical_accuracy: 0.9833 - val_loss: 0.0980 - val_sparse_categorical_accuracy: 0.9626 - learning_rate: 1.2500e-04 ## Epoch 246/500 ## 90/90 - 0s - 1ms/step - loss: 0.0554 - sparse_categorical_accuracy: 0.9809 - val_loss: 0.0978 - val_sparse_categorical_accuracy: 0.9639 - learning_rate: 1.2500e-04 ## Epoch 247/500 ## 90/90 - 0s - 1ms/step - loss: 0.0498 - sparse_categorical_accuracy: 0.9837 - val_loss: 0.1015 - val_sparse_categorical_accuracy: 0.9681 - learning_rate: 1.2500e-04 ## Epoch 248/500 ## 90/90 - 0s - 1ms/step - loss: 0.0493 - sparse_categorical_accuracy: 0.9840 - val_loss: 0.0994 - val_sparse_categorical_accuracy: 0.9612 - learning_rate: 1.2500e-04 ## Epoch 249/500 ## 90/90 - 0s - 1ms/step - loss: 0.0543 - sparse_categorical_accuracy: 0.9847 - val_loss: 0.1054 - val_sparse_categorical_accuracy: 0.9695 - learning_rate: 1.2500e-04 ## Epoch 250/500 ## 90/90 - 0s - 1ms/step - loss: 0.0550 - sparse_categorical_accuracy: 0.9802 - val_loss: 0.0989 - val_sparse_categorical_accuracy: 0.9626 - learning_rate: 1.2500e-04 ## Epoch 251/500 ## 90/90 - 0s - 1ms/step - loss: 0.0539 - sparse_categorical_accuracy: 0.9819 - val_loss: 0.1018 - val_sparse_categorical_accuracy: 0.9667 - learning_rate: 1.2500e-04 ## Epoch 252/500 ## 90/90 - 0s - 1ms/step - loss: 0.0542 - sparse_categorical_accuracy: 0.9830 - val_loss: 0.0975 - val_sparse_categorical_accuracy: 0.9667 - learning_rate: 1.0000e-04 ## Epoch 253/500 ## 90/90 - 0s - 1ms/step - loss: 0.0538 - sparse_categorical_accuracy: 0.9830 - val_loss: 0.1065 - val_sparse_categorical_accuracy: 0.9723 - learning_rate: 1.0000e-04 ## Epoch 254/500 ## 90/90 - 0s - 1ms/step - loss: 0.0542 - sparse_categorical_accuracy: 0.9823 - val_loss: 0.1145 - val_sparse_categorical_accuracy: 0.9528 - learning_rate: 1.0000e-04 ## Epoch 255/500 ## 90/90 - 0s - 1ms/step - loss: 0.0506 - sparse_categorical_accuracy: 0.9819 - val_loss: 0.0976 - val_sparse_categorical_accuracy: 0.9681 - learning_rate: 1.0000e-04 ## Epoch 256/500 ## 90/90 - 0s - 1ms/step - loss: 0.0521 - sparse_categorical_accuracy: 0.9847 - val_loss: 0.1117 - val_sparse_categorical_accuracy: 0.9626 - learning_rate: 1.0000e-04 ## Epoch 257/500 ## 90/90 - 0s - 1ms/step - loss: 0.0504 - sparse_categorical_accuracy: 0.9854 - val_loss: 0.1201 - val_sparse_categorical_accuracy: 0.9626 - learning_rate: 1.0000e-04 ## Epoch 258/500 ## 90/90 - 0s - 1ms/step - loss: 0.0498 - sparse_categorical_accuracy: 0.9851 - val_loss: 0.1039 - val_sparse_categorical_accuracy: 0.9584 - learning_rate: 1.0000e-04 ## Epoch 259/500 ## 90/90 - 0s - 1ms/step - loss: 0.0479 - sparse_categorical_accuracy: 0.9868 - val_loss: 0.1055 - val_sparse_categorical_accuracy: 0.9695 - learning_rate: 1.0000e-04 ## Epoch 260/500 ## 90/90 - 0s - 1ms/step - loss: 0.0499 - sparse_categorical_accuracy: 0.9840 - val_loss: 0.1046 - val_sparse_categorical_accuracy: 0.9556 - learning_rate: 1.0000e-04 ## Epoch 261/500 ## 90/90 - 0s - 1ms/step - loss: 0.0529 - sparse_categorical_accuracy: 0.9823 - val_loss: 0.1022 - val_sparse_categorical_accuracy: 0.9667 - learning_rate: 1.0000e-04 ## Epoch 262/500 ## 90/90 - 0s - 1ms/step - loss: 0.0492 - sparse_categorical_accuracy: 0.9840 - val_loss: 0.1063 - val_sparse_categorical_accuracy: 0.9681 - learning_rate: 1.0000e-04 ## Epoch 263/500 ## 90/90 - 0s - 1ms/step - loss: 0.0500 - sparse_categorical_accuracy: 0.9833 - val_loss: 0.1006 - val_sparse_categorical_accuracy: 0.9626 - learning_rate: 1.0000e-04 ## Epoch 264/500 ## 90/90 - 0s - 1ms/step - loss: 0.0533 - sparse_categorical_accuracy: 0.9823 - val_loss: 0.1088 - val_sparse_categorical_accuracy: 0.9709 - learning_rate: 1.0000e-04 ## Epoch 265/500 ## 90/90 - 0s - 1ms/step - loss: 0.0549 - sparse_categorical_accuracy: 0.9802 - val_loss: 0.1010 - val_sparse_categorical_accuracy: 0.9709 - learning_rate: 1.0000e-04 ## Epoch 266/500 ## 90/90 - 0s - 1ms/step - loss: 0.0512 - sparse_categorical_accuracy: 0.9830 - val_loss: 0.1016 - val_sparse_categorical_accuracy: 0.9570 - learning_rate: 1.0000e-04 ## Epoch 267/500 ## 90/90 - 0s - 1ms/step - loss: 0.0496 - sparse_categorical_accuracy: 0.9847 - val_loss: 0.1012 - val_sparse_categorical_accuracy: 0.9570 - learning_rate: 1.0000e-04 ## Epoch 268/500 ## 90/90 - 0s - 1ms/step - loss: 0.0477 - sparse_categorical_accuracy: 0.9872 - val_loss: 0.1016 - val_sparse_categorical_accuracy: 0.9570 - learning_rate: 1.0000e-04 ## Epoch 269/500 ## 90/90 - 0s - 1ms/step - loss: 0.0523 - sparse_categorical_accuracy: 0.9861 - val_loss: 0.1012 - val_sparse_categorical_accuracy: 0.9723 - learning_rate: 1.0000e-04 ## Epoch 270/500 ## 90/90 - 0s - 1ms/step - loss: 0.0474 - sparse_categorical_accuracy: 0.9889 - val_loss: 0.0975 - val_sparse_categorical_accuracy: 0.9653 - learning_rate: 1.0000e-04 ## Epoch 271/500 ## 90/90 - 0s - 1ms/step - loss: 0.0489 - sparse_categorical_accuracy: 0.9840 - val_loss: 0.1055 - val_sparse_categorical_accuracy: 0.9695 - learning_rate: 1.0000e-04 ## Epoch 272/500 ## 90/90 - 0s - 1ms/step - loss: 0.0503 - sparse_categorical_accuracy: 0.9826 - val_loss: 0.1050 - val_sparse_categorical_accuracy: 0.9709 - learning_rate: 1.0000e-04 ## Epoch 273/500 ## 90/90 - 0s - 1ms/step - loss: 0.0514 - sparse_categorical_accuracy: 0.9844 - val_loss: 0.0980 - val_sparse_categorical_accuracy: 0.9695 - learning_rate: 1.0000e-04 ## Epoch 274/500 ## 90/90 - 0s - 1ms/step - loss: 0.0479 - sparse_categorical_accuracy: 0.9854 - val_loss: 0.1049 - val_sparse_categorical_accuracy: 0.9639 - learning_rate: 1.0000e-04 ## Epoch 275/500 ## 90/90 - 0s - 1ms/step - loss: 0.0467 - sparse_categorical_accuracy: 0.9840 - val_loss: 0.1473 - val_sparse_categorical_accuracy: 0.9556 - learning_rate: 1.0000e-04 ## Epoch 276/500 ## 90/90 - 0s - 1ms/step - loss: 0.0504 - sparse_categorical_accuracy: 0.9819 - val_loss: 0.1065 - val_sparse_categorical_accuracy: 0.9695 - learning_rate: 1.0000e-04 ## Epoch 277/500 ## 90/90 - 0s - 1ms/step - loss: 0.0505 - sparse_categorical_accuracy: 0.9847 - val_loss: 0.1187 - val_sparse_categorical_accuracy: 0.9612 - learning_rate: 1.0000e-04 ## Epoch 278/500 ## 90/90 - 0s - 1ms/step - loss: 0.0498 - sparse_categorical_accuracy: 0.9840 - val_loss: 0.1267 - val_sparse_categorical_accuracy: 0.9598 - learning_rate: 1.0000e-04 ## Epoch 279/500 ## 90/90 - 0s - 1ms/step - loss: 0.0517 - sparse_categorical_accuracy: 0.9830 - val_loss: 0.1037 - val_sparse_categorical_accuracy: 0.9653 - learning_rate: 1.0000e-04 ## Epoch 280/500 ## 90/90 - 0s - 1ms/step - loss: 0.0501 - sparse_categorical_accuracy: 0.9830 - val_loss: 0.0985 - val_sparse_categorical_accuracy: 0.9667 - learning_rate: 1.0000e-04 ## Epoch 281/500 ## 90/90 - 0s - 1ms/step - loss: 0.0468 - sparse_categorical_accuracy: 0.9865 - val_loss: 0.1119 - val_sparse_categorical_accuracy: 0.9653 - learning_rate: 1.0000e-04 ## Epoch 281: early stopping"},{"path":"https://keras3.posit.co/dev/articles/examples/timeseries/timeseries_classification_from_scratch.html","id":"evaluate-model-on-test-data","dir":"Articles > Examples > Timeseries","previous_headings":"","what":"Evaluate model on test data","title":"Timeseries classification from scratch","text":"","code":"model <- load_model(\"best_model.keras\")  results <- model |> evaluate(x_test, y_test) ## 42/42 - 1s - 16ms/step - loss: 0.0938 - sparse_categorical_accuracy: 0.9697 str(results) ## List of 2 ##  $ loss                       : num 0.0938 ##  $ sparse_categorical_accuracy: num 0.97 cat(   \"Test accuracy: \", results$sparse_categorical_accuracy, \"\\n\",   \"Test loss: \", results$loss, \"\\n\",   sep = \"\" ) ## Test accuracy: 0.969697 ## Test loss: 0.09383929"},{"path":"https://keras3.posit.co/dev/articles/examples/timeseries/timeseries_classification_from_scratch.html","id":"plot-the-models-training-history","dir":"Articles > Examples > Timeseries","previous_headings":"","what":"Plot the model’s training history","title":"Timeseries classification from scratch","text":"Plot just training validation accuracy: can see training accuracy reaches almost 0.95 100 epochs. However, observing validation accuracy can see network still needs training reaches almost 0.97 validation training accuracy 200 epochs. Beyond 200th epoch, continue training, validation accuracy start decreasing training accuracy continue increasing: model starts overfitting.","code":"plot(history) plot(history, metric = \"sparse_categorical_accuracy\") +   # scale x axis to actual number of epochs run before early stopping   ggplot2::xlim(0, length(history$metrics$loss))"},{"path":"https://keras3.posit.co/dev/articles/examples/vision/autoencoder.html","id":"introduction","dir":"Articles > Examples > Vision","previous_headings":"","what":"Introduction","title":"Convolutional autoencoder for image denoising","text":"example demonstrates implement deep convolutional autoencoder image denoising, mapping noisy digits images MNIST dataset clean digits images. implementation based original blog post titled Building Autoencoders Keras François Chollet.","code":""},{"path":"https://keras3.posit.co/dev/articles/examples/vision/autoencoder.html","id":"setup","dir":"Articles > Examples > Vision","previous_headings":"","what":"Setup","title":"Convolutional autoencoder for image denoising","text":"","code":"library(keras3)  # Normalizes the supplied array and reshapes it. preprocess <- function(array) {   array_reshape(array/255, c(dim(array)[1], 28, 28, 1)) }  # Adds random noise to each image in the supplied array. noise <- function(array) {   noise_factor <- 0.4   noisy_array <- array + noise_factor * random_normal(dim(array))   op_clip(noisy_array, 0.0, 1.0) }  display <- function(array1, array2) {   n <- 2   indices <- sample.int(dim(array1)[1], n)   images1 <- as.array(array1)[indices, , , ]   images2 <- as.array(array2)[indices, , , ]    par(mfrow = c(2, n), mar = c(0, 0, 0, 0))   for (i in seq_len(n)) {     plot(as.raster(images1[i, , ]))     plot(as.raster(images2[i, , ]))   } }"},{"path":"https://keras3.posit.co/dev/articles/examples/vision/autoencoder.html","id":"prepare-the-data","dir":"Articles > Examples > Vision","previous_headings":"","what":"Prepare the data","title":"Convolutional autoencoder for image denoising","text":"","code":"# Since we only need images from the dataset to encode and decode, we # won't use the labels. c(c(train_data, .), c(test_data, .)) %<-% dataset_mnist()  # Normalize and reshape the data train_data <- preprocess(train_data) test_data <- preprocess(test_data)  # Create a copy of the data with added noise noisy_train_data <- noise(train_data) noisy_test_data <- noise(test_data)  # Display the train data and a version of it with added noise display(train_data, noisy_train_data)"},{"path":"https://keras3.posit.co/dev/articles/examples/vision/autoencoder.html","id":"build-the-autoencoder","dir":"Articles > Examples > Vision","previous_headings":"","what":"Build the autoencoder","title":"Convolutional autoencoder for image denoising","text":"going use Functional API build convolutional autoencoder. Now can train autoencoder using train_data input data target. Notice setting validation data using format. Let’s predict test dataset display original image together prediction autoencoder. Notice predictions pretty close original images, although quite . Now know autoencoder works, let’s retrain using noisy data input clean data target. want autoencoder learn denoise images. Let’s now predict noisy data display results autoencoder. Notice autoencoder amazing job removing noise input images.","code":"input <- keras_input(shape = c(28, 28, 1))  # Encoder enc <- input |>   layer_conv_2d(filters = 32, kernel_size = c(3, 3),                 activation = \"relu\", padding = \"same\") |>   layer_max_pooling_2d(pool_size = c(2, 2), padding = \"same\") |>   layer_conv_2d(filters = 32, kernel_size = c(3, 3),                 activation = \"relu\", padding = \"same\") |>   layer_max_pooling_2d(pool_size = c(2, 2), padding = \"same\")  # Decoder dec <- enc |>   layer_conv_2d_transpose(filters = 32, kernel_size = c(3, 3), strides = 2,                           activation = \"relu\", padding = \"same\") |>   layer_conv_2d_transpose(filters = 32, kernel_size = c(3, 3), strides = 2,                           activation = \"relu\", padding = \"same\") |>   layer_conv_2d(filters = 1, kernel_size = c(3, 3),                 activation = \"sigmoid\", padding = \"same\")  # Autoencoder autoencoder <- keras_model(input, dec) autoencoder |> compile(optimizer = \"adam\", loss = \"binary_crossentropy\") autoencoder |> summary() ## Model: \"functional\" ## ┏━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━┓ ## ┃ Layer (type)                    ┃ Output Shape           ┃       Param # ┃ ## ┡━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━┩ ## │ input_layer (InputLayer)        │ (None, 28, 28, 1)      │             0 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ conv2d (Conv2D)                 │ (None, 28, 28, 32)     │           320 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ max_pooling2d (MaxPooling2D)    │ (None, 14, 14, 32)     │             0 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ conv2d_1 (Conv2D)               │ (None, 14, 14, 32)     │         9,248 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ max_pooling2d_1 (MaxPooling2D)  │ (None, 7, 7, 32)       │             0 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ conv2d_transpose                │ (None, 14, 14, 32)     │         9,248 │ ## │ (Conv2DTranspose)               │                        │               │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ conv2d_transpose_1              │ (None, 28, 28, 32)     │         9,248 │ ## │ (Conv2DTranspose)               │                        │               │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ conv2d_2 (Conv2D)               │ (None, 28, 28, 1)      │           289 │ ## └─────────────────────────────────┴────────────────────────┴───────────────┘ ##  Total params: 28,353 (110.75 KB) ##  Trainable params: 28,353 (110.75 KB) ##  Non-trainable params: 0 (0.00 B) autoencoder |> fit(   x = train_data,   y = train_data,   epochs = 50,   batch_size = 128,   shuffle = TRUE,   validation_data = list(test_data, test_data), ) ## Epoch 1/50 ## 469/469 - 6s - 12ms/step - loss: 0.1371 - val_loss: 0.0727 ## Epoch 2/50 ## 469/469 - 2s - 3ms/step - loss: 0.0712 - val_loss: 0.0692 ## Epoch 3/50 ## 469/469 - 2s - 3ms/step - loss: 0.0690 - val_loss: 0.0678 ## Epoch 4/50 ## 469/469 - 2s - 3ms/step - loss: 0.0679 - val_loss: 0.0671 ## Epoch 5/50 ## 469/469 - 2s - 3ms/step - loss: 0.0672 - val_loss: 0.0665 ## Epoch 6/50 ## 469/469 - 2s - 3ms/step - loss: 0.0666 - val_loss: 0.0660 ## Epoch 7/50 ## 469/469 - 2s - 3ms/step - loss: 0.0662 - val_loss: 0.0656 ## Epoch 8/50 ## 469/469 - 2s - 3ms/step - loss: 0.0659 - val_loss: 0.0652 ## Epoch 9/50 ## 469/469 - 2s - 3ms/step - loss: 0.0656 - val_loss: 0.0650 ## Epoch 10/50 ## 469/469 - 1s - 3ms/step - loss: 0.0653 - val_loss: 0.0648 ## Epoch 11/50 ## 469/469 - 2s - 3ms/step - loss: 0.0651 - val_loss: 0.0646 ## Epoch 12/50 ## 469/469 - 1s - 3ms/step - loss: 0.0649 - val_loss: 0.0644 ## Epoch 13/50 ## 469/469 - 2s - 3ms/step - loss: 0.0648 - val_loss: 0.0643 ## Epoch 14/50 ## 469/469 - 1s - 3ms/step - loss: 0.0646 - val_loss: 0.0641 ## Epoch 15/50 ## 469/469 - 1s - 3ms/step - loss: 0.0645 - val_loss: 0.0640 ## Epoch 16/50 ## 469/469 - 1s - 3ms/step - loss: 0.0644 - val_loss: 0.0639 ## Epoch 17/50 ## 469/469 - 2s - 3ms/step - loss: 0.0642 - val_loss: 0.0638 ## Epoch 18/50 ## 469/469 - 2s - 3ms/step - loss: 0.0641 - val_loss: 0.0637 ## Epoch 19/50 ## 469/469 - 1s - 3ms/step - loss: 0.0640 - val_loss: 0.0636 ## Epoch 20/50 ## 469/469 - 2s - 3ms/step - loss: 0.0639 - val_loss: 0.0635 ## Epoch 21/50 ## 469/469 - 2s - 3ms/step - loss: 0.0639 - val_loss: 0.0635 ## Epoch 22/50 ## 469/469 - 1s - 3ms/step - loss: 0.0638 - val_loss: 0.0634 ## Epoch 23/50 ## 469/469 - 2s - 3ms/step - loss: 0.0637 - val_loss: 0.0634 ## Epoch 24/50 ## 469/469 - 2s - 3ms/step - loss: 0.0637 - val_loss: 0.0633 ## Epoch 25/50 ## 469/469 - 2s - 3ms/step - loss: 0.0636 - val_loss: 0.0632 ## Epoch 26/50 ## 469/469 - 2s - 3ms/step - loss: 0.0636 - val_loss: 0.0632 ## Epoch 27/50 ## 469/469 - 2s - 3ms/step - loss: 0.0635 - val_loss: 0.0631 ## Epoch 28/50 ## 469/469 - 2s - 3ms/step - loss: 0.0635 - val_loss: 0.0631 ## Epoch 29/50 ## 469/469 - 2s - 3ms/step - loss: 0.0634 - val_loss: 0.0631 ## Epoch 30/50 ## 469/469 - 1s - 3ms/step - loss: 0.0634 - val_loss: 0.0630 ## Epoch 31/50 ## 469/469 - 2s - 3ms/step - loss: 0.0633 - val_loss: 0.0630 ## Epoch 32/50 ## 469/469 - 2s - 3ms/step - loss: 0.0633 - val_loss: 0.0630 ## Epoch 33/50 ## 469/469 - 2s - 3ms/step - loss: 0.0633 - val_loss: 0.0629 ## Epoch 34/50 ## 469/469 - 2s - 3ms/step - loss: 0.0632 - val_loss: 0.0629 ## Epoch 35/50 ## 469/469 - 1s - 3ms/step - loss: 0.0632 - val_loss: 0.0628 ## Epoch 36/50 ## 469/469 - 2s - 3ms/step - loss: 0.0632 - val_loss: 0.0628 ## Epoch 37/50 ## 469/469 - 1s - 3ms/step - loss: 0.0631 - val_loss: 0.0628 ## Epoch 38/50 ## 469/469 - 2s - 3ms/step - loss: 0.0631 - val_loss: 0.0628 ## Epoch 39/50 ## 469/469 - 1s - 3ms/step - loss: 0.0631 - val_loss: 0.0627 ## Epoch 40/50 ## 469/469 - 2s - 3ms/step - loss: 0.0630 - val_loss: 0.0627 ## Epoch 41/50 ## 469/469 - 2s - 3ms/step - loss: 0.0630 - val_loss: 0.0627 ## Epoch 42/50 ## 469/469 - 2s - 3ms/step - loss: 0.0630 - val_loss: 0.0626 ## Epoch 43/50 ## 469/469 - 2s - 3ms/step - loss: 0.0630 - val_loss: 0.0626 ## Epoch 44/50 ## 469/469 - 2s - 3ms/step - loss: 0.0629 - val_loss: 0.0626 ## Epoch 45/50 ## 469/469 - 2s - 3ms/step - loss: 0.0629 - val_loss: 0.0626 ## Epoch 46/50 ## 469/469 - 2s - 3ms/step - loss: 0.0629 - val_loss: 0.0625 ## Epoch 47/50 ## 469/469 - 2s - 3ms/step - loss: 0.0629 - val_loss: 0.0625 ## Epoch 48/50 ## 469/469 - 1s - 3ms/step - loss: 0.0628 - val_loss: 0.0625 ## Epoch 49/50 ## 469/469 - 2s - 3ms/step - loss: 0.0628 - val_loss: 0.0625 ## Epoch 50/50 ## 469/469 - 2s - 3ms/step - loss: 0.0628 - val_loss: 0.0625 predictions <- autoencoder |> predict(test_data) ## 313/313 - 1s - 2ms/step display(test_data, predictions) autoencoder |> fit(   x = noisy_train_data,   y = train_data,   epochs = 100,   batch_size = 128,   shuffle = TRUE,   validation_data = list(noisy_test_data, test_data), ) ## Epoch 1/100 ## 469/469 - 2s - 3ms/step - loss: 0.1006 - val_loss: 0.0939 ## Epoch 2/100 ## 469/469 - 2s - 3ms/step - loss: 0.0936 - val_loss: 0.0919 ## Epoch 3/100 ## 469/469 - 2s - 3ms/step - loss: 0.0921 - val_loss: 0.0908 ## Epoch 4/100 ## 469/469 - 2s - 3ms/step - loss: 0.0912 - val_loss: 0.0902 ## Epoch 5/100 ## 469/469 - 2s - 3ms/step - loss: 0.0905 - val_loss: 0.0896 ## Epoch 6/100 ## 469/469 - 2s - 3ms/step - loss: 0.0900 - val_loss: 0.0891 ## Epoch 7/100 ## 469/469 - 2s - 3ms/step - loss: 0.0895 - val_loss: 0.0887 ## Epoch 8/100 ## 469/469 - 2s - 3ms/step - loss: 0.0891 - val_loss: 0.0883 ## Epoch 9/100 ## 469/469 - 2s - 3ms/step - loss: 0.0887 - val_loss: 0.0880 ## Epoch 10/100 ## 469/469 - 2s - 3ms/step - loss: 0.0884 - val_loss: 0.0878 ## Epoch 11/100 ## 469/469 - 2s - 3ms/step - loss: 0.0882 - val_loss: 0.0875 ## Epoch 12/100 ## 469/469 - 2s - 3ms/step - loss: 0.0879 - val_loss: 0.0873 ## Epoch 13/100 ## 469/469 - 2s - 3ms/step - loss: 0.0877 - val_loss: 0.0871 ## Epoch 14/100 ## 469/469 - 2s - 3ms/step - loss: 0.0875 - val_loss: 0.0869 ## Epoch 15/100 ## 469/469 - 2s - 3ms/step - loss: 0.0873 - val_loss: 0.0867 ## Epoch 16/100 ## 469/469 - 2s - 3ms/step - loss: 0.0872 - val_loss: 0.0866 ## Epoch 17/100 ## 469/469 - 2s - 3ms/step - loss: 0.0870 - val_loss: 0.0864 ## Epoch 18/100 ## 469/469 - 2s - 3ms/step - loss: 0.0869 - val_loss: 0.0863 ## Epoch 19/100 ## 469/469 - 2s - 3ms/step - loss: 0.0868 - val_loss: 0.0862 ## Epoch 20/100 ## 469/469 - 2s - 3ms/step - loss: 0.0866 - val_loss: 0.0861 ## Epoch 21/100 ## 469/469 - 2s - 3ms/step - loss: 0.0865 - val_loss: 0.0860 ## Epoch 22/100 ## 469/469 - 2s - 3ms/step - loss: 0.0864 - val_loss: 0.0859 ## Epoch 23/100 ## 469/469 - 2s - 3ms/step - loss: 0.0863 - val_loss: 0.0858 ## Epoch 24/100 ## 469/469 - 1s - 3ms/step - loss: 0.0862 - val_loss: 0.0857 ## Epoch 25/100 ## 469/469 - 2s - 3ms/step - loss: 0.0862 - val_loss: 0.0856 ## Epoch 26/100 ## 469/469 - 2s - 3ms/step - loss: 0.0861 - val_loss: 0.0856 ## Epoch 27/100 ## 469/469 - 2s - 3ms/step - loss: 0.0860 - val_loss: 0.0855 ## Epoch 28/100 ## 469/469 - 2s - 3ms/step - loss: 0.0859 - val_loss: 0.0855 ## Epoch 29/100 ## 469/469 - 2s - 3ms/step - loss: 0.0859 - val_loss: 0.0854 ## Epoch 30/100 ## 469/469 - 2s - 3ms/step - loss: 0.0858 - val_loss: 0.0853 ## Epoch 31/100 ## 469/469 - 2s - 3ms/step - loss: 0.0858 - val_loss: 0.0853 ## Epoch 32/100 ## 469/469 - 2s - 3ms/step - loss: 0.0857 - val_loss: 0.0853 ## Epoch 33/100 ## 469/469 - 2s - 3ms/step - loss: 0.0857 - val_loss: 0.0852 ## Epoch 34/100 ## 469/469 - 2s - 3ms/step - loss: 0.0856 - val_loss: 0.0852 ## Epoch 35/100 ## 469/469 - 2s - 3ms/step - loss: 0.0856 - val_loss: 0.0851 ## Epoch 36/100 ## 469/469 - 2s - 3ms/step - loss: 0.0855 - val_loss: 0.0851 ## Epoch 37/100 ## 469/469 - 2s - 3ms/step - loss: 0.0855 - val_loss: 0.0851 ## Epoch 38/100 ## 469/469 - 2s - 3ms/step - loss: 0.0855 - val_loss: 0.0850 ## Epoch 39/100 ## 469/469 - 2s - 3ms/step - loss: 0.0854 - val_loss: 0.0850 ## Epoch 40/100 ## 469/469 - 2s - 3ms/step - loss: 0.0854 - val_loss: 0.0850 ## Epoch 41/100 ## 469/469 - 2s - 3ms/step - loss: 0.0854 - val_loss: 0.0850 ## Epoch 42/100 ## 469/469 - 2s - 3ms/step - loss: 0.0853 - val_loss: 0.0849 ## Epoch 43/100 ## 469/469 - 2s - 3ms/step - loss: 0.0853 - val_loss: 0.0849 ## Epoch 44/100 ## 469/469 - 2s - 3ms/step - loss: 0.0853 - val_loss: 0.0849 ## Epoch 45/100 ## 469/469 - 2s - 3ms/step - loss: 0.0853 - val_loss: 0.0849 ## Epoch 46/100 ## 469/469 - 2s - 3ms/step - loss: 0.0852 - val_loss: 0.0848 ## Epoch 47/100 ## 469/469 - 2s - 3ms/step - loss: 0.0852 - val_loss: 0.0848 ## Epoch 48/100 ## 469/469 - 2s - 3ms/step - loss: 0.0852 - val_loss: 0.0848 ## Epoch 49/100 ## 469/469 - 2s - 3ms/step - loss: 0.0852 - val_loss: 0.0848 ## Epoch 50/100 ## 469/469 - 2s - 3ms/step - loss: 0.0851 - val_loss: 0.0848 ## Epoch 51/100 ## 469/469 - 2s - 3ms/step - loss: 0.0851 - val_loss: 0.0847 ## Epoch 52/100 ## 469/469 - 2s - 3ms/step - loss: 0.0851 - val_loss: 0.0847 ## Epoch 53/100 ## 469/469 - 2s - 3ms/step - loss: 0.0851 - val_loss: 0.0847 ## Epoch 54/100 ## 469/469 - 2s - 3ms/step - loss: 0.0850 - val_loss: 0.0847 ## Epoch 55/100 ## 469/469 - 2s - 3ms/step - loss: 0.0850 - val_loss: 0.0847 ## Epoch 56/100 ## 469/469 - 2s - 3ms/step - loss: 0.0850 - val_loss: 0.0846 ## Epoch 57/100 ## 469/469 - 2s - 3ms/step - loss: 0.0850 - val_loss: 0.0846 ## Epoch 58/100 ## 469/469 - 2s - 3ms/step - loss: 0.0850 - val_loss: 0.0846 ## Epoch 59/100 ## 469/469 - 2s - 3ms/step - loss: 0.0849 - val_loss: 0.0846 ## Epoch 60/100 ## 469/469 - 2s - 3ms/step - loss: 0.0849 - val_loss: 0.0846 ## Epoch 61/100 ## 469/469 - 2s - 3ms/step - loss: 0.0849 - val_loss: 0.0846 ## Epoch 62/100 ## 469/469 - 2s - 3ms/step - loss: 0.0849 - val_loss: 0.0846 ## Epoch 63/100 ## 469/469 - 1s - 3ms/step - loss: 0.0849 - val_loss: 0.0845 ## Epoch 64/100 ## 469/469 - 2s - 3ms/step - loss: 0.0849 - val_loss: 0.0845 ## Epoch 65/100 ## 469/469 - 2s - 3ms/step - loss: 0.0848 - val_loss: 0.0845 ## Epoch 66/100 ## 469/469 - 2s - 3ms/step - loss: 0.0848 - val_loss: 0.0845 ## Epoch 67/100 ## 469/469 - 2s - 3ms/step - loss: 0.0848 - val_loss: 0.0845 ## Epoch 68/100 ## 469/469 - 2s - 3ms/step - loss: 0.0848 - val_loss: 0.0845 ## Epoch 69/100 ## 469/469 - 2s - 3ms/step - loss: 0.0848 - val_loss: 0.0845 ## Epoch 70/100 ## 469/469 - 2s - 3ms/step - loss: 0.0848 - val_loss: 0.0845 ## Epoch 71/100 ## 469/469 - 2s - 3ms/step - loss: 0.0848 - val_loss: 0.0844 ## Epoch 72/100 ## 469/469 - 2s - 3ms/step - loss: 0.0847 - val_loss: 0.0844 ## Epoch 73/100 ## 469/469 - 2s - 3ms/step - loss: 0.0847 - val_loss: 0.0844 ## Epoch 74/100 ## 469/469 - 2s - 3ms/step - loss: 0.0847 - val_loss: 0.0844 ## Epoch 75/100 ## 469/469 - 1s - 3ms/step - loss: 0.0847 - val_loss: 0.0844 ## Epoch 76/100 ## 469/469 - 2s - 3ms/step - loss: 0.0847 - val_loss: 0.0844 ## Epoch 77/100 ## 469/469 - 2s - 3ms/step - loss: 0.0847 - val_loss: 0.0844 ## Epoch 78/100 ## 469/469 - 2s - 3ms/step - loss: 0.0847 - val_loss: 0.0844 ## Epoch 79/100 ## 469/469 - 2s - 3ms/step - loss: 0.0847 - val_loss: 0.0844 ## Epoch 80/100 ## 469/469 - 2s - 3ms/step - loss: 0.0847 - val_loss: 0.0844 ## Epoch 81/100 ## 469/469 - 2s - 3ms/step - loss: 0.0846 - val_loss: 0.0844 ## Epoch 82/100 ## 469/469 - 2s - 3ms/step - loss: 0.0846 - val_loss: 0.0844 ## Epoch 83/100 ## 469/469 - 2s - 3ms/step - loss: 0.0846 - val_loss: 0.0844 ## Epoch 84/100 ## 469/469 - 2s - 3ms/step - loss: 0.0846 - val_loss: 0.0843 ## Epoch 85/100 ## 469/469 - 2s - 3ms/step - loss: 0.0846 - val_loss: 0.0843 ## Epoch 86/100 ## 469/469 - 2s - 3ms/step - loss: 0.0846 - val_loss: 0.0843 ## Epoch 87/100 ## 469/469 - 1s - 3ms/step - loss: 0.0846 - val_loss: 0.0843 ## Epoch 88/100 ## 469/469 - 2s - 3ms/step - loss: 0.0846 - val_loss: 0.0843 ## Epoch 89/100 ## 469/469 - 2s - 3ms/step - loss: 0.0846 - val_loss: 0.0843 ## Epoch 90/100 ## 469/469 - 2s - 3ms/step - loss: 0.0846 - val_loss: 0.0843 ## Epoch 91/100 ## 469/469 - 2s - 3ms/step - loss: 0.0845 - val_loss: 0.0843 ## Epoch 92/100 ## 469/469 - 2s - 3ms/step - loss: 0.0845 - val_loss: 0.0843 ## Epoch 93/100 ## 469/469 - 2s - 3ms/step - loss: 0.0845 - val_loss: 0.0843 ## Epoch 94/100 ## 469/469 - 2s - 3ms/step - loss: 0.0845 - val_loss: 0.0843 ## Epoch 95/100 ## 469/469 - 2s - 3ms/step - loss: 0.0845 - val_loss: 0.0843 ## Epoch 96/100 ## 469/469 - 1s - 3ms/step - loss: 0.0845 - val_loss: 0.0842 ## Epoch 97/100 ## 469/469 - 2s - 3ms/step - loss: 0.0845 - val_loss: 0.0842 ## Epoch 98/100 ## 469/469 - 2s - 3ms/step - loss: 0.0845 - val_loss: 0.0842 ## Epoch 99/100 ## 469/469 - 2s - 3ms/step - loss: 0.0845 - val_loss: 0.0842 ## Epoch 100/100 ## 469/469 - 1s - 3ms/step - loss: 0.0845 - val_loss: 0.0842 predictions <- autoencoder |> predict(noisy_test_data) ## 313/313 - 0s - 775us/step display(noisy_test_data, predictions)"},{"path":"https://keras3.posit.co/dev/articles/examples/vision/mnist_convnet.html","id":"setup","dir":"Articles > Examples > Vision","previous_headings":"","what":"Setup","title":"Simple MNIST convnet","text":"","code":"library(keras3)"},{"path":"https://keras3.posit.co/dev/articles/examples/vision/mnist_convnet.html","id":"prepare-the-data","dir":"Articles > Examples > Vision","previous_headings":"","what":"Prepare the data","title":"Simple MNIST convnet","text":"","code":"# Model / data parameters num_classes <- 10 input_shape <- c(28, 28, 1)  # Load the data and split it between train and test sets c(c(x_train, y_train), c(x_test, y_test)) %<-% dataset_mnist()  # Scale images to the [0, 1] range x_train <- x_train / 255 x_test <- x_test / 255 # Make sure images have shape (28, 28, 1) x_train <- op_expand_dims(x_train, -1) x_test <- op_expand_dims(x_test, -1)   dim(x_train) ## [1] 60000    28    28     1 dim(x_test) ## [1] 10000    28    28     1 # convert class vectors to binary class matrices y_train <- to_categorical(y_train, num_classes) y_test <- to_categorical(y_test, num_classes)"},{"path":"https://keras3.posit.co/dev/articles/examples/vision/mnist_convnet.html","id":"build-the-model","dir":"Articles > Examples > Vision","previous_headings":"","what":"Build the model","title":"Simple MNIST convnet","text":"","code":"model <- keras_model_sequential(input_shape = input_shape) model |>   layer_conv_2d(filters = 32, kernel_size = c(3, 3), activation = \"relu\") |>   layer_max_pooling_2d(pool_size = c(2, 2)) |>   layer_conv_2d(filters = 64, kernel_size = c(3, 3), activation = \"relu\") |>   layer_max_pooling_2d(pool_size = c(2, 2)) |>   layer_flatten() |>   layer_dropout(rate = 0.5) |>   layer_dense(units = num_classes, activation = \"softmax\")  summary(model) ## Model: \"sequential\" ## ┏━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━┓ ## ┃ Layer (type)                    ┃ Output Shape           ┃       Param # ┃ ## ┡━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━┩ ## │ conv2d (Conv2D)                 │ (None, 26, 26, 32)     │           320 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ max_pooling2d (MaxPooling2D)    │ (None, 13, 13, 32)     │             0 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ conv2d_1 (Conv2D)               │ (None, 11, 11, 64)     │        18,496 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ max_pooling2d_1 (MaxPooling2D)  │ (None, 5, 5, 64)       │             0 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ flatten (Flatten)               │ (None, 1600)           │             0 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ dropout (Dropout)               │ (None, 1600)           │             0 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ dense (Dense)                   │ (None, 10)             │        16,010 │ ## └─────────────────────────────────┴────────────────────────┴───────────────┘ ##  Total params: 34,826 (136.04 KB) ##  Trainable params: 34,826 (136.04 KB) ##  Non-trainable params: 0 (0.00 B)"},{"path":"https://keras3.posit.co/dev/articles/examples/vision/mnist_convnet.html","id":"train-the-model","dir":"Articles > Examples > Vision","previous_headings":"","what":"Train the model","title":"Simple MNIST convnet","text":"","code":"batch_size <- 128 epochs <- 15  model |> compile(   loss = \"categorical_crossentropy\",   optimizer = \"adam\",   metrics = \"accuracy\" )  model |> fit(   x_train, y_train,   batch_size = batch_size,   epochs = epochs,   validation_split = 0.1 ) ## Epoch 1/15 ## 422/422 - 5s - 11ms/step - accuracy: 0.8896 - loss: 0.3635 - val_accuracy: 0.9783 - val_loss: 0.0791 ## Epoch 2/15 ## 422/422 - 1s - 2ms/step - accuracy: 0.9666 - loss: 0.1111 - val_accuracy: 0.9850 - val_loss: 0.0548 ## Epoch 3/15 ## 422/422 - 1s - 2ms/step - accuracy: 0.9744 - loss: 0.0824 - val_accuracy: 0.9883 - val_loss: 0.0441 ## Epoch 4/15 ## 422/422 - 1s - 2ms/step - accuracy: 0.9786 - loss: 0.0694 - val_accuracy: 0.9895 - val_loss: 0.0401 ## Epoch 5/15 ## 422/422 - 1s - 2ms/step - accuracy: 0.9804 - loss: 0.0625 - val_accuracy: 0.9903 - val_loss: 0.0353 ## Epoch 6/15 ## 422/422 - 1s - 2ms/step - accuracy: 0.9824 - loss: 0.0557 - val_accuracy: 0.9907 - val_loss: 0.0332 ## Epoch 7/15 ## 422/422 - 1s - 2ms/step - accuracy: 0.9838 - loss: 0.0500 - val_accuracy: 0.9917 - val_loss: 0.0312 ## Epoch 8/15 ## 422/422 - 1s - 2ms/step - accuracy: 0.9850 - loss: 0.0480 - val_accuracy: 0.9918 - val_loss: 0.0304 ## Epoch 9/15 ## 422/422 - 1s - 2ms/step - accuracy: 0.9860 - loss: 0.0445 - val_accuracy: 0.9918 - val_loss: 0.0303 ## Epoch 10/15 ## 422/422 - 1s - 2ms/step - accuracy: 0.9861 - loss: 0.0442 - val_accuracy: 0.9912 - val_loss: 0.0293 ## Epoch 11/15 ## 422/422 - 1s - 2ms/step - accuracy: 0.9876 - loss: 0.0390 - val_accuracy: 0.9915 - val_loss: 0.0307 ## Epoch 12/15 ## 422/422 - 1s - 2ms/step - accuracy: 0.9876 - loss: 0.0374 - val_accuracy: 0.9925 - val_loss: 0.0288 ## Epoch 13/15 ## 422/422 - 1s - 2ms/step - accuracy: 0.9887 - loss: 0.0346 - val_accuracy: 0.9920 - val_loss: 0.0291 ## Epoch 14/15 ## 422/422 - 1s - 2ms/step - accuracy: 0.9887 - loss: 0.0343 - val_accuracy: 0.9923 - val_loss: 0.0276 ## Epoch 15/15 ## 422/422 - 1s - 2ms/step - accuracy: 0.9897 - loss: 0.0323 - val_accuracy: 0.9922 - val_loss: 0.0286"},{"path":"https://keras3.posit.co/dev/articles/examples/vision/mnist_convnet.html","id":"evaluate-the-trained-model","dir":"Articles > Examples > Vision","previous_headings":"","what":"Evaluate the trained model","title":"Simple MNIST convnet","text":"","code":"score <- model |> evaluate(x_test, y_test, verbose = 0) score ## $accuracy ## [1] 0.9915 ## ## $loss ## [1] 0.02471612"},{"path":"https://keras3.posit.co/dev/articles/examples/vision/mnist_siamese_graph.html","id":"introduction","dir":"Articles > Examples > Vision","previous_headings":"","what":"Introduction","title":"Train a Siamese MLP on pairs of digits from the MNIST dataset.","text":"Siamese Networks neural networks share weights two sister networks, producing embedding vectors respective inputs. supervised similarity learning, networks trained maximize contrast (distance) embeddings inputs different classes, minimizing distance embeddings similar classes, resulting embedding spaces reflect class segmentation training inputs. implementation loosely follows Hadsell-et-al.’06 [1] (see paper mode details) euclidean distance replaced subtraction layer one fully-connect (FC) layer. [1] “Dimensionality Reduction Learning Invariant Mapping” https://yann.lecun.com/exdb/publis/pdf/hadsell-chopra-lecun-06.pdf Gets 98.11% test accuracy 20 epochs. 3 seconds per epoch AMD Ryzen 7 PRO 4750U (CPU)","code":"library(keras3) contrastive_loss <- function(y_true, y_pred) {     # Contrastive loss from Hadsell-et-al.'06     # https://yann.lecun.com/exdb/publis/pdf/hadsell-chopra-lecun-06.pdf     margin = 1     margin_square = op_square(op_maximum(margin - (y_pred), 0))     op_mean((1 - y_true) * op_square(y_pred) + (y_true) * margin_square) }"},{"path":"https://keras3.posit.co/dev/articles/examples/vision/mnist_siamese_graph.html","id":"create-pairs-of-images","dir":"Articles > Examples > Vision","previous_headings":"","what":"Create pairs of images","title":"Train a Siamese MLP on pairs of digits from the MNIST dataset.","text":"train model differentiate digits different classes. example, digit 0 needs differentiated rest digits (1 9), digit 1 - 0 2 9, . carry , select N random images class (example, digit 0) pair N random images another class B (example, digit 1). , can repeat process classes digits (digit 9). paired digit 0 digits, can repeat process remaining classes rest digits (1 9).","code":"create_pairs <- function(x, y) {     # Positive and negative pair creation.     # Alternates between positive and negative pairs.     digit_indices <- tapply(1:length(y), y, list)     y1 <- y     y2 <- sapply(y, function(a) sample(0:9,1,prob=0.1+0.8*(0:9==a)))     idx1 <- 1:nrow(x)     idx2 <- sapply(as.character(y2), function(a) sample(digit_indices[[a]],1))     is_same  <- 1*(y1==y2)     list(pair1 = x[idx1,], pair2 = x[idx2,], y = is_same) }  compute_accuracy <- function(predictions, labels) {     # Compute classification accuracy with a fixed threshold on distances.     mean(labels[predictions > 0.5]) } # the data, shuffled and split between train and test sets mnist   <- dataset_mnist() x_train <- mnist$train$x y_train <- mnist$train$y x_test  <- mnist$test$x y_test  <- mnist$test$y x_train <- array_reshape(x_train, c(nrow(x_train), 784)) x_test  <- array_reshape(x_test, c(nrow(x_test), 784)) x_train <- x_train / 255 x_test  <- x_test / 255  # create training+test positive and negative pairs tr <- create_pairs(x_train, y_train) te <- create_pairs(x_test,  y_test)  names(tr) ## [1] \"pair1\" \"pair2\" \"y\""},{"path":"https://keras3.posit.co/dev/articles/examples/vision/mnist_siamese_graph.html","id":"network-definition","dir":"Articles > Examples > Vision","previous_headings":"","what":"Network definition","title":"Train a Siamese MLP on pairs of digits from the MNIST dataset.","text":"","code":"# input layers input_dim = 784 input_1 <- layer_input(shape = c(input_dim)) input_2 <- layer_input(shape = c(input_dim))  # definition of the base network that will be shared base_network <- keras_model_sequential() %>%     layer_dense(units = 128, activation = 'relu') %>%     layer_dropout(rate = 0.1) %>%     layer_dense(units = 128, activation = 'relu') %>%     layer_dropout(rate = 0.1) %>%     layer_dense(units = 128, activation = 'relu')  # because we re-use the same instance `base_network`, the weights of # the network will be shared across the two branches branch_1 <- base_network(input_1) branch_2 <- base_network(input_2)  # merging layer out <- layer_subtract(list(branch_1, branch_2)) %>%     layer_dropout(rate = 0.1) %>%     layer_dense(units = 16, activation = 'relu') %>%     layer_dense(1, activation = \"sigmoid\")  # create and compile model model <- keras_model(list(input_1, input_2), out)"},{"path":"https://keras3.posit.co/dev/articles/examples/vision/mnist_siamese_graph.html","id":"train","dir":"Articles > Examples > Vision","previous_headings":"","what":"Train","title":"Train a Siamese MLP on pairs of digits from the MNIST dataset.","text":"","code":"model %>% compile(     optimizer = \"rmsprop\",     #loss = \"binary_crossentropy\",     loss = contrastive_loss,     metrics = metric_binary_accuracy )  history <- model %>% fit(     list(tr$pair1, tr$pair2), tr$y,     batch_size = 128,     epochs = 20,     validation_data = list(         list(te$pair1, te$pair2),         te$y     ) ) ## Epoch 1/20 ## 469/469 - 6s - 13ms/step - binary_accuracy: 0.7569 - loss: 0.1640 - val_binary_accuracy: 0.8798 - val_loss: 0.0974 ## Epoch 2/20 ## 469/469 - 1s - 2ms/step - binary_accuracy: 0.8899 - loss: 0.0866 - val_binary_accuracy: 0.9227 - val_loss: 0.0626 ## Epoch 3/20 ## 469/469 - 1s - 2ms/step - binary_accuracy: 0.9233 - loss: 0.0603 - val_binary_accuracy: 0.9398 - val_loss: 0.0497 ## Epoch 4/20 ## 469/469 - 1s - 2ms/step - binary_accuracy: 0.9423 - loss: 0.0460 - val_binary_accuracy: 0.9480 - val_loss: 0.0399 ## Epoch 5/20 ## 469/469 - 1s - 2ms/step - binary_accuracy: 0.9561 - loss: 0.0356 - val_binary_accuracy: 0.9578 - val_loss: 0.0330 ## Epoch 6/20 ## 469/469 - 1s - 2ms/step - binary_accuracy: 0.9637 - loss: 0.0295 - val_binary_accuracy: 0.9613 - val_loss: 0.0313 ## Epoch 7/20 ## 469/469 - 1s - 2ms/step - binary_accuracy: 0.9686 - loss: 0.0252 - val_binary_accuracy: 0.9645 - val_loss: 0.0275 ## Epoch 8/20 ## 469/469 - 1s - 2ms/step - binary_accuracy: 0.9730 - loss: 0.0217 - val_binary_accuracy: 0.9672 - val_loss: 0.0264 ## Epoch 9/20 ## 469/469 - 1s - 2ms/step - binary_accuracy: 0.9768 - loss: 0.0190 - val_binary_accuracy: 0.9686 - val_loss: 0.0246 ## Epoch 10/20 ## 469/469 - 1s - 2ms/step - binary_accuracy: 0.9793 - loss: 0.0169 - val_binary_accuracy: 0.9701 - val_loss: 0.0242 ## Epoch 11/20 ## 469/469 - 1s - 2ms/step - binary_accuracy: 0.9805 - loss: 0.0158 - val_binary_accuracy: 0.9719 - val_loss: 0.0233 ## Epoch 12/20 ## 469/469 - 1s - 2ms/step - binary_accuracy: 0.9822 - loss: 0.0148 - val_binary_accuracy: 0.9718 - val_loss: 0.0230 ## Epoch 13/20 ## 469/469 - 1s - 2ms/step - binary_accuracy: 0.9833 - loss: 0.0135 - val_binary_accuracy: 0.9702 - val_loss: 0.0234 ## Epoch 14/20 ## 469/469 - 1s - 2ms/step - binary_accuracy: 0.9846 - loss: 0.0129 - val_binary_accuracy: 0.9730 - val_loss: 0.0220 ## Epoch 15/20 ## 469/469 - 1s - 2ms/step - binary_accuracy: 0.9857 - loss: 0.0117 - val_binary_accuracy: 0.9759 - val_loss: 0.0193 ## Epoch 16/20 ## 469/469 - 1s - 2ms/step - binary_accuracy: 0.9868 - loss: 0.0111 - val_binary_accuracy: 0.9749 - val_loss: 0.0217 ## Epoch 17/20 ## 469/469 - 1s - 2ms/step - binary_accuracy: 0.9872 - loss: 0.0106 - val_binary_accuracy: 0.9741 - val_loss: 0.0211 ## Epoch 18/20 ## 469/469 - 1s - 2ms/step - binary_accuracy: 0.9880 - loss: 0.0099 - val_binary_accuracy: 0.9746 - val_loss: 0.0201 ## Epoch 19/20 ## 469/469 - 1s - 2ms/step - binary_accuracy: 0.9879 - loss: 0.0100 - val_binary_accuracy: 0.9756 - val_loss: 0.0207 ## Epoch 20/20 ## 469/469 - 1s - 2ms/step - binary_accuracy: 0.9888 - loss: 0.0091 - val_binary_accuracy: 0.9776 - val_loss: 0.0184 plot(history)"},{"path":"https://keras3.posit.co/dev/articles/examples/vision/mnist_siamese_graph.html","id":"evaluate","dir":"Articles > Examples > Vision","previous_headings":"","what":"Evaluate","title":"Train a Siamese MLP on pairs of digits from the MNIST dataset.","text":"","code":"# compute final accuracy on training and test sets  tr_pred <- predict(model, list(tr$pair1, tr$pair2))[,1] ## 1875/1875 - 2s - 1ms/step tr_acc  <- compute_accuracy(tr_pred, tr$y) te_pred <- predict(model, list(te$pair1, te$pair2))[,1] ## 313/313 - 1s - 2ms/step te_acc  <- compute_accuracy(te_pred, te$y)  sprintf('* Accuracy on training set: %0.2f%%', (100 * tr_acc)) ## [1] \"* Accuracy on training set: 99.67%\" sprintf('* Accuracy on test set: %0.2f%%', (100 * te_acc)) ## [1] \"* Accuracy on test set: 98.25%\""},{"path":"https://keras3.posit.co/dev/articles/examples/vision/mnist_siamese_graph.html","id":"plots","dir":"Articles > Examples > Vision","previous_headings":"","what":"Plots","title":"Train a Siamese MLP on pairs of digits from the MNIST dataset.","text":"","code":"par(mfrow=c(1,1)) vioplot::vioplot( te_pred ~ te$y ) i=3 visualizePair <- function(i) {     image(rbind(matrix( te$pair1[i,],28,28)[,28:1], matrix( te$pair2[i,],28,28)[,28:1]))     title(paste(\"true:\", te$y[i],\"|  pred:\", round(te_pred[i],5))) } par(mfrow=c(3,3)) lapply(1:9, visualizePair) ## [[1]] ## NULL ## ## [[2]] ## NULL ## ## [[3]] ## NULL ## ## [[4]] ## NULL ## ## [[5]] ## NULL ## ## [[6]] ## NULL ## ## [[7]] ## NULL ## ## [[8]] ## NULL ## ## [[9]] ## NULL"},{"path":"https://keras3.posit.co/dev/articles/examples/vision/oxford_pets_image_segmentation.html","id":"download-the-data","dir":"Articles > Examples > Vision","previous_headings":"","what":"Download the data","title":"Image segmentation with a U-Net-like architecture","text":"","code":"options(timeout = 5000) download.file(   \"https://www.robots.ox.ac.uk/~vgg/data/pets/data/images.tar.gz\",   \"datasets/images.tar.gz\" ) download.file(   \"https://www.robots.ox.ac.uk/~vgg/data/pets/data/annotations.tar.gz\",   \"datasets/annotations.tar.gz\" )  untar(\"datasets/images.tar.gz\", exdir = \"datasets\") untar(\"datasets/annotations.tar.gz\", exdir = \"datasets\")"},{"path":"https://keras3.posit.co/dev/articles/examples/vision/oxford_pets_image_segmentation.html","id":"prepare-paths-of-input-images-and-target-segmentation-masks","dir":"Articles > Examples > Vision","previous_headings":"","what":"Prepare paths of input images and target segmentation masks","title":"Image segmentation with a U-Net-like architecture","text":"","code":"library(keras3) input_dir <- \"datasets/images/\" target_dir <- \"datasets/annotations/trimaps/\" img_size <- c(160, 160) num_classes <- 3 batch_size <- 32  input_img_paths <- fs::dir_ls(input_dir, glob = \"*.jpg\") |> sort() target_img_paths <- fs::dir_ls(target_dir, glob = \"*.png\") |> sort()  cat(\"Number of samples:\", length(input_img_paths), \"\\n\") ## Number of samples: 7390 for (i in 1:10) {   cat(input_img_paths[i], \"|\", target_img_paths[i], \"\\n\") } ## datasets/images/Abyssinian_1.jpg | datasets/annotations/trimaps/Abyssinian_1.png ## datasets/images/Abyssinian_10.jpg | datasets/annotations/trimaps/Abyssinian_10.png ## datasets/images/Abyssinian_100.jpg | datasets/annotations/trimaps/Abyssinian_100.png ## datasets/images/Abyssinian_101.jpg | datasets/annotations/trimaps/Abyssinian_101.png ## datasets/images/Abyssinian_102.jpg | datasets/annotations/trimaps/Abyssinian_102.png ## datasets/images/Abyssinian_103.jpg | datasets/annotations/trimaps/Abyssinian_103.png ## datasets/images/Abyssinian_104.jpg | datasets/annotations/trimaps/Abyssinian_104.png ## datasets/images/Abyssinian_105.jpg | datasets/annotations/trimaps/Abyssinian_105.png ## datasets/images/Abyssinian_106.jpg | datasets/annotations/trimaps/Abyssinian_106.png ## datasets/images/Abyssinian_107.jpg | datasets/annotations/trimaps/Abyssinian_107.png"},{"path":"https://keras3.posit.co/dev/articles/examples/vision/oxford_pets_image_segmentation.html","id":"what-does-one-input-image-and-corresponding-segmentation-mask-look-like","dir":"Articles > Examples > Vision","previous_headings":"","what":"What does one input image and corresponding segmentation mask look like?","title":"Image segmentation with a U-Net-like architecture","text":"","code":"# Display input image #10 input_img_paths[10] |>   jpeg::readJPEG() |>   as.raster() |>   plot() target_img_paths[10] |>   png::readPNG() |>   magrittr::multiply_by(255)|>   as.raster(max = 3) |>   plot()"},{"path":"https://keras3.posit.co/dev/articles/examples/vision/oxford_pets_image_segmentation.html","id":"prepare-dataset-to-load-vectorize-batches-of-data","dir":"Articles > Examples > Vision","previous_headings":"","what":"Prepare dataset to load & vectorize batches of data","title":"Image segmentation with a U-Net-like architecture","text":"","code":"library(tensorflow, exclude = c(\"shape\", \"set_random_seed\")) library(tfdatasets, exclude = \"shape\")   # Returns a tf_dataset get_dataset <- function(batch_size, img_size, input_img_paths, target_img_paths,                         max_dataset_len = NULL) {    img_size <- as.integer(img_size)    load_img_masks <- function(input_img_path, target_img_path) {     input_img <- input_img_path |>       tf$io$read_file() |>       tf$io$decode_jpeg(channels = 3) |>       tf$image$resize(img_size) |>       tf$image$convert_image_dtype(\"float32\")      target_img <- target_img_path |>       tf$io$read_file() |>       tf$io$decode_png(channels = 1) |>       tf$image$resize(img_size, method = \"nearest\") |>       tf$image$convert_image_dtype(\"uint8\")      # Ground truth labels are 1, 2, 3. Subtract one to make them 0, 1, 2:     target_img <- target_img - 1L      list(input_img, target_img)   }    if (!is.null(max_dataset_len)) {     input_img_paths <- input_img_paths[1:max_dataset_len]     target_img_paths <- target_img_paths[1:max_dataset_len]   }    list(input_img_paths, target_img_paths) |>     tensor_slices_dataset() |>     dataset_map(load_img_masks, num_parallel_calls = tf$data$AUTOTUNE)|>     dataset_batch(batch_size) }"},{"path":"https://keras3.posit.co/dev/articles/examples/vision/oxford_pets_image_segmentation.html","id":"prepare-u-net-xception-style-model","dir":"Articles > Examples > Vision","previous_headings":"","what":"Prepare U-Net Xception-style model","title":"Image segmentation with a U-Net-like architecture","text":"","code":"get_model <- function(img_size, num_classes) {    inputs <- keras_input(shape = c(img_size, 3))    ### [First half of the network: downsampling inputs] ###    # Entry block   x <- inputs |>     layer_conv_2d(filters = 32, kernel_size = 3, strides = 2, padding = \"same\") |>     layer_batch_normalization() |>     layer_activation(\"relu\")    previous_block_activation <- x  # Set aside residual    for (filters in c(64, 128, 256)) {     x <- x |>       layer_activation(\"relu\") |>       layer_separable_conv_2d(filters = filters, kernel_size = 3, padding = \"same\") |>       layer_batch_normalization() |>        layer_activation(\"relu\") |>       layer_separable_conv_2d(filters = filters, kernel_size = 3, padding = \"same\") |>       layer_batch_normalization() |>        layer_max_pooling_2d(pool_size = 3, strides = 2, padding = \"same\")      residual <- previous_block_activation |>       layer_conv_2d(filters = filters, kernel_size = 1, strides = 2, padding = \"same\")      x <- layer_add(x, residual)  # Add back residual     previous_block_activation <- x  # Set aside next residual   }    ### [Second half of the network: upsampling inputs] ###    for (filters in c(256, 128, 64, 32)) {     x <- x |>       layer_activation(\"relu\") |>       layer_conv_2d_transpose(filters = filters, kernel_size = 3, padding = \"same\") |>       layer_batch_normalization() |>        layer_activation(\"relu\") |>       layer_conv_2d_transpose(filters = filters, kernel_size = 3, padding = \"same\") |>       layer_batch_normalization() |>        layer_upsampling_2d(size = 2)      # Project residual     residual <- previous_block_activation |>       layer_upsampling_2d(size = 2) |>       layer_conv_2d(filters = filters, kernel_size = 1, padding = \"same\")      x <- layer_add(x, residual)     # Add back residual     previous_block_activation <- x  # Set aside next residual   }    # Add a per-pixel classification layer   outputs <- x |>     layer_conv_2d(num_classes, 3, activation = \"softmax\", padding = \"same\")    # Define the model   keras_model(inputs, outputs) }  # Build model model <- get_model(img_size, num_classes) summary(model) ## Model: \"functional\" ## ┏━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━┳━━━━━━━━━━━━━━━━┳━━━━━━━┓ ## ┃ Layer (type)      ┃ Output Shape    ┃   Param # ┃ Connected to   ┃ Trai… ┃ ## ┡━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━╇━━━━━━━━━━━━━━━━╇━━━━━━━┩ ## │ input_layer       │ (None, 160,     │         0 │ -              │   -   │ ## │ (InputLayer)      │ 160, 3)         │           │                │       │ ## ├───────────────────┼─────────────────┼───────────┼────────────────┼───────┤ ## │ conv2d (Conv2D)   │ (None, 80, 80,  │       896 │ input_layer[0… │   Y   │ ## │                   │ 32)             │           │                │       │ ## ├───────────────────┼─────────────────┼───────────┼────────────────┼───────┤ ## │ batch_normalizat… │ (None, 80, 80,  │       128 │ conv2d[0][0]   │   Y   │ ## │ (BatchNormalizat… │ 32)             │           │                │       │ ## ├───────────────────┼─────────────────┼───────────┼────────────────┼───────┤ ## │ activation        │ (None, 80, 80,  │         0 │ batch_normali… │   -   │ ## │ (Activation)      │ 32)             │           │                │       │ ## ├───────────────────┼─────────────────┼───────────┼────────────────┼───────┤ ## │ activation_1      │ (None, 80, 80,  │         0 │ activation[0]… │   -   │ ## │ (Activation)      │ 32)             │           │                │       │ ## ├───────────────────┼─────────────────┼───────────┼────────────────┼───────┤ ## │ separable_conv2d  │ (None, 80, 80,  │     2,400 │ activation_1[… │   Y   │ ## │ (SeparableConv2D) │ 64)             │           │                │       │ ## ├───────────────────┼─────────────────┼───────────┼────────────────┼───────┤ ## │ batch_normalizat… │ (None, 80, 80,  │       256 │ separable_con… │   Y   │ ## │ (BatchNormalizat… │ 64)             │           │                │       │ ## ├───────────────────┼─────────────────┼───────────┼────────────────┼───────┤ ## │ activation_2      │ (None, 80, 80,  │         0 │ batch_normali… │   -   │ ## │ (Activation)      │ 64)             │           │                │       │ ## ├───────────────────┼─────────────────┼───────────┼────────────────┼───────┤ ## │ separable_conv2d… │ (None, 80, 80,  │     4,736 │ activation_2[… │   Y   │ ## │ (SeparableConv2D) │ 64)             │           │                │       │ ## ├───────────────────┼─────────────────┼───────────┼────────────────┼───────┤ ## │ batch_normalizat… │ (None, 80, 80,  │       256 │ separable_con… │   Y   │ ## │ (BatchNormalizat… │ 64)             │           │                │       │ ## ├───────────────────┼─────────────────┼───────────┼────────────────┼───────┤ ## │ max_pooling2d     │ (None, 40, 40,  │         0 │ batch_normali… │   -   │ ## │ (MaxPooling2D)    │ 64)             │           │                │       │ ## ├───────────────────┼─────────────────┼───────────┼────────────────┼───────┤ ## │ conv2d_1 (Conv2D) │ (None, 40, 40,  │     2,112 │ activation[0]… │   Y   │ ## │                   │ 64)             │           │                │       │ ## ├───────────────────┼─────────────────┼───────────┼────────────────┼───────┤ ## │ add (Add)         │ (None, 40, 40,  │         0 │ max_pooling2d… │   -   │ ## │                   │ 64)             │           │ conv2d_1[0][0] │       │ ## ├───────────────────┼─────────────────┼───────────┼────────────────┼───────┤ ## │ activation_3      │ (None, 40, 40,  │         0 │ add[0][0]      │   -   │ ## │ (Activation)      │ 64)             │           │                │       │ ## ├───────────────────┼─────────────────┼───────────┼────────────────┼───────┤ ## │ separable_conv2d… │ (None, 40, 40,  │     8,896 │ activation_3[… │   Y   │ ## │ (SeparableConv2D) │ 128)            │           │                │       │ ## ├───────────────────┼─────────────────┼───────────┼────────────────┼───────┤ ## │ batch_normalizat… │ (None, 40, 40,  │       512 │ separable_con… │   Y   │ ## │ (BatchNormalizat… │ 128)            │           │                │       │ ## ├───────────────────┼─────────────────┼───────────┼────────────────┼───────┤ ## │ activation_4      │ (None, 40, 40,  │         0 │ batch_normali… │   -   │ ## │ (Activation)      │ 128)            │           │                │       │ ## ├───────────────────┼─────────────────┼───────────┼────────────────┼───────┤ ## │ separable_conv2d… │ (None, 40, 40,  │    17,664 │ activation_4[… │   Y   │ ## │ (SeparableConv2D) │ 128)            │           │                │       │ ## ├───────────────────┼─────────────────┼───────────┼────────────────┼───────┤ ## │ batch_normalizat… │ (None, 40, 40,  │       512 │ separable_con… │   Y   │ ## │ (BatchNormalizat… │ 128)            │           │                │       │ ## ├───────────────────┼─────────────────┼───────────┼────────────────┼───────┤ ## │ max_pooling2d_1   │ (None, 20, 20,  │         0 │ batch_normali… │   -   │ ## │ (MaxPooling2D)    │ 128)            │           │                │       │ ## ├───────────────────┼─────────────────┼───────────┼────────────────┼───────┤ ## │ conv2d_2 (Conv2D) │ (None, 20, 20,  │     8,320 │ add[0][0]      │   Y   │ ## │                   │ 128)            │           │                │       │ ## ├───────────────────┼─────────────────┼───────────┼────────────────┼───────┤ ## │ add_1 (Add)       │ (None, 20, 20,  │         0 │ max_pooling2d… │   -   │ ## │                   │ 128)            │           │ conv2d_2[0][0] │       │ ## ├───────────────────┼─────────────────┼───────────┼────────────────┼───────┤ ## │ activation_5      │ (None, 20, 20,  │         0 │ add_1[0][0]    │   -   │ ## │ (Activation)      │ 128)            │           │                │       │ ## ├───────────────────┼─────────────────┼───────────┼────────────────┼───────┤ ## │ separable_conv2d… │ (None, 20, 20,  │    34,176 │ activation_5[… │   Y   │ ## │ (SeparableConv2D) │ 256)            │           │                │       │ ## ├───────────────────┼─────────────────┼───────────┼────────────────┼───────┤ ## │ batch_normalizat… │ (None, 20, 20,  │     1,024 │ separable_con… │   Y   │ ## │ (BatchNormalizat… │ 256)            │           │                │       │ ## ├───────────────────┼─────────────────┼───────────┼────────────────┼───────┤ ## │ activation_6      │ (None, 20, 20,  │         0 │ batch_normali… │   -   │ ## │ (Activation)      │ 256)            │           │                │       │ ## ├───────────────────┼─────────────────┼───────────┼────────────────┼───────┤ ## │ separable_conv2d… │ (None, 20, 20,  │    68,096 │ activation_6[… │   Y   │ ## │ (SeparableConv2D) │ 256)            │           │                │       │ ## ├───────────────────┼─────────────────┼───────────┼────────────────┼───────┤ ## │ batch_normalizat… │ (None, 20, 20,  │     1,024 │ separable_con… │   Y   │ ## │ (BatchNormalizat… │ 256)            │           │                │       │ ## ├───────────────────┼─────────────────┼───────────┼────────────────┼───────┤ ## │ max_pooling2d_2   │ (None, 10, 10,  │         0 │ batch_normali… │   -   │ ## │ (MaxPooling2D)    │ 256)            │           │                │       │ ## ├───────────────────┼─────────────────┼───────────┼────────────────┼───────┤ ## │ conv2d_3 (Conv2D) │ (None, 10, 10,  │    33,024 │ add_1[0][0]    │   Y   │ ## │                   │ 256)            │           │                │       │ ## ├───────────────────┼─────────────────┼───────────┼────────────────┼───────┤ ## │ add_2 (Add)       │ (None, 10, 10,  │         0 │ max_pooling2d… │   -   │ ## │                   │ 256)            │           │ conv2d_3[0][0] │       │ ## ├───────────────────┼─────────────────┼───────────┼────────────────┼───────┤ ## │ activation_7      │ (None, 10, 10,  │         0 │ add_2[0][0]    │   -   │ ## │ (Activation)      │ 256)            │           │                │       │ ## ├───────────────────┼─────────────────┼───────────┼────────────────┼───────┤ ## │ conv2d_transpose  │ (None, 10, 10,  │   590,080 │ activation_7[… │   Y   │ ## │ (Conv2DTranspose) │ 256)            │           │                │       │ ## ├───────────────────┼─────────────────┼───────────┼────────────────┼───────┤ ## │ batch_normalizat… │ (None, 10, 10,  │     1,024 │ conv2d_transp… │   Y   │ ## │ (BatchNormalizat… │ 256)            │           │                │       │ ## ├───────────────────┼─────────────────┼───────────┼────────────────┼───────┤ ## │ activation_8      │ (None, 10, 10,  │         0 │ batch_normali… │   -   │ ## │ (Activation)      │ 256)            │           │                │       │ ## ├───────────────────┼─────────────────┼───────────┼────────────────┼───────┤ ## │ conv2d_transpose… │ (None, 10, 10,  │   590,080 │ activation_8[… │   Y   │ ## │ (Conv2DTranspose) │ 256)            │           │                │       │ ## ├───────────────────┼─────────────────┼───────────┼────────────────┼───────┤ ## │ batch_normalizat… │ (None, 10, 10,  │     1,024 │ conv2d_transp… │   Y   │ ## │ (BatchNormalizat… │ 256)            │           │                │       │ ## ├───────────────────┼─────────────────┼───────────┼────────────────┼───────┤ ## │ up_sampling2d_1   │ (None, 20, 20,  │         0 │ add_2[0][0]    │   -   │ ## │ (UpSampling2D)    │ 256)            │           │                │       │ ## ├───────────────────┼─────────────────┼───────────┼────────────────┼───────┤ ## │ up_sampling2d     │ (None, 20, 20,  │         0 │ batch_normali… │   -   │ ## │ (UpSampling2D)    │ 256)            │           │                │       │ ## ├───────────────────┼─────────────────┼───────────┼────────────────┼───────┤ ## │ conv2d_4 (Conv2D) │ (None, 20, 20,  │    65,792 │ up_sampling2d… │   Y   │ ## │                   │ 256)            │           │                │       │ ## ├───────────────────┼─────────────────┼───────────┼────────────────┼───────┤ ## │ add_3 (Add)       │ (None, 20, 20,  │         0 │ up_sampling2d… │   -   │ ## │                   │ 256)            │           │ conv2d_4[0][0] │       │ ## ├───────────────────┼─────────────────┼───────────┼────────────────┼───────┤ ## │ activation_9      │ (None, 20, 20,  │         0 │ add_3[0][0]    │   -   │ ## │ (Activation)      │ 256)            │           │                │       │ ## ├───────────────────┼─────────────────┼───────────┼────────────────┼───────┤ ## │ conv2d_transpose… │ (None, 20, 20,  │   295,040 │ activation_9[… │   Y   │ ## │ (Conv2DTranspose) │ 128)            │           │                │       │ ## ├───────────────────┼─────────────────┼───────────┼────────────────┼───────┤ ## │ batch_normalizat… │ (None, 20, 20,  │       512 │ conv2d_transp… │   Y   │ ## │ (BatchNormalizat… │ 128)            │           │                │       │ ## ├───────────────────┼─────────────────┼───────────┼────────────────┼───────┤ ## │ activation_10     │ (None, 20, 20,  │         0 │ batch_normali… │   -   │ ## │ (Activation)      │ 128)            │           │                │       │ ## ├───────────────────┼─────────────────┼───────────┼────────────────┼───────┤ ## │ conv2d_transpose… │ (None, 20, 20,  │   147,584 │ activation_10… │   Y   │ ## │ (Conv2DTranspose) │ 128)            │           │                │       │ ## ├───────────────────┼─────────────────┼───────────┼────────────────┼───────┤ ## │ batch_normalizat… │ (None, 20, 20,  │       512 │ conv2d_transp… │   Y   │ ## │ (BatchNormalizat… │ 128)            │           │                │       │ ## ├───────────────────┼─────────────────┼───────────┼────────────────┼───────┤ ## │ up_sampling2d_3   │ (None, 40, 40,  │         0 │ add_3[0][0]    │   -   │ ## │ (UpSampling2D)    │ 256)            │           │                │       │ ## ├───────────────────┼─────────────────┼───────────┼────────────────┼───────┤ ## │ up_sampling2d_2   │ (None, 40, 40,  │         0 │ batch_normali… │   -   │ ## │ (UpSampling2D)    │ 128)            │           │                │       │ ## ├───────────────────┼─────────────────┼───────────┼────────────────┼───────┤ ## │ conv2d_5 (Conv2D) │ (None, 40, 40,  │    32,896 │ up_sampling2d… │   Y   │ ## │                   │ 128)            │           │                │       │ ## ├───────────────────┼─────────────────┼───────────┼────────────────┼───────┤ ## │ add_4 (Add)       │ (None, 40, 40,  │         0 │ up_sampling2d… │   -   │ ## │                   │ 128)            │           │ conv2d_5[0][0] │       │ ## ├───────────────────┼─────────────────┼───────────┼────────────────┼───────┤ ## │ activation_11     │ (None, 40, 40,  │         0 │ add_4[0][0]    │   -   │ ## │ (Activation)      │ 128)            │           │                │       │ ## ├───────────────────┼─────────────────┼───────────┼────────────────┼───────┤ ## │ conv2d_transpose… │ (None, 40, 40,  │    73,792 │ activation_11… │   Y   │ ## │ (Conv2DTranspose) │ 64)             │           │                │       │ ## ├───────────────────┼─────────────────┼───────────┼────────────────┼───────┤ ## │ batch_normalizat… │ (None, 40, 40,  │       256 │ conv2d_transp… │   Y   │ ## │ (BatchNormalizat… │ 64)             │           │                │       │ ## ├───────────────────┼─────────────────┼───────────┼────────────────┼───────┤ ## │ activation_12     │ (None, 40, 40,  │         0 │ batch_normali… │   -   │ ## │ (Activation)      │ 64)             │           │                │       │ ## ├───────────────────┼─────────────────┼───────────┼────────────────┼───────┤ ## │ conv2d_transpose… │ (None, 40, 40,  │    36,928 │ activation_12… │   Y   │ ## │ (Conv2DTranspose) │ 64)             │           │                │       │ ## ├───────────────────┼─────────────────┼───────────┼────────────────┼───────┤ ## │ batch_normalizat… │ (None, 40, 40,  │       256 │ conv2d_transp… │   Y   │ ## │ (BatchNormalizat… │ 64)             │           │                │       │ ## ├───────────────────┼─────────────────┼───────────┼────────────────┼───────┤ ## │ up_sampling2d_5   │ (None, 80, 80,  │         0 │ add_4[0][0]    │   -   │ ## │ (UpSampling2D)    │ 128)            │           │                │       │ ## ├───────────────────┼─────────────────┼───────────┼────────────────┼───────┤ ## │ up_sampling2d_4   │ (None, 80, 80,  │         0 │ batch_normali… │   -   │ ## │ (UpSampling2D)    │ 64)             │           │                │       │ ## ├───────────────────┼─────────────────┼───────────┼────────────────┼───────┤ ## │ conv2d_6 (Conv2D) │ (None, 80, 80,  │     8,256 │ up_sampling2d… │   Y   │ ## │                   │ 64)             │           │                │       │ ## ├───────────────────┼─────────────────┼───────────┼────────────────┼───────┤ ## │ add_5 (Add)       │ (None, 80, 80,  │         0 │ up_sampling2d… │   -   │ ## │                   │ 64)             │           │ conv2d_6[0][0] │       │ ## ├───────────────────┼─────────────────┼───────────┼────────────────┼───────┤ ## │ activation_13     │ (None, 80, 80,  │         0 │ add_5[0][0]    │   -   │ ## │ (Activation)      │ 64)             │           │                │       │ ## ├───────────────────┼─────────────────┼───────────┼────────────────┼───────┤ ## │ conv2d_transpose… │ (None, 80, 80,  │    18,464 │ activation_13… │   Y   │ ## │ (Conv2DTranspose) │ 32)             │           │                │       │ ## ├───────────────────┼─────────────────┼───────────┼────────────────┼───────┤ ## │ batch_normalizat… │ (None, 80, 80,  │       128 │ conv2d_transp… │   Y   │ ## │ (BatchNormalizat… │ 32)             │           │                │       │ ## ├───────────────────┼─────────────────┼───────────┼────────────────┼───────┤ ## │ activation_14     │ (None, 80, 80,  │         0 │ batch_normali… │   -   │ ## │ (Activation)      │ 32)             │           │                │       │ ## ├───────────────────┼─────────────────┼───────────┼────────────────┼───────┤ ## │ conv2d_transpose… │ (None, 80, 80,  │     9,248 │ activation_14… │   Y   │ ## │ (Conv2DTranspose) │ 32)             │           │                │       │ ## ├───────────────────┼─────────────────┼───────────┼────────────────┼───────┤ ## │ batch_normalizat… │ (None, 80, 80,  │       128 │ conv2d_transp… │   Y   │ ## │ (BatchNormalizat… │ 32)             │           │                │       │ ## ├───────────────────┼─────────────────┼───────────┼────────────────┼───────┤ ## │ up_sampling2d_7   │ (None, 160,     │         0 │ add_5[0][0]    │   -   │ ## │ (UpSampling2D)    │ 160, 64)        │           │                │       │ ## ├───────────────────┼─────────────────┼───────────┼────────────────┼───────┤ ## │ up_sampling2d_6   │ (None, 160,     │         0 │ batch_normali… │   -   │ ## │ (UpSampling2D)    │ 160, 32)        │           │                │       │ ## ├───────────────────┼─────────────────┼───────────┼────────────────┼───────┤ ## │ conv2d_7 (Conv2D) │ (None, 160,     │     2,080 │ up_sampling2d… │   Y   │ ## │                   │ 160, 32)        │           │                │       │ ## ├───────────────────┼─────────────────┼───────────┼────────────────┼───────┤ ## │ add_6 (Add)       │ (None, 160,     │         0 │ up_sampling2d… │   -   │ ## │                   │ 160, 32)        │           │ conv2d_7[0][0] │       │ ## ├───────────────────┼─────────────────┼───────────┼────────────────┼───────┤ ## │ conv2d_8 (Conv2D) │ (None, 160,     │       867 │ add_6[0][0]    │   Y   │ ## │                   │ 160, 3)         │           │                │       │ ## └───────────────────┴─────────────────┴───────────┴────────────────┴───────┘ ##  Total params: 2,058,979 (7.85 MB) ##  Trainable params: 2,055,203 (7.84 MB) ##  Non-trainable params: 3,776 (14.75 KB)"},{"path":"https://keras3.posit.co/dev/articles/examples/vision/oxford_pets_image_segmentation.html","id":"set-aside-a-validation-split","dir":"Articles > Examples > Vision","previous_headings":"","what":"Set aside a validation split","title":"Image segmentation with a U-Net-like architecture","text":"","code":"# Split our img paths into a training and a validation set val_samples <- 1000 val_samples <- sample.int(length(input_img_paths), val_samples)  train_input_img_paths <- input_img_paths[-val_samples] train_target_img_paths <- target_img_paths[-val_samples]  val_input_img_paths <- input_img_paths[val_samples] val_target_img_paths <- target_img_paths[val_samples]  # Instantiate dataset for each split # Limit input files in `max_dataset_len` for faster epoch training time. # Remove the `max_dataset_len` arg when running with full dataset. train_dataset <- get_dataset(   batch_size,   img_size,   train_input_img_paths,   train_target_img_paths,   max_dataset_len = 1000 ) valid_dataset <- get_dataset(   batch_size, img_size, val_input_img_paths, val_target_img_paths )"},{"path":"https://keras3.posit.co/dev/articles/examples/vision/oxford_pets_image_segmentation.html","id":"train-the-model","dir":"Articles > Examples > Vision","previous_headings":"","what":"Train the model","title":"Image segmentation with a U-Net-like architecture","text":"","code":"# Configure the model for training. # We use the \"sparse\" version of categorical_crossentropy # because our target data is integers. model |> compile(   optimizer = optimizer_adam(1e-4),   loss = \"sparse_categorical_crossentropy\" )  callbacks <- list(   callback_model_checkpoint(     \"models/oxford_segmentation.keras\", save_best_only = TRUE   ) )  # Train the model, doing validation at the end of each epoch. epochs <- 50 model |> fit(     train_dataset,     epochs=epochs,     validation_data=valid_dataset,     callbacks=callbacks,     verbose=2 ) ## Epoch 1/50 ## 32/32 - 49s - 2s/step - loss: 1.4283 - val_loss: 1.5505 ## Epoch 2/50 ## 32/32 - 2s - 73ms/step - loss: 0.9223 - val_loss: 1.9896 ## Epoch 3/50 ## 32/32 - 2s - 72ms/step - loss: 0.7764 - val_loss: 2.5134 ## Epoch 4/50 ## 32/32 - 2s - 73ms/step - loss: 0.7200 - val_loss: 3.0166 ## Epoch 5/50 ## 32/32 - 2s - 74ms/step - loss: 0.6848 - val_loss: 3.2910 ## Epoch 6/50 ## 32/32 - 2s - 71ms/step - loss: 0.6557 - val_loss: 3.4529 ## Epoch 7/50 ## 32/32 - 2s - 74ms/step - loss: 0.6303 - val_loss: 3.5613 ## Epoch 8/50 ## 32/32 - 2s - 71ms/step - loss: 0.6083 - val_loss: 3.6520 ## Epoch 9/50 ## 32/32 - 2s - 68ms/step - loss: 0.5894 - val_loss: 3.7289 ## Epoch 10/50 ## 32/32 - 2s - 76ms/step - loss: 0.5726 - val_loss: 3.7945 ## Epoch 11/50 ## 32/32 - 2s - 75ms/step - loss: 0.5566 - val_loss: 3.8264 ## Epoch 12/50 ## 32/32 - 2s - 73ms/step - loss: 0.5406 - val_loss: 3.8009 ## Epoch 13/50 ## 32/32 - 2s - 73ms/step - loss: 0.5240 - val_loss: 3.7206 ## Epoch 14/50 ## 32/32 - 2s - 76ms/step - loss: 0.5060 - val_loss: 3.5849 ## Epoch 15/50 ## 32/32 - 2s - 76ms/step - loss: 0.4858 - val_loss: 3.4372 ## Epoch 16/50 ## 32/32 - 2s - 73ms/step - loss: 0.4636 - val_loss: 3.2371 ## Epoch 17/50 ## 32/32 - 2s - 78ms/step - loss: 0.4395 - val_loss: 3.0008 ## Epoch 18/50 ## 32/32 - 2s - 75ms/step - loss: 0.4140 - val_loss: 2.6982 ## Epoch 19/50 ## 32/32 - 2s - 72ms/step - loss: 0.3878 - val_loss: 2.3533 ## Epoch 20/50 ## 32/32 - 2s - 74ms/step - loss: 0.3623 - val_loss: 1.9573 ## Epoch 21/50 ## 32/32 - 2s - 73ms/step - loss: 0.3391 - val_loss: 1.6047 ## Epoch 22/50 ## 32/32 - 3s - 83ms/step - loss: 0.3202 - val_loss: 1.3140 ## Epoch 23/50 ## 32/32 - 2s - 77ms/step - loss: 0.3083 - val_loss: 1.0783 ## Epoch 24/50 ## 32/32 - 3s - 80ms/step - loss: 0.3097 - val_loss: 1.0180 ## Epoch 25/50 ## 32/32 - 2s - 77ms/step - loss: 0.3446 - val_loss: 0.9350 ## Epoch 26/50 ## 32/32 - 2s - 70ms/step - loss: 0.3603 - val_loss: 0.9866 ## Epoch 27/50 ## 32/32 - 3s - 82ms/step - loss: 0.3268 - val_loss: 0.8704 ## Epoch 28/50 ## 32/32 - 2s - 72ms/step - loss: 0.2880 - val_loss: 1.0167 ## Epoch 29/50 ## 32/32 - 2s - 75ms/step - loss: 0.2735 - val_loss: 1.1785 ## Epoch 30/50 ## 32/32 - 2s - 75ms/step - loss: 0.2686 - val_loss: 1.1817 ## Epoch 31/50 ## 32/32 - 2s - 74ms/step - loss: 0.2718 - val_loss: 1.1267 ## Epoch 32/50 ## 32/32 - 2s - 72ms/step - loss: 0.2867 - val_loss: 1.1271 ## Epoch 33/50 ## 32/32 - 2s - 75ms/step - loss: 0.3123 - val_loss: 1.2541 ## Epoch 34/50 ## 32/32 - 2s - 75ms/step - loss: 0.2997 - val_loss: 1.2340 ## Epoch 35/50 ## 32/32 - 2s - 74ms/step - loss: 0.2851 - val_loss: 1.0755 ## Epoch 36/50 ## 32/32 - 2s - 73ms/step - loss: 0.2781 - val_loss: 1.0022 ## Epoch 37/50 ## 32/32 - 2s - 75ms/step - loss: 0.2755 - val_loss: 1.5445 ## Epoch 38/50 ## 32/32 - 2s - 76ms/step - loss: 0.2629 - val_loss: 1.1026 ## Epoch 39/50 ## 32/32 - 2s - 75ms/step - loss: 0.2428 - val_loss: 1.2205 ## Epoch 40/50 ## 32/32 - 2s - 74ms/step - loss: 0.2409 - val_loss: 1.2005 ## Epoch 41/50 ## 32/32 - 2s - 74ms/step - loss: 0.2453 - val_loss: 1.2381 ## Epoch 42/50 ## 32/32 - 2s - 73ms/step - loss: 0.2363 - val_loss: 1.5562 ## Epoch 43/50 ## 32/32 - 2s - 75ms/step - loss: 0.2337 - val_loss: 1.3185 ## Epoch 44/50 ## 32/32 - 2s - 77ms/step - loss: 0.2222 - val_loss: 1.1867 ## Epoch 45/50 ## 32/32 - 2s - 77ms/step - loss: 0.2151 - val_loss: 1.0701 ## Epoch 46/50 ## 32/32 - 2s - 75ms/step - loss: 0.2099 - val_loss: 1.1109 ## Epoch 47/50 ## 32/32 - 2s - 74ms/step - loss: 0.2062 - val_loss: 1.0900 ## Epoch 48/50 ## 32/32 - 2s - 74ms/step - loss: 0.2038 - val_loss: 1.1194 ## Epoch 49/50 ## 32/32 - 2s - 71ms/step - loss: 0.1964 - val_loss: 1.0576 ## Epoch 50/50 ## 32/32 - 2s - 73ms/step - loss: 0.1871 - val_loss: 1.0851"},{"path":"https://keras3.posit.co/dev/articles/examples/vision/oxford_pets_image_segmentation.html","id":"visualize-predictions","dir":"Articles > Examples > Vision","previous_headings":"","what":"Visualize predictions","title":"Image segmentation with a U-Net-like architecture","text":"","code":"model <- load_model(\"models/oxford_segmentation.keras\") # Generate predictions for all images in the validation set val_dataset <- get_dataset(   batch_size, img_size, val_input_img_paths, val_target_img_paths ) val_preds <- predict(model, val_dataset) ## 32/32 - 4s - 117ms/step display_mask <- function(i) {   # Quick utility to display a model's prediction.   mask <- val_preds[i,,,] %>%     apply(c(1,2), which.max) %>%     array_reshape(dim = c(img_size, 1))   mask <- abind::abind(mask, mask, mask, along = 3)   plot(as.raster(mask, max = 3)) }  # Display results for validation image #10 i <- 10  par(mfrow = c(1, 3)) # Display input image input_img_paths[i] |>   jpeg::readJPEG() |>   as.raster() |>   plot()  # Display ground-truth target mask target_img_paths[i] |>   png::readPNG() |>   magrittr::multiply_by(255)|>   as.raster(max = 3) |>   plot()  # Display mask predicted by our model display_mask(i)  # Note that the model only sees inputs at 150x150."},{"path":"https://keras3.posit.co/dev/articles/functional_api.html","id":"setup","dir":"Articles","previous_headings":"","what":"Setup","title":"The Functional API","text":"","code":"library(keras3)"},{"path":"https://keras3.posit.co/dev/articles/functional_api.html","id":"introduction","dir":"Articles","previous_headings":"","what":"Introduction","title":"The Functional API","text":"Keras functional API way create models flexible sequential API. functional API can handle models non-linear topology, shared layers, even multiple inputs outputs. main idea deep learning model usually directed acyclic graph (DAG) layers. functional API way build graphs layers. Consider following model: basic graph three layers. build model using functional API, start creating input node: shape data set 784-dimensional vector. batch size always omitted since shape sample specified. , example, image input shape (32, 32, 3), use: inputs returned contains information shape dtype input data feed model. ’s shape: ’s dtype: create new node graph layers calling layer inputs object: “layer call” action like drawing arrow “inputs” layer created. ’re “passing” inputs dense layer, get x output. Let’s add layers graph layers: point, can create Model specifying inputs outputs graph layers: Let’s check model summary looks like: can also plot model graph:  , optionally, display input output shapes layer plotted graph:  figure code almost identical. code version, connection arrows replaced call operation. “graph layers” intuitive mental image deep learning model, functional API way create models closely mirrors .","code":"(input: 784-dimensional vectors)        ↧ [Dense (64 units, relu activation)]        ↧ [Dense (64 units, relu activation)]        ↧ [Dense (10 units, softmax activation)]        ↧ (output: logits of a probability distribution over 10 classes) inputs <- keras_input(shape = c(784)) # Just for demonstration purposes. img_inputs <- keras_input(shape = c(32, 32, 3)) shape(inputs) ## shape(NA, 784) inputs$dtype ## [1] \"float32\" dense <- layer_dense(units = 64, activation=\"relu\") x <- dense(inputs) outputs <- x |>   layer_dense(units = 64, activation = \"relu\") |>   layer_dense(units = 10) model <- keras_model(inputs = inputs, outputs = outputs, name = \"mnist_model\") summary(model) ## Model: \"mnist_model\" ## ┏━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━┓ ## ┃ Layer (type)                    ┃ Output Shape           ┃       Param # ┃ ## ┡━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━┩ ## │ input_layer (InputLayer)        │ (None, 784)            │             0 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ dense (Dense)                   │ (None, 64)             │        50,240 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ dense_1 (Dense)                 │ (None, 64)             │         4,160 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ dense_2 (Dense)                 │ (None, 10)             │           650 │ ## └─────────────────────────────────┴────────────────────────┴───────────────┘ ##  Total params: 55,050 (215.04 KB) ##  Trainable params: 55,050 (215.04 KB) ##  Non-trainable params: 0 (0.00 B) plot(model) plot(model, show_shapes = TRUE)"},{"path":"https://keras3.posit.co/dev/articles/functional_api.html","id":"training-evaluation-and-inference","dir":"Articles","previous_headings":"","what":"Training, evaluation, and inference","title":"The Functional API","text":"Training, evaluation, inference work exactly way models built using functional API Sequential models. Model class offers built-training loop (fit() method) built-evaluation loop (evaluate() method). Note can easily customize loops implement training routines beyond supervised learning (e.g. GANs). , load MNIST image data, reshape vectors, fit model data (monitoring performance validation split), evaluate model test data: reading, see training evaluation guide.","code":"c(c(x_train, y_train), c(x_test, y_test)) %<-% dataset_mnist()  x_train <- array_reshape(x_train, c(60000, 784)) / 255 x_test <- array_reshape(x_test, c(10000, 784)) / 255  model |> compile(   loss = loss_sparse_categorical_crossentropy(from_logits = TRUE),   optimizer = optimizer_rmsprop(),   metrics = \"accuracy\" )  history <- model |> fit(     x_train, y_train, batch_size = 64, epochs = 2, validation_split = 0.2 ) ## Epoch 1/2 ## 750/750 - 2s - 3ms/step - accuracy: 0.8980 - loss: 0.3540 - val_accuracy: 0.9444 - val_loss: 0.1898 ## Epoch 2/2 ## 750/750 - 1s - 1ms/step - accuracy: 0.9512 - loss: 0.1633 - val_accuracy: 0.9605 - val_loss: 0.1387 test_scores <- model |> evaluate(x_test, y_test, verbose=2) ## 313/313 - 1s - 2ms/step - accuracy: 0.9598 - loss: 0.1322 cat(\"Test loss:\", test_scores$loss, \"\\n\") cat(\"Test accuracy:\", test_scores$accuracy, \"\\n\") ## Test loss: 0.1321879 ## Test accuracy: 0.9598"},{"path":"https://keras3.posit.co/dev/articles/functional_api.html","id":"save-and-serialize","dir":"Articles","previous_headings":"","what":"Save and serialize","title":"The Functional API","text":"Saving model serialization work way models built using functional API Sequential models. standard way save functional model call model.save() save entire model single file. can later recreate model file, even code built model longer available. saved file includes : - model architecture - model weight values (learned training) - model training config, (passed compile()) - optimizer state, (restart training left ) details, read model serialization & saving guide.","code":"model |> save_model(\"my_model.keras\") rm(model) # Recreate the exact same model purely from the file: model <- load_model(\"my_model.keras\")"},{"path":"https://keras3.posit.co/dev/articles/functional_api.html","id":"use-the-same-graph-of-layers-to-define-multiple-models","dir":"Articles","previous_headings":"","what":"Use the same graph of layers to define multiple models","title":"The Functional API","text":"functional API, models created specifying inputs outputs graph layers. means single graph layers can used generate multiple models. example , use stack layers instantiate two models: encoder model turns image inputs 16-dimensional vectors, end--end autoencoder model training. , decoding architecture strictly symmetrical encoding architecture, output shape input shape (28, 28, 1). reverse conv_2d layer conv_2d_transpose layer, reverse max_pooling_2d layer upsampling_2d layer.","code":"encoder_input <- keras_input(shape = c(28, 28, 1), name=\"img\") encoder_output <- encoder_input |>   layer_conv_2d(16, 3, activation = \"relu\") |>   layer_conv_2d(32, 3, activation = \"relu\") |>   layer_max_pooling_2d(3) |>   layer_conv_2d(32, 3, activation = \"relu\") |>   layer_conv_2d(16, 3, activation = \"relu\") |>   layer_global_max_pooling_2d()  encoder <- keras_model(encoder_input, encoder_output, name=\"encoder\") summary(encoder) ## Model: \"encoder\" ## ┏━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━┓ ## ┃ Layer (type)                    ┃ Output Shape           ┃       Param # ┃ ## ┡━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━┩ ## │ img (InputLayer)                │ (None, 28, 28, 1)      │             0 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ conv2d (Conv2D)                 │ (None, 26, 26, 16)     │           160 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ conv2d_1 (Conv2D)               │ (None, 24, 24, 32)     │         4,640 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ max_pooling2d (MaxPooling2D)    │ (None, 8, 8, 32)       │             0 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ conv2d_2 (Conv2D)               │ (None, 6, 6, 32)       │         9,248 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ conv2d_3 (Conv2D)               │ (None, 4, 4, 16)       │         4,624 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ global_max_pooling2d            │ (None, 16)             │             0 │ ## │ (GlobalMaxPooling2D)            │                        │               │ ## └─────────────────────────────────┴────────────────────────┴───────────────┘ ##  Total params: 18,672 (72.94 KB) ##  Trainable params: 18,672 (72.94 KB) ##  Non-trainable params: 0 (0.00 B) decoder_output <- encoder_output |>   layer_reshape(c(4, 4, 1)) |>   layer_conv_2d_transpose(16, 3, activation = \"relu\") |>   layer_conv_2d_transpose(32, 3, activation = \"relu\") |>   layer_upsampling_2d(3) |>   layer_conv_2d_transpose(16, 3, activation = \"relu\") |>   layer_conv_2d_transpose(1, 3, activation = \"relu\")  autoencoder <- keras_model(encoder_input, decoder_output, name=\"autoencoder\") summary(autoencoder) ## Model: \"autoencoder\" ## ┏━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━┓ ## ┃ Layer (type)                    ┃ Output Shape           ┃       Param # ┃ ## ┡━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━┩ ## │ img (InputLayer)                │ (None, 28, 28, 1)      │             0 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ conv2d (Conv2D)                 │ (None, 26, 26, 16)     │           160 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ conv2d_1 (Conv2D)               │ (None, 24, 24, 32)     │         4,640 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ max_pooling2d (MaxPooling2D)    │ (None, 8, 8, 32)       │             0 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ conv2d_2 (Conv2D)               │ (None, 6, 6, 32)       │         9,248 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ conv2d_3 (Conv2D)               │ (None, 4, 4, 16)       │         4,624 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ global_max_pooling2d            │ (None, 16)             │             0 │ ## │ (GlobalMaxPooling2D)            │                        │               │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ reshape (Reshape)               │ (None, 4, 4, 1)        │             0 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ conv2d_transpose                │ (None, 6, 6, 16)       │           160 │ ## │ (Conv2DTranspose)               │                        │               │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ conv2d_transpose_1              │ (None, 8, 8, 32)       │         4,640 │ ## │ (Conv2DTranspose)               │                        │               │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ up_sampling2d (UpSampling2D)    │ (None, 24, 24, 32)     │             0 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ conv2d_transpose_2              │ (None, 26, 26, 16)     │         4,624 │ ## │ (Conv2DTranspose)               │                        │               │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ conv2d_transpose_3              │ (None, 28, 28, 1)      │           145 │ ## │ (Conv2DTranspose)               │                        │               │ ## └─────────────────────────────────┴────────────────────────┴───────────────┘ ##  Total params: 28,241 (110.32 KB) ##  Trainable params: 28,241 (110.32 KB) ##  Non-trainable params: 0 (0.00 B)"},{"path":"https://keras3.posit.co/dev/articles/functional_api.html","id":"all-models-are-callable-just-like-layers","dir":"Articles","previous_headings":"","what":"All models are callable, just like layers","title":"The Functional API","text":"can treat model layer invoking Input output another layer. calling model aren’t just reusing architecture model, ’re also reusing weights. see action, ’s different take autoencoder example creates encoder model, decoder model, chains two calls obtain autoencoder model: can see, model can nested: model can contain sub-models (since model just like layer). common use case model nesting ensembling. example, ’s ensemble set models single model averages predictions:","code":"encoder_input <- keras_input(shape = c(28, 28, 1), name=\"img\") encoder_output <- encoder_input |>   layer_conv_2d(16, 3, activation = \"relu\") |>   layer_conv_2d(32, 3, activation = \"relu\") |>   layer_max_pooling_2d(3) |>   layer_conv_2d(32, 3, activation = \"relu\") |>   layer_conv_2d(16, 3, activation = \"relu\") |>   layer_global_max_pooling_2d()  encoder <- keras_model(encoder_input, encoder_output, name=\"encoder\") summary(encoder) ## Model: \"encoder\" ## ┏━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━┓ ## ┃ Layer (type)                    ┃ Output Shape           ┃       Param # ┃ ## ┡━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━┩ ## │ img (InputLayer)                │ (None, 28, 28, 1)      │             0 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ conv2d_4 (Conv2D)               │ (None, 26, 26, 16)     │           160 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ conv2d_5 (Conv2D)               │ (None, 24, 24, 32)     │         4,640 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ max_pooling2d_1 (MaxPooling2D)  │ (None, 8, 8, 32)       │             0 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ conv2d_6 (Conv2D)               │ (None, 6, 6, 32)       │         9,248 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ conv2d_7 (Conv2D)               │ (None, 4, 4, 16)       │         4,624 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ global_max_pooling2d_1          │ (None, 16)             │             0 │ ## │ (GlobalMaxPooling2D)            │                        │               │ ## └─────────────────────────────────┴────────────────────────┴───────────────┘ ##  Total params: 18,672 (72.94 KB) ##  Trainable params: 18,672 (72.94 KB) ##  Non-trainable params: 0 (0.00 B) decoder_input <- keras_input(shape = c(16), name = \"encoded_img\") decoder_output <- decoder_input |>   layer_reshape(c(4, 4, 1)) |>   layer_conv_2d_transpose(16, 3, activation = \"relu\") |>   layer_conv_2d_transpose(32, 3, activation = \"relu\") |>   layer_upsampling_2d(3) |>   layer_conv_2d_transpose(16, 3, activation = \"relu\") |>   layer_conv_2d_transpose(1, 3, activation = \"relu\")  decoder <- keras_model(decoder_input, decoder_output, name = \"decoder\") summary(decoder) ## Model: \"decoder\" ## ┏━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━┓ ## ┃ Layer (type)                    ┃ Output Shape           ┃       Param # ┃ ## ┡━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━┩ ## │ encoded_img (InputLayer)        │ (None, 16)             │             0 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ reshape_1 (Reshape)             │ (None, 4, 4, 1)        │             0 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ conv2d_transpose_4              │ (None, 6, 6, 16)       │           160 │ ## │ (Conv2DTranspose)               │                        │               │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ conv2d_transpose_5              │ (None, 8, 8, 32)       │         4,640 │ ## │ (Conv2DTranspose)               │                        │               │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ up_sampling2d_1 (UpSampling2D)  │ (None, 24, 24, 32)     │             0 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ conv2d_transpose_6              │ (None, 26, 26, 16)     │         4,624 │ ## │ (Conv2DTranspose)               │                        │               │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ conv2d_transpose_7              │ (None, 28, 28, 1)      │           145 │ ## │ (Conv2DTranspose)               │                        │               │ ## └─────────────────────────────────┴────────────────────────┴───────────────┘ ##  Total params: 9,569 (37.38 KB) ##  Trainable params: 9,569 (37.38 KB) ##  Non-trainable params: 0 (0.00 B) autoencoder_input <- keras_input(shape = c(28, 28, 1), name = \"img\") encoded_img <- encoder(autoencoder_input) decoded_img <- decoder(encoded_img) autoencoder <- keras_model(autoencoder_input, decoded_img,                            name = \"autoencoder\") summary(autoencoder) ## Model: \"autoencoder\" ## ┏━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━┓ ## ┃ Layer (type)                    ┃ Output Shape           ┃       Param # ┃ ## ┡━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━┩ ## │ img (InputLayer)                │ (None, 28, 28, 1)      │             0 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ encoder (Functional)            │ (None, 16)             │        18,672 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ decoder (Functional)            │ (None, 28, 28, 1)      │         9,569 │ ## └─────────────────────────────────┴────────────────────────┴───────────────┘ ##  Total params: 28,241 (110.32 KB) ##  Trainable params: 28,241 (110.32 KB) ##  Non-trainable params: 0 (0.00 B) get_model <- function() {   inputs <- keras_input(shape = 128)   outputs <- inputs |> layer_dense(1)   keras_model(inputs, outputs) }  model1 <- get_model() model2 <- get_model() model3 <- get_model()  inputs <- keras_input(shape = 128) y1 <- model1(inputs) y2 <- model2(inputs) y3 <- model3(inputs) outputs <- layer_average(list(y1, y2, y3)) ensemble_model <- keras_model(inputs = inputs, outputs = outputs)"},{"path":[]},{"path":"https://keras3.posit.co/dev/articles/functional_api.html","id":"models-with-multiple-inputs-and-outputs","dir":"Articles","previous_headings":"Manipulate complex graph topologies","what":"Models with multiple inputs and outputs","title":"The Functional API","text":"functional API makes easy manipulate multiple inputs outputs. handled Sequential API. example, ’re building system ranking customer issue tickets priority routing correct department, model three inputs: title ticket (text input), text body ticket (text input), tags added user (categorical input) model two outputs: priority score 0 1 (scalar sigmoid output), department handle ticket (softmax output set departments). can build model lines functional API: Now plot model:  compiling model, can assign different losses output. can even assign different weights loss – modulate contribution total training loss. Since output layers different names, also specify losses loss weights corresponding layer names: Train model passing lists NumPy arrays inputs targets: calling fit Dataset object, yield either list lists like list(list(title_data, body_data, tags_data), list(priority_targets, dept_targets)) list named lists like list(list(title = title_data, body = body_data, tags = tags_data), list(priority = priority_targets, department = dept_targets)). detailed explanation, refer training evaluation guide.","code":"num_tags <- 12  # Number of unique issue tags num_words <- 10000  # Size of vocabulary obtained when preprocessing text data num_departments <- 4  # Number of departments for predictions  title_input <- # Variable-length sequence of ints   keras_input(shape(NA), name = \"title\") body_input <-  # Variable-length sequence of ints   keras_input(shape(NA), name = \"body\") tags_input <-  # Binary vectors of size `num_tags`   keras_input(shape = num_tags, name = \"tags\")  # Embed each word in the title into a 64-dimensional vector title_features <- layer_embedding(title_input, num_words, 64) # Embed each word in the text into a 64-dimensional vector body_features <- layer_embedding(body_input, num_words, 64)  # Reduce sequence of embedded words in the title # into a single 128-dimensional vector title_features <- layer_lstm(title_features, 128) # Reduce sequence of embedded words in the body # into a single 32-dimensional vector body_features <- layer_lstm(body_features, 32)  # Merge all available features into a single large vector via concatenation x <- layer_concatenate(title_features, body_features, tags_input)  # Stick a logistic regression for priority prediction on top of the features priority_pred <- layer_dense(x, 1, name = \"priority\")  # Stick a department classifier on top of the features department_pred <- layer_dense(x, num_departments, name = \"department\")  # Instantiate an end-to-end model predicting both priority and department model <- keras_model(   inputs = list(title_input, body_input, tags_input),   outputs = list(priority = priority_pred, department = department_pred) ) plot(model, show_shapes = TRUE) model |> compile(   optimizer = optimizer_rmsprop(1e-3),   loss = list(     loss_binary_crossentropy(from_logits = TRUE),     loss_categorical_crossentropy(from_logits = TRUE)   ),   loss_weights = c(1.0, 0.2) ) model |> compile(   optimizer = optimizer_rmsprop(1e-3),   loss = list(     priority = loss_binary_crossentropy(from_logits = TRUE),     department = loss_categorical_crossentropy(from_logits = TRUE)   ),   loss_weights = list(priority = 1.0, department = 0.2) ) # Dummy input data title_data <- random_integer(c(1280, 12), 0, num_words) body_data <- random_integer(c(1280, 100), 0, num_words) tags_data <- random_integer(c(1280, num_tags), 0, 2)  # Dummy target data priority_targets <- random_normal(c(1280, 1)) dept_targets <- random_integer(c(1280, num_departments), 0, 2)  model |> fit(   list(title = title_data, body = body_data, tags = tags_data),   list(priority = priority_targets, department = dept_targets),   epochs = 2,   batch_size = 32 ) ## Epoch 1/2 ## 40/40 - 3s - 64ms/step - department_loss: 2.8465 - loss: 0.7669 - priority_loss: 0.1976 ## Epoch 2/2 ## 40/40 - 0s - 5ms/step - department_loss: 2.8554 - loss: 0.7538 - priority_loss: 0.1828"},{"path":"https://keras3.posit.co/dev/articles/functional_api.html","id":"a-toy-resnet-model","dir":"Articles","previous_headings":"Manipulate complex graph topologies","what":"A toy ResNet model","title":"The Functional API","text":"addition models multiple inputs outputs, functional API makes easy manipulate non-linear connectivity topologies – models layers connected sequentially, Sequential API handle. common use case residual connections. Let’s build toy ResNet model CIFAR10 demonstrate : Plot model:  Now train model:","code":"inputs <- keras_input(shape = c(32, 32, 3), name = \"img\") block_1_output <- inputs |>   layer_conv_2d(32, kernel_size = 3, activation = \"relu\") |>   layer_conv_2d(64, kernel_size = 3, activation = \"relu\") |>   layer_max_pooling_2d(pool_size = 3)  block_2_output <- block_1_output |>   layer_conv_2d(32, kernel_size = 3, activation = \"relu\", padding = \"same\") |>   layer_conv_2d(64, kernel_size = 3, activation = \"relu\", padding = \"same\") |>   layer_add(block_1_output)  block_3_output <- block_2_output |>   layer_conv_2d(64, kernel_size = 3, activation = \"relu\", padding = \"same\") |>   layer_conv_2d(64, kernel_size = 3, activation = \"relu\", padding = \"same\") |>   layer_add(block_2_output)  outputs <- block_3_output |>   layer_conv_2d(64, 3, activation = \"relu\") |>   layer_global_average_pooling_2d() |>   layer_dense(256, activation = \"relu\") |>   layer_dropout(0.5) |>   layer_dense(10)  model <- keras_model(inputs, outputs, name = \"toy_resnet\") summary(model) ## Model: \"toy_resnet\" ## ┏━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━┓ ## ┃ Layer (type)        ┃ Output Shape      ┃    Param # ┃ Connected to      ┃ ## ┡━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━┩ ## │ img (InputLayer)    │ (None, 32, 32, 3) │          0 │ -                 │ ## ├─────────────────────┼───────────────────┼────────────┼───────────────────┤ ## │ conv2d_8 (Conv2D)   │ (None, 30, 30,    │        896 │ img[0][0]         │ ## │                     │ 32)               │            │                   │ ## ├─────────────────────┼───────────────────┼────────────┼───────────────────┤ ## │ conv2d_9 (Conv2D)   │ (None, 28, 28,    │     18,496 │ conv2d_8[0][0]    │ ## │                     │ 64)               │            │                   │ ## ├─────────────────────┼───────────────────┼────────────┼───────────────────┤ ## │ max_pooling2d_2     │ (None, 9, 9, 64)  │          0 │ conv2d_9[0][0]    │ ## │ (MaxPooling2D)      │                   │            │                   │ ## ├─────────────────────┼───────────────────┼────────────┼───────────────────┤ ## │ conv2d_10 (Conv2D)  │ (None, 9, 9, 32)  │     18,464 │ max_pooling2d_2[… │ ## ├─────────────────────┼───────────────────┼────────────┼───────────────────┤ ## │ conv2d_11 (Conv2D)  │ (None, 9, 9, 64)  │     18,496 │ conv2d_10[0][0]   │ ## ├─────────────────────┼───────────────────┼────────────┼───────────────────┤ ## │ add (Add)           │ (None, 9, 9, 64)  │          0 │ conv2d_11[0][0],  │ ## │                     │                   │            │ max_pooling2d_2[… │ ## ├─────────────────────┼───────────────────┼────────────┼───────────────────┤ ## │ conv2d_12 (Conv2D)  │ (None, 9, 9, 64)  │     36,928 │ add[0][0]         │ ## ├─────────────────────┼───────────────────┼────────────┼───────────────────┤ ## │ conv2d_13 (Conv2D)  │ (None, 9, 9, 64)  │     36,928 │ conv2d_12[0][0]   │ ## ├─────────────────────┼───────────────────┼────────────┼───────────────────┤ ## │ add_1 (Add)         │ (None, 9, 9, 64)  │          0 │ conv2d_13[0][0],  │ ## │                     │                   │            │ add[0][0]         │ ## ├─────────────────────┼───────────────────┼────────────┼───────────────────┤ ## │ conv2d_14 (Conv2D)  │ (None, 7, 7, 64)  │     36,928 │ add_1[0][0]       │ ## ├─────────────────────┼───────────────────┼────────────┼───────────────────┤ ## │ global_average_poo… │ (None, 64)        │          0 │ conv2d_14[0][0]   │ ## │ (GlobalAveragePool… │                   │            │                   │ ## ├─────────────────────┼───────────────────┼────────────┼───────────────────┤ ## │ dense_6 (Dense)     │ (None, 256)       │     16,640 │ global_average_p… │ ## ├─────────────────────┼───────────────────┼────────────┼───────────────────┤ ## │ dropout (Dropout)   │ (None, 256)       │          0 │ dense_6[0][0]     │ ## ├─────────────────────┼───────────────────┼────────────┼───────────────────┤ ## │ dense_7 (Dense)     │ (None, 10)        │      2,570 │ dropout[0][0]     │ ## └─────────────────────┴───────────────────┴────────────┴───────────────────┘ ##  Total params: 186,346 (727.91 KB) ##  Trainable params: 186,346 (727.91 KB) ##  Non-trainable params: 0 (0.00 B) plot(model, show_shapes = TRUE) c(c(x_train, y_train), c(x_test, y_test)) %<-% dataset_cifar10()  x_train <- x_train / 255.0 x_test <- x_test / 255.0  model |> compile(   optimizer = optimizer_rmsprop(1e-3),   loss = loss_sparse_categorical_crossentropy(from_logits = TRUE),   metrics = \"acc\" ) # We restrict the data to the first 1000 samples so as to limit the # guide render time. # Try to train on the entire dataset until convergence! model |> fit(   x_train[1:1000, , , ],   y_train[1:1000, ],   batch_size = 64,   epochs = 1,   validation_split = 0.2 ) ## 13/13 - 6s - 430ms/step - acc: 0.1250 - loss: 2.2998 - val_acc: 0.1250 - val_loss: 2.2939"},{"path":"https://keras3.posit.co/dev/articles/functional_api.html","id":"shared-layers","dir":"Articles","previous_headings":"","what":"Shared layers","title":"The Functional API","text":"Another good use functional API models use shared layers. Shared layers layer instances reused multiple times model – learn features correspond multiple paths graph--layers. Shared layers often used encode inputs similar spaces (say, two different pieces text feature similar vocabulary). enable sharing information across different inputs, make possible train model less data. given word seen one inputs, benefit processing inputs pass shared layer. share layer functional API, call layer instance multiple times. instance, ’s Embedding layer shared across two different text inputs:","code":"# Embedding for 1000 unique words mapped to 128-dimensional vectors shared_embedding <- layer_embedding(input_dim = 1000, output_dim = 128)  # Variable-length sequence of integers text_input_a <- keras_input(shape = shape(NA), dtype=\"int32\")  # Variable-length sequence of integers text_input_b <- keras_input(shape = shape(NA), dtype=\"int32\")  # Reuse the same layer to encode both inputs encoded_input_a <- shared_embedding(text_input_a) encoded_input_b <- shared_embedding(text_input_b)"},{"path":"https://keras3.posit.co/dev/articles/functional_api.html","id":"extract-and-reuse-nodes-in-the-graph-of-layers","dir":"Articles","previous_headings":"","what":"Extract and reuse nodes in the graph of layers","title":"The Functional API","text":"graph layers manipulating static data structure, can accessed inspected. able plot functional models images. also means can access activations intermediate layers (“nodes” graph) reuse elsewhere – useful something like feature extraction. Let’s look example. VGG19 model weights pretrained ImageNet: intermediate activations model, obtained querying graph data structure: Use features create new feature-extraction model returns values intermediate layer activations: comes handy tasks like neural style transfer, among things.","code":"vgg19 <- application_vgg19() features_list <- lapply(vgg19$layers, function(x) x$output) feat_extraction_model <- keras_model(inputs = vgg19$input,                                      outputs = features_list)  img <- random_normal(c(1, 224, 224, 3)) extracted_features <- feat_extraction_model(img)"},{"path":"https://keras3.posit.co/dev/articles/functional_api.html","id":"extend-the-api-using-custom-layers","dir":"Articles","previous_headings":"","what":"Extend the API using custom layers","title":"The Functional API","text":"keras includes wide range built-layers, example: Convolutional layers: conv_1d, conv_2d, conv_3d, conv_2d_transpose Pooling layers: max_pooling_1d, max_pooling_2d, max_pooling_3d, average_pooling_3d RNN layers: gru, lstm, conv_lstm_2d batch_normalization, dropout, embedding, etc. don’t find need, ’s easy extend API creating layers. layers subclass Layer class implement: call method, specifies computation done layer. build method, creates weights layer (just style convention since can create weights initialize, well). learn creating layers scratch, read custom layers models guide. following basic implementation layer_dense(): serialization support custom layer, define get_config() method returns constructor arguments layer instance: Optionally, implement class method from_config(cls, config) used recreating layer instance given config dictionary. default implementation from_config :","code":"custom_dense <- Layer(   classname = \"CustomDense\",   initialize = function(units = 32) {     super$initialize()     self$units <- as.integer(units)   },   build = function(input_shape) {     self$w <- self$add_weight(       shape = shape(input_shape[[2]], self$units),       initializer = \"random_normal\",       trainable = TRUE,     )     self$b <- self$add_weight(       shape = shape(self$units),             initializer=\"random_normal\",       trainable = TRUE         )   },   call = function(inputs) {     op_matmul(inputs, self$w) + self$b   } )  inputs <- keras_input(c(4)) outputs <- custom_dense(inputs, 10)  model <- keras_model(inputs, outputs) custom_dense <- Layer(   classname = \"CustomDense\",    initialize = function(units = 32, ...) {     super$initialize()     self$units <- as.integer(units)   },    build = function(input_shape) {     self$w <- self$add_weight(       shape = shape(input_shape[[2]], self$units),       initializer = \"random_normal\",       trainable = TRUE,     )     self$b <- self$add_weight(       shape = shape(self$units),             initializer=\"random_normal\",       trainable = TRUE         )   },    call = function(inputs) {     op_matmul(inputs, self$w) + self$b   },    get_config = function() {     list(units = self$units)   } )  inputs <- keras_input(c(4)) outputs <- custom_dense(inputs, 10)  model <- keras_model(inputs, outputs) config <- get_config(model)  new_model <-   from_config(config, custom_objects = list(CustomDense = custom_dense)) from_config <- function(cls, config) {   do.call(cls, config) }"},{"path":"https://keras3.posit.co/dev/articles/functional_api.html","id":"when-to-use-the-functional-api","dir":"Articles","previous_headings":"","what":"When to use the functional API","title":"The Functional API","text":"use Keras functional API create new model, just subclass Model class directly? general, functional API higher-level, easier safer, number features subclassed models support. However, model subclassing provides greater flexibility building models easily expressible directed acyclic graphs layers. example, implement Tree-RNN functional API subclass Model directly. -depth look differences functional API model subclassing, read Symbolic Imperative APIs TensorFlow 2.0?.","code":""},{"path":"https://keras3.posit.co/dev/articles/functional_api.html","id":"functional-api-strengths","dir":"Articles","previous_headings":"When to use the functional API","what":"Functional API strengths:","title":"The Functional API","text":"following properties also true Sequential models (also data structures), true subclassed models (R Python (byte)code, data structures).","code":""},{"path":"https://keras3.posit.co/dev/articles/functional_api.html","id":"less-verbose","dir":"Articles","previous_headings":"When to use the functional API > Functional API strengths:","what":"Less verbose","title":"The Functional API","text":"super$initialize(...), call = function(...), self$..., etc. Compare: subclassed version:","code":"inputs <- keras_input(shape = shape(32)) outputs <- inputs |>   layer_dense(64, activation = \"relu\") |>   layer_dense(10) mlp <- keras_model(inputs, outputs) MLP <- Model(   classname = \"MLP\",   initialize = function(...) {     super$initialize(...)     self$dense_1 <- layer_dense(units = 64, activation = \"relu\")     self$dense_2 <- layer_dense(units = 10)   },   call = function(inputs) {     inputs |>       self$dense_1() |>       self$dense_2()   } )  # Instantiate the model. mlp <- MLP() # Necessary to create the model's state. # The model doesn't have a state until it's called at least once. out <- mlp(op_zeros(c(1, 32)))"},{"path":"https://keras3.posit.co/dev/articles/functional_api.html","id":"model-validation-while-defining-its-connectivity-graph","dir":"Articles","previous_headings":"When to use the functional API > Functional API strengths:","what":"Model validation while defining its connectivity graph","title":"The Functional API","text":"functional API, input specification (shape dtype) created advance (using Input). Every time call layer, layer checks specification passed matches assumptions, raise helpful error message . guarantees model can build functional API run. debugging – convergence-related debugging – happens statically model construction execution time. similar type checking compiler.","code":""},{"path":"https://keras3.posit.co/dev/articles/functional_api.html","id":"a-functional-model-is-plottable-and-inspectable","dir":"Articles","previous_headings":"When to use the functional API > Functional API strengths:","what":"A functional model is plottable and inspectable","title":"The Functional API","text":"can plot model graph, can easily access intermediate nodes graph. example, extract reuse activations intermediate layers (seen previous example):","code":"features_list <- lapply(vgg19$layers, function(x) x$output) feat_extraction_model <- keras_model(inputs = vgg19$input,                                      outputs = features_list)"},{"path":"https://keras3.posit.co/dev/articles/functional_api.html","id":"a-functional-model-can-be-serialized-or-cloned","dir":"Articles","previous_headings":"When to use the functional API > Functional API strengths:","what":"A functional model can be serialized or cloned","title":"The Functional API","text":"functional model data structure rather piece code, safely serializable can saved single file allows recreate exact model without access original code. See serialization & saving guide. serialize subclassed model, necessary implementer specify get_config() from_config() method model level.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/articles/functional_api.html","id":"it-does-not-support-dynamic-architectures","dir":"Articles","previous_headings":"When to use the functional API > Functional API weakness:","what":"It does not support dynamic architectures","title":"The Functional API","text":"functional API treats models DAGs layers. true deep learning architectures, – example, recursive networks Tree RNNs follow assumption implemented functional API.","code":""},{"path":"https://keras3.posit.co/dev/articles/functional_api.html","id":"mix-and-match-api-styles","dir":"Articles","previous_headings":"","what":"Mix-and-match API styles","title":"The Functional API","text":"Choosing functional API Model subclassing isn’t binary decision restricts one category models. models keras API can interact , whether ’re Sequential models, functional models, subclassed models written scratch. can always use functional model Sequential model part subclassed model layer: can use subclassed layer model functional API long implements call method follows one following patterns: call(inputs, ...) – inputs tensor nested structure tensors (e.g. list tensors), ... non-tensor arguments (non-inputs). call(inputs, training = NULL, ...) – training boolean indicating whether layer behave training mode inference mode. call(inputs, mask = NULL, ...) – mask boolean mask tensor (useful RNNs, instance). call(inputs, training = NULL, mask = NULL, ...) – course, can masking training-specific behavior time. Additionally, implement get_config() method custom Layer model, functional models create still serializable cloneable. ’s quick example custom RNN, written scratch, used functional model:","code":"units <- 32 timesteps <- 10 input_dim <- 5  # Define a Functional model inputs <- keras_input(shape(NA, units)) outputs <- inputs |>   layer_global_average_pooling_1d() |>   layer_dense(units = 1)  model <- keras_model(inputs, outputs)  layer_custom_rnn <- Layer(   classname = \"CustomRNN\",   initialize = function(...) {     super$initialize(...)     self$units <- units     self$projection_1 <- layer_dense(units = units, activation = \"tanh\")     self$projection_2 <- layer_dense(units = units, activation = \"tanh\")     self$classifier <- model   },   call = function(inputs, ...) {     outputs <- list()     state <- op_zeros(c(shape(inputs)[[1]], self$units))     for (t in 1:(shape(inputs)[[2]])) {       x <- inputs[, t, ]       h <- self$projection_1(x)       y <- h + self$projection_2(state)       state <- y       outputs[[t]] <- y     }     features <- op_stack(outputs, axis = 2)     self$classifier(features)   } )  rnn <- layer_custom_rnn() out <- rnn(op_zeros(c(1, timesteps, input_dim))) units <- 32 timesteps <- 10 input_dim <- 5 batch_size <- 16  layer_custom_rnn <- Layer(   \"custom_rnn\",   initialize = function(...) {     super$initialize(...)     self$units <- units     self$projection_1 <- layer_dense(units = units, activation = \"tanh\")     self$projection_2 <- layer_dense(units = units, activation = \"tanh\")     self$classifier <- layer_dense(units = 1)   },   call = function(inputs, ...) {     outputs <- list()     state <- op_zeros(c(shape(inputs)[[1]], self$units))     for (t in 1:(shape(inputs)[[2]])) {       x <- inputs[, t, ]       h <- self$projection_1(x)       y <- h + self$projection_2(state)       state <- y       outputs[[t]] <- y     }     features <- op_stack(outputs, axis = 2)     self$classifier(features)   } )  # Note that you specify a static batch size for the inputs with the `batch_shape` # arg, because the inner computation of `layer_custom_rnn()` requires a static batch size # (when you create the `state` zeros tensor). inputs <- keras_input(batch_shape = shape(batch_size, timesteps, input_dim)) outputs <- inputs |>   layer_conv_1d(filters = 32, kernel_size = 3) |>   layer_custom_rnn()  model <- keras_model(inputs, outputs) out <- model(op_zeros(c(1, 10, 5)))"},{"path":"https://keras3.posit.co/dev/articles/getting_started.html","id":"overview","dir":"Articles","previous_headings":"","what":"Overview","title":"Getting Started with Keras","text":"Keras high-level neural networks API developed focus enabling fast experimentation. able go idea result least possible delay key good research. Keras following key features: Allows code run CPU GPU, seamlessly. User-friendly API makes easy quickly prototype deep learning models. Built-support convolutional networks (computer vision), recurrent networks (sequence processing), combination . Supports arbitrary network architectures: multi-input multi-output models, layer sharing, model sharing, etc. means Keras appropriate building essentially deep learning model, memory network neural Turing machine. website provides documentation R interface Keras. See main Keras website https://keras.io additional information project.","code":""},{"path":"https://keras3.posit.co/dev/articles/getting_started.html","id":"installation","dir":"Articles","previous_headings":"","what":"Installation","title":"Getting Started with Keras","text":"First, install keras R package: install development version : Keras R interface requires backend engine installed. TensorFlow default. provide default installation Keras TensorFlow GPU capable, GPU available. want customized installation, e.g. see documentation install_keras() installation section.","code":"install.packages(\"keras3\") remotes::install_github(\"rstudio/keras3\") keras3::install_keras(backend = \"tensorflow\")"},{"path":"https://keras3.posit.co/dev/articles/getting_started.html","id":"mnist-example","dir":"Articles","previous_headings":"","what":"MNIST Example","title":"Getting Started with Keras","text":"can learn basics Keras walking simple example: recognizing handwritten digits MNIST dataset. MNIST consists 28 x 28 grayscale images handwritten digits like :  dataset also includes labels image, telling us digit . example, labels images 5, 0, 4, 1.","code":""},{"path":"https://keras3.posit.co/dev/articles/getting_started.html","id":"preparing-the-data","dir":"Articles","previous_headings":"MNIST Example","what":"Preparing the Data","title":"Getting Started with Keras","text":"MNIST dataset included Keras can accessed using dataset_mnist() function. load dataset create variables test training data: x data 3-d array (images, width, height) grayscale values. prepare data training convert 3-d arrays matrices reshaping width height single dimension (28x28 images flattened length 784 vectors). , convert grayscale values integers ranging 0 255 floating point values ranging 0 1: Note use array_reshape() function rather dim<-() function reshape array. data re-interpreted using row-major semantics (opposed R’s default column-major semantics), turn compatible way numerical libraries called Keras interpret array dimensions. y data integer vector values ranging 0 9. prepare data training one-hot encode vectors binary class matrices using Keras to_categorical() function:","code":"library(keras3) mnist <- dataset_mnist() x_train <- mnist$train$x y_train <- mnist$train$y x_test <- mnist$test$x y_test <- mnist$test$y # reshape x_train <- array_reshape(x_train, c(nrow(x_train), 784)) x_test <- array_reshape(x_test, c(nrow(x_test), 784)) # rescale x_train <- x_train / 255 x_test <- x_test / 255 y_train <- to_categorical(y_train, 10) y_test <- to_categorical(y_test, 10)"},{"path":"https://keras3.posit.co/dev/articles/getting_started.html","id":"defining-the-model","dir":"Articles","previous_headings":"MNIST Example","what":"Defining the Model","title":"Getting Started with Keras","text":"core data structure Keras model, way organize layers. simplest type model Sequential model, linear stack layers. begin creating sequential model adding layers using pipe (|>) operator: input_shape argument first layer specifies shape input data (length 784 numeric vector representing grayscale image). final layer outputs length 10 numeric vector (probabilities digit) using softmax activation function. Use summary() function print details model: Next, compile model appropriate loss function, optimizer, metrics:","code":"model <- keras_model_sequential(input_shape = c(784)) model |>   layer_dense(units = 256, activation = 'relu') |>   layer_dropout(rate = 0.4) |>   layer_dense(units = 128, activation = 'relu') |>   layer_dropout(rate = 0.3) |>   layer_dense(units = 10, activation = 'softmax') summary(model) ## Model: \"sequential\" ## ┏━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━┓ ## ┃ Layer (type)                    ┃ Output Shape           ┃       Param # ┃ ## ┡━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━┩ ## │ dense (Dense)                   │ (None, 256)            │       200,960 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ dropout (Dropout)               │ (None, 256)            │             0 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ dense_1 (Dense)                 │ (None, 128)            │        32,896 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ dropout_1 (Dropout)             │ (None, 128)            │             0 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ dense_2 (Dense)                 │ (None, 10)             │         1,290 │ ## └─────────────────────────────────┴────────────────────────┴───────────────┘ ##  Total params: 235,146 (918.54 KB) ##  Trainable params: 235,146 (918.54 KB) ##  Non-trainable params: 0 (0.00 B) plot(model) model |> compile(   loss = 'categorical_crossentropy',   optimizer = optimizer_rmsprop(),   metrics = c('accuracy') )"},{"path":"https://keras3.posit.co/dev/articles/getting_started.html","id":"training-and-evaluation","dir":"Articles","previous_headings":"MNIST Example","what":"Training and Evaluation","title":"Getting Started with Keras","text":"Use fit() function train model 30 epochs using batches 128 images: history object returned fit() includes loss accuracy metrics can plot: Evaluate model’s performance test data: Generate predictions new data: Keras provides vocabulary building deep learning models simple, elegant, intuitive. Building question answering system, image classification model, neural Turing machine, model just straightforward.","code":"history <- model |> fit(   x_train, y_train,   epochs = 30, batch_size = 128,   validation_split = 0.2 ) plot(history) model |> evaluate(x_test, y_test) ## 313/313 - 1s - 2ms/step - accuracy: 0.9810 - loss: 0.0842 ## $accuracy ## [1] 0.981 ## ## $loss ## [1] 0.08419506 probs <- model |> predict(x_test) ## 313/313 - 0s - 1ms/step max.col(probs) - 1L ##   [1] 7 2 1 0 4 1 4 9 5 9 0 6 9 0 1 5 9 7 3 4 9 6 6 5 4 0 7 4 0 1 3 1 3 4 7 ##  [36] 2 7 1 2 1 1 7 4 2 3 5 1 2 4 4 6 3 5 5 6 0 4 1 9 5 7 8 9 3 7 4 6 4 3 0 ##  [71] 7 0 2 9 1 7 3 2 9 7 7 6 2 7 8 4 7 3 6 1 3 6 9 3 1 4 1 7 6 9 ##  [ reached 'max' / getOption(\"max.print\") -- omitted 9900 entries ]"},{"path":"https://keras3.posit.co/dev/articles/getting_started.html","id":"deep-learning-with-r-book","dir":"Articles","previous_headings":"MNIST Example","what":"Deep Learning with R Book","title":"Getting Started with Keras","text":"want comprehensive introduction Keras concepts practice deep learning, recommend Deep Learning R, 2nd Edition book Manning. book collaboration François Chollet, creator (Python) Keras, J.J. Allaire, wrote original R interface Keras, Tomasz Kalinowski, maintainer R interface Keras. book presumes significant knowledge machine learning deep learning, goes way basic theory advanced practical applications, using R interface Keras.","code":""},{"path":"https://keras3.posit.co/dev/articles/getting_started.html","id":"why-this-name-keras","dir":"Articles","previous_headings":"","what":"Why this name, Keras?","title":"Getting Started with Keras","text":"Keras (κέρας) means horn Greek. reference literary image ancient Greek Latin literature, first found Odyssey, dream spirits (Oneiroi, singular Oneiros) divided deceive men false visions, arrive Earth gate ivory, announce future come pass, arrive gate horn. ’s play words κέρας (horn) / κραίνω (fulfill), ἐλέφας (ivory) / ἐλεφαίρομαι (deceive). Keras initially developed part research effort project ONEIROS (Open-ended Neuro-Electronic Intelligent Robot Operating System). “Oneiroi beyond unravelling –can sure tale tell? men look comes pass. Two gates give passage fleeting Oneiroi; one made horn, one ivory. Oneiroi pass sawn ivory deceitful, bearing message fulfilled; come polished horn truth behind , accomplished men see .” Homer, Odyssey 19. 562 ff (Shewring translation).","code":""},{"path":"https://keras3.posit.co/dev/articles/intro_to_keras_for_engineers.html","id":"introduction","dir":"Articles","previous_headings":"","what":"Introduction","title":"Introduction to Keras for engineers","text":"Keras 3 deep learning framework works TensorFlow, JAX, PyTorch interchangeably. notebook walk key Keras 3 workflows. Let’s start installing Keras 3:","code":"install.packages(\"keras3\") keras3::install_keras()"},{"path":"https://keras3.posit.co/dev/articles/intro_to_keras_for_engineers.html","id":"setup","dir":"Articles","previous_headings":"","what":"Setup","title":"Introduction to Keras for engineers","text":"’re going using tensorflow backend – can edit string \"jax\" \"torch\" hit “Restart runtime”, whole notebook run just ! entire guide backend-agnostic.","code":"library(tensorflow, exclude = c(\"shape\", \"set_random_seed\")) library(keras3)  # Note that you must configure the backend # before calling any other keras functions. # The backend cannot be changed once the # package is imported. use_backend(\"tensorflow\")"},{"path":"https://keras3.posit.co/dev/articles/intro_to_keras_for_engineers.html","id":"a-first-example-a-mnist-convnet","dir":"Articles","previous_headings":"","what":"A first example: A MNIST convnet","title":"Introduction to Keras for engineers","text":"Let’s start Hello World ML: training convnet classify MNIST digits. ’s data: ’s model. Different model-building options Keras offers include: Sequential API (use ) Functional API (typical) Writing models via subclassing (advanced use cases) ’s model summary: use compile() method specify optimizer, loss function, metrics monitor. Note JAX TensorFlow backends, XLA compilation turned default. Let’s train evaluate model. ’ll set aside validation split 15% data training monitor generalization unseen data. training, saving model end epoch. can also save model latest state like : reload like : Next, can query predictions class probabilities predict(): ’s basics!","code":"# Load the data and split it between train and test sets c(c(x_train, y_train), c(x_test, y_test)) %<-% keras3::dataset_mnist()  # Scale images to the [0, 1] range x_train <- x_train / 255 x_test <- x_test / 255  # Make sure images have shape (28, 28, 1) x_train <- array_reshape(x_train, c(-1, 28, 28, 1)) x_test <- array_reshape(x_test, c(-1, 28, 28, 1))  dim(x_train) ## [1] 60000    28    28     1 dim(x_test) ## [1] 10000    28    28     1 # Model parameters num_classes <- 10 input_shape <- c(28, 28, 1)  model <- keras_model_sequential(input_shape = input_shape) model |>   layer_conv_2d(filters = 64, kernel_size = c(3, 3), activation = \"relu\") |>   layer_conv_2d(filters = 64, kernel_size = c(3, 3), activation = \"relu\") |>   layer_max_pooling_2d(pool_size = c(2, 2)) |>   layer_conv_2d(filters = 128, kernel_size = c(3, 3), activation = \"relu\") |>   layer_conv_2d(filters = 128, kernel_size = c(3, 3), activation = \"relu\") |>   layer_global_average_pooling_2d() |>   layer_dropout(rate = 0.5) |>   layer_dense(units = num_classes, activation = \"softmax\") summary(model) ## Model: \"sequential\" ## ┏━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━┓ ## ┃ Layer (type)                    ┃ Output Shape           ┃       Param # ┃ ## ┡━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━┩ ## │ conv2d (Conv2D)                 │ (None, 26, 26, 64)     │           640 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ conv2d_1 (Conv2D)               │ (None, 24, 24, 64)     │        36,928 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ max_pooling2d (MaxPooling2D)    │ (None, 12, 12, 64)     │             0 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ conv2d_2 (Conv2D)               │ (None, 10, 10, 128)    │        73,856 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ conv2d_3 (Conv2D)               │ (None, 8, 8, 128)      │       147,584 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ global_average_pooling2d        │ (None, 128)            │             0 │ ## │ (GlobalAveragePooling2D)        │                        │               │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ dropout (Dropout)               │ (None, 128)            │             0 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ dense (Dense)                   │ (None, 10)             │         1,290 │ ## └─────────────────────────────────┴────────────────────────┴───────────────┘ ##  Total params: 260,298 (1016.79 KB) ##  Trainable params: 260,298 (1016.79 KB) ##  Non-trainable params: 0 (0.00 B) model |> compile(   optimizer = \"adam\",   loss = \"sparse_categorical_crossentropy\",   metrics = list(     metric_sparse_categorical_accuracy(name = \"acc\")   ) ) batch_size <- 128 epochs <- 10  callbacks <- list(   callback_model_checkpoint(filepath=\"model_at_epoch_{epoch}.keras\"),   callback_early_stopping(monitor=\"val_loss\", patience=2) )  model |> fit(   x_train, y_train,   batch_size = batch_size,   epochs = epochs,   validation_split = 0.15,   callbacks = callbacks ) ## Epoch 1/10 ## 399/399 - 7s - 17ms/step - acc: 0.7464 - loss: 0.7476 - val_acc: 0.9627 - val_loss: 0.1272 ## Epoch 2/10 ## 399/399 - 2s - 5ms/step - acc: 0.9376 - loss: 0.2072 - val_acc: 0.9771 - val_loss: 0.0776 ## Epoch 3/10 ## 399/399 - 2s - 5ms/step - acc: 0.9571 - loss: 0.1475 - val_acc: 0.9816 - val_loss: 0.0633 ## Epoch 4/10 ## 399/399 - 2s - 5ms/step - acc: 0.9652 - loss: 0.1175 - val_acc: 0.9863 - val_loss: 0.0488 ## Epoch 5/10 ## 399/399 - 2s - 5ms/step - acc: 0.9713 - loss: 0.1004 - val_acc: 0.9870 - val_loss: 0.0469 ## Epoch 6/10 ## 399/399 - 2s - 5ms/step - acc: 0.9754 - loss: 0.0854 - val_acc: 0.9879 - val_loss: 0.0407 ## Epoch 7/10 ## 399/399 - 2s - 5ms/step - acc: 0.9767 - loss: 0.0788 - val_acc: 0.9899 - val_loss: 0.0378 ## Epoch 8/10 ## 399/399 - 2s - 5ms/step - acc: 0.9795 - loss: 0.0688 - val_acc: 0.9867 - val_loss: 0.0438 ## Epoch 9/10 ## 399/399 - 2s - 5ms/step - acc: 0.9799 - loss: 0.0666 - val_acc: 0.9893 - val_loss: 0.0396 score <- model |> evaluate(x_test, y_test, verbose = 0) save_model(model, \"final_model.keras\", overwrite=TRUE) model <- load_model(\"final_model.keras\") predictions <- model |> predict(x_test) ## 313/313 - 1s - 2ms/step dim(predictions) ## [1] 10000    10"},{"path":"https://keras3.posit.co/dev/articles/intro_to_keras_for_engineers.html","id":"writing-cross-framework-custom-components","dir":"Articles","previous_headings":"","what":"Writing cross-framework custom components","title":"Introduction to Keras for engineers","text":"Keras enables write custom Layers, Models, Metrics, Losses, Optimizers work across TensorFlow, JAX, PyTorch codebase. Let’s take look custom layers first. op_ namespace contains: implementation NumPy API, e.g. op_stack op_matmul. set neural network specific ops absent NumPy, op_conv op_binary_crossentropy. Let’s make custom Dense layer works backends: Next, let’s make custom Dropout layer relies random_* namespace: Next, let’s write custom subclassed model uses two custom layers: Let’s compile fit :","code":"layer_my_dense <- Layer(   classname = \"MyDense\",   initialize = function(units, activation = NULL, name = NULL, ...) {     super$initialize(name = name, ...)     self$units <- units     self$activation <- activation   },   build = function(input_shape) {     input_dim <- tail(input_shape, 1)     self$w <- self$add_weight(       shape = shape(input_dim, self$units),       initializer = initializer_glorot_normal(),       name = \"kernel\",       trainable = TRUE     )     self$b <- self$add_weight(       shape = shape(self$units),       initializer = initializer_zeros(),       name = \"bias\",       trainable = TRUE     )   },   call = function(inputs) {     # Use Keras ops to create backend-agnostic layers/metrics/etc.     x <- op_matmul(inputs, self$w) + self$b     if (!is.null(self$activation))       x <- self$activation(x)     x   } ) layer_my_dropout <- Layer(   \"MyDropout\",   initialize = function(rate, name = NULL, seed = NULL, ...) {     super$initialize(name = name)     self$rate <- rate     # Use seed_generator for managing RNG state.     # It is a state element and its seed variable is     # tracked as part of `layer$variables`.     self$seed_generator <- random_seed_generator(seed)   },   call = function(inputs) {     # Use `keras3::random_*` for random ops.     random_dropout(inputs, self$rate, seed = self$seed_generator)   } ) MyModel <- Model(   \"MyModel\",   initialize = function(num_classes, ...) {     super$initialize(...)     self$conv_base <-       keras_model_sequential() |>       layer_conv_2d(64, kernel_size = c(3, 3), activation = \"relu\") |>       layer_conv_2d(64, kernel_size = c(3, 3), activation = \"relu\") |>       layer_max_pooling_2d(pool_size = c(2, 2)) |>       layer_conv_2d(128, kernel_size = c(3, 3), activation = \"relu\") |>       layer_conv_2d(128, kernel_size = c(3, 3), activation = \"relu\") |>       layer_global_average_pooling_2d()      self$dp <- layer_my_dropout(rate = 0.5)     self$dense <- layer_my_dense(units = num_classes,                                  activation = activation_softmax)   },   call = function(inputs) {     inputs |>       self$conv_base() |>       self$dp() |>       self$dense()   } ) model <- MyModel(num_classes = 10) model |> compile(   loss = loss_sparse_categorical_crossentropy(),   optimizer = optimizer_adam(learning_rate = 1e-3),   metrics = list(     metric_sparse_categorical_accuracy(name = \"acc\")   ) )  model |> fit(   x_train, y_train,   batch_size = batch_size,   epochs = 1, # For speed   validation_split = 0.15 ) ## 399/399 - 6s - 15ms/step - acc: 0.7355 - loss: 0.7721 - val_acc: 0.9287 - val_loss: 0.2384"},{"path":"https://keras3.posit.co/dev/articles/intro_to_keras_for_engineers.html","id":"training-models-on-arbitrary-data-sources","dir":"Articles","previous_headings":"","what":"Training models on arbitrary data sources","title":"Introduction to Keras for engineers","text":"Keras models can trained evaluated wide variety data sources, independently backend ’re using. includes: Arrays Dataframes TensorFlow tf_dataset objects PyTorch DataLoader objects Keras PyDataset objects work whether ’re using TensorFlow, JAX, PyTorch Keras backend. Let’s try tf_dataset:","code":"library(tfdatasets, exclude = \"shape\")  train_dataset <- list(x_train, y_train) |>   tensor_slices_dataset() |>   dataset_batch(batch_size) |>   dataset_prefetch(buffer_size = tf$data$AUTOTUNE)  test_dataset <- list(x_test, y_test) |>   tensor_slices_dataset() |>   dataset_batch(batch_size) |>   dataset_prefetch(buffer_size = tf$data$AUTOTUNE)  model <- MyModel(num_classes = 10) model |> compile(   loss = loss_sparse_categorical_crossentropy(),   optimizer = optimizer_adam(learning_rate = 1e-3),   metrics = list(     metric_sparse_categorical_accuracy(name = \"acc\")   ) )  model |> fit(train_dataset, epochs = 1, validation_data = test_dataset) ## 469/469 - 7s - 15ms/step - acc: 0.7502 - loss: 0.7455 - val_acc: 0.9107 - val_loss: 0.2982"},{"path":"https://keras3.posit.co/dev/articles/intro_to_keras_for_engineers.html","id":"further-reading","dir":"Articles","previous_headings":"","what":"Further reading","title":"Introduction to Keras for engineers","text":"concludes short overview new multi-backend capabilities Keras 3. Next, can learn :","code":""},{"path":"https://keras3.posit.co/dev/articles/intro_to_keras_for_engineers.html","id":"how-to-customize-what-happens-in-fit","dir":"Articles","previous_headings":"Further reading","what":"How to customize what happens in fit()","title":"Introduction to Keras for engineers","text":"Want implement non-standard training algorithm still want benefit power usability fit()? ’s easy customize fit() support arbitrary use cases: Customizing happens fit() TensorFlow","code":""},{"path":"https://keras3.posit.co/dev/articles/intro_to_keras_for_engineers.html","id":"how-to-write-custom-training-loops","dir":"Articles","previous_headings":"","what":"How to write custom training loops","title":"Introduction to Keras for engineers","text":"Writing training loop scratch TensorFlow","code":""},{"path":"https://keras3.posit.co/dev/articles/intro_to_keras_for_engineers.html","id":"how-to-distribute-training","dir":"Articles","previous_headings":"","what":"How to distribute training","title":"Introduction to Keras for engineers","text":"Guide distributed training TensorFlow Enjoy library! 🚀","code":""},{"path":"https://keras3.posit.co/dev/articles/intro_to_keras_for_researchers.html","id":"setup","dir":"Articles","previous_headings":"","what":"Setup","title":"Introduction to Keras for Researchers","text":"","code":"library(keras3) library(tensorflow)"},{"path":"https://keras3.posit.co/dev/articles/intro_to_keras_for_researchers.html","id":"introduction","dir":"Articles","previous_headings":"","what":"Introduction","title":"Introduction to Keras for Researchers","text":"machine learning researcher? publish NeurIPS push state---art CV NLP? guide serve first introduction core Keras & TensorFlow API concepts. guide, learn : Tensors, variables, gradients TensorFlow Creating layers subclassing [Layer] class Writing low-level training loops Tracking losses created layers via add_loss() method Tracking metrics low-level training loop Speeding execution compiled [tensorflow::tf_function()] Executing layers training inference mode Keras Functional API also see Keras API action two end--end research examples: Variational Autoencoder, Hypernetwork.","code":""},{"path":"https://keras3.posit.co/dev/articles/intro_to_keras_for_researchers.html","id":"tensors","dir":"Articles","previous_headings":"","what":"Tensors","title":"Introduction to Keras for Researchers","text":"TensorFlow infrastructure layer differentiable programming. heart, ’s framework manipulating N-dimensional arrays (tensors), much like NumPy. However, three key differences NumPy TensorFlow: TensorFlow can leverage hardware accelerators GPUs TPUs. TensorFlow can automatically compute gradient arbitrary differentiable tensor expressions. TensorFlow computation can distributed large numbers devices single machine, large number machines (potentially multiple devices ). Let’s take look object core TensorFlow: Tensor. ’s constant tensor: can get value R array calling .array(): features attributes dtype shape: common way create constant tensors via tf$ones tf$zeros: can also create random constant tensors:","code":"x <- tf$constant(rbind(c(5, 2), c(1, 3))) print(x) ## tf.Tensor( ## [[5. 2.] ##  [1. 3.]], shape=(2, 2), dtype=float64) as.array(x) ##      [,1] [,2] ## [1,]    5    2 ## [2,]    1    3 x$dtype ## tf.float64 x$shape ## TensorShape([2, 2]) tf$ones(shape = shape(2, 1)) ## tf.Tensor( ## [[1.] ##  [1.]], shape=(2, 1), dtype=float32) tf$zeros(shape = shape(2, 1)) ## tf.Tensor( ## [[0.] ##  [0.]], shape=(2, 1), dtype=float32) x <- random_normal(shape = c(2, 2), mean = 0.0, stddev = 1.0) x <- random_uniform(shape = c(2, 2), minval = 0, maxval = 10)"},{"path":"https://keras3.posit.co/dev/articles/intro_to_keras_for_researchers.html","id":"variables","dir":"Articles","previous_headings":"","what":"Variables","title":"Introduction to Keras for Researchers","text":"Variables special tensors used store mutable state (weights neural network). create Variable using initial value: update value Variable using methods $assign(value), $assign_add(increment), $assign_sub(decrement):","code":"initial_value <- random_normal(shape=c(2, 2)) a <- tf$Variable(initial_value) print(a) ## <tf.Variable 'Variable:0' shape=(2, 2) dtype=float32, numpy= ## array([[ 0.9057419 ,  0.7916686 ], ##        [ 0.28754202, -0.5408822 ]], dtype=float32)> new_value <- random_normal(shape=c(2, 2)) a$assign(new_value) ## <tf.Variable 'UnreadVariable' shape=(2, 2) dtype=float32, numpy= ## array([[-0.3405368 , -2.1463926 ], ##        [ 1.2602988 ,  0.12241419]], dtype=float32)> added_value <- random_normal(shape=c(2, 2)) a$assign_add(added_value) ## <tf.Variable 'UnreadVariable' shape=(2, 2) dtype=float32, numpy= ## array([[ 0.04820395, -2.6854615 ], ##        [ 0.23246336,  1.4535258 ]], dtype=float32)>"},{"path":"https://keras3.posit.co/dev/articles/intro_to_keras_for_researchers.html","id":"doing-math-in-tensorflow","dir":"Articles","previous_headings":"","what":"Doing math in TensorFlow","title":"Introduction to Keras for Researchers","text":"’ve used NumPy, math TensorFlow look familiar. main difference TensorFlow code can run GPU TPU.","code":"a <- random_normal(shape=c(2, 2)) b <- random_normal(shape=c(2, 2))  c <- a + b d <- tf$square(c) e <- tf$exp(d)"},{"path":"https://keras3.posit.co/dev/articles/intro_to_keras_for_researchers.html","id":"gradients","dir":"Articles","previous_headings":"","what":"Gradients","title":"Introduction to Keras for Researchers","text":"’s another big difference R: can automatically retrieve gradient differentiable expression. Just open GradientTape, start “watching” tensor via tape$watch(), compose differentiable expression using tensor input: default, variables watched automatically, don’t need manually watch : Note can compute higher-order derivatives nesting tapes:","code":"a <- random_normal(shape=c(2, 2)) b <- random_normal(shape=c(2, 2))  with(tf$GradientTape() %as% tape, {   tape$watch(a)  # Start recording the history of operations applied to `a`   c <- tf$sqrt(tf$square(a) + tf$square(b))  # Do some math using `a`   # What's the gradient of `c` with respect to `a`?   dc_da <- tape$gradient(c, a)   print(dc_da) }) ## tf.Tensor( ## [[ 0.9969011  -0.7707146 ] ##  [ 0.23378514  0.96255165]], shape=(2, 2), dtype=float32) a <- tf$Variable(a)  with(tf$GradientTape() %as% tape, {   c <- tf$sqrt(tf$square(a) + tf$square(b))   dc_da <- tape$gradient(c, a)   print(dc_da) }) ## tf.Tensor( ## [[ 0.9969011  -0.7707146 ] ##  [ 0.23378514  0.96255165]], shape=(2, 2), dtype=float32) with(tf$GradientTape() %as% outer_tape, {   with(tf$GradientTape() %as% tape, {     c <- tf$sqrt(tf$square(a) + tf$square(b))     dc_da <- tape$gradient(c, a)   })   d2c_da2 <- outer_tape$gradient(dc_da, a)   print(d2c_da2) }) ## tf.Tensor( ## [[3.3447742e-03 7.1282005e-01] ##  [5.7464113e+00 5.5013180e-02]], shape=(2, 2), dtype=float32)"},{"path":"https://keras3.posit.co/dev/articles/intro_to_keras_for_researchers.html","id":"keras-layers","dir":"Articles","previous_headings":"","what":"Keras layers","title":"Introduction to Keras for Researchers","text":"TensorFlow infrastructure layer differentiable programming, dealing tensors, variables, gradients, Keras user interface deep learning, dealing layers, models, optimizers, loss functions, metrics, . Keras serves high-level API TensorFlow: Keras makes TensorFlow simple productive. Layer class fundamental abstraction Keras. Layer encapsulates state (weights) computation (defined call method). simple layer looks like . self$add_weight() method gives shortcut creating weights: use Layer instance much like R function: weight variables (created initialize) automatically tracked weights property: many built-layers available, Dense Conv2D LSTM fancier ones like Conv3DTranspose ConvLSTM2D. smart reusing built-functionality.","code":"Linear <- new_layer_class(   \"Linear\",   initialize = function(units = 32, input_dim = 32) {     super$initialize()     self$w <- self$add_weight(       shape = shape(input_dim, units),       initializer = \"random_normal\",       trainable = TRUE     )     self$b <- self$add_weight(       shape = shape(units),       initializer = \"zeros\",       trainable = TRUE     )   },   call = function(inputs) {     tf$matmul(inputs, self$w) + self$b   } ) # Instantiate our layer. linear_layer <- Linear(units=4, input_dim=2)  # The layer can be treated as a function. # Here we call it on some data. y <- linear_layer(tf$ones(shape(2, 2))) linear_layer$weights ## [[1]] ## <KerasVariable shape=(2, 4), dtype=float32, path=linear/variable> ## ## [[2]] ## <KerasVariable shape=(4), dtype=float32, path=linear/variable_1>"},{"path":"https://keras3.posit.co/dev/articles/intro_to_keras_for_researchers.html","id":"layer-weight-creation-in-buildinput_shape","dir":"Articles","previous_headings":"","what":"Layer weight creation in build(input_shape)","title":"Introduction to Keras for Researchers","text":"’s often good idea defer weight creation build() method, don’t need specify input dim/shape layer construction time:","code":"Linear <- new_layer_class(   \"Linear\",   initialize = function(units = 32) {     super$initialize()     self$units <- units   },   build = function(input_shape) {     self$w <- self$add_weight(       shape = shape(input_shape[-1], self$units),       initializer = \"random_normal\",       trainable = TRUE     )     self$b <- self$add_weight(       shape = shape(self$units),       initializer = \"zeros\",       trainable = TRUE     )   },   call = function(inputs) {     tf$matmul(inputs, self$w) + self$b   } )  # Instantiate our layer. linear_layer <- Linear(units = 4)  # This will also call `build(input_shape)` and create the weights. y <- linear_layer(tf$ones(shape(2, 2)))"},{"path":"https://keras3.posit.co/dev/articles/intro_to_keras_for_researchers.html","id":"layer-gradients","dir":"Articles","previous_headings":"","what":"Layer gradients","title":"Introduction to Keras for Researchers","text":"can automatically retrieve gradients weights layer calling inside GradientTape. Using gradients, can update weights layer, either manually, using optimizer object. course, can modify gradients using , need .","code":"# Prepare a dataset. c(c(x_train, y_train), .) %<-% dataset_mnist()  x_train <- array_reshape(x_train, c(60000, 784)) / 255  dataset <- tfdatasets::tensor_slices_dataset(list(x_train, y_train)) %>%   tfdatasets::dataset_shuffle(buffer_size=1024) %>%   tfdatasets::dataset_batch(64)  # Instantiate our linear layer (defined above) with 10 units. linear_layer <- Linear(units = 10)  # Instantiate a logistic loss function that expects integer targets. loss_fn <- loss_sparse_categorical_crossentropy(from_logits=TRUE)  # Instantiate an optimizer. optimizer <- optimizer_sgd(learning_rate=1e-3)  # Iterate over the batches of the dataset. coro::loop(for(data in dataset) {   # Open a GradientTape.   with(tf$GradientTape() %as% tape, {     # Forward pass.     logits <- linear_layer(data[[1]])      # Loss value for this batch.     loss_value <- loss_fn(data[[2]], logits)   })    # Get gradients of the loss wrt the weights.   gradients <- tape$gradient(loss_value, linear_layer$trainable_weights)    # Update the weights of our linear layer.   optimizer$apply_gradients(zip_lists(gradients, linear_layer$trainable_weights)) }) loss_value ## tf.Tensor(1.2819729, shape=(), dtype=float32)"},{"path":"https://keras3.posit.co/dev/articles/intro_to_keras_for_researchers.html","id":"trainable-and-non-trainable-weights","dir":"Articles","previous_headings":"","what":"Trainable and non-trainable weights","title":"Introduction to Keras for Researchers","text":"Weights created layers can either trainable non-trainable. ’re exposed trainable_weights non_trainable_weights respectively. ’s layer non-trainable weight:","code":"ComputeSum <- new_layer_class(   \"ComputeSum\",   initialize = function(input_dim) {     super$initialize()     # Create a non-trainable weight.     self$total <- self$add_weight(       initializer = \"zeros\",       shape = shape(input_dim),       trainable = FALSE     )   },   call = function(inputs) {     self$total$assign_add(tf$reduce_sum(inputs, axis=0L))     self$total   } )  my_sum <- ComputeSum(input_dim = 2) x <- tf$ones(shape(2, 2))  as.array(my_sum(x)) ## [1] 2 2 as.array(my_sum(x)) ## [1] 4 4 my_sum$trainable_weights ## list()"},{"path":"https://keras3.posit.co/dev/articles/intro_to_keras_for_researchers.html","id":"layers-that-own-layers","dir":"Articles","previous_headings":"","what":"Layers that own layers","title":"Introduction to Keras for Researchers","text":"Layers can recursively nested create bigger computation blocks. layer track weights sublayers (trainable non-trainable). Note manually-created MLP equivalent following built-option:","code":"# Let's reuse the Linear class # with a `build` method that we defined above.  MLP <- new_layer_class(   \"MLP\",   initialize = function() {     super$initialize()     self$linear_1 <- Linear(units = 32)     self$linear_2 <- Linear(units = 32)     self$linear_3 <- Linear(units = 10)   },   call = function(inputs) {     x <- self$linear_1(inputs)     x <- tf$nn$relu(x)     x <- self$linear_2(x)     x <- tf$nn$relu(x)     return(self$linear_3(x))   } )  mlp <- MLP()  # The first call to the `mlp` object will create the weights. y <- mlp(tf$ones(shape=shape(3, 64)))  # Weights are recursively tracked. length(mlp$weights) ## [1] 6 mlp <- keras_model_sequential() %>%   layer_dense(units = 32, activation = \"relu\") %>%   layer_dense(units = 32, activation = \"relu\") %>%   layer_dense(units = 10)"},{"path":"https://keras3.posit.co/dev/articles/intro_to_keras_for_researchers.html","id":"tracking-losses-created-by-layers","dir":"Articles","previous_headings":"","what":"Tracking losses created by layers","title":"Introduction to Keras for Researchers","text":"Layers can create losses forward pass via add_loss() method. especially useful regularization losses. losses created sublayers recursively tracked parent layers. ’s layer creates activity regularization loss: model incorporating layer track regularization loss: losses cleared top-level layer start forward pass – don’t accumulate. layer.losses always contains losses created last forward pass. typically use losses summing computing gradients writing training loop.","code":"# A layer that creates an activity sparsity regularization loss ActivityRegularization <- new_layer_class(   \"ActivityRegularization\",   initialize = function(rate=1e-2) {     super$initialize()     self$rate <- rate   },   call = function(inputs) {     self$add_loss(self$rate * tf$reduce_sum(tf$abs(inputs)))     inputs   } ) # Let's use the loss layer in a MLP block. SparseMLP <- new_layer_class(   \"SparseMLP\",   initialize = function() {     super$initialize()     self$linear_1 <- Linear(units = 32)     self$reg <- ActivityRegularization(rate = 1e-2)     self$linear_3 <- Linear(units = 10)   },   call = function(inputs) {     x <- self$linear_1(inputs)     x <- tf$nn$relu(x)     x <- self$reg(x)     return(self$linear_3(x))   } )  mlp <- SparseMLP() y <- mlp(tf$ones(shape(10, 10)))  mlp$losses  # List containing one float32 scalar ## [[1]] ## tf.Tensor(0.18065463, shape=(), dtype=float32) # Losses correspond to the *last* forward pass. mlp <- SparseMLP() mlp(tf$ones(shape(10, 10))) ## tf.Tensor( ## [[ 0.0388482  -0.03920118  0.01624808 -0.01361975 -0.01354899  0.07107338 ##   -0.01077365  0.05688906 -0.02838149 -0.04084621] ##  [ 0.0388482  -0.03920118  0.01624808 -0.01361975 -0.01354899  0.07107338 ##   -0.01077365  0.05688906 -0.02838149 -0.04084621] ##  [ 0.0388482  -0.03920118  0.01624808 -0.01361975 -0.01354899  0.07107338 ##   -0.01077365  0.05688906 -0.02838149 -0.04084621] ##  [ 0.0388482  -0.03920118  0.01624808 -0.01361975 -0.01354899  0.07107338 ##   -0.01077365  0.05688906 -0.02838149 -0.04084621] ##  [ 0.0388482  -0.03920118  0.01624808 -0.01361975 -0.01354899  0.07107338 ##   -0.01077365  0.05688906 -0.02838149 -0.04084621] ##  [ 0.0388482  -0.03920118  0.01624808 -0.01361975 -0.01354899  0.07107338 ##   -0.01077365  0.05688906 -0.02838149 -0.04084621] ##  [ 0.0388482  -0.03920118  0.01624808 -0.01361975 -0.01354899  0.07107338 ##   -0.01077365  0.05688906 -0.02838149 -0.04084621] ##  [ 0.0388482  -0.03920118  0.01624808 -0.01361975 -0.01354899  0.07107338 ##   -0.01077365  0.05688906 -0.02838149 -0.04084621] ##  [ 0.0388482  -0.03920118  0.01624808 -0.01361975 -0.01354899  0.07107338 ##   -0.01077365  0.05688906 -0.02838149 -0.04084621] ##  [ 0.0388482  -0.03920118  0.01624808 -0.01361975 -0.01354899  0.07107338 ##   -0.01077365  0.05688906 -0.02838149 -0.04084621]], shape=(10, 10), dtype=float32) length(mlp$losses) ## [1] 1 mlp(tf$ones(shape(10, 10))) ## tf.Tensor( ## [[ 0.0388482  -0.03920118  0.01624808 -0.01361975 -0.01354899  0.07107338 ##   -0.01077365  0.05688906 -0.02838149 -0.04084621] ##  [ 0.0388482  -0.03920118  0.01624808 -0.01361975 -0.01354899  0.07107338 ##   -0.01077365  0.05688906 -0.02838149 -0.04084621] ##  [ 0.0388482  -0.03920118  0.01624808 -0.01361975 -0.01354899  0.07107338 ##   -0.01077365  0.05688906 -0.02838149 -0.04084621] ##  [ 0.0388482  -0.03920118  0.01624808 -0.01361975 -0.01354899  0.07107338 ##   -0.01077365  0.05688906 -0.02838149 -0.04084621] ##  [ 0.0388482  -0.03920118  0.01624808 -0.01361975 -0.01354899  0.07107338 ##   -0.01077365  0.05688906 -0.02838149 -0.04084621] ##  [ 0.0388482  -0.03920118  0.01624808 -0.01361975 -0.01354899  0.07107338 ##   -0.01077365  0.05688906 -0.02838149 -0.04084621] ##  [ 0.0388482  -0.03920118  0.01624808 -0.01361975 -0.01354899  0.07107338 ##   -0.01077365  0.05688906 -0.02838149 -0.04084621] ##  [ 0.0388482  -0.03920118  0.01624808 -0.01361975 -0.01354899  0.07107338 ##   -0.01077365  0.05688906 -0.02838149 -0.04084621] ##  [ 0.0388482  -0.03920118  0.01624808 -0.01361975 -0.01354899  0.07107338 ##   -0.01077365  0.05688906 -0.02838149 -0.04084621] ##  [ 0.0388482  -0.03920118  0.01624808 -0.01361975 -0.01354899  0.07107338 ##   -0.01077365  0.05688906 -0.02838149 -0.04084621]], shape=(10, 10), dtype=float32) length(mlp$losses)  # No accumulation. ## [1] 1 # Let's demonstrate how to use these losses in a training loop.  # Prepare a dataset. c(c(x_train, y_train), .) %<-% dataset_mnist() x_train <- array_reshape(x_train, c(60000, 784)) / 255  dataset <- tfdatasets::tensor_slices_dataset(list(x_train, y_train)) %>%   tfdatasets::dataset_shuffle(buffer_size=1024) %>%   tfdatasets::dataset_batch(64)  # A new MLP. mlp <- SparseMLP()  # Loss and optimizer. loss_fn <- loss_sparse_categorical_crossentropy(from_logits=TRUE) optimizer <- optimizer_sgd(learning_rate=1e-3)  coro::loop(for(data in dataset) {   x <- data[[1]]   y <- data[[2]]   with(tf$GradientTape() %as% tape, {     # Forward pass.     logits <- mlp(x)      # External loss value for this batch.     loss <- loss_fn(y, logits)      # Add the losses created during the forward pass.     loss <- loss + Reduce(`+`, mlp$losses)      # Get gradients of the loss wrt the weights.     gradients <- tape$gradient(loss, mlp$trainable_weights)      # Update the weights of our linear layer.     optimizer$apply_gradients(zip_lists(gradients, mlp$trainable_weights))   }) })"},{"path":"https://keras3.posit.co/dev/articles/intro_to_keras_for_researchers.html","id":"keeping-track-of-training-metrics","dir":"Articles","previous_headings":"","what":"Keeping track of training metrics","title":"Introduction to Keras for Researchers","text":"Keras offers broad range built-metrics, like metric_auc metric_precision_at_recall. ’s also easy create metrics lines code. use metric custom training loop, : Instantiate metric object, e.g. metric = metric_auc() Call metric$udpate_state(targets, predictions) method batch data Query result via metric$result() Reset metric’s state end epoch start evaluation via metric$reset_state() ’s simple example: can also define metrics subclassing keras.metrics.Metric. need override three functions called : Override update_state() update statistic values. Override result() return metric value. Override reset_state() reset metric initial state. example implement F1-score metric (support sample weighting). Let’s test-drive :","code":"# Instantiate a metric object accuracy <- metric_sparse_categorical_accuracy()  # Prepare our layer, loss, and optimizer. model <- keras_model_sequential() %>%   layer_dense(units = 32, activation = \"relu\") %>%   layer_dense(units = 32, activation = \"relu\") %>%   layer_dense(units = 10) loss_fn <- loss_sparse_categorical_crossentropy(from_logits = TRUE) optimizer <- optimizer_adam(learning_rate=1e-3)  for (epoch in seq_len(2)) {   coro::loop(for (data in dataset) {     x <- data[[1]]     y <- data[[2]]     with(tf$GradientTape() %as% tape, {       # Forward pass.       logits <- model(x)        # External loss value for this batch.       loss_value <- loss_fn(y, logits)     })      # Update the state of the `accuracy` metric.     accuracy$update_state(y, logits)      # Update the weights of the model to minimize the loss value.     gradients <- tape$gradient(loss_value, model$trainable_weights)     optimizer$apply_gradients(zip_lists(gradients, model$trainable_weights))    })   cat(\"Epoch:\", epoch, \"Accuracy:\", as.numeric(accuracy$result()), \"\\n\")   accuracy$reset_state() } ## Epoch: 1 Accuracy: 0.8757833 ## Epoch: 2 Accuracy: 0.93915 F1Score <- new_metric_class(   \"F1Score\",   initialize = function(self, name=\"f1_score\", dtype=\"float32\", threshold=0.5, ...) {     super$initialize(name=name, dtype=dtype, ...)     self$threshold <- threshold     self$true_positives <- self$add_weight(       name=\"tp\", dtype=dtype, initializer=\"zeros\"     )     self$false_positives <- self$add_weight(       name=\"fp\", dtype=dtype, initializer=\"zeros\"     )     self$false_negatives <- self$add_weight(       name=\"fn\", dtype=dtype, initializer=\"zeros\"     )   },   update_state = function(y_true, y_pred, sample_weight=NULL) {     y_pred <- tf$math$greater_equal(y_pred, self$threshold)     y_true <- tf$cast(y_true, tf$bool)     y_pred <- tf$cast(y_pred, tf$bool)      true_positives <- tf$cast(y_true & y_pred, self$dtype)     false_positives <- tf$cast((!y_true) & y_pred, self$dtype)     false_negatives <- tf$cast(y_true & (!y_pred), self$dtype)      if (!is.null(sample_weight)) {       sample_weight <- tf$cast(sample_weight, self$dtype)       true_positives <- true_positives * sample_weight       false_positives <- false_positives * sample_weight       false_negatives <- false_negatives * sample_weight     }      self$true_positives$assign_add(tf$reduce_sum(true_positives))     self$false_positives$assign_add(tf$reduce_sum(false_positives))     self$false_negatives$assign_add(tf$reduce_sum(false_negatives))   },    result = function() {     precision <- self$true_positives / (self$true_positives + self$false_positives)     recall <- self$true_positives / (self$true_positives + self$false_negatives)     f1_score <- 2 * precision * recall / (precision + recall)     f1_score   },    reset_state = function() {     self$true_positives$assign(0)     self$false_positives$assign(0)     self$false_negatives$assign(0)   } ) m <- F1Score() m$update_state(c(0, 1, 0, 0), c(0.3, 0.5, 0.8, 0.9)) cat(\"Intermediate result:\", as.numeric(m$result()), \"\\n\") ## Intermediate result: 0.5 m$update_state(c(1, 1, 1, 1), c(0.1, 0.7, 0.6, 0.0)) cat(\"Final result:\", as.numeric(m$result()), \"\\n\") ## Final result: 0.6"},{"path":"https://keras3.posit.co/dev/articles/intro_to_keras_for_researchers.html","id":"compiled-functions","dir":"Articles","previous_headings":"","what":"Compiled functions","title":"Introduction to Keras for Researchers","text":"Running eagerly great debugging, get better performance compiling computation static graphs. Static graphs researcher’s best friends. can compile function wrapping tf.function decorator.","code":"# Prepare our layer, loss, and optimizer. model <- keras_model_sequential() %>%   layer_dense(units = 32, activation = \"relu\") %>%   layer_dense(units = 32, activation = \"relu\") %>%   layer_dense(units = 10) loss_fn <- loss_sparse_categorical_crossentropy(from_logits = TRUE) optimizer <- optimizer_adam(learning_rate=1e-3)  # Create a training step function. train_on_batch <- tf_function(function(x, y) {   with(tf$GradientTape() %as% tape, {     # Forward pass.     logits <- model(x)     # External loss value for this batch.     loss_value <- loss_fn(y, logits)   })   # Update the weights of the model to minimize the loss value.   gradients <- tape$gradient(loss_value, model$trainable_weights)   optimizer$apply_gradients(zip_lists(gradients, model$trainable_weights))   loss_value })   # Prepare a dataset. c(c(x_train, y_train), .) %<-% dataset_mnist() x_train <- array_reshape(x_train, c(60000, 784)) / 255  dataset <- tfdatasets::tensor_slices_dataset(list(x_train, y_train)) %>%   tfdatasets::dataset_shuffle(buffer_size=1024) %>%   tfdatasets::dataset_batch(64)  i <- 0 coro::loop(for (data in dataset) {   i <- i + 1   x <- data[[1]]   y <- data[[2]]   loss <- train_on_batch(x, y)   if (i %% 100 == 0)     cat(\"Loss:\", as.numeric(loss), \"\\n\") }) ## Loss: 0.551749 ## Loss: 0.2131135 ## Loss: 0.2765952 ## Loss: 0.1296219 ## Loss: 0.2657076 ## Loss: 0.2683381 ## Loss: 0.1570166 ## Loss: 0.3139241 ## Loss: 0.08981849"},{"path":"https://keras3.posit.co/dev/articles/intro_to_keras_for_researchers.html","id":"training-mode-inference-mode","dir":"Articles","previous_headings":"","what":"Training mode & inference mode","title":"Introduction to Keras for Researchers","text":"layers, particular BatchNormalization layer Dropout layer, different behaviors training inference. layers, standard practice expose training (boolean) argument call method. exposing argument call, enable built-training evaluation loops (e.g. fit) correctly use layer training inference modes.","code":"Dropout <- new_layer_class(   \"Dropout\",   initialize = function(rate) {     super$initialize()     self$rate <- rate   },   call = function(inputs, training = NULL) {     if (!is.null(training) && training) {       return(tf$nn$dropout(inputs, rate = self$rate))     }     inputs   } )  MLPWithDropout <- new_layer_class(   \"MLPWithDropout\",   initialize = function() {     super$initialize()     self$linear_1 <- Linear(units = 32)     self$dropout <- Dropout(rate = 0.5)     self$linear_3 <- Linear(units = 10)   },   call = function(inputs, training = NULL) {     x <- self$linear_1(inputs)     x <- tf$nn$relu(x)     x <- self$dropout(x, training = training)     self$linear_3(x)   } )  mlp <- MLPWithDropout() y_train <- mlp(tf$ones(shape(2, 2)), training=TRUE) y_test <- mlp(tf$ones(shape(2, 2)), training=FALSE)"},{"path":"https://keras3.posit.co/dev/articles/intro_to_keras_for_researchers.html","id":"the-functional-api-for-model-building","dir":"Articles","previous_headings":"","what":"The Functional API for model-building","title":"Introduction to Keras for Researchers","text":"build deep learning models, don’t use object-oriented programming time. layers ’ve seen far can also composed functionally, like (call “Functional API”): Functional API tends concise subclassing, provides advantages (generally advantages functional, typed languages provide untyped OO development). However, can used define DAGs layers – recursive networks defined Layer subclasses instead. Learn Functional API . research workflows, may often find mix--matching OO models Functional models. Note Model class also features built-training & evaluation loops: fit(), predict() evaluate() (configured via compile() method). built-functions give access following built-training infrastructure features: Callbacks. can leverage built-callbacks early-stopping, model checkpointing, monitoring training TensorBoard. can also implement custom callbacks needed. Distributed training. can easily scale training multiple GPUs, TPU, even multiple machines tf.distribute API – changes code. Step fusing. steps_per_execution argument Model.compile(), can process multiple batches single tf.function call, greatly improves device utilization TPUs. won’t go details, provide simple code example . leverages built-training infrastructure implement MNIST example . can always subclass Model class (works exactly like subclassing Layer) want leverage built-training loops OO models. Just override Model$train_step() customize happens fit() retaining support built-infrastructure features outlined – callbacks, zero-code distribution support, step fusing support. may also override test_step() customize happens evaluate(), override predict_step() customize happens predict(). information, please refer guide.","code":"# We use an `Input` object to describe the shape and dtype of the inputs. # This is the deep learning equivalent of *declaring a type*. # The shape argument is per-sample; it does not include the batch size. # The functional API focused on defining per-sample transformations. # The model we create will automatically batch the per-sample transformations, # so that it can be called on batches of data. inputs <- layer_input(shape = 16, dtype = \"float32\")  # We call layers on these \"type\" objects # and they return updated types (new shapes/dtypes). outputs <- inputs %>%   Linear(units = 32) %>% # We are reusing the Linear layer we defined earlier.   Dropout(rate = 0.5) %>% # We are reusing the Dropout layer we defined earlier.   Linear(units = 10)  # A functional `Model` can be defined by specifying inputs and outputs. # A model is itself a layer like any other. model <- keras_model(inputs, outputs)  # A functional model already has weights, before being called on any data. # That's because we defined its input shape in advance (in `Input`). length(model$weights) ## [1] 4 # Let's call our model on some data, for fun. y <- model(tf$ones(shape(2, 16))) y$shape ## TensorShape([2, 10]) # You can pass a `training` argument in `__call__` # (it will get passed down to the Dropout layer). y <- model(tf$ones(shape(2, 16)), training=TRUE) inputs <- layer_input(shape = 784, dtype=\"float32\") outputs <- inputs %>%   layer_dense(units = 32, activation = \"relu\") %>%   layer_dense(units = 32, activation = \"relu\") %>%   layer_dense(units = 10) model <- keras_model(inputs, outputs)  # Specify the loss, optimizer, and metrics with `compile()`. model %>% compile(     loss = loss_sparse_categorical_crossentropy(from_logits=TRUE),     optimizer=optimizer_adam(learning_rate=1e-3),     metrics=list(metric_sparse_categorical_accuracy()), )  # Train the model with the dataset for 2 epochs. model %>% fit(dataset, epochs=2) ## Epoch 1/2 ## 938/938 - 4s - 4ms/step - loss: 0.3958 - sparse_categorical_accuracy: 0.8866 ## Epoch 2/2 ## 938/938 - 1s - 960us/step - loss: 0.1888 - sparse_categorical_accuracy: 0.9443 predictions <- model %>% predict(dataset) ## 938/938 - 1s - 1ms/step model %>% evaluate(dataset) ## 938/938 - 1s - 1ms/step - loss: 0.1763 - sparse_categorical_accuracy: 0.9454 ## $loss ## [1] 0.1763445 ## ## $sparse_categorical_accuracy ## [1] 0.9454167 CustomModel <- new_model_class(   \"CustomModel\",   initialize = function(...) {     super$initialize(...)     self$loss_tracker <- metric_mean(name=\"loss\")     self$accuracy <- metric_sparse_categorical_accuracy()     self$loss_fn <- loss_sparse_categorical_crossentropy(from_logits=TRUE)     self$optimizer <- optimizer_adam(learning_rate=1e-3)   },   train_step = function(data) {     c(x, y = NULL, sample_weight = NULL) %<-% data     with(tf$GradientTape() %as% tape, {       y_pred <- self(x, training=TRUE)       loss <- self$loss_fn(y = y, y_pred = y_pred, sample_weight=sample_weight)     })     gradients <- tape$gradient(loss, self$trainable_variables)     self$optimizer$apply_gradients(       zip_lists(gradients, self$trainable_variables)     )      # Update metrics (includes the metric that tracks the loss)     self$loss_tracker$update_state(loss)     self$accuracy$update_state(y, y_pred, sample_weight=sample_weight)     # Return a list mapping metric names to current value     list(       loss = self$loss_tracker$result(),       accuracy = self$accuracy$result()     )   },   metrics = mark_active(function() {     list(self$loss_tracker, self$accuracy)   }) )  inputs <- layer_input(shape = 784, dtype=\"float32\") outputs <- inputs %>%   layer_dense(units = 32, activation = \"relu\") %>%   layer_dense(units = 32, activation = \"relu\") %>%   layer_dense(units = 10) model <- CustomModel(inputs, outputs) model %>% compile() model %>% fit(dataset, epochs=2) ## Epoch 1/2 ## 938/938 - 2s - 2ms/step - loss: 0.3869 - sparse_categorical_accuracy: 0.8924 ## Epoch 2/2 ## 938/938 - 1s - 1ms/step - loss: 0.2163 - sparse_categorical_accuracy: 0.9370"},{"path":"https://keras3.posit.co/dev/articles/intro_to_keras_for_researchers.html","id":"end-to-end-experiment-example-1-variational-autoencoders-","dir":"Articles","previous_headings":"","what":"End-to-end experiment example 1: variational autoencoders.","title":"Introduction to Keras for Researchers","text":"things ’ve learned far: Layer encapsulates state (created __init__ build) computation (defined call). Layers can recursively nested create new, bigger computation blocks. can easily write highly hackable training loops opening GradientTape, calling model inside tape’s scope, retrieving gradients applying via optimizer. can speed training loops using @tf.function decorator. Layers can create track losses (typically regularization losses) via self.add_loss(). Let’s put things together end--end example: ’re going implement Variational AutoEncoder (VAE). ’ll train MNIST digits. VAE subclass Layer, built nested composition layers subclass Layer. feature regularization loss (KL divergence). model definition. First, Encoder class, uses Sampling layer map MNIST digit latent-space triplet (z_mean, z_log_var, z). Next, Decoder class, maps probabilistic latent space coordinates back MNIST digit. Finally, VariationalAutoEncoder composes together encoder decoder, creates KL divergence regularization loss via add_loss(). Now, let’s write training loop. training step decorated @tf.function compile super fast graph function. can see, building training type model Keras quick painless.","code":"Sampling <- new_layer_class(   \"Sampling\",   call = function(inputs) {     c(z_mean, z_log_var) %<-% inputs     batch <- op_shape(z_mean)[[1]]     dim <- op_shape(z_mean)[[2]]     epsilon <- random_normal(shape = c(batch, dim))     z_mean + op_exp(0.5 * z_log_var) * epsilon   } )  Encoder <- new_layer_class(   \"Encoder\",   initialize = function(latent_dim = 32, intermediate_dim = 64, ...) {     super$initialize(...)     self$dense_proj <- layer_dense(units = intermediate_dim, activation = \"relu\")     self$dense_mean <- layer_dense(units = latent_dim)     self$dense_log_var <- layer_dense(units = latent_dim)     self$sampling <- Sampling()   },   call = function(inputs) {     x <- self$dense_proj(inputs)     z_mean <- self$dense_mean(x)     z_log_var <- self$dense_log_var(x)     z <- self$sampling(list(z_mean, z_log_var))     list(z_mean, z_log_var, z)   } ) Decoder <- new_layer_class(   \"Decoder\",   initialize = function(original_dim, intermediate_dim = 64, ...) {     super$initialize(...)     self$dense_proj <- layer_dense(units = intermediate_dim, activation = \"relu\")     self$dense_output <- layer_dense(units = original_dim, activation = \"sigmoid\")   },   call = function(inputs) {     x <- self$dense_proj(inputs)     self$dense_output(x)   } ) VariationalAutoEncoder <- new_model_class(   \"VariationalAutoEncoder\",   initialize = function(original_dim,         intermediate_dim=64,         latent_dim=32,         name=\"autoencoder\", ...) {     super$initialize(name = name, ...)     self$original_dim <- original_dim     self$encoder <- Encoder(       latent_dim = latent_dim,       intermediate_dim = intermediate_dim     )     self$decoder <- Decoder(       original_dim = original_dim,       intermediate_dim = intermediate_dim     )   },   call = function(inputs) {     c(z_mean, z_log_var, z) %<-% self$encoder(inputs)     reconstructed <- self$decoder(z)     # Add KL divergence regularization loss.     kl_loss <- -0.5 * op_mean(       z_log_var - op_square(z_mean) - op_exp(z_log_var) + 1     )     self$add_loss(kl_loss)     reconstructed   } ) # Our model. vae <- VariationalAutoEncoder(   original_dim = 784,   intermediate_dim = 64,   latent_dim = 32 )  # Loss and optimizer. loss_fn <- loss_mean_squared_error() optimizer = optimizer_adam(learning_rate=1e-3)  # Prepare a dataset. c(c(x_train, .), .) %<-% dataset_mnist() x_train <- array_reshape(x_train, c(60000, 784)) / 255  dataset <- tfdatasets::tensor_slices_dataset(x_train) %>%   tfdatasets::dataset_shuffle(buffer_size=1024) %>%   tfdatasets::dataset_batch(32)   training_step <- tf_function(function(x) {   with(tf$GradientTape() %as% tape, {     reconstructed <- vae(x)  # Compute input reconstruction.     # Compute loss.     loss <- loss_fn(x, reconstructed)     loss <- loss + op_sum(vae$losses)  # Add KLD term.   })   # Update the weights of the VAE.   grads <- tape$gradient(loss, vae$trainable_weights)   optimizer$apply_gradients(zip_lists(grads, vae$trainable_weights))   loss })  losses <- c()  # Keep track of the losses over time. coro::loop(for(data in dataset) {   loss <- training_step(data)    # Logging.   losses[length(losses) + 1] <- as.numeric(loss)   if (length(losses) %% 100 == 0) {     cat(\"Step:\", length(losses), \"Loss:\", mean(losses), \"\\n\")   }   # Stop after 1000 steps.   # Training the model to convergence is left   # as an exercise to the reader.   if (length(losses) >= 1000) {     break   } }) ## Step: 100 Loss: 0.1270978 ## Step: 200 Loss: 0.1003238 ## Step: 300 Loss: 0.09001128 ## Step: 400 Loss: 0.08493649 ## Step: 500 Loss: 0.08171404 ## Step: 600 Loss: 0.07926706 ## Step: 700 Loss: 0.07790599 ## Step: 800 Loss: 0.07670419 ## Step: 900 Loss: 0.07570736 ## Step: 1000 Loss: 0.07476593"},{"path":"https://keras3.posit.co/dev/articles/intro_to_keras_for_researchers.html","id":"end-to-end-experiment-example-2-hypernetworks-","dir":"Articles","previous_headings":"","what":"End-to-end experiment example 2: hypernetworks.","title":"Introduction to Keras for Researchers","text":"Let’s take look another kind research experiment: hypernetworks. idea use small deep neural network (hypernetwork) generate weights larger network (main network). Let’s implement really trivial hypernetwork: ’ll use small 2-layer network generate weights larger 3-layer network. training loop. batch data: use hypernetwork generate array weight coefficients, weights_pred reshape coefficients kernel & bias tensors main_network run forward pass main_network compute actual MNIST predictions run backprop weights hypernetwork minimize final classification loss Implementing arbitrary research ideas Keras straightforward highly productive. Imagine trying 25 ideas per day (20 minutes per experiment average)! Keras designed go idea results fast possible, believe key great research. hope enjoyed quick introduction. Let us know build Keras!","code":"input_dim <- 784 classes <- 10  # This is the main network we'll actually use to predict labels. inputs <- layer_input(shape = input_dim) dense1 <- layer_dense(units = 64, activation = \"relu\") dense1$built <- TRUE  dense2 <- layer_dense(units = classes) dense2$built <- TRUE  outputs <- inputs %>% dense1() %>% dense2() main_network <- keras_model(inputs, outputs)  # This is the number of weight coefficients to generate. Each layer in the # main network requires output_dim * input_dim + output_dim coefficients. num_weights_to_generate <- (classes * 64 + classes) + (64 * input_dim + 64)  # This is the hypernetwork that generates the weights of the `main_network` above. hypernetwork <- keras_model_sequential() %>%   layer_dense(units=16, activation=\"relu\") %>%   layer_dense(units=num_weights_to_generate, activation=\"sigmoid\") # Loss and optimizer. loss_fn <- loss_sparse_categorical_crossentropy(from_logits = TRUE) optimizer <- optimizer_adam(learning_rate=1e-4)  # Prepare a dataset. c(c(x_train, y_train), .) %<-% dataset_mnist() x_train <- array_reshape(x_train, c(60000, 784)) / 255  dataset <- tfdatasets::tensor_slices_dataset(list(x_train, y_train)) %>%   tfdatasets::dataset_shuffle(buffer_size=1024) %>%   # We'll use a batch size of 1 for this experiment.   tfdatasets::dataset_batch(1)  train_step <- function(x, y) {   with(tf$GradientTape() %as% tape, {     weights_pred <- hypernetwork(x)      # Reshape them to the expected shapes for w and b for the outer model.     # Layer 1 kernel.     start_index <- 1     w1_shape <- c(input_dim, 64)     w1_coeffs <- weights_pred[, start_index:(start_index + prod(w1_shape) - 1)]     w1 <- tf$reshape(w1_coeffs, as.integer(w1_shape))     start_index <- start_index + prod(w1_shape)      # Layer 1 bias.     b1_shape <- c(64)     b1_coeffs <- weights_pred[, start_index:(start_index + prod(b1_shape) - 1)]     b1 <- tf$reshape(b1_coeffs, as.integer(b1_shape))     start_index <- start_index + prod(b1_shape)      # Layer 2 kernel.     w2_shape <- c(64, classes)     w2_coeffs <- weights_pred[, start_index:(start_index + prod(w2_shape) - 1)]     w2 <- tf$reshape(w2_coeffs, as.integer(w2_shape))     start_index <- start_index + prod(w2_shape)      # Layer 2 bias.     b2_shape <- c(classes)     b2_coeffs <- weights_pred[, start_index:(start_index + prod(b2_shape) - 1)]     b2 <- tf$reshape(b2_coeffs, as.integer(b2_shape))     start_index <- start_index + prod(b2_shape)      # Set the weight predictions as the weight variables on the outer model.     dense1$kernel <- w1     dense1$bias <- b1     dense2$kernel <- w2     dense2$bias <- b2      # Inference on the outer model.     preds <- main_network(x)     loss <- loss_fn(y, preds)   })    grads <- tape$gradient(loss, hypernetwork$trainable_weights)   optimizer$apply_gradients(zip_lists(grads, hypernetwork$trainable_weights))   loss }  losses <- c()  # Keep track of the losses over time. coro::loop(for (data in dataset) {   x <- data[[1]]   y <- data[[2]]   loss <- train_step(x, y)    # Logging.   losses[length(losses) + 1] <- as.numeric(loss)   if (length(losses) %% 100 == 0) {     cat(\"Step:\", length(losses), \"Loss:\", mean(losses), \"\\n\")   }   # Stop after 1000 steps.   # Training the model to convergence is left   # as an exercise to the reader.   if (length(losses) >= 1000) {     break   } }) ## Step: 100 Loss: 2.536778 ## Step: 200 Loss: 2.236472 ## Step: 300 Loss: 2.119417 ## Step: 400 Loss: 2.040341 ## Step: 500 Loss: 1.949125 ## Step: 600 Loss: 1.859384 ## Step: 700 Loss: 1.845726 ## Step: 800 Loss: 1.820594 ## Step: 900 Loss: 1.771334 ## Step: 1000 Loss: 1.730648"},{"path":"https://keras3.posit.co/dev/articles/making_new_layers_and_models_via_subclassing.html","id":"introduction","dir":"Articles","previous_headings":"","what":"Introduction","title":"Making new layers and models via subclassing","text":"guide cover everything need know build subclassed layers models. particular, ’ll learn following features: Layer class add_weight() method Trainable non-trainable weights build() method Making sure layers can used backend add_loss() method training argument call() mask argument call() Making sure layers can serialized Let’s dive .","code":""},{"path":"https://keras3.posit.co/dev/articles/making_new_layers_and_models_via_subclassing.html","id":"setup","dir":"Articles","previous_headings":"","what":"Setup","title":"Making new layers and models via subclassing","text":"","code":"library(keras3) library(tensorflow, exclude = c(\"set_random_seed\", \"shape\")) library(tfdatasets, exclude = \"shape\")"},{"path":"https://keras3.posit.co/dev/articles/making_new_layers_and_models_via_subclassing.html","id":"the-layer-class-the-combination-of-state-weights-and-some-computation","dir":"Articles","previous_headings":"","what":"The Layer class: the combination of state (weights) and some computation","title":"Making new layers and models via subclassing","text":"One central abstractions Keras Layer class. layer encapsulates state (layer’s “weights”) transformation inputs outputs (“call”, layer’s forward pass). ’s densely-connected layer. two state variables: variables w b. use layer calling tensor input(s), much like R function. Note weights w b automatically tracked layer upon set layer attributes:","code":"layer_linear <- Layer(\"Linear\",    initialize = function(units = 32, input_dim = 32, ...) {     super$initialize(...)     self$w <- self$add_weight(       shape = shape(input_dim, units),       initializer = \"random_normal\",       trainable = TRUE     )     self$b <- self$add_weight(       shape = shape(units),       initializer = \"zeros\",       trainable = TRUE     )   },    call = function(inputs) {     op_matmul(inputs, self$w) + self$b   } ) x <- op_ones(c(2, 2)) linear_layer <- layer_linear(units = 4, input_dim = 2) y <- linear_layer(x) print(y) ## tf.Tensor( ## [[0.02153057 0.15450525 0.0205495  0.04493225] ##  [0.02153057 0.15450525 0.0205495  0.04493225]], shape=(2, 4), dtype=float32) linear_layer$weights ## [[1]] ## <Variable path=linear/variable, shape=(2, 4), dtype=float32, value=[[-0.06251299  0.05335509  0.01485647 -0.00985784] ##  [ 0.08404355  0.10115016  0.00569303  0.05479009]]> ## ## [[2]] ## <Variable path=linear/variable_1, shape=(4), dtype=float32, value=[0. 0. 0. 0.]>"},{"path":"https://keras3.posit.co/dev/articles/making_new_layers_and_models_via_subclassing.html","id":"layers-can-have-non-trainable-weights","dir":"Articles","previous_headings":"","what":"Layers can have non-trainable weights","title":"Making new layers and models via subclassing","text":"Besides trainable weights, can add non-trainable weights layer well. weights meant taken account backpropagation, training layer. ’s add use non-trainable weight: ’s part layer$weights, gets categorized non-trainable weight:","code":"layer_compute_sum <- Layer(   \"ComputeSum\",   initialize = function(input_dim) {     super$initialize()     self$total <- self$add_weight(       initializer = \"zeros\",       shape = shape(input_dim),       trainable = FALSE     )   },   call = function(inputs) {     self$total$assign_add(op_sum(inputs, axis = 1))     self$total   } )  x <- op_ones(c(2, 2)) my_sum <- layer_compute_sum(input_dim = 2) y <- my_sum(x) print(as.array(y)) ## [1] 2 2 y <- my_sum(x) print(as.array(y)) ## [1] 4 4 cat(\"weights:\", length(my_sum$weights)) ## weights: 1 cat(\"non-trainable weights:\", length(my_sum$non_trainable_weights)) ## non-trainable weights: 1 # It's not included in the trainable weights: cat(\"trainable_weights:\", length(my_sum$trainable_weights)) ## trainable_weights: 0"},{"path":"https://keras3.posit.co/dev/articles/making_new_layers_and_models_via_subclassing.html","id":"best-practice-deferring-weight-creation-until-the-shape-of-the-inputs-is-known","dir":"Articles","previous_headings":"","what":"Best practice: deferring weight creation until the shape of the inputs is known","title":"Making new layers and models via subclassing","text":"Linear layer took input_dim argument used compute shape weights w b initialize(): many cases, may know advance size inputs, like lazily create weights value becomes known, time instantiating layer. Keras API, recommend creating layer weights build(self, inputs_shape) method layer. Like : call() method layer automatically run build first time called. now layer ’s lazy thus easier use: Implementing build() separately shown nicely separates creating weights using weights every call.","code":"layer_linear <- Layer(\"Linear\",    initialize = function(units = 32, input_dim = 32, ...) {     super$initialize(...)     self$w <- self$add_weight(       shape = shape(input_dim, units),       initializer = \"random_normal\",       trainable = TRUE     )     self$b <- self$add_weight(       shape = shape(units),       initializer = \"zeros\",       trainable = TRUE     )   },    call = function(inputs) {     op_matmul(inputs, self$w) + self$b   } ) layer_linear <- Layer(   \"Linear\",   initialize = function(units = 32, ...) {     self$units <- as.integer(units)     super$initialize(...)   },   build = function(input_shape) {     self$w <- self$add_weight(       shape = shape(tail(input_shape, 1), self$units),       initializer = \"random_normal\",       trainable = TRUE     )     self$b <- self$add_weight(       shape = shape(self$units),       initializer = \"zeros\",       trainable = TRUE     )   },   call = function(inputs) {     op_matmul(inputs, self$w) + self$b   } ) # At instantiation, we don't know on what inputs this is going to get called linear_layer <- layer_linear(units = 32)  # The layer's weights are created dynamically the first time the layer is called y <- linear_layer(x)"},{"path":"https://keras3.posit.co/dev/articles/making_new_layers_and_models_via_subclassing.html","id":"layers-are-recursively-composable","dir":"Articles","previous_headings":"","what":"Layers are recursively composable","title":"Making new layers and models via subclassing","text":"assign Layer instance attribute another Layer, outer layer start tracking weights created inner layer. recommend creating sublayers initialize() method leave first call() trigger building weights.","code":"MLPBlock <- Layer(   \"MLPBlock\",   initialize = function() {     super$initialize()     self$linear_1 <- layer_linear(units = 32)     self$linear_2 <- layer_linear(units = 32)     self$linear_3 <- layer_linear(units = 1)   },   call = function(inputs) {     inputs |>       self$linear_1() |>       activation_relu() |>       self$linear_2() |>       activation_relu() |>       self$linear_3()   } )  mlp <- MLPBlock() # The first call to the `mlp` will create the weights y <- mlp(op_ones(shape = c(3, 64)))  cat(\"weights:\", length(mlp$weights), \"\\n\") ## weights: 6 cat(\"trainable weights:\", length(mlp$trainable_weights), \"\\n\") ## trainable weights: 6"},{"path":"https://keras3.posit.co/dev/articles/making_new_layers_and_models_via_subclassing.html","id":"backend-agnostic-layers-and-backend-specific-layers","dir":"Articles","previous_headings":"","what":"Backend-agnostic layers and backend-specific layers","title":"Making new layers and models via subclassing","text":"long layer uses APIs ops namespace (ie. using functions starting op_), (Keras namespaces activations_*, random_*, layer_*), can used backend – TensorFlow, JAX, PyTorch. layers ’ve seen far guide work Keras backends. ops namespace gives access : NumPy API, e.g. op_matmul, op_sum, op_reshape, op_stack, etc. Neural networks-specific APIs op_softmax, op_conv, op_binary_crossentropy, op_relu, etc. can also use backend-native APIs layers (tf$nn functions), , layer usable backend question. instance, write following JAX-specific layer using jax$numpy: equivalent TensorFlow-specific layer: equivalent PyTorch-specific layer: cross-backend compatibility tremendously useful property, strongly recommend seek always make layers backend-agnostic leveraging Keras APIs.","code":"# keras3::install_keras(backend = c(\"jax\")) jax <- reticulate::import(\"jax\")  Linear <- new_layer_class(   ...   call = function(inputs) {     jax$numpy$matmul(inputs, self$w) + self$b   } ) library(tensorflow)  Linear <- new_layer_class(   ...   call = function(inputs) {     tf$matmul(inputs, self$w) + self$b   } ) torch <- reticulate::import(\"torch\")  Linear <- new_layer_class(   ...   call = function(inputs) {     torch$matmul(inputs, self$w) + self$b   } )"},{"path":"https://keras3.posit.co/dev/articles/making_new_layers_and_models_via_subclassing.html","id":"the-add_loss-method","dir":"Articles","previous_headings":"","what":"The add_loss() method","title":"Making new layers and models via subclassing","text":"writing call() method layer, can create loss tensors want use later, writing training loop. doable calling self$add_loss(value): losses (including created inner layer) can retrieved via layer$losses. property reset start every call top-level layer, layer$losses always contains loss values created last forward pass. addition, loss property also contains regularization losses created weights inner layer: losses meant taken account writing custom training loops. also work seamlessly fit() (get automatically summed added main loss, ):","code":"# A layer that creates an activity regularization loss layer_activity_regularization <- Layer(   \"ActivityRegularizationLayer\",   initialize = function(rate = 1e-2) {     self$rate <- as.numeric(rate)     super$initialize()   },   call = function(inputs) {     self$add_loss(self$rate * op_mean(inputs))     inputs   } ) layer_outer <- Layer(   \"OuterLayer\",   initialize = function() {     super$initialize()     self$activity_reg <- layer_activity_regularization(rate = 1e-2)   },   call = function(inputs) {     self$activity_reg(inputs)     inputs   } )  layer <- layer_outer() # No losses yet since the layer has never been called cat(\"losses:\", length(layer$losses), \"\\n\") ## losses: 0 x <- layer(op_zeros(c(1, 1))) # We created one loss value cat(\"losses:\", length(layer$losses), \"\\n\") ## losses: 1 # `layer$losses` gets reset at the start of each call x <- layer(op_zeros(c(1, 1))) # This is the loss created during the call above cat(\"losses:\", length(layer$losses), \"\\n\") ## losses: 1 layer_outer_with_kernel_regularizer <- Layer(   \"OuterLayerWithKernelRegularizer\",   initialize = function() {     super$initialize()     self$dense <- layer_dense(units = 32,                               kernel_regularizer = regularizer_l2(1e-3))   },   call = function(inputs) {     self$dense(inputs)   } )  layer <- layer_outer_with_kernel_regularizer() x <- layer(op_zeros(c(1, 1)))  # This is `1e-3 * sum(layer$dense$kernel ** 2)`, # created by the `kernel_regularizer` above. print(layer$losses) ## [[1]] ## tf.Tensor(0.002025157, shape=(), dtype=float32) inputs <- keras_input(shape = 3) outputs <- inputs |> layer_activity_regularization() model <- keras_model(inputs, outputs)  # If there is a loss passed in `compile`, the regularization # losses get added to it model |> compile(optimizer = \"adam\", loss = \"mse\") model |> fit(random_normal(c(2, 3)), random_normal(c(2, 3)), epochs = 1) ## 1/1 - 0s - 131ms/step - loss: 1.9081 # It's also possible not to pass any loss in `compile`, # since the model already has a loss to minimize, via the `add_loss` # call during the forward pass! model |> compile(optimizer = \"adam\") model |> fit(random_normal(c(2, 3)), random_normal(c(2, 3)), epochs = 1) ## 1/1 - 0s - 106ms/step - loss: 1.6613"},{"path":"https://keras3.posit.co/dev/articles/making_new_layers_and_models_via_subclassing.html","id":"you-can-optionally-enable-serialization-on-your-layers","dir":"Articles","previous_headings":"","what":"You can optionally enable serialization on your layers","title":"Making new layers and models via subclassing","text":"need custom layers serializable part Functional model, can optionally implement get_config() method: Note initialize() method base Layer class takes keyword arguments, particular name dtype. ’s good practice pass arguments parent class initialize() include layer config: need flexibility deserializing layer config, can also override from_config() class method. base implementation from_config(): learn serialization saving, see complete guide saving serializing models.","code":"layer_linear <- Layer(   \"Linear\",   initialize = function(units = 32) {     self$units <- as.integer(units)     super$initialize()   },   build = function(input_shape) {     self$w <- self$add_weight(       shape = shape(tail(input_shape, 1), self$units),       initializer = \"random_normal\",       trainable = TRUE     )     self$b <- self$add_weight(       shape = shape(self$units),       initializer = \"zeros\",       trainable = TRUE     )   },   call = function(inputs) {     op_matmul(inputs, self$w) + self$b   },   get_config = function() {     list(units = self$units)   } )  # Now you can recreate the layer from its config: layer <- layer_linear(units = 64) config <- get_config(layer) str(config) ## List of 1 ##  $ units: int 64 ##  - attr(*, \"__class__\")=<class '<r-globalenv>.Linear'> new_layer <- from_config(config) Linear <- new_layer_class(   \"Linear\",   initialize = function(units = 32, ...) {     self$units <- as.integer(units)     super$initialize(...)   },   build = function(input_shape) {     self$w <- self$add_weight(       shape = shape(tail(input_shape, 1), self$units),       initializer = \"random_normal\",       trainable = TRUE     )     self$b <- self$add_weight(       shape = shape(self$units),       initializer = \"zeros\",       trainable = TRUE     )   },   call = function(inputs) {     op_matmul(inputs, self$w) + self$b   },   get_config = function() {     list(units = self$units)   } )  layer <- Linear(units = 64) config <- get_config(layer) str(config) ## List of 1 ##  $ units: int 64 ##  - attr(*, \"__class__\")=<class '<r-globalenv>.Linear'> new_layer <- from_config(config) Layer(   ...,   from_config = function(config) {     # calling `__class__`() creates a new instance and calls initialize()     do.call(`__class__`, config)   } )"},{"path":"https://keras3.posit.co/dev/articles/making_new_layers_and_models_via_subclassing.html","id":"privileged-training-argument-in-the-call-method","dir":"Articles","previous_headings":"","what":"Privileged training argument in the call() method","title":"Making new layers and models via subclassing","text":"layers, particular BatchNormalization layer Dropout layer, different behaviors training inference. layers, standard practice expose training (boolean) argument call() method. exposing argument call(), enable built-training evaluation loops (e.g. fit()) correctly use layer training inference.","code":"layer_custom_dropout <- Layer(   \"CustomDropout\",   initialize = function(rate, ...) {     super$initialize(...)     self$rate <- rate     self$seed_generator <- random_seed_generator(1337)   },   call = function(inputs, training = NULL) {     if (isTRUE(training))       return(random_dropout(inputs, rate = self$rate,                             seed = self.seed_generator))     inputs   } )"},{"path":"https://keras3.posit.co/dev/articles/making_new_layers_and_models_via_subclassing.html","id":"privileged-mask-argument-in-the-call-method","dir":"Articles","previous_headings":"","what":"Privileged mask argument in the call() method","title":"Making new layers and models via subclassing","text":"privileged argument supported call() mask argument. find Keras RNN layers. mask boolean tensor (one boolean value per timestep input) used skip certain input timesteps processing timeseries data. Keras automatically pass correct mask argument call() layers support , mask generated prior layer. Mask-generating layers Embedding layer configured mask_zero = TRUE, Masking layer.","code":""},{"path":"https://keras3.posit.co/dev/articles/making_new_layers_and_models_via_subclassing.html","id":"the-model-class","dir":"Articles","previous_headings":"","what":"The Model class","title":"Making new layers and models via subclassing","text":"general, use Layer class define inner computation blocks, use Model class define outer model – object train. instance, ResNet50 model, several ResNet blocks subclassing Layer, single Model encompassing entire ResNet50 network. Model class API Layer, following differences: exposes built-training, evaluation, prediction loops (fit(), evaluate(), predict()). exposes list inner layers, via model$layers property. exposes saving serialization APIs (save(), save_weights()…) Effectively, Layer class corresponds refer literature “layer” (“convolution layer” “recurrent layer”) “block” (“ResNet block” “Inception block”). Meanwhile, Model class corresponds referred literature “model” (“deep learning model”) “network” (“deep neural network”). ’re wondering, “use Layer class Model class?”, ask : need call fit() ? need call save() ? , go Model. (either class just block bigger system, writing training & saving code ), use Layer. instance, take mini-resnet example , use build Model train fit(), save save_weights():","code":"ResNet <- Model(   \"ResNet\",   initialize = function(num_classes = 1000, ...) {     super$initialize(...)     self$block_1 <- layer_resnet_block()     self$block_2 <- layer_resnet_block()     self$global_pool <- layer_global_average_pooling_2d()     self$classifier <- layer_dense(num_classes)   },   call = function(inputs) {     inputs |>       self$block_1() |>       self$block_2() |>       self$global_pool() |>       self$classifier()   } )  resnet <- ResNet() dataset <- ... resnet |> fit(dataset, epochs=10) resnet |> save_model(\"filepath.keras\")"},{"path":"https://keras3.posit.co/dev/articles/making_new_layers_and_models_via_subclassing.html","id":"putting-it-all-together-an-end-to-end-example","dir":"Articles","previous_headings":"","what":"Putting it all together: an end-to-end example","title":"Making new layers and models via subclassing","text":"’s ’ve learned far: Layer encapsulate state (created initialize() build()) computation (defined call()). Layers can recursively nested create new, bigger computation blocks. Layers backend-agnostic long use Keras APIs. can use backend-native APIs (jax$numpy, torch$nn tf$nn), layer usable specific backend. Layers can create track losses (typically regularization losses) via add_loss(). outer container, thing want train, Model. Model just like Layer, added training serialization utilities. Let’s put things together end--end example: ’re going implement Variational AutoEncoder (VAE) backend-agnostic fashion – runs TensorFlow, JAX, PyTorch. ’ll train MNIST digits. VAE subclass Model, built nested composition layers subclass Layer. feature regularization loss (KL divergence). Let’s train MNIST using fit() API:","code":"layer_sampling <- Layer(   \"Sampling\",   initialize = function(...) {     super$initialize(...)     self$seed_generator <- random_seed_generator(1337)   },   call = function(inputs) {     c(z_mean, z_log_var) %<-% inputs     batch <- op_shape(z_mean)[[1]]     dim <- op_shape(z_mean)[[2]]     epsilon <- random_normal(shape = c(batch, dim),                              seed=self$seed_generator)     z_mean + op_exp(0.5 * z_log_var) * epsilon   } )  # Maps MNIST digits to a triplet (z_mean, z_log_var, z). layer_encoder <- Layer(   \"Encoder\",   initialize = function(latent_dim = 32, intermediate_dim = 64, ...) {     super$initialize(...)     self$dense_proj <-       layer_dense(units = intermediate_dim,  activation = \"relu\")     self$dense_mean <- layer_dense(units = latent_dim)     self$dense_log_var <- layer_dense(units = latent_dim)     self$sampling <- layer_sampling()   },   call = function(inputs) {     x <- self$dense_proj(inputs)     z_mean <- self$dense_mean(x)     z_log_var <- self$dense_log_var(x)     z <- self$sampling(list(z_mean, z_log_var))     list(z_mean, z_log_var, z)   } )  # Converts z, the encoded digit vector, back into a readable digit. layer_decoder <- Layer(   \"Decoder\",   initialize = function(original_dim, intermediate_dim = 64, ...) {     super$initialize(...)     self$dense_proj <-       layer_dense(units = intermediate_dim, activation = \"relu\")     self$dense_output <-       layer_dense(units = original_dim, activation = \"sigmoid\")   },   call = function(inputs) {     x <- self$dense_proj(inputs)     self$dense_output(x)   } )  # Combines the encoder and decoder into an end-to-end model for training. VariationalAutoEncoder <- Model(   \"VariationalAutoEncoder\",    initialize = function(original_dim, intermediate_dim = 64, latent_dim = 32,                         name = \"autoencoder\", ...) {     super$initialize(name = name, ...)     self$original_dim <- original_dim     self$encoder <- layer_encoder(latent_dim = latent_dim,                             intermediate_dim = intermediate_dim)     self$decoder <- layer_decoder(original_dim = original_dim,                             intermediate_dim = intermediate_dim)   },    call = function(inputs) {     c(z_mean, z_log_var, z) %<-% self$encoder(inputs)     reconstructed <- self$decoder(z)     # Add KL divergence regularization loss.     kl_loss <- -0.5 * op_mean(z_log_var - op_square(z_mean) - op_exp(z_log_var) + 1)     self$add_loss(kl_loss)     reconstructed   } ) c(c(x_train, .), .) %<-% dataset_mnist() x_train <- x_train |>   op_reshape(c(60000, 784)) |>   op_cast(\"float32\") |>   op_divide(255)  original_dim <- 784 vae <- VariationalAutoEncoder(   original_dim = 784,   intermediate_dim = 64,   latent_dim = 32 )  optimizer <- optimizer_adam(learning_rate = 1e-3) vae |> compile(optimizer, loss = loss_mean_squared_error())  vae |> fit(x_train, x_train, epochs = 2, batch_size = 64) ## Epoch 1/2 ## 938/938 - 4s - 4ms/step - loss: 0.0748 ## Epoch 2/2 ## 938/938 - 1s - 1ms/step - loss: 0.0676"},{"path":"https://keras3.posit.co/dev/articles/sequential_model.html","id":"setup","dir":"Articles","previous_headings":"","what":"Setup","title":"The Sequential model","text":"","code":"library(keras3)"},{"path":"https://keras3.posit.co/dev/articles/sequential_model.html","id":"when-to-use-a-sequential-model","dir":"Articles","previous_headings":"","what":"When to use a Sequential model","title":"The Sequential model","text":"Sequential model appropriate plain stack layers layer exactly one input tensor one output tensor. Schematically, following Sequential model: equivalent function: Sequential model appropriate : model multiple inputs multiple outputs layers multiple inputs multiple outputs need layer sharing want non-linear topology (e.g. residual connection, multi-branch model)","code":"model <- keras_model_sequential() |>   layer_dense(units = 2, activation = \"relu\", name = \"layer1\") |>   layer_dense(units = 3, activation = \"relu\", name = \"layer2\") |>   layer_dense(units = 4, name = \"layer3\")  # Call model on a test input x <- op_ones(c(3, 3)) y <- model(x) # Create 3 layers layer1 <- layer_dense(units = 2, activation=\"relu\", name=\"layer1\") layer2 <- layer_dense(units = 3, activation=\"relu\", name=\"layer2\") layer3 <- layer_dense(units = 4, name=\"layer3\")  # Call layers on a test input x <- op_ones(c(3, 3)) y <- x |> layer1() |> layer2() |> layer3()"},{"path":"https://keras3.posit.co/dev/articles/sequential_model.html","id":"creating-a-sequential-model","dir":"Articles","previous_headings":"","what":"Creating a Sequential model","title":"The Sequential model","text":"can create Sequential model piping layers keras_model_sequential() object: passing list layers keras_model_sequential(): layers accessible via layers attribute: can also create Sequential model incrementally: Note ’s also corresponding pop_layer() method remove layers: Sequential model behaves much like stack layers. Also note Sequential constructor accepts name argument, just like layer model Keras. useful annotate TensorBoard graphs semantically meaningful names.","code":"model <- keras_model_sequential() |>   layer_dense(units = 2, activation = \"relu\") |>   layer_dense(units = 3, activation = \"relu\") |>   layer_dense(units = 4) model <- keras_model_sequential(layers = list(   layer_dense(units = 2, activation = \"relu\"),   layer_dense(units = 3, activation = \"relu\"),   layer_dense(units = 4) )) model$layers ## [[1]] ## <Dense name=dense_3, built=False> ##  signature: (*args, **kwargs) ## ## [[2]] ## <Dense name=dense_4, built=False> ##  signature: (*args, **kwargs) ## ## [[3]] ## <Dense name=dense_5, built=False> ##  signature: (*args, **kwargs) model <- keras_model_sequential() model |> layer_dense(units = 2, activation=\"relu\") model |> layer_dense(units = 3, activation=\"relu\") model |> layer_dense(units = 4) model |> pop_layer() ## <Dense name=dense_8, built=False> ##  signature: (*args, **kwargs) length(model$layers)  # 2 ## [1] 2 model <- keras_model_sequential(name = \"my_sequential\") model |> layer_dense(units = 2, activation=\"relu\", name = \"layer1\") model |> layer_dense(units = 3, activation=\"relu\", name = \"layer2\") model |> layer_dense(units = 4, name = \"layer3\")"},{"path":"https://keras3.posit.co/dev/articles/sequential_model.html","id":"specifying-the-input-shape-in-advance","dir":"Articles","previous_headings":"","what":"Specifying the input shape in advance","title":"The Sequential model","text":"Generally, layers Keras need know shape inputs order able create weights. create layer like , initially, weights: creates weights first time called input, since shape weights depends shape inputs: Naturally, also applies Sequential models. instantiate Sequential model without input shape, isn’t “built”: weights (calling model$weights results error stating just ). weights created model first sees input data: model “built”, can call summary() method display contents: However, can useful building Sequential model incrementally able display summary model far, including current output shape. case, start model passing input_shape argument model, knows input shape start: Models built predefined input shape like always weights (even seeing data) always defined output shape. general, ’s recommended best practice always specify input shape Sequential model advance know .","code":"layer <- layer_dense(units = 3) layer$weights  # Empty ## list() # Call layer on a test input x <- op_ones(c(1, 4)) y <- layer(x) layer$weights  # Now it has weights, of shape (4, 3) and (3,) ## [[1]] ## <Variable path=dense_9/kernel, shape=(4, 3), dtype=float32, value=[[ 0.48581433  0.78749573  0.61015   ] ##  [ 0.7962619   0.7261175  -0.8046875 ] ##  [-0.6189915   0.37973273  0.50559556] ##  [-0.5455791  -0.60714126  0.19791973]]> ## ## [[2]] ## <Variable path=dense_9/bias, shape=(3), dtype=float32, value=[0. 0. 0.]> model <- keras_model_sequential() |>   layer_dense(units = 2, activation = \"relu\") |>   layer_dense(units = 3, activation = \"relu\") |>   layer_dense(units = 4) # No weights at this stage!  # At this point, you can't do this: # model$weights   # Call the model on a test input x <- op_ones(c(1, 4)) y <- model(x) length(model$weights) ## [1] 6 summary(model) ## Model: \"sequential_4\" ## ┏━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━┓ ## ┃ Layer (type)                    ┃ Output Shape           ┃       Param # ┃ ## ┡━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━┩ ## │ dense_10 (Dense)                │ (1, 2)                 │            10 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ dense_11 (Dense)                │ (1, 3)                 │             9 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ dense_12 (Dense)                │ (1, 4)                 │            16 │ ## └─────────────────────────────────┴────────────────────────┴───────────────┘ ##  Total params: 35 (140.00 B) ##  Trainable params: 35 (140.00 B) ##  Non-trainable params: 0 (0.00 B) model <- keras_model_sequential(input_shape = 4) |>   layer_dense(units = 2, activation = \"relu\") summary(model) ## Model: \"sequential_5\" ## ┏━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━┓ ## ┃ Layer (type)                    ┃ Output Shape           ┃       Param # ┃ ## ┡━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━┩ ## │ dense_13 (Dense)                │ (None, 2)              │            10 │ ## └─────────────────────────────────┴────────────────────────┴───────────────┘ ##  Total params: 10 (40.00 B) ##  Trainable params: 10 (40.00 B) ##  Non-trainable params: 0 (0.00 B) model$layers ## [[1]] ## <Dense name=dense_13, built=True> ##  signature: (*args, **kwargs)"},{"path":"https://keras3.posit.co/dev/articles/sequential_model.html","id":"a-common-debugging-workflow-add-layers-summary","dir":"Articles","previous_headings":"","what":"A common debugging workflow: add layers + summary()","title":"The Sequential model","text":"building new Sequential architecture, ’s useful incrementally stack layers |> frequently print model summaries. instance, enables monitor stack Conv2D MaxPooling2D layers downsampling image feature maps: practical, right? Note |> equivalent calling model$add(), modifies model -place, don’t need reassign model symbol step.","code":"model <- keras_model_sequential(input_shape = c(250, 250, 3)) |>   layer_conv_2d(filters = 32, kernel_size = 5, strides = 2, activation = \"relu\") |>   layer_conv_2d(filters = 32, kernel_size = 3, activation = \"relu\") |>   layer_max_pooling_2d(pool_size = c(3, 3))  # Can you guess what the current output shape is at this point? Probably not. # Let's just print it: summary(model) ## Model: \"sequential_6\" ## ┏━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━┓ ## ┃ Layer (type)                    ┃ Output Shape           ┃       Param # ┃ ## ┡━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━┩ ## │ conv2d (Conv2D)                 │ (None, 123, 123, 32)   │         2,432 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ conv2d_1 (Conv2D)               │ (None, 121, 121, 32)   │         9,248 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ max_pooling2d (MaxPooling2D)    │ (None, 40, 40, 32)     │             0 │ ## └─────────────────────────────────┴────────────────────────┴───────────────┘ ##  Total params: 11,680 (45.62 KB) ##  Trainable params: 11,680 (45.62 KB) ##  Non-trainable params: 0 (0.00 B) # The answer was: (40, 40, 32), so we can keep downsampling...  model |>   layer_conv_2d(filters = 32, kernel_size = 3, activation = \"relu\") |>   layer_conv_2d(filters = 32, kernel_size = 3, activation = \"relu\") |>   layer_max_pooling_2d(pool_size = 3) |>   layer_conv_2d(filters = 32, kernel_size = 3, activation = \"relu\") |>   layer_conv_2d(filters = 32, kernel_size = 3, activation = \"relu\") |>   layer_max_pooling_2d(pool_size = 2)  # And now? summary(model) ## Model: \"sequential_6\" ## ┏━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━┓ ## ┃ Layer (type)                    ┃ Output Shape           ┃       Param # ┃ ## ┡━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━┩ ## │ conv2d (Conv2D)                 │ (None, 123, 123, 32)   │         2,432 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ conv2d_1 (Conv2D)               │ (None, 121, 121, 32)   │         9,248 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ max_pooling2d (MaxPooling2D)    │ (None, 40, 40, 32)     │             0 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ conv2d_2 (Conv2D)               │ (None, 38, 38, 32)     │         9,248 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ conv2d_3 (Conv2D)               │ (None, 36, 36, 32)     │         9,248 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ max_pooling2d_1 (MaxPooling2D)  │ (None, 12, 12, 32)     │             0 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ conv2d_4 (Conv2D)               │ (None, 10, 10, 32)     │         9,248 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ conv2d_5 (Conv2D)               │ (None, 8, 8, 32)       │         9,248 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ max_pooling2d_2 (MaxPooling2D)  │ (None, 4, 4, 32)       │             0 │ ## └─────────────────────────────────┴────────────────────────┴───────────────┘ ##  Total params: 48,672 (190.12 KB) ##  Trainable params: 48,672 (190.12 KB) ##  Non-trainable params: 0 (0.00 B) # Now that we have 4x4 feature maps, time to apply global max pooling. model |>   layer_global_max_pooling_2d()  # Finally, we add a classification layer. model |>   layer_dense(units = 10, activation = \"softmax\")"},{"path":"https://keras3.posit.co/dev/articles/sequential_model.html","id":"what-to-do-once-you-have-a-model","dir":"Articles","previous_headings":"","what":"What to do once you have a model","title":"The Sequential model","text":"model architecture ready, want : Train model, evaluate , run inference. See guide training & evaluation built-loops Save model disk restore . See guide serialization & saving.","code":""},{"path":"https://keras3.posit.co/dev/articles/sequential_model.html","id":"feature-extraction-with-a-sequential-model","dir":"Articles","previous_headings":"","what":"Feature extraction with a Sequential model","title":"The Sequential model","text":"Sequential model built, behaves like Functional API model. means every layer input output attribute. attributes can used neat things, like quickly creating model extracts outputs intermediate layers Sequential model: ’s similar example extract features one layer:","code":"initial_model <- keras_model_sequential(input_shape = c(250, 250, 3)) |>   layer_conv_2d(filters = 32, kernel_size = 5, strides = 2, activation = \"relu\") |>   layer_conv_2d(filters = 32, kernel_size = 3, activation = \"relu\") |>   layer_conv_2d(filters = 32, kernel_size = 3, activation = \"relu\")   feature_extractor <- keras_model(     inputs = initial_model$inputs,     outputs = lapply(initial_model$layers, function(x) x$output), )  # Call feature extractor on test input. x <- op_ones(c(1, 250, 250, 3)) features <- feature_extractor(x) initial_model <-   keras_model_sequential(input_shape = c(250, 250, 3)) |>   layer_conv_2d(filters = 32, kernel_size = 5, strides = 2,                 activation = \"relu\") |>   layer_conv_2d(filters = 32, kernel_size = 3, activation = \"relu\",                 name = \"my_intermediate_layer\") |>   layer_conv_2d(filters = 32, kernel_size = 3, activation = \"relu\")  feature_extractor <- keras_model(   inputs = initial_model$inputs,   outputs = get_layer(initial_model, \"my_intermediate_layer\")$output, )  # Call feature extractor on test input. x <- op_ones(c(1, 250, 250, 3)) features <- feature_extractor(x)"},{"path":"https://keras3.posit.co/dev/articles/sequential_model.html","id":"transfer-learning-with-a-sequential-model","dir":"Articles","previous_headings":"","what":"Transfer learning with a Sequential model","title":"The Sequential model","text":"Transfer learning consists freezing bottom layers model training top layers. aren’t familiar , make sure read guide transfer learning. two common transfer learning blueprint involving Sequential models. First, let’s say Sequential model, want freeze layers except last one. case, can call freeze_weights(). Alternatively, can iterate model$layers set layer$trainable <- FALSE layer, except last one. Like : Another common blueprint use Sequential model stack pre-trained model freshly initialized classification layers. Like : transfer learning, probably find frequently using two patterns. ’s need know Sequential models! find building models Keras, see: Guide Functional API Guide making new Layers & Models via subclassing","code":"model <- keras_model_sequential(input_shape = 784) |>   layer_dense(units = 32, activation = \"relu\") |>   layer_dense(units = 32, activation = \"relu\") |>   layer_dense(units = 32, activation = \"relu\") |>   layer_dense(units = 10) # Presumably you would want to first load pre-trained weights. model |> load_model_weights(...) # Freeze all layers except the last one. model |> freeze_weights(from = 1, to = -2) model # note the \"Trainable\" column now visible in the summary table ## Model: \"sequential_9\" ## ┏━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━┳━━━━━━━┓ ## ┃ Layer (type)                ┃ Output Shape          ┃    Param # ┃ Trai… ┃ ## ┡━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━╇━━━━━━━┩ ## │ dense_15 (Dense)            │ (None, 32)            │     25,120 │   N   │ ## ├─────────────────────────────┼───────────────────────┼────────────┼───────┤ ## │ dense_16 (Dense)            │ (None, 32)            │      1,056 │   N   │ ## ├─────────────────────────────┼───────────────────────┼────────────┼───────┤ ## │ dense_17 (Dense)            │ (None, 32)            │      1,056 │   N   │ ## ├─────────────────────────────┼───────────────────────┼────────────┼───────┤ ## │ dense_18 (Dense)            │ (None, 10)            │        330 │   Y   │ ## └─────────────────────────────┴───────────────────────┴────────────┴───────┘ ##  Total params: 27,562 (107.66 KB) ##  Trainable params: 330 (1.29 KB) ##  Non-trainable params: 27,232 (106.38 KB) # Another way to freeze all layers except the last one. for (layer in model$layers[-length(model$layers)]) {   layer$trainable <- FALSE }  # Recompile and train (this will only update the weights of the last layer). model |> compile(...) model |> fit(...) # Load a convolutional base with pre-trained weights base_model <- application_xception(weights = 'imagenet',                                    include_top = FALSE,                                    pooling = 'avg')  # Freeze the base model freeze_weights(base_model)  # Use a Sequential model to add a trainable classifier on top model <- keras_model_sequential() |>   base_model() |>   layer_dense(1000) # Compile & train model |> compile(...) model |> fit(...)"},{"path":"https://keras3.posit.co/dev/articles/serialization_and_saving.html","id":"introduction","dir":"Articles","previous_headings":"","what":"Introduction","title":"Save, serialize, and export models","text":"Keras model consists multiple components: architecture, configuration, specifies layers model contain, ’re connected. set weights values (“state model”). optimizer (defined compiling model). set losses metrics (defined compiling model). Keras API saves pieces together unified format, marked .keras extension. zip archive consisting following: JSON-based configuration file (config.json): Records model, layer, trackables’ configuration. H5-based state file, model.weights.h5 (whole model), directory keys layers weights. metadata file JSON, storing things current Keras version. Let’s take look works.","code":""},{"path":"https://keras3.posit.co/dev/articles/serialization_and_saving.html","id":"how-to-save-and-load-a-model","dir":"Articles","previous_headings":"","what":"How to save and load a model","title":"Save, serialize, and export models","text":"10 seconds read guide, ’s need know. Saving Keras model: Loading model back: Now, let’s look details.","code":"# Get model (Sequential, Functional Model, or Model subclass) model <- ...  # The filename needs to end with the .keras extension model |> save_model('path/to/location.keras') model <- load_model('path/to/location.keras')"},{"path":"https://keras3.posit.co/dev/articles/serialization_and_saving.html","id":"setup","dir":"Articles","previous_headings":"","what":"Setup","title":"Save, serialize, and export models","text":"","code":"library(keras3)"},{"path":"https://keras3.posit.co/dev/articles/serialization_and_saving.html","id":"saving","dir":"Articles","previous_headings":"","what":"Saving","title":"Save, serialize, and export models","text":"section saving entire model single file. file include: model’s architecture/config model’s weight values (learned training) model’s compilation information (compile() called) optimizer state, (enables restart training left)","code":""},{"path":"https://keras3.posit.co/dev/articles/serialization_and_saving.html","id":"apis","dir":"Articles","previous_headings":"Saving","what":"APIs","title":"Save, serialize, and export models","text":"can save model save_model(). can load back load_model(). supported format Keras 3 “Keras v3” format, uses .keras extension. Example:","code":"get_model <- function() {   # Create a simple model.   inputs <- keras_input(shape(32))   outputs <- inputs |> layer_dense(1)   model <-  keras_model(inputs, outputs)   model |> compile(optimizer = optimizer_adam(), loss = \"mean_squared_error\")   model }  model <- get_model()  # Train the model. test_input <- random_uniform(c(128, 32)) test_target <- random_uniform(c(128, 1)) model |> fit(test_input, test_target)  # Calling `save('my_model.keras')` creates a zip archive `my_model.keras`. model |> save_model(\"my_model.keras\")  # It can be used to reconstruct the model identically. reconstructed_model <- load_model(\"my_model.keras\")  # Let's check: stopifnot(all.equal(   model |> predict(test_input),   reconstructed_model |> predict(test_input) ))"},{"path":"https://keras3.posit.co/dev/articles/serialization_and_saving.html","id":"custom-objects","dir":"Articles","previous_headings":"Saving","what":"Custom objects","title":"Save, serialize, and export models","text":"section covers basic workflows handling custom layers, functions, models Keras saving reloading. saving model includes custom objects, subclassed Layer, must define get_config() method object class. arguments passed constructor (initialize() method) custom object aren’t simple objects (anything types like ints, strings, etc.), must also explicitly deserialize arguments from_config() class method. Like : Please see Defining config methods section details examples. saved .keras file lightweight store Python code custom objects. Therefore, reload model, load_model requires access definition custom objects used one following methods: Registering custom objects (preferred), Passing custom objects directly loading, Using custom object scope examples workflow:","code":"layer_custom <- Layer(   \"CustomLayer\",   initialize = function(sublayer, ...) {     super$initialize(...)     self$sublayer <- sublayer   },   call = function(x) {     self$sublayer(x)   },   get_config = function() {     base_config <- super$get_config()     config <- list(       sublayer = serialize_keras_object(self$sublayer)     )     c(base_config, config)   },   from_config = function(cls, config) {     sublayer_config <- config$sublayer     sublayer <- deserialize_keras_object(sublayer_config)     cls(sublayer, !!!config)   } )"},{"path":"https://keras3.posit.co/dev/articles/serialization_and_saving.html","id":"registering-custom-objects-preferred","dir":"Articles","previous_headings":"Saving > Custom objects","what":"Registering custom objects (preferred)","title":"Save, serialize, and export models","text":"preferred method, custom object registration greatly simplifies saving loading code. Calling register_keras_serializable() custom object registers object globally master list, allowing Keras recognize object loading model. Let’s create custom model involving custom layer custom activation function demonstrate . Example:","code":"# Clear all previously registered custom objects set_custom_objects(clear = TRUE) ## named list() layer_custom <- Layer(   \"CustomLayer\",   initialize = function(self, factor) {     super$initialize()     self$factor = factor   },    call = function(self, x) {     x * self$factor   },    get_config = function(self) {     list(factor = self$factor)   } )  # Upon registration, you can optionally specify a package or a name. # If left blank, the package defaults to \"Custom\" and the name defaults to # the class name. register_keras_serializable(layer_custom, package = \"MyLayers\")  custom_fn <- keras3:::py_func2(function(x) x^2, name = \"custom_fn\", convert = TRUE)  register_keras_serializable(custom_fn, name=\"custom_fn\", package=\"my_package\")   # Create the model. get_model <- function() {   inputs <- keras_input(shape(4))   mid <- inputs |> layer_custom(0.5)   outputs <- mid |> layer_dense(1, activation = custom_fn)   model <- keras_model(inputs, outputs)   model |> compile(optimizer = \"rmsprop\", loss = \"mean_squared_error\")   model }   # Train the model. train_model <- function(model) {   input <- random_uniform(c(4, 4))   target <- random_uniform(c(4, 1))   model |> fit(input, target, verbose = FALSE, epochs = 1)   model }  test_input <- random_uniform(c(4, 4)) test_target <- random_uniform(c(4, 1))  model <- get_model() |> train_model() model |> save_model(\"custom_model.keras\", overwrite = TRUE)  # Now, we can simply load without worrying about our custom objects. reconstructed_model <- load_model(\"custom_model.keras\")  # Let's check: stopifnot(all.equal(   model |> predict(test_input, verbose = FALSE),   reconstructed_model |> predict(test_input, verbose = FALSE) ))"},{"path":"https://keras3.posit.co/dev/articles/serialization_and_saving.html","id":"passing-custom-objects-to-load_model","dir":"Articles","previous_headings":"Saving > Custom objects","what":"Passing custom objects to load_model()","title":"Save, serialize, and export models","text":"","code":"model <- get_model() |> train_model()  # Calling `save_model('my_model.keras')` creates a zip archive `my_model.keras`. model |> save_model(\"custom_model.keras\", overwrite = TRUE)  # Upon loading, pass a named list containing the custom objects used in the # `custom_objects` argument of `load_model()`. reconstructed_model <-  load_model(  \"custom_model.keras\",   custom_objects = list(CustomLayer = layer_custom,                         custom_fn = custom_fn), )  # Let's check: stopifnot(all.equal(   model |> predict(test_input, verbose = FALSE),   reconstructed_model |> predict(test_input, verbose = FALSE) ))"},{"path":"https://keras3.posit.co/dev/articles/serialization_and_saving.html","id":"using-a-custom-object-scope","dir":"Articles","previous_headings":"Saving > Custom objects","what":"Using a custom object scope","title":"Save, serialize, and export models","text":"code within custom object scope able recognize custom objects passed scope argument. Therefore, loading model within scope allow loading custom objects. Example:","code":"model <- get_model() |> train_model() model |> save_model(\"custom_model.keras\", overwrite = TRUE)  # Pass the custom objects dictionary to a custom object scope and place # the `keras.models.load_model()` call within the scope. custom_objects <- list(CustomLayer = layer_custom, custom_fn = custom_fn)  with_custom_object_scope(custom_objects, {   reconstructed_model <- load_model(\"custom_model.keras\") })  # Let's check: stopifnot(all.equal(   model |> predict(test_input, verbose = FALSE),   reconstructed_model |> predict(test_input, verbose = FALSE) ))"},{"path":"https://keras3.posit.co/dev/articles/serialization_and_saving.html","id":"model-serialization","dir":"Articles","previous_headings":"Saving","what":"Model serialization","title":"Save, serialize, and export models","text":"section saving model’s configuration, without state. model’s configuration (architecture) specifies layers model contains, layers connected. configuration model, model can created freshly initialized state (weights compilation information).","code":""},{"path":"https://keras3.posit.co/dev/articles/serialization_and_saving.html","id":"apis-1","dir":"Articles","previous_headings":"Saving > Model serialization","what":"APIs","title":"Save, serialize, and export models","text":"following serialization APIs available: clone_model(model): make (randomly initialized) copy model. get_config() cls.from_config(): retrieve configuration layer model, recreate model instance config, respectively. keras.models.model_to_json() keras.models.model_from_json(): similar, JSON strings. keras.saving.serialize_keras_object(): retrieve configuration arbitrary Keras object. keras.saving.deserialize_keras_object(): recreate object instance configuration.","code":""},{"path":"https://keras3.posit.co/dev/articles/serialization_and_saving.html","id":"in-memory-model-cloning","dir":"Articles","previous_headings":"Saving > Model serialization","what":"In-memory model cloning","title":"Save, serialize, and export models","text":"can -memory cloning model via clone_model(). equivalent getting config recreating model config (preserve compilation information layer weights values). Example:","code":"new_model <- clone_model(model)"},{"path":"https://keras3.posit.co/dev/articles/serialization_and_saving.html","id":"get_config-and-from_config","dir":"Articles","previous_headings":"Saving > Model serialization","what":"get_config() and from_config()","title":"Save, serialize, and export models","text":"Calling get_config(model) get_config(layer) return named list containing configuration model layer, respectively. define get_config() contain arguments needed initialize() method model layer. loading time, from_config(config) method call initialize() arguments reconstruct model layer. Layer example: Now let’s reconstruct layer using from_config() method: Sequential model example: Functional model example:","code":"layer <- layer_dense(, 3, activation=\"relu\") layer_config <- get_config(layer) str(layer_config) ## List of 12 ##  $ name              : chr \"dense_4\" ##  $ trainable         : logi TRUE ##  $ dtype             :List of 4 ##   ..$ module         : chr \"keras\" ##   ..$ class_name     : chr \"DTypePolicy\" ##   ..$ config         :List of 1 ##   .. ..$ name: chr \"float32\" ##   ..$ registered_name: NULL ##  $ units             : int 3 ##  $ activation        : chr \"relu\" ##  $ use_bias          : logi TRUE ##  $ kernel_initializer:List of 4 ##   ..$ module         : chr \"keras.initializers\" ##   ..$ class_name     : chr \"GlorotUniform\" ##   ..$ config         :List of 1 ##   .. ..$ seed: NULL ##   ..$ registered_name: NULL ##  $ bias_initializer  :List of 4 ##   ..$ module         : chr \"keras.initializers\" ##   ..$ class_name     : chr \"Zeros\" ##   ..$ config         : Named list() ##   ..$ registered_name: NULL ##  $ kernel_regularizer: NULL ##  $ bias_regularizer  : NULL ##  $ kernel_constraint : NULL ##  $ bias_constraint   : NULL ##  - attr(*, \"__class__\")=<class 'keras.src.layers.core.dense.Dense'> new_layer <- from_config(layer_config) model <- keras_model_sequential(input_shape = c(32)) |>   layer_dense(1) config <- get_config(model) new_model <- from_config(config) inputs <- keras_input(c(32)) outputs <- inputs |> layer_dense(1) model <- keras_model(inputs, outputs) config <- get_config(model) new_model <- from_config(config)"},{"path":"https://keras3.posit.co/dev/articles/serialization_and_saving.html","id":"save_model_config-and-load_model_config","dir":"Articles","previous_headings":"Saving > Model serialization","what":"save_model_config() and load_model_config()","title":"Save, serialize, and export models","text":"similar get_config / from_config, except turns model JSON file, can loaded without original model class. also specific models, isn’t meant layers. Example:","code":"model <- keras_model_sequential(input_shape = c(32)) |>   layer_dense(1) save_model_config(model, \"model_config.json\") new_model <- load_model_config(\"model_config.json\") unlink(\"model_config.json\")"},{"path":"https://keras3.posit.co/dev/articles/serialization_and_saving.html","id":"arbitrary-object-serialization-and-deserialization","dir":"Articles","previous_headings":"Saving > Model serialization","what":"Arbitrary object serialization and deserialization","title":"Save, serialize, and export models","text":"serialize_keras_object() deserialize_keras_object() APIs general-purpose APIs can used serialize deserialize Keras object custom object. foundation saving model architecture behind serialize()/deserialize() calls keras. Example: Note serialization format containing necessary information proper reconstruction: module containing name Keras module identifying module object comes class_name containing name object’s class. config information needed reconstruct object registered_name custom objects. See . Now can reconstruct regularizer.","code":"my_reg <- regularizer_l1(0.005) config <- serialize_keras_object(my_reg) str(config) ## List of 4 ##  $ module         : chr \"keras.regularizers\" ##  $ class_name     : chr \"L1\" ##  $ config         :List of 1 ##   ..$ l1: num 0.005 ##  $ registered_name: NULL new_reg <- deserialize_keras_object(config) new_reg ## <keras.src.regularizers.regularizers.L1 object> ##  signature: (x)"},{"path":"https://keras3.posit.co/dev/articles/serialization_and_saving.html","id":"model-weights-saving","dir":"Articles","previous_headings":"Saving","what":"Model weights saving","title":"Save, serialize, and export models","text":"can choose save & load model’s weights. can useful : need model inference: case won’t need restart training, don’t need compilation information optimizer state. transfer learning: case training new model reusing state prior model, don’t need compilation information prior model.","code":""},{"path":"https://keras3.posit.co/dev/articles/serialization_and_saving.html","id":"apis-for-in-memory-weight-transfer","dir":"Articles","previous_headings":"Saving > Model weights saving","what":"APIs for in-memory weight transfer","title":"Save, serialize, and export models","text":"Weights can copied different objects using get_weights() set_weights(): get_weights(<layer>): Returns list arrays weight values. set_weights(<layer>weights): Sets model/layer weights values provided (arrays). Examples: Transferring weights one layer another, memory Transferring weights one model another model compatible architecture, memory case stateless layers stateless layers change order number weights, models can compatible architectures even extra/missing stateless layers.","code":"create_layer <- function() {   layer <- layer_dense(, 64, activation = \"relu\", name = \"dense_2\")   layer$build(shape(NA, 784))   layer }  layer_1 <- create_layer() layer_2 <- create_layer()  # Copy weights from layer 1 to layer 2 layer_2 |> set_weights(get_weights(layer_1)) # Create a simple functional model inputs <- keras_input(shape=c(784), name=\"digits\") outputs <- inputs |>   layer_dense(64, activation = \"relu\", name = \"dense_1\") |>   layer_dense(64, activation = \"relu\", name = \"dense_2\") |>   layer_dense(10, name = \"predictions\") functional_model <- keras_model(inputs = inputs, outputs = outputs,                                name = \"3_layer_mlp\")  # Define a subclassed model with the same architecture SubclassedModel <- new_model_class(   \"SubclassedModel\",   initialize = function(output_dim, name = NULL) {     super$initialize(name = name)     self$output_dim <- output_dim |> as.integer()     self$dense_1 <- layer_dense(, 64, activation = \"relu\",                                 name = \"dense_1\")     self$dense_2 <- layer_dense(, 64, activation = \"relu\",                                 name = \"dense_2\")     self$dense_3 <- layer_dense(, self$output_dim,                                 name = \"predictions\")   },    call = function(inputs) {     inputs |>       self$dense_1() |>       self$dense_2() |>       self$dense_3()   },    get_config = function(self) {     list(output_dim = self$output_dim,          name = self$name)   } )   subclassed_model <- SubclassedModel(10) # Call the subclassed model once to create the weights. subclassed_model(op_ones(c(1, 784))) |> invisible()  # Copy weights from functional_model to subclassed_model. set_weights(subclassed_model, get_weights(functional_model))  stopifnot(all.equal(   get_weights(functional_model),   get_weights(subclassed_model) )) input <- keras_input(shape = c(784), name = \"digits\") output <- input |>   layer_dense(64, activation = \"relu\", name = \"dense_1\") |>   layer_dense(64, activation = \"relu\", name = \"dense_2\") |>   layer_dense(10, name = \"predictions\") functional_model <- keras_model(inputs, outputs,                                 name = \"3_layer_mlp\")  input <- keras_input(shape = c(784), name = \"digits\") output <- input |>   layer_dense(64, activation = \"relu\", name = \"dense_1\") |>   layer_dense(64, activation = \"relu\", name = \"dense_2\") |>   # Add a dropout layer, which does not contain any weights.   layer_dropout(0.5) |>   layer_dense(10, name = \"predictions\")  functional_model_with_dropout <-   keras_model(input, output, name = \"3_layer_mlp\")  set_weights(functional_model_with_dropout,             get_weights(functional_model))"},{"path":"https://keras3.posit.co/dev/articles/serialization_and_saving.html","id":"apis-for-saving-weights-to-disk-loading-them-back","dir":"Articles","previous_headings":"Saving > Model weights saving","what":"APIs for saving weights to disk & loading them back","title":"Save, serialize, and export models","text":"Weights can saved disk calling save_model_weights(filepath). filename end .weights.h5. Example: Note using freeze_weights() may result different output get_weights(layer) ordering model contains nested layers.","code":"sequential_model = keras_model_sequential(input_shape = c(784),                                           input_name = \"digits\") |>   layer_dense(64, activation = \"relu\", name = \"dense_1\") |>   layer_dense(64, activation = \"relu\", name = \"dense_2\") |>   layer_dense(10, name = \"predictions\") sequential_model |> save_model_weights(\"my_model.weights.h5\") sequential_model |> load_model_weights(\"my_model.weights.h5\")"},{"path":"https://keras3.posit.co/dev/articles/serialization_and_saving.html","id":"transfer-learning-example","dir":"Articles","previous_headings":"Saving > Model weights saving > APIs for saving weights to disk & loading them back","what":"Transfer learning example","title":"Save, serialize, and export models","text":"loading pretrained weights weights file, recommended load weights original checkpointed model, extract desired weights/layers new model. Example:","code":"create_functional_model <- function() {   inputs <- keras_input(shape = c(784), name = \"digits\")   outputs <- inputs |>     layer_dense(64, activation = \"relu\", name = \"dense_1\") |>     layer_dense(64, activation = \"relu\", name = \"dense_2\") |>     layer_dense(10, name = \"predictions\")   keras_model(inputs, outputs, name = \"3_layer_mlp\") }  functional_model <- create_functional_model() functional_model |> save_model_weights(\"pretrained.weights.h5\")  # In a separate program: pretrained_model <- create_functional_model() pretrained_model |> load_model_weights(\"pretrained.weights.h5\")  # Create a new model by extracting layers from the original model: extracted_layers <- pretrained_model$layers |> head(-1) model <- keras_model_sequential(layers = extracted_layers) |>   layer_dense(5, name = \"dense_3\") summary(model) ## Model: \"sequential_4\" ## ┏━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━┓ ## ┃ Layer (type)                    ┃ Output Shape           ┃       Param # ┃ ## ┡━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━┩ ## │ dense_1 (Dense)                 │ (None, 64)             │        50,240 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ dense_2 (Dense)                 │ (None, 64)             │         4,160 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ dense_3 (Dense)                 │ (None, 5)              │           325 │ ## └─────────────────────────────────┴────────────────────────┴───────────────┘ ##  Total params: 54,725 (213.77 KB) ##  Trainable params: 54,725 (213.77 KB) ##  Non-trainable params: 0 (0.00 B)"},{"path":[]},{"path":"https://keras3.posit.co/dev/articles/serialization_and_saving.html","id":"defining-the-config-methods","dir":"Articles","previous_headings":"Saving > Appendix: Handling custom objects","what":"Defining the config methods","title":"Save, serialize, and export models","text":"Specifications: get_config() return JSON-serializable named list order compatible Keras architecture model-saving APIs. from_config(config) (class method) return new layer model object created config. default implementation returns .call(cls, config). NOTE: constructor arguments already serializable, e.g. strings ints, non-custom Keras objects, overriding from_config() necessary. However, complex objects layers models passed initialize(), deserialization must handled explicitly either initialize overriding from_config() method. Example: Note overriding from_config unnecessary MyDense hidden_units, kernel_initializer, kernel_regularizer ints, strings, built-Keras object, respectively. means default from_config implementation cls(!!!config) work intended. complex objects, layers models passed initialize(), example, must explicitly deserialize objects. Let’s take look example model from_config override necessary. Example:","code":"layer_my_dense <- register_keras_serializable(   package = \"MyLayers\", name = \"KernelMult\",   object = Layer(     \"MyDense\",     initialize = function(units,                           ...,                           kernel_regularizer = NULL,                           kernel_initializer = NULL,                           nested_model = NULL) {       super$initialize(...)       self$hidden_units <- units       self$kernel_regularizer <- kernel_regularizer       self$kernel_initializer <- kernel_initializer       self$nested_model <- nested_model     },     get_config = function() {       config <- super$get_config()       # Update the config with the custom layer's parameters       config <- modifyList(config, list(         units = self$hidden_units,         kernel_regularizer = self$kernel_regularizer,         kernel_initializer = self$kernel_initializer,         nested_model = self$nested_model       ))       config     },     build = function(input_shape) {       input_units <- tail(input_shape, 1)       self$kernel <- self$add_weight(         name = \"kernel\",         shape = shape(input_units, self$hidden_units),         regularizer = self$kernel_regularizer,         initializer = self$kernel_initializer,       )     },     call = function(inputs) {       op_matmul(inputs, self$kernel)     }   ) )   layer <- layer_my_dense(units = 16,                         kernel_regularizer = \"l1\",                         kernel_initializer = \"ones\") layer3 <- layer_my_dense(units = 64, nested_model = layer)  config <- serialize_keras_object(layer3) str(config) ## List of 4 ##  $ module         : chr \"<r-globalenv>\" ##  $ class_name     : chr \"MyDense\" ##  $ config         :List of 5 ##   ..$ name        : chr \"my_dense_1\" ##   ..$ trainable   : logi TRUE ##   ..$ dtype       :List of 4 ##   .. ..$ module         : chr \"keras\" ##   .. ..$ class_name     : chr \"DTypePolicy\" ##   .. ..$ config         :List of 1 ##   .. .. ..$ name: chr \"float32\" ##   .. ..$ registered_name: NULL ##   ..$ units       : num 64 ##   ..$ nested_model:List of 4 ##   .. ..$ module         : chr \"<r-globalenv>\" ##   .. ..$ class_name     : chr \"MyDense\" ##   .. ..$ config         :List of 6 ##   .. .. ..$ name              : chr \"my_dense\" ##   .. .. ..$ trainable         : logi TRUE ##   .. .. ..$ dtype             :List of 4 ##   .. .. .. ..$ module         : chr \"keras\" ##   .. .. .. ..$ class_name     : chr \"DTypePolicy\" ##   .. .. .. ..$ config         :List of 1 ##   .. .. .. .. ..$ name: chr \"float32\" ##   .. .. .. ..$ registered_name: NULL ##   .. .. ..$ units             : num 16 ##   .. .. ..$ kernel_regularizer: chr \"l1\" ##   .. .. ..$ kernel_initializer: chr \"ones\" ##   .. ..$ registered_name: chr \"MyLayers>KernelMult\" ##  $ registered_name: chr \"MyLayers>KernelMult\" new_layer <- deserialize_keras_object(config) new_layer ## <MyDense name=my_dense_1, built=False> ##  signature: (*args, **kwargs) `%||%` <- \\(x, y) if(is.null(x)) y else x layer_custom_model <- register_keras_serializable(   package = \"ComplexModels\",   object = Layer(     \"CustomModel\",     initialize = function(first_layer, second_layer = NULL, ...) {       super$initialize(...)       self$first_layer <- first_layer       self$second_layer <- second_layer %||% layer_dense(, 8)     },      get_config = function() {       config <- super$get_config()       config <- modifyList(config, list(         first_layer = self$first_layer,         second_layer = self$second_layer       ))       config     },      from_config = function(config) {       config$first_layer %<>% deserialize_keras_object()       config$second_layer %<>% deserialize_keras_object()       # note that the class is available in methods under the classname symbol,       # (`CustomModel` for this class), and also under the symbol `__class__`       cls(!!!config)       # CustomModel(!!!config)     },     call = function(self, inputs) {       inputs |>         self$first_layer() |>         self$second_layer()     }   ) )  # Let's make our first layer the custom layer from the previous example (MyDense) inputs <- keras_input(c(32)) outputs <-  inputs |> layer_custom_model(first_layer=layer) model <- keras_model(inputs, outputs)  config <- get_config(model) new_model <- from_config(config)"},{"path":"https://keras3.posit.co/dev/articles/serialization_and_saving.html","id":"how-custom-objects-are-serialized","dir":"Articles","previous_headings":"Saving > Appendix: Handling custom objects","what":"How custom objects are serialized","title":"Save, serialize, and export models","text":"serialization format special key custom objects registered via register_keras_serializable(). registered_name key allows easy retrieval loading/deserialization time also allowing users add custom naming. Let’s take look config serializing custom layer MyDense defined . Example: shown, registered_name key contains lookup information Keras master list, including package MyLayers custom name KernelMult gave calling register_keras_serializables(). Take look custom class definition/registration . Note class_name key contains original name class, allowing proper re-initialization from_config. Additionally, note module key NULL since custom object.","code":"layer <- layer_my_dense(   units = 16,   kernel_regularizer = regularizer_l1_l2(l1 = 1e-5, l2 = 1e-4),   kernel_initializer = \"ones\", ) config <- serialize_keras_object(layer) str(config) ## List of 4 ##  $ module         : chr \"<r-globalenv>\" ##  $ class_name     : chr \"MyDense\" ##  $ config         :List of 6 ##   ..$ name              : chr \"my_dense_2\" ##   ..$ trainable         : logi TRUE ##   ..$ dtype             :List of 4 ##   .. ..$ module         : chr \"keras\" ##   .. ..$ class_name     : chr \"DTypePolicy\" ##   .. ..$ config         :List of 1 ##   .. .. ..$ name: chr \"float32\" ##   .. ..$ registered_name: NULL ##   ..$ units             : num 16 ##   ..$ kernel_regularizer:List of 4 ##   .. ..$ module         : chr \"keras.regularizers\" ##   .. ..$ class_name     : chr \"L1L2\" ##   .. ..$ config         :List of 2 ##   .. .. ..$ l1: num 1e-05 ##   .. .. ..$ l2: num 1e-04 ##   .. ..$ registered_name: NULL ##   ..$ kernel_initializer: chr \"ones\" ##  $ registered_name: chr \"MyLayers>KernelMult\""},{"path":"https://keras3.posit.co/dev/articles/training_with_built_in_methods.html","id":"setup","dir":"Articles","previous_headings":"","what":"Setup","title":"Training & evaluation with the built-in methods","text":"","code":"library(keras3)"},{"path":"https://keras3.posit.co/dev/articles/training_with_built_in_methods.html","id":"introduction","dir":"Articles","previous_headings":"","what":"Introduction","title":"Training & evaluation with the built-in methods","text":"guide covers training, evaluation, prediction (inference) models using built-APIs training & validation (fit(), evaluate() predict()). interested leveraging fit() specifying training step function, see Customizing happens fit() guide. interested writing training & evaluation loops scratch, see guide Writing training loop scratch. general, whether using built-loops writing , model training & evaluation works strictly way across every kind Keras model – Sequential models, models built Functional API, models written scratch via model subclassing.","code":""},{"path":"https://keras3.posit.co/dev/articles/training_with_built_in_methods.html","id":"api-overview-a-first-end-to-end-example","dir":"Articles","previous_headings":"","what":"API overview: a first end-to-end example","title":"Training & evaluation with the built-in methods","text":"passing data built-training loops model, either use: Arrays (data small fits memory) tf_dataset objects PyTorch DataLoader instances next paragraphs, ’ll use MNIST dataset NumPy arrays, order demonstrate use optimizers, losses, metrics. Afterwards, ’ll take close look options. Let’s consider following model (, build Functional API, Sequential model subclassed model well): ’s typical end--end workflow looks like, consisting : Training Validation holdout set generated original training data Evaluation test data ’ll use MNIST data example. specify training configuration (optimizer, loss, metrics): call fit(), train model slicing data “batches” size batch_size, repeatedly iterating entire dataset given number epochs. returned history object holds record loss values metric values training: evaluate model test data via evaluate(): Now, let’s review piece workflow detail.","code":"inputs <- keras_input(shape = 784, name=\"digits\") outputs <- inputs |>   layer_dense(units = 64, activation = \"relu\", name = \"dense_1\") |>   layer_dense(units = 64, activation = \"relu\", name = \"dense_2\") |>   layer_dense(units = 10, activation = \"softmax\", name = \"predictions\") model <- keras_model(inputs = inputs, outputs = outputs) summary(model) ## Model: \"functional\" ## ┏━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━┓ ## ┃ Layer (type)                    ┃ Output Shape           ┃       Param # ┃ ## ┡━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━┩ ## │ digits (InputLayer)             │ (None, 784)            │             0 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ dense_1 (Dense)                 │ (None, 64)             │        50,240 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ dense_2 (Dense)                 │ (None, 64)             │         4,160 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ predictions (Dense)             │ (None, 10)             │           650 │ ## └─────────────────────────────────┴────────────────────────┴───────────────┘ ##  Total params: 55,050 (215.04 KB) ##  Trainable params: 55,050 (215.04 KB) ##  Non-trainable params: 0 (0.00 B) c(c(x_train, y_train), c(x_test, y_test)) %<-% dataset_mnist()  # Preprocess the data (these are NumPy arrays) x_train <- array_reshape(x_train, c(60000, 784)) / 255 x_test <- array_reshape(x_test, c(10000, 784)) / 255  # Reserve 10,000 samples for validation x_val <- x_train[1:10000,] y_val <- y_train[1:10000] x_train <- x_train[-c(1:10000),] y_train <- y_train[-c(1:10000)] model |> compile(   # Optimizer   optimizer = optimizer_rmsprop(),   # Loss function to minimize   loss = loss_sparse_categorical_crossentropy(),   # List of metrics to monitor   metrics = list(metric_sparse_categorical_accuracy()) ) history <- model |> fit(   x_train, y_train,   batch_size = 64,   epochs = 2,   # We pass some validation for   # monitoring validation loss and metrics   # at the end of each epoch   validation_data = list(x_val, y_val) ) ## Epoch 1/2 ## 782/782 - 3s - 3ms/step - loss: 0.3410 - sparse_categorical_accuracy: 0.9034 - val_loss: 0.1855 - val_sparse_categorical_accuracy: 0.9460 ## Epoch 2/2 ## 782/782 - 1s - 1ms/step - loss: 0.1590 - sparse_categorical_accuracy: 0.9538 - val_loss: 0.1323 - val_sparse_categorical_accuracy: 0.9621 history ## ## Final epoch (plot to see history): ##                            loss: 0.159 ##     sparse_categorical_accuracy: 0.9538 ##                        val_loss: 0.1323 ## val_sparse_categorical_accuracy: 0.9621 # Evaluate the model on the test data using `evaluate` results <- model |> evaluate(x_test, y_test, batch_size=128) ## 79/79 - 0s - 4ms/step - loss: 0.1273 - sparse_categorical_accuracy: 0.9621 str(results) ## List of 2 ##  $ loss                       : num 0.127 ##  $ sparse_categorical_accuracy: num 0.962 # Generate predictions (probabilities -- the output of the last layer) # on new data using `predict` predictions <- model |> predict(x_test[1:2,]) ## 1/1 - 0s - 172ms/step dim(predictions) ## [1]  2 10"},{"path":"https://keras3.posit.co/dev/articles/training_with_built_in_methods.html","id":"the-compile-method-specifying-a-loss-metrics-and-an-optimizer","dir":"Articles","previous_headings":"","what":"The compile() method: specifying a loss, metrics, and an optimizer","title":"Training & evaluation with the built-in methods","text":"train model fit(), need specify loss function, optimizer, optionally, metrics monitor. pass model arguments compile() method: metrics argument list – model can number metrics. model multiple outputs, can specify different losses metrics output, can modulate contribution output total loss model. find details Passing data multi-input, multi-output models section. Note ’re satisfied default settings, many cases optimizer, loss, metrics can specified via string identifiers shortcut: later reuse, let’s put model definition compile step functions; call several times across different examples guide.","code":"model |> compile(   optimizer = optimizer_rmsprop(learning_rate = 1e-3),   loss = loss_sparse_categorical_crossentropy(),   metrics = list(metric_sparse_categorical_accuracy()) ) model |> compile(   optimizer = \"rmsprop\",   loss = \"sparse_categorical_crossentropy\",   metrics = c(\"sparse_categorical_accuracy\") ) get_uncompiled_model <- function() {   inputs <- keras_input(shape = 784, name = \"digits\")   outputs <- inputs |>     layer_dense(units = 64, activation = \"relu\", name = \"dense_1\") |>     layer_dense(units = 64, activation = \"relu\", name = \"dense_2\") |>     layer_dense(units = 10, activation = \"softmax\", name = \"predictions\")   keras_model(inputs = inputs, outputs = outputs) }  get_compiled_model <- function() {   model <- get_uncompiled_model()   model |> compile(     optimizer = \"rmsprop\",     loss = \"sparse_categorical_crossentropy\",     metrics = c(\"sparse_categorical_accuracy\")   )   model }"},{"path":"https://keras3.posit.co/dev/articles/training_with_built_in_methods.html","id":"many-built-in-optimizers-losses-and-metrics-are-available","dir":"Articles","previous_headings":"The compile() method: specifying a loss, metrics, and an optimizer","what":"Many built-in optimizers, losses, and metrics are available","title":"Training & evaluation with the built-in methods","text":"general, won’t create losses, metrics, optimizers scratch, need likely already part Keras API: Optimizers: [optimizer_sgd()] (without momentum) [optimizer_rmsprop()] [optimizer_adam()] etc. Losses: [loss_mean_squared_error()] [loss_kl_divergence()] [loss_cosine_similarity()] etc. Metrics: [metric_auc()] [metric_precision()] [metric_recall()] etc.","code":""},{"path":"https://keras3.posit.co/dev/articles/training_with_built_in_methods.html","id":"custom-losses","dir":"Articles","previous_headings":"The compile() method: specifying a loss, metrics, and an optimizer","what":"Custom losses","title":"Training & evaluation with the built-in methods","text":"need create custom loss, Keras provides three ways . first method involves creating function accepts inputs y_true y_pred. following example shows loss function computes mean squared error real data predictions: need loss function takes parameters beside y_true y_pred, can subclass Keras base Loss class using [Loss()] implement following two methods: initialize(): accept parameters pass call loss function call(y_true, y_pred): use targets (y_true) model predictions (y_pred) compute model’s loss Let’s say want use mean squared error, added term de-incentivize prediction values far 0.5 (assume categorical targets one-hot encoded take values 0 1). creates incentive model confident, may help reduce overfitting (won’t know works try!). ’s :","code":"custom_mean_squared_error <- function(y_true, y_pred) {   op_mean(op_square(y_true - y_pred), axis = -1) }  model <- get_uncompiled_model() model |> compile(optimizer = \"adam\",                  loss = custom_mean_squared_error)  # We need to one-hot encode the labels to use MSE y_train_one_hot <- op_one_hot(y_train, num_classes = 10) model |> fit(x_train, y_train_one_hot, batch_size = 64, epochs = 2) ## Epoch 1/2 ## 782/782 - 2s - 2ms/step - loss: 0.0157 ## Epoch 2/2 ## 782/782 - 1s - 1ms/step - loss: 0.0074 loss_custom_mse <- Loss(   classname = \"CustomMSE\",   initialize = function(regularization_factor = 0.1, name = \"custom_mse\") {     super$initialize(name = name)     self$regularization_factor <- regularization_factor   },   call = function(y_true, y_pred) {     mse <- op_mean(op_square(y_true - y_pred), axis = -1)     reg <- op_mean(op_square(0.5 - y_pred), axis = -1)     mse + reg * self$regularization_factor   } )  model <- get_uncompiled_model() model |> compile(optimizer=\"adam\", loss = loss_custom_mse())  y_train_one_hot <- op_one_hot(y_train, num_classes=10) model |> fit(x_train, y_train_one_hot, batch_size=64, epochs=1) ## 782/782 - 2s - 2ms/step - loss: 0.0388"},{"path":"https://keras3.posit.co/dev/articles/training_with_built_in_methods.html","id":"custom-metrics","dir":"Articles","previous_headings":"The compile() method: specifying a loss, metrics, and an optimizer","what":"Custom metrics","title":"Training & evaluation with the built-in methods","text":"need metric isn’t part API, can easily create custom metrics subclassing Keras base Metric class using [Metric()]. need implement 4 methods: initialize(), create state variables metric. update_state(y_true, y_pred, sample_weight = NULL), uses targets y_true model predictions y_pred update state variables. result(), uses state variables compute final results. reset_state(), reinitializes state metric. State update results computation kept separate (update_state() result(), respectively) cases, results computation might expensive done periodically. ’s simple example showing implement CategoricalTruePositives metric counts many samples correctly classified belonging given class:","code":"metric_categorical_true_positives <- Metric(   \"CategoricalTruePositives\",    initialize = function(name = \"categorical_true_positives\", ...) {     super$initialize(name = name, ...)     self$true_positives <- self$add_variable(shape = shape(),                                              name = \"ctp\",                                              initializer = \"zeros\")   },    update_state = function(y_true, y_pred, sample_weight = NULL) {     y_pred <- op_argmax(y_pred, axis = 2) |> op_reshape(c(-1, 1))     values <- op_cast(y_true, \"int32\") == op_cast(y_pred, \"int32\")     values <- op_cast(values, \"float32\")     if (!is.null(sample_weight)) {       sample_weight <- op_cast(sample_weight, \"float32\")       values <- op_multiply(values, sample_weight)     }     self$true_positives$assign_add(op_sum(values))   },    result = function() {     self$true_positives$value   },    reset_state = function() {     self$true_positives$assign(0.0)   } )  model <- get_uncompiled_model() model |> compile(   optimizer = optimizer_rmsprop(learning_rate = 1e-3),   loss = loss_sparse_categorical_crossentropy(),   metrics = c(metric_categorical_true_positives()) ) history <- model |> fit(x_train, y_train, batch_size = 64, epochs = 3) ## Epoch 1/3 ## 782/782 - 2s - 2ms/step - categorical_true_positives: 285017.0000 - loss: 0.3444 ## Epoch 2/3 ## 782/782 - 1s - 1ms/step - categorical_true_positives: 284268.0000 - loss: 0.1658 ## Epoch 3/3 ## 782/782 - 1s - 973us/step - categorical_true_positives: 284173.0000 - loss: 0.1207"},{"path":"https://keras3.posit.co/dev/articles/training_with_built_in_methods.html","id":"handling-losses-and-metrics-that-dont-fit-the-standard-signature","dir":"Articles","previous_headings":"The compile() method: specifying a loss, metrics, and an optimizer","what":"Handling losses and metrics that don’t fit the standard signature","title":"Training & evaluation with the built-in methods","text":"overwhelming majority losses metrics can computed y_true y_pred, y_pred output model – . instance, regularization loss may require activation layer (targets case), activation may model output. cases, can call self$add_loss(loss_value) inside call method custom layer. Losses added way get added “main” loss training (one passed compile()). ’s simple example adds activity regularization (note activity regularization built-Keras layers – layer just sake providing concrete example): Note pass losses via add_loss(), becomes possible call compile() without loss function, since model already loss minimize. Consider following LogisticEndpoint layer: takes inputs targets & logits, tracks crossentropy loss via add_loss(). can use model two inputs (input data & targets), compiled without loss argument, like : information training multi-input models, see section Passing data multi-input, multi-output models.","code":"layer_custom_activity_regularizer <- Layer(   \"ActivityRegularization\",   call = function(inputs) {     self$add_loss(op_sum(inputs) * 0.1)     inputs  # Pass-through layer.   } )  inputs <- keras_input(shape = 784, name = \"digits\") outputs <- inputs |>   layer_dense(units = 32, activation = \"relu\", name = \"dense_1\") |>   layer_custom_activity_regularizer() |>   layer_dense(units = 64, activation = \"relu\", name = \"dense_2\") |>   layer_dense(units = 10, name = \"predictions\")  model <- keras_model(inputs = inputs, outputs = outputs) model |> compile(optimizer = optimizer_rmsprop(learning_rate = 1e-3),                  loss = loss_sparse_categorical_crossentropy(from_logits = TRUE))  # The displayed loss will be much higher than before # due to the regularization component. model |> fit(x_train, y_train, batch_size = 64, epochs = 1) ## 782/782 - 2s - 2ms/step - loss: 2.3721 layer_logistic_endpoint <- Layer(   \"LogisticEndpoint\",   initialize = function(name = NULL) {     super$initialize(name = name)     self$loss_fn <- loss_binary_crossentropy(from_logits = TRUE)   },   call = function(targets, logits, sample_weights = NULL) {     # Compute the training-time loss value and add it     # to the layer using `self.add_loss()`.     loss <- self$loss_fn(targets, logits, sample_weights)     self$add_loss(loss)      # Return the inference-time prediction tensor (for `predict()`).     op_softmax(logits)   } ) inputs <- keras_input(shape = 3, name = \"inputs\") targets <- keras_input(shape = 10, name = \"targets\")  logits <- inputs |> layer_dense(10) predictions <- layer_logistic_endpoint(name = \"predictions\")(targets, logits)  model <- keras_model(inputs = list(inputs, targets),                      outputs = predictions) model |> compile(optimizer = \"adam\")  # No loss argument!  data <- list(   inputs = random_normal(c(3, 3)),   targets = random_normal(c(3, 10)) ) model |> fit(data, epochs = 1) ## 1/1 - 0s - 439ms/step - loss: 0.9638"},{"path":"https://keras3.posit.co/dev/articles/training_with_built_in_methods.html","id":"automatically-setting-apart-a-validation-holdout-set","dir":"Articles","previous_headings":"The compile() method: specifying a loss, metrics, and an optimizer","what":"Automatically setting apart a validation holdout set","title":"Training & evaluation with the built-in methods","text":"first end--end example saw, used validation_data argument pass list arrays list(x_val, y_val) model evaluating validation loss validation metrics end epoch. ’s another option: argument validation_split allows automatically reserve part training data validation. argument value represents fraction data reserved validation, set number higher 0 lower 1. instance, validation_split = 0.2 means “use 20% data validation”, validation_split = 0.6 means “use 60% data validation”. way validation computed taking last x% samples arrays received fit() call, shuffling. Note can use validation_split training NumPy data.","code":"model <- get_compiled_model() model |> fit(x_train, y_train,              batch_size = 64,              validation_split = 0.2, epochs = 1) ## 625/625 - 2s - 3ms/step - loss: 0.3817 - sparse_categorical_accuracy: 0.8919 - val_loss: 0.1953 - val_sparse_categorical_accuracy: 0.9431"},{"path":"https://keras3.posit.co/dev/articles/training_with_built_in_methods.html","id":"training-evaluation-using-tf-dataset-objects","dir":"Articles","previous_headings":"","what":"Training & evaluation using TF Dataset objects","title":"Training & evaluation with the built-in methods","text":"past paragraphs, ’ve seen handle losses, metrics, optimizers, ’ve seen use validation_data validation_split arguments fit(), data passed arrays. Another option use iterator-like, tf.data.Dataset, PyTorch DataLoader, R generator function. Let’s take look former. tfdatasets R package containes set utilities loading preprocessing data way ’s fast scalable. complete guide creating Datasets, see tf.data documentation. can use tf.data train Keras models regardless backend ’re using – whether ’s JAX, PyTorch, TensorFlow. can pass Dataset instance directly methods fit(), evaluate(), predict(): Note Dataset reset end epoch, can reused next epoch. want run training specific number batches Dataset, can pass steps_per_epoch argument, specifies many training steps model run using Dataset moving next epoch. can also pass Dataset instance validation_data argument fit(): end epoch, model iterate validation dataset compute validation loss validation metrics. want run validation specific number batches dataset, can pass validation_steps argument, specifies many validation steps model run validation dataset interrupting validation moving next epoch: Note validation dataset reset use (always evaluating samples epoch epoch). argument validation_split (generating holdout set training data) supported training Dataset objects, since feature requires ability index samples datasets, possible general Dataset API.","code":"library(tfdatasets, exclude = \"shape\") model <- get_compiled_model()  # First, let's create a training Dataset instance. # For the sake of our example, we'll use the same MNIST data as before. train_dataset <- tensor_slices_dataset(list(x_train, y_train))  # Shuffle and slice the dataset. train_dataset <- train_dataset |>   dataset_shuffle(buffer_size=1024) |>   dataset_batch(64)  # Now we get a test dataset. test_dataset <-   tensor_slices_dataset(list(x_test, y_test)) |>   dataset_batch(64)  # Since the dataset already takes care of batching, # we don't pass a `batch_size` argument. model |> fit(train_dataset, epochs = 3) ## Epoch 1/3 ## 782/782 - 2s - 2ms/step - loss: 0.3365 - sparse_categorical_accuracy: 0.9041 ## Epoch 2/3 ## 782/782 - 1s - 1ms/step - loss: 0.1606 - sparse_categorical_accuracy: 0.9521 ## Epoch 3/3 ## 782/782 - 1s - 1ms/step - loss: 0.1178 - sparse_categorical_accuracy: 0.9646 # You can also evaluate or predict on a dataset. result <- model |> evaluate(test_dataset) ## 157/157 - 1s - 3ms/step - loss: 0.1213 - sparse_categorical_accuracy: 0.9596 result ## $loss ## [1] 0.1213005 ## ## $sparse_categorical_accuracy ## [1] 0.9596 model <- get_compiled_model()  # Prepare the training dataset train_dataset <- tensor_slices_dataset(list(x_train, y_train)) train_dataset <- train_dataset |>   dataset_shuffle(buffer_size = 1024) |>   dataset_batch(64)  # Only use the 100 batches per epoch (that's 64 * 100 samples) model |> fit(train_dataset, epochs = 3, steps_per_epoch = 100) ## Epoch 1/3 ## 100/100 - 1s - 8ms/step - loss: 0.8017 - sparse_categorical_accuracy: 0.7806 ## Epoch 2/3 ## 100/100 - 0s - 1ms/step - loss: 0.3661 - sparse_categorical_accuracy: 0.9006 ## Epoch 3/3 ## 100/100 - 0s - 960us/step - loss: 0.3009 - sparse_categorical_accuracy: 0.9106 model <- get_compiled_model()  # Prepare the training dataset train_dataset <- tensor_slices_dataset(list(x_train, y_train)) train_dataset <- train_dataset |>   dataset_shuffle(buffer_size=1024) |>   dataset_batch(64)  # Prepare the validation dataset val_dataset <- tensor_slices_dataset(list(x_val, y_val)) val_dataset <- val_dataset |> dataset_batch(64)  model |> fit(train_dataset, epochs = 1, validation_data = val_dataset) ## 782/782 - 2s - 3ms/step - loss: 0.3428 - sparse_categorical_accuracy: 0.9022 - val_loss: 0.2337 - val_sparse_categorical_accuracy: 0.9291 model  <- get_compiled_model()  # Prepare the training dataset train_dataset <- tensor_slices_dataset(list(x_train, y_train)) train_dataset <- train_dataset |>   dataset_shuffle(buffer_size = 1024) |>   dataset_batch(64)  # Prepare the validation dataset val_dataset <- tensor_slices_dataset(list(x_val, y_val)) val_dataset <- val_dataset |> dataset_batch(64)  model %>% fit(   train_dataset,   epochs = 1,   # Only run validation using the first 10 batches of the dataset   # using the `validation_steps` argument   validation_data = val_dataset,   validation_steps = 10, ) ## 782/782 - 2s - 3ms/step - loss: 0.3391 - sparse_categorical_accuracy: 0.9035 - val_loss: 0.1997 - val_sparse_categorical_accuracy: 0.9391"},{"path":"https://keras3.posit.co/dev/articles/training_with_built_in_methods.html","id":"using-sample-weighting-and-class-weighting","dir":"Articles","previous_headings":"","what":"Using sample weighting and class weighting","title":"Training & evaluation with the built-in methods","text":"default settings weight sample decided frequency dataset. two methods weight data, independent sample frequency: Class weights Sample weights","code":""},{"path":"https://keras3.posit.co/dev/articles/training_with_built_in_methods.html","id":"class-weights","dir":"Articles","previous_headings":"Using sample weighting and class weighting","what":"Class weights","title":"Training & evaluation with the built-in methods","text":"set passing named list class_weight argument fit(). list maps class indices weight used samples belonging class. can used balance classes without resampling, train model gives importance particular class. instance, class “0” half represented class “1” data, use model |> fit(..., class_weight = c(\"0\" = 1, \"1\" = 0.5)). ’s R example use class weights sample weights give importance correct classification class #5 (digit “5” MNIST dataset).","code":"class_weight <- c(     \"0\" = 1.0,     \"1\" = 1.0,     \"2\" = 1.0,     \"3\" = 1.0,     \"4\" = 1.0,     # Set weight \"2\" for class \"5\",     # making this class 2x more important     \"5\" = 2.0,     \"6\" = 1.0,     \"7\" = 1.0,     \"8\" = 1.0,     \"9\" = 1.0 )  model <- get_compiled_model() model |> fit(x_train, y_train,              class_weight = class_weight,              batch_size = 64, epochs = 1) ## 782/782 - 2s - 2ms/step - loss: 0.3713 - sparse_categorical_accuracy: 0.9018"},{"path":"https://keras3.posit.co/dev/articles/training_with_built_in_methods.html","id":"sample-weights","dir":"Articles","previous_headings":"Using sample weighting and class weighting","what":"Sample weights","title":"Training & evaluation with the built-in methods","text":"fine grained control, building classifier, can use sample_weights. training R arrays: Pass sample_weight argument fit(). training tf_dataset sort iterator: yield (input_batch, label_batch, sample_weight_batch) tuples. “sample weights” array array numbers specify much weight sample batch computing total loss. commonly used imbalanced classification problems (idea give weight rarely-seen classes). weights used ones zeros, array can used mask loss function (entirely discarding contribution certain samples total loss). ’s matching Dataset example:","code":"sample_weight <- rep(1.0, length(y_train)) sample_weight[y_train == 5] <- 2.0  model <- get_compiled_model() model |> fit(   x_train, y_train,   sample_weight = sample_weight,   batch_size = 64, epochs = 1 ) ## 782/782 - 2s - 2ms/step - loss: 0.3740 - sparse_categorical_accuracy: 0.9015 sample_weight <- rep(1.0, length(y_train)) sample_weight[y_train == 5] <- 2.0  # Create a Dataset that includes sample weights # (3rd element in the return tuple). train_dataset <- tensor_slices_dataset(list(     x_train, y_train, sample_weight ))  # Shuffle and slice the dataset. train_dataset <- train_dataset |>   dataset_shuffle(buffer_size = 1024) |>   dataset_batch(64)  model <- get_compiled_model() model |> fit(train_dataset, epochs = 1) ## 782/782 - 2s - 2ms/step - loss: 0.3654 - sparse_categorical_accuracy: 0.9057"},{"path":"https://keras3.posit.co/dev/articles/training_with_built_in_methods.html","id":"passing-data-to-multi-input-multi-output-models","dir":"Articles","previous_headings":"","what":"Passing data to multi-input, multi-output models","title":"Training & evaluation with the built-in methods","text":"previous examples, considering model single input (tensor shape (764)) single output (prediction tensor shape (10)). models multiple inputs outputs? Consider following model, image input shape (32, 32, 3) (’s (height, width, channels)) time series input shape (NA, 10) (’s (timesteps, features)). model two outputs computed combination inputs: “score” (shape (1)) probability distribution five classes (shape (5)). Let’s plot model, can clearly see ’re (note shapes shown plot batch shapes, rather per-sample shapes). plot chunk unnamed-chunk-26 compilation time, can specify different losses different outputs, passing loss functions list: passed single loss function model, loss function applied every output (appropriate ). Likewise metrics: Since gave names output layers, also specify per-output losses metrics via named list: recommend use names 2 outputs. ’s possible give different weights different output-specific losses (instance, one might wish privilege “score” loss example, giving 2x importance class loss), using loss_weights argument: also choose compute loss certain outputs, outputs meant prediction training: Passing data multi-input multi-output model fit() works similar way specifying loss function compile: can pass lists arrays (1:1 mapping outputs received loss function) dicts mapping output names arrays. ’s Dataset use case: similarly R arrays, Dataset return tuple named lists (dicts).","code":"image_input <- keras_input(c(32, 32, 3), name = \"img_input\") timeseries_input <- keras_input(c(NA, 10), name = \"ts_input\")  x1 <- image_input |>   layer_conv_2d(filters = 3, kernel_size = c(3, 3)) |>   layer_global_max_pooling_2d()  x2 <- timeseries_input |>   layer_conv_1d(filters = 3, kernel_size = 3) |>   layer_global_max_pooling_1d()  x <- layer_concatenate(x1, x2)  score_output <- layer_dense(x, 1, name = \"score_output\") class_output <- layer_dense(x, 5, name = \"class_output\")  model <- keras_model(   inputs = list(image_input, timeseries_input),   outputs = list(score_output, class_output) ) plot(model, show_shapes = TRUE) model |> compile(   optimizer = optimizer_rmsprop(1e-3),   loss = list(     loss_mean_squared_error(),     loss_categorical_crossentropy()   ) ) model |> compile(   optimizer = optimizer_rmsprop(1e-3),   loss = list(     loss_mean_squared_error(),     loss_categorical_crossentropy()   ),   metrics = list(     list(       metric_mean_absolute_error(),       metric_mean_absolute_percentage_error()     ),     list(metric_categorical_accuracy())   ) ) model |> compile(   optimizer = optimizer_rmsprop(1e-3),   loss = list(     score_output = loss_mean_squared_error(),     class_output = loss_categorical_crossentropy()   ),   metrics = list(     score_output = list(       metric_mean_absolute_error(),       metric_mean_absolute_percentage_error()     ),     class_output = list(metric_categorical_accuracy())   ) ) model |> compile(   optimizer = optimizer_rmsprop(1e-3),   loss = list(     score_output = loss_mean_squared_error(),     class_output = loss_categorical_crossentropy()   ),   metrics = list(     score_output = list(       metric_mean_absolute_error(),       metric_mean_absolute_percentage_error()     ),     class_output = list(metric_categorical_accuracy())   ),   loss_weights = list(score_output = 2.0, class_output = 1.0) ) # loss list, positional version model |> compile(   optimizer = optimizer_rmsprop(1e-3),   loss = list(NULL, loss_categorical_crossentropy()) )  # Or loss list, named version model |> compile(   optimizer = optimizer_rmsprop(1e-3),   loss = list(class_output = loss_categorical_crossentropy()) ) model |> compile(   optimizer = optimizer_rmsprop(1e-3),   loss = list(     loss_mean_squared_error(),     loss_binary_crossentropy()   ) )  # Generate dummy data img_data <- random_normal(c(100, 32, 32, 3)) ts_data <- random_normal(c(100, 20, 10)) score_targets <- random_normal(c(100, 1)) class_targets <- random_normal(c(100, 5))  # Fit on unnamed lists (positional matching) model |> fit(     list(img_data, ts_data),     list(score_targets, class_targets),     batch_size=32,     epochs=1 ) ## 4/4 - 2s - 540ms/step - class_output_categorical_accuracy: 0.2300 - class_output_loss: 1.8162 - loss: 3.8764 - score_output_loss: 1.6820 - score_output_mean_absolute_error: 0.8826 - score_output_mean_absolute_percentage_error: 1826.5676 # Alternatively, fit on named lists (names matching) model <- keras_model(   inputs = list(image_input, timeseries_input),   outputs = list(score_output = score_output,                  class_output = class_output) ) |> compile(   optimizer = optimizer_rmsprop(1e-3),   loss = list(     loss_mean_squared_error(),     loss_binary_crossentropy()   ) )  model |> fit(   list(img_input = img_data, ts_input = ts_data),   list(score_output = score_targets, class_output = class_targets),   batch_size = 32,   epochs = 1 ) ## 4/4 - 2s - 392ms/step - class_output_loss: 4.1489 - loss: 4.5443 - score_output_loss: -4.1125e-01 train_dataset <- tensor_slices_dataset(list(   list(img_input = img_data, ts_input = ts_data),   list(score_output = score_targets, class_output = class_targets) )) train_dataset <- train_dataset |>   dataset_shuffle(buffer_size = 1024) |>   dataset_batch(64)  model |> fit(train_dataset, epochs = 1) ## 2/2 - 1s - 706ms/step - class_output_loss: 3.8547 - loss: 3.5725 - score_output_loss: -1.5209e-01"},{"path":"https://keras3.posit.co/dev/articles/training_with_built_in_methods.html","id":"using-callbacks","dir":"Articles","previous_headings":"","what":"Using callbacks","title":"Training & evaluation with the built-in methods","text":"Callbacks Keras objects called different points training (start epoch, end batch, end epoch, etc.). can used implement certain behaviors, : validation different points training (beyond built-per-epoch validation) Checkpointing model regular intervals exceeds certain accuracy threshold Changing learning rate model training seems plateauing fine-tuning top layers training seems plateauing Sending email instant message notifications training ends certain performance threshold exceeded Etc. Callbacks can passed list call fit():","code":"model <- get_compiled_model()  callbacks <- list(   callback_early_stopping(     # Stop training when `val_loss` is no longer improving     monitor = \"val_loss\",     # \"no longer improving\" being defined as \"no better than 1e-2 less\"     min_delta = 1e-2,     # \"no longer improving\" being further defined as \"for at least 2 epochs\"     patience = 2,     verbose = 1   ) ) model |> fit(   x_train,   y_train,   epochs = 20,   batch_size = 64,   callbacks = callbacks,   validation_split = 0.2, ) ## Epoch 1/20 ## 625/625 - 2s - 3ms/step - loss: 0.3695 - sparse_categorical_accuracy: 0.8961 - val_loss: 0.1873 - val_sparse_categorical_accuracy: 0.9469 ## Epoch 2/20 ## 625/625 - 1s - 1ms/step - loss: 0.1752 - sparse_categorical_accuracy: 0.9488 - val_loss: 0.1405 - val_sparse_categorical_accuracy: 0.9576 ## Epoch 3/20 ## 625/625 - 1s - 1ms/step - loss: 0.1277 - sparse_categorical_accuracy: 0.9626 - val_loss: 0.1217 - val_sparse_categorical_accuracy: 0.9644 ## Epoch 4/20 ## 625/625 - 1s - 1ms/step - loss: 0.1009 - sparse_categorical_accuracy: 0.9702 - val_loss: 0.1138 - val_sparse_categorical_accuracy: 0.9667 ## Epoch 5/20 ## 625/625 - 1s - 1ms/step - loss: 0.0823 - sparse_categorical_accuracy: 0.9758 - val_loss: 0.1086 - val_sparse_categorical_accuracy: 0.9679 ## Epoch 6/20 ## 625/625 - 1s - 1ms/step - loss: 0.0682 - sparse_categorical_accuracy: 0.9800 - val_loss: 0.1084 - val_sparse_categorical_accuracy: 0.9685 ## Epoch 7/20 ## 625/625 - 1s - 1ms/step - loss: 0.0573 - sparse_categorical_accuracy: 0.9834 - val_loss: 0.1100 - val_sparse_categorical_accuracy: 0.9698 ## Epoch 8/20 ## 625/625 - 1s - 1ms/step - loss: 0.0485 - sparse_categorical_accuracy: 0.9868 - val_loss: 0.1113 - val_sparse_categorical_accuracy: 0.9702 ## Epoch 8: early stopping"},{"path":"https://keras3.posit.co/dev/articles/training_with_built_in_methods.html","id":"many-built-in-callbacks-are-available","dir":"Articles","previous_headings":"Using callbacks","what":"Many built-in callbacks are available","title":"Training & evaluation with the built-in methods","text":"many built-callbacks already available Keras, : callback_model_checkpoint(): Periodically save model. callback_early_stopping(): Stop training training longer improving validation metrics. callback_tensorboard(): periodically write model logs can visualized TensorBoard (details section “Visualization”). callback_csv_logger(): streams loss metrics data CSV file. etc. See callbacks documentation complete list.","code":""},{"path":"https://keras3.posit.co/dev/articles/training_with_built_in_methods.html","id":"writing-your-own-callback","dir":"Articles","previous_headings":"Using callbacks","what":"Writing your own callback","title":"Training & evaluation with the built-in methods","text":"can create custom callback subclassing base [Callback()] class. callback access associated model class property self$model. Make sure read complete guide writing custom callbacks. ’s simple example saving list per-batch loss values training:","code":"callback_loss_history <- Callback(   classname = \"LossHistory\",   initialize = function(file = \"per_training_batch_losses.txt\", ...) {     super$initialize(...)     private$file <- file   },   on_train_begin = function(logs = NULL) {     private$per_batch_losses <- fastmap::faststack()   },   on_train_batch_begin = function(batch, logs = NULL) {     private$per_batch_losses$push(logs$loss)   },   on_train_end = function(logs = NULL) {     per_batch_losses <- private$per_batch_losses$as_list() |> as.numeric()     write(per_batch_losses, private$file)   } )"},{"path":"https://keras3.posit.co/dev/articles/training_with_built_in_methods.html","id":"checkpointing-models","dir":"Articles","previous_headings":"","what":"Checkpointing models","title":"Training & evaluation with the built-in methods","text":"’re training model relatively large datasets, ’s crucial save checkpoints model frequent intervals. easiest way achieve [callback_model_checkpoint()]: ModelCheckpoint callback can used implement fault-tolerance: ability restart training last saved state model case training gets randomly interrupted. ’s basic example: call also write callback saving restoring models. complete guide serialization saving, see guide saving serializing Models.","code":"model <- get_compiled_model()  callbacks <- list(   callback_model_checkpoint(     # Path where to save the model     # The two parameters below mean that we will overwrite     # the current checkpoint if and only if     # the `val_loss` score has improved.     # The saved model name will include the current epoch.     filepath = \"mymodel_{epoch}.keras\",     save_best_only = TRUE,     # Only save a model if `val_loss` has improved.     monitor = \"val_loss\",     verbose = 1   ) ) model |> fit(   x_train, y_train,   epochs = 2, batch_size = 64,   callbacks = callbacks,   validation_split = 0.2 ) ## Epoch 1/2 ## ## Epoch 1: val_loss improved from None to 0.19488, saving model to mymodel_1.keras ## 625/625 - 2s - 3ms/step - loss: 0.3793 - sparse_categorical_accuracy: 0.8936 - val_loss: 0.1949 - val_sparse_categorical_accuracy: 0.9447 ## Epoch 2/2 ## ## Epoch 2: val_loss improved from 0.19488 to 0.14489, saving model to mymodel_2.keras ## 625/625 - 1s - 1ms/step - loss: 0.1777 - sparse_categorical_accuracy: 0.9475 - val_loss: 0.1449 - val_sparse_categorical_accuracy: 0.9590 # Prepare a directory to store all the checkpoints. checkpoint_dir <- \"./ckpt\" fs::dir_create(checkpoint_dir)  make_or_restore_model <- function() {   # Either restore the latest (best) model, or create a fresh one   # if there is no checkpoint available.   checkpoints <- Sys.glob(file.path(checkpoint_dir, \"model-loss=*.keras\"))    if (length(checkpoints) > 0) {     checkpoint_losses <- sub(\"^model-loss=([0-9.]+)\\\\.keras$\", \"\\\\1\",                              basename(checkpoints)) |> as.numeric()     best_checkpoint <- checkpoints[which.min(checkpoint_losses)]     load_model(best_checkpoint)   } else {     get_compiled_model()   } }  model <- make_or_restore_model() callbacks <- list(   # This callback saves the model every 100 batches.   # We include the training loss in the saved model name.   callback_model_checkpoint(     filepath = file.path(checkpoint_dir, \"model-loss={loss:.2f}.keras\"),     save_freq = 100   ) ) model |> fit(x_train, y_train, epochs = 1, callbacks = callbacks) ## 1563/1563 - 3s - 2ms/step - loss: 0.2932 - sparse_categorical_accuracy: 0.9145"},{"path":"https://keras3.posit.co/dev/articles/training_with_built_in_methods.html","id":"using-learning-rate-schedules","dir":"Articles","previous_headings":"","what":"Using learning rate schedules","title":"Training & evaluation with the built-in methods","text":"common pattern training deep learning models gradually reduce learning training progresses. generally known “learning rate decay”. learning decay schedule static (fixed advance, function current epoch current batch index), dynamic (responding current behavior model, particular validation loss).","code":""},{"path":"https://keras3.posit.co/dev/articles/training_with_built_in_methods.html","id":"passing-a-schedule-to-an-optimizer","dir":"Articles","previous_headings":"Using learning rate schedules","what":"Passing a schedule to an optimizer","title":"Training & evaluation with the built-in methods","text":"can easily use static learning rate decay schedule passing schedule object learning_rate argument optimizer: Several built-schedules available: ExponentialDecay, PiecewiseConstantDecay, PolynomialDecay, InverseTimeDecay.","code":"initial_learning_rate <- 0.1 lr_schedule <- learning_rate_schedule_exponential_decay(     initial_learning_rate, decay_steps=100000, decay_rate=0.96,     staircase=TRUE )  optimizer <- optimizer_rmsprop(learning_rate = lr_schedule)"},{"path":"https://keras3.posit.co/dev/articles/training_with_built_in_methods.html","id":"using-callbacks-to-implement-a-dynamic-learning-rate-schedule","dir":"Articles","previous_headings":"Using learning rate schedules","what":"Using callbacks to implement a dynamic learning rate schedule","title":"Training & evaluation with the built-in methods","text":"dynamic learning rate schedule (instance, decreasing learning rate validation loss longer improving) achieved schedule objects, since optimizer access validation metrics. However, callbacks access metrics, including validation metrics! can thus achieve pattern using callback modifies current learning rate optimizer. fact, even built-[callback_reduce_lr_on_plateau()].","code":""},{"path":"https://keras3.posit.co/dev/articles/training_with_built_in_methods.html","id":"visualizing-loss-and-metrics-during-training-with-tensorboard","dir":"Articles","previous_headings":"","what":"Visualizing loss and metrics during training with TensorBoard","title":"Training & evaluation with the built-in methods","text":"best way keep eye model training use TensorBoard – browser-based application can run locally provides : Live plots loss metrics training evaluation (optionally) Visualizations histograms layer activations (optionally) 3D visualizations embedding spaces learned layer_embedding() installed TensorFlow pip, able launch TensorBoard command line: R using:","code":"tensorboard --logdir=/full_path_to_your_logs tensorflow::tensorboard(logdir = \"/full_path_to_your_logs\")"},{"path":"https://keras3.posit.co/dev/articles/training_with_built_in_methods.html","id":"using-the-tensorboard-callback","dir":"Articles","previous_headings":"Visualizing loss and metrics during training with TensorBoard","what":"Using the TensorBoard callback","title":"Training & evaluation with the built-in methods","text":"easiest way use TensorBoard Keras model fit() method [callback_tensorboard()]. simplest case, just specify want callback write logs, ’re good go: information, see callback_tensorboard().","code":"tb_callback <- callback_tensorboard(   log_dir = \"/full_path_to_your_logs\",   histogram_freq = 0, # How often to log histogram visualizations   embeddings_freq = 0, # How often to log embedding visualizations   update_freq = \"epoch\", # How often to write logs (default: once per epoch) )"},{"path":"https://keras3.posit.co/dev/articles/transfer_learning.html","id":"setup","dir":"Articles","previous_headings":"","what":"Setup","title":"Transfer learning & fine-tuning","text":"","code":"library(keras3)"},{"path":"https://keras3.posit.co/dev/articles/transfer_learning.html","id":"introduction","dir":"Articles","previous_headings":"","what":"Introduction","title":"Transfer learning & fine-tuning","text":"Transfer learning consists taking features learned one problem, leveraging new, similar problem. instance, features model learned identify raccoon may useful kick-start model meant identify tanukis. Transfer learning usually done tasks dataset little data train full-scale model scratch. common incarnation transfer learning context deep learning following workflow: Take layers previously trained model. Freeze , avoid destroying information contain future training rounds. Add new, trainable layers top frozen layers. learn turn old features predictions new dataset. Train new layers dataset. last, optional step, fine-tuning, consists unfreezing entire model obtained (part ), re-training new data low learning rate. can potentially achieve meaningful improvements, incrementally adapting pretrained features new data. First, go Keras trainable API detail, underlies transfer learning & fine-tuning workflows. , ’ll demonstrate typical workflow taking model pretrained ImageNet dataset, retraining Kaggle “cats vs dogs” classification dataset. adapted Deep Learning R 2016 blog post “building powerful image classification models using little data”.","code":""},{"path":"https://keras3.posit.co/dev/articles/transfer_learning.html","id":"freezing-layers-understanding-the-trainable-attribute","dir":"Articles","previous_headings":"","what":"Freezing layers: understanding the trainable attribute","title":"Transfer learning & fine-tuning","text":"Layers & models three weight attributes: weights list weights variables layer. trainable_weights list meant updated (via gradient descent) minimize loss training. non_trainable_weights list aren’t meant trained. Typically updated model forward pass. Example: Dense layer 2 trainable weights (kernel & bias) general, weights trainable weights. built-layer non-trainable weights [layer_batch_normalization()]. uses non-trainable weights keep track mean variance inputs training. learn use non-trainable weights custom layers, see guide writing new layers scratch. Example: BatchNormalization layer 2 trainable weights 2 non-trainable weights Layers & models also feature boolean attribute trainable. value can changed. Setting layer$trainable FALSE moves layer’s weights trainable non-trainable. called “freezing” layer: state frozen layer won’t updated training (either training fit() training custom loop relies trainable_weights apply gradient updates). Example: setting trainable False trainable weight becomes non-trainable, value longer updated training. confuse layer$trainable attribute argument training layer$call() (controls whether layer run forward pass inference mode training mode). information, see Keras FAQ.","code":"layer <- layer_dense(units = 3) layer$build(shape(NULL, 4))  # Create the weights  length(layer$weights) ## [1] 2 length(layer$trainable_weights) ## [1] 2 length(layer$non_trainable_weights) ## [1] 0 layer <- layer_batch_normalization() layer$build(shape(NA, 4))  # Create the weights  length(layer$weights) ## [1] 4 length(layer$trainable_weights) ## [1] 2 length(layer$non_trainable_weights) ## [1] 2 layer <- layer_dense(units = 3) layer$build(shape(NULL, 4))  # Create the weights layer$trainable <- FALSE  # Freeze the layer  length(layer$weights) ## [1] 2 length(layer$trainable_weights) ## [1] 0 length(layer$non_trainable_weights) ## [1] 2 # Make a model with 2 layers layer1 <- layer_dense(units = 3, activation = \"relu\") layer2 <- layer_dense(units = 3, activation = \"sigmoid\") model <- keras_model_sequential(input_shape = 3) |>   layer1() |>   layer2()  # Freeze the first layer layer1$trainable <- FALSE  # Keep a copy of the weights of layer1 for later reference # (get_weights() returns a list of R arrays, #  layer$weights returns a list of KerasVariables) initial_layer1_weights_values <- get_weights(layer1)  # Train the model model |> compile(optimizer = \"adam\", loss = \"mse\") model |> fit(random_normal(c(2, 3)), random_normal(c(2, 3)), epochs = 1) ## 1/1 - 0s - 436ms/step - loss: 2.8362 # Check that the weights of layer1 have not changed during training final_layer1_weights_values <- get_weights(layer1)  all.equal(initial_layer1_weights_values,           final_layer1_weights_values) ## [1] TRUE"},{"path":"https://keras3.posit.co/dev/articles/transfer_learning.html","id":"recursive-setting-of-the-trainable-attribute","dir":"Articles","previous_headings":"","what":"Recursive setting of the trainable attribute","title":"Transfer learning & fine-tuning","text":"set $trainable <- FALSE model layer sublayers, children layers become non-trainable well. Example:","code":"inner_model <- keras_model_sequential(input_shape = 3) |>   layer_dense(units = 3, activation = \"relu\") |>   layer_dense(units = 3, activation = \"relu\")  model <- keras_model_sequential(input_shape = 3) |>   inner_model() |>   layer_dense(units = 3, activation = \"sigmoid\")  model$trainable <- FALSE  # Freeze the outer model  inner_model$trainable  # All layers in `model` are now frozen ## [1] FALSE inner_model$layers[[1]]$trainable  # `trainable` is propagated recursively ## [1] FALSE"},{"path":"https://keras3.posit.co/dev/articles/transfer_learning.html","id":"the-typical-transfer-learning-workflow","dir":"Articles","previous_headings":"","what":"The typical transfer-learning workflow","title":"Transfer learning & fine-tuning","text":"leads us typical transfer learning workflow can implemented Keras: Instantiate base model load pre-trained weights . Freeze layers base model setting trainable <- FALSE. Create new model top output one (several) layers base model. Train new model new dataset. Note alternative, lightweight workflow also : Instantiate base model load pre-trained weights . Run new dataset record output one (several) layers base model. called feature extraction. Use output input data new, smaller model. key advantage second workflow run base model data, rather per epoch training. ’s lot faster & cheaper. issue second workflow, though, doesn’t allow dynamically modify input data new model training, required data augmentation, instance. Transfer learning typically used tasks new dataset little data train full-scale model scratch, scenarios data augmentation important. follows, focus first workflow. ’s first workflow looks like Keras: First, instantiate base model pre-trained weights. , freeze base model. Create new model top. Train model new data.","code":"base_model <- application_xception(   weights = 'imagenet', # Load weights pre-trained on ImageNet.   input_shape = c(150, 150, 3),   include_top = FALSE  # Do not include the ImageNet classifier at the top. ) base_model$trainable <- FALSE inputs <- keras_input(shape = c(150, 150, 3)) # We make sure that the base_model is running in inference mode here, # by passing `training <- FALSE`. This is important for fine-tuning, as you will # learn in a few paragraphs. outputs <- inputs |>   base_model(training = FALSE) |>   # Convert features of shape `base_model$output_shape[-1]` to vectors   layer_global_average_pooling_2d() |>   # A Dense classifier with a single unit (binary classification)   layer_dense(1)  model <- keras_model(inputs, outputs) model |> compile(   optimizer = optimizer_adam(),   loss = loss_binary_crossentropy(from_logits = TRUE),   metrics = list(metric_binary_accuracy()) ) model |> fit(new_dataset, epochs = 20,              callbacks = ..., validation_data = ...)"},{"path":"https://keras3.posit.co/dev/articles/transfer_learning.html","id":"fine-tuning","dir":"Articles","previous_headings":"","what":"Fine-tuning","title":"Transfer learning & fine-tuning","text":"model converged new data, can try unfreeze part base model retrain whole model end--end low learning rate. optional last step can potentially give incremental improvements. also potentially lead quick overfitting – keep mind. critical step model frozen layers trained convergence. mix randomly-initialized trainable layers trainable layers hold pre-trained features, randomly-initialized layers cause large gradient updates training, destroy pre-trained features. ’s also critical use low learning rate stage, training much larger model first round training, dataset typically small. result, risk overfitting quickly apply large weight updates. , want readapt pretrained weights incremental way. implement fine-tuning whole base model: Important note compile() trainable Calling compile() model meant “freeze” behavior model. implies trainable attribute values time model compiled preserved throughout lifetime model, compile called . Hence, change trainable value, make sure call compile() model changes taken account. Important notes BatchNormalization layer Many image models contain BatchNormalization layers. layer special case every imaginable count. things keep mind. BatchNormalization contains 2 non-trainable weights get updated training. variables tracking mean variance inputs. set bn_layer$trainable <- FALSE, BatchNormalization layer run inference mode, update mean & variance statistics. case layers general, weight trainability & inference/training modes two orthogonal concepts. two tied case BatchNormalization layer. unfreeze model contains BatchNormalization layers order fine-tuning, keep BatchNormalization layers inference mode passing training = FALSE calling base model. Otherwise updates applied non-trainable weights suddenly destroy model learned. ’ll see pattern action end--end example end guide.","code":"# Unfreeze the base model base_model$trainable <- TRUE  # It's important to recompile your model after you make any changes # to the `trainable` attribute of any inner layer, so that your changes # are take into account model |> compile(   optimizer = optimizer_adam(1e-5), # Very low learning rate   loss = loss_binary_crossentropy(from_logits = TRUE),   metrics = c(metric_binary_accuracy()) )  # Train end-to-end. Be careful to stop before you overfit! model |> fit(new_dataset, epochs = 10,              callbacks = ..., validation_data = ...)"},{"path":"https://keras3.posit.co/dev/articles/transfer_learning.html","id":"an-end-to-end-example-fine-tuning-an-image-classification-model-on-a-cats-vs--dogs-dataset","dir":"Articles","previous_headings":"","what":"An end-to-end example: fine-tuning an image classification model on a cats vs. dogs dataset","title":"Transfer learning & fine-tuning","text":"solidify concepts, let’s walk concrete end--end transfer learning & fine-tuning example. load Xception model, pre-trained ImageNet, use Kaggle “cats vs. dogs” classification dataset.","code":""},{"path":"https://keras3.posit.co/dev/articles/transfer_learning.html","id":"getting-the-data","dir":"Articles","previous_headings":"An end-to-end example: fine-tuning an image classification model on a cats vs. dogs dataset","what":"Getting the data","title":"Transfer learning & fine-tuning","text":"First, let’s fetch cats vs. dogs dataset using TFDS. dataset, ’ll probably want use utility [image_dataset_from_directory()] generate similar labeled dataset objects set images disk filed class-specific folders. Transfer learning useful working small datasets. keep dataset small, use 40% original training data (25,000 images) training, 10% validation, 10% testing. first 9 images training dataset – can see, ’re different sizes. can also see label 1 “dog” label 0 “cat”.","code":"reticulate::py_require(\"tensorflow-datasets\") tfds <- reticulate::import(\"tensorflow_datasets\")  c(train_ds, validation_ds, test_ds) %<-% tfds$load(   \"cats_vs_dogs\",   # Reserve 10% for validation and 10% for test   split = c(\"train[:40%]\", \"train[40%:50%]\", \"train[50%:60%]\"),   as_supervised = TRUE  # Include labels )  length(train_ds) ## [1] 9305 library(tfdatasets, exclude = \"shape\")  par(mfrow = c(3, 3), mar = c(1,0,1.5,0)) train_ds |>   dataset_take(9) |>   as_array_iterator() |>   iterate(function(batch) {     c(image, label) %<-% batch     plot(as.raster(image, max = 255L))     title(sprintf(       \"label: %s   size: %s\",       label, paste(dim(image), collapse = \" x \")))   })"},{"path":"https://keras3.posit.co/dev/articles/transfer_learning.html","id":"standardizing-the-data","dir":"Articles","previous_headings":"An end-to-end example: fine-tuning an image classification model on a cats vs. dogs dataset","what":"Standardizing the data","title":"Transfer learning & fine-tuning","text":"raw images variety sizes. addition, pixel consists 3 integer values 0 255 (RGB level values). isn’t great fit feeding neural network. need 2 things: Standardize fixed image size. pick 150x150. Normalize pixel values -1 1. ’ll using Normalization layer part model . general, ’s good practice develop models take raw data input, opposed models take already-preprocessed data. reason , model expects preprocessed data, time export model use elsewhere (web browser, mobile app), ’ll need reimplement exact preprocessing pipeline. gets tricky quickly. least possible amount preprocessing hitting model. , ’ll image resizing data pipeline (deep neural network can process contiguous batches data), ’ll input value scaling part model, create . Let’s resize images 150x150:","code":"resize_fn <- layer_resizing(width = 150, height = 150) resize_pair <- function(x, y) list(resize_fn(x), y)  train_ds <- train_ds |> dataset_map(resize_pair) validation_ds <- validation_ds |> dataset_map(resize_pair) test_ds <- test_ds |> dataset_map(resize_pair)"},{"path":"https://keras3.posit.co/dev/articles/transfer_learning.html","id":"using-random-data-augmentation","dir":"Articles","previous_headings":"An end-to-end example: fine-tuning an image classification model on a cats vs. dogs dataset","what":"Using random data augmentation","title":"Transfer learning & fine-tuning","text":"don’t large image dataset, ’s good practice artificially introduce sample diversity applying random yet realistic transformations training images, random horizontal flipping small random rotations. helps expose model different aspects training data slowing overfitting. Let’s batch data use prefetching optimize loading speed. Let’s visualize first image first batch looks like various random transformations:","code":"data_augmentation <- keras_model_sequential() |>   layer_random_flip(\"horizontal\") |>   layer_random_rotation(.1)  train_ds <- train_ds %>%   dataset_map(function(x, y) list(data_augmentation(x), y)) library(tensorflow, exclude = c(\"shape\", \"set_random_seed\")) batch_size <- 64  train_ds <- train_ds |>   dataset_batch(batch_size) |>   dataset_prefetch()  validation_ds <- validation_ds |>   dataset_batch(batch_size) |>   dataset_prefetch()  test_ds <- test_ds |>   dataset_batch(batch_size) |>   dataset_prefetch() batch <- train_ds |>   dataset_take(1) |>   as_iterator() |>   iter_next()  c(images, labels) %<-% batch first_image <- images[1, all_dims(), drop = TRUE] augmented_image <- data_augmentation(first_image, training = TRUE)  plot_image <- function(image, main = deparse1(substitute(image))) {   image |>     as.array() |> # convert from tensor to R array     as.raster(max = 255) |>     plot()    if(!is.null(main))     title(main) }  par(mfrow = c(2, 2), mar = c(1, 1, 1.5, 1)) plot_image(first_image) plot_image(augmented_image) plot_image(data_augmentation(first_image, training = TRUE), \"augmented 2\") plot_image(data_augmentation(first_image, training = TRUE), \"augmented 3\")"},{"path":"https://keras3.posit.co/dev/articles/transfer_learning.html","id":"build-a-model","dir":"Articles","previous_headings":"","what":"Build a model","title":"Transfer learning & fine-tuning","text":"Now let’s built model follows blueprint ’ve explained earlier. Note : add Rescaling layer scale input values (initially [0, 255] range) [-1, 1] range. add Dropout layer classification layer, regularization. make sure pass training=FALSE calling base model, runs inference mode, batchnorm statistics don’t get updated even unfreeze base model fine-tuning.","code":"base_model <- application_xception(   weights = \"imagenet\", # Load weights pre-trained on ImageNet.   input_shape = c(150, 150, 3),   include_top = FALSE, # Do not include the ImageNet classifier at the top. )  # Freeze the base_model base_model$trainable <- FALSE  # Create new model on top inputs <- keras_input(shape = c(150, 150, 3))  # Pre-trained Xception weights requires that input be scaled # from (0, 255) to a range of (-1., +1.), the rescaling layer # outputs: `(inputs * scale) + offset` scale_layer <- layer_rescaling(scale = 1 / 127.5, offset = -1) x <- scale_layer(inputs)  # The base model contains batchnorm layers. We want to keep them in inference mode # when we unfreeze the base model for fine-tuning, so we make sure that the # base_model is running in inference mode here. outputs <- x |>   base_model(training = FALSE) |>   layer_global_average_pooling_2d() |>   layer_dropout(0.2) |>   layer_dense(1)  model <- keras_model(inputs, outputs)  summary(model, show_trainable = TRUE) ## Model: \"functional_8\" ## ┏━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━┳━━━━━━━┓ ## ┃ Layer (type)                ┃ Output Shape          ┃    Param # ┃ Trai… ┃ ## ┡━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━╇━━━━━━━┩ ## │ input_layer_7 (InputLayer)  │ (None, 150, 150, 3)   │          0 │   -   │ ## ├─────────────────────────────┼───────────────────────┼────────────┼───────┤ ## │ rescaling (Rescaling)       │ (None, 150, 150, 3)   │          0 │   -   │ ## ├─────────────────────────────┼───────────────────────┼────────────┼───────┤ ## │ xception (Functional)       │ (None, 5, 5, 2048)    │ 20,861,480 │   N   │ ## ├─────────────────────────────┼───────────────────────┼────────────┼───────┤ ## │ global_average_pooling2d_1  │ (None, 2048)          │          0 │   -   │ ## │ (GlobalAveragePooling2D)    │                       │            │       │ ## ├─────────────────────────────┼───────────────────────┼────────────┼───────┤ ## │ dropout (Dropout)           │ (None, 2048)          │          0 │   -   │ ## ├─────────────────────────────┼───────────────────────┼────────────┼───────┤ ## │ dense_8 (Dense)             │ (None, 1)             │      2,049 │   Y   │ ## └─────────────────────────────┴───────────────────────┴────────────┴───────┘ ##  Total params: 20,863,529 (79.59 MB) ##  Trainable params: 2,049 (8.00 KB) ##  Non-trainable params: 20,861,480 (79.58 MB)"},{"path":"https://keras3.posit.co/dev/articles/transfer_learning.html","id":"train-the-top-layer","dir":"Articles","previous_headings":"","what":"Train the top layer","title":"Transfer learning & fine-tuning","text":"","code":"model |> compile(   optimizer = optimizer_adam(),   loss = loss_binary_crossentropy(from_logits = TRUE),   metrics = list(metric_binary_accuracy()) )  epochs <- 1 model |> fit(train_ds, epochs = epochs, validation_data = validation_ds) ## 146/146 - 57s - 389ms/step - binary_accuracy: 0.9183 - loss: 0.1887 - val_binary_accuracy: 0.9669 - val_loss: 0.0926"},{"path":"https://keras3.posit.co/dev/articles/transfer_learning.html","id":"do-a-round-of-fine-tuning-of-the-entire-model","dir":"Articles","previous_headings":"","what":"Do a round of fine-tuning of the entire model","title":"Transfer learning & fine-tuning","text":"Finally, let’s unfreeze base model train entire model end--end low learning rate. Importantly, although base model becomes trainable, still running inference mode since passed training=FALSE calling built model. means batch normalization layers inside won’t update batch statistics. , wreck havoc representations learned model far. 10 epochs, fine-tuning gains us nice improvement . Let’s evaluate model test dataset:","code":"# Unfreeze the base_model. Note that it keeps running in inference mode # since we passed `training=FALSE` when calling it. This means that # the batchnorm layers will not update their batch statistics. # This prevents the batchnorm layers from undoing all the training # we've done so far. base_model$trainable <- TRUE summary(model, show_trainable = TRUE) ## Model: \"functional_8\" ## ┏━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━┳━━━━━━━┓ ## ┃ Layer (type)                ┃ Output Shape          ┃    Param # ┃ Trai… ┃ ## ┡━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━╇━━━━━━━┩ ## │ input_layer_7 (InputLayer)  │ (None, 150, 150, 3)   │          0 │   -   │ ## ├─────────────────────────────┼───────────────────────┼────────────┼───────┤ ## │ rescaling (Rescaling)       │ (None, 150, 150, 3)   │          0 │   -   │ ## ├─────────────────────────────┼───────────────────────┼────────────┼───────┤ ## │ xception (Functional)       │ (None, 5, 5, 2048)    │ 20,861,480 │   Y   │ ## ├─────────────────────────────┼───────────────────────┼────────────┼───────┤ ## │ global_average_pooling2d_1  │ (None, 2048)          │          0 │   -   │ ## │ (GlobalAveragePooling2D)    │                       │            │       │ ## ├─────────────────────────────┼───────────────────────┼────────────┼───────┤ ## │ dropout (Dropout)           │ (None, 2048)          │          0 │   -   │ ## ├─────────────────────────────┼───────────────────────┼────────────┼───────┤ ## │ dense_8 (Dense)             │ (None, 1)             │      2,049 │   Y   │ ## └─────────────────────────────┴───────────────────────┴────────────┴───────┘ ##  Total params: 20,867,629 (79.60 MB) ##  Trainable params: 20,809,001 (79.38 MB) ##  Non-trainable params: 54,528 (213.00 KB) ##  Optimizer params: 4,100 (16.02 KB) model |> compile(   optimizer = optimizer_adam(1e-5), # Low learning rate   loss = loss_binary_crossentropy(from_logits = TRUE),   metrics = list(metric_binary_accuracy()) )  epochs <- 1 model |> fit(train_ds, epochs = epochs, validation_data = validation_ds) ## 146/146 - 86s - 592ms/step - binary_accuracy: 0.8660 - loss: 0.3212 - val_binary_accuracy: 0.9652 - val_loss: 0.1022 model |> evaluate(test_ds) ## 37/37 - 2s - 43ms/step - binary_accuracy: 0.9540 - loss: 0.1102 ## $binary_accuracy ## [1] 0.9539983 ## ## $loss ## [1] 0.1102476"},{"path":"https://keras3.posit.co/dev/articles/understanding_masking_and_padding.html","id":"setup","dir":"Articles","previous_headings":"","what":"Setup","title":"Understanding masking & padding","text":"","code":"library(keras3)"},{"path":"https://keras3.posit.co/dev/articles/understanding_masking_and_padding.html","id":"introduction","dir":"Articles","previous_headings":"","what":"Introduction","title":"Understanding masking & padding","text":"Masking way tell sequence-processing layers certain timesteps input missing, thus skipped processing data. Padding special form masking masked steps start end sequence. Padding comes need encode sequence data contiguous batches: order make sequences batch fit given standard length, necessary pad truncate sequences. Let’s take close look.","code":""},{"path":"https://keras3.posit.co/dev/articles/understanding_masking_and_padding.html","id":"padding-sequence-data","dir":"Articles","previous_headings":"","what":"Padding sequence data","title":"Understanding masking & padding","text":"processing sequence data, common individual samples different lengths. Consider following example (text tokenized words): vocabulary lookup, data might vectorized integers, e.g.: data nested list individual samples length 3, 5, 6, respectively. Since input data deep learning model must single tensor (shape e.g. (batch_size, 6, vocab_size) case), samples shorter longest item need padded placeholder value (alternatively, one might also truncate long samples padding short samples). Keras provides utility function truncate pad Python lists common length: pad_sequences.","code":"data <- list(   c(\"Hello\", \"world\", \"!\"),   c(\"How\", \"are\", \"you\", \"doing\", \"today\"),   c(\"The\", \"weather\", \"will\", \"be\", \"nice\", \"tomorrow\") ) data <- list(   c(71, 1331, 4231),   c(73, 8, 3215, 55, 927),   c(83, 91, 1, 645, 1253, 927) ) raw_inputs <- list(   c(711, 632, 71),   c(73, 8, 3215, 55, 927),   c(83, 91, 1, 645, 1253, 927) )  # By default, this will pad using 0s; it is configurable via the # \"value\" parameter. # Note that you could use \"pre\" padding (at the beginning) or # \"post\" padding (at the end). # We recommend using \"post\" padding when working with RNN layers # (in order to be able to use the # CuDNN implementation of the layers). padded_inputs <- pad_sequences(raw_inputs, padding=\"post\") padded_inputs ##      [,1] [,2] [,3] [,4] [,5] [,6] ## [1,]  711  632   71    0    0    0 ## [2,]   73    8 3215   55  927    0 ## [3,]   83   91    1  645 1253  927"},{"path":"https://keras3.posit.co/dev/articles/understanding_masking_and_padding.html","id":"masking","dir":"Articles","previous_headings":"","what":"Masking","title":"Understanding masking & padding","text":"Now samples uniform length, model must informed part data actually padding ignored. mechanism masking. three ways introduce input masks Keras models: Add layer_masking layer. Configure layer_embedding layer mask_zero=TRUE. Pass mask argument manually calling layers support argument (e.g. RNN layers).","code":""},{"path":"https://keras3.posit.co/dev/articles/understanding_masking_and_padding.html","id":"mask-generating-layers-embedding-and-masking","dir":"Articles","previous_headings":"","what":"Mask-generating layers: Embedding and Masking","title":"Understanding masking & padding","text":"hood, layers create mask tensor (2D tensor shape (batch, sequence_length)), attach tensor output returned Masking Embedding layer. can see printed result, mask 2D boolean tensor shape (batch_size, sequence_length), individual FALSE entry indicates corresponding timestep ignored processing.","code":"embedding <- layer_embedding(input_dim=5000, output_dim=16, mask_zero=TRUE) masked_output <- embedding(padded_inputs)  masked_output$`_keras_mask` ## tf.Tensor( ## [[ True  True  True False False False] ##  [ True  True  True  True  True False] ##  [ True  True  True  True  True  True]], shape=(3, 6), dtype=bool) masking_layer <- layer_masking() # Simulate the embedding lookup by expanding the 2D input to 3D, # with embedding dimension of 10. unmasked_embedding <- op_cast(     op_tile(op_expand_dims(padded_inputs, axis=-1), c(1L, 1L, 10L)),     dtype=\"float32\" )  masked_embedding <- masking_layer(unmasked_embedding) masked_embedding$`_keras_mask` ## tf.Tensor( ## [[ True  True  True False False False] ##  [ True  True  True  True  True False] ##  [ True  True  True  True  True  True]], shape=(3, 6), dtype=bool)"},{"path":"https://keras3.posit.co/dev/articles/understanding_masking_and_padding.html","id":"mask-propagation-in-the-functional-api-and-sequential-api","dir":"Articles","previous_headings":"","what":"Mask propagation in the Functional API and Sequential API","title":"Understanding masking & padding","text":"using Functional API Sequential API, mask generated Embedding Masking layer propagated network layer capable using (example, RNN layers). Keras automatically fetch mask corresponding input pass layer knows use . instance, following Sequential model, LSTM layer automatically receive mask, means ignore padded values: also case following Functional API model:","code":"model <- keras_model_sequential() %>%   layer_embedding(input_dim=5000, output_dim=16, mask_zero=TRUE) %>%   layer_lstm(units=32) inputs <- keras_input(shape = shape(NULL), dtype=\"int32\") outputs <- inputs %>%   layer_embedding(input_dim=5000, output_dim=16, mask_zero=TRUE) %>%   layer_lstm(units=32)  model <- keras_model(inputs, outputs)"},{"path":"https://keras3.posit.co/dev/articles/understanding_masking_and_padding.html","id":"passing-mask-tensors-directly-to-layers","dir":"Articles","previous_headings":"","what":"Passing mask tensors directly to layers","title":"Understanding masking & padding","text":"Layers can handle masks (LSTM layer) mask argument call method. Meanwhile, layers produce mask (e.g. Embedding) expose compute_mask(input, previous_mask) method can call. Thus, can pass output compute_mask() method mask-producing layer call method mask-consuming layer, like :","code":"MyLayer <- new_layer_class(   \"MyLayer\",   initialize = function(...) {     super$initialize(...)     self$embedding <- layer_embedding(       input_dim=5000, output_dim=16, mask_zero=TRUE     )     self$lstm <- layer_lstm(units=32)   },   call = function(inputs) {     inputs %>%       self$embedding() %>%       # Note that you could also prepare a `mask` tensor manually.       # It only needs to be a boolean tensor       # with the right shape, i.e. (batch_size, timesteps).       self$lstm(mask=self$embedding$compute_mask(inputs))   } )  layer <- MyLayer() x <- random_integer(c(32, 10), 0, 100) layer(x) ## tf.Tensor( ## [[ 0.00130048 -0.00113367 -0.00715671 ... -0.00107615 -0.00162071 ##    0.00135018] ##  [-0.004185    0.00726349  0.00520932 ...  0.00119117  0.00230441 ##    0.00174123] ##  [-0.00537032 -0.00164898 -0.00238435 ... -0.00154158 -0.0038603 ##   -0.00105811] ##  ... ##  [ 0.00622133 -0.00905907 -0.00599518 ...  0.00025823 -0.00142478 ##   -0.00125036] ##  [-0.00523904  0.00336683 -0.00299453 ...  0.00876719  0.00172074 ##    0.00903089] ##  [-0.00393721  0.00058538  0.00503809 ... -0.00203075  0.00325885 ##   -0.00299755]], shape=(32, 32), dtype=float32)"},{"path":"https://keras3.posit.co/dev/articles/understanding_masking_and_padding.html","id":"supporting-masking-in-your-custom-layers","dir":"Articles","previous_headings":"","what":"Supporting masking in your custom layers","title":"Understanding masking & padding","text":"Sometimes, may need write layers generate mask (like Embedding), layers need modify current mask. instance, layer produces tensor different time dimension input, Concatenate layer concatenates time dimension, need modify current mask downstream layers able properly take masked timesteps account. , layer implement layer.compute_mask() method, produces new mask given input current mask. example TemporalSplit layer needs modify current mask. another example CustomEmbedding layer capable generating mask input values: Note: details format limitations related masking, see serialization guide.","code":"TemporalSplit <- new_layer_class(   \"TemporalSplit\",   call = function(inputs) {     # Expect the input to be 3D and mask to be 2D, split the input tensor into 2     # subtensors along the time axis (axis 1).     op_split(inputs, 2, axis=2)   },   compute_mask = function(inputs, mask = NULL) {     # Also split the mask into 2 if it presents.     if (!is.null(mask)) {       op_split(mask, 2, axis=2)     } else {       NULL     }   } )  c(first_half, second_half) %<-% TemporalSplit(masked_embedding) first_half$`_keras_mask` ## tf.Tensor( ## [[ True  True  True] ##  [ True  True  True] ##  [ True  True  True]], shape=(3, 3), dtype=bool) second_half$`_keras_mask` ## tf.Tensor( ## [[False False False] ##  [ True  True False] ##  [ True  True  True]], shape=(3, 3), dtype=bool) CustomEmbedding <- new_layer_class(   \"CustomEmbedding\",   initialize = function(input_dim, output_dim, mask_zero=FALSE, ...) {     super$initialize(...)     self$input_dim <- as.integer(input_dim)     self$output_dim <- as.integer(output_dim)     self$mask_zero <- mask_zero   },   build = function(input_shape) {     self$embeddings <- self$add_weight(       shape=c(self$input_dim, self$output_dim),       initializer=\"random_normal\",       dtype=\"float32\"     )   },   call = function(inputs) {     inputs <- op_cast(inputs, \"int32\")     op_take(self$embeddings, inputs)   },   compute_mask = function(inputs, mask=NULL) {     if (!self$mask_zero) {       NULL     } else {       op_not_equal(inputs, 0)     }   } )  layer <- CustomEmbedding(input_dim = 10, output_dim = 32, mask_zero=TRUE) x <- random_integer(c(3, 10), 0, 9)  y <- layer(x) mask <- layer$compute_mask(x)  mask ## tf.Tensor( ## [[ True  True  True  True  True  True  True  True  True  True] ##  [ True  True  True  True  True  True  True  True False  True] ##  [ True  True  True  True False  True  True  True  True  True]], shape=(3, 10), dtype=bool)"},{"path":"https://keras3.posit.co/dev/articles/understanding_masking_and_padding.html","id":"opting-in-to-mask-propagation-on-compatible-layers","dir":"Articles","previous_headings":"","what":"Opting-in to mask propagation on compatible layers","title":"Understanding masking & padding","text":"layers don’t modify time dimension, don’t need modify current mask. However, may still want able propagate current mask, unchanged, next layer. opt-behavior. default, custom layer destroy current mask (since framework way tell whether propagating mask safe ). custom layer modify time dimension, want able propagate current input mask, set self.supports_masking = True layer constructor. case, default behavior compute_mask() just pass current mask . ’s example layer whitelisted mask propagation: can now use custom layer -mask-generating layer (like Embedding) mask-consuming layer (like LSTM), pass mask along reaches mask-consuming layer.","code":"MyActivation <- new_layer_class(   \"MyActivation\",   initialize = function(...) {     super$initialize(...)     self$supports_masking <- TRUE   },   call = function(inputs) {     op_relu(inputs)   } ) inputs <- keras_input(shape = shape(NULL), dtype=\"int32\") outputs <- inputs %>%   layer_embedding(input_dim=5000, output_dim=16, mask_zero=TRUE) %>%   MyActivation() %>%   layer_lstm(units=32)  model <- keras_model(inputs, outputs) y <- model(random_integer(c(32, 100), 0, 5000))"},{"path":"https://keras3.posit.co/dev/articles/understanding_masking_and_padding.html","id":"writing-layers-that-need-mask-information","dir":"Articles","previous_headings":"","what":"Writing layers that need mask information","title":"Understanding masking & padding","text":"layers mask consumers: accept mask argument call use determine whether skip certain time steps. write layer, can simply add mask=None argument call signature. mask associated inputs passed layer whenever available. ’s simple example : layer computes softmax time dimension (axis 1) input sequence, discarding masked timesteps.","code":"TemporalSoftmax <- new_layer_class(   \"TemporalSoftmax\",   initialize = function(...) {     super$initialize(...)     self$supports_masking <- TRUE   },   call = function(inputs, mask=NULL) {     if (is.null(mask)) {       stop(\"`TemporalSoftmax` layer requires a previous layer to support masking.\")     }     broadcast_float_mask <- op_expand_dims(op_cast(mask, \"float32\"), -1)     inputs_exp <- op_exp(inputs) * broadcast_float_mask     inputs_sum <- op_sum(inputs_exp * broadcast_float_mask, axis=-1, keepdims=TRUE)     inputs_exp / inputs_sum   } )  inputs <- keras_input(shape = shape(NULL), dtype=\"int32\") outputs <- inputs %>%   layer_embedding(input_dim=10, output_dim=32, mask_zero=TRUE) %>%   layer_dense(1) %>%   TemporalSoftmax()  model <- keras_model(inputs, outputs) y <- model(random_integer(c(32, 100), 0, 10))"},{"path":"https://keras3.posit.co/dev/articles/understanding_masking_and_padding.html","id":"summary","dir":"Articles","previous_headings":"","what":"Summary","title":"Understanding masking & padding","text":"need know padding & masking Keras. recap: “Masking” layers able know skip / ignore certain timesteps sequence inputs. layers mask-generators: Embedding can generate mask input values (mask_zero=TRUE), can Masking layer. layers mask-consumers: expose mask argument call method. case RNN layers. Functional API Sequential API, mask information propagated automatically. using layers standalone way, can pass mask arguments layers manually. can easily write layers modify current mask, generate new mask, consume mask associated inputs.","code":""},{"path":"https://keras3.posit.co/dev/articles/writing_a_custom_training_loop_in_tensorflow.html","id":"setup","dir":"Articles","previous_headings":"","what":"Setup","title":"Writing a training loop from scratch in TensorFlow","text":"","code":"library(tensorflow, exclude = c(\"shape\", \"set_random_seed\")) library(tfdatasets, exclude = \"shape\") library(keras3)  # This guide can only be run with the TensorFlow backend. use_backend(\"tensorflow\")"},{"path":"https://keras3.posit.co/dev/articles/writing_a_custom_training_loop_in_tensorflow.html","id":"introduction","dir":"Articles","previous_headings":"","what":"Introduction","title":"Writing a training loop from scratch in TensorFlow","text":"Keras provides default training evaluation loops, fit() evaluate(). usage covered guide Training & evaluation built-methods. want customize learning algorithm model still leveraging convenience fit() (instance, train GAN using fit()), can subclass Model class implement train_step() method, called repeatedly fit(). covered guide Customizing happens fit(). Now, want low-level control training & evaluation, write training & evaluation loops scratch. guide .","code":""},{"path":"https://keras3.posit.co/dev/articles/writing_a_custom_training_loop_in_tensorflow.html","id":"using-the-gradienttape-a-first-end-to-end-example","dir":"Articles","previous_headings":"","what":"Using the GradientTape: a first end-to-end example","title":"Writing a training loop from scratch in TensorFlow","text":"Calling model inside GradientTape scope enables retrieve gradients trainable weights layer respect loss value. Using optimizer instance, can use gradients update variables (can retrieve using model$trainable_weights). Let’s consider simple MNIST model: Let’s train using mini-batch gradient custom training loop. First, ’re going need optimizer, loss function, dataset: ’s training loop: open loop iterates epochs epoch, iterate dataset, batches batch, open GradientTape() scope Inside scope, call model (forward pass) compute loss Outside scope, retrieve gradients weights model regard loss Finally, use optimizer update weights model based gradients","code":"get_model <- function() {   inputs <- keras_input(shape = 784, name = \"digits\")   outputs <- inputs |>     layer_dense(units = 64, activation = \"relu\") |>     layer_dense(units = 64, activation = \"relu\") |>     layer_dense(units = 10, name = \"predictions\")   keras_model(inputs = inputs, outputs = outputs) }  model <- get_model() # Instantiate an optimizer. optimizer <- optimizer_adam(learning_rate = 1e-3) # Instantiate a loss function. loss_fn <- loss_sparse_categorical_crossentropy(from_logits = TRUE)  # Prepare the training dataset. batch_size <- 64 c(c(x_train, y_train), c(x_test, y_test)) %<-% dataset_mnist() x_train <- array_reshape(x_train, c(-1, 784)) x_test <- array_reshape(x_test, c(-1, 784))  # Reserve 10,000 samples for validation. val_i <- sample.int(nrow(x_train), 10000) x_val <- x_train[val_i,] y_val <- y_train[val_i] x_train %<>% .[-val_i,] y_train %<>% .[-val_i]  # Prepare the training dataset. train_dataset <- list(x_train, y_train) |>   tensor_slices_dataset() |>   dataset_shuffle(buffer_size = 1024) |>   dataset_batch(batch_size)  # Prepare the validation dataset. val_dataset <- list(x_val, y_val) |>   tensor_slices_dataset() |>   dataset_batch(batch_size) epochs <- 3 for (epoch in seq_len(epochs)) {   cat(\"Start of epoch \", epoch, \"\\n\")    # Iterate over the batches of the dataset.   step <- 0   iterator <- as_iterator(train_dataset)   while (!is.null(batch <- iter_next(iterator))) {     c(x_batch_train, y_batch_train) %<-% batch     step <- step + 1     # Open a GradientTape to record the operations run     # during the forward pass, which enables auto-differentiation.     with(tf$GradientTape() %as% tape, {       # Run the forward pass of the layer.       # The operations that the layer applies       # to its inputs are going to be recorded       # on the GradientTape.       logits <- model(x_batch_train, training = TRUE) # Logits for this minibatch        # Compute the loss value for this minibatch.       loss_value <- loss_fn(y_batch_train, logits)     })      # Use the gradient tape to automatically retrieve     # the gradients of the trainable variables with respect to the loss.     gradients <- tape$gradient(loss_value, model$trainable_weights)      # Run one step of gradient descent by updating     # the value of the variables to minimize the loss.     optimizer$apply(gradients, model$trainable_weights)      # Log every 200 batches.     if (step %% 200 == 0) {       cat(sprintf(\"Training loss (for one batch) at step %d: %.4f\\n\",                   step, loss_value))       cat(sprintf(\"Seen so far: %d samples\\n\", (step * batch_size)))     }   } } ## Start of epoch  1 ## Training loss (for one batch) at step 200: 1.1675 ## Seen so far: 12800 samples ## Training loss (for one batch) at step 400: 1.6450 ## Seen so far: 25600 samples ## Training loss (for one batch) at step 600: 0.7421 ## Seen so far: 38400 samples ## Start of epoch  2 ## Training loss (for one batch) at step 200: 0.7254 ## Seen so far: 12800 samples ## Training loss (for one batch) at step 400: 0.2476 ## Seen so far: 25600 samples ## Training loss (for one batch) at step 600: 0.3578 ## Seen so far: 38400 samples ## Start of epoch  3 ## Training loss (for one batch) at step 200: 0.2974 ## Seen so far: 12800 samples ## Training loss (for one batch) at step 400: 0.4441 ## Seen so far: 25600 samples ## Training loss (for one batch) at step 600: 0.1461 ## Seen so far: 38400 samples"},{"path":"https://keras3.posit.co/dev/articles/writing_a_custom_training_loop_in_tensorflow.html","id":"low-level-handling-of-metrics","dir":"Articles","previous_headings":"","what":"Low-level handling of metrics","title":"Writing a training loop from scratch in TensorFlow","text":"Let’s add metrics monitoring basic loop. can readily reuse built-metrics (custom ones wrote) training loops written scratch. ’s flow: Instantiate metric start loop Call metric$update_state() batch Call metric$result() need display current value metric Call metric$reset_state() need clear state metric (typically end epoch) Let’s use knowledge compute SparseCategoricalAccuracy validation data end epoch: ’s training & evaluation loop:","code":"# Get a fresh model model <- get_model()  # Instantiate an optimizer to train the model. optimizer <- optimizer_adam(learning_rate = 1e-3) # Instantiate a loss function. loss_fn <- loss_sparse_categorical_crossentropy(from_logits = TRUE)  # Prepare the metrics. train_acc_metric <- metric_sparse_categorical_accuracy() val_acc_metric <- metric_sparse_categorical_accuracy() epochs <- 2 time <- Sys.time() for (epoch in seq_len(epochs)) {   cat(\"Start of epoch \", epoch, \"\\n\")    # Iterate over the batches of the dataset.   step <- 0   train_dataset_iterator <- as_iterator(train_dataset)   while (!is.null(train_batch <- iter_next(train_dataset_iterator))) {     c(x_batch_train, y_batch_train) %<-% train_batch     step <- step + 1      with(tf$GradientTape() %as% tape, {       logits <- model(x_batch_train, training = TRUE)       loss_value <- loss_fn(y_batch_train, logits)     })     gradients <- tape$gradient(loss_value, model$trainable_weights)     optimizer$apply(gradients, model$trainable_weights)      # Update training metric.     train_acc_metric$update_state(y_batch_train, logits)     if (step %% 200 == 0) {       cat(sprintf(         \"Training loss (for one batch) at step %d: %.4f\\n\", step, loss_value))       cat(sprintf(\"Seen so far: %d samples \\n\", (step * batch_size)))     }   }    # Display metrics at the end of each epoch.   train_acc <- train_acc_metric$result()   cat(sprintf(\"Training acc over epoch: %.4f\\n\", train_acc))    # Reset training metrics at the end of each epoch   train_acc_metric$reset_state()    # Run a validation loop at the end of each epoch.   iterate(val_dataset, function(val_batch) {     c(x_batch_val, y_batch_val) %<-% val_batch     val_logits <- model(x_batch_val, training = FALSE)     val_acc_metric$update_state(y_batch_val, val_logits)   })   val_acc <- val_acc_metric$result()   val_acc_metric$reset_state()   cat(sprintf(\"Validation acc: %.4f\\n\", val_acc)) } ## Start of epoch  1 ## Training loss (for one batch) at step 200: 1.9564 ## Seen so far: 12800 samples ## Training loss (for one batch) at step 400: 1.3562 ## Seen so far: 25600 samples ## Training loss (for one batch) at step 600: 0.4850 ## Seen so far: 38400 samples ## Training acc over epoch: 0.7836 ## Validation acc: 0.8681 ## Start of epoch  2 ## Training loss (for one batch) at step 200: 0.7679 ## Seen so far: 12800 samples ## Training loss (for one batch) at step 400: 0.5674 ## Seen so far: 25600 samples ## Training loss (for one batch) at step 600: 0.4705 ## Seen so far: 38400 samples ## Training acc over epoch: 0.8838 ## Validation acc: 0.9049 Sys.time() - time ## Time difference of 39.32149 secs"},{"path":"https://keras3.posit.co/dev/articles/writing_a_custom_training_loop_in_tensorflow.html","id":"speeding-up-your-training-step-with-tf_function","dir":"Articles","previous_headings":"","what":"Speeding-up your training step with tf_function()","title":"Writing a training loop from scratch in TensorFlow","text":"default runtime TensorFlow 2 eager execution. , training loop executes eagerly. great debugging, graph compilation definite performance advantage. Describing computation static graph enables framework apply global performance optimizations. impossible framework constrained greedily execute one operation another, knowledge comes next. can compile static graph function takes tensors input. Just add @tf.function decorator , like : Let’s evaluation step: Now, let’s re-run training loop compiled training step: Much faster, isn’t ?","code":"train_step <- tf_function(function(x, y) {   with(tf$GradientTape() %as% tape, {     logits <- model(x, training = TRUE)     loss_value <- loss_fn(y, logits)   })   gradients <- tape$gradient(loss_value, model$trainable_weights)   optimizer$apply(gradients, model$trainable_weights)   train_acc_metric$update_state(y, logits)   invisible(NULL) # return nothing }) test_step <- tf_function(function(x, y) {   val_logits <- model(x, training=FALSE)   val_acc_metric$update_state(y, val_logits)   invisible(NULL) # return nothing }) epochs <- 2 time <- Sys.time() for (epoch in seq_len(epochs)) {   cat(\"Start of epoch \", epoch, \"\\n\")    # Iterate over the batches of the dataset.   step <- 0   while (!is.null(batch <- iter_next(iterator))) {     c(x_batch_train, y_batch_train) %<-% batch     step <- step + 1     train_step(x_batch_train, y_batch_train)      if (step %% 200 == 0) {       cat(sprintf(         \"Training loss (for one batch) at step %d: %.4f\\n\", step, loss_value       ))       cat(sprintf(\"Seen so far: %d samples \\n\", (step * batch_size)))     }   }    # Display metrics at the end of each epoch.   train_acc <- train_acc_metric$result()   cat(sprintf(\"Training acc over epoch: %.4f\\n\", train_acc))     # Reset training metrics at the end of each epoch   train_acc_metric$reset_state()    # Run a validation loop at the end of each epoch.    iterate(val_dataset, function(val_batch) {     c(x_batch_val, y_batch_val) %<-% val_batch     test_step(x_batch_val, y_batch_val)   })    val_acc <- val_acc_metric$result()   val_acc_metric$reset_state()   cat(sprintf(\"Validation acc: %.4f\\n\", val_acc)) } ## Start of epoch  1 ## Training acc over epoch: 0.0000 ## Validation acc: 0.9049 ## Start of epoch  2 ## Training acc over epoch: 0.0000 ## Validation acc: 0.9049 Sys.time() - time ## Time difference of 0.4121809 secs"},{"path":"https://keras3.posit.co/dev/articles/writing_a_custom_training_loop_in_tensorflow.html","id":"low-level-handling-of-losses-tracked-by-the-model","dir":"Articles","previous_headings":"","what":"Low-level handling of losses tracked by the model","title":"Writing a training loop from scratch in TensorFlow","text":"Layers models recursively track losses created forward pass layers call self$add_loss(value). resulting list scalar loss values available via property model$losses end forward pass. want using loss components, sum add main loss training step. Consider layer, creates activity regularization loss: Let’s build really simple model uses : ’s training step look like now:","code":"layer_activity_regularization <- Layer(   \"ActivityRegularizationLayer\",   call = function(inputs) {     self$add_loss(0.1 * op_mean(inputs))     inputs   } ) inputs <- keras_input(shape = 784, name=\"digits\") outputs <- inputs |>   layer_dense(units = 64, activation = \"relu\") |>   layer_activity_regularization() |>   layer_dense(units = 64, activation = \"relu\") |>   layer_dense(units = 10, name = \"predictions\") model <- keras_model(inputs = inputs, outputs = outputs) train_step <- tf_function(function(x, y) {   with(tf$GradientTape() %as% tape, {     logits <- model(x, training = TRUE)     loss_value <- Reduce(`+`, c(loss_fn(y, logits),                                 model$losses))   })   gradients <- tape$gradient(loss_value, model$trainable_weights)   optimizer$apply(gradients, model$trainable_weights)   train_acc_metric$update_state(y, logits)   invisible(NULL) })"},{"path":"https://keras3.posit.co/dev/articles/writing_a_custom_training_loop_in_tensorflow.html","id":"summary","dir":"Articles","previous_headings":"","what":"Summary","title":"Writing a training loop from scratch in TensorFlow","text":"Now know everything know using built-training loops writing scratch. conclude, ’s simple end--end example ties together everything ’ve learned guide: DCGAN trained MNIST digits.","code":""},{"path":"https://keras3.posit.co/dev/articles/writing_a_custom_training_loop_in_tensorflow.html","id":"end-to-end-example-a-gan-training-loop-from-scratch","dir":"Articles","previous_headings":"","what":"End-to-end example: a GAN training loop from scratch","title":"Writing a training loop from scratch in TensorFlow","text":"may familiar Generative Adversarial Networks (GANs). GANs can generate new images look almost real, learning latent distribution training dataset images (“latent space” images). GAN made two parts: “generator” model maps points latent space points image space, “discriminator” model, classifier can tell difference real images (training dataset) fake images (output generator network). GAN training loop looks like : Train discriminator. Sample batch random points latent space. Turn points fake images via “generator” model. Get batch real images combine generated images. Train “discriminator” model classify generated vs. real images. Train generator. Sample random points latent space. Turn points fake images via “generator” network. Get batch real images combine generated images. Train “generator” model “fool” discriminator classify fake images real. much detailed overview GANs works, see Deep Learning Python. Let’s implement training loop. First, create discriminator meant classify fake vs real digits: let’s create generator network, turns latent vectors outputs shape (28, 28, 1) (representing MNIST digits): ’s key bit: training loop. can see quite straightforward. training step function takes 17 lines. Let’s train GAN, repeatedly calling train_step batches images. Since discriminator generator convnets, ’re going want run code GPU. ’s ! ’ll get nice-looking fake MNIST digits just ~30s training GPU.","code":"# Create the discriminator discriminator <-   keras_model_sequential(name = \"discriminator\",                          input_shape = c(28, 28, 1)) |>   layer_conv_2d(filters = 64, kernel_size = c(3, 3),                 strides = c(2, 2), padding = \"same\") |>   layer_activation_leaky_relu(negative_slope = 0.2) |>   layer_conv_2d(filters = 128, kernel_size = c(3, 3),                 strides = c(2, 2), padding = \"same\") |>   layer_activation_leaky_relu(negative_slope = 0.2) |>   layer_global_max_pooling_2d() |>   layer_dense(units = 1)  summary(discriminator) ## Model: \"discriminator\" ## ┏━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━┓ ## ┃ Layer (type)                    ┃ Output Shape           ┃       Param # ┃ ## ┡━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━┩ ## │ conv2d (Conv2D)                 │ (None, 14, 14, 64)     │           640 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ leaky_re_lu (LeakyReLU)         │ (None, 14, 14, 64)     │             0 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ conv2d_1 (Conv2D)               │ (None, 7, 7, 128)      │        73,856 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ leaky_re_lu_1 (LeakyReLU)       │ (None, 7, 7, 128)      │             0 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ global_max_pooling2d            │ (None, 128)            │             0 │ ## │ (GlobalMaxPooling2D)            │                        │               │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ dense_6 (Dense)                 │ (None, 1)              │           129 │ ## └─────────────────────────────────┴────────────────────────┴───────────────┘ ##  Total params: 74,625 (291.50 KB) ##  Trainable params: 74,625 (291.50 KB) ##  Non-trainable params: 0 (0.00 B) latent_dim <- 128L  generator <-   keras_model_sequential(name = \"generator\",                          input_shape = latent_dim) |>   layer_dense(7 * 7 * 128) |>   layer_activation_leaky_relu(negative_slope = 0.2) |>   layer_reshape(target_shape = c(7, 7, 128)) |>   layer_conv_2d_transpose(filters = 128, kernel_size = c(4, 4),                           strides = c(2, 2), padding = \"same\") |>   layer_activation_leaky_relu(negative_slope = 0.2) |>   layer_conv_2d_transpose(filters = 128, kernel_size = c(4, 4),                           strides = c(2, 2), padding = \"same\") |>   layer_activation_leaky_relu(negative_slope = 0.2) |>   layer_conv_2d(filters = 1, kernel_size = c(7, 7), padding = \"same\",                 activation = \"sigmoid\")  summary(generator) ## Model: \"generator\" ## ┏━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━┓ ## ┃ Layer (type)                    ┃ Output Shape           ┃       Param # ┃ ## ┡━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━┩ ## │ dense_7 (Dense)                 │ (None, 6272)           │       809,088 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ leaky_re_lu_2 (LeakyReLU)       │ (None, 6272)           │             0 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ reshape (Reshape)               │ (None, 7, 7, 128)      │             0 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ conv2d_transpose                │ (None, 14, 14, 128)    │       262,272 │ ## │ (Conv2DTranspose)               │                        │               │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ leaky_re_lu_3 (LeakyReLU)       │ (None, 14, 14, 128)    │             0 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ conv2d_transpose_1              │ (None, 28, 28, 128)    │       262,272 │ ## │ (Conv2DTranspose)               │                        │               │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ leaky_re_lu_4 (LeakyReLU)       │ (None, 28, 28, 128)    │             0 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ conv2d_2 (Conv2D)               │ (None, 28, 28, 1)      │         6,273 │ ## └─────────────────────────────────┴────────────────────────┴───────────────┘ ##  Total params: 1,339,905 (5.11 MB) ##  Trainable params: 1,339,905 (5.11 MB) ##  Non-trainable params: 0 (0.00 B) # Instantiate one optimizer for the discriminator and another for the generator. d_optimizer <- optimizer_adam(learning_rate = 0.0003) g_optimizer <- optimizer_adam(learning_rate = 0.0004)  # Instantiate a loss function. loss_fn <- loss_binary_crossentropy(from_logits = TRUE)  train_step <- tf_function(function(real_images) {   # Sample random points in the latent space   c(batch_size, ...) %<-% op_shape(real_images)   random_latent_vectors <-     tf$random$normal(shape(batch_size, latent_dim))    # Decode them to fake images   generated_images <- generator(random_latent_vectors)    # Combine them with real images   combined_images <- tf$concat(list(generated_images, real_images),                                axis = 0L)    # Assemble labels discriminating real from fake images   labels <- tf$concat(list(tf$ones(shape(batch_size, 1)),                            tf$zeros(shape(batch_size, 1))),                       axis = 0L)    # Add random noise to the labels - important trick!   labels %<>% `+`(tf$random$uniform(tf$shape(.), maxval = 0.05))    # Train the discriminator   with(tf$GradientTape() %as% tape, {     predictions <- discriminator(combined_images)     d_loss <- loss_fn(labels, predictions)   })   grads <- tape$gradient(d_loss, discriminator$trainable_weights)   d_optimizer$apply(grads, discriminator$trainable_weights)    # Sample random points in the latent space   random_latent_vectors <-     tf$random$normal(shape(batch_size, latent_dim))    # Assemble labels that say \"all real images\"   misleading_labels <- tf$zeros(shape(batch_size, 1))    # Train the generator (note that we should *not* update the weights   # of the discriminator)!   with(tf$GradientTape() %as% tape, {     predictions <- discriminator(generator(random_latent_vectors))     g_loss <- loss_fn(misleading_labels, predictions)   })    grads <- tape$gradient(g_loss, generator$trainable_weights)   g_optimizer$apply(grads, generator$trainable_weights)    list(d_loss, g_loss, generated_images) }) # Prepare the dataset. We use both the training & test MNIST digits. batch_size <- 64 c(c(x_train, .), c(x_test, .)) %<-% dataset_mnist() all_digits <- op_concatenate(list(x_train, x_test)) all_digits <- op_reshape(all_digits, c(-1, 28, 28, 1)) dataset <- all_digits |>   tensor_slices_dataset() |>   dataset_map(\\(x) op_cast(x, \"float32\") / 255) |>   dataset_shuffle(buffer_size = 1024) |>   dataset_batch(batch_size = batch_size)  epochs <- 1 # In practice you need at least 20 epochs to generate nice digits. save_dir <- \"./\"  for (epoch in seq_len(epochs)) {   cat(\"Start epoch: \", epoch, \"\\n\")   step <- 0   train_iterator <- as_iterator(dataset)   while (!is.null(real_images <- iter_next(train_iterator))) {     step <- step + 1     # Train the discriminator & generator on one batch of real images.     c(d_loss, g_loss, generated_images) %<-% train_step(real_images)      # Logging.     if (step %% 200 == 0) {       # Print metrics       cat(sprintf(\"discriminator loss at step %d: %.2f\\n\", step, d_loss))       cat(sprintf(\"adversarial loss at step %d: %.2f\\n\", step, g_loss))     }      # To limit execution time we stop after 10 steps.     # Remove the lines below to actually train the model!     if (step > 10)       break   } } ## Start epoch:  1"},{"path":"https://keras3.posit.co/dev/articles/writing_your_own_callbacks.html","id":"introduction","dir":"Articles","previous_headings":"","what":"Introduction","title":"Writing your own callbacks","text":"callback powerful tool customize behavior Keras model training, evaluation, inference. Examples include keras.callbacks.TensorBoard visualize training progress results TensorBoard, keras.callbacks.ModelCheckpoint periodically save model training. guide, learn Keras callback , can , can build . provide demos simple callback applications get started.","code":""},{"path":"https://keras3.posit.co/dev/articles/writing_your_own_callbacks.html","id":"setup","dir":"Articles","previous_headings":"","what":"Setup","title":"Writing your own callbacks","text":"","code":"library(keras3)"},{"path":"https://keras3.posit.co/dev/articles/writing_your_own_callbacks.html","id":"keras-callbacks-overview","dir":"Articles","previous_headings":"","what":"Keras callbacks overview","title":"Writing your own callbacks","text":"callbacks subclass keras.callbacks.Callback class, override set methods called various stages training, testing, predicting. Callbacks useful get view internal states statistics model training. can pass list callbacks (keyword argument callbacks) following model methods: fit() evaluate() predict()","code":""},{"path":[]},{"path":[]},{"path":"https://keras3.posit.co/dev/articles/writing_your_own_callbacks.html","id":"on_traintestpredict_beginlogs-null","dir":"Articles","previous_headings":"An overview of callback methods > Global methods","what":"on_(train|test|predict)_begin(logs = NULL)","title":"Writing your own callbacks","text":"Called beginning fit/evaluate/predict.","code":""},{"path":"https://keras3.posit.co/dev/articles/writing_your_own_callbacks.html","id":"on_traintestpredict_endlogs-null","dir":"Articles","previous_headings":"An overview of callback methods > Global methods","what":"on_(train|test|predict)_end(logs = NULL)","title":"Writing your own callbacks","text":"Called end fit/evaluate/predict.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/articles/writing_your_own_callbacks.html","id":"on_traintestpredict_batch_beginbatch-logs-null","dir":"Articles","previous_headings":"An overview of callback methods > Batch-level methods for training/testing/predicting","what":"on_(train|test|predict)_batch_begin(batch, logs = NULL)","title":"Writing your own callbacks","text":"Called right processing batch training/testing/predicting.","code":""},{"path":"https://keras3.posit.co/dev/articles/writing_your_own_callbacks.html","id":"on_traintestpredict_batch_endbatch-logs-null","dir":"Articles","previous_headings":"An overview of callback methods > Batch-level methods for training/testing/predicting","what":"on_(train|test|predict)_batch_end(batch, logs = NULL)","title":"Writing your own callbacks","text":"Called end training/testing/predicting batch. Within method, logs named list containing metrics results.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/articles/writing_your_own_callbacks.html","id":"on_epoch_beginepoch-logs-null","dir":"Articles","previous_headings":"An overview of callback methods > Epoch-level methods (training only)","what":"on_epoch_begin(epoch, logs = NULL)","title":"Writing your own callbacks","text":"Called beginning epoch training.","code":""},{"path":"https://keras3.posit.co/dev/articles/writing_your_own_callbacks.html","id":"on_epoch_endepoch-logs-null","dir":"Articles","previous_headings":"An overview of callback methods > Epoch-level methods (training only)","what":"on_epoch_end(epoch, logs = NULL)","title":"Writing your own callbacks","text":"Called end epoch training.","code":""},{"path":"https://keras3.posit.co/dev/articles/writing_your_own_callbacks.html","id":"a-basic-example","dir":"Articles","previous_headings":"","what":"A basic example","title":"Writing your own callbacks","text":"Let’s take look concrete example. get started, let’s import tensorflow define simple Sequential Keras model: , load MNIST data training testing Keras datasets API: Now, define simple custom callback logs: fit/evaluate/predict starts & ends epoch starts & ends training batch starts & ends evaluation (test) batch starts & ends inference (prediction) batch starts & ends Let’s try :","code":"# Define the Keras model to add callbacks to get_model <- function() {   model <- keras_model_sequential()   model |> layer_dense(units = 1)   model |> compile(     optimizer = optimizer_rmsprop(learning_rate = 0.1),     loss = \"mean_squared_error\",     metrics = \"mean_absolute_error\"   )   model } # Load example MNIST data and pre-process it mnist <- dataset_mnist()  flatten_and_rescale <- function(x) {   x <- array_reshape(x, c(-1, 784))   x <- x / 255   x }  mnist$train$x <- flatten_and_rescale(mnist$train$x) mnist$test$x  <- flatten_and_rescale(mnist$test$x)  # limit to 1000 samples n <- 1000 mnist$train$x <- mnist$train$x[1:n,] mnist$train$y <- mnist$train$y[1:n] mnist$test$x  <- mnist$test$x[1:n,] mnist$test$y  <- mnist$test$y[1:n] show <- function(msg, logs) {   cat(glue::glue(msg, .envir = parent.frame()),       \"got logs: \", sep = \"; \")   str(logs); cat(\"\\n\") }  callback_custom <- Callback(   \"CustomCallback\",   on_train_begin         = \\(logs = NULL)        show(\"Starting training\", logs),   on_epoch_begin         = \\(epoch, logs = NULL) show(\"Start epoch {epoch} of training\", logs),   on_train_batch_begin   = \\(batch, logs = NULL) show(\"...Training: start of batch {batch}\", logs),   on_train_batch_end     = \\(batch, logs = NULL) show(\"...Training: end of batch {batch}\",  logs),   on_epoch_end           = \\(epoch, logs = NULL) show(\"End epoch {epoch} of training\", logs),   on_train_end           = \\(logs = NULL)        show(\"Stop training\", logs),     on_test_begin          = \\(logs = NULL)        show(\"Start testing\", logs),   on_test_batch_begin    = \\(batch, logs = NULL) show(\"...Evaluating: start of batch {batch}\", logs),   on_test_batch_end      = \\(batch, logs = NULL) show(\"...Evaluating: end of batch {batch}\", logs),   on_test_end            = \\(logs = NULL)        show(\"Stop testing\", logs),    on_predict_begin       = \\(logs = NULL)        show(\"Start predicting\", logs),   on_predict_end         = \\(logs = NULL)        show(\"Stop predicting\", logs),   on_predict_batch_begin = \\(batch, logs = NULL) show(\"...Predicting: start of batch {batch}\", logs),   on_predict_batch_end   = \\(batch, logs = NULL) show(\"...Predicting: end of batch {batch}\", logs), ) model <- get_model() model |> fit(   mnist$train$x, mnist$train$y,   batch_size = 128,   epochs = 2,   verbose = 0,   validation_split = 0.5,   callbacks = list(callback_custom()) ) ## Starting training; got logs:  Named list() ## ## Start epoch 1 of training; got logs:  Named list() ## ## ...Training: start of batch 1; got logs:  Named list() ## ## ...Training: end of batch 1; got logs: List of 2 ##  $ loss               : num 25.9 ##  $ mean_absolute_error: num 4.19 ## ## ...Training: start of batch 2; got logs:  Named list() ## ## ...Training: end of batch 2; got logs: List of 2 ##  $ loss               : num 433 ##  $ mean_absolute_error: num 15.5 ## ## ...Training: start of batch 3; got logs:  Named list() ## ## ...Training: end of batch 3; got logs: List of 2 ##  $ loss               : num 297 ##  $ mean_absolute_error: num 11.8 ## ## ...Training: start of batch 4; got logs:  Named list() ## ## ...Training: end of batch 4; got logs: List of 2 ##  $ loss               : num 231 ##  $ mean_absolute_error: num 9.68 ## ## Start testing; got logs:  Named list() ## ## ...Evaluating: start of batch 1; got logs:  Named list() ## ## ...Evaluating: end of batch 1; got logs: List of 2 ##  $ loss               : num 8.1 ##  $ mean_absolute_error: num 2.3 ## ## ...Evaluating: start of batch 2; got logs:  Named list() ## ## ...Evaluating: end of batch 2; got logs: List of 2 ##  $ loss               : num 7.58 ##  $ mean_absolute_error: num 2.23 ## ## ...Evaluating: start of batch 3; got logs:  Named list() ## ## ...Evaluating: end of batch 3; got logs: List of 2 ##  $ loss               : num 7.38 ##  $ mean_absolute_error: num 2.21 ## ## ...Evaluating: start of batch 4; got logs:  Named list() ## ## ...Evaluating: end of batch 4; got logs: List of 2 ##  $ loss               : num 7.3 ##  $ mean_absolute_error: num 2.21 ## ## Stop testing; got logs: List of 2 ##  $ loss               : num 7.3 ##  $ mean_absolute_error: num 2.21 ## ## End epoch 1 of training; got logs: List of 4 ##  $ loss                   : num 231 ##  $ mean_absolute_error    : num 9.68 ##  $ val_loss               : num 7.3 ##  $ val_mean_absolute_error: num 2.21 ## ## Start epoch 2 of training; got logs:  Named list() ## ## ...Training: start of batch 1; got logs:  Named list() ## ## ...Training: end of batch 1; got logs: List of 2 ##  $ loss               : num 7.44 ##  $ mean_absolute_error: num 2.27 ## ## ...Training: start of batch 2; got logs:  Named list() ## ## ...Training: end of batch 2; got logs: List of 2 ##  $ loss               : num 6.81 ##  $ mean_absolute_error: num 2.16 ## ## ...Training: start of batch 3; got logs:  Named list() ## ## ...Training: end of batch 3; got logs: List of 2 ##  $ loss               : num 6.12 ##  $ mean_absolute_error: num 2.06 ## ## ...Training: start of batch 4; got logs:  Named list() ## ## ...Training: end of batch 4; got logs: List of 2 ##  $ loss               : num 6.08 ##  $ mean_absolute_error: num 2.04 ## ## Start testing; got logs:  Named list() ## ## ...Evaluating: start of batch 1; got logs:  Named list() ## ## ...Evaluating: end of batch 1; got logs: List of 2 ##  $ loss               : num 5.54 ##  $ mean_absolute_error: num 1.92 ## ## ...Evaluating: start of batch 2; got logs:  Named list() ## ## ...Evaluating: end of batch 2; got logs: List of 2 ##  $ loss               : num 5.31 ##  $ mean_absolute_error: num 1.87 ## ## ...Evaluating: start of batch 3; got logs:  Named list() ## ## ...Evaluating: end of batch 3; got logs: List of 2 ##  $ loss               : num 5.11 ##  $ mean_absolute_error: num 1.8 ## ## ...Evaluating: start of batch 4; got logs:  Named list() ## ## ...Evaluating: end of batch 4; got logs: List of 2 ##  $ loss               : num 5.15 ##  $ mean_absolute_error: num 1.82 ## ## Stop testing; got logs: List of 2 ##  $ loss               : num 5.15 ##  $ mean_absolute_error: num 1.82 ## ## End epoch 2 of training; got logs: List of 4 ##  $ loss                   : num 6.08 ##  $ mean_absolute_error    : num 2.04 ##  $ val_loss               : num 5.15 ##  $ val_mean_absolute_error: num 1.82 ## ## Stop training; got logs: List of 4 ##  $ loss                   : num 6.08 ##  $ mean_absolute_error    : num 2.04 ##  $ val_loss               : num 5.15 ##  $ val_mean_absolute_error: num 1.82 res <- model |> evaluate(   mnist$test$x, mnist$test$y,   batch_size = 128, verbose = 0,   callbacks = list(callback_custom()) ) ## Start testing; got logs:  Named list() ## ## ...Evaluating: start of batch 1; got logs:  Named list() ## ## ...Evaluating: end of batch 1; got logs: List of 2 ##  $ loss               : num 5.2 ##  $ mean_absolute_error: num 1.84 ## ## ...Evaluating: start of batch 2; got logs:  Named list() ## ## ...Evaluating: end of batch 2; got logs: List of 2 ##  $ loss               : num 4.62 ##  $ mean_absolute_error: num 1.73 ## ## ...Evaluating: start of batch 3; got logs:  Named list() ## ## ...Evaluating: end of batch 3; got logs: List of 2 ##  $ loss               : num 4.61 ##  $ mean_absolute_error: num 1.74 ## ## ...Evaluating: start of batch 4; got logs:  Named list() ## ## ...Evaluating: end of batch 4; got logs: List of 2 ##  $ loss               : num 4.65 ##  $ mean_absolute_error: num 1.75 ## ## ...Evaluating: start of batch 5; got logs:  Named list() ## ## ...Evaluating: end of batch 5; got logs: List of 2 ##  $ loss               : num 4.84 ##  $ mean_absolute_error: num 1.77 ## ## ...Evaluating: start of batch 6; got logs:  Named list() ## ## ...Evaluating: end of batch 6; got logs: List of 2 ##  $ loss               : num 4.76 ##  $ mean_absolute_error: num 1.76 ## ## ...Evaluating: start of batch 7; got logs:  Named list() ## ## ...Evaluating: end of batch 7; got logs: List of 2 ##  $ loss               : num 4.74 ##  $ mean_absolute_error: num 1.76 ## ## ...Evaluating: start of batch 8; got logs:  Named list() ## ## ...Evaluating: end of batch 8; got logs: List of 2 ##  $ loss               : num 4.67 ##  $ mean_absolute_error: num 1.75 ## ## Stop testing; got logs: List of 2 ##  $ loss               : num 4.67 ##  $ mean_absolute_error: num 1.75 res <- model |> predict(   mnist$test$x,   batch_size = 128, verbose = 0,   callbacks = list(callback_custom()) ) ## Start predicting; got logs:  Named list() ## ## ...Predicting: start of batch 1; got logs:  Named list() ## ## ...Predicting: end of batch 1; got logs: List of 1 ##  $ outputs:<tf.Tensor: shape=(128, 1), dtype=float32, numpy=…> ## ## ...Predicting: start of batch 2; got logs:  Named list() ## ## ...Predicting: end of batch 2; got logs: List of 1 ##  $ outputs:<tf.Tensor: shape=(128, 1), dtype=float32, numpy=…> ## ## ...Predicting: start of batch 3; got logs:  Named list() ## ## ...Predicting: end of batch 3; got logs: List of 1 ##  $ outputs:<tf.Tensor: shape=(128, 1), dtype=float32, numpy=…> ## ## ...Predicting: start of batch 4; got logs:  Named list() ## ## ...Predicting: end of batch 4; got logs: List of 1 ##  $ outputs:<tf.Tensor: shape=(128, 1), dtype=float32, numpy=…> ## ## ...Predicting: start of batch 5; got logs:  Named list() ## ## ...Predicting: end of batch 5; got logs: List of 1 ##  $ outputs:<tf.Tensor: shape=(128, 1), dtype=float32, numpy=…> ## ## ...Predicting: start of batch 6; got logs:  Named list() ## ## ...Predicting: end of batch 6; got logs: List of 1 ##  $ outputs:<tf.Tensor: shape=(128, 1), dtype=float32, numpy=…> ## ## ...Predicting: start of batch 7; got logs:  Named list() ## ## ...Predicting: end of batch 7; got logs: List of 1 ##  $ outputs:<tf.Tensor: shape=(128, 1), dtype=float32, numpy=…> ## ## ...Predicting: start of batch 8; got logs:  Named list() ## ## ...Predicting: end of batch 8; got logs: List of 1 ##  $ outputs:<tf.Tensor: shape=(104, 1), dtype=float32, numpy=…> ## ## Stop predicting; got logs:  Named list()"},{"path":"https://keras3.posit.co/dev/articles/writing_your_own_callbacks.html","id":"usage-of-logs-list","dir":"Articles","previous_headings":"A basic example","what":"Usage of logs list","title":"Writing your own callbacks","text":"logs named list contains loss value, metrics end batch epoch. Example includes loss mean absolute error. information callbacks, can check Keras callback API documentation.","code":"callback_print_loss_and_mae <- Callback(   \"LossAndErrorPrintingCallback\",    on_train_batch_end = function(batch, logs = NULL)     cat(sprintf(\"Up to batch %i, the average loss is %7.2f.\\n\",                 batch,  logs$loss)),    on_test_batch_end = function(batch, logs = NULL)     cat(sprintf(\"Up to batch %i, the average loss is %7.2f.\\n\",                 batch, logs$loss)),    on_epoch_end = function(epoch, logs = NULL)     cat(sprintf(       \"The average loss for epoch %2i is %9.2f and mean absolute error is %7.2f.\\n\",       epoch, logs$loss, logs$mean_absolute_error     )) )   model <- get_model() model |> fit(   mnist$train$x, mnist$train$y,   epochs = 2, verbose = 0, batch_size = 128,   callbacks = list(callback_print_loss_and_mae()) ) ## Up to batch 1, the average loss is   25.12. ## Up to batch 2, the average loss is  398.92. ## Up to batch 3, the average loss is  274.04. ## Up to batch 4, the average loss is  208.32. ## Up to batch 5, the average loss is  168.15. ## Up to batch 6, the average loss is  141.31. ## Up to batch 7, the average loss is  122.19. ## Up to batch 8, the average loss is  110.05. ## The average loss for epoch  1 is    110.05 and mean absolute error is    5.79. ## Up to batch 1, the average loss is    4.71. ## Up to batch 2, the average loss is    4.74. ## Up to batch 3, the average loss is    4.81. ## Up to batch 4, the average loss is    5.07. ## Up to batch 5, the average loss is    5.08. ## Up to batch 6, the average loss is    5.09. ## Up to batch 7, the average loss is    5.19. ## Up to batch 8, the average loss is    5.51. ## The average loss for epoch  2 is      5.51 and mean absolute error is    1.90. res = model |> evaluate(   mnist$test$x, mnist$test$y,   verbose = 0, batch_size = 128,   callbacks = list(callback_print_loss_and_mae()) ) ## Up to batch 1, the average loss is   15.86. ## Up to batch 2, the average loss is   16.13. ## Up to batch 3, the average loss is   16.02. ## Up to batch 4, the average loss is   16.11. ## Up to batch 5, the average loss is   16.23. ## Up to batch 6, the average loss is   16.68. ## Up to batch 7, the average loss is   16.61. ## Up to batch 8, the average loss is   16.54."},{"path":"https://keras3.posit.co/dev/articles/writing_your_own_callbacks.html","id":"usage-of-selfmodel-attribute","dir":"Articles","previous_headings":"","what":"Usage of self$model attribute","title":"Writing your own callbacks","text":"addition receiving log information one methods called, callbacks access model associated current round training/evaluation/inference: self$model. things can self$model callback: Set self$model$stop_training <- TRUE immediately interrupt training. Mutate hyperparameters optimizer (available self$model$optimizer), self$model$optimizer$learning_rate. Save model period intervals. Record output model |> predict() test samples end epoch, use sanity check training. Extract visualizations intermediate features end epoch, monitor model learning time. etc. Let’s see action couple examples.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/articles/writing_your_own_callbacks.html","id":"early-stopping-at-minimum-loss","dir":"Articles","previous_headings":"Examples of Keras callback applications","what":"Early stopping at minimum loss","title":"Writing your own callbacks","text":"first example shows creation Callback stops training minimum loss reached, setting attribute self$model$stop_training (boolean). Optionally, can provide argument patience specify many epochs wait stopping reached local minimum. callback_early_stopping() provides complete general implementation.","code":"callback_early_stopping_at_min_loss <- Callback(   \"EarlyStoppingAtMinLoss\",   `__doc__` =     \"Stop training when the loss is at its min, i.e. the loss stops decreasing.      Arguments:         patience: Number of epochs to wait after min has been hit. After this         number of no improvement, training stops.     \",    initialize = function(patience = 0) {     super$initialize()     self$patience <- patience     # best_weights to store the weights at which the minimum loss occurs.     self$best_weights <- NULL   },    on_train_begin = function(logs = NULL) {     # The number of epoch it has waited when loss is no longer minimum.     self$wait <- 0     # The epoch the training stops at.     self$stopped_epoch <- 0     # Initialize the best as infinity.     self$best <- Inf   },    on_epoch_end = function(epoch, logs = NULL) {     current <- logs$loss     if (current < self$best) {       self$best <- current       self$wait <- 0L       # Record the best weights if current results is better (less).       self$best_weights <- get_weights(self$model)     } else {       add(self$wait) <- 1L       if (self$wait >= self$patience) {         self$stopped_epoch <- epoch         self$model$stop_training <- TRUE         cat(\"Restoring model weights from the end of the best epoch.\\n\")         model$set_weights(self$best_weights)       }     }   },    on_train_end = function(logs = NULL)     if (self$stopped_epoch > 0)       cat(sprintf(\"Epoch %05d: early stopping\\n\", self$stopped_epoch + 1)) ) `add<-` <- `+`   model <- get_model() model |> fit(   mnist$train$x,   mnist$train$y,   epochs = 30,   batch_size = 64,   verbose = 0,   callbacks = list(callback_print_loss_and_mae(),                    callback_early_stopping_at_min_loss()) ) ## Up to batch 1, the average loss is   30.54. ## Up to batch 2, the average loss is  513.27. ## Up to batch 3, the average loss is  352.60. ## Up to batch 4, the average loss is  266.37. ## Up to batch 5, the average loss is  214.68. ## Up to batch 6, the average loss is  179.97. ## Up to batch 7, the average loss is  155.06. ## Up to batch 8, the average loss is  136.59. ## Up to batch 9, the average loss is  121.96. ## Up to batch 10, the average loss is  110.28. ## Up to batch 11, the average loss is  100.72. ## Up to batch 12, the average loss is   92.71. ## Up to batch 13, the average loss is   85.95. ## Up to batch 14, the average loss is   80.21. ## Up to batch 15, the average loss is   75.17. ## Up to batch 16, the average loss is   72.48. ## The average loss for epoch  1 is     72.48 and mean absolute error is    4.08. ## Up to batch 1, the average loss is    7.98. ## Up to batch 2, the average loss is    9.92. ## Up to batch 3, the average loss is   12.88. ## Up to batch 4, the average loss is   16.61. ## Up to batch 5, the average loss is   20.49. ## Up to batch 6, the average loss is   26.14. ## Up to batch 7, the average loss is   30.44. ## Up to batch 8, the average loss is   33.76. ## Up to batch 9, the average loss is   36.32. ## Up to batch 10, the average loss is   35.26. ## Up to batch 11, the average loss is   34.22. ## Up to batch 12, the average loss is   33.53. ## Up to batch 13, the average loss is   32.84. ## Up to batch 14, the average loss is   31.80. ## Up to batch 15, the average loss is   31.39. ## Up to batch 16, the average loss is   31.45. ## The average loss for epoch  2 is     31.45 and mean absolute error is    4.82. ## Up to batch 1, the average loss is   39.60. ## Up to batch 2, the average loss is   41.95. ## Up to batch 3, the average loss is   41.29. ## Up to batch 4, the average loss is   36.77. ## Up to batch 5, the average loss is   32.08. ## Up to batch 6, the average loss is   28.17. ## Up to batch 7, the average loss is   25.33. ## Up to batch 8, the average loss is   23.56. ## Up to batch 9, the average loss is   22.28. ## Up to batch 10, the average loss is   21.22. ## Up to batch 11, the average loss is   20.87. ## Up to batch 12, the average loss is   22.25. ## Up to batch 13, the average loss is   25.08. ## Up to batch 14, the average loss is   27.87. ## Up to batch 15, the average loss is   31.72. ## Up to batch 16, the average loss is   33.21. ## The average loss for epoch  3 is     33.21 and mean absolute error is    4.79. ## Restoring model weights from the end of the best epoch. ## Epoch 00004: early stopping"},{"path":"https://keras3.posit.co/dev/articles/writing_your_own_callbacks.html","id":"learning-rate-scheduling","dir":"Articles","previous_headings":"Examples of Keras callback applications","what":"Learning rate scheduling","title":"Writing your own callbacks","text":"example, show custom Callback can used dynamically change learning rate optimizer course training. See keras$callbacks$LearningRateScheduler general implementations (RStudio, press F1 cursor LearningRateScheduler browser open page).","code":"callback_custom_learning_rate_scheduler <- Callback(   \"CustomLearningRateScheduler\",   `__doc__` =   \"Learning rate scheduler which sets the learning rate according to schedule.      Arguments:         schedule: a function that takes an epoch index             (integer, indexed from 0) and current learning rate             as inputs and returns a new learning rate as output (float).     \",    initialize = function(schedule) {     super$initialize()     self$schedule <- schedule   },    on_epoch_begin = function(epoch, logs = NULL) {     ## When in doubt about what types of objects are in scope (e.g., self$model)     ## use a debugger to interact with the actual objects at the console!     # browser()      if (!\"learning_rate\" %in% names(self$model$optimizer))       stop('Optimizer must have a \"learning_rate\" attribute.')      # # Get the current learning rate from model's optimizer.     # use as.numeric() to convert the keras variablea to an R numeric     lr <- as.numeric(self$model$optimizer$learning_rate)     # # Call schedule function to get the scheduled learning rate.     scheduled_lr <- self$schedule(epoch, lr)     # # Set the value back to the optimizer before this epoch starts     optimizer <- self$model$optimizer     optimizer$learning_rate <- scheduled_lr     cat(sprintf(\"\\nEpoch %03d: Learning rate is %6.4f.\\n\", epoch, scheduled_lr))   } )  LR_SCHEDULE <- tibble::tribble(   ~start_epoch, ~learning_rate,              0,            0.1,              3,           0.05,              6,           0.01,              9,          0.005,             12,          0.001,   )  last <- function(x) x[length(x)] lr_schedule <- function(epoch, learning_rate) {   \"Helper function to retrieve the scheduled learning rate based on epoch.\"   with(LR_SCHEDULE, learning_rate[last(which(epoch >= start_epoch))]) }  model <- get_model() model |> fit(   mnist$train$x,   mnist$train$y,   epochs = 14,   batch_size = 64,   verbose = 0,   callbacks = list(     callback_print_loss_and_mae(),     callback_custom_learning_rate_scheduler(lr_schedule)   ) ) ## ## Epoch 001: Learning rate is 0.1000. ## Up to batch 1, the average loss is   29.36. ## Up to batch 2, the average loss is  513.95. ## Up to batch 3, the average loss is  352.70. ## Up to batch 4, the average loss is  266.46. ## Up to batch 5, the average loss is  214.73. ## Up to batch 6, the average loss is  180.00. ## Up to batch 7, the average loss is  155.05. ## Up to batch 8, the average loss is  136.64. ## Up to batch 9, the average loss is  121.97. ## Up to batch 10, the average loss is  110.30. ## Up to batch 11, the average loss is  100.76. ## Up to batch 12, the average loss is   92.74. ## Up to batch 13, the average loss is   85.95. ## Up to batch 14, the average loss is   80.18. ## Up to batch 15, the average loss is   75.11. ## Up to batch 16, the average loss is   72.38. ## The average loss for epoch  1 is     72.38 and mean absolute error is    4.04. ## ## Epoch 002: Learning rate is 0.1000. ## Up to batch 1, the average loss is    6.95. ## Up to batch 2, the average loss is    8.71. ## Up to batch 3, the average loss is   11.42. ## Up to batch 4, the average loss is   15.15. ## Up to batch 5, the average loss is   19.28. ## Up to batch 6, the average loss is   25.54. ## Up to batch 7, the average loss is   30.38. ## Up to batch 8, the average loss is   33.95. ## Up to batch 9, the average loss is   36.58. ## Up to batch 10, the average loss is   35.46. ## Up to batch 11, the average loss is   34.34. ## Up to batch 12, the average loss is   33.51. ## Up to batch 13, the average loss is   32.67. ## Up to batch 14, the average loss is   31.54. ## Up to batch 15, the average loss is   31.05. ## Up to batch 16, the average loss is   31.09. ## The average loss for epoch  2 is     31.09 and mean absolute error is    4.77. ## ## Epoch 003: Learning rate is 0.0500. ## Up to batch 1, the average loss is   40.40. ## Up to batch 2, the average loss is   22.33. ## Up to batch 3, the average loss is   16.18. ## Up to batch 4, the average loss is   13.09. ## Up to batch 5, the average loss is   11.48. ## Up to batch 6, the average loss is   10.21. ## Up to batch 7, the average loss is    9.22. ## Up to batch 8, the average loss is    8.70. ## Up to batch 9, the average loss is    8.16. ## Up to batch 10, the average loss is    7.80. ## Up to batch 11, the average loss is    7.50. ## Up to batch 12, the average loss is    7.17. ## Up to batch 13, the average loss is    6.89. ## Up to batch 14, the average loss is    6.70. ## Up to batch 15, the average loss is    6.52. ## Up to batch 16, the average loss is    6.54. ## The average loss for epoch  3 is      6.54 and mean absolute error is    1.93. ## ## Epoch 004: Learning rate is 0.0500. ## Up to batch 1, the average loss is    8.74. ## Up to batch 2, the average loss is    8.34. ## Up to batch 3, the average loss is    9.09. ## Up to batch 4, the average loss is    9.72. ## Up to batch 5, the average loss is   10.48. ## Up to batch 6, the average loss is   11.69. ## Up to batch 7, the average loss is   11.83. ## Up to batch 8, the average loss is   11.56. ## Up to batch 9, the average loss is   11.24. ## Up to batch 10, the average loss is   10.84. ## Up to batch 11, the average loss is   10.66. ## Up to batch 12, the average loss is   10.44. ## Up to batch 13, the average loss is   10.21. ## Up to batch 14, the average loss is   10.06. ## Up to batch 15, the average loss is   10.00. ## Up to batch 16, the average loss is   10.20. ## The average loss for epoch  4 is     10.20 and mean absolute error is    2.71. ## ## Epoch 005: Learning rate is 0.0500. ## Up to batch 1, the average loss is   17.26. ## Up to batch 2, the average loss is   14.09. ## Up to batch 3, the average loss is   12.67. ## Up to batch 4, the average loss is   11.44. ## Up to batch 5, the average loss is   10.54. ## Up to batch 6, the average loss is   10.10. ## Up to batch 7, the average loss is    9.53. ## Up to batch 8, the average loss is    9.17. ## Up to batch 9, the average loss is    8.78. ## Up to batch 10, the average loss is    8.49. ## Up to batch 11, the average loss is    8.50. ## Up to batch 12, the average loss is    8.59. ## Up to batch 13, the average loss is    8.68. ## Up to batch 14, the average loss is    8.86. ## Up to batch 15, the average loss is    9.17. ## Up to batch 16, the average loss is    9.53. ## The average loss for epoch  5 is      9.53 and mean absolute error is    2.58. ## ## Epoch 006: Learning rate is 0.0100. ## Up to batch 1, the average loss is   17.04. ## Up to batch 2, the average loss is   14.85. ## Up to batch 3, the average loss is   11.53. ## Up to batch 4, the average loss is    9.65. ## Up to batch 5, the average loss is    8.44. ## Up to batch 6, the average loss is    7.50. ## Up to batch 7, the average loss is    6.74. ## Up to batch 8, the average loss is    6.56. ## Up to batch 9, the average loss is    6.18. ## Up to batch 10, the average loss is    5.87. ## Up to batch 11, the average loss is    5.63. ## Up to batch 12, the average loss is    5.45. ## Up to batch 13, the average loss is    5.23. ## Up to batch 14, the average loss is    5.12. ## Up to batch 15, the average loss is    4.96. ## Up to batch 16, the average loss is    4.91. ## The average loss for epoch  6 is      4.91 and mean absolute error is    1.67. ## ## Epoch 007: Learning rate is 0.0100. ## Up to batch 1, the average loss is    3.65. ## Up to batch 2, the average loss is    3.04. ## Up to batch 3, the average loss is    2.88. ## Up to batch 4, the average loss is    2.85. ## Up to batch 5, the average loss is    2.88. ## Up to batch 6, the average loss is    2.81. ## Up to batch 7, the average loss is    2.70. ## Up to batch 8, the average loss is    2.96. ## Up to batch 9, the average loss is    2.96. ## Up to batch 10, the average loss is    2.93. ## Up to batch 11, the average loss is    2.95. ## Up to batch 12, the average loss is    2.98. ## Up to batch 13, the average loss is    2.97. ## Up to batch 14, the average loss is    3.01. ## Up to batch 15, the average loss is    3.00. ## Up to batch 16, the average loss is    3.05. ## The average loss for epoch  7 is      3.05 and mean absolute error is    1.34. ## ## Epoch 008: Learning rate is 0.0100. ## Up to batch 1, the average loss is    3.69. ## Up to batch 2, the average loss is    3.21. ## Up to batch 3, the average loss is    3.00. ## Up to batch 4, the average loss is    2.91. ## Up to batch 5, the average loss is    2.94. ## Up to batch 6, the average loss is    2.85. ## Up to batch 7, the average loss is    2.72. ## Up to batch 8, the average loss is    2.95. ## Up to batch 9, the average loss is    2.97. ## Up to batch 10, the average loss is    2.93. ## Up to batch 11, the average loss is    2.96. ## Up to batch 12, the average loss is    2.98. ## Up to batch 13, the average loss is    2.99. ## Up to batch 14, the average loss is    3.05. ## Up to batch 15, the average loss is    3.08. ## Up to batch 16, the average loss is    3.14. ## The average loss for epoch  8 is      3.14 and mean absolute error is    1.36. ## ## Epoch 009: Learning rate is 0.0050. ## Up to batch 1, the average loss is    3.71. ## Up to batch 2, the average loss is    2.93. ## Up to batch 3, the average loss is    2.76. ## Up to batch 4, the average loss is    2.70. ## Up to batch 5, the average loss is    2.76. ## Up to batch 6, the average loss is    2.69. ## Up to batch 7, the average loss is    2.57. ## Up to batch 8, the average loss is    2.79. ## Up to batch 9, the average loss is    2.80. ## Up to batch 10, the average loss is    2.77. ## Up to batch 11, the average loss is    2.79. ## Up to batch 12, the average loss is    2.80. ## Up to batch 13, the average loss is    2.78. ## Up to batch 14, the average loss is    2.81. ## Up to batch 15, the average loss is    2.80. ## Up to batch 16, the average loss is    2.83. ## The average loss for epoch  9 is      2.83 and mean absolute error is    1.28. ## ## Epoch 010: Learning rate is 0.0050. ## Up to batch 1, the average loss is    3.02. ## Up to batch 2, the average loss is    2.69. ## Up to batch 3, the average loss is    2.58. ## Up to batch 4, the average loss is    2.57. ## Up to batch 5, the average loss is    2.65. ## Up to batch 6, the average loss is    2.60. ## Up to batch 7, the average loss is    2.48. ## Up to batch 8, the average loss is    2.72. ## Up to batch 9, the average loss is    2.74. ## Up to batch 10, the average loss is    2.71. ## Up to batch 11, the average loss is    2.74. ## Up to batch 12, the average loss is    2.75. ## Up to batch 13, the average loss is    2.74. ## Up to batch 14, the average loss is    2.77. ## Up to batch 15, the average loss is    2.77. ## Up to batch 16, the average loss is    2.80. ## The average loss for epoch 10 is      2.80 and mean absolute error is    1.27. ## ## Epoch 011: Learning rate is 0.0050. ## Up to batch 1, the average loss is    3.01. ## Up to batch 2, the average loss is    2.69. ## Up to batch 3, the average loss is    2.58. ## Up to batch 4, the average loss is    2.56. ## Up to batch 5, the average loss is    2.63. ## Up to batch 6, the average loss is    2.58. ## Up to batch 7, the average loss is    2.47. ## Up to batch 8, the average loss is    2.70. ## Up to batch 9, the average loss is    2.72. ## Up to batch 10, the average loss is    2.69. ## Up to batch 11, the average loss is    2.71. ## Up to batch 12, the average loss is    2.72. ## Up to batch 13, the average loss is    2.71. ## Up to batch 14, the average loss is    2.75. ## Up to batch 15, the average loss is    2.74. ## Up to batch 16, the average loss is    2.77. ## The average loss for epoch 11 is      2.77 and mean absolute error is    1.27. ## ## Epoch 012: Learning rate is 0.0010. ## Up to batch 1, the average loss is    2.96. ## Up to batch 2, the average loss is    2.53. ## Up to batch 3, the average loss is    2.47. ## Up to batch 4, the average loss is    2.46. ## Up to batch 5, the average loss is    2.54. ## Up to batch 6, the average loss is    2.48. ## Up to batch 7, the average loss is    2.39. ## Up to batch 8, the average loss is    2.60. ## Up to batch 9, the average loss is    2.62. ## Up to batch 10, the average loss is    2.59. ## Up to batch 11, the average loss is    2.61. ## Up to batch 12, the average loss is    2.62. ## Up to batch 13, the average loss is    2.60. ## Up to batch 14, the average loss is    2.64. ## Up to batch 15, the average loss is    2.62. ## Up to batch 16, the average loss is    2.64. ## The average loss for epoch 12 is      2.64 and mean absolute error is    1.24. ## ## Epoch 013: Learning rate is 0.0010. ## Up to batch 1, the average loss is    2.82. ## Up to batch 2, the average loss is    2.46. ## Up to batch 3, the average loss is    2.42. ## Up to batch 4, the average loss is    2.42. ## Up to batch 5, the average loss is    2.50. ## Up to batch 6, the average loss is    2.45. ## Up to batch 7, the average loss is    2.36. ## Up to batch 8, the average loss is    2.57. ## Up to batch 9, the average loss is    2.59. ## Up to batch 10, the average loss is    2.57. ## Up to batch 11, the average loss is    2.59. ## Up to batch 12, the average loss is    2.60. ## Up to batch 13, the average loss is    2.59. ## Up to batch 14, the average loss is    2.62. ## Up to batch 15, the average loss is    2.61. ## Up to batch 16, the average loss is    2.63. ## The average loss for epoch 13 is      2.63 and mean absolute error is    1.23. ## ## Epoch 014: Learning rate is 0.0010. ## Up to batch 1, the average loss is    2.79. ## Up to batch 2, the average loss is    2.44. ## Up to batch 3, the average loss is    2.40. ## Up to batch 4, the average loss is    2.41. ## Up to batch 5, the average loss is    2.49. ## Up to batch 6, the average loss is    2.44. ## Up to batch 7, the average loss is    2.34. ## Up to batch 8, the average loss is    2.56. ## Up to batch 9, the average loss is    2.58. ## Up to batch 10, the average loss is    2.56. ## Up to batch 11, the average loss is    2.58. ## Up to batch 12, the average loss is    2.59. ## Up to batch 13, the average loss is    2.58. ## Up to batch 14, the average loss is    2.61. ## Up to batch 15, the average loss is    2.60. ## Up to batch 16, the average loss is    2.62. ## The average loss for epoch 14 is      2.62 and mean absolute error is    1.23."},{"path":"https://keras3.posit.co/dev/articles/writing_your_own_callbacks.html","id":"built-in-keras-callbacks","dir":"Articles","previous_headings":"Examples of Keras callback applications","what":"Built-in Keras callbacks","title":"Writing your own callbacks","text":"sure check existing Keras callbacks reading API docs. Applications include logging CSV, saving model, visualizing metrics TensorBoard, lot !","code":""},{"path":"https://keras3.posit.co/dev/authors.html","id":null,"dir":"","previous_headings":"","what":"Authors","title":"Authors and Citation","text":"Tomasz Kalinowski. Author, copyright holder, maintainer. Daniel Falbel. Contributor, copyright holder. JJ Allaire. Author, copyright holder. François Chollet. Author, copyright holder. Posit Software, PBC. Copyright holder, funder. Google. Copyright holder, funder. Yuan Tang. Contributor, copyright holder. Wouter Van Der Bijl. Contributor, copyright holder. Martin Studer. Contributor, copyright holder. Sigrid Keydana. Contributor.","code":""},{"path":"https://keras3.posit.co/dev/authors.html","id":"citation","dir":"","previous_headings":"","what":"Citation","title":"Authors and Citation","text":"Kalinowski T, Allaire J, Chollet F (2025). keras3: R Interface 'Keras'. R package version 1.4.0.9000, https://keras3.posit.co/.","code":"@Manual{,   title = {keras3: R Interface to 'Keras'},   author = {Tomasz Kalinowski and JJ Allaire and François Chollet},   year = {2025},   note = {R package version 1.4.0.9000},   url = {https://keras3.posit.co/}, }"},{"path":"https://keras3.posit.co/dev/index.html","id":"r-interface-to-keras","dir":"","previous_headings":"","what":"R Interface to Keras","title":"R Interface to Keras","text":"Keras high-level neural networks API developed focus enabling fast experimentation. able go idea result least possible delay key good research. Keras following key features: Allows code run CPU GPU, seamlessly. User-friendly API makes easy quickly prototype deep learning models. Built-support convolutional networks (computer vision), recurrent networks (sequence processing), combination . Supports arbitrary network architectures: multi-input multi-output models, layer sharing, model sharing, etc. means Keras appropriate building essentially deep learning model, memory network neural Turing machine. See package website https://keras3.posit.co complete documentation.","code":""},{"path":"https://keras3.posit.co/dev/reference/Callback.html","id":null,"dir":"Reference","previous_headings":"","what":"Define a custom Callback class — Callback","title":"Define a custom Callback class — Callback","text":"Callbacks can passed keras methods fit(), evaluate(), predict() order hook various stages model training, evaluation, inference lifecycle. create custom callback, call Callback() override method associated stage interest.","code":""},{"path":"https://keras3.posit.co/dev/reference/Callback.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Define a custom Callback class — Callback","text":"","code":"Callback(   classname,   on_epoch_begin = NULL,   on_epoch_end = NULL,   on_train_begin = NULL,   on_train_end = NULL,   on_train_batch_begin = NULL,   on_train_batch_end = NULL,   on_test_begin = NULL,   on_test_end = NULL,   on_test_batch_begin = NULL,   on_test_batch_end = NULL,   on_predict_begin = NULL,   on_predict_end = NULL,   on_predict_batch_begin = NULL,   on_predict_batch_end = NULL,   ...,   public = list(),   private = list(),   inherit = NULL,   parent_env = parent.frame() )"},{"path":"https://keras3.posit.co/dev/reference/Callback.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Define a custom Callback class — Callback","text":"classname String, name custom class. (Conventionally, CamelCase). on_epoch_begin   Called start epoch. Subclasses override actions run. function called TRAIN mode. Args: epoch: Integer, index epoch. logs: Named List. Currently data passed argument method may change future. on_epoch_end   Called end epoch. Subclasses override actions run. function called TRAIN mode. Args: epoch: Integer, index epoch. logs: Named List, metric results training epoch, validation epoch validation performed. Validation result keys prefixed val_. training epoch, values Model's metrics returned. Example: list(loss = 0.2, accuracy = 0.7). on_train_begin   Called beginning training. Subclasses override actions run. Args: logs: Named list. Currently data passed argument method may change future. on_train_end   Called end training. Subclasses override actions run. Args: logs: Named list. Currently output last call on_epoch_end() passed argument method may change future. on_train_batch_begin   Called beginning training batch fit() methods. Subclasses override actions run. Note steps_per_execution argument compile Model set N, method called every N batches. Args: batch: Integer, index batch within current epoch. logs: Named list. Currently data passed argument method may change future. on_train_batch_end   Called end training batch fit() methods. Subclasses override actions run. Note steps_per_execution argument compile Model set N, method called every N batches. Args: batch: Integer, index batch within current epoch. logs: Named list. Aggregated metric results batch. on_test_begin   Called beginning evaluation validation. Subclasses override actions run. Args: logs: Named list. Currently data passed argument method may change future. on_test_end   Called end evaluation validation. Subclasses override actions run. Args: logs: Named list. Currently output last call on_test_batch_end() passed argument method may change future. on_test_batch_begin   Called beginning batch evaluate() methods. Also called beginning validation batch fit() methods, validation data provided. Subclasses override actions run. Note steps_per_execution argument compile() Model set N, method called every N batches. Args: batch: Integer, index batch within current epoch. logs: Named list. Currently data passed argument method may change future. on_test_batch_end   Called end batch evaluate() methods. Also called end validation batch fit() methods, validation data provided. Subclasses override actions run. Note steps_per_execution argument compile() Model set N, method called every N batches. Args: batch: Integer, index batch within current epoch. logs: Named list. Aggregated metric results batch. on_predict_begin   Called beginning prediction. Subclasses override actions run. Args: logs: Named list. Currently data passed argument method may change future. on_predict_end   Called end prediction. Subclasses override actions run. Args: logs: Named list. Currently data passed argument method may change future. on_predict_batch_begin   Called beginning batch predict() methods. Subclasses override actions run. Note steps_per_execution argument compile() Model set N, method called every N batches. Args: batch: Integer, index batch within current epoch. logs: Named list. Currently data passed argument method may change future. on_predict_batch_end   Called end batch predict() methods. Subclasses override actions run. Note steps_per_execution argument compile Model set N, method called every N batches. Args: batch: Integer, index batch within current epoch. logs: Named list. Aggregated metric results batch. ..., public Additional methods public members custom class. private Named list R objects (typically, functions) include instance private environments. private methods symbols scope public methods (See section \"Symbols Scope\"). instance private environment. objects private invisible Keras framework Python runtime. inherit custom class subclass. default, base keras class. parent_env R environment class methods grandparent.","code":"\\(epoch, logs = NULL) \\(epoch, logs = NULL) \\(logs = NULL) \\(logs = NULL) \\(batch, logs = NULL) \\(batch, logs=NULL) \\(logs = NULL) \\(logs = NULL) \\(batch, logs = NULL) \\(batch, logs = NULL) \\(logs = NULL) \\(logs = NULL) \\(batch, logs = NULL) \\(batch, logs = NULL)"},{"path":"https://keras3.posit.co/dev/reference/Callback.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Define a custom Callback class — Callback","text":"function returns custom Callback instances, similar builtin callback functions.","code":""},{"path":"https://keras3.posit.co/dev/reference/Callback.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Define a custom Callback class — Callback","text":"R function custom methods (public private) following symbols scope: self: Layer instance. super: Layer superclass. private: R environment specific class instance. objects defined invisible Keras framework. __class__ current class type object. also available alias symbol, value supplied Layer(classname = )","code":"training_finished <- FALSE callback_mark_finished <- Callback(\"MarkFinished\",   on_train_end = function(logs = NULL) {     training_finished <<- TRUE   } )  model <- keras_model_sequential(input_shape = c(1)) |>   layer_dense(1) model |> compile(loss = 'mean_squared_error') model |> fit(op_ones(c(1, 1)), op_ones(c(1, 1)),              callbacks = callback_mark_finished()) stopifnot(isTRUE(training_finished))"},{"path":"https://keras3.posit.co/dev/reference/Callback.html","id":"attributes-accessible-via-self-","dir":"Reference","previous_headings":"","what":"Attributes (accessible via self$)","title":"Define a custom Callback class — Callback","text":"params: Named list, Training parameters (e.g. verbosity, batch size, number epochs, ...). model: Instance Model. Reference model trained. logs named list callback methods take argument contain keys quantities relevant current batch epoch (see method-specific docstrings).","code":""},{"path":"https://keras3.posit.co/dev/reference/Callback.html","id":"symbols-in-scope","dir":"Reference","previous_headings":"","what":"Symbols in scope","title":"Define a custom Callback class — Callback","text":"R function custom methods (public private) following symbols scope: self: custom class instance. super: custom class superclass. private: R environment specific class instance. objects assigned invisible Keras framework. __class__ .symbol(classname): custom class type object.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/Constraint.html","id":null,"dir":"Reference","previous_headings":"","what":"Define a custom Constraint class — Constraint","title":"Define a custom Constraint class — Constraint","text":"Base class weight constraints. Constraint() instance works like stateless function. Users subclass Constraint class override call() method, takes single weight parameter return projected version parameter (e.g. normalized clipped). Constraints can used various Keras layers via kernel_constraint bias_constraint arguments. simple example non-negative weight constraint:     Usage layer:","code":"constraint_nonnegative <- Constraint(\"NonNegative\",   call = function(w) {     w * op_cast(w >= 0, dtype = w$dtype)   } ) weight <- op_convert_to_tensor(c(-1, 1)) constraint_nonnegative()(weight) ## tf.Tensor([-0.  1.], shape=(2), dtype=float64) layer_dense(units = 4, kernel_constraint = constraint_nonnegative()) ## <Dense name=dense, built=False> ##  signature: (*args, **kwargs)"},{"path":"https://keras3.posit.co/dev/reference/Constraint.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Define a custom Constraint class — Constraint","text":"","code":"Constraint(   classname,   call = NULL,   get_config = NULL,   ...,   public = list(),   private = list(),   inherit = NULL,   parent_env = parent.frame() )"},{"path":"https://keras3.posit.co/dev/reference/Constraint.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Define a custom Constraint class — Constraint","text":"classname String, name custom class. (Conventionally, CamelCase). call   Applies constraint input weight variable. default, inputs weight variable modified. Users override method implement projection function. Args: w: Input weight variable. Returns: Projected variable (default, returns unmodified inputs). get_config   Function returns named list object config. constraint config named list (JSON-serializable) can used reinstantiate object (via .call(<constraint_class>, <config>)). ..., public Additional methods public members custom class. private Named list R objects (typically, functions) include instance private environments. private methods symbols scope public methods (See section \"Symbols Scope\"). instance private environment. objects private invisible Keras framework Python runtime. inherit custom class subclass. default, base keras class. parent_env R environment class methods grandparent.","code":"\\(w) \\()"},{"path":"https://keras3.posit.co/dev/reference/Constraint.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Define a custom Constraint class — Constraint","text":"function returns Constraint instances, similar builtin constraint functions like constraint_maxnorm().","code":""},{"path":"https://keras3.posit.co/dev/reference/Constraint.html","id":"symbols-in-scope","dir":"Reference","previous_headings":"","what":"Symbols in scope","title":"Define a custom Constraint class — Constraint","text":"R function custom methods (public private) following symbols scope: self: custom class instance. super: custom class superclass. private: R environment specific class instance. objects assigned invisible Keras framework. __class__ .symbol(classname): custom class type object.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/Layer.html","id":null,"dir":"Reference","previous_headings":"","what":"Define a custom Layer class. — Layer","title":"Define a custom Layer class. — Layer","text":"layer callable object takes input one tensors outputs one tensors. involves computation, defined call() method, state (weight variables). State can created: initialize(), instance via self$add_weight(); optional build() method, invoked first call() layer, supplies shape(s) input(s), may known initialization time. Layers recursively composable: assign Layer instance attribute another Layer, outer layer start tracking weights created inner layer. Nested layers instantiated initialize() method build() method. Users just instantiate layer treat callable.","code":""},{"path":"https://keras3.posit.co/dev/reference/Layer.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Define a custom Layer class. — Layer","text":"","code":"Layer(   classname,   initialize = NULL,   call = NULL,   build = NULL,   get_config = NULL,   ...,   public = list(),   private = list(),   inherit = NULL,   parent_env = parent.frame() )"},{"path":"https://keras3.posit.co/dev/reference/Layer.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Define a custom Layer class. — Layer","text":"classname String, name custom class. (Conventionally, CamelCase). initialize, call, build, get_config Recommended methods implement. See description details sections. ..., public Additional methods public members custom class. private Named list R objects (typically, functions) include instance private environments. private methods symbols scope public methods (See section \"Symbols Scope\"). instance private environment. objects private invisible Keras framework Python runtime. inherit custom class subclass. default, base keras class. parent_env R environment class methods grandparent.","code":""},{"path":"https://keras3.posit.co/dev/reference/Layer.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Define a custom Layer class. — Layer","text":"composing layer constructor, similar behavior layer functions like layer_dense(). first argument returned function object, enabling initialize()ing call() layer one step composing layer pipe, like   initialize() layer instance call() , pass missing NULL value object, pass arguments initialize() name.","code":"layer_foo <- Layer(\"Foo\", ....) output <- inputs |> layer_foo() layer <- layer_dense(units = 2, activation = \"relu\") layer <- layer_dense(NULL, 2, activation = \"relu\") layer <- layer_dense(, 2, activation = \"relu\")  # then you can call() the layer in a separate step outputs <- inputs |> layer()"},{"path":"https://keras3.posit.co/dev/reference/Layer.html","id":"symbols-in-scope","dir":"Reference","previous_headings":"","what":"Symbols in scope","title":"Define a custom Layer class. — Layer","text":"R function custom methods (public private) following symbols scope: self: custom class instance. super: custom class superclass. private: R environment specific class instance. objects assigned invisible Keras framework. __class__ .symbol(classname): custom class type object.","code":""},{"path":"https://keras3.posit.co/dev/reference/Layer.html","id":"attributes","dir":"Reference","previous_headings":"","what":"Attributes","title":"Define a custom Layer class. — Layer","text":"name: name layer (string). dtype: Dtype layer's weights. Alias layer$variable_dtype. variable_dtype: Dtype layer's weights. compute_dtype: dtype layer's computations. Layers automatically cast inputs dtype, causes computations output also dtype. mixed precision used keras$mixed_precision$DTypePolicy, different variable_dtype. trainable_weights: List variables included backprop. non_trainable_weights: List variables included backprop. weights: concatenation lists trainable_weights non_trainable_weights (order). trainable: Whether layer trained (boolean), .e. whether potentially-trainable weights returned part layer$trainable_weights. input_spec: Optional (list ) InputSpec object(s) specifying constraints inputs can accepted layer. recommend custom Layers implement following methods: initialize(): Defines custom layer attributes, creates layer weights depend input shapes, using add_weight(), state. build(input_shape): method can used create weights depend shape(s) input(s), using add_weight(), state. Calling call() automatically build layer (built yet) calling build(). call(...): Method called making sure build() called. call() performs logic applying layer input arguments. Two reserved arguments can optionally use call() : training (boolean, whether call inference mode training mode). mask (boolean tensor encoding masked timesteps input, used e.g. RNN layers). typical signature method call(inputs), user optionally add training mask layer need . get_config(): Returns named list containing configuration used initialize layer. list names differ arguments initialize(), override from_config() well. method used saving layer model contains layer.","code":""},{"path":"https://keras3.posit.co/dev/reference/Layer.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Define a custom Layer class. — Layer","text":"basic example: layer two variables, w b, returns y <- (w %*% x) + b. shows implement build() call(). Variables set attributes layer tracked weights layers (layer$weights).   Besides trainable weights, updated via backpropagation training, layers can also non-trainable weights. weights meant updated manually call(). example layer computes running sum inputs:","code":"layer_simple_dense <- Layer(   \"SimpleDense\",   initialize = function(units = 32) {     super$initialize()     self$units <- units   },    # Create the state of the layer (weights)   build = function(input_shape) {     self$kernel <- self$add_weight(       shape = shape(tail(input_shape, 1), self$units),       initializer = \"glorot_uniform\",       trainable = TRUE,       name = \"kernel\"     )     self$bias = self$add_weight(       shape = shape(self$units),       initializer = \"zeros\",       trainable = TRUE,       name = \"bias\"     )   },    # Defines the computation   call = function(self, inputs) {     op_matmul(inputs, self$kernel) + self$bias   } )  # Instantiates the layer. # Supply missing `object` arg to skip invoking `call()` and instead return # the Layer instance linear_layer <- layer_simple_dense(, 4)  # This will call `build(input_shape)` and create the weights, # and then invoke `call()`. y <- linear_layer(op_ones(c(2, 2))) stopifnot(length(linear_layer$weights) == 2)  # These weights are trainable, so they're listed in `trainable_weights`: stopifnot(length(linear_layer$trainable_weights) == 2) layer_compute_sum <- Layer(   classname = \"ComputeSum\",    initialize = function(input_dim) {     super$initialize()      # Create a non-trainable weight.     self$total <- self$add_weight(       shape = shape(),       initializer = \"zeros\",       trainable = FALSE,       name = \"total\"     )   },    call = function(inputs) {     self$total$assign(self$total + op_sum(inputs))     self$total   } )  my_sum <- layer_compute_sum(, 2) x <- op_ones(c(2, 2)) y <- my_sum(x)  stopifnot(exprs = {   all.equal(my_sum$weights,               list(my_sum$total))   all.equal(my_sum$non_trainable_weights, list(my_sum$total))   all.equal(my_sum$trainable_weights,     list()) })"},{"path":"https://keras3.posit.co/dev/reference/Layer.html","id":"methods-available","dir":"Reference","previous_headings":"","what":"Methods available","title":"Define a custom Layer class. — Layer","text":"Initialize self. method typically called custom initialize() method. Example:   Args: trainable: Boolean, whether layer's variables trainable. name: String name layer. dtype: dtype layer's computations weights. Can also keras$DTypePolicy, allows computation weight dtype differ. Defaults NULL. NULL means use config_dtype_policy(), \"float32\" policy unless set different value (via config_set_dtype_policy()).    Can called inside call() method add scalar loss. Example:         Add weight variable layer. Alias add_weight().    Add weight variable layer. Args: shape: shape variable (defined shape()) Must fully-defined (NA/NULL/-1 entries). Defaults () (scalar) unspecified. initializer: Initializer object use populate initial variable value, string name built-initializer (e.g. \"random_normal\"). unspecified, defaults \"glorot_uniform\" floating-point variables \"zeros\" types (e.g. int, bool). dtype: Dtype variable create, e.g. \"float32\". unspecified, defaults layer's variable dtype (defaults \"float32\" unspecified). trainable: Boolean, whether variable trainable via backprop whether updates managed manually. Defaults TRUE. autocast: Boolean, whether autocast layers variables accessing . Defaults TRUE. regularizer: Regularizer object call apply penalty weight. penalties summed loss function optimization. Defaults NULL. constraint: Constraint object call variable optimizer update, string name built-constraint. Defaults NULL. aggregation: Optional string, one NULL, \"none\", \"mean\", \"sum\" \"only_first_replica\". Annotates variable type multi-replica aggregation used variable writing custom data parallel training loops. Defaults \"none\". overwrite_with_gradient: Boolean, whether overwrite variable computed gradient. Useful float8 training. Defaults FALSE. name: String name variable. Useful debugging purposes. Returns: backend tensor, wrapped KerasVariable class. KerasVariable class Methods: assign(value) assign_add(value) assign_sub(value) numpy() (calling .array(<variable>) preferred) Properties/Attributes: value dtype ndim shape (calling shape(<variable>) preferred) trainable       Builds layer's states supplied config (named list args). default, method calls .call(build, config$input_shape) method, creates weights based layer's input shape supplied config. config contains information needed load layer's state, override method. Args: config: Named list containing input shape associated layer.    See description                Count total number scalars composing weights. Returns: integer count.    Returns named list layer's input shape. method returns config (named list) can used build_from_config(config) create states (e.g. Variables Lookup tables) needed layer. default, config contains input shape layer built . writing custom layer creates state unusual way, override method make sure state already created Keras attempts load value upon model loading. Returns: named list containing input shape associated layer.    Returns config object. object config named list (serializable) containing information needed re-instantiate . config expected serializable JSON, expected consist (potentially complex, nested) structure names lists consisting simple objects like strings, ints.    Return values layer$weights list R NumPy arrays.    Currently, Dense, EinsumDense Embedding layers support -place quantization via quantize() method. Example:            Enable rematerialization dynamically layer's call method. Args: layer_call: original call method layer. ...: additional args Returns: rematerialized version layer's call method.    Loads state layer. can override method take full control state layer loaded upon calling load_model(). Args: store: Named list state model loaded.    Saves state layer. can override method take full control state layer saved upon calling save_model(). Args: store: Named list state model saved.    Sets values weights list R NumPy arrays.    Call layer without side effects. Args: trainable_variables: List trainable variables model. non_trainable_variables: List non-trainable variables model. ...: Positional named arguments passed call(). return_losses: TRUE, stateless_call() return list losses created call() part return values. Returns: unnamed list. default, returns list(outputs, non_trainable_variables). return_losses = TRUE, returns list(outputs, non_trainable_variables, losses). Note: non_trainable_variables include non-trainable weights BatchNormalization statistics, also RNG seed state (random operations part layer, dropout), Metric state (metrics attached layer). elements state layer. Example:         Creates layer config. class method, meaning, R function self symbol (class instance) scope. Use __class__ classname symbol provided Layer() constructed) resolve class definition. default implementation :   method reverse get_config(), capable instantiating layer config named list. handle layer connectivity (handled Network), weights (handled set_weights()). Args: config: named list, typically output get_config(). Returns: layer instance.","code":"initialize(...,            activity_regularizer = NULL,            trainable = TRUE,            dtype = NULL,            autocast = TRUE,            name = NULL) layer_my_layer <- Layer(\"MyLayer\",   initialize = function(units, ..., dtype = NULL, name = NULL) {     super$initialize(..., dtype = dtype, name = name)     # .... finish initializing `self` instance   } ) add_loss(loss) Layer(\"MyLayer\",   ...   call = function(x) {     self$add_loss(op_sum(x))     x   } add_metric(...) add_variable(...) add_weight(shape = NULL,            initializer = NULL,            dtype = NULL,            trainable = TRUE,            autocast = TRUE,            regularizer = NULL,            constraint = NULL,            aggregation = 'none',            overwrite_with_gradient = FALSE,            name = NULL) build(input_shape) build_from_config(config) call(...) compute_mask(inputs, previous_mask) compute_output_shape(...) compute_output_spec(...) symbolic_call(...) count_params() get_build_config() get_config() get_weights() quantize(mode, type_check = TRUE) model$quantize(\"int8\") # quantize model in-place model |> predict(data) # faster inference quantized_build(input_shape, mode) quantized_call(...) rematerialized_call(layer_call, ...) load_own_variables(store) save_own_variables(store) set_weights(weights) stateless_call(trainable_variables, non_trainable_variables,                ..., return_losses = FALSE) model <- ... data <- ... trainable_variables <- model$trainable_variables non_trainable_variables <- model$non_trainable_variables # Call the model with zero side effects c(outputs, non_trainable_variables) %<-% model$stateless_call(     trainable_variables,     non_trainable_variables,     data ) # Attach the updated state to the model # (until you do this, the model is still in its pre-call state). purrr::walk2(   model$non_trainable_variables, non_trainable_variables,   \\(variable, value) variable$assign(value)) symbolic_call(...) from_config(config) from_config = function(config) {   do.call(`__class__`, config) }"},{"path":"https://keras3.posit.co/dev/reference/Layer.html","id":"readonly-properties-","dir":"Reference","previous_headings":"","what":"Readonly properties:","title":"Define a custom Layer class. — Layer","text":"compute_dtype dtype computations performed layer. dtype Alias layer$variable_dtype. input_dtype dtype layer inputs converted . losses List scalar losses add_loss(), regularizers sublayers. metrics List metrics. metrics_variables List metric variables. non_trainable_variables List non-trainable layer state. extends layer$non_trainable_weights include state used layer including state metrics SeedGenerators. non_trainable_weights List non-trainable weight variables layer. weights updated optimizer training. Unlike, layer$non_trainable_variables excludes metric state random seeds. trainable_variables List trainable layer state. equivalent layer$trainable_weights. trainable_weights List trainable weight variables layer. weights get updated optimizer training. path path layer. layer built yet, NULL. quantization_mode quantization mode layer, NULL quantized. variable_dtype dtype state (weights) layer. variables List layer state, including random seeds. extends layer$weights include state used layer including SeedGenerators. Note metrics variables included , use metrics_variables visit metric variables. weights List weight variables layer. Unlike, layer$variables excludes metric state random seeds. input Retrieves input tensor(s) symbolic operation. returns tensor(s) corresponding first time operation called. Returns: Input tensor list input tensors. output Retrieves output tensor(s) layer. returns tensor(s) corresponding first time operation called. Returns: Output tensor list output tensors.","code":""},{"path":"https://keras3.posit.co/dev/reference/Layer.html","id":"data-descriptors-attributes-","dir":"Reference","previous_headings":"","what":"Data descriptors (Attributes):","title":"Define a custom Layer class. — Layer","text":"dtype_policy input_spec supports_masking Whether layer supports computing mask using compute_mask. trainable Settable boolean, whether layer trainable .","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/LearningRateSchedule.html","id":null,"dir":"Reference","previous_headings":"","what":"Define a custom LearningRateSchedule class — LearningRateSchedule","title":"Define a custom LearningRateSchedule class — LearningRateSchedule","text":"Subclass Keras LearningRateSchedule base class. can use learning rate schedule modulate learning rate optimizer changes time. Several built-learning rate schedules available, learning_rate_schedule_exponential_decay() learning_rate_schedule_piecewise_constant_decay():   LearningRateSchedule() instance can passed learning_rate argument optimizer. implement schedule object, implement call method, takes step argument (scalar integer backend tensor, current training step count). Note step 0-based (.e., first step 0). Like Keras object, can also optionally make object serializable implementing get_config() from_config() methods.","code":"lr_schedule <- learning_rate_schedule_exponential_decay(   initial_learning_rate = 1e-2,   decay_steps = 10000,   decay_rate = 0.9 ) optimizer <- optimizer_sgd(learning_rate = lr_schedule)"},{"path":"https://keras3.posit.co/dev/reference/LearningRateSchedule.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Define a custom LearningRateSchedule class — LearningRateSchedule","text":"","code":"LearningRateSchedule(   classname,   call = NULL,   initialize = NULL,   get_config = NULL,   ...,   public = list(),   private = list(),   inherit = NULL,   parent_env = parent.frame() )"},{"path":"https://keras3.posit.co/dev/reference/LearningRateSchedule.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Define a custom LearningRateSchedule class — LearningRateSchedule","text":"classname String, name custom class. (Conventionally, CamelCase). call, initialize, get_config Recommended methods implement. See description details sections. ..., public Additional methods public members custom class. private Named list R objects (typically, functions) include instance private environments. private methods symbols scope public methods (See section \"Symbols Scope\"). instance private environment. objects private invisible Keras framework Python runtime. inherit custom class subclass. default, base keras class. parent_env R environment class methods grandparent.","code":""},{"path":"https://keras3.posit.co/dev/reference/LearningRateSchedule.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Define a custom LearningRateSchedule class — LearningRateSchedule","text":"function returns LearningRateSchedule instances, similar built-learning_rate_schedule_* family functions.","code":""},{"path":"https://keras3.posit.co/dev/reference/LearningRateSchedule.html","id":"example","dir":"Reference","previous_headings":"","what":"Example","title":"Define a custom LearningRateSchedule class — LearningRateSchedule","text":"","code":"my_custom_learning_rate_schedule <- LearningRateSchedule(   classname = \"MyLRSchedule\",    initialize = function(initial_learning_rate) {     self$initial_learning_rate <- initial_learning_rate   },    call = function(step) {     # note that `step` is a tensor     # and call() will be traced via tf_function() or similar.      str(step) # <KerasVariable shape=(), dtype=int64, path=SGD/iteration>      # print 'step' every 1000 steps     op_cond((step %% 1000) == 0,             \\() {tensorflow::tf$print(step); NULL},             \\() {NULL})     self$initial_learning_rate / (step + 1)   } )  optimizer <- optimizer_sgd(   learning_rate = my_custom_learning_rate_schedule(0.1) )  # You can also call schedule instances directly # (e.g., for interactive testing, or if implementing a custom optimizer) schedule <- my_custom_learning_rate_schedule(0.1) step <- keras$Variable(initializer = op_ones,                        shape = shape(),                        dtype = \"int64\") schedule(step) ## <Variable path=variable, shape=(), dtype=int64, value=1> ## tf.Tensor(0.0, shape=(), dtype=float64)"},{"path":"https://keras3.posit.co/dev/reference/LearningRateSchedule.html","id":"methods-available-","dir":"Reference","previous_headings":"","what":"Methods available:","title":"Define a custom LearningRateSchedule class — LearningRateSchedule","text":"","code":"get_config()"},{"path":"https://keras3.posit.co/dev/reference/LearningRateSchedule.html","id":"symbols-in-scope","dir":"Reference","previous_headings":"","what":"Symbols in scope","title":"Define a custom LearningRateSchedule class — LearningRateSchedule","text":"R function custom methods (public private) following symbols scope: self: custom class instance. super: custom class superclass. private: R environment specific class instance. objects assigned invisible Keras framework. __class__ .symbol(classname): custom class type object.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/Loss.html","id":null,"dir":"Reference","previous_headings":"","what":"Subclass the base Loss class — Loss","title":"Subclass the base Loss class — Loss","text":"Use define custom loss class. Note, cases need subclass Loss define custom loss: can also pass bare R function, named R function defined custom_metric(), loss function compile().","code":""},{"path":"https://keras3.posit.co/dev/reference/Loss.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Subclass the base Loss class — Loss","text":"","code":"Loss(   classname,   call = NULL,   ...,   public = list(),   private = list(),   inherit = NULL,   parent_env = parent.frame() )"},{"path":"https://keras3.posit.co/dev/reference/Loss.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Subclass the base Loss class — Loss","text":"classname String, name custom class. (Conventionally, CamelCase). call   Method implemented subclasses: Function contains logic loss calculation using y_true, y_pred. ..., public Additional methods public members custom class. private Named list R objects (typically, functions) include instance private environments. private methods symbols scope public methods (See section \"Symbols Scope\"). instance private environment. objects private invisible Keras framework Python runtime. inherit custom class subclass. default, base keras class. parent_env R environment class methods grandparent.","code":"function(y_true, y_pred)"},{"path":"https://keras3.posit.co/dev/reference/Loss.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Subclass the base Loss class — Loss","text":"function returns Loss instances, similar builtin loss functions.","code":""},{"path":"https://keras3.posit.co/dev/reference/Loss.html","id":"details","dir":"Reference","previous_headings":"","what":"Details","title":"Subclass the base Loss class — Loss","text":"Example subclass implementation:","code":"loss_custom_mse <- Loss(   classname = \"CustomMeanSquaredError\",   call = function(y_true, y_pred) {     op_mean(op_square(y_pred - y_true), axis = -1)   } )  # Usage in compile() model <- keras_model_sequential(input_shape = 10) |> layer_dense(10) model |> compile(loss = loss_custom_mse())  # Standalone usage mse <- loss_custom_mse(name = \"my_custom_mse_instance\")  y_true <- op_arange(20) |> op_reshape(c(4, 5)) y_pred <- op_arange(20) |> op_reshape(c(4, 5)) * 2 (loss <- mse(y_true, y_pred)) ## tf.Tensor(143.5, shape=(), dtype=float32) loss2 <- (y_pred - y_true)^2 |>   op_mean(axis = -1) |>   op_mean()  stopifnot(all.equal(as.array(loss), as.array(loss2)))  sample_weight <-array(c(.25, .25, 1, 1)) (weighted_loss <- mse(y_true, y_pred, sample_weight = sample_weight)) ## tf.Tensor(129.0625, shape=(), dtype=float32) weighted_loss2 <- (y_true - y_pred)^2 |>   op_mean(axis = -1) |>   op_multiply(sample_weight) |>   op_mean()  stopifnot(all.equal(as.array(weighted_loss),                     as.array(weighted_loss2)))"},{"path":"https://keras3.posit.co/dev/reference/Loss.html","id":"methods-defined-by-base-loss-class-","dir":"Reference","previous_headings":"","what":"Methods defined by base Loss class:","title":"Subclass the base Loss class — Loss","text":"Args: name: Optional name loss instance. reduction: Type reduction apply loss. almost cases \"sum_over_batch_size\". Supported options \"sum\", \"sum_over_batch_size\", \"mean\", \"mean_with_sample_weight\" NULL. \"sum\" sums loss, \"sum_over_batch_size\" \"mean\" sum loss divide sample size, \"mean_with_sample_weight\" sums loss divides sum sample weights. \"none\" NULL perform aggregation. Defaults \"sum_over_batch_size\". dtype: dtype loss's computations. Defaults NULL, means using config_floatx(). config_floatx() \"float32\" unless set different value (via config_set_floatx()). keras$DTypePolicy provided, compute_dtype utilized.    Call loss instance function, optionally sample_weight.","code":"initialize(name=NULL, reduction=\"sum_over_batch_size\", dtype=NULL) __call__(y_true, y_pred, sample_weight=NULL) get_config()"},{"path":"https://keras3.posit.co/dev/reference/Loss.html","id":"readonly-properties-","dir":"Reference","previous_headings":"","what":"Readonly properties:","title":"Subclass the base Loss class — Loss","text":"","code":"dtype"},{"path":"https://keras3.posit.co/dev/reference/Loss.html","id":"symbols-in-scope","dir":"Reference","previous_headings":"","what":"Symbols in scope","title":"Subclass the base Loss class — Loss","text":"R function custom methods (public private) following symbols scope: self: custom class instance. super: custom class superclass. private: R environment specific class instance. objects assigned invisible Keras framework. __class__ .symbol(classname): custom class type object.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/Metric.html","id":null,"dir":"Reference","previous_headings":"","what":"Subclass the base Metric class — Metric","title":"Subclass the base Metric class — Metric","text":"Metric object encapsulates metric logic state can used track model performance training. returned family metric functions start prefix metric_*, well returned custom metrics defined Metric().","code":""},{"path":"https://keras3.posit.co/dev/reference/Metric.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Subclass the base Metric class — Metric","text":"","code":"Metric(   classname,   initialize = NULL,   update_state = NULL,   result = NULL,   ...,   public = list(),   private = list(),   inherit = NULL,   parent_env = parent.frame() )"},{"path":"https://keras3.posit.co/dev/reference/Metric.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Subclass the base Metric class — Metric","text":"classname String, name custom class. (Conventionally, CamelCase). initialize, update_state, result Recommended methods implement. See description section. ..., public Additional methods public members custom class. private Named list R objects (typically, functions) include instance private environments. private methods symbols scope public methods (See section \"Symbols Scope\"). instance private environment. objects private invisible Keras framework Python runtime. inherit custom class subclass. default, base keras class. parent_env R environment class methods grandparent.","code":""},{"path":"https://keras3.posit.co/dev/reference/Metric.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Subclass the base Metric class — Metric","text":"function returns Metric instances, similar builtin metric functions.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/Metric.html","id":"usage-with-compile-","dir":"Reference","previous_headings":"","what":"Usage with compile():","title":"Subclass the base Metric class — Metric","text":"","code":"model |> compile(   optimizer = 'sgd',   loss = 'mse',   metrics = c(metric_SOME_METRIC(), metric_SOME_OTHER_METRIC()) )"},{"path":"https://keras3.posit.co/dev/reference/Metric.html","id":"standalone-usage-","dir":"Reference","previous_headings":"","what":"Standalone usage:","title":"Subclass the base Metric class — Metric","text":"","code":"m <- metric_SOME_METRIC() for (e in seq(epochs)) {   for (i in seq(train_steps)) {     c(y_true, y_pred, sample_weight = NULL) %<-% ...     m$update_state(y_true, y_pred, sample_weight)   }   cat('Final epoch result: ', as.numeric(m$result()), \"\\n\")   m$reset_state() }"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/Metric.html","id":"usage-with-compile--1","dir":"Reference","previous_headings":"","what":"Usage with compile():","title":"Subclass the base Metric class — Metric","text":"implemented subclasses (custom metrics): initialize(): state variables created method calling self$add_variable() like: self$var <- self$add_variable(...). update_state(): Updates state variables like: self$var$assign(...). result(): Computes returns scalar value named list scalar values metric state variables. Example subclass implementation:","code":"model <- keras_model_sequential() model |>   layer_dense(64, activation = \"relu\") |>   layer_dense(64, activation = \"relu\") |>   layer_dense(10, activation = \"softmax\") model |>   compile(optimizer = optimizer_rmsprop(0.01),           loss = loss_categorical_crossentropy(),           metrics = metric_categorical_accuracy())  data <- random_uniform(c(1000, 32)) labels <- random_uniform(c(1000, 10))  model |> fit(data, labels, verbose = 0) metric_binary_true_positives <- Metric(   classname = \"BinaryTruePositives\",    initialize = function(name = 'binary_true_positives', ...) {     super$initialize(name = name, ...)     self$true_positives <-       self$add_weight(shape = shape(),                       initializer = 'zeros',                       name = 'true_positives')   },    update_state = function(y_true, y_pred, sample_weight = NULL) {     y_true <- op_cast(y_true, \"bool\")     y_pred <- op_cast(y_pred, \"bool\")      values <- y_true & y_pred # `&` calls op_logical_and()     values <- op_cast(values, self$dtype)     if (!is.null(sample_weight)) {       sample_weight <- op_cast(sample_weight, self$dtype)       sample_weight <- op_broadcast_to(sample_weight, shape(values))       values <- values * sample_weight # `*` calls op_multiply()     }     self$true_positives$assign(self$true_positives + op_sum(values))   },    result = function() {     self$true_positives   } ) model <- keras_model_sequential(input_shape = 32) |> layer_dense(10) model |> compile(loss = loss_binary_crossentropy(),                  metrics = list(metric_binary_true_positives())) model |> fit(data, labels, verbose = 0)"},{"path":"https://keras3.posit.co/dev/reference/Metric.html","id":"methods-defined-by-the-base-metric-class-","dir":"Reference","previous_headings":"","what":"Methods defined by the base Metric class:","title":"Subclass the base Metric class — Metric","text":"Calling metric instance self like m(...) equivalent calling:      Initialize self. Args: name: Optional name metric instance. dtype: dtype metric's computations. Defaults NULL, means using config_floatx(). config_floatx() \"float32\" unless set different value (via config_set_floatx()). keras$DTypePolicy provided, compute_dtype utilized.          Return serializable config metric.    Reset metric state variables. function called epochs/steps, metric evaluated training.    Compute current metric value. Returns: scalar tensor, named list scalar tensors.             Accumulate statistics metric.","code":"__call__(...) function(...) {   m$update_state(...)   m$result() } initialize(dtype=NULL, name=NULL) add_variable(shape, initializer, dtype=NULL, aggregation = 'sum', name=NULL) add_weight(shape=shape(), initializer=NULL, dtype=NULL, name=NULL) get_config() reset_state() result() stateless_result(metric_variables) stateless_reset_state() stateless_update_state(metric_variables, ...) update_state(...)"},{"path":"https://keras3.posit.co/dev/reference/Metric.html","id":"readonly-properties","dir":"Reference","previous_headings":"","what":"Readonly properties","title":"Subclass the base Metric class — Metric","text":"dtype variables","code":""},{"path":"https://keras3.posit.co/dev/reference/Metric.html","id":"symbols-in-scope","dir":"Reference","previous_headings":"","what":"Symbols in scope","title":"Subclass the base Metric class — Metric","text":"R function custom methods (public private) following symbols scope: self: custom class instance. super: custom class superclass. private: R environment specific class instance. objects assigned invisible Keras framework. __class__ .symbol(classname): custom class type object.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/Model.html","id":null,"dir":"Reference","previous_headings":"","what":"Subclass the base Keras Model Class — Model","title":"Subclass the base Keras Model Class — Model","text":"advanced use cases need subclass base Model type, e.g., want override train_step() method. just want create define keras model, prefer keras_model() keras_model_sequential(). just want encapsulate custom logic state, need customize training behavior (besides calling self$add_loss() call() method), prefer Layer().","code":""},{"path":"https://keras3.posit.co/dev/reference/Model.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Subclass the base Keras Model Class — Model","text":"","code":"Model(   classname,   initialize = NULL,   call = NULL,   train_step = NULL,   predict_step = NULL,   test_step = NULL,   compute_loss = NULL,   compute_metrics = NULL,   ...,   public = list(),   private = list(),   inherit = NULL,   parent_env = parent.frame() )"},{"path":"https://keras3.posit.co/dev/reference/Model.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Subclass the base Keras Model Class — Model","text":"classname String, name custom class. (Conventionally, CamelCase). initialize, call, train_step, predict_step, test_step, compute_loss, compute_metrics Optional methods can overridden. ..., public Additional methods public members custom class. private Named list R objects (typically, functions) include instance private environments. private methods symbols scope public methods (See section \"Symbols Scope\"). instance private environment. objects private invisible Keras framework Python runtime. inherit custom class subclass. default, base keras class. parent_env R environment class methods grandparent.","code":""},{"path":"https://keras3.posit.co/dev/reference/Model.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Subclass the base Keras Model Class — Model","text":"model constructor function, can call create instance new model type.","code":""},{"path":"https://keras3.posit.co/dev/reference/Model.html","id":"symbols-in-scope","dir":"Reference","previous_headings":"","what":"Symbols in scope","title":"Subclass the base Keras Model Class — Model","text":"R function custom methods (public private) following symbols scope: self: custom class instance. super: custom class superclass. private: R environment specific class instance. objects assigned invisible Keras framework. __class__ .symbol(classname): custom class type object.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/activation_celu.html","id":null,"dir":"Reference","previous_headings":"","what":"Continuously Differentiable Exponential Linear Unit. — activation_celu","title":"Continuously Differentiable Exponential Linear Unit. — activation_celu","text":"CeLU activation function defined : celu(x) = alpha * (exp(x / alpha) - 1) x < 0,celu(x) = x x >= 0. alpha scaling parameter controls activation's shape.","code":""},{"path":"https://keras3.posit.co/dev/reference/activation_celu.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Continuously Differentiable Exponential Linear Unit. — activation_celu","text":"","code":"activation_celu(x, alpha = 1)"},{"path":"https://keras3.posit.co/dev/reference/activation_celu.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Continuously Differentiable Exponential Linear Unit. — activation_celu","text":"x Input tensor. alpha  value CeLU formulation. Defaults 1.0.","code":""},{"path":"https://keras3.posit.co/dev/reference/activation_celu.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Continuously Differentiable Exponential Linear Unit. — activation_celu","text":"tensor, result applying activation input tensor x.","code":""},{"path":"https://keras3.posit.co/dev/reference/activation_celu.html","id":"reference","dir":"Reference","previous_headings":"","what":"Reference","title":"Continuously Differentiable Exponential Linear Unit. — activation_celu","text":"Barron, J. T., 2017","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/activation_elu.html","id":null,"dir":"Reference","previous_headings":"","what":"Exponential Linear Unit. — activation_elu","title":"Exponential Linear Unit. — activation_elu","text":"exponential linear unit (ELU) alpha > 0 defined : x x > 0 alpha * exp(x) - 1 x < 0 ELUs negative values pushes mean activations closer zero. Mean activations closer zero enable faster learning bring gradient closer natural gradient. ELUs saturate negative value argument gets smaller. Saturation means small derivative decreases variation information propagated next layer.","code":""},{"path":"https://keras3.posit.co/dev/reference/activation_elu.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Exponential Linear Unit. — activation_elu","text":"","code":"activation_elu(x, alpha = 1)"},{"path":"https://keras3.posit.co/dev/reference/activation_elu.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Exponential Linear Unit. — activation_elu","text":"x Input tensor. alpha scalar, slope positive section. Defaults 1.0.","code":""},{"path":"https://keras3.posit.co/dev/reference/activation_elu.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Exponential Linear Unit. — activation_elu","text":"tensor, result applying activation input tensor x.","code":""},{"path":"https://keras3.posit.co/dev/reference/activation_elu.html","id":"reference","dir":"Reference","previous_headings":"","what":"Reference","title":"Exponential Linear Unit. — activation_elu","text":"Clevert et al., 2016","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/activation_exponential.html","id":null,"dir":"Reference","previous_headings":"","what":"Exponential activation function. — activation_exponential","title":"Exponential activation function. — activation_exponential","text":"Exponential activation function.","code":""},{"path":"https://keras3.posit.co/dev/reference/activation_exponential.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Exponential activation function. — activation_exponential","text":"","code":"activation_exponential(x)"},{"path":"https://keras3.posit.co/dev/reference/activation_exponential.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Exponential activation function. — activation_exponential","text":"x Input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/activation_exponential.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Exponential activation function. — activation_exponential","text":"tensor, result applying activation input tensor x.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/activation_gelu.html","id":null,"dir":"Reference","previous_headings":"","what":"Gaussian error linear unit (GELU) activation function. — activation_gelu","title":"Gaussian error linear unit (GELU) activation function. — activation_gelu","text":"Gaussian error linear unit (GELU) defined : gelu(x) = x * P(X <= x) P(X) ~ N(0, 1), .e. gelu(x) = 0.5 * x * (1 + erf(x / sqrt(2))). GELU weights inputs value, rather gating inputs sign ReLU.","code":""},{"path":"https://keras3.posit.co/dev/reference/activation_gelu.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Gaussian error linear unit (GELU) activation function. — activation_gelu","text":"","code":"activation_gelu(x, approximate = FALSE)"},{"path":"https://keras3.posit.co/dev/reference/activation_gelu.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Gaussian error linear unit (GELU) activation function. — activation_gelu","text":"x Input tensor. approximate bool, whether enable approximation.","code":""},{"path":"https://keras3.posit.co/dev/reference/activation_gelu.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Gaussian error linear unit (GELU) activation function. — activation_gelu","text":"tensor, result applying activation input tensor x.","code":""},{"path":"https://keras3.posit.co/dev/reference/activation_gelu.html","id":"reference","dir":"Reference","previous_headings":"","what":"Reference","title":"Gaussian error linear unit (GELU) activation function. — activation_gelu","text":"Hendrycks et al., 2016","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/activation_glu.html","id":null,"dir":"Reference","previous_headings":"","what":"Gated Linear Unit (GLU) activation function. — activation_glu","title":"Gated Linear Unit (GLU) activation function. — activation_glu","text":"GLU activation function defined : glu(x) = * sigmoid(b), x split two equal parts b along given axis.","code":""},{"path":"https://keras3.posit.co/dev/reference/activation_glu.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Gated Linear Unit (GLU) activation function. — activation_glu","text":"","code":"activation_glu(x, axis = -1L)"},{"path":"https://keras3.posit.co/dev/reference/activation_glu.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Gated Linear Unit (GLU) activation function. — activation_glu","text":"x Input tensor. axis axis along split input tensor. Defaults -1.","code":""},{"path":"https://keras3.posit.co/dev/reference/activation_glu.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Gated Linear Unit (GLU) activation function. — activation_glu","text":"tensor, result applying activation input tensor x.","code":""},{"path":"https://keras3.posit.co/dev/reference/activation_glu.html","id":"reference","dir":"Reference","previous_headings":"","what":"Reference","title":"Gated Linear Unit (GLU) activation function. — activation_glu","text":"Dauphin et al., 2017","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/activation_hard_shrink.html","id":null,"dir":"Reference","previous_headings":"","what":"Hard Shrink activation function. — activation_hard_shrink","title":"Hard Shrink activation function. — activation_hard_shrink","text":"defined : hard_shrink(x) = x |x| > threshold, hard_shrink(x) = 0 otherwise.","code":""},{"path":"https://keras3.posit.co/dev/reference/activation_hard_shrink.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Hard Shrink activation function. — activation_hard_shrink","text":"","code":"activation_hard_shrink(x, threshold = 0.5)"},{"path":"https://keras3.posit.co/dev/reference/activation_hard_shrink.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Hard Shrink activation function. — activation_hard_shrink","text":"x Input tensor. threshold Threshold value. Defaults 0.5.","code":""},{"path":"https://keras3.posit.co/dev/reference/activation_hard_shrink.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Hard Shrink activation function. — activation_hard_shrink","text":"tensor, result applying activation input tensor x.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/activation_hard_sigmoid.html","id":null,"dir":"Reference","previous_headings":"","what":"Hard sigmoid activation function. — activation_hard_sigmoid","title":"Hard sigmoid activation function. — activation_hard_sigmoid","text":"hard sigmoid activation defined : 0 x <= -3 1 x >= 3 (x/6) + 0.5 -3 < x < 3 faster, piecewise linear approximation sigmoid activation.","code":""},{"path":"https://keras3.posit.co/dev/reference/activation_hard_sigmoid.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Hard sigmoid activation function. — activation_hard_sigmoid","text":"","code":"activation_hard_sigmoid(x)"},{"path":"https://keras3.posit.co/dev/reference/activation_hard_sigmoid.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Hard sigmoid activation function. — activation_hard_sigmoid","text":"x Input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/activation_hard_sigmoid.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Hard sigmoid activation function. — activation_hard_sigmoid","text":"tensor, result applying activation input tensor x.","code":""},{"path":"https://keras3.posit.co/dev/reference/activation_hard_sigmoid.html","id":"reference","dir":"Reference","previous_headings":"","what":"Reference","title":"Hard sigmoid activation function. — activation_hard_sigmoid","text":"Wikipedia \"Hard sigmoid\"","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/activation_hard_silu.html","id":null,"dir":"Reference","previous_headings":"","what":"Hard SiLU activation function, also known as Hard Swish. — activation_hard_silu","title":"Hard SiLU activation function, also known as Hard Swish. — activation_hard_silu","text":"defined : 0 x < -3 x x > 3 x * (x + 3) / 6 -3 <= x <= 3 faster, piecewise linear approximation silu activation.","code":""},{"path":"https://keras3.posit.co/dev/reference/activation_hard_silu.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Hard SiLU activation function, also known as Hard Swish. — activation_hard_silu","text":"","code":"activation_hard_silu(x)  activation_hard_swish(x)"},{"path":"https://keras3.posit.co/dev/reference/activation_hard_silu.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Hard SiLU activation function, also known as Hard Swish. — activation_hard_silu","text":"x Input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/activation_hard_silu.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Hard SiLU activation function, also known as Hard Swish. — activation_hard_silu","text":"tensor, result applying activation input tensor x.","code":""},{"path":"https://keras3.posit.co/dev/reference/activation_hard_silu.html","id":"reference","dir":"Reference","previous_headings":"","what":"Reference","title":"Hard SiLU activation function, also known as Hard Swish. — activation_hard_silu","text":"Howard, 2019","code":""},{"path":"https://keras3.posit.co/dev/reference/activation_hard_tanh.html","id":null,"dir":"Reference","previous_headings":"","what":"HardTanh activation function. — activation_hard_tanh","title":"HardTanh activation function. — activation_hard_tanh","text":"defined : hard_tanh(x) = -1 x < -1, hard_tanh(x) = x -1 <= x <= 1, hard_tanh(x) = 1 x > 1.","code":""},{"path":"https://keras3.posit.co/dev/reference/activation_hard_tanh.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"HardTanh activation function. — activation_hard_tanh","text":"","code":"activation_hard_tanh(x)"},{"path":"https://keras3.posit.co/dev/reference/activation_hard_tanh.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"HardTanh activation function. — activation_hard_tanh","text":"x Input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/activation_hard_tanh.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"HardTanh activation function. — activation_hard_tanh","text":"tensor, result applying activation input tensor x.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/activation_leaky_relu.html","id":null,"dir":"Reference","previous_headings":"","what":"Leaky relu activation function. — activation_leaky_relu","title":"Leaky relu activation function. — activation_leaky_relu","text":"Leaky relu activation function.","code":""},{"path":"https://keras3.posit.co/dev/reference/activation_leaky_relu.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Leaky relu activation function. — activation_leaky_relu","text":"","code":"activation_leaky_relu(x, negative_slope = 0.2)"},{"path":"https://keras3.posit.co/dev/reference/activation_leaky_relu.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Leaky relu activation function. — activation_leaky_relu","text":"x Input tensor. negative_slope float controls slope values lower threshold.","code":""},{"path":"https://keras3.posit.co/dev/reference/activation_leaky_relu.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Leaky relu activation function. — activation_leaky_relu","text":"tensor, result applying activation input tensor x.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/activation_linear.html","id":null,"dir":"Reference","previous_headings":"","what":"Linear activation function (pass-through). — activation_linear","title":"Linear activation function (pass-through). — activation_linear","text":"\"linear\" activation identity function: returns input, unmodified.","code":""},{"path":"https://keras3.posit.co/dev/reference/activation_linear.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Linear activation function (pass-through). — activation_linear","text":"","code":"activation_linear(x)"},{"path":"https://keras3.posit.co/dev/reference/activation_linear.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Linear activation function (pass-through). — activation_linear","text":"x Input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/activation_linear.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Linear activation function (pass-through). — activation_linear","text":"tensor, result applying activation input tensor x.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/activation_log_sigmoid.html","id":null,"dir":"Reference","previous_headings":"","what":"Logarithm of the sigmoid activation function. — activation_log_sigmoid","title":"Logarithm of the sigmoid activation function. — activation_log_sigmoid","text":"defined f(x) = log(1 / (1 + exp(-x))).","code":""},{"path":"https://keras3.posit.co/dev/reference/activation_log_sigmoid.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Logarithm of the sigmoid activation function. — activation_log_sigmoid","text":"","code":"activation_log_sigmoid(x)"},{"path":"https://keras3.posit.co/dev/reference/activation_log_sigmoid.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Logarithm of the sigmoid activation function. — activation_log_sigmoid","text":"x Input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/activation_log_sigmoid.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Logarithm of the sigmoid activation function. — activation_log_sigmoid","text":"tensor, result applying activation input tensor x.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/activation_log_softmax.html","id":null,"dir":"Reference","previous_headings":"","what":"Log-Softmax activation function. — activation_log_softmax","title":"Log-Softmax activation function. — activation_log_softmax","text":"input vector handled independently. axis argument sets axis input function applied along.","code":""},{"path":"https://keras3.posit.co/dev/reference/activation_log_softmax.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Log-Softmax activation function. — activation_log_softmax","text":"","code":"activation_log_softmax(x, axis = -1L)"},{"path":"https://keras3.posit.co/dev/reference/activation_log_softmax.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Log-Softmax activation function. — activation_log_softmax","text":"x Input tensor. axis Integer, axis along softmax applied.","code":""},{"path":"https://keras3.posit.co/dev/reference/activation_log_softmax.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Log-Softmax activation function. — activation_log_softmax","text":"tensor, result applying activation input tensor x.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/activation_mish.html","id":null,"dir":"Reference","previous_headings":"","what":"Mish activation function. — activation_mish","title":"Mish activation function. — activation_mish","text":"defined : mish(x) = x * tanh(softplus(x)) softplus defined : softplus(x) = log(exp(x) + 1)","code":""},{"path":"https://keras3.posit.co/dev/reference/activation_mish.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Mish activation function. — activation_mish","text":"","code":"activation_mish(x)"},{"path":"https://keras3.posit.co/dev/reference/activation_mish.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Mish activation function. — activation_mish","text":"x Input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/activation_mish.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Mish activation function. — activation_mish","text":"tensor, result applying activation input tensor x.","code":""},{"path":"https://keras3.posit.co/dev/reference/activation_mish.html","id":"reference","dir":"Reference","previous_headings":"","what":"Reference","title":"Mish activation function. — activation_mish","text":"Misra, 2019","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/activation_relu.html","id":null,"dir":"Reference","previous_headings":"","what":"Applies the rectified linear unit activation function. — activation_relu","title":"Applies the rectified linear unit activation function. — activation_relu","text":"default values, returns standard ReLU activation: max(x, 0), element-wise maximum 0 input tensor. Modifying default parameters allows use non-zero thresholds, change max value activation, use non-zero multiple input values threshold.","code":""},{"path":"https://keras3.posit.co/dev/reference/activation_relu.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Applies the rectified linear unit activation function. — activation_relu","text":"","code":"activation_relu(x, negative_slope = 0, max_value = NULL, threshold = 0)"},{"path":"https://keras3.posit.co/dev/reference/activation_relu.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Applies the rectified linear unit activation function. — activation_relu","text":"x Input tensor. negative_slope numeric controls slope values lower threshold. max_value numeric sets saturation threshold (largest value function return). threshold numeric giving threshold value activation function values damped set zero.","code":""},{"path":"https://keras3.posit.co/dev/reference/activation_relu.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Applies the rectified linear unit activation function. — activation_relu","text":"tensor shape dtype input x.","code":""},{"path":"https://keras3.posit.co/dev/reference/activation_relu.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Applies the rectified linear unit activation function. — activation_relu","text":"","code":"x <- c(-10, -5, 0, 5, 10) activation_relu(x) ## tf.Tensor([ 0.  0.  0.  5. 10.], shape=(5), dtype=float32) activation_relu(x, negative_slope = 0.5) ## tf.Tensor([-5.  -2.5  0.   5.  10. ], shape=(5), dtype=float32) activation_relu(x, max_value = 5) ## tf.Tensor([0. 0. 0. 5. 5.], shape=(5), dtype=float32) activation_relu(x, threshold = 5) ## tf.Tensor([-0. -0.  0.  0. 10.], shape=(5), dtype=float32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/activation_relu6.html","id":null,"dir":"Reference","previous_headings":"","what":"Relu6 activation function. — activation_relu6","title":"Relu6 activation function. — activation_relu6","text":"ReLU function, truncated maximum value 6.","code":""},{"path":"https://keras3.posit.co/dev/reference/activation_relu6.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Relu6 activation function. — activation_relu6","text":"","code":"activation_relu6(x)"},{"path":"https://keras3.posit.co/dev/reference/activation_relu6.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Relu6 activation function. — activation_relu6","text":"x Input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/activation_relu6.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Relu6 activation function. — activation_relu6","text":"tensor, result applying activation input tensor x.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/activation_selu.html","id":null,"dir":"Reference","previous_headings":"","what":"Scaled Exponential Linear Unit (SELU). — activation_selu","title":"Scaled Exponential Linear Unit (SELU). — activation_selu","text":"Scaled Exponential Linear Unit (SELU) activation function defined : scale * x x > 0 scale * alpha * (exp(x) - 1) x < 0 alpha scale pre-defined constants (alpha = 1.67326324 scale = 1.05070098). Basically, SELU activation function multiplies scale (> 1) output activation_elu function ensure slope larger one positive inputs. values alpha scale chosen mean variance inputs preserved two consecutive layers long weights initialized correctly (see initializer_lecun_normal()) number input units \"large enough\" (see reference paper information).","code":""},{"path":"https://keras3.posit.co/dev/reference/activation_selu.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Scaled Exponential Linear Unit (SELU). — activation_selu","text":"","code":"activation_selu(x)"},{"path":"https://keras3.posit.co/dev/reference/activation_selu.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Scaled Exponential Linear Unit (SELU). — activation_selu","text":"x Input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/activation_selu.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Scaled Exponential Linear Unit (SELU). — activation_selu","text":"tensor, result applying activation input tensor x.","code":""},{"path":"https://keras3.posit.co/dev/reference/activation_selu.html","id":"notes","dir":"Reference","previous_headings":"","what":"Notes","title":"Scaled Exponential Linear Unit (SELU). — activation_selu","text":"used together initializer_lecun_normal(). used together dropout variant layer_alpha_dropout() (legacy, depracated).","code":""},{"path":"https://keras3.posit.co/dev/reference/activation_selu.html","id":"reference","dir":"Reference","previous_headings":"","what":"Reference","title":"Scaled Exponential Linear Unit (SELU). — activation_selu","text":"Klambauer et al., 2017","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/activation_sigmoid.html","id":null,"dir":"Reference","previous_headings":"","what":"Sigmoid activation function. — activation_sigmoid","title":"Sigmoid activation function. — activation_sigmoid","text":"defined : sigmoid(x) = 1 / (1 + exp(-x)). small values (<-5), sigmoid returns value close zero, large values (>5) result function gets close 1. Sigmoid equivalent 2-element softmax, second element assumed zero. sigmoid function always returns value 0 1.","code":""},{"path":"https://keras3.posit.co/dev/reference/activation_sigmoid.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Sigmoid activation function. — activation_sigmoid","text":"","code":"activation_sigmoid(x)"},{"path":"https://keras3.posit.co/dev/reference/activation_sigmoid.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Sigmoid activation function. — activation_sigmoid","text":"x Input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/activation_sigmoid.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Sigmoid activation function. — activation_sigmoid","text":"tensor, result applying activation input tensor x.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/activation_silu.html","id":null,"dir":"Reference","previous_headings":"","what":"Swish (or Silu) activation function. — activation_silu","title":"Swish (or Silu) activation function. — activation_silu","text":"defined : swish(x) = x * sigmoid(x). Swish (Silu) activation function smooth, non-monotonic function unbounded bounded .","code":""},{"path":"https://keras3.posit.co/dev/reference/activation_silu.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Swish (or Silu) activation function. — activation_silu","text":"","code":"activation_silu(x)"},{"path":"https://keras3.posit.co/dev/reference/activation_silu.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Swish (or Silu) activation function. — activation_silu","text":"x Input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/activation_silu.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Swish (or Silu) activation function. — activation_silu","text":"tensor, result applying activation input tensor x.","code":""},{"path":"https://keras3.posit.co/dev/reference/activation_silu.html","id":"reference","dir":"Reference","previous_headings":"","what":"Reference","title":"Swish (or Silu) activation function. — activation_silu","text":"Ramachandran et al., 2017","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/activation_soft_shrink.html","id":null,"dir":"Reference","previous_headings":"","what":"Soft Shrink activation function. — activation_soft_shrink","title":"Soft Shrink activation function. — activation_soft_shrink","text":"defined : soft_shrink(x) = x - threshold x > threshold, soft_shrink(x) = x + threshold x < -threshold, soft_shrink(x) = 0 otherwise.","code":""},{"path":"https://keras3.posit.co/dev/reference/activation_soft_shrink.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Soft Shrink activation function. — activation_soft_shrink","text":"","code":"activation_soft_shrink(x, threshold = 0.5)"},{"path":"https://keras3.posit.co/dev/reference/activation_soft_shrink.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Soft Shrink activation function. — activation_soft_shrink","text":"x Input tensor. threshold Threshold value. Defaults 0.5.","code":""},{"path":"https://keras3.posit.co/dev/reference/activation_soft_shrink.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Soft Shrink activation function. — activation_soft_shrink","text":"tensor, result applying activation input tensor x.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/activation_softmax.html","id":null,"dir":"Reference","previous_headings":"","what":"Softmax converts a vector of values to a probability distribution. — activation_softmax","title":"Softmax converts a vector of values to a probability distribution. — activation_softmax","text":"elements output vector range [0, 1] sum 1. input vector handled independently. axis argument sets axis input function applied along. Softmax often used activation last layer classification network result interpreted probability distribution. softmax vector x computed exp(x) / sum(exp(x)). input values log-odds resulting probability.","code":""},{"path":"https://keras3.posit.co/dev/reference/activation_softmax.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Softmax converts a vector of values to a probability distribution. — activation_softmax","text":"","code":"activation_softmax(x, axis = -1L)"},{"path":"https://keras3.posit.co/dev/reference/activation_softmax.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Softmax converts a vector of values to a probability distribution. — activation_softmax","text":"x Input tensor. axis Integer, axis along softmax applied.","code":""},{"path":"https://keras3.posit.co/dev/reference/activation_softmax.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Softmax converts a vector of values to a probability distribution. — activation_softmax","text":"tensor, result applying activation input tensor x.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/activation_softplus.html","id":null,"dir":"Reference","previous_headings":"","what":"Softplus activation function. — activation_softplus","title":"Softplus activation function. — activation_softplus","text":"defined : softplus(x) = log(exp(x) + 1).","code":""},{"path":"https://keras3.posit.co/dev/reference/activation_softplus.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Softplus activation function. — activation_softplus","text":"","code":"activation_softplus(x)"},{"path":"https://keras3.posit.co/dev/reference/activation_softplus.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Softplus activation function. — activation_softplus","text":"x Input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/activation_softplus.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Softplus activation function. — activation_softplus","text":"tensor, result applying activation input tensor x.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/activation_softsign.html","id":null,"dir":"Reference","previous_headings":"","what":"Softsign activation function. — activation_softsign","title":"Softsign activation function. — activation_softsign","text":"Softsign defined : softsign(x) = x / (abs(x) + 1).","code":""},{"path":"https://keras3.posit.co/dev/reference/activation_softsign.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Softsign activation function. — activation_softsign","text":"","code":"activation_softsign(x)"},{"path":"https://keras3.posit.co/dev/reference/activation_softsign.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Softsign activation function. — activation_softsign","text":"x Input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/activation_softsign.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Softsign activation function. — activation_softsign","text":"tensor, result applying activation input tensor x.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/activation_sparse_plus.html","id":null,"dir":"Reference","previous_headings":"","what":"SparsePlus activation function. — activation_sparse_plus","title":"SparsePlus activation function. — activation_sparse_plus","text":"SparsePlus defined : sparse_plus(x) = 0 x <= -1. sparse_plus(x) = (1/4) * (x + 1)^2 -1 < x < 1. sparse_plus(x) = x x >= 1.","code":""},{"path":"https://keras3.posit.co/dev/reference/activation_sparse_plus.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"SparsePlus activation function. — activation_sparse_plus","text":"","code":"activation_sparse_plus(x)"},{"path":"https://keras3.posit.co/dev/reference/activation_sparse_plus.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"SparsePlus activation function. — activation_sparse_plus","text":"x Input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/activation_sparse_plus.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"SparsePlus activation function. — activation_sparse_plus","text":"tensor, result applying activation input tensor x.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/activation_sparse_sigmoid.html","id":null,"dir":"Reference","previous_headings":"","what":"Sparse sigmoid activation function. — activation_sparse_sigmoid","title":"Sparse sigmoid activation function. — activation_sparse_sigmoid","text":"defined f(x) = 0 x <= -1, f(x) = 0.5 * (x + 1) -1 < x < 1, f(x) = 1 x >= 1.","code":""},{"path":"https://keras3.posit.co/dev/reference/activation_sparse_sigmoid.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Sparse sigmoid activation function. — activation_sparse_sigmoid","text":"","code":"activation_sparse_sigmoid(x)"},{"path":"https://keras3.posit.co/dev/reference/activation_sparse_sigmoid.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Sparse sigmoid activation function. — activation_sparse_sigmoid","text":"x Input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/activation_sparse_sigmoid.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Sparse sigmoid activation function. — activation_sparse_sigmoid","text":"tensor, result applying activation input tensor x.","code":""},{"path":"https://keras3.posit.co/dev/reference/activation_sparse_sigmoid.html","id":"reference","dir":"Reference","previous_headings":"","what":"Reference","title":"Sparse sigmoid activation function. — activation_sparse_sigmoid","text":"M. Blondel, . F. T. Martins, V. Niculae, 2019","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/activation_sparsemax.html","id":null,"dir":"Reference","previous_headings":"","what":"Sparsemax activation function. — activation_sparsemax","title":"Sparsemax activation function. — activation_sparsemax","text":"batch , class j, sparsemax activation function defined : sparsemax(x)[, j] = max(x[, j] - (x[, :]), 0).","code":""},{"path":"https://keras3.posit.co/dev/reference/activation_sparsemax.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Sparsemax activation function. — activation_sparsemax","text":"","code":"activation_sparsemax(x, axis = -1L)"},{"path":"https://keras3.posit.co/dev/reference/activation_sparsemax.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Sparsemax activation function. — activation_sparsemax","text":"x Input tensor. axis int, axis along sparsemax operation applied. (1-based)","code":""},{"path":"https://keras3.posit.co/dev/reference/activation_sparsemax.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Sparsemax activation function. — activation_sparsemax","text":"tensor, output sparsemax transformation. type shape x.","code":""},{"path":"https://keras3.posit.co/dev/reference/activation_sparsemax.html","id":"reference","dir":"Reference","previous_headings":"","what":"Reference","title":"Sparsemax activation function. — activation_sparsemax","text":"Martins et.al., 2016","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/activation_squareplus.html","id":null,"dir":"Reference","previous_headings":"","what":"Squareplus activation function. — activation_squareplus","title":"Squareplus activation function. — activation_squareplus","text":"Squareplus activation function defined : f(x) = (x + sqrt(x^2 + b)) / 2 b smoothness parameter.","code":""},{"path":"https://keras3.posit.co/dev/reference/activation_squareplus.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Squareplus activation function. — activation_squareplus","text":"","code":"activation_squareplus(x, b = 4L)"},{"path":"https://keras3.posit.co/dev/reference/activation_squareplus.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Squareplus activation function. — activation_squareplus","text":"x Input tensor. b Smoothness parameter. Defaults 4.","code":""},{"path":"https://keras3.posit.co/dev/reference/activation_squareplus.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Squareplus activation function. — activation_squareplus","text":"tensor, result applying activation input tensor x.","code":""},{"path":"https://keras3.posit.co/dev/reference/activation_squareplus.html","id":"reference","dir":"Reference","previous_headings":"","what":"Reference","title":"Squareplus activation function. — activation_squareplus","text":"Ramachandran et al., 2021","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/activation_tanh.html","id":null,"dir":"Reference","previous_headings":"","what":"Hyperbolic tangent activation function. — activation_tanh","title":"Hyperbolic tangent activation function. — activation_tanh","text":"defined : tanh(x) = sinh(x) / cosh(x), .e. tanh(x) = ((exp(x) - exp(-x)) / (exp(x) + exp(-x))).","code":""},{"path":"https://keras3.posit.co/dev/reference/activation_tanh.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Hyperbolic tangent activation function. — activation_tanh","text":"","code":"activation_tanh(x)"},{"path":"https://keras3.posit.co/dev/reference/activation_tanh.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Hyperbolic tangent activation function. — activation_tanh","text":"x Input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/activation_tanh.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Hyperbolic tangent activation function. — activation_tanh","text":"tensor, result applying activation input tensor x.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/activation_tanh_shrink.html","id":null,"dir":"Reference","previous_headings":"","what":"Tanh shrink activation function. — activation_tanh_shrink","title":"Tanh shrink activation function. — activation_tanh_shrink","text":"defined : f(x) = x - tanh(x).","code":""},{"path":"https://keras3.posit.co/dev/reference/activation_tanh_shrink.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Tanh shrink activation function. — activation_tanh_shrink","text":"","code":"activation_tanh_shrink(x)"},{"path":"https://keras3.posit.co/dev/reference/activation_tanh_shrink.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Tanh shrink activation function. — activation_tanh_shrink","text":"x Input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/activation_tanh_shrink.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Tanh shrink activation function. — activation_tanh_shrink","text":"tensor, result applying activation input tensor x.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/activation_threshold.html","id":null,"dir":"Reference","previous_headings":"","what":"Threshold activation function. — activation_threshold","title":"Threshold activation function. — activation_threshold","text":"defined : threshold(x) = x x > threshold, threshold(x) = default_value otherwise.","code":""},{"path":"https://keras3.posit.co/dev/reference/activation_threshold.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Threshold activation function. — activation_threshold","text":"","code":"activation_threshold(x, threshold, default_value)"},{"path":"https://keras3.posit.co/dev/reference/activation_threshold.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Threshold activation function. — activation_threshold","text":"x Input tensor. threshold value decides retain replace x. default_value Value assign x <= threshold.","code":""},{"path":"https://keras3.posit.co/dev/reference/activation_threshold.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Threshold activation function. — activation_threshold","text":"tensor, result applying activation input tensor x.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/active_property.html","id":null,"dir":"Reference","previous_headings":"","what":"Create an active property class method — active_property","title":"Create an active property class method — active_property","text":"Create active property class method","code":""},{"path":"https://keras3.posit.co/dev/reference/active_property.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Create an active property class method — active_property","text":"","code":"active_property(fn)"},{"path":"https://keras3.posit.co/dev/reference/active_property.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Create an active property class method — active_property","text":"fn R function","code":""},{"path":"https://keras3.posit.co/dev/reference/active_property.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Create an active property class method — active_property","text":"fn, additional R attribute cause fn converted active property converted method custom subclass.","code":""},{"path":"https://keras3.posit.co/dev/reference/active_property.html","id":"example","dir":"Reference","previous_headings":"","what":"Example","title":"Create an active property class method — active_property","text":"","code":"layer_foo <- Model(\"Foo\", ...,   metrics = active_property(function() {     list(self$d_loss_metric,          self$g_loss_metric)   }))"},{"path":"https://keras3.posit.co/dev/reference/adapt.html","id":null,"dir":"Reference","previous_headings":"","what":"Fits the state of the preprocessing layer to the data being passed — adapt","title":"Fits the state of the preprocessing layer to the data being passed — adapt","text":"Fits state preprocessing layer data passed","code":""},{"path":"https://keras3.posit.co/dev/reference/adapt.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Fits the state of the preprocessing layer to the data being passed — adapt","text":"","code":"adapt(object, data, ..., batch_size = NULL, steps = NULL)"},{"path":"https://keras3.posit.co/dev/reference/adapt.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Fits the state of the preprocessing layer to the data being passed — adapt","text":"object Preprocessing layer object data data train . can passed either tf.data.Dataset R array. ... Used forwards backwards compatibility. Passed underlying method. batch_size Integer NULL. Number asamples per state update. unspecified, batch_size default 32. specify batch_size data form TF Dataset generator (since generate batches). steps Integer NULL. Total number steps (batches samples) training input tensors TensorFlow data tensors, default NULL equal number samples dataset divided batch size, 1 determined. x tf.data.Dataset, steps NULL, epoch run input dataset exhausted. passing infinitely repeating dataset, must specify steps argument. argument supported array inputs.","code":""},{"path":"https://keras3.posit.co/dev/reference/adapt.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Fits the state of the preprocessing layer to the data being passed — adapt","text":"Returns object, invisibly.","code":""},{"path":"https://keras3.posit.co/dev/reference/adapt.html","id":"details","dir":"Reference","previous_headings":"","what":"Details","title":"Fits the state of the preprocessing layer to the data being passed — adapt","text":"calling adapt layer, preprocessing layer's state update training. order make preprocessing layers efficient distribution context, kept constant respect compiled tf.Graphs call layer. affect layer use adapting layer , adapt layer multiple times need take care re-compile compiled functions follows: adding preprocessing layer keras model, need call compile(model) subsequent call adapt(). calling preprocessing layer inside tfdatasets::dataset_map(), call dataset_map() input Dataset adapt(). using tensorflow::tf_function() directly calls preprocessing layer, need call tf_function() callable subsequent call adapt(). keras_model() example multiple adapts:         tfdatasets example multiple adapts:","code":"layer <- layer_normalization(axis = NULL) adapt(layer, c(0, 2)) model <- keras_model_sequential() |> layer() predict(model, c(0, 1, 2), verbose = FALSE) # [1] -1  0  1 ## [1] -1  0  1 adapt(layer, c(-1, 1)) compile(model)  # This is needed to re-compile model.predict! predict(model, c(0, 1, 2), verbose = FALSE) # [1] 0 1 2 ## [1] 0 1 2 layer <- layer_normalization(axis = NULL) adapt(layer, c(0, 2)) input_ds <- tfdatasets::range_dataset(0, 3) normalized_ds <- input_ds |>   tfdatasets::dataset_map(layer) str(tfdatasets::iterate(normalized_ds)) ## List of 3 ##  $ :<tf.Tensor: shape=(1), dtype=float32, numpy=array([-1.], dtype=float32)> ##  $ :<tf.Tensor: shape=(1), dtype=float32, numpy=array([0.], dtype=float32)> ##  $ :<tf.Tensor: shape=(1), dtype=float32, numpy=array([1.], dtype=float32)> adapt(layer, c(-1, 1)) normalized_ds <- input_ds |>   tfdatasets::dataset_map(layer) # Re-map over the input dataset.  normalized_ds |>   tfdatasets::as_array_iterator() |>   tfdatasets::iterate(simplify = FALSE) |>   str() ## List of 3 ##  $ : num [1(1d)] 0 ##  $ : num [1(1d)] 1 ##  $ : num [1(1d)] 2"},{"path":"https://keras3.posit.co/dev/reference/application_convnext_base.html","id":null,"dir":"Reference","previous_headings":"","what":"Instantiates the ConvNeXtBase architecture. — application_convnext_base","title":"Instantiates the ConvNeXtBase architecture. — application_convnext_base","text":"Instantiates ConvNeXtBase architecture.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_convnext_base.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Instantiates the ConvNeXtBase architecture. — application_convnext_base","text":"","code":"application_convnext_base(   include_top = TRUE,   include_preprocessing = TRUE,   weights = \"imagenet\",   input_tensor = NULL,   input_shape = NULL,   pooling = NULL,   classes = 1000L,   classifier_activation = \"softmax\",   name = \"convnext_base\" )"},{"path":"https://keras3.posit.co/dev/reference/application_convnext_base.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Instantiates the ConvNeXtBase architecture. — application_convnext_base","text":"include_top Whether include fully-connected layer top network. Defaults TRUE. include_preprocessing Boolean, whether include preprocessing layer bottom network. weights One NULL (random initialization), \"imagenet\" (pre-training ImageNet-1k), path weights file loaded. Defaults \"imagenet\". input_tensor Optional Keras tensor (.e. output keras_input()) use image input model. input_shape Optional shape tuple, specified include_top FALSE. exactly 3 inputs channels. pooling Optional pooling mode feature extraction include_top FALSE. Defaults NULL. NULL means output model 4D tensor output last convolutional layer. avg means global average pooling applied output last convolutional layer, thus output model 2D tensor. max means global max pooling applied. classes Optional number classes classify images , specified include_top TRUE, weights argument specified. Defaults 1000 (number ImageNet classes). classifier_activation str callable. activation function use \"top\" layer. Ignored unless include_top=TRUE. Set classifier_activation=NULL return logits \"top\" layer. Defaults \"softmax\". loading pretrained weights, classifier_activation can NULL \"softmax\". name name model (string).","code":""},{"path":"https://keras3.posit.co/dev/reference/application_convnext_base.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Instantiates the ConvNeXtBase architecture. — application_convnext_base","text":"model instance.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_convnext_base.html","id":"references","dir":"Reference","previous_headings":"","what":"References","title":"Instantiates the ConvNeXtBase architecture. — application_convnext_base","text":"ConvNet 2020s (CVPR 2022) image classification use cases, see page detailed examples. transfer learning use cases, make sure read guide transfer learning & fine-tuning. base, large, xlarge models first pre-trained ImageNet-21k dataset fine-tuned ImageNet-1k dataset. pre-trained parameters models assembled official repository. get sense parameters converted Keras compatible parameters, please refer repository.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_convnext_base.html","id":"note","dir":"Reference","previous_headings":"","what":"Note","title":"Instantiates the ConvNeXtBase architecture. — application_convnext_base","text":"Keras Application expects specific kind input preprocessing. ConvNeXt, preprocessing included model using Normalization layer.  ConvNeXt models expect inputs float uint8 tensors pixels values [0-255] range. calling summary() method instantiating ConvNeXt model, prefer setting expand_nested argument summary() TRUE better investigate instantiated model.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/application_convnext_large.html","id":null,"dir":"Reference","previous_headings":"","what":"Instantiates the ConvNeXtLarge architecture. — application_convnext_large","title":"Instantiates the ConvNeXtLarge architecture. — application_convnext_large","text":"Instantiates ConvNeXtLarge architecture.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_convnext_large.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Instantiates the ConvNeXtLarge architecture. — application_convnext_large","text":"","code":"application_convnext_large(   include_top = TRUE,   include_preprocessing = TRUE,   weights = \"imagenet\",   input_tensor = NULL,   input_shape = NULL,   pooling = NULL,   classes = 1000L,   classifier_activation = \"softmax\",   name = \"convnext_large\" )"},{"path":"https://keras3.posit.co/dev/reference/application_convnext_large.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Instantiates the ConvNeXtLarge architecture. — application_convnext_large","text":"include_top Whether include fully-connected layer top network. Defaults TRUE. include_preprocessing Boolean, whether include preprocessing layer bottom network. weights One NULL (random initialization), \"imagenet\" (pre-training ImageNet-1k), path weights file loaded. Defaults \"imagenet\". input_tensor Optional Keras tensor (.e. output keras_input()) use image input model. input_shape Optional shape tuple, specified include_top FALSE. exactly 3 inputs channels. pooling Optional pooling mode feature extraction include_top FALSE. Defaults NULL. NULL means output model 4D tensor output last convolutional layer. avg means global average pooling applied output last convolutional layer, thus output model 2D tensor. max means global max pooling applied. classes Optional number classes classify images , specified include_top TRUE, weights argument specified. Defaults 1000 (number ImageNet classes). classifier_activation str callable. activation function use \"top\" layer. Ignored unless include_top=TRUE. Set classifier_activation=NULL return logits \"top\" layer. Defaults \"softmax\". loading pretrained weights, classifier_activation can NULL \"softmax\". name name model (string).","code":""},{"path":"https://keras3.posit.co/dev/reference/application_convnext_large.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Instantiates the ConvNeXtLarge architecture. — application_convnext_large","text":"model instance.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_convnext_large.html","id":"references","dir":"Reference","previous_headings":"","what":"References","title":"Instantiates the ConvNeXtLarge architecture. — application_convnext_large","text":"ConvNet 2020s (CVPR 2022) image classification use cases, see page detailed examples. transfer learning use cases, make sure read guide transfer learning & fine-tuning. base, large, xlarge models first pre-trained ImageNet-21k dataset fine-tuned ImageNet-1k dataset. pre-trained parameters models assembled official repository. get sense parameters converted Keras compatible parameters, please refer repository.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_convnext_large.html","id":"note","dir":"Reference","previous_headings":"","what":"Note","title":"Instantiates the ConvNeXtLarge architecture. — application_convnext_large","text":"Keras Application expects specific kind input preprocessing. ConvNeXt, preprocessing included model using Normalization layer.  ConvNeXt models expect inputs float uint8 tensors pixels values [0-255] range. calling summary() method instantiating ConvNeXt model, prefer setting expand_nested argument summary() TRUE better investigate instantiated model.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/application_convnext_small.html","id":null,"dir":"Reference","previous_headings":"","what":"Instantiates the ConvNeXtSmall architecture. — application_convnext_small","title":"Instantiates the ConvNeXtSmall architecture. — application_convnext_small","text":"Instantiates ConvNeXtSmall architecture.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_convnext_small.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Instantiates the ConvNeXtSmall architecture. — application_convnext_small","text":"","code":"application_convnext_small(   include_top = TRUE,   include_preprocessing = TRUE,   weights = \"imagenet\",   input_tensor = NULL,   input_shape = NULL,   pooling = NULL,   classes = 1000L,   classifier_activation = \"softmax\",   name = \"convnext_small\" )"},{"path":"https://keras3.posit.co/dev/reference/application_convnext_small.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Instantiates the ConvNeXtSmall architecture. — application_convnext_small","text":"include_top Whether include fully-connected layer top network. Defaults TRUE. include_preprocessing Boolean, whether include preprocessing layer bottom network. weights One NULL (random initialization), \"imagenet\" (pre-training ImageNet-1k), path weights file loaded. Defaults \"imagenet\". input_tensor Optional Keras tensor (.e. output keras_input()) use image input model. input_shape Optional shape tuple, specified include_top FALSE. exactly 3 inputs channels. pooling Optional pooling mode feature extraction include_top FALSE. Defaults NULL. NULL means output model 4D tensor output last convolutional layer. avg means global average pooling applied output last convolutional layer, thus output model 2D tensor. max means global max pooling applied. classes Optional number classes classify images , specified include_top TRUE, weights argument specified. Defaults 1000 (number ImageNet classes). classifier_activation str callable. activation function use \"top\" layer. Ignored unless include_top=TRUE. Set classifier_activation=NULL return logits \"top\" layer. Defaults \"softmax\". loading pretrained weights, classifier_activation can NULL \"softmax\". name name model (string).","code":""},{"path":"https://keras3.posit.co/dev/reference/application_convnext_small.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Instantiates the ConvNeXtSmall architecture. — application_convnext_small","text":"model instance.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_convnext_small.html","id":"references","dir":"Reference","previous_headings":"","what":"References","title":"Instantiates the ConvNeXtSmall architecture. — application_convnext_small","text":"ConvNet 2020s (CVPR 2022) image classification use cases, see page detailed examples. transfer learning use cases, make sure read guide transfer learning & fine-tuning. base, large, xlarge models first pre-trained ImageNet-21k dataset fine-tuned ImageNet-1k dataset. pre-trained parameters models assembled official repository. get sense parameters converted Keras compatible parameters, please refer repository.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_convnext_small.html","id":"note","dir":"Reference","previous_headings":"","what":"Note","title":"Instantiates the ConvNeXtSmall architecture. — application_convnext_small","text":"Keras Application expects specific kind input preprocessing. ConvNeXt, preprocessing included model using Normalization layer.  ConvNeXt models expect inputs float uint8 tensors pixels values [0-255] range. calling summary() method instantiating ConvNeXt model, prefer setting expand_nested argument summary() TRUE better investigate instantiated model.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/application_convnext_tiny.html","id":null,"dir":"Reference","previous_headings":"","what":"Instantiates the ConvNeXtTiny architecture. — application_convnext_tiny","title":"Instantiates the ConvNeXtTiny architecture. — application_convnext_tiny","text":"Instantiates ConvNeXtTiny architecture.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_convnext_tiny.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Instantiates the ConvNeXtTiny architecture. — application_convnext_tiny","text":"","code":"application_convnext_tiny(   include_top = TRUE,   include_preprocessing = TRUE,   weights = \"imagenet\",   input_tensor = NULL,   input_shape = NULL,   pooling = NULL,   classes = 1000L,   classifier_activation = \"softmax\",   name = \"convnext_tiny\" )"},{"path":"https://keras3.posit.co/dev/reference/application_convnext_tiny.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Instantiates the ConvNeXtTiny architecture. — application_convnext_tiny","text":"include_top Whether include fully-connected layer top network. Defaults TRUE. include_preprocessing Boolean, whether include preprocessing layer bottom network. weights One NULL (random initialization), \"imagenet\" (pre-training ImageNet-1k), path weights file loaded. Defaults \"imagenet\". input_tensor Optional Keras tensor (.e. output keras_input()) use image input model. input_shape Optional shape tuple, specified include_top FALSE. exactly 3 inputs channels. pooling Optional pooling mode feature extraction include_top FALSE. Defaults NULL. NULL means output model 4D tensor output last convolutional layer. avg means global average pooling applied output last convolutional layer, thus output model 2D tensor. max means global max pooling applied. classes Optional number classes classify images , specified include_top TRUE, weights argument specified. Defaults 1000 (number ImageNet classes). classifier_activation str callable. activation function use \"top\" layer. Ignored unless include_top=TRUE. Set classifier_activation=NULL return logits \"top\" layer. Defaults \"softmax\". loading pretrained weights, classifier_activation can NULL \"softmax\". name name model (string).","code":""},{"path":"https://keras3.posit.co/dev/reference/application_convnext_tiny.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Instantiates the ConvNeXtTiny architecture. — application_convnext_tiny","text":"model instance.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_convnext_tiny.html","id":"references","dir":"Reference","previous_headings":"","what":"References","title":"Instantiates the ConvNeXtTiny architecture. — application_convnext_tiny","text":"ConvNet 2020s (CVPR 2022) image classification use cases, see page detailed examples. transfer learning use cases, make sure read guide transfer learning & fine-tuning. base, large, xlarge models first pre-trained ImageNet-21k dataset fine-tuned ImageNet-1k dataset. pre-trained parameters models assembled official repository. get sense parameters converted Keras compatible parameters, please refer repository.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_convnext_tiny.html","id":"note","dir":"Reference","previous_headings":"","what":"Note","title":"Instantiates the ConvNeXtTiny architecture. — application_convnext_tiny","text":"Keras Application expects specific kind input preprocessing. ConvNeXt, preprocessing included model using Normalization layer.  ConvNeXt models expect inputs float uint8 tensors pixels values [0-255] range. calling summary() method instantiating ConvNeXt model, prefer setting expand_nested argument summary() TRUE better investigate instantiated model.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/application_convnext_xlarge.html","id":null,"dir":"Reference","previous_headings":"","what":"Instantiates the ConvNeXtXLarge architecture. — application_convnext_xlarge","title":"Instantiates the ConvNeXtXLarge architecture. — application_convnext_xlarge","text":"Instantiates ConvNeXtXLarge architecture.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_convnext_xlarge.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Instantiates the ConvNeXtXLarge architecture. — application_convnext_xlarge","text":"","code":"application_convnext_xlarge(   include_top = TRUE,   include_preprocessing = TRUE,   weights = \"imagenet\",   input_tensor = NULL,   input_shape = NULL,   pooling = NULL,   classes = 1000L,   classifier_activation = \"softmax\",   name = \"convnext_xlarge\" )"},{"path":"https://keras3.posit.co/dev/reference/application_convnext_xlarge.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Instantiates the ConvNeXtXLarge architecture. — application_convnext_xlarge","text":"include_top Whether include fully-connected layer top network. Defaults TRUE. include_preprocessing Boolean, whether include preprocessing layer bottom network. weights One NULL (random initialization), \"imagenet\" (pre-training ImageNet-1k), path weights file loaded. Defaults \"imagenet\". input_tensor Optional Keras tensor (.e. output keras_input()) use image input model. input_shape Optional shape tuple, specified include_top FALSE. exactly 3 inputs channels. pooling Optional pooling mode feature extraction include_top FALSE. Defaults NULL. NULL means output model 4D tensor output last convolutional layer. avg means global average pooling applied output last convolutional layer, thus output model 2D tensor. max means global max pooling applied. classes Optional number classes classify images , specified include_top TRUE, weights argument specified. Defaults 1000 (number ImageNet classes). classifier_activation str callable. activation function use \"top\" layer. Ignored unless include_top=TRUE. Set classifier_activation=NULL return logits \"top\" layer. Defaults \"softmax\". loading pretrained weights, classifier_activation can NULL \"softmax\". name name model (string).","code":""},{"path":"https://keras3.posit.co/dev/reference/application_convnext_xlarge.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Instantiates the ConvNeXtXLarge architecture. — application_convnext_xlarge","text":"model instance.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_convnext_xlarge.html","id":"references","dir":"Reference","previous_headings":"","what":"References","title":"Instantiates the ConvNeXtXLarge architecture. — application_convnext_xlarge","text":"ConvNet 2020s (CVPR 2022) image classification use cases, see page detailed examples. transfer learning use cases, make sure read guide transfer learning & fine-tuning. base, large, xlarge models first pre-trained ImageNet-21k dataset fine-tuned ImageNet-1k dataset. pre-trained parameters models assembled official repository. get sense parameters converted Keras compatible parameters, please refer repository.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_convnext_xlarge.html","id":"note","dir":"Reference","previous_headings":"","what":"Note","title":"Instantiates the ConvNeXtXLarge architecture. — application_convnext_xlarge","text":"Keras Application expects specific kind input preprocessing. ConvNeXt, preprocessing included model using Normalization layer.  ConvNeXt models expect inputs float uint8 tensors pixels values [0-255] range. calling summary() method instantiating ConvNeXt model, prefer setting expand_nested argument summary() TRUE better investigate instantiated model.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/application_densenet121.html","id":null,"dir":"Reference","previous_headings":"","what":"Instantiates the Densenet121 architecture. — application_densenet121","title":"Instantiates the Densenet121 architecture. — application_densenet121","text":"Instantiates Densenet121 architecture.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_densenet121.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Instantiates the Densenet121 architecture. — application_densenet121","text":"","code":"application_densenet121(   include_top = TRUE,   weights = \"imagenet\",   input_tensor = NULL,   input_shape = NULL,   pooling = NULL,   classes = 1000L,   classifier_activation = \"softmax\",   name = \"densenet121\" )"},{"path":"https://keras3.posit.co/dev/reference/application_densenet121.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Instantiates the Densenet121 architecture. — application_densenet121","text":"include_top whether include fully-connected layer top network. weights one NULL (random initialization), \"imagenet\" (pre-training ImageNet), path weights file loaded. input_tensor optional Keras tensor (.e. output keras_input()) use image input model. input_shape optional shape tuple, specified include_top FALSE (otherwise input shape (224, 224, 3) ('channels_last' data format) (3, 224, 224) ('channels_first' data format). exactly 3 inputs channels, width height smaller 32. E.g. (200, 200, 3) one valid value. pooling Optional pooling mode feature extraction include_top FALSE. NULL means output model 4D tensor output last convolutional block. avg means global average pooling applied output last convolutional block, thus output model 2D tensor. max means global max pooling applied. classes optional number classes classify images , specified include_top TRUE, weights argument specified. Defaults 1000. classifier_activation str callable. activation function use \"top\" layer. Ignored unless include_top=TRUE. Set classifier_activation=NULL return logits \"top\" layer. loading pretrained weights, classifier_activation can NULL \"softmax\". name name model (string).","code":""},{"path":"https://keras3.posit.co/dev/reference/application_densenet121.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Instantiates the Densenet121 architecture. — application_densenet121","text":"Keras model instance.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_densenet121.html","id":"reference","dir":"Reference","previous_headings":"","what":"Reference","title":"Instantiates the Densenet121 architecture. — application_densenet121","text":"Densely Connected Convolutional Networks (CVPR 2017) Optionally loads weights pre-trained ImageNet. Note data format convention used model one specified Keras config ~/.keras/keras.json.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_densenet121.html","id":"note","dir":"Reference","previous_headings":"","what":"Note","title":"Instantiates the Densenet121 architecture. — application_densenet121","text":"Keras Application expects specific kind input preprocessing. DenseNet, call application_preprocess_inputs() inputs passing model.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/application_densenet169.html","id":null,"dir":"Reference","previous_headings":"","what":"Instantiates the Densenet169 architecture. — application_densenet169","title":"Instantiates the Densenet169 architecture. — application_densenet169","text":"Instantiates Densenet169 architecture.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_densenet169.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Instantiates the Densenet169 architecture. — application_densenet169","text":"","code":"application_densenet169(   include_top = TRUE,   weights = \"imagenet\",   input_tensor = NULL,   input_shape = NULL,   pooling = NULL,   classes = 1000L,   classifier_activation = \"softmax\",   name = \"densenet169\" )"},{"path":"https://keras3.posit.co/dev/reference/application_densenet169.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Instantiates the Densenet169 architecture. — application_densenet169","text":"include_top whether include fully-connected layer top network. weights one NULL (random initialization), \"imagenet\" (pre-training ImageNet), path weights file loaded. input_tensor optional Keras tensor (.e. output keras_input()) use image input model. input_shape optional shape tuple, specified include_top FALSE (otherwise input shape (224, 224, 3) ('channels_last' data format) (3, 224, 224) ('channels_first' data format). exactly 3 inputs channels, width height smaller 32. E.g. (200, 200, 3) one valid value. pooling Optional pooling mode feature extraction include_top FALSE. NULL means output model 4D tensor output last convolutional block. avg means global average pooling applied output last convolutional block, thus output model 2D tensor. max means global max pooling applied. classes optional number classes classify images , specified include_top TRUE, weights argument specified. Defaults 1000. classifier_activation str callable. activation function use \"top\" layer. Ignored unless include_top=TRUE. Set classifier_activation=NULL return logits \"top\" layer. loading pretrained weights, classifier_activation can NULL \"softmax\". name name model (string).","code":""},{"path":"https://keras3.posit.co/dev/reference/application_densenet169.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Instantiates the Densenet169 architecture. — application_densenet169","text":"Keras model instance.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_densenet169.html","id":"reference","dir":"Reference","previous_headings":"","what":"Reference","title":"Instantiates the Densenet169 architecture. — application_densenet169","text":"Densely Connected Convolutional Networks (CVPR 2017) Optionally loads weights pre-trained ImageNet. Note data format convention used model one specified Keras config ~/.keras/keras.json.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_densenet169.html","id":"note","dir":"Reference","previous_headings":"","what":"Note","title":"Instantiates the Densenet169 architecture. — application_densenet169","text":"Keras Application expects specific kind input preprocessing. DenseNet, call application_preprocess_inputs() inputs passing model.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/application_densenet201.html","id":null,"dir":"Reference","previous_headings":"","what":"Instantiates the Densenet201 architecture. — application_densenet201","title":"Instantiates the Densenet201 architecture. — application_densenet201","text":"Instantiates Densenet201 architecture.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_densenet201.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Instantiates the Densenet201 architecture. — application_densenet201","text":"","code":"application_densenet201(   include_top = TRUE,   weights = \"imagenet\",   input_tensor = NULL,   input_shape = NULL,   pooling = NULL,   classes = 1000L,   classifier_activation = \"softmax\",   name = \"densenet201\" )"},{"path":"https://keras3.posit.co/dev/reference/application_densenet201.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Instantiates the Densenet201 architecture. — application_densenet201","text":"include_top whether include fully-connected layer top network. weights one NULL (random initialization), \"imagenet\" (pre-training ImageNet), path weights file loaded. input_tensor optional Keras tensor (.e. output keras_input()) use image input model. input_shape optional shape tuple, specified include_top FALSE (otherwise input shape (224, 224, 3) ('channels_last' data format) (3, 224, 224) ('channels_first' data format). exactly 3 inputs channels, width height smaller 32. E.g. (200, 200, 3) one valid value. pooling Optional pooling mode feature extraction include_top FALSE. NULL means output model 4D tensor output last convolutional block. avg means global average pooling applied output last convolutional block, thus output model 2D tensor. max means global max pooling applied. classes optional number classes classify images , specified include_top TRUE, weights argument specified. Defaults 1000. classifier_activation str callable. activation function use \"top\" layer. Ignored unless include_top=TRUE. Set classifier_activation=NULL return logits \"top\" layer. loading pretrained weights, classifier_activation can NULL \"softmax\". name name model (string).","code":""},{"path":"https://keras3.posit.co/dev/reference/application_densenet201.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Instantiates the Densenet201 architecture. — application_densenet201","text":"Keras model instance.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_densenet201.html","id":"reference","dir":"Reference","previous_headings":"","what":"Reference","title":"Instantiates the Densenet201 architecture. — application_densenet201","text":"Densely Connected Convolutional Networks (CVPR 2017) Optionally loads weights pre-trained ImageNet. Note data format convention used model one specified Keras config ~/.keras/keras.json.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_densenet201.html","id":"note","dir":"Reference","previous_headings":"","what":"Note","title":"Instantiates the Densenet201 architecture. — application_densenet201","text":"Keras Application expects specific kind input preprocessing. DenseNet, call application_preprocess_inputs() inputs passing model.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_b0.html","id":null,"dir":"Reference","previous_headings":"","what":"Instantiates the EfficientNetB0 architecture. — application_efficientnet_b0","title":"Instantiates the EfficientNetB0 architecture. — application_efficientnet_b0","text":"Instantiates EfficientNetB0 architecture.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_b0.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Instantiates the EfficientNetB0 architecture. — application_efficientnet_b0","text":"","code":"application_efficientnet_b0(   include_top = TRUE,   weights = \"imagenet\",   input_tensor = NULL,   input_shape = NULL,   pooling = NULL,   classes = 1000L,   classifier_activation = \"softmax\",   name = \"efficientnetb0\",   ... )"},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_b0.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Instantiates the EfficientNetB0 architecture. — application_efficientnet_b0","text":"include_top Whether include fully-connected layer top network. Defaults TRUE. weights One NULL (random initialization), \"imagenet\" (pre-training ImageNet), path weights file loaded. Defaults \"imagenet\". input_tensor Optional Keras tensor (.e. output keras_input()) use image input model. input_shape Optional shape tuple, specified include_top FALSE. exactly 3 inputs channels. pooling Optional pooling mode feature extraction include_top FALSE. Defaults NULL. NULL means output model 4D tensor output last convolutional layer. avg means global average pooling applied output last convolutional layer, thus output model 2D tensor. max means global max pooling applied. classes Optional number classes classify images , specified include_top TRUE, weights argument specified. 1000 many ImageNet classes . Defaults 1000. classifier_activation str callable. activation function use \"top\" layer. Ignored unless include_top=TRUE. Set classifier_activation=NULL return logits \"top\" layer. Defaults 'softmax'. loading pretrained weights, classifier_activation can NULL \"softmax\". name name model (string). ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_b0.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Instantiates the EfficientNetB0 architecture. — application_efficientnet_b0","text":"model instance.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_b0.html","id":"reference","dir":"Reference","previous_headings":"","what":"Reference","title":"Instantiates the EfficientNetB0 architecture. — application_efficientnet_b0","text":"EfficientNet: Rethinking Model Scaling Convolutional Neural Networks (ICML 2019) function returns Keras image classification model, optionally loaded weights pre-trained ImageNet. image classification use cases, see page detailed examples. transfer learning use cases, make sure read guide transfer learning & fine-tuning.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_b0.html","id":"note","dir":"Reference","previous_headings":"","what":"Note","title":"Instantiates the EfficientNetB0 architecture. — application_efficientnet_b0","text":"Keras Application expects specific kind input preprocessing. EfficientNet, input preprocessing included part model (Rescaling layer), thus application_preprocess_inputs() actually pass-function. EfficientNet models expect inputs float tensors pixels values [0-255] range.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_b1.html","id":null,"dir":"Reference","previous_headings":"","what":"Instantiates the EfficientNetB1 architecture. — application_efficientnet_b1","title":"Instantiates the EfficientNetB1 architecture. — application_efficientnet_b1","text":"Instantiates EfficientNetB1 architecture.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_b1.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Instantiates the EfficientNetB1 architecture. — application_efficientnet_b1","text":"","code":"application_efficientnet_b1(   include_top = TRUE,   weights = \"imagenet\",   input_tensor = NULL,   input_shape = NULL,   pooling = NULL,   classes = 1000L,   classifier_activation = \"softmax\",   name = \"efficientnetb1\",   ... )"},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_b1.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Instantiates the EfficientNetB1 architecture. — application_efficientnet_b1","text":"include_top Whether include fully-connected layer top network. Defaults TRUE. weights One NULL (random initialization), \"imagenet\" (pre-training ImageNet), path weights file loaded. Defaults \"imagenet\". input_tensor Optional Keras tensor (.e. output keras_input()) use image input model. input_shape Optional shape tuple, specified include_top FALSE. exactly 3 inputs channels. pooling Optional pooling mode feature extraction include_top FALSE. Defaults NULL. NULL means output model 4D tensor output last convolutional layer. avg means global average pooling applied output last convolutional layer, thus output model 2D tensor. max means global max pooling applied. classes Optional number classes classify images , specified include_top TRUE, weights argument specified. 1000 many ImageNet classes . Defaults 1000. classifier_activation str callable. activation function use \"top\" layer. Ignored unless include_top=TRUE. Set classifier_activation=NULL return logits \"top\" layer. Defaults 'softmax'. loading pretrained weights, classifier_activation can NULL \"softmax\". name name model (string). ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_b1.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Instantiates the EfficientNetB1 architecture. — application_efficientnet_b1","text":"model instance.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_b1.html","id":"reference","dir":"Reference","previous_headings":"","what":"Reference","title":"Instantiates the EfficientNetB1 architecture. — application_efficientnet_b1","text":"EfficientNet: Rethinking Model Scaling Convolutional Neural Networks (ICML 2019) function returns Keras image classification model, optionally loaded weights pre-trained ImageNet. image classification use cases, see page detailed examples. transfer learning use cases, make sure read guide transfer learning & fine-tuning.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_b1.html","id":"note","dir":"Reference","previous_headings":"","what":"Note","title":"Instantiates the EfficientNetB1 architecture. — application_efficientnet_b1","text":"Keras Application expects specific kind input preprocessing. EfficientNet, input preprocessing included part model (Rescaling layer), thus application_preprocess_inputs() actually pass-function. EfficientNet models expect inputs float tensors pixels values [0-255] range.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_b2.html","id":null,"dir":"Reference","previous_headings":"","what":"Instantiates the EfficientNetB2 architecture. — application_efficientnet_b2","title":"Instantiates the EfficientNetB2 architecture. — application_efficientnet_b2","text":"Instantiates EfficientNetB2 architecture.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_b2.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Instantiates the EfficientNetB2 architecture. — application_efficientnet_b2","text":"","code":"application_efficientnet_b2(   include_top = TRUE,   weights = \"imagenet\",   input_tensor = NULL,   input_shape = NULL,   pooling = NULL,   classes = 1000L,   classifier_activation = \"softmax\",   name = \"efficientnetb2\",   ... )"},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_b2.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Instantiates the EfficientNetB2 architecture. — application_efficientnet_b2","text":"include_top Whether include fully-connected layer top network. Defaults TRUE. weights One NULL (random initialization), \"imagenet\" (pre-training ImageNet), path weights file loaded. Defaults \"imagenet\". input_tensor Optional Keras tensor (.e. output keras_input()) use image input model. input_shape Optional shape tuple, specified include_top FALSE. exactly 3 inputs channels. pooling Optional pooling mode feature extraction include_top FALSE. Defaults NULL. NULL means output model 4D tensor output last convolutional layer. avg means global average pooling applied output last convolutional layer, thus output model 2D tensor. max means global max pooling applied. classes Optional number classes classify images , specified include_top TRUE, weights argument specified. 1000 many ImageNet classes . Defaults 1000. classifier_activation str callable. activation function use \"top\" layer. Ignored unless include_top=TRUE. Set classifier_activation=NULL return logits \"top\" layer. Defaults 'softmax'. loading pretrained weights, classifier_activation can NULL \"softmax\". name name model (string). ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_b2.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Instantiates the EfficientNetB2 architecture. — application_efficientnet_b2","text":"model instance.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_b2.html","id":"reference","dir":"Reference","previous_headings":"","what":"Reference","title":"Instantiates the EfficientNetB2 architecture. — application_efficientnet_b2","text":"EfficientNet: Rethinking Model Scaling Convolutional Neural Networks (ICML 2019) function returns Keras image classification model, optionally loaded weights pre-trained ImageNet. image classification use cases, see page detailed examples. transfer learning use cases, make sure read guide transfer learning & fine-tuning.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_b2.html","id":"note","dir":"Reference","previous_headings":"","what":"Note","title":"Instantiates the EfficientNetB2 architecture. — application_efficientnet_b2","text":"Keras Application expects specific kind input preprocessing. EfficientNet, input preprocessing included part model (Rescaling layer), thus application_preprocess_inputs() actually pass-function. EfficientNet models expect inputs float tensors pixels values [0-255] range.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_b3.html","id":null,"dir":"Reference","previous_headings":"","what":"Instantiates the EfficientNetB3 architecture. — application_efficientnet_b3","title":"Instantiates the EfficientNetB3 architecture. — application_efficientnet_b3","text":"Instantiates EfficientNetB3 architecture.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_b3.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Instantiates the EfficientNetB3 architecture. — application_efficientnet_b3","text":"","code":"application_efficientnet_b3(   include_top = TRUE,   weights = \"imagenet\",   input_tensor = NULL,   input_shape = NULL,   pooling = NULL,   classes = 1000L,   classifier_activation = \"softmax\",   name = \"efficientnetb3\",   ... )"},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_b3.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Instantiates the EfficientNetB3 architecture. — application_efficientnet_b3","text":"include_top Whether include fully-connected layer top network. Defaults TRUE. weights One NULL (random initialization), \"imagenet\" (pre-training ImageNet), path weights file loaded. Defaults \"imagenet\". input_tensor Optional Keras tensor (.e. output keras_input()) use image input model. input_shape Optional shape tuple, specified include_top FALSE. exactly 3 inputs channels. pooling Optional pooling mode feature extraction include_top FALSE. Defaults NULL. NULL means output model 4D tensor output last convolutional layer. avg means global average pooling applied output last convolutional layer, thus output model 2D tensor. max means global max pooling applied. classes Optional number classes classify images , specified include_top TRUE, weights argument specified. 1000 many ImageNet classes . Defaults 1000. classifier_activation str callable. activation function use \"top\" layer. Ignored unless include_top=TRUE. Set classifier_activation=NULL return logits \"top\" layer. Defaults 'softmax'. loading pretrained weights, classifier_activation can NULL \"softmax\". name name model (string). ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_b3.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Instantiates the EfficientNetB3 architecture. — application_efficientnet_b3","text":"model instance.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_b3.html","id":"reference","dir":"Reference","previous_headings":"","what":"Reference","title":"Instantiates the EfficientNetB3 architecture. — application_efficientnet_b3","text":"EfficientNet: Rethinking Model Scaling Convolutional Neural Networks (ICML 2019) function returns Keras image classification model, optionally loaded weights pre-trained ImageNet. image classification use cases, see page detailed examples. transfer learning use cases, make sure read guide transfer learning & fine-tuning.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_b3.html","id":"note","dir":"Reference","previous_headings":"","what":"Note","title":"Instantiates the EfficientNetB3 architecture. — application_efficientnet_b3","text":"Keras Application expects specific kind input preprocessing. EfficientNet, input preprocessing included part model (Rescaling layer), thus application_preprocess_inputs() actually pass-function. EfficientNet models expect inputs float tensors pixels values [0-255] range.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_b4.html","id":null,"dir":"Reference","previous_headings":"","what":"Instantiates the EfficientNetB4 architecture. — application_efficientnet_b4","title":"Instantiates the EfficientNetB4 architecture. — application_efficientnet_b4","text":"Instantiates EfficientNetB4 architecture.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_b4.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Instantiates the EfficientNetB4 architecture. — application_efficientnet_b4","text":"","code":"application_efficientnet_b4(   include_top = TRUE,   weights = \"imagenet\",   input_tensor = NULL,   input_shape = NULL,   pooling = NULL,   classes = 1000L,   classifier_activation = \"softmax\",   name = \"efficientnetb4\",   ... )"},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_b4.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Instantiates the EfficientNetB4 architecture. — application_efficientnet_b4","text":"include_top Whether include fully-connected layer top network. Defaults TRUE. weights One NULL (random initialization), \"imagenet\" (pre-training ImageNet), path weights file loaded. Defaults \"imagenet\". input_tensor Optional Keras tensor (.e. output keras_input()) use image input model. input_shape Optional shape tuple, specified include_top FALSE. exactly 3 inputs channels. pooling Optional pooling mode feature extraction include_top FALSE. Defaults NULL. NULL means output model 4D tensor output last convolutional layer. avg means global average pooling applied output last convolutional layer, thus output model 2D tensor. max means global max pooling applied. classes Optional number classes classify images , specified include_top TRUE, weights argument specified. 1000 many ImageNet classes . Defaults 1000. classifier_activation str callable. activation function use \"top\" layer. Ignored unless include_top=TRUE. Set classifier_activation=NULL return logits \"top\" layer. Defaults 'softmax'. loading pretrained weights, classifier_activation can NULL \"softmax\". name name model (string). ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_b4.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Instantiates the EfficientNetB4 architecture. — application_efficientnet_b4","text":"model instance.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_b4.html","id":"reference","dir":"Reference","previous_headings":"","what":"Reference","title":"Instantiates the EfficientNetB4 architecture. — application_efficientnet_b4","text":"EfficientNet: Rethinking Model Scaling Convolutional Neural Networks (ICML 2019) function returns Keras image classification model, optionally loaded weights pre-trained ImageNet. image classification use cases, see page detailed examples. transfer learning use cases, make sure read guide transfer learning & fine-tuning.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_b4.html","id":"note","dir":"Reference","previous_headings":"","what":"Note","title":"Instantiates the EfficientNetB4 architecture. — application_efficientnet_b4","text":"Keras Application expects specific kind input preprocessing. EfficientNet, input preprocessing included part model (Rescaling layer), thus application_preprocess_inputs() actually pass-function. EfficientNet models expect inputs float tensors pixels values [0-255] range.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_b5.html","id":null,"dir":"Reference","previous_headings":"","what":"Instantiates the EfficientNetB5 architecture. — application_efficientnet_b5","title":"Instantiates the EfficientNetB5 architecture. — application_efficientnet_b5","text":"Instantiates EfficientNetB5 architecture.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_b5.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Instantiates the EfficientNetB5 architecture. — application_efficientnet_b5","text":"","code":"application_efficientnet_b5(   include_top = TRUE,   weights = \"imagenet\",   input_tensor = NULL,   input_shape = NULL,   pooling = NULL,   classes = 1000L,   classifier_activation = \"softmax\",   name = \"efficientnetb5\",   ... )"},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_b5.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Instantiates the EfficientNetB5 architecture. — application_efficientnet_b5","text":"include_top Whether include fully-connected layer top network. Defaults TRUE. weights One NULL (random initialization), \"imagenet\" (pre-training ImageNet), path weights file loaded. Defaults \"imagenet\". input_tensor Optional Keras tensor (.e. output keras_input()) use image input model. input_shape Optional shape tuple, specified include_top FALSE. exactly 3 inputs channels. pooling Optional pooling mode feature extraction include_top FALSE. Defaults NULL. NULL means output model 4D tensor output last convolutional layer. avg means global average pooling applied output last convolutional layer, thus output model 2D tensor. max means global max pooling applied. classes Optional number classes classify images , specified include_top TRUE, weights argument specified. 1000 many ImageNet classes . Defaults 1000. classifier_activation str callable. activation function use \"top\" layer. Ignored unless include_top=TRUE. Set classifier_activation=NULL return logits \"top\" layer. Defaults 'softmax'. loading pretrained weights, classifier_activation can NULL \"softmax\". name name model (string). ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_b5.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Instantiates the EfficientNetB5 architecture. — application_efficientnet_b5","text":"model instance.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_b5.html","id":"reference","dir":"Reference","previous_headings":"","what":"Reference","title":"Instantiates the EfficientNetB5 architecture. — application_efficientnet_b5","text":"EfficientNet: Rethinking Model Scaling Convolutional Neural Networks (ICML 2019) function returns Keras image classification model, optionally loaded weights pre-trained ImageNet. image classification use cases, see page detailed examples. transfer learning use cases, make sure read guide transfer learning & fine-tuning.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_b5.html","id":"note","dir":"Reference","previous_headings":"","what":"Note","title":"Instantiates the EfficientNetB5 architecture. — application_efficientnet_b5","text":"Keras Application expects specific kind input preprocessing. EfficientNet, input preprocessing included part model (Rescaling layer), thus application_preprocess_inputs() actually pass-function. EfficientNet models expect inputs float tensors pixels values [0-255] range.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_b6.html","id":null,"dir":"Reference","previous_headings":"","what":"Instantiates the EfficientNetB6 architecture. — application_efficientnet_b6","title":"Instantiates the EfficientNetB6 architecture. — application_efficientnet_b6","text":"Instantiates EfficientNetB6 architecture.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_b6.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Instantiates the EfficientNetB6 architecture. — application_efficientnet_b6","text":"","code":"application_efficientnet_b6(   include_top = TRUE,   weights = \"imagenet\",   input_tensor = NULL,   input_shape = NULL,   pooling = NULL,   classes = 1000L,   classifier_activation = \"softmax\",   name = \"efficientnetb6\",   ... )"},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_b6.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Instantiates the EfficientNetB6 architecture. — application_efficientnet_b6","text":"include_top Whether include fully-connected layer top network. Defaults TRUE. weights One NULL (random initialization), \"imagenet\" (pre-training ImageNet), path weights file loaded. Defaults \"imagenet\". input_tensor Optional Keras tensor (.e. output keras_input()) use image input model. input_shape Optional shape tuple, specified include_top FALSE. exactly 3 inputs channels. pooling Optional pooling mode feature extraction include_top FALSE. Defaults NULL. NULL means output model 4D tensor output last convolutional layer. avg means global average pooling applied output last convolutional layer, thus output model 2D tensor. max means global max pooling applied. classes Optional number classes classify images , specified include_top TRUE, weights argument specified. 1000 many ImageNet classes . Defaults 1000. classifier_activation str callable. activation function use \"top\" layer. Ignored unless include_top=TRUE. Set classifier_activation=NULL return logits \"top\" layer. Defaults 'softmax'. loading pretrained weights, classifier_activation can NULL \"softmax\". name name model (string). ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_b6.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Instantiates the EfficientNetB6 architecture. — application_efficientnet_b6","text":"model instance.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_b6.html","id":"reference","dir":"Reference","previous_headings":"","what":"Reference","title":"Instantiates the EfficientNetB6 architecture. — application_efficientnet_b6","text":"EfficientNet: Rethinking Model Scaling Convolutional Neural Networks (ICML 2019) function returns Keras image classification model, optionally loaded weights pre-trained ImageNet. image classification use cases, see page detailed examples. transfer learning use cases, make sure read guide transfer learning & fine-tuning.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_b6.html","id":"note","dir":"Reference","previous_headings":"","what":"Note","title":"Instantiates the EfficientNetB6 architecture. — application_efficientnet_b6","text":"Keras Application expects specific kind input preprocessing. EfficientNet, input preprocessing included part model (Rescaling layer), thus application_preprocess_inputs() actually pass-function. EfficientNet models expect inputs float tensors pixels values [0-255] range.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_b7.html","id":null,"dir":"Reference","previous_headings":"","what":"Instantiates the EfficientNetB7 architecture. — application_efficientnet_b7","title":"Instantiates the EfficientNetB7 architecture. — application_efficientnet_b7","text":"Instantiates EfficientNetB7 architecture.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_b7.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Instantiates the EfficientNetB7 architecture. — application_efficientnet_b7","text":"","code":"application_efficientnet_b7(   include_top = TRUE,   weights = \"imagenet\",   input_tensor = NULL,   input_shape = NULL,   pooling = NULL,   classes = 1000L,   classifier_activation = \"softmax\",   name = \"efficientnetb7\",   ... )"},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_b7.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Instantiates the EfficientNetB7 architecture. — application_efficientnet_b7","text":"include_top Whether include fully-connected layer top network. Defaults TRUE. weights One NULL (random initialization), \"imagenet\" (pre-training ImageNet), path weights file loaded. Defaults \"imagenet\". input_tensor Optional Keras tensor (.e. output keras_input()) use image input model. input_shape Optional shape tuple, specified include_top FALSE. exactly 3 inputs channels. pooling Optional pooling mode feature extraction include_top FALSE. Defaults NULL. NULL means output model 4D tensor output last convolutional layer. avg means global average pooling applied output last convolutional layer, thus output model 2D tensor. max means global max pooling applied. classes Optional number classes classify images , specified include_top TRUE, weights argument specified. 1000 many ImageNet classes . Defaults 1000. classifier_activation str callable. activation function use \"top\" layer. Ignored unless include_top=TRUE. Set classifier_activation=NULL return logits \"top\" layer. Defaults 'softmax'. loading pretrained weights, classifier_activation can NULL \"softmax\". name name model (string). ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_b7.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Instantiates the EfficientNetB7 architecture. — application_efficientnet_b7","text":"model instance.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_b7.html","id":"reference","dir":"Reference","previous_headings":"","what":"Reference","title":"Instantiates the EfficientNetB7 architecture. — application_efficientnet_b7","text":"EfficientNet: Rethinking Model Scaling Convolutional Neural Networks (ICML 2019) function returns Keras image classification model, optionally loaded weights pre-trained ImageNet. image classification use cases, see page detailed examples. transfer learning use cases, make sure read guide transfer learning & fine-tuning.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_b7.html","id":"note","dir":"Reference","previous_headings":"","what":"Note","title":"Instantiates the EfficientNetB7 architecture. — application_efficientnet_b7","text":"Keras Application expects specific kind input preprocessing. EfficientNet, input preprocessing included part model (Rescaling layer), thus application_preprocess_inputs() actually pass-function. EfficientNet models expect inputs float tensors pixels values [0-255] range.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_v2b0.html","id":null,"dir":"Reference","previous_headings":"","what":"Instantiates the EfficientNetV2B0 architecture. — application_efficientnet_v2b0","title":"Instantiates the EfficientNetV2B0 architecture. — application_efficientnet_v2b0","text":"Instantiates EfficientNetV2B0 architecture.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_v2b0.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Instantiates the EfficientNetV2B0 architecture. — application_efficientnet_v2b0","text":"","code":"application_efficientnet_v2b0(   include_top = TRUE,   weights = \"imagenet\",   input_tensor = NULL,   input_shape = NULL,   pooling = NULL,   classes = 1000L,   classifier_activation = \"softmax\",   include_preprocessing = TRUE,   name = \"efficientnetv2-b0\" )"},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_v2b0.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Instantiates the EfficientNetV2B0 architecture. — application_efficientnet_v2b0","text":"include_top Boolean, whether include fully-connected layer top network. Defaults TRUE. weights One NULL (random initialization), \"imagenet\" (pre-training ImageNet), path weights file loaded. Defaults \"imagenet\". input_tensor Optional Keras tensor (.e. output keras_input()) use image input model. input_shape Optional shape tuple, specified include_top FALSE. exactly 3 inputs channels. pooling Optional pooling mode feature extraction include_top FALSE. Defaults NULL. NULL means output model 4D tensor output last convolutional layer. \"avg\" means global average pooling applied output last convolutional layer, thus output model 2D tensor. \"max\" means global max pooling applied. classes Optional number classes classify images , specified include_top TRUE, weights argument specified. Defaults 1000 (number ImageNet classes). classifier_activation string callable. activation function use \"top\" layer. Ignored unless include_top=TRUE. Set classifier_activation=NULL return logits \"top\" layer. Defaults \"softmax\". loading pretrained weights, classifier_activation can NULL \"softmax\". include_preprocessing Boolean, whether include preprocessing layer bottom network. name name model (string).","code":""},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_v2b0.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Instantiates the EfficientNetV2B0 architecture. — application_efficientnet_v2b0","text":"model instance.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_v2b0.html","id":"reference","dir":"Reference","previous_headings":"","what":"Reference","title":"Instantiates the EfficientNetV2B0 architecture. — application_efficientnet_v2b0","text":"EfficientNetV2: Smaller Models Faster Training (ICML 2021) function returns Keras image classification model, optionally loaded weights pre-trained ImageNet. image classification use cases, see page detailed examples. transfer learning use cases, make sure read guide transfer learning & fine-tuning.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_v2b0.html","id":"note","dir":"Reference","previous_headings":"","what":"Note","title":"Instantiates the EfficientNetV2B0 architecture. — application_efficientnet_v2b0","text":"Keras Application expects specific kind input preprocessing. EfficientNetV2, default input preprocessing included part model (Rescaling layer), thus application_preprocess_inputs() actually pass-function. use case, EfficientNetV2 models expect inputs float tensors pixels values [0, 255] range. time, preprocessing part model (.e. Rescaling layer) can disabled setting include_preprocessing argument FALSE. preprocessing disabled EfficientNetV2 models expect inputs float tensors pixels values [-1, 1] range.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_v2b1.html","id":null,"dir":"Reference","previous_headings":"","what":"Instantiates the EfficientNetV2B1 architecture. — application_efficientnet_v2b1","title":"Instantiates the EfficientNetV2B1 architecture. — application_efficientnet_v2b1","text":"Instantiates EfficientNetV2B1 architecture.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_v2b1.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Instantiates the EfficientNetV2B1 architecture. — application_efficientnet_v2b1","text":"","code":"application_efficientnet_v2b1(   include_top = TRUE,   weights = \"imagenet\",   input_tensor = NULL,   input_shape = NULL,   pooling = NULL,   classes = 1000L,   classifier_activation = \"softmax\",   include_preprocessing = TRUE,   name = \"efficientnetv2-b1\" )"},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_v2b1.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Instantiates the EfficientNetV2B1 architecture. — application_efficientnet_v2b1","text":"include_top Boolean, whether include fully-connected layer top network. Defaults TRUE. weights One NULL (random initialization), \"imagenet\" (pre-training ImageNet), path weights file loaded. Defaults \"imagenet\". input_tensor Optional Keras tensor (.e. output keras_input()) use image input model. input_shape Optional shape tuple, specified include_top FALSE. exactly 3 inputs channels. pooling Optional pooling mode feature extraction include_top FALSE. Defaults NULL. NULL means output model 4D tensor output last convolutional layer. \"avg\" means global average pooling applied output last convolutional layer, thus output model 2D tensor. \"max\" means global max pooling applied. classes Optional number classes classify images , specified include_top TRUE, weights argument specified. Defaults 1000 (number ImageNet classes). classifier_activation string callable. activation function use \"top\" layer. Ignored unless include_top=TRUE. Set classifier_activation=NULL return logits \"top\" layer. Defaults \"softmax\". loading pretrained weights, classifier_activation can NULL \"softmax\". include_preprocessing Boolean, whether include preprocessing layer bottom network. name name model (string).","code":""},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_v2b1.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Instantiates the EfficientNetV2B1 architecture. — application_efficientnet_v2b1","text":"model instance.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_v2b1.html","id":"reference","dir":"Reference","previous_headings":"","what":"Reference","title":"Instantiates the EfficientNetV2B1 architecture. — application_efficientnet_v2b1","text":"EfficientNetV2: Smaller Models Faster Training (ICML 2021) function returns Keras image classification model, optionally loaded weights pre-trained ImageNet. image classification use cases, see page detailed examples. transfer learning use cases, make sure read guide transfer learning & fine-tuning.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_v2b1.html","id":"note","dir":"Reference","previous_headings":"","what":"Note","title":"Instantiates the EfficientNetV2B1 architecture. — application_efficientnet_v2b1","text":"Keras Application expects specific kind input preprocessing. EfficientNetV2, default input preprocessing included part model (Rescaling layer), thus application_preprocess_inputs() actually pass-function. use case, EfficientNetV2 models expect inputs float tensors pixels values [0, 255] range. time, preprocessing part model (.e. Rescaling layer) can disabled setting include_preprocessing argument FALSE. preprocessing disabled EfficientNetV2 models expect inputs float tensors pixels values [-1, 1] range.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_v2b2.html","id":null,"dir":"Reference","previous_headings":"","what":"Instantiates the EfficientNetV2B2 architecture. — application_efficientnet_v2b2","title":"Instantiates the EfficientNetV2B2 architecture. — application_efficientnet_v2b2","text":"Instantiates EfficientNetV2B2 architecture.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_v2b2.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Instantiates the EfficientNetV2B2 architecture. — application_efficientnet_v2b2","text":"","code":"application_efficientnet_v2b2(   include_top = TRUE,   weights = \"imagenet\",   input_tensor = NULL,   input_shape = NULL,   pooling = NULL,   classes = 1000L,   classifier_activation = \"softmax\",   include_preprocessing = TRUE,   name = \"efficientnetv2-b2\" )"},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_v2b2.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Instantiates the EfficientNetV2B2 architecture. — application_efficientnet_v2b2","text":"include_top Boolean, whether include fully-connected layer top network. Defaults TRUE. weights One NULL (random initialization), \"imagenet\" (pre-training ImageNet), path weights file loaded. Defaults \"imagenet\". input_tensor Optional Keras tensor (.e. output keras_input()) use image input model. input_shape Optional shape tuple, specified include_top FALSE. exactly 3 inputs channels. pooling Optional pooling mode feature extraction include_top FALSE. Defaults NULL. NULL means output model 4D tensor output last convolutional layer. \"avg\" means global average pooling applied output last convolutional layer, thus output model 2D tensor. \"max\" means global max pooling applied. classes Optional number classes classify images , specified include_top TRUE, weights argument specified. Defaults 1000 (number ImageNet classes). classifier_activation string callable. activation function use \"top\" layer. Ignored unless include_top=TRUE. Set classifier_activation=NULL return logits \"top\" layer. Defaults \"softmax\". loading pretrained weights, classifier_activation can NULL \"softmax\". include_preprocessing Boolean, whether include preprocessing layer bottom network. name name model (string).","code":""},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_v2b2.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Instantiates the EfficientNetV2B2 architecture. — application_efficientnet_v2b2","text":"model instance.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_v2b2.html","id":"reference","dir":"Reference","previous_headings":"","what":"Reference","title":"Instantiates the EfficientNetV2B2 architecture. — application_efficientnet_v2b2","text":"EfficientNetV2: Smaller Models Faster Training (ICML 2021) function returns Keras image classification model, optionally loaded weights pre-trained ImageNet. image classification use cases, see page detailed examples. transfer learning use cases, make sure read guide transfer learning & fine-tuning.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_v2b2.html","id":"note","dir":"Reference","previous_headings":"","what":"Note","title":"Instantiates the EfficientNetV2B2 architecture. — application_efficientnet_v2b2","text":"Keras Application expects specific kind input preprocessing. EfficientNetV2, default input preprocessing included part model (Rescaling layer), thus application_preprocess_inputs() actually pass-function. use case, EfficientNetV2 models expect inputs float tensors pixels values [0, 255] range. time, preprocessing part model (.e. Rescaling layer) can disabled setting include_preprocessing argument FALSE. preprocessing disabled EfficientNetV2 models expect inputs float tensors pixels values [-1, 1] range.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_v2b3.html","id":null,"dir":"Reference","previous_headings":"","what":"Instantiates the EfficientNetV2B3 architecture. — application_efficientnet_v2b3","title":"Instantiates the EfficientNetV2B3 architecture. — application_efficientnet_v2b3","text":"Instantiates EfficientNetV2B3 architecture.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_v2b3.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Instantiates the EfficientNetV2B3 architecture. — application_efficientnet_v2b3","text":"","code":"application_efficientnet_v2b3(   include_top = TRUE,   weights = \"imagenet\",   input_tensor = NULL,   input_shape = NULL,   pooling = NULL,   classes = 1000L,   classifier_activation = \"softmax\",   include_preprocessing = TRUE,   name = \"efficientnetv2-b3\" )"},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_v2b3.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Instantiates the EfficientNetV2B3 architecture. — application_efficientnet_v2b3","text":"include_top Boolean, whether include fully-connected layer top network. Defaults TRUE. weights One NULL (random initialization), \"imagenet\" (pre-training ImageNet), path weights file loaded. Defaults \"imagenet\". input_tensor Optional Keras tensor (.e. output keras_input()) use image input model. input_shape Optional shape tuple, specified include_top FALSE. exactly 3 inputs channels. pooling Optional pooling mode feature extraction include_top FALSE. Defaults NULL. NULL means output model 4D tensor output last convolutional layer. \"avg\" means global average pooling applied output last convolutional layer, thus output model 2D tensor. \"max\" means global max pooling applied. classes Optional number classes classify images , specified include_top TRUE, weights argument specified. Defaults 1000 (number ImageNet classes). classifier_activation string callable. activation function use \"top\" layer. Ignored unless include_top=TRUE. Set classifier_activation=NULL return logits \"top\" layer. Defaults \"softmax\". loading pretrained weights, classifier_activation can NULL \"softmax\". include_preprocessing Boolean, whether include preprocessing layer bottom network. name name model (string).","code":""},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_v2b3.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Instantiates the EfficientNetV2B3 architecture. — application_efficientnet_v2b3","text":"model instance.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_v2b3.html","id":"reference","dir":"Reference","previous_headings":"","what":"Reference","title":"Instantiates the EfficientNetV2B3 architecture. — application_efficientnet_v2b3","text":"EfficientNetV2: Smaller Models Faster Training (ICML 2021) function returns Keras image classification model, optionally loaded weights pre-trained ImageNet. image classification use cases, see page detailed examples. transfer learning use cases, make sure read guide transfer learning & fine-tuning.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_v2b3.html","id":"note","dir":"Reference","previous_headings":"","what":"Note","title":"Instantiates the EfficientNetV2B3 architecture. — application_efficientnet_v2b3","text":"Keras Application expects specific kind input preprocessing. EfficientNetV2, default input preprocessing included part model (Rescaling layer), thus application_preprocess_inputs() actually pass-function. use case, EfficientNetV2 models expect inputs float tensors pixels values [0, 255] range. time, preprocessing part model (.e. Rescaling layer) can disabled setting include_preprocessing argument FALSE. preprocessing disabled EfficientNetV2 models expect inputs float tensors pixels values [-1, 1] range.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_v2l.html","id":null,"dir":"Reference","previous_headings":"","what":"Instantiates the EfficientNetV2L architecture. — application_efficientnet_v2l","title":"Instantiates the EfficientNetV2L architecture. — application_efficientnet_v2l","text":"Instantiates EfficientNetV2L architecture.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_v2l.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Instantiates the EfficientNetV2L architecture. — application_efficientnet_v2l","text":"","code":"application_efficientnet_v2l(   include_top = TRUE,   weights = \"imagenet\",   input_tensor = NULL,   input_shape = NULL,   pooling = NULL,   classes = 1000L,   classifier_activation = \"softmax\",   include_preprocessing = TRUE,   name = \"efficientnetv2-l\" )"},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_v2l.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Instantiates the EfficientNetV2L architecture. — application_efficientnet_v2l","text":"include_top Boolean, whether include fully-connected layer top network. Defaults TRUE. weights One NULL (random initialization), \"imagenet\" (pre-training ImageNet), path weights file loaded. Defaults \"imagenet\". input_tensor Optional Keras tensor (.e. output keras_input()) use image input model. input_shape Optional shape tuple, specified include_top FALSE. exactly 3 inputs channels. pooling Optional pooling mode feature extraction include_top FALSE. Defaults NULL. NULL means output model 4D tensor output last convolutional layer. \"avg\" means global average pooling applied output last convolutional layer, thus output model 2D tensor. \"max\" means global max pooling applied. classes Optional number classes classify images , specified include_top TRUE, weights argument specified. Defaults 1000 (number ImageNet classes). classifier_activation string callable. activation function use \"top\" layer. Ignored unless include_top=TRUE. Set classifier_activation=NULL return logits \"top\" layer. Defaults \"softmax\". loading pretrained weights, classifier_activation can NULL \"softmax\". include_preprocessing Boolean, whether include preprocessing layer bottom network. name name model (string).","code":""},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_v2l.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Instantiates the EfficientNetV2L architecture. — application_efficientnet_v2l","text":"model instance.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_v2l.html","id":"reference","dir":"Reference","previous_headings":"","what":"Reference","title":"Instantiates the EfficientNetV2L architecture. — application_efficientnet_v2l","text":"EfficientNetV2: Smaller Models Faster Training (ICML 2021) function returns Keras image classification model, optionally loaded weights pre-trained ImageNet. image classification use cases, see page detailed examples. transfer learning use cases, make sure read guide transfer learning & fine-tuning.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_v2l.html","id":"note","dir":"Reference","previous_headings":"","what":"Note","title":"Instantiates the EfficientNetV2L architecture. — application_efficientnet_v2l","text":"Keras Application expects specific kind input preprocessing. EfficientNetV2, default input preprocessing included part model (Rescaling layer), thus application_preprocess_inputs() actually pass-function. use case, EfficientNetV2 models expect inputs float tensors pixels values [0, 255] range. time, preprocessing part model (.e. Rescaling layer) can disabled setting include_preprocessing argument FALSE. preprocessing disabled EfficientNetV2 models expect inputs float tensors pixels values [-1, 1] range.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_v2m.html","id":null,"dir":"Reference","previous_headings":"","what":"Instantiates the EfficientNetV2M architecture. — application_efficientnet_v2m","title":"Instantiates the EfficientNetV2M architecture. — application_efficientnet_v2m","text":"Instantiates EfficientNetV2M architecture.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_v2m.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Instantiates the EfficientNetV2M architecture. — application_efficientnet_v2m","text":"","code":"application_efficientnet_v2m(   include_top = TRUE,   weights = \"imagenet\",   input_tensor = NULL,   input_shape = NULL,   pooling = NULL,   classes = 1000L,   classifier_activation = \"softmax\",   include_preprocessing = TRUE,   name = \"efficientnetv2-m\" )"},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_v2m.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Instantiates the EfficientNetV2M architecture. — application_efficientnet_v2m","text":"include_top Boolean, whether include fully-connected layer top network. Defaults TRUE. weights One NULL (random initialization), \"imagenet\" (pre-training ImageNet), path weights file loaded. Defaults \"imagenet\". input_tensor Optional Keras tensor (.e. output keras_input()) use image input model. input_shape Optional shape tuple, specified include_top FALSE. exactly 3 inputs channels. pooling Optional pooling mode feature extraction include_top FALSE. Defaults NULL. NULL means output model 4D tensor output last convolutional layer. \"avg\" means global average pooling applied output last convolutional layer, thus output model 2D tensor. \"max\" means global max pooling applied. classes Optional number classes classify images , specified include_top TRUE, weights argument specified. Defaults 1000 (number ImageNet classes). classifier_activation string callable. activation function use \"top\" layer. Ignored unless include_top=TRUE. Set classifier_activation=NULL return logits \"top\" layer. Defaults \"softmax\". loading pretrained weights, classifier_activation can NULL \"softmax\". include_preprocessing Boolean, whether include preprocessing layer bottom network. name name model (string).","code":""},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_v2m.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Instantiates the EfficientNetV2M architecture. — application_efficientnet_v2m","text":"model instance.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_v2m.html","id":"reference","dir":"Reference","previous_headings":"","what":"Reference","title":"Instantiates the EfficientNetV2M architecture. — application_efficientnet_v2m","text":"EfficientNetV2: Smaller Models Faster Training (ICML 2021) function returns Keras image classification model, optionally loaded weights pre-trained ImageNet. image classification use cases, see page detailed examples. transfer learning use cases, make sure read guide transfer learning & fine-tuning.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_v2m.html","id":"note","dir":"Reference","previous_headings":"","what":"Note","title":"Instantiates the EfficientNetV2M architecture. — application_efficientnet_v2m","text":"Keras Application expects specific kind input preprocessing. EfficientNetV2, default input preprocessing included part model (Rescaling layer), thus application_preprocess_inputs() actually pass-function. use case, EfficientNetV2 models expect inputs float tensors pixels values [0, 255] range. time, preprocessing part model (.e. Rescaling layer) can disabled setting include_preprocessing argument FALSE. preprocessing disabled EfficientNetV2 models expect inputs float tensors pixels values [-1, 1] range.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_v2s.html","id":null,"dir":"Reference","previous_headings":"","what":"Instantiates the EfficientNetV2S architecture. — application_efficientnet_v2s","title":"Instantiates the EfficientNetV2S architecture. — application_efficientnet_v2s","text":"Instantiates EfficientNetV2S architecture.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_v2s.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Instantiates the EfficientNetV2S architecture. — application_efficientnet_v2s","text":"","code":"application_efficientnet_v2s(   include_top = TRUE,   weights = \"imagenet\",   input_tensor = NULL,   input_shape = NULL,   pooling = NULL,   classes = 1000L,   classifier_activation = \"softmax\",   include_preprocessing = TRUE,   name = \"efficientnetv2-s\" )"},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_v2s.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Instantiates the EfficientNetV2S architecture. — application_efficientnet_v2s","text":"include_top Boolean, whether include fully-connected layer top network. Defaults TRUE. weights One NULL (random initialization), \"imagenet\" (pre-training ImageNet), path weights file loaded. Defaults \"imagenet\". input_tensor Optional Keras tensor (.e. output keras_input()) use image input model. input_shape Optional shape tuple, specified include_top FALSE. exactly 3 inputs channels. pooling Optional pooling mode feature extraction include_top FALSE. Defaults NULL. NULL means output model 4D tensor output last convolutional layer. \"avg\" means global average pooling applied output last convolutional layer, thus output model 2D tensor. \"max\" means global max pooling applied. classes Optional number classes classify images , specified include_top TRUE, weights argument specified. Defaults 1000 (number ImageNet classes). classifier_activation string callable. activation function use \"top\" layer. Ignored unless include_top=TRUE. Set classifier_activation=NULL return logits \"top\" layer. Defaults \"softmax\". loading pretrained weights, classifier_activation can NULL \"softmax\". include_preprocessing Boolean, whether include preprocessing layer bottom network. name name model (string).","code":""},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_v2s.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Instantiates the EfficientNetV2S architecture. — application_efficientnet_v2s","text":"model instance.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_v2s.html","id":"reference","dir":"Reference","previous_headings":"","what":"Reference","title":"Instantiates the EfficientNetV2S architecture. — application_efficientnet_v2s","text":"EfficientNetV2: Smaller Models Faster Training (ICML 2021) function returns Keras image classification model, optionally loaded weights pre-trained ImageNet. image classification use cases, see page detailed examples. transfer learning use cases, make sure read guide transfer learning & fine-tuning.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_v2s.html","id":"note","dir":"Reference","previous_headings":"","what":"Note","title":"Instantiates the EfficientNetV2S architecture. — application_efficientnet_v2s","text":"Keras Application expects specific kind input preprocessing. EfficientNetV2, default input preprocessing included part model (Rescaling layer), thus application_preprocess_inputs() actually pass-function. use case, EfficientNetV2 models expect inputs float tensors pixels values [0, 255] range. time, preprocessing part model (.e. Rescaling layer) can disabled setting include_preprocessing argument FALSE. preprocessing disabled EfficientNetV2 models expect inputs float tensors pixels values [-1, 1] range.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/application_inception_resnet_v2.html","id":null,"dir":"Reference","previous_headings":"","what":"Instantiates the Inception-ResNet v2 architecture. — application_inception_resnet_v2","title":"Instantiates the Inception-ResNet v2 architecture. — application_inception_resnet_v2","text":"Instantiates Inception-ResNet v2 architecture.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_inception_resnet_v2.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Instantiates the Inception-ResNet v2 architecture. — application_inception_resnet_v2","text":"","code":"application_inception_resnet_v2(   include_top = TRUE,   weights = \"imagenet\",   input_tensor = NULL,   input_shape = NULL,   pooling = NULL,   classes = 1000L,   classifier_activation = \"softmax\",   name = \"inception_resnet_v2\" )"},{"path":"https://keras3.posit.co/dev/reference/application_inception_resnet_v2.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Instantiates the Inception-ResNet v2 architecture. — application_inception_resnet_v2","text":"include_top whether include fully-connected layer top network. weights one NULL (random initialization), \"imagenet\" (pre-training ImageNet), path weights file loaded. input_tensor optional Keras tensor (.e. output keras_input()) use image input model. input_shape optional shape tuple, specified include_top FALSE (otherwise input shape (299, 299, 3) ('channels_last' data format) (3, 299, 299) ('channels_first' data format). exactly 3 inputs channels, width height smaller 75. E.g. (150, 150, 3) one valid value. pooling Optional pooling mode feature extraction include_top FALSE. NULL means output model 4D tensor output last convolutional block. 'avg' means global average pooling applied output last convolutional block, thus output model 2D tensor. 'max' means global max pooling applied. classes optional number classes classify images , specified include_top TRUE, weights argument specified. classifier_activation str callable. activation function use \"top\" layer. Ignored unless include_top=TRUE. Set classifier_activation=NULL return logits \"top\" layer. loading pretrained weights, classifier_activation can NULL \"softmax\". name name model (string).","code":""},{"path":"https://keras3.posit.co/dev/reference/application_inception_resnet_v2.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Instantiates the Inception-ResNet v2 architecture. — application_inception_resnet_v2","text":"model instance.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_inception_resnet_v2.html","id":"reference","dir":"Reference","previous_headings":"","what":"Reference","title":"Instantiates the Inception-ResNet v2 architecture. — application_inception_resnet_v2","text":"Inception-v4, Inception-ResNet Impact Residual Connections Learning (AAAI 2017) function returns Keras image classification model, optionally loaded weights pre-trained ImageNet. image classification use cases, see page detailed examples. transfer learning use cases, make sure read guide transfer learning & fine-tuning.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_inception_resnet_v2.html","id":"note","dir":"Reference","previous_headings":"","what":"Note","title":"Instantiates the Inception-ResNet v2 architecture. — application_inception_resnet_v2","text":"Keras Application expects specific kind input preprocessing. InceptionResNetV2, call application_preprocess_inputs() inputs passing model. application_preprocess_inputs() scale input pixels -1 1.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/application_inception_v3.html","id":null,"dir":"Reference","previous_headings":"","what":"Instantiates the Inception v3 architecture. — application_inception_v3","title":"Instantiates the Inception v3 architecture. — application_inception_v3","text":"Instantiates Inception v3 architecture.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_inception_v3.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Instantiates the Inception v3 architecture. — application_inception_v3","text":"","code":"application_inception_v3(   include_top = TRUE,   weights = \"imagenet\",   input_tensor = NULL,   input_shape = NULL,   pooling = NULL,   classes = 1000L,   classifier_activation = \"softmax\",   name = \"inception_v3\" )"},{"path":"https://keras3.posit.co/dev/reference/application_inception_v3.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Instantiates the Inception v3 architecture. — application_inception_v3","text":"include_top Boolean, whether include fully-connected layer top, last layer network. Defaults TRUE. weights One NULL (random initialization), imagenet (pre-training ImageNet), path weights file loaded. Defaults \"imagenet\". input_tensor Optional Keras tensor (.e. output keras_input()) use image input model. input_tensor useful sharing inputs multiple different networks. Defaults NULL. input_shape Optional shape tuple, specified include_top FALSE (otherwise input shape (299, 299, 3) (channels_last data format) (3, 299, 299) (channels_first data format). exactly 3 inputs channels, width height smaller 75. E.g. (150, 150, 3) one valid value. input_shape ignored input_tensor provided. pooling Optional pooling mode feature extraction include_top FALSE. NULL (default) means output model 4D tensor output last convolutional block. avg means global average pooling applied output last convolutional block, thus output model 2D tensor. max means global max pooling applied. classes optional number classes classify images , specified include_top TRUE, weights argument specified. Defaults 1000. classifier_activation str callable. activation function use \"top\" layer. Ignored unless include_top=TRUE. Set classifier_activation=NULL return logits \"top\" layer. loading pretrained weights, classifier_activation can NULL \"softmax\". name name model (string).","code":""},{"path":"https://keras3.posit.co/dev/reference/application_inception_v3.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Instantiates the Inception v3 architecture. — application_inception_v3","text":"model instance.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_inception_v3.html","id":"reference","dir":"Reference","previous_headings":"","what":"Reference","title":"Instantiates the Inception v3 architecture. — application_inception_v3","text":"Rethinking Inception Architecture Computer Vision (CVPR 2016) function returns Keras image classification model, optionally loaded weights pre-trained ImageNet. image classification use cases, see page detailed examples. transfer learning use cases, make sure read guide transfer learning & fine-tuning.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_inception_v3.html","id":"note","dir":"Reference","previous_headings":"","what":"Note","title":"Instantiates the Inception v3 architecture. — application_inception_v3","text":"Keras Application expects specific kind input preprocessing. InceptionV3, call application_preprocess_inputs() inputs passing model. application_preprocess_inputs() scale input pixels -1 1.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/application_mobilenet.html","id":null,"dir":"Reference","previous_headings":"","what":"Instantiates the MobileNet architecture. — application_mobilenet","title":"Instantiates the MobileNet architecture. — application_mobilenet","text":"Instantiates MobileNet architecture.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_mobilenet.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Instantiates the MobileNet architecture. — application_mobilenet","text":"","code":"application_mobilenet(   input_shape = NULL,   alpha = 1,   depth_multiplier = 1L,   dropout = 0.001,   include_top = TRUE,   weights = \"imagenet\",   input_tensor = NULL,   pooling = NULL,   classes = 1000L,   classifier_activation = \"softmax\",   name = NULL )"},{"path":"https://keras3.posit.co/dev/reference/application_mobilenet.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Instantiates the MobileNet architecture. — application_mobilenet","text":"input_shape Optional shape tuple, specified include_top FALSE (otherwise input shape (224, 224, 3) (\"channels_last\" data format) (3, 224, 224) (\"channels_first\" data format). exactly 3 inputs channels, width height smaller 32. E.g. (200, 200, 3) one valid value. Defaults NULL. input_shape ignored input_tensor provided. alpha Controls width network. known width multiplier MobileNet paper. alpha < 1.0, proportionally decreases number filters layer. alpha > 1.0, proportionally increases number filters layer. alpha == 1, default number filters paper used layer. Defaults 1.0. depth_multiplier Depth multiplier depthwise convolution. called resolution multiplier MobileNet paper. Defaults 1.0. dropout Dropout rate. Defaults 0.001. include_top Boolean, whether include fully-connected layer top network. Defaults TRUE. weights One NULL (random initialization), \"imagenet\" (pre-training ImageNet), path weights file loaded. Defaults \"imagenet\". input_tensor Optional Keras tensor (.e. output keras_input()) use image input model. input_tensor useful sharing inputs multiple different networks. Defaults NULL. pooling Optional pooling mode feature extraction include_top FALSE. NULL (default) means output model 4D tensor output last convolutional block. avg means global average pooling applied output last convolutional block, thus output model 2D tensor. max means global max pooling applied. classes Optional number classes classify images , specified include_top TRUE, weights argument specified. Defaults 1000. classifier_activation str callable. activation function use \"top\" layer. Ignored unless include_top=TRUE. Set classifier_activation=NULL return logits \"top\" layer. loading pretrained weights, classifier_activation can NULL \"softmax\". name name model (string).","code":""},{"path":"https://keras3.posit.co/dev/reference/application_mobilenet.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Instantiates the MobileNet architecture. — application_mobilenet","text":"model instance.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_mobilenet.html","id":"reference","dir":"Reference","previous_headings":"","what":"Reference","title":"Instantiates the MobileNet architecture. — application_mobilenet","text":"MobileNets: Efficient Convolutional Neural Networks Mobile Vision Applications function returns Keras image classification model, optionally loaded weights pre-trained ImageNet. image classification use cases, see page detailed examples. transfer learning use cases, make sure read guide transfer learning & fine-tuning.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_mobilenet.html","id":"note","dir":"Reference","previous_headings":"","what":"Note","title":"Instantiates the MobileNet architecture. — application_mobilenet","text":"Keras Application expects specific kind input preprocessing. MobileNet, call application_preprocess_inputs() inputs passing model. application_preprocess_inputs() scale input pixels -1 1.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/application_mobilenet_v2.html","id":null,"dir":"Reference","previous_headings":"","what":"Instantiates the MobileNetV2 architecture. — application_mobilenet_v2","title":"Instantiates the MobileNetV2 architecture. — application_mobilenet_v2","text":"MobileNetV2 similar original MobileNet, except uses inverted residual blocks bottlenecking features. drastically lower parameter count original MobileNet. MobileNets support input size greater 32 x 32, larger image sizes offering better performance.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_mobilenet_v2.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Instantiates the MobileNetV2 architecture. — application_mobilenet_v2","text":"","code":"application_mobilenet_v2(   input_shape = NULL,   alpha = 1,   include_top = TRUE,   weights = \"imagenet\",   input_tensor = NULL,   pooling = NULL,   classes = 1000L,   classifier_activation = \"softmax\",   name = NULL )"},{"path":"https://keras3.posit.co/dev/reference/application_mobilenet_v2.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Instantiates the MobileNetV2 architecture. — application_mobilenet_v2","text":"input_shape Optional shape tuple, specified include_top FALSE (otherwise input shape (224, 224, 3) (\"channels_last\" data format) (3, 224, 224) (\"channels_first\" data format). exactly 3 inputs channels, width height smaller 32. E.g. (200, 200, 3) one valid value. Defaults NULL. input_shape ignored input_tensor provided. alpha Controls width network. known width multiplier MobileNet paper. alpha < 1.0, proportionally decreases number filters layer. alpha > 1.0, proportionally increases number filters layer. alpha == 1, default number filters paper used layer. Defaults 1.0. include_top Boolean, whether include fully-connected layer top network. Defaults TRUE. weights One NULL (random initialization), \"imagenet\" (pre-training ImageNet), path weights file loaded. Defaults \"imagenet\". input_tensor Optional Keras tensor (.e. output keras_input()) use image input model. input_tensor useful sharing inputs multiple different networks. Defaults NULL. pooling Optional pooling mode feature extraction include_top FALSE. NULL (default) means output model 4D tensor output last convolutional block. avg means global average pooling applied output last convolutional block, thus output model 2D tensor. max means global max pooling applied. classes Optional number classes classify images , specified include_top TRUE, weights argument specified. Defaults 1000. classifier_activation str callable. activation function use \"top\" layer. Ignored unless include_top=TRUE. Set classifier_activation=NULL return logits \"top\" layer. loading pretrained weights, classifier_activation can NULL \"softmax\". name name model (string).","code":""},{"path":"https://keras3.posit.co/dev/reference/application_mobilenet_v2.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Instantiates the MobileNetV2 architecture. — application_mobilenet_v2","text":"model instance.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_mobilenet_v2.html","id":"reference","dir":"Reference","previous_headings":"","what":"Reference","title":"Instantiates the MobileNetV2 architecture. — application_mobilenet_v2","text":"MobileNetV2: Inverted Residuals Linear Bottlenecks (CVPR 2018) function returns Keras image classification model, optionally loaded weights pre-trained ImageNet. image classification use cases, see page detailed examples. transfer learning use cases, make sure read guide transfer learning & fine-tuning.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_mobilenet_v2.html","id":"note","dir":"Reference","previous_headings":"","what":"Note","title":"Instantiates the MobileNetV2 architecture. — application_mobilenet_v2","text":"Keras Application expects specific kind input preprocessing. MobileNetV2, call application_preprocess_inputs() inputs passing model. application_preprocess_inputs() scale input pixels -1 1.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/application_mobilenet_v3_large.html","id":null,"dir":"Reference","previous_headings":"","what":"Instantiates the MobileNetV3Large architecture. — application_mobilenet_v3_large","title":"Instantiates the MobileNetV3Large architecture. — application_mobilenet_v3_large","text":"Instantiates MobileNetV3Large architecture.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_mobilenet_v3_large.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Instantiates the MobileNetV3Large architecture. — application_mobilenet_v3_large","text":"","code":"application_mobilenet_v3_large(   input_shape = NULL,   alpha = 1,   minimalistic = FALSE,   include_top = TRUE,   weights = \"imagenet\",   input_tensor = NULL,   classes = 1000L,   pooling = NULL,   dropout_rate = 0.2,   classifier_activation = \"softmax\",   include_preprocessing = TRUE,   name = \"MobileNetV3Large\" )"},{"path":"https://keras3.posit.co/dev/reference/application_mobilenet_v3_large.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Instantiates the MobileNetV3Large architecture. — application_mobilenet_v3_large","text":"input_shape Optional shape tuple, specified like use model input image resolution (224, 224, 3). exactly 3 inputs channels. can also omit option like infer input_shape input_tensor. choose include input_tensor input_shape input_shape used match, shapes match throw error. E.g. (160, 160, 3) one valid value. alpha controls width network. known depth multiplier MobileNetV3 paper, name kept consistency MobileNetV1 Keras. alpha < 1.0, proportionally decreases number filters layer. alpha > 1.0, proportionally increases number filters layer. alpha == 1, default number filters paper used layer. minimalistic addition large small models module also contains -called minimalistic models, models per-layer dimensions characteristic MobilenetV3 however, utilize advanced blocks (squeeze--excite units, hard-swish, 5x5 convolutions). models less efficient CPU, much performant GPU/DSP. include_top Boolean, whether include fully-connected layer top network. Defaults TRUE. weights String, one NULL (random initialization), \"imagenet\" (pre-training ImageNet), path weights file loaded. input_tensor Optional Keras tensor (.e. output keras_input()) use image input model. classes Integer, optional number classes classify images , specified include_top TRUE, weights argument specified. pooling String, optional pooling mode feature extraction include_top FALSE. NULL means output model 4D tensor output last convolutional block. avg means global average pooling applied output last convolutional block, thus output model 2D tensor. max means global max pooling applied. dropout_rate fraction input units drop last layer. classifier_activation str callable. activation function use \"top\" layer. Ignored unless include_top=TRUE. Set classifier_activation=NULL return logits \"top\" layer. loading pretrained weights, classifier_activation can NULL \"softmax\". include_preprocessing Boolean, whether include preprocessing layer (Rescaling) bottom network. Defaults TRUE. name name model (string).","code":""},{"path":"https://keras3.posit.co/dev/reference/application_mobilenet_v3_large.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Instantiates the MobileNetV3Large architecture. — application_mobilenet_v3_large","text":"model instance.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_mobilenet_v3_large.html","id":"reference","dir":"Reference","previous_headings":"","what":"Reference","title":"Instantiates the MobileNetV3Large architecture. — application_mobilenet_v3_large","text":"Searching MobileNetV3 (ICCV 2019)","code":""},{"path":"https://keras3.posit.co/dev/reference/application_mobilenet_v3_large.html","id":"the-following-table-describes-the-performance-of-mobilenets-v-","dir":"Reference","previous_headings":"","what":"The following table describes the performance of MobileNets v3:","title":"Instantiates the MobileNetV3Large architecture. — application_mobilenet_v3_large","text":"MACs stands Multiply Adds image classification use cases, see page detailed examples. transfer learning use cases, make sure read guide transfer learning & fine-tuning.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_mobilenet_v3_large.html","id":"note","dir":"Reference","previous_headings":"","what":"Note","title":"Instantiates the MobileNetV3Large architecture. — application_mobilenet_v3_large","text":"Keras Application expects specific kind input preprocessing. MobileNetV3, default input preprocessing included part model (Rescaling layer), thus application_preprocess_inputs() actually pass-function. use case, MobileNetV3 models expect inputs float tensors pixels values [0-255] range. time, preprocessing part model (.e. Rescaling layer) can disabled setting include_preprocessing argument FALSE. preprocessing disabled MobileNetV3 models expect inputs float tensors pixels values [-1, 1] range.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_mobilenet_v3_large.html","id":"call-arguments","dir":"Reference","previous_headings":"","what":"Call Arguments","title":"Instantiates the MobileNetV3Large architecture. — application_mobilenet_v3_large","text":"inputs: floating point numpy.array backend-native tensor, 4D 3 color channels, values range [0, 255] include_preprocessing TRUE range [-1, 1] otherwise.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/application_mobilenet_v3_small.html","id":null,"dir":"Reference","previous_headings":"","what":"Instantiates the MobileNetV3Small architecture. — application_mobilenet_v3_small","title":"Instantiates the MobileNetV3Small architecture. — application_mobilenet_v3_small","text":"Instantiates MobileNetV3Small architecture.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_mobilenet_v3_small.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Instantiates the MobileNetV3Small architecture. — application_mobilenet_v3_small","text":"","code":"application_mobilenet_v3_small(   input_shape = NULL,   alpha = 1,   minimalistic = FALSE,   include_top = TRUE,   weights = \"imagenet\",   input_tensor = NULL,   classes = 1000L,   pooling = NULL,   dropout_rate = 0.2,   classifier_activation = \"softmax\",   include_preprocessing = TRUE,   name = \"MobileNetV3Small\" )"},{"path":"https://keras3.posit.co/dev/reference/application_mobilenet_v3_small.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Instantiates the MobileNetV3Small architecture. — application_mobilenet_v3_small","text":"input_shape Optional shape tuple, specified like use model input image resolution (224, 224, 3). exactly 3 inputs channels. can also omit option like infer input_shape input_tensor. choose include input_tensor input_shape input_shape used match, shapes match throw error. E.g. (160, 160, 3) one valid value. alpha controls width network. known depth multiplier MobileNetV3 paper, name kept consistency MobileNetV1 Keras. alpha < 1.0, proportionally decreases number filters layer. alpha > 1.0, proportionally increases number filters layer. alpha == 1, default number filters paper used layer. minimalistic addition large small models module also contains -called minimalistic models, models per-layer dimensions characteristic MobilenetV3 however, utilize advanced blocks (squeeze--excite units, hard-swish, 5x5 convolutions). models less efficient CPU, much performant GPU/DSP. include_top Boolean, whether include fully-connected layer top network. Defaults TRUE. weights String, one NULL (random initialization), \"imagenet\" (pre-training ImageNet), path weights file loaded. input_tensor Optional Keras tensor (.e. output keras_input()) use image input model. classes Integer, optional number classes classify images , specified include_top TRUE, weights argument specified. pooling String, optional pooling mode feature extraction include_top FALSE. NULL means output model 4D tensor output last convolutional block. avg means global average pooling applied output last convolutional block, thus output model 2D tensor. max means global max pooling applied. dropout_rate fraction input units drop last layer. classifier_activation str callable. activation function use \"top\" layer. Ignored unless include_top=TRUE. Set classifier_activation=NULL return logits \"top\" layer. loading pretrained weights, classifier_activation can NULL \"softmax\". include_preprocessing Boolean, whether include preprocessing layer (Rescaling) bottom network. Defaults TRUE. name name model (string).","code":""},{"path":"https://keras3.posit.co/dev/reference/application_mobilenet_v3_small.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Instantiates the MobileNetV3Small architecture. — application_mobilenet_v3_small","text":"model instance.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_mobilenet_v3_small.html","id":"reference","dir":"Reference","previous_headings":"","what":"Reference","title":"Instantiates the MobileNetV3Small architecture. — application_mobilenet_v3_small","text":"Searching MobileNetV3 (ICCV 2019)","code":""},{"path":"https://keras3.posit.co/dev/reference/application_mobilenet_v3_small.html","id":"the-following-table-describes-the-performance-of-mobilenets-v-","dir":"Reference","previous_headings":"","what":"The following table describes the performance of MobileNets v3:","title":"Instantiates the MobileNetV3Small architecture. — application_mobilenet_v3_small","text":"MACs stands Multiply Adds image classification use cases, see page detailed examples. transfer learning use cases, make sure read guide transfer learning & fine-tuning.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_mobilenet_v3_small.html","id":"note","dir":"Reference","previous_headings":"","what":"Note","title":"Instantiates the MobileNetV3Small architecture. — application_mobilenet_v3_small","text":"Keras Application expects specific kind input preprocessing. MobileNetV3, default input preprocessing included part model (Rescaling layer), thus application_preprocess_inputs() actually pass-function. use case, MobileNetV3 models expect inputs float tensors pixels values [0-255] range. time, preprocessing part model (.e. Rescaling layer) can disabled setting include_preprocessing argument FALSE. preprocessing disabled MobileNetV3 models expect inputs float tensors pixels values [-1, 1] range.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_mobilenet_v3_small.html","id":"call-arguments","dir":"Reference","previous_headings":"","what":"Call Arguments","title":"Instantiates the MobileNetV3Small architecture. — application_mobilenet_v3_small","text":"inputs: floating point numpy.array backend-native tensor, 4D 3 color channels, values range [0, 255] include_preprocessing TRUE range [-1, 1] otherwise.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/application_nasnet_large.html","id":null,"dir":"Reference","previous_headings":"","what":"Instantiates a NASNet model in ImageNet mode. — application_nasnet_large","title":"Instantiates a NASNet model in ImageNet mode. — application_nasnet_large","text":"Instantiates NASNet model ImageNet mode.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_nasnet_large.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Instantiates a NASNet model in ImageNet mode. — application_nasnet_large","text":"","code":"application_nasnet_large(   input_shape = NULL,   include_top = TRUE,   weights = \"imagenet\",   input_tensor = NULL,   pooling = NULL,   classes = 1000L,   classifier_activation = \"softmax\",   name = \"nasnet_large\" )"},{"path":"https://keras3.posit.co/dev/reference/application_nasnet_large.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Instantiates a NASNet model in ImageNet mode. — application_nasnet_large","text":"input_shape Optional shape tuple, specified include_top FALSE (otherwise input shape (331, 331, 3) NASNetLarge. exactly 3 inputs channels, width height smaller 32. E.g. (224, 224, 3) one valid value. include_top Whether include fully-connected layer top network. weights NULL (random initialization) imagenet (ImageNet weights).  loading imagenet weights, input_shape (331, 331, 3) input_tensor Optional Keras tensor (.e. output keras_input()) use image input model. pooling Optional pooling mode feature extraction include_top FALSE. NULL means output model 4D tensor output last convolutional layer. avg means global average pooling applied output last convolutional layer, thus output model 2D tensor. max means global max pooling applied. classes Optional number classes classify images , specified include_top TRUE, weights argument specified. classifier_activation str callable. activation function use \"top\" layer. Ignored unless include_top=TRUE. Set classifier_activation=NULL return logits \"top\" layer.  loading pretrained weights, classifier_activation can NULL \"softmax\". name name model (string).","code":""},{"path":"https://keras3.posit.co/dev/reference/application_nasnet_large.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Instantiates a NASNet model in ImageNet mode. — application_nasnet_large","text":"Keras model instance.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_nasnet_large.html","id":"reference","dir":"Reference","previous_headings":"","what":"Reference","title":"Instantiates a NASNet model in ImageNet mode. — application_nasnet_large","text":"Learning Transferable Architectures Scalable Image Recognition (CVPR 2018) Optionally loads weights pre-trained ImageNet. Note data format convention used model one specified Keras config ~/.keras/keras.json.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_nasnet_large.html","id":"note","dir":"Reference","previous_headings":"","what":"Note","title":"Instantiates a NASNet model in ImageNet mode. — application_nasnet_large","text":"Keras Application expects specific kind input preprocessing. NASNet, call application_preprocess_inputs() inputs passing model.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/application_nasnet_mobile.html","id":null,"dir":"Reference","previous_headings":"","what":"Instantiates a Mobile NASNet model in ImageNet mode. — application_nasnet_mobile","title":"Instantiates a Mobile NASNet model in ImageNet mode. — application_nasnet_mobile","text":"Instantiates Mobile NASNet model ImageNet mode.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_nasnet_mobile.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Instantiates a Mobile NASNet model in ImageNet mode. — application_nasnet_mobile","text":"","code":"application_nasnet_mobile(   input_shape = NULL,   include_top = TRUE,   weights = \"imagenet\",   input_tensor = NULL,   pooling = NULL,   classes = 1000L,   classifier_activation = \"softmax\",   name = \"nasnet_mobile\" )"},{"path":"https://keras3.posit.co/dev/reference/application_nasnet_mobile.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Instantiates a Mobile NASNet model in ImageNet mode. — application_nasnet_mobile","text":"input_shape Optional shape tuple, specified include_top FALSE (otherwise input shape (224, 224, 3) NASNetMobile exactly 3 inputs channels, width height smaller 32. E.g. (224, 224, 3) one valid value. include_top Whether include fully-connected layer top network. weights NULL (random initialization) imagenet (ImageNet weights). loading imagenet weights, input_shape (224, 224, 3) input_tensor Optional Keras tensor (.e. output keras_input()) use image input model. pooling Optional pooling mode feature extraction include_top FALSE. NULL means output model 4D tensor output last convolutional layer. avg means global average pooling applied output last convolutional layer, thus output model 2D tensor. max means global max pooling applied. classes Optional number classes classify images , specified include_top TRUE, weights argument specified. classifier_activation str callable. activation function use \"top\" layer. Ignored unless include_top=TRUE. Set classifier_activation=NULL return logits \"top\" layer.  loading pretrained weights, classifier_activation can NULL \"softmax\". name name model (string).","code":""},{"path":"https://keras3.posit.co/dev/reference/application_nasnet_mobile.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Instantiates a Mobile NASNet model in ImageNet mode. — application_nasnet_mobile","text":"Keras model instance.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_nasnet_mobile.html","id":"reference","dir":"Reference","previous_headings":"","what":"Reference","title":"Instantiates a Mobile NASNet model in ImageNet mode. — application_nasnet_mobile","text":"Learning Transferable Architectures Scalable Image Recognition (CVPR 2018) Optionally loads weights pre-trained ImageNet. Note data format convention used model one specified Keras config ~/.keras/keras.json.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_nasnet_mobile.html","id":"note","dir":"Reference","previous_headings":"","what":"Note","title":"Instantiates a Mobile NASNet model in ImageNet mode. — application_nasnet_mobile","text":"Keras Application expects specific kind input preprocessing. NASNet, call application_preprocess_inputs() inputs passing model.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/application_resnet101.html","id":null,"dir":"Reference","previous_headings":"","what":"Instantiates the ResNet101 architecture. — application_resnet101","title":"Instantiates the ResNet101 architecture. — application_resnet101","text":"Instantiates ResNet101 architecture.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_resnet101.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Instantiates the ResNet101 architecture. — application_resnet101","text":"","code":"application_resnet101(   include_top = TRUE,   weights = \"imagenet\",   input_tensor = NULL,   input_shape = NULL,   pooling = NULL,   classes = 1000L,   classifier_activation = \"softmax\",   name = \"resnet101\" )"},{"path":"https://keras3.posit.co/dev/reference/application_resnet101.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Instantiates the ResNet101 architecture. — application_resnet101","text":"include_top whether include fully-connected layer top network. weights one NULL (random initialization), \"imagenet\" (pre-training ImageNet), path weights file loaded. input_tensor optional Keras tensor (.e. output keras_input()) use image input model. input_shape optional shape tuple, specified include_top FALSE (otherwise input shape (224, 224, 3) (\"channels_last\" data format) (3, 224, 224) (\"channels_first\" data format). exactly 3 inputs channels, width height smaller 32. E.g. (200, 200, 3) one valid value. pooling Optional pooling mode feature extraction include_top FALSE. NULL means output model 4D tensor output last convolutional block. avg means global average pooling applied output last convolutional block, thus output model 2D tensor. max means global max pooling applied. classes optional number classes classify images , specified include_top TRUE, weights argument specified. classifier_activation str callable. activation function use \"top\" layer. Ignored unless include_top=TRUE. Set classifier_activation=NULL return logits \"top\" layer. loading pretrained weights, classifier_activation can NULL \"softmax\". name name model (string).","code":""},{"path":"https://keras3.posit.co/dev/reference/application_resnet101.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Instantiates the ResNet101 architecture. — application_resnet101","text":"Model instance.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_resnet101.html","id":"reference","dir":"Reference","previous_headings":"","what":"Reference","title":"Instantiates the ResNet101 architecture. — application_resnet101","text":"Deep Residual Learning Image Recognition (CVPR 2015) image classification use cases, see page detailed examples. transfer learning use cases, make sure read guide transfer learning & fine-tuning.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_resnet101.html","id":"note","dir":"Reference","previous_headings":"","what":"Note","title":"Instantiates the ResNet101 architecture. — application_resnet101","text":"Keras Application expects specific kind input preprocessing. ResNet, call application_preprocess_inputs() inputs passing model. application_preprocess_inputs() convert input images RGB BGR, zero-center color channel respect ImageNet dataset, without scaling.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/application_resnet101_v2.html","id":null,"dir":"Reference","previous_headings":"","what":"Instantiates the ResNet101V2 architecture. — application_resnet101_v2","title":"Instantiates the ResNet101V2 architecture. — application_resnet101_v2","text":"Instantiates ResNet101V2 architecture.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_resnet101_v2.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Instantiates the ResNet101V2 architecture. — application_resnet101_v2","text":"","code":"application_resnet101_v2(   include_top = TRUE,   weights = \"imagenet\",   input_tensor = NULL,   input_shape = NULL,   pooling = NULL,   classes = 1000L,   classifier_activation = \"softmax\",   name = \"resnet101v2\" )"},{"path":"https://keras3.posit.co/dev/reference/application_resnet101_v2.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Instantiates the ResNet101V2 architecture. — application_resnet101_v2","text":"include_top whether include fully-connected layer top network. weights one NULL (random initialization), \"imagenet\" (pre-training ImageNet), path weights file loaded. input_tensor optional Keras tensor (.e. output keras_input()) use image input model. input_shape optional shape tuple, specified include_top FALSE (otherwise input shape (224, 224, 3) (\"channels_last\" data format) (3, 224, 224) (\"channels_first\" data format). exactly 3 inputs channels, width height smaller 32. E.g. (200, 200, 3) one valid value. pooling Optional pooling mode feature extraction include_top FALSE. NULL means output model 4D tensor output last convolutional block. avg means global average pooling applied output last convolutional block, thus output model 2D tensor. max means global max pooling applied. classes optional number classes classify images , specified include_top TRUE, weights argument specified. classifier_activation str callable. activation function use \"top\" layer. Ignored unless include_top=TRUE. Set classifier_activation=NULL return logits \"top\" layer. loading pretrained weights, classifier_activation can NULL \"softmax\". name name model (string).","code":""},{"path":"https://keras3.posit.co/dev/reference/application_resnet101_v2.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Instantiates the ResNet101V2 architecture. — application_resnet101_v2","text":"Model instance.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_resnet101_v2.html","id":"reference","dir":"Reference","previous_headings":"","what":"Reference","title":"Instantiates the ResNet101V2 architecture. — application_resnet101_v2","text":"Identity Mappings Deep Residual Networks (CVPR 2016) image classification use cases, see page detailed examples. transfer learning use cases, make sure read guide transfer learning & fine-tuning.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_resnet101_v2.html","id":"note","dir":"Reference","previous_headings":"","what":"Note","title":"Instantiates the ResNet101V2 architecture. — application_resnet101_v2","text":"Keras Application expects specific kind input preprocessing. ResNet, call application_preprocess_inputs() inputs passing model. application_preprocess_inputs() scale input pixels -1 1.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/application_resnet152.html","id":null,"dir":"Reference","previous_headings":"","what":"Instantiates the ResNet152 architecture. — application_resnet152","title":"Instantiates the ResNet152 architecture. — application_resnet152","text":"Instantiates ResNet152 architecture.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_resnet152.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Instantiates the ResNet152 architecture. — application_resnet152","text":"","code":"application_resnet152(   include_top = TRUE,   weights = \"imagenet\",   input_tensor = NULL,   input_shape = NULL,   pooling = NULL,   classes = 1000L,   classifier_activation = \"softmax\",   name = \"resnet152\" )"},{"path":"https://keras3.posit.co/dev/reference/application_resnet152.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Instantiates the ResNet152 architecture. — application_resnet152","text":"include_top whether include fully-connected layer top network. weights one NULL (random initialization), \"imagenet\" (pre-training ImageNet), path weights file loaded. input_tensor optional Keras tensor (.e. output keras_input()) use image input model. input_shape optional shape tuple, specified include_top FALSE (otherwise input shape (224, 224, 3) (\"channels_last\" data format) (3, 224, 224) (\"channels_first\" data format). exactly 3 inputs channels, width height smaller 32. E.g. (200, 200, 3) one valid value. pooling Optional pooling mode feature extraction include_top FALSE. NULL means output model 4D tensor output last convolutional block. avg means global average pooling applied output last convolutional block, thus output model 2D tensor. max means global max pooling applied. classes optional number classes classify images , specified include_top TRUE, weights argument specified. classifier_activation str callable. activation function use \"top\" layer. Ignored unless include_top=TRUE. Set classifier_activation=NULL return logits \"top\" layer. loading pretrained weights, classifier_activation can NULL \"softmax\". name name model (string).","code":""},{"path":"https://keras3.posit.co/dev/reference/application_resnet152.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Instantiates the ResNet152 architecture. — application_resnet152","text":"Model instance.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_resnet152.html","id":"reference","dir":"Reference","previous_headings":"","what":"Reference","title":"Instantiates the ResNet152 architecture. — application_resnet152","text":"Deep Residual Learning Image Recognition (CVPR 2015) image classification use cases, see page detailed examples. transfer learning use cases, make sure read guide transfer learning & fine-tuning.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_resnet152.html","id":"note","dir":"Reference","previous_headings":"","what":"Note","title":"Instantiates the ResNet152 architecture. — application_resnet152","text":"Keras Application expects specific kind input preprocessing. ResNet, call application_preprocess_inputs() inputs passing model. application_preprocess_inputs() convert input images RGB BGR, zero-center color channel respect ImageNet dataset, without scaling.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/application_resnet152_v2.html","id":null,"dir":"Reference","previous_headings":"","what":"Instantiates the ResNet152V2 architecture. — application_resnet152_v2","title":"Instantiates the ResNet152V2 architecture. — application_resnet152_v2","text":"Instantiates ResNet152V2 architecture.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_resnet152_v2.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Instantiates the ResNet152V2 architecture. — application_resnet152_v2","text":"","code":"application_resnet152_v2(   include_top = TRUE,   weights = \"imagenet\",   input_tensor = NULL,   input_shape = NULL,   pooling = NULL,   classes = 1000L,   classifier_activation = \"softmax\",   name = \"resnet152v2\" )"},{"path":"https://keras3.posit.co/dev/reference/application_resnet152_v2.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Instantiates the ResNet152V2 architecture. — application_resnet152_v2","text":"include_top whether include fully-connected layer top network. weights one NULL (random initialization), \"imagenet\" (pre-training ImageNet), path weights file loaded. input_tensor optional Keras tensor (.e. output keras_input()) use image input model. input_shape optional shape tuple, specified include_top FALSE (otherwise input shape (224, 224, 3) (\"channels_last\" data format) (3, 224, 224) (\"channels_first\" data format). exactly 3 inputs channels, width height smaller 32. E.g. (200, 200, 3) one valid value. pooling Optional pooling mode feature extraction include_top FALSE. NULL means output model 4D tensor output last convolutional block. avg means global average pooling applied output last convolutional block, thus output model 2D tensor. max means global max pooling applied. classes optional number classes classify images , specified include_top TRUE, weights argument specified. classifier_activation str callable. activation function use \"top\" layer. Ignored unless include_top=TRUE. Set classifier_activation=NULL return logits \"top\" layer. loading pretrained weights, classifier_activation can NULL \"softmax\". name name model (string).","code":""},{"path":"https://keras3.posit.co/dev/reference/application_resnet152_v2.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Instantiates the ResNet152V2 architecture. — application_resnet152_v2","text":"Model instance.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_resnet152_v2.html","id":"reference","dir":"Reference","previous_headings":"","what":"Reference","title":"Instantiates the ResNet152V2 architecture. — application_resnet152_v2","text":"Identity Mappings Deep Residual Networks (CVPR 2016) image classification use cases, see page detailed examples. transfer learning use cases, make sure read guide transfer learning & fine-tuning.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_resnet152_v2.html","id":"note","dir":"Reference","previous_headings":"","what":"Note","title":"Instantiates the ResNet152V2 architecture. — application_resnet152_v2","text":"Keras Application expects specific kind input preprocessing. ResNet, call application_preprocess_inputs() inputs passing model. application_preprocess_inputs() scale input pixels -1 1.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/application_resnet50.html","id":null,"dir":"Reference","previous_headings":"","what":"Instantiates the ResNet50 architecture. — application_resnet50","title":"Instantiates the ResNet50 architecture. — application_resnet50","text":"Instantiates ResNet50 architecture.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_resnet50.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Instantiates the ResNet50 architecture. — application_resnet50","text":"","code":"application_resnet50(   include_top = TRUE,   weights = \"imagenet\",   input_tensor = NULL,   input_shape = NULL,   pooling = NULL,   classes = 1000L,   classifier_activation = \"softmax\",   name = \"resnet50\" )"},{"path":"https://keras3.posit.co/dev/reference/application_resnet50.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Instantiates the ResNet50 architecture. — application_resnet50","text":"include_top whether include fully-connected layer top network. weights one NULL (random initialization), \"imagenet\" (pre-training ImageNet), path weights file loaded. input_tensor optional Keras tensor (.e. output keras_input()) use image input model. input_shape optional shape tuple, specified include_top FALSE (otherwise input shape (224, 224, 3) (\"channels_last\" data format) (3, 224, 224) (\"channels_first\" data format). exactly 3 inputs channels, width height smaller 32. E.g. (200, 200, 3) one valid value. pooling Optional pooling mode feature extraction include_top FALSE. NULL means output model 4D tensor output last convolutional block. avg means global average pooling applied output last convolutional block, thus output model 2D tensor. max means global max pooling applied. classes optional number classes classify images , specified include_top TRUE, weights argument specified. Defaults 1000. classifier_activation str callable. activation function use \"top\" layer. Ignored unless include_top=TRUE. Set classifier_activation=NULL return logits \"top\" layer. loading pretrained weights, classifier_activation can NULL \"softmax\". name name model (string).","code":""},{"path":"https://keras3.posit.co/dev/reference/application_resnet50.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Instantiates the ResNet50 architecture. — application_resnet50","text":"Model instance.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_resnet50.html","id":"reference","dir":"Reference","previous_headings":"","what":"Reference","title":"Instantiates the ResNet50 architecture. — application_resnet50","text":"Deep Residual Learning Image Recognition (CVPR 2015) image classification use cases, see page detailed examples. transfer learning use cases, make sure read guide transfer learning & fine-tuning.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_resnet50.html","id":"note","dir":"Reference","previous_headings":"","what":"Note","title":"Instantiates the ResNet50 architecture. — application_resnet50","text":"Keras Application expects specific kind input preprocessing. ResNet, call application_preprocess_inputs() inputs passing model. application_preprocess_inputs() convert input images RGB BGR, zero-center color channel respect ImageNet dataset, without scaling.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/application_resnet50_v2.html","id":null,"dir":"Reference","previous_headings":"","what":"Instantiates the ResNet50V2 architecture. — application_resnet50_v2","title":"Instantiates the ResNet50V2 architecture. — application_resnet50_v2","text":"Instantiates ResNet50V2 architecture.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_resnet50_v2.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Instantiates the ResNet50V2 architecture. — application_resnet50_v2","text":"","code":"application_resnet50_v2(   include_top = TRUE,   weights = \"imagenet\",   input_tensor = NULL,   input_shape = NULL,   pooling = NULL,   classes = 1000L,   classifier_activation = \"softmax\",   name = \"resnet50v2\" )"},{"path":"https://keras3.posit.co/dev/reference/application_resnet50_v2.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Instantiates the ResNet50V2 architecture. — application_resnet50_v2","text":"include_top whether include fully-connected layer top network. weights one NULL (random initialization), \"imagenet\" (pre-training ImageNet), path weights file loaded. input_tensor optional Keras tensor (.e. output keras_input()) use image input model. input_shape optional shape tuple, specified include_top FALSE (otherwise input shape (224, 224, 3) (\"channels_last\" data format) (3, 224, 224) (\"channels_first\" data format). exactly 3 inputs channels, width height smaller 32. E.g. (200, 200, 3) one valid value. pooling Optional pooling mode feature extraction include_top FALSE. NULL means output model 4D tensor output last convolutional block. avg means global average pooling applied output last convolutional block, thus output model 2D tensor. max means global max pooling applied. classes optional number classes classify images , specified include_top TRUE, weights argument specified. classifier_activation str callable. activation function use \"top\" layer. Ignored unless include_top=TRUE. Set classifier_activation=NULL return logits \"top\" layer. loading pretrained weights, classifier_activation can NULL \"softmax\". name name model (string).","code":""},{"path":"https://keras3.posit.co/dev/reference/application_resnet50_v2.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Instantiates the ResNet50V2 architecture. — application_resnet50_v2","text":"Model instance.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_resnet50_v2.html","id":"reference","dir":"Reference","previous_headings":"","what":"Reference","title":"Instantiates the ResNet50V2 architecture. — application_resnet50_v2","text":"Identity Mappings Deep Residual Networks (CVPR 2016) image classification use cases, see page detailed examples. transfer learning use cases, make sure read guide transfer learning & fine-tuning.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_resnet50_v2.html","id":"note","dir":"Reference","previous_headings":"","what":"Note","title":"Instantiates the ResNet50V2 architecture. — application_resnet50_v2","text":"Keras Application expects specific kind input preprocessing. ResNet, call application_preprocess_inputs() inputs passing model. application_preprocess_inputs() scale input pixels -1 1.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/application_vgg16.html","id":null,"dir":"Reference","previous_headings":"","what":"Instantiates the VGG16 model. — application_vgg16","title":"Instantiates the VGG16 model. — application_vgg16","text":"Instantiates VGG16 model.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_vgg16.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Instantiates the VGG16 model. — application_vgg16","text":"","code":"application_vgg16(   include_top = TRUE,   weights = \"imagenet\",   input_tensor = NULL,   input_shape = NULL,   pooling = NULL,   classes = 1000L,   classifier_activation = \"softmax\",   name = \"vgg16\" )"},{"path":"https://keras3.posit.co/dev/reference/application_vgg16.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Instantiates the VGG16 model. — application_vgg16","text":"include_top whether include 3 fully-connected layers top network. weights one NULL (random initialization), \"imagenet\" (pre-training ImageNet), path weights file loaded. input_tensor optional Keras tensor (.e. output keras_input()) use image input model. input_shape optional shape tuple, specified include_top FALSE (otherwise input shape (224, 224, 3) (channels_last data format) (3, 224, 224) (\"channels_first\" data format). exactly 3 input channels, width height smaller 32. E.g. (200, 200, 3) one valid value. pooling Optional pooling mode feature extraction include_top FALSE. NULL means output model 4D tensor output last convolutional block. avg means global average pooling applied output last convolutional block, thus output model 2D tensor. max means global max pooling applied. classes optional number classes classify images , specified include_top TRUE, weights argument specified. classifier_activation str callable. activation function use \"top\" layer. Ignored unless include_top=TRUE. Set classifier_activation=NULL return logits \"top\" layer.  loading pretrained weights, classifier_activation can NULL \"softmax\". name name model (string).","code":""},{"path":"https://keras3.posit.co/dev/reference/application_vgg16.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Instantiates the VGG16 model. — application_vgg16","text":"Model instance.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_vgg16.html","id":"reference","dir":"Reference","previous_headings":"","what":"Reference","title":"Instantiates the VGG16 model. — application_vgg16","text":"Deep Convolutional Networks Large-Scale Image Recognition (ICLR 2015) image classification use cases, see page detailed examples. transfer learning use cases, make sure read guide transfer learning & fine-tuning. default input size model 224x224.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_vgg16.html","id":"note","dir":"Reference","previous_headings":"","what":"Note","title":"Instantiates the VGG16 model. — application_vgg16","text":"Keras Application expects specific kind input preprocessing. VGG16, call application_preprocess_inputs() inputs passing model. application_preprocess_inputs() convert input images RGB BGR, zero-center color channel respect ImageNet dataset, without scaling.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/application_vgg19.html","id":null,"dir":"Reference","previous_headings":"","what":"Instantiates the VGG19 model. — application_vgg19","title":"Instantiates the VGG19 model. — application_vgg19","text":"Instantiates VGG19 model.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_vgg19.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Instantiates the VGG19 model. — application_vgg19","text":"","code":"application_vgg19(   include_top = TRUE,   weights = \"imagenet\",   input_tensor = NULL,   input_shape = NULL,   pooling = NULL,   classes = 1000L,   classifier_activation = \"softmax\",   name = \"vgg19\" )"},{"path":"https://keras3.posit.co/dev/reference/application_vgg19.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Instantiates the VGG19 model. — application_vgg19","text":"include_top whether include 3 fully-connected layers top network. weights one NULL (random initialization), \"imagenet\" (pre-training ImageNet), path weights file loaded. input_tensor optional Keras tensor (.e. output keras_input()) use image input model. input_shape optional shape tuple, specified include_top FALSE (otherwise input shape (224, 224, 3) (channels_last data format) (3, 224, 224) (\"channels_first\" data format). exactly 3 input channels, width height smaller 32. E.g. (200, 200, 3) one valid value. pooling Optional pooling mode feature extraction include_top FALSE. NULL means output model 4D tensor output last convolutional block. avg means global average pooling applied output last convolutional block, thus output model 2D tensor. max means global max pooling applied. classes optional number classes classify images , specified include_top TRUE, weights argument specified. classifier_activation str callable. activation function use \"top\" layer. Ignored unless include_top=TRUE. Set classifier_activation=NULL return logits \"top\" layer.  loading pretrained weights, classifier_activation can NULL \"softmax\". name name model (string).","code":""},{"path":"https://keras3.posit.co/dev/reference/application_vgg19.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Instantiates the VGG19 model. — application_vgg19","text":"model instance.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_vgg19.html","id":"reference","dir":"Reference","previous_headings":"","what":"Reference","title":"Instantiates the VGG19 model. — application_vgg19","text":"Deep Convolutional Networks Large-Scale Image Recognition (ICLR 2015) image classification use cases, see page detailed examples. transfer learning use cases, make sure read guide transfer learning & fine-tuning. default input size model 224x224.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_vgg19.html","id":"note","dir":"Reference","previous_headings":"","what":"Note","title":"Instantiates the VGG19 model. — application_vgg19","text":"Keras Application expects specific kind input preprocessing. VGG19, call application_preprocess_inputs() inputs passing model. application_preprocess_inputs() convert input images RGB BGR, zero-center color channel respect ImageNet dataset, without scaling.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/application_xception.html","id":null,"dir":"Reference","previous_headings":"","what":"Instantiates the Xception architecture. — application_xception","title":"Instantiates the Xception architecture. — application_xception","text":"Instantiates Xception architecture.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_xception.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Instantiates the Xception architecture. — application_xception","text":"","code":"application_xception(   include_top = TRUE,   weights = \"imagenet\",   input_tensor = NULL,   input_shape = NULL,   pooling = NULL,   classes = 1000L,   classifier_activation = \"softmax\",   name = \"xception\" )"},{"path":"https://keras3.posit.co/dev/reference/application_xception.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Instantiates the Xception architecture. — application_xception","text":"include_top whether include 3 fully-connected layers top network. weights one NULL (random initialization), \"imagenet\" (pre-training ImageNet), path weights file loaded. input_tensor optional Keras tensor (.e. output keras_input()) use image input model. input_shape optional shape tuple, specified include_top FALSE (otherwise input shape (299, 299, 3). exactly 3 inputs channels, width height smaller 71. E.g. (150, 150, 3) one valid value. pooling Optional pooling mode feature extraction include_top FALSE. NULL means output model 4D tensor output last convolutional block. avg means global average pooling applied output last convolutional block, thus output model 2D tensor. max means global max pooling applied. classes optional number classes classify images , specified include_top TRUE, weights argument specified. classifier_activation str callable. activation function use \"top\" layer. Ignored unless include_top=TRUE. Set classifier_activation=NULL return logits \"top\" layer.  loading pretrained weights, classifier_activation can NULL \"softmax\". name name model (string).","code":""},{"path":"https://keras3.posit.co/dev/reference/application_xception.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Instantiates the Xception architecture. — application_xception","text":"model instance.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_xception.html","id":"reference","dir":"Reference","previous_headings":"","what":"Reference","title":"Instantiates the Xception architecture. — application_xception","text":"Xception: Deep Learning Depthwise Separable Convolutions (CVPR 2017) image classification use cases, see page detailed examples. transfer learning use cases, make sure read guide transfer learning & fine-tuning. default input image size model 299x299.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_xception.html","id":"note","dir":"Reference","previous_headings":"","what":"Note","title":"Instantiates the Xception architecture. — application_xception","text":"Keras Application expects specific kind input preprocessing. Xception, call application_preprocess_inputs() inputs passing model. application_preprocess_inputs() scale input pixels -1 1.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/audio_dataset_from_directory.html","id":null,"dir":"Reference","previous_headings":"","what":"Generates a tf.data.Dataset from audio files in a directory. — audio_dataset_from_directory","title":"Generates a tf.data.Dataset from audio files in a directory. — audio_dataset_from_directory","text":"directory structure :   calling audio_dataset_from_directory(main_directory, labels = 'inferred') return tf.data.Dataset yields batches audio files subdirectories class_a class_b, together labels 0 1 (0 corresponding class_a 1 corresponding class_b). .wav files supported time.","code":"main_directory/ ...class_a/ ......a_audio_1.wav ......a_audio_2.wav ...class_b/ ......b_audio_1.wav ......b_audio_2.wav"},{"path":"https://keras3.posit.co/dev/reference/audio_dataset_from_directory.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Generates a tf.data.Dataset from audio files in a directory. — audio_dataset_from_directory","text":"","code":"audio_dataset_from_directory(   directory,   labels = \"inferred\",   label_mode = \"int\",   class_names = NULL,   batch_size = 32L,   sampling_rate = NULL,   output_sequence_length = NULL,   ragged = FALSE,   shuffle = TRUE,   seed = NULL,   validation_split = NULL,   subset = NULL,   follow_links = FALSE,   verbose = TRUE )"},{"path":"https://keras3.posit.co/dev/reference/audio_dataset_from_directory.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Generates a tf.data.Dataset from audio files in a directory. — audio_dataset_from_directory","text":"directory Directory data located. labels \"inferred\", contain subdirectories, containing audio files class. Otherwise, directory structure ignored. labels Either \"inferred\" (labels generated directory structure), NULL (labels), list/tuple integer labels size number audio files found directory. Labels sorted according alphanumeric order audio file paths (obtained via os.walk(directory) Python). label_mode String describing encoding labels. Options : \"int\": means labels encoded integers (e.g. sparse_categorical_crossentropy loss). \"categorical\" means labels encoded categorical vector (e.g. categorical_crossentropy loss) \"binary\" means labels (can 2) encoded float32 scalars values 0 1 (e.g. binary_crossentropy). NULL (labels). class_names valid \"labels\" \"inferred\". explicit list class names (must match names subdirectories). Used control order classes (otherwise alphanumerical order used). batch_size Size batches data. Default: 32. NULL, data batched (dataset yield individual samples). sampling_rate Audio sampling rate (samples per second). output_sequence_length Maximum length audio sequence. Audio files longer truncated output_sequence_length. set NULL, sequences batch padded length longest sequence batch. ragged Whether return Ragged dataset (sequence length). Defaults FALSE. shuffle Whether shuffle data. Defaults TRUE. set FALSE, sorts data alphanumeric order. seed Optional random seed shuffling transformations. validation_split Optional float 0 1, fraction data reserve validation. subset Subset data return. One \"training\", \"validation\" \"\". used validation_split set. follow_links Whether visits subdirectories pointed symlinks. Defaults FALSE. verbose Whether display number information classes number files found. Defaults TRUE.","code":""},{"path":"https://keras3.posit.co/dev/reference/audio_dataset_from_directory.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Generates a tf.data.Dataset from audio files in a directory. — audio_dataset_from_directory","text":"tf.data.Dataset object. label_mode NULL, yields string tensors shape (batch_size,), containing contents batch audio files. Otherwise, yields tuple (audio, labels), audio shape (batch_size, sequence_length, num_channels) labels follows format described . Rules regarding labels format: label_mode int, labels int32 tensor shape (batch_size,). label_mode binary, labels float32 tensor 1s 0s shape (batch_size, 1). label_mode categorical, labels float32 tensor shape (batch_size, num_classes), representing one-hot encoding class index.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/back-compat.html","id":null,"dir":"Reference","previous_headings":"","what":"Backward compatibility — application_nasnetlarge","title":"Backward compatibility — application_nasnetlarge","text":"Backward compatibility Backward compatibility","code":""},{"path":"https://keras3.posit.co/dev/reference/back-compat.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Backward compatibility — application_nasnetlarge","text":"","code":"application_nasnetlarge(...)  application_nasnetmobile(...)"},{"path":"https://keras3.posit.co/dev/reference/back-compat.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Backward compatibility — application_nasnetlarge","text":"... passed application_nasnet_large() application_nasnet_mobile().","code":""},{"path":"https://keras3.posit.co/dev/reference/bidirectional.html","id":null,"dir":"Reference","previous_headings":"","what":"layer_bidirectional — bidirectional","title":"layer_bidirectional — bidirectional","text":"bidirectional() alias layer_bidirectional(). See ?layer_bidirectional() full documentation.","code":""},{"path":"https://keras3.posit.co/dev/reference/bidirectional.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"layer_bidirectional — bidirectional","text":"","code":"bidirectional(   object,   layer,   merge_mode = \"concat\",   weights = NULL,   backward_layer = NULL,   ... )"},{"path":"https://keras3.posit.co/dev/reference/bidirectional.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"layer_bidirectional — bidirectional","text":"object Object compose layer . tensor, array, sequential model. layer RNN instance, layer_lstm() layer_gru(). also Layer() instance meets following criteria: sequence-processing layer (accepts 3D+ inputs). go_backwards, return_sequences return_state attribute (semantics RNN class). input_spec attribute. Implement serialization via get_config() from_config(). Note recommended way create new RNN layers write custom RNN cell use layer_rnn(), instead subclassing Layer() directly. return_sequences TRUE, output masked timestep zero regardless layer's original zero_output_for_mask value. merge_mode Mode outputs forward backward RNNs combined. One {\"sum\", \"mul\", \"concat\", \"ave\", NULL}. NULL, outputs combined, returned list. Defaults \"concat\". weights see description backward_layer Optional RNN, Layer() instance used handle backwards input processing. backward_layer provided, layer instance passed layer argument used generate backward layer automatically. Note provided backward_layer layer properties matching layer argument, particular values stateful, return_states, return_sequences, etc. addition, backward_layer layer different go_backwards argument values. ValueError raised requirements met. ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/bidirectional.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"layer_bidirectional — bidirectional","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/callback_backup_and_restore.html","id":null,"dir":"Reference","previous_headings":"","what":"Callback to back up and restore the training state. — callback_backup_and_restore","title":"Callback to back up and restore the training state. — callback_backup_and_restore","text":"callback_backup_and_restore() callback intended recover training interruption happened middle fit execution, backing training states temporary checkpoint file, end epoch. backup overwrites previously written checkpoint file, given time one checkpoint file backup/restoring purpose. training restarts completion, training state (includes model weights epoch number) restored recently saved state beginning new fit run. completion fit run, temporary checkpoint file deleted. Note user responsible bring jobs back interruption. callback important backup restore mechanism fault tolerance purpose, model restored previous checkpoint expected one used back . user changes arguments passed compile fit, checkpoint saved fault tolerance can become invalid.","code":""},{"path":"https://keras3.posit.co/dev/reference/callback_backup_and_restore.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Callback to back up and restore the training state. — callback_backup_and_restore","text":"","code":"callback_backup_and_restore(   backup_dir,   save_freq = \"epoch\",   double_checkpoint = FALSE,   delete_checkpoint = TRUE )"},{"path":"https://keras3.posit.co/dev/reference/callback_backup_and_restore.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Callback to back up and restore the training state. — callback_backup_and_restore","text":"backup_dir String, path directory store data needed restore model. directory reused elsewhere store files, e.g. backup_and_restore callback another training run, another callback (e.g. callback_model_checkpoint) training run. save_freq \"epoch\", integer, FALSE. set \"epoch\", callback saves checkpoint end epoch. set integer, callback saves checkpoint every save_freq batches. Set save_freq = FALSE using preemption checkpointing (.e. save_before_preemption = TRUE). double_checkpoint Boolean. enabled, BackupAndRestore callback save 2 last training states (current previous). interruption current state loaded due IO error (e.g. file corrupted) try restore previous one. behaviour consume twice space disk, increase fault tolerance. Defaults FALSE. delete_checkpoint Boolean. backup_and_restore callback works saving checkpoint back training state. delete_checkpoint = TRUE, checkpoint deleted training finished. Use FALSE like keep checkpoint future usage. Defaults TRUE.","code":""},{"path":"https://keras3.posit.co/dev/reference/callback_backup_and_restore.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Callback to back up and restore the training state. — callback_backup_and_restore","text":"Callback instance can passed fit.keras.src.models.model.Model().","code":""},{"path":"https://keras3.posit.co/dev/reference/callback_backup_and_restore.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Callback to back up and restore the training state. — callback_backup_and_restore","text":"","code":"callback_interrupting <- new_callback_class(   \"InterruptingCallback\",   on_epoch_begin = function(epoch, logs = NULL) {     if (epoch == 4) {       stop('Interrupting!')     }   } )  backup_dir <- tempfile() callback <- callback_backup_and_restore(backup_dir = backup_dir) model <- keras_model_sequential() %>%   layer_dense(10) model %>% compile(optimizer = optimizer_sgd(), loss = 'mse')  # ensure model is built (i.e., weights are initialized) for # callback_backup_and_restore() model(op_ones(c(5, 20))) |> invisible()  tryCatch({   model %>% fit(x = op_ones(c(5, 20)),                 y = op_zeros(5),                 epochs = 10, batch_size = 1,                 callbacks = list(callback, callback_interrupting()),                 verbose = 0) }, python.builtin.RuntimeError = function(e) message(\"Interrupted!\")) ## Interrupted! model$history$epoch ## [1] 0 1 2 # model$history %>% keras3:::to_keras_training_history() %>% as.data.frame() %>% print()  history <- model %>% fit(x = op_ones(c(5, 20)),                          y = op_zeros(5),                          epochs = 10, batch_size = 1,                          callbacks = list(callback),                          verbose = 0)  # Only 6 more epochs are run, since first training got interrupted at # zero-indexed epoch 4, second training will continue from 4 to 9. nrow(as.data.frame(history)) ## [1] 10"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/callback_csv_logger.html","id":null,"dir":"Reference","previous_headings":"","what":"Callback that streams epoch results to a CSV file. — callback_csv_logger","title":"Callback that streams epoch results to a CSV file. — callback_csv_logger","text":"Supports values can represented string, including 1D iterables atomic vectors.","code":""},{"path":"https://keras3.posit.co/dev/reference/callback_csv_logger.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Callback that streams epoch results to a CSV file. — callback_csv_logger","text":"","code":"callback_csv_logger(filename, separator = \",\", append = FALSE)"},{"path":"https://keras3.posit.co/dev/reference/callback_csv_logger.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Callback that streams epoch results to a CSV file. — callback_csv_logger","text":"filename Filename CSV file, e.g. 'run/log.csv'. separator String used separate elements CSV file. append Boolean. TRUE: append file exists (useful continuing training). FALSE: overwrite existing file.","code":""},{"path":"https://keras3.posit.co/dev/reference/callback_csv_logger.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Callback that streams epoch results to a CSV file. — callback_csv_logger","text":"Callback instance can passed fit.keras.src.models.model.Model().","code":""},{"path":"https://keras3.posit.co/dev/reference/callback_csv_logger.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Callback that streams epoch results to a CSV file. — callback_csv_logger","text":"","code":"csv_logger <- callback_csv_logger('training.log') model %>% fit(X_train, Y_train, callbacks = list(csv_logger))"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/callback_early_stopping.html","id":null,"dir":"Reference","previous_headings":"","what":"Stop training when a monitored metric has stopped improving. — callback_early_stopping","title":"Stop training when a monitored metric has stopped improving. — callback_early_stopping","text":"Assuming goal training minimize loss. , metric monitored 'loss', mode 'min'. model$fit() training loop check end every epoch whether loss longer decreasing, considering min_delta patience applicable. found longer decreasing, model$stop_training marked TRUE training terminates. quantity monitored needs available logs list. make , pass loss metrics model$compile().","code":""},{"path":"https://keras3.posit.co/dev/reference/callback_early_stopping.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Stop training when a monitored metric has stopped improving. — callback_early_stopping","text":"","code":"callback_early_stopping(   monitor = \"val_loss\",   min_delta = 0L,   patience = 0L,   verbose = 0L,   mode = \"auto\",   baseline = NULL,   restore_best_weights = FALSE,   start_from_epoch = 0L )"},{"path":"https://keras3.posit.co/dev/reference/callback_early_stopping.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Stop training when a monitored metric has stopped improving. — callback_early_stopping","text":"monitor Quantity monitored. Defaults \"val_loss\". min_delta Minimum change monitored quantity qualify improvement, .e. absolute change less min_delta, count improvement. Defaults 0. patience Number epochs improvement training stopped. Defaults 0. verbose Verbosity mode, 0 1. Mode 0 silent, mode 1 displays messages callback takes action. Defaults 0. mode One {\"auto\", \"min\", \"max\"}. min mode, training stop quantity monitored stopped decreasing; \"max\" mode stop quantity monitored stopped increasing; \"auto\" mode, direction automatically inferred name monitored quantity. Defaults \"auto\". baseline Baseline value monitored quantity. NULL, training stop model show improvement baseline. Defaults NULL. restore_best_weights Whether restore model weights epoch best value monitored quantity. FALSE, model weights obtained last step training used. epoch restored regardless performance relative baseline. epoch improves baseline, training run patience epochs restore weights best epoch set. Defaults FALSE. start_from_epoch Number epochs wait starting monitor improvement. allows warm-period improvement expected thus training stopped. Defaults 0.","code":""},{"path":"https://keras3.posit.co/dev/reference/callback_early_stopping.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Stop training when a monitored metric has stopped improving. — callback_early_stopping","text":"Callback instance can passed fit.keras.src.models.model.Model().","code":""},{"path":"https://keras3.posit.co/dev/reference/callback_early_stopping.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Stop training when a monitored metric has stopped improving. — callback_early_stopping","text":"","code":"callback <- callback_early_stopping(monitor = 'loss',                                    patience = 3) # This callback will stop the training when there is no improvement in # the loss for three consecutive epochs. model <- keras_model_sequential() %>%   layer_dense(10) model %>% compile(optimizer = optimizer_sgd(), loss = 'mse') history <- model %>% fit(x = op_ones(c(5, 20)),                          y = op_zeros(5),                          epochs = 10, batch_size = 1,                          callbacks = list(callback),                          verbose = 0) nrow(as.data.frame(history))  # Only 4 epochs are run. ## [1] 10"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/callback_lambda.html","id":null,"dir":"Reference","previous_headings":"","what":"Callback for creating simple, custom callbacks on-the-fly. — callback_lambda","title":"Callback for creating simple, custom callbacks on-the-fly. — callback_lambda","text":"callback constructed anonymous functions called appropriate time (Model.{fit | evaluate | predict}). Note callbacks expects positional arguments, : on_epoch_begin on_epoch_end expect two positional arguments: epoch, logs on_train_begin on_train_end expect one positional argument: logs on_train_batch_begin on_train_batch_end expect positional argument batch named argument logs See Callback class definition full list functions expected arguments.","code":""},{"path":"https://keras3.posit.co/dev/reference/callback_lambda.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Callback for creating simple, custom callbacks on-the-fly. — callback_lambda","text":"","code":"callback_lambda(   on_epoch_begin = NULL,   on_epoch_end = NULL,   on_train_begin = NULL,   on_train_end = NULL,   on_train_batch_begin = NULL,   on_train_batch_end = NULL,   ... )"},{"path":"https://keras3.posit.co/dev/reference/callback_lambda.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Callback for creating simple, custom callbacks on-the-fly. — callback_lambda","text":"on_epoch_begin called beginning every epoch. on_epoch_end called end every epoch. on_train_begin called beginning model training. on_train_end called end model training. on_train_batch_begin called beginning every train batch. on_train_batch_end called end every train batch. ... function Callback() want override passing function_name = function. example, callback_lambda(.., on_train_end = train_end_fn). custom function needs arguments ones defined Callback().","code":""},{"path":"https://keras3.posit.co/dev/reference/callback_lambda.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Callback for creating simple, custom callbacks on-the-fly. — callback_lambda","text":"Callback instance can passed fit.keras.src.models.model.Model().","code":""},{"path":"https://keras3.posit.co/dev/reference/callback_lambda.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Callback for creating simple, custom callbacks on-the-fly. — callback_lambda","text":"","code":"# Print the batch number at the beginning of every batch. batch_print_callback <- callback_lambda(   on_train_batch_begin = function(batch, logs) {     print(batch)   } )  # Stream the epoch loss to a file in new-line delimited JSON format # (one valid JSON object per line) json_log <- file('loss_log.json', open = 'wt') json_logging_callback <- callback_lambda(   on_epoch_end = function(epoch, logs) {     jsonlite::write_json(       list(epoch = epoch, loss = logs$loss),       json_log,       append = TRUE     )   },   on_train_end = function(logs) {     close(json_log)   } )  # Terminate some processes after having finished model training. processes <- ... cleanup_callback <- callback_lambda(   on_train_end = function(logs) {     for (p in processes) {       if (is_alive(p)) {         terminate(p)       }     }   } )  model %>% fit(   ...,   callbacks = list(     batch_print_callback,     json_logging_callback,     cleanup_callback   ) )"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/callback_learning_rate_scheduler.html","id":null,"dir":"Reference","previous_headings":"","what":"Learning rate scheduler. — callback_learning_rate_scheduler","title":"Learning rate scheduler. — callback_learning_rate_scheduler","text":"beginning every epoch, callback gets updated learning rate value schedule function provided, current epoch current learning rate, applies updated learning rate optimizer.","code":""},{"path":"https://keras3.posit.co/dev/reference/callback_learning_rate_scheduler.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Learning rate scheduler. — callback_learning_rate_scheduler","text":"","code":"callback_learning_rate_scheduler(schedule, verbose = 0L)"},{"path":"https://keras3.posit.co/dev/reference/callback_learning_rate_scheduler.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Learning rate scheduler. — callback_learning_rate_scheduler","text":"schedule function takes epoch index (integer, indexed 0) current learning rate (float) inputs returns new learning rate output (float). verbose Integer. 0: quiet, 1: log update messages.","code":""},{"path":"https://keras3.posit.co/dev/reference/callback_learning_rate_scheduler.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Learning rate scheduler. — callback_learning_rate_scheduler","text":"Callback instance can passed fit.keras.src.models.model.Model().","code":""},{"path":"https://keras3.posit.co/dev/reference/callback_learning_rate_scheduler.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Learning rate scheduler. — callback_learning_rate_scheduler","text":"","code":"# This function keeps the initial learning rate steady for the first ten epochs # and decreases it exponentially after that. scheduler <- function(epoch, lr) {   if (epoch < 10)     return(lr)   else     return(lr * exp(-0.1)) }  model <- keras_model_sequential() |> layer_dense(units = 10) model |> compile(optimizer = optimizer_sgd(), loss = 'mse') model$optimizer$learning_rate |> as.array() |> round(5) ## [1] 0.01 callback <- callback_learning_rate_scheduler(schedule = scheduler) history <- model |> fit(x = array(runif(100), c(5, 20)),                         y = array(0, c(5, 1)),                         epochs = 15, callbacks = list(callback), verbose = 0) model$optimizer$learning_rate |> as.array() |> round(5) ## [1] 0.00607"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/callback_model_checkpoint.html","id":null,"dir":"Reference","previous_headings":"","what":"Callback to save the Keras model or model weights at some frequency. — callback_model_checkpoint","title":"Callback to save the Keras model or model weights at some frequency. — callback_model_checkpoint","text":"callback_model_checkpoint() used conjunction training using model |> fit() save model weights (checkpoint file) interval, model weights can loaded later continue training state saved. options callback provides include: Whether keep model achieved \"best performance\" far, whether save model end every epoch regardless performance. Definition \"best\"; quantity monitor whether maximized minimized. frequency save . Currently, callback supports saving end every epoch, fixed number training batches. Whether weights saved, whole model saved.","code":""},{"path":"https://keras3.posit.co/dev/reference/callback_model_checkpoint.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Callback to save the Keras model or model weights at some frequency. — callback_model_checkpoint","text":"","code":"callback_model_checkpoint(   filepath,   monitor = \"val_loss\",   verbose = 0L,   save_best_only = FALSE,   save_weights_only = FALSE,   mode = \"auto\",   save_freq = \"epoch\",   initial_value_threshold = NULL )"},{"path":"https://keras3.posit.co/dev/reference/callback_model_checkpoint.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Callback to save the Keras model or model weights at some frequency. — callback_model_checkpoint","text":"filepath string, path save model file. filepath can contain named formatting options, filled value epoch keys logs (passed on_epoch_end). filepath name needs end \".weights.h5\" save_weights_only = TRUE end \".keras\" \".h5\" checkpoint saving whole model (default). example: filepath \"{epoch:02d}-{val_loss:.2f}.keras\" \"{epoch:02d}-{val_loss:.2f}.weights.h5\", model checkpoints saved epoch number validation loss filename. directory filepath reused callbacks avoid conflicts. monitor metric name monitor. Typically metrics set model |> compile() method. Note: Prefix name \"val_\" monitor validation metrics. Use \"loss\" \"val_loss\" monitor model's total loss. specify metrics strings, like \"accuracy\", pass string (without \"val_\" prefix). pass Metric objects (created one metric_*()), monitor set metric$name. sure metric names can check contents history$metrics list returned history <- model |> fit() Multi-output models set additional prefixes metric names. verbose Verbosity mode, 0 1. Mode 0 silent, mode 1 displays messages callback takes action. save_best_only save_best_only = TRUE, saves model considered \"best\" latest best model according quantity monitored overwritten. filepath contain formatting options like {epoch} filepath overwritten new better model. save_weights_only TRUE, model's weights saved (model |> save_model_weights(filepath)), else full model saved (model |> save_model(filepath)). mode one {\"auto\", \"min\", \"max\"}. save_best_only = TRUE, decision overwrite current save file made based either maximization minimization monitored quantity. val_acc, \"max\", val_loss \"min\", etc. \"auto\" mode, direction automatically inferred name monitored quantity. save_freq \"epoch\" integer. using \"epoch\", callback saves model epoch. using integer, callback saves model end many batches. Model compiled steps_per_execution = N, saving criteria checked every Nth batch. Note saving aligned epochs, monitored metric may potentially less reliable (reflect little 1 batch, since metrics get reset every epoch). Defaults \"epoch\". initial_value_threshold Floating point initial \"best\" value metric monitored. applies save_best_value = TRUE. overwrites model weights already saved performance current model better value.","code":""},{"path":"https://keras3.posit.co/dev/reference/callback_model_checkpoint.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Callback to save the Keras model or model weights at some frequency. — callback_model_checkpoint","text":"Callback instance can passed fit.keras.src.models.model.Model().","code":""},{"path":"https://keras3.posit.co/dev/reference/callback_model_checkpoint.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Callback to save the Keras model or model weights at some frequency. — callback_model_checkpoint","text":"","code":"model <- keras_model_sequential(input_shape = c(10)) |>   layer_dense(1, activation = \"sigmoid\") |>   compile(loss = \"binary_crossentropy\", optimizer = \"adam\",           metrics = c('accuracy'))  EPOCHS <- 10 checkpoint_filepath <- tempfile('checkpoint-model-', fileext = \".keras\") model_checkpoint_callback <- callback_model_checkpoint(   filepath = checkpoint_filepath,   monitor = 'val_accuracy',   mode = 'max',   save_best_only = TRUE )  # Model is saved at the end of every epoch, if it's the best seen so far. model |> fit(x = random_uniform(c(2, 10)), y = op_ones(2, 1),              epochs = EPOCHS, validation_split = .5, verbose = 0,              callbacks = list(model_checkpoint_callback))  # The model (that are considered the best) can be loaded as - load_model(checkpoint_filepath) ## Model: \"sequential\" ## +---------------------------------+------------------------+---------------+ ## | Layer (type)                    | Output Shape           |       Param # | ## +=================================+========================+===============+ ## | dense (Dense)                   | (None, 1)              |            11 | ## +---------------------------------+------------------------+---------------+ ##  Total params: 35 (144.00 B) ##  Trainable params: 11 (44.00 B) ##  Non-trainable params: 0 (0.00 B) ##  Optimizer params: 24 (100.00 B) # Alternatively, one could checkpoint just the model weights as - checkpoint_filepath <- tempfile('checkpoint-', fileext = \".weights.h5\") model_checkpoint_callback <- callback_model_checkpoint(   filepath = checkpoint_filepath,   save_weights_only = TRUE,   monitor = 'val_accuracy',   mode = 'max',   save_best_only = TRUE )  # Model weights are saved at the end of every epoch, if it's the best seen # so far. # same as above model |> fit(x = random_uniform(c(2, 10)), y = op_ones(2, 1),              epochs = EPOCHS, validation_split = .5, verbose = 0,              callbacks = list(model_checkpoint_callback))  # The model weights (that are considered the best) can be loaded model |> load_model_weights(checkpoint_filepath)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/callback_reduce_lr_on_plateau.html","id":null,"dir":"Reference","previous_headings":"","what":"Reduce learning rate when a metric has stopped improving. — callback_reduce_lr_on_plateau","title":"Reduce learning rate when a metric has stopped improving. — callback_reduce_lr_on_plateau","text":"Models often benefit reducing learning rate factor 2-10 learning stagnates. callback monitors quantity improvement seen 'patience' number epochs, learning rate reduced.","code":""},{"path":"https://keras3.posit.co/dev/reference/callback_reduce_lr_on_plateau.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Reduce learning rate when a metric has stopped improving. — callback_reduce_lr_on_plateau","text":"","code":"callback_reduce_lr_on_plateau(   monitor = \"val_loss\",   factor = 0.1,   patience = 10L,   verbose = 0L,   mode = \"auto\",   min_delta = 1e-04,   cooldown = 0L,   min_lr = 0,   ... )"},{"path":"https://keras3.posit.co/dev/reference/callback_reduce_lr_on_plateau.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Reduce learning rate when a metric has stopped improving. — callback_reduce_lr_on_plateau","text":"monitor String. Quantity monitored. factor Float. Factor learning rate reduced. new_lr = lr * factor. patience Integer. Number epochs improvement learning rate reduced. verbose Integer. 0: quiet, 1: update messages. mode String. One {'auto', 'min', 'max'}. 'min' mode, learning rate reduced quantity monitored stopped decreasing; 'max' mode reduced quantity monitored stopped increasing; 'auto' mode, direction automatically inferred name monitored quantity. min_delta Float. Threshold measuring new optimum, focus significant changes. cooldown Integer. Number epochs wait resuming normal operation learning rate reduced. min_lr Float. Lower bound learning rate. ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/callback_reduce_lr_on_plateau.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Reduce learning rate when a metric has stopped improving. — callback_reduce_lr_on_plateau","text":"Callback instance can passed fit.keras.src.models.model.Model().","code":""},{"path":"https://keras3.posit.co/dev/reference/callback_reduce_lr_on_plateau.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Reduce learning rate when a metric has stopped improving. — callback_reduce_lr_on_plateau","text":"","code":"reduce_lr <- callback_reduce_lr_on_plateau(monitor = 'val_loss', factor = 0.2,                                            patience = 5, min_lr = 0.001) model %>% fit(x_train, y_train, callbacks = list(reduce_lr))"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/callback_remote_monitor.html","id":null,"dir":"Reference","previous_headings":"","what":"Callback used to stream events to a server. — callback_remote_monitor","title":"Callback used to stream events to a server. — callback_remote_monitor","text":"Requires requests library. Events sent root + '/publish/epoch/end/' default. Calls HTTP POST, data argument JSON-encoded named list event data. send_as_json = TRUE, content type request \"application/json\". Otherwise serialized JSON sent within form.","code":""},{"path":"https://keras3.posit.co/dev/reference/callback_remote_monitor.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Callback used to stream events to a server. — callback_remote_monitor","text":"","code":"callback_remote_monitor(   root = \"http://localhost:9000\",   path = \"/publish/epoch/end/\",   field = \"data\",   headers = NULL,   send_as_json = FALSE )"},{"path":"https://keras3.posit.co/dev/reference/callback_remote_monitor.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Callback used to stream events to a server. — callback_remote_monitor","text":"root String; root url target server. path String; path relative root events sent. field String; JSON field data stored. field used payload sent within form (.e. send_as_json = FALSE). headers Named list; optional custom HTTP headers. send_as_json Boolean; whether request sent \"application/json\".","code":""},{"path":"https://keras3.posit.co/dev/reference/callback_remote_monitor.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Callback used to stream events to a server. — callback_remote_monitor","text":"Callback instance can passed fit.keras.src.models.model.Model().","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/callback_swap_ema_weights.html","id":null,"dir":"Reference","previous_headings":"","what":"Swaps model weights and EMA weights before and after evaluation. — callback_swap_ema_weights","title":"Swaps model weights and EMA weights before and after evaluation. — callback_swap_ema_weights","text":"callbacks replaces model's weight values values optimizer's EMA weights (exponential moving average past model weights values, implementing \"Polyak averaging\") model evaluation, restores previous weights evaluation. SwapEMAWeights callback used conjunction optimizer sets use_ema = TRUE. Note weights swapped -place order save memory. behavior undefined modify EMA weights model weights callbacks.","code":""},{"path":"https://keras3.posit.co/dev/reference/callback_swap_ema_weights.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Swaps model weights and EMA weights before and after evaluation. — callback_swap_ema_weights","text":"","code":"callback_swap_ema_weights(swap_on_epoch = FALSE)"},{"path":"https://keras3.posit.co/dev/reference/callback_swap_ema_weights.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Swaps model weights and EMA weights before and after evaluation. — callback_swap_ema_weights","text":"swap_on_epoch Whether perform swapping on_epoch_begin() on_epoch_end(). useful want use EMA weights callbacks callback_model_checkpoint(). Defaults FALSE.","code":""},{"path":"https://keras3.posit.co/dev/reference/callback_swap_ema_weights.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Swaps model weights and EMA weights before and after evaluation. — callback_swap_ema_weights","text":"Callback instance can passed fit.keras.src.models.model.Model().","code":""},{"path":"https://keras3.posit.co/dev/reference/callback_swap_ema_weights.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Swaps model weights and EMA weights before and after evaluation. — callback_swap_ema_weights","text":"","code":"# Remember to set `use_ema=TRUE` in the optimizer optimizer <- optimizer_sgd(use_ema = TRUE) model |> compile(optimizer = optimizer, loss = ..., metrics = ...)  # Metrics will be computed with EMA weights model |> fit(X_train, Y_train,              callbacks = c(callback_swap_ema_weights()))  # If you want to save model checkpoint with EMA weights, you can set # `swap_on_epoch=TRUE` and place ModelCheckpoint after SwapEMAWeights. model |> fit(   X_train, Y_train,   callbacks = c(     callback_swap_ema_weights(swap_on_epoch = TRUE),     callback_model_checkpoint(...)   ) )"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/callback_tensorboard.html","id":null,"dir":"Reference","previous_headings":"","what":"Enable visualizations for TensorBoard. — callback_tensorboard","title":"Enable visualizations for TensorBoard. — callback_tensorboard","text":"TensorBoard visualization tool provided TensorFlow. TensorFlow installation required use callback. callback logs events TensorBoard, including: Metrics summary plots Training graph visualization Weight histograms Sampled profiling used model |> evaluate() regular validation addition epoch summaries, summary records evaluation metrics vs model$optimizer$iterations written. metric names prepended evaluation, model$optimizer$iterations step visualized TensorBoard. installed TensorFlow pip reticulate::py_install(), able launch TensorBoard command line:   R tensorflow::tensorboard(). can find information TensorBoard .","code":"tensorboard --logdir=path_to_your_logs"},{"path":"https://keras3.posit.co/dev/reference/callback_tensorboard.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Enable visualizations for TensorBoard. — callback_tensorboard","text":"","code":"callback_tensorboard(   log_dir = \"logs\",   histogram_freq = 0L,   write_graph = TRUE,   write_images = FALSE,   write_steps_per_second = FALSE,   update_freq = \"epoch\",   profile_batch = 0L,   embeddings_freq = 0L,   embeddings_metadata = NULL )"},{"path":"https://keras3.posit.co/dev/reference/callback_tensorboard.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Enable visualizations for TensorBoard. — callback_tensorboard","text":"log_dir path directory save log files parsed TensorBoard. e.g., log_dir = file.path(working_dir, 'logs'). directory reused callbacks. histogram_freq frequency (epochs) compute weight histograms layers model. set 0, histograms computed. Validation data (split) must specified histogram visualizations. write_graph (supported time) Whether visualize graph TensorBoard. Note log file can become quite large write_graph set TRUE. write_images whether write model weights visualize image TensorBoard. write_steps_per_second whether log training steps per second TensorBoard. supports epoch batch frequency logging. update_freq \"batch\" \"epoch\" integer. using \"epoch\", writes losses metrics TensorBoard every epoch. using integer, say 1000, metrics losses (including custom ones added Model.compile) logged TensorBoard every 1000 batches. \"batch\" synonym 1, meaning written every batch. Note however writing frequently TensorBoard can slow training, especially used distribution strategies incur additional synchronization overhead. Batch-level summary writing also available via train_step override. Please see TensorBoard Scalars tutorial details. profile_batch Profile batch(es) sample compute characteristics. profile_batch must non-negative integer tuple integers. pair positive integers signify range batches profile. default, profiling disabled. embeddings_freq frequency (epochs) embedding layers visualized. set 0, embeddings visualized. embeddings_metadata Named list maps embedding layer names filename file save metadata embedding layer. case metadata file used embedding layers, single filename can passed.","code":""},{"path":"https://keras3.posit.co/dev/reference/callback_tensorboard.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Enable visualizations for TensorBoard. — callback_tensorboard","text":"Callback instance can passed fit.keras.src.models.model.Model().","code":""},{"path":"https://keras3.posit.co/dev/reference/callback_tensorboard.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Enable visualizations for TensorBoard. — callback_tensorboard","text":"Custom batch-level summaries subclassed Model:   Custom batch-level summaries Functional API Model:   Profiling:","code":"tensorboard_callback <- callback_tensorboard(log_dir = \"./logs\") model %>% fit(x_train, y_train, epochs = 2, callbacks = list(tensorboard_callback)) # Then run the tensorboard command to view the visualizations. MyModel <- new_model_class(\"MyModel\",   initialize = function() {     self$dense <- layer_dense(units = 10)   },   call = function(x) {     outputs <- x |> self$dense()     tf$summary$histogram('outputs', outputs)     outputs   } )  model <- MyModel() model |> compile(optimizer = 'sgd', loss = 'mse')  # Make sure to set `update_freq = N` to log a batch-level summary every N # batches. In addition to any `tf$summary` contained in `model$call()`, # metrics added in `model |>compile` will be logged every N batches. tb_callback <- callback_tensorboard(log_dir = './logs', update_freq = 1) model |> fit(x_train, y_train, callbacks = list(tb_callback)) my_summary <- function(x) {   tf$summary$histogram('x', x)   x }  inputs <- layer_input(10) outputs <- inputs |>   layer_dense(10) |>   layer_lambda(my_summary)  model <- keras_model(inputs, outputs) model |> compile(optimizer = 'sgd', loss = 'mse')  # Make sure to set `update_freq = N` to log a batch-level summary every N # batches. In addition to any `tf.summary` contained in `Model.call`, # metrics added in `Model.compile` will be logged every N batches. tb_callback <- callback_tensorboard(log_dir = './logs', update_freq = 1) model |> fit(x_train, y_train, callbacks = list(tb_callback)) # Profile a single batch, e.g. the 5th batch. tensorboard_callback <- callback_tensorboard(   log_dir = './logs', profile_batch = 5) model |> fit(x_train, y_train, epochs = 2,              callbacks = list(tensorboard_callback))  # Profile a range of batches, e.g. from 10 to 20. tensorboard_callback <- callback_tensorboard(   log_dir = './logs', profile_batch = c(10, 20)) model |> fit(x_train, y_train, epochs = 2,              callbacks = list(tensorboard_callback))"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/callback_terminate_on_nan.html","id":null,"dir":"Reference","previous_headings":"","what":"Callback that terminates training when a NaN loss is encountered. — callback_terminate_on_nan","title":"Callback that terminates training when a NaN loss is encountered. — callback_terminate_on_nan","text":"Callback terminates training NaN loss encountered.","code":""},{"path":"https://keras3.posit.co/dev/reference/callback_terminate_on_nan.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Callback that terminates training when a NaN loss is encountered. — callback_terminate_on_nan","text":"","code":"callback_terminate_on_nan()"},{"path":"https://keras3.posit.co/dev/reference/callback_terminate_on_nan.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Callback that terminates training when a NaN loss is encountered. — callback_terminate_on_nan","text":"Callback instance can passed fit.keras.src.models.model.Model().","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/clear_session.html","id":null,"dir":"Reference","previous_headings":"","what":"Resets all state generated by Keras. — clear_session","title":"Resets all state generated by Keras. — clear_session","text":"Keras manages global state, uses implement Functional model-building API uniquify autogenerated layer names. creating many models loop, global state consume increasing amount memory time, may want clear . Calling clear_session() releases global state: helps avoid clutter old models layers, especially memory limited. Example 1: calling clear_session() creating models loop   Example 2: resetting layer name generation counter","code":"for (i in 1:100) {   # Without `clear_session()`, each iteration of this loop will   # slightly increase the size of the global state managed by Keras   model <- keras_model_sequential()   for (j in 1:10) {     model <- model |> layer_dense(units = 10)   } }  for (i in 1:100) {   # With `clear_session()` called at the beginning,   # Keras starts with a blank state at each iteration   # and memory consumption is constant over time.   clear_session()   model <- keras_model_sequential()   for (j in 1:10) {     model <- model |> layer_dense(units = 10)   } } layers <- lapply(1:10, \\(i) layer_dense(units = 10))  new_layer <- layer_dense(units = 10) print(new_layer$name) ## [1] \"dense_10\" clear_session() new_layer <- layer_dense(units = 10) print(new_layer$name) ## [1] \"dense\""},{"path":"https://keras3.posit.co/dev/reference/clear_session.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Resets all state generated by Keras. — clear_session","text":"","code":"clear_session(free_memory = TRUE)"},{"path":"https://keras3.posit.co/dev/reference/clear_session.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Resets all state generated by Keras. — clear_session","text":"free_memory Whether call Python garbage collection. usually good practice call make sure memory used deleted objects immediately freed. However, may take seconds execute, using clear_session() short loop, may want skip .","code":""},{"path":"https://keras3.posit.co/dev/reference/clear_session.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Resets all state generated by Keras. — clear_session","text":"NULL, invisibly, called side effects.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/clone_model.html","id":null,"dir":"Reference","previous_headings":"","what":"Clone a Functional or Sequential Model instance. — clone_model","title":"Clone a Functional or Sequential Model instance. — clone_model","text":"Model cloning similar calling model new inputs, except creates new layers (thus new weights) instead sharing weights existing layers. Note clone_model() preserve uniqueness shared objects within model (e.g. single variable attached two distinct layers restored two separate variables).","code":""},{"path":"https://keras3.posit.co/dev/reference/clone_model.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Clone a Functional or Sequential Model instance. — clone_model","text":"","code":"clone_model(   model,   input_tensors = NULL,   clone_function = NULL,   call_function = NULL,   recursive = FALSE,   ... )"},{"path":"https://keras3.posit.co/dev/reference/clone_model.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Clone a Functional or Sequential Model instance. — clone_model","text":"model Instance Model (Functional model Sequential model). input_tensors Optional list input tensors build model upon. provided, new keras_input() objects created. clone_function Callable signature function(layer) used clone layer target model (except Input instances). takes argument layer instance cloned, returns corresponding layer instance used model copy. unspecified, callable defaults following serialization/deserialization function: function(layer) layer$`__class__`$from_config(layer$get_config()). passing custom callable, can customize copy model, e.g. wrapping certain layers interest (might want replace LSTM instances equivalent Bidirectional(LSTM(...)) instances, example). Defaults NULL. call_function Callable signature function(layer, ...) used call cloned layer set inputs. takes layer instance, call arguments, returns call outputs. unspecified, callable defaults regular call() method: function(layer, ...) .call(layer, list(...)). passing custom callable, can insert new layers given layer. recursive Note, argument can used Functional models. Boolean. Whether recursively clone Sequential Functional models encountered original Sequential/Functional model. FALSE, inner models cloned calling clone_function(). TRUE, inner models cloned calling clone_model() clone_function, call_function, recursive arguments. Note case, call_function propagated Sequential model (since applicable Sequential models). ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/clone_model.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Clone a Functional or Sequential Model instance. — clone_model","text":"instance Model reproducing behavior original model, top new inputs tensors, using newly instantiated weights. cloned model may behave differently original model custom clone_function call_function modifies layer layer call.","code":""},{"path":"https://keras3.posit.co/dev/reference/clone_model.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Clone a Functional or Sequential Model instance. — clone_model","text":"Using clone_function make model deterministic setting random seed everywhere:   Using call_function add Dropout layer Dense layer (without recreating new layers):     Note subclassed models cloned default, since internal layer structure known. achieve equivalent functionality clone_model case subclassed model, simply make sure model class implements get_config() (optionally from_config()), call:   case subclassed model, using custom clone_function.","code":"# Create a test Sequential model. model <- keras_model_sequential(input_shape = c(728)) |>   layer_dense(32, activation = 'relu') |>   layer_dense(1, activation = 'sigmoid')  # Create a copy of the test model (with freshly initialized weights). new_model <- clone_model(model) clone_function <- function(layer) {   config <- layer$get_config()   if (\"seed\" %in% names(config))     config$seed <- 1337L   layer$`__class__`$from_config(config) }  new_model <- clone_model(model, clone_function = clone_function) call_function <- function(layer, ...) {   out <- layer(...)   if (inherits(layer, keras$layers$Dense))     out <- out |> layer_dropout(0.5)   out }  inputs <- keras_input(c(728)) outputs <- inputs |>   layer_dense(32, activation = 'relu') |>   layer_dense(1, activation = 'sigmoid') model <- keras_model(inputs, outputs)  new_model <- clone_model(   model,   clone_function = function(x) x, # Reuse the same layers.   call_function = call_function, ) new_model ## Model: \"functional_4\" ## +-----------------------------------+--------------------------+---------------+ ## | Layer (type)                      | Output Shape             |       Param # | ## +===================================+==========================+===============+ ## | keras_tensor_8 (InputLayer)       | (None, 728)              |             0 | ## +-----------------------------------+--------------------------+---------------+ ## | dense_2 (Dense)                   | (None, 32)               |        23,328 | ## +-----------------------------------+--------------------------+---------------+ ## | dropout (Dropout)                 | (None, 32)               |             0 | ## +-----------------------------------+--------------------------+---------------+ ## | dense_3 (Dense)                   | (None, 1)                |            33 | ## +-----------------------------------+--------------------------+---------------+ ## | dropout_1 (Dropout)               | (None, 1)                |             0 | ## +-----------------------------------+--------------------------+---------------+ ##  Total params: 23,361 (91.25 KB) ##  Trainable params: 23,361 (91.25 KB) ##  Non-trainable params: 0 (0.00 B) new_model <- model$`__class__`$from_config(model$get_config())"},{"path":"https://keras3.posit.co/dev/reference/compile.keras.src.models.model.Model.html","id":null,"dir":"Reference","previous_headings":"","what":"Configure a model for training. — compile.keras.src.models.model.Model","title":"Configure a model for training. — compile.keras.src.models.model.Model","text":"Configure model training.","code":""},{"path":"https://keras3.posit.co/dev/reference/compile.keras.src.models.model.Model.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Configure a model for training. — compile.keras.src.models.model.Model","text":"","code":"# S3 method for class 'keras.src.models.model.Model' compile(   object,   optimizer = \"rmsprop\",   loss = NULL,   metrics = NULL,   ...,   loss_weights = NULL,   weighted_metrics = NULL,   run_eagerly = FALSE,   steps_per_execution = 1L,   jit_compile = \"auto\",   auto_scale_loss = TRUE )"},{"path":"https://keras3.posit.co/dev/reference/compile.keras.src.models.model.Model.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Configure a model for training. — compile.keras.src.models.model.Model","text":"object Keras model object optimizer String (name optimizer) optimizer instance. See optimizer_* family. loss Loss function. May : string (name builtin loss function), custom function, Loss instance (returned loss_* family functions). loss function callable signature loss = fn(y_true, y_pred), y_true ground truth values, y_pred model's predictions. y_true shape (batch_size, d1, .. dN) (except case sparse loss functions sparse categorical crossentropy expects integer arrays shape (batch_size, d1, .. dN-1)). y_pred shape (batch_size, d1, .. dN). loss function return float tensor. metrics List metrics evaluated model training testing. can : string (name built-function), function, optionally \"name\" attribute Metric() instance. See metric_* family functions. Typically use metrics = c('accuracy'). function callable signature result = fn(y_true, y_pred). specify different metrics different outputs multi-output model, also pass named list, metrics = list(= 'accuracy', b = c('accuracy', 'mse')). can also pass list specify metric list metrics output, metrics = list(c('accuracy'), c('accuracy', 'mse')) metrics = list('accuracy', c('accuracy', 'mse')). pass strings 'accuracy' 'acc', convert one metric_binary_accuracy(), metric_categorical_accuracy(), metric_sparse_categorical_accuracy() based shapes targets model output. similar conversion done strings \"crossentropy\" \"ce\" well. metrics passed evaluated without sample weighting; like sample weighting apply, can specify metrics via weighted_metrics argument instead. providing anonymous R function, can customize printed name training assigning attr(<fn>, \"name\") <- \"my_custom_metric_name\", calling custom_metric(\"my_custom_metric_name\", <fn>) ... Additional arguments passed compile() model method. loss_weights Optional list (named unnamed) specifying scalar coefficients (R numerics) weight loss contributions different model outputs. loss value minimized model weighted sum individual losses, weighted loss_weights coefficients.  unnamed list, expected 1:1 mapping model's outputs. named list, expected map output names (strings) scalar coefficients. weighted_metrics List metrics evaluated weighted sample_weight class_weight training testing. run_eagerly Bool. TRUE, model's forward pass never compiled. recommended leave FALSE training (best performance), set TRUE debugging. steps_per_execution Int. number batches run single compiled function call. Running multiple batches inside single compiled function call can greatly improve performance TPUs small models large R/Python overhead. , one full epoch run execution. number larger size epoch passed, execution truncated size epoch. Note steps_per_execution set N, Callback$on_batch_begin Callback$on_batch_end methods called every N batches (.e. /compiled function execution). supported PyTorch backend. jit_compile Bool \"auto\". Whether use XLA compilation compiling model. jax tensorflow backends, jit_compile=\"auto\" enables XLA compilation model supports , disabled otherwise. torch backend, \"auto\" default eager execution jit_compile=True run torch.compile \"inductor\" backend. auto_scale_loss Bool. TRUE model dtype policy \"mixed_float16\", passed optimizer automatically wrapped LossScaleOptimizer, dynamically scale loss prevent underflow.","code":""},{"path":"https://keras3.posit.co/dev/reference/compile.keras.src.models.model.Model.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Configure a model for training. — compile.keras.src.models.model.Model","text":"called primarily side effect modifying object -place. first argument object also returned, invisibly, enable usage pipe.","code":""},{"path":"https://keras3.posit.co/dev/reference/compile.keras.src.models.model.Model.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Configure a model for training. — compile.keras.src.models.model.Model","text":"","code":"model |> compile(   optimizer = optimizer_adam(learning_rate = 1e-3),   loss = loss_binary_crossentropy(),   metrics = c(metric_binary_accuracy(),               metric_false_negatives()) )"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/config_backend.html","id":null,"dir":"Reference","previous_headings":"","what":"Publicly accessible method for determining the current backend. — config_backend","title":"Publicly accessible method for determining the current backend. — config_backend","text":"Publicly accessible method determining current backend.","code":""},{"path":"https://keras3.posit.co/dev/reference/config_backend.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Publicly accessible method for determining the current backend. — config_backend","text":"","code":"config_backend()"},{"path":"https://keras3.posit.co/dev/reference/config_backend.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Publicly accessible method for determining the current backend. — config_backend","text":"String, name backend Keras currently using. One \"tensorflow\", \"torch\", \"jax\".","code":""},{"path":"https://keras3.posit.co/dev/reference/config_backend.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Publicly accessible method for determining the current backend. — config_backend","text":"","code":"config_backend() ## [1] \"tensorflow\""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/config_disable_flash_attention.html","id":null,"dir":"Reference","previous_headings":"","what":"Disable flash attention. — config_disable_flash_attention","title":"Disable flash attention. — config_disable_flash_attention","text":"Flash attention offers performance optimization attention layers, making especially useful large language models (LLMs) benefit faster memory-efficient attention computations. disabled, supported layers like MultiHeadAttention use flash attention faster computations.","code":""},{"path":"https://keras3.posit.co/dev/reference/config_disable_flash_attention.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Disable flash attention. — config_disable_flash_attention","text":"","code":"config_disable_flash_attention()"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/config_disable_interactive_logging.html","id":null,"dir":"Reference","previous_headings":"","what":"Turn off interactive logging. — config_disable_interactive_logging","title":"Turn off interactive logging. — config_disable_interactive_logging","text":"interactive logging disabled, Keras sends logs absl.logging. best option using Keras non-interactive way, running training inference job server.","code":""},{"path":"https://keras3.posit.co/dev/reference/config_disable_interactive_logging.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Turn off interactive logging. — config_disable_interactive_logging","text":"","code":"config_disable_interactive_logging()"},{"path":"https://keras3.posit.co/dev/reference/config_disable_interactive_logging.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Turn off interactive logging. — config_disable_interactive_logging","text":"return value, called side effects.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/config_disable_traceback_filtering.html","id":null,"dir":"Reference","previous_headings":"","what":"Turn off traceback filtering. — config_disable_traceback_filtering","title":"Turn off traceback filtering. — config_disable_traceback_filtering","text":"Raw Keras tracebacks (also known stack traces) involve many internal frames, can challenging read , actionable end users. default, Keras filters internal frames exceptions raises, keep traceback short, readable, focused actionable (code). See also config_enable_traceback_filtering() config_is_traceback_filtering_enabled(). previously disabled traceback filtering via config_disable_traceback_filtering(), can re-enable via config_enable_traceback_filtering().","code":""},{"path":"https://keras3.posit.co/dev/reference/config_disable_traceback_filtering.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Turn off traceback filtering. — config_disable_traceback_filtering","text":"","code":"config_disable_traceback_filtering()"},{"path":"https://keras3.posit.co/dev/reference/config_disable_traceback_filtering.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Turn off traceback filtering. — config_disable_traceback_filtering","text":"return value, called side effects.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/config_dtype_policy.html","id":null,"dir":"Reference","previous_headings":"","what":"Returns the current default dtype policy object. — config_dtype_policy","title":"Returns the current default dtype policy object. — config_dtype_policy","text":"Returns current default dtype policy object.","code":""},{"path":"https://keras3.posit.co/dev/reference/config_dtype_policy.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Returns the current default dtype policy object. — config_dtype_policy","text":"","code":"config_dtype_policy()"},{"path":"https://keras3.posit.co/dev/reference/config_dtype_policy.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Returns the current default dtype policy object. — config_dtype_policy","text":"DTypePolicy object.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/config_enable_flash_attention.html","id":null,"dir":"Reference","previous_headings":"","what":"Enable flash attention. — config_enable_flash_attention","title":"Enable flash attention. — config_enable_flash_attention","text":"Flash attention offers performance optimization attention layers, making especially useful large language models (LLMs) benefit faster memory-efficient attention computations. enabled, supported layers like layer_multi_head_attention attempt use flash attention faster computations. default, feature enabled. Note enabling flash attention guarantee always used. Typically, inputs must float16 bfloat16 dtype, input layout requirements may vary depending backend.","code":""},{"path":"https://keras3.posit.co/dev/reference/config_enable_flash_attention.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Enable flash attention. — config_enable_flash_attention","text":"","code":"config_enable_flash_attention()"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/config_enable_interactive_logging.html","id":null,"dir":"Reference","previous_headings":"","what":"Turn on interactive logging. — config_enable_interactive_logging","title":"Turn on interactive logging. — config_enable_interactive_logging","text":"interactive logging enabled, Keras displays logs via stdout. provides best experience using Keras interactive environment shell notebook.","code":""},{"path":"https://keras3.posit.co/dev/reference/config_enable_interactive_logging.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Turn on interactive logging. — config_enable_interactive_logging","text":"","code":"config_enable_interactive_logging()"},{"path":"https://keras3.posit.co/dev/reference/config_enable_interactive_logging.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Turn on interactive logging. — config_enable_interactive_logging","text":"return value, called side effects.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/config_enable_traceback_filtering.html","id":null,"dir":"Reference","previous_headings":"","what":"Turn on traceback filtering. — config_enable_traceback_filtering","title":"Turn on traceback filtering. — config_enable_traceback_filtering","text":"Raw Keras tracebacks (also known stack traces) involve many internal frames, can challenging read , actionable end users. default, Keras filters internal frames exceptions raises, keep traceback short, readable, focused actionable (code). See also config_disable_traceback_filtering() config_is_traceback_filtering_enabled(). previously disabled traceback filtering via config_disable_traceback_filtering(), can re-enable via config_enable_traceback_filtering().","code":""},{"path":"https://keras3.posit.co/dev/reference/config_enable_traceback_filtering.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Turn on traceback filtering. — config_enable_traceback_filtering","text":"","code":"config_enable_traceback_filtering()"},{"path":"https://keras3.posit.co/dev/reference/config_enable_traceback_filtering.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Turn on traceback filtering. — config_enable_traceback_filtering","text":"return value, called side effects.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/config_enable_unsafe_deserialization.html","id":null,"dir":"Reference","previous_headings":"","what":"Disables safe mode globally, allowing deserialization of lambdas. — config_enable_unsafe_deserialization","title":"Disables safe mode globally, allowing deserialization of lambdas. — config_enable_unsafe_deserialization","text":"Disables safe mode globally, allowing deserialization lambdas.","code":""},{"path":"https://keras3.posit.co/dev/reference/config_enable_unsafe_deserialization.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Disables safe mode globally, allowing deserialization of lambdas. — config_enable_unsafe_deserialization","text":"","code":"config_enable_unsafe_deserialization()"},{"path":"https://keras3.posit.co/dev/reference/config_enable_unsafe_deserialization.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Disables safe mode globally, allowing deserialization of lambdas. — config_enable_unsafe_deserialization","text":"return value, called side effects.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/config_epsilon.html","id":null,"dir":"Reference","previous_headings":"","what":"Return the value of the fuzz factor used in numeric expressions. — config_epsilon","title":"Return the value of the fuzz factor used in numeric expressions. — config_epsilon","text":"Return value fuzz factor used numeric expressions.","code":""},{"path":"https://keras3.posit.co/dev/reference/config_epsilon.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Return the value of the fuzz factor used in numeric expressions. — config_epsilon","text":"","code":"config_epsilon()"},{"path":"https://keras3.posit.co/dev/reference/config_epsilon.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Return the value of the fuzz factor used in numeric expressions. — config_epsilon","text":"float.","code":""},{"path":"https://keras3.posit.co/dev/reference/config_epsilon.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Return the value of the fuzz factor used in numeric expressions. — config_epsilon","text":"","code":"config_epsilon() ## [1] 1e-07"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/config_floatx.html","id":null,"dir":"Reference","previous_headings":"","what":"Return the default float type, as a string. — config_floatx","title":"Return the default float type, as a string. — config_floatx","text":"E.g. 'bfloat16' 'float16', 'float32', 'float64'.","code":""},{"path":"https://keras3.posit.co/dev/reference/config_floatx.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Return the default float type, as a string. — config_floatx","text":"","code":"config_floatx()"},{"path":"https://keras3.posit.co/dev/reference/config_floatx.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Return the default float type, as a string. — config_floatx","text":"String, current default float type.","code":""},{"path":"https://keras3.posit.co/dev/reference/config_floatx.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Return the default float type, as a string. — config_floatx","text":"","code":"keras3::config_floatx() ## [1] \"float32\""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/config_image_data_format.html","id":null,"dir":"Reference","previous_headings":"","what":"Return the default image data format convention. — config_image_data_format","title":"Return the default image data format convention. — config_image_data_format","text":"Return default image data format convention.","code":""},{"path":"https://keras3.posit.co/dev/reference/config_image_data_format.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Return the default image data format convention. — config_image_data_format","text":"","code":"config_image_data_format()"},{"path":"https://keras3.posit.co/dev/reference/config_image_data_format.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Return the default image data format convention. — config_image_data_format","text":"string, either 'channels_first' 'channels_last'.","code":""},{"path":"https://keras3.posit.co/dev/reference/config_image_data_format.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Return the default image data format convention. — config_image_data_format","text":"","code":"config_image_data_format() ## [1] \"channels_last\""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/config_is_flash_attention_enabled.html","id":null,"dir":"Reference","previous_headings":"","what":"Checks whether flash attention is globally enabled in Keras. — config_is_flash_attention_enabled","title":"Checks whether flash attention is globally enabled in Keras. — config_is_flash_attention_enabled","text":"Flash attention performance-optimized method computing attention large models, transformers, allowing faster memory-efficient operations. function checks global Keras configuration determine flash attention enabled compatible layers (e.g., MultiHeadAttention). Note enabling flash attention guarantee always used. Typically, inputs must float16 bfloat16 dtype, input layout requirements may vary depending backend.","code":""},{"path":"https://keras3.posit.co/dev/reference/config_is_flash_attention_enabled.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Checks whether flash attention is globally enabled in Keras. — config_is_flash_attention_enabled","text":"","code":"config_is_flash_attention_enabled()"},{"path":"https://keras3.posit.co/dev/reference/config_is_flash_attention_enabled.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Checks whether flash attention is globally enabled in Keras. — config_is_flash_attention_enabled","text":"FALSE disabled; otherwise, indicates enabled.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/config_is_interactive_logging_enabled.html","id":null,"dir":"Reference","previous_headings":"","what":"Check if interactive logging is enabled. — config_is_interactive_logging_enabled","title":"Check if interactive logging is enabled. — config_is_interactive_logging_enabled","text":"switch writing logs stdout absl.logging, may use config_enable_interactive_logging() config_disable_interactive_logging().","code":""},{"path":"https://keras3.posit.co/dev/reference/config_is_interactive_logging_enabled.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Check if interactive logging is enabled. — config_is_interactive_logging_enabled","text":"","code":"config_is_interactive_logging_enabled()"},{"path":"https://keras3.posit.co/dev/reference/config_is_interactive_logging_enabled.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Check if interactive logging is enabled. — config_is_interactive_logging_enabled","text":"Boolean, TRUE interactive logging enabled, FALSE otherwise.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/config_is_nnx_enabled.html","id":null,"dir":"Reference","previous_headings":"","what":"Check whether NNX-specific features are enabled on the JAX backend. — config_is_nnx_enabled","title":"Check whether NNX-specific features are enabled on the JAX backend. — config_is_nnx_enabled","text":"Check whether NNX-specific features enabled JAX backend.","code":""},{"path":"https://keras3.posit.co/dev/reference/config_is_nnx_enabled.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Check whether NNX-specific features are enabled on the JAX backend. — config_is_nnx_enabled","text":"","code":"config_is_nnx_enabled()"},{"path":"https://keras3.posit.co/dev/reference/config_is_nnx_enabled.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Check whether NNX-specific features are enabled on the JAX backend. — config_is_nnx_enabled","text":"Logical flag; TRUE NNX backend features enabled, FALSE otherwise.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/config_is_traceback_filtering_enabled.html","id":null,"dir":"Reference","previous_headings":"","what":"Check if traceback filtering is enabled. — config_is_traceback_filtering_enabled","title":"Check if traceback filtering is enabled. — config_is_traceback_filtering_enabled","text":"Raw Keras tracebacks (also known stack traces) involve many internal frames, can challenging read , actionable end users. default, Keras filters internal frames exceptions raises, keep traceback short, readable, focused actionable (code). See also config_enable_traceback_filtering() config_disable_traceback_filtering(). previously disabled traceback filtering via config_disable_traceback_filtering(), can re-enable via config_enable_traceback_filtering().","code":""},{"path":"https://keras3.posit.co/dev/reference/config_is_traceback_filtering_enabled.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Check if traceback filtering is enabled. — config_is_traceback_filtering_enabled","text":"","code":"config_is_traceback_filtering_enabled()"},{"path":"https://keras3.posit.co/dev/reference/config_is_traceback_filtering_enabled.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Check if traceback filtering is enabled. — config_is_traceback_filtering_enabled","text":"Boolean, TRUE traceback filtering enabled, FALSE otherwise.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/config_max_epochs.html","id":null,"dir":"Reference","previous_headings":"","what":"Configure the default training loop limits. — config_max_epochs","title":"Configure the default training loop limits. — config_max_epochs","text":"helpers control caps built-training loops obey running fit(), evaluate(), predict(). values can also provided via KERAS_MAX_EPOCHS KERAS_MAX_STEPS_PER_EPOCH environment variables quickly constrain run without modifying source code.","code":""},{"path":"https://keras3.posit.co/dev/reference/config_max_epochs.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Configure the default training loop limits. — config_max_epochs","text":"","code":"config_max_epochs()  config_set_max_epochs(max_epochs)  config_max_steps_per_epoch()  config_set_max_steps_per_epoch(max_steps_per_epoch)"},{"path":"https://keras3.posit.co/dev/reference/config_max_epochs.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Configure the default training loop limits. — config_max_epochs","text":"max_epochs Integer upper bound epochs processed built-training loops. Use NULL remove cap. max_steps_per_epoch Integer upper bound steps processed per epoch built-training loops. Use NULL remove cap.","code":""},{"path":"https://keras3.posit.co/dev/reference/config_max_epochs.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Configure the default training loop limits. — config_max_epochs","text":"config_max_epochs() config_max_steps_per_epoch() return current integer limits (NULL cap unset). setter variants return NULL invisibly called side effects.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/config_set_backend.html","id":null,"dir":"Reference","previous_headings":"","what":"Reload the backend (and the Keras package). — config_set_backend","title":"Reload the backend (and the Keras package). — config_set_backend","text":"Reload backend (Keras package).","code":""},{"path":"https://keras3.posit.co/dev/reference/config_set_backend.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Reload the backend (and the Keras package). — config_set_backend","text":"","code":"config_set_backend(backend)"},{"path":"https://keras3.posit.co/dev/reference/config_set_backend.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Reload the backend (and the Keras package). — config_set_backend","text":"backend String","code":""},{"path":"https://keras3.posit.co/dev/reference/config_set_backend.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Reload the backend (and the Keras package). — config_set_backend","text":"Nothing, function called side effect.","code":""},{"path":"https://keras3.posit.co/dev/reference/config_set_backend.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Reload the backend (and the Keras package). — config_set_backend","text":"","code":"config_set_backend(\"jax\")"},{"path":"https://keras3.posit.co/dev/reference/config_set_backend.html","id":"warning","dir":"Reference","previous_headings":"","what":"WARNING","title":"Reload the backend (and the Keras package). — config_set_backend","text":"Using function dangerous done carefully. Changing backend convert type already-instantiated objects. Thus, layers / tensors / etc. already created longer usable without errors. strongly recommended keep around Keras-originated objects instances created calling config_set_backend(). includes function class instance uses Keras functionality. code needs re-executed calling config_set_backend().","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/config_set_dtype_policy.html","id":null,"dir":"Reference","previous_headings":"","what":"Sets the default dtype policy globally. — config_set_dtype_policy","title":"Sets the default dtype policy globally. — config_set_dtype_policy","text":"Sets default dtype policy globally.","code":""},{"path":"https://keras3.posit.co/dev/reference/config_set_dtype_policy.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Sets the default dtype policy globally. — config_set_dtype_policy","text":"","code":"config_set_dtype_policy(policy)"},{"path":"https://keras3.posit.co/dev/reference/config_set_dtype_policy.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Sets the default dtype policy globally. — config_set_dtype_policy","text":"policy string DTypePolicy object.","code":""},{"path":"https://keras3.posit.co/dev/reference/config_set_dtype_policy.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Sets the default dtype policy globally. — config_set_dtype_policy","text":"return value, called side effects.","code":""},{"path":"https://keras3.posit.co/dev/reference/config_set_dtype_policy.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Sets the default dtype policy globally. — config_set_dtype_policy","text":"","code":"config_set_dtype_policy(\"mixed_float16\")"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/config_set_epsilon.html","id":null,"dir":"Reference","previous_headings":"","what":"Set the value of the fuzz factor used in numeric expressions. — config_set_epsilon","title":"Set the value of the fuzz factor used in numeric expressions. — config_set_epsilon","text":"Set value fuzz factor used numeric expressions.","code":""},{"path":"https://keras3.posit.co/dev/reference/config_set_epsilon.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Set the value of the fuzz factor used in numeric expressions. — config_set_epsilon","text":"","code":"config_set_epsilon(value)"},{"path":"https://keras3.posit.co/dev/reference/config_set_epsilon.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Set the value of the fuzz factor used in numeric expressions. — config_set_epsilon","text":"value float. New value epsilon.","code":""},{"path":"https://keras3.posit.co/dev/reference/config_set_epsilon.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Set the value of the fuzz factor used in numeric expressions. — config_set_epsilon","text":"return value, called side effects.","code":""},{"path":"https://keras3.posit.co/dev/reference/config_set_epsilon.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Set the value of the fuzz factor used in numeric expressions. — config_set_epsilon","text":"","code":"config_epsilon() ## [1] 1e-07 config_set_epsilon(1e-5) config_epsilon() ## [1] 1e-05 # Set it back to the default value. config_set_epsilon(1e-7)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/config_set_floatx.html","id":null,"dir":"Reference","previous_headings":"","what":"Set the default float dtype. — config_set_floatx","title":"Set the default float dtype. — config_set_floatx","text":"Set default float dtype.","code":""},{"path":"https://keras3.posit.co/dev/reference/config_set_floatx.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Set the default float dtype. — config_set_floatx","text":"","code":"config_set_floatx(value)"},{"path":"https://keras3.posit.co/dev/reference/config_set_floatx.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Set the default float dtype. — config_set_floatx","text":"value String; 'bfloat16', 'float16', 'float32', 'float64'.","code":""},{"path":"https://keras3.posit.co/dev/reference/config_set_floatx.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Set the default float dtype. — config_set_floatx","text":"return value, called side effects.","code":""},{"path":"https://keras3.posit.co/dev/reference/config_set_floatx.html","id":"note","dir":"Reference","previous_headings":"","what":"Note","title":"Set the default float dtype. — config_set_floatx","text":"recommended set \"float16\" training, likely cause numeric stability issues. Instead, mixed precision, leverages mix float16 float32. can configured calling keras3::keras$mixed_precision$set_dtype_policy('mixed_float16').","code":""},{"path":"https://keras3.posit.co/dev/reference/config_set_floatx.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Set the default float dtype. — config_set_floatx","text":"","code":"config_floatx() ## [1] \"float32\" config_set_floatx('float64') config_floatx() ## [1] \"float64\" # Set it back to float32 config_set_floatx('float32')"},{"path":"https://keras3.posit.co/dev/reference/config_set_floatx.html","id":"raises","dir":"Reference","previous_headings":"","what":"Raises","title":"Set the default float dtype. — config_set_floatx","text":"ValueError: case invalid value.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/config_set_image_data_format.html","id":null,"dir":"Reference","previous_headings":"","what":"Set the value of the image data format convention. — config_set_image_data_format","title":"Set the value of the image data format convention. — config_set_image_data_format","text":"Set value image data format convention.","code":""},{"path":"https://keras3.posit.co/dev/reference/config_set_image_data_format.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Set the value of the image data format convention. — config_set_image_data_format","text":"","code":"config_set_image_data_format(data_format)"},{"path":"https://keras3.posit.co/dev/reference/config_set_image_data_format.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Set the value of the image data format convention. — config_set_image_data_format","text":"data_format string. 'channels_first' 'channels_last'.","code":""},{"path":"https://keras3.posit.co/dev/reference/config_set_image_data_format.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Set the value of the image data format convention. — config_set_image_data_format","text":"return value, called side effects.","code":""},{"path":"https://keras3.posit.co/dev/reference/config_set_image_data_format.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Set the value of the image data format convention. — config_set_image_data_format","text":"","code":"config_image_data_format() ## [1] \"channels_last\" # 'channels_last' keras3::config_set_image_data_format('channels_first') config_image_data_format() ## [1] \"channels_first\" # Set it back to `'channels_last'` keras3::config_set_image_data_format('channels_last')"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/constraint_maxnorm.html","id":null,"dir":"Reference","previous_headings":"","what":"MaxNorm weight constraint. — constraint_maxnorm","title":"MaxNorm weight constraint. — constraint_maxnorm","text":"Constrains weights incident hidden unit norm less equal desired value.","code":""},{"path":"https://keras3.posit.co/dev/reference/constraint_maxnorm.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"MaxNorm weight constraint. — constraint_maxnorm","text":"","code":"constraint_maxnorm(max_value = 2L, axis = 1L)"},{"path":"https://keras3.posit.co/dev/reference/constraint_maxnorm.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"MaxNorm weight constraint. — constraint_maxnorm","text":"max_value maximum norm value incoming weights. axis integer, axis along calculate weight norms. instance, Dense layer weight matrix shape (input_dim, output_dim), set axis 0 constrain weight vector length (input_dim,). Conv2D layer data_format = \"channels_last\", weight tensor shape (rows, cols, input_depth, output_depth), set axis [0, 1, 2] constrain weights filter tensor size (rows, cols, input_depth).","code":""},{"path":"https://keras3.posit.co/dev/reference/constraint_maxnorm.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"MaxNorm weight constraint. — constraint_maxnorm","text":"Constraint instance, callable can passed layer constructors used directly calling tensors.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/constraint_minmaxnorm.html","id":null,"dir":"Reference","previous_headings":"","what":"MinMaxNorm weight constraint. — constraint_minmaxnorm","title":"MinMaxNorm weight constraint. — constraint_minmaxnorm","text":"Constrains weights incident hidden unit norm lower bound upper bound.","code":""},{"path":"https://keras3.posit.co/dev/reference/constraint_minmaxnorm.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"MinMaxNorm weight constraint. — constraint_minmaxnorm","text":"","code":"constraint_minmaxnorm(min_value = 0, max_value = 1, rate = 1, axis = 1L)"},{"path":"https://keras3.posit.co/dev/reference/constraint_minmaxnorm.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"MinMaxNorm weight constraint. — constraint_minmaxnorm","text":"min_value minimum norm incoming weights. max_value maximum norm incoming weights. rate rate enforcing constraint: weights rescaled yield op_clip? (1 - rate) * norm + rate * op_clip(norm, min_value, max_value). Effectively, means rate = 1.0 stands strict enforcement constraint, rate<1.0 means weights rescaled step slowly move towards value inside desired interval. axis integer, axis along calculate weight norms. instance, Dense layer weight matrix shape (input_dim, output_dim), set axis 0 constrain weight vector length (input_dim,). Conv2D layer data_format = \"channels_last\", weight tensor shape (rows, cols, input_depth, output_depth), set axis [0, 1, 2] constrain weights filter tensor size (rows, cols, input_depth).","code":""},{"path":"https://keras3.posit.co/dev/reference/constraint_minmaxnorm.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"MinMaxNorm weight constraint. — constraint_minmaxnorm","text":"Constraint instance, callable can passed layer constructors used directly calling tensors.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/constraint_nonneg.html","id":null,"dir":"Reference","previous_headings":"","what":"Constrains the weights to be non-negative. — constraint_nonneg","title":"Constrains the weights to be non-negative. — constraint_nonneg","text":"Constrains weights non-negative.","code":""},{"path":"https://keras3.posit.co/dev/reference/constraint_nonneg.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Constrains the weights to be non-negative. — constraint_nonneg","text":"","code":"constraint_nonneg()"},{"path":"https://keras3.posit.co/dev/reference/constraint_nonneg.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Constrains the weights to be non-negative. — constraint_nonneg","text":"Constraint instance, callable can passed layer constructors used directly calling tensors.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/constraint_unitnorm.html","id":null,"dir":"Reference","previous_headings":"","what":"Constrains the weights incident to each hidden unit to have unit norm. — constraint_unitnorm","title":"Constrains the weights incident to each hidden unit to have unit norm. — constraint_unitnorm","text":"Constrains weights incident hidden unit unit norm.","code":""},{"path":"https://keras3.posit.co/dev/reference/constraint_unitnorm.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Constrains the weights incident to each hidden unit to have unit norm. — constraint_unitnorm","text":"","code":"constraint_unitnorm(axis = 1L)"},{"path":"https://keras3.posit.co/dev/reference/constraint_unitnorm.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Constrains the weights incident to each hidden unit to have unit norm. — constraint_unitnorm","text":"axis integer, axis along calculate weight norms. instance, Dense layer weight matrix shape (input_dim, output_dim), set axis 0 constrain weight vector length (input_dim,). Conv2D layer data_format = \"channels_last\", weight tensor shape (rows, cols, input_depth, output_depth), set axis [0, 1, 2] constrain weights filter tensor size (rows, cols, input_depth).","code":""},{"path":"https://keras3.posit.co/dev/reference/constraint_unitnorm.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Constrains the weights incident to each hidden unit to have unit norm. — constraint_unitnorm","text":"Constraint instance, callable can passed layer constructors used directly calling tensors.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/count_params.html","id":null,"dir":"Reference","previous_headings":"","what":"Count the total number of scalars composing the weights. — count_params","title":"Count the total number of scalars composing the weights. — count_params","text":"Count total number scalars composing weights.","code":""},{"path":"https://keras3.posit.co/dev/reference/count_params.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Count the total number of scalars composing the weights. — count_params","text":"","code":"count_params(object)"},{"path":"https://keras3.posit.co/dev/reference/count_params.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Count the total number of scalars composing the weights. — count_params","text":"object Layer model object","code":""},{"path":"https://keras3.posit.co/dev/reference/count_params.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Count the total number of scalars composing the weights. — count_params","text":"integer count","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/custom_metric.html","id":null,"dir":"Reference","previous_headings":"","what":"Custom metric function — custom_metric","title":"Custom metric function — custom_metric","text":"Custom metric function","code":""},{"path":"https://keras3.posit.co/dev/reference/custom_metric.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Custom metric function — custom_metric","text":"","code":"custom_metric(name, metric_fn)"},{"path":"https://keras3.posit.co/dev/reference/custom_metric.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Custom metric function — custom_metric","text":"name name used show training progress output metric_fn R function signature function(y_true, y_pred) accepts tensors.","code":""},{"path":"https://keras3.posit.co/dev/reference/custom_metric.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Custom metric function — custom_metric","text":"callable function __name__ attribute.","code":""},{"path":"https://keras3.posit.co/dev/reference/custom_metric.html","id":"details","dir":"Reference","previous_headings":"","what":"Details","title":"Custom metric function — custom_metric","text":"can provide arbitrary R function custom metric. Note y_true y_pred parameters tensors, computations use op_* tensor functions. Use custom_metric() function define custom metric. Note name ('mean_pred') provided custom metric function: name used within training progress output. want save load model custom metrics, also call register_keras_serializable(), specify metric call load_model(). example: load_model(\"my_model.keras\", c('mean_pred' = metric_mean_pred)). Alternatively, can wrap code call with_custom_object_scope() allow refer metric name just like built keras metrics. Alternative ways supplying custom metrics: custom_metric(): Arbitrary R function. metric_mean_wrapper(): Wrap arbitrary R function Metric instance. Create custom Metric() subclass.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/dataset_boston_housing.html","id":null,"dir":"Reference","previous_headings":"","what":"Boston housing price regression dataset — dataset_boston_housing","title":"Boston housing price regression dataset — dataset_boston_housing","text":"Dataset taken StatLib library maintained Carnegie Mellon University.","code":""},{"path":"https://keras3.posit.co/dev/reference/dataset_boston_housing.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Boston housing price regression dataset — dataset_boston_housing","text":"","code":"dataset_boston_housing(   path = \"boston_housing.npz\",   test_split = 0.2,   seed = 113L )"},{"path":"https://keras3.posit.co/dev/reference/dataset_boston_housing.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Boston housing price regression dataset — dataset_boston_housing","text":"path Path cache dataset locally (relative ~/.keras/datasets). test_split fraction data reserve test set. seed Random seed shuffling data computing test split.","code":""},{"path":"https://keras3.posit.co/dev/reference/dataset_boston_housing.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Boston housing price regression dataset — dataset_boston_housing","text":"Lists training test data: train$x, train$y, test$x, test$y. Samples contain 13 attributes houses different locations around Boston suburbs late 1970s. Targets median values houses location (k$).","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/dataset_california_housing.html","id":null,"dir":"Reference","previous_headings":"","what":"Loads the California Housing dataset. — dataset_california_housing","title":"Loads the California Housing dataset. — dataset_california_housing","text":"dataset obtained StatLib repository. continuous regression dataset 20,640 samples 8 features . target variable scalar: median house value California districts, dollars. 8 input features following: MedInc: median income block group HouseAge: median house age block group AveRooms: average number rooms per household AveBedrms: average number bedrooms per household Population: block group population AveOccup: average number household members Latitude: block group latitude Longitude: block group longitude dataset derived 1990 U.S. census, using one row per census block group. block group smallest geographical unit U.S. Census Bureau publishes sample data (block group typically population 600 3,000 people). household group people residing within home. Since average number rooms bedrooms dataset provided per household, columns may take surprisingly large values block groups households many empty houses, vacation resorts.","code":""},{"path":"https://keras3.posit.co/dev/reference/dataset_california_housing.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Loads the California Housing dataset. — dataset_california_housing","text":"","code":"dataset_california_housing(   version = \"large\",   path = \"california_housing.npz\",   test_split = 0.2,   seed = 113L )"},{"path":"https://keras3.posit.co/dev/reference/dataset_california_housing.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Loads the California Housing dataset. — dataset_california_housing","text":"version \"small\" \"large\". small version contains 600 samples, large version contains 20,640 samples. purpose small version serve approximate replacement deprecated boston_housing dataset. path path cache dataset locally (relative Sys.getenv(\"KERAS_HOME\")). test_split fraction data reserve test set. seed Random seed shuffling data computing test split.","code":""},{"path":"https://keras3.posit.co/dev/reference/dataset_california_housing.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Loads the California Housing dataset. — dataset_california_housing","text":"Nested list arrays: (x_train, y_train), (x_test, y_test). x_train, x_test: arrays shape (num_samples, 8) containing either training samples (x_train), test samples (y_train). y_train, y_test: arrays shape (num_samples) containing target scalars. targets float scalars typically 25,000 500,000 represent home prices dollars.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/dataset_cifar10.html","id":null,"dir":"Reference","previous_headings":"","what":"CIFAR10 small image classification — dataset_cifar10","title":"CIFAR10 small image classification — dataset_cifar10","text":"Dataset 50,000 32x32 color training images, labeled 10 categories, 10,000 test images.","code":""},{"path":"https://keras3.posit.co/dev/reference/dataset_cifar10.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"CIFAR10 small image classification — dataset_cifar10","text":"","code":"dataset_cifar10()"},{"path":"https://keras3.posit.co/dev/reference/dataset_cifar10.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"CIFAR10 small image classification — dataset_cifar10","text":"Lists training test data: train$x, train$y, test$x, test$y. x data array RGB image data shape (num_samples, 3, 32, 32). y data array category labels (integers range 0-9) shape (num_samples).","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/dataset_cifar100.html","id":null,"dir":"Reference","previous_headings":"","what":"CIFAR100 small image classification — dataset_cifar100","title":"CIFAR100 small image classification — dataset_cifar100","text":"Dataset 50,000 32x32 color training images, labeled 100 categories, 10,000 test images.","code":""},{"path":"https://keras3.posit.co/dev/reference/dataset_cifar100.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"CIFAR100 small image classification — dataset_cifar100","text":"","code":"dataset_cifar100(label_mode = c(\"fine\", \"coarse\"))"},{"path":"https://keras3.posit.co/dev/reference/dataset_cifar100.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"CIFAR100 small image classification — dataset_cifar100","text":"label_mode one \"fine\", \"coarse\".","code":""},{"path":"https://keras3.posit.co/dev/reference/dataset_cifar100.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"CIFAR100 small image classification — dataset_cifar100","text":"Lists training test data: train$x, train$y, test$x, test$y. x data array RGB image data shape (num_samples, 3, 32, 32). y data array category labels shape (num_samples).","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/dataset_fashion_mnist.html","id":null,"dir":"Reference","previous_headings":"","what":"Fashion-MNIST database of fashion articles — dataset_fashion_mnist","title":"Fashion-MNIST database of fashion articles — dataset_fashion_mnist","text":"Dataset 60,000 28x28 grayscale images 10 fashion article classes, along test set 10,000 images. dataset can used drop-replacement MNIST. class labels encoded integers 0-9 correspond T-shirt/top, Trouser, Pullover, Dress, Coat, Sandal, Shirt,","code":""},{"path":"https://keras3.posit.co/dev/reference/dataset_fashion_mnist.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Fashion-MNIST database of fashion articles — dataset_fashion_mnist","text":"","code":"dataset_fashion_mnist()"},{"path":"https://keras3.posit.co/dev/reference/dataset_fashion_mnist.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Fashion-MNIST database of fashion articles — dataset_fashion_mnist","text":"Lists training test data: train$x, train$y, test$x, test$y, x array grayscale image data shape (num_samples, 28, 28) y array article labels (integers range 0-9) shape (num_samples).","code":""},{"path":"https://keras3.posit.co/dev/reference/dataset_fashion_mnist.html","id":"details","dir":"Reference","previous_headings":"","what":"Details","title":"Fashion-MNIST database of fashion articles — dataset_fashion_mnist","text":"Dataset 60,000 28x28 grayscale images 10 fashion categories, along test set 10,000 images. dataset can used drop-replacement MNIST. class labels : 0 - T-shirt/top 1 - Trouser 2 - Pullover 3 - Dress 4 - Coat 5 - Sandal 6 - Shirt 7 - Sneaker 8 - Bag 9 - Ankle boot","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/dataset_imdb.html","id":null,"dir":"Reference","previous_headings":"","what":"IMDB Movie reviews sentiment classification — dataset_imdb","title":"IMDB Movie reviews sentiment classification — dataset_imdb","text":"Dataset 25,000 movies reviews IMDB, labeled sentiment (positive/negative). Reviews preprocessed, review encoded sequence word indexes (integers). convenience, words indexed overall frequency dataset, instance integer \"3\" encodes 3rd frequent word data. allows quick filtering operations : \"consider top 10,000 common words, eliminate top 20 common words\".","code":""},{"path":"https://keras3.posit.co/dev/reference/dataset_imdb.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"IMDB Movie reviews sentiment classification — dataset_imdb","text":"","code":"dataset_imdb(   path = \"imdb.npz\",   num_words = NULL,   skip_top = 0L,   maxlen = NULL,   seed = 113L,   start_char = 1L,   oov_char = 2L,   index_from = 3L )  dataset_imdb_word_index(path = \"imdb_word_index.json\")"},{"path":"https://keras3.posit.co/dev/reference/dataset_imdb.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"IMDB Movie reviews sentiment classification — dataset_imdb","text":"path cache data (relative ~/.keras/dataset). num_words Max number words include. Words ranked often occur (training set) frequent words kept skip_top Skip top N frequently occuring words (may informative). maxlen sequences longer filtered . seed random seed sample shuffling. start_char start sequence marked character. Set 1 0 usually padding character. oov_char Words cut num_words skip_top limit replaced character. index_from Index actual words index higher.","code":""},{"path":"https://keras3.posit.co/dev/reference/dataset_imdb.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"IMDB Movie reviews sentiment classification — dataset_imdb","text":"Lists training test data: train$x, train$y, test$x, test$y. x data includes integer sequences. num_words argument specific, maximum possible index value num_words-1. maxlen argument specified, largest possible sequence length maxlen. y data includes set integer labels (0 1). dataset_imdb_word_index() function returns list names words values integer.","code":""},{"path":"https://keras3.posit.co/dev/reference/dataset_imdb.html","id":"details","dir":"Reference","previous_headings":"","what":"Details","title":"IMDB Movie reviews sentiment classification — dataset_imdb","text":"convention, \"0\" stand specific word, instead used encode unknown word.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/dataset_mnist.html","id":null,"dir":"Reference","previous_headings":"","what":"MNIST database of handwritten digits — dataset_mnist","title":"MNIST database of handwritten digits — dataset_mnist","text":"Dataset 60,000 28x28 grayscale images 10 digits, along test set 10,000 images.","code":""},{"path":"https://keras3.posit.co/dev/reference/dataset_mnist.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"MNIST database of handwritten digits — dataset_mnist","text":"","code":"dataset_mnist(path = \"mnist.npz\")"},{"path":"https://keras3.posit.co/dev/reference/dataset_mnist.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"MNIST database of handwritten digits — dataset_mnist","text":"path Path cache dataset locally (relative ~/.keras/datasets).","code":""},{"path":"https://keras3.posit.co/dev/reference/dataset_mnist.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"MNIST database of handwritten digits — dataset_mnist","text":"Lists training test data: train$x, train$y, test$x, test$y, x array grayscale image data shape (num_samples, 28, 28) y array digit labels (integers range 0-9) shape (num_samples).","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/dataset_reuters.html","id":null,"dir":"Reference","previous_headings":"","what":"Reuters newswire topics classification — dataset_reuters","title":"Reuters newswire topics classification — dataset_reuters","text":"Dataset 11,228 newswires Reuters, labeled 46 topics. dataset_imdb() , wire encoded sequence word indexes (conventions).","code":""},{"path":"https://keras3.posit.co/dev/reference/dataset_reuters.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Reuters newswire topics classification — dataset_reuters","text":"","code":"dataset_reuters(   path = \"reuters.npz\",   num_words = NULL,   skip_top = 0L,   maxlen = NULL,   test_split = 0.2,   seed = 113L,   start_char = 1L,   oov_char = 2L,   index_from = 3L )  dataset_reuters_word_index(path = \"reuters_word_index.pkl\")"},{"path":"https://keras3.posit.co/dev/reference/dataset_reuters.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Reuters newswire topics classification — dataset_reuters","text":"path cache data (relative ~/.keras/dataset). num_words Max number words include. Words ranked often occur (training set) frequent words kept skip_top Skip top N frequently occuring words (may informative). maxlen Truncate sequences length. test_split Fraction dataset used test data. seed Random seed sample shuffling. start_char start sequence marked character. Set 1 0 usually padding character. oov_char words cut num_words skip_top limit replaced character. index_from index actual words index higher.","code":""},{"path":"https://keras3.posit.co/dev/reference/dataset_reuters.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Reuters newswire topics classification — dataset_reuters","text":"Lists training test data: train$x, train$y, test$x, test$y format dataset_imdb(). dataset_reuters_word_index() function returns list names words values integer. e.g. word_index[[\"giraffe\"]] might return 1234.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/deserialize_keras_object.html","id":null,"dir":"Reference","previous_headings":"","what":"Retrieve the object by deserializing the config dict. — deserialize_keras_object","title":"Retrieve the object by deserializing the config dict. — deserialize_keras_object","text":"config dict Python dictionary consists set key-value pairs, represents Keras object, Optimizer, Layer, Metrics, etc. saving loading library uses following keys record information Keras object: class_name: String. name class, exactly defined source code, \"LossesContainer\". config: Named List. Library-defined user-defined key-value pairs store configuration object, obtained object$get_config(). module: String. path python module. Built-Keras classes expect prefix keras. registered_name: String. key class registered via register_keras_serializable(package, name) API. key format '{package}>{name}', package name arguments passed register_keras_serializable(). name provided, uses class name. registered_name successfully resolves class (registered), class_name config values config dict used. registered_name used non-built-classes. example, following config list represents built-Adam optimizer relevant config:     class exported Keras namespace, library tracks module class_name. example:   following config represents user-customized MeanSquaredError loss:","code":"config <- list(   class_name = \"Adam\",   config = list(     amsgrad = FALSE,     beta_1 = 0.8999999761581421,     beta_2 = 0.9990000128746033,     epsilon = 1e-07,     learning_rate = 0.0010000000474974513,     name = \"Adam\"   ),   module = \"keras.optimizers\",   registered_name = NULL ) # Returns an `Adam` instance identical to the original one. deserialize_keras_object(config) ## <keras.src.optimizers.adam.Adam object at 0x0> config <- list(   class_name = \"MetricsList\",   config =  list(     ...   ),   module = \"keras.trainers.compile_utils\",   registered_name = \"MetricsList\" )  # Returns a `MetricsList` instance identical to the original one. deserialize_keras_object(config) # define a custom object loss_modified_mse <- Loss(   \"ModifiedMeanSquaredError\",   inherit = loss_mean_squared_error)  # register the custom object register_keras_serializable(loss_modified_mse)  # confirm object is registered get_custom_objects() ## $`keras3>ModifiedMeanSquaredError` ## <class '<r-namespace:keras3>.ModifiedMeanSquaredError'> ##  signature: ( ##    reduction='sum_over_batch_size', ##    name='mean_squared_error', ##    dtype=None ## ) get_registered_name(loss_modified_mse) ## [1] \"keras3>ModifiedMeanSquaredError\" # now custom object instances can be serialized full_config <- serialize_keras_object(loss_modified_mse())  # the `config` arguments will be passed to loss_modified_mse() str(full_config) ## List of 4 ##  $ module         : chr \"<r-namespace:keras3>\" ##  $ class_name     : chr \"ModifiedMeanSquaredError\" ##  $ config         :List of 2 ##   ..$ name     : chr \"mean_squared_error\" ##   ..$ reduction: chr \"sum_over_batch_size\" ##  $ registered_name: chr \"keras3>ModifiedMeanSquaredError\" # and custom object instances can be deserialized deserialize_keras_object(full_config) ## <LossFunctionWrapper(<function mean_squared_error at 0x0>, kwargs={})> ##  signature: (y_true, y_pred, sample_weight=None) # Returns the `ModifiedMeanSquaredError` object"},{"path":"https://keras3.posit.co/dev/reference/deserialize_keras_object.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Retrieve the object by deserializing the config dict. — deserialize_keras_object","text":"","code":"deserialize_keras_object(config, custom_objects = NULL, safe_mode = TRUE, ...)"},{"path":"https://keras3.posit.co/dev/reference/deserialize_keras_object.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Retrieve the object by deserializing the config dict. — deserialize_keras_object","text":"config Named list describing object. custom_objects Named list containing mapping custom object names corresponding classes functions. safe_mode Boolean, whether disallow unsafe lambda deserialization. safe_mode=FALSE, loading object potential trigger arbitrary code execution. argument applicable Keras v3 model format. Defaults TRUE. ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/deserialize_keras_object.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Retrieve the object by deserializing the config dict. — deserialize_keras_object","text":"object described config dictionary.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/evaluate.keras.src.models.model.Model.html","id":null,"dir":"Reference","previous_headings":"","what":"Evaluate a Keras Model — evaluate.keras.src.models.model.Model","title":"Evaluate a Keras Model — evaluate.keras.src.models.model.Model","text":"functions returns loss value metrics values model test mode. Computation done batches (see batch_size arg.)","code":""},{"path":"https://keras3.posit.co/dev/reference/evaluate.keras.src.models.model.Model.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Evaluate a Keras Model — evaluate.keras.src.models.model.Model","text":"","code":"# S3 method for class 'keras.src.models.model.Model' evaluate(   object,   x = NULL,   y = NULL,   ...,   batch_size = NULL,   verbose = getOption(\"keras.verbose\", default = \"auto\"),   sample_weight = NULL,   steps = NULL,   callbacks = NULL )"},{"path":"https://keras3.posit.co/dev/reference/evaluate.keras.src.models.model.Model.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Evaluate a Keras Model — evaluate.keras.src.models.model.Model","text":"object Keras model object x Input data. can : R array (array-like), list arrays (case model multiple inputs). backend-native tensor, list tensors (case model multiple inputs). named list mapping input names corresponding array/tensors, model named inputs. tf.data.Dataset. return tuple either (inputs, targets) (inputs, targets, sample_weights). generator returning (inputs, targets) (inputs, targets, sample_weights). y Target data. Like input data x, either R array(s) backend-native tensor(s). x tf.data.Dataset generator function, y specified (since targets obtained iterator/dataset). ... forward/backward compatability. batch_size Integer NULL. Number samples per batch computation. unspecified, batch_size default 32. specify batch_size data form tf dataset generator (since generate batches). verbose \"auto\", 0, 1, 2. Verbosity mode. 0 = silent, 1 = progress bar, 2 = single line. \"auto\" becomes 1 cases, 2 knitr render running distributed training server. Note progress bar particularly useful logged file, verbose=2 recommended running interactively (e.g. production environment). Defaults \"auto\". sample_weight Optional array tensor weights training samples, used weighting loss function (training ). can either pass flat (1D) array tensor length input samples (1:1 mapping weights samples), case temporal data, can pass 2D array tensor shape (samples, sequence_length) apply different weight every timestep every sample. argument supported x tf.data.Dataset, Python generator function. Instead, provide sample_weights third element x. Note sample weighting apply metrics specified via metrics argument compile(). apply sample weighting metrics, can specify via weighted_metrics compile() instead. steps Integer NULL. Total number steps (batches samples) declaring evaluation round finished. Ignored default value NULL. x tf.data.Dataset steps NULL, evaluation run dataset exhausted. case infinitely repeating dataset, run indefinitely. callbacks List Callback instances. List callbacks apply evaluation.","code":""},{"path":"https://keras3.posit.co/dev/reference/evaluate.keras.src.models.model.Model.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Evaluate a Keras Model — evaluate.keras.src.models.model.Model","text":"Scalar test loss (model single output metrics) list scalars (model multiple outputs /metrics). attribute model$metrics_names give display labels scalar outputs.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/export_savedmodel.keras.src.models.model.Model.html","id":null,"dir":"Reference","previous_headings":"","what":"Export the model as an artifact for inference. — export_savedmodel.keras.src.models.model.Model","title":"Export the model as an artifact for inference. — export_savedmodel.keras.src.models.model.Model","text":"(e.g. via TF-Serving). Note: can currently used TensorFlow JAX backends. method lets export model lightweight SavedModel artifact contains model's forward pass (call() method) can served via e.g. TF-Serving. forward pass registered name serve() (see example ). original code model (including custom layers may used) longer necessary reload artifact – entirely standalone. Note: feature currently supported TensorFlow, JAX Torch backends. Note: aware exported artifact may contain information local file system using format=\"onnx\", verbose=TRUE Torch backend.","code":""},{"path":"https://keras3.posit.co/dev/reference/export_savedmodel.keras.src.models.model.Model.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Export the model as an artifact for inference. — export_savedmodel.keras.src.models.model.Model","text":"","code":"# S3 method for class 'keras.src.models.model.Model' export_savedmodel(   object,   export_dir_base,   ...,   format = \"tf_saved_model\",   verbose = NULL,   input_signature = NULL )"},{"path":"https://keras3.posit.co/dev/reference/export_savedmodel.keras.src.models.model.Model.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Export the model as an artifact for inference. — export_savedmodel.keras.src.models.model.Model","text":"object keras model. export_dir_base string, file path save artifact. ... Additional keyword arguments: Specific JAX backend format=\"tf_saved_model\": is_static: Optional bool. Indicates whether fn static. Set FALSE fn involves state updates (e.g., RNG seeds counters). jax2tf_kwargs: Optional dict. Arguments jax2tf.convert. See documentation jax2tf.convert. native_serialization polymorphic_shapes provided, automatically computed. format string. export format. Supported values: \"tf_saved_model\" \"onnx\".  Defaults \"tf_saved_model\". verbose Bool. Whether print variables exported model. Defaults NULL, uses default value set different backends formats. input_signature Optional. Specifies shape dtype model inputs. Can structure keras.InputSpec, tf.TensorSpec, backend.KerasTensor, backend tensor. provided, automatically computed. Defaults NULL.","code":""},{"path":"https://keras3.posit.co/dev/reference/export_savedmodel.keras.src.models.model.Model.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Export the model as an artifact for inference. — export_savedmodel.keras.src.models.model.Model","text":"called primarily side effect exporting object. first argument, object also returned, invisibly, enable usage pipe.","code":""},{"path":"https://keras3.posit.co/dev/reference/export_savedmodel.keras.src.models.model.Model.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Export the model as an artifact for inference. — export_savedmodel.keras.src.models.model.Model","text":"export ONNX inference.","code":"# Create the artifact model |> tensorflow::export_savedmodel(\"path/to/location\")  # Later, in a different process/environment... library(tensorflow) reloaded_artifact <- tf$saved_model$load(\"path/to/location\") predictions <- reloaded_artifact$serve(input_data)  # see tfdeploy::serve_savedmodel() for serving a model over a local web api. # Export the model as a ONNX artifact model |> export_savedmodel(\"path/to/location\", format = \"onnx\")  # Load the artifact in a different process/environment onnxruntime <- reticulate::import(\"onnxruntime\") ort_session <- onnxruntime$InferenceSession(\"path/to/location\") input_data <- list(....) names(input_data) <- sapply(ort_session$get_inputs(), `[[`, \"name\") predictions <- ort_session$run(NULL, input_data)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/fit.keras.src.models.model.Model.html","id":null,"dir":"Reference","previous_headings":"","what":"Train a model for a fixed number of epochs (dataset iterations). — fit.keras.src.models.model.Model","title":"Train a model for a fixed number of epochs (dataset iterations). — fit.keras.src.models.model.Model","text":"Train model fixed number epochs (dataset iterations).","code":""},{"path":"https://keras3.posit.co/dev/reference/fit.keras.src.models.model.Model.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Train a model for a fixed number of epochs (dataset iterations). — fit.keras.src.models.model.Model","text":"","code":"# S3 method for class 'keras.src.models.model.Model' fit(   object,   x = NULL,   y = NULL,   ...,   batch_size = NULL,   epochs = 1L,   callbacks = NULL,   validation_split = 0,   validation_data = NULL,   shuffle = TRUE,   class_weight = NULL,   sample_weight = NULL,   initial_epoch = 1L,   steps_per_epoch = NULL,   validation_steps = NULL,   validation_batch_size = NULL,   validation_freq = 1L,   verbose = getOption(\"keras.verbose\", default = \"auto\"),   view_metrics = getOption(\"keras.view_metrics\", default = \"auto\") )"},{"path":"https://keras3.posit.co/dev/reference/fit.keras.src.models.model.Model.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Train a model for a fixed number of epochs (dataset iterations). — fit.keras.src.models.model.Model","text":"object Keras model object x Input data. can : array (array-like), list arrays (case model multiple inputs). backend-native tensor, list tensors (case model multiple inputs). named list mapping input names corresponding array/tensors, model named inputs. tf.data.Dataset. return tuple either (inputs, targets) (inputs, targets, sample_weights). generator returning (inputs, targets) (inputs, targets, sample_weights). y Target data. Like input data x, can either array(s) backend-native tensor(s). x TF Dataset generator, y specified (since targets obtained x). ... Additional arguments passed model fit() method. batch_size Integer NULL. Number samples per gradient update. unspecified, batch_size default 32. specify batch_size data form TF Datasets generators, (since generate batches). epochs Integer. Number epochs train model. epoch iteration entire x y data provided (unless steps_per_epoch flag set something NULL). Note conjunction initial_epoch, epochs understood \"final epoch\". model trained number iterations given epochs, merely epoch index epochs reached. callbacks List Callback() instances. List callbacks apply training. See callback_*. validation_split Fraction training data used validation data. model set apart fraction training data, train , evaluate loss model metrics data end epoch. validation data selected last samples x y data provided, shuffling. argument supported x y made arrays tensors. validation_data validation_split provided, validation_data override validation_split. validation_data Data evaluate loss model metrics end epoch. model trained data. Thus, note fact validation loss data provided using validation_split validation_data affected regularization layers like noise dropout. validation_data override validation_split. can : tuple (x_val, y_val) arrays tensors. tuple (x_val, y_val, val_sample_weights) arrays. generator returning (inputs, targets) (inputs, targets, sample_weights). shuffle Boolean, whether shuffle training data epoch. argument ignored x generator TF Dataset. class_weight Optional named list mapping class indices (integers, 0-based) weight (float) value, used weighting loss function (training ). can useful tell model \"pay attention\" samples -represented class. class_weight specified targets rank 2 greater, either y must one-hot encoded, explicit final dimension 1 must included sparse class labels. sample_weight Optional array tensor weights training samples, used weighting loss function (training ). can either pass flat (1D) array tensor length input samples (1:1 mapping weights samples), case temporal data, can pass 2D array tensor shape (samples, sequence_length) apply different weight every timestep every sample. argument supported x tf.data.Dataset, Python generator function. Instead, provide sample_weights third element x. Note sample weighting apply metrics specified via metrics argument compile(). apply sample weighting metrics, can specify via weighted_metrics compile() instead. initial_epoch Integer. Epoch start training (useful resuming previous training run). steps_per_epoch Integer NULL. Total number steps (batches samples) declaring one epoch finished starting next epoch. training input tensors arrays, default NULL means value used number samples dataset divided batch size, 1 determined. x tf.data.Dataset Python generator function, epoch run input dataset exhausted. passing infinitely repeating dataset, must specify steps_per_epoch argument, otherwise training run indefinitely. validation_steps relevant validation_data provided. Total number steps (batches samples) draw stopping performing validation end every epoch. validation_steps NULL, validation run validation_data dataset exhausted. case infinitely repeating dataset, run indefinitely. validation_steps specified part dataset consumed, evaluation start beginning dataset epoch. ensures validation samples used every time. validation_batch_size Integer NULL. Number samples per validation batch. unspecified, default batch_size. specify validation_batch_size data form TF Datasets generator instances (since generate batches). validation_freq relevant validation data provided. Specifies many training epochs run new validation run performed, e.g. validation_freq=2 runs validation every 2 epochs. verbose \"auto\", 0, 1, 2. Verbosity mode. 0 = silent, 1 = progress bar, 2 = one line per epoch. \"auto\" becomes 1 cases, 2 knitr render running distributed training server. Note progress bar particularly useful logged file, verbose=2 recommended running interactively (e.g., production environment). Defaults \"auto\". view_metrics View realtime plot training metrics (epoch). default (\"auto\") display plot running within RStudio, metrics specified model compile(), epochs > 1 verbose > 0. Set global options(keras.view_metrics = ) option establish different default.","code":""},{"path":"https://keras3.posit.co/dev/reference/fit.keras.src.models.model.Model.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Train a model for a fixed number of epochs (dataset iterations). — fit.keras.src.models.model.Model","text":"keras_training_history object, named list: list(params = <params>, metrics = <metrics>\"), S3 methods print(), plot(), .data.frame(). metrics field record training loss values metrics values successive epochs, well validation loss values validation metrics values (applicable).","code":""},{"path":"https://keras3.posit.co/dev/reference/fit.keras.src.models.model.Model.html","id":"details","dir":"Reference","previous_headings":"","what":"Details","title":"Train a model for a fixed number of epochs (dataset iterations). — fit.keras.src.models.model.Model","text":"Unpacking behavior iterator-like inputs: common pattern pass iterator like object tf.data.Dataset generator fit(), fact yield features (x) optionally targets (y) sample weights (sample_weight). Keras requires output iterator-likes unambiguous. iterator return tuple() length 1, 2, 3, optional second third elements used y sample_weight respectively. type provided wrapped length-one tuple(), effectively treating everything x. yielding named lists, still adhere top-level tuple structure, e.g. tuple(list(x0 = x0, x = x1), y). Keras attempt separate features, targets, weights keys single dict.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/freeze_weights.html","id":null,"dir":"Reference","previous_headings":"","what":"Freeze and unfreeze weights — freeze_weights","title":"Freeze and unfreeze weights — freeze_weights","text":"Freeze weights model layer longer trainable.","code":""},{"path":"https://keras3.posit.co/dev/reference/freeze_weights.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Freeze and unfreeze weights — freeze_weights","text":"","code":"freeze_weights(object, from = NULL, to = NULL, which = NULL)  unfreeze_weights(object, from = NULL, to = NULL, which = NULL)"},{"path":"https://keras3.posit.co/dev/reference/freeze_weights.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Freeze and unfreeze weights — freeze_weights","text":"object Keras model layer object Layer instance, layer name, layer index within model Layer instance, layer name, layer index within model layer names, integer positions, layers, logical vector (length(object$layers)), function returning logical vector.","code":""},{"path":"https://keras3.posit.co/dev/reference/freeze_weights.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Freeze and unfreeze weights — freeze_weights","text":"input object frozen weights returned, invisibly. Note, object modified place, return value provided make usage pipe convenient.","code":""},{"path":"https://keras3.posit.co/dev/reference/freeze_weights.html","id":"note","dir":"Reference","previous_headings":"","what":"Note","title":"Freeze and unfreeze weights — freeze_weights","text":"layer arguments inclusive. applied model, freeze unfreeze global operation layers model (.e. layers within specified range set opposite value, e.g. unfrozen call freeze). Models must compiled weights frozen unfrozen.","code":""},{"path":"https://keras3.posit.co/dev/reference/freeze_weights.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Freeze and unfreeze weights — freeze_weights","text":"","code":"# instantiate a VGG16 model conv_base <- application_vgg16(   weights = \"imagenet\",   include_top = FALSE,   input_shape = c(150, 150, 3) )  # freeze it's weights freeze_weights(conv_base)  # Note the \"Trainable\" column conv_base ## Model: \"vgg16\" ## +-----------------------------+-----------------------+------------+-------+ ## | Layer (type)                | Output Shape          |    Param # | Trai… | ## +=============================+=======================+============+=======+ ## | input_layer (InputLayer)    | (None, 150, 150, 3)   |          0 |   -   | ## +-----------------------------+-----------------------+------------+-------+ ## | block1_conv1 (Conv2D)       | (None, 150, 150, 64)  |      1,792 |   N   | ## +-----------------------------+-----------------------+------------+-------+ ## | block1_conv2 (Conv2D)       | (None, 150, 150, 64)  |     36,928 |   N   | ## +-----------------------------+-----------------------+------------+-------+ ## | block1_pool (MaxPooling2D)  | (None, 75, 75, 64)    |          0 |   -   | ## +-----------------------------+-----------------------+------------+-------+ ## | block2_conv1 (Conv2D)       | (None, 75, 75, 128)   |     73,856 |   N   | ## +-----------------------------+-----------------------+------------+-------+ ## | block2_conv2 (Conv2D)       | (None, 75, 75, 128)   |    147,584 |   N   | ## +-----------------------------+-----------------------+------------+-------+ ## | block2_pool (MaxPooling2D)  | (None, 37, 37, 128)   |          0 |   -   | ## +-----------------------------+-----------------------+------------+-------+ ## | block3_conv1 (Conv2D)       | (None, 37, 37, 256)   |    295,168 |   N   | ## +-----------------------------+-----------------------+------------+-------+ ## | block3_conv2 (Conv2D)       | (None, 37, 37, 256)   |    590,080 |   N   | ## +-----------------------------+-----------------------+------------+-------+ ## | block3_conv3 (Conv2D)       | (None, 37, 37, 256)   |    590,080 |   N   | ## +-----------------------------+-----------------------+------------+-------+ ## | block3_pool (MaxPooling2D)  | (None, 18, 18, 256)   |          0 |   -   | ## +-----------------------------+-----------------------+------------+-------+ ## | block4_conv1 (Conv2D)       | (None, 18, 18, 512)   |  1,180,160 |   N   | ## +-----------------------------+-----------------------+------------+-------+ ## | block4_conv2 (Conv2D)       | (None, 18, 18, 512)   |  2,359,808 |   N   | ## +-----------------------------+-----------------------+------------+-------+ ## | block4_conv3 (Conv2D)       | (None, 18, 18, 512)   |  2,359,808 |   N   | ## +-----------------------------+-----------------------+------------+-------+ ## | block4_pool (MaxPooling2D)  | (None, 9, 9, 512)     |          0 |   -   | ## +-----------------------------+-----------------------+------------+-------+ ## | block5_conv1 (Conv2D)       | (None, 9, 9, 512)     |  2,359,808 |   N   | ## +-----------------------------+-----------------------+------------+-------+ ## | block5_conv2 (Conv2D)       | (None, 9, 9, 512)     |  2,359,808 |   N   | ## +-----------------------------+-----------------------+------------+-------+ ## | block5_conv3 (Conv2D)       | (None, 9, 9, 512)     |  2,359,808 |   N   | ## +-----------------------------+-----------------------+------------+-------+ ## | block5_pool (MaxPooling2D)  | (None, 4, 4, 512)     |          0 |   -   | ## +-----------------------------+-----------------------+------------+-------+ ##  Total params: 14,714,688 (56.13 MB) ##  Trainable params: 0 (0.00 B) ##  Non-trainable params: 14,714,688 (56.13 MB) # create a composite model that includes the base + more layers model <- keras_model_sequential(input_batch_shape = shape(conv_base$input)) |>   conv_base() |>   layer_flatten() |>   layer_dense(units = 256, activation = \"relu\") |>   layer_dense(units = 1, activation = \"sigmoid\")  # compile model |> compile(   loss = \"binary_crossentropy\",   optimizer = optimizer_rmsprop(learning_rate = 2e-5),   metrics = c(\"accuracy\") )  model ## Model: \"sequential\" ## +-----------------------------+-----------------------+------------+-------+ ## | Layer (type)                | Output Shape          |    Param # | Trai… | ## +=============================+=======================+============+=======+ ## | vgg16 (Functional)          | (None, 4, 4, 512)     | 14,714,688 |   N   | ## +-----------------------------+-----------------------+------------+-------+ ## | flatten (Flatten)           | (None, 8192)          |          0 |   -   | ## +-----------------------------+-----------------------+------------+-------+ ## | dense (Dense)               | (None, 256)           |  2,097,408 |   Y   | ## +-----------------------------+-----------------------+------------+-------+ ## | dense_1 (Dense)             | (None, 1)             |        257 |   Y   | ## +-----------------------------+-----------------------+------------+-------+ ##  Total params: 16,812,353 (64.13 MB) ##  Trainable params: 2,097,665 (8.00 MB) ##  Non-trainable params: 14,714,688 (56.13 MB) print(model, expand_nested = TRUE) ## Model: \"sequential\" ## +-----------------------------+-----------------------+------------+-------+ ## | Layer (type)                | Output Shape          |    Param # | Trai… | ## +=============================+=======================+============+=======+ ## | vgg16 (Functional)          | (None, 4, 4, 512)     | 14,714,688 |   N   | ## +-----------------------------+-----------------------+------------+-------+ ## |    > input_layer            | (None, 150, 150, 3)   |          0 |   -   | ## | (InputLayer)                |                       |            |       | ## +-----------------------------+-----------------------+------------+-------+ ## |    > block1_conv1 (Conv2D)  | (None, 150, 150, 64)  |      1,792 |   N   | ## +-----------------------------+-----------------------+------------+-------+ ## |    > block1_conv2 (Conv2D)  | (None, 150, 150, 64)  |     36,928 |   N   | ## +-----------------------------+-----------------------+------------+-------+ ## |    > block1_pool            | (None, 75, 75, 64)    |          0 |   -   | ## | (MaxPooling2D)              |                       |            |       | ## +-----------------------------+-----------------------+------------+-------+ ## |    > block2_conv1 (Conv2D)  | (None, 75, 75, 128)   |     73,856 |   N   | ## +-----------------------------+-----------------------+------------+-------+ ## |    > block2_conv2 (Conv2D)  | (None, 75, 75, 128)   |    147,584 |   N   | ## +-----------------------------+-----------------------+------------+-------+ ## |    > block2_pool            | (None, 37, 37, 128)   |          0 |   -   | ## | (MaxPooling2D)              |                       |            |       | ## +-----------------------------+-----------------------+------------+-------+ ## |    > block3_conv1 (Conv2D)  | (None, 37, 37, 256)   |    295,168 |   N   | ## +-----------------------------+-----------------------+------------+-------+ ## |    > block3_conv2 (Conv2D)  | (None, 37, 37, 256)   |    590,080 |   N   | ## +-----------------------------+-----------------------+------------+-------+ ## |    > block3_conv3 (Conv2D)  | (None, 37, 37, 256)   |    590,080 |   N   | ## +-----------------------------+-----------------------+------------+-------+ ## |    > block3_pool            | (None, 18, 18, 256)   |          0 |   -   | ## | (MaxPooling2D)              |                       |            |       | ## +-----------------------------+-----------------------+------------+-------+ ## |    > block4_conv1 (Conv2D)  | (None, 18, 18, 512)   |  1,180,160 |   N   | ## +-----------------------------+-----------------------+------------+-------+ ## |    > block4_conv2 (Conv2D)  | (None, 18, 18, 512)   |  2,359,808 |   N   | ## +-----------------------------+-----------------------+------------+-------+ ## |    > block4_conv3 (Conv2D)  | (None, 18, 18, 512)   |  2,359,808 |   N   | ## +-----------------------------+-----------------------+------------+-------+ ## |    > block4_pool            | (None, 9, 9, 512)     |          0 |   -   | ## | (MaxPooling2D)              |                       |            |       | ## +-----------------------------+-----------------------+------------+-------+ ## |    > block5_conv1 (Conv2D)  | (None, 9, 9, 512)     |  2,359,808 |   N   | ## +-----------------------------+-----------------------+------------+-------+ ## |    > block5_conv2 (Conv2D)  | (None, 9, 9, 512)     |  2,359,808 |   N   | ## +-----------------------------+-----------------------+------------+-------+ ## |    > block5_conv3 (Conv2D)  | (None, 9, 9, 512)     |  2,359,808 |   N   | ## +-----------------------------+-----------------------+------------+-------+ ## |    > block5_pool            | (None, 4, 4, 512)     |          0 |   -   | ## | (MaxPooling2D)              |                       |            |       | ## +-----------------------------+-----------------------+------------+-------+ ## | flatten (Flatten)           | (None, 8192)          |          0 |   -   | ## +-----------------------------+-----------------------+------------+-------+ ## | dense (Dense)               | (None, 256)           |  2,097,408 |   Y   | ## +-----------------------------+-----------------------+------------+-------+ ## | dense_1 (Dense)             | (None, 1)             |        257 |   Y   | ## +-----------------------------+-----------------------+------------+-------+ ##  Total params: 16,812,353 (64.13 MB) ##  Trainable params: 2,097,665 (8.00 MB) ##  Non-trainable params: 14,714,688 (56.13 MB) # unfreeze weights from \"block5_conv1\" on unfreeze_weights(conv_base, from = \"block5_conv1\")  # compile again since we froze or unfroze weights model |> compile(   loss = \"binary_crossentropy\",   optimizer = optimizer_rmsprop(learning_rate = 2e-5),   metrics = c(\"accuracy\") )  conv_base ## Model: \"vgg16\" ## +-----------------------------+-----------------------+------------+-------+ ## | Layer (type)                | Output Shape          |    Param # | Trai… | ## +=============================+=======================+============+=======+ ## | input_layer (InputLayer)    | (None, 150, 150, 3)   |          0 |   -   | ## +-----------------------------+-----------------------+------------+-------+ ## | block1_conv1 (Conv2D)       | (None, 150, 150, 64)  |      1,792 |   N   | ## +-----------------------------+-----------------------+------------+-------+ ## | block1_conv2 (Conv2D)       | (None, 150, 150, 64)  |     36,928 |   N   | ## +-----------------------------+-----------------------+------------+-------+ ## | block1_pool (MaxPooling2D)  | (None, 75, 75, 64)    |          0 |   -   | ## +-----------------------------+-----------------------+------------+-------+ ## | block2_conv1 (Conv2D)       | (None, 75, 75, 128)   |     73,856 |   N   | ## +-----------------------------+-----------------------+------------+-------+ ## | block2_conv2 (Conv2D)       | (None, 75, 75, 128)   |    147,584 |   N   | ## +-----------------------------+-----------------------+------------+-------+ ## | block2_pool (MaxPooling2D)  | (None, 37, 37, 128)   |          0 |   -   | ## +-----------------------------+-----------------------+------------+-------+ ## | block3_conv1 (Conv2D)       | (None, 37, 37, 256)   |    295,168 |   N   | ## +-----------------------------+-----------------------+------------+-------+ ## | block3_conv2 (Conv2D)       | (None, 37, 37, 256)   |    590,080 |   N   | ## +-----------------------------+-----------------------+------------+-------+ ## | block3_conv3 (Conv2D)       | (None, 37, 37, 256)   |    590,080 |   N   | ## +-----------------------------+-----------------------+------------+-------+ ## | block3_pool (MaxPooling2D)  | (None, 18, 18, 256)   |          0 |   -   | ## +-----------------------------+-----------------------+------------+-------+ ## | block4_conv1 (Conv2D)       | (None, 18, 18, 512)   |  1,180,160 |   N   | ## +-----------------------------+-----------------------+------------+-------+ ## | block4_conv2 (Conv2D)       | (None, 18, 18, 512)   |  2,359,808 |   N   | ## +-----------------------------+-----------------------+------------+-------+ ## | block4_conv3 (Conv2D)       | (None, 18, 18, 512)   |  2,359,808 |   N   | ## +-----------------------------+-----------------------+------------+-------+ ## | block4_pool (MaxPooling2D)  | (None, 9, 9, 512)     |          0 |   -   | ## +-----------------------------+-----------------------+------------+-------+ ## | block5_conv1 (Conv2D)       | (None, 9, 9, 512)     |  2,359,808 |   Y   | ## +-----------------------------+-----------------------+------------+-------+ ## | block5_conv2 (Conv2D)       | (None, 9, 9, 512)     |  2,359,808 |   Y   | ## +-----------------------------+-----------------------+------------+-------+ ## | block5_conv3 (Conv2D)       | (None, 9, 9, 512)     |  2,359,808 |   Y   | ## +-----------------------------+-----------------------+------------+-------+ ## | block5_pool (MaxPooling2D)  | (None, 4, 4, 512)     |          0 |   -   | ## +-----------------------------+-----------------------+------------+-------+ ##  Total params: 14,714,688 (56.13 MB) ##  Trainable params: 7,079,424 (27.01 MB) ##  Non-trainable params: 7,635,264 (29.13 MB) print(model, expand_nested = TRUE) ## Model: \"sequential\" ## +-----------------------------+-----------------------+------------+-------+ ## | Layer (type)                | Output Shape          |    Param # | Trai… | ## +=============================+=======================+============+=======+ ## | vgg16 (Functional)          | (None, 4, 4, 512)     | 14,714,688 |   Y   | ## +-----------------------------+-----------------------+------------+-------+ ## |    > input_layer            | (None, 150, 150, 3)   |          0 |   -   | ## | (InputLayer)                |                       |            |       | ## +-----------------------------+-----------------------+------------+-------+ ## |    > block1_conv1 (Conv2D)  | (None, 150, 150, 64)  |      1,792 |   N   | ## +-----------------------------+-----------------------+------------+-------+ ## |    > block1_conv2 (Conv2D)  | (None, 150, 150, 64)  |     36,928 |   N   | ## +-----------------------------+-----------------------+------------+-------+ ## |    > block1_pool            | (None, 75, 75, 64)    |          0 |   -   | ## | (MaxPooling2D)              |                       |            |       | ## +-----------------------------+-----------------------+------------+-------+ ## |    > block2_conv1 (Conv2D)  | (None, 75, 75, 128)   |     73,856 |   N   | ## +-----------------------------+-----------------------+------------+-------+ ## |    > block2_conv2 (Conv2D)  | (None, 75, 75, 128)   |    147,584 |   N   | ## +-----------------------------+-----------------------+------------+-------+ ## |    > block2_pool            | (None, 37, 37, 128)   |          0 |   -   | ## | (MaxPooling2D)              |                       |            |       | ## +-----------------------------+-----------------------+------------+-------+ ## |    > block3_conv1 (Conv2D)  | (None, 37, 37, 256)   |    295,168 |   N   | ## +-----------------------------+-----------------------+------------+-------+ ## |    > block3_conv2 (Conv2D)  | (None, 37, 37, 256)   |    590,080 |   N   | ## +-----------------------------+-----------------------+------------+-------+ ## |    > block3_conv3 (Conv2D)  | (None, 37, 37, 256)   |    590,080 |   N   | ## +-----------------------------+-----------------------+------------+-------+ ## |    > block3_pool            | (None, 18, 18, 256)   |          0 |   -   | ## | (MaxPooling2D)              |                       |            |       | ## +-----------------------------+-----------------------+------------+-------+ ## |    > block4_conv1 (Conv2D)  | (None, 18, 18, 512)   |  1,180,160 |   N   | ## +-----------------------------+-----------------------+------------+-------+ ## |    > block4_conv2 (Conv2D)  | (None, 18, 18, 512)   |  2,359,808 |   N   | ## +-----------------------------+-----------------------+------------+-------+ ## |    > block4_conv3 (Conv2D)  | (None, 18, 18, 512)   |  2,359,808 |   N   | ## +-----------------------------+-----------------------+------------+-------+ ## |    > block4_pool            | (None, 9, 9, 512)     |          0 |   -   | ## | (MaxPooling2D)              |                       |            |       | ## +-----------------------------+-----------------------+------------+-------+ ## |    > block5_conv1 (Conv2D)  | (None, 9, 9, 512)     |  2,359,808 |   Y   | ## +-----------------------------+-----------------------+------------+-------+ ## |    > block5_conv2 (Conv2D)  | (None, 9, 9, 512)     |  2,359,808 |   Y   | ## +-----------------------------+-----------------------+------------+-------+ ## |    > block5_conv3 (Conv2D)  | (None, 9, 9, 512)     |  2,359,808 |   Y   | ## +-----------------------------+-----------------------+------------+-------+ ## |    > block5_pool            | (None, 4, 4, 512)     |          0 |   -   | ## | (MaxPooling2D)              |                       |            |       | ## +-----------------------------+-----------------------+------------+-------+ ## | flatten (Flatten)           | (None, 8192)          |          0 |   -   | ## +-----------------------------+-----------------------+------------+-------+ ## | dense (Dense)               | (None, 256)           |  2,097,408 |   Y   | ## +-----------------------------+-----------------------+------------+-------+ ## | dense_1 (Dense)             | (None, 1)             |        257 |   Y   | ## +-----------------------------+-----------------------+------------+-------+ ##  Total params: 16,812,353 (64.13 MB) ##  Trainable params: 9,177,089 (35.01 MB) ##  Non-trainable params: 7,635,264 (29.13 MB) # freeze only the last 5 layers freeze_weights(conv_base, from = -5) conv_base ## Model: \"vgg16\" ## +-----------------------------+-----------------------+------------+-------+ ## | Layer (type)                | Output Shape          |    Param # | Trai… | ## +=============================+=======================+============+=======+ ## | input_layer (InputLayer)    | (None, 150, 150, 3)   |          0 |   -   | ## +-----------------------------+-----------------------+------------+-------+ ## | block1_conv1 (Conv2D)       | (None, 150, 150, 64)  |      1,792 |   Y   | ## +-----------------------------+-----------------------+------------+-------+ ## | block1_conv2 (Conv2D)       | (None, 150, 150, 64)  |     36,928 |   Y   | ## +-----------------------------+-----------------------+------------+-------+ ## | block1_pool (MaxPooling2D)  | (None, 75, 75, 64)    |          0 |   -   | ## +-----------------------------+-----------------------+------------+-------+ ## | block2_conv1 (Conv2D)       | (None, 75, 75, 128)   |     73,856 |   Y   | ## +-----------------------------+-----------------------+------------+-------+ ## | block2_conv2 (Conv2D)       | (None, 75, 75, 128)   |    147,584 |   Y   | ## +-----------------------------+-----------------------+------------+-------+ ## | block2_pool (MaxPooling2D)  | (None, 37, 37, 128)   |          0 |   -   | ## +-----------------------------+-----------------------+------------+-------+ ## | block3_conv1 (Conv2D)       | (None, 37, 37, 256)   |    295,168 |   Y   | ## +-----------------------------+-----------------------+------------+-------+ ## | block3_conv2 (Conv2D)       | (None, 37, 37, 256)   |    590,080 |   Y   | ## +-----------------------------+-----------------------+------------+-------+ ## | block3_conv3 (Conv2D)       | (None, 37, 37, 256)   |    590,080 |   Y   | ## +-----------------------------+-----------------------+------------+-------+ ## | block3_pool (MaxPooling2D)  | (None, 18, 18, 256)   |          0 |   -   | ## +-----------------------------+-----------------------+------------+-------+ ## | block4_conv1 (Conv2D)       | (None, 18, 18, 512)   |  1,180,160 |   Y   | ## +-----------------------------+-----------------------+------------+-------+ ## | block4_conv2 (Conv2D)       | (None, 18, 18, 512)   |  2,359,808 |   Y   | ## +-----------------------------+-----------------------+------------+-------+ ## | block4_conv3 (Conv2D)       | (None, 18, 18, 512)   |  2,359,808 |   Y   | ## +-----------------------------+-----------------------+------------+-------+ ## | block4_pool (MaxPooling2D)  | (None, 9, 9, 512)     |          0 |   -   | ## +-----------------------------+-----------------------+------------+-------+ ## | block5_conv1 (Conv2D)       | (None, 9, 9, 512)     |  2,359,808 |   N   | ## +-----------------------------+-----------------------+------------+-------+ ## | block5_conv2 (Conv2D)       | (None, 9, 9, 512)     |  2,359,808 |   N   | ## +-----------------------------+-----------------------+------------+-------+ ## | block5_conv3 (Conv2D)       | (None, 9, 9, 512)     |  2,359,808 |   N   | ## +-----------------------------+-----------------------+------------+-------+ ## | block5_pool (MaxPooling2D)  | (None, 4, 4, 512)     |          0 |   -   | ## +-----------------------------+-----------------------+------------+-------+ ##  Total params: 14,714,688 (56.13 MB) ##  Trainable params: 7,635,264 (29.13 MB) ##  Non-trainable params: 7,079,424 (27.01 MB) # freeze only the last 5 layers, a different way unfreeze_weights(conv_base, to = -6) conv_base ## Model: \"vgg16\" ## +-----------------------------+-----------------------+------------+-------+ ## | Layer (type)                | Output Shape          |    Param # | Trai… | ## +=============================+=======================+============+=======+ ## | input_layer (InputLayer)    | (None, 150, 150, 3)   |          0 |   -   | ## +-----------------------------+-----------------------+------------+-------+ ## | block1_conv1 (Conv2D)       | (None, 150, 150, 64)  |      1,792 |   Y   | ## +-----------------------------+-----------------------+------------+-------+ ## | block1_conv2 (Conv2D)       | (None, 150, 150, 64)  |     36,928 |   Y   | ## +-----------------------------+-----------------------+------------+-------+ ## | block1_pool (MaxPooling2D)  | (None, 75, 75, 64)    |          0 |   -   | ## +-----------------------------+-----------------------+------------+-------+ ## | block2_conv1 (Conv2D)       | (None, 75, 75, 128)   |     73,856 |   Y   | ## +-----------------------------+-----------------------+------------+-------+ ## | block2_conv2 (Conv2D)       | (None, 75, 75, 128)   |    147,584 |   Y   | ## +-----------------------------+-----------------------+------------+-------+ ## | block2_pool (MaxPooling2D)  | (None, 37, 37, 128)   |          0 |   -   | ## +-----------------------------+-----------------------+------------+-------+ ## | block3_conv1 (Conv2D)       | (None, 37, 37, 256)   |    295,168 |   Y   | ## +-----------------------------+-----------------------+------------+-------+ ## | block3_conv2 (Conv2D)       | (None, 37, 37, 256)   |    590,080 |   Y   | ## +-----------------------------+-----------------------+------------+-------+ ## | block3_conv3 (Conv2D)       | (None, 37, 37, 256)   |    590,080 |   Y   | ## +-----------------------------+-----------------------+------------+-------+ ## | block3_pool (MaxPooling2D)  | (None, 18, 18, 256)   |          0 |   -   | ## +-----------------------------+-----------------------+------------+-------+ ## | block4_conv1 (Conv2D)       | (None, 18, 18, 512)   |  1,180,160 |   Y   | ## +-----------------------------+-----------------------+------------+-------+ ## | block4_conv2 (Conv2D)       | (None, 18, 18, 512)   |  2,359,808 |   Y   | ## +-----------------------------+-----------------------+------------+-------+ ## | block4_conv3 (Conv2D)       | (None, 18, 18, 512)   |  2,359,808 |   Y   | ## +-----------------------------+-----------------------+------------+-------+ ## | block4_pool (MaxPooling2D)  | (None, 9, 9, 512)     |          0 |   -   | ## +-----------------------------+-----------------------+------------+-------+ ## | block5_conv1 (Conv2D)       | (None, 9, 9, 512)     |  2,359,808 |   N   | ## +-----------------------------+-----------------------+------------+-------+ ## | block5_conv2 (Conv2D)       | (None, 9, 9, 512)     |  2,359,808 |   N   | ## +-----------------------------+-----------------------+------------+-------+ ## | block5_conv3 (Conv2D)       | (None, 9, 9, 512)     |  2,359,808 |   N   | ## +-----------------------------+-----------------------+------------+-------+ ## | block5_pool (MaxPooling2D)  | (None, 4, 4, 512)     |          0 |   -   | ## +-----------------------------+-----------------------+------------+-------+ ##  Total params: 14,714,688 (56.13 MB) ##  Trainable params: 7,635,264 (29.13 MB) ##  Non-trainable params: 7,079,424 (27.01 MB) # Freeze only layers of a certain type, e.g, BatchNorm layers # batch_norm_layer_class_name <- class(layer_batch_normalization())[1] # is_batch_norm_layer <- function(x) inherits(x, batch_norm_layer_class_name) is_batch_norm_layer <- function(x) inherits(x, keras$layers$BatchNormalization)  model <- application_efficientnet_v2b0() freeze_weights(model, which = is_batch_norm_layer) # print(model)  # equivalent to: for(layer in model$layers) {   if(is_batch_norm_layer(layer))     layer$trainable <- FALSE   else     layer$trainable <- TRUE }"},{"path":"https://keras3.posit.co/dev/reference/get_config.html","id":null,"dir":"Reference","previous_headings":"","what":"Layer/Model configuration — get_config","title":"Layer/Model configuration — get_config","text":"layer config object returned get_config() contains configuration layer model. layer model can reinstantiated later (without trained weights) configuration using from_config(). config include connectivity information, class name (handled externally).","code":""},{"path":"https://keras3.posit.co/dev/reference/get_config.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Layer/Model configuration — get_config","text":"","code":"get_config(object)  from_config(config, custom_objects = NULL)"},{"path":"https://keras3.posit.co/dev/reference/get_config.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Layer/Model configuration — get_config","text":"object Layer model object config Object layer model configuration custom_objects list custom objects needed instantiate layer, e.g., custom layers defined new_layer_class() similar.","code":""},{"path":"https://keras3.posit.co/dev/reference/get_config.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Layer/Model configuration — get_config","text":"get_config() returns object configuration, from_config() returns re-instantiation object.","code":""},{"path":"https://keras3.posit.co/dev/reference/get_config.html","id":"note","dir":"Reference","previous_headings":"","what":"Note","title":"Layer/Model configuration — get_config","text":"Objects returned get_config() serializable via RDS. want save restore model across sessions, can use save_model_config() (model configuration , weights) save_model() save model configuration weights filesystem.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/get_custom_objects.html","id":null,"dir":"Reference","previous_headings":"","what":"Get/set the currently registered custom objects. — get_custom_objects","title":"Get/set the currently registered custom objects. — get_custom_objects","text":"Custom objects set using custom_object_scope() added global list custom objects, appear returned list.","code":""},{"path":"https://keras3.posit.co/dev/reference/get_custom_objects.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Get/set the currently registered custom objects. — get_custom_objects","text":"","code":"get_custom_objects()  set_custom_objects(objects = named_list(), clear = TRUE)"},{"path":"https://keras3.posit.co/dev/reference/get_custom_objects.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Get/set the currently registered custom objects. — get_custom_objects","text":"objects named list custom objects, returned get_custom_objects() set_custom_objects(). clear bool, whether clear custom object registry populating objects.","code":""},{"path":"https://keras3.posit.co/dev/reference/get_custom_objects.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Get/set the currently registered custom objects. — get_custom_objects","text":"R named list mapping registered names registered objects. set_custom_objects() returns registry values updating, invisibly.","code":""},{"path":"https://keras3.posit.co/dev/reference/get_custom_objects.html","id":"note","dir":"Reference","previous_headings":"","what":"Note","title":"Get/set the currently registered custom objects. — get_custom_objects","text":"register_keras_serializable() preferred set_custom_objects() registering new objects.","code":""},{"path":"https://keras3.posit.co/dev/reference/get_custom_objects.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Get/set the currently registered custom objects. — get_custom_objects","text":"can use set_custom_objects() restore previous registry state.","code":"get_custom_objects() # within a function, if you want to temporarily modify the registry, function() {   orig_objects <- set_custom_objects(clear = TRUE)   on.exit(set_custom_objects(orig_objects))    ## temporarily modify the global registry   # register_keras_serializable(....)   # ....  <do work>   # on.exit(), the previous registry state is restored. }"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/get_file.html","id":null,"dir":"Reference","previous_headings":"","what":"Downloads a file from a URL if it not already in the cache. — get_file","title":"Downloads a file from a URL if it not already in the cache. — get_file","text":"default file url origin downloaded cache_dir ~/.keras, placed cache_subdir datasets, given filename fname. final location file example.txt therefore ~/.keras/datasets/example.txt. Files .tar, .tar.gz, .tar.bz, .zip formats can also extracted. Passing hash verify file download. command line programs shasum sha256sum can compute hash.","code":""},{"path":"https://keras3.posit.co/dev/reference/get_file.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Downloads a file from a URL if it not already in the cache. — get_file","text":"","code":"get_file(   fname = NULL,   origin = NULL,   ...,   file_hash = NULL,   cache_subdir = \"datasets\",   hash_algorithm = \"auto\",   extract = FALSE,   archive_format = \"auto\",   cache_dir = NULL,   force_download = FALSE )"},{"path":"https://keras3.posit.co/dev/reference/get_file.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Downloads a file from a URL if it not already in the cache. — get_file","text":"fname target single file, desired local name file. NULL, name file origin used. downloading extracting directory archive, provided fname used extraction directory name (extension). origin Original URL file. ... forward/backward compatability. file_hash expected hash string file download. sha256 md5 hash algorithms supported. cache_subdir Subdirectory Keras cache dir file saved. absolute path, e.g. \"/path//folder\" specified, file saved location. hash_algorithm Select hash algorithm verify file. options \"md5', \"sha256', \"auto'. default 'auto' detects hash algorithm use. extract TRUE, extracts archive. applicable compressed archive files like tar zip. archive_format Archive format try extracting file. Options \"auto', \"tar', \"zip', NULL. \"tar\" includes tar, tar.gz, tar.bz files. default \"auto\" corresponds c(\"tar\", \"zip\"). NULL empty list return matches found. cache_dir Location store cached files, NULL defaults Sys.getenv(\"KERAS_HOME\", \"~/.keras/\"). force_download TRUE, file always re-downloaded regardless cache state.","code":""},{"path":"https://keras3.posit.co/dev/reference/get_file.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Downloads a file from a URL if it not already in the cache. — get_file","text":"Path downloaded file. ** Warning malicious downloads ** Downloading something Internet carries risk. NEVER download file/archive trust source. recommend specify file_hash argument (hash source file known) make sure file getting one expect.","code":""},{"path":"https://keras3.posit.co/dev/reference/get_file.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Downloads a file from a URL if it not already in the cache. — get_file","text":"","code":"path_to_downloaded_file <- get_file(     \"flower_photos\",     origin = \"https://storage.googleapis.com/download.tensorflow.org/example_images/flower_photos.tgz\",     extract = TRUE )"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/get_layer.html","id":null,"dir":"Reference","previous_headings":"","what":"Retrieves a layer based on either its name (unique) or index. — get_layer","title":"Retrieves a layer based on either its name (unique) or index. — get_layer","text":"Indices based order horizontal graph traversal (bottom-) 1-based. name index provided, index take precedence.","code":""},{"path":"https://keras3.posit.co/dev/reference/get_layer.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Retrieves a layer based on either its name (unique) or index. — get_layer","text":"","code":"get_layer(object, name = NULL, index = NULL)"},{"path":"https://keras3.posit.co/dev/reference/get_layer.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Retrieves a layer based on either its name (unique) or index. — get_layer","text":"object Keras model object name String, name layer. index Integer, index layer (1-based). Also valid negative values, count end model.","code":""},{"path":"https://keras3.posit.co/dev/reference/get_layer.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Retrieves a layer based on either its name (unique) or index. — get_layer","text":"layer instance.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/get_registered_name.html","id":null,"dir":"Reference","previous_headings":"","what":"Returns the name registered to an object within the Keras framework. — get_registered_name","title":"Returns the name registered to an object within the Keras framework. — get_registered_name","text":"function part Keras serialization deserialization framework. maps objects string names associated objects serialization/deserialization.","code":""},{"path":"https://keras3.posit.co/dev/reference/get_registered_name.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Returns the name registered to an object within the Keras framework. — get_registered_name","text":"","code":"get_registered_name(obj)"},{"path":"https://keras3.posit.co/dev/reference/get_registered_name.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Returns the name registered to an object within the Keras framework. — get_registered_name","text":"obj object look .","code":""},{"path":"https://keras3.posit.co/dev/reference/get_registered_name.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Returns the name registered to an object within the Keras framework. — get_registered_name","text":"name associated object, default name object registered.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/get_registered_object.html","id":null,"dir":"Reference","previous_headings":"","what":"Returns the class associated with name if it is registered with Keras. — get_registered_object","title":"Returns the class associated with name if it is registered with Keras. — get_registered_object","text":"function part Keras serialization deserialization framework. maps strings objects associated serialization/deserialization.","code":""},{"path":"https://keras3.posit.co/dev/reference/get_registered_object.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Returns the class associated with name if it is registered with Keras. — get_registered_object","text":"","code":"get_registered_object(name, custom_objects = NULL, module_objects = NULL)"},{"path":"https://keras3.posit.co/dev/reference/get_registered_object.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Returns the class associated with name if it is registered with Keras. — get_registered_object","text":"name name look . custom_objects named list custom objects look name . Generally, custom_objects provided user. module_objects named list custom objects look name . Generally, module_objects provided midlevel library implementers.","code":""},{"path":"https://keras3.posit.co/dev/reference/get_registered_object.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Returns the class associated with name if it is registered with Keras. — get_registered_object","text":"instantiable class associated name, NULL class exists.","code":""},{"path":"https://keras3.posit.co/dev/reference/get_registered_object.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Returns the class associated with name if it is registered with Keras. — get_registered_object","text":"","code":"from_config <- function(cls, config, custom_objects = NULL) {   if ('my_custom_object_name' \\%in\\% names(config)) {     config$hidden_cls <- get_registered_object(       config$my_custom_object_name,       custom_objects = custom_objects)   } }"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/get_source_inputs.html","id":null,"dir":"Reference","previous_headings":"","what":"Returns the list of input tensors necessary to compute tensor. — get_source_inputs","title":"Returns the list of input tensors necessary to compute tensor. — get_source_inputs","text":"Output always list tensors (potentially 1 element).","code":""},{"path":"https://keras3.posit.co/dev/reference/get_source_inputs.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Returns the list of input tensors necessary to compute tensor. — get_source_inputs","text":"","code":"get_source_inputs(tensor)"},{"path":"https://keras3.posit.co/dev/reference/get_source_inputs.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Returns the list of input tensors necessary to compute tensor. — get_source_inputs","text":"tensor tensor start .","code":""},{"path":"https://keras3.posit.co/dev/reference/get_source_inputs.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Returns the list of input tensors necessary to compute tensor. — get_source_inputs","text":"List input tensors.","code":""},{"path":"https://keras3.posit.co/dev/reference/get_source_inputs.html","id":"example","dir":"Reference","previous_headings":"","what":"Example","title":"Returns the list of input tensors necessary to compute tensor. — get_source_inputs","text":"","code":"input <- keras_input(c(3)) output <- input |> layer_dense(4) |> op_multiply(5) reticulate::py_id(get_source_inputs(output)[[1]]) == reticulate::py_id(input) ## [1] TRUE"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/get_state_tree.html","id":null,"dir":"Reference","previous_headings":"","what":"Retrieves tree-like structure of model variables. — get_state_tree","title":"Retrieves tree-like structure of model variables. — get_state_tree","text":"method allows retrieval different model variables (trainable, non-trainable, optimizer, metrics). variables returned nested dictionary format, keys correspond variable names values nested representations variables.","code":""},{"path":"https://keras3.posit.co/dev/reference/get_state_tree.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Retrieves tree-like structure of model variables. — get_state_tree","text":"","code":"get_state_tree(object, value_format = \"backend_tensor\")"},{"path":"https://keras3.posit.co/dev/reference/get_state_tree.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Retrieves tree-like structure of model variables. — get_state_tree","text":"object Keras Model. value_format One \"backend_tensor\", \"numpy_array\", \"array\". kind array return leaves nested state tree.","code":""},{"path":"https://keras3.posit.co/dev/reference/get_state_tree.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Retrieves tree-like structure of model variables. — get_state_tree","text":"named list containing nested representations requested variables. names variable names, values corresponding nested named lists.","code":""},{"path":"https://keras3.posit.co/dev/reference/get_state_tree.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Retrieves tree-like structure of model variables. — get_state_tree","text":"state_tree list returned looks like:   example:","code":"model <- keras_model_sequential(name = \"my_sequential\",                                 input_shape = c(1),                                 input_name = \"my_input\") |>   layer_dense(1, activation = \"sigmoid\", name = \"my_dense\")  model |> compile(optimizer=\"adam\", loss=\"mse\", metrics=c(\"mae\")) model |> fit(matrix(1), matrix(1), verbose = 0) state_tree <- model |> get_state_tree() list(   metrics_variables = list(     loss = list(       count = ...,       total = ...     ),     mean_absolute_error = list(       count = ...,       total = ...     )   ),   trainable_variables = list(     my_sequential = list(       my_dense = list(         bias = ...,         kernel = ...       )     )   ),   non_trainable_variables = list(),   optimizer_variables = list(     adam = list(       iteration = ...,       learning_rate = ...,       my_sequential_my_dense_bias_momentum = ...,       my_sequential_my_dense_bias_velocity = ...,       my_sequential_my_dense_kernel_momentum = ...,       my_sequential_my_dense_kernel_velocity = ...     )   ) ) str(state_tree) ## List of 4 ##  $ trainable_variables    :List of 1 ##   ..$ my_sequential:List of 1 ##   .. ..$ my_dense:List of 2 ##   .. .. ..$ kernel:<tf.Variable 'my_sequential/my_dense/kernel:0' shape=(1, 1) dtype=float32, numpy=array([[-0.8338491]], dtype=float32)> ##   .. .. ..$ bias  :<tf.Variable 'my_sequential/my_dense/bias:0' shape=(1) dtype=float32, numpy=array([0.00099998], dtype=float32)> ##  $ non_trainable_variables: Named list() ##  $ optimizer_variables    :List of 1 ##   ..$ adam:List of 6 ##   .. ..$ iteration                             :<tf.Variable 'adam/iteration:0' shape=() dtype=int64, numpy=1> ##   .. ..$ learning_rate                         :<tf.Variable 'adam/learning_rate:0' shape=() dtype=float32, numpy=0.0010000000474974513> ##   .. ..$ my_sequential_my_dense_kernel_momentum:<tf.Variable 'adam/my_sequential_my_dense_kernel_momentum:0' shape=(1, 1) dtype=float32, numpy=array([[-0.02943518]], dtype=float32)> ##   .. ..$ my_sequential_my_dense_kernel_velocity:<tf.Variable 'adam/my_sequential_my_dense_kernel_velocity:0' shape=(1, 1) dtype=float32, numpy=array([[8.664299e-05]], dtype=float32)> ##   .. ..$ my_sequential_my_dense_bias_momentum  :<tf.Variable 'adam/my_sequential_my_dense_bias_momentum:0' shape=(1) dtype=float32, numpy=array([-0.02943518], dtype=float32)> ##   .. ..$ my_sequential_my_dense_bias_velocity  :<tf.Variable 'adam/my_sequential_my_dense_bias_velocity:0' shape=(1) dtype=float32, numpy=array([8.664299e-05], dtype=float32)> ##  $ metrics_variables      :List of 2 ##   ..$ loss               :List of 2 ##   .. ..$ total:<tf.Variable 'loss/total:0' shape=() dtype=float32, numpy=0.4863377809524536> ##   .. ..$ count:<tf.Variable 'loss/count:0' shape=() dtype=float32, numpy=1.0> ##   ..$ mean_absolute_error:List of 2 ##   .. ..$ total:<tf.Variable 'mean_absolute_error/total:0' shape=() dtype=float32, numpy=0.6973792314529419> ##   .. ..$ count:<tf.Variable 'mean_absolute_error_1/count:0' shape=() dtype=float32, numpy=1.0>"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/get_weights.html","id":null,"dir":"Reference","previous_headings":"","what":"Layer/Model weights as R arrays — get_weights","title":"Layer/Model weights as R arrays — get_weights","text":"Layer/Model weights R arrays","code":""},{"path":"https://keras3.posit.co/dev/reference/get_weights.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Layer/Model weights as R arrays — get_weights","text":"","code":"get_weights(object, trainable = NA)  set_weights(object, weights)"},{"path":"https://keras3.posit.co/dev/reference/get_weights.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Layer/Model weights as R arrays — get_weights","text":"object Layer model object trainable NA (default), weights returned. TRUE, weights trainable variables returned. FALSE, weights non-trainable variables returned. weights Weights R array","code":""},{"path":"https://keras3.posit.co/dev/reference/get_weights.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Layer/Model weights as R arrays — get_weights","text":"list R arrays.","code":""},{"path":"https://keras3.posit.co/dev/reference/get_weights.html","id":"note","dir":"Reference","previous_headings":"","what":"Note","title":"Layer/Model weights as R arrays — get_weights","text":"can access Layer/Model KerasVariables (also backend-native tensors like tf.Variable) object$weights, object$trainable_weights, object$non_trainable_weights","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/grapes-py_class-grapes.html","id":null,"dir":"Reference","previous_headings":"","what":"Make a python class constructor — %py_class%","title":"Make a python class constructor — %py_class%","text":"Make python class constructor","code":""},{"path":"https://keras3.posit.co/dev/reference/grapes-py_class-grapes.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Make a python class constructor — %py_class%","text":"","code":"spec %py_class% body"},{"path":"https://keras3.posit.co/dev/reference/grapes-py_class-grapes.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Make a python class constructor — %py_class%","text":"spec bare symbol MyClassName, call MyClassName(SuperClass) body expression can evaluated construct class methods.","code":""},{"path":"https://keras3.posit.co/dev/reference/grapes-py_class-grapes.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Make a python class constructor — %py_class%","text":"python class constructor, invisibly. Note, constructor also assigned parent frame.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/grapes-py_class-grapes.html","id":"ref-examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Make a python class constructor — %py_class%","text":"","code":"if (FALSE) { # \\dontrun{ MyClass %py_class% {   initialize <- function(x) {     print(\"Hi from MyClass$initialize()!\")     self$x <- x   }   my_method <- function() {     self$x   } }  my_class_instance <- MyClass(42) my_class_instance$my_method()  MyClass2(MyClass) %py_class% {   \"This will be a __doc__ string for MyClass2\"    initialize <- function(...) {     \"This will be the __doc__ string for the MyClass2.__init__() method\"     print(\"Hi from MyClass2$initialize()!\")     super$initialize(...)   } }  my_class_instance2 <- MyClass2(42) my_class_instance2$my_method()  reticulate::py_help(MyClass2) # see the __doc__ strings and more!  # In addition to `self`, there is also `private` available. # This is an R environment unique to each class instance, where you can # store objects that you don't want converted to Python, but still want # available from methods. You can also assign methods to private, and # `self` and `private` will be available in private methods.  MyClass %py_class% {    initialize <- function(x) {     print(\"Hi from MyClass$initialize()!\")     private$y <- paste(\"A Private field:\", x)   }    get_private_field <- function() {     private$y   }    private$a_private_method <- function() {     cat(\"a_private_method() was called.\\n\")     cat(\"private$y is \", sQuote(private$y), \"\\n\")   }    call_private_method <- function()     private$a_private_method()    # equivalent of @property decorator in python   an_active_property %<-active% function(x = NULL) {     if(!is.null(x)) {       cat(\"`an_active_property` was assigned\", x, \"\\n\")       return(x)     } else {       cat(\"`an_active_property` was accessed\\n\")       return(42)     }   } }  inst1 <- MyClass(1) inst2 <- MyClass(2) inst1$get_private_field() inst2$get_private_field() inst1$call_private_method() inst2$call_private_method() inst1$an_active_property inst1$an_active_property <- 11 } # }"},{"path":"https://keras3.posit.co/dev/reference/grapes-set-active-grapes.html","id":null,"dir":"Reference","previous_headings":"","what":"Make an Active Binding — %<-active%","title":"Make an Active Binding — %<-active%","text":"Make Active Binding","code":""},{"path":"https://keras3.posit.co/dev/reference/grapes-set-active-grapes.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Make an Active Binding — %<-active%","text":"","code":"sym %<-active% value"},{"path":"https://keras3.posit.co/dev/reference/grapes-set-active-grapes.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Make an Active Binding — %<-active%","text":"sym symbol bind value function call value sym accessed.","code":""},{"path":"https://keras3.posit.co/dev/reference/grapes-set-active-grapes.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Make an Active Binding — %<-active%","text":"value, invisibly","code":""},{"path":"https://keras3.posit.co/dev/reference/grapes-set-active-grapes.html","id":"details","dir":"Reference","previous_headings":"","what":"Details","title":"Make an Active Binding — %<-active%","text":"Active bindings defined %py_class% converted @property decorated methods.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/grapes-set-active-grapes.html","id":"ref-examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Make an Active Binding — %<-active%","text":"","code":"set.seed(1234) x %<-active% function(value) {   message(\"Evaluating function of active binding\")   if(missing(value))     runif(1)   else    message(\"Received: \", value) } x #> Evaluating function of active binding #> [1] 0.1137034 x #> Evaluating function of active binding #> [1] 0.6222994 x <- \"foo\" #> Evaluating function of active binding #> Received: foo x <- \"foo\" #> Evaluating function of active binding #> Received: foo x #> Evaluating function of active binding #> [1] 0.6092747 rm(x) # cleanup"},{"path":"https://keras3.posit.co/dev/reference/image_array_save.html","id":null,"dir":"Reference","previous_headings":"","what":"Saves an image stored as an array to a path or file object. — image_array_save","title":"Saves an image stored as an array to a path or file object. — image_array_save","text":"Saves image stored array path file object.","code":""},{"path":"https://keras3.posit.co/dev/reference/image_array_save.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Saves an image stored as an array to a path or file object. — image_array_save","text":"","code":"image_array_save(   x,   path,   data_format = NULL,   file_format = NULL,   scale = TRUE,   ... )"},{"path":"https://keras3.posit.co/dev/reference/image_array_save.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Saves an image stored as an array to a path or file object. — image_array_save","text":"x array. path Path file object. data_format Image data format, either \"channels_first\" \"channels_last\". file_format Optional file format override. omitted, format use determined filename extension. file object used instead filename, parameter always used. scale Whether rescale image values within [0, 255]. ... Additional keyword arguments passed PIL.Image.save().","code":""},{"path":"https://keras3.posit.co/dev/reference/image_array_save.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Saves an image stored as an array to a path or file object. — image_array_save","text":"Called primarily side effects. input x returned, invisibly, enable usage pipe.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/image_dataset_from_directory.html","id":null,"dir":"Reference","previous_headings":"","what":"Generates a tf.data.Dataset from image files in a directory. — image_dataset_from_directory","title":"Generates a tf.data.Dataset from image files in a directory. — image_dataset_from_directory","text":"directory structure :   calling image_dataset_from_directory(main_directory, labels = 'inferred') return tf.data.Dataset yields batches images subdirectories class_a class_b, together labels 0 1 (0 corresponding class_a 1 corresponding class_b). Supported image formats: .jpeg, .jpg, .png, .bmp, .gif. Animated gifs truncated first frame.","code":"main_directory/ ...class_a/ ......a_image_1.jpg ......a_image_2.jpg ...class_b/ ......b_image_1.jpg ......b_image_2.jpg"},{"path":"https://keras3.posit.co/dev/reference/image_dataset_from_directory.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Generates a tf.data.Dataset from image files in a directory. — image_dataset_from_directory","text":"","code":"image_dataset_from_directory(   directory,   labels = \"inferred\",   label_mode = \"int\",   class_names = NULL,   color_mode = \"rgb\",   batch_size = 32L,   image_size = c(256L, 256L),   shuffle = TRUE,   seed = NULL,   validation_split = NULL,   subset = NULL,   interpolation = \"bilinear\",   follow_links = FALSE,   crop_to_aspect_ratio = FALSE,   pad_to_aspect_ratio = FALSE,   data_format = NULL,   verbose = TRUE )"},{"path":"https://keras3.posit.co/dev/reference/image_dataset_from_directory.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Generates a tf.data.Dataset from image files in a directory. — image_dataset_from_directory","text":"directory Directory data located. labels \"inferred\", contain subdirectories, containing images class. Otherwise, directory structure ignored. labels Either \"inferred\" (labels generated directory structure), NULL (labels), list/tuple integer labels size number image files found directory. Labels sorted according alphanumeric order image file paths (obtained via os.walk(directory) Python). label_mode String describing encoding labels. Options : \"int\": means labels encoded integers (e.g. sparse_categorical_crossentropy loss). \"categorical\" means labels encoded categorical vector (e.g. categorical_crossentropy loss). \"binary\" means labels (can 2) encoded float32 scalars values 0 1 (e.g. binary_crossentropy). NULL (labels). class_names valid labels \"inferred\". explicit list class names (must match names subdirectories). Used control order classes (otherwise alphanumerical order used). color_mode One \"grayscale\", \"rgb\", \"rgba\". Whether images converted 1, 3, 4 channels. Defaults \"rgb\". batch_size Size batches data. Defaults 32. NULL, data batched (dataset yield individual samples). image_size Size resize images read disk, specified (height, width). Since pipeline processes batches images must size, must provided. Defaults (256, 256). shuffle Whether shuffle data. Defaults TRUE. set FALSE, sorts data alphanumeric order. seed Optional random seed shuffling transformations. validation_split Optional float 0 1, fraction data reserve validation. subset Subset data return. One \"training\", \"validation\", \"\". used validation_split set. subset = \"\", utility returns tuple two datasets (training validation datasets respectively). interpolation String, interpolation method used resizing images. Supports \"bilinear\", \"nearest\", \"bicubic\", \"area\", \"lanczos3\", \"lanczos5\", \"gaussian\", \"mitchellcubic\". Defaults \"bilinear\". follow_links Whether visit subdirectories pointed symlinks. Defaults FALSE. crop_to_aspect_ratio TRUE, resize images without aspect ratio distortion. original aspect ratio differs target aspect ratio, output image cropped return largest possible window image (size image_size) matches target aspect ratio. default (crop_to_aspect_ratio = FALSE), aspect ratio may preserved. pad_to_aspect_ratio TRUE, resize images without aspect ratio distortion. original aspect ratio differs target aspect ratio, output image padded return largest possible window image (size image_size) matches target aspect ratio. default (pad_to_aspect_ratio=FALSE), aspect ratio may preserved. data_format NULL uses config_image_data_format() otherwise either 'channel_last' 'channel_first'. verbose Whether display number information classes number files found. Defaults TRUE.","code":""},{"path":"https://keras3.posit.co/dev/reference/image_dataset_from_directory.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Generates a tf.data.Dataset from image files in a directory. — image_dataset_from_directory","text":"tf.data.Dataset object. label_mode NULL, yields float32 tensors shape (batch_size, image_size[1], image_size[2], num_channels), encoding images (see rules regarding num_channels). Otherwise, yields tuple (images, labels), images shape (batch_size, image_size[1], image_size[2], num_channels), labels follows format described . Rules regarding labels format: label_mode \"int\", labels int32 tensor shape (batch_size,). label_mode \"binary\", labels float32 tensor 1s 0s shape (batch_size, 1). label_mode \"categorical\", labels float32 tensor shape (batch_size, num_classes), representing one-hot encoding class index. Rules regarding number channels yielded images: color_mode \"grayscale\", 1 channel image tensors. color_mode \"rgb\", 3 channels image tensors. color_mode \"rgba\", 4 channels image tensors.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/image_from_array.html","id":null,"dir":"Reference","previous_headings":"","what":"Converts a 3D array to a PIL Image instance. — image_from_array","title":"Converts a 3D array to a PIL Image instance. — image_from_array","text":"Converts 3D array PIL Image instance.","code":""},{"path":"https://keras3.posit.co/dev/reference/image_from_array.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Converts a 3D array to a PIL Image instance. — image_from_array","text":"","code":"image_from_array(x, data_format = NULL, scale = TRUE, dtype = NULL)"},{"path":"https://keras3.posit.co/dev/reference/image_from_array.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Converts a 3D array to a PIL Image instance. — image_from_array","text":"x Input data, form can converted array. data_format Image data format, can either \"channels_first\" \"channels_last\". Defaults NULL, case global setting config_image_data_format() used (unless changed , defaults \"channels_last\"). scale Whether rescale image minimum maximum values 0 255 respectively. Defaults TRUE. dtype Dtype use. NULL means global setting config_floatx() used (unless changed , defaults \"float32\"). Defaults NULL.","code":""},{"path":"https://keras3.posit.co/dev/reference/image_from_array.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Converts a 3D array to a PIL Image instance. — image_from_array","text":"PIL Image instance.","code":""},{"path":"https://keras3.posit.co/dev/reference/image_from_array.html","id":"example","dir":"Reference","previous_headings":"","what":"Example","title":"Converts a 3D array to a PIL Image instance. — image_from_array","text":"","code":"img <- array(runif(30000), dim = c(100, 100, 3)) pil_img <- image_from_array(img) pil_img ## <PIL.Image.Image image mode=RGB size=100x100 at 0x0>"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/image_load.html","id":null,"dir":"Reference","previous_headings":"","what":"Loads an image into PIL format. — image_load","title":"Loads an image into PIL format. — image_load","text":"Loads image PIL format.","code":""},{"path":"https://keras3.posit.co/dev/reference/image_load.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Loads an image into PIL format. — image_load","text":"","code":"image_load(   path,   color_mode = \"rgb\",   target_size = NULL,   interpolation = \"nearest\",   keep_aspect_ratio = FALSE )"},{"path":"https://keras3.posit.co/dev/reference/image_load.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Loads an image into PIL format. — image_load","text":"path Path image file. color_mode One \"grayscale\", \"rgb\", \"rgba\". Default: \"rgb\". desired image format. target_size Either NULL (default original size) tuple ints (img_height, img_width). interpolation Interpolation method used resample image target size different loaded image. Supported methods \"nearest\", \"bilinear\", \"bicubic\". PIL version 1.1.3 newer installed, \"lanczos\" also supported. PIL version 3.4.0 newer installed, \"box\" \"hamming\" also supported. default, \"nearest\" used. keep_aspect_ratio Boolean, whether resize images target size without aspect ratio distortion. image cropped center target aspect ratio resizing.","code":""},{"path":"https://keras3.posit.co/dev/reference/image_load.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Loads an image into PIL format. — image_load","text":"PIL Image instance.","code":""},{"path":"https://keras3.posit.co/dev/reference/image_load.html","id":"example","dir":"Reference","previous_headings":"","what":"Example","title":"Loads an image into PIL format. — image_load","text":"","code":"image_path <- get_file(origin = \"https://www.r-project.org/logo/Rlogo.png\") (image <- image_load(image_path)) ## <PIL.Image.Image image mode=RGB size=724x561 at 0x0> input_arr <- image_to_array(image) str(input_arr) ##  num [1:561, 1:724, 1:3] 0 0 0 0 0 0 0 0 0 0 ... input_arr %<>% array_reshape(dim = c(1, dim(input_arr))) # Convert single image to a batch. model |> predict(input_arr)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/image_smart_resize.html","id":null,"dir":"Reference","previous_headings":"","what":"Resize images to a target size without aspect ratio distortion. — image_smart_resize","title":"Resize images to a target size without aspect ratio distortion. — image_smart_resize","text":"Image datasets typically yield images different size. However, images need batched can processed Keras layers. batched, images need share height width. simply , TF (JAX equivalent):   However, , distort aspect ratio images, since general aspect ratio size. fine many cases, always (e.g. image generation models can problem). Note passing argument preserve_aspect_ratio = TRUE tf$image$resize() preserve aspect ratio, cost longer respecting provided target size. calls :   output images actually (200, 200), distorted. Instead, parts image fit within target size get cropped . resizing process : Take largest centered crop image aspect ratio target size. instance, size = c(200, 200) input image size (340, 500), take crop (340, 340) centered along width. Resize cropped image target size. example , resize (340, 340) crop (200, 200).","code":"size <- c(200, 200) ds <- ds$map(\\(img) tf$image$resize(img, size)) size <- c(200, 200) ds <- ds$map(\\(img) image_smart_resize(img, size))"},{"path":"https://keras3.posit.co/dev/reference/image_smart_resize.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Resize images to a target size without aspect ratio distortion. — image_smart_resize","text":"","code":"image_smart_resize(   x,   size,   interpolation = \"bilinear\",   data_format = \"channels_last\",   backend_module = NULL )"},{"path":"https://keras3.posit.co/dev/reference/image_smart_resize.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Resize images to a target size without aspect ratio distortion. — image_smart_resize","text":"x Input image batch images (tensor array). Must format (height, width, channels) (batch_size, height, width, channels). size Tuple (height, width) integer. Target size. interpolation String, interpolation use resizing. Supports \"bilinear\", \"nearest\", \"bicubic\", \"lanczos3\", \"lanczos5\". Defaults 'bilinear'. data_format \"channels_last\" \"channels_first\". backend_module Backend module use (different default backend).","code":""},{"path":"https://keras3.posit.co/dev/reference/image_smart_resize.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Resize images to a target size without aspect ratio distortion. — image_smart_resize","text":"Array shape (size[1], size[2], channels). input image array, output array, backend-native tensor, output backend-native tensor.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/image_to_array.html","id":null,"dir":"Reference","previous_headings":"","what":"Converts a PIL Image instance to a matrix. — image_to_array","title":"Converts a PIL Image instance to a matrix. — image_to_array","text":"Converts PIL Image instance matrix.","code":""},{"path":"https://keras3.posit.co/dev/reference/image_to_array.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Converts a PIL Image instance to a matrix. — image_to_array","text":"","code":"image_to_array(img, data_format = NULL, dtype = NULL)"},{"path":"https://keras3.posit.co/dev/reference/image_to_array.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Converts a PIL Image instance to a matrix. — image_to_array","text":"img Input PIL Image instance. data_format Image data format, can either \"channels_first\" \"channels_last\". Defaults NULL, case global setting config_image_data_format() used (unless changed , defaults \"channels_last\"). dtype Dtype use. NULL \"double\" return R double array. \"integer\" returns R integer array. values (e.g., \"float32\") return NumPy array.","code":""},{"path":"https://keras3.posit.co/dev/reference/image_to_array.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Converts a PIL Image instance to a matrix. — image_to_array","text":"3D array.","code":""},{"path":"https://keras3.posit.co/dev/reference/image_to_array.html","id":"example","dir":"Reference","previous_headings":"","what":"Example","title":"Converts a PIL Image instance to a matrix. — image_to_array","text":"","code":"image_path <- get_file(origin = \"https://www.r-project.org/logo/Rlogo.png\") (img <- image_load(image_path)) ## <PIL.Image.Image image mode=RGB size=724x561 at 0x0> array <- image_to_array(img) str(array) ##  num [1:561, 1:724, 1:3] 0 0 0 0 0 0 0 0 0 0 ..."},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/imagenet_decode_predictions.html","id":null,"dir":"Reference","previous_headings":"","what":"Decodes the prediction of an ImageNet model. — imagenet_decode_predictions","title":"Decodes the prediction of an ImageNet model. — imagenet_decode_predictions","text":"Decodes prediction ImageNet model.","code":""},{"path":"https://keras3.posit.co/dev/reference/imagenet_decode_predictions.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Decodes the prediction of an ImageNet model. — imagenet_decode_predictions","text":"","code":"imagenet_decode_predictions(preds, top = 5)"},{"path":"https://keras3.posit.co/dev/reference/imagenet_decode_predictions.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Decodes the prediction of an ImageNet model. — imagenet_decode_predictions","text":"preds Tensor encoding batch predictions. top integer, many top-guesses return.","code":""},{"path":"https://keras3.posit.co/dev/reference/imagenet_decode_predictions.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Decodes the prediction of an ImageNet model. — imagenet_decode_predictions","text":"List data frames variables class_name, class_description, score (one data frame per sample batch input).","code":""},{"path":"https://keras3.posit.co/dev/reference/imagenet_preprocess_input.html","id":null,"dir":"Reference","previous_headings":"","what":"Preprocesses a tensor or array encoding a batch of images. — imagenet_preprocess_input","title":"Preprocesses a tensor or array encoding a batch of images. — imagenet_preprocess_input","text":"Preprocesses tensor array encoding batch images.","code":""},{"path":"https://keras3.posit.co/dev/reference/imagenet_preprocess_input.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Preprocesses a tensor or array encoding a batch of images. — imagenet_preprocess_input","text":"","code":"imagenet_preprocess_input(x, data_format = NULL, mode = \"caffe\")"},{"path":"https://keras3.posit.co/dev/reference/imagenet_preprocess_input.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Preprocesses a tensor or array encoding a batch of images. — imagenet_preprocess_input","text":"x Input Numpy symbolic tensor, 3D 4D. data_format Data format image tensor/array. mode One \"caffe\", \"tf\", \"torch\" caffe: convert images RGB BGR, zero-center color channel respect ImageNet dataset, without scaling. tf: scale pixels -1 1, sample-wise. torch: scale pixels 0 1 normalize channel respect ImageNet dataset.","code":""},{"path":"https://keras3.posit.co/dev/reference/imagenet_preprocess_input.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Preprocesses a tensor or array encoding a batch of images. — imagenet_preprocess_input","text":"Preprocessed tensor array.","code":""},{"path":"https://keras3.posit.co/dev/reference/initializer_constant.html","id":null,"dir":"Reference","previous_headings":"","what":"Initializer that generates tensors with constant values. — initializer_constant","title":"Initializer that generates tensors with constant values. — initializer_constant","text":"scalar values allowed. constant value provided must convertible dtype requested calling initializer.","code":""},{"path":"https://keras3.posit.co/dev/reference/initializer_constant.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Initializer that generates tensors with constant values. — initializer_constant","text":"","code":"initializer_constant(value = 0)"},{"path":"https://keras3.posit.co/dev/reference/initializer_constant.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Initializer that generates tensors with constant values. — initializer_constant","text":"value numeric scalar.","code":""},{"path":"https://keras3.posit.co/dev/reference/initializer_constant.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Initializer that generates tensors with constant values. — initializer_constant","text":"Initializer instance can passed layer variable constructors, called directly shape return Tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/initializer_constant.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Initializer that generates tensors with constant values. — initializer_constant","text":"","code":"# Standalone usage: initializer <- initializer_constant(10) values <- initializer(shape = c(2, 2)) # Usage in a Keras layer: initializer <- initializer_constant(10) layer <- layer_dense(units = 3, kernel_initializer = initializer)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/initializer_glorot_normal.html","id":null,"dir":"Reference","previous_headings":"","what":"The Glorot normal initializer, also called Xavier normal initializer. — initializer_glorot_normal","title":"The Glorot normal initializer, also called Xavier normal initializer. — initializer_glorot_normal","text":"Draws samples truncated normal distribution centered 0 stddev = sqrt(2 / (fan_in + fan_out)) fan_in number input units weight tensor fan_out number output units weight tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/initializer_glorot_normal.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"The Glorot normal initializer, also called Xavier normal initializer. — initializer_glorot_normal","text":"","code":"initializer_glorot_normal(seed = NULL)"},{"path":"https://keras3.posit.co/dev/reference/initializer_glorot_normal.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"The Glorot normal initializer, also called Xavier normal initializer. — initializer_glorot_normal","text":"seed integer instance random_seed_generator(). Used make behavior initializer deterministic. Note initializer seeded integer NULL (unseeded) produce random values across multiple calls. get different random values across multiple calls, use seed instance random_seed_generator().","code":""},{"path":"https://keras3.posit.co/dev/reference/initializer_glorot_normal.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"The Glorot normal initializer, also called Xavier normal initializer. — initializer_glorot_normal","text":"Initializer instance can passed layer variable constructors, called directly shape return Tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/initializer_glorot_normal.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"The Glorot normal initializer, also called Xavier normal initializer. — initializer_glorot_normal","text":"","code":"# Standalone usage: initializer <- initializer_glorot_normal() values <- initializer(shape = c(2, 2)) # Usage in a Keras layer: initializer <- initializer_glorot_normal() layer <- layer_dense(units = 3, kernel_initializer = initializer)"},{"path":"https://keras3.posit.co/dev/reference/initializer_glorot_normal.html","id":"reference","dir":"Reference","previous_headings":"","what":"Reference","title":"The Glorot normal initializer, also called Xavier normal initializer. — initializer_glorot_normal","text":"Glorot et al., 2010","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/initializer_glorot_uniform.html","id":null,"dir":"Reference","previous_headings":"","what":"The Glorot uniform initializer, also called Xavier uniform initializer. — initializer_glorot_uniform","title":"The Glorot uniform initializer, also called Xavier uniform initializer. — initializer_glorot_uniform","text":"Draws samples uniform distribution within [-limit, limit], limit = sqrt(6 / (fan_in + fan_out)) (fan_in number input units weight tensor fan_out number output units).","code":""},{"path":"https://keras3.posit.co/dev/reference/initializer_glorot_uniform.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"The Glorot uniform initializer, also called Xavier uniform initializer. — initializer_glorot_uniform","text":"","code":"initializer_glorot_uniform(seed = NULL)"},{"path":"https://keras3.posit.co/dev/reference/initializer_glorot_uniform.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"The Glorot uniform initializer, also called Xavier uniform initializer. — initializer_glorot_uniform","text":"seed integer instance random_seed_generator(). Used make behavior initializer deterministic. Note initializer seeded integer NULL (unseeded) produce random values across multiple calls. get different random values across multiple calls, use seed instance random_seed_generator().","code":""},{"path":"https://keras3.posit.co/dev/reference/initializer_glorot_uniform.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"The Glorot uniform initializer, also called Xavier uniform initializer. — initializer_glorot_uniform","text":"Initializer instance can passed layer variable constructors, called directly shape return Tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/initializer_glorot_uniform.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"The Glorot uniform initializer, also called Xavier uniform initializer. — initializer_glorot_uniform","text":"","code":"# Standalone usage: initializer <- initializer_glorot_uniform() values <- initializer(shape = c(2, 2)) # Usage in a Keras layer: initializer <- initializer_glorot_uniform() layer <- layer_dense(units = 3, kernel_initializer = initializer)"},{"path":"https://keras3.posit.co/dev/reference/initializer_glorot_uniform.html","id":"reference","dir":"Reference","previous_headings":"","what":"Reference","title":"The Glorot uniform initializer, also called Xavier uniform initializer. — initializer_glorot_uniform","text":"Glorot et al., 2010","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/initializer_he_normal.html","id":null,"dir":"Reference","previous_headings":"","what":"He normal initializer. — initializer_he_normal","title":"He normal initializer. — initializer_he_normal","text":"draws samples truncated normal distribution centered 0 stddev = sqrt(2 / fan_in) fan_in number input units weight tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/initializer_he_normal.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"He normal initializer. — initializer_he_normal","text":"","code":"initializer_he_normal(seed = NULL)"},{"path":"https://keras3.posit.co/dev/reference/initializer_he_normal.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"He normal initializer. — initializer_he_normal","text":"seed integer instance random_seed_generator(). Used make behavior initializer deterministic. Note initializer seeded integer NULL (unseeded) produce random values across multiple calls. get different random values across multiple calls, use seed instance random_seed_generator().","code":""},{"path":"https://keras3.posit.co/dev/reference/initializer_he_normal.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"He normal initializer. — initializer_he_normal","text":"Initializer instance can passed layer variable constructors, called directly shape return Tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/initializer_he_normal.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"He normal initializer. — initializer_he_normal","text":"","code":"# Standalone usage: initializer <- initializer_he_normal() values <- initializer(shape = c(2, 2)) # Usage in a Keras layer: initializer <- initializer_he_normal() layer <- layer_dense(units = 3, kernel_initializer = initializer)"},{"path":"https://keras3.posit.co/dev/reference/initializer_he_normal.html","id":"reference","dir":"Reference","previous_headings":"","what":"Reference","title":"He normal initializer. — initializer_he_normal","text":"et al., 2015","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/initializer_he_uniform.html","id":null,"dir":"Reference","previous_headings":"","what":"He uniform variance scaling initializer. — initializer_he_uniform","title":"He uniform variance scaling initializer. — initializer_he_uniform","text":"Draws samples uniform distribution within [-limit, limit], limit = sqrt(6 / fan_in) (fan_in number input units weight tensor).","code":""},{"path":"https://keras3.posit.co/dev/reference/initializer_he_uniform.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"He uniform variance scaling initializer. — initializer_he_uniform","text":"","code":"initializer_he_uniform(seed = NULL)"},{"path":"https://keras3.posit.co/dev/reference/initializer_he_uniform.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"He uniform variance scaling initializer. — initializer_he_uniform","text":"seed integer instance random_seed_generator(). Used make behavior initializer deterministic. Note initializer seeded integer NULL (unseeded) produce random values across multiple calls. get different random values across multiple calls, use seed instance random_seed_generator().","code":""},{"path":"https://keras3.posit.co/dev/reference/initializer_he_uniform.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"He uniform variance scaling initializer. — initializer_he_uniform","text":"Initializer instance can passed layer variable constructors, called directly shape return Tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/initializer_he_uniform.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"He uniform variance scaling initializer. — initializer_he_uniform","text":"","code":"# Standalone usage: initializer <- initializer_he_uniform() values <- initializer(shape = c(2, 2)) # Usage in a Keras layer: initializer <- initializer_he_uniform() layer <- layer_dense(units = 3, kernel_initializer = initializer)"},{"path":"https://keras3.posit.co/dev/reference/initializer_he_uniform.html","id":"reference","dir":"Reference","previous_headings":"","what":"Reference","title":"He uniform variance scaling initializer. — initializer_he_uniform","text":"et al., 2015","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/initializer_identity.html","id":null,"dir":"Reference","previous_headings":"","what":"Initializer that generates the identity matrix. — initializer_identity","title":"Initializer that generates the identity matrix. — initializer_identity","text":"usable generating 2D matrices.","code":""},{"path":"https://keras3.posit.co/dev/reference/initializer_identity.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Initializer that generates the identity matrix. — initializer_identity","text":"","code":"initializer_identity(gain = 1)"},{"path":"https://keras3.posit.co/dev/reference/initializer_identity.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Initializer that generates the identity matrix. — initializer_identity","text":"gain Multiplicative factor apply identity matrix.","code":""},{"path":"https://keras3.posit.co/dev/reference/initializer_identity.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Initializer that generates the identity matrix. — initializer_identity","text":"Initializer instance can passed layer variable constructors, called directly shape return Tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/initializer_identity.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Initializer that generates the identity matrix. — initializer_identity","text":"","code":"# Standalone usage: initializer <- initializer_identity() values <- initializer(shape = c(2, 2)) # Usage in a Keras layer: initializer <- initializer_identity() layer <- layer_dense(units = 3, kernel_initializer = initializer)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/initializer_lecun_normal.html","id":null,"dir":"Reference","previous_headings":"","what":"Lecun normal initializer. — initializer_lecun_normal","title":"Lecun normal initializer. — initializer_lecun_normal","text":"Initializers allow pre-specify initialization strategy, encoded Initializer object, without knowing shape dtype variable initialized. Draws samples truncated normal distribution centered 0 stddev = sqrt(1 / fan_in) fan_in number input units weight tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/initializer_lecun_normal.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Lecun normal initializer. — initializer_lecun_normal","text":"","code":"initializer_lecun_normal(seed = NULL)"},{"path":"https://keras3.posit.co/dev/reference/initializer_lecun_normal.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Lecun normal initializer. — initializer_lecun_normal","text":"seed integer instance random_seed_generator(). Used make behavior initializer deterministic. Note initializer seeded integer NULL (unseeded) produce random values across multiple calls. get different random values across multiple calls, use seed instance random_seed_generator().","code":""},{"path":"https://keras3.posit.co/dev/reference/initializer_lecun_normal.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Lecun normal initializer. — initializer_lecun_normal","text":"Initializer instance can passed layer variable constructors, called directly shape return Tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/initializer_lecun_normal.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Lecun normal initializer. — initializer_lecun_normal","text":"","code":"# Standalone usage: initializer <- initializer_lecun_normal() values <- initializer(shape = c(2, 2)) # Usage in a Keras layer: initializer <- initializer_lecun_normal() layer <- layer_dense(units = 3, kernel_initializer = initializer)"},{"path":"https://keras3.posit.co/dev/reference/initializer_lecun_normal.html","id":"reference","dir":"Reference","previous_headings":"","what":"Reference","title":"Lecun normal initializer. — initializer_lecun_normal","text":"Klambauer et al., 2017","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/initializer_lecun_uniform.html","id":null,"dir":"Reference","previous_headings":"","what":"Lecun uniform initializer. — initializer_lecun_uniform","title":"Lecun uniform initializer. — initializer_lecun_uniform","text":"Draws samples uniform distribution within [-limit, limit], limit = sqrt(3 / fan_in) (fan_in number input units weight tensor).","code":""},{"path":"https://keras3.posit.co/dev/reference/initializer_lecun_uniform.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Lecun uniform initializer. — initializer_lecun_uniform","text":"","code":"initializer_lecun_uniform(seed = NULL)"},{"path":"https://keras3.posit.co/dev/reference/initializer_lecun_uniform.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Lecun uniform initializer. — initializer_lecun_uniform","text":"seed integer instance random_seed_generator(). Used make behavior initializer deterministic. Note initializer seeded integer NULL (unseeded) produce random values across multiple calls. get different random values across multiple calls, use seed instance random_seed_generator().","code":""},{"path":"https://keras3.posit.co/dev/reference/initializer_lecun_uniform.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Lecun uniform initializer. — initializer_lecun_uniform","text":"Initializer instance can passed layer variable constructors, called directly shape return Tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/initializer_lecun_uniform.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Lecun uniform initializer. — initializer_lecun_uniform","text":"","code":"# Standalone usage: initializer <- initializer_lecun_uniform() values <- initializer(shape = c(2, 2)) # Usage in a Keras layer: initializer <- initializer_lecun_uniform() layer <- layer_dense(units = 3, kernel_initializer = initializer)"},{"path":"https://keras3.posit.co/dev/reference/initializer_lecun_uniform.html","id":"reference","dir":"Reference","previous_headings":"","what":"Reference","title":"Lecun uniform initializer. — initializer_lecun_uniform","text":"Klambauer et al., 2017","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/initializer_ones.html","id":null,"dir":"Reference","previous_headings":"","what":"Initializer that generates tensors initialized to 1. — initializer_ones","title":"Initializer that generates tensors initialized to 1. — initializer_ones","text":"Also available via shortcut function ones.","code":""},{"path":"https://keras3.posit.co/dev/reference/initializer_ones.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Initializer that generates tensors initialized to 1. — initializer_ones","text":"","code":"initializer_ones()"},{"path":"https://keras3.posit.co/dev/reference/initializer_ones.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Initializer that generates tensors initialized to 1. — initializer_ones","text":"Initializer instance can passed layer variable constructors, called directly shape return Tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/initializer_ones.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Initializer that generates tensors initialized to 1. — initializer_ones","text":"","code":"# Standalone usage: initializer <- initializer_ones() values <- initializer(shape = c(2, 2)) # Usage in a Keras layer: initializer <- initializer_ones() layer <- layer_dense(units = 3, kernel_initializer = initializer)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/initializer_orthogonal.html","id":null,"dir":"Reference","previous_headings":"","what":"Initializer that generates an orthogonal matrix. — initializer_orthogonal","title":"Initializer that generates an orthogonal matrix. — initializer_orthogonal","text":"shape tensor initialize two-dimensional, initialized orthogonal matrix obtained QR decomposition matrix random numbers drawn normal distribution. matrix fewer rows columns output orthogonal rows. Otherwise, output orthogonal columns. shape tensor initialize two-dimensional, matrix shape (shape[1] * ... * shape[n - 1], shape[n]) initialized, n length shape vector. matrix subsequently reshaped give tensor desired shape.","code":""},{"path":"https://keras3.posit.co/dev/reference/initializer_orthogonal.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Initializer that generates an orthogonal matrix. — initializer_orthogonal","text":"","code":"initializer_orthogonal(gain = 1, seed = NULL)"},{"path":"https://keras3.posit.co/dev/reference/initializer_orthogonal.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Initializer that generates an orthogonal matrix. — initializer_orthogonal","text":"gain Multiplicative factor apply orthogonal matrix. seed integer. Used make behavior initializer deterministic.","code":""},{"path":"https://keras3.posit.co/dev/reference/initializer_orthogonal.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Initializer that generates an orthogonal matrix. — initializer_orthogonal","text":"Initializer instance can passed layer variable constructors, called directly shape return Tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/initializer_orthogonal.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Initializer that generates an orthogonal matrix. — initializer_orthogonal","text":"","code":"# Standalone usage: initializer <- initializer_orthogonal() values <- initializer(shape = c(2, 2)) # Usage in a Keras layer: initializer <- initializer_orthogonal() layer <- layer_dense(units = 3, kernel_initializer = initializer)"},{"path":"https://keras3.posit.co/dev/reference/initializer_orthogonal.html","id":"reference","dir":"Reference","previous_headings":"","what":"Reference","title":"Initializer that generates an orthogonal matrix. — initializer_orthogonal","text":"Saxe et al., 2014","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/initializer_random_normal.html","id":null,"dir":"Reference","previous_headings":"","what":"Random normal initializer. — initializer_random_normal","title":"Random normal initializer. — initializer_random_normal","text":"Draws samples normal distribution given parameters.","code":""},{"path":"https://keras3.posit.co/dev/reference/initializer_random_normal.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Random normal initializer. — initializer_random_normal","text":"","code":"initializer_random_normal(mean = 0, stddev = 0.05, seed = NULL)"},{"path":"https://keras3.posit.co/dev/reference/initializer_random_normal.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Random normal initializer. — initializer_random_normal","text":"mean numeric scalar. Mean random values generate. stddev numeric scalar. Standard deviation random values generate. seed integer instance random_seed_generator(). Used make behavior initializer deterministic. Note initializer seeded integer NULL (unseeded) produce random values across multiple calls. get different random values across multiple calls, use seed instance random_seed_generator().","code":""},{"path":"https://keras3.posit.co/dev/reference/initializer_random_normal.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Random normal initializer. — initializer_random_normal","text":"Initializer instance can passed layer variable constructors, called directly shape return Tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/initializer_random_normal.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Random normal initializer. — initializer_random_normal","text":"","code":"# Standalone usage: initializer <- initializer_random_normal(mean = 0.0, stddev = 1.0) values <- initializer(shape = c(2, 2)) # Usage in a Keras layer: initializer <- initializer_random_normal(mean = 0.0, stddev = 1.0) layer <- layer_dense(units = 3, kernel_initializer = initializer)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/initializer_random_uniform.html","id":null,"dir":"Reference","previous_headings":"","what":"Random uniform initializer. — initializer_random_uniform","title":"Random uniform initializer. — initializer_random_uniform","text":"Draws samples uniform distribution given parameters.","code":""},{"path":"https://keras3.posit.co/dev/reference/initializer_random_uniform.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Random uniform initializer. — initializer_random_uniform","text":"","code":"initializer_random_uniform(minval = -0.05, maxval = 0.05, seed = NULL)"},{"path":"https://keras3.posit.co/dev/reference/initializer_random_uniform.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Random uniform initializer. — initializer_random_uniform","text":"minval numeric scalar scalar keras tensor. Lower bound range random values generate (inclusive). maxval numeric scalar scalar keras tensor. Upper bound range random values generate (exclusive). seed integer instance random_seed_generator(). Used make behavior initializer deterministic. Note initializer seeded integer NULL (unseeded) produce random values across multiple calls. get different random values across multiple calls, use seed instance random_seed_generator().","code":""},{"path":"https://keras3.posit.co/dev/reference/initializer_random_uniform.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Random uniform initializer. — initializer_random_uniform","text":"Initializer instance can passed layer variable constructors, called directly shape return Tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/initializer_random_uniform.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Random uniform initializer. — initializer_random_uniform","text":"","code":"# Standalone usage: initializer <- initializer_random_uniform(minval = 0.0, maxval = 1.0) values <- initializer(shape = c(2, 2)) # Usage in a Keras layer: initializer <- initializer_random_uniform(minval = 0.0, maxval = 1.0) layer <- layer_dense(units = 3, kernel_initializer = initializer)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/initializer_stft.html","id":null,"dir":"Reference","previous_headings":"","what":"Initializer of Conv kernels for Short-term Fourier Transformation (STFT). — initializer_stft","title":"Initializer of Conv kernels for Short-term Fourier Transformation (STFT). — initializer_stft","text":"Since formula involves complex numbers, class compute either real imaginary components final output. Additionally, initializer supports windowing functions across time dimension commonly used STFT. Windowing functions Python module scipy.signal.windows supported, including common hann hamming windowing functions. layer supports periodic windows scaling-based normalization. primarily intended use STFTSpectrogram layer.","code":""},{"path":"https://keras3.posit.co/dev/reference/initializer_stft.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Initializer of Conv kernels for Short-term Fourier Transformation (STFT). — initializer_stft","text":"","code":"initializer_stft(   side = \"real\",   window = \"hann\",   scaling = \"density\",   periodic = FALSE )"},{"path":"https://keras3.posit.co/dev/reference/initializer_stft.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Initializer of Conv kernels for Short-term Fourier Transformation (STFT). — initializer_stft","text":"side String, \"real\" \"imag\" deciding kernel compute real side imaginary side output. Defaults \"real\". window String name windowing function scipy.signal.windows module, array_like window values, NULL windowing. scaling String, \"density\" \"spectrum\" scaling window normalization, either L2 L1 normalization. NULL scaling. periodic Boolean, True, window function treated periodic. Defaults FALSE.","code":""},{"path":"https://keras3.posit.co/dev/reference/initializer_stft.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Initializer of Conv kernels for Short-term Fourier Transformation (STFT). — initializer_stft","text":"Initializer instance can passed layer variable constructors, called directly shape return Tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/initializer_stft.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Initializer of Conv kernels for Short-term Fourier Transformation (STFT). — initializer_stft","text":"","code":"# Standalone usage: initializer <- initializer_stft(\"real\", \"hann\", \"density\", FALSE) values <- initializer(shape = c(128, 1, 513))"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/initializer_truncated_normal.html","id":null,"dir":"Reference","previous_headings":"","what":"Initializer that generates a truncated normal distribution. — initializer_truncated_normal","title":"Initializer that generates a truncated normal distribution. — initializer_truncated_normal","text":"values generated similar values RandomNormal initializer, except values two standard deviations mean discarded re-drawn.","code":""},{"path":"https://keras3.posit.co/dev/reference/initializer_truncated_normal.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Initializer that generates a truncated normal distribution. — initializer_truncated_normal","text":"","code":"initializer_truncated_normal(mean = 0, stddev = 0.05, seed = NULL)"},{"path":"https://keras3.posit.co/dev/reference/initializer_truncated_normal.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Initializer that generates a truncated normal distribution. — initializer_truncated_normal","text":"mean numeric scalar. Mean random values generate. stddev numeric scalar. Standard deviation random values generate. seed integer instance random_seed_generator(). Used make behavior initializer deterministic. Note initializer seeded integer NULL (unseeded) produce random values across multiple calls. get different random values across multiple calls, use seed instance random_seed_generator().","code":""},{"path":"https://keras3.posit.co/dev/reference/initializer_truncated_normal.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Initializer that generates a truncated normal distribution. — initializer_truncated_normal","text":"Initializer instance can passed layer variable constructors, called directly shape return Tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/initializer_truncated_normal.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Initializer that generates a truncated normal distribution. — initializer_truncated_normal","text":"","code":"# Standalone usage: initializer <- initializer_truncated_normal(mean = 0, stddev = 1) values <- initializer(shape = c(2, 2)) # Usage in a Keras layer: initializer <- initializer_truncated_normal(mean = 0, stddev = 1) layer <- layer_dense(units = 3, kernel_initializer = initializer)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/initializer_variance_scaling.html","id":null,"dir":"Reference","previous_headings":"","what":"Initializer that adapts its scale to the shape of its input tensors. — initializer_variance_scaling","title":"Initializer that adapts its scale to the shape of its input tensors. — initializer_variance_scaling","text":"distribution = \"truncated_normal\" \"untruncated_normal\", samples drawn truncated/untruncated normal distribution mean zero standard deviation (truncation, used) stddev = sqrt(scale / n), n : number input units weight tensor, mode = \"fan_in\" number output units, mode = \"fan_out\" average numbers input output units, mode = \"fan_avg\" distribution = \"uniform\", samples drawn uniform distribution within [-limit, limit], limit = sqrt(3 * scale / n).","code":""},{"path":"https://keras3.posit.co/dev/reference/initializer_variance_scaling.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Initializer that adapts its scale to the shape of its input tensors. — initializer_variance_scaling","text":"","code":"initializer_variance_scaling(   scale = 1,   mode = \"fan_in\",   distribution = \"truncated_normal\",   seed = NULL )"},{"path":"https://keras3.posit.co/dev/reference/initializer_variance_scaling.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Initializer that adapts its scale to the shape of its input tensors. — initializer_variance_scaling","text":"scale Scaling factor (positive float). mode One \"fan_in\", \"fan_out\", \"fan_avg\". distribution Random distribution use. One \"truncated_normal\", \"untruncated_normal\", \"uniform\". seed integer instance random_seed_generator(). Used make behavior initializer deterministic. Note initializer seeded integer NULL (unseeded) produce random values across multiple calls. get different random values across multiple calls, use seed instance random_seed_generator().","code":""},{"path":"https://keras3.posit.co/dev/reference/initializer_variance_scaling.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Initializer that adapts its scale to the shape of its input tensors. — initializer_variance_scaling","text":"Initializer instance can passed layer variable constructors, called directly shape return Tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/initializer_variance_scaling.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Initializer that adapts its scale to the shape of its input tensors. — initializer_variance_scaling","text":"","code":"# Standalone usage: initializer <- initializer_variance_scaling(scale = 0.1, mode = 'fan_in',                                             distribution = 'uniform') values <- initializer(shape = c(2, 2)) # Usage in a Keras layer: initializer <- initializer_variance_scaling(scale = 0.1, mode = 'fan_in',                                             distribution = 'uniform') layer <- layer_dense(units = 3, kernel_initializer = initializer)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/initializer_zeros.html","id":null,"dir":"Reference","previous_headings":"","what":"Initializer that generates tensors initialized to 0. — initializer_zeros","title":"Initializer that generates tensors initialized to 0. — initializer_zeros","text":"Initializer generates tensors initialized 0.","code":""},{"path":"https://keras3.posit.co/dev/reference/initializer_zeros.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Initializer that generates tensors initialized to 0. — initializer_zeros","text":"","code":"initializer_zeros()"},{"path":"https://keras3.posit.co/dev/reference/initializer_zeros.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Initializer that generates tensors initialized to 0. — initializer_zeros","text":"Initializer instance can passed layer variable constructors, called directly shape return Tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/initializer_zeros.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Initializer that generates tensors initialized to 0. — initializer_zeros","text":"","code":"# Standalone usage: initializer <- initializer_zeros() values <- initializer(shape = c(2, 2)) # Usage in a Keras layer: initializer <- initializer_zeros() layer <- layer_dense(units = 3, kernel_initializer = initializer)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/install_keras.html","id":null,"dir":"Reference","previous_headings":"","what":"Install Keras — install_keras","title":"Install Keras — install_keras","text":"function install Keras along selected backend, including Python dependencies.","code":""},{"path":"https://keras3.posit.co/dev/reference/install_keras.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Install Keras — install_keras","text":"","code":"install_keras(   envname = \"r-keras\",   ...,   extra_packages = c(\"scipy\", \"pandas\", \"Pillow\", \"pydot\", \"ipython\",     \"tensorflow_datasets\"),   python_version = \">=3.9,<=3.11\",   backend = c(\"tensorflow\", \"jax\"),   gpu = NA,   restart_session = TRUE )"},{"path":"https://keras3.posit.co/dev/reference/install_keras.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Install Keras — install_keras","text":"envname Name path Python virtual environment ... reserved future compatibility. extra_packages Additional Python packages install alongside Keras python_version Passed reticulate::virtualenv_starter() backend backend(s) install. Accepted values include \"tensorflow\", \"jax\" \"torch\" gpu whether install GPU capable version backend. restart_session Whether restart R session installing (note occur within RStudio).","code":""},{"path":"https://keras3.posit.co/dev/reference/install_keras.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Install Keras — install_keras","text":"return value, called side effects.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/keras.html","id":null,"dir":"Reference","previous_headings":"","what":"Main Keras module — keras","title":"Main Keras module — keras","text":"keras module object equivalent reticulate::import(\"keras\") provided mainly convenience.","code":""},{"path":"https://keras3.posit.co/dev/reference/keras.html","id":"format","dir":"Reference","previous_headings":"","what":"Format","title":"Main Keras module — keras","text":"object class python.builtin.module","code":""},{"path":"https://keras3.posit.co/dev/reference/keras.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Main Keras module — keras","text":"keras Python module","code":""},{"path":"https://keras3.posit.co/dev/reference/keras3-package.html","id":null,"dir":"Reference","previous_headings":"","what":"keras3: R Interface to 'Keras' — keras3-package","title":"keras3: R Interface to 'Keras' — keras3-package","text":"Interface 'Keras' https://keras.io, high-level neural networks API. 'Keras' developed focus enabling fast experimentation, supports convolution based networks recurrent networks (well combinations two), runs seamlessly CPU GPU devices. Keras high-level neural networks API, developed focus enabling fast experimentation. Keras following key features:","code":""},{"path":"https://keras3.posit.co/dev/reference/keras3-package.html","id":"details","dir":"Reference","previous_headings":"","what":"Details","title":"keras3: R Interface to 'Keras' — keras3-package","text":"Allows code run CPU GPU, seamlessly. User-friendly API makes easy quickly prototype deep learning models. Built-support convolutional networks (computer vision), recurrent networks (sequence processing), combination . Supports arbitrary network architectures: multi-input multi-output models, layer sharing, model sharing, etc. means Keras appropriate building essentially deep learning model, memory network neural Turing machine. capable running top multiple back-ends including TensorFlow, Jax, PyTorch. See package website https://keras3.posit.co complete documentation.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/keras3-package.html","id":"author","dir":"Reference","previous_headings":"","what":"Author","title":"keras3: R Interface to 'Keras' — keras3-package","text":"Maintainer: Tomasz Kalinowski tomasz@posit.co [copyright holder] Authors: JJ Allaire [copyright holder] François Chollet [copyright holder] contributors: Daniel Falbel daniel@posit.co [contributor, copyright holder] Posit Software, PBC [copyright holder, funder] Google [copyright holder, funder] Yuan Tang terrytangyuan@gmail.com (ORCID) [contributor, copyright holder] Wouter Van Der Bijl [contributor, copyright holder] Martin Studer [contributor, copyright holder] Sigrid Keydana [contributor]","code":""},{"path":"https://keras3.posit.co/dev/reference/keras_input.html","id":null,"dir":"Reference","previous_headings":"","what":"Create a Keras tensor (Functional API input). — keras_input","title":"Create a Keras tensor (Functional API input). — keras_input","text":"Keras tensor symbolic tensor-like object, augment certain attributes allow us build Keras model just knowing inputs outputs model. instance, , b c Keras tensors, becomes possible : model <- keras_model(input = c(, b), output = c)","code":""},{"path":"https://keras3.posit.co/dev/reference/keras_input.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Create a Keras tensor (Functional API input). — keras_input","text":"","code":"keras_input(   shape = NULL,   batch_size = NULL,   dtype = NULL,   sparse = NULL,   ragged = NULL,   batch_shape = NULL,   name = NULL,   tensor = NULL,   optional = FALSE )"},{"path":"https://keras3.posit.co/dev/reference/keras_input.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Create a Keras tensor (Functional API input). — keras_input","text":"shape shape list (list integers NULL objects), including batch size. instance, shape = c(32) indicates expected input batches 32-dimensional vectors. Elements list can NULL NA; NULL/NA elements represent dimensions shape known may vary (e.g. sequence length). batch_size Optional static batch size (integer). dtype data type expected input, string (e.g. \"float32\", \"int32\"...) sparse boolean specifying whether expected input sparse tensors. Note , sparse FALSE, sparse tensors can still passed input - densified default value 0. feature supported TensorFlow backend. Defaults FALSE. ragged boolean specifying whether expected input ragged tensors. Note , ragged FALSE, ragged tensors can still passed input - densified default value 0. feature supported TensorFlow backend. Defaults FALSE. batch_shape Optional shape list (list integers NULL objects), including batch size. name Optional name string layer. unique model (reuse name twice). autogenerated provided. tensor Optional existing tensor wrap Input layer. set, layer use tensor rather creating new placeholder tensor. optional Boolean, whether input optional . optional input can accept NULL values.","code":""},{"path":"https://keras3.posit.co/dev/reference/keras_input.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Create a Keras tensor (Functional API input). — keras_input","text":"Keras tensor, can passed inputs argument (keras_model()).","code":""},{"path":"https://keras3.posit.co/dev/reference/keras_input.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Create a Keras tensor (Functional API input). — keras_input","text":"","code":"# This is a logistic regression in Keras input <- layer_input(shape=c(32)) output <- input |> layer_dense(16, activation='softmax') model <- keras_model(input, output)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/keras_model.html","id":null,"dir":"Reference","previous_headings":"","what":"Keras Model (Functional API) — keras_model","title":"Keras Model (Functional API) — keras_model","text":"model directed acyclic graph layers.","code":""},{"path":"https://keras3.posit.co/dev/reference/keras_model.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Keras Model (Functional API) — keras_model","text":"","code":"keras_model(inputs = NULL, outputs = NULL, ...)"},{"path":"https://keras3.posit.co/dev/reference/keras_model.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Keras Model (Functional API) — keras_model","text":"inputs Input tensor(s) (keras_input()) outputs Output tensors (calling layers inputs) ... additional arguments","code":""},{"path":"https://keras3.posit.co/dev/reference/keras_model.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Keras Model (Functional API) — keras_model","text":"Model instance.","code":""},{"path":"https://keras3.posit.co/dev/reference/keras_model.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Keras Model (Functional API) — keras_model","text":"","code":"library(keras3)  # input tensor inputs <- keras_input(shape = c(784))  # outputs compose input + dense layers predictions <- inputs |>   layer_dense(units = 64, activation = 'relu') |>   layer_dense(units = 64, activation = 'relu') |>   layer_dense(units = 10, activation = 'softmax')  # create and compile model model <- keras_model(inputs = inputs, outputs = predictions) model |> compile(   optimizer = 'rmsprop',   loss = 'categorical_crossentropy',   metrics = c('accuracy') )"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/keras_model_sequential.html","id":null,"dir":"Reference","previous_headings":"","what":"Keras Model composed of a linear stack of layers — keras_model_sequential","title":"Keras Model composed of a linear stack of layers — keras_model_sequential","text":"Keras Model composed linear stack layers","code":""},{"path":"https://keras3.posit.co/dev/reference/keras_model_sequential.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Keras Model composed of a linear stack of layers — keras_model_sequential","text":"","code":"keras_model_sequential(   input_shape = NULL,   name = NULL,   ...,   input_dtype = NULL,   input_batch_size = NULL,   input_sparse = NULL,   input_ragged = NULL,   input_batch_shape = NULL,   input_name = NULL,   input_tensor = NULL,   input_optional = FALSE,   trainable = TRUE,   layers = list() )"},{"path":"https://keras3.posit.co/dev/reference/keras_model_sequential.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Keras Model composed of a linear stack of layers — keras_model_sequential","text":"input_shape shape integer vector, including batch size. instance, shape=c(32) indicates expected input batches 32-dimensional vectors. Elements shape can NA; NA elements represent dimensions shape known may vary (e.g. sequence length). name Name model ... additional arguments passed keras.layers.InputLayer. input_dtype data type expected input, string (e.g. \"float32\", \"int32\"...) input_batch_size Optional static batch size (integer). input_sparse boolean specifying whether expected input sparse tensors. Note , sparse FALSE, sparse tensors can still passed input - densified default value 0. feature supported TensorFlow backend. Defaults FALSE. input_ragged bool. input_batch_shape optional way specify batch_size input_shape one argument. input_name Optional name string input layer. unique model (reuse name twice). autogenerated provided. input_tensor Optional existing tensor wrap InputLayer. set, layer use tensor rather creating new placeholder tensor. input_optional Boolean, whether input optional . optional input can accept NULL values. trainable Boolean, whether model's variables trainable. can also change trainable status model/layer freeze_weights() unfreeze_weights(). layers List layers add model.","code":""},{"path":"https://keras3.posit.co/dev/reference/keras_model_sequential.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Keras Model composed of a linear stack of layers — keras_model_sequential","text":"Sequential model instance.","code":""},{"path":"https://keras3.posit.co/dev/reference/keras_model_sequential.html","id":"note","dir":"Reference","previous_headings":"","what":"Note","title":"Keras Model composed of a linear stack of layers — keras_model_sequential","text":"input_shape omitted, model layer shapes, including final model output shape, known model built, either calling model input tensor/array like model(input), (possibly via fit()/evaluate()/predict()), explicitly calling model$build(input_shape).","code":""},{"path":"https://keras3.posit.co/dev/reference/keras_model_sequential.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Keras Model composed of a linear stack of layers — keras_model_sequential","text":"","code":"model <- keras_model_sequential(input_shape = c(784)) model |>   layer_dense(units = 32) |>   layer_activation('relu') |>   layer_dense(units = 10) |>   layer_activation('softmax')  model |> compile(   optimizer = 'rmsprop',   loss = 'categorical_crossentropy',   metrics = c('accuracy') )  model ## Model: \"sequential\" ## +---------------------------------+------------------------+---------------+ ## | Layer (type)                    | Output Shape           |       Param # | ## +=================================+========================+===============+ ## | dense (Dense)                   | (None, 32)             |        25,120 | ## +---------------------------------+------------------------+---------------+ ## | activation (Activation)         | (None, 32)             |             0 | ## +---------------------------------+------------------------+---------------+ ## | dense_1 (Dense)                 | (None, 10)             |           330 | ## +---------------------------------+------------------------+---------------+ ## | activation_1 (Activation)       | (None, 10)             |             0 | ## +---------------------------------+------------------------+---------------+ ##  Total params: 25,450 (99.41 KB) ##  Trainable params: 25,450 (99.41 KB) ##  Non-trainable params: 0 (0.00 B)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/keras_variable.html","id":null,"dir":"Reference","previous_headings":"","what":"Represents a backend-agnostic variable in Keras. — keras_variable","title":"Represents a backend-agnostic variable in Keras. — keras_variable","text":"Variable acts container state. holds tensor value can updated. JAX backend, variables used implement \"functionalization\", pattern lifting stateful operations piece computation turn stateless function.","code":""},{"path":"https://keras3.posit.co/dev/reference/keras_variable.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Represents a backend-agnostic variable in Keras. — keras_variable","text":"","code":"keras_variable(   initializer,   shape = NULL,   dtype = NULL,   trainable = TRUE,   autocast = TRUE,   aggregation = \"none\",   synchronization = \"auto\",   name = NULL,   ... )"},{"path":"https://keras3.posit.co/dev/reference/keras_variable.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Represents a backend-agnostic variable in Keras. — keras_variable","text":"initializer Initial value callable initialization. callable used, take arguments shape dtype. shape Optional. Tuple variable's shape. Required initializer callable. dtype Optional. Data type variable. Defaults global float dtype type (\"float32\" never configured). trainable Optional. Boolean indicating variable trainable. Defaults TRUE. autocast Optional. Boolean indicating whether variable supports autocasting. TRUE, layer may first convert variable compute data type accessed. Defaults TRUE. aggregation Optional string, one NULL, \"none\", \"mean\", \"sum\" \"only_first_replica\" specifying distributed variable aggregated. serves semantic annotation, taken account downstream backends users. Defaults \"none\". synchronization Optional string specifying distributed values synchronized. Defaults \"auto\". name Optional. unique name variable. Automatically generated set. ... Additional backend-specific keyword arguments forwarded keras$Variable().","code":""},{"path":"https://keras3.posit.co/dev/reference/keras_variable.html","id":"attributes","dir":"Reference","previous_headings":"","what":"Attributes","title":"Represents a backend-agnostic variable in Keras. — keras_variable","text":"shape: shape variable (tuple integers). ndim: number dimensions variable (integer). dtype: data type variable (string). trainable: Whether variable trainable (boolean). autocast: Whether variable supports autocasting (boolean). aggregation: distributed variable aggregated (string). synchronization: Strategy synchronizing variable across devices (string). value: current value variable (NumPy array tensor). name: name variable (string). path: path variable within Keras model layer (string).","code":""},{"path":"https://keras3.posit.co/dev/reference/keras_variable.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Represents a backend-agnostic variable in Keras. — keras_variable","text":"Initializing Variable NumPy array:   Using Keras initializer create Variable:           Marking Variable non-trainable:","code":"initial_array <- array(1, c(3, 3)) variable_from_array <- keras_variable(initializer = initial_array) variable_from_initializer <- keras_variable(   initializer = initializer_ones(),   shape = c(3, 3),   dtype = \"float32\" ) new_value <- array(0, c(3, 3)) variable_from_array$assign(new_value) ## tf.Tensor( ## [[0. 0. 0.] ##  [0. 0. 0.] ##  [0. 0. 0.]], shape=(3, 3), dtype=float64) # To modify a subset of values value <- variable_from_array$value value@r[1,] <- 99 invisible(variable_from_array$assign(value)) variable_from_array ## <Variable path=variable, shape=(3, 3), dtype=float64, value=[[99. 99. 99.] ##  [ 0.  0.  0.] ##  [ 0.  0.  0.]]> non_trainable_variable <- keras_variable(   initializer = array(1, c(3, 3)),   dtype = \"float32\",   trainable = FALSE )"},{"path":"https://keras3.posit.co/dev/reference/layer_activation.html","id":null,"dir":"Reference","previous_headings":"","what":"Applies an activation function to an output. — layer_activation","title":"Applies an activation function to an output. — layer_activation","text":"Applies activation function output.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_activation.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Applies an activation function to an output. — layer_activation","text":"","code":"layer_activation(object, activation, ...)"},{"path":"https://keras3.posit.co/dev/reference/layer_activation.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Applies an activation function to an output. — layer_activation","text":"object Object compose layer . tensor, array, sequential model. activation Activation function. callable, name activation keras3::activation_* namespace. ... Base layer keyword arguments, name dtype.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_activation.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Applies an activation function to an output. — layer_activation","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_activation.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Applies an activation function to an output. — layer_activation","text":"","code":"x <- array(c(-3, -1, 0, 2)) layer <- layer_activation(activation = 'relu') layer(x) ## tf.Tensor([0. 0. 0. 2.], shape=(4), dtype=float32) layer <- layer_activation(activation = activation_relu) layer(x) ## tf.Tensor([0. 0. 0. 2.], shape=(4), dtype=float32) layer <- layer_activation(activation = op_relu) layer(x) ## tf.Tensor([0. 0. 0. 2.], shape=(4), dtype=float32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_activation_elu.html","id":null,"dir":"Reference","previous_headings":"","what":"Applies an Exponential Linear Unit function to an output. — layer_activation_elu","title":"Applies an Exponential Linear Unit function to an output. — layer_activation_elu","text":"Formula:","code":"f(x) = alpha * (exp(x) - 1.) for x < 0 f(x) = x for x >= 0"},{"path":"https://keras3.posit.co/dev/reference/layer_activation_elu.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Applies an Exponential Linear Unit function to an output. — layer_activation_elu","text":"","code":"layer_activation_elu(object, alpha = 1, ...)"},{"path":"https://keras3.posit.co/dev/reference/layer_activation_elu.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Applies an Exponential Linear Unit function to an output. — layer_activation_elu","text":"object Object compose layer . tensor, array, sequential model. alpha float, slope negative section. Defaults 1.0. ... Base layer keyword arguments, name dtype.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_activation_elu.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Applies an Exponential Linear Unit function to an output. — layer_activation_elu","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_activation_leaky_relu.html","id":null,"dir":"Reference","previous_headings":"","what":"Leaky version of a Rectified Linear Unit activation layer. — layer_activation_leaky_relu","title":"Leaky version of a Rectified Linear Unit activation layer. — layer_activation_leaky_relu","text":"layer allows small gradient unit active. Formula:","code":"f <- function(x) ifelse(x >= 0, x, alpha * x)"},{"path":"https://keras3.posit.co/dev/reference/layer_activation_leaky_relu.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Leaky version of a Rectified Linear Unit activation layer. — layer_activation_leaky_relu","text":"","code":"layer_activation_leaky_relu(object, negative_slope = 0.3, ...)"},{"path":"https://keras3.posit.co/dev/reference/layer_activation_leaky_relu.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Leaky version of a Rectified Linear Unit activation layer. — layer_activation_leaky_relu","text":"object Object compose layer . tensor, array, sequential model. negative_slope Float >= 0.0. Negative slope coefficient. Defaults 0.3. ... Base layer keyword arguments, name dtype.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_activation_leaky_relu.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Leaky version of a Rectified Linear Unit activation layer. — layer_activation_leaky_relu","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_activation_leaky_relu.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Leaky version of a Rectified Linear Unit activation layer. — layer_activation_leaky_relu","text":"","code":"leaky_relu_layer <- layer_activation_leaky_relu(negative_slope=0.5) input <- array(c(-10, -5, 0.0, 5, 10)) result <- leaky_relu_layer(input) as.array(result) ## [1] -5.0 -2.5  0.0  5.0 10.0"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_activation_parametric_relu.html","id":null,"dir":"Reference","previous_headings":"","what":"Parametric Rectified Linear Unit activation layer. — layer_activation_parametric_relu","title":"Parametric Rectified Linear Unit activation layer. — layer_activation_parametric_relu","text":"Formula:   alpha learned array shape x.","code":"f <- function(x) ifelse(x >= 0, x, alpha * x)"},{"path":"https://keras3.posit.co/dev/reference/layer_activation_parametric_relu.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Parametric Rectified Linear Unit activation layer. — layer_activation_parametric_relu","text":"","code":"layer_activation_parametric_relu(   object,   alpha_initializer = \"Zeros\",   alpha_regularizer = NULL,   alpha_constraint = NULL,   shared_axes = NULL,   ... )"},{"path":"https://keras3.posit.co/dev/reference/layer_activation_parametric_relu.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Parametric Rectified Linear Unit activation layer. — layer_activation_parametric_relu","text":"object Object compose layer . tensor, array, sequential model. alpha_initializer Initializer function weights. alpha_regularizer Regularizer weights. alpha_constraint Constraint weights. shared_axes axes along share learnable parameters activation function. example, incoming feature maps 2D convolution output shape (batch, height, width, channels), wish share parameters across space filter one set parameters, set shared_axes=[1, 2]. ... Base layer keyword arguments, name dtype.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_activation_parametric_relu.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Parametric Rectified Linear Unit activation layer. — layer_activation_parametric_relu","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_activation_relu.html","id":null,"dir":"Reference","previous_headings":"","what":"Rectified Linear Unit activation function layer. — layer_activation_relu","title":"Rectified Linear Unit activation function layer. — layer_activation_relu","text":"Formula:","code":"f <- function(x, max_value = Inf, negative_slope = 0, threshold = 0) {  x <- max(x,0)  if (x >= max_value)    max_value  else if (threshold <= x && x < max_value)    x  else    negative_slope * (x - threshold) }"},{"path":"https://keras3.posit.co/dev/reference/layer_activation_relu.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Rectified Linear Unit activation function layer. — layer_activation_relu","text":"","code":"layer_activation_relu(   object,   max_value = NULL,   negative_slope = 0,   threshold = 0,   ... )"},{"path":"https://keras3.posit.co/dev/reference/layer_activation_relu.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Rectified Linear Unit activation function layer. — layer_activation_relu","text":"object Object compose layer . tensor, array, sequential model. max_value Float >= 0. Maximum activation value. NULL means unlimited. Defaults NULL. negative_slope Float >= 0. Negative slope coefficient. Defaults 0.0. threshold Float >= 0. Threshold value thresholded activation. Defaults 0.0. ... Base layer keyword arguments, name dtype.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_activation_relu.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Rectified Linear Unit activation function layer. — layer_activation_relu","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_activation_relu.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Rectified Linear Unit activation function layer. — layer_activation_relu","text":"","code":"relu_layer <- layer_activation_relu(max_value = 10,                                     negative_slope = 0.5,                                     threshold = 0) input <- array(c(-10, -5, 0.0, 5, 10)) result <- relu_layer(input) as.array(result) ## [1] -5.0 -2.5  0.0  5.0 10.0"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_activation_softmax.html","id":null,"dir":"Reference","previous_headings":"","what":"Softmax activation layer. — layer_activation_softmax","title":"Softmax activation layer. — layer_activation_softmax","text":"Formula:","code":"exp_x = exp(x - max(x)) f(x) = exp_x / sum(exp_x)"},{"path":"https://keras3.posit.co/dev/reference/layer_activation_softmax.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Softmax activation layer. — layer_activation_softmax","text":"","code":"layer_activation_softmax(object, axis = -1L, ...)"},{"path":"https://keras3.posit.co/dev/reference/layer_activation_softmax.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Softmax activation layer. — layer_activation_softmax","text":"object Object compose layer . tensor, array, sequential model. axis Integer, list Integers, axis along softmax normalization applied. ... Base layer keyword arguments, name dtype.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_activation_softmax.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Softmax activation layer. — layer_activation_softmax","text":"Softmaxed output shape inputs.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_activation_softmax.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Softmax activation layer. — layer_activation_softmax","text":"","code":"softmax_layer <- layer_activation_softmax() input <- op_array(c(1, 2, 1)) softmax_layer(input) ## tf.Tensor([0.21194157 0.5761169  0.21194157], shape=(3), dtype=float32)"},{"path":"https://keras3.posit.co/dev/reference/layer_activation_softmax.html","id":"call-arguments","dir":"Reference","previous_headings":"","what":"Call Arguments","title":"Softmax activation layer. — layer_activation_softmax","text":"inputs: inputs (logits) softmax layer. mask: boolean mask shape inputs. mask specifies 1 keep 0 mask. Defaults NULL.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_activity_regularization.html","id":null,"dir":"Reference","previous_headings":"","what":"Layer that applies an update to the cost function based input activity. — layer_activity_regularization","title":"Layer that applies an update to the cost function based input activity. — layer_activity_regularization","text":"Layer applies update cost function based input activity.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_activity_regularization.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Layer that applies an update to the cost function based input activity. — layer_activity_regularization","text":"","code":"layer_activity_regularization(object, l1 = 0, l2 = 0, ...)"},{"path":"https://keras3.posit.co/dev/reference/layer_activity_regularization.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Layer that applies an update to the cost function based input activity. — layer_activity_regularization","text":"object Object compose layer . tensor, array, sequential model. l1 L1 regularization factor (positive float). l2 L2 regularization factor (positive float). ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_activity_regularization.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Layer that applies an update to the cost function based input activity. — layer_activity_regularization","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_activity_regularization.html","id":"input-shape","dir":"Reference","previous_headings":"","what":"Input Shape","title":"Layer that applies an update to the cost function based input activity. — layer_activity_regularization","text":"Arbitrary. Use keyword argument input_shape (tuple integers, include samples axis) using layer first layer model.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_activity_regularization.html","id":"output-shape","dir":"Reference","previous_headings":"","what":"Output Shape","title":"Layer that applies an update to the cost function based input activity. — layer_activity_regularization","text":"shape input.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_add.html","id":null,"dir":"Reference","previous_headings":"","what":"Performs elementwise addition operation. — layer_add","title":"Performs elementwise addition operation. — layer_add","text":"takes input list tensors, shape, returns single tensor (also shape).","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_add.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Performs elementwise addition operation. — layer_add","text":"","code":"layer_add(inputs, ...)"},{"path":"https://keras3.posit.co/dev/reference/layer_add.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Performs elementwise addition operation. — layer_add","text":"inputs layers combine ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_add.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Performs elementwise addition operation. — layer_add","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_add.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Performs elementwise addition operation. — layer_add","text":"Usage Keras model:","code":"input_shape <- c(1, 2, 3) x1 <- op_ones(input_shape) x2 <- op_ones(input_shape) layer_add(x1, x2) ## tf.Tensor( ## [[[2. 2. 2.] ##   [2. 2. 2.]]], shape=(1, 2, 3), dtype=float32) input1 <- layer_input(shape = c(16)) x1 <- input1 |> layer_dense(8, activation = 'relu')  input2 <- layer_input(shape = c(32)) x2 <- input2 |> layer_dense(8, activation = 'relu')  # equivalent to `added = layer_add([x1, x2))` added <- layer_add(x1, x2) output <- added |> layer_dense(4)  model <- keras_model(inputs = c(input1, input2), outputs = output)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_additive_attention.html","id":null,"dir":"Reference","previous_headings":"","what":"Additive attention layer, a.k.a. Bahdanau-style attention. — layer_additive_attention","title":"Additive attention layer, a.k.a. Bahdanau-style attention. — layer_additive_attention","text":"Inputs list 2 3 elements: query tensor shape (batch_size, Tq, dim). value tensor shape (batch_size, Tv, dim). optional key tensor shape (batch_size, Tv, dim). none supplied, value used key. calculation follows steps: Calculate attention scores using query key shape (batch_size, Tq, Tv) non-linear sum scores = reduce_sum(tanh(query + key), axis=-1). Use scores calculate softmax distribution shape (batch_size, Tq, Tv). Use softmax distribution create linear combination value shape (batch_size, Tq, dim).","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_additive_attention.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Additive attention layer, a.k.a. Bahdanau-style attention. — layer_additive_attention","text":"","code":"layer_additive_attention(object, use_scale = TRUE, dropout = 0, ...)"},{"path":"https://keras3.posit.co/dev/reference/layer_additive_attention.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Additive attention layer, a.k.a. Bahdanau-style attention. — layer_additive_attention","text":"object Object compose layer . tensor, array, sequential model. use_scale TRUE, create scalar variable scale attention scores. dropout Float 0 1. Fraction units drop attention scores. Defaults 0.0. ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_additive_attention.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Additive attention layer, a.k.a. Bahdanau-style attention. — layer_additive_attention","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_additive_attention.html","id":"call-arguments","dir":"Reference","previous_headings":"","what":"Call Arguments","title":"Additive attention layer, a.k.a. Bahdanau-style attention. — layer_additive_attention","text":"inputs: List following tensors: query: Query tensor shape (batch_size, Tq, dim). value: Value tensor shape (batch_size, Tv, dim). key: Optional key tensor shape (batch_size, Tv, dim). given, use value key value, common case. mask: List following tensors: query_mask: boolean mask tensor shape (batch_size, Tq). given, output zero positions mask==FALSE. value_mask: boolean mask tensor shape (batch_size, Tv). given, apply mask values positions mask==FALSE contribute result. return_attention_scores: bool, TRUE, returns attention scores (masking softmax) additional output argument. training: Python boolean indicating whether layer behave training mode (adding dropout) inference mode (dropout). use_causal_mask: Boolean. Set TRUE decoder self-attention. Adds mask position attend positions j > . prevents flow information future towards past. Defaults FALSE.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_additive_attention.html","id":"output","dir":"Reference","previous_headings":"","what":"Output","title":"Additive attention layer, a.k.a. Bahdanau-style attention. — layer_additive_attention","text":"Attention outputs shape (batch_size, Tq, dim). (Optional) Attention scores masking softmax shape (batch_size, Tq, Tv).","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_alpha_dropout.html","id":null,"dir":"Reference","previous_headings":"","what":"Applies Alpha Dropout to the input. — layer_alpha_dropout","title":"Applies Alpha Dropout to the input. — layer_alpha_dropout","text":"Alpha Dropout Dropout keeps mean variance inputs original values, order ensure self-normalizing property even dropout. Alpha Dropout fits well Scaled Exponential Linear Units (SELU) randomly setting activations negative saturation value.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_alpha_dropout.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Applies Alpha Dropout to the input. — layer_alpha_dropout","text":"","code":"layer_alpha_dropout(object, rate, noise_shape = NULL, seed = NULL, ...)"},{"path":"https://keras3.posit.co/dev/reference/layer_alpha_dropout.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Applies Alpha Dropout to the input. — layer_alpha_dropout","text":"object Object compose layer . tensor, array, sequential model. rate Float 0 1. multiplicative noise standard deviation sqrt(rate / (1 - rate)). noise_shape 1D integer tensor representing shape binary alpha dropout mask multiplied input. instance, inputs shape (batch_size, timesteps, features) want alpha dropout mask timesteps, can use noise_shape = (batch_size, 1, features). seed integer use random seed. ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_alpha_dropout.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Applies Alpha Dropout to the input. — layer_alpha_dropout","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_alpha_dropout.html","id":"call-arguments","dir":"Reference","previous_headings":"","what":"Call Arguments","title":"Applies Alpha Dropout to the input. — layer_alpha_dropout","text":"inputs: Input tensor (rank). training: R boolean indicating whether layer behave training mode (adding alpha dropout) inference mode (nothing).","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_attention.html","id":null,"dir":"Reference","previous_headings":"","what":"Dot-product attention layer, a.k.a. Luong-style attention. — layer_attention","title":"Dot-product attention layer, a.k.a. Luong-style attention. — layer_attention","text":"Inputs list 2 3 elements: query tensor shape (batch_size, Tq, dim). value tensor shape (batch_size, Tv, dim). optional key tensor shape (batch_size, Tv, dim). none supplied, value used key. calculation follows steps: Calculate attention scores using query key shape (batch_size, Tq, Tv). Use scores calculate softmax distribution shape (batch_size, Tq, Tv). Use softmax distribution create linear combination value shape (batch_size, Tq, dim).","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_attention.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Dot-product attention layer, a.k.a. Luong-style attention. — layer_attention","text":"","code":"layer_attention(   object,   use_scale = FALSE,   score_mode = \"dot\",   dropout = 0,   seed = NULL,   ... )"},{"path":"https://keras3.posit.co/dev/reference/layer_attention.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Dot-product attention layer, a.k.a. Luong-style attention. — layer_attention","text":"object Object compose layer . tensor, array, sequential model. use_scale TRUE, create scalar variable scale attention scores. score_mode Function use compute attention scores, one {\"dot\", \"concat\"}. \"dot\" refers dot product query key vectors. \"concat\" refers hyperbolic tangent concatenation query key vectors. dropout Float 0 1. Fraction units drop attention scores. Defaults 0.0. seed integer use random seed case dropout. ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_attention.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Dot-product attention layer, a.k.a. Luong-style attention. — layer_attention","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_attention.html","id":"call-arguments","dir":"Reference","previous_headings":"","what":"Call Arguments","title":"Dot-product attention layer, a.k.a. Luong-style attention. — layer_attention","text":"inputs: List following tensors: query: Query tensor shape (batch_size, Tq, dim). value: Value tensor shape (batch_size, Tv, dim). key: Optional key tensor shape (batch_size, Tv, dim). given, use value key value, common case. mask: List following tensors: query_mask: boolean mask tensor shape (batch_size, Tq). given, output zero positions mask==FALSE. value_mask: boolean mask tensor shape (batch_size, Tv). given, apply mask values positions mask==FALSE contribute result. return_attention_scores: bool, TRUE, returns attention scores (masking softmax) additional output argument. training: Python boolean indicating whether layer behave training mode (adding dropout) inference mode (dropout). use_causal_mask: Boolean. Set TRUE decoder self-attention. Adds mask position attend positions j > . prevents flow information future towards past. Defaults FALSE.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_attention.html","id":"output","dir":"Reference","previous_headings":"","what":"Output","title":"Dot-product attention layer, a.k.a. Luong-style attention. — layer_attention","text":"Attention outputs shape (batch_size, Tq, dim). (Optional) Attention scores masking softmax shape (batch_size, Tq, Tv).","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_aug_mix.html","id":null,"dir":"Reference","previous_headings":"","what":"Performs the AugMix data augmentation technique. — layer_aug_mix","title":"Performs the AugMix data augmentation technique. — layer_aug_mix","text":"AugMix aims produce images variety preserving image semantics local statistics. augmentation process, augmentation applied across images batch num_chains different ways, chain consisting chain_depth augmentations.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_aug_mix.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Performs the AugMix data augmentation technique. — layer_aug_mix","text":"","code":"layer_aug_mix(   object,   value_range = c(0L, 255L),   num_chains = 3L,   chain_depth = 3L,   factor = 0.3,   alpha = 1,   all_ops = TRUE,   interpolation = \"bilinear\",   seed = NULL,   data_format = NULL,   ... )"},{"path":"https://keras3.posit.co/dev/reference/layer_aug_mix.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Performs the AugMix data augmentation technique. — layer_aug_mix","text":"object Object compose layer . tensor, array, sequential model. value_range range values incoming images . Represented two number tuple written (low, high). typically either (0, 1) (0, 255) depending preprocessing pipeline set . num_chains integer representing number different chains mixed, defaults 3. chain_depth integer representing maximum number transformations applied chain. actual number transformations chain sampled randomly range [0, chain_depth]. Defaults 3. factor strength augmentation normalized value 0 1. Default 0.3. alpha float value used probability coefficients Beta Dirichlet distributions, defaults 1.0. all_ops Use operations (including random_brightness, random_color_degeneration, random_contrast random_sharpness). Default True. interpolation interpolation method use resizing operations. Options include \"nearest\", \"bilinear\". Default \"bilinear\". seed Integer. Used create random seed. data_format string, either \"channels_last\" \"channels_first\". ordering dimensions inputs. \"channels_last\" corresponds inputs shape (batch, height, width, channels) \"channels_first\" corresponds inputs shape (batch, channels, height, width). defaults image_data_format value found Keras config file ~/.keras/keras.json. never set , \"channels_last\". ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_aug_mix.html","id":"references","dir":"Reference","previous_headings":"","what":"References","title":"Performs the AugMix data augmentation technique. — layer_aug_mix","text":"AugMix paper Official Code","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_auto_contrast.html","id":null,"dir":"Reference","previous_headings":"","what":"Performs the auto-contrast operation on an image. — layer_auto_contrast","title":"Performs the auto-contrast operation on an image. — layer_auto_contrast","text":"Auto contrast stretches values image across entire available value_range. makes differences pixels obvious. example image values [0, 1] range [0, 255], auto contrast change 1 values 255. layer active training inference time.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_auto_contrast.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Performs the auto-contrast operation on an image. — layer_auto_contrast","text":"","code":"layer_auto_contrast(object, value_range = tuple(0L, 255L), ...)"},{"path":"https://keras3.posit.co/dev/reference/layer_auto_contrast.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Performs the auto-contrast operation on an image. — layer_auto_contrast","text":"object Object compose layer . tensor, array, sequential model. value_range Range values incoming images . Represented two number tuple written (low, high). typically either (0, 1) (0, 255) depending preprocessing pipeline set . Defaults (0, 255). ... forward/backward compatability.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_average.html","id":null,"dir":"Reference","previous_headings":"","what":"Averages a list of inputs element-wise.. — layer_average","title":"Averages a list of inputs element-wise.. — layer_average","text":"takes input list tensors, shape, returns single tensor (also shape).","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_average.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Averages a list of inputs element-wise.. — layer_average","text":"","code":"layer_average(inputs, ...)"},{"path":"https://keras3.posit.co/dev/reference/layer_average.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Averages a list of inputs element-wise.. — layer_average","text":"inputs layers combine ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_average.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Averages a list of inputs element-wise.. — layer_average","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_average.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Averages a list of inputs element-wise.. — layer_average","text":"Usage Keras model:","code":"input_shape <- c(1, 2, 3) x1 <- op_ones(input_shape) x2 <- op_zeros(input_shape) layer_average(x1, x2) ## tf.Tensor( ## [[[0.5 0.5 0.5] ##   [0.5 0.5 0.5]]], shape=(1, 2, 3), dtype=float32) input1 <- layer_input(shape = c(16)) x1 <- input1 |> layer_dense(8, activation = 'relu')  input2 <- layer_input(shape = c(32)) x2 <- input2 |> layer_dense(8, activation = 'relu')  added <- layer_average(x1, x2) output <- added |> layer_dense(4)  model <- keras_model(inputs = c(input1, input2), outputs = output)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_average_pooling_1d.html","id":null,"dir":"Reference","previous_headings":"","what":"Average pooling for temporal data. — layer_average_pooling_1d","title":"Average pooling for temporal data. — layer_average_pooling_1d","text":"Downsamples input representation taking average value window defined pool_size. window shifted strides.  resulting output using \"valid\" padding option shape : output_shape = (input_shape - pool_size + 1) / strides) resulting output shape using \"\" padding option : output_shape = input_shape / strides","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_average_pooling_1d.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Average pooling for temporal data. — layer_average_pooling_1d","text":"","code":"layer_average_pooling_1d(   object,   pool_size,   strides = NULL,   padding = \"valid\",   data_format = NULL,   name = NULL,   ... )"},{"path":"https://keras3.posit.co/dev/reference/layer_average_pooling_1d.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Average pooling for temporal data. — layer_average_pooling_1d","text":"object Object compose layer . tensor, array, sequential model. pool_size int, size max pooling window. strides int NULL. Specifies much pooling window moves pooling step. NULL, default pool_size. padding string, either \"valid\" \"\" (case-insensitive). \"valid\" means padding. \"\" results padding evenly left/right /input output height/width dimension input. data_format string, either \"channels_last\" \"channels_first\". ordering dimensions inputs. \"channels_last\" corresponds inputs shape (batch, steps, features) \"channels_first\" corresponds inputs shape (batch, features, steps). defaults image_data_format value found Keras config file ~/.keras/keras.json. never set , \"channels_last\". name String, name object ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_average_pooling_1d.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Average pooling for temporal data. — layer_average_pooling_1d","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_average_pooling_1d.html","id":"input-shape","dir":"Reference","previous_headings":"","what":"Input Shape","title":"Average pooling for temporal data. — layer_average_pooling_1d","text":"data_format=\"channels_last\": 3D tensor shape (batch_size, steps, features). data_format=\"channels_first\": 3D tensor shape (batch_size, features, steps).","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_average_pooling_1d.html","id":"output-shape","dir":"Reference","previous_headings":"","what":"Output Shape","title":"Average pooling for temporal data. — layer_average_pooling_1d","text":"data_format=\"channels_last\": 3D tensor shape (batch_size, downsampled_steps, features). data_format=\"channels_first\": 3D tensor shape (batch_size, features, downsampled_steps).","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_average_pooling_1d.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Average pooling for temporal data. — layer_average_pooling_1d","text":"strides=1 padding=\"valid\":     strides=2 padding=\"valid\":     strides=1 padding=\"\":","code":"x <- op_array(c(1., 2., 3., 4., 5.)) |> op_reshape(c(1, 5, 1)) output <- x |>   layer_average_pooling_1d(pool_size = 2,                            strides = 1,                            padding = \"valid\") output ## tf.Tensor( ## [[[1.5] ##   [2.5] ##   [3.5] ##   [4.5]]], shape=(1, 4, 1), dtype=float32) x <- op_array(c(1., 2., 3., 4., 5.)) |> op_reshape(c(1, 5, 1)) output <- x |>   layer_average_pooling_1d(pool_size = 2,                            strides = 2,                            padding = \"valid\") output ## tf.Tensor( ## [[[1.5] ##   [3.5]]], shape=(1, 2, 1), dtype=float32) x <- op_array(c(1., 2., 3., 4., 5.)) |> op_reshape(c(1, 5, 1)) output <- x |>   layer_average_pooling_1d(pool_size = 2,                            strides = 1,                            padding = \"same\") output ## tf.Tensor( ## [[[1.5] ##   [2.5] ##   [3.5] ##   [4.5] ##   [5. ]]], shape=(1, 5, 1), dtype=float32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_average_pooling_2d.html","id":null,"dir":"Reference","previous_headings":"","what":"Average pooling operation for 2D spatial data. — layer_average_pooling_2d","title":"Average pooling operation for 2D spatial data. — layer_average_pooling_2d","text":"Downsamples input along spatial dimensions (height width) taking average value input window (size defined pool_size) channel input. window shifted strides along dimension. resulting output using \"valid\" padding option spatial shape (number rows columns) : output_shape = math.floor((input_shape - pool_size) / strides) + 1 (input_shape >= pool_size) resulting output shape using \"\" padding option : output_shape = input_shape","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_average_pooling_2d.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Average pooling operation for 2D spatial data. — layer_average_pooling_2d","text":"","code":"layer_average_pooling_2d(   object,   pool_size,   strides = NULL,   padding = \"valid\",   data_format = NULL,   name = NULL,   ... )"},{"path":"https://keras3.posit.co/dev/reference/layer_average_pooling_2d.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Average pooling operation for 2D spatial data. — layer_average_pooling_2d","text":"object Object compose layer . tensor, array, sequential model. pool_size int list 2 integers, factors downscale (dim1, dim2). one integer specified, window length used dimensions. strides int list 2 integers, NULL. Strides values. NULL, default pool_size. one int specified, stride size used dimensions. padding string, either \"valid\" \"\" (case-insensitive). \"valid\" means padding. \"\" results padding evenly left/right /input output height/width dimension input. data_format string, either \"channels_last\" \"channels_first\". ordering dimensions inputs. \"channels_last\" corresponds inputs shape (batch, height, width, channels) \"channels_first\" corresponds inputs shape (batch, channels, height, width). defaults image_data_format value found Keras config file ~/.keras/keras.json. never set , \"channels_last\". name String, name object ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_average_pooling_2d.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Average pooling operation for 2D spatial data. — layer_average_pooling_2d","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_average_pooling_2d.html","id":"input-shape","dir":"Reference","previous_headings":"","what":"Input Shape","title":"Average pooling operation for 2D spatial data. — layer_average_pooling_2d","text":"data_format=\"channels_last\": 4D tensor shape (batch_size, height, width, channels). data_format=\"channels_first\": 4D tensor shape (batch_size, channels, height, width).","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_average_pooling_2d.html","id":"output-shape","dir":"Reference","previous_headings":"","what":"Output Shape","title":"Average pooling operation for 2D spatial data. — layer_average_pooling_2d","text":"data_format=\"channels_last\": 4D tensor shape (batch_size, pooled_height, pooled_width, channels). data_format=\"channels_first\": 4D tensor shape (batch_size, channels, pooled_height, pooled_width).","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_average_pooling_2d.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Average pooling operation for 2D spatial data. — layer_average_pooling_2d","text":"strides=(1, 1) padding=\"valid\":     strides=(2, 2) padding=\"valid\":     stride=(1, 1) padding=\"\":","code":"x <- op_array(1:9, \"float32\") |> op_reshape(c(1, 3, 3, 1)) output <- x |>   layer_average_pooling_2d(pool_size = c(2, 2),                            strides = c(1, 1),                            padding = \"valid\") output ## tf.Tensor( ## [[[[3.] ##    [4.]] ## ##   [[6.] ##    [7.]]]], shape=(1, 2, 2, 1), dtype=float32) x <- op_array(1:12, \"float32\") |> op_reshape(c(1, 3, 4, 1)) output <- x |>   layer_average_pooling_2d(pool_size = c(2, 2),                            strides = c(2, 2),                            padding = \"valid\") output ## tf.Tensor( ## [[[[3.5] ##    [5.5]]]], shape=(1, 1, 2, 1), dtype=float32) x <- op_array(1:9, \"float32\") |> op_reshape(c(1, 3, 3, 1)) output <- x |>   layer_average_pooling_2d(pool_size = c(2, 2),                            strides = c(1, 1),                            padding = \"same\") output ## tf.Tensor( ## [[[[3. ] ##    [4. ] ##    [4.5]] ## ##   [[6. ] ##    [7. ] ##    [7.5]] ## ##   [[7.5] ##    [8.5] ##    [9. ]]]], shape=(1, 3, 3, 1), dtype=float32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_average_pooling_3d.html","id":null,"dir":"Reference","previous_headings":"","what":"Average pooling operation for 3D data (spatial or spatio-temporal). — layer_average_pooling_3d","title":"Average pooling operation for 3D data (spatial or spatio-temporal). — layer_average_pooling_3d","text":"Downsamples input along spatial dimensions (depth, height, width) taking average value input window (size defined pool_size) channel input. window shifted strides along dimension.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_average_pooling_3d.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Average pooling operation for 3D data (spatial or spatio-temporal). — layer_average_pooling_3d","text":"","code":"layer_average_pooling_3d(   object,   pool_size,   strides = NULL,   padding = \"valid\",   data_format = NULL,   name = NULL,   ... )"},{"path":"https://keras3.posit.co/dev/reference/layer_average_pooling_3d.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Average pooling operation for 3D data (spatial or spatio-temporal). — layer_average_pooling_3d","text":"object Object compose layer . tensor, array, sequential model. pool_size int list 3 integers, factors downscale (dim1, dim2, dim3). one integer specified, window length used dimensions. strides int list 3 integers, NULL. Strides values. NULL, default pool_size. one int specified, stride size used dimensions. padding string, either \"valid\" \"\" (case-insensitive). \"valid\" means padding. \"\" results padding evenly left/right /input output height/width dimension input. data_format string, either \"channels_last\" \"channels_first\". ordering dimensions inputs. \"channels_last\" corresponds inputs shape (batch, spatial_dim1, spatial_dim2, spatial_dim3, channels) \"channels_first\" corresponds inputs shape (batch, channels, spatial_dim1, spatial_dim2, spatial_dim3). defaults image_data_format value found Keras config file ~/.keras/keras.json. never set , \"channels_last\". name String, name object ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_average_pooling_3d.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Average pooling operation for 3D data (spatial or spatio-temporal). — layer_average_pooling_3d","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_average_pooling_3d.html","id":"input-shape","dir":"Reference","previous_headings":"","what":"Input Shape","title":"Average pooling operation for 3D data (spatial or spatio-temporal). — layer_average_pooling_3d","text":"data_format=\"channels_last\": 5D tensor shape: (batch_size, spatial_dim1, spatial_dim2, spatial_dim3, channels) data_format=\"channels_first\": 5D tensor shape: (batch_size, channels, spatial_dim1, spatial_dim2, spatial_dim3)","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_average_pooling_3d.html","id":"output-shape","dir":"Reference","previous_headings":"","what":"Output Shape","title":"Average pooling operation for 3D data (spatial or spatio-temporal). — layer_average_pooling_3d","text":"data_format=\"channels_last\": 5D tensor shape: (batch_size, pooled_dim1, pooled_dim2, pooled_dim3, channels) data_format=\"channels_first\": 5D tensor shape: (batch_size, channels, pooled_dim1, pooled_dim2, pooled_dim3)","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_average_pooling_3d.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Average pooling operation for 3D data (spatial or spatio-temporal). — layer_average_pooling_3d","text":"","code":"depth <- height <- width <- 30 channels <- 3  inputs <- layer_input(shape = c(depth, height, width, channels)) outputs <- inputs |> layer_average_pooling_3d(pool_size = 3) outputs # Shape: (batch_size, 10, 10, 10, 3) ## <KerasTensor shape=(None, 10, 10, 10, 3), dtype=float32, sparse=False, ragged=False, name=keras_tensor_1>"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_batch_normalization.html","id":null,"dir":"Reference","previous_headings":"","what":"Layer that normalizes its inputs. — layer_batch_normalization","title":"Layer that normalizes its inputs. — layer_batch_normalization","text":"Batch normalization applies transformation maintains mean output close 0 output standard deviation close 1. Importantly, batch normalization works differently training inference. training (.e. using fit() calling layer/model argument training = TRUE), layer normalizes output using mean standard deviation current batch inputs. say, channel normalized, layer returns gamma * (batch - mean(batch)) / sqrt(var(batch) + epsilon) + beta, : epsilon small constant (configurable part constructor arguments) gamma learned scaling factor (initialized 1), can disabled passing scale = FALSE constructor. beta learned offset factor (initialized 0), can disabled passing center = FALSE constructor. inference (.e. using evaluate() predict() calling layer/model argument training = FALSE (default), layer normalizes output using moving average mean standard deviation batches seen training. say, returns gamma * (batch - self$moving_mean) / sqrt(self$moving_var+epsilon) + beta. self$moving_mean self$moving_var non-trainable variables updated time layer called training mode, : moving_mean = moving_mean * momentum + mean(batch) * (1 - momentum) moving_var = moving_var * momentum + var(batch) * (1 - momentum) , layer normalize inputs inference trained data similar statistics inference data. setting layer$trainable <- FALSE BatchNormalization layer: meaning setting layer$trainable <- FALSE freeze layer, .e. internal state change training: trainable weights updated fit() train_on_batch(), state updates run. Usually, necessarily mean layer run inference mode (normally controlled training argument can passed calling layer). \"Frozen state\" \"inference mode\" two separate concepts. However, case BatchNormalization layer, setting trainable <- FALSE layer means layer subsequently run inference mode (meaning use moving mean moving variance normalize current batch, rather using mean variance current batch). Note : Setting trainable model containing layers recursively set trainable value inner layers. value trainable attribute changed calling compile() model, new value take effect model compile() called .","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_batch_normalization.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Layer that normalizes its inputs. — layer_batch_normalization","text":"","code":"layer_batch_normalization(   object,   axis = -1L,   momentum = 0.99,   epsilon = 0.001,   center = TRUE,   scale = TRUE,   beta_initializer = \"zeros\",   gamma_initializer = \"ones\",   moving_mean_initializer = \"zeros\",   moving_variance_initializer = \"ones\",   beta_regularizer = NULL,   gamma_regularizer = NULL,   beta_constraint = NULL,   gamma_constraint = NULL,   synchronized = FALSE,   ... )"},{"path":"https://keras3.posit.co/dev/reference/layer_batch_normalization.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Layer that normalizes its inputs. — layer_batch_normalization","text":"object Object compose layer . tensor, array, sequential model. axis Integer, axis normalized (typically features axis). instance, Conv2D layer data_format = \"channels_first\", use axis = 2. momentum Momentum moving average. epsilon Small float added variance avoid dividing zero. center TRUE, add offset beta normalized tensor. FALSE, beta ignored. scale TRUE, multiply gamma. FALSE, gamma used. next layer linear can disabled since scaling done next layer. beta_initializer Initializer beta weight. gamma_initializer Initializer gamma weight. moving_mean_initializer Initializer moving mean. moving_variance_initializer Initializer moving variance. beta_regularizer Optional regularizer beta weight. gamma_regularizer Optional regularizer gamma weight. beta_constraint Optional constraint beta weight. gamma_constraint Optional constraint gamma weight. synchronized applicable TensorFlow backend. TRUE, synchronizes global batch statistics (mean variance) layer across devices training step distributed training strategy. FALSE, replica uses local batch statistics. ... Base layer keyword arguments (e.g. name dtype).","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_batch_normalization.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Layer that normalizes its inputs. — layer_batch_normalization","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_batch_normalization.html","id":"call-arguments","dir":"Reference","previous_headings":"","what":"Call Arguments","title":"Layer that normalizes its inputs. — layer_batch_normalization","text":"inputs: Input tensor (rank). training: R boolean indicating whether layer behave training mode inference mode. training = TRUE: layer normalize inputs using mean variance current batch inputs. training = FALSE: layer normalize inputs using mean variance moving statistics, learned training. mask: Binary tensor shape broadcastable inputs tensor, TRUE values indicating positions mean variance computed. Masked elements current inputs taken account mean variance computation training. prior unmasked element values taken account momentum expires.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_batch_normalization.html","id":"reference","dir":"Reference","previous_headings":"","what":"Reference","title":"Layer that normalizes its inputs. — layer_batch_normalization","text":"Ioffe Szegedy, 2015.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_bidirectional.html","id":null,"dir":"Reference","previous_headings":"","what":"Bidirectional wrapper for RNNs. — layer_bidirectional","title":"Bidirectional wrapper for RNNs. — layer_bidirectional","text":"Bidirectional wrapper RNNs.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_bidirectional.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Bidirectional wrapper for RNNs. — layer_bidirectional","text":"","code":"layer_bidirectional(   object,   layer,   merge_mode = \"concat\",   weights = NULL,   backward_layer = NULL,   ... )"},{"path":"https://keras3.posit.co/dev/reference/layer_bidirectional.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Bidirectional wrapper for RNNs. — layer_bidirectional","text":"object Object compose layer . tensor, array, sequential model. layer RNN instance, layer_lstm() layer_gru(). also Layer() instance meets following criteria: sequence-processing layer (accepts 3D+ inputs). go_backwards, return_sequences return_state attribute (semantics RNN class). input_spec attribute. Implement serialization via get_config() from_config(). Note recommended way create new RNN layers write custom RNN cell use layer_rnn(), instead subclassing Layer() directly. return_sequences TRUE, output masked timestep zero regardless layer's original zero_output_for_mask value. merge_mode Mode outputs forward backward RNNs combined. One {\"sum\", \"mul\", \"concat\", \"ave\", NULL}. NULL, outputs combined, returned list. Defaults \"concat\". weights see description backward_layer Optional RNN, Layer() instance used handle backwards input processing. backward_layer provided, layer instance passed layer argument used generate backward layer automatically. Note provided backward_layer layer properties matching layer argument, particular values stateful, return_states, return_sequences, etc. addition, backward_layer layer different go_backwards argument values. ValueError raised requirements met. ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_bidirectional.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Bidirectional wrapper for RNNs. — layer_bidirectional","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_bidirectional.html","id":"call-arguments","dir":"Reference","previous_headings":"","what":"Call Arguments","title":"Bidirectional wrapper for RNNs. — layer_bidirectional","text":"call arguments layer wrapped RNN layer. Beware passing initial_state argument call layer, first half list elements initial_state list passed forward RNN call last half list elements passed backward RNN call.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_bidirectional.html","id":"note","dir":"Reference","previous_headings":"","what":"Note","title":"Bidirectional wrapper for RNNs. — layer_bidirectional","text":"instantiating Bidirectional layer existing RNN layer instance reuse weights state RNN layer instance – Bidirectional layer freshly initialized weights.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_bidirectional.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Bidirectional wrapper for RNNs. — layer_bidirectional","text":"","code":"model <- keras_model_sequential(input_shape = c(5, 10)) %>%   layer_bidirectional(layer_lstm(units = 10, return_sequences = TRUE)) %>%   layer_bidirectional(layer_lstm(units = 10)) %>%   layer_dense(5, activation = \"softmax\")  model %>% compile(loss = \"categorical_crossentropy\",                   optimizer = \"rmsprop\")  # With custom backward layer forward_layer <- layer_lstm(units = 10, return_sequences = TRUE) backward_layer <- layer_lstm(units = 10, activation = \"relu\",                              return_sequences = TRUE, go_backwards = TRUE)  model <- keras_model_sequential(input_shape = c(5, 10)) %>%   bidirectional(forward_layer, backward_layer = backward_layer) %>%   layer_dense(5, activation = \"softmax\")  model %>% compile(loss = \"categorical_crossentropy\",                   optimizer = \"rmsprop\")"},{"path":"https://keras3.posit.co/dev/reference/layer_bidirectional.html","id":"states","dir":"Reference","previous_headings":"","what":"States","title":"Bidirectional wrapper for RNNs. — layer_bidirectional","text":"Bidirectional layer instance property states, can access layer$states. can also reset states using reset_state()","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_category_encoding.html","id":null,"dir":"Reference","previous_headings":"","what":"A preprocessing layer which encodes integer features. — layer_category_encoding","title":"A preprocessing layer which encodes integer features. — layer_category_encoding","text":"layer provides options condensing data categorical encoding total number tokens known advance. accepts integer values inputs, outputs dense sparse representation inputs. integer inputs total number tokens known, use layer_integer_lookup() instead. Note: layer safe use inside tf.data pipeline (independently backend using).","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_category_encoding.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"A preprocessing layer which encodes integer features. — layer_category_encoding","text":"","code":"layer_category_encoding(   object,   num_tokens = NULL,   output_mode = \"multi_hot\",   sparse = FALSE,   ... )"},{"path":"https://keras3.posit.co/dev/reference/layer_category_encoding.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"A preprocessing layer which encodes integer features. — layer_category_encoding","text":"object Object compose layer . tensor, array, sequential model. num_tokens total number tokens layer support. inputs layer must integers range 0 <= value < num_tokens, error thrown. output_mode Specification output layer. Values can \"one_hot\", \"multi_hot\" \"count\", configuring layer follows: - \"one_hot\": Encodes individual element input array num_tokens size, containing 1 element index. last dimension size 1, encode dimension. last dimension size 1, append new dimension encoded output. - \"multi_hot\": Encodes sample input single array num_tokens size, containing 1 vocabulary term present sample. Treats last dimension sample dimension, input shape (..., sample_length), output shape (..., num_tokens). - \"count\": Like \"multi_hot\", int array contains count number times token index appeared sample. output modes, currently output rank 2 supported. Defaults \"multi_hot\". sparse Whether return sparse tensor; backends support sparse tensors. ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_category_encoding.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"A preprocessing layer which encodes integer features. — layer_category_encoding","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_category_encoding.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"A preprocessing layer which encodes integer features. — layer_category_encoding","text":"One-hot encoding data     Multi-hot encoding data     Using weighted inputs \"count\" mode","code":"layer <- layer_category_encoding(num_tokens = 4, output_mode = \"one_hot\") x <- op_array(c(3, 2, 0, 1), \"int32\") layer(x) ## tf.Tensor( ## [[0. 0. 0. 1.] ##  [0. 0. 1. 0.] ##  [1. 0. 0. 0.] ##  [0. 1. 0. 0.]], shape=(4, 4), dtype=float32) layer <- layer_category_encoding(num_tokens = 4, output_mode = \"multi_hot\") x <- op_array(rbind(c(0, 1),                    c(0, 0),                    c(1, 2),                    c(3, 1)), \"int32\") layer(x) ## tf.Tensor( ## [[1. 1. 0. 0.] ##  [1. 0. 0. 0.] ##  [0. 1. 1. 0.] ##  [0. 1. 0. 1.]], shape=(4, 4), dtype=float32) layer <- layer_category_encoding(num_tokens = 4, output_mode = \"count\") count_weights <- op_array(rbind(c(.1, .2),                                c(.1, .1),                                c(.2, .3),                                c(.4, .2))) x <- op_array(rbind(c(0, 1),                    c(0, 0),                    c(1, 2),                    c(3, 1)), \"int32\") layer(x, count_weights = count_weights) #   array([[01, 02, 0. , 0. ], #          [02, 0. , 0. , 0. ], #          [0. , 02, 03, 0. ], #          [0. , 02, 0. , 04]]>"},{"path":"https://keras3.posit.co/dev/reference/layer_category_encoding.html","id":"call-arguments","dir":"Reference","previous_headings":"","what":"Call Arguments","title":"A preprocessing layer which encodes integer features. — layer_category_encoding","text":"inputs: 1D 2D tensor integer inputs. count_weights: tensor shape inputs indicating weight sample value summing count mode. used \"multi_hot\" \"one_hot\" modes.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_center_crop.html","id":null,"dir":"Reference","previous_headings":"","what":"A preprocessing layer which crops images. — layer_center_crop","title":"A preprocessing layer which crops images. — layer_center_crop","text":"layers crops central portion images target size. image smaller target size, resized cropped return largest possible window image matches target aspect ratio. Input pixel values can range (e.g. [0., 1.) [0, 255]).","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_center_crop.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"A preprocessing layer which crops images. — layer_center_crop","text":"","code":"layer_center_crop(object, height, width, data_format = NULL, ...)"},{"path":"https://keras3.posit.co/dev/reference/layer_center_crop.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"A preprocessing layer which crops images. — layer_center_crop","text":"object Object compose layer . tensor, array, sequential model. height Integer, height output shape. width Integer, width output shape. data_format string, either \"channels_last\" \"channels_first\". ordering dimensions inputs. \"channels_last\" corresponds inputs shape (batch, height, width, channels) \"channels_first\" corresponds inputs shape (batch, channels, height, width). defaults image_data_format value found Keras config file ~/.keras/keras.json. never set , \"channels_last\". ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_center_crop.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"A preprocessing layer which crops images. — layer_center_crop","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_center_crop.html","id":"input-shape","dir":"Reference","previous_headings":"","what":"Input Shape","title":"A preprocessing layer which crops images. — layer_center_crop","text":"3D (unbatched) 4D (batched) tensor shape: (..., height, width, channels), \"channels_last\" format, (..., channels, height, width), \"channels_first\" format.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_center_crop.html","id":"output-shape","dir":"Reference","previous_headings":"","what":"Output Shape","title":"A preprocessing layer which crops images. — layer_center_crop","text":"3D (unbatched) 4D (batched) tensor shape: (..., target_height, target_width, channels), (..., channels, target_height, target_width), \"channels_first\" format. input height/width even target height/width odd (inversely), input image left-padded 1 pixel. Note: layer safe use inside tf.data pipeline (independently backend using).","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_concatenate.html","id":null,"dir":"Reference","previous_headings":"","what":"Concatenates a list of inputs. — layer_concatenate","title":"Concatenates a list of inputs. — layer_concatenate","text":"takes input list tensors, shape except concatenation axis, returns single tensor concatenation inputs.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_concatenate.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Concatenates a list of inputs. — layer_concatenate","text":"","code":"layer_concatenate(inputs, ..., axis = -1L)"},{"path":"https://keras3.posit.co/dev/reference/layer_concatenate.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Concatenates a list of inputs. — layer_concatenate","text":"inputs layers combine ... Standard layer keyword arguments. axis Axis along concatenate.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_concatenate.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Concatenates a list of inputs. — layer_concatenate","text":"tensor, concatenation inputs alongside axis axis.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_concatenate.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Concatenates a list of inputs. — layer_concatenate","text":"Usage Keras model:","code":"x <- op_arange(20) |> op_reshape(c(2, 2, 5)) y <- op_arange(21, 40) |> op_reshape(c(2, 2, 5)) layer_concatenate(x, y, axis = 2) ## tf.Tensor( ## [[[ 1.  2.  3.  4.  5.] ##   [ 6.  7.  8.  9. 10.] ##   [21. 22. 23. 24. 25.] ##   [26. 27. 28. 29. 30.]] ## ##  [[11. 12. 13. 14. 15.] ##   [16. 17. 18. 19. 20.] ##   [31. 32. 33. 34. 35.] ##   [36. 37. 38. 39. 40.]]], shape=(2, 4, 5), dtype=float32) x1 <- op_arange(10)     |> op_reshape(c(5, 2)) |> layer_dense(8) x2 <- op_arange(11, 20) |> op_reshape(c(5, 2)) |> layer_dense(8) y <- layer_concatenate(x1, x2)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_conv_1d.html","id":null,"dir":"Reference","previous_headings":"","what":"1D convolution layer (e.g. temporal convolution). — layer_conv_1d","title":"1D convolution layer (e.g. temporal convolution). — layer_conv_1d","text":"layer creates convolution kernel convolved layer input single spatial (temporal) dimension produce tensor outputs. use_bias TRUE, bias vector created added outputs. Finally, activation NULL, applied outputs well.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_conv_1d.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"1D convolution layer (e.g. temporal convolution). — layer_conv_1d","text":"","code":"layer_conv_1d(   object,   filters,   kernel_size,   strides = 1L,   padding = \"valid\",   data_format = NULL,   dilation_rate = 1L,   groups = 1L,   activation = NULL,   use_bias = TRUE,   kernel_initializer = \"glorot_uniform\",   bias_initializer = \"zeros\",   kernel_regularizer = NULL,   bias_regularizer = NULL,   activity_regularizer = NULL,   kernel_constraint = NULL,   bias_constraint = NULL,   ... )"},{"path":"https://keras3.posit.co/dev/reference/layer_conv_1d.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"1D convolution layer (e.g. temporal convolution). — layer_conv_1d","text":"object Object compose layer . tensor, array, sequential model. filters int, dimension output space (number filters convolution). kernel_size int list 1 integer, specifying size convolution window. strides int list 1 integer, specifying stride length convolution. strides > 1 incompatible dilation_rate > 1. padding string, \"valid\", \"\" \"causal\"(case-insensitive). \"valid\" means padding. \"\" results padding evenly left/right /input. padding=\"\" strides=1, output size input. \"causal\" results causal (dilated) convolutions, e.g. output[t] depend ontail(input, t+1). Useful modeling temporal data model violate temporal order. See WaveNet: Generative Model Raw Audio, section2.1. data_format string, either \"channels_last\" \"channels_first\". ordering dimensions inputs. \"channels_last\" corresponds inputs shape (batch, steps, features) \"channels_first\" corresponds inputs shape (batch, features, steps). defaults image_data_format value found Keras config file ~/.keras/keras.json. never set , \"channels_last\". dilation_rate int list 1 integers, specifying dilation rate use dilated convolution. groups positive int specifying number groups input split along channel axis. group convolved separately filters // groups filters. output concatenation groups results along channel axis. Input channels filters must divisible groups. activation Activation function. NULL, activation applied. use_bias bool, TRUE, bias added output. kernel_initializer Initializer convolution kernel. NULL, default initializer (\"glorot_uniform\") used. bias_initializer Initializer bias vector. NULL, default initializer (\"zeros\") used. kernel_regularizer Optional regularizer convolution kernel. bias_regularizer Optional regularizer bias vector. activity_regularizer Optional regularizer function output. kernel_constraint Optional projection function applied kernel updated Optimizer (e.g. used implement norm constraints value constraints layer weights). function must take input unprojected variable must return projected variable (must shape). Constraints safe use asynchronous distributed training. bias_constraint Optional projection function applied bias updated Optimizer. ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_conv_1d.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"1D convolution layer (e.g. temporal convolution). — layer_conv_1d","text":"3D tensor representing activation(conv1d(inputs, kernel) + bias).","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_conv_1d.html","id":"input-shape","dir":"Reference","previous_headings":"","what":"Input Shape","title":"1D convolution layer (e.g. temporal convolution). — layer_conv_1d","text":"data_format=\"channels_last\": 3D tensor shape: (batch_shape, steps, channels) data_format=\"channels_first\": 3D tensor shape: (batch_shape, channels, steps)","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_conv_1d.html","id":"output-shape","dir":"Reference","previous_headings":"","what":"Output Shape","title":"1D convolution layer (e.g. temporal convolution). — layer_conv_1d","text":"data_format=\"channels_last\": 3D tensor shape: (batch_shape, new_steps, filters) data_format=\"channels_first\": 3D tensor shape: (batch_shape, filters, new_steps)","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_conv_1d.html","id":"raises","dir":"Reference","previous_headings":"","what":"Raises","title":"1D convolution layer (e.g. temporal convolution). — layer_conv_1d","text":"ValueError: strides > 1 dilation_rate > 1.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_conv_1d.html","id":"example","dir":"Reference","previous_headings":"","what":"Example","title":"1D convolution layer (e.g. temporal convolution). — layer_conv_1d","text":"","code":"# The inputs are 128-length vectors with 10 timesteps, and the # batch size is 4. x <- random_uniform(c(4, 10, 128)) y <- x |> layer_conv_1d(32, 3, activation='relu') shape(y) ## shape(4, 8, 32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_conv_1d_transpose.html","id":null,"dir":"Reference","previous_headings":"","what":"1D transposed convolution layer. — layer_conv_1d_transpose","title":"1D transposed convolution layer. — layer_conv_1d_transpose","text":"need transposed convolutions generally arise desire use transformation going opposite direction normal convolution, .e., something shape output convolution something shape input maintaining connectivity pattern compatible said convolution.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_conv_1d_transpose.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"1D transposed convolution layer. — layer_conv_1d_transpose","text":"","code":"layer_conv_1d_transpose(   object,   filters,   kernel_size,   strides = 1L,   padding = \"valid\",   output_padding = NULL,   data_format = NULL,   dilation_rate = 1L,   activation = NULL,   use_bias = TRUE,   kernel_initializer = \"glorot_uniform\",   bias_initializer = \"zeros\",   kernel_regularizer = NULL,   bias_regularizer = NULL,   activity_regularizer = NULL,   kernel_constraint = NULL,   bias_constraint = NULL,   ... )"},{"path":"https://keras3.posit.co/dev/reference/layer_conv_1d_transpose.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"1D transposed convolution layer. — layer_conv_1d_transpose","text":"object Object compose layer . tensor, array, sequential model. filters int, dimension output space (number filters transpose convolution). kernel_size int list 1 integer, specifying size transposed convolution window. strides int list 1 integer, specifying stride length transposed convolution. strides > 1 incompatible dilation_rate > 1. padding string, either \"valid\" \"\" (case-insensitive). \"valid\" means padding. \"\" results padding evenly left/right /input output height/width dimension input. output_padding Scalar integer. Amount padding add output length. Must less stride. NULL (default) output size inferred. data_format string, either \"channels_last\" \"channels_first\". ordering dimensions inputs. \"channels_last\" corresponds inputs shape (batch, steps, features) \"channels_first\" corresponds inputs shape (batch, features, steps). defaults image_data_format value found Keras config file ~/.keras/keras.json. never set , \"channels_last\". dilation_rate Scalar integer. Specifies dilation rate. Values 1 currently require strides = 1 rates greater 1 supported. activation Activation function. NULL, activation applied. use_bias bool, TRUE, bias added output. kernel_initializer Initializer convolution kernel. NULL, default initializer (\"glorot_uniform\") used. bias_initializer Initializer bias vector. NULL, default initializer (\"zeros\") used. kernel_regularizer Optional regularizer convolution kernel. bias_regularizer Optional regularizer bias vector. activity_regularizer Optional regularizer function output. kernel_constraint Optional projection function applied kernel updated Optimizer (e.g. used implement norm constraints value constraints layer weights). function must take input unprojected variable must return projected variable (must shape). Constraints safe use asynchronous distributed training. bias_constraint Optional projection function applied bias updated Optimizer. ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_conv_1d_transpose.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"1D transposed convolution layer. — layer_conv_1d_transpose","text":"3D tensor representing activation(conv1d_transpose(inputs, kernel) + bias).","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_conv_1d_transpose.html","id":"input-shape","dir":"Reference","previous_headings":"","what":"Input Shape","title":"1D transposed convolution layer. — layer_conv_1d_transpose","text":"data_format=\"channels_last\": 3D tensor shape: (batch_shape, steps, channels) data_format=\"channels_first\": 3D tensor shape: (batch_shape, channels, steps)","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_conv_1d_transpose.html","id":"output-shape","dir":"Reference","previous_headings":"","what":"Output Shape","title":"1D transposed convolution layer. — layer_conv_1d_transpose","text":"data_format=\"channels_last\": 3D tensor shape: (batch_shape, new_steps, filters) data_format=\"channels_first\": 3D tensor shape: (batch_shape, filters, new_steps)","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_conv_1d_transpose.html","id":"raises","dir":"Reference","previous_headings":"","what":"Raises","title":"1D transposed convolution layer. — layer_conv_1d_transpose","text":"ValueError: strides > 1 dilation_rate > 1.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_conv_1d_transpose.html","id":"references","dir":"Reference","previous_headings":"","what":"References","title":"1D transposed convolution layer. — layer_conv_1d_transpose","text":"guide convolution arithmetic deep learning Deconvolutional Networks","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_conv_1d_transpose.html","id":"example","dir":"Reference","previous_headings":"","what":"Example","title":"1D transposed convolution layer. — layer_conv_1d_transpose","text":"","code":"x <- random_uniform(c(4, 10, 128)) y <- x |> layer_conv_1d_transpose(32, 3, 2, activation='relu') shape(y) ## shape(4, 21, 32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_conv_2d.html","id":null,"dir":"Reference","previous_headings":"","what":"2D convolution layer. — layer_conv_2d","title":"2D convolution layer. — layer_conv_2d","text":"layer creates convolution kernel convolved layer input 2D spatial (temporal) dimension (height width) produce tensor outputs. use_bias TRUE, bias vector created added outputs. Finally, activation NULL, applied outputs well. Note numerical precision: general Keras operation execution results identical across backends 1e-7 precision float32, Conv2D operations may show larger variations. Due large number element-wise multiplications additions convolution operations, especially large inputs kernel sizes, accumulated floating-point differences can exceed 1e-7 threshold. variations particularly noticeable using different backends (e.g., TensorFlow vs JAX) different hardware.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_conv_2d.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"2D convolution layer. — layer_conv_2d","text":"","code":"layer_conv_2d(   object,   filters,   kernel_size,   strides = list(1L, 1L),   padding = \"valid\",   data_format = NULL,   dilation_rate = list(1L, 1L),   groups = 1L,   activation = NULL,   use_bias = TRUE,   kernel_initializer = \"glorot_uniform\",   bias_initializer = \"zeros\",   kernel_regularizer = NULL,   bias_regularizer = NULL,   activity_regularizer = NULL,   kernel_constraint = NULL,   bias_constraint = NULL,   ... )"},{"path":"https://keras3.posit.co/dev/reference/layer_conv_2d.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"2D convolution layer. — layer_conv_2d","text":"object Object compose layer . tensor, array, sequential model. filters int, dimension output space (number filters convolution). kernel_size int list 2 integer, specifying size convolution window. strides int list 2 integer, specifying stride length convolution. strides > 1 incompatible dilation_rate > 1. padding string, either \"valid\" \"\" (case-insensitive). \"valid\" means padding. \"\" results padding evenly left/right /input. padding=\"\" strides=1, output size input. data_format string, either \"channels_last\" \"channels_first\". ordering dimensions inputs. \"channels_last\" corresponds inputs shape (batch_size, height, width, channels) \"channels_first\" corresponds inputs shape (batch_size, channels, height, width). defaults image_data_format value found Keras config file ~/.keras/keras.json. never set , \"channels_last\". dilation_rate int list 2 integers, specifying dilation rate use dilated convolution. groups positive int specifying number groups input split along channel axis. group convolved separately filters // groups filters. output concatenation groups results along channel axis. Input channels filters must divisible groups. activation Activation function. NULL, activation applied. use_bias bool, TRUE, bias added output. kernel_initializer Initializer convolution kernel. NULL, default initializer (\"glorot_uniform\") used. bias_initializer Initializer bias vector. NULL, default initializer (\"zeros\") used. kernel_regularizer Optional regularizer convolution kernel. bias_regularizer Optional regularizer bias vector. activity_regularizer Optional regularizer function output. kernel_constraint Optional projection function applied kernel updated Optimizer (e.g. used implement norm constraints value constraints layer weights). function must take input unprojected variable must return projected variable (must shape). Constraints safe use asynchronous distributed training. bias_constraint Optional projection function applied bias updated Optimizer. ... forward/backward compatability. output_padding Scalar integer vector two integers. Amount padding add output depth, height, width. element must smaller corresponding stride. NULL (default) output size inferred.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_conv_2d.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"2D convolution layer. — layer_conv_2d","text":"4D tensor representing activation(conv2d(inputs, kernel) + bias).","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_conv_2d.html","id":"input-shape","dir":"Reference","previous_headings":"","what":"Input Shape","title":"2D convolution layer. — layer_conv_2d","text":"data_format=\"channels_last\": 4D tensor shape: (batch_size, height, width, channels) data_format=\"channels_first\": 4D tensor shape: (batch_size, channels, height, width)","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_conv_2d.html","id":"output-shape","dir":"Reference","previous_headings":"","what":"Output Shape","title":"2D convolution layer. — layer_conv_2d","text":"data_format=\"channels_last\": 4D tensor shape: (batch_size, new_height, new_width, filters) data_format=\"channels_first\": 4D tensor shape: (batch_size, filters, new_height, new_width)","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_conv_2d.html","id":"raises","dir":"Reference","previous_headings":"","what":"Raises","title":"2D convolution layer. — layer_conv_2d","text":"ValueError: strides > 1 dilation_rate > 1.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_conv_2d.html","id":"example","dir":"Reference","previous_headings":"","what":"Example","title":"2D convolution layer. — layer_conv_2d","text":"","code":"x <- random_uniform(c(4, 10, 10, 128)) y <- x |> layer_conv_2d(32, 3, activation='relu') shape(y) ## shape(4, 8, 8, 32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_conv_2d_transpose.html","id":null,"dir":"Reference","previous_headings":"","what":"2D transposed convolution layer. — layer_conv_2d_transpose","title":"2D transposed convolution layer. — layer_conv_2d_transpose","text":"need transposed convolutions generally arise desire use transformation going opposite direction normal convolution, .e., something shape output convolution something shape input maintaining connectivity pattern compatible said convolution.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_conv_2d_transpose.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"2D transposed convolution layer. — layer_conv_2d_transpose","text":"","code":"layer_conv_2d_transpose(   object,   filters,   kernel_size,   strides = list(1L, 1L),   padding = \"valid\",   output_padding = NULL,   data_format = NULL,   dilation_rate = list(1L, 1L),   activation = NULL,   use_bias = TRUE,   kernel_initializer = \"glorot_uniform\",   bias_initializer = \"zeros\",   kernel_regularizer = NULL,   bias_regularizer = NULL,   activity_regularizer = NULL,   kernel_constraint = NULL,   bias_constraint = NULL,   ... )"},{"path":"https://keras3.posit.co/dev/reference/layer_conv_2d_transpose.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"2D transposed convolution layer. — layer_conv_2d_transpose","text":"object Object compose layer . tensor, array, sequential model. filters int, dimension output space (number filters transposed convolution). kernel_size int list 1 integer, specifying size transposed convolution window. strides int list 1 integer, specifying stride length transposed convolution. strides > 1 incompatible dilation_rate > 1. padding string, either \"valid\" \"\" (case-insensitive). \"valid\" means padding. \"\" results padding evenly left/right /input. padding=\"\" strides=1, output size input. data_format string, either \"channels_last\" \"channels_first\". ordering dimensions inputs. \"channels_last\" corresponds inputs shape (batch_size, height, width, channels) \"channels_first\" corresponds inputs shape (batch_size, channels, height, width). defaults image_data_format value found Keras config file ~/.keras/keras.json. never set , \"channels_last\". dilation_rate Scalar integer vector 2 integers specifying dilation rate. Values 1 require strides = 1; different rates per dimension supported. activation Activation function. NULL, activation applied. use_bias bool, TRUE, bias added output. kernel_initializer Initializer convolution kernel. NULL, default initializer (\"glorot_uniform\") used. bias_initializer Initializer bias vector. NULL, default initializer (\"zeros\") used. kernel_regularizer Optional regularizer convolution kernel. bias_regularizer Optional regularizer bias vector. activity_regularizer Optional regularizer function output. kernel_constraint Optional projection function applied kernel updated Optimizer (e.g. used implement norm constraints value constraints layer weights). function must take input unprojected variable must return projected variable (must shape). Constraints safe use asynchronous distributed training. bias_constraint Optional projection function applied bias updated Optimizer. ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_conv_2d_transpose.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"2D transposed convolution layer. — layer_conv_2d_transpose","text":"4D tensor representing activation(conv2d_transpose(inputs, kernel) + bias).","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_conv_2d_transpose.html","id":"input-shape","dir":"Reference","previous_headings":"","what":"Input Shape","title":"2D transposed convolution layer. — layer_conv_2d_transpose","text":"data_format=\"channels_last\": 4D tensor shape: (batch_size, height, width, channels) data_format=\"channels_first\": 4D tensor shape: (batch_size, channels, height, width)","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_conv_2d_transpose.html","id":"output-shape","dir":"Reference","previous_headings":"","what":"Output Shape","title":"2D transposed convolution layer. — layer_conv_2d_transpose","text":"data_format=\"channels_last\": 4D tensor shape: (batch_size, new_height, new_width, filters) data_format=\"channels_first\": 4D tensor shape: (batch_size, filters, new_height, new_width)","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_conv_2d_transpose.html","id":"raises","dir":"Reference","previous_headings":"","what":"Raises","title":"2D transposed convolution layer. — layer_conv_2d_transpose","text":"ValueError: strides > 1 dilation_rate > 1.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_conv_2d_transpose.html","id":"references","dir":"Reference","previous_headings":"","what":"References","title":"2D transposed convolution layer. — layer_conv_2d_transpose","text":"guide convolution arithmetic deep learning Deconvolutional Networks","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_conv_2d_transpose.html","id":"example","dir":"Reference","previous_headings":"","what":"Example","title":"2D transposed convolution layer. — layer_conv_2d_transpose","text":"","code":"x <- random_uniform(c(4, 10, 8, 128)) y <- x |> layer_conv_2d_transpose(32, 2, 2, activation='relu') shape(y) ## shape(4, 20, 16, 32) # (4, 20, 16, 32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_conv_3d.html","id":null,"dir":"Reference","previous_headings":"","what":"3D convolution layer. — layer_conv_3d","title":"3D convolution layer. — layer_conv_3d","text":"layer creates convolution kernel convolved layer input 3D spatial (temporal) dimension (width,height depth) produce tensor outputs. use_bias TRUE, bias vector created added outputs. Finally, activation NULL, applied outputs well.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_conv_3d.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"3D convolution layer. — layer_conv_3d","text":"","code":"layer_conv_3d(   object,   filters,   kernel_size,   strides = list(1L, 1L, 1L),   padding = \"valid\",   data_format = NULL,   dilation_rate = list(1L, 1L, 1L),   groups = 1L,   activation = NULL,   use_bias = TRUE,   kernel_initializer = \"glorot_uniform\",   bias_initializer = \"zeros\",   kernel_regularizer = NULL,   bias_regularizer = NULL,   activity_regularizer = NULL,   kernel_constraint = NULL,   bias_constraint = NULL,   ... )"},{"path":"https://keras3.posit.co/dev/reference/layer_conv_3d.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"3D convolution layer. — layer_conv_3d","text":"object Object compose layer . tensor, array, sequential model. filters int, dimension output space (number filters convolution). kernel_size int list 3 integer, specifying size convolution window. strides int list 3 integer, specifying stride length convolution. strides > 1 incompatible dilation_rate > 1. padding string, either \"valid\" \"\" (case-insensitive). \"valid\" means padding. \"\" results padding evenly left/right /input. padding=\"\" strides=1, output size input. data_format string, either \"channels_last\" \"channels_first\". ordering dimensions inputs. \"channels_last\" corresponds inputs shape (batch_size, spatial_dim1, spatial_dim2, spatial_dim3, channels) \"channels_first\" corresponds inputs shape (batch_size, channels, spatial_dim1, spatial_dim2, spatial_dim3). defaults image_data_format value found Keras config file ~/.keras/keras.json. never set , \"channels_last\". dilation_rate int vector 3 ints, specifying dilation rate use dilated convolution. groups positive int specifying number groups input split along channel axis. group convolved separately filters %/% groups filters. output concatenation groups results along channel axis. Input channels filters must divisible groups. activation Activation function. NULL, activation applied. use_bias bool, TRUE, bias added output. kernel_initializer Initializer convolution kernel. NULL, default initializer (\"glorot_uniform\") used. bias_initializer Initializer bias vector. NULL, default initializer (\"zeros\") used. kernel_regularizer Optional regularizer convolution kernel. bias_regularizer Optional regularizer bias vector. activity_regularizer Optional regularizer function output. kernel_constraint Optional projection function applied kernel updated Optimizer (e.g. used implement norm constraints value constraints layer weights). function must take input unprojected variable must return projected variable (must shape). Constraints safe use asynchronous distributed training. bias_constraint Optional projection function applied bias updated Optimizer. ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_conv_3d.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"3D convolution layer. — layer_conv_3d","text":"5D tensor representing activation(conv3d(inputs, kernel) + bias).","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_conv_3d.html","id":"input-shape","dir":"Reference","previous_headings":"","what":"Input Shape","title":"3D convolution layer. — layer_conv_3d","text":"data_format=\"channels_last\": 5D tensor shape: (batch_size, spatial_dim1, spatial_dim2, spatial_dim3, channels) data_format=\"channels_first\": 5D tensor shape: (batch_size, channels, spatial_dim1, spatial_dim2, spatial_dim3)","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_conv_3d.html","id":"output-shape","dir":"Reference","previous_headings":"","what":"Output Shape","title":"3D convolution layer. — layer_conv_3d","text":"data_format=\"channels_last\": 5D tensor shape: (batch_size, new_spatial_dim1, new_spatial_dim2, new_spatial_dim3, filters) data_format=\"channels_first\": 5D tensor shape: (batch_size, filters, new_spatial_dim1, new_spatial_dim2, new_spatial_dim3)","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_conv_3d.html","id":"raises","dir":"Reference","previous_headings":"","what":"Raises","title":"3D convolution layer. — layer_conv_3d","text":"ValueError: strides > 1 dilation_rate > 1.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_conv_3d.html","id":"example","dir":"Reference","previous_headings":"","what":"Example","title":"3D convolution layer. — layer_conv_3d","text":"","code":"x <- random_uniform(c(4, 10, 10, 10, 128)) y <- x |> layer_conv_3d(32, 3, activation = 'relu') shape(y) ## shape(4, 8, 8, 8, 32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_conv_3d_transpose.html","id":null,"dir":"Reference","previous_headings":"","what":"3D transposed convolution layer. — layer_conv_3d_transpose","title":"3D transposed convolution layer. — layer_conv_3d_transpose","text":"need transposed convolutions generally arise desire use transformation going opposite direction normal convolution, .e., something shape output convolution something shape input maintaining connectivity pattern compatible said convolution.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_conv_3d_transpose.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"3D transposed convolution layer. — layer_conv_3d_transpose","text":"","code":"layer_conv_3d_transpose(   object,   filters,   kernel_size,   strides = list(1L, 1L, 1L),   padding = \"valid\",   output_padding = NULL,   data_format = NULL,   dilation_rate = list(1L, 1L, 1L),   activation = NULL,   use_bias = TRUE,   kernel_initializer = \"glorot_uniform\",   bias_initializer = \"zeros\",   kernel_regularizer = NULL,   bias_regularizer = NULL,   activity_regularizer = NULL,   kernel_constraint = NULL,   bias_constraint = NULL,   ... )"},{"path":"https://keras3.posit.co/dev/reference/layer_conv_3d_transpose.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"3D transposed convolution layer. — layer_conv_3d_transpose","text":"object Object compose layer . tensor, array, sequential model. filters int, dimension output space (number filters transposed convolution). kernel_size int list 1 integer, specifying size transposed convolution window. strides int list 1 integer, specifying stride length transposed convolution. strides > 1 incompatible dilation_rate > 1. padding string, either \"valid\" \"\" (case-insensitive). \"valid\" means padding. \"\" results padding evenly left/right /input. padding=\"\" strides=1, output size input. data_format string, either \"channels_last\" \"channels_first\". ordering dimensions inputs. \"channels_last\" corresponds inputs shape (batch_size, spatial_dim1, spatial_dim2, spatial_dim3, channels) \"channels_first\" corresponds inputs shape (batch_size, channels, spatial_dim1, spatial_dim2, spatial_dim3). defaults image_data_format value found Keras config file ~/.keras/keras.json. never set , \"channels_last\". dilation_rate Scalar integer vector 3 integers specifying dilation rate. Values 1 require strides = 1; different rates per dimension supported. activation Activation function. NULL, activation applied. use_bias bool, TRUE, bias added output. kernel_initializer Initializer convolution kernel. NULL, default initializer (\"glorot_uniform\") used. bias_initializer Initializer bias vector. NULL, default initializer (\"zeros\") used. kernel_regularizer Optional regularizer convolution kernel. bias_regularizer Optional regularizer bias vector. activity_regularizer Optional regularizer function output. kernel_constraint Optional projection function applied kernel updated Optimizer (e.g. used implement norm constraints value constraints layer weights). function must take input unprojected variable must return projected variable (must shape). Constraints safe use asynchronous distributed training. bias_constraint Optional projection function applied bias updated Optimizer. ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_conv_3d_transpose.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"3D transposed convolution layer. — layer_conv_3d_transpose","text":"5D tensor representing activation(conv3d(inputs, kernel) + bias).","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_conv_3d_transpose.html","id":"input-shape","dir":"Reference","previous_headings":"","what":"Input Shape","title":"3D transposed convolution layer. — layer_conv_3d_transpose","text":"data_format=\"channels_last\": 5D tensor shape: (batch_size, spatial_dim1, spatial_dim2, spatial_dim3, channels) data_format=\"channels_first\": 5D tensor shape: (batch_size, channels, spatial_dim1, spatial_dim2, spatial_dim3)","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_conv_3d_transpose.html","id":"output-shape","dir":"Reference","previous_headings":"","what":"Output Shape","title":"3D transposed convolution layer. — layer_conv_3d_transpose","text":"data_format=\"channels_last\": 5D tensor shape: (batch_size, new_spatial_dim1, new_spatial_dim2, new_spatial_dim3, filters) data_format=\"channels_first\": 5D tensor shape: (batch_size, filters, new_spatial_dim1, new_spatial_dim2, new_spatial_dim3)","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_conv_3d_transpose.html","id":"raises","dir":"Reference","previous_headings":"","what":"Raises","title":"3D transposed convolution layer. — layer_conv_3d_transpose","text":"ValueError: strides > 1 dilation_rate > 1.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_conv_3d_transpose.html","id":"references","dir":"Reference","previous_headings":"","what":"References","title":"3D transposed convolution layer. — layer_conv_3d_transpose","text":"guide convolution arithmetic deep learning Deconvolutional Networks","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_conv_3d_transpose.html","id":"example","dir":"Reference","previous_headings":"","what":"Example","title":"3D transposed convolution layer. — layer_conv_3d_transpose","text":"","code":"x <- random_uniform(c(4, 10, 8, 12, 128)) y <- x |> layer_conv_3d_transpose(32, 2, 2, activation = 'relu') shape(y) ## shape(4, 20, 16, 24, 32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_conv_lstm_1d.html","id":null,"dir":"Reference","previous_headings":"","what":"1D Convolutional LSTM. — layer_conv_lstm_1d","title":"1D Convolutional LSTM. — layer_conv_lstm_1d","text":"Similar LSTM layer, input transformations recurrent transformations convolutional.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_conv_lstm_1d.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"1D Convolutional LSTM. — layer_conv_lstm_1d","text":"","code":"layer_conv_lstm_1d(   object,   filters,   kernel_size,   strides = 1L,   padding = \"valid\",   data_format = NULL,   dilation_rate = 1L,   activation = \"tanh\",   recurrent_activation = \"sigmoid\",   use_bias = TRUE,   kernel_initializer = \"glorot_uniform\",   recurrent_initializer = \"orthogonal\",   bias_initializer = \"zeros\",   unit_forget_bias = TRUE,   kernel_regularizer = NULL,   recurrent_regularizer = NULL,   bias_regularizer = NULL,   activity_regularizer = NULL,   kernel_constraint = NULL,   recurrent_constraint = NULL,   bias_constraint = NULL,   dropout = 0,   recurrent_dropout = 0,   seed = NULL,   return_sequences = FALSE,   return_state = FALSE,   go_backwards = FALSE,   stateful = FALSE,   ...,   unroll = NULL )"},{"path":"https://keras3.posit.co/dev/reference/layer_conv_lstm_1d.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"1D Convolutional LSTM. — layer_conv_lstm_1d","text":"object Object compose layer . tensor, array, sequential model. filters int, dimension output space (number filters convolution). kernel_size int tuple/list 1 integer, specifying size convolution window. strides int tuple/list 1 integer, specifying stride length convolution. strides > 1 incompatible dilation_rate > 1. padding string, \"valid\" \"\" (case-insensitive). \"valid\" means padding. \"\" results padding evenly left/right /input output height/width dimension input. data_format string, either \"channels_last\" \"channels_first\". ordering dimensions inputs. \"channels_last\" corresponds inputs shape (batch, steps, features) \"channels_first\" corresponds inputs shape (batch, features, steps). defaults image_data_format value found Keras config file ~/.keras/keras.json. never set , \"channels_last\". dilation_rate int tuple/list 1 integers, specifying dilation rate use dilated convolution. activation Activation function use. default hyperbolic tangent activation function applied (tanh(x)). recurrent_activation Activation function use recurrent step. use_bias Boolean, whether layer uses bias vector. kernel_initializer Initializer kernel weights matrix, used linear transformation inputs. recurrent_initializer Initializer recurrent_kernel weights matrix, used linear transformation recurrent state. bias_initializer Initializer bias vector. unit_forget_bias Boolean. TRUE, add 1 bias forget gate initialization. Use combination bias_initializer=\"zeros\". recommended Jozefowicz et al., 2015 kernel_regularizer Regularizer function applied kernel weights matrix. recurrent_regularizer Regularizer function applied recurrent_kernel weights matrix. bias_regularizer Regularizer function applied bias vector. activity_regularizer Regularizer function applied . kernel_constraint Constraint function applied kernel weights matrix. recurrent_constraint Constraint function applied recurrent_kernel weights matrix. bias_constraint Constraint function applied bias vector. dropout Float 0 1. Fraction units drop linear transformation inputs. recurrent_dropout Float 0 1. Fraction units drop linear transformation recurrent state. seed Random seed dropout. return_sequences Boolean. Whether return last output output sequence, full sequence. Default: FALSE. return_state Boolean. Whether return last state addition output. Default: FALSE. go_backwards Boolean (default: FALSE). TRUE, process input sequence backwards return reversed sequence. stateful Boolean (default FALSE). TRUE, last state sample index batch used initial state sample index following batch. ... forward/backward compatability. unroll Boolean (default: FALSE). TRUE, network unrolled, else symbolic loop used. Unrolling can speed-RNN, although tends memory-intensive. Unrolling suitable short sequences.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_conv_lstm_1d.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"1D Convolutional LSTM. — layer_conv_lstm_1d","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_conv_lstm_1d.html","id":"call-arguments","dir":"Reference","previous_headings":"","what":"Call Arguments","title":"1D Convolutional LSTM. — layer_conv_lstm_1d","text":"inputs: 4D tensor. initial_state: List initial state tensors passed first call cell. mask: Binary tensor shape (samples, timesteps) indicating whether given timestep masked. training: Python boolean indicating whether layer behave training mode inference mode. relevant dropout recurrent_dropout set.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_conv_lstm_1d.html","id":"input-shape","dir":"Reference","previous_headings":"","what":"Input Shape","title":"1D Convolutional LSTM. — layer_conv_lstm_1d","text":"data_format=\"channels_first\": 4D tensor shape: (samples, time, channels, rows) data_format=\"channels_last\": 4D tensor shape: (samples, time, rows, channels)","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_conv_lstm_1d.html","id":"output-shape","dir":"Reference","previous_headings":"","what":"Output Shape","title":"1D Convolutional LSTM. — layer_conv_lstm_1d","text":"return_state: list tensors. first tensor output. remaining tensors last states, 3D tensor shape: (samples, filters, new_rows) data_format='channels_first' shape: (samples, new_rows, filters) data_format='channels_last'. rows values might changed due padding. return_sequences: 4D tensor shape: (samples, timesteps, filters, new_rows) data_format='channels_first' shape: (samples, timesteps, new_rows, filters) data_format='channels_last'. Else, 3D tensor shape: (samples, filters, new_rows) data_format='channels_first' shape: (samples, new_rows, filters) data_format='channels_last'.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_conv_lstm_1d.html","id":"references","dir":"Reference","previous_headings":"","what":"References","title":"1D Convolutional LSTM. — layer_conv_lstm_1d","text":"Shi et al., 2015 (current implementation include feedback loop cells output).","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_conv_lstm_2d.html","id":null,"dir":"Reference","previous_headings":"","what":"2D Convolutional LSTM. — layer_conv_lstm_2d","title":"2D Convolutional LSTM. — layer_conv_lstm_2d","text":"Similar LSTM layer, input transformations recurrent transformations convolutional.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_conv_lstm_2d.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"2D Convolutional LSTM. — layer_conv_lstm_2d","text":"","code":"layer_conv_lstm_2d(   object,   filters,   kernel_size,   strides = 1L,   padding = \"valid\",   data_format = NULL,   dilation_rate = 1L,   activation = \"tanh\",   recurrent_activation = \"sigmoid\",   use_bias = TRUE,   kernel_initializer = \"glorot_uniform\",   recurrent_initializer = \"orthogonal\",   bias_initializer = \"zeros\",   unit_forget_bias = TRUE,   kernel_regularizer = NULL,   recurrent_regularizer = NULL,   bias_regularizer = NULL,   activity_regularizer = NULL,   kernel_constraint = NULL,   recurrent_constraint = NULL,   bias_constraint = NULL,   dropout = 0,   recurrent_dropout = 0,   seed = NULL,   return_sequences = FALSE,   return_state = FALSE,   go_backwards = FALSE,   stateful = FALSE,   ...,   unroll = NULL )"},{"path":"https://keras3.posit.co/dev/reference/layer_conv_lstm_2d.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"2D Convolutional LSTM. — layer_conv_lstm_2d","text":"object Object compose layer . tensor, array, sequential model. filters int, dimension output space (number filters convolution). kernel_size int tuple/list 2 integers, specifying size convolution window. strides int tuple/list 2 integers, specifying stride length convolution. strides > 1 incompatible dilation_rate > 1. padding string, \"valid\" \"\" (case-insensitive). \"valid\" means padding. \"\" results padding evenly left/right /input output height/width dimension input. data_format string, either \"channels_last\" \"channels_first\". ordering dimensions inputs. \"channels_last\" corresponds inputs shape (batch, steps, features) \"channels_first\" corresponds inputs shape (batch, features, steps). defaults image_data_format value found Keras config file ~/.keras/keras.json. never set , \"channels_last\". dilation_rate int tuple/list 2 integers, specifying dilation rate use dilated convolution. activation Activation function use. default hyperbolic tangent activation function applied (tanh(x)). recurrent_activation Activation function use recurrent step. use_bias Boolean, whether layer uses bias vector. kernel_initializer Initializer kernel weights matrix, used linear transformation inputs. recurrent_initializer Initializer recurrent_kernel weights matrix, used linear transformation recurrent state. bias_initializer Initializer bias vector. unit_forget_bias Boolean. TRUE, add 1 bias forget gate initialization. Use combination bias_initializer=\"zeros\". recommended Jozefowicz et al., 2015 kernel_regularizer Regularizer function applied kernel weights matrix. recurrent_regularizer Regularizer function applied recurrent_kernel weights matrix. bias_regularizer Regularizer function applied bias vector. activity_regularizer Regularizer function applied . kernel_constraint Constraint function applied kernel weights matrix. recurrent_constraint Constraint function applied recurrent_kernel weights matrix. bias_constraint Constraint function applied bias vector. dropout Float 0 1. Fraction units drop linear transformation inputs. recurrent_dropout Float 0 1. Fraction units drop linear transformation recurrent state. seed Random seed dropout. return_sequences Boolean. Whether return last output output sequence, full sequence. Default: FALSE. return_state Boolean. Whether return last state addition output. Default: FALSE. go_backwards Boolean (default: FALSE). TRUE, process input sequence backwards return reversed sequence. stateful Boolean (default FALSE). TRUE, last state sample index batch used initial state sample index following batch. ... forward/backward compatability. unroll Boolean (default: FALSE). TRUE, network unrolled, else symbolic loop used. Unrolling can speed-RNN, although tends memory-intensive. Unrolling suitable short sequences.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_conv_lstm_2d.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"2D Convolutional LSTM. — layer_conv_lstm_2d","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_conv_lstm_2d.html","id":"call-arguments","dir":"Reference","previous_headings":"","what":"Call Arguments","title":"2D Convolutional LSTM. — layer_conv_lstm_2d","text":"inputs: 5D tensor. mask: Binary tensor shape (samples, timesteps) indicating whether given timestep masked. training: Python boolean indicating whether layer behave training mode inference mode. relevant dropout recurrent_dropout set. initial_state: List initial state tensors passed first call cell.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_conv_lstm_2d.html","id":"input-shape","dir":"Reference","previous_headings":"","what":"Input Shape","title":"2D Convolutional LSTM. — layer_conv_lstm_2d","text":"data_format='channels_first': 5D tensor shape: (samples, time, channels, rows, cols) data_format='channels_last': 5D tensor shape: (samples, time, rows, cols, channels)","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_conv_lstm_2d.html","id":"output-shape","dir":"Reference","previous_headings":"","what":"Output Shape","title":"2D Convolutional LSTM. — layer_conv_lstm_2d","text":"return_state: list tensors. first tensor output. remaining tensors last states, 4D tensor shape: (samples, filters, new_rows, new_cols) data_format='channels_first' shape: (samples, new_rows, new_cols, filters) data_format='channels_last'. rows cols values might changed due padding. return_sequences: 5D tensor shape: (samples, timesteps, filters, new_rows, new_cols) data_format='channels_first' shape: (samples, timesteps, new_rows, new_cols, filters) data_format='channels_last'. Else, 4D tensor shape: (samples, filters, new_rows, new_cols) data_format='channels_first' shape: (samples, new_rows, new_cols, filters) data_format='channels_last'.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_conv_lstm_2d.html","id":"references","dir":"Reference","previous_headings":"","what":"References","title":"2D Convolutional LSTM. — layer_conv_lstm_2d","text":"Shi et al., 2015 (current implementation include feedback loop cells output).","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_conv_lstm_3d.html","id":null,"dir":"Reference","previous_headings":"","what":"3D Convolutional LSTM. — layer_conv_lstm_3d","title":"3D Convolutional LSTM. — layer_conv_lstm_3d","text":"Similar LSTM layer, input transformations recurrent transformations convolutional.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_conv_lstm_3d.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"3D Convolutional LSTM. — layer_conv_lstm_3d","text":"","code":"layer_conv_lstm_3d(   object,   filters,   kernel_size,   strides = 1L,   padding = \"valid\",   data_format = NULL,   dilation_rate = 1L,   activation = \"tanh\",   recurrent_activation = \"sigmoid\",   use_bias = TRUE,   kernel_initializer = \"glorot_uniform\",   recurrent_initializer = \"orthogonal\",   bias_initializer = \"zeros\",   unit_forget_bias = TRUE,   kernel_regularizer = NULL,   recurrent_regularizer = NULL,   bias_regularizer = NULL,   activity_regularizer = NULL,   kernel_constraint = NULL,   recurrent_constraint = NULL,   bias_constraint = NULL,   dropout = 0,   recurrent_dropout = 0,   seed = NULL,   return_sequences = FALSE,   return_state = FALSE,   go_backwards = FALSE,   stateful = FALSE,   ...,   unroll = NULL )"},{"path":"https://keras3.posit.co/dev/reference/layer_conv_lstm_3d.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"3D Convolutional LSTM. — layer_conv_lstm_3d","text":"object Object compose layer . tensor, array, sequential model. filters int, dimension output space (number filters convolution). kernel_size int tuple/list 3 integers, specifying size convolution window. strides int tuple/list 3 integers, specifying stride length convolution. strides > 1 incompatible dilation_rate > 1. padding string, \"valid\" \"\" (case-insensitive). \"valid\" means padding. \"\" results padding evenly left/right /input output height/width dimension input. data_format string, either \"channels_last\" \"channels_first\". ordering dimensions inputs. \"channels_last\" corresponds inputs shape (batch, steps, features) \"channels_first\" corresponds inputs shape (batch, features, steps). defaults image_data_format value found Keras config file ~/.keras/keras.json. never set , \"channels_last\". dilation_rate int tuple/list 3 integers, specifying dilation rate use dilated convolution. activation Activation function use. default hyperbolic tangent activation function applied (tanh(x)). recurrent_activation Activation function use recurrent step. use_bias Boolean, whether layer uses bias vector. kernel_initializer Initializer kernel weights matrix, used linear transformation inputs. recurrent_initializer Initializer recurrent_kernel weights matrix, used linear transformation recurrent state. bias_initializer Initializer bias vector. unit_forget_bias Boolean. TRUE, add 1 bias forget gate initialization. Use combination bias_initializer=\"zeros\". recommended Jozefowicz et al., 2015 kernel_regularizer Regularizer function applied kernel weights matrix. recurrent_regularizer Regularizer function applied recurrent_kernel weights matrix. bias_regularizer Regularizer function applied bias vector. activity_regularizer Regularizer function applied . kernel_constraint Constraint function applied kernel weights matrix. recurrent_constraint Constraint function applied recurrent_kernel weights matrix. bias_constraint Constraint function applied bias vector. dropout Float 0 1. Fraction units drop linear transformation inputs. recurrent_dropout Float 0 1. Fraction units drop linear transformation recurrent state. seed Random seed dropout. return_sequences Boolean. Whether return last output output sequence, full sequence. Default: FALSE. return_state Boolean. Whether return last state addition output. Default: FALSE. go_backwards Boolean (default: FALSE). TRUE, process input sequence backwards return reversed sequence. stateful Boolean (default FALSE). TRUE, last state sample index batch used initial state sample index following batch. ... forward/backward compatability. unroll Boolean (default: FALSE). TRUE, network unrolled, else symbolic loop used. Unrolling can speed-RNN, although tends memory-intensive. Unrolling suitable short sequences.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_conv_lstm_3d.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"3D Convolutional LSTM. — layer_conv_lstm_3d","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_conv_lstm_3d.html","id":"call-arguments","dir":"Reference","previous_headings":"","what":"Call Arguments","title":"3D Convolutional LSTM. — layer_conv_lstm_3d","text":"inputs: 6D tensor. mask: Binary tensor shape (samples, timesteps) indicating whether given timestep masked. training: Python boolean indicating whether layer behave training mode inference mode. relevant dropout recurrent_dropout set. initial_state: List initial state tensors passed first call cell.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_conv_lstm_3d.html","id":"input-shape","dir":"Reference","previous_headings":"","what":"Input Shape","title":"3D Convolutional LSTM. — layer_conv_lstm_3d","text":"data_format='channels_first': 5D tensor shape: (samples, time, channels, *spatial_dims) data_format='channels_last': 5D tensor shape: (samples, time, *spatial_dims, channels)","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_conv_lstm_3d.html","id":"output-shape","dir":"Reference","previous_headings":"","what":"Output Shape","title":"3D Convolutional LSTM. — layer_conv_lstm_3d","text":"return_state: list tensors. first tensor output. remaining tensors last states, 4D tensor shape: (samples, filters, *spatial_dims) data_format='channels_first' shape: (samples, *spatial_dims, filters) data_format='channels_last'. return_sequences: 5D tensor shape: (samples, timesteps, filters, *spatial_dims) data_format='channels_first' shape: (samples, timesteps, *spatial_dims, filters) data_format='channels_last'. Else, 4D tensor shape: (samples, filters, *spatial_dims) data_format='channels_first' shape: (samples, *spatial_dims, filters) data_format='channels_last'.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_conv_lstm_3d.html","id":"references","dir":"Reference","previous_headings":"","what":"References","title":"3D Convolutional LSTM. — layer_conv_lstm_3d","text":"Shi et al., 2015 (current implementation include feedback loop cells output).","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_cropping_1d.html","id":null,"dir":"Reference","previous_headings":"","what":"Cropping layer for 1D input (e.g. temporal sequence). — layer_cropping_1d","title":"Cropping layer for 1D input (e.g. temporal sequence). — layer_cropping_1d","text":"crops along time dimension (axis 2).","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_cropping_1d.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Cropping layer for 1D input (e.g. temporal sequence). — layer_cropping_1d","text":"","code":"layer_cropping_1d(object, cropping = list(1L, 1L), ...)"},{"path":"https://keras3.posit.co/dev/reference/layer_cropping_1d.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Cropping layer for 1D input (e.g. temporal sequence). — layer_cropping_1d","text":"object Object compose layer . tensor, array, sequential model. cropping Int, list int (length 2). int: many units trimmed beginning end cropping dimension (axis 1). list 2 ints: many units trimmed beginning end cropping dimension ((left_crop, right_crop)). ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_cropping_1d.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Cropping layer for 1D input (e.g. temporal sequence). — layer_cropping_1d","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_cropping_1d.html","id":"example","dir":"Reference","previous_headings":"","what":"Example","title":"Cropping layer for 1D input (e.g. temporal sequence). — layer_cropping_1d","text":"","code":"input_shape <- c(2, 3, 2) x <- op_arange(prod(input_shape)) |> op_reshape(input_shape) x ## tf.Tensor( ## [[[ 1.  2.] ##   [ 3.  4.] ##   [ 5.  6.]] ## ##  [[ 7.  8.] ##   [ 9. 10.] ##   [11. 12.]]], shape=(2, 3, 2), dtype=float32) y <- x |> layer_cropping_1d(cropping = 1) y ## tf.Tensor( ## [[[ 3.  4.]] ## ##  [[ 9. 10.]]], shape=(2, 1, 2), dtype=float32)"},{"path":"https://keras3.posit.co/dev/reference/layer_cropping_1d.html","id":"input-shape","dir":"Reference","previous_headings":"","what":"Input Shape","title":"Cropping layer for 1D input (e.g. temporal sequence). — layer_cropping_1d","text":"3D tensor shape (batch_size, axis_to_crop, features)","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_cropping_1d.html","id":"output-shape","dir":"Reference","previous_headings":"","what":"Output Shape","title":"Cropping layer for 1D input (e.g. temporal sequence). — layer_cropping_1d","text":"3D tensor shape (batch_size, cropped_axis, features)","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_cropping_2d.html","id":null,"dir":"Reference","previous_headings":"","what":"Cropping layer for 2D input (e.g. picture). — layer_cropping_2d","title":"Cropping layer for 2D input (e.g. picture). — layer_cropping_2d","text":"crops along spatial dimensions, .e. height width.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_cropping_2d.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Cropping layer for 2D input (e.g. picture). — layer_cropping_2d","text":"","code":"layer_cropping_2d(   object,   cropping = list(list(0L, 0L), list(0L, 0L)),   data_format = NULL,   ... )"},{"path":"https://keras3.posit.co/dev/reference/layer_cropping_2d.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Cropping layer for 2D input (e.g. picture). — layer_cropping_2d","text":"object Object compose layer . tensor, array, sequential model. cropping Int, list 2 ints, list 2 lists 2 ints. int: symmetric cropping applied height width. list 2 ints: interpreted two different symmetric cropping values height width: (symmetric_height_crop, symmetric_width_crop). list 2 lists 2 ints: interpreted ((top_crop, bottom_crop), (left_crop, right_crop)). data_format string, one \"channels_last\" (default) \"channels_first\". ordering dimensions inputs. \"channels_last\" corresponds inputs shape (batch_size, height, width, channels) \"channels_first\" corresponds inputs shape (batch_size, channels, height, width). unspecified, uses image_data_format value found Keras config file ~/.keras/keras.json (exists). Defaults \"channels_last\". ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_cropping_2d.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Cropping layer for 2D input (e.g. picture). — layer_cropping_2d","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_cropping_2d.html","id":"example","dir":"Reference","previous_headings":"","what":"Example","title":"Cropping layer for 2D input (e.g. picture). — layer_cropping_2d","text":"","code":"input_shape <- c(2, 28, 28, 3) x <- op_arange(prod(input_shape), dtype ='int32') |> op_reshape(input_shape) y <- x |> layer_cropping_2d(cropping=list(c(2, 2), c(4, 4))) shape(y) ## shape(2, 24, 20, 3)"},{"path":"https://keras3.posit.co/dev/reference/layer_cropping_2d.html","id":"input-shape","dir":"Reference","previous_headings":"","what":"Input Shape","title":"Cropping layer for 2D input (e.g. picture). — layer_cropping_2d","text":"4D tensor shape: data_format \"channels_last\": (batch_size, height, width, channels) data_format \"channels_first\": (batch_size, channels, height, width)","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_cropping_2d.html","id":"output-shape","dir":"Reference","previous_headings":"","what":"Output Shape","title":"Cropping layer for 2D input (e.g. picture). — layer_cropping_2d","text":"4D tensor shape: data_format \"channels_last\": (batch_size, cropped_height, cropped_width, channels) data_format \"channels_first\": (batch_size, channels, cropped_height, cropped_width)","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_cropping_3d.html","id":null,"dir":"Reference","previous_headings":"","what":"Cropping layer for 3D data (e.g. spatial or spatio-temporal). — layer_cropping_3d","title":"Cropping layer for 3D data (e.g. spatial or spatio-temporal). — layer_cropping_3d","text":"Cropping layer 3D data (e.g. spatial spatio-temporal).","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_cropping_3d.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Cropping layer for 3D data (e.g. spatial or spatio-temporal). — layer_cropping_3d","text":"","code":"layer_cropping_3d(   object,   cropping = list(list(1L, 1L), list(1L, 1L), list(1L, 1L)),   data_format = NULL,   ... )"},{"path":"https://keras3.posit.co/dev/reference/layer_cropping_3d.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Cropping layer for 3D data (e.g. spatial or spatio-temporal). — layer_cropping_3d","text":"object Object compose layer . tensor, array, sequential model. cropping Int, list 3 ints, list 3 lists 2 ints. int: symmetric cropping applied depth, height, width. list 3 ints: interpreted three different symmetric cropping values depth, height, width: (symmetric_dim1_crop, symmetric_dim2_crop, symmetric_dim3_crop). list 3 lists 2 ints: interpreted ((left_dim1_crop, right_dim1_crop), (left_dim2_crop, right_dim2_crop), (left_dim3_crop, right_dim3_crop)). data_format string, one \"channels_last\" (default) \"channels_first\". ordering dimensions inputs. \"channels_last\" corresponds inputs shape (batch_size, spatial_dim1, spatial_dim2, spatial_dim3, channels) \"channels_first\" corresponds inputs shape (batch_size, channels, spatial_dim1, spatial_dim2, spatial_dim3). unspecified, uses image_data_format value found Keras config file ~/.keras/keras.json (exists). Defaults \"channels_last\". ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_cropping_3d.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Cropping layer for 3D data (e.g. spatial or spatio-temporal). — layer_cropping_3d","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_cropping_3d.html","id":"example","dir":"Reference","previous_headings":"","what":"Example","title":"Cropping layer for 3D data (e.g. spatial or spatio-temporal). — layer_cropping_3d","text":"","code":"input_shape <- c(2, 28, 28, 10, 3) x <- input_shape %>% { op_reshape(seq(prod(.)), .) } y <- x |> layer_cropping_3d(cropping = c(2, 4, 2)) shape(y) ## shape(2, 24, 20, 6, 3)"},{"path":"https://keras3.posit.co/dev/reference/layer_cropping_3d.html","id":"input-shape","dir":"Reference","previous_headings":"","what":"Input Shape","title":"Cropping layer for 3D data (e.g. spatial or spatio-temporal). — layer_cropping_3d","text":"5D tensor shape: data_format \"channels_last\": (batch_size, first_axis_to_crop, second_axis_to_crop, third_axis_to_crop, channels) data_format \"channels_first\": (batch_size, channels, first_axis_to_crop, second_axis_to_crop, third_axis_to_crop)","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_cropping_3d.html","id":"output-shape","dir":"Reference","previous_headings":"","what":"Output Shape","title":"Cropping layer for 3D data (e.g. spatial or spatio-temporal). — layer_cropping_3d","text":"5D tensor shape: data_format \"channels_last\": (batch_size, first_cropped_axis, second_cropped_axis, third_cropped_axis, channels) data_format \"channels_first\": (batch_size, channels, first_cropped_axis, second_cropped_axis, third_cropped_axis)","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_cut_mix.html","id":null,"dir":"Reference","previous_headings":"","what":"CutMix data augmentation technique. — layer_cut_mix","title":"CutMix data augmentation technique. — layer_cut_mix","text":"CutMix data augmentation method patches cut pasted two images dataset, labels also mixed proportionally area patches.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_cut_mix.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"CutMix data augmentation technique. — layer_cut_mix","text":"","code":"layer_cut_mix(object, factor = 1, seed = NULL, data_format = NULL, ...)"},{"path":"https://keras3.posit.co/dev/reference/layer_cut_mix.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"CutMix data augmentation technique. — layer_cut_mix","text":"object Object compose layer . tensor, array, sequential model. factor single float tuple two floats 0 1. tuple numbers passed, factor sampled two values. single float passed, value 0 passed float sampled. values define range mixing weight sampled. higher factor increases variability patch sizes, leading diverse larger mixed patches. Defaults 1. seed Integer. Used create random seed. data_format string, either \"channels_last\" \"channels_first\". ordering dimensions inputs. \"channels_last\" corresponds inputs shape (batch, height, width, channels) \"channels_first\" corresponds inputs shape (batch, channels, height, width). defaults image_data_format value found Keras config file ~/.keras/keras.json. never set , \"channels_last\". ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_cut_mix.html","id":"references","dir":"Reference","previous_headings":"","what":"References","title":"CutMix data augmentation technique. — layer_cut_mix","text":"CutMix paper.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_dense.html","id":null,"dir":"Reference","previous_headings":"","what":"Just your regular densely-connected NN layer. — layer_dense","title":"Just your regular densely-connected NN layer. — layer_dense","text":"Dense implements operation: output = activation(dot(input, kernel) + bias) activation element-wise activation function passed activation argument, kernel weights matrix created layer, bias bias vector created layer (applicable use_bias TRUE).","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_dense.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Just your regular densely-connected NN layer. — layer_dense","text":"","code":"layer_dense(   object,   units,   activation = NULL,   use_bias = TRUE,   kernel_initializer = \"glorot_uniform\",   bias_initializer = \"zeros\",   kernel_regularizer = NULL,   bias_regularizer = NULL,   activity_regularizer = NULL,   kernel_constraint = NULL,   bias_constraint = NULL,   lora_rank = NULL,   lora_alpha = NULL,   ... )"},{"path":"https://keras3.posit.co/dev/reference/layer_dense.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Just your regular densely-connected NN layer. — layer_dense","text":"object Object compose layer . tensor, array, sequential model. units Positive integer, dimensionality output space. activation Activation function use. specify anything, activation applied (ie. \"linear\" activation: (x) = x). use_bias Boolean, whether layer uses bias vector. kernel_initializer Initializer kernel weights matrix. bias_initializer Initializer bias vector. kernel_regularizer Regularizer function applied kernel weights matrix. bias_regularizer Regularizer function applied bias vector. activity_regularizer Regularizer function applied output layer (\"activation\"). kernel_constraint Constraint function applied kernel weights matrix. bias_constraint Constraint function applied bias vector. lora_rank Optional integer. set, layer's forward pass implement LoRA (Low-Rank Adaptation) provided rank. LoRA sets layer's kernel non-trainable replaces delta original kernel, obtained via multiplying two lower-rank trainable matrices. can useful reduce computation cost fine-tuning large dense layers. can also enable LoRA existing Dense layer calling layer$enable_lora(rank). lora_alpha Optional integer. Scales low-rank adaptation delta forward pass. delta scaled lora_alpha / lora_rank, letting tune LoRA adjustment strength independently lora_rank. ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_dense.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Just your regular densely-connected NN layer. — layer_dense","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_dense.html","id":"note","dir":"Reference","previous_headings":"","what":"Note","title":"Just your regular densely-connected NN layer. — layer_dense","text":"input layer rank greater 2, Dense computes dot product inputs kernel along last axis inputs axis 0 kernel (using tf.tensordot). example, input dimensions (batch_size, d0, d1), create kernel shape (d1, units), kernel operates along axis 2 input, every sub-tensor shape (1, 1, d1) (batch_size * d0 sub-tensors). output case shape (batch_size, d0, units).","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_dense.html","id":"input-shape","dir":"Reference","previous_headings":"","what":"Input Shape","title":"Just your regular densely-connected NN layer. — layer_dense","text":"N-D tensor shape: (batch_size, ..., input_dim). common situation 2D input shape (batch_size, input_dim).","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_dense.html","id":"output-shape","dir":"Reference","previous_headings":"","what":"Output Shape","title":"Just your regular densely-connected NN layer. — layer_dense","text":"N-D tensor shape: (batch_size, ..., units). instance, 2D input shape (batch_size, input_dim), output shape (batch_size, units).","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_dense.html","id":"methods","dir":"Reference","previous_headings":"","what":"Methods","title":"Just your regular densely-connected NN layer. — layer_dense","text":"","code":"enable_lora(   rank,   lora_alpha = NULL,   a_initializer = 'he_uniform',   b_initializer = 'zeros' ) quantize(mode, type_check = TRUE)"},{"path":"https://keras3.posit.co/dev/reference/layer_dense.html","id":"readonly-properties-","dir":"Reference","previous_headings":"","what":"Readonly properties:","title":"Just your regular densely-connected NN layer. — layer_dense","text":"kernel","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_depthwise_conv_1d.html","id":null,"dir":"Reference","previous_headings":"","what":"1D depthwise convolution layer. — layer_depthwise_conv_1d","title":"1D depthwise convolution layer. — layer_depthwise_conv_1d","text":"Depthwise convolution type convolution input channel convolved different kernel (called depthwise kernel). can understand depthwise convolution first step depthwise separable convolution. implemented via following steps: Split input individual channels. Convolve channel individual depthwise kernel depth_multiplier output channels. Concatenate convolved outputs along channels axis. Unlike regular 1D convolution, depthwise convolution mix information across different input channels. depth_multiplier argument determines many filters applied one input channel. , controls amount output channels generated per input channel depthwise step.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_depthwise_conv_1d.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"1D depthwise convolution layer. — layer_depthwise_conv_1d","text":"","code":"layer_depthwise_conv_1d(   object,   kernel_size,   strides = 1L,   padding = \"valid\",   depth_multiplier = 1L,   data_format = NULL,   dilation_rate = 1L,   activation = NULL,   use_bias = TRUE,   depthwise_initializer = \"glorot_uniform\",   bias_initializer = \"zeros\",   depthwise_regularizer = NULL,   bias_regularizer = NULL,   activity_regularizer = NULL,   depthwise_constraint = NULL,   bias_constraint = NULL,   ... )"},{"path":"https://keras3.posit.co/dev/reference/layer_depthwise_conv_1d.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"1D depthwise convolution layer. — layer_depthwise_conv_1d","text":"object Object compose layer . tensor, array, sequential model. kernel_size int list 1 integer, specifying size depthwise convolution window. strides int list 1 integer, specifying stride length convolution. strides > 1 incompatible dilation_rate > 1. padding string, either \"valid\" \"\" (case-insensitive). \"valid\" means padding. \"\" results padding evenly left/right /input. padding=\"\" strides=1, output size input. depth_multiplier number depthwise convolution output channels input channel. total number depthwise convolution output channels equal input_channel * depth_multiplier. data_format string, either \"channels_last\" \"channels_first\". ordering dimensions inputs. \"channels_last\" corresponds inputs shape (batch, steps, features) \"channels_first\" corresponds inputs shape (batch, features, steps). defaults image_data_format value found Keras config file ~/.keras/keras.json. never set , \"channels_last\". dilation_rate int list 1 integers, specifying dilation rate use dilated convolution. activation Activation function. NULL, activation applied. use_bias bool, TRUE, bias added output. depthwise_initializer Initializer convolution kernel. NULL, default initializer (\"glorot_uniform\") used. bias_initializer Initializer bias vector. NULL, default initializer (\"zeros\") used. depthwise_regularizer Optional regularizer convolution kernel. bias_regularizer Optional regularizer bias vector. activity_regularizer Optional regularizer function output. depthwise_constraint Optional projection function applied kernel updated Optimizer (e.g. used implement norm constraints value constraints layer weights). function must take input unprojected variable must return projected variable (must shape). Constraints safe use asynchronous distributed training. bias_constraint Optional projection function applied bias updated Optimizer. ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_depthwise_conv_1d.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"1D depthwise convolution layer. — layer_depthwise_conv_1d","text":"3D tensor representing activation(depthwise_conv1d(inputs, kernel) + bias).","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_depthwise_conv_1d.html","id":"input-shape","dir":"Reference","previous_headings":"","what":"Input Shape","title":"1D depthwise convolution layer. — layer_depthwise_conv_1d","text":"data_format=\"channels_last\": 3D tensor shape: (batch_shape, steps, channels) data_format=\"channels_first\": 3D tensor shape: (batch_shape, channels, steps)","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_depthwise_conv_1d.html","id":"output-shape","dir":"Reference","previous_headings":"","what":"Output Shape","title":"1D depthwise convolution layer. — layer_depthwise_conv_1d","text":"data_format=\"channels_last\": 3D tensor shape: (batch_shape, new_steps, channels * depth_multiplier) data_format=\"channels_first\": 3D tensor shape: (batch_shape, channels * depth_multiplier, new_steps)","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_depthwise_conv_1d.html","id":"raises","dir":"Reference","previous_headings":"","what":"Raises","title":"1D depthwise convolution layer. — layer_depthwise_conv_1d","text":"ValueError: strides > 1 dilation_rate > 1.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_depthwise_conv_1d.html","id":"example","dir":"Reference","previous_headings":"","what":"Example","title":"1D depthwise convolution layer. — layer_depthwise_conv_1d","text":"","code":"x <- random_uniform(c(4, 10, 12)) y <- x |> layer_depthwise_conv_1d(   kernel_size = 3,   depth_multiplier = 3,   activation = 'relu' ) shape(y) ## shape(4, 8, 36)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_depthwise_conv_2d.html","id":null,"dir":"Reference","previous_headings":"","what":"2D depthwise convolution layer. — layer_depthwise_conv_2d","title":"2D depthwise convolution layer. — layer_depthwise_conv_2d","text":"Depthwise convolution type convolution input channel convolved different kernel (called depthwise kernel). can understand depthwise convolution first step depthwise separable convolution. implemented via following steps: Split input individual channels. Convolve channel individual depthwise kernel depth_multiplier output channels. Concatenate convolved outputs along channels axis. Unlike regular 2D convolution, depthwise convolution mix information across different input channels. depth_multiplier argument determines many filters applied one input channel. , controls amount output channels generated per input channel depthwise step.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_depthwise_conv_2d.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"2D depthwise convolution layer. — layer_depthwise_conv_2d","text":"","code":"layer_depthwise_conv_2d(   object,   kernel_size,   strides = list(1L, 1L),   padding = \"valid\",   depth_multiplier = 1L,   data_format = NULL,   dilation_rate = list(1L, 1L),   activation = NULL,   use_bias = TRUE,   depthwise_initializer = \"glorot_uniform\",   bias_initializer = \"zeros\",   depthwise_regularizer = NULL,   bias_regularizer = NULL,   activity_regularizer = NULL,   depthwise_constraint = NULL,   bias_constraint = NULL,   ... )"},{"path":"https://keras3.posit.co/dev/reference/layer_depthwise_conv_2d.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"2D depthwise convolution layer. — layer_depthwise_conv_2d","text":"object Object compose layer . tensor, array, sequential model. kernel_size int list 2 integer, specifying size depthwise convolution window. strides int list 2 integer, specifying stride length depthwise convolution. strides > 1 incompatible dilation_rate > 1. padding string, either \"valid\" \"\" (case-insensitive). \"valid\" means padding. \"\" results padding evenly left/right /input. padding=\"\" strides=1, output size input. depth_multiplier number depthwise convolution output channels input channel. total number depthwise convolution output channels equal input_channel * depth_multiplier. data_format string, either \"channels_last\" \"channels_first\". ordering dimensions inputs. \"channels_last\" corresponds inputs shape (batch, height, width, channels) \"channels_first\" corresponds inputs shape (batch, channels, height, width). defaults image_data_format value found Keras config file ~/.keras/keras.json. never set , \"channels_last\". dilation_rate int list 2 integers, specifying dilation rate use dilated convolution. activation Activation function. NULL, activation applied. use_bias bool, TRUE, bias added output. depthwise_initializer Initializer convolution kernel. NULL, default initializer (\"glorot_uniform\") used. bias_initializer Initializer bias vector. NULL, default initializer (\"zeros\") used. depthwise_regularizer Optional regularizer convolution kernel. bias_regularizer Optional regularizer bias vector. activity_regularizer Optional regularizer function output. depthwise_constraint Optional projection function applied kernel updated Optimizer (e.g. used implement norm constraints value constraints layer weights). function must take input unprojected variable must return projected variable (must shape). Constraints safe use asynchronous distributed training. bias_constraint Optional projection function applied bias updated Optimizer. ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_depthwise_conv_2d.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"2D depthwise convolution layer. — layer_depthwise_conv_2d","text":"4D tensor representing activation(depthwise_conv2d(inputs, kernel) + bias).","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_depthwise_conv_2d.html","id":"input-shape","dir":"Reference","previous_headings":"","what":"Input Shape","title":"2D depthwise convolution layer. — layer_depthwise_conv_2d","text":"data_format=\"channels_last\": 4D tensor shape: (batch_size, height, width, channels) data_format=\"channels_first\": 4D tensor shape: (batch_size, channels, height, width)","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_depthwise_conv_2d.html","id":"output-shape","dir":"Reference","previous_headings":"","what":"Output Shape","title":"2D depthwise convolution layer. — layer_depthwise_conv_2d","text":"data_format=\"channels_last\": 4D tensor shape: (batch_size, new_height, new_width, channels * depth_multiplier) data_format=\"channels_first\": 4D tensor shape: (batch_size, channels * depth_multiplier, new_height, new_width)","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_depthwise_conv_2d.html","id":"raises","dir":"Reference","previous_headings":"","what":"Raises","title":"2D depthwise convolution layer. — layer_depthwise_conv_2d","text":"ValueError: strides > 1 dilation_rate > 1.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_depthwise_conv_2d.html","id":"example","dir":"Reference","previous_headings":"","what":"Example","title":"2D depthwise convolution layer. — layer_depthwise_conv_2d","text":"","code":"x <- random_uniform(c(4, 10, 10, 12)) y <- x |> layer_depthwise_conv_2d(kernel_size = 3, activation = 'relu') shape(y) ## shape(4, 8, 8, 12)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_discretization.html","id":null,"dir":"Reference","previous_headings":"","what":"A preprocessing layer which buckets continuous features by ranges. — layer_discretization","title":"A preprocessing layer which buckets continuous features by ranges. — layer_discretization","text":"layer place element input data one several contiguous ranges output integer index indicating range element placed . Note: layer safe use inside tf.data pipeline (independently backend using).","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_discretization.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"A preprocessing layer which buckets continuous features by ranges. — layer_discretization","text":"","code":"layer_discretization(   object,   bin_boundaries = NULL,   num_bins = NULL,   epsilon = 0.01,   output_mode = \"int\",   sparse = FALSE,   dtype = NULL,   name = NULL )"},{"path":"https://keras3.posit.co/dev/reference/layer_discretization.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"A preprocessing layer which buckets continuous features by ranges. — layer_discretization","text":"object Object compose layer . tensor, array, sequential model. bin_boundaries list bin boundaries. leftmost rightmost bins always extend -Inf Inf, bin_boundaries = c(0, 1, 2) generates bins (-Inf, 0), [0, 1), [1, 2), [2, +Inf). option set, adapt() called. num_bins integer number bins compute. option set, bin_boundaries set adapt() called learn bin boundaries. epsilon Error tolerance, typically small fraction close zero (e.g. 0.01). Higher values epsilon increase quantile approximation, hence result unequal buckets, improve performance resource consumption. output_mode Specification output layer. Values can \"int\", \"one_hot\", \"multi_hot\", \"count\" configuring layer follows: \"int\": Return discretized bin indices directly. \"one_hot\": Encodes individual element input array size num_bins, containing 1 input's bin index. last dimension size 1, encode dimension.  last dimension size 1, append new dimension encoded output. \"multi_hot\": Encodes sample input single array size num_bins, containing 1 bin index index present sample. Treats last dimension sample dimension, input shape (..., sample_length), output shape (..., num_tokens). \"count\": \"multi_hot\", int array contains count number times bin index appeared sample. Defaults \"int\". sparse Boolean. applicable \"one_hot\", \"multi_hot\", \"count\" output modes. supported TensorFlow backend. TRUE, returns SparseTensor instead dense Tensor. Defaults FALSE. dtype datatype (e.g., \"float32\"). name String, name object","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_discretization.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"A preprocessing layer which buckets continuous features by ranges. — layer_discretization","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_discretization.html","id":"input-shape","dir":"Reference","previous_headings":"","what":"Input Shape","title":"A preprocessing layer which buckets continuous features by ranges. — layer_discretization","text":"array dimension 2 higher.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_discretization.html","id":"output-shape","dir":"Reference","previous_headings":"","what":"Output Shape","title":"A preprocessing layer which buckets continuous features by ranges. — layer_discretization","text":"input shape.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_discretization.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"A preprocessing layer which buckets continuous features by ranges. — layer_discretization","text":"Discretize float values based provided buckets.     Discretize float values based number buckets compute.","code":"input <- op_array(rbind(c(-1.5, 1, 3.4, 0.5),                        c(0, 3, 1.3, 0),                        c(-.5, 0, .5, 1),                        c(1.5, 2, 2.5, 3))) output <- input |> layer_discretization(bin_boundaries = c(0, 1, 2)) output ## tf.Tensor( ## [[0 2 3 1] ##  [1 3 2 1] ##  [0 1 1 2] ##  [2 3 3 3]], shape=(4, 4), dtype=int64) layer <- layer_discretization(num_bins = 4, epsilon = 0.01) layer |> adapt(input) layer(input) ## tf.Tensor( ## [[0 2 3 1] ##  [1 3 2 1] ##  [0 1 1 2] ##  [2 3 3 3]], shape=(4, 4), dtype=int64)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_dot.html","id":null,"dir":"Reference","previous_headings":"","what":"Computes element-wise dot product of two tensors. — layer_dot","title":"Computes element-wise dot product of two tensors. — layer_dot","text":"takes list inputs size 2, axes corresponding input along dot product performed. say x y two input tensors shapes (2, 3, 5) (2, 10, 3). batch dimension size inputs, axes correspond dimensions size corresponding inputs. e.g. axes = c(1, 2), dot product x, y result tensor shape (2, 5, 10)","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_dot.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes element-wise dot product of two tensors. — layer_dot","text":"","code":"layer_dot(inputs, ..., axes, normalize = FALSE)"},{"path":"https://keras3.posit.co/dev/reference/layer_dot.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Computes element-wise dot product of two tensors. — layer_dot","text":"inputs layers combine ... Standard layer keyword arguments. axes Integer list integers, axis axes along take dot product. list, two integers corresponding desired axis first input desired axis second input, respectively. Note size two selected axes must match. normalize Whether L2-normalize samples along dot product axis taking dot product. set TRUE, output dot product cosine proximity two samples.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_dot.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Computes element-wise dot product of two tensors. — layer_dot","text":"tensor, dot product samples inputs.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_dot.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Computes element-wise dot product of two tensors. — layer_dot","text":"Usage Keras model:","code":"x <- op_reshape(0:9,   c(1, 5, 2)) y <- op_reshape(10:19, c(1, 2, 5)) layer_dot(x, y, axes=c(2, 3)) ## tf.Tensor( ## [[[260 360] ##   [320 445]]], shape=(1, 2, 2), dtype=int32) x1 <- op_reshape(0:9, c(5, 2)) |> layer_dense(8) x2 <- op_reshape(10:19, c(5, 2)) |> layer_dense(8) shape(x1) ## shape(5, 8) shape(x2) ## shape(5, 8) y <- layer_dot(x1, x2, axes=2) shape(y) ## shape(5, 1)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_dropout.html","id":null,"dir":"Reference","previous_headings":"","what":"Applies dropout to the input. — layer_dropout","title":"Applies dropout to the input. — layer_dropout","text":"Dropout layer randomly sets input units 0 frequency rate step training time, helps prevent overfitting. Inputs set 0 scaled 1 / (1 - rate) sum inputs unchanged. Note Dropout layer applies training set TRUE call(), values dropped inference. using model.fit, training appropriately set TRUE automatically. contexts, can set argument explicitly TRUE calling layer. (contrast setting trainable=FALSE Dropout layer. trainable affect layer's behavior, Dropout variables/weights can frozen training.)","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_dropout.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Applies dropout to the input. — layer_dropout","text":"","code":"layer_dropout(object, rate, noise_shape = NULL, seed = NULL, ...)"},{"path":"https://keras3.posit.co/dev/reference/layer_dropout.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Applies dropout to the input. — layer_dropout","text":"object Object compose layer . tensor, array, sequential model. rate Float 0 1. Fraction input units drop. noise_shape 1D integer tensor representing shape binary dropout mask multiplied input. instance, inputs shape (batch_size, timesteps, features) want dropout mask timesteps, can use noise_shape=(batch_size, 1, features). seed R integer use random seed. ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_dropout.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Applies dropout to the input. — layer_dropout","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_dropout.html","id":"call-arguments","dir":"Reference","previous_headings":"","what":"Call Arguments","title":"Applies dropout to the input. — layer_dropout","text":"inputs: Input tensor (rank). training: Python boolean indicating whether layer behave training mode (adding dropout) inference mode (nothing).","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_einsum_dense.html","id":null,"dir":"Reference","previous_headings":"","what":"A layer that uses einsum as the backing computation. — layer_einsum_dense","title":"A layer that uses einsum as the backing computation. — layer_einsum_dense","text":"layer can perform einsum calculations arbitrary dimensionality.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_einsum_dense.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"A layer that uses einsum as the backing computation. — layer_einsum_dense","text":"","code":"layer_einsum_dense(   object,   equation,   output_shape,   activation = NULL,   bias_axes = NULL,   kernel_initializer = \"glorot_uniform\",   bias_initializer = \"zeros\",   kernel_regularizer = NULL,   bias_regularizer = NULL,   kernel_constraint = NULL,   bias_constraint = NULL,   lora_rank = NULL,   lora_alpha = NULL,   ... )"},{"path":"https://keras3.posit.co/dev/reference/layer_einsum_dense.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"A layer that uses einsum as the backing computation. — layer_einsum_dense","text":"object Object compose layer . tensor, array, sequential model. equation equation describing einsum perform. equation must valid einsum string form ab,bc->ac, ...ab,bc->...ac, ab...,bc->ac... 'ab', 'bc', 'ac' can valid einsum axis expression sequence. output_shape expected shape output tensor (excluding batch dimension dimensions represented ellipses). can specify NA NULL dimension unknown can inferred input shape. activation Activation function use. specify anything, activation applied (, \"linear\" activation: (x) = x). bias_axes string containing output dimension(s) apply bias . character bias_axes string correspond character output portion equation string. kernel_initializer Initializer kernel weights matrix. bias_initializer Initializer bias vector. kernel_regularizer Regularizer function applied kernel weights matrix. bias_regularizer Regularizer function applied bias vector. kernel_constraint Constraint function applied kernel weights matrix. bias_constraint Constraint function applied bias vector. lora_rank Optional integer. set, layer's forward pass implement LoRA (Low-Rank Adaptation) provided rank. LoRA sets layer's kernel non-trainable replaces delta original kernel, obtained via multiplying two lower-rank trainable matrices (factorization happens last dimension). can useful reduce computation cost fine-tuning large dense layers. can also enable LoRA existing EinsumDense layer calling layer$enable_lora(rank). lora_alpha Optional integer. Scales low-rank adaptation delta forward pass. delta scaled lora_alpha / lora_rank, letting tune LoRA adjustment strength independently lora_rank. ... Base layer keyword arguments, name dtype.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_einsum_dense.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"A layer that uses einsum as the backing computation. — layer_einsum_dense","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_einsum_dense.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"A layer that uses einsum as the backing computation. — layer_einsum_dense","text":"Biased dense layer einsums example shows instantiate standard Keras dense layer using einsum operations. example equivalent layer_Dense(64, use_bias=TRUE).     Applying dense layer sequence example shows instantiate layer applies dense operation every element sequence. , output_shape two values (since two non-batch dimensions output); first dimension output_shape NA, sequence dimension b unknown shape.     Applying dense layer sequence using ellipses example shows instantiate layer applies dense operation every element sequence, uses ellipsis notation instead specifying batch sequence dimensions. using ellipsis notation specified one axis, output_shape arg single value. instantiated way, layer can handle number sequence dimensions - including case sequence dimension exists.","code":"input <- layer_input(shape = c(32)) output <- input |>   layer_einsum_dense(\"ab,bc->ac\",                      output_shape = 64,                      bias_axes = \"c\") output # shape(NA, 64) ## <KerasTensor shape=(None, 64), dtype=float32, sparse=False, ragged=False, name=keras_tensor_1> input <- layer_input(shape = c(32, 128)) output <- input |>   layer_einsum_dense(\"abc,cd->abd\",                      output_shape = c(NA, 64),                      bias_axes = \"d\") output  # shape(NA, 32, 64) ## <KerasTensor shape=(None, None, 64), dtype=float32, sparse=False, ragged=False, name=keras_tensor_3> input <- layer_input(shape = c(32, 128)) output <- input |>   layer_einsum_dense(\"...x,xy->...y\",                      output_shape = 64,                      bias_axes = \"y\")  output  # shape(NA, 32, 64) ## <KerasTensor shape=(None, 32, 64), dtype=float32, sparse=False, ragged=False, name=keras_tensor_5>"},{"path":"https://keras3.posit.co/dev/reference/layer_einsum_dense.html","id":"methods","dir":"Reference","previous_headings":"","what":"Methods","title":"A layer that uses einsum as the backing computation. — layer_einsum_dense","text":"","code":"enable_lora(   rank,   a_initializer = 'he_uniform',   b_initializer = 'zeros' ) quantize(mode, type_check = TRUE)"},{"path":"https://keras3.posit.co/dev/reference/layer_einsum_dense.html","id":"readonly-properties-","dir":"Reference","previous_headings":"","what":"Readonly properties:","title":"A layer that uses einsum as the backing computation. — layer_einsum_dense","text":"kernel","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_embedding.html","id":null,"dir":"Reference","previous_headings":"","what":"Turns nonnegative integers (indexes) into dense vectors of fixed size. — layer_embedding","title":"Turns nonnegative integers (indexes) into dense vectors of fixed size. — layer_embedding","text":"e.g. rbind(4L, 20L) \\(\\rightarrow\\) rbind(c(0.25, 0.1), c(0.6, -0.2)) layer can used nonnegative integer inputs fixed range.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_embedding.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Turns nonnegative integers (indexes) into dense vectors of fixed size. — layer_embedding","text":"","code":"layer_embedding(   object,   input_dim,   output_dim,   embeddings_initializer = \"uniform\",   embeddings_regularizer = NULL,   embeddings_constraint = NULL,   mask_zero = FALSE,   weights = NULL,   lora_rank = NULL,   lora_alpha = NULL,   ... )"},{"path":"https://keras3.posit.co/dev/reference/layer_embedding.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Turns nonnegative integers (indexes) into dense vectors of fixed size. — layer_embedding","text":"object Object compose layer . tensor, array, sequential model. input_dim Integer. Size vocabulary, .e. maximum integer index + 1. output_dim Integer. Dimension dense embedding. embeddings_initializer Initializer embeddings matrix (see keras3::initializer_*). embeddings_regularizer Regularizer function applied embeddings matrix (see keras3::regularizer_*). embeddings_constraint Constraint function applied embeddings matrix (see keras3::constraint_*). mask_zero Boolean, whether input value 0 special \"padding\" value masked . useful using recurrent layers may take variable length input. TRUE, subsequent layers model need support masking exception raised. mask_zero set TRUE, consequence, index 0 used vocabulary (input_dim equal size vocabulary + 1). weights Optional floating-point matrix size (input_dim, output_dim). initial embeddings values use. lora_rank Optional integer. set, layer's forward pass implement LoRA (Low-Rank Adaptation) provided rank. LoRA sets layer's embeddings matrix non-trainable replaces delta original matrix, obtained via multiplying two lower-rank trainable matrices. can useful reduce computation cost fine-tuning large embedding layers. can also enable LoRA existing Embedding layer instance calling layer$enable_lora(rank). lora_alpha Optional integer. Scales low-rank adaptation delta forward pass. delta scaled lora_alpha / lora_rank, letting tune LoRA adjustment strength independently lora_rank. ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_embedding.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Turns nonnegative integers (indexes) into dense vectors of fixed size. — layer_embedding","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_embedding.html","id":"example","dir":"Reference","previous_headings":"","what":"Example","title":"Turns nonnegative integers (indexes) into dense vectors of fixed size. — layer_embedding","text":"","code":"model <- keras_model_sequential() |>   layer_embedding(1000, 64)  # The model will take as input an integer matrix of size (batch,input_length), # and the largest integer (i.e. word index) in the input # should be no larger than 999 (vocabulary size). # Now model$output_shape is (NA, 10, 64), where `NA` is the batch # dimension.  input_array <- random_integer(shape = c(32, 10), minval = 0, maxval = 1000) model |> compile('rmsprop', 'mse') output_array <- model |> predict(input_array, verbose = 0) dim(output_array)    # (32, 10, 64) ## [1] 32 10 64"},{"path":"https://keras3.posit.co/dev/reference/layer_embedding.html","id":"input-shape","dir":"Reference","previous_headings":"","what":"Input Shape","title":"Turns nonnegative integers (indexes) into dense vectors of fixed size. — layer_embedding","text":"2D tensor shape: (batch_size, input_length).","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_embedding.html","id":"output-shape","dir":"Reference","previous_headings":"","what":"Output Shape","title":"Turns nonnegative integers (indexes) into dense vectors of fixed size. — layer_embedding","text":"3D tensor shape: (batch_size, input_length, output_dim).","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_embedding.html","id":"methods","dir":"Reference","previous_headings":"","what":"Methods","title":"Turns nonnegative integers (indexes) into dense vectors of fixed size. — layer_embedding","text":"","code":"enable_lora(   rank,   lora_alpha = NULL,   a_initializer = 'he_uniform',   b_initializer = 'zeros' ) compute_output_spec(...) quantize(mode, type_check = TRUE) quantized_build(input_shape, mode) quantized_call(...)"},{"path":"https://keras3.posit.co/dev/reference/layer_embedding.html","id":"readonly-properties-","dir":"Reference","previous_headings":"","what":"Readonly properties:","title":"Turns nonnegative integers (indexes) into dense vectors of fixed size. — layer_embedding","text":"embeddings","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_equalization.html","id":null,"dir":"Reference","previous_headings":"","what":"Preprocessing layer for histogram equalization on image channels. — layer_equalization","title":"Preprocessing layer for histogram equalization on image channels. — layer_equalization","text":"Histogram equalization technique adjust image intensities enhance contrast effectively spreading frequent intensity values. layer applies equalization channel-wise basis, can improve visibility details images. layer works grayscale color images, performing equalization independently color channel. inference time, equalization consistently applied. Note: layer safe use inside tf.data pipeline (independently backend using).","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_equalization.html","id":"input-shape","dir":"Reference","previous_headings":"","what":"Input Shape","title":"Preprocessing layer for histogram equalization on image channels. — layer_equalization","text":"3D (unbatched) 4D (batched) tensor shape: (..., height, width, channels), \"channels_last\" format, (..., channels, height, width), \"channels_first\" format.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_equalization.html","id":"output-shape","dir":"Reference","previous_headings":"","what":"Output Shape","title":"Preprocessing layer for histogram equalization on image channels. — layer_equalization","text":"3D (unbatched) 4D (batched) tensor shape: (..., target_height, target_width, channels), (..., channels, target_height, target_width), \"channels_first\" format.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_equalization.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Preprocessing layer for histogram equalization on image channels. — layer_equalization","text":"","code":"layer_equalization(   object,   value_range = list(0L, 255L),   bins = 256L,   data_format = NULL,   ... )"},{"path":"https://keras3.posit.co/dev/reference/layer_equalization.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Preprocessing layer for histogram equalization on image channels. — layer_equalization","text":"object Object compose layer . tensor, array, sequential model. value_range Optional list/tuple 2 floats specifying lower upper limits input data values. Defaults [0, 255]. input image scaled, use appropriate range (e.g., [0.0, 1.0]). equalization scaled range, output values clipped accordingly. bins Integer specifying number histogram bins use equalization. Defaults 256, suitable 8-bit images. Larger values can provide granular intensity redistribution. data_format String, one \"channels_last\" (default) \"channels_first\". ordering dimensions inputs. \"channels_last\" corresponds inputs shape (batch, height, width, channels) \"channels_first\" corresponds inputs shape (batch, channels, height, width). ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_equalization.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Preprocessing layer for histogram equalization on image channels. — layer_equalization","text":"","code":"# Create an equalization layer for standard 8-bit images equalizer <- layer_equalization()  # An image with uneven intensity distribution image <- np_array(...) # your input image  # Apply histogram equalization equalized_image <- equalizer(image)  # For images with custom value range custom_equalizer <- layer_equalization(     value_range=c(0.0, 1.0),  # for normalized images     bins=128  # fewer bins for more subtle equalization ) custom_equalized <- custom_equalizer(normalized_image)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_feature_space.html","id":null,"dir":"Reference","previous_headings":"","what":"One-stop utility for preprocessing and encoding structured data. — layer_feature_space","title":"One-stop utility for preprocessing and encoding structured data. — layer_feature_space","text":"Available feature types: Note features can referred string name, e.g. \"integer_categorical\". using string name, default argument values used.","code":"# Plain float values. feature_float(name = NULL)  # Float values to be preprocessed via featurewise standardization # (i.e. via a `layer_normalization()` layer). feature_float_normalized(name = NULL)  # Float values to be preprocessed via linear rescaling # (i.e. via a `layer_rescaling` layer). feature_float_rescaled(scale = 1., offset = 0., name = NULL)  # Float values to be discretized. By default, the discrete # representation will then be one-hot encoded. feature_float_discretized(   num_bins,   bin_boundaries = NULL,   output_mode = \"one_hot\",   name = NULL )  # Integer values to be indexed. By default, the discrete # representation will then be one-hot encoded. feature_integer_categorical(   max_tokens = NULL,   num_oov_indices = 1,   output_mode = \"one_hot\",   name = NULL )  # String values to be indexed. By default, the discrete # representation will then be one-hot encoded. feature_string_categorical(   max_tokens = NULL,   num_oov_indices = 1,   output_mode = \"one_hot\",   name = NULL )  # Integer values to be hashed into a fixed number of bins. # By default, the discrete representation will then be one-hot encoded. feature_integer_hashed(num_bins, output_mode = \"one_hot\", name = NULL)  # String values to be hashed into a fixed number of bins. # By default, the discrete representation will then be one-hot encoded. feature_string_hashed(num_bins, output_mode = \"one_hot\", name = NULL)"},{"path":"https://keras3.posit.co/dev/reference/layer_feature_space.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"One-stop utility for preprocessing and encoding structured data. — layer_feature_space","text":"","code":"layer_feature_space(   object,   features,   output_mode = \"concat\",   crosses = NULL,   crossing_dim = 32L,   hashing_dim = 32L,   num_discretization_bins = 32L,   name = NULL,   feature_names = NULL )  feature_cross(feature_names, crossing_dim, output_mode = \"one_hot\")  feature_custom(dtype, preprocessor, output_mode)  feature_float(name = NULL)  feature_float_rescaled(scale = 1, offset = 0, name = NULL)  feature_float_normalized(name = NULL)  feature_float_discretized(   num_bins,   bin_boundaries = NULL,   output_mode = \"one_hot\",   name = NULL )  feature_integer_categorical(   max_tokens = NULL,   num_oov_indices = 1,   output_mode = \"one_hot\",   name = NULL )  feature_string_categorical(   max_tokens = NULL,   num_oov_indices = 1,   output_mode = \"one_hot\",   name = NULL )  feature_string_hashed(num_bins, output_mode = \"one_hot\", name = NULL)  feature_integer_hashed(num_bins, output_mode = \"one_hot\", name = NULL)"},{"path":"https://keras3.posit.co/dev/reference/layer_feature_space.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"One-stop utility for preprocessing and encoding structured data. — layer_feature_space","text":"object see description features see description output_mode string. layer_feature_space(), one \"concat\" \"dict\". concat mode, features get concatenated together single vector. dict mode, FeatureSpace returns named list individually encoded features (names input list names). feature_* functions, one : \"int\" \"one_hot\" \"float\". crosses List features crossed together, e.g. crosses=list(c(\"feature_1\", \"feature_2\")). features \"crossed\" hashing combined value fixed-length vector. crossing_dim Default vector size hashing crossed features. Defaults 32. hashing_dim Default vector size hashing features type \"integer_hashed\" \"string_hashed\". Defaults 32. num_discretization_bins Default number bins used discretizing features type \"float_discretized\". Defaults 32. name String, name object feature_names Named list mapping names features type specification, e.g. list(my_feature = \"integer_categorical\") list(my_feature = feature_integer_categorical()). complete list supported types, see \"Available feature types\" paragraph . dtype string, output dtype feature. E.g., \"float32\". preprocessor callable. scale, offset Passed layer_rescaling() num_bins, bin_boundaries Passed layer_discretization() max_tokens, num_oov_indices Passed layer_integer_lookup() feature_integer_categorical() layer_string_lookup() feature_string_categorical().","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_feature_space.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"One-stop utility for preprocessing and encoding structured data. — layer_feature_space","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_feature_space.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"One-stop utility for preprocessing and encoding structured data. — layer_feature_space","text":"Basic usage named list input data:   Basic usage tf.data:   Basic usage Keras Functional API:           Customizing feature feature cross:   Returning dict (named list) integer-encoded features:   Specifying Keras preprocessing layer:   Retrieving underlying Keras preprocessing layers:   Saving reloading FeatureSpace:","code":"raw_data <- list(   float_values = c(0.0, 0.1, 0.2, 0.3),   string_values = c(\"zero\", \"one\", \"two\", \"three\"),   int_values = as.integer(c(0, 1, 2, 3)) )  dataset <- tfdatasets::tensor_slices_dataset(raw_data)  feature_space <- layer_feature_space(   features = list(     float_values = \"float_normalized\",     string_values = \"string_categorical\",     int_values = \"integer_categorical\"   ),   crosses = list(c(\"string_values\", \"int_values\")),   output_mode = \"concat\" )  # Before you start using the feature_space(), # you must `adapt()` it on some data. feature_space |> adapt(dataset)  # You can call the feature_space() on a named list of # data (batched or unbatched). output_vector <- feature_space(raw_data) library(tfdatasets) # Unlabeled data preprocessed_ds <- unlabeled_dataset |>   dataset_map(feature_space)  # Labeled data preprocessed_ds <- labeled_dataset |>   dataset_map(function(x, y) tuple(feature_space(x), y)) # Retrieve a named list of Keras layer_input() objects (inputs <- feature_space$get_inputs()) ## $float_values ## <KerasTensor shape=(None, 1), dtype=float32, sparse=False, ragged=False, name=float_values> ## ## $string_values ## <KerasTensor shape=(None, 1), dtype=string, sparse=False, ragged=False, name=string_values> ## ## $int_values ## <KerasTensor shape=(None, 1), dtype=int32, sparse=False, ragged=False, name=int_values> # Retrieve the corresponding encoded Keras tensors (encoded_features <- feature_space$get_encoded_features()) ## <KerasTensor shape=(None, 43), dtype=float32, sparse=False, ragged=False, name=keras_tensor_7> # Build a Functional model outputs <- encoded_features |> layer_dense(1, activation = \"sigmoid\") model <- keras_model(inputs, outputs) feature_space <- layer_feature_space(   features = list(     float_values = feature_float_normalized(),     string_values = feature_string_categorical(max_tokens = 10),     int_values = feature_integer_categorical(max_tokens = 10)   ),   crosses = list(     feature_cross(c(\"string_values\", \"int_values\"), crossing_dim = 32)   ),   output_mode = \"concat\" ) feature_space <- layer_feature_space(   features = list(     \"string_values\" = feature_string_categorical(output_mode = \"int\"),     \"int_values\" = feature_integer_categorical(output_mode = \"int\")   ),   crosses = list(     feature_cross(       feature_names = c(\"string_values\", \"int_values\"),       crossing_dim = 32,       output_mode = \"int\"     )   ),   output_mode = \"dict\" ) # Let's say that one of the features is a short text paragraph that # we want to encode as a vector (one vector per paragraph) via TF-IDF. data <- list(text = c(\"1st string\", \"2nd string\", \"3rd string\"))  # There's a Keras layer for this: layer_text_vectorization() custom_layer <- layer_text_vectorization(output_mode = \"tf_idf\")  # We can use feature_custom() to create a custom feature # that will use our preprocessing layer. feature_space <- layer_feature_space(   features = list(     text = feature_custom(preprocessor = custom_layer,                           dtype = \"string\",                           output_mode = \"float\"     )   ),   output_mode = \"concat\" ) feature_space |> adapt(tfdatasets::tensor_slices_dataset(data)) output_vector <- feature_space(data) # The preprocessing layer of each feature is available in `$preprocessors`. preprocessing_layer <- feature_space$preprocessors$feature1  # The crossing layer of each feature cross is available in `$crossers`. # It's an instance of layer_hashed_crossing() crossing_layer <- feature_space$crossers[[\"feature1_X_feature2\"]] feature_space$save(\"featurespace.keras\") reloaded_feature_space <- keras$models$load_model(\"featurespace.keras\")"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_flatten.html","id":null,"dir":"Reference","previous_headings":"","what":"Flattens the input. Does not affect the batch size. — layer_flatten","title":"Flattens the input. Does not affect the batch size. — layer_flatten","text":"Flattens input. affect batch size.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_flatten.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Flattens the input. Does not affect the batch size. — layer_flatten","text":"","code":"layer_flatten(object, data_format = NULL, ...)"},{"path":"https://keras3.posit.co/dev/reference/layer_flatten.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Flattens the input. Does not affect the batch size. — layer_flatten","text":"object Object compose layer . tensor, array, sequential model. data_format string, one \"channels_last\" (default) \"channels_first\". ordering dimensions inputs. \"channels_last\" corresponds inputs shape (batch, ..., channels) \"channels_first\" corresponds inputs shape (batch, channels, ...). unspecified, uses image_data_format value found Keras config file ~/.keras/keras.json (exists). Defaults \"channels_last\". ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_flatten.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Flattens the input. Does not affect the batch size. — layer_flatten","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_flatten.html","id":"note","dir":"Reference","previous_headings":"","what":"Note","title":"Flattens the input. Does not affect the batch size. — layer_flatten","text":"inputs shaped (batch) without feature axis, flattening adds extra channel dimension output shape (batch, 1).","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_flatten.html","id":"example","dir":"Reference","previous_headings":"","what":"Example","title":"Flattens the input. Does not affect the batch size. — layer_flatten","text":"","code":"x <- layer_input(shape=c(10, 64)) y <- x |> layer_flatten() shape(y) ## shape(NA, 640)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_flax_module_wrapper.html","id":null,"dir":"Reference","previous_headings":"","what":"Keras Layer that wraps a Flax module. — layer_flax_module_wrapper","title":"Keras Layer that wraps a Flax module. — layer_flax_module_wrapper","text":"layer enables use Flax components form flax.linen.Module instances within Keras using JAX backend Keras. module method use forward pass can specified via method argument __call__ default. method must take following arguments exact names: self method bound module, case default __call__, module otherwise pass module. inputs: inputs model, JAX array PyTree arrays. training (optional): argument specifying training mode inference mode, TRUE passed training mode. FlaxLayer handles non-trainable state model required RNGs automatically. Note mutable parameter flax.linen.Module.apply() set DenyList([\"params\"]), therefore making assumption variables outside \"params\" collection non-trainable weights. example shows create FlaxLayer Flax Module default __call__ method training argument:   example shows wrap module method conform required signature. allows multiple input arguments training argument different name values. additionally shows use function bound module.","code":"# keras3::use_backend(\"jax\") # py_install(\"flax\", \"r-keras\")  if(config_backend() == \"jax\" &&    reticulate::py_module_available(\"flax\")) {  flax <- import(\"flax\")  MyFlaxModule(flax$linen$Module) %py_class% {   `__call__` <- flax$linen$compact(\\(self, inputs) {     inputs |>       (flax$linen$Conv(features = 32L, kernel_size = tuple(3L, 3L)))() |>       flax$linen$relu() |>       flax$linen$avg_pool(window_shape = tuple(2L, 2L),                           strides = tuple(2L, 2L)) |>       # flatten all except batch_size axis       (\\(x) x$reshape(tuple(x$shape[[1]], -1L)))() |>       (flax$linen$Dense(features = 200L))() |>       flax$linen$relu() |>       (flax$linen$Dense(features = 10L))() |>       flax$linen$softmax()   }) }  # typical usage: input <- keras_input(c(28, 28, 3)) output <- input |>   layer_flax_module_wrapper(MyFlaxModule())  model <- keras_model(input, output)  # to instantiate the layer before composing: flax_module <- MyFlaxModule() keras_layer <- layer_flax_module_wrapper(module = flax_module)  input <- keras_input(c(28, 28, 3)) output <- input |>   keras_layer()  model <- keras_model(input, output)  } flax <- import(\"flax\")  MyFlaxModule(flax$linen$Module) \\%py_class\\% {   forward <-     flax$linen$compact(\\(self, inputs1, input2, deterministic) {       # do work ....       outputs # return     }) }  my_flax_module_wrapper <- function(module, inputs, training) {   c(input1, input2) \\%<-\\% inputs   module$forward(input1, input2,!training) }  flax_module <- MyFlaxModule() keras_layer <- layer_flax_module_wrapper(module = flax_module,                                          method = my_flax_module_wrapper)"},{"path":"https://keras3.posit.co/dev/reference/layer_flax_module_wrapper.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Keras Layer that wraps a Flax module. — layer_flax_module_wrapper","text":"","code":"layer_flax_module_wrapper(object, module, method = NULL, variables = NULL, ...)"},{"path":"https://keras3.posit.co/dev/reference/layer_flax_module_wrapper.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Keras Layer that wraps a Flax module. — layer_flax_module_wrapper","text":"object Object compose layer . tensor, array, sequential model. module instance flax.linen.Module subclass. method method call model. generally method Module. provided, __call__ method used. method can also function defined Module, case must take Module first argument. used Module.init Module.apply. Details documented method argument flax.linen.Module.apply(). variables dict (named R list) containing variables module format returned flax.linen.Module.init(). contain \"params\" key , applicable, keys collections variables non-trainable state. allows passing trained parameters learned non-trainable state controlling initialization. NULL passed, module's init function called build time initialize variables model. ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_flax_module_wrapper.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Keras Layer that wraps a Flax module. — layer_flax_module_wrapper","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_gaussian_dropout.html","id":null,"dir":"Reference","previous_headings":"","what":"Apply multiplicative 1-centered Gaussian noise. — layer_gaussian_dropout","title":"Apply multiplicative 1-centered Gaussian noise. — layer_gaussian_dropout","text":"regularization layer, active training time.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_gaussian_dropout.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Apply multiplicative 1-centered Gaussian noise. — layer_gaussian_dropout","text":"","code":"layer_gaussian_dropout(object, rate, seed = NULL, ...)"},{"path":"https://keras3.posit.co/dev/reference/layer_gaussian_dropout.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Apply multiplicative 1-centered Gaussian noise. — layer_gaussian_dropout","text":"object Object compose layer . tensor, array, sequential model. rate Float, drop probability (Dropout). multiplicative noise standard deviation sqrt(rate / (1 - rate)). seed Integer, optional random seed enable deterministic behavior. ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_gaussian_dropout.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Apply multiplicative 1-centered Gaussian noise. — layer_gaussian_dropout","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_gaussian_dropout.html","id":"call-arguments","dir":"Reference","previous_headings":"","what":"Call Arguments","title":"Apply multiplicative 1-centered Gaussian noise. — layer_gaussian_dropout","text":"inputs: Input tensor (rank). training: Python boolean indicating whether layer behave training mode (adding dropout) inference mode (nothing).","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_gaussian_noise.html","id":null,"dir":"Reference","previous_headings":"","what":"Apply additive zero-centered Gaussian noise. — layer_gaussian_noise","title":"Apply additive zero-centered Gaussian noise. — layer_gaussian_noise","text":"useful mitigate overfitting (see form random data augmentation). Gaussian Noise (GS) natural choice corruption process real valued inputs. regularization layer, active training time.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_gaussian_noise.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Apply additive zero-centered Gaussian noise. — layer_gaussian_noise","text":"","code":"layer_gaussian_noise(object, stddev, seed = NULL, ...)"},{"path":"https://keras3.posit.co/dev/reference/layer_gaussian_noise.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Apply additive zero-centered Gaussian noise. — layer_gaussian_noise","text":"object Object compose layer . tensor, array, sequential model. stddev Float, standard deviation noise distribution. seed Integer, optional random seed enable deterministic behavior. ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_gaussian_noise.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Apply additive zero-centered Gaussian noise. — layer_gaussian_noise","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_gaussian_noise.html","id":"call-arguments","dir":"Reference","previous_headings":"","what":"Call Arguments","title":"Apply additive zero-centered Gaussian noise. — layer_gaussian_noise","text":"inputs: Input tensor (rank). training: Python boolean indicating whether layer behave training mode (adding noise) inference mode (nothing).","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_global_average_pooling_1d.html","id":null,"dir":"Reference","previous_headings":"","what":"Global average pooling operation for temporal data. — layer_global_average_pooling_1d","title":"Global average pooling operation for temporal data. — layer_global_average_pooling_1d","text":"Global average pooling operation temporal data.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_global_average_pooling_1d.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Global average pooling operation for temporal data. — layer_global_average_pooling_1d","text":"","code":"layer_global_average_pooling_1d(   object,   data_format = NULL,   keepdims = FALSE,   ... )"},{"path":"https://keras3.posit.co/dev/reference/layer_global_average_pooling_1d.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Global average pooling operation for temporal data. — layer_global_average_pooling_1d","text":"object Object compose layer . tensor, array, sequential model. data_format string, either \"channels_last\" \"channels_first\". ordering dimensions inputs. \"channels_last\" corresponds inputs shape (batch, steps, features) \"channels_first\" corresponds inputs shape (batch, features, steps). defaults image_data_format value found Keras config file ~/.keras/keras.json. never set , \"channels_last\". keepdims boolean, whether keep temporal dimension . keepdims FALSE (default), rank tensor reduced spatial dimensions. keepdims TRUE, temporal dimension retained length 1. behavior tf$reduce_mean() op_mean(). ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_global_average_pooling_1d.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Global average pooling operation for temporal data. — layer_global_average_pooling_1d","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_global_average_pooling_1d.html","id":"call-arguments","dir":"Reference","previous_headings":"","what":"Call Arguments","title":"Global average pooling operation for temporal data. — layer_global_average_pooling_1d","text":"inputs: 3D tensor. mask: Binary tensor shape (batch_size, steps) indicating whether given step masked (excluded average).","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_global_average_pooling_1d.html","id":"input-shape","dir":"Reference","previous_headings":"","what":"Input Shape","title":"Global average pooling operation for temporal data. — layer_global_average_pooling_1d","text":"data_format='channels_last': 3D tensor shape: (batch_size, steps, features) data_format='channels_first': 3D tensor shape: (batch_size, features, steps)","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_global_average_pooling_1d.html","id":"output-shape","dir":"Reference","previous_headings":"","what":"Output Shape","title":"Global average pooling operation for temporal data. — layer_global_average_pooling_1d","text":"keepdims=FALSE: 2D tensor shape (batch_size, features). keepdims=TRUE: data_format=\"channels_last\": 3D tensor shape (batch_size, 1, features) data_format=\"channels_first\": 3D tensor shape (batch_size, features, 1)","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_global_average_pooling_1d.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Global average pooling operation for temporal data. — layer_global_average_pooling_1d","text":"","code":"x <- random_uniform(c(2, 3, 4)) y <- x |> layer_global_average_pooling_1d() shape(y) ## shape(2, 4)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_global_average_pooling_2d.html","id":null,"dir":"Reference","previous_headings":"","what":"Global average pooling operation for 2D data. — layer_global_average_pooling_2d","title":"Global average pooling operation for 2D data. — layer_global_average_pooling_2d","text":"Global average pooling operation 2D data.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_global_average_pooling_2d.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Global average pooling operation for 2D data. — layer_global_average_pooling_2d","text":"","code":"layer_global_average_pooling_2d(   object,   data_format = NULL,   keepdims = FALSE,   ... )"},{"path":"https://keras3.posit.co/dev/reference/layer_global_average_pooling_2d.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Global average pooling operation for 2D data. — layer_global_average_pooling_2d","text":"object Object compose layer . tensor, array, sequential model. data_format string, either \"channels_last\" \"channels_first\". ordering dimensions inputs. \"channels_last\" corresponds inputs shape (batch, height, width, channels) \"channels_first\" corresponds inputs shape (batch, features, height, weight). defaults image_data_format value found Keras config file ~/.keras/keras.json. never set , \"channels_last\". keepdims boolean, whether keep temporal dimension . keepdims FALSE (default), rank tensor reduced spatial dimensions. keepdims TRUE, spatial dimension retained length 1. behavior tf$reduce_mean() op_mean(). ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_global_average_pooling_2d.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Global average pooling operation for 2D data. — layer_global_average_pooling_2d","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_global_average_pooling_2d.html","id":"input-shape","dir":"Reference","previous_headings":"","what":"Input Shape","title":"Global average pooling operation for 2D data. — layer_global_average_pooling_2d","text":"data_format='channels_last': 4D tensor shape: (batch_size, height, width, channels) data_format='channels_first': 4D tensor shape: (batch_size, channels, height, width)","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_global_average_pooling_2d.html","id":"output-shape","dir":"Reference","previous_headings":"","what":"Output Shape","title":"Global average pooling operation for 2D data. — layer_global_average_pooling_2d","text":"keepdims=FALSE: 2D tensor shape (batch_size, channels). keepdims=TRUE: data_format=\"channels_last\": 4D tensor shape (batch_size, 1, 1, channels) data_format=\"channels_first\": 4D tensor shape (batch_size, channels, 1, 1)","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_global_average_pooling_2d.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Global average pooling operation for 2D data. — layer_global_average_pooling_2d","text":"","code":"x <- random_uniform(c(2, 4, 5, 3)) y <- x |> layer_global_average_pooling_2d() shape(y) ## shape(2, 3)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_global_average_pooling_3d.html","id":null,"dir":"Reference","previous_headings":"","what":"Global average pooling operation for 3D data. — layer_global_average_pooling_3d","title":"Global average pooling operation for 3D data. — layer_global_average_pooling_3d","text":"Global average pooling operation 3D data.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_global_average_pooling_3d.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Global average pooling operation for 3D data. — layer_global_average_pooling_3d","text":"","code":"layer_global_average_pooling_3d(   object,   data_format = NULL,   keepdims = FALSE,   ... )"},{"path":"https://keras3.posit.co/dev/reference/layer_global_average_pooling_3d.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Global average pooling operation for 3D data. — layer_global_average_pooling_3d","text":"object Object compose layer . tensor, array, sequential model. data_format string, either \"channels_last\" \"channels_first\". ordering dimensions inputs. \"channels_last\" corresponds inputs shape (batch, spatial_dim1, spatial_dim2, spatial_dim3, channels) \"channels_first\" corresponds inputs shape (batch, channels, spatial_dim1, spatial_dim2, spatial_dim3). defaults image_data_format value found Keras config file ~/.keras/keras.json. never set , \"channels_last\". keepdims boolean, whether keep temporal dimension . keepdims FALSE (default), rank tensor reduced spatial dimensions. keepdims TRUE, spatial dimension retained length 1. behavior tf$reduce_mean() op_mean(). ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_global_average_pooling_3d.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Global average pooling operation for 3D data. — layer_global_average_pooling_3d","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_global_average_pooling_3d.html","id":"input-shape","dir":"Reference","previous_headings":"","what":"Input Shape","title":"Global average pooling operation for 3D data. — layer_global_average_pooling_3d","text":"data_format='channels_last': 5D tensor shape: (batch_size, spatial_dim1, spatial_dim2, spatial_dim3, channels) data_format='channels_first': 5D tensor shape: (batch_size, channels, spatial_dim1, spatial_dim2, spatial_dim3)","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_global_average_pooling_3d.html","id":"output-shape","dir":"Reference","previous_headings":"","what":"Output Shape","title":"Global average pooling operation for 3D data. — layer_global_average_pooling_3d","text":"keepdims=FALSE: 2D tensor shape (batch_size, channels). keepdims=TRUE: data_format=\"channels_last\": 5D tensor shape (batch_size, 1, 1, 1, channels) data_format=\"channels_first\": 5D tensor shape (batch_size, channels, 1, 1, 1)","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_global_average_pooling_3d.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Global average pooling operation for 3D data. — layer_global_average_pooling_3d","text":"","code":"x <- random_uniform(c(2, 4, 5, 4, 3)) y <- x |> layer_global_average_pooling_3d() shape(y) ## shape(2, 3)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_global_max_pooling_1d.html","id":null,"dir":"Reference","previous_headings":"","what":"Global max pooling operation for temporal data. — layer_global_max_pooling_1d","title":"Global max pooling operation for temporal data. — layer_global_max_pooling_1d","text":"Global max pooling operation temporal data.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_global_max_pooling_1d.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Global max pooling operation for temporal data. — layer_global_max_pooling_1d","text":"","code":"layer_global_max_pooling_1d(object, data_format = NULL, keepdims = FALSE, ...)"},{"path":"https://keras3.posit.co/dev/reference/layer_global_max_pooling_1d.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Global max pooling operation for temporal data. — layer_global_max_pooling_1d","text":"object Object compose layer . tensor, array, sequential model. data_format string, either \"channels_last\" \"channels_first\". ordering dimensions inputs. \"channels_last\" corresponds inputs shape (batch, steps, features) \"channels_first\" corresponds inputs shape (batch, features, steps). defaults image_data_format value found Keras config file ~/.keras/keras.json. never set , \"channels_last\". keepdims boolean, whether keep temporal dimension . keepdims FALSE (default), rank tensor reduced spatial dimensions. keepdims TRUE, temporal dimension retained length 1. behavior tf$reduce_mean() op_mean(). ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_global_max_pooling_1d.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Global max pooling operation for temporal data. — layer_global_max_pooling_1d","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_global_max_pooling_1d.html","id":"input-shape","dir":"Reference","previous_headings":"","what":"Input Shape","title":"Global max pooling operation for temporal data. — layer_global_max_pooling_1d","text":"data_format='channels_last': 3D tensor shape: (batch_size, steps, features) data_format='channels_first': 3D tensor shape: (batch_size, features, steps)","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_global_max_pooling_1d.html","id":"output-shape","dir":"Reference","previous_headings":"","what":"Output Shape","title":"Global max pooling operation for temporal data. — layer_global_max_pooling_1d","text":"keepdims=FALSE: 2D tensor shape (batch_size, features). keepdims=TRUE: data_format=\"channels_last\": 3D tensor shape (batch_size, 1, features) data_format=\"channels_first\": 3D tensor shape (batch_size, features, 1)","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_global_max_pooling_1d.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Global max pooling operation for temporal data. — layer_global_max_pooling_1d","text":"","code":"x <- random_uniform(c(2, 3, 4)) y <- x |> layer_global_max_pooling_1d() shape(y) ## shape(2, 4)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_global_max_pooling_2d.html","id":null,"dir":"Reference","previous_headings":"","what":"Global max pooling operation for 2D data. — layer_global_max_pooling_2d","title":"Global max pooling operation for 2D data. — layer_global_max_pooling_2d","text":"Global max pooling operation 2D data.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_global_max_pooling_2d.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Global max pooling operation for 2D data. — layer_global_max_pooling_2d","text":"","code":"layer_global_max_pooling_2d(object, data_format = NULL, keepdims = FALSE, ...)"},{"path":"https://keras3.posit.co/dev/reference/layer_global_max_pooling_2d.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Global max pooling operation for 2D data. — layer_global_max_pooling_2d","text":"object Object compose layer . tensor, array, sequential model. data_format string, either \"channels_last\" \"channels_first\". ordering dimensions inputs. \"channels_last\" corresponds inputs shape (batch, height, width, channels) \"channels_first\" corresponds inputs shape (batch, features, height, weight). defaults image_data_format value found Keras config file ~/.keras/keras.json. never set , \"channels_last\". keepdims boolean, whether keep temporal dimension . keepdims FALSE (default), rank tensor reduced spatial dimensions. keepdims TRUE, spatial dimension retained length 1. behavior tf$reduce_mean() op_mean(). ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_global_max_pooling_2d.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Global max pooling operation for 2D data. — layer_global_max_pooling_2d","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_global_max_pooling_2d.html","id":"input-shape","dir":"Reference","previous_headings":"","what":"Input Shape","title":"Global max pooling operation for 2D data. — layer_global_max_pooling_2d","text":"data_format='channels_last': 4D tensor shape: (batch_size, height, width, channels) data_format='channels_first': 4D tensor shape: (batch_size, channels, height, width)","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_global_max_pooling_2d.html","id":"output-shape","dir":"Reference","previous_headings":"","what":"Output Shape","title":"Global max pooling operation for 2D data. — layer_global_max_pooling_2d","text":"keepdims=FALSE: 2D tensor shape (batch_size, channels). keepdims=TRUE: data_format=\"channels_last\": 4D tensor shape (batch_size, 1, 1, channels) data_format=\"channels_first\": 4D tensor shape (batch_size, channels, 1, 1)","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_global_max_pooling_2d.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Global max pooling operation for 2D data. — layer_global_max_pooling_2d","text":"","code":"x <- random_uniform(c(2, 4, 5, 3)) y <- x |> layer_global_max_pooling_2d() shape(y) ## shape(2, 3)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_global_max_pooling_3d.html","id":null,"dir":"Reference","previous_headings":"","what":"Global max pooling operation for 3D data. — layer_global_max_pooling_3d","title":"Global max pooling operation for 3D data. — layer_global_max_pooling_3d","text":"Global max pooling operation 3D data.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_global_max_pooling_3d.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Global max pooling operation for 3D data. — layer_global_max_pooling_3d","text":"","code":"layer_global_max_pooling_3d(object, data_format = NULL, keepdims = FALSE, ...)"},{"path":"https://keras3.posit.co/dev/reference/layer_global_max_pooling_3d.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Global max pooling operation for 3D data. — layer_global_max_pooling_3d","text":"object Object compose layer . tensor, array, sequential model. data_format string, either \"channels_last\" \"channels_first\". ordering dimensions inputs. \"channels_last\" corresponds inputs shape (batch, spatial_dim1, spatial_dim2, spatial_dim3, channels) \"channels_first\" corresponds inputs shape (batch, channels, spatial_dim1, spatial_dim2, spatial_dim3). defaults image_data_format value found Keras config file ~/.keras/keras.json. never set , \"channels_last\". keepdims boolean, whether keep temporal dimension . keepdims FALSE (default), rank tensor reduced spatial dimensions. keepdims TRUE, spatial dimension retained length 1. behavior tf$reduce_mean() op_mean(). ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_global_max_pooling_3d.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Global max pooling operation for 3D data. — layer_global_max_pooling_3d","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_global_max_pooling_3d.html","id":"input-shape","dir":"Reference","previous_headings":"","what":"Input Shape","title":"Global max pooling operation for 3D data. — layer_global_max_pooling_3d","text":"data_format='channels_last': 5D tensor shape: (batch_size, spatial_dim1, spatial_dim2, spatial_dim3, channels) data_format='channels_first': 5D tensor shape: (batch_size, channels, spatial_dim1, spatial_dim2, spatial_dim3)","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_global_max_pooling_3d.html","id":"output-shape","dir":"Reference","previous_headings":"","what":"Output Shape","title":"Global max pooling operation for 3D data. — layer_global_max_pooling_3d","text":"keepdims=FALSE: 2D tensor shape (batch_size, channels). keepdims=TRUE: data_format=\"channels_last\": 5D tensor shape (batch_size, 1, 1, 1, channels) data_format=\"channels_first\": 5D tensor shape (batch_size, channels, 1, 1, 1)","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_global_max_pooling_3d.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Global max pooling operation for 3D data. — layer_global_max_pooling_3d","text":"","code":"x <- random_uniform(c(2, 4, 5, 4, 3)) y <- x |> layer_global_max_pooling_3d() shape(y) ## shape(2, 3)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_group_normalization.html","id":null,"dir":"Reference","previous_headings":"","what":"Group normalization layer. — layer_group_normalization","title":"Group normalization layer. — layer_group_normalization","text":"Group Normalization divides channels groups computes within group mean variance normalization. Empirically, accuracy stable batch norm wide range small batch sizes, learning rate adjusted linearly batch sizes. Relation Layer Normalization: number groups set 1, operation becomes nearly identical Layer Normalization (see Layer Normalization docs details). Relation Instance Normalization: number groups set input dimension (number groups equal number channels), operation becomes identical Instance Normalization. can achieve via groups=-1.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_group_normalization.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Group normalization layer. — layer_group_normalization","text":"","code":"layer_group_normalization(   object,   groups = 32L,   axis = -1L,   epsilon = 0.001,   center = TRUE,   scale = TRUE,   beta_initializer = \"zeros\",   gamma_initializer = \"ones\",   beta_regularizer = NULL,   gamma_regularizer = NULL,   beta_constraint = NULL,   gamma_constraint = NULL,   ... )"},{"path":"https://keras3.posit.co/dev/reference/layer_group_normalization.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Group normalization layer. — layer_group_normalization","text":"object Object compose layer . tensor, array, sequential model. groups Integer, number groups Group Normalization. Can range [1, N] N input dimension. input dimension must divisible number groups. Defaults 32. axis Integer List/Tuple. axis axes normalize across. Typically, features axis/axes. left-axes typically batch axis/axes. -1 last dimension input. Defaults -1. epsilon Small float added variance avoid dividing zero. Defaults 1e-3. center TRUE, add offset beta normalized tensor. FALSE, beta ignored. Defaults TRUE. scale TRUE, multiply gamma. FALSE, gamma used. next layer linear (also e.g. relu), can disabled since scaling done next layer. Defaults TRUE. beta_initializer Initializer beta weight. Defaults zeros. gamma_initializer Initializer gamma weight. Defaults ones. beta_regularizer Optional regularizer beta weight. NULL default. gamma_regularizer Optional regularizer gamma weight. NULL default. beta_constraint Optional constraint beta weight. NULL default. gamma_constraint Optional constraint gamma weight. NULL default. ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_group_normalization.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Group normalization layer. — layer_group_normalization","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_group_normalization.html","id":"input-shape","dir":"Reference","previous_headings":"","what":"Input Shape","title":"Group normalization layer. — layer_group_normalization","text":"Arbitrary. Use keyword argument input_shape (tuple integers, include samples axis) using layer first layer model.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_group_normalization.html","id":"output-shape","dir":"Reference","previous_headings":"","what":"Output Shape","title":"Group normalization layer. — layer_group_normalization","text":"shape input. **kwargs: Base layer keyword arguments (e.g. name dtype).","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_group_normalization.html","id":"reference","dir":"Reference","previous_headings":"","what":"Reference","title":"Group normalization layer. — layer_group_normalization","text":"Yuxin Wu & Kaiming , 2018","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_group_query_attention.html","id":null,"dir":"Reference","previous_headings":"","what":"Grouped Query Attention layer. — layer_group_query_attention","title":"Grouped Query Attention layer. — layer_group_query_attention","text":"implementation grouped-query attention introduced Ainslie et al., 2023. num_key_value_heads denotes number groups, setting num_key_value_heads 1 equivalent multi-query attention, num_key_value_heads equal num_query_heads equivalent multi-head attention. layer first projects query, key, value tensors. , key value repeated match number heads query. , query scaled dot-producted key tensors. softmaxed obtain attention probabilities. value tensors interpolated probabilities concatenated back single tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_group_query_attention.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Grouped Query Attention layer. — layer_group_query_attention","text":"","code":"layer_group_query_attention(   object,   head_dim,   num_query_heads,   num_key_value_heads,   dropout = 0,   use_bias = TRUE,   flash_attention = NULL,   kernel_initializer = \"glorot_uniform\",   bias_initializer = \"zeros\",   kernel_regularizer = NULL,   bias_regularizer = NULL,   activity_regularizer = NULL,   kernel_constraint = NULL,   bias_constraint = NULL,   seed = NULL,   ... )"},{"path":"https://keras3.posit.co/dev/reference/layer_group_query_attention.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Grouped Query Attention layer. — layer_group_query_attention","text":"object Object compose layer . tensor, array, sequential model. head_dim Size attention head. num_query_heads Number query attention heads. num_key_value_heads Number key value attention heads. dropout Dropout probability. use_bias Boolean, whether dense layers use bias vectors/matrices. flash_attention NULL, layer attempts use flash attention faster memory-efficient attention computations possible. behavior can configured using config_enable_flash_attention() config_disable_flash_attention(). kernel_initializer Initializer dense layer kernels. bias_initializer Initializer dense layer biases. kernel_regularizer Regularizer dense layer kernels. bias_regularizer Regularizer dense layer biases. activity_regularizer Regularizer dense layer activity. kernel_constraint Constraint dense layer kernels. bias_constraint Constraint dense layer kernels. seed Optional integer seed dropout layer. ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_group_query_attention.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Grouped Query Attention layer. — layer_group_query_attention","text":"attention_output: Result computation, shape (batch_dim, target_seq_len, feature_dim), target_seq_len target sequence length feature_dim query input last dim. attention_scores: (Optional) attention coefficients shape (batch_dim, num_query_heads, target_seq_len, source_seq_len).","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_group_query_attention.html","id":"call-arguments","dir":"Reference","previous_headings":"","what":"Call Arguments","title":"Grouped Query Attention layer. — layer_group_query_attention","text":"query: Query tensor shape (batch_dim, target_seq_len, feature_dim), batch_dim batch size, target_seq_len length target sequence, feature_dim dimension feature. value: Value tensor shape (batch_dim, source_seq_len, feature_dim), batch_dim batch size, source_seq_len length source sequence, feature_dim dimension feature. key: Optional key tensor shape (batch_dim, source_seq_len, feature_dim). given, use value key value, common case. attention_mask: boolean mask shape (batch_dim, target_seq_len, source_seq_len), prevents attention certain positions. boolean mask specifies query elements can attend key elements, 1 indicates attention 0 indicates attention. Broadcasting can happen missing batch dimensions head dimension. return_attention_scores: boolean indicate whether output (attention_output, attention_scores) TRUE, attention_output FALSE. Defaults FALSE. training: Python boolean indicating whether layer behave training mode (adding dropout) inference mode (dropout). go either using training mode parent layer/model FALSE (inference) parent layer. use_causal_mask: boolean indicate whether apply causal mask prevent tokens attending future tokens (e.g., used decoder Transformer).","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_gru.html","id":null,"dir":"Reference","previous_headings":"","what":"Gated Recurrent Unit - Cho et al. 2014. — layer_gru","title":"Gated Recurrent Unit - Cho et al. 2014. — layer_gru","text":"Based available runtime hardware constraints, layer choose different implementations (cuDNN-based backend-native) maximize performance. GPU available arguments layer meet requirement cuDNN kernel (see details), layer use fast cuDNN implementation using TensorFlow backend. requirements use cuDNN implementation : activation == tanh recurrent_activation == sigmoid recurrent_dropout == 0 unroll FALSE use_bias TRUE reset_after TRUE Inputs, use masking, strictly right-padded. Eager execution enabled outermost context. two variants GRU implementation. default one based v3 reset gate applied hidden state matrix multiplication. one based original order reversed. second variant compatible CuDNNGRU (GPU-) allows inference CPU. Thus separate biases kernel recurrent_kernel. use variant, set reset_after=TRUE recurrent_activation='sigmoid'. example:","code":"inputs <- random_uniform(c(32, 10, 8)) outputs <- inputs |> layer_gru(4) shape(outputs) ## shape(32, 4) # (32, 4) gru <- layer_gru(, 4, return_sequences = TRUE, return_state = TRUE) c(whole_sequence_output, final_state) %<-% gru(inputs) shape(whole_sequence_output) ## shape(32, 10, 4) shape(final_state) ## shape(32, 4)"},{"path":"https://keras3.posit.co/dev/reference/layer_gru.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Gated Recurrent Unit - Cho et al. 2014. — layer_gru","text":"","code":"layer_gru(   object,   units,   activation = \"tanh\",   recurrent_activation = \"sigmoid\",   use_bias = TRUE,   kernel_initializer = \"glorot_uniform\",   recurrent_initializer = \"orthogonal\",   bias_initializer = \"zeros\",   kernel_regularizer = NULL,   recurrent_regularizer = NULL,   bias_regularizer = NULL,   activity_regularizer = NULL,   kernel_constraint = NULL,   recurrent_constraint = NULL,   bias_constraint = NULL,   dropout = 0,   recurrent_dropout = 0,   seed = NULL,   return_sequences = FALSE,   return_state = FALSE,   go_backwards = FALSE,   stateful = FALSE,   unroll = FALSE,   reset_after = TRUE,   use_cudnn = \"auto\",   ... )"},{"path":"https://keras3.posit.co/dev/reference/layer_gru.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Gated Recurrent Unit - Cho et al. 2014. — layer_gru","text":"object Object compose layer . tensor, array, sequential model. units Positive integer, dimensionality output space. activation Activation function use. Default: hyperbolic tangent (tanh). pass NULL, activation applied (ie. \"linear\" activation: (x) = x). recurrent_activation Activation function use recurrent step. Default: sigmoid (sigmoid). pass NULL, activation applied (ie. \"linear\" activation: (x) = x). use_bias Boolean, (default TRUE), whether layer use bias vector. kernel_initializer Initializer kernel weights matrix, used linear transformation inputs. Default: \"glorot_uniform\". recurrent_initializer Initializer recurrent_kernel weights matrix, used linear transformation recurrent state. Default: \"orthogonal\". bias_initializer Initializer bias vector. Default: \"zeros\". kernel_regularizer Regularizer function applied kernel weights matrix. Default: NULL. recurrent_regularizer Regularizer function applied recurrent_kernel weights matrix. Default: NULL. bias_regularizer Regularizer function applied bias vector. Default: NULL. activity_regularizer Regularizer function applied output layer (\"activation\"). Default: NULL. kernel_constraint Constraint function applied kernel weights matrix. Default: NULL. recurrent_constraint Constraint function applied recurrent_kernel weights matrix. Default: NULL. bias_constraint Constraint function applied bias vector. Default: NULL. dropout Float 0 1. Fraction units drop linear transformation inputs. Default: 0. recurrent_dropout Float 0 1. Fraction units drop linear transformation recurrent state. Default: 0. seed Random seed dropout. return_sequences Boolean. Whether return last output output sequence, full sequence. Default: FALSE. return_state Boolean. Whether return last state addition output. Default: FALSE. go_backwards Boolean (default FALSE). TRUE, process input sequence backwards return reversed sequence. stateful Boolean (default: FALSE). TRUE, last state sample index batch used initial state sample index following batch. unroll Boolean (default: FALSE). TRUE, network unrolled, else symbolic loop used. Unrolling can speed-RNN, although tends memory-intensive. Unrolling suitable short sequences. reset_after GRU convention (whether apply reset gate matrix multiplication). FALSE \"\", TRUE \"\" (default cuDNN compatible). use_cudnn Whether use cuDNN-backed implementation. \"auto\" attempt use cuDNN feasible, fallback default implementation . ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_gru.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Gated Recurrent Unit - Cho et al. 2014. — layer_gru","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_gru.html","id":"call-arguments","dir":"Reference","previous_headings":"","what":"Call Arguments","title":"Gated Recurrent Unit - Cho et al. 2014. — layer_gru","text":"inputs: 3D tensor, shape (batch, timesteps, feature). mask: Binary tensor shape (samples, timesteps) indicating whether given timestep masked  (optional). individual TRUE entry indicates corresponding timestep utilized, FALSE entry indicates corresponding timestep ignored. Defaults NULL. training: Python boolean indicating whether layer behave training mode inference mode. argument passed cell calling . relevant dropout recurrent_dropout used  (optional). Defaults NULL. initial_state: List initial state tensors passed first call cell (optional, NULL causes creation zero-filled initial state tensors). Defaults NULL.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_hashed_crossing.html","id":null,"dir":"Reference","previous_headings":"","what":"A preprocessing layer which crosses features using the ","title":"A preprocessing layer which crosses features using the ","text":"layer performs crosses categorical features using \"hashing trick\". Conceptually, transformation can thought : hash(concatenate(features)) %% num_bins. layer currently performs crosses scalar inputs batches scalar inputs. Valid input shapes (batch_size, 1), (batch_size) (). Note: layer wraps tf.keras.layers.HashedCrossing. used part compiled computation graph model backend TensorFlow. can however used backend running eagerly. can also always used part input preprocessing pipeline backend (outside model ), recommend use layer. Note: layer safe use inside tfdatasets pipeline (independently backend using).","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_hashed_crossing.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"A preprocessing layer which crosses features using the ","text":"","code":"layer_hashed_crossing(   object,   num_bins,   output_mode = \"int\",   sparse = FALSE,   name = NULL,   dtype = NULL,   ... )"},{"path":"https://keras3.posit.co/dev/reference/layer_hashed_crossing.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"A preprocessing layer which crosses features using the ","text":"object Object compose layer . tensor, array, sequential model. num_bins Number hash bins. output_mode Specification output layer. Values can \"int\", \"one_hot\" configuring layer follows: \"int\": Return integer bin indices directly. \"one_hot\": Encodes individual element input array size num_bins, containing 1 input's bin index. Defaults \"int\". sparse Boolean. applicable \"one_hot\" mode valid using TensorFlow backend. TRUE, returns SparseTensor instead dense Tensor. Defaults FALSE. name String, name object dtype datatype (e.g., \"float32\"). ... Keyword arguments construct layer.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_hashed_crossing.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"A preprocessing layer which crosses features using the ","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_hashed_crossing.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"A preprocessing layer which crosses features using the ","text":"Crossing two scalar features.     Crossing one-hotting two scalar features.","code":"feat1 <- c('A', 'B', 'A', 'B', 'A') |> as.array() feat2 <- c(101, 101, 101, 102, 102) |> as.integer() |> as.array() layer <- layer_hashed_crossing(num_bins = 5) layer(list(feat1, feat2)) ## tf.Tensor([1 4 1 1 3], shape=(5), dtype=int64) layer <- layer_hashed_crossing(num_bins = 5, output_mode = 'one_hot') layer(list(feat1, feat2)) ## tf.Tensor( ## [[0. 1. 0. 0. 0.] ##  [0. 0. 0. 0. 1.] ##  [0. 1. 0. 0. 0.] ##  [0. 1. 0. 0. 0.] ##  [0. 0. 0. 1. 0.]], shape=(5, 5), dtype=float32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_hashing.html","id":null,"dir":"Reference","previous_headings":"","what":"A preprocessing layer which hashes and bins categorical features. — layer_hashing","title":"A preprocessing layer which hashes and bins categorical features. — layer_hashing","text":"layer transforms categorical inputs hashed output. element-wise converts ints strings ints fixed range. stable hash function uses tensorflow::ops::Fingerprint produce output consistently across platforms. layer uses FarmHash64 default, provides consistent hashed output across different platforms stable across invocations, regardless device context, mixing input bits thoroughly. want obfuscate hashed output, can also pass random salt argument constructor. case, layer use SipHash64 hash function, salt value serving additional input hash function. Note: layer internally uses TensorFlow. used part compiled computation graph model backend TensorFlow. can however used backend running eagerly. can also always used part input preprocessing pipeline backend (outside model ), recommend use layer. Note: layer safe use inside tf.data pipeline (independently backend using). Example (FarmHash64)     Example (FarmHash64) mask value     Example (SipHash64)     Example (Siphash64 single integer, salt=[133, 133])","code":"layer <- layer_hashing(num_bins = 3) inp <- c('A', 'B', 'C', 'D', 'E') |> array(dim = c(5, 1)) layer(inp) ## tf.Tensor( ## [[1] ##  [0] ##  [1] ##  [1] ##  [2]], shape=(5, 1), dtype=int64) layer <- layer_hashing(num_bins=3, mask_value='') inp <- c('A', 'B', '', 'C', 'D') |> array(dim = c(5, 1)) layer(inp) ## tf.Tensor( ## [[1] ##  [1] ##  [0] ##  [2] ##  [2]], shape=(5, 1), dtype=int64) layer <- layer_hashing(num_bins=3, salt=c(133, 137)) inp <- c('A', 'B', 'C', 'D', 'E') |> array(dim = c(5, 1)) layer(inp) ## tf.Tensor( ## [[1] ##  [2] ##  [1] ##  [0] ##  [2]], shape=(5, 1), dtype=int64) layer <- layer_hashing(num_bins=3, salt=133) inp <- c('A', 'B', 'C', 'D', 'E') |> array(dim = c(5, 1)) layer(inp) ## tf.Tensor( ## [[0] ##  [0] ##  [2] ##  [1] ##  [0]], shape=(5, 1), dtype=int64)"},{"path":"https://keras3.posit.co/dev/reference/layer_hashing.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"A preprocessing layer which hashes and bins categorical features. — layer_hashing","text":"","code":"layer_hashing(   object,   num_bins,   mask_value = NULL,   salt = NULL,   output_mode = \"int\",   sparse = FALSE,   ... )"},{"path":"https://keras3.posit.co/dev/reference/layer_hashing.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"A preprocessing layer which hashes and bins categorical features. — layer_hashing","text":"object Object compose layer . tensor, array, sequential model. num_bins Number hash bins. Note includes mask_value bin, effective number bins (num_bins - 1) mask_value set. mask_value value represents masked inputs, mapped index 0. NULL means mask term added hashing start index 0. Defaults NULL. salt single unsigned integer NULL. passed, hash function used SipHash64, values used additional input (known \"salt\" cryptography). non-zero. NULL, uses FarmHash64 hash function. also supports list 2 unsigned integer numbers, see reference paper details. Defaults NULL. output_mode Specification output layer. Values can \"int\", \"one_hot\", \"multi_hot\", \"count\" configuring layer follows: \"int\": Return integer bin indices directly. \"one_hot\": Encodes individual element input array size num_bins, containing 1 input's bin index. last dimension size 1, encode dimension. last dimension size 1, append new dimension encoded output. \"multi_hot\": Encodes sample input single array size num_bins, containing 1 bin index index present sample. Treats last dimension sample dimension, input shape (..., sample_length), output shape (..., num_tokens). \"count\": \"multi_hot\", int array contains count number times bin index appeared sample. Defaults \"int\". sparse Boolean. applicable \"one_hot\", \"multi_hot\", \"count\" output modes. supported TensorFlow backend. TRUE, returns SparseTensor instead dense Tensor. Defaults FALSE. ... Keyword arguments construct layer.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_hashing.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"A preprocessing layer which hashes and bins categorical features. — layer_hashing","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_hashing.html","id":"input-shape","dir":"Reference","previous_headings":"","what":"Input Shape","title":"A preprocessing layer which hashes and bins categorical features. — layer_hashing","text":"single string, list strings, int32 int64 tensor shape (batch_size, ...,).","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_hashing.html","id":"output-shape","dir":"Reference","previous_headings":"","what":"Output Shape","title":"A preprocessing layer which hashes and bins categorical features. — layer_hashing","text":"int32 tensor shape (batch_size, ...).","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_hashing.html","id":"reference","dir":"Reference","previous_headings":"","what":"Reference","title":"A preprocessing layer which hashes and bins categorical features. — layer_hashing","text":"SipHash salt","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_identity.html","id":null,"dir":"Reference","previous_headings":"","what":"Identity layer. — layer_identity","title":"Identity layer. — layer_identity","text":"layer used placeholder operation performed. layer just returns inputs argument output.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_identity.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Identity layer. — layer_identity","text":"","code":"layer_identity(object, ...)"},{"path":"https://keras3.posit.co/dev/reference/layer_identity.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Identity layer. — layer_identity","text":"object Object compose layer . tensor, array, sequential model. ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_identity.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Identity layer. — layer_identity","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_input.html","id":null,"dir":"Reference","previous_headings":"","what":"keras_input — layer_input","title":"keras_input — layer_input","text":"layer_input() alias keras_input(). See ?keras_input() full documentation.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_input.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"keras_input — layer_input","text":"","code":"layer_input(   shape = NULL,   batch_size = NULL,   dtype = NULL,   sparse = NULL,   ragged = NULL,   batch_shape = NULL,   name = NULL,   tensor = NULL,   optional = FALSE )"},{"path":"https://keras3.posit.co/dev/reference/layer_input.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"keras_input — layer_input","text":"shape shape list (list integers NULL objects), including batch size. instance, shape = c(32) indicates expected input batches 32-dimensional vectors. Elements list can NULL NA; NULL/NA elements represent dimensions shape known may vary (e.g. sequence length). batch_size Optional static batch size (integer). dtype data type expected input, string (e.g. \"float32\", \"int32\"...) sparse boolean specifying whether expected input sparse tensors. Note , sparse FALSE, sparse tensors can still passed input - densified default value 0. feature supported TensorFlow backend. Defaults FALSE. ragged boolean specifying whether expected input ragged tensors. Note , ragged FALSE, ragged tensors can still passed input - densified default value 0. feature supported TensorFlow backend. Defaults FALSE. batch_shape Optional shape list (list integers NULL objects), including batch size. name Optional name string layer. unique model (reuse name twice). autogenerated provided. tensor Optional existing tensor wrap Input layer. set, layer use tensor rather creating new placeholder tensor. optional Boolean, whether input optional . optional input can accept NULL values.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_input.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"keras_input — layer_input","text":"Keras tensor, can passed inputs argument (keras_model()).","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_integer_lookup.html","id":null,"dir":"Reference","previous_headings":"","what":"A preprocessing layer that maps integers to (possibly encoded) indices. — layer_integer_lookup","title":"A preprocessing layer that maps integers to (possibly encoded) indices. — layer_integer_lookup","text":"layer maps set arbitrary integer input tokens indexed integer output via table-based vocabulary lookup. layer's output indices contiguously arranged maximum vocab size, even input tokens non-continguous unbounded. layer supports multiple options encoding output via output_mode, optional support --vocabulary (OOV) tokens masking. vocabulary layer must either supplied construction learned via adapt(). adapt(), layer analyze data set, determine frequency individual integer tokens, create vocabulary . vocabulary capped size, frequent tokens used create vocabulary others treated OOV. two possible output modes layer.  output_mode \"int\", input integers converted index vocabulary (integer).  output_mode \"multi_hot\", \"count\", \"tf_idf\", input integers encoded array dimension corresponds element vocabulary. vocabulary can optionally contain mask token well OOV token (can optionally occupy multiple indices vocabulary, set num_oov_indices). position tokens vocabulary fixed. output_mode \"int\", vocabulary begin mask token index 0, followed OOV indices, followed rest vocabulary. output_mode \"multi_hot\", \"count\", \"tf_idf\" vocabulary begin OOV indices instances mask token dropped. Note: layer uses TensorFlow internally. used part compiled computation graph model backend TensorFlow. can however used backend running eagerly. can also always used part input preprocessing pipeline backend (outside model ), recommend use layer. Note: layer safe use inside tf.data pipeline (independently backend using). Note: working layer outputs directly (e.g., passing outputs another layer, using lower-level operations like [ op_*): returned indices 0-based. However, default settings, first (0) index OOV token, returned indices offset 1 may appear 1-based.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_integer_lookup.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"A preprocessing layer that maps integers to (possibly encoded) indices. — layer_integer_lookup","text":"","code":"layer_integer_lookup(   object,   max_tokens = NULL,   num_oov_indices = 1L,   mask_token = NULL,   oov_token = -1L,   vocabulary = NULL,   vocabulary_dtype = \"int64\",   idf_weights = NULL,   invert = FALSE,   output_mode = \"int\",   sparse = FALSE,   pad_to_max_tokens = FALSE,   name = NULL,   ... )"},{"path":"https://keras3.posit.co/dev/reference/layer_integer_lookup.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"A preprocessing layer that maps integers to (possibly encoded) indices. — layer_integer_lookup","text":"object Object compose layer . tensor, array, sequential model. max_tokens Maximum size vocabulary layer. specified adapting vocabulary setting pad_to_max_tokens=TRUE. NULL, cap size vocabulary. Note size includes OOV mask tokens. Defaults NULL. num_oov_indices number --vocabulary tokens use. value 1, OOV inputs modulated determine OOV value. value 0, OOV inputs cause error calling layer. Defaults 1. mask_token integer token represents masked inputs. output_mode \"int\", token included vocabulary mapped index 0. output modes, token appear vocabulary instances mask token input dropped. set NULL, mask term added. Defaults NULL. oov_token used invert TRUE. token return OOV indices. Defaults -1. vocabulary Optional. Either array integers string path text file. passing array, can pass list, list, 1D NumPy array, 1D tensor containing integer vocabulary terms. passing file path, file contain one line per term vocabulary. argument set, need adapt() layer. vocabulary_dtype dtype vocabulary terms, example \"int64\" \"int32\". Defaults \"int64\". idf_weights valid output_mode \"tf_idf\". list, list, 1D NumPy array, 1D tensor length vocabulary, containing floating point inverse document frequency weights, multiplied per sample term counts final TF-IDF weight. vocabulary argument set, output_mode \"tf_idf\", argument must supplied. invert valid output_mode \"int\". TRUE, layer map indices vocabulary items instead mapping vocabulary items indices. Defaults FALSE. output_mode Specification output layer. Values can \"int\", \"one_hot\", \"multi_hot\", \"count\", \"tf_idf\" configuring layer follows: \"int\": Return vocabulary indices input tokens. \"one_hot\": Encodes individual element input array size vocabulary, containing 1 element index. last dimension size 1, encode dimension. last dimension size 1, append new dimension encoded output. \"multi_hot\": Encodes sample input single array size vocabulary, containing 1 vocabulary term present sample. Treats last dimension sample dimension, input shape (..., sample_length), output shape (..., num_tokens). \"count\": \"multi_hot\", int array contains count number times token index appeared sample. \"tf_idf\": \"multi_hot\", TF-IDF algorithm applied find value token slot. \"int\" output, shape input output supported. output modes, currently output rank 2 supported. Defaults \"int\". sparse Boolean. applicable \"multi_hot\", \"count\", \"tf_idf\" output modes. supported TensorFlow backend. TRUE, returns SparseTensor instead dense Tensor. Defaults FALSE. pad_to_max_tokens applicable output_mode \"multi_hot\", \"count\", \"tf_idf\". TRUE, output feature axis padded max_tokens even number unique tokens vocabulary less max_tokens, resulting tensor shape (batch_size, max_tokens) regardless vocabulary size. Defaults FALSE. name String, name object ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_integer_lookup.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"A preprocessing layer that maps integers to (possibly encoded) indices. — layer_integer_lookup","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_integer_lookup.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"A preprocessing layer that maps integers to (possibly encoded) indices. — layer_integer_lookup","text":"Creating lookup layer known vocabulary example creates lookup layer pre-existing vocabulary.     Creating lookup layer adapted vocabulary example creates lookup layer generates vocabulary analyzing dataset.     Note OOV token -1 added vocabulary. remaining tokens sorted frequency (42, 2 occurrences, first) inverse sort order.     Lookups multiple OOV indices example demonstrates use lookup layer multiple OOV indices.  layer created one OOV index, OOV tokens hashed number OOV buckets, distributing OOV tokens deterministic fashion across set.     Note output OOV token 37 1, output OOV token 1000 0. -vocab terms output index increased 1 earlier examples (12 maps 2, etc) order make space extra OOV token. One-hot output Configure layer output_mode='one_hot'. Note first num_oov_indices dimensions ont_hot encoding represent OOV values.     Multi-hot output Configure layer output_mode = 'multi_hot'. Note first num_oov_indices dimensions multi_hot encoding represent OOV tokens     Token count output Configure layer output_mode='count'. multi_hot output, first num_oov_indices dimensions output represent OOV tokens.     TF-IDF output Configure layer output_mode='tf_idf'. multi_hot output, first num_oov_indices dimensions output represent OOV tokens. token bin output token_count * idf_weight, idf weights inverse document frequency weights per token. provided along vocabulary. Note idf_weight OOV tokens default average idf weights passed .     specify idf weights oov tokens, need pass entire vocabulary including leading oov token.     adapting layer \"tf_idf\" mode, input sample considered document, IDF weight per token calculated : log(1 + num_documents / (1 + token_document_count)). Inverse lookup example demonstrates map indices tokens using layer. (can also use adapt() inverse = TRUE, simplicity pass vocab example.)     Note first index correspond oov token default. Forward inverse lookup pairs example demonstrates use vocabulary standard lookup layer create inverse lookup layer.     example, input token 1000 resulted output -1, since 1000 vocabulary - got represented OOV, OOV tokens returned -1 inverse layer. Also, note inverse work, must already set forward layer vocabulary either directly via adapt() calling get_vocabulary().","code":"vocab <- c(12, 36, 1138, 42) |> as.integer() data <- op_array(rbind(c(12, 1138, 42),                        c(42, 1000, 36)))  # Note OOV tokens out <- data |> layer_integer_lookup(vocabulary = vocab) out ## tf.Tensor( ## [[1 3 4] ##  [4 0 2]], shape=(2, 3), dtype=int64) data <- op_array(rbind(c(12, 1138, 42),                        c(42, 1000, 36)))  # Note OOV tokens layer <- layer_integer_lookup() layer |> adapt(data) layer |> get_vocabulary() |> str() ## List of 6 ##  $ : int -1 ##  $ : num 42 ##  $ : num 1138 ##  $ : num 1000 ##  $ : num 36 ##  $ : num 12 layer(data) ## tf.Tensor( ## [[5 2 1] ##  [1 3 4]], shape=(2, 3), dtype=int64) vocab <- c(12, 36, 1138, 42) |> as.integer() data <- op_array(rbind(c(12, 1138, 42),                        c(37, 1000, 36)))  # Note OOV tokens out <- data |>   layer_integer_lookup(vocabulary = vocab,                        num_oov_indices = 2) out ## tf.Tensor( ## [[2 4 5] ##  [1 0 3]], shape=(2, 3), dtype=int64) vocab <- c(12, 36, 1138, 42) |> as.integer() data <- op_array(c(12, 36, 1138, 42, 7), 'int32')  # Note OOV tokens layer <- layer_integer_lookup(vocabulary = vocab,                               output_mode = 'one_hot') layer(data) ## tf.Tensor( ## [[0 1 0 0 0] ##  [0 0 1 0 0] ##  [0 0 0 1 0] ##  [0 0 0 0 1] ##  [1 0 0 0 0]], shape=(5, 5), dtype=int64) vocab <- c(12, 36, 1138, 42) |> as.integer() data <- op_array(rbind(c(12, 1138, 42, 42),                       c(42,    7, 36,  7)), \"int64\")  # Note OOV tokens layer <- layer_integer_lookup(vocabulary = vocab,                               output_mode = 'multi_hot') layer(data) ## tf.Tensor( ## [[0 1 0 1 1] ##  [1 0 1 0 1]], shape=(2, 5), dtype=int64) vocab <- c(12, 36, 1138, 42) |> as.integer() data <- rbind(c(12, 1138, 42, 42),               c(42,    7, 36,  7)) |> op_array(\"int64\") layer <- layer_integer_lookup(vocabulary = vocab,                               output_mode = 'count') layer(data) ## tf.Tensor( ## [[0 1 0 1 2] ##  [2 0 1 0 1]], shape=(2, 5), dtype=int64) vocab <- c(12, 36, 1138, 42) |> as.integer() idf_weights <- c(0.25, 0.75, 0.6, 0.4) data <- rbind(c(12, 1138, 42, 42),               c(42,    7, 36,  7)) |> op_array(\"int64\") layer <- layer_integer_lookup(output_mode = 'tf_idf',                               vocabulary = vocab,                               idf_weights = idf_weights) layer(data) ## tf.Tensor( ## [[0.   0.25 0.   0.6  0.8 ] ##  [1.   0.   0.75 0.   0.4 ]], shape=(2, 5), dtype=float32) vocab <- c(-1, 12, 36, 1138, 42) |> as.integer() idf_weights <- c(0.9, 0.25, 0.75, 0.6, 0.4) data <- rbind(c(12, 1138, 42, 42),               c(42,    7, 36,  7)) |> op_array(\"int64\") layer <- layer_integer_lookup(output_mode = 'tf_idf',                               vocabulary = vocab,                               idf_weights = idf_weights) layer(data) ## tf.Tensor( ## [[0.   0.25 0.   0.6  0.8 ] ##  [1.8  0.   0.75 0.   0.4 ]], shape=(2, 5), dtype=float32) vocab <- c(12, 36, 1138, 42) |> as.integer() data <- op_array(c(1, 3, 4,                    4, 0, 2)) |> op_reshape(c(2,-1)) |> op_cast(\"int32\") layer <- layer_integer_lookup(vocabulary = vocab, invert = TRUE) layer(data) ## tf.Tensor( ## [[  12 1138   42] ##  [  42   -1   36]], shape=(2, 3), dtype=int64) vocab <- c(12, 36, 1138, 42) |> as.integer() data <- op_array(rbind(c(12, 1138, 42), c(42, 1000, 36)), \"int32\") layer <- layer_integer_lookup(vocabulary = vocab) i_layer <- layer_integer_lookup(vocabulary = get_vocabulary(layer),                                 invert = TRUE) int_data <- layer(data) i_layer(int_data) ## tf.Tensor( ## [[  12 1138   42] ##  [  42   -1   36]], shape=(2, 3), dtype=int64)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_jax_model_wrapper.html","id":null,"dir":"Reference","previous_headings":"","what":"Keras Layer that wraps a JAX model. — layer_jax_model_wrapper","title":"Keras Layer that wraps a JAX model. — layer_jax_model_wrapper","text":"layer enables use JAX components within Keras using JAX backend Keras.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_jax_model_wrapper.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Keras Layer that wraps a JAX model. — layer_jax_model_wrapper","text":"","code":"layer_jax_model_wrapper(   object,   call_fn,   init_fn = NULL,   params = NULL,   state = NULL,   seed = NULL,   ...,   dtype = NULL )"},{"path":"https://keras3.posit.co/dev/reference/layer_jax_model_wrapper.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Keras Layer that wraps a JAX model. — layer_jax_model_wrapper","text":"object Object compose layer . tensor, array, sequential model. call_fn function call model. See description list arguments takes outputs returns. init_fn function call initialize model. See description list arguments takes outputs returns. NULL, params /state must provided. params PyTree containing model trainable parameters. allows passing trained parameters controlling initialization. params state NULL, init_fn() called build time initialize trainable parameters model. state PyTree containing model non-trainable state. allows passing learned state controlling initialization. params state NULL, call_fn() takes state argument, init_fn() called build time initialize non-trainable state model. seed Seed random number generator. Optional. ... forward/backward compatability. dtype dtype layer's computations weights. Can also keras.DTypePolicy. Optional. Defaults default policy.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_jax_model_wrapper.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Keras Layer that wraps a JAX model. — layer_jax_model_wrapper","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_jax_model_wrapper.html","id":"model-function","dir":"Reference","previous_headings":"","what":"Model function","title":"Keras Layer that wraps a JAX model. — layer_jax_model_wrapper","text":"layer accepts JAX models form function, call_fn(), must take following arguments exact names: params: trainable parameters model. state (optional): non-trainable state model. Can omitted model non-trainable state. rng (optional): jax.random.PRNGKey instance. Can omitted model need RNGs, neither training inference. inputs: inputs model, JAX array PyTree arrays. training (optional): argument specifying training mode inference mode, TRUE passed training mode. Can omitted model behaves training mode inference mode. inputs argument mandatory. Inputs model must provided via single argument. JAX model takes multiple inputs separate arguments, must combined single structure, instance tuple() dict().","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_jax_model_wrapper.html","id":"model-weights-initialization","dir":"Reference","previous_headings":"","what":"Model weights initialization","title":"Keras Layer that wraps a JAX model. — layer_jax_model_wrapper","text":"initialization params state model can handled layer, case init_fn() argument must provided. allows model initialized dynamically right shape. Alternatively, shape known, params argument optionally state argument can used create already initialized model. init_fn() function, provided, must take following arguments exact names: rng: jax.random.PRNGKey instance. inputs: JAX array PyTree arrays placeholder values provide shape inputs. training (optional): argument specifying training mode inference mode. True always passed init_fn. Can omitted regardless whether call_fn training argument.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_jax_model_wrapper.html","id":"models-with-non-trainable-state","dir":"Reference","previous_headings":"","what":"Models with non-trainable state","title":"Keras Layer that wraps a JAX model. — layer_jax_model_wrapper","text":"JAX models non-trainable state: call_fn() must state argument call_fn() must return tuple() containing outputs model new non-trainable state model init_fn() must return tuple() containing initial trainable params model initial non-trainable state model. code shows possible combination call_fn() init_fn() signatures model non-trainable state. example, model training argument rng argument call_fn().","code":"stateful_call <- function(params, state, rng, inputs, training) {   outputs <- ....   new_state <- ....   tuple(outputs, new_state) }  stateful_init <- function(rng, inputs) {   initial_params <- ....   initial_state <- ....   tuple(initial_params, initial_state) }"},{"path":"https://keras3.posit.co/dev/reference/layer_jax_model_wrapper.html","id":"models-without-non-trainable-state","dir":"Reference","previous_headings":"","what":"Models without non-trainable state","title":"Keras Layer that wraps a JAX model. — layer_jax_model_wrapper","text":"JAX models non-trainable state: call_fn() must state argument call_fn() must return outputs model init_fn() must return initial trainable params model. code shows possible combination call_fn() init_fn() signatures model without non-trainable state. example, model training argument rng argument call_fn().","code":"stateful_call <- function(pparams, inputs) {   outputs <- ....   outputs }  stateful_init <- function(rng, inputs) {   initial_params <- ....   initial_params }"},{"path":"https://keras3.posit.co/dev/reference/layer_jax_model_wrapper.html","id":"conforming-to-the-required-signature","dir":"Reference","previous_headings":"","what":"Conforming to the required signature","title":"Keras Layer that wraps a JAX model. — layer_jax_model_wrapper","text":"model different signature one required JaxLayer, one can easily write wrapper method adapt arguments. example shows model multiple inputs separate arguments, expects multiple RNGs dict, deterministic argument opposite meaning training. conform, inputs combined single structure using tuple, RNG split used populate expected dict, Boolean flag negated:","code":"jax <- import(\"jax\") my_model_fn <- function(params, rngs, input1, input2, deterministic) {   ....   if (!deterministic) {     dropout_rng <- rngs$dropout     keep <- jax$random$bernoulli(dropout_rng, dropout_rate, x$shape)     x <- jax$numpy$where(keep, x / dropout_rate, 0)     ....   }   ....   return(outputs) }  my_model_wrapper_fn <- function(params, rng, inputs, training) {   c(input1, input2) %<-% inputs   c(rng1, rng2) %<-% jax$random$split(rng)   rngs <-  list(dropout = rng1, preprocessing = rng2)   deterministic <-  !training   my_model_fn(params, rngs, input1, input2, deterministic) }  keras_layer <- layer_jax_model_wrapper(call_fn = my_model_wrapper_fn,                                        params = initial_params)"},{"path":"https://keras3.posit.co/dev/reference/layer_jax_model_wrapper.html","id":"usage-with-haiku-modules","dir":"Reference","previous_headings":"","what":"Usage with Haiku modules","title":"Keras Layer that wraps a JAX model. — layer_jax_model_wrapper","text":"JaxLayer enables use Haiku components form haiku.Module. achieved transforming module per Haiku pattern passing module.apply call_fn parameter module.init init_fn parameter needed. model non-trainable state, transformed haiku.transform_with_state. model non-trainable state, transformed haiku.transform. Additionally, optionally, module use RNGs \"apply\", can transformed haiku.without_apply_rng. following example shows create JaxLayer Haiku module uses random number generators via hk.next_rng_key() takes training positional argument:","code":"# reticulate::py_install(\"haiku\", \"r-keras\") hk <- import(\"haiku\") MyHaikuModule(hk$Module) \\%py_class\\% {    `__call__` <- \\(self, x, training) {     x <- hk$Conv2D(32L, tuple(3L, 3L))(x)     x <- jax$nn$relu(x)     x <- hk$AvgPool(tuple(1L, 2L, 2L, 1L),                     tuple(1L, 2L, 2L, 1L), \"VALID\")(x)     x <- hk$Flatten()(x)     x <- hk$Linear(200L)(x)     if (training)       x <- hk$dropout(rng = hk$next_rng_key(), rate = 0.3, x = x)     x <- jax$nn$relu(x)     x <- hk$Linear(10L)(x)     x <- jax$nn$softmax(x)     x   }  }  my_haiku_module_fn <- function(inputs, training) {   module <- MyHaikuModule()   module(inputs, training) }  transformed_module <- hk$transform(my_haiku_module_fn)  keras_layer <-   layer_jax_model_wrapper(call_fn = transformed_module$apply,                           init_fn = transformed_module$init)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_lambda.html","id":null,"dir":"Reference","previous_headings":"","what":"Wraps arbitrary expressions as a Layer object. — layer_lambda","title":"Wraps arbitrary expressions as a Layer object. — layer_lambda","text":"layer_lambda() layer exists arbitrary expressions can used Layer constructing Sequential Functional API models. Lambda layers best suited simple operations quick experimentation. advanced use cases, prefer writing new subclasses Layer using new_layer_class().","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_lambda.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Wraps arbitrary expressions as a Layer object. — layer_lambda","text":"","code":"layer_lambda(   object,   f,   output_shape = NULL,   mask = NULL,   arguments = NULL,   ... )"},{"path":"https://keras3.posit.co/dev/reference/layer_lambda.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Wraps arbitrary expressions as a Layer object. — layer_lambda","text":"object Object compose layer . tensor, array, sequential model. f function evaluated. Takes input tensor first argument. output_shape Expected output shape function. argument can usually inferred explicitly provided. Can list function. list, specifies first dimension onward; sample dimension assumed either input: output_shape = c(input_shape[1], output_shape) , input NULL sample dimension also NULL: output_shape = c(NA, output_shape). function, specifies entire shape function input shape: output_shape = f(input_shape). mask Either NULL (indicating masking) callable signature compute_mask layer method, tensor returned output mask regardless input . arguments Optional named list arguments passed function. ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_lambda.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Wraps arbitrary expressions as a Layer object. — layer_lambda","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_lambda.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Wraps arbitrary expressions as a Layer object. — layer_lambda","text":"","code":"# add a x -> x^2 layer model <- keras_model_sequential() model |> layer_lambda(\\(x) x^2)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_layer_normalization.html","id":null,"dir":"Reference","previous_headings":"","what":"Layer normalization layer (Ba et al., 2016). — layer_layer_normalization","title":"Layer normalization layer (Ba et al., 2016). — layer_layer_normalization","text":"Normalize activations previous layer given example batch independently, rather across batch like Batch Normalization. .e. applies transformation maintains mean activation within example close 0 activation standard deviation close 1. scale center enabled, layer scale normalized outputs broadcasting trainable variable gamma, center outputs broadcasting trainable variable beta. gamma default ones tensor beta default zeros tensor, centering scaling -ops training begun. , scaling centering enabled normalization equations follows: Let intermediate activations mini-batch inputs. sample x batch inputs, compute mean variance sample, normalize value sample (including small factor epsilon numerical stability), finally, transform normalized output gamma beta, learned parameters:   gamma beta span axes inputs specified axis, part inputs' shape must fully defined. example:         Note implementations layer normalization may choose define gamma beta separate set axes axes normalized across. example, Group Normalization (Wu et al. 2018) group size 1 corresponds layer_layer_normalization() normalizes across height, width, channel gamma beta span channel dimension. , layer_layer_normalization() implementation match layer_group_normalization() layer group size set 1.","code":"outputs <- inputs |> apply(1, function(x) {   x_normalized <- (x - mean(x)) /                   sqrt(var(x) + epsilon)   x_normalized * gamma + beta }) layer <- layer_layer_normalization(axis = c(2, 3, 4))  layer(op_ones(c(5, 20, 30, 40))) |> invisible() # build() shape(layer$beta) ## shape(20, 30, 40) shape(layer$gamma) ## shape(20, 30, 40)"},{"path":"https://keras3.posit.co/dev/reference/layer_layer_normalization.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Layer normalization layer (Ba et al., 2016). — layer_layer_normalization","text":"","code":"layer_layer_normalization(   object,   axis = -1L,   epsilon = 0.001,   center = TRUE,   scale = TRUE,   beta_initializer = \"zeros\",   gamma_initializer = \"ones\",   beta_regularizer = NULL,   gamma_regularizer = NULL,   beta_constraint = NULL,   gamma_constraint = NULL,   ... )"},{"path":"https://keras3.posit.co/dev/reference/layer_layer_normalization.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Layer normalization layer (Ba et al., 2016). — layer_layer_normalization","text":"object Object compose layer . tensor, array, sequential model. axis Integer list. axis axes normalize across. Typically, features axis/axes. left-axes typically batch axis/axes. -1 last dimension input. Defaults -1. epsilon Small float added variance avoid dividing zero. Defaults 1e-3. center TRUE, add offset beta normalized tensor. FALSE, beta ignored. Defaults TRUE. scale TRUE, multiply gamma. FALSE, gamma used. next layer linear (also e.g. layer_activation_relu()), can disabled since scaling done next layer. Defaults TRUE. beta_initializer Initializer beta weight. Defaults zeros. gamma_initializer Initializer gamma weight. Defaults ones. beta_regularizer Optional regularizer beta weight. NULL default. gamma_regularizer Optional regularizer gamma weight. NULL default. beta_constraint Optional constraint beta weight. NULL default. gamma_constraint Optional constraint gamma weight. NULL default. ... Base layer keyword arguments (e.g. name dtype).","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_layer_normalization.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Layer normalization layer (Ba et al., 2016). — layer_layer_normalization","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_layer_normalization.html","id":"reference","dir":"Reference","previous_headings":"","what":"Reference","title":"Layer normalization layer (Ba et al., 2016). — layer_layer_normalization","text":"Lei Ba et al., 2016.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_lstm.html","id":null,"dir":"Reference","previous_headings":"","what":"Long Short-Term Memory layer - Hochreiter 1997. — layer_lstm","title":"Long Short-Term Memory layer - Hochreiter 1997. — layer_lstm","text":"Based available runtime hardware constraints, layer choose different implementations (cuDNN-based backend-native) maximize performance. GPU available arguments layer meet requirement cuDNN kernel (see details), layer use fast cuDNN implementation using TensorFlow backend. requirements use cuDNN implementation : activation == tanh recurrent_activation == sigmoid recurrent_dropout == 0 unroll FALSE use_bias TRUE Inputs, use masking, strictly right-padded. Eager execution enabled outermost context. example:","code":"input <- random_uniform(c(32, 10, 8)) output <- input |> layer_lstm(4) shape(output) ## shape(32, 4) lstm <- layer_lstm(units = 4, return_sequences = TRUE, return_state = TRUE) c(whole_seq_output, final_memory_state, final_carry_state) %<-% lstm(input) shape(whole_seq_output) ## shape(32, 10, 4) shape(final_memory_state) ## shape(32, 4) shape(final_carry_state) ## shape(32, 4)"},{"path":"https://keras3.posit.co/dev/reference/layer_lstm.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Long Short-Term Memory layer - Hochreiter 1997. — layer_lstm","text":"","code":"layer_lstm(   object,   units,   activation = \"tanh\",   recurrent_activation = \"sigmoid\",   use_bias = TRUE,   kernel_initializer = \"glorot_uniform\",   recurrent_initializer = \"orthogonal\",   bias_initializer = \"zeros\",   unit_forget_bias = TRUE,   kernel_regularizer = NULL,   recurrent_regularizer = NULL,   bias_regularizer = NULL,   activity_regularizer = NULL,   kernel_constraint = NULL,   recurrent_constraint = NULL,   bias_constraint = NULL,   dropout = 0,   recurrent_dropout = 0,   seed = NULL,   return_sequences = FALSE,   return_state = FALSE,   go_backwards = FALSE,   stateful = FALSE,   unroll = FALSE,   use_cudnn = \"auto\",   ... )"},{"path":"https://keras3.posit.co/dev/reference/layer_lstm.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Long Short-Term Memory layer - Hochreiter 1997. — layer_lstm","text":"object Object compose layer . tensor, array, sequential model. units Positive integer, dimensionality output space. activation Activation function use. Default: hyperbolic tangent (tanh). pass NULL, activation applied (ie. \"linear\" activation: (x) = x). recurrent_activation Activation function use recurrent step. Default: sigmoid (sigmoid). pass NULL, activation applied (ie. \"linear\" activation: (x) = x). use_bias Boolean, (default TRUE), whether layer use bias vector. kernel_initializer Initializer kernel weights matrix, used linear transformation inputs. Default: \"glorot_uniform\". recurrent_initializer Initializer recurrent_kernel weights matrix, used linear transformation recurrent state. Default: \"orthogonal\". bias_initializer Initializer bias vector. Default: \"zeros\". unit_forget_bias Boolean (default TRUE). TRUE, add 1 bias forget gate initialization. Setting TRUE also force bias_initializer=\"zeros\". recommended Jozefowicz et al. kernel_regularizer Regularizer function applied kernel weights matrix. Default: NULL. recurrent_regularizer Regularizer function applied recurrent_kernel weights matrix. Default: NULL. bias_regularizer Regularizer function applied bias vector. Default: NULL. activity_regularizer Regularizer function applied output layer (\"activation\"). Default: NULL. kernel_constraint Constraint function applied kernel weights matrix. Default: NULL. recurrent_constraint Constraint function applied recurrent_kernel weights matrix. Default: NULL. bias_constraint Constraint function applied bias vector. Default: NULL. dropout Float 0 1. Fraction units drop linear transformation inputs. Default: 0. recurrent_dropout Float 0 1. Fraction units drop linear transformation recurrent state. Default: 0. seed Random seed dropout. return_sequences Boolean. Whether return last output output sequence, full sequence. Default: FALSE. return_state Boolean. Whether return last state addition output. Default: FALSE. go_backwards Boolean (default: FALSE). TRUE, process input sequence backwards return reversed sequence. stateful Boolean (default: FALSE). TRUE, last state sample index batch used initial state sample index following batch. unroll Boolean (default FALSE). TRUE, network unrolled, else symbolic loop used. Unrolling can speed-RNN, although tends memory-intensive. Unrolling suitable short sequences. use_cudnn Whether use cuDNN-backed implementation. \"auto\" attempt use cuDNN feasible, fallback default implementation . ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_lstm.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Long Short-Term Memory layer - Hochreiter 1997. — layer_lstm","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_lstm.html","id":"call-arguments","dir":"Reference","previous_headings":"","what":"Call Arguments","title":"Long Short-Term Memory layer - Hochreiter 1997. — layer_lstm","text":"inputs: 3D tensor, shape (batch, timesteps, feature). mask: Binary tensor shape (samples, timesteps) indicating whether given timestep masked  (optional). individual TRUE entry indicates corresponding timestep utilized, FALSE entry indicates corresponding timestep ignored. Defaults NULL. training: Boolean indicating whether layer behave training mode inference mode. argument passed cell calling . relevant dropout recurrent_dropout used  (optional). Defaults NULL. initial_state: List initial state tensors passed first call cell (optional, NULL causes creation zero-filled initial state tensors). Defaults NULL.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_masking.html","id":null,"dir":"Reference","previous_headings":"","what":"Masks a sequence by using a mask value to skip timesteps. — layer_masking","title":"Masks a sequence by using a mask value to skip timesteps. — layer_masking","text":"timestep input tensor (second dimension tensor), values input tensor timestep equal mask_value, timestep masked (skipped) downstream layers (long support masking). downstream layer support masking yet receives input mask, exception raised.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_masking.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Masks a sequence by using a mask value to skip timesteps. — layer_masking","text":"","code":"layer_masking(object, mask_value = 0, ...)"},{"path":"https://keras3.posit.co/dev/reference/layer_masking.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Masks a sequence by using a mask value to skip timesteps. — layer_masking","text":"object Object compose layer . tensor, array, sequential model. mask_value see description ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_masking.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Masks a sequence by using a mask value to skip timesteps. — layer_masking","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_masking.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Masks a sequence by using a mask value to skip timesteps. — layer_masking","text":"Consider array x shape c(samples, timesteps, features), fed LSTM layer. want mask timestep #3 #5 lack data timesteps. can: Set x[, 3, ] <- 0. x[, 5, ] <- 0. Insert layer_masking() layer mask_value = 0. LSTM layer:","code":"c(samples, timesteps, features) %<-% c(32, 10, 8) inputs <- c(samples, timesteps, features) %>% { array(runif(prod(.)), dim = .) } inputs[, 3, ] <- 0 inputs[, 5, ] <- 0  model <- keras_model_sequential() %>%   layer_masking(mask_value = 0) %>%   layer_lstm(32)  output <- model(inputs) # The time step 3 and 5 will be skipped from LSTM calculation."},{"path":"https://keras3.posit.co/dev/reference/layer_masking.html","id":"note","dir":"Reference","previous_headings":"","what":"Note","title":"Masks a sequence by using a mask value to skip timesteps. — layer_masking","text":"Keras masking convention, masked timestep denoted mask value FALSE, non-masked (.e. usable) timestep denoted mask value TRUE.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_max_num_bounding_boxes.html","id":null,"dir":"Reference","previous_headings":"","what":"Ensure the maximum number of bounding boxes. — layer_max_num_bounding_boxes","title":"Ensure the maximum number of bounding boxes. — layer_max_num_bounding_boxes","text":"Ensure maximum number bounding boxes.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_max_num_bounding_boxes.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Ensure the maximum number of bounding boxes. — layer_max_num_bounding_boxes","text":"","code":"layer_max_num_bounding_boxes(object, max_number, fill_value = -1L, ...)"},{"path":"https://keras3.posit.co/dev/reference/layer_max_num_bounding_boxes.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Ensure the maximum number of bounding boxes. — layer_max_num_bounding_boxes","text":"object Object compose layer . tensor, array, sequential model. max_number Desired output number bounding boxes. fill_value fill value boxes labels bounding_boxes. Defaults -1. ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_max_num_bounding_boxes.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Ensure the maximum number of bounding boxes. — layer_max_num_bounding_boxes","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_max_pooling_1d.html","id":null,"dir":"Reference","previous_headings":"","what":"Max pooling operation for 1D temporal data. — layer_max_pooling_1d","title":"Max pooling operation for 1D temporal data. — layer_max_pooling_1d","text":"Downsamples input representation taking maximum value spatial window size pool_size. window shifted strides. resulting output using \"valid\" padding option shape : output_shape = (input_shape - pool_size + 1) / strides). resulting output shape using \"\" padding option : output_shape = input_shape / strides","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_max_pooling_1d.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Max pooling operation for 1D temporal data. — layer_max_pooling_1d","text":"","code":"layer_max_pooling_1d(   object,   pool_size = 2L,   strides = NULL,   padding = \"valid\",   data_format = NULL,   name = NULL,   ... )"},{"path":"https://keras3.posit.co/dev/reference/layer_max_pooling_1d.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Max pooling operation for 1D temporal data. — layer_max_pooling_1d","text":"object Object compose layer . tensor, array, sequential model. pool_size int, size max pooling window. strides int NULL. Specifies much pooling window moves pooling step. NULL, default pool_size. padding string, either \"valid\" \"\" (case-insensitive). \"valid\" means padding. \"\" results padding evenly left/right /input output height/width dimension input. data_format string, either \"channels_last\" \"channels_first\". ordering dimensions inputs. \"channels_last\" corresponds inputs shape (batch, steps, features) \"channels_first\" corresponds inputs shape (batch, features, steps). defaults image_data_format value found Keras config file ~/.keras/keras.json. never set , \"channels_last\". name String, name object ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_max_pooling_1d.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Max pooling operation for 1D temporal data. — layer_max_pooling_1d","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_max_pooling_1d.html","id":"input-shape","dir":"Reference","previous_headings":"","what":"Input Shape","title":"Max pooling operation for 1D temporal data. — layer_max_pooling_1d","text":"data_format=\"channels_last\": 3D tensor shape (batch_size, steps, features). data_format=\"channels_first\": 3D tensor shape (batch_size, features, steps).","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_max_pooling_1d.html","id":"output-shape","dir":"Reference","previous_headings":"","what":"Output Shape","title":"Max pooling operation for 1D temporal data. — layer_max_pooling_1d","text":"data_format=\"channels_last\": 3D tensor shape (batch_size, downsampled_steps, features). data_format=\"channels_first\": 3D tensor shape (batch_size, features, downsampled_steps).","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_max_pooling_1d.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Max pooling operation for 1D temporal data. — layer_max_pooling_1d","text":"strides=1 padding=\"valid\":     strides=2 padding=\"valid\":     strides=1 padding=\"\":","code":"x <- op_reshape(c(1, 2, 3, 4, 5),                c(1, 5, 1)) max_pool_1d <- layer_max_pooling_1d(pool_size = 2,                                     strides = 1,                                     padding = \"valid\") max_pool_1d(x) ## tf.Tensor( ## [[[2.] ##   [3.] ##   [4.] ##   [5.]]], shape=(1, 4, 1), dtype=float32) x <- op_reshape(c(1, 2, 3, 4, 5),                c(1, 5, 1)) max_pool_1d <- layer_max_pooling_1d(pool_size = 2,                                     strides = 2,                                     padding = \"valid\") max_pool_1d(x) ## tf.Tensor( ## [[[2.] ##   [4.]]], shape=(1, 2, 1), dtype=float32) x <- op_reshape(c(1, 2, 3, 4, 5),                c(1, 5, 1)) max_pool_1d <- layer_max_pooling_1d(pool_size = 2,                                     strides = 1,                                     padding = \"same\") max_pool_1d(x) ## tf.Tensor( ## [[[2.] ##   [3.] ##   [4.] ##   [5.] ##   [5.]]], shape=(1, 5, 1), dtype=float32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_max_pooling_2d.html","id":null,"dir":"Reference","previous_headings":"","what":"Max pooling operation for 2D spatial data. — layer_max_pooling_2d","title":"Max pooling operation for 2D spatial data. — layer_max_pooling_2d","text":"Downsamples input along spatial dimensions (height width) taking maximum value input window (size defined pool_size) channel input. window shifted strides along dimension. resulting output using \"valid\" padding option spatial shape (number rows columns) : output_shape = floor((input_shape - pool_size) / strides) + 1 (input_shape >= pool_size) resulting output shape using \"\" padding option : output_shape = floor((input_shape - 1) / strides) + 1","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_max_pooling_2d.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Max pooling operation for 2D spatial data. — layer_max_pooling_2d","text":"","code":"layer_max_pooling_2d(   object,   pool_size = list(2L, 2L),   strides = NULL,   padding = \"valid\",   data_format = NULL,   name = NULL,   ... )"},{"path":"https://keras3.posit.co/dev/reference/layer_max_pooling_2d.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Max pooling operation for 2D spatial data. — layer_max_pooling_2d","text":"object Object compose layer . tensor, array, sequential model. pool_size int list 2 integers, factors downscale (dim1, dim2). one integer specified, window length used dimensions. strides int list 2 integers, NULL. Strides values. NULL, default pool_size. one int specified, stride size used dimensions. padding string, either \"valid\" \"\" (case-insensitive). \"valid\" means padding. \"\" results padding evenly left/right /input output height/width dimension input. data_format string, either \"channels_last\" \"channels_first\". ordering dimensions inputs. \"channels_last\" corresponds inputs shape (batch, height, width, channels) \"channels_first\" corresponds inputs shape (batch, channels, height, width). defaults image_data_format value found Keras config file ~/.keras/keras.json. never set , \"channels_last\". name String, name object ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_max_pooling_2d.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Max pooling operation for 2D spatial data. — layer_max_pooling_2d","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_max_pooling_2d.html","id":"input-shape","dir":"Reference","previous_headings":"","what":"Input Shape","title":"Max pooling operation for 2D spatial data. — layer_max_pooling_2d","text":"data_format=\"channels_last\": 4D tensor shape (batch_size, height, width, channels). data_format=\"channels_first\": 4D tensor shape (batch_size, channels, height, width).","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_max_pooling_2d.html","id":"output-shape","dir":"Reference","previous_headings":"","what":"Output Shape","title":"Max pooling operation for 2D spatial data. — layer_max_pooling_2d","text":"data_format=\"channels_last\": 4D tensor shape (batch_size, pooled_height, pooled_width, channels). data_format=\"channels_first\": 4D tensor shape (batch_size, channels, pooled_height, pooled_width).","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_max_pooling_2d.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Max pooling operation for 2D spatial data. — layer_max_pooling_2d","text":"strides = (1, 1) padding = \"valid\":     strides = c(2, 2) padding = \"valid\":     stride = (1, 1) padding = \"\":","code":"x <- rbind(c(1., 2., 3.),            c(4., 5., 6.),            c(7., 8., 9.)) |> op_reshape(c(1, 3, 3, 1)) max_pool_2d <- layer_max_pooling_2d(pool_size = c(2, 2),                                     strides = c(1, 1),                                     padding = \"valid\") max_pool_2d(x) ## tf.Tensor( ## [[[[5.] ##    [6.]] ## ##   [[8.] ##    [9.]]]], shape=(1, 2, 2, 1), dtype=float32) x <- rbind(c(1., 2., 3., 4.),            c(5., 6., 7., 8.),            c(9., 10., 11., 12.)) |> op_reshape(c(1, 3, 4, 1)) max_pool_2d <- layer_max_pooling_2d(pool_size = c(2, 2),                                     strides = c(2, 2),                                     padding = \"valid\") max_pool_2d(x) ## tf.Tensor( ## [[[[6.] ##    [8.]]]], shape=(1, 1, 2, 1), dtype=float32) x <- rbind(c(1., 2., 3.),            c(4., 5., 6.),            c(7., 8., 9.)) |> op_reshape(c(1, 3, 3, 1)) max_pool_2d <- layer_max_pooling_2d(pool_size = c(2, 2),                                     strides = c(1, 1),                                     padding = \"same\") max_pool_2d(x) ## tf.Tensor( ## [[[[5.] ##    [6.] ##    [6.]] ## ##   [[8.] ##    [9.] ##    [9.]] ## ##   [[8.] ##    [9.] ##    [9.]]]], shape=(1, 3, 3, 1), dtype=float32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_max_pooling_3d.html","id":null,"dir":"Reference","previous_headings":"","what":"Max pooling operation for 3D data (spatial or spatio-temporal). — layer_max_pooling_3d","title":"Max pooling operation for 3D data (spatial or spatio-temporal). — layer_max_pooling_3d","text":"Downsamples input along spatial dimensions (depth, height, width) taking maximum value input window (size defined pool_size) channel input. window shifted strides along dimension.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_max_pooling_3d.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Max pooling operation for 3D data (spatial or spatio-temporal). — layer_max_pooling_3d","text":"","code":"layer_max_pooling_3d(   object,   pool_size = list(2L, 2L, 2L),   strides = NULL,   padding = \"valid\",   data_format = NULL,   name = NULL,   ... )"},{"path":"https://keras3.posit.co/dev/reference/layer_max_pooling_3d.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Max pooling operation for 3D data (spatial or spatio-temporal). — layer_max_pooling_3d","text":"object Object compose layer . tensor, array, sequential model. pool_size int list 3 integers, factors downscale (dim1, dim2, dim3). one integer specified, window length used dimensions. strides int list 3 integers, NULL. Strides values. NULL, default pool_size. one int specified, stride size used dimensions. padding string, either \"valid\" \"\" (case-insensitive). \"valid\" means padding. \"\" results padding evenly left/right /input output height/width dimension input. data_format string, either \"channels_last\" \"channels_first\". ordering dimensions inputs. \"channels_last\" corresponds inputs shape (batch, spatial_dim1, spatial_dim2, spatial_dim3, channels) \"channels_first\" corresponds inputs shape (batch, channels, spatial_dim1, spatial_dim2, spatial_dim3). defaults image_data_format value found Keras config file ~/.keras/keras.json. never set , \"channels_last\". name String, name object ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_max_pooling_3d.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Max pooling operation for 3D data (spatial or spatio-temporal). — layer_max_pooling_3d","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_max_pooling_3d.html","id":"input-shape","dir":"Reference","previous_headings":"","what":"Input Shape","title":"Max pooling operation for 3D data (spatial or spatio-temporal). — layer_max_pooling_3d","text":"data_format=\"channels_last\": 5D tensor shape: (batch_size, spatial_dim1, spatial_dim2, spatial_dim3, channels) data_format=\"channels_first\": 5D tensor shape: (batch_size, channels, spatial_dim1, spatial_dim2, spatial_dim3)","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_max_pooling_3d.html","id":"output-shape","dir":"Reference","previous_headings":"","what":"Output Shape","title":"Max pooling operation for 3D data (spatial or spatio-temporal). — layer_max_pooling_3d","text":"data_format=\"channels_last\": 5D tensor shape: (batch_size, pooled_dim1, pooled_dim2, pooled_dim3, channels) data_format=\"channels_first\": 5D tensor shape: (batch_size, channels, pooled_dim1, pooled_dim2, pooled_dim3)","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_max_pooling_3d.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Max pooling operation for 3D data (spatial or spatio-temporal). — layer_max_pooling_3d","text":"","code":"depth <- 30 height <- 30 width <- 30 channels <- 3  inputs <- layer_input(shape=c(depth, height, width, channels)) layer <- layer_max_pooling_3d(pool_size=3) outputs <- inputs |> layer() outputs ## <KerasTensor shape=(None, 10, 10, 10, 3), dtype=float32, sparse=False, ragged=False, name=keras_tensor_1>"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_maximum.html","id":null,"dir":"Reference","previous_headings":"","what":"Computes element-wise maximum on a list of inputs. — layer_maximum","title":"Computes element-wise maximum on a list of inputs. — layer_maximum","text":"takes input list tensors, shape, returns single tensor (also shape).","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_maximum.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes element-wise maximum on a list of inputs. — layer_maximum","text":"","code":"layer_maximum(inputs, ...)"},{"path":"https://keras3.posit.co/dev/reference/layer_maximum.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Computes element-wise maximum on a list of inputs. — layer_maximum","text":"inputs layers combine ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_maximum.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Computes element-wise maximum on a list of inputs. — layer_maximum","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_maximum.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Computes element-wise maximum on a list of inputs. — layer_maximum","text":"Usage Keras model:","code":"input_shape <- c(2, 3, 4) x1 <- random_uniform(input_shape) x2 <- random_uniform(input_shape) y <- layer_maximum(x1, x2) input1 <- layer_input(shape = c(16)) x1 <- input1 |> layer_dense(8, activation = 'relu') input2 <- layer_input(shape = c(32)) x2 <- input2 |> layer_dense(8, activation = 'relu') # equivalent to `y <- layer_maximum(x1, x2)` y <- layer_maximum(x1, x2) out <- y |> layer_dense(4) model <- keras_model(inputs = c(input1, input2), outputs = out)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_mel_spectrogram.html","id":null,"dir":"Reference","previous_headings":"","what":"A preprocessing layer to convert raw audio signals to Mel spectrograms. — layer_mel_spectrogram","title":"A preprocessing layer to convert raw audio signals to Mel spectrograms. — layer_mel_spectrogram","text":"layer takes float32/float64 single batched audio signal inputs computes Mel spectrogram using Short-Time Fourier Transform Mel scaling. input 1D (unbatched) 2D (batched) tensor representing audio signals. output 2D 3D tensor representing Mel spectrograms. spectrogram image-like representation shows frequency spectrum signal time. uses x-axis represent time, y-axis represent frequency, pixel represent intensity. Mel spectrograms special type spectrogram use mel scale, approximates humans perceive sound. commonly used speech music processing tasks like speech recognition, speaker identification, music genre classification.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_mel_spectrogram.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"A preprocessing layer to convert raw audio signals to Mel spectrograms. — layer_mel_spectrogram","text":"","code":"layer_mel_spectrogram(   object,   fft_length = 2048L,   sequence_stride = 512L,   sequence_length = NULL,   window = \"hann\",   sampling_rate = 16000L,   num_mel_bins = 128L,   min_freq = 20,   max_freq = NULL,   power_to_db = TRUE,   top_db = 80,   mag_exp = 2,   min_power = 1e-10,   ref_power = 1,   ... )"},{"path":"https://keras3.posit.co/dev/reference/layer_mel_spectrogram.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"A preprocessing layer to convert raw audio signals to Mel spectrograms. — layer_mel_spectrogram","text":"object Object compose layer . tensor, array, sequential model. fft_length Integer, size FFT window. sequence_stride Integer, number samples successive STFT columns. sequence_length Integer, size window used applying window audio frame. NULL, defaults fft_length. window String, name window function use. Available values \"hann\" \"hamming\". window tensor, used directly window length must sequence_length. window NULL, windowing used. Defaults \"hann\". sampling_rate Integer, sample rate input signal. num_mel_bins Integer, number mel bins generate. min_freq Float, minimum frequency mel bins. max_freq Float, maximum frequency mel bins. NULL, defaults sampling_rate / 2. power_to_db TRUE, convert power spectrogram decibels. top_db Float, minimum negative cut-max(10 * log10(S)) - top_db. mag_exp Float, exponent magnitude spectrogram. 1 magnitude, 2 power, etc. Default 2. min_power Float, minimum value power ref_power. ref_power Float, power scaled relative 10 * log10(S / ref_power). ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_mel_spectrogram.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"A preprocessing layer to convert raw audio signals to Mel spectrograms. — layer_mel_spectrogram","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_mel_spectrogram.html","id":"references","dir":"Reference","previous_headings":"","what":"References","title":"A preprocessing layer to convert raw audio signals to Mel spectrograms. — layer_mel_spectrogram","text":"Spectrogram, Mel scale.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_mel_spectrogram.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"A preprocessing layer to convert raw audio signals to Mel spectrograms. — layer_mel_spectrogram","text":"Unbatched audio signal   Batched audio signal","code":"layer <- layer_mel_spectrogram(   num_mel_bins = 64,   sampling_rate = 8000,   sequence_stride = 256,   fft_length = 2048 ) layer(random_uniform(shape = c(16000))) |> shape() layer <- layer_mel_spectrogram(   num_mel_bins = 80,   sampling_rate = 8000,   sequence_stride = 128,   fft_length = 2048 ) layer(random_uniform(shape = c(2, 16000))) |> shape()"},{"path":"https://keras3.posit.co/dev/reference/layer_mel_spectrogram.html","id":"input-shape","dir":"Reference","previous_headings":"","what":"Input Shape","title":"A preprocessing layer to convert raw audio signals to Mel spectrograms. — layer_mel_spectrogram","text":"1D (unbatched) 2D (batched) tensor shape:(..., samples).","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_mel_spectrogram.html","id":"output-shape","dir":"Reference","previous_headings":"","what":"Output Shape","title":"A preprocessing layer to convert raw audio signals to Mel spectrograms. — layer_mel_spectrogram","text":"2D (unbatched) 3D (batched) tensor shape:(..., num_mel_bins, time).","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_minimum.html","id":null,"dir":"Reference","previous_headings":"","what":"Computes elementwise minimum on a list of inputs. — layer_minimum","title":"Computes elementwise minimum on a list of inputs. — layer_minimum","text":"takes input list tensors, shape, returns single tensor (also shape).","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_minimum.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes elementwise minimum on a list of inputs. — layer_minimum","text":"","code":"layer_minimum(inputs, ...)"},{"path":"https://keras3.posit.co/dev/reference/layer_minimum.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Computes elementwise minimum on a list of inputs. — layer_minimum","text":"inputs layers combine ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_minimum.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Computes elementwise minimum on a list of inputs. — layer_minimum","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_minimum.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Computes elementwise minimum on a list of inputs. — layer_minimum","text":"Usage Keras model:","code":"input_shape <- c(2, 3, 4) x1 <- random_uniform(input_shape) x2 <- random_uniform(input_shape) y <- layer_minimum(x1, x2) input1 <- layer_input(shape = c(16)) x1 <- input1 |> layer_dense(8, activation = 'relu') input2 <- layer_input(shape = c(32)) x2 <- input2 |> layer_dense(8, activation = 'relu') # equivalent to `y <- layer_minimum(x1, x2)` y <- layer_minimum(x1, x2) out <- y |> layer_dense(4) model <- keras_model(inputs = c(input1, input2), outputs = out)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_mix_up.html","id":null,"dir":"Reference","previous_headings":"","what":"MixUp implements the MixUp data augmentation technique. — layer_mix_up","title":"MixUp implements the MixUp data augmentation technique. — layer_mix_up","text":"MixUp implements MixUp data augmentation technique.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_mix_up.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"MixUp implements the MixUp data augmentation technique. — layer_mix_up","text":"","code":"layer_mix_up(object, alpha = 0.2, data_format = NULL, seed = NULL, ...)"},{"path":"https://keras3.posit.co/dev/reference/layer_mix_up.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"MixUp implements the MixUp data augmentation technique. — layer_mix_up","text":"object Object compose layer . tensor, array, sequential model. alpha Float 0 1. Controls blending strength. Smaller values mean less mixing, larger values allow  blending images. Defaults 0.2, recommended ImageNet1k classification. data_format String, one \"channels_last\" (default) \"channels_first\". ordering dimensions inputs. \"channels_last\" corresponds inputs shape (batch, height, width, channels) \"channels_first\" corresponds inputs shape (batch, channels, height, width). seed Integer. Used create random seed. ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_mix_up.html","id":"references","dir":"Reference","previous_headings":"","what":"References","title":"MixUp implements the MixUp data augmentation technique. — layer_mix_up","text":"MixUp paper. MixUp Object Detection paper.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_mix_up.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"MixUp implements the MixUp data augmentation technique. — layer_mix_up","text":"","code":"c(c(images, labels), .) %<-% dataset_cifar10() c(images, labels) %<-% list(images[1:8,,,], labels[1:8,]) labels <- labels |> op_one_hot(10) |> op_cast(\"float32\") mix_up <- layer_mix_up(alpha=0.2) output <- mix_up(list(images = images, labels = labels))"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_multi_head_attention.html","id":null,"dir":"Reference","previous_headings":"","what":"Multi Head Attention layer. — layer_multi_head_attention","title":"Multi Head Attention layer. — layer_multi_head_attention","text":"implementation multi-headed attention described paper \"Attention Need\" Vaswani et al., 2017. query, key, value , self-attention. timestep query attends corresponding sequence key, returns fixed-width vector. layer first projects query, key value. (effectively) list tensors length num_attention_heads, corresponding shapes (batch_size, <query dimensions>, key_dim), (batch_size, <key/value dimensions>, key_dim), (batch_size, <key/value dimensions>, value_dim). , query key tensors dot-producted scaled. softmaxed obtain attention probabilities. value tensors interpolated probabilities, concatenated back single tensor. Finally, result tensor last dimension value_dim can take linear projection return.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_multi_head_attention.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Multi Head Attention layer. — layer_multi_head_attention","text":"","code":"layer_multi_head_attention(   inputs,   num_heads,   key_dim,   value_dim = NULL,   dropout = 0,   use_bias = TRUE,   output_shape = NULL,   attention_axes = NULL,   flash_attention = NULL,   kernel_initializer = \"glorot_uniform\",   bias_initializer = \"zeros\",   kernel_regularizer = NULL,   bias_regularizer = NULL,   activity_regularizer = NULL,   kernel_constraint = NULL,   bias_constraint = NULL,   seed = NULL,   ... )"},{"path":"https://keras3.posit.co/dev/reference/layer_multi_head_attention.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Multi Head Attention layer. — layer_multi_head_attention","text":"inputs see description num_heads Number attention heads. key_dim Size attention head query key. value_dim Size attention head value. dropout Dropout probability. use_bias Boolean, whether dense layers use bias vectors/matrices. output_shape expected shape output tensor, besides batch sequence dims. specified, projects back query feature dim (query input's last dimension). attention_axes axes attention applied. NULL means attention axes, batch, heads, features. flash_attention NULL, layer attempts use flash attention faster memory-efficient attention computations possible. behavior can configured using config_enable_flash_attention() config_disable_flash_attention(). kernel_initializer Initializer dense layer kernels. bias_initializer Initializer dense layer biases. kernel_regularizer Regularizer dense layer kernels. bias_regularizer Regularizer dense layer biases. activity_regularizer Regularizer dense layer activity. kernel_constraint Constraint dense layer kernels. bias_constraint Constraint dense layer kernels. seed Optional integer seed dropout layer. ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_multi_head_attention.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Multi Head Attention layer. — layer_multi_head_attention","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_multi_head_attention.html","id":"call-arguments","dir":"Reference","previous_headings":"","what":"Call Arguments","title":"Multi Head Attention layer. — layer_multi_head_attention","text":"query: Query tensor shape (B, T, dim), B batch size, T target sequence length, dim feature dimension. value: Value tensor shape (B, S, dim), B batch size, S source sequence length, dim feature dimension. key: Optional key tensor shape (B, S, dim). given, use value key value, common case. attention_mask: boolean mask shape (B, T, S), prevents attention certain positions. boolean mask specifies query elements can attend key elements, 1 indicates attention 0 indicates attention. Broadcasting can happen missing batch dimensions head dimension. return_attention_scores: boolean indicate whether output (attention_output, attention_scores) TRUE, attention_output FALSE. Defaults FALSE. training: Python boolean indicating whether layer behave training mode (adding dropout) inference mode (dropout). go either using training mode parent layer/model, FALSE (inference) parent layer. use_causal_mask: boolean indicate whether apply causal mask prevent tokens attending future tokens (e.g., used decoder Transformer).","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_multi_head_attention.html","id":"call-return","dir":"Reference","previous_headings":"","what":"Call return","title":"Multi Head Attention layer. — layer_multi_head_attention","text":"attention_output: result computation, shape (B, T, E), T target sequence shapes E query input last dimension output_shape NULL. Otherwise, multi-head outputs projected shape specified output_shape. attention_scores: (Optional) multi-head attention coefficients attention axes.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_multi_head_attention.html","id":"properties","dir":"Reference","previous_headings":"","what":"Properties","title":"Multi Head Attention layer. — layer_multi_head_attention","text":"MultiHeadAttention Layer instance following additional read-properties: attention_axes dropout key_dense key_dim num_heads output_dense output_shape query_dense use_bias value_dense value_dim","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_multiply.html","id":null,"dir":"Reference","previous_headings":"","what":"Performs elementwise multiplication. — layer_multiply","title":"Performs elementwise multiplication. — layer_multiply","text":"takes input list tensors, shape, returns single tensor (also shape).","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_multiply.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Performs elementwise multiplication. — layer_multiply","text":"","code":"layer_multiply(inputs, ...)"},{"path":"https://keras3.posit.co/dev/reference/layer_multiply.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Performs elementwise multiplication. — layer_multiply","text":"inputs layers combine ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_multiply.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Performs elementwise multiplication. — layer_multiply","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_multiply.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Performs elementwise multiplication. — layer_multiply","text":"Usage Keras model:","code":"input_shape <- c(2, 3, 4) x1 <- random_uniform(input_shape) x2 <- random_uniform(input_shape) y <- layer_multiply(x1, x2) input1 <- layer_input(shape = c(16)) x1 <- input1 |> layer_dense(8, activation = 'relu') input2 <- layer_input(shape = c(32)) x2 <- input2 |> layer_dense(8, activation = 'relu') # equivalent to `y <- layer_multiply(x1, x2)` y <- layer_multiply(x1, x2) out <- y |> layer_dense(4) model <- keras_model(inputs = c(input1, input2), outputs = out)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_normalization.html","id":null,"dir":"Reference","previous_headings":"","what":"A preprocessing layer that normalizes continuous features. — layer_normalization","title":"A preprocessing layer that normalizes continuous features. — layer_normalization","text":"layer shift scale inputs distribution centered around 0 standard deviation 1. accomplishes precomputing mean variance data, calling (input - mean) / sqrt(var) runtime. mean variance values layer must either supplied construction learned via adapt(). adapt() compute mean variance data store layer's weights. adapt() called fit(), evaluate(), predict().","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_normalization.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"A preprocessing layer that normalizes continuous features. — layer_normalization","text":"","code":"layer_normalization(   object,   axis = -1L,   mean = NULL,   variance = NULL,   invert = FALSE,   ... )"},{"path":"https://keras3.posit.co/dev/reference/layer_normalization.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"A preprocessing layer that normalizes continuous features. — layer_normalization","text":"object Object compose layer . tensor, array, sequential model. axis Integer, list integers, NULL. axis axes separate mean variance index shape. example, shape (NULL, 5) axis=1, layer track 5 separate mean variance values last axis. axis set NULL, layer normalize elements input scalar mean variance. -1, last axis input assumed feature dimension normalized per index. Note specific case batched scalar inputs axis batch axis, default normalize index batch separately. case, consider passing axis=NULL. Defaults -1. mean mean value(s) use normalization. passed value(s) broadcast shape kept axes ; value(s) broadcast, error raised layer's build() method called. variance variance value(s) use normalization. passed value(s) broadcast shape kept axes ; value(s) broadcast, error raised layer's build() method called. invert TRUE, layer apply inverse transformation inputs: turn normalized input back original form. ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_normalization.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"A preprocessing layer that normalizes continuous features. — layer_normalization","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_normalization.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"A preprocessing layer that normalizes continuous features. — layer_normalization","text":"Calculate global mean variance analyzing dataset adapt().     Calculate mean variance index last axis.     Pass mean variance directly.     Use layer de-normalize inputs (adapting layer).","code":"adapt_data <- op_array(c(1., 2., 3., 4., 5.), dtype='float32') input_data <- op_array(c(1., 2., 3.), dtype='float32') layer <- layer_normalization(axis = NULL) layer %>% adapt(adapt_data) layer(input_data) ## tf.Tensor([-1.4142135  -0.70710677  0.        ], shape=(3), dtype=float32) adapt_data <- op_array(rbind(c(0., 7., 4.),                        c(2., 9., 6.),                        c(0., 7., 4.),                        c(2., 9., 6.)), dtype='float32') input_data <- op_array(matrix(c(0., 7., 4.), nrow = 1), dtype='float32') layer <- layer_normalization(axis=-1) layer %>% adapt(adapt_data) layer(input_data) ## tf.Tensor([[-1. -1. -1.]], shape=(1, 3), dtype=float32) input_data <- op_array(rbind(1, 2, 3), dtype='float32') layer <- layer_normalization(mean=3., variance=2.) layer(input_data) ## tf.Tensor( ## [[-1.4142135 ] ##  [-0.70710677] ##  [ 0.        ]], shape=(3, 1), dtype=float32) adapt_data <- op_array(rbind(c(0., 7., 4.),                        c(2., 9., 6.),                        c(0., 7., 4.),                        c(2., 9., 6.)), dtype='float32') input_data <- op_array(c(1., 2., 3.), dtype='float32') layer <- layer_normalization(axis=-1, invert=TRUE) layer %>% adapt(adapt_data) layer(input_data) ## tf.Tensor([[ 2. 10.  8.]], shape=(1, 3), dtype=float32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_permute.html","id":null,"dir":"Reference","previous_headings":"","what":"Permutes the dimensions of the input according to a given pattern. — layer_permute","title":"Permutes the dimensions of the input according to a given pattern. — layer_permute","text":"Useful e.g. connecting RNNs convnets.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_permute.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Permutes the dimensions of the input according to a given pattern. — layer_permute","text":"","code":"layer_permute(object, dims, ...)"},{"path":"https://keras3.posit.co/dev/reference/layer_permute.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Permutes the dimensions of the input according to a given pattern. — layer_permute","text":"object Object compose layer . tensor, array, sequential model. dims List integers. Permutation pattern include batch dimension. Indexing starts 1. instance, (1, 3, 2) permutes second third dimensions input. ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_permute.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Permutes the dimensions of the input according to a given pattern. — layer_permute","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_permute.html","id":"input-shape","dir":"Reference","previous_headings":"","what":"Input Shape","title":"Permutes the dimensions of the input according to a given pattern. — layer_permute","text":"Arbitrary.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_permute.html","id":"output-shape","dir":"Reference","previous_headings":"","what":"Output Shape","title":"Permutes the dimensions of the input according to a given pattern. — layer_permute","text":"input shape, dimensions re-ordered according specified pattern.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_permute.html","id":"example","dir":"Reference","previous_headings":"","what":"Example","title":"Permutes the dimensions of the input according to a given pattern. — layer_permute","text":"","code":"x <- layer_input(shape=c(10, 64)) y <- layer_permute(x, c(2, 1)) shape(y) ## shape(NA, 64, 10)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_pipeline.html","id":null,"dir":"Reference","previous_headings":"","what":"Applies a series of layers to an input. — layer_pipeline","title":"Applies a series of layers to an input. — layer_pipeline","text":"class useful build preprocessing pipeline, particular image data augmentation pipeline. Compared Sequential model, Pipeline features important differences: Model, just plain layer. layers pipeline compatible tf.data, pipeline also remain tf.data compatible. say, pipeline attempt convert inputs backend-native tensors tf.data context (unlike Sequential model).","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_pipeline.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Applies a series of layers to an input. — layer_pipeline","text":"","code":"layer_pipeline(layers, name = NULL)"},{"path":"https://keras3.posit.co/dev/reference/layer_pipeline.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Applies a series of layers to an input. — layer_pipeline","text":"layers list layers. name String, name object","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_pipeline.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Applies a series of layers to an input. — layer_pipeline","text":"","code":"preprocessing_pipeline <- layer_pipeline(c(   layer_auto_contrast(, ),   layer_random_zoom(, 0.2),   layer_random_rotation(, 0.2) ))  # `ds` is a tf.data.Dataset of images ds <- tfdatasets::tensor_slices_dataset(1:100) |>   tfdatasets::dataset_map(\\(.x) {     random_normal(c(28, 28))   }) |>   tfdatasets::dataset_batch(32)   #|>   # tfdatasets::dataset_take(4) |>   # iterate() |> str()  preprocessed_ds <- ds |>   tfdatasets::dataset_map(preprocessing_pipeline, num_parallel_calls = 4)"},{"path":"https://keras3.posit.co/dev/reference/layer_rand_augment.html","id":null,"dir":"Reference","previous_headings":"","what":"RandAugment performs the Rand Augment operation on input images. — layer_rand_augment","title":"RandAugment performs the Rand Augment operation on input images. — layer_rand_augment","text":"layer can thought --one image augmentation layer. policy implemented layer benchmarked extensively effective wide variety datasets.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_rand_augment.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"RandAugment performs the Rand Augment operation on input images. — layer_rand_augment","text":"","code":"layer_rand_augment(   object,   value_range = list(0L, 255L),   num_ops = 2L,   factor = 0.5,   interpolation = \"bilinear\",   seed = NULL,   data_format = NULL,   ... )"},{"path":"https://keras3.posit.co/dev/reference/layer_rand_augment.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"RandAugment performs the Rand Augment operation on input images. — layer_rand_augment","text":"object Object compose layer . tensor, array, sequential model. value_range range values input image can take. Default (0, 255). Typically, (0, 1) normalized images (0, 255) raw images. num_ops number augmentation operations apply sequentially image. Default 2. factor strength augmentation normalized value 0 1. Default 0.5. interpolation interpolation method use resizing operations. Options include nearest, bilinear. Default bilinear. seed Integer. Used create random seed. data_format String, one \"channels_last\" (default) \"channels_first\". ordering dimensions inputs. \"channels_last\" corresponds inputs shape (batch, height, width, channels) \"channels_first\" corresponds inputs shape (batch, channels, height, width). ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_rand_augment.html","id":"references","dir":"Reference","previous_headings":"","what":"References","title":"RandAugment performs the Rand Augment operation on input images. — layer_rand_augment","text":"RandAugment","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_random_brightness.html","id":null,"dir":"Reference","previous_headings":"","what":"A preprocessing layer which randomly adjusts brightness during training. — layer_random_brightness","title":"A preprocessing layer which randomly adjusts brightness during training. — layer_random_brightness","text":"layer randomly increase/reduce brightness input RGB images. inference time, output identical input. Call layer training=TRUE adjust brightness input. Note: layer safe use inside tf.data pipeline (independently backend using).","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_random_brightness.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"A preprocessing layer which randomly adjusts brightness during training. — layer_random_brightness","text":"","code":"layer_random_brightness(   object,   factor,   value_range = list(0L, 255L),   seed = NULL,   ... )"},{"path":"https://keras3.posit.co/dev/reference/layer_random_brightness.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"A preprocessing layer which randomly adjusts brightness during training. — layer_random_brightness","text":"object Object compose layer . tensor, array, sequential model. factor Float list 2 floats -1.0 1.0. factor used determine lower bound upper bound brightness adjustment. float value chosen randomly limits. -1.0 chosen, output image black, 1.0 chosen, image fully white. one float provided, eg, 0.2, -0.2 used lower bound 0.2 used upper bound. value_range Optional list 2 floats lower upper limit values input data. make change, use c(0.0, 1.0), e.g., image input scaled layer. Defaults c(0.0, 255.0). brightness adjustment scaled range, output values clipped range. seed optional integer, fixed RNG behavior. ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_random_brightness.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"A preprocessing layer which randomly adjusts brightness during training. — layer_random_brightness","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_random_brightness.html","id":"inputs","dir":"Reference","previous_headings":"","what":"Inputs","title":"A preprocessing layer which randomly adjusts brightness during training. — layer_random_brightness","text":"3D (HWC) 4D (NHWC) tensor, float int dtype. Input pixel values can range (e.g. [0., 1.) [0, 255])","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_random_brightness.html","id":"output","dir":"Reference","previous_headings":"","what":"Output","title":"A preprocessing layer which randomly adjusts brightness during training. — layer_random_brightness","text":"3D (HWC) 4D (NHWC) tensor brightness adjusted based factor. default, layer output floats. output value clipped range [0, 255], valid range RGB colors, rescaled based value_range needed.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_random_brightness.html","id":"example","dir":"Reference","previous_headings":"","what":"Example","title":"A preprocessing layer which randomly adjusts brightness during training. — layer_random_brightness","text":"","code":"random_bright <- layer_random_brightness(factor=0.2, seed = 1)  # An image with shape [2, 2, 3] image <- array(1:12, dim=c(2, 2, 3))  # Assume we randomly select the factor to be 0.1, then it will apply # 0.1 * 255 to all the channel output <- random_bright(image, training=TRUE) output ## tf.Tensor( ## [[[39 43 47] ##   [41 45 49]] ## ##  [[40 44 48] ##   [42 46 50]]], shape=(2, 2, 3), dtype=int32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_random_color_degeneration.html","id":null,"dir":"Reference","previous_headings":"","what":"Randomly performs the color degeneration operation on given images. — layer_random_color_degeneration","title":"Randomly performs the color degeneration operation on given images. — layer_random_color_degeneration","text":"sharpness operation first converts image gray scale, back color. takes weighted average original image degenerated image. makes colors appear dull.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_random_color_degeneration.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Randomly performs the color degeneration operation on given images. — layer_random_color_degeneration","text":"","code":"layer_random_color_degeneration(   object,   factor,   value_range = list(0L, 255L),   data_format = NULL,   seed = NULL,   ... )"},{"path":"https://keras3.posit.co/dev/reference/layer_random_color_degeneration.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Randomly performs the color degeneration operation on given images. — layer_random_color_degeneration","text":"object Object compose layer . tensor, array, sequential model. factor tuple two floats single float. factor controls extent image sharpness impacted. factor=0.0 makes layer perform -op operation, value 1.0 uses degenerated result entirely. Values 0 1 result linear interpolation original image sharpened image. Values 0.0 1.0. tuple used, factor sampled two values every image augmented. single float used, value 0.0 passed float sampled. order ensure value always , please pass tuple two identical floats: (0.5, 0.5). value_range range values input image can take. Default (0, 255). Typically, (0, 1) normalized images (0, 255) raw images. data_format String, one \"channels_last\" (default) \"channels_first\". ordering dimensions inputs. \"channels_last\" corresponds inputs shape (batch, height, width, channels) \"channels_first\" corresponds inputs shape (batch, channels, height, width). seed Integer. Used create random seed. ... forward/backward compatability.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_random_color_jitter.html","id":null,"dir":"Reference","previous_headings":"","what":"Randomly apply brightness, contrast, saturation — layer_random_color_jitter","title":"Randomly apply brightness, contrast, saturation — layer_random_color_jitter","text":"hue image processing operation sequentially randomly input.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_random_color_jitter.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Randomly apply brightness, contrast, saturation — layer_random_color_jitter","text":"","code":"layer_random_color_jitter(   object,   value_range = list(0L, 255L),   brightness_factor = NULL,   contrast_factor = NULL,   saturation_factor = NULL,   hue_factor = NULL,   seed = NULL,   data_format = NULL,   ... )"},{"path":"https://keras3.posit.co/dev/reference/layer_random_color_jitter.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Randomly apply brightness, contrast, saturation — layer_random_color_jitter","text":"object Object compose layer . tensor, array, sequential model. value_range range values incoming images . Represented two number tuple written [low, high]. typically either [0, 1] [0, 255] depending preprocessing pipeline set . brightness_factor Float list/tuple 2 floats -1.0 1.0. factor used determine lower bound upper bound brightness adjustment. float value chosen randomly limits. -1.0 chosen, output image black, 1.0 chosen, image fully white. one float provided, eg, 0.2, -0.2 used lower bound 0.2 used upper bound. contrast_factor positive float represented fraction value, tuple size 2 representing lower upper bound. represented single float, lower = upper. contrast factor randomly picked [1.0 - lower, 1.0 + upper]. pixel x channel, output (x - mean) * factor + mean mean mean value channel. saturation_factor tuple two floats single float. factor controls extent image saturation impacted. factor=0.5 makes layer perform -op operation. factor=0.0 makes image fully grayscale. factor=1.0 makes image fully saturated. Values 0.0 1.0. tuple used, factor sampled two values every image augmented. single float used, value 0.0 passed float sampled. ensure value always , pass tuple two identical floats: (0.5, 0.5). hue_factor single float tuple two floats. factor controls extent image hue impacted. factor=0.0 makes layer perform -op operation, value 1.0 performs aggressive contrast adjustment available. tuple used, factor sampled two values every image augmented. single float used, value 0.0 passed float sampled. order ensure value always , please pass tuple two identical floats: (0.5, 0.5). seed Integer. Used create random seed. data_format String, one \"channels_last\" (default) \"channels_first\". ordering dimensions inputs. \"channels_last\" corresponds inputs shape (batch, height, width, channels) \"channels_first\" corresponds inputs shape (batch, channels, height, width). ... forward/backward compatability.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_random_contrast.html","id":null,"dir":"Reference","previous_headings":"","what":"A preprocessing layer which randomly adjusts contrast during training. — layer_random_contrast","title":"A preprocessing layer which randomly adjusts contrast during training. — layer_random_contrast","text":"layer randomly adjust contrast image images random factor. Contrast adjusted independently channel image training. channel, layer computes mean image pixels channel adjusts component x pixel (x - mean) * contrast_factor + mean. Input pixel values can range (e.g. [0., 1.) [0, 255]) integer floating point dtype. default, layer output floats. Note: layer safe use inside tf.data pipeline (independently backend using).","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_random_contrast.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"A preprocessing layer which randomly adjusts contrast during training. — layer_random_contrast","text":"","code":"layer_random_contrast(   object,   factor,   value_range = c(0L, 255L),   seed = NULL,   ... )"},{"path":"https://keras3.posit.co/dev/reference/layer_random_contrast.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"A preprocessing layer which randomly adjusts contrast during training. — layer_random_contrast","text":"object Object compose layer . tensor, array, sequential model. factor positive float represented fraction value, tuple size 2 representing lower upper bound. represented single float, lower = upper. contrast factor randomly picked [1.0 - lower, 1.0 + upper]. pixel x channel, output (x - mean) * factor + mean mean mean value channel. value_range range values incoming images . Represented two-number tuple written tuple(low, high). typically either [0, 1] [0, 255] depending preprocessing pipeline set . seed Integer. Used create random seed. ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_random_contrast.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"A preprocessing layer which randomly adjusts contrast during training. — layer_random_contrast","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_random_contrast.html","id":"input-shape","dir":"Reference","previous_headings":"","what":"Input Shape","title":"A preprocessing layer which randomly adjusts contrast during training. — layer_random_contrast","text":"3D (unbatched) 4D (batched) tensor shape: (..., height, width, channels), \"channels_last\" format.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_random_contrast.html","id":"output-shape","dir":"Reference","previous_headings":"","what":"Output Shape","title":"A preprocessing layer which randomly adjusts contrast during training. — layer_random_contrast","text":"3D (unbatched) 4D (batched) tensor shape: (..., height, width, channels), \"channels_last\" format.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_random_crop.html","id":null,"dir":"Reference","previous_headings":"","what":"A preprocessing layer which randomly crops images during training. — layer_random_crop","title":"A preprocessing layer which randomly crops images during training. — layer_random_crop","text":"training, layer randomly choose location crop images target size. layer crop images batch cropping location. inference time, training input image smaller target size, input resized cropped return largest possible window image matches target aspect ratio. need apply random cropping inference time, set training TRUE calling layer. Input pixel values can range (e.g. [0., 1.) [0, 255]) integer floating point dtype. default, layer output floats. Note: layer safe use inside tf.data pipeline (independently backend using).","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_random_crop.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"A preprocessing layer which randomly crops images during training. — layer_random_crop","text":"","code":"layer_random_crop(   object,   height,   width,   seed = NULL,   data_format = NULL,   name = NULL,   ... )"},{"path":"https://keras3.posit.co/dev/reference/layer_random_crop.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"A preprocessing layer which randomly crops images during training. — layer_random_crop","text":"object Object compose layer . tensor, array, sequential model. height Integer, height output shape. width Integer, width output shape. seed Integer. Used create random seed. data_format string, either \"channels_last\" \"channels_first\". ordering dimensions inputs. \"channels_last\" corresponds inputs shape (batch, height, width, channels) \"channels_first\" corresponds inputs shape (batch, channels, height, width). defaults image_data_format value found Keras config file ~/.keras/keras.json. never set , \"channels_last\". name String, name object ... Base layer keyword arguments, name dtype.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_random_crop.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"A preprocessing layer which randomly crops images during training. — layer_random_crop","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_random_crop.html","id":"input-shape","dir":"Reference","previous_headings":"","what":"Input Shape","title":"A preprocessing layer which randomly crops images during training. — layer_random_crop","text":"3D (unbatched) 4D (batched) tensor shape: (..., height, width, channels), \"channels_last\" format.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_random_crop.html","id":"output-shape","dir":"Reference","previous_headings":"","what":"Output Shape","title":"A preprocessing layer which randomly crops images during training. — layer_random_crop","text":"3D (unbatched) 4D (batched) tensor shape: (..., target_height, target_width, channels).","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_random_elastic_transform.html","id":null,"dir":"Reference","previous_headings":"","what":"A preprocessing layer that applies random elastic transformations. — layer_random_elastic_transform","title":"A preprocessing layer that applies random elastic transformations. — layer_random_elastic_transform","text":"layer distorts input images applying elastic deformations, simulating physically realistic transformation. magnitude distortion controlled scale parameter, factor determines probability applying transformation.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_random_elastic_transform.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"A preprocessing layer that applies random elastic transformations. — layer_random_elastic_transform","text":"","code":"layer_random_elastic_transform(   object,   factor = 1,   scale = 1,   interpolation = \"bilinear\",   fill_mode = \"reflect\",   fill_value = 0,   value_range = list(0L, 255L),   seed = NULL,   data_format = NULL,   ... )"},{"path":"https://keras3.posit.co/dev/reference/layer_random_elastic_transform.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"A preprocessing layer that applies random elastic transformations. — layer_random_elastic_transform","text":"object Object compose layer . tensor, array, sequential model. factor single float tuple two floats. factor controls probability applying transformation. factor = 0.0 ensures transformation applied. factor = 1.0 means transformation always applied. tuple (min, max) provided, probability value sampled min max image. single float provided, probability sampled 0.0 given float. Default 1.0. scale float tuple two floats defining magnitude distortion applied. tuple (min, max) provided, random scale value sampled within range. single float provided, random scale value sampled 0.0 given float. Default 1.0. interpolation Interpolation mode. Supported values: \"nearest\", \"bilinear\". fill_mode Points outside boundaries input filled according given mode. Available methods \"constant\", \"nearest\", \"wrap\" \"reflect\". Defaults \"reflect\". \"reflect\": (d c b | b c d | d c b ) input extended reflecting edge last pixel. \"constant\": (k k k k | b c d | k k k k) input extended filling values beyond edge constant value k specified fill_value. \"wrap\": (b c d | b c d | b c d) input extended wrapping around opposite edge. \"nearest\": (| b c d | d d d d) input extended nearest pixel. using torch backend, \"reflect\" redirected \"mirror\" torch support \"reflect\". torch backend also support \"wrap\". fill_value float representing value fill outside boundaries fill_mode = \"constant\". value_range range values incoming images . Represented two-number tuple written [low, high]. typically either [0, 1] [0, 255] depending preprocessing pipeline set . seed Integer. Used create random seed. data_format string, either \"channels_last\" \"channels_first\". ordering dimensions inputs. \"channels_last\" corresponds inputs shape (batch, height, width, channels) \"channels_first\" corresponds inputs shape (batch, channels, height, width). defaults image_data_format value found Keras config file ~/.keras/keras.json. never set , \"channels_last\". ... forward/backward compatability.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_random_erasing.html","id":null,"dir":"Reference","previous_headings":"","what":"Random Erasing data augmentation technique. — layer_random_erasing","title":"Random Erasing data augmentation technique. — layer_random_erasing","text":"Random Erasing data augmentation method random patches image erased (replaced constant value noise) training improve generalization.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_random_erasing.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Random Erasing data augmentation technique. — layer_random_erasing","text":"","code":"layer_random_erasing(   object,   factor = 1,   scale = list(0.02, 0.33),   fill_value = NULL,   value_range = list(0L, 255L),   seed = NULL,   data_format = NULL,   ... )"},{"path":"https://keras3.posit.co/dev/reference/layer_random_erasing.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Random Erasing data augmentation technique. — layer_random_erasing","text":"object Object compose layer . tensor, array, sequential model. factor single float tuple two floats. factor controls probability applying transformation. factor=0.0 ensures erasing applied. factor=1.0 means erasing always applied. tuple (min, max) provided, probability value sampled min max image. single float provided, probability sampled 0.0 given float. Default 1.0. scale tuple two floats representing aspect ratio range erased patch. defines width--height ratio patch erased. can help control rw shape erased region. Default (0.02, 0.33). fill_value value fill erased region . can set constant value None sample random value normal distribution. Default None. value_range range values incoming images . Represented two-number tuple written [low, high]. typically either [0, 1] [0, 255] depending preprocessing pipeline set . seed Integer. Used create random seed. data_format string, either \"channels_last\" \"channels_first\". ordering dimensions inputs. \"channels_last\" corresponds inputs shape (batch, height, width, channels) \"channels_first\" corresponds inputs shape (batch, channels, height, width). defaults image_data_format value found Keras config file ~/.keras/keras.json. never set , \"channels_last\". ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_random_erasing.html","id":"references","dir":"Reference","previous_headings":"","what":"References","title":"Random Erasing data augmentation technique. — layer_random_erasing","text":"Random Erasing paper.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_random_flip.html","id":null,"dir":"Reference","previous_headings":"","what":"A preprocessing layer which randomly flips images during training. — layer_random_flip","title":"A preprocessing layer which randomly flips images during training. — layer_random_flip","text":"layer flip images horizontally vertically based mode attribute. inference time, output identical input. Call layer training=TRUE flip input. Input pixel values can range (e.g. [0., 1.) [0, 255]) integer floating point dtype. default, layer output floats. Note: layer safe use inside tf.data pipeline (independently backend using).","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_random_flip.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"A preprocessing layer which randomly flips images during training. — layer_random_flip","text":"","code":"layer_random_flip(   object,   mode = \"horizontal_and_vertical\",   seed = NULL,   data_format = NULL,   ... )"},{"path":"https://keras3.posit.co/dev/reference/layer_random_flip.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"A preprocessing layer which randomly flips images during training. — layer_random_flip","text":"object Object compose layer . tensor, array, sequential model. mode String indicating flip mode use. Can \"horizontal\", \"vertical\", \"horizontal_and_vertical\". \"horizontal\" left-right flip \"vertical\" top-bottom flip. Defaults \"horizontal_and_vertical\" seed Integer. Used create random seed. data_format string, either \"channels_last\" \"channels_first\". ordering dimensions inputs. \"channels_last\" corresponds inputs shape (batch, height, width, channels) \"channels_first\" corresponds inputs shape (batch, channels, height, width). defaults image_data_format value found Keras config file ~/.keras/keras.json. never set , \"channels_last\". ... Base layer keyword arguments, name dtype.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_random_flip.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"A preprocessing layer which randomly flips images during training. — layer_random_flip","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_random_flip.html","id":"input-shape","dir":"Reference","previous_headings":"","what":"Input Shape","title":"A preprocessing layer which randomly flips images during training. — layer_random_flip","text":"3D (unbatched) 4D (batched) tensor shape: (..., height, width, channels), \"channels_last\" format.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_random_flip.html","id":"output-shape","dir":"Reference","previous_headings":"","what":"Output Shape","title":"A preprocessing layer which randomly flips images during training. — layer_random_flip","text":"3D (unbatched) 4D (batched) tensor shape: (..., height, width, channels), \"channels_last\" format.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_random_gaussian_blur.html","id":null,"dir":"Reference","previous_headings":"","what":"Applies random Gaussian blur to images for data augmentation. — layer_random_gaussian_blur","title":"Applies random Gaussian blur to images for data augmentation. — layer_random_gaussian_blur","text":"layer performs Gaussian blur operation input images randomly selected degree blurring, controlled factor sigma arguments.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_random_gaussian_blur.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Applies random Gaussian blur to images for data augmentation. — layer_random_gaussian_blur","text":"","code":"layer_random_gaussian_blur(   object,   factor = 1,   kernel_size = 3L,   sigma = 1,   value_range = list(0L, 255L),   data_format = NULL,   seed = NULL,   ... )"},{"path":"https://keras3.posit.co/dev/reference/layer_random_gaussian_blur.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Applies random Gaussian blur to images for data augmentation. — layer_random_gaussian_blur","text":"object Object compose layer . tensor, array, sequential model. factor single float tuple two floats. factor controls extent image hue impacted. factor=0.0 makes layer perform -op operation, value 1.0 performs aggressive blurring available. tuple used, factor sampled two values every image augmented. single float used, value 0.0 passed float sampled. Default 1.0. kernel_size Integer. Size Gaussian kernel used blurring. Must odd integer. Default 3. sigma Float tuple two floats. Standard deviation Gaussian kernel. Controls intensity blur. tuple provided, value sampled two image. Default 1.0. value_range range values incoming images . Represented two-number tuple written [low, high]. typically either [0, 1] [0, 255] depending preprocessing pipeline set . data_format string, either \"channels_last\" \"channels_first\". ordering dimensions inputs. \"channels_last\" corresponds inputs shape (batch, height, width, channels) \"channels_first\" corresponds inputs shape (batch, channels, height, width). defaults image_data_format value found Keras config file ~/.keras/keras.json. never set , \"channels_last\". seed Integer. Used create random seed. ... forward/backward compatability.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_random_grayscale.html","id":null,"dir":"Reference","previous_headings":"","what":"Preprocessing layer for random conversion of RGB images to grayscale. — layer_random_grayscale","title":"Preprocessing layer for random conversion of RGB images to grayscale. — layer_random_grayscale","text":"layer randomly converts input images grayscale specified factor. applied, maintains original number channels sets channels grayscale value. can useful data augmentation training models robust color variations. conversion preserves perceived luminance original color image using standard RGB grayscale conversion coefficients. Images selected conversion remain unchanged. Note: layer safe use inside tf.data pipeline (independently backend using).","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_random_grayscale.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Preprocessing layer for random conversion of RGB images to grayscale. — layer_random_grayscale","text":"","code":"layer_random_grayscale(   object,   factor = 0.5,   data_format = NULL,   seed = NULL,   ... )"},{"path":"https://keras3.posit.co/dev/reference/layer_random_grayscale.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Preprocessing layer for random conversion of RGB images to grayscale. — layer_random_grayscale","text":"object Object compose layer . tensor, array, sequential model. factor Float 0 1, specifying factor converting image grayscale. Defaults 0.5. value 1.0 means images converted, 0.0 means images converted. data_format String, one \"channels_last\" (default) \"channels_first\". ordering dimensions inputs. \"channels_last\" corresponds inputs shape (batch, height, width, channels) \"channels_first\" corresponds inputs shape (batch, channels, height, width). seed Initial seed random number generator ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_random_grayscale.html","id":"input-shape","dir":"Reference","previous_headings":"","what":"Input Shape","title":"Preprocessing layer for random conversion of RGB images to grayscale. — layer_random_grayscale","text":"3D (unbatched) 4D (batched) tensor shape: (..., height, width, channels), \"channels_last\" format, (..., channels, height, width), \"channels_first\" format.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_random_grayscale.html","id":"output-shape","dir":"Reference","previous_headings":"","what":"Output Shape","title":"Preprocessing layer for random conversion of RGB images to grayscale. — layer_random_grayscale","text":"input shape. output maintains number channels input, even grayscale-converted images channels value.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_random_hue.html","id":null,"dir":"Reference","previous_headings":"","what":"Randomly adjusts the hue on given images. — layer_random_hue","title":"Randomly adjusts the hue on given images. — layer_random_hue","text":"layer randomly increase/reduce hue input RGB images. image hue adjusted converting image(s) HSV rotating hue channel (H) delta. image converted back RGB.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_random_hue.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Randomly adjusts the hue on given images. — layer_random_hue","text":"","code":"layer_random_hue(   object,   factor,   value_range = list(0L, 255L),   data_format = NULL,   seed = NULL,   ... )"},{"path":"https://keras3.posit.co/dev/reference/layer_random_hue.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Randomly adjusts the hue on given images. — layer_random_hue","text":"object Object compose layer . tensor, array, sequential model. factor single float tuple two floats. factor controls extent image hue impacted. factor=0.0 makes layer perform -op operation, value 1.0 performs aggressive contrast adjustment available. tuple used, factor sampled two values every image augmented. single float used, value 0.0 passed float sampled. order ensure value always , please pass tuple two identical floats: (0.5, 0.5). value_range range values incoming images . Represented two-number tuple written [low, high]. typically either [0, 1] [0, 255] depending preprocessing pipeline set . data_format String, one \"channels_last\" (default) \"channels_first\". ordering dimensions inputs. \"channels_last\" corresponds inputs shape (batch, height, width, channels) \"channels_first\" corresponds inputs shape (batch, channels, height, width). seed Integer. Used create random seed. ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_random_hue.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Randomly adjusts the hue on given images. — layer_random_hue","text":"","code":"c(c(images, labels), .) %<-% dataset_cifar10() random_hue <- layer_random_hue(factor=0.5, value_range=c(0, 1)) images <- op_cast(images[1:8,,,], \"float32\") augmented_images_batch = random_hue(images)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_random_invert.html","id":null,"dir":"Reference","previous_headings":"","what":"Preprocessing layer for random inversion of image colors. — layer_random_invert","title":"Preprocessing layer for random inversion of image colors. — layer_random_invert","text":"layer randomly inverts colors input images specified probability range. applied, image chance colors inverted, pixel values transformed complementary values. Images selected inversion remain unchanged.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_random_invert.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Preprocessing layer for random inversion of image colors. — layer_random_invert","text":"","code":"layer_random_invert(   object,   factor = 1,   value_range = list(0L, 255L),   seed = NULL,   data_format = NULL,   ... )"},{"path":"https://keras3.posit.co/dev/reference/layer_random_invert.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Preprocessing layer for random inversion of image colors. — layer_random_invert","text":"object Object compose layer . tensor, array, sequential model. factor single float tuple two floats. factor controls probability inverting image colors. tuple provided, value sampled two values image, factor[0] minimum factor[1] maximum probability. single float provided, value 0.0 provided float sampled. Defaults (0, 1). value_range tuple list two elements. first value represents lower bound values passed images, second represents upper bound. Images passed layer values within value_range. Defaults (0, 255). seed Integer. Used create random seed. data_format string, either \"channels_last\" \"channels_first\". ordering dimensions inputs. \"channels_last\" corresponds inputs shape (batch, height, width, channels) \"channels_first\" corresponds inputs shape (batch, channels, height, width). defaults image_data_format value found Keras config file ~/.keras/keras.json. never set , \"channels_last\". ... forward/backward compatability.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_random_perspective.html","id":null,"dir":"Reference","previous_headings":"","what":"A preprocessing layer that applies random perspective transformations. — layer_random_perspective","title":"A preprocessing layer that applies random perspective transformations. — layer_random_perspective","text":"layer distorts perspective input images shifting corner points, simulating 3D-like transformation. amount distortion controlled factor scale parameters.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_random_perspective.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"A preprocessing layer that applies random perspective transformations. — layer_random_perspective","text":"","code":"layer_random_perspective(   object,   factor = 1,   scale = 1,   interpolation = \"bilinear\",   fill_value = 0,   seed = NULL,   data_format = NULL,   ... )"},{"path":"https://keras3.posit.co/dev/reference/layer_random_perspective.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"A preprocessing layer that applies random perspective transformations. — layer_random_perspective","text":"object Object compose layer . tensor, array, sequential model. factor float tuple two floats. Represents probability applying perspective transformation image batch. factor=0.0 ensures transformation applied. factor=1.0 means transformation always applied. tuple (min, max) provided, probability randomly sampled min max image. single float given, probability sampled 0.0 provided float. Default 1.0. scale float defining relative amount perspective shift. Determines much image corners displaced, affecting intensity perspective effect. interpolation Interpolation mode. Supported values: \"nearest\", \"bilinear\". fill_value float represents value filled outside boundaries fill_mode=\"constant\". seed Integer. Used create random seed. data_format string, either \"channels_last\" \"channels_first\". ordering dimensions inputs. \"channels_last\" corresponds inputs shape (batch, height, width, channels) \"channels_first\" corresponds inputs shape (batch, channels, height, width). defaults image_data_format value found Keras config file ~/.keras/keras.json. never set , \"channels_last\". ... forward/backward compatability.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_random_posterization.html","id":null,"dir":"Reference","previous_headings":"","what":"Reduces the number of bits for each color channel. — layer_random_posterization","title":"Reduces the number of bits for each color channel. — layer_random_posterization","text":"Reduces number bits color channel.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_random_posterization.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Reduces the number of bits for each color channel. — layer_random_posterization","text":"","code":"layer_random_posterization(   object,   factor,   value_range = list(0L, 255L),   data_format = NULL,   seed = NULL,   ... )"},{"path":"https://keras3.posit.co/dev/reference/layer_random_posterization.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Reduces the number of bits for each color channel. — layer_random_posterization","text":"object Object compose layer . tensor, array, sequential model. factor integer, number bits keep channel. Must value 1-8. value_range tuple list two elements. first value represents lower bound values passed images, second represents upper bound. Images passed layer values within value_range. Defaults (0, 255). data_format String, one \"channels_last\" (default) \"channels_first\". ordering dimensions inputs. \"channels_last\" corresponds inputs shape (batch, height, width, channels) \"channels_first\" corresponds inputs shape (batch, channels, height, width). seed Initial seed random number generator ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_random_posterization.html","id":"references","dir":"Reference","previous_headings":"","what":"References","title":"Reduces the number of bits for each color channel. — layer_random_posterization","text":"AutoAugment: Learning Augmentation Policies Data RandAugment: Practical automated data augmentation reduced search space","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_random_rotation.html","id":null,"dir":"Reference","previous_headings":"","what":"A preprocessing layer which randomly rotates images during training. — layer_random_rotation","title":"A preprocessing layer which randomly rotates images during training. — layer_random_rotation","text":"layer apply random rotations image, filling empty space according fill_mode. default, random rotations applied training. inference time, layer nothing. need apply random rotations inference time, pass training = TRUE calling layer. Input pixel values can range (e.g. [0., 1.) [0, 255]) integer floating point dtype. default, layer output floats. Note: layer safe use inside tf.data pipeline (independently backend using).","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_random_rotation.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"A preprocessing layer which randomly rotates images during training. — layer_random_rotation","text":"","code":"layer_random_rotation(   object,   factor,   fill_mode = \"reflect\",   interpolation = \"bilinear\",   seed = NULL,   fill_value = 0,   data_format = NULL,   ... )"},{"path":"https://keras3.posit.co/dev/reference/layer_random_rotation.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"A preprocessing layer which randomly rotates images during training. — layer_random_rotation","text":"object Object compose layer . tensor, array, sequential model. factor float represented fraction 2 Pi, tuple size 2 representing lower upper bound rotating clockwise counter-clockwise. positive values means rotating counter clock-wise, negative value means clock-wise. represented single float, value used upper lower bound. instance, factor=c(-0.2, 0.3) results output rotation random amount range [-20% * 360, 30% * 360]. factor=0.2 results output rotating random amount range [-20% * 360, 20% * 360]. fill_mode Points outside boundaries input filled according given mode (one {\"constant\", \"reflect\", \"wrap\", \"nearest\"}). reflect: (d c b | b c d | d c b ) input extended reflecting edge last pixel. constant: (k k k k | b c d | k k k k) input extended filling values beyond edge constant value k = 0. wrap: (b c d | b c d | b c d) input extended wrapping around opposite edge. nearest: (| b c d | d d d d) input extended nearest pixel. interpolation Interpolation mode. Supported values: \"nearest\", \"bilinear\". seed Integer. Used create random seed. fill_value float represents value filled outside boundaries fill_mode=\"constant\". data_format string, either \"channels_last\" \"channels_first\". ordering dimensions inputs. \"channels_last\" corresponds inputs shape (batch, height, width, channels) \"channels_first\" corresponds inputs shape (batch, channels, height, width). defaults image_data_format value found Keras config file ~/.keras/keras.json. never set , \"channels_last\". ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_random_rotation.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"A preprocessing layer which randomly rotates images during training. — layer_random_rotation","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_random_rotation.html","id":"input-shape","dir":"Reference","previous_headings":"","what":"Input Shape","title":"A preprocessing layer which randomly rotates images during training. — layer_random_rotation","text":"3D (unbatched) 4D (batched) tensor shape: (..., height, width, channels), \"channels_last\" format","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_random_rotation.html","id":"output-shape","dir":"Reference","previous_headings":"","what":"Output Shape","title":"A preprocessing layer which randomly rotates images during training. — layer_random_rotation","text":"3D (unbatched) 4D (batched) tensor shape: (..., height, width, channels), \"channels_last\" format","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_random_saturation.html","id":null,"dir":"Reference","previous_headings":"","what":"Randomly adjusts the saturation on given images. — layer_random_saturation","title":"Randomly adjusts the saturation on given images. — layer_random_saturation","text":"layer randomly increase/reduce saturation input RGB images.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_random_saturation.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Randomly adjusts the saturation on given images. — layer_random_saturation","text":"","code":"layer_random_saturation(   object,   factor,   value_range = list(0L, 255L),   data_format = NULL,   seed = NULL,   ... )"},{"path":"https://keras3.posit.co/dev/reference/layer_random_saturation.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Randomly adjusts the saturation on given images. — layer_random_saturation","text":"object Object compose layer . tensor, array, sequential model. factor tuple two floats single float. factor controls extent image saturation impacted. factor=0.5 makes layer perform -op operation. factor=0.0 makes image fully grayscale. factor=1.0 makes image fully saturated. Values 0.0 1.0. tuple used, factor sampled two values every image augmented. single float used, value 0.0 passed float sampled. ensure value always , pass tuple two identical floats: (0.5, 0.5). value_range range values incoming images . Represented two-number tuple written [low, high]. typically either [0, 1] [0, 255] depending preprocessing pipeline set . data_format String, one \"channels_last\" (default) \"channels_first\". ordering dimensions inputs. \"channels_last\" corresponds inputs shape (batch, height, width, channels) \"channels_first\" corresponds inputs shape (batch, channels, height, width). seed Integer. Used create random seed. ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_random_saturation.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Randomly adjusts the saturation on given images. — layer_random_saturation","text":"","code":"c(c(images, labels), .) %<-% dataset_cifar10() images <- images[1:8, , , ] |> op_cast(\"float32\") random_saturation <- layer_random_saturation(factor = 0.2) augmented_images <- random_saturation(images)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_random_sharpness.html","id":null,"dir":"Reference","previous_headings":"","what":"Randomly performs the sharpness operation on given images. — layer_random_sharpness","title":"Randomly performs the sharpness operation on given images. — layer_random_sharpness","text":"sharpness operation first performs blur, blends original image processed image. operation adjusts clarity edges image, ranging blurred enhanced sharpness.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_random_sharpness.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Randomly performs the sharpness operation on given images. — layer_random_sharpness","text":"","code":"layer_random_sharpness(   object,   factor,   value_range = list(0L, 255L),   data_format = NULL,   seed = NULL,   ... )"},{"path":"https://keras3.posit.co/dev/reference/layer_random_sharpness.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Randomly performs the sharpness operation on given images. — layer_random_sharpness","text":"object Object compose layer . tensor, array, sequential model. factor tuple two floats single float. factor controls extent image sharpness impacted. factor=0.0 results fully blurred image, factor=0.5 applies operation (preserving original image), factor=1.0 enhances sharpness beyond original. Values 0.0 1.0. tuple used, factor sampled two values every image augmented. single float used, value 0.0 passed float sampled. ensure value always , pass tuple two identical floats: (0.5, 0.5). value_range range values incoming images . Represented two-number tuple written [low, high]. typically either [0, 1] [0, 255] depending preprocessing pipeline set . data_format String, one \"channels_last\" (default) \"channels_first\". ordering dimensions inputs. \"channels_last\" corresponds inputs shape (batch, height, width, channels) \"channels_first\" corresponds inputs shape (batch, channels, height, width). seed Integer. Used create random seed. ... forward/backward compatability.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_random_shear.html","id":null,"dir":"Reference","previous_headings":"","what":"A preprocessing layer that randomly applies shear transformations — layer_random_shear","title":"A preprocessing layer that randomly applies shear transformations — layer_random_shear","text":"images. layer shears input images along x-axis /y-axis randomly selected factor within specified range. shear transformation applied image independently batch. Empty regions created transformation filled according fill_mode fill_value parameters.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_random_shear.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"A preprocessing layer that randomly applies shear transformations — layer_random_shear","text":"","code":"layer_random_shear(   object,   x_factor = 0,   y_factor = 0,   interpolation = \"bilinear\",   fill_mode = \"reflect\",   fill_value = 0,   data_format = NULL,   seed = NULL,   ... )"},{"path":"https://keras3.posit.co/dev/reference/layer_random_shear.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"A preprocessing layer that randomly applies shear transformations — layer_random_shear","text":"object Object compose layer . tensor, array, sequential model. x_factor tuple two floats. augmented image, value sampled provided range. float passed, range interpreted (0, x_factor). Values represent percentage image shear . example, 0.3 shears pixels 30% way across image. provided values positive. y_factor tuple two floats. augmented image, value sampled provided range. float passed, range interpreted (0, y_factor). Values represent percentage image shear . example, 0.3 shears pixels 30% way across image. provided values positive. interpolation Interpolation mode. Supported values: \"nearest\", \"bilinear\". fill_mode Points outside boundaries input filled according given mode. Available methods \"constant\", \"nearest\", \"wrap\" \"reflect\". Defaults \"reflect\". \"reflect\": (d c b | b c d | d c b ) input extended reflecting edge last pixel. \"constant\": (k k k k | b c d | k k k k) input extended filling values beyond edge constant value k specified fill_value. \"wrap\": (b c d | b c d | b c d) input extended wrapping around opposite edge. \"nearest\": (| b c d | d d d d) input extended nearest pixel. Note using torch backend, \"reflect\" redirected \"mirror\" (c d c b | b c d | c b b) torch support \"reflect\". Note torch backend support \"wrap\". fill_value float representing value filled outside boundaries fill_mode=\"constant\". data_format String, one \"channels_last\" (default) \"channels_first\". ordering dimensions inputs. \"channels_last\" corresponds inputs shape (batch, height, width, channels) \"channels_first\" corresponds inputs shape (batch, channels, height, width). seed Integer. Used create random seed. ... forward/backward compatability.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_random_translation.html","id":null,"dir":"Reference","previous_headings":"","what":"A preprocessing layer which randomly translates images during training. — layer_random_translation","title":"A preprocessing layer which randomly translates images during training. — layer_random_translation","text":"layer apply random translations image training, filling empty space according fill_mode. Input pixel values can range (e.g. [0., 1.) [0, 255]) integer floating point dtype. default, layer output floats.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_random_translation.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"A preprocessing layer which randomly translates images during training. — layer_random_translation","text":"","code":"layer_random_translation(   object,   height_factor,   width_factor,   fill_mode = \"reflect\",   interpolation = \"bilinear\",   seed = NULL,   fill_value = 0,   data_format = NULL,   ... )"},{"path":"https://keras3.posit.co/dev/reference/layer_random_translation.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"A preprocessing layer which randomly translates images during training. — layer_random_translation","text":"object Object compose layer . tensor, array, sequential model. height_factor float represented fraction value, tuple size 2 representing lower upper bound shifting vertically. negative value means shifting image , positive value means shifting image . represented single positive float, value used upper lower bound. instance, height_factor=(-0.2, 0.3) results output shifted random amount range [-20%, +30%]. height_factor=0.2 results output height shifted random amount range [-20%, +20%]. width_factor float represented fraction value, tuple size 2 representing lower upper bound shifting horizontally. negative value means shifting image left, positive value means shifting image right. represented single positive float, value used upper lower bound. instance, width_factor=(-0.2, 0.3) results output shifted left 20%, shifted right 30%. width_factor=0.2 results output height shifted left right 20%. fill_mode Points outside boundaries input filled according given mode. Available methods \"constant\", \"nearest\", \"wrap\" \"reflect\". Defaults \"reflect\". \"reflect\": (d c b | b c d | d c b ) input extended reflecting edge last pixel. \"constant\": (k k k k | b c d | k k k k) input extended filling values beyond edge constant value k specified fill_value. \"wrap\": (b c d | b c d | b c d) input extended wrapping around opposite edge. \"nearest\": (| b c d | d d d d) input extended nearest pixel. Note using torch backend, \"reflect\" redirected \"mirror\" (c d c b | b c d | c b b) torch support \"reflect\". Note torch backend support \"wrap\". interpolation Interpolation mode. Supported values: \"nearest\", \"bilinear\". seed Integer. Used create random seed. fill_value float represents value filled outside boundaries fill_mode=\"constant\". data_format string, either \"channels_last\" \"channels_first\". ordering dimensions inputs. \"channels_last\" corresponds inputs shape (batch, height, width, channels) \"channels_first\" corresponds inputs shape (batch, channels, height, width). defaults image_data_format value found Keras config file ~/.keras/keras.json. never set , \"channels_last\". ... Base layer keyword arguments, name dtype.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_random_translation.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"A preprocessing layer which randomly translates images during training. — layer_random_translation","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_random_translation.html","id":"input-shape","dir":"Reference","previous_headings":"","what":"Input Shape","title":"A preprocessing layer which randomly translates images during training. — layer_random_translation","text":"3D (unbatched) 4D (batched) tensor shape: (..., height, width, channels), \"channels_last\" format, (..., channels, height, width), \"channels_first\" format.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_random_translation.html","id":"output-shape","dir":"Reference","previous_headings":"","what":"Output Shape","title":"A preprocessing layer which randomly translates images during training. — layer_random_translation","text":"3D (unbatched) 4D (batched) tensor shape: (..., target_height, target_width, channels), (..., channels, target_height, target_width), \"channels_first\" format. Note: layer safe use inside tf.data pipeline (independently backend using).","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_random_zoom.html","id":null,"dir":"Reference","previous_headings":"","what":"A preprocessing layer which randomly zooms images during training. — layer_random_zoom","title":"A preprocessing layer which randomly zooms images during training. — layer_random_zoom","text":"layer randomly zoom axis image independently, filling empty space according fill_mode. Input pixel values can range (e.g. [0., 1.) [0, 255]) integer floating point dtype. default, layer output floats.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_random_zoom.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"A preprocessing layer which randomly zooms images during training. — layer_random_zoom","text":"","code":"layer_random_zoom(   object,   height_factor,   width_factor = NULL,   fill_mode = \"reflect\",   interpolation = \"bilinear\",   seed = NULL,   fill_value = 0,   data_format = NULL,   ... )"},{"path":"https://keras3.posit.co/dev/reference/layer_random_zoom.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"A preprocessing layer which randomly zooms images during training. — layer_random_zoom","text":"object Object compose layer . tensor, array, sequential model. height_factor float represented fraction value, list size 2 representing lower upper bound zooming vertically. represented single float, value used upper lower bound. positive value means zooming , negative value means zooming . instance, height_factor=c(0.2, 0.3) result output zoomed random amount range [+20%, +30%]. height_factor=c(-0.3, -0.2) result output zoomed random amount range [+20%, +30%]. width_factor float represented fraction value, list size 2 representing lower upper bound zooming horizontally. represented single float, value used upper lower bound. instance, width_factor=c(0.2, 0.3) result output zooming 20% 30%. width_factor=c(-0.3, -0.2) result output zooming 20% 30%. NULL means .e., zooming vertical horizontal directions preserving aspect ratio. Defaults NULL. fill_mode Points outside boundaries input filled according given mode. Available methods \"constant\", \"nearest\", \"wrap\" \"reflect\". Defaults \"reflect\". \"reflect\": (d c b | b c d | d c b ) input extended reflecting edge last pixel. \"constant\": (k k k k | b c d | k k k k) input extended filling values beyond edge constant value k specified fill_value. \"wrap\": (b c d | b c d | b c d) input extended wrapping around opposite edge. \"nearest\": (| b c d | d d d d) input extended nearest pixel. Note using torch backend, \"reflect\" redirected \"mirror\" (c d c b | b c d | c b b) torch support \"reflect\". Note torch backend support \"wrap\". interpolation Interpolation mode. Supported values: \"nearest\", \"bilinear\". seed Integer. Used create random seed. fill_value float represents value filled outside boundaries fill_mode=\"constant\". data_format string, either \"channels_last\" \"channels_first\". ordering dimensions inputs. \"channels_last\" corresponds inputs shape (batch, height, width, channels) \"channels_first\" corresponds inputs shape (batch, channels, height, width). defaults image_data_format value found Keras config file ~/.keras/keras.json. never set , \"channels_last\". ... Base layer keyword arguments, name dtype.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_random_zoom.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"A preprocessing layer which randomly zooms images during training. — layer_random_zoom","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_random_zoom.html","id":"input-shape","dir":"Reference","previous_headings":"","what":"Input Shape","title":"A preprocessing layer which randomly zooms images during training. — layer_random_zoom","text":"3D (unbatched) 4D (batched) tensor shape: (..., height, width, channels), \"channels_last\" format, (..., channels, height, width), \"channels_first\" format.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_random_zoom.html","id":"output-shape","dir":"Reference","previous_headings":"","what":"Output Shape","title":"A preprocessing layer which randomly zooms images during training. — layer_random_zoom","text":"3D (unbatched) 4D (batched) tensor shape: (..., target_height, target_width, channels), (..., channels, target_height, target_width), \"channels_first\" format. Note: layer safe use inside tf.data pipeline (independently backend using).","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_random_zoom.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"A preprocessing layer which randomly zooms images during training. — layer_random_zoom","text":"","code":"input_img <- random_uniform(c(32, 224, 224, 3)) layer <- layer_random_zoom(height_factor = .5, width_factor = .2) out_img <- layer(input_img)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_repeat_vector.html","id":null,"dir":"Reference","previous_headings":"","what":"Repeats the input n times. — layer_repeat_vector","title":"Repeats the input n times. — layer_repeat_vector","text":"Repeats input n times.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_repeat_vector.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Repeats the input n times. — layer_repeat_vector","text":"","code":"layer_repeat_vector(object, n, ...)"},{"path":"https://keras3.posit.co/dev/reference/layer_repeat_vector.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Repeats the input n times. — layer_repeat_vector","text":"object Object compose layer . tensor, array, sequential model. n Integer, repetition factor. ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_repeat_vector.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Repeats the input n times. — layer_repeat_vector","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_repeat_vector.html","id":"example","dir":"Reference","previous_headings":"","what":"Example","title":"Repeats the input n times. — layer_repeat_vector","text":"","code":"x <- layer_input(shape = 32) y <- layer_repeat_vector(x, n = 3) shape(y) ## shape(NA, 3, 32)"},{"path":"https://keras3.posit.co/dev/reference/layer_repeat_vector.html","id":"input-shape","dir":"Reference","previous_headings":"","what":"Input Shape","title":"Repeats the input n times. — layer_repeat_vector","text":"2D tensor shape (batch_size, features).","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_repeat_vector.html","id":"output-shape","dir":"Reference","previous_headings":"","what":"Output Shape","title":"Repeats the input n times. — layer_repeat_vector","text":"3D tensor shape (batch_size, n, features).","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_rescaling.html","id":null,"dir":"Reference","previous_headings":"","what":"A preprocessing layer which rescales input values to a new range. — layer_rescaling","title":"A preprocessing layer which rescales input values to a new range. — layer_rescaling","text":"layer rescales every value input (often image) multiplying scale adding offset. instance: rescale input [0, 255] range [0, 1] range, pass scale=1./255. rescale input [0, 255] range [-1, 1] range, pass scale=1./127.5, offset=-1. rescaling applied training inference. Inputs can integer floating point dtype, default layer output floats. Note: layer safe use inside tf.data pipeline (independently backend using).","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_rescaling.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"A preprocessing layer which rescales input values to a new range. — layer_rescaling","text":"","code":"layer_rescaling(object, scale, offset = 0, ...)"},{"path":"https://keras3.posit.co/dev/reference/layer_rescaling.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"A preprocessing layer which rescales input values to a new range. — layer_rescaling","text":"object Object compose layer . tensor, array, sequential model. scale Numeric scalar, vector, array. scale apply inputs. scalar, scale applied every feature channel; vector array, scaling applied per channel. offset Numeric scalar, vector, array. offset apply inputs. scalar, offset applied every feature channel; vector array, shift applied per channel. ... Base layer keyword arguments, name dtype.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_rescaling.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"A preprocessing layer which rescales input values to a new range. — layer_rescaling","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_reshape.html","id":null,"dir":"Reference","previous_headings":"","what":"Layer that reshapes inputs into the given shape. — layer_reshape","title":"Layer that reshapes inputs into the given shape. — layer_reshape","text":"Layer reshapes inputs given shape.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_reshape.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Layer that reshapes inputs into the given shape. — layer_reshape","text":"","code":"layer_reshape(object, target_shape, ...)"},{"path":"https://keras3.posit.co/dev/reference/layer_reshape.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Layer that reshapes inputs into the given shape. — layer_reshape","text":"object Object compose layer . tensor, array, sequential model. target_shape Target shape. List integers, include samples dimension (batch size). ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_reshape.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Layer that reshapes inputs into the given shape. — layer_reshape","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_reshape.html","id":"input-shape","dir":"Reference","previous_headings":"","what":"Input Shape","title":"Layer that reshapes inputs into the given shape. — layer_reshape","text":"Arbitrary, although dimensions input shape must known/fixed. Use keyword argument input_shape (list integers, include samples/batch size axis) using layer first layer model.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_reshape.html","id":"output-shape","dir":"Reference","previous_headings":"","what":"Output Shape","title":"Layer that reshapes inputs into the given shape. — layer_reshape","text":"(batch_size, *target_shape)","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_reshape.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Layer that reshapes inputs into the given shape. — layer_reshape","text":"","code":"x <- layer_input(shape = 12) y <- layer_reshape(x, c(3, 4)) shape(y) ## shape(NA, 3, 4) # also supports shape inference using `-1` as dimension y <- layer_reshape(x, c(-1, 2, 2)) shape(y) ## shape(NA, 3, 2, 2)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_resizing.html","id":null,"dir":"Reference","previous_headings":"","what":"A preprocessing layer which resizes images. — layer_resizing","title":"A preprocessing layer which resizes images. — layer_resizing","text":"layer resizes image input target height width. input 4D (batched) 3D (unbatched) tensor \"channels_last\" format. Input pixel values can range (e.g. [0., 1.) [0, 255]).","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_resizing.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"A preprocessing layer which resizes images. — layer_resizing","text":"","code":"layer_resizing(   object,   height,   width,   interpolation = \"bilinear\",   crop_to_aspect_ratio = FALSE,   pad_to_aspect_ratio = FALSE,   fill_mode = \"constant\",   fill_value = 0,   data_format = NULL,   ...,   antialias = FALSE )"},{"path":"https://keras3.posit.co/dev/reference/layer_resizing.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"A preprocessing layer which resizes images. — layer_resizing","text":"object Object compose layer . tensor, array, sequential model. height Integer, height output shape. width Integer, width output shape. interpolation String, interpolation method. Supports \"bilinear\", \"nearest\", \"bicubic\", \"lanczos3\", \"lanczos5\". Defaults \"bilinear\". crop_to_aspect_ratio TRUE, resize images without aspect ratio distortion. original aspect ratio differs target aspect ratio, output image cropped return largest possible window image (size (height, width)) matches target aspect ratio. default (crop_to_aspect_ratio=FALSE), aspect ratio may preserved. pad_to_aspect_ratio TRUE, pad images without aspect ratio distortion. original aspect ratio differs target aspect ratio, output image evenly padded short side. fill_mode using pad_to_aspect_ratio=TRUE, padded areas filled according given mode. \"constant\" supported time (fill constant value, equal fill_value). fill_value Float. Padding value use pad_to_aspect_ratio=TRUE. data_format string, either \"channels_last\" \"channels_first\". ordering dimensions inputs. \"channels_last\" corresponds inputs shape (batch, height, width, channels) \"channels_first\" corresponds inputs shape (batch, channels, height, width). defaults image_data_format value found Keras config file ~/.keras/keras.json. never set , \"channels_last\". ... Base layer keyword arguments, name dtype. antialias bool","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_resizing.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"A preprocessing layer which resizes images. — layer_resizing","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_resizing.html","id":"input-shape","dir":"Reference","previous_headings":"","what":"Input Shape","title":"A preprocessing layer which resizes images. — layer_resizing","text":"3D (unbatched) 4D (batched) tensor shape: (..., height, width, channels), \"channels_last\" format, (..., channels, height, width), \"channels_first\" format.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_resizing.html","id":"output-shape","dir":"Reference","previous_headings":"","what":"Output Shape","title":"A preprocessing layer which resizes images. — layer_resizing","text":"3D (unbatched) 4D (batched) tensor shape: (..., target_height, target_width, channels), (..., channels, target_height, target_width), \"channels_first\" format. Note: layer safe use inside tf.data pipeline (independently backend using).","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_rms_normalization.html","id":null,"dir":"Reference","previous_headings":"","what":"Root Mean Square (RMS) Normalization layer. — layer_rms_normalization","title":"Root Mean Square (RMS) Normalization layer. — layer_rms_normalization","text":"layer normalizes input tensor based RMS value. Keras layer performs operation described Root Mean Square Layer Normalization Biao Zhang et al. scale enabled, layer scale normalized outputs via learnable scaling factor. , scaling enabled, normalization equations follows: Let intermediate activations mini-batch inputs.   example:","code":"rms_normalization(x) = x * rsqrt(mean(square(x))) * scale layer <- layer_rms_normalization() layer$build(shape(5, 20, 30, 10)) op_shape(layer$scale$shape) ## shape(1)  op_shape(layer(op_array(runif(10)))) ## shape(10)"},{"path":"https://keras3.posit.co/dev/reference/layer_rms_normalization.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Root Mean Square (RMS) Normalization layer. — layer_rms_normalization","text":"","code":"layer_rms_normalization(object, axis = -1L, epsilon = 1e-06, ...)"},{"path":"https://keras3.posit.co/dev/reference/layer_rms_normalization.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Root Mean Square (RMS) Normalization layer. — layer_rms_normalization","text":"object Object compose layer . tensor, array, sequential model. axis int. axis perform normalization. epsilon float. small number add avoid division zero. ... forward/backward compatability.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_rnn.html","id":null,"dir":"Reference","previous_headings":"","what":"Base class for recurrent layers — layer_rnn","title":"Base class for recurrent layers — layer_rnn","text":"Base class recurrent layers","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_rnn.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Base class for recurrent layers — layer_rnn","text":"","code":"layer_rnn(   object,   cell,   return_sequences = FALSE,   return_state = FALSE,   go_backwards = FALSE,   stateful = FALSE,   unroll = FALSE,   zero_output_for_mask = FALSE,   ... )"},{"path":"https://keras3.posit.co/dev/reference/layer_rnn.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Base class for recurrent layers — layer_rnn","text":"object Object compose layer . tensor, array, sequential model. cell RNN cell instance list RNN cell instances. RNN cell class : call(input_at_t, states_at_t) method, returning (output_at_t, states_at_t_plus_1). call method cell can also take optional argument constants, see section \"Note passing external constants\" . state_size attribute. can single integer (single state) case size recurrent state. can also list integers (one size per state). output_size attribute, single integer. get_initial_state(batch_size=NULL) method creates tensor meant fed call() initial state, user specify initial state via means. returned initial state shape (batch_size, cell.state_size). cell might choose create tensor full zeros, values based cell's implementation. inputs input tensor RNN layer, shape (batch_size, timesteps, features). method implemented cell, RNN layer create zero filled tensor shape (batch_size, cell$state_size). case cell list RNN cell instances, cells stacked top RNN, resulting efficient stacked RNN. return_sequences Boolean (default FALSE). Whether return last output output sequence, full sequence. return_state Boolean (default FALSE). Whether return last state addition output. go_backwards Boolean (default FALSE). TRUE, process input sequence backwards return reversed sequence. stateful Boolean (default FALSE). TRUE, last state sample index batch used initial state sample index following batch. unroll Boolean (default FALSE). TRUE, network unrolled, else symbolic loop used. Unrolling can speed-RNN, although tends memory-intensive. Unrolling suitable short sequences. zero_output_for_mask Boolean (default FALSE). Whether output use zeros masked timesteps. Note field used return_sequences TRUE mask provided. can useful want reuse raw output sequence RNN without interference masked timesteps, e.g., merging bidirectional RNNs. ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_rnn.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Base class for recurrent layers — layer_rnn","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_rnn.html","id":"call-arguments","dir":"Reference","previous_headings":"","what":"Call Arguments","title":"Base class for recurrent layers — layer_rnn","text":"sequences: 3-D tensor shape (batch_size, timesteps, features). initial_state: List initial state tensors passed first call cell. mask: Binary tensor shape [batch_size, timesteps] indicating whether given timestep masked. individual TRUE entry indicates corresponding timestep utilized, FALSE entry indicates corresponding timestep ignored. training: Python boolean indicating whether layer behave training mode inference mode. argument passed cell calling . use cells use dropout.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_rnn.html","id":"input-shape","dir":"Reference","previous_headings":"","what":"Input Shape","title":"Base class for recurrent layers — layer_rnn","text":"3-D tensor shape (batch_size, timesteps, features).","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_rnn.html","id":"output-shape","dir":"Reference","previous_headings":"","what":"Output Shape","title":"Base class for recurrent layers — layer_rnn","text":"return_state: list tensors. first tensor output. remaining tensors last states, shape (batch_size, state_size), state_size high dimension tensor shape. return_sequences: 3D tensor shape (batch_size, timesteps, output_size).","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_rnn.html","id":"masking-","dir":"Reference","previous_headings":"","what":"Masking:","title":"Base class for recurrent layers — layer_rnn","text":"layer supports masking input data variable number timesteps. introduce masks data, use layer_embedding() layer mask_zero parameter set TRUE. Note using statefulness RNNs: can set RNN layers 'stateful', means states computed samples one batch reused initial states samples next batch. assumes one--one mapping samples different successive batches. enable statefulness: Specify stateful=TRUE layer constructor. Specify fixed batch size model, passing batch_size=... layer_input() layer(s) model. Remember also specify batch_size=... calling fit(), otherwise use generator-like data source like tf.data.Dataset. Specify shuffle=FALSE calling fit(), since batches expected temporally ordered. reset states model, call reset_state() either specific layer, entire model. Note specifying initial state RNNs: can specify initial state RNN layers symbolically passing named argument initial_state layer reset_state(). value initial_state tensor list tensors representing initial state RNN layer.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_rnn.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Base class for recurrent layers — layer_rnn","text":"First, define RNN Cell, layer subclass.   use cell RNN layer:","code":"rnn_cell_minimal <- Layer(   \"MinimalRNNCell\",    initialize = function(units, ...) {     super$initialize(...)     self$units <- as.integer(units)     self$state_size <- as.integer(units)   },    build = function(input_shape) {     self$kernel <- self$add_weight(       shape = shape(tail(input_shape, 1), self$units),       initializer = 'uniform',       name = 'kernel'     )     self$recurrent_kernel <- self$add_weight(       shape = shape(self$units, self$units),       initializer = 'uniform',       name = 'recurrent_kernel'     )   },    call = function(inputs, states) {     prev_output <- states[[1]]     h <- op_matmul(inputs, self$kernel)     output <- h + op_matmul(prev_output, self$recurrent_kernel)     list(output, list(output))   } ) cell <- rnn_cell_minimal(units = 32) x <- layer_input(shape = shape(NULL, 5)) layer <- layer_rnn(cell = cell) y <- layer(x) cells <- list(rnn_cell_minimal(units = 32), rnn_cell_minimal(units = 4)) x <- layer_input(shape = shape(NULL, 5)) layer <- layer_rnn(cell = cells) y <- layer(x)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_separable_conv_1d.html","id":null,"dir":"Reference","previous_headings":"","what":"1D separable convolution layer. — layer_separable_conv_1d","title":"1D separable convolution layer. — layer_separable_conv_1d","text":"layer performs depthwise convolution acts separately channels, followed pointwise convolution mixes channels. use_bias TRUE bias initializer provided, adds bias vector output. optionally applies activation function produce final output.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_separable_conv_1d.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"1D separable convolution layer. — layer_separable_conv_1d","text":"","code":"layer_separable_conv_1d(   object,   filters,   kernel_size,   strides = 1L,   padding = \"valid\",   data_format = NULL,   dilation_rate = 1L,   depth_multiplier = 1L,   activation = NULL,   use_bias = TRUE,   depthwise_initializer = \"glorot_uniform\",   pointwise_initializer = \"glorot_uniform\",   bias_initializer = \"zeros\",   depthwise_regularizer = NULL,   pointwise_regularizer = NULL,   bias_regularizer = NULL,   activity_regularizer = NULL,   depthwise_constraint = NULL,   pointwise_constraint = NULL,   bias_constraint = NULL,   ... )"},{"path":"https://keras3.posit.co/dev/reference/layer_separable_conv_1d.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"1D separable convolution layer. — layer_separable_conv_1d","text":"object Object compose layer . tensor, array, sequential model. filters int, dimensionality output space (.e. number filters pointwise convolution). kernel_size int list 1 integers, specifying size depthwise convolution window. strides int list 1 integers, specifying stride length depthwise convolution. one int specified, stride size used dimensions. strides > 1 incompatible dilation_rate > 1. padding string, either \"valid\" \"\" (case-insensitive). \"valid\" means padding. \"\" results padding evenly left/right /input. padding=\"\" strides=1, output size input. data_format string, either \"channels_last\" \"channels_first\". ordering dimensions inputs. \"channels_last\" corresponds inputs shape (batch, steps, features) \"channels_first\" corresponds inputs shape (batch, features, steps). defaults image_data_format value found Keras config file ~/.keras/keras.json. never set , \"channels_last\". dilation_rate int list 1 integers, specifying dilation rate use dilated convolution. one int specified, dilation rate used dimensions. depth_multiplier number depthwise convolution output channels input channel. total number depthwise convolution output channels equal input_channel * depth_multiplier. activation Activation function. NULL, activation applied. use_bias bool, TRUE, bias added output. depthwise_initializer initializer depthwise convolution kernel. NULL, default initializer (\"glorot_uniform\") used. pointwise_initializer initializer pointwise convolution kernel. NULL, default initializer (\"glorot_uniform\") used. bias_initializer initializer bias vector. NULL, default initializer ('\"zeros\"') used. depthwise_regularizer Optional regularizer depthwise convolution kernel. pointwise_regularizer Optional regularizer pointwise convolution kernel. bias_regularizer Optional regularizer bias vector. activity_regularizer Optional regularizer function output. depthwise_constraint Optional projection function applied depthwise kernel updated Optimizer (e.g. used norm constraints value constraints layer weights). function must take input unprojected variable must return projected variable (must shape). pointwise_constraint Optional projection function applied pointwise kernel updated Optimizer. bias_constraint Optional projection function applied bias updated Optimizer. ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_separable_conv_1d.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"1D separable convolution layer. — layer_separable_conv_1d","text":"3D tensor representing activation(separable_conv1d(inputs, kernel) + bias).","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_separable_conv_1d.html","id":"input-shape","dir":"Reference","previous_headings":"","what":"Input Shape","title":"1D separable convolution layer. — layer_separable_conv_1d","text":"data_format=\"channels_last\": 3D tensor shape: (batch_shape, steps, channels) data_format=\"channels_first\": 3D tensor shape: (batch_shape, channels, steps)","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_separable_conv_1d.html","id":"output-shape","dir":"Reference","previous_headings":"","what":"Output Shape","title":"1D separable convolution layer. — layer_separable_conv_1d","text":"data_format=\"channels_last\": 3D tensor shape: (batch_shape, new_steps, filters) data_format=\"channels_first\": 3D tensor shape: (batch_shape, filters, new_steps)","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_separable_conv_1d.html","id":"example","dir":"Reference","previous_headings":"","what":"Example","title":"1D separable convolution layer. — layer_separable_conv_1d","text":"","code":"x <- random_uniform(c(4, 10, 12)) y <- layer_separable_conv_1d(x, 3, 2, 2, activation='relu') shape(y) ## shape(4, 5, 3)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_separable_conv_2d.html","id":null,"dir":"Reference","previous_headings":"","what":"2D separable convolution layer. — layer_separable_conv_2d","title":"2D separable convolution layer. — layer_separable_conv_2d","text":"layer performs depthwise convolution acts separately channels, followed pointwise convolution mixes channels. use_bias TRUE bias initializer provided, adds bias vector output. optionally applies activation function produce final output.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_separable_conv_2d.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"2D separable convolution layer. — layer_separable_conv_2d","text":"","code":"layer_separable_conv_2d(   object,   filters,   kernel_size,   strides = list(1L, 1L),   padding = \"valid\",   data_format = NULL,   dilation_rate = list(1L, 1L),   depth_multiplier = 1L,   activation = NULL,   use_bias = TRUE,   depthwise_initializer = \"glorot_uniform\",   pointwise_initializer = \"glorot_uniform\",   bias_initializer = \"zeros\",   depthwise_regularizer = NULL,   pointwise_regularizer = NULL,   bias_regularizer = NULL,   activity_regularizer = NULL,   depthwise_constraint = NULL,   pointwise_constraint = NULL,   bias_constraint = NULL,   ... )"},{"path":"https://keras3.posit.co/dev/reference/layer_separable_conv_2d.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"2D separable convolution layer. — layer_separable_conv_2d","text":"object Object compose layer . tensor, array, sequential model. filters int, dimensionality output space (.e. number filters pointwise convolution). kernel_size int list 2 integers, specifying size depthwise convolution window. strides int list 2 integers, specifying stride length depthwise convolution. one int specified, stride size used dimensions. strides > 1 incompatible dilation_rate > 1. padding string, either \"valid\" \"\" (case-insensitive). \"valid\" means padding. \"\" results padding evenly left/right /input. padding=\"\" strides=1, output size input. data_format string, either \"channels_last\" \"channels_first\". ordering dimensions inputs. \"channels_last\" corresponds inputs shape (batch, height, width, channels) \"channels_first\" corresponds inputs shape (batch, channels, height, width). defaults image_data_format value found Keras config file ~/.keras/keras.json. never set , \"channels_last\". dilation_rate int list 2 integers, specifying dilation rate use dilated convolution. one int specified, dilation rate used dimensions. depth_multiplier number depthwise convolution output channels input channel. total number depthwise convolution output channels equal input_channel * depth_multiplier. activation Activation function. NULL, activation applied. use_bias bool, TRUE, bias added output. depthwise_initializer initializer depthwise convolution kernel. NULL, default initializer (\"glorot_uniform\") used. pointwise_initializer initializer pointwise convolution kernel. NULL, default initializer (\"glorot_uniform\") used. bias_initializer initializer bias vector. NULL, default initializer ('\"zeros\"') used. depthwise_regularizer Optional regularizer depthwise convolution kernel. pointwise_regularizer Optional regularizer pointwise convolution kernel. bias_regularizer Optional regularizer bias vector. activity_regularizer Optional regularizer function output. depthwise_constraint Optional projection function applied depthwise kernel updated Optimizer (e.g. used norm constraints value constraints layer weights). function must take input unprojected variable must return projected variable (must shape). pointwise_constraint Optional projection function applied pointwise kernel updated Optimizer. bias_constraint Optional projection function applied bias updated Optimizer. ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_separable_conv_2d.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"2D separable convolution layer. — layer_separable_conv_2d","text":"4D tensor representing activation(separable_conv2d(inputs, kernel) + bias).","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_separable_conv_2d.html","id":"input-shape","dir":"Reference","previous_headings":"","what":"Input Shape","title":"2D separable convolution layer. — layer_separable_conv_2d","text":"data_format=\"channels_last\": 4D tensor shape: (batch_size, height, width, channels) data_format=\"channels_first\": 4D tensor shape: (batch_size, channels, height, width)","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_separable_conv_2d.html","id":"output-shape","dir":"Reference","previous_headings":"","what":"Output Shape","title":"2D separable convolution layer. — layer_separable_conv_2d","text":"data_format=\"channels_last\": 4D tensor shape: (batch_size, new_height, new_width, filters) data_format=\"channels_first\": 4D tensor shape: (batch_size, filters, new_height, new_width)","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_separable_conv_2d.html","id":"example","dir":"Reference","previous_headings":"","what":"Example","title":"2D separable convolution layer. — layer_separable_conv_2d","text":"","code":"x <- random_uniform(c(4, 10, 10, 12)) y <- layer_separable_conv_2d(x, 3, c(4, 3), 2, activation='relu') shape(y) ## shape(4, 4, 4, 3)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_simple_rnn.html","id":null,"dir":"Reference","previous_headings":"","what":"Fully-connected RNN where the output is to be fed back as the new input. — layer_simple_rnn","title":"Fully-connected RNN where the output is to be fed back as the new input. — layer_simple_rnn","text":"Fully-connected RNN output fed back new input.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_simple_rnn.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Fully-connected RNN where the output is to be fed back as the new input. — layer_simple_rnn","text":"","code":"layer_simple_rnn(   object,   units,   activation = \"tanh\",   use_bias = TRUE,   kernel_initializer = \"glorot_uniform\",   recurrent_initializer = \"orthogonal\",   bias_initializer = \"zeros\",   kernel_regularizer = NULL,   recurrent_regularizer = NULL,   bias_regularizer = NULL,   activity_regularizer = NULL,   kernel_constraint = NULL,   recurrent_constraint = NULL,   bias_constraint = NULL,   dropout = 0,   recurrent_dropout = 0,   return_sequences = FALSE,   return_state = FALSE,   go_backwards = FALSE,   stateful = FALSE,   unroll = FALSE,   seed = NULL,   ... )"},{"path":"https://keras3.posit.co/dev/reference/layer_simple_rnn.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Fully-connected RNN where the output is to be fed back as the new input. — layer_simple_rnn","text":"object Object compose layer . tensor, array, sequential model. units Positive integer, dimensionality output space. activation Activation function use. Default: hyperbolic tangent (tanh). pass NULL, activation applied (ie. \"linear\" activation: (x) = x). use_bias Boolean, (default TRUE), whether layer uses bias vector. kernel_initializer Initializer kernel weights matrix, used linear transformation inputs. Default: \"glorot_uniform\". recurrent_initializer Initializer recurrent_kernel weights matrix, used linear transformation recurrent state.  Default: \"orthogonal\". bias_initializer Initializer bias vector. Default: \"zeros\". kernel_regularizer Regularizer function applied kernel weights matrix. Default: NULL. recurrent_regularizer Regularizer function applied recurrent_kernel weights matrix. Default: NULL. bias_regularizer Regularizer function applied bias vector. Default: NULL. activity_regularizer Regularizer function applied output layer (\"activation\"). Default: NULL. kernel_constraint Constraint function applied kernel weights matrix. Default: NULL. recurrent_constraint Constraint function applied recurrent_kernel weights matrix.  Default: NULL. bias_constraint Constraint function applied bias vector. Default: NULL. dropout Float 0 1. Fraction units drop linear transformation inputs. Default: 0. recurrent_dropout Float 0 1. Fraction units drop linear transformation recurrent state. Default: 0. return_sequences Boolean. Whether return last output output sequence, full sequence. Default: FALSE. return_state Boolean. Whether return last state addition output. Default: FALSE. go_backwards Boolean (default: FALSE). TRUE, process input sequence backwards return reversed sequence. stateful Boolean (default: FALSE). TRUE, last state sample index batch used initial state sample index following batch. unroll Boolean (default: FALSE). TRUE, network unrolled, else symbolic loop used. Unrolling can speed-RNN, although tends memory-intensive. Unrolling suitable short sequences. seed Initial seed random number generator ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_simple_rnn.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Fully-connected RNN where the output is to be fed back as the new input. — layer_simple_rnn","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_simple_rnn.html","id":"call-arguments","dir":"Reference","previous_headings":"","what":"Call Arguments","title":"Fully-connected RNN where the output is to be fed back as the new input. — layer_simple_rnn","text":"sequence: 3D tensor, shape [batch, timesteps, feature]. mask: Binary tensor shape [batch, timesteps] indicating whether given timestep masked. individual TRUE entry indicates corresponding timestep utilized, FALSE entry indicates corresponding timestep ignored. training: Python boolean indicating whether layer behave training mode inference mode. argument passed cell calling . relevant dropout recurrent_dropout used. initial_state: List initial state tensors passed first call cell.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_simple_rnn.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Fully-connected RNN where the output is to be fed back as the new input. — layer_simple_rnn","text":"","code":"inputs <- random_uniform(c(32, 10, 8)) simple_rnn <- layer_simple_rnn(units = 4) output <- simple_rnn(inputs)  # The output has shape `(32, 4)`. simple_rnn <- layer_simple_rnn(     units = 4, return_sequences=TRUE, return_state=TRUE ) # whole_sequence_output has shape `(32, 10, 4)`. # final_state has shape `(32, 4)`. c(whole_sequence_output, final_state) %<-% simple_rnn(inputs)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_solarization.html","id":null,"dir":"Reference","previous_headings":"","what":"Applies (max_value - pixel + min_value) for each pixel in the image. — layer_solarization","title":"Applies (max_value - pixel + min_value) for each pixel in the image. — layer_solarization","text":"created without threshold parameter, layer performs solarization values. created specified threshold layer augments pixels threshold value.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_solarization.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Applies (max_value - pixel + min_value) for each pixel in the image. — layer_solarization","text":"","code":"layer_solarization(   object,   addition_factor = 0,   threshold_factor = 0,   value_range = tuple(0L, 255L),   seed = NULL,   ... )"},{"path":"https://keras3.posit.co/dev/reference/layer_solarization.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Applies (max_value - pixel + min_value) for each pixel in the image. — layer_solarization","text":"object Object compose layer . tensor, array, sequential model. addition_factor (Optional)  tuple two floats single float, 0 1. augmented image value sampled provided range. float passed, range interpreted (0, addition_factor). specified, value (times value range input images, e.g. 255), added pixel solarization thresholding. Defaults 0.0. threshold_factor (Optional)  tuple two floats single float. augmented image value sampled provided range. float passed, range interpreted (0, threshold_factor). specified, pixel values threshold solarized. value_range tuple list two elements. first value represents lower bound values input images, second represents upper bound. Images passed layer values within value_range. Typical values pass (0, 255) (RGB image) (0., 1.) (scaled image). seed Integer. Used create random seed. ... Base layer keyword arguments, name dtype.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_solarization.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Applies (max_value - pixel + min_value) for each pixel in the image. — layer_solarization","text":"","code":"c(c(images, labels), .) %<-% dataset_cifar10() str(images) ##  int [1:50000, 1:32, 1:32, 1:3] 59 154 255 28 170 159 164 28 134 125 ... str(images[1, 1, 1, ]) ##  int [1:3] 59 62 63 # Note that images are Tensor with values in the range [0, 255] solarization <- layer_solarization(value_range = c(0, 255)) images <- solarization(images) |> as.array() str(images[1, 1, 1, ]) ##  num [1:3] 196 193 192"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_spatial_dropout_1d.html","id":null,"dir":"Reference","previous_headings":"","what":"Spatial 1D version of Dropout. — layer_spatial_dropout_1d","title":"Spatial 1D version of Dropout. — layer_spatial_dropout_1d","text":"layer performs function Dropout, however, drops entire 1D feature maps instead individual elements. adjacent frames within feature maps strongly correlated (normally case early convolution layers) regular dropout regularize activations otherwise just result effective learning rate decrease. case, SpatialDropout1D help promote independence feature maps used instead.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_spatial_dropout_1d.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Spatial 1D version of Dropout. — layer_spatial_dropout_1d","text":"","code":"layer_spatial_dropout_1d(object, rate, seed = NULL, name = NULL, dtype = NULL)"},{"path":"https://keras3.posit.co/dev/reference/layer_spatial_dropout_1d.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Spatial 1D version of Dropout. — layer_spatial_dropout_1d","text":"object Object compose layer . tensor, array, sequential model. rate Float 0 1. Fraction input units drop. seed Initial seed random number generator name String, name object dtype datatype (e.g., \"float32\").","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_spatial_dropout_1d.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Spatial 1D version of Dropout. — layer_spatial_dropout_1d","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_spatial_dropout_1d.html","id":"call-arguments","dir":"Reference","previous_headings":"","what":"Call Arguments","title":"Spatial 1D version of Dropout. — layer_spatial_dropout_1d","text":"inputs: 3D tensor. training: Python boolean indicating whether layer behave training mode (applying dropout) inference mode (pass-).","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_spatial_dropout_1d.html","id":"input-shape","dir":"Reference","previous_headings":"","what":"Input Shape","title":"Spatial 1D version of Dropout. — layer_spatial_dropout_1d","text":"3D tensor shape: (samples, timesteps, channels)","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_spatial_dropout_1d.html","id":"output-shape","dir":"Reference","previous_headings":"","what":"Output Shape","title":"Spatial 1D version of Dropout. — layer_spatial_dropout_1d","text":"input.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_spatial_dropout_1d.html","id":"reference","dir":"Reference","previous_headings":"","what":"Reference","title":"Spatial 1D version of Dropout. — layer_spatial_dropout_1d","text":"Tompson et al., 2014","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_spatial_dropout_2d.html","id":null,"dir":"Reference","previous_headings":"","what":"Spatial 2D version of Dropout. — layer_spatial_dropout_2d","title":"Spatial 2D version of Dropout. — layer_spatial_dropout_2d","text":"version performs function Dropout, however, drops entire 2D feature maps instead individual elements. adjacent pixels within feature maps strongly correlated (normally case early convolution layers) regular dropout regularize activations otherwise just result effective learning rate decrease. case, SpatialDropout2D help promote independence feature maps used instead.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_spatial_dropout_2d.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Spatial 2D version of Dropout. — layer_spatial_dropout_2d","text":"","code":"layer_spatial_dropout_2d(   object,   rate,   data_format = NULL,   seed = NULL,   name = NULL,   dtype = NULL )"},{"path":"https://keras3.posit.co/dev/reference/layer_spatial_dropout_2d.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Spatial 2D version of Dropout. — layer_spatial_dropout_2d","text":"object Object compose layer . tensor, array, sequential model. rate Float 0 1. Fraction input units drop. data_format \"channels_first\" \"channels_last\". \"channels_first\" mode, channels dimension (depth) index 1, \"channels_last\" mode index 3. defaults image_data_format value found Keras config file ~/.keras/keras.json. never set , \"channels_last\". seed Initial seed random number generator name String, name object dtype datatype (e.g., \"float32\").","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_spatial_dropout_2d.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Spatial 2D version of Dropout. — layer_spatial_dropout_2d","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_spatial_dropout_2d.html","id":"call-arguments","dir":"Reference","previous_headings":"","what":"Call Arguments","title":"Spatial 2D version of Dropout. — layer_spatial_dropout_2d","text":"inputs: 4D tensor. training: Python boolean indicating whether layer behave training mode (applying dropout) inference mode (pass-).","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_spatial_dropout_2d.html","id":"input-shape","dir":"Reference","previous_headings":"","what":"Input Shape","title":"Spatial 2D version of Dropout. — layer_spatial_dropout_2d","text":"4D tensor shape: (samples, channels, rows, cols) data_format='channels_first' 4D tensor shape: (samples, rows, cols, channels) data_format='channels_last'.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_spatial_dropout_2d.html","id":"output-shape","dir":"Reference","previous_headings":"","what":"Output Shape","title":"Spatial 2D version of Dropout. — layer_spatial_dropout_2d","text":"input.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_spatial_dropout_2d.html","id":"reference","dir":"Reference","previous_headings":"","what":"Reference","title":"Spatial 2D version of Dropout. — layer_spatial_dropout_2d","text":"Tompson et al., 2014","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_spatial_dropout_3d.html","id":null,"dir":"Reference","previous_headings":"","what":"Spatial 3D version of Dropout. — layer_spatial_dropout_3d","title":"Spatial 3D version of Dropout. — layer_spatial_dropout_3d","text":"version performs function Dropout, however, drops entire 3D feature maps instead individual elements. adjacent voxels within feature maps strongly correlated (normally case early convolution layers) regular dropout regularize activations otherwise just result effective learning rate decrease. case, SpatialDropout3D help promote independence feature maps used instead.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_spatial_dropout_3d.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Spatial 3D version of Dropout. — layer_spatial_dropout_3d","text":"","code":"layer_spatial_dropout_3d(   object,   rate,   data_format = NULL,   seed = NULL,   name = NULL,   dtype = NULL )"},{"path":"https://keras3.posit.co/dev/reference/layer_spatial_dropout_3d.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Spatial 3D version of Dropout. — layer_spatial_dropout_3d","text":"object Object compose layer . tensor, array, sequential model. rate Float 0 1. Fraction input units drop. data_format \"channels_first\" \"channels_last\". \"channels_first\" mode, channels dimension (depth) index 1, \"channels_last\" mode index 4. defaults image_data_format value found Keras config file ~/.keras/keras.json. never set , \"channels_last\". seed Initial seed random number generator name String, name object dtype datatype (e.g., \"float32\").","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_spatial_dropout_3d.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Spatial 3D version of Dropout. — layer_spatial_dropout_3d","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_spatial_dropout_3d.html","id":"call-arguments","dir":"Reference","previous_headings":"","what":"Call Arguments","title":"Spatial 3D version of Dropout. — layer_spatial_dropout_3d","text":"inputs: 5D tensor. training: Python boolean indicating whether layer behave training mode (applying dropout) inference mode (pass-).","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_spatial_dropout_3d.html","id":"input-shape","dir":"Reference","previous_headings":"","what":"Input Shape","title":"Spatial 3D version of Dropout. — layer_spatial_dropout_3d","text":"5D tensor shape: (samples, channels, dim1, dim2, dim3) data_format='channels_first' 5D tensor shape: (samples, dim1, dim2, dim3, channels) data_format='channels_last'.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_spatial_dropout_3d.html","id":"output-shape","dir":"Reference","previous_headings":"","what":"Output Shape","title":"Spatial 3D version of Dropout. — layer_spatial_dropout_3d","text":"input.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_spatial_dropout_3d.html","id":"reference","dir":"Reference","previous_headings":"","what":"Reference","title":"Spatial 3D version of Dropout. — layer_spatial_dropout_3d","text":"Tompson et al., 2014","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_spectral_normalization.html","id":null,"dir":"Reference","previous_headings":"","what":"Performs spectral normalization on the weights of a target layer. — layer_spectral_normalization","title":"Performs spectral normalization on the weights of a target layer. — layer_spectral_normalization","text":"wrapper controls Lipschitz constant weights layer constraining spectral norm, can stabilize training GANs.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_spectral_normalization.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Performs spectral normalization on the weights of a target layer. — layer_spectral_normalization","text":"","code":"layer_spectral_normalization(object, layer, power_iterations = 1L, ...)"},{"path":"https://keras3.posit.co/dev/reference/layer_spectral_normalization.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Performs spectral normalization on the weights of a target layer. — layer_spectral_normalization","text":"object Object compose layer . tensor, array, sequential model. layer Layer instance either kernel (e.g. layer_conv_2d, layer_dense...) embeddings attribute (layer_embedding layer). power_iterations int, number iterations normalization. ... Base wrapper keyword arguments.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_spectral_normalization.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Performs spectral normalization on the weights of a target layer. — layer_spectral_normalization","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_spectral_normalization.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Performs spectral normalization on the weights of a target layer. — layer_spectral_normalization","text":"Wrap layer_conv_2d:     Wrap layer_dense:","code":"x <- random_uniform(c(1, 10, 10, 1)) conv2d <- layer_spectral_normalization(   layer = layer_conv_2d(filters = 2, kernel_size = 2) ) y <- conv2d(x) shape(y) ## shape(1, 9, 9, 2) x <- random_uniform(c(1, 10, 10, 1)) dense <- layer_spectral_normalization(layer = layer_dense(units = 10)) y <- dense(x) shape(y) ## shape(1, 10, 10, 10)"},{"path":"https://keras3.posit.co/dev/reference/layer_spectral_normalization.html","id":"reference","dir":"Reference","previous_headings":"","what":"Reference","title":"Performs spectral normalization on the weights of a target layer. — layer_spectral_normalization","text":"Spectral Normalization GAN.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_stft_spectrogram.html","id":null,"dir":"Reference","previous_headings":"","what":"Layer to compute the Short-Time Fourier Transform (STFT) on a 1D signal. — layer_stft_spectrogram","title":"Layer to compute the Short-Time Fourier Transform (STFT) on a 1D signal. — layer_stft_spectrogram","text":"layer computes Spectrograms input signal produce spectrogram. layers layer computes Spectrograms based Short-Time Fourier Transform (STFT) utilizing convolution kernels, allows parallelization GPUs trainable kernels fine-tuning support. layer allows different modes output (e.g., log-scaled magnitude, phase, power spectral density, etc.) provides flexibility windowing, padding, scaling options STFT calculation.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_stft_spectrogram.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Layer to compute the Short-Time Fourier Transform (STFT) on a 1D signal. — layer_stft_spectrogram","text":"","code":"layer_stft_spectrogram(   object,   mode = \"log\",   frame_length = 256L,   frame_step = NULL,   fft_length = NULL,   window = \"hann\",   periodic = FALSE,   scaling = \"density\",   padding = \"valid\",   expand_dims = FALSE,   data_format = NULL,   ... )"},{"path":"https://keras3.posit.co/dev/reference/layer_stft_spectrogram.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Layer to compute the Short-Time Fourier Transform (STFT) on a 1D signal. — layer_stft_spectrogram","text":"object Object compose layer . tensor, array, sequential model. mode String, output type spectrogram. Can one \"log\", \"magnitude\", \"psd\", \"real\", \"imag\", \"angle\", \"stft\". Defaults \"log\". frame_length Integer, length frame (window) STFT samples. Defaults 256. frame_step Integer, step size (hop length) consecutive frames. provided, defaults half frame_length. Defaults frame_length %/% 2. fft_length Integer, size frequency bins used Fast-Fourier Transform (FFT) apply frame. greater equal frame_length.  Recommended power two. Defaults smallest power two greater equal frame_length. window (String array_like), windowing function apply frame. Can \"hann\" (default), \"hamming\", custom window provided array_like. periodic Boolean, TRUE, window function treated periodic. Defaults FALSE. scaling String, type scaling applied window. Can \"density\", \"spectrum\", None. Default \"density\". padding String, padding strategy. Can \"valid\" \"\". Defaults \"valid\". expand_dims Boolean, TRUE, expand output spectrograms two dimensions compatible image models. Defaults FALSE. data_format String, either \"channels_last\" \"channels_first\". ordering dimensions inputs. \"channels_last\" corresponds inputs shape (batch, height, width, channels) \"channels_first\" corresponds inputs shape (batch, channels, height, weight). Defaults \"channels_last\". ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_stft_spectrogram.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Layer to compute the Short-Time Fourier Transform (STFT) on a 1D signal. — layer_stft_spectrogram","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_stft_spectrogram.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Layer to compute the Short-Time Fourier Transform (STFT) on a 1D signal. — layer_stft_spectrogram","text":"Apply non-trainable preprocessing layer 3 audio tracks 1 channel, 10 seconds sampled 16 kHz.     Apply trainable processing layer 3 stereo audio tracks 2 channels, 10 seconds sampled 16 kHz. initialized non-trainable layer, can trained jointly within model.     Similar last example, add extra dimension output image used image models. apply signal 3 input channels output image tensor, hence directly applicable image model.","code":"layer <- layer_stft_spectrogram(   mode = 'log',   frame_length = 256,   frame_step = 128, # 50% overlap   fft_length = 512,   window = \"hann\",   padding = \"valid\",   trainable = FALSE # non-trainable, preprocessing only) ) random_uniform(shape=c(3, 160000, 1)) |> layer() |> op_shape() ## shape(3, 1249, 257) layer <- layer_stft_spectrogram(   mode = 'log',   frame_length = 256,   frame_step = 128,   # 50% overlap   fft_length = 512,   window = \"hamming\", # hamming windowing function   padding = \"same\",   # padding to preserve the time dimension   trainable = TRUE,   # trainable, this is the default in keras ) random_uniform(shape=c(3, 160000, 2)) |> layer() |> op_shape() ## shape(3, 1250, 514) layer <- layer_stft_spectrogram(   mode = 'log',   frame_length = 256,   frame_step = 128,   fft_length = 512,   padding = \"same\",   expand_dims = TRUE  # this adds the extra dimension ) random_uniform(shape=c(3, 160000, 3)) |> layer() |> op_shape() ## shape(3, 1250, 257, 3)"},{"path":"https://keras3.posit.co/dev/reference/layer_stft_spectrogram.html","id":"raises","dir":"Reference","previous_headings":"","what":"Raises","title":"Layer to compute the Short-Time Fourier Transform (STFT) on a 1D signal. — layer_stft_spectrogram","text":"ValueError: invalid value provided \"mode\", \"scaling\", \"padding\", input arguments. TypeError: input data type one \"float16\", \"float32\", \"float64\".","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_stft_spectrogram.html","id":"input-shape","dir":"Reference","previous_headings":"","what":"Input Shape","title":"Layer to compute the Short-Time Fourier Transform (STFT) on a 1D signal. — layer_stft_spectrogram","text":"3D tensor shape (batch_size, time_length, input_channels), data_format==\"channels_last\", shape (batch_size, input_channels, time_length) data_format==\"channels_first\", time_length length input signal, input_channels number input channels. kernels applied channel independently.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_stft_spectrogram.html","id":"output-shape","dir":"Reference","previous_headings":"","what":"Output Shape","title":"Layer to compute the Short-Time Fourier Transform (STFT) on a 1D signal. — layer_stft_spectrogram","text":"data_format==\"channels_first\" && !expand_dims, 3D tensor: (batch_size, input_channels * freq_channels, new_time_length) data_format==\"channels_last\" && !expand_dims, 3D tensor: (batch_size, new_time_length, input_channels * freq_channels) data_format==\"channels_first\" && expand_dims, 4D tensor: (batch_size, input_channels, new_time_length, freq_channels) data_format==\"channels_last\" && expand_dims, 4D tensor: (batch_size, new_time_length, freq_channels, input_channels) new_time_length depends padding, freq_channels number FFT bins (fft_length %/% 2 + 1).","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_string_lookup.html","id":null,"dir":"Reference","previous_headings":"","what":"A preprocessing layer that maps strings to (possibly encoded) indices. — layer_string_lookup","title":"A preprocessing layer that maps strings to (possibly encoded) indices. — layer_string_lookup","text":"layer translates set arbitrary strings integer output via table-based vocabulary lookup. layer perform splitting transformation input strings. layer can split tokenize natural language, see layer_text_vectorization. vocabulary layer must either supplied construction learned via adapt(). adapt(), layer analyze data set, determine frequency individual strings tokens, create vocabulary . vocabulary capped size, frequent tokens used create vocabulary others treated --vocabulary (OOV). two possible output modes layer. output_mode \"int\", input strings converted (0-based) index vocabulary (integer). output_mode \"multi_hot\", \"count\", \"tf_idf\", input strings encoded array dimension corresponds element vocabulary. vocabulary can optionally contain mask token well OOV token (can optionally occupy multiple indices vocabulary, set num_oov_indices). position tokens vocabulary fixed. output_mode \"int\", vocabulary begin mask token (set), followed OOV indices, followed rest vocabulary. output_mode \"multi_hot\", \"count\", \"tf_idf\" vocabulary begin OOV indices instances mask token dropped. Note: layer uses TensorFlow internally. used part compiled computation graph model backend TensorFlow. can however used backend running eagerly. can also always used part input preprocessing pipeline backend (outside model ), recommend use layer. Note: layer safe use inside tf.data pipeline (independently backend using). Note: working layer outputs directly (e.g., passing outputs another layer, using lower-level operations like [ op_*): returned indices 0-based. However, default settings, first (0) index OOV token, returned indices offset 1 may appear 1-based.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_string_lookup.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"A preprocessing layer that maps strings to (possibly encoded) indices. — layer_string_lookup","text":"","code":"layer_string_lookup(   object,   max_tokens = NULL,   num_oov_indices = 1L,   mask_token = NULL,   oov_token = \"[UNK]\",   vocabulary = NULL,   idf_weights = NULL,   invert = FALSE,   output_mode = \"int\",   pad_to_max_tokens = FALSE,   sparse = FALSE,   encoding = \"utf-8\",   name = NULL,   ...,   vocabulary_dtype = NULL )"},{"path":"https://keras3.posit.co/dev/reference/layer_string_lookup.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"A preprocessing layer that maps strings to (possibly encoded) indices. — layer_string_lookup","text":"object Object compose layer . tensor, array, sequential model. max_tokens Maximum size vocabulary layer. specified adapting vocabulary setting pad_to_max_tokens=TRUE. NULL, cap size vocabulary. Note size includes OOV mask tokens. Defaults NULL. num_oov_indices number --vocabulary tokens use. value 1, OOV inputs modulated determine OOV value. value 0, OOV inputs cause error calling layer. Defaults 1. mask_token token represents masked inputs. output_mode \"int\", token included vocabulary mapped index 0. output modes, token appear vocabulary instances mask token input dropped. set NULL, mask term added. Defaults NULL. oov_token used invert TRUE. token return OOV indices. Defaults \"[UNK]\". vocabulary Optional. Either array integers string path text file. passing array, can pass list, list, 1D NumPy array, 1D tensor containing integer vocabulary terms. passing file path, file contain one line per term vocabulary. argument set, need adapt() layer. idf_weights valid output_mode \"tf_idf\". list, list, 1D NumPy array, 1D tensor length vocabulary, containing floating point inverse document frequency weights, multiplied per sample term counts final TF-IDF weight. vocabulary argument set, output_mode \"tf_idf\", argument must supplied. invert valid output_mode \"int\". TRUE, layer map indices vocabulary items instead mapping vocabulary items indices. Defaults FALSE. output_mode Specification output layer. Values can \"int\", \"one_hot\", \"multi_hot\", \"count\", \"tf_idf\" configuring layer follows: \"int\": Return vocabulary indices input tokens. \"one_hot\": Encodes individual element input array size vocabulary, containing 1 element index. last dimension size 1, encode dimension. last dimension size 1, append new dimension encoded output. \"multi_hot\": Encodes sample input single array size vocabulary, containing 1 vocabulary term present sample. Treats last dimension sample dimension, input shape (..., sample_length), output shape (..., num_tokens). \"count\": \"multi_hot\", int array contains count number times token index appeared sample. \"tf_idf\": \"multi_hot\", TF-IDF algorithm applied find value token slot. \"int\" output, shape input output supported. output modes, currently output rank 2 supported. Defaults \"int\". pad_to_max_tokens applicable output_mode \"multi_hot\", \"count\", \"tf_idf\". TRUE, output feature axis padded max_tokens even number unique tokens vocabulary less max_tokens, resulting tensor shape (batch_size, max_tokens) regardless vocabulary size. Defaults FALSE. sparse Boolean. applicable \"multi_hot\", \"count\", \"tf_idf\" output modes. supported TensorFlow backend. TRUE, returns SparseTensor instead dense Tensor. Defaults FALSE. encoding Optional. text encoding use interpret input strings. Defaults \"utf-8\". name String, name object ... forward/backward compatability. vocabulary_dtype dtype vocabulary terms, example \"int64\" \"int32\". Defaults \"int64\".","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_string_lookup.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"A preprocessing layer that maps strings to (possibly encoded) indices. — layer_string_lookup","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_string_lookup.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"A preprocessing layer that maps strings to (possibly encoded) indices. — layer_string_lookup","text":"Creating lookup layer known vocabulary example creates lookup layer pre-existing vocabulary.     Creating lookup layer adapted vocabulary example creates lookup layer generates vocabulary analyzing dataset.     Note OOV token \"[UNK]\" added vocabulary. remaining tokens sorted frequency (\"d\", 2 occurrences, first) inverse sort order.     Lookups multiple OOV indices example demonstrates use lookup layer multiple OOV indices.  layer created one OOV index, OOV values hashed number OOV buckets, distributing OOV values deterministic fashion across set.     Note output OOV value 'm' 0, output OOV value \"z\" 1. -vocab terms output index increased 1 earlier examples (maps 2, etc) order make space extra OOV value. One-hot output Configure layer output_mode='one_hot'. Note first num_oov_indices dimensions ont_hot encoding represent OOV values.     Multi-hot output Configure layer output_mode='multi_hot'. Note first num_oov_indices dimensions multi_hot encoding represent OOV values.     Token count output Configure layer output_mode='count'. multi_hot output, first num_oov_indices dimensions output represent OOV values.     TF-IDF output Configure layer output_mode=\"tf_idf\". multi_hot output, first num_oov_indices dimensions output represent OOV values. token bin output token_count * idf_weight, idf weights inverse document frequency weights per token. provided along vocabulary. Note idf_weight OOV values default average idf weights passed .     specify idf weights oov values, need pass entire vocabulary including leading oov token.     adapting layer \"tf_idf\" mode, input sample considered document, IDF weight per token calculated log(1 + num_documents / (1 + token_document_count)). Inverse lookup example demonstrates map indices strings using layer. (can also use adapt() inverse=TRUE, simplicity pass vocab example.)     Note first index correspond oov token default. Forward inverse lookup pairs example demonstrates use vocabulary standard lookup layer create inverse lookup layer.     example, input value \"z\" resulted output \"[UNK]\", since 1000 vocabulary - got represented OOV, OOV values returned \"[UNK]\" inverse layer. Also, note inverse work, must already set forward layer vocabulary either directly via adapt() calling get_vocabulary().","code":"vocab <- c(\"a\", \"b\", \"c\", \"d\") data <- rbind(c(\"a\", \"c\", \"d\"), c(\"d\", \"z\", \"b\")) layer <- layer_string_lookup(vocabulary=vocab) layer(data) ## tf.Tensor( ## [[1 3 4] ##  [4 0 2]], shape=(2, 3), dtype=int64) data <- rbind(c(\"a\", \"c\", \"d\"), c(\"d\", \"z\", \"b\")) layer <- layer_string_lookup() layer %>% adapt(data) get_vocabulary(layer) ## [1] \"[UNK]\" \"d\"     \"z\"     \"c\"     \"b\"     \"a\" data <- rbind(c(\"a\", \"c\", \"d\"), c(\"d\", \"z\", \"b\")) layer <- layer_string_lookup() layer %>% adapt(data) layer(data) ## tf.Tensor( ## [[5 3 1] ##  [1 2 4]], shape=(2, 3), dtype=int64) vocab <- c(\"a\", \"b\", \"c\", \"d\") data <- rbind(c(\"a\", \"c\", \"d\"), c(\"m\", \"z\", \"b\")) layer <- layer_string_lookup(vocabulary = vocab, num_oov_indices = 2) layer(data) ## tf.Tensor( ## [[2 4 5] ##  [0 1 3]], shape=(2, 3), dtype=int64) vocab <- c(\"a\", \"b\", \"c\", \"d\") data <- c(\"a\", \"b\", \"c\", \"d\", \"z\") layer <- layer_string_lookup(vocabulary = vocab, output_mode = 'one_hot') layer(data) ## tf.Tensor( ## [[0 1 0 0 0] ##  [0 0 1 0 0] ##  [0 0 0 1 0] ##  [0 0 0 0 1] ##  [1 0 0 0 0]], shape=(5, 5), dtype=int64) vocab <- c(\"a\", \"b\", \"c\", \"d\") data <- rbind(c(\"a\", \"c\", \"d\", \"d\"), c(\"d\", \"z\", \"b\", \"z\")) layer <- layer_string_lookup(vocabulary = vocab, output_mode = 'multi_hot') layer(data) ## tf.Tensor( ## [[0 1 0 1 1] ##  [1 0 1 0 1]], shape=(2, 5), dtype=int64) vocab <- c(\"a\", \"b\", \"c\", \"d\") data <- rbind(c(\"a\", \"c\", \"d\", \"d\"), c(\"d\", \"z\", \"b\", \"z\")) layer <- layer_string_lookup(vocabulary = vocab, output_mode = 'count') layer(data) ## tf.Tensor( ## [[0 1 0 1 2] ##  [2 0 1 0 1]], shape=(2, 5), dtype=int64) vocab <- c(\"a\", \"b\", \"c\", \"d\") idf_weights <- c(0.25, 0.75, 0.6, 0.4) data <- rbind(c(\"a\", \"c\", \"d\", \"d\"), c(\"d\", \"z\", \"b\", \"z\")) layer <- layer_string_lookup(output_mode = \"tf_idf\") layer %>% set_vocabulary(vocab, idf_weights=idf_weights) layer(data) ## tf.Tensor( ## [[0.   0.25 0.   0.6  0.8 ] ##  [1.   0.   0.75 0.   0.4 ]], shape=(2, 5), dtype=float32) vocab <- c(\"[UNK]\", \"a\", \"b\", \"c\", \"d\") idf_weights <- c(0.9, 0.25, 0.75, 0.6, 0.4) data <- rbind(c(\"a\", \"c\", \"d\", \"d\"), c(\"d\", \"z\", \"b\", \"z\")) layer <- layer_string_lookup(output_mode = \"tf_idf\") layer %>% set_vocabulary(vocab, idf_weights=idf_weights) layer(data) ## tf.Tensor( ## [[0.   0.25 0.   0.6  0.8 ] ##  [1.8  0.   0.75 0.   0.4 ]], shape=(2, 5), dtype=float32) vocab <- c(\"a\", \"b\", \"c\", \"d\") data <- rbind(c(1, 3, 4), c(4, 0, 2)) layer <- layer_string_lookup(vocabulary = vocab, invert = TRUE) layer(data) ## tf.Tensor( ## [[b'a' b'c' b'd'] ##  [b'd' b'[UNK]' b'b']], shape=(2, 3), dtype=string) vocab <- c(\"a\", \"b\", \"c\", \"d\") data <- rbind(c(\"a\", \"c\", \"d\"), c(\"d\", \"z\", \"b\")) layer <- layer_string_lookup(vocabulary = vocab) i_layer <- layer_string_lookup(vocabulary = vocab, invert = TRUE) int_data <- layer(data) i_layer(int_data) ## tf.Tensor( ## [[b'a' b'c' b'd'] ##  [b'd' b'[UNK]' b'b']], shape=(2, 3), dtype=string)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_subtract.html","id":null,"dir":"Reference","previous_headings":"","what":"Performs elementwise subtraction. — layer_subtract","title":"Performs elementwise subtraction. — layer_subtract","text":"takes input list tensors size 2 shape, returns single tensor (inputs[0] - inputs[1)) shape.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_subtract.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Performs elementwise subtraction. — layer_subtract","text":"","code":"layer_subtract(inputs, ...)"},{"path":"https://keras3.posit.co/dev/reference/layer_subtract.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Performs elementwise subtraction. — layer_subtract","text":"inputs layers combine ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_subtract.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Performs elementwise subtraction. — layer_subtract","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_subtract.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Performs elementwise subtraction. — layer_subtract","text":"Usage Keras model:","code":"input_shape <- c(2, 3, 4) x1 <- random_uniform(input_shape) x2 <- random_uniform(input_shape) y <- layer_subtract(list(x1, x2)) input1 <- layer_input(shape = 16) x1 <- layer_dense(input1, units = 8, activation = 'relu') input2 <- layer_input(shape = 32) x2 <- layer_dense(input2, units = 8, activation = 'relu') subtracted <- layer_subtract(list(x1, x2)) out <- layer_dense(subtracted, units = 4) model <- keras_model(inputs = list(input1, input2), outputs = out)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_text_vectorization.html","id":null,"dir":"Reference","previous_headings":"","what":"A preprocessing layer which maps text features to integer sequences. — layer_text_vectorization","title":"A preprocessing layer which maps text features to integer sequences. — layer_text_vectorization","text":"layer basic options managing text Keras model. transforms batch strings (one example = one string) either list token indices (one example = 1D tensor integer token indices) dense representation (one example = 1D tensor float values representing data example's tokens). layer meant handle natural language inputs. handle simple string inputs (categorical strings pre-tokenized strings) see layer_string_lookup(). vocabulary layer must either supplied construction learned via adapt(). layer adapted, analyze dataset, determine frequency individual string values, create vocabulary . vocabulary can unlimited size capped, depending configuration options layer; unique values input maximum vocabulary size, frequent terms used create vocabulary. processing example contains following steps: Standardize example (usually lowercasing + punctuation stripping) Split example substrings (usually words) Recombine substrings tokens (usually ngrams) Index tokens (associate unique int value token) Transform example using index, either vector ints dense float vector. notes passing callables customize splitting normalization layer: callable can passed Layer, want serialize object pass functions registered Keras serializables (see register_keras_serializable() details). using custom callable standardize, data received callable exactly passed layer. callable return tensor shape input. using custom callable split, data received callable 1st dimension squeezed - instead list(\"string split\", \"another string split\"), Callable see c(\"string split\", \"another string split\"). callable return tf.Tensor dtype string first dimension containing split tokens - example, see something like list(c(\"string\", \"\", \"split\"), c(\"another\", \"string\", \"\", \"split\")). Note: layer uses TensorFlow internally. used part compiled computation graph model backend TensorFlow. can however used backend running eagerly. can also always used part input preprocessing pipeline backend (outside model ), recommend use layer. Note: layer safe use inside tf.data pipeline (independently backend using).","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_text_vectorization.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"A preprocessing layer which maps text features to integer sequences. — layer_text_vectorization","text":"","code":"layer_text_vectorization(   object,   max_tokens = NULL,   standardize = \"lower_and_strip_punctuation\",   split = \"whitespace\",   ngrams = NULL,   output_mode = \"int\",   output_sequence_length = NULL,   pad_to_max_tokens = FALSE,   vocabulary = NULL,   idf_weights = NULL,   sparse = FALSE,   ragged = FALSE,   encoding = \"utf-8\",   name = NULL,   ... )  get_vocabulary(object, include_special_tokens = TRUE)  set_vocabulary(object, vocabulary, idf_weights = NULL, ...)"},{"path":"https://keras3.posit.co/dev/reference/layer_text_vectorization.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"A preprocessing layer which maps text features to integer sequences. — layer_text_vectorization","text":"object Object compose layer . tensor, array, sequential model. max_tokens Maximum size vocabulary layer. specified adapting vocabulary setting pad_to_max_tokens=TRUE. Note vocabulary contains 1 OOV token, effective number tokens (max_tokens - 1 - (1 output_mode == \"int\" else 0)). standardize Optional specification standardization apply input text. Values can : NULL: standardization. \"lower_and_strip_punctuation\": Text lowercased punctuation removed. \"lower\": Text lowercased. \"strip_punctuation\": punctuation removed. Callable: Inputs passed callable function, standardized returned. split Optional specification splitting input text. Values can : NULL: splitting. \"whitespace\": Split whitespace. \"character\": Split unicode character. Callable: Standardized inputs passed callable function, split returned. ngrams Optional specification ngrams create possibly-split input text. Values can NULL, integer list integers; passing integer create ngrams integer, passing list integers create ngrams specified values list. Passing NULL means ngrams created. output_mode Optional specification output layer. Values can \"int\", \"multi_hot\", \"count\" \"tf_idf\", configuring layer follows: \"int\": Outputs integer indices, one integer index per split string token. output_mode == \"int\", 0 reserved masked locations; reduces vocab size max_tokens - 2 instead max_tokens - 1. \"multi_hot\": Outputs single int array per batch, either vocab_size max_tokens size, containing 1s elements token mapped index exists least batch item. \"count\": Like \"multi_hot\", int array contains count number times token index appeared batch item. \"tf_idf\": Like \"multi_hot\", TF-IDF algorithm applied find value token slot. \"int\" output, shape input output supported. output modes, currently rank 1 inputs (rank 2 outputs splitting) supported. output_sequence_length valid INT mode. set, output time dimension padded truncated exactly output_sequence_length values, resulting tensor shape (batch_size, output_sequence_length) regardless many tokens resulted splitting step. Defaults NULL. ragged TRUE output_sequence_length may still truncate output. pad_to_max_tokens valid  \"multi_hot\", \"count\", \"tf_idf\" modes. TRUE, output feature axis padded max_tokens even number unique tokens vocabulary less max_tokens, resulting tensor shape (batch_size, max_tokens) regardless vocabulary size. Defaults FALSE. vocabulary Optional. Either array strings string path text file. passing array, can pass list, list, 1D NumPy array, 1D tensor containing string vocabulary terms. passing file path, file contain one line per term vocabulary. argument set, need adapt() layer. idf_weights R vector, 1D numpy array, 1D tensor inverse document frequency weights equal length vocabulary. Must set output_mode \"tf_idf\". set otherwise. sparse Boolean. applicable \"multi_hot\", \"count\", \"tf_idf\" output modes. supported TensorFlow backend. TRUE, returns SparseTensor instead dense Tensor. Defaults FALSE. ragged Boolean. applicable \"int\" output mode. supported TensorFlow backend. TRUE, returns RaggedTensor instead dense Tensor, sequence may different length string splitting. Defaults FALSE. encoding Optional. text encoding use interpret input strings. Defaults \"utf-8\". name String, name object ... forward/backward compatability. include_special_tokens TRUE, returned vocabulary include padding OOV tokens, term's index vocabulary equal term's index calling layer. FALSE, returned vocabulary include padding OOV tokens.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_text_vectorization.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"A preprocessing layer which maps text features to integer sequences. — layer_text_vectorization","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_text_vectorization.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"A preprocessing layer which maps text features to integer sequences. — layer_text_vectorization","text":"example instantiates TextVectorization layer lowercases text, splits whitespace, strips punctuation, outputs integer vocab indices.         example instantiates TextVectorization layer passing list vocabulary terms layer's initialize() method.","code":"max_tokens <- 5000  # Maximum vocab size. max_len <- 4  # Sequence length to pad the outputs to. # Create the layer. vectorize_layer <- layer_text_vectorization(     max_tokens = max_tokens,     output_mode = 'int',     output_sequence_length = max_len) # Now that the vocab layer has been created, call `adapt` on the # list of strings to create the vocabulary. vectorize_layer %>% adapt(c(\"foo bar\", \"bar baz\", \"baz bada boom\")) # Now, the layer can map strings to integers -- you can use an # embedding layer to map these integers to learned embeddings. input_data <- rbind(\"foo qux bar\", \"qux baz\") vectorize_layer(input_data) ## tf.Tensor( ## [[4 1 3 0] ##  [1 2 0 0]], shape=(2, 4), dtype=int64) vocab_data <- c(\"earth\", \"wind\", \"and\", \"fire\") max_len <- 4  # Sequence length to pad the outputs to. # Create the layer, passing the vocab directly. You can also pass the # vocabulary arg a path to a file containing one vocabulary word per # line. vectorize_layer <- layer_text_vectorization(     max_tokens = max_tokens,     output_mode = 'int',     output_sequence_length = max_len,     vocabulary = vocab_data) # Because we've passed the vocabulary directly, we don't need to adapt # the layer - the vocabulary is already set. The vocabulary contains the # padding token ('') and OOV token ('[UNK]') # as well as the passed tokens. vectorize_layer %>% get_vocabulary() ## [1] \"\"      \"[UNK]\" \"earth\" \"wind\"  \"and\"   \"fire\" # ['', '[UNK]', 'earth', 'wind', 'and', 'fire']"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_tfsm.html","id":null,"dir":"Reference","previous_headings":"","what":"Reload a Keras model/layer that was saved via export_savedmodel(). — layer_tfsm","title":"Reload a Keras model/layer that was saved via export_savedmodel(). — layer_tfsm","text":"Reload Keras model/layer saved via export_savedmodel().","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_tfsm.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Reload a Keras model/layer that was saved via export_savedmodel(). — layer_tfsm","text":"","code":"layer_tfsm(   object,   filepath,   call_endpoint = \"serve\",   call_training_endpoint = NULL,   trainable = TRUE,   name = NULL,   dtype = NULL )"},{"path":"https://keras3.posit.co/dev/reference/layer_tfsm.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Reload a Keras model/layer that was saved via export_savedmodel(). — layer_tfsm","text":"object Object compose layer . tensor, array, sequential model. filepath string, path SavedModel. call_endpoint Name endpoint use call() method reloaded layer. SavedModel created via export_savedmodel(), default endpoint name 'serve'. cases may named 'serving_default'. call_training_endpoint see description trainable see description name String, name object dtype datatype (e.g., \"float32\").","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_tfsm.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Reload a Keras model/layer that was saved via export_savedmodel(). — layer_tfsm","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_tfsm.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Reload a Keras model/layer that was saved via export_savedmodel(). — layer_tfsm","text":"reloaded object can used like regular Keras layer, supports training/fine-tuning trainable weights. Note reloaded object retains none internal structure custom methods original object – brand new layer created around saved function. Limitations: call endpoints single inputs tensor argument (may optionally named list/list tensors) supported. endpoints multiple separate input tensor arguments, consider subclassing layer_tfsm implementing call() method custom signature. need training-time behavior differ inference-time behavior (.e. need reloaded object support training=TRUE argument __call__()), make sure training-time call function saved standalone endpoint artifact, provide name layer_tfsm via call_training_endpoint argument.","code":"model <- keras_model_sequential(input_shape = c(784)) |> layer_dense(10) model |> export_savedmodel(\"path/to/artifact\") ## Saved artifact at 'path/to/artifact'. The following endpoints are available: ## ## * Endpoint 'serve' ##   args_0 (POSITIONAL_ONLY): TensorSpec(shape=(None, 784), dtype=tf.float32, name='keras_tensor') ## Output Type: ##   TensorSpec(shape=(None, 10), dtype=tf.float32, name=None) ## Captures: ##   131671756274576: TensorSpec(shape=(), dtype=tf.resource, name=None) ##   131671756265936: TensorSpec(shape=(), dtype=tf.resource, name=None) reloaded_layer <- layer_tfsm(filepath = \"path/to/artifact\") input <- random_normal(c(2, 784)) output <- reloaded_layer(input) stopifnot(all.equal(as.array(output), as.array(model(input))))"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_time_distributed.html","id":null,"dir":"Reference","previous_headings":"","what":"This wrapper allows to apply a layer to every temporal slice of an input. — layer_time_distributed","title":"This wrapper allows to apply a layer to every temporal slice of an input. — layer_time_distributed","text":"Every input least 3D, dimension index one first input considered temporal dimension. Consider batch 32 video samples, sample 128x128 RGB image channels_last data format, across 10 timesteps. batch input shape (32, 10, 128, 128, 3). can use TimeDistributed apply Conv2D layer 10 timesteps, independently:     layer_time_distributed applies instance layer_conv2d timestamps, set weights used timestamp.","code":"inputs <- keras_input(shape = c(10, 128, 128, 3), batch_size = 32) conv_2d_layer <- layer_conv_2d(filters = 64, kernel_size = c(3, 3)) outputs <- layer_time_distributed(inputs, layer = conv_2d_layer) shape(outputs) ## shape(32, 10, 126, 126, 64)"},{"path":"https://keras3.posit.co/dev/reference/layer_time_distributed.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"This wrapper allows to apply a layer to every temporal slice of an input. — layer_time_distributed","text":"","code":"layer_time_distributed(object, layer, ...)"},{"path":"https://keras3.posit.co/dev/reference/layer_time_distributed.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"This wrapper allows to apply a layer to every temporal slice of an input. — layer_time_distributed","text":"object Object compose layer . tensor, array, sequential model. layer Layer instance. ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_time_distributed.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"This wrapper allows to apply a layer to every temporal slice of an input. — layer_time_distributed","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_time_distributed.html","id":"call-arguments","dir":"Reference","previous_headings":"","what":"Call Arguments","title":"This wrapper allows to apply a layer to every temporal slice of an input. — layer_time_distributed","text":"inputs: Input tensor shape (batch, time, ...) nested tensors, shape (batch, time, ...). training: Boolean indicating whether layer behave training mode inference mode. argument passed wrapped layer (layer supports argument). mask: Binary tensor shape (samples, timesteps) indicating whether given timestep masked. argument passed wrapped layer (layer supports argument).","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_torch_module_wrapper.html","id":null,"dir":"Reference","previous_headings":"","what":"Torch module wrapper layer. — layer_torch_module_wrapper","title":"Torch module wrapper layer. — layer_torch_module_wrapper","text":"layer_torch_module_wrapper wrapper class can turn torch.nn.Module Keras layer, particular making parameters trackable Keras. layer_torch_module_wrapper() compatible PyTorch backend used TensorFlow JAX backends.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_torch_module_wrapper.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Torch module wrapper layer. — layer_torch_module_wrapper","text":"","code":"layer_torch_module_wrapper(   object,   module,   output_shape = NULL,   name = NULL,   ... )"},{"path":"https://keras3.posit.co/dev/reference/layer_torch_module_wrapper.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Torch module wrapper layer. — layer_torch_module_wrapper","text":"object Object compose layer . tensor, array, sequential model. module torch.nn.Module instance. LazyModule instance, parameters must initialized passing instance layer_torch_module_wrapper (e.g. calling ). output_shape Shape output layer. Helps Keras infer shapes. name name layer (string). ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_torch_module_wrapper.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Torch module wrapper layer. — layer_torch_module_wrapper","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_torch_module_wrapper.html","id":"example","dir":"Reference","previous_headings":"","what":"Example","title":"Torch module wrapper layer. — layer_torch_module_wrapper","text":"example layer_torch_module_wrapper() can used vanilla PyTorch modules.","code":"# reticulate::py_install( #   packages = c(\"torch\", \"torchvision\", \"torchaudio\"), #   envname = \"r-keras\", #   pip_options = c(\"--index-url https://download.pytorch.org/whl/cpu\") # ) library(keras3) use_backend(\"torch\") torch <- reticulate::import(\"torch\") nn <- reticulate::import(\"torch.nn\") nnf <- reticulate::import(\"torch.nn.functional\")  Classifier(keras$Model) \\%py_class\\% {   initialize <- function(...) {     super$initialize(...)      self$conv1 <- layer_torch_module_wrapper(module = nn$Conv2d(       in_channels = 1L,       out_channels = 32L,       kernel_size = tuple(3L, 3L)     ))     self$conv2 <- layer_torch_module_wrapper(module = nn$Conv2d(       in_channels = 32L,       out_channels = 64L,       kernel_size = tuple(3L, 3L)     ))     self$pool <- nn$MaxPool2d(kernel_size = tuple(2L, 2L))     self$flatten <- nn$Flatten()     self$dropout <- nn$Dropout(p = 0.5)     self$fc <-       layer_torch_module_wrapper(module = nn$Linear(1600L, 10L))   }    call <- function(inputs) {     x <- nnf$relu(self$conv1(inputs))     x <- self$pool(x)     x <- nnf$relu(self$conv2(x))     x <- self$pool(x)     x <- self$flatten(x)     x <- self$dropout(x)     x <- self$fc(x)     nnf$softmax(x, dim = 1L)   } } model <- Classifier() model$build(shape(1, 28, 28)) cat(\"Output shape:\", format(shape(model(torch$ones(1L, 1L, 28L, 28L)))))  model |> compile(loss = \"sparse_categorical_crossentropy\",                  optimizer = \"adam\",                  metrics = \"accuracy\") model |> fit(train_loader, epochs = 5)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_unit_normalization.html","id":null,"dir":"Reference","previous_headings":"","what":"Unit normalization layer. — layer_unit_normalization","title":"Unit normalization layer. — layer_unit_normalization","text":"Normalize batch inputs input batch L2 norm equal 1 (across axes specified axis).","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_unit_normalization.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Unit normalization layer. — layer_unit_normalization","text":"","code":"layer_unit_normalization(object, axis = -1L, ...)"},{"path":"https://keras3.posit.co/dev/reference/layer_unit_normalization.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Unit normalization layer. — layer_unit_normalization","text":"object Object compose layer . tensor, array, sequential model. axis Integer list. axis axes normalize across. Typically, features axis axes. left-axes typically batch axis axes. -1 last dimension input. Defaults -1. ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_unit_normalization.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Unit normalization layer. — layer_unit_normalization","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_unit_normalization.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Unit normalization layer. — layer_unit_normalization","text":"","code":"data <- 1:6 |> op_array(\"float32\") |> op_reshape(c(2, 3)) normalized_data <- layer_unit_normalization(data) op_sum(normalized_data[1, ]^2) ## tf.Tensor(0.9999999, shape=(), dtype=float32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_upsampling_1d.html","id":null,"dir":"Reference","previous_headings":"","what":"Upsampling layer for 1D inputs. — layer_upsampling_1d","title":"Upsampling layer for 1D inputs. — layer_upsampling_1d","text":"Repeats temporal step size times along time axis.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_upsampling_1d.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Upsampling layer for 1D inputs. — layer_upsampling_1d","text":"","code":"layer_upsampling_1d(object, size = 2L, ...)"},{"path":"https://keras3.posit.co/dev/reference/layer_upsampling_1d.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Upsampling layer for 1D inputs. — layer_upsampling_1d","text":"object Object compose layer . tensor, array, sequential model. size Integer. Upsampling factor. ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_upsampling_1d.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Upsampling layer for 1D inputs. — layer_upsampling_1d","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_upsampling_1d.html","id":"example","dir":"Reference","previous_headings":"","what":"Example","title":"Upsampling layer for 1D inputs. — layer_upsampling_1d","text":"","code":"input_shape <- c(2, 2, 3) x <- seq_len(prod(input_shape)) %>% op_reshape(input_shape) x ## tf.Tensor( ## [[[ 1  2  3] ##   [ 4  5  6]] ## ##  [[ 7  8  9] ##   [10 11 12]]], shape=(2, 2, 3), dtype=int32) y <- layer_upsampling_1d(x, size = 2) y ## tf.Tensor( ## [[[ 1  2  3] ##   [ 1  2  3] ##   [ 4  5  6] ##   [ 4  5  6]] ## ##  [[ 7  8  9] ##   [ 7  8  9] ##   [10 11 12] ##   [10 11 12]]], shape=(2, 4, 3), dtype=int32)"},{"path":"https://keras3.posit.co/dev/reference/layer_upsampling_1d.html","id":"input-shape","dir":"Reference","previous_headings":"","what":"Input Shape","title":"Upsampling layer for 1D inputs. — layer_upsampling_1d","text":"3D tensor shape: (batch_size, steps, features).","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_upsampling_1d.html","id":"output-shape","dir":"Reference","previous_headings":"","what":"Output Shape","title":"Upsampling layer for 1D inputs. — layer_upsampling_1d","text":"3D tensor shape: (batch_size, upsampled_steps, features).","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_upsampling_2d.html","id":null,"dir":"Reference","previous_headings":"","what":"Upsampling layer for 2D inputs. — layer_upsampling_2d","title":"Upsampling layer for 2D inputs. — layer_upsampling_2d","text":"implementation uses interpolative resizing, given resize method (specified interpolation argument). Use interpolation=nearest repeat rows columns data.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_upsampling_2d.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Upsampling layer for 2D inputs. — layer_upsampling_2d","text":"","code":"layer_upsampling_2d(   object,   size = list(2L, 2L),   data_format = NULL,   interpolation = \"nearest\",   ... )"},{"path":"https://keras3.posit.co/dev/reference/layer_upsampling_2d.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Upsampling layer for 2D inputs. — layer_upsampling_2d","text":"object Object compose layer . tensor, array, sequential model. size Int, list 2 integers. upsampling factors rows columns. data_format string, one \"channels_last\" (default) \"channels_first\". ordering dimensions inputs. \"channels_last\" corresponds inputs shape (batch_size, height, width, channels) \"channels_first\" corresponds inputs shape (batch_size, channels, height, width). unspecified, uses image_data_format value found Keras config file ~/.keras/keras.json (exists) else \"channels_last\". Defaults \"channels_last\". interpolation string, one \"bicubic\", \"bilinear\", \"lanczos3\", \"lanczos5\", \"nearest\". ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_upsampling_2d.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Upsampling layer for 2D inputs. — layer_upsampling_2d","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_upsampling_2d.html","id":"example","dir":"Reference","previous_headings":"","what":"Example","title":"Upsampling layer for 2D inputs. — layer_upsampling_2d","text":"","code":"input_shape <- c(2, 2, 1, 3) x <- op_reshape(seq_len(prod(input_shape)), input_shape) print(x) ## tf.Tensor( ## [[[[ 1  2  3]] ## ##   [[ 4  5  6]]] ## ## ##  [[[ 7  8  9]] ## ##   [[10 11 12]]]], shape=(2, 2, 1, 3), dtype=int32) y <- layer_upsampling_2d(x, size = c(1, 2)) print(y) ## tf.Tensor( ## [[[[ 1  2  3] ##    [ 1  2  3]] ## ##   [[ 4  5  6] ##    [ 4  5  6]]] ## ## ##  [[[ 7  8  9] ##    [ 7  8  9]] ## ##   [[10 11 12] ##    [10 11 12]]]], shape=(2, 2, 2, 3), dtype=int32)"},{"path":"https://keras3.posit.co/dev/reference/layer_upsampling_2d.html","id":"input-shape","dir":"Reference","previous_headings":"","what":"Input Shape","title":"Upsampling layer for 2D inputs. — layer_upsampling_2d","text":"4D tensor shape: data_format \"channels_last\": (batch_size, rows, cols, channels) data_format \"channels_first\": (batch_size, channels, rows, cols)","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_upsampling_2d.html","id":"output-shape","dir":"Reference","previous_headings":"","what":"Output Shape","title":"Upsampling layer for 2D inputs. — layer_upsampling_2d","text":"4D tensor shape: data_format \"channels_last\": (batch_size, upsampled_rows, upsampled_cols, channels) data_format \"channels_first\": (batch_size, channels, upsampled_rows, upsampled_cols)","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_upsampling_3d.html","id":null,"dir":"Reference","previous_headings":"","what":"Upsampling layer for 3D inputs. — layer_upsampling_3d","title":"Upsampling layer for 3D inputs. — layer_upsampling_3d","text":"Repeats 1st, 2nd 3rd dimensions data size[0], size[1] size[2] respectively.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_upsampling_3d.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Upsampling layer for 3D inputs. — layer_upsampling_3d","text":"","code":"layer_upsampling_3d(object, size = list(2L, 2L, 2L), data_format = NULL, ...)"},{"path":"https://keras3.posit.co/dev/reference/layer_upsampling_3d.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Upsampling layer for 3D inputs. — layer_upsampling_3d","text":"object Object compose layer . tensor, array, sequential model. size Int, list 3 integers. upsampling factors dim1, dim2 dim3. data_format string, one \"channels_last\" (default) \"channels_first\". ordering dimensions inputs. \"channels_last\" corresponds inputs shape (batch_size, spatial_dim1, spatial_dim2, spatial_dim3, channels) \"channels_first\" corresponds inputs shape (batch_size, channels, spatial_dim1, spatial_dim2, spatial_dim3). unspecified, uses image_data_format value found Keras config file ~/.keras/keras.json (exists) else \"channels_last\". Defaults \"channels_last\". ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_upsampling_3d.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Upsampling layer for 3D inputs. — layer_upsampling_3d","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_upsampling_3d.html","id":"example","dir":"Reference","previous_headings":"","what":"Example","title":"Upsampling layer for 3D inputs. — layer_upsampling_3d","text":"","code":"input_shape <- c(2, 1, 2, 1, 3) x <- array(1, dim = input_shape) y <- layer_upsampling_3d(x, size = c(2, 2, 2)) shape(y) ## shape(2, 2, 4, 2, 3)"},{"path":"https://keras3.posit.co/dev/reference/layer_upsampling_3d.html","id":"input-shape","dir":"Reference","previous_headings":"","what":"Input Shape","title":"Upsampling layer for 3D inputs. — layer_upsampling_3d","text":"5D tensor shape: data_format \"channels_last\": (batch_size, dim1, dim2, dim3, channels) data_format \"channels_first\": (batch_size, channels, dim1, dim2, dim3)","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_upsampling_3d.html","id":"output-shape","dir":"Reference","previous_headings":"","what":"Output Shape","title":"Upsampling layer for 3D inputs. — layer_upsampling_3d","text":"5D tensor shape: data_format \"channels_last\": (batch_size, upsampled_dim1, upsampled_dim2, upsampled_dim3, channels) data_format \"channels_first\": (batch_size, channels, upsampled_dim1, upsampled_dim2, upsampled_dim3)","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_zero_padding_1d.html","id":null,"dir":"Reference","previous_headings":"","what":"Zero-padding layer for 1D input (e.g. temporal sequence). — layer_zero_padding_1d","title":"Zero-padding layer for 1D input (e.g. temporal sequence). — layer_zero_padding_1d","text":"Zero-padding layer 1D input (e.g. temporal sequence).","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_zero_padding_1d.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Zero-padding layer for 1D input (e.g. temporal sequence). — layer_zero_padding_1d","text":"","code":"layer_zero_padding_1d(object, padding = 1L, data_format = NULL, ...)"},{"path":"https://keras3.posit.co/dev/reference/layer_zero_padding_1d.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Zero-padding layer for 1D input (e.g. temporal sequence). — layer_zero_padding_1d","text":"object Object compose layer . tensor, array, sequential model. padding Int, list int (length 2). int: many zeros add beginning end padding dimension (axis 2). list 2 ints: many zeros add beginning end padding dimension ((left_pad, right_pad)). data_format string, one \"channels_last\" (default) \"channels_first\". ordering dimensions inputs. \"channels_last\" corresponds inputs shape (batch_size, axis_to_pad, channels) \"channels_first\" corresponds inputs shape (batch_size, channels, axis_to_pad). unspecified, uses image_data_format value found Keras config file ~/.keras/keras.json (exists). Defaults \"channels_last\". ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_zero_padding_1d.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Zero-padding layer for 1D input (e.g. temporal sequence). — layer_zero_padding_1d","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_zero_padding_1d.html","id":"example","dir":"Reference","previous_headings":"","what":"Example","title":"Zero-padding layer for 1D input (e.g. temporal sequence). — layer_zero_padding_1d","text":"","code":"input_shape <- c(2, 2, 3) x <- op_reshape(seq_len(prod(input_shape)), input_shape) x ## tf.Tensor( ## [[[ 1  2  3] ##   [ 4  5  6]] ## ##  [[ 7  8  9] ##   [10 11 12]]], shape=(2, 2, 3), dtype=int32) y <- layer_zero_padding_1d(x, padding = 2) y ## tf.Tensor( ## [[[ 0  0  0] ##   [ 0  0  0] ##   [ 1  2  3] ##   [ 4  5  6] ##   [ 0  0  0] ##   [ 0  0  0]] ## ##  [[ 0  0  0] ##   [ 0  0  0] ##   [ 7  8  9] ##   [10 11 12] ##   [ 0  0  0] ##   [ 0  0  0]]], shape=(2, 6, 3), dtype=int32)"},{"path":"https://keras3.posit.co/dev/reference/layer_zero_padding_1d.html","id":"input-shape","dir":"Reference","previous_headings":"","what":"Input Shape","title":"Zero-padding layer for 1D input (e.g. temporal sequence). — layer_zero_padding_1d","text":"3D tensor shape: data_format \"channels_last\": (batch_size, axis_to_pad, features) data_format \"channels_first\": (batch_size, features, axis_to_pad)","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_zero_padding_1d.html","id":"output-shape","dir":"Reference","previous_headings":"","what":"Output Shape","title":"Zero-padding layer for 1D input (e.g. temporal sequence). — layer_zero_padding_1d","text":"3D tensor shape: data_format \"channels_last\": (batch_size, padded_axis, features) data_format \"channels_first\": (batch_size, features, padded_axis)","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_zero_padding_2d.html","id":null,"dir":"Reference","previous_headings":"","what":"Zero-padding layer for 2D input (e.g. picture). — layer_zero_padding_2d","title":"Zero-padding layer for 2D input (e.g. picture). — layer_zero_padding_2d","text":"layer can add rows columns zeros top, bottom, left right side image tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_zero_padding_2d.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Zero-padding layer for 2D input (e.g. picture). — layer_zero_padding_2d","text":"","code":"layer_zero_padding_2d(object, padding = list(1L, 1L), data_format = NULL, ...)"},{"path":"https://keras3.posit.co/dev/reference/layer_zero_padding_2d.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Zero-padding layer for 2D input (e.g. picture). — layer_zero_padding_2d","text":"object Object compose layer . tensor, array, sequential model. padding Int, list 2 ints, list 2 lists 2 ints. int: symmetric padding applied height width. list 2 ints: interpreted two different symmetric padding values height width: (symmetric_height_pad, symmetric_width_pad). list 2 lists 2 ints: interpreted ((top_pad, bottom_pad), (left_pad, right_pad)). data_format string, one \"channels_last\" (default) \"channels_first\". ordering dimensions inputs. \"channels_last\" corresponds inputs shape (batch_size, height, width, channels) \"channels_first\" corresponds inputs shape (batch_size, channels, height, width). unspecified, uses image_data_format value found Keras config file ~/.keras/keras.json (exists). Defaults \"channels_last\". ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_zero_padding_2d.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Zero-padding layer for 2D input (e.g. picture). — layer_zero_padding_2d","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_zero_padding_2d.html","id":"example","dir":"Reference","previous_headings":"","what":"Example","title":"Zero-padding layer for 2D input (e.g. picture). — layer_zero_padding_2d","text":"","code":"input_shape <- c(1, 1, 2, 2) x <- op_reshape(seq_len(prod(input_shape)), input_shape) x ## tf.Tensor( ## [[[[1 2] ##    [3 4]]]], shape=(1, 1, 2, 2), dtype=int32) y <- layer_zero_padding_2d(x, padding = 1) y ## tf.Tensor( ## [[[[0 0] ##    [0 0] ##    [0 0] ##    [0 0]] ## ##   [[0 0] ##    [1 2] ##    [3 4] ##    [0 0]] ## ##   [[0 0] ##    [0 0] ##    [0 0] ##    [0 0]]]], shape=(1, 3, 4, 2), dtype=int32)"},{"path":"https://keras3.posit.co/dev/reference/layer_zero_padding_2d.html","id":"input-shape","dir":"Reference","previous_headings":"","what":"Input Shape","title":"Zero-padding layer for 2D input (e.g. picture). — layer_zero_padding_2d","text":"4D tensor shape: data_format \"channels_last\": (batch_size, height, width, channels) data_format \"channels_first\": (batch_size, channels, height, width)","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_zero_padding_2d.html","id":"output-shape","dir":"Reference","previous_headings":"","what":"Output Shape","title":"Zero-padding layer for 2D input (e.g. picture). — layer_zero_padding_2d","text":"4D tensor shape: data_format \"channels_last\": (batch_size, padded_height, padded_width, channels) data_format \"channels_first\": (batch_size, channels, padded_height, padded_width)","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_zero_padding_3d.html","id":null,"dir":"Reference","previous_headings":"","what":"Zero-padding layer for 3D data (spatial or spatio-temporal). — layer_zero_padding_3d","title":"Zero-padding layer for 3D data (spatial or spatio-temporal). — layer_zero_padding_3d","text":"Zero-padding layer 3D data (spatial spatio-temporal).","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_zero_padding_3d.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Zero-padding layer for 3D data (spatial or spatio-temporal). — layer_zero_padding_3d","text":"","code":"layer_zero_padding_3d(   object,   padding = list(list(1L, 1L), list(1L, 1L), list(1L, 1L)),   data_format = NULL,   ... )"},{"path":"https://keras3.posit.co/dev/reference/layer_zero_padding_3d.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Zero-padding layer for 3D data (spatial or spatio-temporal). — layer_zero_padding_3d","text":"object Object compose layer . tensor, array, sequential model. padding Int, list 3 ints, list 3 lists 2 ints. int: symmetric padding applied depth, height, width. list 3 ints: interpreted three different symmetric padding values depth, height, width: (symmetric_dim1_pad, symmetric_dim2_pad, symmetric_dim3_pad). list 3 lists 2 ints: interpreted ((left_dim1_pad, right_dim1_pad), (left_dim2_pad, right_dim2_pad), (left_dim3_pad, right_dim3_pad)). data_format string, one \"channels_last\" (default) \"channels_first\". ordering dimensions inputs. \"channels_last\" corresponds inputs shape (batch_size, spatial_dim1, spatial_dim2, spatial_dim3, channels) \"channels_first\" corresponds inputs shape (batch_size, channels, spatial_dim1, spatial_dim2, spatial_dim3). unspecified, uses image_data_format value found Keras config file ~/.keras/keras.json (exists). Defaults \"channels_last\". ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_zero_padding_3d.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Zero-padding layer for 3D data (spatial or spatio-temporal). — layer_zero_padding_3d","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_zero_padding_3d.html","id":"example","dir":"Reference","previous_headings":"","what":"Example","title":"Zero-padding layer for 3D data (spatial or spatio-temporal). — layer_zero_padding_3d","text":"","code":"input_shape <- c(1, 1, 2, 2, 3) x <- op_reshape(seq_len(prod(input_shape)), input_shape) x ## tf.Tensor( ## [[[[[ 1  2  3] ##     [ 4  5  6]] ## ##    [[ 7  8  9] ##     [10 11 12]]]]], shape=(1, 1, 2, 2, 3), dtype=int32) y <- layer_zero_padding_3d(x, padding = 2) shape(y) ## shape(1, 5, 6, 6, 3)"},{"path":"https://keras3.posit.co/dev/reference/layer_zero_padding_3d.html","id":"input-shape","dir":"Reference","previous_headings":"","what":"Input Shape","title":"Zero-padding layer for 3D data (spatial or spatio-temporal). — layer_zero_padding_3d","text":"5D tensor shape: data_format \"channels_last\": (batch_size, first_axis_to_pad, second_axis_to_pad, third_axis_to_pad, depth) data_format \"channels_first\": (batch_size, depth, first_axis_to_pad, second_axis_to_pad, third_axis_to_pad)","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_zero_padding_3d.html","id":"output-shape","dir":"Reference","previous_headings":"","what":"Output Shape","title":"Zero-padding layer for 3D data (spatial or spatio-temporal). — layer_zero_padding_3d","text":"5D tensor shape: data_format \"channels_last\": (batch_size, first_padded_axis, second_padded_axis, third_axis_to_pad, depth) data_format \"channels_first\": (batch_size, depth, first_padded_axis, second_padded_axis, third_axis_to_pad)","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/learning_rate_schedule_cosine_decay.html","id":null,"dir":"Reference","previous_headings":"","what":"A LearningRateSchedule that uses a cosine decay with optional warmup. — learning_rate_schedule_cosine_decay","title":"A LearningRateSchedule that uses a cosine decay with optional warmup. — learning_rate_schedule_cosine_decay","text":"See Loshchilov & Hutter, ICLR2016, SGDR: Stochastic Gradient Descent Warm Restarts. idea linear warmup learning rate, see Goyal et al.. begin training model, often want initial increase learning rate followed decay. warmup_target int, schedule applies linear increase per optimizer step learning rate initial_learning_rate warmup_target duration warmup_steps. Afterwards, applies cosine decay function taking learning rate warmup_target alpha duration decay_steps. warmup_target NULL skip warmup decay take learning rate initial_learning_rate alpha. requires step value  compute learning rate. can just pass backend variable increment training step. schedule 1-arg callable produces warmup followed decayed learning rate passed current optimizer step. can useful changing learning rate value across different invocations optimizer functions. warmup computed :   decay computed :   Example usage without warmup:   Example usage warmup:   can pass schedule directly optimizer learning rate. learning rate schedule also serializable deserializable using keras$optimizers$schedules$serialize keras$optimizers$schedules$deserialize.","code":"warmup_learning_rate <- function(step) {   completed_fraction <- step / warmup_steps   total_delta <- target_warmup - initial_learning_rate   completed_fraction * total_delta } if (is.null(warmup_target)) {   initial_decay_lr <- initial_learning_rate } else {   initial_decay_lr <- warmup_target }  decayed_learning_rate <- function(step) {   step <- min(step, decay_steps)   cosine_decay <- 0.5 * (1 + cos(pi * step / decay_steps))   decayed <- (1 - alpha) * cosine_decay + alpha   initial_decay_lr * decayed } decay_steps <- 1000 initial_learning_rate <- 0.1 lr_decayed_fn <- learning_rate_schedule_cosine_decay(     initial_learning_rate, decay_steps) decay_steps <- 1000 initial_learning_rate <- 0 warmup_steps <- 1000 target_learning_rate <- 0.1 lr_warmup_decayed_fn <- learning_rate_schedule_cosine_decay(     initial_learning_rate, decay_steps, warmup_target = target_learning_rate,     warmup_steps = warmup_steps )"},{"path":"https://keras3.posit.co/dev/reference/learning_rate_schedule_cosine_decay.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"A LearningRateSchedule that uses a cosine decay with optional warmup. — learning_rate_schedule_cosine_decay","text":"","code":"learning_rate_schedule_cosine_decay(   initial_learning_rate,   decay_steps,   alpha = 0,   name = \"CosineDecay\",   warmup_target = NULL,   warmup_steps = 0L )"},{"path":"https://keras3.posit.co/dev/reference/learning_rate_schedule_cosine_decay.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"A LearningRateSchedule that uses a cosine decay with optional warmup. — learning_rate_schedule_cosine_decay","text":"initial_learning_rate float. initial learning rate. decay_steps int. Number steps decay . alpha float. Minimum learning rate value decay fraction initial_learning_rate. name String. Optional name operation.  Defaults \"CosineDecay\". warmup_target float. target learning rate warmup phase. cast initial_learning_rate datatype. Setting NULL skip warmup begins decay phase initial_learning_rate. Otherwise scheduler warmup initial_learning_rate warmup_target. warmup_steps int. Number steps warmup .","code":""},{"path":"https://keras3.posit.co/dev/reference/learning_rate_schedule_cosine_decay.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"A LearningRateSchedule that uses a cosine decay with optional warmup. — learning_rate_schedule_cosine_decay","text":"1-arg callable learning rate schedule takes current optimizer step outputs decayed learning rate, scalar tensor type initial_learning_rate.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/learning_rate_schedule_cosine_decay_restarts.html","id":null,"dir":"Reference","previous_headings":"","what":"A LearningRateSchedule that uses a cosine decay schedule with restarts. — learning_rate_schedule_cosine_decay_restarts","title":"A LearningRateSchedule that uses a cosine decay schedule with restarts. — learning_rate_schedule_cosine_decay_restarts","text":"See Loshchilov & Hutter, ICLR2016, SGDR: Stochastic Gradient Descent Warm Restarts. training model, often useful lower learning rate training progresses. schedule applies cosine decay function restarts optimizer step, given provided initial learning rate. requires step value compute decayed learning rate. can just pass backend variable increment training step. schedule 1-arg callable produces decayed learning rate passed current optimizer step. can useful changing learning rate value across different invocations optimizer functions. learning rate multiplier first decays 1 alpha first_decay_steps steps. , warm restart performed. new warm restart runs t_mul times steps m_mul times initial learning rate new learning rate.","code":""},{"path":"https://keras3.posit.co/dev/reference/learning_rate_schedule_cosine_decay_restarts.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"A LearningRateSchedule that uses a cosine decay schedule with restarts. — learning_rate_schedule_cosine_decay_restarts","text":"","code":"learning_rate_schedule_cosine_decay_restarts(   initial_learning_rate,   first_decay_steps,   t_mul = 2,   m_mul = 1,   alpha = 0,   name = \"SGDRDecay\" )"},{"path":"https://keras3.posit.co/dev/reference/learning_rate_schedule_cosine_decay_restarts.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"A LearningRateSchedule that uses a cosine decay schedule with restarts. — learning_rate_schedule_cosine_decay_restarts","text":"initial_learning_rate float. initial learning rate. first_decay_steps integer. Number steps decay . t_mul float. Used derive number iterations -th period. m_mul float. Used derive initial learning rate -th period. alpha float. Minimum learning rate value fraction initial_learning_rate. name String. Optional name operation. Defaults \"SGDRDecay\".","code":""},{"path":"https://keras3.posit.co/dev/reference/learning_rate_schedule_cosine_decay_restarts.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"A LearningRateSchedule that uses a cosine decay schedule with restarts. — learning_rate_schedule_cosine_decay_restarts","text":"1-arg callable learning rate schedule takes current optimizer step outputs decayed learning rate, scalar tensor type initial_learning_rate.","code":""},{"path":"https://keras3.posit.co/dev/reference/learning_rate_schedule_cosine_decay_restarts.html","id":"example","dir":"Reference","previous_headings":"","what":"Example","title":"A LearningRateSchedule that uses a cosine decay schedule with restarts. — learning_rate_schedule_cosine_decay_restarts","text":"can pass schedule directly optimizer learning rate. learning rate schedule also serializable deserializable using keras$optimizers$schedules$serialize keras$optimizers$schedules$deserialize.","code":"first_decay_steps <- 1000 lr_decayed_fn <- learning_rate_schedule_cosine_decay_restarts(         0.001,         first_decay_steps)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/learning_rate_schedule_exponential_decay.html","id":null,"dir":"Reference","previous_headings":"","what":"A LearningRateSchedule that uses an exponential decay schedule. — learning_rate_schedule_exponential_decay","title":"A LearningRateSchedule that uses an exponential decay schedule. — learning_rate_schedule_exponential_decay","text":"training model, often useful lower learning rate training progresses. schedule applies exponential decay function optimizer step, given provided initial learning rate. schedule 1-arg callable produces decayed learning rate passed current optimizer step. can useful changing learning rate value across different invocations optimizer functions. computed :   argument staircase TRUE, step / decay_steps integer division decayed learning rate follows staircase function. can pass schedule directly optimizer learning rate.","code":"decayed_learning_rate <- function(step) {   initial_learning_rate * decay_rate ^ (step / decay_steps) }"},{"path":"https://keras3.posit.co/dev/reference/learning_rate_schedule_exponential_decay.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"A LearningRateSchedule that uses an exponential decay schedule. — learning_rate_schedule_exponential_decay","text":"","code":"learning_rate_schedule_exponential_decay(   initial_learning_rate,   decay_steps,   decay_rate,   staircase = FALSE,   name = \"ExponentialDecay\" )"},{"path":"https://keras3.posit.co/dev/reference/learning_rate_schedule_exponential_decay.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"A LearningRateSchedule that uses an exponential decay schedule. — learning_rate_schedule_exponential_decay","text":"initial_learning_rate float. initial learning rate. decay_steps integer. Must positive. See decay computation . decay_rate float. decay rate. staircase Boolean.  TRUE decay learning rate discrete intervals. name String.  Optional name operation.  Defaults \"ExponentialDecay\".","code":""},{"path":"https://keras3.posit.co/dev/reference/learning_rate_schedule_exponential_decay.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"A LearningRateSchedule that uses an exponential decay schedule. — learning_rate_schedule_exponential_decay","text":"1-arg callable learning rate schedule takes current optimizer step outputs decayed learning rate, scalar tensor type initial_learning_rate.","code":""},{"path":"https://keras3.posit.co/dev/reference/learning_rate_schedule_exponential_decay.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"A LearningRateSchedule that uses an exponential decay schedule. — learning_rate_schedule_exponential_decay","text":"fitting Keras model, decay every 100000 steps base 0.96:   learning rate schedule also serializable deserializable using keras$optimizers$schedules$serialize keras$optimizers$schedules$deserialize.","code":"initial_learning_rate <- 0.1 lr_schedule <- learning_rate_schedule_exponential_decay(     initial_learning_rate,     decay_steps=100000,     decay_rate=0.96,     staircase=TRUE)  model %>% compile(   optimizer = optimizer_sgd(learning_rate = lr_schedule),   loss = 'sparse_categorical_crossentropy',   metrics = c('accuracy'))  model %>% fit(data, labels, epochs=5)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/learning_rate_schedule_inverse_time_decay.html","id":null,"dir":"Reference","previous_headings":"","what":"A LearningRateSchedule that uses an inverse time decay schedule. — learning_rate_schedule_inverse_time_decay","title":"A LearningRateSchedule that uses an inverse time decay schedule. — learning_rate_schedule_inverse_time_decay","text":"training model, often useful lower learning rate training progresses. schedule applies inverse decay function optimizer step, given provided initial learning rate. requires step value compute decayed learning rate. can just pass backend variable increment training step. schedule 1-arg callable produces decayed learning rate passed current optimizer step. can useful changing learning rate value across different invocations optimizer functions. computed :   , staircase TRUE, :   can pass schedule directly optimizer_* learning rate.","code":"decayed_learning_rate <- function(step) {   initial_learning_rate / (1 + decay_rate * step / decay_step) } decayed_learning_rate <- function(step) {   initial_learning_rate /            (1 + decay_rate * floor(step / decay_step)) }"},{"path":"https://keras3.posit.co/dev/reference/learning_rate_schedule_inverse_time_decay.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"A LearningRateSchedule that uses an inverse time decay schedule. — learning_rate_schedule_inverse_time_decay","text":"","code":"learning_rate_schedule_inverse_time_decay(   initial_learning_rate,   decay_steps,   decay_rate,   staircase = FALSE,   name = \"InverseTimeDecay\" )"},{"path":"https://keras3.posit.co/dev/reference/learning_rate_schedule_inverse_time_decay.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"A LearningRateSchedule that uses an inverse time decay schedule. — learning_rate_schedule_inverse_time_decay","text":"initial_learning_rate float. initial learning rate. decay_steps often apply decay. decay_rate number.  decay rate. staircase Whether apply decay discrete staircase, o pposed continuous, fashion. name String.  Optional name operation.  Defaults \"InverseTimeDecay\".","code":""},{"path":"https://keras3.posit.co/dev/reference/learning_rate_schedule_inverse_time_decay.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"A LearningRateSchedule that uses an inverse time decay schedule. — learning_rate_schedule_inverse_time_decay","text":"1-arg callable learning rate schedule takes current optimizer step outputs decayed learning rate, scalar tensor type initial_learning_rate.","code":""},{"path":"https://keras3.posit.co/dev/reference/learning_rate_schedule_inverse_time_decay.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"A LearningRateSchedule that uses an inverse time decay schedule. — learning_rate_schedule_inverse_time_decay","text":"Fit Keras model decaying 1/t rate 0.5:","code":"... initial_learning_rate <- 0.1 decay_steps <- 1.0 decay_rate <- 0.5 learning_rate_fn <- learning_rate_schedule_inverse_time_decay(     initial_learning_rate, decay_steps, decay_rate)  model %>% compile(   optimizer = optimizer_sgd(learning_rate=learning_rate_fn),   loss = 'sparse_categorical_crossentropy',   metrics = 'accuracy') )  model %>% fit(data, labels, epochs=5)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/learning_rate_schedule_piecewise_constant_decay.html","id":null,"dir":"Reference","previous_headings":"","what":"A LearningRateSchedule that uses a piecewise constant decay schedule. — learning_rate_schedule_piecewise_constant_decay","title":"A LearningRateSchedule that uses a piecewise constant decay schedule. — learning_rate_schedule_piecewise_constant_decay","text":"function returns 1-arg callable compute piecewise constant passed current optimizer step. can useful changing learning rate value across different invocations optimizer functions.","code":""},{"path":"https://keras3.posit.co/dev/reference/learning_rate_schedule_piecewise_constant_decay.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"A LearningRateSchedule that uses a piecewise constant decay schedule. — learning_rate_schedule_piecewise_constant_decay","text":"","code":"learning_rate_schedule_piecewise_constant_decay(   boundaries,   values,   name = \"PiecewiseConstant\" )"},{"path":"https://keras3.posit.co/dev/reference/learning_rate_schedule_piecewise_constant_decay.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"A LearningRateSchedule that uses a piecewise constant decay schedule. — learning_rate_schedule_piecewise_constant_decay","text":"boundaries list Python numbers strictly increasing entries, elements type optimizer step. values list Python numbers specifies values intervals defined boundaries. one element boundaries, elements type. name string. Optional name operation. Defaults \"PiecewiseConstant\".","code":""},{"path":"https://keras3.posit.co/dev/reference/learning_rate_schedule_piecewise_constant_decay.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"A LearningRateSchedule that uses a piecewise constant decay schedule. — learning_rate_schedule_piecewise_constant_decay","text":"1-arg callable learning rate schedule takes current optimizer step outputs decayed learning rate, scalar tensor type boundary tensors. output 1-arg function takes step values[0] step <= boundaries[0], values[1] step > boundaries[0] step <= boundaries[1], ..., values[-1] step > boundaries[-1].","code":""},{"path":"https://keras3.posit.co/dev/reference/learning_rate_schedule_piecewise_constant_decay.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"A LearningRateSchedule that uses a piecewise constant decay schedule. — learning_rate_schedule_piecewise_constant_decay","text":"use learning rate 1.0 first 100001 steps, 0.5 next 10000 steps, 0.1 additional steps.   can pass schedule directly optimizer learning rate. learning rate schedule also serializable deserializable using keras$optimizers$schedules$serialize keras$optimizers$schedules$deserialize.","code":"step <- 0 boundaries <- c(100000, 110000) values <- c(1.0, 0.5, 0.1) learning_rate_fn <- learning_rate_schedule_piecewise_constant_decay(   boundaries, values)  # Later, whenever we perform an optimization step, we pass in the step. learning_rate <- learning_rate_fn(step)"},{"path":"https://keras3.posit.co/dev/reference/learning_rate_schedule_piecewise_constant_decay.html","id":"raises","dir":"Reference","previous_headings":"","what":"Raises","title":"A LearningRateSchedule that uses a piecewise constant decay schedule. — learning_rate_schedule_piecewise_constant_decay","text":"ValueError: number elements boundaries values lists match.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/learning_rate_schedule_polynomial_decay.html","id":null,"dir":"Reference","previous_headings":"","what":"A LearningRateSchedule that uses a polynomial decay schedule. — learning_rate_schedule_polynomial_decay","title":"A LearningRateSchedule that uses a polynomial decay schedule. — learning_rate_schedule_polynomial_decay","text":"commonly observed monotonically decreasing learning rate, whose degree change carefully chosen, results better performing model. schedule applies polynomial decay function optimizer step, given provided initial_learning_rate, reach end_learning_rate given decay_steps. requires step value compute decayed learning rate. can just pass backend variable increment training step. schedule 1-arg callable produces decayed learning rate passed current optimizer step. can useful changing learning rate value across different invocations optimizer functions. computed :   cycle TRUE multiple decay_steps used, first one bigger step.   can pass schedule directly Optimizer learning rate.","code":"decayed_learning_rate <- function(step) {   step = min(step, decay_steps)   ((initial_learning_rate - end_learning_rate) *     (1 - step / decay_steps) ^ (power)) +     end_learning_rate } decayed_learning_rate <- function(step) {   decay_steps = decay_steps * ceil(step / decay_steps)   ((initial_learning_rate - end_learning_rate) *       (1 - step / decay_steps) ^ (power)) +     end_learning_rate }"},{"path":"https://keras3.posit.co/dev/reference/learning_rate_schedule_polynomial_decay.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"A LearningRateSchedule that uses a polynomial decay schedule. — learning_rate_schedule_polynomial_decay","text":"","code":"learning_rate_schedule_polynomial_decay(   initial_learning_rate,   decay_steps,   end_learning_rate = 1e-04,   power = 1,   cycle = FALSE,   name = \"PolynomialDecay\" )"},{"path":"https://keras3.posit.co/dev/reference/learning_rate_schedule_polynomial_decay.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"A LearningRateSchedule that uses a polynomial decay schedule. — learning_rate_schedule_polynomial_decay","text":"initial_learning_rate float. initial learning rate. decay_steps integer. Must positive. See decay computation . end_learning_rate float. minimal end learning rate. power float. power polynomial. Defaults 1.0. cycle boolean, whether cycle beyond decay_steps. name String.  Optional name operation. Defaults \"PolynomialDecay\".","code":""},{"path":"https://keras3.posit.co/dev/reference/learning_rate_schedule_polynomial_decay.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"A LearningRateSchedule that uses a polynomial decay schedule. — learning_rate_schedule_polynomial_decay","text":"1-arg callable learning rate schedule takes current optimizer step outputs decayed learning rate, scalar tensor type initial_learning_rate.","code":""},{"path":"https://keras3.posit.co/dev/reference/learning_rate_schedule_polynomial_decay.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"A LearningRateSchedule that uses a polynomial decay schedule. — learning_rate_schedule_polynomial_decay","text":"Fit model decaying 0.1 0.01 10000 steps using sqrt (.e. power=0.5):   learning rate schedule also serializable deserializable using keras$optimizers$schedules$serialize keras$optimizers$schedules$deserialize.","code":"... starter_learning_rate <- 0.1 end_learning_rate <- 0.01 decay_steps <- 10000 learning_rate_fn <- learning_rate_schedule_polynomial_decay(     starter_learning_rate,     decay_steps,     end_learning_rate,     power=0.5)  model %>% compile(   optimizer = optimizer_sgd(learning_rate=learning_rate_fn),   loss = 'sparse_categorical_crossentropy',   metrics = 'accuracy' )  model %>% fit(data, labels, epochs=5)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/load_model.html","id":null,"dir":"Reference","previous_headings":"","what":"Loads a model saved via save_model(). — load_model","title":"Loads a model saved via save_model(). — load_model","text":"Loads model saved via save_model().","code":""},{"path":"https://keras3.posit.co/dev/reference/load_model.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Loads a model saved via save_model(). — load_model","text":"","code":"load_model(model, custom_objects = NULL, compile = TRUE, safe_mode = TRUE)"},{"path":"https://keras3.posit.co/dev/reference/load_model.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Loads a model saved via save_model(). — load_model","text":"model string, path saved model file, raw vector, returned save_model(filepath = NULL) custom_objects Optional named list mapping names custom classes functions considered deserialization. compile Boolean, whether compile model loading. safe_mode Boolean, whether disallow unsafe lambda deserialization. safe_mode=FALSE, loading object potential trigger arbitrary code execution. argument applicable Keras v3 model format. Defaults TRUE.","code":""},{"path":"https://keras3.posit.co/dev/reference/load_model.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Loads a model saved via save_model(). — load_model","text":"Keras model instance. original model compiled, argument compile = TRUE set, returned model compiled. Otherwise, model left uncompiled.","code":""},{"path":"https://keras3.posit.co/dev/reference/load_model.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Loads a model saved via save_model(). — load_model","text":"Note model variables may different name values (var$name property, e.g. \"dense_1/kernel:0\") reloaded. recommended use layer attributes access specific variables, e.g. model |> get_layer(\"dense_1\") |> _$kernel.","code":"model <- keras_model_sequential(input_shape = c(3)) |>   layer_dense(5) |>   layer_activation_softmax()  model |> save_model(\"model.keras\") loaded_model <- load_model(\"model.keras\") x <- random_uniform(c(10, 3)) stopifnot(all.equal(   model |> predict(x),   loaded_model |> predict(x) ))"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/load_model_weights.html","id":null,"dir":"Reference","previous_headings":"","what":"Load the weights from a single file or sharded files. — load_model_weights","title":"Load the weights from a single file or sharded files. — load_model_weights","text":"Weights loaded based network's topology. means architecture weights saved. Note layers weights taken account topological ordering, adding removing layers fine long weights. Sharding loading sharded weights, specify filepath ending \".weights.json\" (configuration file), corresponding shard files (*_xxxxx.weights.h5) located alongside . Partial weight loading modified model, instance adding new layer (weights) changing shape weights layer, can choose ignore errors continue loading setting skip_mismatch=TRUE. case layer mismatching weights skipped. warning displayed skipped layer.","code":""},{"path":"https://keras3.posit.co/dev/reference/load_model_weights.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Load the weights from a single file or sharded files. — load_model_weights","text":"","code":"load_model_weights(model, filepath, skip_mismatch = FALSE, ...)"},{"path":"https://keras3.posit.co/dev/reference/load_model_weights.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Load the weights from a single file or sharded files. — load_model_weights","text":"model keras model. filepath Path path-like object weights. Accepts .weights.h5, legacy .h5, sharded weights .weights.json manifest sitting alongside shard files (*_xxxxx.weights.h5). skip_mismatch Boolean, whether skip loading layers mismatch number weights, mismatch shape weights. ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/load_model_weights.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Load the weights from a single file or sharded files. — load_model_weights","text":"called primarily side effects. model returned, invisibly, enable usage pipe.","code":""},{"path":"https://keras3.posit.co/dev/reference/load_model_weights.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Load the weights from a single file or sharded files. — load_model_weights","text":"","code":"model |> load_model_weights(\"model.weights.h5\") model |> load_model_weights(\"model.weights.json\")"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/loss_binary_crossentropy.html","id":null,"dir":"Reference","previous_headings":"","what":"Computes the cross-entropy loss between true labels and predicted labels. — loss_binary_crossentropy","title":"Computes the cross-entropy loss between true labels and predicted labels. — loss_binary_crossentropy","text":"Use cross-entropy loss binary (0 1) classification applications. loss function requires following inputs: y_true (true label): either 0 1. y_pred (predicted value): model's prediction, .e, single floating-point value either represents logit, (.e, value [-inf, inf] from_logits=TRUE) probability (.e, value [0., 1.] from_logits=FALSE).","code":""},{"path":"https://keras3.posit.co/dev/reference/loss_binary_crossentropy.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes the cross-entropy loss between true labels and predicted labels. — loss_binary_crossentropy","text":"","code":"loss_binary_crossentropy(   y_true,   y_pred,   from_logits = FALSE,   label_smoothing = 0,   axis = -1L,   ...,   reduction = \"sum_over_batch_size\",   name = \"binary_crossentropy\",   dtype = NULL )"},{"path":"https://keras3.posit.co/dev/reference/loss_binary_crossentropy.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Computes the cross-entropy loss between true labels and predicted labels. — loss_binary_crossentropy","text":"y_true Ground truth values. shape = [batch_size, d0, .. dN]. y_pred predicted values. shape = [batch_size, d0, .. dN]. from_logits Whether interpret y_pred tensor logit values. default, assume y_pred probabilities (.e., values [0, 1)). label_smoothing Float range [0, 1]. 0, smoothing occurs. > 0, compute loss predicted labels smoothed version true labels, smoothing squeezes labels towards 0.5. Larger values label_smoothing correspond heavier smoothing. axis axis along compute crossentropy (features axis). Defaults -1. ... forward/backward compatability. reduction Type reduction apply loss. almost cases \"sum_over_batch_size\". Supported options \"sum\", \"sum_over_batch_size\", \"mean\", \"mean_with_sample_weight\" NULL. \"sum\" sums loss, \"sum_over_batch_size\" \"mean\" sum loss divide sample size, \"mean_with_sample_weight\" sums loss divides sum sample weights. \"none\" NULL perform aggregation. Defaults \"sum_over_batch_size\". name Optional name loss instance. dtype dtype loss's computations. Defaults NULL, means using config_floatx(). config_floatx() \"float32\" unless set different value (via config_set_floatx()). keras$DTypePolicy provided, compute_dtype utilized.","code":""},{"path":"https://keras3.posit.co/dev/reference/loss_binary_crossentropy.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Computes the cross-entropy loss between true labels and predicted labels. — loss_binary_crossentropy","text":"Binary crossentropy loss value. shape = [batch_size, d0, .. dN-1].","code":""},{"path":"https://keras3.posit.co/dev/reference/loss_binary_crossentropy.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Computes the cross-entropy loss between true labels and predicted labels. — loss_binary_crossentropy","text":"Recommended Usage: (set from_logits=TRUE) compile() API:   standalone function:                     Default Usage: (set from_logits=FALSE)","code":"y_true <- rbind(c(0, 1), c(0, 0)) y_pred <- rbind(c(0.6, 0.4), c(0.4, 0.6)) loss <- loss_binary_crossentropy(y_true, y_pred) loss ## tf.Tensor([0.91629076 0.7135582 ], shape=(2), dtype=float32) model %>% compile(     loss = loss_binary_crossentropy(from_logits=TRUE),     ... ) # Example 1: (batch_size = 1, number of samples = 4) y_true <- op_array(c(0, 1, 0, 0)) y_pred <- op_array(c(-18.6, 0.51, 2.94, -12.8)) bce <- loss_binary_crossentropy(from_logits = TRUE) bce(y_true, y_pred) ## tf.Tensor(0.865458, shape=(), dtype=float32) # Example 2: (batch_size = 2, number of samples = 4) y_true <- rbind(c(0, 1), c(0, 0)) y_pred <- rbind(c(-18.6, 0.51), c(2.94, -12.8)) # Using default 'auto'/'sum_over_batch_size' reduction type. bce <- loss_binary_crossentropy(from_logits = TRUE) bce(y_true, y_pred) ## tf.Tensor(0.865458, shape=(), dtype=float32) # Using 'sample_weight' attribute bce(y_true, y_pred, sample_weight = c(0.8, 0.2)) ## tf.Tensor(0.2436386, shape=(), dtype=float32) # 0.243 # Using 'sum' reduction` type. bce <- loss_binary_crossentropy(from_logits = TRUE, reduction = \"sum\") bce(y_true, y_pred) ## tf.Tensor(1.730916, shape=(), dtype=float32) # Using 'none' reduction type. bce <- loss_binary_crossentropy(from_logits = TRUE, reduction = NULL) bce(y_true, y_pred) ## tf.Tensor([0.23515666 1.4957594 ], shape=(2), dtype=float32) # Make the following updates to the above \"Recommended Usage\" section # 1. Set `from_logits=FALSE` loss_binary_crossentropy() # OR ...('from_logits=FALSE') ## <LossFunctionWrapper(<function binary_crossentropy at 0x0>, kwargs={'from_logits': False, 'label_smoothing': 0.0, 'axis': -1})> ##  signature: (y_true, y_pred, sample_weight=None) # 2. Update `y_pred` to use probabilities instead of logits y_pred <- c(0.6, 0.3, 0.2, 0.8) # OR [[0.6, 0.3], [0.2, 0.8]]"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/loss_binary_focal_crossentropy.html","id":null,"dir":"Reference","previous_headings":"","what":"Computes focal cross-entropy loss between true labels and predictions. — loss_binary_focal_crossentropy","title":"Computes focal cross-entropy loss between true labels and predictions. — loss_binary_focal_crossentropy","text":"According Lin et al., 2018, helps apply focal factor -weight easy examples focus hard examples. default, focal tensor computed follows: focal_factor = (1 - output)^gamma class 1 focal_factor = output^gamma class 0 gamma focusing parameter. gamma = 0, focal effect binary crossentropy loss. apply_class_balancing == TRUE, function also takes account weight balancing factor binary classes 0 1 follows: weight = alpha class 1 (target == 1) weight = 1 - alpha class 0 alpha float range [0, 1]. Binary cross-entropy loss often used binary (0 1) classification tasks. loss function requires following inputs: y_true (true label): either 0 1. y_pred (predicted value): model's prediction, .e, single floating-point value either represents logit, (.e, value [-inf, inf] from_logits=TRUE) probability (.e, value [0., 1.] from_logits=FALSE). According Lin et al., 2018, helps apply \"focal factor\" -weight easy examples focus hard examples. default, focal tensor computed follows: focal_factor = (1 - output) ** gamma class 1 focal_factor = output ** gamma class 0 gamma focusing parameter. gamma=0, function equivalent binary crossentropy loss.","code":""},{"path":"https://keras3.posit.co/dev/reference/loss_binary_focal_crossentropy.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes focal cross-entropy loss between true labels and predictions. — loss_binary_focal_crossentropy","text":"","code":"loss_binary_focal_crossentropy(   y_true,   y_pred,   apply_class_balancing = FALSE,   alpha = 0.25,   gamma = 2,   from_logits = FALSE,   label_smoothing = 0,   axis = -1L,   ...,   reduction = \"sum_over_batch_size\",   name = \"binary_focal_crossentropy\",   dtype = NULL )"},{"path":"https://keras3.posit.co/dev/reference/loss_binary_focal_crossentropy.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Computes focal cross-entropy loss between true labels and predictions. — loss_binary_focal_crossentropy","text":"y_true Ground truth values, shape (batch_size, d0, .. dN). y_pred predicted values, shape (batch_size, d0, .. dN). apply_class_balancing bool, whether apply weight balancing binary classes 0 1. alpha weight balancing factor class 1, default 0.25 mentioned reference Lin et al., 2018.  weight class 0 1.0 - alpha. gamma focusing parameter used compute focal factor, default 2.0 mentioned reference Lin et al., 2018. from_logits Whether interpret y_pred tensor logit values. default, assume y_pred probabilities (.e., values [0, 1]). label_smoothing Float [0, 1]. 0, smoothing occurs. > 0, compute loss predicted labels smoothed version true labels, smoothing squeezes labels towards 0.5. Larger values label_smoothing correspond heavier smoothing. axis axis along compute crossentropy (features axis). Defaults -1. ... forward/backward compatability. reduction Type reduction apply loss. almost cases \"sum_over_batch_size\". Supported options \"sum\", \"sum_over_batch_size\", \"mean\", \"mean_with_sample_weight\" NULL. \"sum\" sums loss, \"sum_over_batch_size\" \"mean\" sum loss divide sample size, \"mean_with_sample_weight\" sums loss divides sum sample weights. \"none\" NULL perform aggregation. Defaults \"sum_over_batch_size\". name Optional name loss instance. dtype dtype loss's computations. Defaults NULL, means using config_floatx(). config_floatx() \"float32\" unless set different value (via config_set_floatx()). keras$DTypePolicy provided, compute_dtype utilized.","code":""},{"path":"https://keras3.posit.co/dev/reference/loss_binary_focal_crossentropy.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Computes focal cross-entropy loss between true labels and predictions. — loss_binary_focal_crossentropy","text":"Binary focal crossentropy loss value shape = [batch_size, d0, .. dN-1].","code":""},{"path":"https://keras3.posit.co/dev/reference/loss_binary_focal_crossentropy.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Computes focal cross-entropy loss between true labels and predictions. — loss_binary_focal_crossentropy","text":"compile() API:   standalone function:","code":"y_true <- rbind(c(0, 1), c(0, 0)) y_pred <- rbind(c(0.6, 0.4), c(0.4, 0.6)) loss <- loss_binary_focal_crossentropy(y_true, y_pred, gamma = 2) loss ## tf.Tensor([0.32986468 0.20579839], shape=(2), dtype=float32) model %>% compile(     loss = loss_binary_focal_crossentropy(         gamma = 2.0, from_logits = TRUE),     ... ) # Example 1: (batch_size = 1, number of samples = 4) y_true <- op_array(c(0, 1, 0, 0)) y_pred <- op_array(c(-18.6, 0.51, 2.94, -12.8)) loss <- loss_binary_focal_crossentropy(gamma = 2, from_logits = TRUE) loss(y_true, y_pred) ## tf.Tensor(0.6912122, shape=(), dtype=float32) # Apply class weight loss <- loss_binary_focal_crossentropy(   apply_class_balancing = TRUE, gamma = 2, from_logits = TRUE) loss(y_true, y_pred) ## tf.Tensor(0.5101333, shape=(), dtype=float32) # Example 2: (batch_size = 2, number of samples = 4) y_true <- rbind(c(0, 1), c(0, 0)) y_pred <- rbind(c(-18.6, 0.51), c(2.94, -12.8)) # Using default 'auto'/'sum_over_batch_size' reduction type. loss <- loss_binary_focal_crossentropy(     gamma = 3, from_logits = TRUE) loss(y_true, y_pred) ## tf.Tensor(0.6469951, shape=(), dtype=float32) # Apply class weight loss <- loss_binary_focal_crossentropy(      apply_class_balancing = TRUE, gamma = 3, from_logits = TRUE) loss(y_true, y_pred) ## tf.Tensor(0.48214132, shape=(), dtype=float32) # Using 'sample_weight' attribute with focal effect loss <- loss_binary_focal_crossentropy(     gamma = 3, from_logits = TRUE) loss(y_true, y_pred, sample_weight = c(0.8, 0.2)) ## tf.Tensor(0.13312504, shape=(), dtype=float32) # Apply class weight loss <- loss_binary_focal_crossentropy(      apply_class_balancing = TRUE, gamma = 3, from_logits = TRUE) loss(y_true, y_pred, sample_weight = c(0.8, 0.2)) ## tf.Tensor(0.09735977, shape=(), dtype=float32) # Using 'sum' reduction` type. loss <- loss_binary_focal_crossentropy(     gamma = 4, from_logits = TRUE,     reduction = \"sum\") loss(y_true, y_pred) ## tf.Tensor(1.2218808, shape=(), dtype=float32) # Apply class weight loss <- loss_binary_focal_crossentropy(     apply_class_balancing = TRUE, gamma = 4, from_logits = TRUE,     reduction = \"sum\") loss(y_true, y_pred) ## tf.Tensor(0.9140807, shape=(), dtype=float32) # Using 'none' reduction type. loss <- loss_binary_focal_crossentropy(     gamma = 5, from_logits = TRUE,     reduction = NULL) loss(y_true, y_pred) ## tf.Tensor([0.00174837 1.1561027 ], shape=(2), dtype=float32) # Apply class weight loss <- loss_binary_focal_crossentropy(     apply_class_balancing = TRUE, gamma = 5, from_logits = TRUE,     reduction = NULL) loss(y_true, y_pred) ## tf.Tensor([4.3709317e-04 8.6707699e-01], shape=(2), dtype=float32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/loss_categorical_crossentropy.html","id":null,"dir":"Reference","previous_headings":"","what":"Computes the crossentropy loss between the labels and predictions. — loss_categorical_crossentropy","title":"Computes the crossentropy loss between the labels and predictions. — loss_categorical_crossentropy","text":"Use crossentropy loss function two label classes. expect labels provided one_hot representation. want provide labels integers, please use SparseCategoricalCrossentropy loss. num_classes floating point values per feature, .e., shape y_pred y_true [batch_size, num_classes].","code":""},{"path":"https://keras3.posit.co/dev/reference/loss_categorical_crossentropy.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes the crossentropy loss between the labels and predictions. — loss_categorical_crossentropy","text":"","code":"loss_categorical_crossentropy(   y_true,   y_pred,   from_logits = FALSE,   label_smoothing = 0,   axis = -1L,   ...,   reduction = \"sum_over_batch_size\",   name = \"categorical_crossentropy\",   dtype = NULL )"},{"path":"https://keras3.posit.co/dev/reference/loss_categorical_crossentropy.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Computes the crossentropy loss between the labels and predictions. — loss_categorical_crossentropy","text":"y_true Tensor one-hot true targets. y_pred Tensor predicted targets. from_logits Whether y_pred expected logits tensor. default, assume y_pred encodes probability distribution. label_smoothing Float [0, 1]. > 0, label values smoothed, meaning confidence label values relaxed. example, 0.1, use 0.1 / num_classes non-target labels 0.9 + 0.1 / num_classes target labels. axis axis along compute crossentropy (features axis). Defaults -1. ... forward/backward compatability. reduction Type reduction apply loss. almost cases \"sum_over_batch_size\". Supported options \"sum\", \"sum_over_batch_size\", \"mean\", \"mean_with_sample_weight\" NULL. \"sum\" sums loss, \"sum_over_batch_size\" \"mean\" sum loss divide sample size, \"mean_with_sample_weight\" sums loss divides sum sample weights. \"none\" NULL perform aggregation. Defaults \"sum_over_batch_size\". name Optional name loss instance. dtype dtype loss's computations. Defaults NULL, means using config_floatx(). config_floatx() \"float32\" unless set different value (via config_set_floatx()). keras$DTypePolicy provided, compute_dtype utilized.","code":""},{"path":"https://keras3.posit.co/dev/reference/loss_categorical_crossentropy.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Computes the crossentropy loss between the labels and predictions. — loss_categorical_crossentropy","text":"Categorical crossentropy loss value.","code":""},{"path":"https://keras3.posit.co/dev/reference/loss_categorical_crossentropy.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Computes the crossentropy loss between the labels and predictions. — loss_categorical_crossentropy","text":"Standalone usage:                 Usage compile() API:","code":"y_true <- rbind(c(0, 1, 0), c(0, 0, 1)) y_pred <- rbind(c(0.05, 0.95, 0), c(0.1, 0.8, 0.1)) loss <- loss_categorical_crossentropy(y_true, y_pred) loss ## tf.Tensor([0.05129331 2.3025851 ], shape=(2), dtype=float32) y_true <- rbind(c(0, 1, 0), c(0, 0, 1)) y_pred <- rbind(c(0.05, 0.95, 0), c(0.1, 0.8, 0.1)) # Using 'auto'/'sum_over_batch_size' reduction type. cce <- loss_categorical_crossentropy() cce(y_true, y_pred) ## tf.Tensor(1.1769392, shape=(), dtype=float32) # Calling with 'sample_weight'. cce(y_true, y_pred, sample_weight = op_array(c(0.3, 0.7))) ## tf.Tensor(0.8135988, shape=(), dtype=float32) # Using 'sum' reduction type. cce <- loss_categorical_crossentropy(reduction = \"sum\") cce(y_true, y_pred) ## tf.Tensor(2.3538785, shape=(), dtype=float32) # Using 'none' reduction type. cce <- loss_categorical_crossentropy(reduction = NULL) cce(y_true, y_pred) ## tf.Tensor([0.05129331 2.3025851 ], shape=(2), dtype=float32) model %>% compile(optimizer = 'sgd',               loss=loss_categorical_crossentropy())"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/loss_categorical_focal_crossentropy.html","id":null,"dir":"Reference","previous_headings":"","what":"Computes the alpha balanced focal crossentropy loss. — loss_categorical_focal_crossentropy","title":"Computes the alpha balanced focal crossentropy loss. — loss_categorical_focal_crossentropy","text":"Use crossentropy loss function two label classes want handle class imbalance without using class_weights. expect labels provided one_hot representation. According Lin et al., 2018, helps apply focal factor -weight easy examples focus hard examples. general formula focal loss (FL) follows: FL(p_t) = (1 - p_t)^gamma * log(p_t) p_t defined follows: p_t = output y_true == 1, else 1 - output (1 - p_t)^gamma modulating_factor, gamma focusing parameter. gamma = 0, focal effect cross entropy. gamma reduces importance given simple examples smooth manner. authors use alpha-balanced variant focal loss (FL) paper: FL(p_t) = -alpha * (1 - p_t)^gamma * log(p_t) alpha weight factor classes. alpha = 1, loss able handle class imbalance properly classes weight. can constant list constants. alpha list, must length number classes. formula can generalized : FL(p_t) = alpha * (1 - p_t)^gamma * CrossEntropy(y_true, y_pred) minus comes CrossEntropy(y_true, y_pred) (CE). Extending multi-class case straightforward: FL(p_t) = alpha * (1 - p_t) ** gamma * CategoricalCE(y_true, y_pred) snippet , num_classes floating pointing values per example. shape y_pred y_true (batch_size, num_classes).","code":""},{"path":"https://keras3.posit.co/dev/reference/loss_categorical_focal_crossentropy.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes the alpha balanced focal crossentropy loss. — loss_categorical_focal_crossentropy","text":"","code":"loss_categorical_focal_crossentropy(   y_true,   y_pred,   alpha = 0.25,   gamma = 2,   from_logits = FALSE,   label_smoothing = 0,   axis = -1L,   ...,   reduction = \"sum_over_batch_size\",   name = \"categorical_focal_crossentropy\",   dtype = NULL )"},{"path":"https://keras3.posit.co/dev/reference/loss_categorical_focal_crossentropy.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Computes the alpha balanced focal crossentropy loss. — loss_categorical_focal_crossentropy","text":"y_true Tensor one-hot true targets. y_pred Tensor predicted targets. alpha weight balancing factor classes, default 0.25 mentioned reference. can list floats scalar. multi-class case, alpha may set inverse class frequency using compute_class_weight sklearn.utils. gamma focusing parameter, default 2.0 mentioned reference. helps gradually reduce importance given simple examples smooth manner. gamma = 0, focal effect categorical crossentropy. from_logits Whether output expected logits tensor. default, consider output encodes probability distribution. label_smoothing Float [0, 1]. > 0, label values smoothed, meaning confidence label values relaxed. example, 0.1, use 0.1 / num_classes non-target labels 0.9 + 0.1 / num_classes target labels. axis axis along compute crossentropy (features axis). Defaults -1. ... forward/backward compatability. reduction Type reduction apply loss. almost cases \"sum_over_batch_size\". Supported options \"sum\", \"sum_over_batch_size\", \"mean\", \"mean_with_sample_weight\" NULL. \"sum\" sums loss, \"sum_over_batch_size\" \"mean\" sum loss divide sample size, \"mean_with_sample_weight\" sums loss divides sum sample weights. \"none\" NULL perform aggregation. Defaults \"sum_over_batch_size\". name Optional name loss instance. dtype dtype loss's computations. Defaults NULL, means using config_floatx(). config_floatx() \"float32\" unless set different value (via config_set_floatx()). keras$DTypePolicy provided, compute_dtype utilized.","code":""},{"path":"https://keras3.posit.co/dev/reference/loss_categorical_focal_crossentropy.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Computes the alpha balanced focal crossentropy loss. — loss_categorical_focal_crossentropy","text":"Categorical focal crossentropy loss value.","code":""},{"path":"https://keras3.posit.co/dev/reference/loss_categorical_focal_crossentropy.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Computes the alpha balanced focal crossentropy loss. — loss_categorical_focal_crossentropy","text":"Standalone usage:                 Usage compile() API:","code":"y_true <- rbind(c(0, 1, 0), c(0, 0, 1)) y_pred <- rbind(c(0.05, 0.95, 0), c(0.1, 0.8, 0.1)) loss <- loss_categorical_focal_crossentropy(y_true, y_pred) loss ## tf.Tensor([3.2058331e-05 4.6627346e-01], shape=(2), dtype=float32) y_true <- rbind(c(0, 1, 0), c(0, 0, 1)) y_pred <- rbind(c(0.05, 0.95, 0), c(0.1, 0.8, 0.1)) # Using 'auto'/'sum_over_batch_size' reduction type. cce <- loss_categorical_focal_crossentropy() cce(y_true, y_pred) ## tf.Tensor(0.23315276, shape=(), dtype=float32) # Calling with 'sample_weight'. cce(y_true, y_pred, sample_weight = op_array(c(0.3, 0.7))) ## tf.Tensor(0.16320053, shape=(), dtype=float32) # Using 'sum' reduction type. cce <- loss_categorical_focal_crossentropy(reduction = \"sum\") cce(y_true, y_pred) ## tf.Tensor(0.46630552, shape=(), dtype=float32) # Using 'none' reduction type. cce <- loss_categorical_focal_crossentropy(reduction = NULL) cce(y_true, y_pred) ## tf.Tensor([3.2058331e-05 4.6627346e-01], shape=(2), dtype=float32) model %>% compile(   optimizer = 'adam',   loss = loss_categorical_focal_crossentropy())"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/loss_categorical_generalized_cross_entropy.html","id":null,"dir":"Reference","previous_headings":"","what":"Computes the generalized cross entropy loss. — loss_categorical_generalized_cross_entropy","title":"Computes the generalized cross entropy loss. — loss_categorical_generalized_cross_entropy","text":"generalized cross entropy (GCE) loss offers robustness noisy labels interpolating categorical cross entropy (q -> 0) mean absolute error (q -> 1). true-class probability p noise parameter q, loss loss = (1 - p^q) / q.","code":""},{"path":"https://keras3.posit.co/dev/reference/loss_categorical_generalized_cross_entropy.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes the generalized cross entropy loss. — loss_categorical_generalized_cross_entropy","text":"","code":"loss_categorical_generalized_cross_entropy(   y_true,   y_pred,   q = 0.5,   ...,   reduction = \"sum_over_batch_size\",   name = \"categorical_generalized_cross_entropy\",   dtype = NULL )"},{"path":"https://keras3.posit.co/dev/reference/loss_categorical_generalized_cross_entropy.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Computes the generalized cross entropy loss. — loss_categorical_generalized_cross_entropy","text":"y_true Integer class indices shape (batch_size) (batch_size, 1). y_pred Predicted class probabilities shape (batch_size, num_classes). q Float (0, 1). Controls transition cross entropy mean absolute error. Defaults 0.5. q approaches 0: behaves like categorical cross entropy. q approaches 1: behaves like mean absolute error. ... forward/backward compatibility. reduction Type reduction apply loss. almost cases \"sum_over_batch_size\". Supported options \"sum\", \"sum_over_batch_size\", \"mean\", \"mean_with_sample_weight\" NULL. \"sum\" sums loss, \"sum_over_batch_size\" \"mean\" sum loss divide sample size, \"mean_with_sample_weight\" sums loss divides sum sample weights. \"none\" NULL perform aggregation. Defaults \"sum_over_batch_size\". name Optional name loss instance. dtype Dtype used loss computations. Defaults config_floatx() (global float type).","code":""},{"path":"https://keras3.posit.co/dev/reference/loss_categorical_generalized_cross_entropy.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Computes the generalized cross entropy loss. — loss_categorical_generalized_cross_entropy","text":"Generalized cross entropy loss value(s).","code":""},{"path":"https://keras3.posit.co/dev/reference/loss_categorical_generalized_cross_entropy.html","id":"references","dir":"Reference","previous_headings":"","what":"References","title":"Computes the generalized cross entropy loss. — loss_categorical_generalized_cross_entropy","text":"Zhang & Sabuncu (2018), \"Generalized Cross Entropy Loss Training Deep Neural Networks Noisy Labels\"","code":""},{"path":"https://keras3.posit.co/dev/reference/loss_categorical_generalized_cross_entropy.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Computes the generalized cross entropy loss. — loss_categorical_generalized_cross_entropy","text":"","code":"y_true <- c(0L, 1L, 0L, 1L) y_pred <- rbind(   c(0.7, 0.3),   c(0.2, 0.8),   c(0.6, 0.4),   c(0.4, 0.6) ) gce <- loss_categorical_generalized_cross_entropy(q = 0.7) gce(y_true, y_pred) ## tf.Tensor(0.34529287, shape=(), dtype=float32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/loss_categorical_hinge.html","id":null,"dir":"Reference","previous_headings":"","what":"Computes the categorical hinge loss between y_true & y_pred. — loss_categorical_hinge","title":"Computes the categorical hinge loss between y_true & y_pred. — loss_categorical_hinge","text":"Formula:   neg=maximum((1-y_true)*y_pred) pos=sum(y_true*y_pred)","code":"loss <- maximum(neg - pos + 1, 0)"},{"path":"https://keras3.posit.co/dev/reference/loss_categorical_hinge.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes the categorical hinge loss between y_true & y_pred. — loss_categorical_hinge","text":"","code":"loss_categorical_hinge(   y_true,   y_pred,   ...,   reduction = \"sum_over_batch_size\",   name = \"categorical_hinge\",   dtype = NULL )"},{"path":"https://keras3.posit.co/dev/reference/loss_categorical_hinge.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Computes the categorical hinge loss between y_true & y_pred. — loss_categorical_hinge","text":"y_true ground truth values. y_true values expected either {-1, +1} {0, 1} (.e. one-hot-encoded tensor) shape <- [batch_size, d0, .. dN]. y_pred predicted values shape = [batch_size, d0, .. dN]. ... forward/backward compatability. reduction Type reduction apply loss. almost cases \"sum_over_batch_size\". Supported options \"sum\", \"sum_over_batch_size\", \"mean\", \"mean_with_sample_weight\" NULL. \"sum\" sums loss, \"sum_over_batch_size\" \"mean\" sum loss divide sample size, \"mean_with_sample_weight\" sums loss divides sum sample weights. \"none\" NULL perform aggregation. Defaults \"sum_over_batch_size\". name Optional name loss instance. dtype dtype loss's computations. Defaults NULL, means using config_floatx(). config_floatx() \"float32\" unless set different value (via config_set_floatx()). keras$DTypePolicy provided, compute_dtype utilized.","code":""},{"path":"https://keras3.posit.co/dev/reference/loss_categorical_hinge.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Computes the categorical hinge loss between y_true & y_pred. — loss_categorical_hinge","text":"Categorical hinge loss values shape = [batch_size, d0, .. dN-1].","code":""},{"path":"https://keras3.posit.co/dev/reference/loss_categorical_hinge.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Computes the categorical hinge loss between y_true & y_pred. — loss_categorical_hinge","text":"","code":"y_true <- rbind(c(0, 1), c(0, 0)) y_pred <- rbind(c(0.6, 0.4), c(0.4, 0.6)) loss <- loss_categorical_hinge(y_true, y_pred)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/loss_circle.html","id":null,"dir":"Reference","previous_headings":"","what":"Computes Circle Loss between integer labels and L2-normalized embeddings. — loss_circle","title":"Computes Circle Loss between integer labels and L2-normalized embeddings. — loss_circle","text":"designed minimize within-class distances maximize -class distances L2 normalized embedding space. metric learning loss designed minimize within-class distance maximize -class distance flexible manner dynamically adjusting penalty strength based optimization status similarity score. use Circle Loss effectively, model output embeddings without activation function (Dense layer activation=NULL) followed UnitNormalization layer ensure unit-norm embeddings.","code":""},{"path":"https://keras3.posit.co/dev/reference/loss_circle.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes Circle Loss between integer labels and L2-normalized embeddings. — loss_circle","text":"","code":"loss_circle(   y_true,   y_pred,   ref_labels = NULL,   ref_embeddings = NULL,   remove_diagonal = TRUE,   gamma = 80L,   margin = 0.4,   ...,   reduction = \"sum_over_batch_size\",   name = \"circle\",   dtype = NULL )"},{"path":"https://keras3.posit.co/dev/reference/loss_circle.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Computes Circle Loss between integer labels and L2-normalized embeddings. — loss_circle","text":"y_true Tensor ground truth labels integer format. y_pred Tensor predicted L2 normalized embeddings. ref_labels Optional integer tensor labels reference embeddings. NULL, defaults y_true. ref_embeddings Optional tensor L2 normalized reference embeddings. NULL, defaults y_pred. remove_diagonal Boolean, whether remove self-similarities positive mask. Defaults TRUE. gamma Scaling factor determines largest scale similarity score. Defaults 80. margin relaxation factor, distance, negatives weighted positives weighted. Similarly, distance negatives weighted positive weighted. Defaults 0.4. ... forward/backward compatability. reduction Type reduction apply loss. almost cases \"sum_over_batch_size\". Supported options \"sum\", \"sum_over_batch_size\", \"mean\", \"mean_with_sample_weight\" NULL. \"sum\" sums loss, \"sum_over_batch_size\" \"mean\" sum loss divide sample size, \"mean_with_sample_weight\" sums loss divides sum sample weights. \"none\" NULL perform aggregation. Defaults \"sum_over_batch_size\". name Optional name loss instance. dtype dtype loss's computations. Defaults NULL, means using config_floatx(). config_floatx() \"float32\" unless set different value (via config_set_floatx()). keras.DTypePolicy provided, compute_dtype utilized.","code":""},{"path":"https://keras3.posit.co/dev/reference/loss_circle.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Computes Circle Loss between integer labels and L2-normalized embeddings. — loss_circle","text":"Circle loss value.","code":""},{"path":"https://keras3.posit.co/dev/reference/loss_circle.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Computes Circle Loss between integer labels and L2-normalized embeddings. — loss_circle","text":"Usage compile() API:","code":"model <- keras_model_sequential(input_shape = c(224, 224, 3)) |>   layer_conv_2d(16, c(3, 3), activation = 'relu') |>   layer_flatten() |>   layer_dense(64, activation = NULL) |>   # No activation   layer_unit_normalization()  # L2 normalization  model |>   compile(optimizer = \"adam\", loss = loss_circle())"},{"path":"https://keras3.posit.co/dev/reference/loss_circle.html","id":"reference","dir":"Reference","previous_headings":"","what":"Reference","title":"Computes Circle Loss between integer labels and L2-normalized embeddings. — loss_circle","text":"Yifan Sun et al., 2020","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/loss_cosine_similarity.html","id":null,"dir":"Reference","previous_headings":"","what":"Computes the cosine similarity between y_true & y_pred. — loss_cosine_similarity","title":"Computes the cosine similarity between y_true & y_pred. — loss_cosine_similarity","text":"Formula:   Note number -1 1. negative number -1 0, 0 indicates orthogonality values closer -1 indicate greater similarity. makes usable loss function setting try maximize proximity predictions targets. either y_true y_pred zero vector, cosine similarity 0 regardless proximity predictions targets.","code":"loss <- -sum(l2_norm(y_true) * l2_norm(y_pred))"},{"path":"https://keras3.posit.co/dev/reference/loss_cosine_similarity.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes the cosine similarity between y_true & y_pred. — loss_cosine_similarity","text":"","code":"loss_cosine_similarity(   y_true,   y_pred,   axis = -1L,   ...,   reduction = \"sum_over_batch_size\",   name = \"cosine_similarity\",   dtype = NULL )"},{"path":"https://keras3.posit.co/dev/reference/loss_cosine_similarity.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Computes the cosine similarity between y_true & y_pred. — loss_cosine_similarity","text":"y_true Tensor true targets. y_pred Tensor predicted targets. axis axis along cosine similarity computed (features axis). Defaults -1. ... forward/backward compatability. reduction Type reduction apply loss. almost cases \"sum_over_batch_size\". Supported options \"sum\", \"sum_over_batch_size\", \"mean\", \"mean_with_sample_weight\" NULL. \"sum\" sums loss, \"sum_over_batch_size\" \"mean\" sum loss divide sample size, \"mean_with_sample_weight\" sums loss divides sum sample weights. \"none\" NULL perform aggregation. Defaults \"sum_over_batch_size\". name Optional name loss instance. dtype dtype loss's computations. Defaults NULL, means using config_floatx(). config_floatx() \"float32\" unless set different value (via config_set_floatx()). keras$DTypePolicy provided, compute_dtype utilized.","code":""},{"path":"https://keras3.posit.co/dev/reference/loss_cosine_similarity.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Computes the cosine similarity between y_true & y_pred. — loss_cosine_similarity","text":"Cosine similarity tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/loss_cosine_similarity.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Computes the cosine similarity between y_true & y_pred. — loss_cosine_similarity","text":"","code":"y_true <- rbind(c(0., 1.), c(1., 1.), c(1., 1.)) y_pred <- rbind(c(1., 0.), c(1., 1.), c(-1., -1.)) loss <- loss_cosine_similarity(y_true, y_pred, axis=-1) loss ## tf.Tensor([-0.         -0.99999994  0.99999994], shape=(3), dtype=float32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/loss_ctc.html","id":null,"dir":"Reference","previous_headings":"","what":"CTC (Connectionist Temporal Classification) loss. — loss_ctc","title":"CTC (Connectionist Temporal Classification) loss. — loss_ctc","text":"CTC (Connectionist Temporal Classification) loss.","code":""},{"path":"https://keras3.posit.co/dev/reference/loss_ctc.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"CTC (Connectionist Temporal Classification) loss. — loss_ctc","text":"","code":"loss_ctc(   y_true,   y_pred,   ...,   reduction = \"sum_over_batch_size\",   name = \"ctc\",   dtype = NULL )"},{"path":"https://keras3.posit.co/dev/reference/loss_ctc.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"CTC (Connectionist Temporal Classification) loss. — loss_ctc","text":"y_true tensor shape (batch_size, target_max_length) containing true labels integer format. 0 always represents blank/mask index used classes. y_pred tensor shape (batch_size, output_max_length, num_classes) containing logits (output model). normalized via softmax. ... forward/backward compatability. reduction Type reduction apply loss. almost cases \"sum_over_batch_size\". Supported options \"sum\", \"sum_over_batch_size\", \"mean\", \"mean_with_sample_weight\" NULL. \"sum\" sums loss, \"sum_over_batch_size\" \"mean\" sum loss divide sample size, \"mean_with_sample_weight\" sums loss divides sum sample weights. \"none\" NULL perform aggregation. Defaults \"sum_over_batch_size\". name Optional name loss instance. dtype dtype loss's computations. Defaults NULL, means using config_floatx(). config_floatx() \"float32\" unless set different value (via config_set_floatx()). keras$DTypePolicy provided, compute_dtype utilized.","code":""},{"path":"https://keras3.posit.co/dev/reference/loss_ctc.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"CTC (Connectionist Temporal Classification) loss. — loss_ctc","text":"CTC loss value.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/loss_dice.html","id":null,"dir":"Reference","previous_headings":"","what":"Computes the Dice loss value between y_true and y_pred. — loss_dice","title":"Computes the Dice loss value between y_true and y_pred. — loss_dice","text":"Formula:   Formula:","code":"loss = 1 - (2 * sum(y_true * y_pred)) / (sum(y_true) + sum(y_pred)) loss = 1 - (2 * sum(y_true * y_pred)) / (sum(y_true) + sum(y_pred))"},{"path":"https://keras3.posit.co/dev/reference/loss_dice.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes the Dice loss value between y_true and y_pred. — loss_dice","text":"","code":"loss_dice(   y_true,   y_pred,   ...,   reduction = \"sum_over_batch_size\",   name = \"dice\",   axis = NULL,   dtype = NULL )"},{"path":"https://keras3.posit.co/dev/reference/loss_dice.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Computes the Dice loss value between y_true and y_pred. — loss_dice","text":"y_true Tensor true targets. y_pred Tensor predicted targets. ... forward/backward compatability. reduction Type reduction apply loss. almost cases \"sum_over_batch_size\". Supported options \"sum\", \"sum_over_batch_size\", \"mean\", \"mean_with_sample_weight\" NULL. \"sum\" sums loss, \"sum_over_batch_size\" \"mean\" sum loss divide sample size, \"mean_with_sample_weight\" sums loss divides sum sample weights. \"none\" NULL perform aggregation. Defaults \"sum_over_batch_size\". name String, name object axis List dimensions loss calculated. Defaults NULL. dtype dtype loss's computations. Defaults NULL, means using config_floatx(). config_floatx() \"float32\" unless set different value (via config_set_floatx()). keras$DTypePolicy provided, compute_dtype utilized.","code":""},{"path":"https://keras3.posit.co/dev/reference/loss_dice.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Computes the Dice loss value between y_true and y_pred. — loss_dice","text":"y_true y_pred provided, Dice loss value. Otherwise, Loss() instance.","code":""},{"path":"https://keras3.posit.co/dev/reference/loss_dice.html","id":"example","dir":"Reference","previous_headings":"","what":"Example","title":"Computes the Dice loss value between y_true and y_pred. — loss_dice","text":"","code":"y_true <- array(c(1, 1, 0, 0,                   1, 1, 0, 0), dim = c(2, 2, 2, 1)) y_pred <- array(c(0, 0.4, 0,   0,                   1,   0, 1, 0.9), dim = c(2, 2, 2, 1))  axis <- c(2, 3, 4) loss_fn <- loss_dice(axis = axis, reduction = NULL) loss <- loss_fn(y_true, y_pred) stopifnot(shape(loss) == shape(2)) loss ## tf.Tensor([0.5        0.75757575], shape=(2), dtype=float32) loss_fn <- loss_dice() loss <- loss_fn(y_true, y_pred) stopifnot(shape(loss) == shape()) loss ## tf.Tensor(0.6164384, shape=(), dtype=float32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/loss_hinge.html","id":null,"dir":"Reference","previous_headings":"","what":"Computes the hinge loss between y_true & y_pred. — loss_hinge","title":"Computes the hinge loss between y_true & y_pred. — loss_hinge","text":"Formula:   y_true values expected -1 1. binary (0 1) labels provided convert -1 1.","code":"loss <- mean(maximum(1 - y_true * y_pred, 0), axis=-1)"},{"path":"https://keras3.posit.co/dev/reference/loss_hinge.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes the hinge loss between y_true & y_pred. — loss_hinge","text":"","code":"loss_hinge(   y_true,   y_pred,   ...,   reduction = \"sum_over_batch_size\",   name = \"hinge\",   dtype = NULL )"},{"path":"https://keras3.posit.co/dev/reference/loss_hinge.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Computes the hinge loss between y_true & y_pred. — loss_hinge","text":"y_true ground truth values. y_true values expected -1 1. binary (0 1) labels provided converted -1 1 shape = [batch_size, d0, .. dN]. y_pred predicted values shape = [batch_size, d0, .. dN]. ... forward/backward compatability. reduction Type reduction apply loss. almost cases \"sum_over_batch_size\". Supported options \"sum\", \"sum_over_batch_size\", \"mean\", \"mean_with_sample_weight\" NULL. \"sum\" sums loss, \"sum_over_batch_size\" \"mean\" sum loss divide sample size, \"mean_with_sample_weight\" sums loss divides sum sample weights. \"none\" NULL perform aggregation. Defaults \"sum_over_batch_size\". name Optional name loss instance. dtype dtype loss's computations. Defaults NULL, means using config_floatx(). config_floatx() \"float32\" unless set different value (via config_set_floatx()). keras$DTypePolicy provided, compute_dtype utilized.","code":""},{"path":"https://keras3.posit.co/dev/reference/loss_hinge.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Computes the hinge loss between y_true & y_pred. — loss_hinge","text":"Hinge loss values shape = [batch_size, d0, .. dN-1].","code":""},{"path":"https://keras3.posit.co/dev/reference/loss_hinge.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Computes the hinge loss between y_true & y_pred. — loss_hinge","text":"","code":"y_true <- array(sample(c(-1,1), 6, replace = TRUE), dim = c(2, 3)) y_pred <- random_uniform(c(2, 3)) loss <- loss_hinge(y_true, y_pred) loss ## tf.Tensor([1.0610152  0.93285507], shape=(2), dtype=float32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/loss_huber.html","id":null,"dir":"Reference","previous_headings":"","what":"Computes the Huber loss between y_true & y_pred. — loss_huber","title":"Computes the Huber loss between y_true & y_pred. — loss_huber","text":"Formula:   See: Huber loss.","code":"for (x in error) {   if (abs(x) <= delta){     loss <- c(loss, (0.5 * x^2))   } else if (abs(x) > delta) {     loss <- c(loss, (delta * abs(x) - 0.5 * delta^2))   } } loss <- mean(loss)"},{"path":"https://keras3.posit.co/dev/reference/loss_huber.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes the Huber loss between y_true & y_pred. — loss_huber","text":"","code":"loss_huber(   y_true,   y_pred,   delta = 1,   ...,   reduction = \"sum_over_batch_size\",   name = \"huber_loss\",   dtype = NULL )"},{"path":"https://keras3.posit.co/dev/reference/loss_huber.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Computes the Huber loss between y_true & y_pred. — loss_huber","text":"y_true tensor true targets. y_pred tensor predicted targets. delta float, point Huber loss function changes quadratic linear. Defaults 1.0. ... forward/backward compatability. reduction Type reduction apply loss. Options \"sum\", \"sum_over_batch_size\" NULL. Defaults \"sum_over_batch_size\". name Optional name instance. dtype dtype loss's computations. Defaults NULL, means using config_floatx(). config_floatx() \"float32\" unless set different value (via config_set_floatx()). keras$DTypePolicy provided, compute_dtype utilized.","code":""},{"path":"https://keras3.posit.co/dev/reference/loss_huber.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Computes the Huber loss between y_true & y_pred. — loss_huber","text":"Tensor one scalar loss entry per sample.","code":""},{"path":"https://keras3.posit.co/dev/reference/loss_huber.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Computes the Huber loss between y_true & y_pred. — loss_huber","text":"","code":"y_true <- rbind(c(0, 1), c(0, 0)) y_pred <- rbind(c(0.6, 0.4), c(0.4, 0.6)) loss <- loss_huber(y_true, y_pred)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/loss_kl_divergence.html","id":null,"dir":"Reference","previous_headings":"","what":"Computes Kullback-Leibler divergence loss between y_true & y_pred. — loss_kl_divergence","title":"Computes Kullback-Leibler divergence loss between y_true & y_pred. — loss_kl_divergence","text":"Formula:   y_true y_pred expected probability distributions, values 0 1. get clipped [0, 1] range.","code":"loss <- y_true * log(y_true / y_pred)"},{"path":"https://keras3.posit.co/dev/reference/loss_kl_divergence.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes Kullback-Leibler divergence loss between y_true & y_pred. — loss_kl_divergence","text":"","code":"loss_kl_divergence(   y_true,   y_pred,   ...,   reduction = \"sum_over_batch_size\",   name = \"kl_divergence\",   dtype = NULL )"},{"path":"https://keras3.posit.co/dev/reference/loss_kl_divergence.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Computes Kullback-Leibler divergence loss between y_true & y_pred. — loss_kl_divergence","text":"y_true Tensor true targets. y_pred Tensor predicted targets. ... forward/backward compatability. reduction Type reduction apply loss. almost cases \"sum_over_batch_size\". Supported options \"sum\", \"sum_over_batch_size\", \"mean\", \"mean_with_sample_weight\" NULL. \"sum\" sums loss, \"sum_over_batch_size\" \"mean\" sum loss divide sample size, \"mean_with_sample_weight\" sums loss divides sum sample weights. \"none\" NULL perform aggregation. Defaults \"sum_over_batch_size\". name Optional name loss instance. dtype dtype loss's computations. Defaults NULL, means using config_floatx(). config_floatx() \"float32\" unless set different value (via config_set_floatx()). keras$DTypePolicy provided, compute_dtype utilized.","code":""},{"path":"https://keras3.posit.co/dev/reference/loss_kl_divergence.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Computes Kullback-Leibler divergence loss between y_true & y_pred. — loss_kl_divergence","text":"KL Divergence loss values shape = [batch_size, d0, .. dN-1].","code":""},{"path":"https://keras3.posit.co/dev/reference/loss_kl_divergence.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Computes Kullback-Leibler divergence loss between y_true & y_pred. — loss_kl_divergence","text":"","code":"y_true <- random_uniform(c(2, 3), 0, 2) y_pred <- random_uniform(c(2,3)) loss <- loss_kl_divergence(y_true, y_pred) loss ## tf.Tensor([ 2.4290292 -0.6284211], shape=(2), dtype=float32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/loss_log_cosh.html","id":null,"dir":"Reference","previous_headings":"","what":"Computes the logarithm of the hyperbolic cosine of the prediction error. — loss_log_cosh","title":"Computes the logarithm of the hyperbolic cosine of the prediction error. — loss_log_cosh","text":"Formula:   Note log(cosh(x)) approximately equal (x ** 2) / 2 small x abs(x) - log(2) large x. means 'logcosh' works mostly like mean squared error, strongly affected occasional wildly incorrect prediction.","code":"loss <- mean(log(cosh(y_pred - y_true)), axis=-1)"},{"path":"https://keras3.posit.co/dev/reference/loss_log_cosh.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes the logarithm of the hyperbolic cosine of the prediction error. — loss_log_cosh","text":"","code":"loss_log_cosh(   y_true,   y_pred,   ...,   reduction = \"sum_over_batch_size\",   name = \"log_cosh\",   dtype = NULL )"},{"path":"https://keras3.posit.co/dev/reference/loss_log_cosh.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Computes the logarithm of the hyperbolic cosine of the prediction error. — loss_log_cosh","text":"y_true Ground truth values shape = [batch_size, d0, .. dN]. y_pred predicted values shape = [batch_size, d0, .. dN]. ... forward/backward compatability. reduction Type reduction apply loss. Options \"sum\", \"sum_over_batch_size\" NULL. Defaults \"sum_over_batch_size\". name Optional name instance. dtype dtype loss's computations. Defaults NULL, means using config_floatx(). config_floatx() \"float32\" unless set different value (via config_set_floatx()). keras$DTypePolicy provided, compute_dtype utilized.","code":""},{"path":"https://keras3.posit.co/dev/reference/loss_log_cosh.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Computes the logarithm of the hyperbolic cosine of the prediction error. — loss_log_cosh","text":"Logcosh error values shape = [batch_size, d0, .. dN-1].","code":""},{"path":"https://keras3.posit.co/dev/reference/loss_log_cosh.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Computes the logarithm of the hyperbolic cosine of the prediction error. — loss_log_cosh","text":"","code":"y_true <- rbind(c(0., 1.), c(0., 0.)) y_pred <- rbind(c(1., 1.), c(0., 0.)) loss <- loss_log_cosh(y_true, y_pred) # 0.108"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/loss_mean_absolute_error.html","id":null,"dir":"Reference","previous_headings":"","what":"Computes the mean of absolute difference between labels and predictions. — loss_mean_absolute_error","title":"Computes the mean of absolute difference between labels and predictions. — loss_mean_absolute_error","text":"Formula:","code":"loss <- mean(abs(y_true - y_pred))"},{"path":"https://keras3.posit.co/dev/reference/loss_mean_absolute_error.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes the mean of absolute difference between labels and predictions. — loss_mean_absolute_error","text":"","code":"loss_mean_absolute_error(   y_true,   y_pred,   ...,   reduction = \"sum_over_batch_size\",   name = \"mean_absolute_error\",   dtype = NULL )"},{"path":"https://keras3.posit.co/dev/reference/loss_mean_absolute_error.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Computes the mean of absolute difference between labels and predictions. — loss_mean_absolute_error","text":"y_true Ground truth values shape = [batch_size, d0, .. dN]. y_pred predicted values shape = [batch_size, d0, .. dN]. ... forward/backward compatability. reduction Type reduction apply loss. almost cases \"sum_over_batch_size\". Supported options \"sum\", \"sum_over_batch_size\", \"mean\", \"mean_with_sample_weight\" NULL. \"sum\" sums loss, \"sum_over_batch_size\" \"mean\" sum loss divide sample size, \"mean_with_sample_weight\" sums loss divides sum sample weights. \"none\" NULL perform aggregation. Defaults \"sum_over_batch_size\". name Optional name loss instance. dtype dtype loss's computations. Defaults NULL, means using config_floatx(). config_floatx() \"float32\" unless set different value (via config_set_floatx()). keras$DTypePolicy provided, compute_dtype utilized.","code":""},{"path":"https://keras3.posit.co/dev/reference/loss_mean_absolute_error.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Computes the mean of absolute difference between labels and predictions. — loss_mean_absolute_error","text":"Mean absolute error values shape = [batch_size, d0, .. dN-1].","code":""},{"path":"https://keras3.posit.co/dev/reference/loss_mean_absolute_error.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Computes the mean of absolute difference between labels and predictions. — loss_mean_absolute_error","text":"","code":"y_true <- random_uniform(c(2, 3), 0, 2) y_pred <- random_uniform(c(2, 3)) loss <- loss_mean_absolute_error(y_true, y_pred)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/loss_mean_absolute_percentage_error.html","id":null,"dir":"Reference","previous_headings":"","what":"Computes the mean absolute percentage error between y_true and y_pred. — loss_mean_absolute_percentage_error","title":"Computes the mean absolute percentage error between y_true and y_pred. — loss_mean_absolute_percentage_error","text":"Formula:   Division zero prevented dividing max(y_true, epsilon) epsilon = config_epsilon() (default 1e-7).","code":"loss <- 100 * op_mean(op_abs((y_true - y_pred) / y_true),                       axis=-1)"},{"path":"https://keras3.posit.co/dev/reference/loss_mean_absolute_percentage_error.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes the mean absolute percentage error between y_true and y_pred. — loss_mean_absolute_percentage_error","text":"","code":"loss_mean_absolute_percentage_error(   y_true,   y_pred,   ...,   reduction = \"sum_over_batch_size\",   name = \"mean_absolute_percentage_error\",   dtype = NULL )"},{"path":"https://keras3.posit.co/dev/reference/loss_mean_absolute_percentage_error.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Computes the mean absolute percentage error between y_true and y_pred. — loss_mean_absolute_percentage_error","text":"y_true Ground truth values shape = [batch_size, d0, .. dN]. y_pred predicted values shape = [batch_size, d0, .. dN]. ... forward/backward compatability. reduction Type reduction apply loss. almost cases \"sum_over_batch_size\". Supported options \"sum\", \"sum_over_batch_size\", \"mean\", \"mean_with_sample_weight\" NULL. \"sum\" sums loss, \"sum_over_batch_size\" \"mean\" sum loss divide sample size, \"mean_with_sample_weight\" sums loss divides sum sample weights. \"none\" NULL perform aggregation. Defaults \"sum_over_batch_size\". name Optional name loss instance. dtype dtype loss's computations. Defaults NULL, means using config_floatx(). config_floatx() \"float32\" unless set different value (via config_set_floatx()). keras$DTypePolicy provided, compute_dtype utilized.","code":""},{"path":"https://keras3.posit.co/dev/reference/loss_mean_absolute_percentage_error.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Computes the mean absolute percentage error between y_true and y_pred. — loss_mean_absolute_percentage_error","text":"Mean absolute percentage error values shape = [batch_size, d0, ..dN-1].","code":""},{"path":"https://keras3.posit.co/dev/reference/loss_mean_absolute_percentage_error.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Computes the mean absolute percentage error between y_true and y_pred. — loss_mean_absolute_percentage_error","text":"","code":"y_true <- random_uniform(c(2, 3)) y_pred <- random_uniform(c(2, 3)) loss <- loss_mean_absolute_percentage_error(y_true, y_pred)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/loss_mean_squared_error.html","id":null,"dir":"Reference","previous_headings":"","what":"Computes the mean of squares of errors between labels and predictions. — loss_mean_squared_error","title":"Computes the mean of squares of errors between labels and predictions. — loss_mean_squared_error","text":"Formula:","code":"loss <- mean(square(y_true - y_pred))"},{"path":"https://keras3.posit.co/dev/reference/loss_mean_squared_error.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes the mean of squares of errors between labels and predictions. — loss_mean_squared_error","text":"","code":"loss_mean_squared_error(   y_true,   y_pred,   ...,   reduction = \"sum_over_batch_size\",   name = \"mean_squared_error\",   dtype = NULL )"},{"path":"https://keras3.posit.co/dev/reference/loss_mean_squared_error.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Computes the mean of squares of errors between labels and predictions. — loss_mean_squared_error","text":"y_true Ground truth values shape = [batch_size, d0, .. dN]. y_pred predicted values shape = [batch_size, d0, .. dN]. ... forward/backward compatability. reduction Type reduction apply loss. almost cases \"sum_over_batch_size\". Supported options \"sum\", \"sum_over_batch_size\", \"mean\", \"mean_with_sample_weight\" NULL. \"sum\" sums loss, \"sum_over_batch_size\" \"mean\" sum loss divide sample size, \"mean_with_sample_weight\" sums loss divides sum sample weights. \"none\" NULL perform aggregation. Defaults \"sum_over_batch_size\". name Optional name loss instance. dtype dtype loss's computations. Defaults NULL, means using config_floatx(). config_floatx() \"float32\" unless set different value (via config_set_floatx()). keras$DTypePolicy provided, compute_dtype utilized.","code":""},{"path":"https://keras3.posit.co/dev/reference/loss_mean_squared_error.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Computes the mean of squares of errors between labels and predictions. — loss_mean_squared_error","text":"Mean squared error values shape = [batch_size, d0, .. dN-1].","code":""},{"path":"https://keras3.posit.co/dev/reference/loss_mean_squared_error.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Computes the mean of squares of errors between labels and predictions. — loss_mean_squared_error","text":"","code":"y_true <- random_uniform(c(2, 3), 0, 2) y_pred <- random_uniform(c(2, 3)) loss <- loss_mean_squared_error(y_true, y_pred)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/loss_mean_squared_logarithmic_error.html","id":null,"dir":"Reference","previous_headings":"","what":"Computes the mean squared logarithmic error between y_true and y_pred. — loss_mean_squared_logarithmic_error","title":"Computes the mean squared logarithmic error between y_true and y_pred. — loss_mean_squared_logarithmic_error","text":"Note y_pred y_true less equal 0. Negative values 0 values replaced config_epsilon() (default 1e-7). Formula:","code":"loss <- mean(square(log(y_true + 1) - log(y_pred + 1)))"},{"path":"https://keras3.posit.co/dev/reference/loss_mean_squared_logarithmic_error.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes the mean squared logarithmic error between y_true and y_pred. — loss_mean_squared_logarithmic_error","text":"","code":"loss_mean_squared_logarithmic_error(   y_true,   y_pred,   ...,   reduction = \"sum_over_batch_size\",   name = \"mean_squared_logarithmic_error\",   dtype = NULL )"},{"path":"https://keras3.posit.co/dev/reference/loss_mean_squared_logarithmic_error.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Computes the mean squared logarithmic error between y_true and y_pred. — loss_mean_squared_logarithmic_error","text":"y_true Ground truth values shape = [batch_size, d0, .. dN]. y_pred predicted values shape = [batch_size, d0, .. dN]. ... forward/backward compatability. reduction Type reduction apply loss. almost cases \"sum_over_batch_size\". Supported options \"sum\", \"sum_over_batch_size\", \"mean\", \"mean_with_sample_weight\" NULL. \"sum\" sums loss, \"sum_over_batch_size\" \"mean\" sum loss divide sample size, \"mean_with_sample_weight\" sums loss divides sum sample weights. \"none\" NULL perform aggregation. Defaults \"sum_over_batch_size\". name Optional name loss instance. dtype dtype loss's computations. Defaults NULL, means using config_floatx(). config_floatx() \"float32\" unless set different value (via config_set_floatx()). keras$DTypePolicy provided, compute_dtype utilized.","code":""},{"path":"https://keras3.posit.co/dev/reference/loss_mean_squared_logarithmic_error.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Computes the mean squared logarithmic error between y_true and y_pred. — loss_mean_squared_logarithmic_error","text":"Mean squared logarithmic error values shape = [batch_size, d0, .. dN-1].","code":""},{"path":"https://keras3.posit.co/dev/reference/loss_mean_squared_logarithmic_error.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Computes the mean squared logarithmic error between y_true and y_pred. — loss_mean_squared_logarithmic_error","text":"","code":"y_true <- random_uniform(c(2, 3), 0, 2) y_pred <- random_uniform(c(2, 3)) loss <- loss_mean_squared_logarithmic_error(y_true, y_pred)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/loss_poisson.html","id":null,"dir":"Reference","previous_headings":"","what":"Computes the Poisson loss between y_true & y_pred. — loss_poisson","title":"Computes the Poisson loss between y_true & y_pred. — loss_poisson","text":"Formula:","code":"loss <- y_pred - y_true * log(y_pred)"},{"path":"https://keras3.posit.co/dev/reference/loss_poisson.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes the Poisson loss between y_true & y_pred. — loss_poisson","text":"","code":"loss_poisson(   y_true,   y_pred,   ...,   reduction = \"sum_over_batch_size\",   name = \"poisson\",   dtype = NULL )"},{"path":"https://keras3.posit.co/dev/reference/loss_poisson.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Computes the Poisson loss between y_true & y_pred. — loss_poisson","text":"y_true Ground truth values. shape = [batch_size, d0, .. dN]. y_pred predicted values. shape = [batch_size, d0, .. dN]. ... forward/backward compatability. reduction Type reduction apply loss. almost cases \"sum_over_batch_size\". Supported options \"sum\", \"sum_over_batch_size\", \"mean\", \"mean_with_sample_weight\" NULL. \"sum\" sums loss, \"sum_over_batch_size\" \"mean\" sum loss divide sample size, \"mean_with_sample_weight\" sums loss divides sum sample weights. \"none\" NULL perform aggregation. Defaults \"sum_over_batch_size\". name Optional name loss instance. dtype dtype loss's computations. Defaults NULL, means using config_floatx(). config_floatx() \"float32\" unless set different value (via config_set_floatx()). keras$DTypePolicy provided, compute_dtype utilized.","code":""},{"path":"https://keras3.posit.co/dev/reference/loss_poisson.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Computes the Poisson loss between y_true & y_pred. — loss_poisson","text":"Poisson loss values shape = [batch_size, d0, .. dN-1].","code":""},{"path":"https://keras3.posit.co/dev/reference/loss_poisson.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Computes the Poisson loss between y_true & y_pred. — loss_poisson","text":"","code":"y_true <- random_uniform(c(2, 3), 0, 2) y_pred <- random_uniform(c(2, 3)) loss <- loss_poisson(y_true, y_pred) loss ## tf.Tensor([1.6422468  0.81166863], shape=(2), dtype=float32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/loss_sparse_categorical_crossentropy.html","id":null,"dir":"Reference","previous_headings":"","what":"Computes the crossentropy loss between the labels and predictions. — loss_sparse_categorical_crossentropy","title":"Computes the crossentropy loss between the labels and predictions. — loss_sparse_categorical_crossentropy","text":"Use crossentropy loss function two label classes.  expect labels provided integers. want provide labels using one-hot representation, please use CategoricalCrossentropy loss.  # classes floating point values per feature y_pred single floating point value per feature y_true. snippet , single floating point value per example y_true num_classes floating pointing values per example y_pred. shape y_true [batch_size] shape y_pred [batch_size, num_classes].","code":""},{"path":"https://keras3.posit.co/dev/reference/loss_sparse_categorical_crossentropy.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes the crossentropy loss between the labels and predictions. — loss_sparse_categorical_crossentropy","text":"","code":"loss_sparse_categorical_crossentropy(   y_true,   y_pred,   from_logits = FALSE,   ignore_class = NULL,   axis = -1L,   ...,   reduction = \"sum_over_batch_size\",   name = \"sparse_categorical_crossentropy\",   dtype = NULL )"},{"path":"https://keras3.posit.co/dev/reference/loss_sparse_categorical_crossentropy.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Computes the crossentropy loss between the labels and predictions. — loss_sparse_categorical_crossentropy","text":"y_true Ground truth values. y_pred predicted values. from_logits Whether y_pred expected logits tensor. default, assume y_pred encodes probability distribution. ignore_class Optional integer. ID class ignored loss computation. useful, example, segmentation problems featuring \"void\" class (commonly -1 255) segmentation maps. default (ignore_class=NULL), classes considered. axis Defaults -1. dimension along entropy computed. ... forward/backward compatability. reduction Type reduction apply loss. almost cases \"sum_over_batch_size\". Supported options \"sum\", \"sum_over_batch_size\", \"mean\", \"mean_with_sample_weight\" NULL. \"sum\" sums loss, \"sum_over_batch_size\" \"mean\" sum loss divide sample size, \"mean_with_sample_weight\" sums loss divides sum sample weights. \"none\" NULL perform aggregation. Defaults \"sum_over_batch_size\". name Optional name loss instance. dtype dtype loss's computations. Defaults NULL, means using config_floatx(). config_floatx() \"float32\" unless set different value (via config_set_floatx()). keras$DTypePolicy provided, compute_dtype utilized.","code":""},{"path":"https://keras3.posit.co/dev/reference/loss_sparse_categorical_crossentropy.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Computes the crossentropy loss between the labels and predictions. — loss_sparse_categorical_crossentropy","text":"Sparse categorical crossentropy loss value.","code":""},{"path":"https://keras3.posit.co/dev/reference/loss_sparse_categorical_crossentropy.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Computes the crossentropy loss between the labels and predictions. — loss_sparse_categorical_crossentropy","text":"Usage compile() API:","code":"y_true <- op_array(c(1L, 2L)) y_pred <- op_array(rbind(c(0.05, 0.95, 0), c(0.1, 0.8, 0.1))) loss <- loss_sparse_categorical_crossentropy(y_true, y_pred) loss ## tf.Tensor([0.05129339 2.30258509], shape=(2), dtype=float64) y_true <- op_array(c(1L, 2L)) y_pred <- op_array(rbind(c(0.05, 0.95, 0), c(0.1, 0.8, 0.1))) # Using 'auto'/'sum_over_batch_size' reduction type. scce <- loss_sparse_categorical_crossentropy() scce(y_true, y_pred) ## tf.Tensor(1.1769392, shape=(), dtype=float32) # Calling with 'sample_weight'. scce(y_true, y_pred, sample_weight = op_array(c(0.3, 0.7))) ## tf.Tensor(0.8135988, shape=(), dtype=float32) # Using 'sum' reduction type. scce <- loss_sparse_categorical_crossentropy(reduction=\"sum\") scce(y_true, y_pred) ## tf.Tensor(2.3538785, shape=(), dtype=float32) # Using 'none' reduction type. scce <- loss_sparse_categorical_crossentropy(reduction=NULL) scce(y_true, y_pred) ## tf.Tensor([0.05129344 2.3025851 ], shape=(2), dtype=float32) model %>% compile(optimizer = 'sgd',                   loss = loss_sparse_categorical_crossentropy())"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/loss_squared_hinge.html","id":null,"dir":"Reference","previous_headings":"","what":"Computes the squared hinge loss between y_true & y_pred. — loss_squared_hinge","title":"Computes the squared hinge loss between y_true & y_pred. — loss_squared_hinge","text":"Formula:   y_true values expected -1 1. binary (0 1) labels provided convert -1 1.","code":"loss <- square(maximum(1 - y_true * y_pred, 0))"},{"path":"https://keras3.posit.co/dev/reference/loss_squared_hinge.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes the squared hinge loss between y_true & y_pred. — loss_squared_hinge","text":"","code":"loss_squared_hinge(   y_true,   y_pred,   ...,   reduction = \"sum_over_batch_size\",   name = \"squared_hinge\",   dtype = NULL )"},{"path":"https://keras3.posit.co/dev/reference/loss_squared_hinge.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Computes the squared hinge loss between y_true & y_pred. — loss_squared_hinge","text":"y_true ground truth values. y_true values expected -1 1. binary (0 1) labels provided convert -1 1 shape = [batch_size, d0, .. dN]. y_pred predicted values shape = [batch_size, d0, .. dN]. ... forward/backward compatability. reduction Type reduction apply loss. almost cases \"sum_over_batch_size\". Supported options \"sum\", \"sum_over_batch_size\", \"mean\", \"mean_with_sample_weight\" NULL. \"sum\" sums loss, \"sum_over_batch_size\" \"mean\" sum loss divide sample size, \"mean_with_sample_weight\" sums loss divides sum sample weights. \"none\" NULL perform aggregation. Defaults \"sum_over_batch_size\". name Optional name loss instance. dtype dtype loss's computations. Defaults NULL, means using config_floatx(). config_floatx() \"float32\" unless set different value (via config_set_floatx()). keras$DTypePolicy provided, compute_dtype utilized.","code":""},{"path":"https://keras3.posit.co/dev/reference/loss_squared_hinge.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Computes the squared hinge loss between y_true & y_pred. — loss_squared_hinge","text":"Squared hinge loss values shape = [batch_size, d0, .. dN-1].","code":""},{"path":"https://keras3.posit.co/dev/reference/loss_squared_hinge.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Computes the squared hinge loss between y_true & y_pred. — loss_squared_hinge","text":"","code":"y_true <- array(sample(c(-1,1), 6, replace = TRUE), dim = c(2, 3)) y_pred <- random_uniform(c(2, 3)) loss <- loss_squared_hinge(y_true, y_pred)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/loss_tversky.html","id":null,"dir":"Reference","previous_headings":"","what":"Computes the Tversky loss value between y_true and y_pred. — loss_tversky","title":"Computes the Tversky loss value between y_true and y_pred. — loss_tversky","text":"loss function weighted alpha beta coefficients penalize false positives false negatives. alpha=0.5 beta=0.5, loss value becomes equivalent Dice Loss. loss function weighted alpha beta coefficients penalize false positives false negatives. alpha=0.5 beta=0.5, loss value becomes equivalent Dice Loss.","code":""},{"path":"https://keras3.posit.co/dev/reference/loss_tversky.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes the Tversky loss value between y_true and y_pred. — loss_tversky","text":"","code":"loss_tversky(   y_true,   y_pred,   ...,   alpha = 0.5,   beta = 0.5,   reduction = \"sum_over_batch_size\",   name = \"tversky\",   axis = NULL,   dtype = NULL )"},{"path":"https://keras3.posit.co/dev/reference/loss_tversky.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Computes the Tversky loss value between y_true and y_pred. — loss_tversky","text":"y_true tensor true targets. y_pred tensor predicted targets. ... forward/backward compatability. alpha coefficient controlling incidence false positives. Defaults 0.5. beta coefficient controlling incidence false negatives. Defaults 0.5. reduction Type reduction apply loss. almost cases \"sum_over_batch_size\". Supported options \"sum\", \"sum_over_batch_size\", \"mean\", \"mean_with_sample_weight\" NULL. \"sum\" sums loss, \"sum_over_batch_size\" \"mean\" sum loss divide sample size, \"mean_with_sample_weight\" sums loss divides sum sample weights. \"none\" NULL perform aggregation. Defaults \"sum_over_batch_size\". name Optional name loss instance. (string) axis Axis (1-based) dtype dtype loss's computations. Defaults NULL, means using config_floatx(). config_floatx() \"float32\" unless set different value (via config_set_floatx()). keras$DTypePolicy provided, compute_dtype utilized.","code":""},{"path":"https://keras3.posit.co/dev/reference/loss_tversky.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Computes the Tversky loss value between y_true and y_pred. — loss_tversky","text":"Tversky loss value.","code":""},{"path":"https://keras3.posit.co/dev/reference/loss_tversky.html","id":"reference","dir":"Reference","previous_headings":"","what":"Reference","title":"Computes the Tversky loss value between y_true and y_pred. — loss_tversky","text":"Salehi et al., 2017","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/mark_active.html","id":null,"dir":"Reference","previous_headings":"","what":"active_property — mark_active","title":"active_property — mark_active","text":"mark_active() alias active_property(). See ?active_property() full documentation.","code":""},{"path":"https://keras3.posit.co/dev/reference/mark_active.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"active_property — mark_active","text":"","code":"mark_active(fn)"},{"path":"https://keras3.posit.co/dev/reference/mark_active.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"active_property — mark_active","text":"fn R function","code":""},{"path":"https://keras3.posit.co/dev/reference/mark_active.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"active_property — mark_active","text":"fn, additional R attribute cause fn converted active property converted method custom subclass.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_auc.html","id":null,"dir":"Reference","previous_headings":"","what":"Approximates the AUC (Area under the curve) of the ROC or PR curves. — metric_auc","title":"Approximates the AUC (Area under the curve) of the ROC or PR curves. — metric_auc","text":"AUC (Area curve) ROC (Receiver operating characteristic; default) PR (Precision Recall) curves quality measures binary classifiers. Unlike accuracy, like cross-entropy losses, ROC-AUC PR-AUC evaluate operational points model. class approximates AUCs using Riemann sum. metric accumulation phrase, predictions accumulated within predefined buckets value. AUC computed interpolating per-bucket averages. buckets define evaluated operational points. metric creates four local variables, true_positives, true_negatives, false_positives false_negatives used compute AUC.  discretize AUC curve, linearly spaced set thresholds used compute pairs recall precision values. area ROC-curve therefore computed using height recall values false positive rate, area PR-curve computed using height precision values recall. value ultimately returned auc, idempotent operation computes area discretized curve precision versus recall values (computed using aforementioned variables). num_thresholds variable controls degree discretization larger numbers thresholds closely approximating true AUC. quality approximation may vary dramatically depending num_thresholds. thresholds parameter can used manually specify thresholds split predictions evenly. best approximation real AUC, predictions distributed approximately uniformly range [0, 1] (from_logits=FALSE). quality AUC approximation may poor case. Setting summation_method 'minoring' 'majoring' can help quantify error approximation providing lower upper bound estimate AUC. sample_weight NULL, weights default 1. Use sample_weight 0 mask values.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_auc.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Approximates the AUC (Area under the curve) of the ROC or PR curves. — metric_auc","text":"","code":"metric_auc(   ...,   num_thresholds = 200L,   curve = \"ROC\",   summation_method = \"interpolation\",   name = NULL,   dtype = NULL,   thresholds = NULL,   multi_label = FALSE,   num_labels = NULL,   label_weights = NULL,   from_logits = FALSE )"},{"path":"https://keras3.posit.co/dev/reference/metric_auc.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Approximates the AUC (Area under the curve) of the ROC or PR curves. — metric_auc","text":"... forward/backward compatability. num_thresholds (Optional) number thresholds use discretizing roc curve. Values must > 1. Defaults 200. curve (Optional) Specifies name curve computed, 'ROC' (default) 'PR' Precision-Recall-curve. summation_method (Optional) Specifies Riemann summation method used. 'interpolation' (default) applies mid-point summation scheme ROC.  PR-AUC, interpolates (true/false) positives ratio precision (see Davis & Goadrich 2006 details); 'minoring' applies left summation increasing intervals right summation decreasing intervals; 'majoring' opposite. name (Optional) string name metric instance. dtype (Optional) data type metric result. thresholds (Optional) list floating point values use thresholds discretizing curve. set, num_thresholds parameter ignored. Values [0, 1]. Endpoint thresholds equal {-epsilon, 1+epsilon} small positive epsilon value automatically included correctly handle predictions equal exactly 0 1. multi_label boolean indicating whether multilabel data treated , wherein AUC computed separately label averaged across labels, (FALSE) data flattened single label AUC computation. latter case, multilabel data passed AUC, label-prediction pair treated individual data point. set `FALSE“ multi-class data. num_labels (Optional) number labels, used multi_label TRUE. num_labels specified, state variables get created first call update_state. label_weights (Optional) list, array, tensor non-negative weights used compute AUCs multilabel data. multi_label TRUE, weights applied individual label AUCs averaged produce multi-label AUC. FALSE, used weight individual label predictions computing confusion matrix flattened data. Note unlike class_weights class_weights weights example depending value label, whereas label_weights depends index label flattening; therefore label_weights used multi-class data. from_logits boolean indicating whether predictions (y_pred update_state) probabilities sigmoid logits. rule thumb, using keras loss, from_logits constructor argument loss match AUC from_logits constructor argument.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_auc.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Approximates the AUC (Area under the curve) of the ROC or PR curves. — metric_auc","text":"Metric instance returned. Metric instance can passed directly compile(metrics = ), used standalone object. See ?Metric example usage.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_auc.html","id":"usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Approximates the AUC (Area under the curve) of the ROC or PR curves. — metric_auc","text":"Standalone usage:         Usage compile() API:","code":"m <- metric_auc(num_thresholds = 3) m$update_state(c(0,   0,   1,   1),                c(0, 0.5, 0.3, 0.9)) # threshold values are [0 - 1e-7, 0.5, 1 + 1e-7] # tp = [2, 1, 0], fp = [2, 0, 0], fn = [0, 1, 2], tn = [0, 2, 2] # tp_rate = recall = [1, 0.5, 0], fp_rate = [1, 0, 0] # auc = ((((1 + 0.5) / 2) * (1 - 0)) + (((0.5 + 0) / 2) * (0 - 0))) #     = 0.75 m$result() ## tf.Tensor(0.75, shape=(), dtype=float32) m$reset_state() m$update_state(c(0,   0,   1,   1),                c(0, 0.5, 0.3, 0.9),                sample_weight=c(1, 0, 0, 1)) m$result() ## tf.Tensor(1.0, shape=(), dtype=float32) # Reports the AUC of a model outputting a probability. model |> compile(   optimizer = 'sgd',   loss = loss_binary_crossentropy(),   metrics = list(metric_auc()) )  # Reports the AUC of a model outputting a logit. model |> compile(   optimizer = 'sgd',   loss = loss_binary_crossentropy(from_logits = TRUE),   metrics = list(metric_auc(from_logits = TRUE)) )"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/metric_binary_accuracy.html","id":null,"dir":"Reference","previous_headings":"","what":"Calculates how often predictions match binary labels. — metric_binary_accuracy","title":"Calculates how often predictions match binary labels. — metric_binary_accuracy","text":"metric creates two local variables, total count used compute frequency y_pred matches y_true. frequency ultimately returned binary accuracy: idempotent operation simply divides total count. sample_weight NULL, weights default 1. Use sample_weight 0 mask values.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_binary_accuracy.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Calculates how often predictions match binary labels. — metric_binary_accuracy","text":"","code":"metric_binary_accuracy(   y_true,   y_pred,   threshold = 0.5,   ...,   name = \"binary_accuracy\",   dtype = NULL )"},{"path":"https://keras3.posit.co/dev/reference/metric_binary_accuracy.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Calculates how often predictions match binary labels. — metric_binary_accuracy","text":"y_true Tensor true targets. y_pred Tensor predicted targets. threshold (Optional) Float representing threshold deciding whether prediction values 1 0. ... forward/backward compatability. name (Optional) string name metric instance. dtype (Optional) data type metric result.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_binary_accuracy.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Calculates how often predictions match binary labels. — metric_binary_accuracy","text":"y_true y_pred missing, Metric instance returned. Metric instance can passed directly compile(metrics = ), used standalone object. See ?Metric example usage. y_true y_pred provided, tensor computed value returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_binary_accuracy.html","id":"usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Calculates how often predictions match binary labels. — metric_binary_accuracy","text":"Standalone usage:             Usage compile() API:","code":"m <- metric_binary_accuracy() m$update_state(rbind(1, 1, 0, 0), rbind(0.98, 1, 0, 0.6)) m$result() ## tf.Tensor(0.75, shape=(), dtype=float32) # 0.75 m$reset_state() m$update_state(rbind(1, 1, 0, 0), rbind(0.98, 1, 0, 0.6),                sample_weight = c(1, 0, 0, 1)) m$result() ## tf.Tensor(0.5, shape=(), dtype=float32) # 0.5 model %>% compile(optimizer='sgd',                   loss='binary_crossentropy',                   metrics=list(metric_binary_accuracy()))"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/metric_binary_crossentropy.html","id":null,"dir":"Reference","previous_headings":"","what":"Computes the crossentropy metric between the labels and predictions. — metric_binary_crossentropy","title":"Computes the crossentropy metric between the labels and predictions. — metric_binary_crossentropy","text":"crossentropy metric class used two label classes (0 1).","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_binary_crossentropy.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes the crossentropy metric between the labels and predictions. — metric_binary_crossentropy","text":"","code":"metric_binary_crossentropy(   y_true,   y_pred,   from_logits = FALSE,   label_smoothing = 0,   axis = -1L,   ...,   name = \"binary_crossentropy\",   dtype = NULL )"},{"path":"https://keras3.posit.co/dev/reference/metric_binary_crossentropy.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Computes the crossentropy metric between the labels and predictions. — metric_binary_crossentropy","text":"y_true Ground truth values. shape = [batch_size, d0, .. dN]. y_pred predicted values. shape = [batch_size, d0, .. dN]. from_logits (Optional) Whether output expected logits tensor. default, consider output encodes probability distribution. label_smoothing (Optional) Float [0, 1]. > 0, label values smoothed, meaning confidence label values relaxed. e.g. label_smoothing=0.2 means use value 0.1 label \"0\" 0.9 label \"1\". axis axis along mean computed. Defaults -1. ... forward/backward compatability. name (Optional) string name metric instance. dtype (Optional) data type metric result.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_binary_crossentropy.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Computes the crossentropy metric between the labels and predictions. — metric_binary_crossentropy","text":"y_true y_pred missing, Metric instance returned. Metric instance can passed directly compile(metrics = ), used standalone object. See ?Metric example usage. y_true y_pred provided, tensor computed value returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_binary_crossentropy.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Computes the crossentropy metric between the labels and predictions. — metric_binary_crossentropy","text":"Standalone usage:         Usage compile() API:","code":"m <- metric_binary_crossentropy() m$update_state(rbind(c(0, 1), c(0, 0)), rbind(c(0.6, 0.4), c(0.4, 0.6))) m$result() ## tf.Tensor(0.8149245, shape=(), dtype=float32) m$reset_state() m$update_state(rbind(c(0, 1), c(0, 0)), rbind(c(0.6, 0.4), c(0.4, 0.6)),                sample_weight=c(1, 0)) m$result() ## tf.Tensor(0.91629076, shape=(), dtype=float32) model %>% compile(     optimizer = 'sgd',     loss = 'mse',     metrics = list(metric_binary_crossentropy()))"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/metric_binary_focal_crossentropy.html","id":null,"dir":"Reference","previous_headings":"","what":"Computes the binary focal crossentropy loss. — metric_binary_focal_crossentropy","title":"Computes the binary focal crossentropy loss. — metric_binary_focal_crossentropy","text":"According Lin et al., 2018, helps apply focal factor -weight easy examples focus hard examples. default, focal tensor computed follows: focal_factor = (1 - output)^gamma class 1 focal_factor = output^gamma class 0 gamma focusing parameter. gamma = 0, focal effect binary crossentropy loss. apply_class_balancing == TRUE, function also takes account weight balancing factor binary classes 0 1 follows: weight = alpha class 1 (target == 1) weight = 1 - alpha class 0 alpha float range [0, 1].","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_binary_focal_crossentropy.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes the binary focal crossentropy loss. — metric_binary_focal_crossentropy","text":"","code":"metric_binary_focal_crossentropy(   y_true,   y_pred,   apply_class_balancing = FALSE,   alpha = 0.25,   gamma = 2,   from_logits = FALSE,   label_smoothing = 0,   axis = -1L )"},{"path":"https://keras3.posit.co/dev/reference/metric_binary_focal_crossentropy.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Computes the binary focal crossentropy loss. — metric_binary_focal_crossentropy","text":"y_true Ground truth values, shape (batch_size, d0, .. dN). y_pred predicted values, shape (batch_size, d0, .. dN). apply_class_balancing bool, whether apply weight balancing binary classes 0 1. alpha weight balancing factor class 1, default 0.25 mentioned reference. weight class 0 1.0 - alpha. gamma focusing parameter, default 2.0 mentioned reference. from_logits Whether y_pred expected logits tensor. default, assume y_pred encodes probability distribution. label_smoothing Float [0, 1]. > 0 smooth labels squeezing towards 0.5, , using 1. - 0.5 * label_smoothing target class 0.5 * label_smoothing non-target class. axis axis along mean computed. Defaults -1.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_binary_focal_crossentropy.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Computes the binary focal crossentropy loss. — metric_binary_focal_crossentropy","text":"Binary focal crossentropy loss value shape = [batch_size, d0, .. dN-1].","code":""},{"path":[]},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/metric_binary_iou.html","id":null,"dir":"Reference","previous_headings":"","what":"Computes the Intersection-Over-Union metric for class 0 and/or 1. — metric_binary_iou","title":"Computes the Intersection-Over-Union metric for class 0 and/or 1. — metric_binary_iou","text":"Formula:   Intersection--Union common evaluation metric semantic image segmentation. compute IoUs, predictions accumulated confusion matrix, weighted sample_weight metric calculated . sample_weight NULL, weights default 1. Use sample_weight 0 mask values. class can used compute IoUs binary classification task predictions provided logits. First threshold applied predicted values threshold converted class 0 threshold converted class 1. IoUs classes 0 1 computed, mean IoUs classes specified target_class_ids returned.","code":"iou <- true_positives / (true_positives + false_positives + false_negatives)"},{"path":"https://keras3.posit.co/dev/reference/metric_binary_iou.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes the Intersection-Over-Union metric for class 0 and/or 1. — metric_binary_iou","text":"","code":"metric_binary_iou(   ...,   target_class_ids = list(0L, 1L),   threshold = 0.5,   name = NULL,   dtype = NULL )"},{"path":"https://keras3.posit.co/dev/reference/metric_binary_iou.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Computes the Intersection-Over-Union metric for class 0 and/or 1. — metric_binary_iou","text":"... forward/backward compatability. target_class_ids list list target class ids metric returned. Options 0, 1, c(0, 1). 0 (1), IoU metric class 0 (class 1, respectively) returned. c(0, 1), mean IoUs two classes returned. threshold threshold applies prediction logits convert either predicted class 0 logit threshold predicted class 1 logit threshold. name (Optional) string name metric instance. dtype (Optional) data type metric result.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_binary_iou.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Computes the Intersection-Over-Union metric for class 0 and/or 1. — metric_binary_iou","text":"Metric instance returned. Metric instance can passed directly compile(metrics = ), used standalone object. See ?Metric example usage.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_binary_iou.html","id":"note","dir":"Reference","previous_headings":"","what":"Note","title":"Computes the Intersection-Over-Union metric for class 0 and/or 1. — metric_binary_iou","text":"threshold=0, metric behavior IoU.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_binary_iou.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Computes the Intersection-Over-Union metric for class 0 and/or 1. — metric_binary_iou","text":"Standalone usage:                 Usage compile() API:","code":"m <- metric_binary_iou(target_class_ids=c(0L, 1L), threshold = 0.3) m$update_state(c(0, 1, 0, 1), c(0.1, 0.2, 0.4, 0.7)) ## tf.Tensor( ## [[1 1] ##  [1 1]], shape=(2, 2), dtype=int64) m$result() ## tf.Tensor(0.33333334, shape=(), dtype=float32) m$reset_state() m$update_state(c(0, 1, 0, 1), c(0.1, 0.2, 0.4, 0.7),                sample_weight = 10 * c(0.2, 0.3, 0.4, 0.1)) ## tf.Tensor( ## [[2 4] ##  [3 1]], shape=(2, 2), dtype=int64) m$result() ## tf.Tensor(0.1736111, shape=(), dtype=float32) model %>% compile(     optimizer = 'sgd',     loss = 'mse',     metrics = list(metric_binary_iou(         target_class_ids = 0L,         threshold = 0.5     )) )"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/metric_categorical_accuracy.html","id":null,"dir":"Reference","previous_headings":"","what":"Calculates how often predictions match one-hot labels. — metric_categorical_accuracy","title":"Calculates how often predictions match one-hot labels. — metric_categorical_accuracy","text":"can provide logits classes y_pred, since argmax logits probabilities . metric creates two local variables, total count used compute frequency y_pred matches y_true. frequency ultimately returned categorical accuracy: idempotent operation simply divides total count. y_pred y_true passed vectors probabilities, rather labels. necessary, use op_one_hot expand y_true vector. sample_weight NULL, weights default 1. Use sample_weight 0 mask values.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_categorical_accuracy.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Calculates how often predictions match one-hot labels. — metric_categorical_accuracy","text":"","code":"metric_categorical_accuracy(   y_true,   y_pred,   ...,   name = \"categorical_accuracy\",   dtype = NULL )"},{"path":"https://keras3.posit.co/dev/reference/metric_categorical_accuracy.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Calculates how often predictions match one-hot labels. — metric_categorical_accuracy","text":"y_true Tensor true targets. y_pred Tensor predicted targets. ... forward/backward compatability. name (Optional) string name metric instance. dtype (Optional) data type metric result.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_categorical_accuracy.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Calculates how often predictions match one-hot labels. — metric_categorical_accuracy","text":"y_true y_pred missing, Metric instance returned. Metric instance can passed directly compile(metrics = ), used standalone object. See ?Metric example usage. y_true y_pred provided, tensor computed value returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_categorical_accuracy.html","id":"usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Calculates how often predictions match one-hot labels. — metric_categorical_accuracy","text":"Standalone usage:           Usage compile() API:","code":"m <- metric_categorical_accuracy() m$update_state(rbind(c(0, 0, 1), c(0, 1, 0)), rbind(c(0.1, 0.9, 0.8),                 c(0.05, 0.95, 0))) m$result() ## tf.Tensor(0.5, shape=(), dtype=float32) m$reset_state() m$update_state(rbind(c(0, 0, 1), c(0, 1, 0)), rbind(c(0.1, 0.9, 0.8),                c(0.05, 0.95, 0)),                sample_weight = c(0.7, 0.3)) m$result() ## tf.Tensor(0.3, shape=(), dtype=float32) # 0.3 model %>% compile(optimizer = 'sgd',                   loss = 'categorical_crossentropy',                   metrics = list(metric_categorical_accuracy()))"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/metric_categorical_crossentropy.html","id":null,"dir":"Reference","previous_headings":"","what":"Computes the crossentropy metric between the labels and predictions. — metric_categorical_crossentropy","title":"Computes the crossentropy metric between the labels and predictions. — metric_categorical_crossentropy","text":"crossentropy metric class used multiple label classes (2 ). assumes labels one-hot encoded, e.g., labels values c(2, 0, 1), y_true rbind(c([0, 0, 1), c(1, 0, 0), c(0, 1, 0)).","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_categorical_crossentropy.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes the crossentropy metric between the labels and predictions. — metric_categorical_crossentropy","text":"","code":"metric_categorical_crossentropy(   y_true,   y_pred,   from_logits = FALSE,   label_smoothing = 0,   axis = -1L,   ...,   name = \"categorical_crossentropy\",   dtype = NULL )"},{"path":"https://keras3.posit.co/dev/reference/metric_categorical_crossentropy.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Computes the crossentropy metric between the labels and predictions. — metric_categorical_crossentropy","text":"y_true Tensor one-hot true targets. y_pred Tensor predicted targets. from_logits (Optional) Whether output expected logits tensor. default, consider output encodes probability distribution. label_smoothing (Optional) Float [0, 1]. > 0, label values smoothed, meaning confidence label values relaxed. e.g. label_smoothing=0.2 means use value 0.1 label \"0\" 0.9 label \"1\". axis (Optional) Defaults -1. dimension along entropy computed. ... forward/backward compatability. name (Optional) string name metric instance. dtype (Optional) data type metric result.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_categorical_crossentropy.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Computes the crossentropy metric between the labels and predictions. — metric_categorical_crossentropy","text":"y_true y_pred missing, Metric instance returned. Metric instance can passed directly compile(metrics = ), used standalone object. See ?Metric example usage. y_true y_pred provided, tensor computed value returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_categorical_crossentropy.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Computes the crossentropy metric between the labels and predictions. — metric_categorical_crossentropy","text":"Standalone usage:           Usage compile() API:","code":"# EPSILON = 1e-7, y = y_true, y` = y_pred # y` = clip_op_clip_by_value(output, EPSILON, 1. - EPSILON) # y` = rbind(c(0.05, 0.95, EPSILON), c(0.1, 0.8, 0.1)) # xent = -sum(y * log(y'), axis = -1) #      = -((log 0.95), (log 0.1)) #      = [0.051, 2.302] # Reduced xent = (0.051 + 2.302) / 2  m <- metric_categorical_crossentropy() m$update_state(rbind(c(0, 1, 0), c(0, 0, 1)),                rbind(c(0.05, 0.95, 0), c(0.1, 0.8, 0.1))) m$result() ## tf.Tensor(1.1769392, shape=(), dtype=float32) # 1.1769392 m$reset_state() m$update_state(rbind(c(0, 1, 0), c(0, 0, 1)),                rbind(c(0.05, 0.95, 0), c(0.1, 0.8, 0.1)),                sample_weight = c(0.3, 0.7)) m$result() ## tf.Tensor(1.6271976, shape=(), dtype=float32) model %>% compile(   optimizer = 'sgd',   loss = 'mse',   metrics = list(metric_categorical_crossentropy()))"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/metric_categorical_focal_crossentropy.html","id":null,"dir":"Reference","previous_headings":"","what":"Computes the categorical focal crossentropy loss. — metric_categorical_focal_crossentropy","title":"Computes the categorical focal crossentropy loss. — metric_categorical_focal_crossentropy","text":"Computes categorical focal crossentropy loss.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_categorical_focal_crossentropy.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes the categorical focal crossentropy loss. — metric_categorical_focal_crossentropy","text":"","code":"metric_categorical_focal_crossentropy(   y_true,   y_pred,   alpha = 0.25,   gamma = 2,   from_logits = FALSE,   label_smoothing = 0,   axis = -1L )"},{"path":"https://keras3.posit.co/dev/reference/metric_categorical_focal_crossentropy.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Computes the categorical focal crossentropy loss. — metric_categorical_focal_crossentropy","text":"y_true Tensor one-hot true targets. y_pred Tensor predicted targets. alpha weight balancing factor classes, default 0.25 mentioned reference. can list floats scalar. multi-class case, alpha may set inverse class frequency using compute_class_weight sklearn.utils. gamma focusing parameter, default 2.0 mentioned reference. helps gradually reduce importance given simple examples smooth manner. gamma = 0, focal effect categorical crossentropy. from_logits Whether y_pred expected logits tensor. default, assume y_pred encodes probability distribution. label_smoothing Float [0, 1]. > 0 smooth labels. example, 0.1, use 0.1 / num_classes non-target labels 0.9 + 0.1 / num_classes target labels. axis Defaults -1. dimension along entropy computed.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_categorical_focal_crossentropy.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Computes the categorical focal crossentropy loss. — metric_categorical_focal_crossentropy","text":"Categorical focal crossentropy loss value.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_categorical_focal_crossentropy.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Computes the categorical focal crossentropy loss. — metric_categorical_focal_crossentropy","text":"","code":"y_true <- rbind(c(0, 1, 0), c(0, 0, 1)) y_pred <- rbind(c(0.05, 0.9, 0.05), c(0.1, 0.85, 0.05)) loss <- loss_categorical_focal_crossentropy(y_true, y_pred) loss ## tf.Tensor([2.6340148e-04 6.7591208e-01], shape=(2), dtype=float32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/metric_categorical_hinge.html","id":null,"dir":"Reference","previous_headings":"","what":"Computes the categorical hinge metric between y_true and y_pred. — metric_categorical_hinge","title":"Computes the categorical hinge metric between y_true and y_pred. — metric_categorical_hinge","text":"Formula:   neg=maximum((1-y_true)*y_pred) pos=sum(y_true*y_pred)","code":"loss <- maximum(neg - pos + 1, 0)"},{"path":"https://keras3.posit.co/dev/reference/metric_categorical_hinge.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes the categorical hinge metric between y_true and y_pred. — metric_categorical_hinge","text":"","code":"metric_categorical_hinge(   y_true,   y_pred,   ...,   name = \"categorical_hinge\",   dtype = NULL )"},{"path":"https://keras3.posit.co/dev/reference/metric_categorical_hinge.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Computes the categorical hinge metric between y_true and y_pred. — metric_categorical_hinge","text":"y_true ground truth values. y_true values expected either {-1, +1} {0, 1} (.e. one-hot-encoded tensor) shape = [batch_size, d0, .. dN]. y_pred predicted values shape = [batch_size, d0, .. dN]. ... forward/backward compatability. name (Optional) string name metric instance. dtype (Optional) data type metric result.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_categorical_hinge.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Computes the categorical hinge metric between y_true and y_pred. — metric_categorical_hinge","text":"Categorical hinge loss values shape = [batch_size, d0, .. dN-1].","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_categorical_hinge.html","id":"usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes the categorical hinge metric between y_true and y_pred. — metric_categorical_hinge","text":"Standalone usage:","code":"m <- metric_categorical_hinge() m$update_state(rbind(c(0, 1), c(0, 0)), rbind(c(0.6, 0.4), c(0.4, 0.6))) m$result() ## tf.Tensor(1.4000001, shape=(), dtype=float32) m$reset_state() m$update_state(rbind(c(0, 1), c(0, 0)), rbind(c(0.6, 0.4), c(0.4, 0.6)),                sample_weight = c(1, 0)) m$result() ## tf.Tensor(1.2, shape=(), dtype=float32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/metric_concordance_correlation.html","id":null,"dir":"Reference","previous_headings":"","what":"Calculates the Concordance Correlation Coefficient (CCC). — metric_concordance_correlation","title":"Calculates the Concordance Correlation Coefficient (CCC). — metric_concordance_correlation","text":"Formula:   CCC evaluates agreement true values (y_true) predicted values (y_pred) considering precision accuracy. coefficient ranges -1 1, value 1 indicates perfect agreement. metric useful regression tasks important assess well predictions match true values, taking account correlation proximity 45-degree line perfect concordance.","code":"loss <- mean(   2 * (y_true - mean(y_true)) * (y_pred - mean(y_pred)) /     (var(y_true) + var(y_pred) + (mean(y_true) - mean(y_pred))^2) )"},{"path":"https://keras3.posit.co/dev/reference/metric_concordance_correlation.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Calculates the Concordance Correlation Coefficient (CCC). — metric_concordance_correlation","text":"","code":"metric_concordance_correlation(   y_true,   y_pred,   axis = -1L,   ...,   name = \"concordance_correlation\",   dtype = NULL )"},{"path":"https://keras3.posit.co/dev/reference/metric_concordance_correlation.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Calculates the Concordance Correlation Coefficient (CCC). — metric_concordance_correlation","text":"y_true Tensor true targets. y_pred Tensor predicted targets. axis (Optional) integer tuple integers axis/axes along compute metric. Defaults -1. ... forward/backward compatability. name (Optional) string name metric instance. dtype (Optional) data type metric result.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_concordance_correlation.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Calculates the Concordance Correlation Coefficient (CCC). — metric_concordance_correlation","text":"Usage compile() API:","code":"ccc <- metric_concordance_correlation(axis=-1) y_true <- rbind(c(0, 1, 0.5),                 c(1, 1, 0.2)) y_pred <- rbind(c(0.1, 0.9, 0.5),                 c(1, 0.9, 0.2)) ccc$update_state(y_true, y_pred) ccc$result() ## tf.Tensor(0.9816317, shape=(), dtype=float32) model |> compile(   optimizer = 'sgd',   loss = 'mean_squared_error',   metrics = c(metric_concordance_correlation()) )"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/metric_cosine_similarity.html","id":null,"dir":"Reference","previous_headings":"","what":"Computes the cosine similarity between the labels and predictions. — metric_cosine_similarity","title":"Computes the cosine similarity between the labels and predictions. — metric_cosine_similarity","text":"Formula:   See: Cosine Similarity. metric keeps average cosine similarity predictions labels stream data.","code":"loss <- sum(l2_norm(y_true) * l2_norm(y_pred))"},{"path":"https://keras3.posit.co/dev/reference/metric_cosine_similarity.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes the cosine similarity between the labels and predictions. — metric_cosine_similarity","text":"","code":"metric_cosine_similarity(   ...,   name = \"cosine_similarity\",   dtype = NULL,   axis = -1L )"},{"path":"https://keras3.posit.co/dev/reference/metric_cosine_similarity.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Computes the cosine similarity between the labels and predictions. — metric_cosine_similarity","text":"... forward/backward compatability. name (Optional) string name metric instance. dtype (Optional) data type metric result. axis (Optional) Defaults -1. dimension along cosine similarity computed.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_cosine_similarity.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Computes the cosine similarity between the labels and predictions. — metric_cosine_similarity","text":"Metric instance returned. Metric instance can passed directly compile(metrics = ), used standalone object. See ?Metric example usage.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_cosine_similarity.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Computes the cosine similarity between the labels and predictions. — metric_cosine_similarity","text":"Standalone usage:         Usage compile() API:","code":"m <- metric_cosine_similarity(axis=2) m$update_state(rbind(c(0., 1.), c(1., 1.)), rbind(c(1., 0.), c(1., 1.))) m$result() ## tf.Tensor(0.5, shape=(), dtype=float32) m$reset_state() m$update_state(rbind(c(0., 1.), c(1., 1.)), rbind(c(1., 0.), c(1., 1.)),                sample_weight = c(0.3, 0.7)) m$result() ## tf.Tensor(0.7, shape=(), dtype=float32) model %>% compile(   optimizer = 'sgd',   loss = 'mse',   metrics = list(metric_cosine_similarity(axis=2)))"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/metric_f1_score.html","id":null,"dir":"Reference","previous_headings":"","what":"Computes F-1 Score. — metric_f1_score","title":"Computes F-1 Score. — metric_f1_score","text":"Formula:   harmonic mean precision recall. output range [0, 1]. works multi-class multi-label classification.","code":"f1_score <- 2 * (precision * recall) / (precision + recall)"},{"path":"https://keras3.posit.co/dev/reference/metric_f1_score.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes F-1 Score. — metric_f1_score","text":"","code":"metric_f1_score(   ...,   average = NULL,   threshold = NULL,   name = \"f1_score\",   dtype = NULL )"},{"path":"https://keras3.posit.co/dev/reference/metric_f1_score.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Computes F-1 Score. — metric_f1_score","text":"... forward/backward compatability. average Type averaging performed data. Acceptable values NULL, \"micro\", \"macro\" \"weighted\". Defaults NULL. NULL, averaging performed result() return score class. \"micro\", compute metrics globally counting total true positives, false negatives false positives. \"macro\", compute metrics label, return unweighted mean. take label imbalance account. \"weighted\", compute metrics label, return average weighted support (number true instances label). alters \"macro\" account label imbalance. can result score precision recall. threshold Elements y_pred greater threshold converted 1, rest 0. threshold NULL, argmax y_pred converted 1, rest 0. name Optional. String name metric instance. dtype Optional. Data type metric result.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_f1_score.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Computes F-1 Score. — metric_f1_score","text":"Metric instance returned. Metric instance can passed directly compile(metrics = ), used standalone object. See ?Metric example usage.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_f1_score.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Computes F-1 Score. — metric_f1_score","text":"","code":"metric <- metric_f1_score(threshold = 0.5) y_true <- rbind(c(1, 1, 1),                 c(1, 0, 0),                 c(1, 1, 0)) y_pred <- rbind(c(0.2, 0.6, 0.7),                 c(0.2, 0.6, 0.6),                 c(0.6, 0.8, 0.0)) metric$update_state(y_true, y_pred) result <- metric$result() result ## tf.Tensor([0.49999997 0.79999995 0.66666657], shape=(3), dtype=float32)"},{"path":"https://keras3.posit.co/dev/reference/metric_f1_score.html","id":"returns","dir":"Reference","previous_headings":"","what":"Returns","title":"Computes F-1 Score. — metric_f1_score","text":"F-1 Score: float.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/metric_false_negatives.html","id":null,"dir":"Reference","previous_headings":"","what":"Calculates the number of false negatives. — metric_false_negatives","title":"Calculates the number of false negatives. — metric_false_negatives","text":"sample_weight given, calculates sum weights false negatives. metric creates one local variable, accumulator used keep track number false negatives. sample_weight NULL, weights default 1. Use sample_weight 0 mask values.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_false_negatives.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Calculates the number of false negatives. — metric_false_negatives","text":"","code":"metric_false_negatives(..., thresholds = NULL, name = NULL, dtype = NULL)"},{"path":"https://keras3.posit.co/dev/reference/metric_false_negatives.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Calculates the number of false negatives. — metric_false_negatives","text":"... forward/backward compatability. thresholds (Optional) Defaults 0.5. float value, Python list float threshold values [0, 1]. threshold compared prediction values determine truth value predictions (.e., threshold TRUE, FALSE). used loss function sets from_logits=TRUE (.e. sigmoid applied predictions), thresholds set 0. One metric value generated threshold value. name (Optional) string name metric instance. dtype (Optional) data type metric result.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_false_negatives.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Calculates the number of false negatives. — metric_false_negatives","text":"Metric instance returned. Metric instance can passed directly compile(metrics = ), used standalone object. See ?Metric example usage.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_false_negatives.html","id":"usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Calculates the number of false negatives. — metric_false_negatives","text":"Standalone usage:","code":"m <- metric_false_negatives() m$update_state(c(0, 1, 1, 1), c(0, 1, 0, 0)) m$result() ## tf.Tensor(2.0, shape=(), dtype=float32) m$reset_state() m$update_state(c(0, 1, 1, 1), c(0, 1, 0, 0), sample_weight=c(0, 0, 1, 0)) m$result() ## tf.Tensor(1.0, shape=(), dtype=float32) # 1.0"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/metric_false_positives.html","id":null,"dir":"Reference","previous_headings":"","what":"Calculates the number of false positives. — metric_false_positives","title":"Calculates the number of false positives. — metric_false_positives","text":"sample_weight given, calculates sum weights false positives. metric creates one local variable, accumulator used keep track number false positives. sample_weight NULL, weights default 1. Use sample_weight 0 mask values.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_false_positives.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Calculates the number of false positives. — metric_false_positives","text":"","code":"metric_false_positives(..., thresholds = NULL, name = NULL, dtype = NULL)"},{"path":"https://keras3.posit.co/dev/reference/metric_false_positives.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Calculates the number of false positives. — metric_false_positives","text":"... forward/backward compatability. thresholds (Optional) Defaults 0.5. float value, Python list float threshold values [0, 1]. threshold compared prediction values determine truth value predictions (.e., threshold TRUE, FALSE). used loss function sets from_logits=TRUE (.e. sigmoid applied predictions), thresholds set 0. One metric value generated threshold value. name (Optional) string name metric instance. dtype (Optional) data type metric result.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_false_positives.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Calculates the number of false positives. — metric_false_positives","text":"Metric instance returned. Metric instance can passed directly compile(metrics = ), used standalone object. See ?Metric example usage.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_false_positives.html","id":"usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Calculates the number of false positives. — metric_false_positives","text":"Standalone usage:","code":"m <- metric_false_positives() m$update_state(c(0, 1, 0, 0), c(0, 0, 1, 1)) m$result() ## tf.Tensor(2.0, shape=(), dtype=float32) m$reset_state() m$update_state(c(0, 1, 0, 0), c(0, 0, 1, 1), sample_weight = c(0, 0, 1, 0)) m$result() ## tf.Tensor(1.0, shape=(), dtype=float32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/metric_fbeta_score.html","id":null,"dir":"Reference","previous_headings":"","what":"Computes F-Beta score. — metric_fbeta_score","title":"Computes F-Beta score. — metric_fbeta_score","text":"Formula:   weighted harmonic mean precision recall. output range [0, 1]. works multi-class multi-label classification.","code":"b2 <- beta^2 f_beta_score <- (1 + b2) * (precision * recall) / (precision * b2 + recall)"},{"path":"https://keras3.posit.co/dev/reference/metric_fbeta_score.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes F-Beta score. — metric_fbeta_score","text":"","code":"metric_fbeta_score(   ...,   average = NULL,   beta = 1,   threshold = NULL,   name = \"fbeta_score\",   dtype = NULL )"},{"path":"https://keras3.posit.co/dev/reference/metric_fbeta_score.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Computes F-Beta score. — metric_fbeta_score","text":"... forward/backward compatability. average Type averaging performed across per-class results multi-class case. Acceptable values NULL, \"micro\", \"macro\" \"weighted\". Defaults NULL. NULL, averaging performed result() return score class. \"micro\", compute metrics globally counting total true positives, false negatives false positives. \"macro\", compute metrics label, return unweighted mean. take label imbalance account. \"weighted\", compute metrics label, return average weighted support (number true instances label). alters \"macro\" account label imbalance. can result score precision recall. beta Determines weight given recall harmonic mean precision recall (see pseudocode equation ). Defaults 1. threshold Elements y_pred greater threshold converted 1, rest 0. threshold NULL, argmax y_pred converted 1, rest 0. name Optional. String name metric instance. dtype Optional. Data type metric result.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_fbeta_score.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Computes F-Beta score. — metric_fbeta_score","text":"Metric instance returned. Metric instance can passed directly compile(metrics = ), used standalone object. See ?Metric example usage.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_fbeta_score.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Computes F-Beta score. — metric_fbeta_score","text":"","code":"metric <- metric_fbeta_score(beta = 2.0, threshold = 0.5) y_true <- rbind(c(1, 1, 1),                 c(1, 0, 0),                 c(1, 1, 0)) y_pred <- rbind(c(0.2, 0.6, 0.7),                 c(0.2, 0.6, 0.6),                 c(0.6, 0.8, 0.0)) metric$update_state(y_true, y_pred) metric$result() ## tf.Tensor([0.3846154  0.90909094 0.8333332 ], shape=(3), dtype=float32)"},{"path":"https://keras3.posit.co/dev/reference/metric_fbeta_score.html","id":"returns","dir":"Reference","previous_headings":"","what":"Returns","title":"Computes F-Beta score. — metric_fbeta_score","text":"F-Beta Score: float.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/metric_hinge.html","id":null,"dir":"Reference","previous_headings":"","what":"Computes the hinge metric between y_true and y_pred. — metric_hinge","title":"Computes the hinge metric between y_true and y_pred. — metric_hinge","text":"Formula:   y_true values expected -1 1. binary (0 1) labels provided convert -1 1.","code":"loss <- mean(maximum(1 - y_true * y_pred, 0), axis=-1)"},{"path":"https://keras3.posit.co/dev/reference/metric_hinge.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes the hinge metric between y_true and y_pred. — metric_hinge","text":"","code":"metric_hinge(y_true, y_pred, ..., name = \"hinge\", dtype = NULL)"},{"path":"https://keras3.posit.co/dev/reference/metric_hinge.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Computes the hinge metric between y_true and y_pred. — metric_hinge","text":"y_true ground truth values. y_true values expected -1 1. binary (0 1) labels provided converted -1 1 shape = [batch_size, d0, .. dN]. y_pred predicted values shape = [batch_size, d0, .. dN]. ... forward/backward compatability. name (Optional) string name metric instance. dtype (Optional) data type metric result.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_hinge.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Computes the hinge metric between y_true and y_pred. — metric_hinge","text":"y_true y_pred missing, Metric instance returned. Metric instance can passed directly compile(metrics = ), used standalone object. See ?Metric example usage. y_true y_pred provided, tensor computed value returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_hinge.html","id":"usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes the hinge metric between y_true and y_pred. — metric_hinge","text":"Standalone usage:","code":"m <- metric_hinge() m$update_state(rbind(c(0, 1), c(0, 0)), rbind(c(0.6, 0.4), c(0.4, 0.6))) m$result() ## tf.Tensor(1.3, shape=(), dtype=float32) m$reset_state() m$update_state(rbind(c(0, 1), c(0, 0)), rbind(c(0.6, 0.4), c(0.4, 0.6)),                sample_weight = c(1, 0)) m$result() ## tf.Tensor(1.1, shape=(), dtype=float32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/metric_huber.html","id":null,"dir":"Reference","previous_headings":"","what":"Computes Huber loss value. — metric_huber","title":"Computes Huber loss value. — metric_huber","text":"Formula:   See: Huber loss.","code":"for (x in error) {   if (abs(x) <= delta){     loss <- c(loss, (0.5 * x^2))   } else if (abs(x) > delta) {     loss <- c(loss, (delta * abs(x) - 0.5 * delta^2))   } } loss <- mean(loss)"},{"path":"https://keras3.posit.co/dev/reference/metric_huber.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes Huber loss value. — metric_huber","text":"","code":"metric_huber(y_true, y_pred, delta = 1)"},{"path":"https://keras3.posit.co/dev/reference/metric_huber.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Computes Huber loss value. — metric_huber","text":"y_true tensor true targets. y_pred tensor predicted targets. delta float, point Huber loss function changes quadratic linear. Defaults 1.0.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_huber.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Computes Huber loss value. — metric_huber","text":"Tensor one scalar loss entry per sample.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_huber.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Computes Huber loss value. — metric_huber","text":"","code":"y_true <- rbind(c(0, 1), c(0, 0)) y_pred <- rbind(c(0.6, 0.4), c(0.4, 0.6)) loss <- loss_huber(y_true, y_pred)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/metric_iou.html","id":null,"dir":"Reference","previous_headings":"","what":"Computes the Intersection-Over-Union metric for specific target classes. — metric_iou","title":"Computes the Intersection-Over-Union metric for specific target classes. — metric_iou","text":"Formula:   Intersection--Union common evaluation metric semantic image segmentation. compute IoUs, predictions accumulated confusion matrix, weighted sample_weight metric calculated . sample_weight NULL, weights default 1. Use sample_weight 0 mask values. Note, class first computes IoUs individual classes, returns mean IoUs classes specified target_class_ids. target_class_ids one id value, IoU specific class returned.","code":"iou <- true_positives / (true_positives + false_positives + false_negatives)"},{"path":"https://keras3.posit.co/dev/reference/metric_iou.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes the Intersection-Over-Union metric for specific target classes. — metric_iou","text":"","code":"metric_iou(   ...,   num_classes,   target_class_ids,   name = NULL,   dtype = NULL,   ignore_class = NULL,   sparse_y_true = TRUE,   sparse_y_pred = TRUE,   axis = -1L )"},{"path":"https://keras3.posit.co/dev/reference/metric_iou.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Computes the Intersection-Over-Union metric for specific target classes. — metric_iou","text":"... forward/backward compatability. num_classes possible number labels prediction task can . target_class_ids list target class ids metric returned. compute IoU specific class, list single id value provided. name (Optional) string name metric instance. dtype (Optional) data type metric result. ignore_class Optional integer. ID class ignored metric computation. useful, example, segmentation problems featuring \"void\" class (commonly -1 255) segmentation maps. default (ignore_class=NULL), classes considered. sparse_y_true Whether labels encoded using integers dense floating point vectors. FALSE, argmax function used determine sample's likely associated label. sparse_y_pred Whether predictions encoded using integers dense floating point vectors. FALSE, argmax function used determine sample's likely associated label. axis (Optional) -1 dimension containing logits. Defaults -1.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_iou.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Computes the Intersection-Over-Union metric for specific target classes. — metric_iou","text":"Metric instance returned. Metric instance can passed directly compile(metrics = ), used standalone object. See ?Metric example usage.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_iou.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Computes the Intersection-Over-Union metric for specific target classes. — metric_iou","text":"Standalone usage:                 Usage compile() API:","code":"m <- metric_iou(num_classes = 2L, target_class_ids = list(0L)) m$update_state(c(0, 0, 1, 1), c(0, 1, 0, 1)) ## tf.Tensor( ## [[1 1] ##  [1 1]], shape=(2, 2), dtype=int64) m$result() ## tf.Tensor(0.3333333, shape=(), dtype=float32) m$reset_state() m$update_state(c(0, 0, 1, 1), c(0, 1, 0, 1),                sample_weight = 10 * c(0.3, 0.3, 0.3, 0.1)) ## tf.Tensor( ## [[3 3] ##  [3 1]], shape=(2, 2), dtype=int64) m$result() ## tf.Tensor(0.3333333, shape=(), dtype=float32) model %>% compile(   optimizer = 'sgd',   loss = 'mse',   metrics = list(metric_iou(num_classes = 2L, target_class_ids = list(0L))))"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/metric_kl_divergence.html","id":null,"dir":"Reference","previous_headings":"","what":"Computes Kullback-Leibler divergence metric between y_true and — metric_kl_divergence","title":"Computes Kullback-Leibler divergence metric between y_true and — metric_kl_divergence","text":"Formula:   y_true y_pred expected probability distributions, values 0 1. get clipped [0, 1] range.","code":"loss <- y_true * log(y_true / y_pred)"},{"path":"https://keras3.posit.co/dev/reference/metric_kl_divergence.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes Kullback-Leibler divergence metric between y_true and — metric_kl_divergence","text":"","code":"metric_kl_divergence(y_true, y_pred, ..., name = \"kl_divergence\", dtype = NULL)"},{"path":"https://keras3.posit.co/dev/reference/metric_kl_divergence.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Computes Kullback-Leibler divergence metric between y_true and — metric_kl_divergence","text":"y_true Tensor true targets. y_pred Tensor predicted targets. ... forward/backward compatability. name (Optional) string name metric instance. dtype (Optional) data type metric result.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_kl_divergence.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Computes Kullback-Leibler divergence metric between y_true and — metric_kl_divergence","text":"y_true y_pred missing, Metric instance returned. Metric instance can passed directly compile(metrics = ), used standalone object. See ?Metric example usage. y_true y_pred provided, tensor computed value returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_kl_divergence.html","id":"usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes Kullback-Leibler divergence metric between y_true and — metric_kl_divergence","text":"Standalone usage:         Usage compile() API:","code":"m <- metric_kl_divergence() m$update_state(rbind(c(0, 1), c(0, 0)), rbind(c(0.6, 0.4), c(0.4, 0.6))) m$result() ## tf.Tensor(0.45814303, shape=(), dtype=float32) m$reset_state() m$update_state(rbind(c(0, 1), c(0, 0)), rbind(c(0.6, 0.4), c(0.4, 0.6)),                sample_weight = c(1, 0)) m$result() ## tf.Tensor(0.91628915, shape=(), dtype=float32) model %>% compile(optimizer = 'sgd',                   loss = 'mse',                   metrics = list(metric_kl_divergence()))"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/metric_log_cosh.html","id":null,"dir":"Reference","previous_headings":"","what":"Logarithm of the hyperbolic cosine of the prediction error. — metric_log_cosh","title":"Logarithm of the hyperbolic cosine of the prediction error. — metric_log_cosh","text":"Formula:   Note log(cosh(x)) approximately equal (x ** 2) / 2 small x abs(x) - log(2) large x. means 'logcosh' works mostly like mean squared error, strongly affected occasional wildly incorrect prediction.","code":"loss <- mean(log(cosh(y_pred - y_true)), axis=-1)"},{"path":"https://keras3.posit.co/dev/reference/metric_log_cosh.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Logarithm of the hyperbolic cosine of the prediction error. — metric_log_cosh","text":"","code":"metric_log_cosh(y_true, y_pred)"},{"path":"https://keras3.posit.co/dev/reference/metric_log_cosh.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Logarithm of the hyperbolic cosine of the prediction error. — metric_log_cosh","text":"y_true Ground truth values shape = [batch_size, d0, .. dN]. y_pred predicted values shape = [batch_size, d0, .. dN].","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_log_cosh.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Logarithm of the hyperbolic cosine of the prediction error. — metric_log_cosh","text":"Logcosh error values shape = [batch_size, d0, .. dN-1].","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_log_cosh.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Logarithm of the hyperbolic cosine of the prediction error. — metric_log_cosh","text":"","code":"y_true <- rbind(c(0., 1.), c(0., 0.)) y_pred <- rbind(c(1., 1.), c(0., 0.)) loss <- metric_log_cosh(y_true, y_pred) loss ## tf.Tensor([0.2168904 0.       ], shape=(2), dtype=float32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/metric_log_cosh_error.html","id":null,"dir":"Reference","previous_headings":"","what":"Computes the logarithm of the hyperbolic cosine of the prediction error. — metric_log_cosh_error","title":"Computes the logarithm of the hyperbolic cosine of the prediction error. — metric_log_cosh_error","text":"Formula:","code":"error <- y_pred - y_true logcosh <- mean(log((exp(error) + exp(-error))/2), axis=-1)"},{"path":"https://keras3.posit.co/dev/reference/metric_log_cosh_error.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes the logarithm of the hyperbolic cosine of the prediction error. — metric_log_cosh_error","text":"","code":"metric_log_cosh_error(..., name = \"logcosh\", dtype = NULL)"},{"path":"https://keras3.posit.co/dev/reference/metric_log_cosh_error.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Computes the logarithm of the hyperbolic cosine of the prediction error. — metric_log_cosh_error","text":"... forward/backward compatability. name (Optional) string name metric instance. dtype (Optional) data type metric result.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_log_cosh_error.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Computes the logarithm of the hyperbolic cosine of the prediction error. — metric_log_cosh_error","text":"Metric instance returned. Metric instance can passed directly compile(metrics = ), used standalone object. See ?Metric example usage.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_log_cosh_error.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Computes the logarithm of the hyperbolic cosine of the prediction error. — metric_log_cosh_error","text":"Standalone usage:         Usage compile() API:","code":"m <- metric_log_cosh_error() m$update_state(rbind(c(0, 1), c(0, 0)), rbind(c(1, 1), c(0, 0))) m$result() ## tf.Tensor(0.108445205, shape=(), dtype=float32) m$reset_state() m$update_state(rbind(c(0, 1), c(0, 0)), rbind(c(1, 1), c(0, 0)),                sample_weight = c(1, 0)) m$result() ## tf.Tensor(0.21689041, shape=(), dtype=float32) model %>% compile(optimizer = 'sgd',                   loss = 'mse',                   metrics = list(metric_log_cosh_error()))"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/metric_mean.html","id":null,"dir":"Reference","previous_headings":"","what":"Compute the (weighted) mean of the given values. — metric_mean","title":"Compute the (weighted) mean of the given values. — metric_mean","text":"example, values c(1, 3, 5, 7) mean 4. sample_weight specified c(1, 1, 0, 0) mean 2. metric creates two variables, total count. mean value returned simply total divided count.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_mean.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Compute the (weighted) mean of the given values. — metric_mean","text":"","code":"metric_mean(..., name = \"mean\", dtype = NULL)"},{"path":"https://keras3.posit.co/dev/reference/metric_mean.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Compute the (weighted) mean of the given values. — metric_mean","text":"... forward/backward compatability. name (Optional) string name metric instance. dtype (Optional) data type metric result.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_mean.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Compute the (weighted) mean of the given values. — metric_mean","text":"Metric instance returned. Metric instance can passed directly compile(metrics = ), used standalone object. See ?Metric example usage.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_mean.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Compute the (weighted) mean of the given values. — metric_mean","text":"","code":"m <- metric_mean() m$update_state(c(1, 3, 5, 7)) m$result() ## tf.Tensor(4.0, shape=(), dtype=float32) # calling a metric directly is equivalent to calling # m$update_state(); m$result() m <- metric_mean() m(c(1, 3, 5, 7)) ## tf.Tensor(4.0, shape=(), dtype=float32) m$reset_state() m$update_state(c(1, 3, 5, 7), sample_weight = c(1, 1, 0, 0)) m$result() ## tf.Tensor(2.0, shape=(), dtype=float32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/metric_mean_absolute_error.html","id":null,"dir":"Reference","previous_headings":"","what":"Computes the mean absolute error between the labels and predictions. — metric_mean_absolute_error","title":"Computes the mean absolute error between the labels and predictions. — metric_mean_absolute_error","text":"Formula:","code":"loss <- mean(abs(y_true - y_pred))"},{"path":"https://keras3.posit.co/dev/reference/metric_mean_absolute_error.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes the mean absolute error between the labels and predictions. — metric_mean_absolute_error","text":"","code":"metric_mean_absolute_error(   y_true,   y_pred,   ...,   name = \"mean_absolute_error\",   dtype = NULL )"},{"path":"https://keras3.posit.co/dev/reference/metric_mean_absolute_error.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Computes the mean absolute error between the labels and predictions. — metric_mean_absolute_error","text":"y_true Ground truth values shape = [batch_size, d0, .. dN]. y_pred predicted values shape = [batch_size, d0, .. dN]. ... forward/backward compatability. name (Optional) string name metric instance. dtype (Optional) data type metric result.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_mean_absolute_error.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Computes the mean absolute error between the labels and predictions. — metric_mean_absolute_error","text":"y_true y_pred missing, Metric instance returned. Metric instance can passed directly compile(metrics = ), used standalone object. See ?Metric example usage. y_true y_pred provided, tensor computed value returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_mean_absolute_error.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Computes the mean absolute error between the labels and predictions. — metric_mean_absolute_error","text":"Standalone usage:         Usage compile() API:","code":"m <- metric_mean_absolute_error() m$update_state(rbind(c(0, 1), c(0, 0)), rbind(c(1, 1), c(0, 0))) m$result() ## tf.Tensor(0.25, shape=(), dtype=float32) m$reset_state() m$update_state(rbind(c(0, 1), c(0, 0)), rbind(c(1, 1), c(0, 0)),                sample_weight = c(1, 0)) m$result() ## tf.Tensor(0.5, shape=(), dtype=float32) model %>% compile(     optimizer = 'sgd',     loss = 'mse',     metrics = list(metric_mean_absolute_error()))"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/metric_mean_absolute_percentage_error.html","id":null,"dir":"Reference","previous_headings":"","what":"Computes mean absolute percentage error between y_true and y_pred. — metric_mean_absolute_percentage_error","title":"Computes mean absolute percentage error between y_true and y_pred. — metric_mean_absolute_percentage_error","text":"Formula:   Division zero prevented dividing maximum(y_true, epsilon) epsilon = keras$backend$epsilon() (default 1e-7).","code":"loss <- 100 * mean(abs((y_true - y_pred) / y_true), axis=-1)"},{"path":"https://keras3.posit.co/dev/reference/metric_mean_absolute_percentage_error.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes mean absolute percentage error between y_true and y_pred. — metric_mean_absolute_percentage_error","text":"","code":"metric_mean_absolute_percentage_error(   y_true,   y_pred,   ...,   name = \"mean_absolute_percentage_error\",   dtype = NULL )"},{"path":"https://keras3.posit.co/dev/reference/metric_mean_absolute_percentage_error.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Computes mean absolute percentage error between y_true and y_pred. — metric_mean_absolute_percentage_error","text":"y_true Ground truth values shape = [batch_size, d0, .. dN]. y_pred predicted values shape = [batch_size, d0, .. dN]. ... forward/backward compatability. name (Optional) string name metric instance. dtype (Optional) data type metric result.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_mean_absolute_percentage_error.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Computes mean absolute percentage error between y_true and y_pred. — metric_mean_absolute_percentage_error","text":"y_true y_pred missing, Metric instance returned. Metric instance can passed directly compile(metrics = ), used standalone object. See ?Metric example usage. y_true y_pred provided, tensor computed value returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_mean_absolute_percentage_error.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Computes mean absolute percentage error between y_true and y_pred. — metric_mean_absolute_percentage_error","text":"Standalone usage:         Usage compile() API:","code":"m <- metric_mean_absolute_percentage_error() m$update_state(rbind(c(0, 1), c(0, 0)), rbind(c(1, 1), c(0, 0))) m$result() ## tf.Tensor(2.5e+08, shape=(), dtype=float32) m$reset_state() m$update_state(rbind(c(0, 1), c(0, 0)), rbind(c(1, 1), c(0, 0)),                sample_weight = c(1, 0)) m$result() ## tf.Tensor(5e+08, shape=(), dtype=float32) model %>% compile(     optimizer = 'sgd',     loss = 'mse',     metrics = list(metric_mean_absolute_percentage_error()))"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/metric_mean_iou.html","id":null,"dir":"Reference","previous_headings":"","what":"Computes the mean Intersection-Over-Union metric. — metric_mean_iou","title":"Computes the mean Intersection-Over-Union metric. — metric_mean_iou","text":"Formula:   Intersection--Union common evaluation metric semantic image segmentation. compute IoUs, predictions accumulated confusion matrix, weighted sample_weight metric calculated . sample_weight NULL, weights default 1. Use sample_weight 0 mask values. Note class first computes IoUs individual classes, returns mean values.","code":"iou <- true_positives / (true_positives + false_positives + false_negatives)"},{"path":"https://keras3.posit.co/dev/reference/metric_mean_iou.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes the mean Intersection-Over-Union metric. — metric_mean_iou","text":"","code":"metric_mean_iou(   ...,   num_classes,   name = NULL,   dtype = NULL,   ignore_class = NULL,   sparse_y_true = TRUE,   sparse_y_pred = TRUE,   axis = -1L )"},{"path":"https://keras3.posit.co/dev/reference/metric_mean_iou.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Computes the mean Intersection-Over-Union metric. — metric_mean_iou","text":"... forward/backward compatability. num_classes possible number labels prediction task can . value must provided, since confusion matrix dimension = [num_classes, num_classes] allocated. name (Optional) string name metric instance. dtype (Optional) data type metric result. ignore_class Optional integer. ID class ignored metric computation. useful, example, segmentation problems featuring \"void\" class (commonly -1 255) segmentation maps. default (ignore_class=NULL), classes considered. sparse_y_true Whether labels encoded using integers dense floating point vectors. FALSE, argmax function used determine sample's likely associated label. sparse_y_pred Whether predictions encoded using integers dense floating point vectors. FALSE, argmax function used determine sample's likely associated label. axis (Optional) dimension containing logits. Defaults -1.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_mean_iou.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Computes the mean Intersection-Over-Union metric. — metric_mean_iou","text":"Metric instance returned. Metric instance can passed directly compile(metrics = ), used standalone object. See ?Metric example usage.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_mean_iou.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Computes the mean Intersection-Over-Union metric. — metric_mean_iou","text":"Standalone usage:                 Usage compile() API:","code":"# cm = [[1, 1], #       [1, 1]] # sum_row = [2, 2], sum_col = [2, 2], true_positives = [1, 1] # iou = true_positives / (sum_row + sum_col - true_positives)) # result = (1 / (2 + 2 - 1) + 1 / (2 + 2 - 1)) / 2 = 0.33 m <- metric_mean_iou(num_classes = 2) m$update_state(c(0, 0, 1, 1), c(0, 1, 0, 1)) ## tf.Tensor( ## [[1 1] ##  [1 1]], shape=(2, 2), dtype=int64) m$result() ## tf.Tensor(0.33333334, shape=(), dtype=float32) m$reset_state() m$update_state(c(0, 0, 1, 1), c(0, 1, 0, 1),                sample_weight= 10 * c(0.3, 0.3, 0.3, 0.1)) ## tf.Tensor( ## [[3 3] ##  [3 1]], shape=(2, 2), dtype=int64) m$result() ## tf.Tensor(0.23809525, shape=(), dtype=float32) model %>% compile(   optimizer = 'sgd',   loss = 'mse',   metrics = list(metric_mean_iou(num_classes=2)))"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/metric_mean_squared_error.html","id":null,"dir":"Reference","previous_headings":"","what":"Computes the mean squared error between y_true and y_pred. — metric_mean_squared_error","title":"Computes the mean squared error between y_true and y_pred. — metric_mean_squared_error","text":"Formula:","code":"loss <- mean(square(y_true - y_pred))"},{"path":"https://keras3.posit.co/dev/reference/metric_mean_squared_error.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes the mean squared error between y_true and y_pred. — metric_mean_squared_error","text":"","code":"metric_mean_squared_error(   y_true,   y_pred,   ...,   name = \"mean_squared_error\",   dtype = NULL )"},{"path":"https://keras3.posit.co/dev/reference/metric_mean_squared_error.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Computes the mean squared error between y_true and y_pred. — metric_mean_squared_error","text":"y_true Ground truth values shape = [batch_size, d0, .. dN]. y_pred predicted values shape = [batch_size, d0, .. dN]. ... forward/backward compatability. name (Optional) string name metric instance. dtype (Optional) data type metric result.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_mean_squared_error.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Computes the mean squared error between y_true and y_pred. — metric_mean_squared_error","text":"y_true y_pred missing, Metric instance returned. Metric instance can passed directly compile(metrics = ), used standalone object. See ?Metric example usage. y_true y_pred provided, tensor computed value returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_mean_squared_error.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Computes the mean squared error between y_true and y_pred. — metric_mean_squared_error","text":"","code":"m <- metric_mean_squared_error() m$update_state(rbind(c(0, 1), c(0, 0)), rbind(c(1, 1), c(0, 0))) m$result() ## tf.Tensor(0.25, shape=(), dtype=float32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/metric_mean_squared_logarithmic_error.html","id":null,"dir":"Reference","previous_headings":"","what":"Computes mean squared logarithmic error between y_true and y_pred. — metric_mean_squared_logarithmic_error","title":"Computes mean squared logarithmic error between y_true and y_pred. — metric_mean_squared_logarithmic_error","text":"Formula:   Note y_pred y_true less equal 0. Negative values 0 values replaced keras$backend$epsilon() (default 1e-7).","code":"loss <- mean(square(log(y_true + 1) - log(y_pred + 1)), axis=-1)"},{"path":"https://keras3.posit.co/dev/reference/metric_mean_squared_logarithmic_error.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes mean squared logarithmic error between y_true and y_pred. — metric_mean_squared_logarithmic_error","text":"","code":"metric_mean_squared_logarithmic_error(   y_true,   y_pred,   ...,   name = \"mean_squared_logarithmic_error\",   dtype = NULL )"},{"path":"https://keras3.posit.co/dev/reference/metric_mean_squared_logarithmic_error.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Computes mean squared logarithmic error between y_true and y_pred. — metric_mean_squared_logarithmic_error","text":"y_true Ground truth values shape = [batch_size, d0, .. dN]. y_pred predicted values shape = [batch_size, d0, .. dN]. ... forward/backward compatability. name (Optional) string name metric instance. dtype (Optional) data type metric result.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_mean_squared_logarithmic_error.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Computes mean squared logarithmic error between y_true and y_pred. — metric_mean_squared_logarithmic_error","text":"y_true y_pred missing, Metric instance returned. Metric instance can passed directly compile(metrics = ), used standalone object. See ?Metric example usage. y_true y_pred provided, tensor computed value returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_mean_squared_logarithmic_error.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Computes mean squared logarithmic error between y_true and y_pred. — metric_mean_squared_logarithmic_error","text":"Standalone usage:         Usage compile() API:","code":"m <- metric_mean_squared_logarithmic_error() m$update_state(rbind(c(0, 1), c(0, 0)), rbind(c(1, 1), c(0, 0))) m$result() ## tf.Tensor(0.12011322, shape=(), dtype=float32) m$reset_state() m$update_state(rbind(c(0, 1), c(0, 0)), rbind(c(1, 1), c(0, 0)),                sample_weight = c(1, 0)) m$result() ## tf.Tensor(0.24022643, shape=(), dtype=float32) model %>% compile(   optimizer = 'sgd',   loss = 'mse',   metrics = list(metric_mean_squared_logarithmic_error()))"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/metric_mean_wrapper.html","id":null,"dir":"Reference","previous_headings":"","what":"Wrap a stateless metric function with the Mean metric. — metric_mean_wrapper","title":"Wrap a stateless metric function with the Mean metric. — metric_mean_wrapper","text":"use class quickly build mean metric function. function needs signature fn(y_true, y_pred) return per-sample loss array. metric_mean_wrapper$result() return average metric value across samples seen far. example:","code":"mse <- function(y_true, y_pred) {   (y_true - y_pred)^2 }  mse_metric <- metric_mean_wrapper(fn = mse) mse_metric$update_state(c(0, 1), c(1, 1)) mse_metric$result() ## tf.Tensor(0.5, shape=(), dtype=float32)"},{"path":"https://keras3.posit.co/dev/reference/metric_mean_wrapper.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Wrap a stateless metric function with the Mean metric. — metric_mean_wrapper","text":"","code":"metric_mean_wrapper(..., fn, name = NULL, dtype = NULL)"},{"path":"https://keras3.posit.co/dev/reference/metric_mean_wrapper.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Wrap a stateless metric function with the Mean metric. — metric_mean_wrapper","text":"... Keyword arguments pass fn. fn metric function wrap, signature fn(y_true, y_pred). name (Optional) string name metric instance. dtype (Optional) data type metric result.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_mean_wrapper.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Wrap a stateless metric function with the Mean metric. — metric_mean_wrapper","text":"Metric instance returned. Metric instance can passed directly compile(metrics = ), used standalone object. See ?Metric example usage.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/metric_one_hot_iou.html","id":null,"dir":"Reference","previous_headings":"","what":"Computes the Intersection-Over-Union metric for one-hot encoded labels. — metric_one_hot_iou","title":"Computes the Intersection-Over-Union metric for one-hot encoded labels. — metric_one_hot_iou","text":"Formula:   Intersection--Union common evaluation metric semantic image segmentation. compute IoUs, predictions accumulated confusion matrix, weighted sample_weight metric calculated . sample_weight NULL, weights default 1. Use sample_weight 0 mask values. class can used compute IoU multi-class classification tasks labels one-hot encoded (last axis one dimension per class). Note predictions also shape. compute IoU, first labels predictions converted back integer format taking argmax class axis. computation steps base IoU class apply. Note, one channel labels predictions, class class IoU. case, use IoU instead. Also, make sure num_classes equal number classes data, avoid \"labels bound\" error confusion matrix computed.","code":"iou <- true_positives / (true_positives + false_positives + false_negatives)"},{"path":"https://keras3.posit.co/dev/reference/metric_one_hot_iou.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes the Intersection-Over-Union metric for one-hot encoded labels. — metric_one_hot_iou","text":"","code":"metric_one_hot_iou(   ...,   num_classes,   target_class_ids,   name = NULL,   dtype = NULL,   ignore_class = NULL,   sparse_y_pred = FALSE,   axis = -1L )"},{"path":"https://keras3.posit.co/dev/reference/metric_one_hot_iou.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Computes the Intersection-Over-Union metric for one-hot encoded labels. — metric_one_hot_iou","text":"... forward/backward compatability. num_classes possible number labels prediction task can . target_class_ids list list target class ids metric returned. compute IoU specific class, list (list) single id value provided. name (Optional) string name metric instance. dtype (Optional) data type metric result. ignore_class Optional integer. ID class ignored metric computation. useful, example, segmentation problems featuring \"void\" class (commonly -1 255) segmentation maps. default (ignore_class=NULL), classes considered. sparse_y_pred Whether predictions encoded using integers dense floating point vectors. FALSE, argmax function used determine sample's likely associated label. axis (Optional) dimension containing logits. Defaults -1.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_one_hot_iou.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Computes the Intersection-Over-Union metric for one-hot encoded labels. — metric_one_hot_iou","text":"Metric instance returned. Metric instance can passed directly compile(metrics = ), used standalone object. See ?Metric example usage.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_one_hot_iou.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Computes the Intersection-Over-Union metric for one-hot encoded labels. — metric_one_hot_iou","text":"Standalone usage:         Usage compile() API:","code":"y_true <- rbind(c(0, 0, 1),                 c(1, 0, 0),                 c(0, 1, 0),                 c(1, 0, 0)) y_pred <- rbind(c(0.2, 0.3, 0.5),                 c(0.1, 0.2, 0.7),                 c(0.5, 0.3, 0.1),                 c(0.1, 0.4, 0.5)) sample_weight <- c(0.1, 0.2, 0.3, 0.4)  m <- metric_one_hot_iou(num_classes = 3, target_class_ids = c(0, 2)) m$update_state(y_true = y_true,                y_pred = y_pred,                sample_weight = 10*sample_weight) ## tf.Tensor( ## [[0 0 6] ##  [3 0 0] ##  [0 0 1]], shape=(3, 3), dtype=int64) m$result() ## tf.Tensor(0.071428575, shape=(), dtype=float32) model %>% compile(   optimizer = 'sgd',   loss = 'mse',   metrics = list(metric_one_hot_iou(     num_classes = 3L,     target_class_id = list(1L)   )) )"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/metric_one_hot_mean_iou.html","id":null,"dir":"Reference","previous_headings":"","what":"Computes mean Intersection-Over-Union metric for one-hot encoded labels. — metric_one_hot_mean_iou","title":"Computes mean Intersection-Over-Union metric for one-hot encoded labels. — metric_one_hot_mean_iou","text":"Formula:   Intersection--Union common evaluation metric semantic image segmentation. compute IoUs, predictions accumulated confusion matrix, weighted sample_weight metric calculated . sample_weight NULL, weights default 1. Use sample_weight 0 mask values. class can used compute mean IoU multi-class classification tasks labels one-hot encoded (last axis one dimension per class). Note predictions also shape. compute mean IoU, first labels predictions converted back integer format taking argmax class axis. computation steps base MeanIoU class apply. Note, one channel labels predictions, class class metric_mean_iou. case, use metric_mean_iou instead. Also, make sure num_classes equal number classes data, avoid \"labels bound\" error confusion matrix computed.","code":"iou <- true_positives / (true_positives + false_positives + false_negatives)"},{"path":"https://keras3.posit.co/dev/reference/metric_one_hot_mean_iou.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes mean Intersection-Over-Union metric for one-hot encoded labels. — metric_one_hot_mean_iou","text":"","code":"metric_one_hot_mean_iou(   ...,   num_classes,   name = NULL,   dtype = NULL,   ignore_class = NULL,   sparse_y_pred = FALSE,   axis = -1L )"},{"path":"https://keras3.posit.co/dev/reference/metric_one_hot_mean_iou.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Computes mean Intersection-Over-Union metric for one-hot encoded labels. — metric_one_hot_mean_iou","text":"... forward/backward compatability. num_classes possible number labels prediction task can . name (Optional) string name metric instance. dtype (Optional) data type metric result. ignore_class Optional integer. ID class ignored metric computation. useful, example, segmentation problems featuring \"void\" class (commonly -1 255) segmentation maps. default (ignore_class=NULL), classes considered. sparse_y_pred Whether predictions encoded using natural numbers probability distribution vectors. FALSE, argmax function used determine sample's likely associated label. axis (Optional) dimension containing logits. Defaults -1.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_one_hot_mean_iou.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Computes mean Intersection-Over-Union metric for one-hot encoded labels. — metric_one_hot_mean_iou","text":"Metric instance returned. Metric instance can passed directly compile(metrics = ), used standalone object. See ?Metric example usage.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_one_hot_mean_iou.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Computes mean Intersection-Over-Union metric for one-hot encoded labels. — metric_one_hot_mean_iou","text":"Standalone usage:         Usage compile() API:","code":"y_true <- rbind(c(0, 0, 1), c(1, 0, 0), c(0, 1, 0), c(1, 0, 0)) y_pred <- rbind(c(0.2, 0.3, 0.5), c(0.1, 0.2, 0.7), c(0.5, 0.3, 0.1),                 c(0.1, 0.4, 0.5)) sample_weight <- 10 * c(0.1, 0.2, 0.3, 0.4) m <- metric_one_hot_mean_iou(num_classes = 3L) m$update_state(     y_true = y_true, y_pred = y_pred, sample_weight = sample_weight) ## tf.Tensor( ## [[0 0 6] ##  [3 0 0] ##  [0 0 1]], shape=(3, 3), dtype=int64) m$result() ## tf.Tensor(0.04761905, shape=(), dtype=float32) model %>% compile(     optimizer = 'sgd',     loss = 'mse',     metrics = list(metric_one_hot_mean_iou(num_classes = 3L)))"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/metric_pearson_correlation.html","id":null,"dir":"Reference","previous_headings":"","what":"Calculates the Pearson Correlation Coefficient (PCC). — metric_pearson_correlation","title":"Calculates the Pearson Correlation Coefficient (PCC). — metric_pearson_correlation","text":"Formula:   PCC measures linear relationship true values (y_true) predicted values (y_pred). coefficient ranges -1 1, value 1 implies perfect positive linear correlation, 0 indicates linear correlation, -1 indicates perfect negative linear correlation. metric widely used regression tasks strength linear relationship predictions true labels important evaluation criterion.","code":"loss = mean(l2norm(y_true - mean(y_true) * l2norm(y_pred - mean(y_pred)))"},{"path":"https://keras3.posit.co/dev/reference/metric_pearson_correlation.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Calculates the Pearson Correlation Coefficient (PCC). — metric_pearson_correlation","text":"","code":"metric_pearson_correlation(   y_true,   y_pred,   axis = -1L,   ...,   name = \"pearson_correlation\",   dtype = NULL )"},{"path":"https://keras3.posit.co/dev/reference/metric_pearson_correlation.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Calculates the Pearson Correlation Coefficient (PCC). — metric_pearson_correlation","text":"y_true Tensor true targets. y_pred Tensor predicted targets. axis (Optional) integer tuple integers axis/axes along compute metric. Defaults -1. ... forward/backward compatability. name (Optional) string name metric instance. dtype (Optional) data type metric result.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_pearson_correlation.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Calculates the Pearson Correlation Coefficient (PCC). — metric_pearson_correlation","text":"Usage compile() API:","code":"pcc <- metric_pearson_correlation(axis = -1) y_true <- rbind(c(0, 1, 0.5),                 c(1, 1, 0.2)) y_pred <- rbind(c(0.1, 0.9, 0.5),                 c(1, 0.9, 0.2)) pcc$update_state(y_true, y_pred) pcc$result() ## tf.Tensor(0.99669963, shape=(), dtype=float32) # equivalent operation using R's stats::cor() mean(sapply(1:nrow(y_true), function(i) {   cor(y_true[i, ], y_pred[i, ]) })) ## [1] 0.9966996 model |> compile(   optimizer = 'sgd',   loss = 'mean_squared_error',   metrics = c(keras.metrics.PearsonCorrelation()) )"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/metric_poisson.html","id":null,"dir":"Reference","previous_headings":"","what":"Computes the Poisson metric between y_true and y_pred. — metric_poisson","title":"Computes the Poisson metric between y_true and y_pred. — metric_poisson","text":"Formula:","code":"metric <- y_pred - y_true * log(y_pred)"},{"path":"https://keras3.posit.co/dev/reference/metric_poisson.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes the Poisson metric between y_true and y_pred. — metric_poisson","text":"","code":"metric_poisson(y_true, y_pred, ..., name = \"poisson\", dtype = NULL)"},{"path":"https://keras3.posit.co/dev/reference/metric_poisson.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Computes the Poisson metric between y_true and y_pred. — metric_poisson","text":"y_true Ground truth values. shape = [batch_size, d0, .. dN]. y_pred predicted values. shape = [batch_size, d0, .. dN]. ... forward/backward compatability. name (Optional) string name metric instance. dtype (Optional) data type metric result.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_poisson.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Computes the Poisson metric between y_true and y_pred. — metric_poisson","text":"y_true y_pred missing, Metric instance returned. Metric instance can passed directly compile(metrics = ), used standalone object. See ?Metric example usage. y_true y_pred provided, tensor computed value returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_poisson.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Computes the Poisson metric between y_true and y_pred. — metric_poisson","text":"Standalone usage:         Usage compile() API:","code":"m <- metric_poisson() m$update_state(rbind(c(0, 1), c(0, 0)), rbind(c(1, 1), c(0, 0))) m$result() ## tf.Tensor(0.49999997, shape=(), dtype=float32) m$reset_state() m$update_state(rbind(c(0, 1), c(0, 0)), rbind(c(1, 1), c(0, 0)),                sample_weight = c(1, 0)) m$result() ## tf.Tensor(0.99999994, shape=(), dtype=float32) model %>% compile(   optimizer = 'sgd',   loss = 'mse',   metrics = list(metric_poisson()) )"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/metric_precision.html","id":null,"dir":"Reference","previous_headings":"","what":"Computes the precision of the predictions with respect to the labels. — metric_precision","title":"Computes the precision of the predictions with respect to the labels. — metric_precision","text":"metric creates two local variables, true_positives false_positives used compute precision. value ultimately returned precision, idempotent operation simply divides true_positives sum true_positives false_positives. sample_weight NULL, weights default 1. Use sample_weight 0 mask values. top_k set, calculate precision often average class among top-k classes highest predicted values batch entry correct can found label entry. class_id specified, calculate precision considering entries batch class_id threshold /top-k highest predictions, computing fraction class_id indeed correct label.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_precision.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes the precision of the predictions with respect to the labels. — metric_precision","text":"","code":"metric_precision(   ...,   thresholds = NULL,   top_k = NULL,   class_id = NULL,   name = NULL,   dtype = NULL )"},{"path":"https://keras3.posit.co/dev/reference/metric_precision.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Computes the precision of the predictions with respect to the labels. — metric_precision","text":"... forward/backward compatability. thresholds (Optional) float value, Python list float threshold values [0, 1]. threshold compared prediction values determine truth value predictions (.e., threshold TRUE, FALSE). used loss function sets from_logits=TRUE (.e. sigmoid applied predictions), thresholds set 0. One metric value generated threshold value. neither thresholds top_k set, default calculate precision thresholds=0.5. top_k (Optional) Unset default. int value specifying top-k predictions consider calculating precision. class_id (Optional) Integer class ID want binary metrics. must half-open interval [0, num_classes), num_classes last dimension predictions. name (Optional) string name metric instance. dtype (Optional) data type metric result.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_precision.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Computes the precision of the predictions with respect to the labels. — metric_precision","text":"Metric instance returned. Metric instance can passed directly compile(metrics = ), used standalone object. See ?Metric example usage.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_precision.html","id":"usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes the precision of the predictions with respect to the labels. — metric_precision","text":"Standalone usage:                 Usage compile() API:   Usage loss from_logits=TRUE:","code":"m <- metric_precision() m$update_state(c(0, 1, 1, 1),                c(1, 0, 1, 1)) m$result() |> as.double() |> signif() ## [1] 0.666667 m$reset_state() m$update_state(c(0, 1, 1, 1),                c(1, 0, 1, 1),                sample_weight = c(0, 0, 1, 0)) m$result() |> as.double() |> signif() ## [1] 1 # With top_k=2, it will calculate precision over y_true[1:2] # and y_pred[1:2] m <- metric_precision(top_k = 2) m$update_state(c(0, 0, 1, 1), c(1, 1, 1, 1)) m$result() ## tf.Tensor(0.0, shape=(), dtype=float32) # With top_k=4, it will calculate precision over y_true[1:4] # and y_pred[1:4] m <- metric_precision(top_k = 4) m$update_state(c(0, 0, 1, 1), c(1, 1, 1, 1)) m$result() ## tf.Tensor(0.5, shape=(), dtype=float32) model |> compile(   optimizer = 'sgd',   loss = 'binary_crossentropy',   metrics = list(metric_precision()) ) model |> compile(   optimizer = 'adam',   loss = loss_binary_crossentropy(from_logits = TRUE),   metrics = list(metric_precision(thresholds = 0)) )"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/metric_precision_at_recall.html","id":null,"dir":"Reference","previous_headings":"","what":"Computes best precision where recall is >= specified value. — metric_precision_at_recall","title":"Computes best precision where recall is >= specified value. — metric_precision_at_recall","text":"metric creates four local variables, true_positives, true_negatives, false_positives false_negatives used compute precision given recall. threshold given recall value computed used evaluate corresponding precision. sample_weight NULL, weights default 1. Use sample_weight 0 mask values. class_id specified, calculate precision considering entries batch class_id threshold predictions, computing fraction class_id indeed correct label.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_precision_at_recall.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes best precision where recall is >= specified value. — metric_precision_at_recall","text":"","code":"metric_precision_at_recall(   ...,   recall,   num_thresholds = 200L,   class_id = NULL,   name = NULL,   dtype = NULL )"},{"path":"https://keras3.posit.co/dev/reference/metric_precision_at_recall.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Computes best precision where recall is >= specified value. — metric_precision_at_recall","text":"... forward/backward compatability. recall scalar value range [0, 1]. num_thresholds (Optional) Defaults 200. number thresholds use matching given recall. class_id (Optional) Integer class ID want binary metrics. must half-open interval [0, num_classes), num_classes last dimension predictions. name (Optional) string name metric instance. dtype (Optional) data type metric result.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_precision_at_recall.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Computes best precision where recall is >= specified value. — metric_precision_at_recall","text":"Metric instance returned. Metric instance can passed directly compile(metrics = ), used standalone object. See ?Metric example usage.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_precision_at_recall.html","id":"usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes best precision where recall is >= specified value. — metric_precision_at_recall","text":"Standalone usage:         Usage compile() API:","code":"m <- metric_precision_at_recall(recall = 0.5) m$update_state(c(0,   0,   0,   1,   1),                c(0, 0.3, 0.8, 0.3, 0.8)) m$result() ## tf.Tensor(0.5, shape=(), dtype=float32) m$reset_state() m$update_state(c(0,   0,   0,   1,   1),                c(0, 0.3, 0.8, 0.3, 0.8),                sample_weight = c(2, 2, 2, 1, 1)) m$result() ## tf.Tensor(0.33333334, shape=(), dtype=float32) model |> compile(   optimizer = 'sgd',   loss = 'binary_crossentropy',   metrics = list(metric_precision_at_recall(recall = 0.8)) )"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/metric_r2_score.html","id":null,"dir":"Reference","previous_headings":"","what":"Computes R2 score. — metric_r2_score","title":"Computes R2 score. — metric_r2_score","text":"Formula:   also called coefficient determination. indicates close fitted regression line ground-truth data. highest score possible 1.0. indicates predictors perfectly accounts variation target. score 0.0 indicates predictors account variation target. can also negative model worse random. metric can also compute \"Adjusted R2\" score.","code":"sum_squares_residuals <- sum((y_true - y_pred) ** 2) sum_squares <- sum((y_true - mean(y_true)) ** 2) R2 <- 1 - sum_squares_residuals / sum_squares"},{"path":"https://keras3.posit.co/dev/reference/metric_r2_score.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes R2 score. — metric_r2_score","text":"","code":"metric_r2_score(   ...,   class_aggregation = \"uniform_average\",   num_regressors = 0L,   name = \"r2_score\",   dtype = NULL )"},{"path":"https://keras3.posit.co/dev/reference/metric_r2_score.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Computes R2 score. — metric_r2_score","text":"... forward/backward compatability. class_aggregation Specifies aggregate scores corresponding different output classes (target dimensions), .e. different dimensions last axis predictions. Equivalent multioutput argument Scikit-Learn. one NULL (aggregation), \"uniform_average\", \"variance_weighted_average\". num_regressors Number independent regressors used (\"Adjusted R2\" score). 0 standard R2 score. Defaults 0. name Optional. string name metric instance. dtype Optional. data type metric result.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_r2_score.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Computes R2 score. — metric_r2_score","text":"Metric instance returned. Metric instance can passed directly compile(metrics = ), used standalone object. See ?Metric example usage.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_r2_score.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Computes R2 score. — metric_r2_score","text":"","code":"y_true <- rbind(1, 4, 3) y_pred <- rbind(2, 4, 4) metric <- metric_r2_score() metric$update_state(y_true, y_pred) metric$result() ## tf.Tensor(0.57142854, shape=(), dtype=float32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/metric_recall.html","id":null,"dir":"Reference","previous_headings":"","what":"Computes the recall of the predictions with respect to the labels. — metric_recall","title":"Computes the recall of the predictions with respect to the labels. — metric_recall","text":"metric creates two local variables, true_positives false_negatives, used compute recall. value ultimately returned recall, idempotent operation simply divides true_positives sum true_positives false_negatives. sample_weight NULL, weights default 1. Use sample_weight 0 mask values. top_k set, recall computed often average class among labels batch entry top-k predictions. class_id specified, calculate recall considering entries batch class_id label, computing fraction class_id threshold /top-k predictions.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_recall.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes the recall of the predictions with respect to the labels. — metric_recall","text":"","code":"metric_recall(   ...,   thresholds = NULL,   top_k = NULL,   class_id = NULL,   name = NULL,   dtype = NULL )"},{"path":"https://keras3.posit.co/dev/reference/metric_recall.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Computes the recall of the predictions with respect to the labels. — metric_recall","text":"... forward/backward compatability. thresholds (Optional) float value, Python list float threshold values [0, 1]. threshold compared prediction values determine truth value predictions (.e., threshold TRUE, FALSE). used loss function sets from_logits=TRUE (.e. sigmoid applied predictions), thresholds set 0. One metric value generated threshold value. neither thresholds top_k set, default calculate recall thresholds=0.5. top_k (Optional) Unset default. int value specifying top-k predictions consider calculating recall. class_id (Optional) Integer class ID want binary metrics. must half-open interval [0, num_classes), num_classes last dimension predictions. name (Optional) string name metric instance. dtype (Optional) data type metric result.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_recall.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Computes the recall of the predictions with respect to the labels. — metric_recall","text":"Metric instance returned. Metric instance can passed directly compile(metrics = ), used standalone object. See ?Metric example usage.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_recall.html","id":"usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes the recall of the predictions with respect to the labels. — metric_recall","text":"Standalone usage:         Usage compile() API:   Usage loss from_logits=TRUE:","code":"m <- metric_recall() m$update_state(c(0, 1, 1, 1),                c(1, 0, 1, 1)) m$result() ## tf.Tensor(0.6666667, shape=(), dtype=float32) m$reset_state() m$update_state(c(0, 1, 1, 1),                c(1, 0, 1, 1),                sample_weight = c(0, 0, 1, 0)) m$result() ## tf.Tensor(1.0, shape=(), dtype=float32) model |> compile(   optimizer = 'sgd',   loss = 'binary_crossentropy',   metrics = list(metric_recall()) ) model |> compile(   optimizer = 'adam',   loss = loss_binary_crossentropy(from_logits = TRUE),   metrics = list(metric_recall(thresholds = 0)) )"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/metric_recall_at_precision.html","id":null,"dir":"Reference","previous_headings":"","what":"Computes best recall where precision is >= specified value. — metric_recall_at_precision","title":"Computes best recall where precision is >= specified value. — metric_recall_at_precision","text":"given score-label-distribution required precision might achievable, case 0.0 returned recall. metric creates four local variables, true_positives, true_negatives, false_positives false_negatives used compute recall given precision. threshold given precision value computed used evaluate corresponding recall. sample_weight NULL, weights default 1. Use sample_weight 0 mask values. class_id specified, calculate precision considering entries batch class_id threshold predictions, computing fraction class_id indeed correct label.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_recall_at_precision.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes best recall where precision is >= specified value. — metric_recall_at_precision","text":"","code":"metric_recall_at_precision(   ...,   precision,   num_thresholds = 200L,   class_id = NULL,   name = NULL,   dtype = NULL )"},{"path":"https://keras3.posit.co/dev/reference/metric_recall_at_precision.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Computes best recall where precision is >= specified value. — metric_recall_at_precision","text":"... forward/backward compatability. precision scalar value range [0, 1]. num_thresholds (Optional) Defaults 200. number thresholds use matching given precision. class_id (Optional) Integer class ID want binary metrics. must half-open interval [0, num_classes), num_classes last dimension predictions. name (Optional) string name metric instance. dtype (Optional) data type metric result.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_recall_at_precision.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Computes best recall where precision is >= specified value. — metric_recall_at_precision","text":"Metric instance returned. Metric instance can passed directly compile(metrics = ), used standalone object. See ?Metric example usage.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_recall_at_precision.html","id":"usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes best recall where precision is >= specified value. — metric_recall_at_precision","text":"Standalone usage:         Usage compile() API:","code":"m <- metric_recall_at_precision(precision = 0.8) m$update_state(c(0,   0,   1,   1),                c(0, 0.5, 0.3, 0.9)) m$result() ## tf.Tensor(0.5, shape=(), dtype=float32) m$reset_state() m$update_state(c(0,   0,   1,   1),                c(0, 0.5, 0.3, 0.9),                sample_weight = c(1, 0, 0, 1)) m$result() ## tf.Tensor(1.0, shape=(), dtype=float32) model |> compile(   optimizer = 'sgd',   loss = 'binary_crossentropy',   metrics = list(metric_recall_at_precision(precision = 0.8)) )"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/metric_root_mean_squared_error.html","id":null,"dir":"Reference","previous_headings":"","what":"Computes root mean squared error metric between y_true and y_pred. — metric_root_mean_squared_error","title":"Computes root mean squared error metric between y_true and y_pred. — metric_root_mean_squared_error","text":"Formula:","code":"loss <- sqrt(mean((y_pred - y_true) ^ 2))"},{"path":"https://keras3.posit.co/dev/reference/metric_root_mean_squared_error.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes root mean squared error metric between y_true and y_pred. — metric_root_mean_squared_error","text":"","code":"metric_root_mean_squared_error(   ...,   name = \"root_mean_squared_error\",   dtype = NULL )"},{"path":"https://keras3.posit.co/dev/reference/metric_root_mean_squared_error.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Computes root mean squared error metric between y_true and y_pred. — metric_root_mean_squared_error","text":"... forward/backward compatability. name (Optional) string name metric instance. dtype (Optional) data type metric result.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_root_mean_squared_error.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Computes root mean squared error metric between y_true and y_pred. — metric_root_mean_squared_error","text":"Metric instance returned. Metric instance can passed directly compile(metrics = ), used standalone object. See ?Metric example usage.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_root_mean_squared_error.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Computes root mean squared error metric between y_true and y_pred. — metric_root_mean_squared_error","text":"Standalone usage:         Usage compile() API:","code":"m <- metric_root_mean_squared_error() m$update_state(rbind(c(0, 1), c(0, 0)), rbind(c(1, 1), c(0, 0))) m$result() ## tf.Tensor(0.5, shape=(), dtype=float32) m$reset_state() m$update_state(rbind(c(0, 1), c(0, 0)), rbind(c(1, 1), c(0, 0)),                sample_weight = c(1, 0)) m$result() ## tf.Tensor(0.70710677, shape=(), dtype=float32) model %>% compile(   optimizer = 'sgd',   loss = 'mse',   metrics = list(metric_root_mean_squared_error()))"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/metric_sensitivity_at_specificity.html","id":null,"dir":"Reference","previous_headings":"","what":"Computes best sensitivity where specificity is >= specified value. — metric_sensitivity_at_specificity","title":"Computes best sensitivity where specificity is >= specified value. — metric_sensitivity_at_specificity","text":"Sensitivity measures proportion actual positives correctly identified (tp / (tp + fn)). Specificity measures proportion actual negatives correctly identified (tn / (tn + fp)). metric creates four local variables, true_positives, true_negatives, false_positives false_negatives used compute sensitivity given specificity. threshold given specificity value computed used evaluate corresponding sensitivity. sample_weight NULL, weights default 1. Use sample_weight 0 mask values. class_id specified, calculate precision considering entries batch class_id threshold predictions, computing fraction class_id indeed correct label. additional information specificity sensitivity, see following.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_sensitivity_at_specificity.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes best sensitivity where specificity is >= specified value. — metric_sensitivity_at_specificity","text":"","code":"metric_sensitivity_at_specificity(   ...,   specificity,   num_thresholds = 200L,   class_id = NULL,   name = NULL,   dtype = NULL )"},{"path":"https://keras3.posit.co/dev/reference/metric_sensitivity_at_specificity.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Computes best sensitivity where specificity is >= specified value. — metric_sensitivity_at_specificity","text":"... forward/backward compatability. specificity scalar value range [0, 1]. num_thresholds (Optional) Defaults 200. number thresholds use matching given specificity. class_id (Optional) Integer class ID want binary metrics. must half-open interval [0, num_classes), num_classes last dimension predictions. name (Optional) string name metric instance. dtype (Optional) data type metric result.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_sensitivity_at_specificity.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Computes best sensitivity where specificity is >= specified value. — metric_sensitivity_at_specificity","text":"Metric instance returned. Metric instance can passed directly compile(metrics = ), used standalone object. See ?Metric example usage.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_sensitivity_at_specificity.html","id":"usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes best sensitivity where specificity is >= specified value. — metric_sensitivity_at_specificity","text":"Standalone usage:         Usage compile() API:","code":"m <- metric_sensitivity_at_specificity(specificity = 0.5) m$update_state(c(0,   0,   0,   1,   1),                c(0, 0.3, 0.8, 0.3, 0.8)) m$result() ## tf.Tensor(0.5, shape=(), dtype=float32) m$reset_state() m$update_state(c(0,   0,   0,   1,   1),                c(0, 0.3, 0.8, 0.3, 0.8),                sample_weight = c(1, 1, 2, 2, 1)) m$result() ## tf.Tensor(0.33333334, shape=(), dtype=float32) model |> compile(   optimizer = 'sgd',   loss = 'binary_crossentropy',   metrics = list(metric_sensitivity_at_specificity()) )"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/metric_sparse_categorical_accuracy.html","id":null,"dir":"Reference","previous_headings":"","what":"Calculates how often predictions match integer labels. — metric_sparse_categorical_accuracy","title":"Calculates how often predictions match integer labels. — metric_sparse_categorical_accuracy","text":"can provide logits classes y_pred, since argmax logits probabilities . metric creates two local variables, total count used compute frequency y_pred matches y_true. frequency ultimately returned sparse categorical accuracy: idempotent operation simply divides total count. sample_weight NULL, weights default 1. Use sample_weight 0 mask values.","code":"acc <- sample_weight %*% (y_true == which.max(y_pred))"},{"path":"https://keras3.posit.co/dev/reference/metric_sparse_categorical_accuracy.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Calculates how often predictions match integer labels. — metric_sparse_categorical_accuracy","text":"","code":"metric_sparse_categorical_accuracy(   y_true,   y_pred,   ...,   name = \"sparse_categorical_accuracy\",   dtype = NULL )"},{"path":"https://keras3.posit.co/dev/reference/metric_sparse_categorical_accuracy.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Calculates how often predictions match integer labels. — metric_sparse_categorical_accuracy","text":"y_true Tensor true targets. y_pred Tensor predicted targets. ... forward/backward compatability. name (Optional) string name metric instance. dtype (Optional) data type metric result.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_sparse_categorical_accuracy.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Calculates how often predictions match integer labels. — metric_sparse_categorical_accuracy","text":"y_true y_pred missing, Metric instance returned. Metric instance can passed directly compile(metrics = ), used standalone object. See ?Metric example usage. y_true y_pred provided, tensor computed value returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_sparse_categorical_accuracy.html","id":"usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Calculates how often predictions match integer labels. — metric_sparse_categorical_accuracy","text":"Standalone usage:         Usage compile() API:","code":"m <- metric_sparse_categorical_accuracy() m$update_state(rbind(2L, 1L), rbind(c(0.1, 0.6, 0.3), c(0.05, 0.95, 0))) m$result() ## tf.Tensor(0.5, shape=(), dtype=float32) m$reset_state() m$update_state(rbind(2L, 1L), rbind(c(0.1, 0.6, 0.3), c(0.05, 0.95, 0)),                sample_weight = c(0.7, 0.3)) m$result() ## tf.Tensor(0.3, shape=(), dtype=float32) model |> compile(optimizer = 'sgd',                  loss = 'sparse_categorical_crossentropy',                  metrics = list(metric_sparse_categorical_accuracy()))"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/metric_sparse_categorical_crossentropy.html","id":null,"dir":"Reference","previous_headings":"","what":"Computes the crossentropy metric between the labels and predictions. — metric_sparse_categorical_crossentropy","title":"Computes the crossentropy metric between the labels and predictions. — metric_sparse_categorical_crossentropy","text":"Use crossentropy metric two label classes. expects labels provided integers. want provide labels one-hot encoded, please use metric_categorical_crossentropy() metric instead. num_classes floating point values per feature y_pred single floating point value per feature y_true.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_sparse_categorical_crossentropy.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes the crossentropy metric between the labels and predictions. — metric_sparse_categorical_crossentropy","text":"","code":"metric_sparse_categorical_crossentropy(   y_true,   y_pred,   from_logits = FALSE,   ignore_class = NULL,   axis = -1L,   ...,   name = \"sparse_categorical_crossentropy\",   dtype = NULL )"},{"path":"https://keras3.posit.co/dev/reference/metric_sparse_categorical_crossentropy.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Computes the crossentropy metric between the labels and predictions. — metric_sparse_categorical_crossentropy","text":"y_true Ground truth values. y_pred predicted values. from_logits (Optional) Whether output expected logits tensor. default, consider output encodes probability distribution. ignore_class Optional integer. ID class ignored loss computation. useful, example, segmentation problems featuring \"void\" class (commonly -1 255) segmentation maps. default (ignore_class=NULL), classes considered. axis (Optional) Defaults -1. dimension along entropy computed. ... forward/backward compatability. name (Optional) string name metric instance. dtype (Optional) data type metric result.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_sparse_categorical_crossentropy.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Computes the crossentropy metric between the labels and predictions. — metric_sparse_categorical_crossentropy","text":"y_true y_pred missing, Metric instance returned. Metric instance can passed directly compile(metrics = ), used standalone object. See ?Metric example usage. y_true y_pred provided, tensor computed value returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_sparse_categorical_crossentropy.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Computes the crossentropy metric between the labels and predictions. — metric_sparse_categorical_crossentropy","text":"Standalone usage:         Usage compile() API:","code":"m <- metric_sparse_categorical_crossentropy() m$update_state(array(c(1, 2)),                rbind(c(0.05, 0.95, 0), c(0.1, 0.8, 0.1))) m$result() ## tf.Tensor(1.1769392, shape=(), dtype=float32) m$reset_state() m$update_state(array(c(1, 2)),                rbind(c(0.05, 0.95, 0), c(0.1, 0.8, 0.1)),                sample_weight = c(0.3, 0.7)) m$result() ## tf.Tensor(1.6271976, shape=(), dtype=float32) model |> compile(   optimizer = 'sgd',   loss = 'mse',   metrics = list(metric_sparse_categorical_crossentropy()) )"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/metric_sparse_top_k_categorical_accuracy.html","id":null,"dir":"Reference","previous_headings":"","what":"Computes how often integer targets are in the top K predictions. — metric_sparse_top_k_categorical_accuracy","title":"Computes how often integer targets are in the top K predictions. — metric_sparse_top_k_categorical_accuracy","text":"Computes often integer targets top K predictions. default, arguments expected update_state() : y_true: tensor shape (batch_size) representing indices true categories. y_pred: tensor shape (batch_size, num_categories) containing scores sample possible categories. from_sorted_ids=TRUE, arguments expected update_state : y_true: tensor shape (batch_size) representing indices IDs true categories. y_pred: tensor shape (batch_size, N) containing indices IDs top N categories sorted order highest score lowest score. N must greater equal k. from_sorted_ids=TRUE option can efficient set categories large model optimized way retrieve top ones either without scoring without maintaining scores possible categories.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_sparse_top_k_categorical_accuracy.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes how often integer targets are in the top K predictions. — metric_sparse_top_k_categorical_accuracy","text":"","code":"metric_sparse_top_k_categorical_accuracy(   y_true,   y_pred,   k = 5L,   ...,   name = \"sparse_top_k_categorical_accuracy\",   dtype = NULL,   from_sorted_ids = FALSE )"},{"path":"https://keras3.posit.co/dev/reference/metric_sparse_top_k_categorical_accuracy.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Computes how often integer targets are in the top K predictions. — metric_sparse_top_k_categorical_accuracy","text":"y_true Tensor true targets. y_pred Tensor predicted targets. k (Optional) Number top elements look computing accuracy. Defaults 5. ... forward/backward compatability. name (Optional) string name metric instance. dtype (Optional) data type metric result. from_sorted_ids (Optional) FALSE, default, tensor passed y_pred contains unsorted scores possible categories. TRUE, y_pred contains indices IDs top categories.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_sparse_top_k_categorical_accuracy.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Computes how often integer targets are in the top K predictions. — metric_sparse_top_k_categorical_accuracy","text":"y_true y_pred missing, Metric instance returned. Metric instance can passed directly compile(metrics = ), used standalone object. See ?Metric example usage. y_true y_pred provided, tensor computed value returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_sparse_top_k_categorical_accuracy.html","id":"usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes how often integer targets are in the top K predictions. — metric_sparse_top_k_categorical_accuracy","text":"Standalone usage:             Usage compile() API:","code":"m <- metric_sparse_top_k_categorical_accuracy(k = 1L) m$update_state(   rbind(2, 1),   op_array(rbind(c(0.1, 0.9, 0.8), c(0.05, 0.95, 0)), dtype = \"float32\") ) m$result() ## tf.Tensor(0.5, shape=(), dtype=float32) m$reset_state() m$update_state(   rbind(2, 1),   op_array(rbind(c(0.1, 0.9, 0.8), c(0.05, 0.95, 0)), dtype = \"float32\"),   sample_weight = c(0.7, 0.3) ) m$result() ## tf.Tensor(0.3, shape=(), dtype=float32) m <- metric_sparse_top_k_categorical_accuracy(k = 1, from_sorted_ids = TRUE) m$update_state(array(c(2, 1)), rbind(c(1, 0, 3),                                      c(1, 2, 3))) m$result() ## tf.Tensor(0.5, shape=(), dtype=float32) model %>% compile(optimizer = 'sgd',                   loss = 'sparse_categorical_crossentropy',                   metrics = list(metric_sparse_top_k_categorical_accuracy()))"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/metric_specificity_at_sensitivity.html","id":null,"dir":"Reference","previous_headings":"","what":"Computes best specificity where sensitivity is >= specified value. — metric_specificity_at_sensitivity","title":"Computes best specificity where sensitivity is >= specified value. — metric_specificity_at_sensitivity","text":"Sensitivity measures proportion actual positives correctly identified (tp / (tp + fn)). Specificity measures proportion actual negatives correctly identified (tn / (tn + fp)). metric creates four local variables, true_positives, true_negatives, false_positives false_negatives used compute specificity given sensitivity. threshold given sensitivity value computed used evaluate corresponding specificity. sample_weight NULL, weights default 1. Use sample_weight 0 mask values. class_id specified, calculate precision considering entries batch class_id threshold predictions, computing fraction class_id indeed correct label. additional information specificity sensitivity, see following.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_specificity_at_sensitivity.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes best specificity where sensitivity is >= specified value. — metric_specificity_at_sensitivity","text":"","code":"metric_specificity_at_sensitivity(   ...,   sensitivity,   num_thresholds = 200L,   class_id = NULL,   name = NULL,   dtype = NULL )"},{"path":"https://keras3.posit.co/dev/reference/metric_specificity_at_sensitivity.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Computes best specificity where sensitivity is >= specified value. — metric_specificity_at_sensitivity","text":"... forward/backward compatability. sensitivity scalar value range [0, 1]. num_thresholds (Optional) Defaults 200. number thresholds use matching given sensitivity. class_id (Optional) Integer class ID want binary metrics. must half-open interval [0, num_classes), num_classes last dimension predictions. name (Optional) string name metric instance. dtype (Optional) data type metric result.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_specificity_at_sensitivity.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Computes best specificity where sensitivity is >= specified value. — metric_specificity_at_sensitivity","text":"Metric instance returned. Metric instance can passed directly compile(metrics = ), used standalone object. See ?Metric example usage.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_specificity_at_sensitivity.html","id":"usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes best specificity where sensitivity is >= specified value. — metric_specificity_at_sensitivity","text":"Standalone usage:         Usage compile() API:","code":"m <- metric_specificity_at_sensitivity(sensitivity = 0.5) m$update_state(c(0,   0,   0,   1,   1),                c(0, 0.3, 0.8, 0.3, 0.8)) m$result() ## tf.Tensor(0.6666667, shape=(), dtype=float32) m$reset_state() m$update_state(c(0,   0,   0,   1,   1),                c(0, 0.3, 0.8, 0.3, 0.8),                sample_weight = c(1, 1, 2, 2, 2)) m$result() ## tf.Tensor(0.5, shape=(), dtype=float32) model |> compile(   optimizer = 'sgd',   loss = 'binary_crossentropy',   metrics = list(metric_sensitivity_at_specificity()) )"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/metric_squared_hinge.html","id":null,"dir":"Reference","previous_headings":"","what":"Computes the hinge metric between y_true and y_pred. — metric_squared_hinge","title":"Computes the hinge metric between y_true and y_pred. — metric_squared_hinge","text":"Formula:   y_true values expected -1 1. binary (0 1) labels provided convert -1 1.","code":"loss <- mean(square(maximum(1 - y_true * y_pred, 0)))"},{"path":"https://keras3.posit.co/dev/reference/metric_squared_hinge.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes the hinge metric between y_true and y_pred. — metric_squared_hinge","text":"","code":"metric_squared_hinge(y_true, y_pred, ..., name = \"squared_hinge\", dtype = NULL)"},{"path":"https://keras3.posit.co/dev/reference/metric_squared_hinge.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Computes the hinge metric between y_true and y_pred. — metric_squared_hinge","text":"y_true ground truth values. y_true values expected -1 1. binary (0 1) labels provided convert -1 1 shape = [batch_size, d0, .. dN]. y_pred predicted values shape = [batch_size, d0, .. dN]. ... forward/backward compatability. name (Optional) string name metric instance. dtype (Optional) data type metric result.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_squared_hinge.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Computes the hinge metric between y_true and y_pred. — metric_squared_hinge","text":"y_true y_pred missing, Metric instance returned. Metric instance can passed directly compile(metrics = ), used standalone object. See ?Metric example usage. y_true y_pred provided, tensor computed value returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_squared_hinge.html","id":"usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes the hinge metric between y_true and y_pred. — metric_squared_hinge","text":"Standalone usage:","code":"m <- metric_squared_hinge() m$update_state(rbind(c(0, 1), c(0, 0)), rbind(c(0.6, 0.4), c(0.4, 0.6))) m$result() ## tf.Tensor(1.86, shape=(), dtype=float32) m$reset_state() m$update_state(rbind(c(0, 1), c(0, 0)), rbind(c(0.6, 0.4), c(0.4, 0.6)),                sample_weight = c(1, 0)) m$result() ## tf.Tensor(1.46, shape=(), dtype=float32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/metric_sum.html","id":null,"dir":"Reference","previous_headings":"","what":"Compute the (weighted) sum of the given values. — metric_sum","title":"Compute the (weighted) sum of the given values. — metric_sum","text":"example, values [1, 3, 5, 7] sum 16. sample_weight specified [1, 1, 0, 0] sum 4. metric creates one variable, total. ultimately returned sum value.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_sum.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Compute the (weighted) sum of the given values. — metric_sum","text":"","code":"metric_sum(..., name = \"sum\", dtype = NULL)"},{"path":"https://keras3.posit.co/dev/reference/metric_sum.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Compute the (weighted) sum of the given values. — metric_sum","text":"... forward/backward compatability. name (Optional) string name metric instance. dtype (Optional) data type metric result.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_sum.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Compute the (weighted) sum of the given values. — metric_sum","text":"Metric instance returned. Metric instance can passed directly compile(metrics = ), used standalone object. See ?Metric example usage.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_sum.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Compute the (weighted) sum of the given values. — metric_sum","text":"","code":"m <- metric_sum() m$update_state(c(1, 3, 5, 7)) m$result() ## tf.Tensor(16.0, shape=(), dtype=float32) m <- metric_sum() m$update_state(c(1, 3, 5, 7), sample_weight = c(1, 1, 0, 0)) m$result() ## tf.Tensor(4.0, shape=(), dtype=float32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/metric_top_k_categorical_accuracy.html","id":null,"dir":"Reference","previous_headings":"","what":"Computes how often targets are in the top K predictions. — metric_top_k_categorical_accuracy","title":"Computes how often targets are in the top K predictions. — metric_top_k_categorical_accuracy","text":"Computes often targets top K predictions.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_top_k_categorical_accuracy.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes how often targets are in the top K predictions. — metric_top_k_categorical_accuracy","text":"","code":"metric_top_k_categorical_accuracy(   y_true,   y_pred,   k = 5L,   ...,   name = \"top_k_categorical_accuracy\",   dtype = NULL )"},{"path":"https://keras3.posit.co/dev/reference/metric_top_k_categorical_accuracy.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Computes how often targets are in the top K predictions. — metric_top_k_categorical_accuracy","text":"y_true Tensor true targets. y_pred Tensor predicted targets. k (Optional) Number top elements look computing accuracy. Defaults 5. ... forward/backward compatability. name (Optional) string name metric instance. dtype (Optional) data type metric result.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_top_k_categorical_accuracy.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Computes how often targets are in the top K predictions. — metric_top_k_categorical_accuracy","text":"y_true y_pred missing, Metric instance returned. Metric instance can passed directly compile(metrics = ), used standalone object. See ?Metric example usage. y_true y_pred provided, tensor computed value returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_top_k_categorical_accuracy.html","id":"usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes how often targets are in the top K predictions. — metric_top_k_categorical_accuracy","text":"Standalone usage:         Usage compile() API:","code":"m <- metric_top_k_categorical_accuracy(k = 1) m$update_state(   rbind(c(0, 0, 1), c(0, 1, 0)),   op_array(rbind(c(0.1, 0.9, 0.8), c(0.05, 0.95, 0)), dtype = \"float32\") ) m$result() ## tf.Tensor(0.5, shape=(), dtype=float32) m$reset_state() m$update_state(   rbind(c(0, 0, 1), c(0, 1, 0)),   op_array(rbind(c(0.1, 0.9, 0.8), c(0.05, 0.95, 0)), dtype = \"float32\"),   sample_weight = c(0.7, 0.3)) m$result() ## tf.Tensor(0.3, shape=(), dtype=float32) model |> compile(optimizer = 'sgd',                  loss = 'categorical_crossentropy',                  metrics = list(metric_top_k_categorical_accuracy()))"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/metric_true_negatives.html","id":null,"dir":"Reference","previous_headings":"","what":"Calculates the number of true negatives. — metric_true_negatives","title":"Calculates the number of true negatives. — metric_true_negatives","text":"sample_weight given, calculates sum weights true negatives. metric creates one local variable, accumulator used keep track number true negatives. sample_weight NULL, weights default 1. Use sample_weight 0 mask values.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_true_negatives.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Calculates the number of true negatives. — metric_true_negatives","text":"","code":"metric_true_negatives(..., thresholds = NULL, name = NULL, dtype = NULL)"},{"path":"https://keras3.posit.co/dev/reference/metric_true_negatives.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Calculates the number of true negatives. — metric_true_negatives","text":"... forward/backward compatability. thresholds (Optional) Defaults 0.5. float value, Python list float threshold values [0, 1]. threshold compared prediction values determine truth value predictions (.e., threshold TRUE, FALSE). used loss function sets from_logits=TRUE (.e. sigmoid applied predictions), thresholds set 0. One metric value generated threshold value. name (Optional) string name metric instance. dtype (Optional) data type metric result.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_true_negatives.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Calculates the number of true negatives. — metric_true_negatives","text":"Metric instance returned. Metric instance can passed directly compile(metrics = ), used standalone object. See ?Metric example usage.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_true_negatives.html","id":"usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Calculates the number of true negatives. — metric_true_negatives","text":"Standalone usage:","code":"m <- metric_true_negatives() m$update_state(c(0, 1, 0, 0), c(1, 1, 0, 0)) m$result() ## tf.Tensor(2.0, shape=(), dtype=float32) m$reset_state() m$update_state(c(0, 1, 0, 0), c(1, 1, 0, 0), sample_weight = c(0, 0, 1, 0)) m$result() ## tf.Tensor(1.0, shape=(), dtype=float32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/metric_true_positives.html","id":null,"dir":"Reference","previous_headings":"","what":"Calculates the number of true positives. — metric_true_positives","title":"Calculates the number of true positives. — metric_true_positives","text":"sample_weight given, calculates sum weights true positives. metric creates one local variable, true_positives used keep track number true positives. sample_weight NULL, weights default 1. Use sample_weight 0 mask values.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_true_positives.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Calculates the number of true positives. — metric_true_positives","text":"","code":"metric_true_positives(..., thresholds = NULL, name = NULL, dtype = NULL)"},{"path":"https://keras3.posit.co/dev/reference/metric_true_positives.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Calculates the number of true positives. — metric_true_positives","text":"... forward/backward compatability. thresholds (Optional) Defaults 0.5. float value, Python list float threshold values [0, 1]. threshold compared prediction values determine truth value predictions (.e., threshold TRUE, FALSE). used loss function sets from_logits=TRUE (.e. sigmoid applied predictions), thresholds set 0. One metric value generated threshold value. name (Optional) string name metric instance. dtype (Optional) data type metric result.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_true_positives.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Calculates the number of true positives. — metric_true_positives","text":"Metric instance returned. Metric instance can passed directly compile(metrics = ), used standalone object. See ?Metric example usage.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_true_positives.html","id":"usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Calculates the number of true positives. — metric_true_positives","text":"Standalone usage:","code":"m <- metric_true_positives() m$update_state(c(0, 1, 1, 1), c(1, 0, 1, 1)) m$result() ## tf.Tensor(2.0, shape=(), dtype=float32) m$reset_state() m$update_state(c(0, 1, 1, 1), c(1, 0, 1, 1), sample_weight = c(0, 0, 1, 0)) m$result() ## tf.Tensor(1.0, shape=(), dtype=float32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/multi-assign.html","id":null,"dir":"Reference","previous_headings":"","what":"Assign values to names — %<-%","title":"Assign values to names — %<-%","text":"See %<-% details.","code":""},{"path":"https://keras3.posit.co/dev/reference/multi-assign.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Assign values to names — %<-%","text":"","code":"x %<-% value"},{"path":"https://keras3.posit.co/dev/reference/multi-assign.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Assign values to names — %<-%","text":"right-hand-side argument, value, invisibly. called primarily side-effect assigning symbols current frame.","code":""},{"path":"https://keras3.posit.co/dev/reference/new_callback_class.html","id":null,"dir":"Reference","previous_headings":"","what":"Callback — new_callback_class","title":"Callback — new_callback_class","text":"new_callback_class() alias Callback(). See ?Callback() full documentation.","code":""},{"path":"https://keras3.posit.co/dev/reference/new_callback_class.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Callback — new_callback_class","text":"","code":"new_callback_class(   classname,   on_epoch_begin = NULL,   on_epoch_end = NULL,   on_train_begin = NULL,   on_train_end = NULL,   on_train_batch_begin = NULL,   on_train_batch_end = NULL,   on_test_begin = NULL,   on_test_end = NULL,   on_test_batch_begin = NULL,   on_test_batch_end = NULL,   on_predict_begin = NULL,   on_predict_end = NULL,   on_predict_batch_begin = NULL,   on_predict_batch_end = NULL,   ...,   public = list(),   private = list(),   inherit = NULL,   parent_env = parent.frame() )"},{"path":"https://keras3.posit.co/dev/reference/new_callback_class.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Callback — new_callback_class","text":"classname String, name custom class. (Conventionally, CamelCase). on_epoch_begin   Called start epoch. Subclasses override actions run. function called TRAIN mode. Args: epoch: Integer, index epoch. logs: Named List. Currently data passed argument method may change future. on_epoch_end   Called end epoch. Subclasses override actions run. function called TRAIN mode. Args: epoch: Integer, index epoch. logs: Named List, metric results training epoch, validation epoch validation performed. Validation result keys prefixed val_. training epoch, values Model's metrics returned. Example: list(loss = 0.2, accuracy = 0.7). on_train_begin   Called beginning training. Subclasses override actions run. Args: logs: Named list. Currently data passed argument method may change future. on_train_end   Called end training. Subclasses override actions run. Args: logs: Named list. Currently output last call on_epoch_end() passed argument method may change future. on_train_batch_begin   Called beginning training batch fit() methods. Subclasses override actions run. Note steps_per_execution argument compile Model set N, method called every N batches. Args: batch: Integer, index batch within current epoch. logs: Named list. Currently data passed argument method may change future. on_train_batch_end   Called end training batch fit() methods. Subclasses override actions run. Note steps_per_execution argument compile Model set N, method called every N batches. Args: batch: Integer, index batch within current epoch. logs: Named list. Aggregated metric results batch. on_test_begin   Called beginning evaluation validation. Subclasses override actions run. Args: logs: Named list. Currently data passed argument method may change future. on_test_end   Called end evaluation validation. Subclasses override actions run. Args: logs: Named list. Currently output last call on_test_batch_end() passed argument method may change future. on_test_batch_begin   Called beginning batch evaluate() methods. Also called beginning validation batch fit() methods, validation data provided. Subclasses override actions run. Note steps_per_execution argument compile() Model set N, method called every N batches. Args: batch: Integer, index batch within current epoch. logs: Named list. Currently data passed argument method may change future. on_test_batch_end   Called end batch evaluate() methods. Also called end validation batch fit() methods, validation data provided. Subclasses override actions run. Note steps_per_execution argument compile() Model set N, method called every N batches. Args: batch: Integer, index batch within current epoch. logs: Named list. Aggregated metric results batch. on_predict_begin   Called beginning prediction. Subclasses override actions run. Args: logs: Named list. Currently data passed argument method may change future. on_predict_end   Called end prediction. Subclasses override actions run. Args: logs: Named list. Currently data passed argument method may change future. on_predict_batch_begin   Called beginning batch predict() methods. Subclasses override actions run. Note steps_per_execution argument compile() Model set N, method called every N batches. Args: batch: Integer, index batch within current epoch. logs: Named list. Currently data passed argument method may change future. on_predict_batch_end   Called end batch predict() methods. Subclasses override actions run. Note steps_per_execution argument compile Model set N, method called every N batches. Args: batch: Integer, index batch within current epoch. logs: Named list. Aggregated metric results batch. ..., public Additional methods public members custom class. private Named list R objects (typically, functions) include instance private environments. private methods symbols scope public methods (See section \"Symbols Scope\"). instance private environment. objects private invisible Keras framework Python runtime. inherit custom class subclass. default, base keras class. parent_env R environment class methods grandparent.","code":"\\(epoch, logs = NULL) \\(epoch, logs = NULL) \\(logs = NULL) \\(logs = NULL) \\(batch, logs = NULL) \\(batch, logs=NULL) \\(logs = NULL) \\(logs = NULL) \\(batch, logs = NULL) \\(batch, logs = NULL) \\(logs = NULL) \\(logs = NULL) \\(batch, logs = NULL) \\(batch, logs = NULL)"},{"path":"https://keras3.posit.co/dev/reference/new_callback_class.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Callback — new_callback_class","text":"function returns custom Callback instances, similar builtin callback functions.","code":""},{"path":"https://keras3.posit.co/dev/reference/new_layer_class.html","id":null,"dir":"Reference","previous_headings":"","what":"Layer — new_layer_class","title":"Layer — new_layer_class","text":"new_layer_class() alias Layer(). See ?Layer() full documentation.","code":""},{"path":"https://keras3.posit.co/dev/reference/new_layer_class.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Layer — new_layer_class","text":"","code":"new_layer_class(   classname,   initialize = NULL,   call = NULL,   build = NULL,   get_config = NULL,   ...,   public = list(),   private = list(),   inherit = NULL,   parent_env = parent.frame() )"},{"path":"https://keras3.posit.co/dev/reference/new_layer_class.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Layer — new_layer_class","text":"classname String, name custom class. (Conventionally, CamelCase). initialize, call, build, get_config Recommended methods implement. See description details sections. ..., public Additional methods public members custom class. private Named list R objects (typically, functions) include instance private environments. private methods symbols scope public methods (See section \"Symbols Scope\"). instance private environment. objects private invisible Keras framework Python runtime. inherit custom class subclass. default, base keras class. parent_env R environment class methods grandparent.","code":""},{"path":"https://keras3.posit.co/dev/reference/new_layer_class.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Layer — new_layer_class","text":"composing layer constructor, similar behavior layer functions like layer_dense(). first argument returned function object, enabling initialize()ing call() layer one step composing layer pipe, like   initialize() layer instance call() , pass missing NULL value object, pass arguments initialize() name.","code":"layer_foo <- Layer(\"Foo\", ....) output <- inputs |> layer_foo() layer <- layer_dense(units = 2, activation = \"relu\") layer <- layer_dense(NULL, 2, activation = \"relu\") layer <- layer_dense(, 2, activation = \"relu\")  # then you can call() the layer in a separate step outputs <- inputs |> layer()"},{"path":"https://keras3.posit.co/dev/reference/new_learning_rate_schedule_class.html","id":null,"dir":"Reference","previous_headings":"","what":"LearningRateSchedule — new_learning_rate_schedule_class","title":"LearningRateSchedule — new_learning_rate_schedule_class","text":"new_learning_rate_schedule_class() alias LearningRateSchedule(). See ?LearningRateSchedule() full documentation.","code":""},{"path":"https://keras3.posit.co/dev/reference/new_learning_rate_schedule_class.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"LearningRateSchedule — new_learning_rate_schedule_class","text":"","code":"new_learning_rate_schedule_class(   classname,   call = NULL,   initialize = NULL,   get_config = NULL,   ...,   public = list(),   private = list(),   inherit = NULL,   parent_env = parent.frame() )"},{"path":"https://keras3.posit.co/dev/reference/new_learning_rate_schedule_class.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"LearningRateSchedule — new_learning_rate_schedule_class","text":"classname String, name custom class. (Conventionally, CamelCase). call, initialize, get_config Recommended methods implement. See description details sections. ..., public Additional methods public members custom class. private Named list R objects (typically, functions) include instance private environments. private methods symbols scope public methods (See section \"Symbols Scope\"). instance private environment. objects private invisible Keras framework Python runtime. inherit custom class subclass. default, base keras class. parent_env R environment class methods grandparent.","code":""},{"path":"https://keras3.posit.co/dev/reference/new_learning_rate_schedule_class.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"LearningRateSchedule — new_learning_rate_schedule_class","text":"function returns LearningRateSchedule instances, similar built-learning_rate_schedule_* family functions.","code":""},{"path":"https://keras3.posit.co/dev/reference/new_loss_class.html","id":null,"dir":"Reference","previous_headings":"","what":"Loss — new_loss_class","title":"Loss — new_loss_class","text":"new_loss_class() alias Loss(). See ?Loss() full documentation.","code":""},{"path":"https://keras3.posit.co/dev/reference/new_loss_class.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Loss — new_loss_class","text":"","code":"new_loss_class(   classname,   call = NULL,   ...,   public = list(),   private = list(),   inherit = NULL,   parent_env = parent.frame() )"},{"path":"https://keras3.posit.co/dev/reference/new_loss_class.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Loss — new_loss_class","text":"classname String, name custom class. (Conventionally, CamelCase). call   Method implemented subclasses: Function contains logic loss calculation using y_true, y_pred. ..., public Additional methods public members custom class. private Named list R objects (typically, functions) include instance private environments. private methods symbols scope public methods (See section \"Symbols Scope\"). instance private environment. objects private invisible Keras framework Python runtime. inherit custom class subclass. default, base keras class. parent_env R environment class methods grandparent.","code":"function(y_true, y_pred)"},{"path":"https://keras3.posit.co/dev/reference/new_loss_class.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Loss — new_loss_class","text":"function returns Loss instances, similar builtin loss functions.","code":""},{"path":"https://keras3.posit.co/dev/reference/new_metric_class.html","id":null,"dir":"Reference","previous_headings":"","what":"Metric — new_metric_class","title":"Metric — new_metric_class","text":"new_metric_class() alias Metric(). See ?Metric() full documentation.","code":""},{"path":"https://keras3.posit.co/dev/reference/new_metric_class.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Metric — new_metric_class","text":"","code":"new_metric_class(   classname,   initialize = NULL,   update_state = NULL,   result = NULL,   ...,   public = list(),   private = list(),   inherit = NULL,   parent_env = parent.frame() )"},{"path":"https://keras3.posit.co/dev/reference/new_metric_class.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Metric — new_metric_class","text":"classname String, name custom class. (Conventionally, CamelCase). initialize, update_state, result Recommended methods implement. See description section. ..., public Additional methods public members custom class. private Named list R objects (typically, functions) include instance private environments. private methods symbols scope public methods (See section \"Symbols Scope\"). instance private environment. objects private invisible Keras framework Python runtime. inherit custom class subclass. default, base keras class. parent_env R environment class methods grandparent.","code":""},{"path":"https://keras3.posit.co/dev/reference/new_metric_class.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Metric — new_metric_class","text":"function returns Metric instances, similar builtin metric functions.","code":""},{"path":"https://keras3.posit.co/dev/reference/new_model_class.html","id":null,"dir":"Reference","previous_headings":"","what":"Model — new_model_class","title":"Model — new_model_class","text":"new_model_class() alias Model(). See ?Model() full documentation.","code":""},{"path":"https://keras3.posit.co/dev/reference/new_model_class.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Model — new_model_class","text":"","code":"new_model_class(   classname,   initialize = NULL,   call = NULL,   train_step = NULL,   predict_step = NULL,   test_step = NULL,   compute_loss = NULL,   compute_metrics = NULL,   ...,   public = list(),   private = list(),   inherit = NULL,   parent_env = parent.frame() )"},{"path":"https://keras3.posit.co/dev/reference/new_model_class.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Model — new_model_class","text":"classname String, name custom class. (Conventionally, CamelCase). initialize, call, train_step, predict_step, test_step, compute_loss, compute_metrics Optional methods can overridden. ..., public Additional methods public members custom class. private Named list R objects (typically, functions) include instance private environments. private methods symbols scope public methods (See section \"Symbols Scope\"). instance private environment. objects private invisible Keras framework Python runtime. inherit custom class subclass. default, base keras class. parent_env R environment class methods grandparent.","code":""},{"path":"https://keras3.posit.co/dev/reference/new_model_class.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Model — new_model_class","text":"model constructor function, can call create instance new model type.","code":""},{"path":"https://keras3.posit.co/dev/reference/newaxis.html","id":null,"dir":"Reference","previous_headings":"","what":"New axis — newaxis","title":"New axis — newaxis","text":"alias NULL. meant used [ tensors, expand dimensions tensor","code":""},{"path":"https://keras3.posit.co/dev/reference/newaxis.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"New axis — newaxis","text":"","code":"newaxis"},{"path":"https://keras3.posit.co/dev/reference/newaxis.html","id":"format","dir":"Reference","previous_headings":"","what":"Format","title":"New axis — newaxis","text":"object class NULL length 0.","code":""},{"path":"https://keras3.posit.co/dev/reference/newaxis.html","id":"details","dir":"Reference","previous_headings":"","what":"Details","title":"New axis — newaxis","text":"","code":"x <- op_convert_to_tensor(1:10)  op_shape(x) op_shape(x[]) op_shape(x[newaxis]) op_shape(x@py[newaxis]) op_shape(x@r[newaxis])  op_shape(x[newaxis, .., newaxis]) op_shape(x@py[newaxis, .., newaxis]) op_shape(x@r[newaxis, .., newaxis])"},{"path":"https://keras3.posit.co/dev/reference/normalize.html","id":null,"dir":"Reference","previous_headings":"","what":"Normalizes an array. — normalize","title":"Normalizes an array. — normalize","text":"input R array, R array returned. backend tensor, backend tensor returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/normalize.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Normalizes an array. — normalize","text":"","code":"normalize(x, axis = -1L, order = 2L)"},{"path":"https://keras3.posit.co/dev/reference/normalize.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Normalizes an array. — normalize","text":"x Array normalize. axis axis along normalize. order Normalization order (e.g. order=2 L2 norm).","code":""},{"path":"https://keras3.posit.co/dev/reference/normalize.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Normalizes an array. — normalize","text":"normalized copy array.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_abs.html","id":null,"dir":"Reference","previous_headings":"","what":"Compute the absolute value element-wise. — op_abs","title":"Compute the absolute value element-wise. — op_abs","text":"Compute absolute value element-wise.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_abs.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Compute the absolute value element-wise. — op_abs","text":"","code":"op_abs(x)"},{"path":"https://keras3.posit.co/dev/reference/op_abs.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Compute the absolute value element-wise. — op_abs","text":"x Input tensor","code":""},{"path":"https://keras3.posit.co/dev/reference/op_abs.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Compute the absolute value element-wise. — op_abs","text":"array containing absolute value element x.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_abs.html","id":"example","dir":"Reference","previous_headings":"","what":"Example","title":"Compute the absolute value element-wise. — op_abs","text":"","code":"x <- op_convert_to_tensor(c(-1.2, 1.2)) op_abs(x) ## tf.Tensor([1.2 1.2], shape=(2), dtype=float64)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_add.html","id":null,"dir":"Reference","previous_headings":"","what":"Add arguments element-wise. — op_add","title":"Add arguments element-wise. — op_add","text":"Add arguments element-wise.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_add.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Add arguments element-wise. — op_add","text":"","code":"op_add(x1, x2)"},{"path":"https://keras3.posit.co/dev/reference/op_add.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Add arguments element-wise. — op_add","text":"x1 First input tensor. x2 Second input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_add.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Add arguments element-wise. — op_add","text":"tensor containing element-wise sum x1 x2.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_add.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Add arguments element-wise. — op_add","text":"op_add also broadcasts shapes:     Note function automatically called using R operator + tensors.","code":"x1 <- op_convert_to_tensor(c(1, 4)) x2 <- op_convert_to_tensor(c(5, 6)) op_add(x1, x2) ## tf.Tensor([ 6. 10.], shape=(2), dtype=float64) # alias for x1 + x2 x1 + x2 ## tf.Tensor([ 6. 10.], shape=(2), dtype=float64) x1 <- op_convert_to_tensor(array(c(5, 5, 4, 6), dim =c(2, 2))) x2 <- op_convert_to_tensor(c(5, 6)) op_add(x1, x2) ## tf.Tensor( ## [[10. 10.] ##  [10. 12.]], shape=(2, 2), dtype=float64) x <- op_ones(c(3)) op_add(x, x) ## tf.Tensor([2. 2. 2.], shape=(3), dtype=float32) x + x ## tf.Tensor([2. 2. 2.], shape=(3), dtype=float32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_all.html","id":null,"dir":"Reference","previous_headings":"","what":"Test whether all array elements along a given axis evaluate to TRUE. — op_all","title":"Test whether all array elements along a given axis evaluate to TRUE. — op_all","text":"Test whether array elements along given axis evaluate TRUE.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_all.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Test whether all array elements along a given axis evaluate to TRUE. — op_all","text":"","code":"op_all(x, axis = NULL, keepdims = FALSE)"},{"path":"https://keras3.posit.co/dev/reference/op_all.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Test whether all array elements along a given axis evaluate to TRUE. — op_all","text":"x Input tensor. axis integer tuple integers represent axis along logical reduction performed. default (axis = NULL) perform logical dimensions input array. axis may negative, case counts last first axis. keepdims TRUE, axes reduced left result dimensions size one. option, result broadcast correctly input array. Defaults FALSE.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_all.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Test whether all array elements along a given axis evaluate to TRUE. — op_all","text":"tensor containing logical reduction axis.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_all.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Test whether all array elements along a given axis evaluate to TRUE. — op_all","text":"keepdims = TRUE outputs tensor dimensions reduced one.","code":"x <- op_convert_to_tensor(c(TRUE, FALSE)) op_all(x) ## tf.Tensor(False, shape=(), dtype=bool) (x <- op_convert_to_tensor(array(c(TRUE, FALSE, TRUE, TRUE, TRUE, TRUE), dim = c(3, 2)))) ## tf.Tensor( ## [[ True  True] ##  [False  True] ##  [ True  True]], shape=(3, 2), dtype=bool) op_all(x, axis = 1) ## tf.Tensor([False  True], shape=(2), dtype=bool) op_all(x, keepdims = TRUE) ## tf.Tensor([[False]], shape=(1, 1), dtype=bool)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_angle.html","id":null,"dir":"Reference","previous_headings":"","what":"Element-wise angle of a complex tensor. — op_angle","title":"Element-wise angle of a complex tensor. — op_angle","text":"Returns phase angle (radians) element x.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_angle.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Element-wise angle of a complex tensor. — op_angle","text":"","code":"op_angle(x)"},{"path":"https://keras3.posit.co/dev/reference/op_angle.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Element-wise angle of a complex tensor. — op_angle","text":"x Input tensor. Can real complex.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_angle.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Element-wise angle of a complex tensor. — op_angle","text":"Tensor shape x, containing element-wise angles.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_angle.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Element-wise angle of a complex tensor. — op_angle","text":"","code":"x <- op_convert_to_tensor(matrix(c(1 + 3i, 2 - 5i, 4 - 3i, 3 + 2i), nrow = 2)) op_angle(x) ## tf.Tensor( ## [[ 1.24904577 -0.64350111] ##  [-1.19028995  0.5880026 ]], shape=(2, 2), dtype=float64)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_any.html","id":null,"dir":"Reference","previous_headings":"","what":"Test whether any array element along a given axis evaluates to TRUE. — op_any","title":"Test whether any array element along a given axis evaluates to TRUE. — op_any","text":"Test whether array element along given axis evaluates TRUE.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_any.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Test whether any array element along a given axis evaluates to TRUE. — op_any","text":"","code":"op_any(x, axis = NULL, keepdims = FALSE)"},{"path":"https://keras3.posit.co/dev/reference/op_any.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Test whether any array element along a given axis evaluates to TRUE. — op_any","text":"x Input tensor. axis integer tuple integers represent axis along logical reduction performed. default (axis = NULL) perform logical dimensions input array. axis may negative, case counts last first axis. keepdims TRUE, axes reduced left result dimensions size one. option, result broadcast correctly input array. Defaults FALSE.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_any.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Test whether any array element along a given axis evaluates to TRUE. — op_any","text":"tensor containing logical reduction axis.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_any.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Test whether any array element along a given axis evaluates to TRUE. — op_any","text":"keepdims = TRUE outputs tensor dimensions reduced one.","code":"x <- op_array(c(TRUE, FALSE)) op_any(x) ## tf.Tensor(True, shape=(), dtype=bool) (x <- op_reshape(c(FALSE, FALSE, FALSE,                    TRUE, FALSE, FALSE),                  c(2, 3))) ## tf.Tensor( ## [[False False False] ##  [ True False False]], shape=(2, 3), dtype=bool) op_any(x, axis = 1) ## tf.Tensor([ True False False], shape=(3), dtype=bool) op_any(x, axis = 2) ## tf.Tensor([False  True], shape=(2), dtype=bool) op_any(x, axis = -1) ## tf.Tensor([False  True], shape=(2), dtype=bool) op_any(x, keepdims = TRUE) ## tf.Tensor([[ True]], shape=(1, 1), dtype=bool) op_any(x, 1, keepdims = TRUE) ## tf.Tensor([[ True False False]], shape=(1, 3), dtype=bool)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_append.html","id":null,"dir":"Reference","previous_headings":"","what":"Append tensor x2 to the end of tensor x1. — op_append","title":"Append tensor x2 to the end of tensor x1. — op_append","text":"Append tensor x2 end tensor x1.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_append.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Append tensor x2 to the end of tensor x1. — op_append","text":"","code":"op_append(x1, x2, axis = NULL)"},{"path":"https://keras3.posit.co/dev/reference/op_append.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Append tensor x2 to the end of tensor x1. — op_append","text":"x1 First input tensor. x2 Second input tensor. axis Axis along tensor x2 appended tensor x1. NULL, tensors flattened use.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_append.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Append tensor x2 to the end of tensor x1. — op_append","text":"tensor values x2 appended x1.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_append.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Append tensor x2 to the end of tensor x1. — op_append","text":"axis specified, x1 x2 must compatible shapes.","code":"x1 <- op_convert_to_tensor(c(1, 2, 3)) x2 <- op_convert_to_tensor(rbind(c(4, 5, 6), c(7, 8, 9))) op_append(x1, x2) ## tf.Tensor([1. 2. 3. 4. 5. 6. 7. 8. 9.], shape=(9), dtype=float64) x1 <- op_convert_to_tensor(rbind(c(1, 2, 3), c(4, 5, 6))) x2 <- op_convert_to_tensor(rbind(c(7, 8, 9))) op_append(x1, x2, axis = 1) ## tf.Tensor( ## [[1. 2. 3.] ##  [4. 5. 6.] ##  [7. 8. 9.]], shape=(3, 3), dtype=float64) x3 <- op_convert_to_tensor(c(7, 8, 9)) try(op_append(x1, x3, axis = 1)) ## Error in py_call_impl(callable, call_args$unnamed, call_args$named) : ##   tensorflow.python.framework.errors_impl.InvalidArgumentError: {{function_node __wrapped__ConcatV2_N_2_device_/job:localhost/replica:0/task:0/device:CPU:0}} ConcatOp : Ranks of all input tensors should match: shape[0] = [2,3] vs. shape[1] = [3] [Op:ConcatV2] name: concat"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_arange.html","id":null,"dir":"Reference","previous_headings":"","what":"Return evenly spaced values within a given interval. — op_arange","title":"Return evenly spaced values within a given interval. — op_arange","text":"arange can called varying number positional arguments: arange(end): Values generated within half-open interval [1, end] (words, interval including start end). arange(start, end): Values generated within interval [start, end]. arange(start, end, step): Values generated within interval [start, end], spacing values given step.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_arange.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Return evenly spaced values within a given interval. — op_arange","text":"","code":"op_arange(start, end, step = 1L, dtype = NULL, include_end = TRUE)"},{"path":"https://keras3.posit.co/dev/reference/op_arange.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Return evenly spaced values within a given interval. — op_arange","text":"start Integer real, representing start interval. interval includes value. end Integer real, representing end interval. include_end FALSE, interval include value, except cases step integer floating point round-affects length . Defaults NULL. step Integer real, represent spacing values. output , distance two adjacent values, [+1] - []. default step size 1. step specified position argument, start must also given. dtype type output array. dtype given, infer data type input arguments. include_end TRUE FALSE. FALSE, end included output sequence.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_arange.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Return evenly spaced values within a given interval. — op_arange","text":"Tensor evenly spaced values. floating point arguments, length result ceiling((end - start)/step). floating point overflow, rule may result last element greater end.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_arange.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Return evenly spaced values within a given interval. — op_arange","text":"","code":"op_arange(3L) ## tf.Tensor([1 2 3], shape=(3), dtype=int32) op_arange(3) # float ## tf.Tensor([1. 2. 3.], shape=(3), dtype=float32) op_arange(3, dtype = 'int32') #int ## tf.Tensor([1 2 3], shape=(3), dtype=int32) op_arange(3L, 7L) ## tf.Tensor([3 4 5 6 7], shape=(5), dtype=int32) op_arange(3L, 7L, 2L) ## tf.Tensor([3 5 7], shape=(3), dtype=int32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_arccos.html","id":null,"dir":"Reference","previous_headings":"","what":"Trigonometric inverse cosine, element-wise. — op_arccos","title":"Trigonometric inverse cosine, element-wise. — op_arccos","text":"inverse cos , y = cos(x), x = arccos(y).","code":""},{"path":"https://keras3.posit.co/dev/reference/op_arccos.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Trigonometric inverse cosine, element-wise. — op_arccos","text":"","code":"op_arccos(x)"},{"path":"https://keras3.posit.co/dev/reference/op_arccos.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Trigonometric inverse cosine, element-wise. — op_arccos","text":"x Input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_arccos.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Trigonometric inverse cosine, element-wise. — op_arccos","text":"Tensor angle ray intersecting unit circle given x-coordinate radians [0, pi].","code":""},{"path":"https://keras3.posit.co/dev/reference/op_arccos.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Trigonometric inverse cosine, element-wise. — op_arccos","text":"","code":"x <- op_convert_to_tensor(c(1, -1)) op_arccos(x) ## tf.Tensor([0.         3.14159265], shape=(2), dtype=float64)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_arccosh.html","id":null,"dir":"Reference","previous_headings":"","what":"Inverse hyperbolic cosine, element-wise. — op_arccosh","title":"Inverse hyperbolic cosine, element-wise. — op_arccosh","text":"Inverse hyperbolic cosine, element-wise.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_arccosh.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Inverse hyperbolic cosine, element-wise. — op_arccosh","text":"","code":"op_arccosh(x)"},{"path":"https://keras3.posit.co/dev/reference/op_arccosh.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Inverse hyperbolic cosine, element-wise. — op_arccosh","text":"x Input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_arccosh.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Inverse hyperbolic cosine, element-wise. — op_arccosh","text":"Output tensor shape x.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_arccosh.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Inverse hyperbolic cosine, element-wise. — op_arccosh","text":"","code":"x <- op_convert_to_tensor(c(10, 100)) op_arccosh(x) ## tf.Tensor([2.99322285 5.29829237], shape=(2), dtype=float64)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_arcsin.html","id":null,"dir":"Reference","previous_headings":"","what":"Inverse sine, element-wise. — op_arcsin","title":"Inverse sine, element-wise. — op_arcsin","text":"Inverse sine, element-wise.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_arcsin.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Inverse sine, element-wise. — op_arcsin","text":"","code":"op_arcsin(x)"},{"path":"https://keras3.posit.co/dev/reference/op_arcsin.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Inverse sine, element-wise. — op_arcsin","text":"x Input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_arcsin.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Inverse sine, element-wise. — op_arcsin","text":"Tensor inverse sine element x, radians closed interval [-pi/2, pi/2].","code":""},{"path":"https://keras3.posit.co/dev/reference/op_arcsin.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Inverse sine, element-wise. — op_arcsin","text":"","code":"x <- op_convert_to_tensor(c(1, -1, 0)) op_arcsin(x) ## tf.Tensor([ 1.57079633 -1.57079633  0.        ], shape=(3), dtype=float64)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_arcsinh.html","id":null,"dir":"Reference","previous_headings":"","what":"Inverse hyperbolic sine, element-wise. — op_arcsinh","title":"Inverse hyperbolic sine, element-wise. — op_arcsinh","text":"Inverse hyperbolic sine, element-wise.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_arcsinh.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Inverse hyperbolic sine, element-wise. — op_arcsinh","text":"","code":"op_arcsinh(x)"},{"path":"https://keras3.posit.co/dev/reference/op_arcsinh.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Inverse hyperbolic sine, element-wise. — op_arcsinh","text":"x Input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_arcsinh.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Inverse hyperbolic sine, element-wise. — op_arcsinh","text":"Output tensor shape x.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_arcsinh.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Inverse hyperbolic sine, element-wise. — op_arcsinh","text":"","code":"x <- op_convert_to_tensor(c(1, -1, 0)) op_arcsinh(x) ## tf.Tensor([ 0.88137359 -0.88137359  0.        ], shape=(3), dtype=float64)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_arctan.html","id":null,"dir":"Reference","previous_headings":"","what":"Trigonometric inverse tangent, element-wise. — op_arctan","title":"Trigonometric inverse tangent, element-wise. — op_arctan","text":"Trigonometric inverse tangent, element-wise.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_arctan.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Trigonometric inverse tangent, element-wise. — op_arctan","text":"","code":"op_arctan(x)"},{"path":"https://keras3.posit.co/dev/reference/op_arctan.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Trigonometric inverse tangent, element-wise. — op_arctan","text":"x Input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_arctan.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Trigonometric inverse tangent, element-wise. — op_arctan","text":"Tensor inverse tangent element x, interval [-pi/2, pi/2].","code":""},{"path":"https://keras3.posit.co/dev/reference/op_arctan.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Trigonometric inverse tangent, element-wise. — op_arctan","text":"","code":"x <- op_convert_to_tensor(c(0, 1)) op_arctan(x) ## tf.Tensor([0.         0.78539816], shape=(2), dtype=float64)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_arctan2.html","id":null,"dir":"Reference","previous_headings":"","what":"Element-wise arc tangent of x1/x2 choosing the quadrant correctly. — op_arctan2","title":"Element-wise arc tangent of x1/x2 choosing the quadrant correctly. — op_arctan2","text":"quadrant (.e., branch) chosen arctan2(x1, x2) signed angle radians ray ending origin passing point (1, 0), ray ending origin passing point (x2, x1). (Note role reversal: \"y-coordinate\" first function parameter, \"x-coordinate\" second.) IEEE convention, function defined x2 = +/-0 either x1 x2 = +/-inf.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_arctan2.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Element-wise arc tangent of x1/x2 choosing the quadrant correctly. — op_arctan2","text":"","code":"op_arctan2(x1, x2)"},{"path":"https://keras3.posit.co/dev/reference/op_arctan2.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Element-wise arc tangent of x1/x2 choosing the quadrant correctly. — op_arctan2","text":"x1 First input tensor. x2 Second input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_arctan2.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Element-wise arc tangent of x1/x2 choosing the quadrant correctly. — op_arctan2","text":"Tensor angles radians, range [-pi, pi].","code":""},{"path":"https://keras3.posit.co/dev/reference/op_arctan2.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Element-wise arc tangent of x1/x2 choosing the quadrant correctly. — op_arctan2","text":"Consider four points different quadrants:     Note order parameters. arctan2 defined also x2 = 0 several points, obtaining values range [-pi, pi]:","code":"x <- op_array(c(-1, 1, 1, -1)) y <- op_array(c(-1, -1, 1, 1)) op_arctan2(y, x) * 180 / pi ## tf.Tensor([-135.  -45.   45.  135.], shape=(4), dtype=float32) op_arctan2(     op_array(c(1, -1)),     op_array(c(0, 0)) ) ## tf.Tensor([ 1.5707964 -1.5707964], shape=(2), dtype=float32) op_arctan2(     op_array(c(0, 0, Inf)),     op_array(c(+0, -0, Inf)) ) ## tf.Tensor([0.        3.1415927 0.7853982], shape=(3), dtype=float32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_arctanh.html","id":null,"dir":"Reference","previous_headings":"","what":"Inverse hyperbolic tangent, element-wise. — op_arctanh","title":"Inverse hyperbolic tangent, element-wise. — op_arctanh","text":"Inverse hyperbolic tangent, element-wise.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_arctanh.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Inverse hyperbolic tangent, element-wise. — op_arctanh","text":"","code":"op_arctanh(x)"},{"path":"https://keras3.posit.co/dev/reference/op_arctanh.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Inverse hyperbolic tangent, element-wise. — op_arctanh","text":"x Input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_arctanh.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Inverse hyperbolic tangent, element-wise. — op_arctanh","text":"Output tensor shape x.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_argmax.html","id":null,"dir":"Reference","previous_headings":"","what":"Returns the indices of the maximum values along an axis. — op_argmax","title":"Returns the indices of the maximum values along an axis. — op_argmax","text":"Returns indices maximum values along axis.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_argmax.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Returns the indices of the maximum values along an axis. — op_argmax","text":"","code":"op_argmax(x, axis = NULL, keepdims = FALSE, zero_indexed = FALSE)"},{"path":"https://keras3.posit.co/dev/reference/op_argmax.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Returns the indices of the maximum values along an axis. — op_argmax","text":"x Input tensor. axis default, index flattened tensor, otherwise along specified axis. keepdims set TRUE, axes reduced left result dimensions size one. Defaults FALSE. zero_indexed TRUE, returned indices zero-based (0 encodes first position); FALSE (default), returned indices one-based (1 encodes first position).","code":""},{"path":"https://keras3.posit.co/dev/reference/op_argmax.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Returns the indices of the maximum values along an axis. — op_argmax","text":"Tensor indices. shape x, dimension along axis removed. Note returned integer 1-based (.e., argmax first index position, returned value 1)","code":""},{"path":"https://keras3.posit.co/dev/reference/op_argmax.html","id":"note","dir":"Reference","previous_headings":"","what":"Note","title":"Returns the indices of the maximum values along an axis. — op_argmax","text":"similar R max.col(x) case 2-d array (matrix), nd-array, apply(x, axis, .max)","code":""},{"path":"https://keras3.posit.co/dev/reference/op_argmax.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Returns the indices of the maximum values along an axis. — op_argmax","text":"","code":"x <- op_array(rbind(c(11, 12, 10),                     c(12, 11, 10))) x ## tf.Tensor( ## [[11. 12. 10.] ##  [12. 11. 10.]], shape=(2, 3), dtype=float64) op_argmax(x) ## tf.Tensor(2, shape=(), dtype=int32) op_argmax(x, axis = 1) ## tf.Tensor([2 1 1], shape=(3), dtype=int32) op_argmax(x, axis = 2) ## tf.Tensor([2 1], shape=(2), dtype=int32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_argmin.html","id":null,"dir":"Reference","previous_headings":"","what":"Returns the indices of the minimum values along an axis. — op_argmin","title":"Returns the indices of the minimum values along an axis. — op_argmin","text":"Returns indices minimum values along axis.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_argmin.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Returns the indices of the minimum values along an axis. — op_argmin","text":"","code":"op_argmin(x, axis = NULL, keepdims = FALSE, zero_indexed = FALSE)"},{"path":"https://keras3.posit.co/dev/reference/op_argmin.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Returns the indices of the minimum values along an axis. — op_argmin","text":"x Input tensor. axis default, index flattened tensor, otherwise along specified axis. keepdims set TRUE, axes reduced left result dimensions size one. Defaults FALSE. zero_indexed TRUE, returned indices zero-based (0 encodes first position); FALSE (default), returned indices one-based (1 encodes first position).","code":""},{"path":"https://keras3.posit.co/dev/reference/op_argmin.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Returns the indices of the minimum values along an axis. — op_argmin","text":"Tensor indices. shape x, dimension along axis removed.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_argmin.html","id":"note","dir":"Reference","previous_headings":"","what":"Note","title":"Returns the indices of the minimum values along an axis. — op_argmin","text":"similar R expression apply(x, axis, .min), x R array.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_argmin.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Returns the indices of the minimum values along an axis. — op_argmin","text":"","code":"x <- op_arange(6L) |> op_reshape(c(2, 3)) |> op_add(10) x ## tf.Tensor( ## [[11. 12. 13.] ##  [14. 15. 16.]], shape=(2, 3), dtype=float32) op_argmin(x) ## tf.Tensor(1, shape=(), dtype=int32) op_argmin(x, axis = 1) ## tf.Tensor([1 1 1], shape=(3), dtype=int32) op_argmin(x, axis = 2) ## tf.Tensor([1 1], shape=(2), dtype=int32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_argpartition.html","id":null,"dir":"Reference","previous_headings":"","what":"Performs an indirect partition along the given axis. — op_argpartition","title":"Performs an indirect partition along the given axis. — op_argpartition","text":"returns array indices shape x index data along given axis partitioned order.","code":"x <- op_convert_to_tensor(c(9, 3, 6, 2, 8, 5, 7, 1, 10, 4)) x@r[op_argpartition(x, 3)] ## tf.Tensor([ 1.  2.  3. 10.  9.  8.  7.  6.  5.  4.], shape=(10), dtype=float64) x@r[op_argpartition(x, 5)] ## tf.Tensor([ 1.  2.  3.  4.  5. 10.  9.  8.  7.  6.], shape=(10), dtype=float64) x@r[op_argpartition(x, 7)] ## tf.Tensor([ 1.  2.  3.  4.  5.  6.  7. 10.  9.  8.], shape=(10), dtype=float64)"},{"path":"https://keras3.posit.co/dev/reference/op_argpartition.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Performs an indirect partition along the given axis. — op_argpartition","text":"","code":"op_argpartition(x, kth, axis = -1L, zero_indexed = FALSE)"},{"path":"https://keras3.posit.co/dev/reference/op_argpartition.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Performs an indirect partition along the given axis. — op_argpartition","text":"x Array sort. kth Element index partition . k-th element final sorted position smaller elements moved larger elements behind . order elements partitions undefined. provided sequence k-th partition sorted position . axis Axis along sort. default -1 (last axis). NULL, flattened array used. zero_indexed TRUE, returned indices zero-based (0 encodes first position); FALSE (default), returned indices one-based (1 encodes first position).","code":""},{"path":"https://keras3.posit.co/dev/reference/op_argpartition.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Performs an indirect partition along the given axis. — op_argpartition","text":"Array indices partition x along specified axis.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_argsort.html","id":null,"dir":"Reference","previous_headings":"","what":"Returns the indices that would sort a tensor. — op_argsort","title":"Returns the indices that would sort a tensor. — op_argsort","text":"Returns indices sort tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_argsort.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Returns the indices that would sort a tensor. — op_argsort","text":"","code":"op_argsort(x, axis = -1L, zero_indexed = FALSE)"},{"path":"https://keras3.posit.co/dev/reference/op_argsort.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Returns the indices that would sort a tensor. — op_argsort","text":"x Input tensor. axis Axis along sort. Defaults -1 (last axis). NULL, flattened tensor used. zero_indexed TRUE, returned indices zero-based (0 encodes first position); FALSE (default), returned indices one-based (1 encodes first position).","code":""},{"path":"https://keras3.posit.co/dev/reference/op_argsort.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Returns the indices that would sort a tensor. — op_argsort","text":"Tensor indices sort x along specified axis.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_argsort.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Returns the indices that would sort a tensor. — op_argsort","text":"One dimensional array:         Two-dimensional array:","code":"x <- op_array(c(3, 1, 2)) + .1 op_argsort(x) ## tf.Tensor([2 3 1], shape=(3), dtype=int32) x@r[op_argsort(x)] == op_sort(x) ## tf.Tensor([ True  True  True], shape=(3), dtype=bool) x <- op_array(rbind(c(0, 3),                     c(3, 2),                     c(4, 5))) + .1 (i <- op_argsort(x, axis = 1)) ## tf.Tensor( ## [[1 2] ##  [2 1] ##  [3 3]], shape=(3, 2), dtype=int32) x@r[i@r[, 1], ] # sort x-rows using first col of x ## tf.Tensor( ## [[0.1 3.1] ##  [3.1 2.1] ##  [4.1 5.1]], shape=(3, 2), dtype=float64) x@r[i@r[, 2], ] # sort x-rows using second col of x ## tf.Tensor( ## [[3.1 2.1] ##  [0.1 3.1] ##  [4.1 5.1]], shape=(3, 2), dtype=float64) (i <- op_argsort(x, axis = 2)) ## tf.Tensor( ## [[1 2] ##  [2 1] ##  [1 2]], shape=(3, 2), dtype=int32) x@r[, i@r[2,]] # sort x-cols using second row of x ## tf.Tensor( ## [[3.1 0.1] ##  [2.1 3.1] ##  [5.1 4.1]], shape=(3, 2), dtype=float64)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_array.html","id":null,"dir":"Reference","previous_headings":"","what":"Create a tensor. — op_array","title":"Create a tensor. — op_array","text":"Create tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_array.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Create a tensor. — op_array","text":"","code":"op_array(x, dtype = NULL)"},{"path":"https://keras3.posit.co/dev/reference/op_array.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Create a tensor. — op_array","text":"x Input tensor. dtype desired data-type tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_array.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Create a tensor. — op_array","text":"tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_array.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Create a tensor. — op_array","text":"","code":"op_array(c(1, 2, 3)) ## tf.Tensor([1. 2. 3.], shape=(3), dtype=float32) op_array(c(1, 2, 3), dtype = \"float32\") ## tf.Tensor([1. 2. 3.], shape=(3), dtype=float32) op_array(c(1, 2, 3), dtype = \"int32\") ## tf.Tensor([1 2 3], shape=(3), dtype=int32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_associative_scan.html","id":null,"dir":"Reference","previous_headings":"","what":"Performs a scan with an associative binary operation, in parallel. — op_associative_scan","title":"Performs a scan with an associative binary operation, in parallel. — op_associative_scan","text":"operation similar op_scan(), key difference op_associative_scan() parallel implementation potentially significant performance benefits, especially jit compiled. catch can used f binary associative operation (.e. must verify f(, f(b, c)) == f(f(, b), c)). introduction associative scans, refer paper: Blelloch, Guy E. 1990. Prefix Sums Applications.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_associative_scan.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Performs a scan with an associative binary operation, in parallel. — op_associative_scan","text":"","code":"op_associative_scan(f, elems, reverse = FALSE, axis = 1L)"},{"path":"https://keras3.posit.co/dev/reference/op_associative_scan.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Performs a scan with an associative binary operation, in parallel. — op_associative_scan","text":"f callable implementing associative binary operation signature  r = f(, b). Function f must associative, .e., must satisfy equation f(, f(b, c)) == f(f(, b), c). inputs result (possibly nested tree structures ) array(s) matching elems. array dimension place axis dimension. f applied elementwise axis dimension. result r shape (structure) two inputs b. elems (possibly nested tree structure ) array(s), axis dimension size num_elems. reverse boolean stating scan reversed respect axis dimension. axis integer identifying axis scan occur.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_associative_scan.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Performs a scan with an associative binary operation, in parallel. — op_associative_scan","text":"(possibly nested tree structure ) array(s) shape structure elems, k'th element axis result recursively applying f combine first k elements elems along axis. example, given elems = list(, b, c, ...), result list(, f(, b), f(f(, b), c), ...).","code":""},{"path":"https://keras3.posit.co/dev/reference/op_associative_scan.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Performs a scan with an associative binary operation, in parallel. — op_associative_scan","text":"","code":"sum_fn <- function(x, y) x + y xs <- op_arange(5L) op_associative_scan(sum_fn, xs) ## tf.Tensor([ 1  3  6 10 15], shape=(5), dtype=int32) sum_fn <- function(x, y) {   str(list(x = x, y = y))   map2(x, y, \\(.x, .y) .x + .y) }  xs <- list(op_array(1:2),            op_array(1:2),            op_array(1:2)) ys <- op_associative_scan(sum_fn, xs, axis = 1) ## List of 2 ##  $ x:List of 3 ##   ..$ :<tf.Tensor: shape=(1), dtype=int32, numpy=array([1], dtype=int32)> ##   ..$ :<tf.Tensor: shape=(1), dtype=int32, numpy=array([1], dtype=int32)> ##   ..$ :<tf.Tensor: shape=(1), dtype=int32, numpy=array([1], dtype=int32)> ##  $ y:List of 3 ##   ..$ :<tf.Tensor: shape=(1), dtype=int32, numpy=array([2], dtype=int32)> ##   ..$ :<tf.Tensor: shape=(1), dtype=int32, numpy=array([2], dtype=int32)> ##   ..$ :<tf.Tensor: shape=(1), dtype=int32, numpy=array([2], dtype=int32)> ys ## [[1]] ## tf.Tensor([1 3], shape=(2), dtype=int32) ## ## [[2]] ## tf.Tensor([1 3], shape=(2), dtype=int32) ## ## [[3]] ## tf.Tensor([1 3], shape=(2), dtype=int32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_average.html","id":null,"dir":"Reference","previous_headings":"","what":"Compute the weighted average along the specified axis. — op_average","title":"Compute the weighted average along the specified axis. — op_average","text":"Compute weighted average along specified axis.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_average.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Compute the weighted average along the specified axis. — op_average","text":"","code":"op_average(x, axis = NULL, weights = NULL)"},{"path":"https://keras3.posit.co/dev/reference/op_average.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Compute the weighted average along the specified axis. — op_average","text":"x Input tensor. axis Integer along average x. default, axis = NULL, average elements input tensor. axis negative counts last first axis. weights Tensor weights associated values x. value x contributes average according associated weight. weights array can either 1-D (case length must size along given axis) shape x. weights = NULL (default), data x assumed weight equal one. 1-D calculation : avg = sum(* weights) / sum(weights). constraint weights sum(weights) must 0.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_average.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Compute the weighted average along the specified axis. — op_average","text":"Return average along specified axis.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_average.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Compute the weighted average along the specified axis. — op_average","text":"","code":"data <- op_arange(1, 5, dtype = \"int32\") data ## tf.Tensor([1 2 3 4 5], shape=(5), dtype=int32) op_average(data) ## tf.Tensor(3.0, shape=(), dtype=float32) op_average(   op_arange(1, 11),   weights = op_arange(10, 0, -1) ) ## tf.Tensor(4.0, shape=(), dtype=float32) data <- op_arange(6) |> op_reshape(c(3, 2)) data ## tf.Tensor( ## [[1. 2.] ##  [3. 4.] ##  [5. 6.]], shape=(3, 2), dtype=float32) op_average(   data,   axis = 2,   weights = op_array(c(1/4, 3/4)) ) ## tf.Tensor([1.75 3.75 5.75], shape=(3), dtype=float32) # Error: Axis must be specified when shapes of x and weights differ. try(op_average(   data,   weights = op_array(c(1/4, 3/4)) )) ## Error in op_average(data, weights = op_array(c(1/4, 3/4))) : ##   Axis must be specified when shapes of x and weights differ."},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_average_pool.html","id":null,"dir":"Reference","previous_headings":"","what":"Average pooling operation. — op_average_pool","title":"Average pooling operation. — op_average_pool","text":"Average pooling operation.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_average_pool.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Average pooling operation. — op_average_pool","text":"","code":"op_average_pool(   inputs,   pool_size,   strides = NULL,   padding = \"valid\",   data_format = NULL )"},{"path":"https://keras3.posit.co/dev/reference/op_average_pool.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Average pooling operation. — op_average_pool","text":"inputs Tensor rank N+2. inputs shape (batch_size,) + inputs_spatial_shape + (num_channels,) data_format = \"channels_last\", (batch_size, num_channels) + inputs_spatial_shape data_format = \"channels_first\". Pooling happens spatial dimensions . pool_size int tuple/list integers size len(inputs_spatial_shape), specifying size pooling window spatial dimension input tensor. pool_size int, every spatial dimension shares pool_size. strides int tuple/list integers size len(inputs_spatial_shape). stride sliding window spatial dimension input tensor. strides int, every spatial dimension shares strides. padding string, either \"valid\" \"\". \"valid\" means padding applied, \"\" results padding evenly left/right /input output height/width dimension input strides = 1. data_format string, either \"channels_last\" \"channels_first\". data_format determines ordering dimensions inputs. data_format = \"channels_last\", inputs shape (batch_size, ..., channels) data_format = \"channels_first\", inputs shape (batch_size, channels, ...).","code":""},{"path":"https://keras3.posit.co/dev/reference/op_average_pool.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Average pooling operation. — op_average_pool","text":"tensor rank N+2, result average pooling operation.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_bartlett.html","id":null,"dir":"Reference","previous_headings":"","what":"Bartlett window function. — op_bartlett","title":"Bartlett window function. — op_bartlett","text":"Bartlett window triangular window rises falls linearly.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_bartlett.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Bartlett window function. — op_bartlett","text":"","code":"op_bartlett(x)"},{"path":"https://keras3.posit.co/dev/reference/op_bartlett.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Bartlett window function. — op_bartlett","text":"x Length window. Must positive integer.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_bartlett.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Bartlett window function. — op_bartlett","text":"1D tensor containing window values.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_bartlett.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Bartlett window function. — op_bartlett","text":"","code":"op_bartlett(5) ## tf.Tensor([0.  0.5 1.  0.5 0. ], shape=(5), dtype=float32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_batch_normalization.html","id":null,"dir":"Reference","previous_headings":"","what":"Normalizes x by mean and variance. — op_batch_normalization","title":"Normalizes x by mean and variance. — op_batch_normalization","text":"op typically used batch normalization step neural network. normalizes input tensor along given axis.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_batch_normalization.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Normalizes x by mean and variance. — op_batch_normalization","text":"","code":"op_batch_normalization(   x,   mean,   variance,   axis,   offset = NULL,   scale = NULL,   epsilon = 0.001 )"},{"path":"https://keras3.posit.co/dev/reference/op_batch_normalization.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Normalizes x by mean and variance. — op_batch_normalization","text":"x Input tensor. mean mean vector length axis dimension input thensor. variance variance vector length axis dimension input tensor. axis Integer, axis normalized. offset offset vector length axis dimension input tensor. NULL, offset added normalized tensor. Defaults NULL. scale scale vector length axis dimension input tensor. NULL, normalized tensor multiplied scale. Defaults NULL. epsilon Small float added variance avoid dividing zero. Defaults 1e-3.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_batch_normalization.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Normalizes x by mean and variance. — op_batch_normalization","text":"normalized tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_batch_normalization.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Normalizes x by mean and variance. — op_batch_normalization","text":"","code":"x <- op_convert_to_tensor(rbind(c(0.1, 0.2, 0.3),                                 c(0.4, 0.5, 0.6),                                 c(0.7, 0.8, 0.9))) op_batch_normalization(   x,   mean = c(0.4, 0.5, 0.6),   variance = c(0.67, 0.67, 0.67),   axis = -1 ) ## tf.Tensor( ## [[-0.36623513 -0.36623513 -0.36623513] ##  [ 0.          0.          0.        ] ##  [ 0.36623513  0.36623513  0.36623513]], shape=(3, 3), dtype=float64)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_binary_crossentropy.html","id":null,"dir":"Reference","previous_headings":"","what":"Computes binary cross-entropy loss between target and output tensor. — op_binary_crossentropy","title":"Computes binary cross-entropy loss between target and output tensor. — op_binary_crossentropy","text":"binary cross-entropy loss commonly used binary classification tasks input sample belongs one two classes. measures dissimilarity target output probabilities logits.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_binary_crossentropy.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes binary cross-entropy loss between target and output tensor. — op_binary_crossentropy","text":"","code":"op_binary_crossentropy(target, output, from_logits = FALSE)"},{"path":"https://keras3.posit.co/dev/reference/op_binary_crossentropy.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Computes binary cross-entropy loss between target and output tensor. — op_binary_crossentropy","text":"target target tensor representing true binary labels. shape match shape output tensor. output output tensor representing predicted probabilities logits. shape match shape target tensor. from_logits (optional) Whether output tensor logits probabilities. Set TRUE output represents logits; otherwise, set FALSE output represents probabilities. Defaults FALSE.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_binary_crossentropy.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Computes binary cross-entropy loss between target and output tensor. — op_binary_crossentropy","text":"Integer tensor: computed binary cross-entropy loss target output.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_binary_crossentropy.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Computes binary cross-entropy loss between target and output tensor. — op_binary_crossentropy","text":"","code":"target <- op_array(c(0, 1, 1, 0)) output <- op_array(c(0.1, 0.9, 0.8, 0.2)) op_binary_crossentropy(target, output) ## tf.Tensor([0.10536055 0.10536055 0.22314353 0.22314353], shape=(4), dtype=float32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_bincount.html","id":null,"dir":"Reference","previous_headings":"","what":"Count the number of occurrences of each value in a tensor of integers. — op_bincount","title":"Count the number of occurrences of each value in a tensor of integers. — op_bincount","text":"Count number occurrences value tensor integers.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_bincount.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Count the number of occurrences of each value in a tensor of integers. — op_bincount","text":"","code":"op_bincount(x, weights = NULL, minlength = 0L, sparse = FALSE)"},{"path":"https://keras3.posit.co/dev/reference/op_bincount.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Count the number of occurrences of each value in a tensor of integers. — op_bincount","text":"x Input tensor. must dimension 1, must contain non-negative integer(s). weights Weight tensor. must length x. default value NULL. specified, x weighted , .e. n = x[], [n] += weight[] instead default behavior [n] += 1. minlength integer. default value 0. specified, least number bins output tensor. greater max(x) + 1, value output index higher max(x) set 0. sparse Whether return sparse tensor; backends support sparse tensors.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_bincount.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Count the number of occurrences of each value in a tensor of integers. — op_bincount","text":"1D tensor element gives number occurrence(s) index value x. length maximum max(x) + 1 minlength.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_bincount.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Count the number of occurrences of each value in a tensor of integers. — op_bincount","text":"","code":"(x <- op_array(c(1, 2, 2, 3), dtype = \"uint8\")) ## tf.Tensor([1 2 2 3], shape=(4), dtype=uint8) op_bincount(x) ## tf.Tensor([0 1 2 1], shape=(4), dtype=int32) (weights <- x / 2) ## tf.Tensor([0.5 1.  1.  1.5], shape=(4), dtype=float32) op_bincount(x, weights = weights) ## tf.Tensor([0.  0.5 2.  1.5], shape=(4), dtype=float32) minlength <- as.integer(op_max(x) + 1 + 2) # 6 op_bincount(x, minlength = minlength) ## tf.Tensor([0 1 2 1 0 0], shape=(6), dtype=int32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_bitwise_and.html","id":null,"dir":"Reference","previous_headings":"","what":"Compute the bit-wise AND of two arrays element-wise. — op_bitwise_and","title":"Compute the bit-wise AND of two arrays element-wise. — op_bitwise_and","text":"Computes bit-wise underlying binary representation integers input arrays. ufunc implements C/Python operator &.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_bitwise_and.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Compute the bit-wise AND of two arrays element-wise. — op_bitwise_and","text":"","code":"op_bitwise_and(x, y)"},{"path":"https://keras3.posit.co/dev/reference/op_bitwise_and.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Compute the bit-wise AND of two arrays element-wise. — op_bitwise_and","text":"x Input integer tensor. y Input integer tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_bitwise_and.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Compute the bit-wise AND of two arrays element-wise. — op_bitwise_and","text":"Result tensor.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_bitwise_invert.html","id":null,"dir":"Reference","previous_headings":"","what":"Compute bit-wise inversion, or bit-wise NOT, element-wise. — op_bitwise_invert","title":"Compute bit-wise inversion, or bit-wise NOT, element-wise. — op_bitwise_invert","text":"Computes bit-wise underlying binary representation integers input arrays. ufunc implements C/Python operator ~.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_bitwise_invert.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Compute bit-wise inversion, or bit-wise NOT, element-wise. — op_bitwise_invert","text":"","code":"op_bitwise_invert(x)"},{"path":"https://keras3.posit.co/dev/reference/op_bitwise_invert.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Compute bit-wise inversion, or bit-wise NOT, element-wise. — op_bitwise_invert","text":"x Input integer tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_bitwise_invert.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Compute bit-wise inversion, or bit-wise NOT, element-wise. — op_bitwise_invert","text":"Result tensor.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_bitwise_left_shift.html","id":null,"dir":"Reference","previous_headings":"","what":"Shift the bits of an integer to the left. — op_bitwise_left_shift","title":"Shift the bits of an integer to the left. — op_bitwise_left_shift","text":"Bits shifted left appending y 0s right x. Since internal representation numbers binary format, operation equivalent multiplying x 2**y.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_bitwise_left_shift.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Shift the bits of an integer to the left. — op_bitwise_left_shift","text":"","code":"op_bitwise_left_shift(x, y)"},{"path":"https://keras3.posit.co/dev/reference/op_bitwise_left_shift.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Shift the bits of an integer to the left. — op_bitwise_left_shift","text":"x Input integer tensor. y Input integer tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_bitwise_left_shift.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Shift the bits of an integer to the left. — op_bitwise_left_shift","text":"Result tensor.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_bitwise_not.html","id":null,"dir":"Reference","previous_headings":"","what":"Compute bit-wise inversion, or bit-wise NOT, element-wise. — op_bitwise_not","title":"Compute bit-wise inversion, or bit-wise NOT, element-wise. — op_bitwise_not","text":"Computes bit-wise underlying binary representation integers input arrays. ufunc implements C/Python operator ~.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_bitwise_not.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Compute bit-wise inversion, or bit-wise NOT, element-wise. — op_bitwise_not","text":"","code":"op_bitwise_not(x)"},{"path":"https://keras3.posit.co/dev/reference/op_bitwise_not.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Compute bit-wise inversion, or bit-wise NOT, element-wise. — op_bitwise_not","text":"x Input integer tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_bitwise_not.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Compute bit-wise inversion, or bit-wise NOT, element-wise. — op_bitwise_not","text":"Result tensor.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_bitwise_or.html","id":null,"dir":"Reference","previous_headings":"","what":"Compute the bit-wise OR of two arrays element-wise. — op_bitwise_or","title":"Compute the bit-wise OR of two arrays element-wise. — op_bitwise_or","text":"Computes bit-wise underlying binary representation integers input arrays. ufunc implements C/Python operator |.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_bitwise_or.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Compute the bit-wise OR of two arrays element-wise. — op_bitwise_or","text":"","code":"op_bitwise_or(x, y)"},{"path":"https://keras3.posit.co/dev/reference/op_bitwise_or.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Compute the bit-wise OR of two arrays element-wise. — op_bitwise_or","text":"x Input integer tensor. y Input integer tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_bitwise_or.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Compute the bit-wise OR of two arrays element-wise. — op_bitwise_or","text":"Result tensor.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_bitwise_right_shift.html","id":null,"dir":"Reference","previous_headings":"","what":"Shift the bits of an integer to the right. — op_bitwise_right_shift","title":"Shift the bits of an integer to the right. — op_bitwise_right_shift","text":"Bits shifted right y. internal representation numbers binary format, operation equivalent dividing x 2**y.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_bitwise_right_shift.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Shift the bits of an integer to the right. — op_bitwise_right_shift","text":"","code":"op_bitwise_right_shift(x, y)"},{"path":"https://keras3.posit.co/dev/reference/op_bitwise_right_shift.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Shift the bits of an integer to the right. — op_bitwise_right_shift","text":"x Input integer tensor. y Input integer tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_bitwise_right_shift.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Shift the bits of an integer to the right. — op_bitwise_right_shift","text":"Result tensor.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_bitwise_xor.html","id":null,"dir":"Reference","previous_headings":"","what":"Compute the bit-wise XOR of two arrays element-wise. — op_bitwise_xor","title":"Compute the bit-wise XOR of two arrays element-wise. — op_bitwise_xor","text":"Computes bit-wise XOR underlying binary representation integers input arrays. ufunc implements C/Python operator ^.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_bitwise_xor.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Compute the bit-wise XOR of two arrays element-wise. — op_bitwise_xor","text":"","code":"op_bitwise_xor(x, y)"},{"path":"https://keras3.posit.co/dev/reference/op_bitwise_xor.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Compute the bit-wise XOR of two arrays element-wise. — op_bitwise_xor","text":"x Input integer tensor. y Input integer tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_bitwise_xor.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Compute the bit-wise XOR of two arrays element-wise. — op_bitwise_xor","text":"Result tensor.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_blackman.html","id":null,"dir":"Reference","previous_headings":"","what":"Blackman window function. — op_blackman","title":"Blackman window function. — op_blackman","text":"Blackman window taper formed using weighted cosine.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_blackman.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Blackman window function. — op_blackman","text":"","code":"op_blackman(x)"},{"path":"https://keras3.posit.co/dev/reference/op_blackman.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Blackman window function. — op_blackman","text":"x Length window. Must positive integer.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_blackman.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Blackman window function. — op_blackman","text":"1D tensor containing window values.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_blackman.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Blackman window function. — op_blackman","text":"[-1.4901161e-08  3.4000003e-01  9.9999994e-01  3.3999997e-01","code":"op_blackman(5) ## tf.Tensor( ## [-1.4901161e-08  3.4000003e-01  9.9999994e-01  3.3999997e-01 ##  -1.4901161e-08], shape=(5), dtype=float32)"},{"path":[]},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_broadcast_to.html","id":null,"dir":"Reference","previous_headings":"","what":"Broadcast a tensor to a new shape. — op_broadcast_to","title":"Broadcast a tensor to a new shape. — op_broadcast_to","text":"Broadcast tensor new shape.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_broadcast_to.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Broadcast a tensor to a new shape. — op_broadcast_to","text":"","code":"op_broadcast_to(x, shape)"},{"path":"https://keras3.posit.co/dev/reference/op_broadcast_to.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Broadcast a tensor to a new shape. — op_broadcast_to","text":"x tensor broadcast. shape shape desired tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_broadcast_to.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Broadcast a tensor to a new shape. — op_broadcast_to","text":"tensor desired shape.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_broadcast_to.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Broadcast a tensor to a new shape. — op_broadcast_to","text":"","code":"x <- op_array(c(1, 2, 3)) op_broadcast_to(x, shape = c(3, 3)) ## tf.Tensor( ## [[1. 2. 3.] ##  [1. 2. 3.] ##  [1. 2. 3.]], shape=(3, 3), dtype=float32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_cast.html","id":null,"dir":"Reference","previous_headings":"","what":"Cast a tensor to the desired dtype. — op_cast","title":"Cast a tensor to the desired dtype. — op_cast","text":"Cast tensor desired dtype.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_cast.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Cast a tensor to the desired dtype. — op_cast","text":"","code":"op_cast(x, dtype)"},{"path":"https://keras3.posit.co/dev/reference/op_cast.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Cast a tensor to the desired dtype. — op_cast","text":"x tensor variable. dtype target type.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_cast.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Cast a tensor to the desired dtype. — op_cast","text":"tensor specified dtype.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_cast.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Cast a tensor to the desired dtype. — op_cast","text":"","code":"(x <- op_arange(4)) ## tf.Tensor([1. 2. 3. 4.], shape=(4), dtype=float32) op_cast(x, dtype = \"float16\") ## tf.Tensor([1. 2. 3. 4.], shape=(4), dtype=float16)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_categorical_crossentropy.html","id":null,"dir":"Reference","previous_headings":"","what":"Computes categorical cross-entropy loss between target and output tensor. — op_categorical_crossentropy","title":"Computes categorical cross-entropy loss between target and output tensor. — op_categorical_crossentropy","text":"categorical cross-entropy loss commonly used multi-class classification tasks input sample can belong one multiple classes. measures dissimilarity target output probabilities logits.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_categorical_crossentropy.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes categorical cross-entropy loss between target and output tensor. — op_categorical_crossentropy","text":"","code":"op_categorical_crossentropy(target, output, from_logits = FALSE, axis = -1L)"},{"path":"https://keras3.posit.co/dev/reference/op_categorical_crossentropy.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Computes categorical cross-entropy loss between target and output tensor. — op_categorical_crossentropy","text":"target target tensor representing true categorical labels. shape match shape output tensor except last dimension. output output tensor representing predicted probabilities logits. shape match shape target tensor except last dimension. from_logits (optional) Whether output tensor logits probabilities. Set TRUE output represents logits; otherwise, set FALSE output represents probabilities. Defaults FALSE. axis (optional) axis along categorical cross-entropy computed. Defaults -1, corresponds last dimension tensors.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_categorical_crossentropy.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Computes categorical cross-entropy loss between target and output tensor. — op_categorical_crossentropy","text":"Integer tensor: computed categorical cross-entropy loss target output.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_categorical_crossentropy.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Computes categorical cross-entropy loss between target and output tensor. — op_categorical_crossentropy","text":"","code":"target <- op_array(rbind(c(1, 0, 0),                         c(0, 1, 0),                         c(0, 0, 1))) output <- op_array(rbind(c(0.9, 0.05, 0.05),                         c(0.1, 0.8, 0.1),                         c(0.2, 0.3, 0.5))) op_categorical_crossentropy(target, output) ## tf.Tensor([0.10536052 0.22314355 0.69314718], shape=(3), dtype=float64)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_cbrt.html","id":null,"dir":"Reference","previous_headings":"","what":"Computes the cube root of the input tensor, element-wise. — op_cbrt","title":"Computes the cube root of the input tensor, element-wise. — op_cbrt","text":"Returns real-valued cube root x, handling negative inputs real domain.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_cbrt.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes the cube root of the input tensor, element-wise. — op_cbrt","text":"","code":"op_cbrt(x)"},{"path":"https://keras3.posit.co/dev/reference/op_cbrt.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Computes the cube root of the input tensor, element-wise. — op_cbrt","text":"x Input tensor.Examples","code":"op_cbrt(c(-8, 0, 8)) ## tf.Tensor([-2.  0.  2.], shape=(3), dtype=float32)"},{"path":"https://keras3.posit.co/dev/reference/op_cbrt.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Computes the cube root of the input tensor, element-wise. — op_cbrt","text":"tensor containing cube root element x.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_ceil.html","id":null,"dir":"Reference","previous_headings":"","what":"Return the ceiling of the input, element-wise. — op_ceil","title":"Return the ceiling of the input, element-wise. — op_ceil","text":"ceil scalar x smallest integer , >= x.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_ceil.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Return the ceiling of the input, element-wise. — op_ceil","text":"","code":"op_ceil(x)"},{"path":"https://keras3.posit.co/dev/reference/op_ceil.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Return the ceiling of the input, element-wise. — op_ceil","text":"x Input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_ceil.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Return the ceiling of the input, element-wise. — op_ceil","text":"ceiling element x, float dtype.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_celu.html","id":null,"dir":"Reference","previous_headings":"","what":"Continuously-differentiable exponential linear unit. — op_celu","title":"Continuously-differentiable exponential linear unit. — op_celu","text":"defined : f(x) =  alpha * (exp(x / alpha) - 1) x < 0, f(x) = x x >= 0.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_celu.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Continuously-differentiable exponential linear unit. — op_celu","text":"","code":"op_celu(x, alpha = 1)"},{"path":"https://keras3.posit.co/dev/reference/op_celu.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Continuously-differentiable exponential linear unit. — op_celu","text":"x Input tensor. alpha value CELU formulation. Defaults 1.0.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_celu.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Continuously-differentiable exponential linear unit. — op_celu","text":"tensor shape x.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_celu.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Continuously-differentiable exponential linear unit. — op_celu","text":"","code":"x <- op_array(c(-1., 0., 1.)) op_celu(x) ## tf.Tensor([-0.63212055  0.          1.        ], shape=(3), dtype=float32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_cholesky.html","id":null,"dir":"Reference","previous_headings":"","what":"Computes the Cholesky decomposition of a positive semi-definite matrix. — op_cholesky","title":"Computes the Cholesky decomposition of a positive semi-definite matrix. — op_cholesky","text":"Computes Cholesky decomposition positive semi-definite matrix.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_cholesky.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes the Cholesky decomposition of a positive semi-definite matrix. — op_cholesky","text":"","code":"op_cholesky(x)"},{"path":"https://keras3.posit.co/dev/reference/op_cholesky.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Computes the Cholesky decomposition of a positive semi-definite matrix. — op_cholesky","text":"x Input tensor shape (..., M, M).","code":""},{"path":"https://keras3.posit.co/dev/reference/op_cholesky.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Computes the Cholesky decomposition of a positive semi-definite matrix. — op_cholesky","text":"tensor shape (..., M, M) representing lower triangular Cholesky factor x.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_clip.html","id":null,"dir":"Reference","previous_headings":"","what":"Clip (limit) the values in a tensor. — op_clip","title":"Clip (limit) the values in a tensor. — op_clip","text":"Given interval, values outside interval clipped interval edges. example, interval [0, 1] specified, values smaller 0 become 0, values larger 1 become 1.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_clip.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Clip (limit) the values in a tensor. — op_clip","text":"","code":"op_clip(x, x_min, x_max)"},{"path":"https://keras3.posit.co/dev/reference/op_clip.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Clip (limit) the values in a tensor. — op_clip","text":"x Input tensor. x_min Minimum value. x_max Maximum value.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_clip.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Clip (limit) the values in a tensor. — op_clip","text":"clipped tensor.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_concatenate.html","id":null,"dir":"Reference","previous_headings":"","what":"Join a sequence of tensors along an existing axis. — op_concatenate","title":"Join a sequence of tensors along an existing axis. — op_concatenate","text":"Join sequence tensors along existing axis.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_concatenate.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Join a sequence of tensors along an existing axis. — op_concatenate","text":"","code":"op_concatenate(xs, axis = 1L)"},{"path":"https://keras3.posit.co/dev/reference/op_concatenate.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Join a sequence of tensors along an existing axis. — op_concatenate","text":"xs sequence tensors concatenate. axis axis along tensors joined. Defaults 0.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_concatenate.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Join a sequence of tensors along an existing axis. — op_concatenate","text":"concatenated tensor.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_cond.html","id":null,"dir":"Reference","previous_headings":"","what":"Conditionally applies true_fn or false_fn. — op_cond","title":"Conditionally applies true_fn or false_fn. — op_cond","text":"Conditionally applies true_fn false_fn.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_cond.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Conditionally applies true_fn or false_fn. — op_cond","text":"","code":"op_cond(pred, true_fn, false_fn)"},{"path":"https://keras3.posit.co/dev/reference/op_cond.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Conditionally applies true_fn or false_fn. — op_cond","text":"pred Boolean scalar type true_fn Callable returning output pred == TRUE case. false_fn Callable returning output pred == FALSE case.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_cond.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Conditionally applies true_fn or false_fn. — op_cond","text":"output either true_fn false_fn depending pred.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_cond.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Conditionally applies true_fn or false_fn. — op_cond","text":"","code":"fn <- tensorflow::tf_function(function(x) {   op_cond(x > 0,     true_fn = \\() x + 1,     false_fn = \\() x - 1) })  fn(tensorflow::as_tensor(1)) ## tf.Tensor(2.0, shape=(), dtype=float64) fn(tensorflow::as_tensor(-1)) ## tf.Tensor(-2.0, shape=(), dtype=float64) # # Conditional side-effect (print only, no return value). file <- tempfile(fileext = \".txt\") fn <- tensorflow::tf_function(function(epochs) {   op_fori_loop(     0, epochs,     body_fun = \\(epoch, state) {       op_cond(epoch %% 20 == 0,               \\() {                 tensorflow::tf$print(                   \"epoch:\", epoch,                   output_stream = paste0(\"file://\", file))                 NULL               },               \\() {NULL})       state     },     init_val = tensorflow::as_tensor(0, \"float32\")) })  fn(tensorflow::as_tensor(100, \"float32\")) ## tf.Tensor(0.0, shape=(), dtype=float32) readLines(file) ## [1] \"epoch: 0\"  \"epoch: 20\" \"epoch: 40\" \"epoch: 60\" \"epoch: 80\" # cleanup unlink(file)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_conj.html","id":null,"dir":"Reference","previous_headings":"","what":"Returns the complex conjugate, element-wise. — op_conj","title":"Returns the complex conjugate, element-wise. — op_conj","text":"complex conjugate complex number obtained changing sign imaginary part.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_conj.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Returns the complex conjugate, element-wise. — op_conj","text":"","code":"op_conj(x)"},{"path":"https://keras3.posit.co/dev/reference/op_conj.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Returns the complex conjugate, element-wise. — op_conj","text":"x Input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_conj.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Returns the complex conjugate, element-wise. — op_conj","text":"complex conjugate element x.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_conv.html","id":null,"dir":"Reference","previous_headings":"","what":"General N-D convolution. — op_conv","title":"General N-D convolution. — op_conv","text":"ops supports 1D, 2D 3D convolution.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_conv.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"General N-D convolution. — op_conv","text":"","code":"op_conv(   inputs,   kernel,   strides = 1L,   padding = \"valid\",   data_format = NULL,   dilation_rate = 1L )"},{"path":"https://keras3.posit.co/dev/reference/op_conv.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"General N-D convolution. — op_conv","text":"inputs Tensor rank N+2. inputs shape (batch_size,) + inputs_spatial_shape + (num_channels,) data_format = \"channels_last\", (batch_size, num_channels) + inputs_spatial_shape data_format = \"channels_first\". kernel Tensor rank N+2. kernel shape (kernel_spatial_shape, num_input_channels, num_output_channels). num_input_channels match number channels inputs. strides int int tuple/list len(inputs_spatial_shape), specifying strides convolution along spatial dimension. strides int, every spatial dimension shares strides. Defaults 1. padding string, either \"valid\" \"\". \"valid\" means padding applied, \"\" results padding evenly left/right /input output height/width dimension input strides = 1. data_format string, either \"channels_last\" \"channels_first\". data_format determines ordering dimensions inputs. data_format = \"channels_last\", inputs shape (batch_size, ..., channels) data_format = \"channels_first\", inputs shape (batch_size, channels, ...). dilation_rate int int tuple/list len(inputs_spatial_shape), specifying dilation rate use dilated convolution. dilation_rate int, every spatial dimension shares dilation_rate.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_conv.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"General N-D convolution. — op_conv","text":"tensor rank N+2, result conv operation.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_conv_transpose.html","id":null,"dir":"Reference","previous_headings":"","what":"General N-D convolution transpose. — op_conv_transpose","title":"General N-D convolution transpose. — op_conv_transpose","text":"Also known de-convolution. ops supports 1D, 2D 3D convolution.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_conv_transpose.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"General N-D convolution transpose. — op_conv_transpose","text":"","code":"op_conv_transpose(   inputs,   kernel,   strides = 1L,   padding = \"valid\",   output_padding = NULL,   data_format = NULL,   dilation_rate = 1L )"},{"path":"https://keras3.posit.co/dev/reference/op_conv_transpose.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"General N-D convolution transpose. — op_conv_transpose","text":"inputs Tensor rank N+2. inputs shape (batch_size,) + inputs_spatial_shape + (num_channels,) data_format = \"channels_last\", (batch_size, num_channels) + inputs_spatial_shape data_format = \"channels_first\". kernel Tensor rank N+2. kernel shape [kernel_spatial_shape, num_output_channels, num_input_channels], num_input_channels match number channels inputs. strides int int tuple/list len(inputs_spatial_shape), specifying strides convolution along spatial dimension. strides int, every spatial dimension shares strides. padding string, either \"valid\" \"\". \"valid\" means padding applied, \"\" results padding evenly left/right /input output height/width dimension input strides = 1. output_padding int int tuple/list len(inputs_spatial_shape), specifying amount padding along height width output tensor. Can single integer specify value spatial dimensions. amount output padding along given dimension must lower stride along dimension. set NULL (default), output shape inferred. data_format string, either \"channels_last\" \"channels_first\". data_format determines ordering dimensions inputs. data_format = \"channels_last\", inputs shape (batch_size, ..., channels) data_format = \"channels_first\", inputs shape (batch_size, channels, ...). dilation_rate int int tuple/list len(inputs_spatial_shape), specifying dilation rate use dilated convolution. dilation_rate int, every spatial dimension shares dilation_rate.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_conv_transpose.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"General N-D convolution transpose. — op_conv_transpose","text":"tensor rank N+2, result conv operation.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_convert_to_numpy.html","id":null,"dir":"Reference","previous_headings":"","what":"Convert a tensor to an R or NumPy array. — op_convert_to_numpy","title":"Convert a tensor to an R or NumPy array. — op_convert_to_numpy","text":"Convert tensor R NumPy array.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_convert_to_numpy.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Convert a tensor to an R or NumPy array. — op_convert_to_numpy","text":"","code":"op_convert_to_numpy(x)  op_convert_to_array(x)"},{"path":"https://keras3.posit.co/dev/reference/op_convert_to_numpy.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Convert a tensor to an R or NumPy array. — op_convert_to_numpy","text":"x tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_convert_to_numpy.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Convert a tensor to an R or NumPy array. — op_convert_to_numpy","text":"NumPy array R array.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_convert_to_tensor.html","id":null,"dir":"Reference","previous_headings":"","what":"Convert an array to a tensor. — op_convert_to_tensor","title":"Convert an array to a tensor. — op_convert_to_tensor","text":"Native tensors current backend left unchanged unless dtype, sparse ragged arguments set.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_convert_to_tensor.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Convert an array to a tensor. — op_convert_to_tensor","text":"","code":"op_convert_to_tensor(x, dtype = NULL, sparse = NULL, ragged = NULL)"},{"path":"https://keras3.posit.co/dev/reference/op_convert_to_tensor.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Convert an array to a tensor. — op_convert_to_tensor","text":"x R NumPy array (can nested) backend tensor. dtype target type. NULL, type x used. sparse Whether keep sparse tensors. FALSE cause sparse tensors densified. default value NULL means sparse tensors kept backend supports . ragged Whether keep ragged tensors. FALSE cause ragged tensors densified. default value NULL means ragged tensors kept backend supports .","code":""},{"path":"https://keras3.posit.co/dev/reference/op_convert_to_tensor.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Convert an array to a tensor. — op_convert_to_tensor","text":"backend tensor specified dtype sparseness.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_convert_to_tensor.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Convert an array to a tensor. — op_convert_to_tensor","text":"","code":"x <- array(c(1, 2, 3)) y <- op_convert_to_tensor(x) y ## tf.Tensor([1. 2. 3.], shape=(3), dtype=float64) op_convert_to_tensor(c(1, 3, 2, 0), \"int32\") ## tf.Tensor([1 3 2 0], shape=(4), dtype=int32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_copy.html","id":null,"dir":"Reference","previous_headings":"","what":"Returns a copy of x. — op_copy","title":"Returns a copy of x. — op_copy","text":"Returns copy x.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_copy.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Returns a copy of x. — op_copy","text":"","code":"op_copy(x)"},{"path":"https://keras3.posit.co/dev/reference/op_copy.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Returns a copy of x. — op_copy","text":"x Input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_copy.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Returns a copy of x. — op_copy","text":"copy x.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_corrcoef.html","id":null,"dir":"Reference","previous_headings":"","what":"Compute the Pearson correlation coefficient matrix. — op_corrcoef","title":"Compute the Pearson correlation coefficient matrix. — op_corrcoef","text":"Compute Pearson correlation coefficient matrix.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_corrcoef.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Compute the Pearson correlation coefficient matrix. — op_corrcoef","text":"","code":"op_corrcoef(x)"},{"path":"https://keras3.posit.co/dev/reference/op_corrcoef.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Compute the Pearson correlation coefficient matrix. — op_corrcoef","text":"x 2D tensor shape (N, D), N number variables D number observations.Examples","code":"x <- op_array(matrix(c(1, 2, 3,                        2, 3, 4), nrow = 2, byrow = TRUE)) op_corrcoef(x) ## tf.Tensor( ## [[1. 1.] ##  [1. 1.]], shape=(2, 2), dtype=float64)"},{"path":"https://keras3.posit.co/dev/reference/op_corrcoef.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Compute the Pearson correlation coefficient matrix. — op_corrcoef","text":"tensor shape (N, N) representing correlation matrix.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_correlate.html","id":null,"dir":"Reference","previous_headings":"","what":"Compute the cross-correlation of two 1-dimensional tensors. — op_correlate","title":"Compute the cross-correlation of two 1-dimensional tensors. — op_correlate","text":"Compute cross-correlation two 1-dimensional tensors.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_correlate.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Compute the cross-correlation of two 1-dimensional tensors. — op_correlate","text":"","code":"op_correlate(x1, x2, mode = \"valid\")"},{"path":"https://keras3.posit.co/dev/reference/op_correlate.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Compute the cross-correlation of two 1-dimensional tensors. — op_correlate","text":"x1 First 1-dimensional input tensor length M. x2 Second 1-dimensional input tensor length N. mode Either \"valid\", \"\" \"full\". default mode set \"valid\", returns output length max(M, N) - min(M, N) + 1. \"\" returns output length max(M, N). \"full\" mode returns convolution point overlap, output length N+M-1.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_correlate.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Compute the cross-correlation of two 1-dimensional tensors. — op_correlate","text":"Output tensor, cross-correlation x1 x2.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_cos.html","id":null,"dir":"Reference","previous_headings":"","what":"Cosine, element-wise. — op_cos","title":"Cosine, element-wise. — op_cos","text":"Cosine, element-wise.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_cos.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Cosine, element-wise. — op_cos","text":"","code":"op_cos(x)"},{"path":"https://keras3.posit.co/dev/reference/op_cos.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Cosine, element-wise. — op_cos","text":"x Input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_cos.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Cosine, element-wise. — op_cos","text":"corresponding cosine values.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_cosh.html","id":null,"dir":"Reference","previous_headings":"","what":"Hyperbolic cosine, element-wise. — op_cosh","title":"Hyperbolic cosine, element-wise. — op_cosh","text":"Hyperbolic cosine, element-wise.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_cosh.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Hyperbolic cosine, element-wise. — op_cosh","text":"","code":"op_cosh(x)"},{"path":"https://keras3.posit.co/dev/reference/op_cosh.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Hyperbolic cosine, element-wise. — op_cosh","text":"x Input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_cosh.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Hyperbolic cosine, element-wise. — op_cosh","text":"Output tensor shape x.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_count_nonzero.html","id":null,"dir":"Reference","previous_headings":"","what":"Counts the number of non-zero values in x along the given axis. — op_count_nonzero","title":"Counts the number of non-zero values in x along the given axis. — op_count_nonzero","text":"axis specified non-zeros tensor counted.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_count_nonzero.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Counts the number of non-zero values in x along the given axis. — op_count_nonzero","text":"","code":"op_count_nonzero(x, axis = NULL)"},{"path":"https://keras3.posit.co/dev/reference/op_count_nonzero.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Counts the number of non-zero values in x along the given axis. — op_count_nonzero","text":"x Input tensor. axis Axis tuple axes along count number non-zeros. Defaults NULL.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_count_nonzero.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Counts the number of non-zero values in x along the given axis. — op_count_nonzero","text":"integer tensor integers.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_count_nonzero.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Counts the number of non-zero values in x along the given axis. — op_count_nonzero","text":"","code":"x <- op_array(rbind(c(0, 1, 7, 0),                    c(3, 0, 2, 19))) op_count_nonzero(x) ## tf.Tensor(5, shape=(), dtype=int32) op_count_nonzero(x, axis = 1) ## tf.Tensor([1 1 2 1], shape=(4), dtype=int32) op_count_nonzero(x, axis = 2) ## tf.Tensor([2 3], shape=(2), dtype=int32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_cross.html","id":null,"dir":"Reference","previous_headings":"","what":"Returns the cross product of two (arrays of) vectors. — op_cross","title":"Returns the cross product of two (arrays of) vectors. — op_cross","text":"cross product x1 x2 R^3 vector perpendicular x1 x2. x1 x2 arrays vectors, vectors defined last axis x1 x2 default, axes can dimensions 2 3. dimension either x1 x2 2, third component input vector assumed zero cross product calculated accordingly. cases input vectors dimension 2, z-component cross product returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_cross.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Returns the cross product of two (arrays of) vectors. — op_cross","text":"","code":"op_cross(x1, x2, axisa = -1L, axisb = -1L, axisc = -1L, axis = NULL)"},{"path":"https://keras3.posit.co/dev/reference/op_cross.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Returns the cross product of two (arrays of) vectors. — op_cross","text":"x1 Components first vector(s). x2 Components second vector(s). axisa Axis x1 defines vector(s). Defaults -1. axisb Axis x2 defines vector(s). Defaults -1. axisc Axis result containing cross product vector(s). Ignored input vectors dimension 2, return scalar. default, last axis. axis defined, axis x1, x2 result defines vector(s) cross product(s). Overrides axisa, axisb axisc.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_cross.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Returns the cross product of two (arrays of) vectors. — op_cross","text":"Vector cross product(s).","code":""},{"path":"https://keras3.posit.co/dev/reference/op_cross.html","id":"note","dir":"Reference","previous_headings":"","what":"Note","title":"Returns the cross product of two (arrays of) vectors. — op_cross","text":"Torch backend support two dimensional vectors, arguments axisa, axisb axisc. Use axis instead.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_ctc_decode.html","id":null,"dir":"Reference","previous_headings":"","what":"Decodes the output of a CTC model. — op_ctc_decode","title":"Decodes the output of a CTC model. — op_ctc_decode","text":"Decodes output CTC model.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_ctc_decode.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Decodes the output of a CTC model. — op_ctc_decode","text":"","code":"op_ctc_decode(   inputs,   sequence_lengths,   strategy = \"greedy\",   beam_width = 100L,   top_paths = 1L,   merge_repeated = TRUE,   mask_index = 1L )"},{"path":"https://keras3.posit.co/dev/reference/op_ctc_decode.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Decodes the output of a CTC model. — op_ctc_decode","text":"inputs tensor shape (batch_size, max_length, num_classes) containing logits (output model). normalized via softmax. sequence_lengths tensor shape (batch_size) containing sequence lengths batch. strategy string decoding strategy. Supported values \"greedy\" \"beam_search\". beam_width integer scalar beam width used beam search. Defaults 100. top_paths integer scalar, number top paths return. Defaults 1. merge_repeated boolean scalar, whether merge repeated labels output. Defaults TRUE. mask_index integer scalar, (1-based) index mask character vocabulary. Defaults 1.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_ctc_decode.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Decodes the output of a CTC model. — op_ctc_decode","text":"list containing: tensor representing list decoded sequences. strategy=\"greedy\", shape (1, batch_size, max_length). strategy=\"beam_search\", shape (top_paths, batch_size, max_length). Note : -1 indicates blank label. strategy=\"greedy\", tensor shape (batch_size, 1) representing negative sum probability logits sequence. strategy=\"beam_search\", tensor shape (batch_size, top_paths) representing log probability sequence.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_ctc_loss.html","id":null,"dir":"Reference","previous_headings":"","what":"CTC (Connectionist Temporal Classification) loss. — op_ctc_loss","title":"CTC (Connectionist Temporal Classification) loss. — op_ctc_loss","text":"CTC (Connectionist Temporal Classification) loss.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_ctc_loss.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"CTC (Connectionist Temporal Classification) loss. — op_ctc_loss","text":"","code":"op_ctc_loss(target, output, target_length, output_length, mask_index = 1L)"},{"path":"https://keras3.posit.co/dev/reference/op_ctc_loss.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"CTC (Connectionist Temporal Classification) loss. — op_ctc_loss","text":"target tensor shape (batch_size, max_length) containing true labels integer format. output tensor shape (batch_size, max_length, num_classes) containing logits (output model). target_length tensor shape (batch_size) containing true label lengths. output_length tensor shape (batch_size) containing output lengths. mask_index index mask character vocabulary. Defaults 1 (first element vocabulary).","code":""},{"path":"https://keras3.posit.co/dev/reference/op_ctc_loss.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"CTC (Connectionist Temporal Classification) loss. — op_ctc_loss","text":"tensor, shape (batch_size), loss values.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_cumprod.html","id":null,"dir":"Reference","previous_headings":"","what":"Return the cumulative product of elements along a given axis. — op_cumprod","title":"Return the cumulative product of elements along a given axis. — op_cumprod","text":"Return cumulative product elements along given axis.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_cumprod.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Return the cumulative product of elements along a given axis. — op_cumprod","text":"","code":"op_cumprod(x, axis = NULL, dtype = NULL)"},{"path":"https://keras3.posit.co/dev/reference/op_cumprod.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Return the cumulative product of elements along a given axis. — op_cumprod","text":"x Input tensor. axis Axis along cumulative product computed. default input flattened. dtype dtype returned tensor. Defaults x$dtype.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_cumprod.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Return the cumulative product of elements along a given axis. — op_cumprod","text":"Output tensor.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_cumsum.html","id":null,"dir":"Reference","previous_headings":"","what":"Returns the cumulative sum of elements along a given axis. — op_cumsum","title":"Returns the cumulative sum of elements along a given axis. — op_cumsum","text":"Returns cumulative sum elements along given axis.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_cumsum.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Returns the cumulative sum of elements along a given axis. — op_cumsum","text":"","code":"op_cumsum(x, axis = NULL, dtype = NULL)"},{"path":"https://keras3.posit.co/dev/reference/op_cumsum.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Returns the cumulative sum of elements along a given axis. — op_cumsum","text":"x Input tensor. axis Axis along cumulative sum computed. default input flattened. dtype dtype returned tensor. Defaults x$dtype.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_cumsum.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Returns the cumulative sum of elements along a given axis. — op_cumsum","text":"Output tensor.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_custom_gradient.html","id":null,"dir":"Reference","previous_headings":"","what":"Decorator to define a function with a custom gradient. — op_custom_gradient","title":"Decorator to define a function with a custom gradient. — op_custom_gradient","text":"decorator allows fine grained control gradients sequence operations. may useful multiple reasons, including providing efficient numerically stable gradient sequence operations.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_custom_gradient.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Decorator to define a function with a custom gradient. — op_custom_gradient","text":"","code":"op_custom_gradient(f)"},{"path":"https://keras3.posit.co/dev/reference/op_custom_gradient.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Decorator to define a function with a custom gradient. — op_custom_gradient","text":"f Function f(...) returns tuple (output, grad_fn) : ... sequence unnamed arguments, tensor input nested structure tensor inputs function. output (potentially nested structure ) tensor outputs applying operations forward_fn f() .... grad_fn function signature grad_fn(..., upstream) returns list tensors size (flattened) ...: derivatives tensors output respect tensors .... upstream tensor sequence tensors holding initial value gradients tensor output.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_custom_gradient.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Decorator to define a function with a custom gradient. — op_custom_gradient","text":"function h(...) returns value f(...)[[1]] whose gradient determined f(...)[[2]].","code":""},{"path":"https://keras3.posit.co/dev/reference/op_custom_gradient.html","id":"note","dir":"Reference","previous_headings":"","what":"Note","title":"Decorator to define a function with a custom gradient. — op_custom_gradient","text":"Note grad function returns gradient computation requires ... well upstream named argument, depending backend set. JAX TensorFlow backends, requires one argument, whereas might use upstream argument case PyTorch backend. working TensorFlow/JAX backend, grad(upstream) sufficient. PyTorch, grad function requires ... well upstream, e.g. grad <- \\(..., upstream). Follow example use op_custom_gradient() way compatible backends.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_custom_gradient.html","id":"example","dir":"Reference","previous_headings":"","what":"Example","title":"Decorator to define a function with a custom gradient. — op_custom_gradient","text":"Backend-agnostic example.","code":"log1pexp <- op_custom_gradient(\\(x) {      e <- op_exp(x)      grad <- function(..., upstream = NULL) {       upstream <- upstream %||% ..1       op_multiply(upstream, 1.0 - 1.0 / op_add(1, e))     }      tuple(op_log(1 + e), grad)  })  if(config_backend() == \"tensorflow\") {   tf <- tensorflow::tf   x <- op_convert_to_tensor(100.0)   with(tf$GradientTape() %as% tape, {     tape$watch(x)     y <- log1pexp(x)   })   dy_dx <- tape$gradient(y, x)   stopifnot(as.numeric(dy_dx) == 1) }"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_deg2rad.html","id":null,"dir":"Reference","previous_headings":"","what":"Convert angles from degrees to radians. — op_deg2rad","title":"Convert angles from degrees to radians. — op_deg2rad","text":"conversion defined : rad = deg * (pi / 180).","code":""},{"path":"https://keras3.posit.co/dev/reference/op_deg2rad.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Convert angles from degrees to radians. — op_deg2rad","text":"","code":"op_deg2rad(x)"},{"path":"https://keras3.posit.co/dev/reference/op_deg2rad.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Convert angles from degrees to radians. — op_deg2rad","text":"x Input tensor angles degrees.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_deg2rad.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Convert angles from degrees to radians. — op_deg2rad","text":"tensor containing angles converted radians.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_deg2rad.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Convert angles from degrees to radians. — op_deg2rad","text":"","code":"op_deg2rad(c(0, 90, 180)) ## tf.Tensor([0.        1.5707964 3.1415927], shape=(3), dtype=float32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_depthwise_conv.html","id":null,"dir":"Reference","previous_headings":"","what":"General N-D depthwise convolution. — op_depthwise_conv","title":"General N-D depthwise convolution. — op_depthwise_conv","text":"ops supports 1D 2D depthwise convolution.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_depthwise_conv.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"General N-D depthwise convolution. — op_depthwise_conv","text":"","code":"op_depthwise_conv(   inputs,   kernel,   strides = 1L,   padding = \"valid\",   data_format = NULL,   dilation_rate = 1L )"},{"path":"https://keras3.posit.co/dev/reference/op_depthwise_conv.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"General N-D depthwise convolution. — op_depthwise_conv","text":"inputs Tensor rank N+2. inputs shape (batch_size,) + inputs_spatial_shape + (num_channels,) data_format = \"channels_last\", (batch_size, num_channels) + inputs_spatial_shape data_format = \"channels_first\". kernel Tensor rank N+2. kernel shape [kernel_spatial_shape, num_input_channels, num_channels_multiplier], num_input_channels match number channels inputs. strides int int tuple/list len(inputs_spatial_shape), specifying strides convolution along spatial dimension. strides int, every spatial dimension shares strides. padding string, either \"valid\" \"\". \"valid\" means padding applied, \"\" results padding evenly left/right /input output height/width dimension input strides = 1. data_format string, either \"channels_last\" \"channels_first\". data_format determines ordering dimensions inputs. data_format = \"channels_last\", inputs shape (batch_size, ..., channels) data_format = \"channels_first\", inputs shape (batch_size, channels, ...). dilation_rate int int tuple/list len(inputs_spatial_shape), specifying dilation rate use dilated convolution. dilation_rate int, every spatial dimension shares dilation_rate.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_depthwise_conv.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"General N-D depthwise convolution. — op_depthwise_conv","text":"tensor rank N+2, result depthwise conv operation.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_det.html","id":null,"dir":"Reference","previous_headings":"","what":"Computes the determinant of a square tensor. — op_det","title":"Computes the determinant of a square tensor. — op_det","text":"Computes determinant square tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_det.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes the determinant of a square tensor. — op_det","text":"","code":"op_det(x)"},{"path":"https://keras3.posit.co/dev/reference/op_det.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Computes the determinant of a square tensor. — op_det","text":"x Input tensor shape (..., M, M).","code":""},{"path":"https://keras3.posit.co/dev/reference/op_det.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Computes the determinant of a square tensor. — op_det","text":"tensor shape (...) representing determinant x.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_diag.html","id":null,"dir":"Reference","previous_headings":"","what":"Extract a diagonal or construct a diagonal array. — op_diag","title":"Extract a diagonal or construct a diagonal array. — op_diag","text":"Extract diagonal construct diagonal array.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_diag.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Extract a diagonal or construct a diagonal array. — op_diag","text":"","code":"op_diag(x, k = 0L)"},{"path":"https://keras3.posit.co/dev/reference/op_diag.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Extract a diagonal or construct a diagonal array. — op_diag","text":"x Input tensor. x 2-D, returns k-th diagonal x. x 1-D, return 2-D tensor x k-th diagonal. k diagonal consider. Defaults 0. Use k > 0 diagonals main diagonal, k < 0 diagonals main diagonal.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_diag.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Extract a diagonal or construct a diagonal array. — op_diag","text":"extracted diagonal constructed diagonal tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_diag.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Extract a diagonal or construct a diagonal array. — op_diag","text":"","code":"x <- op_arange(9L) |> op_reshape(c(3, 3)) x ## tf.Tensor( ## [[1 2 3] ##  [4 5 6] ##  [7 8 9]], shape=(3, 3), dtype=int32) op_diag(x) ## tf.Tensor([1 5 9], shape=(3), dtype=int32) op_diag(x, k = 1) ## tf.Tensor([2 6], shape=(2), dtype=int32) op_diag(x, k = -1) ## tf.Tensor([4 8], shape=(2), dtype=int32) op_diag(op_diag(x)) ## tf.Tensor( ## [[1 0 0] ##  [0 5 0] ##  [0 0 9]], shape=(3, 3), dtype=int32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_diagflat.html","id":null,"dir":"Reference","previous_headings":"","what":"Create a two-dimensional array with the flattened input diagonal. — op_diagflat","title":"Create a two-dimensional array with the flattened input diagonal. — op_diagflat","text":"k-th diagonal.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_diagflat.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Create a two-dimensional array with the flattened input diagonal. — op_diagflat","text":"","code":"op_diagflat(x, k = 0L)"},{"path":"https://keras3.posit.co/dev/reference/op_diagflat.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Create a two-dimensional array with the flattened input diagonal. — op_diagflat","text":"x Input tensor flattened placed diagonal. k diagonal place flattened input. Defaults 0. Use k > 0 diagonals main diagonal, k < 0 diagonals main diagonal.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_diagflat.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Create a two-dimensional array with the flattened input diagonal. — op_diagflat","text":"2-D tensor flattened input specified diagonal.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_diagonal.html","id":null,"dir":"Reference","previous_headings":"","what":"Return specified diagonals. — op_diagonal","title":"Return specified diagonals. — op_diagonal","text":"x 2-D, returns diagonal x given offset, .e., collection elements form x[, +offset]. x two dimensions, axes specified axis1 axis2 used determine 2-D sub-array whose diagonal returned. shape resulting array can determined removing axis1 axis2 appending index right equal size resulting diagonals.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_diagonal.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Return specified diagonals. — op_diagonal","text":"","code":"op_diagonal(x, offset = 0L, axis1 = 1L, axis2 = 2L)"},{"path":"https://keras3.posit.co/dev/reference/op_diagonal.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Return specified diagonals. — op_diagonal","text":"x Input tensor. offset Offset diagonal main diagonal. Can positive negative. Defaults 0 (main diagonal). axis1 Axis used first axis 2-D sub-arrays. Defaults 1 (first axis). axis2 Axis used second axis 2-D sub-arrays. Defaults 2 (second axis).","code":""},{"path":"https://keras3.posit.co/dev/reference/op_diagonal.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Return specified diagonals. — op_diagonal","text":"Tensor diagonals.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_diagonal.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Return specified diagonals. — op_diagonal","text":"","code":"x <- op_arange(4L) |> op_reshape(c(2, 2)) x ## tf.Tensor( ## [[1 2] ##  [3 4]], shape=(2, 2), dtype=int32) op_diagonal(x) ## tf.Tensor([1 4], shape=(2), dtype=int32) op_diagonal(x, offset = 1) ## tf.Tensor([2], shape=(1), dtype=int32) x <- op_array(1:8) |> op_reshape(c(2, 2, 2)) x ## tf.Tensor( ## [[[1 2] ##   [3 4]] ## ##  [[5 6] ##   [7 8]]], shape=(2, 2, 2), dtype=int32) x |> op_diagonal(0) ## tf.Tensor( ## [[1 7] ##  [2 8]], shape=(2, 2), dtype=int32) x |> op_diagonal(0, 1, 2) # same as above, the default ## tf.Tensor( ## [[1 7] ##  [2 8]], shape=(2, 2), dtype=int32) x |> op_diagonal(0, 2, 3) ## tf.Tensor( ## [[1 4] ##  [5 8]], shape=(2, 2), dtype=int32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_diff.html","id":null,"dir":"Reference","previous_headings":"","what":"Calculate the n-th discrete difference along the given axis. — op_diff","title":"Calculate the n-th discrete difference along the given axis. — op_diff","text":"first difference given [] = [+1] - [] along given axis, higher differences calculated using diff recursively.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_diff.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Calculate the n-th discrete difference along the given axis. — op_diff","text":"","code":"op_diff(a, n = 1L, axis = -1L)"},{"path":"https://keras3.posit.co/dev/reference/op_diff.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Calculate the n-th discrete difference along the given axis. — op_diff","text":"Input tensor. n number times values differenced. Defaults 1. axis Axis compute discrete difference(s) along. Defaults -1 (last axis).","code":""},{"path":"https://keras3.posit.co/dev/reference/op_diff.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Calculate the n-th discrete difference along the given axis. — op_diff","text":"Tensor diagonals.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_diff.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Calculate the n-th discrete difference along the given axis. — op_diff","text":"","code":"x <- op_array(c(1, 2, 4, 7, 0)) op_diff(x) ## tf.Tensor([ 1.  2.  3. -7.], shape=(4), dtype=float32) op_diff(x, n = 2) ## tf.Tensor([  1.   1. -10.], shape=(3), dtype=float32) x <- op_array(rbind(c(1, 3, 6, 10),                     c(0, 5, 6, 8))) op_diff(x) ## tf.Tensor( ## [[2. 3. 4.] ##  [5. 1. 2.]], shape=(2, 3), dtype=float64) op_diff(x, axis = 1) ## tf.Tensor([[-1.  2.  0. -2.]], shape=(1, 4), dtype=float64)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_digitize.html","id":null,"dir":"Reference","previous_headings":"","what":"Returns the indices of the bins to which each value in x belongs. — op_digitize","title":"Returns the indices of the bins to which each value in x belongs. — op_digitize","text":"Returns indices bins value x belongs.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_digitize.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Returns the indices of the bins to which each value in x belongs. — op_digitize","text":"","code":"op_digitize(x, bins, zero_indexed = FALSE)"},{"path":"https://keras3.posit.co/dev/reference/op_digitize.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Returns the indices of the bins to which each value in x belongs. — op_digitize","text":"x Input array binned. bins Array bins. one-dimensional monotonically increasing. zero_indexed TRUE, returned indices zero-based (0 encodes first bin); FALSE (default), returned indices one-based (1 encodes first bin).","code":""},{"path":"https://keras3.posit.co/dev/reference/op_digitize.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Returns the indices of the bins to which each value in x belongs. — op_digitize","text":"Output array indices, shape x.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_digitize.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Returns the indices of the bins to which each value in x belongs. — op_digitize","text":"","code":"x <- op_array(c(-1,    0, 1,    3, 3.5,      4.5,  1.6,    8)) bins <- array(c(    0,       3,         4.5,            7)) op_digitize(x, bins) ## tf.Tensor([1 2 2 3 3 4 2 5], shape=(8), dtype=int32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_divide.html","id":null,"dir":"Reference","previous_headings":"","what":"Divide arguments element-wise. — op_divide","title":"Divide arguments element-wise. — op_divide","text":"Note function automatically called using R operator * tensor.","code":"(x <- op_arange(4)) ## tf.Tensor([1. 2. 3. 4.], shape=(4), dtype=float32) op_divide(x, 2) ## tf.Tensor([0.5 1.  1.5 2. ], shape=(4), dtype=float32) x / 2 ## tf.Tensor([0.5 1.  1.5 2. ], shape=(4), dtype=float32)"},{"path":"https://keras3.posit.co/dev/reference/op_divide.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Divide arguments element-wise. — op_divide","text":"","code":"op_divide(x1, x2)"},{"path":"https://keras3.posit.co/dev/reference/op_divide.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Divide arguments element-wise. — op_divide","text":"x1 First input tensor. x2 Second input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_divide.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Divide arguments element-wise. — op_divide","text":"Output tensor, quotient x1/x2, element-wise.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_divide.html","id":"example","dir":"Reference","previous_headings":"","what":"Example","title":"Divide arguments element-wise. — op_divide","text":"","code":"op_divide(3, 2) ## tf.Tensor(1.5, shape=(), dtype=float32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_divide_no_nan.html","id":null,"dir":"Reference","previous_headings":"","what":"Safe element-wise division which returns 0 where the denominator is 0. — op_divide_no_nan","title":"Safe element-wise division which returns 0 where the denominator is 0. — op_divide_no_nan","text":"Safe element-wise division returns 0 denominator 0.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_divide_no_nan.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Safe element-wise division which returns 0 where the denominator is 0. — op_divide_no_nan","text":"","code":"op_divide_no_nan(x1, x2)"},{"path":"https://keras3.posit.co/dev/reference/op_divide_no_nan.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Safe element-wise division which returns 0 where the denominator is 0. — op_divide_no_nan","text":"x1 First input tensor. x2 Second input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_divide_no_nan.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Safe element-wise division which returns 0 where the denominator is 0. — op_divide_no_nan","text":"quotient x1/x2, element-wise, zero x2 zero.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_dot.html","id":null,"dir":"Reference","previous_headings":"","what":"Dot product of two tensors. — op_dot","title":"Dot product of two tensors. — op_dot","text":"x1 x2 1-D tensors, inner product vectors (without complex conjugation). x1 x2 2-D tensors, matrix multiplication. either x1 x2 0-D (scalar), equivalent x1 * x2. x1 N-D tensor x2 1-D tensor, sum product last axis x1 x2. x1 N-D tensor x2 M-D tensor (M >= 2), sum product last axis x1 second--last axis x2: dot(x1, x2)[,j,k,m] = sum([,j,:] * b[k,:,m]).","code":""},{"path":"https://keras3.posit.co/dev/reference/op_dot.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Dot product of two tensors. — op_dot","text":"","code":"op_dot(x1, x2)"},{"path":"https://keras3.posit.co/dev/reference/op_dot.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Dot product of two tensors. — op_dot","text":"x1 First argument. x2 Second argument.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_dot.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Dot product of two tensors. — op_dot","text":"Dot product x1 x2.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_dot.html","id":"note","dir":"Reference","previous_headings":"","what":"Note","title":"Dot product of two tensors. — op_dot","text":"Torch backend accept 0-D tensors arguments.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_dot_product_attention.html","id":null,"dir":"Reference","previous_headings":"","what":"Scaled dot product attention function. — op_dot_product_attention","title":"Scaled dot product attention function. — op_dot_product_attention","text":"Computes attention function Q (query), K (key), V(value): attention(Q, K, V) = softmax(Q * K / sqrt(d)) * V. define logits output Q * K probs output softmax. Throughout function, utilize following notation represent shape array: B: batch size S: length key/value T: length query N: number attention heads H: dimensions attention head K: number key/value heads G: number groups, equals N // K","code":""},{"path":"https://keras3.posit.co/dev/reference/op_dot_product_attention.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Scaled dot product attention function. — op_dot_product_attention","text":"","code":"op_dot_product_attention(   query,   key,   value,   bias = NULL,   mask = NULL,   scale = NULL,   is_causal = FALSE,   flash_attention = NULL,   attn_logits_soft_cap = NULL )"},{"path":"https://keras3.posit.co/dev/reference/op_dot_product_attention.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Scaled dot product attention function. — op_dot_product_attention","text":"query query array shape (B, T, N, H). key key array shape (B, S, K, H). K equals N, multi-headed attention (MHA) performed. Otherwise, grouped query attention (GQA) performed N multiple K. multi-query attention (MQA) performed K==1 (special case GQA). value value array shape key. bias Optional bias array added logits. shape must broadcastable (B, N, T, S). mask Optional mask array used filter logits. boolean mask TRUE indicates element take part attention. additive mask, users pass bias. shape must broadcastable (B, N, T, S). scale Optional scale logits. NULL, scale set 1.0 / sqrt(H). is_causal Whether apply causal mask. flash_attention Whether use flash attention. NULL, attempt use flash attention required conditions met. Typically, inputs must float16 bfloat16 dtype input layout requirements may vary depending backend. attn_logits_soft_cap Optional numeric cap attention logits softmax. supported JAX TPU backend.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_dot_product_attention.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Scaled dot product attention function. — op_dot_product_attention","text":"array attention output shape query.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_dot_product_attention.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Scaled dot product attention function. — op_dot_product_attention","text":"","code":"query = random_normal(c(2, 4, 8, 16)) key = random_normal(c(2, 6, 8, 16)) value = random_normal(c(2, 6, 8, 16)) op_dot_product_attention(query, key, value) |> op_shape() ## shape(2, 4, 8, 16)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_dtype.html","id":null,"dir":"Reference","previous_headings":"","what":"Return the dtype of the tensor input as a standardized string. — op_dtype","title":"Return the dtype of the tensor input as a standardized string. — op_dtype","text":"Note due standardization, dtype compare equal backend-specific version dtype.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_dtype.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Return the dtype of the tensor input as a standardized string. — op_dtype","text":"","code":"op_dtype(x)"},{"path":"https://keras3.posit.co/dev/reference/op_dtype.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Return the dtype of the tensor input as a standardized string. — op_dtype","text":"x tensor. function try access dtype attribute input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_dtype.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Return the dtype of the tensor input as a standardized string. — op_dtype","text":"string indicating dtype input tensor, e.g. \"float32\".","code":""},{"path":"https://keras3.posit.co/dev/reference/op_dtype.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Return the dtype of the tensor input as a standardized string. — op_dtype","text":"","code":"x <- op_zeros(c(8, 12)) op_dtype(x) ## [1] \"float32\""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_eig.html","id":null,"dir":"Reference","previous_headings":"","what":"Computes the eigenvalues and eigenvectors of a square matrix. — op_eig","title":"Computes the eigenvalues and eigenvectors of a square matrix. — op_eig","text":"Computes eigenvalues eigenvectors square matrix.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_eig.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes the eigenvalues and eigenvectors of a square matrix. — op_eig","text":"","code":"op_eig(x)"},{"path":"https://keras3.posit.co/dev/reference/op_eig.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Computes the eigenvalues and eigenvectors of a square matrix. — op_eig","text":"x Input tensor shape (..., M, M).","code":""},{"path":"https://keras3.posit.co/dev/reference/op_eig.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Computes the eigenvalues and eigenvectors of a square matrix. — op_eig","text":"list two tensors: tensor shape (..., M) containing eigenvalues tensor shape (..., M, M) containing eigenvectors.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_eigh.html","id":null,"dir":"Reference","previous_headings":"","what":"Computes the eigenvalues and eigenvectors of a complex Hermitian. — op_eigh","title":"Computes the eigenvalues and eigenvectors of a complex Hermitian. — op_eigh","text":"Computes eigenvalues eigenvectors complex Hermitian.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_eigh.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes the eigenvalues and eigenvectors of a complex Hermitian. — op_eigh","text":"","code":"op_eigh(x)"},{"path":"https://keras3.posit.co/dev/reference/op_eigh.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Computes the eigenvalues and eigenvectors of a complex Hermitian. — op_eigh","text":"x Input tensor shape (..., M, M).","code":""},{"path":"https://keras3.posit.co/dev/reference/op_eigh.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Computes the eigenvalues and eigenvectors of a complex Hermitian. — op_eigh","text":"list two tensors: tensor shape (..., M) containing eigenvalues tensor shape (..., M, M) containing eigenvectors.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_einsum.html","id":null,"dir":"Reference","previous_headings":"","what":"Evaluates the Einstein summation convention on the operands. — op_einsum","title":"Evaluates the Einstein summation convention on the operands. — op_einsum","text":"Evaluates Einstein summation convention operands.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_einsum.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Evaluates the Einstein summation convention on the operands. — op_einsum","text":"","code":"op_einsum(subscripts, ...)"},{"path":"https://keras3.posit.co/dev/reference/op_einsum.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Evaluates the Einstein summation convention on the operands. — op_einsum","text":"subscripts Specifies subscripts summation comma separated list subscript labels. implicit (classical Einstein summation) calculation performed unless explicit indicator -> included well subscript labels precise output form. ... operands compute Einstein sum unnamed arguments. Additional named arguments forwarded underlying backend.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_einsum.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Evaluates the Einstein summation convention on the operands. — op_einsum","text":"calculation based Einstein summation convention.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_einsum.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Evaluates the Einstein summation convention on the operands. — op_einsum","text":"Trace matrix:     Extract diagonal:     Sum axis:     higher dimensional tensors summing single axis can done ellipsis:     Compute matrix transpose reorder number axes:         Matrix vector multiplication:","code":"a <- op_arange(25) |> op_reshape(c(5, 5)) b <- op_arange(5) c <- op_arange(6) |> op_reshape(c(2, 3)) op_einsum(\"ii\", a) op_trace(a) ## tf.Tensor(65.0, shape=(), dtype=float32) ## tf.Tensor(65.0, shape=(), dtype=float32) op_einsum(\"ii -> i\", a) op_diag(a) ## tf.Tensor([ 1.  7. 13. 19. 25.], shape=(5), dtype=float32) ## tf.Tensor([ 1.  7. 13. 19. 25.], shape=(5), dtype=float32) op_einsum(\"ij -> i\", a) op_sum(a, axis = 2) ## tf.Tensor([ 15.  40.  65.  90. 115.], shape=(5), dtype=float32) ## tf.Tensor([ 15.  40.  65.  90. 115.], shape=(5), dtype=float32) op_einsum(\"...j -> ...\", a) op_sum(a, axis = -1) ## tf.Tensor([ 15.  40.  65.  90. 115.], shape=(5), dtype=float32) ## tf.Tensor([ 15.  40.  65.  90. 115.], shape=(5), dtype=float32) op_einsum(\"ji\", c) # return c unchanged ## tf.Tensor( ## [[1. 2. 3.] ##  [4. 5. 6.]], shape=(2, 3), dtype=float32) op_einsum(\"ij -> ji\", c) # transpose op_transpose(c)          # same as above ## tf.Tensor( ## [[1. 4.] ##  [2. 5.] ##  [3. 6.]], shape=(3, 2), dtype=float32) ## tf.Tensor( ## [[1. 4.] ##  [2. 5.] ##  [3. 6.]], shape=(3, 2), dtype=float32) op_einsum(\"ij, j\", a, b) op_einsum(\"...j, j\", a, b) a %*% b op_matmul(a, b) ## tf.Tensor([ 55. 130. 205. 280. 355.], shape=(5), dtype=float32) ## tf.Tensor([ 55. 130. 205. 280. 355.], shape=(5), dtype=float32) ## tf.Tensor([ 55. 130. 205. 280. 355.], shape=(5), dtype=float32) ## tf.Tensor([ 55. 130. 205. 280. 355.], shape=(5), dtype=float32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_elu.html","id":null,"dir":"Reference","previous_headings":"","what":"Exponential Linear Unit activation function. — op_elu","title":"Exponential Linear Unit activation function. — op_elu","text":"defined : f(x) =  alpha * (exp(x) - 1.) x < 0, f(x) = x x >= 0.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_elu.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Exponential Linear Unit activation function. — op_elu","text":"","code":"op_elu(x, alpha = 1)"},{"path":"https://keras3.posit.co/dev/reference/op_elu.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Exponential Linear Unit activation function. — op_elu","text":"x Input tensor. alpha scalar, slope positive section. Defaults 1.0.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_elu.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Exponential Linear Unit activation function. — op_elu","text":"tensor shape x.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_elu.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Exponential Linear Unit activation function. — op_elu","text":"","code":"x <- op_array(c(-1., 0., 1.)) op_elu(x) ## tf.Tensor([-0.63212055  0.          1.        ], shape=(3), dtype=float32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_empty.html","id":null,"dir":"Reference","previous_headings":"","what":"Return a tensor of given shape and type filled with uninitialized data. — op_empty","title":"Return a tensor of given shape and type filled with uninitialized data. — op_empty","text":"Return tensor given shape type filled uninitialized data.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_empty.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Return a tensor of given shape and type filled with uninitialized data. — op_empty","text":"","code":"op_empty(shape, dtype = NULL)"},{"path":"https://keras3.posit.co/dev/reference/op_empty.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Return a tensor of given shape and type filled with uninitialized data. — op_empty","text":"shape Shape empty tensor. dtype Desired data type empty tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_empty.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Return a tensor of given shape and type filled with uninitialized data. — op_empty","text":"empty tensor.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_equal.html","id":null,"dir":"Reference","previous_headings":"","what":"Returns (x1 == x2) element-wise. — op_equal","title":"Returns (x1 == x2) element-wise. — op_equal","text":"Note function automatically called using R operator == tensor.","code":"(x <- op_arange(4)) ## tf.Tensor([1. 2. 3. 4.], shape=(4), dtype=float32) op_equal(x, 2) ## tf.Tensor([False  True False False], shape=(4), dtype=bool) x == 2 ## tf.Tensor([False  True False False], shape=(4), dtype=bool)"},{"path":"https://keras3.posit.co/dev/reference/op_equal.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Returns (x1 == x2) element-wise. — op_equal","text":"","code":"op_equal(x1, x2)"},{"path":"https://keras3.posit.co/dev/reference/op_equal.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Returns (x1 == x2) element-wise. — op_equal","text":"x1 Tensor compare. x2 Tensor compare.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_equal.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Returns (x1 == x2) element-wise. — op_equal","text":"Output tensor, element-wise comparison x1 x2.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_erf.html","id":null,"dir":"Reference","previous_headings":"","what":"Computes the error function of x, element-wise. — op_erf","title":"Computes the error function of x, element-wise. — op_erf","text":"Computes error function x, element-wise.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_erf.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes the error function of x, element-wise. — op_erf","text":"","code":"op_erf(x)"},{"path":"https://keras3.posit.co/dev/reference/op_erf.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Computes the error function of x, element-wise. — op_erf","text":"x Input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_erf.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Computes the error function of x, element-wise. — op_erf","text":"tensor dtype x.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_erf.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Computes the error function of x, element-wise. — op_erf","text":"","code":"x <- op_array(c(-3, -2, -1, 0, 1)) op_erf(x) ## tf.Tensor([-0.999978   -0.99532235 -0.84270084  0.          0.84270084], shape=(5), dtype=float32) # array([-0.99998 , -0.99532, -0.842701,  0.,  0.842701], dtype=float32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_erfinv.html","id":null,"dir":"Reference","previous_headings":"","what":"Computes the inverse error function of x, element-wise. — op_erfinv","title":"Computes the inverse error function of x, element-wise. — op_erfinv","text":"Computes inverse error function x, element-wise.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_erfinv.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes the inverse error function of x, element-wise. — op_erfinv","text":"","code":"op_erfinv(x)"},{"path":"https://keras3.posit.co/dev/reference/op_erfinv.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Computes the inverse error function of x, element-wise. — op_erfinv","text":"x Input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_erfinv.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Computes the inverse error function of x, element-wise. — op_erfinv","text":"tensor dtype x.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_erfinv.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Computes the inverse error function of x, element-wise. — op_erfinv","text":"","code":"x <- op_array(c(-0.5, -0.2, -0.1, 0.0, 0.3)) op_erfinv(x) ## tf.Tensor([-0.4769363  -0.17914344 -0.088856    0.          0.27246267], shape=(5), dtype=float32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_exp.html","id":null,"dir":"Reference","previous_headings":"","what":"Calculate the exponential of all elements in the input tensor. — op_exp","title":"Calculate the exponential of all elements in the input tensor. — op_exp","text":"Calculate exponential elements input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_exp.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Calculate the exponential of all elements in the input tensor. — op_exp","text":"","code":"op_exp(x)"},{"path":"https://keras3.posit.co/dev/reference/op_exp.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Calculate the exponential of all elements in the input tensor. — op_exp","text":"x Input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_exp.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Calculate the exponential of all elements in the input tensor. — op_exp","text":"Output tensor, element-wise exponential x.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_exp2.html","id":null,"dir":"Reference","previous_headings":"","what":"Calculate the base-2 exponential of all elements in the input tensor. — op_exp2","title":"Calculate the base-2 exponential of all elements in the input tensor. — op_exp2","text":"Calculate base-2 exponential elements input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_exp2.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Calculate the base-2 exponential of all elements in the input tensor. — op_exp2","text":"","code":"op_exp2(x)"},{"path":"https://keras3.posit.co/dev/reference/op_exp2.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Calculate the base-2 exponential of all elements in the input tensor. — op_exp2","text":"x Input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_exp2.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Calculate the base-2 exponential of all elements in the input tensor. — op_exp2","text":"Output tensor, element-wise base-2 exponential x.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_expand_dims.html","id":null,"dir":"Reference","previous_headings":"","what":"Expand the shape of a tensor. — op_expand_dims","title":"Expand the shape of a tensor. — op_expand_dims","text":"Insert new axis axis position expanded tensor shape.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_expand_dims.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Expand the shape of a tensor. — op_expand_dims","text":"","code":"op_expand_dims(x, axis)"},{"path":"https://keras3.posit.co/dev/reference/op_expand_dims.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Expand the shape of a tensor. — op_expand_dims","text":"x Input tensor. axis Position expanded axes new axis (axes) placed.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_expand_dims.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Expand the shape of a tensor. — op_expand_dims","text":"Output tensor number dimensions increased.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_expm1.html","id":null,"dir":"Reference","previous_headings":"","what":"Calculate exp(x) - 1 for all elements in the tensor. — op_expm1","title":"Calculate exp(x) - 1 for all elements in the tensor. — op_expm1","text":"Calculate exp(x) - 1 elements tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_expm1.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Calculate exp(x) - 1 for all elements in the tensor. — op_expm1","text":"","code":"op_expm1(x)"},{"path":"https://keras3.posit.co/dev/reference/op_expm1.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Calculate exp(x) - 1 for all elements in the tensor. — op_expm1","text":"x Input values.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_expm1.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Calculate exp(x) - 1 for all elements in the tensor. — op_expm1","text":"Output tensor, element-wise exponential minus one.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_extract_sequences.html","id":null,"dir":"Reference","previous_headings":"","what":"Expands the dimension of last axis into sequences of sequence_length. — op_extract_sequences","title":"Expands the dimension of last axis into sequences of sequence_length. — op_extract_sequences","text":"Slides window size sequence_length last axis input stride sequence_stride, replacing last axis [num_sequences, sequence_length] sequences. dimension along last axis N, number sequences can computed : num_sequences = 1 + (N - sequence_length) // sequence_stride","code":""},{"path":"https://keras3.posit.co/dev/reference/op_extract_sequences.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Expands the dimension of last axis into sequences of sequence_length. — op_extract_sequences","text":"","code":"op_extract_sequences(x, sequence_length, sequence_stride)"},{"path":"https://keras3.posit.co/dev/reference/op_extract_sequences.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Expands the dimension of last axis into sequences of sequence_length. — op_extract_sequences","text":"x Input tensor. sequence_length integer representing sequences length. sequence_stride integer representing sequences hop size.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_extract_sequences.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Expands the dimension of last axis into sequences of sequence_length. — op_extract_sequences","text":"tensor sequences shape [..., num_sequences, sequence_length].","code":""},{"path":"https://keras3.posit.co/dev/reference/op_extract_sequences.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Expands the dimension of last axis into sequences of sequence_length. — op_extract_sequences","text":"","code":"x <- op_convert_to_tensor(1:6) op_extract_sequences(x, 3, 2) ## tf.Tensor( ## [[1 2 3] ##  [3 4 5]], shape=(2, 3), dtype=int32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_eye.html","id":null,"dir":"Reference","previous_headings":"","what":"Return a 2-D tensor with ones on the diagonal and zeros elsewhere. — op_eye","title":"Return a 2-D tensor with ones on the diagonal and zeros elsewhere. — op_eye","text":"Return 2-D tensor ones diagonal zeros elsewhere.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_eye.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Return a 2-D tensor with ones on the diagonal and zeros elsewhere. — op_eye","text":"","code":"op_eye(N, M = NULL, k = 0L, dtype = NULL)"},{"path":"https://keras3.posit.co/dev/reference/op_eye.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Return a 2-D tensor with ones on the diagonal and zeros elsewhere. — op_eye","text":"N Number rows output. M Number columns output. NULL, defaults N. k Index diagonal: 0 (default) refers main diagonal, positive value refers upper diagonal, negative value lower diagonal. dtype Data type returned tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_eye.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Return a 2-D tensor with ones on the diagonal and zeros elsewhere. — op_eye","text":"Tensor ones k-th diagonal zeros elsewhere.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_fft.html","id":null,"dir":"Reference","previous_headings":"","what":"Computes the Fast Fourier Transform along last axis of input. — op_fft","title":"Computes the Fast Fourier Transform along last axis of input. — op_fft","text":"Computes Fast Fourier Transform along last axis input.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_fft.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes the Fast Fourier Transform along last axis of input. — op_fft","text":"","code":"op_fft(x)"},{"path":"https://keras3.posit.co/dev/reference/op_fft.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Computes the Fast Fourier Transform along last axis of input. — op_fft","text":"x list real imaginary parts input tensor. tensors provided floating type.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_fft.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Computes the Fast Fourier Transform along last axis of input. — op_fft","text":"list containing two tensors - real imaginary parts output tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_fft.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Computes the Fast Fourier Transform along last axis of input. — op_fft","text":"","code":"x = c(op_array(c(1., 2.)),       op_array(c(0., 1.))) op_fft(x) ## [[1]] ## tf.Tensor([ 3. -1.], shape=(2), dtype=float32) ## ## [[2]] ## tf.Tensor([ 1. -1.], shape=(2), dtype=float32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_fft2.html","id":null,"dir":"Reference","previous_headings":"","what":"Computes the 2D Fast Fourier Transform along the last two axes of input. — op_fft2","title":"Computes the 2D Fast Fourier Transform along the last two axes of input. — op_fft2","text":"Computes 2D Fast Fourier Transform along last two axes input.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_fft2.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes the 2D Fast Fourier Transform along the last two axes of input. — op_fft2","text":"","code":"op_fft2(x)"},{"path":"https://keras3.posit.co/dev/reference/op_fft2.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Computes the 2D Fast Fourier Transform along the last two axes of input. — op_fft2","text":"x list real imaginary parts input tensor. tensors provided floating type.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_fft2.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Computes the 2D Fast Fourier Transform along the last two axes of input. — op_fft2","text":"list containing two tensors - real imaginary parts output.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_fft2.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Computes the 2D Fast Fourier Transform along the last two axes of input. — op_fft2","text":"","code":"x <- c(op_array(rbind(c(1, 2),                      c(2, 1))),        op_array(rbind(c(0, 1),                      c(1, 0)))) op_fft2(x) ## [[1]] ## tf.Tensor( ## [[ 6.  0.] ##  [ 0. -2.]], shape=(2, 2), dtype=float64) ## ## [[2]] ## tf.Tensor( ## [[ 2.  0.] ##  [ 0. -2.]], shape=(2, 2), dtype=float64)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_flip.html","id":null,"dir":"Reference","previous_headings":"","what":"Reverse the order of elements in the tensor along the given axis. — op_flip","title":"Reverse the order of elements in the tensor along the given axis. — op_flip","text":"shape tensor preserved, elements reordered.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_flip.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Reverse the order of elements in the tensor along the given axis. — op_flip","text":"","code":"op_flip(x, axis = NULL)"},{"path":"https://keras3.posit.co/dev/reference/op_flip.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Reverse the order of elements in the tensor along the given axis. — op_flip","text":"x Input tensor. axis Axis axes along flip tensor. default, axis = NULL, flip axes input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_flip.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Reverse the order of elements in the tensor along the given axis. — op_flip","text":"Output tensor entries axis reversed.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_floor.html","id":null,"dir":"Reference","previous_headings":"","what":"Return the floor of the input, element-wise. — op_floor","title":"Return the floor of the input, element-wise. — op_floor","text":"floor scalar x largest integer , <= x.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_floor.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Return the floor of the input, element-wise. — op_floor","text":"","code":"op_floor(x)"},{"path":"https://keras3.posit.co/dev/reference/op_floor.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Return the floor of the input, element-wise. — op_floor","text":"x Input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_floor.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Return the floor of the input, element-wise. — op_floor","text":"Output tensor, element-wise floor x.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_floor_divide.html","id":null,"dir":"Reference","previous_headings":"","what":"Returns the largest integer smaller or equal to the division of inputs. — op_floor_divide","title":"Returns the largest integer smaller or equal to the division of inputs. — op_floor_divide","text":"Note function automatically called using R operator %/% tensor.","code":"(x <- op_arange(10)) ## tf.Tensor([ 1.  2.  3.  4.  5.  6.  7.  8.  9. 10.], shape=(10), dtype=float32) op_floor_divide(x, 2) ## tf.Tensor([0. 1. 1. 2. 2. 3. 3. 4. 4. 5.], shape=(10), dtype=float32) x %/% 2 ## tf.Tensor([0. 1. 1. 2. 2. 3. 3. 4. 4. 5.], shape=(10), dtype=float32)"},{"path":"https://keras3.posit.co/dev/reference/op_floor_divide.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Returns the largest integer smaller or equal to the division of inputs. — op_floor_divide","text":"","code":"op_floor_divide(x1, x2)"},{"path":"https://keras3.posit.co/dev/reference/op_floor_divide.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Returns the largest integer smaller or equal to the division of inputs. — op_floor_divide","text":"x1 Numerator. x2 Denominator.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_floor_divide.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Returns the largest integer smaller or equal to the division of inputs. — op_floor_divide","text":"Output tensor, y <- floor(x1/x2)","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_fori_loop.html","id":null,"dir":"Reference","previous_headings":"","what":"For loop implementation. — op_fori_loop","title":"For loop implementation. — op_fori_loop","text":"loop implementation.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_fori_loop.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"For loop implementation. — op_fori_loop","text":"","code":"op_fori_loop(lower, upper, body_fun, init_val)"},{"path":"https://keras3.posit.co/dev/reference/op_fori_loop.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"For loop implementation. — op_fori_loop","text":"lower initial value loop variable. upper upper bound loop variable. body_fun callable represents loop body. Must take two arguments: loop variable loop state. loop state updated returned function. init_val initial value loop state.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_fori_loop.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"For loop implementation. — op_fori_loop","text":"final state loop.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_fori_loop.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"For loop implementation. — op_fori_loop","text":"","code":"lower <- 0L upper <- 10L body_fun <- function(i, state) state + i init_state <- 0L final_state <- op_fori_loop(lower, upper, body_fun, init_state) final_state ## tf.Tensor(45, shape=(), dtype=int32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_full.html","id":null,"dir":"Reference","previous_headings":"","what":"Return a new tensor of given shape and type, filled with fill_value. — op_full","title":"Return a new tensor of given shape and type, filled with fill_value. — op_full","text":"Return new tensor given shape type, filled fill_value.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_full.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Return a new tensor of given shape and type, filled with fill_value. — op_full","text":"","code":"op_full(shape, fill_value, dtype = NULL)"},{"path":"https://keras3.posit.co/dev/reference/op_full.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Return a new tensor of given shape and type, filled with fill_value. — op_full","text":"shape Shape new tensor. fill_value Fill value. dtype Desired data type tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_full.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Return a new tensor of given shape and type, filled with fill_value. — op_full","text":"Output tensor.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_full_like.html","id":null,"dir":"Reference","previous_headings":"","what":"Return a full tensor with the same shape and type as the given tensor. — op_full_like","title":"Return a full tensor with the same shape and type as the given tensor. — op_full_like","text":"Return full tensor shape type given tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_full_like.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Return a full tensor with the same shape and type as the given tensor. — op_full_like","text":"","code":"op_full_like(x, fill_value, dtype = NULL)"},{"path":"https://keras3.posit.co/dev/reference/op_full_like.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Return a full tensor with the same shape and type as the given tensor. — op_full_like","text":"x Input tensor. fill_value Fill value. dtype Overrides data type result.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_full_like.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Return a full tensor with the same shape and type as the given tensor. — op_full_like","text":"Tensor fill_value shape type x.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_gelu.html","id":null,"dir":"Reference","previous_headings":"","what":"Gaussian Error Linear Unit (GELU) activation function. — op_gelu","title":"Gaussian Error Linear Unit (GELU) activation function. — op_gelu","text":"approximate TRUE, defined : f(x) = 0.5 * x * (1 + tanh(sqrt(2 / pi) * (x + 0.044715 * x^3))) approximate FALSE, defined : f(x) = x * P(X <= x) = 0.5 * x * (1 + erf(x / sqrt(2))), P(X) ~ N(0, 1).","code":""},{"path":"https://keras3.posit.co/dev/reference/op_gelu.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Gaussian Error Linear Unit (GELU) activation function. — op_gelu","text":"","code":"op_gelu(x, approximate = TRUE)"},{"path":"https://keras3.posit.co/dev/reference/op_gelu.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Gaussian Error Linear Unit (GELU) activation function. — op_gelu","text":"x Input tensor. approximate Approximate version GELU activation. Defaults TRUE.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_gelu.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Gaussian Error Linear Unit (GELU) activation function. — op_gelu","text":"tensor shape x.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_gelu.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Gaussian Error Linear Unit (GELU) activation function. — op_gelu","text":"","code":"x <- op_array(c(-1., 0., 1.)) op_gelu(x) ## tf.Tensor([-0.158808  0.        0.841192], shape=(3), dtype=float32) op_gelu(x, FALSE) ## tf.Tensor([-0.15865526  0.          0.8413447 ], shape=(3), dtype=float32) x <- seq(-5, 5, .1) plot(x, op_gelu(x),      type = \"l\", #, frame.plot = FALSE,      panel.first = grid())"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_get_item.html","id":null,"dir":"Reference","previous_headings":"","what":"Return x[key]. — op_get_item","title":"Return x[key]. — op_get_item","text":"Return x[key].","code":""},{"path":"https://keras3.posit.co/dev/reference/op_get_item.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Return x[key]. — op_get_item","text":"","code":"op_get_item(x, key)"},{"path":"https://keras3.posit.co/dev/reference/op_get_item.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Return x[key]. — op_get_item","text":"x dictionary-like object key Generally, string, object __hash__ method acceptable.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_get_item.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Return x[key]. — op_get_item","text":"key.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_get_item.html","id":"note","dir":"Reference","previous_headings":"","what":"Note","title":"Return x[key]. — op_get_item","text":"Generally, calling x[[key]] x$key preferable.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_glu.html","id":null,"dir":"Reference","previous_headings":"","what":"Gated Linear Unit (GLU) activation function. — op_glu","title":"Gated Linear Unit (GLU) activation function. — op_glu","text":"defined : f(x) = * sigmoid(b) x split b along given axis.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_glu.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Gated Linear Unit (GLU) activation function. — op_glu","text":"","code":"op_glu(x, axis = -1L)"},{"path":"https://keras3.posit.co/dev/reference/op_glu.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Gated Linear Unit (GLU) activation function. — op_glu","text":"x Input tensor. axis axis along split input tensor. Defaults -1.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_glu.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Gated Linear Unit (GLU) activation function. — op_glu","text":"tensor shape half input.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_glu.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Gated Linear Unit (GLU) activation function. — op_glu","text":"","code":"x <- op_array(c(-1., 0., 1. , 1.)) op_glu(x) ## tf.Tensor([-0.7310586  0.       ], shape=(2), dtype=float32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_greater.html","id":null,"dir":"Reference","previous_headings":"","what":"Return the truth value of x1 > x2 element-wise. — op_greater","title":"Return the truth value of x1 > x2 element-wise. — op_greater","text":"Note function automatically called using R operator > tensor.","code":"(x <- op_arange(4)) ## tf.Tensor([1. 2. 3. 4.], shape=(4), dtype=float32) op_greater(x, 2) ## tf.Tensor([False False  True  True], shape=(4), dtype=bool) x > 2 ## tf.Tensor([False False  True  True], shape=(4), dtype=bool)"},{"path":"https://keras3.posit.co/dev/reference/op_greater.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Return the truth value of x1 > x2 element-wise. — op_greater","text":"","code":"op_greater(x1, x2)"},{"path":"https://keras3.posit.co/dev/reference/op_greater.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Return the truth value of x1 > x2 element-wise. — op_greater","text":"x1 First input tensor. x2 Second input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_greater.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Return the truth value of x1 > x2 element-wise. — op_greater","text":"Output tensor, element-wise comparison x1 x2.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_greater_equal.html","id":null,"dir":"Reference","previous_headings":"","what":"Return the truth value of x1 >= x2 element-wise. — op_greater_equal","title":"Return the truth value of x1 >= x2 element-wise. — op_greater_equal","text":"Note function automatically called using R operator >= tensor.","code":"(x <- op_arange(4)) ## tf.Tensor([1. 2. 3. 4.], shape=(4), dtype=float32) op_greater_equal(x, 2) ## tf.Tensor([False  True  True  True], shape=(4), dtype=bool) x >= 2 ## tf.Tensor([False  True  True  True], shape=(4), dtype=bool)"},{"path":"https://keras3.posit.co/dev/reference/op_greater_equal.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Return the truth value of x1 >= x2 element-wise. — op_greater_equal","text":"","code":"op_greater_equal(x1, x2)"},{"path":"https://keras3.posit.co/dev/reference/op_greater_equal.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Return the truth value of x1 >= x2 element-wise. — op_greater_equal","text":"x1 First input tensor. x2 Second input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_greater_equal.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Return the truth value of x1 >= x2 element-wise. — op_greater_equal","text":"Output tensor, element-wise comparison x1 x2.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_hamming.html","id":null,"dir":"Reference","previous_headings":"","what":"Hamming window function. — op_hamming","title":"Hamming window function. — op_hamming","text":"Hamming window defined : w[n] = 0.54 - 0.46 * cos(2 * pi * n / (N - 1)) 0 <= n <= N - 1.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_hamming.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Hamming window function. — op_hamming","text":"","code":"op_hamming(x)"},{"path":"https://keras3.posit.co/dev/reference/op_hamming.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Hamming window function. — op_hamming","text":"x Length window. Must positive integer.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_hamming.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Hamming window function. — op_hamming","text":"1D tensor containing window values.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_hamming.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Hamming window function. — op_hamming","text":"","code":"op_hamming(5) ## tf.Tensor([0.08000001 0.54       1.         0.54       0.08000001], shape=(5), dtype=float32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_hanning.html","id":null,"dir":"Reference","previous_headings":"","what":"Hanning window function. — op_hanning","title":"Hanning window function. — op_hanning","text":"Hanning window defined : w[n] = 0.5 - 0.5 * cos(2 * pi * n / (N - 1)) 0 <= n <= N - 1.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_hanning.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Hanning window function. — op_hanning","text":"","code":"op_hanning(x)"},{"path":"https://keras3.posit.co/dev/reference/op_hanning.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Hanning window function. — op_hanning","text":"x Length window. Must positive integer.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_hanning.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Hanning window function. — op_hanning","text":"1D tensor containing window values.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_hanning.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Hanning window function. — op_hanning","text":"","code":"op_hanning(5) ## tf.Tensor([0.  0.5 1.  0.5 0. ], shape=(5), dtype=float32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_hard_shrink.html","id":null,"dir":"Reference","previous_headings":"","what":"Hard Shrink activation function. — op_hard_shrink","title":"Hard Shrink activation function. — op_hard_shrink","text":"Hard Shrink function thresholding operation defined : f(x) = x |x| > threshold, f(x) = 0 otherwise.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_hard_shrink.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Hard Shrink activation function. — op_hard_shrink","text":"","code":"op_hard_shrink(x, threshold = 0.5)"},{"path":"https://keras3.posit.co/dev/reference/op_hard_shrink.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Hard Shrink activation function. — op_hard_shrink","text":"x Input tensor. threshold Threshold value. Defaults 0.5.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_hard_shrink.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Hard Shrink activation function. — op_hard_shrink","text":"tensor shape x.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_hard_shrink.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Hard Shrink activation function. — op_hard_shrink","text":"","code":"x <- op_array(c(-0.5, 0., 1.)) op_hard_shrink(x) ## tf.Tensor([0. 0. 1.], shape=(3), dtype=float32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_hard_sigmoid.html","id":null,"dir":"Reference","previous_headings":"","what":"Hard sigmoid activation function. — op_hard_sigmoid","title":"Hard sigmoid activation function. — op_hard_sigmoid","text":"defined : 0 x < -2.5, 1 x > 2.5, (0.2 * x) + 0.5 -2.5 <= x <= 2.5.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_hard_sigmoid.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Hard sigmoid activation function. — op_hard_sigmoid","text":"","code":"op_hard_sigmoid(x)"},{"path":"https://keras3.posit.co/dev/reference/op_hard_sigmoid.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Hard sigmoid activation function. — op_hard_sigmoid","text":"x Input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_hard_sigmoid.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Hard sigmoid activation function. — op_hard_sigmoid","text":"tensor shape x.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_hard_sigmoid.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Hard sigmoid activation function. — op_hard_sigmoid","text":"","code":"x <- op_array(c(-1., 0., 1.)) op_hard_sigmoid(x) ## tf.Tensor([0.33333334 0.5        0.6666667 ], shape=(3), dtype=float32) x <- as.array(seq(-5, 5, .1)) plot(x, op_hard_sigmoid(x),      type = 'l', panel.first = grid(), frame.plot = FALSE)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_hard_silu.html","id":null,"dir":"Reference","previous_headings":"","what":"Hard SiLU activation function, also known as Hard Swish. — op_hard_silu","title":"Hard SiLU activation function, also known as Hard Swish. — op_hard_silu","text":"defined : 0 x < -3 x x > 3 x * (x + 3) / 6 -3 <= x <= 3 faster, piecewise linear approximation silu activation.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_hard_silu.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Hard SiLU activation function, also known as Hard Swish. — op_hard_silu","text":"","code":"op_hard_silu(x)  op_hard_swish(x)"},{"path":"https://keras3.posit.co/dev/reference/op_hard_silu.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Hard SiLU activation function, also known as Hard Swish. — op_hard_silu","text":"x Input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_hard_silu.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Hard SiLU activation function, also known as Hard Swish. — op_hard_silu","text":"tensor shape x.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_hard_silu.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Hard SiLU activation function, also known as Hard Swish. — op_hard_silu","text":"","code":"x <- op_convert_to_tensor(c(-3.0, -1.0, 0.0, 1.0, 3.0)) op_hard_silu(x) ## tf.Tensor([-0.         -0.33333333  0.          0.66666667  3.        ], shape=(5), dtype=float64)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_hard_tanh.html","id":null,"dir":"Reference","previous_headings":"","what":"Applies the HardTanh function element-wise. — op_hard_tanh","title":"Applies the HardTanh function element-wise. — op_hard_tanh","text":"defined : f(x) = -1 x < -1, f(x) = x -1 <= x <= 1, f(x) = 1 x > 1.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_hard_tanh.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Applies the HardTanh function element-wise. — op_hard_tanh","text":"","code":"op_hard_tanh(x)"},{"path":"https://keras3.posit.co/dev/reference/op_hard_tanh.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Applies the HardTanh function element-wise. — op_hard_tanh","text":"x Input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_hard_tanh.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Applies the HardTanh function element-wise. — op_hard_tanh","text":"Output tensor shape x values clamped -1 1.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_hard_tanh.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Applies the HardTanh function element-wise. — op_hard_tanh","text":"","code":"x <- op_array(c(-2., -1., 0., 1., 2.)) op_hard_tanh(x) ## tf.Tensor([-1. -1.  0.  1.  1.], shape=(5), dtype=float32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_heaviside.html","id":null,"dir":"Reference","previous_headings":"","what":"Heaviside step function. — op_heaviside","title":"Heaviside step function. — op_heaviside","text":"Heaviside step function defined : heaviside(x1, x2) = 0 x1 < 0, heaviside(x1, x2) = 1 x1 > 0, heaviside(x1, x2) = x2 x1 == 0.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_heaviside.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Heaviside step function. — op_heaviside","text":"","code":"op_heaviside(x1, x2)"},{"path":"https://keras3.posit.co/dev/reference/op_heaviside.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Heaviside step function. — op_heaviside","text":"x1 Tensor input. x2 Value use x1 == 0.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_heaviside.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Heaviside step function. — op_heaviside","text":"tensor broadcast x1 x2 containing 0, 1, x2.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_heaviside.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Heaviside step function. — op_heaviside","text":"","code":"x1 <- op_array(c(-2, 0, 3)) op_heaviside(x1, 0.5) ## tf.Tensor([0.  0.5 1. ], shape=(3), dtype=float32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_histogram.html","id":null,"dir":"Reference","previous_headings":"","what":"Computes a histogram of the data tensor x. — op_histogram","title":"Computes a histogram of the data tensor x. — op_histogram","text":"Computes histogram data tensor x.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_histogram.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes a histogram of the data tensor x. — op_histogram","text":"","code":"op_histogram(x, bins = 10L, range = NULL)"},{"path":"https://keras3.posit.co/dev/reference/op_histogram.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Computes a histogram of the data tensor x. — op_histogram","text":"x Input tensor. bins integer representing number histogram bins. Defaults 10. range pair numbers representing lower upper range bins. specified, use min max x.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_histogram.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Computes a histogram of the data tensor x. — op_histogram","text":"list two tensors containing: tensor representing counts elements bin. tensor representing bin edges.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_histogram.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Computes a histogram of the data tensor x. — op_histogram","text":"","code":"input_tensor <- random_uniform(8) c(counts, edges) %<-% op_histogram(input_tensor)  counts #> tf.Tensor([2. 1. 1. 1. 0. 1. 0. 0. 0. 2.], shape=(10), dtype=float32) edges #> tf.Tensor( #> [0.01647806 0.10867032 0.20086257 0.29305482 0.38524708 0.47743934 #>  0.5696316  0.66182387 0.7540161  0.84620833 0.9384006 ], shape=(11), dtype=float32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_hstack.html","id":null,"dir":"Reference","previous_headings":"","what":"Stack tensors in sequence horizontally (column wise). — op_hstack","title":"Stack tensors in sequence horizontally (column wise). — op_hstack","text":"equivalent concatenation along first axis 1-D tensors, along second axis tensors.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_hstack.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Stack tensors in sequence horizontally (column wise). — op_hstack","text":"","code":"op_hstack(xs, ...)"},{"path":"https://keras3.posit.co/dev/reference/op_hstack.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Stack tensors in sequence horizontally (column wise). — op_hstack","text":"xs, ... list tensors.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_hstack.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Stack tensors in sequence horizontally (column wise). — op_hstack","text":"tensor formed stacking given tensors.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_identity.html","id":null,"dir":"Reference","previous_headings":"","what":"Return the identity tensor. — op_identity","title":"Return the identity tensor. — op_identity","text":"identity tensor square tensor ones main diagonal zeros elsewhere.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_identity.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Return the identity tensor. — op_identity","text":"","code":"op_identity(n, dtype = NULL)"},{"path":"https://keras3.posit.co/dev/reference/op_identity.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Return the identity tensor. — op_identity","text":"n Number rows (columns) n x n output tensor. dtype Data type output tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_identity.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Return the identity tensor. — op_identity","text":"identity tensor.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_ifft2.html","id":null,"dir":"Reference","previous_headings":"","what":"Computes the 2D Inverse Fast Fourier Transform along the last two axes of — op_ifft2","title":"Computes the 2D Inverse Fast Fourier Transform along the last two axes of — op_ifft2","text":"input.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_ifft2.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes the 2D Inverse Fast Fourier Transform along the last two axes of — op_ifft2","text":"","code":"op_ifft2(x)"},{"path":"https://keras3.posit.co/dev/reference/op_ifft2.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Computes the 2D Inverse Fast Fourier Transform along the last two axes of — op_ifft2","text":"x Tuple real imaginary parts input tensor. tensors tuple floating type.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_ifft2.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Computes the 2D Inverse Fast Fourier Transform along the last two axes of — op_ifft2","text":"tuple containing two tensors - real imaginary parts output.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_ifft2.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Computes the 2D Inverse Fast Fourier Transform along the last two axes of — op_ifft2","text":"","code":"x <- list(op_array(rbind(c(1, 2), c(2, 1))),           op_array(rbind(c(0, 1), c(1, 0)))) op_ifft2(x) ## [[1]] ## tf.Tensor( ## [[ 1.5  0. ] ##  [ 0.  -0.5]], shape=(2, 2), dtype=float64) ## ## [[2]] ## tf.Tensor( ## [[ 0.5 -0. ] ##  [-0.  -0.5]], shape=(2, 2), dtype=float64)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_imag.html","id":null,"dir":"Reference","previous_headings":"","what":"Return the imaginary part of the complex argument. — op_imag","title":"Return the imaginary part of the complex argument. — op_imag","text":"Return imaginary part complex argument.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_imag.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Return the imaginary part of the complex argument. — op_imag","text":"","code":"op_imag(x)"},{"path":"https://keras3.posit.co/dev/reference/op_imag.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Return the imaginary part of the complex argument. — op_imag","text":"x Input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_imag.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Return the imaginary part of the complex argument. — op_imag","text":"imaginary component complex argument.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_image_affine_transform.html","id":null,"dir":"Reference","previous_headings":"","what":"Applies the given transform(s) to the image(s). — op_image_affine_transform","title":"Applies the given transform(s) to the image(s). — op_image_affine_transform","text":"Applies given transform(s) image(s).","code":""},{"path":"https://keras3.posit.co/dev/reference/op_image_affine_transform.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Applies the given transform(s) to the image(s). — op_image_affine_transform","text":"","code":"op_image_affine_transform(   images,   transform,   interpolation = \"bilinear\",   fill_mode = \"constant\",   fill_value = 0L,   data_format = NULL )"},{"path":"https://keras3.posit.co/dev/reference/op_image_affine_transform.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Applies the given transform(s) to the image(s). — op_image_affine_transform","text":"images Input image batch images. Must 3D 4D. transform Projective transform matrix/matrices. vector length 8 tensor size N x 8. one row transform [a0, a1, a2, b0, b1, b2, c0, c1], maps output point (x, y) transformed input point (x', y') = ((a0 x + a1 y + a2) / k, (b0 x + b1 y + b2) / k), k = c0 x + c1 y + 1. transform inverted compared transform mapping input points output points. Note gradients backpropagated transformation parameters. Note c0 c1 effective using TensorFlow backend considered 0 using backends. interpolation Interpolation method. Available methods \"nearest\", \"bilinear\". Defaults \"bilinear\". fill_mode Points outside boundaries input filled according given mode. Available methods \"constant\", \"nearest\", \"wrap\" \"reflect\". Defaults \"constant\". \"reflect\": (d c b | b c d | d c b ) input extended reflecting edge last pixel. \"constant\": (k k k k | b c d | k k k k) input extended filling values beyond edge constant value k specified fill_value. \"wrap\": (b c d | b c d | b c d) input extended wrapping around opposite edge. \"nearest\": (| b c d | d d d d) input extended nearest pixel. fill_value Value used points outside boundaries input fill_mode = \"constant\". Defaults 0. data_format string specifying data format input tensor. can either \"channels_last\" \"channels_first\". \"channels_last\" corresponds inputs shape (batch, height, width, channels), \"channels_first\" corresponds inputs shape (batch, channels, height, width). specified, value default config_image_data_format().","code":""},{"path":"https://keras3.posit.co/dev/reference/op_image_affine_transform.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Applies the given transform(s) to the image(s). — op_image_affine_transform","text":"Applied affine transform image batch images.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_image_affine_transform.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Applies the given transform(s) to the image(s). — op_image_affine_transform","text":"","code":"x <- random_uniform(c(2, 64, 80, 3)) # batch of 2 RGB images transform <- op_array(rbind(c(1.5, 0, -20, 0, 1.5, -16, 0, 0),  # zoom                            c(1, 0, -20, 0, 1, -16, 0, 0)))  # translation)) y <- op_image_affine_transform(x, transform) shape(y) ## shape(2, 64, 80, 3) # (2, 64, 80, 3) x <- random_uniform(c(64, 80, 3)) # single RGB image transform <- op_array(c(1.0, 0.5, -20, 0.5, 1.0, -16, 0, 0))  # shear y <- op_image_affine_transform(x, transform) shape(y) ## shape(64, 80, 3) # (64, 80, 3) x <- random_uniform(c(2, 3, 64, 80)) # batch of 2 RGB images transform <- op_array(rbind(   c(1.5, 0,-20, 0, 1.5,-16, 0, 0),  # zoom   c(1, 0,-20, 0, 1,-16, 0, 0)  # translation )) y <- op_image_affine_transform(x, transform, data_format = \"channels_first\") shape(y) ## shape(2, 3, 64, 80) # (2, 3, 64, 80)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_image_crop.html","id":null,"dir":"Reference","previous_headings":"","what":"Crop images to a specified height and width. — op_image_crop","title":"Crop images to a specified height and width. — op_image_crop","text":"Crop images specified height width.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_image_crop.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Crop images to a specified height and width. — op_image_crop","text":"","code":"op_image_crop(   images,   top_cropping = NULL,   left_cropping = NULL,   bottom_cropping = NULL,   right_cropping = NULL,   target_height = NULL,   target_width = NULL,   data_format = NULL )"},{"path":"https://keras3.posit.co/dev/reference/op_image_crop.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Crop images to a specified height and width. — op_image_crop","text":"images Input image batch images. Must 3D 4D. top_cropping Number columns crop top. left_cropping Number columns crop left. bottom_cropping Number columns crop bottom. right_cropping Number columns crop right. target_height Height output images. target_width Width output images. data_format string specifying data format input tensor. can either \"channels_last\" \"channels_first\". \"channels_last\" corresponds inputs shape (batch, height, width, channels), \"channels_first\" corresponds inputs shape (batch, channels, height, width). specified, value default config_image_data_format().","code":""},{"path":"https://keras3.posit.co/dev/reference/op_image_crop.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Crop images to a specified height and width. — op_image_crop","text":"Cropped image batch images.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_image_crop.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Crop images to a specified height and width. — op_image_crop","text":"","code":"images <- op_reshape(op_arange(1, 28, dtype=\"float32\"), c(3, 3, 3)) images[, , 1] # print the first channel of the images  cropped_images <- op_image_crop(images, 0, 0, 2, 2) cropped_images[, , 1] # print the first channel of the cropped images"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_image_elastic_transform.html","id":null,"dir":"Reference","previous_headings":"","what":"Applies elastic deformation to the image(s). — op_image_elastic_transform","title":"Applies elastic deformation to the image(s). — op_image_elastic_transform","text":"Apply random elastic deformation 3D 4D image tensors.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_image_elastic_transform.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Applies elastic deformation to the image(s). — op_image_elastic_transform","text":"","code":"op_image_elastic_transform(   images,   alpha = 20,   sigma = 5,   interpolation = \"bilinear\",   fill_mode = \"reflect\",   fill_value = 0,   seed = NULL,   data_format = NULL )"},{"path":"https://keras3.posit.co/dev/reference/op_image_elastic_transform.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Applies elastic deformation to the image(s). — op_image_elastic_transform","text":"images Input image batch images. Must 3D 4D. alpha Scaling factor controls intensity deformation. sigma Standard deviation Gaussian filter used smoothing displacement fields. interpolation Interpolation method. Available methods \"nearest\", \"bilinear\". Defaults \"bilinear\". fill_mode Points outside boundaries input filled according given mode. Available methods \"constant\", \"nearest\", \"wrap\" \"reflect\". Defaults \"reflect\". \"reflect\": (d c b | b c d | d c b ) input extended reflecting edge last pixel. \"constant\": (k k k k | b c d | k k k k) input extended filling values beyond edge constant value k specified fill_value. \"wrap\": (b c d | b c d | b c d) input extended wrapping around opposite edge. \"nearest\": (| b c d | d d d d) input extended nearest pixel. fill_value Value used points outside boundaries input fill_mode=\"constant\". Defaults 0. seed Optional integer seed random number generator. data_format string specifying data format input tensor. can either \"channels_last\" \"channels_first\". \"channels_last\" corresponds inputs shape (batch, height, width, channels), \"channels_first\" corresponds inputs shape (batch, channels, height, width). specified, value default keras.config.image_data_format.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_image_elastic_transform.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Applies elastic deformation to the image(s). — op_image_elastic_transform","text":"Transformed image batch images elastic deformation.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_image_elastic_transform.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Applies elastic deformation to the image(s). — op_image_elastic_transform","text":"","code":"x <- random_uniform(c(2, 64, 80, 3))  # batch of 2 RGB images y <- op_image_elastic_transform(x) op_shape(y) ## shape(2, 64, 80, 3) x <- random_uniform(c(64, 80, 3))  # single RGB image y <- op_image_elastic_transform(x) op_shape(y) ## shape(64, 80, 3) x <- random_uniform(c(2, 3, 64, 80))  # batch of 2 RGB images y <- op_image_elastic_transform(   x,   data_format = \"channels_first\",   seed = 123 ) op_shape(y) ## shape(2, 3, 64, 80)"},{"path":"https://keras3.posit.co/dev/reference/op_image_extract_patches.html","id":null,"dir":"Reference","previous_headings":"","what":"Extracts patches from the image(s). — op_image_extract_patches","title":"Extracts patches from the image(s). — op_image_extract_patches","text":"Extracts patches image(s).","code":""},{"path":"https://keras3.posit.co/dev/reference/op_image_extract_patches.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Extracts patches from the image(s). — op_image_extract_patches","text":"","code":"op_image_extract_patches(   images,   size,   strides = NULL,   dilation_rate = 1L,   padding = \"valid\",   data_format = \"channels_last\" )"},{"path":"https://keras3.posit.co/dev/reference/op_image_extract_patches.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Extracts patches from the image(s). — op_image_extract_patches","text":"images Input image batch images. Must 3D 4D. size Patch size int list (patch_height, patch_width) strides strides along height width. specified, NULL, defaults value size. dilation_rate input stride, specifying far two consecutive patch samples input. value 1, strides must 1. NOTE: strides > 1 supported conjunction dilation_rate > 1 padding type padding algorithm use: \"\" \"valid\". data_format string specifying data format input tensor. can either \"channels_last\" \"channels_first\". \"channels_last\" corresponds inputs shape (batch, height, width, channels), \"channels_first\" corresponds inputs shape (batch, channels, height, width). specified, value default config_image_data_format().","code":""},{"path":"https://keras3.posit.co/dev/reference/op_image_extract_patches.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Extracts patches from the image(s). — op_image_extract_patches","text":"Extracted patches 3D (batched) 4D (batched)","code":""},{"path":"https://keras3.posit.co/dev/reference/op_image_extract_patches.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Extracts patches from the image(s). — op_image_extract_patches","text":"","code":"image <- random_uniform(c(2, 20, 20, 3), dtype = \"float32\") # batch of 2 RGB images patches <- op_image_extract_patches(image, c(5, 5)) shape(patches) ## shape(2, 4, 4, 75) # (2, 4, 4, 75) image <- random_uniform(c(20, 20, 3), dtype = \"float32\") # 1 RGB image patches <- op_image_extract_patches(image, c(3, 3), c(1, 1)) shape(patches) ## shape(18, 18, 27) # (18, 18, 27)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_image_gaussian_blur.html","id":null,"dir":"Reference","previous_headings":"","what":"Applies a Gaussian blur to the image(s). — op_image_gaussian_blur","title":"Applies a Gaussian blur to the image(s). — op_image_gaussian_blur","text":"Applies Gaussian blur image(s).","code":""},{"path":"https://keras3.posit.co/dev/reference/op_image_gaussian_blur.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Applies a Gaussian blur to the image(s). — op_image_gaussian_blur","text":"","code":"op_image_gaussian_blur(   images,   kernel_size = list(3L, 3L),   sigma = list(1, 1),   data_format = NULL )"},{"path":"https://keras3.posit.co/dev/reference/op_image_gaussian_blur.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Applies a Gaussian blur to the image(s). — op_image_gaussian_blur","text":"images Input image batch images. Must 3D 4D. kernel_size tuple two integers, specifying height width Gaussian kernel. sigma tuple two floats, specifying standard deviation Gaussian kernel along height width. data_format string specifying data format input tensor. can either \"channels_last\" \"channels_first\". \"channels_last\" corresponds inputs shape (batch, height, width, channels), \"channels_first\" corresponds inputs shape (batch, channels, height, width). specified, value default keras.config.image_data_format.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_image_gaussian_blur.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Applies a Gaussian blur to the image(s). — op_image_gaussian_blur","text":"Blurred image batch images.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_image_gaussian_blur.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Applies a Gaussian blur to the image(s). — op_image_gaussian_blur","text":"","code":"x <- op_ones(c(2, 64, 80, 3))  # batch of 2 RGB images y <- op_image_gaussian_blur(x) op_shape(y) ## shape(2, 64, 80, 3) x <- op_ones(c(64, 80, 3))  # single RGB image y <- op_image_gaussian_blur(x) op_shape(y) ## shape(64, 80, 3) x <- op_ones(c(2, 3, 64, 80))  # batch of 2 RGB images, channels_first y <- op_image_gaussian_blur(x, data_format = \"channels_first\") op_shape(y) ## shape(2, 3, 64, 80)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_image_hsv_to_rgb.html","id":null,"dir":"Reference","previous_headings":"","what":"Convert HSV images to RGB. — op_image_hsv_to_rgb","title":"Convert HSV images to RGB. — op_image_hsv_to_rgb","text":"images must float dtype, output well defined values images [0, 1].","code":""},{"path":"https://keras3.posit.co/dev/reference/op_image_hsv_to_rgb.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Convert HSV images to RGB. — op_image_hsv_to_rgb","text":"","code":"op_image_hsv_to_rgb(images, data_format = NULL)"},{"path":"https://keras3.posit.co/dev/reference/op_image_hsv_to_rgb.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Convert HSV images to RGB. — op_image_hsv_to_rgb","text":"images Input image batch images. Must 3D 4D. data_format string specifying data format input tensor. can either \"channels_last\" \"channels_first\". \"channels_last\" corresponds inputs shape (batch, height, width, channels), \"channels_first\" corresponds inputs shape (batch, channels, height, width). specified, value default config_image_data_format().","code":""},{"path":"https://keras3.posit.co/dev/reference/op_image_hsv_to_rgb.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Convert HSV images to RGB. — op_image_hsv_to_rgb","text":"RGB image batch RGB images.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_image_hsv_to_rgb.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Convert HSV images to RGB. — op_image_hsv_to_rgb","text":"","code":"x <- random_uniform(c(2, 4, 4, 3)) y <- op_image_hsv_to_rgb(x) shape(y) ## shape(2, 4, 4, 3) x <- random_uniform(c(4, 4, 3)) # Single HSV image y <- op_image_hsv_to_rgb(x) shape(y) ## shape(4, 4, 3) x <- random_uniform(c(2, 3, 4, 4)) y <- op_image_hsv_to_rgb(x, data_format=\"channels_first\") shape(y) ## shape(2, 3, 4, 4)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_image_map_coordinates.html","id":null,"dir":"Reference","previous_headings":"","what":"Map the input array to new coordinates by interpolation. — op_image_map_coordinates","title":"Map the input array to new coordinates by interpolation. — op_image_map_coordinates","text":"Note interpolation near boundaries differs scipy function, fixed outstanding bug scipy/issues/2640.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_image_map_coordinates.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Map the input array to new coordinates by interpolation. — op_image_map_coordinates","text":"","code":"op_image_map_coordinates(   inputs,   coordinates,   order,   fill_mode = \"constant\",   fill_value = 0L )"},{"path":"https://keras3.posit.co/dev/reference/op_image_map_coordinates.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Map the input array to new coordinates by interpolation. — op_image_map_coordinates","text":"inputs input array. coordinates coordinates inputs evaluated. order order spline interpolation. order must 0 1. 0 indicates nearest neighbor 1 indicates linear interpolation. fill_mode Points outside boundaries inputs filled according given mode. Available methods \"constant\", \"nearest\", \"wrap\" \"mirror\" \"reflect\". Defaults \"constant\". \"constant\": (k k k k | b c d | k k k k) inputs extended filling values beyond edge constant value k specified fill_value. \"nearest\": (| b c d | d d d d) inputs extended nearest pixel. \"wrap\": (b c d | b c d | b c d) inputs extended wrapping around opposite edge. \"mirror\": (c d c b | b c d | c b b) inputs extended mirroring edge. \"reflect\": (d c b | b c d | d c b ) inputs extended reflecting edge last pixel. fill_value Value used points outside boundaries inputs fill_mode = \"constant\". Defaults 0.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_image_map_coordinates.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Map the input array to new coordinates by interpolation. — op_image_map_coordinates","text":"Output input batch inputs.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_image_pad.html","id":null,"dir":"Reference","previous_headings":"","what":"Pad images with zeros to the specified height and width. — op_image_pad","title":"Pad images with zeros to the specified height and width. — op_image_pad","text":"Pad images zeros specified height width.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_image_pad.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Pad images with zeros to the specified height and width. — op_image_pad","text":"","code":"op_image_pad(   images,   top_padding = NULL,   left_padding = NULL,   bottom_padding = NULL,   right_padding = NULL,   target_height = NULL,   target_width = NULL,   data_format = NULL )"},{"path":"https://keras3.posit.co/dev/reference/op_image_pad.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Pad images with zeros to the specified height and width. — op_image_pad","text":"images Input image batch images. Must 3D 4D. top_padding Number rows zeros add top. left_padding Number columns zeros add left. bottom_padding Number rows zeros add bottom. right_padding Number columns zeros add right. target_height Height output images. target_width Width output images. data_format string specifying data format input tensor. can either \"channels_last\" \"channels_first\". \"channels_last\" corresponds inputs shape (batch, height, width, channels), \"channels_first\" corresponds inputs shape (batch, channels, height, width). specified, value default config_image_data_format().","code":""},{"path":"https://keras3.posit.co/dev/reference/op_image_pad.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Pad images with zeros to the specified height and width. — op_image_pad","text":"Padded image batch images.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_image_pad.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Pad images with zeros to the specified height and width. — op_image_pad","text":"","code":"images <- random_uniform(c(15, 25, 3)) padded_images <- op_image_pad(     images, 2, 3, target_height = 20, target_width = 30 ) shape(padded_images) ## shape(20, 30, 3) batch_images <- random_uniform(c(2, 15, 25, 3)) padded_batch <- op_image_pad(batch_images, 2, 3,                              target_height = 20,                              target_width = 30) shape(padded_batch) ## shape(2, 20, 30, 3)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_image_perspective_transform.html","id":null,"dir":"Reference","previous_headings":"","what":"Applies a perspective transformation to the image(s). — op_image_perspective_transform","title":"Applies a perspective transformation to the image(s). — op_image_perspective_transform","text":"Applies perspective transformation image(s).","code":""},{"path":"https://keras3.posit.co/dev/reference/op_image_perspective_transform.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Applies a perspective transformation to the image(s). — op_image_perspective_transform","text":"","code":"op_image_perspective_transform(   images,   start_points,   end_points,   interpolation = \"bilinear\",   fill_value = 0L,   data_format = NULL )"},{"path":"https://keras3.posit.co/dev/reference/op_image_perspective_transform.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Applies a perspective transformation to the image(s). — op_image_perspective_transform","text":"images Input image batch images. Must 3D 4D. start_points tensor shape (N, 4, 2) (4, 2), representing source points original image define transformation. end_points tensor shape (N, 4, 2) (4, 2), representing target points output image transformation. interpolation Interpolation method. Available methods \"nearest\", \"bilinear\". Defaults \"bilinear\". fill_value Value used points outside boundaries input extrapolation needed. Defaults 0. data_format string specifying data format input tensor. can either \"channels_last\" \"channels_first\". \"channels_last\" corresponds inputs shape (batch, height, width, channels), \"channels_first\" corresponds inputs shape (batch, channels, height, width). specified, value default keras.config.image_data_format.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_image_perspective_transform.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Applies a perspective transformation to the image(s). — op_image_perspective_transform","text":"Applied perspective transform image batch images.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_image_perspective_transform.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Applies a perspective transformation to the image(s). — op_image_perspective_transform","text":"","code":"# Batch of 2 RGB images (channels_last) x <- op_ones(c(2, 64, 80, 3)) start_points <- op_stack(list(   rbind(c(0, 0), c(0, 64), c(80, 0), c(80, 64)),   rbind(c(0, 0), c(0, 64), c(80, 0), c(80, 64)) )) end_points <- op_stack(list(   rbind(c(3, 5), c(7, 64), c(76, -10), c(84, 61)),   rbind(c(8, 10), c(10, 61), c(65, 3), c(88, 43)) )) y <- op_image_perspective_transform(x, start_points, end_points) op_shape(y) ## shape(2, 64, 80, 3) # Single RGB image (channels_last) x <- op_ones(c(64, 80, 3)) start_points <- rbind(c(0, 0), c(0, 64), c(80, 0), c(80, 64)) end_points <- rbind(c(3, 5), c(7, 64), c(76, -10), c(84, 61)) y <- op_image_perspective_transform(x, start_points, end_points) op_shape(y) ## shape(64, 80, 3) # Batch of 2 RGB images (channels_first) x <- op_ones(c(2, 3, 64, 80)) start_points <- op_stack(list(   rbind(c(0, 0), c(0, 64), c(80, 0), c(80, 64)),   rbind(c(0, 0), c(0, 64), c(80, 0), c(80, 64)) )) end_points <- op_stack(list(   rbind(c(3, 5), c(7, 64), c(76, -10), c(84, 61)),   rbind(c(8, 10), c(10, 61), c(65, 3), c(88, 43)) )) y <- op_image_perspective_transform(   x, start_points, end_points,   data_format = \"channels_first\" ) op_shape(y) ## shape(2, 3, 64, 80)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_image_resize.html","id":null,"dir":"Reference","previous_headings":"","what":"Resize images to size using the specified interpolation method. — op_image_resize","title":"Resize images to size using the specified interpolation method. — op_image_resize","text":"Resize images size using specified interpolation method.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_image_resize.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Resize images to size using the specified interpolation method. — op_image_resize","text":"","code":"op_image_resize(   images,   size,   interpolation = \"bilinear\",   antialias = FALSE,   crop_to_aspect_ratio = FALSE,   pad_to_aspect_ratio = FALSE,   fill_mode = \"constant\",   fill_value = 0,   data_format = NULL )"},{"path":"https://keras3.posit.co/dev/reference/op_image_resize.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Resize images to size using the specified interpolation method. — op_image_resize","text":"images Input image batch images. Must 3D 4D. size Size output image (height, width) format. interpolation Interpolation method. Available methods \"nearest\", \"bilinear\", \"bicubic\". Defaults \"bilinear\". antialias Whether use antialiasing filter downsampling image. Defaults FALSE. crop_to_aspect_ratio TRUE, resize images without aspect ratio distortion. original aspect ratio differs target aspect ratio, output image cropped return largest possible window image (size (height, width)) matches target aspect ratio. default (crop_to_aspect_ratio=FALSE), aspect ratio may preserved. pad_to_aspect_ratio TRUE, pad images without aspect ratio distortion. original aspect ratio differs target aspect ratio, output image evenly padded short side. fill_mode using pad_to_aspect_ratio=TRUE, padded areas filled according given mode. \"constant\" supported time (fill constant value, equal fill_value). fill_value Float. Padding value use pad_to_aspect_ratio=TRUE. data_format string specifying data format input tensor. can either \"channels_last\" \"channels_first\". \"channels_last\" corresponds inputs shape (batch, height, width, channels), \"channels_first\" corresponds inputs shape (batch, channels, height, width). specified, value default config_image_data_format().","code":""},{"path":"https://keras3.posit.co/dev/reference/op_image_resize.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Resize images to size using the specified interpolation method. — op_image_resize","text":"Resized image batch images.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_image_resize.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Resize images to size using the specified interpolation method. — op_image_resize","text":"","code":"x <- random_uniform(c(2, 4, 4, 3)) # batch of 2 RGB images y <- op_image_resize(x, c(2, 2)) shape(y) ## shape(2, 2, 2, 3) x <- random_uniform(c(4, 4, 3)) # single RGB image y <- op_image_resize(x, c(2, 2)) shape(y) ## shape(2, 2, 3) x <- random_uniform(c(2, 3, 4, 4)) # batch of 2 RGB images y <- op_image_resize(x, c(2, 2), data_format = \"channels_first\") shape(y) ## shape(2, 3, 2, 2)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_image_rgb_to_grayscale.html","id":null,"dir":"Reference","previous_headings":"","what":"Convert RGB images to grayscale. — op_image_rgb_to_grayscale","title":"Convert RGB images to grayscale. — op_image_rgb_to_grayscale","text":"function converts RGB images grayscale images. supports 3D 4D tensors, last dimension represents channels.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_image_rgb_to_grayscale.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Convert RGB images to grayscale. — op_image_rgb_to_grayscale","text":"","code":"op_image_rgb_to_grayscale(images, data_format = NULL)"},{"path":"https://keras3.posit.co/dev/reference/op_image_rgb_to_grayscale.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Convert RGB images to grayscale. — op_image_rgb_to_grayscale","text":"images Input image batch images. Must 3D 4D. data_format string specifying data format input tensor. can either \"channels_last\" \"channels_first\". \"channels_last\" corresponds inputs shape (batch, height, width, channels), \"channels_first\" corresponds inputs shape (batch, channels, height, width). specified, value default config_image_data_format().","code":""},{"path":"https://keras3.posit.co/dev/reference/op_image_rgb_to_grayscale.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Convert RGB images to grayscale. — op_image_rgb_to_grayscale","text":"Grayscale image batch grayscale images.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_image_rgb_to_grayscale.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Convert RGB images to grayscale. — op_image_rgb_to_grayscale","text":"","code":"x <- random_uniform(c(2, 4, 4, 3)) y <- op_image_rgb_to_grayscale(x) shape(y) ## shape(2, 4, 4, 1) x <- random_uniform(c(4, 4, 3)) # Single RGB image y = op_image_rgb_to_grayscale(x) shape(y) ## shape(4, 4, 1) x <- random_uniform(c(2, 3, 4, 4)) y <- op_image_rgb_to_grayscale(x, data_format=\"channels_first\") shape(y) ## shape(2, 1, 4, 4)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_image_rgb_to_hsv.html","id":null,"dir":"Reference","previous_headings":"","what":"Convert RGB images to HSV. — op_image_rgb_to_hsv","title":"Convert RGB images to HSV. — op_image_rgb_to_hsv","text":"images must float dtype, output well defined values images [0, 1]. HSV values [0, 1]. hue 0 corresponds pure red, 1/3 pure green, 2/3 pure blue.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_image_rgb_to_hsv.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Convert RGB images to HSV. — op_image_rgb_to_hsv","text":"","code":"op_image_rgb_to_hsv(images, data_format = NULL)"},{"path":"https://keras3.posit.co/dev/reference/op_image_rgb_to_hsv.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Convert RGB images to HSV. — op_image_rgb_to_hsv","text":"images Input image batch images. Must 3D 4D. data_format string specifying data format input tensor. can either \"channels_last\" \"channels_first\". \"channels_last\" corresponds inputs shape (batch, height, width, channels), \"channels_first\" corresponds inputs shape (batch, channels, height, width). specified, value default config_image_data_format().","code":""},{"path":"https://keras3.posit.co/dev/reference/op_image_rgb_to_hsv.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Convert RGB images to HSV. — op_image_rgb_to_hsv","text":"HSV image batch HSV images.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_image_rgb_to_hsv.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Convert RGB images to HSV. — op_image_rgb_to_hsv","text":"","code":"x <- random_uniform(c(2, 4, 4, 3)) y <- op_image_rgb_to_hsv(x) shape(y) ## shape(2, 4, 4, 3) x <- random_uniform(c(4, 4, 3)) # Single RGB image y <- op_image_rgb_to_hsv(x) shape(y) ## shape(4, 4, 3) x <- random_uniform(c(2, 3, 4, 4)) y <- op_image_rgb_to_hsv(x, data_format = \"channels_first\") shape(y) ## shape(2, 3, 4, 4)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_in_top_k.html","id":null,"dir":"Reference","previous_headings":"","what":"Checks if the targets are in the top-k predictions. — op_in_top_k","title":"Checks if the targets are in the top-k predictions. — op_in_top_k","text":"Checks targets top-k predictions.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_in_top_k.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Checks if the targets are in the top-k predictions. — op_in_top_k","text":"","code":"op_in_top_k(targets, predictions, k)"},{"path":"https://keras3.posit.co/dev/reference/op_in_top_k.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Checks if the targets are in the top-k predictions. — op_in_top_k","text":"targets tensor true labels. predictions tensor predicted labels. k integer representing number predictions consider.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_in_top_k.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Checks if the targets are in the top-k predictions. — op_in_top_k","text":"boolean tensor shape targets, element indicates whether corresponding target top-k predictions.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_in_top_k.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Checks if the targets are in the top-k predictions. — op_in_top_k","text":"","code":"targets <- op_array(c(2, 5, 3), \"int32\") predictions <- op_array(dtype = \"float32\", rbind(   c(0.1, 0.4, 0.6, 0.9, 0.5),   c(0.1, 0.7, 0.9, 0.8, 0.3),   c(0.1, 0.6, 0.9, 0.9, 0.5) )) op_in_top_k(targets, predictions, k = 3L) ## tf.Tensor([ True False  True], shape=(3), dtype=bool)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_inner.html","id":null,"dir":"Reference","previous_headings":"","what":"Return the inner product of two tensors. — op_inner","title":"Return the inner product of two tensors. — op_inner","text":"Ordinary inner product vectors 1-D tensors (without complex conjugation), higher dimensions sum product last axes. Multidimensional arrays treated vectors flattening last axes. resulting dot product performed last axes.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_inner.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Return the inner product of two tensors. — op_inner","text":"","code":"op_inner(x1, x2)"},{"path":"https://keras3.posit.co/dev/reference/op_inner.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Return the inner product of two tensors. — op_inner","text":"x1 First input tensor. x2 Second input tensor. last dimension x1 x2 must match.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_inner.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Return the inner product of two tensors. — op_inner","text":"Output tensor. shape output determined broadcasting shapes x1 x2 removing last axes.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_inv.html","id":null,"dir":"Reference","previous_headings":"","what":"Computes the inverse of a square tensor. — op_inv","title":"Computes the inverse of a square tensor. — op_inv","text":"Computes inverse square tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_inv.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes the inverse of a square tensor. — op_inv","text":"","code":"op_inv(x)"},{"path":"https://keras3.posit.co/dev/reference/op_inv.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Computes the inverse of a square tensor. — op_inv","text":"x Input tensor shape (..., M, M).","code":""},{"path":"https://keras3.posit.co/dev/reference/op_inv.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Computes the inverse of a square tensor. — op_inv","text":"tensor shape (..., M, M) representing inverse x.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_irfft.html","id":null,"dir":"Reference","previous_headings":"","what":"Inverse real-valued Fast Fourier transform along the last axis. — op_irfft","title":"Inverse real-valued Fast Fourier transform along the last axis. — op_irfft","text":"Computes inverse 1D Discrete Fourier Transform real-valued signal inner-dimension input. inner-dimension input assumed result RFFT: fft_length / 2 + 1 unique components DFT real-valued signal. fft_length provided, computed size inner-dimension input (fft_length = 2 * (inner - 1)). FFT length used compute odd, provided since inferred properly. Along axis IRFFT computed , fft_length / 2 + 1 smaller corresponding dimension input, dimension cropped. larger, dimension padded zeros.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_irfft.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Inverse real-valued Fast Fourier transform along the last axis. — op_irfft","text":"","code":"op_irfft(x, fft_length = NULL)"},{"path":"https://keras3.posit.co/dev/reference/op_irfft.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Inverse real-valued Fast Fourier transform along the last axis. — op_irfft","text":"x List real imaginary parts input tensor. tensors list floating type. fft_length integer representing number fft length. specified, inferred length last axis x. Defaults NULL.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_irfft.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Inverse real-valued Fast Fourier transform along the last axis. — op_irfft","text":"tensor containing inverse real-valued Fast Fourier Transform along last axis x.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_irfft.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Inverse real-valued Fast Fourier transform along the last axis. — op_irfft","text":"","code":"real <- op_array(c(0, 1, 2, 3, 4)) imag <- op_array(c(0, 1, 2, 3, 4)) op_irfft(c(real, imag)) #> tf.Tensor( #> [ 2.         -2.0606601   0.5        -0.35355338  0.          0.06066012 #>  -0.5         0.35355338], shape=(8), dtype=float32) all.equal(op_irfft(op_rfft(real, 5), 5), real) #> [1] TRUE"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_is_tensor.html","id":null,"dir":"Reference","previous_headings":"","what":"Check whether the given object is a tensor. — op_is_tensor","title":"Check whether the given object is a tensor. — op_is_tensor","text":"Check whether given object tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_is_tensor.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Check whether the given object is a tensor. — op_is_tensor","text":"","code":"op_is_tensor(x)"},{"path":"https://keras3.posit.co/dev/reference/op_is_tensor.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Check whether the given object is a tensor. — op_is_tensor","text":"x variable.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_is_tensor.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Check whether the given object is a tensor. — op_is_tensor","text":"TRUE x tensor, otherwise FALSE.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_is_tensor.html","id":"note","dir":"Reference","previous_headings":"","what":"Note","title":"Check whether the given object is a tensor. — op_is_tensor","text":"checks backend specific tensors passing TensorFlow tensor return FALSE backend PyTorch JAX.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_isclose.html","id":null,"dir":"Reference","previous_headings":"","what":"Return whether two tensors are element-wise almost equal. — op_isclose","title":"Return whether two tensors are element-wise almost equal. — op_isclose","text":"Return whether two tensors element-wise almost equal.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_isclose.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Return whether two tensors are element-wise almost equal. — op_isclose","text":"","code":"op_isclose(x1, x2, rtol = 1e-05, atol = 1e-08, equal_nan = FALSE)"},{"path":"https://keras3.posit.co/dev/reference/op_isclose.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Return whether two tensors are element-wise almost equal. — op_isclose","text":"x1 First input tensor. x2 Second input tensor. rtol Relative tolerance. atol Absolute tolerance. equal_nan TRUE, element-wise NaNs considered equal.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_isclose.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Return whether two tensors are element-wise almost equal. — op_isclose","text":"Output boolean tensor.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_isfinite.html","id":null,"dir":"Reference","previous_headings":"","what":"Return whether a tensor is finite, element-wise. — op_isfinite","title":"Return whether a tensor is finite, element-wise. — op_isfinite","text":"Real values finite NaN, positive infinity, negative infinity. Complex values finite real imaginary parts finite.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_isfinite.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Return whether a tensor is finite, element-wise. — op_isfinite","text":"","code":"op_isfinite(x)"},{"path":"https://keras3.posit.co/dev/reference/op_isfinite.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Return whether a tensor is finite, element-wise. — op_isfinite","text":"x Input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_isfinite.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Return whether a tensor is finite, element-wise. — op_isfinite","text":"Output boolean tensor.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_isinf.html","id":null,"dir":"Reference","previous_headings":"","what":"Test element-wise for positive or negative infinity. — op_isinf","title":"Test element-wise for positive or negative infinity. — op_isinf","text":"Test element-wise positive negative infinity.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_isinf.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Test element-wise for positive or negative infinity. — op_isinf","text":"","code":"op_isinf(x)"},{"path":"https://keras3.posit.co/dev/reference/op_isinf.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Test element-wise for positive or negative infinity. — op_isinf","text":"x Input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_isinf.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Test element-wise for positive or negative infinity. — op_isinf","text":"Output boolean tensor.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_isnan.html","id":null,"dir":"Reference","previous_headings":"","what":"Test element-wise for NaN and return result as a boolean tensor. — op_isnan","title":"Test element-wise for NaN and return result as a boolean tensor. — op_isnan","text":"Test element-wise NaN return result boolean tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_isnan.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Test element-wise for NaN and return result as a boolean tensor. — op_isnan","text":"","code":"op_isnan(x)"},{"path":"https://keras3.posit.co/dev/reference/op_isnan.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Test element-wise for NaN and return result as a boolean tensor. — op_isnan","text":"x Input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_isnan.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Test element-wise for NaN and return result as a boolean tensor. — op_isnan","text":"Output boolean tensor.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_istft.html","id":null,"dir":"Reference","previous_headings":"","what":"Inverse Short-Time Fourier Transform along the last axis of the input. — op_istft","title":"Inverse Short-Time Fourier Transform along the last axis of the input. — op_istft","text":"reconstruct original waveform, parameters stft.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_istft.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Inverse Short-Time Fourier Transform along the last axis of the input. — op_istft","text":"","code":"op_istft(   x,   sequence_length,   sequence_stride,   fft_length,   length = NULL,   window = \"hann\",   center = TRUE )"},{"path":"https://keras3.posit.co/dev/reference/op_istft.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Inverse Short-Time Fourier Transform along the last axis of the input. — op_istft","text":"x Tuple real imaginary parts input tensor. tensors list floating type. sequence_length integer representing sequence length. sequence_stride integer representing sequence hop size. fft_length integer representing size FFT produced stft. type int32. length integer representing output clipped exactly length. specified, padding clipping take place. Defaults NULL. window string, tensor window NULL. window string, available values \"hann\" \"hamming\". window tensor, used directly window length must sequence_length. window NULL, windowing used. Defaults \"hann\". center Whether x padded sides t-th sequence centered time t * sequence_stride. Defaults TRUE.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_istft.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Inverse Short-Time Fourier Transform along the last axis of the input. — op_istft","text":"tensor containing inverse Short-Time Fourier Transform along last axis x.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_istft.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Inverse Short-Time Fourier Transform along the last axis of the input. — op_istft","text":"","code":"x <- op_convert_to_tensor(c(0, 1, 2, 3, 4)) op_istft(op_stft(x, 1, 1, 1), 1, 1, 1) ## tf.Tensor([], shape=(0), dtype=float64) # array([0.0, 1.0, 2.0, 3.0, 4.0])"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_kaiser.html","id":null,"dir":"Reference","previous_headings":"","what":"Kaiser window function. — op_kaiser","title":"Kaiser window function. — op_kaiser","text":"Kaiser window defined : w[n] = I0(beta * sqrt(1 - (2 * n / (N - 1) - 1)^2)) / I0(beta) I0 modified zeroth-order Bessel function first kind.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_kaiser.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Kaiser window function. — op_kaiser","text":"","code":"op_kaiser(x, beta)"},{"path":"https://keras3.posit.co/dev/reference/op_kaiser.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Kaiser window function. — op_kaiser","text":"x Length window. Must positive integer. beta Shape parameter window.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_kaiser.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Kaiser window function. — op_kaiser","text":"1D tensor containing window values.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_kaiser.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Kaiser window function. — op_kaiser","text":"","code":"op_kaiser(5, beta = 14) ## tf.Tensor([7.7268669e-06 1.6493215e-01 1.0000000e+00 1.6493215e-01 7.7268669e-06], shape=(5), dtype=float32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_layer_normalization.html","id":null,"dir":"Reference","previous_headings":"","what":"Layer normalization (Ba et al., 2016). — op_layer_normalization","title":"Layer normalization (Ba et al., 2016). — op_layer_normalization","text":"Normalizes activations x example independently centering mean 0 scaling unit variance along specified axis.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_layer_normalization.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Layer normalization (Ba et al., 2016). — op_layer_normalization","text":"","code":"op_layer_normalization(   x,   gamma = NULL,   beta = NULL,   axis = -1L,   epsilon = NULL,   ... )"},{"path":"https://keras3.posit.co/dev/reference/op_layer_normalization.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Layer normalization (Ba et al., 2016). — op_layer_normalization","text":"x Input tensor. gamma Optional scaling factor applied normalized output. beta Optional offset added normalized output. axis Axis axes along compute statistics. Defaults -1. epsilon Small constant added variance numerical stability. ... forward/backward compatibility.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_layer_normalization.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Layer normalization (Ba et al., 2016). — op_layer_normalization","text":"Tensor shape x containing normalized values.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_layer_normalization.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Layer normalization (Ba et al., 2016). — op_layer_normalization","text":"","code":"x <- op_arange(5, dtype = \"float32\") op_layer_normalization(x) ## tf.Tensor([-1.4142134 -0.7071067  0.         0.7071068  1.4142137], shape=(5), dtype=float32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_leaky_relu.html","id":null,"dir":"Reference","previous_headings":"","what":"Leaky version of a Rectified Linear Unit activation function. — op_leaky_relu","title":"Leaky version of a Rectified Linear Unit activation function. — op_leaky_relu","text":"allows small gradient unit active, defined : f(x) = alpha * x x < 0 f(x) = x x >= 0.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_leaky_relu.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Leaky version of a Rectified Linear Unit activation function. — op_leaky_relu","text":"","code":"op_leaky_relu(x, negative_slope = 0.2)"},{"path":"https://keras3.posit.co/dev/reference/op_leaky_relu.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Leaky version of a Rectified Linear Unit activation function. — op_leaky_relu","text":"x Input tensor. negative_slope Slope activation function x < 0. Defaults 0.2.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_leaky_relu.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Leaky version of a Rectified Linear Unit activation function. — op_leaky_relu","text":"tensor shape x.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_leaky_relu.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Leaky version of a Rectified Linear Unit activation function. — op_leaky_relu","text":"","code":"x <- op_array(c(-1., 0., 1.)) op_leaky_relu(x) ## tf.Tensor([-0.2  0.   1. ], shape=(3), dtype=float32) # array([-0.2,  0. ,  1. ], shape=(3,), dtype=float64) x <- seq(-5, 5, .1) plot(x, op_leaky_relu(x),      type = 'l', panel.first = grid())"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_left_shift.html","id":null,"dir":"Reference","previous_headings":"","what":"Shift the bits of an integer to the left. — op_left_shift","title":"Shift the bits of an integer to the left. — op_left_shift","text":"Bits shifted left appending y 0s right x. Since internal representation numbers binary format, operation equivalent multiplying x 2^y.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_left_shift.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Shift the bits of an integer to the left. — op_left_shift","text":"","code":"op_left_shift(x, y)"},{"path":"https://keras3.posit.co/dev/reference/op_left_shift.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Shift the bits of an integer to the left. — op_left_shift","text":"x Input integer tensor. y Input integer tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_left_shift.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Shift the bits of an integer to the left. — op_left_shift","text":"Result tensor.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_less.html","id":null,"dir":"Reference","previous_headings":"","what":"Return the truth value of x1 < x2 element-wise. — op_less","title":"Return the truth value of x1 < x2 element-wise. — op_less","text":"Note function automatically called using R operator < tensor.","code":"(x <- op_arange(4)) ## tf.Tensor([1. 2. 3. 4.], shape=(4), dtype=float32) op_less(x, 2) ## tf.Tensor([ True False False False], shape=(4), dtype=bool) x < 2 ## tf.Tensor([ True False False False], shape=(4), dtype=bool)"},{"path":"https://keras3.posit.co/dev/reference/op_less.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Return the truth value of x1 < x2 element-wise. — op_less","text":"","code":"op_less(x1, x2)"},{"path":"https://keras3.posit.co/dev/reference/op_less.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Return the truth value of x1 < x2 element-wise. — op_less","text":"x1 First input tensor. x2 Second input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_less.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Return the truth value of x1 < x2 element-wise. — op_less","text":"Output tensor, element-wise comparison x1 x2.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_less_equal.html","id":null,"dir":"Reference","previous_headings":"","what":"Return the truth value of x1 <= x2 element-wise. — op_less_equal","title":"Return the truth value of x1 <= x2 element-wise. — op_less_equal","text":"Note function automatically called using R operator <= tensor.","code":"(x <- op_arange(4)) ## tf.Tensor([1. 2. 3. 4.], shape=(4), dtype=float32) op_less_equal(x, 2) ## tf.Tensor([ True  True False False], shape=(4), dtype=bool) x <= 2 ## tf.Tensor([ True  True False False], shape=(4), dtype=bool)"},{"path":"https://keras3.posit.co/dev/reference/op_less_equal.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Return the truth value of x1 <= x2 element-wise. — op_less_equal","text":"","code":"op_less_equal(x1, x2)"},{"path":"https://keras3.posit.co/dev/reference/op_less_equal.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Return the truth value of x1 <= x2 element-wise. — op_less_equal","text":"x1 First input tensor. x2 Second input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_less_equal.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Return the truth value of x1 <= x2 element-wise. — op_less_equal","text":"Output tensor, element-wise comparison x1 x2.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_linspace.html","id":null,"dir":"Reference","previous_headings":"","what":"Return evenly spaced numbers over a specified interval. — op_linspace","title":"Return evenly spaced numbers over a specified interval. — op_linspace","text":"Returns num evenly spaced samples, calculated interval [start, stop]. endpoint interval can optionally excluded.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_linspace.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Return evenly spaced numbers over a specified interval. — op_linspace","text":"","code":"op_linspace(   start,   stop,   num = 50L,   endpoint = TRUE,   retstep = FALSE,   dtype = NULL,   axis = 1L )"},{"path":"https://keras3.posit.co/dev/reference/op_linspace.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Return evenly spaced numbers over a specified interval. — op_linspace","text":"start starting value sequence. stop end value sequence, unless endpoint set FALSE. case, sequence consists last num + 1 evenly spaced samples, stop excluded. Note step size changes endpoint FALSE. num Number samples generate. Defaults 50. Must non-negative. endpoint TRUE, stop last sample. Otherwise, included. Defaults TRUE. retstep TRUE, return (samples, step), step spacing samples. dtype type output tensor. axis axis result store samples. Relevant start stop array-like. Defaults 1, first axis.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_linspace.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Return evenly spaced numbers over a specified interval. — op_linspace","text":"tensor evenly spaced numbers. retstep TRUE, returns (samples, step)","code":""},{"path":"https://keras3.posit.co/dev/reference/op_linspace.html","id":"note","dir":"Reference","previous_headings":"","what":"Note","title":"Return evenly spaced numbers over a specified interval. — op_linspace","text":"Torch backend support axis argument.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_log.html","id":null,"dir":"Reference","previous_headings":"","what":"Natural logarithm, element-wise. — op_log","title":"Natural logarithm, element-wise. — op_log","text":"Natural logarithm, element-wise.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_log.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Natural logarithm, element-wise. — op_log","text":"","code":"op_log(x)"},{"path":"https://keras3.posit.co/dev/reference/op_log.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Natural logarithm, element-wise. — op_log","text":"x Input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_log.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Natural logarithm, element-wise. — op_log","text":"Output tensor, element-wise natural logarithm x.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_log10.html","id":null,"dir":"Reference","previous_headings":"","what":"Return the base 10 logarithm of the input tensor, element-wise. — op_log10","title":"Return the base 10 logarithm of the input tensor, element-wise. — op_log10","text":"Return base 10 logarithm input tensor, element-wise.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_log10.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Return the base 10 logarithm of the input tensor, element-wise. — op_log10","text":"","code":"op_log10(x)"},{"path":"https://keras3.posit.co/dev/reference/op_log10.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Return the base 10 logarithm of the input tensor, element-wise. — op_log10","text":"x Input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_log10.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Return the base 10 logarithm of the input tensor, element-wise. — op_log10","text":"Output tensor, element-wise base 10 logarithm x.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_log1p.html","id":null,"dir":"Reference","previous_headings":"","what":"Returns the natural logarithm of one plus the x, element-wise. — op_log1p","title":"Returns the natural logarithm of one plus the x, element-wise. — op_log1p","text":"Calculates log(1 + x).","code":""},{"path":"https://keras3.posit.co/dev/reference/op_log1p.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Returns the natural logarithm of one plus the x, element-wise. — op_log1p","text":"","code":"op_log1p(x)"},{"path":"https://keras3.posit.co/dev/reference/op_log1p.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Returns the natural logarithm of one plus the x, element-wise. — op_log1p","text":"x Input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_log1p.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Returns the natural logarithm of one plus the x, element-wise. — op_log1p","text":"Output tensor, element-wise natural logarithm 1 + x.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_log2.html","id":null,"dir":"Reference","previous_headings":"","what":"Base-2 logarithm of x, element-wise. — op_log2","title":"Base-2 logarithm of x, element-wise. — op_log2","text":"Base-2 logarithm x, element-wise.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_log2.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Base-2 logarithm of x, element-wise. — op_log2","text":"","code":"op_log2(x)"},{"path":"https://keras3.posit.co/dev/reference/op_log2.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Base-2 logarithm of x, element-wise. — op_log2","text":"x Input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_log2.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Base-2 logarithm of x, element-wise. — op_log2","text":"Output tensor, element-wise base-2 logarithm x.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_log_sigmoid.html","id":null,"dir":"Reference","previous_headings":"","what":"Logarithm of the sigmoid activation function. — op_log_sigmoid","title":"Logarithm of the sigmoid activation function. — op_log_sigmoid","text":"defined f(x) = log(1 / (1 + exp(-x))).","code":""},{"path":"https://keras3.posit.co/dev/reference/op_log_sigmoid.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Logarithm of the sigmoid activation function. — op_log_sigmoid","text":"","code":"op_log_sigmoid(x)"},{"path":"https://keras3.posit.co/dev/reference/op_log_sigmoid.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Logarithm of the sigmoid activation function. — op_log_sigmoid","text":"x Input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_log_sigmoid.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Logarithm of the sigmoid activation function. — op_log_sigmoid","text":"tensor shape x.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_log_sigmoid.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Logarithm of the sigmoid activation function. — op_log_sigmoid","text":"","code":"x <- op_convert_to_tensor(c(-0.541391, 0.0, 0.50, 5.0)) op_log_sigmoid(x) ## tf.Tensor([-1.00004181 -0.69314718 -0.47407698 -0.00671535], shape=(4), dtype=float64)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_log_softmax.html","id":null,"dir":"Reference","previous_headings":"","what":"Log-softmax activation function. — op_log_softmax","title":"Log-softmax activation function. — op_log_softmax","text":"defined : f(x) = x - max(x) - log(sum(exp(x - max(x))))","code":""},{"path":"https://keras3.posit.co/dev/reference/op_log_softmax.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Log-softmax activation function. — op_log_softmax","text":"","code":"op_log_softmax(x, axis = -1L)"},{"path":"https://keras3.posit.co/dev/reference/op_log_softmax.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Log-softmax activation function. — op_log_softmax","text":"x Input tensor. axis Integer, axis along log-softmax applied. Defaults -1.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_log_softmax.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Log-softmax activation function. — op_log_softmax","text":"tensor shape x.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_log_softmax.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Log-softmax activation function. — op_log_softmax","text":"","code":"x <- op_array(c(-1., 0., 1.)) op_log_softmax(x) ## tf.Tensor([-2.407606   -1.4076059  -0.40760595], shape=(3), dtype=float32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_logaddexp.html","id":null,"dir":"Reference","previous_headings":"","what":"Logarithm of the sum of exponentiations of the inputs. — op_logaddexp","title":"Logarithm of the sum of exponentiations of the inputs. — op_logaddexp","text":"Calculates log(exp(x1) + exp(x2)).","code":""},{"path":"https://keras3.posit.co/dev/reference/op_logaddexp.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Logarithm of the sum of exponentiations of the inputs. — op_logaddexp","text":"","code":"op_logaddexp(x1, x2)"},{"path":"https://keras3.posit.co/dev/reference/op_logaddexp.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Logarithm of the sum of exponentiations of the inputs. — op_logaddexp","text":"x1 Input tensor. x2 Input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_logaddexp.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Logarithm of the sum of exponentiations of the inputs. — op_logaddexp","text":"Output tensor, element-wise logarithm sum exponentiations inputs.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_logdet.html","id":null,"dir":"Reference","previous_headings":"","what":"Computes log of the determinant of a hermitian positive definite matrix. — op_logdet","title":"Computes log of the determinant of a hermitian positive definite matrix. — op_logdet","text":"Computes log determinant hermitian positive definite matrix.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_logdet.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes log of the determinant of a hermitian positive definite matrix. — op_logdet","text":"","code":"op_logdet(x)"},{"path":"https://keras3.posit.co/dev/reference/op_logdet.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Computes log of the determinant of a hermitian positive definite matrix. — op_logdet","text":"x Input matrix. must 2D square.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_logdet.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Computes log of the determinant of a hermitian positive definite matrix. — op_logdet","text":"natural log determinant matrix.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_logical_and.html","id":null,"dir":"Reference","previous_headings":"","what":"Computes the element-wise logical AND of the given input tensors. — op_logical_and","title":"Computes the element-wise logical AND of the given input tensors. — op_logical_and","text":"Zeros treated FALSE non-zeros treated TRUE.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_logical_and.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes the element-wise logical AND of the given input tensors. — op_logical_and","text":"","code":"op_logical_and(x1, x2)"},{"path":"https://keras3.posit.co/dev/reference/op_logical_and.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Computes the element-wise logical AND of the given input tensors. — op_logical_and","text":"x1 Input tensor. x2 Input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_logical_and.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Computes the element-wise logical AND of the given input tensors. — op_logical_and","text":"Output tensor, element-wise logical inputs.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_logical_and.html","id":"details","dir":"Reference","previous_headings":"","what":"Details","title":"Computes the element-wise logical AND of the given input tensors. — op_logical_and","text":"Note function automatically called using R operator & tensor.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_logical_not.html","id":null,"dir":"Reference","previous_headings":"","what":"Computes the element-wise NOT of the given input tensor. — op_logical_not","title":"Computes the element-wise NOT of the given input tensor. — op_logical_not","text":"Zeros treated FALSE non-zeros treated TRUE. Note function automatically called using R operator ! tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_logical_not.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes the element-wise NOT of the given input tensor. — op_logical_not","text":"","code":"op_logical_not(x)"},{"path":"https://keras3.posit.co/dev/reference/op_logical_not.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Computes the element-wise NOT of the given input tensor. — op_logical_not","text":"x Input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_logical_not.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Computes the element-wise NOT of the given input tensor. — op_logical_not","text":"Output tensor, element-wise logical input.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_logical_or.html","id":null,"dir":"Reference","previous_headings":"","what":"Computes the element-wise logical OR of the given input tensors. — op_logical_or","title":"Computes the element-wise logical OR of the given input tensors. — op_logical_or","text":"Zeros treated FALSE non-zeros treated TRUE. Note function automatically called using R operator | tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_logical_or.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes the element-wise logical OR of the given input tensors. — op_logical_or","text":"","code":"op_logical_or(x1, x2)"},{"path":"https://keras3.posit.co/dev/reference/op_logical_or.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Computes the element-wise logical OR of the given input tensors. — op_logical_or","text":"x1 Input tensor. x2 Input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_logical_or.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Computes the element-wise logical OR of the given input tensors. — op_logical_or","text":"Output tensor, element-wise logical inputs.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_logical_xor.html","id":null,"dir":"Reference","previous_headings":"","what":"Compute the truth value of x1 XOR x2, element-wise. — op_logical_xor","title":"Compute the truth value of x1 XOR x2, element-wise. — op_logical_xor","text":"Compute truth value x1 XOR x2, element-wise.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_logical_xor.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Compute the truth value of x1 XOR x2, element-wise. — op_logical_xor","text":"","code":"op_logical_xor(x1, x2)"},{"path":"https://keras3.posit.co/dev/reference/op_logical_xor.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Compute the truth value of x1 XOR x2, element-wise. — op_logical_xor","text":"x1 First input tensor. x2 Second input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_logical_xor.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Compute the truth value of x1 XOR x2, element-wise. — op_logical_xor","text":"Output boolean tensor.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_logspace.html","id":null,"dir":"Reference","previous_headings":"","what":"Returns numbers spaced evenly on a log scale. — op_logspace","title":"Returns numbers spaced evenly on a log scale. — op_logspace","text":"linear space, sequence starts base ** start ends base ** stop (see endpoint ).","code":""},{"path":"https://keras3.posit.co/dev/reference/op_logspace.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Returns numbers spaced evenly on a log scale. — op_logspace","text":"","code":"op_logspace(   start,   stop,   num = 50L,   endpoint = TRUE,   base = 10L,   dtype = NULL,   axis = 1L )"},{"path":"https://keras3.posit.co/dev/reference/op_logspace.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Returns numbers spaced evenly on a log scale. — op_logspace","text":"start starting value sequence. stop final value sequence, unless endpoint FALSE. case, num + 1 values spaced interval log-space, last (sequence length num) returned. num Number samples generate. Defaults 50. endpoint TRUE, stop last sample. Otherwise, included. Defaults TRUE. base base log space. Defaults 10. dtype type output tensor. axis axis result store samples. Relevant start stop array-like.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_logspace.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Returns numbers spaced evenly on a log scale. — op_logspace","text":"tensor evenly spaced samples log scale.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_logspace.html","id":"note","dir":"Reference","previous_headings":"","what":"Note","title":"Returns numbers spaced evenly on a log scale. — op_logspace","text":"Torch backend support axis argument.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_logsumexp.html","id":null,"dir":"Reference","previous_headings":"","what":"Computes the logarithm of sum of exponentials of elements in a tensor. — op_logsumexp","title":"Computes the logarithm of sum of exponentials of elements in a tensor. — op_logsumexp","text":"Computes logarithm sum exponentials elements tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_logsumexp.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes the logarithm of sum of exponentials of elements in a tensor. — op_logsumexp","text":"","code":"op_logsumexp(x, axis = NULL, keepdims = FALSE)"},{"path":"https://keras3.posit.co/dev/reference/op_logsumexp.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Computes the logarithm of sum of exponentials of elements in a tensor. — op_logsumexp","text":"x Input tensor. axis integer list integers specifying axis/axes along compute sum. NULL, sum computed elements. Defaults NULL. keepdims boolean indicating whether keep dimensions input tensor computing sum. Defaults FALSE.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_logsumexp.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Computes the logarithm of sum of exponentials of elements in a tensor. — op_logsumexp","text":"tensor containing logarithm sum exponentials elements x.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_logsumexp.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Computes the logarithm of sum of exponentials of elements in a tensor. — op_logsumexp","text":"","code":"x <- op_convert_to_tensor(c(1, 2, 3)) op_logsumexp(x) ## tf.Tensor(3.4076059644443806, shape=(), dtype=float64)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_lstsq.html","id":null,"dir":"Reference","previous_headings":"","what":"Return the least-squares solution to a linear matrix equation. — op_lstsq","title":"Return the least-squares solution to a linear matrix equation. — op_lstsq","text":"Computes vector x approximately solves equation %*% x = b. equation may -, well-, -determined (.e., number linearly independent rows can less , equal , greater number linearly independent columns). square full rank, x (round-error) exact solution equation. Else, x minimizes L2 norm b - %*% x. multiple minimizing solutions, one smallest L2 norm  returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_lstsq.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Return the least-squares solution to a linear matrix equation. — op_lstsq","text":"","code":"op_lstsq(a, b, rcond = NULL)"},{"path":"https://keras3.posit.co/dev/reference/op_lstsq.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Return the least-squares solution to a linear matrix equation. — op_lstsq","text":"\"Coefficient\" matrix shape (M, N). b Ordinate \"dependent variable\" values, shape (M) (M, K). b two-dimensional, least-squares solution calculated K columns b. rcond Cut-ratio small singular values . purposes rank determination, singular values treated zero smaller rcond times largest singular value .","code":""},{"path":"https://keras3.posit.co/dev/reference/op_lstsq.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Return the least-squares solution to a linear matrix equation. — op_lstsq","text":"Tensor shape (N) (N, K) containing least-squares solutions. NOTE: output differs numpy.linalg.lstsq(). NumPy returns tuple four elements, first least-squares solutions others essentially never used. Keras returns first value. done ensure consistency across backends (achieved values) simplify API.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_lu_factor.html","id":null,"dir":"Reference","previous_headings":"","what":"Computes the lower-upper decomposition of a square matrix. — op_lu_factor","title":"Computes the lower-upper decomposition of a square matrix. — op_lu_factor","text":"Computes lower-upper decomposition square matrix.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_lu_factor.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes the lower-upper decomposition of a square matrix. — op_lu_factor","text":"","code":"op_lu_factor(x)"},{"path":"https://keras3.posit.co/dev/reference/op_lu_factor.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Computes the lower-upper decomposition of a square matrix. — op_lu_factor","text":"x tensor shape (..., M, M).","code":""},{"path":"https://keras3.posit.co/dev/reference/op_lu_factor.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Computes the lower-upper decomposition of a square matrix. — op_lu_factor","text":"tuple two tensors: tensor shape (..., M, M) containing lower upper triangular matrices tensor shape (..., M) containing pivots.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_map.html","id":null,"dir":"Reference","previous_headings":"","what":"Map a function over leading array axes. — op_map","title":"Map a function over leading array axes. — op_map","text":"Like purrr::map() base::lapply(), except inputs outputs form stacked arrays. Consider using op_vectorized_map() transform instead, unless need apply function element element reduced memory usage heterogeneous computation control flow primitives. xs array type, semantics op_map() match implementation:","code":"op_map <- function(xs, f) {   xs |>     op_unstack() |>     lapply(f) |>     op_stack() }"},{"path":"https://keras3.posit.co/dev/reference/op_map.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Map a function over leading array axes. — op_map","text":"","code":"op_map(xs, f)"},{"path":"https://keras3.posit.co/dev/reference/op_map.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Map a function over leading array axes. — op_map","text":"xs Values map along leading axis. f Callable defines function apply element-wise first axis axes xs.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_map.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Map a function over leading array axes. — op_map","text":"Mapped values.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_map.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Map a function over leading array axes. — op_map","text":"","code":"f <- function(x) x^2 xs <- op_arange(10) ys <- op_map(xs, f) ys ## tf.Tensor([  1.   4.   9.  16.  25.  36.  49.  64.  81. 100.], shape=(10), dtype=float32) f <- function(x) list(y1 = x^2, y2 = x * 10)  # Can have nested outputs ys <- op_map(xs, f) ys$y1 ## tf.Tensor([  1.   4.   9.  16.  25.  36.  49.  64.  81. 100.], shape=(10), dtype=float32) ys$y2 ## tf.Tensor([ 10.  20.  30.  40.  50.  60.  70.  80.  90. 100.], shape=(10), dtype=float32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_matmul.html","id":null,"dir":"Reference","previous_headings":"","what":"Matrix product of two tensors. — op_matmul","title":"Matrix product of two tensors. — op_matmul","text":"tensors 1-dimensional, dot product (scalar) returned. either tensor N-D, N > 2, treated stack matrices residing last two indexes broadcast accordingly. first tensor 1-D, promoted matrix prepending 1 dimensions. matrix multiplication prepended 1 removed. second tensor 1-D, promoted matrix appending 1 dimensions. matrix multiplication appended 1 removed.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_matmul.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Matrix product of two tensors. — op_matmul","text":"","code":"op_matmul(x1, x2)"},{"path":"https://keras3.posit.co/dev/reference/op_matmul.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Matrix product of two tensors. — op_matmul","text":"x1 First tensor. x2 Second tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_matmul.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Matrix product of two tensors. — op_matmul","text":"Output tensor, matrix product inputs.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_max.html","id":null,"dir":"Reference","previous_headings":"","what":"Return the maximum of a tensor or maximum along an axis. — op_max","title":"Return the maximum of a tensor or maximum along an axis. — op_max","text":"Return maximum tensor maximum along axis.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_max.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Return the maximum of a tensor or maximum along an axis. — op_max","text":"","code":"op_max(x, axis = NULL, keepdims = FALSE, initial = NULL)"},{"path":"https://keras3.posit.co/dev/reference/op_max.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Return the maximum of a tensor or maximum along an axis. — op_max","text":"x Input tensor. axis Axis axes along operate. default, flattened input used. keepdims set TRUE, axes reduced left result dimensions size one. Defaults FALSE. initial minimum value output element. Defaults NULL.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_max.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Return the maximum of a tensor or maximum along an axis. — op_max","text":"Maximum x.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_max.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Return the maximum of a tensor or maximum along an axis. — op_max","text":"","code":"(x <- op_convert_to_tensor(rbind(c(1, 3, 5), c(1, 5, 2)))) ## tf.Tensor( ## [[1. 3. 5.] ##  [1. 5. 2.]], shape=(2, 3), dtype=float64) op_max(x) ## tf.Tensor(5.0, shape=(), dtype=float64) op_max(x, axis = 1) ## tf.Tensor([1. 5. 5.], shape=(3), dtype=float64) op_max(x, axis = 1, keepdims = TRUE) ## tf.Tensor([[1. 5. 5.]], shape=(1, 3), dtype=float64)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_max_pool.html","id":null,"dir":"Reference","previous_headings":"","what":"Max pooling operation. — op_max_pool","title":"Max pooling operation. — op_max_pool","text":"Max pooling operation.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_max_pool.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Max pooling operation. — op_max_pool","text":"","code":"op_max_pool(   inputs,   pool_size,   strides = NULL,   padding = \"valid\",   data_format = NULL )"},{"path":"https://keras3.posit.co/dev/reference/op_max_pool.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Max pooling operation. — op_max_pool","text":"inputs Tensor rank N+2. inputs shape (batch_size,) + inputs_spatial_shape + (num_channels,) data_format = \"channels_last\", (batch_size, num_channels) + inputs_spatial_shape data_format = \"channels_first\". Pooling happens spatial dimensions . pool_size int tuple/list integers size len(inputs_spatial_shape), specifying size pooling window spatial dimension input tensor. pool_size int, every spatial dimension shares pool_size. strides int tuple/list integers size len(inputs_spatial_shape). stride sliding window spatial dimension input tensor. strides int, every spatial dimension shares strides. padding string, either \"valid\" \"\". \"valid\" means padding applied, \"\" results padding evenly left/right /input output height/width dimension input strides = 1. data_format string, either \"channels_last\" \"channels_first\". data_format determines ordering dimensions inputs. data_format = \"channels_last\", inputs shape (batch_size, ..., channels) data_format = \"channels_first\", inputs shape (batch_size, channels, ...).","code":""},{"path":"https://keras3.posit.co/dev/reference/op_max_pool.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Max pooling operation. — op_max_pool","text":"tensor rank N+2, result max pooling operation.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_maximum.html","id":null,"dir":"Reference","previous_headings":"","what":"Element-wise maximum of x1 and x2. — op_maximum","title":"Element-wise maximum of x1 and x2. — op_maximum","text":"Element-wise maximum x1 x2.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_maximum.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Element-wise maximum of x1 and x2. — op_maximum","text":"","code":"op_maximum(x1, x2)  op_pmax(x1, x2)"},{"path":"https://keras3.posit.co/dev/reference/op_maximum.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Element-wise maximum of x1 and x2. — op_maximum","text":"x1 First tensor. x2 Second tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_maximum.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Element-wise maximum of x1 and x2. — op_maximum","text":"Output tensor, element-wise maximum x1 x2.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_mean.html","id":null,"dir":"Reference","previous_headings":"","what":"Compute the arithmetic mean along the specified axes. — op_mean","title":"Compute the arithmetic mean along the specified axes. — op_mean","text":"Compute arithmetic mean along specified axes.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_mean.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Compute the arithmetic mean along the specified axes. — op_mean","text":"","code":"op_mean(x, axis = NULL, keepdims = FALSE)"},{"path":"https://keras3.posit.co/dev/reference/op_mean.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Compute the arithmetic mean along the specified axes. — op_mean","text":"x Input tensor. axis Axis axes along means computed. default compute mean flattened tensor. keepdims set TRUE, axes reduced left result dimensions size one.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_mean.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Compute the arithmetic mean along the specified axes. — op_mean","text":"Output tensor containing mean values.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_median.html","id":null,"dir":"Reference","previous_headings":"","what":"Compute the median along the specified axis. — op_median","title":"Compute the median along the specified axis. — op_median","text":"Compute median along specified axis.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_median.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Compute the median along the specified axis. — op_median","text":"","code":"op_median(x, axis = NULL, keepdims = FALSE)"},{"path":"https://keras3.posit.co/dev/reference/op_median.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Compute the median along the specified axis. — op_median","text":"x Input tensor. axis Axis axes along medians computed. Defaults axis = NULL compute median(s) along flattened version array. keepdims set TRUE, axes reduce left result dimensions size one.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_median.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Compute the median along the specified axis. — op_median","text":"output tensor.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_meshgrid.html","id":null,"dir":"Reference","previous_headings":"","what":"Creates grids of coordinates from coordinate vectors. — op_meshgrid","title":"Creates grids of coordinates from coordinate vectors. — op_meshgrid","text":"Given N 1-D tensors T0, T1, ..., TN-1 inputs corresponding lengths S0, S1, ..., SN-1, creates N N-dimensional tensors G0, G1, ..., GN-1 shape (S0, ..., SN-1) output Gi constructed expanding Ti result shape.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_meshgrid.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Creates grids of coordinates from coordinate vectors. — op_meshgrid","text":"","code":"op_meshgrid(..., indexing = \"xy\")"},{"path":"https://keras3.posit.co/dev/reference/op_meshgrid.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Creates grids of coordinates from coordinate vectors. — op_meshgrid","text":"... 1-D tensors representing coordinates grid. indexing \"xy\" \"ij\". \"xy\" cartesian; \"ij\" matrix indexing output. Defaults \"xy\".","code":""},{"path":"https://keras3.posit.co/dev/reference/op_meshgrid.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Creates grids of coordinates from coordinate vectors. — op_meshgrid","text":"Sequence N tensors.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_meshgrid.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Creates grids of coordinates from coordinate vectors. — op_meshgrid","text":"","code":"x <- op_array(c(1, 2, 3), \"int32\") y <- op_array(c(4, 5, 6), \"int32\") c(grid_x, grid_y) %<-% op_meshgrid(x, y, indexing = \"ij\") grid_x ## tf.Tensor( ## [[1 1 1] ##  [2 2 2] ##  [3 3 3]], shape=(3, 3), dtype=int32) # array([[1, 1, 1], #        [2, 2, 2], #        [3, 3, 3])) grid_y ## tf.Tensor( ## [[4 5 6] ##  [4 5 6] ##  [4 5 6]], shape=(3, 3), dtype=int32) # array([[4, 5, 6], #        [4, 5, 6], #        [4, 5, 6]))"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_min.html","id":null,"dir":"Reference","previous_headings":"","what":"Return the minimum of a tensor or minimum along an axis. — op_min","title":"Return the minimum of a tensor or minimum along an axis. — op_min","text":"Return minimum tensor minimum along axis.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_min.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Return the minimum of a tensor or minimum along an axis. — op_min","text":"","code":"op_min(x, axis = NULL, keepdims = FALSE, initial = NULL)"},{"path":"https://keras3.posit.co/dev/reference/op_min.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Return the minimum of a tensor or minimum along an axis. — op_min","text":"x Input tensor. axis Axis axes along operate. default, flattened input used. keepdims set TRUE, axes reduced left result dimensions size one. Defaults FALSE. initial maximum value output element. Defaults NULL.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_min.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Return the minimum of a tensor or minimum along an axis. — op_min","text":"Minimum x.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_min.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Return the minimum of a tensor or minimum along an axis. — op_min","text":"","code":"(x <- op_convert_to_tensor(rbind(c(1, 3, 5), c(1, 5, 2)))) ## tf.Tensor( ## [[1. 3. 5.] ##  [1. 5. 2.]], shape=(2, 3), dtype=float64) op_min(x) ## tf.Tensor(1.0, shape=(), dtype=float64) op_min(x, axis = 1) ## tf.Tensor([1. 3. 2.], shape=(3), dtype=float64) op_min(x, axis = 1, keepdims = TRUE) ## tf.Tensor([[1. 3. 2.]], shape=(1, 3), dtype=float64)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_minimum.html","id":null,"dir":"Reference","previous_headings":"","what":"Element-wise minimum of x1 and x2. — op_minimum","title":"Element-wise minimum of x1 and x2. — op_minimum","text":"Element-wise minimum x1 x2.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_minimum.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Element-wise minimum of x1 and x2. — op_minimum","text":"","code":"op_minimum(x1, x2)  op_pmin(x1, x2)"},{"path":"https://keras3.posit.co/dev/reference/op_minimum.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Element-wise minimum of x1 and x2. — op_minimum","text":"x1 First tensor. x2 Second tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_minimum.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Element-wise minimum of x1 and x2. — op_minimum","text":"Output tensor, element-wise minimum x1 x2.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_mod.html","id":null,"dir":"Reference","previous_headings":"","what":"Returns the element-wise remainder of division. — op_mod","title":"Returns the element-wise remainder of division. — op_mod","text":"Note function automatically called using R operator %% tensor.","code":"(x <- op_arange(10)) ## tf.Tensor([ 1.  2.  3.  4.  5.  6.  7.  8.  9. 10.], shape=(10), dtype=float32) op_mod(x, 3) ## tf.Tensor([1. 2. 0. 1. 2. 0. 1. 2. 0. 1.], shape=(10), dtype=float32) x %% 3 ## tf.Tensor([1. 2. 0. 1. 2. 0. 1. 2. 0. 1.], shape=(10), dtype=float32)"},{"path":"https://keras3.posit.co/dev/reference/op_mod.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Returns the element-wise remainder of division. — op_mod","text":"","code":"op_mod(x1, x2)"},{"path":"https://keras3.posit.co/dev/reference/op_mod.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Returns the element-wise remainder of division. — op_mod","text":"x1 First tensor. x2 Second tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_mod.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Returns the element-wise remainder of division. — op_mod","text":"Output tensor, element-wise remainder division.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_moments.html","id":null,"dir":"Reference","previous_headings":"","what":"Calculates the mean and variance of x. — op_moments","title":"Calculates the mean and variance of x. — op_moments","text":"mean variance calculated aggregating contents x across axes. x 1-D axes = c(1) just mean variance vector.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_moments.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Calculates the mean and variance of x. — op_moments","text":"","code":"op_moments(x, axes, keepdims = FALSE, synchronized = FALSE)"},{"path":"https://keras3.posit.co/dev/reference/op_moments.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Calculates the mean and variance of x. — op_moments","text":"x Input tensor. axes list axes compute mean variance. keepdims set TRUE, axes reduced left result dimensions size one. synchronized applicable TensorFlow backend. TRUE, synchronizes global batch statistics (mean variance) across devices training step distributed training strategy. FALSE, replica uses local batch statistics.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_moments.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Calculates the mean and variance of x. — op_moments","text":"list containing two tensors - mean variance.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_moments.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Calculates the mean and variance of x. — op_moments","text":"","code":"x <- op_convert_to_tensor(c(0, 1, 2, 3, 100), dtype = \"float32\") op_moments(x, axes = c(1)) ## [[1]] ## tf.Tensor(21.2, shape=(), dtype=float32) ## ## [[2]] ## tf.Tensor(1553.3601, shape=(), dtype=float32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_moveaxis.html","id":null,"dir":"Reference","previous_headings":"","what":"Move axes of a tensor to new positions. — op_moveaxis","title":"Move axes of a tensor to new positions. — op_moveaxis","text":"axes remain original order.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_moveaxis.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Move axes of a tensor to new positions. — op_moveaxis","text":"","code":"op_moveaxis(x, source, destination)"},{"path":"https://keras3.posit.co/dev/reference/op_moveaxis.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Move axes of a tensor to new positions. — op_moveaxis","text":"x Tensor whose axes reordered. source Original positions axes move. must unique. destination Destinations positions original axes. must also unique.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_moveaxis.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Move axes of a tensor to new positions. — op_moveaxis","text":"Tensor moved axes.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_multi_hot.html","id":null,"dir":"Reference","previous_headings":"","what":"Encodes integer labels as multi-hot vectors. — op_multi_hot","title":"Encodes integer labels as multi-hot vectors. — op_multi_hot","text":"function encodes integer labels multi-hot vectors, label mapped binary value resulting vector.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_multi_hot.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Encodes integer labels as multi-hot vectors. — op_multi_hot","text":"","code":"op_multi_hot(   inputs,   num_classes,   axis = -1L,   dtype = NULL,   sparse = FALSE,   ... )"},{"path":"https://keras3.posit.co/dev/reference/op_multi_hot.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Encodes integer labels as multi-hot vectors. — op_multi_hot","text":"inputs Tensor integer labels converted multi-hot vectors. num_classes Integer, total number unique classes. axis (optional) Axis along multi-hot encoding added. Defaults -1, corresponds last dimension. dtype (optional) data type resulting tensor. Default backend's float type. sparse Whether return sparse tensor; backends support sparse tensors. ... forward/backwards compatability","code":""},{"path":"https://keras3.posit.co/dev/reference/op_multi_hot.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Encodes integer labels as multi-hot vectors. — op_multi_hot","text":"Tensor: multi-hot encoded tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_multi_hot.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Encodes integer labels as multi-hot vectors. — op_multi_hot","text":"","code":"data <- op_convert_to_tensor(c(0, 4)) op_multi_hot(data, num_classes = 5) ## tf.Tensor([1. 0. 0. 0. 1.], shape=(5), dtype=float32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_multiply.html","id":null,"dir":"Reference","previous_headings":"","what":"Multiply arguments element-wise. — op_multiply","title":"Multiply arguments element-wise. — op_multiply","text":"Note function automatically called using R operator * tensor.","code":"(x <- op_arange(4)) ## tf.Tensor([1. 2. 3. 4.], shape=(4), dtype=float32) op_multiply(x, x) ## tf.Tensor([ 1.  4.  9. 16.], shape=(4), dtype=float32) x * x ## tf.Tensor([ 1.  4.  9. 16.], shape=(4), dtype=float32)"},{"path":"https://keras3.posit.co/dev/reference/op_multiply.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Multiply arguments element-wise. — op_multiply","text":"","code":"op_multiply(x1, x2)"},{"path":"https://keras3.posit.co/dev/reference/op_multiply.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Multiply arguments element-wise. — op_multiply","text":"x1 First input tensor. x2 Second input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_multiply.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Multiply arguments element-wise. — op_multiply","text":"Output tensor, element-wise product x1 x2.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_nan_to_num.html","id":null,"dir":"Reference","previous_headings":"","what":"Replace NaN with zero and infinity with large finite numbers. — op_nan_to_num","title":"Replace NaN with zero and infinity with large finite numbers. — op_nan_to_num","text":"Replace NaN zero infinity large finite numbers.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_nan_to_num.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Replace NaN with zero and infinity with large finite numbers. — op_nan_to_num","text":"","code":"op_nan_to_num(x, nan = 0, posinf = NULL, neginf = NULL)"},{"path":"https://keras3.posit.co/dev/reference/op_nan_to_num.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Replace NaN with zero and infinity with large finite numbers. — op_nan_to_num","text":"x Input data. nan Optional float int. Value replace NaN entries . posinf Optional float int. Value replace positive infinity . neginf Optional float int. Value replace negative infinity .","code":""},{"path":"https://keras3.posit.co/dev/reference/op_nan_to_num.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Replace NaN with zero and infinity with large finite numbers. — op_nan_to_num","text":"x, non-finite values replaced.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_nan_to_num.html","id":"example","dir":"Reference","previous_headings":"","what":"Example","title":"Replace NaN with zero and infinity with large finite numbers. — op_nan_to_num","text":"","code":"(x <- op_convert_to_tensor(c(1, NaN, -Inf, Inf))) ## tf.Tensor([  1.  nan -inf  inf], shape=(4), dtype=float64) op_nan_to_num(x) ## tf.Tensor([ 1.00000000e+000  0.00000000e+000 -1.79769313e+308  1.79769313e+308], shape=(4), dtype=float64) op_nan_to_num(x, nan = -1, posinf = 2, neginf = -2) ## tf.Tensor([ 1. -1. -2.  2.], shape=(4), dtype=float64)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_ndim.html","id":null,"dir":"Reference","previous_headings":"","what":"Return the number of dimensions of a tensor. — op_ndim","title":"Return the number of dimensions of a tensor. — op_ndim","text":"Return number dimensions tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_ndim.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Return the number of dimensions of a tensor. — op_ndim","text":"","code":"op_ndim(x)"},{"path":"https://keras3.posit.co/dev/reference/op_ndim.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Return the number of dimensions of a tensor. — op_ndim","text":"x Input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_ndim.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Return the number of dimensions of a tensor. — op_ndim","text":"number dimensions x.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_negative.html","id":null,"dir":"Reference","previous_headings":"","what":"Numerical negative, element-wise. — op_negative","title":"Numerical negative, element-wise. — op_negative","text":"Note function automatically called using unary R operator - tensor.","code":"(x <- op_arange(4)) ## tf.Tensor([1. 2. 3. 4.], shape=(4), dtype=float32) op_negative(x) ## tf.Tensor([-1. -2. -3. -4.], shape=(4), dtype=float32) -x ## tf.Tensor([-1. -2. -3. -4.], shape=(4), dtype=float32)"},{"path":"https://keras3.posit.co/dev/reference/op_negative.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Numerical negative, element-wise. — op_negative","text":"","code":"op_negative(x)"},{"path":"https://keras3.posit.co/dev/reference/op_negative.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Numerical negative, element-wise. — op_negative","text":"x Input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_negative.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Numerical negative, element-wise. — op_negative","text":"Output tensor, y = -x.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_nonzero.html","id":null,"dir":"Reference","previous_headings":"","what":"Return the indices of the elements that are non-zero. — op_nonzero","title":"Return the indices of the elements that are non-zero. — op_nonzero","text":"Return indices elements non-zero.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_nonzero.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Return the indices of the elements that are non-zero. — op_nonzero","text":"","code":"op_nonzero(x)"},{"path":"https://keras3.posit.co/dev/reference/op_nonzero.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Return the indices of the elements that are non-zero. — op_nonzero","text":"x Input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_nonzero.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Return the indices of the elements that are non-zero. — op_nonzero","text":"Indices elements non-zero.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_nonzero.html","id":"example","dir":"Reference","previous_headings":"","what":"Example","title":"Return the indices of the elements that are non-zero. — op_nonzero","text":"op_nonzero() indices can used <tensor>@r[","code":"(x <- op_scatter(indices = rbind(1, 5, 10), values = c(1, 2, 3), shape = c(10))) ## tf.Tensor([1. 0. 0. 0. 2. 0. 0. 0. 0. 3.], shape=(10), dtype=float64) (nz <- op_nonzero(x)) ## tf.Tensor( ## [[ 1] ##  [ 5] ##  [10]], shape=(3, 1), dtype=int32) x@r[nz] ## tf.Tensor([1. 2. 3.], shape=(3), dtype=float64) # same as `x@r[nz]` x@r[x != 0] ## tf.Tensor([1. 2. 3.], shape=(3), dtype=float64) x@r[op_cast(x, \"bool\")] ## tf.Tensor([1. 2. 3.], shape=(3), dtype=float64) # 2d example (x2 <- op_stack(c(x, op_roll(x, 1), op_roll(x, 2)))) ## tf.Tensor( ## [[1. 0. 0. 0. 2. 0. 0. 0. 0. 3.] ##  [3. 1. 0. 0. 0. 2. 0. 0. 0. 0.] ##  [0. 3. 1. 0. 0. 0. 2. 0. 0. 0.]], shape=(3, 10), dtype=float64) x2@r[op_nonzero(x2)] ## tf.Tensor([1. 2. 3. 3. 1. 2. 3. 1. 2.], shape=(9), dtype=float64) x3 <- op_stack(c(x2, x2*1.1)) x3@r[op_nonzero(x3)] ## tf.Tensor([1.  2.  3.  3.  1.  2.  3.  1.  2.  1.1 2.2 3.3 3.3 1.1 2.2 3.3 1.1 2.2], shape=(18), dtype=float64)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_norm.html","id":null,"dir":"Reference","previous_headings":"","what":"Matrix or vector norm. — op_norm","title":"Matrix or vector norm. — op_norm","text":"function able return one eight different matrix norms, one infinite number vector norms (described ), depending value ord parameter.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_norm.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Matrix or vector norm. — op_norm","text":"","code":"op_norm(x, ord = NULL, axis = NULL, keepdims = FALSE)"},{"path":"https://keras3.posit.co/dev/reference/op_norm.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Matrix or vector norm. — op_norm","text":"x Input tensor. ord Order norm (see table Notes). default NULL. axis axis integer, specifies axis x along compute vector norms. axis length 2 vector, specifies axes hold 2-D matrices, matrix norms matrices computed. keepdims set TRUE, axes reduced left result dimensions size one.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_norm.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Matrix or vector norm. — op_norm","text":"Norm matrix vector(s).","code":""},{"path":"https://keras3.posit.co/dev/reference/op_norm.html","id":"note","dir":"Reference","previous_headings":"","what":"Note","title":"Matrix or vector norm. — op_norm","text":"values ord < 1, result , strictly speaking, mathematical 'norm', may still useful various numerical purposes. following norms can calculated: matrices: ord=NULL: Frobenius norm ord=\"fro\": Frobenius norm ord=\"nuc\": nuclear norm ord=Inf: max(sum(abs(x), axis=2)) ord=-Inf: min(sum(abs(x), axis=2)) ord=0: supported ord=1: max(sum(abs(x), axis=1)) ord=-1: min(sum(abs(x), axis=1)) ord=2: 2-norm (largest sing. value) ord=-2: smallest singular value : supported vectors: ord=NULL: 2-norm ord=\"fro\": supported ord=\"nuc\": supported ord=Inf: max(abs(x)) ord=-Inf: min(abs(x)) ord=0: sum(x != 0) ord=1: ord=-1: ord=2: ord=-2: : sum(abs(x)^ord)^(1/ord)","code":""},{"path":"https://keras3.posit.co/dev/reference/op_norm.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Matrix or vector norm. — op_norm","text":"","code":"x <- op_reshape(op_arange(9, dtype=\"float32\") - 4, c(3, 3)) op_norm(x) ## tf.Tensor(8.306623, shape=(), dtype=float32) # 7.7459664"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_normalize.html","id":null,"dir":"Reference","previous_headings":"","what":"Normalizes x over the specified axis. — op_normalize","title":"Normalizes x over the specified axis. — op_normalize","text":"defined : op_normalize(x) = x / max(norm(x), epsilon).","code":""},{"path":"https://keras3.posit.co/dev/reference/op_normalize.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Normalizes x over the specified axis. — op_normalize","text":"","code":"op_normalize(x, axis = -1L, order = 2L, epsilon = NULL)"},{"path":"https://keras3.posit.co/dev/reference/op_normalize.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Normalizes x over the specified axis. — op_normalize","text":"x Input tensor. axis axis axes along perform normalization. Default -1. order exponent value norm formulation. Defaults 2. epsilon lower bound value norm. Defaults config_epsilon().","code":""},{"path":"https://keras3.posit.co/dev/reference/op_normalize.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Normalizes x over the specified axis. — op_normalize","text":"normalized array.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_normalize.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Normalizes x over the specified axis. — op_normalize","text":"","code":"x <- op_convert_to_tensor(rbind(c(1, 2, 3), c(4, 5, 6))) x_norm <- op_normalize(x) x_norm ## tf.Tensor( ## [[0.26726124 0.53452248 0.80178373] ##  [0.45584231 0.56980288 0.68376346]], shape=(2, 3), dtype=float64)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_not_equal.html","id":null,"dir":"Reference","previous_headings":"","what":"Return (x1 != x2) element-wise. — op_not_equal","title":"Return (x1 != x2) element-wise. — op_not_equal","text":"Note function automatically called using R operator != tensor.","code":"(x <- op_arange(4)) ## tf.Tensor([1. 2. 3. 4.], shape=(4), dtype=float32) op_not_equal(x, 2) ## tf.Tensor([ True False  True  True], shape=(4), dtype=bool) x != 2 ## tf.Tensor([ True False  True  True], shape=(4), dtype=bool)"},{"path":"https://keras3.posit.co/dev/reference/op_not_equal.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Return (x1 != x2) element-wise. — op_not_equal","text":"","code":"op_not_equal(x1, x2)"},{"path":"https://keras3.posit.co/dev/reference/op_not_equal.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Return (x1 != x2) element-wise. — op_not_equal","text":"x1 First input tensor. x2 Second input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_not_equal.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Return (x1 != x2) element-wise. — op_not_equal","text":"Output tensor, element-wise comparison x1 x2.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_one_hot.html","id":null,"dir":"Reference","previous_headings":"","what":"Converts integer tensor x into a one-hot tensor. — op_one_hot","title":"Converts integer tensor x into a one-hot tensor. — op_one_hot","text":"one-hot encoding representation integer value converted binary vector length equal num_classes, index corresponding integer value marked 1, indices marked 0.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_one_hot.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Converts integer tensor x into a one-hot tensor. — op_one_hot","text":"","code":"op_one_hot(   x,   num_classes,   axis = -1L,   dtype = NULL,   sparse = FALSE,   zero_indexed = FALSE )"},{"path":"https://keras3.posit.co/dev/reference/op_one_hot.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Converts integer tensor x into a one-hot tensor. — op_one_hot","text":"x Integer tensor encoded. shape can arbitrary, dtype integer. R factors coerced integer. num_classes Number classes one-hot encoding. x factor num_classes NULL missing, levels(x) used. axis Axis along encoding performed. -1 represents last axis. Defaults -1. dtype (Optional) Data type output tensor. provided, defaults default data type backend. sparse Whether return sparse tensor; backends support sparse tensors. zero_indexed TRUE, treats indices zero-based (0 encodes first position); FALSE (default), treats indices one-based (1 encodes first position).","code":""},{"path":"https://keras3.posit.co/dev/reference/op_one_hot.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Converts integer tensor x into a one-hot tensor. — op_one_hot","text":"Integer tensor: One-hot encoded tensor shape x except specified axis dimension, length num_classes. dtype output tensor determined dtype default data type backend.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_one_hot.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Converts integer tensor x into a one-hot tensor. — op_one_hot","text":"","code":"x <- op_array(c(2, 4, 3, 1), \"int32\") op_one_hot(x, num_classes = 4) ## tf.Tensor( ## [[0. 1. 0. 0.] ##  [0. 0. 0. 1.] ##  [0. 0. 1. 0.] ##  [1. 0. 0. 0.]], shape=(4, 4), dtype=float32) op_one_hot(x - 1, num_classes = 4, zero_indexed = TRUE) ## tf.Tensor( ## [[0. 1. 0. 0.] ##  [0. 0. 0. 1.] ##  [0. 0. 1. 0.] ##  [1. 0. 0. 0.]], shape=(4, 4), dtype=float32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_ones.html","id":null,"dir":"Reference","previous_headings":"","what":"Return a new tensor of given shape and type, filled with ones. — op_ones","title":"Return a new tensor of given shape and type, filled with ones. — op_ones","text":"Return new tensor given shape type, filled ones.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_ones.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Return a new tensor of given shape and type, filled with ones. — op_ones","text":"","code":"op_ones(shape, dtype = NULL)"},{"path":"https://keras3.posit.co/dev/reference/op_ones.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Return a new tensor of given shape and type, filled with ones. — op_ones","text":"shape Shape new tensor. dtype Desired data type tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_ones.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Return a new tensor of given shape and type, filled with ones. — op_ones","text":"Tensor ones given shape dtype.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_ones_like.html","id":null,"dir":"Reference","previous_headings":"","what":"Return a tensor of ones with the same shape and type of x. — op_ones_like","title":"Return a tensor of ones with the same shape and type of x. — op_ones_like","text":"Return tensor ones shape type x.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_ones_like.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Return a tensor of ones with the same shape and type of x. — op_ones_like","text":"","code":"op_ones_like(x, dtype = NULL)"},{"path":"https://keras3.posit.co/dev/reference/op_ones_like.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Return a tensor of ones with the same shape and type of x. — op_ones_like","text":"x Input tensor. dtype Overrides data type result.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_ones_like.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Return a tensor of ones with the same shape and type of x. — op_ones_like","text":"tensor ones shape type x.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_outer.html","id":null,"dir":"Reference","previous_headings":"","what":"Compute the outer product of two vectors. — op_outer","title":"Compute the outer product of two vectors. — op_outer","text":"Given two vectors x1 x2, outer product :","code":"out[i, j] = x1[i] * x2[j]"},{"path":"https://keras3.posit.co/dev/reference/op_outer.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Compute the outer product of two vectors. — op_outer","text":"","code":"op_outer(x1, x2)"},{"path":"https://keras3.posit.co/dev/reference/op_outer.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Compute the outer product of two vectors. — op_outer","text":"x1 First input tensor. x2 Second input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_outer.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Compute the outer product of two vectors. — op_outer","text":"Outer product x1 x2.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_pad.html","id":null,"dir":"Reference","previous_headings":"","what":"Pad a tensor. — op_pad","title":"Pad a tensor. — op_pad","text":"Pad tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_pad.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Pad a tensor. — op_pad","text":"","code":"op_pad(x, pad_width, mode = \"constant\", constant_values = NULL)"},{"path":"https://keras3.posit.co/dev/reference/op_pad.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Pad a tensor. — op_pad","text":"x Tensor pad. pad_width Number values padded edges axis. ((before_1, after_1), ...(before_N, after_N)) unique pad widths axis. ((, ),) yields pad axis. (pad,) int shortcut = = pad width axes. mode One \"constant\", \"edge\", \"linear_ramp\", \"maximum\", \"mean\", \"median\", \"minimum\", \"reflect\", \"symmetric\", \"wrap\", \"empty\", \"circular\". Defaults \"constant\". constant_values Value pad mode == \"constant\". Defaults 0. ValueError raised NULL mode != \"constant\".","code":""},{"path":"https://keras3.posit.co/dev/reference/op_pad.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Pad a tensor. — op_pad","text":"Padded tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_pad.html","id":"note","dir":"Reference","previous_headings":"","what":"Note","title":"Pad a tensor. — op_pad","text":"Torch backend supports modes \"constant\", \"reflect\", \"symmetric\" \"circular\". Torch backend supports \"circular\" mode. Note: Tensorflow backend supports modes \"constant\", \"reflect\" \"symmetric\".","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_polar.html","id":null,"dir":"Reference","previous_headings":"","what":"Constructs a complex tensor whose elements are Cartesian — op_polar","title":"Constructs a complex tensor whose elements are Cartesian — op_polar","text":"coordinates corresponding polar coordinates absolute value abs angle angle. operation numerically equivalent torch.polar(). equivalent scipy.lingalg.polar() performs Singular Value Decomposition. Given magnitude (abs_) angle (angle), function computes corresponding complex number form real + imaginary * 1i, : real = abs_ * cos(angle) imaginary = abs_ * sin(angle)","code":""},{"path":"https://keras3.posit.co/dev/reference/op_polar.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Constructs a complex tensor whose elements are Cartesian — op_polar","text":"","code":"op_polar(abs_, angle)"},{"path":"https://keras3.posit.co/dev/reference/op_polar.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Constructs a complex tensor whose elements are Cartesian — op_polar","text":"abs_ magnitude (absolute value) complex number. angle angle (radians) complex number.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_polar.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Constructs a complex tensor whose elements are Cartesian — op_polar","text":"complex number (array complex numbers) shape abs_ angle.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_polar.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Constructs a complex tensor whose elements are Cartesian — op_polar","text":"","code":"abs_ <- random_normal(c(1, 2)) angle <- random_normal(c(1, 2)) op_shape(op_polar(abs_, angle)) ## shape(1, 2) op_polar(abs_, angle) ## tf.Tensor([[-0.10252866-1.2460487j -0.2586355 -1.0352845j]], shape=(1, 2), dtype=complex64)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_power.html","id":null,"dir":"Reference","previous_headings":"","what":"First tensor elements raised to powers from second tensor, element-wise. — op_power","title":"First tensor elements raised to powers from second tensor, element-wise. — op_power","text":"Note function automatically called using R operator ^ tensor.","code":"(x <- op_arange(4)) ## tf.Tensor([1. 2. 3. 4.], shape=(4), dtype=float32) op_power(2, x) ## tf.Tensor([ 2.  4.  8. 16.], shape=(4), dtype=float32) 2 ^ x ## tf.Tensor([ 2.  4.  8. 16.], shape=(4), dtype=float32)"},{"path":"https://keras3.posit.co/dev/reference/op_power.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"First tensor elements raised to powers from second tensor, element-wise. — op_power","text":"","code":"op_power(x1, x2)"},{"path":"https://keras3.posit.co/dev/reference/op_power.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"First tensor elements raised to powers from second tensor, element-wise. — op_power","text":"x1 bases. x2 exponents.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_power.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"First tensor elements raised to powers from second tensor, element-wise. — op_power","text":"Output tensor, bases x1 raised exponents x2.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_prod.html","id":null,"dir":"Reference","previous_headings":"","what":"Return the product of tensor elements over a given axis. — op_prod","title":"Return the product of tensor elements over a given axis. — op_prod","text":"Return product tensor elements given axis.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_prod.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Return the product of tensor elements over a given axis. — op_prod","text":"","code":"op_prod(x, axis = NULL, keepdims = FALSE, dtype = NULL)"},{"path":"https://keras3.posit.co/dev/reference/op_prod.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Return the product of tensor elements over a given axis. — op_prod","text":"x Input tensor. axis Axis axes along product performed. default, axis = NULL, compute product elements input tensor. keepdims set TRUE, axes reduce left result dimensions size one. dtype Data type returned tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_prod.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Return the product of tensor elements over a given axis. — op_prod","text":"Product elements x given axis axes.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_psnr.html","id":null,"dir":"Reference","previous_headings":"","what":"Peak Signal-to-Noise Ratio (PSNR) function. — op_psnr","title":"Peak Signal-to-Noise Ratio (PSNR) function. — op_psnr","text":"function computes Peak Signal--Noise Ratio two signals, x1 x2. PSNR measure quality reconstructed signal. higher PSNR, closer reconstructed signal original signal. Note can become negative signal power smaller noise power.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_psnr.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Peak Signal-to-Noise Ratio (PSNR) function. — op_psnr","text":"","code":"op_psnr(x1, x2, max_val)"},{"path":"https://keras3.posit.co/dev/reference/op_psnr.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Peak Signal-to-Noise Ratio (PSNR) function. — op_psnr","text":"x1 first input signal. x2 second input signal. Must shape x1. max_val maximum possible value signals.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_psnr.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Peak Signal-to-Noise Ratio (PSNR) function. — op_psnr","text":"float: PSNR value x1 x2.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_psnr.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Peak Signal-to-Noise Ratio (PSNR) function. — op_psnr","text":"","code":"x1 <- random_normal(c(2, 4, 4, 3)) x2 <- random_normal(c(2, 4, 4, 3)) max_val <- 1.0 op_psnr(x1, x2, max_val) ## tf.Tensor(-3.5293808, shape=(), dtype=float32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_qr.html","id":null,"dir":"Reference","previous_headings":"","what":"Computes the QR decomposition of a tensor. — op_qr","title":"Computes the QR decomposition of a tensor. — op_qr","text":"Computes QR decomposition tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_qr.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes the QR decomposition of a tensor. — op_qr","text":"","code":"op_qr(x, mode = \"reduced\")"},{"path":"https://keras3.posit.co/dev/reference/op_qr.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Computes the QR decomposition of a tensor. — op_qr","text":"x Input tensor shape (..., M, N). mode string specifying mode QR decomposition. 'reduced': Returns reduced QR decomposition. (default) 'complete': Returns complete QR decomposition.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_qr.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Computes the QR decomposition of a tensor. — op_qr","text":"list containing two tensors. first tensor shape (..., M, K) orthogonal matrix q second tensor shape (..., K, N)upper triangular matrixr, K = min(M, N)`.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_qr.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Computes the QR decomposition of a tensor. — op_qr","text":"","code":"x <- op_convert_to_tensor(rbind(c(1, 2), c(3, 4), c(5, 6))) op_qr(x) ## $q ## tf.Tensor( ## [[-0.16903085  0.89708523] ##  [-0.50709255  0.27602622] ##  [-0.84515425 -0.34503278]], shape=(3, 2), dtype=float64) ## ## $r ## tf.Tensor( ## [[-5.91607978 -7.43735744] ##  [ 0.          0.82807867]], shape=(2, 2), dtype=float64) c(q, r) %<-% op_qr(x)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_quantile.html","id":null,"dir":"Reference","previous_headings":"","what":"Compute the q-th quantile(s) of the data along the specified axis. — op_quantile","title":"Compute the q-th quantile(s) of the data along the specified axis. — op_quantile","text":"Compute q-th quantile(s) data along specified axis.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_quantile.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Compute the q-th quantile(s) of the data along the specified axis. — op_quantile","text":"","code":"op_quantile(x, q, axis = NULL, method = \"linear\", keepdims = FALSE)"},{"path":"https://keras3.posit.co/dev/reference/op_quantile.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Compute the q-th quantile(s) of the data along the specified axis. — op_quantile","text":"x Input tensor. q Probability sequence probabilities quantiles compute. Values must 0 1 inclusive. axis Axis axes along quantiles computed. Defaults axis=NULL compute quantile(s) along flattened version array. method string specifies method use estimating quantile. Available methods \"linear\", \"lower\", \"higher\", \"midpoint\", \"nearest\". Defaults \"linear\". desired quantile lies two data points < j: \"linear\": + (j - ) * fraction, fraction fractional part index surrounded j. \"lower\": . \"higher\": j. \"midpoint\": (+ j) / 2 \"nearest\": j, whichever nearest. keepdims set TRUE, axes reduce left result dimensions size one.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_quantile.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Compute the q-th quantile(s) of the data along the specified axis. — op_quantile","text":"quantile(s). q single probability axis = NULL, result scalar. multiple probabilities levels given, first axis result corresponds quantiles. axes axes remain reduction x.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_ravel.html","id":null,"dir":"Reference","previous_headings":"","what":"Return a contiguous flattened tensor. — op_ravel","title":"Return a contiguous flattened tensor. — op_ravel","text":"1-D tensor, containing elements input, returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_ravel.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Return a contiguous flattened tensor. — op_ravel","text":"","code":"op_ravel(x)"},{"path":"https://keras3.posit.co/dev/reference/op_ravel.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Return a contiguous flattened tensor. — op_ravel","text":"x Input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_ravel.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Return a contiguous flattened tensor. — op_ravel","text":"Output tensor.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_real.html","id":null,"dir":"Reference","previous_headings":"","what":"Return the real part of the complex argument. — op_real","title":"Return the real part of the complex argument. — op_real","text":"Return real part complex argument.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_real.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Return the real part of the complex argument. — op_real","text":"","code":"op_real(x)"},{"path":"https://keras3.posit.co/dev/reference/op_real.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Return the real part of the complex argument. — op_real","text":"x Input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_real.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Return the real part of the complex argument. — op_real","text":"real component complex argument.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_rearrange.html","id":null,"dir":"Reference","previous_headings":"","what":"Rearranges the axes of a Keras tensor according to a specified pattern, — op_rearrange","title":"Rearranges the axes of a Keras tensor according to a specified pattern, — op_rearrange","text":"einops-style.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_rearrange.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Rearranges the axes of a Keras tensor according to a specified pattern, — op_rearrange","text":"","code":"op_rearrange(tensor, pattern, ...)"},{"path":"https://keras3.posit.co/dev/reference/op_rearrange.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Rearranges the axes of a Keras tensor according to a specified pattern, — op_rearrange","text":"tensor Input Keras tensor. pattern String describing rearrangement einops notation. ... axes_lengths, named arguments specifying lengths axes axes decomposition used.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_rearrange.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Rearranges the axes of a Keras tensor according to a specified pattern, — op_rearrange","text":"Tensor: Keras tensor rearranged axes. Follows logic : decomposition needed, reshape match decomposed dimensions. Permute known inferred axes match form output. Reshape match desired output shape. Example Usage:","code":"images <- op_ones(c(32, 30, 40, 3)) # BHWC format  # Reordering to BCHW op_rearrange(images, 'b h w c -> b c h w') |> op_shape() ## shape(32, 3, 30, 40) # \"Merge\" along first axis - concat images from a batch op_rearrange(images, 'b h w c -> (b h) w c') |> op_shape() ## shape(960, 40, 3) # \"Merge\" along second axis - concat images horizontally op_rearrange(images, 'b h w c -> h (b w) c') |> op_shape() ## shape(30, 1280, 3) # Flatten images into a CHW vector op_rearrange(images, 'b h w c -> b (c h w)') |> op_shape() ## shape(32, 3600) # Decompose H and W axes into 4 smaller patches op_rearrange(images, 'b (h1 h) (w1 w) c -> (b h1 w1) h w c', h1 = 2, w1 = 2) |> op_shape() ## shape(128, 15, 20, 3) # Space-to-depth decomposition of input axes op_rearrange(images, 'b (h h1) (w w1) c -> b h w (c h1 w1)', h1 = 2, w1 = 2) |> op_shape() ## shape(32, 15, 20, 12)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_reciprocal.html","id":null,"dir":"Reference","previous_headings":"","what":"Return the reciprocal of the argument, element-wise. — op_reciprocal","title":"Return the reciprocal of the argument, element-wise. — op_reciprocal","text":"Calculates 1/x.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_reciprocal.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Return the reciprocal of the argument, element-wise. — op_reciprocal","text":"","code":"op_reciprocal(x)"},{"path":"https://keras3.posit.co/dev/reference/op_reciprocal.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Return the reciprocal of the argument, element-wise. — op_reciprocal","text":"x Input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_reciprocal.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Return the reciprocal of the argument, element-wise. — op_reciprocal","text":"Output tensor, element-wise reciprocal x.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_relu.html","id":null,"dir":"Reference","previous_headings":"","what":"Rectified linear unit activation function. — op_relu","title":"Rectified linear unit activation function. — op_relu","text":"defined f(x) = max(0, x).","code":""},{"path":"https://keras3.posit.co/dev/reference/op_relu.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Rectified linear unit activation function. — op_relu","text":"","code":"op_relu(x)"},{"path":"https://keras3.posit.co/dev/reference/op_relu.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Rectified linear unit activation function. — op_relu","text":"x Input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_relu.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Rectified linear unit activation function. — op_relu","text":"tensor shape x.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_relu.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Rectified linear unit activation function. — op_relu","text":"","code":"x1 <- op_convert_to_tensor(c(-1, 0, 1, 0.2)) op_relu(x1) ## tf.Tensor([0.  0.  1.  0.2], shape=(4), dtype=float64) x <- seq(-10, 10, .1) plot(x, op_relu(x))"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_relu6.html","id":null,"dir":"Reference","previous_headings":"","what":"Rectified linear unit activation function with upper bound of 6. — op_relu6","title":"Rectified linear unit activation function with upper bound of 6. — op_relu6","text":"defined f(x) = op_clip(x, 0, 6).","code":""},{"path":"https://keras3.posit.co/dev/reference/op_relu6.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Rectified linear unit activation function with upper bound of 6. — op_relu6","text":"","code":"op_relu6(x)"},{"path":"https://keras3.posit.co/dev/reference/op_relu6.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Rectified linear unit activation function with upper bound of 6. — op_relu6","text":"x Input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_relu6.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Rectified linear unit activation function with upper bound of 6. — op_relu6","text":"tensor shape x.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_relu6.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Rectified linear unit activation function with upper bound of 6. — op_relu6","text":"","code":"x <- op_convert_to_tensor(c(-3, -2, 0.1, 0.2, 6, 8)) op_relu6(x) ## tf.Tensor([0.  0.  0.1 0.2 6.  6. ], shape=(6), dtype=float64) x <- seq(-10, 10, .1) plot(x, op_relu6(x))"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_repeat.html","id":null,"dir":"Reference","previous_headings":"","what":"Repeat each element of a tensor after themselves. — op_repeat","title":"Repeat each element of a tensor after themselves. — op_repeat","text":"Repeat element tensor .","code":""},{"path":"https://keras3.posit.co/dev/reference/op_repeat.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Repeat each element of a tensor after themselves. — op_repeat","text":"","code":"op_repeat(x, repeats, axis = NULL)"},{"path":"https://keras3.posit.co/dev/reference/op_repeat.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Repeat each element of a tensor after themselves. — op_repeat","text":"x Input tensor. repeats number repetitions element. axis axis along repeat values. default, use flattened input array, return flat output array.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_repeat.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Repeat each element of a tensor after themselves. — op_repeat","text":"Output tensor.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_reshape.html","id":null,"dir":"Reference","previous_headings":"","what":"Gives a new shape to a tensor without changing its data. — op_reshape","title":"Gives a new shape to a tensor without changing its data. — op_reshape","text":"Gives new shape tensor without changing data.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_reshape.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Gives a new shape to a tensor without changing its data. — op_reshape","text":"","code":"op_reshape(x, newshape)"},{"path":"https://keras3.posit.co/dev/reference/op_reshape.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Gives a new shape to a tensor without changing its data. — op_reshape","text":"x Input tensor. newshape new shape compatible original shape. One shape dimension can -1 case value inferred length array remaining dimensions.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_reshape.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Gives a new shape to a tensor without changing its data. — op_reshape","text":"reshaped tensor.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_rfft.html","id":null,"dir":"Reference","previous_headings":"","what":"Real-valued Fast Fourier Transform along the last axis of the input. — op_rfft","title":"Real-valued Fast Fourier Transform along the last axis of the input. — op_rfft","text":"Computes 1D Discrete Fourier Transform real-valued signal inner-dimension input. Since Discrete Fourier Transform real-valued signal Hermitian-symmetric, RFFT returns fft_length / 2 + 1 unique components FFT: zero-frequency term, followed fft_length / 2 positive-frequency terms. Along axis RFFT computed , fft_length smaller corresponding dimension input, dimension cropped. larger, dimension padded zeros.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_rfft.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Real-valued Fast Fourier Transform along the last axis of the input. — op_rfft","text":"","code":"op_rfft(x, fft_length = NULL)"},{"path":"https://keras3.posit.co/dev/reference/op_rfft.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Real-valued Fast Fourier Transform along the last axis of the input. — op_rfft","text":"x Input tensor. fft_length integer representing number fft length. specified, inferred length last axis x. Defaults NULL.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_rfft.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Real-valued Fast Fourier Transform along the last axis of the input. — op_rfft","text":"list containing two tensors - real imaginary parts output.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_rfft.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Real-valued Fast Fourier Transform along the last axis of the input. — op_rfft","text":"","code":"x <- op_convert_to_tensor(c(0, 1, 2, 3, 4)) op_rfft(x) ## [[1]] ## tf.Tensor([10.  -2.5 -2.5], shape=(3), dtype=float64) ## ## [[2]] ## tf.Tensor([0.         3.4409548  0.81229924], shape=(3), dtype=float64) op_rfft(x, 3) ## [[1]] ## tf.Tensor([ 3.  -1.5], shape=(2), dtype=float64) ## ## [[2]] ## tf.Tensor([0.        0.8660254], shape=(2), dtype=float64)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_right_shift.html","id":null,"dir":"Reference","previous_headings":"","what":"Shift the bits of an integer to the right. — op_right_shift","title":"Shift the bits of an integer to the right. — op_right_shift","text":"Bits shifted right y. internal representation numbers binary format, operation equivalent dividing x 2^y.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_right_shift.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Shift the bits of an integer to the right. — op_right_shift","text":"","code":"op_right_shift(x, y)"},{"path":"https://keras3.posit.co/dev/reference/op_right_shift.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Shift the bits of an integer to the right. — op_right_shift","text":"x Input integer tensor. y Input integer tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_right_shift.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Shift the bits of an integer to the right. — op_right_shift","text":"Result tensor.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_rms_normalization.html","id":null,"dir":"Reference","previous_headings":"","what":"Performs Root Mean Square (RMS) normalization on x. — op_rms_normalization","title":"Performs Root Mean Square (RMS) normalization on x. — op_rms_normalization","text":"Keras operation implements operation described Root Mean Square Layer Normalization Biao Zhang et al. operation different LayerNormalization RMS scaling. defined rms_normalization(x) = x * rsqrt(mean(square(x))) * scale","code":""},{"path":"https://keras3.posit.co/dev/reference/op_rms_normalization.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Performs Root Mean Square (RMS) normalization on x. — op_rms_normalization","text":"","code":"op_rms_normalization(x, scale = NULL, axis = -1L, epsilon = NULL)"},{"path":"https://keras3.posit.co/dev/reference/op_rms_normalization.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Performs Root Mean Square (RMS) normalization on x. — op_rms_normalization","text":"x Input tensor. scale Optional scaling factor normalization. axis axis axes along perform normalization. Defaults -1. epsilon lower bound value norm. Defaults config_epsilon().","code":""},{"path":"https://keras3.posit.co/dev/reference/op_rms_normalization.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Performs Root Mean Square (RMS) normalization on x. — op_rms_normalization","text":"normalized array.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_rms_normalization.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Performs Root Mean Square (RMS) normalization on x. — op_rms_normalization","text":"[-1.1701705   0.9987454   0.27809578 -0.18452731  1.5731977   1.8934134","code":"x <- random_normal(c(1, 10)) op_rms_normalization(x) ## tf.Tensor( ## [[-1.1701705   0.9987454   0.27809578 -0.18452731  1.5731977   1.8934134 ##    0.10656697  1.0256069   0.6185074  -0.12678868]], shape=(1, 10), dtype=float32) op_rms_normalization(x, scale = 10) ## tf.Tensor( ## [[-11.701706    9.987453    2.7809577  -1.845273   15.731977   18.934134 ##     1.0656698  10.256069    6.185074   -1.2678868]], shape=(1, 10), dtype=float32)"},{"path":"https://keras3.posit.co/dev/reference/op_rms_normalization.html","id":"-r-a-","dir":"Reference","previous_headings":"","what":"0.10656697  1.0256069   0.6185074  -0.12678868]: R:-1.1701705%20%20%200.9987454%20%20%200.27809578%20-0.18452731%20%201.5731977%20%20%201.8934134%0A##%20%20%20%200.10656697%20%201.0256069%20%20%200.6185074%20%20-0.12678868","title":"Performs Root Mean Square (RMS) normalization on x. — op_rms_normalization","text":"[-11.701706    9.987453    2.7809577  -1.845273   15.731977   18.934134","code":""},{"path":[]},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_roll.html","id":null,"dir":"Reference","previous_headings":"","what":"Roll tensor elements along a given axis. — op_roll","title":"Roll tensor elements along a given axis. — op_roll","text":"Elements roll beyond last position re-introduced first.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_roll.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Roll tensor elements along a given axis. — op_roll","text":"","code":"op_roll(x, shift, axis = NULL)"},{"path":"https://keras3.posit.co/dev/reference/op_roll.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Roll tensor elements along a given axis. — op_roll","text":"x Input tensor. shift number places elements shifted. axis axis along elements shifted. default, array flattened shifting, original shape restored.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_roll.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Roll tensor elements along a given axis. — op_roll","text":"Output tensor.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_rot90.html","id":null,"dir":"Reference","previous_headings":"","what":"Rotate an array by 90 degrees in the plane specified by axes. — op_rot90","title":"Rotate an array by 90 degrees in the plane specified by axes. — op_rot90","text":"function rotates array counterclockwise 90 degrees k times plane specified axes. Supports arrays two dimensions.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_rot90.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Rotate an array by 90 degrees in the plane specified by axes. — op_rot90","text":"","code":"op_rot90(array, k = 1L, axes = list(1L, 2L))"},{"path":"https://keras3.posit.co/dev/reference/op_rot90.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Rotate an array by 90 degrees in the plane specified by axes. — op_rot90","text":"array Input array rotate. k Number times array rotated 90 degrees. axes tuple two integers specifying plane rotation (defaults (1, 2)).","code":""},{"path":"https://keras3.posit.co/dev/reference/op_rot90.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Rotate an array by 90 degrees in the plane specified by axes. — op_rot90","text":"Rotated array.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_rot90.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Rotate an array by 90 degrees in the plane specified by axes. — op_rot90","text":"","code":"m <- 1:4 |> op_reshape(c(2, 2)) m ## tf.Tensor( ## [[1 2] ##  [3 4]], shape=(2, 2), dtype=int32) op_rot90(m) ## tf.Tensor( ## [[2 4] ##  [1 3]], shape=(2, 2), dtype=int32) m <- 1:8 |> op_reshape(c(2, 2, 2)) m ## tf.Tensor( ## [[[1 2] ##   [3 4]] ## ##  [[5 6] ##   [7 8]]], shape=(2, 2, 2), dtype=int32) op_rot90(m, k = 1, axes = c(2, 3)) ## tf.Tensor( ## [[[2 4] ##   [1 3]] ## ##  [[6 8] ##   [5 7]]], shape=(2, 2, 2), dtype=int32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_round.html","id":null,"dir":"Reference","previous_headings":"","what":"Evenly round to the given number of decimals. — op_round","title":"Evenly round to the given number of decimals. — op_round","text":"Evenly round given number decimals.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_round.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Evenly round to the given number of decimals. — op_round","text":"","code":"op_round(x, decimals = 0L)"},{"path":"https://keras3.posit.co/dev/reference/op_round.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Evenly round to the given number of decimals. — op_round","text":"x Input tensor. decimals Number decimal places round . Defaults 0.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_round.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Evenly round to the given number of decimals. — op_round","text":"Output tensor.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_rsqrt.html","id":null,"dir":"Reference","previous_headings":"","what":"Computes reciprocal of square root of x element-wise. — op_rsqrt","title":"Computes reciprocal of square root of x element-wise. — op_rsqrt","text":"Computes reciprocal square root x element-wise.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_rsqrt.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes reciprocal of square root of x element-wise. — op_rsqrt","text":"","code":"op_rsqrt(x)"},{"path":"https://keras3.posit.co/dev/reference/op_rsqrt.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Computes reciprocal of square root of x element-wise. — op_rsqrt","text":"x input tensor","code":""},{"path":"https://keras3.posit.co/dev/reference/op_rsqrt.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Computes reciprocal of square root of x element-wise. — op_rsqrt","text":"tensor dtype x.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_rsqrt.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Computes reciprocal of square root of x element-wise. — op_rsqrt","text":"","code":"x <- op_convert_to_tensor(c(1, 10, 100)) op_rsqrt(x) ## tf.Tensor([1.         0.31622777 0.1       ], shape=(3), dtype=float64) # array([1, 0.31622776, 0.1], dtype=float32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_saturate_cast.html","id":null,"dir":"Reference","previous_headings":"","what":"Performs a safe saturating cast to the desired dtype. — op_saturate_cast","title":"Performs a safe saturating cast to the desired dtype. — op_saturate_cast","text":"Saturating cast prevents data type overflow casting dtype smaller values range. E.g. op_cast(c(-1, 256), \"float32\") |> op_cast(\"uint8\") returns c(255, 0), op_cast(c(-1, 256), \"float32\") |> op_saturate_cast(\"uint8\") returns c(0, 255).","code":""},{"path":"https://keras3.posit.co/dev/reference/op_saturate_cast.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Performs a safe saturating cast to the desired dtype. — op_saturate_cast","text":"","code":"op_saturate_cast(x, dtype)"},{"path":"https://keras3.posit.co/dev/reference/op_saturate_cast.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Performs a safe saturating cast to the desired dtype. — op_saturate_cast","text":"x tensor variable. dtype target type.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_saturate_cast.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Performs a safe saturating cast to the desired dtype. — op_saturate_cast","text":"safely casted tensor specified dtype.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_saturate_cast.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Performs a safe saturating cast to the desired dtype. — op_saturate_cast","text":"Image resizing bicubic interpolation may produce values outside original range.     Casting resized image back uint8 cause overflow.     Saturate casting uint8 clip values uint8 range casting cause overflow.","code":"image2x2 <- np_array(as.integer(c(0, 1, 254, 255)), \"uint8\") |>   array_reshape(c(1, 2, 2, 1)) image4x4 <- image2x2 |>   tensorflow::tf$image$resize(shape(4, 4), method=\"bicubic\") image4x4 |> as.array() |> drop() ##           [,1]      [,2]      [,3]      [,4] ## [1,] -22.50000 -22.20462 -21.61891 -21.32353 ## [2,]  52.52605  52.82143  53.40715  53.70253 ## [3,] 201.29752 201.59288 202.17859 202.47395 ## [4,] 276.32355 276.61893 277.20465 277.50006 image4x4_casted <- op_cast(image4x4, \"uint8\") image4x4_casted |> as.array() |> drop() ##      [,1] [,2] [,3] [,4] ## [1,]  234  234  235  235 ## [2,]   52   52   53   53 ## [3,]  201  201  202  202 ## [4,]   20   20   21   21 image4x4_saturate_casted <- image4x4 |> op_saturate_cast(\"uint8\") image4x4_saturate_casted |> as.array() |> drop() ##      [,1] [,2] [,3] [,4] ## [1,]    0    0    0    0 ## [2,]   52   52   53   53 ## [3,]  201  201  202  202 ## [4,]  255  255  255  255"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_scan.html","id":null,"dir":"Reference","previous_headings":"","what":"Scan a function over leading array axes while carrying along state. — op_scan","title":"Scan a function over leading array axes while carrying along state. — op_scan","text":"type xs array type NULL, type ys array type, semantics op_scan() given roughly implementation:   loop-carried value carry (init) must hold fixed shape dtype across iterations. TensorFlow, y must match carry shape dtype. required backends.","code":"op_scan <- function(f, init, xs = NULL, length = NULL) {   xs <- xs %||% vector(\"list\", length)   if(!is.list(xs))     xs <- op_unstack(xs)   ys <- vector(\"list\", length(xs))   carry <- init   for (i in seq_along(xs)) {     c(carry, y) %<-% f(carry, xs[[i]])     ys[[i]] <- y   }   list(carry, op_stack(ys)) }"},{"path":"https://keras3.posit.co/dev/reference/op_scan.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Scan a function over leading array axes while carrying along state. — op_scan","text":"","code":"op_scan(f, init, xs = NULL, length = NULL, reverse = FALSE, unroll = 1L)"},{"path":"https://keras3.posit.co/dev/reference/op_scan.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Scan a function over leading array axes while carrying along state. — op_scan","text":"f Callable defines logic loop iteration. accepts two arguments first value loop carry second slice xs along leading axis. callable returns pair first represents new value loop carry second represents slice output. init initial loop carry value. can scalar, tensor, nested structure. must match structure first element returned f. xs Optional value scan along leading axis. can tensor nested structure. xs provided, must specify length define number loop iterations. Defaults NULL. length Optional integer specifying number loop iterations. length provided, defaults sizes leading axis arrays xs. Defaults NULL. reverse Optional boolean specifying whether run scan iteration forward reverse, equivalent reversing leading axes arrays xs ys. unroll Optional positive integer boolean specifying many scan iterations unroll within single iteration loop. integer provided, determines many unrolled loop iterations run within single rolled iteration loop. boolean provided, determine loop completely unrolled (unroll=TRUE) left completely unrolled (unroll=FALSE). Note unrolling supported JAX TensorFlow backends.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_scan.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Scan a function over leading array axes while carrying along state. — op_scan","text":"pair first element represents final loop carry value second element represents stacked outputs f scanned leading axis inputs.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_scan.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Scan a function over leading array axes while carrying along state. — op_scan","text":"","code":"sum_fn <- function(c, x) list(c + x, c + x) init <- op_array(0L) xs <- op_array(1:5) c(carry, result) %<-% op_scan(sum_fn, init, xs) carry ## tf.Tensor(15, shape=(), dtype=int32) result ## tf.Tensor([ 1  3  6 10 15], shape=(5), dtype=int32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_scatter.html","id":null,"dir":"Reference","previous_headings":"","what":"Returns a tensor of shape shape where indices are set to values. — op_scatter","title":"Returns a tensor of shape shape where indices are set to values. — op_scatter","text":"high level, operation zeros[indices] = updates returns output. equivalent :","code":"output <- op_scatter_update(op_zeros(shape), indices, values)"},{"path":"https://keras3.posit.co/dev/reference/op_scatter.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Returns a tensor of shape shape where indices are set to values. — op_scatter","text":"","code":"op_scatter(indices, values, shape)"},{"path":"https://keras3.posit.co/dev/reference/op_scatter.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Returns a tensor of shape shape where indices are set to values. — op_scatter","text":"indices tensor list specifying indices values values. values tensor, values set indices. shape Shape output tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_scatter.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Returns a tensor of shape shape where indices are set to values. — op_scatter","text":"tensor shape shape indices set values.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_scatter.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Returns a tensor of shape shape where indices are set to values. — op_scatter","text":"","code":"op_scatter(indices = rbind(1), values = 1, shape = c(5)) ## tf.Tensor([1. 0. 0. 0. 0.], shape=(5), dtype=float64) op_scatter(indices = rbind(1, 3), values = 1:2, shape = c(5)) ## tf.Tensor([1 0 2 0 0], shape=(5), dtype=int32) indices <- rbind(c(1, 2), c(2, 2)) values <- op_array(c(1, 1)) op_scatter(indices, values, shape = c(2, 2)) ## tf.Tensor( ## [[0. 1.] ##  [0. 1.]], shape=(2, 2), dtype=float32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_scatter_update.html","id":null,"dir":"Reference","previous_headings":"","what":"Update inputs via updates at scattered (sparse) indices. — op_scatter_update","title":"Update inputs via updates at scattered (sparse) indices. — op_scatter_update","text":"high level, operation inputs[indices] <- updates. Assume inputs tensor shape (D1, D2, ..., Dn), 2 main usages scatter_update. indices 2D tensor shape (num_updates, n), num_updates number updates perform, updates 1D tensor shape (num_updates). example, inputs op_zeros(c(4, 4, 4)), want update inputs[2, 3, 4] inputs[1, 2, 4] 1, can use:     2 indices 2D tensor shape (num_updates, k), num_updates number updates perform, k (k <= n) size index indices. updates n - k-D tensor shape (num_updates, shape(inputs)[-(1:k)]). example, inputs <- op_zeros(c(4, 4, 4)), want update inputs[1, 2, ] inputs[2, 3, ] [1, 1, 1, 1], indices shape (num_updates, 2) (k = 2), updates shape (num_updates, 4) (shape(inputs)[3:4] == 4). See code :","code":"inputs <- op_zeros(c(4, 4, 4)) indices <- rbind(c(2, 3, 4),                  c(1, 2, 4)) updates <- op_array(c(1, 1), \"float32\") op_scatter_update(inputs, indices, updates) ## tf.Tensor( ## [[[0. 0. 0. 0.] ##   [0. 0. 0. 1.] ##   [0. 0. 0. 0.] ##   [0. 0. 0. 0.]] ## ##  [[0. 0. 0. 0.] ##   [0. 0. 0. 0.] ##   [0. 0. 0. 1.] ##   [0. 0. 0. 0.]] ## ##  [[0. 0. 0. 0.] ##   [0. 0. 0. 0.] ##   [0. 0. 0. 0.] ##   [0. 0. 0. 0.]] ## ##  [[0. 0. 0. 0.] ##   [0. 0. 0. 0.] ##   [0. 0. 0. 0.] ##   [0. 0. 0. 0.]]], shape=(4, 4, 4), dtype=float32) inputs <- op_zeros(c(4, 4, 4)) indices <- rbind(c(2, 3),                  c(3, 4)) updates <- op_array(rbind(c(1, 1, 1, 1),                           c(1, 1, 1, 1)),                     \"float32\") op_scatter_update(inputs, indices, updates) ## tf.Tensor( ## [[[0. 0. 0. 0.] ##   [0. 0. 0. 0.] ##   [0. 0. 0. 0.] ##   [0. 0. 0. 0.]] ## ##  [[0. 0. 0. 0.] ##   [0. 0. 0. 0.] ##   [1. 1. 1. 1.] ##   [0. 0. 0. 0.]] ## ##  [[0. 0. 0. 0.] ##   [0. 0. 0. 0.] ##   [0. 0. 0. 0.] ##   [1. 1. 1. 1.]] ## ##  [[0. 0. 0. 0.] ##   [0. 0. 0. 0.] ##   [0. 0. 0. 0.] ##   [0. 0. 0. 0.]]], shape=(4, 4, 4), dtype=float32)"},{"path":"https://keras3.posit.co/dev/reference/op_scatter_update.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Update inputs via updates at scattered (sparse) indices. — op_scatter_update","text":"","code":"op_scatter_update(inputs, indices, updates)"},{"path":"https://keras3.posit.co/dev/reference/op_scatter_update.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Update inputs via updates at scattered (sparse) indices. — op_scatter_update","text":"inputs tensor, tensor updated. indices tensor list shape (N, inputs$ndim), specifying indices update. N number indices update, must equal first dimension updates. updates tensor, new values put inputs indices.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_scatter_update.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Update inputs via updates at scattered (sparse) indices. — op_scatter_update","text":"tensor, shape dtype inputs.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_searchsorted.html","id":null,"dir":"Reference","previous_headings":"","what":"Perform a binary search. — op_searchsorted","title":"Perform a binary search. — op_searchsorted","text":"Return insertion indices keep values sorted order placed sorted_sequence.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_searchsorted.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Perform a binary search. — op_searchsorted","text":"","code":"op_searchsorted(sorted_sequence, values, side = \"left\", zero_indexed = FALSE)"},{"path":"https://keras3.posit.co/dev/reference/op_searchsorted.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Perform a binary search. — op_searchsorted","text":"sorted_sequence 1-D input tensor sorted along innermost dimension. values N-D tensor query insertion values. side Either \"left\" \"right\", choosing side insert ties. zero_indexed TRUE, returned indices zero-based (0 encodes first position); FALSE (default), returned indices one-based (1 encodes first position).","code":""},{"path":"https://keras3.posit.co/dev/reference/op_searchsorted.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Perform a binary search. — op_searchsorted","text":"Tensor insertion indices shape values.","code":""},{"path":[]},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_segment_max.html","id":null,"dir":"Reference","previous_headings":"","what":"Computes the max of segments in a tensor. — op_segment_max","title":"Computes the max of segments in a tensor. — op_segment_max","text":"Computes max segments tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_segment_max.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes the max of segments in a tensor. — op_segment_max","text":"","code":"op_segment_max(data, segment_ids, num_segments = NULL, sorted = FALSE)"},{"path":"https://keras3.posit.co/dev/reference/op_segment_max.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Computes the max of segments in a tensor. — op_segment_max","text":"data Input tensor. segment_ids N-D tensor containing segment indices element data. head(op_shape(data), length(op_shape(segment_ids))) match op_shape(segment_ids) num_segments integer representing total number segments. specified, inferred maximum value segment_ids. sorted boolean indicating whether segment_ids sorted. Defaults FALSE.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_segment_max.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Computes the max of segments in a tensor. — op_segment_max","text":"tensor containing max segments, element represents max corresponding segment data.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_segment_max.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Computes the max of segments in a tensor. — op_segment_max","text":"","code":"data <- op_convert_to_tensor(c(1, 2, 10, 20, 100, 200)) segment_ids <- op_array(c(1, 1, 2, 2, 3, 3), \"int32\") num_segments <- 3 op_segment_max(data, segment_ids, num_segments) ## tf.Tensor([  2.  20. 200.], shape=(3), dtype=float64) # array([2, 20, 200], dtype=int32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_segment_sum.html","id":null,"dir":"Reference","previous_headings":"","what":"Computes the sum of segments in a tensor. — op_segment_sum","title":"Computes the sum of segments in a tensor. — op_segment_sum","text":"Computes sum segments tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_segment_sum.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes the sum of segments in a tensor. — op_segment_sum","text":"","code":"op_segment_sum(data, segment_ids, num_segments = NULL, sorted = FALSE)"},{"path":"https://keras3.posit.co/dev/reference/op_segment_sum.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Computes the sum of segments in a tensor. — op_segment_sum","text":"data Input tensor. segment_ids N-D tensor containing segment indices element data. Num dims segment ids strictly smaller equal number dims data. num_segments integer representing total number segments. specified, inferred maximum value segment_ids. sorted boolean indicating whether segment_ids sorted. Defaults FALSE.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_segment_sum.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Computes the sum of segments in a tensor. — op_segment_sum","text":"tensor containing sum segments, element represents sum corresponding segment data.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_segment_sum.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Computes the sum of segments in a tensor. — op_segment_sum","text":"","code":"data <- op_array(c(1, 2, 10, 20, 100, 200)) segment_ids <- op_array(c(1, 1, 2, 2, 3, 3), \"int32\") num_segments <- 3 op_segment_sum(data, segment_ids, num_segments) ## tf.Tensor([  3.  30. 300.], shape=(3), dtype=float32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_select.html","id":null,"dir":"Reference","previous_headings":"","what":"Return elements from choicelist, based on conditions in condlist. — op_select","title":"Return elements from choicelist, based on conditions in condlist. — op_select","text":"Return elements choicelist, based conditions condlist.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_select.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Return elements from choicelist, based on conditions in condlist. — op_select","text":"","code":"op_select(condlist, choicelist, default = 0L)"},{"path":"https://keras3.posit.co/dev/reference/op_select.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Return elements from choicelist, based on conditions in condlist. — op_select","text":"condlist List boolean tensors. list conditions determine array choicelist output elements taken. multiple conditions satisfied, first one encountered condlist used. choicelist List tensors. list tensors output elements taken. list length condlist. default Optional scalar value. element inserted output conditions evaluate FALSE.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_select.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Return elements from choicelist, based on conditions in condlist. — op_select","text":"Tensor output position m m-th element tensor choicelist m-th element corresponding tensor condlist TRUE.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_select.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Return elements from choicelist, based on conditions in condlist. — op_select","text":"","code":"x <- op_arange(6L) condlist <- list(x < 3, x > 3) choicelist <- list(x, x^2) op_select(condlist, choicelist, 42) ## tf.Tensor([ 1  2 42 16 25 36], shape=(6), dtype=int32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_selu.html","id":null,"dir":"Reference","previous_headings":"","what":"Scaled Exponential Linear Unit (SELU) activation function. — op_selu","title":"Scaled Exponential Linear Unit (SELU) activation function. — op_selu","text":"defined : f(x) =  scale * alpha * (exp(x) - 1.) x < 0, f(x) = scale * x x >= 0.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_selu.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Scaled Exponential Linear Unit (SELU) activation function. — op_selu","text":"","code":"op_selu(x)"},{"path":"https://keras3.posit.co/dev/reference/op_selu.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Scaled Exponential Linear Unit (SELU) activation function. — op_selu","text":"x Input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_selu.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Scaled Exponential Linear Unit (SELU) activation function. — op_selu","text":"tensor shape x.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_selu.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Scaled Exponential Linear Unit (SELU) activation function. — op_selu","text":"","code":"x <- op_array(c(-1, 0, 1)) op_selu(x) ## tf.Tensor([-1.1113307  0.         1.050701 ], shape=(3), dtype=float32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_separable_conv.html","id":null,"dir":"Reference","previous_headings":"","what":"General N-D separable convolution. — op_separable_conv","title":"General N-D separable convolution. — op_separable_conv","text":"ops supports 1D 2D separable convolution. separable_conv depthwise conv followed pointwise conv.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_separable_conv.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"General N-D separable convolution. — op_separable_conv","text":"","code":"op_separable_conv(   inputs,   depthwise_kernel,   pointwise_kernel,   strides = 1L,   padding = \"valid\",   data_format = NULL,   dilation_rate = 1L )"},{"path":"https://keras3.posit.co/dev/reference/op_separable_conv.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"General N-D separable convolution. — op_separable_conv","text":"inputs Tensor rank N+2. inputs shape (batch_size,) + inputs_spatial_shape + (num_channels,) data_format=\"channels_last\", (batch_size, num_channels) + inputs_spatial_shape data_format=\"channels_first\". depthwise_kernel Tensor rank N+2. depthwise_kernel shape [kernel_spatial_shape, num_input_channels, num_channels_multiplier], num_input_channels match number channels inputs. pointwise_kernel Tensor rank N+2. pointwise_kernel shape (*ones_like(kernel_spatial_shape), num_input_channels * num_channels_multiplier, num_output_channels). strides int int tuple/list len(inputs_spatial_shape), specifying strides convolution along spatial dimension. strides int, every spatial dimension shares strides. padding string, either \"valid\" \"\". \"valid\" means padding applied, \"\" results padding evenly left/right /input output height/width dimension input strides=1. data_format string, either \"channels_last\" \"channels_first\". data_format determines ordering dimensions inputs. data_format=\"channels_last\", inputs shape (batch_size, ..., channels) data_format=\"channels_first\", inputs shape (batch_size, channels, ...). dilation_rate int int tuple/list len(inputs_spatial_shape), specifying dilation rate use dilated convolution. dilation_rate int, every spatial dimension shares dilation_rate.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_separable_conv.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"General N-D separable convolution. — op_separable_conv","text":"tensor rank N+2, result depthwise conv operation.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_shape.html","id":null,"dir":"Reference","previous_headings":"","what":"Gets the shape of the tensor input. — op_shape","title":"Gets the shape of the tensor input. — op_shape","text":"Gets shape tensor input.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_shape.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Gets the shape of the tensor input. — op_shape","text":"","code":"op_shape(x)"},{"path":"https://keras3.posit.co/dev/reference/op_shape.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Gets the shape of the tensor input. — op_shape","text":"x tensor. function try access shape attribute input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_shape.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Gets the shape of the tensor input. — op_shape","text":"list integers NULL values, indicating shape input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_shape.html","id":"note","dir":"Reference","previous_headings":"","what":"Note","title":"Gets the shape of the tensor input. — op_shape","text":"TensorFlow backend, x tf.Tensor dynamic shape, dimensions dynamic context compiled function tf.Tensor value instead static integer value.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_shape.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Gets the shape of the tensor input. — op_shape","text":"","code":"x <- op_zeros(c(8, 12)) op_shape(x) ## shape(8, 12)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_sigmoid.html","id":null,"dir":"Reference","previous_headings":"","what":"Sigmoid activation function. — op_sigmoid","title":"Sigmoid activation function. — op_sigmoid","text":"defined f(x) = 1 / (1 + exp(-x)).","code":""},{"path":"https://keras3.posit.co/dev/reference/op_sigmoid.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Sigmoid activation function. — op_sigmoid","text":"","code":"op_sigmoid(x)"},{"path":"https://keras3.posit.co/dev/reference/op_sigmoid.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Sigmoid activation function. — op_sigmoid","text":"x Input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_sigmoid.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Sigmoid activation function. — op_sigmoid","text":"tensor shape x.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_sigmoid.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Sigmoid activation function. — op_sigmoid","text":"","code":"x <- op_convert_to_tensor(c(-6, 1, 0, 1, 6)) op_sigmoid(x) ## tf.Tensor([0.00247262 0.73105858 0.5        0.73105858 0.99752738], shape=(5), dtype=float64)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_sign.html","id":null,"dir":"Reference","previous_headings":"","what":"Returns a tensor with the signs of the elements of x. — op_sign","title":"Returns a tensor with the signs of the elements of x. — op_sign","text":"Returns tensor signs elements x.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_sign.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Returns a tensor with the signs of the elements of x. — op_sign","text":"","code":"op_sign(x)"},{"path":"https://keras3.posit.co/dev/reference/op_sign.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Returns a tensor with the signs of the elements of x. — op_sign","text":"x Input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_sign.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Returns a tensor with the signs of the elements of x. — op_sign","text":"Output tensor shape x.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_signbit.html","id":null,"dir":"Reference","previous_headings":"","what":"Return the sign bit of the elements of x. — op_signbit","title":"Return the sign bit of the elements of x. — op_signbit","text":"output boolean tensor contains TRUE sign x negative, FALSE otherwise.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_signbit.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Return the sign bit of the elements of x. — op_signbit","text":"","code":"op_signbit(x)"},{"path":"https://keras3.posit.co/dev/reference/op_signbit.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Return the sign bit of the elements of x. — op_signbit","text":"x Input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_signbit.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Return the sign bit of the elements of x. — op_signbit","text":"Output boolean tensor shape x.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_silu.html","id":null,"dir":"Reference","previous_headings":"","what":"Sigmoid Linear Unit (SiLU) activation function, also known as Swish. — op_silu","title":"Sigmoid Linear Unit (SiLU) activation function, also known as Swish. — op_silu","text":"SiLU activation function computed sigmoid function multiplied input. defined f(x) = x * sigmoid(x).","code":""},{"path":"https://keras3.posit.co/dev/reference/op_silu.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Sigmoid Linear Unit (SiLU) activation function, also known as Swish. — op_silu","text":"","code":"op_silu(x)"},{"path":"https://keras3.posit.co/dev/reference/op_silu.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Sigmoid Linear Unit (SiLU) activation function, also known as Swish. — op_silu","text":"x Input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_silu.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Sigmoid Linear Unit (SiLU) activation function, also known as Swish. — op_silu","text":"tensor shape x.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_silu.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Sigmoid Linear Unit (SiLU) activation function, also known as Swish. — op_silu","text":"","code":"x <- op_convert_to_tensor(c(-6, 1, 0, 1, 6)) op_sigmoid(x) ## tf.Tensor([0.00247262 0.73105858 0.5        0.73105858 0.99752738], shape=(5), dtype=float64) op_silu(x) ## tf.Tensor([-0.01483574  0.73105858  0.          0.73105858  5.98516426], shape=(5), dtype=float64)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_sin.html","id":null,"dir":"Reference","previous_headings":"","what":"Trigonometric sine, element-wise. — op_sin","title":"Trigonometric sine, element-wise. — op_sin","text":"Trigonometric sine, element-wise.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_sin.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Trigonometric sine, element-wise. — op_sin","text":"","code":"op_sin(x)"},{"path":"https://keras3.posit.co/dev/reference/op_sin.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Trigonometric sine, element-wise. — op_sin","text":"x Input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_sin.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Trigonometric sine, element-wise. — op_sin","text":"Output tensor shape x.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_sinh.html","id":null,"dir":"Reference","previous_headings":"","what":"Hyperbolic sine, element-wise. — op_sinh","title":"Hyperbolic sine, element-wise. — op_sinh","text":"Hyperbolic sine, element-wise.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_sinh.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Hyperbolic sine, element-wise. — op_sinh","text":"","code":"op_sinh(x)"},{"path":"https://keras3.posit.co/dev/reference/op_sinh.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Hyperbolic sine, element-wise. — op_sinh","text":"x Input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_sinh.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Hyperbolic sine, element-wise. — op_sinh","text":"Output tensor shape x.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_size.html","id":null,"dir":"Reference","previous_headings":"","what":"Return the number of elements in a tensor. — op_size","title":"Return the number of elements in a tensor. — op_size","text":"Return number elements tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_size.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Return the number of elements in a tensor. — op_size","text":"","code":"op_size(x)"},{"path":"https://keras3.posit.co/dev/reference/op_size.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Return the number of elements in a tensor. — op_size","text":"x Input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_size.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Return the number of elements in a tensor. — op_size","text":"Number elements x.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_slice.html","id":null,"dir":"Reference","previous_headings":"","what":"Return a slice of an input tensor. — op_slice","title":"Return a slice of an input tensor. — op_slice","text":"high level, operation explicit replacement array slicing e.g. inputs[start_indices:(start_indices + shape)]. Unlike slicing via brackets, operation accept tensor start indices backends, useful indices dynamically computed via tensor operations.","code":"(inputs <- op_arange(5*5) |> op_reshape(c(5, 5))) ## tf.Tensor( ## [[ 1.  2.  3.  4.  5.] ##  [ 6.  7.  8.  9. 10.] ##  [11. 12. 13. 14. 15.] ##  [16. 17. 18. 19. 20.] ##  [21. 22. 23. 24. 25.]], shape=(5, 5), dtype=float32) start_indices <- c(3, 3) shape <- c(2, 2) op_slice(inputs, start_indices, shape) ## tf.Tensor( ## [[13. 14.] ##  [18. 19.]], shape=(2, 2), dtype=float32)"},{"path":"https://keras3.posit.co/dev/reference/op_slice.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Return a slice of an input tensor. — op_slice","text":"","code":"op_slice(inputs, start_indices, shape)"},{"path":"https://keras3.posit.co/dev/reference/op_slice.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Return a slice of an input tensor. — op_slice","text":"inputs tensor, tensor sliced. start_indices list length inputs$ndim, specifying starting indices updating. shape full shape returned slice.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_slice.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Return a slice of an input tensor. — op_slice","text":"tensor, shape dtype inputs.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_slice_update.html","id":null,"dir":"Reference","previous_headings":"","what":"Update an input by slicing in a tensor of updated values. — op_slice_update","title":"Update an input by slicing in a tensor of updated values. — op_slice_update","text":"high level, operation inputs[start_indices: start_indices + updates.shape] = updates. Assume inputs tensor shape (D1, D2, ..., Dn), start_indices must list n integers, specifying starting indices. updates must rank inputs, size dim must exceed Di - start_indices[]. example, 2D inputs inputs = op_zeros(c(5, 5)), want update intersection last 2 rows last 2 columns 1, .e., inputs[4:5, 4:5] = op_ones(c(2, 2)), can use code :","code":"inputs <- op_zeros(c(5, 5)) start_indices <- c(3, 3) updates <- op_ones(c(2, 2)) op_slice_update(inputs, start_indices, updates) ## tf.Tensor( ## [[0. 0. 0. 0. 0.] ##  [0. 0. 0. 0. 0.] ##  [0. 0. 1. 1. 0.] ##  [0. 0. 1. 1. 0.] ##  [0. 0. 0. 0. 0.]], shape=(5, 5), dtype=float32)"},{"path":"https://keras3.posit.co/dev/reference/op_slice_update.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Update an input by slicing in a tensor of updated values. — op_slice_update","text":"","code":"op_slice_update(inputs, start_indices, updates)"},{"path":"https://keras3.posit.co/dev/reference/op_slice_update.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Update an input by slicing in a tensor of updated values. — op_slice_update","text":"inputs tensor, tensor updated. start_indices list length inputs$ndim, specifying starting indices updating. updates tensor, new values put inputs indices. updates must rank inputs.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_slice_update.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Update an input by slicing in a tensor of updated values. — op_slice_update","text":"tensor, shape dtype inputs.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_slogdet.html","id":null,"dir":"Reference","previous_headings":"","what":"Compute the sign and natural logarithm of the determinant of a matrix. — op_slogdet","title":"Compute the sign and natural logarithm of the determinant of a matrix. — op_slogdet","text":"Compute sign natural logarithm determinant matrix.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_slogdet.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Compute the sign and natural logarithm of the determinant of a matrix. — op_slogdet","text":"","code":"op_slogdet(x)"},{"path":"https://keras3.posit.co/dev/reference/op_slogdet.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Compute the sign and natural logarithm of the determinant of a matrix. — op_slogdet","text":"x Input matrix. must 2D square.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_slogdet.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Compute the sign and natural logarithm of the determinant of a matrix. — op_slogdet","text":"list: (sign, logabsdet). sign number representing sign determinant. real matrix, 1, 0, -1. complex matrix, complex number absolute value 1 (.e., unit circle), else 0. logabsdet natural log absolute value determinant.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_soft_shrink.html","id":null,"dir":"Reference","previous_headings":"","what":"Soft Shrink activation function. — op_soft_shrink","title":"Soft Shrink activation function. — op_soft_shrink","text":"defined f(x) = x - threshold x > threshold, f(x) = x + threshold x < -threshold, f(x) = 0 otherwise.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_soft_shrink.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Soft Shrink activation function. — op_soft_shrink","text":"","code":"op_soft_shrink(x, threshold = 0.5)"},{"path":"https://keras3.posit.co/dev/reference/op_soft_shrink.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Soft Shrink activation function. — op_soft_shrink","text":"x Input tensor. threshold Threshold value. Defaults 0.5.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_soft_shrink.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Soft Shrink activation function. — op_soft_shrink","text":"tensor shape x.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_soft_shrink.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Soft Shrink activation function. — op_soft_shrink","text":"","code":"x <- op_array(c(-1, 0, 1)) op_soft_shrink(x) ## tf.Tensor([-0.5  0.   0.5], shape=(3), dtype=float32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_softmax.html","id":null,"dir":"Reference","previous_headings":"","what":"Softmax activation function. — op_softmax","title":"Softmax activation function. — op_softmax","text":"elements output vector lie within range (0, 1), total sum exactly 1 (excluding floating point rounding error). vector processed independently. axis argument specifies axis along function applied within input. defined : f(x) = exp(x) / sum(exp(x))","code":""},{"path":"https://keras3.posit.co/dev/reference/op_softmax.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Softmax activation function. — op_softmax","text":"","code":"op_softmax(x, axis = -1L)"},{"path":"https://keras3.posit.co/dev/reference/op_softmax.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Softmax activation function. — op_softmax","text":"x Input tensor. axis Integer, axis along softmax applied.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_softmax.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Softmax activation function. — op_softmax","text":"tensor shape x.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_softmax.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Softmax activation function. — op_softmax","text":"","code":"x <- op_array(c(-1, 0, 1)) op_softmax(x) ## tf.Tensor([0.09003057 0.24472848 0.66524094], shape=(3), dtype=float32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_softplus.html","id":null,"dir":"Reference","previous_headings":"","what":"Softplus activation function. — op_softplus","title":"Softplus activation function. — op_softplus","text":"defined f(x) = log(exp(x) + 1), log natural logarithm exp exponential function.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_softplus.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Softplus activation function. — op_softplus","text":"","code":"op_softplus(x)"},{"path":"https://keras3.posit.co/dev/reference/op_softplus.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Softplus activation function. — op_softplus","text":"x Input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_softplus.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Softplus activation function. — op_softplus","text":"tensor shape x.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_softplus.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Softplus activation function. — op_softplus","text":"","code":"x <- op_convert_to_tensor(c(-0.555, 0, 0.555)) op_softplus(x) ## tf.Tensor([0.45366606 0.69314718 1.00866606], shape=(3), dtype=float64) x <- seq(-10, 10, .1) plot(x, op_softplus(x))"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_softsign.html","id":null,"dir":"Reference","previous_headings":"","what":"Softsign activation function. — op_softsign","title":"Softsign activation function. — op_softsign","text":"defined f(x) = x / (abs(x) + 1).","code":""},{"path":"https://keras3.posit.co/dev/reference/op_softsign.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Softsign activation function. — op_softsign","text":"","code":"op_softsign(x)"},{"path":"https://keras3.posit.co/dev/reference/op_softsign.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Softsign activation function. — op_softsign","text":"x Input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_softsign.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Softsign activation function. — op_softsign","text":"tensor shape x.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_softsign.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Softsign activation function. — op_softsign","text":"","code":"x <- op_convert_to_tensor(c(-0.100, -10.0, 1.0, 0.0, 100.0)) op_softsign(x) ## tf.Tensor([-0.09090909 -0.90909091  0.5         0.          0.99009901], shape=(5), dtype=float64) x <- seq(-10, 10, .1) plot(x, op_softsign(x), ylim = c(-1, 1))"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_solve.html","id":null,"dir":"Reference","previous_headings":"","what":"Solves a linear system of equations given by a x = b. — op_solve","title":"Solves a linear system of equations given by a x = b. — op_solve","text":"Solves x equation %*% x == b.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_solve.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Solves a linear system of equations given by a x = b. — op_solve","text":"","code":"op_solve(a, b)"},{"path":"https://keras3.posit.co/dev/reference/op_solve.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Solves a linear system of equations given by a x = b. — op_solve","text":"tensor shape (..., M, M) representing coefficients matrix. b tensor shape (..., M) (..., M, N) representing right-hand side \"dependent variable\" matrix.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_solve.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Solves a linear system of equations given by a x = b. — op_solve","text":"tensor shape (..., M) (..., M, N) representing solution linear system. Returned shape identical b.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_solve.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Solves a linear system of equations given by a x = b. — op_solve","text":"","code":"a <- op_array(c(1, 2, 4, 5), dtype=\"float32\") |> op_reshape(c(2, 2)) b <- op_array(c(2, 4, 8, 10), dtype=\"float32\") |> op_reshape(c(2, 2)) op_solve(a, b) ## tf.Tensor( ## [[2. 0.] ##  [0. 2.]], shape=(2, 2), dtype=float32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_solve_triangular.html","id":null,"dir":"Reference","previous_headings":"","what":"Solves a linear system of equations given by a %*% x = b. — op_solve_triangular","title":"Solves a linear system of equations given by a %*% x = b. — op_solve_triangular","text":"Solves linear system equations given %*% x = b.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_solve_triangular.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Solves a linear system of equations given by a %*% x = b. — op_solve_triangular","text":"","code":"op_solve_triangular(a, b, lower = FALSE)"},{"path":"https://keras3.posit.co/dev/reference/op_solve_triangular.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Solves a linear system of equations given by a %*% x = b. — op_solve_triangular","text":"tensor shape (..., M, M) representing coefficients matrix. b tensor shape (..., M) (..., M, N) representing right-hand side \"dependent variable\" matrix. lower logical. Use data contained lower triangle . Default use upper triangle.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_solve_triangular.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Solves a linear system of equations given by a %*% x = b. — op_solve_triangular","text":"tensor shape (..., M) (..., M, N) representing solution linear system. Returned shape identical b.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_sort.html","id":null,"dir":"Reference","previous_headings":"","what":"Sorts the elements of x along a given axis in ascending order. — op_sort","title":"Sorts the elements of x along a given axis in ascending order. — op_sort","text":"Sorts elements x along given axis ascending order.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_sort.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Sorts the elements of x along a given axis in ascending order. — op_sort","text":"","code":"op_sort(x, axis = -1L)"},{"path":"https://keras3.posit.co/dev/reference/op_sort.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Sorts the elements of x along a given axis in ascending order. — op_sort","text":"x Input tensor. axis Axis along sort. NULL, tensor flattened sorting. Defaults -1; last axis.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_sort.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Sorts the elements of x along a given axis in ascending order. — op_sort","text":"Sorted tensor.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_sparse_categorical_crossentropy.html","id":null,"dir":"Reference","previous_headings":"","what":"Computes sparse categorical cross-entropy loss. — op_sparse_categorical_crossentropy","title":"Computes sparse categorical cross-entropy loss. — op_sparse_categorical_crossentropy","text":"sparse categorical cross-entropy loss similar categorical cross-entropy, used target tensor contains integer class labels instead one-hot encoded vectors. measures dissimilarity target output probabilities logits.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_sparse_categorical_crossentropy.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes sparse categorical cross-entropy loss. — op_sparse_categorical_crossentropy","text":"","code":"op_sparse_categorical_crossentropy(   target,   output,   from_logits = FALSE,   axis = -1L )"},{"path":"https://keras3.posit.co/dev/reference/op_sparse_categorical_crossentropy.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Computes sparse categorical cross-entropy loss. — op_sparse_categorical_crossentropy","text":"target target tensor representing true class labels integers. shape match shape output tensor except last dimension. output output tensor representing predicted probabilities logits. shape match shape target tensor except last dimension. from_logits (optional) Whether output tensor logits probabilities. Set TRUE output represents logits; otherwise, set FALSE output represents probabilities. Defaults FALSE. axis (optional) axis along sparse categorical cross-entropy computed. Defaults -1, corresponds last dimension tensors.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_sparse_categorical_crossentropy.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Computes sparse categorical cross-entropy loss. — op_sparse_categorical_crossentropy","text":"Integer tensor: computed sparse categorical cross-entropy loss target output.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_sparse_categorical_crossentropy.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Computes sparse categorical cross-entropy loss. — op_sparse_categorical_crossentropy","text":"","code":"target <- op_array(c(0, 1, 2), dtype=\"int32\") output <- op_array(rbind(c(0.9, 0.05, 0.05),                         c(0.1, 0.8,  0.1),                         c(0.2, 0.3,  0.5))) op_sparse_categorical_crossentropy(target, output) ## tf.Tensor([0.10536052 0.22314355 0.69314718], shape=(3), dtype=float64)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_sparse_plus.html","id":null,"dir":"Reference","previous_headings":"","what":"SparsePlus activation function. — op_sparse_plus","title":"SparsePlus activation function. — op_sparse_plus","text":"defined f(x) = 0 x <= -1. f(x) = (1/4) * (x + 1)^2 -1 < x < 1. f(x) = x x >= 1.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_sparse_plus.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"SparsePlus activation function. — op_sparse_plus","text":"","code":"op_sparse_plus(x)"},{"path":"https://keras3.posit.co/dev/reference/op_sparse_plus.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"SparsePlus activation function. — op_sparse_plus","text":"x Input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_sparse_plus.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"SparsePlus activation function. — op_sparse_plus","text":"tensor shape x.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_sparse_plus.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"SparsePlus activation function. — op_sparse_plus","text":"","code":"x <- op_array(c(-1.0, 0.0, 1.0)) op_sparse_plus(x) ## tf.Tensor([0.   0.25 1.  ], shape=(3), dtype=float32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_sparse_sigmoid.html","id":null,"dir":"Reference","previous_headings":"","what":"Sparse sigmoid activation function. — op_sparse_sigmoid","title":"Sparse sigmoid activation function. — op_sparse_sigmoid","text":"defined f(x) = 0 x <= -1, f(x) = 0.5 * (x + 1) -1 < x < 1, f(x) = 1 x >= 1.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_sparse_sigmoid.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Sparse sigmoid activation function. — op_sparse_sigmoid","text":"","code":"op_sparse_sigmoid(x)"},{"path":"https://keras3.posit.co/dev/reference/op_sparse_sigmoid.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Sparse sigmoid activation function. — op_sparse_sigmoid","text":"x Input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_sparse_sigmoid.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Sparse sigmoid activation function. — op_sparse_sigmoid","text":"tensor shape x.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_sparse_sigmoid.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Sparse sigmoid activation function. — op_sparse_sigmoid","text":"","code":"x <- op_array(c(-1.0, 0.0, 1.0)) op_sparse_sigmoid(x) ## tf.Tensor([0.  0.5 1. ], shape=(3), dtype=float32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_sparsemax.html","id":null,"dir":"Reference","previous_headings":"","what":"Sparsemax activation function. — op_sparsemax","title":"Sparsemax activation function. — op_sparsemax","text":"batch , class j, sparsemax activation function defined : sparsemax(x)[, j] = max(x[, j] - (x[, :]), 0).","code":""},{"path":"https://keras3.posit.co/dev/reference/op_sparsemax.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Sparsemax activation function. — op_sparsemax","text":"","code":"op_sparsemax(x, axis = -1L)"},{"path":"https://keras3.posit.co/dev/reference/op_sparsemax.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Sparsemax activation function. — op_sparsemax","text":"x Input tensor. axis int, axis along sparsemax operation applied.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_sparsemax.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Sparsemax activation function. — op_sparsemax","text":"tensor, output sparsemax transformation. type shape x.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_sparsemax.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Sparsemax activation function. — op_sparsemax","text":"","code":"x <- op_array(c(-1., 0., 1.)) op_sparsemax(x) ## tf.Tensor([0. 0. 1.], shape=(3), dtype=float32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_split.html","id":null,"dir":"Reference","previous_headings":"","what":"Split a tensor into chunks. — op_split","title":"Split a tensor into chunks. — op_split","text":"Split tensor chunks.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_split.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Split a tensor into chunks. — op_split","text":"","code":"op_split(x, indices_or_sections, axis = 1L)"},{"path":"https://keras3.posit.co/dev/reference/op_split.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Split a tensor into chunks. — op_split","text":"x Input tensor. indices_or_sections integer, N, tensor split N equal sections along axis. 1-D array sorted integers, entries indicate indices tensor split along axis. axis Axis along split. Defaults 1, first axis.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_split.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Split a tensor into chunks. — op_split","text":"list tensors.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_split.html","id":"note","dir":"Reference","previous_headings":"","what":"Note","title":"Split a tensor into chunks. — op_split","text":"split result equal division using Torch backend.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_split.html","id":"example","dir":"Reference","previous_headings":"","what":"Example","title":"Split a tensor into chunks. — op_split","text":"","code":"x <- op_arange(12)  # pass a scalar integer for n sections op_split(x, 2) ## [[1]] ## tf.Tensor([1. 2. 3. 4. 5. 6.], shape=(6), dtype=float32) ## ## [[2]] ## tf.Tensor([ 7.  8.  9. 10. 11. 12.], shape=(6), dtype=float32) op_split(x, 3) ## [[1]] ## tf.Tensor([1. 2. 3. 4.], shape=(4), dtype=float32) ## ## [[2]] ## tf.Tensor([5. 6. 7. 8.], shape=(4), dtype=float32) ## ## [[3]] ## tf.Tensor([ 9. 10. 11. 12.], shape=(4), dtype=float32) # 1-d array/tensor for indices op_split(x, array(c(3, 8))) ## [[1]] ## tf.Tensor([1. 2.], shape=(2), dtype=float32) ## ## [[2]] ## tf.Tensor([3. 4. 5. 6. 7.], shape=(5), dtype=float32) ## ## [[3]] ## tf.Tensor([ 8.  9. 10. 11. 12.], shape=(5), dtype=float32) op_split(x, array(c(3))) ## [[1]] ## tf.Tensor([1. 2.], shape=(2), dtype=float32) ## ## [[2]] ## tf.Tensor([ 3.  4.  5.  6.  7.  8.  9. 10. 11. 12.], shape=(10), dtype=float32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_sqrt.html","id":null,"dir":"Reference","previous_headings":"","what":"Return the non-negative square root of a tensor, element-wise. — op_sqrt","title":"Return the non-negative square root of a tensor, element-wise. — op_sqrt","text":"Return non-negative square root tensor, element-wise.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_sqrt.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Return the non-negative square root of a tensor, element-wise. — op_sqrt","text":"","code":"op_sqrt(x)"},{"path":"https://keras3.posit.co/dev/reference/op_sqrt.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Return the non-negative square root of a tensor, element-wise. — op_sqrt","text":"x Input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_sqrt.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Return the non-negative square root of a tensor, element-wise. — op_sqrt","text":"Output tensor, non-negative square root x.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_square.html","id":null,"dir":"Reference","previous_headings":"","what":"Return the element-wise square of the input. — op_square","title":"Return the element-wise square of the input. — op_square","text":"Return element-wise square input.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_square.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Return the element-wise square of the input. — op_square","text":"","code":"op_square(x)"},{"path":"https://keras3.posit.co/dev/reference/op_square.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Return the element-wise square of the input. — op_square","text":"x Input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_square.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Return the element-wise square of the input. — op_square","text":"Output tensor, square x.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_squareplus.html","id":null,"dir":"Reference","previous_headings":"","what":"Squareplus activation function. — op_squareplus","title":"Squareplus activation function. — op_squareplus","text":"Squareplus activation function defined : f(x) = (x + sqrt(x^2 + b)) / 2","code":""},{"path":"https://keras3.posit.co/dev/reference/op_squareplus.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Squareplus activation function. — op_squareplus","text":"","code":"op_squareplus(x, b = 4L)"},{"path":"https://keras3.posit.co/dev/reference/op_squareplus.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Squareplus activation function. — op_squareplus","text":"x Input tensor. b Smoothness parameter. Defaults 4.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_squareplus.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Squareplus activation function. — op_squareplus","text":"tensor shape x.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_squareplus.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Squareplus activation function. — op_squareplus","text":"","code":"x <- op_array(c(-1.0, 0.0, 1.0)) op_squareplus(x) ## tf.Tensor([0.618034 1.       1.618034], shape=(3), dtype=float32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_squeeze.html","id":null,"dir":"Reference","previous_headings":"","what":"Remove axes of length one from x. — op_squeeze","title":"Remove axes of length one from x. — op_squeeze","text":"Remove axes length one x.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_squeeze.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Remove axes of length one from x. — op_squeeze","text":"","code":"op_squeeze(x, axis = NULL)"},{"path":"https://keras3.posit.co/dev/reference/op_squeeze.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Remove axes of length one from x. — op_squeeze","text":"x Input tensor. axis Select subset entries length one shape.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_squeeze.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Remove axes of length one from x. — op_squeeze","text":"input tensor subset dimensions length 1 removed.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_stack.html","id":null,"dir":"Reference","previous_headings":"","what":"Join a sequence of tensors along a new axis. — op_stack","title":"Join a sequence of tensors along a new axis. — op_stack","text":"axis parameter specifies index new axis dimensions result.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_stack.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Join a sequence of tensors along a new axis. — op_stack","text":"","code":"op_stack(x, axis = 1L)"},{"path":"https://keras3.posit.co/dev/reference/op_stack.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Join a sequence of tensors along a new axis. — op_stack","text":"x sequence tensors. axis Axis along stack. Defaults 1, first axis.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_stack.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Join a sequence of tensors along a new axis. — op_stack","text":"stacked tensor.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_std.html","id":null,"dir":"Reference","previous_headings":"","what":"Compute the standard deviation along the specified axis. — op_std","title":"Compute the standard deviation along the specified axis. — op_std","text":"Compute standard deviation along specified axis.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_std.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Compute the standard deviation along the specified axis. — op_std","text":"","code":"op_std(x, axis = NULL, keepdims = FALSE)"},{"path":"https://keras3.posit.co/dev/reference/op_std.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Compute the standard deviation along the specified axis. — op_std","text":"x Input tensor. axis Axis along compute standard deviation. Default compute standard deviation flattened tensor. keepdims set TRUE, axes reduced left result dimensions size one.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_std.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Compute the standard deviation along the specified axis. — op_std","text":"Output tensor containing standard deviation values.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_stft.html","id":null,"dir":"Reference","previous_headings":"","what":"Short-Time Fourier Transform along the last axis of the input. — op_stft","title":"Short-Time Fourier Transform along the last axis of the input. — op_stft","text":"STFT computes Fourier transform short overlapping windows input. giving frequency components signal change time.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_stft.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Short-Time Fourier Transform along the last axis of the input. — op_stft","text":"","code":"op_stft(   x,   sequence_length,   sequence_stride,   fft_length,   window = \"hann\",   center = TRUE )"},{"path":"https://keras3.posit.co/dev/reference/op_stft.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Short-Time Fourier Transform along the last axis of the input. — op_stft","text":"x Input tensor. sequence_length integer representing sequence length. sequence_stride integer representing sequence hop size. fft_length integer representing size FFT apply. specified, uses smallest power 2 enclosing sequence_length. window string, tensor window NULL. window string, available values \"hann\" \"hamming\". window tensor, used directly window length must sequence_length. window NULL, windowing used. Defaults \"hann\". center Whether pad x sides t-th sequence centered time t * sequence_stride. Otherwise, t-th sequence begins time t * sequence_stride. Defaults TRUE.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_stft.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Short-Time Fourier Transform along the last axis of the input. — op_stft","text":"list containing two tensors - real imaginary parts STFT output.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_stft.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Short-Time Fourier Transform along the last axis of the input. — op_stft","text":"","code":"x <- op_array(c(0, 1, 2, 3, 4)) op_stft(x, 3, 2, 3) ## [[1]] ## tf.Tensor( ## [[ 0.  0.] ##  [ 2. -1.] ##  [ 4. -2.]], shape=(3, 2), dtype=float32) ## ## [[2]] ## tf.Tensor( ## [[ 0.         0.       ] ##  [ 0.        -1.7320508] ##  [ 0.        -3.4641016]], shape=(3, 2), dtype=float32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_stop_gradient.html","id":null,"dir":"Reference","previous_headings":"","what":"Stops gradient computation. — op_stop_gradient","title":"Stops gradient computation. — op_stop_gradient","text":"Stops gradient computation.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_stop_gradient.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Stops gradient computation. — op_stop_gradient","text":"","code":"op_stop_gradient(variable)"},{"path":"https://keras3.posit.co/dev/reference/op_stop_gradient.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Stops gradient computation. — op_stop_gradient","text":"variable tensor variable gradient computation disabled.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_stop_gradient.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Stops gradient computation. — op_stop_gradient","text":"variable gradient computation disabled.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_stop_gradient.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Stops gradient computation. — op_stop_gradient","text":"","code":"var <- op_convert_to_tensor(c(1, 2, 3), dtype=\"float32\") var <- op_stop_gradient(var)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_subset.html","id":null,"dir":"Reference","previous_headings":"","what":"Subset elements from a tensor — op_subset","title":"Subset elements from a tensor — op_subset","text":"Extract elements tensor using common R-style [ indexing idioms. function can also conveniently accessed via syntax tensor@r[...].","code":""},{"path":"https://keras3.posit.co/dev/reference/op_subset.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Subset elements from a tensor — op_subset","text":"","code":"op_subset(x, ...)  op_subset(x, ...) <- value  op_subset_set(x, ..., value)"},{"path":"https://keras3.posit.co/dev/reference/op_subset.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Subset elements from a tensor — op_subset","text":"x Input tensor. ... Indices specifying elements extract. argument ... can : integer scalar 1-d integer logical vector NULL newaxis .. symbol slice expression using : single argument supplied ..., ..1 can also : logical array shape x integer matrix ncol(..1) == op_rank(x) value new value replace selected subset .","code":""},{"path":"https://keras3.posit.co/dev/reference/op_subset.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Subset elements from a tensor — op_subset","text":"tensor containing subset elements.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_subset.html","id":"details","dir":"Reference","previous_headings":"","what":"Details","title":"Subset elements from a tensor — op_subset","text":"semantics similar R's [, differences:","code":""},{"path":"https://keras3.posit.co/dev/reference/op_subset.html","id":"differences-from-r-s-","dir":"Reference","previous_headings":"","what":"Differences from R's [:","title":"Subset elements from a tensor — op_subset","text":"Negative indices follow Python-style indexing, counting end array. NULL newaxis adds new dimension (equivalent op_expand_dims()). fewer indices dimensions (op_rank(x)) provided, missing dimensions implicitly filled. example, x matrix, x[1] returns first row. .. all_dims() expands include unspecified dimensions (see examples). Extended slicing syntax (:) supported, including: Strided steps: x@r[start:end:step] NA values start end. NA start defaults 1, NA end defaults axis size. logical array matching shape x selects elements row-wise order.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_subset.html","id":"similarities-with-r-s-","dir":"Reference","previous_headings":"","what":"Similarities with R's [:","title":"Subset elements from a tensor — op_subset","text":"Similarities R's [ (differences Python's [): Positive indices 1-based. Slices (x[start:end]) inclusive end. 1-d logical/integer arrays subset along respective axis. Multiple vectors provided different axes return intersected subsets. single integer matrix ncol() == op_rank(x) selects elements coordinates. row matrix specifies location one value, column corresponds axis tensor subsetted.  means use 2-column matrix subset matrix, 3-column matrix subset 3d array, .","code":""},{"path":"https://keras3.posit.co/dev/reference/op_subset.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Subset elements from a tensor — op_subset","text":"","code":"(x <- op_arange(5L) + 10L) ## tf.Tensor([11 12 13 14 15], shape=(5), dtype=int32) # Basic example, get first element op_subset(x, 1) ## tf.Tensor(11, shape=(), dtype=int32) # Use `@r[` syntax x@r[1]           # same as `op_subset(x, 1)` ## tf.Tensor(11, shape=(), dtype=int32) x@r[1:2]         # get the first 2 elements ## tf.Tensor([11 12], shape=(2), dtype=int32) x@r[c(1, 3)]     # first and third element ## tf.Tensor([11 13], shape=(2), dtype=int32) # Negative indices x@r[-1]          # last element ## tf.Tensor(15, shape=(), dtype=int32) x@r[-2]          # second to last element ## tf.Tensor(14, shape=(), dtype=int32) x@r[c(-1, -2)]   # last and second to last elements ## tf.Tensor([15 14], shape=(2), dtype=int32) x@r[c(-2, -1)]   # second to last and last elements ## tf.Tensor([14 15], shape=(2), dtype=int32) x@r[c(1, -1)]    # first and last elements ## tf.Tensor([11 15], shape=(2), dtype=int32) # Slices x@r[1:3]          # first 3 elements ## tf.Tensor([11 12 13], shape=(3), dtype=int32) x@r[NA:3]         # first 3 elements ## tf.Tensor([11 12 13], shape=(3), dtype=int32) x@r[1:5]          # all elements ## tf.Tensor([11 12 13 14 15], shape=(5), dtype=int32) x@r[1:-1]         # all elements ## tf.Tensor([11 12 13 14 15], shape=(5), dtype=int32) x@r[NA:NA]        # all elements ## tf.Tensor([11 12 13 14 15], shape=(5), dtype=int32) x@r[]             # all elements ## tf.Tensor([11 12 13 14 15], shape=(5), dtype=int32) x@r[1:-2]         # drop last element ## tf.Tensor([11 12 13 14], shape=(4), dtype=int32) x@r[NA:-2]        # drop last element ## tf.Tensor([11 12 13 14], shape=(4), dtype=int32) x@r[2:NA]         # drop first element ## tf.Tensor([12 13 14 15], shape=(4), dtype=int32) # 2D array examples xr <- array(1:12, c(3, 4)) x <- op_convert_to_tensor(xr)  # Basic subsetting x@r[1, ]      # first row ## tf.Tensor([ 1  4  7 10], shape=(4), dtype=int32) x@r[1]        # also first row! Missing axes are implicitly inserted ## tf.Tensor([ 1  4  7 10], shape=(4), dtype=int32) x@r[-1]       # last row ## tf.Tensor([ 3  6  9 12], shape=(4), dtype=int32) x@r[, 2]      # second column ## tf.Tensor([4 5 6], shape=(3), dtype=int32) x@r[, 2:2]    # second column, but shape preserved (like [, drop=FALSE]) ## tf.Tensor( ## [[4] ##  [5] ##  [6]], shape=(3, 1), dtype=int32) # Subsetting with a boolean array # Note: extracted elements are selected row-wise, not column-wise mask <- x >= 6 x@r[mask]             # returns a 1D tensor ## tf.Tensor([ 7 10  8 11  6  9 12], shape=(7), dtype=int32) x.r <- as.array(x) mask.r <- as.array(mask) # as.array(x)[mask] selects column-wise. Use `aperm()` to reverse search order. all(aperm(x.r)[aperm(mask.r)] == as.array(x@r[mask])) ## [1] TRUE # Subsetting with a matrix of index positions indices <- rbind(c(1, 1), c(2, 2), c(3, 3)) x@r[indices] # get diagonal elements ## tf.Tensor([1 5 9], shape=(3), dtype=int32) x.r[indices] # same as subsetting an R array ## [1] 1 5 9 # 3D array examples # Image: 4x4 pixels, 3 colors (RGB) # Tensor shape: (img_height, img_width, img_color_channels) shp <- shape(4, 4, 3) x <- op_arange(prod(shp)) |> op_reshape(shp)  # Convert to a batch of images by inserting a new axis # New shape: (batch_size, img_height, img_width, img_color_channels) x@r[newaxis, , , ] |> op_shape() ## shape(1, 4, 4, 3) x@r[newaxis] |> op_shape()  # same as above ## shape(1, 4, 4, 3) x@r[NULL] |> op_shape()     # same as above ## shape(1, 4, 4, 3) x <- x@r[newaxis] # Extract color channels x@r[, , , 1]          # red channel ## tf.Tensor( ## [[[ 1.  4.  7. 10.] ##   [13. 16. 19. 22.] ##   [25. 28. 31. 34.] ##   [37. 40. 43. 46.]]], shape=(1, 4, 4), dtype=float32) x@r[.., 1]            # red channel, same as above using .. shorthand ## tf.Tensor( ## [[[ 1.  4.  7. 10.] ##   [13. 16. 19. 22.] ##   [25. 28. 31. 34.] ##   [37. 40. 43. 46.]]], shape=(1, 4, 4), dtype=float32) x@r[.., 2]            # green channel ## tf.Tensor( ## [[[ 2.  5.  8. 11.] ##   [14. 17. 20. 23.] ##   [26. 29. 32. 35.] ##   [38. 41. 44. 47.]]], shape=(1, 4, 4), dtype=float32) x@r[.., 3]            # blue channel ## tf.Tensor( ## [[[ 3.  6.  9. 12.] ##   [15. 18. 21. 24.] ##   [27. 30. 33. 36.] ##   [39. 42. 45. 48.]]], shape=(1, 4, 4), dtype=float32) # .. expands to all unspecified axes. op_shape(x@r[]) ## shape(1, 4, 4, 3) op_shape(x@r[..]) ## shape(1, 4, 4, 3) op_shape(x@r[1, ..]) ## shape(4, 4, 3) op_shape(x@r[1, .., 1, 1]) ## shape(4) op_shape(x@r[1, 1, 1, .., 1]) ## shape() # op_subset<- uses the same semantics, but note that not all tensors # support modification. E.g., TensorFlow constant tensors cannot be modified, # while TensorFlow Variables can be.  (x <- tensorflow::tf$Variable(matrix(1, nrow = 2, ncol = 3))) ## <tf.Variable 'Variable:0' shape=(2, 3) dtype=float64, numpy= ## array([[1., 1., 1.], ##        [1., 1., 1.]])> op_subset(x, 1) <- 9 x ## <tf.Variable 'UnreadVariable' shape=(2, 3) dtype=float64, numpy= ## array([[9., 9., 9.], ##        [1., 1., 1.]])> x@r[1,1] <- 33 x ## <tf.Variable 'UnreadVariable' shape=(2, 3) dtype=float64, numpy= ## array([[33.,  9.,  9.], ##        [ 1.,  1.,  1.]])>"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_subtract.html","id":null,"dir":"Reference","previous_headings":"","what":"Subtract arguments element-wise. — op_subtract","title":"Subtract arguments element-wise. — op_subtract","text":"Note function automatically called using R operator - tensor.","code":"x <- op_ones(c(3)) op_subtract(x, x) ## tf.Tensor([0. 0. 0.], shape=(3), dtype=float32) x - x ## tf.Tensor([0. 0. 0.], shape=(3), dtype=float32)"},{"path":"https://keras3.posit.co/dev/reference/op_subtract.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Subtract arguments element-wise. — op_subtract","text":"","code":"op_subtract(x1, x2)"},{"path":"https://keras3.posit.co/dev/reference/op_subtract.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Subtract arguments element-wise. — op_subtract","text":"x1 First input tensor. x2 Second input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_subtract.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Subtract arguments element-wise. — op_subtract","text":"Output tensor, element-wise difference x1 x2.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_sum.html","id":null,"dir":"Reference","previous_headings":"","what":"Sum of a tensor over the given axes. — op_sum","title":"Sum of a tensor over the given axes. — op_sum","text":"Sum tensor given axes.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_sum.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Sum of a tensor over the given axes. — op_sum","text":"","code":"op_sum(x, axis = NULL, keepdims = FALSE)"},{"path":"https://keras3.posit.co/dev/reference/op_sum.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Sum of a tensor over the given axes. — op_sum","text":"x Input tensor. axis Axis axes along sum computed. default compute sum flattened tensor. keepdims set TRUE, axes reduced left result dimensions size one.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_sum.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Sum of a tensor over the given axes. — op_sum","text":"Output tensor containing sum.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_svd.html","id":null,"dir":"Reference","previous_headings":"","what":"Computes the singular value decomposition of a matrix. — op_svd","title":"Computes the singular value decomposition of a matrix. — op_svd","text":"Computes singular value decomposition matrix.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_svd.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes the singular value decomposition of a matrix. — op_svd","text":"","code":"op_svd(x, full_matrices = TRUE, compute_uv = TRUE)"},{"path":"https://keras3.posit.co/dev/reference/op_svd.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Computes the singular value decomposition of a matrix. — op_svd","text":"x Input tensor shape (..., M, N). full_matrices Logical compute_uv Logical","code":""},{"path":"https://keras3.posit.co/dev/reference/op_svd.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Computes the singular value decomposition of a matrix. — op_svd","text":"list three tensors: tensor shape (..., M, M) containing left singular vectors, tensor shape (..., M, N) containing singular values tensor shape (..., N, N) containing right singular vectors.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_swapaxes.html","id":null,"dir":"Reference","previous_headings":"","what":"Interchange two axes of a tensor. — op_swapaxes","title":"Interchange two axes of a tensor. — op_swapaxes","text":"Interchange two axes tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_swapaxes.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Interchange two axes of a tensor. — op_swapaxes","text":"","code":"op_swapaxes(x, axis1, axis2)"},{"path":"https://keras3.posit.co/dev/reference/op_swapaxes.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Interchange two axes of a tensor. — op_swapaxes","text":"x Input tensor. axis1 First axis. axis2 Second axis.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_swapaxes.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Interchange two axes of a tensor. — op_swapaxes","text":"tensor axes swapped.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_switch.html","id":null,"dir":"Reference","previous_headings":"","what":"Apply exactly one of the branches given by index. — op_switch","title":"Apply exactly one of the branches given by index. — op_switch","text":"index bounds, clamped within bounds. semantics switch given roughly implementation:","code":"op_switch <- function(index, branches, ...) {   index <- op_clip(index, 1, length(branches))   branches[[index]](...) }"},{"path":"https://keras3.posit.co/dev/reference/op_switch.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Apply exactly one of the branches given by index. — op_switch","text":"","code":"op_switch(index, branches, ...)"},{"path":"https://keras3.posit.co/dev/reference/op_switch.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Apply exactly one of the branches given by index. — op_switch","text":"index integer scalar indicating branch function apply (1-based). branches list functions applied based index. ... Inputs whichever branch applied.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_switch.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Apply exactly one of the branches given by index. — op_switch","text":"outputs branch(...) branch selected based index.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_switch.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Apply exactly one of the branches given by index. — op_switch","text":"","code":"add_fn <- function(x, y) x + y + 100 subtract_fn <- function(x, y) x - y - 100 x <- op_array(2) y <- op_array(0.5) branches <- list(add_fn, subtract_fn) op_switch(0, branches, x, y) # + ## tf.Tensor(102.5, shape=(), dtype=float32) op_switch(1, branches, x, y) # + ## tf.Tensor(102.5, shape=(), dtype=float32) op_switch(2, branches, x, y) # - ## tf.Tensor(-98.5, shape=(), dtype=float32) op_switch(3, branches, x, y) # - ## tf.Tensor(-98.5, shape=(), dtype=float32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_take.html","id":null,"dir":"Reference","previous_headings":"","what":"Take elements from a tensor along an axis. — op_take","title":"Take elements from a tensor along an axis. — op_take","text":"Take elements tensor along axis.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_take.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Take elements from a tensor along an axis. — op_take","text":"","code":"op_take(x, indices, axis = NULL, zero_indexed = FALSE)"},{"path":"https://keras3.posit.co/dev/reference/op_take.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Take elements from a tensor along an axis. — op_take","text":"x Source tensor. indices indices values extract. axis axis select values. default, flattened input tensor used. zero_indexed TRUE, treats indices zero-based (0 encodes first position); FALSE (default), treats indices one-based (1 encodes first position).","code":""},{"path":"https://keras3.posit.co/dev/reference/op_take.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Take elements from a tensor along an axis. — op_take","text":"corresponding tensor values.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_take_along_axis.html","id":null,"dir":"Reference","previous_headings":"","what":"Select values from x at the 1-D indices along the given axis. — op_take_along_axis","title":"Select values from x at the 1-D indices along the given axis. — op_take_along_axis","text":"Select values x 1-D indices along given axis.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_take_along_axis.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Select values from x at the 1-D indices along the given axis. — op_take_along_axis","text":"","code":"op_take_along_axis(x, indices, axis = NULL, zero_indexed = FALSE)"},{"path":"https://keras3.posit.co/dev/reference/op_take_along_axis.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Select values from x at the 1-D indices along the given axis. — op_take_along_axis","text":"x Source tensor. indices indices values extract. axis axis select values. default, flattened input tensor used. zero_indexed TRUE, treats indices zero-based (0 encodes first position); FALSE (default), treats indices one-based (1 encodes first position).","code":""},{"path":"https://keras3.posit.co/dev/reference/op_take_along_axis.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Select values from x at the 1-D indices along the given axis. — op_take_along_axis","text":"corresponding tensor values.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_tan.html","id":null,"dir":"Reference","previous_headings":"","what":"Compute tangent, element-wise. — op_tan","title":"Compute tangent, element-wise. — op_tan","text":"Compute tangent, element-wise.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_tan.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Compute tangent, element-wise. — op_tan","text":"","code":"op_tan(x)"},{"path":"https://keras3.posit.co/dev/reference/op_tan.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Compute tangent, element-wise. — op_tan","text":"x Input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_tan.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Compute tangent, element-wise. — op_tan","text":"Output tensor shape x.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_tanh.html","id":null,"dir":"Reference","previous_headings":"","what":"Hyperbolic tangent, element-wise. — op_tanh","title":"Hyperbolic tangent, element-wise. — op_tanh","text":"Hyperbolic tangent, element-wise.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_tanh.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Hyperbolic tangent, element-wise. — op_tanh","text":"","code":"op_tanh(x)"},{"path":"https://keras3.posit.co/dev/reference/op_tanh.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Hyperbolic tangent, element-wise. — op_tanh","text":"x Input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_tanh.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Hyperbolic tangent, element-wise. — op_tanh","text":"Output tensor shape x.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_tanh_shrink.html","id":null,"dir":"Reference","previous_headings":"","what":"Applies the tanh shrink function element-wise. — op_tanh_shrink","title":"Applies the tanh shrink function element-wise. — op_tanh_shrink","text":"defined : f(x) = x - tanh(x).","code":""},{"path":"https://keras3.posit.co/dev/reference/op_tanh_shrink.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Applies the tanh shrink function element-wise. — op_tanh_shrink","text":"","code":"op_tanh_shrink(x)"},{"path":"https://keras3.posit.co/dev/reference/op_tanh_shrink.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Applies the tanh shrink function element-wise. — op_tanh_shrink","text":"x Input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_tanh_shrink.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Applies the tanh shrink function element-wise. — op_tanh_shrink","text":"Output tensor shape x, element transformed according tanh shrink operation.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_tanh_shrink.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Applies the tanh shrink function element-wise. — op_tanh_shrink","text":"","code":"x <- op_array(c(-1., 0., 1.)) op_tanh_shrink(x) ## tf.Tensor([-0.23840582  0.          0.23840582], shape=(3), dtype=float32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_tensordot.html","id":null,"dir":"Reference","previous_headings":"","what":"Compute the tensor dot product along specified axes. — op_tensordot","title":"Compute the tensor dot product along specified axes. — op_tensordot","text":"Compute tensor dot product along specified axes.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_tensordot.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Compute the tensor dot product along specified axes. — op_tensordot","text":"","code":"op_tensordot(x1, x2, axes = 3L)"},{"path":"https://keras3.posit.co/dev/reference/op_tensordot.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Compute the tensor dot product along specified axes. — op_tensordot","text":"x1 First tensor. x2 Second tensor. axes integer, N, sum last N axes x1 first N axes x2 order. sizes corresponding axes must match. , list axes summed , first sequence applying x1, second x2. sequences must length.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_tensordot.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Compute the tensor dot product along specified axes. — op_tensordot","text":"tensor dot product inputs.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_threshold.html","id":null,"dir":"Reference","previous_headings":"","what":"Threshold activation function. — op_threshold","title":"Threshold activation function. — op_threshold","text":"function thresholds input x follows: f(x) = x x > threshold, f(x) = default_value otherwise.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_threshold.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Threshold activation function. — op_threshold","text":"","code":"op_threshold(x, threshold, default_value)"},{"path":"https://keras3.posit.co/dev/reference/op_threshold.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Threshold activation function. — op_threshold","text":"x Input tensor. threshold value decides retain replace x. default_value Value assign x <= threshold.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_threshold.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Threshold activation function. — op_threshold","text":"tensor shape x.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_threshold.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Threshold activation function. — op_threshold","text":"","code":"x <- op_array(c(-1.0, 0.0, 1.0, 2.0)) op_threshold(x, 1, 0) ## tf.Tensor([0. 0. 0. 2.], shape=(4), dtype=float32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_tile.html","id":null,"dir":"Reference","previous_headings":"","what":"Repeat x the number of times given by repeats. — op_tile","title":"Repeat x the number of times given by repeats. — op_tile","text":"repeats length d, result dimension max(d, x.ndim). x.ndim < d, x promoted d-dimensional prepending new axes. x.ndim > d, repeats promoted x.ndim prepending 1's .","code":""},{"path":"https://keras3.posit.co/dev/reference/op_tile.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Repeat x the number of times given by repeats. — op_tile","text":"","code":"op_tile(x, repeats)"},{"path":"https://keras3.posit.co/dev/reference/op_tile.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Repeat x the number of times given by repeats. — op_tile","text":"x Input tensor. repeats number repetitions x along axis.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_tile.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Repeat x the number of times given by repeats. — op_tile","text":"tiled output tensor.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_top_k.html","id":null,"dir":"Reference","previous_headings":"","what":"Finds the top-k values and their indices in a tensor. — op_top_k","title":"Finds the top-k values and their indices in a tensor. — op_top_k","text":"Finds top-k values indices tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_top_k.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Finds the top-k values and their indices in a tensor. — op_top_k","text":"","code":"op_top_k(x, k, sorted = TRUE, zero_indexed = FALSE)"},{"path":"https://keras3.posit.co/dev/reference/op_top_k.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Finds the top-k values and their indices in a tensor. — op_top_k","text":"x Input tensor. k integer representing number top elements retrieve. sorted boolean indicating whether sort output descending order. Defaults TRUE. zero_indexed TRUE, returned indices zero-based (0 encodes first position); FALSE (default), returned indices one-based (1 encodes first position).","code":""},{"path":"https://keras3.posit.co/dev/reference/op_top_k.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Finds the top-k values and their indices in a tensor. — op_top_k","text":"list containing two tensors. first tensor contains top-k values, second tensor contains indices top-k values input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_top_k.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Finds the top-k values and their indices in a tensor. — op_top_k","text":"","code":"x <- op_array(c(5, 2, 7, 1, 9, 3), \"int32\") op_top_k(x, k = 3) ## $values ## tf.Tensor([9 7 5], shape=(3), dtype=int32) ## ## $indices ## tf.Tensor([5 3 1], shape=(3), dtype=int32) c(values, indices) %<-% op_top_k(x, k = 3) values ## tf.Tensor([9 7 5], shape=(3), dtype=int32) indices ## tf.Tensor([5 3 1], shape=(3), dtype=int32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_trace.html","id":null,"dir":"Reference","previous_headings":"","what":"Return the sum along diagonals of the tensor. — op_trace","title":"Return the sum along diagonals of the tensor. — op_trace","text":"x 2-D, sum along diagonal given offset returned, .e., sum elements x[, +offset] . two dimensions, axes specified axis1 axis2 used determine 2-D sub-arrays whose traces returned. shape resulting tensor x axis1 axis2 removed.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_trace.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Return the sum along diagonals of the tensor. — op_trace","text":"","code":"op_trace(x, offset = 0L, axis1 = 1L, axis2 = 2L)"},{"path":"https://keras3.posit.co/dev/reference/op_trace.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Return the sum along diagonals of the tensor. — op_trace","text":"x Input tensor. offset Offset diagonal main diagonal. Can positive negative. Defaults 0. axis1 Axis used first axis 2-D sub-arrays. Defaults 1. (first axis). axis2 Axis used second axis 2-D sub-arrays. Defaults 2. (second axis).","code":""},{"path":"https://keras3.posit.co/dev/reference/op_trace.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Return the sum along diagonals of the tensor. — op_trace","text":"x 2-D, sum diagonal returned. x larger dimensions, tensor sums along diagonals returned.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_transpose.html","id":null,"dir":"Reference","previous_headings":"","what":"Returns a tensor with axes transposed. — op_transpose","title":"Returns a tensor with axes transposed. — op_transpose","text":"Returns tensor axes transposed.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_transpose.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Returns a tensor with axes transposed. — op_transpose","text":"","code":"op_transpose(x, axes = NULL)"},{"path":"https://keras3.posit.co/dev/reference/op_transpose.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Returns a tensor with axes transposed. — op_transpose","text":"x Input tensor. axes Sequence integers. Permutation dimensions x. default, order axes reversed.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_transpose.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Returns a tensor with axes transposed. — op_transpose","text":"x axes permuted.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_tri.html","id":null,"dir":"Reference","previous_headings":"","what":"Return a tensor with ones at and below a diagonal and zeros elsewhere. — op_tri","title":"Return a tensor with ones at and below a diagonal and zeros elsewhere. — op_tri","text":"Return tensor ones diagonal zeros elsewhere.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_tri.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Return a tensor with ones at and below a diagonal and zeros elsewhere. — op_tri","text":"","code":"op_tri(N, M = NULL, k = 0L, dtype = NULL)"},{"path":"https://keras3.posit.co/dev/reference/op_tri.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Return a tensor with ones at and below a diagonal and zeros elsewhere. — op_tri","text":"N Number rows tensor. M Number columns tensor. k sub-diagonal array filled. k = 0 main diagonal, k < 0 , k > 0 . default 0. dtype Data type returned tensor. default \"float32\".","code":""},{"path":"https://keras3.posit.co/dev/reference/op_tri.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Return a tensor with ones at and below a diagonal and zeros elsewhere. — op_tri","text":"Tensor lower triangle filled ones zeros elsewhere. T[, j] == 1 j <= + k, 0 otherwise.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_tril.html","id":null,"dir":"Reference","previous_headings":"","what":"Return lower triangle of a tensor. — op_tril","title":"Return lower triangle of a tensor. — op_tril","text":"tensors ndim exceeding 2, tril apply final two axes.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_tril.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Return lower triangle of a tensor. — op_tril","text":"","code":"op_tril(x, k = 0L)"},{"path":"https://keras3.posit.co/dev/reference/op_tril.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Return lower triangle of a tensor. — op_tril","text":"x Input tensor. k Diagonal zero elements. Defaults 0. main diagonal. k < 0 , k > 0 .","code":""},{"path":"https://keras3.posit.co/dev/reference/op_tril.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Return lower triangle of a tensor. — op_tril","text":"Lower triangle x, shape data type x.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_triu.html","id":null,"dir":"Reference","previous_headings":"","what":"Return upper triangle of a tensor. — op_triu","title":"Return upper triangle of a tensor. — op_triu","text":"tensors ndim exceeding 2, triu apply final two axes.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_triu.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Return upper triangle of a tensor. — op_triu","text":"","code":"op_triu(x, k = 0L)"},{"path":"https://keras3.posit.co/dev/reference/op_triu.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Return upper triangle of a tensor. — op_triu","text":"x Input tensor. k Diagonal zero elements. Defaults 0. main diagonal. k < 0 , k > 0 .","code":""},{"path":"https://keras3.posit.co/dev/reference/op_triu.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Return upper triangle of a tensor. — op_triu","text":"Upper triangle x, shape data type x.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_trunc.html","id":null,"dir":"Reference","previous_headings":"","what":"Return the truncated value of the input, element-wise. — op_trunc","title":"Return the truncated value of the input, element-wise. — op_trunc","text":"truncated value scalar x nearest integer closer zero x . short, fractional part signed number x discarded.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_trunc.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Return the truncated value of the input, element-wise. — op_trunc","text":"","code":"op_trunc(x)"},{"path":"https://keras3.posit.co/dev/reference/op_trunc.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Return the truncated value of the input, element-wise. — op_trunc","text":"x Input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_trunc.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Return the truncated value of the input, element-wise. — op_trunc","text":"truncated value element x.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_trunc.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Return the truncated value of the input, element-wise. — op_trunc","text":"","code":"x <- op_array(c(-1.7, -1.5, -0.2, 0.2, 1.5, 1.7, 2.0)) op_trunc(x) ## tf.Tensor([-1. -1. -0.  0.  1.  1.  2.], shape=(7), dtype=float32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_unravel_index.html","id":null,"dir":"Reference","previous_headings":"","what":"Convert flat indices to coordinate arrays in a given array shape. — op_unravel_index","title":"Convert flat indices to coordinate arrays in a given array shape. — op_unravel_index","text":"Convert flat indices coordinate arrays given array shape.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_unravel_index.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Convert flat indices to coordinate arrays in a given array shape. — op_unravel_index","text":"","code":"op_unravel_index(indices, shape)"},{"path":"https://keras3.posit.co/dev/reference/op_unravel_index.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Convert flat indices to coordinate arrays in a given array shape. — op_unravel_index","text":"indices integer array integers representing flat indices. shape shape array unravel .","code":""},{"path":"https://keras3.posit.co/dev/reference/op_unravel_index.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Convert flat indices to coordinate arrays in a given array shape. — op_unravel_index","text":"Tuple arrays dimension unraveled indices.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_unravel_index.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Convert flat indices to coordinate arrays in a given array shape. — op_unravel_index","text":"","code":"indices <- c(1, 5) shape <- array(c(3, 3)) op_unravel_index(indices, shape) ## [[1]] ## tf.Tensor([0 1], shape=(2), dtype=int32) ## ## [[2]] ## tf.Tensor([0 1], shape=(2), dtype=int32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_unstack.html","id":null,"dir":"Reference","previous_headings":"","what":"Unpacks the given dimension of a rank-R tensor into rank-(R-1) tensors. — op_unstack","title":"Unpacks the given dimension of a rank-R tensor into rank-(R-1) tensors. — op_unstack","text":"Unpacks given dimension rank-R tensor rank-(R-1) tensors.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_unstack.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Unpacks the given dimension of a rank-R tensor into rank-(R-1) tensors. — op_unstack","text":"","code":"op_unstack(x, num = NULL, axis = 1L)"},{"path":"https://keras3.posit.co/dev/reference/op_unstack.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Unpacks the given dimension of a rank-R tensor into rank-(R-1) tensors. — op_unstack","text":"x input tensor. num length dimension axis. Automatically inferred NULL. axis axis along unpack.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_unstack.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Unpacks the given dimension of a rank-R tensor into rank-(R-1) tensors. — op_unstack","text":"list tensors unpacked along given axis.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_unstack.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Unpacks the given dimension of a rank-R tensor into rank-(R-1) tensors. — op_unstack","text":"[3, 4))]: R:3,%204))","code":"x <- op_array(rbind(c(1, 2),                    c(3, 4))) op_unstack(x, axis=1) ## [[1]] ## tf.Tensor([1. 2.], shape=(2), dtype=float64) ## ## [[2]] ## tf.Tensor([3. 4.], shape=(2), dtype=float64) op_unstack(x, axis=2) ## [[1]] ## tf.Tensor([1. 3.], shape=(2), dtype=float64) ## ## [[2]] ## tf.Tensor([2. 4.], shape=(2), dtype=float64) all.equal(op_unstack(x),           op_unstack(x, axis = 1)) ## [1] TRUE all.equal(op_unstack(x, axis = -1),           op_unstack(x, axis = 2)) ## [1] TRUE # [array([1, 2)), array([3, 4))]"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_var.html","id":null,"dir":"Reference","previous_headings":"","what":"Compute the variance along the specified axes. — op_var","title":"Compute the variance along the specified axes. — op_var","text":"Compute variance along specified axes.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_var.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Compute the variance along the specified axes. — op_var","text":"","code":"op_var(x, axis = NULL, keepdims = FALSE)"},{"path":"https://keras3.posit.co/dev/reference/op_var.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Compute the variance along the specified axes. — op_var","text":"x Input tensor. axis Axis axes along variance computed. default compute variance flattened tensor. keepdims set TRUE, axes reduced left result dimensions size one.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_var.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Compute the variance along the specified axes. — op_var","text":"Output tensor containing variance.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_vdot.html","id":null,"dir":"Reference","previous_headings":"","what":"Return the dot product of two vectors. — op_vdot","title":"Return the dot product of two vectors. — op_vdot","text":"first argument complex, complex conjugate first argument used calculation dot product. Multidimensional tensors flattened dot product taken.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_vdot.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Return the dot product of two vectors. — op_vdot","text":"","code":"op_vdot(x1, x2)"},{"path":"https://keras3.posit.co/dev/reference/op_vdot.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Return the dot product of two vectors. — op_vdot","text":"x1 First input tensor. complex, complex conjugate taken calculation dot product. x2 Second input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_vdot.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Return the dot product of two vectors. — op_vdot","text":"Output tensor.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_vectorize.html","id":null,"dir":"Reference","previous_headings":"","what":"Turn a function into a vectorized function. — op_vectorize","title":"Turn a function into a vectorized function. — op_vectorize","text":"Turn function vectorized function.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_vectorize.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Turn a function into a vectorized function. — op_vectorize","text":"","code":"op_vectorize(func, ..., excluded = NULL, signature = NULL)"},{"path":"https://keras3.posit.co/dev/reference/op_vectorize.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Turn a function into a vectorized function. — op_vectorize","text":"func Callable single tensor argument. ... forward/backward compatability. excluded Optional set integers representing positional arguments function vectorized. passed directly func unmodified. signature Optional generalized universal function signature, e.g., \"(m,n),(n)->(m)\" vectorized matrix-vector multiplication. provided, func called (expected return) arrays shapes given size corresponding core dimensions. default, func assumed take scalar tensors input output.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_vectorize.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Turn a function into a vectorized function. — op_vectorize","text":"new function applies func every element input along axis 1 (batch axis, first axis).","code":""},{"path":"https://keras3.posit.co/dev/reference/op_vectorize.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Turn a function into a vectorized function. — op_vectorize","text":"","code":"# currently does not work w/ tensorflow backend if(config_backend() != \"tensorflow\") {    myfunc <- function(a, b) a + b    vfunc <- op_vectorize(myfunc)   y <- vfunc(c(1, 2, 3, 4), 2)   print(y)   # with Jax backend, y is:   # Array([3., 4., 5., 6.], dtype=float32) }"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_vectorized_map.html","id":null,"dir":"Reference","previous_headings":"","what":"Parallel map of function f on the first axis of tensor(s) elements. — op_vectorized_map","title":"Parallel map of function f on the first axis of tensor(s) elements. — op_vectorized_map","text":"Schematically, op_vectorized_map() maps first dimension provided tensors. elements list tensors, tensors required size first dimension, iterated together.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_vectorized_map.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Parallel map of function f on the first axis of tensor(s) elements. — op_vectorized_map","text":"","code":"op_vectorized_map(elements, f)"},{"path":"https://keras3.posit.co/dev/reference/op_vectorized_map.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Parallel map of function f on the first axis of tensor(s) elements. — op_vectorized_map","text":"elements see description f function taking either tensor, list tensors.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_vectorized_map.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Parallel map of function f on the first axis of tensor(s) elements. — op_vectorized_map","text":"tensor list tensors, result mapping f across elements.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_vectorized_map.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Parallel map of function f on the first axis of tensor(s) elements. — op_vectorized_map","text":"Note f may traced compiled. Meaning, R function may evaluated symbolic tensors using Jax TensorFlow backends, eager tensors. See output str() examples:","code":"(x <- op_arange(12L) |> op_reshape(c(3, 4))) ## tf.Tensor( ## [[ 1  2  3  4] ##  [ 5  6  7  8] ##  [ 9 10 11 12]], shape=(3, 4), dtype=int32) x |> op_vectorized_map(\\(row) {row + 10}) ## tf.Tensor( ## [[11 12 13 14] ##  [15 16 17 18] ##  [19 20 21 22]], shape=(3, 4), dtype=int32) list(x, x, x) |> op_vectorized_map(\\(rows) Reduce(`+`, rows)) ## tf.Tensor( ## [[ 3  6  9 12] ##  [15 18 21 24] ##  [27 30 33 36]], shape=(3, 4), dtype=int32) # simplest case, map f over rows of x, # where .x is 1 row of x input <- x output <- op_vectorized_map(input, function(.x) {   str(.x)   .x + 10 }) ## <tf.Tensor 'loop_body/GatherV2:0' shape=(4) dtype=int32> output ## tf.Tensor( ## [[11 12 13 14] ##  [15 16 17 18] ##  [19 20 21 22]], shape=(3, 4), dtype=int32) # map f over two tensors simultaneously. Here, # `.x` is a list of two # tensors. The return values from each call of `f(row)` are stacked to form the # final output input <- list(x, x) output <- op_vectorized_map(input, function(.x) {   str(.x)   .x[[1]] + 10 }) ## List of 2 ##  $ :<tf.Tensor 'loop_body/GatherV2:0' shape=(4) dtype=int32> ##  $ :<tf.Tensor 'loop_body/GatherV2_1:0' shape=(4) dtype=int32> output ## tf.Tensor( ## [[11 12 13 14] ##  [15 16 17 18] ##  [19 20 21 22]], shape=(3, 4), dtype=int32) # same as above, but now returning two tensors in the final output output <- op_vectorized_map(input, function(.x) {   str(.x)   c(.x1, .x2) %<-% .x   list(.x1+10, .x2+20) }) ## List of 2 ##  $ :<tf.Tensor 'loop_body/GatherV2:0' shape=(4) dtype=int32> ##  $ :<tf.Tensor 'loop_body/GatherV2_1:0' shape=(4) dtype=int32> output ## [[1]] ## tf.Tensor( ## [[11 12 13 14] ##  [15 16 17 18] ##  [19 20 21 22]], shape=(3, 4), dtype=int32) ## ## [[2]] ## tf.Tensor( ## [[21 22 23 24] ##  [25 26 27 28] ##  [29 30 31 32]], shape=(3, 4), dtype=int32) # passing named lists. # WARNING: if passing a named list, the order of elements of `.x` supplied # to `f` is not stable. Only retrieve elements by name. input <- list(name1 = x, name2 = x) output <- op_vectorized_map(input, function(.x) {   str(.x)   list(outname1 = .x$name1 + 10,        outname2 = .x$name2 + 20) }) ## List of 2 ##  $ name1:<tf.Tensor 'loop_body/GatherV2:0' shape=(4) dtype=int32> ##  $ name2:<tf.Tensor 'loop_body/GatherV2_1:0' shape=(4) dtype=int32> output ## $outname1 ## tf.Tensor( ## [[11 12 13 14] ##  [15 16 17 18] ##  [19 20 21 22]], shape=(3, 4), dtype=int32) ## ## $outname2 ## tf.Tensor( ## [[21 22 23 24] ##  [25 26 27 28] ##  [29 30 31 32]], shape=(3, 4), dtype=int32) # passing a tuple() is equivalent to passing an unnamed list() input <- tuple(x, x) output <- op_vectorized_map(input, function(.x) {   str(.x)   list(.x[[1]] + 10) }) ## List of 2 ##  $ :<tf.Tensor 'loop_body/GatherV2:0' shape=(4) dtype=int32> ##  $ :<tf.Tensor 'loop_body/GatherV2_1:0' shape=(4) dtype=int32> output ## [[1]] ## tf.Tensor( ## [[11 12 13 14] ##  [15 16 17 18] ##  [19 20 21 22]], shape=(3, 4), dtype=int32)"},{"path":"https://keras3.posit.co/dev/reference/op_vectorized_map.html","id":"debugging-f","dir":"Reference","previous_headings":"","what":"Debugging f","title":"Parallel map of function f on the first axis of tensor(s) elements. — op_vectorized_map","text":"Even eager contexts, op_vectorized_map() may trace f. case, want eagerly debug f (e.g., browser()), can swap manual (slow) implementation op_vectorized_map(). Note example debug implementation handle edge cases op_vectorized_map(), particular, f returns structure multiple tensors.","code":"op_vectorized_map_debug <- function(elements, fn) {    if (!is.list(elements)) {     # `elements` is a single tensor     batch_size <- op_shape(elements)[[1]]     out <- elements |>       op_split(batch_size) |>       lapply(fn) |>       op_stack()     return(out)   }    # `elements` is a list of tensors   batch_size <- elements[[1]] |> op_shape() |> _[[1]]   elements |>     lapply(\\(e) op_split(e, batch_size)) |>     zip_lists() |>     lapply(fn) |>     op_stack()  }"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_view_as_complex.html","id":null,"dir":"Reference","previous_headings":"","what":"Convert a real tensor with two channels into a complex tensor. — op_view_as_complex","title":"Convert a real tensor with two channels into a complex tensor. — op_view_as_complex","text":"Expects real-valued tensor whose last dimension size 2, holding real imaginary parts. Returns corresponding complex tensor last dimension removed.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_view_as_complex.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Convert a real tensor with two channels into a complex tensor. — op_view_as_complex","text":"","code":"op_view_as_complex(x)"},{"path":"https://keras3.posit.co/dev/reference/op_view_as_complex.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Convert a real tensor with two channels into a complex tensor. — op_view_as_complex","text":"x Real-valued tensor whose trailing dimension encodes complex components.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_view_as_complex.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Convert a real tensor with two channels into a complex tensor. — op_view_as_complex","text":"complex tensor shape op_shape(x)[-length(op_shape(x))].","code":""},{"path":"https://keras3.posit.co/dev/reference/op_view_as_complex.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Convert a real tensor with two channels into a complex tensor. — op_view_as_complex","text":"","code":"x <- op_array(matrix(c(1, 2, 3, 4), nrow = 2, byrow = TRUE)) op_view_as_complex(x) ## tf.Tensor([1.+2.j 3.+4.j], shape=(2), dtype=complex64)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_view_as_real.html","id":null,"dir":"Reference","previous_headings":"","what":"Convert a complex tensor into a stacked real representation. — op_view_as_real","title":"Convert a complex tensor into a stacked real representation. — op_view_as_real","text":"Produces real-valued tensor last dimension gathers real imaginary parts complex input.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_view_as_real.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Convert a complex tensor into a stacked real representation. — op_view_as_real","text":"","code":"op_view_as_real(x)"},{"path":"https://keras3.posit.co/dev/reference/op_view_as_real.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Convert a complex tensor into a stacked real representation. — op_view_as_real","text":"x Complex-valued tensor converted.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_view_as_real.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Convert a complex tensor into a stacked real representation. — op_view_as_real","text":"real tensor shape c(op_shape(x), 2) containing real imaginary parts.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_view_as_real.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Convert a complex tensor into a stacked real representation. — op_view_as_real","text":"","code":"x <- op_array(matrix(c(1, 2, 3, 4), nrow = 2, byrow = TRUE)) z <- op_view_as_complex(x) op_view_as_real(z) ## tf.Tensor( ## [[1. 2.] ##  [3. 4.]], shape=(2, 2), dtype=float32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_vstack.html","id":null,"dir":"Reference","previous_headings":"","what":"Stack tensors in sequence vertically (row wise). — op_vstack","title":"Stack tensors in sequence vertically (row wise). — op_vstack","text":"Stack tensors sequence vertically (row wise).","code":""},{"path":"https://keras3.posit.co/dev/reference/op_vstack.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Stack tensors in sequence vertically (row wise). — op_vstack","text":"","code":"op_vstack(xs, ...)"},{"path":"https://keras3.posit.co/dev/reference/op_vstack.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Stack tensors in sequence vertically (row wise). — op_vstack","text":"xs, ... Sequence tensors.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_vstack.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Stack tensors in sequence vertically (row wise). — op_vstack","text":"Tensor formed stacking given tensors.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_where.html","id":null,"dir":"Reference","previous_headings":"","what":"Return elements chosen from x1 or x2 depending on condition. — op_where","title":"Return elements chosen from x1 or x2 depending on condition. — op_where","text":"Return elements chosen x1 x2 depending condition.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_where.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Return elements chosen from x1 or x2 depending on condition. — op_where","text":"","code":"op_where(condition, x1 = NULL, x2 = NULL)"},{"path":"https://keras3.posit.co/dev/reference/op_where.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Return elements chosen from x1 or x2 depending on condition. — op_where","text":"condition TRUE, yield x1, otherwise yield x2. x1 Values choose condition TRUE. x2 Values choose condition FALSE.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_where.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Return elements chosen from x1 or x2 depending on condition. — op_where","text":"tensor elements x1 condition TRUE, elements x2 condition FALSE.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_while_loop.html","id":null,"dir":"Reference","previous_headings":"","what":"While loop implementation. — op_while_loop","title":"While loop implementation. — op_while_loop","text":"loop implementation.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_while_loop.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"While loop implementation. — op_while_loop","text":"","code":"op_while_loop(cond, body, loop_vars, maximum_iterations = NULL)"},{"path":"https://keras3.posit.co/dev/reference/op_while_loop.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"While loop implementation. — op_while_loop","text":"cond callable represents termination condition loop. Must accept loop_vars like structure argument. loop_vars tuple unnamed list, element loop_vars passed positionally callable. body callable represents loop body. Must accept loop_vars like structure argument, return update value structure. loop_vars tuple unnamed list, element loop_vars passed positionally callable. loop_vars arbitrary nested structure tensor state persist across loop iterations. maximum_iterations Optional maximum number iterations loop run. provided, cond output -ed additional condition ensuring number iterations executed greater maximum_iterations.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_while_loop.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"While loop implementation. — op_while_loop","text":"list tensors, shape dtype loop_vars.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_while_loop.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"While loop implementation. — op_while_loop","text":"","code":"i <- 0 loop_vars <- list(i)  # cond() must return a scalar bool cond <- function(i) i < 10L  # body must return same shape as loop_vars body <- function(i) list(i + 1L)  op_while_loop(cond, body, loop_vars) ## [[1]] ## tf.Tensor(10.0, shape=(), dtype=float32) x <- 0; y <- 1 cond <- \\(x, y) x < 10 body <- \\(x, y) list(x+1, y+1) op_while_loop(cond, body, list(x, y)) ## [[1]] ## tf.Tensor(10.0, shape=(), dtype=float32) ## ## [[2]] ## tf.Tensor(11.0, shape=(), dtype=float32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_zeros.html","id":null,"dir":"Reference","previous_headings":"","what":"Return a new tensor of given shape and type, filled with zeros. — op_zeros","title":"Return a new tensor of given shape and type, filled with zeros. — op_zeros","text":"Return new tensor given shape type, filled zeros.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_zeros.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Return a new tensor of given shape and type, filled with zeros. — op_zeros","text":"","code":"op_zeros(shape, dtype = NULL)"},{"path":"https://keras3.posit.co/dev/reference/op_zeros.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Return a new tensor of given shape and type, filled with zeros. — op_zeros","text":"shape Shape new tensor. dtype Desired data type tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_zeros.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Return a new tensor of given shape and type, filled with zeros. — op_zeros","text":"Tensor zeros given shape dtype.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_zeros_like.html","id":null,"dir":"Reference","previous_headings":"","what":"Return a tensor of zeros with the same shape and type as x. — op_zeros_like","title":"Return a tensor of zeros with the same shape and type as x. — op_zeros_like","text":"Return tensor zeros shape type x.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_zeros_like.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Return a tensor of zeros with the same shape and type as x. — op_zeros_like","text":"","code":"op_zeros_like(x, dtype = NULL)"},{"path":"https://keras3.posit.co/dev/reference/op_zeros_like.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Return a tensor of zeros with the same shape and type as x. — op_zeros_like","text":"x Input tensor. dtype Overrides data type result.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_zeros_like.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Return a tensor of zeros with the same shape and type as x. — op_zeros_like","text":"tensor zeros shape type x.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/optimizer_adadelta.html","id":null,"dir":"Reference","previous_headings":"","what":"Optimizer that implements the Adadelta algorithm. — optimizer_adadelta","title":"Optimizer that implements the Adadelta algorithm. — optimizer_adadelta","text":"Adadelta optimization stochastic gradient descent method based adaptive learning rate per dimension address two drawbacks: continual decay learning rates throughout training. need manually selected global learning rate. Adadelta robust extension Adagrad adapts learning rates based moving window gradient updates, instead accumulating past gradients. way, Adadelta continues learning even many updates done. Compared Adagrad, original version Adadelta set initial learning rate. version, initial learning rate can set, Keras optimizers.","code":""},{"path":"https://keras3.posit.co/dev/reference/optimizer_adadelta.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Optimizer that implements the Adadelta algorithm. — optimizer_adadelta","text":"","code":"optimizer_adadelta(   learning_rate = 0.001,   rho = 0.95,   epsilon = 1e-07,   weight_decay = NULL,   clipnorm = NULL,   clipvalue = NULL,   global_clipnorm = NULL,   use_ema = FALSE,   ema_momentum = 0.99,   ema_overwrite_frequency = NULL,   name = \"adadelta\",   ...,   loss_scale_factor = NULL,   gradient_accumulation_steps = NULL )"},{"path":"https://keras3.posit.co/dev/reference/optimizer_adadelta.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Optimizer that implements the Adadelta algorithm. — optimizer_adadelta","text":"learning_rate float, [LearningRateSchedule()] instance, callable takes arguments returns actual value use. learning rate. Defaults 0.001. Note Adadelta tends benefit higher initial learning rate values compared optimizers. match exact form original paper, use 1.0. rho floating point value. decay rate. Defaults 0.95. epsilon Small floating point value maintaining numerical stability. weight_decay Float. set, weight decay applied. clipnorm Float. set, gradient weight individually clipped norm higher value. clipvalue Float. set, gradient weight clipped higher value. global_clipnorm Float. set, gradient weights clipped global norm higher value. use_ema Boolean, defaults FALSE. TRUE, exponential moving average (EMA) applied. EMA consists computing exponential moving average weights model (weight values change training batch), periodically overwriting weights moving average. ema_momentum Float, defaults 0.99. used use_ema = TRUE. momentum use computing EMA model's weights: new_average = ema_momentum * old_average + (1 - ema_momentum) * current_variable_value. ema_overwrite_frequency Int NULL, defaults NULL. used use_ema = TRUE. Every ema_overwrite_frequency steps iterations, overwrite model variable moving average. NULL, optimizer overwrite model variables middle training, need explicitly overwrite variables end training calling optimizer$finalize_variable_values() (updates model variables -place). using built-fit() training loop, happens automatically last epoch, need anything. name String. name use momentum accumulator weights created optimizer. ... forward/backward compatability. loss_scale_factor Float NULL. float, scale factor multiplied loss computing gradients, inverse scale factor multiplied gradients updating variables. Useful preventing underflow mixed precision training. Alternately, optimizer_loss_scale() automatically set loss scale factor. gradient_accumulation_steps Int NULL. int, model optimizer variables updated every step; instead updated every gradient_accumulation_steps steps, using average value gradients since last update. known \"gradient accumulation\". can useful batch size small, order reduce gradient noise update step. EMA frequency look \"accumulated\" iterations value (optimizer steps // gradient_accumulation_steps). Learning rate schedules look \"real\" iterations value (optimizer steps).","code":""},{"path":"https://keras3.posit.co/dev/reference/optimizer_adadelta.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Optimizer that implements the Adadelta algorithm. — optimizer_adadelta","text":"Optimizer instance","code":""},{"path":"https://keras3.posit.co/dev/reference/optimizer_adadelta.html","id":"reference","dir":"Reference","previous_headings":"","what":"Reference","title":"Optimizer that implements the Adadelta algorithm. — optimizer_adadelta","text":"Zeiler, 2012","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/optimizer_adafactor.html","id":null,"dir":"Reference","previous_headings":"","what":"Optimizer that implements the Adafactor algorithm. — optimizer_adafactor","title":"Optimizer that implements the Adafactor algorithm. — optimizer_adafactor","text":"Adafactor commonly used NLP tasks, advantage taking less memory saves partial information previous gradients. default argument setup based original paper (see reference). gradients dimension > 2, Adafactor optimizer delete last 2 dimensions separately accumulator variables.","code":""},{"path":"https://keras3.posit.co/dev/reference/optimizer_adafactor.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Optimizer that implements the Adafactor algorithm. — optimizer_adafactor","text":"","code":"optimizer_adafactor(   learning_rate = 0.001,   beta_2_decay = -0.8,   epsilon_1 = 1e-30,   epsilon_2 = 0.001,   clip_threshold = 1,   relative_step = TRUE,   weight_decay = NULL,   clipnorm = NULL,   clipvalue = NULL,   global_clipnorm = NULL,   use_ema = FALSE,   ema_momentum = 0.99,   ema_overwrite_frequency = NULL,   name = \"adafactor\",   ...,   loss_scale_factor = NULL,   gradient_accumulation_steps = NULL )"},{"path":"https://keras3.posit.co/dev/reference/optimizer_adafactor.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Optimizer that implements the Adafactor algorithm. — optimizer_adafactor","text":"learning_rate float, LearningRateSchedule() instance, callable takes arguments returns actual value use. learning rate. Defaults 0.001. beta_2_decay float, defaults -0.8. decay rate beta_2. epsilon_1 float, defaults 1e-30. small offset keep denominator away 0. epsilon_2 float, defaults 1e-3. small offset avoid learning rate becoming small time. clip_threshold float, defaults 1.0. Clipping threshold. part Adafactor algorithm, independent clipnorm, clipvalue, global_clipnorm. relative_step bool, defaults TRUE. learning_rate constant relative_step=TRUE, learning rate adjusted based current iterations. default learning rate decay Adafactor. weight_decay Float. set, weight decay applied. clipnorm Float. set, gradient weight individually clipped norm higher value. clipvalue Float. set, gradient weight clipped higher value. global_clipnorm Float. set, gradient weights clipped global norm higher value. use_ema Boolean, defaults FALSE. TRUE, exponential moving average (EMA) applied. EMA consists computing exponential moving average weights model (weight values change training batch), periodically overwriting weights moving average. ema_momentum Float, defaults 0.99. used use_ema = TRUE. momentum use computing EMA model's weights: new_average = ema_momentum * old_average + (1 - ema_momentum) * current_variable_value. ema_overwrite_frequency Int NULL, defaults NULL. used use_ema=TRUE. Every ema_overwrite_frequency steps iterations, overwrite model variable moving average. NULL, optimizer overwrite model variables middle training, need explicitly overwrite variables end training calling optimizer$finalize_variable_values() (updates model variables -place). using built-fit() training loop, happens automatically last epoch, need anything. name String. name use momentum accumulator weights created optimizer. ... forward/backward compatability. loss_scale_factor Float NULL. float, scale factor multiplied loss computing gradients, inverse scale factor multiplied gradients updating variables. Useful preventing underflow mixed precision training. Alternately, optimizer_loss_scale() automatically set loss scale factor. gradient_accumulation_steps Int NULL. int, model optimizer variables updated every step; instead updated every gradient_accumulation_steps steps, using average value gradients since last update. known \"gradient accumulation\". can useful batch size small, order reduce gradient noise update step. EMA frequency look \"accumulated\" iterations value (optimizer steps // gradient_accumulation_steps). Learning rate schedules look \"real\" iterations value (optimizer steps).","code":""},{"path":"https://keras3.posit.co/dev/reference/optimizer_adafactor.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Optimizer that implements the Adafactor algorithm. — optimizer_adafactor","text":"Optimizer instance","code":""},{"path":"https://keras3.posit.co/dev/reference/optimizer_adafactor.html","id":"reference","dir":"Reference","previous_headings":"","what":"Reference","title":"Optimizer that implements the Adafactor algorithm. — optimizer_adafactor","text":"Shazeer, Noam et al., 2018.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/optimizer_adagrad.html","id":null,"dir":"Reference","previous_headings":"","what":"Optimizer that implements the Adagrad algorithm. — optimizer_adagrad","title":"Optimizer that implements the Adagrad algorithm. — optimizer_adagrad","text":"Adagrad optimizer parameter-specific learning rates, adapted relative frequently parameter gets updated training. updates parameter receives, smaller updates.","code":""},{"path":"https://keras3.posit.co/dev/reference/optimizer_adagrad.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Optimizer that implements the Adagrad algorithm. — optimizer_adagrad","text":"","code":"optimizer_adagrad(   learning_rate = 0.001,   initial_accumulator_value = 0.1,   epsilon = 1e-07,   weight_decay = NULL,   clipnorm = NULL,   clipvalue = NULL,   global_clipnorm = NULL,   use_ema = FALSE,   ema_momentum = 0.99,   ema_overwrite_frequency = NULL,   name = \"adagrad\",   ...,   loss_scale_factor = NULL,   gradient_accumulation_steps = NULL )"},{"path":"https://keras3.posit.co/dev/reference/optimizer_adagrad.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Optimizer that implements the Adagrad algorithm. — optimizer_adagrad","text":"learning_rate float, LearningRateSchedule() instance, callable takes arguments returns actual value use. learning rate. Defaults 0.001. Note Adagrad tends benefit higher initial learning rate values compared optimizers. match exact form original paper, use 1.0. initial_accumulator_value Floating point value. Starting value accumulators (per-parameter momentum values). Must non-negative. epsilon Small floating point value maintaining numerical stability. weight_decay Float. set, weight decay applied. clipnorm Float. set, gradient weight individually clipped norm higher value. clipvalue Float. set, gradient weight clipped higher value. global_clipnorm Float. set, gradient weights clipped global norm higher value. use_ema Boolean, defaults FALSE. TRUE, exponential moving average (EMA) applied. EMA consists computing exponential moving average weights model (weight values change training batch), periodically overwriting weights moving average. ema_momentum Float, defaults 0.99. used use_ema=TRUE. momentum use computing EMA model's weights: new_average = ema_momentum * old_average + (1 - ema_momentum) * current_variable_value. ema_overwrite_frequency Int NULL, defaults NULL. used use_ema=TRUE. Every ema_overwrite_frequency steps iterations, overwrite model variable moving average. NULL, optimizer overwrite model variables middle training, need explicitly overwrite variables end training calling optimizer$finalize_variable_values() (updates model variables -place). using built-fit() training loop, happens automatically last epoch, need anything. name String. name use momentum accumulator weights created optimizer. ... forward/backward compatability. loss_scale_factor Float NULL. float, scale factor multiplied loss computing gradients, inverse scale factor multiplied gradients updating variables. Useful preventing underflow mixed precision training. Alternately, optimizer_loss_scale() automatically set loss scale factor. gradient_accumulation_steps Int NULL. int, model optimizer variables updated every step; instead updated every gradient_accumulation_steps steps, using average value gradients since last update. known \"gradient accumulation\". can useful batch size small, order reduce gradient noise update step. EMA frequency look \"accumulated\" iterations value (optimizer steps // gradient_accumulation_steps). Learning rate schedules look \"real\" iterations value (optimizer steps).","code":""},{"path":"https://keras3.posit.co/dev/reference/optimizer_adagrad.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Optimizer that implements the Adagrad algorithm. — optimizer_adagrad","text":"Optimizer instance","code":""},{"path":"https://keras3.posit.co/dev/reference/optimizer_adagrad.html","id":"reference","dir":"Reference","previous_headings":"","what":"Reference","title":"Optimizer that implements the Adagrad algorithm. — optimizer_adagrad","text":"Duchi et al., 2011.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/optimizer_adam.html","id":null,"dir":"Reference","previous_headings":"","what":"Optimizer that implements the Adam algorithm. — optimizer_adam","title":"Optimizer that implements the Adam algorithm. — optimizer_adam","text":"Adam optimization stochastic gradient descent method based adaptive estimation first-order second-order moments. According Kingma et al., 2014, method \"computationally efficient, little memory requirement, invariant diagonal rescaling gradients, well suited problems large terms data/parameters\".","code":""},{"path":"https://keras3.posit.co/dev/reference/optimizer_adam.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Optimizer that implements the Adam algorithm. — optimizer_adam","text":"","code":"optimizer_adam(   learning_rate = 0.001,   beta_1 = 0.9,   beta_2 = 0.999,   epsilon = 1e-07,   amsgrad = FALSE,   weight_decay = NULL,   clipnorm = NULL,   clipvalue = NULL,   global_clipnorm = NULL,   use_ema = FALSE,   ema_momentum = 0.99,   ema_overwrite_frequency = NULL,   name = \"adam\",   ...,   loss_scale_factor = NULL,   gradient_accumulation_steps = NULL )"},{"path":"https://keras3.posit.co/dev/reference/optimizer_adam.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Optimizer that implements the Adam algorithm. — optimizer_adam","text":"learning_rate float, LearningRateSchedule() instance, callable takes arguments returns actual value use. learning rate. Defaults 0.001. beta_1 float value constant float tensor, callable takes arguments returns actual value use. exponential decay rate 1st moment estimates. Defaults 0.9. beta_2 float value constant float tensor, callable takes arguments returns actual value use. exponential decay rate 2nd moment estimates. Defaults 0.999. epsilon small constant numerical stability. epsilon \"epsilon hat\" Kingma Ba paper (formula just Section 2.1), epsilon Algorithm 1 paper. Defaults 1e-7. amsgrad Boolean. Whether apply AMSGrad variant algorithm paper \"Convergence Adam beyond\". Defaults FALSE. weight_decay Float. set, weight decay applied. clipnorm Float. set, gradient weight individually clipped norm higher value. clipvalue Float. set, gradient weight clipped higher value. global_clipnorm Float. set, gradient weights clipped global norm higher value. use_ema Boolean, defaults FALSE. TRUE, exponential moving average (EMA) applied. EMA consists computing exponential moving average weights model (weight values change training batch), periodically overwriting weights moving average. ema_momentum Float, defaults 0.99. used use_ema=TRUE. momentum use computing EMA model's weights: new_average = ema_momentum * old_average + (1 - ema_momentum) * current_variable_value. ema_overwrite_frequency Int NULL, defaults NULL. used use_ema=TRUE. Every ema_overwrite_frequency steps iterations, overwrite model variable moving average. NULL, optimizer overwrite model variables middle training, need explicitly overwrite variables end training calling optimizer$finalize_variable_values() (updates model variables -place). using built-fit() training loop, happens automatically last epoch, need anything. name String. name use momentum accumulator weights created optimizer. ... forward/backward compatability. loss_scale_factor Float NULL. float, scale factor multiplied loss computing gradients, inverse scale factor multiplied gradients updating variables. Useful preventing underflow mixed precision training. Alternately, optimizer_loss_scale() automatically set loss scale factor. gradient_accumulation_steps Int NULL. int, model optimizer variables updated every step; instead updated every gradient_accumulation_steps steps, using average value gradients since last update. known \"gradient accumulation\". can useful batch size small, order reduce gradient noise update step. EMA frequency look \"accumulated\" iterations value (optimizer steps // gradient_accumulation_steps). Learning rate schedules look \"real\" iterations value (optimizer steps).","code":""},{"path":"https://keras3.posit.co/dev/reference/optimizer_adam.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Optimizer that implements the Adam algorithm. — optimizer_adam","text":"Optimizer instance","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/optimizer_adam_w.html","id":null,"dir":"Reference","previous_headings":"","what":"Optimizer that implements the AdamW algorithm. — optimizer_adam_w","title":"Optimizer that implements the AdamW algorithm. — optimizer_adam_w","text":"AdamW optimization stochastic gradient descent method based adaptive estimation first-order second-order moments added method decay weights per techniques discussed paper, 'Decoupled Weight Decay Regularization' Loshchilov, Hutter et al., 2019. According Kingma et al., 2014, underlying Adam method \"computationally efficient, little memory requirement, invariant diagonal rescaling gradients, well suited problems large terms data/parameters\".","code":""},{"path":"https://keras3.posit.co/dev/reference/optimizer_adam_w.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Optimizer that implements the AdamW algorithm. — optimizer_adam_w","text":"","code":"optimizer_adam_w(   learning_rate = 0.001,   weight_decay = 0.004,   beta_1 = 0.9,   beta_2 = 0.999,   epsilon = 1e-07,   amsgrad = FALSE,   clipnorm = NULL,   clipvalue = NULL,   global_clipnorm = NULL,   use_ema = FALSE,   ema_momentum = 0.99,   ema_overwrite_frequency = NULL,   name = \"adamw\",   ...,   loss_scale_factor = NULL,   gradient_accumulation_steps = NULL )"},{"path":"https://keras3.posit.co/dev/reference/optimizer_adam_w.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Optimizer that implements the AdamW algorithm. — optimizer_adam_w","text":"learning_rate float, LearningRateSchedule() instance, callable takes arguments returns actual value use. learning rate. Defaults 0.001. weight_decay Float. set, weight decay applied. beta_1 float value constant float tensor, callable takes arguments returns actual value use. exponential decay rate 1st moment estimates. Defaults 0.9. beta_2 float value constant float tensor, callable takes arguments returns actual value use. exponential decay rate 2nd moment estimates. Defaults 0.999. epsilon small constant numerical stability. epsilon \"epsilon hat\" Kingma Ba paper (formula just Section 2.1), epsilon Algorithm 1 paper. Defaults 1e-7. amsgrad Boolean. Whether apply AMSGrad variant algorithm paper \"Convergence Adam beyond\". Defaults FALSE. clipnorm Float. set, gradient weight individually clipped norm higher value. clipvalue Float. set, gradient weight clipped higher value. global_clipnorm Float. set, gradient weights clipped global norm higher value. use_ema Boolean, defaults FALSE. TRUE, exponential moving average (EMA) applied. EMA consists computing exponential moving average weights model (weight values change training batch), periodically overwriting weights moving average. ema_momentum Float, defaults 0.99. used use_ema=TRUE. momentum use computing EMA model's weights: new_average = ema_momentum * old_average + (1 - ema_momentum) * current_variable_value. ema_overwrite_frequency Int NULL, defaults NULL. used use_ema=TRUE. Every ema_overwrite_frequency steps iterations, overwrite model variable moving average. NULL, optimizer overwrite model variables middle training, need explicitly overwrite variables end training calling optimizer$finalize_variable_values() (updates model variables -place). using built-fit() training loop, happens automatically last epoch, need anything. name String. name use momentum accumulator weights created optimizer. ... forward/backward compatability. loss_scale_factor Float NULL. float, scale factor multiplied loss computing gradients, inverse scale factor multiplied gradients updating variables. Useful preventing underflow mixed precision training. Alternately, optimizer_loss_scale() automatically set loss scale factor. gradient_accumulation_steps Int NULL. int, model optimizer variables updated every step; instead updated every gradient_accumulation_steps steps, using average value gradients since last update. known \"gradient accumulation\". can useful batch size small, order reduce gradient noise update step. EMA frequency look \"accumulated\" iterations value (optimizer steps // gradient_accumulation_steps). Learning rate schedules look \"real\" iterations value (optimizer steps).","code":""},{"path":"https://keras3.posit.co/dev/reference/optimizer_adam_w.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Optimizer that implements the AdamW algorithm. — optimizer_adam_w","text":"Optimizer instance","code":""},{"path":"https://keras3.posit.co/dev/reference/optimizer_adam_w.html","id":"references","dir":"Reference","previous_headings":"","what":"References","title":"Optimizer that implements the AdamW algorithm. — optimizer_adam_w","text":"Loshchilov et al., 2019 Kingma et al., 2014 adam Reddi et al., 2018 amsgrad.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/optimizer_adamax.html","id":null,"dir":"Reference","previous_headings":"","what":"Optimizer that implements the Adamax algorithm. — optimizer_adamax","title":"Optimizer that implements the Adamax algorithm. — optimizer_adamax","text":"Adamax, variant Adam based infinity norm, first-order gradient-based optimization method. Due capability adjusting learning rate based data characteristics, suited learn time-variant process, e.g., speech data dynamically changed noise conditions. Default parameters follow provided paper (see references ). Initialization:   update rule parameter w gradient g described end section 7.1 paper (see reference section):","code":"m <- 0  # Initialize initial 1st moment vector u <- 0  # Initialize the exponentially weighted infinity norm t <- 0  # Initialize timestep t <-  t + 1 m <- beta1 * m + (1 - beta) * g u <- max(beta2 * u, abs(g)) current_lr <- learning_rate / (1 - beta1 ** t) w <- w - current_lr * m / (u + epsilon)"},{"path":"https://keras3.posit.co/dev/reference/optimizer_adamax.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Optimizer that implements the Adamax algorithm. — optimizer_adamax","text":"","code":"optimizer_adamax(   learning_rate = 0.001,   beta_1 = 0.9,   beta_2 = 0.999,   epsilon = 1e-07,   weight_decay = NULL,   clipnorm = NULL,   clipvalue = NULL,   global_clipnorm = NULL,   use_ema = FALSE,   ema_momentum = 0.99,   ema_overwrite_frequency = NULL,   name = \"adamax\",   ...,   loss_scale_factor = NULL,   gradient_accumulation_steps = NULL )"},{"path":"https://keras3.posit.co/dev/reference/optimizer_adamax.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Optimizer that implements the Adamax algorithm. — optimizer_adamax","text":"learning_rate float, LearningRateSchedule() instance, callable takes arguments returns actual value use. learning rate. Defaults 0.001. beta_1 float value constant float tensor. exponential decay rate 1st moment estimates. beta_2 float value constant float tensor. exponential decay rate exponentially weighted infinity norm. epsilon small constant numerical stability. name: String. name use momentum accumulator weights created optimizer. weight_decay Float. set, weight decay applied. clipnorm Float. set, gradient weight individually clipped norm higher value. clipvalue Float. set, gradient weight clipped higher value. global_clipnorm Float. set, gradient weights clipped global norm higher value. use_ema Boolean, defaults FALSE. TRUE, exponential moving average (EMA) applied. EMA consists computing exponential moving average weights model (weight values change training batch), periodically overwriting weights moving average. ema_momentum Float, defaults 0.99. used use_ema=TRUE. momentum use computing EMA model's weights: new_average = ema_momentum * old_average + (1 - ema_momentum) * current_variable_value. ema_overwrite_frequency Int NULL, defaults NULL. used use_ema=TRUE. Every ema_overwrite_frequency steps iterations, overwrite model variable moving average. NULL, optimizer overwrite model variables middle training, need explicitly overwrite variables end training calling optimizer$finalize_variable_values() (updates model variables -place). using built-fit() training loop, happens automatically last epoch, need anything. name String, name object ... forward/backward compatability. loss_scale_factor Float NULL. float, scale factor multiplied loss computing gradients, inverse scale factor multiplied gradients updating variables. Useful preventing underflow mixed precision training. Alternately, optimizer_loss_scale() automatically set loss scale factor. gradient_accumulation_steps Int NULL. int, model optimizer variables updated every step; instead updated every gradient_accumulation_steps steps, using average value gradients since last update. known \"gradient accumulation\". can useful batch size small, order reduce gradient noise update step. EMA frequency look \"accumulated\" iterations value (optimizer steps // gradient_accumulation_steps). Learning rate schedules look \"real\" iterations value (optimizer steps).","code":""},{"path":"https://keras3.posit.co/dev/reference/optimizer_adamax.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Optimizer that implements the Adamax algorithm. — optimizer_adamax","text":"Optimizer instance","code":""},{"path":"https://keras3.posit.co/dev/reference/optimizer_adamax.html","id":"reference","dir":"Reference","previous_headings":"","what":"Reference","title":"Optimizer that implements the Adamax algorithm. — optimizer_adamax","text":"Kingma et al., 2014","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/optimizer_ftrl.html","id":null,"dir":"Reference","previous_headings":"","what":"Optimizer that implements the FTRL algorithm. — optimizer_ftrl","title":"Optimizer that implements the FTRL algorithm. — optimizer_ftrl","text":"\"Follow Regularized Leader\" (FTRL) optimization algorithm developed Google click-rate prediction early 2010s. suitable shallow models large sparse feature spaces. algorithm described McMahan et al., 2013. Keras version support online L2 regularization (L2 regularization described paper ) shrinkage-type L2 regularization (addition L2 penalty loss function). Initialization:   Update rule one variable w:   Notation: lr learning rate g gradient variable lambda_1 L1 regularization strength lambda_2 L2 regularization strength lr_power power scale n. Check documentation l2_shrinkage_regularization_strength parameter details shrinkage enabled, case gradient replaced gradient shrinkage.","code":"n <- 0 sigma <- 0 z <- 0 prev_n <- n n <- n + g^2 sigma <- (n^(-lr_power) - prev_n^(-lr_power)) / lr z <- z + g - sigma * w if (abs(z) < lambda_1) {   w <- 0 } else {   w <- (sgn(z) * lambda_1 - z) / ((beta + sqrt(n)) / alpha + lambda_2) }"},{"path":"https://keras3.posit.co/dev/reference/optimizer_ftrl.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Optimizer that implements the FTRL algorithm. — optimizer_ftrl","text":"","code":"optimizer_ftrl(   learning_rate = 0.001,   learning_rate_power = -0.5,   initial_accumulator_value = 0.1,   l1_regularization_strength = 0,   l2_regularization_strength = 0,   l2_shrinkage_regularization_strength = 0,   beta = 0,   weight_decay = NULL,   clipnorm = NULL,   clipvalue = NULL,   global_clipnorm = NULL,   use_ema = FALSE,   ema_momentum = 0.99,   ema_overwrite_frequency = NULL,   name = \"ftrl\",   ...,   loss_scale_factor = NULL,   gradient_accumulation_steps = NULL )"},{"path":"https://keras3.posit.co/dev/reference/optimizer_ftrl.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Optimizer that implements the FTRL algorithm. — optimizer_ftrl","text":"learning_rate float, LearningRateSchedule() instance, callable takes arguments returns actual value use. learning rate. Defaults 0.001. learning_rate_power float value, must less equal zero. Controls learning rate decreases training. Use zero fixed learning rate. initial_accumulator_value starting value accumulators. zero positive values allowed. l1_regularization_strength float value, must greater equal zero. Defaults 0.0. l2_regularization_strength float value, must greater equal zero. Defaults 0.0. l2_shrinkage_regularization_strength float value, must greater equal zero. differs L2 L2 stabilization penalty, whereas L2 shrinkage magnitude penalty. input sparse shrinkage happen active weights. beta float value, representing beta value paper. Defaults 0.0. weight_decay Float. set, weight decay applied. clipnorm Float. set, gradient weight individually clipped norm higher value. clipvalue Float. set, gradient weight clipped higher value. global_clipnorm Float. set, gradient weights clipped global norm higher value. use_ema Boolean, defaults FALSE. TRUE, exponential moving average (EMA) applied. EMA consists computing exponential moving average weights model (weight values change training batch), periodically overwriting weights moving average. ema_momentum Float, defaults 0.99. used use_ema=TRUE. momentum use computing EMA model's weights: new_average = ema_momentum * old_average + (1 - ema_momentum) * current_variable_value. ema_overwrite_frequency Int NULL, defaults NULL. used use_ema=TRUE. Every ema_overwrite_frequency steps iterations, overwrite model variable moving average. NULL, optimizer overwrite model variables middle training, need explicitly overwrite variables end training calling optimizer$finalize_variable_values() (updates model variables -place). using built-fit() training loop, happens automatically last epoch, need anything. name String. name use momentum accumulator weights created optimizer. ... forward/backward compatability. loss_scale_factor Float NULL. float, scale factor multiplied loss computing gradients, inverse scale factor multiplied gradients updating variables. Useful preventing underflow mixed precision training. Alternately, optimizer_loss_scale automatically set loss scale factor. gradient_accumulation_steps Int NULL. int, model optimizer variables updated every step; instead updated every gradient_accumulation_steps steps, using average value gradients since last update. known \"gradient accumulation\". can useful batch size small, order reduce gradient noise update step. EMA frequency look \"accumulated\" iterations value (optimizer steps // gradient_accumulation_steps). Learning rate schedules look \"real\" iterations value (optimizer steps).","code":""},{"path":"https://keras3.posit.co/dev/reference/optimizer_ftrl.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Optimizer that implements the FTRL algorithm. — optimizer_ftrl","text":"Optimizer instance","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/optimizer_lamb.html","id":null,"dir":"Reference","previous_headings":"","what":"Optimizer that implements the Lamb algorithm. — optimizer_lamb","title":"Optimizer that implements the Lamb algorithm. — optimizer_lamb","text":"Lamb stochastic gradient descent method uses layer-wise adaptive moments adjusts learning rate parameter based ratio norm weight norm gradient helps stabilize training process improves convergence especially large batch sizes.","code":""},{"path":"https://keras3.posit.co/dev/reference/optimizer_lamb.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Optimizer that implements the Lamb algorithm. — optimizer_lamb","text":"","code":"optimizer_lamb(   learning_rate = 0.001,   beta_1 = 0.9,   beta_2 = 0.999,   epsilon = 1e-07,   weight_decay = NULL,   clipnorm = NULL,   clipvalue = NULL,   global_clipnorm = NULL,   use_ema = FALSE,   ema_momentum = 0.99,   ema_overwrite_frequency = NULL,   loss_scale_factor = NULL,   gradient_accumulation_steps = NULL,   name = \"lamb\",   ... )"},{"path":"https://keras3.posit.co/dev/reference/optimizer_lamb.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Optimizer that implements the Lamb algorithm. — optimizer_lamb","text":"learning_rate float, LearningRateSchedule() instance, callable takes arguments returns actual value use. learning rate. Defaults 0.001. beta_1 float value constant float tensor, callable takes arguments returns actual value use. exponential decay rate 1st moment estimates. Defaults 0.9. beta_2 float value constant float tensor, callable takes arguments returns actual value use. exponential decay rate 2nd moment estimates. Defaults 0.999. epsilon small constant numerical stability. Defaults 1e-7. weight_decay Float. set, weight decay applied. clipnorm Float. set, gradient weight individually clipped norm higher value. clipvalue Float. set, gradient weight clipped higher value. global_clipnorm Float. set, gradient weights clipped global norm higher value. use_ema Boolean, defaults FALSE. TRUE, exponential moving average (EMA) applied. EMA consists computing exponential moving average weights model (weight values change training batch), periodically overwriting weights moving average. ema_momentum Float, defaults 0.99. used use_ema = TRUE. momentum use computing EMA model's weights: new_average = ema_momentum * old_average + (1 - ema_momentum) * current_variable_value. ema_overwrite_frequency Int NULL, defaults NULL. used use_ema = TRUE. Every ema_overwrite_frequency steps iterations, overwrite model variable moving average. NULL, optimizer overwrite model variables middle training, need explicitly overwrite variables end training calling optimizer$finalize_variable_values() (updates model variables -place). using built-fit() training loop, happens automatically last epoch, need anything. loss_scale_factor Float NULL. float, scale factor multiplied loss computing gradients, inverse scale factor multiplied gradients updating variables. Useful preventing underflow mixed precision training. Alternately, optimizer_loss_scale() automatically set loss scale factor. gradient_accumulation_steps Int NULL. int, model optimizer variables updated every step; instead updated every gradient_accumulation_steps steps, using average value gradients since last update. known \"gradient accumulation\". can useful batch size small, order reduce gradient noise update step. EMA frequency look \"accumulated\" iterations value (optimizer steps // gradient_accumulation_steps). Learning rate schedules look \"real\" iterations value (optimizer steps). name String. name use momentum accumulator weights created optimizer. ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/optimizer_lamb.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Optimizer that implements the Lamb algorithm. — optimizer_lamb","text":"Optimizer instance","code":""},{"path":"https://keras3.posit.co/dev/reference/optimizer_lamb.html","id":"references","dir":"Reference","previous_headings":"","what":"References","title":"Optimizer that implements the Lamb algorithm. — optimizer_lamb","text":"Yang et al.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/optimizer_lion.html","id":null,"dir":"Reference","previous_headings":"","what":"Optimizer that implements the Lion algorithm. — optimizer_lion","title":"Optimizer that implements the Lion algorithm. — optimizer_lion","text":"Lion optimizer stochastic-gradient-descent method uses sign operator control magnitude update, unlike adaptive optimizers Adam rely second-order moments. makes Lion memory-efficient keeps track momentum. According authors (see reference), performance gain Adam grows batch size. update Lion produced sign operation, resulting larger norm, suitable learning rate Lion typically 3-10x smaller AdamW. weight decay Lion turn 3-10x larger AdamW maintain similar strength (lr * wd).","code":""},{"path":"https://keras3.posit.co/dev/reference/optimizer_lion.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Optimizer that implements the Lion algorithm. — optimizer_lion","text":"","code":"optimizer_lion(   learning_rate = 0.001,   beta_1 = 0.9,   beta_2 = 0.99,   weight_decay = NULL,   clipnorm = NULL,   clipvalue = NULL,   global_clipnorm = NULL,   use_ema = FALSE,   ema_momentum = 0.99,   ema_overwrite_frequency = NULL,   name = \"lion\",   ...,   loss_scale_factor = NULL,   gradient_accumulation_steps = NULL )"},{"path":"https://keras3.posit.co/dev/reference/optimizer_lion.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Optimizer that implements the Lion algorithm. — optimizer_lion","text":"learning_rate float, LearningRateSchedule() instance, callable takes arguments returns actual value use. learning rate. Defaults 0.001. beta_1 float value constant float tensor, callable takes arguments returns actual value use. rate combine current gradient 1st moment estimate. Defaults 0.9. beta_2 float value constant float tensor, callable takes arguments returns actual value use. exponential decay rate 1st moment estimate. Defaults 0.99. weight_decay Float. set, weight decay applied. clipnorm Float. set, gradient weight individually clipped norm higher value. clipvalue Float. set, gradient weight clipped higher value. global_clipnorm Float. set, gradient weights clipped global norm higher value. use_ema Boolean, defaults FALSE. TRUE, exponential moving average (EMA) applied. EMA consists computing exponential moving average weights model (weight values change training batch), periodically overwriting weights moving average. ema_momentum Float, defaults 0.99. used use_ema=TRUE. momentum use computing EMA model's weights: new_average = ema_momentum * old_average + (1 - ema_momentum) * current_variable_value. ema_overwrite_frequency Int NULL, defaults NULL. used use_ema=TRUE. Every ema_overwrite_frequency steps iterations, overwrite model variable moving average. NULL, optimizer overwrite model variables middle training, need explicitly overwrite variables end training calling optimizer$finalize_variable_values() (updates model variables -place). using built-fit() training loop, happens automatically last epoch, need anything. name String. name use momentum accumulator weights created optimizer. ... forward/backward compatability. loss_scale_factor Float NULL. float, scale factor multiplied loss computing gradients, inverse scale factor multiplied gradients updating variables. Useful preventing underflow mixed precision training. Alternately, optimizer_loss_scale() automatically set loss scale factor. gradient_accumulation_steps Int NULL. int, model optimizer variables updated every step; instead updated every gradient_accumulation_steps steps, using average value gradients since last update. known \"gradient accumulation\". can useful batch size small, order reduce gradient noise update step. EMA frequency look \"accumulated\" iterations value (optimizer steps // gradient_accumulation_steps). Learning rate schedules look \"real\" iterations value (optimizer steps).","code":""},{"path":"https://keras3.posit.co/dev/reference/optimizer_lion.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Optimizer that implements the Lion algorithm. — optimizer_lion","text":"Optimizer instance","code":""},{"path":"https://keras3.posit.co/dev/reference/optimizer_lion.html","id":"references","dir":"Reference","previous_headings":"","what":"References","title":"Optimizer that implements the Lion algorithm. — optimizer_lion","text":"Chen et al., 2023 Authors' implementation","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/optimizer_loss_scale.html","id":null,"dir":"Reference","previous_headings":"","what":"An optimizer that dynamically scales the loss to prevent underflow. — optimizer_loss_scale","title":"An optimizer that dynamically scales the loss to prevent underflow. — optimizer_loss_scale","text":"Loss scaling technique prevent numeric underflow intermediate gradients float16 used. prevent underflow, loss multiplied (\"scaled\") certain factor called \"loss scale\", causes intermediate gradients scaled loss scale well. final gradients divided (\"unscaled\") loss scale bring back original value. LossScaleOptimizer wraps another optimizer applies dynamic loss scaling . loss scale dynamically updated time follows: train step, nonfinite gradient encountered, loss scale halved, train step skipped. dynamic_growth_steps occurred since last time loss scale updated, nonfinite gradients occurred, loss scale doubled.","code":""},{"path":"https://keras3.posit.co/dev/reference/optimizer_loss_scale.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"An optimizer that dynamically scales the loss to prevent underflow. — optimizer_loss_scale","text":"","code":"optimizer_loss_scale(   inner_optimizer,   initial_scale = 32768,   dynamic_growth_steps = 2000L,   ...,   name = NULL,   weight_decay = NULL,   clipnorm = NULL,   clipvalue = NULL,   global_clipnorm = NULL,   use_ema = NULL,   ema_momentum = NULL,   ema_overwrite_frequency = NULL,   loss_scale_factor = NULL,   gradient_accumulation_steps = NULL )"},{"path":"https://keras3.posit.co/dev/reference/optimizer_loss_scale.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"An optimizer that dynamically scales the loss to prevent underflow. — optimizer_loss_scale","text":"inner_optimizer keras Optimizer instance wrap. initial_scale Float. initial loss scale. scale updated training. recommended high number, loss scale high gets lowered far quickly loss scale low gets raised. dynamic_growth_steps Int. often update scale upwards. every dynamic_growth_steps steps finite gradients, loss scale doubled. ... forward/backward compatability. name String. name use momentum accumulator weights created optimizer. weight_decay Float. set, weight decay applied. clipnorm Float. set, gradient weight individually clipped norm higher value. clipvalue Float. set, gradient weight clipped higher value. global_clipnorm Float. set, gradient weights clipped global norm higher value. use_ema Boolean, defaults FALSE. TRUE, exponential moving average (EMA) applied. EMA consists computing exponential moving average weights model (weight values change training batch), periodically overwriting weights moving average. ema_momentum Float, defaults 0.99. used use_ema=TRUE. momentum use computing EMA model's weights: new_average = ema_momentum * old_average + (1 - ema_momentum) * current_variable_value. ema_overwrite_frequency Int NULL, defaults NULL. used use_ema=TRUE. Every ema_overwrite_frequency steps iterations, overwrite model variable moving average. NULL, optimizer overwrite model variables middle training, need explicitly overwrite variables end training calling optimizer$finalize_variable_values() (updates model variables -place). using built-fit() training loop, happens automatically last epoch, need anything. loss_scale_factor Float NULL. float, scale factor multiplied loss computing gradients, inverse scale factor multiplied gradients updating variables. Useful preventing underflow mixed precision training. Alternately, optimizer_loss_scale() automatically set loss scale factor. gradient_accumulation_steps Int NULL. int, model optimizer variables updated every step; instead updated every gradient_accumulation_steps steps, using average value gradients since last update. known \"gradient accumulation\". can useful batch size small, order reduce gradient noise update step. EMA frequency look \"accumulated\" iterations value (optimizer steps // gradient_accumulation_steps). Learning rate schedules look \"real\" iterations value (optimizer steps).","code":""},{"path":"https://keras3.posit.co/dev/reference/optimizer_loss_scale.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"An optimizer that dynamically scales the loss to prevent underflow. — optimizer_loss_scale","text":"Optimizer instance","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/optimizer_muon.html","id":null,"dir":"Reference","previous_headings":"","what":"Optimizer that implements the Muon algorithm. — optimizer_muon","title":"Optimizer that implements the Muon algorithm. — optimizer_muon","text":"Note optimizer used following layers: Embedding layer Final output fully connected layer 0,1-D variables optimized using AdamW. Muon optimizer can use Muon update step AdamW update step based following: variable 2D, 3D 4D, AdamW step used. configurable. argument exclude_embeddings (defaults TRUE) set TRUE, AdamW step used. variable name matches expression listed argument exclude_layers (list), AdamW step used. variable uses Muon step. Typically, need pass name densely-connected output layer exclude_layers, e.g. exclude_layers = \"output_dense\".","code":""},{"path":"https://keras3.posit.co/dev/reference/optimizer_muon.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Optimizer that implements the Muon algorithm. — optimizer_muon","text":"","code":"optimizer_muon(   learning_rate = 0.001,   adam_beta_1 = 0.9,   adam_beta_2 = 0.999,   epsilon = 1e-07,   weight_decay = 0.1,   clipnorm = NULL,   clipvalue = NULL,   global_clipnorm = NULL,   use_ema = FALSE,   ema_momentum = 0.99,   ema_overwrite_frequency = NULL,   loss_scale_factor = NULL,   gradient_accumulation_steps = NULL,   name = \"muon\",   exclude_layers = NULL,   exclude_embeddings = TRUE,   muon_a = 3.4445,   muon_b = -4.775,   muon_c = 2.0315,   adam_lr_ratio = 0.1,   momentum = 0.95,   ns_steps = 6L,   nesterov = TRUE,   ... )"},{"path":"https://keras3.posit.co/dev/reference/optimizer_muon.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Optimizer that implements the Muon algorithm. — optimizer_muon","text":"learning_rate float, LearningRateSchedule() instance, callable takes arguments returns actual value use. learning rate. Defaults 0.001. adam_beta_1 float value constant float tensor, callable takes arguments returns actual value use. exponential decay rate 1st moment estimates. Defaults 0.9. adam_beta_2 float value constant float tensor, ora callable takes arguments returns actual value use. exponential decay rate 2nd moment estimates. Defaults 0.999. epsilon small constant numerical stability. \"epsilon hat\" Kingma Ba paper (formula just Section 2.1), epsilon Algorithm 1 paper. used AdamW. Defaults 1e-7. name String, name object exclude_layers List strings, keywords layer names exclude. layers keywords path use AdamW. exclude_embeddings Boolean value. TRUE, embedding layers use AdamW. muon_a Float, parameter muon algorithm. recommended use default value. muon_b Float, parameter b muon algorithm. recommended use default value. muon_c Float, parameter c muon algorithm. recommended use default value. adam_lr_ratio Float, ratio learning rate using Adam main learning rate. recommended set 0.1. momentum Float, momentum used internal SGD. ns_steps Integer, number Newton-Schulz iterations run. nesterov Boolean, whether use Nesterov-style momentum. ... forward/backward compatibility.","code":""},{"path":"https://keras3.posit.co/dev/reference/optimizer_muon.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Optimizer that implements the Muon algorithm. — optimizer_muon","text":"Optimizer instance","code":""},{"path":"https://keras3.posit.co/dev/reference/optimizer_muon.html","id":"references","dir":"Reference","previous_headings":"","what":"References","title":"Optimizer that implements the Muon algorithm. — optimizer_muon","text":"Original implementation Liu et al, 2025","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/optimizer_nadam.html","id":null,"dir":"Reference","previous_headings":"","what":"Optimizer that implements the Nadam algorithm. — optimizer_nadam","title":"Optimizer that implements the Nadam algorithm. — optimizer_nadam","text":"Much like Adam essentially RMSprop momentum, Nadam Adam Nesterov momentum.","code":""},{"path":"https://keras3.posit.co/dev/reference/optimizer_nadam.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Optimizer that implements the Nadam algorithm. — optimizer_nadam","text":"","code":"optimizer_nadam(   learning_rate = 0.001,   beta_1 = 0.9,   beta_2 = 0.999,   epsilon = 1e-07,   weight_decay = NULL,   clipnorm = NULL,   clipvalue = NULL,   global_clipnorm = NULL,   use_ema = FALSE,   ema_momentum = 0.99,   ema_overwrite_frequency = NULL,   name = \"nadam\",   ...,   loss_scale_factor = NULL,   gradient_accumulation_steps = NULL )"},{"path":"https://keras3.posit.co/dev/reference/optimizer_nadam.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Optimizer that implements the Nadam algorithm. — optimizer_nadam","text":"learning_rate float, LearningRateSchedule() instance, callable takes arguments returns actual value use. learning rate. Defaults 0.001. beta_1 float value constant float tensor, callable takes arguments returns actual value use. exponential decay rate 1st moment estimates. Defaults 0.9. beta_2 float value constant float tensor, callable takes arguments returns actual value use. exponential decay rate 2nd moment estimates. Defaults 0.999. epsilon small constant numerical stability. epsilon \"epsilon hat\" Kingma Ba paper (formula just Section 2.1), epsilon Algorithm 1 paper. Defaults 1e-7. weight_decay Float. set, weight decay applied. clipnorm Float. set, gradient weight individually clipped norm higher value. clipvalue Float. set, gradient weight clipped higher value. global_clipnorm Float. set, gradient weights clipped global norm higher value. use_ema Boolean, defaults FALSE. TRUE, exponential moving average (EMA) applied. EMA consists computing exponential moving average weights model (weight values change training batch), periodically overwriting weights moving average. ema_momentum Float, defaults 0.99. used use_ema=TRUE. momentum use computing EMA model's weights: new_average = ema_momentum * old_average + (1 - ema_momentum) * current_variable_value. ema_overwrite_frequency Int NULL, defaults NULL. used use_ema=TRUE. Every ema_overwrite_frequency steps iterations, overwrite model variable moving average. NULL, optimizer overwrite model variables middle training, need explicitly overwrite variables end training calling optimizer$finalize_variable_values() (updates model variables -place). using built-fit() training loop, happens automatically last epoch, need anything. name String. name use momentum accumulator weights created optimizer. ... forward/backward compatability. loss_scale_factor Float NULL. float, scale factor multiplied loss computing gradients, inverse scale factor multiplied gradients updating variables. Useful preventing underflow mixed precision training. Alternately, optimizer_loss_scale() automatically set loss scale factor. gradient_accumulation_steps Int NULL. int, model optimizer variables updated every step; instead updated every gradient_accumulation_steps steps, using average value gradients since last update. known \"gradient accumulation\". can useful batch size small, order reduce gradient noise update step. EMA frequency look \"accumulated\" iterations value (optimizer steps // gradient_accumulation_steps). Learning rate schedules look \"real\" iterations value (optimizer steps).","code":""},{"path":"https://keras3.posit.co/dev/reference/optimizer_nadam.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Optimizer that implements the Nadam algorithm. — optimizer_nadam","text":"Optimizer instance","code":""},{"path":"https://keras3.posit.co/dev/reference/optimizer_nadam.html","id":"reference","dir":"Reference","previous_headings":"","what":"Reference","title":"Optimizer that implements the Nadam algorithm. — optimizer_nadam","text":"Dozat, 2015.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/optimizer_rmsprop.html","id":null,"dir":"Reference","previous_headings":"","what":"Optimizer that implements the RMSprop algorithm. — optimizer_rmsprop","title":"Optimizer that implements the RMSprop algorithm. — optimizer_rmsprop","text":"gist RMSprop : Maintain moving (discounted) average square gradients Divide gradient root average implementation RMSprop uses plain momentum, Nesterov momentum. centered version additionally maintains moving average gradients, uses average estimate variance.","code":""},{"path":"https://keras3.posit.co/dev/reference/optimizer_rmsprop.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Optimizer that implements the RMSprop algorithm. — optimizer_rmsprop","text":"","code":"optimizer_rmsprop(   learning_rate = 0.001,   rho = 0.9,   momentum = 0,   epsilon = 1e-07,   centered = FALSE,   weight_decay = NULL,   clipnorm = NULL,   clipvalue = NULL,   global_clipnorm = NULL,   use_ema = FALSE,   ema_momentum = 0.99,   ema_overwrite_frequency = NULL,   name = \"rmsprop\",   ...,   loss_scale_factor = NULL,   gradient_accumulation_steps = NULL )"},{"path":"https://keras3.posit.co/dev/reference/optimizer_rmsprop.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Optimizer that implements the RMSprop algorithm. — optimizer_rmsprop","text":"learning_rate float, learning_rate_schedule_* instance, callable takes arguments returns actual value use. learning rate. Defaults 0.001. rho float, defaults 0.9. Discounting factor old gradients. momentum float, defaults 0.0. 0.0., optimizer tracks momentum value, decay rate equals 1 - momentum. epsilon small constant numerical stability. epsilon \"epsilon hat\" Kingma Ba paper (formula just Section 2.1), epsilon Algorithm 1 paper. Defaults 1e-7. centered Boolean. TRUE, gradients normalized estimated variance gradient; FALSE, uncentered second moment. Setting TRUE may help training, slightly expensive terms computation memory. Defaults FALSE. weight_decay Float. set, weight decay applied. clipnorm Float. set, gradient weight individually clipped norm higher value. clipvalue Float. set, gradient weight clipped higher value. global_clipnorm Float. set, gradient weights clipped global norm higher value. use_ema Boolean, defaults FALSE. TRUE, exponential moving average (EMA) applied. EMA consists computing exponential moving average weights model (weight values change training batch), periodically overwriting weights moving average. ema_momentum Float, defaults 0.99. used use_ema=TRUE. momentum use computing EMA model's weights: new_average = ema_momentum * old_average + (1 - ema_momentum) * current_variable_value. ema_overwrite_frequency Int NULL, defaults NULL. used use_ema=TRUE. Every ema_overwrite_frequency steps iterations, overwrite model variable moving average. NULL, optimizer overwrite model variables middle training, need explicitly overwrite variables end training calling optimizer$finalize_variable_values() (updates model variables -place). using built-fit() training loop, happens automatically last epoch, need anything. name String. name use momentum accumulator weights created optimizer. ... forward/backward compatability. loss_scale_factor Float NULL. float, scale factor multiplied loss computing gradients, inverse scale factor multiplied gradients updating variables. Useful preventing underflow mixed precision training. Alternately, optimizer_loss_scale() automatically set loss scale factor. gradient_accumulation_steps Int NULL. int, model optimizer variables updated every step; instead updated every gradient_accumulation_steps steps, using average value gradients since last update. known \"gradient accumulation\". can useful batch size small, order reduce gradient noise update step. EMA frequency look \"accumulated\" iterations value (optimizer steps // gradient_accumulation_steps). Learning rate schedules look \"real\" iterations value (optimizer steps).","code":""},{"path":"https://keras3.posit.co/dev/reference/optimizer_rmsprop.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Optimizer that implements the RMSprop algorithm. — optimizer_rmsprop","text":"Optimizer instance","code":""},{"path":"https://keras3.posit.co/dev/reference/optimizer_rmsprop.html","id":"usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Optimizer that implements the RMSprop algorithm. — optimizer_rmsprop","text":"","code":"opt <- optimizer_rmsprop(learning_rate=0.1)"},{"path":"https://keras3.posit.co/dev/reference/optimizer_rmsprop.html","id":"reference","dir":"Reference","previous_headings":"","what":"Reference","title":"Optimizer that implements the RMSprop algorithm. — optimizer_rmsprop","text":"Hinton, 2012","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/optimizer_sgd.html","id":null,"dir":"Reference","previous_headings":"","what":"Gradient descent (with momentum) optimizer. — optimizer_sgd","title":"Gradient descent (with momentum) optimizer. — optimizer_sgd","text":"Update rule parameter w gradient g momentum 0:   Update rule momentum larger 0:   nesterov=TRUE, rule becomes:","code":"w <- w - learning_rate * g velocity <- momentum * velocity - learning_rate * g w <- w + velocity velocity <- momentum * velocity - learning_rate * g w <- w + momentum * velocity - learning_rate * g"},{"path":"https://keras3.posit.co/dev/reference/optimizer_sgd.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Gradient descent (with momentum) optimizer. — optimizer_sgd","text":"","code":"optimizer_sgd(   learning_rate = 0.01,   momentum = 0,   nesterov = FALSE,   weight_decay = NULL,   clipnorm = NULL,   clipvalue = NULL,   global_clipnorm = NULL,   use_ema = FALSE,   ema_momentum = 0.99,   ema_overwrite_frequency = NULL,   name = \"SGD\",   ...,   loss_scale_factor = NULL,   gradient_accumulation_steps = NULL )"},{"path":"https://keras3.posit.co/dev/reference/optimizer_sgd.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Gradient descent (with momentum) optimizer. — optimizer_sgd","text":"learning_rate float, learning_rate_schedule_* instance, callable takes arguments returns actual value use. learning rate. Defaults 0.01. momentum float hyperparameter >= 0 accelerates gradient descent relevant direction dampens oscillations. 0 vanilla gradient descent. Defaults 0.0. nesterov boolean. Whether apply Nesterov momentum. Defaults FALSE. weight_decay Float. set, weight decay applied. clipnorm Float. set, gradient weight individually clipped norm higher value. clipvalue Float. set, gradient weight clipped higher value. global_clipnorm Float. set, gradient weights clipped global norm higher value. use_ema Boolean, defaults FALSE. TRUE, exponential moving average (EMA) applied. EMA consists computing exponential moving average weights model (weight values change training batch), periodically overwriting weights moving average. ema_momentum Float, defaults 0.99. used use_ema=TRUE. momentum use computing EMA model's weights: new_average = ema_momentum * old_average + (1 - ema_momentum) * current_variable_value. ema_overwrite_frequency Int NULL, defaults NULL. used use_ema=TRUE. Every ema_overwrite_frequency steps iterations, overwrite model variable moving average. NULL, optimizer overwrite model variables middle training, need explicitly overwrite variables end training calling optimizer$finalize_variable_values() (updates model variables -place). using built-fit() training loop, happens automatically last epoch, need anything. name String. name use momentum accumulator weights created optimizer. ... forward/backward compatability. loss_scale_factor Float NULL. float, scale factor multiplied loss computing gradients, inverse scale factor multiplied gradients updating variables. Useful preventing underflow mixed precision training. Alternately, optimizer_loss_scale() automatically set loss scale factor. gradient_accumulation_steps Int NULL. int, model optimizer variables updated every step; instead updated every gradient_accumulation_steps steps, using average value gradients since last update. known \"gradient accumulation\". can useful batch size small, order reduce gradient noise update step. EMA frequency look \"accumulated\" iterations value (optimizer steps // gradient_accumulation_steps). Learning rate schedules look \"real\" iterations value (optimizer steps).","code":""},{"path":"https://keras3.posit.co/dev/reference/optimizer_sgd.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Gradient descent (with momentum) optimizer. — optimizer_sgd","text":"Optimizer instance","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/pad_sequences.html","id":null,"dir":"Reference","previous_headings":"","what":"Pads sequences to the same length. — pad_sequences","title":"Pads sequences to the same length. — pad_sequences","text":"function transforms list (length num_samples) sequences (lists integers) 2D NumPy array shape (num_samples, num_timesteps). num_timesteps either maxlen argument provided, length longest sequence list. Sequences shorter num_timesteps padded value num_timesteps long. Sequences longer num_timesteps truncated fit desired length. position padding truncation happens determined arguments padding truncating, respectively. Pre-padding removing values beginning sequence default.","code":"sequence <- list(c(1), c(2, 3), c(4, 5, 6)) pad_sequences(sequence) ##      [,1] [,2] [,3] ## [1,]    0    0    1 ## [2,]    0    2    3 ## [3,]    4    5    6 pad_sequences(sequence, value=-1) ##      [,1] [,2] [,3] ## [1,]   -1   -1    1 ## [2,]   -1    2    3 ## [3,]    4    5    6 pad_sequences(sequence, padding='post') ##      [,1] [,2] [,3] ## [1,]    1    0    0 ## [2,]    2    3    0 ## [3,]    4    5    6 pad_sequences(sequence, maxlen=2) ##      [,1] [,2] ## [1,]    0    1 ## [2,]    2    3 ## [3,]    5    6"},{"path":"https://keras3.posit.co/dev/reference/pad_sequences.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Pads sequences to the same length. — pad_sequences","text":"","code":"pad_sequences(   sequences,   maxlen = NULL,   dtype = \"int32\",   padding = \"pre\",   truncating = \"pre\",   value = 0 )"},{"path":"https://keras3.posit.co/dev/reference/pad_sequences.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Pads sequences to the same length. — pad_sequences","text":"sequences List sequences (sequence list integers). maxlen Optional Int, maximum length sequences. provided, sequences padded length longest individual sequence. dtype (Optional, defaults \"int32\"). Type output sequences. pad sequences variable length strings, can use object. padding String, \"pre\" \"post\" (optional, defaults \"pre\"): pad either sequence. truncating String, \"pre\" \"post\" (optional, defaults \"pre\"): remove values sequences larger maxlen, either beginning end sequences. value Float String, padding value. (Optional, defaults 0)","code":""},{"path":"https://keras3.posit.co/dev/reference/pad_sequences.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Pads sequences to the same length. — pad_sequences","text":"Array shape (len(sequences), maxlen)","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/pipe.html","id":null,"dir":"Reference","previous_headings":"","what":"Pipe operator — %>%","title":"Pipe operator — %>%","text":"See %>% details.","code":""},{"path":"https://keras3.posit.co/dev/reference/pipe.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Pipe operator — %>%","text":"","code":"lhs %>% rhs"},{"path":"https://keras3.posit.co/dev/reference/pipe.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Pipe operator — %>%","text":"commonly, result calling right hand side left hand side argument: rhs(lhs). See magritter vignette , advanced, usages.","code":""},{"path":"https://keras3.posit.co/dev/reference/plot.keras.src.models.model.Model.html","id":null,"dir":"Reference","previous_headings":"","what":"Plot a Keras model — plot.keras.src.models.model.Model","title":"Plot a Keras model — plot.keras.src.models.model.Model","text":"Plot Keras model","code":""},{"path":"https://keras3.posit.co/dev/reference/plot.keras.src.models.model.Model.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Plot a Keras model — plot.keras.src.models.model.Model","text":"","code":"# S3 method for class 'keras.src.models.model.Model' plot(   x,   show_shapes = FALSE,   show_dtype = FALSE,   show_layer_names = FALSE,   ...,   rankdir = \"TB\",   expand_nested = FALSE,   dpi = 200,   layer_range = NULL,   show_layer_activations = FALSE,   show_trainable = NA,   to_file = NULL )"},{"path":"https://keras3.posit.co/dev/reference/plot.keras.src.models.model.Model.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Plot a Keras model — plot.keras.src.models.model.Model","text":"x Keras model instance show_shapes whether display shape information. show_dtype whether display layer dtypes. show_layer_names whether display layer names. ... passed Python keras.utils.model_to_dot(). Used forward backward compatibility. rankdir string specifying format plot: 'TB' creates vertical plot; 'LR' creates horizontal plot. (argument passed PyDot) expand_nested Whether expand nested models clusters. dpi Dots per inch. Increase value image text appears excessively pixelated. layer_range list containing two character strings, starting layer name ending layer name (inclusive) indicating range layers plot generated. also accepts regex patterns instead exact name. case, start predicate first element matches layer_range[1] end predicate last element matches layer_range[2]. default NULL considers layers model. Note must pass range resultant subgraph must complete. show_layer_activations Display layer activations (layers activation property). show_trainable whether display layer trainable. to_file File name plot image. NULL (default), model drawn default graphics device. Otherwise, file saved.","code":""},{"path":"https://keras3.posit.co/dev/reference/plot.keras.src.models.model.Model.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Plot a Keras model — plot.keras.src.models.model.Model","text":"Nothing, called side effects.","code":""},{"path":"https://keras3.posit.co/dev/reference/plot.keras.src.models.model.Model.html","id":"raises","dir":"Reference","previous_headings":"","what":"Raises","title":"Plot a Keras model — plot.keras.src.models.model.Model","text":"ValueError: plot(model) called model built, unless input_shape =  argument supplied keras_model_sequential().","code":""},{"path":"https://keras3.posit.co/dev/reference/plot.keras.src.models.model.Model.html","id":"requirements","dir":"Reference","previous_headings":"","what":"Requirements","title":"Plot a Keras model — plot.keras.src.models.model.Model","text":"function requires pydot graphviz. pydot default installed install_keras(), installed Keras means, can install pydot directly :   can install graphviz directly : https://graphviz.gitlab.io/download/ Linux platforms, can install graphviz via package manager. example, Ubuntu/Debian can install   macOS can install graphviz using brew:   conda environment, can install graphviz :","code":"reticulate::py_install(\"pydot\", pip = TRUE) sudo apt install graphviz brew install graphviz reticulate::conda_install(packages = \"graphviz\") # Restart the R session after install."},{"path":"https://keras3.posit.co/dev/reference/plot.keras_training_history.html","id":null,"dir":"Reference","previous_headings":"","what":"Plot training history — plot.keras_training_history","title":"Plot training history — plot.keras_training_history","text":"Plots metrics recorded training.","code":""},{"path":"https://keras3.posit.co/dev/reference/plot.keras_training_history.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Plot training history — plot.keras_training_history","text":"","code":"# S3 method for class 'keras_training_history' plot(   x,   y,   metrics = NULL,   method = c(\"auto\", \"ggplot2\", \"base\"),   smooth = getOption(\"keras.plot.history.smooth\", TRUE),   theme_bw = getOption(\"keras.plot.history.theme_bw\", FALSE),   ... )"},{"path":"https://keras3.posit.co/dev/reference/plot.keras_training_history.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Plot training history — plot.keras_training_history","text":"x Training history object returned fit.keras.src.models.model.Model(). y Unused. metrics One metrics plot (e.g. c('loss', 'accuracy')). Defaults plotting captured metrics. method Method use plotting. default \"auto\" use ggplot2 available, otherwise use base graphics. smooth Whether loess smooth added plot, available ggplot2 method. number epochs smaller ten, forced false. theme_bw Use ggplot2::theme_bw() plot history black white. ... Additional parameters pass plot() method.","code":""},{"path":"https://keras3.posit.co/dev/reference/plot.keras_training_history.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Plot training history — plot.keras_training_history","text":"method == \"ggplot2\", ggplot object returned. method == \"base\", function draw graphics device return NULL, invisibly.","code":""},{"path":"https://keras3.posit.co/dev/reference/pop_layer.html","id":null,"dir":"Reference","previous_headings":"","what":"Remove the last layer in a Sequential model — pop_layer","title":"Remove the last layer in a Sequential model — pop_layer","text":"Remove last layer Sequential model","code":""},{"path":"https://keras3.posit.co/dev/reference/pop_layer.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Remove the last layer in a Sequential model — pop_layer","text":"","code":"pop_layer(object, rebuild = TRUE)"},{"path":"https://keras3.posit.co/dev/reference/pop_layer.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Remove the last layer in a Sequential model — pop_layer","text":"object Sequential keras model object rebuild bool. Whether rebuild model removing layer. Defaults TRUE.","code":""},{"path":"https://keras3.posit.co/dev/reference/pop_layer.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Remove the last layer in a Sequential model — pop_layer","text":"removed layer.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/predict.keras.src.models.model.Model.html","id":null,"dir":"Reference","previous_headings":"","what":"Generates output predictions for the input samples. — predict.keras.src.models.model.Model","title":"Generates output predictions for the input samples. — predict.keras.src.models.model.Model","text":"Generates output predictions input samples.","code":""},{"path":"https://keras3.posit.co/dev/reference/predict.keras.src.models.model.Model.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Generates output predictions for the input samples. — predict.keras.src.models.model.Model","text":"","code":"# S3 method for class 'keras.src.models.model.Model' predict(   object,   x,   ...,   batch_size = NULL,   verbose = getOption(\"keras.verbose\", default = \"auto\"),   steps = NULL,   callbacks = NULL )"},{"path":"https://keras3.posit.co/dev/reference/predict.keras.src.models.model.Model.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Generates output predictions for the input samples. — predict.keras.src.models.model.Model","text":"object Keras model object x Input samples. can : array (array-like), list arrays (case model multiple inputs). backend-native tensor, list tensors (case model multiple inputs). TF Dataset. Python generator function. ... forward/backward compatability. batch_size Integer NULL. Number samples per batch computation. unspecified, batch_size default 32. specify batch_size input data x form TF Dataset generator (since generate batches). verbose \"auto\", 0, 1, 2. Verbosity mode. 0 = silent, 1 = progress bar, 2 = one line per epoch. \"auto\" becomes 1 cases, 2 knitr render running distributed training server. Note progress bar particularly useful logged file, verbose=2 recommended running interactively (e.g., production environment). Defaults \"auto\". steps Total number steps (batches samples) draw declaring prediction round finished. steps NULL, predict() run x exhausted. case infinitely repeating dataset, predict() run indefinitely. callbacks List Callback instances. List callbacks apply prediction.","code":""},{"path":"https://keras3.posit.co/dev/reference/predict.keras.src.models.model.Model.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Generates output predictions for the input samples. — predict.keras.src.models.model.Model","text":"R array(s) predictions.","code":""},{"path":"https://keras3.posit.co/dev/reference/predict.keras.src.models.model.Model.html","id":"details","dir":"Reference","previous_headings":"","what":"Details","title":"Generates output predictions for the input samples. — predict.keras.src.models.model.Model","text":"Computation done batches. method designed batch processing large numbers inputs. intended use inside loops iterate data process small numbers inputs time. small numbers inputs fit one batch, directly call model model$call faster execution, e.g., model(x), model(x, training = FALSE) layers BatchNormalization behave differently inference.","code":""},{"path":"https://keras3.posit.co/dev/reference/predict.keras.src.models.model.Model.html","id":"note","dir":"Reference","previous_headings":"","what":"Note","title":"Generates output predictions for the input samples. — predict.keras.src.models.model.Model","text":"See FAQ entry details difference Model methods predict() call().","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/predict_on_batch.html","id":null,"dir":"Reference","previous_headings":"","what":"Returns predictions for a single batch of samples. — predict_on_batch","title":"Returns predictions for a single batch of samples. — predict_on_batch","text":"Returns predictions single batch samples.","code":""},{"path":"https://keras3.posit.co/dev/reference/predict_on_batch.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Returns predictions for a single batch of samples. — predict_on_batch","text":"","code":"predict_on_batch(object, x)"},{"path":"https://keras3.posit.co/dev/reference/predict_on_batch.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Returns predictions for a single batch of samples. — predict_on_batch","text":"object Keras model object x Input data. must array-like.","code":""},{"path":"https://keras3.posit.co/dev/reference/predict_on_batch.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Returns predictions for a single batch of samples. — predict_on_batch","text":"Array(s) predictions.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/process_utils.html","id":null,"dir":"Reference","previous_headings":"","what":"Preprocessing and postprocessing utilities — process_utils","title":"Preprocessing and postprocessing utilities — process_utils","text":"functions used preprocess postprocess inputs outputs Keras applications.","code":""},{"path":"https://keras3.posit.co/dev/reference/process_utils.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Preprocessing and postprocessing utilities — process_utils","text":"","code":"application_preprocess_inputs(model, x, ..., data_format = NULL)  application_decode_predictions(model, preds, top = 5L, ...)"},{"path":"https://keras3.posit.co/dev/reference/process_utils.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Preprocessing and postprocessing utilities — process_utils","text":"model Keras model initialized using application_ function. x batch inputs model. x missing, preprocess_input function appropriate model returned. ... Additional arguments passed preprocessing decoding function. data_format Optional data format image tensor/array. NULL means global setting config_image_data_format() used (unless changed , uses \"channels_last\"). Defaults NULL. preds batch outputs model. top number top predictions return.","code":""},{"path":"https://keras3.posit.co/dev/reference/process_utils.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Preprocessing and postprocessing utilities — process_utils","text":"list decoded predictions case application_decode_predictions(). batch preprocessed inputs case application_preprocess_inputs().","code":""},{"path":"https://keras3.posit.co/dev/reference/process_utils.html","id":"functions","dir":"Reference","previous_headings":"","what":"Functions","title":"Preprocessing and postprocessing utilities — process_utils","text":"application_preprocess_inputs(): Pre-process inputs used model application_decode_predictions(): Decode predictions model","code":""},{"path":"https://keras3.posit.co/dev/reference/process_utils.html","id":"ref-examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Preprocessing and postprocessing utilities — process_utils","text":"","code":"if (FALSE) { # \\dontrun{ model <- application_convnext_tiny()  inputs <- random_normal(c(32, 224, 224, 3)) processed_inputs <- application_preprocess_inputs(model, inputs)  preds <- random_normal(c(32, 1000)) decoded_preds <- application_decode_predictions(model, preds)  } # }"},{"path":"https://keras3.posit.co/dev/reference/quantize_weights.html","id":null,"dir":"Reference","previous_headings":"","what":"Quantize the weights of a model. — quantize_weights","title":"Quantize the weights of a model. — quantize_weights","text":"Note model must built first calling method. quantize_weights() recursively call layer$quantize(mode) layers skipped layer implement function. Currently Dense EinsumDense layers support quantization.","code":""},{"path":"https://keras3.posit.co/dev/reference/quantize_weights.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Quantize the weights of a model. — quantize_weights","text":"","code":"quantize_weights(object, mode, ...)"},{"path":"https://keras3.posit.co/dev/reference/quantize_weights.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Quantize the weights of a model. — quantize_weights","text":"object Keras Model Layer. mode mode quantization. 'int8' supported time. ... Passed object quantization method.","code":""},{"path":"https://keras3.posit.co/dev/reference/quantize_weights.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Quantize the weights of a model. — quantize_weights","text":"model, invisibly. Note just convenience usage |>, model modified -place.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/random_beta.html","id":null,"dir":"Reference","previous_headings":"","what":"Draw samples from a Beta distribution. — random_beta","title":"Draw samples from a Beta distribution. — random_beta","text":"values drawn Beta distribution parametrized alpha beta.","code":""},{"path":"https://keras3.posit.co/dev/reference/random_beta.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Draw samples from a Beta distribution. — random_beta","text":"","code":"random_beta(shape, alpha, beta, dtype = NULL, seed = NULL)"},{"path":"https://keras3.posit.co/dev/reference/random_beta.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Draw samples from a Beta distribution. — random_beta","text":"shape shape random values generate. alpha Float array floats representing first parameter alpha. Must broadcastable beta shape. beta Float array floats representing second parameter beta. Must broadcastable alpha shape. dtype Optional dtype tensor. floating point types supported. specified, config_floatx() used, defaults \"float32\" unless configured otherwise (via config_set_floatx(float_dtype)). seed Optional R integer instance random_seed_generator(). default, seed argument NULL, internal global random_seed_generator() used. seed argument can used ensure deterministic (repeatable) random number generation. Note passing integer seed value produce random values call. generate different random values repeated calls, instance random_seed_generator() must provided seed value. Remark concerning JAX backend: tracing functions JAX backend global random_seed_generator() supported. Therefore, tracing default value seed=NULL produce error, seed argument must provided.","code":""},{"path":"https://keras3.posit.co/dev/reference/random_beta.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Draw samples from a Beta distribution. — random_beta","text":"tensor random values.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/random_binomial.html","id":null,"dir":"Reference","previous_headings":"","what":"Draw samples from a Binomial distribution. — random_binomial","title":"Draw samples from a Binomial distribution. — random_binomial","text":"values drawn Binomial distribution specified trial count probability success.","code":""},{"path":"https://keras3.posit.co/dev/reference/random_binomial.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Draw samples from a Binomial distribution. — random_binomial","text":"","code":"random_binomial(shape, counts, probabilities, dtype = NULL, seed = NULL)"},{"path":"https://keras3.posit.co/dev/reference/random_binomial.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Draw samples from a Binomial distribution. — random_binomial","text":"shape shape random values generate. counts number array numbers representing number trials. must broadcastable probabilities. probabilities float array floats representing probability success individual event. must broadcastable counts. dtype Optional dtype tensor. floating point types supported. specified, config_floatx() used, defaults \"float32\" unless configured otherwise (via config_set_floatx(float_dtype)). seed Optional R integer instance random_seed_generator(). default, seed argument NULL, internal global random_seed_generator() used. seed argument can used ensure deterministic (repeatable) random number generation. Note passing integer seed value produce random values call. generate different random values repeated calls, instance random_seed_generator() must provided seed value. Remark concerning JAX backend: tracing functions JAX backend global random_seed_generator() supported. Therefore, tracing default value seed=NULL produce error, seed argument must provided.","code":""},{"path":"https://keras3.posit.co/dev/reference/random_binomial.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Draw samples from a Binomial distribution. — random_binomial","text":"tensor random values.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/random_categorical.html","id":null,"dir":"Reference","previous_headings":"","what":"Draws samples from a categorical distribution. — random_categorical","title":"Draws samples from a categorical distribution. — random_categorical","text":"function takes input logits, 2-D input tensor shape (batch_size, num_classes). row input represents categorical distribution, column index containing log-probability given class. function output 2-D tensor shape (batch_size, num_samples), row contains samples corresponding row logits. column index contains independent samples drawn input distribution.","code":"x <- matrix(c(100, .1, 99), nrow = 1) random_categorical(x, num_samples = 5, seed = 1234) ## tf.Tensor([[3 1 1 3 3]], shape=(1, 5), dtype=int32) random_categorical(x, num_samples = 5, seed = 1234,                    zero_indexed = TRUE) ## tf.Tensor([[2 0 0 2 2]], shape=(1, 5), dtype=int32) op_take(x, random_categorical(x, num_samples = 5, seed = 1234)) ## tf.Tensor([[ 99. 100. 100.  99.  99.]], shape=(1, 5), dtype=float64) op_take(x, random_categorical(x, num_samples = 5, seed = 1234,                               zero_indexed = TRUE),         zero_indexed = TRUE) ## tf.Tensor([[ 99. 100. 100.  99.  99.]], shape=(1, 5), dtype=float64)"},{"path":"https://keras3.posit.co/dev/reference/random_categorical.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Draws samples from a categorical distribution. — random_categorical","text":"","code":"random_categorical(   logits,   num_samples,   dtype = \"int32\",   seed = NULL,   zero_indexed = FALSE )"},{"path":"https://keras3.posit.co/dev/reference/random_categorical.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Draws samples from a categorical distribution. — random_categorical","text":"logits 2-D Tensor shape (batch_size, num_classes). row define categorical distribution unnormalized log-probabilities classes. num_samples Int, number independent samples draw row input. second dimension output tensor's shape. dtype Optional dtype output tensor. seed Optional R integer instance random_seed_generator(). default, seed argument NULL, internal global random_seed_generator() used. seed argument can used ensure deterministic (repeatable) random number generation. Note passing integer seed value produce random values call. generate different random values repeated calls, instance random_seed_generator() must provided seed value. Remark concerning JAX backend: tracing functions JAX backend global random_seed_generator() supported. Therefore, tracing default value seed=NULL produce error, seed argument must provided. zero_indexed TRUE, returned indices zero-based (0 encodes first position); FALSE (default), returned indices one-based (1 encodes first position).","code":""},{"path":"https://keras3.posit.co/dev/reference/random_categorical.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Draws samples from a categorical distribution. — random_categorical","text":"2-D tensor (batch_size, num_samples).","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/random_dropout.html","id":null,"dir":"Reference","previous_headings":"","what":"Randomly set some values in a tensor to 0. — random_dropout","title":"Randomly set some values in a tensor to 0. — random_dropout","text":"Randomly set portion values tensor 0.","code":""},{"path":"https://keras3.posit.co/dev/reference/random_dropout.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Randomly set some values in a tensor to 0. — random_dropout","text":"","code":"random_dropout(inputs, rate, noise_shape = NULL, seed = NULL)"},{"path":"https://keras3.posit.co/dev/reference/random_dropout.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Randomly set some values in a tensor to 0. — random_dropout","text":"inputs tensor rate numeric noise_shape shape() value seed Initial seed random number generator","code":""},{"path":"https://keras3.posit.co/dev/reference/random_dropout.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Randomly set some values in a tensor to 0. — random_dropout","text":"tensor copy inputs values set 0.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/random_gamma.html","id":null,"dir":"Reference","previous_headings":"","what":"Draw random samples from the Gamma distribution. — random_gamma","title":"Draw random samples from the Gamma distribution. — random_gamma","text":"Draw random samples Gamma distribution.","code":""},{"path":"https://keras3.posit.co/dev/reference/random_gamma.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Draw random samples from the Gamma distribution. — random_gamma","text":"","code":"random_gamma(shape, alpha, dtype = NULL, seed = NULL)"},{"path":"https://keras3.posit.co/dev/reference/random_gamma.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Draw random samples from the Gamma distribution. — random_gamma","text":"shape shape random values generate. alpha Float, parameter distribution. dtype Optional dtype tensor. floating point types supported. specified, config_floatx() used, defaults float32 unless configured otherwise (via config_set_floatx(float_dtype)). seed Optional R integer instance random_seed_generator(). default, seed argument NULL, internal global random_seed_generator() used. seed argument can used ensure deterministic (repeatable) random number generation. Note passing integer seed value produce random values call. generate different random values repeated calls, instance random_seed_generator() must provided seed value. Remark concerning JAX backend: tracing functions JAX backend global random_seed_generator() supported. Therefore, tracing default value seed=NULL produce error, seed argument must provided.","code":""},{"path":"https://keras3.posit.co/dev/reference/random_gamma.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Draw random samples from the Gamma distribution. — random_gamma","text":"tensor random values.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/random_integer.html","id":null,"dir":"Reference","previous_headings":"","what":"Draw random integers from a uniform distribution. — random_integer","title":"Draw random integers from a uniform distribution. — random_integer","text":"generated values follow uniform distribution range [minval, maxval). lower bound minval included range, upper bound maxval excluded. dtype must integer type.","code":""},{"path":"https://keras3.posit.co/dev/reference/random_integer.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Draw random integers from a uniform distribution. — random_integer","text":"","code":"random_integer(shape, minval, maxval, dtype = \"int32\", seed = NULL)"},{"path":"https://keras3.posit.co/dev/reference/random_integer.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Draw random integers from a uniform distribution. — random_integer","text":"shape shape random values generate. minval integer, lower bound range random values generate (inclusive). maxval integer, upper bound range random values generate (exclusive). dtype Optional dtype tensor. integer types supported. specified, \"int32\" used. seed Optional R integer instance random_seed_generator(). default, seed argument NULL, internal global random_seed_generator() used. seed argument can used ensure deterministic (repeatable) random number generation. Note passing integer seed value produce random values call. generate different random values repeated calls, instance random_seed_generator() must provided seed value. Remark concerning JAX backend: tracing functions JAX backend global random_seed_generator() supported. Therefore, tracing default value seed=NULL produce error, seed argument must provided.","code":""},{"path":"https://keras3.posit.co/dev/reference/random_integer.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Draw random integers from a uniform distribution. — random_integer","text":"tensor random values.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/random_normal.html","id":null,"dir":"Reference","previous_headings":"","what":"Draw random samples from a normal (Gaussian) distribution. — random_normal","title":"Draw random samples from a normal (Gaussian) distribution. — random_normal","text":"Draw random samples normal (Gaussian) distribution.","code":""},{"path":"https://keras3.posit.co/dev/reference/random_normal.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Draw random samples from a normal (Gaussian) distribution. — random_normal","text":"","code":"random_normal(shape, mean = 0, stddev = 1, dtype = NULL, seed = NULL)"},{"path":"https://keras3.posit.co/dev/reference/random_normal.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Draw random samples from a normal (Gaussian) distribution. — random_normal","text":"shape shape random values generate. mean Float, defaults 0. Mean random values generate. stddev Float, defaults 1. Standard deviation random values generate. dtype Optional dtype tensor. floating point types supported. specified, config_floatx() used, defaults float32 unless configured otherwise (via config_set_floatx(float_dtype)). seed Optional R integer instance random_seed_generator(). default, seed argument NULL, internal global random_seed_generator() used. seed argument can used ensure deterministic (repeatable) random number generation. Note passing integer seed value produce random values call. generate different random values repeated calls, instance random_seed_generator() must provided seed value. Remark concerning JAX backend: tracing functions JAX backend global random_seed_generator() supported. Therefore, tracing default value seed=NULL produce error, seed argument must provided.","code":""},{"path":"https://keras3.posit.co/dev/reference/random_normal.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Draw random samples from a normal (Gaussian) distribution. — random_normal","text":"tensor random values.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/random_seed_generator.html","id":null,"dir":"Reference","previous_headings":"","what":"Generates variable seeds upon each call to a function generating random numbers. — random_seed_generator","title":"Generates variable seeds upon each call to a function generating random numbers. — random_seed_generator","text":"Keras, random number generators (random_normal()) stateless, meaning pass integer seed (seed=42), return values repeated calls. get different values call, SeedGenerator providing state random generator used. Note random number generators default seed NULL, implies internal global SeedGenerator used. need decouple RNG global state can provide local StateGenerator either deterministic random initial state. Remark concerning JAX backen: Note use local StateGenerator seed argument required JIT compilation RNG JAX backend, use global state supported.","code":""},{"path":"https://keras3.posit.co/dev/reference/random_seed_generator.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Generates variable seeds upon each call to a function generating random numbers. — random_seed_generator","text":"","code":"random_seed_generator(seed = NULL, name = NULL, ...)"},{"path":"https://keras3.posit.co/dev/reference/random_seed_generator.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Generates variable seeds upon each call to a function generating random numbers. — random_seed_generator","text":"seed Initial seed random number generator name String, name object ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/random_seed_generator.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Generates variable seeds upon each call to a function generating random numbers. — random_seed_generator","text":"SeedGenerator instance, can passed seed =  argument random tensor generators.","code":""},{"path":"https://keras3.posit.co/dev/reference/random_seed_generator.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Generates variable seeds upon each call to a function generating random numbers. — random_seed_generator","text":"Usage layer:","code":"seed_gen <- random_seed_generator(seed = 42) values <- random_normal(shape = c(2, 3), seed = seed_gen) new_values <- random_normal(shape = c(2, 3), seed = seed_gen) layer_dropout2 <- new_layer_class(   \"dropout2\",   initialize = function(...) {     super$initialize(...)     self$seed_generator <- random_seed_generator(seed = 1337)   },   call = function(x, training = FALSE) {     if (training) {       return(random_dropout(x, rate = 0.5, seed = self$seed_generator))     }     return(x)   } )  out <- layer_dropout(rate = 0.8) out(op_ones(10), training = TRUE) ## tf.Tensor([0. 5. 5. 0. 0. 0. 0. 0. 0. 0.], shape=(10), dtype=float32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/random_shuffle.html","id":null,"dir":"Reference","previous_headings":"","what":"Shuffle the elements of a tensor uniformly at random along an axis. — random_shuffle","title":"Shuffle the elements of a tensor uniformly at random along an axis. — random_shuffle","text":"Shuffle elements tensor uniformly random along axis.","code":""},{"path":"https://keras3.posit.co/dev/reference/random_shuffle.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Shuffle the elements of a tensor uniformly at random along an axis. — random_shuffle","text":"","code":"random_shuffle(x, axis = 1L, seed = NULL)"},{"path":"https://keras3.posit.co/dev/reference/random_shuffle.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Shuffle the elements of a tensor uniformly at random along an axis. — random_shuffle","text":"x tensor shuffled. axis integer specifying axis along shuffle. Defaults 0. seed Optional R integer instance random_seed_generator(). default, seed argument NULL, internal global random_seed_generator() used. seed argument can used ensure deterministic (repeatable) random number generation. Note passing integer seed value produce random values call. generate different random values repeated calls, instance random_seed_generator() must provided seed value. Remark concerning JAX backend: tracing functions JAX backend global random_seed_generator() supported. Therefore, tracing default value seed=NULL produce error, seed argument must provided.","code":""},{"path":"https://keras3.posit.co/dev/reference/random_shuffle.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Shuffle the elements of a tensor uniformly at random along an axis. — random_shuffle","text":"tensor, copy x axis axis shuffled.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/random_truncated_normal.html","id":null,"dir":"Reference","previous_headings":"","what":"Draw samples from a truncated normal distribution. — random_truncated_normal","title":"Draw samples from a truncated normal distribution. — random_truncated_normal","text":"values drawn normal distribution specified mean standard deviation, discarding re-drawing samples two standard deviations mean.","code":""},{"path":"https://keras3.posit.co/dev/reference/random_truncated_normal.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Draw samples from a truncated normal distribution. — random_truncated_normal","text":"","code":"random_truncated_normal(shape, mean = 0, stddev = 1, dtype = NULL, seed = NULL)"},{"path":"https://keras3.posit.co/dev/reference/random_truncated_normal.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Draw samples from a truncated normal distribution. — random_truncated_normal","text":"shape shape random values generate. mean Float, defaults 0. Mean random values generate. stddev Float, defaults 1. Standard deviation random values generate. dtype Optional dtype tensor. floating point types supported. specified, config_floatx() used, defaults float32 unless configured otherwise (via config_set_floatx(float_dtype)) seed Optional R integer instance random_seed_generator(). default, seed argument NULL, internal global random_seed_generator() used. seed argument can used ensure deterministic (repeatable) random number generation. Note passing integer seed value produce random values call. generate different random values repeated calls, instance random_seed_generator() must provided seed value. Remark concerning JAX backend: tracing functions JAX backend global random_seed_generator() supported. Therefore, tracing default value seed=NULL produce error, seed argument must provided.","code":""},{"path":"https://keras3.posit.co/dev/reference/random_truncated_normal.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Draw samples from a truncated normal distribution. — random_truncated_normal","text":"tensor random values.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/random_uniform.html","id":null,"dir":"Reference","previous_headings":"","what":"Draw samples from a uniform distribution. — random_uniform","title":"Draw samples from a uniform distribution. — random_uniform","text":"generated values follow uniform distribution range [minval, maxval). lower bound minval included range, upper bound maxval excluded. dtype must floating point type, default range [0, 1).","code":""},{"path":"https://keras3.posit.co/dev/reference/random_uniform.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Draw samples from a uniform distribution. — random_uniform","text":"","code":"random_uniform(shape, minval = 0, maxval = 1, dtype = NULL, seed = NULL)"},{"path":"https://keras3.posit.co/dev/reference/random_uniform.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Draw samples from a uniform distribution. — random_uniform","text":"shape shape random values generate. minval Float, defaults 0. Lower bound range random values generate (inclusive). maxval Float, defaults 1. Upper bound range random values generate (exclusive). dtype Optional dtype tensor. floating point types supported. specified, config_floatx() used, defaults float32 unless configured otherwise (via config_set_floatx(float_dtype)) seed Optional R integer instance random_seed_generator(). default, seed argument NULL, internal global random_seed_generator() used. seed argument can used ensure deterministic (repeatable) random number generation. Note passing integer seed value produce random values call. generate different random values repeated calls, instance random_seed_generator() must provided seed value. Remark concerning JAX backend: tracing functions JAX backend global random_seed_generator() supported. Therefore, tracing default value seed=NULL produce error, seed argument must provided.","code":""},{"path":"https://keras3.posit.co/dev/reference/random_uniform.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Draw samples from a uniform distribution. — random_uniform","text":"tensor random values.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/reexports.html","id":null,"dir":"Reference","previous_headings":"","what":"Objects exported from other packages — reexports","title":"Objects exported from other packages — reexports","text":"objects imported packages. Follow links see documentation. dotty . generics compile, fit, generate magrittr %<>% reticulate %%, array_reshape, as_iterator, iter_next, iterate, np_array, tuple, use_python, use_virtualenv tensorflow all_dims, as_tensor, evaluate, export_savedmodel, tensorboard tfruns flag_boolean, flag_integer, flag_numeric, flag_string, flags, run_dir","code":""},{"path":"https://keras3.posit.co/dev/reference/register_keras_serializable.html","id":null,"dir":"Reference","previous_headings":"","what":"Registers a custom object with the Keras serialization framework. — register_keras_serializable","title":"Registers a custom object with the Keras serialization framework. — register_keras_serializable","text":"function registers custom class function Keras custom object registry, can serialized deserialized without needing entry user-provided custom_objects argument. also injects function Keras call get object's serializable string key. Note serialized deserialized, classes must implement get_config() method. Functions requirement. object registered key 'package>name' name, defaults object name passed.","code":""},{"path":"https://keras3.posit.co/dev/reference/register_keras_serializable.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Registers a custom object with the Keras serialization framework. — register_keras_serializable","text":"","code":"register_keras_serializable(object, name = NULL, package = NULL)"},{"path":"https://keras3.posit.co/dev/reference/register_keras_serializable.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Registers a custom object with the Keras serialization framework. — register_keras_serializable","text":"object keras object. name name serialize class package. package package class belongs . used key (\"package>name\") identify class. Defaults current package name, \"Custom\" outside package.","code":""},{"path":"https://keras3.posit.co/dev/reference/register_keras_serializable.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Registers a custom object with the Keras serialization framework. — register_keras_serializable","text":"object returned invisibly, convenient piping. primarily called side effects.","code":""},{"path":"https://keras3.posit.co/dev/reference/register_keras_serializable.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Registers a custom object with the Keras serialization framework. — register_keras_serializable","text":"","code":"# Note that `'my_package'` is used as the `package` argument here, and since # the `name` argument is not provided, `'MyDense'` is used as the `name`. layer_my_dense <- Layer(\"MyDense\") register_keras_serializable(layer_my_dense, package = \"my_package\")  MyDense <- environment(layer_my_dense)$`__class__` # the python class obj stopifnot(exprs = {   get_registered_object('my_package>MyDense') == MyDense   get_registered_name(MyDense) == 'my_package>MyDense' })"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/regularizer_l1.html","id":null,"dir":"Reference","previous_headings":"","what":"A regularizer that applies a L1 regularization penalty. — regularizer_l1","title":"A regularizer that applies a L1 regularization penalty. — regularizer_l1","text":"L1 regularization penalty computed : loss = l1 * reduce_sum(abs(x)) L1 may passed layer string identifier:   case, default value used l1=0.01.","code":"dense <- layer_dense(units = 3, kernel_regularizer = 'l1')"},{"path":"https://keras3.posit.co/dev/reference/regularizer_l1.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"A regularizer that applies a L1 regularization penalty. — regularizer_l1","text":"","code":"regularizer_l1(l1 = 0.01)"},{"path":"https://keras3.posit.co/dev/reference/regularizer_l1.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"A regularizer that applies a L1 regularization penalty. — regularizer_l1","text":"l1 float, L1 regularization factor.","code":""},{"path":"https://keras3.posit.co/dev/reference/regularizer_l1.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"A regularizer that applies a L1 regularization penalty. — regularizer_l1","text":"Regularizer instance can passed layer constructors used standalone object.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/regularizer_l1_l2.html","id":null,"dir":"Reference","previous_headings":"","what":"A regularizer that applies both L1 and L2 regularization penalties. — regularizer_l1_l2","title":"A regularizer that applies both L1 and L2 regularization penalties. — regularizer_l1_l2","text":"L1 regularization penalty computed : loss = l1 * reduce_sum(abs(x)) L2 regularization penalty computed loss = l2 * reduce_sum(square(x)) L1L2 may passed layer string identifier:   case, default values used l1=0.01 l2=0.01.","code":"dense <- layer_dense(units = 3, kernel_regularizer = 'L1L2')"},{"path":"https://keras3.posit.co/dev/reference/regularizer_l1_l2.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"A regularizer that applies both L1 and L2 regularization penalties. — regularizer_l1_l2","text":"","code":"regularizer_l1_l2(l1 = 0, l2 = 0)"},{"path":"https://keras3.posit.co/dev/reference/regularizer_l1_l2.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"A regularizer that applies both L1 and L2 regularization penalties. — regularizer_l1_l2","text":"l1 float, L1 regularization factor. l2 float, L2 regularization factor.","code":""},{"path":"https://keras3.posit.co/dev/reference/regularizer_l1_l2.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"A regularizer that applies both L1 and L2 regularization penalties. — regularizer_l1_l2","text":"Regularizer instance can passed layer constructors used standalone object.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/regularizer_l2.html","id":null,"dir":"Reference","previous_headings":"","what":"A regularizer that applies a L2 regularization penalty. — regularizer_l2","title":"A regularizer that applies a L2 regularization penalty. — regularizer_l2","text":"L2 regularization penalty computed : loss = l2 * reduce_sum(square(x)) L2 may passed layer string identifier:   case, default value used l2=0.01.","code":"dense <- layer_dense(units = 3, kernel_regularizer='l2')"},{"path":"https://keras3.posit.co/dev/reference/regularizer_l2.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"A regularizer that applies a L2 regularization penalty. — regularizer_l2","text":"","code":"regularizer_l2(l2 = 0.01)"},{"path":"https://keras3.posit.co/dev/reference/regularizer_l2.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"A regularizer that applies a L2 regularization penalty. — regularizer_l2","text":"l2 float, L2 regularization factor.","code":""},{"path":"https://keras3.posit.co/dev/reference/regularizer_l2.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"A regularizer that applies a L2 regularization penalty. — regularizer_l2","text":"Regularizer instance can passed layer constructors used standalone object.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/regularizer_orthogonal.html","id":null,"dir":"Reference","previous_headings":"","what":"Regularizer that encourages input vectors to be orthogonal to each other. — regularizer_orthogonal","title":"Regularizer that encourages input vectors to be orthogonal to each other. — regularizer_orthogonal","text":"can applied either rows matrix (mode=\"rows\") columns (mode=\"columns\"). applied Dense kernel shape (input_dim, units), rows mode seek make feature vectors (.e. basis output space) orthogonal .","code":""},{"path":"https://keras3.posit.co/dev/reference/regularizer_orthogonal.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Regularizer that encourages input vectors to be orthogonal to each other. — regularizer_orthogonal","text":"","code":"regularizer_orthogonal(factor = 0.01, mode = \"rows\")"},{"path":"https://keras3.posit.co/dev/reference/regularizer_orthogonal.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Regularizer that encourages input vectors to be orthogonal to each other. — regularizer_orthogonal","text":"factor Float. regularization factor. regularization penalty proportional factor times mean dot products L2-normalized rows (mode=\"rows\", columns mode=\"columns\") inputs, excluding product row/column .  Defaults 0.01. mode String, one {\"rows\", \"columns\"}. Defaults \"rows\". rows mode, regularization effect seeks make rows input orthogonal . columns mode, seeks make columns input orthogonal .","code":""},{"path":"https://keras3.posit.co/dev/reference/regularizer_orthogonal.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Regularizer that encourages input vectors to be orthogonal to each other. — regularizer_orthogonal","text":"Regularizer instance can passed layer constructors used standalone object.","code":""},{"path":"https://keras3.posit.co/dev/reference/regularizer_orthogonal.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Regularizer that encourages input vectors to be orthogonal to each other. — regularizer_orthogonal","text":"","code":"regularizer <- regularizer_orthogonal(factor=0.01) layer <- layer_dense(units=4, kernel_regularizer=regularizer)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/reset_state.html","id":null,"dir":"Reference","previous_headings":"","what":"Reset the state for a model, layer or metric. — reset_state","title":"Reset the state for a model, layer or metric. — reset_state","text":"Reset state model, layer metric.","code":""},{"path":"https://keras3.posit.co/dev/reference/reset_state.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Reset the state for a model, layer or metric. — reset_state","text":"","code":"reset_state(object)"},{"path":"https://keras3.posit.co/dev/reference/reset_state.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Reset the state for a model, layer or metric. — reset_state","text":"object Model, Layer, Metric instance Layers resettable state (E.g., adapt()-able preprocessing layers rnn layers resettable state, layer_dense() ). Calling Layer instance without resettable-state error.","code":""},{"path":"https://keras3.posit.co/dev/reference/reset_state.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Reset the state for a model, layer or metric. — reset_state","text":"object, invisibly.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/rnn_cell_gru.html","id":null,"dir":"Reference","previous_headings":"","what":"Cell class for the GRU layer. — rnn_cell_gru","title":"Cell class for the GRU layer. — rnn_cell_gru","text":"class processes one step within whole time sequence input, whereas layer_gru() processes whole sequence.","code":""},{"path":"https://keras3.posit.co/dev/reference/rnn_cell_gru.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Cell class for the GRU layer. — rnn_cell_gru","text":"","code":"rnn_cell_gru(   units,   activation = \"tanh\",   recurrent_activation = \"sigmoid\",   use_bias = TRUE,   kernel_initializer = \"glorot_uniform\",   recurrent_initializer = \"orthogonal\",   bias_initializer = \"zeros\",   kernel_regularizer = NULL,   recurrent_regularizer = NULL,   bias_regularizer = NULL,   kernel_constraint = NULL,   recurrent_constraint = NULL,   bias_constraint = NULL,   dropout = 0,   recurrent_dropout = 0,   reset_after = TRUE,   seed = NULL,   ... )"},{"path":"https://keras3.posit.co/dev/reference/rnn_cell_gru.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Cell class for the GRU layer. — rnn_cell_gru","text":"units Positive integer, dimensionality output space. activation Activation function use. Default: hyperbolic tangent (tanh). pass NULL, activation applied (ie. \"linear\" activation: (x) = x). recurrent_activation Activation function use recurrent step. Default: sigmoid (sigmoid). pass NULL, activation applied (ie. \"linear\" activation: (x) = x). use_bias Boolean, (default TRUE), whether layer use bias vector. kernel_initializer Initializer kernel weights matrix, used linear transformation inputs. Default: \"glorot_uniform\". recurrent_initializer Initializer recurrent_kernel weights matrix, used linear transformation recurrent state. Default: \"orthogonal\". bias_initializer Initializer bias vector. Default: \"zeros\". kernel_regularizer Regularizer function applied kernel weights matrix. Default: NULL. recurrent_regularizer Regularizer function applied recurrent_kernel weights matrix. Default: NULL. bias_regularizer Regularizer function applied bias vector. Default: NULL. kernel_constraint Constraint function applied kernel weights matrix. Default: NULL. recurrent_constraint Constraint function applied recurrent_kernel weights matrix. Default: NULL. bias_constraint Constraint function applied bias vector. Default: NULL. dropout Float 0 1. Fraction units drop linear transformation inputs. Default: 0. recurrent_dropout Float 0 1. Fraction units drop linear transformation recurrent state. Default: 0. reset_after GRU convention (whether apply reset gate matrix multiplication). FALSE = \"\", TRUE = \"\" (default cuDNN compatible). seed Random seed dropout. ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/rnn_cell_gru.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Cell class for the GRU layer. — rnn_cell_gru","text":"Layer instance, intended used layer_rnn().","code":""},{"path":"https://keras3.posit.co/dev/reference/rnn_cell_gru.html","id":"call-arguments","dir":"Reference","previous_headings":"","what":"Call Arguments","title":"Cell class for the GRU layer. — rnn_cell_gru","text":"inputs: 2D tensor, shape (batch, features). states: 2D tensor shape (batch, units), state previous time step. training: Python boolean indicating whether layer behave training mode inference mode. relevant dropout recurrent_dropout used.","code":""},{"path":"https://keras3.posit.co/dev/reference/rnn_cell_gru.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Cell class for the GRU layer. — rnn_cell_gru","text":"","code":"inputs <- random_uniform(c(32, 10, 8)) outputs <- inputs |> layer_rnn(rnn_cell_gru(4)) shape(outputs) ## shape(32, 4) rnn <- layer_rnn(    cell = rnn_cell_gru(4),    return_sequences=TRUE,    return_state=TRUE) c(whole_sequence_output, final_state) %<-% rnn(inputs) shape(whole_sequence_output) ## shape(32, 10, 4) shape(final_state) ## shape(32, 4)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/rnn_cell_lstm.html","id":null,"dir":"Reference","previous_headings":"","what":"Cell class for the LSTM layer. — rnn_cell_lstm","title":"Cell class for the LSTM layer. — rnn_cell_lstm","text":"class processes one step within whole time sequence input, whereas layer_lstm() processes whole sequence.","code":""},{"path":"https://keras3.posit.co/dev/reference/rnn_cell_lstm.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Cell class for the LSTM layer. — rnn_cell_lstm","text":"","code":"rnn_cell_lstm(   units,   activation = \"tanh\",   recurrent_activation = \"sigmoid\",   use_bias = TRUE,   kernel_initializer = \"glorot_uniform\",   recurrent_initializer = \"orthogonal\",   bias_initializer = \"zeros\",   unit_forget_bias = TRUE,   kernel_regularizer = NULL,   recurrent_regularizer = NULL,   bias_regularizer = NULL,   kernel_constraint = NULL,   recurrent_constraint = NULL,   bias_constraint = NULL,   dropout = 0,   recurrent_dropout = 0,   seed = NULL,   ... )"},{"path":"https://keras3.posit.co/dev/reference/rnn_cell_lstm.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Cell class for the LSTM layer. — rnn_cell_lstm","text":"units Positive integer, dimensionality output space. activation Activation function use. Default: hyperbolic tangent (tanh). pass NULL, activation applied (ie. \"linear\" activation: (x) = x). recurrent_activation Activation function use recurrent step. Default: sigmoid (sigmoid). pass NULL, activation applied (ie. \"linear\" activation: (x) = x). use_bias Boolean, (default TRUE), whether layer use bias vector. kernel_initializer Initializer kernel weights matrix, used linear transformation inputs. Default: \"glorot_uniform\". recurrent_initializer Initializer recurrent_kernel weights matrix, used linear transformation recurrent state. Default: \"orthogonal\". bias_initializer Initializer bias vector. Default: \"zeros\". unit_forget_bias Boolean (default TRUE). TRUE, add 1 bias forget gate initialization. Setting TRUE also force bias_initializer=\"zeros\". recommended Jozefowicz et al. kernel_regularizer Regularizer function applied kernel weights matrix. Default: NULL. recurrent_regularizer Regularizer function applied recurrent_kernel weights matrix. Default: NULL. bias_regularizer Regularizer function applied bias vector. Default: NULL. kernel_constraint Constraint function applied kernel weights matrix. Default: NULL. recurrent_constraint Constraint function applied recurrent_kernel weights matrix. Default: NULL. bias_constraint Constraint function applied bias vector. Default: NULL. dropout Float 0 1. Fraction units drop linear transformation inputs. Default: 0. recurrent_dropout Float 0 1. Fraction units drop linear transformation recurrent state. Default: 0. seed Random seed dropout. ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/rnn_cell_lstm.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Cell class for the LSTM layer. — rnn_cell_lstm","text":"Layer instance, intended used layer_rnn().","code":""},{"path":"https://keras3.posit.co/dev/reference/rnn_cell_lstm.html","id":"call-arguments","dir":"Reference","previous_headings":"","what":"Call Arguments","title":"Cell class for the LSTM layer. — rnn_cell_lstm","text":"inputs: 2D tensor, shape (batch, features). states: 2D tensor shape (batch, units), state previous time step. training: Boolean indicating whether layer behave training mode inference mode. relevant dropout recurrent_dropout used.","code":""},{"path":"https://keras3.posit.co/dev/reference/rnn_cell_lstm.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Cell class for the LSTM layer. — rnn_cell_lstm","text":"","code":"inputs <- random_uniform(c(32, 10, 8)) output <- inputs |>   layer_rnn(cell = rnn_cell_lstm(4)) shape(output) ## shape(32, 4) rnn <- layer_rnn(cell = rnn_cell_lstm(4),                  return_sequences = T,                  return_state = T) c(whole_sequence_output, ..final_state) %<-% rnn(inputs) str(whole_sequence_output) ## <tf.Tensor: shape=(32, 10, 4), dtype=float32, numpy=…> str(final_state) ## List of 2 ##  $ :<tf.Tensor: shape=(32, 4), dtype=float32, numpy=…> ##  $ :<tf.Tensor: shape=(32, 4), dtype=float32, numpy=…>"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/rnn_cell_simple.html","id":null,"dir":"Reference","previous_headings":"","what":"Cell class for SimpleRNN. — rnn_cell_simple","title":"Cell class for SimpleRNN. — rnn_cell_simple","text":"class processes one step within whole time sequence input, whereas layer_simple_rnn() processes whole sequence.","code":""},{"path":"https://keras3.posit.co/dev/reference/rnn_cell_simple.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Cell class for SimpleRNN. — rnn_cell_simple","text":"","code":"rnn_cell_simple(   units,   activation = \"tanh\",   use_bias = TRUE,   kernel_initializer = \"glorot_uniform\",   recurrent_initializer = \"orthogonal\",   bias_initializer = \"zeros\",   kernel_regularizer = NULL,   recurrent_regularizer = NULL,   bias_regularizer = NULL,   kernel_constraint = NULL,   recurrent_constraint = NULL,   bias_constraint = NULL,   dropout = 0,   recurrent_dropout = 0,   seed = NULL,   ... )"},{"path":"https://keras3.posit.co/dev/reference/rnn_cell_simple.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Cell class for SimpleRNN. — rnn_cell_simple","text":"units Positive integer, dimensionality output space. activation Activation function use. Default: hyperbolic tangent (tanh). pass NULL, activation applied (ie. \"linear\" activation: (x) = x). use_bias Boolean, (default TRUE), whether layer use bias vector. kernel_initializer Initializer kernel weights matrix, used linear transformation inputs. Default: \"glorot_uniform\". recurrent_initializer Initializer recurrent_kernel weights matrix, used linear transformation recurrent state. Default: \"orthogonal\". bias_initializer Initializer bias vector. Default: \"zeros\". kernel_regularizer Regularizer function applied kernel weights matrix. Default: NULL. recurrent_regularizer Regularizer function applied recurrent_kernel weights matrix. Default: NULL. bias_regularizer Regularizer function applied bias vector. Default: NULL. kernel_constraint Constraint function applied kernel weights matrix. Default: NULL. recurrent_constraint Constraint function applied recurrent_kernel weights matrix. Default: NULL. bias_constraint Constraint function applied bias vector. Default: NULL. dropout Float 0 1. Fraction units drop linear transformation inputs. Default: 0. recurrent_dropout Float 0 1. Fraction units drop linear transformation recurrent state. Default: 0. seed Random seed dropout. ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/rnn_cell_simple.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Cell class for SimpleRNN. — rnn_cell_simple","text":"Layer instance, intended used layer_rnn().","code":""},{"path":"https://keras3.posit.co/dev/reference/rnn_cell_simple.html","id":"call-arguments","dir":"Reference","previous_headings":"","what":"Call Arguments","title":"Cell class for SimpleRNN. — rnn_cell_simple","text":"sequence: 2D tensor, shape (batch, features). states: 2D tensor shape (batch, units), state previous time step. training: Python boolean indicating whether layer behave training mode inference mode. relevant dropout recurrent_dropout used.","code":""},{"path":"https://keras3.posit.co/dev/reference/rnn_cell_simple.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Cell class for SimpleRNN. — rnn_cell_simple","text":"","code":"inputs <- random_uniform(c(32, 10, 8)) rnn <- layer_rnn(cell = rnn_cell_simple(units = 4)) output <- rnn(inputs)  # The output has shape `(32, 4)`. rnn <- layer_rnn(     cell = rnn_cell_simple(units = 4),     return_sequences=TRUE,     return_state=TRUE ) # whole_sequence_output has shape `(32, 10, 4)`. # final_state has shape `(32, 4)`. c(whole_sequence_output, final_state) %<-% rnn(inputs)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/rnn_cells_stack.html","id":null,"dir":"Reference","previous_headings":"","what":"Wrapper allowing a stack of RNN cells to behave as a single cell. — rnn_cells_stack","title":"Wrapper allowing a stack of RNN cells to behave as a single cell. — rnn_cells_stack","text":"Used implement efficient stacked RNNs.","code":""},{"path":"https://keras3.posit.co/dev/reference/rnn_cells_stack.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Wrapper allowing a stack of RNN cells to behave as a single cell. — rnn_cells_stack","text":"","code":"rnn_cells_stack(cells, ...)"},{"path":"https://keras3.posit.co/dev/reference/rnn_cells_stack.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Wrapper allowing a stack of RNN cells to behave as a single cell. — rnn_cells_stack","text":"cells List RNN cell instances. ... Unnamed arguments treated additional cells. Named arguments passed underlying layer.","code":""},{"path":"https://keras3.posit.co/dev/reference/rnn_cells_stack.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Wrapper allowing a stack of RNN cells to behave as a single cell. — rnn_cells_stack","text":"Layer instance, intended used layer_rnn().","code":""},{"path":"https://keras3.posit.co/dev/reference/rnn_cells_stack.html","id":"example","dir":"Reference","previous_headings":"","what":"Example","title":"Wrapper allowing a stack of RNN cells to behave as a single cell. — rnn_cells_stack","text":"","code":"batch_size <- 3 sentence_length <- 5 num_features <- 2 new_shape <- c(batch_size, sentence_length, num_features) x <- array(1:30, dim = new_shape)  rnn_cells <- lapply(1:2, function(x) rnn_cell_lstm(units = 128)) stacked_lstm <- rnn_cells_stack(rnn_cells) lstm_layer <- layer_rnn(cell = stacked_lstm)  result <- lstm_layer(x) str(result) ## <tf.Tensor: shape=(3, 128), dtype=float32, numpy=…>"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/save_model.html","id":null,"dir":"Reference","previous_headings":"","what":"Saves a model as a .keras file. — save_model","title":"Saves a model as a .keras file. — save_model","text":"Saves model .keras file.","code":""},{"path":"https://keras3.posit.co/dev/reference/save_model.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Saves a model as a .keras file. — save_model","text":"","code":"save_model(model, filepath = NULL, overwrite = FALSE, zipped = NULL, ...)"},{"path":"https://keras3.posit.co/dev/reference/save_model.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Saves a model as a .keras file. — save_model","text":"model keras model. filepath string, Path save model. Must end .keras. overwrite Whether overwrite existing model target location, instead ask user via interactive prompt. zipped Whether save model zipped .keras archive (default saving locally), unzipped directory (default saving Hugging Face Hub). ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/save_model.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Saves a model as a .keras file. — save_model","text":"filepath provided, function called primarily side effects, model returned invisibly. filepath provided NULL, serialized model returned R raw vector.","code":""},{"path":"https://keras3.posit.co/dev/reference/save_model.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Saves a model as a .keras file. — save_model","text":"saved .keras file zip archive contains: model's configuration (architecture) model's weights model's optimizer's state () Thus models can reinstantiated exact state.","code":"model <- keras_model_sequential(input_shape = c(3)) |>   layer_dense(5) |>   layer_activation_softmax()  model |> save_model(\"model.keras\") loaded_model <- load_model(\"model.keras\") x <- random_uniform(c(10, 3)) stopifnot(all.equal(   model |> predict(x),   loaded_model |> predict(x) )) zip::zip_list(\"model.keras\")[, \"filename\"] ## [1] \"metadata.json\"    \"config.json\"      \"model.weights.h5\""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/save_model_config.html","id":null,"dir":"Reference","previous_headings":"","what":"Save and load model configuration as JSON — save_model_config","title":"Save and load model configuration as JSON — save_model_config","text":"Save re-load models configurations JSON. Note representation include weights, architecture.","code":""},{"path":"https://keras3.posit.co/dev/reference/save_model_config.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Save and load model configuration as JSON — save_model_config","text":"","code":"save_model_config(model, filepath = NULL, overwrite = FALSE)  load_model_config(filepath, custom_objects = NULL)"},{"path":"https://keras3.posit.co/dev/reference/save_model_config.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Save and load model configuration as JSON — save_model_config","text":"model Model object save filepath path json file model config. overwrite Whether overwrite existing model configuration json filepath, instead ask user via interactive prompt. custom_objects Optional named list mapping names custom classes functions considered deserialization.","code":""},{"path":"https://keras3.posit.co/dev/reference/save_model_config.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Save and load model configuration as JSON — save_model_config","text":"called primarily side effects. model returned, invisibly, enable usage pipe.","code":""},{"path":"https://keras3.posit.co/dev/reference/save_model_config.html","id":"details","dir":"Reference","previous_headings":"","what":"Details","title":"Save and load model configuration as JSON — save_model_config","text":"Note: save_model_config() serializes model JSON using serialize_keras_object(), get_config(). serialize_keras_object() returns superset get_config(), additional information needed create class object needed restore model. See example extract get_config() value saved model.","code":""},{"path":"https://keras3.posit.co/dev/reference/save_model_config.html","id":"example","dir":"Reference","previous_headings":"","what":"Example","title":"Save and load model configuration as JSON — save_model_config","text":"","code":"model <- keras_model_sequential(input_shape = 10) |> layer_dense(10) file <- tempfile(\"model-config-\", fileext = \".json\") save_model_config(model, file)  # load a new model instance with the same architecture but different weights model2 <- load_model_config(file)  stopifnot(exprs = {   all.equal(get_config(model), get_config(model2))    # To extract the `get_config()` value from a saved model config:   all.equal(       get_config(model),       structure(jsonlite::read_json(file)$config,                 \"__class__\" = keras_model_sequential()$`__class__`)   ) })"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/save_model_weights.html","id":null,"dir":"Reference","previous_headings":"","what":"Saves all weights to a single file or sharded files. — save_model_weights","title":"Saves all weights to a single file or sharded files. — save_model_weights","text":"default, weights saved single .weights.h5 file. Enable sharding via max_shard_size split weights across multiple files (GB) produce .weights.json manifest tracks shard metadata. saved sharded files contain: *.weights.json: configuration file containing metadata weight_map entries. *_xxxxxx.weights.h5: weight shards limited max_shard_size.","code":"model <-   keras_model_sequential(input_shape = 2) |>   layer_dense(4)  path_h5 <- tempfile(fileext = \".weights.h5\") path_json <- tempfile(fileext = \".weights.json\")  model |> save_model_weights(path_h5) model |> save_model_weights(path_json, max_shard_size = 0.01)  model |> load_model_weights(path_h5) model |> load_model_weights(path_json)"},{"path":"https://keras3.posit.co/dev/reference/save_model_weights.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Saves all weights to a single file or sharded files. — save_model_weights","text":"","code":"save_model_weights(model, filepath, overwrite = FALSE, max_shard_size = NULL)"},{"path":"https://keras3.posit.co/dev/reference/save_model_weights.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Saves all weights to a single file or sharded files. — save_model_weights","text":"model keras Model object. filepath Path weights saved. Accepts .weights.h5, sharding enabled, .weights.json manifest path. .weights.h5 provided sharding, filename overridden end .weights.json. overwrite Whether overwrite existing weights target location, instead ask user via interactive prompt. max_shard_size Numeric size GB sharded file. Use NULL disable sharding.","code":""},{"path":"https://keras3.posit.co/dev/reference/save_model_weights.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Saves all weights to a single file or sharded files. — save_model_weights","text":"called primarily side effects. model returned, invisibly, enable usage pipe.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/serialize_keras_object.html","id":null,"dir":"Reference","previous_headings":"","what":"Retrieve the full config by serializing the Keras object. — serialize_keras_object","title":"Retrieve the full config by serializing the Keras object. — serialize_keras_object","text":"serialize_keras_object() serializes Keras object named list represents object, reciprocal function deserialize_keras_object(). See deserialize_keras_object() information full config format.","code":""},{"path":"https://keras3.posit.co/dev/reference/serialize_keras_object.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Retrieve the full config by serializing the Keras object. — serialize_keras_object","text":"","code":"serialize_keras_object(obj)"},{"path":"https://keras3.posit.co/dev/reference/serialize_keras_object.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Retrieve the full config by serializing the Keras object. — serialize_keras_object","text":"obj Keras object serialize.","code":""},{"path":"https://keras3.posit.co/dev/reference/serialize_keras_object.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Retrieve the full config by serializing the Keras object. — serialize_keras_object","text":"named list represents object config. config expected contain simple types , can saved json. object can deserialized config via deserialize_keras_object().","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/set_random_seed.html","id":null,"dir":"Reference","previous_headings":"","what":"Sets all random seeds (Python, NumPy, and backend framework, e.g. TF). — set_random_seed","title":"Sets all random seeds (Python, NumPy, and backend framework, e.g. TF). — set_random_seed","text":"can use utility make almost Keras program fully deterministic. limitations apply cases network communications involved (e.g. parameter server distribution), creates additional sources randomness, certain non-deterministic cuDNN ops involved. sets: R session seed: set.seed() Python session seed: import random; random.seed(seed) Python NumPy seed: import numpy; numpy.random.seed(seed) TensorFlow seed: tf$random$set_seed(seed) (TF installed) Torch seed: import(\"torch\")$manual_seed(seed) (backend torch) disables Python hash randomization. Note TensorFlow seed set even using TensorFlow backend framework, since many workflows leverage tf$data pipelines (feature random shuffling). Likewise many workflows might leverage NumPy APIs.","code":""},{"path":"https://keras3.posit.co/dev/reference/set_random_seed.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Sets all random seeds (Python, NumPy, and backend framework, e.g. TF). — set_random_seed","text":"","code":"set_random_seed(seed)"},{"path":"https://keras3.posit.co/dev/reference/set_random_seed.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Sets all random seeds (Python, NumPy, and backend framework, e.g. TF). — set_random_seed","text":"seed Integer, random seed use.","code":""},{"path":"https://keras3.posit.co/dev/reference/set_random_seed.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Sets all random seeds (Python, NumPy, and backend framework, e.g. TF). — set_random_seed","text":"return value, called side effects.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/set_state_tree.html","id":null,"dir":"Reference","previous_headings":"","what":"Assigns values to variables of the model. — set_state_tree","title":"Assigns values to variables of the model. — set_state_tree","text":"method takes dictionary (named list) nested variable values, represents state tree model, assigns corresponding variables model. keys (list names) represent variable names (e.g., 'trainable_variables', 'optimizer_variables'), values nested dictionaries containing variable paths corresponding values.","code":""},{"path":"https://keras3.posit.co/dev/reference/set_state_tree.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Assigns values to variables of the model. — set_state_tree","text":"","code":"set_state_tree(object, state_tree)"},{"path":"https://keras3.posit.co/dev/reference/set_state_tree.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Assigns values to variables of the model. — set_state_tree","text":"object keras model. state_tree dictionary representing state tree model. keys variable names, values nested dictionaries representing variable paths values.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/shape.html","id":null,"dir":"Reference","previous_headings":"","what":"Tensor shape utility — shape","title":"Tensor shape utility — shape","text":"function can used get create tensor shape.","code":""},{"path":"https://keras3.posit.co/dev/reference/shape.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Tensor shape utility — shape","text":"","code":"shape(...)  # S3 method for class 'keras_shape' format(x, ..., prefix = TRUE)  # S3 method for class 'keras_shape' print(x, ...)  # S3 method for class 'keras_shape' x[...]  # S3 method for class 'keras_shape' as.integer(x, ...)  # S3 method for class 'keras_shape' Summary(..., na.rm = FALSE)  # S3 method for class 'keras_shape' as.list(x, ...)  # S3 method for class 'keras_shape' x == y  # S3 method for class 'keras_shape' x != y"},{"path":"https://keras3.posit.co/dev/reference/shape.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Tensor shape utility — shape","text":"... shape specification. Numerics, NULL tensors valid. NULL, NA, -1L can used specify unspecified dim size. Tensors dispatched op_shape() extract tensor shape. Values wrapped () used asis (see examples). objects coerced via .integer(). x, y keras_shape object. prefix Whether format shape object prefix. Defaults \"shape\". na.rm passed Summary group generics like prod(). Unknown axes treated NA.","code":""},{"path":"https://keras3.posit.co/dev/reference/shape.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Tensor shape utility — shape","text":"list \"keras_shape\" class attribute. element list either ) NULL, b) R integer c) scalar integer tensor (e.g., supplied TF tensor unspecified dimension function traced).","code":""},{"path":"https://keras3.posit.co/dev/reference/shape.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Tensor shape utility — shape","text":"3 ways specify unknown dimension     functions take 'shape' argument also coerce shape()     can also use shape() get shape tensor (excepting scalar integer tensors).             Combine expand shapes         Scalar integer tensors treated axis values. commonly encountered tracing function graph mode, axis size might unknown.     useful pattern unpack shape() %<-%, like :                     want resolve shape tensor can potentially scalar integer, can wrap tensor (), use op_shape().","code":"shape(1, 2, 3) ## shape(1, 2, 3) shape(NA,   2, 3) shape(NULL, 2, 3) shape(-1,   2, 3) ## shape(NA, 2, 3) ## shape(NA, 2, 3) ## shape(NA, 2, 3) layer_input(c(1, 2, 3)) layer_input(shape(1, 2, 3)) ## <KerasTensor shape=(None, 1, 2, 3), dtype=float32, sparse=False, ragged=False, name=keras_tensor> ## <KerasTensor shape=(None, 1, 2, 3), dtype=float32, sparse=False, ragged=False, name=keras_tensor_1> symbolic_tensor <- layer_input(shape(1, 2, 3)) shape(symbolic_tensor) ## shape(NA, 1, 2, 3) eager_tensor <- op_ones(c(1,2,3)) shape(eager_tensor) ## shape(1, 2, 3) op_shape(eager_tensor) ## shape(1, 2, 3) shape(symbolic_tensor, 4) ## shape(NA, 1, 2, 3, 4) shape(5, symbolic_tensor, 4) ## shape(5, NA, 1, 2, 3, 4) tfn <- tensorflow::tf_function(function(x) {   print(op_shape(x))   x }, input_signature = list(tensorflow::tf$TensorSpec(shape(1, NA, 3)))) invisible(tfn(op_ones(shape(1, 2, 3)))) ## shape(1, Tensor(\"strided_slice:0\", shape=(), dtype=int32), 3) c(batch_size, seq_len, channels) %<-% shape(x)  # `%<-%` also has support for skipping values # during unpacking with `.` and `...`. For example, # To retrieve just the first and/or last dim: c(batch_size, ...) %<-% shape(x) c(batch_size, ., .) %<-% shape(x) c(..., channels) %<-% shape(x) c(batch_size, ..., channels) %<-% shape(x) c(batch_size, ., channels) %<-% shape(x) echo_print <- function(x) {   message(\"> \", deparse(substitute(x)));   if(!is.null(x)) print(x) } tfn <- tensorflow::tf_function(function(x) {   c(axis1, axis2, axis3) %<-% shape(x)   echo_print(str(list(axis1 = axis1, axis2 = axis2, axis3 = axis3)))    echo_print(shape(axis1))               # use axis1 tensor as axis value   echo_print(shape(axis1, axis2, axis3)) # use axis1 tensor as axis value    # use shape() to compose a new shape, e.g., in multihead attention   n_heads <- 4   echo_print(shape(axis1, axis2, n_heads, axis3/n_heads))    x }, input_signature = list(tensorflow::tf$TensorSpec(shape(NA, 4, 16)))) invisible(tfn(op_ones(shape(2, 4, 16)))) ## > str(list(axis1 = axis1, axis2 = axis2, axis3 = axis3)) ## List of 3 ##  $ axis1:<tf.Tensor 'strided_slice:0' shape=() dtype=int32> ##  $ axis2: int 4 ##  $ axis3: int 16 ## > shape(axis1) ## shape(Tensor(\"strided_slice:0\", shape=(), dtype=int32)) ## > shape(axis1, axis2, axis3) ## shape(Tensor(\"strided_slice:0\", shape=(), dtype=int32), 4, 16) ## > shape(axis1, axis2, n_heads, axis3/n_heads) ## shape(Tensor(\"strided_slice:0\", shape=(), dtype=int32), 4, 4, 4) (x <- op_convert_to_tensor(2L)) ## tf.Tensor(2, shape=(), dtype=int32) # by default, shape() treats scalar integer tensors as axis values shape(x) ## shape(tf.Tensor(2, shape=(), dtype=int32)) # to access the shape of a scalar integer, # call `op_shape()`, or protect with `I()` op_shape(x) ## shape() shape(I(x)) ## shape()"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/split_dataset.html","id":null,"dir":"Reference","previous_headings":"","what":"Splits a dataset into a left half and a right half (e.g. train / test). — split_dataset","title":"Splits a dataset into a left half and a right half (e.g. train / test). — split_dataset","text":"Splits dataset left half right half (e.g. train / test).","code":""},{"path":"https://keras3.posit.co/dev/reference/split_dataset.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Splits a dataset into a left half and a right half (e.g. train / test). — split_dataset","text":"","code":"split_dataset(   dataset,   left_size = NULL,   right_size = NULL,   shuffle = FALSE,   seed = NULL )"},{"path":"https://keras3.posit.co/dev/reference/split_dataset.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Splits a dataset into a left half and a right half (e.g. train / test). — split_dataset","text":"dataset tf$data$Dataset, torch$utils$data$Dataset object, list arrays length. left_size float (range [0, 1]), signifies fraction data pack left dataset. integer, signifies number samples pack left dataset. NULL, defaults complement right_size. Defaults NULL. right_size float (range [0, 1]), signifies fraction data pack right dataset. integer, signifies number samples pack right dataset. NULL, defaults complement left_size. Defaults NULL. shuffle Boolean, whether shuffle data splitting . seed random seed shuffling.","code":""},{"path":"https://keras3.posit.co/dev/reference/split_dataset.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Splits a dataset into a left half and a right half (e.g. train / test). — split_dataset","text":"list two tf$data$Dataset objects: left right splits.","code":""},{"path":"https://keras3.posit.co/dev/reference/split_dataset.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Splits a dataset into a left half and a right half (e.g. train / test). — split_dataset","text":"","code":"data <- random_uniform(c(1000, 4)) c(left_ds, right_ds) %<-% split_dataset(list(data$numpy()), left_size = 0.8) left_ds$cardinality() ## tf.Tensor(800, shape=(), dtype=int64) right_ds$cardinality() ## tf.Tensor(200, shape=(), dtype=int64)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/summary.keras.src.models.model.Model.html","id":null,"dir":"Reference","previous_headings":"","what":"Print a summary of a Keras Model — summary.keras.src.models.model.Model","title":"Print a summary of a Keras Model — summary.keras.src.models.model.Model","text":"Print summary Keras Model","code":""},{"path":"https://keras3.posit.co/dev/reference/summary.keras.src.models.model.Model.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Print a summary of a Keras Model — summary.keras.src.models.model.Model","text":"","code":"# S3 method for class 'keras.src.models.model.Model' summary(object, ...)  # S3 method for class 'keras.src.models.model.Model' format(   x,   line_length = min(getOption(\"width\"), 180),   positions = NULL,   expand_nested = FALSE,   show_trainable = NA,   ...,   layer_range = NULL,   compact = TRUE )  # S3 method for class 'keras.src.models.model.Model' print(x, ...)"},{"path":"https://keras3.posit.co/dev/reference/summary.keras.src.models.model.Model.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Print a summary of a Keras Model — summary.keras.src.models.model.Model","text":"object, x Keras model instance ... summary() print(), passed format(). format(), passed model$summary(). line_length Total length printed lines positions Relative absolute positions log elements line. provided, defaults c(0.33, 0.55, 0.67, 1.0). expand_nested Whether expand nested models. provided, defaults FALSE. show_trainable Whether show layer trainable. provided, defaults FALSE. layer_range list, tuple, vector 2 strings, starting layer name ending layer name (inclusive) indicating range layers printed summary. also accepts regex patterns instead exact name. case, start predicate first element matches layer_range[[1]] end predicate last element matches layer_range[[1]]. default NULL considers layers model. compact Whether remove white-space lines model summary. (Default TRUE)","code":""},{"path":"https://keras3.posit.co/dev/reference/summary.keras.src.models.model.Model.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Print a summary of a Keras Model — summary.keras.src.models.model.Model","text":"format() returns length 1 character vector. print() returns model object invisibly. summary() returns output format() invisibly printing .","code":""},{"path":"https://keras3.posit.co/dev/reference/summary.keras.src.models.model.Model.html","id":"enabling-color-output-in-knitr-rmarkdown-quarto-","dir":"Reference","previous_headings":"","what":"Enabling color output in Knitr (RMarkdown, Quarto)","title":"Print a summary of a Keras Model — summary.keras.src.models.model.Model","text":"order enable color output quarto rmarkdown document html output format (include revealjs presentations), need following setup chunk:","code":"```{r setup, include = FALSE}  options(cli.num_colors = 256)  fansi::set_knit_hooks(knitr::knit_hooks)  options(width = 75) # adjust as needed for format  ```"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/test_on_batch.html","id":null,"dir":"Reference","previous_headings":"","what":"Test the model on a single batch of samples. — test_on_batch","title":"Test the model on a single batch of samples. — test_on_batch","text":"Test model single batch samples.","code":""},{"path":"https://keras3.posit.co/dev/reference/test_on_batch.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Test the model on a single batch of samples. — test_on_batch","text":"","code":"test_on_batch(object, x, y = NULL, sample_weight = NULL, ...)"},{"path":"https://keras3.posit.co/dev/reference/test_on_batch.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Test the model on a single batch of samples. — test_on_batch","text":"object Keras model object x Input data. Must array-like. y Target data. Must array-like. sample_weight Optional array length x, containing weights apply model's loss sample. case temporal data, can pass 2D array shape (samples, sequence_length), apply different weight every timestep every sample. ... forward/backward compatability","code":""},{"path":"https://keras3.posit.co/dev/reference/test_on_batch.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Test the model on a single batch of samples. — test_on_batch","text":"scalar loss value (metrics), named list loss metric values (metrics).","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/text_dataset_from_directory.html","id":null,"dir":"Reference","previous_headings":"","what":"Generates a tf.data.Dataset from text files in a directory. — text_dataset_from_directory","title":"Generates a tf.data.Dataset from text files in a directory. — text_dataset_from_directory","text":"directory structure :   calling text_dataset_from_directory(main_directory, labels='inferred') return tf.data.Dataset yields batches texts subdirectories class_a class_b, together labels 0 1 (0 corresponding class_a 1 corresponding class_b). .txt files supported time.","code":"main_directory/ ...class_a/ ......a_text_1.txt ......a_text_2.txt ...class_b/ ......b_text_1.txt ......b_text_2.txt"},{"path":"https://keras3.posit.co/dev/reference/text_dataset_from_directory.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Generates a tf.data.Dataset from text files in a directory. — text_dataset_from_directory","text":"","code":"text_dataset_from_directory(   directory,   labels = \"inferred\",   label_mode = \"int\",   class_names = NULL,   batch_size = 32L,   max_length = NULL,   shuffle = TRUE,   seed = NULL,   validation_split = NULL,   subset = NULL,   follow_links = FALSE,   verbose = TRUE )"},{"path":"https://keras3.posit.co/dev/reference/text_dataset_from_directory.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Generates a tf.data.Dataset from text files in a directory. — text_dataset_from_directory","text":"directory Directory data located. labels \"inferred\", contain subdirectories, containing text files class. Otherwise, directory structure ignored. labels Either \"inferred\" (labels generated directory structure), NULL (labels), list/tuple integer labels size number text files found directory. Labels sorted according alphanumeric order text file paths (obtained via os.walk(directory) Python). label_mode String describing encoding labels. Options : \"int\": means labels encoded integers (e.g. sparse_categorical_crossentropy loss). \"categorical\" means labels encoded categorical vector (e.g. categorical_crossentropy loss). \"binary\" means labels (can 2) encoded float32 scalars values 0 1 (e.g. binary_crossentropy). NULL (labels). class_names valid \"labels\" \"inferred\". explicit list class names (must match names subdirectories). Used control order classes (otherwise alphanumerical order used). batch_size Size batches data. NULL, data batched (dataset yield individual samples). Defaults 32. max_length Maximum size text string. Texts longer truncated max_length. shuffle Whether shuffle data. set FALSE, sorts data alphanumeric order. Defaults TRUE. seed Optional random seed shuffling transformations. validation_split Optional float 0 1, fraction data reserve validation. subset Subset data return. One \"training\", \"validation\" \"\". used validation_split set. subset=\"\", utility returns tuple two datasets (training validation datasets respectively). follow_links Whether visits subdirectories pointed symlinks. Defaults FALSE. verbose Whether display number information classes number files found. Defaults TRUE.","code":""},{"path":"https://keras3.posit.co/dev/reference/text_dataset_from_directory.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Generates a tf.data.Dataset from text files in a directory. — text_dataset_from_directory","text":"tf.data.Dataset object. label_mode NULL, yields string tensors shape (batch_size,), containing contents batch text files. Otherwise, yields tuple (texts, labels), texts shape (batch_size,) labels follows format described . Rules regarding labels format: label_mode int, labels int32 tensor shape (batch_size,). label_mode binary, labels float32 tensor 1s 0s shape (batch_size, 1). label_mode categorical, labels float32 tensor shape (batch_size, num_classes), representing one-hot encoding class index.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/time_distributed.html","id":null,"dir":"Reference","previous_headings":"","what":"layer_time_distributed — time_distributed","title":"layer_time_distributed — time_distributed","text":"time_distributed() alias layer_time_distributed(). See ?layer_time_distributed() full documentation.","code":""},{"path":"https://keras3.posit.co/dev/reference/time_distributed.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"layer_time_distributed — time_distributed","text":"","code":"time_distributed(object, layer, ...)"},{"path":"https://keras3.posit.co/dev/reference/time_distributed.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"layer_time_distributed — time_distributed","text":"object Object compose layer . tensor, array, sequential model. layer Layer instance. ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/time_distributed.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"layer_time_distributed — time_distributed","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/timeseries_dataset_from_array.html","id":null,"dir":"Reference","previous_headings":"","what":"Creates a dataset of sliding windows over a timeseries provided as array. — timeseries_dataset_from_array","title":"Creates a dataset of sliding windows over a timeseries provided as array. — timeseries_dataset_from_array","text":"function takes sequence data-points gathered equal intervals, along time series parameters length sequences/windows, spacing two sequence/windows, etc., produce batches timeseries inputs targets.","code":""},{"path":"https://keras3.posit.co/dev/reference/timeseries_dataset_from_array.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Creates a dataset of sliding windows over a timeseries provided as array. — timeseries_dataset_from_array","text":"","code":"timeseries_dataset_from_array(   data,   targets,   sequence_length,   sequence_stride = 1L,   sampling_rate = 1L,   batch_size = 128L,   shuffle = FALSE,   seed = NULL,   start_index = NULL,   end_index = NULL )"},{"path":"https://keras3.posit.co/dev/reference/timeseries_dataset_from_array.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Creates a dataset of sliding windows over a timeseries provided as array. — timeseries_dataset_from_array","text":"data array eager tensor containing consecutive data points (timesteps). first dimension expected time dimension. targets Targets corresponding timesteps data. targets[] target corresponding window starts index (see example 2 ). Pass NULL target data (case dataset yield input data). sequence_length Length output sequences (number timesteps). sequence_stride Period successive output sequences. stride s, output samples start index data[], data[+ s], data[+ 2 * s], etc. sampling_rate Period successive individual timesteps within sequences. rate r, timesteps data[], data[+ r], ... data[+ sequence_length] used creating sample sequence. batch_size Number timeseries samples batch (except maybe last one). NULL, data batched (dataset yield individual samples). shuffle Whether shuffle output samples, instead draw chronological order. seed Optional int; random seed shuffling. start_index Optional int; data points earlier (exclusive) start_index used output sequences. useful reserve part data test validation. end_index Optional int; data points later (exclusive) end_index used output sequences. useful reserve part data test validation.","code":""},{"path":"https://keras3.posit.co/dev/reference/timeseries_dataset_from_array.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Creates a dataset of sliding windows over a timeseries provided as array. — timeseries_dataset_from_array","text":"tf$data$Dataset instance. targets passed, dataset yields list (batch_of_sequences, batch_of_targets). , dataset yields batch_of_sequences. Example 1: Consider indices [0, 1, ... 98]. sequence_length=10,  sampling_rate=2, sequence_stride=3, shuffle=FALSE, dataset yield batches sequences composed following indices:   case last 2 data points discarded since full sequence can generated include (next sequence started index 81, thus last step gone 98). Example 2: Temporal regression. Consider array data scalar values, shape (steps,). generate dataset uses past 10 timesteps predict next timestep, use:     Example 3: Temporal regression many--many architectures. Consider two arrays scalar values X Y, shape (100,). resulting dataset consist samples 20 timestamps . samples overlap. generate dataset uses current timestamp predict corresponding target timestep, use:","code":"First sequence:  [0  2  4  6  8 10 12 14 16 18] Second sequence: [3  5  7  9 11 13 15 17 19 21] Third sequence:  [6  8 10 12 14 16 18 20 22 24] ... Last sequence:   [78 80 82 84 86 88 90 92 94 96] data <- op_array(1:20) input_data <- data[1:10] targets <- data[11:20] dataset <- timeseries_dataset_from_array(   input_data, targets, sequence_length=10) iter <- reticulate::as_iterator(dataset) reticulate::iter_next(iter) ## [[1]] ## tf.Tensor([[ 1  2  3  4  5  6  7  8  9 10]], shape=(1, 10), dtype=int32) ## ## [[2]] ## tf.Tensor([11], shape=(1), dtype=int32) X <- op_array(1:100) Y <- X*2  sample_length <- 20 input_dataset <- timeseries_dataset_from_array(     X, NULL, sequence_length=sample_length, sequence_stride=sample_length) target_dataset <- timeseries_dataset_from_array(     Y, NULL, sequence_length=sample_length, sequence_stride=sample_length)   inputs <- reticulate::as_iterator(input_dataset) %>% reticulate::iter_next() targets <- reticulate::as_iterator(target_dataset) %>% reticulate::iter_next()"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/to_categorical.html","id":null,"dir":"Reference","previous_headings":"","what":"Converts a class vector (integers) to binary class matrix. — to_categorical","title":"Converts a class vector (integers) to binary class matrix. — to_categorical","text":"E.g. use loss_categorical_crossentropy().","code":""},{"path":"https://keras3.posit.co/dev/reference/to_categorical.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Converts a class vector (integers) to binary class matrix. — to_categorical","text":"","code":"to_categorical(x, num_classes = NULL)"},{"path":"https://keras3.posit.co/dev/reference/to_categorical.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Converts a class vector (integers) to binary class matrix. — to_categorical","text":"x Array-like class values converted matrix (integers 0 num_classes - 1). R factors coerced integer offset 0-based, .e., .integer(x) - 1L. num_classes Total number classes. NULL, inferred max(x) + 1. Defaults NULL.","code":""},{"path":"https://keras3.posit.co/dev/reference/to_categorical.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Converts a class vector (integers) to binary class matrix. — to_categorical","text":"binary matrix representation input R array. class axis placed last.","code":""},{"path":"https://keras3.posit.co/dev/reference/to_categorical.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Converts a class vector (integers) to binary class matrix. — to_categorical","text":"","code":"a <- to_categorical(c(0, 1, 2, 3), num_classes=4) print(a) ##      [,1] [,2] [,3] [,4] ## [1,]    1    0    0    0 ## [2,]    0    1    0    0 ## [3,]    0    0    1    0 ## [4,]    0    0    0    1 b <- array(c(.9, .04, .03, .03,               .3, .45, .15, .13,               .04, .01, .94, .05,               .12, .21, .5, .17),               dim = c(4, 4)) loss <- op_categorical_crossentropy(a, b) loss ## tf.Tensor([0.41284522 0.45601739 0.54430155 0.80437282], shape=(4), dtype=float64) loss <- op_categorical_crossentropy(a, a) loss ## tf.Tensor([1.00000005e-07 1.00000005e-07 1.00000005e-07 1.00000005e-07], shape=(4), dtype=float64)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/train_on_batch.html","id":null,"dir":"Reference","previous_headings":"","what":"Runs a single gradient update on a single batch of data. — train_on_batch","title":"Runs a single gradient update on a single batch of data. — train_on_batch","text":"Runs single gradient update single batch data.","code":""},{"path":"https://keras3.posit.co/dev/reference/train_on_batch.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Runs a single gradient update on a single batch of data. — train_on_batch","text":"","code":"train_on_batch(object, x, y = NULL, sample_weight = NULL, class_weight = NULL)"},{"path":"https://keras3.posit.co/dev/reference/train_on_batch.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Runs a single gradient update on a single batch of data. — train_on_batch","text":"object Keras model object x Input data. Must array-like. y Target data. Must array-like. sample_weight Optional array length x, containing weights apply model's loss sample. case temporal data, can pass 2D array shape (samples, sequence_length), apply different weight every timestep every sample. class_weight Optional named list mapping class indices (integers, 0-based) weight (float) apply model's loss samples class training. can useful tell model \"pay attention\" samples -represented class. class_weight specified targets rank 2 greater, either y must one-hot encoded, explicit final dimension 1 must included sparse class labels.","code":""},{"path":"https://keras3.posit.co/dev/reference/train_on_batch.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Runs a single gradient update on a single batch of data. — train_on_batch","text":"scalar loss value (metrics), named list loss metric values (metrics). property model$metrics_names give display labels scalar outputs.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/use_backend.html","id":null,"dir":"Reference","previous_headings":"","what":"Configure a Keras backend — use_backend","title":"Configure a Keras backend — use_backend","text":"Configure Keras backend","code":""},{"path":"https://keras3.posit.co/dev/reference/use_backend.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Configure a Keras backend — use_backend","text":"","code":"use_backend(backend, gpu = NA)"},{"path":"https://keras3.posit.co/dev/reference/use_backend.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Configure a Keras backend — use_backend","text":"backend string, can \"tensorflow\", \"jax\", \"numpy\", \"torch\". gpu bool, whether use GPU. NA (default), attempt detect GPU availability Linux. M-series Macs, defaults FALSE TensorFlow TRUE JAX. Windows, defaults FALSE.","code":""},{"path":"https://keras3.posit.co/dev/reference/use_backend.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Configure a Keras backend — use_backend","text":"Called primarily side effects. Returns provided backend, invisibly.","code":""},{"path":"https://keras3.posit.co/dev/reference/use_backend.html","id":"details","dir":"Reference","previous_headings":"","what":"Details","title":"Configure a Keras backend — use_backend","text":"functions allow configuring backend keras use. Note one backend can configured time. function called library(keras3) calling functions within package (see example). experimental support changing backend keras initialized. using config_set_backend().","code":"library(keras3) use_backend(\"tensorflow\")"},{"path":"https://keras3.posit.co/dev/reference/with_custom_object_scope.html","id":null,"dir":"Reference","previous_headings":"","what":"Provide a scope with mappings of names to custom objects — with_custom_object_scope","title":"Provide a scope with mappings of names to custom objects — with_custom_object_scope","text":"Provide scope mappings names custom objects","code":""},{"path":"https://keras3.posit.co/dev/reference/with_custom_object_scope.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Provide a scope with mappings of names to custom objects — with_custom_object_scope","text":"","code":"with_custom_object_scope(objects, expr)"},{"path":"https://keras3.posit.co/dev/reference/with_custom_object_scope.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Provide a scope with mappings of names to custom objects — with_custom_object_scope","text":"objects Named list objects expr Expression evaluate","code":""},{"path":"https://keras3.posit.co/dev/reference/with_custom_object_scope.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Provide a scope with mappings of names to custom objects — with_custom_object_scope","text":"result evaluating expr within custom object scope.","code":""},{"path":"https://keras3.posit.co/dev/reference/with_custom_object_scope.html","id":"details","dir":"Reference","previous_headings":"","what":"Details","title":"Provide a scope with mappings of names to custom objects — with_custom_object_scope","text":"many elements Keras models can customized user objects (e.g. losses, metrics, regularizers, etc.). loading saved models use functions typically need explicitly map names user objects via custom_objects parameter. with_custom_object_scope() function provides alternative lets create named alias user object applies entire block code, automatically recognized loading saved models.","code":""},{"path":"https://keras3.posit.co/dev/reference/with_custom_object_scope.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Provide a scope with mappings of names to custom objects — with_custom_object_scope","text":"","code":"# define custom metric metric_top_3_categorical_accuracy <-   custom_metric(\"top_3_categorical_accuracy\", function(y_true, y_pred) {     metric_top_k_categorical_accuracy(y_true, y_pred, k = 3)   })  with_custom_object_scope(c(top_k_acc = sparse_top_k_cat_acc), {    # ...define model...    # compile model (refer to \"top_k_acc\" by name)   model |> compile(     loss = \"binary_crossentropy\",     optimizer = optimizer_nadam(),     metrics = c(\"top_k_acc\")   )    # save the model   model |> save_model(\"my_model.keras\")    # loading the model within the custom object scope doesn't   # require explicitly providing the custom_object   reloaded_model <- load_model(\"my_model.keras\") })"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/zip_lists.html","id":null,"dir":"Reference","previous_headings":"","what":"Zip lists — zip_lists","title":"Zip lists — zip_lists","text":"conceptually similar zip() Python, R functions purrr::transpose() data.table::transpose() (albeit, accepting elements ... instead single list), one crucial difference: provided objects named, matching done names, positions.","code":""},{"path":"https://keras3.posit.co/dev/reference/zip_lists.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Zip lists — zip_lists","text":"","code":"zip_lists(...)"},{"path":"https://keras3.posit.co/dev/reference/zip_lists.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Zip lists — zip_lists","text":"... R lists atomic vectors, optionally named.","code":""},{"path":"https://keras3.posit.co/dev/reference/zip_lists.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Zip lists — zip_lists","text":"inverted list","code":""},{"path":"https://keras3.posit.co/dev/reference/zip_lists.html","id":"details","dir":"Reference","previous_headings":"","what":"Details","title":"Zip lists — zip_lists","text":"arguments supplied must length. positional matching required, arguments provided must unnamed. matching names, arguments must set names, can different orders.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/zip_lists.html","id":"ref-examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Zip lists — zip_lists","text":"","code":"gradients <- list(\"grad_for_wt_1\", \"grad_for_wt_2\", \"grad_for_wt_3\") weights <- list(\"weight_1\", \"weight_2\", \"weight_3\") str(zip_lists(gradients, weights)) #> List of 3 #>  $ :List of 2 #>   ..$ : chr \"grad_for_wt_1\" #>   ..$ : chr \"weight_1\" #>  $ :List of 2 #>   ..$ : chr \"grad_for_wt_2\" #>   ..$ : chr \"weight_2\" #>  $ :List of 2 #>   ..$ : chr \"grad_for_wt_3\" #>   ..$ : chr \"weight_3\" str(zip_lists(gradient = gradients, weight = weights)) #> List of 3 #>  $ :List of 2 #>   ..$ gradient: chr \"grad_for_wt_1\" #>   ..$ weight  : chr \"weight_1\" #>  $ :List of 2 #>   ..$ gradient: chr \"grad_for_wt_2\" #>   ..$ weight  : chr \"weight_2\" #>  $ :List of 2 #>   ..$ gradient: chr \"grad_for_wt_3\" #>   ..$ weight  : chr \"weight_3\"  names(gradients) <- names(weights) <- paste0(\"layer_\", 1:3) str(zip_lists(gradients, weights[c(3, 1, 2)])) #> List of 3 #>  $ layer_1:List of 2 #>   ..$ : chr \"grad_for_wt_1\" #>   ..$ : chr \"weight_1\" #>  $ layer_2:List of 2 #>   ..$ : chr \"grad_for_wt_2\" #>   ..$ : chr \"weight_2\" #>  $ layer_3:List of 2 #>   ..$ : chr \"grad_for_wt_3\" #>   ..$ : chr \"weight_3\"  names(gradients) <- paste0(\"gradient_\", 1:3) try(zip_lists(gradients, weights)) # error, names don't match #> Error in zip_lists(gradients, weights) :  #>   All names of arguments provided to `zip_lists()` must match. Call `unname()` on each argument if you want positional matching # call unname directly for positional matching str(zip_lists(unname(gradients), unname(weights))) #> List of 3 #>  $ :List of 2 #>   ..$ : chr \"grad_for_wt_1\" #>   ..$ : chr \"weight_1\" #>  $ :List of 2 #>   ..$ : chr \"grad_for_wt_2\" #>   ..$ : chr \"weight_2\" #>  $ :List of 2 #>   ..$ : chr \"grad_for_wt_3\" #>   ..$ : chr \"weight_3\""},{"path":"https://keras3.posit.co/dev/news/index.html","id":"keras3-development-version","dir":"Changelog","previous_headings":"","what":"keras3 (development version)","title":"keras3 (development version)","text":"Added S3 methods JAX array: str, .array, .double, .integer, .numeric. Added str S3 method Keras Variables. layer_reshape() can now accept -1 sentinel automatically calculated axis size. Updated dependencies declared use_backend(\"jax\", gpu=TRUE) compatability keras-hub. Added training loop configuration helpers: config_max_epochs(), config_set_max_epochs(), config_max_steps_per_epoch(), config_set_max_steps_per_epoch(). caps can also set via KERAS_MAX_EPOCHS KERAS_MAX_STEPS_PER_EPOCH environment variables. Added config_is_nnx_enabled() check whether JAX NNX features enabled. LoRA-enabled layers (layer_dense(), layer_embedding(), layer_einsum_dense()) gain lora_alpha argument scale adaptation delta independently chosen rank. keras_variable() now accepts synchronization argument distributed strategies. Layer$add_weight() gains overwrite_with_gradient option layers now provide symbolic_call() method. Transposed convolution utilities now follow latest Keras API: op_conv_transpose() defaults strides = 1 layer_conv_*_transpose() layers expose output_padding precise shape control. layer_torch_module_wrapper() gains output_shape argument help Keras infer shapes wrapping PyTorch modules. save_model_weights() adds max_shard_size argument split large weight files manageable shards. Added elastic deformation utilities images: layer_random_elastic_transform() lower-level op_image_elastic_transform(). Added loss_categorical_generalized_cross_entropy() training noisy labels. Added Muon optimizer via optimizer_muon(). Added complex-valued helpers: S3 Arg() methods tensors, op_angle(), conversions op_view_as_real() / op_view_as_complex(). Added signal window operations: op_bartlett(), op_blackman(), op_hamming(), op_hanning(), op_kaiser(). Expanded numeric operations op_layer_normalization(), op_cbrt(), op_corrcoef(), op_deg2rad(), op_heaviside(), new op_sparse_sigmoid() plus matching activation_sparse_sigmoid(), attn_logits_soft_cap argument op_dot_product_attention(). layer_layer_normalization() removes rms_scaling argument.","code":""},{"path":"https://keras3.posit.co/dev/news/index.html","id":"keras3-140","dir":"Changelog","previous_headings":"","what":"keras3 1.4.0","title":"keras3 1.4.0","text":"CRAN release: 2025-05-04 New op_subset() x@r[...] methods enable tensor subsetting using R’s [ semantics idioms. New subset assignment methods implemented tensors: op_subset(x, ...) <- value x@r[...] <- value Breaking changes: operations prefixed op_ now return 1-based indices default. following functions return consume indices changed: op_argmax(), op_argmin(), op_top_k(), op_argpartition(), op_searchsorted(), op_argsort(), op_digitize(), op_nonzero(), op_split(), op_trace(), op_swapaxes(), op_ctc_decode(), op_ctc_loss(), op_one_hot(), op_arange() op_arange() now matches semantics base::seq(). default starts, includes end value, automatically infers step direction. op_one_hot() now infers num_classes supplied factor. op_hstack() op_vstack() now accept arguments passed via .... application_decode_predictions() now returns processed data frame default decoder function predictions missing. application_preprocess_inputs() returns preprocessor function inputs missing. Various new examples added documentation, including op_scatter(), op_switch(), op_nonzero(). New x@py[...] accessor introduced Python-style 0-based indexing tensors. New Summary group generic method keras_shape, enabling usage like prod(shape(3, 4)) KERAS_HOME now set tools::R_user_dir(\"keras3\", \"cache\") ~/.keras exist KERAS_HOME unset. new op_convert_to_array() convert tensor R array. Added compatibility Keras v3.9.2. New operations added: op_rot90() op_rearrange() (Einops-style) op_signbit() op_polar() op_image_perspective_transform() op_image_gaussian_blur() New layers introduced: layer_rms_normalization() layer_aug_mix() layer_cut_mix() layer_random_invert() layer_random_erasing() layer_random_gaussian_blur() layer_random_perspective() layer_resizing() gains antialias argument. keras_input(), keras_model_sequential(), op_convert_to_tensor() gain ragged argument. layer$pop_layer() gains rebuild argument now returns removed layer. New rematerialized_call() method added Layer objects. Documentation improvements minor fixes. Fixed issue op_shape() sometimes return TensorFlow TensorShape Fixes metric_iou(), op_top_k(), op_eye() called R atomic doubles","code":""},{"path":"https://keras3.posit.co/dev/news/index.html","id":"keras3-130","dir":"Changelog","previous_headings":"","what":"keras3 1.3.0","title":"keras3 1.3.0","text":"CRAN release: 2025-03-03 Keras now uses reticulate::py_require() resolve Python dependencies. Calling install_keras() longer required (still supported). use_backend() gains gpu argument, specify GPU-capable set dependencies resolved py_require(). progress bar fit(), evaluate() predict() now defaults presenting testthat tests. dotty::. now reexported. %*% now dispatches op_matmul() tensorflow tensors, relaxed shape constraints compared tf$matmul(). Fixed issue calling Metric Loss object unnamed arguments error.","code":""},{"path":"https://keras3.posit.co/dev/news/index.html","id":"added-compatibility-with-keras-v380-user-facing-changes-1-3-0","dir":"Changelog","previous_headings":"","what":"Added compatibility with Keras v3.8.0. User-facing changes:","title":"keras3 1.3.0","text":"activation_sparse_plus() activation_sparsemax() activation_threshold() layer_equalization() layer_mix_up() layer_rand_augment() layer_random_color_degeneration() layer_random_color_jitter() layer_random_grayscale() layer_random_hue() layer_random_posterization() layer_random_saturation() layer_random_sharpness() layer_random_shear() op_diagflat() op_sparse_plus() op_sparsemax() op_threshold() op_unravel_index() Add argument axis tversky loss New: ONNX model export export_savedmodel() Doc improvements bug fixes. JAX specific changes: Add support JAX named scope TensorFlow specific changes: Make random_shuffle() XLA compilable","code":""},{"path":[]},{"path":[]},{"path":"https://keras3.posit.co/dev/news/index.html","id":"activations-1-3-0","dir":"Changelog","previous_headings":"Added compatibility with Keras v3.7.0. User-facing changes: > New functions","what":"Activations","title":"keras3 1.3.0","text":"activation_celu() activation_glu() activation_hard_shrink() activation_hard_tanh() activation_log_sigmoid() activation_soft_shrink() activation_squareplus() activation_tanh_shrink()","code":""},{"path":"https://keras3.posit.co/dev/news/index.html","id":"configuration-1-3-0","dir":"Changelog","previous_headings":"Added compatibility with Keras v3.7.0. User-facing changes: > New functions","what":"Configuration","title":"keras3 1.3.0","text":"config_disable_flash_attention() config_enable_flash_attention() config_is_flash_attention_enabled()","code":""},{"path":"https://keras3.posit.co/dev/news/index.html","id":"layers-and-initializers-1-3-0","dir":"Changelog","previous_headings":"Added compatibility with Keras v3.7.0. User-facing changes: > New functions","what":"Layers and Initializers","title":"keras3 1.3.0","text":"initializer_stft() layer_max_num_bounding_boxes() layer_stft_spectrogram()","code":""},{"path":"https://keras3.posit.co/dev/news/index.html","id":"losses-and-metrics-1-3-0","dir":"Changelog","previous_headings":"Added compatibility with Keras v3.7.0. User-facing changes: > New functions","what":"Losses and Metrics","title":"keras3 1.3.0","text":"loss_circle() metric_concordance_correlation() metric_pearson_correlation()","code":""},{"path":"https://keras3.posit.co/dev/news/index.html","id":"operations-1-3-0","dir":"Changelog","previous_headings":"Added compatibility with Keras v3.7.0. User-facing changes: > New functions","what":"Operations","title":"keras3 1.3.0","text":"op_celu() op_exp2() op_glu() op_hard_shrink() op_hard_tanh() op_ifft2() op_inner() op_soft_shrink() op_squareplus() op_tanh_shrink()","code":""},{"path":"https://keras3.posit.co/dev/news/index.html","id":"new-arguments-1-3-0","dir":"Changelog","previous_headings":"Added compatibility with Keras v3.7.0. User-facing changes: > New functions","what":"New arguments","title":"keras3 1.3.0","text":"callback_backup_and_restore(): Added double_checkpoint argument save fallback checkpoint callback_tensorboard(): Added support profile_batch argument layer_group_query_attention(): Added flash_attention seed arguments layer_multi_head_attention(): Added flash_attention argument metric_sparse_top_k_categorical_accuracy(): Added from_sorted_ids argument","code":""},{"path":"https://keras3.posit.co/dev/news/index.html","id":"performance-improvements-1-3-0","dir":"Changelog","previous_headings":"Added compatibility with Keras v3.7.0. User-facing changes:","what":"Performance improvements","title":"keras3 1.3.0","text":"Added native Flash Attention support GPU (via cuDNN) TPU (via Pallas kernel) JAX backend Added opt-native Flash Attention support GPU PyTorch backend Enabled additional kernel fusion via bias_add TensorFlow backend Added support Intel XPU devices PyTorch backend install_keras() changes: GPU available, default now install CPU build TensorFlow GPU build JAX. use GPU current session, call use_backend(\"jax\").","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/news/index.html","id":"breaking-changes-1-3-0","dir":"Changelog","previous_headings":"Added compatibility with Keras v3.6.0. User-facing changes:","what":"Breaking changes:","title":"keras3 1.3.0","text":"using get_file() extract = TRUE untar = TRUE, return value now path extracted directory, rather path archive.","code":""},{"path":"https://keras3.posit.co/dev/news/index.html","id":"other-changes-and-additions-1-3-0","dir":"Changelog","previous_headings":"Added compatibility with Keras v3.6.0. User-facing changes:","what":"Other changes and additions:","title":"keras3 1.3.0","text":"Logging now asynchronous fit(), evaluate(), predict(). enables 100% compact stacking train_step calls accelerators (e.g. running small models TPU). using custom callbacks rely on_batch_end, disable async logging. can re-enable adding self$async_safe <- TRUE callbacks. Note TensorBoard callback considered async-safe default. Default callbacks like progress bar async-safe. New bitwise operations: op_bitwise_and() op_bitwise_invert() op_bitwise_left_shift() op_bitwise_not() op_bitwise_or() op_bitwise_right_shift() op_bitwise_xor() New math operations: op_logdet() op_trunc() op_histogram() New neural network operation: op_dot_product_attention() New image preprocessing layers: layer_auto_contrast() layer_solarization() New Model functions get_state_tree() set_state_tree(), retrieving model variables, including trainable, non-trainable, optimizer variables, metric variables. New layer_pipeline() composing sequence layers. class useful building preprocessing pipeline. Compared keras_model_sequential(), layer_pipeline() key differences: ’s Model, just plain layer. layers pipeline compatible tf.data, pipeline also remain tf.data compatible, regardless backend use. New argument: export_savedmodel(verbose = ) New argument: op_normalize(epsilon = ) Various documentation improvements bug fixes.","code":""},{"path":"https://keras3.posit.co/dev/news/index.html","id":"keras3-120","dir":"Changelog","previous_headings":"","what":"keras3 1.2.0","title":"keras3 1.2.0","text":"CRAN release: 2024-09-05 Added compatibility Keras v3.5.0. User facing changes: op_associative_scan() op_searchsorted() optimizer_lamb() keras$DTypePolicy instances can now supplied dtype argument losses, metrics, layers. Add integration Hugging Face Hub. can now save models Hugging Face Hub directly save_model() load .keras models directly Hugging Face Hub load_model(). Added compatibility NumPy 2.0. Improved keras$distribution API support large models. Bug fixes performance improvements. Add data_format argument layer_zero_padding_1d() layer. Miscellaneous documentation improvements. Bug fixes performance improvements.","code":""},{"path":"https://keras3.posit.co/dev/news/index.html","id":"keras3-110","dir":"Changelog","previous_headings":"","what":"keras3 1.1.0","title":"keras3 1.1.0","text":"CRAN release: 2024-07-17 Fixed issue GPUs found running Windows WSL Linux. (reported #1456, fixed #1459) keras_shape objects (returned keras3::shape()) gain == != methods. Fixed warning tfruns::training_run() unable log optimizer learning rate. Added compatibility Keras v3.4.1 (R user facing changes). Added compatibility Keras v3.4.0. User facing changes: op_argpartition() op_map() op_scan() op_switch() op_dtype() op_lstsq() op_image_hsv_to_rgb() op_image_rgb_to_hsv() Added support arbitrary, deeply nested input/output structures Functional models (e.g. lists lists lists inputs outputs…) keras_input() gains optional argument. keras_model_sequential() gains input_optional argument. Add support float8 inference Dense EinsumDense layers. Enable layer_feature_space() used tfdatasets pipeline even backend isn’t TensorFlow. layer_string_lookup() can now take tf$SparseTensor() input. layer_string_lookup() returns \"int64\" dtype default modes now. Layer() instances gain attributes path quantization_mode. Metric()$variables now recursive. Add training argument Model$compute_loss(). split_dataset() now supports nested structures dataset. applications gain name argument, accept custom name. layer_multi_head_attention() gains seed argument. losses gain dtype argument. loss_dice() gains axis argument. op_ctc_decode(), new default mask_index = 0 op_image_* functions now use default data_format value config_image_data_format() op_isclose() gains arguments rtol, atol, equal_nan. save_model() gains argument zipped. Bugs fixes performance improvements.","code":""},{"path":"https://keras3.posit.co/dev/news/index.html","id":"keras3-100","dir":"Changelog","previous_headings":"","what":"keras3 1.0.0","title":"keras3 1.0.0","text":"CRAN release: 2024-05-21 Chains layer_* calls |> now instantiate layers order %>% pipe chains: left-hand-side first (#1440). iterate(), iter_next() as_iterator() now reexported reticulate. User facing changes upstream Keras v3.3.3: new functions: op_slogdet(), op_psnr() clone_model() gains new args: call_function, recursive Updated example usage. op_ctc_decode() strategy argument new default: \"greedy\". Updated docs. loss_ctc() default name fixed, changed \"ctc\" User facing changes upstream Keras v3.3.2: new function: op_ctc_decode() new function: op_eigh() new function: op_select() new function: op_vectorize() new function: op_image_rgb_to_grayscale() new function: loss_tversky() new args: layer_resizing(pad_to_aspect_ratio, fill_mode, fill_value) new arg: layer_embedding(weights) providing initial weights matrix new args: op_nan_to_num(nan, posinf, neginf) new args: op_image_resize(crop_to_aspect_ratio, pad_to_aspect_ratio, fill_mode, fill_value) new args: op_argmax(keepdims) op_argmin(keepdims) new arg: clear_session(free_memory) clearing without invoking garbage collector. metric_kl_divergence() loss_kl_divergence() clip inputs (y_true y_pred) [0, 1] range. new Layer() attributes: metrics, dtype_policy Added initial support float8 training layer_conv_*d() layers now support LoRa op_digitize() now supports sparse tensors. Models layers now return owned metrics recursively. Add pickling support Keras models. (e.g., via reticulate::py_save_object()) Note pickling recommended, prefer using Keras saving APIs.","code":""},{"path":"https://keras3.posit.co/dev/news/index.html","id":"keras3-020","dir":"Changelog","previous_headings":"","what":"keras3 0.2.0","title":"keras3 0.2.0","text":"CRAN release: 2024-04-18 New functions: quantize_weights(): quantize model layer weights -place. Currently, Dense, EinsumDense, Embedding layers supported (enough cover majority transformers today) layer_mel_spectrogram() layer_flax_module_wrapper() layer_jax_model_wrapper() loss_dice() random_beta() random_binomial() config_set_backend(): change backend Keras initialized. config_dtype_policy() config_set_dtype_policy() New Ops op_custom_gradient() op_batch_normalization() op_image_crop() op_divide_no_nan() op_normalize() op_correlate() ` New family linear algebra ops op_cholesky() op_det() op_eig() op_inv() op_lu_factor() op_norm() op_erfinv() op_solve_triangular() op_svd() audio_dataset_from_directory(), image_dataset_from_directory() text_dataset_from_directory() gain verbose argument (default TRUE) image_dataset_from_directory() gains pad_to_aspect_ratio argument (default FALSE) to_categorical(), op_one_hot(), fit() can now accept R factors, offset 0-based (reported #1055). op_convert_to_numpy() now returns unconverted NumPy arrays. op_array() op_convert_to_tensor() longer error casting R doubles integer types. export_savedmodel() now works Jax backend. Metric()$add_variable() method gains arg: aggregration. Layer()$add_weight() method gains args: autocast, regularizer, aggregation. op_bincount(), op_multi_hot(), op_one_hot(), layer_category_encoding() now support sparse tensors. op_custom_gradient() now supports PyTorch backend layer_lstm() layer_gru() gain arg use_cudnn, default 'auto'. Fixed issue application_preprocess_inputs() error supplied R array input. Doc improvements.","code":""},{"path":"https://keras3.posit.co/dev/news/index.html","id":"keras3-010","dir":"Changelog","previous_headings":"","what":"keras3 0.1.0","title":"keras3 0.1.0","text":"CRAN release: 2024-02-17 package rebuilt Keras 3.0. Refer https://blogs.rstudio.com/ai/posts/2024-05-21-keras3/ overview https://keras3.posit.co current --date documentation.","code":""}]
+[{"path":"https://keras3.posit.co/dev/articles/custom_train_step_in_tensorflow.html","id":"introduction","dir":"Articles","previous_headings":"","what":"Introduction","title":"Customizing what happens in `fit()` with TensorFlow","text":"’re supervised learning, can use fit() everything works smoothly. need take control every little detail, can write training loop entirely scratch. need custom training algorithm, still want benefit convenient features fit(), callbacks, built-distribution support, step fusing? core principle Keras progressive disclosure complexity. always able get lower-level workflows gradual way. shouldn’t fall cliff high-level functionality doesn’t exactly match use case. able gain control small details retaining commensurate amount high-level convenience. need customize fit() , override training step function Model class. function called fit() every batch data. able call fit() usual – running learning algorithm. Note pattern prevent building models Functional API. can whether ’re building Sequential models, Functional API models, subclassed models. Let’s see works.","code":""},{"path":"https://keras3.posit.co/dev/articles/custom_train_step_in_tensorflow.html","id":"setup","dir":"Articles","previous_headings":"","what":"Setup","title":"Customizing what happens in `fit()` with TensorFlow","text":"","code":"library(reticulate) library(tensorflow, exclude = c(\"set_random_seed\", \"shape\")) library(keras3)"},{"path":"https://keras3.posit.co/dev/articles/custom_train_step_in_tensorflow.html","id":"a-first-simple-example","dir":"Articles","previous_headings":"","what":"A first simple example","title":"Customizing what happens in `fit()` with TensorFlow","text":"Let’s start simple example: create new class subclasses Model. just override method train_step(self, data). return dictionary mapping metric names (including loss) current value. input argument data gets passed fit training data: pass arrays, calling fit(x, y, ...), data list (x, y) pass tf_dataset, calling fit(dataset, ...), data gets yielded dataset batch. body train_step() method, implement regular training update, similar already familiar . Importantly, compute loss via self.compute_loss(), wraps loss(es) function(s) passed compile(). Similarly, call metric$update_state(y, y_pred) metrics self$metrics, update state metrics passed compile(), query results self$metrics end retrieve current value. Let’s try :","code":"CustomModel <- new_model_class(   \"CustomModel\",   train_step = function(data) {     c(x, y = NULL, sample_weight = NULL) %<-% data      with(tf$GradientTape() %as% tape, {       y_pred <- self(x, training = TRUE)       loss <- self$compute_loss(y = y, y_pred = y_pred,                                 sample_weight = sample_weight)     })      # Compute gradients     trainable_vars <- self$trainable_variables     gradients <- tape$gradient(loss, trainable_vars)      # Update weights     self$optimizer$apply(gradients, trainable_vars)      # Update metrics (includes the metric that tracks the loss)     for (metric in self$metrics) {       if (metric$name == \"loss\")         metric$update_state(loss)       else         metric$update_state(y, y_pred)     }      # Return a dict mapping metric names to current value     metrics <- lapply(self$metrics, function(m) m$result())     metrics <- setNames(metrics, sapply(self$metrics, function(m) m$name))     metrics   } ) # Construct and compile an instance of CustomModel inputs <- keras_input(shape = 32) outputs <- layer_dense(inputs, 1) model <- CustomModel(inputs, outputs) model |> compile(optimizer = \"adam\", loss = \"mse\", metrics = \"mae\")  # Just use `fit` as usual x <- random_normal(c(1000, 32)) y <- random_normal(c(1000, 1)) model |> fit(x, y, epochs = 3) ## Epoch 1/3 ## 32/32 - 1s - 20ms/step - mae: 1.4339 - loss: 3.2271 ## Epoch 2/3 ## 32/32 - 0s - 3ms/step - mae: 1.3605 - loss: 2.9034 ## Epoch 3/3 ## 32/32 - 0s - 2ms/step - mae: 1.2960 - loss: 2.6272"},{"path":"https://keras3.posit.co/dev/articles/custom_train_step_in_tensorflow.html","id":"going-lower-level","dir":"Articles","previous_headings":"","what":"Going lower-level","title":"Customizing what happens in `fit()` with TensorFlow","text":"Naturally, just skip passing loss function compile(), instead everything manually train_step. Likewise metrics. ’s lower-level example, uses compile() configure optimizer: start creating Metric instances track loss MAE score (__init__()). implement custom train_step() updates state metrics (calling update_state() ), query (via result()) return current average value, displayed progress bar pass callback. Note need call reset_states() metrics epoch! Otherwise calling result() return average since start training, whereas usually work per-epoch averages. Thankfully, framework can us: just list metric want reset metrics property model. model call reset_states() object listed beginning fit() epoch beginning call evaluate().","code":"CustomModel <- new_model_class(   \"CustomModel\",   initialize = function(...) {     super$initialize(...)     self$loss_tracker <- metric_mean(name = \"loss\")     self$mae_metric <- metric_mean_absolute_error(name = \"mae\")     self$loss_fn <- loss_mean_squared_error()   },   train_step = function(data) {     c(x, y = NULL, sample_weight = NULL) %<-% data      with(tf$GradientTape() %as% tape, {       y_pred <- self(x, training = TRUE)       loss <- self$loss_fn(y, y_pred, sample_weight = sample_weight)     })      # Compute gradients     trainable_vars <- self$trainable_variables     gradients <- tape$gradient(loss, trainable_vars)      # Update weights     self$optimizer$apply(gradients, trainable_vars)      # Compute our own metrics     self$loss_tracker$update_state(loss)     self$mae_metric$update_state(y, y_pred)      # Return a dict mapping metric names to current value     list(       loss = self$loss_tracker$result(),       mae = self$mae_metric$result()     )   },   metrics = mark_active(function() {     # We list our `Metric` objects here so that `reset_states()` can be     # called automatically at the start of each epoch     # or at the start of `evaluate()`.     list(self$loss_tracker, self$mae_metric)   }) )   # Construct and compile an instance of CustomModel inputs <- keras_input(shape = 32) outputs <- layer_dense(inputs, 1) model <- CustomModel(inputs, outputs)  # We don't pass a loss or metrics here. model |> compile(optimizer = \"adam\")  # Just use `fit` as usual x <- random_normal(c(1000, 32)) y <- random_normal(c(1000, 1)) model |> fit(x, y, epochs = 3) ## Epoch 1/3 ## 32/32 - 1s - 18ms/step - loss: 2.5170 - mae: 1.2923 ## Epoch 2/3 ## 32/32 - 0s - 2ms/step - loss: 2.2689 - mae: 1.2241 ## Epoch 3/3 ## 32/32 - 0s - 2ms/step - loss: 2.0578 - mae: 1.1633"},{"path":"https://keras3.posit.co/dev/articles/custom_train_step_in_tensorflow.html","id":"supporting-sample_weight-class_weight","dir":"Articles","previous_headings":"","what":"Supporting sample_weight & class_weight","title":"Customizing what happens in `fit()` with TensorFlow","text":"may noticed first basic example didn’t make mention sample weighting. want support fit() arguments sample_weight class_weight, ’d simply following: Unpack sample_weight data argument Pass compute_loss & update_state (course, also just apply manually don’t rely compile() losses & metrics) ’s .","code":"CustomModel <- new_model_class(   \"CustomModel\",   train_step = function(data) {     c(x, y = NULL, sample_weight = NULL) %<-% data      with(tf$GradientTape() %as% tape, {       y_pred <- self(x, training = TRUE)       loss <- self$compute_loss(y = y, y_pred = y_pred,                                 sample_weight = sample_weight)     })      # Compute gradients     trainable_vars <- self$trainable_variables     gradients <- tape$gradient(loss, trainable_vars)      # Update weights     self$optimizer$apply_gradients(zip_lists(gradients, trainable_vars))      # Update metrics (includes the metric that tracks the loss)     for (metric in self$metrics) {       if (metric$name == \"loss\") {         metric$update_state(loss)       } else {         metric$update_state(y, y_pred, sample_weight = sample_weight)       }     }      # Return a dict mapping metric names to current value     metrics <- lapply(self$metrics, function(m) m$result())     metrics <- setNames(metrics, sapply(self$metrics, function(m) m$name))     metrics   } )   # Construct and compile an instance of CustomModel inputs <- keras_input(shape = 32) outputs <- layer_dense(inputs, units = 1) model <- CustomModel(inputs, outputs) model |> compile(optimizer = \"adam\", loss = \"mse\", metrics = \"mae\")  # You can now use sample_weight argument x <- random_normal(c(1000, 32)) y <- random_normal(c(1000, 1)) sw <- random_normal(c(1000, 1)) model |> fit(x, y, sample_weight = sw, epochs = 3) ## Epoch 1/3 ## 32/32 - 1s - 20ms/step - mae: 1.3434 - loss: 0.1681 ## Epoch 2/3 ## 32/32 - 0s - 2ms/step - mae: 1.3364 - loss: 0.1394 ## Epoch 3/3 ## 32/32 - 0s - 2ms/step - mae: 1.3286 - loss: 0.1148"},{"path":"https://keras3.posit.co/dev/articles/custom_train_step_in_tensorflow.html","id":"providing-your-own-evaluation-step","dir":"Articles","previous_headings":"","what":"Providing your own evaluation step","title":"Customizing what happens in `fit()` with TensorFlow","text":"want calls model.evaluate()? override test_step exactly way. ’s looks like:","code":"CustomModel <- new_model_class(   \"CustomModel\",   test_step = function(data) {     # Unpack the data     c(x, y = NULL, sw = NULL) %<-% data     # Compute predictions     y_pred = self(x, training = FALSE)     # Updates the metrics tracking the loss     self$compute_loss(y = y, y_pred = y_pred, sample_weight = sw)     # Update the metrics.     for (metric in self$metrics) {       if (metric$name != \"loss\") {         metric$update_state(y, y_pred, sample_weight = sw)       }     }     # Return a dict mapping metric names to current value.     # Note that it will include the loss (tracked in self.metrics).     metrics <- lapply(self$metrics, function(m) m$result())     metrics <- setNames(metrics, sapply(self$metrics, function(m) m$name))     metrics   } )  # Construct an instance of CustomModel inputs <- keras_input(shape = 32) outputs <- layer_dense(inputs, 1) model <- CustomModel(inputs, outputs) model |> compile(loss = \"mse\", metrics = \"mae\")  # Evaluate with our custom test_step x <- random_normal(c(1000, 32)) y <- random_normal(c(1000, 1)) model |> evaluate(x, y) ## 32/32 - 0s - 10ms/step - mae: 1.3871 - loss: 0.0000e+00 ## $loss ## tf.Tensor(0.0, shape=(), dtype=float32) ## ## $compile_metrics ## $compile_metrics$mae ## tf.Tensor(1.3871489, shape=(), dtype=float32)"},{"path":"https://keras3.posit.co/dev/articles/custom_train_step_in_tensorflow.html","id":"wrapping-up-an-end-to-end-gan-example","dir":"Articles","previous_headings":"","what":"Wrapping up: an end-to-end GAN example","title":"Customizing what happens in `fit()` with TensorFlow","text":"Let’s walk end--end example leverages everything just learned. Let’s consider: generator network meant generate 28x28x1 images. discriminator network meant classify 28x28x1 images two classes (“fake” “real”). One optimizer . loss function train discriminator. ’s feature-complete GAN class, overriding compile() use signature, implementing entire GAN algorithm 17 lines train_step: Let’s test-drive : ideas behind deep learning simple, implementation painful?","code":"# Create the discriminator discriminator <-   keras_model_sequential(name = \"discriminator\", input_shape = c(28, 28, 1)) |>    layer_conv_2d(filters = 64, kernel_size = c(3, 3),                 strides = c(2, 2),  padding = \"same\") |>   layer_activation_leaky_relu(negative_slope = 0.2) |>    layer_conv_2d(filters = 128, kernel_size = c(3, 3),                 strides = c(2, 2), padding = \"same\") |>   layer_activation_leaky_relu(negative_slope = 0.2) |>    layer_global_max_pooling_2d() |>   layer_dense(units = 1)   # Create the generator latent_dim <- 128 generator <-   keras_model_sequential(name = \"generator\", input_shape = latent_dim) |>    layer_dense(7 * 7 * 128) |>   layer_activation_leaky_relu(negative_slope = 0.2) |>    layer_reshape(target_shape = c(7, 7, 128)) |>    layer_conv_2d_transpose(filters = 128, kernel_size = c(4, 4),                           strides = c(2, 2), padding = \"same\") |>   layer_activation_leaky_relu(negative_slope = 0.2) |>    layer_conv_2d_transpose(filters = 128, kernel_size = c(4, 4),                           strides = c(2, 2), padding = \"same\") |>   layer_activation_leaky_relu(negative_slope = 0.2) |>    layer_conv_2d(filters = 1, kernel_size = c(7, 7), padding = \"same\",                 activation = \"sigmoid\") GAN <- Model(   classname = \"GAN\",    initialize = function(discriminator, generator, latent_dim, ...) {     super$initialize(...)     self$discriminator <- discriminator     self$generator <- generator     self$latent_dim <- as.integer(latent_dim)     self$d_loss_tracker <- metric_mean(name = \"d_loss\")     self$g_loss_tracker <- metric_mean(name = \"g_loss\")   },    compile = function(d_optimizer, g_optimizer, loss_fn, ...) {     super$compile(...)     self$d_optimizer <- d_optimizer     self$g_optimizer <- g_optimizer     self$loss_fn <- loss_fn   },   metrics = active_property(function() {     list(self$d_loss_tracker, self$g_loss_tracker)   }),    train_step = function(real_images) {      # Sample random points in the latent space     batch_size <- shape(real_images)[[1]]     random_latent_vectors <-       tf$random$normal(shape(batch_size, self$latent_dim))      # Decode them to fake images     generated_images <- self$generator(random_latent_vectors)      # Combine them with real images     combined_images <- op_concatenate(list(generated_images,                                            real_images))      # Assemble labels discriminating real from fake images     labels <- op_concatenate(list(op_ones(c(batch_size, 1)),                                   op_zeros(c(batch_size, 1))))      # Add random noise to the labels - important trick!     labels %<>% `+`(tf$random$uniform(shape(.), maxval = 0.05))      # Train the discriminator     with(tf$GradientTape() %as% tape, {       predictions <- self$discriminator(combined_images)       d_loss <- self$loss_fn(labels, predictions)     })     grads <- tape$gradient(d_loss, self$discriminator$trainable_weights)     self$d_optimizer$apply_gradients(       zip_lists(grads, self$discriminator$trainable_weights))      # Sample random points in the latent space     random_latent_vectors <-       tf$random$normal(shape(batch_size, self$latent_dim))      # Assemble labels that say \"all real images\"     misleading_labels <- op_zeros(c(batch_size, 1))      # Train the generator (note that we should *not* update the weights     # of the discriminator)!     with(tf$GradientTape() %as% tape, {       predictions <- self$discriminator(self$generator(random_latent_vectors))       g_loss <- self$loss_fn(misleading_labels, predictions)     })     grads <- tape$gradient(g_loss, self$generator$trainable_weights)     self$g_optimizer$apply_gradients(       zip_lists(grads, self$generator$trainable_weights))      list(d_loss = d_loss, g_loss = g_loss)   } ) batch_size <- 64 c(c(x_train, .), c(x_test, .)) %<-% dataset_mnist() all_digits <- op_concatenate(list(x_train, x_test)) all_digits <- op_reshape(all_digits, c(-1, 28, 28, 1)) dataset <- all_digits |>   tfdatasets::tensor_slices_dataset() |>   tfdatasets::dataset_map(\\(x) op_cast(x, \"float32\") / 255) |>   tfdatasets::dataset_shuffle(buffer_size = 1024) |>   tfdatasets::dataset_batch(batch_size = batch_size)  gan <- GAN(discriminator = discriminator,            generator = generator,            latent_dim = latent_dim)  gan |> compile(   d_optimizer = optimizer_adam(learning_rate = 0.0003),   g_optimizer = optimizer_adam(learning_rate = 0.0003),   loss_fn = loss_binary_crossentropy(from_logits = TRUE) )  # To limit the execution time, we only train on 100 batches. You can train on # the entire dataset. You will need about 20 epochs to get nice results. gan |> fit(   tfdatasets::dataset_take(dataset, 100),   epochs = 1 ) ## 100/100 - 5s - 51ms/step - d_loss: 0.0000e+00 - g_loss: 0.0000e+00"},{"path":"https://keras3.posit.co/dev/articles/distributed_training_with_tensorflow.html","id":"introduction","dir":"Articles","previous_headings":"","what":"Introduction","title":"Multi-GPU distributed training with TensorFlow","text":"generally two ways distribute computation across multiple devices: Data parallelism, single model gets replicated multiple devices multiple machines. processes different batches data, merge results. exist many variants setup, differ different model replicas merge results, whether stay sync every batch whether loosely coupled, etc. Model parallelism, different parts single model run different devices, processing single batch data together. works best models naturally-parallel architecture, models feature multiple branches. guide focuses data parallelism, particular synchronous data parallelism, different replicas model stay sync batch process. Synchronicity keeps model convergence behavior identical see single-device training. Specifically, guide teaches use tf.distribute API train Keras models multiple GPUs, minimal changes code, multiple GPUs (typically 2 16) installed single machine (single host, multi-device training). common setup researchers small-scale industry workflows.","code":""},{"path":"https://keras3.posit.co/dev/articles/distributed_training_with_tensorflow.html","id":"setup","dir":"Articles","previous_headings":"","what":"Setup","title":"Multi-GPU distributed training with TensorFlow","text":"","code":"library(keras3) library(tensorflow, exclude = c(\"shape\", \"set_random_seed\")) library(tfdatasets, exclude = \"shape\")"},{"path":"https://keras3.posit.co/dev/articles/distributed_training_with_tensorflow.html","id":"single-host-multi-device-synchronous-training","dir":"Articles","previous_headings":"","what":"Single-host, multi-device synchronous training","title":"Multi-GPU distributed training with TensorFlow","text":"setup, one machine several GPUs (typically 2 16). device run copy model (called replica). simplicity, follows, ’ll assume ’re dealing 8 GPUs, loss generality. works step training: current batch data (called global batch) split 8 different sub-batches (called local batches). instance, global batch 512 samples, 8 local batches 64 samples. 8 replicas independently processes local batch: run forward pass, backward pass, outputting gradient weights respect loss model local batch. weight updates originating local gradients efficiently merged across 8 replicas. done end every step, replicas always stay sync. practice, process synchronously updating weights model replicas handled level individual weight variable. done mirrored variable object. use single-host, multi-device synchronous training Keras model, use tf$distribute$MirroredStrategy API. ’s works: Instantiate MirroredStrategy, optionally configuring specific devices want use (default strategy use GPUs available). Use strategy object open scope, within scope, create Keras objects need contain variables. Typically, means creating & compiling model inside distribution scope. cases, first call fit() may also create variables, ’s good idea put fit() call scope well. Train model via fit() usual. Importantly, recommend use tf.data.Dataset objects load data multi-device distributed workflow. Schematically, looks like : ’s simple end--end runnable example:","code":"# Create a MirroredStrategy. strategy <- tf$distribute$MirroredStrategy() cat(sprintf('Number of devices: %d\\n', strategy$num_replicas_in_sync))  # Open a strategy scope. with(startegy$scope(), {   # Everything that creates variables should be under the strategy scope.   # In general this is only model construction & `compile()`.   model <- Model(...)   model |> compile(...)    # Train the model on all available devices.   model |> fit(train_dataset, validation_data=val_dataset, ...)    # Test the model on all available devices.   model |> evaluate(test_dataset) }) get_compiled_model <- function() {   inputs <- keras_input(shape = 784)   outputs <- inputs |>     layer_dense(units = 256, activation = \"relu\") |>     layer_dense(units = 256, activation = \"relu\") |>     layer_dense(units = 10)   model <- keras_model(inputs, outputs)   model |> compile(     optimizer = optimizer_adam(),     loss = loss_sparse_categorical_crossentropy(from_logits = TRUE),     metrics = list(metric_sparse_categorical_accuracy())   )   model }  get_dataset <- function(batch_size = 64) {    c(c(x_train, y_train), c(x_test, y_test)) %<-% dataset_mnist()   x_train <- array_reshape(x_train, c(-1, 784)) / 255   x_test <- array_reshape(x_test, c(-1, 784)) / 255    # Reserve 10,000 samples for validation.   val_i <- sample.int(nrow(x_train), 10000)   x_val <- x_train[val_i,]   y_val <- y_train[val_i]   x_train = x_train[-val_i,]   y_train = y_train[-val_i]    y_train <- array_reshape(y_train, c(-1, 1))   y_val <- array_reshape(y_val, c(-1, 1))   y_test <- array_reshape(y_test, c(-1, 1))    # Prepare the training dataset.   train_dataset <- list(x_train, y_train) |>     lapply(np_array, \"float32\") |>     tensor_slices_dataset() |>     dataset_batch(batch_size)    # Prepare the validation dataset.   val_dataset <- list(x_val, y_val) |>     lapply(np_array, \"float32\") |>     tensor_slices_dataset() |>     dataset_batch(batch_size)    # Prepare the test dataset.   test_dataset <- list(x_test, y_test) |>     lapply(np_array, \"float32\") |>     tensor_slices_dataset() |>     dataset_batch(batch_size)    list(train_dataset, val_dataset, test_dataset) }  # Create a MirroredStrategy. strategy <- tf$distribute$MirroredStrategy() cat(sprintf('Number of devices: %d\\n', strategy$num_replicas_in_sync)) ## Number of devices: 2 # Open a strategy scope. with(strategy$scope(), {   # Everything that creates variables should be under the strategy scope.   # In general this is only model construction & `compile()`.   model <- get_compiled_model()    c(train_dataset, val_dataset, test_dataset) %<-% get_dataset()    # Train the model on all available devices.   model |> fit(train_dataset, epochs = 2, validation_data = val_dataset)    # Test the model on all available devices.   model |> evaluate(test_dataset)  }) ## Epoch 1/2 ## 782/782 - 6s - 7ms/step - loss: 0.2616 - sparse_categorical_accuracy: 0.9232 - val_loss: 0.1510 - val_sparse_categorical_accuracy: 0.9524 ## Epoch 2/2 ## 782/782 - 4s - 5ms/step - loss: 0.1044 - sparse_categorical_accuracy: 0.9682 - val_loss: 0.1004 - val_sparse_categorical_accuracy: 0.9675 ## 157/157 - 0s - 3ms/step - loss: 0.1004 - sparse_categorical_accuracy: 0.9683 ## $loss ## [1] 0.1004203 ## ## $sparse_categorical_accuracy ## [1] 0.9683"},{"path":"https://keras3.posit.co/dev/articles/distributed_training_with_tensorflow.html","id":"using-callbacks-to-ensure-fault-tolerance","dir":"Articles","previous_headings":"","what":"Using callbacks to ensure fault tolerance","title":"Multi-GPU distributed training with TensorFlow","text":"using distributed training, always make sure strategy recover failure (fault tolerance). simplest way handle pass ModelCheckpoint callback fit(), save model regular intervals (e.g. every 100 batches every epoch). can restart training saved model. ’s simple example:","code":"# Prepare a directory to store all the checkpoints. checkpoint_dir <- \"./ckpt\" if (!dir.exists(checkpoint_dir)) {   dir.create(checkpoint_dir) }  make_or_restore_model <- function() {   # Either restore the latest model, or create a fresh one   # if there is no checkpoint available.   checkpoints <- list.files(checkpoint_dir, pattern = \"ckpt-.*\\\\.keras\",                             full.names = TRUE)    if (length(checkpoints) > 0) {     checkpoint_epochs <- as.integer(sub(\"ckpt-([0-9]+)\\\\.keras\", \"\\\\1\",                                         basename(checkpoints)))     latest_checkpoint <- checkpoints[which.max(checkpoint_epochs)]     load_model(latest_checkpoint)   } else {     get_compiled_model()   } }    run_training <- function(epochs = 1) {   # Create a MirroredStrategy.   strategy <- tf$distribute$MirroredStrategy()    # Open a strategy scope and create/restore the model   with(strategy$scope(), {     model <- make_or_restore_model()      callbacks <- list(       # This callback saves a SavedModel every epoch       # We include the current epoch in the folder name.       callback_model_checkpoint(         filepath = paste0(checkpoint_dir, \"/ckpt-{epoch}.keras\"),         save_freq = \"epoch\"       ))      model |> fit(       train_dataset,       epochs = epochs,       callbacks = callbacks,       validation_data = val_dataset,       verbose = 2     )   }) }  # Running the first time creates the model run_training(epochs = 1) ## 782/782 - 5s - 6ms/step - loss: 0.2575 - sparse_categorical_accuracy: 0.9237 - val_loss: 0.1428 - val_sparse_categorical_accuracy: 0.9537 # Calling the same function again will resume from where we left off run_training(epochs = 1) ## 782/782 - 5s - 6ms/step - loss: 0.1011 - sparse_categorical_accuracy: 0.9693 - val_loss: 0.1185 - val_sparse_categorical_accuracy: 0.9602"},{"path":"https://keras3.posit.co/dev/articles/distributed_training_with_tensorflow.html","id":"tfdata-performance-tips","dir":"Articles","previous_headings":"","what":"tf$data performance tips","title":"Multi-GPU distributed training with TensorFlow","text":"distributed training, efficiency load data can often become critical. tips make sure tf$data pipelines run fast possible. Note dataset batching creating dataset, make sure batched global batch size. instance, 8 GPUs capable running batch 64 samples, call use global batch size 512. Calling dataset_cache() call dataset_cache() dataset, data cached running first iteration data. Every subsequent iteration use cached data. cache can memory (default) local file specify. can improve performance : data expected change iteration iteration reading data remote distributed filesystem reading data local disk, data fit memory workflow significantly IO-bound (e.g. reading & decoding image files). Calling dataset_prefetch(buffer_size) almost always call dataset_prefetch(buffer_size) creating dataset. means data pipeline run asynchronously model, new samples preprocessed stored buffer current batch samples used train model. next batch prefetched GPU memory time current batch . ’s !","code":""},{"path":"https://keras3.posit.co/dev/articles/distribution.html","id":"introduction","dir":"Articles","previous_headings":"","what":"Introduction","title":"Distributed training with Keras 3","text":"Keras distribution API new interface designed facilitate distributed deep learning across variety backends like JAX, TensorFlow PyTorch. powerful API introduces suite tools enabling data model parallelism, allowing efficient scaling deep learning models multiple accelerators hosts. Whether leveraging power GPUs TPUs, API provides streamlined approach initializing distributed environments, defining device meshes, orchestrating layout tensors across computational resources. classes like DataParallel ModelParallel, abstracts complexity involved parallel computation, making easier developers accelerate machine learning workflows.","code":""},{"path":"https://keras3.posit.co/dev/articles/distribution.html","id":"how-it-works","dir":"Articles","previous_headings":"","what":"How it works","title":"Distributed training with Keras 3","text":"Keras distribution API provides global programming model allows developers compose applications operate tensors global context (working single device) automatically managing distribution across many devices. API leverages underlying framework (e.g. JAX) distribute program tensors according sharding directives procedure called single program, multiple data (SPMD) expansion. decoupling application sharding directives, API enables running application single device, multiple devices, even multiple clients, preserving global semantics.","code":""},{"path":"https://keras3.posit.co/dev/articles/distribution.html","id":"setup","dir":"Articles","previous_headings":"","what":"Setup","title":"Distributed training with Keras 3","text":"","code":"# This guide assumes there are 8 GPUs available for testing. If you don't have # 8 gpus available locally, you can set the following envvar to # make xla initialize the CPU as 8 devices, to enable local testing Sys.setenv(\"CUDA_VISIBLE_DEVICES\" = \"\") Sys.setenv(\"XLA_FLAGS\" = \"--xla_force_host_platform_device_count=8\") library(keras3)  # The distribution API is only implemented for the JAX backend for now. use_backend(\"jax\", FALSE) jax <- reticulate::import(\"jax\")  library(tfdatasets, exclude = \"shape\") # For dataset input."},{"path":"https://keras3.posit.co/dev/articles/distribution.html","id":"devicemesh-and-tensorlayout","dir":"Articles","previous_headings":"","what":"DeviceMesh and TensorLayout","title":"Distributed training with Keras 3","text":"keras$distribution$DeviceMesh class Keras distribution API represents cluster computational devices configured distributed computation. aligns similar concepts jax.sharding.Mesh tf.dtensor.Mesh, ’s used map physical devices logical mesh structure. TensorLayout class specifies tensors distributed across DeviceMesh, detailing sharding tensors along specified axes correspond names axes DeviceMesh. can find detailed concept explainers TensorFlow DTensor guide.","code":"# Retrieve the local available gpu devices. devices <- jax$devices() # \"gpu\" str(devices) ## List of 8 ##  $ :TFRT_CPU_0 ##  $ :TFRT_CPU_1 ##  $ :TFRT_CPU_2 ##  $ :TFRT_CPU_3 ##  $ :TFRT_CPU_4 ##  $ :TFRT_CPU_5 ##  $ :TFRT_CPU_6 ##  $ :TFRT_CPU_7 # Define a 2x4 device mesh with data and model parallel axes mesh <- keras$distribution$DeviceMesh(   shape = shape(2, 4),   axis_names = list(\"data\", \"model\"),   devices = devices )  # A 2D layout, which describes how a tensor is distributed across the # mesh. The layout can be visualized as a 2D grid with \"model\" as rows and # \"data\" as columns, and it is a [4, 2] grid when it mapped to the physical # devices on the mesh. layout_2d <- keras$distribution$TensorLayout(   axes = c(\"model\", \"data\"),   device_mesh = mesh )  # A 4D layout which could be used for data parallelism of an image input. replicated_layout_4d <- keras$distribution$TensorLayout(   axes = list(\"data\", NULL, NULL, NULL),   device_mesh = mesh )"},{"path":"https://keras3.posit.co/dev/articles/distribution.html","id":"distribution","dir":"Articles","previous_headings":"","what":"Distribution","title":"Distributed training with Keras 3","text":"Distribution class Keras serves foundational abstract class designed developing custom distribution strategies. encapsulates core logic needed distribute model’s variables, input data, intermediate computations across device mesh. end user, won’t interact directly class, subclasses like DataParallel ModelParallel.","code":""},{"path":"https://keras3.posit.co/dev/articles/distribution.html","id":"dataparallel","dir":"Articles","previous_headings":"","what":"DataParallel","title":"Distributed training with Keras 3","text":"DataParallel class Keras distribution API designed data parallelism strategy distributed training, model weights replicated across devices DeviceMesh, device processes portion input data. sample usage class.","code":"# Create DataParallel with list of devices. # As a shortcut, the devices can be skipped, # and Keras will detect all local available devices. # E.g. data_parallel <- DataParallel() data_parallel <- keras$distribution$DataParallel(devices = devices)  # Or you can choose to create DataParallel with a 1D `DeviceMesh`. mesh_1d <- keras$distribution$DeviceMesh(   shape = shape(8),   axis_names = list(\"data\"),   devices = devices ) data_parallel <- keras$distribution$DataParallel(device_mesh = mesh_1d)  inputs <- random_normal(c(128, 28, 28, 1)) labels <- random_normal(c(128, 10)) dataset <- tensor_slices_dataset(c(inputs, labels)) |>   dataset_batch(16)  # Set the global distribution. keras$distribution$set_distribution(data_parallel)  # Note that all the model weights from here on are replicated to # all the devices of the `DeviceMesh`. This includes the RNG # state, optimizer states, metrics, etc. The dataset fed into `model |> fit()` or # `model |> evaluate()` will be split evenly on the batch dimension, and sent to # all the devices. You don't have to do any manual aggregation of losses, # since all the computation happens in a global context. inputs <- keras_input(shape = c(28, 28, 1)) outputs <- inputs |>   layer_flatten() |>   layer_dense(units = 200, use_bias = FALSE, activation = \"relu\") |>   layer_dropout(0.4) |>   layer_dense(units = 10, activation = \"softmax\")  model <- keras_model(inputs = inputs, outputs = outputs)  model |> compile(loss = \"mse\") model |> fit(dataset, epochs = 3) ## Epoch 1/3 ## 8/8 - 0s - 42ms/step - loss: 1.1575 ## Epoch 2/3 ## 8/8 - 0s - 4ms/step - loss: 1.0557 ## Epoch 3/3 ## 8/8 - 0s - 4ms/step - loss: 1.0179 model |> evaluate(dataset) ## 8/8 - 0s - 9ms/step - loss: 0.9707 ## $loss ## [1] 0.9707004"},{"path":"https://keras3.posit.co/dev/articles/distribution.html","id":"modelparallel-and-layoutmap","dir":"Articles","previous_headings":"","what":"ModelParallel and LayoutMap","title":"Distributed training with Keras 3","text":"ModelParallel mostly useful model weights large fit single accelerator. setting allows spit model weights activation tensors across devices DeviceMesh, enable horizontal scaling large models. Unlike DataParallel model weights fully replicated, weights layout ModelParallel usually need customization best performances. introduce LayoutMap let specify TensorLayout weights intermediate tensors global perspective. LayoutMap dict-like object maps string TensorLayout instances. behaves differently normal dict string key treated regex retrieving value. class allows define naming schema TensorLayout retrieve corresponding TensorLayout instance. Typically, key used query variable$path attribute, identifier variable. shortcut, list axis names also allowed inserting value, converted TensorLayout. LayoutMap can also optionally contain DeviceMesh populate TensorLayout$device_mesh set. retrieving layout key, isn’t exact match, existing keys layout map treated regex matched input key . multiple matches, ValueError raised. matches found, NULL returned. can visualize individual weights sharded also easy change mesh structure tune computation data parallel model parallel. can adjusting shape mesh. changes needed code.","code":"mesh_2d <- keras$distribution$DeviceMesh(   shape = shape(2, 4),   axis_names = c(\"data\", \"model\"),   devices = devices ) layout_map  <- keras$distribution$LayoutMap(mesh_2d)  # The rule below means that for any weights that match with d1/kernel, it # will be sharded with model dimensions (4 devices), same for the d1/bias. # All other weights will be fully replicated. layout_map[\"d1/kernel\"] <- tuple(NULL, \"model\") layout_map[\"d1/bias\"] <- tuple(\"model\")  # You can also set the layout for the layer output like layout_map[\"d2/output\"] <- tuple(\"data\", NULL)  model_parallel <- keras$distribution$ModelParallel(   layout_map = layout_map, batch_dim_name = \"data\" )  keras$distribution$set_distribution(model_parallel)  inputs <- layer_input(shape = c(28, 28, 1)) outputs <- inputs |>   layer_flatten() |>   layer_dense(units = 200, use_bias = FALSE,               activation = \"relu\", name = \"d1\") |>   layer_dropout(0.4) |>   layer_dense(units = 10,               activation = \"softmax\",               name = \"d2\")  model <- keras_model(inputs = inputs, outputs = outputs) d1 <- get_layer(model, \"d1\") d1$kernel$value |> jax$debug$visualize_array_sharding() ## ┌───────┬───────┬───────┬───────┐ ## │       │       │       │       │ ## │       │       │       │       │ ## │       │       │       │       │ ## │       │       │       │       │ ## │CPU 0,4│CPU 1,5│CPU 2,6│CPU 3,7│ ## │       │       │       │       │ ## │       │       │       │       │ ## │       │       │       │       │ ## │       │       │       │       │ ## └───────┴───────┴───────┴───────┘ d2 <- get_layer(model, \"d2\") d2$kernel$value |> jax$debug$visualize_array_sharding() ## ┌───────────────────┐ ## │                   │ ## │                   │ ## │                   │ ## │                   │ ## │CPU 0,1,2,3,4,5,6,7│ ## │                   │ ## │                   │ ## │                   │ ## │                   │ ## └───────────────────┘ d2$bias$value |> jax$debug$visualize_array_sharding() ## ┌───────────────────┐ ## │CPU 0,1,2,3,4,5,6,7│ ## └───────────────────┘ x_batch <- dataset |>   as_iterator() |> iter_next() |>   _[[1]] |> op_convert_to_tensor()  output_array <- model(x_batch) output_array |> jax$debug$visualize_array_sharding() ## ┌─────────────┐ ## │             │ ## │ CPU 0,1,2,3 │ ## │             │ ## │             │ ## ├─────────────┤ ## │             │ ## │ CPU 4,5,6,7 │ ## │             │ ## │             │ ## └─────────────┘ # The data will be sharded across the \"data\" dimension of the method, which # has 2 devices. model |> compile(loss = \"mse\") model |> fit(dataset, epochs = 3) ## Epoch 1/3 ## 8/8 - 0s - 48ms/step - loss: 1.1454 ## Epoch 2/3 ## 8/8 - 0s - 2ms/step - loss: 1.0527 ## Epoch 3/3 ## 8/8 - 0s - 2ms/step - loss: 1.0115 model |> evaluate(dataset) ## 8/8 - 0s - 10ms/step - loss: 0.9674 ## $loss ## [1] 0.9673891 full_data_parallel_mesh <- keras$distribution$DeviceMesh(   shape = shape(8, 1),   axis_names = list(\"data\", \"model\"),   devices = devices ) more_data_parallel_mesh <- keras$distribution$DeviceMesh(   shape = shape(4, 2),   axis_names = list(\"data\", \"model\"),   devices = devices ) more_model_parallel_mesh <- keras$distribution$DeviceMesh(   shape = shape(2, 4),   axis_names = list(\"data\", \"model\"),   devices = devices ) full_model_parallel_mesh <- keras$distribution$DeviceMesh(   shape = shape(1, 8),   axis_names = list(\"data\", \"model\"),   devices = devices )"},{"path":"https://keras3.posit.co/dev/articles/distribution.html","id":"further-reading","dir":"Articles","previous_headings":"ModelParallel and LayoutMap","what":"Further reading","title":"Distributed training with Keras 3","text":"JAX Distributed arrays automatic parallelization JAX sharding module TensorFlow Distributed training DTensors TensorFlow DTensor concepts Using DTensors tf.keras","code":""},{"path":"https://keras3.posit.co/dev/articles/examples/nlp/neural_machine_translation_with_transformer.html","id":"introduction","dir":"Articles > Examples > Nlp","previous_headings":"","what":"Introduction","title":"English-to-Spanish translation with a sequence-to-sequence Transformer","text":"example, ’ll build sequence--sequence Transformer model, ’ll train English--Spanish machine translation task. ’ll learn : Vectorize text using layer_text_vectorization(). Implement layer_transformer_encoder(), layer_transformer_decoder(), layer_positional_embedding(). Prepare data training sequence--sequence model. Use trained model generate translations never-seen-input sentences (sequence--sequence inference). code featured adapted book Deep Learning R, Second Edition (chapter 11: Deep learning text). present example fairly barebones, detailed explanations building block works, well theory behind Transformers, recommend reading book.","code":""},{"path":"https://keras3.posit.co/dev/articles/examples/nlp/neural_machine_translation_with_transformer.html","id":"setup","dir":"Articles > Examples > Nlp","previous_headings":"","what":"Setup","title":"English-to-Spanish translation with a sequence-to-sequence Transformer","text":"","code":"library(glue)  library(tensorflow, exclude = c(\"set_random_seed\", \"shape\")) library(keras3)"},{"path":"https://keras3.posit.co/dev/articles/examples/nlp/neural_machine_translation_with_transformer.html","id":"downloading-the-data","dir":"Articles > Examples > Nlp","previous_headings":"","what":"Downloading the data","title":"English-to-Spanish translation with a sequence-to-sequence Transformer","text":"’ll working English--Spanish translation dataset provided Anki. Let’s download :","code":"zip_path <-   \"http://storage.googleapis.com/download.tensorflow.org/data/spa-eng.zip\" |>   get_file(origin = _, extract = TRUE)  text_path <- fs::path(zip_path, \"spa-eng/spa.txt\")"},{"path":"https://keras3.posit.co/dev/articles/examples/nlp/neural_machine_translation_with_transformer.html","id":"parsing-the-data","dir":"Articles > Examples > Nlp","previous_headings":"","what":"Parsing the data","title":"English-to-Spanish translation with a sequence-to-sequence Transformer","text":"line contains English sentence corresponding Spanish sentence. English sentence source sequence Spanish one target sequence. prepend token \"[start]\" append token \"[end]\" Spanish sentence. ’s sentence pairs look like: Now, let’s split sentence pairs training set, validation set, test set.","code":"text_file <- \"spa-eng/spa.txt\" text_pairs <- text_file %>%   readr::read_tsv(col_names = c(\"english\", \"spanish\"),                   col_types = c(\"cc\")) %>%   within(spanish %<>% paste(\"[start]\", ., \"[end]\")) df <- text_pairs[sample(nrow(text_pairs), 5), ] glue::glue_data(df, r\"(   english: {english}   spanish: {spanish} )\") |> cat(sep = \"\\n\\n\") ## english: I'm staying in Italy. ## spanish: [start] Me estoy quedando en Italia. [end] ## ## english: What's so strange about that? ## spanish: [start] ¿Qué es tan extraño acerca de eso? [end] ## ## english: All of the buses are full. ## spanish: [start] Todos los bondis están llenos. [end] ## ## english: Is this where your mother works? ## spanish: [start] ¿Es aquí donde trabaja tu madre? [end] ## ## english: Take precautions. ## spanish: [start] Ten cuidado. [end] random.shuffle(text_pairs) num_val_samples = int(0.15 * len(text_pairs)) num_train_samples = len(text_pairs) - 2 * num_val_samples train_pairs = text_pairs[:num_train_samples] val_pairs = text_pairs[num_train_samples : num_train_samples + num_val_samples] test_pairs = text_pairs[num_train_samples + num_val_samples :]  print(f\"{len(text_pairs)} total pairs\") print(f\"{len(train_pairs)} training pairs\") print(f\"{len(val_pairs)} validation pairs\") print(f\"{len(test_pairs)} test pairs\") num_test_samples <- num_val_samples <-   round(0.15 * nrow(text_pairs)) num_train_samples <- nrow(text_pairs) - num_val_samples - num_test_samples  pair_group <- sample(c(   rep(\"train\", num_train_samples),   rep(\"test\", num_test_samples),   rep(\"val\", num_val_samples) ))  train_pairs <- text_pairs[pair_group == \"train\", ] test_pairs <- text_pairs[pair_group == \"test\", ] val_pairs <- text_pairs[pair_group == \"val\", ] glue(r\"(   {nrow(text_pairs)} total pairs   {nrow(train_pairs)} training pairs   {nrow(val_pairs)} validation pairs   {nrow(test_pairs)} test pairs )\", .transformer = function(text, envir) {   val <- eval(str2lang(text), envir)   prettyNum(val, big.mark = \",\") }) ## 118,493 total pairs ## 82,945 training pairs ## 17,774 validation pairs ## 17,774 test pairs"},{"path":"https://keras3.posit.co/dev/articles/examples/nlp/neural_machine_translation_with_transformer.html","id":"vectorizing-the-text-data","dir":"Articles > Examples > Nlp","previous_headings":"","what":"Vectorizing the text data","title":"English-to-Spanish translation with a sequence-to-sequence Transformer","text":"’ll use two instances layer_text_vectorization() vectorize text data (one English one Spanish), say, turn original strings integer sequences integer represents index word vocabulary. English layer use default string standardization (strip punctuation characters) splitting scheme (split whitespace), Spanish layer use custom standardization, add character \"¿\" set punctuation characters stripped. Note: production-grade machine translation model, recommend stripping punctuation characters either language. Instead, recommend turning punctuation character token, achieve providing custom split function layer_text_vectorization(). Next, ’ll format datasets. training step, model seek predict target words N+1 (beyond) using source sentence target words 1 N. , training dataset yield tuple (inputs, targets), : inputs dictionary (named list) keys (names) encoder_inputs decoder_inputs. encoder_inputs vectorized source sentence decoder_inputs target sentence “far”, say, words 0 N used predict word N+1 (beyond) target sentence. target target sentence offset one step: provides next words target sentence – model try predict. Let’s take quick look sequence shapes (batches 64 pairs, sequences 20 steps long):","code":"punctuation_regex <- \"[¡¿]|[^[:^punct:][\\\\]]\" # the regex explained: Match ¡, or ¿, or any punctuation character except ] # # [:^punct:]: is a negated POSIX character class. # [:punct:] matches any punctuation character, so [:^punct:] matches any # character that is not a punctuation character. # [^...] negates the whole character class # So [^[:^punct:]] would matche any character that is a punctuation character. # Putting this all together, [^[:^punct:][\\\\]] matches any # punctuation character except the ] character.  custom_standardization <- function(input_string) {   input_string %>%     tf$strings$lower() %>%     tf$strings$regex_replace(punctuation_regex, \"\") }  input_string <- as_tensor(\"[start] ¡corre! [end]\") custom_standardization(input_string) ## tf.Tensor(b'[start] corre [end]', shape=(), dtype=string) vocab_size <- 15000 sequence_length <- 20  # rename to eng_vectorization eng_vectorization <- layer_text_vectorization(   max_tokens = vocab_size,   output_mode = \"int\",   output_sequence_length = sequence_length )  spa_vectorization <- layer_text_vectorization(   max_tokens = vocab_size,   output_mode = \"int\",   output_sequence_length = sequence_length + 1,   standardize = custom_standardization )  adapt(eng_vectorization, train_pairs$english) adapt(spa_vectorization, train_pairs$spanish) format_pair <- function(pair) {   eng <- pair$english |> eng_vectorization()   spa <- pair$spanish |> spa_vectorization()    spa_feature <- spa@r[NA:-2]                                                   # <1>   spa_target <- spa@r[2:NA]                                                     # <2>    features <- list(encoder_inputs = eng, decoder_inputs = spa_feature)   labels <- spa_target   sample_weight <- labels != 0    tuple(features, labels, sample_weight) }  batch_size <- 64  library(tfdatasets, exclude = \"shape\") make_dataset <- function(pairs) {   tensor_slices_dataset(pairs) |>     dataset_map(format_pair, num_parallel_calls = 4) |>     dataset_cache() |>     dataset_shuffle(2048) |>     dataset_batch(batch_size) |>     dataset_prefetch(16) }  train_ds <- make_dataset(train_pairs) val_ds <- make_dataset(val_pairs) c(inputs, targets, weights) %<-% iter_next(as_iterator(train_ds)) str(inputs) ## List of 2 ##  $ encoder_inputs:<tf.Tensor: shape=(64, 20), dtype=int64, numpy=…> ##  $ decoder_inputs:<tf.Tensor: shape=(64, 20), dtype=int64, numpy=…> str(targets) ## <tf.Tensor: shape=(64, 20), dtype=int64, numpy=…>"},{"path":"https://keras3.posit.co/dev/articles/examples/nlp/neural_machine_translation_with_transformer.html","id":"building-the-model","dir":"Articles > Examples > Nlp","previous_headings":"","what":"Building the model","title":"English-to-Spanish translation with a sequence-to-sequence Transformer","text":"sequence--sequence Transformer consists TransformerEncoder TransformerDecoder chained together. make model aware word order, also use PositionalEmbedding layer. source sequence pass TransformerEncoder, produce new representation . new representation passed TransformerDecoder, together target sequence far (target words 1 N). TransformerDecoder seek predict next words target sequence (N+1 beyond). key detail makes possible causal masking (see method get_causal_attention_mask() TransformerDecoder). TransformerDecoder sees entire sequences , thus must make sure uses information target tokens 0 N predicting token N+1 (otherwise, use information future, result model used inference time). Next, assemble end--end model.","code":"layer_transformer_encoder <- Layer(   classname = \"TransformerEncoder\",   initialize = function(embed_dim, dense_dim, num_heads, ...) {     super$initialize(...)     self$embed_dim <- embed_dim     self$dense_dim <- dense_dim     self$num_heads <- num_heads     self$attention <-       layer_multi_head_attention(num_heads = num_heads,                                  key_dim = embed_dim)      self$dense_proj <- keras_model_sequential() %>%       layer_dense(dense_dim, activation = \"relu\") %>%       layer_dense(embed_dim)      self$layernorm_1 <- layer_layer_normalization()     self$layernorm_2 <- layer_layer_normalization()     self$supports_masking <- TRUE   },    call = function(inputs, mask = NULL) {     if (!is.null(mask))       mask <- mask[, NULL, ] |> op_cast(\"int32\")      inputs %>%       { self$attention(., ., attention_mask = mask) + . } %>%       self$layernorm_1() %>%       { self$dense_proj(.) + . } %>%       self$layernorm_2()   },    get_config = function() {     config <- super$get_config()     for(name in c(\"embed_dim\", \"num_heads\", \"dense_dim\"))       config[[name]] <- self[[name]]     config   } )  layer_transformer_decoder <- Layer(   classname = \"TransformerDecoder\",    initialize = function(embed_dim, latent_dim, num_heads, ...) {     super$initialize(...)     self$embed_dim <- embed_dim     self$latent_dim <- latent_dim     self$num_heads <- num_heads     self$attention_1 <- layer_multi_head_attention(num_heads = num_heads,                                                    key_dim = embed_dim)     self$attention_2 <- layer_multi_head_attention(num_heads = num_heads,                                                    key_dim = embed_dim)     self$dense_proj <- keras_model_sequential() %>%       layer_dense(latent_dim, activation = \"relu\") %>%       layer_dense(embed_dim)      self$layernorm_1 <- layer_layer_normalization()     self$layernorm_2 <- layer_layer_normalization()     self$layernorm_3 <- layer_layer_normalization()     self$supports_masking <- TRUE   },    get_config = function() {     config <- super$get_config()     for (name in c(\"embed_dim\", \"num_heads\", \"latent_dim\"))       config[[name]] <- self[[name]]     config   },     get_causal_attention_mask = function(inputs) {     c(batch_size, sequence_length, encoding_length) %<-% op_shape(inputs)      x <- op_arange(0L, sequence_length, include_end = FALSE)     i <- x[, NULL]     j <- x[NULL, ]     mask <- op_cast(i >= j, \"int32\")      repeats <- op_stack(c(batch_size, 1L, 1L))     op_tile(mask[NULL, , ], repeats)   },    call = function(inputs, mask = NULL) {     c(inputs_seq, encoder_outputs) %<-% inputs     causal_mask <- self$get_causal_attention_mask(inputs_seq)      if (is.null(mask)) {       inputs_padding_mask <- NULL       encoder_outputs_padding_mask <- NULL     } else {       c(inputs_padding_mask, encoder_outputs_padding_mask) %<-% mask     }      attention_output_1 <- self$attention_1(       query = inputs_seq,       value = inputs_seq,       key = inputs_seq,       attention_mask = causal_mask,       query_mask = inputs_padding_mask     )     out_1 <- self$layernorm_1(inputs_seq + attention_output_1)      attention_output_2 <- self$attention_2(       query = out_1,       value = encoder_outputs,       key = encoder_outputs,       query_mask = inputs_padding_mask,       key_mask = encoder_outputs_padding_mask     )     out_2 <- self$layernorm_2(out_1 + attention_output_2)      self$layernorm_3(out_2 + self$dense_proj(out_2))   } )  layer_positional_embedding <- Layer(   classname = \"PositionalEmbedding\",    initialize = function(sequence_length, vocab_size, embed_dim, ...) {     super$initialize(...)     self$token_embeddings <- layer_embedding(       input_dim = vocab_size, output_dim = embed_dim     )     self$position_embeddings <- layer_embedding(       input_dim = sequence_length, output_dim = embed_dim     )     self$sequence_length <- sequence_length     self$vocab_size <- vocab_size     self$embed_dim <- embed_dim   },    call = function(inputs) {     c(., len) %<-% op_shape(inputs) # (batch_size, seq_len)     positions <- op_arange(0, len, dtype = \"int32\", include_end = FALSE)     embedded_tokens <- self$token_embeddings(inputs)     embedded_positions <- self$position_embeddings(positions)     embedded_tokens + embedded_positions   },    compute_mask = function(inputs, mask = NULL) {     inputs != 0L   },    get_config = function() {     config <- super$get_config()     for(name in c(\"sequence_length\", \"vocab_size\", \"embed_dim\"))       config[[name]] <- self[[name]]     config   } ) embed_dim <- 256 latent_dim <- 2048 num_heads <- 8  encoder_inputs <- layer_input(shape(NA), dtype = \"int64\",                               name = \"encoder_inputs\") encoder_outputs <- encoder_inputs %>%   layer_positional_embedding(sequence_length, vocab_size, embed_dim) %>%   layer_transformer_encoder(embed_dim, latent_dim, num_heads)  encoder <- keras_model(encoder_inputs, encoder_outputs)  decoder_inputs <-  layer_input(shape(NA), dtype = \"int64\",                                name = \"decoder_inputs\") encoded_seq_inputs <- layer_input(shape(NA, embed_dim),                                   name = \"decoder_state_inputs\")  transformer_decoder <- layer_transformer_decoder(NULL,   embed_dim, latent_dim, num_heads)  decoder_outputs <- decoder_inputs %>%   layer_positional_embedding(sequence_length, vocab_size, embed_dim) %>%   { transformer_decoder(list(., encoded_seq_inputs)) } %>%   layer_dropout(0.5) %>%   layer_dense(vocab_size, activation=\"softmax\")  decoder <- keras_model(inputs = list(decoder_inputs, encoded_seq_inputs),                        outputs = decoder_outputs)  decoder_outputs <- decoder(list(decoder_inputs, encoder_outputs))  transformer <- keras_model(   inputs = list(encoder_inputs = encoder_inputs, decoder_inputs = decoder_inputs),   outputs = decoder_outputs,   name = \"transformer\" )"},{"path":"https://keras3.posit.co/dev/articles/examples/nlp/neural_machine_translation_with_transformer.html","id":"training-our-model","dir":"Articles > Examples > Nlp","previous_headings":"","what":"Training our model","title":"English-to-Spanish translation with a sequence-to-sequence Transformer","text":"’ll use accuracy quick way monitor training progress validation data. Note machine translation typically uses BLEU scores well metrics, rather accuracy. train 1 epoch, get model actually converge train least 30 epochs.","code":"epochs <- 1  # This should be at least 30 for convergence  transformer ## Model: \"transformer\" ## ┏━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━┓ ## ┃ Layer (type)        ┃ Output Shape      ┃    Param # ┃ Connected to      ┃ ## ┡━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━┩ ## │ encoder_inputs      │ (None, None)      │          0 │ -                 │ ## │ (InputLayer)        │                   │            │                   │ ## ├─────────────────────┼───────────────────┼────────────┼───────────────────┤ ## │ positional_embeddi… │ (None, None, 256) │  3,845,120 │ encoder_inputs[0… │ ## │ (PositionalEmbeddi… │                   │            │                   │ ## ├─────────────────────┼───────────────────┼────────────┼───────────────────┤ ## │ not_equal           │ (None, None)      │          0 │ encoder_inputs[0… │ ## │ (NotEqual)          │                   │            │                   │ ## ├─────────────────────┼───────────────────┼────────────┼───────────────────┤ ## │ decoder_inputs      │ (None, None)      │          0 │ -                 │ ## │ (InputLayer)        │                   │            │                   │ ## ├─────────────────────┼───────────────────┼────────────┼───────────────────┤ ## │ transformer_encoder │ (None, None, 256) │  3,155,456 │ positional_embed… │ ## │ (TransformerEncode… │                   │            │ not_equal[0][0]   │ ## ├─────────────────────┼───────────────────┼────────────┼───────────────────┤ ## │ functional_3        │ (None, None,      │ 12,959,640 │ decoder_inputs[0… │ ## │ (Functional)        │ 15000)            │            │ transformer_enco… │ ## │                     │                   │            │ not_equal[0][0]   │ ## └─────────────────────┴───────────────────┴────────────┴───────────────────┘ ##  Total params: 19,960,216 (76.14 MB) ##  Trainable params: 19,960,216 (76.14 MB) ##  Non-trainable params: 0 (0.00 B) transformer |> compile(   \"rmsprop\",   loss = loss_sparse_categorical_crossentropy(ignore_class = 0),   metrics = \"accuracy\" )  transformer |> fit(train_ds, epochs = epochs,                    validation_data = val_ds) ## 1297/1297 - 55s - 42ms/step - accuracy: 0.1592 - loss: 3.8833 - val_accuracy: 0.1691 - val_loss: 3.1841"},{"path":"https://keras3.posit.co/dev/articles/examples/nlp/neural_machine_translation_with_transformer.html","id":"decoding-test-sentences","dir":"Articles > Examples > Nlp","previous_headings":"","what":"Decoding test sentences","title":"English-to-Spanish translation with a sequence-to-sequence Transformer","text":"Finally, let’s demonstrate translate brand new English sentences. simply feed model vectorized English sentence well target token \"[start]\", repeatedly generated next token, hit token \"[end]\". 30 epochs, get results :","code":"spa_vocab <- spa_vectorization |> get_vocabulary() max_decoded_sentence_length <- 20 tf_decode_sequence <- tf_function(function(input_sentence) {   withr::local_options(tensorflow.extract.style = \"python\")    tokenized_input_sentence <- input_sentence %>%     as_tensor(shape = c(1, 1)) %>%     eng_vectorization()   spa_vocab <- as_tensor(spa_vocab)   decoded_sentence <- as_tensor(\"[start]\", shape = c(1, 1))    for (i in tf$range(as.integer(max_decoded_sentence_length))) {      tokenized_target_sentence <-       spa_vectorization(decoded_sentence)[,NA:-1]      next_token_predictions <-       transformer(list(         encoder_inputs = tokenized_input_sentence,         decoder_inputs = tokenized_target_sentence       ))      sampled_token_index <- tf$argmax(next_token_predictions[0, i, ])     sampled_token <- spa_vocab[sampled_token_index]     decoded_sentence <-       tf$strings$join(c(decoded_sentence, sampled_token),                       separator = \" \")      if (sampled_token == \"[end]\")       break   }    decoded_sentence  })  for (i in seq(20)) {      c(input_sentence, correct_translation) %<-%       test_pairs[sample.int(nrow(test_pairs), 1), ]     cat(\"-\\n\")     cat(\"English:\", input_sentence, \"\\n\")     cat(\"Correct Translation:\", tolower(correct_translation), \"\\n\")     cat(\"  Model Translation:\", input_sentence %>% as_tensor() %>%           tf_decode_sequence() %>% as.character(), \"\\n\") } English: I'm sure everything will be fine. Correct Translation: [start] estoy segura de que todo irá bien. [end]   Model Translation: [start] estoy seguro de que todo va bien [end]"},{"path":"https://keras3.posit.co/dev/articles/examples/nlp/text_classification_from_scratch.html","id":"introduction","dir":"Articles > Examples > Nlp","previous_headings":"","what":"Introduction","title":"Text classification from scratch","text":"example shows text classification starting raw text (set text files disk). demonstrate workflow IMDB sentiment classification dataset (unprocessed version). use layer_text_vectorization() word splitting & indexing.","code":""},{"path":"https://keras3.posit.co/dev/articles/examples/nlp/text_classification_from_scratch.html","id":"setup","dir":"Articles > Examples > Nlp","previous_headings":"","what":"Setup","title":"Text classification from scratch","text":"","code":"library(tensorflow, exclude = c(\"shape\", \"set_random_seed\")) library(tfdatasets, exclude = \"shape\") library(keras3)"},{"path":"https://keras3.posit.co/dev/articles/examples/nlp/text_classification_from_scratch.html","id":"load-the-data-imdb-movie-review-sentiment-classification","dir":"Articles > Examples > Nlp","previous_headings":"","what":"Load the data: IMDB movie review sentiment classification","title":"Text classification from scratch","text":"Let’s download data inspect structure. aclImdb folder contains train test subfolder: aclImdb/train/pos aclImdb/train/neg folders contain text files, represents one review (either positive negative): interested pos neg subfolders, let’s delete subfolder text files : can use utility text_dataset_from_directory() generate labeled tf_dataset object set text files disk filed class-specific folders. Let’s use generate training, validation, test datasets. validation training datasets generated two subsets train directory, 20% samples going validation dataset 80% going training dataset. validation dataset addition test dataset useful tuning hyperparameters, model architecture, test dataset used. putting model real world however, retrained using available training data (without creating validation dataset), performance maximized. using validation_split subset arguments, make sure either specify random seed, pass shuffle=FALSE, validation & training splits get overlap. Let’s preview samples:","code":"if (!dir.exists(\"datasets/aclImdb\")) {   dir.create(\"datasets\")   download.file(     \"https://ai.stanford.edu/~amaas/data/sentiment/aclImdb_v1.tar.gz\",     \"datasets/aclImdb_v1.tar.gz\"   )   untar(\"datasets/aclImdb_v1.tar.gz\", exdir = \"datasets\")   unlink(\"datasets/aclImdb/train/unsup\", recursive = TRUE) } head(list.files(\"datasets/aclImdb/test\")) ## [1] \"labeledBow.feat\" \"neg\"             \"pos\"             \"urls_neg.txt\" ## [5] \"urls_pos.txt\" head(list.files(\"datasets/aclImdb/train\")) ## [1] \"labeledBow.feat\" \"neg\"             \"pos\"             \"unsupBow.feat\" ## [5] \"urls_neg.txt\"    \"urls_pos.txt\" writeLines(strwrap(readLines(\"datasets/aclImdb/train/pos/4229_10.txt\"))) ## Don't waste time reading my review. Go out and see this ## astonishingly good episode, which may very well be the best Columbo ## ever written! Ruth Gordon is perfectly cast as the scheming yet ## charming mystery writer who murders her son-in-law to avenge his ## murder of her daughter. Columbo is his usual rumpled, befuddled and ## far-cleverer-than-he-seems self, and this particular installment ## features fantastic chemistry between Gordon and Falk. Ironically, ## this was not written by heralded creators Levinson or Link yet is ## possibly the densest, most thoroughly original and twist-laden ## Columbo plot ever. Utterly satisfying in nearly every department ## and overflowing with droll and witty dialogue and thinking. Truly ## unexpected and inventive climax tops all. 10/10...seek this one out ## on Netflix! unlink(\"datasets/aclImdb/train/unsup\", recursive = TRUE) batch_size <- 32  raw_train_ds <- text_dataset_from_directory(   \"datasets/aclImdb/train\",   batch_size = batch_size,   validation_split = 0.2,   subset = \"training\",   seed = 1337 ) ## Found 25000 files belonging to 2 classes. ## Using 20000 files for training. raw_val_ds <- text_dataset_from_directory(   \"datasets/aclImdb/train\",   batch_size = batch_size,   validation_split = 0.2,   subset = \"validation\",   seed = 1337 ) ## Found 25000 files belonging to 2 classes. ## Using 5000 files for validation. raw_test_ds <- text_dataset_from_directory(   \"datasets/aclImdb/test\",   batch_size = batch_size ) ## Found 25000 files belonging to 2 classes. cat(\"Number of batches in raw_train_ds:\", length(raw_train_ds), \"\\n\") ## Number of batches in raw_train_ds: 625 cat(\"Number of batches in raw_val_ds:\", length(raw_val_ds), \"\\n\") ## Number of batches in raw_val_ds: 157 cat(\"Number of batches in raw_test_ds:\", length(raw_test_ds), \"\\n\") ## Number of batches in raw_test_ds: 782 # It's important to take a look at your raw data to ensure your normalization # and tokenization will work as expected. We can do that by taking a few # examples from the training set and looking at them. # This is one of the places where eager execution shines: # we can just evaluate these tensors using .numpy() # instead of needing to evaluate them in a Session/Graph context. batch <- iter_next(as_iterator(raw_train_ds)) str(batch) ## List of 2 ##  $ :<tf.Tensor: shape=(32), dtype=string, numpy=…> ##  $ :<tf.Tensor: shape=(32), dtype=int32, numpy=…> c(text_batch, label_batch) %<-% batch for (i in 1:3) {   print(text_batch[i])   print(label_batch[i]) } ## tf.Tensor(b\"I have read the novel Reaper of Ben Mezrich a fews years ago and last night I accidentally came to see this adaption.<br /><br />Although it's been years since I read the story the first time, the differences between the novel and the movie are humongous. Very important elements, which made the whole thing plausible are just written out or changed to bad.<br /><br />If the plot sounds interesting to you: go and get the novel. Its much, much, much better.<br /><br />Still 4 out of 10 since it was hard to stop watching because of the great basic plot by Ben Mezrich.\", shape=(), dtype=string) ## tf.Tensor(0, shape=(), dtype=int32) ## tf.Tensor(b'After seeing all the Jesse James, Quantrill, jayhawkers,etc films in the fifties, it is quite a thrill to see this film with a new perspective by director Ang Lee. The scene of the attack of Lawrence, Kansas is awesome. The romantic relationship between Jewel and Toby Mcguire turns out to be one of the best parts and Jonathan Rhys-Meyers is outstanding as the bad guy. All the time this film makes you feel the horror of war, and the desperate situation of the main characters who do not know if they are going to survive the next hours. Definitely worth seeing.', shape=(), dtype=string) ## tf.Tensor(1, shape=(), dtype=int32) ## tf.Tensor(b'AG was an excellent presentation of drama, suspense and thriller that is so rare to American TV. Sheriff Lucas gave many a viewer the willies. We rooted for Caleb as he strove to resist the overtures of Sheriff Lucas. We became engrossed and fearful upon learning of the unthinkable connection between these two characters. The manipulations which weekly gave cause to fear what Lucas would do next were truly surprising. This show lived up to the \"Gothic\" moniker in ways American entertainment has so seldom attempted, much less mastered. The suits definitely made a big mistake in not supporting this show. This show puts shame to the current glut of \"reality\" shows- which are so less than satisfying viewing.The call for a DVD box set is well based. This show is quality viewing for a discerning market hungry for quality viewing. A public that is tiring of over-saturation of mind-numbing reality fare will welcome this gem of real storytelling. Bring on the DVD box set!!', shape=(), dtype=string) ## tf.Tensor(1, shape=(), dtype=int32)"},{"path":"https://keras3.posit.co/dev/articles/examples/nlp/text_classification_from_scratch.html","id":"prepare-the-data","dir":"Articles > Examples > Nlp","previous_headings":"","what":"Prepare the data","title":"Text classification from scratch","text":"particular, remove <br /> tags.","code":"# Having looked at our data above, we see that the raw text contains HTML break # tags of the form '<br />'. These tags will not be removed by the default # standardizer (which doesn't strip HTML). Because of this, we will need to # create a custom standardization function. custom_standardization_fn <- function(string_tensor) {   string_tensor |>     tf$strings$lower() |> # convert to all lowercase     tf$strings$regex_replace(\"<br />\", \" \") |> # remove '<br />' HTML tag     tf$strings$regex_replace(\"[[:punct:]]\", \"\") # remove punctuation }   # Model constants. max_features <- 20000 embedding_dim <- 128 sequence_length <- 500  # Now that we have our custom standardization, we can instantiate our text # vectorization layer. We are using this layer to normalize, split, and map # strings to integers, so we set our 'output_mode' to 'int'. # Note that we're using the default split function, # and the custom standardization defined above. # We also set an explicit maximum sequence length, since the CNNs later in our # model won't support ragged sequences. vectorize_layer <- layer_text_vectorization(   standardize = custom_standardization_fn,   max_tokens = max_features,   output_mode = \"int\",   output_sequence_length = sequence_length, )  # Now that the vectorize_layer has been created, call `adapt` on a text-only # dataset to create the vocabulary. You don't have to batch, but for very large # datasets this means you're not keeping spare copies of the dataset in memory.  # Let's make a text-only dataset (no labels): text_ds <- raw_train_ds |>   dataset_map(\\(x, y) x) # Let's call `adapt`: vectorize_layer |> adapt(text_ds)"},{"path":"https://keras3.posit.co/dev/articles/examples/nlp/text_classification_from_scratch.html","id":"two-options-to-vectorize-the-data","dir":"Articles > Examples > Nlp","previous_headings":"","what":"Two options to vectorize the data","title":"Text classification from scratch","text":"2 ways can use text vectorization layer: Option 1: Make part model, obtain model processes raw strings, like : Option 2: Apply text dataset obtain dataset word indices, feed model expects integer sequences inputs. important difference two option 2 enables asynchronous CPU processing buffering data training GPU. ’re training model GPU, probably want go option get best performance. . export model production, ’d ship model accepts raw strings input, like code snippet option 1 . can done training. last section.","code":"text_input <- keras_input(shape = c(1L), dtype = \"string\", name = 'text') x <- text_input |>   vectorize_layer() |>   layer_embedding(max_features + 1, embedding_dim) vectorize_text <- function(text, label) {   text <- text |>     op_expand_dims(-1) |>     vectorize_layer()   list(text, label) }  # Vectorize the data. train_ds <- raw_train_ds |> dataset_map(vectorize_text) val_ds   <- raw_val_ds   |> dataset_map(vectorize_text) test_ds  <- raw_test_ds  |> dataset_map(vectorize_text)  # Do async prefetching / buffering of the data for best performance on GPU. train_ds <- train_ds |>   dataset_cache() |>   dataset_prefetch(buffer_size = 10) val_ds <- val_ds |>   dataset_cache() |>   dataset_prefetch(buffer_size = 10) test_ds <- test_ds |>   dataset_cache() |>   dataset_prefetch(buffer_size = 10)"},{"path":"https://keras3.posit.co/dev/articles/examples/nlp/text_classification_from_scratch.html","id":"build-a-model","dir":"Articles > Examples > Nlp","previous_headings":"","what":"Build a model","title":"Text classification from scratch","text":"choose simple 1D convnet starting Embedding layer.","code":"# A integer input for vocab indices. inputs <- keras_input(shape = c(NA), dtype = \"int64\")  predictions <- inputs |>   # Next, we add a layer to map those vocab indices into a space of dimensionality   # 'embedding_dim'.   layer_embedding(max_features, embedding_dim) |>   layer_dropout(0.5) |>   # Conv1D + global max pooling   layer_conv_1d(128, 7, padding = \"valid\", activation = \"relu\", strides = 3) |>   layer_conv_1d(128, 7, padding = \"valid\", activation = \"relu\", strides = 3) |>   layer_global_max_pooling_1d() |>   # We add a vanilla hidden layer:   layer_dense(128, activation = \"relu\") |>   layer_dropout(0.5) |>   # We project onto a single unit output layer, and squash it with a sigmoid:   layer_dense(1, activation = \"sigmoid\", name = \"predictions\")  model <- keras_model(inputs, predictions)  summary(model) ## Model: \"functional\" ## ┏━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━┓ ## ┃ Layer (type)                    ┃ Output Shape           ┃       Param # ┃ ## ┡━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━┩ ## │ input_layer (InputLayer)        │ (None, None)           │             0 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ embedding_1 (Embedding)         │ (None, None, 128)      │     2,560,000 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ dropout (Dropout)               │ (None, None, 128)      │             0 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ conv1d (Conv1D)                 │ (None, None, 128)      │       114,816 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ conv1d_1 (Conv1D)               │ (None, None, 128)      │       114,816 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ global_max_pooling1d            │ (None, 128)            │             0 │ ## │ (GlobalMaxPooling1D)            │                        │               │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ dense (Dense)                   │ (None, 128)            │        16,512 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ dropout_1 (Dropout)             │ (None, 128)            │             0 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ predictions (Dense)             │ (None, 1)              │           129 │ ## └─────────────────────────────────┴────────────────────────┴───────────────┘ ##  Total params: 2,806,273 (10.71 MB) ##  Trainable params: 2,806,273 (10.71 MB) ##  Non-trainable params: 0 (0.00 B) # Compile the model with binary crossentropy loss and an adam optimizer. model |> compile(loss = \"binary_crossentropy\",                  optimizer = \"adam\",                  metrics = \"accuracy\")"},{"path":"https://keras3.posit.co/dev/articles/examples/nlp/text_classification_from_scratch.html","id":"train-the-model","dir":"Articles > Examples > Nlp","previous_headings":"","what":"Train the model","title":"Text classification from scratch","text":"","code":"epochs <- 3  # Fit the model using the train and test datasets. model |> fit(train_ds, validation_data = val_ds, epochs = epochs) ## Epoch 1/3 ## 625/625 - 5s - 8ms/step - accuracy: 0.6964 - loss: 0.5261 - val_accuracy: 0.8668 - val_loss: 0.3185 ## Epoch 2/3 ## 625/625 - 2s - 3ms/step - accuracy: 0.9074 - loss: 0.2365 - val_accuracy: 0.8754 - val_loss: 0.3245 ## Epoch 3/3 ## 625/625 - 2s - 3ms/step - accuracy: 0.9551 - loss: 0.1220 - val_accuracy: 0.8284 - val_loss: 0.4862"},{"path":"https://keras3.posit.co/dev/articles/examples/nlp/text_classification_from_scratch.html","id":"evaluate-the-model-on-the-test-set","dir":"Articles > Examples > Nlp","previous_headings":"","what":"Evaluate the model on the test set","title":"Text classification from scratch","text":"","code":"model |> evaluate(test_ds) ## 782/782 - 1s - 2ms/step - accuracy: 0.8115 - loss: 0.5299 ## $accuracy ## [1] 0.81152 ## ## $loss ## [1] 0.5299004"},{"path":"https://keras3.posit.co/dev/articles/examples/nlp/text_classification_from_scratch.html","id":"make-an-end-to-end-model","dir":"Articles > Examples > Nlp","previous_headings":"","what":"Make an end-to-end model","title":"Text classification from scratch","text":"want obtain model capable processing raw strings, can simply create new model (using weights just trained):","code":"# A string input inputs <- keras_input(shape = c(1), dtype = \"string\") # Turn strings into vocab indices indices <- vectorize_layer(inputs) # Turn vocab indices into predictions outputs <- model(indices)  # Our end to end model end_to_end_model <- keras_model(inputs, outputs) end_to_end_model |> compile(   loss = \"binary_crossentropy\",   optimizer = \"adam\",   metrics = c(\"accuracy\") )  # Test it with `raw_test_ds`, which yields raw strings end_to_end_model |> evaluate(raw_test_ds) ## 782/782 - 3s - 4ms/step - accuracy: 0.8115 - loss: 0.5299 ## $accuracy ## [1] 0.81152 ## ## $loss ## [1] 0.5299006"},{"path":"https://keras3.posit.co/dev/articles/examples/structured_data/imbalanced_classification.html","id":"introduction","dir":"Articles > Examples > Structured_data","previous_headings":"","what":"Introduction","title":"Imbalanced classification: credit card fraud detection","text":"example looks Kaggle Credit Card Fraud Detection dataset demonstrate train classification model data highly imbalanced classes. can download data clicking “Download” link, ’re setup kaggle API key \"~/.kaggle/kagle.json\", can run following:","code":"reticulate::py_install(\"kaggle\", pip = TRUE) reticulate::py_available(TRUE) # ensure 'kaggle' is on the PATH system(\"kaggle datasets download -d mlg-ulb/creditcardfraud\") zip::unzip(\"creditcardfraud.zip\", files = \"creditcard.csv\")"},{"path":"https://keras3.posit.co/dev/articles/examples/structured_data/imbalanced_classification.html","id":"first-load-the-data","dir":"Articles > Examples > Structured_data","previous_headings":"","what":"First, load the data","title":"Imbalanced classification: credit card fraud detection","text":"","code":"library(readr) df <- read_csv(\"creditcard.csv\", col_types = cols(   Class = col_integer(),   .default = col_double() )) tibble::glimpse(df) ## Rows: 284,807 ## Columns: 31 ## $ Time   <dbl> 0, 0, 1, 1, 2, 2, 4, 7, 7, 9, 10, 10, 10, 11, 12, 12, 12, 1… ## $ V1     <dbl> -1.3598071, 1.1918571, -1.3583541, -0.9662717, -1.1582331, … ## $ V2     <dbl> -0.07278117, 0.26615071, -1.34016307, -0.18522601, 0.877736… ## $ V3     <dbl> 2.53634674, 0.16648011, 1.77320934, 1.79299334, 1.54871785,… ## $ V4     <dbl> 1.37815522, 0.44815408, 0.37977959, -0.86329128, 0.40303393… ## $ V5     <dbl> -0.33832077, 0.06001765, -0.50319813, -0.01030888, -0.40719… ## $ V6     <dbl> 0.46238778, -0.08236081, 1.80049938, 1.24720317, 0.09592146… ## $ V7     <dbl> 0.239598554, -0.078802983, 0.791460956, 0.237608940, 0.5929… ## $ V8     <dbl> 0.098697901, 0.085101655, 0.247675787, 0.377435875, -0.2705… ## $ V9     <dbl> 0.3637870, -0.2554251, -1.5146543, -1.3870241, 0.8177393, -… ## $ V10    <dbl> 0.09079417, -0.16697441, 0.20764287, -0.05495192, 0.7530744… ## $ V11    <dbl> -0.55159953, 1.61272666, 0.62450146, -0.22648726, -0.822842… ## $ V12    <dbl> -0.61780086, 1.06523531, 0.06608369, 0.17822823, 0.53819555… ## $ V13    <dbl> -0.99138985, 0.48909502, 0.71729273, 0.50775687, 1.34585159… ## $ V14    <dbl> -0.31116935, -0.14377230, -0.16594592, -0.28792375, -1.1196… ## $ V15    <dbl> 1.468176972, 0.635558093, 2.345864949, -0.631418118, 0.1751… ## $ V16    <dbl> -0.47040053, 0.46391704, -2.89008319, -1.05964725, -0.45144… ## $ V17    <dbl> 0.207971242, -0.114804663, 1.109969379, -0.684092786, -0.23… ## $ V18    <dbl> 0.02579058, -0.18336127, -0.12135931, 1.96577500, -0.038194… ## $ V19    <dbl> 0.40399296, -0.14578304, -2.26185710, -1.23262197, 0.803486… ## $ V20    <dbl> 0.25141210, -0.06908314, 0.52497973, -0.20803778, 0.4085423… ## $ V21    <dbl> -0.018306778, -0.225775248, 0.247998153, -0.108300452, -0.0… ## $ V22    <dbl> 0.277837576, -0.638671953, 0.771679402, 0.005273597, 0.7982… ## $ V23    <dbl> -0.110473910, 0.101288021, 0.909412262, -0.190320519, -0.13… ## $ V24    <dbl> 0.06692807, -0.33984648, -0.68928096, -1.17557533, 0.141266… ## $ V25    <dbl> 0.12853936, 0.16717040, -0.32764183, 0.64737603, -0.2060095… ## $ V26    <dbl> -0.18911484, 0.12589453, -0.13909657, -0.22192884, 0.502292… ## $ V27    <dbl> 0.133558377, -0.008983099, -0.055352794, 0.062722849, 0.219… ## $ V28    <dbl> -0.021053053, 0.014724169, -0.059751841, 0.061457629, 0.215… ## $ Amount <dbl> 149.62, 2.69, 378.66, 123.50, 69.99, 3.67, 4.99, 40.80, 93.… ## $ Class  <int> 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,…"},{"path":"https://keras3.posit.co/dev/articles/examples/structured_data/imbalanced_classification.html","id":"prepare-a-validation-set","dir":"Articles > Examples > Structured_data","previous_headings":"","what":"Prepare a validation set","title":"Imbalanced classification: credit card fraud detection","text":"","code":"val_idx <- nrow(df) %>% sample.int(., round( . * 0.2)) val_df <- df[val_idx, ] train_df <- df[-val_idx, ]  cat(\"Number of training samples:\", nrow(train_df), \"\\n\") ## Number of training samples: 227846 cat(\"Number of validation samples:\", nrow(val_df), \"\\n\") ## Number of validation samples: 56961"},{"path":"https://keras3.posit.co/dev/articles/examples/structured_data/imbalanced_classification.html","id":"analyze-class-imbalance-in-the-targets","dir":"Articles > Examples > Structured_data","previous_headings":"","what":"Analyze class imbalance in the targets","title":"Imbalanced classification: credit card fraud detection","text":"","code":"counts <- table(train_df$Class) counts ## ##      0      1 ## 227464    382 cat(sprintf(\"Number of positive samples in training data: %i (%.2f%% of total)\",             counts[\"1\"], 100 * counts[\"1\"] / sum(counts))) ## Number of positive samples in training data: 382 (0.17% of total) weight_for_0 = 1 / counts[\"0\"] weight_for_1 = 1 / counts[\"1\"]"},{"path":"https://keras3.posit.co/dev/articles/examples/structured_data/imbalanced_classification.html","id":"normalize-the-data-using-training-set-statistics","dir":"Articles > Examples > Structured_data","previous_headings":"","what":"Normalize the data using training set statistics","title":"Imbalanced classification: credit card fraud detection","text":"","code":"feature_names <- colnames(train_df) %>% setdiff(\"Class\")  train_features <- as.matrix(train_df[feature_names]) train_targets <- as.matrix(train_df$Class)  val_features <- as.matrix(val_df[feature_names]) val_targets <- as.matrix(val_df$Class)  train_features %<>% scale() val_features %<>% scale(center = attr(train_features, \"scaled:center\"),                         scale = attr(train_features, \"scaled:scale\"))"},{"path":"https://keras3.posit.co/dev/articles/examples/structured_data/imbalanced_classification.html","id":"build-a-binary-classification-model","dir":"Articles > Examples > Structured_data","previous_headings":"","what":"Build a binary classification model","title":"Imbalanced classification: credit card fraud detection","text":"","code":"model <-   keras_model_sequential(input_shape = ncol(train_features)) |>   layer_dense(256, activation = \"relu\") |>   layer_dense(256, activation = \"relu\") |>   layer_dropout(0.3) |>   layer_dense(256, activation = \"relu\") |>   layer_dropout(0.3) |>   layer_dense(1, activation = \"sigmoid\")  model ## Model: \"sequential\" ## ┏━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━┓ ## ┃ Layer (type)                    ┃ Output Shape           ┃       Param # ┃ ## ┡━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━┩ ## │ dense (Dense)                   │ (None, 256)            │         7,936 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ dense_1 (Dense)                 │ (None, 256)            │        65,792 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ dropout (Dropout)               │ (None, 256)            │             0 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ dense_2 (Dense)                 │ (None, 256)            │        65,792 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ dropout_1 (Dropout)             │ (None, 256)            │             0 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ dense_3 (Dense)                 │ (None, 1)              │           257 │ ## └─────────────────────────────────┴────────────────────────┴───────────────┘ ##  Total params: 139,777 (546.00 KB) ##  Trainable params: 139,777 (546.00 KB) ##  Non-trainable params: 0 (0.00 B)"},{"path":"https://keras3.posit.co/dev/articles/examples/structured_data/imbalanced_classification.html","id":"train-the-model-with-class_weight-argument","dir":"Articles > Examples > Structured_data","previous_headings":"","what":"Train the model with class_weight argument","title":"Imbalanced classification: credit card fraud detection","text":"","code":"metrics <- list(   metric_false_negatives(name = \"fn\"),   metric_false_positives(name = \"fp\"),   metric_true_negatives(name = \"tn\"),   metric_true_positives(name = \"tp\"),   metric_precision(name = \"precision\"),   metric_recall(name = \"recall\") ) model |> compile(   optimizer = optimizer_adam(1e-2),   loss = \"binary_crossentropy\",   metrics = metrics ) callbacks <- list(   callback_model_checkpoint(\"fraud_model_at_epoch_{epoch}.keras\") )  class_weight <- list(\"0\" = weight_for_0,                      \"1\" = weight_for_1)  model |> fit(   train_features, train_targets,   validation_data = list(val_features, val_targets),   class_weight = class_weight,   batch_size = 2048,   epochs = 30,   callbacks = callbacks,   verbose = 2 ) ## Epoch 1/30 ## 112/112 - 4s - 37ms/step - fn: 36.0000 - fp: 29717.0000 - loss: 2.3629e-06 - precision: 0.0115 - recall: 0.9058 - tn: 197747.0000 - tp: 346.0000 - val_fn: 12.0000 - val_fp: 1039.0000 - val_loss: 0.0984 - val_precision: 0.0862 - val_recall: 0.8909 - val_tn: 55812.0000 - val_tp: 98.0000 ## Epoch 2/30 ## 112/112 - 1s - 9ms/step - fn: 30.0000 - fp: 6694.0000 - loss: 1.3508e-06 - precision: 0.0500 - recall: 0.9215 - tn: 220770.0000 - tp: 352.0000 - val_fn: 8.0000 - val_fp: 2418.0000 - val_loss: 0.1390 - val_precision: 0.0405 - val_recall: 0.9273 - val_tn: 54433.0000 - val_tp: 102.0000 ## Epoch 3/30 ## 112/112 - 0s - 2ms/step - fn: 27.0000 - fp: 9781.0000 - loss: 1.3090e-06 - precision: 0.0350 - recall: 0.9293 - tn: 217683.0000 - tp: 355.0000 - val_fn: 10.0000 - val_fp: 984.0000 - val_loss: 0.1063 - val_precision: 0.0923 - val_recall: 0.9091 - val_tn: 55867.0000 - val_tp: 100.0000 ## Epoch 4/30 ## 112/112 - 0s - 2ms/step - fn: 32.0000 - fp: 6894.0000 - loss: 1.1099e-06 - precision: 0.0483 - recall: 0.9162 - tn: 220570.0000 - tp: 350.0000 - val_fn: 9.0000 - val_fp: 1557.0000 - val_loss: 0.0857 - val_precision: 0.0609 - val_recall: 0.9182 - val_tn: 55294.0000 - val_tp: 101.0000 ## Epoch 5/30 ## 112/112 - 0s - 2ms/step - fn: 23.0000 - fp: 8892.0000 - loss: 9.2992e-07 - precision: 0.0388 - recall: 0.9398 - tn: 218572.0000 - tp: 359.0000 - val_fn: 7.0000 - val_fp: 1492.0000 - val_loss: 0.0665 - val_precision: 0.0646 - val_recall: 0.9364 - val_tn: 55359.0000 - val_tp: 103.0000 ## Epoch 6/30 ## 112/112 - 0s - 2ms/step - fn: 18.0000 - fp: 8209.0000 - loss: 7.7549e-07 - precision: 0.0425 - recall: 0.9529 - tn: 219255.0000 - tp: 364.0000 - val_fn: 7.0000 - val_fp: 2749.0000 - val_loss: 0.1010 - val_precision: 0.0361 - val_recall: 0.9364 - val_tn: 54102.0000 - val_tp: 103.0000 ## Epoch 7/30 ## 112/112 - 0s - 2ms/step - fn: 15.0000 - fp: 7189.0000 - loss: 6.9032e-07 - precision: 0.0486 - recall: 0.9607 - tn: 220275.0000 - tp: 367.0000 - val_fn: 12.0000 - val_fp: 934.0000 - val_loss: 0.0539 - val_precision: 0.0950 - val_recall: 0.8909 - val_tn: 55917.0000 - val_tp: 98.0000 ## Epoch 8/30 ## 112/112 - 0s - 2ms/step - fn: 18.0000 - fp: 7759.0000 - loss: 8.2432e-07 - precision: 0.0448 - recall: 0.9529 - tn: 219705.0000 - tp: 364.0000 - val_fn: 10.0000 - val_fp: 1481.0000 - val_loss: 0.0777 - val_precision: 0.0633 - val_recall: 0.9091 - val_tn: 55370.0000 - val_tp: 100.0000 ## Epoch 9/30 ## 112/112 - 0s - 2ms/step - fn: 16.0000 - fp: 7565.0000 - loss: 7.2956e-07 - precision: 0.0461 - recall: 0.9581 - tn: 219899.0000 - tp: 366.0000 - val_fn: 8.0000 - val_fp: 2075.0000 - val_loss: 0.0718 - val_precision: 0.0469 - val_recall: 0.9273 - val_tn: 54776.0000 - val_tp: 102.0000 ## Epoch 10/30 ## 112/112 - 0s - 2ms/step - fn: 10.0000 - fp: 7374.0000 - loss: 5.9522e-07 - precision: 0.0480 - recall: 0.9738 - tn: 220090.0000 - tp: 372.0000 - val_fn: 11.0000 - val_fp: 995.0000 - val_loss: 0.0425 - val_precision: 0.0905 - val_recall: 0.9000 - val_tn: 55856.0000 - val_tp: 99.0000 ## Epoch 11/30 ## 112/112 - 0s - 2ms/step - fn: 10.0000 - fp: 6586.0000 - loss: 6.2194e-07 - precision: 0.0535 - recall: 0.9738 - tn: 220878.0000 - tp: 372.0000 - val_fn: 9.0000 - val_fp: 5097.0000 - val_loss: 0.2103 - val_precision: 0.0194 - val_recall: 0.9182 - val_tn: 51754.0000 - val_tp: 101.0000 ## Epoch 12/30 ## 112/112 - 0s - 2ms/step - fn: 6.0000 - fp: 8285.0000 - loss: 6.0991e-07 - precision: 0.0434 - recall: 0.9843 - tn: 219179.0000 - tp: 376.0000 - val_fn: 9.0000 - val_fp: 1926.0000 - val_loss: 0.0781 - val_precision: 0.0498 - val_recall: 0.9182 - val_tn: 54925.0000 - val_tp: 101.0000 ## Epoch 13/30 ## 112/112 - 0s - 2ms/step - fn: 7.0000 - fp: 6864.0000 - loss: 5.7087e-07 - precision: 0.0518 - recall: 0.9817 - tn: 220600.0000 - tp: 375.0000 - val_fn: 8.0000 - val_fp: 1900.0000 - val_loss: 0.0837 - val_precision: 0.0509 - val_recall: 0.9273 - val_tn: 54951.0000 - val_tp: 102.0000 ## Epoch 14/30 ## 112/112 - 0s - 2ms/step - fn: 7.0000 - fp: 7663.0000 - loss: 5.7664e-07 - precision: 0.0467 - recall: 0.9817 - tn: 219801.0000 - tp: 375.0000 - val_fn: 8.0000 - val_fp: 1540.0000 - val_loss: 0.0656 - val_precision: 0.0621 - val_recall: 0.9273 - val_tn: 55311.0000 - val_tp: 102.0000 ## Epoch 15/30 ## 112/112 - 0s - 2ms/step - fn: 6.0000 - fp: 4607.0000 - loss: 3.9004e-07 - precision: 0.0755 - recall: 0.9843 - tn: 222857.0000 - tp: 376.0000 - val_fn: 11.0000 - val_fp: 2546.0000 - val_loss: 0.0994 - val_precision: 0.0374 - val_recall: 0.9000 - val_tn: 54305.0000 - val_tp: 99.0000 ## Epoch 16/30 ## 112/112 - 0s - 2ms/step - fn: 6.0000 - fp: 6195.0000 - loss: 5.1274e-07 - precision: 0.0572 - recall: 0.9843 - tn: 221269.0000 - tp: 376.0000 - val_fn: 9.0000 - val_fp: 1652.0000 - val_loss: 0.0693 - val_precision: 0.0576 - val_recall: 0.9182 - val_tn: 55199.0000 - val_tp: 101.0000 ## Epoch 17/30 ## 112/112 - 0s - 2ms/step - fn: 6.0000 - fp: 5367.0000 - loss: 4.6364e-07 - precision: 0.0655 - recall: 0.9843 - tn: 222097.0000 - tp: 376.0000 - val_fn: 10.0000 - val_fp: 1378.0000 - val_loss: 0.0575 - val_precision: 0.0677 - val_recall: 0.9091 - val_tn: 55473.0000 - val_tp: 100.0000 ## Epoch 18/30 ## 112/112 - 0s - 2ms/step - fn: 2.0000 - fp: 3696.0000 - loss: 2.9266e-07 - precision: 0.0932 - recall: 0.9948 - tn: 223768.0000 - tp: 380.0000 - val_fn: 11.0000 - val_fp: 824.0000 - val_loss: 0.0375 - val_precision: 0.1073 - val_recall: 0.9000 - val_tn: 56027.0000 - val_tp: 99.0000 ## Epoch 19/30 ## 112/112 - 0s - 2ms/step - fn: 3.0000 - fp: 3638.0000 - loss: 3.6945e-07 - precision: 0.0943 - recall: 0.9921 - tn: 223826.0000 - tp: 379.0000 - val_fn: 4.0000 - val_fp: 4639.0000 - val_loss: 0.5475 - val_precision: 0.0223 - val_recall: 0.9636 - val_tn: 52212.0000 - val_tp: 106.0000 ## Epoch 20/30 ## 112/112 - 0s - 2ms/step - fn: 26.0000 - fp: 11743.0000 - loss: 4.2168e-06 - precision: 0.0294 - recall: 0.9319 - tn: 215721.0000 - tp: 356.0000 - val_fn: 104.0000 - val_fp: 395.0000 - val_loss: 0.0507 - val_precision: 0.0150 - val_recall: 0.0545 - val_tn: 56456.0000 - val_tp: 6.0000 ## Epoch 21/30 ## 112/112 - 0s - 2ms/step - fn: 41.0000 - fp: 8860.0000 - loss: 5.8042e-06 - precision: 0.0371 - recall: 0.8927 - tn: 218604.0000 - tp: 341.0000 - val_fn: 10.0000 - val_fp: 2791.0000 - val_loss: 0.2183 - val_precision: 0.0346 - val_recall: 0.9091 - val_tn: 54060.0000 - val_tp: 100.0000 ## Epoch 22/30 ## 112/112 - 0s - 2ms/step - fn: 15.0000 - fp: 9031.0000 - loss: 1.0670e-06 - precision: 0.0391 - recall: 0.9607 - tn: 218433.0000 - tp: 367.0000 - val_fn: 10.0000 - val_fp: 1275.0000 - val_loss: 0.0764 - val_precision: 0.0727 - val_recall: 0.9091 - val_tn: 55576.0000 - val_tp: 100.0000 ## Epoch 23/30 ## 112/112 - 0s - 2ms/step - fn: 10.0000 - fp: 5141.0000 - loss: 6.9193e-07 - precision: 0.0675 - recall: 0.9738 - tn: 222323.0000 - tp: 372.0000 - val_fn: 11.0000 - val_fp: 1872.0000 - val_loss: 0.0824 - val_precision: 0.0502 - val_recall: 0.9000 - val_tn: 54979.0000 - val_tp: 99.0000 ## Epoch 24/30 ## 112/112 - 0s - 2ms/step - fn: 2.0000 - fp: 4232.0000 - loss: 4.2135e-07 - precision: 0.0824 - recall: 0.9948 - tn: 223232.0000 - tp: 380.0000 - val_fn: 11.0000 - val_fp: 976.0000 - val_loss: 0.0488 - val_precision: 0.0921 - val_recall: 0.9000 - val_tn: 55875.0000 - val_tp: 99.0000 ## Epoch 25/30 ## 112/112 - 0s - 2ms/step - fn: 3.0000 - fp: 4149.0000 - loss: 3.7232e-07 - precision: 0.0837 - recall: 0.9921 - tn: 223315.0000 - tp: 379.0000 - val_fn: 13.0000 - val_fp: 359.0000 - val_loss: 0.0229 - val_precision: 0.2127 - val_recall: 0.8818 - val_tn: 56492.0000 - val_tp: 97.0000 ## Epoch 26/30 ## 112/112 - 0s - 2ms/step - fn: 3.0000 - fp: 3757.0000 - loss: 3.5368e-07 - precision: 0.0916 - recall: 0.9921 - tn: 223707.0000 - tp: 379.0000 - val_fn: 12.0000 - val_fp: 1111.0000 - val_loss: 0.0569 - val_precision: 0.0811 - val_recall: 0.8909 - val_tn: 55740.0000 - val_tp: 98.0000 ## Epoch 27/30 ## 112/112 - 0s - 2ms/step - fn: 2.0000 - fp: 3220.0000 - loss: 2.5512e-07 - precision: 0.1056 - recall: 0.9948 - tn: 224244.0000 - tp: 380.0000 - val_fn: 11.0000 - val_fp: 839.0000 - val_loss: 0.0396 - val_precision: 0.1055 - val_recall: 0.9000 - val_tn: 56012.0000 - val_tp: 99.0000 ## Epoch 28/30 ## 112/112 - 0s - 2ms/step - fn: 0.0000e+00 - fp: 2287.0000 - loss: 1.8121e-07 - precision: 0.1431 - recall: 1.0000 - tn: 225177.0000 - tp: 382.0000 - val_fn: 12.0000 - val_fp: 425.0000 - val_loss: 0.0272 - val_precision: 0.1874 - val_recall: 0.8909 - val_tn: 56426.0000 - val_tp: 98.0000 ## Epoch 29/30 ## 112/112 - 0s - 2ms/step - fn: 0.0000e+00 - fp: 1435.0000 - loss: 1.3034e-07 - precision: 0.2102 - recall: 1.0000 - tn: 226029.0000 - tp: 382.0000 - val_fn: 13.0000 - val_fp: 321.0000 - val_loss: 0.0250 - val_precision: 0.2321 - val_recall: 0.8818 - val_tn: 56530.0000 - val_tp: 97.0000 ## Epoch 30/30 ## 112/112 - 0s - 2ms/step - fn: 0.0000e+00 - fp: 1365.0000 - loss: 1.2132e-07 - precision: 0.2187 - recall: 1.0000 - tn: 226099.0000 - tp: 382.0000 - val_fn: 12.0000 - val_fp: 241.0000 - val_loss: 0.0183 - val_precision: 0.2891 - val_recall: 0.8909 - val_tn: 56610.0000 - val_tp: 98.0000 val_pred <- model %>%   predict(val_features) %>%   { as.integer(. > 0.5) } ## 1781/1781 - 1s - 427us/step pred_correct <- val_df$Class == val_pred cat(sprintf(\"Validation accuracy: %.2f\", mean(pred_correct))) ## Validation accuracy: 1.00 fraudulent <- val_df$Class == 1  n_fraudulent_detected <- sum(fraudulent & pred_correct) n_fraudulent_missed <- sum(fraudulent & !pred_correct) n_legitimate_flagged <- sum(!fraudulent & !pred_correct)"},{"path":"https://keras3.posit.co/dev/articles/examples/structured_data/imbalanced_classification.html","id":"conclusions","dir":"Articles > Examples > Structured_data","previous_headings":"","what":"Conclusions","title":"Imbalanced classification: credit card fraud detection","text":"end training, 56,961 validation transactions, : Correctly identifying 98 fraudulent Missing 12 fraudulent transactions cost incorrectly flagging 241 legitimate transactions real world, one put even higher weight class 1, reflect False Negatives costly False Positives. Next time credit card gets declined online purchase – .","code":""},{"path":"https://keras3.posit.co/dev/articles/examples/structured_data/structured_data_classification_with_feature_space.html","id":"introduction","dir":"Articles > Examples > Structured_data","previous_headings":"","what":"Introduction","title":"Structured data classification with FeatureSpace","text":"example demonstrates structured data classification (also known tabular data classification), starting raw CSV file. data includes numerical features, integer categorical features, string categorical features. use utility layer_feature_space() index, preprocess, encode features. code adapted example Structured data classification scratch. previous example managed low-level feature preprocessing encoding Keras preprocessing layers, example delegate everything layer_feature_space(), making workflow extremely quick easy.","code":""},{"path":"https://keras3.posit.co/dev/articles/examples/structured_data/structured_data_classification_with_feature_space.html","id":"the-dataset","dir":"Articles > Examples > Structured_data","previous_headings":"Introduction","what":"The dataset","title":"Structured data classification with FeatureSpace","text":"dataset provided Cleveland Clinic Foundation Heart Disease. ’s CSV file 303 rows. row contains information patient (sample), column describes attribute patient (feature). use features predict whether patient heart disease (binary classification). ’s description feature:","code":""},{"path":"https://keras3.posit.co/dev/articles/examples/structured_data/structured_data_classification_with_feature_space.html","id":"setup","dir":"Articles > Examples > Structured_data","previous_headings":"","what":"Setup","title":"Structured data classification with FeatureSpace","text":"","code":"library(readr) library(dplyr, warn.conflicts = FALSE) library(keras3) library(tensorflow, exclude = c(\"shape\", \"set_random_seed\")) library(tfdatasets, exclude = \"shape\")  conflicted::conflicts_prefer(   keras3::shape(),   keras3::set_random_seed(),   dplyr::filter(),   .quiet = TRUE )  use_backend(\"tensorflow\")"},{"path":"https://keras3.posit.co/dev/articles/examples/structured_data/structured_data_classification_with_feature_space.html","id":"preparing-the-data","dir":"Articles > Examples > Structured_data","previous_headings":"","what":"Preparing the data","title":"Structured data classification with FeatureSpace","text":"Let’s download data load Pandas dataframe: dataset includes 303 samples 14 columns per sample (13 features, plus target label) ’s preview samples: last column, “target”, indicates whether patient heart disease (1) (0). Let’s split data training validation set: Let’s generate tf_dataset objects dataframe: tf_dataset yields tuple (input, target) input dictionary (named list) features target value 0 1: Let’s batch datasets:","code":"file_url <-   \"http://storage.googleapis.com/download.tensorflow.org/data/heart.csv\" df <- read_csv(file_url, col_types = cols(   oldpeak = col_double(),   thal = col_character(),   .default = col_integer() ))  # the dataset has two malformed rows, filter them out df <- df |> filter(!thal %in% c(\"1\", \"2\")) glimpse(df) ## Rows: 301 ## Columns: 14 ## $ age      <int> 63, 67, 67, 37, 41, 56, 62, 57, 63, 53, 57, 56, 56, 44, 5… ## $ sex      <int> 1, 1, 1, 1, 0, 1, 0, 0, 1, 1, 1, 0, 1, 1, 1, 1, 1, 1, 0, … ## $ cp       <int> 1, 4, 4, 3, 2, 2, 4, 4, 4, 4, 4, 2, 3, 2, 3, 3, 2, 4, 3, … ## $ trestbps <int> 145, 160, 120, 130, 130, 120, 140, 120, 130, 140, 140, 14… ## $ chol     <int> 233, 286, 229, 250, 204, 236, 268, 354, 254, 203, 192, 29… ## $ fbs      <int> 1, 0, 0, 0, 0, 0, 0, 0, 0, 1, 0, 0, 1, 0, 1, 0, 0, 0, 0, … ## $ restecg  <int> 2, 2, 2, 0, 2, 0, 2, 0, 2, 2, 0, 2, 2, 0, 0, 0, 0, 0, 0, … ## $ thalach  <int> 150, 108, 129, 187, 172, 178, 160, 163, 147, 155, 148, 15… ## $ exang    <int> 0, 1, 1, 0, 0, 0, 0, 1, 0, 1, 0, 0, 1, 0, 0, 0, 0, 0, 0, … ## $ oldpeak  <dbl> 2.3, 1.5, 2.6, 3.5, 1.4, 0.8, 3.6, 0.6, 1.4, 3.1, 0.4, 1.… ## $ slope    <int> 3, 2, 2, 3, 1, 1, 3, 1, 2, 3, 2, 2, 2, 1, 1, 1, 3, 1, 1, … ## $ ca       <int> 0, 3, 2, 0, 0, 0, 2, 0, 1, 0, 0, 0, 1, 0, 0, 0, 0, 0, 0, … ## $ thal     <chr> \"fixed\", \"normal\", \"reversible\", \"normal\", \"normal\", \"nor… ## $ target   <int> 0, 1, 0, 0, 0, 0, 1, 0, 1, 0, 0, 0, 1, 0, 0, 0, 0, 0, 0, … df ## # A tibble: 301 × 14 ##      age   sex    cp trestbps  chol   fbs restecg thalach exang oldpeak ##    <int> <int> <int>    <int> <int> <int>   <int>   <int> <int>   <dbl> ##  1    63     1     1      145   233     1       2     150     0     2.3 ##  2    67     1     4      160   286     0       2     108     1     1.5 ##  3    67     1     4      120   229     0       2     129     1     2.6 ##  4    37     1     3      130   250     0       0     187     0     3.5 ##  5    41     0     2      130   204     0       2     172     0     1.4 ##  6    56     1     2      120   236     0       0     178     0     0.8 ##  7    62     0     4      140   268     0       2     160     0     3.6 ##  8    57     0     4      120   354     0       0     163     1     0.6 ##  9    63     1     4      130   254     0       2     147     0     1.4 ## 10    53     1     4      140   203     1       2     155     1     3.1 ## # ℹ 291 more rows ## # ℹ 4 more variables: slope <int>, ca <int>, thal <chr>, target <int> val_idx <- nrow(df) %>% sample.int(., . * 0.2) val_df <- df[val_idx, ] train_df <- df[-val_idx, ]  cat(sprintf(   \"Using %d samples for training and %d for validation\",   nrow(train_df), nrow(val_df) )) ## Using 241 samples for training and 60 for validation dataframe_to_dataset <- function(df) {   labels <- df |> pull(target) |> as.integer()   inputs <- df |> select(-target) |> as.list()    ds <- tensor_slices_dataset(list(inputs, labels)) |>     dataset_shuffle(nrow(df))    ds }  train_ds <- dataframe_to_dataset(train_df) val_ds <- dataframe_to_dataset(val_df) c(x, y) %<-% iter_next(as_iterator(train_ds)) cat(\"Input: \"); str(x) cat(\"Target: \"); str(y) ## Input: List of 13 ##  $ age     :<tf.Tensor: shape=(), dtype=int32, numpy=68> ##  $ sex     :<tf.Tensor: shape=(), dtype=int32, numpy=1> ##  $ cp      :<tf.Tensor: shape=(), dtype=int32, numpy=4> ##  $ trestbps:<tf.Tensor: shape=(), dtype=int32, numpy=144> ##  $ chol    :<tf.Tensor: shape=(), dtype=int32, numpy=193> ##  $ fbs     :<tf.Tensor: shape=(), dtype=int32, numpy=1> ##  $ restecg :<tf.Tensor: shape=(), dtype=int32, numpy=0> ##  $ thalach :<tf.Tensor: shape=(), dtype=int32, numpy=141> ##  $ exang   :<tf.Tensor: shape=(), dtype=int32, numpy=0> ##  $ oldpeak :<tf.Tensor: shape=(), dtype=float32, numpy=3.4000000953674316> ##  $ slope   :<tf.Tensor: shape=(), dtype=int32, numpy=2> ##  $ ca      :<tf.Tensor: shape=(), dtype=int32, numpy=2> ##  $ thal    :<tf.Tensor: shape=(), dtype=string, numpy=b'reversible'> ## Target: <tf.Tensor: shape=(), dtype=int32, numpy=1> train_ds <- train_ds |> dataset_batch(32) val_ds <- val_ds |> dataset_batch(32)"},{"path":"https://keras3.posit.co/dev/articles/examples/structured_data/structured_data_classification_with_feature_space.html","id":"configuring-a-featurespace","dir":"Articles > Examples > Structured_data","previous_headings":"","what":"Configuring a FeatureSpace","title":"Structured data classification with FeatureSpace","text":"configure feature preprocessed, instantiate layer_feature_space(), pass dictionary (named list unique names) maps name features string describes feature type. “integer categorical” features \"FBS\", one “string categorical” feature (\"thal\"), numerical features, ’d like normalize – except \"age\", ’d like discretize number bins. also use crosses argument capture feature interactions categorical features, say, create additional features represent value co-occurrences categorical features. can compute feature crosses like arbitrary sets categorical features – just tuples two features. resulting co-occurences hashed fixed-sized vector, don’t need worry whether co-occurence space large.","code":"feature_space <- layer_feature_space(   features = list(     # Categorical features encoded as integers     sex = \"integer_categorical\",     cp = \"integer_categorical\",     fbs = \"integer_categorical\",     restecg = \"integer_categorical\",     exang = \"integer_categorical\",     ca = \"integer_categorical\",     # Categorical feature encoded as string     thal = \"string_categorical\",     # Numerical features to discretize     age = \"float_discretized\",     # Numerical features to normalize     trestbps = \"float_normalized\",     chol = \"float_normalized\",     thalach = \"float_normalized\",     oldpeak = \"float_normalized\",     slope = \"float_normalized\"   ),   # We create additional features by hashing   # value co-occurrences for the   # following groups of categorical features.   crosses = list(c(\"sex\", \"age\"), c(\"thal\", \"ca\")),   # The hashing space for these co-occurrences   # wil be 32-dimensional.   crossing_dim = 32,   # Our utility will one-hot encode all categorical   # features and concat all features into a single   # vector (one vector per sample).   output_mode = \"concat\" )"},{"path":"https://keras3.posit.co/dev/articles/examples/structured_data/structured_data_classification_with_feature_space.html","id":"further-customizing-a-featurespace","dir":"Articles > Examples > Structured_data","previous_headings":"","what":"Further customizing a FeatureSpace","title":"Structured data classification with FeatureSpace","text":"Specifying feature type via string name quick easy, sometimes may want configure preprocessing feature. instance, case, categorical features don’t large set possible values – ’s handful values per feature (e.g. 1 0 feature \"FBS\"), possible values represented training set. result, don’t need reserve index represent “vocabulary” values features – default behavior. , just specify num_oov_indices=0 features tell feature preprocessor skip “vocabulary” indexing. customizations access include specifying number bins discretizing features type \"float_discretized\", dimensionality hashing space feature crossing.","code":"feature_space <- layer_feature_space(   features = list(     # Categorical features encoded as integers     sex       = feature_integer_categorical(num_oov_indices = 0),     cp        = feature_integer_categorical(num_oov_indices = 0),     fbs       = feature_integer_categorical(num_oov_indices = 0),     restecg   = feature_integer_categorical(num_oov_indices = 0),     exang     = feature_integer_categorical(num_oov_indices = 0),     ca        = feature_integer_categorical(num_oov_indices = 0),     # Categorical feature encoded as string     thal      = feature_string_categorical(num_oov_indices = 0),     # Numerical features to discretize     age       = feature_float_discretized(num_bins = 30),     # Numerical features to normalize     trestbps  = feature_float_normalized(),     chol      = feature_float_normalized(),     thalach   = feature_float_normalized(),     oldpeak   = feature_float_normalized(),     slope     = feature_float_normalized()   ),   # Specify feature cross with a custom crossing dim.   crosses = list(     feature_cross(       feature_names = c(\"sex\", \"age\"),       crossing_dim = 64     ),     feature_cross(       feature_names = c(\"thal\", \"ca\"),       crossing_dim = 16     )   ),   output_mode = \"concat\" )"},{"path":"https://keras3.posit.co/dev/articles/examples/structured_data/structured_data_classification_with_feature_space.html","id":"adapt-the-featurespace-to-the-training-data","dir":"Articles > Examples > Structured_data","previous_headings":"","what":"Adapt the FeatureSpace to the training data","title":"Structured data classification with FeatureSpace","text":"start using FeatureSpace build model, adapt training data. adapt(), FeatureSpace : Index set possible values categorical features. Compute mean variance numerical features normalize. Compute value boundaries different bins numerical features discretize. Note adapt() called tf_dataset yields dicts (named lists) feature values – labels. point, FeatureSpace can called dict raw feature values, return single concatenate vector sample, combining encoded features feature crosses.","code":"train_ds_with_no_labels <- train_ds |> dataset_map(\\(x, y) x) feature_space |> adapt(train_ds_with_no_labels) c(x, y) %<-% iter_next(as_iterator(train_ds)) preprocessed_x <- feature_space(x) preprocessed_x ## tf.Tensor( ## [[0. 0. 0. ... 0. 0. 0.] ##  [0. 0. 0. ... 0. 0. 0.] ##  [0. 0. 0. ... 0. 0. 0.] ##  ... ##  [0. 0. 0. ... 0. 0. 0.] ##  [0. 0. 0. ... 0. 0. 0.] ##  [0. 0. 0. ... 0. 0. 0.]], shape=(32, 136), dtype=float32)"},{"path":"https://keras3.posit.co/dev/articles/examples/structured_data/structured_data_classification_with_feature_space.html","id":"two-ways-to-manage-preprocessing-as-part-of-the-tf-data-pipeline-or-in-the-model-itself","dir":"Articles > Examples > Structured_data","previous_headings":"","what":"Two ways to manage preprocessing: as part of the tf.data pipeline, or in the model itself","title":"Structured data classification with FeatureSpace","text":"two ways can leverage FeatureSpace:","code":""},{"path":"https://keras3.posit.co/dev/articles/examples/structured_data/structured_data_classification_with_feature_space.html","id":"asynchronous-preprocessing-in-tf-data","dir":"Articles > Examples > Structured_data","previous_headings":"Two ways to manage preprocessing: as part of the tf.data pipeline, or in the model itself","what":"Asynchronous preprocessing in tf.data","title":"Structured data classification with FeatureSpace","text":"can make part data pipeline, model. enables asynchronous parallel preprocessing data CPU hits model. ’re training GPU TPU, want speed preprocessing. Usually, always right thing training.","code":""},{"path":"https://keras3.posit.co/dev/articles/examples/structured_data/structured_data_classification_with_feature_space.html","id":"synchronous-preprocessing-in-the-model","dir":"Articles > Examples > Structured_data","previous_headings":"Two ways to manage preprocessing: as part of the tf.data pipeline, or in the model itself","what":"Synchronous preprocessing in the model","title":"Structured data classification with FeatureSpace","text":"can make part model. means model expect dicts raw feature values, preprocessing batch done synchronously (blocking manner) rest forward pass. want end--end model can process raw feature values – keep mind model able run CPU, since types feature preprocessing (e.g. string preprocessing) GPU TPU compatible. GPU / TPU performance-sensitive settings. general, want -model preprocessing inference CPU. case, apply FeatureSpace tf.data pipeline training, inference end--end model includes FeatureSpace. Let’s create training validation dataset preprocessed batches:","code":"preprocessed_train_ds <- train_ds |>   dataset_map(\\(x, y) list(feature_space(x), y),               num_parallel_calls = tf$data$AUTOTUNE) |>   dataset_prefetch(tf$data$AUTOTUNE)  preprocessed_val_ds <- val_ds |>   dataset_map(\\(x, y) list(feature_space(x), y),               num_parallel_calls = tf$data$AUTOTUNE) |>   dataset_prefetch(tf$data$AUTOTUNE)"},{"path":"https://keras3.posit.co/dev/articles/examples/structured_data/structured_data_classification_with_feature_space.html","id":"build-a-model","dir":"Articles > Examples > Structured_data","previous_headings":"","what":"Build a model","title":"Structured data classification with FeatureSpace","text":"Time build model – rather two models: training model expects preprocessed features (one sample = one vector) inference model expects raw features (one sample = dict raw feature values)","code":"dict_inputs <- feature_space$get_inputs() encoded_features <- feature_space$get_encoded_features()  predictions <- encoded_features |>   layer_dense(32, activation=\"relu\") |>   layer_dropout(0.5) |>   layer_dense(1, activation=\"sigmoid\")  training_model <- keras_model(inputs = encoded_features,                               outputs = predictions) training_model |> compile(optimizer = \"adam\",                           loss = \"binary_crossentropy\",                           metrics = \"accuracy\")  inference_model <- keras_model(inputs = dict_inputs,                                outputs = predictions)"},{"path":"https://keras3.posit.co/dev/articles/examples/structured_data/structured_data_classification_with_feature_space.html","id":"train-the-model","dir":"Articles > Examples > Structured_data","previous_headings":"","what":"Train the model","title":"Structured data classification with FeatureSpace","text":"Let’s train model 20 epochs. Note feature preprocessing happening part tf.data pipeline, part model. quickly get 80% validation accuracy.","code":"training_model |> fit(   preprocessed_train_ds,   epochs = 20,   validation_data = preprocessed_val_ds,   verbose = 2 ) ## Epoch 1/20 ## 8/8 - 2s - 262ms/step - accuracy: 0.4108 - loss: 0.7398 - val_accuracy: 0.3667 - val_loss: 0.7218 ## Epoch 2/20 ## 8/8 - 0s - 26ms/step - accuracy: 0.5934 - loss: 0.6798 - val_accuracy: 0.5333 - val_loss: 0.6670 ## Epoch 3/20 ## 8/8 - 0s - 26ms/step - accuracy: 0.6473 - loss: 0.6384 - val_accuracy: 0.6333 - val_loss: 0.6207 ## Epoch 4/20 ## 8/8 - 0s - 26ms/step - accuracy: 0.7012 - loss: 0.5829 - val_accuracy: 0.7167 - val_loss: 0.5800 ## Epoch 5/20 ## 8/8 - 0s - 27ms/step - accuracy: 0.6846 - loss: 0.5780 - val_accuracy: 0.7500 - val_loss: 0.5445 ## Epoch 6/20 ## 8/8 - 0s - 26ms/step - accuracy: 0.7718 - loss: 0.5595 - val_accuracy: 0.7667 - val_loss: 0.5134 ## Epoch 7/20 ## 8/8 - 0s - 26ms/step - accuracy: 0.7801 - loss: 0.5166 - val_accuracy: 0.8167 - val_loss: 0.4868 ## Epoch 8/20 ## 8/8 - 0s - 26ms/step - accuracy: 0.7718 - loss: 0.5053 - val_accuracy: 0.8333 - val_loss: 0.4626 ## Epoch 9/20 ## 8/8 - 0s - 28ms/step - accuracy: 0.8091 - loss: 0.4774 - val_accuracy: 0.8333 - val_loss: 0.4401 ## Epoch 10/20 ## 8/8 - 0s - 27ms/step - accuracy: 0.8050 - loss: 0.4682 - val_accuracy: 0.8500 - val_loss: 0.4197 ## Epoch 11/20 ## 8/8 - 0s - 26ms/step - accuracy: 0.8133 - loss: 0.4541 - val_accuracy: 0.8667 - val_loss: 0.4011 ## Epoch 12/20 ## 8/8 - 0s - 27ms/step - accuracy: 0.8216 - loss: 0.4360 - val_accuracy: 0.8667 - val_loss: 0.3843 ## Epoch 13/20 ## 8/8 - 0s - 26ms/step - accuracy: 0.8174 - loss: 0.4081 - val_accuracy: 0.8667 - val_loss: 0.3684 ## Epoch 14/20 ## 8/8 - 0s - 27ms/step - accuracy: 0.8423 - loss: 0.4034 - val_accuracy: 0.8667 - val_loss: 0.3531 ## Epoch 15/20 ## 8/8 - 0s - 26ms/step - accuracy: 0.8174 - loss: 0.3956 - val_accuracy: 0.8833 - val_loss: 0.3397 ## Epoch 16/20 ## 8/8 - 0s - 27ms/step - accuracy: 0.8423 - loss: 0.3952 - val_accuracy: 0.8833 - val_loss: 0.3274 ## Epoch 17/20 ## 8/8 - 0s - 27ms/step - accuracy: 0.8548 - loss: 0.3445 - val_accuracy: 0.8833 - val_loss: 0.3169 ## Epoch 18/20 ## 8/8 - 0s - 27ms/step - accuracy: 0.8382 - loss: 0.3609 - val_accuracy: 0.8833 - val_loss: 0.3080 ## Epoch 19/20 ## 8/8 - 0s - 26ms/step - accuracy: 0.8714 - loss: 0.3412 - val_accuracy: 0.8833 - val_loss: 0.2986 ## Epoch 20/20 ## 8/8 - 0s - 26ms/step - accuracy: 0.8506 - loss: 0.3541 - val_accuracy: 0.9000 - val_loss: 0.2905"},{"path":"https://keras3.posit.co/dev/articles/examples/structured_data/structured_data_classification_with_feature_space.html","id":"inference-on-new-data-with-the-end-to-end-model","dir":"Articles > Examples > Structured_data","previous_headings":"","what":"Inference on new data with the end-to-end model","title":"Structured data classification with FeatureSpace","text":"Now, can use inference model (includes FeatureSpace) make predictions based dicts raw features values, follows:","code":"sample <- list(   age = 60,   sex = 1,   cp = 1,   trestbps = 145,   chol = 233,   fbs = 1,   restecg = 2,   thalach = 150,   exang = 0,   oldpeak = 2.3,   slope = 3,   ca = 0,   thal = \"fixed\" )  input_dict <- lapply(sample, \\(x) op_convert_to_tensor(array(x))) predictions <- inference_model |> predict(input_dict) ## 1/1 - 0s - 298ms/step glue::glue(r\"---(   This particular patient had a {(100 * predictions) |> signif(3)}% probability   of having a heart disease, as evaluated by our model. )---\") ## This particular patient had a 48.6% probability ## of having a heart disease, as evaluated by our model."},{"path":"https://keras3.posit.co/dev/articles/examples/timeseries/timeseries_anomaly_detection.html","id":"introduction","dir":"Articles > Examples > Timeseries","previous_headings":"","what":"Introduction","title":"Timeseries anomaly detection using an Autoencoder","text":"script demonstrates can use reconstruction convolutional autoencoder model detect anomalies timeseries data.","code":""},{"path":"https://keras3.posit.co/dev/articles/examples/timeseries/timeseries_anomaly_detection.html","id":"setup","dir":"Articles > Examples > Timeseries","previous_headings":"","what":"Setup","title":"Timeseries anomaly detection using an Autoencoder","text":"","code":"library(dplyr, warn.conflicts = FALSE) library(ggplot2) theme_set(theme_minimal())  library(listarrays) library(tfdatasets, exclude = c(\"shape\")) library(keras3)"},{"path":"https://keras3.posit.co/dev/articles/examples/timeseries/timeseries_anomaly_detection.html","id":"load-the-data","dir":"Articles > Examples > Timeseries","previous_headings":"","what":"Load the data","title":"Timeseries anomaly detection using an Autoencoder","text":"use Numenta Anomaly Benchmark(NAB) dataset. provides artificial timeseries data containing labeled anomalous periods behavior. Data ordered, timestamped, single-valued metrics. use art_daily_small_noise.csv file training art_daily_jumpsup.csv file testing. simplicity dataset allows us demonstrate anomaly detection.","code":"get_data <- function(url_suffix) {   url_root <- \"https://raw.githubusercontent.com/numenta/NAB/master/data/\"   url <- paste0(url_root, url_suffix)   file <- get_file(origin = url) # cache file locally   # parse csv; 2 columns with types: datetime (T), double (d)   readr::read_csv(file, col_types = \"Td\") }  df_small_noise   <- get_data(\"artificialNoAnomaly/art_daily_small_noise.csv\") df_daily_jumpsup <- get_data(\"artificialWithAnomaly/art_daily_jumpsup.csv\")"},{"path":"https://keras3.posit.co/dev/articles/examples/timeseries/timeseries_anomaly_detection.html","id":"quick-look-at-the-data","dir":"Articles > Examples > Timeseries","previous_headings":"","what":"Quick look at the data","title":"Timeseries anomaly detection using an Autoencoder","text":"","code":"df_small_noise ## # A tibble: 4,032 × 2 ##    timestamp           value ##    <dttm>              <dbl> ##  1 2014-04-01 00:00:00  18.3 ##  2 2014-04-01 00:05:00  22.0 ##  3 2014-04-01 00:10:00  18.6 ##  4 2014-04-01 00:15:00  22.0 ##  5 2014-04-01 00:20:00  21.9 ##  6 2014-04-01 00:25:00  21.2 ##  7 2014-04-01 00:30:00  20.6 ##  8 2014-04-01 00:35:00  20.3 ##  9 2014-04-01 00:40:00  21.5 ## 10 2014-04-01 00:45:00  19.2 ## # ℹ 4,022 more rows df_daily_jumpsup ## # A tibble: 4,032 × 2 ##    timestamp           value ##    <dttm>              <dbl> ##  1 2014-04-01 00:00:00  19.8 ##  2 2014-04-01 00:05:00  20.5 ##  3 2014-04-01 00:10:00  20.0 ##  4 2014-04-01 00:15:00  21.5 ##  5 2014-04-01 00:20:00  20.2 ##  6 2014-04-01 00:25:00  19.9 ##  7 2014-04-01 00:30:00  21.7 ##  8 2014-04-01 00:35:00  20.9 ##  9 2014-04-01 00:40:00  18.4 ## 10 2014-04-01 00:45:00  18.7 ## # ℹ 4,022 more rows"},{"path":[]},{"path":"https://keras3.posit.co/dev/articles/examples/timeseries/timeseries_anomaly_detection.html","id":"timeseries-data-without-anomalies","dir":"Articles > Examples > Timeseries","previous_headings":"Visualize the data","what":"Timeseries data without anomalies","title":"Timeseries anomaly detection using an Autoencoder","text":"use following data training.","code":"plot_ts <- function(df) {   ggplot(df, aes(x = timestamp, y = value)) + geom_line() +     scale_x_datetime(date_breaks = \"1 day\", date_labels = \"%b-%d\") }  plot_ts(df_small_noise) + ggtitle(\"Without Anomaly\")"},{"path":"https://keras3.posit.co/dev/articles/examples/timeseries/timeseries_anomaly_detection.html","id":"timeseries-data-with-anomalies","dir":"Articles > Examples > Timeseries","previous_headings":"Visualize the data","what":"Timeseries data with anomalies","title":"Timeseries anomaly detection using an Autoencoder","text":"use following data testing see sudden jump data detected anomaly.","code":"plot_ts(df_daily_jumpsup) + ggtitle(\"With Anomaly\")"},{"path":"https://keras3.posit.co/dev/articles/examples/timeseries/timeseries_anomaly_detection.html","id":"prepare-training-data","dir":"Articles > Examples > Timeseries","previous_headings":"","what":"Prepare training data","title":"Timeseries anomaly detection using an Autoencoder","text":"Get data values training timeseries data file normalize value data. value every 5 mins 14 days. 24 * 60 / 5 = 288 timesteps per day 288 * 14 = 4032 data points total","code":"df_train <- df_small_noise |>   mutate(value = (value - mean(value)) / sd(value))  cat(\"Number of training samples:\", nrow(df_train), \"\\n\") ## Number of training samples: 4032"},{"path":"https://keras3.posit.co/dev/articles/examples/timeseries/timeseries_anomaly_detection.html","id":"create-sequences","dir":"Articles > Examples > Timeseries","previous_headings":"Prepare training data","what":"Create sequences","title":"Timeseries anomaly detection using an Autoencoder","text":"Create sequences combining TIME_STEPS contiguous data values training data.","code":"TIME_STEPS <- 288  as_dataset <- function(df) {   x <- as.matrix(df$value)   ds <- timeseries_dataset_from_array(x, NULL, sequence_length = TIME_STEPS)   # Because the dataset is small, cast TF Dataset to an R array for convenience.   ds |> as_array_iterator() |> iterate() |> bind_on_rows() }  x_train <- as_dataset(df_train) writeLines(sprintf(\"Training input shape: (%s)\", toString(dim(x_train)))) ## Training input shape: (3745, 288, 1)"},{"path":"https://keras3.posit.co/dev/articles/examples/timeseries/timeseries_anomaly_detection.html","id":"build-a-model","dir":"Articles > Examples > Timeseries","previous_headings":"","what":"Build a model","title":"Timeseries anomaly detection using an Autoencoder","text":"build convolutional reconstruction autoencoder model. model take input shape (batch_size, sequence_length, num_features) return output shape. case, sequence_length 288 num_features 1.","code":"model <- keras_model_sequential(input_shape = c(TIME_STEPS, 1)) |>   layer_conv_1d(     filters = 32, kernel_size = 7, padding = \"same\",     strides = 2, activation = \"relu\"   ) |>   layer_dropout(rate = 0.2) |>   layer_conv_1d(     filters = 16, kernel_size = 7, padding = \"same\",     strides = 2, activation = \"relu\"   ) |>   layer_conv_1d_transpose(     filters = 16, kernel_size = 7, padding = \"same\",     strides = 2, activation = \"relu\"   ) |>   layer_dropout(rate = 0.2) |>   layer_conv_1d_transpose(     filters = 32, kernel_size = 7, padding = \"same\",     strides = 2, activation = \"relu\"   ) |>   layer_conv_1d_transpose(filters = 1, kernel_size = 7, padding = \"same\")  model |> compile(optimizer=optimizer_adam(learning_rate=0.001), loss=\"mse\") model ## Model: \"sequential\" ## ┏━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━┓ ## ┃ Layer (type)                    ┃ Output Shape           ┃       Param # ┃ ## ┡━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━┩ ## │ conv1d (Conv1D)                 │ (None, 144, 32)        │           256 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ dropout (Dropout)               │ (None, 144, 32)        │             0 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ conv1d_1 (Conv1D)               │ (None, 72, 16)         │         3,600 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ conv1d_transpose                │ (None, 144, 16)        │         1,808 │ ## │ (Conv1DTranspose)               │                        │               │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ dropout_1 (Dropout)             │ (None, 144, 16)        │             0 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ conv1d_transpose_1              │ (None, 288, 32)        │         3,616 │ ## │ (Conv1DTranspose)               │                        │               │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ conv1d_transpose_2              │ (None, 288, 1)         │           225 │ ## │ (Conv1DTranspose)               │                        │               │ ## └─────────────────────────────────┴────────────────────────┴───────────────┘ ##  Total params: 9,505 (37.13 KB) ##  Trainable params: 9,505 (37.13 KB) ##  Non-trainable params: 0 (0.00 B)"},{"path":"https://keras3.posit.co/dev/articles/examples/timeseries/timeseries_anomaly_detection.html","id":"train-the-model","dir":"Articles > Examples > Timeseries","previous_headings":"","what":"Train the model","title":"Timeseries anomaly detection using an Autoencoder","text":"Please note using x_train input target since reconstruction model. Let’s plot training validation loss see training went.","code":"history = model |> fit(   x_train, x_train,   epochs = 50,   validation_split = 0.1,   callbacks = c(     callback_early_stopping(       monitor = \"val_loss\", patience = 5, mode = \"min\"     )   ) ) ## Epoch 1/50 ## 106/106 - 4s - 39ms/step - loss: 0.1863 - val_loss: 0.0316 ## Epoch 2/50 ## 106/106 - 0s - 2ms/step - loss: 0.0333 - val_loss: 0.0233 ## Epoch 3/50 ## 106/106 - 0s - 2ms/step - loss: 0.0248 - val_loss: 0.0201 ## Epoch 4/50 ## 106/106 - 0s - 2ms/step - loss: 0.0209 - val_loss: 0.0186 ## Epoch 5/50 ## 106/106 - 0s - 2ms/step - loss: 0.0179 - val_loss: 0.0146 ## Epoch 6/50 ## 106/106 - 0s - 2ms/step - loss: 0.0150 - val_loss: 0.0112 ## Epoch 7/50 ## 106/106 - 0s - 2ms/step - loss: 0.0127 - val_loss: 0.0094 ## Epoch 8/50 ## 106/106 - 0s - 2ms/step - loss: 0.0109 - val_loss: 0.0088 ## Epoch 9/50 ## 106/106 - 0s - 2ms/step - loss: 0.0096 - val_loss: 0.0083 ## Epoch 10/50 ## 106/106 - 0s - 2ms/step - loss: 0.0086 - val_loss: 0.0069 ## Epoch 11/50 ## 106/106 - 0s - 2ms/step - loss: 0.0078 - val_loss: 0.0061 ## Epoch 12/50 ## 106/106 - 0s - 2ms/step - loss: 0.0073 - val_loss: 0.0057 ## Epoch 13/50 ## 106/106 - 0s - 2ms/step - loss: 0.0068 - val_loss: 0.0053 ## Epoch 14/50 ## 106/106 - 0s - 2ms/step - loss: 0.0064 - val_loss: 0.0050 ## Epoch 15/50 ## 106/106 - 0s - 2ms/step - loss: 0.0061 - val_loss: 0.0046 ## Epoch 16/50 ## 106/106 - 0s - 2ms/step - loss: 0.0058 - val_loss: 0.0040 ## Epoch 17/50 ## 106/106 - 0s - 2ms/step - loss: 0.0056 - val_loss: 0.0041 ## Epoch 18/50 ## 106/106 - 0s - 2ms/step - loss: 0.0053 - val_loss: 0.0035 ## Epoch 19/50 ## 106/106 - 0s - 2ms/step - loss: 0.0050 - val_loss: 0.0037 ## Epoch 20/50 ## 106/106 - 0s - 2ms/step - loss: 0.0048 - val_loss: 0.0032 ## Epoch 21/50 ## 106/106 - 0s - 2ms/step - loss: 0.0046 - val_loss: 0.0031 ## Epoch 22/50 ## 106/106 - 0s - 2ms/step - loss: 0.0044 - val_loss: 0.0032 ## Epoch 23/50 ## 106/106 - 0s - 2ms/step - loss: 0.0042 - val_loss: 0.0033 ## Epoch 24/50 ## 106/106 - 0s - 2ms/step - loss: 0.0041 - val_loss: 0.0031 ## Epoch 25/50 ## 106/106 - 0s - 2ms/step - loss: 0.0039 - val_loss: 0.0033 ## Epoch 26/50 ## 106/106 - 0s - 2ms/step - loss: 0.0037 - val_loss: 0.0030 ## Epoch 27/50 ## 106/106 - 0s - 2ms/step - loss: 0.0035 - val_loss: 0.0032 ## Epoch 28/50 ## 106/106 - 0s - 2ms/step - loss: 0.0034 - val_loss: 0.0025 ## Epoch 29/50 ## 106/106 - 0s - 2ms/step - loss: 0.0033 - val_loss: 0.0028 ## Epoch 30/50 ## 106/106 - 0s - 2ms/step - loss: 0.0032 - val_loss: 0.0030 ## Epoch 31/50 ## 106/106 - 0s - 2ms/step - loss: 0.0031 - val_loss: 0.0024 ## Epoch 32/50 ## 106/106 - 0s - 2ms/step - loss: 0.0030 - val_loss: 0.0028 ## Epoch 33/50 ## 106/106 - 0s - 2ms/step - loss: 0.0029 - val_loss: 0.0027 ## Epoch 34/50 ## 106/106 - 0s - 2ms/step - loss: 0.0028 - val_loss: 0.0022 ## Epoch 35/50 ## 106/106 - 0s - 2ms/step - loss: 0.0027 - val_loss: 0.0024 ## Epoch 36/50 ## 106/106 - 0s - 2ms/step - loss: 0.0026 - val_loss: 0.0025 ## Epoch 37/50 ## 106/106 - 0s - 2ms/step - loss: 0.0026 - val_loss: 0.0027 ## Epoch 38/50 ## 106/106 - 0s - 2ms/step - loss: 0.0025 - val_loss: 0.0024 ## Epoch 39/50 ## 106/106 - 0s - 2ms/step - loss: 0.0025 - val_loss: 0.0024 plot(history)"},{"path":"https://keras3.posit.co/dev/articles/examples/timeseries/timeseries_anomaly_detection.html","id":"detecting-anomalies","dir":"Articles > Examples > Timeseries","previous_headings":"","what":"Detecting anomalies","title":"Timeseries anomaly detection using an Autoencoder","text":"detect anomalies determining well model can reconstruct input data. Find MAE loss training samples. Find max MAE loss value. worst model performed trying reconstruct sample. make threshold anomaly detection. reconstruction loss sample greater threshold value can infer model seeing pattern isn’t familiar . label sample anomaly.","code":"# Get train MAE loss. x_train_pred <- model |> predict(x_train) ## 118/118 - 0s - 4ms/step train_mae_loss <- apply(abs(x_train_pred - x_train), 1, mean)  hist(train_mae_loss, breaks = 50) # Get reconstruction loss threshold. threshold <- max(train_mae_loss) cat(\"Reconstruction error threshold: \", threshold, \"\\n\") ## Reconstruction error threshold:  0.03994837"},{"path":"https://keras3.posit.co/dev/articles/examples/timeseries/timeseries_anomaly_detection.html","id":"compare-recontruction","dir":"Articles > Examples > Timeseries","previous_headings":"Detecting anomalies","what":"Compare recontruction","title":"Timeseries anomaly detection using an Autoencoder","text":"Just fun, let’s see model recontructed first sample. 288 timesteps day 1 training dataset.","code":"# Checking how the first sequence is learnt plot(NULL, NULL, ylab = 'Value',      xlim = c(0, TIME_STEPS),      ylim = range(c(x_train[1,,], x_train_pred[1,,]))) lines(x_train[1,,]) lines(x_train_pred[1,,], col = 'red') legend(\"topleft\", lty = 1,        legend = c(\"actual\", \"predicted\"),        col = c(\"black\", \"red\"))"},{"path":"https://keras3.posit.co/dev/articles/examples/timeseries/timeseries_anomaly_detection.html","id":"prepare-test-data","dir":"Articles > Examples > Timeseries","previous_headings":"Detecting anomalies","what":"Prepare test data","title":"Timeseries anomaly detection using an Autoencoder","text":"","code":"df_test <- df_daily_jumpsup |>   mutate(value =            (value - mean(df_small_noise$value)) /              sd(df_small_noise$value))  df_test |> head() plot_ts(df_test) # Create sequences from test values. x_test <- as_dataset(df_test)  # Get test MAE loss. x_test_pred <- model |> predict(x_test) test_mae_loss <- apply(abs(x_test_pred - x_test), 1, mean)  hist(test_mae_loss, breaks = 50, xlab = \"test MAE loss\", ylab = \"No of samples\") # Detect all the samples which are anomalies. anomalies <- test_mae_loss > threshold cat(\"Number of anomaly samples:\", sum(anomalies), \"\\n\") cat(\"Indices of anomaly samples:\", which(anomalies), \"\\n\", fill = TRUE) ## # A tibble: 6 × 2 ##   timestamp            value ##   <dttm>               <dbl> ## 1 2014-04-01 00:00:00 -0.808 ## 2 2014-04-01 00:05:00 -0.781 ## 3 2014-04-01 00:10:00 -0.801 ## 4 2014-04-01 00:15:00 -0.746 ## 5 2014-04-01 00:20:00 -0.792 ## 6 2014-04-01 00:25:00 -0.802 ## 118/118 - 0s - 808us/step ## Number of anomaly samples: 483 ## Indices of anomaly samples: 216 217 218 219 220 221 396 507 793 794 795 797 ## 799 861 974 1659 1944 1945 1946 1947 2016 2020 2036 2044 2046 2052 2053 ## 2054 2056 2057 2058 2060 2061 2062 2064 2065 2066 2068 2069 2070 2072 2073 ## 2074 2076 2077 2078 2080 2081 2082 2084 2085 2086 2088 2089 2090 2092 2093 ## 2094 2096 2097 2098 2100 2101 2102 2104 2105 2108 2112 2113 2116 2117 2120 ## 2121 2124 2126 2141 2521 2522 2523 2525 2546 2698 2700 2702 2703 2704 2705 ## 2706 2707 2708 2709 2710 2711 2712 2713 2714 2715 2716 2717 2718 2719 2720 ## 2721 2722 2723 2724 2725 2726 2727 2728 2729 2730 2731 2732 2733 2734 2735 ## 2736 2737 2738 2739 2740 2741 2742 2743 2744 2745 2746 2747 2748 2749 2750 ## 2751 2752 2753 2754 2755 2756 2757 2758 2759 2760 2761 2762 2763 2764 2765 ## 2766 2767 2768 2769 2770 2771 2772 2773 2774 2775 2776 2777 2778 2779 2780 ## 2781 2782 2783 2784 2785 2786 2787 2788 2789 2790 2791 2792 2793 2794 2795 ## 2796 2797 2798 2799 2800 2801 2802 2803 2804 2805 2806 2807 2808 2809 2810 ## 2811 2812 2813 2814 2815 2816 2817 2818 2819 2820 2821 2822 2823 2824 2825 ## 2826 2827 2828 2829 2830 2831 2832 2833 2834 2835 2836 2837 2838 2839 2840 ## 2841 2842 2843 2844 2845 2846 2847 2848 2849 2850 2851 2852 2853 2854 2855 ## 2856 2857 2858 2859 2860 2861 2862 2863 2864 2865 2866 2867 2868 2869 2870 ## 2871 2872 2873 2874 2875 2876 2877 2878 2879 2880 2881 2882 2883 2884 2885 ## 2886 2887 2888 2889 2890 2891 2892 2893 2894 2895 2896 2897 2898 2899 2900 ## 2901 2902 2903 2904 2905 2906 2907 2908 2909 2910 2911 2912 2913 2914 2915 ## 2916 2917 2918 2919 2920 2921 2922 2923 2924 2925 2926 2927 2928 2929 2930 ## 2931 2932 2933 2934 2935 2936 2937 2938 2939 2940 2941 2942 2943 2944 2945 ## 2946 2947 2948 2949 2950 2951 2952 2953 2954 2955 2956 2957 2958 2959 2960 ## 2961 2962 2963 2964 2965 2966 2967 2968 2969 2970 2971 2972 2973 2974 2975 ## 2976 2977 2978 2979 2980 2981 2982 2983 2984 2985 2986 2987 2988 2989 2990 ## 2991 2992 2993 2994 2995 2996 2997 2998 2999 3000 3001 3002 3003 3004 3005 ## 3006 3007 3008 3009 3010 3011 3012 3013 3014 3015 3016 3017 3018 3019 3020 ## 3021 3022 3023 3024 3025 3026 3027 3028 3029 3030 3031 3032 3033 3034 3035 ## 3036 3037 3038 3039 3040 3041 3042 3043 3044 3045 3046 3047 3048 3049 3050 ## 3051 3052 3053 3054 3055 3056 3057 3058 3059 3060 3061 3062 3063 3064 3065 ## 3066 3067 3068 3069 3070 3071 3072 3073 3074 3075 3076 3077 3078 3079 3080 ## 3081 3082 3083 3084 3085 3086 3087 3088 3089 3090 3091 3092 3093 3094 3095 ## 3096 3097 3098 3099 3100 3101"},{"path":"https://keras3.posit.co/dev/articles/examples/timeseries/timeseries_anomaly_detection.html","id":"plot-anomalies","dir":"Articles > Examples > Timeseries","previous_headings":"","what":"Plot anomalies","title":"Timeseries anomaly detection using an Autoencoder","text":"now know samples data anomalies. , find corresponding timestamps original test data. using following method : Let’s say time_steps = 3 10 training values. x_train look like : 0, 1, 2 1, 2, 3 2, 3, 4 3, 4, 5 4, 5, 6 5, 6, 7 6, 7, 8 7, 8, 9 except initial final time_steps-1 data values, appear time_steps number samples. , know samples [(3, 4, 5), (4, 5, 6), (5, 6, 7)] anomalies, can say data point 5 anomaly. Let’s overlay anomalies original test data plot.","code":"is_anomaly <- test_mae_loss > threshold is_anomaly <- is_anomaly &   zoo::rollsum(is_anomaly, TIME_STEPS,                align = \"right\", na.pad = TRUE) >= TIME_STEPS  with(df_test, {   plot(value ~ timestamp, type = 'l', xaxt = 'n', las = 2)   axis.POSIXct(1, at = seq(timestamp[1], tail(timestamp, 1), by = \"days\"),                format = \"%b-%d\") })  with(df_test[which(is_anomaly),], {   points(value ~ timestamp, col = \"red\") })"},{"path":"https://keras3.posit.co/dev/articles/examples/timeseries/timeseries_classification_from_scratch.html","id":"introduction","dir":"Articles > Examples > Timeseries","previous_headings":"","what":"Introduction","title":"Timeseries classification from scratch","text":"example shows timeseries classification scratch, starting raw CSV timeseries files disk. demonstrate workflow FordA dataset UCR/UEA archive.","code":""},{"path":"https://keras3.posit.co/dev/articles/examples/timeseries/timeseries_classification_from_scratch.html","id":"setup","dir":"Articles > Examples > Timeseries","previous_headings":"","what":"Setup","title":"Timeseries classification from scratch","text":"","code":"library(keras3) use_backend(\"jax\")"},{"path":[]},{"path":"https://keras3.posit.co/dev/articles/examples/timeseries/timeseries_classification_from_scratch.html","id":"dataset-description","dir":"Articles > Examples > Timeseries","previous_headings":"Load the data: the FordA dataset","what":"Dataset description","title":"Timeseries classification from scratch","text":"dataset using called FordA. data comes UCR archive. dataset contains 3601 training instances another 1320 testing instances. timeseries corresponds measurement engine noise captured motor sensor. task, goal automatically detect presence specific issue engine. problem balanced binary classification task. full description dataset can found .","code":""},{"path":"https://keras3.posit.co/dev/articles/examples/timeseries/timeseries_classification_from_scratch.html","id":"read-the-tsv-data","dir":"Articles > Examples > Timeseries","previous_headings":"Load the data: the FordA dataset","what":"Read the TSV data","title":"Timeseries classification from scratch","text":"use FordA_TRAIN file training FordA_TEST file testing. simplicity dataset allows us demonstrate effectively use ConvNets timeseries classification. file, first column corresponds label.","code":"get_data <- function(path) {   if(path |> startsWith(\"https://\"))     path <- get_file(origin = path)  # cache file locally    data <- readr::read_tsv(     path, col_names = FALSE,     # Each row is: one integer (the label),     # followed by 500 doubles (the timeseries)     col_types = paste0(\"i\", strrep(\"d\", 500))   )    y <- as.matrix(data[[1]])   x <- as.matrix(data[,-1])   dimnames(x) <- dimnames(y) <- NULL    list(x, y) }  root_url <- \"https://raw.githubusercontent.com/hfawaz/cd-diagram/master/FordA/\" c(x_train, y_train) %<-% get_data(paste0(root_url, \"FordA_TRAIN.tsv\")) c(x_test, y_test) %<-% get_data(paste0(root_url, \"FordA_TEST.tsv\"))  str(keras3:::named_list(   x_train, y_train,   x_test, y_test )) ## List of 4 ##  $ x_train: num [1:3601, 1:500] -0.797 0.805 0.728 -0.234 -0.171 ... ##  $ y_train: int [1:3601, 1] -1 1 -1 -1 -1 1 1 1 1 1 ... ##  $ x_test : num [1:1320, 1:500] -0.14 0.334 0.717 1.24 -1.159 ... ##  $ y_test : int [1:1320, 1] -1 -1 -1 1 -1 1 -1 -1 1 1 ..."},{"path":"https://keras3.posit.co/dev/articles/examples/timeseries/timeseries_classification_from_scratch.html","id":"visualize-the-data","dir":"Articles > Examples > Timeseries","previous_headings":"","what":"Visualize the data","title":"Timeseries classification from scratch","text":"visualize one timeseries example class dataset.","code":"plot(NULL, main = \"Timeseries Data\",      xlab = \"Timepoints\",  ylab = \"Values\",      xlim = c(1, ncol(x_test)),      ylim = range(x_test)) grid() lines(x_test[match(-1, y_test), ], col = \"blue\") lines(x_test[match( 1, y_test), ], col = \"red\") legend(\"topright\", legend=c(\"label -1\", \"label 1\"), col=c(\"blue\", \"red\"), lty=1)"},{"path":"https://keras3.posit.co/dev/articles/examples/timeseries/timeseries_classification_from_scratch.html","id":"standardize-the-data","dir":"Articles > Examples > Timeseries","previous_headings":"","what":"Standardize the data","title":"Timeseries classification from scratch","text":"timeseries already single length (500). However, values usually various ranges. ideal neural network; general seek make input values normalized. specific dataset, data already z-normalized: timeseries sample mean equal zero standard deviation equal one. type normalization common timeseries classification problems, see Bagnall et al. (2016). Note timeseries data used univariate, meaning one channel per timeseries example. therefore transform timeseries multivariate one one channel using simple reshaping via numpy. allow us construct model easily applicable multivariate time series. Finally, order use sparse_categorical_crossentropy, count number classes beforehand. Now shuffle training set using validation_split option later training. Standardize labels positive integers. expected labels 0 1.","code":"dim(x_train) <- c(dim(x_train), 1) dim(x_test) <- c(dim(x_test), 1) num_classes <- length(unique(y_train)) c(x_train, y_train) %<-% listarrays::shuffle_rows(x_train, y_train) # idx <- sample.int(nrow(x_train)) # x_train %<>% .[idx,, ,drop = FALSE] # y_train %<>% .[idx,  ,drop = FALSE] y_train[y_train == -1L] <- 0L y_test[y_test == -1L] <- 0L"},{"path":"https://keras3.posit.co/dev/articles/examples/timeseries/timeseries_classification_from_scratch.html","id":"build-a-model","dir":"Articles > Examples > Timeseries","previous_headings":"","what":"Build a model","title":"Timeseries classification from scratch","text":"build Fully Convolutional Neural Network originally proposed paper. implementation based TF 2 version provided . following hyperparameters (kernel_size, filters, usage BatchNorm) found via random search using KerasTuner. plot chunk unnamed-chunk-9","code":"make_model <- function(input_shape) {   inputs <- keras_input(input_shape)    outputs <- inputs |>     # conv1     layer_conv_1d(filters = 64, kernel_size = 3, padding = \"same\") |>     layer_batch_normalization() |>     layer_activation_relu() |>     # conv2     layer_conv_1d(filters = 64, kernel_size = 3, padding = \"same\") |>     layer_batch_normalization() |>     layer_activation_relu() |>     # conv3     layer_conv_1d(filters = 64, kernel_size = 3, padding = \"same\") |>     layer_batch_normalization() |>     layer_activation_relu() |>     # pooling     layer_global_average_pooling_1d() |>     # final output     layer_dense(num_classes, activation = \"softmax\")    keras_model(inputs, outputs) }  model <- make_model(input_shape = dim(x_train)[-1]) model ## Model: \"functional\" ## ┏━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━┳━━━━━━━┓ ## ┃ Layer (type)                ┃ Output Shape          ┃    Param # ┃ Trai… ┃ ## ┡━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━╇━━━━━━━┩ ## │ input_layer (InputLayer)    │ (None, 500, 1)        │          0 │   -   │ ## ├─────────────────────────────┼───────────────────────┼────────────┼───────┤ ## │ conv1d (Conv1D)             │ (None, 500, 64)       │        256 │   Y   │ ## ├─────────────────────────────┼───────────────────────┼────────────┼───────┤ ## │ batch_normalization         │ (None, 500, 64)       │        256 │   Y   │ ## │ (BatchNormalization)        │                       │            │       │ ## ├─────────────────────────────┼───────────────────────┼────────────┼───────┤ ## │ re_lu (ReLU)                │ (None, 500, 64)       │          0 │   -   │ ## ├─────────────────────────────┼───────────────────────┼────────────┼───────┤ ## │ conv1d_1 (Conv1D)           │ (None, 500, 64)       │     12,352 │   Y   │ ## ├─────────────────────────────┼───────────────────────┼────────────┼───────┤ ## │ batch_normalization_1       │ (None, 500, 64)       │        256 │   Y   │ ## │ (BatchNormalization)        │                       │            │       │ ## ├─────────────────────────────┼───────────────────────┼────────────┼───────┤ ## │ re_lu_1 (ReLU)              │ (None, 500, 64)       │          0 │   -   │ ## ├─────────────────────────────┼───────────────────────┼────────────┼───────┤ ## │ conv1d_2 (Conv1D)           │ (None, 500, 64)       │     12,352 │   Y   │ ## ├─────────────────────────────┼───────────────────────┼────────────┼───────┤ ## │ batch_normalization_2       │ (None, 500, 64)       │        256 │   Y   │ ## │ (BatchNormalization)        │                       │            │       │ ## ├─────────────────────────────┼───────────────────────┼────────────┼───────┤ ## │ re_lu_2 (ReLU)              │ (None, 500, 64)       │          0 │   -   │ ## ├─────────────────────────────┼───────────────────────┼────────────┼───────┤ ## │ global_average_pooling1d    │ (None, 64)            │          0 │   -   │ ## │ (GlobalAveragePooling1D)    │                       │            │       │ ## ├─────────────────────────────┼───────────────────────┼────────────┼───────┤ ## │ dense (Dense)               │ (None, 2)             │        130 │   Y   │ ## └─────────────────────────────┴───────────────────────┴────────────┴───────┘ ##  Total params: 25,858 (101.01 KB) ##  Trainable params: 25,474 (99.51 KB) ##  Non-trainable params: 384 (1.50 KB) plot(model, show_shapes = TRUE)"},{"path":"https://keras3.posit.co/dev/articles/examples/timeseries/timeseries_classification_from_scratch.html","id":"train-the-model","dir":"Articles > Examples > Timeseries","previous_headings":"","what":"Train the model","title":"Timeseries classification from scratch","text":"","code":"epochs <- 500 batch_size <- 32  callbacks <- c(   callback_model_checkpoint(     \"best_model.keras\", save_best_only = TRUE,     monitor = \"val_loss\"   ),   callback_reduce_lr_on_plateau(     monitor = \"val_loss\", factor = 0.5,     patience = 20, min_lr = 0.0001   ),   callback_early_stopping(     monitor = \"val_loss\", patience = 50,     verbose = 1   ) )   model |> compile(   optimizer = \"adam\",   loss = \"sparse_categorical_crossentropy\",   metrics = \"sparse_categorical_accuracy\" )  history <- model |> fit(   x_train, y_train,   batch_size = batch_size,   epochs = epochs,   callbacks = callbacks,   validation_split = 0.2 ) ## Epoch 1/500 ## 90/90 - 3s - 32ms/step - loss: 0.5310 - sparse_categorical_accuracy: 0.7191 - val_loss: 0.7907 - val_sparse_categorical_accuracy: 0.4896 - learning_rate: 1.0000e-03 ## Epoch 2/500 ## 90/90 - 1s - 11ms/step - loss: 0.4799 - sparse_categorical_accuracy: 0.7590 - val_loss: 0.9127 - val_sparse_categorical_accuracy: 0.4896 - learning_rate: 1.0000e-03 ## Epoch 3/500 ## 90/90 - 0s - 2ms/step - loss: 0.4590 - sparse_categorical_accuracy: 0.7757 - val_loss: 0.8196 - val_sparse_categorical_accuracy: 0.4896 - learning_rate: 1.0000e-03 ## Epoch 4/500 ## 90/90 - 0s - 2ms/step - loss: 0.4088 - sparse_categorical_accuracy: 0.7983 - val_loss: 0.6425 - val_sparse_categorical_accuracy: 0.5534 - learning_rate: 1.0000e-03 ## Epoch 5/500 ## 90/90 - 0s - 2ms/step - loss: 0.4141 - sparse_categorical_accuracy: 0.7889 - val_loss: 0.5519 - val_sparse_categorical_accuracy: 0.6588 - learning_rate: 1.0000e-03 ## Epoch 6/500 ## 90/90 - 0s - 2ms/step - loss: 0.3929 - sparse_categorical_accuracy: 0.8163 - val_loss: 0.4225 - val_sparse_categorical_accuracy: 0.8294 - learning_rate: 1.0000e-03 ## Epoch 7/500 ## 90/90 - 0s - 1ms/step - loss: 0.3870 - sparse_categorical_accuracy: 0.8146 - val_loss: 0.6112 - val_sparse_categorical_accuracy: 0.6949 - learning_rate: 1.0000e-03 ## Epoch 8/500 ## 90/90 - 0s - 1ms/step - loss: 0.3717 - sparse_categorical_accuracy: 0.8247 - val_loss: 0.4257 - val_sparse_categorical_accuracy: 0.7559 - learning_rate: 1.0000e-03 ## Epoch 9/500 ## 90/90 - 0s - 2ms/step - loss: 0.3698 - sparse_categorical_accuracy: 0.8306 - val_loss: 0.3653 - val_sparse_categorical_accuracy: 0.8350 - learning_rate: 1.0000e-03 ## Epoch 10/500 ## 90/90 - 0s - 1ms/step - loss: 0.3634 - sparse_categorical_accuracy: 0.8260 - val_loss: 0.4390 - val_sparse_categorical_accuracy: 0.7836 - learning_rate: 1.0000e-03 ## Epoch 11/500 ## 90/90 - 0s - 1ms/step - loss: 0.3676 - sparse_categorical_accuracy: 0.8281 - val_loss: 0.4205 - val_sparse_categorical_accuracy: 0.7822 - learning_rate: 1.0000e-03 ## Epoch 12/500 ## 90/90 - 0s - 1ms/step - loss: 0.3467 - sparse_categorical_accuracy: 0.8406 - val_loss: 0.3697 - val_sparse_categorical_accuracy: 0.8460 - learning_rate: 1.0000e-03 ## Epoch 13/500 ## 90/90 - 0s - 2ms/step - loss: 0.3310 - sparse_categorical_accuracy: 0.8594 - val_loss: 0.3341 - val_sparse_categorical_accuracy: 0.8530 - learning_rate: 1.0000e-03 ## Epoch 14/500 ## 90/90 - 0s - 1ms/step - loss: 0.3350 - sparse_categorical_accuracy: 0.8514 - val_loss: 0.5723 - val_sparse_categorical_accuracy: 0.7198 - learning_rate: 1.0000e-03 ## Epoch 15/500 ## 90/90 - 0s - 1ms/step - loss: 0.3295 - sparse_categorical_accuracy: 0.8497 - val_loss: 1.0273 - val_sparse_categorical_accuracy: 0.5465 - learning_rate: 1.0000e-03 ## Epoch 16/500 ## 90/90 - 0s - 2ms/step - loss: 0.3244 - sparse_categorical_accuracy: 0.8587 - val_loss: 0.3219 - val_sparse_categorical_accuracy: 0.8544 - learning_rate: 1.0000e-03 ## Epoch 17/500 ## 90/90 - 0s - 1ms/step - loss: 0.3082 - sparse_categorical_accuracy: 0.8646 - val_loss: 0.4187 - val_sparse_categorical_accuracy: 0.7906 - learning_rate: 1.0000e-03 ## Epoch 18/500 ## 90/90 - 0s - 1ms/step - loss: 0.3022 - sparse_categorical_accuracy: 0.8715 - val_loss: 0.6302 - val_sparse_categorical_accuracy: 0.6824 - learning_rate: 1.0000e-03 ## Epoch 19/500 ## 90/90 - 0s - 1ms/step - loss: 0.2996 - sparse_categorical_accuracy: 0.8764 - val_loss: 0.3537 - val_sparse_categorical_accuracy: 0.8447 - learning_rate: 1.0000e-03 ## Epoch 20/500 ## 90/90 - 0s - 1ms/step - loss: 0.3053 - sparse_categorical_accuracy: 0.8715 - val_loss: 0.4925 - val_sparse_categorical_accuracy: 0.7309 - learning_rate: 1.0000e-03 ## Epoch 21/500 ## 90/90 - 0s - 1ms/step - loss: 0.2827 - sparse_categorical_accuracy: 0.8826 - val_loss: 0.3893 - val_sparse_categorical_accuracy: 0.7809 - learning_rate: 1.0000e-03 ## Epoch 22/500 ## 90/90 - 0s - 2ms/step - loss: 0.2892 - sparse_categorical_accuracy: 0.8795 - val_loss: 0.3128 - val_sparse_categorical_accuracy: 0.8863 - learning_rate: 1.0000e-03 ## Epoch 23/500 ## 90/90 - 0s - 2ms/step - loss: 0.2718 - sparse_categorical_accuracy: 0.8889 - val_loss: 0.3083 - val_sparse_categorical_accuracy: 0.8571 - learning_rate: 1.0000e-03 ## Epoch 24/500 ## 90/90 - 0s - 1ms/step - loss: 0.2691 - sparse_categorical_accuracy: 0.8934 - val_loss: 0.5609 - val_sparse_categorical_accuracy: 0.7420 - learning_rate: 1.0000e-03 ## Epoch 25/500 ## 90/90 - 0s - 1ms/step - loss: 0.2635 - sparse_categorical_accuracy: 0.8948 - val_loss: 0.5264 - val_sparse_categorical_accuracy: 0.7573 - learning_rate: 1.0000e-03 ## Epoch 26/500 ## 90/90 - 0s - 1ms/step - loss: 0.3109 - sparse_categorical_accuracy: 0.8667 - val_loss: 0.3934 - val_sparse_categorical_accuracy: 0.7878 - learning_rate: 1.0000e-03 ## Epoch 27/500 ## 90/90 - 0s - 2ms/step - loss: 0.2599 - sparse_categorical_accuracy: 0.8969 - val_loss: 0.2974 - val_sparse_categorical_accuracy: 0.8558 - learning_rate: 1.0000e-03 ## Epoch 28/500 ## 90/90 - 0s - 1ms/step - loss: 0.2805 - sparse_categorical_accuracy: 0.8823 - val_loss: 0.4026 - val_sparse_categorical_accuracy: 0.8183 - learning_rate: 1.0000e-03 ## Epoch 29/500 ## 90/90 - 0s - 2ms/step - loss: 0.2517 - sparse_categorical_accuracy: 0.8948 - val_loss: 0.2699 - val_sparse_categorical_accuracy: 0.8849 - learning_rate: 1.0000e-03 ## Epoch 30/500 ## 90/90 - 0s - 1ms/step - loss: 0.2581 - sparse_categorical_accuracy: 0.8920 - val_loss: 0.2743 - val_sparse_categorical_accuracy: 0.8752 - learning_rate: 1.0000e-03 ## Epoch 31/500 ## 90/90 - 0s - 1ms/step - loss: 0.2550 - sparse_categorical_accuracy: 0.8986 - val_loss: 0.2936 - val_sparse_categorical_accuracy: 0.8655 - learning_rate: 1.0000e-03 ## Epoch 32/500 ## 90/90 - 0s - 2ms/step - loss: 0.2528 - sparse_categorical_accuracy: 0.8934 - val_loss: 0.2528 - val_sparse_categorical_accuracy: 0.8904 - learning_rate: 1.0000e-03 ## Epoch 33/500 ## 90/90 - 0s - 1ms/step - loss: 0.2390 - sparse_categorical_accuracy: 0.9062 - val_loss: 0.2961 - val_sparse_categorical_accuracy: 0.8571 - learning_rate: 1.0000e-03 ## Epoch 34/500 ## 90/90 - 0s - 1ms/step - loss: 0.2327 - sparse_categorical_accuracy: 0.9049 - val_loss: 0.3039 - val_sparse_categorical_accuracy: 0.8558 - learning_rate: 1.0000e-03 ## Epoch 35/500 ## 90/90 - 0s - 1ms/step - loss: 0.2447 - sparse_categorical_accuracy: 0.9021 - val_loss: 0.3123 - val_sparse_categorical_accuracy: 0.8710 - learning_rate: 1.0000e-03 ## Epoch 36/500 ## 90/90 - 0s - 1ms/step - loss: 0.2340 - sparse_categorical_accuracy: 0.9069 - val_loss: 0.2912 - val_sparse_categorical_accuracy: 0.8821 - learning_rate: 1.0000e-03 ## Epoch 37/500 ## 90/90 - 0s - 1ms/step - loss: 0.2352 - sparse_categorical_accuracy: 0.8997 - val_loss: 0.3983 - val_sparse_categorical_accuracy: 0.8017 - learning_rate: 1.0000e-03 ## Epoch 38/500 ## 90/90 - 0s - 1ms/step - loss: 0.2294 - sparse_categorical_accuracy: 0.9139 - val_loss: 0.3342 - val_sparse_categorical_accuracy: 0.8447 - learning_rate: 1.0000e-03 ## Epoch 39/500 ## 90/90 - 0s - 1ms/step - loss: 0.2268 - sparse_categorical_accuracy: 0.9135 - val_loss: 0.4730 - val_sparse_categorical_accuracy: 0.7975 - learning_rate: 1.0000e-03 ## Epoch 40/500 ## 90/90 - 0s - 1ms/step - loss: 0.2202 - sparse_categorical_accuracy: 0.9111 - val_loss: 0.3007 - val_sparse_categorical_accuracy: 0.8738 - learning_rate: 1.0000e-03 ## Epoch 41/500 ## 90/90 - 0s - 1ms/step - loss: 0.2064 - sparse_categorical_accuracy: 0.9233 - val_loss: 0.2924 - val_sparse_categorical_accuracy: 0.8849 - learning_rate: 1.0000e-03 ## Epoch 42/500 ## 90/90 - 0s - 1ms/step - loss: 0.1966 - sparse_categorical_accuracy: 0.9316 - val_loss: 0.8303 - val_sparse_categorical_accuracy: 0.6935 - learning_rate: 1.0000e-03 ## Epoch 43/500 ## 90/90 - 0s - 2ms/step - loss: 0.2074 - sparse_categorical_accuracy: 0.9212 - val_loss: 0.2334 - val_sparse_categorical_accuracy: 0.9043 - learning_rate: 1.0000e-03 ## Epoch 44/500 ## 90/90 - 0s - 1ms/step - loss: 0.1961 - sparse_categorical_accuracy: 0.9313 - val_loss: 0.2343 - val_sparse_categorical_accuracy: 0.9182 - learning_rate: 1.0000e-03 ## Epoch 45/500 ## 90/90 - 0s - 1ms/step - loss: 0.1923 - sparse_categorical_accuracy: 0.9302 - val_loss: 1.1167 - val_sparse_categorical_accuracy: 0.6755 - learning_rate: 1.0000e-03 ## Epoch 46/500 ## 90/90 - 0s - 1ms/step - loss: 0.1823 - sparse_categorical_accuracy: 0.9347 - val_loss: 0.8034 - val_sparse_categorical_accuracy: 0.6935 - learning_rate: 1.0000e-03 ## Epoch 47/500 ## 90/90 - 0s - 1ms/step - loss: 0.1635 - sparse_categorical_accuracy: 0.9444 - val_loss: 0.7104 - val_sparse_categorical_accuracy: 0.7115 - learning_rate: 1.0000e-03 ## Epoch 48/500 ## 90/90 - 0s - 2ms/step - loss: 0.1587 - sparse_categorical_accuracy: 0.9434 - val_loss: 0.2001 - val_sparse_categorical_accuracy: 0.9196 - learning_rate: 1.0000e-03 ## Epoch 49/500 ## 90/90 - 0s - 1ms/step - loss: 0.1492 - sparse_categorical_accuracy: 0.9483 - val_loss: 0.3399 - val_sparse_categorical_accuracy: 0.8294 - learning_rate: 1.0000e-03 ## Epoch 50/500 ## 90/90 - 0s - 1ms/step - loss: 0.1738 - sparse_categorical_accuracy: 0.9382 - val_loss: 0.2680 - val_sparse_categorical_accuracy: 0.8890 - learning_rate: 1.0000e-03 ## Epoch 51/500 ## 90/90 - 0s - 1ms/step - loss: 0.1530 - sparse_categorical_accuracy: 0.9486 - val_loss: 0.2891 - val_sparse_categorical_accuracy: 0.8544 - learning_rate: 1.0000e-03 ## Epoch 52/500 ## 90/90 - 0s - 1ms/step - loss: 0.1320 - sparse_categorical_accuracy: 0.9538 - val_loss: 1.7182 - val_sparse_categorical_accuracy: 0.7032 - learning_rate: 1.0000e-03 ## Epoch 53/500 ## 90/90 - 0s - 1ms/step - loss: 0.1277 - sparse_categorical_accuracy: 0.9559 - val_loss: 1.4269 - val_sparse_categorical_accuracy: 0.7254 - learning_rate: 1.0000e-03 ## Epoch 54/500 ## 90/90 - 0s - 1ms/step - loss: 0.1249 - sparse_categorical_accuracy: 0.9611 - val_loss: 0.7962 - val_sparse_categorical_accuracy: 0.7365 - learning_rate: 1.0000e-03 ## Epoch 55/500 ## 90/90 - 0s - 1ms/step - loss: 0.1214 - sparse_categorical_accuracy: 0.9611 - val_loss: 0.2158 - val_sparse_categorical_accuracy: 0.9071 - learning_rate: 1.0000e-03 ## Epoch 56/500 ## 90/90 - 0s - 1ms/step - loss: 0.1258 - sparse_categorical_accuracy: 0.9604 - val_loss: 0.6349 - val_sparse_categorical_accuracy: 0.7642 - learning_rate: 1.0000e-03 ## Epoch 57/500 ## 90/90 - 0s - 1ms/step - loss: 0.1268 - sparse_categorical_accuracy: 0.9604 - val_loss: 0.2129 - val_sparse_categorical_accuracy: 0.9071 - learning_rate: 1.0000e-03 ## Epoch 58/500 ## 90/90 - 0s - 1ms/step - loss: 0.1088 - sparse_categorical_accuracy: 0.9639 - val_loss: 0.7215 - val_sparse_categorical_accuracy: 0.6657 - learning_rate: 1.0000e-03 ## Epoch 59/500 ## 90/90 - 0s - 2ms/step - loss: 0.1219 - sparse_categorical_accuracy: 0.9569 - val_loss: 0.1600 - val_sparse_categorical_accuracy: 0.9320 - learning_rate: 1.0000e-03 ## Epoch 60/500 ## 90/90 - 0s - 1ms/step - loss: 0.1190 - sparse_categorical_accuracy: 0.9642 - val_loss: 0.3411 - val_sparse_categorical_accuracy: 0.8558 - learning_rate: 1.0000e-03 ## Epoch 61/500 ## 90/90 - 0s - 1ms/step - loss: 0.1177 - sparse_categorical_accuracy: 0.9653 - val_loss: 0.3984 - val_sparse_categorical_accuracy: 0.8502 - learning_rate: 1.0000e-03 ## Epoch 62/500 ## 90/90 - 0s - 1ms/step - loss: 0.1309 - sparse_categorical_accuracy: 0.9542 - val_loss: 0.3462 - val_sparse_categorical_accuracy: 0.8211 - learning_rate: 1.0000e-03 ## Epoch 63/500 ## 90/90 - 0s - 1ms/step - loss: 0.1291 - sparse_categorical_accuracy: 0.9514 - val_loss: 0.2303 - val_sparse_categorical_accuracy: 0.8960 - learning_rate: 1.0000e-03 ## Epoch 64/500 ## 90/90 - 0s - 1ms/step - loss: 0.1200 - sparse_categorical_accuracy: 0.9628 - val_loss: 0.1735 - val_sparse_categorical_accuracy: 0.9223 - learning_rate: 1.0000e-03 ## Epoch 65/500 ## 90/90 - 0s - 1ms/step - loss: 0.1138 - sparse_categorical_accuracy: 0.9625 - val_loss: 0.5628 - val_sparse_categorical_accuracy: 0.7920 - learning_rate: 1.0000e-03 ## Epoch 66/500 ## 90/90 - 0s - 2ms/step - loss: 0.1065 - sparse_categorical_accuracy: 0.9667 - val_loss: 0.1540 - val_sparse_categorical_accuracy: 0.9417 - learning_rate: 1.0000e-03 ## Epoch 67/500 ## 90/90 - 0s - 1ms/step - loss: 0.1156 - sparse_categorical_accuracy: 0.9618 - val_loss: 0.2319 - val_sparse_categorical_accuracy: 0.9043 - learning_rate: 1.0000e-03 ## Epoch 68/500 ## 90/90 - 0s - 1ms/step - loss: 0.1100 - sparse_categorical_accuracy: 0.9597 - val_loss: 0.8668 - val_sparse_categorical_accuracy: 0.7406 - learning_rate: 1.0000e-03 ## Epoch 69/500 ## 90/90 - 0s - 2ms/step - loss: 0.1165 - sparse_categorical_accuracy: 0.9618 - val_loss: 0.1460 - val_sparse_categorical_accuracy: 0.9515 - learning_rate: 1.0000e-03 ## Epoch 70/500 ## 90/90 - 0s - 1ms/step - loss: 0.1178 - sparse_categorical_accuracy: 0.9608 - val_loss: 0.8988 - val_sparse_categorical_accuracy: 0.7046 - learning_rate: 1.0000e-03 ## Epoch 71/500 ## 90/90 - 0s - 1ms/step - loss: 0.1068 - sparse_categorical_accuracy: 0.9642 - val_loss: 0.3645 - val_sparse_categorical_accuracy: 0.8585 - learning_rate: 1.0000e-03 ## Epoch 72/500 ## 90/90 - 0s - 1ms/step - loss: 0.1106 - sparse_categorical_accuracy: 0.9653 - val_loss: 0.2813 - val_sparse_categorical_accuracy: 0.8932 - learning_rate: 1.0000e-03 ## Epoch 73/500 ## 90/90 - 0s - 1ms/step - loss: 0.1117 - sparse_categorical_accuracy: 0.9632 - val_loss: 1.8641 - val_sparse_categorical_accuracy: 0.6741 - learning_rate: 1.0000e-03 ## Epoch 74/500 ## 90/90 - 0s - 1ms/step - loss: 0.1058 - sparse_categorical_accuracy: 0.9625 - val_loss: 1.4979 - val_sparse_categorical_accuracy: 0.6893 - learning_rate: 1.0000e-03 ## Epoch 75/500 ## 90/90 - 0s - 1ms/step - loss: 0.1223 - sparse_categorical_accuracy: 0.9594 - val_loss: 0.1508 - val_sparse_categorical_accuracy: 0.9473 - learning_rate: 1.0000e-03 ## Epoch 76/500 ## 90/90 - 0s - 2ms/step - loss: 0.1140 - sparse_categorical_accuracy: 0.9601 - val_loss: 0.1400 - val_sparse_categorical_accuracy: 0.9320 - learning_rate: 1.0000e-03 ## Epoch 77/500 ## 90/90 - 0s - 1ms/step - loss: 0.1050 - sparse_categorical_accuracy: 0.9639 - val_loss: 0.2169 - val_sparse_categorical_accuracy: 0.9126 - learning_rate: 1.0000e-03 ## Epoch 78/500 ## 90/90 - 0s - 1ms/step - loss: 0.0986 - sparse_categorical_accuracy: 0.9712 - val_loss: 0.3240 - val_sparse_categorical_accuracy: 0.8724 - learning_rate: 1.0000e-03 ## Epoch 79/500 ## 90/90 - 0s - 1ms/step - loss: 0.1081 - sparse_categorical_accuracy: 0.9632 - val_loss: 0.1525 - val_sparse_categorical_accuracy: 0.9362 - learning_rate: 1.0000e-03 ## Epoch 80/500 ## 90/90 - 0s - 1ms/step - loss: 0.0995 - sparse_categorical_accuracy: 0.9667 - val_loss: 0.3123 - val_sparse_categorical_accuracy: 0.8641 - learning_rate: 1.0000e-03 ## Epoch 81/500 ## 90/90 - 0s - 1ms/step - loss: 0.1058 - sparse_categorical_accuracy: 0.9632 - val_loss: 0.2851 - val_sparse_categorical_accuracy: 0.8779 - learning_rate: 1.0000e-03 ## Epoch 82/500 ## 90/90 - 0s - 1ms/step - loss: 0.0958 - sparse_categorical_accuracy: 0.9681 - val_loss: 0.1417 - val_sparse_categorical_accuracy: 0.9362 - learning_rate: 1.0000e-03 ## Epoch 83/500 ## 90/90 - 0s - 1ms/step - loss: 0.1005 - sparse_categorical_accuracy: 0.9660 - val_loss: 0.7325 - val_sparse_categorical_accuracy: 0.7157 - learning_rate: 1.0000e-03 ## Epoch 84/500 ## 90/90 - 0s - 2ms/step - loss: 0.0966 - sparse_categorical_accuracy: 0.9684 - val_loss: 0.1322 - val_sparse_categorical_accuracy: 0.9404 - learning_rate: 1.0000e-03 ## Epoch 85/500 ## 90/90 - 0s - 2ms/step - loss: 0.1018 - sparse_categorical_accuracy: 0.9667 - val_loss: 0.1202 - val_sparse_categorical_accuracy: 0.9473 - learning_rate: 1.0000e-03 ## Epoch 86/500 ## 90/90 - 0s - 1ms/step - loss: 0.0995 - sparse_categorical_accuracy: 0.9649 - val_loss: 0.2956 - val_sparse_categorical_accuracy: 0.8766 - learning_rate: 1.0000e-03 ## Epoch 87/500 ## 90/90 - 0s - 1ms/step - loss: 0.1068 - sparse_categorical_accuracy: 0.9632 - val_loss: 0.3050 - val_sparse_categorical_accuracy: 0.8710 - learning_rate: 1.0000e-03 ## Epoch 88/500 ## 90/90 - 0s - 1ms/step - loss: 0.1002 - sparse_categorical_accuracy: 0.9674 - val_loss: 0.1675 - val_sparse_categorical_accuracy: 0.9196 - learning_rate: 1.0000e-03 ## Epoch 89/500 ## 90/90 - 0s - 1ms/step - loss: 0.1042 - sparse_categorical_accuracy: 0.9608 - val_loss: 0.7157 - val_sparse_categorical_accuracy: 0.7115 - learning_rate: 1.0000e-03 ## Epoch 90/500 ## 90/90 - 0s - 1ms/step - loss: 0.0998 - sparse_categorical_accuracy: 0.9694 - val_loss: 0.1576 - val_sparse_categorical_accuracy: 0.9251 - learning_rate: 1.0000e-03 ## Epoch 91/500 ## 90/90 - 0s - 1ms/step - loss: 0.1187 - sparse_categorical_accuracy: 0.9583 - val_loss: 0.4125 - val_sparse_categorical_accuracy: 0.8031 - learning_rate: 1.0000e-03 ## Epoch 92/500 ## 90/90 - 0s - 1ms/step - loss: 0.0973 - sparse_categorical_accuracy: 0.9688 - val_loss: 0.3084 - val_sparse_categorical_accuracy: 0.8696 - learning_rate: 1.0000e-03 ## Epoch 93/500 ## 90/90 - 0s - 1ms/step - loss: 0.0972 - sparse_categorical_accuracy: 0.9677 - val_loss: 0.1446 - val_sparse_categorical_accuracy: 0.9334 - learning_rate: 1.0000e-03 ## Epoch 94/500 ## 90/90 - 0s - 1ms/step - loss: 0.0965 - sparse_categorical_accuracy: 0.9632 - val_loss: 0.1265 - val_sparse_categorical_accuracy: 0.9584 - learning_rate: 1.0000e-03 ## Epoch 95/500 ## 90/90 - 0s - 2ms/step - loss: 0.0927 - sparse_categorical_accuracy: 0.9667 - val_loss: 0.1507 - val_sparse_categorical_accuracy: 0.9376 - learning_rate: 1.0000e-03 ## Epoch 96/500 ## 90/90 - 0s - 1ms/step - loss: 0.1050 - sparse_categorical_accuracy: 0.9618 - val_loss: 0.5005 - val_sparse_categorical_accuracy: 0.8086 - learning_rate: 1.0000e-03 ## Epoch 97/500 ## 90/90 - 0s - 1ms/step - loss: 0.1134 - sparse_categorical_accuracy: 0.9597 - val_loss: 0.1732 - val_sparse_categorical_accuracy: 0.9390 - learning_rate: 1.0000e-03 ## Epoch 98/500 ## 90/90 - 0s - 1ms/step - loss: 0.0946 - sparse_categorical_accuracy: 0.9698 - val_loss: 1.8677 - val_sparse_categorical_accuracy: 0.6949 - learning_rate: 1.0000e-03 ## Epoch 99/500 ## 90/90 - 0s - 1ms/step - loss: 0.1018 - sparse_categorical_accuracy: 0.9632 - val_loss: 1.1308 - val_sparse_categorical_accuracy: 0.6782 - learning_rate: 1.0000e-03 ## Epoch 100/500 ## 90/90 - 0s - 1ms/step - loss: 0.0990 - sparse_categorical_accuracy: 0.9674 - val_loss: 0.3846 - val_sparse_categorical_accuracy: 0.8558 - learning_rate: 1.0000e-03 ## Epoch 101/500 ## 90/90 - 0s - 1ms/step - loss: 0.0891 - sparse_categorical_accuracy: 0.9705 - val_loss: 0.1480 - val_sparse_categorical_accuracy: 0.9362 - learning_rate: 1.0000e-03 ## Epoch 102/500 ## 90/90 - 0s - 1ms/step - loss: 0.1003 - sparse_categorical_accuracy: 0.9677 - val_loss: 0.1713 - val_sparse_categorical_accuracy: 0.9390 - learning_rate: 1.0000e-03 ## Epoch 103/500 ## 90/90 - 0s - 1ms/step - loss: 0.0910 - sparse_categorical_accuracy: 0.9722 - val_loss: 0.1415 - val_sparse_categorical_accuracy: 0.9542 - learning_rate: 1.0000e-03 ## Epoch 104/500 ## 90/90 - 0s - 1ms/step - loss: 0.0998 - sparse_categorical_accuracy: 0.9622 - val_loss: 0.5677 - val_sparse_categorical_accuracy: 0.7753 - learning_rate: 1.0000e-03 ## Epoch 105/500 ## 90/90 - 0s - 1ms/step - loss: 0.1037 - sparse_categorical_accuracy: 0.9649 - val_loss: 0.9369 - val_sparse_categorical_accuracy: 0.6241 - learning_rate: 1.0000e-03 ## Epoch 106/500 ## 90/90 - 1s - 10ms/step - loss: 0.0850 - sparse_categorical_accuracy: 0.9708 - val_loss: 0.1388 - val_sparse_categorical_accuracy: 0.9487 - learning_rate: 5.0000e-04 ## Epoch 107/500 ## 90/90 - 0s - 1ms/step - loss: 0.0833 - sparse_categorical_accuracy: 0.9743 - val_loss: 0.1413 - val_sparse_categorical_accuracy: 0.9376 - learning_rate: 5.0000e-04 ## Epoch 108/500 ## 90/90 - 0s - 1ms/step - loss: 0.0775 - sparse_categorical_accuracy: 0.9767 - val_loss: 0.1223 - val_sparse_categorical_accuracy: 0.9445 - learning_rate: 5.0000e-04 ## Epoch 109/500 ## 90/90 - 0s - 1ms/step - loss: 0.0837 - sparse_categorical_accuracy: 0.9753 - val_loss: 0.1275 - val_sparse_categorical_accuracy: 0.9528 - learning_rate: 5.0000e-04 ## Epoch 110/500 ## 90/90 - 0s - 1ms/step - loss: 0.0859 - sparse_categorical_accuracy: 0.9705 - val_loss: 0.1209 - val_sparse_categorical_accuracy: 0.9487 - learning_rate: 5.0000e-04 ## Epoch 111/500 ## 90/90 - 0s - 2ms/step - loss: 0.0780 - sparse_categorical_accuracy: 0.9757 - val_loss: 0.1142 - val_sparse_categorical_accuracy: 0.9584 - learning_rate: 5.0000e-04 ## Epoch 112/500 ## 90/90 - 0s - 1ms/step - loss: 0.0801 - sparse_categorical_accuracy: 0.9736 - val_loss: 0.1316 - val_sparse_categorical_accuracy: 0.9459 - learning_rate: 5.0000e-04 ## Epoch 113/500 ## 90/90 - 0s - 1ms/step - loss: 0.0740 - sparse_categorical_accuracy: 0.9753 - val_loss: 0.3385 - val_sparse_categorical_accuracy: 0.8544 - learning_rate: 5.0000e-04 ## Epoch 114/500 ## 90/90 - 0s - 1ms/step - loss: 0.0907 - sparse_categorical_accuracy: 0.9722 - val_loss: 0.1222 - val_sparse_categorical_accuracy: 0.9612 - learning_rate: 5.0000e-04 ## Epoch 115/500 ## 90/90 - 0s - 2ms/step - loss: 0.0767 - sparse_categorical_accuracy: 0.9733 - val_loss: 0.1115 - val_sparse_categorical_accuracy: 0.9584 - learning_rate: 5.0000e-04 ## Epoch 116/500 ## 90/90 - 0s - 1ms/step - loss: 0.0786 - sparse_categorical_accuracy: 0.9750 - val_loss: 0.1667 - val_sparse_categorical_accuracy: 0.9431 - learning_rate: 5.0000e-04 ## Epoch 117/500 ## 90/90 - 0s - 1ms/step - loss: 0.0760 - sparse_categorical_accuracy: 0.9750 - val_loss: 0.1519 - val_sparse_categorical_accuracy: 0.9376 - learning_rate: 5.0000e-04 ## Epoch 118/500 ## 90/90 - 0s - 1ms/step - loss: 0.0797 - sparse_categorical_accuracy: 0.9722 - val_loss: 0.1555 - val_sparse_categorical_accuracy: 0.9459 - learning_rate: 5.0000e-04 ## Epoch 119/500 ## 90/90 - 0s - 1ms/step - loss: 0.0793 - sparse_categorical_accuracy: 0.9729 - val_loss: 0.1252 - val_sparse_categorical_accuracy: 0.9598 - learning_rate: 5.0000e-04 ## Epoch 120/500 ## 90/90 - 0s - 1ms/step - loss: 0.0803 - sparse_categorical_accuracy: 0.9747 - val_loss: 0.2298 - val_sparse_categorical_accuracy: 0.9015 - learning_rate: 5.0000e-04 ## Epoch 121/500 ## 90/90 - 0s - 1ms/step - loss: 0.0793 - sparse_categorical_accuracy: 0.9736 - val_loss: 0.1140 - val_sparse_categorical_accuracy: 0.9528 - learning_rate: 5.0000e-04 ## Epoch 122/500 ## 90/90 - 0s - 1ms/step - loss: 0.0795 - sparse_categorical_accuracy: 0.9740 - val_loss: 0.1154 - val_sparse_categorical_accuracy: 0.9542 - learning_rate: 5.0000e-04 ## Epoch 123/500 ## 90/90 - 0s - 1ms/step - loss: 0.0768 - sparse_categorical_accuracy: 0.9740 - val_loss: 0.3421 - val_sparse_categorical_accuracy: 0.8724 - learning_rate: 5.0000e-04 ## Epoch 124/500 ## 90/90 - 0s - 1ms/step - loss: 0.0848 - sparse_categorical_accuracy: 0.9694 - val_loss: 0.1126 - val_sparse_categorical_accuracy: 0.9598 - learning_rate: 5.0000e-04 ## Epoch 125/500 ## 90/90 - 0s - 1ms/step - loss: 0.0800 - sparse_categorical_accuracy: 0.9722 - val_loss: 0.1821 - val_sparse_categorical_accuracy: 0.9293 - learning_rate: 5.0000e-04 ## Epoch 126/500 ## 90/90 - 0s - 2ms/step - loss: 0.0787 - sparse_categorical_accuracy: 0.9753 - val_loss: 0.1105 - val_sparse_categorical_accuracy: 0.9570 - learning_rate: 5.0000e-04 ## Epoch 127/500 ## 90/90 - 0s - 2ms/step - loss: 0.0799 - sparse_categorical_accuracy: 0.9760 - val_loss: 0.1091 - val_sparse_categorical_accuracy: 0.9584 - learning_rate: 5.0000e-04 ## Epoch 128/500 ## 90/90 - 0s - 1ms/step - loss: 0.0811 - sparse_categorical_accuracy: 0.9736 - val_loss: 0.1287 - val_sparse_categorical_accuracy: 0.9515 - learning_rate: 5.0000e-04 ## Epoch 129/500 ## 90/90 - 0s - 1ms/step - loss: 0.0797 - sparse_categorical_accuracy: 0.9736 - val_loss: 0.1143 - val_sparse_categorical_accuracy: 0.9584 - learning_rate: 5.0000e-04 ## Epoch 130/500 ## 90/90 - 0s - 1ms/step - loss: 0.0850 - sparse_categorical_accuracy: 0.9750 - val_loss: 0.1581 - val_sparse_categorical_accuracy: 0.9376 - learning_rate: 5.0000e-04 ## Epoch 131/500 ## 90/90 - 0s - 1ms/step - loss: 0.0707 - sparse_categorical_accuracy: 0.9792 - val_loss: 0.1129 - val_sparse_categorical_accuracy: 0.9639 - learning_rate: 5.0000e-04 ## Epoch 132/500 ## 90/90 - 0s - 1ms/step - loss: 0.0725 - sparse_categorical_accuracy: 0.9740 - val_loss: 0.1425 - val_sparse_categorical_accuracy: 0.9501 - learning_rate: 5.0000e-04 ## Epoch 133/500 ## 90/90 - 0s - 2ms/step - loss: 0.0705 - sparse_categorical_accuracy: 0.9764 - val_loss: 0.1098 - val_sparse_categorical_accuracy: 0.9612 - learning_rate: 5.0000e-04 ## Epoch 134/500 ## 90/90 - 0s - 1ms/step - loss: 0.0767 - sparse_categorical_accuracy: 0.9736 - val_loss: 0.1884 - val_sparse_categorical_accuracy: 0.9376 - learning_rate: 5.0000e-04 ## Epoch 135/500 ## 90/90 - 0s - 1ms/step - loss: 0.0775 - sparse_categorical_accuracy: 0.9743 - val_loss: 0.2899 - val_sparse_categorical_accuracy: 0.8988 - learning_rate: 5.0000e-04 ## Epoch 136/500 ## 90/90 - 0s - 1ms/step - loss: 0.0807 - sparse_categorical_accuracy: 0.9701 - val_loss: 0.1152 - val_sparse_categorical_accuracy: 0.9584 - learning_rate: 5.0000e-04 ## Epoch 137/500 ## 90/90 - 0s - 1ms/step - loss: 0.0839 - sparse_categorical_accuracy: 0.9750 - val_loss: 0.3530 - val_sparse_categorical_accuracy: 0.8793 - learning_rate: 5.0000e-04 ## Epoch 138/500 ## 90/90 - 0s - 1ms/step - loss: 0.0770 - sparse_categorical_accuracy: 0.9743 - val_loss: 0.1267 - val_sparse_categorical_accuracy: 0.9473 - learning_rate: 5.0000e-04 ## Epoch 139/500 ## 90/90 - 0s - 2ms/step - loss: 0.0845 - sparse_categorical_accuracy: 0.9719 - val_loss: 0.1043 - val_sparse_categorical_accuracy: 0.9626 - learning_rate: 5.0000e-04 ## Epoch 140/500 ## 90/90 - 0s - 1ms/step - loss: 0.0701 - sparse_categorical_accuracy: 0.9799 - val_loss: 0.1351 - val_sparse_categorical_accuracy: 0.9556 - learning_rate: 5.0000e-04 ## Epoch 141/500 ## 90/90 - 0s - 1ms/step - loss: 0.0712 - sparse_categorical_accuracy: 0.9764 - val_loss: 0.1121 - val_sparse_categorical_accuracy: 0.9612 - learning_rate: 5.0000e-04 ## Epoch 142/500 ## 90/90 - 0s - 1ms/step - loss: 0.0692 - sparse_categorical_accuracy: 0.9778 - val_loss: 0.1139 - val_sparse_categorical_accuracy: 0.9639 - learning_rate: 5.0000e-04 ## Epoch 143/500 ## 90/90 - 0s - 1ms/step - loss: 0.0781 - sparse_categorical_accuracy: 0.9736 - val_loss: 0.1202 - val_sparse_categorical_accuracy: 0.9515 - learning_rate: 5.0000e-04 ## Epoch 144/500 ## 90/90 - 0s - 1ms/step - loss: 0.0699 - sparse_categorical_accuracy: 0.9781 - val_loss: 0.1131 - val_sparse_categorical_accuracy: 0.9501 - learning_rate: 5.0000e-04 ## Epoch 145/500 ## 90/90 - 0s - 1ms/step - loss: 0.0712 - sparse_categorical_accuracy: 0.9771 - val_loss: 0.1097 - val_sparse_categorical_accuracy: 0.9542 - learning_rate: 5.0000e-04 ## Epoch 146/500 ## 90/90 - 0s - 1ms/step - loss: 0.0730 - sparse_categorical_accuracy: 0.9736 - val_loss: 0.1421 - val_sparse_categorical_accuracy: 0.9445 - learning_rate: 5.0000e-04 ## Epoch 147/500 ## 90/90 - 0s - 1ms/step - loss: 0.0757 - sparse_categorical_accuracy: 0.9740 - val_loss: 0.3533 - val_sparse_categorical_accuracy: 0.8752 - learning_rate: 5.0000e-04 ## Epoch 148/500 ## 90/90 - 0s - 1ms/step - loss: 0.0818 - sparse_categorical_accuracy: 0.9701 - val_loss: 0.7787 - val_sparse_categorical_accuracy: 0.7836 - learning_rate: 5.0000e-04 ## Epoch 149/500 ## 90/90 - 0s - 1ms/step - loss: 0.0717 - sparse_categorical_accuracy: 0.9771 - val_loss: 0.1214 - val_sparse_categorical_accuracy: 0.9515 - learning_rate: 5.0000e-04 ## Epoch 150/500 ## 90/90 - 0s - 1ms/step - loss: 0.0736 - sparse_categorical_accuracy: 0.9757 - val_loss: 0.1445 - val_sparse_categorical_accuracy: 0.9362 - learning_rate: 5.0000e-04 ## Epoch 151/500 ## 90/90 - 0s - 1ms/step - loss: 0.0725 - sparse_categorical_accuracy: 0.9729 - val_loss: 0.1175 - val_sparse_categorical_accuracy: 0.9528 - learning_rate: 5.0000e-04 ## Epoch 152/500 ## 90/90 - 0s - 1ms/step - loss: 0.0726 - sparse_categorical_accuracy: 0.9750 - val_loss: 0.1329 - val_sparse_categorical_accuracy: 0.9515 - learning_rate: 5.0000e-04 ## Epoch 153/500 ## 90/90 - 0s - 1ms/step - loss: 0.0727 - sparse_categorical_accuracy: 0.9757 - val_loss: 0.1806 - val_sparse_categorical_accuracy: 0.9293 - learning_rate: 5.0000e-04 ## Epoch 154/500 ## 90/90 - 0s - 1ms/step - loss: 0.0684 - sparse_categorical_accuracy: 0.9778 - val_loss: 0.3892 - val_sparse_categorical_accuracy: 0.8682 - learning_rate: 5.0000e-04 ## Epoch 155/500 ## 90/90 - 0s - 1ms/step - loss: 0.0734 - sparse_categorical_accuracy: 0.9753 - val_loss: 0.1122 - val_sparse_categorical_accuracy: 0.9542 - learning_rate: 5.0000e-04 ## Epoch 156/500 ## 90/90 - 0s - 1ms/step - loss: 0.0758 - sparse_categorical_accuracy: 0.9774 - val_loss: 0.1077 - val_sparse_categorical_accuracy: 0.9653 - learning_rate: 5.0000e-04 ## Epoch 157/500 ## 90/90 - 0s - 1ms/step - loss: 0.0734 - sparse_categorical_accuracy: 0.9747 - val_loss: 0.1594 - val_sparse_categorical_accuracy: 0.9376 - learning_rate: 5.0000e-04 ## Epoch 158/500 ## 90/90 - 0s - 1ms/step - loss: 0.0753 - sparse_categorical_accuracy: 0.9750 - val_loss: 0.1992 - val_sparse_categorical_accuracy: 0.9168 - learning_rate: 5.0000e-04 ## Epoch 159/500 ## 90/90 - 0s - 1ms/step - loss: 0.0740 - sparse_categorical_accuracy: 0.9740 - val_loss: 0.1483 - val_sparse_categorical_accuracy: 0.9362 - learning_rate: 5.0000e-04 ## Epoch 160/500 ## 90/90 - 0s - 2ms/step - loss: 0.0677 - sparse_categorical_accuracy: 0.9778 - val_loss: 0.1041 - val_sparse_categorical_accuracy: 0.9653 - learning_rate: 2.5000e-04 ## Epoch 161/500 ## 90/90 - 0s - 1ms/step - loss: 0.0686 - sparse_categorical_accuracy: 0.9778 - val_loss: 0.1327 - val_sparse_categorical_accuracy: 0.9515 - learning_rate: 2.5000e-04 ## Epoch 162/500 ## 90/90 - 0s - 1ms/step - loss: 0.0634 - sparse_categorical_accuracy: 0.9792 - val_loss: 0.1059 - val_sparse_categorical_accuracy: 0.9667 - learning_rate: 2.5000e-04 ## Epoch 163/500 ## 90/90 - 0s - 1ms/step - loss: 0.0595 - sparse_categorical_accuracy: 0.9792 - val_loss: 0.1236 - val_sparse_categorical_accuracy: 0.9612 - learning_rate: 2.5000e-04 ## Epoch 164/500 ## 90/90 - 0s - 1ms/step - loss: 0.0637 - sparse_categorical_accuracy: 0.9806 - val_loss: 0.1185 - val_sparse_categorical_accuracy: 0.9626 - learning_rate: 2.5000e-04 ## Epoch 165/500 ## 90/90 - 0s - 1ms/step - loss: 0.0607 - sparse_categorical_accuracy: 0.9816 - val_loss: 0.1480 - val_sparse_categorical_accuracy: 0.9473 - learning_rate: 2.5000e-04 ## Epoch 166/500 ## 90/90 - 0s - 1ms/step - loss: 0.0594 - sparse_categorical_accuracy: 0.9799 - val_loss: 0.1044 - val_sparse_categorical_accuracy: 0.9653 - learning_rate: 2.5000e-04 ## Epoch 167/500 ## 90/90 - 0s - 1ms/step - loss: 0.0629 - sparse_categorical_accuracy: 0.9788 - val_loss: 0.1343 - val_sparse_categorical_accuracy: 0.9445 - learning_rate: 2.5000e-04 ## Epoch 168/500 ## 90/90 - 0s - 1ms/step - loss: 0.0740 - sparse_categorical_accuracy: 0.9750 - val_loss: 0.1081 - val_sparse_categorical_accuracy: 0.9667 - learning_rate: 2.5000e-04 ## Epoch 169/500 ## 90/90 - 0s - 1ms/step - loss: 0.0649 - sparse_categorical_accuracy: 0.9799 - val_loss: 0.1136 - val_sparse_categorical_accuracy: 0.9681 - learning_rate: 2.5000e-04 ## Epoch 170/500 ## 90/90 - 0s - 2ms/step - loss: 0.0649 - sparse_categorical_accuracy: 0.9823 - val_loss: 0.1037 - val_sparse_categorical_accuracy: 0.9695 - learning_rate: 2.5000e-04 ## Epoch 171/500 ## 90/90 - 0s - 1ms/step - loss: 0.0609 - sparse_categorical_accuracy: 0.9799 - val_loss: 0.1239 - val_sparse_categorical_accuracy: 0.9653 - learning_rate: 2.5000e-04 ## Epoch 172/500 ## 90/90 - 0s - 1ms/step - loss: 0.0636 - sparse_categorical_accuracy: 0.9795 - val_loss: 0.1037 - val_sparse_categorical_accuracy: 0.9528 - learning_rate: 2.5000e-04 ## Epoch 173/500 ## 90/90 - 0s - 1ms/step - loss: 0.0608 - sparse_categorical_accuracy: 0.9816 - val_loss: 0.1296 - val_sparse_categorical_accuracy: 0.9626 - learning_rate: 2.5000e-04 ## Epoch 174/500 ## 90/90 - 0s - 2ms/step - loss: 0.0624 - sparse_categorical_accuracy: 0.9792 - val_loss: 0.0981 - val_sparse_categorical_accuracy: 0.9639 - learning_rate: 2.5000e-04 ## Epoch 175/500 ## 90/90 - 0s - 1ms/step - loss: 0.0666 - sparse_categorical_accuracy: 0.9785 - val_loss: 0.1135 - val_sparse_categorical_accuracy: 0.9487 - learning_rate: 2.5000e-04 ## Epoch 176/500 ## 90/90 - 0s - 1ms/step - loss: 0.0625 - sparse_categorical_accuracy: 0.9771 - val_loss: 0.1024 - val_sparse_categorical_accuracy: 0.9584 - learning_rate: 2.5000e-04 ## Epoch 177/500 ## 90/90 - 0s - 1ms/step - loss: 0.0616 - sparse_categorical_accuracy: 0.9806 - val_loss: 0.1036 - val_sparse_categorical_accuracy: 0.9570 - learning_rate: 2.5000e-04 ## Epoch 178/500 ## 90/90 - 0s - 1ms/step - loss: 0.0590 - sparse_categorical_accuracy: 0.9823 - val_loss: 0.2086 - val_sparse_categorical_accuracy: 0.9154 - learning_rate: 2.5000e-04 ## Epoch 179/500 ## 90/90 - 0s - 1ms/step - loss: 0.0602 - sparse_categorical_accuracy: 0.9788 - val_loss: 0.1181 - val_sparse_categorical_accuracy: 0.9653 - learning_rate: 2.5000e-04 ## Epoch 180/500 ## 90/90 - 0s - 1ms/step - loss: 0.0615 - sparse_categorical_accuracy: 0.9816 - val_loss: 0.1587 - val_sparse_categorical_accuracy: 0.9459 - learning_rate: 2.5000e-04 ## Epoch 181/500 ## 90/90 - 0s - 1ms/step - loss: 0.0647 - sparse_categorical_accuracy: 0.9813 - val_loss: 0.1836 - val_sparse_categorical_accuracy: 0.9362 - learning_rate: 2.5000e-04 ## Epoch 182/500 ## 90/90 - 0s - 1ms/step - loss: 0.0614 - sparse_categorical_accuracy: 0.9813 - val_loss: 0.2156 - val_sparse_categorical_accuracy: 0.9223 - learning_rate: 2.5000e-04 ## Epoch 183/500 ## 90/90 - 0s - 1ms/step - loss: 0.0637 - sparse_categorical_accuracy: 0.9778 - val_loss: 0.1028 - val_sparse_categorical_accuracy: 0.9584 - learning_rate: 2.5000e-04 ## Epoch 184/500 ## 90/90 - 0s - 1ms/step - loss: 0.0572 - sparse_categorical_accuracy: 0.9823 - val_loss: 0.1347 - val_sparse_categorical_accuracy: 0.9431 - learning_rate: 2.5000e-04 ## Epoch 185/500 ## 90/90 - 0s - 1ms/step - loss: 0.0593 - sparse_categorical_accuracy: 0.9826 - val_loss: 0.1402 - val_sparse_categorical_accuracy: 0.9459 - learning_rate: 2.5000e-04 ## Epoch 186/500 ## 90/90 - 0s - 1ms/step - loss: 0.0659 - sparse_categorical_accuracy: 0.9774 - val_loss: 0.1873 - val_sparse_categorical_accuracy: 0.9293 - learning_rate: 2.5000e-04 ## Epoch 187/500 ## 90/90 - 0s - 1ms/step - loss: 0.0606 - sparse_categorical_accuracy: 0.9799 - val_loss: 0.1210 - val_sparse_categorical_accuracy: 0.9570 - learning_rate: 2.5000e-04 ## Epoch 188/500 ## 90/90 - 0s - 1ms/step - loss: 0.0610 - sparse_categorical_accuracy: 0.9813 - val_loss: 0.1250 - val_sparse_categorical_accuracy: 0.9473 - learning_rate: 2.5000e-04 ## Epoch 189/500 ## 90/90 - 0s - 1ms/step - loss: 0.0570 - sparse_categorical_accuracy: 0.9830 - val_loss: 0.1227 - val_sparse_categorical_accuracy: 0.9667 - learning_rate: 2.5000e-04 ## Epoch 190/500 ## 90/90 - 0s - 1ms/step - loss: 0.0678 - sparse_categorical_accuracy: 0.9771 - val_loss: 0.1367 - val_sparse_categorical_accuracy: 0.9584 - learning_rate: 2.5000e-04 ## Epoch 191/500 ## 90/90 - 0s - 1ms/step - loss: 0.0580 - sparse_categorical_accuracy: 0.9802 - val_loss: 0.1005 - val_sparse_categorical_accuracy: 0.9653 - learning_rate: 2.5000e-04 ## Epoch 192/500 ## 90/90 - 0s - 1ms/step - loss: 0.0651 - sparse_categorical_accuracy: 0.9778 - val_loss: 0.1170 - val_sparse_categorical_accuracy: 0.9639 - learning_rate: 2.5000e-04 ## Epoch 193/500 ## 90/90 - 0s - 1ms/step - loss: 0.0593 - sparse_categorical_accuracy: 0.9823 - val_loss: 0.1235 - val_sparse_categorical_accuracy: 0.9473 - learning_rate: 2.5000e-04 ## Epoch 194/500 ## 90/90 - 0s - 1ms/step - loss: 0.0588 - sparse_categorical_accuracy: 0.9792 - val_loss: 0.1089 - val_sparse_categorical_accuracy: 0.9570 - learning_rate: 2.5000e-04 ## Epoch 195/500 ## 90/90 - 0s - 2ms/step - loss: 0.0564 - sparse_categorical_accuracy: 0.9833 - val_loss: 0.0979 - val_sparse_categorical_accuracy: 0.9681 - learning_rate: 1.2500e-04 ## Epoch 196/500 ## 90/90 - 0s - 1ms/step - loss: 0.0584 - sparse_categorical_accuracy: 0.9788 - val_loss: 0.1169 - val_sparse_categorical_accuracy: 0.9584 - learning_rate: 1.2500e-04 ## Epoch 197/500 ## 90/90 - 0s - 1ms/step - loss: 0.0553 - sparse_categorical_accuracy: 0.9802 - val_loss: 0.1380 - val_sparse_categorical_accuracy: 0.9556 - learning_rate: 1.2500e-04 ## Epoch 198/500 ## 90/90 - 0s - 1ms/step - loss: 0.0541 - sparse_categorical_accuracy: 0.9868 - val_loss: 0.1261 - val_sparse_categorical_accuracy: 0.9556 - learning_rate: 1.2500e-04 ## Epoch 199/500 ## 90/90 - 0s - 1ms/step - loss: 0.0575 - sparse_categorical_accuracy: 0.9806 - val_loss: 0.1010 - val_sparse_categorical_accuracy: 0.9681 - learning_rate: 1.2500e-04 ## Epoch 200/500 ## 90/90 - 0s - 1ms/step - loss: 0.0568 - sparse_categorical_accuracy: 0.9816 - val_loss: 0.0996 - val_sparse_categorical_accuracy: 0.9612 - learning_rate: 1.2500e-04 ## Epoch 201/500 ## 90/90 - 0s - 1ms/step - loss: 0.0547 - sparse_categorical_accuracy: 0.9819 - val_loss: 0.1096 - val_sparse_categorical_accuracy: 0.9653 - learning_rate: 1.2500e-04 ## Epoch 202/500 ## 90/90 - 0s - 1ms/step - loss: 0.0547 - sparse_categorical_accuracy: 0.9847 - val_loss: 0.1019 - val_sparse_categorical_accuracy: 0.9556 - learning_rate: 1.2500e-04 ## Epoch 203/500 ## 90/90 - 0s - 1ms/step - loss: 0.0562 - sparse_categorical_accuracy: 0.9816 - val_loss: 0.0985 - val_sparse_categorical_accuracy: 0.9639 - learning_rate: 1.2500e-04 ## Epoch 204/500 ## 90/90 - 0s - 2ms/step - loss: 0.0519 - sparse_categorical_accuracy: 0.9830 - val_loss: 0.0978 - val_sparse_categorical_accuracy: 0.9681 - learning_rate: 1.2500e-04 ## Epoch 205/500 ## 90/90 - 0s - 1ms/step - loss: 0.0558 - sparse_categorical_accuracy: 0.9802 - val_loss: 0.1035 - val_sparse_categorical_accuracy: 0.9667 - learning_rate: 1.2500e-04 ## Epoch 206/500 ## 90/90 - 0s - 1ms/step - loss: 0.0551 - sparse_categorical_accuracy: 0.9816 - val_loss: 0.1006 - val_sparse_categorical_accuracy: 0.9709 - learning_rate: 1.2500e-04 ## Epoch 207/500 ## 90/90 - 0s - 1ms/step - loss: 0.0593 - sparse_categorical_accuracy: 0.9816 - val_loss: 0.1023 - val_sparse_categorical_accuracy: 0.9556 - learning_rate: 1.2500e-04 ## Epoch 208/500 ## 90/90 - 0s - 2ms/step - loss: 0.0556 - sparse_categorical_accuracy: 0.9816 - val_loss: 0.1314 - val_sparse_categorical_accuracy: 0.9626 - learning_rate: 1.2500e-04 ## Epoch 209/500 ## 90/90 - 0s - 1ms/step - loss: 0.0539 - sparse_categorical_accuracy: 0.9809 - val_loss: 0.1529 - val_sparse_categorical_accuracy: 0.9487 - learning_rate: 1.2500e-04 ## Epoch 210/500 ## 90/90 - 0s - 1ms/step - loss: 0.0566 - sparse_categorical_accuracy: 0.9816 - val_loss: 0.0993 - val_sparse_categorical_accuracy: 0.9653 - learning_rate: 1.2500e-04 ## Epoch 211/500 ## 90/90 - 0s - 1ms/step - loss: 0.0552 - sparse_categorical_accuracy: 0.9837 - val_loss: 0.0984 - val_sparse_categorical_accuracy: 0.9639 - learning_rate: 1.2500e-04 ## Epoch 212/500 ## 90/90 - 0s - 1ms/step - loss: 0.0548 - sparse_categorical_accuracy: 0.9826 - val_loss: 0.0999 - val_sparse_categorical_accuracy: 0.9584 - learning_rate: 1.2500e-04 ## Epoch 213/500 ## 90/90 - 0s - 2ms/step - loss: 0.0573 - sparse_categorical_accuracy: 0.9840 - val_loss: 0.0968 - val_sparse_categorical_accuracy: 0.9667 - learning_rate: 1.2500e-04 ## Epoch 214/500 ## 90/90 - 0s - 1ms/step - loss: 0.0568 - sparse_categorical_accuracy: 0.9830 - val_loss: 0.1000 - val_sparse_categorical_accuracy: 0.9639 - learning_rate: 1.2500e-04 ## Epoch 215/500 ## 90/90 - 0s - 1ms/step - loss: 0.0534 - sparse_categorical_accuracy: 0.9816 - val_loss: 0.1031 - val_sparse_categorical_accuracy: 0.9667 - learning_rate: 1.2500e-04 ## Epoch 216/500 ## 90/90 - 0s - 1ms/step - loss: 0.0531 - sparse_categorical_accuracy: 0.9847 - val_loss: 0.1042 - val_sparse_categorical_accuracy: 0.9681 - learning_rate: 1.2500e-04 ## Epoch 217/500 ## 90/90 - 0s - 1ms/step - loss: 0.0532 - sparse_categorical_accuracy: 0.9826 - val_loss: 0.1096 - val_sparse_categorical_accuracy: 0.9695 - learning_rate: 1.2500e-04 ## Epoch 218/500 ## 90/90 - 0s - 1ms/step - loss: 0.0577 - sparse_categorical_accuracy: 0.9809 - val_loss: 0.1009 - val_sparse_categorical_accuracy: 0.9612 - learning_rate: 1.2500e-04 ## Epoch 219/500 ## 90/90 - 0s - 1ms/step - loss: 0.0539 - sparse_categorical_accuracy: 0.9823 - val_loss: 0.1072 - val_sparse_categorical_accuracy: 0.9695 - learning_rate: 1.2500e-04 ## Epoch 220/500 ## 90/90 - 0s - 1ms/step - loss: 0.0529 - sparse_categorical_accuracy: 0.9851 - val_loss: 0.1005 - val_sparse_categorical_accuracy: 0.9695 - learning_rate: 1.2500e-04 ## Epoch 221/500 ## 90/90 - 0s - 1ms/step - loss: 0.0554 - sparse_categorical_accuracy: 0.9799 - val_loss: 0.1036 - val_sparse_categorical_accuracy: 0.9695 - learning_rate: 1.2500e-04 ## Epoch 222/500 ## 90/90 - 0s - 1ms/step - loss: 0.0562 - sparse_categorical_accuracy: 0.9809 - val_loss: 0.1037 - val_sparse_categorical_accuracy: 0.9736 - learning_rate: 1.2500e-04 ## Epoch 223/500 ## 90/90 - 0s - 1ms/step - loss: 0.0544 - sparse_categorical_accuracy: 0.9823 - val_loss: 0.1005 - val_sparse_categorical_accuracy: 0.9736 - learning_rate: 1.2500e-04 ## Epoch 224/500 ## 90/90 - 0s - 1ms/step - loss: 0.0509 - sparse_categorical_accuracy: 0.9847 - val_loss: 0.0969 - val_sparse_categorical_accuracy: 0.9626 - learning_rate: 1.2500e-04 ## Epoch 225/500 ## 90/90 - 0s - 1ms/step - loss: 0.0569 - sparse_categorical_accuracy: 0.9823 - val_loss: 0.1261 - val_sparse_categorical_accuracy: 0.9612 - learning_rate: 1.2500e-04 ## Epoch 226/500 ## 90/90 - 0s - 1ms/step - loss: 0.0519 - sparse_categorical_accuracy: 0.9826 - val_loss: 0.1019 - val_sparse_categorical_accuracy: 0.9570 - learning_rate: 1.2500e-04 ## Epoch 227/500 ## 90/90 - 0s - 1ms/step - loss: 0.0563 - sparse_categorical_accuracy: 0.9819 - val_loss: 0.1037 - val_sparse_categorical_accuracy: 0.9584 - learning_rate: 1.2500e-04 ## Epoch 228/500 ## 90/90 - 0s - 1ms/step - loss: 0.0518 - sparse_categorical_accuracy: 0.9844 - val_loss: 0.1127 - val_sparse_categorical_accuracy: 0.9515 - learning_rate: 1.2500e-04 ## Epoch 229/500 ## 90/90 - 0s - 1ms/step - loss: 0.0521 - sparse_categorical_accuracy: 0.9865 - val_loss: 0.0973 - val_sparse_categorical_accuracy: 0.9639 - learning_rate: 1.2500e-04 ## Epoch 230/500 ## 90/90 - 0s - 1ms/step - loss: 0.0520 - sparse_categorical_accuracy: 0.9847 - val_loss: 0.0999 - val_sparse_categorical_accuracy: 0.9626 - learning_rate: 1.2500e-04 ## Epoch 231/500 ## 90/90 - 0s - 2ms/step - loss: 0.0569 - sparse_categorical_accuracy: 0.9813 - val_loss: 0.0958 - val_sparse_categorical_accuracy: 0.9626 - learning_rate: 1.2500e-04 ## Epoch 232/500 ## 90/90 - 0s - 1ms/step - loss: 0.0530 - sparse_categorical_accuracy: 0.9851 - val_loss: 0.1024 - val_sparse_categorical_accuracy: 0.9681 - learning_rate: 1.2500e-04 ## Epoch 233/500 ## 90/90 - 0s - 1ms/step - loss: 0.0523 - sparse_categorical_accuracy: 0.9844 - val_loss: 0.1019 - val_sparse_categorical_accuracy: 0.9709 - learning_rate: 1.2500e-04 ## Epoch 234/500 ## 90/90 - 0s - 1ms/step - loss: 0.0539 - sparse_categorical_accuracy: 0.9837 - val_loss: 0.1595 - val_sparse_categorical_accuracy: 0.9459 - learning_rate: 1.2500e-04 ## Epoch 235/500 ## 90/90 - 0s - 1ms/step - loss: 0.0543 - sparse_categorical_accuracy: 0.9802 - val_loss: 0.1028 - val_sparse_categorical_accuracy: 0.9653 - learning_rate: 1.2500e-04 ## Epoch 236/500 ## 90/90 - 0s - 1ms/step - loss: 0.0540 - sparse_categorical_accuracy: 0.9826 - val_loss: 0.1068 - val_sparse_categorical_accuracy: 0.9653 - learning_rate: 1.2500e-04 ## Epoch 237/500 ## 90/90 - 0s - 1ms/step - loss: 0.0543 - sparse_categorical_accuracy: 0.9809 - val_loss: 0.1528 - val_sparse_categorical_accuracy: 0.9459 - learning_rate: 1.2500e-04 ## Epoch 238/500 ## 90/90 - 0s - 1ms/step - loss: 0.0527 - sparse_categorical_accuracy: 0.9861 - val_loss: 0.0977 - val_sparse_categorical_accuracy: 0.9695 - learning_rate: 1.2500e-04 ## Epoch 239/500 ## 90/90 - 0s - 1ms/step - loss: 0.0537 - sparse_categorical_accuracy: 0.9813 - val_loss: 0.0980 - val_sparse_categorical_accuracy: 0.9598 - learning_rate: 1.2500e-04 ## Epoch 240/500 ## 90/90 - 0s - 1ms/step - loss: 0.0512 - sparse_categorical_accuracy: 0.9854 - val_loss: 0.1054 - val_sparse_categorical_accuracy: 0.9709 - learning_rate: 1.2500e-04 ## Epoch 241/500 ## 90/90 - 0s - 1ms/step - loss: 0.0541 - sparse_categorical_accuracy: 0.9837 - val_loss: 0.1325 - val_sparse_categorical_accuracy: 0.9487 - learning_rate: 1.2500e-04 ## Epoch 242/500 ## 90/90 - 0s - 1ms/step - loss: 0.0535 - sparse_categorical_accuracy: 0.9826 - val_loss: 0.1028 - val_sparse_categorical_accuracy: 0.9681 - learning_rate: 1.2500e-04 ## Epoch 243/500 ## 90/90 - 0s - 1ms/step - loss: 0.0555 - sparse_categorical_accuracy: 0.9823 - val_loss: 0.0991 - val_sparse_categorical_accuracy: 0.9626 - learning_rate: 1.2500e-04 ## Epoch 244/500 ## 90/90 - 0s - 1ms/step - loss: 0.0572 - sparse_categorical_accuracy: 0.9795 - val_loss: 0.0986 - val_sparse_categorical_accuracy: 0.9639 - learning_rate: 1.2500e-04 ## Epoch 245/500 ## 90/90 - 0s - 1ms/step - loss: 0.0520 - sparse_categorical_accuracy: 0.9833 - val_loss: 0.0980 - val_sparse_categorical_accuracy: 0.9626 - learning_rate: 1.2500e-04 ## Epoch 246/500 ## 90/90 - 0s - 1ms/step - loss: 0.0554 - sparse_categorical_accuracy: 0.9809 - val_loss: 0.0978 - val_sparse_categorical_accuracy: 0.9639 - learning_rate: 1.2500e-04 ## Epoch 247/500 ## 90/90 - 0s - 1ms/step - loss: 0.0498 - sparse_categorical_accuracy: 0.9837 - val_loss: 0.1015 - val_sparse_categorical_accuracy: 0.9681 - learning_rate: 1.2500e-04 ## Epoch 248/500 ## 90/90 - 0s - 1ms/step - loss: 0.0493 - sparse_categorical_accuracy: 0.9840 - val_loss: 0.0994 - val_sparse_categorical_accuracy: 0.9612 - learning_rate: 1.2500e-04 ## Epoch 249/500 ## 90/90 - 0s - 1ms/step - loss: 0.0543 - sparse_categorical_accuracy: 0.9847 - val_loss: 0.1054 - val_sparse_categorical_accuracy: 0.9695 - learning_rate: 1.2500e-04 ## Epoch 250/500 ## 90/90 - 0s - 1ms/step - loss: 0.0550 - sparse_categorical_accuracy: 0.9802 - val_loss: 0.0989 - val_sparse_categorical_accuracy: 0.9626 - learning_rate: 1.2500e-04 ## Epoch 251/500 ## 90/90 - 0s - 1ms/step - loss: 0.0539 - sparse_categorical_accuracy: 0.9819 - val_loss: 0.1018 - val_sparse_categorical_accuracy: 0.9667 - learning_rate: 1.2500e-04 ## Epoch 252/500 ## 90/90 - 0s - 1ms/step - loss: 0.0542 - sparse_categorical_accuracy: 0.9830 - val_loss: 0.0975 - val_sparse_categorical_accuracy: 0.9667 - learning_rate: 1.0000e-04 ## Epoch 253/500 ## 90/90 - 0s - 1ms/step - loss: 0.0538 - sparse_categorical_accuracy: 0.9830 - val_loss: 0.1065 - val_sparse_categorical_accuracy: 0.9723 - learning_rate: 1.0000e-04 ## Epoch 254/500 ## 90/90 - 0s - 1ms/step - loss: 0.0542 - sparse_categorical_accuracy: 0.9823 - val_loss: 0.1145 - val_sparse_categorical_accuracy: 0.9528 - learning_rate: 1.0000e-04 ## Epoch 255/500 ## 90/90 - 0s - 1ms/step - loss: 0.0506 - sparse_categorical_accuracy: 0.9819 - val_loss: 0.0976 - val_sparse_categorical_accuracy: 0.9681 - learning_rate: 1.0000e-04 ## Epoch 256/500 ## 90/90 - 0s - 1ms/step - loss: 0.0521 - sparse_categorical_accuracy: 0.9847 - val_loss: 0.1117 - val_sparse_categorical_accuracy: 0.9626 - learning_rate: 1.0000e-04 ## Epoch 257/500 ## 90/90 - 0s - 1ms/step - loss: 0.0504 - sparse_categorical_accuracy: 0.9854 - val_loss: 0.1201 - val_sparse_categorical_accuracy: 0.9626 - learning_rate: 1.0000e-04 ## Epoch 258/500 ## 90/90 - 0s - 1ms/step - loss: 0.0498 - sparse_categorical_accuracy: 0.9851 - val_loss: 0.1039 - val_sparse_categorical_accuracy: 0.9584 - learning_rate: 1.0000e-04 ## Epoch 259/500 ## 90/90 - 0s - 1ms/step - loss: 0.0479 - sparse_categorical_accuracy: 0.9868 - val_loss: 0.1055 - val_sparse_categorical_accuracy: 0.9695 - learning_rate: 1.0000e-04 ## Epoch 260/500 ## 90/90 - 0s - 1ms/step - loss: 0.0499 - sparse_categorical_accuracy: 0.9840 - val_loss: 0.1046 - val_sparse_categorical_accuracy: 0.9556 - learning_rate: 1.0000e-04 ## Epoch 261/500 ## 90/90 - 0s - 1ms/step - loss: 0.0529 - sparse_categorical_accuracy: 0.9823 - val_loss: 0.1022 - val_sparse_categorical_accuracy: 0.9667 - learning_rate: 1.0000e-04 ## Epoch 262/500 ## 90/90 - 0s - 1ms/step - loss: 0.0492 - sparse_categorical_accuracy: 0.9840 - val_loss: 0.1063 - val_sparse_categorical_accuracy: 0.9681 - learning_rate: 1.0000e-04 ## Epoch 263/500 ## 90/90 - 0s - 1ms/step - loss: 0.0500 - sparse_categorical_accuracy: 0.9833 - val_loss: 0.1006 - val_sparse_categorical_accuracy: 0.9626 - learning_rate: 1.0000e-04 ## Epoch 264/500 ## 90/90 - 0s - 1ms/step - loss: 0.0533 - sparse_categorical_accuracy: 0.9823 - val_loss: 0.1088 - val_sparse_categorical_accuracy: 0.9709 - learning_rate: 1.0000e-04 ## Epoch 265/500 ## 90/90 - 0s - 1ms/step - loss: 0.0549 - sparse_categorical_accuracy: 0.9802 - val_loss: 0.1010 - val_sparse_categorical_accuracy: 0.9709 - learning_rate: 1.0000e-04 ## Epoch 266/500 ## 90/90 - 0s - 1ms/step - loss: 0.0512 - sparse_categorical_accuracy: 0.9830 - val_loss: 0.1016 - val_sparse_categorical_accuracy: 0.9570 - learning_rate: 1.0000e-04 ## Epoch 267/500 ## 90/90 - 0s - 1ms/step - loss: 0.0496 - sparse_categorical_accuracy: 0.9847 - val_loss: 0.1012 - val_sparse_categorical_accuracy: 0.9570 - learning_rate: 1.0000e-04 ## Epoch 268/500 ## 90/90 - 0s - 1ms/step - loss: 0.0477 - sparse_categorical_accuracy: 0.9872 - val_loss: 0.1016 - val_sparse_categorical_accuracy: 0.9570 - learning_rate: 1.0000e-04 ## Epoch 269/500 ## 90/90 - 0s - 1ms/step - loss: 0.0523 - sparse_categorical_accuracy: 0.9861 - val_loss: 0.1012 - val_sparse_categorical_accuracy: 0.9723 - learning_rate: 1.0000e-04 ## Epoch 270/500 ## 90/90 - 0s - 1ms/step - loss: 0.0474 - sparse_categorical_accuracy: 0.9889 - val_loss: 0.0975 - val_sparse_categorical_accuracy: 0.9653 - learning_rate: 1.0000e-04 ## Epoch 271/500 ## 90/90 - 0s - 1ms/step - loss: 0.0489 - sparse_categorical_accuracy: 0.9840 - val_loss: 0.1055 - val_sparse_categorical_accuracy: 0.9695 - learning_rate: 1.0000e-04 ## Epoch 272/500 ## 90/90 - 0s - 1ms/step - loss: 0.0503 - sparse_categorical_accuracy: 0.9826 - val_loss: 0.1050 - val_sparse_categorical_accuracy: 0.9709 - learning_rate: 1.0000e-04 ## Epoch 273/500 ## 90/90 - 0s - 1ms/step - loss: 0.0514 - sparse_categorical_accuracy: 0.9844 - val_loss: 0.0980 - val_sparse_categorical_accuracy: 0.9695 - learning_rate: 1.0000e-04 ## Epoch 274/500 ## 90/90 - 0s - 1ms/step - loss: 0.0479 - sparse_categorical_accuracy: 0.9854 - val_loss: 0.1049 - val_sparse_categorical_accuracy: 0.9639 - learning_rate: 1.0000e-04 ## Epoch 275/500 ## 90/90 - 0s - 1ms/step - loss: 0.0467 - sparse_categorical_accuracy: 0.9840 - val_loss: 0.1473 - val_sparse_categorical_accuracy: 0.9556 - learning_rate: 1.0000e-04 ## Epoch 276/500 ## 90/90 - 0s - 1ms/step - loss: 0.0504 - sparse_categorical_accuracy: 0.9819 - val_loss: 0.1065 - val_sparse_categorical_accuracy: 0.9695 - learning_rate: 1.0000e-04 ## Epoch 277/500 ## 90/90 - 0s - 1ms/step - loss: 0.0505 - sparse_categorical_accuracy: 0.9847 - val_loss: 0.1187 - val_sparse_categorical_accuracy: 0.9612 - learning_rate: 1.0000e-04 ## Epoch 278/500 ## 90/90 - 0s - 1ms/step - loss: 0.0498 - sparse_categorical_accuracy: 0.9840 - val_loss: 0.1267 - val_sparse_categorical_accuracy: 0.9598 - learning_rate: 1.0000e-04 ## Epoch 279/500 ## 90/90 - 0s - 1ms/step - loss: 0.0517 - sparse_categorical_accuracy: 0.9830 - val_loss: 0.1037 - val_sparse_categorical_accuracy: 0.9653 - learning_rate: 1.0000e-04 ## Epoch 280/500 ## 90/90 - 0s - 1ms/step - loss: 0.0501 - sparse_categorical_accuracy: 0.9830 - val_loss: 0.0985 - val_sparse_categorical_accuracy: 0.9667 - learning_rate: 1.0000e-04 ## Epoch 281/500 ## 90/90 - 0s - 1ms/step - loss: 0.0468 - sparse_categorical_accuracy: 0.9865 - val_loss: 0.1119 - val_sparse_categorical_accuracy: 0.9653 - learning_rate: 1.0000e-04 ## Epoch 281: early stopping"},{"path":"https://keras3.posit.co/dev/articles/examples/timeseries/timeseries_classification_from_scratch.html","id":"evaluate-model-on-test-data","dir":"Articles > Examples > Timeseries","previous_headings":"","what":"Evaluate model on test data","title":"Timeseries classification from scratch","text":"","code":"model <- load_model(\"best_model.keras\")  results <- model |> evaluate(x_test, y_test) ## 42/42 - 1s - 16ms/step - loss: 0.0938 - sparse_categorical_accuracy: 0.9697 str(results) ## List of 2 ##  $ loss                       : num 0.0938 ##  $ sparse_categorical_accuracy: num 0.97 cat(   \"Test accuracy: \", results$sparse_categorical_accuracy, \"\\n\",   \"Test loss: \", results$loss, \"\\n\",   sep = \"\" ) ## Test accuracy: 0.969697 ## Test loss: 0.09383929"},{"path":"https://keras3.posit.co/dev/articles/examples/timeseries/timeseries_classification_from_scratch.html","id":"plot-the-models-training-history","dir":"Articles > Examples > Timeseries","previous_headings":"","what":"Plot the model’s training history","title":"Timeseries classification from scratch","text":"Plot just training validation accuracy: can see training accuracy reaches almost 0.95 100 epochs. However, observing validation accuracy can see network still needs training reaches almost 0.97 validation training accuracy 200 epochs. Beyond 200th epoch, continue training, validation accuracy start decreasing training accuracy continue increasing: model starts overfitting.","code":"plot(history) plot(history, metric = \"sparse_categorical_accuracy\") +   # scale x axis to actual number of epochs run before early stopping   ggplot2::xlim(0, length(history$metrics$loss))"},{"path":"https://keras3.posit.co/dev/articles/examples/vision/autoencoder.html","id":"introduction","dir":"Articles > Examples > Vision","previous_headings":"","what":"Introduction","title":"Convolutional autoencoder for image denoising","text":"example demonstrates implement deep convolutional autoencoder image denoising, mapping noisy digits images MNIST dataset clean digits images. implementation based original blog post titled Building Autoencoders Keras François Chollet.","code":""},{"path":"https://keras3.posit.co/dev/articles/examples/vision/autoencoder.html","id":"setup","dir":"Articles > Examples > Vision","previous_headings":"","what":"Setup","title":"Convolutional autoencoder for image denoising","text":"","code":"library(keras3)  # Normalizes the supplied array and reshapes it. preprocess <- function(array) {   array_reshape(array/255, c(dim(array)[1], 28, 28, 1)) }  # Adds random noise to each image in the supplied array. noise <- function(array) {   noise_factor <- 0.4   noisy_array <- array + noise_factor * random_normal(dim(array))   op_clip(noisy_array, 0.0, 1.0) }  display <- function(array1, array2) {   n <- 2   indices <- sample.int(dim(array1)[1], n)   images1 <- as.array(array1)[indices, , , ]   images2 <- as.array(array2)[indices, , , ]    par(mfrow = c(2, n), mar = c(0, 0, 0, 0))   for (i in seq_len(n)) {     plot(as.raster(images1[i, , ]))     plot(as.raster(images2[i, , ]))   } }"},{"path":"https://keras3.posit.co/dev/articles/examples/vision/autoencoder.html","id":"prepare-the-data","dir":"Articles > Examples > Vision","previous_headings":"","what":"Prepare the data","title":"Convolutional autoencoder for image denoising","text":"","code":"# Since we only need images from the dataset to encode and decode, we # won't use the labels. c(c(train_data, .), c(test_data, .)) %<-% dataset_mnist()  # Normalize and reshape the data train_data <- preprocess(train_data) test_data <- preprocess(test_data)  # Create a copy of the data with added noise noisy_train_data <- noise(train_data) noisy_test_data <- noise(test_data)  # Display the train data and a version of it with added noise display(train_data, noisy_train_data)"},{"path":"https://keras3.posit.co/dev/articles/examples/vision/autoencoder.html","id":"build-the-autoencoder","dir":"Articles > Examples > Vision","previous_headings":"","what":"Build the autoencoder","title":"Convolutional autoencoder for image denoising","text":"going use Functional API build convolutional autoencoder. Now can train autoencoder using train_data input data target. Notice setting validation data using format. Let’s predict test dataset display original image together prediction autoencoder. Notice predictions pretty close original images, although quite . Now know autoencoder works, let’s retrain using noisy data input clean data target. want autoencoder learn denoise images. Let’s now predict noisy data display results autoencoder. Notice autoencoder amazing job removing noise input images.","code":"input <- keras_input(shape = c(28, 28, 1))  # Encoder enc <- input |>   layer_conv_2d(filters = 32, kernel_size = c(3, 3),                 activation = \"relu\", padding = \"same\") |>   layer_max_pooling_2d(pool_size = c(2, 2), padding = \"same\") |>   layer_conv_2d(filters = 32, kernel_size = c(3, 3),                 activation = \"relu\", padding = \"same\") |>   layer_max_pooling_2d(pool_size = c(2, 2), padding = \"same\")  # Decoder dec <- enc |>   layer_conv_2d_transpose(filters = 32, kernel_size = c(3, 3), strides = 2,                           activation = \"relu\", padding = \"same\") |>   layer_conv_2d_transpose(filters = 32, kernel_size = c(3, 3), strides = 2,                           activation = \"relu\", padding = \"same\") |>   layer_conv_2d(filters = 1, kernel_size = c(3, 3),                 activation = \"sigmoid\", padding = \"same\")  # Autoencoder autoencoder <- keras_model(input, dec) autoencoder |> compile(optimizer = \"adam\", loss = \"binary_crossentropy\") autoencoder |> summary() ## Model: \"functional\" ## ┏━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━┓ ## ┃ Layer (type)                    ┃ Output Shape           ┃       Param # ┃ ## ┡━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━┩ ## │ input_layer (InputLayer)        │ (None, 28, 28, 1)      │             0 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ conv2d (Conv2D)                 │ (None, 28, 28, 32)     │           320 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ max_pooling2d (MaxPooling2D)    │ (None, 14, 14, 32)     │             0 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ conv2d_1 (Conv2D)               │ (None, 14, 14, 32)     │         9,248 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ max_pooling2d_1 (MaxPooling2D)  │ (None, 7, 7, 32)       │             0 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ conv2d_transpose                │ (None, 14, 14, 32)     │         9,248 │ ## │ (Conv2DTranspose)               │                        │               │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ conv2d_transpose_1              │ (None, 28, 28, 32)     │         9,248 │ ## │ (Conv2DTranspose)               │                        │               │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ conv2d_2 (Conv2D)               │ (None, 28, 28, 1)      │           289 │ ## └─────────────────────────────────┴────────────────────────┴───────────────┘ ##  Total params: 28,353 (110.75 KB) ##  Trainable params: 28,353 (110.75 KB) ##  Non-trainable params: 0 (0.00 B) autoencoder |> fit(   x = train_data,   y = train_data,   epochs = 50,   batch_size = 128,   shuffle = TRUE,   validation_data = list(test_data, test_data), ) ## Epoch 1/50 ## 469/469 - 6s - 12ms/step - loss: 0.1371 - val_loss: 0.0727 ## Epoch 2/50 ## 469/469 - 2s - 3ms/step - loss: 0.0712 - val_loss: 0.0692 ## Epoch 3/50 ## 469/469 - 2s - 3ms/step - loss: 0.0690 - val_loss: 0.0678 ## Epoch 4/50 ## 469/469 - 2s - 3ms/step - loss: 0.0679 - val_loss: 0.0671 ## Epoch 5/50 ## 469/469 - 2s - 3ms/step - loss: 0.0672 - val_loss: 0.0665 ## Epoch 6/50 ## 469/469 - 2s - 3ms/step - loss: 0.0666 - val_loss: 0.0660 ## Epoch 7/50 ## 469/469 - 2s - 3ms/step - loss: 0.0662 - val_loss: 0.0656 ## Epoch 8/50 ## 469/469 - 2s - 3ms/step - loss: 0.0659 - val_loss: 0.0652 ## Epoch 9/50 ## 469/469 - 2s - 3ms/step - loss: 0.0656 - val_loss: 0.0650 ## Epoch 10/50 ## 469/469 - 1s - 3ms/step - loss: 0.0653 - val_loss: 0.0648 ## Epoch 11/50 ## 469/469 - 2s - 3ms/step - loss: 0.0651 - val_loss: 0.0646 ## Epoch 12/50 ## 469/469 - 1s - 3ms/step - loss: 0.0649 - val_loss: 0.0644 ## Epoch 13/50 ## 469/469 - 2s - 3ms/step - loss: 0.0648 - val_loss: 0.0643 ## Epoch 14/50 ## 469/469 - 1s - 3ms/step - loss: 0.0646 - val_loss: 0.0641 ## Epoch 15/50 ## 469/469 - 1s - 3ms/step - loss: 0.0645 - val_loss: 0.0640 ## Epoch 16/50 ## 469/469 - 1s - 3ms/step - loss: 0.0644 - val_loss: 0.0639 ## Epoch 17/50 ## 469/469 - 2s - 3ms/step - loss: 0.0642 - val_loss: 0.0638 ## Epoch 18/50 ## 469/469 - 2s - 3ms/step - loss: 0.0641 - val_loss: 0.0637 ## Epoch 19/50 ## 469/469 - 1s - 3ms/step - loss: 0.0640 - val_loss: 0.0636 ## Epoch 20/50 ## 469/469 - 2s - 3ms/step - loss: 0.0639 - val_loss: 0.0635 ## Epoch 21/50 ## 469/469 - 2s - 3ms/step - loss: 0.0639 - val_loss: 0.0635 ## Epoch 22/50 ## 469/469 - 1s - 3ms/step - loss: 0.0638 - val_loss: 0.0634 ## Epoch 23/50 ## 469/469 - 2s - 3ms/step - loss: 0.0637 - val_loss: 0.0634 ## Epoch 24/50 ## 469/469 - 2s - 3ms/step - loss: 0.0637 - val_loss: 0.0633 ## Epoch 25/50 ## 469/469 - 2s - 3ms/step - loss: 0.0636 - val_loss: 0.0632 ## Epoch 26/50 ## 469/469 - 2s - 3ms/step - loss: 0.0636 - val_loss: 0.0632 ## Epoch 27/50 ## 469/469 - 2s - 3ms/step - loss: 0.0635 - val_loss: 0.0631 ## Epoch 28/50 ## 469/469 - 2s - 3ms/step - loss: 0.0635 - val_loss: 0.0631 ## Epoch 29/50 ## 469/469 - 2s - 3ms/step - loss: 0.0634 - val_loss: 0.0631 ## Epoch 30/50 ## 469/469 - 1s - 3ms/step - loss: 0.0634 - val_loss: 0.0630 ## Epoch 31/50 ## 469/469 - 2s - 3ms/step - loss: 0.0633 - val_loss: 0.0630 ## Epoch 32/50 ## 469/469 - 2s - 3ms/step - loss: 0.0633 - val_loss: 0.0630 ## Epoch 33/50 ## 469/469 - 2s - 3ms/step - loss: 0.0633 - val_loss: 0.0629 ## Epoch 34/50 ## 469/469 - 2s - 3ms/step - loss: 0.0632 - val_loss: 0.0629 ## Epoch 35/50 ## 469/469 - 1s - 3ms/step - loss: 0.0632 - val_loss: 0.0628 ## Epoch 36/50 ## 469/469 - 2s - 3ms/step - loss: 0.0632 - val_loss: 0.0628 ## Epoch 37/50 ## 469/469 - 1s - 3ms/step - loss: 0.0631 - val_loss: 0.0628 ## Epoch 38/50 ## 469/469 - 2s - 3ms/step - loss: 0.0631 - val_loss: 0.0628 ## Epoch 39/50 ## 469/469 - 1s - 3ms/step - loss: 0.0631 - val_loss: 0.0627 ## Epoch 40/50 ## 469/469 - 2s - 3ms/step - loss: 0.0630 - val_loss: 0.0627 ## Epoch 41/50 ## 469/469 - 2s - 3ms/step - loss: 0.0630 - val_loss: 0.0627 ## Epoch 42/50 ## 469/469 - 2s - 3ms/step - loss: 0.0630 - val_loss: 0.0626 ## Epoch 43/50 ## 469/469 - 2s - 3ms/step - loss: 0.0630 - val_loss: 0.0626 ## Epoch 44/50 ## 469/469 - 2s - 3ms/step - loss: 0.0629 - val_loss: 0.0626 ## Epoch 45/50 ## 469/469 - 2s - 3ms/step - loss: 0.0629 - val_loss: 0.0626 ## Epoch 46/50 ## 469/469 - 2s - 3ms/step - loss: 0.0629 - val_loss: 0.0625 ## Epoch 47/50 ## 469/469 - 2s - 3ms/step - loss: 0.0629 - val_loss: 0.0625 ## Epoch 48/50 ## 469/469 - 1s - 3ms/step - loss: 0.0628 - val_loss: 0.0625 ## Epoch 49/50 ## 469/469 - 2s - 3ms/step - loss: 0.0628 - val_loss: 0.0625 ## Epoch 50/50 ## 469/469 - 2s - 3ms/step - loss: 0.0628 - val_loss: 0.0625 predictions <- autoencoder |> predict(test_data) ## 313/313 - 1s - 2ms/step display(test_data, predictions) autoencoder |> fit(   x = noisy_train_data,   y = train_data,   epochs = 100,   batch_size = 128,   shuffle = TRUE,   validation_data = list(noisy_test_data, test_data), ) ## Epoch 1/100 ## 469/469 - 2s - 3ms/step - loss: 0.1006 - val_loss: 0.0939 ## Epoch 2/100 ## 469/469 - 2s - 3ms/step - loss: 0.0936 - val_loss: 0.0919 ## Epoch 3/100 ## 469/469 - 2s - 3ms/step - loss: 0.0921 - val_loss: 0.0908 ## Epoch 4/100 ## 469/469 - 2s - 3ms/step - loss: 0.0912 - val_loss: 0.0902 ## Epoch 5/100 ## 469/469 - 2s - 3ms/step - loss: 0.0905 - val_loss: 0.0896 ## Epoch 6/100 ## 469/469 - 2s - 3ms/step - loss: 0.0900 - val_loss: 0.0891 ## Epoch 7/100 ## 469/469 - 2s - 3ms/step - loss: 0.0895 - val_loss: 0.0887 ## Epoch 8/100 ## 469/469 - 2s - 3ms/step - loss: 0.0891 - val_loss: 0.0883 ## Epoch 9/100 ## 469/469 - 2s - 3ms/step - loss: 0.0887 - val_loss: 0.0880 ## Epoch 10/100 ## 469/469 - 2s - 3ms/step - loss: 0.0884 - val_loss: 0.0878 ## Epoch 11/100 ## 469/469 - 2s - 3ms/step - loss: 0.0882 - val_loss: 0.0875 ## Epoch 12/100 ## 469/469 - 2s - 3ms/step - loss: 0.0879 - val_loss: 0.0873 ## Epoch 13/100 ## 469/469 - 2s - 3ms/step - loss: 0.0877 - val_loss: 0.0871 ## Epoch 14/100 ## 469/469 - 2s - 3ms/step - loss: 0.0875 - val_loss: 0.0869 ## Epoch 15/100 ## 469/469 - 2s - 3ms/step - loss: 0.0873 - val_loss: 0.0867 ## Epoch 16/100 ## 469/469 - 2s - 3ms/step - loss: 0.0872 - val_loss: 0.0866 ## Epoch 17/100 ## 469/469 - 2s - 3ms/step - loss: 0.0870 - val_loss: 0.0864 ## Epoch 18/100 ## 469/469 - 2s - 3ms/step - loss: 0.0869 - val_loss: 0.0863 ## Epoch 19/100 ## 469/469 - 2s - 3ms/step - loss: 0.0868 - val_loss: 0.0862 ## Epoch 20/100 ## 469/469 - 2s - 3ms/step - loss: 0.0866 - val_loss: 0.0861 ## Epoch 21/100 ## 469/469 - 2s - 3ms/step - loss: 0.0865 - val_loss: 0.0860 ## Epoch 22/100 ## 469/469 - 2s - 3ms/step - loss: 0.0864 - val_loss: 0.0859 ## Epoch 23/100 ## 469/469 - 2s - 3ms/step - loss: 0.0863 - val_loss: 0.0858 ## Epoch 24/100 ## 469/469 - 1s - 3ms/step - loss: 0.0862 - val_loss: 0.0857 ## Epoch 25/100 ## 469/469 - 2s - 3ms/step - loss: 0.0862 - val_loss: 0.0856 ## Epoch 26/100 ## 469/469 - 2s - 3ms/step - loss: 0.0861 - val_loss: 0.0856 ## Epoch 27/100 ## 469/469 - 2s - 3ms/step - loss: 0.0860 - val_loss: 0.0855 ## Epoch 28/100 ## 469/469 - 2s - 3ms/step - loss: 0.0859 - val_loss: 0.0855 ## Epoch 29/100 ## 469/469 - 2s - 3ms/step - loss: 0.0859 - val_loss: 0.0854 ## Epoch 30/100 ## 469/469 - 2s - 3ms/step - loss: 0.0858 - val_loss: 0.0853 ## Epoch 31/100 ## 469/469 - 2s - 3ms/step - loss: 0.0858 - val_loss: 0.0853 ## Epoch 32/100 ## 469/469 - 2s - 3ms/step - loss: 0.0857 - val_loss: 0.0853 ## Epoch 33/100 ## 469/469 - 2s - 3ms/step - loss: 0.0857 - val_loss: 0.0852 ## Epoch 34/100 ## 469/469 - 2s - 3ms/step - loss: 0.0856 - val_loss: 0.0852 ## Epoch 35/100 ## 469/469 - 2s - 3ms/step - loss: 0.0856 - val_loss: 0.0851 ## Epoch 36/100 ## 469/469 - 2s - 3ms/step - loss: 0.0855 - val_loss: 0.0851 ## Epoch 37/100 ## 469/469 - 2s - 3ms/step - loss: 0.0855 - val_loss: 0.0851 ## Epoch 38/100 ## 469/469 - 2s - 3ms/step - loss: 0.0855 - val_loss: 0.0850 ## Epoch 39/100 ## 469/469 - 2s - 3ms/step - loss: 0.0854 - val_loss: 0.0850 ## Epoch 40/100 ## 469/469 - 2s - 3ms/step - loss: 0.0854 - val_loss: 0.0850 ## Epoch 41/100 ## 469/469 - 2s - 3ms/step - loss: 0.0854 - val_loss: 0.0850 ## Epoch 42/100 ## 469/469 - 2s - 3ms/step - loss: 0.0853 - val_loss: 0.0849 ## Epoch 43/100 ## 469/469 - 2s - 3ms/step - loss: 0.0853 - val_loss: 0.0849 ## Epoch 44/100 ## 469/469 - 2s - 3ms/step - loss: 0.0853 - val_loss: 0.0849 ## Epoch 45/100 ## 469/469 - 2s - 3ms/step - loss: 0.0853 - val_loss: 0.0849 ## Epoch 46/100 ## 469/469 - 2s - 3ms/step - loss: 0.0852 - val_loss: 0.0848 ## Epoch 47/100 ## 469/469 - 2s - 3ms/step - loss: 0.0852 - val_loss: 0.0848 ## Epoch 48/100 ## 469/469 - 2s - 3ms/step - loss: 0.0852 - val_loss: 0.0848 ## Epoch 49/100 ## 469/469 - 2s - 3ms/step - loss: 0.0852 - val_loss: 0.0848 ## Epoch 50/100 ## 469/469 - 2s - 3ms/step - loss: 0.0851 - val_loss: 0.0848 ## Epoch 51/100 ## 469/469 - 2s - 3ms/step - loss: 0.0851 - val_loss: 0.0847 ## Epoch 52/100 ## 469/469 - 2s - 3ms/step - loss: 0.0851 - val_loss: 0.0847 ## Epoch 53/100 ## 469/469 - 2s - 3ms/step - loss: 0.0851 - val_loss: 0.0847 ## Epoch 54/100 ## 469/469 - 2s - 3ms/step - loss: 0.0850 - val_loss: 0.0847 ## Epoch 55/100 ## 469/469 - 2s - 3ms/step - loss: 0.0850 - val_loss: 0.0847 ## Epoch 56/100 ## 469/469 - 2s - 3ms/step - loss: 0.0850 - val_loss: 0.0846 ## Epoch 57/100 ## 469/469 - 2s - 3ms/step - loss: 0.0850 - val_loss: 0.0846 ## Epoch 58/100 ## 469/469 - 2s - 3ms/step - loss: 0.0850 - val_loss: 0.0846 ## Epoch 59/100 ## 469/469 - 2s - 3ms/step - loss: 0.0849 - val_loss: 0.0846 ## Epoch 60/100 ## 469/469 - 2s - 3ms/step - loss: 0.0849 - val_loss: 0.0846 ## Epoch 61/100 ## 469/469 - 2s - 3ms/step - loss: 0.0849 - val_loss: 0.0846 ## Epoch 62/100 ## 469/469 - 2s - 3ms/step - loss: 0.0849 - val_loss: 0.0846 ## Epoch 63/100 ## 469/469 - 1s - 3ms/step - loss: 0.0849 - val_loss: 0.0845 ## Epoch 64/100 ## 469/469 - 2s - 3ms/step - loss: 0.0849 - val_loss: 0.0845 ## Epoch 65/100 ## 469/469 - 2s - 3ms/step - loss: 0.0848 - val_loss: 0.0845 ## Epoch 66/100 ## 469/469 - 2s - 3ms/step - loss: 0.0848 - val_loss: 0.0845 ## Epoch 67/100 ## 469/469 - 2s - 3ms/step - loss: 0.0848 - val_loss: 0.0845 ## Epoch 68/100 ## 469/469 - 2s - 3ms/step - loss: 0.0848 - val_loss: 0.0845 ## Epoch 69/100 ## 469/469 - 2s - 3ms/step - loss: 0.0848 - val_loss: 0.0845 ## Epoch 70/100 ## 469/469 - 2s - 3ms/step - loss: 0.0848 - val_loss: 0.0845 ## Epoch 71/100 ## 469/469 - 2s - 3ms/step - loss: 0.0848 - val_loss: 0.0844 ## Epoch 72/100 ## 469/469 - 2s - 3ms/step - loss: 0.0847 - val_loss: 0.0844 ## Epoch 73/100 ## 469/469 - 2s - 3ms/step - loss: 0.0847 - val_loss: 0.0844 ## Epoch 74/100 ## 469/469 - 2s - 3ms/step - loss: 0.0847 - val_loss: 0.0844 ## Epoch 75/100 ## 469/469 - 1s - 3ms/step - loss: 0.0847 - val_loss: 0.0844 ## Epoch 76/100 ## 469/469 - 2s - 3ms/step - loss: 0.0847 - val_loss: 0.0844 ## Epoch 77/100 ## 469/469 - 2s - 3ms/step - loss: 0.0847 - val_loss: 0.0844 ## Epoch 78/100 ## 469/469 - 2s - 3ms/step - loss: 0.0847 - val_loss: 0.0844 ## Epoch 79/100 ## 469/469 - 2s - 3ms/step - loss: 0.0847 - val_loss: 0.0844 ## Epoch 80/100 ## 469/469 - 2s - 3ms/step - loss: 0.0847 - val_loss: 0.0844 ## Epoch 81/100 ## 469/469 - 2s - 3ms/step - loss: 0.0846 - val_loss: 0.0844 ## Epoch 82/100 ## 469/469 - 2s - 3ms/step - loss: 0.0846 - val_loss: 0.0844 ## Epoch 83/100 ## 469/469 - 2s - 3ms/step - loss: 0.0846 - val_loss: 0.0844 ## Epoch 84/100 ## 469/469 - 2s - 3ms/step - loss: 0.0846 - val_loss: 0.0843 ## Epoch 85/100 ## 469/469 - 2s - 3ms/step - loss: 0.0846 - val_loss: 0.0843 ## Epoch 86/100 ## 469/469 - 2s - 3ms/step - loss: 0.0846 - val_loss: 0.0843 ## Epoch 87/100 ## 469/469 - 1s - 3ms/step - loss: 0.0846 - val_loss: 0.0843 ## Epoch 88/100 ## 469/469 - 2s - 3ms/step - loss: 0.0846 - val_loss: 0.0843 ## Epoch 89/100 ## 469/469 - 2s - 3ms/step - loss: 0.0846 - val_loss: 0.0843 ## Epoch 90/100 ## 469/469 - 2s - 3ms/step - loss: 0.0846 - val_loss: 0.0843 ## Epoch 91/100 ## 469/469 - 2s - 3ms/step - loss: 0.0845 - val_loss: 0.0843 ## Epoch 92/100 ## 469/469 - 2s - 3ms/step - loss: 0.0845 - val_loss: 0.0843 ## Epoch 93/100 ## 469/469 - 2s - 3ms/step - loss: 0.0845 - val_loss: 0.0843 ## Epoch 94/100 ## 469/469 - 2s - 3ms/step - loss: 0.0845 - val_loss: 0.0843 ## Epoch 95/100 ## 469/469 - 2s - 3ms/step - loss: 0.0845 - val_loss: 0.0843 ## Epoch 96/100 ## 469/469 - 1s - 3ms/step - loss: 0.0845 - val_loss: 0.0842 ## Epoch 97/100 ## 469/469 - 2s - 3ms/step - loss: 0.0845 - val_loss: 0.0842 ## Epoch 98/100 ## 469/469 - 2s - 3ms/step - loss: 0.0845 - val_loss: 0.0842 ## Epoch 99/100 ## 469/469 - 2s - 3ms/step - loss: 0.0845 - val_loss: 0.0842 ## Epoch 100/100 ## 469/469 - 1s - 3ms/step - loss: 0.0845 - val_loss: 0.0842 predictions <- autoencoder |> predict(noisy_test_data) ## 313/313 - 0s - 775us/step display(noisy_test_data, predictions)"},{"path":"https://keras3.posit.co/dev/articles/examples/vision/mnist_convnet.html","id":"setup","dir":"Articles > Examples > Vision","previous_headings":"","what":"Setup","title":"Simple MNIST convnet","text":"","code":"library(keras3)"},{"path":"https://keras3.posit.co/dev/articles/examples/vision/mnist_convnet.html","id":"prepare-the-data","dir":"Articles > Examples > Vision","previous_headings":"","what":"Prepare the data","title":"Simple MNIST convnet","text":"","code":"# Model / data parameters num_classes <- 10 input_shape <- c(28, 28, 1)  # Load the data and split it between train and test sets c(c(x_train, y_train), c(x_test, y_test)) %<-% dataset_mnist()  # Scale images to the [0, 1] range x_train <- x_train / 255 x_test <- x_test / 255 # Make sure images have shape (28, 28, 1) x_train <- op_expand_dims(x_train, -1) x_test <- op_expand_dims(x_test, -1)   dim(x_train) ## [1] 60000    28    28     1 dim(x_test) ## [1] 10000    28    28     1 # convert class vectors to binary class matrices y_train <- to_categorical(y_train, num_classes) y_test <- to_categorical(y_test, num_classes)"},{"path":"https://keras3.posit.co/dev/articles/examples/vision/mnist_convnet.html","id":"build-the-model","dir":"Articles > Examples > Vision","previous_headings":"","what":"Build the model","title":"Simple MNIST convnet","text":"","code":"model <- keras_model_sequential(input_shape = input_shape) model |>   layer_conv_2d(filters = 32, kernel_size = c(3, 3), activation = \"relu\") |>   layer_max_pooling_2d(pool_size = c(2, 2)) |>   layer_conv_2d(filters = 64, kernel_size = c(3, 3), activation = \"relu\") |>   layer_max_pooling_2d(pool_size = c(2, 2)) |>   layer_flatten() |>   layer_dropout(rate = 0.5) |>   layer_dense(units = num_classes, activation = \"softmax\")  summary(model) ## Model: \"sequential\" ## ┏━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━┓ ## ┃ Layer (type)                    ┃ Output Shape           ┃       Param # ┃ ## ┡━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━┩ ## │ conv2d (Conv2D)                 │ (None, 26, 26, 32)     │           320 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ max_pooling2d (MaxPooling2D)    │ (None, 13, 13, 32)     │             0 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ conv2d_1 (Conv2D)               │ (None, 11, 11, 64)     │        18,496 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ max_pooling2d_1 (MaxPooling2D)  │ (None, 5, 5, 64)       │             0 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ flatten (Flatten)               │ (None, 1600)           │             0 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ dropout (Dropout)               │ (None, 1600)           │             0 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ dense (Dense)                   │ (None, 10)             │        16,010 │ ## └─────────────────────────────────┴────────────────────────┴───────────────┘ ##  Total params: 34,826 (136.04 KB) ##  Trainable params: 34,826 (136.04 KB) ##  Non-trainable params: 0 (0.00 B)"},{"path":"https://keras3.posit.co/dev/articles/examples/vision/mnist_convnet.html","id":"train-the-model","dir":"Articles > Examples > Vision","previous_headings":"","what":"Train the model","title":"Simple MNIST convnet","text":"","code":"batch_size <- 128 epochs <- 15  model |> compile(   loss = \"categorical_crossentropy\",   optimizer = \"adam\",   metrics = \"accuracy\" )  model |> fit(   x_train, y_train,   batch_size = batch_size,   epochs = epochs,   validation_split = 0.1 ) ## Epoch 1/15 ## 422/422 - 5s - 11ms/step - accuracy: 0.8896 - loss: 0.3635 - val_accuracy: 0.9783 - val_loss: 0.0791 ## Epoch 2/15 ## 422/422 - 1s - 2ms/step - accuracy: 0.9666 - loss: 0.1111 - val_accuracy: 0.9850 - val_loss: 0.0548 ## Epoch 3/15 ## 422/422 - 1s - 2ms/step - accuracy: 0.9744 - loss: 0.0824 - val_accuracy: 0.9883 - val_loss: 0.0441 ## Epoch 4/15 ## 422/422 - 1s - 2ms/step - accuracy: 0.9786 - loss: 0.0694 - val_accuracy: 0.9895 - val_loss: 0.0401 ## Epoch 5/15 ## 422/422 - 1s - 2ms/step - accuracy: 0.9804 - loss: 0.0625 - val_accuracy: 0.9903 - val_loss: 0.0353 ## Epoch 6/15 ## 422/422 - 1s - 2ms/step - accuracy: 0.9824 - loss: 0.0557 - val_accuracy: 0.9907 - val_loss: 0.0332 ## Epoch 7/15 ## 422/422 - 1s - 2ms/step - accuracy: 0.9838 - loss: 0.0500 - val_accuracy: 0.9917 - val_loss: 0.0312 ## Epoch 8/15 ## 422/422 - 1s - 2ms/step - accuracy: 0.9850 - loss: 0.0480 - val_accuracy: 0.9918 - val_loss: 0.0304 ## Epoch 9/15 ## 422/422 - 1s - 2ms/step - accuracy: 0.9860 - loss: 0.0445 - val_accuracy: 0.9918 - val_loss: 0.0303 ## Epoch 10/15 ## 422/422 - 1s - 2ms/step - accuracy: 0.9861 - loss: 0.0442 - val_accuracy: 0.9912 - val_loss: 0.0293 ## Epoch 11/15 ## 422/422 - 1s - 2ms/step - accuracy: 0.9876 - loss: 0.0390 - val_accuracy: 0.9915 - val_loss: 0.0307 ## Epoch 12/15 ## 422/422 - 1s - 2ms/step - accuracy: 0.9876 - loss: 0.0374 - val_accuracy: 0.9925 - val_loss: 0.0288 ## Epoch 13/15 ## 422/422 - 1s - 2ms/step - accuracy: 0.9887 - loss: 0.0346 - val_accuracy: 0.9920 - val_loss: 0.0291 ## Epoch 14/15 ## 422/422 - 1s - 2ms/step - accuracy: 0.9887 - loss: 0.0343 - val_accuracy: 0.9923 - val_loss: 0.0276 ## Epoch 15/15 ## 422/422 - 1s - 2ms/step - accuracy: 0.9897 - loss: 0.0323 - val_accuracy: 0.9922 - val_loss: 0.0286"},{"path":"https://keras3.posit.co/dev/articles/examples/vision/mnist_convnet.html","id":"evaluate-the-trained-model","dir":"Articles > Examples > Vision","previous_headings":"","what":"Evaluate the trained model","title":"Simple MNIST convnet","text":"","code":"score <- model |> evaluate(x_test, y_test, verbose = 0) score ## $accuracy ## [1] 0.9915 ## ## $loss ## [1] 0.02471612"},{"path":"https://keras3.posit.co/dev/articles/examples/vision/mnist_siamese_graph.html","id":"introduction","dir":"Articles > Examples > Vision","previous_headings":"","what":"Introduction","title":"Train a Siamese MLP on pairs of digits from the MNIST dataset.","text":"Siamese Networks neural networks share weights two sister networks, producing embedding vectors respective inputs. supervised similarity learning, networks trained maximize contrast (distance) embeddings inputs different classes, minimizing distance embeddings similar classes, resulting embedding spaces reflect class segmentation training inputs. implementation loosely follows Hadsell-et-al.’06 [1] (see paper mode details) euclidean distance replaced subtraction layer one fully-connect (FC) layer. [1] “Dimensionality Reduction Learning Invariant Mapping” https://yann.lecun.com/exdb/publis/pdf/hadsell-chopra-lecun-06.pdf Gets 98.11% test accuracy 20 epochs. 3 seconds per epoch AMD Ryzen 7 PRO 4750U (CPU)","code":"library(keras3) contrastive_loss <- function(y_true, y_pred) {     # Contrastive loss from Hadsell-et-al.'06     # https://yann.lecun.com/exdb/publis/pdf/hadsell-chopra-lecun-06.pdf     margin = 1     margin_square = op_square(op_maximum(margin - (y_pred), 0))     op_mean((1 - y_true) * op_square(y_pred) + (y_true) * margin_square) }"},{"path":"https://keras3.posit.co/dev/articles/examples/vision/mnist_siamese_graph.html","id":"create-pairs-of-images","dir":"Articles > Examples > Vision","previous_headings":"","what":"Create pairs of images","title":"Train a Siamese MLP on pairs of digits from the MNIST dataset.","text":"train model differentiate digits different classes. example, digit 0 needs differentiated rest digits (1 9), digit 1 - 0 2 9, . carry , select N random images class (example, digit 0) pair N random images another class B (example, digit 1). , can repeat process classes digits (digit 9). paired digit 0 digits, can repeat process remaining classes rest digits (1 9).","code":"create_pairs <- function(x, y) {     # Positive and negative pair creation.     # Alternates between positive and negative pairs.     digit_indices <- tapply(1:length(y), y, list)     y1 <- y     y2 <- sapply(y, function(a) sample(0:9,1,prob=0.1+0.8*(0:9==a)))     idx1 <- 1:nrow(x)     idx2 <- sapply(as.character(y2), function(a) sample(digit_indices[[a]],1))     is_same  <- 1*(y1==y2)     list(pair1 = x[idx1,], pair2 = x[idx2,], y = is_same) }  compute_accuracy <- function(predictions, labels) {     # Compute classification accuracy with a fixed threshold on distances.     mean(labels[predictions > 0.5]) } # the data, shuffled and split between train and test sets mnist   <- dataset_mnist() x_train <- mnist$train$x y_train <- mnist$train$y x_test  <- mnist$test$x y_test  <- mnist$test$y x_train <- array_reshape(x_train, c(nrow(x_train), 784)) x_test  <- array_reshape(x_test, c(nrow(x_test), 784)) x_train <- x_train / 255 x_test  <- x_test / 255  # create training+test positive and negative pairs tr <- create_pairs(x_train, y_train) te <- create_pairs(x_test,  y_test)  names(tr) ## [1] \"pair1\" \"pair2\" \"y\""},{"path":"https://keras3.posit.co/dev/articles/examples/vision/mnist_siamese_graph.html","id":"network-definition","dir":"Articles > Examples > Vision","previous_headings":"","what":"Network definition","title":"Train a Siamese MLP on pairs of digits from the MNIST dataset.","text":"","code":"# input layers input_dim = 784 input_1 <- layer_input(shape = c(input_dim)) input_2 <- layer_input(shape = c(input_dim))  # definition of the base network that will be shared base_network <- keras_model_sequential() %>%     layer_dense(units = 128, activation = 'relu') %>%     layer_dropout(rate = 0.1) %>%     layer_dense(units = 128, activation = 'relu') %>%     layer_dropout(rate = 0.1) %>%     layer_dense(units = 128, activation = 'relu')  # because we re-use the same instance `base_network`, the weights of # the network will be shared across the two branches branch_1 <- base_network(input_1) branch_2 <- base_network(input_2)  # merging layer out <- layer_subtract(list(branch_1, branch_2)) %>%     layer_dropout(rate = 0.1) %>%     layer_dense(units = 16, activation = 'relu') %>%     layer_dense(1, activation = \"sigmoid\")  # create and compile model model <- keras_model(list(input_1, input_2), out)"},{"path":"https://keras3.posit.co/dev/articles/examples/vision/mnist_siamese_graph.html","id":"train","dir":"Articles > Examples > Vision","previous_headings":"","what":"Train","title":"Train a Siamese MLP on pairs of digits from the MNIST dataset.","text":"","code":"model %>% compile(     optimizer = \"rmsprop\",     #loss = \"binary_crossentropy\",     loss = contrastive_loss,     metrics = metric_binary_accuracy )  history <- model %>% fit(     list(tr$pair1, tr$pair2), tr$y,     batch_size = 128,     epochs = 20,     validation_data = list(         list(te$pair1, te$pair2),         te$y     ) ) ## Epoch 1/20 ## 469/469 - 6s - 13ms/step - binary_accuracy: 0.7569 - loss: 0.1640 - val_binary_accuracy: 0.8798 - val_loss: 0.0974 ## Epoch 2/20 ## 469/469 - 1s - 2ms/step - binary_accuracy: 0.8899 - loss: 0.0866 - val_binary_accuracy: 0.9227 - val_loss: 0.0626 ## Epoch 3/20 ## 469/469 - 1s - 2ms/step - binary_accuracy: 0.9233 - loss: 0.0603 - val_binary_accuracy: 0.9398 - val_loss: 0.0497 ## Epoch 4/20 ## 469/469 - 1s - 2ms/step - binary_accuracy: 0.9423 - loss: 0.0460 - val_binary_accuracy: 0.9480 - val_loss: 0.0399 ## Epoch 5/20 ## 469/469 - 1s - 2ms/step - binary_accuracy: 0.9561 - loss: 0.0356 - val_binary_accuracy: 0.9578 - val_loss: 0.0330 ## Epoch 6/20 ## 469/469 - 1s - 2ms/step - binary_accuracy: 0.9637 - loss: 0.0295 - val_binary_accuracy: 0.9613 - val_loss: 0.0313 ## Epoch 7/20 ## 469/469 - 1s - 2ms/step - binary_accuracy: 0.9686 - loss: 0.0252 - val_binary_accuracy: 0.9645 - val_loss: 0.0275 ## Epoch 8/20 ## 469/469 - 1s - 2ms/step - binary_accuracy: 0.9730 - loss: 0.0217 - val_binary_accuracy: 0.9672 - val_loss: 0.0264 ## Epoch 9/20 ## 469/469 - 1s - 2ms/step - binary_accuracy: 0.9768 - loss: 0.0190 - val_binary_accuracy: 0.9686 - val_loss: 0.0246 ## Epoch 10/20 ## 469/469 - 1s - 2ms/step - binary_accuracy: 0.9793 - loss: 0.0169 - val_binary_accuracy: 0.9701 - val_loss: 0.0242 ## Epoch 11/20 ## 469/469 - 1s - 2ms/step - binary_accuracy: 0.9805 - loss: 0.0158 - val_binary_accuracy: 0.9719 - val_loss: 0.0233 ## Epoch 12/20 ## 469/469 - 1s - 2ms/step - binary_accuracy: 0.9822 - loss: 0.0148 - val_binary_accuracy: 0.9718 - val_loss: 0.0230 ## Epoch 13/20 ## 469/469 - 1s - 2ms/step - binary_accuracy: 0.9833 - loss: 0.0135 - val_binary_accuracy: 0.9702 - val_loss: 0.0234 ## Epoch 14/20 ## 469/469 - 1s - 2ms/step - binary_accuracy: 0.9846 - loss: 0.0129 - val_binary_accuracy: 0.9730 - val_loss: 0.0220 ## Epoch 15/20 ## 469/469 - 1s - 2ms/step - binary_accuracy: 0.9857 - loss: 0.0117 - val_binary_accuracy: 0.9759 - val_loss: 0.0193 ## Epoch 16/20 ## 469/469 - 1s - 2ms/step - binary_accuracy: 0.9868 - loss: 0.0111 - val_binary_accuracy: 0.9749 - val_loss: 0.0217 ## Epoch 17/20 ## 469/469 - 1s - 2ms/step - binary_accuracy: 0.9872 - loss: 0.0106 - val_binary_accuracy: 0.9741 - val_loss: 0.0211 ## Epoch 18/20 ## 469/469 - 1s - 2ms/step - binary_accuracy: 0.9880 - loss: 0.0099 - val_binary_accuracy: 0.9746 - val_loss: 0.0201 ## Epoch 19/20 ## 469/469 - 1s - 2ms/step - binary_accuracy: 0.9879 - loss: 0.0100 - val_binary_accuracy: 0.9756 - val_loss: 0.0207 ## Epoch 20/20 ## 469/469 - 1s - 2ms/step - binary_accuracy: 0.9888 - loss: 0.0091 - val_binary_accuracy: 0.9776 - val_loss: 0.0184 plot(history)"},{"path":"https://keras3.posit.co/dev/articles/examples/vision/mnist_siamese_graph.html","id":"evaluate","dir":"Articles > Examples > Vision","previous_headings":"","what":"Evaluate","title":"Train a Siamese MLP on pairs of digits from the MNIST dataset.","text":"","code":"# compute final accuracy on training and test sets  tr_pred <- predict(model, list(tr$pair1, tr$pair2))[,1] ## 1875/1875 - 2s - 1ms/step tr_acc  <- compute_accuracy(tr_pred, tr$y) te_pred <- predict(model, list(te$pair1, te$pair2))[,1] ## 313/313 - 1s - 2ms/step te_acc  <- compute_accuracy(te_pred, te$y)  sprintf('* Accuracy on training set: %0.2f%%', (100 * tr_acc)) ## [1] \"* Accuracy on training set: 99.67%\" sprintf('* Accuracy on test set: %0.2f%%', (100 * te_acc)) ## [1] \"* Accuracy on test set: 98.25%\""},{"path":"https://keras3.posit.co/dev/articles/examples/vision/mnist_siamese_graph.html","id":"plots","dir":"Articles > Examples > Vision","previous_headings":"","what":"Plots","title":"Train a Siamese MLP on pairs of digits from the MNIST dataset.","text":"","code":"par(mfrow=c(1,1)) vioplot::vioplot( te_pred ~ te$y ) i=3 visualizePair <- function(i) {     image(rbind(matrix( te$pair1[i,],28,28)[,28:1], matrix( te$pair2[i,],28,28)[,28:1]))     title(paste(\"true:\", te$y[i],\"|  pred:\", round(te_pred[i],5))) } par(mfrow=c(3,3)) lapply(1:9, visualizePair) ## [[1]] ## NULL ## ## [[2]] ## NULL ## ## [[3]] ## NULL ## ## [[4]] ## NULL ## ## [[5]] ## NULL ## ## [[6]] ## NULL ## ## [[7]] ## NULL ## ## [[8]] ## NULL ## ## [[9]] ## NULL"},{"path":"https://keras3.posit.co/dev/articles/examples/vision/oxford_pets_image_segmentation.html","id":"download-the-data","dir":"Articles > Examples > Vision","previous_headings":"","what":"Download the data","title":"Image segmentation with a U-Net-like architecture","text":"","code":"options(timeout = 5000) download.file(   \"https://www.robots.ox.ac.uk/~vgg/data/pets/data/images.tar.gz\",   \"datasets/images.tar.gz\" ) download.file(   \"https://www.robots.ox.ac.uk/~vgg/data/pets/data/annotations.tar.gz\",   \"datasets/annotations.tar.gz\" )  untar(\"datasets/images.tar.gz\", exdir = \"datasets\") untar(\"datasets/annotations.tar.gz\", exdir = \"datasets\")"},{"path":"https://keras3.posit.co/dev/articles/examples/vision/oxford_pets_image_segmentation.html","id":"prepare-paths-of-input-images-and-target-segmentation-masks","dir":"Articles > Examples > Vision","previous_headings":"","what":"Prepare paths of input images and target segmentation masks","title":"Image segmentation with a U-Net-like architecture","text":"","code":"library(keras3) input_dir <- \"datasets/images/\" target_dir <- \"datasets/annotations/trimaps/\" img_size <- c(160, 160) num_classes <- 3 batch_size <- 32  input_img_paths <- fs::dir_ls(input_dir, glob = \"*.jpg\") |> sort() target_img_paths <- fs::dir_ls(target_dir, glob = \"*.png\") |> sort()  cat(\"Number of samples:\", length(input_img_paths), \"\\n\") ## Number of samples: 7390 for (i in 1:10) {   cat(input_img_paths[i], \"|\", target_img_paths[i], \"\\n\") } ## datasets/images/Abyssinian_1.jpg | datasets/annotations/trimaps/Abyssinian_1.png ## datasets/images/Abyssinian_10.jpg | datasets/annotations/trimaps/Abyssinian_10.png ## datasets/images/Abyssinian_100.jpg | datasets/annotations/trimaps/Abyssinian_100.png ## datasets/images/Abyssinian_101.jpg | datasets/annotations/trimaps/Abyssinian_101.png ## datasets/images/Abyssinian_102.jpg | datasets/annotations/trimaps/Abyssinian_102.png ## datasets/images/Abyssinian_103.jpg | datasets/annotations/trimaps/Abyssinian_103.png ## datasets/images/Abyssinian_104.jpg | datasets/annotations/trimaps/Abyssinian_104.png ## datasets/images/Abyssinian_105.jpg | datasets/annotations/trimaps/Abyssinian_105.png ## datasets/images/Abyssinian_106.jpg | datasets/annotations/trimaps/Abyssinian_106.png ## datasets/images/Abyssinian_107.jpg | datasets/annotations/trimaps/Abyssinian_107.png"},{"path":"https://keras3.posit.co/dev/articles/examples/vision/oxford_pets_image_segmentation.html","id":"what-does-one-input-image-and-corresponding-segmentation-mask-look-like","dir":"Articles > Examples > Vision","previous_headings":"","what":"What does one input image and corresponding segmentation mask look like?","title":"Image segmentation with a U-Net-like architecture","text":"","code":"# Display input image #10 input_img_paths[10] |>   jpeg::readJPEG() |>   as.raster() |>   plot() target_img_paths[10] |>   png::readPNG() |>   magrittr::multiply_by(255)|>   as.raster(max = 3) |>   plot()"},{"path":"https://keras3.posit.co/dev/articles/examples/vision/oxford_pets_image_segmentation.html","id":"prepare-dataset-to-load-vectorize-batches-of-data","dir":"Articles > Examples > Vision","previous_headings":"","what":"Prepare dataset to load & vectorize batches of data","title":"Image segmentation with a U-Net-like architecture","text":"","code":"library(tensorflow, exclude = c(\"shape\", \"set_random_seed\")) library(tfdatasets, exclude = \"shape\")   # Returns a tf_dataset get_dataset <- function(batch_size, img_size, input_img_paths, target_img_paths,                         max_dataset_len = NULL) {    img_size <- as.integer(img_size)    load_img_masks <- function(input_img_path, target_img_path) {     input_img <- input_img_path |>       tf$io$read_file() |>       tf$io$decode_jpeg(channels = 3) |>       tf$image$resize(img_size) |>       tf$image$convert_image_dtype(\"float32\")      target_img <- target_img_path |>       tf$io$read_file() |>       tf$io$decode_png(channels = 1) |>       tf$image$resize(img_size, method = \"nearest\") |>       tf$image$convert_image_dtype(\"uint8\")      # Ground truth labels are 1, 2, 3. Subtract one to make them 0, 1, 2:     target_img <- target_img - 1L      list(input_img, target_img)   }    if (!is.null(max_dataset_len)) {     input_img_paths <- input_img_paths[1:max_dataset_len]     target_img_paths <- target_img_paths[1:max_dataset_len]   }    list(input_img_paths, target_img_paths) |>     tensor_slices_dataset() |>     dataset_map(load_img_masks, num_parallel_calls = tf$data$AUTOTUNE)|>     dataset_batch(batch_size) }"},{"path":"https://keras3.posit.co/dev/articles/examples/vision/oxford_pets_image_segmentation.html","id":"prepare-u-net-xception-style-model","dir":"Articles > Examples > Vision","previous_headings":"","what":"Prepare U-Net Xception-style model","title":"Image segmentation with a U-Net-like architecture","text":"","code":"get_model <- function(img_size, num_classes) {    inputs <- keras_input(shape = c(img_size, 3))    ### [First half of the network: downsampling inputs] ###    # Entry block   x <- inputs |>     layer_conv_2d(filters = 32, kernel_size = 3, strides = 2, padding = \"same\") |>     layer_batch_normalization() |>     layer_activation(\"relu\")    previous_block_activation <- x  # Set aside residual    for (filters in c(64, 128, 256)) {     x <- x |>       layer_activation(\"relu\") |>       layer_separable_conv_2d(filters = filters, kernel_size = 3, padding = \"same\") |>       layer_batch_normalization() |>        layer_activation(\"relu\") |>       layer_separable_conv_2d(filters = filters, kernel_size = 3, padding = \"same\") |>       layer_batch_normalization() |>        layer_max_pooling_2d(pool_size = 3, strides = 2, padding = \"same\")      residual <- previous_block_activation |>       layer_conv_2d(filters = filters, kernel_size = 1, strides = 2, padding = \"same\")      x <- layer_add(x, residual)  # Add back residual     previous_block_activation <- x  # Set aside next residual   }    ### [Second half of the network: upsampling inputs] ###    for (filters in c(256, 128, 64, 32)) {     x <- x |>       layer_activation(\"relu\") |>       layer_conv_2d_transpose(filters = filters, kernel_size = 3, padding = \"same\") |>       layer_batch_normalization() |>        layer_activation(\"relu\") |>       layer_conv_2d_transpose(filters = filters, kernel_size = 3, padding = \"same\") |>       layer_batch_normalization() |>        layer_upsampling_2d(size = 2)      # Project residual     residual <- previous_block_activation |>       layer_upsampling_2d(size = 2) |>       layer_conv_2d(filters = filters, kernel_size = 1, padding = \"same\")      x <- layer_add(x, residual)     # Add back residual     previous_block_activation <- x  # Set aside next residual   }    # Add a per-pixel classification layer   outputs <- x |>     layer_conv_2d(num_classes, 3, activation = \"softmax\", padding = \"same\")    # Define the model   keras_model(inputs, outputs) }  # Build model model <- get_model(img_size, num_classes) summary(model) ## Model: \"functional\" ## ┏━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━┳━━━━━━━━━━━━━━━━┳━━━━━━━┓ ## ┃ Layer (type)      ┃ Output Shape    ┃   Param # ┃ Connected to   ┃ Trai… ┃ ## ┡━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━╇━━━━━━━━━━━━━━━━╇━━━━━━━┩ ## │ input_layer       │ (None, 160,     │         0 │ -              │   -   │ ## │ (InputLayer)      │ 160, 3)         │           │                │       │ ## ├───────────────────┼─────────────────┼───────────┼────────────────┼───────┤ ## │ conv2d (Conv2D)   │ (None, 80, 80,  │       896 │ input_layer[0… │   Y   │ ## │                   │ 32)             │           │                │       │ ## ├───────────────────┼─────────────────┼───────────┼────────────────┼───────┤ ## │ batch_normalizat… │ (None, 80, 80,  │       128 │ conv2d[0][0]   │   Y   │ ## │ (BatchNormalizat… │ 32)             │           │                │       │ ## ├───────────────────┼─────────────────┼───────────┼────────────────┼───────┤ ## │ activation        │ (None, 80, 80,  │         0 │ batch_normali… │   -   │ ## │ (Activation)      │ 32)             │           │                │       │ ## ├───────────────────┼─────────────────┼───────────┼────────────────┼───────┤ ## │ activation_1      │ (None, 80, 80,  │         0 │ activation[0]… │   -   │ ## │ (Activation)      │ 32)             │           │                │       │ ## ├───────────────────┼─────────────────┼───────────┼────────────────┼───────┤ ## │ separable_conv2d  │ (None, 80, 80,  │     2,400 │ activation_1[… │   Y   │ ## │ (SeparableConv2D) │ 64)             │           │                │       │ ## ├───────────────────┼─────────────────┼───────────┼────────────────┼───────┤ ## │ batch_normalizat… │ (None, 80, 80,  │       256 │ separable_con… │   Y   │ ## │ (BatchNormalizat… │ 64)             │           │                │       │ ## ├───────────────────┼─────────────────┼───────────┼────────────────┼───────┤ ## │ activation_2      │ (None, 80, 80,  │         0 │ batch_normali… │   -   │ ## │ (Activation)      │ 64)             │           │                │       │ ## ├───────────────────┼─────────────────┼───────────┼────────────────┼───────┤ ## │ separable_conv2d… │ (None, 80, 80,  │     4,736 │ activation_2[… │   Y   │ ## │ (SeparableConv2D) │ 64)             │           │                │       │ ## ├───────────────────┼─────────────────┼───────────┼────────────────┼───────┤ ## │ batch_normalizat… │ (None, 80, 80,  │       256 │ separable_con… │   Y   │ ## │ (BatchNormalizat… │ 64)             │           │                │       │ ## ├───────────────────┼─────────────────┼───────────┼────────────────┼───────┤ ## │ max_pooling2d     │ (None, 40, 40,  │         0 │ batch_normali… │   -   │ ## │ (MaxPooling2D)    │ 64)             │           │                │       │ ## ├───────────────────┼─────────────────┼───────────┼────────────────┼───────┤ ## │ conv2d_1 (Conv2D) │ (None, 40, 40,  │     2,112 │ activation[0]… │   Y   │ ## │                   │ 64)             │           │                │       │ ## ├───────────────────┼─────────────────┼───────────┼────────────────┼───────┤ ## │ add (Add)         │ (None, 40, 40,  │         0 │ max_pooling2d… │   -   │ ## │                   │ 64)             │           │ conv2d_1[0][0] │       │ ## ├───────────────────┼─────────────────┼───────────┼────────────────┼───────┤ ## │ activation_3      │ (None, 40, 40,  │         0 │ add[0][0]      │   -   │ ## │ (Activation)      │ 64)             │           │                │       │ ## ├───────────────────┼─────────────────┼───────────┼────────────────┼───────┤ ## │ separable_conv2d… │ (None, 40, 40,  │     8,896 │ activation_3[… │   Y   │ ## │ (SeparableConv2D) │ 128)            │           │                │       │ ## ├───────────────────┼─────────────────┼───────────┼────────────────┼───────┤ ## │ batch_normalizat… │ (None, 40, 40,  │       512 │ separable_con… │   Y   │ ## │ (BatchNormalizat… │ 128)            │           │                │       │ ## ├───────────────────┼─────────────────┼───────────┼────────────────┼───────┤ ## │ activation_4      │ (None, 40, 40,  │         0 │ batch_normali… │   -   │ ## │ (Activation)      │ 128)            │           │                │       │ ## ├───────────────────┼─────────────────┼───────────┼────────────────┼───────┤ ## │ separable_conv2d… │ (None, 40, 40,  │    17,664 │ activation_4[… │   Y   │ ## │ (SeparableConv2D) │ 128)            │           │                │       │ ## ├───────────────────┼─────────────────┼───────────┼────────────────┼───────┤ ## │ batch_normalizat… │ (None, 40, 40,  │       512 │ separable_con… │   Y   │ ## │ (BatchNormalizat… │ 128)            │           │                │       │ ## ├───────────────────┼─────────────────┼───────────┼────────────────┼───────┤ ## │ max_pooling2d_1   │ (None, 20, 20,  │         0 │ batch_normali… │   -   │ ## │ (MaxPooling2D)    │ 128)            │           │                │       │ ## ├───────────────────┼─────────────────┼───────────┼────────────────┼───────┤ ## │ conv2d_2 (Conv2D) │ (None, 20, 20,  │     8,320 │ add[0][0]      │   Y   │ ## │                   │ 128)            │           │                │       │ ## ├───────────────────┼─────────────────┼───────────┼────────────────┼───────┤ ## │ add_1 (Add)       │ (None, 20, 20,  │         0 │ max_pooling2d… │   -   │ ## │                   │ 128)            │           │ conv2d_2[0][0] │       │ ## ├───────────────────┼─────────────────┼───────────┼────────────────┼───────┤ ## │ activation_5      │ (None, 20, 20,  │         0 │ add_1[0][0]    │   -   │ ## │ (Activation)      │ 128)            │           │                │       │ ## ├───────────────────┼─────────────────┼───────────┼────────────────┼───────┤ ## │ separable_conv2d… │ (None, 20, 20,  │    34,176 │ activation_5[… │   Y   │ ## │ (SeparableConv2D) │ 256)            │           │                │       │ ## ├───────────────────┼─────────────────┼───────────┼────────────────┼───────┤ ## │ batch_normalizat… │ (None, 20, 20,  │     1,024 │ separable_con… │   Y   │ ## │ (BatchNormalizat… │ 256)            │           │                │       │ ## ├───────────────────┼─────────────────┼───────────┼────────────────┼───────┤ ## │ activation_6      │ (None, 20, 20,  │         0 │ batch_normali… │   -   │ ## │ (Activation)      │ 256)            │           │                │       │ ## ├───────────────────┼─────────────────┼───────────┼────────────────┼───────┤ ## │ separable_conv2d… │ (None, 20, 20,  │    68,096 │ activation_6[… │   Y   │ ## │ (SeparableConv2D) │ 256)            │           │                │       │ ## ├───────────────────┼─────────────────┼───────────┼────────────────┼───────┤ ## │ batch_normalizat… │ (None, 20, 20,  │     1,024 │ separable_con… │   Y   │ ## │ (BatchNormalizat… │ 256)            │           │                │       │ ## ├───────────────────┼─────────────────┼───────────┼────────────────┼───────┤ ## │ max_pooling2d_2   │ (None, 10, 10,  │         0 │ batch_normali… │   -   │ ## │ (MaxPooling2D)    │ 256)            │           │                │       │ ## ├───────────────────┼─────────────────┼───────────┼────────────────┼───────┤ ## │ conv2d_3 (Conv2D) │ (None, 10, 10,  │    33,024 │ add_1[0][0]    │   Y   │ ## │                   │ 256)            │           │                │       │ ## ├───────────────────┼─────────────────┼───────────┼────────────────┼───────┤ ## │ add_2 (Add)       │ (None, 10, 10,  │         0 │ max_pooling2d… │   -   │ ## │                   │ 256)            │           │ conv2d_3[0][0] │       │ ## ├───────────────────┼─────────────────┼───────────┼────────────────┼───────┤ ## │ activation_7      │ (None, 10, 10,  │         0 │ add_2[0][0]    │   -   │ ## │ (Activation)      │ 256)            │           │                │       │ ## ├───────────────────┼─────────────────┼───────────┼────────────────┼───────┤ ## │ conv2d_transpose  │ (None, 10, 10,  │   590,080 │ activation_7[… │   Y   │ ## │ (Conv2DTranspose) │ 256)            │           │                │       │ ## ├───────────────────┼─────────────────┼───────────┼────────────────┼───────┤ ## │ batch_normalizat… │ (None, 10, 10,  │     1,024 │ conv2d_transp… │   Y   │ ## │ (BatchNormalizat… │ 256)            │           │                │       │ ## ├───────────────────┼─────────────────┼───────────┼────────────────┼───────┤ ## │ activation_8      │ (None, 10, 10,  │         0 │ batch_normali… │   -   │ ## │ (Activation)      │ 256)            │           │                │       │ ## ├───────────────────┼─────────────────┼───────────┼────────────────┼───────┤ ## │ conv2d_transpose… │ (None, 10, 10,  │   590,080 │ activation_8[… │   Y   │ ## │ (Conv2DTranspose) │ 256)            │           │                │       │ ## ├───────────────────┼─────────────────┼───────────┼────────────────┼───────┤ ## │ batch_normalizat… │ (None, 10, 10,  │     1,024 │ conv2d_transp… │   Y   │ ## │ (BatchNormalizat… │ 256)            │           │                │       │ ## ├───────────────────┼─────────────────┼───────────┼────────────────┼───────┤ ## │ up_sampling2d_1   │ (None, 20, 20,  │         0 │ add_2[0][0]    │   -   │ ## │ (UpSampling2D)    │ 256)            │           │                │       │ ## ├───────────────────┼─────────────────┼───────────┼────────────────┼───────┤ ## │ up_sampling2d     │ (None, 20, 20,  │         0 │ batch_normali… │   -   │ ## │ (UpSampling2D)    │ 256)            │           │                │       │ ## ├───────────────────┼─────────────────┼───────────┼────────────────┼───────┤ ## │ conv2d_4 (Conv2D) │ (None, 20, 20,  │    65,792 │ up_sampling2d… │   Y   │ ## │                   │ 256)            │           │                │       │ ## ├───────────────────┼─────────────────┼───────────┼────────────────┼───────┤ ## │ add_3 (Add)       │ (None, 20, 20,  │         0 │ up_sampling2d… │   -   │ ## │                   │ 256)            │           │ conv2d_4[0][0] │       │ ## ├───────────────────┼─────────────────┼───────────┼────────────────┼───────┤ ## │ activation_9      │ (None, 20, 20,  │         0 │ add_3[0][0]    │   -   │ ## │ (Activation)      │ 256)            │           │                │       │ ## ├───────────────────┼─────────────────┼───────────┼────────────────┼───────┤ ## │ conv2d_transpose… │ (None, 20, 20,  │   295,040 │ activation_9[… │   Y   │ ## │ (Conv2DTranspose) │ 128)            │           │                │       │ ## ├───────────────────┼─────────────────┼───────────┼────────────────┼───────┤ ## │ batch_normalizat… │ (None, 20, 20,  │       512 │ conv2d_transp… │   Y   │ ## │ (BatchNormalizat… │ 128)            │           │                │       │ ## ├───────────────────┼─────────────────┼───────────┼────────────────┼───────┤ ## │ activation_10     │ (None, 20, 20,  │         0 │ batch_normali… │   -   │ ## │ (Activation)      │ 128)            │           │                │       │ ## ├───────────────────┼─────────────────┼───────────┼────────────────┼───────┤ ## │ conv2d_transpose… │ (None, 20, 20,  │   147,584 │ activation_10… │   Y   │ ## │ (Conv2DTranspose) │ 128)            │           │                │       │ ## ├───────────────────┼─────────────────┼───────────┼────────────────┼───────┤ ## │ batch_normalizat… │ (None, 20, 20,  │       512 │ conv2d_transp… │   Y   │ ## │ (BatchNormalizat… │ 128)            │           │                │       │ ## ├───────────────────┼─────────────────┼───────────┼────────────────┼───────┤ ## │ up_sampling2d_3   │ (None, 40, 40,  │         0 │ add_3[0][0]    │   -   │ ## │ (UpSampling2D)    │ 256)            │           │                │       │ ## ├───────────────────┼─────────────────┼───────────┼────────────────┼───────┤ ## │ up_sampling2d_2   │ (None, 40, 40,  │         0 │ batch_normali… │   -   │ ## │ (UpSampling2D)    │ 128)            │           │                │       │ ## ├───────────────────┼─────────────────┼───────────┼────────────────┼───────┤ ## │ conv2d_5 (Conv2D) │ (None, 40, 40,  │    32,896 │ up_sampling2d… │   Y   │ ## │                   │ 128)            │           │                │       │ ## ├───────────────────┼─────────────────┼───────────┼────────────────┼───────┤ ## │ add_4 (Add)       │ (None, 40, 40,  │         0 │ up_sampling2d… │   -   │ ## │                   │ 128)            │           │ conv2d_5[0][0] │       │ ## ├───────────────────┼─────────────────┼───────────┼────────────────┼───────┤ ## │ activation_11     │ (None, 40, 40,  │         0 │ add_4[0][0]    │   -   │ ## │ (Activation)      │ 128)            │           │                │       │ ## ├───────────────────┼─────────────────┼───────────┼────────────────┼───────┤ ## │ conv2d_transpose… │ (None, 40, 40,  │    73,792 │ activation_11… │   Y   │ ## │ (Conv2DTranspose) │ 64)             │           │                │       │ ## ├───────────────────┼─────────────────┼───────────┼────────────────┼───────┤ ## │ batch_normalizat… │ (None, 40, 40,  │       256 │ conv2d_transp… │   Y   │ ## │ (BatchNormalizat… │ 64)             │           │                │       │ ## ├───────────────────┼─────────────────┼───────────┼────────────────┼───────┤ ## │ activation_12     │ (None, 40, 40,  │         0 │ batch_normali… │   -   │ ## │ (Activation)      │ 64)             │           │                │       │ ## ├───────────────────┼─────────────────┼───────────┼────────────────┼───────┤ ## │ conv2d_transpose… │ (None, 40, 40,  │    36,928 │ activation_12… │   Y   │ ## │ (Conv2DTranspose) │ 64)             │           │                │       │ ## ├───────────────────┼─────────────────┼───────────┼────────────────┼───────┤ ## │ batch_normalizat… │ (None, 40, 40,  │       256 │ conv2d_transp… │   Y   │ ## │ (BatchNormalizat… │ 64)             │           │                │       │ ## ├───────────────────┼─────────────────┼───────────┼────────────────┼───────┤ ## │ up_sampling2d_5   │ (None, 80, 80,  │         0 │ add_4[0][0]    │   -   │ ## │ (UpSampling2D)    │ 128)            │           │                │       │ ## ├───────────────────┼─────────────────┼───────────┼────────────────┼───────┤ ## │ up_sampling2d_4   │ (None, 80, 80,  │         0 │ batch_normali… │   -   │ ## │ (UpSampling2D)    │ 64)             │           │                │       │ ## ├───────────────────┼─────────────────┼───────────┼────────────────┼───────┤ ## │ conv2d_6 (Conv2D) │ (None, 80, 80,  │     8,256 │ up_sampling2d… │   Y   │ ## │                   │ 64)             │           │                │       │ ## ├───────────────────┼─────────────────┼───────────┼────────────────┼───────┤ ## │ add_5 (Add)       │ (None, 80, 80,  │         0 │ up_sampling2d… │   -   │ ## │                   │ 64)             │           │ conv2d_6[0][0] │       │ ## ├───────────────────┼─────────────────┼───────────┼────────────────┼───────┤ ## │ activation_13     │ (None, 80, 80,  │         0 │ add_5[0][0]    │   -   │ ## │ (Activation)      │ 64)             │           │                │       │ ## ├───────────────────┼─────────────────┼───────────┼────────────────┼───────┤ ## │ conv2d_transpose… │ (None, 80, 80,  │    18,464 │ activation_13… │   Y   │ ## │ (Conv2DTranspose) │ 32)             │           │                │       │ ## ├───────────────────┼─────────────────┼───────────┼────────────────┼───────┤ ## │ batch_normalizat… │ (None, 80, 80,  │       128 │ conv2d_transp… │   Y   │ ## │ (BatchNormalizat… │ 32)             │           │                │       │ ## ├───────────────────┼─────────────────┼───────────┼────────────────┼───────┤ ## │ activation_14     │ (None, 80, 80,  │         0 │ batch_normali… │   -   │ ## │ (Activation)      │ 32)             │           │                │       │ ## ├───────────────────┼─────────────────┼───────────┼────────────────┼───────┤ ## │ conv2d_transpose… │ (None, 80, 80,  │     9,248 │ activation_14… │   Y   │ ## │ (Conv2DTranspose) │ 32)             │           │                │       │ ## ├───────────────────┼─────────────────┼───────────┼────────────────┼───────┤ ## │ batch_normalizat… │ (None, 80, 80,  │       128 │ conv2d_transp… │   Y   │ ## │ (BatchNormalizat… │ 32)             │           │                │       │ ## ├───────────────────┼─────────────────┼───────────┼────────────────┼───────┤ ## │ up_sampling2d_7   │ (None, 160,     │         0 │ add_5[0][0]    │   -   │ ## │ (UpSampling2D)    │ 160, 64)        │           │                │       │ ## ├───────────────────┼─────────────────┼───────────┼────────────────┼───────┤ ## │ up_sampling2d_6   │ (None, 160,     │         0 │ batch_normali… │   -   │ ## │ (UpSampling2D)    │ 160, 32)        │           │                │       │ ## ├───────────────────┼─────────────────┼───────────┼────────────────┼───────┤ ## │ conv2d_7 (Conv2D) │ (None, 160,     │     2,080 │ up_sampling2d… │   Y   │ ## │                   │ 160, 32)        │           │                │       │ ## ├───────────────────┼─────────────────┼───────────┼────────────────┼───────┤ ## │ add_6 (Add)       │ (None, 160,     │         0 │ up_sampling2d… │   -   │ ## │                   │ 160, 32)        │           │ conv2d_7[0][0] │       │ ## ├───────────────────┼─────────────────┼───────────┼────────────────┼───────┤ ## │ conv2d_8 (Conv2D) │ (None, 160,     │       867 │ add_6[0][0]    │   Y   │ ## │                   │ 160, 3)         │           │                │       │ ## └───────────────────┴─────────────────┴───────────┴────────────────┴───────┘ ##  Total params: 2,058,979 (7.85 MB) ##  Trainable params: 2,055,203 (7.84 MB) ##  Non-trainable params: 3,776 (14.75 KB)"},{"path":"https://keras3.posit.co/dev/articles/examples/vision/oxford_pets_image_segmentation.html","id":"set-aside-a-validation-split","dir":"Articles > Examples > Vision","previous_headings":"","what":"Set aside a validation split","title":"Image segmentation with a U-Net-like architecture","text":"","code":"# Split our img paths into a training and a validation set val_samples <- 1000 val_samples <- sample.int(length(input_img_paths), val_samples)  train_input_img_paths <- input_img_paths[-val_samples] train_target_img_paths <- target_img_paths[-val_samples]  val_input_img_paths <- input_img_paths[val_samples] val_target_img_paths <- target_img_paths[val_samples]  # Instantiate dataset for each split # Limit input files in `max_dataset_len` for faster epoch training time. # Remove the `max_dataset_len` arg when running with full dataset. train_dataset <- get_dataset(   batch_size,   img_size,   train_input_img_paths,   train_target_img_paths,   max_dataset_len = 1000 ) valid_dataset <- get_dataset(   batch_size, img_size, val_input_img_paths, val_target_img_paths )"},{"path":"https://keras3.posit.co/dev/articles/examples/vision/oxford_pets_image_segmentation.html","id":"train-the-model","dir":"Articles > Examples > Vision","previous_headings":"","what":"Train the model","title":"Image segmentation with a U-Net-like architecture","text":"","code":"# Configure the model for training. # We use the \"sparse\" version of categorical_crossentropy # because our target data is integers. model |> compile(   optimizer = optimizer_adam(1e-4),   loss = \"sparse_categorical_crossentropy\" )  callbacks <- list(   callback_model_checkpoint(     \"models/oxford_segmentation.keras\", save_best_only = TRUE   ) )  # Train the model, doing validation at the end of each epoch. epochs <- 50 model |> fit(     train_dataset,     epochs=epochs,     validation_data=valid_dataset,     callbacks=callbacks,     verbose=2 ) ## Epoch 1/50 ## 32/32 - 49s - 2s/step - loss: 1.4283 - val_loss: 1.5505 ## Epoch 2/50 ## 32/32 - 2s - 73ms/step - loss: 0.9223 - val_loss: 1.9896 ## Epoch 3/50 ## 32/32 - 2s - 72ms/step - loss: 0.7764 - val_loss: 2.5134 ## Epoch 4/50 ## 32/32 - 2s - 73ms/step - loss: 0.7200 - val_loss: 3.0166 ## Epoch 5/50 ## 32/32 - 2s - 74ms/step - loss: 0.6848 - val_loss: 3.2910 ## Epoch 6/50 ## 32/32 - 2s - 71ms/step - loss: 0.6557 - val_loss: 3.4529 ## Epoch 7/50 ## 32/32 - 2s - 74ms/step - loss: 0.6303 - val_loss: 3.5613 ## Epoch 8/50 ## 32/32 - 2s - 71ms/step - loss: 0.6083 - val_loss: 3.6520 ## Epoch 9/50 ## 32/32 - 2s - 68ms/step - loss: 0.5894 - val_loss: 3.7289 ## Epoch 10/50 ## 32/32 - 2s - 76ms/step - loss: 0.5726 - val_loss: 3.7945 ## Epoch 11/50 ## 32/32 - 2s - 75ms/step - loss: 0.5566 - val_loss: 3.8264 ## Epoch 12/50 ## 32/32 - 2s - 73ms/step - loss: 0.5406 - val_loss: 3.8009 ## Epoch 13/50 ## 32/32 - 2s - 73ms/step - loss: 0.5240 - val_loss: 3.7206 ## Epoch 14/50 ## 32/32 - 2s - 76ms/step - loss: 0.5060 - val_loss: 3.5849 ## Epoch 15/50 ## 32/32 - 2s - 76ms/step - loss: 0.4858 - val_loss: 3.4372 ## Epoch 16/50 ## 32/32 - 2s - 73ms/step - loss: 0.4636 - val_loss: 3.2371 ## Epoch 17/50 ## 32/32 - 2s - 78ms/step - loss: 0.4395 - val_loss: 3.0008 ## Epoch 18/50 ## 32/32 - 2s - 75ms/step - loss: 0.4140 - val_loss: 2.6982 ## Epoch 19/50 ## 32/32 - 2s - 72ms/step - loss: 0.3878 - val_loss: 2.3533 ## Epoch 20/50 ## 32/32 - 2s - 74ms/step - loss: 0.3623 - val_loss: 1.9573 ## Epoch 21/50 ## 32/32 - 2s - 73ms/step - loss: 0.3391 - val_loss: 1.6047 ## Epoch 22/50 ## 32/32 - 3s - 83ms/step - loss: 0.3202 - val_loss: 1.3140 ## Epoch 23/50 ## 32/32 - 2s - 77ms/step - loss: 0.3083 - val_loss: 1.0783 ## Epoch 24/50 ## 32/32 - 3s - 80ms/step - loss: 0.3097 - val_loss: 1.0180 ## Epoch 25/50 ## 32/32 - 2s - 77ms/step - loss: 0.3446 - val_loss: 0.9350 ## Epoch 26/50 ## 32/32 - 2s - 70ms/step - loss: 0.3603 - val_loss: 0.9866 ## Epoch 27/50 ## 32/32 - 3s - 82ms/step - loss: 0.3268 - val_loss: 0.8704 ## Epoch 28/50 ## 32/32 - 2s - 72ms/step - loss: 0.2880 - val_loss: 1.0167 ## Epoch 29/50 ## 32/32 - 2s - 75ms/step - loss: 0.2735 - val_loss: 1.1785 ## Epoch 30/50 ## 32/32 - 2s - 75ms/step - loss: 0.2686 - val_loss: 1.1817 ## Epoch 31/50 ## 32/32 - 2s - 74ms/step - loss: 0.2718 - val_loss: 1.1267 ## Epoch 32/50 ## 32/32 - 2s - 72ms/step - loss: 0.2867 - val_loss: 1.1271 ## Epoch 33/50 ## 32/32 - 2s - 75ms/step - loss: 0.3123 - val_loss: 1.2541 ## Epoch 34/50 ## 32/32 - 2s - 75ms/step - loss: 0.2997 - val_loss: 1.2340 ## Epoch 35/50 ## 32/32 - 2s - 74ms/step - loss: 0.2851 - val_loss: 1.0755 ## Epoch 36/50 ## 32/32 - 2s - 73ms/step - loss: 0.2781 - val_loss: 1.0022 ## Epoch 37/50 ## 32/32 - 2s - 75ms/step - loss: 0.2755 - val_loss: 1.5445 ## Epoch 38/50 ## 32/32 - 2s - 76ms/step - loss: 0.2629 - val_loss: 1.1026 ## Epoch 39/50 ## 32/32 - 2s - 75ms/step - loss: 0.2428 - val_loss: 1.2205 ## Epoch 40/50 ## 32/32 - 2s - 74ms/step - loss: 0.2409 - val_loss: 1.2005 ## Epoch 41/50 ## 32/32 - 2s - 74ms/step - loss: 0.2453 - val_loss: 1.2381 ## Epoch 42/50 ## 32/32 - 2s - 73ms/step - loss: 0.2363 - val_loss: 1.5562 ## Epoch 43/50 ## 32/32 - 2s - 75ms/step - loss: 0.2337 - val_loss: 1.3185 ## Epoch 44/50 ## 32/32 - 2s - 77ms/step - loss: 0.2222 - val_loss: 1.1867 ## Epoch 45/50 ## 32/32 - 2s - 77ms/step - loss: 0.2151 - val_loss: 1.0701 ## Epoch 46/50 ## 32/32 - 2s - 75ms/step - loss: 0.2099 - val_loss: 1.1109 ## Epoch 47/50 ## 32/32 - 2s - 74ms/step - loss: 0.2062 - val_loss: 1.0900 ## Epoch 48/50 ## 32/32 - 2s - 74ms/step - loss: 0.2038 - val_loss: 1.1194 ## Epoch 49/50 ## 32/32 - 2s - 71ms/step - loss: 0.1964 - val_loss: 1.0576 ## Epoch 50/50 ## 32/32 - 2s - 73ms/step - loss: 0.1871 - val_loss: 1.0851"},{"path":"https://keras3.posit.co/dev/articles/examples/vision/oxford_pets_image_segmentation.html","id":"visualize-predictions","dir":"Articles > Examples > Vision","previous_headings":"","what":"Visualize predictions","title":"Image segmentation with a U-Net-like architecture","text":"","code":"model <- load_model(\"models/oxford_segmentation.keras\") # Generate predictions for all images in the validation set val_dataset <- get_dataset(   batch_size, img_size, val_input_img_paths, val_target_img_paths ) val_preds <- predict(model, val_dataset) ## 32/32 - 4s - 117ms/step display_mask <- function(i) {   # Quick utility to display a model's prediction.   mask <- val_preds[i,,,] %>%     apply(c(1,2), which.max) %>%     array_reshape(dim = c(img_size, 1))   mask <- abind::abind(mask, mask, mask, along = 3)   plot(as.raster(mask, max = 3)) }  # Display results for validation image #10 i <- 10  par(mfrow = c(1, 3)) # Display input image input_img_paths[i] |>   jpeg::readJPEG() |>   as.raster() |>   plot()  # Display ground-truth target mask target_img_paths[i] |>   png::readPNG() |>   magrittr::multiply_by(255)|>   as.raster(max = 3) |>   plot()  # Display mask predicted by our model display_mask(i)  # Note that the model only sees inputs at 150x150."},{"path":"https://keras3.posit.co/dev/articles/functional_api.html","id":"setup","dir":"Articles","previous_headings":"","what":"Setup","title":"The Functional API","text":"","code":"library(keras3)"},{"path":"https://keras3.posit.co/dev/articles/functional_api.html","id":"introduction","dir":"Articles","previous_headings":"","what":"Introduction","title":"The Functional API","text":"Keras functional API way create models flexible sequential API. functional API can handle models non-linear topology, shared layers, even multiple inputs outputs. main idea deep learning model usually directed acyclic graph (DAG) layers. functional API way build graphs layers. Consider following model: basic graph three layers. build model using functional API, start creating input node: shape data set 784-dimensional vector. batch size always omitted since shape sample specified. , example, image input shape (32, 32, 3), use: inputs returned contains information shape dtype input data feed model. ’s shape: ’s dtype: create new node graph layers calling layer inputs object: “layer call” action like drawing arrow “inputs” layer created. ’re “passing” inputs dense layer, get x output. Let’s add layers graph layers: point, can create Model specifying inputs outputs graph layers: Let’s check model summary looks like: can also plot model graph:  , optionally, display input output shapes layer plotted graph:  figure code almost identical. code version, connection arrows replaced call operation. “graph layers” intuitive mental image deep learning model, functional API way create models closely mirrors .","code":"(input: 784-dimensional vectors)        ↧ [Dense (64 units, relu activation)]        ↧ [Dense (64 units, relu activation)]        ↧ [Dense (10 units, softmax activation)]        ↧ (output: logits of a probability distribution over 10 classes) inputs <- keras_input(shape = c(784)) # Just for demonstration purposes. img_inputs <- keras_input(shape = c(32, 32, 3)) shape(inputs) ## shape(NA, 784) inputs$dtype ## [1] \"float32\" dense <- layer_dense(units = 64, activation=\"relu\") x <- dense(inputs) outputs <- x |>   layer_dense(units = 64, activation = \"relu\") |>   layer_dense(units = 10) model <- keras_model(inputs = inputs, outputs = outputs, name = \"mnist_model\") summary(model) ## Model: \"mnist_model\" ## ┏━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━┓ ## ┃ Layer (type)                    ┃ Output Shape           ┃       Param # ┃ ## ┡━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━┩ ## │ input_layer (InputLayer)        │ (None, 784)            │             0 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ dense (Dense)                   │ (None, 64)             │        50,240 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ dense_1 (Dense)                 │ (None, 64)             │         4,160 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ dense_2 (Dense)                 │ (None, 10)             │           650 │ ## └─────────────────────────────────┴────────────────────────┴───────────────┘ ##  Total params: 55,050 (215.04 KB) ##  Trainable params: 55,050 (215.04 KB) ##  Non-trainable params: 0 (0.00 B) plot(model) plot(model, show_shapes = TRUE)"},{"path":"https://keras3.posit.co/dev/articles/functional_api.html","id":"training-evaluation-and-inference","dir":"Articles","previous_headings":"","what":"Training, evaluation, and inference","title":"The Functional API","text":"Training, evaluation, inference work exactly way models built using functional API Sequential models. Model class offers built-training loop (fit() method) built-evaluation loop (evaluate() method). Note can easily customize loops implement training routines beyond supervised learning (e.g. GANs). , load MNIST image data, reshape vectors, fit model data (monitoring performance validation split), evaluate model test data: reading, see training evaluation guide.","code":"c(c(x_train, y_train), c(x_test, y_test)) %<-% dataset_mnist()  x_train <- array_reshape(x_train, c(60000, 784)) / 255 x_test <- array_reshape(x_test, c(10000, 784)) / 255  model |> compile(   loss = loss_sparse_categorical_crossentropy(from_logits = TRUE),   optimizer = optimizer_rmsprop(),   metrics = \"accuracy\" )  history <- model |> fit(     x_train, y_train, batch_size = 64, epochs = 2, validation_split = 0.2 ) ## Epoch 1/2 ## 750/750 - 2s - 3ms/step - accuracy: 0.8980 - loss: 0.3540 - val_accuracy: 0.9444 - val_loss: 0.1898 ## Epoch 2/2 ## 750/750 - 1s - 1ms/step - accuracy: 0.9512 - loss: 0.1633 - val_accuracy: 0.9605 - val_loss: 0.1387 test_scores <- model |> evaluate(x_test, y_test, verbose=2) ## 313/313 - 1s - 2ms/step - accuracy: 0.9598 - loss: 0.1322 cat(\"Test loss:\", test_scores$loss, \"\\n\") cat(\"Test accuracy:\", test_scores$accuracy, \"\\n\") ## Test loss: 0.1321879 ## Test accuracy: 0.9598"},{"path":"https://keras3.posit.co/dev/articles/functional_api.html","id":"save-and-serialize","dir":"Articles","previous_headings":"","what":"Save and serialize","title":"The Functional API","text":"Saving model serialization work way models built using functional API Sequential models. standard way save functional model call model.save() save entire model single file. can later recreate model file, even code built model longer available. saved file includes : - model architecture - model weight values (learned training) - model training config, (passed compile()) - optimizer state, (restart training left ) details, read model serialization & saving guide.","code":"model |> save_model(\"my_model.keras\") rm(model) # Recreate the exact same model purely from the file: model <- load_model(\"my_model.keras\")"},{"path":"https://keras3.posit.co/dev/articles/functional_api.html","id":"use-the-same-graph-of-layers-to-define-multiple-models","dir":"Articles","previous_headings":"","what":"Use the same graph of layers to define multiple models","title":"The Functional API","text":"functional API, models created specifying inputs outputs graph layers. means single graph layers can used generate multiple models. example , use stack layers instantiate two models: encoder model turns image inputs 16-dimensional vectors, end--end autoencoder model training. , decoding architecture strictly symmetrical encoding architecture, output shape input shape (28, 28, 1). reverse conv_2d layer conv_2d_transpose layer, reverse max_pooling_2d layer upsampling_2d layer.","code":"encoder_input <- keras_input(shape = c(28, 28, 1), name=\"img\") encoder_output <- encoder_input |>   layer_conv_2d(16, 3, activation = \"relu\") |>   layer_conv_2d(32, 3, activation = \"relu\") |>   layer_max_pooling_2d(3) |>   layer_conv_2d(32, 3, activation = \"relu\") |>   layer_conv_2d(16, 3, activation = \"relu\") |>   layer_global_max_pooling_2d()  encoder <- keras_model(encoder_input, encoder_output, name=\"encoder\") summary(encoder) ## Model: \"encoder\" ## ┏━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━┓ ## ┃ Layer (type)                    ┃ Output Shape           ┃       Param # ┃ ## ┡━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━┩ ## │ img (InputLayer)                │ (None, 28, 28, 1)      │             0 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ conv2d (Conv2D)                 │ (None, 26, 26, 16)     │           160 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ conv2d_1 (Conv2D)               │ (None, 24, 24, 32)     │         4,640 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ max_pooling2d (MaxPooling2D)    │ (None, 8, 8, 32)       │             0 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ conv2d_2 (Conv2D)               │ (None, 6, 6, 32)       │         9,248 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ conv2d_3 (Conv2D)               │ (None, 4, 4, 16)       │         4,624 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ global_max_pooling2d            │ (None, 16)             │             0 │ ## │ (GlobalMaxPooling2D)            │                        │               │ ## └─────────────────────────────────┴────────────────────────┴───────────────┘ ##  Total params: 18,672 (72.94 KB) ##  Trainable params: 18,672 (72.94 KB) ##  Non-trainable params: 0 (0.00 B) decoder_output <- encoder_output |>   layer_reshape(c(4, 4, 1)) |>   layer_conv_2d_transpose(16, 3, activation = \"relu\") |>   layer_conv_2d_transpose(32, 3, activation = \"relu\") |>   layer_upsampling_2d(3) |>   layer_conv_2d_transpose(16, 3, activation = \"relu\") |>   layer_conv_2d_transpose(1, 3, activation = \"relu\")  autoencoder <- keras_model(encoder_input, decoder_output, name=\"autoencoder\") summary(autoencoder) ## Model: \"autoencoder\" ## ┏━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━┓ ## ┃ Layer (type)                    ┃ Output Shape           ┃       Param # ┃ ## ┡━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━┩ ## │ img (InputLayer)                │ (None, 28, 28, 1)      │             0 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ conv2d (Conv2D)                 │ (None, 26, 26, 16)     │           160 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ conv2d_1 (Conv2D)               │ (None, 24, 24, 32)     │         4,640 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ max_pooling2d (MaxPooling2D)    │ (None, 8, 8, 32)       │             0 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ conv2d_2 (Conv2D)               │ (None, 6, 6, 32)       │         9,248 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ conv2d_3 (Conv2D)               │ (None, 4, 4, 16)       │         4,624 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ global_max_pooling2d            │ (None, 16)             │             0 │ ## │ (GlobalMaxPooling2D)            │                        │               │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ reshape (Reshape)               │ (None, 4, 4, 1)        │             0 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ conv2d_transpose                │ (None, 6, 6, 16)       │           160 │ ## │ (Conv2DTranspose)               │                        │               │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ conv2d_transpose_1              │ (None, 8, 8, 32)       │         4,640 │ ## │ (Conv2DTranspose)               │                        │               │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ up_sampling2d (UpSampling2D)    │ (None, 24, 24, 32)     │             0 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ conv2d_transpose_2              │ (None, 26, 26, 16)     │         4,624 │ ## │ (Conv2DTranspose)               │                        │               │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ conv2d_transpose_3              │ (None, 28, 28, 1)      │           145 │ ## │ (Conv2DTranspose)               │                        │               │ ## └─────────────────────────────────┴────────────────────────┴───────────────┘ ##  Total params: 28,241 (110.32 KB) ##  Trainable params: 28,241 (110.32 KB) ##  Non-trainable params: 0 (0.00 B)"},{"path":"https://keras3.posit.co/dev/articles/functional_api.html","id":"all-models-are-callable-just-like-layers","dir":"Articles","previous_headings":"","what":"All models are callable, just like layers","title":"The Functional API","text":"can treat model layer invoking Input output another layer. calling model aren’t just reusing architecture model, ’re also reusing weights. see action, ’s different take autoencoder example creates encoder model, decoder model, chains two calls obtain autoencoder model: can see, model can nested: model can contain sub-models (since model just like layer). common use case model nesting ensembling. example, ’s ensemble set models single model averages predictions:","code":"encoder_input <- keras_input(shape = c(28, 28, 1), name=\"img\") encoder_output <- encoder_input |>   layer_conv_2d(16, 3, activation = \"relu\") |>   layer_conv_2d(32, 3, activation = \"relu\") |>   layer_max_pooling_2d(3) |>   layer_conv_2d(32, 3, activation = \"relu\") |>   layer_conv_2d(16, 3, activation = \"relu\") |>   layer_global_max_pooling_2d()  encoder <- keras_model(encoder_input, encoder_output, name=\"encoder\") summary(encoder) ## Model: \"encoder\" ## ┏━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━┓ ## ┃ Layer (type)                    ┃ Output Shape           ┃       Param # ┃ ## ┡━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━┩ ## │ img (InputLayer)                │ (None, 28, 28, 1)      │             0 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ conv2d_4 (Conv2D)               │ (None, 26, 26, 16)     │           160 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ conv2d_5 (Conv2D)               │ (None, 24, 24, 32)     │         4,640 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ max_pooling2d_1 (MaxPooling2D)  │ (None, 8, 8, 32)       │             0 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ conv2d_6 (Conv2D)               │ (None, 6, 6, 32)       │         9,248 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ conv2d_7 (Conv2D)               │ (None, 4, 4, 16)       │         4,624 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ global_max_pooling2d_1          │ (None, 16)             │             0 │ ## │ (GlobalMaxPooling2D)            │                        │               │ ## └─────────────────────────────────┴────────────────────────┴───────────────┘ ##  Total params: 18,672 (72.94 KB) ##  Trainable params: 18,672 (72.94 KB) ##  Non-trainable params: 0 (0.00 B) decoder_input <- keras_input(shape = c(16), name = \"encoded_img\") decoder_output <- decoder_input |>   layer_reshape(c(4, 4, 1)) |>   layer_conv_2d_transpose(16, 3, activation = \"relu\") |>   layer_conv_2d_transpose(32, 3, activation = \"relu\") |>   layer_upsampling_2d(3) |>   layer_conv_2d_transpose(16, 3, activation = \"relu\") |>   layer_conv_2d_transpose(1, 3, activation = \"relu\")  decoder <- keras_model(decoder_input, decoder_output, name = \"decoder\") summary(decoder) ## Model: \"decoder\" ## ┏━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━┓ ## ┃ Layer (type)                    ┃ Output Shape           ┃       Param # ┃ ## ┡━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━┩ ## │ encoded_img (InputLayer)        │ (None, 16)             │             0 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ reshape_1 (Reshape)             │ (None, 4, 4, 1)        │             0 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ conv2d_transpose_4              │ (None, 6, 6, 16)       │           160 │ ## │ (Conv2DTranspose)               │                        │               │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ conv2d_transpose_5              │ (None, 8, 8, 32)       │         4,640 │ ## │ (Conv2DTranspose)               │                        │               │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ up_sampling2d_1 (UpSampling2D)  │ (None, 24, 24, 32)     │             0 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ conv2d_transpose_6              │ (None, 26, 26, 16)     │         4,624 │ ## │ (Conv2DTranspose)               │                        │               │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ conv2d_transpose_7              │ (None, 28, 28, 1)      │           145 │ ## │ (Conv2DTranspose)               │                        │               │ ## └─────────────────────────────────┴────────────────────────┴───────────────┘ ##  Total params: 9,569 (37.38 KB) ##  Trainable params: 9,569 (37.38 KB) ##  Non-trainable params: 0 (0.00 B) autoencoder_input <- keras_input(shape = c(28, 28, 1), name = \"img\") encoded_img <- encoder(autoencoder_input) decoded_img <- decoder(encoded_img) autoencoder <- keras_model(autoencoder_input, decoded_img,                            name = \"autoencoder\") summary(autoencoder) ## Model: \"autoencoder\" ## ┏━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━┓ ## ┃ Layer (type)                    ┃ Output Shape           ┃       Param # ┃ ## ┡━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━┩ ## │ img (InputLayer)                │ (None, 28, 28, 1)      │             0 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ encoder (Functional)            │ (None, 16)             │        18,672 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ decoder (Functional)            │ (None, 28, 28, 1)      │         9,569 │ ## └─────────────────────────────────┴────────────────────────┴───────────────┘ ##  Total params: 28,241 (110.32 KB) ##  Trainable params: 28,241 (110.32 KB) ##  Non-trainable params: 0 (0.00 B) get_model <- function() {   inputs <- keras_input(shape = 128)   outputs <- inputs |> layer_dense(1)   keras_model(inputs, outputs) }  model1 <- get_model() model2 <- get_model() model3 <- get_model()  inputs <- keras_input(shape = 128) y1 <- model1(inputs) y2 <- model2(inputs) y3 <- model3(inputs) outputs <- layer_average(list(y1, y2, y3)) ensemble_model <- keras_model(inputs = inputs, outputs = outputs)"},{"path":[]},{"path":"https://keras3.posit.co/dev/articles/functional_api.html","id":"models-with-multiple-inputs-and-outputs","dir":"Articles","previous_headings":"Manipulate complex graph topologies","what":"Models with multiple inputs and outputs","title":"The Functional API","text":"functional API makes easy manipulate multiple inputs outputs. handled Sequential API. example, ’re building system ranking customer issue tickets priority routing correct department, model three inputs: title ticket (text input), text body ticket (text input), tags added user (categorical input) model two outputs: priority score 0 1 (scalar sigmoid output), department handle ticket (softmax output set departments). can build model lines functional API: Now plot model:  compiling model, can assign different losses output. can even assign different weights loss – modulate contribution total training loss. Since output layers different names, also specify losses loss weights corresponding layer names: Train model passing lists NumPy arrays inputs targets: calling fit Dataset object, yield either list lists like list(list(title_data, body_data, tags_data), list(priority_targets, dept_targets)) list named lists like list(list(title = title_data, body = body_data, tags = tags_data), list(priority = priority_targets, department = dept_targets)). detailed explanation, refer training evaluation guide.","code":"num_tags <- 12  # Number of unique issue tags num_words <- 10000  # Size of vocabulary obtained when preprocessing text data num_departments <- 4  # Number of departments for predictions  title_input <- # Variable-length sequence of ints   keras_input(shape(NA), name = \"title\") body_input <-  # Variable-length sequence of ints   keras_input(shape(NA), name = \"body\") tags_input <-  # Binary vectors of size `num_tags`   keras_input(shape = num_tags, name = \"tags\")  # Embed each word in the title into a 64-dimensional vector title_features <- layer_embedding(title_input, num_words, 64) # Embed each word in the text into a 64-dimensional vector body_features <- layer_embedding(body_input, num_words, 64)  # Reduce sequence of embedded words in the title # into a single 128-dimensional vector title_features <- layer_lstm(title_features, 128) # Reduce sequence of embedded words in the body # into a single 32-dimensional vector body_features <- layer_lstm(body_features, 32)  # Merge all available features into a single large vector via concatenation x <- layer_concatenate(title_features, body_features, tags_input)  # Stick a logistic regression for priority prediction on top of the features priority_pred <- layer_dense(x, 1, name = \"priority\")  # Stick a department classifier on top of the features department_pred <- layer_dense(x, num_departments, name = \"department\")  # Instantiate an end-to-end model predicting both priority and department model <- keras_model(   inputs = list(title_input, body_input, tags_input),   outputs = list(priority = priority_pred, department = department_pred) ) plot(model, show_shapes = TRUE) model |> compile(   optimizer = optimizer_rmsprop(1e-3),   loss = list(     loss_binary_crossentropy(from_logits = TRUE),     loss_categorical_crossentropy(from_logits = TRUE)   ),   loss_weights = c(1.0, 0.2) ) model |> compile(   optimizer = optimizer_rmsprop(1e-3),   loss = list(     priority = loss_binary_crossentropy(from_logits = TRUE),     department = loss_categorical_crossentropy(from_logits = TRUE)   ),   loss_weights = list(priority = 1.0, department = 0.2) ) # Dummy input data title_data <- random_integer(c(1280, 12), 0, num_words) body_data <- random_integer(c(1280, 100), 0, num_words) tags_data <- random_integer(c(1280, num_tags), 0, 2)  # Dummy target data priority_targets <- random_normal(c(1280, 1)) dept_targets <- random_integer(c(1280, num_departments), 0, 2)  model |> fit(   list(title = title_data, body = body_data, tags = tags_data),   list(priority = priority_targets, department = dept_targets),   epochs = 2,   batch_size = 32 ) ## Epoch 1/2 ## 40/40 - 3s - 64ms/step - department_loss: 2.8465 - loss: 0.7669 - priority_loss: 0.1976 ## Epoch 2/2 ## 40/40 - 0s - 5ms/step - department_loss: 2.8554 - loss: 0.7538 - priority_loss: 0.1828"},{"path":"https://keras3.posit.co/dev/articles/functional_api.html","id":"a-toy-resnet-model","dir":"Articles","previous_headings":"Manipulate complex graph topologies","what":"A toy ResNet model","title":"The Functional API","text":"addition models multiple inputs outputs, functional API makes easy manipulate non-linear connectivity topologies – models layers connected sequentially, Sequential API handle. common use case residual connections. Let’s build toy ResNet model CIFAR10 demonstrate : Plot model:  Now train model:","code":"inputs <- keras_input(shape = c(32, 32, 3), name = \"img\") block_1_output <- inputs |>   layer_conv_2d(32, kernel_size = 3, activation = \"relu\") |>   layer_conv_2d(64, kernel_size = 3, activation = \"relu\") |>   layer_max_pooling_2d(pool_size = 3)  block_2_output <- block_1_output |>   layer_conv_2d(32, kernel_size = 3, activation = \"relu\", padding = \"same\") |>   layer_conv_2d(64, kernel_size = 3, activation = \"relu\", padding = \"same\") |>   layer_add(block_1_output)  block_3_output <- block_2_output |>   layer_conv_2d(64, kernel_size = 3, activation = \"relu\", padding = \"same\") |>   layer_conv_2d(64, kernel_size = 3, activation = \"relu\", padding = \"same\") |>   layer_add(block_2_output)  outputs <- block_3_output |>   layer_conv_2d(64, 3, activation = \"relu\") |>   layer_global_average_pooling_2d() |>   layer_dense(256, activation = \"relu\") |>   layer_dropout(0.5) |>   layer_dense(10)  model <- keras_model(inputs, outputs, name = \"toy_resnet\") summary(model) ## Model: \"toy_resnet\" ## ┏━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━┓ ## ┃ Layer (type)        ┃ Output Shape      ┃    Param # ┃ Connected to      ┃ ## ┡━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━┩ ## │ img (InputLayer)    │ (None, 32, 32, 3) │          0 │ -                 │ ## ├─────────────────────┼───────────────────┼────────────┼───────────────────┤ ## │ conv2d_8 (Conv2D)   │ (None, 30, 30,    │        896 │ img[0][0]         │ ## │                     │ 32)               │            │                   │ ## ├─────────────────────┼───────────────────┼────────────┼───────────────────┤ ## │ conv2d_9 (Conv2D)   │ (None, 28, 28,    │     18,496 │ conv2d_8[0][0]    │ ## │                     │ 64)               │            │                   │ ## ├─────────────────────┼───────────────────┼────────────┼───────────────────┤ ## │ max_pooling2d_2     │ (None, 9, 9, 64)  │          0 │ conv2d_9[0][0]    │ ## │ (MaxPooling2D)      │                   │            │                   │ ## ├─────────────────────┼───────────────────┼────────────┼───────────────────┤ ## │ conv2d_10 (Conv2D)  │ (None, 9, 9, 32)  │     18,464 │ max_pooling2d_2[… │ ## ├─────────────────────┼───────────────────┼────────────┼───────────────────┤ ## │ conv2d_11 (Conv2D)  │ (None, 9, 9, 64)  │     18,496 │ conv2d_10[0][0]   │ ## ├─────────────────────┼───────────────────┼────────────┼───────────────────┤ ## │ add (Add)           │ (None, 9, 9, 64)  │          0 │ conv2d_11[0][0],  │ ## │                     │                   │            │ max_pooling2d_2[… │ ## ├─────────────────────┼───────────────────┼────────────┼───────────────────┤ ## │ conv2d_12 (Conv2D)  │ (None, 9, 9, 64)  │     36,928 │ add[0][0]         │ ## ├─────────────────────┼───────────────────┼────────────┼───────────────────┤ ## │ conv2d_13 (Conv2D)  │ (None, 9, 9, 64)  │     36,928 │ conv2d_12[0][0]   │ ## ├─────────────────────┼───────────────────┼────────────┼───────────────────┤ ## │ add_1 (Add)         │ (None, 9, 9, 64)  │          0 │ conv2d_13[0][0],  │ ## │                     │                   │            │ add[0][0]         │ ## ├─────────────────────┼───────────────────┼────────────┼───────────────────┤ ## │ conv2d_14 (Conv2D)  │ (None, 7, 7, 64)  │     36,928 │ add_1[0][0]       │ ## ├─────────────────────┼───────────────────┼────────────┼───────────────────┤ ## │ global_average_poo… │ (None, 64)        │          0 │ conv2d_14[0][0]   │ ## │ (GlobalAveragePool… │                   │            │                   │ ## ├─────────────────────┼───────────────────┼────────────┼───────────────────┤ ## │ dense_6 (Dense)     │ (None, 256)       │     16,640 │ global_average_p… │ ## ├─────────────────────┼───────────────────┼────────────┼───────────────────┤ ## │ dropout (Dropout)   │ (None, 256)       │          0 │ dense_6[0][0]     │ ## ├─────────────────────┼───────────────────┼────────────┼───────────────────┤ ## │ dense_7 (Dense)     │ (None, 10)        │      2,570 │ dropout[0][0]     │ ## └─────────────────────┴───────────────────┴────────────┴───────────────────┘ ##  Total params: 186,346 (727.91 KB) ##  Trainable params: 186,346 (727.91 KB) ##  Non-trainable params: 0 (0.00 B) plot(model, show_shapes = TRUE) c(c(x_train, y_train), c(x_test, y_test)) %<-% dataset_cifar10()  x_train <- x_train / 255.0 x_test <- x_test / 255.0  model |> compile(   optimizer = optimizer_rmsprop(1e-3),   loss = loss_sparse_categorical_crossentropy(from_logits = TRUE),   metrics = \"acc\" ) # We restrict the data to the first 1000 samples so as to limit the # guide render time. # Try to train on the entire dataset until convergence! model |> fit(   x_train[1:1000, , , ],   y_train[1:1000, ],   batch_size = 64,   epochs = 1,   validation_split = 0.2 ) ## 13/13 - 6s - 430ms/step - acc: 0.1250 - loss: 2.2998 - val_acc: 0.1250 - val_loss: 2.2939"},{"path":"https://keras3.posit.co/dev/articles/functional_api.html","id":"shared-layers","dir":"Articles","previous_headings":"","what":"Shared layers","title":"The Functional API","text":"Another good use functional API models use shared layers. Shared layers layer instances reused multiple times model – learn features correspond multiple paths graph--layers. Shared layers often used encode inputs similar spaces (say, two different pieces text feature similar vocabulary). enable sharing information across different inputs, make possible train model less data. given word seen one inputs, benefit processing inputs pass shared layer. share layer functional API, call layer instance multiple times. instance, ’s Embedding layer shared across two different text inputs:","code":"# Embedding for 1000 unique words mapped to 128-dimensional vectors shared_embedding <- layer_embedding(input_dim = 1000, output_dim = 128)  # Variable-length sequence of integers text_input_a <- keras_input(shape = shape(NA), dtype=\"int32\")  # Variable-length sequence of integers text_input_b <- keras_input(shape = shape(NA), dtype=\"int32\")  # Reuse the same layer to encode both inputs encoded_input_a <- shared_embedding(text_input_a) encoded_input_b <- shared_embedding(text_input_b)"},{"path":"https://keras3.posit.co/dev/articles/functional_api.html","id":"extract-and-reuse-nodes-in-the-graph-of-layers","dir":"Articles","previous_headings":"","what":"Extract and reuse nodes in the graph of layers","title":"The Functional API","text":"graph layers manipulating static data structure, can accessed inspected. able plot functional models images. also means can access activations intermediate layers (“nodes” graph) reuse elsewhere – useful something like feature extraction. Let’s look example. VGG19 model weights pretrained ImageNet: intermediate activations model, obtained querying graph data structure: Use features create new feature-extraction model returns values intermediate layer activations: comes handy tasks like neural style transfer, among things.","code":"vgg19 <- application_vgg19() features_list <- lapply(vgg19$layers, function(x) x$output) feat_extraction_model <- keras_model(inputs = vgg19$input,                                      outputs = features_list)  img <- random_normal(c(1, 224, 224, 3)) extracted_features <- feat_extraction_model(img)"},{"path":"https://keras3.posit.co/dev/articles/functional_api.html","id":"extend-the-api-using-custom-layers","dir":"Articles","previous_headings":"","what":"Extend the API using custom layers","title":"The Functional API","text":"keras includes wide range built-layers, example: Convolutional layers: conv_1d, conv_2d, conv_3d, conv_2d_transpose Pooling layers: max_pooling_1d, max_pooling_2d, max_pooling_3d, average_pooling_3d RNN layers: gru, lstm, conv_lstm_2d batch_normalization, dropout, embedding, etc. don’t find need, ’s easy extend API creating layers. layers subclass Layer class implement: call method, specifies computation done layer. build method, creates weights layer (just style convention since can create weights initialize, well). learn creating layers scratch, read custom layers models guide. following basic implementation layer_dense(): serialization support custom layer, define get_config() method returns constructor arguments layer instance: Optionally, implement class method from_config(cls, config) used recreating layer instance given config dictionary. default implementation from_config :","code":"custom_dense <- Layer(   classname = \"CustomDense\",   initialize = function(units = 32) {     super$initialize()     self$units <- as.integer(units)   },   build = function(input_shape) {     self$w <- self$add_weight(       shape = shape(input_shape[[2]], self$units),       initializer = \"random_normal\",       trainable = TRUE,     )     self$b <- self$add_weight(       shape = shape(self$units),             initializer=\"random_normal\",       trainable = TRUE         )   },   call = function(inputs) {     op_matmul(inputs, self$w) + self$b   } )  inputs <- keras_input(c(4)) outputs <- custom_dense(inputs, 10)  model <- keras_model(inputs, outputs) custom_dense <- Layer(   classname = \"CustomDense\",    initialize = function(units = 32, ...) {     super$initialize()     self$units <- as.integer(units)   },    build = function(input_shape) {     self$w <- self$add_weight(       shape = shape(input_shape[[2]], self$units),       initializer = \"random_normal\",       trainable = TRUE,     )     self$b <- self$add_weight(       shape = shape(self$units),             initializer=\"random_normal\",       trainable = TRUE         )   },    call = function(inputs) {     op_matmul(inputs, self$w) + self$b   },    get_config = function() {     list(units = self$units)   } )  inputs <- keras_input(c(4)) outputs <- custom_dense(inputs, 10)  model <- keras_model(inputs, outputs) config <- get_config(model)  new_model <-   from_config(config, custom_objects = list(CustomDense = custom_dense)) from_config <- function(cls, config) {   do.call(cls, config) }"},{"path":"https://keras3.posit.co/dev/articles/functional_api.html","id":"when-to-use-the-functional-api","dir":"Articles","previous_headings":"","what":"When to use the functional API","title":"The Functional API","text":"use Keras functional API create new model, just subclass Model class directly? general, functional API higher-level, easier safer, number features subclassed models support. However, model subclassing provides greater flexibility building models easily expressible directed acyclic graphs layers. example, implement Tree-RNN functional API subclass Model directly. -depth look differences functional API model subclassing, read Symbolic Imperative APIs TensorFlow 2.0?.","code":""},{"path":"https://keras3.posit.co/dev/articles/functional_api.html","id":"functional-api-strengths","dir":"Articles","previous_headings":"When to use the functional API","what":"Functional API strengths:","title":"The Functional API","text":"following properties also true Sequential models (also data structures), true subclassed models (R Python (byte)code, data structures).","code":""},{"path":"https://keras3.posit.co/dev/articles/functional_api.html","id":"less-verbose","dir":"Articles","previous_headings":"When to use the functional API > Functional API strengths:","what":"Less verbose","title":"The Functional API","text":"super$initialize(...), call = function(...), self$..., etc. Compare: subclassed version:","code":"inputs <- keras_input(shape = shape(32)) outputs <- inputs |>   layer_dense(64, activation = \"relu\") |>   layer_dense(10) mlp <- keras_model(inputs, outputs) MLP <- Model(   classname = \"MLP\",   initialize = function(...) {     super$initialize(...)     self$dense_1 <- layer_dense(units = 64, activation = \"relu\")     self$dense_2 <- layer_dense(units = 10)   },   call = function(inputs) {     inputs |>       self$dense_1() |>       self$dense_2()   } )  # Instantiate the model. mlp <- MLP() # Necessary to create the model's state. # The model doesn't have a state until it's called at least once. out <- mlp(op_zeros(c(1, 32)))"},{"path":"https://keras3.posit.co/dev/articles/functional_api.html","id":"model-validation-while-defining-its-connectivity-graph","dir":"Articles","previous_headings":"When to use the functional API > Functional API strengths:","what":"Model validation while defining its connectivity graph","title":"The Functional API","text":"functional API, input specification (shape dtype) created advance (using Input). Every time call layer, layer checks specification passed matches assumptions, raise helpful error message . guarantees model can build functional API run. debugging – convergence-related debugging – happens statically model construction execution time. similar type checking compiler.","code":""},{"path":"https://keras3.posit.co/dev/articles/functional_api.html","id":"a-functional-model-is-plottable-and-inspectable","dir":"Articles","previous_headings":"When to use the functional API > Functional API strengths:","what":"A functional model is plottable and inspectable","title":"The Functional API","text":"can plot model graph, can easily access intermediate nodes graph. example, extract reuse activations intermediate layers (seen previous example):","code":"features_list <- lapply(vgg19$layers, function(x) x$output) feat_extraction_model <- keras_model(inputs = vgg19$input,                                      outputs = features_list)"},{"path":"https://keras3.posit.co/dev/articles/functional_api.html","id":"a-functional-model-can-be-serialized-or-cloned","dir":"Articles","previous_headings":"When to use the functional API > Functional API strengths:","what":"A functional model can be serialized or cloned","title":"The Functional API","text":"functional model data structure rather piece code, safely serializable can saved single file allows recreate exact model without access original code. See serialization & saving guide. serialize subclassed model, necessary implementer specify get_config() from_config() method model level.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/articles/functional_api.html","id":"it-does-not-support-dynamic-architectures","dir":"Articles","previous_headings":"When to use the functional API > Functional API weakness:","what":"It does not support dynamic architectures","title":"The Functional API","text":"functional API treats models DAGs layers. true deep learning architectures, – example, recursive networks Tree RNNs follow assumption implemented functional API.","code":""},{"path":"https://keras3.posit.co/dev/articles/functional_api.html","id":"mix-and-match-api-styles","dir":"Articles","previous_headings":"","what":"Mix-and-match API styles","title":"The Functional API","text":"Choosing functional API Model subclassing isn’t binary decision restricts one category models. models keras API can interact , whether ’re Sequential models, functional models, subclassed models written scratch. can always use functional model Sequential model part subclassed model layer: can use subclassed layer model functional API long implements call method follows one following patterns: call(inputs, ...) – inputs tensor nested structure tensors (e.g. list tensors), ... non-tensor arguments (non-inputs). call(inputs, training = NULL, ...) – training boolean indicating whether layer behave training mode inference mode. call(inputs, mask = NULL, ...) – mask boolean mask tensor (useful RNNs, instance). call(inputs, training = NULL, mask = NULL, ...) – course, can masking training-specific behavior time. Additionally, implement get_config() method custom Layer model, functional models create still serializable cloneable. ’s quick example custom RNN, written scratch, used functional model:","code":"units <- 32 timesteps <- 10 input_dim <- 5  # Define a Functional model inputs <- keras_input(shape(NA, units)) outputs <- inputs |>   layer_global_average_pooling_1d() |>   layer_dense(units = 1)  model <- keras_model(inputs, outputs)  layer_custom_rnn <- Layer(   classname = \"CustomRNN\",   initialize = function(...) {     super$initialize(...)     self$units <- units     self$projection_1 <- layer_dense(units = units, activation = \"tanh\")     self$projection_2 <- layer_dense(units = units, activation = \"tanh\")     self$classifier <- model   },   call = function(inputs, ...) {     outputs <- list()     state <- op_zeros(c(shape(inputs)[[1]], self$units))     for (t in 1:(shape(inputs)[[2]])) {       x <- inputs[, t, ]       h <- self$projection_1(x)       y <- h + self$projection_2(state)       state <- y       outputs[[t]] <- y     }     features <- op_stack(outputs, axis = 2)     self$classifier(features)   } )  rnn <- layer_custom_rnn() out <- rnn(op_zeros(c(1, timesteps, input_dim))) units <- 32 timesteps <- 10 input_dim <- 5 batch_size <- 16  layer_custom_rnn <- Layer(   \"custom_rnn\",   initialize = function(...) {     super$initialize(...)     self$units <- units     self$projection_1 <- layer_dense(units = units, activation = \"tanh\")     self$projection_2 <- layer_dense(units = units, activation = \"tanh\")     self$classifier <- layer_dense(units = 1)   },   call = function(inputs, ...) {     outputs <- list()     state <- op_zeros(c(shape(inputs)[[1]], self$units))     for (t in 1:(shape(inputs)[[2]])) {       x <- inputs[, t, ]       h <- self$projection_1(x)       y <- h + self$projection_2(state)       state <- y       outputs[[t]] <- y     }     features <- op_stack(outputs, axis = 2)     self$classifier(features)   } )  # Note that you specify a static batch size for the inputs with the `batch_shape` # arg, because the inner computation of `layer_custom_rnn()` requires a static batch size # (when you create the `state` zeros tensor). inputs <- keras_input(batch_shape = shape(batch_size, timesteps, input_dim)) outputs <- inputs |>   layer_conv_1d(filters = 32, kernel_size = 3) |>   layer_custom_rnn()  model <- keras_model(inputs, outputs) out <- model(op_zeros(c(1, 10, 5)))"},{"path":"https://keras3.posit.co/dev/articles/getting_started.html","id":"overview","dir":"Articles","previous_headings":"","what":"Overview","title":"Getting Started with Keras","text":"Keras high-level neural networks API developed focus enabling fast experimentation. able go idea result least possible delay key good research. Keras following key features: Allows code run CPU GPU, seamlessly. User-friendly API makes easy quickly prototype deep learning models. Built-support convolutional networks (computer vision), recurrent networks (sequence processing), combination . Supports arbitrary network architectures: multi-input multi-output models, layer sharing, model sharing, etc. means Keras appropriate building essentially deep learning model, memory network neural Turing machine. website provides documentation R interface Keras. See main Keras website https://keras.io additional information project.","code":""},{"path":"https://keras3.posit.co/dev/articles/getting_started.html","id":"installation","dir":"Articles","previous_headings":"","what":"Installation","title":"Getting Started with Keras","text":"First, install keras R package: install development version : Keras R interface requires backend engine installed. TensorFlow default. provide default installation Keras TensorFlow GPU capable, GPU available. want customized installation, e.g. see documentation install_keras() installation section.","code":"install.packages(\"keras3\") remotes::install_github(\"rstudio/keras3\") keras3::install_keras(backend = \"tensorflow\")"},{"path":"https://keras3.posit.co/dev/articles/getting_started.html","id":"mnist-example","dir":"Articles","previous_headings":"","what":"MNIST Example","title":"Getting Started with Keras","text":"can learn basics Keras walking simple example: recognizing handwritten digits MNIST dataset. MNIST consists 28 x 28 grayscale images handwritten digits like :  dataset also includes labels image, telling us digit . example, labels images 5, 0, 4, 1.","code":""},{"path":"https://keras3.posit.co/dev/articles/getting_started.html","id":"preparing-the-data","dir":"Articles","previous_headings":"MNIST Example","what":"Preparing the Data","title":"Getting Started with Keras","text":"MNIST dataset included Keras can accessed using dataset_mnist() function. load dataset create variables test training data: x data 3-d array (images, width, height) grayscale values. prepare data training convert 3-d arrays matrices reshaping width height single dimension (28x28 images flattened length 784 vectors). , convert grayscale values integers ranging 0 255 floating point values ranging 0 1: Note use array_reshape() function rather dim<-() function reshape array. data re-interpreted using row-major semantics (opposed R’s default column-major semantics), turn compatible way numerical libraries called Keras interpret array dimensions. y data integer vector values ranging 0 9. prepare data training one-hot encode vectors binary class matrices using Keras to_categorical() function:","code":"library(keras3) mnist <- dataset_mnist() x_train <- mnist$train$x y_train <- mnist$train$y x_test <- mnist$test$x y_test <- mnist$test$y # reshape x_train <- array_reshape(x_train, c(nrow(x_train), 784)) x_test <- array_reshape(x_test, c(nrow(x_test), 784)) # rescale x_train <- x_train / 255 x_test <- x_test / 255 y_train <- to_categorical(y_train, 10) y_test <- to_categorical(y_test, 10)"},{"path":"https://keras3.posit.co/dev/articles/getting_started.html","id":"defining-the-model","dir":"Articles","previous_headings":"MNIST Example","what":"Defining the Model","title":"Getting Started with Keras","text":"core data structure Keras model, way organize layers. simplest type model Sequential model, linear stack layers. begin creating sequential model adding layers using pipe (|>) operator: input_shape argument first layer specifies shape input data (length 784 numeric vector representing grayscale image). final layer outputs length 10 numeric vector (probabilities digit) using softmax activation function. Use summary() function print details model: Next, compile model appropriate loss function, optimizer, metrics:","code":"model <- keras_model_sequential(input_shape = c(784)) model |>   layer_dense(units = 256, activation = 'relu') |>   layer_dropout(rate = 0.4) |>   layer_dense(units = 128, activation = 'relu') |>   layer_dropout(rate = 0.3) |>   layer_dense(units = 10, activation = 'softmax') summary(model) ## Model: \"sequential\" ## ┏━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━┓ ## ┃ Layer (type)                    ┃ Output Shape           ┃       Param # ┃ ## ┡━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━┩ ## │ dense (Dense)                   │ (None, 256)            │       200,960 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ dropout (Dropout)               │ (None, 256)            │             0 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ dense_1 (Dense)                 │ (None, 128)            │        32,896 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ dropout_1 (Dropout)             │ (None, 128)            │             0 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ dense_2 (Dense)                 │ (None, 10)             │         1,290 │ ## └─────────────────────────────────┴────────────────────────┴───────────────┘ ##  Total params: 235,146 (918.54 KB) ##  Trainable params: 235,146 (918.54 KB) ##  Non-trainable params: 0 (0.00 B) plot(model) model |> compile(   loss = 'categorical_crossentropy',   optimizer = optimizer_rmsprop(),   metrics = c('accuracy') )"},{"path":"https://keras3.posit.co/dev/articles/getting_started.html","id":"training-and-evaluation","dir":"Articles","previous_headings":"MNIST Example","what":"Training and Evaluation","title":"Getting Started with Keras","text":"Use fit() function train model 30 epochs using batches 128 images: history object returned fit() includes loss accuracy metrics can plot: Evaluate model’s performance test data: Generate predictions new data: Keras provides vocabulary building deep learning models simple, elegant, intuitive. Building question answering system, image classification model, neural Turing machine, model just straightforward.","code":"history <- model |> fit(   x_train, y_train,   epochs = 30, batch_size = 128,   validation_split = 0.2 ) plot(history) model |> evaluate(x_test, y_test) ## 313/313 - 1s - 2ms/step - accuracy: 0.9810 - loss: 0.0842 ## $accuracy ## [1] 0.981 ## ## $loss ## [1] 0.08419506 probs <- model |> predict(x_test) ## 313/313 - 0s - 1ms/step max.col(probs) - 1L ##   [1] 7 2 1 0 4 1 4 9 5 9 0 6 9 0 1 5 9 7 3 4 9 6 6 5 4 0 7 4 0 1 3 1 3 4 7 ##  [36] 2 7 1 2 1 1 7 4 2 3 5 1 2 4 4 6 3 5 5 6 0 4 1 9 5 7 8 9 3 7 4 6 4 3 0 ##  [71] 7 0 2 9 1 7 3 2 9 7 7 6 2 7 8 4 7 3 6 1 3 6 9 3 1 4 1 7 6 9 ##  [ reached 'max' / getOption(\"max.print\") -- omitted 9900 entries ]"},{"path":"https://keras3.posit.co/dev/articles/getting_started.html","id":"deep-learning-with-r-book","dir":"Articles","previous_headings":"MNIST Example","what":"Deep Learning with R Book","title":"Getting Started with Keras","text":"want comprehensive introduction Keras concepts practice deep learning, recommend Deep Learning R, 2nd Edition book Manning. book collaboration François Chollet, creator (Python) Keras, J.J. Allaire, wrote original R interface Keras, Tomasz Kalinowski, maintainer R interface Keras. book presumes significant knowledge machine learning deep learning, goes way basic theory advanced practical applications, using R interface Keras.","code":""},{"path":"https://keras3.posit.co/dev/articles/getting_started.html","id":"why-this-name-keras","dir":"Articles","previous_headings":"","what":"Why this name, Keras?","title":"Getting Started with Keras","text":"Keras (κέρας) means horn Greek. reference literary image ancient Greek Latin literature, first found Odyssey, dream spirits (Oneiroi, singular Oneiros) divided deceive men false visions, arrive Earth gate ivory, announce future come pass, arrive gate horn. ’s play words κέρας (horn) / κραίνω (fulfill), ἐλέφας (ivory) / ἐλεφαίρομαι (deceive). Keras initially developed part research effort project ONEIROS (Open-ended Neuro-Electronic Intelligent Robot Operating System). “Oneiroi beyond unravelling –can sure tale tell? men look comes pass. Two gates give passage fleeting Oneiroi; one made horn, one ivory. Oneiroi pass sawn ivory deceitful, bearing message fulfilled; come polished horn truth behind , accomplished men see .” Homer, Odyssey 19. 562 ff (Shewring translation).","code":""},{"path":"https://keras3.posit.co/dev/articles/intro_to_keras_for_engineers.html","id":"introduction","dir":"Articles","previous_headings":"","what":"Introduction","title":"Introduction to Keras for engineers","text":"Keras 3 deep learning framework works TensorFlow, JAX, PyTorch interchangeably. notebook walk key Keras 3 workflows. Let’s start installing Keras 3:","code":"install.packages(\"keras3\") keras3::install_keras()"},{"path":"https://keras3.posit.co/dev/articles/intro_to_keras_for_engineers.html","id":"setup","dir":"Articles","previous_headings":"","what":"Setup","title":"Introduction to Keras for engineers","text":"’re going using tensorflow backend – can edit string \"jax\" \"torch\" hit “Restart runtime”, whole notebook run just ! entire guide backend-agnostic.","code":"library(tensorflow, exclude = c(\"shape\", \"set_random_seed\")) library(keras3)  # Note that you must configure the backend # before calling any other keras functions. # The backend cannot be changed once the # package is imported. use_backend(\"tensorflow\")"},{"path":"https://keras3.posit.co/dev/articles/intro_to_keras_for_engineers.html","id":"a-first-example-a-mnist-convnet","dir":"Articles","previous_headings":"","what":"A first example: A MNIST convnet","title":"Introduction to Keras for engineers","text":"Let’s start Hello World ML: training convnet classify MNIST digits. ’s data: ’s model. Different model-building options Keras offers include: Sequential API (use ) Functional API (typical) Writing models via subclassing (advanced use cases) ’s model summary: use compile() method specify optimizer, loss function, metrics monitor. Note JAX TensorFlow backends, XLA compilation turned default. Let’s train evaluate model. ’ll set aside validation split 15% data training monitor generalization unseen data. training, saving model end epoch. can also save model latest state like : reload like : Next, can query predictions class probabilities predict(): ’s basics!","code":"# Load the data and split it between train and test sets c(c(x_train, y_train), c(x_test, y_test)) %<-% keras3::dataset_mnist()  # Scale images to the [0, 1] range x_train <- x_train / 255 x_test <- x_test / 255  # Make sure images have shape (28, 28, 1) x_train <- array_reshape(x_train, c(-1, 28, 28, 1)) x_test <- array_reshape(x_test, c(-1, 28, 28, 1))  dim(x_train) ## [1] 60000    28    28     1 dim(x_test) ## [1] 10000    28    28     1 # Model parameters num_classes <- 10 input_shape <- c(28, 28, 1)  model <- keras_model_sequential(input_shape = input_shape) model |>   layer_conv_2d(filters = 64, kernel_size = c(3, 3), activation = \"relu\") |>   layer_conv_2d(filters = 64, kernel_size = c(3, 3), activation = \"relu\") |>   layer_max_pooling_2d(pool_size = c(2, 2)) |>   layer_conv_2d(filters = 128, kernel_size = c(3, 3), activation = \"relu\") |>   layer_conv_2d(filters = 128, kernel_size = c(3, 3), activation = \"relu\") |>   layer_global_average_pooling_2d() |>   layer_dropout(rate = 0.5) |>   layer_dense(units = num_classes, activation = \"softmax\") summary(model) ## Model: \"sequential\" ## ┏━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━┓ ## ┃ Layer (type)                    ┃ Output Shape           ┃       Param # ┃ ## ┡━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━┩ ## │ conv2d (Conv2D)                 │ (None, 26, 26, 64)     │           640 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ conv2d_1 (Conv2D)               │ (None, 24, 24, 64)     │        36,928 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ max_pooling2d (MaxPooling2D)    │ (None, 12, 12, 64)     │             0 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ conv2d_2 (Conv2D)               │ (None, 10, 10, 128)    │        73,856 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ conv2d_3 (Conv2D)               │ (None, 8, 8, 128)      │       147,584 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ global_average_pooling2d        │ (None, 128)            │             0 │ ## │ (GlobalAveragePooling2D)        │                        │               │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ dropout (Dropout)               │ (None, 128)            │             0 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ dense (Dense)                   │ (None, 10)             │         1,290 │ ## └─────────────────────────────────┴────────────────────────┴───────────────┘ ##  Total params: 260,298 (1016.79 KB) ##  Trainable params: 260,298 (1016.79 KB) ##  Non-trainable params: 0 (0.00 B) model |> compile(   optimizer = \"adam\",   loss = \"sparse_categorical_crossentropy\",   metrics = list(     metric_sparse_categorical_accuracy(name = \"acc\")   ) ) batch_size <- 128 epochs <- 10  callbacks <- list(   callback_model_checkpoint(filepath=\"model_at_epoch_{epoch}.keras\"),   callback_early_stopping(monitor=\"val_loss\", patience=2) )  model |> fit(   x_train, y_train,   batch_size = batch_size,   epochs = epochs,   validation_split = 0.15,   callbacks = callbacks ) ## Epoch 1/10 ## 399/399 - 7s - 17ms/step - acc: 0.7464 - loss: 0.7476 - val_acc: 0.9627 - val_loss: 0.1272 ## Epoch 2/10 ## 399/399 - 2s - 5ms/step - acc: 0.9376 - loss: 0.2072 - val_acc: 0.9771 - val_loss: 0.0776 ## Epoch 3/10 ## 399/399 - 2s - 5ms/step - acc: 0.9571 - loss: 0.1475 - val_acc: 0.9816 - val_loss: 0.0633 ## Epoch 4/10 ## 399/399 - 2s - 5ms/step - acc: 0.9652 - loss: 0.1175 - val_acc: 0.9863 - val_loss: 0.0488 ## Epoch 5/10 ## 399/399 - 2s - 5ms/step - acc: 0.9713 - loss: 0.1004 - val_acc: 0.9870 - val_loss: 0.0469 ## Epoch 6/10 ## 399/399 - 2s - 5ms/step - acc: 0.9754 - loss: 0.0854 - val_acc: 0.9879 - val_loss: 0.0407 ## Epoch 7/10 ## 399/399 - 2s - 5ms/step - acc: 0.9767 - loss: 0.0788 - val_acc: 0.9899 - val_loss: 0.0378 ## Epoch 8/10 ## 399/399 - 2s - 5ms/step - acc: 0.9795 - loss: 0.0688 - val_acc: 0.9867 - val_loss: 0.0438 ## Epoch 9/10 ## 399/399 - 2s - 5ms/step - acc: 0.9799 - loss: 0.0666 - val_acc: 0.9893 - val_loss: 0.0396 score <- model |> evaluate(x_test, y_test, verbose = 0) save_model(model, \"final_model.keras\", overwrite=TRUE) model <- load_model(\"final_model.keras\") predictions <- model |> predict(x_test) ## 313/313 - 1s - 2ms/step dim(predictions) ## [1] 10000    10"},{"path":"https://keras3.posit.co/dev/articles/intro_to_keras_for_engineers.html","id":"writing-cross-framework-custom-components","dir":"Articles","previous_headings":"","what":"Writing cross-framework custom components","title":"Introduction to Keras for engineers","text":"Keras enables write custom Layers, Models, Metrics, Losses, Optimizers work across TensorFlow, JAX, PyTorch codebase. Let’s take look custom layers first. op_ namespace contains: implementation NumPy API, e.g. op_stack op_matmul. set neural network specific ops absent NumPy, op_conv op_binary_crossentropy. Let’s make custom Dense layer works backends: Next, let’s make custom Dropout layer relies random_* namespace: Next, let’s write custom subclassed model uses two custom layers: Let’s compile fit :","code":"layer_my_dense <- Layer(   classname = \"MyDense\",   initialize = function(units, activation = NULL, name = NULL, ...) {     super$initialize(name = name, ...)     self$units <- units     self$activation <- activation   },   build = function(input_shape) {     input_dim <- tail(input_shape, 1)     self$w <- self$add_weight(       shape = shape(input_dim, self$units),       initializer = initializer_glorot_normal(),       name = \"kernel\",       trainable = TRUE     )     self$b <- self$add_weight(       shape = shape(self$units),       initializer = initializer_zeros(),       name = \"bias\",       trainable = TRUE     )   },   call = function(inputs) {     # Use Keras ops to create backend-agnostic layers/metrics/etc.     x <- op_matmul(inputs, self$w) + self$b     if (!is.null(self$activation))       x <- self$activation(x)     x   } ) layer_my_dropout <- Layer(   \"MyDropout\",   initialize = function(rate, name = NULL, seed = NULL, ...) {     super$initialize(name = name)     self$rate <- rate     # Use seed_generator for managing RNG state.     # It is a state element and its seed variable is     # tracked as part of `layer$variables`.     self$seed_generator <- random_seed_generator(seed)   },   call = function(inputs) {     # Use `keras3::random_*` for random ops.     random_dropout(inputs, self$rate, seed = self$seed_generator)   } ) MyModel <- Model(   \"MyModel\",   initialize = function(num_classes, ...) {     super$initialize(...)     self$conv_base <-       keras_model_sequential() |>       layer_conv_2d(64, kernel_size = c(3, 3), activation = \"relu\") |>       layer_conv_2d(64, kernel_size = c(3, 3), activation = \"relu\") |>       layer_max_pooling_2d(pool_size = c(2, 2)) |>       layer_conv_2d(128, kernel_size = c(3, 3), activation = \"relu\") |>       layer_conv_2d(128, kernel_size = c(3, 3), activation = \"relu\") |>       layer_global_average_pooling_2d()      self$dp <- layer_my_dropout(rate = 0.5)     self$dense <- layer_my_dense(units = num_classes,                                  activation = activation_softmax)   },   call = function(inputs) {     inputs |>       self$conv_base() |>       self$dp() |>       self$dense()   } ) model <- MyModel(num_classes = 10) model |> compile(   loss = loss_sparse_categorical_crossentropy(),   optimizer = optimizer_adam(learning_rate = 1e-3),   metrics = list(     metric_sparse_categorical_accuracy(name = \"acc\")   ) )  model |> fit(   x_train, y_train,   batch_size = batch_size,   epochs = 1, # For speed   validation_split = 0.15 ) ## 399/399 - 6s - 15ms/step - acc: 0.7355 - loss: 0.7721 - val_acc: 0.9287 - val_loss: 0.2384"},{"path":"https://keras3.posit.co/dev/articles/intro_to_keras_for_engineers.html","id":"training-models-on-arbitrary-data-sources","dir":"Articles","previous_headings":"","what":"Training models on arbitrary data sources","title":"Introduction to Keras for engineers","text":"Keras models can trained evaluated wide variety data sources, independently backend ’re using. includes: Arrays Dataframes TensorFlow tf_dataset objects PyTorch DataLoader objects Keras PyDataset objects work whether ’re using TensorFlow, JAX, PyTorch Keras backend. Let’s try tf_dataset:","code":"library(tfdatasets, exclude = \"shape\")  train_dataset <- list(x_train, y_train) |>   tensor_slices_dataset() |>   dataset_batch(batch_size) |>   dataset_prefetch(buffer_size = tf$data$AUTOTUNE)  test_dataset <- list(x_test, y_test) |>   tensor_slices_dataset() |>   dataset_batch(batch_size) |>   dataset_prefetch(buffer_size = tf$data$AUTOTUNE)  model <- MyModel(num_classes = 10) model |> compile(   loss = loss_sparse_categorical_crossentropy(),   optimizer = optimizer_adam(learning_rate = 1e-3),   metrics = list(     metric_sparse_categorical_accuracy(name = \"acc\")   ) )  model |> fit(train_dataset, epochs = 1, validation_data = test_dataset) ## 469/469 - 7s - 15ms/step - acc: 0.7502 - loss: 0.7455 - val_acc: 0.9107 - val_loss: 0.2982"},{"path":"https://keras3.posit.co/dev/articles/intro_to_keras_for_engineers.html","id":"further-reading","dir":"Articles","previous_headings":"","what":"Further reading","title":"Introduction to Keras for engineers","text":"concludes short overview new multi-backend capabilities Keras 3. Next, can learn :","code":""},{"path":"https://keras3.posit.co/dev/articles/intro_to_keras_for_engineers.html","id":"how-to-customize-what-happens-in-fit","dir":"Articles","previous_headings":"Further reading","what":"How to customize what happens in fit()","title":"Introduction to Keras for engineers","text":"Want implement non-standard training algorithm still want benefit power usability fit()? ’s easy customize fit() support arbitrary use cases: Customizing happens fit() TensorFlow","code":""},{"path":"https://keras3.posit.co/dev/articles/intro_to_keras_for_engineers.html","id":"how-to-write-custom-training-loops","dir":"Articles","previous_headings":"","what":"How to write custom training loops","title":"Introduction to Keras for engineers","text":"Writing training loop scratch TensorFlow","code":""},{"path":"https://keras3.posit.co/dev/articles/intro_to_keras_for_engineers.html","id":"how-to-distribute-training","dir":"Articles","previous_headings":"","what":"How to distribute training","title":"Introduction to Keras for engineers","text":"Guide distributed training TensorFlow Enjoy library! 🚀","code":""},{"path":"https://keras3.posit.co/dev/articles/intro_to_keras_for_researchers.html","id":"setup","dir":"Articles","previous_headings":"","what":"Setup","title":"Introduction to Keras for Researchers","text":"","code":"library(keras3) library(tensorflow)"},{"path":"https://keras3.posit.co/dev/articles/intro_to_keras_for_researchers.html","id":"introduction","dir":"Articles","previous_headings":"","what":"Introduction","title":"Introduction to Keras for Researchers","text":"machine learning researcher? publish NeurIPS push state---art CV NLP? guide serve first introduction core Keras & TensorFlow API concepts. guide, learn : Tensors, variables, gradients TensorFlow Creating layers subclassing [Layer] class Writing low-level training loops Tracking losses created layers via add_loss() method Tracking metrics low-level training loop Speeding execution compiled [tensorflow::tf_function()] Executing layers training inference mode Keras Functional API also see Keras API action two end--end research examples: Variational Autoencoder, Hypernetwork.","code":""},{"path":"https://keras3.posit.co/dev/articles/intro_to_keras_for_researchers.html","id":"tensors","dir":"Articles","previous_headings":"","what":"Tensors","title":"Introduction to Keras for Researchers","text":"TensorFlow infrastructure layer differentiable programming. heart, ’s framework manipulating N-dimensional arrays (tensors), much like NumPy. However, three key differences NumPy TensorFlow: TensorFlow can leverage hardware accelerators GPUs TPUs. TensorFlow can automatically compute gradient arbitrary differentiable tensor expressions. TensorFlow computation can distributed large numbers devices single machine, large number machines (potentially multiple devices ). Let’s take look object core TensorFlow: Tensor. ’s constant tensor: can get value R array calling .array(): features attributes dtype shape: common way create constant tensors via tf$ones tf$zeros: can also create random constant tensors:","code":"x <- tf$constant(rbind(c(5, 2), c(1, 3))) print(x) ## tf.Tensor( ## [[5. 2.] ##  [1. 3.]], shape=(2, 2), dtype=float64) as.array(x) ##      [,1] [,2] ## [1,]    5    2 ## [2,]    1    3 x$dtype ## tf.float64 x$shape ## TensorShape([2, 2]) tf$ones(shape = shape(2, 1)) ## tf.Tensor( ## [[1.] ##  [1.]], shape=(2, 1), dtype=float32) tf$zeros(shape = shape(2, 1)) ## tf.Tensor( ## [[0.] ##  [0.]], shape=(2, 1), dtype=float32) x <- random_normal(shape = c(2, 2), mean = 0.0, stddev = 1.0) x <- random_uniform(shape = c(2, 2), minval = 0, maxval = 10)"},{"path":"https://keras3.posit.co/dev/articles/intro_to_keras_for_researchers.html","id":"variables","dir":"Articles","previous_headings":"","what":"Variables","title":"Introduction to Keras for Researchers","text":"Variables special tensors used store mutable state (weights neural network). create Variable using initial value: update value Variable using methods $assign(value), $assign_add(increment), $assign_sub(decrement):","code":"initial_value <- random_normal(shape=c(2, 2)) a <- tf$Variable(initial_value) print(a) ## <tf.Variable 'Variable:0' shape=(2, 2) dtype=float32, numpy= ## array([[ 0.9057419 ,  0.7916686 ], ##        [ 0.28754202, -0.5408822 ]], dtype=float32)> new_value <- random_normal(shape=c(2, 2)) a$assign(new_value) ## <tf.Variable 'UnreadVariable' shape=(2, 2) dtype=float32, numpy= ## array([[-0.3405368 , -2.1463926 ], ##        [ 1.2602988 ,  0.12241419]], dtype=float32)> added_value <- random_normal(shape=c(2, 2)) a$assign_add(added_value) ## <tf.Variable 'UnreadVariable' shape=(2, 2) dtype=float32, numpy= ## array([[ 0.04820395, -2.6854615 ], ##        [ 0.23246336,  1.4535258 ]], dtype=float32)>"},{"path":"https://keras3.posit.co/dev/articles/intro_to_keras_for_researchers.html","id":"doing-math-in-tensorflow","dir":"Articles","previous_headings":"","what":"Doing math in TensorFlow","title":"Introduction to Keras for Researchers","text":"’ve used NumPy, math TensorFlow look familiar. main difference TensorFlow code can run GPU TPU.","code":"a <- random_normal(shape=c(2, 2)) b <- random_normal(shape=c(2, 2))  c <- a + b d <- tf$square(c) e <- tf$exp(d)"},{"path":"https://keras3.posit.co/dev/articles/intro_to_keras_for_researchers.html","id":"gradients","dir":"Articles","previous_headings":"","what":"Gradients","title":"Introduction to Keras for Researchers","text":"’s another big difference R: can automatically retrieve gradient differentiable expression. Just open GradientTape, start “watching” tensor via tape$watch(), compose differentiable expression using tensor input: default, variables watched automatically, don’t need manually watch : Note can compute higher-order derivatives nesting tapes:","code":"a <- random_normal(shape=c(2, 2)) b <- random_normal(shape=c(2, 2))  with(tf$GradientTape() %as% tape, {   tape$watch(a)  # Start recording the history of operations applied to `a`   c <- tf$sqrt(tf$square(a) + tf$square(b))  # Do some math using `a`   # What's the gradient of `c` with respect to `a`?   dc_da <- tape$gradient(c, a)   print(dc_da) }) ## tf.Tensor( ## [[ 0.9969011  -0.7707146 ] ##  [ 0.23378514  0.96255165]], shape=(2, 2), dtype=float32) a <- tf$Variable(a)  with(tf$GradientTape() %as% tape, {   c <- tf$sqrt(tf$square(a) + tf$square(b))   dc_da <- tape$gradient(c, a)   print(dc_da) }) ## tf.Tensor( ## [[ 0.9969011  -0.7707146 ] ##  [ 0.23378514  0.96255165]], shape=(2, 2), dtype=float32) with(tf$GradientTape() %as% outer_tape, {   with(tf$GradientTape() %as% tape, {     c <- tf$sqrt(tf$square(a) + tf$square(b))     dc_da <- tape$gradient(c, a)   })   d2c_da2 <- outer_tape$gradient(dc_da, a)   print(d2c_da2) }) ## tf.Tensor( ## [[3.3447742e-03 7.1282005e-01] ##  [5.7464113e+00 5.5013180e-02]], shape=(2, 2), dtype=float32)"},{"path":"https://keras3.posit.co/dev/articles/intro_to_keras_for_researchers.html","id":"keras-layers","dir":"Articles","previous_headings":"","what":"Keras layers","title":"Introduction to Keras for Researchers","text":"TensorFlow infrastructure layer differentiable programming, dealing tensors, variables, gradients, Keras user interface deep learning, dealing layers, models, optimizers, loss functions, metrics, . Keras serves high-level API TensorFlow: Keras makes TensorFlow simple productive. Layer class fundamental abstraction Keras. Layer encapsulates state (weights) computation (defined call method). simple layer looks like . self$add_weight() method gives shortcut creating weights: use Layer instance much like R function: weight variables (created initialize) automatically tracked weights property: many built-layers available, Dense Conv2D LSTM fancier ones like Conv3DTranspose ConvLSTM2D. smart reusing built-functionality.","code":"Linear <- new_layer_class(   \"Linear\",   initialize = function(units = 32, input_dim = 32) {     super$initialize()     self$w <- self$add_weight(       shape = shape(input_dim, units),       initializer = \"random_normal\",       trainable = TRUE     )     self$b <- self$add_weight(       shape = shape(units),       initializer = \"zeros\",       trainable = TRUE     )   },   call = function(inputs) {     tf$matmul(inputs, self$w) + self$b   } ) # Instantiate our layer. linear_layer <- Linear(units=4, input_dim=2)  # The layer can be treated as a function. # Here we call it on some data. y <- linear_layer(tf$ones(shape(2, 2))) linear_layer$weights ## [[1]] ## <KerasVariable shape=(2, 4), dtype=float32, path=linear/variable> ## ## [[2]] ## <KerasVariable shape=(4), dtype=float32, path=linear/variable_1>"},{"path":"https://keras3.posit.co/dev/articles/intro_to_keras_for_researchers.html","id":"layer-weight-creation-in-buildinput_shape","dir":"Articles","previous_headings":"","what":"Layer weight creation in build(input_shape)","title":"Introduction to Keras for Researchers","text":"’s often good idea defer weight creation build() method, don’t need specify input dim/shape layer construction time:","code":"Linear <- new_layer_class(   \"Linear\",   initialize = function(units = 32) {     super$initialize()     self$units <- units   },   build = function(input_shape) {     self$w <- self$add_weight(       shape = shape(input_shape[-1], self$units),       initializer = \"random_normal\",       trainable = TRUE     )     self$b <- self$add_weight(       shape = shape(self$units),       initializer = \"zeros\",       trainable = TRUE     )   },   call = function(inputs) {     tf$matmul(inputs, self$w) + self$b   } )  # Instantiate our layer. linear_layer <- Linear(units = 4)  # This will also call `build(input_shape)` and create the weights. y <- linear_layer(tf$ones(shape(2, 2)))"},{"path":"https://keras3.posit.co/dev/articles/intro_to_keras_for_researchers.html","id":"layer-gradients","dir":"Articles","previous_headings":"","what":"Layer gradients","title":"Introduction to Keras for Researchers","text":"can automatically retrieve gradients weights layer calling inside GradientTape. Using gradients, can update weights layer, either manually, using optimizer object. course, can modify gradients using , need .","code":"# Prepare a dataset. c(c(x_train, y_train), .) %<-% dataset_mnist()  x_train <- array_reshape(x_train, c(60000, 784)) / 255  dataset <- tfdatasets::tensor_slices_dataset(list(x_train, y_train)) %>%   tfdatasets::dataset_shuffle(buffer_size=1024) %>%   tfdatasets::dataset_batch(64)  # Instantiate our linear layer (defined above) with 10 units. linear_layer <- Linear(units = 10)  # Instantiate a logistic loss function that expects integer targets. loss_fn <- loss_sparse_categorical_crossentropy(from_logits=TRUE)  # Instantiate an optimizer. optimizer <- optimizer_sgd(learning_rate=1e-3)  # Iterate over the batches of the dataset. coro::loop(for(data in dataset) {   # Open a GradientTape.   with(tf$GradientTape() %as% tape, {     # Forward pass.     logits <- linear_layer(data[[1]])      # Loss value for this batch.     loss_value <- loss_fn(data[[2]], logits)   })    # Get gradients of the loss wrt the weights.   gradients <- tape$gradient(loss_value, linear_layer$trainable_weights)    # Update the weights of our linear layer.   optimizer$apply_gradients(zip_lists(gradients, linear_layer$trainable_weights)) }) loss_value ## tf.Tensor(1.2819729, shape=(), dtype=float32)"},{"path":"https://keras3.posit.co/dev/articles/intro_to_keras_for_researchers.html","id":"trainable-and-non-trainable-weights","dir":"Articles","previous_headings":"","what":"Trainable and non-trainable weights","title":"Introduction to Keras for Researchers","text":"Weights created layers can either trainable non-trainable. ’re exposed trainable_weights non_trainable_weights respectively. ’s layer non-trainable weight:","code":"ComputeSum <- new_layer_class(   \"ComputeSum\",   initialize = function(input_dim) {     super$initialize()     # Create a non-trainable weight.     self$total <- self$add_weight(       initializer = \"zeros\",       shape = shape(input_dim),       trainable = FALSE     )   },   call = function(inputs) {     self$total$assign_add(tf$reduce_sum(inputs, axis=0L))     self$total   } )  my_sum <- ComputeSum(input_dim = 2) x <- tf$ones(shape(2, 2))  as.array(my_sum(x)) ## [1] 2 2 as.array(my_sum(x)) ## [1] 4 4 my_sum$trainable_weights ## list()"},{"path":"https://keras3.posit.co/dev/articles/intro_to_keras_for_researchers.html","id":"layers-that-own-layers","dir":"Articles","previous_headings":"","what":"Layers that own layers","title":"Introduction to Keras for Researchers","text":"Layers can recursively nested create bigger computation blocks. layer track weights sublayers (trainable non-trainable). Note manually-created MLP equivalent following built-option:","code":"# Let's reuse the Linear class # with a `build` method that we defined above.  MLP <- new_layer_class(   \"MLP\",   initialize = function() {     super$initialize()     self$linear_1 <- Linear(units = 32)     self$linear_2 <- Linear(units = 32)     self$linear_3 <- Linear(units = 10)   },   call = function(inputs) {     x <- self$linear_1(inputs)     x <- tf$nn$relu(x)     x <- self$linear_2(x)     x <- tf$nn$relu(x)     return(self$linear_3(x))   } )  mlp <- MLP()  # The first call to the `mlp` object will create the weights. y <- mlp(tf$ones(shape=shape(3, 64)))  # Weights are recursively tracked. length(mlp$weights) ## [1] 6 mlp <- keras_model_sequential() %>%   layer_dense(units = 32, activation = \"relu\") %>%   layer_dense(units = 32, activation = \"relu\") %>%   layer_dense(units = 10)"},{"path":"https://keras3.posit.co/dev/articles/intro_to_keras_for_researchers.html","id":"tracking-losses-created-by-layers","dir":"Articles","previous_headings":"","what":"Tracking losses created by layers","title":"Introduction to Keras for Researchers","text":"Layers can create losses forward pass via add_loss() method. especially useful regularization losses. losses created sublayers recursively tracked parent layers. ’s layer creates activity regularization loss: model incorporating layer track regularization loss: losses cleared top-level layer start forward pass – don’t accumulate. layer.losses always contains losses created last forward pass. typically use losses summing computing gradients writing training loop.","code":"# A layer that creates an activity sparsity regularization loss ActivityRegularization <- new_layer_class(   \"ActivityRegularization\",   initialize = function(rate=1e-2) {     super$initialize()     self$rate <- rate   },   call = function(inputs) {     self$add_loss(self$rate * tf$reduce_sum(tf$abs(inputs)))     inputs   } ) # Let's use the loss layer in a MLP block. SparseMLP <- new_layer_class(   \"SparseMLP\",   initialize = function() {     super$initialize()     self$linear_1 <- Linear(units = 32)     self$reg <- ActivityRegularization(rate = 1e-2)     self$linear_3 <- Linear(units = 10)   },   call = function(inputs) {     x <- self$linear_1(inputs)     x <- tf$nn$relu(x)     x <- self$reg(x)     return(self$linear_3(x))   } )  mlp <- SparseMLP() y <- mlp(tf$ones(shape(10, 10)))  mlp$losses  # List containing one float32 scalar ## [[1]] ## tf.Tensor(0.18065463, shape=(), dtype=float32) # Losses correspond to the *last* forward pass. mlp <- SparseMLP() mlp(tf$ones(shape(10, 10))) ## tf.Tensor( ## [[ 0.0388482  -0.03920118  0.01624808 -0.01361975 -0.01354899  0.07107338 ##   -0.01077365  0.05688906 -0.02838149 -0.04084621] ##  [ 0.0388482  -0.03920118  0.01624808 -0.01361975 -0.01354899  0.07107338 ##   -0.01077365  0.05688906 -0.02838149 -0.04084621] ##  [ 0.0388482  -0.03920118  0.01624808 -0.01361975 -0.01354899  0.07107338 ##   -0.01077365  0.05688906 -0.02838149 -0.04084621] ##  [ 0.0388482  -0.03920118  0.01624808 -0.01361975 -0.01354899  0.07107338 ##   -0.01077365  0.05688906 -0.02838149 -0.04084621] ##  [ 0.0388482  -0.03920118  0.01624808 -0.01361975 -0.01354899  0.07107338 ##   -0.01077365  0.05688906 -0.02838149 -0.04084621] ##  [ 0.0388482  -0.03920118  0.01624808 -0.01361975 -0.01354899  0.07107338 ##   -0.01077365  0.05688906 -0.02838149 -0.04084621] ##  [ 0.0388482  -0.03920118  0.01624808 -0.01361975 -0.01354899  0.07107338 ##   -0.01077365  0.05688906 -0.02838149 -0.04084621] ##  [ 0.0388482  -0.03920118  0.01624808 -0.01361975 -0.01354899  0.07107338 ##   -0.01077365  0.05688906 -0.02838149 -0.04084621] ##  [ 0.0388482  -0.03920118  0.01624808 -0.01361975 -0.01354899  0.07107338 ##   -0.01077365  0.05688906 -0.02838149 -0.04084621] ##  [ 0.0388482  -0.03920118  0.01624808 -0.01361975 -0.01354899  0.07107338 ##   -0.01077365  0.05688906 -0.02838149 -0.04084621]], shape=(10, 10), dtype=float32) length(mlp$losses) ## [1] 1 mlp(tf$ones(shape(10, 10))) ## tf.Tensor( ## [[ 0.0388482  -0.03920118  0.01624808 -0.01361975 -0.01354899  0.07107338 ##   -0.01077365  0.05688906 -0.02838149 -0.04084621] ##  [ 0.0388482  -0.03920118  0.01624808 -0.01361975 -0.01354899  0.07107338 ##   -0.01077365  0.05688906 -0.02838149 -0.04084621] ##  [ 0.0388482  -0.03920118  0.01624808 -0.01361975 -0.01354899  0.07107338 ##   -0.01077365  0.05688906 -0.02838149 -0.04084621] ##  [ 0.0388482  -0.03920118  0.01624808 -0.01361975 -0.01354899  0.07107338 ##   -0.01077365  0.05688906 -0.02838149 -0.04084621] ##  [ 0.0388482  -0.03920118  0.01624808 -0.01361975 -0.01354899  0.07107338 ##   -0.01077365  0.05688906 -0.02838149 -0.04084621] ##  [ 0.0388482  -0.03920118  0.01624808 -0.01361975 -0.01354899  0.07107338 ##   -0.01077365  0.05688906 -0.02838149 -0.04084621] ##  [ 0.0388482  -0.03920118  0.01624808 -0.01361975 -0.01354899  0.07107338 ##   -0.01077365  0.05688906 -0.02838149 -0.04084621] ##  [ 0.0388482  -0.03920118  0.01624808 -0.01361975 -0.01354899  0.07107338 ##   -0.01077365  0.05688906 -0.02838149 -0.04084621] ##  [ 0.0388482  -0.03920118  0.01624808 -0.01361975 -0.01354899  0.07107338 ##   -0.01077365  0.05688906 -0.02838149 -0.04084621] ##  [ 0.0388482  -0.03920118  0.01624808 -0.01361975 -0.01354899  0.07107338 ##   -0.01077365  0.05688906 -0.02838149 -0.04084621]], shape=(10, 10), dtype=float32) length(mlp$losses)  # No accumulation. ## [1] 1 # Let's demonstrate how to use these losses in a training loop.  # Prepare a dataset. c(c(x_train, y_train), .) %<-% dataset_mnist() x_train <- array_reshape(x_train, c(60000, 784)) / 255  dataset <- tfdatasets::tensor_slices_dataset(list(x_train, y_train)) %>%   tfdatasets::dataset_shuffle(buffer_size=1024) %>%   tfdatasets::dataset_batch(64)  # A new MLP. mlp <- SparseMLP()  # Loss and optimizer. loss_fn <- loss_sparse_categorical_crossentropy(from_logits=TRUE) optimizer <- optimizer_sgd(learning_rate=1e-3)  coro::loop(for(data in dataset) {   x <- data[[1]]   y <- data[[2]]   with(tf$GradientTape() %as% tape, {     # Forward pass.     logits <- mlp(x)      # External loss value for this batch.     loss <- loss_fn(y, logits)      # Add the losses created during the forward pass.     loss <- loss + Reduce(`+`, mlp$losses)      # Get gradients of the loss wrt the weights.     gradients <- tape$gradient(loss, mlp$trainable_weights)      # Update the weights of our linear layer.     optimizer$apply_gradients(zip_lists(gradients, mlp$trainable_weights))   }) })"},{"path":"https://keras3.posit.co/dev/articles/intro_to_keras_for_researchers.html","id":"keeping-track-of-training-metrics","dir":"Articles","previous_headings":"","what":"Keeping track of training metrics","title":"Introduction to Keras for Researchers","text":"Keras offers broad range built-metrics, like metric_auc metric_precision_at_recall. ’s also easy create metrics lines code. use metric custom training loop, : Instantiate metric object, e.g. metric = metric_auc() Call metric$udpate_state(targets, predictions) method batch data Query result via metric$result() Reset metric’s state end epoch start evaluation via metric$reset_state() ’s simple example: can also define metrics subclassing keras.metrics.Metric. need override three functions called : Override update_state() update statistic values. Override result() return metric value. Override reset_state() reset metric initial state. example implement F1-score metric (support sample weighting). Let’s test-drive :","code":"# Instantiate a metric object accuracy <- metric_sparse_categorical_accuracy()  # Prepare our layer, loss, and optimizer. model <- keras_model_sequential() %>%   layer_dense(units = 32, activation = \"relu\") %>%   layer_dense(units = 32, activation = \"relu\") %>%   layer_dense(units = 10) loss_fn <- loss_sparse_categorical_crossentropy(from_logits = TRUE) optimizer <- optimizer_adam(learning_rate=1e-3)  for (epoch in seq_len(2)) {   coro::loop(for (data in dataset) {     x <- data[[1]]     y <- data[[2]]     with(tf$GradientTape() %as% tape, {       # Forward pass.       logits <- model(x)        # External loss value for this batch.       loss_value <- loss_fn(y, logits)     })      # Update the state of the `accuracy` metric.     accuracy$update_state(y, logits)      # Update the weights of the model to minimize the loss value.     gradients <- tape$gradient(loss_value, model$trainable_weights)     optimizer$apply_gradients(zip_lists(gradients, model$trainable_weights))    })   cat(\"Epoch:\", epoch, \"Accuracy:\", as.numeric(accuracy$result()), \"\\n\")   accuracy$reset_state() } ## Epoch: 1 Accuracy: 0.8757833 ## Epoch: 2 Accuracy: 0.93915 F1Score <- new_metric_class(   \"F1Score\",   initialize = function(self, name=\"f1_score\", dtype=\"float32\", threshold=0.5, ...) {     super$initialize(name=name, dtype=dtype, ...)     self$threshold <- threshold     self$true_positives <- self$add_weight(       name=\"tp\", dtype=dtype, initializer=\"zeros\"     )     self$false_positives <- self$add_weight(       name=\"fp\", dtype=dtype, initializer=\"zeros\"     )     self$false_negatives <- self$add_weight(       name=\"fn\", dtype=dtype, initializer=\"zeros\"     )   },   update_state = function(y_true, y_pred, sample_weight=NULL) {     y_pred <- tf$math$greater_equal(y_pred, self$threshold)     y_true <- tf$cast(y_true, tf$bool)     y_pred <- tf$cast(y_pred, tf$bool)      true_positives <- tf$cast(y_true & y_pred, self$dtype)     false_positives <- tf$cast((!y_true) & y_pred, self$dtype)     false_negatives <- tf$cast(y_true & (!y_pred), self$dtype)      if (!is.null(sample_weight)) {       sample_weight <- tf$cast(sample_weight, self$dtype)       true_positives <- true_positives * sample_weight       false_positives <- false_positives * sample_weight       false_negatives <- false_negatives * sample_weight     }      self$true_positives$assign_add(tf$reduce_sum(true_positives))     self$false_positives$assign_add(tf$reduce_sum(false_positives))     self$false_negatives$assign_add(tf$reduce_sum(false_negatives))   },    result = function() {     precision <- self$true_positives / (self$true_positives + self$false_positives)     recall <- self$true_positives / (self$true_positives + self$false_negatives)     f1_score <- 2 * precision * recall / (precision + recall)     f1_score   },    reset_state = function() {     self$true_positives$assign(0)     self$false_positives$assign(0)     self$false_negatives$assign(0)   } ) m <- F1Score() m$update_state(c(0, 1, 0, 0), c(0.3, 0.5, 0.8, 0.9)) cat(\"Intermediate result:\", as.numeric(m$result()), \"\\n\") ## Intermediate result: 0.5 m$update_state(c(1, 1, 1, 1), c(0.1, 0.7, 0.6, 0.0)) cat(\"Final result:\", as.numeric(m$result()), \"\\n\") ## Final result: 0.6"},{"path":"https://keras3.posit.co/dev/articles/intro_to_keras_for_researchers.html","id":"compiled-functions","dir":"Articles","previous_headings":"","what":"Compiled functions","title":"Introduction to Keras for Researchers","text":"Running eagerly great debugging, get better performance compiling computation static graphs. Static graphs researcher’s best friends. can compile function wrapping tf.function decorator.","code":"# Prepare our layer, loss, and optimizer. model <- keras_model_sequential() %>%   layer_dense(units = 32, activation = \"relu\") %>%   layer_dense(units = 32, activation = \"relu\") %>%   layer_dense(units = 10) loss_fn <- loss_sparse_categorical_crossentropy(from_logits = TRUE) optimizer <- optimizer_adam(learning_rate=1e-3)  # Create a training step function. train_on_batch <- tf_function(function(x, y) {   with(tf$GradientTape() %as% tape, {     # Forward pass.     logits <- model(x)     # External loss value for this batch.     loss_value <- loss_fn(y, logits)   })   # Update the weights of the model to minimize the loss value.   gradients <- tape$gradient(loss_value, model$trainable_weights)   optimizer$apply_gradients(zip_lists(gradients, model$trainable_weights))   loss_value })   # Prepare a dataset. c(c(x_train, y_train), .) %<-% dataset_mnist() x_train <- array_reshape(x_train, c(60000, 784)) / 255  dataset <- tfdatasets::tensor_slices_dataset(list(x_train, y_train)) %>%   tfdatasets::dataset_shuffle(buffer_size=1024) %>%   tfdatasets::dataset_batch(64)  i <- 0 coro::loop(for (data in dataset) {   i <- i + 1   x <- data[[1]]   y <- data[[2]]   loss <- train_on_batch(x, y)   if (i %% 100 == 0)     cat(\"Loss:\", as.numeric(loss), \"\\n\") }) ## Loss: 0.551749 ## Loss: 0.2131135 ## Loss: 0.2765952 ## Loss: 0.1296219 ## Loss: 0.2657076 ## Loss: 0.2683381 ## Loss: 0.1570166 ## Loss: 0.3139241 ## Loss: 0.08981849"},{"path":"https://keras3.posit.co/dev/articles/intro_to_keras_for_researchers.html","id":"training-mode-inference-mode","dir":"Articles","previous_headings":"","what":"Training mode & inference mode","title":"Introduction to Keras for Researchers","text":"layers, particular BatchNormalization layer Dropout layer, different behaviors training inference. layers, standard practice expose training (boolean) argument call method. exposing argument call, enable built-training evaluation loops (e.g. fit) correctly use layer training inference modes.","code":"Dropout <- new_layer_class(   \"Dropout\",   initialize = function(rate) {     super$initialize()     self$rate <- rate   },   call = function(inputs, training = NULL) {     if (!is.null(training) && training) {       return(tf$nn$dropout(inputs, rate = self$rate))     }     inputs   } )  MLPWithDropout <- new_layer_class(   \"MLPWithDropout\",   initialize = function() {     super$initialize()     self$linear_1 <- Linear(units = 32)     self$dropout <- Dropout(rate = 0.5)     self$linear_3 <- Linear(units = 10)   },   call = function(inputs, training = NULL) {     x <- self$linear_1(inputs)     x <- tf$nn$relu(x)     x <- self$dropout(x, training = training)     self$linear_3(x)   } )  mlp <- MLPWithDropout() y_train <- mlp(tf$ones(shape(2, 2)), training=TRUE) y_test <- mlp(tf$ones(shape(2, 2)), training=FALSE)"},{"path":"https://keras3.posit.co/dev/articles/intro_to_keras_for_researchers.html","id":"the-functional-api-for-model-building","dir":"Articles","previous_headings":"","what":"The Functional API for model-building","title":"Introduction to Keras for Researchers","text":"build deep learning models, don’t use object-oriented programming time. layers ’ve seen far can also composed functionally, like (call “Functional API”): Functional API tends concise subclassing, provides advantages (generally advantages functional, typed languages provide untyped OO development). However, can used define DAGs layers – recursive networks defined Layer subclasses instead. Learn Functional API . research workflows, may often find mix--matching OO models Functional models. Note Model class also features built-training & evaluation loops: fit(), predict() evaluate() (configured via compile() method). built-functions give access following built-training infrastructure features: Callbacks. can leverage built-callbacks early-stopping, model checkpointing, monitoring training TensorBoard. can also implement custom callbacks needed. Distributed training. can easily scale training multiple GPUs, TPU, even multiple machines tf.distribute API – changes code. Step fusing. steps_per_execution argument Model.compile(), can process multiple batches single tf.function call, greatly improves device utilization TPUs. won’t go details, provide simple code example . leverages built-training infrastructure implement MNIST example . can always subclass Model class (works exactly like subclassing Layer) want leverage built-training loops OO models. Just override Model$train_step() customize happens fit() retaining support built-infrastructure features outlined – callbacks, zero-code distribution support, step fusing support. may also override test_step() customize happens evaluate(), override predict_step() customize happens predict(). information, please refer guide.","code":"# We use an `Input` object to describe the shape and dtype of the inputs. # This is the deep learning equivalent of *declaring a type*. # The shape argument is per-sample; it does not include the batch size. # The functional API focused on defining per-sample transformations. # The model we create will automatically batch the per-sample transformations, # so that it can be called on batches of data. inputs <- layer_input(shape = 16, dtype = \"float32\")  # We call layers on these \"type\" objects # and they return updated types (new shapes/dtypes). outputs <- inputs %>%   Linear(units = 32) %>% # We are reusing the Linear layer we defined earlier.   Dropout(rate = 0.5) %>% # We are reusing the Dropout layer we defined earlier.   Linear(units = 10)  # A functional `Model` can be defined by specifying inputs and outputs. # A model is itself a layer like any other. model <- keras_model(inputs, outputs)  # A functional model already has weights, before being called on any data. # That's because we defined its input shape in advance (in `Input`). length(model$weights) ## [1] 4 # Let's call our model on some data, for fun. y <- model(tf$ones(shape(2, 16))) y$shape ## TensorShape([2, 10]) # You can pass a `training` argument in `__call__` # (it will get passed down to the Dropout layer). y <- model(tf$ones(shape(2, 16)), training=TRUE) inputs <- layer_input(shape = 784, dtype=\"float32\") outputs <- inputs %>%   layer_dense(units = 32, activation = \"relu\") %>%   layer_dense(units = 32, activation = \"relu\") %>%   layer_dense(units = 10) model <- keras_model(inputs, outputs)  # Specify the loss, optimizer, and metrics with `compile()`. model %>% compile(     loss = loss_sparse_categorical_crossentropy(from_logits=TRUE),     optimizer=optimizer_adam(learning_rate=1e-3),     metrics=list(metric_sparse_categorical_accuracy()), )  # Train the model with the dataset for 2 epochs. model %>% fit(dataset, epochs=2) ## Epoch 1/2 ## 938/938 - 4s - 4ms/step - loss: 0.3958 - sparse_categorical_accuracy: 0.8866 ## Epoch 2/2 ## 938/938 - 1s - 960us/step - loss: 0.1888 - sparse_categorical_accuracy: 0.9443 predictions <- model %>% predict(dataset) ## 938/938 - 1s - 1ms/step model %>% evaluate(dataset) ## 938/938 - 1s - 1ms/step - loss: 0.1763 - sparse_categorical_accuracy: 0.9454 ## $loss ## [1] 0.1763445 ## ## $sparse_categorical_accuracy ## [1] 0.9454167 CustomModel <- new_model_class(   \"CustomModel\",   initialize = function(...) {     super$initialize(...)     self$loss_tracker <- metric_mean(name=\"loss\")     self$accuracy <- metric_sparse_categorical_accuracy()     self$loss_fn <- loss_sparse_categorical_crossentropy(from_logits=TRUE)     self$optimizer <- optimizer_adam(learning_rate=1e-3)   },   train_step = function(data) {     c(x, y = NULL, sample_weight = NULL) %<-% data     with(tf$GradientTape() %as% tape, {       y_pred <- self(x, training=TRUE)       loss <- self$loss_fn(y = y, y_pred = y_pred, sample_weight=sample_weight)     })     gradients <- tape$gradient(loss, self$trainable_variables)     self$optimizer$apply_gradients(       zip_lists(gradients, self$trainable_variables)     )      # Update metrics (includes the metric that tracks the loss)     self$loss_tracker$update_state(loss)     self$accuracy$update_state(y, y_pred, sample_weight=sample_weight)     # Return a list mapping metric names to current value     list(       loss = self$loss_tracker$result(),       accuracy = self$accuracy$result()     )   },   metrics = mark_active(function() {     list(self$loss_tracker, self$accuracy)   }) )  inputs <- layer_input(shape = 784, dtype=\"float32\") outputs <- inputs %>%   layer_dense(units = 32, activation = \"relu\") %>%   layer_dense(units = 32, activation = \"relu\") %>%   layer_dense(units = 10) model <- CustomModel(inputs, outputs) model %>% compile() model %>% fit(dataset, epochs=2) ## Epoch 1/2 ## 938/938 - 2s - 2ms/step - loss: 0.3869 - sparse_categorical_accuracy: 0.8924 ## Epoch 2/2 ## 938/938 - 1s - 1ms/step - loss: 0.2163 - sparse_categorical_accuracy: 0.9370"},{"path":"https://keras3.posit.co/dev/articles/intro_to_keras_for_researchers.html","id":"end-to-end-experiment-example-1-variational-autoencoders-","dir":"Articles","previous_headings":"","what":"End-to-end experiment example 1: variational autoencoders.","title":"Introduction to Keras for Researchers","text":"things ’ve learned far: Layer encapsulates state (created __init__ build) computation (defined call). Layers can recursively nested create new, bigger computation blocks. can easily write highly hackable training loops opening GradientTape, calling model inside tape’s scope, retrieving gradients applying via optimizer. can speed training loops using @tf.function decorator. Layers can create track losses (typically regularization losses) via self.add_loss(). Let’s put things together end--end example: ’re going implement Variational AutoEncoder (VAE). ’ll train MNIST digits. VAE subclass Layer, built nested composition layers subclass Layer. feature regularization loss (KL divergence). model definition. First, Encoder class, uses Sampling layer map MNIST digit latent-space triplet (z_mean, z_log_var, z). Next, Decoder class, maps probabilistic latent space coordinates back MNIST digit. Finally, VariationalAutoEncoder composes together encoder decoder, creates KL divergence regularization loss via add_loss(). Now, let’s write training loop. training step decorated @tf.function compile super fast graph function. can see, building training type model Keras quick painless.","code":"Sampling <- new_layer_class(   \"Sampling\",   call = function(inputs) {     c(z_mean, z_log_var) %<-% inputs     batch <- op_shape(z_mean)[[1]]     dim <- op_shape(z_mean)[[2]]     epsilon <- random_normal(shape = c(batch, dim))     z_mean + op_exp(0.5 * z_log_var) * epsilon   } )  Encoder <- new_layer_class(   \"Encoder\",   initialize = function(latent_dim = 32, intermediate_dim = 64, ...) {     super$initialize(...)     self$dense_proj <- layer_dense(units = intermediate_dim, activation = \"relu\")     self$dense_mean <- layer_dense(units = latent_dim)     self$dense_log_var <- layer_dense(units = latent_dim)     self$sampling <- Sampling()   },   call = function(inputs) {     x <- self$dense_proj(inputs)     z_mean <- self$dense_mean(x)     z_log_var <- self$dense_log_var(x)     z <- self$sampling(list(z_mean, z_log_var))     list(z_mean, z_log_var, z)   } ) Decoder <- new_layer_class(   \"Decoder\",   initialize = function(original_dim, intermediate_dim = 64, ...) {     super$initialize(...)     self$dense_proj <- layer_dense(units = intermediate_dim, activation = \"relu\")     self$dense_output <- layer_dense(units = original_dim, activation = \"sigmoid\")   },   call = function(inputs) {     x <- self$dense_proj(inputs)     self$dense_output(x)   } ) VariationalAutoEncoder <- new_model_class(   \"VariationalAutoEncoder\",   initialize = function(original_dim,         intermediate_dim=64,         latent_dim=32,         name=\"autoencoder\", ...) {     super$initialize(name = name, ...)     self$original_dim <- original_dim     self$encoder <- Encoder(       latent_dim = latent_dim,       intermediate_dim = intermediate_dim     )     self$decoder <- Decoder(       original_dim = original_dim,       intermediate_dim = intermediate_dim     )   },   call = function(inputs) {     c(z_mean, z_log_var, z) %<-% self$encoder(inputs)     reconstructed <- self$decoder(z)     # Add KL divergence regularization loss.     kl_loss <- -0.5 * op_mean(       z_log_var - op_square(z_mean) - op_exp(z_log_var) + 1     )     self$add_loss(kl_loss)     reconstructed   } ) # Our model. vae <- VariationalAutoEncoder(   original_dim = 784,   intermediate_dim = 64,   latent_dim = 32 )  # Loss and optimizer. loss_fn <- loss_mean_squared_error() optimizer = optimizer_adam(learning_rate=1e-3)  # Prepare a dataset. c(c(x_train, .), .) %<-% dataset_mnist() x_train <- array_reshape(x_train, c(60000, 784)) / 255  dataset <- tfdatasets::tensor_slices_dataset(x_train) %>%   tfdatasets::dataset_shuffle(buffer_size=1024) %>%   tfdatasets::dataset_batch(32)   training_step <- tf_function(function(x) {   with(tf$GradientTape() %as% tape, {     reconstructed <- vae(x)  # Compute input reconstruction.     # Compute loss.     loss <- loss_fn(x, reconstructed)     loss <- loss + op_sum(vae$losses)  # Add KLD term.   })   # Update the weights of the VAE.   grads <- tape$gradient(loss, vae$trainable_weights)   optimizer$apply_gradients(zip_lists(grads, vae$trainable_weights))   loss })  losses <- c()  # Keep track of the losses over time. coro::loop(for(data in dataset) {   loss <- training_step(data)    # Logging.   losses[length(losses) + 1] <- as.numeric(loss)   if (length(losses) %% 100 == 0) {     cat(\"Step:\", length(losses), \"Loss:\", mean(losses), \"\\n\")   }   # Stop after 1000 steps.   # Training the model to convergence is left   # as an exercise to the reader.   if (length(losses) >= 1000) {     break   } }) ## Step: 100 Loss: 0.1270978 ## Step: 200 Loss: 0.1003238 ## Step: 300 Loss: 0.09001128 ## Step: 400 Loss: 0.08493649 ## Step: 500 Loss: 0.08171404 ## Step: 600 Loss: 0.07926706 ## Step: 700 Loss: 0.07790599 ## Step: 800 Loss: 0.07670419 ## Step: 900 Loss: 0.07570736 ## Step: 1000 Loss: 0.07476593"},{"path":"https://keras3.posit.co/dev/articles/intro_to_keras_for_researchers.html","id":"end-to-end-experiment-example-2-hypernetworks-","dir":"Articles","previous_headings":"","what":"End-to-end experiment example 2: hypernetworks.","title":"Introduction to Keras for Researchers","text":"Let’s take look another kind research experiment: hypernetworks. idea use small deep neural network (hypernetwork) generate weights larger network (main network). Let’s implement really trivial hypernetwork: ’ll use small 2-layer network generate weights larger 3-layer network. training loop. batch data: use hypernetwork generate array weight coefficients, weights_pred reshape coefficients kernel & bias tensors main_network run forward pass main_network compute actual MNIST predictions run backprop weights hypernetwork minimize final classification loss Implementing arbitrary research ideas Keras straightforward highly productive. Imagine trying 25 ideas per day (20 minutes per experiment average)! Keras designed go idea results fast possible, believe key great research. hope enjoyed quick introduction. Let us know build Keras!","code":"input_dim <- 784 classes <- 10  # This is the main network we'll actually use to predict labels. inputs <- layer_input(shape = input_dim) dense1 <- layer_dense(units = 64, activation = \"relu\") dense1$built <- TRUE  dense2 <- layer_dense(units = classes) dense2$built <- TRUE  outputs <- inputs %>% dense1() %>% dense2() main_network <- keras_model(inputs, outputs)  # This is the number of weight coefficients to generate. Each layer in the # main network requires output_dim * input_dim + output_dim coefficients. num_weights_to_generate <- (classes * 64 + classes) + (64 * input_dim + 64)  # This is the hypernetwork that generates the weights of the `main_network` above. hypernetwork <- keras_model_sequential() %>%   layer_dense(units=16, activation=\"relu\") %>%   layer_dense(units=num_weights_to_generate, activation=\"sigmoid\") # Loss and optimizer. loss_fn <- loss_sparse_categorical_crossentropy(from_logits = TRUE) optimizer <- optimizer_adam(learning_rate=1e-4)  # Prepare a dataset. c(c(x_train, y_train), .) %<-% dataset_mnist() x_train <- array_reshape(x_train, c(60000, 784)) / 255  dataset <- tfdatasets::tensor_slices_dataset(list(x_train, y_train)) %>%   tfdatasets::dataset_shuffle(buffer_size=1024) %>%   # We'll use a batch size of 1 for this experiment.   tfdatasets::dataset_batch(1)  train_step <- function(x, y) {   with(tf$GradientTape() %as% tape, {     weights_pred <- hypernetwork(x)      # Reshape them to the expected shapes for w and b for the outer model.     # Layer 1 kernel.     start_index <- 1     w1_shape <- c(input_dim, 64)     w1_coeffs <- weights_pred[, start_index:(start_index + prod(w1_shape) - 1)]     w1 <- tf$reshape(w1_coeffs, as.integer(w1_shape))     start_index <- start_index + prod(w1_shape)      # Layer 1 bias.     b1_shape <- c(64)     b1_coeffs <- weights_pred[, start_index:(start_index + prod(b1_shape) - 1)]     b1 <- tf$reshape(b1_coeffs, as.integer(b1_shape))     start_index <- start_index + prod(b1_shape)      # Layer 2 kernel.     w2_shape <- c(64, classes)     w2_coeffs <- weights_pred[, start_index:(start_index + prod(w2_shape) - 1)]     w2 <- tf$reshape(w2_coeffs, as.integer(w2_shape))     start_index <- start_index + prod(w2_shape)      # Layer 2 bias.     b2_shape <- c(classes)     b2_coeffs <- weights_pred[, start_index:(start_index + prod(b2_shape) - 1)]     b2 <- tf$reshape(b2_coeffs, as.integer(b2_shape))     start_index <- start_index + prod(b2_shape)      # Set the weight predictions as the weight variables on the outer model.     dense1$kernel <- w1     dense1$bias <- b1     dense2$kernel <- w2     dense2$bias <- b2      # Inference on the outer model.     preds <- main_network(x)     loss <- loss_fn(y, preds)   })    grads <- tape$gradient(loss, hypernetwork$trainable_weights)   optimizer$apply_gradients(zip_lists(grads, hypernetwork$trainable_weights))   loss }  losses <- c()  # Keep track of the losses over time. coro::loop(for (data in dataset) {   x <- data[[1]]   y <- data[[2]]   loss <- train_step(x, y)    # Logging.   losses[length(losses) + 1] <- as.numeric(loss)   if (length(losses) %% 100 == 0) {     cat(\"Step:\", length(losses), \"Loss:\", mean(losses), \"\\n\")   }   # Stop after 1000 steps.   # Training the model to convergence is left   # as an exercise to the reader.   if (length(losses) >= 1000) {     break   } }) ## Step: 100 Loss: 2.536778 ## Step: 200 Loss: 2.236472 ## Step: 300 Loss: 2.119417 ## Step: 400 Loss: 2.040341 ## Step: 500 Loss: 1.949125 ## Step: 600 Loss: 1.859384 ## Step: 700 Loss: 1.845726 ## Step: 800 Loss: 1.820594 ## Step: 900 Loss: 1.771334 ## Step: 1000 Loss: 1.730648"},{"path":"https://keras3.posit.co/dev/articles/making_new_layers_and_models_via_subclassing.html","id":"introduction","dir":"Articles","previous_headings":"","what":"Introduction","title":"Making new layers and models via subclassing","text":"guide cover everything need know build subclassed layers models. particular, ’ll learn following features: Layer class add_weight() method Trainable non-trainable weights build() method Making sure layers can used backend add_loss() method training argument call() mask argument call() Making sure layers can serialized Let’s dive .","code":""},{"path":"https://keras3.posit.co/dev/articles/making_new_layers_and_models_via_subclassing.html","id":"setup","dir":"Articles","previous_headings":"","what":"Setup","title":"Making new layers and models via subclassing","text":"","code":"library(keras3) library(tensorflow, exclude = c(\"set_random_seed\", \"shape\")) library(tfdatasets, exclude = \"shape\")"},{"path":"https://keras3.posit.co/dev/articles/making_new_layers_and_models_via_subclassing.html","id":"the-layer-class-the-combination-of-state-weights-and-some-computation","dir":"Articles","previous_headings":"","what":"The Layer class: the combination of state (weights) and some computation","title":"Making new layers and models via subclassing","text":"One central abstractions Keras Layer class. layer encapsulates state (layer’s “weights”) transformation inputs outputs (“call”, layer’s forward pass). ’s densely-connected layer. two state variables: variables w b. use layer calling tensor input(s), much like R function. Note weights w b automatically tracked layer upon set layer attributes:","code":"layer_linear <- Layer(\"Linear\",    initialize = function(units = 32, input_dim = 32, ...) {     super$initialize(...)     self$w <- self$add_weight(       shape = shape(input_dim, units),       initializer = \"random_normal\",       trainable = TRUE     )     self$b <- self$add_weight(       shape = shape(units),       initializer = \"zeros\",       trainable = TRUE     )   },    call = function(inputs) {     op_matmul(inputs, self$w) + self$b   } ) x <- op_ones(c(2, 2)) linear_layer <- layer_linear(units = 4, input_dim = 2) y <- linear_layer(x) print(y) ## tf.Tensor( ## [[0.02153057 0.15450525 0.0205495  0.04493225] ##  [0.02153057 0.15450525 0.0205495  0.04493225]], shape=(2, 4), dtype=float32) linear_layer$weights ## [[1]] ## <Variable path=linear/variable, shape=(2, 4), dtype=float32, value=[[-0.06251299  0.05335509  0.01485647 -0.00985784] ##  [ 0.08404355  0.10115016  0.00569303  0.05479009]]> ## ## [[2]] ## <Variable path=linear/variable_1, shape=(4), dtype=float32, value=[0. 0. 0. 0.]>"},{"path":"https://keras3.posit.co/dev/articles/making_new_layers_and_models_via_subclassing.html","id":"layers-can-have-non-trainable-weights","dir":"Articles","previous_headings":"","what":"Layers can have non-trainable weights","title":"Making new layers and models via subclassing","text":"Besides trainable weights, can add non-trainable weights layer well. weights meant taken account backpropagation, training layer. ’s add use non-trainable weight: ’s part layer$weights, gets categorized non-trainable weight:","code":"layer_compute_sum <- Layer(   \"ComputeSum\",   initialize = function(input_dim) {     super$initialize()     self$total <- self$add_weight(       initializer = \"zeros\",       shape = shape(input_dim),       trainable = FALSE     )   },   call = function(inputs) {     self$total$assign_add(op_sum(inputs, axis = 1))     self$total   } )  x <- op_ones(c(2, 2)) my_sum <- layer_compute_sum(input_dim = 2) y <- my_sum(x) print(as.array(y)) ## [1] 2 2 y <- my_sum(x) print(as.array(y)) ## [1] 4 4 cat(\"weights:\", length(my_sum$weights)) ## weights: 1 cat(\"non-trainable weights:\", length(my_sum$non_trainable_weights)) ## non-trainable weights: 1 # It's not included in the trainable weights: cat(\"trainable_weights:\", length(my_sum$trainable_weights)) ## trainable_weights: 0"},{"path":"https://keras3.posit.co/dev/articles/making_new_layers_and_models_via_subclassing.html","id":"best-practice-deferring-weight-creation-until-the-shape-of-the-inputs-is-known","dir":"Articles","previous_headings":"","what":"Best practice: deferring weight creation until the shape of the inputs is known","title":"Making new layers and models via subclassing","text":"Linear layer took input_dim argument used compute shape weights w b initialize(): many cases, may know advance size inputs, like lazily create weights value becomes known, time instantiating layer. Keras API, recommend creating layer weights build(self, inputs_shape) method layer. Like : call() method layer automatically run build first time called. now layer ’s lazy thus easier use: Implementing build() separately shown nicely separates creating weights using weights every call.","code":"layer_linear <- Layer(\"Linear\",    initialize = function(units = 32, input_dim = 32, ...) {     super$initialize(...)     self$w <- self$add_weight(       shape = shape(input_dim, units),       initializer = \"random_normal\",       trainable = TRUE     )     self$b <- self$add_weight(       shape = shape(units),       initializer = \"zeros\",       trainable = TRUE     )   },    call = function(inputs) {     op_matmul(inputs, self$w) + self$b   } ) layer_linear <- Layer(   \"Linear\",   initialize = function(units = 32, ...) {     self$units <- as.integer(units)     super$initialize(...)   },   build = function(input_shape) {     self$w <- self$add_weight(       shape = shape(tail(input_shape, 1), self$units),       initializer = \"random_normal\",       trainable = TRUE     )     self$b <- self$add_weight(       shape = shape(self$units),       initializer = \"zeros\",       trainable = TRUE     )   },   call = function(inputs) {     op_matmul(inputs, self$w) + self$b   } ) # At instantiation, we don't know on what inputs this is going to get called linear_layer <- layer_linear(units = 32)  # The layer's weights are created dynamically the first time the layer is called y <- linear_layer(x)"},{"path":"https://keras3.posit.co/dev/articles/making_new_layers_and_models_via_subclassing.html","id":"layers-are-recursively-composable","dir":"Articles","previous_headings":"","what":"Layers are recursively composable","title":"Making new layers and models via subclassing","text":"assign Layer instance attribute another Layer, outer layer start tracking weights created inner layer. recommend creating sublayers initialize() method leave first call() trigger building weights.","code":"MLPBlock <- Layer(   \"MLPBlock\",   initialize = function() {     super$initialize()     self$linear_1 <- layer_linear(units = 32)     self$linear_2 <- layer_linear(units = 32)     self$linear_3 <- layer_linear(units = 1)   },   call = function(inputs) {     inputs |>       self$linear_1() |>       activation_relu() |>       self$linear_2() |>       activation_relu() |>       self$linear_3()   } )  mlp <- MLPBlock() # The first call to the `mlp` will create the weights y <- mlp(op_ones(shape = c(3, 64)))  cat(\"weights:\", length(mlp$weights), \"\\n\") ## weights: 6 cat(\"trainable weights:\", length(mlp$trainable_weights), \"\\n\") ## trainable weights: 6"},{"path":"https://keras3.posit.co/dev/articles/making_new_layers_and_models_via_subclassing.html","id":"backend-agnostic-layers-and-backend-specific-layers","dir":"Articles","previous_headings":"","what":"Backend-agnostic layers and backend-specific layers","title":"Making new layers and models via subclassing","text":"long layer uses APIs ops namespace (ie. using functions starting op_), (Keras namespaces activations_*, random_*, layer_*), can used backend – TensorFlow, JAX, PyTorch. layers ’ve seen far guide work Keras backends. ops namespace gives access : NumPy API, e.g. op_matmul, op_sum, op_reshape, op_stack, etc. Neural networks-specific APIs op_softmax, op_conv, op_binary_crossentropy, op_relu, etc. can also use backend-native APIs layers (tf$nn functions), , layer usable backend question. instance, write following JAX-specific layer using jax$numpy: equivalent TensorFlow-specific layer: equivalent PyTorch-specific layer: cross-backend compatibility tremendously useful property, strongly recommend seek always make layers backend-agnostic leveraging Keras APIs.","code":"# keras3::install_keras(backend = c(\"jax\")) jax <- reticulate::import(\"jax\")  Linear <- new_layer_class(   ...   call = function(inputs) {     jax$numpy$matmul(inputs, self$w) + self$b   } ) library(tensorflow)  Linear <- new_layer_class(   ...   call = function(inputs) {     tf$matmul(inputs, self$w) + self$b   } ) torch <- reticulate::import(\"torch\")  Linear <- new_layer_class(   ...   call = function(inputs) {     torch$matmul(inputs, self$w) + self$b   } )"},{"path":"https://keras3.posit.co/dev/articles/making_new_layers_and_models_via_subclassing.html","id":"the-add_loss-method","dir":"Articles","previous_headings":"","what":"The add_loss() method","title":"Making new layers and models via subclassing","text":"writing call() method layer, can create loss tensors want use later, writing training loop. doable calling self$add_loss(value): losses (including created inner layer) can retrieved via layer$losses. property reset start every call top-level layer, layer$losses always contains loss values created last forward pass. addition, loss property also contains regularization losses created weights inner layer: losses meant taken account writing custom training loops. also work seamlessly fit() (get automatically summed added main loss, ):","code":"# A layer that creates an activity regularization loss layer_activity_regularization <- Layer(   \"ActivityRegularizationLayer\",   initialize = function(rate = 1e-2) {     self$rate <- as.numeric(rate)     super$initialize()   },   call = function(inputs) {     self$add_loss(self$rate * op_mean(inputs))     inputs   } ) layer_outer <- Layer(   \"OuterLayer\",   initialize = function() {     super$initialize()     self$activity_reg <- layer_activity_regularization(rate = 1e-2)   },   call = function(inputs) {     self$activity_reg(inputs)     inputs   } )  layer <- layer_outer() # No losses yet since the layer has never been called cat(\"losses:\", length(layer$losses), \"\\n\") ## losses: 0 x <- layer(op_zeros(c(1, 1))) # We created one loss value cat(\"losses:\", length(layer$losses), \"\\n\") ## losses: 1 # `layer$losses` gets reset at the start of each call x <- layer(op_zeros(c(1, 1))) # This is the loss created during the call above cat(\"losses:\", length(layer$losses), \"\\n\") ## losses: 1 layer_outer_with_kernel_regularizer <- Layer(   \"OuterLayerWithKernelRegularizer\",   initialize = function() {     super$initialize()     self$dense <- layer_dense(units = 32,                               kernel_regularizer = regularizer_l2(1e-3))   },   call = function(inputs) {     self$dense(inputs)   } )  layer <- layer_outer_with_kernel_regularizer() x <- layer(op_zeros(c(1, 1)))  # This is `1e-3 * sum(layer$dense$kernel ** 2)`, # created by the `kernel_regularizer` above. print(layer$losses) ## [[1]] ## tf.Tensor(0.002025157, shape=(), dtype=float32) inputs <- keras_input(shape = 3) outputs <- inputs |> layer_activity_regularization() model <- keras_model(inputs, outputs)  # If there is a loss passed in `compile`, the regularization # losses get added to it model |> compile(optimizer = \"adam\", loss = \"mse\") model |> fit(random_normal(c(2, 3)), random_normal(c(2, 3)), epochs = 1) ## 1/1 - 0s - 131ms/step - loss: 1.9081 # It's also possible not to pass any loss in `compile`, # since the model already has a loss to minimize, via the `add_loss` # call during the forward pass! model |> compile(optimizer = \"adam\") model |> fit(random_normal(c(2, 3)), random_normal(c(2, 3)), epochs = 1) ## 1/1 - 0s - 106ms/step - loss: 1.6613"},{"path":"https://keras3.posit.co/dev/articles/making_new_layers_and_models_via_subclassing.html","id":"you-can-optionally-enable-serialization-on-your-layers","dir":"Articles","previous_headings":"","what":"You can optionally enable serialization on your layers","title":"Making new layers and models via subclassing","text":"need custom layers serializable part Functional model, can optionally implement get_config() method: Note initialize() method base Layer class takes keyword arguments, particular name dtype. ’s good practice pass arguments parent class initialize() include layer config: need flexibility deserializing layer config, can also override from_config() class method. base implementation from_config(): learn serialization saving, see complete guide saving serializing models.","code":"layer_linear <- Layer(   \"Linear\",   initialize = function(units = 32) {     self$units <- as.integer(units)     super$initialize()   },   build = function(input_shape) {     self$w <- self$add_weight(       shape = shape(tail(input_shape, 1), self$units),       initializer = \"random_normal\",       trainable = TRUE     )     self$b <- self$add_weight(       shape = shape(self$units),       initializer = \"zeros\",       trainable = TRUE     )   },   call = function(inputs) {     op_matmul(inputs, self$w) + self$b   },   get_config = function() {     list(units = self$units)   } )  # Now you can recreate the layer from its config: layer <- layer_linear(units = 64) config <- get_config(layer) str(config) ## List of 1 ##  $ units: int 64 ##  - attr(*, \"__class__\")=<class '<r-globalenv>.Linear'> new_layer <- from_config(config) Linear <- new_layer_class(   \"Linear\",   initialize = function(units = 32, ...) {     self$units <- as.integer(units)     super$initialize(...)   },   build = function(input_shape) {     self$w <- self$add_weight(       shape = shape(tail(input_shape, 1), self$units),       initializer = \"random_normal\",       trainable = TRUE     )     self$b <- self$add_weight(       shape = shape(self$units),       initializer = \"zeros\",       trainable = TRUE     )   },   call = function(inputs) {     op_matmul(inputs, self$w) + self$b   },   get_config = function() {     list(units = self$units)   } )  layer <- Linear(units = 64) config <- get_config(layer) str(config) ## List of 1 ##  $ units: int 64 ##  - attr(*, \"__class__\")=<class '<r-globalenv>.Linear'> new_layer <- from_config(config) Layer(   ...,   from_config = function(config) {     # calling `__class__`() creates a new instance and calls initialize()     do.call(`__class__`, config)   } )"},{"path":"https://keras3.posit.co/dev/articles/making_new_layers_and_models_via_subclassing.html","id":"privileged-training-argument-in-the-call-method","dir":"Articles","previous_headings":"","what":"Privileged training argument in the call() method","title":"Making new layers and models via subclassing","text":"layers, particular BatchNormalization layer Dropout layer, different behaviors training inference. layers, standard practice expose training (boolean) argument call() method. exposing argument call(), enable built-training evaluation loops (e.g. fit()) correctly use layer training inference.","code":"layer_custom_dropout <- Layer(   \"CustomDropout\",   initialize = function(rate, ...) {     super$initialize(...)     self$rate <- rate     self$seed_generator <- random_seed_generator(1337)   },   call = function(inputs, training = NULL) {     if (isTRUE(training))       return(random_dropout(inputs, rate = self$rate,                             seed = self.seed_generator))     inputs   } )"},{"path":"https://keras3.posit.co/dev/articles/making_new_layers_and_models_via_subclassing.html","id":"privileged-mask-argument-in-the-call-method","dir":"Articles","previous_headings":"","what":"Privileged mask argument in the call() method","title":"Making new layers and models via subclassing","text":"privileged argument supported call() mask argument. find Keras RNN layers. mask boolean tensor (one boolean value per timestep input) used skip certain input timesteps processing timeseries data. Keras automatically pass correct mask argument call() layers support , mask generated prior layer. Mask-generating layers Embedding layer configured mask_zero = TRUE, Masking layer.","code":""},{"path":"https://keras3.posit.co/dev/articles/making_new_layers_and_models_via_subclassing.html","id":"the-model-class","dir":"Articles","previous_headings":"","what":"The Model class","title":"Making new layers and models via subclassing","text":"general, use Layer class define inner computation blocks, use Model class define outer model – object train. instance, ResNet50 model, several ResNet blocks subclassing Layer, single Model encompassing entire ResNet50 network. Model class API Layer, following differences: exposes built-training, evaluation, prediction loops (fit(), evaluate(), predict()). exposes list inner layers, via model$layers property. exposes saving serialization APIs (save(), save_weights()…) Effectively, Layer class corresponds refer literature “layer” (“convolution layer” “recurrent layer”) “block” (“ResNet block” “Inception block”). Meanwhile, Model class corresponds referred literature “model” (“deep learning model”) “network” (“deep neural network”). ’re wondering, “use Layer class Model class?”, ask : need call fit() ? need call save() ? , go Model. (either class just block bigger system, writing training & saving code ), use Layer. instance, take mini-resnet example , use build Model train fit(), save save_weights():","code":"ResNet <- Model(   \"ResNet\",   initialize = function(num_classes = 1000, ...) {     super$initialize(...)     self$block_1 <- layer_resnet_block()     self$block_2 <- layer_resnet_block()     self$global_pool <- layer_global_average_pooling_2d()     self$classifier <- layer_dense(num_classes)   },   call = function(inputs) {     inputs |>       self$block_1() |>       self$block_2() |>       self$global_pool() |>       self$classifier()   } )  resnet <- ResNet() dataset <- ... resnet |> fit(dataset, epochs=10) resnet |> save_model(\"filepath.keras\")"},{"path":"https://keras3.posit.co/dev/articles/making_new_layers_and_models_via_subclassing.html","id":"putting-it-all-together-an-end-to-end-example","dir":"Articles","previous_headings":"","what":"Putting it all together: an end-to-end example","title":"Making new layers and models via subclassing","text":"’s ’ve learned far: Layer encapsulate state (created initialize() build()) computation (defined call()). Layers can recursively nested create new, bigger computation blocks. Layers backend-agnostic long use Keras APIs. can use backend-native APIs (jax$numpy, torch$nn tf$nn), layer usable specific backend. Layers can create track losses (typically regularization losses) via add_loss(). outer container, thing want train, Model. Model just like Layer, added training serialization utilities. Let’s put things together end--end example: ’re going implement Variational AutoEncoder (VAE) backend-agnostic fashion – runs TensorFlow, JAX, PyTorch. ’ll train MNIST digits. VAE subclass Model, built nested composition layers subclass Layer. feature regularization loss (KL divergence). Let’s train MNIST using fit() API:","code":"layer_sampling <- Layer(   \"Sampling\",   initialize = function(...) {     super$initialize(...)     self$seed_generator <- random_seed_generator(1337)   },   call = function(inputs) {     c(z_mean, z_log_var) %<-% inputs     batch <- op_shape(z_mean)[[1]]     dim <- op_shape(z_mean)[[2]]     epsilon <- random_normal(shape = c(batch, dim),                              seed=self$seed_generator)     z_mean + op_exp(0.5 * z_log_var) * epsilon   } )  # Maps MNIST digits to a triplet (z_mean, z_log_var, z). layer_encoder <- Layer(   \"Encoder\",   initialize = function(latent_dim = 32, intermediate_dim = 64, ...) {     super$initialize(...)     self$dense_proj <-       layer_dense(units = intermediate_dim,  activation = \"relu\")     self$dense_mean <- layer_dense(units = latent_dim)     self$dense_log_var <- layer_dense(units = latent_dim)     self$sampling <- layer_sampling()   },   call = function(inputs) {     x <- self$dense_proj(inputs)     z_mean <- self$dense_mean(x)     z_log_var <- self$dense_log_var(x)     z <- self$sampling(list(z_mean, z_log_var))     list(z_mean, z_log_var, z)   } )  # Converts z, the encoded digit vector, back into a readable digit. layer_decoder <- Layer(   \"Decoder\",   initialize = function(original_dim, intermediate_dim = 64, ...) {     super$initialize(...)     self$dense_proj <-       layer_dense(units = intermediate_dim, activation = \"relu\")     self$dense_output <-       layer_dense(units = original_dim, activation = \"sigmoid\")   },   call = function(inputs) {     x <- self$dense_proj(inputs)     self$dense_output(x)   } )  # Combines the encoder and decoder into an end-to-end model for training. VariationalAutoEncoder <- Model(   \"VariationalAutoEncoder\",    initialize = function(original_dim, intermediate_dim = 64, latent_dim = 32,                         name = \"autoencoder\", ...) {     super$initialize(name = name, ...)     self$original_dim <- original_dim     self$encoder <- layer_encoder(latent_dim = latent_dim,                             intermediate_dim = intermediate_dim)     self$decoder <- layer_decoder(original_dim = original_dim,                             intermediate_dim = intermediate_dim)   },    call = function(inputs) {     c(z_mean, z_log_var, z) %<-% self$encoder(inputs)     reconstructed <- self$decoder(z)     # Add KL divergence regularization loss.     kl_loss <- -0.5 * op_mean(z_log_var - op_square(z_mean) - op_exp(z_log_var) + 1)     self$add_loss(kl_loss)     reconstructed   } ) c(c(x_train, .), .) %<-% dataset_mnist() x_train <- x_train |>   op_reshape(c(60000, 784)) |>   op_cast(\"float32\") |>   op_divide(255)  original_dim <- 784 vae <- VariationalAutoEncoder(   original_dim = 784,   intermediate_dim = 64,   latent_dim = 32 )  optimizer <- optimizer_adam(learning_rate = 1e-3) vae |> compile(optimizer, loss = loss_mean_squared_error())  vae |> fit(x_train, x_train, epochs = 2, batch_size = 64) ## Epoch 1/2 ## 938/938 - 4s - 4ms/step - loss: 0.0748 ## Epoch 2/2 ## 938/938 - 1s - 1ms/step - loss: 0.0676"},{"path":"https://keras3.posit.co/dev/articles/sequential_model.html","id":"setup","dir":"Articles","previous_headings":"","what":"Setup","title":"The Sequential model","text":"","code":"library(keras3)"},{"path":"https://keras3.posit.co/dev/articles/sequential_model.html","id":"when-to-use-a-sequential-model","dir":"Articles","previous_headings":"","what":"When to use a Sequential model","title":"The Sequential model","text":"Sequential model appropriate plain stack layers layer exactly one input tensor one output tensor. Schematically, following Sequential model: equivalent function: Sequential model appropriate : model multiple inputs multiple outputs layers multiple inputs multiple outputs need layer sharing want non-linear topology (e.g. residual connection, multi-branch model)","code":"model <- keras_model_sequential() |>   layer_dense(units = 2, activation = \"relu\", name = \"layer1\") |>   layer_dense(units = 3, activation = \"relu\", name = \"layer2\") |>   layer_dense(units = 4, name = \"layer3\")  # Call model on a test input x <- op_ones(c(3, 3)) y <- model(x) # Create 3 layers layer1 <- layer_dense(units = 2, activation=\"relu\", name=\"layer1\") layer2 <- layer_dense(units = 3, activation=\"relu\", name=\"layer2\") layer3 <- layer_dense(units = 4, name=\"layer3\")  # Call layers on a test input x <- op_ones(c(3, 3)) y <- x |> layer1() |> layer2() |> layer3()"},{"path":"https://keras3.posit.co/dev/articles/sequential_model.html","id":"creating-a-sequential-model","dir":"Articles","previous_headings":"","what":"Creating a Sequential model","title":"The Sequential model","text":"can create Sequential model piping layers keras_model_sequential() object: passing list layers keras_model_sequential(): layers accessible via layers attribute: can also create Sequential model incrementally: Note ’s also corresponding pop_layer() method remove layers: Sequential model behaves much like stack layers. Also note Sequential constructor accepts name argument, just like layer model Keras. useful annotate TensorBoard graphs semantically meaningful names.","code":"model <- keras_model_sequential() |>   layer_dense(units = 2, activation = \"relu\") |>   layer_dense(units = 3, activation = \"relu\") |>   layer_dense(units = 4) model <- keras_model_sequential(layers = list(   layer_dense(units = 2, activation = \"relu\"),   layer_dense(units = 3, activation = \"relu\"),   layer_dense(units = 4) )) model$layers ## [[1]] ## <Dense name=dense_3, built=False> ##  signature: (*args, **kwargs) ## ## [[2]] ## <Dense name=dense_4, built=False> ##  signature: (*args, **kwargs) ## ## [[3]] ## <Dense name=dense_5, built=False> ##  signature: (*args, **kwargs) model <- keras_model_sequential() model |> layer_dense(units = 2, activation=\"relu\") model |> layer_dense(units = 3, activation=\"relu\") model |> layer_dense(units = 4) model |> pop_layer() ## <Dense name=dense_8, built=False> ##  signature: (*args, **kwargs) length(model$layers)  # 2 ## [1] 2 model <- keras_model_sequential(name = \"my_sequential\") model |> layer_dense(units = 2, activation=\"relu\", name = \"layer1\") model |> layer_dense(units = 3, activation=\"relu\", name = \"layer2\") model |> layer_dense(units = 4, name = \"layer3\")"},{"path":"https://keras3.posit.co/dev/articles/sequential_model.html","id":"specifying-the-input-shape-in-advance","dir":"Articles","previous_headings":"","what":"Specifying the input shape in advance","title":"The Sequential model","text":"Generally, layers Keras need know shape inputs order able create weights. create layer like , initially, weights: creates weights first time called input, since shape weights depends shape inputs: Naturally, also applies Sequential models. instantiate Sequential model without input shape, isn’t “built”: weights (calling model$weights results error stating just ). weights created model first sees input data: model “built”, can call summary() method display contents: However, can useful building Sequential model incrementally able display summary model far, including current output shape. case, start model passing input_shape argument model, knows input shape start: Models built predefined input shape like always weights (even seeing data) always defined output shape. general, ’s recommended best practice always specify input shape Sequential model advance know .","code":"layer <- layer_dense(units = 3) layer$weights  # Empty ## list() # Call layer on a test input x <- op_ones(c(1, 4)) y <- layer(x) layer$weights  # Now it has weights, of shape (4, 3) and (3,) ## [[1]] ## <Variable path=dense_9/kernel, shape=(4, 3), dtype=float32, value=[[ 0.48581433  0.78749573  0.61015   ] ##  [ 0.7962619   0.7261175  -0.8046875 ] ##  [-0.6189915   0.37973273  0.50559556] ##  [-0.5455791  -0.60714126  0.19791973]]> ## ## [[2]] ## <Variable path=dense_9/bias, shape=(3), dtype=float32, value=[0. 0. 0.]> model <- keras_model_sequential() |>   layer_dense(units = 2, activation = \"relu\") |>   layer_dense(units = 3, activation = \"relu\") |>   layer_dense(units = 4) # No weights at this stage!  # At this point, you can't do this: # model$weights   # Call the model on a test input x <- op_ones(c(1, 4)) y <- model(x) length(model$weights) ## [1] 6 summary(model) ## Model: \"sequential_4\" ## ┏━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━┓ ## ┃ Layer (type)                    ┃ Output Shape           ┃       Param # ┃ ## ┡━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━┩ ## │ dense_10 (Dense)                │ (1, 2)                 │            10 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ dense_11 (Dense)                │ (1, 3)                 │             9 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ dense_12 (Dense)                │ (1, 4)                 │            16 │ ## └─────────────────────────────────┴────────────────────────┴───────────────┘ ##  Total params: 35 (140.00 B) ##  Trainable params: 35 (140.00 B) ##  Non-trainable params: 0 (0.00 B) model <- keras_model_sequential(input_shape = 4) |>   layer_dense(units = 2, activation = \"relu\") summary(model) ## Model: \"sequential_5\" ## ┏━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━┓ ## ┃ Layer (type)                    ┃ Output Shape           ┃       Param # ┃ ## ┡━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━┩ ## │ dense_13 (Dense)                │ (None, 2)              │            10 │ ## └─────────────────────────────────┴────────────────────────┴───────────────┘ ##  Total params: 10 (40.00 B) ##  Trainable params: 10 (40.00 B) ##  Non-trainable params: 0 (0.00 B) model$layers ## [[1]] ## <Dense name=dense_13, built=True> ##  signature: (*args, **kwargs)"},{"path":"https://keras3.posit.co/dev/articles/sequential_model.html","id":"a-common-debugging-workflow-add-layers-summary","dir":"Articles","previous_headings":"","what":"A common debugging workflow: add layers + summary()","title":"The Sequential model","text":"building new Sequential architecture, ’s useful incrementally stack layers |> frequently print model summaries. instance, enables monitor stack Conv2D MaxPooling2D layers downsampling image feature maps: practical, right? Note |> equivalent calling model$add(), modifies model -place, don’t need reassign model symbol step.","code":"model <- keras_model_sequential(input_shape = c(250, 250, 3)) |>   layer_conv_2d(filters = 32, kernel_size = 5, strides = 2, activation = \"relu\") |>   layer_conv_2d(filters = 32, kernel_size = 3, activation = \"relu\") |>   layer_max_pooling_2d(pool_size = c(3, 3))  # Can you guess what the current output shape is at this point? Probably not. # Let's just print it: summary(model) ## Model: \"sequential_6\" ## ┏━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━┓ ## ┃ Layer (type)                    ┃ Output Shape           ┃       Param # ┃ ## ┡━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━┩ ## │ conv2d (Conv2D)                 │ (None, 123, 123, 32)   │         2,432 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ conv2d_1 (Conv2D)               │ (None, 121, 121, 32)   │         9,248 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ max_pooling2d (MaxPooling2D)    │ (None, 40, 40, 32)     │             0 │ ## └─────────────────────────────────┴────────────────────────┴───────────────┘ ##  Total params: 11,680 (45.62 KB) ##  Trainable params: 11,680 (45.62 KB) ##  Non-trainable params: 0 (0.00 B) # The answer was: (40, 40, 32), so we can keep downsampling...  model |>   layer_conv_2d(filters = 32, kernel_size = 3, activation = \"relu\") |>   layer_conv_2d(filters = 32, kernel_size = 3, activation = \"relu\") |>   layer_max_pooling_2d(pool_size = 3) |>   layer_conv_2d(filters = 32, kernel_size = 3, activation = \"relu\") |>   layer_conv_2d(filters = 32, kernel_size = 3, activation = \"relu\") |>   layer_max_pooling_2d(pool_size = 2)  # And now? summary(model) ## Model: \"sequential_6\" ## ┏━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━┓ ## ┃ Layer (type)                    ┃ Output Shape           ┃       Param # ┃ ## ┡━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━┩ ## │ conv2d (Conv2D)                 │ (None, 123, 123, 32)   │         2,432 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ conv2d_1 (Conv2D)               │ (None, 121, 121, 32)   │         9,248 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ max_pooling2d (MaxPooling2D)    │ (None, 40, 40, 32)     │             0 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ conv2d_2 (Conv2D)               │ (None, 38, 38, 32)     │         9,248 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ conv2d_3 (Conv2D)               │ (None, 36, 36, 32)     │         9,248 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ max_pooling2d_1 (MaxPooling2D)  │ (None, 12, 12, 32)     │             0 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ conv2d_4 (Conv2D)               │ (None, 10, 10, 32)     │         9,248 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ conv2d_5 (Conv2D)               │ (None, 8, 8, 32)       │         9,248 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ max_pooling2d_2 (MaxPooling2D)  │ (None, 4, 4, 32)       │             0 │ ## └─────────────────────────────────┴────────────────────────┴───────────────┘ ##  Total params: 48,672 (190.12 KB) ##  Trainable params: 48,672 (190.12 KB) ##  Non-trainable params: 0 (0.00 B) # Now that we have 4x4 feature maps, time to apply global max pooling. model |>   layer_global_max_pooling_2d()  # Finally, we add a classification layer. model |>   layer_dense(units = 10, activation = \"softmax\")"},{"path":"https://keras3.posit.co/dev/articles/sequential_model.html","id":"what-to-do-once-you-have-a-model","dir":"Articles","previous_headings":"","what":"What to do once you have a model","title":"The Sequential model","text":"model architecture ready, want : Train model, evaluate , run inference. See guide training & evaluation built-loops Save model disk restore . See guide serialization & saving.","code":""},{"path":"https://keras3.posit.co/dev/articles/sequential_model.html","id":"feature-extraction-with-a-sequential-model","dir":"Articles","previous_headings":"","what":"Feature extraction with a Sequential model","title":"The Sequential model","text":"Sequential model built, behaves like Functional API model. means every layer input output attribute. attributes can used neat things, like quickly creating model extracts outputs intermediate layers Sequential model: ’s similar example extract features one layer:","code":"initial_model <- keras_model_sequential(input_shape = c(250, 250, 3)) |>   layer_conv_2d(filters = 32, kernel_size = 5, strides = 2, activation = \"relu\") |>   layer_conv_2d(filters = 32, kernel_size = 3, activation = \"relu\") |>   layer_conv_2d(filters = 32, kernel_size = 3, activation = \"relu\")   feature_extractor <- keras_model(     inputs = initial_model$inputs,     outputs = lapply(initial_model$layers, function(x) x$output), )  # Call feature extractor on test input. x <- op_ones(c(1, 250, 250, 3)) features <- feature_extractor(x) initial_model <-   keras_model_sequential(input_shape = c(250, 250, 3)) |>   layer_conv_2d(filters = 32, kernel_size = 5, strides = 2,                 activation = \"relu\") |>   layer_conv_2d(filters = 32, kernel_size = 3, activation = \"relu\",                 name = \"my_intermediate_layer\") |>   layer_conv_2d(filters = 32, kernel_size = 3, activation = \"relu\")  feature_extractor <- keras_model(   inputs = initial_model$inputs,   outputs = get_layer(initial_model, \"my_intermediate_layer\")$output, )  # Call feature extractor on test input. x <- op_ones(c(1, 250, 250, 3)) features <- feature_extractor(x)"},{"path":"https://keras3.posit.co/dev/articles/sequential_model.html","id":"transfer-learning-with-a-sequential-model","dir":"Articles","previous_headings":"","what":"Transfer learning with a Sequential model","title":"The Sequential model","text":"Transfer learning consists freezing bottom layers model training top layers. aren’t familiar , make sure read guide transfer learning. two common transfer learning blueprint involving Sequential models. First, let’s say Sequential model, want freeze layers except last one. case, can call freeze_weights(). Alternatively, can iterate model$layers set layer$trainable <- FALSE layer, except last one. Like : Another common blueprint use Sequential model stack pre-trained model freshly initialized classification layers. Like : transfer learning, probably find frequently using two patterns. ’s need know Sequential models! find building models Keras, see: Guide Functional API Guide making new Layers & Models via subclassing","code":"model <- keras_model_sequential(input_shape = 784) |>   layer_dense(units = 32, activation = \"relu\") |>   layer_dense(units = 32, activation = \"relu\") |>   layer_dense(units = 32, activation = \"relu\") |>   layer_dense(units = 10) # Presumably you would want to first load pre-trained weights. model |> load_model_weights(...) # Freeze all layers except the last one. model |> freeze_weights(from = 1, to = -2) model # note the \"Trainable\" column now visible in the summary table ## Model: \"sequential_9\" ## ┏━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━┳━━━━━━━┓ ## ┃ Layer (type)                ┃ Output Shape          ┃    Param # ┃ Trai… ┃ ## ┡━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━╇━━━━━━━┩ ## │ dense_15 (Dense)            │ (None, 32)            │     25,120 │   N   │ ## ├─────────────────────────────┼───────────────────────┼────────────┼───────┤ ## │ dense_16 (Dense)            │ (None, 32)            │      1,056 │   N   │ ## ├─────────────────────────────┼───────────────────────┼────────────┼───────┤ ## │ dense_17 (Dense)            │ (None, 32)            │      1,056 │   N   │ ## ├─────────────────────────────┼───────────────────────┼────────────┼───────┤ ## │ dense_18 (Dense)            │ (None, 10)            │        330 │   Y   │ ## └─────────────────────────────┴───────────────────────┴────────────┴───────┘ ##  Total params: 27,562 (107.66 KB) ##  Trainable params: 330 (1.29 KB) ##  Non-trainable params: 27,232 (106.38 KB) # Another way to freeze all layers except the last one. for (layer in model$layers[-length(model$layers)]) {   layer$trainable <- FALSE }  # Recompile and train (this will only update the weights of the last layer). model |> compile(...) model |> fit(...) # Load a convolutional base with pre-trained weights base_model <- application_xception(weights = 'imagenet',                                    include_top = FALSE,                                    pooling = 'avg')  # Freeze the base model freeze_weights(base_model)  # Use a Sequential model to add a trainable classifier on top model <- keras_model_sequential() |>   base_model() |>   layer_dense(1000) # Compile & train model |> compile(...) model |> fit(...)"},{"path":"https://keras3.posit.co/dev/articles/serialization_and_saving.html","id":"introduction","dir":"Articles","previous_headings":"","what":"Introduction","title":"Save, serialize, and export models","text":"Keras model consists multiple components: architecture, configuration, specifies layers model contain, ’re connected. set weights values (“state model”). optimizer (defined compiling model). set losses metrics (defined compiling model). Keras API saves pieces together unified format, marked .keras extension. zip archive consisting following: JSON-based configuration file (config.json): Records model, layer, trackables’ configuration. H5-based state file, model.weights.h5 (whole model), directory keys layers weights. metadata file JSON, storing things current Keras version. Let’s take look works.","code":""},{"path":"https://keras3.posit.co/dev/articles/serialization_and_saving.html","id":"how-to-save-and-load-a-model","dir":"Articles","previous_headings":"","what":"How to save and load a model","title":"Save, serialize, and export models","text":"10 seconds read guide, ’s need know. Saving Keras model: Loading model back: Now, let’s look details.","code":"# Get model (Sequential, Functional Model, or Model subclass) model <- ...  # The filename needs to end with the .keras extension model |> save_model('path/to/location.keras') model <- load_model('path/to/location.keras')"},{"path":"https://keras3.posit.co/dev/articles/serialization_and_saving.html","id":"setup","dir":"Articles","previous_headings":"","what":"Setup","title":"Save, serialize, and export models","text":"","code":"library(keras3)"},{"path":"https://keras3.posit.co/dev/articles/serialization_and_saving.html","id":"saving","dir":"Articles","previous_headings":"","what":"Saving","title":"Save, serialize, and export models","text":"section saving entire model single file. file include: model’s architecture/config model’s weight values (learned training) model’s compilation information (compile() called) optimizer state, (enables restart training left)","code":""},{"path":"https://keras3.posit.co/dev/articles/serialization_and_saving.html","id":"apis","dir":"Articles","previous_headings":"Saving","what":"APIs","title":"Save, serialize, and export models","text":"can save model save_model(). can load back load_model(). supported format Keras 3 “Keras v3” format, uses .keras extension. Example:","code":"get_model <- function() {   # Create a simple model.   inputs <- keras_input(shape(32))   outputs <- inputs |> layer_dense(1)   model <-  keras_model(inputs, outputs)   model |> compile(optimizer = optimizer_adam(), loss = \"mean_squared_error\")   model }  model <- get_model()  # Train the model. test_input <- random_uniform(c(128, 32)) test_target <- random_uniform(c(128, 1)) model |> fit(test_input, test_target)  # Calling `save('my_model.keras')` creates a zip archive `my_model.keras`. model |> save_model(\"my_model.keras\")  # It can be used to reconstruct the model identically. reconstructed_model <- load_model(\"my_model.keras\")  # Let's check: stopifnot(all.equal(   model |> predict(test_input),   reconstructed_model |> predict(test_input) ))"},{"path":"https://keras3.posit.co/dev/articles/serialization_and_saving.html","id":"custom-objects","dir":"Articles","previous_headings":"Saving","what":"Custom objects","title":"Save, serialize, and export models","text":"section covers basic workflows handling custom layers, functions, models Keras saving reloading. saving model includes custom objects, subclassed Layer, must define get_config() method object class. arguments passed constructor (initialize() method) custom object aren’t simple objects (anything types like ints, strings, etc.), must also explicitly deserialize arguments from_config() class method. Like : Please see Defining config methods section details examples. saved .keras file lightweight store Python code custom objects. Therefore, reload model, load_model requires access definition custom objects used one following methods: Registering custom objects (preferred), Passing custom objects directly loading, Using custom object scope examples workflow:","code":"layer_custom <- Layer(   \"CustomLayer\",   initialize = function(sublayer, ...) {     super$initialize(...)     self$sublayer <- sublayer   },   call = function(x) {     self$sublayer(x)   },   get_config = function() {     base_config <- super$get_config()     config <- list(       sublayer = serialize_keras_object(self$sublayer)     )     c(base_config, config)   },   from_config = function(cls, config) {     sublayer_config <- config$sublayer     sublayer <- deserialize_keras_object(sublayer_config)     cls(sublayer, !!!config)   } )"},{"path":"https://keras3.posit.co/dev/articles/serialization_and_saving.html","id":"registering-custom-objects-preferred","dir":"Articles","previous_headings":"Saving > Custom objects","what":"Registering custom objects (preferred)","title":"Save, serialize, and export models","text":"preferred method, custom object registration greatly simplifies saving loading code. Calling register_keras_serializable() custom object registers object globally master list, allowing Keras recognize object loading model. Let’s create custom model involving custom layer custom activation function demonstrate . Example:","code":"# Clear all previously registered custom objects set_custom_objects(clear = TRUE) ## named list() layer_custom <- Layer(   \"CustomLayer\",   initialize = function(self, factor) {     super$initialize()     self$factor = factor   },    call = function(self, x) {     x * self$factor   },    get_config = function(self) {     list(factor = self$factor)   } )  # Upon registration, you can optionally specify a package or a name. # If left blank, the package defaults to \"Custom\" and the name defaults to # the class name. register_keras_serializable(layer_custom, package = \"MyLayers\")  custom_fn <- keras3:::py_func2(function(x) x^2, name = \"custom_fn\", convert = TRUE)  register_keras_serializable(custom_fn, name=\"custom_fn\", package=\"my_package\")   # Create the model. get_model <- function() {   inputs <- keras_input(shape(4))   mid <- inputs |> layer_custom(0.5)   outputs <- mid |> layer_dense(1, activation = custom_fn)   model <- keras_model(inputs, outputs)   model |> compile(optimizer = \"rmsprop\", loss = \"mean_squared_error\")   model }   # Train the model. train_model <- function(model) {   input <- random_uniform(c(4, 4))   target <- random_uniform(c(4, 1))   model |> fit(input, target, verbose = FALSE, epochs = 1)   model }  test_input <- random_uniform(c(4, 4)) test_target <- random_uniform(c(4, 1))  model <- get_model() |> train_model() model |> save_model(\"custom_model.keras\", overwrite = TRUE)  # Now, we can simply load without worrying about our custom objects. reconstructed_model <- load_model(\"custom_model.keras\")  # Let's check: stopifnot(all.equal(   model |> predict(test_input, verbose = FALSE),   reconstructed_model |> predict(test_input, verbose = FALSE) ))"},{"path":"https://keras3.posit.co/dev/articles/serialization_and_saving.html","id":"passing-custom-objects-to-load_model","dir":"Articles","previous_headings":"Saving > Custom objects","what":"Passing custom objects to load_model()","title":"Save, serialize, and export models","text":"","code":"model <- get_model() |> train_model()  # Calling `save_model('my_model.keras')` creates a zip archive `my_model.keras`. model |> save_model(\"custom_model.keras\", overwrite = TRUE)  # Upon loading, pass a named list containing the custom objects used in the # `custom_objects` argument of `load_model()`. reconstructed_model <-  load_model(  \"custom_model.keras\",   custom_objects = list(CustomLayer = layer_custom,                         custom_fn = custom_fn), )  # Let's check: stopifnot(all.equal(   model |> predict(test_input, verbose = FALSE),   reconstructed_model |> predict(test_input, verbose = FALSE) ))"},{"path":"https://keras3.posit.co/dev/articles/serialization_and_saving.html","id":"using-a-custom-object-scope","dir":"Articles","previous_headings":"Saving > Custom objects","what":"Using a custom object scope","title":"Save, serialize, and export models","text":"code within custom object scope able recognize custom objects passed scope argument. Therefore, loading model within scope allow loading custom objects. Example:","code":"model <- get_model() |> train_model() model |> save_model(\"custom_model.keras\", overwrite = TRUE)  # Pass the custom objects dictionary to a custom object scope and place # the `keras.models.load_model()` call within the scope. custom_objects <- list(CustomLayer = layer_custom, custom_fn = custom_fn)  with_custom_object_scope(custom_objects, {   reconstructed_model <- load_model(\"custom_model.keras\") })  # Let's check: stopifnot(all.equal(   model |> predict(test_input, verbose = FALSE),   reconstructed_model |> predict(test_input, verbose = FALSE) ))"},{"path":"https://keras3.posit.co/dev/articles/serialization_and_saving.html","id":"model-serialization","dir":"Articles","previous_headings":"Saving","what":"Model serialization","title":"Save, serialize, and export models","text":"section saving model’s configuration, without state. model’s configuration (architecture) specifies layers model contains, layers connected. configuration model, model can created freshly initialized state (weights compilation information).","code":""},{"path":"https://keras3.posit.co/dev/articles/serialization_and_saving.html","id":"apis-1","dir":"Articles","previous_headings":"Saving > Model serialization","what":"APIs","title":"Save, serialize, and export models","text":"following serialization APIs available: clone_model(model): make (randomly initialized) copy model. get_config() cls.from_config(): retrieve configuration layer model, recreate model instance config, respectively. keras.models.model_to_json() keras.models.model_from_json(): similar, JSON strings. keras.saving.serialize_keras_object(): retrieve configuration arbitrary Keras object. keras.saving.deserialize_keras_object(): recreate object instance configuration.","code":""},{"path":"https://keras3.posit.co/dev/articles/serialization_and_saving.html","id":"in-memory-model-cloning","dir":"Articles","previous_headings":"Saving > Model serialization","what":"In-memory model cloning","title":"Save, serialize, and export models","text":"can -memory cloning model via clone_model(). equivalent getting config recreating model config (preserve compilation information layer weights values). Example:","code":"new_model <- clone_model(model)"},{"path":"https://keras3.posit.co/dev/articles/serialization_and_saving.html","id":"get_config-and-from_config","dir":"Articles","previous_headings":"Saving > Model serialization","what":"get_config() and from_config()","title":"Save, serialize, and export models","text":"Calling get_config(model) get_config(layer) return named list containing configuration model layer, respectively. define get_config() contain arguments needed initialize() method model layer. loading time, from_config(config) method call initialize() arguments reconstruct model layer. Layer example: Now let’s reconstruct layer using from_config() method: Sequential model example: Functional model example:","code":"layer <- layer_dense(, 3, activation=\"relu\") layer_config <- get_config(layer) str(layer_config) ## List of 12 ##  $ name              : chr \"dense_4\" ##  $ trainable         : logi TRUE ##  $ dtype             :List of 4 ##   ..$ module         : chr \"keras\" ##   ..$ class_name     : chr \"DTypePolicy\" ##   ..$ config         :List of 1 ##   .. ..$ name: chr \"float32\" ##   ..$ registered_name: NULL ##  $ units             : int 3 ##  $ activation        : chr \"relu\" ##  $ use_bias          : logi TRUE ##  $ kernel_initializer:List of 4 ##   ..$ module         : chr \"keras.initializers\" ##   ..$ class_name     : chr \"GlorotUniform\" ##   ..$ config         :List of 1 ##   .. ..$ seed: NULL ##   ..$ registered_name: NULL ##  $ bias_initializer  :List of 4 ##   ..$ module         : chr \"keras.initializers\" ##   ..$ class_name     : chr \"Zeros\" ##   ..$ config         : Named list() ##   ..$ registered_name: NULL ##  $ kernel_regularizer: NULL ##  $ bias_regularizer  : NULL ##  $ kernel_constraint : NULL ##  $ bias_constraint   : NULL ##  - attr(*, \"__class__\")=<class 'keras.src.layers.core.dense.Dense'> new_layer <- from_config(layer_config) model <- keras_model_sequential(input_shape = c(32)) |>   layer_dense(1) config <- get_config(model) new_model <- from_config(config) inputs <- keras_input(c(32)) outputs <- inputs |> layer_dense(1) model <- keras_model(inputs, outputs) config <- get_config(model) new_model <- from_config(config)"},{"path":"https://keras3.posit.co/dev/articles/serialization_and_saving.html","id":"save_model_config-and-load_model_config","dir":"Articles","previous_headings":"Saving > Model serialization","what":"save_model_config() and load_model_config()","title":"Save, serialize, and export models","text":"similar get_config / from_config, except turns model JSON file, can loaded without original model class. also specific models, isn’t meant layers. Example:","code":"model <- keras_model_sequential(input_shape = c(32)) |>   layer_dense(1) save_model_config(model, \"model_config.json\") new_model <- load_model_config(\"model_config.json\") unlink(\"model_config.json\")"},{"path":"https://keras3.posit.co/dev/articles/serialization_and_saving.html","id":"arbitrary-object-serialization-and-deserialization","dir":"Articles","previous_headings":"Saving > Model serialization","what":"Arbitrary object serialization and deserialization","title":"Save, serialize, and export models","text":"serialize_keras_object() deserialize_keras_object() APIs general-purpose APIs can used serialize deserialize Keras object custom object. foundation saving model architecture behind serialize()/deserialize() calls keras. Example: Note serialization format containing necessary information proper reconstruction: module containing name Keras module identifying module object comes class_name containing name object’s class. config information needed reconstruct object registered_name custom objects. See . Now can reconstruct regularizer.","code":"my_reg <- regularizer_l1(0.005) config <- serialize_keras_object(my_reg) str(config) ## List of 4 ##  $ module         : chr \"keras.regularizers\" ##  $ class_name     : chr \"L1\" ##  $ config         :List of 1 ##   ..$ l1: num 0.005 ##  $ registered_name: NULL new_reg <- deserialize_keras_object(config) new_reg ## <keras.src.regularizers.regularizers.L1 object> ##  signature: (x)"},{"path":"https://keras3.posit.co/dev/articles/serialization_and_saving.html","id":"model-weights-saving","dir":"Articles","previous_headings":"Saving","what":"Model weights saving","title":"Save, serialize, and export models","text":"can choose save & load model’s weights. can useful : need model inference: case won’t need restart training, don’t need compilation information optimizer state. transfer learning: case training new model reusing state prior model, don’t need compilation information prior model.","code":""},{"path":"https://keras3.posit.co/dev/articles/serialization_and_saving.html","id":"apis-for-in-memory-weight-transfer","dir":"Articles","previous_headings":"Saving > Model weights saving","what":"APIs for in-memory weight transfer","title":"Save, serialize, and export models","text":"Weights can copied different objects using get_weights() set_weights(): get_weights(<layer>): Returns list arrays weight values. set_weights(<layer>weights): Sets model/layer weights values provided (arrays). Examples: Transferring weights one layer another, memory Transferring weights one model another model compatible architecture, memory case stateless layers stateless layers change order number weights, models can compatible architectures even extra/missing stateless layers.","code":"create_layer <- function() {   layer <- layer_dense(, 64, activation = \"relu\", name = \"dense_2\")   layer$build(shape(NA, 784))   layer }  layer_1 <- create_layer() layer_2 <- create_layer()  # Copy weights from layer 1 to layer 2 layer_2 |> set_weights(get_weights(layer_1)) # Create a simple functional model inputs <- keras_input(shape=c(784), name=\"digits\") outputs <- inputs |>   layer_dense(64, activation = \"relu\", name = \"dense_1\") |>   layer_dense(64, activation = \"relu\", name = \"dense_2\") |>   layer_dense(10, name = \"predictions\") functional_model <- keras_model(inputs = inputs, outputs = outputs,                                name = \"3_layer_mlp\")  # Define a subclassed model with the same architecture SubclassedModel <- new_model_class(   \"SubclassedModel\",   initialize = function(output_dim, name = NULL) {     super$initialize(name = name)     self$output_dim <- output_dim |> as.integer()     self$dense_1 <- layer_dense(, 64, activation = \"relu\",                                 name = \"dense_1\")     self$dense_2 <- layer_dense(, 64, activation = \"relu\",                                 name = \"dense_2\")     self$dense_3 <- layer_dense(, self$output_dim,                                 name = \"predictions\")   },    call = function(inputs) {     inputs |>       self$dense_1() |>       self$dense_2() |>       self$dense_3()   },    get_config = function(self) {     list(output_dim = self$output_dim,          name = self$name)   } )   subclassed_model <- SubclassedModel(10) # Call the subclassed model once to create the weights. subclassed_model(op_ones(c(1, 784))) |> invisible()  # Copy weights from functional_model to subclassed_model. set_weights(subclassed_model, get_weights(functional_model))  stopifnot(all.equal(   get_weights(functional_model),   get_weights(subclassed_model) )) input <- keras_input(shape = c(784), name = \"digits\") output <- input |>   layer_dense(64, activation = \"relu\", name = \"dense_1\") |>   layer_dense(64, activation = \"relu\", name = \"dense_2\") |>   layer_dense(10, name = \"predictions\") functional_model <- keras_model(inputs, outputs,                                 name = \"3_layer_mlp\")  input <- keras_input(shape = c(784), name = \"digits\") output <- input |>   layer_dense(64, activation = \"relu\", name = \"dense_1\") |>   layer_dense(64, activation = \"relu\", name = \"dense_2\") |>   # Add a dropout layer, which does not contain any weights.   layer_dropout(0.5) |>   layer_dense(10, name = \"predictions\")  functional_model_with_dropout <-   keras_model(input, output, name = \"3_layer_mlp\")  set_weights(functional_model_with_dropout,             get_weights(functional_model))"},{"path":"https://keras3.posit.co/dev/articles/serialization_and_saving.html","id":"apis-for-saving-weights-to-disk-loading-them-back","dir":"Articles","previous_headings":"Saving > Model weights saving","what":"APIs for saving weights to disk & loading them back","title":"Save, serialize, and export models","text":"Weights can saved disk calling save_model_weights(filepath). filename end .weights.h5. Example: Note using freeze_weights() may result different output get_weights(layer) ordering model contains nested layers.","code":"sequential_model = keras_model_sequential(input_shape = c(784),                                           input_name = \"digits\") |>   layer_dense(64, activation = \"relu\", name = \"dense_1\") |>   layer_dense(64, activation = \"relu\", name = \"dense_2\") |>   layer_dense(10, name = \"predictions\") sequential_model |> save_model_weights(\"my_model.weights.h5\") sequential_model |> load_model_weights(\"my_model.weights.h5\")"},{"path":"https://keras3.posit.co/dev/articles/serialization_and_saving.html","id":"transfer-learning-example","dir":"Articles","previous_headings":"Saving > Model weights saving > APIs for saving weights to disk & loading them back","what":"Transfer learning example","title":"Save, serialize, and export models","text":"loading pretrained weights weights file, recommended load weights original checkpointed model, extract desired weights/layers new model. Example:","code":"create_functional_model <- function() {   inputs <- keras_input(shape = c(784), name = \"digits\")   outputs <- inputs |>     layer_dense(64, activation = \"relu\", name = \"dense_1\") |>     layer_dense(64, activation = \"relu\", name = \"dense_2\") |>     layer_dense(10, name = \"predictions\")   keras_model(inputs, outputs, name = \"3_layer_mlp\") }  functional_model <- create_functional_model() functional_model |> save_model_weights(\"pretrained.weights.h5\")  # In a separate program: pretrained_model <- create_functional_model() pretrained_model |> load_model_weights(\"pretrained.weights.h5\")  # Create a new model by extracting layers from the original model: extracted_layers <- pretrained_model$layers |> head(-1) model <- keras_model_sequential(layers = extracted_layers) |>   layer_dense(5, name = \"dense_3\") summary(model) ## Model: \"sequential_4\" ## ┏━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━┓ ## ┃ Layer (type)                    ┃ Output Shape           ┃       Param # ┃ ## ┡━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━┩ ## │ dense_1 (Dense)                 │ (None, 64)             │        50,240 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ dense_2 (Dense)                 │ (None, 64)             │         4,160 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ dense_3 (Dense)                 │ (None, 5)              │           325 │ ## └─────────────────────────────────┴────────────────────────┴───────────────┘ ##  Total params: 54,725 (213.77 KB) ##  Trainable params: 54,725 (213.77 KB) ##  Non-trainable params: 0 (0.00 B)"},{"path":[]},{"path":"https://keras3.posit.co/dev/articles/serialization_and_saving.html","id":"defining-the-config-methods","dir":"Articles","previous_headings":"Saving > Appendix: Handling custom objects","what":"Defining the config methods","title":"Save, serialize, and export models","text":"Specifications: get_config() return JSON-serializable named list order compatible Keras architecture model-saving APIs. from_config(config) (class method) return new layer model object created config. default implementation returns .call(cls, config). NOTE: constructor arguments already serializable, e.g. strings ints, non-custom Keras objects, overriding from_config() necessary. However, complex objects layers models passed initialize(), deserialization must handled explicitly either initialize overriding from_config() method. Example: Note overriding from_config unnecessary MyDense hidden_units, kernel_initializer, kernel_regularizer ints, strings, built-Keras object, respectively. means default from_config implementation cls(!!!config) work intended. complex objects, layers models passed initialize(), example, must explicitly deserialize objects. Let’s take look example model from_config override necessary. Example:","code":"layer_my_dense <- register_keras_serializable(   package = \"MyLayers\", name = \"KernelMult\",   object = Layer(     \"MyDense\",     initialize = function(units,                           ...,                           kernel_regularizer = NULL,                           kernel_initializer = NULL,                           nested_model = NULL) {       super$initialize(...)       self$hidden_units <- units       self$kernel_regularizer <- kernel_regularizer       self$kernel_initializer <- kernel_initializer       self$nested_model <- nested_model     },     get_config = function() {       config <- super$get_config()       # Update the config with the custom layer's parameters       config <- modifyList(config, list(         units = self$hidden_units,         kernel_regularizer = self$kernel_regularizer,         kernel_initializer = self$kernel_initializer,         nested_model = self$nested_model       ))       config     },     build = function(input_shape) {       input_units <- tail(input_shape, 1)       self$kernel <- self$add_weight(         name = \"kernel\",         shape = shape(input_units, self$hidden_units),         regularizer = self$kernel_regularizer,         initializer = self$kernel_initializer,       )     },     call = function(inputs) {       op_matmul(inputs, self$kernel)     }   ) )   layer <- layer_my_dense(units = 16,                         kernel_regularizer = \"l1\",                         kernel_initializer = \"ones\") layer3 <- layer_my_dense(units = 64, nested_model = layer)  config <- serialize_keras_object(layer3) str(config) ## List of 4 ##  $ module         : chr \"<r-globalenv>\" ##  $ class_name     : chr \"MyDense\" ##  $ config         :List of 5 ##   ..$ name        : chr \"my_dense_1\" ##   ..$ trainable   : logi TRUE ##   ..$ dtype       :List of 4 ##   .. ..$ module         : chr \"keras\" ##   .. ..$ class_name     : chr \"DTypePolicy\" ##   .. ..$ config         :List of 1 ##   .. .. ..$ name: chr \"float32\" ##   .. ..$ registered_name: NULL ##   ..$ units       : num 64 ##   ..$ nested_model:List of 4 ##   .. ..$ module         : chr \"<r-globalenv>\" ##   .. ..$ class_name     : chr \"MyDense\" ##   .. ..$ config         :List of 6 ##   .. .. ..$ name              : chr \"my_dense\" ##   .. .. ..$ trainable         : logi TRUE ##   .. .. ..$ dtype             :List of 4 ##   .. .. .. ..$ module         : chr \"keras\" ##   .. .. .. ..$ class_name     : chr \"DTypePolicy\" ##   .. .. .. ..$ config         :List of 1 ##   .. .. .. .. ..$ name: chr \"float32\" ##   .. .. .. ..$ registered_name: NULL ##   .. .. ..$ units             : num 16 ##   .. .. ..$ kernel_regularizer: chr \"l1\" ##   .. .. ..$ kernel_initializer: chr \"ones\" ##   .. ..$ registered_name: chr \"MyLayers>KernelMult\" ##  $ registered_name: chr \"MyLayers>KernelMult\" new_layer <- deserialize_keras_object(config) new_layer ## <MyDense name=my_dense_1, built=False> ##  signature: (*args, **kwargs) `%||%` <- \\(x, y) if(is.null(x)) y else x layer_custom_model <- register_keras_serializable(   package = \"ComplexModels\",   object = Layer(     \"CustomModel\",     initialize = function(first_layer, second_layer = NULL, ...) {       super$initialize(...)       self$first_layer <- first_layer       self$second_layer <- second_layer %||% layer_dense(, 8)     },      get_config = function() {       config <- super$get_config()       config <- modifyList(config, list(         first_layer = self$first_layer,         second_layer = self$second_layer       ))       config     },      from_config = function(config) {       config$first_layer %<>% deserialize_keras_object()       config$second_layer %<>% deserialize_keras_object()       # note that the class is available in methods under the classname symbol,       # (`CustomModel` for this class), and also under the symbol `__class__`       cls(!!!config)       # CustomModel(!!!config)     },     call = function(self, inputs) {       inputs |>         self$first_layer() |>         self$second_layer()     }   ) )  # Let's make our first layer the custom layer from the previous example (MyDense) inputs <- keras_input(c(32)) outputs <-  inputs |> layer_custom_model(first_layer=layer) model <- keras_model(inputs, outputs)  config <- get_config(model) new_model <- from_config(config)"},{"path":"https://keras3.posit.co/dev/articles/serialization_and_saving.html","id":"how-custom-objects-are-serialized","dir":"Articles","previous_headings":"Saving > Appendix: Handling custom objects","what":"How custom objects are serialized","title":"Save, serialize, and export models","text":"serialization format special key custom objects registered via register_keras_serializable(). registered_name key allows easy retrieval loading/deserialization time also allowing users add custom naming. Let’s take look config serializing custom layer MyDense defined . Example: shown, registered_name key contains lookup information Keras master list, including package MyLayers custom name KernelMult gave calling register_keras_serializables(). Take look custom class definition/registration . Note class_name key contains original name class, allowing proper re-initialization from_config. Additionally, note module key NULL since custom object.","code":"layer <- layer_my_dense(   units = 16,   kernel_regularizer = regularizer_l1_l2(l1 = 1e-5, l2 = 1e-4),   kernel_initializer = \"ones\", ) config <- serialize_keras_object(layer) str(config) ## List of 4 ##  $ module         : chr \"<r-globalenv>\" ##  $ class_name     : chr \"MyDense\" ##  $ config         :List of 6 ##   ..$ name              : chr \"my_dense_2\" ##   ..$ trainable         : logi TRUE ##   ..$ dtype             :List of 4 ##   .. ..$ module         : chr \"keras\" ##   .. ..$ class_name     : chr \"DTypePolicy\" ##   .. ..$ config         :List of 1 ##   .. .. ..$ name: chr \"float32\" ##   .. ..$ registered_name: NULL ##   ..$ units             : num 16 ##   ..$ kernel_regularizer:List of 4 ##   .. ..$ module         : chr \"keras.regularizers\" ##   .. ..$ class_name     : chr \"L1L2\" ##   .. ..$ config         :List of 2 ##   .. .. ..$ l1: num 1e-05 ##   .. .. ..$ l2: num 1e-04 ##   .. ..$ registered_name: NULL ##   ..$ kernel_initializer: chr \"ones\" ##  $ registered_name: chr \"MyLayers>KernelMult\""},{"path":"https://keras3.posit.co/dev/articles/training_with_built_in_methods.html","id":"setup","dir":"Articles","previous_headings":"","what":"Setup","title":"Training & evaluation with the built-in methods","text":"","code":"library(keras3)"},{"path":"https://keras3.posit.co/dev/articles/training_with_built_in_methods.html","id":"introduction","dir":"Articles","previous_headings":"","what":"Introduction","title":"Training & evaluation with the built-in methods","text":"guide covers training, evaluation, prediction (inference) models using built-APIs training & validation (fit(), evaluate() predict()). interested leveraging fit() specifying training step function, see Customizing happens fit() guide. interested writing training & evaluation loops scratch, see guide Writing training loop scratch. general, whether using built-loops writing , model training & evaluation works strictly way across every kind Keras model – Sequential models, models built Functional API, models written scratch via model subclassing.","code":""},{"path":"https://keras3.posit.co/dev/articles/training_with_built_in_methods.html","id":"api-overview-a-first-end-to-end-example","dir":"Articles","previous_headings":"","what":"API overview: a first end-to-end example","title":"Training & evaluation with the built-in methods","text":"passing data built-training loops model, either use: Arrays (data small fits memory) tf_dataset objects PyTorch DataLoader instances next paragraphs, ’ll use MNIST dataset NumPy arrays, order demonstrate use optimizers, losses, metrics. Afterwards, ’ll take close look options. Let’s consider following model (, build Functional API, Sequential model subclassed model well): ’s typical end--end workflow looks like, consisting : Training Validation holdout set generated original training data Evaluation test data ’ll use MNIST data example. specify training configuration (optimizer, loss, metrics): call fit(), train model slicing data “batches” size batch_size, repeatedly iterating entire dataset given number epochs. returned history object holds record loss values metric values training: evaluate model test data via evaluate(): Now, let’s review piece workflow detail.","code":"inputs <- keras_input(shape = 784, name=\"digits\") outputs <- inputs |>   layer_dense(units = 64, activation = \"relu\", name = \"dense_1\") |>   layer_dense(units = 64, activation = \"relu\", name = \"dense_2\") |>   layer_dense(units = 10, activation = \"softmax\", name = \"predictions\") model <- keras_model(inputs = inputs, outputs = outputs) summary(model) ## Model: \"functional\" ## ┏━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━┓ ## ┃ Layer (type)                    ┃ Output Shape           ┃       Param # ┃ ## ┡━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━┩ ## │ digits (InputLayer)             │ (None, 784)            │             0 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ dense_1 (Dense)                 │ (None, 64)             │        50,240 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ dense_2 (Dense)                 │ (None, 64)             │         4,160 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ predictions (Dense)             │ (None, 10)             │           650 │ ## └─────────────────────────────────┴────────────────────────┴───────────────┘ ##  Total params: 55,050 (215.04 KB) ##  Trainable params: 55,050 (215.04 KB) ##  Non-trainable params: 0 (0.00 B) c(c(x_train, y_train), c(x_test, y_test)) %<-% dataset_mnist()  # Preprocess the data (these are NumPy arrays) x_train <- array_reshape(x_train, c(60000, 784)) / 255 x_test <- array_reshape(x_test, c(10000, 784)) / 255  # Reserve 10,000 samples for validation x_val <- x_train[1:10000,] y_val <- y_train[1:10000] x_train <- x_train[-c(1:10000),] y_train <- y_train[-c(1:10000)] model |> compile(   # Optimizer   optimizer = optimizer_rmsprop(),   # Loss function to minimize   loss = loss_sparse_categorical_crossentropy(),   # List of metrics to monitor   metrics = list(metric_sparse_categorical_accuracy()) ) history <- model |> fit(   x_train, y_train,   batch_size = 64,   epochs = 2,   # We pass some validation for   # monitoring validation loss and metrics   # at the end of each epoch   validation_data = list(x_val, y_val) ) ## Epoch 1/2 ## 782/782 - 3s - 3ms/step - loss: 0.3410 - sparse_categorical_accuracy: 0.9034 - val_loss: 0.1855 - val_sparse_categorical_accuracy: 0.9460 ## Epoch 2/2 ## 782/782 - 1s - 1ms/step - loss: 0.1590 - sparse_categorical_accuracy: 0.9538 - val_loss: 0.1323 - val_sparse_categorical_accuracy: 0.9621 history ## ## Final epoch (plot to see history): ##                            loss: 0.159 ##     sparse_categorical_accuracy: 0.9538 ##                        val_loss: 0.1323 ## val_sparse_categorical_accuracy: 0.9621 # Evaluate the model on the test data using `evaluate` results <- model |> evaluate(x_test, y_test, batch_size=128) ## 79/79 - 0s - 4ms/step - loss: 0.1273 - sparse_categorical_accuracy: 0.9621 str(results) ## List of 2 ##  $ loss                       : num 0.127 ##  $ sparse_categorical_accuracy: num 0.962 # Generate predictions (probabilities -- the output of the last layer) # on new data using `predict` predictions <- model |> predict(x_test[1:2,]) ## 1/1 - 0s - 172ms/step dim(predictions) ## [1]  2 10"},{"path":"https://keras3.posit.co/dev/articles/training_with_built_in_methods.html","id":"the-compile-method-specifying-a-loss-metrics-and-an-optimizer","dir":"Articles","previous_headings":"","what":"The compile() method: specifying a loss, metrics, and an optimizer","title":"Training & evaluation with the built-in methods","text":"train model fit(), need specify loss function, optimizer, optionally, metrics monitor. pass model arguments compile() method: metrics argument list – model can number metrics. model multiple outputs, can specify different losses metrics output, can modulate contribution output total loss model. find details Passing data multi-input, multi-output models section. Note ’re satisfied default settings, many cases optimizer, loss, metrics can specified via string identifiers shortcut: later reuse, let’s put model definition compile step functions; call several times across different examples guide.","code":"model |> compile(   optimizer = optimizer_rmsprop(learning_rate = 1e-3),   loss = loss_sparse_categorical_crossentropy(),   metrics = list(metric_sparse_categorical_accuracy()) ) model |> compile(   optimizer = \"rmsprop\",   loss = \"sparse_categorical_crossentropy\",   metrics = c(\"sparse_categorical_accuracy\") ) get_uncompiled_model <- function() {   inputs <- keras_input(shape = 784, name = \"digits\")   outputs <- inputs |>     layer_dense(units = 64, activation = \"relu\", name = \"dense_1\") |>     layer_dense(units = 64, activation = \"relu\", name = \"dense_2\") |>     layer_dense(units = 10, activation = \"softmax\", name = \"predictions\")   keras_model(inputs = inputs, outputs = outputs) }  get_compiled_model <- function() {   model <- get_uncompiled_model()   model |> compile(     optimizer = \"rmsprop\",     loss = \"sparse_categorical_crossentropy\",     metrics = c(\"sparse_categorical_accuracy\")   )   model }"},{"path":"https://keras3.posit.co/dev/articles/training_with_built_in_methods.html","id":"many-built-in-optimizers-losses-and-metrics-are-available","dir":"Articles","previous_headings":"The compile() method: specifying a loss, metrics, and an optimizer","what":"Many built-in optimizers, losses, and metrics are available","title":"Training & evaluation with the built-in methods","text":"general, won’t create losses, metrics, optimizers scratch, need likely already part Keras API: Optimizers: [optimizer_sgd()] (without momentum) [optimizer_rmsprop()] [optimizer_adam()] etc. Losses: [loss_mean_squared_error()] [loss_kl_divergence()] [loss_cosine_similarity()] etc. Metrics: [metric_auc()] [metric_precision()] [metric_recall()] etc.","code":""},{"path":"https://keras3.posit.co/dev/articles/training_with_built_in_methods.html","id":"custom-losses","dir":"Articles","previous_headings":"The compile() method: specifying a loss, metrics, and an optimizer","what":"Custom losses","title":"Training & evaluation with the built-in methods","text":"need create custom loss, Keras provides three ways . first method involves creating function accepts inputs y_true y_pred. following example shows loss function computes mean squared error real data predictions: need loss function takes parameters beside y_true y_pred, can subclass Keras base Loss class using [Loss()] implement following two methods: initialize(): accept parameters pass call loss function call(y_true, y_pred): use targets (y_true) model predictions (y_pred) compute model’s loss Let’s say want use mean squared error, added term de-incentivize prediction values far 0.5 (assume categorical targets one-hot encoded take values 0 1). creates incentive model confident, may help reduce overfitting (won’t know works try!). ’s :","code":"custom_mean_squared_error <- function(y_true, y_pred) {   op_mean(op_square(y_true - y_pred), axis = -1) }  model <- get_uncompiled_model() model |> compile(optimizer = \"adam\",                  loss = custom_mean_squared_error)  # We need to one-hot encode the labels to use MSE y_train_one_hot <- op_one_hot(y_train, num_classes = 10) model |> fit(x_train, y_train_one_hot, batch_size = 64, epochs = 2) ## Epoch 1/2 ## 782/782 - 2s - 2ms/step - loss: 0.0157 ## Epoch 2/2 ## 782/782 - 1s - 1ms/step - loss: 0.0074 loss_custom_mse <- Loss(   classname = \"CustomMSE\",   initialize = function(regularization_factor = 0.1, name = \"custom_mse\") {     super$initialize(name = name)     self$regularization_factor <- regularization_factor   },   call = function(y_true, y_pred) {     mse <- op_mean(op_square(y_true - y_pred), axis = -1)     reg <- op_mean(op_square(0.5 - y_pred), axis = -1)     mse + reg * self$regularization_factor   } )  model <- get_uncompiled_model() model |> compile(optimizer=\"adam\", loss = loss_custom_mse())  y_train_one_hot <- op_one_hot(y_train, num_classes=10) model |> fit(x_train, y_train_one_hot, batch_size=64, epochs=1) ## 782/782 - 2s - 2ms/step - loss: 0.0388"},{"path":"https://keras3.posit.co/dev/articles/training_with_built_in_methods.html","id":"custom-metrics","dir":"Articles","previous_headings":"The compile() method: specifying a loss, metrics, and an optimizer","what":"Custom metrics","title":"Training & evaluation with the built-in methods","text":"need metric isn’t part API, can easily create custom metrics subclassing Keras base Metric class using [Metric()]. need implement 4 methods: initialize(), create state variables metric. update_state(y_true, y_pred, sample_weight = NULL), uses targets y_true model predictions y_pred update state variables. result(), uses state variables compute final results. reset_state(), reinitializes state metric. State update results computation kept separate (update_state() result(), respectively) cases, results computation might expensive done periodically. ’s simple example showing implement CategoricalTruePositives metric counts many samples correctly classified belonging given class:","code":"metric_categorical_true_positives <- Metric(   \"CategoricalTruePositives\",    initialize = function(name = \"categorical_true_positives\", ...) {     super$initialize(name = name, ...)     self$true_positives <- self$add_variable(shape = shape(),                                              name = \"ctp\",                                              initializer = \"zeros\")   },    update_state = function(y_true, y_pred, sample_weight = NULL) {     y_pred <- op_argmax(y_pred, axis = 2) |> op_reshape(c(-1, 1))     values <- op_cast(y_true, \"int32\") == op_cast(y_pred, \"int32\")     values <- op_cast(values, \"float32\")     if (!is.null(sample_weight)) {       sample_weight <- op_cast(sample_weight, \"float32\")       values <- op_multiply(values, sample_weight)     }     self$true_positives$assign_add(op_sum(values))   },    result = function() {     self$true_positives$value   },    reset_state = function() {     self$true_positives$assign(0.0)   } )  model <- get_uncompiled_model() model |> compile(   optimizer = optimizer_rmsprop(learning_rate = 1e-3),   loss = loss_sparse_categorical_crossentropy(),   metrics = c(metric_categorical_true_positives()) ) history <- model |> fit(x_train, y_train, batch_size = 64, epochs = 3) ## Epoch 1/3 ## 782/782 - 2s - 2ms/step - categorical_true_positives: 285017.0000 - loss: 0.3444 ## Epoch 2/3 ## 782/782 - 1s - 1ms/step - categorical_true_positives: 284268.0000 - loss: 0.1658 ## Epoch 3/3 ## 782/782 - 1s - 973us/step - categorical_true_positives: 284173.0000 - loss: 0.1207"},{"path":"https://keras3.posit.co/dev/articles/training_with_built_in_methods.html","id":"handling-losses-and-metrics-that-dont-fit-the-standard-signature","dir":"Articles","previous_headings":"The compile() method: specifying a loss, metrics, and an optimizer","what":"Handling losses and metrics that don’t fit the standard signature","title":"Training & evaluation with the built-in methods","text":"overwhelming majority losses metrics can computed y_true y_pred, y_pred output model – . instance, regularization loss may require activation layer (targets case), activation may model output. cases, can call self$add_loss(loss_value) inside call method custom layer. Losses added way get added “main” loss training (one passed compile()). ’s simple example adds activity regularization (note activity regularization built-Keras layers – layer just sake providing concrete example): Note pass losses via add_loss(), becomes possible call compile() without loss function, since model already loss minimize. Consider following LogisticEndpoint layer: takes inputs targets & logits, tracks crossentropy loss via add_loss(). can use model two inputs (input data & targets), compiled without loss argument, like : information training multi-input models, see section Passing data multi-input, multi-output models.","code":"layer_custom_activity_regularizer <- Layer(   \"ActivityRegularization\",   call = function(inputs) {     self$add_loss(op_sum(inputs) * 0.1)     inputs  # Pass-through layer.   } )  inputs <- keras_input(shape = 784, name = \"digits\") outputs <- inputs |>   layer_dense(units = 32, activation = \"relu\", name = \"dense_1\") |>   layer_custom_activity_regularizer() |>   layer_dense(units = 64, activation = \"relu\", name = \"dense_2\") |>   layer_dense(units = 10, name = \"predictions\")  model <- keras_model(inputs = inputs, outputs = outputs) model |> compile(optimizer = optimizer_rmsprop(learning_rate = 1e-3),                  loss = loss_sparse_categorical_crossentropy(from_logits = TRUE))  # The displayed loss will be much higher than before # due to the regularization component. model |> fit(x_train, y_train, batch_size = 64, epochs = 1) ## 782/782 - 2s - 2ms/step - loss: 2.3721 layer_logistic_endpoint <- Layer(   \"LogisticEndpoint\",   initialize = function(name = NULL) {     super$initialize(name = name)     self$loss_fn <- loss_binary_crossentropy(from_logits = TRUE)   },   call = function(targets, logits, sample_weights = NULL) {     # Compute the training-time loss value and add it     # to the layer using `self.add_loss()`.     loss <- self$loss_fn(targets, logits, sample_weights)     self$add_loss(loss)      # Return the inference-time prediction tensor (for `predict()`).     op_softmax(logits)   } ) inputs <- keras_input(shape = 3, name = \"inputs\") targets <- keras_input(shape = 10, name = \"targets\")  logits <- inputs |> layer_dense(10) predictions <- layer_logistic_endpoint(name = \"predictions\")(targets, logits)  model <- keras_model(inputs = list(inputs, targets),                      outputs = predictions) model |> compile(optimizer = \"adam\")  # No loss argument!  data <- list(   inputs = random_normal(c(3, 3)),   targets = random_normal(c(3, 10)) ) model |> fit(data, epochs = 1) ## 1/1 - 0s - 439ms/step - loss: 0.9638"},{"path":"https://keras3.posit.co/dev/articles/training_with_built_in_methods.html","id":"automatically-setting-apart-a-validation-holdout-set","dir":"Articles","previous_headings":"The compile() method: specifying a loss, metrics, and an optimizer","what":"Automatically setting apart a validation holdout set","title":"Training & evaluation with the built-in methods","text":"first end--end example saw, used validation_data argument pass list arrays list(x_val, y_val) model evaluating validation loss validation metrics end epoch. ’s another option: argument validation_split allows automatically reserve part training data validation. argument value represents fraction data reserved validation, set number higher 0 lower 1. instance, validation_split = 0.2 means “use 20% data validation”, validation_split = 0.6 means “use 60% data validation”. way validation computed taking last x% samples arrays received fit() call, shuffling. Note can use validation_split training NumPy data.","code":"model <- get_compiled_model() model |> fit(x_train, y_train,              batch_size = 64,              validation_split = 0.2, epochs = 1) ## 625/625 - 2s - 3ms/step - loss: 0.3817 - sparse_categorical_accuracy: 0.8919 - val_loss: 0.1953 - val_sparse_categorical_accuracy: 0.9431"},{"path":"https://keras3.posit.co/dev/articles/training_with_built_in_methods.html","id":"training-evaluation-using-tf-dataset-objects","dir":"Articles","previous_headings":"","what":"Training & evaluation using TF Dataset objects","title":"Training & evaluation with the built-in methods","text":"past paragraphs, ’ve seen handle losses, metrics, optimizers, ’ve seen use validation_data validation_split arguments fit(), data passed arrays. Another option use iterator-like, tf.data.Dataset, PyTorch DataLoader, R generator function. Let’s take look former. tfdatasets R package containes set utilities loading preprocessing data way ’s fast scalable. complete guide creating Datasets, see tf.data documentation. can use tf.data train Keras models regardless backend ’re using – whether ’s JAX, PyTorch, TensorFlow. can pass Dataset instance directly methods fit(), evaluate(), predict(): Note Dataset reset end epoch, can reused next epoch. want run training specific number batches Dataset, can pass steps_per_epoch argument, specifies many training steps model run using Dataset moving next epoch. can also pass Dataset instance validation_data argument fit(): end epoch, model iterate validation dataset compute validation loss validation metrics. want run validation specific number batches dataset, can pass validation_steps argument, specifies many validation steps model run validation dataset interrupting validation moving next epoch: Note validation dataset reset use (always evaluating samples epoch epoch). argument validation_split (generating holdout set training data) supported training Dataset objects, since feature requires ability index samples datasets, possible general Dataset API.","code":"library(tfdatasets, exclude = \"shape\") model <- get_compiled_model()  # First, let's create a training Dataset instance. # For the sake of our example, we'll use the same MNIST data as before. train_dataset <- tensor_slices_dataset(list(x_train, y_train))  # Shuffle and slice the dataset. train_dataset <- train_dataset |>   dataset_shuffle(buffer_size=1024) |>   dataset_batch(64)  # Now we get a test dataset. test_dataset <-   tensor_slices_dataset(list(x_test, y_test)) |>   dataset_batch(64)  # Since the dataset already takes care of batching, # we don't pass a `batch_size` argument. model |> fit(train_dataset, epochs = 3) ## Epoch 1/3 ## 782/782 - 2s - 2ms/step - loss: 0.3365 - sparse_categorical_accuracy: 0.9041 ## Epoch 2/3 ## 782/782 - 1s - 1ms/step - loss: 0.1606 - sparse_categorical_accuracy: 0.9521 ## Epoch 3/3 ## 782/782 - 1s - 1ms/step - loss: 0.1178 - sparse_categorical_accuracy: 0.9646 # You can also evaluate or predict on a dataset. result <- model |> evaluate(test_dataset) ## 157/157 - 1s - 3ms/step - loss: 0.1213 - sparse_categorical_accuracy: 0.9596 result ## $loss ## [1] 0.1213005 ## ## $sparse_categorical_accuracy ## [1] 0.9596 model <- get_compiled_model()  # Prepare the training dataset train_dataset <- tensor_slices_dataset(list(x_train, y_train)) train_dataset <- train_dataset |>   dataset_shuffle(buffer_size = 1024) |>   dataset_batch(64)  # Only use the 100 batches per epoch (that's 64 * 100 samples) model |> fit(train_dataset, epochs = 3, steps_per_epoch = 100) ## Epoch 1/3 ## 100/100 - 1s - 8ms/step - loss: 0.8017 - sparse_categorical_accuracy: 0.7806 ## Epoch 2/3 ## 100/100 - 0s - 1ms/step - loss: 0.3661 - sparse_categorical_accuracy: 0.9006 ## Epoch 3/3 ## 100/100 - 0s - 960us/step - loss: 0.3009 - sparse_categorical_accuracy: 0.9106 model <- get_compiled_model()  # Prepare the training dataset train_dataset <- tensor_slices_dataset(list(x_train, y_train)) train_dataset <- train_dataset |>   dataset_shuffle(buffer_size=1024) |>   dataset_batch(64)  # Prepare the validation dataset val_dataset <- tensor_slices_dataset(list(x_val, y_val)) val_dataset <- val_dataset |> dataset_batch(64)  model |> fit(train_dataset, epochs = 1, validation_data = val_dataset) ## 782/782 - 2s - 3ms/step - loss: 0.3428 - sparse_categorical_accuracy: 0.9022 - val_loss: 0.2337 - val_sparse_categorical_accuracy: 0.9291 model  <- get_compiled_model()  # Prepare the training dataset train_dataset <- tensor_slices_dataset(list(x_train, y_train)) train_dataset <- train_dataset |>   dataset_shuffle(buffer_size = 1024) |>   dataset_batch(64)  # Prepare the validation dataset val_dataset <- tensor_slices_dataset(list(x_val, y_val)) val_dataset <- val_dataset |> dataset_batch(64)  model %>% fit(   train_dataset,   epochs = 1,   # Only run validation using the first 10 batches of the dataset   # using the `validation_steps` argument   validation_data = val_dataset,   validation_steps = 10, ) ## 782/782 - 2s - 3ms/step - loss: 0.3391 - sparse_categorical_accuracy: 0.9035 - val_loss: 0.1997 - val_sparse_categorical_accuracy: 0.9391"},{"path":"https://keras3.posit.co/dev/articles/training_with_built_in_methods.html","id":"using-sample-weighting-and-class-weighting","dir":"Articles","previous_headings":"","what":"Using sample weighting and class weighting","title":"Training & evaluation with the built-in methods","text":"default settings weight sample decided frequency dataset. two methods weight data, independent sample frequency: Class weights Sample weights","code":""},{"path":"https://keras3.posit.co/dev/articles/training_with_built_in_methods.html","id":"class-weights","dir":"Articles","previous_headings":"Using sample weighting and class weighting","what":"Class weights","title":"Training & evaluation with the built-in methods","text":"set passing named list class_weight argument fit(). list maps class indices weight used samples belonging class. can used balance classes without resampling, train model gives importance particular class. instance, class “0” half represented class “1” data, use model |> fit(..., class_weight = c(\"0\" = 1, \"1\" = 0.5)). ’s R example use class weights sample weights give importance correct classification class #5 (digit “5” MNIST dataset).","code":"class_weight <- c(     \"0\" = 1.0,     \"1\" = 1.0,     \"2\" = 1.0,     \"3\" = 1.0,     \"4\" = 1.0,     # Set weight \"2\" for class \"5\",     # making this class 2x more important     \"5\" = 2.0,     \"6\" = 1.0,     \"7\" = 1.0,     \"8\" = 1.0,     \"9\" = 1.0 )  model <- get_compiled_model() model |> fit(x_train, y_train,              class_weight = class_weight,              batch_size = 64, epochs = 1) ## 782/782 - 2s - 2ms/step - loss: 0.3713 - sparse_categorical_accuracy: 0.9018"},{"path":"https://keras3.posit.co/dev/articles/training_with_built_in_methods.html","id":"sample-weights","dir":"Articles","previous_headings":"Using sample weighting and class weighting","what":"Sample weights","title":"Training & evaluation with the built-in methods","text":"fine grained control, building classifier, can use sample_weights. training R arrays: Pass sample_weight argument fit(). training tf_dataset sort iterator: yield (input_batch, label_batch, sample_weight_batch) tuples. “sample weights” array array numbers specify much weight sample batch computing total loss. commonly used imbalanced classification problems (idea give weight rarely-seen classes). weights used ones zeros, array can used mask loss function (entirely discarding contribution certain samples total loss). ’s matching Dataset example:","code":"sample_weight <- rep(1.0, length(y_train)) sample_weight[y_train == 5] <- 2.0  model <- get_compiled_model() model |> fit(   x_train, y_train,   sample_weight = sample_weight,   batch_size = 64, epochs = 1 ) ## 782/782 - 2s - 2ms/step - loss: 0.3740 - sparse_categorical_accuracy: 0.9015 sample_weight <- rep(1.0, length(y_train)) sample_weight[y_train == 5] <- 2.0  # Create a Dataset that includes sample weights # (3rd element in the return tuple). train_dataset <- tensor_slices_dataset(list(     x_train, y_train, sample_weight ))  # Shuffle and slice the dataset. train_dataset <- train_dataset |>   dataset_shuffle(buffer_size = 1024) |>   dataset_batch(64)  model <- get_compiled_model() model |> fit(train_dataset, epochs = 1) ## 782/782 - 2s - 2ms/step - loss: 0.3654 - sparse_categorical_accuracy: 0.9057"},{"path":"https://keras3.posit.co/dev/articles/training_with_built_in_methods.html","id":"passing-data-to-multi-input-multi-output-models","dir":"Articles","previous_headings":"","what":"Passing data to multi-input, multi-output models","title":"Training & evaluation with the built-in methods","text":"previous examples, considering model single input (tensor shape (764)) single output (prediction tensor shape (10)). models multiple inputs outputs? Consider following model, image input shape (32, 32, 3) (’s (height, width, channels)) time series input shape (NA, 10) (’s (timesteps, features)). model two outputs computed combination inputs: “score” (shape (1)) probability distribution five classes (shape (5)). Let’s plot model, can clearly see ’re (note shapes shown plot batch shapes, rather per-sample shapes). plot chunk unnamed-chunk-26 compilation time, can specify different losses different outputs, passing loss functions list: passed single loss function model, loss function applied every output (appropriate ). Likewise metrics: Since gave names output layers, also specify per-output losses metrics via named list: recommend use names 2 outputs. ’s possible give different weights different output-specific losses (instance, one might wish privilege “score” loss example, giving 2x importance class loss), using loss_weights argument: also choose compute loss certain outputs, outputs meant prediction training: Passing data multi-input multi-output model fit() works similar way specifying loss function compile: can pass lists arrays (1:1 mapping outputs received loss function) dicts mapping output names arrays. ’s Dataset use case: similarly R arrays, Dataset return tuple named lists (dicts).","code":"image_input <- keras_input(c(32, 32, 3), name = \"img_input\") timeseries_input <- keras_input(c(NA, 10), name = \"ts_input\")  x1 <- image_input |>   layer_conv_2d(filters = 3, kernel_size = c(3, 3)) |>   layer_global_max_pooling_2d()  x2 <- timeseries_input |>   layer_conv_1d(filters = 3, kernel_size = 3) |>   layer_global_max_pooling_1d()  x <- layer_concatenate(x1, x2)  score_output <- layer_dense(x, 1, name = \"score_output\") class_output <- layer_dense(x, 5, name = \"class_output\")  model <- keras_model(   inputs = list(image_input, timeseries_input),   outputs = list(score_output, class_output) ) plot(model, show_shapes = TRUE) model |> compile(   optimizer = optimizer_rmsprop(1e-3),   loss = list(     loss_mean_squared_error(),     loss_categorical_crossentropy()   ) ) model |> compile(   optimizer = optimizer_rmsprop(1e-3),   loss = list(     loss_mean_squared_error(),     loss_categorical_crossentropy()   ),   metrics = list(     list(       metric_mean_absolute_error(),       metric_mean_absolute_percentage_error()     ),     list(metric_categorical_accuracy())   ) ) model |> compile(   optimizer = optimizer_rmsprop(1e-3),   loss = list(     score_output = loss_mean_squared_error(),     class_output = loss_categorical_crossentropy()   ),   metrics = list(     score_output = list(       metric_mean_absolute_error(),       metric_mean_absolute_percentage_error()     ),     class_output = list(metric_categorical_accuracy())   ) ) model |> compile(   optimizer = optimizer_rmsprop(1e-3),   loss = list(     score_output = loss_mean_squared_error(),     class_output = loss_categorical_crossentropy()   ),   metrics = list(     score_output = list(       metric_mean_absolute_error(),       metric_mean_absolute_percentage_error()     ),     class_output = list(metric_categorical_accuracy())   ),   loss_weights = list(score_output = 2.0, class_output = 1.0) ) # loss list, positional version model |> compile(   optimizer = optimizer_rmsprop(1e-3),   loss = list(NULL, loss_categorical_crossentropy()) )  # Or loss list, named version model |> compile(   optimizer = optimizer_rmsprop(1e-3),   loss = list(class_output = loss_categorical_crossentropy()) ) model |> compile(   optimizer = optimizer_rmsprop(1e-3),   loss = list(     loss_mean_squared_error(),     loss_binary_crossentropy()   ) )  # Generate dummy data img_data <- random_normal(c(100, 32, 32, 3)) ts_data <- random_normal(c(100, 20, 10)) score_targets <- random_normal(c(100, 1)) class_targets <- random_normal(c(100, 5))  # Fit on unnamed lists (positional matching) model |> fit(     list(img_data, ts_data),     list(score_targets, class_targets),     batch_size=32,     epochs=1 ) ## 4/4 - 2s - 540ms/step - class_output_categorical_accuracy: 0.2300 - class_output_loss: 1.8162 - loss: 3.8764 - score_output_loss: 1.6820 - score_output_mean_absolute_error: 0.8826 - score_output_mean_absolute_percentage_error: 1826.5676 # Alternatively, fit on named lists (names matching) model <- keras_model(   inputs = list(image_input, timeseries_input),   outputs = list(score_output = score_output,                  class_output = class_output) ) |> compile(   optimizer = optimizer_rmsprop(1e-3),   loss = list(     loss_mean_squared_error(),     loss_binary_crossentropy()   ) )  model |> fit(   list(img_input = img_data, ts_input = ts_data),   list(score_output = score_targets, class_output = class_targets),   batch_size = 32,   epochs = 1 ) ## 4/4 - 2s - 392ms/step - class_output_loss: 4.1489 - loss: 4.5443 - score_output_loss: -4.1125e-01 train_dataset <- tensor_slices_dataset(list(   list(img_input = img_data, ts_input = ts_data),   list(score_output = score_targets, class_output = class_targets) )) train_dataset <- train_dataset |>   dataset_shuffle(buffer_size = 1024) |>   dataset_batch(64)  model |> fit(train_dataset, epochs = 1) ## 2/2 - 1s - 706ms/step - class_output_loss: 3.8547 - loss: 3.5725 - score_output_loss: -1.5209e-01"},{"path":"https://keras3.posit.co/dev/articles/training_with_built_in_methods.html","id":"using-callbacks","dir":"Articles","previous_headings":"","what":"Using callbacks","title":"Training & evaluation with the built-in methods","text":"Callbacks Keras objects called different points training (start epoch, end batch, end epoch, etc.). can used implement certain behaviors, : validation different points training (beyond built-per-epoch validation) Checkpointing model regular intervals exceeds certain accuracy threshold Changing learning rate model training seems plateauing fine-tuning top layers training seems plateauing Sending email instant message notifications training ends certain performance threshold exceeded Etc. Callbacks can passed list call fit():","code":"model <- get_compiled_model()  callbacks <- list(   callback_early_stopping(     # Stop training when `val_loss` is no longer improving     monitor = \"val_loss\",     # \"no longer improving\" being defined as \"no better than 1e-2 less\"     min_delta = 1e-2,     # \"no longer improving\" being further defined as \"for at least 2 epochs\"     patience = 2,     verbose = 1   ) ) model |> fit(   x_train,   y_train,   epochs = 20,   batch_size = 64,   callbacks = callbacks,   validation_split = 0.2, ) ## Epoch 1/20 ## 625/625 - 2s - 3ms/step - loss: 0.3695 - sparse_categorical_accuracy: 0.8961 - val_loss: 0.1873 - val_sparse_categorical_accuracy: 0.9469 ## Epoch 2/20 ## 625/625 - 1s - 1ms/step - loss: 0.1752 - sparse_categorical_accuracy: 0.9488 - val_loss: 0.1405 - val_sparse_categorical_accuracy: 0.9576 ## Epoch 3/20 ## 625/625 - 1s - 1ms/step - loss: 0.1277 - sparse_categorical_accuracy: 0.9626 - val_loss: 0.1217 - val_sparse_categorical_accuracy: 0.9644 ## Epoch 4/20 ## 625/625 - 1s - 1ms/step - loss: 0.1009 - sparse_categorical_accuracy: 0.9702 - val_loss: 0.1138 - val_sparse_categorical_accuracy: 0.9667 ## Epoch 5/20 ## 625/625 - 1s - 1ms/step - loss: 0.0823 - sparse_categorical_accuracy: 0.9758 - val_loss: 0.1086 - val_sparse_categorical_accuracy: 0.9679 ## Epoch 6/20 ## 625/625 - 1s - 1ms/step - loss: 0.0682 - sparse_categorical_accuracy: 0.9800 - val_loss: 0.1084 - val_sparse_categorical_accuracy: 0.9685 ## Epoch 7/20 ## 625/625 - 1s - 1ms/step - loss: 0.0573 - sparse_categorical_accuracy: 0.9834 - val_loss: 0.1100 - val_sparse_categorical_accuracy: 0.9698 ## Epoch 8/20 ## 625/625 - 1s - 1ms/step - loss: 0.0485 - sparse_categorical_accuracy: 0.9868 - val_loss: 0.1113 - val_sparse_categorical_accuracy: 0.9702 ## Epoch 8: early stopping"},{"path":"https://keras3.posit.co/dev/articles/training_with_built_in_methods.html","id":"many-built-in-callbacks-are-available","dir":"Articles","previous_headings":"Using callbacks","what":"Many built-in callbacks are available","title":"Training & evaluation with the built-in methods","text":"many built-callbacks already available Keras, : callback_model_checkpoint(): Periodically save model. callback_early_stopping(): Stop training training longer improving validation metrics. callback_tensorboard(): periodically write model logs can visualized TensorBoard (details section “Visualization”). callback_csv_logger(): streams loss metrics data CSV file. etc. See callbacks documentation complete list.","code":""},{"path":"https://keras3.posit.co/dev/articles/training_with_built_in_methods.html","id":"writing-your-own-callback","dir":"Articles","previous_headings":"Using callbacks","what":"Writing your own callback","title":"Training & evaluation with the built-in methods","text":"can create custom callback subclassing base [Callback()] class. callback access associated model class property self$model. Make sure read complete guide writing custom callbacks. ’s simple example saving list per-batch loss values training:","code":"callback_loss_history <- Callback(   classname = \"LossHistory\",   initialize = function(file = \"per_training_batch_losses.txt\", ...) {     super$initialize(...)     private$file <- file   },   on_train_begin = function(logs = NULL) {     private$per_batch_losses <- fastmap::faststack()   },   on_train_batch_begin = function(batch, logs = NULL) {     private$per_batch_losses$push(logs$loss)   },   on_train_end = function(logs = NULL) {     per_batch_losses <- private$per_batch_losses$as_list() |> as.numeric()     write(per_batch_losses, private$file)   } )"},{"path":"https://keras3.posit.co/dev/articles/training_with_built_in_methods.html","id":"checkpointing-models","dir":"Articles","previous_headings":"","what":"Checkpointing models","title":"Training & evaluation with the built-in methods","text":"’re training model relatively large datasets, ’s crucial save checkpoints model frequent intervals. easiest way achieve [callback_model_checkpoint()]: ModelCheckpoint callback can used implement fault-tolerance: ability restart training last saved state model case training gets randomly interrupted. ’s basic example: call also write callback saving restoring models. complete guide serialization saving, see guide saving serializing Models.","code":"model <- get_compiled_model()  callbacks <- list(   callback_model_checkpoint(     # Path where to save the model     # The two parameters below mean that we will overwrite     # the current checkpoint if and only if     # the `val_loss` score has improved.     # The saved model name will include the current epoch.     filepath = \"mymodel_{epoch}.keras\",     save_best_only = TRUE,     # Only save a model if `val_loss` has improved.     monitor = \"val_loss\",     verbose = 1   ) ) model |> fit(   x_train, y_train,   epochs = 2, batch_size = 64,   callbacks = callbacks,   validation_split = 0.2 ) ## Epoch 1/2 ## ## Epoch 1: val_loss improved from None to 0.19488, saving model to mymodel_1.keras ## 625/625 - 2s - 3ms/step - loss: 0.3793 - sparse_categorical_accuracy: 0.8936 - val_loss: 0.1949 - val_sparse_categorical_accuracy: 0.9447 ## Epoch 2/2 ## ## Epoch 2: val_loss improved from 0.19488 to 0.14489, saving model to mymodel_2.keras ## 625/625 - 1s - 1ms/step - loss: 0.1777 - sparse_categorical_accuracy: 0.9475 - val_loss: 0.1449 - val_sparse_categorical_accuracy: 0.9590 # Prepare a directory to store all the checkpoints. checkpoint_dir <- \"./ckpt\" fs::dir_create(checkpoint_dir)  make_or_restore_model <- function() {   # Either restore the latest (best) model, or create a fresh one   # if there is no checkpoint available.   checkpoints <- Sys.glob(file.path(checkpoint_dir, \"model-loss=*.keras\"))    if (length(checkpoints) > 0) {     checkpoint_losses <- sub(\"^model-loss=([0-9.]+)\\\\.keras$\", \"\\\\1\",                              basename(checkpoints)) |> as.numeric()     best_checkpoint <- checkpoints[which.min(checkpoint_losses)]     load_model(best_checkpoint)   } else {     get_compiled_model()   } }  model <- make_or_restore_model() callbacks <- list(   # This callback saves the model every 100 batches.   # We include the training loss in the saved model name.   callback_model_checkpoint(     filepath = file.path(checkpoint_dir, \"model-loss={loss:.2f}.keras\"),     save_freq = 100   ) ) model |> fit(x_train, y_train, epochs = 1, callbacks = callbacks) ## 1563/1563 - 3s - 2ms/step - loss: 0.2932 - sparse_categorical_accuracy: 0.9145"},{"path":"https://keras3.posit.co/dev/articles/training_with_built_in_methods.html","id":"using-learning-rate-schedules","dir":"Articles","previous_headings":"","what":"Using learning rate schedules","title":"Training & evaluation with the built-in methods","text":"common pattern training deep learning models gradually reduce learning training progresses. generally known “learning rate decay”. learning decay schedule static (fixed advance, function current epoch current batch index), dynamic (responding current behavior model, particular validation loss).","code":""},{"path":"https://keras3.posit.co/dev/articles/training_with_built_in_methods.html","id":"passing-a-schedule-to-an-optimizer","dir":"Articles","previous_headings":"Using learning rate schedules","what":"Passing a schedule to an optimizer","title":"Training & evaluation with the built-in methods","text":"can easily use static learning rate decay schedule passing schedule object learning_rate argument optimizer: Several built-schedules available: ExponentialDecay, PiecewiseConstantDecay, PolynomialDecay, InverseTimeDecay.","code":"initial_learning_rate <- 0.1 lr_schedule <- learning_rate_schedule_exponential_decay(     initial_learning_rate, decay_steps=100000, decay_rate=0.96,     staircase=TRUE )  optimizer <- optimizer_rmsprop(learning_rate = lr_schedule)"},{"path":"https://keras3.posit.co/dev/articles/training_with_built_in_methods.html","id":"using-callbacks-to-implement-a-dynamic-learning-rate-schedule","dir":"Articles","previous_headings":"Using learning rate schedules","what":"Using callbacks to implement a dynamic learning rate schedule","title":"Training & evaluation with the built-in methods","text":"dynamic learning rate schedule (instance, decreasing learning rate validation loss longer improving) achieved schedule objects, since optimizer access validation metrics. However, callbacks access metrics, including validation metrics! can thus achieve pattern using callback modifies current learning rate optimizer. fact, even built-[callback_reduce_lr_on_plateau()].","code":""},{"path":"https://keras3.posit.co/dev/articles/training_with_built_in_methods.html","id":"visualizing-loss-and-metrics-during-training-with-tensorboard","dir":"Articles","previous_headings":"","what":"Visualizing loss and metrics during training with TensorBoard","title":"Training & evaluation with the built-in methods","text":"best way keep eye model training use TensorBoard – browser-based application can run locally provides : Live plots loss metrics training evaluation (optionally) Visualizations histograms layer activations (optionally) 3D visualizations embedding spaces learned layer_embedding() installed TensorFlow pip, able launch TensorBoard command line: R using:","code":"tensorboard --logdir=/full_path_to_your_logs tensorflow::tensorboard(logdir = \"/full_path_to_your_logs\")"},{"path":"https://keras3.posit.co/dev/articles/training_with_built_in_methods.html","id":"using-the-tensorboard-callback","dir":"Articles","previous_headings":"Visualizing loss and metrics during training with TensorBoard","what":"Using the TensorBoard callback","title":"Training & evaluation with the built-in methods","text":"easiest way use TensorBoard Keras model fit() method [callback_tensorboard()]. simplest case, just specify want callback write logs, ’re good go: information, see callback_tensorboard().","code":"tb_callback <- callback_tensorboard(   log_dir = \"/full_path_to_your_logs\",   histogram_freq = 0, # How often to log histogram visualizations   embeddings_freq = 0, # How often to log embedding visualizations   update_freq = \"epoch\", # How often to write logs (default: once per epoch) )"},{"path":"https://keras3.posit.co/dev/articles/transfer_learning.html","id":"setup","dir":"Articles","previous_headings":"","what":"Setup","title":"Transfer learning & fine-tuning","text":"","code":"library(keras3)"},{"path":"https://keras3.posit.co/dev/articles/transfer_learning.html","id":"introduction","dir":"Articles","previous_headings":"","what":"Introduction","title":"Transfer learning & fine-tuning","text":"Transfer learning consists taking features learned one problem, leveraging new, similar problem. instance, features model learned identify raccoon may useful kick-start model meant identify tanukis. Transfer learning usually done tasks dataset little data train full-scale model scratch. common incarnation transfer learning context deep learning following workflow: Take layers previously trained model. Freeze , avoid destroying information contain future training rounds. Add new, trainable layers top frozen layers. learn turn old features predictions new dataset. Train new layers dataset. last, optional step, fine-tuning, consists unfreezing entire model obtained (part ), re-training new data low learning rate. can potentially achieve meaningful improvements, incrementally adapting pretrained features new data. First, go Keras trainable API detail, underlies transfer learning & fine-tuning workflows. , ’ll demonstrate typical workflow taking model pretrained ImageNet dataset, retraining Kaggle “cats vs dogs” classification dataset. adapted Deep Learning R 2016 blog post “building powerful image classification models using little data”.","code":""},{"path":"https://keras3.posit.co/dev/articles/transfer_learning.html","id":"freezing-layers-understanding-the-trainable-attribute","dir":"Articles","previous_headings":"","what":"Freezing layers: understanding the trainable attribute","title":"Transfer learning & fine-tuning","text":"Layers & models three weight attributes: weights list weights variables layer. trainable_weights list meant updated (via gradient descent) minimize loss training. non_trainable_weights list aren’t meant trained. Typically updated model forward pass. Example: Dense layer 2 trainable weights (kernel & bias) general, weights trainable weights. built-layer non-trainable weights [layer_batch_normalization()]. uses non-trainable weights keep track mean variance inputs training. learn use non-trainable weights custom layers, see guide writing new layers scratch. Example: BatchNormalization layer 2 trainable weights 2 non-trainable weights Layers & models also feature boolean attribute trainable. value can changed. Setting layer$trainable FALSE moves layer’s weights trainable non-trainable. called “freezing” layer: state frozen layer won’t updated training (either training fit() training custom loop relies trainable_weights apply gradient updates). Example: setting trainable False trainable weight becomes non-trainable, value longer updated training. confuse layer$trainable attribute argument training layer$call() (controls whether layer run forward pass inference mode training mode). information, see Keras FAQ.","code":"layer <- layer_dense(units = 3) layer$build(shape(NULL, 4))  # Create the weights  length(layer$weights) ## [1] 2 length(layer$trainable_weights) ## [1] 2 length(layer$non_trainable_weights) ## [1] 0 layer <- layer_batch_normalization() layer$build(shape(NA, 4))  # Create the weights  length(layer$weights) ## [1] 4 length(layer$trainable_weights) ## [1] 2 length(layer$non_trainable_weights) ## [1] 2 layer <- layer_dense(units = 3) layer$build(shape(NULL, 4))  # Create the weights layer$trainable <- FALSE  # Freeze the layer  length(layer$weights) ## [1] 2 length(layer$trainable_weights) ## [1] 0 length(layer$non_trainable_weights) ## [1] 2 # Make a model with 2 layers layer1 <- layer_dense(units = 3, activation = \"relu\") layer2 <- layer_dense(units = 3, activation = \"sigmoid\") model <- keras_model_sequential(input_shape = 3) |>   layer1() |>   layer2()  # Freeze the first layer layer1$trainable <- FALSE  # Keep a copy of the weights of layer1 for later reference # (get_weights() returns a list of R arrays, #  layer$weights returns a list of KerasVariables) initial_layer1_weights_values <- get_weights(layer1)  # Train the model model |> compile(optimizer = \"adam\", loss = \"mse\") model |> fit(random_normal(c(2, 3)), random_normal(c(2, 3)), epochs = 1) ## 1/1 - 0s - 436ms/step - loss: 2.8362 # Check that the weights of layer1 have not changed during training final_layer1_weights_values <- get_weights(layer1)  all.equal(initial_layer1_weights_values,           final_layer1_weights_values) ## [1] TRUE"},{"path":"https://keras3.posit.co/dev/articles/transfer_learning.html","id":"recursive-setting-of-the-trainable-attribute","dir":"Articles","previous_headings":"","what":"Recursive setting of the trainable attribute","title":"Transfer learning & fine-tuning","text":"set $trainable <- FALSE model layer sublayers, children layers become non-trainable well. Example:","code":"inner_model <- keras_model_sequential(input_shape = 3) |>   layer_dense(units = 3, activation = \"relu\") |>   layer_dense(units = 3, activation = \"relu\")  model <- keras_model_sequential(input_shape = 3) |>   inner_model() |>   layer_dense(units = 3, activation = \"sigmoid\")  model$trainable <- FALSE  # Freeze the outer model  inner_model$trainable  # All layers in `model` are now frozen ## [1] FALSE inner_model$layers[[1]]$trainable  # `trainable` is propagated recursively ## [1] FALSE"},{"path":"https://keras3.posit.co/dev/articles/transfer_learning.html","id":"the-typical-transfer-learning-workflow","dir":"Articles","previous_headings":"","what":"The typical transfer-learning workflow","title":"Transfer learning & fine-tuning","text":"leads us typical transfer learning workflow can implemented Keras: Instantiate base model load pre-trained weights . Freeze layers base model setting trainable <- FALSE. Create new model top output one (several) layers base model. Train new model new dataset. Note alternative, lightweight workflow also : Instantiate base model load pre-trained weights . Run new dataset record output one (several) layers base model. called feature extraction. Use output input data new, smaller model. key advantage second workflow run base model data, rather per epoch training. ’s lot faster & cheaper. issue second workflow, though, doesn’t allow dynamically modify input data new model training, required data augmentation, instance. Transfer learning typically used tasks new dataset little data train full-scale model scratch, scenarios data augmentation important. follows, focus first workflow. ’s first workflow looks like Keras: First, instantiate base model pre-trained weights. , freeze base model. Create new model top. Train model new data.","code":"base_model <- application_xception(   weights = 'imagenet', # Load weights pre-trained on ImageNet.   input_shape = c(150, 150, 3),   include_top = FALSE  # Do not include the ImageNet classifier at the top. ) base_model$trainable <- FALSE inputs <- keras_input(shape = c(150, 150, 3)) # We make sure that the base_model is running in inference mode here, # by passing `training <- FALSE`. This is important for fine-tuning, as you will # learn in a few paragraphs. outputs <- inputs |>   base_model(training = FALSE) |>   # Convert features of shape `base_model$output_shape[-1]` to vectors   layer_global_average_pooling_2d() |>   # A Dense classifier with a single unit (binary classification)   layer_dense(1)  model <- keras_model(inputs, outputs) model |> compile(   optimizer = optimizer_adam(),   loss = loss_binary_crossentropy(from_logits = TRUE),   metrics = list(metric_binary_accuracy()) ) model |> fit(new_dataset, epochs = 20,              callbacks = ..., validation_data = ...)"},{"path":"https://keras3.posit.co/dev/articles/transfer_learning.html","id":"fine-tuning","dir":"Articles","previous_headings":"","what":"Fine-tuning","title":"Transfer learning & fine-tuning","text":"model converged new data, can try unfreeze part base model retrain whole model end--end low learning rate. optional last step can potentially give incremental improvements. also potentially lead quick overfitting – keep mind. critical step model frozen layers trained convergence. mix randomly-initialized trainable layers trainable layers hold pre-trained features, randomly-initialized layers cause large gradient updates training, destroy pre-trained features. ’s also critical use low learning rate stage, training much larger model first round training, dataset typically small. result, risk overfitting quickly apply large weight updates. , want readapt pretrained weights incremental way. implement fine-tuning whole base model: Important note compile() trainable Calling compile() model meant “freeze” behavior model. implies trainable attribute values time model compiled preserved throughout lifetime model, compile called . Hence, change trainable value, make sure call compile() model changes taken account. Important notes BatchNormalization layer Many image models contain BatchNormalization layers. layer special case every imaginable count. things keep mind. BatchNormalization contains 2 non-trainable weights get updated training. variables tracking mean variance inputs. set bn_layer$trainable <- FALSE, BatchNormalization layer run inference mode, update mean & variance statistics. case layers general, weight trainability & inference/training modes two orthogonal concepts. two tied case BatchNormalization layer. unfreeze model contains BatchNormalization layers order fine-tuning, keep BatchNormalization layers inference mode passing training = FALSE calling base model. Otherwise updates applied non-trainable weights suddenly destroy model learned. ’ll see pattern action end--end example end guide.","code":"# Unfreeze the base model base_model$trainable <- TRUE  # It's important to recompile your model after you make any changes # to the `trainable` attribute of any inner layer, so that your changes # are take into account model |> compile(   optimizer = optimizer_adam(1e-5), # Very low learning rate   loss = loss_binary_crossentropy(from_logits = TRUE),   metrics = c(metric_binary_accuracy()) )  # Train end-to-end. Be careful to stop before you overfit! model |> fit(new_dataset, epochs = 10,              callbacks = ..., validation_data = ...)"},{"path":"https://keras3.posit.co/dev/articles/transfer_learning.html","id":"an-end-to-end-example-fine-tuning-an-image-classification-model-on-a-cats-vs--dogs-dataset","dir":"Articles","previous_headings":"","what":"An end-to-end example: fine-tuning an image classification model on a cats vs. dogs dataset","title":"Transfer learning & fine-tuning","text":"solidify concepts, let’s walk concrete end--end transfer learning & fine-tuning example. load Xception model, pre-trained ImageNet, use Kaggle “cats vs. dogs” classification dataset.","code":""},{"path":"https://keras3.posit.co/dev/articles/transfer_learning.html","id":"getting-the-data","dir":"Articles","previous_headings":"An end-to-end example: fine-tuning an image classification model on a cats vs. dogs dataset","what":"Getting the data","title":"Transfer learning & fine-tuning","text":"First, let’s fetch cats vs. dogs dataset using TFDS. dataset, ’ll probably want use utility [image_dataset_from_directory()] generate similar labeled dataset objects set images disk filed class-specific folders. Transfer learning useful working small datasets. keep dataset small, use 40% original training data (25,000 images) training, 10% validation, 10% testing. first 9 images training dataset – can see, ’re different sizes. can also see label 1 “dog” label 0 “cat”.","code":"reticulate::py_require(\"tensorflow-datasets\") tfds <- reticulate::import(\"tensorflow_datasets\")  c(train_ds, validation_ds, test_ds) %<-% tfds$load(   \"cats_vs_dogs\",   # Reserve 10% for validation and 10% for test   split = c(\"train[:40%]\", \"train[40%:50%]\", \"train[50%:60%]\"),   as_supervised = TRUE  # Include labels )  length(train_ds) ## [1] 9305 library(tfdatasets, exclude = \"shape\")  par(mfrow = c(3, 3), mar = c(1,0,1.5,0)) train_ds |>   dataset_take(9) |>   as_array_iterator() |>   iterate(function(batch) {     c(image, label) %<-% batch     plot(as.raster(image, max = 255L))     title(sprintf(       \"label: %s   size: %s\",       label, paste(dim(image), collapse = \" x \")))   })"},{"path":"https://keras3.posit.co/dev/articles/transfer_learning.html","id":"standardizing-the-data","dir":"Articles","previous_headings":"An end-to-end example: fine-tuning an image classification model on a cats vs. dogs dataset","what":"Standardizing the data","title":"Transfer learning & fine-tuning","text":"raw images variety sizes. addition, pixel consists 3 integer values 0 255 (RGB level values). isn’t great fit feeding neural network. need 2 things: Standardize fixed image size. pick 150x150. Normalize pixel values -1 1. ’ll using Normalization layer part model . general, ’s good practice develop models take raw data input, opposed models take already-preprocessed data. reason , model expects preprocessed data, time export model use elsewhere (web browser, mobile app), ’ll need reimplement exact preprocessing pipeline. gets tricky quickly. least possible amount preprocessing hitting model. , ’ll image resizing data pipeline (deep neural network can process contiguous batches data), ’ll input value scaling part model, create . Let’s resize images 150x150:","code":"resize_fn <- layer_resizing(width = 150, height = 150) resize_pair <- function(x, y) list(resize_fn(x), y)  train_ds <- train_ds |> dataset_map(resize_pair) validation_ds <- validation_ds |> dataset_map(resize_pair) test_ds <- test_ds |> dataset_map(resize_pair)"},{"path":"https://keras3.posit.co/dev/articles/transfer_learning.html","id":"using-random-data-augmentation","dir":"Articles","previous_headings":"An end-to-end example: fine-tuning an image classification model on a cats vs. dogs dataset","what":"Using random data augmentation","title":"Transfer learning & fine-tuning","text":"don’t large image dataset, ’s good practice artificially introduce sample diversity applying random yet realistic transformations training images, random horizontal flipping small random rotations. helps expose model different aspects training data slowing overfitting. Let’s batch data use prefetching optimize loading speed. Let’s visualize first image first batch looks like various random transformations:","code":"data_augmentation <- keras_model_sequential() |>   layer_random_flip(\"horizontal\") |>   layer_random_rotation(.1)  train_ds <- train_ds %>%   dataset_map(function(x, y) list(data_augmentation(x), y)) library(tensorflow, exclude = c(\"shape\", \"set_random_seed\")) batch_size <- 64  train_ds <- train_ds |>   dataset_batch(batch_size) |>   dataset_prefetch()  validation_ds <- validation_ds |>   dataset_batch(batch_size) |>   dataset_prefetch()  test_ds <- test_ds |>   dataset_batch(batch_size) |>   dataset_prefetch() batch <- train_ds |>   dataset_take(1) |>   as_iterator() |>   iter_next()  c(images, labels) %<-% batch first_image <- images[1, all_dims(), drop = TRUE] augmented_image <- data_augmentation(first_image, training = TRUE)  plot_image <- function(image, main = deparse1(substitute(image))) {   image |>     as.array() |> # convert from tensor to R array     as.raster(max = 255) |>     plot()    if(!is.null(main))     title(main) }  par(mfrow = c(2, 2), mar = c(1, 1, 1.5, 1)) plot_image(first_image) plot_image(augmented_image) plot_image(data_augmentation(first_image, training = TRUE), \"augmented 2\") plot_image(data_augmentation(first_image, training = TRUE), \"augmented 3\")"},{"path":"https://keras3.posit.co/dev/articles/transfer_learning.html","id":"build-a-model","dir":"Articles","previous_headings":"","what":"Build a model","title":"Transfer learning & fine-tuning","text":"Now let’s built model follows blueprint ’ve explained earlier. Note : add Rescaling layer scale input values (initially [0, 255] range) [-1, 1] range. add Dropout layer classification layer, regularization. make sure pass training=FALSE calling base model, runs inference mode, batchnorm statistics don’t get updated even unfreeze base model fine-tuning.","code":"base_model <- application_xception(   weights = \"imagenet\", # Load weights pre-trained on ImageNet.   input_shape = c(150, 150, 3),   include_top = FALSE, # Do not include the ImageNet classifier at the top. )  # Freeze the base_model base_model$trainable <- FALSE  # Create new model on top inputs <- keras_input(shape = c(150, 150, 3))  # Pre-trained Xception weights requires that input be scaled # from (0, 255) to a range of (-1., +1.), the rescaling layer # outputs: `(inputs * scale) + offset` scale_layer <- layer_rescaling(scale = 1 / 127.5, offset = -1) x <- scale_layer(inputs)  # The base model contains batchnorm layers. We want to keep them in inference mode # when we unfreeze the base model for fine-tuning, so we make sure that the # base_model is running in inference mode here. outputs <- x |>   base_model(training = FALSE) |>   layer_global_average_pooling_2d() |>   layer_dropout(0.2) |>   layer_dense(1)  model <- keras_model(inputs, outputs)  summary(model, show_trainable = TRUE) ## Model: \"functional_8\" ## ┏━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━┳━━━━━━━┓ ## ┃ Layer (type)                ┃ Output Shape          ┃    Param # ┃ Trai… ┃ ## ┡━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━╇━━━━━━━┩ ## │ input_layer_7 (InputLayer)  │ (None, 150, 150, 3)   │          0 │   -   │ ## ├─────────────────────────────┼───────────────────────┼────────────┼───────┤ ## │ rescaling (Rescaling)       │ (None, 150, 150, 3)   │          0 │   -   │ ## ├─────────────────────────────┼───────────────────────┼────────────┼───────┤ ## │ xception (Functional)       │ (None, 5, 5, 2048)    │ 20,861,480 │   N   │ ## ├─────────────────────────────┼───────────────────────┼────────────┼───────┤ ## │ global_average_pooling2d_1  │ (None, 2048)          │          0 │   -   │ ## │ (GlobalAveragePooling2D)    │                       │            │       │ ## ├─────────────────────────────┼───────────────────────┼────────────┼───────┤ ## │ dropout (Dropout)           │ (None, 2048)          │          0 │   -   │ ## ├─────────────────────────────┼───────────────────────┼────────────┼───────┤ ## │ dense_8 (Dense)             │ (None, 1)             │      2,049 │   Y   │ ## └─────────────────────────────┴───────────────────────┴────────────┴───────┘ ##  Total params: 20,863,529 (79.59 MB) ##  Trainable params: 2,049 (8.00 KB) ##  Non-trainable params: 20,861,480 (79.58 MB)"},{"path":"https://keras3.posit.co/dev/articles/transfer_learning.html","id":"train-the-top-layer","dir":"Articles","previous_headings":"","what":"Train the top layer","title":"Transfer learning & fine-tuning","text":"","code":"model |> compile(   optimizer = optimizer_adam(),   loss = loss_binary_crossentropy(from_logits = TRUE),   metrics = list(metric_binary_accuracy()) )  epochs <- 1 model |> fit(train_ds, epochs = epochs, validation_data = validation_ds) ## 146/146 - 57s - 389ms/step - binary_accuracy: 0.9183 - loss: 0.1887 - val_binary_accuracy: 0.9669 - val_loss: 0.0926"},{"path":"https://keras3.posit.co/dev/articles/transfer_learning.html","id":"do-a-round-of-fine-tuning-of-the-entire-model","dir":"Articles","previous_headings":"","what":"Do a round of fine-tuning of the entire model","title":"Transfer learning & fine-tuning","text":"Finally, let’s unfreeze base model train entire model end--end low learning rate. Importantly, although base model becomes trainable, still running inference mode since passed training=FALSE calling built model. means batch normalization layers inside won’t update batch statistics. , wreck havoc representations learned model far. 10 epochs, fine-tuning gains us nice improvement . Let’s evaluate model test dataset:","code":"# Unfreeze the base_model. Note that it keeps running in inference mode # since we passed `training=FALSE` when calling it. This means that # the batchnorm layers will not update their batch statistics. # This prevents the batchnorm layers from undoing all the training # we've done so far. base_model$trainable <- TRUE summary(model, show_trainable = TRUE) ## Model: \"functional_8\" ## ┏━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━┳━━━━━━━┓ ## ┃ Layer (type)                ┃ Output Shape          ┃    Param # ┃ Trai… ┃ ## ┡━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━╇━━━━━━━┩ ## │ input_layer_7 (InputLayer)  │ (None, 150, 150, 3)   │          0 │   -   │ ## ├─────────────────────────────┼───────────────────────┼────────────┼───────┤ ## │ rescaling (Rescaling)       │ (None, 150, 150, 3)   │          0 │   -   │ ## ├─────────────────────────────┼───────────────────────┼────────────┼───────┤ ## │ xception (Functional)       │ (None, 5, 5, 2048)    │ 20,861,480 │   Y   │ ## ├─────────────────────────────┼───────────────────────┼────────────┼───────┤ ## │ global_average_pooling2d_1  │ (None, 2048)          │          0 │   -   │ ## │ (GlobalAveragePooling2D)    │                       │            │       │ ## ├─────────────────────────────┼───────────────────────┼────────────┼───────┤ ## │ dropout (Dropout)           │ (None, 2048)          │          0 │   -   │ ## ├─────────────────────────────┼───────────────────────┼────────────┼───────┤ ## │ dense_8 (Dense)             │ (None, 1)             │      2,049 │   Y   │ ## └─────────────────────────────┴───────────────────────┴────────────┴───────┘ ##  Total params: 20,867,629 (79.60 MB) ##  Trainable params: 20,809,001 (79.38 MB) ##  Non-trainable params: 54,528 (213.00 KB) ##  Optimizer params: 4,100 (16.02 KB) model |> compile(   optimizer = optimizer_adam(1e-5), # Low learning rate   loss = loss_binary_crossentropy(from_logits = TRUE),   metrics = list(metric_binary_accuracy()) )  epochs <- 1 model |> fit(train_ds, epochs = epochs, validation_data = validation_ds) ## 146/146 - 86s - 592ms/step - binary_accuracy: 0.8660 - loss: 0.3212 - val_binary_accuracy: 0.9652 - val_loss: 0.1022 model |> evaluate(test_ds) ## 37/37 - 2s - 43ms/step - binary_accuracy: 0.9540 - loss: 0.1102 ## $binary_accuracy ## [1] 0.9539983 ## ## $loss ## [1] 0.1102476"},{"path":"https://keras3.posit.co/dev/articles/understanding_masking_and_padding.html","id":"setup","dir":"Articles","previous_headings":"","what":"Setup","title":"Understanding masking & padding","text":"","code":"library(keras3)"},{"path":"https://keras3.posit.co/dev/articles/understanding_masking_and_padding.html","id":"introduction","dir":"Articles","previous_headings":"","what":"Introduction","title":"Understanding masking & padding","text":"Masking way tell sequence-processing layers certain timesteps input missing, thus skipped processing data. Padding special form masking masked steps start end sequence. Padding comes need encode sequence data contiguous batches: order make sequences batch fit given standard length, necessary pad truncate sequences. Let’s take close look.","code":""},{"path":"https://keras3.posit.co/dev/articles/understanding_masking_and_padding.html","id":"padding-sequence-data","dir":"Articles","previous_headings":"","what":"Padding sequence data","title":"Understanding masking & padding","text":"processing sequence data, common individual samples different lengths. Consider following example (text tokenized words): vocabulary lookup, data might vectorized integers, e.g.: data nested list individual samples length 3, 5, 6, respectively. Since input data deep learning model must single tensor (shape e.g. (batch_size, 6, vocab_size) case), samples shorter longest item need padded placeholder value (alternatively, one might also truncate long samples padding short samples). Keras provides utility function truncate pad Python lists common length: pad_sequences.","code":"data <- list(   c(\"Hello\", \"world\", \"!\"),   c(\"How\", \"are\", \"you\", \"doing\", \"today\"),   c(\"The\", \"weather\", \"will\", \"be\", \"nice\", \"tomorrow\") ) data <- list(   c(71, 1331, 4231),   c(73, 8, 3215, 55, 927),   c(83, 91, 1, 645, 1253, 927) ) raw_inputs <- list(   c(711, 632, 71),   c(73, 8, 3215, 55, 927),   c(83, 91, 1, 645, 1253, 927) )  # By default, this will pad using 0s; it is configurable via the # \"value\" parameter. # Note that you could use \"pre\" padding (at the beginning) or # \"post\" padding (at the end). # We recommend using \"post\" padding when working with RNN layers # (in order to be able to use the # CuDNN implementation of the layers). padded_inputs <- pad_sequences(raw_inputs, padding=\"post\") padded_inputs ##      [,1] [,2] [,3] [,4] [,5] [,6] ## [1,]  711  632   71    0    0    0 ## [2,]   73    8 3215   55  927    0 ## [3,]   83   91    1  645 1253  927"},{"path":"https://keras3.posit.co/dev/articles/understanding_masking_and_padding.html","id":"masking","dir":"Articles","previous_headings":"","what":"Masking","title":"Understanding masking & padding","text":"Now samples uniform length, model must informed part data actually padding ignored. mechanism masking. three ways introduce input masks Keras models: Add layer_masking layer. Configure layer_embedding layer mask_zero=TRUE. Pass mask argument manually calling layers support argument (e.g. RNN layers).","code":""},{"path":"https://keras3.posit.co/dev/articles/understanding_masking_and_padding.html","id":"mask-generating-layers-embedding-and-masking","dir":"Articles","previous_headings":"","what":"Mask-generating layers: Embedding and Masking","title":"Understanding masking & padding","text":"hood, layers create mask tensor (2D tensor shape (batch, sequence_length)), attach tensor output returned Masking Embedding layer. can see printed result, mask 2D boolean tensor shape (batch_size, sequence_length), individual FALSE entry indicates corresponding timestep ignored processing.","code":"embedding <- layer_embedding(input_dim=5000, output_dim=16, mask_zero=TRUE) masked_output <- embedding(padded_inputs)  masked_output$`_keras_mask` ## tf.Tensor( ## [[ True  True  True False False False] ##  [ True  True  True  True  True False] ##  [ True  True  True  True  True  True]], shape=(3, 6), dtype=bool) masking_layer <- layer_masking() # Simulate the embedding lookup by expanding the 2D input to 3D, # with embedding dimension of 10. unmasked_embedding <- op_cast(     op_tile(op_expand_dims(padded_inputs, axis=-1), c(1L, 1L, 10L)),     dtype=\"float32\" )  masked_embedding <- masking_layer(unmasked_embedding) masked_embedding$`_keras_mask` ## tf.Tensor( ## [[ True  True  True False False False] ##  [ True  True  True  True  True False] ##  [ True  True  True  True  True  True]], shape=(3, 6), dtype=bool)"},{"path":"https://keras3.posit.co/dev/articles/understanding_masking_and_padding.html","id":"mask-propagation-in-the-functional-api-and-sequential-api","dir":"Articles","previous_headings":"","what":"Mask propagation in the Functional API and Sequential API","title":"Understanding masking & padding","text":"using Functional API Sequential API, mask generated Embedding Masking layer propagated network layer capable using (example, RNN layers). Keras automatically fetch mask corresponding input pass layer knows use . instance, following Sequential model, LSTM layer automatically receive mask, means ignore padded values: also case following Functional API model:","code":"model <- keras_model_sequential() %>%   layer_embedding(input_dim=5000, output_dim=16, mask_zero=TRUE) %>%   layer_lstm(units=32) inputs <- keras_input(shape = shape(NULL), dtype=\"int32\") outputs <- inputs %>%   layer_embedding(input_dim=5000, output_dim=16, mask_zero=TRUE) %>%   layer_lstm(units=32)  model <- keras_model(inputs, outputs)"},{"path":"https://keras3.posit.co/dev/articles/understanding_masking_and_padding.html","id":"passing-mask-tensors-directly-to-layers","dir":"Articles","previous_headings":"","what":"Passing mask tensors directly to layers","title":"Understanding masking & padding","text":"Layers can handle masks (LSTM layer) mask argument call method. Meanwhile, layers produce mask (e.g. Embedding) expose compute_mask(input, previous_mask) method can call. Thus, can pass output compute_mask() method mask-producing layer call method mask-consuming layer, like :","code":"MyLayer <- new_layer_class(   \"MyLayer\",   initialize = function(...) {     super$initialize(...)     self$embedding <- layer_embedding(       input_dim=5000, output_dim=16, mask_zero=TRUE     )     self$lstm <- layer_lstm(units=32)   },   call = function(inputs) {     inputs %>%       self$embedding() %>%       # Note that you could also prepare a `mask` tensor manually.       # It only needs to be a boolean tensor       # with the right shape, i.e. (batch_size, timesteps).       self$lstm(mask=self$embedding$compute_mask(inputs))   } )  layer <- MyLayer() x <- random_integer(c(32, 10), 0, 100) layer(x) ## tf.Tensor( ## [[ 0.00130048 -0.00113367 -0.00715671 ... -0.00107615 -0.00162071 ##    0.00135018] ##  [-0.004185    0.00726349  0.00520932 ...  0.00119117  0.00230441 ##    0.00174123] ##  [-0.00537032 -0.00164898 -0.00238435 ... -0.00154158 -0.0038603 ##   -0.00105811] ##  ... ##  [ 0.00622133 -0.00905907 -0.00599518 ...  0.00025823 -0.00142478 ##   -0.00125036] ##  [-0.00523904  0.00336683 -0.00299453 ...  0.00876719  0.00172074 ##    0.00903089] ##  [-0.00393721  0.00058538  0.00503809 ... -0.00203075  0.00325885 ##   -0.00299755]], shape=(32, 32), dtype=float32)"},{"path":"https://keras3.posit.co/dev/articles/understanding_masking_and_padding.html","id":"supporting-masking-in-your-custom-layers","dir":"Articles","previous_headings":"","what":"Supporting masking in your custom layers","title":"Understanding masking & padding","text":"Sometimes, may need write layers generate mask (like Embedding), layers need modify current mask. instance, layer produces tensor different time dimension input, Concatenate layer concatenates time dimension, need modify current mask downstream layers able properly take masked timesteps account. , layer implement layer.compute_mask() method, produces new mask given input current mask. example TemporalSplit layer needs modify current mask. another example CustomEmbedding layer capable generating mask input values: Note: details format limitations related masking, see serialization guide.","code":"TemporalSplit <- new_layer_class(   \"TemporalSplit\",   call = function(inputs) {     # Expect the input to be 3D and mask to be 2D, split the input tensor into 2     # subtensors along the time axis (axis 1).     op_split(inputs, 2, axis=2)   },   compute_mask = function(inputs, mask = NULL) {     # Also split the mask into 2 if it presents.     if (!is.null(mask)) {       op_split(mask, 2, axis=2)     } else {       NULL     }   } )  c(first_half, second_half) %<-% TemporalSplit(masked_embedding) first_half$`_keras_mask` ## tf.Tensor( ## [[ True  True  True] ##  [ True  True  True] ##  [ True  True  True]], shape=(3, 3), dtype=bool) second_half$`_keras_mask` ## tf.Tensor( ## [[False False False] ##  [ True  True False] ##  [ True  True  True]], shape=(3, 3), dtype=bool) CustomEmbedding <- new_layer_class(   \"CustomEmbedding\",   initialize = function(input_dim, output_dim, mask_zero=FALSE, ...) {     super$initialize(...)     self$input_dim <- as.integer(input_dim)     self$output_dim <- as.integer(output_dim)     self$mask_zero <- mask_zero   },   build = function(input_shape) {     self$embeddings <- self$add_weight(       shape=c(self$input_dim, self$output_dim),       initializer=\"random_normal\",       dtype=\"float32\"     )   },   call = function(inputs) {     inputs <- op_cast(inputs, \"int32\")     op_take(self$embeddings, inputs)   },   compute_mask = function(inputs, mask=NULL) {     if (!self$mask_zero) {       NULL     } else {       op_not_equal(inputs, 0)     }   } )  layer <- CustomEmbedding(input_dim = 10, output_dim = 32, mask_zero=TRUE) x <- random_integer(c(3, 10), 0, 9)  y <- layer(x) mask <- layer$compute_mask(x)  mask ## tf.Tensor( ## [[ True  True  True  True  True  True  True  True  True  True] ##  [ True  True  True  True  True  True  True  True False  True] ##  [ True  True  True  True False  True  True  True  True  True]], shape=(3, 10), dtype=bool)"},{"path":"https://keras3.posit.co/dev/articles/understanding_masking_and_padding.html","id":"opting-in-to-mask-propagation-on-compatible-layers","dir":"Articles","previous_headings":"","what":"Opting-in to mask propagation on compatible layers","title":"Understanding masking & padding","text":"layers don’t modify time dimension, don’t need modify current mask. However, may still want able propagate current mask, unchanged, next layer. opt-behavior. default, custom layer destroy current mask (since framework way tell whether propagating mask safe ). custom layer modify time dimension, want able propagate current input mask, set self.supports_masking = True layer constructor. case, default behavior compute_mask() just pass current mask . ’s example layer whitelisted mask propagation: can now use custom layer -mask-generating layer (like Embedding) mask-consuming layer (like LSTM), pass mask along reaches mask-consuming layer.","code":"MyActivation <- new_layer_class(   \"MyActivation\",   initialize = function(...) {     super$initialize(...)     self$supports_masking <- TRUE   },   call = function(inputs) {     op_relu(inputs)   } ) inputs <- keras_input(shape = shape(NULL), dtype=\"int32\") outputs <- inputs %>%   layer_embedding(input_dim=5000, output_dim=16, mask_zero=TRUE) %>%   MyActivation() %>%   layer_lstm(units=32)  model <- keras_model(inputs, outputs) y <- model(random_integer(c(32, 100), 0, 5000))"},{"path":"https://keras3.posit.co/dev/articles/understanding_masking_and_padding.html","id":"writing-layers-that-need-mask-information","dir":"Articles","previous_headings":"","what":"Writing layers that need mask information","title":"Understanding masking & padding","text":"layers mask consumers: accept mask argument call use determine whether skip certain time steps. write layer, can simply add mask=None argument call signature. mask associated inputs passed layer whenever available. ’s simple example : layer computes softmax time dimension (axis 1) input sequence, discarding masked timesteps.","code":"TemporalSoftmax <- new_layer_class(   \"TemporalSoftmax\",   initialize = function(...) {     super$initialize(...)     self$supports_masking <- TRUE   },   call = function(inputs, mask=NULL) {     if (is.null(mask)) {       stop(\"`TemporalSoftmax` layer requires a previous layer to support masking.\")     }     broadcast_float_mask <- op_expand_dims(op_cast(mask, \"float32\"), -1)     inputs_exp <- op_exp(inputs) * broadcast_float_mask     inputs_sum <- op_sum(inputs_exp * broadcast_float_mask, axis=-1, keepdims=TRUE)     inputs_exp / inputs_sum   } )  inputs <- keras_input(shape = shape(NULL), dtype=\"int32\") outputs <- inputs %>%   layer_embedding(input_dim=10, output_dim=32, mask_zero=TRUE) %>%   layer_dense(1) %>%   TemporalSoftmax()  model <- keras_model(inputs, outputs) y <- model(random_integer(c(32, 100), 0, 10))"},{"path":"https://keras3.posit.co/dev/articles/understanding_masking_and_padding.html","id":"summary","dir":"Articles","previous_headings":"","what":"Summary","title":"Understanding masking & padding","text":"need know padding & masking Keras. recap: “Masking” layers able know skip / ignore certain timesteps sequence inputs. layers mask-generators: Embedding can generate mask input values (mask_zero=TRUE), can Masking layer. layers mask-consumers: expose mask argument call method. case RNN layers. Functional API Sequential API, mask information propagated automatically. using layers standalone way, can pass mask arguments layers manually. can easily write layers modify current mask, generate new mask, consume mask associated inputs.","code":""},{"path":"https://keras3.posit.co/dev/articles/writing_a_custom_training_loop_in_tensorflow.html","id":"setup","dir":"Articles","previous_headings":"","what":"Setup","title":"Writing a training loop from scratch in TensorFlow","text":"","code":"library(tensorflow, exclude = c(\"shape\", \"set_random_seed\")) library(tfdatasets, exclude = \"shape\") library(keras3)  # This guide can only be run with the TensorFlow backend. use_backend(\"tensorflow\")"},{"path":"https://keras3.posit.co/dev/articles/writing_a_custom_training_loop_in_tensorflow.html","id":"introduction","dir":"Articles","previous_headings":"","what":"Introduction","title":"Writing a training loop from scratch in TensorFlow","text":"Keras provides default training evaluation loops, fit() evaluate(). usage covered guide Training & evaluation built-methods. want customize learning algorithm model still leveraging convenience fit() (instance, train GAN using fit()), can subclass Model class implement train_step() method, called repeatedly fit(). covered guide Customizing happens fit(). Now, want low-level control training & evaluation, write training & evaluation loops scratch. guide .","code":""},{"path":"https://keras3.posit.co/dev/articles/writing_a_custom_training_loop_in_tensorflow.html","id":"using-the-gradienttape-a-first-end-to-end-example","dir":"Articles","previous_headings":"","what":"Using the GradientTape: a first end-to-end example","title":"Writing a training loop from scratch in TensorFlow","text":"Calling model inside GradientTape scope enables retrieve gradients trainable weights layer respect loss value. Using optimizer instance, can use gradients update variables (can retrieve using model$trainable_weights). Let’s consider simple MNIST model: Let’s train using mini-batch gradient custom training loop. First, ’re going need optimizer, loss function, dataset: ’s training loop: open loop iterates epochs epoch, iterate dataset, batches batch, open GradientTape() scope Inside scope, call model (forward pass) compute loss Outside scope, retrieve gradients weights model regard loss Finally, use optimizer update weights model based gradients","code":"get_model <- function() {   inputs <- keras_input(shape = 784, name = \"digits\")   outputs <- inputs |>     layer_dense(units = 64, activation = \"relu\") |>     layer_dense(units = 64, activation = \"relu\") |>     layer_dense(units = 10, name = \"predictions\")   keras_model(inputs = inputs, outputs = outputs) }  model <- get_model() # Instantiate an optimizer. optimizer <- optimizer_adam(learning_rate = 1e-3) # Instantiate a loss function. loss_fn <- loss_sparse_categorical_crossentropy(from_logits = TRUE)  # Prepare the training dataset. batch_size <- 64 c(c(x_train, y_train), c(x_test, y_test)) %<-% dataset_mnist() x_train <- array_reshape(x_train, c(-1, 784)) x_test <- array_reshape(x_test, c(-1, 784))  # Reserve 10,000 samples for validation. val_i <- sample.int(nrow(x_train), 10000) x_val <- x_train[val_i,] y_val <- y_train[val_i] x_train %<>% .[-val_i,] y_train %<>% .[-val_i]  # Prepare the training dataset. train_dataset <- list(x_train, y_train) |>   tensor_slices_dataset() |>   dataset_shuffle(buffer_size = 1024) |>   dataset_batch(batch_size)  # Prepare the validation dataset. val_dataset <- list(x_val, y_val) |>   tensor_slices_dataset() |>   dataset_batch(batch_size) epochs <- 3 for (epoch in seq_len(epochs)) {   cat(\"Start of epoch \", epoch, \"\\n\")    # Iterate over the batches of the dataset.   step <- 0   iterator <- as_iterator(train_dataset)   while (!is.null(batch <- iter_next(iterator))) {     c(x_batch_train, y_batch_train) %<-% batch     step <- step + 1     # Open a GradientTape to record the operations run     # during the forward pass, which enables auto-differentiation.     with(tf$GradientTape() %as% tape, {       # Run the forward pass of the layer.       # The operations that the layer applies       # to its inputs are going to be recorded       # on the GradientTape.       logits <- model(x_batch_train, training = TRUE) # Logits for this minibatch        # Compute the loss value for this minibatch.       loss_value <- loss_fn(y_batch_train, logits)     })      # Use the gradient tape to automatically retrieve     # the gradients of the trainable variables with respect to the loss.     gradients <- tape$gradient(loss_value, model$trainable_weights)      # Run one step of gradient descent by updating     # the value of the variables to minimize the loss.     optimizer$apply(gradients, model$trainable_weights)      # Log every 200 batches.     if (step %% 200 == 0) {       cat(sprintf(\"Training loss (for one batch) at step %d: %.4f\\n\",                   step, loss_value))       cat(sprintf(\"Seen so far: %d samples\\n\", (step * batch_size)))     }   } } ## Start of epoch  1 ## Training loss (for one batch) at step 200: 1.1675 ## Seen so far: 12800 samples ## Training loss (for one batch) at step 400: 1.6450 ## Seen so far: 25600 samples ## Training loss (for one batch) at step 600: 0.7421 ## Seen so far: 38400 samples ## Start of epoch  2 ## Training loss (for one batch) at step 200: 0.7254 ## Seen so far: 12800 samples ## Training loss (for one batch) at step 400: 0.2476 ## Seen so far: 25600 samples ## Training loss (for one batch) at step 600: 0.3578 ## Seen so far: 38400 samples ## Start of epoch  3 ## Training loss (for one batch) at step 200: 0.2974 ## Seen so far: 12800 samples ## Training loss (for one batch) at step 400: 0.4441 ## Seen so far: 25600 samples ## Training loss (for one batch) at step 600: 0.1461 ## Seen so far: 38400 samples"},{"path":"https://keras3.posit.co/dev/articles/writing_a_custom_training_loop_in_tensorflow.html","id":"low-level-handling-of-metrics","dir":"Articles","previous_headings":"","what":"Low-level handling of metrics","title":"Writing a training loop from scratch in TensorFlow","text":"Let’s add metrics monitoring basic loop. can readily reuse built-metrics (custom ones wrote) training loops written scratch. ’s flow: Instantiate metric start loop Call metric$update_state() batch Call metric$result() need display current value metric Call metric$reset_state() need clear state metric (typically end epoch) Let’s use knowledge compute SparseCategoricalAccuracy validation data end epoch: ’s training & evaluation loop:","code":"# Get a fresh model model <- get_model()  # Instantiate an optimizer to train the model. optimizer <- optimizer_adam(learning_rate = 1e-3) # Instantiate a loss function. loss_fn <- loss_sparse_categorical_crossentropy(from_logits = TRUE)  # Prepare the metrics. train_acc_metric <- metric_sparse_categorical_accuracy() val_acc_metric <- metric_sparse_categorical_accuracy() epochs <- 2 time <- Sys.time() for (epoch in seq_len(epochs)) {   cat(\"Start of epoch \", epoch, \"\\n\")    # Iterate over the batches of the dataset.   step <- 0   train_dataset_iterator <- as_iterator(train_dataset)   while (!is.null(train_batch <- iter_next(train_dataset_iterator))) {     c(x_batch_train, y_batch_train) %<-% train_batch     step <- step + 1      with(tf$GradientTape() %as% tape, {       logits <- model(x_batch_train, training = TRUE)       loss_value <- loss_fn(y_batch_train, logits)     })     gradients <- tape$gradient(loss_value, model$trainable_weights)     optimizer$apply(gradients, model$trainable_weights)      # Update training metric.     train_acc_metric$update_state(y_batch_train, logits)     if (step %% 200 == 0) {       cat(sprintf(         \"Training loss (for one batch) at step %d: %.4f\\n\", step, loss_value))       cat(sprintf(\"Seen so far: %d samples \\n\", (step * batch_size)))     }   }    # Display metrics at the end of each epoch.   train_acc <- train_acc_metric$result()   cat(sprintf(\"Training acc over epoch: %.4f\\n\", train_acc))    # Reset training metrics at the end of each epoch   train_acc_metric$reset_state()    # Run a validation loop at the end of each epoch.   iterate(val_dataset, function(val_batch) {     c(x_batch_val, y_batch_val) %<-% val_batch     val_logits <- model(x_batch_val, training = FALSE)     val_acc_metric$update_state(y_batch_val, val_logits)   })   val_acc <- val_acc_metric$result()   val_acc_metric$reset_state()   cat(sprintf(\"Validation acc: %.4f\\n\", val_acc)) } ## Start of epoch  1 ## Training loss (for one batch) at step 200: 1.9564 ## Seen so far: 12800 samples ## Training loss (for one batch) at step 400: 1.3562 ## Seen so far: 25600 samples ## Training loss (for one batch) at step 600: 0.4850 ## Seen so far: 38400 samples ## Training acc over epoch: 0.7836 ## Validation acc: 0.8681 ## Start of epoch  2 ## Training loss (for one batch) at step 200: 0.7679 ## Seen so far: 12800 samples ## Training loss (for one batch) at step 400: 0.5674 ## Seen so far: 25600 samples ## Training loss (for one batch) at step 600: 0.4705 ## Seen so far: 38400 samples ## Training acc over epoch: 0.8838 ## Validation acc: 0.9049 Sys.time() - time ## Time difference of 39.32149 secs"},{"path":"https://keras3.posit.co/dev/articles/writing_a_custom_training_loop_in_tensorflow.html","id":"speeding-up-your-training-step-with-tf_function","dir":"Articles","previous_headings":"","what":"Speeding-up your training step with tf_function()","title":"Writing a training loop from scratch in TensorFlow","text":"default runtime TensorFlow 2 eager execution. , training loop executes eagerly. great debugging, graph compilation definite performance advantage. Describing computation static graph enables framework apply global performance optimizations. impossible framework constrained greedily execute one operation another, knowledge comes next. can compile static graph function takes tensors input. Just add @tf.function decorator , like : Let’s evaluation step: Now, let’s re-run training loop compiled training step: Much faster, isn’t ?","code":"train_step <- tf_function(function(x, y) {   with(tf$GradientTape() %as% tape, {     logits <- model(x, training = TRUE)     loss_value <- loss_fn(y, logits)   })   gradients <- tape$gradient(loss_value, model$trainable_weights)   optimizer$apply(gradients, model$trainable_weights)   train_acc_metric$update_state(y, logits)   invisible(NULL) # return nothing }) test_step <- tf_function(function(x, y) {   val_logits <- model(x, training=FALSE)   val_acc_metric$update_state(y, val_logits)   invisible(NULL) # return nothing }) epochs <- 2 time <- Sys.time() for (epoch in seq_len(epochs)) {   cat(\"Start of epoch \", epoch, \"\\n\")    # Iterate over the batches of the dataset.   step <- 0   while (!is.null(batch <- iter_next(iterator))) {     c(x_batch_train, y_batch_train) %<-% batch     step <- step + 1     train_step(x_batch_train, y_batch_train)      if (step %% 200 == 0) {       cat(sprintf(         \"Training loss (for one batch) at step %d: %.4f\\n\", step, loss_value       ))       cat(sprintf(\"Seen so far: %d samples \\n\", (step * batch_size)))     }   }    # Display metrics at the end of each epoch.   train_acc <- train_acc_metric$result()   cat(sprintf(\"Training acc over epoch: %.4f\\n\", train_acc))     # Reset training metrics at the end of each epoch   train_acc_metric$reset_state()    # Run a validation loop at the end of each epoch.    iterate(val_dataset, function(val_batch) {     c(x_batch_val, y_batch_val) %<-% val_batch     test_step(x_batch_val, y_batch_val)   })    val_acc <- val_acc_metric$result()   val_acc_metric$reset_state()   cat(sprintf(\"Validation acc: %.4f\\n\", val_acc)) } ## Start of epoch  1 ## Training acc over epoch: 0.0000 ## Validation acc: 0.9049 ## Start of epoch  2 ## Training acc over epoch: 0.0000 ## Validation acc: 0.9049 Sys.time() - time ## Time difference of 0.4121809 secs"},{"path":"https://keras3.posit.co/dev/articles/writing_a_custom_training_loop_in_tensorflow.html","id":"low-level-handling-of-losses-tracked-by-the-model","dir":"Articles","previous_headings":"","what":"Low-level handling of losses tracked by the model","title":"Writing a training loop from scratch in TensorFlow","text":"Layers models recursively track losses created forward pass layers call self$add_loss(value). resulting list scalar loss values available via property model$losses end forward pass. want using loss components, sum add main loss training step. Consider layer, creates activity regularization loss: Let’s build really simple model uses : ’s training step look like now:","code":"layer_activity_regularization <- Layer(   \"ActivityRegularizationLayer\",   call = function(inputs) {     self$add_loss(0.1 * op_mean(inputs))     inputs   } ) inputs <- keras_input(shape = 784, name=\"digits\") outputs <- inputs |>   layer_dense(units = 64, activation = \"relu\") |>   layer_activity_regularization() |>   layer_dense(units = 64, activation = \"relu\") |>   layer_dense(units = 10, name = \"predictions\") model <- keras_model(inputs = inputs, outputs = outputs) train_step <- tf_function(function(x, y) {   with(tf$GradientTape() %as% tape, {     logits <- model(x, training = TRUE)     loss_value <- Reduce(`+`, c(loss_fn(y, logits),                                 model$losses))   })   gradients <- tape$gradient(loss_value, model$trainable_weights)   optimizer$apply(gradients, model$trainable_weights)   train_acc_metric$update_state(y, logits)   invisible(NULL) })"},{"path":"https://keras3.posit.co/dev/articles/writing_a_custom_training_loop_in_tensorflow.html","id":"summary","dir":"Articles","previous_headings":"","what":"Summary","title":"Writing a training loop from scratch in TensorFlow","text":"Now know everything know using built-training loops writing scratch. conclude, ’s simple end--end example ties together everything ’ve learned guide: DCGAN trained MNIST digits.","code":""},{"path":"https://keras3.posit.co/dev/articles/writing_a_custom_training_loop_in_tensorflow.html","id":"end-to-end-example-a-gan-training-loop-from-scratch","dir":"Articles","previous_headings":"","what":"End-to-end example: a GAN training loop from scratch","title":"Writing a training loop from scratch in TensorFlow","text":"may familiar Generative Adversarial Networks (GANs). GANs can generate new images look almost real, learning latent distribution training dataset images (“latent space” images). GAN made two parts: “generator” model maps points latent space points image space, “discriminator” model, classifier can tell difference real images (training dataset) fake images (output generator network). GAN training loop looks like : Train discriminator. Sample batch random points latent space. Turn points fake images via “generator” model. Get batch real images combine generated images. Train “discriminator” model classify generated vs. real images. Train generator. Sample random points latent space. Turn points fake images via “generator” network. Get batch real images combine generated images. Train “generator” model “fool” discriminator classify fake images real. much detailed overview GANs works, see Deep Learning Python. Let’s implement training loop. First, create discriminator meant classify fake vs real digits: let’s create generator network, turns latent vectors outputs shape (28, 28, 1) (representing MNIST digits): ’s key bit: training loop. can see quite straightforward. training step function takes 17 lines. Let’s train GAN, repeatedly calling train_step batches images. Since discriminator generator convnets, ’re going want run code GPU. ’s ! ’ll get nice-looking fake MNIST digits just ~30s training GPU.","code":"# Create the discriminator discriminator <-   keras_model_sequential(name = \"discriminator\",                          input_shape = c(28, 28, 1)) |>   layer_conv_2d(filters = 64, kernel_size = c(3, 3),                 strides = c(2, 2), padding = \"same\") |>   layer_activation_leaky_relu(negative_slope = 0.2) |>   layer_conv_2d(filters = 128, kernel_size = c(3, 3),                 strides = c(2, 2), padding = \"same\") |>   layer_activation_leaky_relu(negative_slope = 0.2) |>   layer_global_max_pooling_2d() |>   layer_dense(units = 1)  summary(discriminator) ## Model: \"discriminator\" ## ┏━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━┓ ## ┃ Layer (type)                    ┃ Output Shape           ┃       Param # ┃ ## ┡━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━┩ ## │ conv2d (Conv2D)                 │ (None, 14, 14, 64)     │           640 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ leaky_re_lu (LeakyReLU)         │ (None, 14, 14, 64)     │             0 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ conv2d_1 (Conv2D)               │ (None, 7, 7, 128)      │        73,856 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ leaky_re_lu_1 (LeakyReLU)       │ (None, 7, 7, 128)      │             0 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ global_max_pooling2d            │ (None, 128)            │             0 │ ## │ (GlobalMaxPooling2D)            │                        │               │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ dense_6 (Dense)                 │ (None, 1)              │           129 │ ## └─────────────────────────────────┴────────────────────────┴───────────────┘ ##  Total params: 74,625 (291.50 KB) ##  Trainable params: 74,625 (291.50 KB) ##  Non-trainable params: 0 (0.00 B) latent_dim <- 128L  generator <-   keras_model_sequential(name = \"generator\",                          input_shape = latent_dim) |>   layer_dense(7 * 7 * 128) |>   layer_activation_leaky_relu(negative_slope = 0.2) |>   layer_reshape(target_shape = c(7, 7, 128)) |>   layer_conv_2d_transpose(filters = 128, kernel_size = c(4, 4),                           strides = c(2, 2), padding = \"same\") |>   layer_activation_leaky_relu(negative_slope = 0.2) |>   layer_conv_2d_transpose(filters = 128, kernel_size = c(4, 4),                           strides = c(2, 2), padding = \"same\") |>   layer_activation_leaky_relu(negative_slope = 0.2) |>   layer_conv_2d(filters = 1, kernel_size = c(7, 7), padding = \"same\",                 activation = \"sigmoid\")  summary(generator) ## Model: \"generator\" ## ┏━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━┓ ## ┃ Layer (type)                    ┃ Output Shape           ┃       Param # ┃ ## ┡━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━┩ ## │ dense_7 (Dense)                 │ (None, 6272)           │       809,088 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ leaky_re_lu_2 (LeakyReLU)       │ (None, 6272)           │             0 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ reshape (Reshape)               │ (None, 7, 7, 128)      │             0 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ conv2d_transpose                │ (None, 14, 14, 128)    │       262,272 │ ## │ (Conv2DTranspose)               │                        │               │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ leaky_re_lu_3 (LeakyReLU)       │ (None, 14, 14, 128)    │             0 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ conv2d_transpose_1              │ (None, 28, 28, 128)    │       262,272 │ ## │ (Conv2DTranspose)               │                        │               │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ leaky_re_lu_4 (LeakyReLU)       │ (None, 28, 28, 128)    │             0 │ ## ├─────────────────────────────────┼────────────────────────┼───────────────┤ ## │ conv2d_2 (Conv2D)               │ (None, 28, 28, 1)      │         6,273 │ ## └─────────────────────────────────┴────────────────────────┴───────────────┘ ##  Total params: 1,339,905 (5.11 MB) ##  Trainable params: 1,339,905 (5.11 MB) ##  Non-trainable params: 0 (0.00 B) # Instantiate one optimizer for the discriminator and another for the generator. d_optimizer <- optimizer_adam(learning_rate = 0.0003) g_optimizer <- optimizer_adam(learning_rate = 0.0004)  # Instantiate a loss function. loss_fn <- loss_binary_crossentropy(from_logits = TRUE)  train_step <- tf_function(function(real_images) {   # Sample random points in the latent space   c(batch_size, ...) %<-% op_shape(real_images)   random_latent_vectors <-     tf$random$normal(shape(batch_size, latent_dim))    # Decode them to fake images   generated_images <- generator(random_latent_vectors)    # Combine them with real images   combined_images <- tf$concat(list(generated_images, real_images),                                axis = 0L)    # Assemble labels discriminating real from fake images   labels <- tf$concat(list(tf$ones(shape(batch_size, 1)),                            tf$zeros(shape(batch_size, 1))),                       axis = 0L)    # Add random noise to the labels - important trick!   labels %<>% `+`(tf$random$uniform(tf$shape(.), maxval = 0.05))    # Train the discriminator   with(tf$GradientTape() %as% tape, {     predictions <- discriminator(combined_images)     d_loss <- loss_fn(labels, predictions)   })   grads <- tape$gradient(d_loss, discriminator$trainable_weights)   d_optimizer$apply(grads, discriminator$trainable_weights)    # Sample random points in the latent space   random_latent_vectors <-     tf$random$normal(shape(batch_size, latent_dim))    # Assemble labels that say \"all real images\"   misleading_labels <- tf$zeros(shape(batch_size, 1))    # Train the generator (note that we should *not* update the weights   # of the discriminator)!   with(tf$GradientTape() %as% tape, {     predictions <- discriminator(generator(random_latent_vectors))     g_loss <- loss_fn(misleading_labels, predictions)   })    grads <- tape$gradient(g_loss, generator$trainable_weights)   g_optimizer$apply(grads, generator$trainable_weights)    list(d_loss, g_loss, generated_images) }) # Prepare the dataset. We use both the training & test MNIST digits. batch_size <- 64 c(c(x_train, .), c(x_test, .)) %<-% dataset_mnist() all_digits <- op_concatenate(list(x_train, x_test)) all_digits <- op_reshape(all_digits, c(-1, 28, 28, 1)) dataset <- all_digits |>   tensor_slices_dataset() |>   dataset_map(\\(x) op_cast(x, \"float32\") / 255) |>   dataset_shuffle(buffer_size = 1024) |>   dataset_batch(batch_size = batch_size)  epochs <- 1 # In practice you need at least 20 epochs to generate nice digits. save_dir <- \"./\"  for (epoch in seq_len(epochs)) {   cat(\"Start epoch: \", epoch, \"\\n\")   step <- 0   train_iterator <- as_iterator(dataset)   while (!is.null(real_images <- iter_next(train_iterator))) {     step <- step + 1     # Train the discriminator & generator on one batch of real images.     c(d_loss, g_loss, generated_images) %<-% train_step(real_images)      # Logging.     if (step %% 200 == 0) {       # Print metrics       cat(sprintf(\"discriminator loss at step %d: %.2f\\n\", step, d_loss))       cat(sprintf(\"adversarial loss at step %d: %.2f\\n\", step, g_loss))     }      # To limit execution time we stop after 10 steps.     # Remove the lines below to actually train the model!     if (step > 10)       break   } } ## Start epoch:  1"},{"path":"https://keras3.posit.co/dev/articles/writing_your_own_callbacks.html","id":"introduction","dir":"Articles","previous_headings":"","what":"Introduction","title":"Writing your own callbacks","text":"callback powerful tool customize behavior Keras model training, evaluation, inference. Examples include keras.callbacks.TensorBoard visualize training progress results TensorBoard, keras.callbacks.ModelCheckpoint periodically save model training. guide, learn Keras callback , can , can build . provide demos simple callback applications get started.","code":""},{"path":"https://keras3.posit.co/dev/articles/writing_your_own_callbacks.html","id":"setup","dir":"Articles","previous_headings":"","what":"Setup","title":"Writing your own callbacks","text":"","code":"library(keras3)"},{"path":"https://keras3.posit.co/dev/articles/writing_your_own_callbacks.html","id":"keras-callbacks-overview","dir":"Articles","previous_headings":"","what":"Keras callbacks overview","title":"Writing your own callbacks","text":"callbacks subclass keras.callbacks.Callback class, override set methods called various stages training, testing, predicting. Callbacks useful get view internal states statistics model training. can pass list callbacks (keyword argument callbacks) following model methods: fit() evaluate() predict()","code":""},{"path":[]},{"path":[]},{"path":"https://keras3.posit.co/dev/articles/writing_your_own_callbacks.html","id":"on_traintestpredict_beginlogs-null","dir":"Articles","previous_headings":"An overview of callback methods > Global methods","what":"on_(train|test|predict)_begin(logs = NULL)","title":"Writing your own callbacks","text":"Called beginning fit/evaluate/predict.","code":""},{"path":"https://keras3.posit.co/dev/articles/writing_your_own_callbacks.html","id":"on_traintestpredict_endlogs-null","dir":"Articles","previous_headings":"An overview of callback methods > Global methods","what":"on_(train|test|predict)_end(logs = NULL)","title":"Writing your own callbacks","text":"Called end fit/evaluate/predict.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/articles/writing_your_own_callbacks.html","id":"on_traintestpredict_batch_beginbatch-logs-null","dir":"Articles","previous_headings":"An overview of callback methods > Batch-level methods for training/testing/predicting","what":"on_(train|test|predict)_batch_begin(batch, logs = NULL)","title":"Writing your own callbacks","text":"Called right processing batch training/testing/predicting.","code":""},{"path":"https://keras3.posit.co/dev/articles/writing_your_own_callbacks.html","id":"on_traintestpredict_batch_endbatch-logs-null","dir":"Articles","previous_headings":"An overview of callback methods > Batch-level methods for training/testing/predicting","what":"on_(train|test|predict)_batch_end(batch, logs = NULL)","title":"Writing your own callbacks","text":"Called end training/testing/predicting batch. Within method, logs named list containing metrics results.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/articles/writing_your_own_callbacks.html","id":"on_epoch_beginepoch-logs-null","dir":"Articles","previous_headings":"An overview of callback methods > Epoch-level methods (training only)","what":"on_epoch_begin(epoch, logs = NULL)","title":"Writing your own callbacks","text":"Called beginning epoch training.","code":""},{"path":"https://keras3.posit.co/dev/articles/writing_your_own_callbacks.html","id":"on_epoch_endepoch-logs-null","dir":"Articles","previous_headings":"An overview of callback methods > Epoch-level methods (training only)","what":"on_epoch_end(epoch, logs = NULL)","title":"Writing your own callbacks","text":"Called end epoch training.","code":""},{"path":"https://keras3.posit.co/dev/articles/writing_your_own_callbacks.html","id":"a-basic-example","dir":"Articles","previous_headings":"","what":"A basic example","title":"Writing your own callbacks","text":"Let’s take look concrete example. get started, let’s import tensorflow define simple Sequential Keras model: , load MNIST data training testing Keras datasets API: Now, define simple custom callback logs: fit/evaluate/predict starts & ends epoch starts & ends training batch starts & ends evaluation (test) batch starts & ends inference (prediction) batch starts & ends Let’s try :","code":"# Define the Keras model to add callbacks to get_model <- function() {   model <- keras_model_sequential()   model |> layer_dense(units = 1)   model |> compile(     optimizer = optimizer_rmsprop(learning_rate = 0.1),     loss = \"mean_squared_error\",     metrics = \"mean_absolute_error\"   )   model } # Load example MNIST data and pre-process it mnist <- dataset_mnist()  flatten_and_rescale <- function(x) {   x <- array_reshape(x, c(-1, 784))   x <- x / 255   x }  mnist$train$x <- flatten_and_rescale(mnist$train$x) mnist$test$x  <- flatten_and_rescale(mnist$test$x)  # limit to 1000 samples n <- 1000 mnist$train$x <- mnist$train$x[1:n,] mnist$train$y <- mnist$train$y[1:n] mnist$test$x  <- mnist$test$x[1:n,] mnist$test$y  <- mnist$test$y[1:n] show <- function(msg, logs) {   cat(glue::glue(msg, .envir = parent.frame()),       \"got logs: \", sep = \"; \")   str(logs); cat(\"\\n\") }  callback_custom <- Callback(   \"CustomCallback\",   on_train_begin         = \\(logs = NULL)        show(\"Starting training\", logs),   on_epoch_begin         = \\(epoch, logs = NULL) show(\"Start epoch {epoch} of training\", logs),   on_train_batch_begin   = \\(batch, logs = NULL) show(\"...Training: start of batch {batch}\", logs),   on_train_batch_end     = \\(batch, logs = NULL) show(\"...Training: end of batch {batch}\",  logs),   on_epoch_end           = \\(epoch, logs = NULL) show(\"End epoch {epoch} of training\", logs),   on_train_end           = \\(logs = NULL)        show(\"Stop training\", logs),     on_test_begin          = \\(logs = NULL)        show(\"Start testing\", logs),   on_test_batch_begin    = \\(batch, logs = NULL) show(\"...Evaluating: start of batch {batch}\", logs),   on_test_batch_end      = \\(batch, logs = NULL) show(\"...Evaluating: end of batch {batch}\", logs),   on_test_end            = \\(logs = NULL)        show(\"Stop testing\", logs),    on_predict_begin       = \\(logs = NULL)        show(\"Start predicting\", logs),   on_predict_end         = \\(logs = NULL)        show(\"Stop predicting\", logs),   on_predict_batch_begin = \\(batch, logs = NULL) show(\"...Predicting: start of batch {batch}\", logs),   on_predict_batch_end   = \\(batch, logs = NULL) show(\"...Predicting: end of batch {batch}\", logs), ) model <- get_model() model |> fit(   mnist$train$x, mnist$train$y,   batch_size = 128,   epochs = 2,   verbose = 0,   validation_split = 0.5,   callbacks = list(callback_custom()) ) ## Starting training; got logs:  Named list() ## ## Start epoch 1 of training; got logs:  Named list() ## ## ...Training: start of batch 1; got logs:  Named list() ## ## ...Training: end of batch 1; got logs: List of 2 ##  $ loss               : num 25.9 ##  $ mean_absolute_error: num 4.19 ## ## ...Training: start of batch 2; got logs:  Named list() ## ## ...Training: end of batch 2; got logs: List of 2 ##  $ loss               : num 433 ##  $ mean_absolute_error: num 15.5 ## ## ...Training: start of batch 3; got logs:  Named list() ## ## ...Training: end of batch 3; got logs: List of 2 ##  $ loss               : num 297 ##  $ mean_absolute_error: num 11.8 ## ## ...Training: start of batch 4; got logs:  Named list() ## ## ...Training: end of batch 4; got logs: List of 2 ##  $ loss               : num 231 ##  $ mean_absolute_error: num 9.68 ## ## Start testing; got logs:  Named list() ## ## ...Evaluating: start of batch 1; got logs:  Named list() ## ## ...Evaluating: end of batch 1; got logs: List of 2 ##  $ loss               : num 8.1 ##  $ mean_absolute_error: num 2.3 ## ## ...Evaluating: start of batch 2; got logs:  Named list() ## ## ...Evaluating: end of batch 2; got logs: List of 2 ##  $ loss               : num 7.58 ##  $ mean_absolute_error: num 2.23 ## ## ...Evaluating: start of batch 3; got logs:  Named list() ## ## ...Evaluating: end of batch 3; got logs: List of 2 ##  $ loss               : num 7.38 ##  $ mean_absolute_error: num 2.21 ## ## ...Evaluating: start of batch 4; got logs:  Named list() ## ## ...Evaluating: end of batch 4; got logs: List of 2 ##  $ loss               : num 7.3 ##  $ mean_absolute_error: num 2.21 ## ## Stop testing; got logs: List of 2 ##  $ loss               : num 7.3 ##  $ mean_absolute_error: num 2.21 ## ## End epoch 1 of training; got logs: List of 4 ##  $ loss                   : num 231 ##  $ mean_absolute_error    : num 9.68 ##  $ val_loss               : num 7.3 ##  $ val_mean_absolute_error: num 2.21 ## ## Start epoch 2 of training; got logs:  Named list() ## ## ...Training: start of batch 1; got logs:  Named list() ## ## ...Training: end of batch 1; got logs: List of 2 ##  $ loss               : num 7.44 ##  $ mean_absolute_error: num 2.27 ## ## ...Training: start of batch 2; got logs:  Named list() ## ## ...Training: end of batch 2; got logs: List of 2 ##  $ loss               : num 6.81 ##  $ mean_absolute_error: num 2.16 ## ## ...Training: start of batch 3; got logs:  Named list() ## ## ...Training: end of batch 3; got logs: List of 2 ##  $ loss               : num 6.12 ##  $ mean_absolute_error: num 2.06 ## ## ...Training: start of batch 4; got logs:  Named list() ## ## ...Training: end of batch 4; got logs: List of 2 ##  $ loss               : num 6.08 ##  $ mean_absolute_error: num 2.04 ## ## Start testing; got logs:  Named list() ## ## ...Evaluating: start of batch 1; got logs:  Named list() ## ## ...Evaluating: end of batch 1; got logs: List of 2 ##  $ loss               : num 5.54 ##  $ mean_absolute_error: num 1.92 ## ## ...Evaluating: start of batch 2; got logs:  Named list() ## ## ...Evaluating: end of batch 2; got logs: List of 2 ##  $ loss               : num 5.31 ##  $ mean_absolute_error: num 1.87 ## ## ...Evaluating: start of batch 3; got logs:  Named list() ## ## ...Evaluating: end of batch 3; got logs: List of 2 ##  $ loss               : num 5.11 ##  $ mean_absolute_error: num 1.8 ## ## ...Evaluating: start of batch 4; got logs:  Named list() ## ## ...Evaluating: end of batch 4; got logs: List of 2 ##  $ loss               : num 5.15 ##  $ mean_absolute_error: num 1.82 ## ## Stop testing; got logs: List of 2 ##  $ loss               : num 5.15 ##  $ mean_absolute_error: num 1.82 ## ## End epoch 2 of training; got logs: List of 4 ##  $ loss                   : num 6.08 ##  $ mean_absolute_error    : num 2.04 ##  $ val_loss               : num 5.15 ##  $ val_mean_absolute_error: num 1.82 ## ## Stop training; got logs: List of 4 ##  $ loss                   : num 6.08 ##  $ mean_absolute_error    : num 2.04 ##  $ val_loss               : num 5.15 ##  $ val_mean_absolute_error: num 1.82 res <- model |> evaluate(   mnist$test$x, mnist$test$y,   batch_size = 128, verbose = 0,   callbacks = list(callback_custom()) ) ## Start testing; got logs:  Named list() ## ## ...Evaluating: start of batch 1; got logs:  Named list() ## ## ...Evaluating: end of batch 1; got logs: List of 2 ##  $ loss               : num 5.2 ##  $ mean_absolute_error: num 1.84 ## ## ...Evaluating: start of batch 2; got logs:  Named list() ## ## ...Evaluating: end of batch 2; got logs: List of 2 ##  $ loss               : num 4.62 ##  $ mean_absolute_error: num 1.73 ## ## ...Evaluating: start of batch 3; got logs:  Named list() ## ## ...Evaluating: end of batch 3; got logs: List of 2 ##  $ loss               : num 4.61 ##  $ mean_absolute_error: num 1.74 ## ## ...Evaluating: start of batch 4; got logs:  Named list() ## ## ...Evaluating: end of batch 4; got logs: List of 2 ##  $ loss               : num 4.65 ##  $ mean_absolute_error: num 1.75 ## ## ...Evaluating: start of batch 5; got logs:  Named list() ## ## ...Evaluating: end of batch 5; got logs: List of 2 ##  $ loss               : num 4.84 ##  $ mean_absolute_error: num 1.77 ## ## ...Evaluating: start of batch 6; got logs:  Named list() ## ## ...Evaluating: end of batch 6; got logs: List of 2 ##  $ loss               : num 4.76 ##  $ mean_absolute_error: num 1.76 ## ## ...Evaluating: start of batch 7; got logs:  Named list() ## ## ...Evaluating: end of batch 7; got logs: List of 2 ##  $ loss               : num 4.74 ##  $ mean_absolute_error: num 1.76 ## ## ...Evaluating: start of batch 8; got logs:  Named list() ## ## ...Evaluating: end of batch 8; got logs: List of 2 ##  $ loss               : num 4.67 ##  $ mean_absolute_error: num 1.75 ## ## Stop testing; got logs: List of 2 ##  $ loss               : num 4.67 ##  $ mean_absolute_error: num 1.75 res <- model |> predict(   mnist$test$x,   batch_size = 128, verbose = 0,   callbacks = list(callback_custom()) ) ## Start predicting; got logs:  Named list() ## ## ...Predicting: start of batch 1; got logs:  Named list() ## ## ...Predicting: end of batch 1; got logs: List of 1 ##  $ outputs:<tf.Tensor: shape=(128, 1), dtype=float32, numpy=…> ## ## ...Predicting: start of batch 2; got logs:  Named list() ## ## ...Predicting: end of batch 2; got logs: List of 1 ##  $ outputs:<tf.Tensor: shape=(128, 1), dtype=float32, numpy=…> ## ## ...Predicting: start of batch 3; got logs:  Named list() ## ## ...Predicting: end of batch 3; got logs: List of 1 ##  $ outputs:<tf.Tensor: shape=(128, 1), dtype=float32, numpy=…> ## ## ...Predicting: start of batch 4; got logs:  Named list() ## ## ...Predicting: end of batch 4; got logs: List of 1 ##  $ outputs:<tf.Tensor: shape=(128, 1), dtype=float32, numpy=…> ## ## ...Predicting: start of batch 5; got logs:  Named list() ## ## ...Predicting: end of batch 5; got logs: List of 1 ##  $ outputs:<tf.Tensor: shape=(128, 1), dtype=float32, numpy=…> ## ## ...Predicting: start of batch 6; got logs:  Named list() ## ## ...Predicting: end of batch 6; got logs: List of 1 ##  $ outputs:<tf.Tensor: shape=(128, 1), dtype=float32, numpy=…> ## ## ...Predicting: start of batch 7; got logs:  Named list() ## ## ...Predicting: end of batch 7; got logs: List of 1 ##  $ outputs:<tf.Tensor: shape=(128, 1), dtype=float32, numpy=…> ## ## ...Predicting: start of batch 8; got logs:  Named list() ## ## ...Predicting: end of batch 8; got logs: List of 1 ##  $ outputs:<tf.Tensor: shape=(104, 1), dtype=float32, numpy=…> ## ## Stop predicting; got logs:  Named list()"},{"path":"https://keras3.posit.co/dev/articles/writing_your_own_callbacks.html","id":"usage-of-logs-list","dir":"Articles","previous_headings":"A basic example","what":"Usage of logs list","title":"Writing your own callbacks","text":"logs named list contains loss value, metrics end batch epoch. Example includes loss mean absolute error. information callbacks, can check Keras callback API documentation.","code":"callback_print_loss_and_mae <- Callback(   \"LossAndErrorPrintingCallback\",    on_train_batch_end = function(batch, logs = NULL)     cat(sprintf(\"Up to batch %i, the average loss is %7.2f.\\n\",                 batch,  logs$loss)),    on_test_batch_end = function(batch, logs = NULL)     cat(sprintf(\"Up to batch %i, the average loss is %7.2f.\\n\",                 batch, logs$loss)),    on_epoch_end = function(epoch, logs = NULL)     cat(sprintf(       \"The average loss for epoch %2i is %9.2f and mean absolute error is %7.2f.\\n\",       epoch, logs$loss, logs$mean_absolute_error     )) )   model <- get_model() model |> fit(   mnist$train$x, mnist$train$y,   epochs = 2, verbose = 0, batch_size = 128,   callbacks = list(callback_print_loss_and_mae()) ) ## Up to batch 1, the average loss is   25.12. ## Up to batch 2, the average loss is  398.92. ## Up to batch 3, the average loss is  274.04. ## Up to batch 4, the average loss is  208.32. ## Up to batch 5, the average loss is  168.15. ## Up to batch 6, the average loss is  141.31. ## Up to batch 7, the average loss is  122.19. ## Up to batch 8, the average loss is  110.05. ## The average loss for epoch  1 is    110.05 and mean absolute error is    5.79. ## Up to batch 1, the average loss is    4.71. ## Up to batch 2, the average loss is    4.74. ## Up to batch 3, the average loss is    4.81. ## Up to batch 4, the average loss is    5.07. ## Up to batch 5, the average loss is    5.08. ## Up to batch 6, the average loss is    5.09. ## Up to batch 7, the average loss is    5.19. ## Up to batch 8, the average loss is    5.51. ## The average loss for epoch  2 is      5.51 and mean absolute error is    1.90. res = model |> evaluate(   mnist$test$x, mnist$test$y,   verbose = 0, batch_size = 128,   callbacks = list(callback_print_loss_and_mae()) ) ## Up to batch 1, the average loss is   15.86. ## Up to batch 2, the average loss is   16.13. ## Up to batch 3, the average loss is   16.02. ## Up to batch 4, the average loss is   16.11. ## Up to batch 5, the average loss is   16.23. ## Up to batch 6, the average loss is   16.68. ## Up to batch 7, the average loss is   16.61. ## Up to batch 8, the average loss is   16.54."},{"path":"https://keras3.posit.co/dev/articles/writing_your_own_callbacks.html","id":"usage-of-selfmodel-attribute","dir":"Articles","previous_headings":"","what":"Usage of self$model attribute","title":"Writing your own callbacks","text":"addition receiving log information one methods called, callbacks access model associated current round training/evaluation/inference: self$model. things can self$model callback: Set self$model$stop_training <- TRUE immediately interrupt training. Mutate hyperparameters optimizer (available self$model$optimizer), self$model$optimizer$learning_rate. Save model period intervals. Record output model |> predict() test samples end epoch, use sanity check training. Extract visualizations intermediate features end epoch, monitor model learning time. etc. Let’s see action couple examples.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/articles/writing_your_own_callbacks.html","id":"early-stopping-at-minimum-loss","dir":"Articles","previous_headings":"Examples of Keras callback applications","what":"Early stopping at minimum loss","title":"Writing your own callbacks","text":"first example shows creation Callback stops training minimum loss reached, setting attribute self$model$stop_training (boolean). Optionally, can provide argument patience specify many epochs wait stopping reached local minimum. callback_early_stopping() provides complete general implementation.","code":"callback_early_stopping_at_min_loss <- Callback(   \"EarlyStoppingAtMinLoss\",   `__doc__` =     \"Stop training when the loss is at its min, i.e. the loss stops decreasing.      Arguments:         patience: Number of epochs to wait after min has been hit. After this         number of no improvement, training stops.     \",    initialize = function(patience = 0) {     super$initialize()     self$patience <- patience     # best_weights to store the weights at which the minimum loss occurs.     self$best_weights <- NULL   },    on_train_begin = function(logs = NULL) {     # The number of epoch it has waited when loss is no longer minimum.     self$wait <- 0     # The epoch the training stops at.     self$stopped_epoch <- 0     # Initialize the best as infinity.     self$best <- Inf   },    on_epoch_end = function(epoch, logs = NULL) {     current <- logs$loss     if (current < self$best) {       self$best <- current       self$wait <- 0L       # Record the best weights if current results is better (less).       self$best_weights <- get_weights(self$model)     } else {       add(self$wait) <- 1L       if (self$wait >= self$patience) {         self$stopped_epoch <- epoch         self$model$stop_training <- TRUE         cat(\"Restoring model weights from the end of the best epoch.\\n\")         model$set_weights(self$best_weights)       }     }   },    on_train_end = function(logs = NULL)     if (self$stopped_epoch > 0)       cat(sprintf(\"Epoch %05d: early stopping\\n\", self$stopped_epoch + 1)) ) `add<-` <- `+`   model <- get_model() model |> fit(   mnist$train$x,   mnist$train$y,   epochs = 30,   batch_size = 64,   verbose = 0,   callbacks = list(callback_print_loss_and_mae(),                    callback_early_stopping_at_min_loss()) ) ## Up to batch 1, the average loss is   30.54. ## Up to batch 2, the average loss is  513.27. ## Up to batch 3, the average loss is  352.60. ## Up to batch 4, the average loss is  266.37. ## Up to batch 5, the average loss is  214.68. ## Up to batch 6, the average loss is  179.97. ## Up to batch 7, the average loss is  155.06. ## Up to batch 8, the average loss is  136.59. ## Up to batch 9, the average loss is  121.96. ## Up to batch 10, the average loss is  110.28. ## Up to batch 11, the average loss is  100.72. ## Up to batch 12, the average loss is   92.71. ## Up to batch 13, the average loss is   85.95. ## Up to batch 14, the average loss is   80.21. ## Up to batch 15, the average loss is   75.17. ## Up to batch 16, the average loss is   72.48. ## The average loss for epoch  1 is     72.48 and mean absolute error is    4.08. ## Up to batch 1, the average loss is    7.98. ## Up to batch 2, the average loss is    9.92. ## Up to batch 3, the average loss is   12.88. ## Up to batch 4, the average loss is   16.61. ## Up to batch 5, the average loss is   20.49. ## Up to batch 6, the average loss is   26.14. ## Up to batch 7, the average loss is   30.44. ## Up to batch 8, the average loss is   33.76. ## Up to batch 9, the average loss is   36.32. ## Up to batch 10, the average loss is   35.26. ## Up to batch 11, the average loss is   34.22. ## Up to batch 12, the average loss is   33.53. ## Up to batch 13, the average loss is   32.84. ## Up to batch 14, the average loss is   31.80. ## Up to batch 15, the average loss is   31.39. ## Up to batch 16, the average loss is   31.45. ## The average loss for epoch  2 is     31.45 and mean absolute error is    4.82. ## Up to batch 1, the average loss is   39.60. ## Up to batch 2, the average loss is   41.95. ## Up to batch 3, the average loss is   41.29. ## Up to batch 4, the average loss is   36.77. ## Up to batch 5, the average loss is   32.08. ## Up to batch 6, the average loss is   28.17. ## Up to batch 7, the average loss is   25.33. ## Up to batch 8, the average loss is   23.56. ## Up to batch 9, the average loss is   22.28. ## Up to batch 10, the average loss is   21.22. ## Up to batch 11, the average loss is   20.87. ## Up to batch 12, the average loss is   22.25. ## Up to batch 13, the average loss is   25.08. ## Up to batch 14, the average loss is   27.87. ## Up to batch 15, the average loss is   31.72. ## Up to batch 16, the average loss is   33.21. ## The average loss for epoch  3 is     33.21 and mean absolute error is    4.79. ## Restoring model weights from the end of the best epoch. ## Epoch 00004: early stopping"},{"path":"https://keras3.posit.co/dev/articles/writing_your_own_callbacks.html","id":"learning-rate-scheduling","dir":"Articles","previous_headings":"Examples of Keras callback applications","what":"Learning rate scheduling","title":"Writing your own callbacks","text":"example, show custom Callback can used dynamically change learning rate optimizer course training. See keras$callbacks$LearningRateScheduler general implementations (RStudio, press F1 cursor LearningRateScheduler browser open page).","code":"callback_custom_learning_rate_scheduler <- Callback(   \"CustomLearningRateScheduler\",   `__doc__` =   \"Learning rate scheduler which sets the learning rate according to schedule.      Arguments:         schedule: a function that takes an epoch index             (integer, indexed from 0) and current learning rate             as inputs and returns a new learning rate as output (float).     \",    initialize = function(schedule) {     super$initialize()     self$schedule <- schedule   },    on_epoch_begin = function(epoch, logs = NULL) {     ## When in doubt about what types of objects are in scope (e.g., self$model)     ## use a debugger to interact with the actual objects at the console!     # browser()      if (!\"learning_rate\" %in% names(self$model$optimizer))       stop('Optimizer must have a \"learning_rate\" attribute.')      # # Get the current learning rate from model's optimizer.     # use as.numeric() to convert the keras variablea to an R numeric     lr <- as.numeric(self$model$optimizer$learning_rate)     # # Call schedule function to get the scheduled learning rate.     scheduled_lr <- self$schedule(epoch, lr)     # # Set the value back to the optimizer before this epoch starts     optimizer <- self$model$optimizer     optimizer$learning_rate <- scheduled_lr     cat(sprintf(\"\\nEpoch %03d: Learning rate is %6.4f.\\n\", epoch, scheduled_lr))   } )  LR_SCHEDULE <- tibble::tribble(   ~start_epoch, ~learning_rate,              0,            0.1,              3,           0.05,              6,           0.01,              9,          0.005,             12,          0.001,   )  last <- function(x) x[length(x)] lr_schedule <- function(epoch, learning_rate) {   \"Helper function to retrieve the scheduled learning rate based on epoch.\"   with(LR_SCHEDULE, learning_rate[last(which(epoch >= start_epoch))]) }  model <- get_model() model |> fit(   mnist$train$x,   mnist$train$y,   epochs = 14,   batch_size = 64,   verbose = 0,   callbacks = list(     callback_print_loss_and_mae(),     callback_custom_learning_rate_scheduler(lr_schedule)   ) ) ## ## Epoch 001: Learning rate is 0.1000. ## Up to batch 1, the average loss is   29.36. ## Up to batch 2, the average loss is  513.95. ## Up to batch 3, the average loss is  352.70. ## Up to batch 4, the average loss is  266.46. ## Up to batch 5, the average loss is  214.73. ## Up to batch 6, the average loss is  180.00. ## Up to batch 7, the average loss is  155.05. ## Up to batch 8, the average loss is  136.64. ## Up to batch 9, the average loss is  121.97. ## Up to batch 10, the average loss is  110.30. ## Up to batch 11, the average loss is  100.76. ## Up to batch 12, the average loss is   92.74. ## Up to batch 13, the average loss is   85.95. ## Up to batch 14, the average loss is   80.18. ## Up to batch 15, the average loss is   75.11. ## Up to batch 16, the average loss is   72.38. ## The average loss for epoch  1 is     72.38 and mean absolute error is    4.04. ## ## Epoch 002: Learning rate is 0.1000. ## Up to batch 1, the average loss is    6.95. ## Up to batch 2, the average loss is    8.71. ## Up to batch 3, the average loss is   11.42. ## Up to batch 4, the average loss is   15.15. ## Up to batch 5, the average loss is   19.28. ## Up to batch 6, the average loss is   25.54. ## Up to batch 7, the average loss is   30.38. ## Up to batch 8, the average loss is   33.95. ## Up to batch 9, the average loss is   36.58. ## Up to batch 10, the average loss is   35.46. ## Up to batch 11, the average loss is   34.34. ## Up to batch 12, the average loss is   33.51. ## Up to batch 13, the average loss is   32.67. ## Up to batch 14, the average loss is   31.54. ## Up to batch 15, the average loss is   31.05. ## Up to batch 16, the average loss is   31.09. ## The average loss for epoch  2 is     31.09 and mean absolute error is    4.77. ## ## Epoch 003: Learning rate is 0.0500. ## Up to batch 1, the average loss is   40.40. ## Up to batch 2, the average loss is   22.33. ## Up to batch 3, the average loss is   16.18. ## Up to batch 4, the average loss is   13.09. ## Up to batch 5, the average loss is   11.48. ## Up to batch 6, the average loss is   10.21. ## Up to batch 7, the average loss is    9.22. ## Up to batch 8, the average loss is    8.70. ## Up to batch 9, the average loss is    8.16. ## Up to batch 10, the average loss is    7.80. ## Up to batch 11, the average loss is    7.50. ## Up to batch 12, the average loss is    7.17. ## Up to batch 13, the average loss is    6.89. ## Up to batch 14, the average loss is    6.70. ## Up to batch 15, the average loss is    6.52. ## Up to batch 16, the average loss is    6.54. ## The average loss for epoch  3 is      6.54 and mean absolute error is    1.93. ## ## Epoch 004: Learning rate is 0.0500. ## Up to batch 1, the average loss is    8.74. ## Up to batch 2, the average loss is    8.34. ## Up to batch 3, the average loss is    9.09. ## Up to batch 4, the average loss is    9.72. ## Up to batch 5, the average loss is   10.48. ## Up to batch 6, the average loss is   11.69. ## Up to batch 7, the average loss is   11.83. ## Up to batch 8, the average loss is   11.56. ## Up to batch 9, the average loss is   11.24. ## Up to batch 10, the average loss is   10.84. ## Up to batch 11, the average loss is   10.66. ## Up to batch 12, the average loss is   10.44. ## Up to batch 13, the average loss is   10.21. ## Up to batch 14, the average loss is   10.06. ## Up to batch 15, the average loss is   10.00. ## Up to batch 16, the average loss is   10.20. ## The average loss for epoch  4 is     10.20 and mean absolute error is    2.71. ## ## Epoch 005: Learning rate is 0.0500. ## Up to batch 1, the average loss is   17.26. ## Up to batch 2, the average loss is   14.09. ## Up to batch 3, the average loss is   12.67. ## Up to batch 4, the average loss is   11.44. ## Up to batch 5, the average loss is   10.54. ## Up to batch 6, the average loss is   10.10. ## Up to batch 7, the average loss is    9.53. ## Up to batch 8, the average loss is    9.17. ## Up to batch 9, the average loss is    8.78. ## Up to batch 10, the average loss is    8.49. ## Up to batch 11, the average loss is    8.50. ## Up to batch 12, the average loss is    8.59. ## Up to batch 13, the average loss is    8.68. ## Up to batch 14, the average loss is    8.86. ## Up to batch 15, the average loss is    9.17. ## Up to batch 16, the average loss is    9.53. ## The average loss for epoch  5 is      9.53 and mean absolute error is    2.58. ## ## Epoch 006: Learning rate is 0.0100. ## Up to batch 1, the average loss is   17.04. ## Up to batch 2, the average loss is   14.85. ## Up to batch 3, the average loss is   11.53. ## Up to batch 4, the average loss is    9.65. ## Up to batch 5, the average loss is    8.44. ## Up to batch 6, the average loss is    7.50. ## Up to batch 7, the average loss is    6.74. ## Up to batch 8, the average loss is    6.56. ## Up to batch 9, the average loss is    6.18. ## Up to batch 10, the average loss is    5.87. ## Up to batch 11, the average loss is    5.63. ## Up to batch 12, the average loss is    5.45. ## Up to batch 13, the average loss is    5.23. ## Up to batch 14, the average loss is    5.12. ## Up to batch 15, the average loss is    4.96. ## Up to batch 16, the average loss is    4.91. ## The average loss for epoch  6 is      4.91 and mean absolute error is    1.67. ## ## Epoch 007: Learning rate is 0.0100. ## Up to batch 1, the average loss is    3.65. ## Up to batch 2, the average loss is    3.04. ## Up to batch 3, the average loss is    2.88. ## Up to batch 4, the average loss is    2.85. ## Up to batch 5, the average loss is    2.88. ## Up to batch 6, the average loss is    2.81. ## Up to batch 7, the average loss is    2.70. ## Up to batch 8, the average loss is    2.96. ## Up to batch 9, the average loss is    2.96. ## Up to batch 10, the average loss is    2.93. ## Up to batch 11, the average loss is    2.95. ## Up to batch 12, the average loss is    2.98. ## Up to batch 13, the average loss is    2.97. ## Up to batch 14, the average loss is    3.01. ## Up to batch 15, the average loss is    3.00. ## Up to batch 16, the average loss is    3.05. ## The average loss for epoch  7 is      3.05 and mean absolute error is    1.34. ## ## Epoch 008: Learning rate is 0.0100. ## Up to batch 1, the average loss is    3.69. ## Up to batch 2, the average loss is    3.21. ## Up to batch 3, the average loss is    3.00. ## Up to batch 4, the average loss is    2.91. ## Up to batch 5, the average loss is    2.94. ## Up to batch 6, the average loss is    2.85. ## Up to batch 7, the average loss is    2.72. ## Up to batch 8, the average loss is    2.95. ## Up to batch 9, the average loss is    2.97. ## Up to batch 10, the average loss is    2.93. ## Up to batch 11, the average loss is    2.96. ## Up to batch 12, the average loss is    2.98. ## Up to batch 13, the average loss is    2.99. ## Up to batch 14, the average loss is    3.05. ## Up to batch 15, the average loss is    3.08. ## Up to batch 16, the average loss is    3.14. ## The average loss for epoch  8 is      3.14 and mean absolute error is    1.36. ## ## Epoch 009: Learning rate is 0.0050. ## Up to batch 1, the average loss is    3.71. ## Up to batch 2, the average loss is    2.93. ## Up to batch 3, the average loss is    2.76. ## Up to batch 4, the average loss is    2.70. ## Up to batch 5, the average loss is    2.76. ## Up to batch 6, the average loss is    2.69. ## Up to batch 7, the average loss is    2.57. ## Up to batch 8, the average loss is    2.79. ## Up to batch 9, the average loss is    2.80. ## Up to batch 10, the average loss is    2.77. ## Up to batch 11, the average loss is    2.79. ## Up to batch 12, the average loss is    2.80. ## Up to batch 13, the average loss is    2.78. ## Up to batch 14, the average loss is    2.81. ## Up to batch 15, the average loss is    2.80. ## Up to batch 16, the average loss is    2.83. ## The average loss for epoch  9 is      2.83 and mean absolute error is    1.28. ## ## Epoch 010: Learning rate is 0.0050. ## Up to batch 1, the average loss is    3.02. ## Up to batch 2, the average loss is    2.69. ## Up to batch 3, the average loss is    2.58. ## Up to batch 4, the average loss is    2.57. ## Up to batch 5, the average loss is    2.65. ## Up to batch 6, the average loss is    2.60. ## Up to batch 7, the average loss is    2.48. ## Up to batch 8, the average loss is    2.72. ## Up to batch 9, the average loss is    2.74. ## Up to batch 10, the average loss is    2.71. ## Up to batch 11, the average loss is    2.74. ## Up to batch 12, the average loss is    2.75. ## Up to batch 13, the average loss is    2.74. ## Up to batch 14, the average loss is    2.77. ## Up to batch 15, the average loss is    2.77. ## Up to batch 16, the average loss is    2.80. ## The average loss for epoch 10 is      2.80 and mean absolute error is    1.27. ## ## Epoch 011: Learning rate is 0.0050. ## Up to batch 1, the average loss is    3.01. ## Up to batch 2, the average loss is    2.69. ## Up to batch 3, the average loss is    2.58. ## Up to batch 4, the average loss is    2.56. ## Up to batch 5, the average loss is    2.63. ## Up to batch 6, the average loss is    2.58. ## Up to batch 7, the average loss is    2.47. ## Up to batch 8, the average loss is    2.70. ## Up to batch 9, the average loss is    2.72. ## Up to batch 10, the average loss is    2.69. ## Up to batch 11, the average loss is    2.71. ## Up to batch 12, the average loss is    2.72. ## Up to batch 13, the average loss is    2.71. ## Up to batch 14, the average loss is    2.75. ## Up to batch 15, the average loss is    2.74. ## Up to batch 16, the average loss is    2.77. ## The average loss for epoch 11 is      2.77 and mean absolute error is    1.27. ## ## Epoch 012: Learning rate is 0.0010. ## Up to batch 1, the average loss is    2.96. ## Up to batch 2, the average loss is    2.53. ## Up to batch 3, the average loss is    2.47. ## Up to batch 4, the average loss is    2.46. ## Up to batch 5, the average loss is    2.54. ## Up to batch 6, the average loss is    2.48. ## Up to batch 7, the average loss is    2.39. ## Up to batch 8, the average loss is    2.60. ## Up to batch 9, the average loss is    2.62. ## Up to batch 10, the average loss is    2.59. ## Up to batch 11, the average loss is    2.61. ## Up to batch 12, the average loss is    2.62. ## Up to batch 13, the average loss is    2.60. ## Up to batch 14, the average loss is    2.64. ## Up to batch 15, the average loss is    2.62. ## Up to batch 16, the average loss is    2.64. ## The average loss for epoch 12 is      2.64 and mean absolute error is    1.24. ## ## Epoch 013: Learning rate is 0.0010. ## Up to batch 1, the average loss is    2.82. ## Up to batch 2, the average loss is    2.46. ## Up to batch 3, the average loss is    2.42. ## Up to batch 4, the average loss is    2.42. ## Up to batch 5, the average loss is    2.50. ## Up to batch 6, the average loss is    2.45. ## Up to batch 7, the average loss is    2.36. ## Up to batch 8, the average loss is    2.57. ## Up to batch 9, the average loss is    2.59. ## Up to batch 10, the average loss is    2.57. ## Up to batch 11, the average loss is    2.59. ## Up to batch 12, the average loss is    2.60. ## Up to batch 13, the average loss is    2.59. ## Up to batch 14, the average loss is    2.62. ## Up to batch 15, the average loss is    2.61. ## Up to batch 16, the average loss is    2.63. ## The average loss for epoch 13 is      2.63 and mean absolute error is    1.23. ## ## Epoch 014: Learning rate is 0.0010. ## Up to batch 1, the average loss is    2.79. ## Up to batch 2, the average loss is    2.44. ## Up to batch 3, the average loss is    2.40. ## Up to batch 4, the average loss is    2.41. ## Up to batch 5, the average loss is    2.49. ## Up to batch 6, the average loss is    2.44. ## Up to batch 7, the average loss is    2.34. ## Up to batch 8, the average loss is    2.56. ## Up to batch 9, the average loss is    2.58. ## Up to batch 10, the average loss is    2.56. ## Up to batch 11, the average loss is    2.58. ## Up to batch 12, the average loss is    2.59. ## Up to batch 13, the average loss is    2.58. ## Up to batch 14, the average loss is    2.61. ## Up to batch 15, the average loss is    2.60. ## Up to batch 16, the average loss is    2.62. ## The average loss for epoch 14 is      2.62 and mean absolute error is    1.23."},{"path":"https://keras3.posit.co/dev/articles/writing_your_own_callbacks.html","id":"built-in-keras-callbacks","dir":"Articles","previous_headings":"Examples of Keras callback applications","what":"Built-in Keras callbacks","title":"Writing your own callbacks","text":"sure check existing Keras callbacks reading API docs. Applications include logging CSV, saving model, visualizing metrics TensorBoard, lot !","code":""},{"path":"https://keras3.posit.co/dev/authors.html","id":null,"dir":"","previous_headings":"","what":"Authors","title":"Authors and Citation","text":"Tomasz Kalinowski. Author, copyright holder, maintainer. Daniel Falbel. Contributor, copyright holder. JJ Allaire. Author, copyright holder. François Chollet. Author, copyright holder. Posit Software, PBC. Copyright holder, funder. Google. Copyright holder, funder. Yuan Tang. Contributor, copyright holder. Wouter Van Der Bijl. Contributor, copyright holder. Martin Studer. Contributor, copyright holder. Sigrid Keydana. Contributor.","code":""},{"path":"https://keras3.posit.co/dev/authors.html","id":"citation","dir":"","previous_headings":"","what":"Citation","title":"Authors and Citation","text":"Kalinowski T, Allaire J, Chollet F (2025). keras3: R Interface 'Keras'. R package version 1.4.0.9000, https://keras3.posit.co/.","code":"@Manual{,   title = {keras3: R Interface to 'Keras'},   author = {Tomasz Kalinowski and JJ Allaire and François Chollet},   year = {2025},   note = {R package version 1.4.0.9000},   url = {https://keras3.posit.co/}, }"},{"path":"https://keras3.posit.co/dev/index.html","id":"r-interface-to-keras","dir":"","previous_headings":"","what":"R Interface to Keras","title":"R Interface to Keras","text":"Keras high-level neural networks API developed focus enabling fast experimentation. able go idea result least possible delay key good research. Keras following key features: Allows code run CPU GPU, seamlessly. User-friendly API makes easy quickly prototype deep learning models. Built-support convolutional networks (computer vision), recurrent networks (sequence processing), combination . Supports arbitrary network architectures: multi-input multi-output models, layer sharing, model sharing, etc. means Keras appropriate building essentially deep learning model, memory network neural Turing machine. See package website https://keras3.posit.co complete documentation.","code":""},{"path":"https://keras3.posit.co/dev/reference/Callback.html","id":null,"dir":"Reference","previous_headings":"","what":"Define a custom Callback class — Callback","title":"Define a custom Callback class — Callback","text":"Callbacks can passed keras methods fit(), evaluate(), predict() order hook various stages model training, evaluation, inference lifecycle. create custom callback, call Callback() override method associated stage interest.","code":""},{"path":"https://keras3.posit.co/dev/reference/Callback.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Define a custom Callback class — Callback","text":"","code":"Callback(   classname,   on_epoch_begin = NULL,   on_epoch_end = NULL,   on_train_begin = NULL,   on_train_end = NULL,   on_train_batch_begin = NULL,   on_train_batch_end = NULL,   on_test_begin = NULL,   on_test_end = NULL,   on_test_batch_begin = NULL,   on_test_batch_end = NULL,   on_predict_begin = NULL,   on_predict_end = NULL,   on_predict_batch_begin = NULL,   on_predict_batch_end = NULL,   ...,   public = list(),   private = list(),   inherit = NULL,   parent_env = parent.frame() )"},{"path":"https://keras3.posit.co/dev/reference/Callback.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Define a custom Callback class — Callback","text":"classname String, name custom class. (Conventionally, CamelCase). on_epoch_begin   Called start epoch. Subclasses override actions run. function called TRAIN mode. Args: epoch: Integer, index epoch. logs: Named List. Currently data passed argument method may change future. on_epoch_end   Called end epoch. Subclasses override actions run. function called TRAIN mode. Args: epoch: Integer, index epoch. logs: Named List, metric results training epoch, validation epoch validation performed. Validation result keys prefixed val_. training epoch, values Model's metrics returned. Example: list(loss = 0.2, accuracy = 0.7). on_train_begin   Called beginning training. Subclasses override actions run. Args: logs: Named list. Currently data passed argument method may change future. on_train_end   Called end training. Subclasses override actions run. Args: logs: Named list. Currently output last call on_epoch_end() passed argument method may change future. on_train_batch_begin   Called beginning training batch fit() methods. Subclasses override actions run. Note steps_per_execution argument compile Model set N, method called every N batches. Args: batch: Integer, index batch within current epoch. logs: Named list. Currently data passed argument method may change future. on_train_batch_end   Called end training batch fit() methods. Subclasses override actions run. Note steps_per_execution argument compile Model set N, method called every N batches. Args: batch: Integer, index batch within current epoch. logs: Named list. Aggregated metric results batch. on_test_begin   Called beginning evaluation validation. Subclasses override actions run. Args: logs: Named list. Currently data passed argument method may change future. on_test_end   Called end evaluation validation. Subclasses override actions run. Args: logs: Named list. Currently output last call on_test_batch_end() passed argument method may change future. on_test_batch_begin   Called beginning batch evaluate() methods. Also called beginning validation batch fit() methods, validation data provided. Subclasses override actions run. Note steps_per_execution argument compile() Model set N, method called every N batches. Args: batch: Integer, index batch within current epoch. logs: Named list. Currently data passed argument method may change future. on_test_batch_end   Called end batch evaluate() methods. Also called end validation batch fit() methods, validation data provided. Subclasses override actions run. Note steps_per_execution argument compile() Model set N, method called every N batches. Args: batch: Integer, index batch within current epoch. logs: Named list. Aggregated metric results batch. on_predict_begin   Called beginning prediction. Subclasses override actions run. Args: logs: Named list. Currently data passed argument method may change future. on_predict_end   Called end prediction. Subclasses override actions run. Args: logs: Named list. Currently data passed argument method may change future. on_predict_batch_begin   Called beginning batch predict() methods. Subclasses override actions run. Note steps_per_execution argument compile() Model set N, method called every N batches. Args: batch: Integer, index batch within current epoch. logs: Named list. Currently data passed argument method may change future. on_predict_batch_end   Called end batch predict() methods. Subclasses override actions run. Note steps_per_execution argument compile Model set N, method called every N batches. Args: batch: Integer, index batch within current epoch. logs: Named list. Aggregated metric results batch. ..., public Additional methods public members custom class. private Named list R objects (typically, functions) include instance private environments. private methods symbols scope public methods (See section \"Symbols Scope\"). instance private environment. objects private invisible Keras framework Python runtime. inherit custom class subclass. default, base keras class. parent_env R environment class methods grandparent.","code":"\\(epoch, logs = NULL) \\(epoch, logs = NULL) \\(logs = NULL) \\(logs = NULL) \\(batch, logs = NULL) \\(batch, logs=NULL) \\(logs = NULL) \\(logs = NULL) \\(batch, logs = NULL) \\(batch, logs = NULL) \\(logs = NULL) \\(logs = NULL) \\(batch, logs = NULL) \\(batch, logs = NULL)"},{"path":"https://keras3.posit.co/dev/reference/Callback.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Define a custom Callback class — Callback","text":"function returns custom Callback instances, similar builtin callback functions.","code":""},{"path":"https://keras3.posit.co/dev/reference/Callback.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Define a custom Callback class — Callback","text":"R function custom methods (public private) following symbols scope: self: Layer instance. super: Layer superclass. private: R environment specific class instance. objects defined invisible Keras framework. __class__ current class type object. also available alias symbol, value supplied Layer(classname = )","code":"training_finished <- FALSE callback_mark_finished <- Callback(\"MarkFinished\",   on_train_end = function(logs = NULL) {     training_finished <<- TRUE   } )  model <- keras_model_sequential(input_shape = c(1)) |>   layer_dense(1) model |> compile(loss = 'mean_squared_error') model |> fit(op_ones(c(1, 1)), op_ones(c(1, 1)),              callbacks = callback_mark_finished()) stopifnot(isTRUE(training_finished))"},{"path":"https://keras3.posit.co/dev/reference/Callback.html","id":"attributes-accessible-via-self-","dir":"Reference","previous_headings":"","what":"Attributes (accessible via self$)","title":"Define a custom Callback class — Callback","text":"params: Named list, Training parameters (e.g. verbosity, batch size, number epochs, ...). model: Instance Model. Reference model trained. logs named list callback methods take argument contain keys quantities relevant current batch epoch (see method-specific docstrings).","code":""},{"path":"https://keras3.posit.co/dev/reference/Callback.html","id":"symbols-in-scope","dir":"Reference","previous_headings":"","what":"Symbols in scope","title":"Define a custom Callback class — Callback","text":"R function custom methods (public private) following symbols scope: self: custom class instance. super: custom class superclass. private: R environment specific class instance. objects assigned invisible Keras framework. __class__ .symbol(classname): custom class type object.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/Constraint.html","id":null,"dir":"Reference","previous_headings":"","what":"Define a custom Constraint class — Constraint","title":"Define a custom Constraint class — Constraint","text":"Base class weight constraints. Constraint() instance works like stateless function. Users subclass Constraint class override call() method, takes single weight parameter return projected version parameter (e.g. normalized clipped). Constraints can used various Keras layers via kernel_constraint bias_constraint arguments. simple example non-negative weight constraint:     Usage layer:","code":"constraint_nonnegative <- Constraint(\"NonNegative\",   call = function(w) {     w * op_cast(w >= 0, dtype = w$dtype)   } ) weight <- op_convert_to_tensor(c(-1, 1)) constraint_nonnegative()(weight) ## tf.Tensor([-0.  1.], shape=(2), dtype=float64) layer_dense(units = 4, kernel_constraint = constraint_nonnegative()) ## <Dense name=dense, built=False> ##  signature: (*args, **kwargs)"},{"path":"https://keras3.posit.co/dev/reference/Constraint.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Define a custom Constraint class — Constraint","text":"","code":"Constraint(   classname,   call = NULL,   get_config = NULL,   ...,   public = list(),   private = list(),   inherit = NULL,   parent_env = parent.frame() )"},{"path":"https://keras3.posit.co/dev/reference/Constraint.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Define a custom Constraint class — Constraint","text":"classname String, name custom class. (Conventionally, CamelCase). call   Applies constraint input weight variable. default, inputs weight variable modified. Users override method implement projection function. Args: w: Input weight variable. Returns: Projected variable (default, returns unmodified inputs). get_config   Function returns named list object config. constraint config named list (JSON-serializable) can used reinstantiate object (via .call(<constraint_class>, <config>)). ..., public Additional methods public members custom class. private Named list R objects (typically, functions) include instance private environments. private methods symbols scope public methods (See section \"Symbols Scope\"). instance private environment. objects private invisible Keras framework Python runtime. inherit custom class subclass. default, base keras class. parent_env R environment class methods grandparent.","code":"\\(w) \\()"},{"path":"https://keras3.posit.co/dev/reference/Constraint.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Define a custom Constraint class — Constraint","text":"function returns Constraint instances, similar builtin constraint functions like constraint_maxnorm().","code":""},{"path":"https://keras3.posit.co/dev/reference/Constraint.html","id":"symbols-in-scope","dir":"Reference","previous_headings":"","what":"Symbols in scope","title":"Define a custom Constraint class — Constraint","text":"R function custom methods (public private) following symbols scope: self: custom class instance. super: custom class superclass. private: R environment specific class instance. objects assigned invisible Keras framework. __class__ .symbol(classname): custom class type object.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/Layer.html","id":null,"dir":"Reference","previous_headings":"","what":"Define a custom Layer class. — Layer","title":"Define a custom Layer class. — Layer","text":"layer callable object takes input one tensors outputs one tensors. involves computation, defined call() method, state (weight variables). State can created: initialize(), instance via self$add_weight(); optional build() method, invoked first call() layer, supplies shape(s) input(s), may known initialization time. Layers recursively composable: assign Layer instance attribute another Layer, outer layer start tracking weights created inner layer. Nested layers instantiated initialize() method build() method. Users just instantiate layer treat callable.","code":""},{"path":"https://keras3.posit.co/dev/reference/Layer.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Define a custom Layer class. — Layer","text":"","code":"Layer(   classname,   initialize = NULL,   call = NULL,   build = NULL,   get_config = NULL,   ...,   public = list(),   private = list(),   inherit = NULL,   parent_env = parent.frame() )"},{"path":"https://keras3.posit.co/dev/reference/Layer.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Define a custom Layer class. — Layer","text":"classname String, name custom class. (Conventionally, CamelCase). initialize, call, build, get_config Recommended methods implement. See description details sections. ..., public Additional methods public members custom class. private Named list R objects (typically, functions) include instance private environments. private methods symbols scope public methods (See section \"Symbols Scope\"). instance private environment. objects private invisible Keras framework Python runtime. inherit custom class subclass. default, base keras class. parent_env R environment class methods grandparent.","code":""},{"path":"https://keras3.posit.co/dev/reference/Layer.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Define a custom Layer class. — Layer","text":"composing layer constructor, similar behavior layer functions like layer_dense(). first argument returned function object, enabling initialize()ing call() layer one step composing layer pipe, like   initialize() layer instance call() , pass missing NULL value object, pass arguments initialize() name.","code":"layer_foo <- Layer(\"Foo\", ....) output <- inputs |> layer_foo() layer <- layer_dense(units = 2, activation = \"relu\") layer <- layer_dense(NULL, 2, activation = \"relu\") layer <- layer_dense(, 2, activation = \"relu\")  # then you can call() the layer in a separate step outputs <- inputs |> layer()"},{"path":"https://keras3.posit.co/dev/reference/Layer.html","id":"symbols-in-scope","dir":"Reference","previous_headings":"","what":"Symbols in scope","title":"Define a custom Layer class. — Layer","text":"R function custom methods (public private) following symbols scope: self: custom class instance. super: custom class superclass. private: R environment specific class instance. objects assigned invisible Keras framework. __class__ .symbol(classname): custom class type object.","code":""},{"path":"https://keras3.posit.co/dev/reference/Layer.html","id":"attributes","dir":"Reference","previous_headings":"","what":"Attributes","title":"Define a custom Layer class. — Layer","text":"name: name layer (string). dtype: Dtype layer's weights. Alias layer$variable_dtype. variable_dtype: Dtype layer's weights. compute_dtype: dtype layer's computations. Layers automatically cast inputs dtype, causes computations output also dtype. mixed precision used keras$mixed_precision$DTypePolicy, different variable_dtype. trainable_weights: List variables included backprop. non_trainable_weights: List variables included backprop. weights: concatenation lists trainable_weights non_trainable_weights (order). trainable: Whether layer trained (boolean), .e. whether potentially-trainable weights returned part layer$trainable_weights. input_spec: Optional (list ) InputSpec object(s) specifying constraints inputs can accepted layer. recommend custom Layers implement following methods: initialize(): Defines custom layer attributes, creates layer weights depend input shapes, using add_weight(), state. build(input_shape): method can used create weights depend shape(s) input(s), using add_weight(), state. Calling call() automatically build layer (built yet) calling build(). call(...): Method called making sure build() called. call() performs logic applying layer input arguments. Two reserved arguments can optionally use call() : training (boolean, whether call inference mode training mode). mask (boolean tensor encoding masked timesteps input, used e.g. RNN layers). typical signature method call(inputs), user optionally add training mask layer need . get_config(): Returns named list containing configuration used initialize layer. list names differ arguments initialize(), override from_config() well. method used saving layer model contains layer.","code":""},{"path":"https://keras3.posit.co/dev/reference/Layer.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Define a custom Layer class. — Layer","text":"basic example: layer two variables, w b, returns y <- (w %*% x) + b. shows implement build() call(). Variables set attributes layer tracked weights layers (layer$weights).   Besides trainable weights, updated via backpropagation training, layers can also non-trainable weights. weights meant updated manually call(). example layer computes running sum inputs:","code":"layer_simple_dense <- Layer(   \"SimpleDense\",   initialize = function(units = 32) {     super$initialize()     self$units <- units   },    # Create the state of the layer (weights)   build = function(input_shape) {     self$kernel <- self$add_weight(       shape = shape(tail(input_shape, 1), self$units),       initializer = \"glorot_uniform\",       trainable = TRUE,       name = \"kernel\"     )     self$bias = self$add_weight(       shape = shape(self$units),       initializer = \"zeros\",       trainable = TRUE,       name = \"bias\"     )   },    # Defines the computation   call = function(self, inputs) {     op_matmul(inputs, self$kernel) + self$bias   } )  # Instantiates the layer. # Supply missing `object` arg to skip invoking `call()` and instead return # the Layer instance linear_layer <- layer_simple_dense(, 4)  # This will call `build(input_shape)` and create the weights, # and then invoke `call()`. y <- linear_layer(op_ones(c(2, 2))) stopifnot(length(linear_layer$weights) == 2)  # These weights are trainable, so they're listed in `trainable_weights`: stopifnot(length(linear_layer$trainable_weights) == 2) layer_compute_sum <- Layer(   classname = \"ComputeSum\",    initialize = function(input_dim) {     super$initialize()      # Create a non-trainable weight.     self$total <- self$add_weight(       shape = shape(),       initializer = \"zeros\",       trainable = FALSE,       name = \"total\"     )   },    call = function(inputs) {     self$total$assign(self$total + op_sum(inputs))     self$total   } )  my_sum <- layer_compute_sum(, 2) x <- op_ones(c(2, 2)) y <- my_sum(x)  stopifnot(exprs = {   all.equal(my_sum$weights,               list(my_sum$total))   all.equal(my_sum$non_trainable_weights, list(my_sum$total))   all.equal(my_sum$trainable_weights,     list()) })"},{"path":"https://keras3.posit.co/dev/reference/Layer.html","id":"methods-available","dir":"Reference","previous_headings":"","what":"Methods available","title":"Define a custom Layer class. — Layer","text":"Initialize self. method typically called custom initialize() method. Example:   Args: trainable: Boolean, whether layer's variables trainable. name: String name layer. dtype: dtype layer's computations weights. Can also keras$DTypePolicy, allows computation weight dtype differ. Defaults NULL. NULL means use config_dtype_policy(), \"float32\" policy unless set different value (via config_set_dtype_policy()).    Can called inside call() method add scalar loss. Example:         Add weight variable layer. Alias add_weight().    Add weight variable layer. Args: shape: shape variable (defined shape()) Must fully-defined (NA/NULL/-1 entries). Defaults () (scalar) unspecified. initializer: Initializer object use populate initial variable value, string name built-initializer (e.g. \"random_normal\"). unspecified, defaults \"glorot_uniform\" floating-point variables \"zeros\" types (e.g. int, bool). dtype: Dtype variable create, e.g. \"float32\". unspecified, defaults layer's variable dtype (defaults \"float32\" unspecified). trainable: Boolean, whether variable trainable via backprop whether updates managed manually. Defaults TRUE. autocast: Boolean, whether autocast layers variables accessing . Defaults TRUE. regularizer: Regularizer object call apply penalty weight. penalties summed loss function optimization. Defaults NULL. constraint: Constraint object call variable optimizer update, string name built-constraint. Defaults NULL. aggregation: Optional string, one NULL, \"none\", \"mean\", \"sum\" \"only_first_replica\". Annotates variable type multi-replica aggregation used variable writing custom data parallel training loops. Defaults \"none\". overwrite_with_gradient: Boolean, whether overwrite variable computed gradient. Useful float8 training. Defaults FALSE. name: String name variable. Useful debugging purposes. Returns: backend tensor, wrapped KerasVariable class. KerasVariable class Methods: assign(value) assign_add(value) assign_sub(value) numpy() (calling .array(<variable>) preferred) Properties/Attributes: value dtype ndim shape (calling shape(<variable>) preferred) trainable       Builds layer's states supplied config (named list args). default, method calls .call(build, config$input_shape) method, creates weights based layer's input shape supplied config. config contains information needed load layer's state, override method. Args: config: Named list containing input shape associated layer.    See description                Count total number scalars composing weights. Returns: integer count.    Returns named list layer's input shape. method returns config (named list) can used build_from_config(config) create states (e.g. Variables Lookup tables) needed layer. default, config contains input shape layer built . writing custom layer creates state unusual way, override method make sure state already created Keras attempts load value upon model loading. Returns: named list containing input shape associated layer.    Returns config object. object config named list (serializable) containing information needed re-instantiate . config expected serializable JSON, expected consist (potentially complex, nested) structure names lists consisting simple objects like strings, ints.    Return values layer$weights list R NumPy arrays.    Currently, Dense, EinsumDense Embedding layers support -place quantization via quantize() method. Example:            Enable rematerialization dynamically layer's call method. Args: layer_call: original call method layer. ...: additional args Returns: rematerialized version layer's call method.    Loads state layer. can override method take full control state layer loaded upon calling load_model(). Args: store: Named list state model loaded.    Saves state layer. can override method take full control state layer saved upon calling save_model(). Args: store: Named list state model saved.    Sets values weights list R NumPy arrays.    Call layer without side effects. Args: trainable_variables: List trainable variables model. non_trainable_variables: List non-trainable variables model. ...: Positional named arguments passed call(). return_losses: TRUE, stateless_call() return list losses created call() part return values. Returns: unnamed list. default, returns list(outputs, non_trainable_variables). return_losses = TRUE, returns list(outputs, non_trainable_variables, losses). Note: non_trainable_variables include non-trainable weights BatchNormalization statistics, also RNG seed state (random operations part layer, dropout), Metric state (metrics attached layer). elements state layer. Example:         Creates layer config. class method, meaning, R function self symbol (class instance) scope. Use __class__ classname symbol provided Layer() constructed) resolve class definition. default implementation :   method reverse get_config(), capable instantiating layer config named list. handle layer connectivity (handled Network), weights (handled set_weights()). Args: config: named list, typically output get_config(). Returns: layer instance.","code":"initialize(...,            activity_regularizer = NULL,            trainable = TRUE,            dtype = NULL,            autocast = TRUE,            name = NULL) layer_my_layer <- Layer(\"MyLayer\",   initialize = function(units, ..., dtype = NULL, name = NULL) {     super$initialize(..., dtype = dtype, name = name)     # .... finish initializing `self` instance   } ) add_loss(loss) Layer(\"MyLayer\",   ...   call = function(x) {     self$add_loss(op_sum(x))     x   } add_metric(...) add_variable(...) add_weight(shape = NULL,            initializer = NULL,            dtype = NULL,            trainable = TRUE,            autocast = TRUE,            regularizer = NULL,            constraint = NULL,            aggregation = 'none',            overwrite_with_gradient = FALSE,            name = NULL) build(input_shape) build_from_config(config) call(...) compute_mask(inputs, previous_mask) compute_output_shape(...) compute_output_spec(...) symbolic_call(...) count_params() get_build_config() get_config() get_weights() quantize(mode, type_check = TRUE) model$quantize(\"int8\") # quantize model in-place model |> predict(data) # faster inference quantized_build(input_shape, mode) quantized_call(...) rematerialized_call(layer_call, ...) load_own_variables(store) save_own_variables(store) set_weights(weights) stateless_call(trainable_variables, non_trainable_variables,                ..., return_losses = FALSE) model <- ... data <- ... trainable_variables <- model$trainable_variables non_trainable_variables <- model$non_trainable_variables # Call the model with zero side effects c(outputs, non_trainable_variables) %<-% model$stateless_call(     trainable_variables,     non_trainable_variables,     data ) # Attach the updated state to the model # (until you do this, the model is still in its pre-call state). purrr::walk2(   model$non_trainable_variables, non_trainable_variables,   \\(variable, value) variable$assign(value)) symbolic_call(...) from_config(config) from_config = function(config) {   do.call(`__class__`, config) }"},{"path":"https://keras3.posit.co/dev/reference/Layer.html","id":"readonly-properties-","dir":"Reference","previous_headings":"","what":"Readonly properties:","title":"Define a custom Layer class. — Layer","text":"compute_dtype dtype computations performed layer. dtype Alias layer$variable_dtype. input_dtype dtype layer inputs converted . losses List scalar losses add_loss(), regularizers sublayers. metrics List metrics. metrics_variables List metric variables. non_trainable_variables List non-trainable layer state. extends layer$non_trainable_weights include state used layer including state metrics SeedGenerators. non_trainable_weights List non-trainable weight variables layer. weights updated optimizer training. Unlike, layer$non_trainable_variables excludes metric state random seeds. trainable_variables List trainable layer state. equivalent layer$trainable_weights. trainable_weights List trainable weight variables layer. weights get updated optimizer training. path path layer. layer built yet, NULL. quantization_mode quantization mode layer, NULL quantized. variable_dtype dtype state (weights) layer. variables List layer state, including random seeds. extends layer$weights include state used layer including SeedGenerators. Note metrics variables included , use metrics_variables visit metric variables. weights List weight variables layer. Unlike, layer$variables excludes metric state random seeds. input Retrieves input tensor(s) symbolic operation. returns tensor(s) corresponding first time operation called. Returns: Input tensor list input tensors. output Retrieves output tensor(s) layer. returns tensor(s) corresponding first time operation called. Returns: Output tensor list output tensors.","code":""},{"path":"https://keras3.posit.co/dev/reference/Layer.html","id":"data-descriptors-attributes-","dir":"Reference","previous_headings":"","what":"Data descriptors (Attributes):","title":"Define a custom Layer class. — Layer","text":"dtype_policy input_spec supports_masking Whether layer supports computing mask using compute_mask. trainable Settable boolean, whether layer trainable .","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/LearningRateSchedule.html","id":null,"dir":"Reference","previous_headings":"","what":"Define a custom LearningRateSchedule class — LearningRateSchedule","title":"Define a custom LearningRateSchedule class — LearningRateSchedule","text":"Subclass Keras LearningRateSchedule base class. can use learning rate schedule modulate learning rate optimizer changes time. Several built-learning rate schedules available, learning_rate_schedule_exponential_decay() learning_rate_schedule_piecewise_constant_decay():   LearningRateSchedule() instance can passed learning_rate argument optimizer. implement schedule object, implement call method, takes step argument (scalar integer backend tensor, current training step count). Note step 0-based (.e., first step 0). Like Keras object, can also optionally make object serializable implementing get_config() from_config() methods.","code":"lr_schedule <- learning_rate_schedule_exponential_decay(   initial_learning_rate = 1e-2,   decay_steps = 10000,   decay_rate = 0.9 ) optimizer <- optimizer_sgd(learning_rate = lr_schedule)"},{"path":"https://keras3.posit.co/dev/reference/LearningRateSchedule.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Define a custom LearningRateSchedule class — LearningRateSchedule","text":"","code":"LearningRateSchedule(   classname,   call = NULL,   initialize = NULL,   get_config = NULL,   ...,   public = list(),   private = list(),   inherit = NULL,   parent_env = parent.frame() )"},{"path":"https://keras3.posit.co/dev/reference/LearningRateSchedule.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Define a custom LearningRateSchedule class — LearningRateSchedule","text":"classname String, name custom class. (Conventionally, CamelCase). call, initialize, get_config Recommended methods implement. See description details sections. ..., public Additional methods public members custom class. private Named list R objects (typically, functions) include instance private environments. private methods symbols scope public methods (See section \"Symbols Scope\"). instance private environment. objects private invisible Keras framework Python runtime. inherit custom class subclass. default, base keras class. parent_env R environment class methods grandparent.","code":""},{"path":"https://keras3.posit.co/dev/reference/LearningRateSchedule.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Define a custom LearningRateSchedule class — LearningRateSchedule","text":"function returns LearningRateSchedule instances, similar built-learning_rate_schedule_* family functions.","code":""},{"path":"https://keras3.posit.co/dev/reference/LearningRateSchedule.html","id":"example","dir":"Reference","previous_headings":"","what":"Example","title":"Define a custom LearningRateSchedule class — LearningRateSchedule","text":"","code":"my_custom_learning_rate_schedule <- LearningRateSchedule(   classname = \"MyLRSchedule\",    initialize = function(initial_learning_rate) {     self$initial_learning_rate <- initial_learning_rate   },    call = function(step) {     # note that `step` is a tensor     # and call() will be traced via tf_function() or similar.      str(step) # <KerasVariable shape=(), dtype=int64, path=SGD/iteration>      # print 'step' every 1000 steps     op_cond((step %% 1000) == 0,             \\() {tensorflow::tf$print(step); NULL},             \\() {NULL})     self$initial_learning_rate / (step + 1)   } )  optimizer <- optimizer_sgd(   learning_rate = my_custom_learning_rate_schedule(0.1) )  # You can also call schedule instances directly # (e.g., for interactive testing, or if implementing a custom optimizer) schedule <- my_custom_learning_rate_schedule(0.1) step <- keras$Variable(initializer = op_ones,                        shape = shape(),                        dtype = \"int64\") schedule(step) ## <Variable path=variable, shape=(), dtype=int64, value=1> ## tf.Tensor(0.0, shape=(), dtype=float64)"},{"path":"https://keras3.posit.co/dev/reference/LearningRateSchedule.html","id":"methods-available-","dir":"Reference","previous_headings":"","what":"Methods available:","title":"Define a custom LearningRateSchedule class — LearningRateSchedule","text":"","code":"get_config()"},{"path":"https://keras3.posit.co/dev/reference/LearningRateSchedule.html","id":"symbols-in-scope","dir":"Reference","previous_headings":"","what":"Symbols in scope","title":"Define a custom LearningRateSchedule class — LearningRateSchedule","text":"R function custom methods (public private) following symbols scope: self: custom class instance. super: custom class superclass. private: R environment specific class instance. objects assigned invisible Keras framework. __class__ .symbol(classname): custom class type object.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/Loss.html","id":null,"dir":"Reference","previous_headings":"","what":"Subclass the base Loss class — Loss","title":"Subclass the base Loss class — Loss","text":"Use define custom loss class. Note, cases need subclass Loss define custom loss: can also pass bare R function, named R function defined custom_metric(), loss function compile().","code":""},{"path":"https://keras3.posit.co/dev/reference/Loss.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Subclass the base Loss class — Loss","text":"","code":"Loss(   classname,   call = NULL,   ...,   public = list(),   private = list(),   inherit = NULL,   parent_env = parent.frame() )"},{"path":"https://keras3.posit.co/dev/reference/Loss.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Subclass the base Loss class — Loss","text":"classname String, name custom class. (Conventionally, CamelCase). call   Method implemented subclasses: Function contains logic loss calculation using y_true, y_pred. ..., public Additional methods public members custom class. private Named list R objects (typically, functions) include instance private environments. private methods symbols scope public methods (See section \"Symbols Scope\"). instance private environment. objects private invisible Keras framework Python runtime. inherit custom class subclass. default, base keras class. parent_env R environment class methods grandparent.","code":"function(y_true, y_pred)"},{"path":"https://keras3.posit.co/dev/reference/Loss.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Subclass the base Loss class — Loss","text":"function returns Loss instances, similar builtin loss functions.","code":""},{"path":"https://keras3.posit.co/dev/reference/Loss.html","id":"details","dir":"Reference","previous_headings":"","what":"Details","title":"Subclass the base Loss class — Loss","text":"Example subclass implementation:","code":"loss_custom_mse <- Loss(   classname = \"CustomMeanSquaredError\",   call = function(y_true, y_pred) {     op_mean(op_square(y_pred - y_true), axis = -1)   } )  # Usage in compile() model <- keras_model_sequential(input_shape = 10) |> layer_dense(10) model |> compile(loss = loss_custom_mse())  # Standalone usage mse <- loss_custom_mse(name = \"my_custom_mse_instance\")  y_true <- op_arange(20) |> op_reshape(c(4, 5)) y_pred <- op_arange(20) |> op_reshape(c(4, 5)) * 2 (loss <- mse(y_true, y_pred)) ## tf.Tensor(143.5, shape=(), dtype=float32) loss2 <- (y_pred - y_true)^2 |>   op_mean(axis = -1) |>   op_mean()  stopifnot(all.equal(as.array(loss), as.array(loss2)))  sample_weight <-array(c(.25, .25, 1, 1)) (weighted_loss <- mse(y_true, y_pred, sample_weight = sample_weight)) ## tf.Tensor(129.0625, shape=(), dtype=float32) weighted_loss2 <- (y_true - y_pred)^2 |>   op_mean(axis = -1) |>   op_multiply(sample_weight) |>   op_mean()  stopifnot(all.equal(as.array(weighted_loss),                     as.array(weighted_loss2)))"},{"path":"https://keras3.posit.co/dev/reference/Loss.html","id":"methods-defined-by-base-loss-class-","dir":"Reference","previous_headings":"","what":"Methods defined by base Loss class:","title":"Subclass the base Loss class — Loss","text":"Args: name: Optional name loss instance. reduction: Type reduction apply loss. almost cases \"sum_over_batch_size\". Supported options \"sum\", \"sum_over_batch_size\", \"mean\", \"mean_with_sample_weight\" NULL. \"sum\" sums loss, \"sum_over_batch_size\" \"mean\" sum loss divide sample size, \"mean_with_sample_weight\" sums loss divides sum sample weights. \"none\" NULL perform aggregation. Defaults \"sum_over_batch_size\". dtype: dtype loss's computations. Defaults NULL, means using config_floatx(). config_floatx() \"float32\" unless set different value (via config_set_floatx()). keras$DTypePolicy provided, compute_dtype utilized.    Call loss instance function, optionally sample_weight.","code":"initialize(name=NULL, reduction=\"sum_over_batch_size\", dtype=NULL) __call__(y_true, y_pred, sample_weight=NULL) get_config()"},{"path":"https://keras3.posit.co/dev/reference/Loss.html","id":"readonly-properties-","dir":"Reference","previous_headings":"","what":"Readonly properties:","title":"Subclass the base Loss class — Loss","text":"","code":"dtype"},{"path":"https://keras3.posit.co/dev/reference/Loss.html","id":"symbols-in-scope","dir":"Reference","previous_headings":"","what":"Symbols in scope","title":"Subclass the base Loss class — Loss","text":"R function custom methods (public private) following symbols scope: self: custom class instance. super: custom class superclass. private: R environment specific class instance. objects assigned invisible Keras framework. __class__ .symbol(classname): custom class type object.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/Metric.html","id":null,"dir":"Reference","previous_headings":"","what":"Subclass the base Metric class — Metric","title":"Subclass the base Metric class — Metric","text":"Metric object encapsulates metric logic state can used track model performance training. returned family metric functions start prefix metric_*, well returned custom metrics defined Metric().","code":""},{"path":"https://keras3.posit.co/dev/reference/Metric.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Subclass the base Metric class — Metric","text":"","code":"Metric(   classname,   initialize = NULL,   update_state = NULL,   result = NULL,   ...,   public = list(),   private = list(),   inherit = NULL,   parent_env = parent.frame() )"},{"path":"https://keras3.posit.co/dev/reference/Metric.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Subclass the base Metric class — Metric","text":"classname String, name custom class. (Conventionally, CamelCase). initialize, update_state, result Recommended methods implement. See description section. ..., public Additional methods public members custom class. private Named list R objects (typically, functions) include instance private environments. private methods symbols scope public methods (See section \"Symbols Scope\"). instance private environment. objects private invisible Keras framework Python runtime. inherit custom class subclass. default, base keras class. parent_env R environment class methods grandparent.","code":""},{"path":"https://keras3.posit.co/dev/reference/Metric.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Subclass the base Metric class — Metric","text":"function returns Metric instances, similar builtin metric functions.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/Metric.html","id":"usage-with-compile-","dir":"Reference","previous_headings":"","what":"Usage with compile():","title":"Subclass the base Metric class — Metric","text":"","code":"model |> compile(   optimizer = 'sgd',   loss = 'mse',   metrics = c(metric_SOME_METRIC(), metric_SOME_OTHER_METRIC()) )"},{"path":"https://keras3.posit.co/dev/reference/Metric.html","id":"standalone-usage-","dir":"Reference","previous_headings":"","what":"Standalone usage:","title":"Subclass the base Metric class — Metric","text":"","code":"m <- metric_SOME_METRIC() for (e in seq(epochs)) {   for (i in seq(train_steps)) {     c(y_true, y_pred, sample_weight = NULL) %<-% ...     m$update_state(y_true, y_pred, sample_weight)   }   cat('Final epoch result: ', as.numeric(m$result()), \"\\n\")   m$reset_state() }"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/Metric.html","id":"usage-with-compile--1","dir":"Reference","previous_headings":"","what":"Usage with compile():","title":"Subclass the base Metric class — Metric","text":"implemented subclasses (custom metrics): initialize(): state variables created method calling self$add_variable() like: self$var <- self$add_variable(...). update_state(): Updates state variables like: self$var$assign(...). result(): Computes returns scalar value named list scalar values metric state variables. Example subclass implementation:","code":"model <- keras_model_sequential() model |>   layer_dense(64, activation = \"relu\") |>   layer_dense(64, activation = \"relu\") |>   layer_dense(10, activation = \"softmax\") model |>   compile(optimizer = optimizer_rmsprop(0.01),           loss = loss_categorical_crossentropy(),           metrics = metric_categorical_accuracy())  data <- random_uniform(c(1000, 32)) labels <- random_uniform(c(1000, 10))  model |> fit(data, labels, verbose = 0) metric_binary_true_positives <- Metric(   classname = \"BinaryTruePositives\",    initialize = function(name = 'binary_true_positives', ...) {     super$initialize(name = name, ...)     self$true_positives <-       self$add_weight(shape = shape(),                       initializer = 'zeros',                       name = 'true_positives')   },    update_state = function(y_true, y_pred, sample_weight = NULL) {     y_true <- op_cast(y_true, \"bool\")     y_pred <- op_cast(y_pred, \"bool\")      values <- y_true & y_pred # `&` calls op_logical_and()     values <- op_cast(values, self$dtype)     if (!is.null(sample_weight)) {       sample_weight <- op_cast(sample_weight, self$dtype)       sample_weight <- op_broadcast_to(sample_weight, shape(values))       values <- values * sample_weight # `*` calls op_multiply()     }     self$true_positives$assign(self$true_positives + op_sum(values))   },    result = function() {     self$true_positives   } ) model <- keras_model_sequential(input_shape = 32) |> layer_dense(10) model |> compile(loss = loss_binary_crossentropy(),                  metrics = list(metric_binary_true_positives())) model |> fit(data, labels, verbose = 0)"},{"path":"https://keras3.posit.co/dev/reference/Metric.html","id":"methods-defined-by-the-base-metric-class-","dir":"Reference","previous_headings":"","what":"Methods defined by the base Metric class:","title":"Subclass the base Metric class — Metric","text":"Calling metric instance self like m(...) equivalent calling:      Initialize self. Args: name: Optional name metric instance. dtype: dtype metric's computations. Defaults NULL, means using config_floatx(). config_floatx() \"float32\" unless set different value (via config_set_floatx()). keras$DTypePolicy provided, compute_dtype utilized.          Return serializable config metric.    Reset metric state variables. function called epochs/steps, metric evaluated training.    Compute current metric value. Returns: scalar tensor, named list scalar tensors.             Accumulate statistics metric.","code":"__call__(...) function(...) {   m$update_state(...)   m$result() } initialize(dtype=NULL, name=NULL) add_variable(shape, initializer, dtype=NULL, aggregation = 'sum', name=NULL) add_weight(shape=shape(), initializer=NULL, dtype=NULL, name=NULL) get_config() reset_state() result() stateless_result(metric_variables) stateless_reset_state() stateless_update_state(metric_variables, ...) update_state(...)"},{"path":"https://keras3.posit.co/dev/reference/Metric.html","id":"readonly-properties","dir":"Reference","previous_headings":"","what":"Readonly properties","title":"Subclass the base Metric class — Metric","text":"dtype variables","code":""},{"path":"https://keras3.posit.co/dev/reference/Metric.html","id":"symbols-in-scope","dir":"Reference","previous_headings":"","what":"Symbols in scope","title":"Subclass the base Metric class — Metric","text":"R function custom methods (public private) following symbols scope: self: custom class instance. super: custom class superclass. private: R environment specific class instance. objects assigned invisible Keras framework. __class__ .symbol(classname): custom class type object.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/Model.html","id":null,"dir":"Reference","previous_headings":"","what":"Subclass the base Keras Model Class — Model","title":"Subclass the base Keras Model Class — Model","text":"advanced use cases need subclass base Model type, e.g., want override train_step() method. just want create define keras model, prefer keras_model() keras_model_sequential(). just want encapsulate custom logic state, need customize training behavior (besides calling self$add_loss() call() method), prefer Layer().","code":""},{"path":"https://keras3.posit.co/dev/reference/Model.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Subclass the base Keras Model Class — Model","text":"","code":"Model(   classname,   initialize = NULL,   call = NULL,   train_step = NULL,   predict_step = NULL,   test_step = NULL,   compute_loss = NULL,   compute_metrics = NULL,   ...,   public = list(),   private = list(),   inherit = NULL,   parent_env = parent.frame() )"},{"path":"https://keras3.posit.co/dev/reference/Model.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Subclass the base Keras Model Class — Model","text":"classname String, name custom class. (Conventionally, CamelCase). initialize, call, train_step, predict_step, test_step, compute_loss, compute_metrics Optional methods can overridden. ..., public Additional methods public members custom class. private Named list R objects (typically, functions) include instance private environments. private methods symbols scope public methods (See section \"Symbols Scope\"). instance private environment. objects private invisible Keras framework Python runtime. inherit custom class subclass. default, base keras class. parent_env R environment class methods grandparent.","code":""},{"path":"https://keras3.posit.co/dev/reference/Model.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Subclass the base Keras Model Class — Model","text":"model constructor function, can call create instance new model type.","code":""},{"path":"https://keras3.posit.co/dev/reference/Model.html","id":"symbols-in-scope","dir":"Reference","previous_headings":"","what":"Symbols in scope","title":"Subclass the base Keras Model Class — Model","text":"R function custom methods (public private) following symbols scope: self: custom class instance. super: custom class superclass. private: R environment specific class instance. objects assigned invisible Keras framework. __class__ .symbol(classname): custom class type object.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/activation_celu.html","id":null,"dir":"Reference","previous_headings":"","what":"Continuously Differentiable Exponential Linear Unit. — activation_celu","title":"Continuously Differentiable Exponential Linear Unit. — activation_celu","text":"CeLU activation function defined : celu(x) = alpha * (exp(x / alpha) - 1) x < 0,celu(x) = x x >= 0. alpha scaling parameter controls activation's shape.","code":""},{"path":"https://keras3.posit.co/dev/reference/activation_celu.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Continuously Differentiable Exponential Linear Unit. — activation_celu","text":"","code":"activation_celu(x, alpha = 1)"},{"path":"https://keras3.posit.co/dev/reference/activation_celu.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Continuously Differentiable Exponential Linear Unit. — activation_celu","text":"x Input tensor. alpha  value CeLU formulation. Defaults 1.0.","code":""},{"path":"https://keras3.posit.co/dev/reference/activation_celu.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Continuously Differentiable Exponential Linear Unit. — activation_celu","text":"tensor, result applying activation input tensor x.","code":""},{"path":"https://keras3.posit.co/dev/reference/activation_celu.html","id":"reference","dir":"Reference","previous_headings":"","what":"Reference","title":"Continuously Differentiable Exponential Linear Unit. — activation_celu","text":"Barron, J. T., 2017","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/activation_elu.html","id":null,"dir":"Reference","previous_headings":"","what":"Exponential Linear Unit. — activation_elu","title":"Exponential Linear Unit. — activation_elu","text":"exponential linear unit (ELU) alpha > 0 defined : x x > 0 alpha * exp(x) - 1 x < 0 ELUs negative values pushes mean activations closer zero. Mean activations closer zero enable faster learning bring gradient closer natural gradient. ELUs saturate negative value argument gets smaller. Saturation means small derivative decreases variation information propagated next layer.","code":""},{"path":"https://keras3.posit.co/dev/reference/activation_elu.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Exponential Linear Unit. — activation_elu","text":"","code":"activation_elu(x, alpha = 1)"},{"path":"https://keras3.posit.co/dev/reference/activation_elu.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Exponential Linear Unit. — activation_elu","text":"x Input tensor. alpha scalar, slope positive section. Defaults 1.0.","code":""},{"path":"https://keras3.posit.co/dev/reference/activation_elu.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Exponential Linear Unit. — activation_elu","text":"tensor, result applying activation input tensor x.","code":""},{"path":"https://keras3.posit.co/dev/reference/activation_elu.html","id":"reference","dir":"Reference","previous_headings":"","what":"Reference","title":"Exponential Linear Unit. — activation_elu","text":"Clevert et al., 2016","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/activation_exponential.html","id":null,"dir":"Reference","previous_headings":"","what":"Exponential activation function. — activation_exponential","title":"Exponential activation function. — activation_exponential","text":"Exponential activation function.","code":""},{"path":"https://keras3.posit.co/dev/reference/activation_exponential.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Exponential activation function. — activation_exponential","text":"","code":"activation_exponential(x)"},{"path":"https://keras3.posit.co/dev/reference/activation_exponential.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Exponential activation function. — activation_exponential","text":"x Input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/activation_exponential.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Exponential activation function. — activation_exponential","text":"tensor, result applying activation input tensor x.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/activation_gelu.html","id":null,"dir":"Reference","previous_headings":"","what":"Gaussian error linear unit (GELU) activation function. — activation_gelu","title":"Gaussian error linear unit (GELU) activation function. — activation_gelu","text":"Gaussian error linear unit (GELU) defined : gelu(x) = x * P(X <= x) P(X) ~ N(0, 1), .e. gelu(x) = 0.5 * x * (1 + erf(x / sqrt(2))). GELU weights inputs value, rather gating inputs sign ReLU.","code":""},{"path":"https://keras3.posit.co/dev/reference/activation_gelu.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Gaussian error linear unit (GELU) activation function. — activation_gelu","text":"","code":"activation_gelu(x, approximate = FALSE)"},{"path":"https://keras3.posit.co/dev/reference/activation_gelu.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Gaussian error linear unit (GELU) activation function. — activation_gelu","text":"x Input tensor. approximate bool, whether enable approximation.","code":""},{"path":"https://keras3.posit.co/dev/reference/activation_gelu.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Gaussian error linear unit (GELU) activation function. — activation_gelu","text":"tensor, result applying activation input tensor x.","code":""},{"path":"https://keras3.posit.co/dev/reference/activation_gelu.html","id":"reference","dir":"Reference","previous_headings":"","what":"Reference","title":"Gaussian error linear unit (GELU) activation function. — activation_gelu","text":"Hendrycks et al., 2016","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/activation_glu.html","id":null,"dir":"Reference","previous_headings":"","what":"Gated Linear Unit (GLU) activation function. — activation_glu","title":"Gated Linear Unit (GLU) activation function. — activation_glu","text":"GLU activation function defined : glu(x) = * sigmoid(b), x split two equal parts b along given axis.","code":""},{"path":"https://keras3.posit.co/dev/reference/activation_glu.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Gated Linear Unit (GLU) activation function. — activation_glu","text":"","code":"activation_glu(x, axis = -1L)"},{"path":"https://keras3.posit.co/dev/reference/activation_glu.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Gated Linear Unit (GLU) activation function. — activation_glu","text":"x Input tensor. axis axis along split input tensor. Defaults -1.","code":""},{"path":"https://keras3.posit.co/dev/reference/activation_glu.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Gated Linear Unit (GLU) activation function. — activation_glu","text":"tensor, result applying activation input tensor x.","code":""},{"path":"https://keras3.posit.co/dev/reference/activation_glu.html","id":"reference","dir":"Reference","previous_headings":"","what":"Reference","title":"Gated Linear Unit (GLU) activation function. — activation_glu","text":"Dauphin et al., 2017","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/activation_hard_shrink.html","id":null,"dir":"Reference","previous_headings":"","what":"Hard Shrink activation function. — activation_hard_shrink","title":"Hard Shrink activation function. — activation_hard_shrink","text":"defined : hard_shrink(x) = x |x| > threshold, hard_shrink(x) = 0 otherwise.","code":""},{"path":"https://keras3.posit.co/dev/reference/activation_hard_shrink.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Hard Shrink activation function. — activation_hard_shrink","text":"","code":"activation_hard_shrink(x, threshold = 0.5)"},{"path":"https://keras3.posit.co/dev/reference/activation_hard_shrink.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Hard Shrink activation function. — activation_hard_shrink","text":"x Input tensor. threshold Threshold value. Defaults 0.5.","code":""},{"path":"https://keras3.posit.co/dev/reference/activation_hard_shrink.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Hard Shrink activation function. — activation_hard_shrink","text":"tensor, result applying activation input tensor x.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/activation_hard_sigmoid.html","id":null,"dir":"Reference","previous_headings":"","what":"Hard sigmoid activation function. — activation_hard_sigmoid","title":"Hard sigmoid activation function. — activation_hard_sigmoid","text":"hard sigmoid activation defined : 0 x <= -3 1 x >= 3 (x/6) + 0.5 -3 < x < 3 faster, piecewise linear approximation sigmoid activation.","code":""},{"path":"https://keras3.posit.co/dev/reference/activation_hard_sigmoid.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Hard sigmoid activation function. — activation_hard_sigmoid","text":"","code":"activation_hard_sigmoid(x)"},{"path":"https://keras3.posit.co/dev/reference/activation_hard_sigmoid.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Hard sigmoid activation function. — activation_hard_sigmoid","text":"x Input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/activation_hard_sigmoid.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Hard sigmoid activation function. — activation_hard_sigmoid","text":"tensor, result applying activation input tensor x.","code":""},{"path":"https://keras3.posit.co/dev/reference/activation_hard_sigmoid.html","id":"reference","dir":"Reference","previous_headings":"","what":"Reference","title":"Hard sigmoid activation function. — activation_hard_sigmoid","text":"Wikipedia \"Hard sigmoid\"","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/activation_hard_silu.html","id":null,"dir":"Reference","previous_headings":"","what":"Hard SiLU activation function, also known as Hard Swish. — activation_hard_silu","title":"Hard SiLU activation function, also known as Hard Swish. — activation_hard_silu","text":"defined : 0 x < -3 x x > 3 x * (x + 3) / 6 -3 <= x <= 3 faster, piecewise linear approximation silu activation.","code":""},{"path":"https://keras3.posit.co/dev/reference/activation_hard_silu.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Hard SiLU activation function, also known as Hard Swish. — activation_hard_silu","text":"","code":"activation_hard_silu(x)  activation_hard_swish(x)"},{"path":"https://keras3.posit.co/dev/reference/activation_hard_silu.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Hard SiLU activation function, also known as Hard Swish. — activation_hard_silu","text":"x Input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/activation_hard_silu.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Hard SiLU activation function, also known as Hard Swish. — activation_hard_silu","text":"tensor, result applying activation input tensor x.","code":""},{"path":"https://keras3.posit.co/dev/reference/activation_hard_silu.html","id":"reference","dir":"Reference","previous_headings":"","what":"Reference","title":"Hard SiLU activation function, also known as Hard Swish. — activation_hard_silu","text":"Howard, 2019","code":""},{"path":"https://keras3.posit.co/dev/reference/activation_hard_tanh.html","id":null,"dir":"Reference","previous_headings":"","what":"HardTanh activation function. — activation_hard_tanh","title":"HardTanh activation function. — activation_hard_tanh","text":"defined : hard_tanh(x) = -1 x < -1, hard_tanh(x) = x -1 <= x <= 1, hard_tanh(x) = 1 x > 1.","code":""},{"path":"https://keras3.posit.co/dev/reference/activation_hard_tanh.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"HardTanh activation function. — activation_hard_tanh","text":"","code":"activation_hard_tanh(x)"},{"path":"https://keras3.posit.co/dev/reference/activation_hard_tanh.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"HardTanh activation function. — activation_hard_tanh","text":"x Input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/activation_hard_tanh.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"HardTanh activation function. — activation_hard_tanh","text":"tensor, result applying activation input tensor x.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/activation_leaky_relu.html","id":null,"dir":"Reference","previous_headings":"","what":"Leaky relu activation function. — activation_leaky_relu","title":"Leaky relu activation function. — activation_leaky_relu","text":"Leaky relu activation function.","code":""},{"path":"https://keras3.posit.co/dev/reference/activation_leaky_relu.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Leaky relu activation function. — activation_leaky_relu","text":"","code":"activation_leaky_relu(x, negative_slope = 0.2)"},{"path":"https://keras3.posit.co/dev/reference/activation_leaky_relu.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Leaky relu activation function. — activation_leaky_relu","text":"x Input tensor. negative_slope float controls slope values lower threshold.","code":""},{"path":"https://keras3.posit.co/dev/reference/activation_leaky_relu.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Leaky relu activation function. — activation_leaky_relu","text":"tensor, result applying activation input tensor x.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/activation_linear.html","id":null,"dir":"Reference","previous_headings":"","what":"Linear activation function (pass-through). — activation_linear","title":"Linear activation function (pass-through). — activation_linear","text":"\"linear\" activation identity function: returns input, unmodified.","code":""},{"path":"https://keras3.posit.co/dev/reference/activation_linear.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Linear activation function (pass-through). — activation_linear","text":"","code":"activation_linear(x)"},{"path":"https://keras3.posit.co/dev/reference/activation_linear.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Linear activation function (pass-through). — activation_linear","text":"x Input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/activation_linear.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Linear activation function (pass-through). — activation_linear","text":"tensor, result applying activation input tensor x.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/activation_log_sigmoid.html","id":null,"dir":"Reference","previous_headings":"","what":"Logarithm of the sigmoid activation function. — activation_log_sigmoid","title":"Logarithm of the sigmoid activation function. — activation_log_sigmoid","text":"defined f(x) = log(1 / (1 + exp(-x))).","code":""},{"path":"https://keras3.posit.co/dev/reference/activation_log_sigmoid.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Logarithm of the sigmoid activation function. — activation_log_sigmoid","text":"","code":"activation_log_sigmoid(x)"},{"path":"https://keras3.posit.co/dev/reference/activation_log_sigmoid.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Logarithm of the sigmoid activation function. — activation_log_sigmoid","text":"x Input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/activation_log_sigmoid.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Logarithm of the sigmoid activation function. — activation_log_sigmoid","text":"tensor, result applying activation input tensor x.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/activation_log_softmax.html","id":null,"dir":"Reference","previous_headings":"","what":"Log-Softmax activation function. — activation_log_softmax","title":"Log-Softmax activation function. — activation_log_softmax","text":"input vector handled independently. axis argument sets axis input function applied along.","code":""},{"path":"https://keras3.posit.co/dev/reference/activation_log_softmax.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Log-Softmax activation function. — activation_log_softmax","text":"","code":"activation_log_softmax(x, axis = -1L)"},{"path":"https://keras3.posit.co/dev/reference/activation_log_softmax.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Log-Softmax activation function. — activation_log_softmax","text":"x Input tensor. axis Integer, axis along softmax applied.","code":""},{"path":"https://keras3.posit.co/dev/reference/activation_log_softmax.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Log-Softmax activation function. — activation_log_softmax","text":"tensor, result applying activation input tensor x.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/activation_mish.html","id":null,"dir":"Reference","previous_headings":"","what":"Mish activation function. — activation_mish","title":"Mish activation function. — activation_mish","text":"defined : mish(x) = x * tanh(softplus(x)) softplus defined : softplus(x) = log(exp(x) + 1)","code":""},{"path":"https://keras3.posit.co/dev/reference/activation_mish.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Mish activation function. — activation_mish","text":"","code":"activation_mish(x)"},{"path":"https://keras3.posit.co/dev/reference/activation_mish.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Mish activation function. — activation_mish","text":"x Input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/activation_mish.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Mish activation function. — activation_mish","text":"tensor, result applying activation input tensor x.","code":""},{"path":"https://keras3.posit.co/dev/reference/activation_mish.html","id":"reference","dir":"Reference","previous_headings":"","what":"Reference","title":"Mish activation function. — activation_mish","text":"Misra, 2019","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/activation_relu.html","id":null,"dir":"Reference","previous_headings":"","what":"Applies the rectified linear unit activation function. — activation_relu","title":"Applies the rectified linear unit activation function. — activation_relu","text":"default values, returns standard ReLU activation: max(x, 0), element-wise maximum 0 input tensor. Modifying default parameters allows use non-zero thresholds, change max value activation, use non-zero multiple input values threshold.","code":""},{"path":"https://keras3.posit.co/dev/reference/activation_relu.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Applies the rectified linear unit activation function. — activation_relu","text":"","code":"activation_relu(x, negative_slope = 0, max_value = NULL, threshold = 0)"},{"path":"https://keras3.posit.co/dev/reference/activation_relu.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Applies the rectified linear unit activation function. — activation_relu","text":"x Input tensor. negative_slope numeric controls slope values lower threshold. max_value numeric sets saturation threshold (largest value function return). threshold numeric giving threshold value activation function values damped set zero.","code":""},{"path":"https://keras3.posit.co/dev/reference/activation_relu.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Applies the rectified linear unit activation function. — activation_relu","text":"tensor shape dtype input x.","code":""},{"path":"https://keras3.posit.co/dev/reference/activation_relu.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Applies the rectified linear unit activation function. — activation_relu","text":"","code":"x <- c(-10, -5, 0, 5, 10) activation_relu(x) ## tf.Tensor([ 0.  0.  0.  5. 10.], shape=(5), dtype=float32) activation_relu(x, negative_slope = 0.5) ## tf.Tensor([-5.  -2.5  0.   5.  10. ], shape=(5), dtype=float32) activation_relu(x, max_value = 5) ## tf.Tensor([0. 0. 0. 5. 5.], shape=(5), dtype=float32) activation_relu(x, threshold = 5) ## tf.Tensor([-0. -0.  0.  0. 10.], shape=(5), dtype=float32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/activation_relu6.html","id":null,"dir":"Reference","previous_headings":"","what":"Relu6 activation function. — activation_relu6","title":"Relu6 activation function. — activation_relu6","text":"ReLU function, truncated maximum value 6.","code":""},{"path":"https://keras3.posit.co/dev/reference/activation_relu6.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Relu6 activation function. — activation_relu6","text":"","code":"activation_relu6(x)"},{"path":"https://keras3.posit.co/dev/reference/activation_relu6.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Relu6 activation function. — activation_relu6","text":"x Input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/activation_relu6.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Relu6 activation function. — activation_relu6","text":"tensor, result applying activation input tensor x.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/activation_selu.html","id":null,"dir":"Reference","previous_headings":"","what":"Scaled Exponential Linear Unit (SELU). — activation_selu","title":"Scaled Exponential Linear Unit (SELU). — activation_selu","text":"Scaled Exponential Linear Unit (SELU) activation function defined : scale * x x > 0 scale * alpha * (exp(x) - 1) x < 0 alpha scale pre-defined constants (alpha = 1.67326324 scale = 1.05070098). Basically, SELU activation function multiplies scale (> 1) output activation_elu function ensure slope larger one positive inputs. values alpha scale chosen mean variance inputs preserved two consecutive layers long weights initialized correctly (see initializer_lecun_normal()) number input units \"large enough\" (see reference paper information).","code":""},{"path":"https://keras3.posit.co/dev/reference/activation_selu.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Scaled Exponential Linear Unit (SELU). — activation_selu","text":"","code":"activation_selu(x)"},{"path":"https://keras3.posit.co/dev/reference/activation_selu.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Scaled Exponential Linear Unit (SELU). — activation_selu","text":"x Input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/activation_selu.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Scaled Exponential Linear Unit (SELU). — activation_selu","text":"tensor, result applying activation input tensor x.","code":""},{"path":"https://keras3.posit.co/dev/reference/activation_selu.html","id":"notes","dir":"Reference","previous_headings":"","what":"Notes","title":"Scaled Exponential Linear Unit (SELU). — activation_selu","text":"used together initializer_lecun_normal(). used together dropout variant layer_alpha_dropout() (legacy, depracated).","code":""},{"path":"https://keras3.posit.co/dev/reference/activation_selu.html","id":"reference","dir":"Reference","previous_headings":"","what":"Reference","title":"Scaled Exponential Linear Unit (SELU). — activation_selu","text":"Klambauer et al., 2017","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/activation_sigmoid.html","id":null,"dir":"Reference","previous_headings":"","what":"Sigmoid activation function. — activation_sigmoid","title":"Sigmoid activation function. — activation_sigmoid","text":"defined : sigmoid(x) = 1 / (1 + exp(-x)). small values (<-5), sigmoid returns value close zero, large values (>5) result function gets close 1. Sigmoid equivalent 2-element softmax, second element assumed zero. sigmoid function always returns value 0 1.","code":""},{"path":"https://keras3.posit.co/dev/reference/activation_sigmoid.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Sigmoid activation function. — activation_sigmoid","text":"","code":"activation_sigmoid(x)"},{"path":"https://keras3.posit.co/dev/reference/activation_sigmoid.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Sigmoid activation function. — activation_sigmoid","text":"x Input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/activation_sigmoid.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Sigmoid activation function. — activation_sigmoid","text":"tensor, result applying activation input tensor x.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/activation_silu.html","id":null,"dir":"Reference","previous_headings":"","what":"Swish (or Silu) activation function. — activation_silu","title":"Swish (or Silu) activation function. — activation_silu","text":"defined : swish(x) = x * sigmoid(x). Swish (Silu) activation function smooth, non-monotonic function unbounded bounded .","code":""},{"path":"https://keras3.posit.co/dev/reference/activation_silu.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Swish (or Silu) activation function. — activation_silu","text":"","code":"activation_silu(x)"},{"path":"https://keras3.posit.co/dev/reference/activation_silu.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Swish (or Silu) activation function. — activation_silu","text":"x Input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/activation_silu.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Swish (or Silu) activation function. — activation_silu","text":"tensor, result applying activation input tensor x.","code":""},{"path":"https://keras3.posit.co/dev/reference/activation_silu.html","id":"reference","dir":"Reference","previous_headings":"","what":"Reference","title":"Swish (or Silu) activation function. — activation_silu","text":"Ramachandran et al., 2017","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/activation_soft_shrink.html","id":null,"dir":"Reference","previous_headings":"","what":"Soft Shrink activation function. — activation_soft_shrink","title":"Soft Shrink activation function. — activation_soft_shrink","text":"defined : soft_shrink(x) = x - threshold x > threshold, soft_shrink(x) = x + threshold x < -threshold, soft_shrink(x) = 0 otherwise.","code":""},{"path":"https://keras3.posit.co/dev/reference/activation_soft_shrink.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Soft Shrink activation function. — activation_soft_shrink","text":"","code":"activation_soft_shrink(x, threshold = 0.5)"},{"path":"https://keras3.posit.co/dev/reference/activation_soft_shrink.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Soft Shrink activation function. — activation_soft_shrink","text":"x Input tensor. threshold Threshold value. Defaults 0.5.","code":""},{"path":"https://keras3.posit.co/dev/reference/activation_soft_shrink.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Soft Shrink activation function. — activation_soft_shrink","text":"tensor, result applying activation input tensor x.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/activation_softmax.html","id":null,"dir":"Reference","previous_headings":"","what":"Softmax converts a vector of values to a probability distribution. — activation_softmax","title":"Softmax converts a vector of values to a probability distribution. — activation_softmax","text":"elements output vector range [0, 1] sum 1. input vector handled independently. axis argument sets axis input function applied along. Softmax often used activation last layer classification network result interpreted probability distribution. softmax vector x computed exp(x) / sum(exp(x)). input values log-odds resulting probability.","code":""},{"path":"https://keras3.posit.co/dev/reference/activation_softmax.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Softmax converts a vector of values to a probability distribution. — activation_softmax","text":"","code":"activation_softmax(x, axis = -1L)"},{"path":"https://keras3.posit.co/dev/reference/activation_softmax.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Softmax converts a vector of values to a probability distribution. — activation_softmax","text":"x Input tensor. axis Integer, axis along softmax applied.","code":""},{"path":"https://keras3.posit.co/dev/reference/activation_softmax.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Softmax converts a vector of values to a probability distribution. — activation_softmax","text":"tensor, result applying activation input tensor x.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/activation_softplus.html","id":null,"dir":"Reference","previous_headings":"","what":"Softplus activation function. — activation_softplus","title":"Softplus activation function. — activation_softplus","text":"defined : softplus(x) = log(exp(x) + 1).","code":""},{"path":"https://keras3.posit.co/dev/reference/activation_softplus.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Softplus activation function. — activation_softplus","text":"","code":"activation_softplus(x)"},{"path":"https://keras3.posit.co/dev/reference/activation_softplus.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Softplus activation function. — activation_softplus","text":"x Input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/activation_softplus.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Softplus activation function. — activation_softplus","text":"tensor, result applying activation input tensor x.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/activation_softsign.html","id":null,"dir":"Reference","previous_headings":"","what":"Softsign activation function. — activation_softsign","title":"Softsign activation function. — activation_softsign","text":"Softsign defined : softsign(x) = x / (abs(x) + 1).","code":""},{"path":"https://keras3.posit.co/dev/reference/activation_softsign.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Softsign activation function. — activation_softsign","text":"","code":"activation_softsign(x)"},{"path":"https://keras3.posit.co/dev/reference/activation_softsign.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Softsign activation function. — activation_softsign","text":"x Input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/activation_softsign.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Softsign activation function. — activation_softsign","text":"tensor, result applying activation input tensor x.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/activation_sparse_plus.html","id":null,"dir":"Reference","previous_headings":"","what":"SparsePlus activation function. — activation_sparse_plus","title":"SparsePlus activation function. — activation_sparse_plus","text":"SparsePlus defined : sparse_plus(x) = 0 x <= -1. sparse_plus(x) = (1/4) * (x + 1)^2 -1 < x < 1. sparse_plus(x) = x x >= 1.","code":""},{"path":"https://keras3.posit.co/dev/reference/activation_sparse_plus.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"SparsePlus activation function. — activation_sparse_plus","text":"","code":"activation_sparse_plus(x)"},{"path":"https://keras3.posit.co/dev/reference/activation_sparse_plus.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"SparsePlus activation function. — activation_sparse_plus","text":"x Input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/activation_sparse_plus.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"SparsePlus activation function. — activation_sparse_plus","text":"tensor, result applying activation input tensor x.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/activation_sparse_sigmoid.html","id":null,"dir":"Reference","previous_headings":"","what":"Sparse sigmoid activation function. — activation_sparse_sigmoid","title":"Sparse sigmoid activation function. — activation_sparse_sigmoid","text":"defined f(x) = 0 x <= -1, f(x) = 0.5 * (x + 1) -1 < x < 1, f(x) = 1 x >= 1.","code":""},{"path":"https://keras3.posit.co/dev/reference/activation_sparse_sigmoid.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Sparse sigmoid activation function. — activation_sparse_sigmoid","text":"","code":"activation_sparse_sigmoid(x)"},{"path":"https://keras3.posit.co/dev/reference/activation_sparse_sigmoid.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Sparse sigmoid activation function. — activation_sparse_sigmoid","text":"x Input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/activation_sparse_sigmoid.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Sparse sigmoid activation function. — activation_sparse_sigmoid","text":"tensor, result applying activation input tensor x.","code":""},{"path":"https://keras3.posit.co/dev/reference/activation_sparse_sigmoid.html","id":"reference","dir":"Reference","previous_headings":"","what":"Reference","title":"Sparse sigmoid activation function. — activation_sparse_sigmoid","text":"M. Blondel, . F. T. Martins, V. Niculae, 2019","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/activation_sparsemax.html","id":null,"dir":"Reference","previous_headings":"","what":"Sparsemax activation function. — activation_sparsemax","title":"Sparsemax activation function. — activation_sparsemax","text":"batch , class j, sparsemax activation function defined : sparsemax(x)[, j] = max(x[, j] - (x[, :]), 0).","code":""},{"path":"https://keras3.posit.co/dev/reference/activation_sparsemax.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Sparsemax activation function. — activation_sparsemax","text":"","code":"activation_sparsemax(x, axis = -1L)"},{"path":"https://keras3.posit.co/dev/reference/activation_sparsemax.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Sparsemax activation function. — activation_sparsemax","text":"x Input tensor. axis int, axis along sparsemax operation applied. (1-based)","code":""},{"path":"https://keras3.posit.co/dev/reference/activation_sparsemax.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Sparsemax activation function. — activation_sparsemax","text":"tensor, output sparsemax transformation. type shape x.","code":""},{"path":"https://keras3.posit.co/dev/reference/activation_sparsemax.html","id":"reference","dir":"Reference","previous_headings":"","what":"Reference","title":"Sparsemax activation function. — activation_sparsemax","text":"Martins et.al., 2016","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/activation_squareplus.html","id":null,"dir":"Reference","previous_headings":"","what":"Squareplus activation function. — activation_squareplus","title":"Squareplus activation function. — activation_squareplus","text":"Squareplus activation function defined : f(x) = (x + sqrt(x^2 + b)) / 2 b smoothness parameter.","code":""},{"path":"https://keras3.posit.co/dev/reference/activation_squareplus.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Squareplus activation function. — activation_squareplus","text":"","code":"activation_squareplus(x, b = 4L)"},{"path":"https://keras3.posit.co/dev/reference/activation_squareplus.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Squareplus activation function. — activation_squareplus","text":"x Input tensor. b Smoothness parameter. Defaults 4.","code":""},{"path":"https://keras3.posit.co/dev/reference/activation_squareplus.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Squareplus activation function. — activation_squareplus","text":"tensor, result applying activation input tensor x.","code":""},{"path":"https://keras3.posit.co/dev/reference/activation_squareplus.html","id":"reference","dir":"Reference","previous_headings":"","what":"Reference","title":"Squareplus activation function. — activation_squareplus","text":"Ramachandran et al., 2021","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/activation_tanh.html","id":null,"dir":"Reference","previous_headings":"","what":"Hyperbolic tangent activation function. — activation_tanh","title":"Hyperbolic tangent activation function. — activation_tanh","text":"defined : tanh(x) = sinh(x) / cosh(x), .e. tanh(x) = ((exp(x) - exp(-x)) / (exp(x) + exp(-x))).","code":""},{"path":"https://keras3.posit.co/dev/reference/activation_tanh.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Hyperbolic tangent activation function. — activation_tanh","text":"","code":"activation_tanh(x)"},{"path":"https://keras3.posit.co/dev/reference/activation_tanh.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Hyperbolic tangent activation function. — activation_tanh","text":"x Input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/activation_tanh.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Hyperbolic tangent activation function. — activation_tanh","text":"tensor, result applying activation input tensor x.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/activation_tanh_shrink.html","id":null,"dir":"Reference","previous_headings":"","what":"Tanh shrink activation function. — activation_tanh_shrink","title":"Tanh shrink activation function. — activation_tanh_shrink","text":"defined : f(x) = x - tanh(x).","code":""},{"path":"https://keras3.posit.co/dev/reference/activation_tanh_shrink.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Tanh shrink activation function. — activation_tanh_shrink","text":"","code":"activation_tanh_shrink(x)"},{"path":"https://keras3.posit.co/dev/reference/activation_tanh_shrink.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Tanh shrink activation function. — activation_tanh_shrink","text":"x Input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/activation_tanh_shrink.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Tanh shrink activation function. — activation_tanh_shrink","text":"tensor, result applying activation input tensor x.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/activation_threshold.html","id":null,"dir":"Reference","previous_headings":"","what":"Threshold activation function. — activation_threshold","title":"Threshold activation function. — activation_threshold","text":"defined : threshold(x) = x x > threshold, threshold(x) = default_value otherwise.","code":""},{"path":"https://keras3.posit.co/dev/reference/activation_threshold.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Threshold activation function. — activation_threshold","text":"","code":"activation_threshold(x, threshold, default_value)"},{"path":"https://keras3.posit.co/dev/reference/activation_threshold.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Threshold activation function. — activation_threshold","text":"x Input tensor. threshold value decides retain replace x. default_value Value assign x <= threshold.","code":""},{"path":"https://keras3.posit.co/dev/reference/activation_threshold.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Threshold activation function. — activation_threshold","text":"tensor, result applying activation input tensor x.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/active_property.html","id":null,"dir":"Reference","previous_headings":"","what":"Create an active property class method — active_property","title":"Create an active property class method — active_property","text":"Create active property class method","code":""},{"path":"https://keras3.posit.co/dev/reference/active_property.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Create an active property class method — active_property","text":"","code":"active_property(fn)"},{"path":"https://keras3.posit.co/dev/reference/active_property.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Create an active property class method — active_property","text":"fn R function","code":""},{"path":"https://keras3.posit.co/dev/reference/active_property.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Create an active property class method — active_property","text":"fn, additional R attribute cause fn converted active property converted method custom subclass.","code":""},{"path":"https://keras3.posit.co/dev/reference/active_property.html","id":"example","dir":"Reference","previous_headings":"","what":"Example","title":"Create an active property class method — active_property","text":"","code":"layer_foo <- Model(\"Foo\", ...,   metrics = active_property(function() {     list(self$d_loss_metric,          self$g_loss_metric)   }))"},{"path":"https://keras3.posit.co/dev/reference/adapt.html","id":null,"dir":"Reference","previous_headings":"","what":"Fits the state of the preprocessing layer to the data being passed — adapt","title":"Fits the state of the preprocessing layer to the data being passed — adapt","text":"Fits state preprocessing layer data passed","code":""},{"path":"https://keras3.posit.co/dev/reference/adapt.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Fits the state of the preprocessing layer to the data being passed — adapt","text":"","code":"adapt(object, data, ..., batch_size = NULL, steps = NULL)"},{"path":"https://keras3.posit.co/dev/reference/adapt.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Fits the state of the preprocessing layer to the data being passed — adapt","text":"object Preprocessing layer object data data train . can passed either tf.data.Dataset R array. ... Used forwards backwards compatibility. Passed underlying method. batch_size Integer NULL. Number asamples per state update. unspecified, batch_size default 32. specify batch_size data form TF Dataset generator (since generate batches). steps Integer NULL. Total number steps (batches samples) training input tensors TensorFlow data tensors, default NULL equal number samples dataset divided batch size, 1 determined. x tf.data.Dataset, steps NULL, epoch run input dataset exhausted. passing infinitely repeating dataset, must specify steps argument. argument supported array inputs.","code":""},{"path":"https://keras3.posit.co/dev/reference/adapt.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Fits the state of the preprocessing layer to the data being passed — adapt","text":"Returns object, invisibly.","code":""},{"path":"https://keras3.posit.co/dev/reference/adapt.html","id":"details","dir":"Reference","previous_headings":"","what":"Details","title":"Fits the state of the preprocessing layer to the data being passed — adapt","text":"calling adapt layer, preprocessing layer's state update training. order make preprocessing layers efficient distribution context, kept constant respect compiled tf.Graphs call layer. affect layer use adapting layer , adapt layer multiple times need take care re-compile compiled functions follows: adding preprocessing layer keras model, need call compile(model) subsequent call adapt(). calling preprocessing layer inside tfdatasets::dataset_map(), call dataset_map() input Dataset adapt(). using tensorflow::tf_function() directly calls preprocessing layer, need call tf_function() callable subsequent call adapt(). keras_model() example multiple adapts:         tfdatasets example multiple adapts:","code":"layer <- layer_normalization(axis = NULL) adapt(layer, c(0, 2)) model <- keras_model_sequential() |> layer() predict(model, c(0, 1, 2), verbose = FALSE) # [1] -1  0  1 ## [1] -1  0  1 adapt(layer, c(-1, 1)) compile(model)  # This is needed to re-compile model.predict! predict(model, c(0, 1, 2), verbose = FALSE) # [1] 0 1 2 ## [1] 0 1 2 layer <- layer_normalization(axis = NULL) adapt(layer, c(0, 2)) input_ds <- tfdatasets::range_dataset(0, 3) normalized_ds <- input_ds |>   tfdatasets::dataset_map(layer) str(tfdatasets::iterate(normalized_ds)) ## List of 3 ##  $ :<tf.Tensor: shape=(1), dtype=float32, numpy=array([-1.], dtype=float32)> ##  $ :<tf.Tensor: shape=(1), dtype=float32, numpy=array([0.], dtype=float32)> ##  $ :<tf.Tensor: shape=(1), dtype=float32, numpy=array([1.], dtype=float32)> adapt(layer, c(-1, 1)) normalized_ds <- input_ds |>   tfdatasets::dataset_map(layer) # Re-map over the input dataset.  normalized_ds |>   tfdatasets::as_array_iterator() |>   tfdatasets::iterate(simplify = FALSE) |>   str() ## List of 3 ##  $ : num [1(1d)] 0 ##  $ : num [1(1d)] 1 ##  $ : num [1(1d)] 2"},{"path":"https://keras3.posit.co/dev/reference/application_convnext_base.html","id":null,"dir":"Reference","previous_headings":"","what":"Instantiates the ConvNeXtBase architecture. — application_convnext_base","title":"Instantiates the ConvNeXtBase architecture. — application_convnext_base","text":"Instantiates ConvNeXtBase architecture.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_convnext_base.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Instantiates the ConvNeXtBase architecture. — application_convnext_base","text":"","code":"application_convnext_base(   include_top = TRUE,   include_preprocessing = TRUE,   weights = \"imagenet\",   input_tensor = NULL,   input_shape = NULL,   pooling = NULL,   classes = 1000L,   classifier_activation = \"softmax\",   name = \"convnext_base\" )"},{"path":"https://keras3.posit.co/dev/reference/application_convnext_base.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Instantiates the ConvNeXtBase architecture. — application_convnext_base","text":"include_top Whether include fully-connected layer top network. Defaults TRUE. include_preprocessing Boolean, whether include preprocessing layer bottom network. weights One NULL (random initialization), \"imagenet\" (pre-training ImageNet-1k), path weights file loaded. Defaults \"imagenet\". input_tensor Optional Keras tensor (.e. output keras_input()) use image input model. input_shape Optional shape tuple, specified include_top FALSE. exactly 3 inputs channels. pooling Optional pooling mode feature extraction include_top FALSE. Defaults NULL. NULL means output model 4D tensor output last convolutional layer. avg means global average pooling applied output last convolutional layer, thus output model 2D tensor. max means global max pooling applied. classes Optional number classes classify images , specified include_top TRUE, weights argument specified. Defaults 1000 (number ImageNet classes). classifier_activation str callable. activation function use \"top\" layer. Ignored unless include_top=TRUE. Set classifier_activation=NULL return logits \"top\" layer. Defaults \"softmax\". loading pretrained weights, classifier_activation can NULL \"softmax\". name name model (string).","code":""},{"path":"https://keras3.posit.co/dev/reference/application_convnext_base.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Instantiates the ConvNeXtBase architecture. — application_convnext_base","text":"model instance.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_convnext_base.html","id":"references","dir":"Reference","previous_headings":"","what":"References","title":"Instantiates the ConvNeXtBase architecture. — application_convnext_base","text":"ConvNet 2020s (CVPR 2022) image classification use cases, see page detailed examples. transfer learning use cases, make sure read guide transfer learning & fine-tuning. base, large, xlarge models first pre-trained ImageNet-21k dataset fine-tuned ImageNet-1k dataset. pre-trained parameters models assembled official repository. get sense parameters converted Keras compatible parameters, please refer repository.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_convnext_base.html","id":"note","dir":"Reference","previous_headings":"","what":"Note","title":"Instantiates the ConvNeXtBase architecture. — application_convnext_base","text":"Keras Application expects specific kind input preprocessing. ConvNeXt, preprocessing included model using Normalization layer.  ConvNeXt models expect inputs float uint8 tensors pixels values [0-255] range. calling summary() method instantiating ConvNeXt model, prefer setting expand_nested argument summary() TRUE better investigate instantiated model.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/application_convnext_large.html","id":null,"dir":"Reference","previous_headings":"","what":"Instantiates the ConvNeXtLarge architecture. — application_convnext_large","title":"Instantiates the ConvNeXtLarge architecture. — application_convnext_large","text":"Instantiates ConvNeXtLarge architecture.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_convnext_large.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Instantiates the ConvNeXtLarge architecture. — application_convnext_large","text":"","code":"application_convnext_large(   include_top = TRUE,   include_preprocessing = TRUE,   weights = \"imagenet\",   input_tensor = NULL,   input_shape = NULL,   pooling = NULL,   classes = 1000L,   classifier_activation = \"softmax\",   name = \"convnext_large\" )"},{"path":"https://keras3.posit.co/dev/reference/application_convnext_large.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Instantiates the ConvNeXtLarge architecture. — application_convnext_large","text":"include_top Whether include fully-connected layer top network. Defaults TRUE. include_preprocessing Boolean, whether include preprocessing layer bottom network. weights One NULL (random initialization), \"imagenet\" (pre-training ImageNet-1k), path weights file loaded. Defaults \"imagenet\". input_tensor Optional Keras tensor (.e. output keras_input()) use image input model. input_shape Optional shape tuple, specified include_top FALSE. exactly 3 inputs channels. pooling Optional pooling mode feature extraction include_top FALSE. Defaults NULL. NULL means output model 4D tensor output last convolutional layer. avg means global average pooling applied output last convolutional layer, thus output model 2D tensor. max means global max pooling applied. classes Optional number classes classify images , specified include_top TRUE, weights argument specified. Defaults 1000 (number ImageNet classes). classifier_activation str callable. activation function use \"top\" layer. Ignored unless include_top=TRUE. Set classifier_activation=NULL return logits \"top\" layer. Defaults \"softmax\". loading pretrained weights, classifier_activation can NULL \"softmax\". name name model (string).","code":""},{"path":"https://keras3.posit.co/dev/reference/application_convnext_large.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Instantiates the ConvNeXtLarge architecture. — application_convnext_large","text":"model instance.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_convnext_large.html","id":"references","dir":"Reference","previous_headings":"","what":"References","title":"Instantiates the ConvNeXtLarge architecture. — application_convnext_large","text":"ConvNet 2020s (CVPR 2022) image classification use cases, see page detailed examples. transfer learning use cases, make sure read guide transfer learning & fine-tuning. base, large, xlarge models first pre-trained ImageNet-21k dataset fine-tuned ImageNet-1k dataset. pre-trained parameters models assembled official repository. get sense parameters converted Keras compatible parameters, please refer repository.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_convnext_large.html","id":"note","dir":"Reference","previous_headings":"","what":"Note","title":"Instantiates the ConvNeXtLarge architecture. — application_convnext_large","text":"Keras Application expects specific kind input preprocessing. ConvNeXt, preprocessing included model using Normalization layer.  ConvNeXt models expect inputs float uint8 tensors pixels values [0-255] range. calling summary() method instantiating ConvNeXt model, prefer setting expand_nested argument summary() TRUE better investigate instantiated model.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/application_convnext_small.html","id":null,"dir":"Reference","previous_headings":"","what":"Instantiates the ConvNeXtSmall architecture. — application_convnext_small","title":"Instantiates the ConvNeXtSmall architecture. — application_convnext_small","text":"Instantiates ConvNeXtSmall architecture.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_convnext_small.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Instantiates the ConvNeXtSmall architecture. — application_convnext_small","text":"","code":"application_convnext_small(   include_top = TRUE,   include_preprocessing = TRUE,   weights = \"imagenet\",   input_tensor = NULL,   input_shape = NULL,   pooling = NULL,   classes = 1000L,   classifier_activation = \"softmax\",   name = \"convnext_small\" )"},{"path":"https://keras3.posit.co/dev/reference/application_convnext_small.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Instantiates the ConvNeXtSmall architecture. — application_convnext_small","text":"include_top Whether include fully-connected layer top network. Defaults TRUE. include_preprocessing Boolean, whether include preprocessing layer bottom network. weights One NULL (random initialization), \"imagenet\" (pre-training ImageNet-1k), path weights file loaded. Defaults \"imagenet\". input_tensor Optional Keras tensor (.e. output keras_input()) use image input model. input_shape Optional shape tuple, specified include_top FALSE. exactly 3 inputs channels. pooling Optional pooling mode feature extraction include_top FALSE. Defaults NULL. NULL means output model 4D tensor output last convolutional layer. avg means global average pooling applied output last convolutional layer, thus output model 2D tensor. max means global max pooling applied. classes Optional number classes classify images , specified include_top TRUE, weights argument specified. Defaults 1000 (number ImageNet classes). classifier_activation str callable. activation function use \"top\" layer. Ignored unless include_top=TRUE. Set classifier_activation=NULL return logits \"top\" layer. Defaults \"softmax\". loading pretrained weights, classifier_activation can NULL \"softmax\". name name model (string).","code":""},{"path":"https://keras3.posit.co/dev/reference/application_convnext_small.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Instantiates the ConvNeXtSmall architecture. — application_convnext_small","text":"model instance.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_convnext_small.html","id":"references","dir":"Reference","previous_headings":"","what":"References","title":"Instantiates the ConvNeXtSmall architecture. — application_convnext_small","text":"ConvNet 2020s (CVPR 2022) image classification use cases, see page detailed examples. transfer learning use cases, make sure read guide transfer learning & fine-tuning. base, large, xlarge models first pre-trained ImageNet-21k dataset fine-tuned ImageNet-1k dataset. pre-trained parameters models assembled official repository. get sense parameters converted Keras compatible parameters, please refer repository.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_convnext_small.html","id":"note","dir":"Reference","previous_headings":"","what":"Note","title":"Instantiates the ConvNeXtSmall architecture. — application_convnext_small","text":"Keras Application expects specific kind input preprocessing. ConvNeXt, preprocessing included model using Normalization layer.  ConvNeXt models expect inputs float uint8 tensors pixels values [0-255] range. calling summary() method instantiating ConvNeXt model, prefer setting expand_nested argument summary() TRUE better investigate instantiated model.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/application_convnext_tiny.html","id":null,"dir":"Reference","previous_headings":"","what":"Instantiates the ConvNeXtTiny architecture. — application_convnext_tiny","title":"Instantiates the ConvNeXtTiny architecture. — application_convnext_tiny","text":"Instantiates ConvNeXtTiny architecture.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_convnext_tiny.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Instantiates the ConvNeXtTiny architecture. — application_convnext_tiny","text":"","code":"application_convnext_tiny(   include_top = TRUE,   include_preprocessing = TRUE,   weights = \"imagenet\",   input_tensor = NULL,   input_shape = NULL,   pooling = NULL,   classes = 1000L,   classifier_activation = \"softmax\",   name = \"convnext_tiny\" )"},{"path":"https://keras3.posit.co/dev/reference/application_convnext_tiny.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Instantiates the ConvNeXtTiny architecture. — application_convnext_tiny","text":"include_top Whether include fully-connected layer top network. Defaults TRUE. include_preprocessing Boolean, whether include preprocessing layer bottom network. weights One NULL (random initialization), \"imagenet\" (pre-training ImageNet-1k), path weights file loaded. Defaults \"imagenet\". input_tensor Optional Keras tensor (.e. output keras_input()) use image input model. input_shape Optional shape tuple, specified include_top FALSE. exactly 3 inputs channels. pooling Optional pooling mode feature extraction include_top FALSE. Defaults NULL. NULL means output model 4D tensor output last convolutional layer. avg means global average pooling applied output last convolutional layer, thus output model 2D tensor. max means global max pooling applied. classes Optional number classes classify images , specified include_top TRUE, weights argument specified. Defaults 1000 (number ImageNet classes). classifier_activation str callable. activation function use \"top\" layer. Ignored unless include_top=TRUE. Set classifier_activation=NULL return logits \"top\" layer. Defaults \"softmax\". loading pretrained weights, classifier_activation can NULL \"softmax\". name name model (string).","code":""},{"path":"https://keras3.posit.co/dev/reference/application_convnext_tiny.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Instantiates the ConvNeXtTiny architecture. — application_convnext_tiny","text":"model instance.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_convnext_tiny.html","id":"references","dir":"Reference","previous_headings":"","what":"References","title":"Instantiates the ConvNeXtTiny architecture. — application_convnext_tiny","text":"ConvNet 2020s (CVPR 2022) image classification use cases, see page detailed examples. transfer learning use cases, make sure read guide transfer learning & fine-tuning. base, large, xlarge models first pre-trained ImageNet-21k dataset fine-tuned ImageNet-1k dataset. pre-trained parameters models assembled official repository. get sense parameters converted Keras compatible parameters, please refer repository.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_convnext_tiny.html","id":"note","dir":"Reference","previous_headings":"","what":"Note","title":"Instantiates the ConvNeXtTiny architecture. — application_convnext_tiny","text":"Keras Application expects specific kind input preprocessing. ConvNeXt, preprocessing included model using Normalization layer.  ConvNeXt models expect inputs float uint8 tensors pixels values [0-255] range. calling summary() method instantiating ConvNeXt model, prefer setting expand_nested argument summary() TRUE better investigate instantiated model.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/application_convnext_xlarge.html","id":null,"dir":"Reference","previous_headings":"","what":"Instantiates the ConvNeXtXLarge architecture. — application_convnext_xlarge","title":"Instantiates the ConvNeXtXLarge architecture. — application_convnext_xlarge","text":"Instantiates ConvNeXtXLarge architecture.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_convnext_xlarge.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Instantiates the ConvNeXtXLarge architecture. — application_convnext_xlarge","text":"","code":"application_convnext_xlarge(   include_top = TRUE,   include_preprocessing = TRUE,   weights = \"imagenet\",   input_tensor = NULL,   input_shape = NULL,   pooling = NULL,   classes = 1000L,   classifier_activation = \"softmax\",   name = \"convnext_xlarge\" )"},{"path":"https://keras3.posit.co/dev/reference/application_convnext_xlarge.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Instantiates the ConvNeXtXLarge architecture. — application_convnext_xlarge","text":"include_top Whether include fully-connected layer top network. Defaults TRUE. include_preprocessing Boolean, whether include preprocessing layer bottom network. weights One NULL (random initialization), \"imagenet\" (pre-training ImageNet-1k), path weights file loaded. Defaults \"imagenet\". input_tensor Optional Keras tensor (.e. output keras_input()) use image input model. input_shape Optional shape tuple, specified include_top FALSE. exactly 3 inputs channels. pooling Optional pooling mode feature extraction include_top FALSE. Defaults NULL. NULL means output model 4D tensor output last convolutional layer. avg means global average pooling applied output last convolutional layer, thus output model 2D tensor. max means global max pooling applied. classes Optional number classes classify images , specified include_top TRUE, weights argument specified. Defaults 1000 (number ImageNet classes). classifier_activation str callable. activation function use \"top\" layer. Ignored unless include_top=TRUE. Set classifier_activation=NULL return logits \"top\" layer. Defaults \"softmax\". loading pretrained weights, classifier_activation can NULL \"softmax\". name name model (string).","code":""},{"path":"https://keras3.posit.co/dev/reference/application_convnext_xlarge.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Instantiates the ConvNeXtXLarge architecture. — application_convnext_xlarge","text":"model instance.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_convnext_xlarge.html","id":"references","dir":"Reference","previous_headings":"","what":"References","title":"Instantiates the ConvNeXtXLarge architecture. — application_convnext_xlarge","text":"ConvNet 2020s (CVPR 2022) image classification use cases, see page detailed examples. transfer learning use cases, make sure read guide transfer learning & fine-tuning. base, large, xlarge models first pre-trained ImageNet-21k dataset fine-tuned ImageNet-1k dataset. pre-trained parameters models assembled official repository. get sense parameters converted Keras compatible parameters, please refer repository.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_convnext_xlarge.html","id":"note","dir":"Reference","previous_headings":"","what":"Note","title":"Instantiates the ConvNeXtXLarge architecture. — application_convnext_xlarge","text":"Keras Application expects specific kind input preprocessing. ConvNeXt, preprocessing included model using Normalization layer.  ConvNeXt models expect inputs float uint8 tensors pixels values [0-255] range. calling summary() method instantiating ConvNeXt model, prefer setting expand_nested argument summary() TRUE better investigate instantiated model.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/application_densenet121.html","id":null,"dir":"Reference","previous_headings":"","what":"Instantiates the Densenet121 architecture. — application_densenet121","title":"Instantiates the Densenet121 architecture. — application_densenet121","text":"Instantiates Densenet121 architecture.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_densenet121.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Instantiates the Densenet121 architecture. — application_densenet121","text":"","code":"application_densenet121(   include_top = TRUE,   weights = \"imagenet\",   input_tensor = NULL,   input_shape = NULL,   pooling = NULL,   classes = 1000L,   classifier_activation = \"softmax\",   name = \"densenet121\" )"},{"path":"https://keras3.posit.co/dev/reference/application_densenet121.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Instantiates the Densenet121 architecture. — application_densenet121","text":"include_top whether include fully-connected layer top network. weights one NULL (random initialization), \"imagenet\" (pre-training ImageNet), path weights file loaded. input_tensor optional Keras tensor (.e. output keras_input()) use image input model. input_shape optional shape tuple, specified include_top FALSE (otherwise input shape (224, 224, 3) ('channels_last' data format) (3, 224, 224) ('channels_first' data format). exactly 3 inputs channels, width height smaller 32. E.g. (200, 200, 3) one valid value. pooling Optional pooling mode feature extraction include_top FALSE. NULL means output model 4D tensor output last convolutional block. avg means global average pooling applied output last convolutional block, thus output model 2D tensor. max means global max pooling applied. classes optional number classes classify images , specified include_top TRUE, weights argument specified. Defaults 1000. classifier_activation str callable. activation function use \"top\" layer. Ignored unless include_top=TRUE. Set classifier_activation=NULL return logits \"top\" layer. loading pretrained weights, classifier_activation can NULL \"softmax\". name name model (string).","code":""},{"path":"https://keras3.posit.co/dev/reference/application_densenet121.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Instantiates the Densenet121 architecture. — application_densenet121","text":"Keras model instance.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_densenet121.html","id":"reference","dir":"Reference","previous_headings":"","what":"Reference","title":"Instantiates the Densenet121 architecture. — application_densenet121","text":"Densely Connected Convolutional Networks (CVPR 2017) Optionally loads weights pre-trained ImageNet. Note data format convention used model one specified Keras config ~/.keras/keras.json.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_densenet121.html","id":"note","dir":"Reference","previous_headings":"","what":"Note","title":"Instantiates the Densenet121 architecture. — application_densenet121","text":"Keras Application expects specific kind input preprocessing. DenseNet, call application_preprocess_inputs() inputs passing model.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/application_densenet169.html","id":null,"dir":"Reference","previous_headings":"","what":"Instantiates the Densenet169 architecture. — application_densenet169","title":"Instantiates the Densenet169 architecture. — application_densenet169","text":"Instantiates Densenet169 architecture.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_densenet169.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Instantiates the Densenet169 architecture. — application_densenet169","text":"","code":"application_densenet169(   include_top = TRUE,   weights = \"imagenet\",   input_tensor = NULL,   input_shape = NULL,   pooling = NULL,   classes = 1000L,   classifier_activation = \"softmax\",   name = \"densenet169\" )"},{"path":"https://keras3.posit.co/dev/reference/application_densenet169.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Instantiates the Densenet169 architecture. — application_densenet169","text":"include_top whether include fully-connected layer top network. weights one NULL (random initialization), \"imagenet\" (pre-training ImageNet), path weights file loaded. input_tensor optional Keras tensor (.e. output keras_input()) use image input model. input_shape optional shape tuple, specified include_top FALSE (otherwise input shape (224, 224, 3) ('channels_last' data format) (3, 224, 224) ('channels_first' data format). exactly 3 inputs channels, width height smaller 32. E.g. (200, 200, 3) one valid value. pooling Optional pooling mode feature extraction include_top FALSE. NULL means output model 4D tensor output last convolutional block. avg means global average pooling applied output last convolutional block, thus output model 2D tensor. max means global max pooling applied. classes optional number classes classify images , specified include_top TRUE, weights argument specified. Defaults 1000. classifier_activation str callable. activation function use \"top\" layer. Ignored unless include_top=TRUE. Set classifier_activation=NULL return logits \"top\" layer. loading pretrained weights, classifier_activation can NULL \"softmax\". name name model (string).","code":""},{"path":"https://keras3.posit.co/dev/reference/application_densenet169.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Instantiates the Densenet169 architecture. — application_densenet169","text":"Keras model instance.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_densenet169.html","id":"reference","dir":"Reference","previous_headings":"","what":"Reference","title":"Instantiates the Densenet169 architecture. — application_densenet169","text":"Densely Connected Convolutional Networks (CVPR 2017) Optionally loads weights pre-trained ImageNet. Note data format convention used model one specified Keras config ~/.keras/keras.json.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_densenet169.html","id":"note","dir":"Reference","previous_headings":"","what":"Note","title":"Instantiates the Densenet169 architecture. — application_densenet169","text":"Keras Application expects specific kind input preprocessing. DenseNet, call application_preprocess_inputs() inputs passing model.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/application_densenet201.html","id":null,"dir":"Reference","previous_headings":"","what":"Instantiates the Densenet201 architecture. — application_densenet201","title":"Instantiates the Densenet201 architecture. — application_densenet201","text":"Instantiates Densenet201 architecture.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_densenet201.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Instantiates the Densenet201 architecture. — application_densenet201","text":"","code":"application_densenet201(   include_top = TRUE,   weights = \"imagenet\",   input_tensor = NULL,   input_shape = NULL,   pooling = NULL,   classes = 1000L,   classifier_activation = \"softmax\",   name = \"densenet201\" )"},{"path":"https://keras3.posit.co/dev/reference/application_densenet201.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Instantiates the Densenet201 architecture. — application_densenet201","text":"include_top whether include fully-connected layer top network. weights one NULL (random initialization), \"imagenet\" (pre-training ImageNet), path weights file loaded. input_tensor optional Keras tensor (.e. output keras_input()) use image input model. input_shape optional shape tuple, specified include_top FALSE (otherwise input shape (224, 224, 3) ('channels_last' data format) (3, 224, 224) ('channels_first' data format). exactly 3 inputs channels, width height smaller 32. E.g. (200, 200, 3) one valid value. pooling Optional pooling mode feature extraction include_top FALSE. NULL means output model 4D tensor output last convolutional block. avg means global average pooling applied output last convolutional block, thus output model 2D tensor. max means global max pooling applied. classes optional number classes classify images , specified include_top TRUE, weights argument specified. Defaults 1000. classifier_activation str callable. activation function use \"top\" layer. Ignored unless include_top=TRUE. Set classifier_activation=NULL return logits \"top\" layer. loading pretrained weights, classifier_activation can NULL \"softmax\". name name model (string).","code":""},{"path":"https://keras3.posit.co/dev/reference/application_densenet201.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Instantiates the Densenet201 architecture. — application_densenet201","text":"Keras model instance.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_densenet201.html","id":"reference","dir":"Reference","previous_headings":"","what":"Reference","title":"Instantiates the Densenet201 architecture. — application_densenet201","text":"Densely Connected Convolutional Networks (CVPR 2017) Optionally loads weights pre-trained ImageNet. Note data format convention used model one specified Keras config ~/.keras/keras.json.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_densenet201.html","id":"note","dir":"Reference","previous_headings":"","what":"Note","title":"Instantiates the Densenet201 architecture. — application_densenet201","text":"Keras Application expects specific kind input preprocessing. DenseNet, call application_preprocess_inputs() inputs passing model.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_b0.html","id":null,"dir":"Reference","previous_headings":"","what":"Instantiates the EfficientNetB0 architecture. — application_efficientnet_b0","title":"Instantiates the EfficientNetB0 architecture. — application_efficientnet_b0","text":"Instantiates EfficientNetB0 architecture.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_b0.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Instantiates the EfficientNetB0 architecture. — application_efficientnet_b0","text":"","code":"application_efficientnet_b0(   include_top = TRUE,   weights = \"imagenet\",   input_tensor = NULL,   input_shape = NULL,   pooling = NULL,   classes = 1000L,   classifier_activation = \"softmax\",   name = \"efficientnetb0\",   ... )"},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_b0.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Instantiates the EfficientNetB0 architecture. — application_efficientnet_b0","text":"include_top Whether include fully-connected layer top network. Defaults TRUE. weights One NULL (random initialization), \"imagenet\" (pre-training ImageNet), path weights file loaded. Defaults \"imagenet\". input_tensor Optional Keras tensor (.e. output keras_input()) use image input model. input_shape Optional shape tuple, specified include_top FALSE. exactly 3 inputs channels. pooling Optional pooling mode feature extraction include_top FALSE. Defaults NULL. NULL means output model 4D tensor output last convolutional layer. avg means global average pooling applied output last convolutional layer, thus output model 2D tensor. max means global max pooling applied. classes Optional number classes classify images , specified include_top TRUE, weights argument specified. 1000 many ImageNet classes . Defaults 1000. classifier_activation str callable. activation function use \"top\" layer. Ignored unless include_top=TRUE. Set classifier_activation=NULL return logits \"top\" layer. Defaults 'softmax'. loading pretrained weights, classifier_activation can NULL \"softmax\". name name model (string). ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_b0.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Instantiates the EfficientNetB0 architecture. — application_efficientnet_b0","text":"model instance.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_b0.html","id":"reference","dir":"Reference","previous_headings":"","what":"Reference","title":"Instantiates the EfficientNetB0 architecture. — application_efficientnet_b0","text":"EfficientNet: Rethinking Model Scaling Convolutional Neural Networks (ICML 2019) function returns Keras image classification model, optionally loaded weights pre-trained ImageNet. image classification use cases, see page detailed examples. transfer learning use cases, make sure read guide transfer learning & fine-tuning.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_b0.html","id":"note","dir":"Reference","previous_headings":"","what":"Note","title":"Instantiates the EfficientNetB0 architecture. — application_efficientnet_b0","text":"Keras Application expects specific kind input preprocessing. EfficientNet, input preprocessing included part model (Rescaling layer), thus application_preprocess_inputs() actually pass-function. EfficientNet models expect inputs float tensors pixels values [0-255] range.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_b1.html","id":null,"dir":"Reference","previous_headings":"","what":"Instantiates the EfficientNetB1 architecture. — application_efficientnet_b1","title":"Instantiates the EfficientNetB1 architecture. — application_efficientnet_b1","text":"Instantiates EfficientNetB1 architecture.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_b1.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Instantiates the EfficientNetB1 architecture. — application_efficientnet_b1","text":"","code":"application_efficientnet_b1(   include_top = TRUE,   weights = \"imagenet\",   input_tensor = NULL,   input_shape = NULL,   pooling = NULL,   classes = 1000L,   classifier_activation = \"softmax\",   name = \"efficientnetb1\",   ... )"},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_b1.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Instantiates the EfficientNetB1 architecture. — application_efficientnet_b1","text":"include_top Whether include fully-connected layer top network. Defaults TRUE. weights One NULL (random initialization), \"imagenet\" (pre-training ImageNet), path weights file loaded. Defaults \"imagenet\". input_tensor Optional Keras tensor (.e. output keras_input()) use image input model. input_shape Optional shape tuple, specified include_top FALSE. exactly 3 inputs channels. pooling Optional pooling mode feature extraction include_top FALSE. Defaults NULL. NULL means output model 4D tensor output last convolutional layer. avg means global average pooling applied output last convolutional layer, thus output model 2D tensor. max means global max pooling applied. classes Optional number classes classify images , specified include_top TRUE, weights argument specified. 1000 many ImageNet classes . Defaults 1000. classifier_activation str callable. activation function use \"top\" layer. Ignored unless include_top=TRUE. Set classifier_activation=NULL return logits \"top\" layer. Defaults 'softmax'. loading pretrained weights, classifier_activation can NULL \"softmax\". name name model (string). ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_b1.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Instantiates the EfficientNetB1 architecture. — application_efficientnet_b1","text":"model instance.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_b1.html","id":"reference","dir":"Reference","previous_headings":"","what":"Reference","title":"Instantiates the EfficientNetB1 architecture. — application_efficientnet_b1","text":"EfficientNet: Rethinking Model Scaling Convolutional Neural Networks (ICML 2019) function returns Keras image classification model, optionally loaded weights pre-trained ImageNet. image classification use cases, see page detailed examples. transfer learning use cases, make sure read guide transfer learning & fine-tuning.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_b1.html","id":"note","dir":"Reference","previous_headings":"","what":"Note","title":"Instantiates the EfficientNetB1 architecture. — application_efficientnet_b1","text":"Keras Application expects specific kind input preprocessing. EfficientNet, input preprocessing included part model (Rescaling layer), thus application_preprocess_inputs() actually pass-function. EfficientNet models expect inputs float tensors pixels values [0-255] range.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_b2.html","id":null,"dir":"Reference","previous_headings":"","what":"Instantiates the EfficientNetB2 architecture. — application_efficientnet_b2","title":"Instantiates the EfficientNetB2 architecture. — application_efficientnet_b2","text":"Instantiates EfficientNetB2 architecture.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_b2.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Instantiates the EfficientNetB2 architecture. — application_efficientnet_b2","text":"","code":"application_efficientnet_b2(   include_top = TRUE,   weights = \"imagenet\",   input_tensor = NULL,   input_shape = NULL,   pooling = NULL,   classes = 1000L,   classifier_activation = \"softmax\",   name = \"efficientnetb2\",   ... )"},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_b2.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Instantiates the EfficientNetB2 architecture. — application_efficientnet_b2","text":"include_top Whether include fully-connected layer top network. Defaults TRUE. weights One NULL (random initialization), \"imagenet\" (pre-training ImageNet), path weights file loaded. Defaults \"imagenet\". input_tensor Optional Keras tensor (.e. output keras_input()) use image input model. input_shape Optional shape tuple, specified include_top FALSE. exactly 3 inputs channels. pooling Optional pooling mode feature extraction include_top FALSE. Defaults NULL. NULL means output model 4D tensor output last convolutional layer. avg means global average pooling applied output last convolutional layer, thus output model 2D tensor. max means global max pooling applied. classes Optional number classes classify images , specified include_top TRUE, weights argument specified. 1000 many ImageNet classes . Defaults 1000. classifier_activation str callable. activation function use \"top\" layer. Ignored unless include_top=TRUE. Set classifier_activation=NULL return logits \"top\" layer. Defaults 'softmax'. loading pretrained weights, classifier_activation can NULL \"softmax\". name name model (string). ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_b2.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Instantiates the EfficientNetB2 architecture. — application_efficientnet_b2","text":"model instance.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_b2.html","id":"reference","dir":"Reference","previous_headings":"","what":"Reference","title":"Instantiates the EfficientNetB2 architecture. — application_efficientnet_b2","text":"EfficientNet: Rethinking Model Scaling Convolutional Neural Networks (ICML 2019) function returns Keras image classification model, optionally loaded weights pre-trained ImageNet. image classification use cases, see page detailed examples. transfer learning use cases, make sure read guide transfer learning & fine-tuning.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_b2.html","id":"note","dir":"Reference","previous_headings":"","what":"Note","title":"Instantiates the EfficientNetB2 architecture. — application_efficientnet_b2","text":"Keras Application expects specific kind input preprocessing. EfficientNet, input preprocessing included part model (Rescaling layer), thus application_preprocess_inputs() actually pass-function. EfficientNet models expect inputs float tensors pixels values [0-255] range.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_b3.html","id":null,"dir":"Reference","previous_headings":"","what":"Instantiates the EfficientNetB3 architecture. — application_efficientnet_b3","title":"Instantiates the EfficientNetB3 architecture. — application_efficientnet_b3","text":"Instantiates EfficientNetB3 architecture.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_b3.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Instantiates the EfficientNetB3 architecture. — application_efficientnet_b3","text":"","code":"application_efficientnet_b3(   include_top = TRUE,   weights = \"imagenet\",   input_tensor = NULL,   input_shape = NULL,   pooling = NULL,   classes = 1000L,   classifier_activation = \"softmax\",   name = \"efficientnetb3\",   ... )"},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_b3.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Instantiates the EfficientNetB3 architecture. — application_efficientnet_b3","text":"include_top Whether include fully-connected layer top network. Defaults TRUE. weights One NULL (random initialization), \"imagenet\" (pre-training ImageNet), path weights file loaded. Defaults \"imagenet\". input_tensor Optional Keras tensor (.e. output keras_input()) use image input model. input_shape Optional shape tuple, specified include_top FALSE. exactly 3 inputs channels. pooling Optional pooling mode feature extraction include_top FALSE. Defaults NULL. NULL means output model 4D tensor output last convolutional layer. avg means global average pooling applied output last convolutional layer, thus output model 2D tensor. max means global max pooling applied. classes Optional number classes classify images , specified include_top TRUE, weights argument specified. 1000 many ImageNet classes . Defaults 1000. classifier_activation str callable. activation function use \"top\" layer. Ignored unless include_top=TRUE. Set classifier_activation=NULL return logits \"top\" layer. Defaults 'softmax'. loading pretrained weights, classifier_activation can NULL \"softmax\". name name model (string). ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_b3.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Instantiates the EfficientNetB3 architecture. — application_efficientnet_b3","text":"model instance.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_b3.html","id":"reference","dir":"Reference","previous_headings":"","what":"Reference","title":"Instantiates the EfficientNetB3 architecture. — application_efficientnet_b3","text":"EfficientNet: Rethinking Model Scaling Convolutional Neural Networks (ICML 2019) function returns Keras image classification model, optionally loaded weights pre-trained ImageNet. image classification use cases, see page detailed examples. transfer learning use cases, make sure read guide transfer learning & fine-tuning.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_b3.html","id":"note","dir":"Reference","previous_headings":"","what":"Note","title":"Instantiates the EfficientNetB3 architecture. — application_efficientnet_b3","text":"Keras Application expects specific kind input preprocessing. EfficientNet, input preprocessing included part model (Rescaling layer), thus application_preprocess_inputs() actually pass-function. EfficientNet models expect inputs float tensors pixels values [0-255] range.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_b4.html","id":null,"dir":"Reference","previous_headings":"","what":"Instantiates the EfficientNetB4 architecture. — application_efficientnet_b4","title":"Instantiates the EfficientNetB4 architecture. — application_efficientnet_b4","text":"Instantiates EfficientNetB4 architecture.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_b4.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Instantiates the EfficientNetB4 architecture. — application_efficientnet_b4","text":"","code":"application_efficientnet_b4(   include_top = TRUE,   weights = \"imagenet\",   input_tensor = NULL,   input_shape = NULL,   pooling = NULL,   classes = 1000L,   classifier_activation = \"softmax\",   name = \"efficientnetb4\",   ... )"},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_b4.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Instantiates the EfficientNetB4 architecture. — application_efficientnet_b4","text":"include_top Whether include fully-connected layer top network. Defaults TRUE. weights One NULL (random initialization), \"imagenet\" (pre-training ImageNet), path weights file loaded. Defaults \"imagenet\". input_tensor Optional Keras tensor (.e. output keras_input()) use image input model. input_shape Optional shape tuple, specified include_top FALSE. exactly 3 inputs channels. pooling Optional pooling mode feature extraction include_top FALSE. Defaults NULL. NULL means output model 4D tensor output last convolutional layer. avg means global average pooling applied output last convolutional layer, thus output model 2D tensor. max means global max pooling applied. classes Optional number classes classify images , specified include_top TRUE, weights argument specified. 1000 many ImageNet classes . Defaults 1000. classifier_activation str callable. activation function use \"top\" layer. Ignored unless include_top=TRUE. Set classifier_activation=NULL return logits \"top\" layer. Defaults 'softmax'. loading pretrained weights, classifier_activation can NULL \"softmax\". name name model (string). ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_b4.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Instantiates the EfficientNetB4 architecture. — application_efficientnet_b4","text":"model instance.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_b4.html","id":"reference","dir":"Reference","previous_headings":"","what":"Reference","title":"Instantiates the EfficientNetB4 architecture. — application_efficientnet_b4","text":"EfficientNet: Rethinking Model Scaling Convolutional Neural Networks (ICML 2019) function returns Keras image classification model, optionally loaded weights pre-trained ImageNet. image classification use cases, see page detailed examples. transfer learning use cases, make sure read guide transfer learning & fine-tuning.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_b4.html","id":"note","dir":"Reference","previous_headings":"","what":"Note","title":"Instantiates the EfficientNetB4 architecture. — application_efficientnet_b4","text":"Keras Application expects specific kind input preprocessing. EfficientNet, input preprocessing included part model (Rescaling layer), thus application_preprocess_inputs() actually pass-function. EfficientNet models expect inputs float tensors pixels values [0-255] range.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_b5.html","id":null,"dir":"Reference","previous_headings":"","what":"Instantiates the EfficientNetB5 architecture. — application_efficientnet_b5","title":"Instantiates the EfficientNetB5 architecture. — application_efficientnet_b5","text":"Instantiates EfficientNetB5 architecture.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_b5.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Instantiates the EfficientNetB5 architecture. — application_efficientnet_b5","text":"","code":"application_efficientnet_b5(   include_top = TRUE,   weights = \"imagenet\",   input_tensor = NULL,   input_shape = NULL,   pooling = NULL,   classes = 1000L,   classifier_activation = \"softmax\",   name = \"efficientnetb5\",   ... )"},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_b5.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Instantiates the EfficientNetB5 architecture. — application_efficientnet_b5","text":"include_top Whether include fully-connected layer top network. Defaults TRUE. weights One NULL (random initialization), \"imagenet\" (pre-training ImageNet), path weights file loaded. Defaults \"imagenet\". input_tensor Optional Keras tensor (.e. output keras_input()) use image input model. input_shape Optional shape tuple, specified include_top FALSE. exactly 3 inputs channels. pooling Optional pooling mode feature extraction include_top FALSE. Defaults NULL. NULL means output model 4D tensor output last convolutional layer. avg means global average pooling applied output last convolutional layer, thus output model 2D tensor. max means global max pooling applied. classes Optional number classes classify images , specified include_top TRUE, weights argument specified. 1000 many ImageNet classes . Defaults 1000. classifier_activation str callable. activation function use \"top\" layer. Ignored unless include_top=TRUE. Set classifier_activation=NULL return logits \"top\" layer. Defaults 'softmax'. loading pretrained weights, classifier_activation can NULL \"softmax\". name name model (string). ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_b5.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Instantiates the EfficientNetB5 architecture. — application_efficientnet_b5","text":"model instance.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_b5.html","id":"reference","dir":"Reference","previous_headings":"","what":"Reference","title":"Instantiates the EfficientNetB5 architecture. — application_efficientnet_b5","text":"EfficientNet: Rethinking Model Scaling Convolutional Neural Networks (ICML 2019) function returns Keras image classification model, optionally loaded weights pre-trained ImageNet. image classification use cases, see page detailed examples. transfer learning use cases, make sure read guide transfer learning & fine-tuning.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_b5.html","id":"note","dir":"Reference","previous_headings":"","what":"Note","title":"Instantiates the EfficientNetB5 architecture. — application_efficientnet_b5","text":"Keras Application expects specific kind input preprocessing. EfficientNet, input preprocessing included part model (Rescaling layer), thus application_preprocess_inputs() actually pass-function. EfficientNet models expect inputs float tensors pixels values [0-255] range.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_b6.html","id":null,"dir":"Reference","previous_headings":"","what":"Instantiates the EfficientNetB6 architecture. — application_efficientnet_b6","title":"Instantiates the EfficientNetB6 architecture. — application_efficientnet_b6","text":"Instantiates EfficientNetB6 architecture.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_b6.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Instantiates the EfficientNetB6 architecture. — application_efficientnet_b6","text":"","code":"application_efficientnet_b6(   include_top = TRUE,   weights = \"imagenet\",   input_tensor = NULL,   input_shape = NULL,   pooling = NULL,   classes = 1000L,   classifier_activation = \"softmax\",   name = \"efficientnetb6\",   ... )"},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_b6.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Instantiates the EfficientNetB6 architecture. — application_efficientnet_b6","text":"include_top Whether include fully-connected layer top network. Defaults TRUE. weights One NULL (random initialization), \"imagenet\" (pre-training ImageNet), path weights file loaded. Defaults \"imagenet\". input_tensor Optional Keras tensor (.e. output keras_input()) use image input model. input_shape Optional shape tuple, specified include_top FALSE. exactly 3 inputs channels. pooling Optional pooling mode feature extraction include_top FALSE. Defaults NULL. NULL means output model 4D tensor output last convolutional layer. avg means global average pooling applied output last convolutional layer, thus output model 2D tensor. max means global max pooling applied. classes Optional number classes classify images , specified include_top TRUE, weights argument specified. 1000 many ImageNet classes . Defaults 1000. classifier_activation str callable. activation function use \"top\" layer. Ignored unless include_top=TRUE. Set classifier_activation=NULL return logits \"top\" layer. Defaults 'softmax'. loading pretrained weights, classifier_activation can NULL \"softmax\". name name model (string). ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_b6.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Instantiates the EfficientNetB6 architecture. — application_efficientnet_b6","text":"model instance.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_b6.html","id":"reference","dir":"Reference","previous_headings":"","what":"Reference","title":"Instantiates the EfficientNetB6 architecture. — application_efficientnet_b6","text":"EfficientNet: Rethinking Model Scaling Convolutional Neural Networks (ICML 2019) function returns Keras image classification model, optionally loaded weights pre-trained ImageNet. image classification use cases, see page detailed examples. transfer learning use cases, make sure read guide transfer learning & fine-tuning.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_b6.html","id":"note","dir":"Reference","previous_headings":"","what":"Note","title":"Instantiates the EfficientNetB6 architecture. — application_efficientnet_b6","text":"Keras Application expects specific kind input preprocessing. EfficientNet, input preprocessing included part model (Rescaling layer), thus application_preprocess_inputs() actually pass-function. EfficientNet models expect inputs float tensors pixels values [0-255] range.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_b7.html","id":null,"dir":"Reference","previous_headings":"","what":"Instantiates the EfficientNetB7 architecture. — application_efficientnet_b7","title":"Instantiates the EfficientNetB7 architecture. — application_efficientnet_b7","text":"Instantiates EfficientNetB7 architecture.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_b7.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Instantiates the EfficientNetB7 architecture. — application_efficientnet_b7","text":"","code":"application_efficientnet_b7(   include_top = TRUE,   weights = \"imagenet\",   input_tensor = NULL,   input_shape = NULL,   pooling = NULL,   classes = 1000L,   classifier_activation = \"softmax\",   name = \"efficientnetb7\",   ... )"},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_b7.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Instantiates the EfficientNetB7 architecture. — application_efficientnet_b7","text":"include_top Whether include fully-connected layer top network. Defaults TRUE. weights One NULL (random initialization), \"imagenet\" (pre-training ImageNet), path weights file loaded. Defaults \"imagenet\". input_tensor Optional Keras tensor (.e. output keras_input()) use image input model. input_shape Optional shape tuple, specified include_top FALSE. exactly 3 inputs channels. pooling Optional pooling mode feature extraction include_top FALSE. Defaults NULL. NULL means output model 4D tensor output last convolutional layer. avg means global average pooling applied output last convolutional layer, thus output model 2D tensor. max means global max pooling applied. classes Optional number classes classify images , specified include_top TRUE, weights argument specified. 1000 many ImageNet classes . Defaults 1000. classifier_activation str callable. activation function use \"top\" layer. Ignored unless include_top=TRUE. Set classifier_activation=NULL return logits \"top\" layer. Defaults 'softmax'. loading pretrained weights, classifier_activation can NULL \"softmax\". name name model (string). ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_b7.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Instantiates the EfficientNetB7 architecture. — application_efficientnet_b7","text":"model instance.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_b7.html","id":"reference","dir":"Reference","previous_headings":"","what":"Reference","title":"Instantiates the EfficientNetB7 architecture. — application_efficientnet_b7","text":"EfficientNet: Rethinking Model Scaling Convolutional Neural Networks (ICML 2019) function returns Keras image classification model, optionally loaded weights pre-trained ImageNet. image classification use cases, see page detailed examples. transfer learning use cases, make sure read guide transfer learning & fine-tuning.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_b7.html","id":"note","dir":"Reference","previous_headings":"","what":"Note","title":"Instantiates the EfficientNetB7 architecture. — application_efficientnet_b7","text":"Keras Application expects specific kind input preprocessing. EfficientNet, input preprocessing included part model (Rescaling layer), thus application_preprocess_inputs() actually pass-function. EfficientNet models expect inputs float tensors pixels values [0-255] range.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_v2b0.html","id":null,"dir":"Reference","previous_headings":"","what":"Instantiates the EfficientNetV2B0 architecture. — application_efficientnet_v2b0","title":"Instantiates the EfficientNetV2B0 architecture. — application_efficientnet_v2b0","text":"Instantiates EfficientNetV2B0 architecture.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_v2b0.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Instantiates the EfficientNetV2B0 architecture. — application_efficientnet_v2b0","text":"","code":"application_efficientnet_v2b0(   include_top = TRUE,   weights = \"imagenet\",   input_tensor = NULL,   input_shape = NULL,   pooling = NULL,   classes = 1000L,   classifier_activation = \"softmax\",   include_preprocessing = TRUE,   name = \"efficientnetv2-b0\" )"},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_v2b0.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Instantiates the EfficientNetV2B0 architecture. — application_efficientnet_v2b0","text":"include_top Boolean, whether include fully-connected layer top network. Defaults TRUE. weights One NULL (random initialization), \"imagenet\" (pre-training ImageNet), path weights file loaded. Defaults \"imagenet\". input_tensor Optional Keras tensor (.e. output keras_input()) use image input model. input_shape Optional shape tuple, specified include_top FALSE. exactly 3 inputs channels. pooling Optional pooling mode feature extraction include_top FALSE. Defaults NULL. NULL means output model 4D tensor output last convolutional layer. \"avg\" means global average pooling applied output last convolutional layer, thus output model 2D tensor. \"max\" means global max pooling applied. classes Optional number classes classify images , specified include_top TRUE, weights argument specified. Defaults 1000 (number ImageNet classes). classifier_activation string callable. activation function use \"top\" layer. Ignored unless include_top=TRUE. Set classifier_activation=NULL return logits \"top\" layer. Defaults \"softmax\". loading pretrained weights, classifier_activation can NULL \"softmax\". include_preprocessing Boolean, whether include preprocessing layer bottom network. name name model (string).","code":""},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_v2b0.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Instantiates the EfficientNetV2B0 architecture. — application_efficientnet_v2b0","text":"model instance.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_v2b0.html","id":"reference","dir":"Reference","previous_headings":"","what":"Reference","title":"Instantiates the EfficientNetV2B0 architecture. — application_efficientnet_v2b0","text":"EfficientNetV2: Smaller Models Faster Training (ICML 2021) function returns Keras image classification model, optionally loaded weights pre-trained ImageNet. image classification use cases, see page detailed examples. transfer learning use cases, make sure read guide transfer learning & fine-tuning.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_v2b0.html","id":"note","dir":"Reference","previous_headings":"","what":"Note","title":"Instantiates the EfficientNetV2B0 architecture. — application_efficientnet_v2b0","text":"Keras Application expects specific kind input preprocessing. EfficientNetV2, default input preprocessing included part model (Rescaling layer), thus application_preprocess_inputs() actually pass-function. use case, EfficientNetV2 models expect inputs float tensors pixels values [0, 255] range. time, preprocessing part model (.e. Rescaling layer) can disabled setting include_preprocessing argument FALSE. preprocessing disabled EfficientNetV2 models expect inputs float tensors pixels values [-1, 1] range.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_v2b1.html","id":null,"dir":"Reference","previous_headings":"","what":"Instantiates the EfficientNetV2B1 architecture. — application_efficientnet_v2b1","title":"Instantiates the EfficientNetV2B1 architecture. — application_efficientnet_v2b1","text":"Instantiates EfficientNetV2B1 architecture.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_v2b1.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Instantiates the EfficientNetV2B1 architecture. — application_efficientnet_v2b1","text":"","code":"application_efficientnet_v2b1(   include_top = TRUE,   weights = \"imagenet\",   input_tensor = NULL,   input_shape = NULL,   pooling = NULL,   classes = 1000L,   classifier_activation = \"softmax\",   include_preprocessing = TRUE,   name = \"efficientnetv2-b1\" )"},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_v2b1.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Instantiates the EfficientNetV2B1 architecture. — application_efficientnet_v2b1","text":"include_top Boolean, whether include fully-connected layer top network. Defaults TRUE. weights One NULL (random initialization), \"imagenet\" (pre-training ImageNet), path weights file loaded. Defaults \"imagenet\". input_tensor Optional Keras tensor (.e. output keras_input()) use image input model. input_shape Optional shape tuple, specified include_top FALSE. exactly 3 inputs channels. pooling Optional pooling mode feature extraction include_top FALSE. Defaults NULL. NULL means output model 4D tensor output last convolutional layer. \"avg\" means global average pooling applied output last convolutional layer, thus output model 2D tensor. \"max\" means global max pooling applied. classes Optional number classes classify images , specified include_top TRUE, weights argument specified. Defaults 1000 (number ImageNet classes). classifier_activation string callable. activation function use \"top\" layer. Ignored unless include_top=TRUE. Set classifier_activation=NULL return logits \"top\" layer. Defaults \"softmax\". loading pretrained weights, classifier_activation can NULL \"softmax\". include_preprocessing Boolean, whether include preprocessing layer bottom network. name name model (string).","code":""},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_v2b1.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Instantiates the EfficientNetV2B1 architecture. — application_efficientnet_v2b1","text":"model instance.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_v2b1.html","id":"reference","dir":"Reference","previous_headings":"","what":"Reference","title":"Instantiates the EfficientNetV2B1 architecture. — application_efficientnet_v2b1","text":"EfficientNetV2: Smaller Models Faster Training (ICML 2021) function returns Keras image classification model, optionally loaded weights pre-trained ImageNet. image classification use cases, see page detailed examples. transfer learning use cases, make sure read guide transfer learning & fine-tuning.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_v2b1.html","id":"note","dir":"Reference","previous_headings":"","what":"Note","title":"Instantiates the EfficientNetV2B1 architecture. — application_efficientnet_v2b1","text":"Keras Application expects specific kind input preprocessing. EfficientNetV2, default input preprocessing included part model (Rescaling layer), thus application_preprocess_inputs() actually pass-function. use case, EfficientNetV2 models expect inputs float tensors pixels values [0, 255] range. time, preprocessing part model (.e. Rescaling layer) can disabled setting include_preprocessing argument FALSE. preprocessing disabled EfficientNetV2 models expect inputs float tensors pixels values [-1, 1] range.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_v2b2.html","id":null,"dir":"Reference","previous_headings":"","what":"Instantiates the EfficientNetV2B2 architecture. — application_efficientnet_v2b2","title":"Instantiates the EfficientNetV2B2 architecture. — application_efficientnet_v2b2","text":"Instantiates EfficientNetV2B2 architecture.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_v2b2.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Instantiates the EfficientNetV2B2 architecture. — application_efficientnet_v2b2","text":"","code":"application_efficientnet_v2b2(   include_top = TRUE,   weights = \"imagenet\",   input_tensor = NULL,   input_shape = NULL,   pooling = NULL,   classes = 1000L,   classifier_activation = \"softmax\",   include_preprocessing = TRUE,   name = \"efficientnetv2-b2\" )"},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_v2b2.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Instantiates the EfficientNetV2B2 architecture. — application_efficientnet_v2b2","text":"include_top Boolean, whether include fully-connected layer top network. Defaults TRUE. weights One NULL (random initialization), \"imagenet\" (pre-training ImageNet), path weights file loaded. Defaults \"imagenet\". input_tensor Optional Keras tensor (.e. output keras_input()) use image input model. input_shape Optional shape tuple, specified include_top FALSE. exactly 3 inputs channels. pooling Optional pooling mode feature extraction include_top FALSE. Defaults NULL. NULL means output model 4D tensor output last convolutional layer. \"avg\" means global average pooling applied output last convolutional layer, thus output model 2D tensor. \"max\" means global max pooling applied. classes Optional number classes classify images , specified include_top TRUE, weights argument specified. Defaults 1000 (number ImageNet classes). classifier_activation string callable. activation function use \"top\" layer. Ignored unless include_top=TRUE. Set classifier_activation=NULL return logits \"top\" layer. Defaults \"softmax\". loading pretrained weights, classifier_activation can NULL \"softmax\". include_preprocessing Boolean, whether include preprocessing layer bottom network. name name model (string).","code":""},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_v2b2.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Instantiates the EfficientNetV2B2 architecture. — application_efficientnet_v2b2","text":"model instance.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_v2b2.html","id":"reference","dir":"Reference","previous_headings":"","what":"Reference","title":"Instantiates the EfficientNetV2B2 architecture. — application_efficientnet_v2b2","text":"EfficientNetV2: Smaller Models Faster Training (ICML 2021) function returns Keras image classification model, optionally loaded weights pre-trained ImageNet. image classification use cases, see page detailed examples. transfer learning use cases, make sure read guide transfer learning & fine-tuning.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_v2b2.html","id":"note","dir":"Reference","previous_headings":"","what":"Note","title":"Instantiates the EfficientNetV2B2 architecture. — application_efficientnet_v2b2","text":"Keras Application expects specific kind input preprocessing. EfficientNetV2, default input preprocessing included part model (Rescaling layer), thus application_preprocess_inputs() actually pass-function. use case, EfficientNetV2 models expect inputs float tensors pixels values [0, 255] range. time, preprocessing part model (.e. Rescaling layer) can disabled setting include_preprocessing argument FALSE. preprocessing disabled EfficientNetV2 models expect inputs float tensors pixels values [-1, 1] range.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_v2b3.html","id":null,"dir":"Reference","previous_headings":"","what":"Instantiates the EfficientNetV2B3 architecture. — application_efficientnet_v2b3","title":"Instantiates the EfficientNetV2B3 architecture. — application_efficientnet_v2b3","text":"Instantiates EfficientNetV2B3 architecture.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_v2b3.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Instantiates the EfficientNetV2B3 architecture. — application_efficientnet_v2b3","text":"","code":"application_efficientnet_v2b3(   include_top = TRUE,   weights = \"imagenet\",   input_tensor = NULL,   input_shape = NULL,   pooling = NULL,   classes = 1000L,   classifier_activation = \"softmax\",   include_preprocessing = TRUE,   name = \"efficientnetv2-b3\" )"},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_v2b3.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Instantiates the EfficientNetV2B3 architecture. — application_efficientnet_v2b3","text":"include_top Boolean, whether include fully-connected layer top network. Defaults TRUE. weights One NULL (random initialization), \"imagenet\" (pre-training ImageNet), path weights file loaded. Defaults \"imagenet\". input_tensor Optional Keras tensor (.e. output keras_input()) use image input model. input_shape Optional shape tuple, specified include_top FALSE. exactly 3 inputs channels. pooling Optional pooling mode feature extraction include_top FALSE. Defaults NULL. NULL means output model 4D tensor output last convolutional layer. \"avg\" means global average pooling applied output last convolutional layer, thus output model 2D tensor. \"max\" means global max pooling applied. classes Optional number classes classify images , specified include_top TRUE, weights argument specified. Defaults 1000 (number ImageNet classes). classifier_activation string callable. activation function use \"top\" layer. Ignored unless include_top=TRUE. Set classifier_activation=NULL return logits \"top\" layer. Defaults \"softmax\". loading pretrained weights, classifier_activation can NULL \"softmax\". include_preprocessing Boolean, whether include preprocessing layer bottom network. name name model (string).","code":""},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_v2b3.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Instantiates the EfficientNetV2B3 architecture. — application_efficientnet_v2b3","text":"model instance.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_v2b3.html","id":"reference","dir":"Reference","previous_headings":"","what":"Reference","title":"Instantiates the EfficientNetV2B3 architecture. — application_efficientnet_v2b3","text":"EfficientNetV2: Smaller Models Faster Training (ICML 2021) function returns Keras image classification model, optionally loaded weights pre-trained ImageNet. image classification use cases, see page detailed examples. transfer learning use cases, make sure read guide transfer learning & fine-tuning.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_v2b3.html","id":"note","dir":"Reference","previous_headings":"","what":"Note","title":"Instantiates the EfficientNetV2B3 architecture. — application_efficientnet_v2b3","text":"Keras Application expects specific kind input preprocessing. EfficientNetV2, default input preprocessing included part model (Rescaling layer), thus application_preprocess_inputs() actually pass-function. use case, EfficientNetV2 models expect inputs float tensors pixels values [0, 255] range. time, preprocessing part model (.e. Rescaling layer) can disabled setting include_preprocessing argument FALSE. preprocessing disabled EfficientNetV2 models expect inputs float tensors pixels values [-1, 1] range.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_v2l.html","id":null,"dir":"Reference","previous_headings":"","what":"Instantiates the EfficientNetV2L architecture. — application_efficientnet_v2l","title":"Instantiates the EfficientNetV2L architecture. — application_efficientnet_v2l","text":"Instantiates EfficientNetV2L architecture.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_v2l.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Instantiates the EfficientNetV2L architecture. — application_efficientnet_v2l","text":"","code":"application_efficientnet_v2l(   include_top = TRUE,   weights = \"imagenet\",   input_tensor = NULL,   input_shape = NULL,   pooling = NULL,   classes = 1000L,   classifier_activation = \"softmax\",   include_preprocessing = TRUE,   name = \"efficientnetv2-l\" )"},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_v2l.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Instantiates the EfficientNetV2L architecture. — application_efficientnet_v2l","text":"include_top Boolean, whether include fully-connected layer top network. Defaults TRUE. weights One NULL (random initialization), \"imagenet\" (pre-training ImageNet), path weights file loaded. Defaults \"imagenet\". input_tensor Optional Keras tensor (.e. output keras_input()) use image input model. input_shape Optional shape tuple, specified include_top FALSE. exactly 3 inputs channels. pooling Optional pooling mode feature extraction include_top FALSE. Defaults NULL. NULL means output model 4D tensor output last convolutional layer. \"avg\" means global average pooling applied output last convolutional layer, thus output model 2D tensor. \"max\" means global max pooling applied. classes Optional number classes classify images , specified include_top TRUE, weights argument specified. Defaults 1000 (number ImageNet classes). classifier_activation string callable. activation function use \"top\" layer. Ignored unless include_top=TRUE. Set classifier_activation=NULL return logits \"top\" layer. Defaults \"softmax\". loading pretrained weights, classifier_activation can NULL \"softmax\". include_preprocessing Boolean, whether include preprocessing layer bottom network. name name model (string).","code":""},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_v2l.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Instantiates the EfficientNetV2L architecture. — application_efficientnet_v2l","text":"model instance.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_v2l.html","id":"reference","dir":"Reference","previous_headings":"","what":"Reference","title":"Instantiates the EfficientNetV2L architecture. — application_efficientnet_v2l","text":"EfficientNetV2: Smaller Models Faster Training (ICML 2021) function returns Keras image classification model, optionally loaded weights pre-trained ImageNet. image classification use cases, see page detailed examples. transfer learning use cases, make sure read guide transfer learning & fine-tuning.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_v2l.html","id":"note","dir":"Reference","previous_headings":"","what":"Note","title":"Instantiates the EfficientNetV2L architecture. — application_efficientnet_v2l","text":"Keras Application expects specific kind input preprocessing. EfficientNetV2, default input preprocessing included part model (Rescaling layer), thus application_preprocess_inputs() actually pass-function. use case, EfficientNetV2 models expect inputs float tensors pixels values [0, 255] range. time, preprocessing part model (.e. Rescaling layer) can disabled setting include_preprocessing argument FALSE. preprocessing disabled EfficientNetV2 models expect inputs float tensors pixels values [-1, 1] range.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_v2m.html","id":null,"dir":"Reference","previous_headings":"","what":"Instantiates the EfficientNetV2M architecture. — application_efficientnet_v2m","title":"Instantiates the EfficientNetV2M architecture. — application_efficientnet_v2m","text":"Instantiates EfficientNetV2M architecture.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_v2m.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Instantiates the EfficientNetV2M architecture. — application_efficientnet_v2m","text":"","code":"application_efficientnet_v2m(   include_top = TRUE,   weights = \"imagenet\",   input_tensor = NULL,   input_shape = NULL,   pooling = NULL,   classes = 1000L,   classifier_activation = \"softmax\",   include_preprocessing = TRUE,   name = \"efficientnetv2-m\" )"},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_v2m.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Instantiates the EfficientNetV2M architecture. — application_efficientnet_v2m","text":"include_top Boolean, whether include fully-connected layer top network. Defaults TRUE. weights One NULL (random initialization), \"imagenet\" (pre-training ImageNet), path weights file loaded. Defaults \"imagenet\". input_tensor Optional Keras tensor (.e. output keras_input()) use image input model. input_shape Optional shape tuple, specified include_top FALSE. exactly 3 inputs channels. pooling Optional pooling mode feature extraction include_top FALSE. Defaults NULL. NULL means output model 4D tensor output last convolutional layer. \"avg\" means global average pooling applied output last convolutional layer, thus output model 2D tensor. \"max\" means global max pooling applied. classes Optional number classes classify images , specified include_top TRUE, weights argument specified. Defaults 1000 (number ImageNet classes). classifier_activation string callable. activation function use \"top\" layer. Ignored unless include_top=TRUE. Set classifier_activation=NULL return logits \"top\" layer. Defaults \"softmax\". loading pretrained weights, classifier_activation can NULL \"softmax\". include_preprocessing Boolean, whether include preprocessing layer bottom network. name name model (string).","code":""},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_v2m.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Instantiates the EfficientNetV2M architecture. — application_efficientnet_v2m","text":"model instance.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_v2m.html","id":"reference","dir":"Reference","previous_headings":"","what":"Reference","title":"Instantiates the EfficientNetV2M architecture. — application_efficientnet_v2m","text":"EfficientNetV2: Smaller Models Faster Training (ICML 2021) function returns Keras image classification model, optionally loaded weights pre-trained ImageNet. image classification use cases, see page detailed examples. transfer learning use cases, make sure read guide transfer learning & fine-tuning.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_v2m.html","id":"note","dir":"Reference","previous_headings":"","what":"Note","title":"Instantiates the EfficientNetV2M architecture. — application_efficientnet_v2m","text":"Keras Application expects specific kind input preprocessing. EfficientNetV2, default input preprocessing included part model (Rescaling layer), thus application_preprocess_inputs() actually pass-function. use case, EfficientNetV2 models expect inputs float tensors pixels values [0, 255] range. time, preprocessing part model (.e. Rescaling layer) can disabled setting include_preprocessing argument FALSE. preprocessing disabled EfficientNetV2 models expect inputs float tensors pixels values [-1, 1] range.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_v2s.html","id":null,"dir":"Reference","previous_headings":"","what":"Instantiates the EfficientNetV2S architecture. — application_efficientnet_v2s","title":"Instantiates the EfficientNetV2S architecture. — application_efficientnet_v2s","text":"Instantiates EfficientNetV2S architecture.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_v2s.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Instantiates the EfficientNetV2S architecture. — application_efficientnet_v2s","text":"","code":"application_efficientnet_v2s(   include_top = TRUE,   weights = \"imagenet\",   input_tensor = NULL,   input_shape = NULL,   pooling = NULL,   classes = 1000L,   classifier_activation = \"softmax\",   include_preprocessing = TRUE,   name = \"efficientnetv2-s\" )"},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_v2s.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Instantiates the EfficientNetV2S architecture. — application_efficientnet_v2s","text":"include_top Boolean, whether include fully-connected layer top network. Defaults TRUE. weights One NULL (random initialization), \"imagenet\" (pre-training ImageNet), path weights file loaded. Defaults \"imagenet\". input_tensor Optional Keras tensor (.e. output keras_input()) use image input model. input_shape Optional shape tuple, specified include_top FALSE. exactly 3 inputs channels. pooling Optional pooling mode feature extraction include_top FALSE. Defaults NULL. NULL means output model 4D tensor output last convolutional layer. \"avg\" means global average pooling applied output last convolutional layer, thus output model 2D tensor. \"max\" means global max pooling applied. classes Optional number classes classify images , specified include_top TRUE, weights argument specified. Defaults 1000 (number ImageNet classes). classifier_activation string callable. activation function use \"top\" layer. Ignored unless include_top=TRUE. Set classifier_activation=NULL return logits \"top\" layer. Defaults \"softmax\". loading pretrained weights, classifier_activation can NULL \"softmax\". include_preprocessing Boolean, whether include preprocessing layer bottom network. name name model (string).","code":""},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_v2s.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Instantiates the EfficientNetV2S architecture. — application_efficientnet_v2s","text":"model instance.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_v2s.html","id":"reference","dir":"Reference","previous_headings":"","what":"Reference","title":"Instantiates the EfficientNetV2S architecture. — application_efficientnet_v2s","text":"EfficientNetV2: Smaller Models Faster Training (ICML 2021) function returns Keras image classification model, optionally loaded weights pre-trained ImageNet. image classification use cases, see page detailed examples. transfer learning use cases, make sure read guide transfer learning & fine-tuning.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_efficientnet_v2s.html","id":"note","dir":"Reference","previous_headings":"","what":"Note","title":"Instantiates the EfficientNetV2S architecture. — application_efficientnet_v2s","text":"Keras Application expects specific kind input preprocessing. EfficientNetV2, default input preprocessing included part model (Rescaling layer), thus application_preprocess_inputs() actually pass-function. use case, EfficientNetV2 models expect inputs float tensors pixels values [0, 255] range. time, preprocessing part model (.e. Rescaling layer) can disabled setting include_preprocessing argument FALSE. preprocessing disabled EfficientNetV2 models expect inputs float tensors pixels values [-1, 1] range.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/application_inception_resnet_v2.html","id":null,"dir":"Reference","previous_headings":"","what":"Instantiates the Inception-ResNet v2 architecture. — application_inception_resnet_v2","title":"Instantiates the Inception-ResNet v2 architecture. — application_inception_resnet_v2","text":"Instantiates Inception-ResNet v2 architecture.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_inception_resnet_v2.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Instantiates the Inception-ResNet v2 architecture. — application_inception_resnet_v2","text":"","code":"application_inception_resnet_v2(   include_top = TRUE,   weights = \"imagenet\",   input_tensor = NULL,   input_shape = NULL,   pooling = NULL,   classes = 1000L,   classifier_activation = \"softmax\",   name = \"inception_resnet_v2\" )"},{"path":"https://keras3.posit.co/dev/reference/application_inception_resnet_v2.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Instantiates the Inception-ResNet v2 architecture. — application_inception_resnet_v2","text":"include_top whether include fully-connected layer top network. weights one NULL (random initialization), \"imagenet\" (pre-training ImageNet), path weights file loaded. input_tensor optional Keras tensor (.e. output keras_input()) use image input model. input_shape optional shape tuple, specified include_top FALSE (otherwise input shape (299, 299, 3) ('channels_last' data format) (3, 299, 299) ('channels_first' data format). exactly 3 inputs channels, width height smaller 75. E.g. (150, 150, 3) one valid value. pooling Optional pooling mode feature extraction include_top FALSE. NULL means output model 4D tensor output last convolutional block. 'avg' means global average pooling applied output last convolutional block, thus output model 2D tensor. 'max' means global max pooling applied. classes optional number classes classify images , specified include_top TRUE, weights argument specified. classifier_activation str callable. activation function use \"top\" layer. Ignored unless include_top=TRUE. Set classifier_activation=NULL return logits \"top\" layer. loading pretrained weights, classifier_activation can NULL \"softmax\". name name model (string).","code":""},{"path":"https://keras3.posit.co/dev/reference/application_inception_resnet_v2.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Instantiates the Inception-ResNet v2 architecture. — application_inception_resnet_v2","text":"model instance.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_inception_resnet_v2.html","id":"reference","dir":"Reference","previous_headings":"","what":"Reference","title":"Instantiates the Inception-ResNet v2 architecture. — application_inception_resnet_v2","text":"Inception-v4, Inception-ResNet Impact Residual Connections Learning (AAAI 2017) function returns Keras image classification model, optionally loaded weights pre-trained ImageNet. image classification use cases, see page detailed examples. transfer learning use cases, make sure read guide transfer learning & fine-tuning.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_inception_resnet_v2.html","id":"note","dir":"Reference","previous_headings":"","what":"Note","title":"Instantiates the Inception-ResNet v2 architecture. — application_inception_resnet_v2","text":"Keras Application expects specific kind input preprocessing. InceptionResNetV2, call application_preprocess_inputs() inputs passing model. application_preprocess_inputs() scale input pixels -1 1.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/application_inception_v3.html","id":null,"dir":"Reference","previous_headings":"","what":"Instantiates the Inception v3 architecture. — application_inception_v3","title":"Instantiates the Inception v3 architecture. — application_inception_v3","text":"Instantiates Inception v3 architecture.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_inception_v3.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Instantiates the Inception v3 architecture. — application_inception_v3","text":"","code":"application_inception_v3(   include_top = TRUE,   weights = \"imagenet\",   input_tensor = NULL,   input_shape = NULL,   pooling = NULL,   classes = 1000L,   classifier_activation = \"softmax\",   name = \"inception_v3\" )"},{"path":"https://keras3.posit.co/dev/reference/application_inception_v3.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Instantiates the Inception v3 architecture. — application_inception_v3","text":"include_top Boolean, whether include fully-connected layer top, last layer network. Defaults TRUE. weights One NULL (random initialization), imagenet (pre-training ImageNet), path weights file loaded. Defaults \"imagenet\". input_tensor Optional Keras tensor (.e. output keras_input()) use image input model. input_tensor useful sharing inputs multiple different networks. Defaults NULL. input_shape Optional shape tuple, specified include_top FALSE (otherwise input shape (299, 299, 3) (channels_last data format) (3, 299, 299) (channels_first data format). exactly 3 inputs channels, width height smaller 75. E.g. (150, 150, 3) one valid value. input_shape ignored input_tensor provided. pooling Optional pooling mode feature extraction include_top FALSE. NULL (default) means output model 4D tensor output last convolutional block. avg means global average pooling applied output last convolutional block, thus output model 2D tensor. max means global max pooling applied. classes optional number classes classify images , specified include_top TRUE, weights argument specified. Defaults 1000. classifier_activation str callable. activation function use \"top\" layer. Ignored unless include_top=TRUE. Set classifier_activation=NULL return logits \"top\" layer. loading pretrained weights, classifier_activation can NULL \"softmax\". name name model (string).","code":""},{"path":"https://keras3.posit.co/dev/reference/application_inception_v3.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Instantiates the Inception v3 architecture. — application_inception_v3","text":"model instance.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_inception_v3.html","id":"reference","dir":"Reference","previous_headings":"","what":"Reference","title":"Instantiates the Inception v3 architecture. — application_inception_v3","text":"Rethinking Inception Architecture Computer Vision (CVPR 2016) function returns Keras image classification model, optionally loaded weights pre-trained ImageNet. image classification use cases, see page detailed examples. transfer learning use cases, make sure read guide transfer learning & fine-tuning.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_inception_v3.html","id":"note","dir":"Reference","previous_headings":"","what":"Note","title":"Instantiates the Inception v3 architecture. — application_inception_v3","text":"Keras Application expects specific kind input preprocessing. InceptionV3, call application_preprocess_inputs() inputs passing model. application_preprocess_inputs() scale input pixels -1 1.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/application_mobilenet.html","id":null,"dir":"Reference","previous_headings":"","what":"Instantiates the MobileNet architecture. — application_mobilenet","title":"Instantiates the MobileNet architecture. — application_mobilenet","text":"Instantiates MobileNet architecture.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_mobilenet.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Instantiates the MobileNet architecture. — application_mobilenet","text":"","code":"application_mobilenet(   input_shape = NULL,   alpha = 1,   depth_multiplier = 1L,   dropout = 0.001,   include_top = TRUE,   weights = \"imagenet\",   input_tensor = NULL,   pooling = NULL,   classes = 1000L,   classifier_activation = \"softmax\",   name = NULL )"},{"path":"https://keras3.posit.co/dev/reference/application_mobilenet.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Instantiates the MobileNet architecture. — application_mobilenet","text":"input_shape Optional shape tuple, specified include_top FALSE (otherwise input shape (224, 224, 3) (\"channels_last\" data format) (3, 224, 224) (\"channels_first\" data format). exactly 3 inputs channels, width height smaller 32. E.g. (200, 200, 3) one valid value. Defaults NULL. input_shape ignored input_tensor provided. alpha Controls width network. known width multiplier MobileNet paper. alpha < 1.0, proportionally decreases number filters layer. alpha > 1.0, proportionally increases number filters layer. alpha == 1, default number filters paper used layer. Defaults 1.0. depth_multiplier Depth multiplier depthwise convolution. called resolution multiplier MobileNet paper. Defaults 1.0. dropout Dropout rate. Defaults 0.001. include_top Boolean, whether include fully-connected layer top network. Defaults TRUE. weights One NULL (random initialization), \"imagenet\" (pre-training ImageNet), path weights file loaded. Defaults \"imagenet\". input_tensor Optional Keras tensor (.e. output keras_input()) use image input model. input_tensor useful sharing inputs multiple different networks. Defaults NULL. pooling Optional pooling mode feature extraction include_top FALSE. NULL (default) means output model 4D tensor output last convolutional block. avg means global average pooling applied output last convolutional block, thus output model 2D tensor. max means global max pooling applied. classes Optional number classes classify images , specified include_top TRUE, weights argument specified. Defaults 1000. classifier_activation str callable. activation function use \"top\" layer. Ignored unless include_top=TRUE. Set classifier_activation=NULL return logits \"top\" layer. loading pretrained weights, classifier_activation can NULL \"softmax\". name name model (string).","code":""},{"path":"https://keras3.posit.co/dev/reference/application_mobilenet.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Instantiates the MobileNet architecture. — application_mobilenet","text":"model instance.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_mobilenet.html","id":"reference","dir":"Reference","previous_headings":"","what":"Reference","title":"Instantiates the MobileNet architecture. — application_mobilenet","text":"MobileNets: Efficient Convolutional Neural Networks Mobile Vision Applications function returns Keras image classification model, optionally loaded weights pre-trained ImageNet. image classification use cases, see page detailed examples. transfer learning use cases, make sure read guide transfer learning & fine-tuning.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_mobilenet.html","id":"note","dir":"Reference","previous_headings":"","what":"Note","title":"Instantiates the MobileNet architecture. — application_mobilenet","text":"Keras Application expects specific kind input preprocessing. MobileNet, call application_preprocess_inputs() inputs passing model. application_preprocess_inputs() scale input pixels -1 1.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/application_mobilenet_v2.html","id":null,"dir":"Reference","previous_headings":"","what":"Instantiates the MobileNetV2 architecture. — application_mobilenet_v2","title":"Instantiates the MobileNetV2 architecture. — application_mobilenet_v2","text":"MobileNetV2 similar original MobileNet, except uses inverted residual blocks bottlenecking features. drastically lower parameter count original MobileNet. MobileNets support input size greater 32 x 32, larger image sizes offering better performance.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_mobilenet_v2.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Instantiates the MobileNetV2 architecture. — application_mobilenet_v2","text":"","code":"application_mobilenet_v2(   input_shape = NULL,   alpha = 1,   include_top = TRUE,   weights = \"imagenet\",   input_tensor = NULL,   pooling = NULL,   classes = 1000L,   classifier_activation = \"softmax\",   name = NULL )"},{"path":"https://keras3.posit.co/dev/reference/application_mobilenet_v2.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Instantiates the MobileNetV2 architecture. — application_mobilenet_v2","text":"input_shape Optional shape tuple, specified include_top FALSE (otherwise input shape (224, 224, 3) (\"channels_last\" data format) (3, 224, 224) (\"channels_first\" data format). exactly 3 inputs channels, width height smaller 32. E.g. (200, 200, 3) one valid value. Defaults NULL. input_shape ignored input_tensor provided. alpha Controls width network. known width multiplier MobileNet paper. alpha < 1.0, proportionally decreases number filters layer. alpha > 1.0, proportionally increases number filters layer. alpha == 1, default number filters paper used layer. Defaults 1.0. include_top Boolean, whether include fully-connected layer top network. Defaults TRUE. weights One NULL (random initialization), \"imagenet\" (pre-training ImageNet), path weights file loaded. Defaults \"imagenet\". input_tensor Optional Keras tensor (.e. output keras_input()) use image input model. input_tensor useful sharing inputs multiple different networks. Defaults NULL. pooling Optional pooling mode feature extraction include_top FALSE. NULL (default) means output model 4D tensor output last convolutional block. avg means global average pooling applied output last convolutional block, thus output model 2D tensor. max means global max pooling applied. classes Optional number classes classify images , specified include_top TRUE, weights argument specified. Defaults 1000. classifier_activation str callable. activation function use \"top\" layer. Ignored unless include_top=TRUE. Set classifier_activation=NULL return logits \"top\" layer. loading pretrained weights, classifier_activation can NULL \"softmax\". name name model (string).","code":""},{"path":"https://keras3.posit.co/dev/reference/application_mobilenet_v2.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Instantiates the MobileNetV2 architecture. — application_mobilenet_v2","text":"model instance.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_mobilenet_v2.html","id":"reference","dir":"Reference","previous_headings":"","what":"Reference","title":"Instantiates the MobileNetV2 architecture. — application_mobilenet_v2","text":"MobileNetV2: Inverted Residuals Linear Bottlenecks (CVPR 2018) function returns Keras image classification model, optionally loaded weights pre-trained ImageNet. image classification use cases, see page detailed examples. transfer learning use cases, make sure read guide transfer learning & fine-tuning.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_mobilenet_v2.html","id":"note","dir":"Reference","previous_headings":"","what":"Note","title":"Instantiates the MobileNetV2 architecture. — application_mobilenet_v2","text":"Keras Application expects specific kind input preprocessing. MobileNetV2, call application_preprocess_inputs() inputs passing model. application_preprocess_inputs() scale input pixels -1 1.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/application_mobilenet_v3_large.html","id":null,"dir":"Reference","previous_headings":"","what":"Instantiates the MobileNetV3Large architecture. — application_mobilenet_v3_large","title":"Instantiates the MobileNetV3Large architecture. — application_mobilenet_v3_large","text":"Instantiates MobileNetV3Large architecture.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_mobilenet_v3_large.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Instantiates the MobileNetV3Large architecture. — application_mobilenet_v3_large","text":"","code":"application_mobilenet_v3_large(   input_shape = NULL,   alpha = 1,   minimalistic = FALSE,   include_top = TRUE,   weights = \"imagenet\",   input_tensor = NULL,   classes = 1000L,   pooling = NULL,   dropout_rate = 0.2,   classifier_activation = \"softmax\",   include_preprocessing = TRUE,   name = \"MobileNetV3Large\" )"},{"path":"https://keras3.posit.co/dev/reference/application_mobilenet_v3_large.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Instantiates the MobileNetV3Large architecture. — application_mobilenet_v3_large","text":"input_shape Optional shape tuple, specified like use model input image resolution (224, 224, 3). exactly 3 inputs channels. can also omit option like infer input_shape input_tensor. choose include input_tensor input_shape input_shape used match, shapes match throw error. E.g. (160, 160, 3) one valid value. alpha controls width network. known depth multiplier MobileNetV3 paper, name kept consistency MobileNetV1 Keras. alpha < 1.0, proportionally decreases number filters layer. alpha > 1.0, proportionally increases number filters layer. alpha == 1, default number filters paper used layer. minimalistic addition large small models module also contains -called minimalistic models, models per-layer dimensions characteristic MobilenetV3 however, utilize advanced blocks (squeeze--excite units, hard-swish, 5x5 convolutions). models less efficient CPU, much performant GPU/DSP. include_top Boolean, whether include fully-connected layer top network. Defaults TRUE. weights String, one NULL (random initialization), \"imagenet\" (pre-training ImageNet), path weights file loaded. input_tensor Optional Keras tensor (.e. output keras_input()) use image input model. classes Integer, optional number classes classify images , specified include_top TRUE, weights argument specified. pooling String, optional pooling mode feature extraction include_top FALSE. NULL means output model 4D tensor output last convolutional block. avg means global average pooling applied output last convolutional block, thus output model 2D tensor. max means global max pooling applied. dropout_rate fraction input units drop last layer. classifier_activation str callable. activation function use \"top\" layer. Ignored unless include_top=TRUE. Set classifier_activation=NULL return logits \"top\" layer. loading pretrained weights, classifier_activation can NULL \"softmax\". include_preprocessing Boolean, whether include preprocessing layer (Rescaling) bottom network. Defaults TRUE. name name model (string).","code":""},{"path":"https://keras3.posit.co/dev/reference/application_mobilenet_v3_large.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Instantiates the MobileNetV3Large architecture. — application_mobilenet_v3_large","text":"model instance.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_mobilenet_v3_large.html","id":"reference","dir":"Reference","previous_headings":"","what":"Reference","title":"Instantiates the MobileNetV3Large architecture. — application_mobilenet_v3_large","text":"Searching MobileNetV3 (ICCV 2019)","code":""},{"path":"https://keras3.posit.co/dev/reference/application_mobilenet_v3_large.html","id":"the-following-table-describes-the-performance-of-mobilenets-v-","dir":"Reference","previous_headings":"","what":"The following table describes the performance of MobileNets v3:","title":"Instantiates the MobileNetV3Large architecture. — application_mobilenet_v3_large","text":"MACs stands Multiply Adds image classification use cases, see page detailed examples. transfer learning use cases, make sure read guide transfer learning & fine-tuning.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_mobilenet_v3_large.html","id":"note","dir":"Reference","previous_headings":"","what":"Note","title":"Instantiates the MobileNetV3Large architecture. — application_mobilenet_v3_large","text":"Keras Application expects specific kind input preprocessing. MobileNetV3, default input preprocessing included part model (Rescaling layer), thus application_preprocess_inputs() actually pass-function. use case, MobileNetV3 models expect inputs float tensors pixels values [0-255] range. time, preprocessing part model (.e. Rescaling layer) can disabled setting include_preprocessing argument FALSE. preprocessing disabled MobileNetV3 models expect inputs float tensors pixels values [-1, 1] range.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_mobilenet_v3_large.html","id":"call-arguments","dir":"Reference","previous_headings":"","what":"Call Arguments","title":"Instantiates the MobileNetV3Large architecture. — application_mobilenet_v3_large","text":"inputs: floating point numpy.array backend-native tensor, 4D 3 color channels, values range [0, 255] include_preprocessing TRUE range [-1, 1] otherwise.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/application_mobilenet_v3_small.html","id":null,"dir":"Reference","previous_headings":"","what":"Instantiates the MobileNetV3Small architecture. — application_mobilenet_v3_small","title":"Instantiates the MobileNetV3Small architecture. — application_mobilenet_v3_small","text":"Instantiates MobileNetV3Small architecture.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_mobilenet_v3_small.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Instantiates the MobileNetV3Small architecture. — application_mobilenet_v3_small","text":"","code":"application_mobilenet_v3_small(   input_shape = NULL,   alpha = 1,   minimalistic = FALSE,   include_top = TRUE,   weights = \"imagenet\",   input_tensor = NULL,   classes = 1000L,   pooling = NULL,   dropout_rate = 0.2,   classifier_activation = \"softmax\",   include_preprocessing = TRUE,   name = \"MobileNetV3Small\" )"},{"path":"https://keras3.posit.co/dev/reference/application_mobilenet_v3_small.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Instantiates the MobileNetV3Small architecture. — application_mobilenet_v3_small","text":"input_shape Optional shape tuple, specified like use model input image resolution (224, 224, 3). exactly 3 inputs channels. can also omit option like infer input_shape input_tensor. choose include input_tensor input_shape input_shape used match, shapes match throw error. E.g. (160, 160, 3) one valid value. alpha controls width network. known depth multiplier MobileNetV3 paper, name kept consistency MobileNetV1 Keras. alpha < 1.0, proportionally decreases number filters layer. alpha > 1.0, proportionally increases number filters layer. alpha == 1, default number filters paper used layer. minimalistic addition large small models module also contains -called minimalistic models, models per-layer dimensions characteristic MobilenetV3 however, utilize advanced blocks (squeeze--excite units, hard-swish, 5x5 convolutions). models less efficient CPU, much performant GPU/DSP. include_top Boolean, whether include fully-connected layer top network. Defaults TRUE. weights String, one NULL (random initialization), \"imagenet\" (pre-training ImageNet), path weights file loaded. input_tensor Optional Keras tensor (.e. output keras_input()) use image input model. classes Integer, optional number classes classify images , specified include_top TRUE, weights argument specified. pooling String, optional pooling mode feature extraction include_top FALSE. NULL means output model 4D tensor output last convolutional block. avg means global average pooling applied output last convolutional block, thus output model 2D tensor. max means global max pooling applied. dropout_rate fraction input units drop last layer. classifier_activation str callable. activation function use \"top\" layer. Ignored unless include_top=TRUE. Set classifier_activation=NULL return logits \"top\" layer. loading pretrained weights, classifier_activation can NULL \"softmax\". include_preprocessing Boolean, whether include preprocessing layer (Rescaling) bottom network. Defaults TRUE. name name model (string).","code":""},{"path":"https://keras3.posit.co/dev/reference/application_mobilenet_v3_small.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Instantiates the MobileNetV3Small architecture. — application_mobilenet_v3_small","text":"model instance.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_mobilenet_v3_small.html","id":"reference","dir":"Reference","previous_headings":"","what":"Reference","title":"Instantiates the MobileNetV3Small architecture. — application_mobilenet_v3_small","text":"Searching MobileNetV3 (ICCV 2019)","code":""},{"path":"https://keras3.posit.co/dev/reference/application_mobilenet_v3_small.html","id":"the-following-table-describes-the-performance-of-mobilenets-v-","dir":"Reference","previous_headings":"","what":"The following table describes the performance of MobileNets v3:","title":"Instantiates the MobileNetV3Small architecture. — application_mobilenet_v3_small","text":"MACs stands Multiply Adds image classification use cases, see page detailed examples. transfer learning use cases, make sure read guide transfer learning & fine-tuning.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_mobilenet_v3_small.html","id":"note","dir":"Reference","previous_headings":"","what":"Note","title":"Instantiates the MobileNetV3Small architecture. — application_mobilenet_v3_small","text":"Keras Application expects specific kind input preprocessing. MobileNetV3, default input preprocessing included part model (Rescaling layer), thus application_preprocess_inputs() actually pass-function. use case, MobileNetV3 models expect inputs float tensors pixels values [0-255] range. time, preprocessing part model (.e. Rescaling layer) can disabled setting include_preprocessing argument FALSE. preprocessing disabled MobileNetV3 models expect inputs float tensors pixels values [-1, 1] range.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_mobilenet_v3_small.html","id":"call-arguments","dir":"Reference","previous_headings":"","what":"Call Arguments","title":"Instantiates the MobileNetV3Small architecture. — application_mobilenet_v3_small","text":"inputs: floating point numpy.array backend-native tensor, 4D 3 color channels, values range [0, 255] include_preprocessing TRUE range [-1, 1] otherwise.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/application_nasnet_large.html","id":null,"dir":"Reference","previous_headings":"","what":"Instantiates a NASNet model in ImageNet mode. — application_nasnet_large","title":"Instantiates a NASNet model in ImageNet mode. — application_nasnet_large","text":"Instantiates NASNet model ImageNet mode.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_nasnet_large.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Instantiates a NASNet model in ImageNet mode. — application_nasnet_large","text":"","code":"application_nasnet_large(   input_shape = NULL,   include_top = TRUE,   weights = \"imagenet\",   input_tensor = NULL,   pooling = NULL,   classes = 1000L,   classifier_activation = \"softmax\",   name = \"nasnet_large\" )"},{"path":"https://keras3.posit.co/dev/reference/application_nasnet_large.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Instantiates a NASNet model in ImageNet mode. — application_nasnet_large","text":"input_shape Optional shape tuple, specified include_top FALSE (otherwise input shape (331, 331, 3) NASNetLarge. exactly 3 inputs channels, width height smaller 32. E.g. (224, 224, 3) one valid value. include_top Whether include fully-connected layer top network. weights NULL (random initialization) imagenet (ImageNet weights).  loading imagenet weights, input_shape (331, 331, 3) input_tensor Optional Keras tensor (.e. output keras_input()) use image input model. pooling Optional pooling mode feature extraction include_top FALSE. NULL means output model 4D tensor output last convolutional layer. avg means global average pooling applied output last convolutional layer, thus output model 2D tensor. max means global max pooling applied. classes Optional number classes classify images , specified include_top TRUE, weights argument specified. classifier_activation str callable. activation function use \"top\" layer. Ignored unless include_top=TRUE. Set classifier_activation=NULL return logits \"top\" layer.  loading pretrained weights, classifier_activation can NULL \"softmax\". name name model (string).","code":""},{"path":"https://keras3.posit.co/dev/reference/application_nasnet_large.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Instantiates a NASNet model in ImageNet mode. — application_nasnet_large","text":"Keras model instance.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_nasnet_large.html","id":"reference","dir":"Reference","previous_headings":"","what":"Reference","title":"Instantiates a NASNet model in ImageNet mode. — application_nasnet_large","text":"Learning Transferable Architectures Scalable Image Recognition (CVPR 2018) Optionally loads weights pre-trained ImageNet. Note data format convention used model one specified Keras config ~/.keras/keras.json.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_nasnet_large.html","id":"note","dir":"Reference","previous_headings":"","what":"Note","title":"Instantiates a NASNet model in ImageNet mode. — application_nasnet_large","text":"Keras Application expects specific kind input preprocessing. NASNet, call application_preprocess_inputs() inputs passing model.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/application_nasnet_mobile.html","id":null,"dir":"Reference","previous_headings":"","what":"Instantiates a Mobile NASNet model in ImageNet mode. — application_nasnet_mobile","title":"Instantiates a Mobile NASNet model in ImageNet mode. — application_nasnet_mobile","text":"Instantiates Mobile NASNet model ImageNet mode.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_nasnet_mobile.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Instantiates a Mobile NASNet model in ImageNet mode. — application_nasnet_mobile","text":"","code":"application_nasnet_mobile(   input_shape = NULL,   include_top = TRUE,   weights = \"imagenet\",   input_tensor = NULL,   pooling = NULL,   classes = 1000L,   classifier_activation = \"softmax\",   name = \"nasnet_mobile\" )"},{"path":"https://keras3.posit.co/dev/reference/application_nasnet_mobile.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Instantiates a Mobile NASNet model in ImageNet mode. — application_nasnet_mobile","text":"input_shape Optional shape tuple, specified include_top FALSE (otherwise input shape (224, 224, 3) NASNetMobile exactly 3 inputs channels, width height smaller 32. E.g. (224, 224, 3) one valid value. include_top Whether include fully-connected layer top network. weights NULL (random initialization) imagenet (ImageNet weights). loading imagenet weights, input_shape (224, 224, 3) input_tensor Optional Keras tensor (.e. output keras_input()) use image input model. pooling Optional pooling mode feature extraction include_top FALSE. NULL means output model 4D tensor output last convolutional layer. avg means global average pooling applied output last convolutional layer, thus output model 2D tensor. max means global max pooling applied. classes Optional number classes classify images , specified include_top TRUE, weights argument specified. classifier_activation str callable. activation function use \"top\" layer. Ignored unless include_top=TRUE. Set classifier_activation=NULL return logits \"top\" layer.  loading pretrained weights, classifier_activation can NULL \"softmax\". name name model (string).","code":""},{"path":"https://keras3.posit.co/dev/reference/application_nasnet_mobile.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Instantiates a Mobile NASNet model in ImageNet mode. — application_nasnet_mobile","text":"Keras model instance.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_nasnet_mobile.html","id":"reference","dir":"Reference","previous_headings":"","what":"Reference","title":"Instantiates a Mobile NASNet model in ImageNet mode. — application_nasnet_mobile","text":"Learning Transferable Architectures Scalable Image Recognition (CVPR 2018) Optionally loads weights pre-trained ImageNet. Note data format convention used model one specified Keras config ~/.keras/keras.json.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_nasnet_mobile.html","id":"note","dir":"Reference","previous_headings":"","what":"Note","title":"Instantiates a Mobile NASNet model in ImageNet mode. — application_nasnet_mobile","text":"Keras Application expects specific kind input preprocessing. NASNet, call application_preprocess_inputs() inputs passing model.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/application_resnet101.html","id":null,"dir":"Reference","previous_headings":"","what":"Instantiates the ResNet101 architecture. — application_resnet101","title":"Instantiates the ResNet101 architecture. — application_resnet101","text":"Instantiates ResNet101 architecture.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_resnet101.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Instantiates the ResNet101 architecture. — application_resnet101","text":"","code":"application_resnet101(   include_top = TRUE,   weights = \"imagenet\",   input_tensor = NULL,   input_shape = NULL,   pooling = NULL,   classes = 1000L,   classifier_activation = \"softmax\",   name = \"resnet101\" )"},{"path":"https://keras3.posit.co/dev/reference/application_resnet101.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Instantiates the ResNet101 architecture. — application_resnet101","text":"include_top whether include fully-connected layer top network. weights one NULL (random initialization), \"imagenet\" (pre-training ImageNet), path weights file loaded. input_tensor optional Keras tensor (.e. output keras_input()) use image input model. input_shape optional shape tuple, specified include_top FALSE (otherwise input shape (224, 224, 3) (\"channels_last\" data format) (3, 224, 224) (\"channels_first\" data format). exactly 3 inputs channels, width height smaller 32. E.g. (200, 200, 3) one valid value. pooling Optional pooling mode feature extraction include_top FALSE. NULL means output model 4D tensor output last convolutional block. avg means global average pooling applied output last convolutional block, thus output model 2D tensor. max means global max pooling applied. classes optional number classes classify images , specified include_top TRUE, weights argument specified. classifier_activation str callable. activation function use \"top\" layer. Ignored unless include_top=TRUE. Set classifier_activation=NULL return logits \"top\" layer. loading pretrained weights, classifier_activation can NULL \"softmax\". name name model (string).","code":""},{"path":"https://keras3.posit.co/dev/reference/application_resnet101.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Instantiates the ResNet101 architecture. — application_resnet101","text":"Model instance.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_resnet101.html","id":"reference","dir":"Reference","previous_headings":"","what":"Reference","title":"Instantiates the ResNet101 architecture. — application_resnet101","text":"Deep Residual Learning Image Recognition (CVPR 2015) image classification use cases, see page detailed examples. transfer learning use cases, make sure read guide transfer learning & fine-tuning.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_resnet101.html","id":"note","dir":"Reference","previous_headings":"","what":"Note","title":"Instantiates the ResNet101 architecture. — application_resnet101","text":"Keras Application expects specific kind input preprocessing. ResNet, call application_preprocess_inputs() inputs passing model. application_preprocess_inputs() convert input images RGB BGR, zero-center color channel respect ImageNet dataset, without scaling.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/application_resnet101_v2.html","id":null,"dir":"Reference","previous_headings":"","what":"Instantiates the ResNet101V2 architecture. — application_resnet101_v2","title":"Instantiates the ResNet101V2 architecture. — application_resnet101_v2","text":"Instantiates ResNet101V2 architecture.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_resnet101_v2.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Instantiates the ResNet101V2 architecture. — application_resnet101_v2","text":"","code":"application_resnet101_v2(   include_top = TRUE,   weights = \"imagenet\",   input_tensor = NULL,   input_shape = NULL,   pooling = NULL,   classes = 1000L,   classifier_activation = \"softmax\",   name = \"resnet101v2\" )"},{"path":"https://keras3.posit.co/dev/reference/application_resnet101_v2.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Instantiates the ResNet101V2 architecture. — application_resnet101_v2","text":"include_top whether include fully-connected layer top network. weights one NULL (random initialization), \"imagenet\" (pre-training ImageNet), path weights file loaded. input_tensor optional Keras tensor (.e. output keras_input()) use image input model. input_shape optional shape tuple, specified include_top FALSE (otherwise input shape (224, 224, 3) (\"channels_last\" data format) (3, 224, 224) (\"channels_first\" data format). exactly 3 inputs channels, width height smaller 32. E.g. (200, 200, 3) one valid value. pooling Optional pooling mode feature extraction include_top FALSE. NULL means output model 4D tensor output last convolutional block. avg means global average pooling applied output last convolutional block, thus output model 2D tensor. max means global max pooling applied. classes optional number classes classify images , specified include_top TRUE, weights argument specified. classifier_activation str callable. activation function use \"top\" layer. Ignored unless include_top=TRUE. Set classifier_activation=NULL return logits \"top\" layer. loading pretrained weights, classifier_activation can NULL \"softmax\". name name model (string).","code":""},{"path":"https://keras3.posit.co/dev/reference/application_resnet101_v2.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Instantiates the ResNet101V2 architecture. — application_resnet101_v2","text":"Model instance.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_resnet101_v2.html","id":"reference","dir":"Reference","previous_headings":"","what":"Reference","title":"Instantiates the ResNet101V2 architecture. — application_resnet101_v2","text":"Identity Mappings Deep Residual Networks (CVPR 2016) image classification use cases, see page detailed examples. transfer learning use cases, make sure read guide transfer learning & fine-tuning.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_resnet101_v2.html","id":"note","dir":"Reference","previous_headings":"","what":"Note","title":"Instantiates the ResNet101V2 architecture. — application_resnet101_v2","text":"Keras Application expects specific kind input preprocessing. ResNet, call application_preprocess_inputs() inputs passing model. application_preprocess_inputs() scale input pixels -1 1.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/application_resnet152.html","id":null,"dir":"Reference","previous_headings":"","what":"Instantiates the ResNet152 architecture. — application_resnet152","title":"Instantiates the ResNet152 architecture. — application_resnet152","text":"Instantiates ResNet152 architecture.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_resnet152.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Instantiates the ResNet152 architecture. — application_resnet152","text":"","code":"application_resnet152(   include_top = TRUE,   weights = \"imagenet\",   input_tensor = NULL,   input_shape = NULL,   pooling = NULL,   classes = 1000L,   classifier_activation = \"softmax\",   name = \"resnet152\" )"},{"path":"https://keras3.posit.co/dev/reference/application_resnet152.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Instantiates the ResNet152 architecture. — application_resnet152","text":"include_top whether include fully-connected layer top network. weights one NULL (random initialization), \"imagenet\" (pre-training ImageNet), path weights file loaded. input_tensor optional Keras tensor (.e. output keras_input()) use image input model. input_shape optional shape tuple, specified include_top FALSE (otherwise input shape (224, 224, 3) (\"channels_last\" data format) (3, 224, 224) (\"channels_first\" data format). exactly 3 inputs channels, width height smaller 32. E.g. (200, 200, 3) one valid value. pooling Optional pooling mode feature extraction include_top FALSE. NULL means output model 4D tensor output last convolutional block. avg means global average pooling applied output last convolutional block, thus output model 2D tensor. max means global max pooling applied. classes optional number classes classify images , specified include_top TRUE, weights argument specified. classifier_activation str callable. activation function use \"top\" layer. Ignored unless include_top=TRUE. Set classifier_activation=NULL return logits \"top\" layer. loading pretrained weights, classifier_activation can NULL \"softmax\". name name model (string).","code":""},{"path":"https://keras3.posit.co/dev/reference/application_resnet152.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Instantiates the ResNet152 architecture. — application_resnet152","text":"Model instance.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_resnet152.html","id":"reference","dir":"Reference","previous_headings":"","what":"Reference","title":"Instantiates the ResNet152 architecture. — application_resnet152","text":"Deep Residual Learning Image Recognition (CVPR 2015) image classification use cases, see page detailed examples. transfer learning use cases, make sure read guide transfer learning & fine-tuning.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_resnet152.html","id":"note","dir":"Reference","previous_headings":"","what":"Note","title":"Instantiates the ResNet152 architecture. — application_resnet152","text":"Keras Application expects specific kind input preprocessing. ResNet, call application_preprocess_inputs() inputs passing model. application_preprocess_inputs() convert input images RGB BGR, zero-center color channel respect ImageNet dataset, without scaling.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/application_resnet152_v2.html","id":null,"dir":"Reference","previous_headings":"","what":"Instantiates the ResNet152V2 architecture. — application_resnet152_v2","title":"Instantiates the ResNet152V2 architecture. — application_resnet152_v2","text":"Instantiates ResNet152V2 architecture.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_resnet152_v2.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Instantiates the ResNet152V2 architecture. — application_resnet152_v2","text":"","code":"application_resnet152_v2(   include_top = TRUE,   weights = \"imagenet\",   input_tensor = NULL,   input_shape = NULL,   pooling = NULL,   classes = 1000L,   classifier_activation = \"softmax\",   name = \"resnet152v2\" )"},{"path":"https://keras3.posit.co/dev/reference/application_resnet152_v2.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Instantiates the ResNet152V2 architecture. — application_resnet152_v2","text":"include_top whether include fully-connected layer top network. weights one NULL (random initialization), \"imagenet\" (pre-training ImageNet), path weights file loaded. input_tensor optional Keras tensor (.e. output keras_input()) use image input model. input_shape optional shape tuple, specified include_top FALSE (otherwise input shape (224, 224, 3) (\"channels_last\" data format) (3, 224, 224) (\"channels_first\" data format). exactly 3 inputs channels, width height smaller 32. E.g. (200, 200, 3) one valid value. pooling Optional pooling mode feature extraction include_top FALSE. NULL means output model 4D tensor output last convolutional block. avg means global average pooling applied output last convolutional block, thus output model 2D tensor. max means global max pooling applied. classes optional number classes classify images , specified include_top TRUE, weights argument specified. classifier_activation str callable. activation function use \"top\" layer. Ignored unless include_top=TRUE. Set classifier_activation=NULL return logits \"top\" layer. loading pretrained weights, classifier_activation can NULL \"softmax\". name name model (string).","code":""},{"path":"https://keras3.posit.co/dev/reference/application_resnet152_v2.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Instantiates the ResNet152V2 architecture. — application_resnet152_v2","text":"Model instance.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_resnet152_v2.html","id":"reference","dir":"Reference","previous_headings":"","what":"Reference","title":"Instantiates the ResNet152V2 architecture. — application_resnet152_v2","text":"Identity Mappings Deep Residual Networks (CVPR 2016) image classification use cases, see page detailed examples. transfer learning use cases, make sure read guide transfer learning & fine-tuning.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_resnet152_v2.html","id":"note","dir":"Reference","previous_headings":"","what":"Note","title":"Instantiates the ResNet152V2 architecture. — application_resnet152_v2","text":"Keras Application expects specific kind input preprocessing. ResNet, call application_preprocess_inputs() inputs passing model. application_preprocess_inputs() scale input pixels -1 1.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/application_resnet50.html","id":null,"dir":"Reference","previous_headings":"","what":"Instantiates the ResNet50 architecture. — application_resnet50","title":"Instantiates the ResNet50 architecture. — application_resnet50","text":"Instantiates ResNet50 architecture.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_resnet50.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Instantiates the ResNet50 architecture. — application_resnet50","text":"","code":"application_resnet50(   include_top = TRUE,   weights = \"imagenet\",   input_tensor = NULL,   input_shape = NULL,   pooling = NULL,   classes = 1000L,   classifier_activation = \"softmax\",   name = \"resnet50\" )"},{"path":"https://keras3.posit.co/dev/reference/application_resnet50.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Instantiates the ResNet50 architecture. — application_resnet50","text":"include_top whether include fully-connected layer top network. weights one NULL (random initialization), \"imagenet\" (pre-training ImageNet), path weights file loaded. input_tensor optional Keras tensor (.e. output keras_input()) use image input model. input_shape optional shape tuple, specified include_top FALSE (otherwise input shape (224, 224, 3) (\"channels_last\" data format) (3, 224, 224) (\"channels_first\" data format). exactly 3 inputs channels, width height smaller 32. E.g. (200, 200, 3) one valid value. pooling Optional pooling mode feature extraction include_top FALSE. NULL means output model 4D tensor output last convolutional block. avg means global average pooling applied output last convolutional block, thus output model 2D tensor. max means global max pooling applied. classes optional number classes classify images , specified include_top TRUE, weights argument specified. Defaults 1000. classifier_activation str callable. activation function use \"top\" layer. Ignored unless include_top=TRUE. Set classifier_activation=NULL return logits \"top\" layer. loading pretrained weights, classifier_activation can NULL \"softmax\". name name model (string).","code":""},{"path":"https://keras3.posit.co/dev/reference/application_resnet50.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Instantiates the ResNet50 architecture. — application_resnet50","text":"Model instance.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_resnet50.html","id":"reference","dir":"Reference","previous_headings":"","what":"Reference","title":"Instantiates the ResNet50 architecture. — application_resnet50","text":"Deep Residual Learning Image Recognition (CVPR 2015) image classification use cases, see page detailed examples. transfer learning use cases, make sure read guide transfer learning & fine-tuning.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_resnet50.html","id":"note","dir":"Reference","previous_headings":"","what":"Note","title":"Instantiates the ResNet50 architecture. — application_resnet50","text":"Keras Application expects specific kind input preprocessing. ResNet, call application_preprocess_inputs() inputs passing model. application_preprocess_inputs() convert input images RGB BGR, zero-center color channel respect ImageNet dataset, without scaling.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/application_resnet50_v2.html","id":null,"dir":"Reference","previous_headings":"","what":"Instantiates the ResNet50V2 architecture. — application_resnet50_v2","title":"Instantiates the ResNet50V2 architecture. — application_resnet50_v2","text":"Instantiates ResNet50V2 architecture.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_resnet50_v2.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Instantiates the ResNet50V2 architecture. — application_resnet50_v2","text":"","code":"application_resnet50_v2(   include_top = TRUE,   weights = \"imagenet\",   input_tensor = NULL,   input_shape = NULL,   pooling = NULL,   classes = 1000L,   classifier_activation = \"softmax\",   name = \"resnet50v2\" )"},{"path":"https://keras3.posit.co/dev/reference/application_resnet50_v2.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Instantiates the ResNet50V2 architecture. — application_resnet50_v2","text":"include_top whether include fully-connected layer top network. weights one NULL (random initialization), \"imagenet\" (pre-training ImageNet), path weights file loaded. input_tensor optional Keras tensor (.e. output keras_input()) use image input model. input_shape optional shape tuple, specified include_top FALSE (otherwise input shape (224, 224, 3) (\"channels_last\" data format) (3, 224, 224) (\"channels_first\" data format). exactly 3 inputs channels, width height smaller 32. E.g. (200, 200, 3) one valid value. pooling Optional pooling mode feature extraction include_top FALSE. NULL means output model 4D tensor output last convolutional block. avg means global average pooling applied output last convolutional block, thus output model 2D tensor. max means global max pooling applied. classes optional number classes classify images , specified include_top TRUE, weights argument specified. classifier_activation str callable. activation function use \"top\" layer. Ignored unless include_top=TRUE. Set classifier_activation=NULL return logits \"top\" layer. loading pretrained weights, classifier_activation can NULL \"softmax\". name name model (string).","code":""},{"path":"https://keras3.posit.co/dev/reference/application_resnet50_v2.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Instantiates the ResNet50V2 architecture. — application_resnet50_v2","text":"Model instance.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_resnet50_v2.html","id":"reference","dir":"Reference","previous_headings":"","what":"Reference","title":"Instantiates the ResNet50V2 architecture. — application_resnet50_v2","text":"Identity Mappings Deep Residual Networks (CVPR 2016) image classification use cases, see page detailed examples. transfer learning use cases, make sure read guide transfer learning & fine-tuning.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_resnet50_v2.html","id":"note","dir":"Reference","previous_headings":"","what":"Note","title":"Instantiates the ResNet50V2 architecture. — application_resnet50_v2","text":"Keras Application expects specific kind input preprocessing. ResNet, call application_preprocess_inputs() inputs passing model. application_preprocess_inputs() scale input pixels -1 1.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/application_vgg16.html","id":null,"dir":"Reference","previous_headings":"","what":"Instantiates the VGG16 model. — application_vgg16","title":"Instantiates the VGG16 model. — application_vgg16","text":"Instantiates VGG16 model.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_vgg16.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Instantiates the VGG16 model. — application_vgg16","text":"","code":"application_vgg16(   include_top = TRUE,   weights = \"imagenet\",   input_tensor = NULL,   input_shape = NULL,   pooling = NULL,   classes = 1000L,   classifier_activation = \"softmax\",   name = \"vgg16\" )"},{"path":"https://keras3.posit.co/dev/reference/application_vgg16.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Instantiates the VGG16 model. — application_vgg16","text":"include_top whether include 3 fully-connected layers top network. weights one NULL (random initialization), \"imagenet\" (pre-training ImageNet), path weights file loaded. input_tensor optional Keras tensor (.e. output keras_input()) use image input model. input_shape optional shape tuple, specified include_top FALSE (otherwise input shape (224, 224, 3) (channels_last data format) (3, 224, 224) (\"channels_first\" data format). exactly 3 input channels, width height smaller 32. E.g. (200, 200, 3) one valid value. pooling Optional pooling mode feature extraction include_top FALSE. NULL means output model 4D tensor output last convolutional block. avg means global average pooling applied output last convolutional block, thus output model 2D tensor. max means global max pooling applied. classes optional number classes classify images , specified include_top TRUE, weights argument specified. classifier_activation str callable. activation function use \"top\" layer. Ignored unless include_top=TRUE. Set classifier_activation=NULL return logits \"top\" layer.  loading pretrained weights, classifier_activation can NULL \"softmax\". name name model (string).","code":""},{"path":"https://keras3.posit.co/dev/reference/application_vgg16.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Instantiates the VGG16 model. — application_vgg16","text":"Model instance.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_vgg16.html","id":"reference","dir":"Reference","previous_headings":"","what":"Reference","title":"Instantiates the VGG16 model. — application_vgg16","text":"Deep Convolutional Networks Large-Scale Image Recognition (ICLR 2015) image classification use cases, see page detailed examples. transfer learning use cases, make sure read guide transfer learning & fine-tuning. default input size model 224x224.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_vgg16.html","id":"note","dir":"Reference","previous_headings":"","what":"Note","title":"Instantiates the VGG16 model. — application_vgg16","text":"Keras Application expects specific kind input preprocessing. VGG16, call application_preprocess_inputs() inputs passing model. application_preprocess_inputs() convert input images RGB BGR, zero-center color channel respect ImageNet dataset, without scaling.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/application_vgg19.html","id":null,"dir":"Reference","previous_headings":"","what":"Instantiates the VGG19 model. — application_vgg19","title":"Instantiates the VGG19 model. — application_vgg19","text":"Instantiates VGG19 model.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_vgg19.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Instantiates the VGG19 model. — application_vgg19","text":"","code":"application_vgg19(   include_top = TRUE,   weights = \"imagenet\",   input_tensor = NULL,   input_shape = NULL,   pooling = NULL,   classes = 1000L,   classifier_activation = \"softmax\",   name = \"vgg19\" )"},{"path":"https://keras3.posit.co/dev/reference/application_vgg19.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Instantiates the VGG19 model. — application_vgg19","text":"include_top whether include 3 fully-connected layers top network. weights one NULL (random initialization), \"imagenet\" (pre-training ImageNet), path weights file loaded. input_tensor optional Keras tensor (.e. output keras_input()) use image input model. input_shape optional shape tuple, specified include_top FALSE (otherwise input shape (224, 224, 3) (channels_last data format) (3, 224, 224) (\"channels_first\" data format). exactly 3 input channels, width height smaller 32. E.g. (200, 200, 3) one valid value. pooling Optional pooling mode feature extraction include_top FALSE. NULL means output model 4D tensor output last convolutional block. avg means global average pooling applied output last convolutional block, thus output model 2D tensor. max means global max pooling applied. classes optional number classes classify images , specified include_top TRUE, weights argument specified. classifier_activation str callable. activation function use \"top\" layer. Ignored unless include_top=TRUE. Set classifier_activation=NULL return logits \"top\" layer.  loading pretrained weights, classifier_activation can NULL \"softmax\". name name model (string).","code":""},{"path":"https://keras3.posit.co/dev/reference/application_vgg19.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Instantiates the VGG19 model. — application_vgg19","text":"model instance.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_vgg19.html","id":"reference","dir":"Reference","previous_headings":"","what":"Reference","title":"Instantiates the VGG19 model. — application_vgg19","text":"Deep Convolutional Networks Large-Scale Image Recognition (ICLR 2015) image classification use cases, see page detailed examples. transfer learning use cases, make sure read guide transfer learning & fine-tuning. default input size model 224x224.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_vgg19.html","id":"note","dir":"Reference","previous_headings":"","what":"Note","title":"Instantiates the VGG19 model. — application_vgg19","text":"Keras Application expects specific kind input preprocessing. VGG19, call application_preprocess_inputs() inputs passing model. application_preprocess_inputs() convert input images RGB BGR, zero-center color channel respect ImageNet dataset, without scaling.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/application_xception.html","id":null,"dir":"Reference","previous_headings":"","what":"Instantiates the Xception architecture. — application_xception","title":"Instantiates the Xception architecture. — application_xception","text":"Instantiates Xception architecture.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_xception.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Instantiates the Xception architecture. — application_xception","text":"","code":"application_xception(   include_top = TRUE,   weights = \"imagenet\",   input_tensor = NULL,   input_shape = NULL,   pooling = NULL,   classes = 1000L,   classifier_activation = \"softmax\",   name = \"xception\" )"},{"path":"https://keras3.posit.co/dev/reference/application_xception.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Instantiates the Xception architecture. — application_xception","text":"include_top whether include 3 fully-connected layers top network. weights one NULL (random initialization), \"imagenet\" (pre-training ImageNet), path weights file loaded. input_tensor optional Keras tensor (.e. output keras_input()) use image input model. input_shape optional shape tuple, specified include_top FALSE (otherwise input shape (299, 299, 3). exactly 3 inputs channels, width height smaller 71. E.g. (150, 150, 3) one valid value. pooling Optional pooling mode feature extraction include_top FALSE. NULL means output model 4D tensor output last convolutional block. avg means global average pooling applied output last convolutional block, thus output model 2D tensor. max means global max pooling applied. classes optional number classes classify images , specified include_top TRUE, weights argument specified. classifier_activation str callable. activation function use \"top\" layer. Ignored unless include_top=TRUE. Set classifier_activation=NULL return logits \"top\" layer.  loading pretrained weights, classifier_activation can NULL \"softmax\". name name model (string).","code":""},{"path":"https://keras3.posit.co/dev/reference/application_xception.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Instantiates the Xception architecture. — application_xception","text":"model instance.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_xception.html","id":"reference","dir":"Reference","previous_headings":"","what":"Reference","title":"Instantiates the Xception architecture. — application_xception","text":"Xception: Deep Learning Depthwise Separable Convolutions (CVPR 2017) image classification use cases, see page detailed examples. transfer learning use cases, make sure read guide transfer learning & fine-tuning. default input image size model 299x299.","code":""},{"path":"https://keras3.posit.co/dev/reference/application_xception.html","id":"note","dir":"Reference","previous_headings":"","what":"Note","title":"Instantiates the Xception architecture. — application_xception","text":"Keras Application expects specific kind input preprocessing. Xception, call application_preprocess_inputs() inputs passing model. application_preprocess_inputs() scale input pixels -1 1.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/audio_dataset_from_directory.html","id":null,"dir":"Reference","previous_headings":"","what":"Generates a tf.data.Dataset from audio files in a directory. — audio_dataset_from_directory","title":"Generates a tf.data.Dataset from audio files in a directory. — audio_dataset_from_directory","text":"directory structure :   calling audio_dataset_from_directory(main_directory, labels = 'inferred') return tf.data.Dataset yields batches audio files subdirectories class_a class_b, together labels 0 1 (0 corresponding class_a 1 corresponding class_b). .wav files supported time.","code":"main_directory/ ...class_a/ ......a_audio_1.wav ......a_audio_2.wav ...class_b/ ......b_audio_1.wav ......b_audio_2.wav"},{"path":"https://keras3.posit.co/dev/reference/audio_dataset_from_directory.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Generates a tf.data.Dataset from audio files in a directory. — audio_dataset_from_directory","text":"","code":"audio_dataset_from_directory(   directory,   labels = \"inferred\",   label_mode = \"int\",   class_names = NULL,   batch_size = 32L,   sampling_rate = NULL,   output_sequence_length = NULL,   ragged = FALSE,   shuffle = TRUE,   seed = NULL,   validation_split = NULL,   subset = NULL,   follow_links = FALSE,   verbose = TRUE )"},{"path":"https://keras3.posit.co/dev/reference/audio_dataset_from_directory.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Generates a tf.data.Dataset from audio files in a directory. — audio_dataset_from_directory","text":"directory Directory data located. labels \"inferred\", contain subdirectories, containing audio files class. Otherwise, directory structure ignored. labels Either \"inferred\" (labels generated directory structure), NULL (labels), list/tuple integer labels size number audio files found directory. Labels sorted according alphanumeric order audio file paths (obtained via os.walk(directory) Python). label_mode String describing encoding labels. Options : \"int\": means labels encoded integers (e.g. sparse_categorical_crossentropy loss). \"categorical\" means labels encoded categorical vector (e.g. categorical_crossentropy loss) \"binary\" means labels (can 2) encoded float32 scalars values 0 1 (e.g. binary_crossentropy). NULL (labels). class_names valid \"labels\" \"inferred\". explicit list class names (must match names subdirectories). Used control order classes (otherwise alphanumerical order used). batch_size Size batches data. Default: 32. NULL, data batched (dataset yield individual samples). sampling_rate Audio sampling rate (samples per second). output_sequence_length Maximum length audio sequence. Audio files longer truncated output_sequence_length. set NULL, sequences batch padded length longest sequence batch. ragged Whether return Ragged dataset (sequence length). Defaults FALSE. shuffle Whether shuffle data. Defaults TRUE. set FALSE, sorts data alphanumeric order. seed Optional random seed shuffling transformations. validation_split Optional float 0 1, fraction data reserve validation. subset Subset data return. One \"training\", \"validation\" \"\". used validation_split set. follow_links Whether visits subdirectories pointed symlinks. Defaults FALSE. verbose Whether display number information classes number files found. Defaults TRUE.","code":""},{"path":"https://keras3.posit.co/dev/reference/audio_dataset_from_directory.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Generates a tf.data.Dataset from audio files in a directory. — audio_dataset_from_directory","text":"tf.data.Dataset object. label_mode NULL, yields string tensors shape (batch_size,), containing contents batch audio files. Otherwise, yields tuple (audio, labels), audio shape (batch_size, sequence_length, num_channels) labels follows format described . Rules regarding labels format: label_mode int, labels int32 tensor shape (batch_size,). label_mode binary, labels float32 tensor 1s 0s shape (batch_size, 1). label_mode categorical, labels float32 tensor shape (batch_size, num_classes), representing one-hot encoding class index.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/back-compat.html","id":null,"dir":"Reference","previous_headings":"","what":"Backward compatibility — application_nasnetlarge","title":"Backward compatibility — application_nasnetlarge","text":"Backward compatibility Backward compatibility","code":""},{"path":"https://keras3.posit.co/dev/reference/back-compat.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Backward compatibility — application_nasnetlarge","text":"","code":"application_nasnetlarge(...)  application_nasnetmobile(...)"},{"path":"https://keras3.posit.co/dev/reference/back-compat.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Backward compatibility — application_nasnetlarge","text":"... passed application_nasnet_large() application_nasnet_mobile().","code":""},{"path":"https://keras3.posit.co/dev/reference/bidirectional.html","id":null,"dir":"Reference","previous_headings":"","what":"layer_bidirectional — bidirectional","title":"layer_bidirectional — bidirectional","text":"bidirectional() alias layer_bidirectional(). See ?layer_bidirectional() full documentation.","code":""},{"path":"https://keras3.posit.co/dev/reference/bidirectional.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"layer_bidirectional — bidirectional","text":"","code":"bidirectional(   object,   layer,   merge_mode = \"concat\",   weights = NULL,   backward_layer = NULL,   ... )"},{"path":"https://keras3.posit.co/dev/reference/bidirectional.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"layer_bidirectional — bidirectional","text":"object Object compose layer . tensor, array, sequential model. layer RNN instance, layer_lstm() layer_gru(). also Layer() instance meets following criteria: sequence-processing layer (accepts 3D+ inputs). go_backwards, return_sequences return_state attribute (semantics RNN class). input_spec attribute. Implement serialization via get_config() from_config(). Note recommended way create new RNN layers write custom RNN cell use layer_rnn(), instead subclassing Layer() directly. return_sequences TRUE, output masked timestep zero regardless layer's original zero_output_for_mask value. merge_mode Mode outputs forward backward RNNs combined. One {\"sum\", \"mul\", \"concat\", \"ave\", NULL}. NULL, outputs combined, returned list. Defaults \"concat\". weights see description backward_layer Optional RNN, Layer() instance used handle backwards input processing. backward_layer provided, layer instance passed layer argument used generate backward layer automatically. Note provided backward_layer layer properties matching layer argument, particular values stateful, return_states, return_sequences, etc. addition, backward_layer layer different go_backwards argument values. ValueError raised requirements met. ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/bidirectional.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"layer_bidirectional — bidirectional","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/callback_backup_and_restore.html","id":null,"dir":"Reference","previous_headings":"","what":"Callback to back up and restore the training state. — callback_backup_and_restore","title":"Callback to back up and restore the training state. — callback_backup_and_restore","text":"callback_backup_and_restore() callback intended recover training interruption happened middle fit execution, backing training states temporary checkpoint file, end epoch. backup overwrites previously written checkpoint file, given time one checkpoint file backup/restoring purpose. training restarts completion, training state (includes model weights epoch number) restored recently saved state beginning new fit run. completion fit run, temporary checkpoint file deleted. Note user responsible bring jobs back interruption. callback important backup restore mechanism fault tolerance purpose, model restored previous checkpoint expected one used back . user changes arguments passed compile fit, checkpoint saved fault tolerance can become invalid.","code":""},{"path":"https://keras3.posit.co/dev/reference/callback_backup_and_restore.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Callback to back up and restore the training state. — callback_backup_and_restore","text":"","code":"callback_backup_and_restore(   backup_dir,   save_freq = \"epoch\",   double_checkpoint = FALSE,   delete_checkpoint = TRUE )"},{"path":"https://keras3.posit.co/dev/reference/callback_backup_and_restore.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Callback to back up and restore the training state. — callback_backup_and_restore","text":"backup_dir String, path directory store data needed restore model. directory reused elsewhere store files, e.g. backup_and_restore callback another training run, another callback (e.g. callback_model_checkpoint) training run. save_freq \"epoch\", integer, FALSE. set \"epoch\", callback saves checkpoint end epoch. set integer, callback saves checkpoint every save_freq batches. Set save_freq = FALSE using preemption checkpointing (.e. save_before_preemption = TRUE). double_checkpoint Boolean. enabled, BackupAndRestore callback save 2 last training states (current previous). interruption current state loaded due IO error (e.g. file corrupted) try restore previous one. behaviour consume twice space disk, increase fault tolerance. Defaults FALSE. delete_checkpoint Boolean. backup_and_restore callback works saving checkpoint back training state. delete_checkpoint = TRUE, checkpoint deleted training finished. Use FALSE like keep checkpoint future usage. Defaults TRUE.","code":""},{"path":"https://keras3.posit.co/dev/reference/callback_backup_and_restore.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Callback to back up and restore the training state. — callback_backup_and_restore","text":"Callback instance can passed fit.keras.src.models.model.Model().","code":""},{"path":"https://keras3.posit.co/dev/reference/callback_backup_and_restore.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Callback to back up and restore the training state. — callback_backup_and_restore","text":"","code":"callback_interrupting <- new_callback_class(   \"InterruptingCallback\",   on_epoch_begin = function(epoch, logs = NULL) {     if (epoch == 4) {       stop('Interrupting!')     }   } )  backup_dir <- tempfile() callback <- callback_backup_and_restore(backup_dir = backup_dir) model <- keras_model_sequential() %>%   layer_dense(10) model %>% compile(optimizer = optimizer_sgd(), loss = 'mse')  # ensure model is built (i.e., weights are initialized) for # callback_backup_and_restore() model(op_ones(c(5, 20))) |> invisible()  tryCatch({   model %>% fit(x = op_ones(c(5, 20)),                 y = op_zeros(5),                 epochs = 10, batch_size = 1,                 callbacks = list(callback, callback_interrupting()),                 verbose = 0) }, python.builtin.RuntimeError = function(e) message(\"Interrupted!\")) ## Interrupted! model$history$epoch ## [1] 0 1 2 # model$history %>% keras3:::to_keras_training_history() %>% as.data.frame() %>% print()  history <- model %>% fit(x = op_ones(c(5, 20)),                          y = op_zeros(5),                          epochs = 10, batch_size = 1,                          callbacks = list(callback),                          verbose = 0)  # Only 6 more epochs are run, since first training got interrupted at # zero-indexed epoch 4, second training will continue from 4 to 9. nrow(as.data.frame(history)) ## [1] 10"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/callback_csv_logger.html","id":null,"dir":"Reference","previous_headings":"","what":"Callback that streams epoch results to a CSV file. — callback_csv_logger","title":"Callback that streams epoch results to a CSV file. — callback_csv_logger","text":"Supports values can represented string, including 1D iterables atomic vectors.","code":""},{"path":"https://keras3.posit.co/dev/reference/callback_csv_logger.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Callback that streams epoch results to a CSV file. — callback_csv_logger","text":"","code":"callback_csv_logger(filename, separator = \",\", append = FALSE)"},{"path":"https://keras3.posit.co/dev/reference/callback_csv_logger.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Callback that streams epoch results to a CSV file. — callback_csv_logger","text":"filename Filename CSV file, e.g. 'run/log.csv'. separator String used separate elements CSV file. append Boolean. TRUE: append file exists (useful continuing training). FALSE: overwrite existing file.","code":""},{"path":"https://keras3.posit.co/dev/reference/callback_csv_logger.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Callback that streams epoch results to a CSV file. — callback_csv_logger","text":"Callback instance can passed fit.keras.src.models.model.Model().","code":""},{"path":"https://keras3.posit.co/dev/reference/callback_csv_logger.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Callback that streams epoch results to a CSV file. — callback_csv_logger","text":"","code":"csv_logger <- callback_csv_logger('training.log') model %>% fit(X_train, Y_train, callbacks = list(csv_logger))"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/callback_early_stopping.html","id":null,"dir":"Reference","previous_headings":"","what":"Stop training when a monitored metric has stopped improving. — callback_early_stopping","title":"Stop training when a monitored metric has stopped improving. — callback_early_stopping","text":"Assuming goal training minimize loss. , metric monitored 'loss', mode 'min'. model$fit() training loop check end every epoch whether loss longer decreasing, considering min_delta patience applicable. found longer decreasing, model$stop_training marked TRUE training terminates. quantity monitored needs available logs list. make , pass loss metrics model$compile().","code":""},{"path":"https://keras3.posit.co/dev/reference/callback_early_stopping.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Stop training when a monitored metric has stopped improving. — callback_early_stopping","text":"","code":"callback_early_stopping(   monitor = \"val_loss\",   min_delta = 0L,   patience = 0L,   verbose = 0L,   mode = \"auto\",   baseline = NULL,   restore_best_weights = FALSE,   start_from_epoch = 0L )"},{"path":"https://keras3.posit.co/dev/reference/callback_early_stopping.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Stop training when a monitored metric has stopped improving. — callback_early_stopping","text":"monitor Quantity monitored. Defaults \"val_loss\". min_delta Minimum change monitored quantity qualify improvement, .e. absolute change less min_delta, count improvement. Defaults 0. patience Number epochs improvement training stopped. Defaults 0. verbose Verbosity mode, 0 1. Mode 0 silent, mode 1 displays messages callback takes action. Defaults 0. mode One {\"auto\", \"min\", \"max\"}. min mode, training stop quantity monitored stopped decreasing; \"max\" mode stop quantity monitored stopped increasing; \"auto\" mode, direction automatically inferred name monitored quantity. Defaults \"auto\". baseline Baseline value monitored quantity. NULL, training stop model show improvement baseline. Defaults NULL. restore_best_weights Whether restore model weights epoch best value monitored quantity. FALSE, model weights obtained last step training used. epoch restored regardless performance relative baseline. epoch improves baseline, training run patience epochs restore weights best epoch set. Defaults FALSE. start_from_epoch Number epochs wait starting monitor improvement. allows warm-period improvement expected thus training stopped. Defaults 0.","code":""},{"path":"https://keras3.posit.co/dev/reference/callback_early_stopping.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Stop training when a monitored metric has stopped improving. — callback_early_stopping","text":"Callback instance can passed fit.keras.src.models.model.Model().","code":""},{"path":"https://keras3.posit.co/dev/reference/callback_early_stopping.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Stop training when a monitored metric has stopped improving. — callback_early_stopping","text":"","code":"callback <- callback_early_stopping(monitor = 'loss',                                    patience = 3) # This callback will stop the training when there is no improvement in # the loss for three consecutive epochs. model <- keras_model_sequential() %>%   layer_dense(10) model %>% compile(optimizer = optimizer_sgd(), loss = 'mse') history <- model %>% fit(x = op_ones(c(5, 20)),                          y = op_zeros(5),                          epochs = 10, batch_size = 1,                          callbacks = list(callback),                          verbose = 0) nrow(as.data.frame(history))  # Only 4 epochs are run. ## [1] 10"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/callback_lambda.html","id":null,"dir":"Reference","previous_headings":"","what":"Callback for creating simple, custom callbacks on-the-fly. — callback_lambda","title":"Callback for creating simple, custom callbacks on-the-fly. — callback_lambda","text":"callback constructed anonymous functions called appropriate time (Model.{fit | evaluate | predict}). Note callbacks expects positional arguments, : on_epoch_begin on_epoch_end expect two positional arguments: epoch, logs on_train_begin on_train_end expect one positional argument: logs on_train_batch_begin on_train_batch_end expect positional argument batch named argument logs See Callback class definition full list functions expected arguments.","code":""},{"path":"https://keras3.posit.co/dev/reference/callback_lambda.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Callback for creating simple, custom callbacks on-the-fly. — callback_lambda","text":"","code":"callback_lambda(   on_epoch_begin = NULL,   on_epoch_end = NULL,   on_train_begin = NULL,   on_train_end = NULL,   on_train_batch_begin = NULL,   on_train_batch_end = NULL,   ... )"},{"path":"https://keras3.posit.co/dev/reference/callback_lambda.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Callback for creating simple, custom callbacks on-the-fly. — callback_lambda","text":"on_epoch_begin called beginning every epoch. on_epoch_end called end every epoch. on_train_begin called beginning model training. on_train_end called end model training. on_train_batch_begin called beginning every train batch. on_train_batch_end called end every train batch. ... function Callback() want override passing function_name = function. example, callback_lambda(.., on_train_end = train_end_fn). custom function needs arguments ones defined Callback().","code":""},{"path":"https://keras3.posit.co/dev/reference/callback_lambda.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Callback for creating simple, custom callbacks on-the-fly. — callback_lambda","text":"Callback instance can passed fit.keras.src.models.model.Model().","code":""},{"path":"https://keras3.posit.co/dev/reference/callback_lambda.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Callback for creating simple, custom callbacks on-the-fly. — callback_lambda","text":"","code":"# Print the batch number at the beginning of every batch. batch_print_callback <- callback_lambda(   on_train_batch_begin = function(batch, logs) {     print(batch)   } )  # Stream the epoch loss to a file in new-line delimited JSON format # (one valid JSON object per line) json_log <- file('loss_log.json', open = 'wt') json_logging_callback <- callback_lambda(   on_epoch_end = function(epoch, logs) {     jsonlite::write_json(       list(epoch = epoch, loss = logs$loss),       json_log,       append = TRUE     )   },   on_train_end = function(logs) {     close(json_log)   } )  # Terminate some processes after having finished model training. processes <- ... cleanup_callback <- callback_lambda(   on_train_end = function(logs) {     for (p in processes) {       if (is_alive(p)) {         terminate(p)       }     }   } )  model %>% fit(   ...,   callbacks = list(     batch_print_callback,     json_logging_callback,     cleanup_callback   ) )"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/callback_learning_rate_scheduler.html","id":null,"dir":"Reference","previous_headings":"","what":"Learning rate scheduler. — callback_learning_rate_scheduler","title":"Learning rate scheduler. — callback_learning_rate_scheduler","text":"beginning every epoch, callback gets updated learning rate value schedule function provided, current epoch current learning rate, applies updated learning rate optimizer.","code":""},{"path":"https://keras3.posit.co/dev/reference/callback_learning_rate_scheduler.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Learning rate scheduler. — callback_learning_rate_scheduler","text":"","code":"callback_learning_rate_scheduler(schedule, verbose = 0L)"},{"path":"https://keras3.posit.co/dev/reference/callback_learning_rate_scheduler.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Learning rate scheduler. — callback_learning_rate_scheduler","text":"schedule function takes epoch index (integer, indexed 0) current learning rate (float) inputs returns new learning rate output (float). verbose Integer. 0: quiet, 1: log update messages.","code":""},{"path":"https://keras3.posit.co/dev/reference/callback_learning_rate_scheduler.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Learning rate scheduler. — callback_learning_rate_scheduler","text":"Callback instance can passed fit.keras.src.models.model.Model().","code":""},{"path":"https://keras3.posit.co/dev/reference/callback_learning_rate_scheduler.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Learning rate scheduler. — callback_learning_rate_scheduler","text":"","code":"# This function keeps the initial learning rate steady for the first ten epochs # and decreases it exponentially after that. scheduler <- function(epoch, lr) {   if (epoch < 10)     return(lr)   else     return(lr * exp(-0.1)) }  model <- keras_model_sequential() |> layer_dense(units = 10) model |> compile(optimizer = optimizer_sgd(), loss = 'mse') model$optimizer$learning_rate |> as.array() |> round(5) ## [1] 0.01 callback <- callback_learning_rate_scheduler(schedule = scheduler) history <- model |> fit(x = array(runif(100), c(5, 20)),                         y = array(0, c(5, 1)),                         epochs = 15, callbacks = list(callback), verbose = 0) model$optimizer$learning_rate |> as.array() |> round(5) ## [1] 0.00607"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/callback_model_checkpoint.html","id":null,"dir":"Reference","previous_headings":"","what":"Callback to save the Keras model or model weights at some frequency. — callback_model_checkpoint","title":"Callback to save the Keras model or model weights at some frequency. — callback_model_checkpoint","text":"callback_model_checkpoint() used conjunction training using model |> fit() save model weights (checkpoint file) interval, model weights can loaded later continue training state saved. options callback provides include: Whether keep model achieved \"best performance\" far, whether save model end every epoch regardless performance. Definition \"best\"; quantity monitor whether maximized minimized. frequency save . Currently, callback supports saving end every epoch, fixed number training batches. Whether weights saved, whole model saved.","code":""},{"path":"https://keras3.posit.co/dev/reference/callback_model_checkpoint.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Callback to save the Keras model or model weights at some frequency. — callback_model_checkpoint","text":"","code":"callback_model_checkpoint(   filepath,   monitor = \"val_loss\",   verbose = 0L,   save_best_only = FALSE,   save_weights_only = FALSE,   mode = \"auto\",   save_freq = \"epoch\",   initial_value_threshold = NULL )"},{"path":"https://keras3.posit.co/dev/reference/callback_model_checkpoint.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Callback to save the Keras model or model weights at some frequency. — callback_model_checkpoint","text":"filepath string, path save model file. filepath can contain named formatting options, filled value epoch keys logs (passed on_epoch_end). filepath name needs end \".weights.h5\" save_weights_only = TRUE end \".keras\" \".h5\" checkpoint saving whole model (default). example: filepath \"{epoch:02d}-{val_loss:.2f}.keras\" \"{epoch:02d}-{val_loss:.2f}.weights.h5\", model checkpoints saved epoch number validation loss filename. directory filepath reused callbacks avoid conflicts. monitor metric name monitor. Typically metrics set model |> compile() method. Note: Prefix name \"val_\" monitor validation metrics. Use \"loss\" \"val_loss\" monitor model's total loss. specify metrics strings, like \"accuracy\", pass string (without \"val_\" prefix). pass Metric objects (created one metric_*()), monitor set metric$name. sure metric names can check contents history$metrics list returned history <- model |> fit() Multi-output models set additional prefixes metric names. verbose Verbosity mode, 0 1. Mode 0 silent, mode 1 displays messages callback takes action. save_best_only save_best_only = TRUE, saves model considered \"best\" latest best model according quantity monitored overwritten. filepath contain formatting options like {epoch} filepath overwritten new better model. save_weights_only TRUE, model's weights saved (model |> save_model_weights(filepath)), else full model saved (model |> save_model(filepath)). mode one {\"auto\", \"min\", \"max\"}. save_best_only = TRUE, decision overwrite current save file made based either maximization minimization monitored quantity. val_acc, \"max\", val_loss \"min\", etc. \"auto\" mode, direction automatically inferred name monitored quantity. save_freq \"epoch\" integer. using \"epoch\", callback saves model epoch. using integer, callback saves model end many batches. Model compiled steps_per_execution = N, saving criteria checked every Nth batch. Note saving aligned epochs, monitored metric may potentially less reliable (reflect little 1 batch, since metrics get reset every epoch). Defaults \"epoch\". initial_value_threshold Floating point initial \"best\" value metric monitored. applies save_best_value = TRUE. overwrites model weights already saved performance current model better value.","code":""},{"path":"https://keras3.posit.co/dev/reference/callback_model_checkpoint.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Callback to save the Keras model or model weights at some frequency. — callback_model_checkpoint","text":"Callback instance can passed fit.keras.src.models.model.Model().","code":""},{"path":"https://keras3.posit.co/dev/reference/callback_model_checkpoint.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Callback to save the Keras model or model weights at some frequency. — callback_model_checkpoint","text":"","code":"model <- keras_model_sequential(input_shape = c(10)) |>   layer_dense(1, activation = \"sigmoid\") |>   compile(loss = \"binary_crossentropy\", optimizer = \"adam\",           metrics = c('accuracy'))  EPOCHS <- 10 checkpoint_filepath <- tempfile('checkpoint-model-', fileext = \".keras\") model_checkpoint_callback <- callback_model_checkpoint(   filepath = checkpoint_filepath,   monitor = 'val_accuracy',   mode = 'max',   save_best_only = TRUE )  # Model is saved at the end of every epoch, if it's the best seen so far. model |> fit(x = random_uniform(c(2, 10)), y = op_ones(2, 1),              epochs = EPOCHS, validation_split = .5, verbose = 0,              callbacks = list(model_checkpoint_callback))  # The model (that are considered the best) can be loaded as - load_model(checkpoint_filepath) ## Model: \"sequential\" ## +---------------------------------+------------------------+---------------+ ## | Layer (type)                    | Output Shape           |       Param # | ## +=================================+========================+===============+ ## | dense (Dense)                   | (None, 1)              |            11 | ## +---------------------------------+------------------------+---------------+ ##  Total params: 35 (144.00 B) ##  Trainable params: 11 (44.00 B) ##  Non-trainable params: 0 (0.00 B) ##  Optimizer params: 24 (100.00 B) # Alternatively, one could checkpoint just the model weights as - checkpoint_filepath <- tempfile('checkpoint-', fileext = \".weights.h5\") model_checkpoint_callback <- callback_model_checkpoint(   filepath = checkpoint_filepath,   save_weights_only = TRUE,   monitor = 'val_accuracy',   mode = 'max',   save_best_only = TRUE )  # Model weights are saved at the end of every epoch, if it's the best seen # so far. # same as above model |> fit(x = random_uniform(c(2, 10)), y = op_ones(2, 1),              epochs = EPOCHS, validation_split = .5, verbose = 0,              callbacks = list(model_checkpoint_callback))  # The model weights (that are considered the best) can be loaded model |> load_model_weights(checkpoint_filepath)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/callback_reduce_lr_on_plateau.html","id":null,"dir":"Reference","previous_headings":"","what":"Reduce learning rate when a metric has stopped improving. — callback_reduce_lr_on_plateau","title":"Reduce learning rate when a metric has stopped improving. — callback_reduce_lr_on_plateau","text":"Models often benefit reducing learning rate factor 2-10 learning stagnates. callback monitors quantity improvement seen 'patience' number epochs, learning rate reduced.","code":""},{"path":"https://keras3.posit.co/dev/reference/callback_reduce_lr_on_plateau.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Reduce learning rate when a metric has stopped improving. — callback_reduce_lr_on_plateau","text":"","code":"callback_reduce_lr_on_plateau(   monitor = \"val_loss\",   factor = 0.1,   patience = 10L,   verbose = 0L,   mode = \"auto\",   min_delta = 1e-04,   cooldown = 0L,   min_lr = 0,   ... )"},{"path":"https://keras3.posit.co/dev/reference/callback_reduce_lr_on_plateau.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Reduce learning rate when a metric has stopped improving. — callback_reduce_lr_on_plateau","text":"monitor String. Quantity monitored. factor Float. Factor learning rate reduced. new_lr = lr * factor. patience Integer. Number epochs improvement learning rate reduced. verbose Integer. 0: quiet, 1: update messages. mode String. One {'auto', 'min', 'max'}. 'min' mode, learning rate reduced quantity monitored stopped decreasing; 'max' mode reduced quantity monitored stopped increasing; 'auto' mode, direction automatically inferred name monitored quantity. min_delta Float. Threshold measuring new optimum, focus significant changes. cooldown Integer. Number epochs wait resuming normal operation learning rate reduced. min_lr Float. Lower bound learning rate. ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/callback_reduce_lr_on_plateau.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Reduce learning rate when a metric has stopped improving. — callback_reduce_lr_on_plateau","text":"Callback instance can passed fit.keras.src.models.model.Model().","code":""},{"path":"https://keras3.posit.co/dev/reference/callback_reduce_lr_on_plateau.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Reduce learning rate when a metric has stopped improving. — callback_reduce_lr_on_plateau","text":"","code":"reduce_lr <- callback_reduce_lr_on_plateau(monitor = 'val_loss', factor = 0.2,                                            patience = 5, min_lr = 0.001) model %>% fit(x_train, y_train, callbacks = list(reduce_lr))"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/callback_remote_monitor.html","id":null,"dir":"Reference","previous_headings":"","what":"Callback used to stream events to a server. — callback_remote_monitor","title":"Callback used to stream events to a server. — callback_remote_monitor","text":"Requires requests library. Events sent root + '/publish/epoch/end/' default. Calls HTTP POST, data argument JSON-encoded named list event data. send_as_json = TRUE, content type request \"application/json\". Otherwise serialized JSON sent within form.","code":""},{"path":"https://keras3.posit.co/dev/reference/callback_remote_monitor.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Callback used to stream events to a server. — callback_remote_monitor","text":"","code":"callback_remote_monitor(   root = \"http://localhost:9000\",   path = \"/publish/epoch/end/\",   field = \"data\",   headers = NULL,   send_as_json = FALSE )"},{"path":"https://keras3.posit.co/dev/reference/callback_remote_monitor.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Callback used to stream events to a server. — callback_remote_monitor","text":"root String; root url target server. path String; path relative root events sent. field String; JSON field data stored. field used payload sent within form (.e. send_as_json = FALSE). headers Named list; optional custom HTTP headers. send_as_json Boolean; whether request sent \"application/json\".","code":""},{"path":"https://keras3.posit.co/dev/reference/callback_remote_monitor.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Callback used to stream events to a server. — callback_remote_monitor","text":"Callback instance can passed fit.keras.src.models.model.Model().","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/callback_swap_ema_weights.html","id":null,"dir":"Reference","previous_headings":"","what":"Swaps model weights and EMA weights before and after evaluation. — callback_swap_ema_weights","title":"Swaps model weights and EMA weights before and after evaluation. — callback_swap_ema_weights","text":"callbacks replaces model's weight values values optimizer's EMA weights (exponential moving average past model weights values, implementing \"Polyak averaging\") model evaluation, restores previous weights evaluation. SwapEMAWeights callback used conjunction optimizer sets use_ema = TRUE. Note weights swapped -place order save memory. behavior undefined modify EMA weights model weights callbacks.","code":""},{"path":"https://keras3.posit.co/dev/reference/callback_swap_ema_weights.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Swaps model weights and EMA weights before and after evaluation. — callback_swap_ema_weights","text":"","code":"callback_swap_ema_weights(swap_on_epoch = FALSE)"},{"path":"https://keras3.posit.co/dev/reference/callback_swap_ema_weights.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Swaps model weights and EMA weights before and after evaluation. — callback_swap_ema_weights","text":"swap_on_epoch Whether perform swapping on_epoch_begin() on_epoch_end(). useful want use EMA weights callbacks callback_model_checkpoint(). Defaults FALSE.","code":""},{"path":"https://keras3.posit.co/dev/reference/callback_swap_ema_weights.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Swaps model weights and EMA weights before and after evaluation. — callback_swap_ema_weights","text":"Callback instance can passed fit.keras.src.models.model.Model().","code":""},{"path":"https://keras3.posit.co/dev/reference/callback_swap_ema_weights.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Swaps model weights and EMA weights before and after evaluation. — callback_swap_ema_weights","text":"","code":"# Remember to set `use_ema=TRUE` in the optimizer optimizer <- optimizer_sgd(use_ema = TRUE) model |> compile(optimizer = optimizer, loss = ..., metrics = ...)  # Metrics will be computed with EMA weights model |> fit(X_train, Y_train,              callbacks = c(callback_swap_ema_weights()))  # If you want to save model checkpoint with EMA weights, you can set # `swap_on_epoch=TRUE` and place ModelCheckpoint after SwapEMAWeights. model |> fit(   X_train, Y_train,   callbacks = c(     callback_swap_ema_weights(swap_on_epoch = TRUE),     callback_model_checkpoint(...)   ) )"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/callback_tensorboard.html","id":null,"dir":"Reference","previous_headings":"","what":"Enable visualizations for TensorBoard. — callback_tensorboard","title":"Enable visualizations for TensorBoard. — callback_tensorboard","text":"TensorBoard visualization tool provided TensorFlow. TensorFlow installation required use callback. callback logs events TensorBoard, including: Metrics summary plots Training graph visualization Weight histograms Sampled profiling used model |> evaluate() regular validation addition epoch summaries, summary records evaluation metrics vs model$optimizer$iterations written. metric names prepended evaluation, model$optimizer$iterations step visualized TensorBoard. installed TensorFlow pip reticulate::py_install(), able launch TensorBoard command line:   R tensorflow::tensorboard(). can find information TensorBoard .","code":"tensorboard --logdir=path_to_your_logs"},{"path":"https://keras3.posit.co/dev/reference/callback_tensorboard.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Enable visualizations for TensorBoard. — callback_tensorboard","text":"","code":"callback_tensorboard(   log_dir = \"logs\",   histogram_freq = 0L,   write_graph = TRUE,   write_images = FALSE,   write_steps_per_second = FALSE,   update_freq = \"epoch\",   profile_batch = 0L,   embeddings_freq = 0L,   embeddings_metadata = NULL )"},{"path":"https://keras3.posit.co/dev/reference/callback_tensorboard.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Enable visualizations for TensorBoard. — callback_tensorboard","text":"log_dir path directory save log files parsed TensorBoard. e.g., log_dir = file.path(working_dir, 'logs'). directory reused callbacks. histogram_freq frequency (epochs) compute weight histograms layers model. set 0, histograms computed. Validation data (split) must specified histogram visualizations. write_graph (supported time) Whether visualize graph TensorBoard. Note log file can become quite large write_graph set TRUE. write_images whether write model weights visualize image TensorBoard. write_steps_per_second whether log training steps per second TensorBoard. supports epoch batch frequency logging. update_freq \"batch\" \"epoch\" integer. using \"epoch\", writes losses metrics TensorBoard every epoch. using integer, say 1000, metrics losses (including custom ones added Model.compile) logged TensorBoard every 1000 batches. \"batch\" synonym 1, meaning written every batch. Note however writing frequently TensorBoard can slow training, especially used distribution strategies incur additional synchronization overhead. Batch-level summary writing also available via train_step override. Please see TensorBoard Scalars tutorial details. profile_batch Profile batch(es) sample compute characteristics. profile_batch must non-negative integer tuple integers. pair positive integers signify range batches profile. default, profiling disabled. embeddings_freq frequency (epochs) embedding layers visualized. set 0, embeddings visualized. embeddings_metadata Named list maps embedding layer names filename file save metadata embedding layer. case metadata file used embedding layers, single filename can passed.","code":""},{"path":"https://keras3.posit.co/dev/reference/callback_tensorboard.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Enable visualizations for TensorBoard. — callback_tensorboard","text":"Callback instance can passed fit.keras.src.models.model.Model().","code":""},{"path":"https://keras3.posit.co/dev/reference/callback_tensorboard.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Enable visualizations for TensorBoard. — callback_tensorboard","text":"Custom batch-level summaries subclassed Model:   Custom batch-level summaries Functional API Model:   Profiling:","code":"tensorboard_callback <- callback_tensorboard(log_dir = \"./logs\") model %>% fit(x_train, y_train, epochs = 2, callbacks = list(tensorboard_callback)) # Then run the tensorboard command to view the visualizations. MyModel <- new_model_class(\"MyModel\",   initialize = function() {     self$dense <- layer_dense(units = 10)   },   call = function(x) {     outputs <- x |> self$dense()     tf$summary$histogram('outputs', outputs)     outputs   } )  model <- MyModel() model |> compile(optimizer = 'sgd', loss = 'mse')  # Make sure to set `update_freq = N` to log a batch-level summary every N # batches. In addition to any `tf$summary` contained in `model$call()`, # metrics added in `model |>compile` will be logged every N batches. tb_callback <- callback_tensorboard(log_dir = './logs', update_freq = 1) model |> fit(x_train, y_train, callbacks = list(tb_callback)) my_summary <- function(x) {   tf$summary$histogram('x', x)   x }  inputs <- layer_input(10) outputs <- inputs |>   layer_dense(10) |>   layer_lambda(my_summary)  model <- keras_model(inputs, outputs) model |> compile(optimizer = 'sgd', loss = 'mse')  # Make sure to set `update_freq = N` to log a batch-level summary every N # batches. In addition to any `tf.summary` contained in `Model.call`, # metrics added in `Model.compile` will be logged every N batches. tb_callback <- callback_tensorboard(log_dir = './logs', update_freq = 1) model |> fit(x_train, y_train, callbacks = list(tb_callback)) # Profile a single batch, e.g. the 5th batch. tensorboard_callback <- callback_tensorboard(   log_dir = './logs', profile_batch = 5) model |> fit(x_train, y_train, epochs = 2,              callbacks = list(tensorboard_callback))  # Profile a range of batches, e.g. from 10 to 20. tensorboard_callback <- callback_tensorboard(   log_dir = './logs', profile_batch = c(10, 20)) model |> fit(x_train, y_train, epochs = 2,              callbacks = list(tensorboard_callback))"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/callback_terminate_on_nan.html","id":null,"dir":"Reference","previous_headings":"","what":"Callback that terminates training when a NaN loss is encountered. — callback_terminate_on_nan","title":"Callback that terminates training when a NaN loss is encountered. — callback_terminate_on_nan","text":"Callback terminates training NaN loss encountered.","code":""},{"path":"https://keras3.posit.co/dev/reference/callback_terminate_on_nan.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Callback that terminates training when a NaN loss is encountered. — callback_terminate_on_nan","text":"","code":"callback_terminate_on_nan()"},{"path":"https://keras3.posit.co/dev/reference/callback_terminate_on_nan.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Callback that terminates training when a NaN loss is encountered. — callback_terminate_on_nan","text":"Callback instance can passed fit.keras.src.models.model.Model().","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/clear_session.html","id":null,"dir":"Reference","previous_headings":"","what":"Resets all state generated by Keras. — clear_session","title":"Resets all state generated by Keras. — clear_session","text":"Keras manages global state, uses implement Functional model-building API uniquify autogenerated layer names. creating many models loop, global state consume increasing amount memory time, may want clear . Calling clear_session() releases global state: helps avoid clutter old models layers, especially memory limited. Example 1: calling clear_session() creating models loop   Example 2: resetting layer name generation counter","code":"for (i in 1:100) {   # Without `clear_session()`, each iteration of this loop will   # slightly increase the size of the global state managed by Keras   model <- keras_model_sequential()   for (j in 1:10) {     model <- model |> layer_dense(units = 10)   } }  for (i in 1:100) {   # With `clear_session()` called at the beginning,   # Keras starts with a blank state at each iteration   # and memory consumption is constant over time.   clear_session()   model <- keras_model_sequential()   for (j in 1:10) {     model <- model |> layer_dense(units = 10)   } } layers <- lapply(1:10, \\(i) layer_dense(units = 10))  new_layer <- layer_dense(units = 10) print(new_layer$name) ## [1] \"dense_10\" clear_session() new_layer <- layer_dense(units = 10) print(new_layer$name) ## [1] \"dense\""},{"path":"https://keras3.posit.co/dev/reference/clear_session.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Resets all state generated by Keras. — clear_session","text":"","code":"clear_session(free_memory = TRUE)"},{"path":"https://keras3.posit.co/dev/reference/clear_session.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Resets all state generated by Keras. — clear_session","text":"free_memory Whether call Python garbage collection. usually good practice call make sure memory used deleted objects immediately freed. However, may take seconds execute, using clear_session() short loop, may want skip .","code":""},{"path":"https://keras3.posit.co/dev/reference/clear_session.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Resets all state generated by Keras. — clear_session","text":"NULL, invisibly, called side effects.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/clone_model.html","id":null,"dir":"Reference","previous_headings":"","what":"Clone a Functional or Sequential Model instance. — clone_model","title":"Clone a Functional or Sequential Model instance. — clone_model","text":"Model cloning similar calling model new inputs, except creates new layers (thus new weights) instead sharing weights existing layers. Note clone_model() preserve uniqueness shared objects within model (e.g. single variable attached two distinct layers restored two separate variables).","code":""},{"path":"https://keras3.posit.co/dev/reference/clone_model.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Clone a Functional or Sequential Model instance. — clone_model","text":"","code":"clone_model(   model,   input_tensors = NULL,   clone_function = NULL,   call_function = NULL,   recursive = FALSE,   ... )"},{"path":"https://keras3.posit.co/dev/reference/clone_model.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Clone a Functional or Sequential Model instance. — clone_model","text":"model Instance Model (Functional model Sequential model). input_tensors Optional list input tensors build model upon. provided, new keras_input() objects created. clone_function Callable signature function(layer) used clone layer target model (except Input instances). takes argument layer instance cloned, returns corresponding layer instance used model copy. unspecified, callable defaults following serialization/deserialization function: function(layer) layer$`__class__`$from_config(layer$get_config()). passing custom callable, can customize copy model, e.g. wrapping certain layers interest (might want replace LSTM instances equivalent Bidirectional(LSTM(...)) instances, example). Defaults NULL. call_function Callable signature function(layer, ...) used call cloned layer set inputs. takes layer instance, call arguments, returns call outputs. unspecified, callable defaults regular call() method: function(layer, ...) .call(layer, list(...)). passing custom callable, can insert new layers given layer. recursive Note, argument can used Functional models. Boolean. Whether recursively clone Sequential Functional models encountered original Sequential/Functional model. FALSE, inner models cloned calling clone_function(). TRUE, inner models cloned calling clone_model() clone_function, call_function, recursive arguments. Note case, call_function propagated Sequential model (since applicable Sequential models). ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/clone_model.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Clone a Functional or Sequential Model instance. — clone_model","text":"instance Model reproducing behavior original model, top new inputs tensors, using newly instantiated weights. cloned model may behave differently original model custom clone_function call_function modifies layer layer call.","code":""},{"path":"https://keras3.posit.co/dev/reference/clone_model.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Clone a Functional or Sequential Model instance. — clone_model","text":"Using clone_function make model deterministic setting random seed everywhere:   Using call_function add Dropout layer Dense layer (without recreating new layers):     Note subclassed models cloned default, since internal layer structure known. achieve equivalent functionality clone_model case subclassed model, simply make sure model class implements get_config() (optionally from_config()), call:   case subclassed model, using custom clone_function.","code":"# Create a test Sequential model. model <- keras_model_sequential(input_shape = c(728)) |>   layer_dense(32, activation = 'relu') |>   layer_dense(1, activation = 'sigmoid')  # Create a copy of the test model (with freshly initialized weights). new_model <- clone_model(model) clone_function <- function(layer) {   config <- layer$get_config()   if (\"seed\" %in% names(config))     config$seed <- 1337L   layer$`__class__`$from_config(config) }  new_model <- clone_model(model, clone_function = clone_function) call_function <- function(layer, ...) {   out <- layer(...)   if (inherits(layer, keras$layers$Dense))     out <- out |> layer_dropout(0.5)   out }  inputs <- keras_input(c(728)) outputs <- inputs |>   layer_dense(32, activation = 'relu') |>   layer_dense(1, activation = 'sigmoid') model <- keras_model(inputs, outputs)  new_model <- clone_model(   model,   clone_function = function(x) x, # Reuse the same layers.   call_function = call_function, ) new_model ## Model: \"functional_4\" ## +-----------------------------------+--------------------------+---------------+ ## | Layer (type)                      | Output Shape             |       Param # | ## +===================================+==========================+===============+ ## | keras_tensor_8 (InputLayer)       | (None, 728)              |             0 | ## +-----------------------------------+--------------------------+---------------+ ## | dense_2 (Dense)                   | (None, 32)               |        23,328 | ## +-----------------------------------+--------------------------+---------------+ ## | dropout (Dropout)                 | (None, 32)               |             0 | ## +-----------------------------------+--------------------------+---------------+ ## | dense_3 (Dense)                   | (None, 1)                |            33 | ## +-----------------------------------+--------------------------+---------------+ ## | dropout_1 (Dropout)               | (None, 1)                |             0 | ## +-----------------------------------+--------------------------+---------------+ ##  Total params: 23,361 (91.25 KB) ##  Trainable params: 23,361 (91.25 KB) ##  Non-trainable params: 0 (0.00 B) new_model <- model$`__class__`$from_config(model$get_config())"},{"path":"https://keras3.posit.co/dev/reference/compile.keras.src.models.model.Model.html","id":null,"dir":"Reference","previous_headings":"","what":"Configure a model for training. — compile.keras.src.models.model.Model","title":"Configure a model for training. — compile.keras.src.models.model.Model","text":"Configure model training.","code":""},{"path":"https://keras3.posit.co/dev/reference/compile.keras.src.models.model.Model.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Configure a model for training. — compile.keras.src.models.model.Model","text":"","code":"# S3 method for class 'keras.src.models.model.Model' compile(   object,   optimizer = \"rmsprop\",   loss = NULL,   metrics = NULL,   ...,   loss_weights = NULL,   weighted_metrics = NULL,   run_eagerly = FALSE,   steps_per_execution = 1L,   jit_compile = \"auto\",   auto_scale_loss = TRUE )"},{"path":"https://keras3.posit.co/dev/reference/compile.keras.src.models.model.Model.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Configure a model for training. — compile.keras.src.models.model.Model","text":"object Keras model object optimizer String (name optimizer) optimizer instance. See optimizer_* family. loss Loss function. May : string (name builtin loss function), custom function, Loss instance (returned loss_* family functions). loss function callable signature loss = fn(y_true, y_pred), y_true ground truth values, y_pred model's predictions. y_true shape (batch_size, d1, .. dN) (except case sparse loss functions sparse categorical crossentropy expects integer arrays shape (batch_size, d1, .. dN-1)). y_pred shape (batch_size, d1, .. dN). loss function return float tensor. metrics List metrics evaluated model training testing. can : string (name built-function), function, optionally \"name\" attribute Metric() instance. See metric_* family functions. Typically use metrics = c('accuracy'). function callable signature result = fn(y_true, y_pred). specify different metrics different outputs multi-output model, also pass named list, metrics = list(= 'accuracy', b = c('accuracy', 'mse')). can also pass list specify metric list metrics output, metrics = list(c('accuracy'), c('accuracy', 'mse')) metrics = list('accuracy', c('accuracy', 'mse')). pass strings 'accuracy' 'acc', convert one metric_binary_accuracy(), metric_categorical_accuracy(), metric_sparse_categorical_accuracy() based shapes targets model output. similar conversion done strings \"crossentropy\" \"ce\" well. metrics passed evaluated without sample weighting; like sample weighting apply, can specify metrics via weighted_metrics argument instead. providing anonymous R function, can customize printed name training assigning attr(<fn>, \"name\") <- \"my_custom_metric_name\", calling custom_metric(\"my_custom_metric_name\", <fn>) ... Additional arguments passed compile() model method. loss_weights Optional list (named unnamed) specifying scalar coefficients (R numerics) weight loss contributions different model outputs. loss value minimized model weighted sum individual losses, weighted loss_weights coefficients.  unnamed list, expected 1:1 mapping model's outputs. named list, expected map output names (strings) scalar coefficients. weighted_metrics List metrics evaluated weighted sample_weight class_weight training testing. run_eagerly Bool. TRUE, model's forward pass never compiled. recommended leave FALSE training (best performance), set TRUE debugging. steps_per_execution Int. number batches run single compiled function call. Running multiple batches inside single compiled function call can greatly improve performance TPUs small models large R/Python overhead. , one full epoch run execution. number larger size epoch passed, execution truncated size epoch. Note steps_per_execution set N, Callback$on_batch_begin Callback$on_batch_end methods called every N batches (.e. /compiled function execution). supported PyTorch backend. jit_compile Bool \"auto\". Whether use XLA compilation compiling model. jax tensorflow backends, jit_compile=\"auto\" enables XLA compilation model supports , disabled otherwise. torch backend, \"auto\" default eager execution jit_compile=True run torch.compile \"inductor\" backend. auto_scale_loss Bool. TRUE model dtype policy \"mixed_float16\", passed optimizer automatically wrapped LossScaleOptimizer, dynamically scale loss prevent underflow.","code":""},{"path":"https://keras3.posit.co/dev/reference/compile.keras.src.models.model.Model.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Configure a model for training. — compile.keras.src.models.model.Model","text":"called primarily side effect modifying object -place. first argument object also returned, invisibly, enable usage pipe.","code":""},{"path":"https://keras3.posit.co/dev/reference/compile.keras.src.models.model.Model.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Configure a model for training. — compile.keras.src.models.model.Model","text":"","code":"model |> compile(   optimizer = optimizer_adam(learning_rate = 1e-3),   loss = loss_binary_crossentropy(),   metrics = c(metric_binary_accuracy(),               metric_false_negatives()) )"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/config_backend.html","id":null,"dir":"Reference","previous_headings":"","what":"Publicly accessible method for determining the current backend. — config_backend","title":"Publicly accessible method for determining the current backend. — config_backend","text":"Publicly accessible method determining current backend.","code":""},{"path":"https://keras3.posit.co/dev/reference/config_backend.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Publicly accessible method for determining the current backend. — config_backend","text":"","code":"config_backend()"},{"path":"https://keras3.posit.co/dev/reference/config_backend.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Publicly accessible method for determining the current backend. — config_backend","text":"String, name backend Keras currently using. One \"tensorflow\", \"torch\", \"jax\".","code":""},{"path":"https://keras3.posit.co/dev/reference/config_backend.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Publicly accessible method for determining the current backend. — config_backend","text":"","code":"config_backend() ## [1] \"tensorflow\""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/config_disable_flash_attention.html","id":null,"dir":"Reference","previous_headings":"","what":"Disable flash attention. — config_disable_flash_attention","title":"Disable flash attention. — config_disable_flash_attention","text":"Flash attention offers performance optimization attention layers, making especially useful large language models (LLMs) benefit faster memory-efficient attention computations. disabled, supported layers like MultiHeadAttention use flash attention faster computations.","code":""},{"path":"https://keras3.posit.co/dev/reference/config_disable_flash_attention.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Disable flash attention. — config_disable_flash_attention","text":"","code":"config_disable_flash_attention()"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/config_disable_interactive_logging.html","id":null,"dir":"Reference","previous_headings":"","what":"Turn off interactive logging. — config_disable_interactive_logging","title":"Turn off interactive logging. — config_disable_interactive_logging","text":"interactive logging disabled, Keras sends logs absl.logging. best option using Keras non-interactive way, running training inference job server.","code":""},{"path":"https://keras3.posit.co/dev/reference/config_disable_interactive_logging.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Turn off interactive logging. — config_disable_interactive_logging","text":"","code":"config_disable_interactive_logging()"},{"path":"https://keras3.posit.co/dev/reference/config_disable_interactive_logging.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Turn off interactive logging. — config_disable_interactive_logging","text":"return value, called side effects.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/config_disable_traceback_filtering.html","id":null,"dir":"Reference","previous_headings":"","what":"Turn off traceback filtering. — config_disable_traceback_filtering","title":"Turn off traceback filtering. — config_disable_traceback_filtering","text":"Raw Keras tracebacks (also known stack traces) involve many internal frames, can challenging read , actionable end users. default, Keras filters internal frames exceptions raises, keep traceback short, readable, focused actionable (code). See also config_enable_traceback_filtering() config_is_traceback_filtering_enabled(). previously disabled traceback filtering via config_disable_traceback_filtering(), can re-enable via config_enable_traceback_filtering().","code":""},{"path":"https://keras3.posit.co/dev/reference/config_disable_traceback_filtering.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Turn off traceback filtering. — config_disable_traceback_filtering","text":"","code":"config_disable_traceback_filtering()"},{"path":"https://keras3.posit.co/dev/reference/config_disable_traceback_filtering.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Turn off traceback filtering. — config_disable_traceback_filtering","text":"return value, called side effects.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/config_dtype_policy.html","id":null,"dir":"Reference","previous_headings":"","what":"Returns the current default dtype policy object. — config_dtype_policy","title":"Returns the current default dtype policy object. — config_dtype_policy","text":"Returns current default dtype policy object.","code":""},{"path":"https://keras3.posit.co/dev/reference/config_dtype_policy.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Returns the current default dtype policy object. — config_dtype_policy","text":"","code":"config_dtype_policy()"},{"path":"https://keras3.posit.co/dev/reference/config_dtype_policy.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Returns the current default dtype policy object. — config_dtype_policy","text":"DTypePolicy object.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/config_enable_flash_attention.html","id":null,"dir":"Reference","previous_headings":"","what":"Enable flash attention. — config_enable_flash_attention","title":"Enable flash attention. — config_enable_flash_attention","text":"Flash attention offers performance optimization attention layers, making especially useful large language models (LLMs) benefit faster memory-efficient attention computations. enabled, supported layers like layer_multi_head_attention attempt use flash attention faster computations. default, feature enabled. Note enabling flash attention guarantee always used. Typically, inputs must float16 bfloat16 dtype, input layout requirements may vary depending backend.","code":""},{"path":"https://keras3.posit.co/dev/reference/config_enable_flash_attention.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Enable flash attention. — config_enable_flash_attention","text":"","code":"config_enable_flash_attention()"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/config_enable_interactive_logging.html","id":null,"dir":"Reference","previous_headings":"","what":"Turn on interactive logging. — config_enable_interactive_logging","title":"Turn on interactive logging. — config_enable_interactive_logging","text":"interactive logging enabled, Keras displays logs via stdout. provides best experience using Keras interactive environment shell notebook.","code":""},{"path":"https://keras3.posit.co/dev/reference/config_enable_interactive_logging.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Turn on interactive logging. — config_enable_interactive_logging","text":"","code":"config_enable_interactive_logging()"},{"path":"https://keras3.posit.co/dev/reference/config_enable_interactive_logging.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Turn on interactive logging. — config_enable_interactive_logging","text":"return value, called side effects.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/config_enable_traceback_filtering.html","id":null,"dir":"Reference","previous_headings":"","what":"Turn on traceback filtering. — config_enable_traceback_filtering","title":"Turn on traceback filtering. — config_enable_traceback_filtering","text":"Raw Keras tracebacks (also known stack traces) involve many internal frames, can challenging read , actionable end users. default, Keras filters internal frames exceptions raises, keep traceback short, readable, focused actionable (code). See also config_disable_traceback_filtering() config_is_traceback_filtering_enabled(). previously disabled traceback filtering via config_disable_traceback_filtering(), can re-enable via config_enable_traceback_filtering().","code":""},{"path":"https://keras3.posit.co/dev/reference/config_enable_traceback_filtering.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Turn on traceback filtering. — config_enable_traceback_filtering","text":"","code":"config_enable_traceback_filtering()"},{"path":"https://keras3.posit.co/dev/reference/config_enable_traceback_filtering.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Turn on traceback filtering. — config_enable_traceback_filtering","text":"return value, called side effects.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/config_enable_unsafe_deserialization.html","id":null,"dir":"Reference","previous_headings":"","what":"Disables safe mode globally, allowing deserialization of lambdas. — config_enable_unsafe_deserialization","title":"Disables safe mode globally, allowing deserialization of lambdas. — config_enable_unsafe_deserialization","text":"Disables safe mode globally, allowing deserialization lambdas.","code":""},{"path":"https://keras3.posit.co/dev/reference/config_enable_unsafe_deserialization.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Disables safe mode globally, allowing deserialization of lambdas. — config_enable_unsafe_deserialization","text":"","code":"config_enable_unsafe_deserialization()"},{"path":"https://keras3.posit.co/dev/reference/config_enable_unsafe_deserialization.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Disables safe mode globally, allowing deserialization of lambdas. — config_enable_unsafe_deserialization","text":"return value, called side effects.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/config_epsilon.html","id":null,"dir":"Reference","previous_headings":"","what":"Return the value of the fuzz factor used in numeric expressions. — config_epsilon","title":"Return the value of the fuzz factor used in numeric expressions. — config_epsilon","text":"Return value fuzz factor used numeric expressions.","code":""},{"path":"https://keras3.posit.co/dev/reference/config_epsilon.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Return the value of the fuzz factor used in numeric expressions. — config_epsilon","text":"","code":"config_epsilon()"},{"path":"https://keras3.posit.co/dev/reference/config_epsilon.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Return the value of the fuzz factor used in numeric expressions. — config_epsilon","text":"float.","code":""},{"path":"https://keras3.posit.co/dev/reference/config_epsilon.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Return the value of the fuzz factor used in numeric expressions. — config_epsilon","text":"","code":"config_epsilon() ## [1] 1e-07"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/config_floatx.html","id":null,"dir":"Reference","previous_headings":"","what":"Return the default float type, as a string. — config_floatx","title":"Return the default float type, as a string. — config_floatx","text":"E.g. 'bfloat16' 'float16', 'float32', 'float64'.","code":""},{"path":"https://keras3.posit.co/dev/reference/config_floatx.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Return the default float type, as a string. — config_floatx","text":"","code":"config_floatx()"},{"path":"https://keras3.posit.co/dev/reference/config_floatx.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Return the default float type, as a string. — config_floatx","text":"String, current default float type.","code":""},{"path":"https://keras3.posit.co/dev/reference/config_floatx.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Return the default float type, as a string. — config_floatx","text":"","code":"keras3::config_floatx() ## [1] \"float32\""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/config_image_data_format.html","id":null,"dir":"Reference","previous_headings":"","what":"Return the default image data format convention. — config_image_data_format","title":"Return the default image data format convention. — config_image_data_format","text":"Return default image data format convention.","code":""},{"path":"https://keras3.posit.co/dev/reference/config_image_data_format.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Return the default image data format convention. — config_image_data_format","text":"","code":"config_image_data_format()"},{"path":"https://keras3.posit.co/dev/reference/config_image_data_format.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Return the default image data format convention. — config_image_data_format","text":"string, either 'channels_first' 'channels_last'.","code":""},{"path":"https://keras3.posit.co/dev/reference/config_image_data_format.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Return the default image data format convention. — config_image_data_format","text":"","code":"config_image_data_format() ## [1] \"channels_last\""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/config_is_flash_attention_enabled.html","id":null,"dir":"Reference","previous_headings":"","what":"Checks whether flash attention is globally enabled in Keras. — config_is_flash_attention_enabled","title":"Checks whether flash attention is globally enabled in Keras. — config_is_flash_attention_enabled","text":"Flash attention performance-optimized method computing attention large models, transformers, allowing faster memory-efficient operations. function checks global Keras configuration determine flash attention enabled compatible layers (e.g., MultiHeadAttention). Note enabling flash attention guarantee always used. Typically, inputs must float16 bfloat16 dtype, input layout requirements may vary depending backend.","code":""},{"path":"https://keras3.posit.co/dev/reference/config_is_flash_attention_enabled.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Checks whether flash attention is globally enabled in Keras. — config_is_flash_attention_enabled","text":"","code":"config_is_flash_attention_enabled()"},{"path":"https://keras3.posit.co/dev/reference/config_is_flash_attention_enabled.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Checks whether flash attention is globally enabled in Keras. — config_is_flash_attention_enabled","text":"FALSE disabled; otherwise, indicates enabled.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/config_is_interactive_logging_enabled.html","id":null,"dir":"Reference","previous_headings":"","what":"Check if interactive logging is enabled. — config_is_interactive_logging_enabled","title":"Check if interactive logging is enabled. — config_is_interactive_logging_enabled","text":"switch writing logs stdout absl.logging, may use config_enable_interactive_logging() config_disable_interactive_logging().","code":""},{"path":"https://keras3.posit.co/dev/reference/config_is_interactive_logging_enabled.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Check if interactive logging is enabled. — config_is_interactive_logging_enabled","text":"","code":"config_is_interactive_logging_enabled()"},{"path":"https://keras3.posit.co/dev/reference/config_is_interactive_logging_enabled.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Check if interactive logging is enabled. — config_is_interactive_logging_enabled","text":"Boolean, TRUE interactive logging enabled, FALSE otherwise.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/config_is_nnx_enabled.html","id":null,"dir":"Reference","previous_headings":"","what":"Check whether NNX-specific features are enabled on the JAX backend. — config_is_nnx_enabled","title":"Check whether NNX-specific features are enabled on the JAX backend. — config_is_nnx_enabled","text":"Check whether NNX-specific features enabled JAX backend.","code":""},{"path":"https://keras3.posit.co/dev/reference/config_is_nnx_enabled.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Check whether NNX-specific features are enabled on the JAX backend. — config_is_nnx_enabled","text":"","code":"config_is_nnx_enabled()"},{"path":"https://keras3.posit.co/dev/reference/config_is_nnx_enabled.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Check whether NNX-specific features are enabled on the JAX backend. — config_is_nnx_enabled","text":"Logical flag; TRUE NNX backend features enabled, FALSE otherwise.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/config_is_traceback_filtering_enabled.html","id":null,"dir":"Reference","previous_headings":"","what":"Check if traceback filtering is enabled. — config_is_traceback_filtering_enabled","title":"Check if traceback filtering is enabled. — config_is_traceback_filtering_enabled","text":"Raw Keras tracebacks (also known stack traces) involve many internal frames, can challenging read , actionable end users. default, Keras filters internal frames exceptions raises, keep traceback short, readable, focused actionable (code). See also config_enable_traceback_filtering() config_disable_traceback_filtering(). previously disabled traceback filtering via config_disable_traceback_filtering(), can re-enable via config_enable_traceback_filtering().","code":""},{"path":"https://keras3.posit.co/dev/reference/config_is_traceback_filtering_enabled.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Check if traceback filtering is enabled. — config_is_traceback_filtering_enabled","text":"","code":"config_is_traceback_filtering_enabled()"},{"path":"https://keras3.posit.co/dev/reference/config_is_traceback_filtering_enabled.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Check if traceback filtering is enabled. — config_is_traceback_filtering_enabled","text":"Boolean, TRUE traceback filtering enabled, FALSE otherwise.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/config_max_epochs.html","id":null,"dir":"Reference","previous_headings":"","what":"Configure the default training loop limits. — config_max_epochs","title":"Configure the default training loop limits. — config_max_epochs","text":"helpers control caps built-training loops obey running fit(), evaluate(), predict(). values can also provided via KERAS_MAX_EPOCHS KERAS_MAX_STEPS_PER_EPOCH environment variables quickly constrain run without modifying source code.","code":""},{"path":"https://keras3.posit.co/dev/reference/config_max_epochs.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Configure the default training loop limits. — config_max_epochs","text":"","code":"config_max_epochs()  config_set_max_epochs(max_epochs)  config_max_steps_per_epoch()  config_set_max_steps_per_epoch(max_steps_per_epoch)"},{"path":"https://keras3.posit.co/dev/reference/config_max_epochs.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Configure the default training loop limits. — config_max_epochs","text":"max_epochs Integer upper bound epochs processed built-training loops. Use NULL remove cap. max_steps_per_epoch Integer upper bound steps processed per epoch built-training loops. Use NULL remove cap.","code":""},{"path":"https://keras3.posit.co/dev/reference/config_max_epochs.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Configure the default training loop limits. — config_max_epochs","text":"config_max_epochs() config_max_steps_per_epoch() return current integer limits (NULL cap unset). setter variants return NULL invisibly called side effects.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/config_set_backend.html","id":null,"dir":"Reference","previous_headings":"","what":"Reload the backend (and the Keras package). — config_set_backend","title":"Reload the backend (and the Keras package). — config_set_backend","text":"Reload backend (Keras package).","code":""},{"path":"https://keras3.posit.co/dev/reference/config_set_backend.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Reload the backend (and the Keras package). — config_set_backend","text":"","code":"config_set_backend(backend)"},{"path":"https://keras3.posit.co/dev/reference/config_set_backend.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Reload the backend (and the Keras package). — config_set_backend","text":"backend String","code":""},{"path":"https://keras3.posit.co/dev/reference/config_set_backend.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Reload the backend (and the Keras package). — config_set_backend","text":"Nothing, function called side effect.","code":""},{"path":"https://keras3.posit.co/dev/reference/config_set_backend.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Reload the backend (and the Keras package). — config_set_backend","text":"","code":"config_set_backend(\"jax\")"},{"path":"https://keras3.posit.co/dev/reference/config_set_backend.html","id":"warning","dir":"Reference","previous_headings":"","what":"WARNING","title":"Reload the backend (and the Keras package). — config_set_backend","text":"Using function dangerous done carefully. Changing backend convert type already-instantiated objects. Thus, layers / tensors / etc. already created longer usable without errors. strongly recommended keep around Keras-originated objects instances created calling config_set_backend(). includes function class instance uses Keras functionality. code needs re-executed calling config_set_backend().","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/config_set_dtype_policy.html","id":null,"dir":"Reference","previous_headings":"","what":"Sets the default dtype policy globally. — config_set_dtype_policy","title":"Sets the default dtype policy globally. — config_set_dtype_policy","text":"Sets default dtype policy globally.","code":""},{"path":"https://keras3.posit.co/dev/reference/config_set_dtype_policy.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Sets the default dtype policy globally. — config_set_dtype_policy","text":"","code":"config_set_dtype_policy(policy)"},{"path":"https://keras3.posit.co/dev/reference/config_set_dtype_policy.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Sets the default dtype policy globally. — config_set_dtype_policy","text":"policy string DTypePolicy object.","code":""},{"path":"https://keras3.posit.co/dev/reference/config_set_dtype_policy.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Sets the default dtype policy globally. — config_set_dtype_policy","text":"return value, called side effects.","code":""},{"path":"https://keras3.posit.co/dev/reference/config_set_dtype_policy.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Sets the default dtype policy globally. — config_set_dtype_policy","text":"","code":"config_set_dtype_policy(\"mixed_float16\")"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/config_set_epsilon.html","id":null,"dir":"Reference","previous_headings":"","what":"Set the value of the fuzz factor used in numeric expressions. — config_set_epsilon","title":"Set the value of the fuzz factor used in numeric expressions. — config_set_epsilon","text":"Set value fuzz factor used numeric expressions.","code":""},{"path":"https://keras3.posit.co/dev/reference/config_set_epsilon.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Set the value of the fuzz factor used in numeric expressions. — config_set_epsilon","text":"","code":"config_set_epsilon(value)"},{"path":"https://keras3.posit.co/dev/reference/config_set_epsilon.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Set the value of the fuzz factor used in numeric expressions. — config_set_epsilon","text":"value float. New value epsilon.","code":""},{"path":"https://keras3.posit.co/dev/reference/config_set_epsilon.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Set the value of the fuzz factor used in numeric expressions. — config_set_epsilon","text":"return value, called side effects.","code":""},{"path":"https://keras3.posit.co/dev/reference/config_set_epsilon.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Set the value of the fuzz factor used in numeric expressions. — config_set_epsilon","text":"","code":"config_epsilon() ## [1] 1e-07 config_set_epsilon(1e-5) config_epsilon() ## [1] 1e-05 # Set it back to the default value. config_set_epsilon(1e-7)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/config_set_floatx.html","id":null,"dir":"Reference","previous_headings":"","what":"Set the default float dtype. — config_set_floatx","title":"Set the default float dtype. — config_set_floatx","text":"Set default float dtype.","code":""},{"path":"https://keras3.posit.co/dev/reference/config_set_floatx.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Set the default float dtype. — config_set_floatx","text":"","code":"config_set_floatx(value)"},{"path":"https://keras3.posit.co/dev/reference/config_set_floatx.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Set the default float dtype. — config_set_floatx","text":"value String; 'bfloat16', 'float16', 'float32', 'float64'.","code":""},{"path":"https://keras3.posit.co/dev/reference/config_set_floatx.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Set the default float dtype. — config_set_floatx","text":"return value, called side effects.","code":""},{"path":"https://keras3.posit.co/dev/reference/config_set_floatx.html","id":"note","dir":"Reference","previous_headings":"","what":"Note","title":"Set the default float dtype. — config_set_floatx","text":"recommended set \"float16\" training, likely cause numeric stability issues. Instead, mixed precision, leverages mix float16 float32. can configured calling keras3::keras$mixed_precision$set_dtype_policy('mixed_float16').","code":""},{"path":"https://keras3.posit.co/dev/reference/config_set_floatx.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Set the default float dtype. — config_set_floatx","text":"","code":"config_floatx() ## [1] \"float32\" config_set_floatx('float64') config_floatx() ## [1] \"float64\" # Set it back to float32 config_set_floatx('float32')"},{"path":"https://keras3.posit.co/dev/reference/config_set_floatx.html","id":"raises","dir":"Reference","previous_headings":"","what":"Raises","title":"Set the default float dtype. — config_set_floatx","text":"ValueError: case invalid value.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/config_set_image_data_format.html","id":null,"dir":"Reference","previous_headings":"","what":"Set the value of the image data format convention. — config_set_image_data_format","title":"Set the value of the image data format convention. — config_set_image_data_format","text":"Set value image data format convention.","code":""},{"path":"https://keras3.posit.co/dev/reference/config_set_image_data_format.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Set the value of the image data format convention. — config_set_image_data_format","text":"","code":"config_set_image_data_format(data_format)"},{"path":"https://keras3.posit.co/dev/reference/config_set_image_data_format.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Set the value of the image data format convention. — config_set_image_data_format","text":"data_format string. 'channels_first' 'channels_last'.","code":""},{"path":"https://keras3.posit.co/dev/reference/config_set_image_data_format.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Set the value of the image data format convention. — config_set_image_data_format","text":"return value, called side effects.","code":""},{"path":"https://keras3.posit.co/dev/reference/config_set_image_data_format.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Set the value of the image data format convention. — config_set_image_data_format","text":"","code":"config_image_data_format() ## [1] \"channels_last\" # 'channels_last' keras3::config_set_image_data_format('channels_first') config_image_data_format() ## [1] \"channels_first\" # Set it back to `'channels_last'` keras3::config_set_image_data_format('channels_last')"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/constraint_maxnorm.html","id":null,"dir":"Reference","previous_headings":"","what":"MaxNorm weight constraint. — constraint_maxnorm","title":"MaxNorm weight constraint. — constraint_maxnorm","text":"Constrains weights incident hidden unit norm less equal desired value.","code":""},{"path":"https://keras3.posit.co/dev/reference/constraint_maxnorm.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"MaxNorm weight constraint. — constraint_maxnorm","text":"","code":"constraint_maxnorm(max_value = 2L, axis = 1L)"},{"path":"https://keras3.posit.co/dev/reference/constraint_maxnorm.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"MaxNorm weight constraint. — constraint_maxnorm","text":"max_value maximum norm value incoming weights. axis integer, axis along calculate weight norms. instance, Dense layer weight matrix shape (input_dim, output_dim), set axis 0 constrain weight vector length (input_dim,). Conv2D layer data_format = \"channels_last\", weight tensor shape (rows, cols, input_depth, output_depth), set axis [0, 1, 2] constrain weights filter tensor size (rows, cols, input_depth).","code":""},{"path":"https://keras3.posit.co/dev/reference/constraint_maxnorm.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"MaxNorm weight constraint. — constraint_maxnorm","text":"Constraint instance, callable can passed layer constructors used directly calling tensors.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/constraint_minmaxnorm.html","id":null,"dir":"Reference","previous_headings":"","what":"MinMaxNorm weight constraint. — constraint_minmaxnorm","title":"MinMaxNorm weight constraint. — constraint_minmaxnorm","text":"Constrains weights incident hidden unit norm lower bound upper bound.","code":""},{"path":"https://keras3.posit.co/dev/reference/constraint_minmaxnorm.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"MinMaxNorm weight constraint. — constraint_minmaxnorm","text":"","code":"constraint_minmaxnorm(min_value = 0, max_value = 1, rate = 1, axis = 1L)"},{"path":"https://keras3.posit.co/dev/reference/constraint_minmaxnorm.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"MinMaxNorm weight constraint. — constraint_minmaxnorm","text":"min_value minimum norm incoming weights. max_value maximum norm incoming weights. rate rate enforcing constraint: weights rescaled yield op_clip? (1 - rate) * norm + rate * op_clip(norm, min_value, max_value). Effectively, means rate = 1.0 stands strict enforcement constraint, rate<1.0 means weights rescaled step slowly move towards value inside desired interval. axis integer, axis along calculate weight norms. instance, Dense layer weight matrix shape (input_dim, output_dim), set axis 0 constrain weight vector length (input_dim,). Conv2D layer data_format = \"channels_last\", weight tensor shape (rows, cols, input_depth, output_depth), set axis [0, 1, 2] constrain weights filter tensor size (rows, cols, input_depth).","code":""},{"path":"https://keras3.posit.co/dev/reference/constraint_minmaxnorm.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"MinMaxNorm weight constraint. — constraint_minmaxnorm","text":"Constraint instance, callable can passed layer constructors used directly calling tensors.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/constraint_nonneg.html","id":null,"dir":"Reference","previous_headings":"","what":"Constrains the weights to be non-negative. — constraint_nonneg","title":"Constrains the weights to be non-negative. — constraint_nonneg","text":"Constrains weights non-negative.","code":""},{"path":"https://keras3.posit.co/dev/reference/constraint_nonneg.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Constrains the weights to be non-negative. — constraint_nonneg","text":"","code":"constraint_nonneg()"},{"path":"https://keras3.posit.co/dev/reference/constraint_nonneg.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Constrains the weights to be non-negative. — constraint_nonneg","text":"Constraint instance, callable can passed layer constructors used directly calling tensors.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/constraint_unitnorm.html","id":null,"dir":"Reference","previous_headings":"","what":"Constrains the weights incident to each hidden unit to have unit norm. — constraint_unitnorm","title":"Constrains the weights incident to each hidden unit to have unit norm. — constraint_unitnorm","text":"Constrains weights incident hidden unit unit norm.","code":""},{"path":"https://keras3.posit.co/dev/reference/constraint_unitnorm.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Constrains the weights incident to each hidden unit to have unit norm. — constraint_unitnorm","text":"","code":"constraint_unitnorm(axis = 1L)"},{"path":"https://keras3.posit.co/dev/reference/constraint_unitnorm.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Constrains the weights incident to each hidden unit to have unit norm. — constraint_unitnorm","text":"axis integer, axis along calculate weight norms. instance, Dense layer weight matrix shape (input_dim, output_dim), set axis 0 constrain weight vector length (input_dim,). Conv2D layer data_format = \"channels_last\", weight tensor shape (rows, cols, input_depth, output_depth), set axis [0, 1, 2] constrain weights filter tensor size (rows, cols, input_depth).","code":""},{"path":"https://keras3.posit.co/dev/reference/constraint_unitnorm.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Constrains the weights incident to each hidden unit to have unit norm. — constraint_unitnorm","text":"Constraint instance, callable can passed layer constructors used directly calling tensors.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/count_params.html","id":null,"dir":"Reference","previous_headings":"","what":"Count the total number of scalars composing the weights. — count_params","title":"Count the total number of scalars composing the weights. — count_params","text":"Count total number scalars composing weights.","code":""},{"path":"https://keras3.posit.co/dev/reference/count_params.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Count the total number of scalars composing the weights. — count_params","text":"","code":"count_params(object)"},{"path":"https://keras3.posit.co/dev/reference/count_params.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Count the total number of scalars composing the weights. — count_params","text":"object Layer model object","code":""},{"path":"https://keras3.posit.co/dev/reference/count_params.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Count the total number of scalars composing the weights. — count_params","text":"integer count","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/custom_metric.html","id":null,"dir":"Reference","previous_headings":"","what":"Custom metric function — custom_metric","title":"Custom metric function — custom_metric","text":"Custom metric function","code":""},{"path":"https://keras3.posit.co/dev/reference/custom_metric.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Custom metric function — custom_metric","text":"","code":"custom_metric(name, metric_fn)"},{"path":"https://keras3.posit.co/dev/reference/custom_metric.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Custom metric function — custom_metric","text":"name name used show training progress output metric_fn R function signature function(y_true, y_pred) accepts tensors.","code":""},{"path":"https://keras3.posit.co/dev/reference/custom_metric.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Custom metric function — custom_metric","text":"callable function __name__ attribute.","code":""},{"path":"https://keras3.posit.co/dev/reference/custom_metric.html","id":"details","dir":"Reference","previous_headings":"","what":"Details","title":"Custom metric function — custom_metric","text":"can provide arbitrary R function custom metric. Note y_true y_pred parameters tensors, computations use op_* tensor functions. Use custom_metric() function define custom metric. Note name ('mean_pred') provided custom metric function: name used within training progress output. want save load model custom metrics, also call register_keras_serializable(), specify metric call load_model(). example: load_model(\"my_model.keras\", c('mean_pred' = metric_mean_pred)). Alternatively, can wrap code call with_custom_object_scope() allow refer metric name just like built keras metrics. Alternative ways supplying custom metrics: custom_metric(): Arbitrary R function. metric_mean_wrapper(): Wrap arbitrary R function Metric instance. Create custom Metric() subclass.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/dataset_boston_housing.html","id":null,"dir":"Reference","previous_headings":"","what":"Boston housing price regression dataset — dataset_boston_housing","title":"Boston housing price regression dataset — dataset_boston_housing","text":"Dataset taken StatLib library maintained Carnegie Mellon University.","code":""},{"path":"https://keras3.posit.co/dev/reference/dataset_boston_housing.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Boston housing price regression dataset — dataset_boston_housing","text":"","code":"dataset_boston_housing(   path = \"boston_housing.npz\",   test_split = 0.2,   seed = 113L )"},{"path":"https://keras3.posit.co/dev/reference/dataset_boston_housing.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Boston housing price regression dataset — dataset_boston_housing","text":"path Path cache dataset locally (relative ~/.keras/datasets). test_split fraction data reserve test set. seed Random seed shuffling data computing test split.","code":""},{"path":"https://keras3.posit.co/dev/reference/dataset_boston_housing.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Boston housing price regression dataset — dataset_boston_housing","text":"Lists training test data: train$x, train$y, test$x, test$y. Samples contain 13 attributes houses different locations around Boston suburbs late 1970s. Targets median values houses location (k$).","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/dataset_california_housing.html","id":null,"dir":"Reference","previous_headings":"","what":"Loads the California Housing dataset. — dataset_california_housing","title":"Loads the California Housing dataset. — dataset_california_housing","text":"dataset obtained StatLib repository. continuous regression dataset 20,640 samples 8 features . target variable scalar: median house value California districts, dollars. 8 input features following: MedInc: median income block group HouseAge: median house age block group AveRooms: average number rooms per household AveBedrms: average number bedrooms per household Population: block group population AveOccup: average number household members Latitude: block group latitude Longitude: block group longitude dataset derived 1990 U.S. census, using one row per census block group. block group smallest geographical unit U.S. Census Bureau publishes sample data (block group typically population 600 3,000 people). household group people residing within home. Since average number rooms bedrooms dataset provided per household, columns may take surprisingly large values block groups households many empty houses, vacation resorts.","code":""},{"path":"https://keras3.posit.co/dev/reference/dataset_california_housing.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Loads the California Housing dataset. — dataset_california_housing","text":"","code":"dataset_california_housing(   version = \"large\",   path = \"california_housing.npz\",   test_split = 0.2,   seed = 113L )"},{"path":"https://keras3.posit.co/dev/reference/dataset_california_housing.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Loads the California Housing dataset. — dataset_california_housing","text":"version \"small\" \"large\". small version contains 600 samples, large version contains 20,640 samples. purpose small version serve approximate replacement deprecated boston_housing dataset. path path cache dataset locally (relative Sys.getenv(\"KERAS_HOME\")). test_split fraction data reserve test set. seed Random seed shuffling data computing test split.","code":""},{"path":"https://keras3.posit.co/dev/reference/dataset_california_housing.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Loads the California Housing dataset. — dataset_california_housing","text":"Nested list arrays: (x_train, y_train), (x_test, y_test). x_train, x_test: arrays shape (num_samples, 8) containing either training samples (x_train), test samples (y_train). y_train, y_test: arrays shape (num_samples) containing target scalars. targets float scalars typically 25,000 500,000 represent home prices dollars.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/dataset_cifar10.html","id":null,"dir":"Reference","previous_headings":"","what":"CIFAR10 small image classification — dataset_cifar10","title":"CIFAR10 small image classification — dataset_cifar10","text":"Dataset 50,000 32x32 color training images, labeled 10 categories, 10,000 test images.","code":""},{"path":"https://keras3.posit.co/dev/reference/dataset_cifar10.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"CIFAR10 small image classification — dataset_cifar10","text":"","code":"dataset_cifar10()"},{"path":"https://keras3.posit.co/dev/reference/dataset_cifar10.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"CIFAR10 small image classification — dataset_cifar10","text":"Lists training test data: train$x, train$y, test$x, test$y. x data array RGB image data shape (num_samples, 3, 32, 32). y data array category labels (integers range 0-9) shape (num_samples).","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/dataset_cifar100.html","id":null,"dir":"Reference","previous_headings":"","what":"CIFAR100 small image classification — dataset_cifar100","title":"CIFAR100 small image classification — dataset_cifar100","text":"Dataset 50,000 32x32 color training images, labeled 100 categories, 10,000 test images.","code":""},{"path":"https://keras3.posit.co/dev/reference/dataset_cifar100.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"CIFAR100 small image classification — dataset_cifar100","text":"","code":"dataset_cifar100(label_mode = c(\"fine\", \"coarse\"))"},{"path":"https://keras3.posit.co/dev/reference/dataset_cifar100.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"CIFAR100 small image classification — dataset_cifar100","text":"label_mode one \"fine\", \"coarse\".","code":""},{"path":"https://keras3.posit.co/dev/reference/dataset_cifar100.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"CIFAR100 small image classification — dataset_cifar100","text":"Lists training test data: train$x, train$y, test$x, test$y. x data array RGB image data shape (num_samples, 3, 32, 32). y data array category labels shape (num_samples).","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/dataset_fashion_mnist.html","id":null,"dir":"Reference","previous_headings":"","what":"Fashion-MNIST database of fashion articles — dataset_fashion_mnist","title":"Fashion-MNIST database of fashion articles — dataset_fashion_mnist","text":"Dataset 60,000 28x28 grayscale images 10 fashion article classes, along test set 10,000 images. dataset can used drop-replacement MNIST. class labels encoded integers 0-9 correspond T-shirt/top, Trouser, Pullover, Dress, Coat, Sandal, Shirt,","code":""},{"path":"https://keras3.posit.co/dev/reference/dataset_fashion_mnist.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Fashion-MNIST database of fashion articles — dataset_fashion_mnist","text":"","code":"dataset_fashion_mnist()"},{"path":"https://keras3.posit.co/dev/reference/dataset_fashion_mnist.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Fashion-MNIST database of fashion articles — dataset_fashion_mnist","text":"Lists training test data: train$x, train$y, test$x, test$y, x array grayscale image data shape (num_samples, 28, 28) y array article labels (integers range 0-9) shape (num_samples).","code":""},{"path":"https://keras3.posit.co/dev/reference/dataset_fashion_mnist.html","id":"details","dir":"Reference","previous_headings":"","what":"Details","title":"Fashion-MNIST database of fashion articles — dataset_fashion_mnist","text":"Dataset 60,000 28x28 grayscale images 10 fashion categories, along test set 10,000 images. dataset can used drop-replacement MNIST. class labels : 0 - T-shirt/top 1 - Trouser 2 - Pullover 3 - Dress 4 - Coat 5 - Sandal 6 - Shirt 7 - Sneaker 8 - Bag 9 - Ankle boot","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/dataset_imdb.html","id":null,"dir":"Reference","previous_headings":"","what":"IMDB Movie reviews sentiment classification — dataset_imdb","title":"IMDB Movie reviews sentiment classification — dataset_imdb","text":"Dataset 25,000 movies reviews IMDB, labeled sentiment (positive/negative). Reviews preprocessed, review encoded sequence word indexes (integers). convenience, words indexed overall frequency dataset, instance integer \"3\" encodes 3rd frequent word data. allows quick filtering operations : \"consider top 10,000 common words, eliminate top 20 common words\".","code":""},{"path":"https://keras3.posit.co/dev/reference/dataset_imdb.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"IMDB Movie reviews sentiment classification — dataset_imdb","text":"","code":"dataset_imdb(   path = \"imdb.npz\",   num_words = NULL,   skip_top = 0L,   maxlen = NULL,   seed = 113L,   start_char = 1L,   oov_char = 2L,   index_from = 3L )  dataset_imdb_word_index(path = \"imdb_word_index.json\")"},{"path":"https://keras3.posit.co/dev/reference/dataset_imdb.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"IMDB Movie reviews sentiment classification — dataset_imdb","text":"path cache data (relative ~/.keras/dataset). num_words Max number words include. Words ranked often occur (training set) frequent words kept skip_top Skip top N frequently occuring words (may informative). maxlen sequences longer filtered . seed random seed sample shuffling. start_char start sequence marked character. Set 1 0 usually padding character. oov_char Words cut num_words skip_top limit replaced character. index_from Index actual words index higher.","code":""},{"path":"https://keras3.posit.co/dev/reference/dataset_imdb.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"IMDB Movie reviews sentiment classification — dataset_imdb","text":"Lists training test data: train$x, train$y, test$x, test$y. x data includes integer sequences. num_words argument specific, maximum possible index value num_words-1. maxlen argument specified, largest possible sequence length maxlen. y data includes set integer labels (0 1). dataset_imdb_word_index() function returns list names words values integer.","code":""},{"path":"https://keras3.posit.co/dev/reference/dataset_imdb.html","id":"details","dir":"Reference","previous_headings":"","what":"Details","title":"IMDB Movie reviews sentiment classification — dataset_imdb","text":"convention, \"0\" stand specific word, instead used encode unknown word.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/dataset_mnist.html","id":null,"dir":"Reference","previous_headings":"","what":"MNIST database of handwritten digits — dataset_mnist","title":"MNIST database of handwritten digits — dataset_mnist","text":"Dataset 60,000 28x28 grayscale images 10 digits, along test set 10,000 images.","code":""},{"path":"https://keras3.posit.co/dev/reference/dataset_mnist.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"MNIST database of handwritten digits — dataset_mnist","text":"","code":"dataset_mnist(path = \"mnist.npz\")"},{"path":"https://keras3.posit.co/dev/reference/dataset_mnist.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"MNIST database of handwritten digits — dataset_mnist","text":"path Path cache dataset locally (relative ~/.keras/datasets).","code":""},{"path":"https://keras3.posit.co/dev/reference/dataset_mnist.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"MNIST database of handwritten digits — dataset_mnist","text":"Lists training test data: train$x, train$y, test$x, test$y, x array grayscale image data shape (num_samples, 28, 28) y array digit labels (integers range 0-9) shape (num_samples).","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/dataset_reuters.html","id":null,"dir":"Reference","previous_headings":"","what":"Reuters newswire topics classification — dataset_reuters","title":"Reuters newswire topics classification — dataset_reuters","text":"Dataset 11,228 newswires Reuters, labeled 46 topics. dataset_imdb() , wire encoded sequence word indexes (conventions).","code":""},{"path":"https://keras3.posit.co/dev/reference/dataset_reuters.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Reuters newswire topics classification — dataset_reuters","text":"","code":"dataset_reuters(   path = \"reuters.npz\",   num_words = NULL,   skip_top = 0L,   maxlen = NULL,   test_split = 0.2,   seed = 113L,   start_char = 1L,   oov_char = 2L,   index_from = 3L )  dataset_reuters_word_index(path = \"reuters_word_index.pkl\")"},{"path":"https://keras3.posit.co/dev/reference/dataset_reuters.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Reuters newswire topics classification — dataset_reuters","text":"path cache data (relative ~/.keras/dataset). num_words Max number words include. Words ranked often occur (training set) frequent words kept skip_top Skip top N frequently occuring words (may informative). maxlen Truncate sequences length. test_split Fraction dataset used test data. seed Random seed sample shuffling. start_char start sequence marked character. Set 1 0 usually padding character. oov_char words cut num_words skip_top limit replaced character. index_from index actual words index higher.","code":""},{"path":"https://keras3.posit.co/dev/reference/dataset_reuters.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Reuters newswire topics classification — dataset_reuters","text":"Lists training test data: train$x, train$y, test$x, test$y format dataset_imdb(). dataset_reuters_word_index() function returns list names words values integer. e.g. word_index[[\"giraffe\"]] might return 1234.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/deserialize_keras_object.html","id":null,"dir":"Reference","previous_headings":"","what":"Retrieve the object by deserializing the config dict. — deserialize_keras_object","title":"Retrieve the object by deserializing the config dict. — deserialize_keras_object","text":"config dict Python dictionary consists set key-value pairs, represents Keras object, Optimizer, Layer, Metrics, etc. saving loading library uses following keys record information Keras object: class_name: String. name class, exactly defined source code, \"LossesContainer\". config: Named List. Library-defined user-defined key-value pairs store configuration object, obtained object$get_config(). module: String. path python module. Built-Keras classes expect prefix keras. registered_name: String. key class registered via register_keras_serializable(package, name) API. key format '{package}>{name}', package name arguments passed register_keras_serializable(). name provided, uses class name. registered_name successfully resolves class (registered), class_name config values config dict used. registered_name used non-built-classes. example, following config list represents built-Adam optimizer relevant config:     class exported Keras namespace, library tracks module class_name. example:   following config represents user-customized MeanSquaredError loss:","code":"config <- list(   class_name = \"Adam\",   config = list(     amsgrad = FALSE,     beta_1 = 0.8999999761581421,     beta_2 = 0.9990000128746033,     epsilon = 1e-07,     learning_rate = 0.0010000000474974513,     name = \"Adam\"   ),   module = \"keras.optimizers\",   registered_name = NULL ) # Returns an `Adam` instance identical to the original one. deserialize_keras_object(config) ## <keras.src.optimizers.adam.Adam object at 0x0> config <- list(   class_name = \"MetricsList\",   config =  list(     ...   ),   module = \"keras.trainers.compile_utils\",   registered_name = \"MetricsList\" )  # Returns a `MetricsList` instance identical to the original one. deserialize_keras_object(config) # define a custom object loss_modified_mse <- Loss(   \"ModifiedMeanSquaredError\",   inherit = loss_mean_squared_error)  # register the custom object register_keras_serializable(loss_modified_mse)  # confirm object is registered get_custom_objects() ## $`keras3>ModifiedMeanSquaredError` ## <class '<r-namespace:keras3>.ModifiedMeanSquaredError'> ##  signature: ( ##    reduction='sum_over_batch_size', ##    name='mean_squared_error', ##    dtype=None ## ) get_registered_name(loss_modified_mse) ## [1] \"keras3>ModifiedMeanSquaredError\" # now custom object instances can be serialized full_config <- serialize_keras_object(loss_modified_mse())  # the `config` arguments will be passed to loss_modified_mse() str(full_config) ## List of 4 ##  $ module         : chr \"<r-namespace:keras3>\" ##  $ class_name     : chr \"ModifiedMeanSquaredError\" ##  $ config         :List of 2 ##   ..$ name     : chr \"mean_squared_error\" ##   ..$ reduction: chr \"sum_over_batch_size\" ##  $ registered_name: chr \"keras3>ModifiedMeanSquaredError\" # and custom object instances can be deserialized deserialize_keras_object(full_config) ## <LossFunctionWrapper(<function mean_squared_error at 0x0>, kwargs={})> ##  signature: (y_true, y_pred, sample_weight=None) # Returns the `ModifiedMeanSquaredError` object"},{"path":"https://keras3.posit.co/dev/reference/deserialize_keras_object.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Retrieve the object by deserializing the config dict. — deserialize_keras_object","text":"","code":"deserialize_keras_object(config, custom_objects = NULL, safe_mode = TRUE, ...)"},{"path":"https://keras3.posit.co/dev/reference/deserialize_keras_object.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Retrieve the object by deserializing the config dict. — deserialize_keras_object","text":"config Named list describing object. custom_objects Named list containing mapping custom object names corresponding classes functions. safe_mode Boolean, whether disallow unsafe lambda deserialization. safe_mode=FALSE, loading object potential trigger arbitrary code execution. argument applicable Keras v3 model format. Defaults TRUE. ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/deserialize_keras_object.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Retrieve the object by deserializing the config dict. — deserialize_keras_object","text":"object described config dictionary.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/evaluate.keras.src.models.model.Model.html","id":null,"dir":"Reference","previous_headings":"","what":"Evaluate a Keras Model — evaluate.keras.src.models.model.Model","title":"Evaluate a Keras Model — evaluate.keras.src.models.model.Model","text":"functions returns loss value metrics values model test mode. Computation done batches (see batch_size arg.)","code":""},{"path":"https://keras3.posit.co/dev/reference/evaluate.keras.src.models.model.Model.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Evaluate a Keras Model — evaluate.keras.src.models.model.Model","text":"","code":"# S3 method for class 'keras.src.models.model.Model' evaluate(   object,   x = NULL,   y = NULL,   ...,   batch_size = NULL,   verbose = getOption(\"keras.verbose\", default = \"auto\"),   sample_weight = NULL,   steps = NULL,   callbacks = NULL )"},{"path":"https://keras3.posit.co/dev/reference/evaluate.keras.src.models.model.Model.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Evaluate a Keras Model — evaluate.keras.src.models.model.Model","text":"object Keras model object x Input data. can : R array (array-like), list arrays (case model multiple inputs). backend-native tensor, list tensors (case model multiple inputs). named list mapping input names corresponding array/tensors, model named inputs. tf.data.Dataset. return tuple either (inputs, targets) (inputs, targets, sample_weights). generator returning (inputs, targets) (inputs, targets, sample_weights). y Target data. Like input data x, either R array(s) backend-native tensor(s). x tf.data.Dataset generator function, y specified (since targets obtained iterator/dataset). ... forward/backward compatability. batch_size Integer NULL. Number samples per batch computation. unspecified, batch_size default 32. specify batch_size data form tf dataset generator (since generate batches). verbose \"auto\", 0, 1, 2. Verbosity mode. 0 = silent, 1 = progress bar, 2 = single line. \"auto\" becomes 1 cases, 2 knitr render running distributed training server. Note progress bar particularly useful logged file, verbose=2 recommended running interactively (e.g. production environment). Defaults \"auto\". sample_weight Optional array tensor weights training samples, used weighting loss function (training ). can either pass flat (1D) array tensor length input samples (1:1 mapping weights samples), case temporal data, can pass 2D array tensor shape (samples, sequence_length) apply different weight every timestep every sample. argument supported x tf.data.Dataset, Python generator function. Instead, provide sample_weights third element x. Note sample weighting apply metrics specified via metrics argument compile(). apply sample weighting metrics, can specify via weighted_metrics compile() instead. steps Integer NULL. Total number steps (batches samples) declaring evaluation round finished. Ignored default value NULL. x tf.data.Dataset steps NULL, evaluation run dataset exhausted. case infinitely repeating dataset, run indefinitely. callbacks List Callback instances. List callbacks apply evaluation.","code":""},{"path":"https://keras3.posit.co/dev/reference/evaluate.keras.src.models.model.Model.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Evaluate a Keras Model — evaluate.keras.src.models.model.Model","text":"Scalar test loss (model single output metrics) list scalars (model multiple outputs /metrics). attribute model$metrics_names give display labels scalar outputs.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/export_savedmodel.keras.src.models.model.Model.html","id":null,"dir":"Reference","previous_headings":"","what":"Export the model as an artifact for inference. — export_savedmodel.keras.src.models.model.Model","title":"Export the model as an artifact for inference. — export_savedmodel.keras.src.models.model.Model","text":"(e.g. via TF-Serving). Note: can currently used TensorFlow JAX backends. method lets export model lightweight SavedModel artifact contains model's forward pass (call() method) can served via e.g. TF-Serving. forward pass registered name serve() (see example ). original code model (including custom layers may used) longer necessary reload artifact – entirely standalone. Note: feature currently supported TensorFlow, JAX Torch backends. Note: aware exported artifact may contain information local file system using format=\"onnx\", verbose=TRUE Torch backend.","code":""},{"path":"https://keras3.posit.co/dev/reference/export_savedmodel.keras.src.models.model.Model.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Export the model as an artifact for inference. — export_savedmodel.keras.src.models.model.Model","text":"","code":"# S3 method for class 'keras.src.models.model.Model' export_savedmodel(   object,   export_dir_base,   ...,   format = \"tf_saved_model\",   verbose = NULL,   input_signature = NULL )"},{"path":"https://keras3.posit.co/dev/reference/export_savedmodel.keras.src.models.model.Model.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Export the model as an artifact for inference. — export_savedmodel.keras.src.models.model.Model","text":"object keras model. export_dir_base string, file path save artifact. ... Additional keyword arguments: Specific JAX backend format=\"tf_saved_model\": is_static: Optional bool. Indicates whether fn static. Set FALSE fn involves state updates (e.g., RNG seeds counters). jax2tf_kwargs: Optional dict. Arguments jax2tf.convert. See documentation jax2tf.convert. native_serialization polymorphic_shapes provided, automatically computed. format string. export format. Supported values: \"tf_saved_model\" \"onnx\".  Defaults \"tf_saved_model\". verbose Bool. Whether print variables exported model. Defaults NULL, uses default value set different backends formats. input_signature Optional. Specifies shape dtype model inputs. Can structure keras.InputSpec, tf.TensorSpec, backend.KerasTensor, backend tensor. provided, automatically computed. Defaults NULL.","code":""},{"path":"https://keras3.posit.co/dev/reference/export_savedmodel.keras.src.models.model.Model.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Export the model as an artifact for inference. — export_savedmodel.keras.src.models.model.Model","text":"called primarily side effect exporting object. first argument, object also returned, invisibly, enable usage pipe.","code":""},{"path":"https://keras3.posit.co/dev/reference/export_savedmodel.keras.src.models.model.Model.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Export the model as an artifact for inference. — export_savedmodel.keras.src.models.model.Model","text":"export ONNX inference.","code":"# Create the artifact model |> tensorflow::export_savedmodel(\"path/to/location\")  # Later, in a different process/environment... library(tensorflow) reloaded_artifact <- tf$saved_model$load(\"path/to/location\") predictions <- reloaded_artifact$serve(input_data)  # see tfdeploy::serve_savedmodel() for serving a model over a local web api. # Export the model as a ONNX artifact model |> export_savedmodel(\"path/to/location\", format = \"onnx\")  # Load the artifact in a different process/environment onnxruntime <- reticulate::import(\"onnxruntime\") ort_session <- onnxruntime$InferenceSession(\"path/to/location\") input_data <- list(....) names(input_data) <- sapply(ort_session$get_inputs(), `[[`, \"name\") predictions <- ort_session$run(NULL, input_data)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/fit.keras.src.models.model.Model.html","id":null,"dir":"Reference","previous_headings":"","what":"Train a model for a fixed number of epochs (dataset iterations). — fit.keras.src.models.model.Model","title":"Train a model for a fixed number of epochs (dataset iterations). — fit.keras.src.models.model.Model","text":"Train model fixed number epochs (dataset iterations).","code":""},{"path":"https://keras3.posit.co/dev/reference/fit.keras.src.models.model.Model.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Train a model for a fixed number of epochs (dataset iterations). — fit.keras.src.models.model.Model","text":"","code":"# S3 method for class 'keras.src.models.model.Model' fit(   object,   x = NULL,   y = NULL,   ...,   batch_size = NULL,   epochs = 1L,   callbacks = NULL,   validation_split = 0,   validation_data = NULL,   shuffle = TRUE,   class_weight = NULL,   sample_weight = NULL,   initial_epoch = 1L,   steps_per_epoch = NULL,   validation_steps = NULL,   validation_batch_size = NULL,   validation_freq = 1L,   verbose = getOption(\"keras.verbose\", default = \"auto\"),   view_metrics = getOption(\"keras.view_metrics\", default = \"auto\") )"},{"path":"https://keras3.posit.co/dev/reference/fit.keras.src.models.model.Model.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Train a model for a fixed number of epochs (dataset iterations). — fit.keras.src.models.model.Model","text":"object Keras model object x Input data. can : array (array-like), list arrays (case model multiple inputs). backend-native tensor, list tensors (case model multiple inputs). named list mapping input names corresponding array/tensors, model named inputs. tf.data.Dataset. return tuple either (inputs, targets) (inputs, targets, sample_weights). generator returning (inputs, targets) (inputs, targets, sample_weights). y Target data. Like input data x, can either array(s) backend-native tensor(s). x TF Dataset generator, y specified (since targets obtained x). ... Additional arguments passed model fit() method. batch_size Integer NULL. Number samples per gradient update. unspecified, batch_size default 32. specify batch_size data form TF Datasets generators, (since generate batches). epochs Integer. Number epochs train model. epoch iteration entire x y data provided (unless steps_per_epoch flag set something NULL). Note conjunction initial_epoch, epochs understood \"final epoch\". model trained number iterations given epochs, merely epoch index epochs reached. callbacks List Callback() instances. List callbacks apply training. See callback_*. validation_split Fraction training data used validation data. model set apart fraction training data, train , evaluate loss model metrics data end epoch. validation data selected last samples x y data provided, shuffling. argument supported x y made arrays tensors. validation_data validation_split provided, validation_data override validation_split. validation_data Data evaluate loss model metrics end epoch. model trained data. Thus, note fact validation loss data provided using validation_split validation_data affected regularization layers like noise dropout. validation_data override validation_split. can : tuple (x_val, y_val) arrays tensors. tuple (x_val, y_val, val_sample_weights) arrays. generator returning (inputs, targets) (inputs, targets, sample_weights). shuffle Boolean, whether shuffle training data epoch. argument ignored x generator TF Dataset. class_weight Optional named list mapping class indices (integers, 0-based) weight (float) value, used weighting loss function (training ). can useful tell model \"pay attention\" samples -represented class. class_weight specified targets rank 2 greater, either y must one-hot encoded, explicit final dimension 1 must included sparse class labels. sample_weight Optional array tensor weights training samples, used weighting loss function (training ). can either pass flat (1D) array tensor length input samples (1:1 mapping weights samples), case temporal data, can pass 2D array tensor shape (samples, sequence_length) apply different weight every timestep every sample. argument supported x tf.data.Dataset, Python generator function. Instead, provide sample_weights third element x. Note sample weighting apply metrics specified via metrics argument compile(). apply sample weighting metrics, can specify via weighted_metrics compile() instead. initial_epoch Integer. Epoch start training (useful resuming previous training run). steps_per_epoch Integer NULL. Total number steps (batches samples) declaring one epoch finished starting next epoch. training input tensors arrays, default NULL means value used number samples dataset divided batch size, 1 determined. x tf.data.Dataset Python generator function, epoch run input dataset exhausted. passing infinitely repeating dataset, must specify steps_per_epoch argument, otherwise training run indefinitely. validation_steps relevant validation_data provided. Total number steps (batches samples) draw stopping performing validation end every epoch. validation_steps NULL, validation run validation_data dataset exhausted. case infinitely repeating dataset, run indefinitely. validation_steps specified part dataset consumed, evaluation start beginning dataset epoch. ensures validation samples used every time. validation_batch_size Integer NULL. Number samples per validation batch. unspecified, default batch_size. specify validation_batch_size data form TF Datasets generator instances (since generate batches). validation_freq relevant validation data provided. Specifies many training epochs run new validation run performed, e.g. validation_freq=2 runs validation every 2 epochs. verbose \"auto\", 0, 1, 2. Verbosity mode. 0 = silent, 1 = progress bar, 2 = one line per epoch. \"auto\" becomes 1 cases, 2 knitr render running distributed training server. Note progress bar particularly useful logged file, verbose=2 recommended running interactively (e.g., production environment). Defaults \"auto\". view_metrics View realtime plot training metrics (epoch). default (\"auto\") display plot running within RStudio, metrics specified model compile(), epochs > 1 verbose > 0. Set global options(keras.view_metrics = ) option establish different default.","code":""},{"path":"https://keras3.posit.co/dev/reference/fit.keras.src.models.model.Model.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Train a model for a fixed number of epochs (dataset iterations). — fit.keras.src.models.model.Model","text":"keras_training_history object, named list: list(params = <params>, metrics = <metrics>\"), S3 methods print(), plot(), .data.frame(). metrics field record training loss values metrics values successive epochs, well validation loss values validation metrics values (applicable).","code":""},{"path":"https://keras3.posit.co/dev/reference/fit.keras.src.models.model.Model.html","id":"details","dir":"Reference","previous_headings":"","what":"Details","title":"Train a model for a fixed number of epochs (dataset iterations). — fit.keras.src.models.model.Model","text":"Unpacking behavior iterator-like inputs: common pattern pass iterator like object tf.data.Dataset generator fit(), fact yield features (x) optionally targets (y) sample weights (sample_weight). Keras requires output iterator-likes unambiguous. iterator return tuple() length 1, 2, 3, optional second third elements used y sample_weight respectively. type provided wrapped length-one tuple(), effectively treating everything x. yielding named lists, still adhere top-level tuple structure, e.g. tuple(list(x0 = x0, x = x1), y). Keras attempt separate features, targets, weights keys single dict.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/freeze_weights.html","id":null,"dir":"Reference","previous_headings":"","what":"Freeze and unfreeze weights — freeze_weights","title":"Freeze and unfreeze weights — freeze_weights","text":"Freeze weights model layer longer trainable.","code":""},{"path":"https://keras3.posit.co/dev/reference/freeze_weights.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Freeze and unfreeze weights — freeze_weights","text":"","code":"freeze_weights(object, from = NULL, to = NULL, which = NULL)  unfreeze_weights(object, from = NULL, to = NULL, which = NULL)"},{"path":"https://keras3.posit.co/dev/reference/freeze_weights.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Freeze and unfreeze weights — freeze_weights","text":"object Keras model layer object Layer instance, layer name, layer index within model Layer instance, layer name, layer index within model layer names, integer positions, layers, logical vector (length(object$layers)), function returning logical vector.","code":""},{"path":"https://keras3.posit.co/dev/reference/freeze_weights.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Freeze and unfreeze weights — freeze_weights","text":"input object frozen weights returned, invisibly. Note, object modified place, return value provided make usage pipe convenient.","code":""},{"path":"https://keras3.posit.co/dev/reference/freeze_weights.html","id":"note","dir":"Reference","previous_headings":"","what":"Note","title":"Freeze and unfreeze weights — freeze_weights","text":"layer arguments inclusive. applied model, freeze unfreeze global operation layers model (.e. layers within specified range set opposite value, e.g. unfrozen call freeze). Models must compiled weights frozen unfrozen.","code":""},{"path":"https://keras3.posit.co/dev/reference/freeze_weights.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Freeze and unfreeze weights — freeze_weights","text":"","code":"# instantiate a VGG16 model conv_base <- application_vgg16(   weights = \"imagenet\",   include_top = FALSE,   input_shape = c(150, 150, 3) )  # freeze it's weights freeze_weights(conv_base)  # Note the \"Trainable\" column conv_base ## Model: \"vgg16\" ## +-----------------------------+-----------------------+------------+-------+ ## | Layer (type)                | Output Shape          |    Param # | Trai… | ## +=============================+=======================+============+=======+ ## | input_layer (InputLayer)    | (None, 150, 150, 3)   |          0 |   -   | ## +-----------------------------+-----------------------+------------+-------+ ## | block1_conv1 (Conv2D)       | (None, 150, 150, 64)  |      1,792 |   N   | ## +-----------------------------+-----------------------+------------+-------+ ## | block1_conv2 (Conv2D)       | (None, 150, 150, 64)  |     36,928 |   N   | ## +-----------------------------+-----------------------+------------+-------+ ## | block1_pool (MaxPooling2D)  | (None, 75, 75, 64)    |          0 |   -   | ## +-----------------------------+-----------------------+------------+-------+ ## | block2_conv1 (Conv2D)       | (None, 75, 75, 128)   |     73,856 |   N   | ## +-----------------------------+-----------------------+------------+-------+ ## | block2_conv2 (Conv2D)       | (None, 75, 75, 128)   |    147,584 |   N   | ## +-----------------------------+-----------------------+------------+-------+ ## | block2_pool (MaxPooling2D)  | (None, 37, 37, 128)   |          0 |   -   | ## +-----------------------------+-----------------------+------------+-------+ ## | block3_conv1 (Conv2D)       | (None, 37, 37, 256)   |    295,168 |   N   | ## +-----------------------------+-----------------------+------------+-------+ ## | block3_conv2 (Conv2D)       | (None, 37, 37, 256)   |    590,080 |   N   | ## +-----------------------------+-----------------------+------------+-------+ ## | block3_conv3 (Conv2D)       | (None, 37, 37, 256)   |    590,080 |   N   | ## +-----------------------------+-----------------------+------------+-------+ ## | block3_pool (MaxPooling2D)  | (None, 18, 18, 256)   |          0 |   -   | ## +-----------------------------+-----------------------+------------+-------+ ## | block4_conv1 (Conv2D)       | (None, 18, 18, 512)   |  1,180,160 |   N   | ## +-----------------------------+-----------------------+------------+-------+ ## | block4_conv2 (Conv2D)       | (None, 18, 18, 512)   |  2,359,808 |   N   | ## +-----------------------------+-----------------------+------------+-------+ ## | block4_conv3 (Conv2D)       | (None, 18, 18, 512)   |  2,359,808 |   N   | ## +-----------------------------+-----------------------+------------+-------+ ## | block4_pool (MaxPooling2D)  | (None, 9, 9, 512)     |          0 |   -   | ## +-----------------------------+-----------------------+------------+-------+ ## | block5_conv1 (Conv2D)       | (None, 9, 9, 512)     |  2,359,808 |   N   | ## +-----------------------------+-----------------------+------------+-------+ ## | block5_conv2 (Conv2D)       | (None, 9, 9, 512)     |  2,359,808 |   N   | ## +-----------------------------+-----------------------+------------+-------+ ## | block5_conv3 (Conv2D)       | (None, 9, 9, 512)     |  2,359,808 |   N   | ## +-----------------------------+-----------------------+------------+-------+ ## | block5_pool (MaxPooling2D)  | (None, 4, 4, 512)     |          0 |   -   | ## +-----------------------------+-----------------------+------------+-------+ ##  Total params: 14,714,688 (56.13 MB) ##  Trainable params: 0 (0.00 B) ##  Non-trainable params: 14,714,688 (56.13 MB) # create a composite model that includes the base + more layers model <- keras_model_sequential(input_batch_shape = shape(conv_base$input)) |>   conv_base() |>   layer_flatten() |>   layer_dense(units = 256, activation = \"relu\") |>   layer_dense(units = 1, activation = \"sigmoid\")  # compile model |> compile(   loss = \"binary_crossentropy\",   optimizer = optimizer_rmsprop(learning_rate = 2e-5),   metrics = c(\"accuracy\") )  model ## Model: \"sequential\" ## +-----------------------------+-----------------------+------------+-------+ ## | Layer (type)                | Output Shape          |    Param # | Trai… | ## +=============================+=======================+============+=======+ ## | vgg16 (Functional)          | (None, 4, 4, 512)     | 14,714,688 |   N   | ## +-----------------------------+-----------------------+------------+-------+ ## | flatten (Flatten)           | (None, 8192)          |          0 |   -   | ## +-----------------------------+-----------------------+------------+-------+ ## | dense (Dense)               | (None, 256)           |  2,097,408 |   Y   | ## +-----------------------------+-----------------------+------------+-------+ ## | dense_1 (Dense)             | (None, 1)             |        257 |   Y   | ## +-----------------------------+-----------------------+------------+-------+ ##  Total params: 16,812,353 (64.13 MB) ##  Trainable params: 2,097,665 (8.00 MB) ##  Non-trainable params: 14,714,688 (56.13 MB) print(model, expand_nested = TRUE) ## Model: \"sequential\" ## +-----------------------------+-----------------------+------------+-------+ ## | Layer (type)                | Output Shape          |    Param # | Trai… | ## +=============================+=======================+============+=======+ ## | vgg16 (Functional)          | (None, 4, 4, 512)     | 14,714,688 |   N   | ## +-----------------------------+-----------------------+------------+-------+ ## |    > input_layer            | (None, 150, 150, 3)   |          0 |   -   | ## | (InputLayer)                |                       |            |       | ## +-----------------------------+-----------------------+------------+-------+ ## |    > block1_conv1 (Conv2D)  | (None, 150, 150, 64)  |      1,792 |   N   | ## +-----------------------------+-----------------------+------------+-------+ ## |    > block1_conv2 (Conv2D)  | (None, 150, 150, 64)  |     36,928 |   N   | ## +-----------------------------+-----------------------+------------+-------+ ## |    > block1_pool            | (None, 75, 75, 64)    |          0 |   -   | ## | (MaxPooling2D)              |                       |            |       | ## +-----------------------------+-----------------------+------------+-------+ ## |    > block2_conv1 (Conv2D)  | (None, 75, 75, 128)   |     73,856 |   N   | ## +-----------------------------+-----------------------+------------+-------+ ## |    > block2_conv2 (Conv2D)  | (None, 75, 75, 128)   |    147,584 |   N   | ## +-----------------------------+-----------------------+------------+-------+ ## |    > block2_pool            | (None, 37, 37, 128)   |          0 |   -   | ## | (MaxPooling2D)              |                       |            |       | ## +-----------------------------+-----------------------+------------+-------+ ## |    > block3_conv1 (Conv2D)  | (None, 37, 37, 256)   |    295,168 |   N   | ## +-----------------------------+-----------------------+------------+-------+ ## |    > block3_conv2 (Conv2D)  | (None, 37, 37, 256)   |    590,080 |   N   | ## +-----------------------------+-----------------------+------------+-------+ ## |    > block3_conv3 (Conv2D)  | (None, 37, 37, 256)   |    590,080 |   N   | ## +-----------------------------+-----------------------+------------+-------+ ## |    > block3_pool            | (None, 18, 18, 256)   |          0 |   -   | ## | (MaxPooling2D)              |                       |            |       | ## +-----------------------------+-----------------------+------------+-------+ ## |    > block4_conv1 (Conv2D)  | (None, 18, 18, 512)   |  1,180,160 |   N   | ## +-----------------------------+-----------------------+------------+-------+ ## |    > block4_conv2 (Conv2D)  | (None, 18, 18, 512)   |  2,359,808 |   N   | ## +-----------------------------+-----------------------+------------+-------+ ## |    > block4_conv3 (Conv2D)  | (None, 18, 18, 512)   |  2,359,808 |   N   | ## +-----------------------------+-----------------------+------------+-------+ ## |    > block4_pool            | (None, 9, 9, 512)     |          0 |   -   | ## | (MaxPooling2D)              |                       |            |       | ## +-----------------------------+-----------------------+------------+-------+ ## |    > block5_conv1 (Conv2D)  | (None, 9, 9, 512)     |  2,359,808 |   N   | ## +-----------------------------+-----------------------+------------+-------+ ## |    > block5_conv2 (Conv2D)  | (None, 9, 9, 512)     |  2,359,808 |   N   | ## +-----------------------------+-----------------------+------------+-------+ ## |    > block5_conv3 (Conv2D)  | (None, 9, 9, 512)     |  2,359,808 |   N   | ## +-----------------------------+-----------------------+------------+-------+ ## |    > block5_pool            | (None, 4, 4, 512)     |          0 |   -   | ## | (MaxPooling2D)              |                       |            |       | ## +-----------------------------+-----------------------+------------+-------+ ## | flatten (Flatten)           | (None, 8192)          |          0 |   -   | ## +-----------------------------+-----------------------+------------+-------+ ## | dense (Dense)               | (None, 256)           |  2,097,408 |   Y   | ## +-----------------------------+-----------------------+------------+-------+ ## | dense_1 (Dense)             | (None, 1)             |        257 |   Y   | ## +-----------------------------+-----------------------+------------+-------+ ##  Total params: 16,812,353 (64.13 MB) ##  Trainable params: 2,097,665 (8.00 MB) ##  Non-trainable params: 14,714,688 (56.13 MB) # unfreeze weights from \"block5_conv1\" on unfreeze_weights(conv_base, from = \"block5_conv1\")  # compile again since we froze or unfroze weights model |> compile(   loss = \"binary_crossentropy\",   optimizer = optimizer_rmsprop(learning_rate = 2e-5),   metrics = c(\"accuracy\") )  conv_base ## Model: \"vgg16\" ## +-----------------------------+-----------------------+------------+-------+ ## | Layer (type)                | Output Shape          |    Param # | Trai… | ## +=============================+=======================+============+=======+ ## | input_layer (InputLayer)    | (None, 150, 150, 3)   |          0 |   -   | ## +-----------------------------+-----------------------+------------+-------+ ## | block1_conv1 (Conv2D)       | (None, 150, 150, 64)  |      1,792 |   N   | ## +-----------------------------+-----------------------+------------+-------+ ## | block1_conv2 (Conv2D)       | (None, 150, 150, 64)  |     36,928 |   N   | ## +-----------------------------+-----------------------+------------+-------+ ## | block1_pool (MaxPooling2D)  | (None, 75, 75, 64)    |          0 |   -   | ## +-----------------------------+-----------------------+------------+-------+ ## | block2_conv1 (Conv2D)       | (None, 75, 75, 128)   |     73,856 |   N   | ## +-----------------------------+-----------------------+------------+-------+ ## | block2_conv2 (Conv2D)       | (None, 75, 75, 128)   |    147,584 |   N   | ## +-----------------------------+-----------------------+------------+-------+ ## | block2_pool (MaxPooling2D)  | (None, 37, 37, 128)   |          0 |   -   | ## +-----------------------------+-----------------------+------------+-------+ ## | block3_conv1 (Conv2D)       | (None, 37, 37, 256)   |    295,168 |   N   | ## +-----------------------------+-----------------------+------------+-------+ ## | block3_conv2 (Conv2D)       | (None, 37, 37, 256)   |    590,080 |   N   | ## +-----------------------------+-----------------------+------------+-------+ ## | block3_conv3 (Conv2D)       | (None, 37, 37, 256)   |    590,080 |   N   | ## +-----------------------------+-----------------------+------------+-------+ ## | block3_pool (MaxPooling2D)  | (None, 18, 18, 256)   |          0 |   -   | ## +-----------------------------+-----------------------+------------+-------+ ## | block4_conv1 (Conv2D)       | (None, 18, 18, 512)   |  1,180,160 |   N   | ## +-----------------------------+-----------------------+------------+-------+ ## | block4_conv2 (Conv2D)       | (None, 18, 18, 512)   |  2,359,808 |   N   | ## +-----------------------------+-----------------------+------------+-------+ ## | block4_conv3 (Conv2D)       | (None, 18, 18, 512)   |  2,359,808 |   N   | ## +-----------------------------+-----------------------+------------+-------+ ## | block4_pool (MaxPooling2D)  | (None, 9, 9, 512)     |          0 |   -   | ## +-----------------------------+-----------------------+------------+-------+ ## | block5_conv1 (Conv2D)       | (None, 9, 9, 512)     |  2,359,808 |   Y   | ## +-----------------------------+-----------------------+------------+-------+ ## | block5_conv2 (Conv2D)       | (None, 9, 9, 512)     |  2,359,808 |   Y   | ## +-----------------------------+-----------------------+------------+-------+ ## | block5_conv3 (Conv2D)       | (None, 9, 9, 512)     |  2,359,808 |   Y   | ## +-----------------------------+-----------------------+------------+-------+ ## | block5_pool (MaxPooling2D)  | (None, 4, 4, 512)     |          0 |   -   | ## +-----------------------------+-----------------------+------------+-------+ ##  Total params: 14,714,688 (56.13 MB) ##  Trainable params: 7,079,424 (27.01 MB) ##  Non-trainable params: 7,635,264 (29.13 MB) print(model, expand_nested = TRUE) ## Model: \"sequential\" ## +-----------------------------+-----------------------+------------+-------+ ## | Layer (type)                | Output Shape          |    Param # | Trai… | ## +=============================+=======================+============+=======+ ## | vgg16 (Functional)          | (None, 4, 4, 512)     | 14,714,688 |   Y   | ## +-----------------------------+-----------------------+------------+-------+ ## |    > input_layer            | (None, 150, 150, 3)   |          0 |   -   | ## | (InputLayer)                |                       |            |       | ## +-----------------------------+-----------------------+------------+-------+ ## |    > block1_conv1 (Conv2D)  | (None, 150, 150, 64)  |      1,792 |   N   | ## +-----------------------------+-----------------------+------------+-------+ ## |    > block1_conv2 (Conv2D)  | (None, 150, 150, 64)  |     36,928 |   N   | ## +-----------------------------+-----------------------+------------+-------+ ## |    > block1_pool            | (None, 75, 75, 64)    |          0 |   -   | ## | (MaxPooling2D)              |                       |            |       | ## +-----------------------------+-----------------------+------------+-------+ ## |    > block2_conv1 (Conv2D)  | (None, 75, 75, 128)   |     73,856 |   N   | ## +-----------------------------+-----------------------+------------+-------+ ## |    > block2_conv2 (Conv2D)  | (None, 75, 75, 128)   |    147,584 |   N   | ## +-----------------------------+-----------------------+------------+-------+ ## |    > block2_pool            | (None, 37, 37, 128)   |          0 |   -   | ## | (MaxPooling2D)              |                       |            |       | ## +-----------------------------+-----------------------+------------+-------+ ## |    > block3_conv1 (Conv2D)  | (None, 37, 37, 256)   |    295,168 |   N   | ## +-----------------------------+-----------------------+------------+-------+ ## |    > block3_conv2 (Conv2D)  | (None, 37, 37, 256)   |    590,080 |   N   | ## +-----------------------------+-----------------------+------------+-------+ ## |    > block3_conv3 (Conv2D)  | (None, 37, 37, 256)   |    590,080 |   N   | ## +-----------------------------+-----------------------+------------+-------+ ## |    > block3_pool            | (None, 18, 18, 256)   |          0 |   -   | ## | (MaxPooling2D)              |                       |            |       | ## +-----------------------------+-----------------------+------------+-------+ ## |    > block4_conv1 (Conv2D)  | (None, 18, 18, 512)   |  1,180,160 |   N   | ## +-----------------------------+-----------------------+------------+-------+ ## |    > block4_conv2 (Conv2D)  | (None, 18, 18, 512)   |  2,359,808 |   N   | ## +-----------------------------+-----------------------+------------+-------+ ## |    > block4_conv3 (Conv2D)  | (None, 18, 18, 512)   |  2,359,808 |   N   | ## +-----------------------------+-----------------------+------------+-------+ ## |    > block4_pool            | (None, 9, 9, 512)     |          0 |   -   | ## | (MaxPooling2D)              |                       |            |       | ## +-----------------------------+-----------------------+------------+-------+ ## |    > block5_conv1 (Conv2D)  | (None, 9, 9, 512)     |  2,359,808 |   Y   | ## +-----------------------------+-----------------------+------------+-------+ ## |    > block5_conv2 (Conv2D)  | (None, 9, 9, 512)     |  2,359,808 |   Y   | ## +-----------------------------+-----------------------+------------+-------+ ## |    > block5_conv3 (Conv2D)  | (None, 9, 9, 512)     |  2,359,808 |   Y   | ## +-----------------------------+-----------------------+------------+-------+ ## |    > block5_pool            | (None, 4, 4, 512)     |          0 |   -   | ## | (MaxPooling2D)              |                       |            |       | ## +-----------------------------+-----------------------+------------+-------+ ## | flatten (Flatten)           | (None, 8192)          |          0 |   -   | ## +-----------------------------+-----------------------+------------+-------+ ## | dense (Dense)               | (None, 256)           |  2,097,408 |   Y   | ## +-----------------------------+-----------------------+------------+-------+ ## | dense_1 (Dense)             | (None, 1)             |        257 |   Y   | ## +-----------------------------+-----------------------+------------+-------+ ##  Total params: 16,812,353 (64.13 MB) ##  Trainable params: 9,177,089 (35.01 MB) ##  Non-trainable params: 7,635,264 (29.13 MB) # freeze only the last 5 layers freeze_weights(conv_base, from = -5) conv_base ## Model: \"vgg16\" ## +-----------------------------+-----------------------+------------+-------+ ## | Layer (type)                | Output Shape          |    Param # | Trai… | ## +=============================+=======================+============+=======+ ## | input_layer (InputLayer)    | (None, 150, 150, 3)   |          0 |   -   | ## +-----------------------------+-----------------------+------------+-------+ ## | block1_conv1 (Conv2D)       | (None, 150, 150, 64)  |      1,792 |   Y   | ## +-----------------------------+-----------------------+------------+-------+ ## | block1_conv2 (Conv2D)       | (None, 150, 150, 64)  |     36,928 |   Y   | ## +-----------------------------+-----------------------+------------+-------+ ## | block1_pool (MaxPooling2D)  | (None, 75, 75, 64)    |          0 |   -   | ## +-----------------------------+-----------------------+------------+-------+ ## | block2_conv1 (Conv2D)       | (None, 75, 75, 128)   |     73,856 |   Y   | ## +-----------------------------+-----------------------+------------+-------+ ## | block2_conv2 (Conv2D)       | (None, 75, 75, 128)   |    147,584 |   Y   | ## +-----------------------------+-----------------------+------------+-------+ ## | block2_pool (MaxPooling2D)  | (None, 37, 37, 128)   |          0 |   -   | ## +-----------------------------+-----------------------+------------+-------+ ## | block3_conv1 (Conv2D)       | (None, 37, 37, 256)   |    295,168 |   Y   | ## +-----------------------------+-----------------------+------------+-------+ ## | block3_conv2 (Conv2D)       | (None, 37, 37, 256)   |    590,080 |   Y   | ## +-----------------------------+-----------------------+------------+-------+ ## | block3_conv3 (Conv2D)       | (None, 37, 37, 256)   |    590,080 |   Y   | ## +-----------------------------+-----------------------+------------+-------+ ## | block3_pool (MaxPooling2D)  | (None, 18, 18, 256)   |          0 |   -   | ## +-----------------------------+-----------------------+------------+-------+ ## | block4_conv1 (Conv2D)       | (None, 18, 18, 512)   |  1,180,160 |   Y   | ## +-----------------------------+-----------------------+------------+-------+ ## | block4_conv2 (Conv2D)       | (None, 18, 18, 512)   |  2,359,808 |   Y   | ## +-----------------------------+-----------------------+------------+-------+ ## | block4_conv3 (Conv2D)       | (None, 18, 18, 512)   |  2,359,808 |   Y   | ## +-----------------------------+-----------------------+------------+-------+ ## | block4_pool (MaxPooling2D)  | (None, 9, 9, 512)     |          0 |   -   | ## +-----------------------------+-----------------------+------------+-------+ ## | block5_conv1 (Conv2D)       | (None, 9, 9, 512)     |  2,359,808 |   N   | ## +-----------------------------+-----------------------+------------+-------+ ## | block5_conv2 (Conv2D)       | (None, 9, 9, 512)     |  2,359,808 |   N   | ## +-----------------------------+-----------------------+------------+-------+ ## | block5_conv3 (Conv2D)       | (None, 9, 9, 512)     |  2,359,808 |   N   | ## +-----------------------------+-----------------------+------------+-------+ ## | block5_pool (MaxPooling2D)  | (None, 4, 4, 512)     |          0 |   -   | ## +-----------------------------+-----------------------+------------+-------+ ##  Total params: 14,714,688 (56.13 MB) ##  Trainable params: 7,635,264 (29.13 MB) ##  Non-trainable params: 7,079,424 (27.01 MB) # freeze only the last 5 layers, a different way unfreeze_weights(conv_base, to = -6) conv_base ## Model: \"vgg16\" ## +-----------------------------+-----------------------+------------+-------+ ## | Layer (type)                | Output Shape          |    Param # | Trai… | ## +=============================+=======================+============+=======+ ## | input_layer (InputLayer)    | (None, 150, 150, 3)   |          0 |   -   | ## +-----------------------------+-----------------------+------------+-------+ ## | block1_conv1 (Conv2D)       | (None, 150, 150, 64)  |      1,792 |   Y   | ## +-----------------------------+-----------------------+------------+-------+ ## | block1_conv2 (Conv2D)       | (None, 150, 150, 64)  |     36,928 |   Y   | ## +-----------------------------+-----------------------+------------+-------+ ## | block1_pool (MaxPooling2D)  | (None, 75, 75, 64)    |          0 |   -   | ## +-----------------------------+-----------------------+------------+-------+ ## | block2_conv1 (Conv2D)       | (None, 75, 75, 128)   |     73,856 |   Y   | ## +-----------------------------+-----------------------+------------+-------+ ## | block2_conv2 (Conv2D)       | (None, 75, 75, 128)   |    147,584 |   Y   | ## +-----------------------------+-----------------------+------------+-------+ ## | block2_pool (MaxPooling2D)  | (None, 37, 37, 128)   |          0 |   -   | ## +-----------------------------+-----------------------+------------+-------+ ## | block3_conv1 (Conv2D)       | (None, 37, 37, 256)   |    295,168 |   Y   | ## +-----------------------------+-----------------------+------------+-------+ ## | block3_conv2 (Conv2D)       | (None, 37, 37, 256)   |    590,080 |   Y   | ## +-----------------------------+-----------------------+------------+-------+ ## | block3_conv3 (Conv2D)       | (None, 37, 37, 256)   |    590,080 |   Y   | ## +-----------------------------+-----------------------+------------+-------+ ## | block3_pool (MaxPooling2D)  | (None, 18, 18, 256)   |          0 |   -   | ## +-----------------------------+-----------------------+------------+-------+ ## | block4_conv1 (Conv2D)       | (None, 18, 18, 512)   |  1,180,160 |   Y   | ## +-----------------------------+-----------------------+------------+-------+ ## | block4_conv2 (Conv2D)       | (None, 18, 18, 512)   |  2,359,808 |   Y   | ## +-----------------------------+-----------------------+------------+-------+ ## | block4_conv3 (Conv2D)       | (None, 18, 18, 512)   |  2,359,808 |   Y   | ## +-----------------------------+-----------------------+------------+-------+ ## | block4_pool (MaxPooling2D)  | (None, 9, 9, 512)     |          0 |   -   | ## +-----------------------------+-----------------------+------------+-------+ ## | block5_conv1 (Conv2D)       | (None, 9, 9, 512)     |  2,359,808 |   N   | ## +-----------------------------+-----------------------+------------+-------+ ## | block5_conv2 (Conv2D)       | (None, 9, 9, 512)     |  2,359,808 |   N   | ## +-----------------------------+-----------------------+------------+-------+ ## | block5_conv3 (Conv2D)       | (None, 9, 9, 512)     |  2,359,808 |   N   | ## +-----------------------------+-----------------------+------------+-------+ ## | block5_pool (MaxPooling2D)  | (None, 4, 4, 512)     |          0 |   -   | ## +-----------------------------+-----------------------+------------+-------+ ##  Total params: 14,714,688 (56.13 MB) ##  Trainable params: 7,635,264 (29.13 MB) ##  Non-trainable params: 7,079,424 (27.01 MB) # Freeze only layers of a certain type, e.g, BatchNorm layers # batch_norm_layer_class_name <- class(layer_batch_normalization())[1] # is_batch_norm_layer <- function(x) inherits(x, batch_norm_layer_class_name) is_batch_norm_layer <- function(x) inherits(x, keras$layers$BatchNormalization)  model <- application_efficientnet_v2b0() freeze_weights(model, which = is_batch_norm_layer) # print(model)  # equivalent to: for(layer in model$layers) {   if(is_batch_norm_layer(layer))     layer$trainable <- FALSE   else     layer$trainable <- TRUE }"},{"path":"https://keras3.posit.co/dev/reference/get_config.html","id":null,"dir":"Reference","previous_headings":"","what":"Layer/Model configuration — get_config","title":"Layer/Model configuration — get_config","text":"layer config object returned get_config() contains configuration layer model. layer model can reinstantiated later (without trained weights) configuration using from_config(). config include connectivity information, class name (handled externally).","code":""},{"path":"https://keras3.posit.co/dev/reference/get_config.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Layer/Model configuration — get_config","text":"","code":"get_config(object)  from_config(config, custom_objects = NULL)"},{"path":"https://keras3.posit.co/dev/reference/get_config.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Layer/Model configuration — get_config","text":"object Layer model object config Object layer model configuration custom_objects list custom objects needed instantiate layer, e.g., custom layers defined new_layer_class() similar.","code":""},{"path":"https://keras3.posit.co/dev/reference/get_config.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Layer/Model configuration — get_config","text":"get_config() returns object configuration, from_config() returns re-instantiation object.","code":""},{"path":"https://keras3.posit.co/dev/reference/get_config.html","id":"note","dir":"Reference","previous_headings":"","what":"Note","title":"Layer/Model configuration — get_config","text":"Objects returned get_config() serializable via RDS. want save restore model across sessions, can use save_model_config() (model configuration , weights) save_model() save model configuration weights filesystem.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/get_custom_objects.html","id":null,"dir":"Reference","previous_headings":"","what":"Get/set the currently registered custom objects. — get_custom_objects","title":"Get/set the currently registered custom objects. — get_custom_objects","text":"Custom objects set using custom_object_scope() added global list custom objects, appear returned list.","code":""},{"path":"https://keras3.posit.co/dev/reference/get_custom_objects.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Get/set the currently registered custom objects. — get_custom_objects","text":"","code":"get_custom_objects()  set_custom_objects(objects = named_list(), clear = TRUE)"},{"path":"https://keras3.posit.co/dev/reference/get_custom_objects.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Get/set the currently registered custom objects. — get_custom_objects","text":"objects named list custom objects, returned get_custom_objects() set_custom_objects(). clear bool, whether clear custom object registry populating objects.","code":""},{"path":"https://keras3.posit.co/dev/reference/get_custom_objects.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Get/set the currently registered custom objects. — get_custom_objects","text":"R named list mapping registered names registered objects. set_custom_objects() returns registry values updating, invisibly.","code":""},{"path":"https://keras3.posit.co/dev/reference/get_custom_objects.html","id":"note","dir":"Reference","previous_headings":"","what":"Note","title":"Get/set the currently registered custom objects. — get_custom_objects","text":"register_keras_serializable() preferred set_custom_objects() registering new objects.","code":""},{"path":"https://keras3.posit.co/dev/reference/get_custom_objects.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Get/set the currently registered custom objects. — get_custom_objects","text":"can use set_custom_objects() restore previous registry state.","code":"get_custom_objects() # within a function, if you want to temporarily modify the registry, function() {   orig_objects <- set_custom_objects(clear = TRUE)   on.exit(set_custom_objects(orig_objects))    ## temporarily modify the global registry   # register_keras_serializable(....)   # ....  <do work>   # on.exit(), the previous registry state is restored. }"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/get_file.html","id":null,"dir":"Reference","previous_headings":"","what":"Downloads a file from a URL if it not already in the cache. — get_file","title":"Downloads a file from a URL if it not already in the cache. — get_file","text":"default file url origin downloaded cache_dir ~/.keras, placed cache_subdir datasets, given filename fname. final location file example.txt therefore ~/.keras/datasets/example.txt. Files .tar, .tar.gz, .tar.bz, .zip formats can also extracted. Passing hash verify file download. command line programs shasum sha256sum can compute hash.","code":""},{"path":"https://keras3.posit.co/dev/reference/get_file.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Downloads a file from a URL if it not already in the cache. — get_file","text":"","code":"get_file(   fname = NULL,   origin = NULL,   ...,   file_hash = NULL,   cache_subdir = \"datasets\",   hash_algorithm = \"auto\",   extract = FALSE,   archive_format = \"auto\",   cache_dir = NULL,   force_download = FALSE )"},{"path":"https://keras3.posit.co/dev/reference/get_file.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Downloads a file from a URL if it not already in the cache. — get_file","text":"fname target single file, desired local name file. NULL, name file origin used. downloading extracting directory archive, provided fname used extraction directory name (extension). origin Original URL file. ... forward/backward compatability. file_hash expected hash string file download. sha256 md5 hash algorithms supported. cache_subdir Subdirectory Keras cache dir file saved. absolute path, e.g. \"/path//folder\" specified, file saved location. hash_algorithm Select hash algorithm verify file. options \"md5', \"sha256', \"auto'. default 'auto' detects hash algorithm use. extract TRUE, extracts archive. applicable compressed archive files like tar zip. archive_format Archive format try extracting file. Options \"auto', \"tar', \"zip', NULL. \"tar\" includes tar, tar.gz, tar.bz files. default \"auto\" corresponds c(\"tar\", \"zip\"). NULL empty list return matches found. cache_dir Location store cached files, NULL defaults Sys.getenv(\"KERAS_HOME\", \"~/.keras/\"). force_download TRUE, file always re-downloaded regardless cache state.","code":""},{"path":"https://keras3.posit.co/dev/reference/get_file.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Downloads a file from a URL if it not already in the cache. — get_file","text":"Path downloaded file. ** Warning malicious downloads ** Downloading something Internet carries risk. NEVER download file/archive trust source. recommend specify file_hash argument (hash source file known) make sure file getting one expect.","code":""},{"path":"https://keras3.posit.co/dev/reference/get_file.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Downloads a file from a URL if it not already in the cache. — get_file","text":"","code":"path_to_downloaded_file <- get_file(     \"flower_photos\",     origin = \"https://storage.googleapis.com/download.tensorflow.org/example_images/flower_photos.tgz\",     extract = TRUE )"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/get_layer.html","id":null,"dir":"Reference","previous_headings":"","what":"Retrieves a layer based on either its name (unique) or index. — get_layer","title":"Retrieves a layer based on either its name (unique) or index. — get_layer","text":"Indices based order horizontal graph traversal (bottom-) 1-based. name index provided, index take precedence.","code":""},{"path":"https://keras3.posit.co/dev/reference/get_layer.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Retrieves a layer based on either its name (unique) or index. — get_layer","text":"","code":"get_layer(object, name = NULL, index = NULL)"},{"path":"https://keras3.posit.co/dev/reference/get_layer.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Retrieves a layer based on either its name (unique) or index. — get_layer","text":"object Keras model object name String, name layer. index Integer, index layer (1-based). Also valid negative values, count end model.","code":""},{"path":"https://keras3.posit.co/dev/reference/get_layer.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Retrieves a layer based on either its name (unique) or index. — get_layer","text":"layer instance.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/get_registered_name.html","id":null,"dir":"Reference","previous_headings":"","what":"Returns the name registered to an object within the Keras framework. — get_registered_name","title":"Returns the name registered to an object within the Keras framework. — get_registered_name","text":"function part Keras serialization deserialization framework. maps objects string names associated objects serialization/deserialization.","code":""},{"path":"https://keras3.posit.co/dev/reference/get_registered_name.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Returns the name registered to an object within the Keras framework. — get_registered_name","text":"","code":"get_registered_name(obj)"},{"path":"https://keras3.posit.co/dev/reference/get_registered_name.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Returns the name registered to an object within the Keras framework. — get_registered_name","text":"obj object look .","code":""},{"path":"https://keras3.posit.co/dev/reference/get_registered_name.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Returns the name registered to an object within the Keras framework. — get_registered_name","text":"name associated object, default name object registered.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/get_registered_object.html","id":null,"dir":"Reference","previous_headings":"","what":"Returns the class associated with name if it is registered with Keras. — get_registered_object","title":"Returns the class associated with name if it is registered with Keras. — get_registered_object","text":"function part Keras serialization deserialization framework. maps strings objects associated serialization/deserialization.","code":""},{"path":"https://keras3.posit.co/dev/reference/get_registered_object.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Returns the class associated with name if it is registered with Keras. — get_registered_object","text":"","code":"get_registered_object(name, custom_objects = NULL, module_objects = NULL)"},{"path":"https://keras3.posit.co/dev/reference/get_registered_object.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Returns the class associated with name if it is registered with Keras. — get_registered_object","text":"name name look . custom_objects named list custom objects look name . Generally, custom_objects provided user. module_objects named list custom objects look name . Generally, module_objects provided midlevel library implementers.","code":""},{"path":"https://keras3.posit.co/dev/reference/get_registered_object.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Returns the class associated with name if it is registered with Keras. — get_registered_object","text":"instantiable class associated name, NULL class exists.","code":""},{"path":"https://keras3.posit.co/dev/reference/get_registered_object.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Returns the class associated with name if it is registered with Keras. — get_registered_object","text":"","code":"from_config <- function(cls, config, custom_objects = NULL) {   if ('my_custom_object_name' \\%in\\% names(config)) {     config$hidden_cls <- get_registered_object(       config$my_custom_object_name,       custom_objects = custom_objects)   } }"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/get_source_inputs.html","id":null,"dir":"Reference","previous_headings":"","what":"Returns the list of input tensors necessary to compute tensor. — get_source_inputs","title":"Returns the list of input tensors necessary to compute tensor. — get_source_inputs","text":"Output always list tensors (potentially 1 element).","code":""},{"path":"https://keras3.posit.co/dev/reference/get_source_inputs.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Returns the list of input tensors necessary to compute tensor. — get_source_inputs","text":"","code":"get_source_inputs(tensor)"},{"path":"https://keras3.posit.co/dev/reference/get_source_inputs.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Returns the list of input tensors necessary to compute tensor. — get_source_inputs","text":"tensor tensor start .","code":""},{"path":"https://keras3.posit.co/dev/reference/get_source_inputs.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Returns the list of input tensors necessary to compute tensor. — get_source_inputs","text":"List input tensors.","code":""},{"path":"https://keras3.posit.co/dev/reference/get_source_inputs.html","id":"example","dir":"Reference","previous_headings":"","what":"Example","title":"Returns the list of input tensors necessary to compute tensor. — get_source_inputs","text":"","code":"input <- keras_input(c(3)) output <- input |> layer_dense(4) |> op_multiply(5) reticulate::py_id(get_source_inputs(output)[[1]]) == reticulate::py_id(input) ## [1] TRUE"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/get_state_tree.html","id":null,"dir":"Reference","previous_headings":"","what":"Retrieves tree-like structure of model variables. — get_state_tree","title":"Retrieves tree-like structure of model variables. — get_state_tree","text":"method allows retrieval different model variables (trainable, non-trainable, optimizer, metrics). variables returned nested dictionary format, keys correspond variable names values nested representations variables.","code":""},{"path":"https://keras3.posit.co/dev/reference/get_state_tree.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Retrieves tree-like structure of model variables. — get_state_tree","text":"","code":"get_state_tree(object, value_format = \"backend_tensor\")"},{"path":"https://keras3.posit.co/dev/reference/get_state_tree.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Retrieves tree-like structure of model variables. — get_state_tree","text":"object Keras Model. value_format One \"backend_tensor\", \"numpy_array\", \"array\". kind array return leaves nested state tree.","code":""},{"path":"https://keras3.posit.co/dev/reference/get_state_tree.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Retrieves tree-like structure of model variables. — get_state_tree","text":"named list containing nested representations requested variables. names variable names, values corresponding nested named lists.","code":""},{"path":"https://keras3.posit.co/dev/reference/get_state_tree.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Retrieves tree-like structure of model variables. — get_state_tree","text":"state_tree list returned looks like:   example:","code":"model <- keras_model_sequential(name = \"my_sequential\",                                 input_shape = c(1),                                 input_name = \"my_input\") |>   layer_dense(1, activation = \"sigmoid\", name = \"my_dense\")  model |> compile(optimizer=\"adam\", loss=\"mse\", metrics=c(\"mae\")) model |> fit(matrix(1), matrix(1), verbose = 0) state_tree <- model |> get_state_tree() list(   metrics_variables = list(     loss = list(       count = ...,       total = ...     ),     mean_absolute_error = list(       count = ...,       total = ...     )   ),   trainable_variables = list(     my_sequential = list(       my_dense = list(         bias = ...,         kernel = ...       )     )   ),   non_trainable_variables = list(),   optimizer_variables = list(     adam = list(       iteration = ...,       learning_rate = ...,       my_sequential_my_dense_bias_momentum = ...,       my_sequential_my_dense_bias_velocity = ...,       my_sequential_my_dense_kernel_momentum = ...,       my_sequential_my_dense_kernel_velocity = ...     )   ) ) str(state_tree) ## List of 4 ##  $ trainable_variables    :List of 1 ##   ..$ my_sequential:List of 1 ##   .. ..$ my_dense:List of 2 ##   .. .. ..$ kernel:<tf.Variable 'my_sequential/my_dense/kernel:0' shape=(1, 1) dtype=float32, numpy=array([[-0.8338491]], dtype=float32)> ##   .. .. ..$ bias  :<tf.Variable 'my_sequential/my_dense/bias:0' shape=(1) dtype=float32, numpy=array([0.00099998], dtype=float32)> ##  $ non_trainable_variables: Named list() ##  $ optimizer_variables    :List of 1 ##   ..$ adam:List of 6 ##   .. ..$ iteration                             :<tf.Variable 'adam/iteration:0' shape=() dtype=int64, numpy=1> ##   .. ..$ learning_rate                         :<tf.Variable 'adam/learning_rate:0' shape=() dtype=float32, numpy=0.0010000000474974513> ##   .. ..$ my_sequential_my_dense_kernel_momentum:<tf.Variable 'adam/my_sequential_my_dense_kernel_momentum:0' shape=(1, 1) dtype=float32, numpy=array([[-0.02943518]], dtype=float32)> ##   .. ..$ my_sequential_my_dense_kernel_velocity:<tf.Variable 'adam/my_sequential_my_dense_kernel_velocity:0' shape=(1, 1) dtype=float32, numpy=array([[8.664299e-05]], dtype=float32)> ##   .. ..$ my_sequential_my_dense_bias_momentum  :<tf.Variable 'adam/my_sequential_my_dense_bias_momentum:0' shape=(1) dtype=float32, numpy=array([-0.02943518], dtype=float32)> ##   .. ..$ my_sequential_my_dense_bias_velocity  :<tf.Variable 'adam/my_sequential_my_dense_bias_velocity:0' shape=(1) dtype=float32, numpy=array([8.664299e-05], dtype=float32)> ##  $ metrics_variables      :List of 2 ##   ..$ loss               :List of 2 ##   .. ..$ total:<tf.Variable 'loss/total:0' shape=() dtype=float32, numpy=0.4863377809524536> ##   .. ..$ count:<tf.Variable 'loss/count:0' shape=() dtype=float32, numpy=1.0> ##   ..$ mean_absolute_error:List of 2 ##   .. ..$ total:<tf.Variable 'mean_absolute_error/total:0' shape=() dtype=float32, numpy=0.6973792314529419> ##   .. ..$ count:<tf.Variable 'mean_absolute_error_1/count:0' shape=() dtype=float32, numpy=1.0>"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/get_weights.html","id":null,"dir":"Reference","previous_headings":"","what":"Layer/Model weights as R arrays — get_weights","title":"Layer/Model weights as R arrays — get_weights","text":"Layer/Model weights R arrays","code":""},{"path":"https://keras3.posit.co/dev/reference/get_weights.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Layer/Model weights as R arrays — get_weights","text":"","code":"get_weights(object, trainable = NA)  set_weights(object, weights)"},{"path":"https://keras3.posit.co/dev/reference/get_weights.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Layer/Model weights as R arrays — get_weights","text":"object Layer model object trainable NA (default), weights returned. TRUE, weights trainable variables returned. FALSE, weights non-trainable variables returned. weights Weights R array","code":""},{"path":"https://keras3.posit.co/dev/reference/get_weights.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Layer/Model weights as R arrays — get_weights","text":"list R arrays.","code":""},{"path":"https://keras3.posit.co/dev/reference/get_weights.html","id":"note","dir":"Reference","previous_headings":"","what":"Note","title":"Layer/Model weights as R arrays — get_weights","text":"can access Layer/Model KerasVariables (also backend-native tensors like tf.Variable) object$weights, object$trainable_weights, object$non_trainable_weights","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/grapes-py_class-grapes.html","id":null,"dir":"Reference","previous_headings":"","what":"Make a python class constructor — %py_class%","title":"Make a python class constructor — %py_class%","text":"Make python class constructor","code":""},{"path":"https://keras3.posit.co/dev/reference/grapes-py_class-grapes.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Make a python class constructor — %py_class%","text":"","code":"spec %py_class% body"},{"path":"https://keras3.posit.co/dev/reference/grapes-py_class-grapes.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Make a python class constructor — %py_class%","text":"spec bare symbol MyClassName, call MyClassName(SuperClass) body expression can evaluated construct class methods.","code":""},{"path":"https://keras3.posit.co/dev/reference/grapes-py_class-grapes.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Make a python class constructor — %py_class%","text":"python class constructor, invisibly. Note, constructor also assigned parent frame.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/grapes-py_class-grapes.html","id":"ref-examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Make a python class constructor — %py_class%","text":"","code":"if (FALSE) { # \\dontrun{ MyClass %py_class% {   initialize <- function(x) {     print(\"Hi from MyClass$initialize()!\")     self$x <- x   }   my_method <- function() {     self$x   } }  my_class_instance <- MyClass(42) my_class_instance$my_method()  MyClass2(MyClass) %py_class% {   \"This will be a __doc__ string for MyClass2\"    initialize <- function(...) {     \"This will be the __doc__ string for the MyClass2.__init__() method\"     print(\"Hi from MyClass2$initialize()!\")     super$initialize(...)   } }  my_class_instance2 <- MyClass2(42) my_class_instance2$my_method()  reticulate::py_help(MyClass2) # see the __doc__ strings and more!  # In addition to `self`, there is also `private` available. # This is an R environment unique to each class instance, where you can # store objects that you don't want converted to Python, but still want # available from methods. You can also assign methods to private, and # `self` and `private` will be available in private methods.  MyClass %py_class% {    initialize <- function(x) {     print(\"Hi from MyClass$initialize()!\")     private$y <- paste(\"A Private field:\", x)   }    get_private_field <- function() {     private$y   }    private$a_private_method <- function() {     cat(\"a_private_method() was called.\\n\")     cat(\"private$y is \", sQuote(private$y), \"\\n\")   }    call_private_method <- function()     private$a_private_method()    # equivalent of @property decorator in python   an_active_property %<-active% function(x = NULL) {     if(!is.null(x)) {       cat(\"`an_active_property` was assigned\", x, \"\\n\")       return(x)     } else {       cat(\"`an_active_property` was accessed\\n\")       return(42)     }   } }  inst1 <- MyClass(1) inst2 <- MyClass(2) inst1$get_private_field() inst2$get_private_field() inst1$call_private_method() inst2$call_private_method() inst1$an_active_property inst1$an_active_property <- 11 } # }"},{"path":"https://keras3.posit.co/dev/reference/grapes-set-active-grapes.html","id":null,"dir":"Reference","previous_headings":"","what":"Make an Active Binding — %<-active%","title":"Make an Active Binding — %<-active%","text":"Make Active Binding","code":""},{"path":"https://keras3.posit.co/dev/reference/grapes-set-active-grapes.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Make an Active Binding — %<-active%","text":"","code":"sym %<-active% value"},{"path":"https://keras3.posit.co/dev/reference/grapes-set-active-grapes.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Make an Active Binding — %<-active%","text":"sym symbol bind value function call value sym accessed.","code":""},{"path":"https://keras3.posit.co/dev/reference/grapes-set-active-grapes.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Make an Active Binding — %<-active%","text":"value, invisibly","code":""},{"path":"https://keras3.posit.co/dev/reference/grapes-set-active-grapes.html","id":"details","dir":"Reference","previous_headings":"","what":"Details","title":"Make an Active Binding — %<-active%","text":"Active bindings defined %py_class% converted @property decorated methods.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/grapes-set-active-grapes.html","id":"ref-examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Make an Active Binding — %<-active%","text":"","code":"set.seed(1234) x %<-active% function(value) {   message(\"Evaluating function of active binding\")   if(missing(value))     runif(1)   else    message(\"Received: \", value) } x #> Evaluating function of active binding #> [1] 0.1137034 x #> Evaluating function of active binding #> [1] 0.6222994 x <- \"foo\" #> Evaluating function of active binding #> Received: foo x <- \"foo\" #> Evaluating function of active binding #> Received: foo x #> Evaluating function of active binding #> [1] 0.6092747 rm(x) # cleanup"},{"path":"https://keras3.posit.co/dev/reference/image_array_save.html","id":null,"dir":"Reference","previous_headings":"","what":"Saves an image stored as an array to a path or file object. — image_array_save","title":"Saves an image stored as an array to a path or file object. — image_array_save","text":"Saves image stored array path file object.","code":""},{"path":"https://keras3.posit.co/dev/reference/image_array_save.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Saves an image stored as an array to a path or file object. — image_array_save","text":"","code":"image_array_save(   x,   path,   data_format = NULL,   file_format = NULL,   scale = TRUE,   ... )"},{"path":"https://keras3.posit.co/dev/reference/image_array_save.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Saves an image stored as an array to a path or file object. — image_array_save","text":"x array. path Path file object. data_format Image data format, either \"channels_first\" \"channels_last\". file_format Optional file format override. omitted, format use determined filename extension. file object used instead filename, parameter always used. scale Whether rescale image values within [0, 255]. ... Additional keyword arguments passed PIL.Image.save().","code":""},{"path":"https://keras3.posit.co/dev/reference/image_array_save.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Saves an image stored as an array to a path or file object. — image_array_save","text":"Called primarily side effects. input x returned, invisibly, enable usage pipe.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/image_dataset_from_directory.html","id":null,"dir":"Reference","previous_headings":"","what":"Generates a tf.data.Dataset from image files in a directory. — image_dataset_from_directory","title":"Generates a tf.data.Dataset from image files in a directory. — image_dataset_from_directory","text":"directory structure :   calling image_dataset_from_directory(main_directory, labels = 'inferred') return tf.data.Dataset yields batches images subdirectories class_a class_b, together labels 0 1 (0 corresponding class_a 1 corresponding class_b). Supported image formats: .jpeg, .jpg, .png, .bmp, .gif. Animated gifs truncated first frame.","code":"main_directory/ ...class_a/ ......a_image_1.jpg ......a_image_2.jpg ...class_b/ ......b_image_1.jpg ......b_image_2.jpg"},{"path":"https://keras3.posit.co/dev/reference/image_dataset_from_directory.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Generates a tf.data.Dataset from image files in a directory. — image_dataset_from_directory","text":"","code":"image_dataset_from_directory(   directory,   labels = \"inferred\",   label_mode = \"int\",   class_names = NULL,   color_mode = \"rgb\",   batch_size = 32L,   image_size = c(256L, 256L),   shuffle = TRUE,   seed = NULL,   validation_split = NULL,   subset = NULL,   interpolation = \"bilinear\",   follow_links = FALSE,   crop_to_aspect_ratio = FALSE,   pad_to_aspect_ratio = FALSE,   data_format = NULL,   verbose = TRUE )"},{"path":"https://keras3.posit.co/dev/reference/image_dataset_from_directory.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Generates a tf.data.Dataset from image files in a directory. — image_dataset_from_directory","text":"directory Directory data located. labels \"inferred\", contain subdirectories, containing images class. Otherwise, directory structure ignored. labels Either \"inferred\" (labels generated directory structure), NULL (labels), list/tuple integer labels size number image files found directory. Labels sorted according alphanumeric order image file paths (obtained via os.walk(directory) Python). label_mode String describing encoding labels. Options : \"int\": means labels encoded integers (e.g. sparse_categorical_crossentropy loss). \"categorical\" means labels encoded categorical vector (e.g. categorical_crossentropy loss). \"binary\" means labels (can 2) encoded float32 scalars values 0 1 (e.g. binary_crossentropy). NULL (labels). class_names valid labels \"inferred\". explicit list class names (must match names subdirectories). Used control order classes (otherwise alphanumerical order used). color_mode One \"grayscale\", \"rgb\", \"rgba\". Whether images converted 1, 3, 4 channels. Defaults \"rgb\". batch_size Size batches data. Defaults 32. NULL, data batched (dataset yield individual samples). image_size Size resize images read disk, specified (height, width). Since pipeline processes batches images must size, must provided. Defaults (256, 256). shuffle Whether shuffle data. Defaults TRUE. set FALSE, sorts data alphanumeric order. seed Optional random seed shuffling transformations. validation_split Optional float 0 1, fraction data reserve validation. subset Subset data return. One \"training\", \"validation\", \"\". used validation_split set. subset = \"\", utility returns tuple two datasets (training validation datasets respectively). interpolation String, interpolation method used resizing images. Supports \"bilinear\", \"nearest\", \"bicubic\", \"area\", \"lanczos3\", \"lanczos5\", \"gaussian\", \"mitchellcubic\". Defaults \"bilinear\". follow_links Whether visit subdirectories pointed symlinks. Defaults FALSE. crop_to_aspect_ratio TRUE, resize images without aspect ratio distortion. original aspect ratio differs target aspect ratio, output image cropped return largest possible window image (size image_size) matches target aspect ratio. default (crop_to_aspect_ratio = FALSE), aspect ratio may preserved. pad_to_aspect_ratio TRUE, resize images without aspect ratio distortion. original aspect ratio differs target aspect ratio, output image padded return largest possible window image (size image_size) matches target aspect ratio. default (pad_to_aspect_ratio=FALSE), aspect ratio may preserved. data_format NULL uses config_image_data_format() otherwise either 'channel_last' 'channel_first'. verbose Whether display number information classes number files found. Defaults TRUE.","code":""},{"path":"https://keras3.posit.co/dev/reference/image_dataset_from_directory.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Generates a tf.data.Dataset from image files in a directory. — image_dataset_from_directory","text":"tf.data.Dataset object. label_mode NULL, yields float32 tensors shape (batch_size, image_size[1], image_size[2], num_channels), encoding images (see rules regarding num_channels). Otherwise, yields tuple (images, labels), images shape (batch_size, image_size[1], image_size[2], num_channels), labels follows format described . Rules regarding labels format: label_mode \"int\", labels int32 tensor shape (batch_size,). label_mode \"binary\", labels float32 tensor 1s 0s shape (batch_size, 1). label_mode \"categorical\", labels float32 tensor shape (batch_size, num_classes), representing one-hot encoding class index. Rules regarding number channels yielded images: color_mode \"grayscale\", 1 channel image tensors. color_mode \"rgb\", 3 channels image tensors. color_mode \"rgba\", 4 channels image tensors.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/image_from_array.html","id":null,"dir":"Reference","previous_headings":"","what":"Converts a 3D array to a PIL Image instance. — image_from_array","title":"Converts a 3D array to a PIL Image instance. — image_from_array","text":"Converts 3D array PIL Image instance.","code":""},{"path":"https://keras3.posit.co/dev/reference/image_from_array.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Converts a 3D array to a PIL Image instance. — image_from_array","text":"","code":"image_from_array(x, data_format = NULL, scale = TRUE, dtype = NULL)"},{"path":"https://keras3.posit.co/dev/reference/image_from_array.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Converts a 3D array to a PIL Image instance. — image_from_array","text":"x Input data, form can converted array. data_format Image data format, can either \"channels_first\" \"channels_last\". Defaults NULL, case global setting config_image_data_format() used (unless changed , defaults \"channels_last\"). scale Whether rescale image minimum maximum values 0 255 respectively. Defaults TRUE. dtype Dtype use. NULL means global setting config_floatx() used (unless changed , defaults \"float32\"). Defaults NULL.","code":""},{"path":"https://keras3.posit.co/dev/reference/image_from_array.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Converts a 3D array to a PIL Image instance. — image_from_array","text":"PIL Image instance.","code":""},{"path":"https://keras3.posit.co/dev/reference/image_from_array.html","id":"example","dir":"Reference","previous_headings":"","what":"Example","title":"Converts a 3D array to a PIL Image instance. — image_from_array","text":"","code":"img <- array(runif(30000), dim = c(100, 100, 3)) pil_img <- image_from_array(img) pil_img ## <PIL.Image.Image image mode=RGB size=100x100 at 0x0>"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/image_load.html","id":null,"dir":"Reference","previous_headings":"","what":"Loads an image into PIL format. — image_load","title":"Loads an image into PIL format. — image_load","text":"Loads image PIL format.","code":""},{"path":"https://keras3.posit.co/dev/reference/image_load.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Loads an image into PIL format. — image_load","text":"","code":"image_load(   path,   color_mode = \"rgb\",   target_size = NULL,   interpolation = \"nearest\",   keep_aspect_ratio = FALSE )"},{"path":"https://keras3.posit.co/dev/reference/image_load.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Loads an image into PIL format. — image_load","text":"path Path image file. color_mode One \"grayscale\", \"rgb\", \"rgba\". Default: \"rgb\". desired image format. target_size Either NULL (default original size) tuple ints (img_height, img_width). interpolation Interpolation method used resample image target size different loaded image. Supported methods \"nearest\", \"bilinear\", \"bicubic\". PIL version 1.1.3 newer installed, \"lanczos\" also supported. PIL version 3.4.0 newer installed, \"box\" \"hamming\" also supported. default, \"nearest\" used. keep_aspect_ratio Boolean, whether resize images target size without aspect ratio distortion. image cropped center target aspect ratio resizing.","code":""},{"path":"https://keras3.posit.co/dev/reference/image_load.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Loads an image into PIL format. — image_load","text":"PIL Image instance.","code":""},{"path":"https://keras3.posit.co/dev/reference/image_load.html","id":"example","dir":"Reference","previous_headings":"","what":"Example","title":"Loads an image into PIL format. — image_load","text":"","code":"image_path <- get_file(origin = \"https://www.r-project.org/logo/Rlogo.png\") (image <- image_load(image_path)) ## <PIL.Image.Image image mode=RGB size=724x561 at 0x0> input_arr <- image_to_array(image) str(input_arr) ##  num [1:561, 1:724, 1:3] 0 0 0 0 0 0 0 0 0 0 ... input_arr %<>% array_reshape(dim = c(1, dim(input_arr))) # Convert single image to a batch. model |> predict(input_arr)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/image_smart_resize.html","id":null,"dir":"Reference","previous_headings":"","what":"Resize images to a target size without aspect ratio distortion. — image_smart_resize","title":"Resize images to a target size without aspect ratio distortion. — image_smart_resize","text":"Image datasets typically yield images different size. However, images need batched can processed Keras layers. batched, images need share height width. simply , TF (JAX equivalent):   However, , distort aspect ratio images, since general aspect ratio size. fine many cases, always (e.g. image generation models can problem). Note passing argument preserve_aspect_ratio = TRUE tf$image$resize() preserve aspect ratio, cost longer respecting provided target size. calls :   output images actually (200, 200), distorted. Instead, parts image fit within target size get cropped . resizing process : Take largest centered crop image aspect ratio target size. instance, size = c(200, 200) input image size (340, 500), take crop (340, 340) centered along width. Resize cropped image target size. example , resize (340, 340) crop (200, 200).","code":"size <- c(200, 200) ds <- ds$map(\\(img) tf$image$resize(img, size)) size <- c(200, 200) ds <- ds$map(\\(img) image_smart_resize(img, size))"},{"path":"https://keras3.posit.co/dev/reference/image_smart_resize.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Resize images to a target size without aspect ratio distortion. — image_smart_resize","text":"","code":"image_smart_resize(   x,   size,   interpolation = \"bilinear\",   data_format = \"channels_last\",   backend_module = NULL )"},{"path":"https://keras3.posit.co/dev/reference/image_smart_resize.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Resize images to a target size without aspect ratio distortion. — image_smart_resize","text":"x Input image batch images (tensor array). Must format (height, width, channels) (batch_size, height, width, channels). size Tuple (height, width) integer. Target size. interpolation String, interpolation use resizing. Supports \"bilinear\", \"nearest\", \"bicubic\", \"lanczos3\", \"lanczos5\". Defaults 'bilinear'. data_format \"channels_last\" \"channels_first\". backend_module Backend module use (different default backend).","code":""},{"path":"https://keras3.posit.co/dev/reference/image_smart_resize.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Resize images to a target size without aspect ratio distortion. — image_smart_resize","text":"Array shape (size[1], size[2], channels). input image array, output array, backend-native tensor, output backend-native tensor.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/image_to_array.html","id":null,"dir":"Reference","previous_headings":"","what":"Converts a PIL Image instance to a matrix. — image_to_array","title":"Converts a PIL Image instance to a matrix. — image_to_array","text":"Converts PIL Image instance matrix.","code":""},{"path":"https://keras3.posit.co/dev/reference/image_to_array.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Converts a PIL Image instance to a matrix. — image_to_array","text":"","code":"image_to_array(img, data_format = NULL, dtype = NULL)"},{"path":"https://keras3.posit.co/dev/reference/image_to_array.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Converts a PIL Image instance to a matrix. — image_to_array","text":"img Input PIL Image instance. data_format Image data format, can either \"channels_first\" \"channels_last\". Defaults NULL, case global setting config_image_data_format() used (unless changed , defaults \"channels_last\"). dtype Dtype use. NULL \"double\" return R double array. \"integer\" returns R integer array. values (e.g., \"float32\") return NumPy array.","code":""},{"path":"https://keras3.posit.co/dev/reference/image_to_array.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Converts a PIL Image instance to a matrix. — image_to_array","text":"3D array.","code":""},{"path":"https://keras3.posit.co/dev/reference/image_to_array.html","id":"example","dir":"Reference","previous_headings":"","what":"Example","title":"Converts a PIL Image instance to a matrix. — image_to_array","text":"","code":"image_path <- get_file(origin = \"https://www.r-project.org/logo/Rlogo.png\") (img <- image_load(image_path)) ## <PIL.Image.Image image mode=RGB size=724x561 at 0x0> array <- image_to_array(img) str(array) ##  num [1:561, 1:724, 1:3] 0 0 0 0 0 0 0 0 0 0 ..."},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/imagenet_decode_predictions.html","id":null,"dir":"Reference","previous_headings":"","what":"Decodes the prediction of an ImageNet model. — imagenet_decode_predictions","title":"Decodes the prediction of an ImageNet model. — imagenet_decode_predictions","text":"Decodes prediction ImageNet model.","code":""},{"path":"https://keras3.posit.co/dev/reference/imagenet_decode_predictions.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Decodes the prediction of an ImageNet model. — imagenet_decode_predictions","text":"","code":"imagenet_decode_predictions(preds, top = 5)"},{"path":"https://keras3.posit.co/dev/reference/imagenet_decode_predictions.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Decodes the prediction of an ImageNet model. — imagenet_decode_predictions","text":"preds Tensor encoding batch predictions. top integer, many top-guesses return.","code":""},{"path":"https://keras3.posit.co/dev/reference/imagenet_decode_predictions.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Decodes the prediction of an ImageNet model. — imagenet_decode_predictions","text":"List data frames variables class_name, class_description, score (one data frame per sample batch input).","code":""},{"path":"https://keras3.posit.co/dev/reference/imagenet_preprocess_input.html","id":null,"dir":"Reference","previous_headings":"","what":"Preprocesses a tensor or array encoding a batch of images. — imagenet_preprocess_input","title":"Preprocesses a tensor or array encoding a batch of images. — imagenet_preprocess_input","text":"Preprocesses tensor array encoding batch images.","code":""},{"path":"https://keras3.posit.co/dev/reference/imagenet_preprocess_input.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Preprocesses a tensor or array encoding a batch of images. — imagenet_preprocess_input","text":"","code":"imagenet_preprocess_input(x, data_format = NULL, mode = \"caffe\")"},{"path":"https://keras3.posit.co/dev/reference/imagenet_preprocess_input.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Preprocesses a tensor or array encoding a batch of images. — imagenet_preprocess_input","text":"x Input Numpy symbolic tensor, 3D 4D. data_format Data format image tensor/array. mode One \"caffe\", \"tf\", \"torch\" caffe: convert images RGB BGR, zero-center color channel respect ImageNet dataset, without scaling. tf: scale pixels -1 1, sample-wise. torch: scale pixels 0 1 normalize channel respect ImageNet dataset.","code":""},{"path":"https://keras3.posit.co/dev/reference/imagenet_preprocess_input.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Preprocesses a tensor or array encoding a batch of images. — imagenet_preprocess_input","text":"Preprocessed tensor array.","code":""},{"path":"https://keras3.posit.co/dev/reference/initializer_constant.html","id":null,"dir":"Reference","previous_headings":"","what":"Initializer that generates tensors with constant values. — initializer_constant","title":"Initializer that generates tensors with constant values. — initializer_constant","text":"scalar values allowed. constant value provided must convertible dtype requested calling initializer.","code":""},{"path":"https://keras3.posit.co/dev/reference/initializer_constant.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Initializer that generates tensors with constant values. — initializer_constant","text":"","code":"initializer_constant(value = 0)"},{"path":"https://keras3.posit.co/dev/reference/initializer_constant.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Initializer that generates tensors with constant values. — initializer_constant","text":"value numeric scalar.","code":""},{"path":"https://keras3.posit.co/dev/reference/initializer_constant.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Initializer that generates tensors with constant values. — initializer_constant","text":"Initializer instance can passed layer variable constructors, called directly shape return Tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/initializer_constant.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Initializer that generates tensors with constant values. — initializer_constant","text":"","code":"# Standalone usage: initializer <- initializer_constant(10) values <- initializer(shape = c(2, 2)) # Usage in a Keras layer: initializer <- initializer_constant(10) layer <- layer_dense(units = 3, kernel_initializer = initializer)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/initializer_glorot_normal.html","id":null,"dir":"Reference","previous_headings":"","what":"The Glorot normal initializer, also called Xavier normal initializer. — initializer_glorot_normal","title":"The Glorot normal initializer, also called Xavier normal initializer. — initializer_glorot_normal","text":"Draws samples truncated normal distribution centered 0 stddev = sqrt(2 / (fan_in + fan_out)) fan_in number input units weight tensor fan_out number output units weight tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/initializer_glorot_normal.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"The Glorot normal initializer, also called Xavier normal initializer. — initializer_glorot_normal","text":"","code":"initializer_glorot_normal(seed = NULL)"},{"path":"https://keras3.posit.co/dev/reference/initializer_glorot_normal.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"The Glorot normal initializer, also called Xavier normal initializer. — initializer_glorot_normal","text":"seed integer instance random_seed_generator(). Used make behavior initializer deterministic. Note initializer seeded integer NULL (unseeded) produce random values across multiple calls. get different random values across multiple calls, use seed instance random_seed_generator().","code":""},{"path":"https://keras3.posit.co/dev/reference/initializer_glorot_normal.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"The Glorot normal initializer, also called Xavier normal initializer. — initializer_glorot_normal","text":"Initializer instance can passed layer variable constructors, called directly shape return Tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/initializer_glorot_normal.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"The Glorot normal initializer, also called Xavier normal initializer. — initializer_glorot_normal","text":"","code":"# Standalone usage: initializer <- initializer_glorot_normal() values <- initializer(shape = c(2, 2)) # Usage in a Keras layer: initializer <- initializer_glorot_normal() layer <- layer_dense(units = 3, kernel_initializer = initializer)"},{"path":"https://keras3.posit.co/dev/reference/initializer_glorot_normal.html","id":"reference","dir":"Reference","previous_headings":"","what":"Reference","title":"The Glorot normal initializer, also called Xavier normal initializer. — initializer_glorot_normal","text":"Glorot et al., 2010","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/initializer_glorot_uniform.html","id":null,"dir":"Reference","previous_headings":"","what":"The Glorot uniform initializer, also called Xavier uniform initializer. — initializer_glorot_uniform","title":"The Glorot uniform initializer, also called Xavier uniform initializer. — initializer_glorot_uniform","text":"Draws samples uniform distribution within [-limit, limit], limit = sqrt(6 / (fan_in + fan_out)) (fan_in number input units weight tensor fan_out number output units).","code":""},{"path":"https://keras3.posit.co/dev/reference/initializer_glorot_uniform.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"The Glorot uniform initializer, also called Xavier uniform initializer. — initializer_glorot_uniform","text":"","code":"initializer_glorot_uniform(seed = NULL)"},{"path":"https://keras3.posit.co/dev/reference/initializer_glorot_uniform.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"The Glorot uniform initializer, also called Xavier uniform initializer. — initializer_glorot_uniform","text":"seed integer instance random_seed_generator(). Used make behavior initializer deterministic. Note initializer seeded integer NULL (unseeded) produce random values across multiple calls. get different random values across multiple calls, use seed instance random_seed_generator().","code":""},{"path":"https://keras3.posit.co/dev/reference/initializer_glorot_uniform.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"The Glorot uniform initializer, also called Xavier uniform initializer. — initializer_glorot_uniform","text":"Initializer instance can passed layer variable constructors, called directly shape return Tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/initializer_glorot_uniform.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"The Glorot uniform initializer, also called Xavier uniform initializer. — initializer_glorot_uniform","text":"","code":"# Standalone usage: initializer <- initializer_glorot_uniform() values <- initializer(shape = c(2, 2)) # Usage in a Keras layer: initializer <- initializer_glorot_uniform() layer <- layer_dense(units = 3, kernel_initializer = initializer)"},{"path":"https://keras3.posit.co/dev/reference/initializer_glorot_uniform.html","id":"reference","dir":"Reference","previous_headings":"","what":"Reference","title":"The Glorot uniform initializer, also called Xavier uniform initializer. — initializer_glorot_uniform","text":"Glorot et al., 2010","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/initializer_he_normal.html","id":null,"dir":"Reference","previous_headings":"","what":"He normal initializer. — initializer_he_normal","title":"He normal initializer. — initializer_he_normal","text":"draws samples truncated normal distribution centered 0 stddev = sqrt(2 / fan_in) fan_in number input units weight tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/initializer_he_normal.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"He normal initializer. — initializer_he_normal","text":"","code":"initializer_he_normal(seed = NULL)"},{"path":"https://keras3.posit.co/dev/reference/initializer_he_normal.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"He normal initializer. — initializer_he_normal","text":"seed integer instance random_seed_generator(). Used make behavior initializer deterministic. Note initializer seeded integer NULL (unseeded) produce random values across multiple calls. get different random values across multiple calls, use seed instance random_seed_generator().","code":""},{"path":"https://keras3.posit.co/dev/reference/initializer_he_normal.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"He normal initializer. — initializer_he_normal","text":"Initializer instance can passed layer variable constructors, called directly shape return Tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/initializer_he_normal.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"He normal initializer. — initializer_he_normal","text":"","code":"# Standalone usage: initializer <- initializer_he_normal() values <- initializer(shape = c(2, 2)) # Usage in a Keras layer: initializer <- initializer_he_normal() layer <- layer_dense(units = 3, kernel_initializer = initializer)"},{"path":"https://keras3.posit.co/dev/reference/initializer_he_normal.html","id":"reference","dir":"Reference","previous_headings":"","what":"Reference","title":"He normal initializer. — initializer_he_normal","text":"et al., 2015","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/initializer_he_uniform.html","id":null,"dir":"Reference","previous_headings":"","what":"He uniform variance scaling initializer. — initializer_he_uniform","title":"He uniform variance scaling initializer. — initializer_he_uniform","text":"Draws samples uniform distribution within [-limit, limit], limit = sqrt(6 / fan_in) (fan_in number input units weight tensor).","code":""},{"path":"https://keras3.posit.co/dev/reference/initializer_he_uniform.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"He uniform variance scaling initializer. — initializer_he_uniform","text":"","code":"initializer_he_uniform(seed = NULL)"},{"path":"https://keras3.posit.co/dev/reference/initializer_he_uniform.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"He uniform variance scaling initializer. — initializer_he_uniform","text":"seed integer instance random_seed_generator(). Used make behavior initializer deterministic. Note initializer seeded integer NULL (unseeded) produce random values across multiple calls. get different random values across multiple calls, use seed instance random_seed_generator().","code":""},{"path":"https://keras3.posit.co/dev/reference/initializer_he_uniform.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"He uniform variance scaling initializer. — initializer_he_uniform","text":"Initializer instance can passed layer variable constructors, called directly shape return Tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/initializer_he_uniform.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"He uniform variance scaling initializer. — initializer_he_uniform","text":"","code":"# Standalone usage: initializer <- initializer_he_uniform() values <- initializer(shape = c(2, 2)) # Usage in a Keras layer: initializer <- initializer_he_uniform() layer <- layer_dense(units = 3, kernel_initializer = initializer)"},{"path":"https://keras3.posit.co/dev/reference/initializer_he_uniform.html","id":"reference","dir":"Reference","previous_headings":"","what":"Reference","title":"He uniform variance scaling initializer. — initializer_he_uniform","text":"et al., 2015","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/initializer_identity.html","id":null,"dir":"Reference","previous_headings":"","what":"Initializer that generates the identity matrix. — initializer_identity","title":"Initializer that generates the identity matrix. — initializer_identity","text":"usable generating 2D matrices.","code":""},{"path":"https://keras3.posit.co/dev/reference/initializer_identity.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Initializer that generates the identity matrix. — initializer_identity","text":"","code":"initializer_identity(gain = 1)"},{"path":"https://keras3.posit.co/dev/reference/initializer_identity.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Initializer that generates the identity matrix. — initializer_identity","text":"gain Multiplicative factor apply identity matrix.","code":""},{"path":"https://keras3.posit.co/dev/reference/initializer_identity.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Initializer that generates the identity matrix. — initializer_identity","text":"Initializer instance can passed layer variable constructors, called directly shape return Tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/initializer_identity.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Initializer that generates the identity matrix. — initializer_identity","text":"","code":"# Standalone usage: initializer <- initializer_identity() values <- initializer(shape = c(2, 2)) # Usage in a Keras layer: initializer <- initializer_identity() layer <- layer_dense(units = 3, kernel_initializer = initializer)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/initializer_lecun_normal.html","id":null,"dir":"Reference","previous_headings":"","what":"Lecun normal initializer. — initializer_lecun_normal","title":"Lecun normal initializer. — initializer_lecun_normal","text":"Initializers allow pre-specify initialization strategy, encoded Initializer object, without knowing shape dtype variable initialized. Draws samples truncated normal distribution centered 0 stddev = sqrt(1 / fan_in) fan_in number input units weight tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/initializer_lecun_normal.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Lecun normal initializer. — initializer_lecun_normal","text":"","code":"initializer_lecun_normal(seed = NULL)"},{"path":"https://keras3.posit.co/dev/reference/initializer_lecun_normal.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Lecun normal initializer. — initializer_lecun_normal","text":"seed integer instance random_seed_generator(). Used make behavior initializer deterministic. Note initializer seeded integer NULL (unseeded) produce random values across multiple calls. get different random values across multiple calls, use seed instance random_seed_generator().","code":""},{"path":"https://keras3.posit.co/dev/reference/initializer_lecun_normal.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Lecun normal initializer. — initializer_lecun_normal","text":"Initializer instance can passed layer variable constructors, called directly shape return Tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/initializer_lecun_normal.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Lecun normal initializer. — initializer_lecun_normal","text":"","code":"# Standalone usage: initializer <- initializer_lecun_normal() values <- initializer(shape = c(2, 2)) # Usage in a Keras layer: initializer <- initializer_lecun_normal() layer <- layer_dense(units = 3, kernel_initializer = initializer)"},{"path":"https://keras3.posit.co/dev/reference/initializer_lecun_normal.html","id":"reference","dir":"Reference","previous_headings":"","what":"Reference","title":"Lecun normal initializer. — initializer_lecun_normal","text":"Klambauer et al., 2017","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/initializer_lecun_uniform.html","id":null,"dir":"Reference","previous_headings":"","what":"Lecun uniform initializer. — initializer_lecun_uniform","title":"Lecun uniform initializer. — initializer_lecun_uniform","text":"Draws samples uniform distribution within [-limit, limit], limit = sqrt(3 / fan_in) (fan_in number input units weight tensor).","code":""},{"path":"https://keras3.posit.co/dev/reference/initializer_lecun_uniform.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Lecun uniform initializer. — initializer_lecun_uniform","text":"","code":"initializer_lecun_uniform(seed = NULL)"},{"path":"https://keras3.posit.co/dev/reference/initializer_lecun_uniform.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Lecun uniform initializer. — initializer_lecun_uniform","text":"seed integer instance random_seed_generator(). Used make behavior initializer deterministic. Note initializer seeded integer NULL (unseeded) produce random values across multiple calls. get different random values across multiple calls, use seed instance random_seed_generator().","code":""},{"path":"https://keras3.posit.co/dev/reference/initializer_lecun_uniform.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Lecun uniform initializer. — initializer_lecun_uniform","text":"Initializer instance can passed layer variable constructors, called directly shape return Tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/initializer_lecun_uniform.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Lecun uniform initializer. — initializer_lecun_uniform","text":"","code":"# Standalone usage: initializer <- initializer_lecun_uniform() values <- initializer(shape = c(2, 2)) # Usage in a Keras layer: initializer <- initializer_lecun_uniform() layer <- layer_dense(units = 3, kernel_initializer = initializer)"},{"path":"https://keras3.posit.co/dev/reference/initializer_lecun_uniform.html","id":"reference","dir":"Reference","previous_headings":"","what":"Reference","title":"Lecun uniform initializer. — initializer_lecun_uniform","text":"Klambauer et al., 2017","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/initializer_ones.html","id":null,"dir":"Reference","previous_headings":"","what":"Initializer that generates tensors initialized to 1. — initializer_ones","title":"Initializer that generates tensors initialized to 1. — initializer_ones","text":"Also available via shortcut function ones.","code":""},{"path":"https://keras3.posit.co/dev/reference/initializer_ones.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Initializer that generates tensors initialized to 1. — initializer_ones","text":"","code":"initializer_ones()"},{"path":"https://keras3.posit.co/dev/reference/initializer_ones.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Initializer that generates tensors initialized to 1. — initializer_ones","text":"Initializer instance can passed layer variable constructors, called directly shape return Tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/initializer_ones.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Initializer that generates tensors initialized to 1. — initializer_ones","text":"","code":"# Standalone usage: initializer <- initializer_ones() values <- initializer(shape = c(2, 2)) # Usage in a Keras layer: initializer <- initializer_ones() layer <- layer_dense(units = 3, kernel_initializer = initializer)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/initializer_orthogonal.html","id":null,"dir":"Reference","previous_headings":"","what":"Initializer that generates an orthogonal matrix. — initializer_orthogonal","title":"Initializer that generates an orthogonal matrix. — initializer_orthogonal","text":"shape tensor initialize two-dimensional, initialized orthogonal matrix obtained QR decomposition matrix random numbers drawn normal distribution. matrix fewer rows columns output orthogonal rows. Otherwise, output orthogonal columns. shape tensor initialize two-dimensional, matrix shape (shape[1] * ... * shape[n - 1], shape[n]) initialized, n length shape vector. matrix subsequently reshaped give tensor desired shape.","code":""},{"path":"https://keras3.posit.co/dev/reference/initializer_orthogonal.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Initializer that generates an orthogonal matrix. — initializer_orthogonal","text":"","code":"initializer_orthogonal(gain = 1, seed = NULL)"},{"path":"https://keras3.posit.co/dev/reference/initializer_orthogonal.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Initializer that generates an orthogonal matrix. — initializer_orthogonal","text":"gain Multiplicative factor apply orthogonal matrix. seed integer. Used make behavior initializer deterministic.","code":""},{"path":"https://keras3.posit.co/dev/reference/initializer_orthogonal.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Initializer that generates an orthogonal matrix. — initializer_orthogonal","text":"Initializer instance can passed layer variable constructors, called directly shape return Tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/initializer_orthogonal.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Initializer that generates an orthogonal matrix. — initializer_orthogonal","text":"","code":"# Standalone usage: initializer <- initializer_orthogonal() values <- initializer(shape = c(2, 2)) # Usage in a Keras layer: initializer <- initializer_orthogonal() layer <- layer_dense(units = 3, kernel_initializer = initializer)"},{"path":"https://keras3.posit.co/dev/reference/initializer_orthogonal.html","id":"reference","dir":"Reference","previous_headings":"","what":"Reference","title":"Initializer that generates an orthogonal matrix. — initializer_orthogonal","text":"Saxe et al., 2014","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/initializer_random_normal.html","id":null,"dir":"Reference","previous_headings":"","what":"Random normal initializer. — initializer_random_normal","title":"Random normal initializer. — initializer_random_normal","text":"Draws samples normal distribution given parameters.","code":""},{"path":"https://keras3.posit.co/dev/reference/initializer_random_normal.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Random normal initializer. — initializer_random_normal","text":"","code":"initializer_random_normal(mean = 0, stddev = 0.05, seed = NULL)"},{"path":"https://keras3.posit.co/dev/reference/initializer_random_normal.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Random normal initializer. — initializer_random_normal","text":"mean numeric scalar. Mean random values generate. stddev numeric scalar. Standard deviation random values generate. seed integer instance random_seed_generator(). Used make behavior initializer deterministic. Note initializer seeded integer NULL (unseeded) produce random values across multiple calls. get different random values across multiple calls, use seed instance random_seed_generator().","code":""},{"path":"https://keras3.posit.co/dev/reference/initializer_random_normal.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Random normal initializer. — initializer_random_normal","text":"Initializer instance can passed layer variable constructors, called directly shape return Tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/initializer_random_normal.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Random normal initializer. — initializer_random_normal","text":"","code":"# Standalone usage: initializer <- initializer_random_normal(mean = 0.0, stddev = 1.0) values <- initializer(shape = c(2, 2)) # Usage in a Keras layer: initializer <- initializer_random_normal(mean = 0.0, stddev = 1.0) layer <- layer_dense(units = 3, kernel_initializer = initializer)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/initializer_random_uniform.html","id":null,"dir":"Reference","previous_headings":"","what":"Random uniform initializer. — initializer_random_uniform","title":"Random uniform initializer. — initializer_random_uniform","text":"Draws samples uniform distribution given parameters.","code":""},{"path":"https://keras3.posit.co/dev/reference/initializer_random_uniform.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Random uniform initializer. — initializer_random_uniform","text":"","code":"initializer_random_uniform(minval = -0.05, maxval = 0.05, seed = NULL)"},{"path":"https://keras3.posit.co/dev/reference/initializer_random_uniform.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Random uniform initializer. — initializer_random_uniform","text":"minval numeric scalar scalar keras tensor. Lower bound range random values generate (inclusive). maxval numeric scalar scalar keras tensor. Upper bound range random values generate (exclusive). seed integer instance random_seed_generator(). Used make behavior initializer deterministic. Note initializer seeded integer NULL (unseeded) produce random values across multiple calls. get different random values across multiple calls, use seed instance random_seed_generator().","code":""},{"path":"https://keras3.posit.co/dev/reference/initializer_random_uniform.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Random uniform initializer. — initializer_random_uniform","text":"Initializer instance can passed layer variable constructors, called directly shape return Tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/initializer_random_uniform.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Random uniform initializer. — initializer_random_uniform","text":"","code":"# Standalone usage: initializer <- initializer_random_uniform(minval = 0.0, maxval = 1.0) values <- initializer(shape = c(2, 2)) # Usage in a Keras layer: initializer <- initializer_random_uniform(minval = 0.0, maxval = 1.0) layer <- layer_dense(units = 3, kernel_initializer = initializer)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/initializer_stft.html","id":null,"dir":"Reference","previous_headings":"","what":"Initializer of Conv kernels for Short-term Fourier Transformation (STFT). — initializer_stft","title":"Initializer of Conv kernels for Short-term Fourier Transformation (STFT). — initializer_stft","text":"Since formula involves complex numbers, class compute either real imaginary components final output. Additionally, initializer supports windowing functions across time dimension commonly used STFT. Windowing functions Python module scipy.signal.windows supported, including common hann hamming windowing functions. layer supports periodic windows scaling-based normalization. primarily intended use STFTSpectrogram layer.","code":""},{"path":"https://keras3.posit.co/dev/reference/initializer_stft.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Initializer of Conv kernels for Short-term Fourier Transformation (STFT). — initializer_stft","text":"","code":"initializer_stft(   side = \"real\",   window = \"hann\",   scaling = \"density\",   periodic = FALSE )"},{"path":"https://keras3.posit.co/dev/reference/initializer_stft.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Initializer of Conv kernels for Short-term Fourier Transformation (STFT). — initializer_stft","text":"side String, \"real\" \"imag\" deciding kernel compute real side imaginary side output. Defaults \"real\". window String name windowing function scipy.signal.windows module, array_like window values, NULL windowing. scaling String, \"density\" \"spectrum\" scaling window normalization, either L2 L1 normalization. NULL scaling. periodic Boolean, True, window function treated periodic. Defaults FALSE.","code":""},{"path":"https://keras3.posit.co/dev/reference/initializer_stft.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Initializer of Conv kernels for Short-term Fourier Transformation (STFT). — initializer_stft","text":"Initializer instance can passed layer variable constructors, called directly shape return Tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/initializer_stft.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Initializer of Conv kernels for Short-term Fourier Transformation (STFT). — initializer_stft","text":"","code":"# Standalone usage: initializer <- initializer_stft(\"real\", \"hann\", \"density\", FALSE) values <- initializer(shape = c(128, 1, 513))"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/initializer_truncated_normal.html","id":null,"dir":"Reference","previous_headings":"","what":"Initializer that generates a truncated normal distribution. — initializer_truncated_normal","title":"Initializer that generates a truncated normal distribution. — initializer_truncated_normal","text":"values generated similar values RandomNormal initializer, except values two standard deviations mean discarded re-drawn.","code":""},{"path":"https://keras3.posit.co/dev/reference/initializer_truncated_normal.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Initializer that generates a truncated normal distribution. — initializer_truncated_normal","text":"","code":"initializer_truncated_normal(mean = 0, stddev = 0.05, seed = NULL)"},{"path":"https://keras3.posit.co/dev/reference/initializer_truncated_normal.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Initializer that generates a truncated normal distribution. — initializer_truncated_normal","text":"mean numeric scalar. Mean random values generate. stddev numeric scalar. Standard deviation random values generate. seed integer instance random_seed_generator(). Used make behavior initializer deterministic. Note initializer seeded integer NULL (unseeded) produce random values across multiple calls. get different random values across multiple calls, use seed instance random_seed_generator().","code":""},{"path":"https://keras3.posit.co/dev/reference/initializer_truncated_normal.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Initializer that generates a truncated normal distribution. — initializer_truncated_normal","text":"Initializer instance can passed layer variable constructors, called directly shape return Tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/initializer_truncated_normal.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Initializer that generates a truncated normal distribution. — initializer_truncated_normal","text":"","code":"# Standalone usage: initializer <- initializer_truncated_normal(mean = 0, stddev = 1) values <- initializer(shape = c(2, 2)) # Usage in a Keras layer: initializer <- initializer_truncated_normal(mean = 0, stddev = 1) layer <- layer_dense(units = 3, kernel_initializer = initializer)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/initializer_variance_scaling.html","id":null,"dir":"Reference","previous_headings":"","what":"Initializer that adapts its scale to the shape of its input tensors. — initializer_variance_scaling","title":"Initializer that adapts its scale to the shape of its input tensors. — initializer_variance_scaling","text":"distribution = \"truncated_normal\" \"untruncated_normal\", samples drawn truncated/untruncated normal distribution mean zero standard deviation (truncation, used) stddev = sqrt(scale / n), n : number input units weight tensor, mode = \"fan_in\" number output units, mode = \"fan_out\" average numbers input output units, mode = \"fan_avg\" distribution = \"uniform\", samples drawn uniform distribution within [-limit, limit], limit = sqrt(3 * scale / n).","code":""},{"path":"https://keras3.posit.co/dev/reference/initializer_variance_scaling.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Initializer that adapts its scale to the shape of its input tensors. — initializer_variance_scaling","text":"","code":"initializer_variance_scaling(   scale = 1,   mode = \"fan_in\",   distribution = \"truncated_normal\",   seed = NULL )"},{"path":"https://keras3.posit.co/dev/reference/initializer_variance_scaling.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Initializer that adapts its scale to the shape of its input tensors. — initializer_variance_scaling","text":"scale Scaling factor (positive float). mode One \"fan_in\", \"fan_out\", \"fan_avg\". distribution Random distribution use. One \"truncated_normal\", \"untruncated_normal\", \"uniform\". seed integer instance random_seed_generator(). Used make behavior initializer deterministic. Note initializer seeded integer NULL (unseeded) produce random values across multiple calls. get different random values across multiple calls, use seed instance random_seed_generator().","code":""},{"path":"https://keras3.posit.co/dev/reference/initializer_variance_scaling.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Initializer that adapts its scale to the shape of its input tensors. — initializer_variance_scaling","text":"Initializer instance can passed layer variable constructors, called directly shape return Tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/initializer_variance_scaling.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Initializer that adapts its scale to the shape of its input tensors. — initializer_variance_scaling","text":"","code":"# Standalone usage: initializer <- initializer_variance_scaling(scale = 0.1, mode = 'fan_in',                                             distribution = 'uniform') values <- initializer(shape = c(2, 2)) # Usage in a Keras layer: initializer <- initializer_variance_scaling(scale = 0.1, mode = 'fan_in',                                             distribution = 'uniform') layer <- layer_dense(units = 3, kernel_initializer = initializer)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/initializer_zeros.html","id":null,"dir":"Reference","previous_headings":"","what":"Initializer that generates tensors initialized to 0. — initializer_zeros","title":"Initializer that generates tensors initialized to 0. — initializer_zeros","text":"Initializer generates tensors initialized 0.","code":""},{"path":"https://keras3.posit.co/dev/reference/initializer_zeros.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Initializer that generates tensors initialized to 0. — initializer_zeros","text":"","code":"initializer_zeros()"},{"path":"https://keras3.posit.co/dev/reference/initializer_zeros.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Initializer that generates tensors initialized to 0. — initializer_zeros","text":"Initializer instance can passed layer variable constructors, called directly shape return Tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/initializer_zeros.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Initializer that generates tensors initialized to 0. — initializer_zeros","text":"","code":"# Standalone usage: initializer <- initializer_zeros() values <- initializer(shape = c(2, 2)) # Usage in a Keras layer: initializer <- initializer_zeros() layer <- layer_dense(units = 3, kernel_initializer = initializer)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/install_keras.html","id":null,"dir":"Reference","previous_headings":"","what":"Install Keras — install_keras","title":"Install Keras — install_keras","text":"function install Keras along selected backend, including Python dependencies.","code":""},{"path":"https://keras3.posit.co/dev/reference/install_keras.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Install Keras — install_keras","text":"","code":"install_keras(   envname = \"r-keras\",   ...,   extra_packages = c(\"scipy\", \"pandas\", \"Pillow\", \"pydot\", \"ipython\",     \"tensorflow_datasets\"),   python_version = \">=3.9,<=3.11\",   backend = c(\"tensorflow\", \"jax\"),   gpu = NA,   restart_session = TRUE )"},{"path":"https://keras3.posit.co/dev/reference/install_keras.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Install Keras — install_keras","text":"envname Name path Python virtual environment ... reserved future compatibility. extra_packages Additional Python packages install alongside Keras python_version Passed reticulate::virtualenv_starter() backend backend(s) install. Accepted values include \"tensorflow\", \"jax\" \"torch\" gpu whether install GPU capable version backend. restart_session Whether restart R session installing (note occur within RStudio).","code":""},{"path":"https://keras3.posit.co/dev/reference/install_keras.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Install Keras — install_keras","text":"return value, called side effects.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/keras.html","id":null,"dir":"Reference","previous_headings":"","what":"Main Keras module — keras","title":"Main Keras module — keras","text":"keras module object equivalent reticulate::import(\"keras\") provided mainly convenience.","code":""},{"path":"https://keras3.posit.co/dev/reference/keras.html","id":"format","dir":"Reference","previous_headings":"","what":"Format","title":"Main Keras module — keras","text":"object class python.builtin.module","code":""},{"path":"https://keras3.posit.co/dev/reference/keras.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Main Keras module — keras","text":"keras Python module","code":""},{"path":"https://keras3.posit.co/dev/reference/keras3-package.html","id":null,"dir":"Reference","previous_headings":"","what":"keras3: R Interface to 'Keras' — keras3-package","title":"keras3: R Interface to 'Keras' — keras3-package","text":"Interface 'Keras' https://keras.io, high-level neural networks API. 'Keras' developed focus enabling fast experimentation, supports convolution based networks recurrent networks (well combinations two), runs seamlessly CPU GPU devices. Keras high-level neural networks API, developed focus enabling fast experimentation. Keras following key features:","code":""},{"path":"https://keras3.posit.co/dev/reference/keras3-package.html","id":"details","dir":"Reference","previous_headings":"","what":"Details","title":"keras3: R Interface to 'Keras' — keras3-package","text":"Allows code run CPU GPU, seamlessly. User-friendly API makes easy quickly prototype deep learning models. Built-support convolutional networks (computer vision), recurrent networks (sequence processing), combination . Supports arbitrary network architectures: multi-input multi-output models, layer sharing, model sharing, etc. means Keras appropriate building essentially deep learning model, memory network neural Turing machine. capable running top multiple back-ends including TensorFlow, Jax, PyTorch. See package website https://keras3.posit.co complete documentation.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/keras3-package.html","id":"author","dir":"Reference","previous_headings":"","what":"Author","title":"keras3: R Interface to 'Keras' — keras3-package","text":"Maintainer: Tomasz Kalinowski tomasz@posit.co [copyright holder] Authors: JJ Allaire [copyright holder] François Chollet [copyright holder] contributors: Daniel Falbel daniel@posit.co [contributor, copyright holder] Posit Software, PBC [copyright holder, funder] Google [copyright holder, funder] Yuan Tang terrytangyuan@gmail.com (ORCID) [contributor, copyright holder] Wouter Van Der Bijl [contributor, copyright holder] Martin Studer [contributor, copyright holder] Sigrid Keydana [contributor]","code":""},{"path":"https://keras3.posit.co/dev/reference/keras_input.html","id":null,"dir":"Reference","previous_headings":"","what":"Create a Keras tensor (Functional API input). — keras_input","title":"Create a Keras tensor (Functional API input). — keras_input","text":"Keras tensor symbolic tensor-like object, augment certain attributes allow us build Keras model just knowing inputs outputs model. instance, , b c Keras tensors, becomes possible : model <- keras_model(input = c(, b), output = c)","code":""},{"path":"https://keras3.posit.co/dev/reference/keras_input.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Create a Keras tensor (Functional API input). — keras_input","text":"","code":"keras_input(   shape = NULL,   batch_size = NULL,   dtype = NULL,   sparse = NULL,   ragged = NULL,   batch_shape = NULL,   name = NULL,   tensor = NULL,   optional = FALSE )"},{"path":"https://keras3.posit.co/dev/reference/keras_input.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Create a Keras tensor (Functional API input). — keras_input","text":"shape shape list (list integers NULL objects), including batch size. instance, shape = c(32) indicates expected input batches 32-dimensional vectors. Elements list can NULL NA; NULL/NA elements represent dimensions shape known may vary (e.g. sequence length). batch_size Optional static batch size (integer). dtype data type expected input, string (e.g. \"float32\", \"int32\"...) sparse boolean specifying whether expected input sparse tensors. Note , sparse FALSE, sparse tensors can still passed input - densified default value 0. feature supported TensorFlow backend. Defaults FALSE. ragged boolean specifying whether expected input ragged tensors. Note , ragged FALSE, ragged tensors can still passed input - densified default value 0. feature supported TensorFlow backend. Defaults FALSE. batch_shape Optional shape list (list integers NULL objects), including batch size. name Optional name string layer. unique model (reuse name twice). autogenerated provided. tensor Optional existing tensor wrap Input layer. set, layer use tensor rather creating new placeholder tensor. optional Boolean, whether input optional . optional input can accept NULL values.","code":""},{"path":"https://keras3.posit.co/dev/reference/keras_input.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Create a Keras tensor (Functional API input). — keras_input","text":"Keras tensor, can passed inputs argument (keras_model()).","code":""},{"path":"https://keras3.posit.co/dev/reference/keras_input.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Create a Keras tensor (Functional API input). — keras_input","text":"","code":"# This is a logistic regression in Keras input <- layer_input(shape=c(32)) output <- input |> layer_dense(16, activation='softmax') model <- keras_model(input, output)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/keras_model.html","id":null,"dir":"Reference","previous_headings":"","what":"Keras Model (Functional API) — keras_model","title":"Keras Model (Functional API) — keras_model","text":"model directed acyclic graph layers.","code":""},{"path":"https://keras3.posit.co/dev/reference/keras_model.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Keras Model (Functional API) — keras_model","text":"","code":"keras_model(inputs = NULL, outputs = NULL, ...)"},{"path":"https://keras3.posit.co/dev/reference/keras_model.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Keras Model (Functional API) — keras_model","text":"inputs Input tensor(s) (keras_input()) outputs Output tensors (calling layers inputs) ... additional arguments","code":""},{"path":"https://keras3.posit.co/dev/reference/keras_model.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Keras Model (Functional API) — keras_model","text":"Model instance.","code":""},{"path":"https://keras3.posit.co/dev/reference/keras_model.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Keras Model (Functional API) — keras_model","text":"","code":"library(keras3)  # input tensor inputs <- keras_input(shape = c(784))  # outputs compose input + dense layers predictions <- inputs |>   layer_dense(units = 64, activation = 'relu') |>   layer_dense(units = 64, activation = 'relu') |>   layer_dense(units = 10, activation = 'softmax')  # create and compile model model <- keras_model(inputs = inputs, outputs = predictions) model |> compile(   optimizer = 'rmsprop',   loss = 'categorical_crossentropy',   metrics = c('accuracy') )"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/keras_model_sequential.html","id":null,"dir":"Reference","previous_headings":"","what":"Keras Model composed of a linear stack of layers — keras_model_sequential","title":"Keras Model composed of a linear stack of layers — keras_model_sequential","text":"Keras Model composed linear stack layers","code":""},{"path":"https://keras3.posit.co/dev/reference/keras_model_sequential.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Keras Model composed of a linear stack of layers — keras_model_sequential","text":"","code":"keras_model_sequential(   input_shape = NULL,   name = NULL,   ...,   input_dtype = NULL,   input_batch_size = NULL,   input_sparse = NULL,   input_ragged = NULL,   input_batch_shape = NULL,   input_name = NULL,   input_tensor = NULL,   input_optional = FALSE,   trainable = TRUE,   layers = list() )"},{"path":"https://keras3.posit.co/dev/reference/keras_model_sequential.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Keras Model composed of a linear stack of layers — keras_model_sequential","text":"input_shape shape integer vector, including batch size. instance, shape=c(32) indicates expected input batches 32-dimensional vectors. Elements shape can NA; NA elements represent dimensions shape known may vary (e.g. sequence length). name Name model ... additional arguments passed keras.layers.InputLayer. input_dtype data type expected input, string (e.g. \"float32\", \"int32\"...) input_batch_size Optional static batch size (integer). input_sparse boolean specifying whether expected input sparse tensors. Note , sparse FALSE, sparse tensors can still passed input - densified default value 0. feature supported TensorFlow backend. Defaults FALSE. input_ragged bool. input_batch_shape optional way specify batch_size input_shape one argument. input_name Optional name string input layer. unique model (reuse name twice). autogenerated provided. input_tensor Optional existing tensor wrap InputLayer. set, layer use tensor rather creating new placeholder tensor. input_optional Boolean, whether input optional . optional input can accept NULL values. trainable Boolean, whether model's variables trainable. can also change trainable status model/layer freeze_weights() unfreeze_weights(). layers List layers add model.","code":""},{"path":"https://keras3.posit.co/dev/reference/keras_model_sequential.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Keras Model composed of a linear stack of layers — keras_model_sequential","text":"Sequential model instance.","code":""},{"path":"https://keras3.posit.co/dev/reference/keras_model_sequential.html","id":"note","dir":"Reference","previous_headings":"","what":"Note","title":"Keras Model composed of a linear stack of layers — keras_model_sequential","text":"input_shape omitted, model layer shapes, including final model output shape, known model built, either calling model input tensor/array like model(input), (possibly via fit()/evaluate()/predict()), explicitly calling model$build(input_shape).","code":""},{"path":"https://keras3.posit.co/dev/reference/keras_model_sequential.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Keras Model composed of a linear stack of layers — keras_model_sequential","text":"","code":"model <- keras_model_sequential(input_shape = c(784)) model |>   layer_dense(units = 32) |>   layer_activation('relu') |>   layer_dense(units = 10) |>   layer_activation('softmax')  model |> compile(   optimizer = 'rmsprop',   loss = 'categorical_crossentropy',   metrics = c('accuracy') )  model ## Model: \"sequential\" ## +---------------------------------+------------------------+---------------+ ## | Layer (type)                    | Output Shape           |       Param # | ## +=================================+========================+===============+ ## | dense (Dense)                   | (None, 32)             |        25,120 | ## +---------------------------------+------------------------+---------------+ ## | activation (Activation)         | (None, 32)             |             0 | ## +---------------------------------+------------------------+---------------+ ## | dense_1 (Dense)                 | (None, 10)             |           330 | ## +---------------------------------+------------------------+---------------+ ## | activation_1 (Activation)       | (None, 10)             |             0 | ## +---------------------------------+------------------------+---------------+ ##  Total params: 25,450 (99.41 KB) ##  Trainable params: 25,450 (99.41 KB) ##  Non-trainable params: 0 (0.00 B)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/keras_variable.html","id":null,"dir":"Reference","previous_headings":"","what":"Represents a backend-agnostic variable in Keras. — keras_variable","title":"Represents a backend-agnostic variable in Keras. — keras_variable","text":"Variable acts container state. holds tensor value can updated. JAX backend, variables used implement \"functionalization\", pattern lifting stateful operations piece computation turn stateless function.","code":""},{"path":"https://keras3.posit.co/dev/reference/keras_variable.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Represents a backend-agnostic variable in Keras. — keras_variable","text":"","code":"keras_variable(   initializer,   shape = NULL,   dtype = NULL,   trainable = TRUE,   autocast = TRUE,   aggregation = \"none\",   synchronization = \"auto\",   name = NULL,   ... )"},{"path":"https://keras3.posit.co/dev/reference/keras_variable.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Represents a backend-agnostic variable in Keras. — keras_variable","text":"initializer Initial value callable initialization. callable used, take arguments shape dtype. shape Optional. Tuple variable's shape. Required initializer callable. dtype Optional. Data type variable. Defaults global float dtype type (\"float32\" never configured). trainable Optional. Boolean indicating variable trainable. Defaults TRUE. autocast Optional. Boolean indicating whether variable supports autocasting. TRUE, layer may first convert variable compute data type accessed. Defaults TRUE. aggregation Optional string, one NULL, \"none\", \"mean\", \"sum\" \"only_first_replica\" specifying distributed variable aggregated. serves semantic annotation, taken account downstream backends users. Defaults \"none\". synchronization Optional string specifying distributed values synchronized. Defaults \"auto\". name Optional. unique name variable. Automatically generated set. ... Additional backend-specific keyword arguments forwarded keras$Variable().","code":""},{"path":"https://keras3.posit.co/dev/reference/keras_variable.html","id":"attributes","dir":"Reference","previous_headings":"","what":"Attributes","title":"Represents a backend-agnostic variable in Keras. — keras_variable","text":"shape: shape variable (tuple integers). ndim: number dimensions variable (integer). dtype: data type variable (string). trainable: Whether variable trainable (boolean). autocast: Whether variable supports autocasting (boolean). aggregation: distributed variable aggregated (string). synchronization: Strategy synchronizing variable across devices (string). value: current value variable (NumPy array tensor). name: name variable (string). path: path variable within Keras model layer (string).","code":""},{"path":"https://keras3.posit.co/dev/reference/keras_variable.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Represents a backend-agnostic variable in Keras. — keras_variable","text":"Initializing Variable NumPy array:   Using Keras initializer create Variable:           Marking Variable non-trainable:","code":"initial_array <- array(1, c(3, 3)) variable_from_array <- keras_variable(initializer = initial_array) variable_from_initializer <- keras_variable(   initializer = initializer_ones(),   shape = c(3, 3),   dtype = \"float32\" ) new_value <- array(0, c(3, 3)) variable_from_array$assign(new_value) ## tf.Tensor( ## [[0. 0. 0.] ##  [0. 0. 0.] ##  [0. 0. 0.]], shape=(3, 3), dtype=float64) # To modify a subset of values value <- variable_from_array$value value@r[1,] <- 99 invisible(variable_from_array$assign(value)) variable_from_array ## <Variable path=variable, shape=(3, 3), dtype=float64, value=[[99. 99. 99.] ##  [ 0.  0.  0.] ##  [ 0.  0.  0.]]> non_trainable_variable <- keras_variable(   initializer = array(1, c(3, 3)),   dtype = \"float32\",   trainable = FALSE )"},{"path":"https://keras3.posit.co/dev/reference/layer_activation.html","id":null,"dir":"Reference","previous_headings":"","what":"Applies an activation function to an output. — layer_activation","title":"Applies an activation function to an output. — layer_activation","text":"Applies activation function output.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_activation.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Applies an activation function to an output. — layer_activation","text":"","code":"layer_activation(object, activation, ...)"},{"path":"https://keras3.posit.co/dev/reference/layer_activation.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Applies an activation function to an output. — layer_activation","text":"object Object compose layer . tensor, array, sequential model. activation Activation function. callable, name activation keras3::activation_* namespace. ... Base layer keyword arguments, name dtype.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_activation.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Applies an activation function to an output. — layer_activation","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_activation.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Applies an activation function to an output. — layer_activation","text":"","code":"x <- array(c(-3, -1, 0, 2)) layer <- layer_activation(activation = 'relu') layer(x) ## tf.Tensor([0. 0. 0. 2.], shape=(4), dtype=float32) layer <- layer_activation(activation = activation_relu) layer(x) ## tf.Tensor([0. 0. 0. 2.], shape=(4), dtype=float32) layer <- layer_activation(activation = op_relu) layer(x) ## tf.Tensor([0. 0. 0. 2.], shape=(4), dtype=float32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_activation_elu.html","id":null,"dir":"Reference","previous_headings":"","what":"Applies an Exponential Linear Unit function to an output. — layer_activation_elu","title":"Applies an Exponential Linear Unit function to an output. — layer_activation_elu","text":"Formula:","code":"f(x) = alpha * (exp(x) - 1.) for x < 0 f(x) = x for x >= 0"},{"path":"https://keras3.posit.co/dev/reference/layer_activation_elu.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Applies an Exponential Linear Unit function to an output. — layer_activation_elu","text":"","code":"layer_activation_elu(object, alpha = 1, ...)"},{"path":"https://keras3.posit.co/dev/reference/layer_activation_elu.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Applies an Exponential Linear Unit function to an output. — layer_activation_elu","text":"object Object compose layer . tensor, array, sequential model. alpha float, slope negative section. Defaults 1.0. ... Base layer keyword arguments, name dtype.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_activation_elu.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Applies an Exponential Linear Unit function to an output. — layer_activation_elu","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_activation_leaky_relu.html","id":null,"dir":"Reference","previous_headings":"","what":"Leaky version of a Rectified Linear Unit activation layer. — layer_activation_leaky_relu","title":"Leaky version of a Rectified Linear Unit activation layer. — layer_activation_leaky_relu","text":"layer allows small gradient unit active. Formula:","code":"f <- function(x) ifelse(x >= 0, x, alpha * x)"},{"path":"https://keras3.posit.co/dev/reference/layer_activation_leaky_relu.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Leaky version of a Rectified Linear Unit activation layer. — layer_activation_leaky_relu","text":"","code":"layer_activation_leaky_relu(object, negative_slope = 0.3, ...)"},{"path":"https://keras3.posit.co/dev/reference/layer_activation_leaky_relu.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Leaky version of a Rectified Linear Unit activation layer. — layer_activation_leaky_relu","text":"object Object compose layer . tensor, array, sequential model. negative_slope Float >= 0.0. Negative slope coefficient. Defaults 0.3. ... Base layer keyword arguments, name dtype.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_activation_leaky_relu.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Leaky version of a Rectified Linear Unit activation layer. — layer_activation_leaky_relu","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_activation_leaky_relu.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Leaky version of a Rectified Linear Unit activation layer. — layer_activation_leaky_relu","text":"","code":"leaky_relu_layer <- layer_activation_leaky_relu(negative_slope=0.5) input <- array(c(-10, -5, 0.0, 5, 10)) result <- leaky_relu_layer(input) as.array(result) ## [1] -5.0 -2.5  0.0  5.0 10.0"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_activation_parametric_relu.html","id":null,"dir":"Reference","previous_headings":"","what":"Parametric Rectified Linear Unit activation layer. — layer_activation_parametric_relu","title":"Parametric Rectified Linear Unit activation layer. — layer_activation_parametric_relu","text":"Formula:   alpha learned array shape x.","code":"f <- function(x) ifelse(x >= 0, x, alpha * x)"},{"path":"https://keras3.posit.co/dev/reference/layer_activation_parametric_relu.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Parametric Rectified Linear Unit activation layer. — layer_activation_parametric_relu","text":"","code":"layer_activation_parametric_relu(   object,   alpha_initializer = \"Zeros\",   alpha_regularizer = NULL,   alpha_constraint = NULL,   shared_axes = NULL,   ... )"},{"path":"https://keras3.posit.co/dev/reference/layer_activation_parametric_relu.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Parametric Rectified Linear Unit activation layer. — layer_activation_parametric_relu","text":"object Object compose layer . tensor, array, sequential model. alpha_initializer Initializer function weights. alpha_regularizer Regularizer weights. alpha_constraint Constraint weights. shared_axes axes along share learnable parameters activation function. example, incoming feature maps 2D convolution output shape (batch, height, width, channels), wish share parameters across space filter one set parameters, set shared_axes=[1, 2]. ... Base layer keyword arguments, name dtype.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_activation_parametric_relu.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Parametric Rectified Linear Unit activation layer. — layer_activation_parametric_relu","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_activation_relu.html","id":null,"dir":"Reference","previous_headings":"","what":"Rectified Linear Unit activation function layer. — layer_activation_relu","title":"Rectified Linear Unit activation function layer. — layer_activation_relu","text":"Formula:","code":"f <- function(x, max_value = Inf, negative_slope = 0, threshold = 0) {  x <- max(x,0)  if (x >= max_value)    max_value  else if (threshold <= x && x < max_value)    x  else    negative_slope * (x - threshold) }"},{"path":"https://keras3.posit.co/dev/reference/layer_activation_relu.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Rectified Linear Unit activation function layer. — layer_activation_relu","text":"","code":"layer_activation_relu(   object,   max_value = NULL,   negative_slope = 0,   threshold = 0,   ... )"},{"path":"https://keras3.posit.co/dev/reference/layer_activation_relu.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Rectified Linear Unit activation function layer. — layer_activation_relu","text":"object Object compose layer . tensor, array, sequential model. max_value Float >= 0. Maximum activation value. NULL means unlimited. Defaults NULL. negative_slope Float >= 0. Negative slope coefficient. Defaults 0.0. threshold Float >= 0. Threshold value thresholded activation. Defaults 0.0. ... Base layer keyword arguments, name dtype.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_activation_relu.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Rectified Linear Unit activation function layer. — layer_activation_relu","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_activation_relu.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Rectified Linear Unit activation function layer. — layer_activation_relu","text":"","code":"relu_layer <- layer_activation_relu(max_value = 10,                                     negative_slope = 0.5,                                     threshold = 0) input <- array(c(-10, -5, 0.0, 5, 10)) result <- relu_layer(input) as.array(result) ## [1] -5.0 -2.5  0.0  5.0 10.0"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_activation_softmax.html","id":null,"dir":"Reference","previous_headings":"","what":"Softmax activation layer. — layer_activation_softmax","title":"Softmax activation layer. — layer_activation_softmax","text":"Formula:","code":"exp_x = exp(x - max(x)) f(x) = exp_x / sum(exp_x)"},{"path":"https://keras3.posit.co/dev/reference/layer_activation_softmax.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Softmax activation layer. — layer_activation_softmax","text":"","code":"layer_activation_softmax(object, axis = -1L, ...)"},{"path":"https://keras3.posit.co/dev/reference/layer_activation_softmax.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Softmax activation layer. — layer_activation_softmax","text":"object Object compose layer . tensor, array, sequential model. axis Integer, list Integers, axis along softmax normalization applied. ... Base layer keyword arguments, name dtype.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_activation_softmax.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Softmax activation layer. — layer_activation_softmax","text":"Softmaxed output shape inputs.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_activation_softmax.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Softmax activation layer. — layer_activation_softmax","text":"","code":"softmax_layer <- layer_activation_softmax() input <- op_array(c(1, 2, 1)) softmax_layer(input) ## tf.Tensor([0.21194157 0.5761169  0.21194157], shape=(3), dtype=float32)"},{"path":"https://keras3.posit.co/dev/reference/layer_activation_softmax.html","id":"call-arguments","dir":"Reference","previous_headings":"","what":"Call Arguments","title":"Softmax activation layer. — layer_activation_softmax","text":"inputs: inputs (logits) softmax layer. mask: boolean mask shape inputs. mask specifies 1 keep 0 mask. Defaults NULL.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_activity_regularization.html","id":null,"dir":"Reference","previous_headings":"","what":"Layer that applies an update to the cost function based input activity. — layer_activity_regularization","title":"Layer that applies an update to the cost function based input activity. — layer_activity_regularization","text":"Layer applies update cost function based input activity.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_activity_regularization.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Layer that applies an update to the cost function based input activity. — layer_activity_regularization","text":"","code":"layer_activity_regularization(object, l1 = 0, l2 = 0, ...)"},{"path":"https://keras3.posit.co/dev/reference/layer_activity_regularization.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Layer that applies an update to the cost function based input activity. — layer_activity_regularization","text":"object Object compose layer . tensor, array, sequential model. l1 L1 regularization factor (positive float). l2 L2 regularization factor (positive float). ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_activity_regularization.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Layer that applies an update to the cost function based input activity. — layer_activity_regularization","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_activity_regularization.html","id":"input-shape","dir":"Reference","previous_headings":"","what":"Input Shape","title":"Layer that applies an update to the cost function based input activity. — layer_activity_regularization","text":"Arbitrary. Use keyword argument input_shape (tuple integers, include samples axis) using layer first layer model.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_activity_regularization.html","id":"output-shape","dir":"Reference","previous_headings":"","what":"Output Shape","title":"Layer that applies an update to the cost function based input activity. — layer_activity_regularization","text":"shape input.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_add.html","id":null,"dir":"Reference","previous_headings":"","what":"Performs elementwise addition operation. — layer_add","title":"Performs elementwise addition operation. — layer_add","text":"takes input list tensors, shape, returns single tensor (also shape).","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_add.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Performs elementwise addition operation. — layer_add","text":"","code":"layer_add(inputs, ...)"},{"path":"https://keras3.posit.co/dev/reference/layer_add.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Performs elementwise addition operation. — layer_add","text":"inputs layers combine ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_add.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Performs elementwise addition operation. — layer_add","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_add.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Performs elementwise addition operation. — layer_add","text":"Usage Keras model:","code":"input_shape <- c(1, 2, 3) x1 <- op_ones(input_shape) x2 <- op_ones(input_shape) layer_add(x1, x2) ## tf.Tensor( ## [[[2. 2. 2.] ##   [2. 2. 2.]]], shape=(1, 2, 3), dtype=float32) input1 <- layer_input(shape = c(16)) x1 <- input1 |> layer_dense(8, activation = 'relu')  input2 <- layer_input(shape = c(32)) x2 <- input2 |> layer_dense(8, activation = 'relu')  # equivalent to `added = layer_add([x1, x2))` added <- layer_add(x1, x2) output <- added |> layer_dense(4)  model <- keras_model(inputs = c(input1, input2), outputs = output)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_additive_attention.html","id":null,"dir":"Reference","previous_headings":"","what":"Additive attention layer, a.k.a. Bahdanau-style attention. — layer_additive_attention","title":"Additive attention layer, a.k.a. Bahdanau-style attention. — layer_additive_attention","text":"Inputs list 2 3 elements: query tensor shape (batch_size, Tq, dim). value tensor shape (batch_size, Tv, dim). optional key tensor shape (batch_size, Tv, dim). none supplied, value used key. calculation follows steps: Calculate attention scores using query key shape (batch_size, Tq, Tv) non-linear sum scores = reduce_sum(tanh(query + key), axis=-1). Use scores calculate softmax distribution shape (batch_size, Tq, Tv). Use softmax distribution create linear combination value shape (batch_size, Tq, dim).","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_additive_attention.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Additive attention layer, a.k.a. Bahdanau-style attention. — layer_additive_attention","text":"","code":"layer_additive_attention(object, use_scale = TRUE, dropout = 0, ...)"},{"path":"https://keras3.posit.co/dev/reference/layer_additive_attention.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Additive attention layer, a.k.a. Bahdanau-style attention. — layer_additive_attention","text":"object Object compose layer . tensor, array, sequential model. use_scale TRUE, create scalar variable scale attention scores. dropout Float 0 1. Fraction units drop attention scores. Defaults 0.0. ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_additive_attention.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Additive attention layer, a.k.a. Bahdanau-style attention. — layer_additive_attention","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_additive_attention.html","id":"call-arguments","dir":"Reference","previous_headings":"","what":"Call Arguments","title":"Additive attention layer, a.k.a. Bahdanau-style attention. — layer_additive_attention","text":"inputs: List following tensors: query: Query tensor shape (batch_size, Tq, dim). value: Value tensor shape (batch_size, Tv, dim). key: Optional key tensor shape (batch_size, Tv, dim). given, use value key value, common case. mask: List following tensors: query_mask: boolean mask tensor shape (batch_size, Tq). given, output zero positions mask==FALSE. value_mask: boolean mask tensor shape (batch_size, Tv). given, apply mask values positions mask==FALSE contribute result. return_attention_scores: bool, TRUE, returns attention scores (masking softmax) additional output argument. training: Python boolean indicating whether layer behave training mode (adding dropout) inference mode (dropout). use_causal_mask: Boolean. Set TRUE decoder self-attention. Adds mask position attend positions j > . prevents flow information future towards past. Defaults FALSE.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_additive_attention.html","id":"output","dir":"Reference","previous_headings":"","what":"Output","title":"Additive attention layer, a.k.a. Bahdanau-style attention. — layer_additive_attention","text":"Attention outputs shape (batch_size, Tq, dim). (Optional) Attention scores masking softmax shape (batch_size, Tq, Tv).","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_alpha_dropout.html","id":null,"dir":"Reference","previous_headings":"","what":"Applies Alpha Dropout to the input. — layer_alpha_dropout","title":"Applies Alpha Dropout to the input. — layer_alpha_dropout","text":"Alpha Dropout Dropout keeps mean variance inputs original values, order ensure self-normalizing property even dropout. Alpha Dropout fits well Scaled Exponential Linear Units (SELU) randomly setting activations negative saturation value.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_alpha_dropout.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Applies Alpha Dropout to the input. — layer_alpha_dropout","text":"","code":"layer_alpha_dropout(object, rate, noise_shape = NULL, seed = NULL, ...)"},{"path":"https://keras3.posit.co/dev/reference/layer_alpha_dropout.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Applies Alpha Dropout to the input. — layer_alpha_dropout","text":"object Object compose layer . tensor, array, sequential model. rate Float 0 1. multiplicative noise standard deviation sqrt(rate / (1 - rate)). noise_shape 1D integer tensor representing shape binary alpha dropout mask multiplied input. instance, inputs shape (batch_size, timesteps, features) want alpha dropout mask timesteps, can use noise_shape = (batch_size, 1, features). seed integer use random seed. ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_alpha_dropout.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Applies Alpha Dropout to the input. — layer_alpha_dropout","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_alpha_dropout.html","id":"call-arguments","dir":"Reference","previous_headings":"","what":"Call Arguments","title":"Applies Alpha Dropout to the input. — layer_alpha_dropout","text":"inputs: Input tensor (rank). training: R boolean indicating whether layer behave training mode (adding alpha dropout) inference mode (nothing).","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_attention.html","id":null,"dir":"Reference","previous_headings":"","what":"Dot-product attention layer, a.k.a. Luong-style attention. — layer_attention","title":"Dot-product attention layer, a.k.a. Luong-style attention. — layer_attention","text":"Inputs list 2 3 elements: query tensor shape (batch_size, Tq, dim). value tensor shape (batch_size, Tv, dim). optional key tensor shape (batch_size, Tv, dim). none supplied, value used key. calculation follows steps: Calculate attention scores using query key shape (batch_size, Tq, Tv). Use scores calculate softmax distribution shape (batch_size, Tq, Tv). Use softmax distribution create linear combination value shape (batch_size, Tq, dim).","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_attention.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Dot-product attention layer, a.k.a. Luong-style attention. — layer_attention","text":"","code":"layer_attention(   object,   use_scale = FALSE,   score_mode = \"dot\",   dropout = 0,   seed = NULL,   ... )"},{"path":"https://keras3.posit.co/dev/reference/layer_attention.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Dot-product attention layer, a.k.a. Luong-style attention. — layer_attention","text":"object Object compose layer . tensor, array, sequential model. use_scale TRUE, create scalar variable scale attention scores. score_mode Function use compute attention scores, one {\"dot\", \"concat\"}. \"dot\" refers dot product query key vectors. \"concat\" refers hyperbolic tangent concatenation query key vectors. dropout Float 0 1. Fraction units drop attention scores. Defaults 0.0. seed integer use random seed case dropout. ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_attention.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Dot-product attention layer, a.k.a. Luong-style attention. — layer_attention","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_attention.html","id":"call-arguments","dir":"Reference","previous_headings":"","what":"Call Arguments","title":"Dot-product attention layer, a.k.a. Luong-style attention. — layer_attention","text":"inputs: List following tensors: query: Query tensor shape (batch_size, Tq, dim). value: Value tensor shape (batch_size, Tv, dim). key: Optional key tensor shape (batch_size, Tv, dim). given, use value key value, common case. mask: List following tensors: query_mask: boolean mask tensor shape (batch_size, Tq). given, output zero positions mask==FALSE. value_mask: boolean mask tensor shape (batch_size, Tv). given, apply mask values positions mask==FALSE contribute result. return_attention_scores: bool, TRUE, returns attention scores (masking softmax) additional output argument. training: Python boolean indicating whether layer behave training mode (adding dropout) inference mode (dropout). use_causal_mask: Boolean. Set TRUE decoder self-attention. Adds mask position attend positions j > . prevents flow information future towards past. Defaults FALSE.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_attention.html","id":"output","dir":"Reference","previous_headings":"","what":"Output","title":"Dot-product attention layer, a.k.a. Luong-style attention. — layer_attention","text":"Attention outputs shape (batch_size, Tq, dim). (Optional) Attention scores masking softmax shape (batch_size, Tq, Tv).","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_aug_mix.html","id":null,"dir":"Reference","previous_headings":"","what":"Performs the AugMix data augmentation technique. — layer_aug_mix","title":"Performs the AugMix data augmentation technique. — layer_aug_mix","text":"AugMix aims produce images variety preserving image semantics local statistics. augmentation process, augmentation applied across images batch num_chains different ways, chain consisting chain_depth augmentations.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_aug_mix.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Performs the AugMix data augmentation technique. — layer_aug_mix","text":"","code":"layer_aug_mix(   object,   value_range = c(0L, 255L),   num_chains = 3L,   chain_depth = 3L,   factor = 0.3,   alpha = 1,   all_ops = TRUE,   interpolation = \"bilinear\",   seed = NULL,   data_format = NULL,   ... )"},{"path":"https://keras3.posit.co/dev/reference/layer_aug_mix.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Performs the AugMix data augmentation technique. — layer_aug_mix","text":"object Object compose layer . tensor, array, sequential model. value_range range values incoming images . Represented two number tuple written (low, high). typically either (0, 1) (0, 255) depending preprocessing pipeline set . num_chains integer representing number different chains mixed, defaults 3. chain_depth integer representing maximum number transformations applied chain. actual number transformations chain sampled randomly range [0, chain_depth]. Defaults 3. factor strength augmentation normalized value 0 1. Default 0.3. alpha float value used probability coefficients Beta Dirichlet distributions, defaults 1.0. all_ops Use operations (including random_brightness, random_color_degeneration, random_contrast random_sharpness). Default True. interpolation interpolation method use resizing operations. Options include \"nearest\", \"bilinear\". Default \"bilinear\". seed Integer. Used create random seed. data_format string, either \"channels_last\" \"channels_first\". ordering dimensions inputs. \"channels_last\" corresponds inputs shape (batch, height, width, channels) \"channels_first\" corresponds inputs shape (batch, channels, height, width). defaults image_data_format value found Keras config file ~/.keras/keras.json. never set , \"channels_last\". ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_aug_mix.html","id":"references","dir":"Reference","previous_headings":"","what":"References","title":"Performs the AugMix data augmentation technique. — layer_aug_mix","text":"AugMix paper Official Code","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_auto_contrast.html","id":null,"dir":"Reference","previous_headings":"","what":"Performs the auto-contrast operation on an image. — layer_auto_contrast","title":"Performs the auto-contrast operation on an image. — layer_auto_contrast","text":"Auto contrast stretches values image across entire available value_range. makes differences pixels obvious. example image values [0, 1] range [0, 255], auto contrast change 1 values 255. layer active training inference time.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_auto_contrast.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Performs the auto-contrast operation on an image. — layer_auto_contrast","text":"","code":"layer_auto_contrast(object, value_range = tuple(0L, 255L), ...)"},{"path":"https://keras3.posit.co/dev/reference/layer_auto_contrast.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Performs the auto-contrast operation on an image. — layer_auto_contrast","text":"object Object compose layer . tensor, array, sequential model. value_range Range values incoming images . Represented two number tuple written (low, high). typically either (0, 1) (0, 255) depending preprocessing pipeline set . Defaults (0, 255). ... forward/backward compatability.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_average.html","id":null,"dir":"Reference","previous_headings":"","what":"Averages a list of inputs element-wise.. — layer_average","title":"Averages a list of inputs element-wise.. — layer_average","text":"takes input list tensors, shape, returns single tensor (also shape).","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_average.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Averages a list of inputs element-wise.. — layer_average","text":"","code":"layer_average(inputs, ...)"},{"path":"https://keras3.posit.co/dev/reference/layer_average.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Averages a list of inputs element-wise.. — layer_average","text":"inputs layers combine ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_average.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Averages a list of inputs element-wise.. — layer_average","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_average.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Averages a list of inputs element-wise.. — layer_average","text":"Usage Keras model:","code":"input_shape <- c(1, 2, 3) x1 <- op_ones(input_shape) x2 <- op_zeros(input_shape) layer_average(x1, x2) ## tf.Tensor( ## [[[0.5 0.5 0.5] ##   [0.5 0.5 0.5]]], shape=(1, 2, 3), dtype=float32) input1 <- layer_input(shape = c(16)) x1 <- input1 |> layer_dense(8, activation = 'relu')  input2 <- layer_input(shape = c(32)) x2 <- input2 |> layer_dense(8, activation = 'relu')  added <- layer_average(x1, x2) output <- added |> layer_dense(4)  model <- keras_model(inputs = c(input1, input2), outputs = output)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_average_pooling_1d.html","id":null,"dir":"Reference","previous_headings":"","what":"Average pooling for temporal data. — layer_average_pooling_1d","title":"Average pooling for temporal data. — layer_average_pooling_1d","text":"Downsamples input representation taking average value window defined pool_size. window shifted strides.  resulting output using \"valid\" padding option shape : output_shape = (input_shape - pool_size + 1) / strides) resulting output shape using \"\" padding option : output_shape = input_shape / strides","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_average_pooling_1d.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Average pooling for temporal data. — layer_average_pooling_1d","text":"","code":"layer_average_pooling_1d(   object,   pool_size,   strides = NULL,   padding = \"valid\",   data_format = NULL,   name = NULL,   ... )"},{"path":"https://keras3.posit.co/dev/reference/layer_average_pooling_1d.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Average pooling for temporal data. — layer_average_pooling_1d","text":"object Object compose layer . tensor, array, sequential model. pool_size int, size max pooling window. strides int NULL. Specifies much pooling window moves pooling step. NULL, default pool_size. padding string, either \"valid\" \"\" (case-insensitive). \"valid\" means padding. \"\" results padding evenly left/right /input output height/width dimension input. data_format string, either \"channels_last\" \"channels_first\". ordering dimensions inputs. \"channels_last\" corresponds inputs shape (batch, steps, features) \"channels_first\" corresponds inputs shape (batch, features, steps). defaults image_data_format value found Keras config file ~/.keras/keras.json. never set , \"channels_last\". name String, name object ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_average_pooling_1d.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Average pooling for temporal data. — layer_average_pooling_1d","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_average_pooling_1d.html","id":"input-shape","dir":"Reference","previous_headings":"","what":"Input Shape","title":"Average pooling for temporal data. — layer_average_pooling_1d","text":"data_format=\"channels_last\": 3D tensor shape (batch_size, steps, features). data_format=\"channels_first\": 3D tensor shape (batch_size, features, steps).","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_average_pooling_1d.html","id":"output-shape","dir":"Reference","previous_headings":"","what":"Output Shape","title":"Average pooling for temporal data. — layer_average_pooling_1d","text":"data_format=\"channels_last\": 3D tensor shape (batch_size, downsampled_steps, features). data_format=\"channels_first\": 3D tensor shape (batch_size, features, downsampled_steps).","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_average_pooling_1d.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Average pooling for temporal data. — layer_average_pooling_1d","text":"strides=1 padding=\"valid\":     strides=2 padding=\"valid\":     strides=1 padding=\"\":","code":"x <- op_array(c(1., 2., 3., 4., 5.)) |> op_reshape(c(1, 5, 1)) output <- x |>   layer_average_pooling_1d(pool_size = 2,                            strides = 1,                            padding = \"valid\") output ## tf.Tensor( ## [[[1.5] ##   [2.5] ##   [3.5] ##   [4.5]]], shape=(1, 4, 1), dtype=float32) x <- op_array(c(1., 2., 3., 4., 5.)) |> op_reshape(c(1, 5, 1)) output <- x |>   layer_average_pooling_1d(pool_size = 2,                            strides = 2,                            padding = \"valid\") output ## tf.Tensor( ## [[[1.5] ##   [3.5]]], shape=(1, 2, 1), dtype=float32) x <- op_array(c(1., 2., 3., 4., 5.)) |> op_reshape(c(1, 5, 1)) output <- x |>   layer_average_pooling_1d(pool_size = 2,                            strides = 1,                            padding = \"same\") output ## tf.Tensor( ## [[[1.5] ##   [2.5] ##   [3.5] ##   [4.5] ##   [5. ]]], shape=(1, 5, 1), dtype=float32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_average_pooling_2d.html","id":null,"dir":"Reference","previous_headings":"","what":"Average pooling operation for 2D spatial data. — layer_average_pooling_2d","title":"Average pooling operation for 2D spatial data. — layer_average_pooling_2d","text":"Downsamples input along spatial dimensions (height width) taking average value input window (size defined pool_size) channel input. window shifted strides along dimension. resulting output using \"valid\" padding option spatial shape (number rows columns) : output_shape = math.floor((input_shape - pool_size) / strides) + 1 (input_shape >= pool_size) resulting output shape using \"\" padding option : output_shape = input_shape","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_average_pooling_2d.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Average pooling operation for 2D spatial data. — layer_average_pooling_2d","text":"","code":"layer_average_pooling_2d(   object,   pool_size,   strides = NULL,   padding = \"valid\",   data_format = NULL,   name = NULL,   ... )"},{"path":"https://keras3.posit.co/dev/reference/layer_average_pooling_2d.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Average pooling operation for 2D spatial data. — layer_average_pooling_2d","text":"object Object compose layer . tensor, array, sequential model. pool_size int list 2 integers, factors downscale (dim1, dim2). one integer specified, window length used dimensions. strides int list 2 integers, NULL. Strides values. NULL, default pool_size. one int specified, stride size used dimensions. padding string, either \"valid\" \"\" (case-insensitive). \"valid\" means padding. \"\" results padding evenly left/right /input output height/width dimension input. data_format string, either \"channels_last\" \"channels_first\". ordering dimensions inputs. \"channels_last\" corresponds inputs shape (batch, height, width, channels) \"channels_first\" corresponds inputs shape (batch, channels, height, width). defaults image_data_format value found Keras config file ~/.keras/keras.json. never set , \"channels_last\". name String, name object ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_average_pooling_2d.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Average pooling operation for 2D spatial data. — layer_average_pooling_2d","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_average_pooling_2d.html","id":"input-shape","dir":"Reference","previous_headings":"","what":"Input Shape","title":"Average pooling operation for 2D spatial data. — layer_average_pooling_2d","text":"data_format=\"channels_last\": 4D tensor shape (batch_size, height, width, channels). data_format=\"channels_first\": 4D tensor shape (batch_size, channels, height, width).","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_average_pooling_2d.html","id":"output-shape","dir":"Reference","previous_headings":"","what":"Output Shape","title":"Average pooling operation for 2D spatial data. — layer_average_pooling_2d","text":"data_format=\"channels_last\": 4D tensor shape (batch_size, pooled_height, pooled_width, channels). data_format=\"channels_first\": 4D tensor shape (batch_size, channels, pooled_height, pooled_width).","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_average_pooling_2d.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Average pooling operation for 2D spatial data. — layer_average_pooling_2d","text":"strides=(1, 1) padding=\"valid\":     strides=(2, 2) padding=\"valid\":     stride=(1, 1) padding=\"\":","code":"x <- op_array(1:9, \"float32\") |> op_reshape(c(1, 3, 3, 1)) output <- x |>   layer_average_pooling_2d(pool_size = c(2, 2),                            strides = c(1, 1),                            padding = \"valid\") output ## tf.Tensor( ## [[[[3.] ##    [4.]] ## ##   [[6.] ##    [7.]]]], shape=(1, 2, 2, 1), dtype=float32) x <- op_array(1:12, \"float32\") |> op_reshape(c(1, 3, 4, 1)) output <- x |>   layer_average_pooling_2d(pool_size = c(2, 2),                            strides = c(2, 2),                            padding = \"valid\") output ## tf.Tensor( ## [[[[3.5] ##    [5.5]]]], shape=(1, 1, 2, 1), dtype=float32) x <- op_array(1:9, \"float32\") |> op_reshape(c(1, 3, 3, 1)) output <- x |>   layer_average_pooling_2d(pool_size = c(2, 2),                            strides = c(1, 1),                            padding = \"same\") output ## tf.Tensor( ## [[[[3. ] ##    [4. ] ##    [4.5]] ## ##   [[6. ] ##    [7. ] ##    [7.5]] ## ##   [[7.5] ##    [8.5] ##    [9. ]]]], shape=(1, 3, 3, 1), dtype=float32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_average_pooling_3d.html","id":null,"dir":"Reference","previous_headings":"","what":"Average pooling operation for 3D data (spatial or spatio-temporal). — layer_average_pooling_3d","title":"Average pooling operation for 3D data (spatial or spatio-temporal). — layer_average_pooling_3d","text":"Downsamples input along spatial dimensions (depth, height, width) taking average value input window (size defined pool_size) channel input. window shifted strides along dimension.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_average_pooling_3d.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Average pooling operation for 3D data (spatial or spatio-temporal). — layer_average_pooling_3d","text":"","code":"layer_average_pooling_3d(   object,   pool_size,   strides = NULL,   padding = \"valid\",   data_format = NULL,   name = NULL,   ... )"},{"path":"https://keras3.posit.co/dev/reference/layer_average_pooling_3d.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Average pooling operation for 3D data (spatial or spatio-temporal). — layer_average_pooling_3d","text":"object Object compose layer . tensor, array, sequential model. pool_size int list 3 integers, factors downscale (dim1, dim2, dim3). one integer specified, window length used dimensions. strides int list 3 integers, NULL. Strides values. NULL, default pool_size. one int specified, stride size used dimensions. padding string, either \"valid\" \"\" (case-insensitive). \"valid\" means padding. \"\" results padding evenly left/right /input output height/width dimension input. data_format string, either \"channels_last\" \"channels_first\". ordering dimensions inputs. \"channels_last\" corresponds inputs shape (batch, spatial_dim1, spatial_dim2, spatial_dim3, channels) \"channels_first\" corresponds inputs shape (batch, channels, spatial_dim1, spatial_dim2, spatial_dim3). defaults image_data_format value found Keras config file ~/.keras/keras.json. never set , \"channels_last\". name String, name object ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_average_pooling_3d.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Average pooling operation for 3D data (spatial or spatio-temporal). — layer_average_pooling_3d","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_average_pooling_3d.html","id":"input-shape","dir":"Reference","previous_headings":"","what":"Input Shape","title":"Average pooling operation for 3D data (spatial or spatio-temporal). — layer_average_pooling_3d","text":"data_format=\"channels_last\": 5D tensor shape: (batch_size, spatial_dim1, spatial_dim2, spatial_dim3, channels) data_format=\"channels_first\": 5D tensor shape: (batch_size, channels, spatial_dim1, spatial_dim2, spatial_dim3)","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_average_pooling_3d.html","id":"output-shape","dir":"Reference","previous_headings":"","what":"Output Shape","title":"Average pooling operation for 3D data (spatial or spatio-temporal). — layer_average_pooling_3d","text":"data_format=\"channels_last\": 5D tensor shape: (batch_size, pooled_dim1, pooled_dim2, pooled_dim3, channels) data_format=\"channels_first\": 5D tensor shape: (batch_size, channels, pooled_dim1, pooled_dim2, pooled_dim3)","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_average_pooling_3d.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Average pooling operation for 3D data (spatial or spatio-temporal). — layer_average_pooling_3d","text":"","code":"depth <- height <- width <- 30 channels <- 3  inputs <- layer_input(shape = c(depth, height, width, channels)) outputs <- inputs |> layer_average_pooling_3d(pool_size = 3) outputs # Shape: (batch_size, 10, 10, 10, 3) ## <KerasTensor shape=(None, 10, 10, 10, 3), dtype=float32, sparse=False, ragged=False, name=keras_tensor_1>"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_batch_normalization.html","id":null,"dir":"Reference","previous_headings":"","what":"Layer that normalizes its inputs. — layer_batch_normalization","title":"Layer that normalizes its inputs. — layer_batch_normalization","text":"Batch normalization applies transformation maintains mean output close 0 output standard deviation close 1. Importantly, batch normalization works differently training inference. training (.e. using fit() calling layer/model argument training = TRUE), layer normalizes output using mean standard deviation current batch inputs. say, channel normalized, layer returns gamma * (batch - mean(batch)) / sqrt(var(batch) + epsilon) + beta, : epsilon small constant (configurable part constructor arguments) gamma learned scaling factor (initialized 1), can disabled passing scale = FALSE constructor. beta learned offset factor (initialized 0), can disabled passing center = FALSE constructor. inference (.e. using evaluate() predict() calling layer/model argument training = FALSE (default), layer normalizes output using moving average mean standard deviation batches seen training. say, returns gamma * (batch - self$moving_mean) / sqrt(self$moving_var+epsilon) + beta. self$moving_mean self$moving_var non-trainable variables updated time layer called training mode, : moving_mean = moving_mean * momentum + mean(batch) * (1 - momentum) moving_var = moving_var * momentum + var(batch) * (1 - momentum) , layer normalize inputs inference trained data similar statistics inference data. setting layer$trainable <- FALSE BatchNormalization layer: meaning setting layer$trainable <- FALSE freeze layer, .e. internal state change training: trainable weights updated fit() train_on_batch(), state updates run. Usually, necessarily mean layer run inference mode (normally controlled training argument can passed calling layer). \"Frozen state\" \"inference mode\" two separate concepts. However, case BatchNormalization layer, setting trainable <- FALSE layer means layer subsequently run inference mode (meaning use moving mean moving variance normalize current batch, rather using mean variance current batch). Note : Setting trainable model containing layers recursively set trainable value inner layers. value trainable attribute changed calling compile() model, new value take effect model compile() called .","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_batch_normalization.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Layer that normalizes its inputs. — layer_batch_normalization","text":"","code":"layer_batch_normalization(   object,   axis = -1L,   momentum = 0.99,   epsilon = 0.001,   center = TRUE,   scale = TRUE,   beta_initializer = \"zeros\",   gamma_initializer = \"ones\",   moving_mean_initializer = \"zeros\",   moving_variance_initializer = \"ones\",   beta_regularizer = NULL,   gamma_regularizer = NULL,   beta_constraint = NULL,   gamma_constraint = NULL,   synchronized = FALSE,   ... )"},{"path":"https://keras3.posit.co/dev/reference/layer_batch_normalization.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Layer that normalizes its inputs. — layer_batch_normalization","text":"object Object compose layer . tensor, array, sequential model. axis Integer, axis normalized (typically features axis). instance, Conv2D layer data_format = \"channels_first\", use axis = 2. momentum Momentum moving average. epsilon Small float added variance avoid dividing zero. center TRUE, add offset beta normalized tensor. FALSE, beta ignored. scale TRUE, multiply gamma. FALSE, gamma used. next layer linear can disabled since scaling done next layer. beta_initializer Initializer beta weight. gamma_initializer Initializer gamma weight. moving_mean_initializer Initializer moving mean. moving_variance_initializer Initializer moving variance. beta_regularizer Optional regularizer beta weight. gamma_regularizer Optional regularizer gamma weight. beta_constraint Optional constraint beta weight. gamma_constraint Optional constraint gamma weight. synchronized applicable TensorFlow backend. TRUE, synchronizes global batch statistics (mean variance) layer across devices training step distributed training strategy. FALSE, replica uses local batch statistics. ... Base layer keyword arguments (e.g. name dtype).","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_batch_normalization.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Layer that normalizes its inputs. — layer_batch_normalization","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_batch_normalization.html","id":"call-arguments","dir":"Reference","previous_headings":"","what":"Call Arguments","title":"Layer that normalizes its inputs. — layer_batch_normalization","text":"inputs: Input tensor (rank). training: R boolean indicating whether layer behave training mode inference mode. training = TRUE: layer normalize inputs using mean variance current batch inputs. training = FALSE: layer normalize inputs using mean variance moving statistics, learned training. mask: Binary tensor shape broadcastable inputs tensor, TRUE values indicating positions mean variance computed. Masked elements current inputs taken account mean variance computation training. prior unmasked element values taken account momentum expires.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_batch_normalization.html","id":"reference","dir":"Reference","previous_headings":"","what":"Reference","title":"Layer that normalizes its inputs. — layer_batch_normalization","text":"Ioffe Szegedy, 2015.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_bidirectional.html","id":null,"dir":"Reference","previous_headings":"","what":"Bidirectional wrapper for RNNs. — layer_bidirectional","title":"Bidirectional wrapper for RNNs. — layer_bidirectional","text":"Bidirectional wrapper RNNs.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_bidirectional.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Bidirectional wrapper for RNNs. — layer_bidirectional","text":"","code":"layer_bidirectional(   object,   layer,   merge_mode = \"concat\",   weights = NULL,   backward_layer = NULL,   ... )"},{"path":"https://keras3.posit.co/dev/reference/layer_bidirectional.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Bidirectional wrapper for RNNs. — layer_bidirectional","text":"object Object compose layer . tensor, array, sequential model. layer RNN instance, layer_lstm() layer_gru(). also Layer() instance meets following criteria: sequence-processing layer (accepts 3D+ inputs). go_backwards, return_sequences return_state attribute (semantics RNN class). input_spec attribute. Implement serialization via get_config() from_config(). Note recommended way create new RNN layers write custom RNN cell use layer_rnn(), instead subclassing Layer() directly. return_sequences TRUE, output masked timestep zero regardless layer's original zero_output_for_mask value. merge_mode Mode outputs forward backward RNNs combined. One {\"sum\", \"mul\", \"concat\", \"ave\", NULL}. NULL, outputs combined, returned list. Defaults \"concat\". weights see description backward_layer Optional RNN, Layer() instance used handle backwards input processing. backward_layer provided, layer instance passed layer argument used generate backward layer automatically. Note provided backward_layer layer properties matching layer argument, particular values stateful, return_states, return_sequences, etc. addition, backward_layer layer different go_backwards argument values. ValueError raised requirements met. ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_bidirectional.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Bidirectional wrapper for RNNs. — layer_bidirectional","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_bidirectional.html","id":"call-arguments","dir":"Reference","previous_headings":"","what":"Call Arguments","title":"Bidirectional wrapper for RNNs. — layer_bidirectional","text":"call arguments layer wrapped RNN layer. Beware passing initial_state argument call layer, first half list elements initial_state list passed forward RNN call last half list elements passed backward RNN call.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_bidirectional.html","id":"note","dir":"Reference","previous_headings":"","what":"Note","title":"Bidirectional wrapper for RNNs. — layer_bidirectional","text":"instantiating Bidirectional layer existing RNN layer instance reuse weights state RNN layer instance – Bidirectional layer freshly initialized weights.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_bidirectional.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Bidirectional wrapper for RNNs. — layer_bidirectional","text":"","code":"model <- keras_model_sequential(input_shape = c(5, 10)) %>%   layer_bidirectional(layer_lstm(units = 10, return_sequences = TRUE)) %>%   layer_bidirectional(layer_lstm(units = 10)) %>%   layer_dense(5, activation = \"softmax\")  model %>% compile(loss = \"categorical_crossentropy\",                   optimizer = \"rmsprop\")  # With custom backward layer forward_layer <- layer_lstm(units = 10, return_sequences = TRUE) backward_layer <- layer_lstm(units = 10, activation = \"relu\",                              return_sequences = TRUE, go_backwards = TRUE)  model <- keras_model_sequential(input_shape = c(5, 10)) %>%   bidirectional(forward_layer, backward_layer = backward_layer) %>%   layer_dense(5, activation = \"softmax\")  model %>% compile(loss = \"categorical_crossentropy\",                   optimizer = \"rmsprop\")"},{"path":"https://keras3.posit.co/dev/reference/layer_bidirectional.html","id":"states","dir":"Reference","previous_headings":"","what":"States","title":"Bidirectional wrapper for RNNs. — layer_bidirectional","text":"Bidirectional layer instance property states, can access layer$states. can also reset states using reset_state()","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_category_encoding.html","id":null,"dir":"Reference","previous_headings":"","what":"A preprocessing layer which encodes integer features. — layer_category_encoding","title":"A preprocessing layer which encodes integer features. — layer_category_encoding","text":"layer provides options condensing data categorical encoding total number tokens known advance. accepts integer values inputs, outputs dense sparse representation inputs. integer inputs total number tokens known, use layer_integer_lookup() instead. Note: layer safe use inside tf.data pipeline (independently backend using).","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_category_encoding.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"A preprocessing layer which encodes integer features. — layer_category_encoding","text":"","code":"layer_category_encoding(   object,   num_tokens = NULL,   output_mode = \"multi_hot\",   sparse = FALSE,   ... )"},{"path":"https://keras3.posit.co/dev/reference/layer_category_encoding.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"A preprocessing layer which encodes integer features. — layer_category_encoding","text":"object Object compose layer . tensor, array, sequential model. num_tokens total number tokens layer support. inputs layer must integers range 0 <= value < num_tokens, error thrown. output_mode Specification output layer. Values can \"one_hot\", \"multi_hot\" \"count\", configuring layer follows: - \"one_hot\": Encodes individual element input array num_tokens size, containing 1 element index. last dimension size 1, encode dimension. last dimension size 1, append new dimension encoded output. - \"multi_hot\": Encodes sample input single array num_tokens size, containing 1 vocabulary term present sample. Treats last dimension sample dimension, input shape (..., sample_length), output shape (..., num_tokens). - \"count\": Like \"multi_hot\", int array contains count number times token index appeared sample. output modes, currently output rank 2 supported. Defaults \"multi_hot\". sparse Whether return sparse tensor; backends support sparse tensors. ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_category_encoding.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"A preprocessing layer which encodes integer features. — layer_category_encoding","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_category_encoding.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"A preprocessing layer which encodes integer features. — layer_category_encoding","text":"One-hot encoding data     Multi-hot encoding data     Using weighted inputs \"count\" mode","code":"layer <- layer_category_encoding(num_tokens = 4, output_mode = \"one_hot\") x <- op_array(c(3, 2, 0, 1), \"int32\") layer(x) ## tf.Tensor( ## [[0. 0. 0. 1.] ##  [0. 0. 1. 0.] ##  [1. 0. 0. 0.] ##  [0. 1. 0. 0.]], shape=(4, 4), dtype=float32) layer <- layer_category_encoding(num_tokens = 4, output_mode = \"multi_hot\") x <- op_array(rbind(c(0, 1),                    c(0, 0),                    c(1, 2),                    c(3, 1)), \"int32\") layer(x) ## tf.Tensor( ## [[1. 1. 0. 0.] ##  [1. 0. 0. 0.] ##  [0. 1. 1. 0.] ##  [0. 1. 0. 1.]], shape=(4, 4), dtype=float32) layer <- layer_category_encoding(num_tokens = 4, output_mode = \"count\") count_weights <- op_array(rbind(c(.1, .2),                                c(.1, .1),                                c(.2, .3),                                c(.4, .2))) x <- op_array(rbind(c(0, 1),                    c(0, 0),                    c(1, 2),                    c(3, 1)), \"int32\") layer(x, count_weights = count_weights) #   array([[01, 02, 0. , 0. ], #          [02, 0. , 0. , 0. ], #          [0. , 02, 03, 0. ], #          [0. , 02, 0. , 04]]>"},{"path":"https://keras3.posit.co/dev/reference/layer_category_encoding.html","id":"call-arguments","dir":"Reference","previous_headings":"","what":"Call Arguments","title":"A preprocessing layer which encodes integer features. — layer_category_encoding","text":"inputs: 1D 2D tensor integer inputs. count_weights: tensor shape inputs indicating weight sample value summing count mode. used \"multi_hot\" \"one_hot\" modes.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_center_crop.html","id":null,"dir":"Reference","previous_headings":"","what":"A preprocessing layer which crops images. — layer_center_crop","title":"A preprocessing layer which crops images. — layer_center_crop","text":"layers crops central portion images target size. image smaller target size, resized cropped return largest possible window image matches target aspect ratio. Input pixel values can range (e.g. [0., 1.) [0, 255]).","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_center_crop.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"A preprocessing layer which crops images. — layer_center_crop","text":"","code":"layer_center_crop(object, height, width, data_format = NULL, ...)"},{"path":"https://keras3.posit.co/dev/reference/layer_center_crop.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"A preprocessing layer which crops images. — layer_center_crop","text":"object Object compose layer . tensor, array, sequential model. height Integer, height output shape. width Integer, width output shape. data_format string, either \"channels_last\" \"channels_first\". ordering dimensions inputs. \"channels_last\" corresponds inputs shape (batch, height, width, channels) \"channels_first\" corresponds inputs shape (batch, channels, height, width). defaults image_data_format value found Keras config file ~/.keras/keras.json. never set , \"channels_last\". ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_center_crop.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"A preprocessing layer which crops images. — layer_center_crop","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_center_crop.html","id":"input-shape","dir":"Reference","previous_headings":"","what":"Input Shape","title":"A preprocessing layer which crops images. — layer_center_crop","text":"3D (unbatched) 4D (batched) tensor shape: (..., height, width, channels), \"channels_last\" format, (..., channels, height, width), \"channels_first\" format.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_center_crop.html","id":"output-shape","dir":"Reference","previous_headings":"","what":"Output Shape","title":"A preprocessing layer which crops images. — layer_center_crop","text":"3D (unbatched) 4D (batched) tensor shape: (..., target_height, target_width, channels), (..., channels, target_height, target_width), \"channels_first\" format. input height/width even target height/width odd (inversely), input image left-padded 1 pixel. Note: layer safe use inside tf.data pipeline (independently backend using).","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_concatenate.html","id":null,"dir":"Reference","previous_headings":"","what":"Concatenates a list of inputs. — layer_concatenate","title":"Concatenates a list of inputs. — layer_concatenate","text":"takes input list tensors, shape except concatenation axis, returns single tensor concatenation inputs.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_concatenate.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Concatenates a list of inputs. — layer_concatenate","text":"","code":"layer_concatenate(inputs, ..., axis = -1L)"},{"path":"https://keras3.posit.co/dev/reference/layer_concatenate.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Concatenates a list of inputs. — layer_concatenate","text":"inputs layers combine ... Standard layer keyword arguments. axis Axis along concatenate.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_concatenate.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Concatenates a list of inputs. — layer_concatenate","text":"tensor, concatenation inputs alongside axis axis.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_concatenate.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Concatenates a list of inputs. — layer_concatenate","text":"Usage Keras model:","code":"x <- op_arange(20) |> op_reshape(c(2, 2, 5)) y <- op_arange(21, 40) |> op_reshape(c(2, 2, 5)) layer_concatenate(x, y, axis = 2) ## tf.Tensor( ## [[[ 1.  2.  3.  4.  5.] ##   [ 6.  7.  8.  9. 10.] ##   [21. 22. 23. 24. 25.] ##   [26. 27. 28. 29. 30.]] ## ##  [[11. 12. 13. 14. 15.] ##   [16. 17. 18. 19. 20.] ##   [31. 32. 33. 34. 35.] ##   [36. 37. 38. 39. 40.]]], shape=(2, 4, 5), dtype=float32) x1 <- op_arange(10)     |> op_reshape(c(5, 2)) |> layer_dense(8) x2 <- op_arange(11, 20) |> op_reshape(c(5, 2)) |> layer_dense(8) y <- layer_concatenate(x1, x2)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_conv_1d.html","id":null,"dir":"Reference","previous_headings":"","what":"1D convolution layer (e.g. temporal convolution). — layer_conv_1d","title":"1D convolution layer (e.g. temporal convolution). — layer_conv_1d","text":"layer creates convolution kernel convolved layer input single spatial (temporal) dimension produce tensor outputs. use_bias TRUE, bias vector created added outputs. Finally, activation NULL, applied outputs well.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_conv_1d.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"1D convolution layer (e.g. temporal convolution). — layer_conv_1d","text":"","code":"layer_conv_1d(   object,   filters,   kernel_size,   strides = 1L,   padding = \"valid\",   data_format = NULL,   dilation_rate = 1L,   groups = 1L,   activation = NULL,   use_bias = TRUE,   kernel_initializer = \"glorot_uniform\",   bias_initializer = \"zeros\",   kernel_regularizer = NULL,   bias_regularizer = NULL,   activity_regularizer = NULL,   kernel_constraint = NULL,   bias_constraint = NULL,   ... )"},{"path":"https://keras3.posit.co/dev/reference/layer_conv_1d.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"1D convolution layer (e.g. temporal convolution). — layer_conv_1d","text":"object Object compose layer . tensor, array, sequential model. filters int, dimension output space (number filters convolution). kernel_size int list 1 integer, specifying size convolution window. strides int list 1 integer, specifying stride length convolution. strides > 1 incompatible dilation_rate > 1. padding string, \"valid\", \"\" \"causal\"(case-insensitive). \"valid\" means padding. \"\" results padding evenly left/right /input. padding=\"\" strides=1, output size input. \"causal\" results causal (dilated) convolutions, e.g. output[t] depend ontail(input, t+1). Useful modeling temporal data model violate temporal order. See WaveNet: Generative Model Raw Audio, section2.1. data_format string, either \"channels_last\" \"channels_first\". ordering dimensions inputs. \"channels_last\" corresponds inputs shape (batch, steps, features) \"channels_first\" corresponds inputs shape (batch, features, steps). defaults image_data_format value found Keras config file ~/.keras/keras.json. never set , \"channels_last\". dilation_rate int list 1 integers, specifying dilation rate use dilated convolution. groups positive int specifying number groups input split along channel axis. group convolved separately filters // groups filters. output concatenation groups results along channel axis. Input channels filters must divisible groups. activation Activation function. NULL, activation applied. use_bias bool, TRUE, bias added output. kernel_initializer Initializer convolution kernel. NULL, default initializer (\"glorot_uniform\") used. bias_initializer Initializer bias vector. NULL, default initializer (\"zeros\") used. kernel_regularizer Optional regularizer convolution kernel. bias_regularizer Optional regularizer bias vector. activity_regularizer Optional regularizer function output. kernel_constraint Optional projection function applied kernel updated Optimizer (e.g. used implement norm constraints value constraints layer weights). function must take input unprojected variable must return projected variable (must shape). Constraints safe use asynchronous distributed training. bias_constraint Optional projection function applied bias updated Optimizer. ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_conv_1d.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"1D convolution layer (e.g. temporal convolution). — layer_conv_1d","text":"3D tensor representing activation(conv1d(inputs, kernel) + bias).","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_conv_1d.html","id":"input-shape","dir":"Reference","previous_headings":"","what":"Input Shape","title":"1D convolution layer (e.g. temporal convolution). — layer_conv_1d","text":"data_format=\"channels_last\": 3D tensor shape: (batch_shape, steps, channels) data_format=\"channels_first\": 3D tensor shape: (batch_shape, channels, steps)","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_conv_1d.html","id":"output-shape","dir":"Reference","previous_headings":"","what":"Output Shape","title":"1D convolution layer (e.g. temporal convolution). — layer_conv_1d","text":"data_format=\"channels_last\": 3D tensor shape: (batch_shape, new_steps, filters) data_format=\"channels_first\": 3D tensor shape: (batch_shape, filters, new_steps)","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_conv_1d.html","id":"raises","dir":"Reference","previous_headings":"","what":"Raises","title":"1D convolution layer (e.g. temporal convolution). — layer_conv_1d","text":"ValueError: strides > 1 dilation_rate > 1.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_conv_1d.html","id":"example","dir":"Reference","previous_headings":"","what":"Example","title":"1D convolution layer (e.g. temporal convolution). — layer_conv_1d","text":"","code":"# The inputs are 128-length vectors with 10 timesteps, and the # batch size is 4. x <- random_uniform(c(4, 10, 128)) y <- x |> layer_conv_1d(32, 3, activation='relu') shape(y) ## shape(4, 8, 32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_conv_1d_transpose.html","id":null,"dir":"Reference","previous_headings":"","what":"1D transposed convolution layer. — layer_conv_1d_transpose","title":"1D transposed convolution layer. — layer_conv_1d_transpose","text":"need transposed convolutions generally arise desire use transformation going opposite direction normal convolution, .e., something shape output convolution something shape input maintaining connectivity pattern compatible said convolution.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_conv_1d_transpose.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"1D transposed convolution layer. — layer_conv_1d_transpose","text":"","code":"layer_conv_1d_transpose(   object,   filters,   kernel_size,   strides = 1L,   padding = \"valid\",   output_padding = NULL,   data_format = NULL,   dilation_rate = 1L,   activation = NULL,   use_bias = TRUE,   kernel_initializer = \"glorot_uniform\",   bias_initializer = \"zeros\",   kernel_regularizer = NULL,   bias_regularizer = NULL,   activity_regularizer = NULL,   kernel_constraint = NULL,   bias_constraint = NULL,   ... )"},{"path":"https://keras3.posit.co/dev/reference/layer_conv_1d_transpose.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"1D transposed convolution layer. — layer_conv_1d_transpose","text":"object Object compose layer . tensor, array, sequential model. filters int, dimension output space (number filters transpose convolution). kernel_size int list 1 integer, specifying size transposed convolution window. strides int list 1 integer, specifying stride length transposed convolution. strides > 1 incompatible dilation_rate > 1. padding string, either \"valid\" \"\" (case-insensitive). \"valid\" means padding. \"\" results padding evenly left/right /input output height/width dimension input. output_padding Scalar integer. Amount padding add output length. Must less stride. NULL (default) output size inferred. data_format string, either \"channels_last\" \"channels_first\". ordering dimensions inputs. \"channels_last\" corresponds inputs shape (batch, steps, features) \"channels_first\" corresponds inputs shape (batch, features, steps). defaults image_data_format value found Keras config file ~/.keras/keras.json. never set , \"channels_last\". dilation_rate Scalar integer. Specifies dilation rate. Values 1 currently require strides = 1 rates greater 1 supported. activation Activation function. NULL, activation applied. use_bias bool, TRUE, bias added output. kernel_initializer Initializer convolution kernel. NULL, default initializer (\"glorot_uniform\") used. bias_initializer Initializer bias vector. NULL, default initializer (\"zeros\") used. kernel_regularizer Optional regularizer convolution kernel. bias_regularizer Optional regularizer bias vector. activity_regularizer Optional regularizer function output. kernel_constraint Optional projection function applied kernel updated Optimizer (e.g. used implement norm constraints value constraints layer weights). function must take input unprojected variable must return projected variable (must shape). Constraints safe use asynchronous distributed training. bias_constraint Optional projection function applied bias updated Optimizer. ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_conv_1d_transpose.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"1D transposed convolution layer. — layer_conv_1d_transpose","text":"3D tensor representing activation(conv1d_transpose(inputs, kernel) + bias).","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_conv_1d_transpose.html","id":"input-shape","dir":"Reference","previous_headings":"","what":"Input Shape","title":"1D transposed convolution layer. — layer_conv_1d_transpose","text":"data_format=\"channels_last\": 3D tensor shape: (batch_shape, steps, channels) data_format=\"channels_first\": 3D tensor shape: (batch_shape, channels, steps)","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_conv_1d_transpose.html","id":"output-shape","dir":"Reference","previous_headings":"","what":"Output Shape","title":"1D transposed convolution layer. — layer_conv_1d_transpose","text":"data_format=\"channels_last\": 3D tensor shape: (batch_shape, new_steps, filters) data_format=\"channels_first\": 3D tensor shape: (batch_shape, filters, new_steps)","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_conv_1d_transpose.html","id":"raises","dir":"Reference","previous_headings":"","what":"Raises","title":"1D transposed convolution layer. — layer_conv_1d_transpose","text":"ValueError: strides > 1 dilation_rate > 1.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_conv_1d_transpose.html","id":"references","dir":"Reference","previous_headings":"","what":"References","title":"1D transposed convolution layer. — layer_conv_1d_transpose","text":"guide convolution arithmetic deep learning Deconvolutional Networks","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_conv_1d_transpose.html","id":"example","dir":"Reference","previous_headings":"","what":"Example","title":"1D transposed convolution layer. — layer_conv_1d_transpose","text":"","code":"x <- random_uniform(c(4, 10, 128)) y <- x |> layer_conv_1d_transpose(32, 3, 2, activation='relu') shape(y) ## shape(4, 21, 32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_conv_2d.html","id":null,"dir":"Reference","previous_headings":"","what":"2D convolution layer. — layer_conv_2d","title":"2D convolution layer. — layer_conv_2d","text":"layer creates convolution kernel convolved layer input 2D spatial (temporal) dimension (height width) produce tensor outputs. use_bias TRUE, bias vector created added outputs. Finally, activation NULL, applied outputs well. Note numerical precision: general Keras operation execution results identical across backends 1e-7 precision float32, Conv2D operations may show larger variations. Due large number element-wise multiplications additions convolution operations, especially large inputs kernel sizes, accumulated floating-point differences can exceed 1e-7 threshold. variations particularly noticeable using different backends (e.g., TensorFlow vs JAX) different hardware.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_conv_2d.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"2D convolution layer. — layer_conv_2d","text":"","code":"layer_conv_2d(   object,   filters,   kernel_size,   strides = list(1L, 1L),   padding = \"valid\",   data_format = NULL,   dilation_rate = list(1L, 1L),   groups = 1L,   activation = NULL,   use_bias = TRUE,   kernel_initializer = \"glorot_uniform\",   bias_initializer = \"zeros\",   kernel_regularizer = NULL,   bias_regularizer = NULL,   activity_regularizer = NULL,   kernel_constraint = NULL,   bias_constraint = NULL,   ... )"},{"path":"https://keras3.posit.co/dev/reference/layer_conv_2d.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"2D convolution layer. — layer_conv_2d","text":"object Object compose layer . tensor, array, sequential model. filters int, dimension output space (number filters convolution). kernel_size int list 2 integer, specifying size convolution window. strides int list 2 integer, specifying stride length convolution. strides > 1 incompatible dilation_rate > 1. padding string, either \"valid\" \"\" (case-insensitive). \"valid\" means padding. \"\" results padding evenly left/right /input. padding=\"\" strides=1, output size input. data_format string, either \"channels_last\" \"channels_first\". ordering dimensions inputs. \"channels_last\" corresponds inputs shape (batch_size, height, width, channels) \"channels_first\" corresponds inputs shape (batch_size, channels, height, width). defaults image_data_format value found Keras config file ~/.keras/keras.json. never set , \"channels_last\". dilation_rate int list 2 integers, specifying dilation rate use dilated convolution. groups positive int specifying number groups input split along channel axis. group convolved separately filters // groups filters. output concatenation groups results along channel axis. Input channels filters must divisible groups. activation Activation function. NULL, activation applied. use_bias bool, TRUE, bias added output. kernel_initializer Initializer convolution kernel. NULL, default initializer (\"glorot_uniform\") used. bias_initializer Initializer bias vector. NULL, default initializer (\"zeros\") used. kernel_regularizer Optional regularizer convolution kernel. bias_regularizer Optional regularizer bias vector. activity_regularizer Optional regularizer function output. kernel_constraint Optional projection function applied kernel updated Optimizer (e.g. used implement norm constraints value constraints layer weights). function must take input unprojected variable must return projected variable (must shape). Constraints safe use asynchronous distributed training. bias_constraint Optional projection function applied bias updated Optimizer. ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_conv_2d.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"2D convolution layer. — layer_conv_2d","text":"4D tensor representing activation(conv2d(inputs, kernel) + bias).","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_conv_2d.html","id":"input-shape","dir":"Reference","previous_headings":"","what":"Input Shape","title":"2D convolution layer. — layer_conv_2d","text":"data_format=\"channels_last\": 4D tensor shape: (batch_size, height, width, channels) data_format=\"channels_first\": 4D tensor shape: (batch_size, channels, height, width)","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_conv_2d.html","id":"output-shape","dir":"Reference","previous_headings":"","what":"Output Shape","title":"2D convolution layer. — layer_conv_2d","text":"data_format=\"channels_last\": 4D tensor shape: (batch_size, new_height, new_width, filters) data_format=\"channels_first\": 4D tensor shape: (batch_size, filters, new_height, new_width)","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_conv_2d.html","id":"raises","dir":"Reference","previous_headings":"","what":"Raises","title":"2D convolution layer. — layer_conv_2d","text":"ValueError: strides > 1 dilation_rate > 1.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_conv_2d.html","id":"example","dir":"Reference","previous_headings":"","what":"Example","title":"2D convolution layer. — layer_conv_2d","text":"","code":"x <- random_uniform(c(4, 10, 10, 128)) y <- x |> layer_conv_2d(32, 3, activation='relu') shape(y) ## shape(4, 8, 8, 32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_conv_2d_transpose.html","id":null,"dir":"Reference","previous_headings":"","what":"2D transposed convolution layer. — layer_conv_2d_transpose","title":"2D transposed convolution layer. — layer_conv_2d_transpose","text":"need transposed convolutions generally arise desire use transformation going opposite direction normal convolution, .e., something shape output convolution something shape input maintaining connectivity pattern compatible said convolution.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_conv_2d_transpose.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"2D transposed convolution layer. — layer_conv_2d_transpose","text":"","code":"layer_conv_2d_transpose(   object,   filters,   kernel_size,   strides = list(1L, 1L),   padding = \"valid\",   output_padding = NULL,   data_format = NULL,   dilation_rate = list(1L, 1L),   activation = NULL,   use_bias = TRUE,   kernel_initializer = \"glorot_uniform\",   bias_initializer = \"zeros\",   kernel_regularizer = NULL,   bias_regularizer = NULL,   activity_regularizer = NULL,   kernel_constraint = NULL,   bias_constraint = NULL,   ... )"},{"path":"https://keras3.posit.co/dev/reference/layer_conv_2d_transpose.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"2D transposed convolution layer. — layer_conv_2d_transpose","text":"object Object compose layer . tensor, array, sequential model. filters int, dimension output space (number filters transposed convolution). kernel_size int list 1 integer, specifying size transposed convolution window. strides int list 1 integer, specifying stride length transposed convolution. strides > 1 incompatible dilation_rate > 1. padding string, either \"valid\" \"\" (case-insensitive). \"valid\" means padding. \"\" results padding evenly left/right /input. padding=\"\" strides=1, output size input. output_padding Scalar integer vector two integers. Amount padding add height width output tensor. element must smaller corresponding stride. NULL (default) output size inferred. data_format string, either \"channels_last\" \"channels_first\". ordering dimensions inputs. \"channels_last\" corresponds inputs shape (batch_size, height, width, channels) \"channels_first\" corresponds inputs shape (batch_size, channels, height, width). defaults image_data_format value found Keras config file ~/.keras/keras.json. never set , \"channels_last\". dilation_rate Scalar integer vector 2 integers specifying dilation rate. Values 1 require strides = 1; different rates per dimension supported. activation Activation function. NULL, activation applied. use_bias bool, TRUE, bias added output. kernel_initializer Initializer convolution kernel. NULL, default initializer (\"glorot_uniform\") used. bias_initializer Initializer bias vector. NULL, default initializer (\"zeros\") used. kernel_regularizer Optional regularizer convolution kernel. bias_regularizer Optional regularizer bias vector. activity_regularizer Optional regularizer function output. kernel_constraint Optional projection function applied kernel updated Optimizer (e.g. used implement norm constraints value constraints layer weights). function must take input unprojected variable must return projected variable (must shape). Constraints safe use asynchronous distributed training. bias_constraint Optional projection function applied bias updated Optimizer. ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_conv_2d_transpose.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"2D transposed convolution layer. — layer_conv_2d_transpose","text":"4D tensor representing activation(conv2d_transpose(inputs, kernel) + bias).","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_conv_2d_transpose.html","id":"input-shape","dir":"Reference","previous_headings":"","what":"Input Shape","title":"2D transposed convolution layer. — layer_conv_2d_transpose","text":"data_format=\"channels_last\": 4D tensor shape: (batch_size, height, width, channels) data_format=\"channels_first\": 4D tensor shape: (batch_size, channels, height, width)","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_conv_2d_transpose.html","id":"output-shape","dir":"Reference","previous_headings":"","what":"Output Shape","title":"2D transposed convolution layer. — layer_conv_2d_transpose","text":"data_format=\"channels_last\": 4D tensor shape: (batch_size, new_height, new_width, filters) data_format=\"channels_first\": 4D tensor shape: (batch_size, filters, new_height, new_width)","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_conv_2d_transpose.html","id":"raises","dir":"Reference","previous_headings":"","what":"Raises","title":"2D transposed convolution layer. — layer_conv_2d_transpose","text":"ValueError: strides > 1 dilation_rate > 1.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_conv_2d_transpose.html","id":"references","dir":"Reference","previous_headings":"","what":"References","title":"2D transposed convolution layer. — layer_conv_2d_transpose","text":"guide convolution arithmetic deep learning Deconvolutional Networks","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_conv_2d_transpose.html","id":"example","dir":"Reference","previous_headings":"","what":"Example","title":"2D transposed convolution layer. — layer_conv_2d_transpose","text":"","code":"x <- random_uniform(c(4, 10, 8, 128)) y <- x |> layer_conv_2d_transpose(32, 2, 2, activation='relu') shape(y) ## shape(4, 20, 16, 32) # (4, 20, 16, 32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_conv_3d.html","id":null,"dir":"Reference","previous_headings":"","what":"3D convolution layer. — layer_conv_3d","title":"3D convolution layer. — layer_conv_3d","text":"layer creates convolution kernel convolved layer input 3D spatial (temporal) dimension (width,height depth) produce tensor outputs. use_bias TRUE, bias vector created added outputs. Finally, activation NULL, applied outputs well.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_conv_3d.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"3D convolution layer. — layer_conv_3d","text":"","code":"layer_conv_3d(   object,   filters,   kernel_size,   strides = list(1L, 1L, 1L),   padding = \"valid\",   data_format = NULL,   dilation_rate = list(1L, 1L, 1L),   groups = 1L,   activation = NULL,   use_bias = TRUE,   kernel_initializer = \"glorot_uniform\",   bias_initializer = \"zeros\",   kernel_regularizer = NULL,   bias_regularizer = NULL,   activity_regularizer = NULL,   kernel_constraint = NULL,   bias_constraint = NULL,   ... )"},{"path":"https://keras3.posit.co/dev/reference/layer_conv_3d.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"3D convolution layer. — layer_conv_3d","text":"object Object compose layer . tensor, array, sequential model. filters int, dimension output space (number filters convolution). kernel_size int list 3 integer, specifying size convolution window. strides int list 3 integer, specifying stride length convolution. strides > 1 incompatible dilation_rate > 1. padding string, either \"valid\" \"\" (case-insensitive). \"valid\" means padding. \"\" results padding evenly left/right /input. padding=\"\" strides=1, output size input. data_format string, either \"channels_last\" \"channels_first\". ordering dimensions inputs. \"channels_last\" corresponds inputs shape (batch_size, spatial_dim1, spatial_dim2, spatial_dim3, channels) \"channels_first\" corresponds inputs shape (batch_size, channels, spatial_dim1, spatial_dim2, spatial_dim3). defaults image_data_format value found Keras config file ~/.keras/keras.json. never set , \"channels_last\". dilation_rate int vector 3 ints, specifying dilation rate use dilated convolution. groups positive int specifying number groups input split along channel axis. group convolved separately filters %/% groups filters. output concatenation groups results along channel axis. Input channels filters must divisible groups. activation Activation function. NULL, activation applied. use_bias bool, TRUE, bias added output. kernel_initializer Initializer convolution kernel. NULL, default initializer (\"glorot_uniform\") used. bias_initializer Initializer bias vector. NULL, default initializer (\"zeros\") used. kernel_regularizer Optional regularizer convolution kernel. bias_regularizer Optional regularizer bias vector. activity_regularizer Optional regularizer function output. kernel_constraint Optional projection function applied kernel updated Optimizer (e.g. used implement norm constraints value constraints layer weights). function must take input unprojected variable must return projected variable (must shape). Constraints safe use asynchronous distributed training. bias_constraint Optional projection function applied bias updated Optimizer. ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_conv_3d.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"3D convolution layer. — layer_conv_3d","text":"5D tensor representing activation(conv3d(inputs, kernel) + bias).","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_conv_3d.html","id":"input-shape","dir":"Reference","previous_headings":"","what":"Input Shape","title":"3D convolution layer. — layer_conv_3d","text":"data_format=\"channels_last\": 5D tensor shape: (batch_size, spatial_dim1, spatial_dim2, spatial_dim3, channels) data_format=\"channels_first\": 5D tensor shape: (batch_size, channels, spatial_dim1, spatial_dim2, spatial_dim3)","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_conv_3d.html","id":"output-shape","dir":"Reference","previous_headings":"","what":"Output Shape","title":"3D convolution layer. — layer_conv_3d","text":"data_format=\"channels_last\": 5D tensor shape: (batch_size, new_spatial_dim1, new_spatial_dim2, new_spatial_dim3, filters) data_format=\"channels_first\": 5D tensor shape: (batch_size, filters, new_spatial_dim1, new_spatial_dim2, new_spatial_dim3)","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_conv_3d.html","id":"raises","dir":"Reference","previous_headings":"","what":"Raises","title":"3D convolution layer. — layer_conv_3d","text":"ValueError: strides > 1 dilation_rate > 1.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_conv_3d.html","id":"example","dir":"Reference","previous_headings":"","what":"Example","title":"3D convolution layer. — layer_conv_3d","text":"","code":"x <- random_uniform(c(4, 10, 10, 10, 128)) y <- x |> layer_conv_3d(32, 3, activation = 'relu') shape(y) ## shape(4, 8, 8, 8, 32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_conv_3d_transpose.html","id":null,"dir":"Reference","previous_headings":"","what":"3D transposed convolution layer. — layer_conv_3d_transpose","title":"3D transposed convolution layer. — layer_conv_3d_transpose","text":"need transposed convolutions generally arise desire use transformation going opposite direction normal convolution, .e., something shape output convolution something shape input maintaining connectivity pattern compatible said convolution.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_conv_3d_transpose.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"3D transposed convolution layer. — layer_conv_3d_transpose","text":"","code":"layer_conv_3d_transpose(   object,   filters,   kernel_size,   strides = list(1L, 1L, 1L),   padding = \"valid\",   output_padding = NULL,   data_format = NULL,   dilation_rate = list(1L, 1L, 1L),   activation = NULL,   use_bias = TRUE,   kernel_initializer = \"glorot_uniform\",   bias_initializer = \"zeros\",   kernel_regularizer = NULL,   bias_regularizer = NULL,   activity_regularizer = NULL,   kernel_constraint = NULL,   bias_constraint = NULL,   ... )"},{"path":"https://keras3.posit.co/dev/reference/layer_conv_3d_transpose.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"3D transposed convolution layer. — layer_conv_3d_transpose","text":"object Object compose layer . tensor, array, sequential model. filters int, dimension output space (number filters transposed convolution). kernel_size int list 1 integer, specifying size transposed convolution window. strides int list 1 integer, specifying stride length transposed convolution. strides > 1 incompatible dilation_rate > 1. padding string, either \"valid\" \"\" (case-insensitive). \"valid\" means padding. \"\" results padding evenly left/right /input. padding=\"\" strides=1, output size input. output_padding Scalar integer vector three integers. Amount padding add depth, height, width output tensor. element must smaller corresponding stride. NULL (default) output size inferred. data_format string, either \"channels_last\" \"channels_first\". ordering dimensions inputs. \"channels_last\" corresponds inputs shape (batch_size, spatial_dim1, spatial_dim2, spatial_dim3, channels) \"channels_first\" corresponds inputs shape (batch_size, channels, spatial_dim1, spatial_dim2, spatial_dim3). defaults image_data_format value found Keras config file ~/.keras/keras.json. never set , \"channels_last\". dilation_rate Scalar integer vector 3 integers specifying dilation rate. Values 1 require strides = 1; different rates per dimension supported. activation Activation function. NULL, activation applied. use_bias bool, TRUE, bias added output. kernel_initializer Initializer convolution kernel. NULL, default initializer (\"glorot_uniform\") used. bias_initializer Initializer bias vector. NULL, default initializer (\"zeros\") used. kernel_regularizer Optional regularizer convolution kernel. bias_regularizer Optional regularizer bias vector. activity_regularizer Optional regularizer function output. kernel_constraint Optional projection function applied kernel updated Optimizer (e.g. used implement norm constraints value constraints layer weights). function must take input unprojected variable must return projected variable (must shape). Constraints safe use asynchronous distributed training. bias_constraint Optional projection function applied bias updated Optimizer. ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_conv_3d_transpose.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"3D transposed convolution layer. — layer_conv_3d_transpose","text":"5D tensor representing activation(conv3d(inputs, kernel) + bias).","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_conv_3d_transpose.html","id":"input-shape","dir":"Reference","previous_headings":"","what":"Input Shape","title":"3D transposed convolution layer. — layer_conv_3d_transpose","text":"data_format=\"channels_last\": 5D tensor shape: (batch_size, spatial_dim1, spatial_dim2, spatial_dim3, channels) data_format=\"channels_first\": 5D tensor shape: (batch_size, channels, spatial_dim1, spatial_dim2, spatial_dim3)","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_conv_3d_transpose.html","id":"output-shape","dir":"Reference","previous_headings":"","what":"Output Shape","title":"3D transposed convolution layer. — layer_conv_3d_transpose","text":"data_format=\"channels_last\": 5D tensor shape: (batch_size, new_spatial_dim1, new_spatial_dim2, new_spatial_dim3, filters) data_format=\"channels_first\": 5D tensor shape: (batch_size, filters, new_spatial_dim1, new_spatial_dim2, new_spatial_dim3)","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_conv_3d_transpose.html","id":"raises","dir":"Reference","previous_headings":"","what":"Raises","title":"3D transposed convolution layer. — layer_conv_3d_transpose","text":"ValueError: strides > 1 dilation_rate > 1.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_conv_3d_transpose.html","id":"references","dir":"Reference","previous_headings":"","what":"References","title":"3D transposed convolution layer. — layer_conv_3d_transpose","text":"guide convolution arithmetic deep learning Deconvolutional Networks","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_conv_3d_transpose.html","id":"example","dir":"Reference","previous_headings":"","what":"Example","title":"3D transposed convolution layer. — layer_conv_3d_transpose","text":"","code":"x <- random_uniform(c(4, 10, 8, 12, 128)) y <- x |> layer_conv_3d_transpose(32, 2, 2, activation = 'relu') shape(y) ## shape(4, 20, 16, 24, 32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_conv_lstm_1d.html","id":null,"dir":"Reference","previous_headings":"","what":"1D Convolutional LSTM. — layer_conv_lstm_1d","title":"1D Convolutional LSTM. — layer_conv_lstm_1d","text":"Similar LSTM layer, input transformations recurrent transformations convolutional.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_conv_lstm_1d.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"1D Convolutional LSTM. — layer_conv_lstm_1d","text":"","code":"layer_conv_lstm_1d(   object,   filters,   kernel_size,   strides = 1L,   padding = \"valid\",   data_format = NULL,   dilation_rate = 1L,   activation = \"tanh\",   recurrent_activation = \"sigmoid\",   use_bias = TRUE,   kernel_initializer = \"glorot_uniform\",   recurrent_initializer = \"orthogonal\",   bias_initializer = \"zeros\",   unit_forget_bias = TRUE,   kernel_regularizer = NULL,   recurrent_regularizer = NULL,   bias_regularizer = NULL,   activity_regularizer = NULL,   kernel_constraint = NULL,   recurrent_constraint = NULL,   bias_constraint = NULL,   dropout = 0,   recurrent_dropout = 0,   seed = NULL,   return_sequences = FALSE,   return_state = FALSE,   go_backwards = FALSE,   stateful = FALSE,   ...,   unroll = NULL )"},{"path":"https://keras3.posit.co/dev/reference/layer_conv_lstm_1d.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"1D Convolutional LSTM. — layer_conv_lstm_1d","text":"object Object compose layer . tensor, array, sequential model. filters int, dimension output space (number filters convolution). kernel_size int tuple/list 1 integer, specifying size convolution window. strides int tuple/list 1 integer, specifying stride length convolution. strides > 1 incompatible dilation_rate > 1. padding string, \"valid\" \"\" (case-insensitive). \"valid\" means padding. \"\" results padding evenly left/right /input output height/width dimension input. data_format string, either \"channels_last\" \"channels_first\". ordering dimensions inputs. \"channels_last\" corresponds inputs shape (batch, steps, features) \"channels_first\" corresponds inputs shape (batch, features, steps). defaults image_data_format value found Keras config file ~/.keras/keras.json. never set , \"channels_last\". dilation_rate int tuple/list 1 integers, specifying dilation rate use dilated convolution. activation Activation function use. default hyperbolic tangent activation function applied (tanh(x)). recurrent_activation Activation function use recurrent step. use_bias Boolean, whether layer uses bias vector. kernel_initializer Initializer kernel weights matrix, used linear transformation inputs. recurrent_initializer Initializer recurrent_kernel weights matrix, used linear transformation recurrent state. bias_initializer Initializer bias vector. unit_forget_bias Boolean. TRUE, add 1 bias forget gate initialization. Use combination bias_initializer=\"zeros\". recommended Jozefowicz et al., 2015 kernel_regularizer Regularizer function applied kernel weights matrix. recurrent_regularizer Regularizer function applied recurrent_kernel weights matrix. bias_regularizer Regularizer function applied bias vector. activity_regularizer Regularizer function applied . kernel_constraint Constraint function applied kernel weights matrix. recurrent_constraint Constraint function applied recurrent_kernel weights matrix. bias_constraint Constraint function applied bias vector. dropout Float 0 1. Fraction units drop linear transformation inputs. recurrent_dropout Float 0 1. Fraction units drop linear transformation recurrent state. seed Random seed dropout. return_sequences Boolean. Whether return last output output sequence, full sequence. Default: FALSE. return_state Boolean. Whether return last state addition output. Default: FALSE. go_backwards Boolean (default: FALSE). TRUE, process input sequence backwards return reversed sequence. stateful Boolean (default FALSE). TRUE, last state sample index batch used initial state sample index following batch. ... forward/backward compatability. unroll Boolean (default: FALSE). TRUE, network unrolled, else symbolic loop used. Unrolling can speed-RNN, although tends memory-intensive. Unrolling suitable short sequences.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_conv_lstm_1d.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"1D Convolutional LSTM. — layer_conv_lstm_1d","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_conv_lstm_1d.html","id":"call-arguments","dir":"Reference","previous_headings":"","what":"Call Arguments","title":"1D Convolutional LSTM. — layer_conv_lstm_1d","text":"inputs: 4D tensor. initial_state: List initial state tensors passed first call cell. mask: Binary tensor shape (samples, timesteps) indicating whether given timestep masked. training: Python boolean indicating whether layer behave training mode inference mode. relevant dropout recurrent_dropout set.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_conv_lstm_1d.html","id":"input-shape","dir":"Reference","previous_headings":"","what":"Input Shape","title":"1D Convolutional LSTM. — layer_conv_lstm_1d","text":"data_format=\"channels_first\": 4D tensor shape: (samples, time, channels, rows) data_format=\"channels_last\": 4D tensor shape: (samples, time, rows, channels)","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_conv_lstm_1d.html","id":"output-shape","dir":"Reference","previous_headings":"","what":"Output Shape","title":"1D Convolutional LSTM. — layer_conv_lstm_1d","text":"return_state: list tensors. first tensor output. remaining tensors last states, 3D tensor shape: (samples, filters, new_rows) data_format='channels_first' shape: (samples, new_rows, filters) data_format='channels_last'. rows values might changed due padding. return_sequences: 4D tensor shape: (samples, timesteps, filters, new_rows) data_format='channels_first' shape: (samples, timesteps, new_rows, filters) data_format='channels_last'. Else, 3D tensor shape: (samples, filters, new_rows) data_format='channels_first' shape: (samples, new_rows, filters) data_format='channels_last'.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_conv_lstm_1d.html","id":"references","dir":"Reference","previous_headings":"","what":"References","title":"1D Convolutional LSTM. — layer_conv_lstm_1d","text":"Shi et al., 2015 (current implementation include feedback loop cells output).","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_conv_lstm_2d.html","id":null,"dir":"Reference","previous_headings":"","what":"2D Convolutional LSTM. — layer_conv_lstm_2d","title":"2D Convolutional LSTM. — layer_conv_lstm_2d","text":"Similar LSTM layer, input transformations recurrent transformations convolutional.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_conv_lstm_2d.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"2D Convolutional LSTM. — layer_conv_lstm_2d","text":"","code":"layer_conv_lstm_2d(   object,   filters,   kernel_size,   strides = 1L,   padding = \"valid\",   data_format = NULL,   dilation_rate = 1L,   activation = \"tanh\",   recurrent_activation = \"sigmoid\",   use_bias = TRUE,   kernel_initializer = \"glorot_uniform\",   recurrent_initializer = \"orthogonal\",   bias_initializer = \"zeros\",   unit_forget_bias = TRUE,   kernel_regularizer = NULL,   recurrent_regularizer = NULL,   bias_regularizer = NULL,   activity_regularizer = NULL,   kernel_constraint = NULL,   recurrent_constraint = NULL,   bias_constraint = NULL,   dropout = 0,   recurrent_dropout = 0,   seed = NULL,   return_sequences = FALSE,   return_state = FALSE,   go_backwards = FALSE,   stateful = FALSE,   ...,   unroll = NULL )"},{"path":"https://keras3.posit.co/dev/reference/layer_conv_lstm_2d.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"2D Convolutional LSTM. — layer_conv_lstm_2d","text":"object Object compose layer . tensor, array, sequential model. filters int, dimension output space (number filters convolution). kernel_size int tuple/list 2 integers, specifying size convolution window. strides int tuple/list 2 integers, specifying stride length convolution. strides > 1 incompatible dilation_rate > 1. padding string, \"valid\" \"\" (case-insensitive). \"valid\" means padding. \"\" results padding evenly left/right /input output height/width dimension input. data_format string, either \"channels_last\" \"channels_first\". ordering dimensions inputs. \"channels_last\" corresponds inputs shape (batch, steps, features) \"channels_first\" corresponds inputs shape (batch, features, steps). defaults image_data_format value found Keras config file ~/.keras/keras.json. never set , \"channels_last\". dilation_rate int tuple/list 2 integers, specifying dilation rate use dilated convolution. activation Activation function use. default hyperbolic tangent activation function applied (tanh(x)). recurrent_activation Activation function use recurrent step. use_bias Boolean, whether layer uses bias vector. kernel_initializer Initializer kernel weights matrix, used linear transformation inputs. recurrent_initializer Initializer recurrent_kernel weights matrix, used linear transformation recurrent state. bias_initializer Initializer bias vector. unit_forget_bias Boolean. TRUE, add 1 bias forget gate initialization. Use combination bias_initializer=\"zeros\". recommended Jozefowicz et al., 2015 kernel_regularizer Regularizer function applied kernel weights matrix. recurrent_regularizer Regularizer function applied recurrent_kernel weights matrix. bias_regularizer Regularizer function applied bias vector. activity_regularizer Regularizer function applied . kernel_constraint Constraint function applied kernel weights matrix. recurrent_constraint Constraint function applied recurrent_kernel weights matrix. bias_constraint Constraint function applied bias vector. dropout Float 0 1. Fraction units drop linear transformation inputs. recurrent_dropout Float 0 1. Fraction units drop linear transformation recurrent state. seed Random seed dropout. return_sequences Boolean. Whether return last output output sequence, full sequence. Default: FALSE. return_state Boolean. Whether return last state addition output. Default: FALSE. go_backwards Boolean (default: FALSE). TRUE, process input sequence backwards return reversed sequence. stateful Boolean (default FALSE). TRUE, last state sample index batch used initial state sample index following batch. ... forward/backward compatability. unroll Boolean (default: FALSE). TRUE, network unrolled, else symbolic loop used. Unrolling can speed-RNN, although tends memory-intensive. Unrolling suitable short sequences.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_conv_lstm_2d.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"2D Convolutional LSTM. — layer_conv_lstm_2d","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_conv_lstm_2d.html","id":"call-arguments","dir":"Reference","previous_headings":"","what":"Call Arguments","title":"2D Convolutional LSTM. — layer_conv_lstm_2d","text":"inputs: 5D tensor. mask: Binary tensor shape (samples, timesteps) indicating whether given timestep masked. training: Python boolean indicating whether layer behave training mode inference mode. relevant dropout recurrent_dropout set. initial_state: List initial state tensors passed first call cell.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_conv_lstm_2d.html","id":"input-shape","dir":"Reference","previous_headings":"","what":"Input Shape","title":"2D Convolutional LSTM. — layer_conv_lstm_2d","text":"data_format='channels_first': 5D tensor shape: (samples, time, channels, rows, cols) data_format='channels_last': 5D tensor shape: (samples, time, rows, cols, channels)","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_conv_lstm_2d.html","id":"output-shape","dir":"Reference","previous_headings":"","what":"Output Shape","title":"2D Convolutional LSTM. — layer_conv_lstm_2d","text":"return_state: list tensors. first tensor output. remaining tensors last states, 4D tensor shape: (samples, filters, new_rows, new_cols) data_format='channels_first' shape: (samples, new_rows, new_cols, filters) data_format='channels_last'. rows cols values might changed due padding. return_sequences: 5D tensor shape: (samples, timesteps, filters, new_rows, new_cols) data_format='channels_first' shape: (samples, timesteps, new_rows, new_cols, filters) data_format='channels_last'. Else, 4D tensor shape: (samples, filters, new_rows, new_cols) data_format='channels_first' shape: (samples, new_rows, new_cols, filters) data_format='channels_last'.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_conv_lstm_2d.html","id":"references","dir":"Reference","previous_headings":"","what":"References","title":"2D Convolutional LSTM. — layer_conv_lstm_2d","text":"Shi et al., 2015 (current implementation include feedback loop cells output).","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_conv_lstm_3d.html","id":null,"dir":"Reference","previous_headings":"","what":"3D Convolutional LSTM. — layer_conv_lstm_3d","title":"3D Convolutional LSTM. — layer_conv_lstm_3d","text":"Similar LSTM layer, input transformations recurrent transformations convolutional.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_conv_lstm_3d.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"3D Convolutional LSTM. — layer_conv_lstm_3d","text":"","code":"layer_conv_lstm_3d(   object,   filters,   kernel_size,   strides = 1L,   padding = \"valid\",   data_format = NULL,   dilation_rate = 1L,   activation = \"tanh\",   recurrent_activation = \"sigmoid\",   use_bias = TRUE,   kernel_initializer = \"glorot_uniform\",   recurrent_initializer = \"orthogonal\",   bias_initializer = \"zeros\",   unit_forget_bias = TRUE,   kernel_regularizer = NULL,   recurrent_regularizer = NULL,   bias_regularizer = NULL,   activity_regularizer = NULL,   kernel_constraint = NULL,   recurrent_constraint = NULL,   bias_constraint = NULL,   dropout = 0,   recurrent_dropout = 0,   seed = NULL,   return_sequences = FALSE,   return_state = FALSE,   go_backwards = FALSE,   stateful = FALSE,   ...,   unroll = NULL )"},{"path":"https://keras3.posit.co/dev/reference/layer_conv_lstm_3d.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"3D Convolutional LSTM. — layer_conv_lstm_3d","text":"object Object compose layer . tensor, array, sequential model. filters int, dimension output space (number filters convolution). kernel_size int tuple/list 3 integers, specifying size convolution window. strides int tuple/list 3 integers, specifying stride length convolution. strides > 1 incompatible dilation_rate > 1. padding string, \"valid\" \"\" (case-insensitive). \"valid\" means padding. \"\" results padding evenly left/right /input output height/width dimension input. data_format string, either \"channels_last\" \"channels_first\". ordering dimensions inputs. \"channels_last\" corresponds inputs shape (batch, steps, features) \"channels_first\" corresponds inputs shape (batch, features, steps). defaults image_data_format value found Keras config file ~/.keras/keras.json. never set , \"channels_last\". dilation_rate int tuple/list 3 integers, specifying dilation rate use dilated convolution. activation Activation function use. default hyperbolic tangent activation function applied (tanh(x)). recurrent_activation Activation function use recurrent step. use_bias Boolean, whether layer uses bias vector. kernel_initializer Initializer kernel weights matrix, used linear transformation inputs. recurrent_initializer Initializer recurrent_kernel weights matrix, used linear transformation recurrent state. bias_initializer Initializer bias vector. unit_forget_bias Boolean. TRUE, add 1 bias forget gate initialization. Use combination bias_initializer=\"zeros\". recommended Jozefowicz et al., 2015 kernel_regularizer Regularizer function applied kernel weights matrix. recurrent_regularizer Regularizer function applied recurrent_kernel weights matrix. bias_regularizer Regularizer function applied bias vector. activity_regularizer Regularizer function applied . kernel_constraint Constraint function applied kernel weights matrix. recurrent_constraint Constraint function applied recurrent_kernel weights matrix. bias_constraint Constraint function applied bias vector. dropout Float 0 1. Fraction units drop linear transformation inputs. recurrent_dropout Float 0 1. Fraction units drop linear transformation recurrent state. seed Random seed dropout. return_sequences Boolean. Whether return last output output sequence, full sequence. Default: FALSE. return_state Boolean. Whether return last state addition output. Default: FALSE. go_backwards Boolean (default: FALSE). TRUE, process input sequence backwards return reversed sequence. stateful Boolean (default FALSE). TRUE, last state sample index batch used initial state sample index following batch. ... forward/backward compatability. unroll Boolean (default: FALSE). TRUE, network unrolled, else symbolic loop used. Unrolling can speed-RNN, although tends memory-intensive. Unrolling suitable short sequences.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_conv_lstm_3d.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"3D Convolutional LSTM. — layer_conv_lstm_3d","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_conv_lstm_3d.html","id":"call-arguments","dir":"Reference","previous_headings":"","what":"Call Arguments","title":"3D Convolutional LSTM. — layer_conv_lstm_3d","text":"inputs: 6D tensor. mask: Binary tensor shape (samples, timesteps) indicating whether given timestep masked. training: Python boolean indicating whether layer behave training mode inference mode. relevant dropout recurrent_dropout set. initial_state: List initial state tensors passed first call cell.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_conv_lstm_3d.html","id":"input-shape","dir":"Reference","previous_headings":"","what":"Input Shape","title":"3D Convolutional LSTM. — layer_conv_lstm_3d","text":"data_format='channels_first': 5D tensor shape: (samples, time, channels, *spatial_dims) data_format='channels_last': 5D tensor shape: (samples, time, *spatial_dims, channels)","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_conv_lstm_3d.html","id":"output-shape","dir":"Reference","previous_headings":"","what":"Output Shape","title":"3D Convolutional LSTM. — layer_conv_lstm_3d","text":"return_state: list tensors. first tensor output. remaining tensors last states, 4D tensor shape: (samples, filters, *spatial_dims) data_format='channels_first' shape: (samples, *spatial_dims, filters) data_format='channels_last'. return_sequences: 5D tensor shape: (samples, timesteps, filters, *spatial_dims) data_format='channels_first' shape: (samples, timesteps, *spatial_dims, filters) data_format='channels_last'. Else, 4D tensor shape: (samples, filters, *spatial_dims) data_format='channels_first' shape: (samples, *spatial_dims, filters) data_format='channels_last'.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_conv_lstm_3d.html","id":"references","dir":"Reference","previous_headings":"","what":"References","title":"3D Convolutional LSTM. — layer_conv_lstm_3d","text":"Shi et al., 2015 (current implementation include feedback loop cells output).","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_cropping_1d.html","id":null,"dir":"Reference","previous_headings":"","what":"Cropping layer for 1D input (e.g. temporal sequence). — layer_cropping_1d","title":"Cropping layer for 1D input (e.g. temporal sequence). — layer_cropping_1d","text":"crops along time dimension (axis 2).","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_cropping_1d.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Cropping layer for 1D input (e.g. temporal sequence). — layer_cropping_1d","text":"","code":"layer_cropping_1d(object, cropping = list(1L, 1L), ...)"},{"path":"https://keras3.posit.co/dev/reference/layer_cropping_1d.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Cropping layer for 1D input (e.g. temporal sequence). — layer_cropping_1d","text":"object Object compose layer . tensor, array, sequential model. cropping Int, list int (length 2). int: many units trimmed beginning end cropping dimension (axis 1). list 2 ints: many units trimmed beginning end cropping dimension ((left_crop, right_crop)). ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_cropping_1d.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Cropping layer for 1D input (e.g. temporal sequence). — layer_cropping_1d","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_cropping_1d.html","id":"example","dir":"Reference","previous_headings":"","what":"Example","title":"Cropping layer for 1D input (e.g. temporal sequence). — layer_cropping_1d","text":"","code":"input_shape <- c(2, 3, 2) x <- op_arange(prod(input_shape)) |> op_reshape(input_shape) x ## tf.Tensor( ## [[[ 1.  2.] ##   [ 3.  4.] ##   [ 5.  6.]] ## ##  [[ 7.  8.] ##   [ 9. 10.] ##   [11. 12.]]], shape=(2, 3, 2), dtype=float32) y <- x |> layer_cropping_1d(cropping = 1) y ## tf.Tensor( ## [[[ 3.  4.]] ## ##  [[ 9. 10.]]], shape=(2, 1, 2), dtype=float32)"},{"path":"https://keras3.posit.co/dev/reference/layer_cropping_1d.html","id":"input-shape","dir":"Reference","previous_headings":"","what":"Input Shape","title":"Cropping layer for 1D input (e.g. temporal sequence). — layer_cropping_1d","text":"3D tensor shape (batch_size, axis_to_crop, features)","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_cropping_1d.html","id":"output-shape","dir":"Reference","previous_headings":"","what":"Output Shape","title":"Cropping layer for 1D input (e.g. temporal sequence). — layer_cropping_1d","text":"3D tensor shape (batch_size, cropped_axis, features)","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_cropping_2d.html","id":null,"dir":"Reference","previous_headings":"","what":"Cropping layer for 2D input (e.g. picture). — layer_cropping_2d","title":"Cropping layer for 2D input (e.g. picture). — layer_cropping_2d","text":"crops along spatial dimensions, .e. height width.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_cropping_2d.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Cropping layer for 2D input (e.g. picture). — layer_cropping_2d","text":"","code":"layer_cropping_2d(   object,   cropping = list(list(0L, 0L), list(0L, 0L)),   data_format = NULL,   ... )"},{"path":"https://keras3.posit.co/dev/reference/layer_cropping_2d.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Cropping layer for 2D input (e.g. picture). — layer_cropping_2d","text":"object Object compose layer . tensor, array, sequential model. cropping Int, list 2 ints, list 2 lists 2 ints. int: symmetric cropping applied height width. list 2 ints: interpreted two different symmetric cropping values height width: (symmetric_height_crop, symmetric_width_crop). list 2 lists 2 ints: interpreted ((top_crop, bottom_crop), (left_crop, right_crop)). data_format string, one \"channels_last\" (default) \"channels_first\". ordering dimensions inputs. \"channels_last\" corresponds inputs shape (batch_size, height, width, channels) \"channels_first\" corresponds inputs shape (batch_size, channels, height, width). unspecified, uses image_data_format value found Keras config file ~/.keras/keras.json (exists). Defaults \"channels_last\". ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_cropping_2d.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Cropping layer for 2D input (e.g. picture). — layer_cropping_2d","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_cropping_2d.html","id":"example","dir":"Reference","previous_headings":"","what":"Example","title":"Cropping layer for 2D input (e.g. picture). — layer_cropping_2d","text":"","code":"input_shape <- c(2, 28, 28, 3) x <- op_arange(prod(input_shape), dtype ='int32') |> op_reshape(input_shape) y <- x |> layer_cropping_2d(cropping=list(c(2, 2), c(4, 4))) shape(y) ## shape(2, 24, 20, 3)"},{"path":"https://keras3.posit.co/dev/reference/layer_cropping_2d.html","id":"input-shape","dir":"Reference","previous_headings":"","what":"Input Shape","title":"Cropping layer for 2D input (e.g. picture). — layer_cropping_2d","text":"4D tensor shape: data_format \"channels_last\": (batch_size, height, width, channels) data_format \"channels_first\": (batch_size, channels, height, width)","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_cropping_2d.html","id":"output-shape","dir":"Reference","previous_headings":"","what":"Output Shape","title":"Cropping layer for 2D input (e.g. picture). — layer_cropping_2d","text":"4D tensor shape: data_format \"channels_last\": (batch_size, cropped_height, cropped_width, channels) data_format \"channels_first\": (batch_size, channels, cropped_height, cropped_width)","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_cropping_3d.html","id":null,"dir":"Reference","previous_headings":"","what":"Cropping layer for 3D data (e.g. spatial or spatio-temporal). — layer_cropping_3d","title":"Cropping layer for 3D data (e.g. spatial or spatio-temporal). — layer_cropping_3d","text":"Cropping layer 3D data (e.g. spatial spatio-temporal).","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_cropping_3d.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Cropping layer for 3D data (e.g. spatial or spatio-temporal). — layer_cropping_3d","text":"","code":"layer_cropping_3d(   object,   cropping = list(list(1L, 1L), list(1L, 1L), list(1L, 1L)),   data_format = NULL,   ... )"},{"path":"https://keras3.posit.co/dev/reference/layer_cropping_3d.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Cropping layer for 3D data (e.g. spatial or spatio-temporal). — layer_cropping_3d","text":"object Object compose layer . tensor, array, sequential model. cropping Int, list 3 ints, list 3 lists 2 ints. int: symmetric cropping applied depth, height, width. list 3 ints: interpreted three different symmetric cropping values depth, height, width: (symmetric_dim1_crop, symmetric_dim2_crop, symmetric_dim3_crop). list 3 lists 2 ints: interpreted ((left_dim1_crop, right_dim1_crop), (left_dim2_crop, right_dim2_crop), (left_dim3_crop, right_dim3_crop)). data_format string, one \"channels_last\" (default) \"channels_first\". ordering dimensions inputs. \"channels_last\" corresponds inputs shape (batch_size, spatial_dim1, spatial_dim2, spatial_dim3, channels) \"channels_first\" corresponds inputs shape (batch_size, channels, spatial_dim1, spatial_dim2, spatial_dim3). unspecified, uses image_data_format value found Keras config file ~/.keras/keras.json (exists). Defaults \"channels_last\". ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_cropping_3d.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Cropping layer for 3D data (e.g. spatial or spatio-temporal). — layer_cropping_3d","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_cropping_3d.html","id":"example","dir":"Reference","previous_headings":"","what":"Example","title":"Cropping layer for 3D data (e.g. spatial or spatio-temporal). — layer_cropping_3d","text":"","code":"input_shape <- c(2, 28, 28, 10, 3) x <- input_shape %>% { op_reshape(seq(prod(.)), .) } y <- x |> layer_cropping_3d(cropping = c(2, 4, 2)) shape(y) ## shape(2, 24, 20, 6, 3)"},{"path":"https://keras3.posit.co/dev/reference/layer_cropping_3d.html","id":"input-shape","dir":"Reference","previous_headings":"","what":"Input Shape","title":"Cropping layer for 3D data (e.g. spatial or spatio-temporal). — layer_cropping_3d","text":"5D tensor shape: data_format \"channels_last\": (batch_size, first_axis_to_crop, second_axis_to_crop, third_axis_to_crop, channels) data_format \"channels_first\": (batch_size, channels, first_axis_to_crop, second_axis_to_crop, third_axis_to_crop)","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_cropping_3d.html","id":"output-shape","dir":"Reference","previous_headings":"","what":"Output Shape","title":"Cropping layer for 3D data (e.g. spatial or spatio-temporal). — layer_cropping_3d","text":"5D tensor shape: data_format \"channels_last\": (batch_size, first_cropped_axis, second_cropped_axis, third_cropped_axis, channels) data_format \"channels_first\": (batch_size, channels, first_cropped_axis, second_cropped_axis, third_cropped_axis)","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_cut_mix.html","id":null,"dir":"Reference","previous_headings":"","what":"CutMix data augmentation technique. — layer_cut_mix","title":"CutMix data augmentation technique. — layer_cut_mix","text":"CutMix data augmentation method patches cut pasted two images dataset, labels also mixed proportionally area patches.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_cut_mix.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"CutMix data augmentation technique. — layer_cut_mix","text":"","code":"layer_cut_mix(object, factor = 1, seed = NULL, data_format = NULL, ...)"},{"path":"https://keras3.posit.co/dev/reference/layer_cut_mix.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"CutMix data augmentation technique. — layer_cut_mix","text":"object Object compose layer . tensor, array, sequential model. factor single float tuple two floats 0 1. tuple numbers passed, factor sampled two values. single float passed, value 0 passed float sampled. values define range mixing weight sampled. higher factor increases variability patch sizes, leading diverse larger mixed patches. Defaults 1. seed Integer. Used create random seed. data_format string, either \"channels_last\" \"channels_first\". ordering dimensions inputs. \"channels_last\" corresponds inputs shape (batch, height, width, channels) \"channels_first\" corresponds inputs shape (batch, channels, height, width). defaults image_data_format value found Keras config file ~/.keras/keras.json. never set , \"channels_last\". ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_cut_mix.html","id":"references","dir":"Reference","previous_headings":"","what":"References","title":"CutMix data augmentation technique. — layer_cut_mix","text":"CutMix paper.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_dense.html","id":null,"dir":"Reference","previous_headings":"","what":"Just your regular densely-connected NN layer. — layer_dense","title":"Just your regular densely-connected NN layer. — layer_dense","text":"Dense implements operation: output = activation(dot(input, kernel) + bias) activation element-wise activation function passed activation argument, kernel weights matrix created layer, bias bias vector created layer (applicable use_bias TRUE).","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_dense.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Just your regular densely-connected NN layer. — layer_dense","text":"","code":"layer_dense(   object,   units,   activation = NULL,   use_bias = TRUE,   kernel_initializer = \"glorot_uniform\",   bias_initializer = \"zeros\",   kernel_regularizer = NULL,   bias_regularizer = NULL,   activity_regularizer = NULL,   kernel_constraint = NULL,   bias_constraint = NULL,   lora_rank = NULL,   lora_alpha = NULL,   ... )"},{"path":"https://keras3.posit.co/dev/reference/layer_dense.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Just your regular densely-connected NN layer. — layer_dense","text":"object Object compose layer . tensor, array, sequential model. units Positive integer, dimensionality output space. activation Activation function use. specify anything, activation applied (ie. \"linear\" activation: (x) = x). use_bias Boolean, whether layer uses bias vector. kernel_initializer Initializer kernel weights matrix. bias_initializer Initializer bias vector. kernel_regularizer Regularizer function applied kernel weights matrix. bias_regularizer Regularizer function applied bias vector. activity_regularizer Regularizer function applied output layer (\"activation\"). kernel_constraint Constraint function applied kernel weights matrix. bias_constraint Constraint function applied bias vector. lora_rank Optional integer. set, layer's forward pass implement LoRA (Low-Rank Adaptation) provided rank. LoRA sets layer's kernel non-trainable replaces delta original kernel, obtained via multiplying two lower-rank trainable matrices. can useful reduce computation cost fine-tuning large dense layers. can also enable LoRA existing Dense layer calling layer$enable_lora(rank). lora_alpha Optional integer. Scales low-rank adaptation delta forward pass. delta scaled lora_alpha / lora_rank, letting tune LoRA adjustment strength independently lora_rank. ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_dense.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Just your regular densely-connected NN layer. — layer_dense","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_dense.html","id":"note","dir":"Reference","previous_headings":"","what":"Note","title":"Just your regular densely-connected NN layer. — layer_dense","text":"input layer rank greater 2, Dense computes dot product inputs kernel along last axis inputs axis 0 kernel (using tf.tensordot). example, input dimensions (batch_size, d0, d1), create kernel shape (d1, units), kernel operates along axis 2 input, every sub-tensor shape (1, 1, d1) (batch_size * d0 sub-tensors). output case shape (batch_size, d0, units).","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_dense.html","id":"input-shape","dir":"Reference","previous_headings":"","what":"Input Shape","title":"Just your regular densely-connected NN layer. — layer_dense","text":"N-D tensor shape: (batch_size, ..., input_dim). common situation 2D input shape (batch_size, input_dim).","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_dense.html","id":"output-shape","dir":"Reference","previous_headings":"","what":"Output Shape","title":"Just your regular densely-connected NN layer. — layer_dense","text":"N-D tensor shape: (batch_size, ..., units). instance, 2D input shape (batch_size, input_dim), output shape (batch_size, units).","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_dense.html","id":"methods","dir":"Reference","previous_headings":"","what":"Methods","title":"Just your regular densely-connected NN layer. — layer_dense","text":"","code":"enable_lora(   rank,   lora_alpha = NULL,   a_initializer = 'he_uniform',   b_initializer = 'zeros' ) quantize(mode, type_check = TRUE)"},{"path":"https://keras3.posit.co/dev/reference/layer_dense.html","id":"readonly-properties-","dir":"Reference","previous_headings":"","what":"Readonly properties:","title":"Just your regular densely-connected NN layer. — layer_dense","text":"kernel","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_depthwise_conv_1d.html","id":null,"dir":"Reference","previous_headings":"","what":"1D depthwise convolution layer. — layer_depthwise_conv_1d","title":"1D depthwise convolution layer. — layer_depthwise_conv_1d","text":"Depthwise convolution type convolution input channel convolved different kernel (called depthwise kernel). can understand depthwise convolution first step depthwise separable convolution. implemented via following steps: Split input individual channels. Convolve channel individual depthwise kernel depth_multiplier output channels. Concatenate convolved outputs along channels axis. Unlike regular 1D convolution, depthwise convolution mix information across different input channels. depth_multiplier argument determines many filters applied one input channel. , controls amount output channels generated per input channel depthwise step.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_depthwise_conv_1d.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"1D depthwise convolution layer. — layer_depthwise_conv_1d","text":"","code":"layer_depthwise_conv_1d(   object,   kernel_size,   strides = 1L,   padding = \"valid\",   depth_multiplier = 1L,   data_format = NULL,   dilation_rate = 1L,   activation = NULL,   use_bias = TRUE,   depthwise_initializer = \"glorot_uniform\",   bias_initializer = \"zeros\",   depthwise_regularizer = NULL,   bias_regularizer = NULL,   activity_regularizer = NULL,   depthwise_constraint = NULL,   bias_constraint = NULL,   ... )"},{"path":"https://keras3.posit.co/dev/reference/layer_depthwise_conv_1d.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"1D depthwise convolution layer. — layer_depthwise_conv_1d","text":"object Object compose layer . tensor, array, sequential model. kernel_size int list 1 integer, specifying size depthwise convolution window. strides int list 1 integer, specifying stride length convolution. strides > 1 incompatible dilation_rate > 1. padding string, either \"valid\" \"\" (case-insensitive). \"valid\" means padding. \"\" results padding evenly left/right /input. padding=\"\" strides=1, output size input. depth_multiplier number depthwise convolution output channels input channel. total number depthwise convolution output channels equal input_channel * depth_multiplier. data_format string, either \"channels_last\" \"channels_first\". ordering dimensions inputs. \"channels_last\" corresponds inputs shape (batch, steps, features) \"channels_first\" corresponds inputs shape (batch, features, steps). defaults image_data_format value found Keras config file ~/.keras/keras.json. never set , \"channels_last\". dilation_rate int list 1 integers, specifying dilation rate use dilated convolution. activation Activation function. NULL, activation applied. use_bias bool, TRUE, bias added output. depthwise_initializer Initializer convolution kernel. NULL, default initializer (\"glorot_uniform\") used. bias_initializer Initializer bias vector. NULL, default initializer (\"zeros\") used. depthwise_regularizer Optional regularizer convolution kernel. bias_regularizer Optional regularizer bias vector. activity_regularizer Optional regularizer function output. depthwise_constraint Optional projection function applied kernel updated Optimizer (e.g. used implement norm constraints value constraints layer weights). function must take input unprojected variable must return projected variable (must shape). Constraints safe use asynchronous distributed training. bias_constraint Optional projection function applied bias updated Optimizer. ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_depthwise_conv_1d.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"1D depthwise convolution layer. — layer_depthwise_conv_1d","text":"3D tensor representing activation(depthwise_conv1d(inputs, kernel) + bias).","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_depthwise_conv_1d.html","id":"input-shape","dir":"Reference","previous_headings":"","what":"Input Shape","title":"1D depthwise convolution layer. — layer_depthwise_conv_1d","text":"data_format=\"channels_last\": 3D tensor shape: (batch_shape, steps, channels) data_format=\"channels_first\": 3D tensor shape: (batch_shape, channels, steps)","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_depthwise_conv_1d.html","id":"output-shape","dir":"Reference","previous_headings":"","what":"Output Shape","title":"1D depthwise convolution layer. — layer_depthwise_conv_1d","text":"data_format=\"channels_last\": 3D tensor shape: (batch_shape, new_steps, channels * depth_multiplier) data_format=\"channels_first\": 3D tensor shape: (batch_shape, channels * depth_multiplier, new_steps)","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_depthwise_conv_1d.html","id":"raises","dir":"Reference","previous_headings":"","what":"Raises","title":"1D depthwise convolution layer. — layer_depthwise_conv_1d","text":"ValueError: strides > 1 dilation_rate > 1.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_depthwise_conv_1d.html","id":"example","dir":"Reference","previous_headings":"","what":"Example","title":"1D depthwise convolution layer. — layer_depthwise_conv_1d","text":"","code":"x <- random_uniform(c(4, 10, 12)) y <- x |> layer_depthwise_conv_1d(   kernel_size = 3,   depth_multiplier = 3,   activation = 'relu' ) shape(y) ## shape(4, 8, 36)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_depthwise_conv_2d.html","id":null,"dir":"Reference","previous_headings":"","what":"2D depthwise convolution layer. — layer_depthwise_conv_2d","title":"2D depthwise convolution layer. — layer_depthwise_conv_2d","text":"Depthwise convolution type convolution input channel convolved different kernel (called depthwise kernel). can understand depthwise convolution first step depthwise separable convolution. implemented via following steps: Split input individual channels. Convolve channel individual depthwise kernel depth_multiplier output channels. Concatenate convolved outputs along channels axis. Unlike regular 2D convolution, depthwise convolution mix information across different input channels. depth_multiplier argument determines many filters applied one input channel. , controls amount output channels generated per input channel depthwise step.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_depthwise_conv_2d.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"2D depthwise convolution layer. — layer_depthwise_conv_2d","text":"","code":"layer_depthwise_conv_2d(   object,   kernel_size,   strides = list(1L, 1L),   padding = \"valid\",   depth_multiplier = 1L,   data_format = NULL,   dilation_rate = list(1L, 1L),   activation = NULL,   use_bias = TRUE,   depthwise_initializer = \"glorot_uniform\",   bias_initializer = \"zeros\",   depthwise_regularizer = NULL,   bias_regularizer = NULL,   activity_regularizer = NULL,   depthwise_constraint = NULL,   bias_constraint = NULL,   ... )"},{"path":"https://keras3.posit.co/dev/reference/layer_depthwise_conv_2d.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"2D depthwise convolution layer. — layer_depthwise_conv_2d","text":"object Object compose layer . tensor, array, sequential model. kernel_size int list 2 integer, specifying size depthwise convolution window. strides int list 2 integer, specifying stride length depthwise convolution. strides > 1 incompatible dilation_rate > 1. padding string, either \"valid\" \"\" (case-insensitive). \"valid\" means padding. \"\" results padding evenly left/right /input. padding=\"\" strides=1, output size input. depth_multiplier number depthwise convolution output channels input channel. total number depthwise convolution output channels equal input_channel * depth_multiplier. data_format string, either \"channels_last\" \"channels_first\". ordering dimensions inputs. \"channels_last\" corresponds inputs shape (batch, height, width, channels) \"channels_first\" corresponds inputs shape (batch, channels, height, width). defaults image_data_format value found Keras config file ~/.keras/keras.json. never set , \"channels_last\". dilation_rate int list 2 integers, specifying dilation rate use dilated convolution. activation Activation function. NULL, activation applied. use_bias bool, TRUE, bias added output. depthwise_initializer Initializer convolution kernel. NULL, default initializer (\"glorot_uniform\") used. bias_initializer Initializer bias vector. NULL, default initializer (\"zeros\") used. depthwise_regularizer Optional regularizer convolution kernel. bias_regularizer Optional regularizer bias vector. activity_regularizer Optional regularizer function output. depthwise_constraint Optional projection function applied kernel updated Optimizer (e.g. used implement norm constraints value constraints layer weights). function must take input unprojected variable must return projected variable (must shape). Constraints safe use asynchronous distributed training. bias_constraint Optional projection function applied bias updated Optimizer. ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_depthwise_conv_2d.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"2D depthwise convolution layer. — layer_depthwise_conv_2d","text":"4D tensor representing activation(depthwise_conv2d(inputs, kernel) + bias).","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_depthwise_conv_2d.html","id":"input-shape","dir":"Reference","previous_headings":"","what":"Input Shape","title":"2D depthwise convolution layer. — layer_depthwise_conv_2d","text":"data_format=\"channels_last\": 4D tensor shape: (batch_size, height, width, channels) data_format=\"channels_first\": 4D tensor shape: (batch_size, channels, height, width)","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_depthwise_conv_2d.html","id":"output-shape","dir":"Reference","previous_headings":"","what":"Output Shape","title":"2D depthwise convolution layer. — layer_depthwise_conv_2d","text":"data_format=\"channels_last\": 4D tensor shape: (batch_size, new_height, new_width, channels * depth_multiplier) data_format=\"channels_first\": 4D tensor shape: (batch_size, channels * depth_multiplier, new_height, new_width)","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_depthwise_conv_2d.html","id":"raises","dir":"Reference","previous_headings":"","what":"Raises","title":"2D depthwise convolution layer. — layer_depthwise_conv_2d","text":"ValueError: strides > 1 dilation_rate > 1.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_depthwise_conv_2d.html","id":"example","dir":"Reference","previous_headings":"","what":"Example","title":"2D depthwise convolution layer. — layer_depthwise_conv_2d","text":"","code":"x <- random_uniform(c(4, 10, 10, 12)) y <- x |> layer_depthwise_conv_2d(kernel_size = 3, activation = 'relu') shape(y) ## shape(4, 8, 8, 12)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_discretization.html","id":null,"dir":"Reference","previous_headings":"","what":"A preprocessing layer which buckets continuous features by ranges. — layer_discretization","title":"A preprocessing layer which buckets continuous features by ranges. — layer_discretization","text":"layer place element input data one several contiguous ranges output integer index indicating range element placed . Note: layer safe use inside tf.data pipeline (independently backend using).","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_discretization.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"A preprocessing layer which buckets continuous features by ranges. — layer_discretization","text":"","code":"layer_discretization(   object,   bin_boundaries = NULL,   num_bins = NULL,   epsilon = 0.01,   output_mode = \"int\",   sparse = FALSE,   dtype = NULL,   name = NULL )"},{"path":"https://keras3.posit.co/dev/reference/layer_discretization.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"A preprocessing layer which buckets continuous features by ranges. — layer_discretization","text":"object Object compose layer . tensor, array, sequential model. bin_boundaries list bin boundaries. leftmost rightmost bins always extend -Inf Inf, bin_boundaries = c(0, 1, 2) generates bins (-Inf, 0), [0, 1), [1, 2), [2, +Inf). option set, adapt() called. num_bins integer number bins compute. option set, bin_boundaries set adapt() called learn bin boundaries. epsilon Error tolerance, typically small fraction close zero (e.g. 0.01). Higher values epsilon increase quantile approximation, hence result unequal buckets, improve performance resource consumption. output_mode Specification output layer. Values can \"int\", \"one_hot\", \"multi_hot\", \"count\" configuring layer follows: \"int\": Return discretized bin indices directly. \"one_hot\": Encodes individual element input array size num_bins, containing 1 input's bin index. last dimension size 1, encode dimension.  last dimension size 1, append new dimension encoded output. \"multi_hot\": Encodes sample input single array size num_bins, containing 1 bin index index present sample. Treats last dimension sample dimension, input shape (..., sample_length), output shape (..., num_tokens). \"count\": \"multi_hot\", int array contains count number times bin index appeared sample. Defaults \"int\". sparse Boolean. applicable \"one_hot\", \"multi_hot\", \"count\" output modes. supported TensorFlow backend. TRUE, returns SparseTensor instead dense Tensor. Defaults FALSE. dtype datatype (e.g., \"float32\"). name String, name object","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_discretization.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"A preprocessing layer which buckets continuous features by ranges. — layer_discretization","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_discretization.html","id":"input-shape","dir":"Reference","previous_headings":"","what":"Input Shape","title":"A preprocessing layer which buckets continuous features by ranges. — layer_discretization","text":"array dimension 2 higher.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_discretization.html","id":"output-shape","dir":"Reference","previous_headings":"","what":"Output Shape","title":"A preprocessing layer which buckets continuous features by ranges. — layer_discretization","text":"input shape.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_discretization.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"A preprocessing layer which buckets continuous features by ranges. — layer_discretization","text":"Discretize float values based provided buckets.     Discretize float values based number buckets compute.","code":"input <- op_array(rbind(c(-1.5, 1, 3.4, 0.5),                        c(0, 3, 1.3, 0),                        c(-.5, 0, .5, 1),                        c(1.5, 2, 2.5, 3))) output <- input |> layer_discretization(bin_boundaries = c(0, 1, 2)) output ## tf.Tensor( ## [[0 2 3 1] ##  [1 3 2 1] ##  [0 1 1 2] ##  [2 3 3 3]], shape=(4, 4), dtype=int64) layer <- layer_discretization(num_bins = 4, epsilon = 0.01) layer |> adapt(input) layer(input) ## tf.Tensor( ## [[0 2 3 1] ##  [1 3 2 1] ##  [0 1 1 2] ##  [2 3 3 3]], shape=(4, 4), dtype=int64)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_dot.html","id":null,"dir":"Reference","previous_headings":"","what":"Computes element-wise dot product of two tensors. — layer_dot","title":"Computes element-wise dot product of two tensors. — layer_dot","text":"takes list inputs size 2, axes corresponding input along dot product performed. say x y two input tensors shapes (2, 3, 5) (2, 10, 3). batch dimension size inputs, axes correspond dimensions size corresponding inputs. e.g. axes = c(1, 2), dot product x, y result tensor shape (2, 5, 10)","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_dot.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes element-wise dot product of two tensors. — layer_dot","text":"","code":"layer_dot(inputs, ..., axes, normalize = FALSE)"},{"path":"https://keras3.posit.co/dev/reference/layer_dot.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Computes element-wise dot product of two tensors. — layer_dot","text":"inputs layers combine ... Standard layer keyword arguments. axes Integer list integers, axis axes along take dot product. list, two integers corresponding desired axis first input desired axis second input, respectively. Note size two selected axes must match. normalize Whether L2-normalize samples along dot product axis taking dot product. set TRUE, output dot product cosine proximity two samples.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_dot.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Computes element-wise dot product of two tensors. — layer_dot","text":"tensor, dot product samples inputs.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_dot.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Computes element-wise dot product of two tensors. — layer_dot","text":"Usage Keras model:","code":"x <- op_reshape(0:9,   c(1, 5, 2)) y <- op_reshape(10:19, c(1, 2, 5)) layer_dot(x, y, axes=c(2, 3)) ## tf.Tensor( ## [[[260 360] ##   [320 445]]], shape=(1, 2, 2), dtype=int32) x1 <- op_reshape(0:9, c(5, 2)) |> layer_dense(8) x2 <- op_reshape(10:19, c(5, 2)) |> layer_dense(8) shape(x1) ## shape(5, 8) shape(x2) ## shape(5, 8) y <- layer_dot(x1, x2, axes=2) shape(y) ## shape(5, 1)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_dropout.html","id":null,"dir":"Reference","previous_headings":"","what":"Applies dropout to the input. — layer_dropout","title":"Applies dropout to the input. — layer_dropout","text":"Dropout layer randomly sets input units 0 frequency rate step training time, helps prevent overfitting. Inputs set 0 scaled 1 / (1 - rate) sum inputs unchanged. Note Dropout layer applies training set TRUE call(), values dropped inference. using model.fit, training appropriately set TRUE automatically. contexts, can set argument explicitly TRUE calling layer. (contrast setting trainable=FALSE Dropout layer. trainable affect layer's behavior, Dropout variables/weights can frozen training.)","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_dropout.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Applies dropout to the input. — layer_dropout","text":"","code":"layer_dropout(object, rate, noise_shape = NULL, seed = NULL, ...)"},{"path":"https://keras3.posit.co/dev/reference/layer_dropout.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Applies dropout to the input. — layer_dropout","text":"object Object compose layer . tensor, array, sequential model. rate Float 0 1. Fraction input units drop. noise_shape 1D integer tensor representing shape binary dropout mask multiplied input. instance, inputs shape (batch_size, timesteps, features) want dropout mask timesteps, can use noise_shape=(batch_size, 1, features). seed R integer use random seed. ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_dropout.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Applies dropout to the input. — layer_dropout","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_dropout.html","id":"call-arguments","dir":"Reference","previous_headings":"","what":"Call Arguments","title":"Applies dropout to the input. — layer_dropout","text":"inputs: Input tensor (rank). training: Python boolean indicating whether layer behave training mode (adding dropout) inference mode (nothing).","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_einsum_dense.html","id":null,"dir":"Reference","previous_headings":"","what":"A layer that uses einsum as the backing computation. — layer_einsum_dense","title":"A layer that uses einsum as the backing computation. — layer_einsum_dense","text":"layer can perform einsum calculations arbitrary dimensionality.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_einsum_dense.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"A layer that uses einsum as the backing computation. — layer_einsum_dense","text":"","code":"layer_einsum_dense(   object,   equation,   output_shape,   activation = NULL,   bias_axes = NULL,   kernel_initializer = \"glorot_uniform\",   bias_initializer = \"zeros\",   kernel_regularizer = NULL,   bias_regularizer = NULL,   kernel_constraint = NULL,   bias_constraint = NULL,   lora_rank = NULL,   lora_alpha = NULL,   ... )"},{"path":"https://keras3.posit.co/dev/reference/layer_einsum_dense.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"A layer that uses einsum as the backing computation. — layer_einsum_dense","text":"object Object compose layer . tensor, array, sequential model. equation equation describing einsum perform. equation must valid einsum string form ab,bc->ac, ...ab,bc->...ac, ab...,bc->ac... 'ab', 'bc', 'ac' can valid einsum axis expression sequence. output_shape expected shape output tensor (excluding batch dimension dimensions represented ellipses). can specify NA NULL dimension unknown can inferred input shape. activation Activation function use. specify anything, activation applied (, \"linear\" activation: (x) = x). bias_axes string containing output dimension(s) apply bias . character bias_axes string correspond character output portion equation string. kernel_initializer Initializer kernel weights matrix. bias_initializer Initializer bias vector. kernel_regularizer Regularizer function applied kernel weights matrix. bias_regularizer Regularizer function applied bias vector. kernel_constraint Constraint function applied kernel weights matrix. bias_constraint Constraint function applied bias vector. lora_rank Optional integer. set, layer's forward pass implement LoRA (Low-Rank Adaptation) provided rank. LoRA sets layer's kernel non-trainable replaces delta original kernel, obtained via multiplying two lower-rank trainable matrices (factorization happens last dimension). can useful reduce computation cost fine-tuning large dense layers. can also enable LoRA existing EinsumDense layer calling layer$enable_lora(rank). lora_alpha Optional integer. Scales low-rank adaptation delta forward pass. delta scaled lora_alpha / lora_rank, letting tune LoRA adjustment strength independently lora_rank. ... Base layer keyword arguments, name dtype.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_einsum_dense.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"A layer that uses einsum as the backing computation. — layer_einsum_dense","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_einsum_dense.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"A layer that uses einsum as the backing computation. — layer_einsum_dense","text":"Biased dense layer einsums example shows instantiate standard Keras dense layer using einsum operations. example equivalent layer_Dense(64, use_bias=TRUE).     Applying dense layer sequence example shows instantiate layer applies dense operation every element sequence. , output_shape two values (since two non-batch dimensions output); first dimension output_shape NA, sequence dimension b unknown shape.     Applying dense layer sequence using ellipses example shows instantiate layer applies dense operation every element sequence, uses ellipsis notation instead specifying batch sequence dimensions. using ellipsis notation specified one axis, output_shape arg single value. instantiated way, layer can handle number sequence dimensions - including case sequence dimension exists.","code":"input <- layer_input(shape = c(32)) output <- input |>   layer_einsum_dense(\"ab,bc->ac\",                      output_shape = 64,                      bias_axes = \"c\") output # shape(NA, 64) ## <KerasTensor shape=(None, 64), dtype=float32, sparse=False, ragged=False, name=keras_tensor_1> input <- layer_input(shape = c(32, 128)) output <- input |>   layer_einsum_dense(\"abc,cd->abd\",                      output_shape = c(NA, 64),                      bias_axes = \"d\") output  # shape(NA, 32, 64) ## <KerasTensor shape=(None, None, 64), dtype=float32, sparse=False, ragged=False, name=keras_tensor_3> input <- layer_input(shape = c(32, 128)) output <- input |>   layer_einsum_dense(\"...x,xy->...y\",                      output_shape = 64,                      bias_axes = \"y\")  output  # shape(NA, 32, 64) ## <KerasTensor shape=(None, 32, 64), dtype=float32, sparse=False, ragged=False, name=keras_tensor_5>"},{"path":"https://keras3.posit.co/dev/reference/layer_einsum_dense.html","id":"methods","dir":"Reference","previous_headings":"","what":"Methods","title":"A layer that uses einsum as the backing computation. — layer_einsum_dense","text":"","code":"enable_lora(   rank,   a_initializer = 'he_uniform',   b_initializer = 'zeros' ) quantize(mode, type_check = TRUE)"},{"path":"https://keras3.posit.co/dev/reference/layer_einsum_dense.html","id":"readonly-properties-","dir":"Reference","previous_headings":"","what":"Readonly properties:","title":"A layer that uses einsum as the backing computation. — layer_einsum_dense","text":"kernel","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_embedding.html","id":null,"dir":"Reference","previous_headings":"","what":"Turns nonnegative integers (indexes) into dense vectors of fixed size. — layer_embedding","title":"Turns nonnegative integers (indexes) into dense vectors of fixed size. — layer_embedding","text":"e.g. rbind(4L, 20L) \\(\\rightarrow\\) rbind(c(0.25, 0.1), c(0.6, -0.2)) layer can used nonnegative integer inputs fixed range.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_embedding.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Turns nonnegative integers (indexes) into dense vectors of fixed size. — layer_embedding","text":"","code":"layer_embedding(   object,   input_dim,   output_dim,   embeddings_initializer = \"uniform\",   embeddings_regularizer = NULL,   embeddings_constraint = NULL,   mask_zero = FALSE,   weights = NULL,   lora_rank = NULL,   lora_alpha = NULL,   ... )"},{"path":"https://keras3.posit.co/dev/reference/layer_embedding.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Turns nonnegative integers (indexes) into dense vectors of fixed size. — layer_embedding","text":"object Object compose layer . tensor, array, sequential model. input_dim Integer. Size vocabulary, .e. maximum integer index + 1. output_dim Integer. Dimension dense embedding. embeddings_initializer Initializer embeddings matrix (see keras3::initializer_*). embeddings_regularizer Regularizer function applied embeddings matrix (see keras3::regularizer_*). embeddings_constraint Constraint function applied embeddings matrix (see keras3::constraint_*). mask_zero Boolean, whether input value 0 special \"padding\" value masked . useful using recurrent layers may take variable length input. TRUE, subsequent layers model need support masking exception raised. mask_zero set TRUE, consequence, index 0 used vocabulary (input_dim equal size vocabulary + 1). weights Optional floating-point matrix size (input_dim, output_dim). initial embeddings values use. lora_rank Optional integer. set, layer's forward pass implement LoRA (Low-Rank Adaptation) provided rank. LoRA sets layer's embeddings matrix non-trainable replaces delta original matrix, obtained via multiplying two lower-rank trainable matrices. can useful reduce computation cost fine-tuning large embedding layers. can also enable LoRA existing Embedding layer instance calling layer$enable_lora(rank). lora_alpha Optional integer. Scales low-rank adaptation delta forward pass. delta scaled lora_alpha / lora_rank, letting tune LoRA adjustment strength independently lora_rank. ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_embedding.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Turns nonnegative integers (indexes) into dense vectors of fixed size. — layer_embedding","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_embedding.html","id":"example","dir":"Reference","previous_headings":"","what":"Example","title":"Turns nonnegative integers (indexes) into dense vectors of fixed size. — layer_embedding","text":"","code":"model <- keras_model_sequential() |>   layer_embedding(1000, 64)  # The model will take as input an integer matrix of size (batch,input_length), # and the largest integer (i.e. word index) in the input # should be no larger than 999 (vocabulary size). # Now model$output_shape is (NA, 10, 64), where `NA` is the batch # dimension.  input_array <- random_integer(shape = c(32, 10), minval = 0, maxval = 1000) model |> compile('rmsprop', 'mse') output_array <- model |> predict(input_array, verbose = 0) dim(output_array)    # (32, 10, 64) ## [1] 32 10 64"},{"path":"https://keras3.posit.co/dev/reference/layer_embedding.html","id":"input-shape","dir":"Reference","previous_headings":"","what":"Input Shape","title":"Turns nonnegative integers (indexes) into dense vectors of fixed size. — layer_embedding","text":"2D tensor shape: (batch_size, input_length).","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_embedding.html","id":"output-shape","dir":"Reference","previous_headings":"","what":"Output Shape","title":"Turns nonnegative integers (indexes) into dense vectors of fixed size. — layer_embedding","text":"3D tensor shape: (batch_size, input_length, output_dim).","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_embedding.html","id":"methods","dir":"Reference","previous_headings":"","what":"Methods","title":"Turns nonnegative integers (indexes) into dense vectors of fixed size. — layer_embedding","text":"","code":"enable_lora(   rank,   lora_alpha = NULL,   a_initializer = 'he_uniform',   b_initializer = 'zeros' ) compute_output_spec(...) quantize(mode, type_check = TRUE) quantized_build(input_shape, mode) quantized_call(...)"},{"path":"https://keras3.posit.co/dev/reference/layer_embedding.html","id":"readonly-properties-","dir":"Reference","previous_headings":"","what":"Readonly properties:","title":"Turns nonnegative integers (indexes) into dense vectors of fixed size. — layer_embedding","text":"embeddings","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_equalization.html","id":null,"dir":"Reference","previous_headings":"","what":"Preprocessing layer for histogram equalization on image channels. — layer_equalization","title":"Preprocessing layer for histogram equalization on image channels. — layer_equalization","text":"Histogram equalization technique adjust image intensities enhance contrast effectively spreading frequent intensity values. layer applies equalization channel-wise basis, can improve visibility details images. layer works grayscale color images, performing equalization independently color channel. inference time, equalization consistently applied. Note: layer safe use inside tf.data pipeline (independently backend using).","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_equalization.html","id":"input-shape","dir":"Reference","previous_headings":"","what":"Input Shape","title":"Preprocessing layer for histogram equalization on image channels. — layer_equalization","text":"3D (unbatched) 4D (batched) tensor shape: (..., height, width, channels), \"channels_last\" format, (..., channels, height, width), \"channels_first\" format.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_equalization.html","id":"output-shape","dir":"Reference","previous_headings":"","what":"Output Shape","title":"Preprocessing layer for histogram equalization on image channels. — layer_equalization","text":"3D (unbatched) 4D (batched) tensor shape: (..., target_height, target_width, channels), (..., channels, target_height, target_width), \"channels_first\" format.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_equalization.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Preprocessing layer for histogram equalization on image channels. — layer_equalization","text":"","code":"layer_equalization(   object,   value_range = list(0L, 255L),   bins = 256L,   data_format = NULL,   ... )"},{"path":"https://keras3.posit.co/dev/reference/layer_equalization.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Preprocessing layer for histogram equalization on image channels. — layer_equalization","text":"object Object compose layer . tensor, array, sequential model. value_range Optional list/tuple 2 floats specifying lower upper limits input data values. Defaults [0, 255]. input image scaled, use appropriate range (e.g., [0.0, 1.0]). equalization scaled range, output values clipped accordingly. bins Integer specifying number histogram bins use equalization. Defaults 256, suitable 8-bit images. Larger values can provide granular intensity redistribution. data_format String, one \"channels_last\" (default) \"channels_first\". ordering dimensions inputs. \"channels_last\" corresponds inputs shape (batch, height, width, channels) \"channels_first\" corresponds inputs shape (batch, channels, height, width). ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_equalization.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Preprocessing layer for histogram equalization on image channels. — layer_equalization","text":"","code":"# Create an equalization layer for standard 8-bit images equalizer <- layer_equalization()  # An image with uneven intensity distribution image <- np_array(...) # your input image  # Apply histogram equalization equalized_image <- equalizer(image)  # For images with custom value range custom_equalizer <- layer_equalization(     value_range=c(0.0, 1.0),  # for normalized images     bins=128  # fewer bins for more subtle equalization ) custom_equalized <- custom_equalizer(normalized_image)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_feature_space.html","id":null,"dir":"Reference","previous_headings":"","what":"One-stop utility for preprocessing and encoding structured data. — layer_feature_space","title":"One-stop utility for preprocessing and encoding structured data. — layer_feature_space","text":"Available feature types: Note features can referred string name, e.g. \"integer_categorical\". using string name, default argument values used.","code":"# Plain float values. feature_float(name = NULL)  # Float values to be preprocessed via featurewise standardization # (i.e. via a `layer_normalization()` layer). feature_float_normalized(name = NULL)  # Float values to be preprocessed via linear rescaling # (i.e. via a `layer_rescaling` layer). feature_float_rescaled(scale = 1., offset = 0., name = NULL)  # Float values to be discretized. By default, the discrete # representation will then be one-hot encoded. feature_float_discretized(   num_bins,   bin_boundaries = NULL,   output_mode = \"one_hot\",   name = NULL )  # Integer values to be indexed. By default, the discrete # representation will then be one-hot encoded. feature_integer_categorical(   max_tokens = NULL,   num_oov_indices = 1,   output_mode = \"one_hot\",   name = NULL )  # String values to be indexed. By default, the discrete # representation will then be one-hot encoded. feature_string_categorical(   max_tokens = NULL,   num_oov_indices = 1,   output_mode = \"one_hot\",   name = NULL )  # Integer values to be hashed into a fixed number of bins. # By default, the discrete representation will then be one-hot encoded. feature_integer_hashed(num_bins, output_mode = \"one_hot\", name = NULL)  # String values to be hashed into a fixed number of bins. # By default, the discrete representation will then be one-hot encoded. feature_string_hashed(num_bins, output_mode = \"one_hot\", name = NULL)"},{"path":"https://keras3.posit.co/dev/reference/layer_feature_space.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"One-stop utility for preprocessing and encoding structured data. — layer_feature_space","text":"","code":"layer_feature_space(   object,   features,   output_mode = \"concat\",   crosses = NULL,   crossing_dim = 32L,   hashing_dim = 32L,   num_discretization_bins = 32L,   name = NULL,   feature_names = NULL )  feature_cross(feature_names, crossing_dim, output_mode = \"one_hot\")  feature_custom(dtype, preprocessor, output_mode)  feature_float(name = NULL)  feature_float_rescaled(scale = 1, offset = 0, name = NULL)  feature_float_normalized(name = NULL)  feature_float_discretized(   num_bins,   bin_boundaries = NULL,   output_mode = \"one_hot\",   name = NULL )  feature_integer_categorical(   max_tokens = NULL,   num_oov_indices = 1,   output_mode = \"one_hot\",   name = NULL )  feature_string_categorical(   max_tokens = NULL,   num_oov_indices = 1,   output_mode = \"one_hot\",   name = NULL )  feature_string_hashed(num_bins, output_mode = \"one_hot\", name = NULL)  feature_integer_hashed(num_bins, output_mode = \"one_hot\", name = NULL)"},{"path":"https://keras3.posit.co/dev/reference/layer_feature_space.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"One-stop utility for preprocessing and encoding structured data. — layer_feature_space","text":"object see description features see description output_mode string. layer_feature_space(), one \"concat\" \"dict\". concat mode, features get concatenated together single vector. dict mode, FeatureSpace returns named list individually encoded features (names input list names). feature_* functions, one : \"int\" \"one_hot\" \"float\". crosses List features crossed together, e.g. crosses=list(c(\"feature_1\", \"feature_2\")). features \"crossed\" hashing combined value fixed-length vector. crossing_dim Default vector size hashing crossed features. Defaults 32. hashing_dim Default vector size hashing features type \"integer_hashed\" \"string_hashed\". Defaults 32. num_discretization_bins Default number bins used discretizing features type \"float_discretized\". Defaults 32. name String, name object feature_names Named list mapping names features type specification, e.g. list(my_feature = \"integer_categorical\") list(my_feature = feature_integer_categorical()). complete list supported types, see \"Available feature types\" paragraph . dtype string, output dtype feature. E.g., \"float32\". preprocessor callable. scale, offset Passed layer_rescaling() num_bins, bin_boundaries Passed layer_discretization() max_tokens, num_oov_indices Passed layer_integer_lookup() feature_integer_categorical() layer_string_lookup() feature_string_categorical().","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_feature_space.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"One-stop utility for preprocessing and encoding structured data. — layer_feature_space","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_feature_space.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"One-stop utility for preprocessing and encoding structured data. — layer_feature_space","text":"Basic usage named list input data:   Basic usage tf.data:   Basic usage Keras Functional API:           Customizing feature feature cross:   Returning dict (named list) integer-encoded features:   Specifying Keras preprocessing layer:   Retrieving underlying Keras preprocessing layers:   Saving reloading FeatureSpace:","code":"raw_data <- list(   float_values = c(0.0, 0.1, 0.2, 0.3),   string_values = c(\"zero\", \"one\", \"two\", \"three\"),   int_values = as.integer(c(0, 1, 2, 3)) )  dataset <- tfdatasets::tensor_slices_dataset(raw_data)  feature_space <- layer_feature_space(   features = list(     float_values = \"float_normalized\",     string_values = \"string_categorical\",     int_values = \"integer_categorical\"   ),   crosses = list(c(\"string_values\", \"int_values\")),   output_mode = \"concat\" )  # Before you start using the feature_space(), # you must `adapt()` it on some data. feature_space |> adapt(dataset)  # You can call the feature_space() on a named list of # data (batched or unbatched). output_vector <- feature_space(raw_data) library(tfdatasets) # Unlabeled data preprocessed_ds <- unlabeled_dataset |>   dataset_map(feature_space)  # Labeled data preprocessed_ds <- labeled_dataset |>   dataset_map(function(x, y) tuple(feature_space(x), y)) # Retrieve a named list of Keras layer_input() objects (inputs <- feature_space$get_inputs()) ## $float_values ## <KerasTensor shape=(None, 1), dtype=float32, sparse=False, ragged=False, name=float_values> ## ## $string_values ## <KerasTensor shape=(None, 1), dtype=string, sparse=False, ragged=False, name=string_values> ## ## $int_values ## <KerasTensor shape=(None, 1), dtype=int32, sparse=False, ragged=False, name=int_values> # Retrieve the corresponding encoded Keras tensors (encoded_features <- feature_space$get_encoded_features()) ## <KerasTensor shape=(None, 43), dtype=float32, sparse=False, ragged=False, name=keras_tensor_7> # Build a Functional model outputs <- encoded_features |> layer_dense(1, activation = \"sigmoid\") model <- keras_model(inputs, outputs) feature_space <- layer_feature_space(   features = list(     float_values = feature_float_normalized(),     string_values = feature_string_categorical(max_tokens = 10),     int_values = feature_integer_categorical(max_tokens = 10)   ),   crosses = list(     feature_cross(c(\"string_values\", \"int_values\"), crossing_dim = 32)   ),   output_mode = \"concat\" ) feature_space <- layer_feature_space(   features = list(     \"string_values\" = feature_string_categorical(output_mode = \"int\"),     \"int_values\" = feature_integer_categorical(output_mode = \"int\")   ),   crosses = list(     feature_cross(       feature_names = c(\"string_values\", \"int_values\"),       crossing_dim = 32,       output_mode = \"int\"     )   ),   output_mode = \"dict\" ) # Let's say that one of the features is a short text paragraph that # we want to encode as a vector (one vector per paragraph) via TF-IDF. data <- list(text = c(\"1st string\", \"2nd string\", \"3rd string\"))  # There's a Keras layer for this: layer_text_vectorization() custom_layer <- layer_text_vectorization(output_mode = \"tf_idf\")  # We can use feature_custom() to create a custom feature # that will use our preprocessing layer. feature_space <- layer_feature_space(   features = list(     text = feature_custom(preprocessor = custom_layer,                           dtype = \"string\",                           output_mode = \"float\"     )   ),   output_mode = \"concat\" ) feature_space |> adapt(tfdatasets::tensor_slices_dataset(data)) output_vector <- feature_space(data) # The preprocessing layer of each feature is available in `$preprocessors`. preprocessing_layer <- feature_space$preprocessors$feature1  # The crossing layer of each feature cross is available in `$crossers`. # It's an instance of layer_hashed_crossing() crossing_layer <- feature_space$crossers[[\"feature1_X_feature2\"]] feature_space$save(\"featurespace.keras\") reloaded_feature_space <- keras$models$load_model(\"featurespace.keras\")"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_flatten.html","id":null,"dir":"Reference","previous_headings":"","what":"Flattens the input. Does not affect the batch size. — layer_flatten","title":"Flattens the input. Does not affect the batch size. — layer_flatten","text":"Flattens input. affect batch size.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_flatten.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Flattens the input. Does not affect the batch size. — layer_flatten","text":"","code":"layer_flatten(object, data_format = NULL, ...)"},{"path":"https://keras3.posit.co/dev/reference/layer_flatten.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Flattens the input. Does not affect the batch size. — layer_flatten","text":"object Object compose layer . tensor, array, sequential model. data_format string, one \"channels_last\" (default) \"channels_first\". ordering dimensions inputs. \"channels_last\" corresponds inputs shape (batch, ..., channels) \"channels_first\" corresponds inputs shape (batch, channels, ...). unspecified, uses image_data_format value found Keras config file ~/.keras/keras.json (exists). Defaults \"channels_last\". ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_flatten.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Flattens the input. Does not affect the batch size. — layer_flatten","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_flatten.html","id":"note","dir":"Reference","previous_headings":"","what":"Note","title":"Flattens the input. Does not affect the batch size. — layer_flatten","text":"inputs shaped (batch) without feature axis, flattening adds extra channel dimension output shape (batch, 1).","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_flatten.html","id":"example","dir":"Reference","previous_headings":"","what":"Example","title":"Flattens the input. Does not affect the batch size. — layer_flatten","text":"","code":"x <- layer_input(shape=c(10, 64)) y <- x |> layer_flatten() shape(y) ## shape(NA, 640)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_flax_module_wrapper.html","id":null,"dir":"Reference","previous_headings":"","what":"Keras Layer that wraps a Flax module. — layer_flax_module_wrapper","title":"Keras Layer that wraps a Flax module. — layer_flax_module_wrapper","text":"layer enables use Flax components form flax.linen.Module instances within Keras using JAX backend Keras. module method use forward pass can specified via method argument __call__ default. method must take following arguments exact names: self method bound module, case default __call__, module otherwise pass module. inputs: inputs model, JAX array PyTree arrays. training (optional): argument specifying training mode inference mode, TRUE passed training mode. FlaxLayer handles non-trainable state model required RNGs automatically. Note mutable parameter flax.linen.Module.apply() set DenyList([\"params\"]), therefore making assumption variables outside \"params\" collection non-trainable weights. example shows create FlaxLayer Flax Module default __call__ method training argument:   example shows wrap module method conform required signature. allows multiple input arguments training argument different name values. additionally shows use function bound module.","code":"# keras3::use_backend(\"jax\") # py_install(\"flax\", \"r-keras\")  if(config_backend() == \"jax\" &&    reticulate::py_module_available(\"flax\")) {  flax <- import(\"flax\")  MyFlaxModule(flax$linen$Module) %py_class% {   `__call__` <- flax$linen$compact(\\(self, inputs) {     inputs |>       (flax$linen$Conv(features = 32L, kernel_size = tuple(3L, 3L)))() |>       flax$linen$relu() |>       flax$linen$avg_pool(window_shape = tuple(2L, 2L),                           strides = tuple(2L, 2L)) |>       # flatten all except batch_size axis       (\\(x) x$reshape(tuple(x$shape[[1]], -1L)))() |>       (flax$linen$Dense(features = 200L))() |>       flax$linen$relu() |>       (flax$linen$Dense(features = 10L))() |>       flax$linen$softmax()   }) }  # typical usage: input <- keras_input(c(28, 28, 3)) output <- input |>   layer_flax_module_wrapper(MyFlaxModule())  model <- keras_model(input, output)  # to instantiate the layer before composing: flax_module <- MyFlaxModule() keras_layer <- layer_flax_module_wrapper(module = flax_module)  input <- keras_input(c(28, 28, 3)) output <- input |>   keras_layer()  model <- keras_model(input, output)  } flax <- import(\"flax\")  MyFlaxModule(flax$linen$Module) \\%py_class\\% {   forward <-     flax$linen$compact(\\(self, inputs1, input2, deterministic) {       # do work ....       outputs # return     }) }  my_flax_module_wrapper <- function(module, inputs, training) {   c(input1, input2) \\%<-\\% inputs   module$forward(input1, input2,!training) }  flax_module <- MyFlaxModule() keras_layer <- layer_flax_module_wrapper(module = flax_module,                                          method = my_flax_module_wrapper)"},{"path":"https://keras3.posit.co/dev/reference/layer_flax_module_wrapper.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Keras Layer that wraps a Flax module. — layer_flax_module_wrapper","text":"","code":"layer_flax_module_wrapper(object, module, method = NULL, variables = NULL, ...)"},{"path":"https://keras3.posit.co/dev/reference/layer_flax_module_wrapper.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Keras Layer that wraps a Flax module. — layer_flax_module_wrapper","text":"object Object compose layer . tensor, array, sequential model. module instance flax.linen.Module subclass. method method call model. generally method Module. provided, __call__ method used. method can also function defined Module, case must take Module first argument. used Module.init Module.apply. Details documented method argument flax.linen.Module.apply(). variables dict (named R list) containing variables module format returned flax.linen.Module.init(). contain \"params\" key , applicable, keys collections variables non-trainable state. allows passing trained parameters learned non-trainable state controlling initialization. NULL passed, module's init function called build time initialize variables model. ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_flax_module_wrapper.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Keras Layer that wraps a Flax module. — layer_flax_module_wrapper","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_gaussian_dropout.html","id":null,"dir":"Reference","previous_headings":"","what":"Apply multiplicative 1-centered Gaussian noise. — layer_gaussian_dropout","title":"Apply multiplicative 1-centered Gaussian noise. — layer_gaussian_dropout","text":"regularization layer, active training time.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_gaussian_dropout.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Apply multiplicative 1-centered Gaussian noise. — layer_gaussian_dropout","text":"","code":"layer_gaussian_dropout(object, rate, seed = NULL, ...)"},{"path":"https://keras3.posit.co/dev/reference/layer_gaussian_dropout.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Apply multiplicative 1-centered Gaussian noise. — layer_gaussian_dropout","text":"object Object compose layer . tensor, array, sequential model. rate Float, drop probability (Dropout). multiplicative noise standard deviation sqrt(rate / (1 - rate)). seed Integer, optional random seed enable deterministic behavior. ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_gaussian_dropout.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Apply multiplicative 1-centered Gaussian noise. — layer_gaussian_dropout","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_gaussian_dropout.html","id":"call-arguments","dir":"Reference","previous_headings":"","what":"Call Arguments","title":"Apply multiplicative 1-centered Gaussian noise. — layer_gaussian_dropout","text":"inputs: Input tensor (rank). training: Python boolean indicating whether layer behave training mode (adding dropout) inference mode (nothing).","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_gaussian_noise.html","id":null,"dir":"Reference","previous_headings":"","what":"Apply additive zero-centered Gaussian noise. — layer_gaussian_noise","title":"Apply additive zero-centered Gaussian noise. — layer_gaussian_noise","text":"useful mitigate overfitting (see form random data augmentation). Gaussian Noise (GS) natural choice corruption process real valued inputs. regularization layer, active training time.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_gaussian_noise.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Apply additive zero-centered Gaussian noise. — layer_gaussian_noise","text":"","code":"layer_gaussian_noise(object, stddev, seed = NULL, ...)"},{"path":"https://keras3.posit.co/dev/reference/layer_gaussian_noise.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Apply additive zero-centered Gaussian noise. — layer_gaussian_noise","text":"object Object compose layer . tensor, array, sequential model. stddev Float, standard deviation noise distribution. seed Integer, optional random seed enable deterministic behavior. ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_gaussian_noise.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Apply additive zero-centered Gaussian noise. — layer_gaussian_noise","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_gaussian_noise.html","id":"call-arguments","dir":"Reference","previous_headings":"","what":"Call Arguments","title":"Apply additive zero-centered Gaussian noise. — layer_gaussian_noise","text":"inputs: Input tensor (rank). training: Python boolean indicating whether layer behave training mode (adding noise) inference mode (nothing).","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_global_average_pooling_1d.html","id":null,"dir":"Reference","previous_headings":"","what":"Global average pooling operation for temporal data. — layer_global_average_pooling_1d","title":"Global average pooling operation for temporal data. — layer_global_average_pooling_1d","text":"Global average pooling operation temporal data.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_global_average_pooling_1d.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Global average pooling operation for temporal data. — layer_global_average_pooling_1d","text":"","code":"layer_global_average_pooling_1d(   object,   data_format = NULL,   keepdims = FALSE,   ... )"},{"path":"https://keras3.posit.co/dev/reference/layer_global_average_pooling_1d.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Global average pooling operation for temporal data. — layer_global_average_pooling_1d","text":"object Object compose layer . tensor, array, sequential model. data_format string, either \"channels_last\" \"channels_first\". ordering dimensions inputs. \"channels_last\" corresponds inputs shape (batch, steps, features) \"channels_first\" corresponds inputs shape (batch, features, steps). defaults image_data_format value found Keras config file ~/.keras/keras.json. never set , \"channels_last\". keepdims boolean, whether keep temporal dimension . keepdims FALSE (default), rank tensor reduced spatial dimensions. keepdims TRUE, temporal dimension retained length 1. behavior tf$reduce_mean() op_mean(). ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_global_average_pooling_1d.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Global average pooling operation for temporal data. — layer_global_average_pooling_1d","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_global_average_pooling_1d.html","id":"call-arguments","dir":"Reference","previous_headings":"","what":"Call Arguments","title":"Global average pooling operation for temporal data. — layer_global_average_pooling_1d","text":"inputs: 3D tensor. mask: Binary tensor shape (batch_size, steps) indicating whether given step masked (excluded average).","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_global_average_pooling_1d.html","id":"input-shape","dir":"Reference","previous_headings":"","what":"Input Shape","title":"Global average pooling operation for temporal data. — layer_global_average_pooling_1d","text":"data_format='channels_last': 3D tensor shape: (batch_size, steps, features) data_format='channels_first': 3D tensor shape: (batch_size, features, steps)","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_global_average_pooling_1d.html","id":"output-shape","dir":"Reference","previous_headings":"","what":"Output Shape","title":"Global average pooling operation for temporal data. — layer_global_average_pooling_1d","text":"keepdims=FALSE: 2D tensor shape (batch_size, features). keepdims=TRUE: data_format=\"channels_last\": 3D tensor shape (batch_size, 1, features) data_format=\"channels_first\": 3D tensor shape (batch_size, features, 1)","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_global_average_pooling_1d.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Global average pooling operation for temporal data. — layer_global_average_pooling_1d","text":"","code":"x <- random_uniform(c(2, 3, 4)) y <- x |> layer_global_average_pooling_1d() shape(y) ## shape(2, 4)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_global_average_pooling_2d.html","id":null,"dir":"Reference","previous_headings":"","what":"Global average pooling operation for 2D data. — layer_global_average_pooling_2d","title":"Global average pooling operation for 2D data. — layer_global_average_pooling_2d","text":"Global average pooling operation 2D data.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_global_average_pooling_2d.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Global average pooling operation for 2D data. — layer_global_average_pooling_2d","text":"","code":"layer_global_average_pooling_2d(   object,   data_format = NULL,   keepdims = FALSE,   ... )"},{"path":"https://keras3.posit.co/dev/reference/layer_global_average_pooling_2d.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Global average pooling operation for 2D data. — layer_global_average_pooling_2d","text":"object Object compose layer . tensor, array, sequential model. data_format string, either \"channels_last\" \"channels_first\". ordering dimensions inputs. \"channels_last\" corresponds inputs shape (batch, height, width, channels) \"channels_first\" corresponds inputs shape (batch, features, height, weight). defaults image_data_format value found Keras config file ~/.keras/keras.json. never set , \"channels_last\". keepdims boolean, whether keep temporal dimension . keepdims FALSE (default), rank tensor reduced spatial dimensions. keepdims TRUE, spatial dimension retained length 1. behavior tf$reduce_mean() op_mean(). ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_global_average_pooling_2d.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Global average pooling operation for 2D data. — layer_global_average_pooling_2d","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_global_average_pooling_2d.html","id":"input-shape","dir":"Reference","previous_headings":"","what":"Input Shape","title":"Global average pooling operation for 2D data. — layer_global_average_pooling_2d","text":"data_format='channels_last': 4D tensor shape: (batch_size, height, width, channels) data_format='channels_first': 4D tensor shape: (batch_size, channels, height, width)","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_global_average_pooling_2d.html","id":"output-shape","dir":"Reference","previous_headings":"","what":"Output Shape","title":"Global average pooling operation for 2D data. — layer_global_average_pooling_2d","text":"keepdims=FALSE: 2D tensor shape (batch_size, channels). keepdims=TRUE: data_format=\"channels_last\": 4D tensor shape (batch_size, 1, 1, channels) data_format=\"channels_first\": 4D tensor shape (batch_size, channels, 1, 1)","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_global_average_pooling_2d.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Global average pooling operation for 2D data. — layer_global_average_pooling_2d","text":"","code":"x <- random_uniform(c(2, 4, 5, 3)) y <- x |> layer_global_average_pooling_2d() shape(y) ## shape(2, 3)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_global_average_pooling_3d.html","id":null,"dir":"Reference","previous_headings":"","what":"Global average pooling operation for 3D data. — layer_global_average_pooling_3d","title":"Global average pooling operation for 3D data. — layer_global_average_pooling_3d","text":"Global average pooling operation 3D data.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_global_average_pooling_3d.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Global average pooling operation for 3D data. — layer_global_average_pooling_3d","text":"","code":"layer_global_average_pooling_3d(   object,   data_format = NULL,   keepdims = FALSE,   ... )"},{"path":"https://keras3.posit.co/dev/reference/layer_global_average_pooling_3d.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Global average pooling operation for 3D data. — layer_global_average_pooling_3d","text":"object Object compose layer . tensor, array, sequential model. data_format string, either \"channels_last\" \"channels_first\". ordering dimensions inputs. \"channels_last\" corresponds inputs shape (batch, spatial_dim1, spatial_dim2, spatial_dim3, channels) \"channels_first\" corresponds inputs shape (batch, channels, spatial_dim1, spatial_dim2, spatial_dim3). defaults image_data_format value found Keras config file ~/.keras/keras.json. never set , \"channels_last\". keepdims boolean, whether keep temporal dimension . keepdims FALSE (default), rank tensor reduced spatial dimensions. keepdims TRUE, spatial dimension retained length 1. behavior tf$reduce_mean() op_mean(). ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_global_average_pooling_3d.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Global average pooling operation for 3D data. — layer_global_average_pooling_3d","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_global_average_pooling_3d.html","id":"input-shape","dir":"Reference","previous_headings":"","what":"Input Shape","title":"Global average pooling operation for 3D data. — layer_global_average_pooling_3d","text":"data_format='channels_last': 5D tensor shape: (batch_size, spatial_dim1, spatial_dim2, spatial_dim3, channels) data_format='channels_first': 5D tensor shape: (batch_size, channels, spatial_dim1, spatial_dim2, spatial_dim3)","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_global_average_pooling_3d.html","id":"output-shape","dir":"Reference","previous_headings":"","what":"Output Shape","title":"Global average pooling operation for 3D data. — layer_global_average_pooling_3d","text":"keepdims=FALSE: 2D tensor shape (batch_size, channels). keepdims=TRUE: data_format=\"channels_last\": 5D tensor shape (batch_size, 1, 1, 1, channels) data_format=\"channels_first\": 5D tensor shape (batch_size, channels, 1, 1, 1)","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_global_average_pooling_3d.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Global average pooling operation for 3D data. — layer_global_average_pooling_3d","text":"","code":"x <- random_uniform(c(2, 4, 5, 4, 3)) y <- x |> layer_global_average_pooling_3d() shape(y) ## shape(2, 3)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_global_max_pooling_1d.html","id":null,"dir":"Reference","previous_headings":"","what":"Global max pooling operation for temporal data. — layer_global_max_pooling_1d","title":"Global max pooling operation for temporal data. — layer_global_max_pooling_1d","text":"Global max pooling operation temporal data.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_global_max_pooling_1d.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Global max pooling operation for temporal data. — layer_global_max_pooling_1d","text":"","code":"layer_global_max_pooling_1d(object, data_format = NULL, keepdims = FALSE, ...)"},{"path":"https://keras3.posit.co/dev/reference/layer_global_max_pooling_1d.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Global max pooling operation for temporal data. — layer_global_max_pooling_1d","text":"object Object compose layer . tensor, array, sequential model. data_format string, either \"channels_last\" \"channels_first\". ordering dimensions inputs. \"channels_last\" corresponds inputs shape (batch, steps, features) \"channels_first\" corresponds inputs shape (batch, features, steps). defaults image_data_format value found Keras config file ~/.keras/keras.json. never set , \"channels_last\". keepdims boolean, whether keep temporal dimension . keepdims FALSE (default), rank tensor reduced spatial dimensions. keepdims TRUE, temporal dimension retained length 1. behavior tf$reduce_mean() op_mean(). ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_global_max_pooling_1d.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Global max pooling operation for temporal data. — layer_global_max_pooling_1d","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_global_max_pooling_1d.html","id":"input-shape","dir":"Reference","previous_headings":"","what":"Input Shape","title":"Global max pooling operation for temporal data. — layer_global_max_pooling_1d","text":"data_format='channels_last': 3D tensor shape: (batch_size, steps, features) data_format='channels_first': 3D tensor shape: (batch_size, features, steps)","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_global_max_pooling_1d.html","id":"output-shape","dir":"Reference","previous_headings":"","what":"Output Shape","title":"Global max pooling operation for temporal data. — layer_global_max_pooling_1d","text":"keepdims=FALSE: 2D tensor shape (batch_size, features). keepdims=TRUE: data_format=\"channels_last\": 3D tensor shape (batch_size, 1, features) data_format=\"channels_first\": 3D tensor shape (batch_size, features, 1)","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_global_max_pooling_1d.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Global max pooling operation for temporal data. — layer_global_max_pooling_1d","text":"","code":"x <- random_uniform(c(2, 3, 4)) y <- x |> layer_global_max_pooling_1d() shape(y) ## shape(2, 4)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_global_max_pooling_2d.html","id":null,"dir":"Reference","previous_headings":"","what":"Global max pooling operation for 2D data. — layer_global_max_pooling_2d","title":"Global max pooling operation for 2D data. — layer_global_max_pooling_2d","text":"Global max pooling operation 2D data.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_global_max_pooling_2d.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Global max pooling operation for 2D data. — layer_global_max_pooling_2d","text":"","code":"layer_global_max_pooling_2d(object, data_format = NULL, keepdims = FALSE, ...)"},{"path":"https://keras3.posit.co/dev/reference/layer_global_max_pooling_2d.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Global max pooling operation for 2D data. — layer_global_max_pooling_2d","text":"object Object compose layer . tensor, array, sequential model. data_format string, either \"channels_last\" \"channels_first\". ordering dimensions inputs. \"channels_last\" corresponds inputs shape (batch, height, width, channels) \"channels_first\" corresponds inputs shape (batch, features, height, weight). defaults image_data_format value found Keras config file ~/.keras/keras.json. never set , \"channels_last\". keepdims boolean, whether keep temporal dimension . keepdims FALSE (default), rank tensor reduced spatial dimensions. keepdims TRUE, spatial dimension retained length 1. behavior tf$reduce_mean() op_mean(). ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_global_max_pooling_2d.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Global max pooling operation for 2D data. — layer_global_max_pooling_2d","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_global_max_pooling_2d.html","id":"input-shape","dir":"Reference","previous_headings":"","what":"Input Shape","title":"Global max pooling operation for 2D data. — layer_global_max_pooling_2d","text":"data_format='channels_last': 4D tensor shape: (batch_size, height, width, channels) data_format='channels_first': 4D tensor shape: (batch_size, channels, height, width)","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_global_max_pooling_2d.html","id":"output-shape","dir":"Reference","previous_headings":"","what":"Output Shape","title":"Global max pooling operation for 2D data. — layer_global_max_pooling_2d","text":"keepdims=FALSE: 2D tensor shape (batch_size, channels). keepdims=TRUE: data_format=\"channels_last\": 4D tensor shape (batch_size, 1, 1, channels) data_format=\"channels_first\": 4D tensor shape (batch_size, channels, 1, 1)","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_global_max_pooling_2d.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Global max pooling operation for 2D data. — layer_global_max_pooling_2d","text":"","code":"x <- random_uniform(c(2, 4, 5, 3)) y <- x |> layer_global_max_pooling_2d() shape(y) ## shape(2, 3)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_global_max_pooling_3d.html","id":null,"dir":"Reference","previous_headings":"","what":"Global max pooling operation for 3D data. — layer_global_max_pooling_3d","title":"Global max pooling operation for 3D data. — layer_global_max_pooling_3d","text":"Global max pooling operation 3D data.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_global_max_pooling_3d.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Global max pooling operation for 3D data. — layer_global_max_pooling_3d","text":"","code":"layer_global_max_pooling_3d(object, data_format = NULL, keepdims = FALSE, ...)"},{"path":"https://keras3.posit.co/dev/reference/layer_global_max_pooling_3d.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Global max pooling operation for 3D data. — layer_global_max_pooling_3d","text":"object Object compose layer . tensor, array, sequential model. data_format string, either \"channels_last\" \"channels_first\". ordering dimensions inputs. \"channels_last\" corresponds inputs shape (batch, spatial_dim1, spatial_dim2, spatial_dim3, channels) \"channels_first\" corresponds inputs shape (batch, channels, spatial_dim1, spatial_dim2, spatial_dim3). defaults image_data_format value found Keras config file ~/.keras/keras.json. never set , \"channels_last\". keepdims boolean, whether keep temporal dimension . keepdims FALSE (default), rank tensor reduced spatial dimensions. keepdims TRUE, spatial dimension retained length 1. behavior tf$reduce_mean() op_mean(). ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_global_max_pooling_3d.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Global max pooling operation for 3D data. — layer_global_max_pooling_3d","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_global_max_pooling_3d.html","id":"input-shape","dir":"Reference","previous_headings":"","what":"Input Shape","title":"Global max pooling operation for 3D data. — layer_global_max_pooling_3d","text":"data_format='channels_last': 5D tensor shape: (batch_size, spatial_dim1, spatial_dim2, spatial_dim3, channels) data_format='channels_first': 5D tensor shape: (batch_size, channels, spatial_dim1, spatial_dim2, spatial_dim3)","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_global_max_pooling_3d.html","id":"output-shape","dir":"Reference","previous_headings":"","what":"Output Shape","title":"Global max pooling operation for 3D data. — layer_global_max_pooling_3d","text":"keepdims=FALSE: 2D tensor shape (batch_size, channels). keepdims=TRUE: data_format=\"channels_last\": 5D tensor shape (batch_size, 1, 1, 1, channels) data_format=\"channels_first\": 5D tensor shape (batch_size, channels, 1, 1, 1)","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_global_max_pooling_3d.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Global max pooling operation for 3D data. — layer_global_max_pooling_3d","text":"","code":"x <- random_uniform(c(2, 4, 5, 4, 3)) y <- x |> layer_global_max_pooling_3d() shape(y) ## shape(2, 3)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_group_normalization.html","id":null,"dir":"Reference","previous_headings":"","what":"Group normalization layer. — layer_group_normalization","title":"Group normalization layer. — layer_group_normalization","text":"Group Normalization divides channels groups computes within group mean variance normalization. Empirically, accuracy stable batch norm wide range small batch sizes, learning rate adjusted linearly batch sizes. Relation Layer Normalization: number groups set 1, operation becomes nearly identical Layer Normalization (see Layer Normalization docs details). Relation Instance Normalization: number groups set input dimension (number groups equal number channels), operation becomes identical Instance Normalization. can achieve via groups=-1.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_group_normalization.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Group normalization layer. — layer_group_normalization","text":"","code":"layer_group_normalization(   object,   groups = 32L,   axis = -1L,   epsilon = 0.001,   center = TRUE,   scale = TRUE,   beta_initializer = \"zeros\",   gamma_initializer = \"ones\",   beta_regularizer = NULL,   gamma_regularizer = NULL,   beta_constraint = NULL,   gamma_constraint = NULL,   ... )"},{"path":"https://keras3.posit.co/dev/reference/layer_group_normalization.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Group normalization layer. — layer_group_normalization","text":"object Object compose layer . tensor, array, sequential model. groups Integer, number groups Group Normalization. Can range [1, N] N input dimension. input dimension must divisible number groups. Defaults 32. axis Integer List/Tuple. axis axes normalize across. Typically, features axis/axes. left-axes typically batch axis/axes. -1 last dimension input. Defaults -1. epsilon Small float added variance avoid dividing zero. Defaults 1e-3. center TRUE, add offset beta normalized tensor. FALSE, beta ignored. Defaults TRUE. scale TRUE, multiply gamma. FALSE, gamma used. next layer linear (also e.g. relu), can disabled since scaling done next layer. Defaults TRUE. beta_initializer Initializer beta weight. Defaults zeros. gamma_initializer Initializer gamma weight. Defaults ones. beta_regularizer Optional regularizer beta weight. NULL default. gamma_regularizer Optional regularizer gamma weight. NULL default. beta_constraint Optional constraint beta weight. NULL default. gamma_constraint Optional constraint gamma weight. NULL default. ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_group_normalization.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Group normalization layer. — layer_group_normalization","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_group_normalization.html","id":"input-shape","dir":"Reference","previous_headings":"","what":"Input Shape","title":"Group normalization layer. — layer_group_normalization","text":"Arbitrary. Use keyword argument input_shape (tuple integers, include samples axis) using layer first layer model.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_group_normalization.html","id":"output-shape","dir":"Reference","previous_headings":"","what":"Output Shape","title":"Group normalization layer. — layer_group_normalization","text":"shape input. **kwargs: Base layer keyword arguments (e.g. name dtype).","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_group_normalization.html","id":"reference","dir":"Reference","previous_headings":"","what":"Reference","title":"Group normalization layer. — layer_group_normalization","text":"Yuxin Wu & Kaiming , 2018","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_group_query_attention.html","id":null,"dir":"Reference","previous_headings":"","what":"Grouped Query Attention layer. — layer_group_query_attention","title":"Grouped Query Attention layer. — layer_group_query_attention","text":"implementation grouped-query attention introduced Ainslie et al., 2023. num_key_value_heads denotes number groups, setting num_key_value_heads 1 equivalent multi-query attention, num_key_value_heads equal num_query_heads equivalent multi-head attention. layer first projects query, key, value tensors. , key value repeated match number heads query. , query scaled dot-producted key tensors. softmaxed obtain attention probabilities. value tensors interpolated probabilities concatenated back single tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_group_query_attention.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Grouped Query Attention layer. — layer_group_query_attention","text":"","code":"layer_group_query_attention(   object,   head_dim,   num_query_heads,   num_key_value_heads,   dropout = 0,   use_bias = TRUE,   flash_attention = NULL,   kernel_initializer = \"glorot_uniform\",   bias_initializer = \"zeros\",   kernel_regularizer = NULL,   bias_regularizer = NULL,   activity_regularizer = NULL,   kernel_constraint = NULL,   bias_constraint = NULL,   seed = NULL,   ... )"},{"path":"https://keras3.posit.co/dev/reference/layer_group_query_attention.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Grouped Query Attention layer. — layer_group_query_attention","text":"object Object compose layer . tensor, array, sequential model. head_dim Size attention head. num_query_heads Number query attention heads. num_key_value_heads Number key value attention heads. dropout Dropout probability. use_bias Boolean, whether dense layers use bias vectors/matrices. flash_attention NULL, layer attempts use flash attention faster memory-efficient attention computations possible. behavior can configured using config_enable_flash_attention() config_disable_flash_attention(). kernel_initializer Initializer dense layer kernels. bias_initializer Initializer dense layer biases. kernel_regularizer Regularizer dense layer kernels. bias_regularizer Regularizer dense layer biases. activity_regularizer Regularizer dense layer activity. kernel_constraint Constraint dense layer kernels. bias_constraint Constraint dense layer kernels. seed Optional integer seed dropout layer. ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_group_query_attention.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Grouped Query Attention layer. — layer_group_query_attention","text":"attention_output: Result computation, shape (batch_dim, target_seq_len, feature_dim), target_seq_len target sequence length feature_dim query input last dim. attention_scores: (Optional) attention coefficients shape (batch_dim, num_query_heads, target_seq_len, source_seq_len).","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_group_query_attention.html","id":"call-arguments","dir":"Reference","previous_headings":"","what":"Call Arguments","title":"Grouped Query Attention layer. — layer_group_query_attention","text":"query: Query tensor shape (batch_dim, target_seq_len, feature_dim), batch_dim batch size, target_seq_len length target sequence, feature_dim dimension feature. value: Value tensor shape (batch_dim, source_seq_len, feature_dim), batch_dim batch size, source_seq_len length source sequence, feature_dim dimension feature. key: Optional key tensor shape (batch_dim, source_seq_len, feature_dim). given, use value key value, common case. attention_mask: boolean mask shape (batch_dim, target_seq_len, source_seq_len), prevents attention certain positions. boolean mask specifies query elements can attend key elements, 1 indicates attention 0 indicates attention. Broadcasting can happen missing batch dimensions head dimension. return_attention_scores: boolean indicate whether output (attention_output, attention_scores) TRUE, attention_output FALSE. Defaults FALSE. training: Python boolean indicating whether layer behave training mode (adding dropout) inference mode (dropout). go either using training mode parent layer/model FALSE (inference) parent layer. use_causal_mask: boolean indicate whether apply causal mask prevent tokens attending future tokens (e.g., used decoder Transformer).","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_gru.html","id":null,"dir":"Reference","previous_headings":"","what":"Gated Recurrent Unit - Cho et al. 2014. — layer_gru","title":"Gated Recurrent Unit - Cho et al. 2014. — layer_gru","text":"Based available runtime hardware constraints, layer choose different implementations (cuDNN-based backend-native) maximize performance. GPU available arguments layer meet requirement cuDNN kernel (see details), layer use fast cuDNN implementation using TensorFlow backend. requirements use cuDNN implementation : activation == tanh recurrent_activation == sigmoid recurrent_dropout == 0 unroll FALSE use_bias TRUE reset_after TRUE Inputs, use masking, strictly right-padded. Eager execution enabled outermost context. two variants GRU implementation. default one based v3 reset gate applied hidden state matrix multiplication. one based original order reversed. second variant compatible CuDNNGRU (GPU-) allows inference CPU. Thus separate biases kernel recurrent_kernel. use variant, set reset_after=TRUE recurrent_activation='sigmoid'. example:","code":"inputs <- random_uniform(c(32, 10, 8)) outputs <- inputs |> layer_gru(4) shape(outputs) ## shape(32, 4) # (32, 4) gru <- layer_gru(, 4, return_sequences = TRUE, return_state = TRUE) c(whole_sequence_output, final_state) %<-% gru(inputs) shape(whole_sequence_output) ## shape(32, 10, 4) shape(final_state) ## shape(32, 4)"},{"path":"https://keras3.posit.co/dev/reference/layer_gru.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Gated Recurrent Unit - Cho et al. 2014. — layer_gru","text":"","code":"layer_gru(   object,   units,   activation = \"tanh\",   recurrent_activation = \"sigmoid\",   use_bias = TRUE,   kernel_initializer = \"glorot_uniform\",   recurrent_initializer = \"orthogonal\",   bias_initializer = \"zeros\",   kernel_regularizer = NULL,   recurrent_regularizer = NULL,   bias_regularizer = NULL,   activity_regularizer = NULL,   kernel_constraint = NULL,   recurrent_constraint = NULL,   bias_constraint = NULL,   dropout = 0,   recurrent_dropout = 0,   seed = NULL,   return_sequences = FALSE,   return_state = FALSE,   go_backwards = FALSE,   stateful = FALSE,   unroll = FALSE,   reset_after = TRUE,   use_cudnn = \"auto\",   ... )"},{"path":"https://keras3.posit.co/dev/reference/layer_gru.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Gated Recurrent Unit - Cho et al. 2014. — layer_gru","text":"object Object compose layer . tensor, array, sequential model. units Positive integer, dimensionality output space. activation Activation function use. Default: hyperbolic tangent (tanh). pass NULL, activation applied (ie. \"linear\" activation: (x) = x). recurrent_activation Activation function use recurrent step. Default: sigmoid (sigmoid). pass NULL, activation applied (ie. \"linear\" activation: (x) = x). use_bias Boolean, (default TRUE), whether layer use bias vector. kernel_initializer Initializer kernel weights matrix, used linear transformation inputs. Default: \"glorot_uniform\". recurrent_initializer Initializer recurrent_kernel weights matrix, used linear transformation recurrent state. Default: \"orthogonal\". bias_initializer Initializer bias vector. Default: \"zeros\". kernel_regularizer Regularizer function applied kernel weights matrix. Default: NULL. recurrent_regularizer Regularizer function applied recurrent_kernel weights matrix. Default: NULL. bias_regularizer Regularizer function applied bias vector. Default: NULL. activity_regularizer Regularizer function applied output layer (\"activation\"). Default: NULL. kernel_constraint Constraint function applied kernel weights matrix. Default: NULL. recurrent_constraint Constraint function applied recurrent_kernel weights matrix. Default: NULL. bias_constraint Constraint function applied bias vector. Default: NULL. dropout Float 0 1. Fraction units drop linear transformation inputs. Default: 0. recurrent_dropout Float 0 1. Fraction units drop linear transformation recurrent state. Default: 0. seed Random seed dropout. return_sequences Boolean. Whether return last output output sequence, full sequence. Default: FALSE. return_state Boolean. Whether return last state addition output. Default: FALSE. go_backwards Boolean (default FALSE). TRUE, process input sequence backwards return reversed sequence. stateful Boolean (default: FALSE). TRUE, last state sample index batch used initial state sample index following batch. unroll Boolean (default: FALSE). TRUE, network unrolled, else symbolic loop used. Unrolling can speed-RNN, although tends memory-intensive. Unrolling suitable short sequences. reset_after GRU convention (whether apply reset gate matrix multiplication). FALSE \"\", TRUE \"\" (default cuDNN compatible). use_cudnn Whether use cuDNN-backed implementation. \"auto\" attempt use cuDNN feasible, fallback default implementation . ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_gru.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Gated Recurrent Unit - Cho et al. 2014. — layer_gru","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_gru.html","id":"call-arguments","dir":"Reference","previous_headings":"","what":"Call Arguments","title":"Gated Recurrent Unit - Cho et al. 2014. — layer_gru","text":"inputs: 3D tensor, shape (batch, timesteps, feature). mask: Binary tensor shape (samples, timesteps) indicating whether given timestep masked  (optional). individual TRUE entry indicates corresponding timestep utilized, FALSE entry indicates corresponding timestep ignored. Defaults NULL. training: Python boolean indicating whether layer behave training mode inference mode. argument passed cell calling . relevant dropout recurrent_dropout used  (optional). Defaults NULL. initial_state: List initial state tensors passed first call cell (optional, NULL causes creation zero-filled initial state tensors). Defaults NULL.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_hashed_crossing.html","id":null,"dir":"Reference","previous_headings":"","what":"A preprocessing layer which crosses features using the ","title":"A preprocessing layer which crosses features using the ","text":"layer performs crosses categorical features using \"hashing trick\". Conceptually, transformation can thought : hash(concatenate(features)) %% num_bins. layer currently performs crosses scalar inputs batches scalar inputs. Valid input shapes (batch_size, 1), (batch_size) (). Note: layer wraps tf.keras.layers.HashedCrossing. used part compiled computation graph model backend TensorFlow. can however used backend running eagerly. can also always used part input preprocessing pipeline backend (outside model ), recommend use layer. Note: layer safe use inside tfdatasets pipeline (independently backend using).","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_hashed_crossing.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"A preprocessing layer which crosses features using the ","text":"","code":"layer_hashed_crossing(   object,   num_bins,   output_mode = \"int\",   sparse = FALSE,   name = NULL,   dtype = NULL,   ... )"},{"path":"https://keras3.posit.co/dev/reference/layer_hashed_crossing.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"A preprocessing layer which crosses features using the ","text":"object Object compose layer . tensor, array, sequential model. num_bins Number hash bins. output_mode Specification output layer. Values can \"int\", \"one_hot\" configuring layer follows: \"int\": Return integer bin indices directly. \"one_hot\": Encodes individual element input array size num_bins, containing 1 input's bin index. Defaults \"int\". sparse Boolean. applicable \"one_hot\" mode valid using TensorFlow backend. TRUE, returns SparseTensor instead dense Tensor. Defaults FALSE. name String, name object dtype datatype (e.g., \"float32\"). ... Keyword arguments construct layer.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_hashed_crossing.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"A preprocessing layer which crosses features using the ","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_hashed_crossing.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"A preprocessing layer which crosses features using the ","text":"Crossing two scalar features.     Crossing one-hotting two scalar features.","code":"feat1 <- c('A', 'B', 'A', 'B', 'A') |> as.array() feat2 <- c(101, 101, 101, 102, 102) |> as.integer() |> as.array() layer <- layer_hashed_crossing(num_bins = 5) layer(list(feat1, feat2)) ## tf.Tensor([1 4 1 1 3], shape=(5), dtype=int64) layer <- layer_hashed_crossing(num_bins = 5, output_mode = 'one_hot') layer(list(feat1, feat2)) ## tf.Tensor( ## [[0. 1. 0. 0. 0.] ##  [0. 0. 0. 0. 1.] ##  [0. 1. 0. 0. 0.] ##  [0. 1. 0. 0. 0.] ##  [0. 0. 0. 1. 0.]], shape=(5, 5), dtype=float32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_hashing.html","id":null,"dir":"Reference","previous_headings":"","what":"A preprocessing layer which hashes and bins categorical features. — layer_hashing","title":"A preprocessing layer which hashes and bins categorical features. — layer_hashing","text":"layer transforms categorical inputs hashed output. element-wise converts ints strings ints fixed range. stable hash function uses tensorflow::ops::Fingerprint produce output consistently across platforms. layer uses FarmHash64 default, provides consistent hashed output across different platforms stable across invocations, regardless device context, mixing input bits thoroughly. want obfuscate hashed output, can also pass random salt argument constructor. case, layer use SipHash64 hash function, salt value serving additional input hash function. Note: layer internally uses TensorFlow. used part compiled computation graph model backend TensorFlow. can however used backend running eagerly. can also always used part input preprocessing pipeline backend (outside model ), recommend use layer. Note: layer safe use inside tf.data pipeline (independently backend using). Example (FarmHash64)     Example (FarmHash64) mask value     Example (SipHash64)     Example (Siphash64 single integer, salt=[133, 133])","code":"layer <- layer_hashing(num_bins = 3) inp <- c('A', 'B', 'C', 'D', 'E') |> array(dim = c(5, 1)) layer(inp) ## tf.Tensor( ## [[1] ##  [0] ##  [1] ##  [1] ##  [2]], shape=(5, 1), dtype=int64) layer <- layer_hashing(num_bins=3, mask_value='') inp <- c('A', 'B', '', 'C', 'D') |> array(dim = c(5, 1)) layer(inp) ## tf.Tensor( ## [[1] ##  [1] ##  [0] ##  [2] ##  [2]], shape=(5, 1), dtype=int64) layer <- layer_hashing(num_bins=3, salt=c(133, 137)) inp <- c('A', 'B', 'C', 'D', 'E') |> array(dim = c(5, 1)) layer(inp) ## tf.Tensor( ## [[1] ##  [2] ##  [1] ##  [0] ##  [2]], shape=(5, 1), dtype=int64) layer <- layer_hashing(num_bins=3, salt=133) inp <- c('A', 'B', 'C', 'D', 'E') |> array(dim = c(5, 1)) layer(inp) ## tf.Tensor( ## [[0] ##  [0] ##  [2] ##  [1] ##  [0]], shape=(5, 1), dtype=int64)"},{"path":"https://keras3.posit.co/dev/reference/layer_hashing.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"A preprocessing layer which hashes and bins categorical features. — layer_hashing","text":"","code":"layer_hashing(   object,   num_bins,   mask_value = NULL,   salt = NULL,   output_mode = \"int\",   sparse = FALSE,   ... )"},{"path":"https://keras3.posit.co/dev/reference/layer_hashing.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"A preprocessing layer which hashes and bins categorical features. — layer_hashing","text":"object Object compose layer . tensor, array, sequential model. num_bins Number hash bins. Note includes mask_value bin, effective number bins (num_bins - 1) mask_value set. mask_value value represents masked inputs, mapped index 0. NULL means mask term added hashing start index 0. Defaults NULL. salt single unsigned integer NULL. passed, hash function used SipHash64, values used additional input (known \"salt\" cryptography). non-zero. NULL, uses FarmHash64 hash function. also supports list 2 unsigned integer numbers, see reference paper details. Defaults NULL. output_mode Specification output layer. Values can \"int\", \"one_hot\", \"multi_hot\", \"count\" configuring layer follows: \"int\": Return integer bin indices directly. \"one_hot\": Encodes individual element input array size num_bins, containing 1 input's bin index. last dimension size 1, encode dimension. last dimension size 1, append new dimension encoded output. \"multi_hot\": Encodes sample input single array size num_bins, containing 1 bin index index present sample. Treats last dimension sample dimension, input shape (..., sample_length), output shape (..., num_tokens). \"count\": \"multi_hot\", int array contains count number times bin index appeared sample. Defaults \"int\". sparse Boolean. applicable \"one_hot\", \"multi_hot\", \"count\" output modes. supported TensorFlow backend. TRUE, returns SparseTensor instead dense Tensor. Defaults FALSE. ... Keyword arguments construct layer.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_hashing.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"A preprocessing layer which hashes and bins categorical features. — layer_hashing","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_hashing.html","id":"input-shape","dir":"Reference","previous_headings":"","what":"Input Shape","title":"A preprocessing layer which hashes and bins categorical features. — layer_hashing","text":"single string, list strings, int32 int64 tensor shape (batch_size, ...,).","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_hashing.html","id":"output-shape","dir":"Reference","previous_headings":"","what":"Output Shape","title":"A preprocessing layer which hashes and bins categorical features. — layer_hashing","text":"int32 tensor shape (batch_size, ...).","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_hashing.html","id":"reference","dir":"Reference","previous_headings":"","what":"Reference","title":"A preprocessing layer which hashes and bins categorical features. — layer_hashing","text":"SipHash salt","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_identity.html","id":null,"dir":"Reference","previous_headings":"","what":"Identity layer. — layer_identity","title":"Identity layer. — layer_identity","text":"layer used placeholder operation performed. layer just returns inputs argument output.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_identity.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Identity layer. — layer_identity","text":"","code":"layer_identity(object, ...)"},{"path":"https://keras3.posit.co/dev/reference/layer_identity.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Identity layer. — layer_identity","text":"object Object compose layer . tensor, array, sequential model. ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_identity.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Identity layer. — layer_identity","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_input.html","id":null,"dir":"Reference","previous_headings":"","what":"keras_input — layer_input","title":"keras_input — layer_input","text":"layer_input() alias keras_input(). See ?keras_input() full documentation.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_input.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"keras_input — layer_input","text":"","code":"layer_input(   shape = NULL,   batch_size = NULL,   dtype = NULL,   sparse = NULL,   ragged = NULL,   batch_shape = NULL,   name = NULL,   tensor = NULL,   optional = FALSE )"},{"path":"https://keras3.posit.co/dev/reference/layer_input.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"keras_input — layer_input","text":"shape shape list (list integers NULL objects), including batch size. instance, shape = c(32) indicates expected input batches 32-dimensional vectors. Elements list can NULL NA; NULL/NA elements represent dimensions shape known may vary (e.g. sequence length). batch_size Optional static batch size (integer). dtype data type expected input, string (e.g. \"float32\", \"int32\"...) sparse boolean specifying whether expected input sparse tensors. Note , sparse FALSE, sparse tensors can still passed input - densified default value 0. feature supported TensorFlow backend. Defaults FALSE. ragged boolean specifying whether expected input ragged tensors. Note , ragged FALSE, ragged tensors can still passed input - densified default value 0. feature supported TensorFlow backend. Defaults FALSE. batch_shape Optional shape list (list integers NULL objects), including batch size. name Optional name string layer. unique model (reuse name twice). autogenerated provided. tensor Optional existing tensor wrap Input layer. set, layer use tensor rather creating new placeholder tensor. optional Boolean, whether input optional . optional input can accept NULL values.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_input.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"keras_input — layer_input","text":"Keras tensor, can passed inputs argument (keras_model()).","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_integer_lookup.html","id":null,"dir":"Reference","previous_headings":"","what":"A preprocessing layer that maps integers to (possibly encoded) indices. — layer_integer_lookup","title":"A preprocessing layer that maps integers to (possibly encoded) indices. — layer_integer_lookup","text":"layer maps set arbitrary integer input tokens indexed integer output via table-based vocabulary lookup. layer's output indices contiguously arranged maximum vocab size, even input tokens non-continguous unbounded. layer supports multiple options encoding output via output_mode, optional support --vocabulary (OOV) tokens masking. vocabulary layer must either supplied construction learned via adapt(). adapt(), layer analyze data set, determine frequency individual integer tokens, create vocabulary . vocabulary capped size, frequent tokens used create vocabulary others treated OOV. two possible output modes layer.  output_mode \"int\", input integers converted index vocabulary (integer).  output_mode \"multi_hot\", \"count\", \"tf_idf\", input integers encoded array dimension corresponds element vocabulary. vocabulary can optionally contain mask token well OOV token (can optionally occupy multiple indices vocabulary, set num_oov_indices). position tokens vocabulary fixed. output_mode \"int\", vocabulary begin mask token index 0, followed OOV indices, followed rest vocabulary. output_mode \"multi_hot\", \"count\", \"tf_idf\" vocabulary begin OOV indices instances mask token dropped. Note: layer uses TensorFlow internally. used part compiled computation graph model backend TensorFlow. can however used backend running eagerly. can also always used part input preprocessing pipeline backend (outside model ), recommend use layer. Note: layer safe use inside tf.data pipeline (independently backend using). Note: working layer outputs directly (e.g., passing outputs another layer, using lower-level operations like [ op_*): returned indices 0-based. However, default settings, first (0) index OOV token, returned indices offset 1 may appear 1-based.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_integer_lookup.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"A preprocessing layer that maps integers to (possibly encoded) indices. — layer_integer_lookup","text":"","code":"layer_integer_lookup(   object,   max_tokens = NULL,   num_oov_indices = 1L,   mask_token = NULL,   oov_token = -1L,   vocabulary = NULL,   vocabulary_dtype = \"int64\",   idf_weights = NULL,   invert = FALSE,   output_mode = \"int\",   sparse = FALSE,   pad_to_max_tokens = FALSE,   name = NULL,   ... )"},{"path":"https://keras3.posit.co/dev/reference/layer_integer_lookup.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"A preprocessing layer that maps integers to (possibly encoded) indices. — layer_integer_lookup","text":"object Object compose layer . tensor, array, sequential model. max_tokens Maximum size vocabulary layer. specified adapting vocabulary setting pad_to_max_tokens=TRUE. NULL, cap size vocabulary. Note size includes OOV mask tokens. Defaults NULL. num_oov_indices number --vocabulary tokens use. value 1, OOV inputs modulated determine OOV value. value 0, OOV inputs cause error calling layer. Defaults 1. mask_token integer token represents masked inputs. output_mode \"int\", token included vocabulary mapped index 0. output modes, token appear vocabulary instances mask token input dropped. set NULL, mask term added. Defaults NULL. oov_token used invert TRUE. token return OOV indices. Defaults -1. vocabulary Optional. Either array integers string path text file. passing array, can pass list, list, 1D NumPy array, 1D tensor containing integer vocabulary terms. passing file path, file contain one line per term vocabulary. argument set, need adapt() layer. vocabulary_dtype dtype vocabulary terms, example \"int64\" \"int32\". Defaults \"int64\". idf_weights valid output_mode \"tf_idf\". list, list, 1D NumPy array, 1D tensor length vocabulary, containing floating point inverse document frequency weights, multiplied per sample term counts final TF-IDF weight. vocabulary argument set, output_mode \"tf_idf\", argument must supplied. invert valid output_mode \"int\". TRUE, layer map indices vocabulary items instead mapping vocabulary items indices. Defaults FALSE. output_mode Specification output layer. Values can \"int\", \"one_hot\", \"multi_hot\", \"count\", \"tf_idf\" configuring layer follows: \"int\": Return vocabulary indices input tokens. \"one_hot\": Encodes individual element input array size vocabulary, containing 1 element index. last dimension size 1, encode dimension. last dimension size 1, append new dimension encoded output. \"multi_hot\": Encodes sample input single array size vocabulary, containing 1 vocabulary term present sample. Treats last dimension sample dimension, input shape (..., sample_length), output shape (..., num_tokens). \"count\": \"multi_hot\", int array contains count number times token index appeared sample. \"tf_idf\": \"multi_hot\", TF-IDF algorithm applied find value token slot. \"int\" output, shape input output supported. output modes, currently output rank 2 supported. Defaults \"int\". sparse Boolean. applicable \"multi_hot\", \"count\", \"tf_idf\" output modes. supported TensorFlow backend. TRUE, returns SparseTensor instead dense Tensor. Defaults FALSE. pad_to_max_tokens applicable output_mode \"multi_hot\", \"count\", \"tf_idf\". TRUE, output feature axis padded max_tokens even number unique tokens vocabulary less max_tokens, resulting tensor shape (batch_size, max_tokens) regardless vocabulary size. Defaults FALSE. name String, name object ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_integer_lookup.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"A preprocessing layer that maps integers to (possibly encoded) indices. — layer_integer_lookup","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_integer_lookup.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"A preprocessing layer that maps integers to (possibly encoded) indices. — layer_integer_lookup","text":"Creating lookup layer known vocabulary example creates lookup layer pre-existing vocabulary.     Creating lookup layer adapted vocabulary example creates lookup layer generates vocabulary analyzing dataset.     Note OOV token -1 added vocabulary. remaining tokens sorted frequency (42, 2 occurrences, first) inverse sort order.     Lookups multiple OOV indices example demonstrates use lookup layer multiple OOV indices.  layer created one OOV index, OOV tokens hashed number OOV buckets, distributing OOV tokens deterministic fashion across set.     Note output OOV token 37 1, output OOV token 1000 0. -vocab terms output index increased 1 earlier examples (12 maps 2, etc) order make space extra OOV token. One-hot output Configure layer output_mode='one_hot'. Note first num_oov_indices dimensions ont_hot encoding represent OOV values.     Multi-hot output Configure layer output_mode = 'multi_hot'. Note first num_oov_indices dimensions multi_hot encoding represent OOV tokens     Token count output Configure layer output_mode='count'. multi_hot output, first num_oov_indices dimensions output represent OOV tokens.     TF-IDF output Configure layer output_mode='tf_idf'. multi_hot output, first num_oov_indices dimensions output represent OOV tokens. token bin output token_count * idf_weight, idf weights inverse document frequency weights per token. provided along vocabulary. Note idf_weight OOV tokens default average idf weights passed .     specify idf weights oov tokens, need pass entire vocabulary including leading oov token.     adapting layer \"tf_idf\" mode, input sample considered document, IDF weight per token calculated : log(1 + num_documents / (1 + token_document_count)). Inverse lookup example demonstrates map indices tokens using layer. (can also use adapt() inverse = TRUE, simplicity pass vocab example.)     Note first index correspond oov token default. Forward inverse lookup pairs example demonstrates use vocabulary standard lookup layer create inverse lookup layer.     example, input token 1000 resulted output -1, since 1000 vocabulary - got represented OOV, OOV tokens returned -1 inverse layer. Also, note inverse work, must already set forward layer vocabulary either directly via adapt() calling get_vocabulary().","code":"vocab <- c(12, 36, 1138, 42) |> as.integer() data <- op_array(rbind(c(12, 1138, 42),                        c(42, 1000, 36)))  # Note OOV tokens out <- data |> layer_integer_lookup(vocabulary = vocab) out ## tf.Tensor( ## [[1 3 4] ##  [4 0 2]], shape=(2, 3), dtype=int64) data <- op_array(rbind(c(12, 1138, 42),                        c(42, 1000, 36)))  # Note OOV tokens layer <- layer_integer_lookup() layer |> adapt(data) layer |> get_vocabulary() |> str() ## List of 6 ##  $ : int -1 ##  $ : num 42 ##  $ : num 1138 ##  $ : num 1000 ##  $ : num 36 ##  $ : num 12 layer(data) ## tf.Tensor( ## [[5 2 1] ##  [1 3 4]], shape=(2, 3), dtype=int64) vocab <- c(12, 36, 1138, 42) |> as.integer() data <- op_array(rbind(c(12, 1138, 42),                        c(37, 1000, 36)))  # Note OOV tokens out <- data |>   layer_integer_lookup(vocabulary = vocab,                        num_oov_indices = 2) out ## tf.Tensor( ## [[2 4 5] ##  [1 0 3]], shape=(2, 3), dtype=int64) vocab <- c(12, 36, 1138, 42) |> as.integer() data <- op_array(c(12, 36, 1138, 42, 7), 'int32')  # Note OOV tokens layer <- layer_integer_lookup(vocabulary = vocab,                               output_mode = 'one_hot') layer(data) ## tf.Tensor( ## [[0 1 0 0 0] ##  [0 0 1 0 0] ##  [0 0 0 1 0] ##  [0 0 0 0 1] ##  [1 0 0 0 0]], shape=(5, 5), dtype=int64) vocab <- c(12, 36, 1138, 42) |> as.integer() data <- op_array(rbind(c(12, 1138, 42, 42),                       c(42,    7, 36,  7)), \"int64\")  # Note OOV tokens layer <- layer_integer_lookup(vocabulary = vocab,                               output_mode = 'multi_hot') layer(data) ## tf.Tensor( ## [[0 1 0 1 1] ##  [1 0 1 0 1]], shape=(2, 5), dtype=int64) vocab <- c(12, 36, 1138, 42) |> as.integer() data <- rbind(c(12, 1138, 42, 42),               c(42,    7, 36,  7)) |> op_array(\"int64\") layer <- layer_integer_lookup(vocabulary = vocab,                               output_mode = 'count') layer(data) ## tf.Tensor( ## [[0 1 0 1 2] ##  [2 0 1 0 1]], shape=(2, 5), dtype=int64) vocab <- c(12, 36, 1138, 42) |> as.integer() idf_weights <- c(0.25, 0.75, 0.6, 0.4) data <- rbind(c(12, 1138, 42, 42),               c(42,    7, 36,  7)) |> op_array(\"int64\") layer <- layer_integer_lookup(output_mode = 'tf_idf',                               vocabulary = vocab,                               idf_weights = idf_weights) layer(data) ## tf.Tensor( ## [[0.   0.25 0.   0.6  0.8 ] ##  [1.   0.   0.75 0.   0.4 ]], shape=(2, 5), dtype=float32) vocab <- c(-1, 12, 36, 1138, 42) |> as.integer() idf_weights <- c(0.9, 0.25, 0.75, 0.6, 0.4) data <- rbind(c(12, 1138, 42, 42),               c(42,    7, 36,  7)) |> op_array(\"int64\") layer <- layer_integer_lookup(output_mode = 'tf_idf',                               vocabulary = vocab,                               idf_weights = idf_weights) layer(data) ## tf.Tensor( ## [[0.   0.25 0.   0.6  0.8 ] ##  [1.8  0.   0.75 0.   0.4 ]], shape=(2, 5), dtype=float32) vocab <- c(12, 36, 1138, 42) |> as.integer() data <- op_array(c(1, 3, 4,                    4, 0, 2)) |> op_reshape(c(2,-1)) |> op_cast(\"int32\") layer <- layer_integer_lookup(vocabulary = vocab, invert = TRUE) layer(data) ## tf.Tensor( ## [[  12 1138   42] ##  [  42   -1   36]], shape=(2, 3), dtype=int64) vocab <- c(12, 36, 1138, 42) |> as.integer() data <- op_array(rbind(c(12, 1138, 42), c(42, 1000, 36)), \"int32\") layer <- layer_integer_lookup(vocabulary = vocab) i_layer <- layer_integer_lookup(vocabulary = get_vocabulary(layer),                                 invert = TRUE) int_data <- layer(data) i_layer(int_data) ## tf.Tensor( ## [[  12 1138   42] ##  [  42   -1   36]], shape=(2, 3), dtype=int64)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_jax_model_wrapper.html","id":null,"dir":"Reference","previous_headings":"","what":"Keras Layer that wraps a JAX model. — layer_jax_model_wrapper","title":"Keras Layer that wraps a JAX model. — layer_jax_model_wrapper","text":"layer enables use JAX components within Keras using JAX backend Keras.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_jax_model_wrapper.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Keras Layer that wraps a JAX model. — layer_jax_model_wrapper","text":"","code":"layer_jax_model_wrapper(   object,   call_fn,   init_fn = NULL,   params = NULL,   state = NULL,   seed = NULL,   ...,   dtype = NULL )"},{"path":"https://keras3.posit.co/dev/reference/layer_jax_model_wrapper.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Keras Layer that wraps a JAX model. — layer_jax_model_wrapper","text":"object Object compose layer . tensor, array, sequential model. call_fn function call model. See description list arguments takes outputs returns. init_fn function call initialize model. See description list arguments takes outputs returns. NULL, params /state must provided. params PyTree containing model trainable parameters. allows passing trained parameters controlling initialization. params state NULL, init_fn() called build time initialize trainable parameters model. state PyTree containing model non-trainable state. allows passing learned state controlling initialization. params state NULL, call_fn() takes state argument, init_fn() called build time initialize non-trainable state model. seed Seed random number generator. Optional. ... forward/backward compatability. dtype dtype layer's computations weights. Can also keras.DTypePolicy. Optional. Defaults default policy.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_jax_model_wrapper.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Keras Layer that wraps a JAX model. — layer_jax_model_wrapper","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_jax_model_wrapper.html","id":"model-function","dir":"Reference","previous_headings":"","what":"Model function","title":"Keras Layer that wraps a JAX model. — layer_jax_model_wrapper","text":"layer accepts JAX models form function, call_fn(), must take following arguments exact names: params: trainable parameters model. state (optional): non-trainable state model. Can omitted model non-trainable state. rng (optional): jax.random.PRNGKey instance. Can omitted model need RNGs, neither training inference. inputs: inputs model, JAX array PyTree arrays. training (optional): argument specifying training mode inference mode, TRUE passed training mode. Can omitted model behaves training mode inference mode. inputs argument mandatory. Inputs model must provided via single argument. JAX model takes multiple inputs separate arguments, must combined single structure, instance tuple() dict().","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_jax_model_wrapper.html","id":"model-weights-initialization","dir":"Reference","previous_headings":"","what":"Model weights initialization","title":"Keras Layer that wraps a JAX model. — layer_jax_model_wrapper","text":"initialization params state model can handled layer, case init_fn() argument must provided. allows model initialized dynamically right shape. Alternatively, shape known, params argument optionally state argument can used create already initialized model. init_fn() function, provided, must take following arguments exact names: rng: jax.random.PRNGKey instance. inputs: JAX array PyTree arrays placeholder values provide shape inputs. training (optional): argument specifying training mode inference mode. True always passed init_fn. Can omitted regardless whether call_fn training argument.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_jax_model_wrapper.html","id":"models-with-non-trainable-state","dir":"Reference","previous_headings":"","what":"Models with non-trainable state","title":"Keras Layer that wraps a JAX model. — layer_jax_model_wrapper","text":"JAX models non-trainable state: call_fn() must state argument call_fn() must return tuple() containing outputs model new non-trainable state model init_fn() must return tuple() containing initial trainable params model initial non-trainable state model. code shows possible combination call_fn() init_fn() signatures model non-trainable state. example, model training argument rng argument call_fn().","code":"stateful_call <- function(params, state, rng, inputs, training) {   outputs <- ....   new_state <- ....   tuple(outputs, new_state) }  stateful_init <- function(rng, inputs) {   initial_params <- ....   initial_state <- ....   tuple(initial_params, initial_state) }"},{"path":"https://keras3.posit.co/dev/reference/layer_jax_model_wrapper.html","id":"models-without-non-trainable-state","dir":"Reference","previous_headings":"","what":"Models without non-trainable state","title":"Keras Layer that wraps a JAX model. — layer_jax_model_wrapper","text":"JAX models non-trainable state: call_fn() must state argument call_fn() must return outputs model init_fn() must return initial trainable params model. code shows possible combination call_fn() init_fn() signatures model without non-trainable state. example, model training argument rng argument call_fn().","code":"stateful_call <- function(pparams, inputs) {   outputs <- ....   outputs }  stateful_init <- function(rng, inputs) {   initial_params <- ....   initial_params }"},{"path":"https://keras3.posit.co/dev/reference/layer_jax_model_wrapper.html","id":"conforming-to-the-required-signature","dir":"Reference","previous_headings":"","what":"Conforming to the required signature","title":"Keras Layer that wraps a JAX model. — layer_jax_model_wrapper","text":"model different signature one required JaxLayer, one can easily write wrapper method adapt arguments. example shows model multiple inputs separate arguments, expects multiple RNGs dict, deterministic argument opposite meaning training. conform, inputs combined single structure using tuple, RNG split used populate expected dict, Boolean flag negated:","code":"jax <- import(\"jax\") my_model_fn <- function(params, rngs, input1, input2, deterministic) {   ....   if (!deterministic) {     dropout_rng <- rngs$dropout     keep <- jax$random$bernoulli(dropout_rng, dropout_rate, x$shape)     x <- jax$numpy$where(keep, x / dropout_rate, 0)     ....   }   ....   return(outputs) }  my_model_wrapper_fn <- function(params, rng, inputs, training) {   c(input1, input2) %<-% inputs   c(rng1, rng2) %<-% jax$random$split(rng)   rngs <-  list(dropout = rng1, preprocessing = rng2)   deterministic <-  !training   my_model_fn(params, rngs, input1, input2, deterministic) }  keras_layer <- layer_jax_model_wrapper(call_fn = my_model_wrapper_fn,                                        params = initial_params)"},{"path":"https://keras3.posit.co/dev/reference/layer_jax_model_wrapper.html","id":"usage-with-haiku-modules","dir":"Reference","previous_headings":"","what":"Usage with Haiku modules","title":"Keras Layer that wraps a JAX model. — layer_jax_model_wrapper","text":"JaxLayer enables use Haiku components form haiku.Module. achieved transforming module per Haiku pattern passing module.apply call_fn parameter module.init init_fn parameter needed. model non-trainable state, transformed haiku.transform_with_state. model non-trainable state, transformed haiku.transform. Additionally, optionally, module use RNGs \"apply\", can transformed haiku.without_apply_rng. following example shows create JaxLayer Haiku module uses random number generators via hk.next_rng_key() takes training positional argument:","code":"# reticulate::py_install(\"haiku\", \"r-keras\") hk <- import(\"haiku\") MyHaikuModule(hk$Module) \\%py_class\\% {    `__call__` <- \\(self, x, training) {     x <- hk$Conv2D(32L, tuple(3L, 3L))(x)     x <- jax$nn$relu(x)     x <- hk$AvgPool(tuple(1L, 2L, 2L, 1L),                     tuple(1L, 2L, 2L, 1L), \"VALID\")(x)     x <- hk$Flatten()(x)     x <- hk$Linear(200L)(x)     if (training)       x <- hk$dropout(rng = hk$next_rng_key(), rate = 0.3, x = x)     x <- jax$nn$relu(x)     x <- hk$Linear(10L)(x)     x <- jax$nn$softmax(x)     x   }  }  my_haiku_module_fn <- function(inputs, training) {   module <- MyHaikuModule()   module(inputs, training) }  transformed_module <- hk$transform(my_haiku_module_fn)  keras_layer <-   layer_jax_model_wrapper(call_fn = transformed_module$apply,                           init_fn = transformed_module$init)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_lambda.html","id":null,"dir":"Reference","previous_headings":"","what":"Wraps arbitrary expressions as a Layer object. — layer_lambda","title":"Wraps arbitrary expressions as a Layer object. — layer_lambda","text":"layer_lambda() layer exists arbitrary expressions can used Layer constructing Sequential Functional API models. Lambda layers best suited simple operations quick experimentation. advanced use cases, prefer writing new subclasses Layer using new_layer_class().","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_lambda.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Wraps arbitrary expressions as a Layer object. — layer_lambda","text":"","code":"layer_lambda(   object,   f,   output_shape = NULL,   mask = NULL,   arguments = NULL,   ... )"},{"path":"https://keras3.posit.co/dev/reference/layer_lambda.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Wraps arbitrary expressions as a Layer object. — layer_lambda","text":"object Object compose layer . tensor, array, sequential model. f function evaluated. Takes input tensor first argument. output_shape Expected output shape function. argument can usually inferred explicitly provided. Can list function. list, specifies first dimension onward; sample dimension assumed either input: output_shape = c(input_shape[1], output_shape) , input NULL sample dimension also NULL: output_shape = c(NA, output_shape). function, specifies entire shape function input shape: output_shape = f(input_shape). mask Either NULL (indicating masking) callable signature compute_mask layer method, tensor returned output mask regardless input . arguments Optional named list arguments passed function. ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_lambda.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Wraps arbitrary expressions as a Layer object. — layer_lambda","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_lambda.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Wraps arbitrary expressions as a Layer object. — layer_lambda","text":"","code":"# add a x -> x^2 layer model <- keras_model_sequential() model |> layer_lambda(\\(x) x^2)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_layer_normalization.html","id":null,"dir":"Reference","previous_headings":"","what":"Layer normalization layer (Ba et al., 2016). — layer_layer_normalization","title":"Layer normalization layer (Ba et al., 2016). — layer_layer_normalization","text":"Normalize activations previous layer given example batch independently, rather across batch like Batch Normalization. .e. applies transformation maintains mean activation within example close 0 activation standard deviation close 1. scale center enabled, layer scale normalized outputs broadcasting trainable variable gamma, center outputs broadcasting trainable variable beta. gamma default ones tensor beta default zeros tensor, centering scaling -ops training begun. , scaling centering enabled normalization equations follows: Let intermediate activations mini-batch inputs. sample x batch inputs, compute mean variance sample, normalize value sample (including small factor epsilon numerical stability), finally, transform normalized output gamma beta, learned parameters:   gamma beta span axes inputs specified axis, part inputs' shape must fully defined. example:         Note implementations layer normalization may choose define gamma beta separate set axes axes normalized across. example, Group Normalization (Wu et al. 2018) group size 1 corresponds layer_layer_normalization() normalizes across height, width, channel gamma beta span channel dimension. , layer_layer_normalization() implementation match layer_group_normalization() layer group size set 1.","code":"outputs <- inputs |> apply(1, function(x) {   x_normalized <- (x - mean(x)) /                   sqrt(var(x) + epsilon)   x_normalized * gamma + beta }) layer <- layer_layer_normalization(axis = c(2, 3, 4))  layer(op_ones(c(5, 20, 30, 40))) |> invisible() # build() shape(layer$beta) ## shape(20, 30, 40) shape(layer$gamma) ## shape(20, 30, 40)"},{"path":"https://keras3.posit.co/dev/reference/layer_layer_normalization.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Layer normalization layer (Ba et al., 2016). — layer_layer_normalization","text":"","code":"layer_layer_normalization(   object,   axis = -1L,   epsilon = 0.001,   center = TRUE,   scale = TRUE,   beta_initializer = \"zeros\",   gamma_initializer = \"ones\",   beta_regularizer = NULL,   gamma_regularizer = NULL,   beta_constraint = NULL,   gamma_constraint = NULL,   ... )"},{"path":"https://keras3.posit.co/dev/reference/layer_layer_normalization.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Layer normalization layer (Ba et al., 2016). — layer_layer_normalization","text":"object Object compose layer . tensor, array, sequential model. axis Integer list. axis axes normalize across. Typically, features axis/axes. left-axes typically batch axis/axes. -1 last dimension input. Defaults -1. epsilon Small float added variance avoid dividing zero. Defaults 1e-3. center TRUE, add offset beta normalized tensor. FALSE, beta ignored. Defaults TRUE. scale TRUE, multiply gamma. FALSE, gamma used. next layer linear (also e.g. layer_activation_relu()), can disabled since scaling done next layer. Defaults TRUE. beta_initializer Initializer beta weight. Defaults zeros. gamma_initializer Initializer gamma weight. Defaults ones. beta_regularizer Optional regularizer beta weight. NULL default. gamma_regularizer Optional regularizer gamma weight. NULL default. beta_constraint Optional constraint beta weight. NULL default. gamma_constraint Optional constraint gamma weight. NULL default. ... Base layer keyword arguments (e.g. name dtype).","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_layer_normalization.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Layer normalization layer (Ba et al., 2016). — layer_layer_normalization","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_layer_normalization.html","id":"reference","dir":"Reference","previous_headings":"","what":"Reference","title":"Layer normalization layer (Ba et al., 2016). — layer_layer_normalization","text":"Lei Ba et al., 2016.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_lstm.html","id":null,"dir":"Reference","previous_headings":"","what":"Long Short-Term Memory layer - Hochreiter 1997. — layer_lstm","title":"Long Short-Term Memory layer - Hochreiter 1997. — layer_lstm","text":"Based available runtime hardware constraints, layer choose different implementations (cuDNN-based backend-native) maximize performance. GPU available arguments layer meet requirement cuDNN kernel (see details), layer use fast cuDNN implementation using TensorFlow backend. requirements use cuDNN implementation : activation == tanh recurrent_activation == sigmoid recurrent_dropout == 0 unroll FALSE use_bias TRUE Inputs, use masking, strictly right-padded. Eager execution enabled outermost context. example:","code":"input <- random_uniform(c(32, 10, 8)) output <- input |> layer_lstm(4) shape(output) ## shape(32, 4) lstm <- layer_lstm(units = 4, return_sequences = TRUE, return_state = TRUE) c(whole_seq_output, final_memory_state, final_carry_state) %<-% lstm(input) shape(whole_seq_output) ## shape(32, 10, 4) shape(final_memory_state) ## shape(32, 4) shape(final_carry_state) ## shape(32, 4)"},{"path":"https://keras3.posit.co/dev/reference/layer_lstm.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Long Short-Term Memory layer - Hochreiter 1997. — layer_lstm","text":"","code":"layer_lstm(   object,   units,   activation = \"tanh\",   recurrent_activation = \"sigmoid\",   use_bias = TRUE,   kernel_initializer = \"glorot_uniform\",   recurrent_initializer = \"orthogonal\",   bias_initializer = \"zeros\",   unit_forget_bias = TRUE,   kernel_regularizer = NULL,   recurrent_regularizer = NULL,   bias_regularizer = NULL,   activity_regularizer = NULL,   kernel_constraint = NULL,   recurrent_constraint = NULL,   bias_constraint = NULL,   dropout = 0,   recurrent_dropout = 0,   seed = NULL,   return_sequences = FALSE,   return_state = FALSE,   go_backwards = FALSE,   stateful = FALSE,   unroll = FALSE,   use_cudnn = \"auto\",   ... )"},{"path":"https://keras3.posit.co/dev/reference/layer_lstm.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Long Short-Term Memory layer - Hochreiter 1997. — layer_lstm","text":"object Object compose layer . tensor, array, sequential model. units Positive integer, dimensionality output space. activation Activation function use. Default: hyperbolic tangent (tanh). pass NULL, activation applied (ie. \"linear\" activation: (x) = x). recurrent_activation Activation function use recurrent step. Default: sigmoid (sigmoid). pass NULL, activation applied (ie. \"linear\" activation: (x) = x). use_bias Boolean, (default TRUE), whether layer use bias vector. kernel_initializer Initializer kernel weights matrix, used linear transformation inputs. Default: \"glorot_uniform\". recurrent_initializer Initializer recurrent_kernel weights matrix, used linear transformation recurrent state. Default: \"orthogonal\". bias_initializer Initializer bias vector. Default: \"zeros\". unit_forget_bias Boolean (default TRUE). TRUE, add 1 bias forget gate initialization. Setting TRUE also force bias_initializer=\"zeros\". recommended Jozefowicz et al. kernel_regularizer Regularizer function applied kernel weights matrix. Default: NULL. recurrent_regularizer Regularizer function applied recurrent_kernel weights matrix. Default: NULL. bias_regularizer Regularizer function applied bias vector. Default: NULL. activity_regularizer Regularizer function applied output layer (\"activation\"). Default: NULL. kernel_constraint Constraint function applied kernel weights matrix. Default: NULL. recurrent_constraint Constraint function applied recurrent_kernel weights matrix. Default: NULL. bias_constraint Constraint function applied bias vector. Default: NULL. dropout Float 0 1. Fraction units drop linear transformation inputs. Default: 0. recurrent_dropout Float 0 1. Fraction units drop linear transformation recurrent state. Default: 0. seed Random seed dropout. return_sequences Boolean. Whether return last output output sequence, full sequence. Default: FALSE. return_state Boolean. Whether return last state addition output. Default: FALSE. go_backwards Boolean (default: FALSE). TRUE, process input sequence backwards return reversed sequence. stateful Boolean (default: FALSE). TRUE, last state sample index batch used initial state sample index following batch. unroll Boolean (default FALSE). TRUE, network unrolled, else symbolic loop used. Unrolling can speed-RNN, although tends memory-intensive. Unrolling suitable short sequences. use_cudnn Whether use cuDNN-backed implementation. \"auto\" attempt use cuDNN feasible, fallback default implementation . ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_lstm.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Long Short-Term Memory layer - Hochreiter 1997. — layer_lstm","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_lstm.html","id":"call-arguments","dir":"Reference","previous_headings":"","what":"Call Arguments","title":"Long Short-Term Memory layer - Hochreiter 1997. — layer_lstm","text":"inputs: 3D tensor, shape (batch, timesteps, feature). mask: Binary tensor shape (samples, timesteps) indicating whether given timestep masked  (optional). individual TRUE entry indicates corresponding timestep utilized, FALSE entry indicates corresponding timestep ignored. Defaults NULL. training: Boolean indicating whether layer behave training mode inference mode. argument passed cell calling . relevant dropout recurrent_dropout used  (optional). Defaults NULL. initial_state: List initial state tensors passed first call cell (optional, NULL causes creation zero-filled initial state tensors). Defaults NULL.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_masking.html","id":null,"dir":"Reference","previous_headings":"","what":"Masks a sequence by using a mask value to skip timesteps. — layer_masking","title":"Masks a sequence by using a mask value to skip timesteps. — layer_masking","text":"timestep input tensor (second dimension tensor), values input tensor timestep equal mask_value, timestep masked (skipped) downstream layers (long support masking). downstream layer support masking yet receives input mask, exception raised.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_masking.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Masks a sequence by using a mask value to skip timesteps. — layer_masking","text":"","code":"layer_masking(object, mask_value = 0, ...)"},{"path":"https://keras3.posit.co/dev/reference/layer_masking.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Masks a sequence by using a mask value to skip timesteps. — layer_masking","text":"object Object compose layer . tensor, array, sequential model. mask_value see description ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_masking.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Masks a sequence by using a mask value to skip timesteps. — layer_masking","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_masking.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Masks a sequence by using a mask value to skip timesteps. — layer_masking","text":"Consider array x shape c(samples, timesteps, features), fed LSTM layer. want mask timestep #3 #5 lack data timesteps. can: Set x[, 3, ] <- 0. x[, 5, ] <- 0. Insert layer_masking() layer mask_value = 0. LSTM layer:","code":"c(samples, timesteps, features) %<-% c(32, 10, 8) inputs <- c(samples, timesteps, features) %>% { array(runif(prod(.)), dim = .) } inputs[, 3, ] <- 0 inputs[, 5, ] <- 0  model <- keras_model_sequential() %>%   layer_masking(mask_value = 0) %>%   layer_lstm(32)  output <- model(inputs) # The time step 3 and 5 will be skipped from LSTM calculation."},{"path":"https://keras3.posit.co/dev/reference/layer_masking.html","id":"note","dir":"Reference","previous_headings":"","what":"Note","title":"Masks a sequence by using a mask value to skip timesteps. — layer_masking","text":"Keras masking convention, masked timestep denoted mask value FALSE, non-masked (.e. usable) timestep denoted mask value TRUE.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_max_num_bounding_boxes.html","id":null,"dir":"Reference","previous_headings":"","what":"Ensure the maximum number of bounding boxes. — layer_max_num_bounding_boxes","title":"Ensure the maximum number of bounding boxes. — layer_max_num_bounding_boxes","text":"Ensure maximum number bounding boxes.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_max_num_bounding_boxes.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Ensure the maximum number of bounding boxes. — layer_max_num_bounding_boxes","text":"","code":"layer_max_num_bounding_boxes(object, max_number, fill_value = -1L, ...)"},{"path":"https://keras3.posit.co/dev/reference/layer_max_num_bounding_boxes.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Ensure the maximum number of bounding boxes. — layer_max_num_bounding_boxes","text":"object Object compose layer . tensor, array, sequential model. max_number Desired output number bounding boxes. fill_value fill value boxes labels bounding_boxes. Defaults -1. ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_max_num_bounding_boxes.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Ensure the maximum number of bounding boxes. — layer_max_num_bounding_boxes","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_max_pooling_1d.html","id":null,"dir":"Reference","previous_headings":"","what":"Max pooling operation for 1D temporal data. — layer_max_pooling_1d","title":"Max pooling operation for 1D temporal data. — layer_max_pooling_1d","text":"Downsamples input representation taking maximum value spatial window size pool_size. window shifted strides. resulting output using \"valid\" padding option shape : output_shape = (input_shape - pool_size + 1) / strides). resulting output shape using \"\" padding option : output_shape = input_shape / strides","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_max_pooling_1d.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Max pooling operation for 1D temporal data. — layer_max_pooling_1d","text":"","code":"layer_max_pooling_1d(   object,   pool_size = 2L,   strides = NULL,   padding = \"valid\",   data_format = NULL,   name = NULL,   ... )"},{"path":"https://keras3.posit.co/dev/reference/layer_max_pooling_1d.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Max pooling operation for 1D temporal data. — layer_max_pooling_1d","text":"object Object compose layer . tensor, array, sequential model. pool_size int, size max pooling window. strides int NULL. Specifies much pooling window moves pooling step. NULL, default pool_size. padding string, either \"valid\" \"\" (case-insensitive). \"valid\" means padding. \"\" results padding evenly left/right /input output height/width dimension input. data_format string, either \"channels_last\" \"channels_first\". ordering dimensions inputs. \"channels_last\" corresponds inputs shape (batch, steps, features) \"channels_first\" corresponds inputs shape (batch, features, steps). defaults image_data_format value found Keras config file ~/.keras/keras.json. never set , \"channels_last\". name String, name object ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_max_pooling_1d.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Max pooling operation for 1D temporal data. — layer_max_pooling_1d","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_max_pooling_1d.html","id":"input-shape","dir":"Reference","previous_headings":"","what":"Input Shape","title":"Max pooling operation for 1D temporal data. — layer_max_pooling_1d","text":"data_format=\"channels_last\": 3D tensor shape (batch_size, steps, features). data_format=\"channels_first\": 3D tensor shape (batch_size, features, steps).","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_max_pooling_1d.html","id":"output-shape","dir":"Reference","previous_headings":"","what":"Output Shape","title":"Max pooling operation for 1D temporal data. — layer_max_pooling_1d","text":"data_format=\"channels_last\": 3D tensor shape (batch_size, downsampled_steps, features). data_format=\"channels_first\": 3D tensor shape (batch_size, features, downsampled_steps).","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_max_pooling_1d.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Max pooling operation for 1D temporal data. — layer_max_pooling_1d","text":"strides=1 padding=\"valid\":     strides=2 padding=\"valid\":     strides=1 padding=\"\":","code":"x <- op_reshape(c(1, 2, 3, 4, 5),                c(1, 5, 1)) max_pool_1d <- layer_max_pooling_1d(pool_size = 2,                                     strides = 1,                                     padding = \"valid\") max_pool_1d(x) ## tf.Tensor( ## [[[2.] ##   [3.] ##   [4.] ##   [5.]]], shape=(1, 4, 1), dtype=float32) x <- op_reshape(c(1, 2, 3, 4, 5),                c(1, 5, 1)) max_pool_1d <- layer_max_pooling_1d(pool_size = 2,                                     strides = 2,                                     padding = \"valid\") max_pool_1d(x) ## tf.Tensor( ## [[[2.] ##   [4.]]], shape=(1, 2, 1), dtype=float32) x <- op_reshape(c(1, 2, 3, 4, 5),                c(1, 5, 1)) max_pool_1d <- layer_max_pooling_1d(pool_size = 2,                                     strides = 1,                                     padding = \"same\") max_pool_1d(x) ## tf.Tensor( ## [[[2.] ##   [3.] ##   [4.] ##   [5.] ##   [5.]]], shape=(1, 5, 1), dtype=float32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_max_pooling_2d.html","id":null,"dir":"Reference","previous_headings":"","what":"Max pooling operation for 2D spatial data. — layer_max_pooling_2d","title":"Max pooling operation for 2D spatial data. — layer_max_pooling_2d","text":"Downsamples input along spatial dimensions (height width) taking maximum value input window (size defined pool_size) channel input. window shifted strides along dimension. resulting output using \"valid\" padding option spatial shape (number rows columns) : output_shape = floor((input_shape - pool_size) / strides) + 1 (input_shape >= pool_size) resulting output shape using \"\" padding option : output_shape = floor((input_shape - 1) / strides) + 1","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_max_pooling_2d.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Max pooling operation for 2D spatial data. — layer_max_pooling_2d","text":"","code":"layer_max_pooling_2d(   object,   pool_size = list(2L, 2L),   strides = NULL,   padding = \"valid\",   data_format = NULL,   name = NULL,   ... )"},{"path":"https://keras3.posit.co/dev/reference/layer_max_pooling_2d.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Max pooling operation for 2D spatial data. — layer_max_pooling_2d","text":"object Object compose layer . tensor, array, sequential model. pool_size int list 2 integers, factors downscale (dim1, dim2). one integer specified, window length used dimensions. strides int list 2 integers, NULL. Strides values. NULL, default pool_size. one int specified, stride size used dimensions. padding string, either \"valid\" \"\" (case-insensitive). \"valid\" means padding. \"\" results padding evenly left/right /input output height/width dimension input. data_format string, either \"channels_last\" \"channels_first\". ordering dimensions inputs. \"channels_last\" corresponds inputs shape (batch, height, width, channels) \"channels_first\" corresponds inputs shape (batch, channels, height, width). defaults image_data_format value found Keras config file ~/.keras/keras.json. never set , \"channels_last\". name String, name object ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_max_pooling_2d.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Max pooling operation for 2D spatial data. — layer_max_pooling_2d","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_max_pooling_2d.html","id":"input-shape","dir":"Reference","previous_headings":"","what":"Input Shape","title":"Max pooling operation for 2D spatial data. — layer_max_pooling_2d","text":"data_format=\"channels_last\": 4D tensor shape (batch_size, height, width, channels). data_format=\"channels_first\": 4D tensor shape (batch_size, channels, height, width).","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_max_pooling_2d.html","id":"output-shape","dir":"Reference","previous_headings":"","what":"Output Shape","title":"Max pooling operation for 2D spatial data. — layer_max_pooling_2d","text":"data_format=\"channels_last\": 4D tensor shape (batch_size, pooled_height, pooled_width, channels). data_format=\"channels_first\": 4D tensor shape (batch_size, channels, pooled_height, pooled_width).","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_max_pooling_2d.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Max pooling operation for 2D spatial data. — layer_max_pooling_2d","text":"strides = (1, 1) padding = \"valid\":     strides = c(2, 2) padding = \"valid\":     stride = (1, 1) padding = \"\":","code":"x <- rbind(c(1., 2., 3.),            c(4., 5., 6.),            c(7., 8., 9.)) |> op_reshape(c(1, 3, 3, 1)) max_pool_2d <- layer_max_pooling_2d(pool_size = c(2, 2),                                     strides = c(1, 1),                                     padding = \"valid\") max_pool_2d(x) ## tf.Tensor( ## [[[[5.] ##    [6.]] ## ##   [[8.] ##    [9.]]]], shape=(1, 2, 2, 1), dtype=float32) x <- rbind(c(1., 2., 3., 4.),            c(5., 6., 7., 8.),            c(9., 10., 11., 12.)) |> op_reshape(c(1, 3, 4, 1)) max_pool_2d <- layer_max_pooling_2d(pool_size = c(2, 2),                                     strides = c(2, 2),                                     padding = \"valid\") max_pool_2d(x) ## tf.Tensor( ## [[[[6.] ##    [8.]]]], shape=(1, 1, 2, 1), dtype=float32) x <- rbind(c(1., 2., 3.),            c(4., 5., 6.),            c(7., 8., 9.)) |> op_reshape(c(1, 3, 3, 1)) max_pool_2d <- layer_max_pooling_2d(pool_size = c(2, 2),                                     strides = c(1, 1),                                     padding = \"same\") max_pool_2d(x) ## tf.Tensor( ## [[[[5.] ##    [6.] ##    [6.]] ## ##   [[8.] ##    [9.] ##    [9.]] ## ##   [[8.] ##    [9.] ##    [9.]]]], shape=(1, 3, 3, 1), dtype=float32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_max_pooling_3d.html","id":null,"dir":"Reference","previous_headings":"","what":"Max pooling operation for 3D data (spatial or spatio-temporal). — layer_max_pooling_3d","title":"Max pooling operation for 3D data (spatial or spatio-temporal). — layer_max_pooling_3d","text":"Downsamples input along spatial dimensions (depth, height, width) taking maximum value input window (size defined pool_size) channel input. window shifted strides along dimension.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_max_pooling_3d.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Max pooling operation for 3D data (spatial or spatio-temporal). — layer_max_pooling_3d","text":"","code":"layer_max_pooling_3d(   object,   pool_size = list(2L, 2L, 2L),   strides = NULL,   padding = \"valid\",   data_format = NULL,   name = NULL,   ... )"},{"path":"https://keras3.posit.co/dev/reference/layer_max_pooling_3d.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Max pooling operation for 3D data (spatial or spatio-temporal). — layer_max_pooling_3d","text":"object Object compose layer . tensor, array, sequential model. pool_size int list 3 integers, factors downscale (dim1, dim2, dim3). one integer specified, window length used dimensions. strides int list 3 integers, NULL. Strides values. NULL, default pool_size. one int specified, stride size used dimensions. padding string, either \"valid\" \"\" (case-insensitive). \"valid\" means padding. \"\" results padding evenly left/right /input output height/width dimension input. data_format string, either \"channels_last\" \"channels_first\". ordering dimensions inputs. \"channels_last\" corresponds inputs shape (batch, spatial_dim1, spatial_dim2, spatial_dim3, channels) \"channels_first\" corresponds inputs shape (batch, channels, spatial_dim1, spatial_dim2, spatial_dim3). defaults image_data_format value found Keras config file ~/.keras/keras.json. never set , \"channels_last\". name String, name object ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_max_pooling_3d.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Max pooling operation for 3D data (spatial or spatio-temporal). — layer_max_pooling_3d","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_max_pooling_3d.html","id":"input-shape","dir":"Reference","previous_headings":"","what":"Input Shape","title":"Max pooling operation for 3D data (spatial or spatio-temporal). — layer_max_pooling_3d","text":"data_format=\"channels_last\": 5D tensor shape: (batch_size, spatial_dim1, spatial_dim2, spatial_dim3, channels) data_format=\"channels_first\": 5D tensor shape: (batch_size, channels, spatial_dim1, spatial_dim2, spatial_dim3)","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_max_pooling_3d.html","id":"output-shape","dir":"Reference","previous_headings":"","what":"Output Shape","title":"Max pooling operation for 3D data (spatial or spatio-temporal). — layer_max_pooling_3d","text":"data_format=\"channels_last\": 5D tensor shape: (batch_size, pooled_dim1, pooled_dim2, pooled_dim3, channels) data_format=\"channels_first\": 5D tensor shape: (batch_size, channels, pooled_dim1, pooled_dim2, pooled_dim3)","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_max_pooling_3d.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Max pooling operation for 3D data (spatial or spatio-temporal). — layer_max_pooling_3d","text":"","code":"depth <- 30 height <- 30 width <- 30 channels <- 3  inputs <- layer_input(shape=c(depth, height, width, channels)) layer <- layer_max_pooling_3d(pool_size=3) outputs <- inputs |> layer() outputs ## <KerasTensor shape=(None, 10, 10, 10, 3), dtype=float32, sparse=False, ragged=False, name=keras_tensor_1>"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_maximum.html","id":null,"dir":"Reference","previous_headings":"","what":"Computes element-wise maximum on a list of inputs. — layer_maximum","title":"Computes element-wise maximum on a list of inputs. — layer_maximum","text":"takes input list tensors, shape, returns single tensor (also shape).","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_maximum.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes element-wise maximum on a list of inputs. — layer_maximum","text":"","code":"layer_maximum(inputs, ...)"},{"path":"https://keras3.posit.co/dev/reference/layer_maximum.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Computes element-wise maximum on a list of inputs. — layer_maximum","text":"inputs layers combine ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_maximum.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Computes element-wise maximum on a list of inputs. — layer_maximum","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_maximum.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Computes element-wise maximum on a list of inputs. — layer_maximum","text":"Usage Keras model:","code":"input_shape <- c(2, 3, 4) x1 <- random_uniform(input_shape) x2 <- random_uniform(input_shape) y <- layer_maximum(x1, x2) input1 <- layer_input(shape = c(16)) x1 <- input1 |> layer_dense(8, activation = 'relu') input2 <- layer_input(shape = c(32)) x2 <- input2 |> layer_dense(8, activation = 'relu') # equivalent to `y <- layer_maximum(x1, x2)` y <- layer_maximum(x1, x2) out <- y |> layer_dense(4) model <- keras_model(inputs = c(input1, input2), outputs = out)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_mel_spectrogram.html","id":null,"dir":"Reference","previous_headings":"","what":"A preprocessing layer to convert raw audio signals to Mel spectrograms. — layer_mel_spectrogram","title":"A preprocessing layer to convert raw audio signals to Mel spectrograms. — layer_mel_spectrogram","text":"layer takes float32/float64 single batched audio signal inputs computes Mel spectrogram using Short-Time Fourier Transform Mel scaling. input 1D (unbatched) 2D (batched) tensor representing audio signals. output 2D 3D tensor representing Mel spectrograms. spectrogram image-like representation shows frequency spectrum signal time. uses x-axis represent time, y-axis represent frequency, pixel represent intensity. Mel spectrograms special type spectrogram use mel scale, approximates humans perceive sound. commonly used speech music processing tasks like speech recognition, speaker identification, music genre classification.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_mel_spectrogram.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"A preprocessing layer to convert raw audio signals to Mel spectrograms. — layer_mel_spectrogram","text":"","code":"layer_mel_spectrogram(   object,   fft_length = 2048L,   sequence_stride = 512L,   sequence_length = NULL,   window = \"hann\",   sampling_rate = 16000L,   num_mel_bins = 128L,   min_freq = 20,   max_freq = NULL,   power_to_db = TRUE,   top_db = 80,   mag_exp = 2,   min_power = 1e-10,   ref_power = 1,   ... )"},{"path":"https://keras3.posit.co/dev/reference/layer_mel_spectrogram.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"A preprocessing layer to convert raw audio signals to Mel spectrograms. — layer_mel_spectrogram","text":"object Object compose layer . tensor, array, sequential model. fft_length Integer, size FFT window. sequence_stride Integer, number samples successive STFT columns. sequence_length Integer, size window used applying window audio frame. NULL, defaults fft_length. window String, name window function use. Available values \"hann\" \"hamming\". window tensor, used directly window length must sequence_length. window NULL, windowing used. Defaults \"hann\". sampling_rate Integer, sample rate input signal. num_mel_bins Integer, number mel bins generate. min_freq Float, minimum frequency mel bins. max_freq Float, maximum frequency mel bins. NULL, defaults sampling_rate / 2. power_to_db TRUE, convert power spectrogram decibels. top_db Float, minimum negative cut-max(10 * log10(S)) - top_db. mag_exp Float, exponent magnitude spectrogram. 1 magnitude, 2 power, etc. Default 2. min_power Float, minimum value power ref_power. ref_power Float, power scaled relative 10 * log10(S / ref_power). ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_mel_spectrogram.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"A preprocessing layer to convert raw audio signals to Mel spectrograms. — layer_mel_spectrogram","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_mel_spectrogram.html","id":"references","dir":"Reference","previous_headings":"","what":"References","title":"A preprocessing layer to convert raw audio signals to Mel spectrograms. — layer_mel_spectrogram","text":"Spectrogram, Mel scale.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_mel_spectrogram.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"A preprocessing layer to convert raw audio signals to Mel spectrograms. — layer_mel_spectrogram","text":"Unbatched audio signal   Batched audio signal","code":"layer <- layer_mel_spectrogram(   num_mel_bins = 64,   sampling_rate = 8000,   sequence_stride = 256,   fft_length = 2048 ) layer(random_uniform(shape = c(16000))) |> shape() layer <- layer_mel_spectrogram(   num_mel_bins = 80,   sampling_rate = 8000,   sequence_stride = 128,   fft_length = 2048 ) layer(random_uniform(shape = c(2, 16000))) |> shape()"},{"path":"https://keras3.posit.co/dev/reference/layer_mel_spectrogram.html","id":"input-shape","dir":"Reference","previous_headings":"","what":"Input Shape","title":"A preprocessing layer to convert raw audio signals to Mel spectrograms. — layer_mel_spectrogram","text":"1D (unbatched) 2D (batched) tensor shape:(..., samples).","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_mel_spectrogram.html","id":"output-shape","dir":"Reference","previous_headings":"","what":"Output Shape","title":"A preprocessing layer to convert raw audio signals to Mel spectrograms. — layer_mel_spectrogram","text":"2D (unbatched) 3D (batched) tensor shape:(..., num_mel_bins, time).","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_minimum.html","id":null,"dir":"Reference","previous_headings":"","what":"Computes elementwise minimum on a list of inputs. — layer_minimum","title":"Computes elementwise minimum on a list of inputs. — layer_minimum","text":"takes input list tensors, shape, returns single tensor (also shape).","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_minimum.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes elementwise minimum on a list of inputs. — layer_minimum","text":"","code":"layer_minimum(inputs, ...)"},{"path":"https://keras3.posit.co/dev/reference/layer_minimum.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Computes elementwise minimum on a list of inputs. — layer_minimum","text":"inputs layers combine ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_minimum.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Computes elementwise minimum on a list of inputs. — layer_minimum","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_minimum.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Computes elementwise minimum on a list of inputs. — layer_minimum","text":"Usage Keras model:","code":"input_shape <- c(2, 3, 4) x1 <- random_uniform(input_shape) x2 <- random_uniform(input_shape) y <- layer_minimum(x1, x2) input1 <- layer_input(shape = c(16)) x1 <- input1 |> layer_dense(8, activation = 'relu') input2 <- layer_input(shape = c(32)) x2 <- input2 |> layer_dense(8, activation = 'relu') # equivalent to `y <- layer_minimum(x1, x2)` y <- layer_minimum(x1, x2) out <- y |> layer_dense(4) model <- keras_model(inputs = c(input1, input2), outputs = out)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_mix_up.html","id":null,"dir":"Reference","previous_headings":"","what":"MixUp implements the MixUp data augmentation technique. — layer_mix_up","title":"MixUp implements the MixUp data augmentation technique. — layer_mix_up","text":"MixUp implements MixUp data augmentation technique.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_mix_up.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"MixUp implements the MixUp data augmentation technique. — layer_mix_up","text":"","code":"layer_mix_up(object, alpha = 0.2, data_format = NULL, seed = NULL, ...)"},{"path":"https://keras3.posit.co/dev/reference/layer_mix_up.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"MixUp implements the MixUp data augmentation technique. — layer_mix_up","text":"object Object compose layer . tensor, array, sequential model. alpha Float 0 1. Controls blending strength. Smaller values mean less mixing, larger values allow  blending images. Defaults 0.2, recommended ImageNet1k classification. data_format String, one \"channels_last\" (default) \"channels_first\". ordering dimensions inputs. \"channels_last\" corresponds inputs shape (batch, height, width, channels) \"channels_first\" corresponds inputs shape (batch, channels, height, width). seed Integer. Used create random seed. ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_mix_up.html","id":"references","dir":"Reference","previous_headings":"","what":"References","title":"MixUp implements the MixUp data augmentation technique. — layer_mix_up","text":"MixUp paper. MixUp Object Detection paper.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_mix_up.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"MixUp implements the MixUp data augmentation technique. — layer_mix_up","text":"","code":"c(c(images, labels), .) %<-% dataset_cifar10() c(images, labels) %<-% list(images[1:8,,,], labels[1:8,]) labels <- labels |> op_one_hot(10) |> op_cast(\"float32\") mix_up <- layer_mix_up(alpha=0.2) output <- mix_up(list(images = images, labels = labels))"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_multi_head_attention.html","id":null,"dir":"Reference","previous_headings":"","what":"Multi Head Attention layer. — layer_multi_head_attention","title":"Multi Head Attention layer. — layer_multi_head_attention","text":"implementation multi-headed attention described paper \"Attention Need\" Vaswani et al., 2017. query, key, value , self-attention. timestep query attends corresponding sequence key, returns fixed-width vector. layer first projects query, key value. (effectively) list tensors length num_attention_heads, corresponding shapes (batch_size, <query dimensions>, key_dim), (batch_size, <key/value dimensions>, key_dim), (batch_size, <key/value dimensions>, value_dim). , query key tensors dot-producted scaled. softmaxed obtain attention probabilities. value tensors interpolated probabilities, concatenated back single tensor. Finally, result tensor last dimension value_dim can take linear projection return.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_multi_head_attention.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Multi Head Attention layer. — layer_multi_head_attention","text":"","code":"layer_multi_head_attention(   inputs,   num_heads,   key_dim,   value_dim = NULL,   dropout = 0,   use_bias = TRUE,   output_shape = NULL,   attention_axes = NULL,   flash_attention = NULL,   kernel_initializer = \"glorot_uniform\",   bias_initializer = \"zeros\",   kernel_regularizer = NULL,   bias_regularizer = NULL,   activity_regularizer = NULL,   kernel_constraint = NULL,   bias_constraint = NULL,   seed = NULL,   ... )"},{"path":"https://keras3.posit.co/dev/reference/layer_multi_head_attention.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Multi Head Attention layer. — layer_multi_head_attention","text":"inputs see description num_heads Number attention heads. key_dim Size attention head query key. value_dim Size attention head value. dropout Dropout probability. use_bias Boolean, whether dense layers use bias vectors/matrices. output_shape expected shape output tensor, besides batch sequence dims. specified, projects back query feature dim (query input's last dimension). attention_axes axes attention applied. NULL means attention axes, batch, heads, features. flash_attention NULL, layer attempts use flash attention faster memory-efficient attention computations possible. behavior can configured using config_enable_flash_attention() config_disable_flash_attention(). kernel_initializer Initializer dense layer kernels. bias_initializer Initializer dense layer biases. kernel_regularizer Regularizer dense layer kernels. bias_regularizer Regularizer dense layer biases. activity_regularizer Regularizer dense layer activity. kernel_constraint Constraint dense layer kernels. bias_constraint Constraint dense layer kernels. seed Optional integer seed dropout layer. ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_multi_head_attention.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Multi Head Attention layer. — layer_multi_head_attention","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_multi_head_attention.html","id":"call-arguments","dir":"Reference","previous_headings":"","what":"Call Arguments","title":"Multi Head Attention layer. — layer_multi_head_attention","text":"query: Query tensor shape (B, T, dim), B batch size, T target sequence length, dim feature dimension. value: Value tensor shape (B, S, dim), B batch size, S source sequence length, dim feature dimension. key: Optional key tensor shape (B, S, dim). given, use value key value, common case. attention_mask: boolean mask shape (B, T, S), prevents attention certain positions. boolean mask specifies query elements can attend key elements, 1 indicates attention 0 indicates attention. Broadcasting can happen missing batch dimensions head dimension. return_attention_scores: boolean indicate whether output (attention_output, attention_scores) TRUE, attention_output FALSE. Defaults FALSE. training: Python boolean indicating whether layer behave training mode (adding dropout) inference mode (dropout). go either using training mode parent layer/model, FALSE (inference) parent layer. use_causal_mask: boolean indicate whether apply causal mask prevent tokens attending future tokens (e.g., used decoder Transformer).","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_multi_head_attention.html","id":"call-return","dir":"Reference","previous_headings":"","what":"Call return","title":"Multi Head Attention layer. — layer_multi_head_attention","text":"attention_output: result computation, shape (B, T, E), T target sequence shapes E query input last dimension output_shape NULL. Otherwise, multi-head outputs projected shape specified output_shape. attention_scores: (Optional) multi-head attention coefficients attention axes.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_multi_head_attention.html","id":"properties","dir":"Reference","previous_headings":"","what":"Properties","title":"Multi Head Attention layer. — layer_multi_head_attention","text":"MultiHeadAttention Layer instance following additional read-properties: attention_axes dropout key_dense key_dim num_heads output_dense output_shape query_dense use_bias value_dense value_dim","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_multiply.html","id":null,"dir":"Reference","previous_headings":"","what":"Performs elementwise multiplication. — layer_multiply","title":"Performs elementwise multiplication. — layer_multiply","text":"takes input list tensors, shape, returns single tensor (also shape).","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_multiply.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Performs elementwise multiplication. — layer_multiply","text":"","code":"layer_multiply(inputs, ...)"},{"path":"https://keras3.posit.co/dev/reference/layer_multiply.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Performs elementwise multiplication. — layer_multiply","text":"inputs layers combine ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_multiply.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Performs elementwise multiplication. — layer_multiply","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_multiply.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Performs elementwise multiplication. — layer_multiply","text":"Usage Keras model:","code":"input_shape <- c(2, 3, 4) x1 <- random_uniform(input_shape) x2 <- random_uniform(input_shape) y <- layer_multiply(x1, x2) input1 <- layer_input(shape = c(16)) x1 <- input1 |> layer_dense(8, activation = 'relu') input2 <- layer_input(shape = c(32)) x2 <- input2 |> layer_dense(8, activation = 'relu') # equivalent to `y <- layer_multiply(x1, x2)` y <- layer_multiply(x1, x2) out <- y |> layer_dense(4) model <- keras_model(inputs = c(input1, input2), outputs = out)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_normalization.html","id":null,"dir":"Reference","previous_headings":"","what":"A preprocessing layer that normalizes continuous features. — layer_normalization","title":"A preprocessing layer that normalizes continuous features. — layer_normalization","text":"layer shift scale inputs distribution centered around 0 standard deviation 1. accomplishes precomputing mean variance data, calling (input - mean) / sqrt(var) runtime. mean variance values layer must either supplied construction learned via adapt(). adapt() compute mean variance data store layer's weights. adapt() called fit(), evaluate(), predict().","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_normalization.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"A preprocessing layer that normalizes continuous features. — layer_normalization","text":"","code":"layer_normalization(   object,   axis = -1L,   mean = NULL,   variance = NULL,   invert = FALSE,   ... )"},{"path":"https://keras3.posit.co/dev/reference/layer_normalization.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"A preprocessing layer that normalizes continuous features. — layer_normalization","text":"object Object compose layer . tensor, array, sequential model. axis Integer, list integers, NULL. axis axes separate mean variance index shape. example, shape (NULL, 5) axis=1, layer track 5 separate mean variance values last axis. axis set NULL, layer normalize elements input scalar mean variance. -1, last axis input assumed feature dimension normalized per index. Note specific case batched scalar inputs axis batch axis, default normalize index batch separately. case, consider passing axis=NULL. Defaults -1. mean mean value(s) use normalization. passed value(s) broadcast shape kept axes ; value(s) broadcast, error raised layer's build() method called. variance variance value(s) use normalization. passed value(s) broadcast shape kept axes ; value(s) broadcast, error raised layer's build() method called. invert TRUE, layer apply inverse transformation inputs: turn normalized input back original form. ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_normalization.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"A preprocessing layer that normalizes continuous features. — layer_normalization","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_normalization.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"A preprocessing layer that normalizes continuous features. — layer_normalization","text":"Calculate global mean variance analyzing dataset adapt().     Calculate mean variance index last axis.     Pass mean variance directly.     Use layer de-normalize inputs (adapting layer).","code":"adapt_data <- op_array(c(1., 2., 3., 4., 5.), dtype='float32') input_data <- op_array(c(1., 2., 3.), dtype='float32') layer <- layer_normalization(axis = NULL) layer %>% adapt(adapt_data) layer(input_data) ## tf.Tensor([-1.4142135  -0.70710677  0.        ], shape=(3), dtype=float32) adapt_data <- op_array(rbind(c(0., 7., 4.),                        c(2., 9., 6.),                        c(0., 7., 4.),                        c(2., 9., 6.)), dtype='float32') input_data <- op_array(matrix(c(0., 7., 4.), nrow = 1), dtype='float32') layer <- layer_normalization(axis=-1) layer %>% adapt(adapt_data) layer(input_data) ## tf.Tensor([[-1. -1. -1.]], shape=(1, 3), dtype=float32) input_data <- op_array(rbind(1, 2, 3), dtype='float32') layer <- layer_normalization(mean=3., variance=2.) layer(input_data) ## tf.Tensor( ## [[-1.4142135 ] ##  [-0.70710677] ##  [ 0.        ]], shape=(3, 1), dtype=float32) adapt_data <- op_array(rbind(c(0., 7., 4.),                        c(2., 9., 6.),                        c(0., 7., 4.),                        c(2., 9., 6.)), dtype='float32') input_data <- op_array(c(1., 2., 3.), dtype='float32') layer <- layer_normalization(axis=-1, invert=TRUE) layer %>% adapt(adapt_data) layer(input_data) ## tf.Tensor([[ 2. 10.  8.]], shape=(1, 3), dtype=float32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_permute.html","id":null,"dir":"Reference","previous_headings":"","what":"Permutes the dimensions of the input according to a given pattern. — layer_permute","title":"Permutes the dimensions of the input according to a given pattern. — layer_permute","text":"Useful e.g. connecting RNNs convnets.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_permute.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Permutes the dimensions of the input according to a given pattern. — layer_permute","text":"","code":"layer_permute(object, dims, ...)"},{"path":"https://keras3.posit.co/dev/reference/layer_permute.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Permutes the dimensions of the input according to a given pattern. — layer_permute","text":"object Object compose layer . tensor, array, sequential model. dims List integers. Permutation pattern include batch dimension. Indexing starts 1. instance, (1, 3, 2) permutes second third dimensions input. ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_permute.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Permutes the dimensions of the input according to a given pattern. — layer_permute","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_permute.html","id":"input-shape","dir":"Reference","previous_headings":"","what":"Input Shape","title":"Permutes the dimensions of the input according to a given pattern. — layer_permute","text":"Arbitrary.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_permute.html","id":"output-shape","dir":"Reference","previous_headings":"","what":"Output Shape","title":"Permutes the dimensions of the input according to a given pattern. — layer_permute","text":"input shape, dimensions re-ordered according specified pattern.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_permute.html","id":"example","dir":"Reference","previous_headings":"","what":"Example","title":"Permutes the dimensions of the input according to a given pattern. — layer_permute","text":"","code":"x <- layer_input(shape=c(10, 64)) y <- layer_permute(x, c(2, 1)) shape(y) ## shape(NA, 64, 10)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_pipeline.html","id":null,"dir":"Reference","previous_headings":"","what":"Applies a series of layers to an input. — layer_pipeline","title":"Applies a series of layers to an input. — layer_pipeline","text":"class useful build preprocessing pipeline, particular image data augmentation pipeline. Compared Sequential model, Pipeline features important differences: Model, just plain layer. layers pipeline compatible tf.data, pipeline also remain tf.data compatible. say, pipeline attempt convert inputs backend-native tensors tf.data context (unlike Sequential model).","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_pipeline.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Applies a series of layers to an input. — layer_pipeline","text":"","code":"layer_pipeline(layers, name = NULL)"},{"path":"https://keras3.posit.co/dev/reference/layer_pipeline.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Applies a series of layers to an input. — layer_pipeline","text":"layers list layers. name String, name object","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_pipeline.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Applies a series of layers to an input. — layer_pipeline","text":"","code":"preprocessing_pipeline <- layer_pipeline(c(   layer_auto_contrast(, ),   layer_random_zoom(, 0.2),   layer_random_rotation(, 0.2) ))  # `ds` is a tf.data.Dataset of images ds <- tfdatasets::tensor_slices_dataset(1:100) |>   tfdatasets::dataset_map(\\(.x) {     random_normal(c(28, 28))   }) |>   tfdatasets::dataset_batch(32)   #|>   # tfdatasets::dataset_take(4) |>   # iterate() |> str()  preprocessed_ds <- ds |>   tfdatasets::dataset_map(preprocessing_pipeline, num_parallel_calls = 4)"},{"path":"https://keras3.posit.co/dev/reference/layer_rand_augment.html","id":null,"dir":"Reference","previous_headings":"","what":"RandAugment performs the Rand Augment operation on input images. — layer_rand_augment","title":"RandAugment performs the Rand Augment operation on input images. — layer_rand_augment","text":"layer can thought --one image augmentation layer. policy implemented layer benchmarked extensively effective wide variety datasets.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_rand_augment.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"RandAugment performs the Rand Augment operation on input images. — layer_rand_augment","text":"","code":"layer_rand_augment(   object,   value_range = list(0L, 255L),   num_ops = 2L,   factor = 0.5,   interpolation = \"bilinear\",   seed = NULL,   data_format = NULL,   ... )"},{"path":"https://keras3.posit.co/dev/reference/layer_rand_augment.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"RandAugment performs the Rand Augment operation on input images. — layer_rand_augment","text":"object Object compose layer . tensor, array, sequential model. value_range range values input image can take. Default (0, 255). Typically, (0, 1) normalized images (0, 255) raw images. num_ops number augmentation operations apply sequentially image. Default 2. factor strength augmentation normalized value 0 1. Default 0.5. interpolation interpolation method use resizing operations. Options include nearest, bilinear. Default bilinear. seed Integer. Used create random seed. data_format String, one \"channels_last\" (default) \"channels_first\". ordering dimensions inputs. \"channels_last\" corresponds inputs shape (batch, height, width, channels) \"channels_first\" corresponds inputs shape (batch, channels, height, width). ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_rand_augment.html","id":"references","dir":"Reference","previous_headings":"","what":"References","title":"RandAugment performs the Rand Augment operation on input images. — layer_rand_augment","text":"RandAugment","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_random_brightness.html","id":null,"dir":"Reference","previous_headings":"","what":"A preprocessing layer which randomly adjusts brightness during training. — layer_random_brightness","title":"A preprocessing layer which randomly adjusts brightness during training. — layer_random_brightness","text":"layer randomly increase/reduce brightness input RGB images. inference time, output identical input. Call layer training=TRUE adjust brightness input. Note: layer safe use inside tf.data pipeline (independently backend using).","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_random_brightness.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"A preprocessing layer which randomly adjusts brightness during training. — layer_random_brightness","text":"","code":"layer_random_brightness(   object,   factor,   value_range = list(0L, 255L),   seed = NULL,   ... )"},{"path":"https://keras3.posit.co/dev/reference/layer_random_brightness.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"A preprocessing layer which randomly adjusts brightness during training. — layer_random_brightness","text":"object Object compose layer . tensor, array, sequential model. factor Float list 2 floats -1.0 1.0. factor used determine lower bound upper bound brightness adjustment. float value chosen randomly limits. -1.0 chosen, output image black, 1.0 chosen, image fully white. one float provided, eg, 0.2, -0.2 used lower bound 0.2 used upper bound. value_range Optional list 2 floats lower upper limit values input data. make change, use c(0.0, 1.0), e.g., image input scaled layer. Defaults c(0.0, 255.0). brightness adjustment scaled range, output values clipped range. seed optional integer, fixed RNG behavior. ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_random_brightness.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"A preprocessing layer which randomly adjusts brightness during training. — layer_random_brightness","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_random_brightness.html","id":"inputs","dir":"Reference","previous_headings":"","what":"Inputs","title":"A preprocessing layer which randomly adjusts brightness during training. — layer_random_brightness","text":"3D (HWC) 4D (NHWC) tensor, float int dtype. Input pixel values can range (e.g. [0., 1.) [0, 255])","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_random_brightness.html","id":"output","dir":"Reference","previous_headings":"","what":"Output","title":"A preprocessing layer which randomly adjusts brightness during training. — layer_random_brightness","text":"3D (HWC) 4D (NHWC) tensor brightness adjusted based factor. default, layer output floats. output value clipped range [0, 255], valid range RGB colors, rescaled based value_range needed.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_random_brightness.html","id":"example","dir":"Reference","previous_headings":"","what":"Example","title":"A preprocessing layer which randomly adjusts brightness during training. — layer_random_brightness","text":"","code":"random_bright <- layer_random_brightness(factor=0.2, seed = 1)  # An image with shape [2, 2, 3] image <- array(1:12, dim=c(2, 2, 3))  # Assume we randomly select the factor to be 0.1, then it will apply # 0.1 * 255 to all the channel output <- random_bright(image, training=TRUE) output ## tf.Tensor( ## [[[39 43 47] ##   [41 45 49]] ## ##  [[40 44 48] ##   [42 46 50]]], shape=(2, 2, 3), dtype=int32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_random_color_degeneration.html","id":null,"dir":"Reference","previous_headings":"","what":"Randomly performs the color degeneration operation on given images. — layer_random_color_degeneration","title":"Randomly performs the color degeneration operation on given images. — layer_random_color_degeneration","text":"sharpness operation first converts image gray scale, back color. takes weighted average original image degenerated image. makes colors appear dull.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_random_color_degeneration.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Randomly performs the color degeneration operation on given images. — layer_random_color_degeneration","text":"","code":"layer_random_color_degeneration(   object,   factor,   value_range = list(0L, 255L),   data_format = NULL,   seed = NULL,   ... )"},{"path":"https://keras3.posit.co/dev/reference/layer_random_color_degeneration.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Randomly performs the color degeneration operation on given images. — layer_random_color_degeneration","text":"object Object compose layer . tensor, array, sequential model. factor tuple two floats single float. factor controls extent image sharpness impacted. factor=0.0 makes layer perform -op operation, value 1.0 uses degenerated result entirely. Values 0 1 result linear interpolation original image sharpened image. Values 0.0 1.0. tuple used, factor sampled two values every image augmented. single float used, value 0.0 passed float sampled. order ensure value always , please pass tuple two identical floats: (0.5, 0.5). value_range range values input image can take. Default (0, 255). Typically, (0, 1) normalized images (0, 255) raw images. data_format String, one \"channels_last\" (default) \"channels_first\". ordering dimensions inputs. \"channels_last\" corresponds inputs shape (batch, height, width, channels) \"channels_first\" corresponds inputs shape (batch, channels, height, width). seed Integer. Used create random seed. ... forward/backward compatability.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_random_color_jitter.html","id":null,"dir":"Reference","previous_headings":"","what":"Randomly apply brightness, contrast, saturation — layer_random_color_jitter","title":"Randomly apply brightness, contrast, saturation — layer_random_color_jitter","text":"hue image processing operation sequentially randomly input.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_random_color_jitter.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Randomly apply brightness, contrast, saturation — layer_random_color_jitter","text":"","code":"layer_random_color_jitter(   object,   value_range = list(0L, 255L),   brightness_factor = NULL,   contrast_factor = NULL,   saturation_factor = NULL,   hue_factor = NULL,   seed = NULL,   data_format = NULL,   ... )"},{"path":"https://keras3.posit.co/dev/reference/layer_random_color_jitter.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Randomly apply brightness, contrast, saturation — layer_random_color_jitter","text":"object Object compose layer . tensor, array, sequential model. value_range range values incoming images . Represented two number tuple written [low, high]. typically either [0, 1] [0, 255] depending preprocessing pipeline set . brightness_factor Float list/tuple 2 floats -1.0 1.0. factor used determine lower bound upper bound brightness adjustment. float value chosen randomly limits. -1.0 chosen, output image black, 1.0 chosen, image fully white. one float provided, eg, 0.2, -0.2 used lower bound 0.2 used upper bound. contrast_factor positive float represented fraction value, tuple size 2 representing lower upper bound. represented single float, lower = upper. contrast factor randomly picked [1.0 - lower, 1.0 + upper]. pixel x channel, output (x - mean) * factor + mean mean mean value channel. saturation_factor tuple two floats single float. factor controls extent image saturation impacted. factor=0.5 makes layer perform -op operation. factor=0.0 makes image fully grayscale. factor=1.0 makes image fully saturated. Values 0.0 1.0. tuple used, factor sampled two values every image augmented. single float used, value 0.0 passed float sampled. ensure value always , pass tuple two identical floats: (0.5, 0.5). hue_factor single float tuple two floats. factor controls extent image hue impacted. factor=0.0 makes layer perform -op operation, value 1.0 performs aggressive contrast adjustment available. tuple used, factor sampled two values every image augmented. single float used, value 0.0 passed float sampled. order ensure value always , please pass tuple two identical floats: (0.5, 0.5). seed Integer. Used create random seed. data_format String, one \"channels_last\" (default) \"channels_first\". ordering dimensions inputs. \"channels_last\" corresponds inputs shape (batch, height, width, channels) \"channels_first\" corresponds inputs shape (batch, channels, height, width). ... forward/backward compatability.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_random_contrast.html","id":null,"dir":"Reference","previous_headings":"","what":"A preprocessing layer which randomly adjusts contrast during training. — layer_random_contrast","title":"A preprocessing layer which randomly adjusts contrast during training. — layer_random_contrast","text":"layer randomly adjust contrast image images random factor. Contrast adjusted independently channel image training. channel, layer computes mean image pixels channel adjusts component x pixel (x - mean) * contrast_factor + mean. Input pixel values can range (e.g. [0., 1.) [0, 255]) integer floating point dtype. default, layer output floats. Note: layer safe use inside tf.data pipeline (independently backend using).","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_random_contrast.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"A preprocessing layer which randomly adjusts contrast during training. — layer_random_contrast","text":"","code":"layer_random_contrast(   object,   factor,   value_range = c(0L, 255L),   seed = NULL,   ... )"},{"path":"https://keras3.posit.co/dev/reference/layer_random_contrast.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"A preprocessing layer which randomly adjusts contrast during training. — layer_random_contrast","text":"object Object compose layer . tensor, array, sequential model. factor positive float represented fraction value, tuple size 2 representing lower upper bound. represented single float, lower = upper. contrast factor randomly picked [1.0 - lower, 1.0 + upper]. pixel x channel, output (x - mean) * factor + mean mean mean value channel. value_range range values incoming images . Represented two-number tuple written tuple(low, high). typically either [0, 1] [0, 255] depending preprocessing pipeline set . seed Integer. Used create random seed. ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_random_contrast.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"A preprocessing layer which randomly adjusts contrast during training. — layer_random_contrast","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_random_contrast.html","id":"input-shape","dir":"Reference","previous_headings":"","what":"Input Shape","title":"A preprocessing layer which randomly adjusts contrast during training. — layer_random_contrast","text":"3D (unbatched) 4D (batched) tensor shape: (..., height, width, channels), \"channels_last\" format.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_random_contrast.html","id":"output-shape","dir":"Reference","previous_headings":"","what":"Output Shape","title":"A preprocessing layer which randomly adjusts contrast during training. — layer_random_contrast","text":"3D (unbatched) 4D (batched) tensor shape: (..., height, width, channels), \"channels_last\" format.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_random_crop.html","id":null,"dir":"Reference","previous_headings":"","what":"A preprocessing layer which randomly crops images during training. — layer_random_crop","title":"A preprocessing layer which randomly crops images during training. — layer_random_crop","text":"training, layer randomly choose location crop images target size. layer crop images batch cropping location. inference time, training input image smaller target size, input resized cropped return largest possible window image matches target aspect ratio. need apply random cropping inference time, set training TRUE calling layer. Input pixel values can range (e.g. [0., 1.) [0, 255]) integer floating point dtype. default, layer output floats. Note: layer safe use inside tf.data pipeline (independently backend using).","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_random_crop.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"A preprocessing layer which randomly crops images during training. — layer_random_crop","text":"","code":"layer_random_crop(   object,   height,   width,   seed = NULL,   data_format = NULL,   name = NULL,   ... )"},{"path":"https://keras3.posit.co/dev/reference/layer_random_crop.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"A preprocessing layer which randomly crops images during training. — layer_random_crop","text":"object Object compose layer . tensor, array, sequential model. height Integer, height output shape. width Integer, width output shape. seed Integer. Used create random seed. data_format string, either \"channels_last\" \"channels_first\". ordering dimensions inputs. \"channels_last\" corresponds inputs shape (batch, height, width, channels) \"channels_first\" corresponds inputs shape (batch, channels, height, width). defaults image_data_format value found Keras config file ~/.keras/keras.json. never set , \"channels_last\". name String, name object ... Base layer keyword arguments, name dtype.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_random_crop.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"A preprocessing layer which randomly crops images during training. — layer_random_crop","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_random_crop.html","id":"input-shape","dir":"Reference","previous_headings":"","what":"Input Shape","title":"A preprocessing layer which randomly crops images during training. — layer_random_crop","text":"3D (unbatched) 4D (batched) tensor shape: (..., height, width, channels), \"channels_last\" format.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_random_crop.html","id":"output-shape","dir":"Reference","previous_headings":"","what":"Output Shape","title":"A preprocessing layer which randomly crops images during training. — layer_random_crop","text":"3D (unbatched) 4D (batched) tensor shape: (..., target_height, target_width, channels).","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_random_elastic_transform.html","id":null,"dir":"Reference","previous_headings":"","what":"A preprocessing layer that applies random elastic transformations. — layer_random_elastic_transform","title":"A preprocessing layer that applies random elastic transformations. — layer_random_elastic_transform","text":"layer distorts input images applying elastic deformations, simulating physically realistic transformation. magnitude distortion controlled scale parameter, factor determines probability applying transformation.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_random_elastic_transform.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"A preprocessing layer that applies random elastic transformations. — layer_random_elastic_transform","text":"","code":"layer_random_elastic_transform(   object,   factor = 1,   scale = 1,   interpolation = \"bilinear\",   fill_mode = \"reflect\",   fill_value = 0,   value_range = list(0L, 255L),   seed = NULL,   data_format = NULL,   ... )"},{"path":"https://keras3.posit.co/dev/reference/layer_random_elastic_transform.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"A preprocessing layer that applies random elastic transformations. — layer_random_elastic_transform","text":"object Object compose layer . tensor, array, sequential model. factor single float tuple two floats. factor controls probability applying transformation. factor = 0.0 ensures transformation applied. factor = 1.0 means transformation always applied. tuple (min, max) provided, probability value sampled min max image. single float provided, probability sampled 0.0 given float. Default 1.0. scale float tuple two floats defining magnitude distortion applied. tuple (min, max) provided, random scale value sampled within range. single float provided, random scale value sampled 0.0 given float. Default 1.0. interpolation Interpolation mode. Supported values: \"nearest\", \"bilinear\". fill_mode Points outside boundaries input filled according given mode. Available methods \"constant\", \"nearest\", \"wrap\" \"reflect\". Defaults \"reflect\". \"reflect\": (d c b | b c d | d c b ) input extended reflecting edge last pixel. \"constant\": (k k k k | b c d | k k k k) input extended filling values beyond edge constant value k specified fill_value. \"wrap\": (b c d | b c d | b c d) input extended wrapping around opposite edge. \"nearest\": (| b c d | d d d d) input extended nearest pixel. using torch backend, \"reflect\" redirected \"mirror\" torch support \"reflect\". torch backend also support \"wrap\". fill_value float representing value fill outside boundaries fill_mode = \"constant\". value_range range values incoming images . Represented two-number tuple written [low, high]. typically either [0, 1] [0, 255] depending preprocessing pipeline set . seed Integer. Used create random seed. data_format string, either \"channels_last\" \"channels_first\". ordering dimensions inputs. \"channels_last\" corresponds inputs shape (batch, height, width, channels) \"channels_first\" corresponds inputs shape (batch, channels, height, width). defaults image_data_format value found Keras config file ~/.keras/keras.json. never set , \"channels_last\". ... forward/backward compatability.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_random_erasing.html","id":null,"dir":"Reference","previous_headings":"","what":"Random Erasing data augmentation technique. — layer_random_erasing","title":"Random Erasing data augmentation technique. — layer_random_erasing","text":"Random Erasing data augmentation method random patches image erased (replaced constant value noise) training improve generalization.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_random_erasing.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Random Erasing data augmentation technique. — layer_random_erasing","text":"","code":"layer_random_erasing(   object,   factor = 1,   scale = list(0.02, 0.33),   fill_value = NULL,   value_range = list(0L, 255L),   seed = NULL,   data_format = NULL,   ... )"},{"path":"https://keras3.posit.co/dev/reference/layer_random_erasing.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Random Erasing data augmentation technique. — layer_random_erasing","text":"object Object compose layer . tensor, array, sequential model. factor single float tuple two floats. factor controls probability applying transformation. factor=0.0 ensures erasing applied. factor=1.0 means erasing always applied. tuple (min, max) provided, probability value sampled min max image. single float provided, probability sampled 0.0 given float. Default 1.0. scale tuple two floats representing aspect ratio range erased patch. defines width--height ratio patch erased. can help control rw shape erased region. Default (0.02, 0.33). fill_value value fill erased region . can set constant value None sample random value normal distribution. Default None. value_range range values incoming images . Represented two-number tuple written [low, high]. typically either [0, 1] [0, 255] depending preprocessing pipeline set . seed Integer. Used create random seed. data_format string, either \"channels_last\" \"channels_first\". ordering dimensions inputs. \"channels_last\" corresponds inputs shape (batch, height, width, channels) \"channels_first\" corresponds inputs shape (batch, channels, height, width). defaults image_data_format value found Keras config file ~/.keras/keras.json. never set , \"channels_last\". ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_random_erasing.html","id":"references","dir":"Reference","previous_headings":"","what":"References","title":"Random Erasing data augmentation technique. — layer_random_erasing","text":"Random Erasing paper.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_random_flip.html","id":null,"dir":"Reference","previous_headings":"","what":"A preprocessing layer which randomly flips images during training. — layer_random_flip","title":"A preprocessing layer which randomly flips images during training. — layer_random_flip","text":"layer flip images horizontally vertically based mode attribute. inference time, output identical input. Call layer training=TRUE flip input. Input pixel values can range (e.g. [0., 1.) [0, 255]) integer floating point dtype. default, layer output floats. Note: layer safe use inside tf.data pipeline (independently backend using).","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_random_flip.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"A preprocessing layer which randomly flips images during training. — layer_random_flip","text":"","code":"layer_random_flip(   object,   mode = \"horizontal_and_vertical\",   seed = NULL,   data_format = NULL,   ... )"},{"path":"https://keras3.posit.co/dev/reference/layer_random_flip.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"A preprocessing layer which randomly flips images during training. — layer_random_flip","text":"object Object compose layer . tensor, array, sequential model. mode String indicating flip mode use. Can \"horizontal\", \"vertical\", \"horizontal_and_vertical\". \"horizontal\" left-right flip \"vertical\" top-bottom flip. Defaults \"horizontal_and_vertical\" seed Integer. Used create random seed. data_format string, either \"channels_last\" \"channels_first\". ordering dimensions inputs. \"channels_last\" corresponds inputs shape (batch, height, width, channels) \"channels_first\" corresponds inputs shape (batch, channels, height, width). defaults image_data_format value found Keras config file ~/.keras/keras.json. never set , \"channels_last\". ... Base layer keyword arguments, name dtype.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_random_flip.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"A preprocessing layer which randomly flips images during training. — layer_random_flip","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_random_flip.html","id":"input-shape","dir":"Reference","previous_headings":"","what":"Input Shape","title":"A preprocessing layer which randomly flips images during training. — layer_random_flip","text":"3D (unbatched) 4D (batched) tensor shape: (..., height, width, channels), \"channels_last\" format.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_random_flip.html","id":"output-shape","dir":"Reference","previous_headings":"","what":"Output Shape","title":"A preprocessing layer which randomly flips images during training. — layer_random_flip","text":"3D (unbatched) 4D (batched) tensor shape: (..., height, width, channels), \"channels_last\" format.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_random_gaussian_blur.html","id":null,"dir":"Reference","previous_headings":"","what":"Applies random Gaussian blur to images for data augmentation. — layer_random_gaussian_blur","title":"Applies random Gaussian blur to images for data augmentation. — layer_random_gaussian_blur","text":"layer performs Gaussian blur operation input images randomly selected degree blurring, controlled factor sigma arguments.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_random_gaussian_blur.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Applies random Gaussian blur to images for data augmentation. — layer_random_gaussian_blur","text":"","code":"layer_random_gaussian_blur(   object,   factor = 1,   kernel_size = 3L,   sigma = 1,   value_range = list(0L, 255L),   data_format = NULL,   seed = NULL,   ... )"},{"path":"https://keras3.posit.co/dev/reference/layer_random_gaussian_blur.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Applies random Gaussian blur to images for data augmentation. — layer_random_gaussian_blur","text":"object Object compose layer . tensor, array, sequential model. factor single float tuple two floats. factor controls extent image hue impacted. factor=0.0 makes layer perform -op operation, value 1.0 performs aggressive blurring available. tuple used, factor sampled two values every image augmented. single float used, value 0.0 passed float sampled. Default 1.0. kernel_size Integer. Size Gaussian kernel used blurring. Must odd integer. Default 3. sigma Float tuple two floats. Standard deviation Gaussian kernel. Controls intensity blur. tuple provided, value sampled two image. Default 1.0. value_range range values incoming images . Represented two-number tuple written [low, high]. typically either [0, 1] [0, 255] depending preprocessing pipeline set . data_format string, either \"channels_last\" \"channels_first\". ordering dimensions inputs. \"channels_last\" corresponds inputs shape (batch, height, width, channels) \"channels_first\" corresponds inputs shape (batch, channels, height, width). defaults image_data_format value found Keras config file ~/.keras/keras.json. never set , \"channels_last\". seed Integer. Used create random seed. ... forward/backward compatability.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_random_grayscale.html","id":null,"dir":"Reference","previous_headings":"","what":"Preprocessing layer for random conversion of RGB images to grayscale. — layer_random_grayscale","title":"Preprocessing layer for random conversion of RGB images to grayscale. — layer_random_grayscale","text":"layer randomly converts input images grayscale specified factor. applied, maintains original number channels sets channels grayscale value. can useful data augmentation training models robust color variations. conversion preserves perceived luminance original color image using standard RGB grayscale conversion coefficients. Images selected conversion remain unchanged. Note: layer safe use inside tf.data pipeline (independently backend using).","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_random_grayscale.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Preprocessing layer for random conversion of RGB images to grayscale. — layer_random_grayscale","text":"","code":"layer_random_grayscale(   object,   factor = 0.5,   data_format = NULL,   seed = NULL,   ... )"},{"path":"https://keras3.posit.co/dev/reference/layer_random_grayscale.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Preprocessing layer for random conversion of RGB images to grayscale. — layer_random_grayscale","text":"object Object compose layer . tensor, array, sequential model. factor Float 0 1, specifying factor converting image grayscale. Defaults 0.5. value 1.0 means images converted, 0.0 means images converted. data_format String, one \"channels_last\" (default) \"channels_first\". ordering dimensions inputs. \"channels_last\" corresponds inputs shape (batch, height, width, channels) \"channels_first\" corresponds inputs shape (batch, channels, height, width). seed Initial seed random number generator ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_random_grayscale.html","id":"input-shape","dir":"Reference","previous_headings":"","what":"Input Shape","title":"Preprocessing layer for random conversion of RGB images to grayscale. — layer_random_grayscale","text":"3D (unbatched) 4D (batched) tensor shape: (..., height, width, channels), \"channels_last\" format, (..., channels, height, width), \"channels_first\" format.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_random_grayscale.html","id":"output-shape","dir":"Reference","previous_headings":"","what":"Output Shape","title":"Preprocessing layer for random conversion of RGB images to grayscale. — layer_random_grayscale","text":"input shape. output maintains number channels input, even grayscale-converted images channels value.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_random_hue.html","id":null,"dir":"Reference","previous_headings":"","what":"Randomly adjusts the hue on given images. — layer_random_hue","title":"Randomly adjusts the hue on given images. — layer_random_hue","text":"layer randomly increase/reduce hue input RGB images. image hue adjusted converting image(s) HSV rotating hue channel (H) delta. image converted back RGB.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_random_hue.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Randomly adjusts the hue on given images. — layer_random_hue","text":"","code":"layer_random_hue(   object,   factor,   value_range = list(0L, 255L),   data_format = NULL,   seed = NULL,   ... )"},{"path":"https://keras3.posit.co/dev/reference/layer_random_hue.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Randomly adjusts the hue on given images. — layer_random_hue","text":"object Object compose layer . tensor, array, sequential model. factor single float tuple two floats. factor controls extent image hue impacted. factor=0.0 makes layer perform -op operation, value 1.0 performs aggressive contrast adjustment available. tuple used, factor sampled two values every image augmented. single float used, value 0.0 passed float sampled. order ensure value always , please pass tuple two identical floats: (0.5, 0.5). value_range range values incoming images . Represented two-number tuple written [low, high]. typically either [0, 1] [0, 255] depending preprocessing pipeline set . data_format String, one \"channels_last\" (default) \"channels_first\". ordering dimensions inputs. \"channels_last\" corresponds inputs shape (batch, height, width, channels) \"channels_first\" corresponds inputs shape (batch, channels, height, width). seed Integer. Used create random seed. ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_random_hue.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Randomly adjusts the hue on given images. — layer_random_hue","text":"","code":"c(c(images, labels), .) %<-% dataset_cifar10() random_hue <- layer_random_hue(factor=0.5, value_range=c(0, 1)) images <- op_cast(images[1:8,,,], \"float32\") augmented_images_batch = random_hue(images)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_random_invert.html","id":null,"dir":"Reference","previous_headings":"","what":"Preprocessing layer for random inversion of image colors. — layer_random_invert","title":"Preprocessing layer for random inversion of image colors. — layer_random_invert","text":"layer randomly inverts colors input images specified probability range. applied, image chance colors inverted, pixel values transformed complementary values. Images selected inversion remain unchanged.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_random_invert.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Preprocessing layer for random inversion of image colors. — layer_random_invert","text":"","code":"layer_random_invert(   object,   factor = 1,   value_range = list(0L, 255L),   seed = NULL,   data_format = NULL,   ... )"},{"path":"https://keras3.posit.co/dev/reference/layer_random_invert.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Preprocessing layer for random inversion of image colors. — layer_random_invert","text":"object Object compose layer . tensor, array, sequential model. factor single float tuple two floats. factor controls probability inverting image colors. tuple provided, value sampled two values image, factor[0] minimum factor[1] maximum probability. single float provided, value 0.0 provided float sampled. Defaults (0, 1). value_range tuple list two elements. first value represents lower bound values passed images, second represents upper bound. Images passed layer values within value_range. Defaults (0, 255). seed Integer. Used create random seed. data_format string, either \"channels_last\" \"channels_first\". ordering dimensions inputs. \"channels_last\" corresponds inputs shape (batch, height, width, channels) \"channels_first\" corresponds inputs shape (batch, channels, height, width). defaults image_data_format value found Keras config file ~/.keras/keras.json. never set , \"channels_last\". ... forward/backward compatability.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_random_perspective.html","id":null,"dir":"Reference","previous_headings":"","what":"A preprocessing layer that applies random perspective transformations. — layer_random_perspective","title":"A preprocessing layer that applies random perspective transformations. — layer_random_perspective","text":"layer distorts perspective input images shifting corner points, simulating 3D-like transformation. amount distortion controlled factor scale parameters.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_random_perspective.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"A preprocessing layer that applies random perspective transformations. — layer_random_perspective","text":"","code":"layer_random_perspective(   object,   factor = 1,   scale = 1,   interpolation = \"bilinear\",   fill_value = 0,   seed = NULL,   data_format = NULL,   ... )"},{"path":"https://keras3.posit.co/dev/reference/layer_random_perspective.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"A preprocessing layer that applies random perspective transformations. — layer_random_perspective","text":"object Object compose layer . tensor, array, sequential model. factor float tuple two floats. Represents probability applying perspective transformation image batch. factor=0.0 ensures transformation applied. factor=1.0 means transformation always applied. tuple (min, max) provided, probability randomly sampled min max image. single float given, probability sampled 0.0 provided float. Default 1.0. scale float defining relative amount perspective shift. Determines much image corners displaced, affecting intensity perspective effect. interpolation Interpolation mode. Supported values: \"nearest\", \"bilinear\". fill_value float represents value filled outside boundaries fill_mode=\"constant\". seed Integer. Used create random seed. data_format string, either \"channels_last\" \"channels_first\". ordering dimensions inputs. \"channels_last\" corresponds inputs shape (batch, height, width, channels) \"channels_first\" corresponds inputs shape (batch, channels, height, width). defaults image_data_format value found Keras config file ~/.keras/keras.json. never set , \"channels_last\". ... forward/backward compatability.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_random_posterization.html","id":null,"dir":"Reference","previous_headings":"","what":"Reduces the number of bits for each color channel. — layer_random_posterization","title":"Reduces the number of bits for each color channel. — layer_random_posterization","text":"Reduces number bits color channel.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_random_posterization.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Reduces the number of bits for each color channel. — layer_random_posterization","text":"","code":"layer_random_posterization(   object,   factor,   value_range = list(0L, 255L),   data_format = NULL,   seed = NULL,   ... )"},{"path":"https://keras3.posit.co/dev/reference/layer_random_posterization.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Reduces the number of bits for each color channel. — layer_random_posterization","text":"object Object compose layer . tensor, array, sequential model. factor integer, number bits keep channel. Must value 1-8. value_range tuple list two elements. first value represents lower bound values passed images, second represents upper bound. Images passed layer values within value_range. Defaults (0, 255). data_format String, one \"channels_last\" (default) \"channels_first\". ordering dimensions inputs. \"channels_last\" corresponds inputs shape (batch, height, width, channels) \"channels_first\" corresponds inputs shape (batch, channels, height, width). seed Initial seed random number generator ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_random_posterization.html","id":"references","dir":"Reference","previous_headings":"","what":"References","title":"Reduces the number of bits for each color channel. — layer_random_posterization","text":"AutoAugment: Learning Augmentation Policies Data RandAugment: Practical automated data augmentation reduced search space","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_random_rotation.html","id":null,"dir":"Reference","previous_headings":"","what":"A preprocessing layer which randomly rotates images during training. — layer_random_rotation","title":"A preprocessing layer which randomly rotates images during training. — layer_random_rotation","text":"layer apply random rotations image, filling empty space according fill_mode. default, random rotations applied training. inference time, layer nothing. need apply random rotations inference time, pass training = TRUE calling layer. Input pixel values can range (e.g. [0., 1.) [0, 255]) integer floating point dtype. default, layer output floats. Note: layer safe use inside tf.data pipeline (independently backend using).","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_random_rotation.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"A preprocessing layer which randomly rotates images during training. — layer_random_rotation","text":"","code":"layer_random_rotation(   object,   factor,   fill_mode = \"reflect\",   interpolation = \"bilinear\",   seed = NULL,   fill_value = 0,   data_format = NULL,   ... )"},{"path":"https://keras3.posit.co/dev/reference/layer_random_rotation.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"A preprocessing layer which randomly rotates images during training. — layer_random_rotation","text":"object Object compose layer . tensor, array, sequential model. factor float represented fraction 2 Pi, tuple size 2 representing lower upper bound rotating clockwise counter-clockwise. positive values means rotating counter clock-wise, negative value means clock-wise. represented single float, value used upper lower bound. instance, factor=c(-0.2, 0.3) results output rotation random amount range [-20% * 360, 30% * 360]. factor=0.2 results output rotating random amount range [-20% * 360, 20% * 360]. fill_mode Points outside boundaries input filled according given mode (one {\"constant\", \"reflect\", \"wrap\", \"nearest\"}). reflect: (d c b | b c d | d c b ) input extended reflecting edge last pixel. constant: (k k k k | b c d | k k k k) input extended filling values beyond edge constant value k = 0. wrap: (b c d | b c d | b c d) input extended wrapping around opposite edge. nearest: (| b c d | d d d d) input extended nearest pixel. interpolation Interpolation mode. Supported values: \"nearest\", \"bilinear\". seed Integer. Used create random seed. fill_value float represents value filled outside boundaries fill_mode=\"constant\". data_format string, either \"channels_last\" \"channels_first\". ordering dimensions inputs. \"channels_last\" corresponds inputs shape (batch, height, width, channels) \"channels_first\" corresponds inputs shape (batch, channels, height, width). defaults image_data_format value found Keras config file ~/.keras/keras.json. never set , \"channels_last\". ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_random_rotation.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"A preprocessing layer which randomly rotates images during training. — layer_random_rotation","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_random_rotation.html","id":"input-shape","dir":"Reference","previous_headings":"","what":"Input Shape","title":"A preprocessing layer which randomly rotates images during training. — layer_random_rotation","text":"3D (unbatched) 4D (batched) tensor shape: (..., height, width, channels), \"channels_last\" format","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_random_rotation.html","id":"output-shape","dir":"Reference","previous_headings":"","what":"Output Shape","title":"A preprocessing layer which randomly rotates images during training. — layer_random_rotation","text":"3D (unbatched) 4D (batched) tensor shape: (..., height, width, channels), \"channels_last\" format","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_random_saturation.html","id":null,"dir":"Reference","previous_headings":"","what":"Randomly adjusts the saturation on given images. — layer_random_saturation","title":"Randomly adjusts the saturation on given images. — layer_random_saturation","text":"layer randomly increase/reduce saturation input RGB images.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_random_saturation.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Randomly adjusts the saturation on given images. — layer_random_saturation","text":"","code":"layer_random_saturation(   object,   factor,   value_range = list(0L, 255L),   data_format = NULL,   seed = NULL,   ... )"},{"path":"https://keras3.posit.co/dev/reference/layer_random_saturation.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Randomly adjusts the saturation on given images. — layer_random_saturation","text":"object Object compose layer . tensor, array, sequential model. factor tuple two floats single float. factor controls extent image saturation impacted. factor=0.5 makes layer perform -op operation. factor=0.0 makes image fully grayscale. factor=1.0 makes image fully saturated. Values 0.0 1.0. tuple used, factor sampled two values every image augmented. single float used, value 0.0 passed float sampled. ensure value always , pass tuple two identical floats: (0.5, 0.5). value_range range values incoming images . Represented two-number tuple written [low, high]. typically either [0, 1] [0, 255] depending preprocessing pipeline set . data_format String, one \"channels_last\" (default) \"channels_first\". ordering dimensions inputs. \"channels_last\" corresponds inputs shape (batch, height, width, channels) \"channels_first\" corresponds inputs shape (batch, channels, height, width). seed Integer. Used create random seed. ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_random_saturation.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Randomly adjusts the saturation on given images. — layer_random_saturation","text":"","code":"c(c(images, labels), .) %<-% dataset_cifar10() images <- images[1:8, , , ] |> op_cast(\"float32\") random_saturation <- layer_random_saturation(factor = 0.2) augmented_images <- random_saturation(images)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_random_sharpness.html","id":null,"dir":"Reference","previous_headings":"","what":"Randomly performs the sharpness operation on given images. — layer_random_sharpness","title":"Randomly performs the sharpness operation on given images. — layer_random_sharpness","text":"sharpness operation first performs blur, blends original image processed image. operation adjusts clarity edges image, ranging blurred enhanced sharpness.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_random_sharpness.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Randomly performs the sharpness operation on given images. — layer_random_sharpness","text":"","code":"layer_random_sharpness(   object,   factor,   value_range = list(0L, 255L),   data_format = NULL,   seed = NULL,   ... )"},{"path":"https://keras3.posit.co/dev/reference/layer_random_sharpness.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Randomly performs the sharpness operation on given images. — layer_random_sharpness","text":"object Object compose layer . tensor, array, sequential model. factor tuple two floats single float. factor controls extent image sharpness impacted. factor=0.0 results fully blurred image, factor=0.5 applies operation (preserving original image), factor=1.0 enhances sharpness beyond original. Values 0.0 1.0. tuple used, factor sampled two values every image augmented. single float used, value 0.0 passed float sampled. ensure value always , pass tuple two identical floats: (0.5, 0.5). value_range range values incoming images . Represented two-number tuple written [low, high]. typically either [0, 1] [0, 255] depending preprocessing pipeline set . data_format String, one \"channels_last\" (default) \"channels_first\". ordering dimensions inputs. \"channels_last\" corresponds inputs shape (batch, height, width, channels) \"channels_first\" corresponds inputs shape (batch, channels, height, width). seed Integer. Used create random seed. ... forward/backward compatability.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_random_shear.html","id":null,"dir":"Reference","previous_headings":"","what":"A preprocessing layer that randomly applies shear transformations — layer_random_shear","title":"A preprocessing layer that randomly applies shear transformations — layer_random_shear","text":"images. layer shears input images along x-axis /y-axis randomly selected factor within specified range. shear transformation applied image independently batch. Empty regions created transformation filled according fill_mode fill_value parameters.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_random_shear.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"A preprocessing layer that randomly applies shear transformations — layer_random_shear","text":"","code":"layer_random_shear(   object,   x_factor = 0,   y_factor = 0,   interpolation = \"bilinear\",   fill_mode = \"reflect\",   fill_value = 0,   data_format = NULL,   seed = NULL,   ... )"},{"path":"https://keras3.posit.co/dev/reference/layer_random_shear.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"A preprocessing layer that randomly applies shear transformations — layer_random_shear","text":"object Object compose layer . tensor, array, sequential model. x_factor tuple two floats. augmented image, value sampled provided range. float passed, range interpreted (0, x_factor). Values represent percentage image shear . example, 0.3 shears pixels 30% way across image. provided values positive. y_factor tuple two floats. augmented image, value sampled provided range. float passed, range interpreted (0, y_factor). Values represent percentage image shear . example, 0.3 shears pixels 30% way across image. provided values positive. interpolation Interpolation mode. Supported values: \"nearest\", \"bilinear\". fill_mode Points outside boundaries input filled according given mode. Available methods \"constant\", \"nearest\", \"wrap\" \"reflect\". Defaults \"reflect\". \"reflect\": (d c b | b c d | d c b ) input extended reflecting edge last pixel. \"constant\": (k k k k | b c d | k k k k) input extended filling values beyond edge constant value k specified fill_value. \"wrap\": (b c d | b c d | b c d) input extended wrapping around opposite edge. \"nearest\": (| b c d | d d d d) input extended nearest pixel. Note using torch backend, \"reflect\" redirected \"mirror\" (c d c b | b c d | c b b) torch support \"reflect\". Note torch backend support \"wrap\". fill_value float representing value filled outside boundaries fill_mode=\"constant\". data_format String, one \"channels_last\" (default) \"channels_first\". ordering dimensions inputs. \"channels_last\" corresponds inputs shape (batch, height, width, channels) \"channels_first\" corresponds inputs shape (batch, channels, height, width). seed Integer. Used create random seed. ... forward/backward compatability.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_random_translation.html","id":null,"dir":"Reference","previous_headings":"","what":"A preprocessing layer which randomly translates images during training. — layer_random_translation","title":"A preprocessing layer which randomly translates images during training. — layer_random_translation","text":"layer apply random translations image training, filling empty space according fill_mode. Input pixel values can range (e.g. [0., 1.) [0, 255]) integer floating point dtype. default, layer output floats.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_random_translation.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"A preprocessing layer which randomly translates images during training. — layer_random_translation","text":"","code":"layer_random_translation(   object,   height_factor,   width_factor,   fill_mode = \"reflect\",   interpolation = \"bilinear\",   seed = NULL,   fill_value = 0,   data_format = NULL,   ... )"},{"path":"https://keras3.posit.co/dev/reference/layer_random_translation.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"A preprocessing layer which randomly translates images during training. — layer_random_translation","text":"object Object compose layer . tensor, array, sequential model. height_factor float represented fraction value, tuple size 2 representing lower upper bound shifting vertically. negative value means shifting image , positive value means shifting image . represented single positive float, value used upper lower bound. instance, height_factor=(-0.2, 0.3) results output shifted random amount range [-20%, +30%]. height_factor=0.2 results output height shifted random amount range [-20%, +20%]. width_factor float represented fraction value, tuple size 2 representing lower upper bound shifting horizontally. negative value means shifting image left, positive value means shifting image right. represented single positive float, value used upper lower bound. instance, width_factor=(-0.2, 0.3) results output shifted left 20%, shifted right 30%. width_factor=0.2 results output height shifted left right 20%. fill_mode Points outside boundaries input filled according given mode. Available methods \"constant\", \"nearest\", \"wrap\" \"reflect\". Defaults \"reflect\". \"reflect\": (d c b | b c d | d c b ) input extended reflecting edge last pixel. \"constant\": (k k k k | b c d | k k k k) input extended filling values beyond edge constant value k specified fill_value. \"wrap\": (b c d | b c d | b c d) input extended wrapping around opposite edge. \"nearest\": (| b c d | d d d d) input extended nearest pixel. Note using torch backend, \"reflect\" redirected \"mirror\" (c d c b | b c d | c b b) torch support \"reflect\". Note torch backend support \"wrap\". interpolation Interpolation mode. Supported values: \"nearest\", \"bilinear\". seed Integer. Used create random seed. fill_value float represents value filled outside boundaries fill_mode=\"constant\". data_format string, either \"channels_last\" \"channels_first\". ordering dimensions inputs. \"channels_last\" corresponds inputs shape (batch, height, width, channels) \"channels_first\" corresponds inputs shape (batch, channels, height, width). defaults image_data_format value found Keras config file ~/.keras/keras.json. never set , \"channels_last\". ... Base layer keyword arguments, name dtype.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_random_translation.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"A preprocessing layer which randomly translates images during training. — layer_random_translation","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_random_translation.html","id":"input-shape","dir":"Reference","previous_headings":"","what":"Input Shape","title":"A preprocessing layer which randomly translates images during training. — layer_random_translation","text":"3D (unbatched) 4D (batched) tensor shape: (..., height, width, channels), \"channels_last\" format, (..., channels, height, width), \"channels_first\" format.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_random_translation.html","id":"output-shape","dir":"Reference","previous_headings":"","what":"Output Shape","title":"A preprocessing layer which randomly translates images during training. — layer_random_translation","text":"3D (unbatched) 4D (batched) tensor shape: (..., target_height, target_width, channels), (..., channels, target_height, target_width), \"channels_first\" format. Note: layer safe use inside tf.data pipeline (independently backend using).","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_random_zoom.html","id":null,"dir":"Reference","previous_headings":"","what":"A preprocessing layer which randomly zooms images during training. — layer_random_zoom","title":"A preprocessing layer which randomly zooms images during training. — layer_random_zoom","text":"layer randomly zoom axis image independently, filling empty space according fill_mode. Input pixel values can range (e.g. [0., 1.) [0, 255]) integer floating point dtype. default, layer output floats.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_random_zoom.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"A preprocessing layer which randomly zooms images during training. — layer_random_zoom","text":"","code":"layer_random_zoom(   object,   height_factor,   width_factor = NULL,   fill_mode = \"reflect\",   interpolation = \"bilinear\",   seed = NULL,   fill_value = 0,   data_format = NULL,   ... )"},{"path":"https://keras3.posit.co/dev/reference/layer_random_zoom.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"A preprocessing layer which randomly zooms images during training. — layer_random_zoom","text":"object Object compose layer . tensor, array, sequential model. height_factor float represented fraction value, list size 2 representing lower upper bound zooming vertically. represented single float, value used upper lower bound. positive value means zooming , negative value means zooming . instance, height_factor=c(0.2, 0.3) result output zoomed random amount range [+20%, +30%]. height_factor=c(-0.3, -0.2) result output zoomed random amount range [+20%, +30%]. width_factor float represented fraction value, list size 2 representing lower upper bound zooming horizontally. represented single float, value used upper lower bound. instance, width_factor=c(0.2, 0.3) result output zooming 20% 30%. width_factor=c(-0.3, -0.2) result output zooming 20% 30%. NULL means .e., zooming vertical horizontal directions preserving aspect ratio. Defaults NULL. fill_mode Points outside boundaries input filled according given mode. Available methods \"constant\", \"nearest\", \"wrap\" \"reflect\". Defaults \"reflect\". \"reflect\": (d c b | b c d | d c b ) input extended reflecting edge last pixel. \"constant\": (k k k k | b c d | k k k k) input extended filling values beyond edge constant value k specified fill_value. \"wrap\": (b c d | b c d | b c d) input extended wrapping around opposite edge. \"nearest\": (| b c d | d d d d) input extended nearest pixel. Note using torch backend, \"reflect\" redirected \"mirror\" (c d c b | b c d | c b b) torch support \"reflect\". Note torch backend support \"wrap\". interpolation Interpolation mode. Supported values: \"nearest\", \"bilinear\". seed Integer. Used create random seed. fill_value float represents value filled outside boundaries fill_mode=\"constant\". data_format string, either \"channels_last\" \"channels_first\". ordering dimensions inputs. \"channels_last\" corresponds inputs shape (batch, height, width, channels) \"channels_first\" corresponds inputs shape (batch, channels, height, width). defaults image_data_format value found Keras config file ~/.keras/keras.json. never set , \"channels_last\". ... Base layer keyword arguments, name dtype.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_random_zoom.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"A preprocessing layer which randomly zooms images during training. — layer_random_zoom","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_random_zoom.html","id":"input-shape","dir":"Reference","previous_headings":"","what":"Input Shape","title":"A preprocessing layer which randomly zooms images during training. — layer_random_zoom","text":"3D (unbatched) 4D (batched) tensor shape: (..., height, width, channels), \"channels_last\" format, (..., channels, height, width), \"channels_first\" format.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_random_zoom.html","id":"output-shape","dir":"Reference","previous_headings":"","what":"Output Shape","title":"A preprocessing layer which randomly zooms images during training. — layer_random_zoom","text":"3D (unbatched) 4D (batched) tensor shape: (..., target_height, target_width, channels), (..., channels, target_height, target_width), \"channels_first\" format. Note: layer safe use inside tf.data pipeline (independently backend using).","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_random_zoom.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"A preprocessing layer which randomly zooms images during training. — layer_random_zoom","text":"","code":"input_img <- random_uniform(c(32, 224, 224, 3)) layer <- layer_random_zoom(height_factor = .5, width_factor = .2) out_img <- layer(input_img)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_repeat_vector.html","id":null,"dir":"Reference","previous_headings":"","what":"Repeats the input n times. — layer_repeat_vector","title":"Repeats the input n times. — layer_repeat_vector","text":"Repeats input n times.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_repeat_vector.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Repeats the input n times. — layer_repeat_vector","text":"","code":"layer_repeat_vector(object, n, ...)"},{"path":"https://keras3.posit.co/dev/reference/layer_repeat_vector.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Repeats the input n times. — layer_repeat_vector","text":"object Object compose layer . tensor, array, sequential model. n Integer, repetition factor. ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_repeat_vector.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Repeats the input n times. — layer_repeat_vector","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_repeat_vector.html","id":"example","dir":"Reference","previous_headings":"","what":"Example","title":"Repeats the input n times. — layer_repeat_vector","text":"","code":"x <- layer_input(shape = 32) y <- layer_repeat_vector(x, n = 3) shape(y) ## shape(NA, 3, 32)"},{"path":"https://keras3.posit.co/dev/reference/layer_repeat_vector.html","id":"input-shape","dir":"Reference","previous_headings":"","what":"Input Shape","title":"Repeats the input n times. — layer_repeat_vector","text":"2D tensor shape (batch_size, features).","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_repeat_vector.html","id":"output-shape","dir":"Reference","previous_headings":"","what":"Output Shape","title":"Repeats the input n times. — layer_repeat_vector","text":"3D tensor shape (batch_size, n, features).","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_rescaling.html","id":null,"dir":"Reference","previous_headings":"","what":"A preprocessing layer which rescales input values to a new range. — layer_rescaling","title":"A preprocessing layer which rescales input values to a new range. — layer_rescaling","text":"layer rescales every value input (often image) multiplying scale adding offset. instance: rescale input [0, 255] range [0, 1] range, pass scale=1./255. rescale input [0, 255] range [-1, 1] range, pass scale=1./127.5, offset=-1. rescaling applied training inference. Inputs can integer floating point dtype, default layer output floats. Note: layer safe use inside tf.data pipeline (independently backend using).","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_rescaling.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"A preprocessing layer which rescales input values to a new range. — layer_rescaling","text":"","code":"layer_rescaling(object, scale, offset = 0, ...)"},{"path":"https://keras3.posit.co/dev/reference/layer_rescaling.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"A preprocessing layer which rescales input values to a new range. — layer_rescaling","text":"object Object compose layer . tensor, array, sequential model. scale Numeric scalar, vector, array. scale apply inputs. scalar, scale applied every feature channel; vector array, scaling applied per channel. offset Numeric scalar, vector, array. offset apply inputs. scalar, offset applied every feature channel; vector array, shift applied per channel. ... Base layer keyword arguments, name dtype.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_rescaling.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"A preprocessing layer which rescales input values to a new range. — layer_rescaling","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_reshape.html","id":null,"dir":"Reference","previous_headings":"","what":"Layer that reshapes inputs into the given shape. — layer_reshape","title":"Layer that reshapes inputs into the given shape. — layer_reshape","text":"Layer reshapes inputs given shape.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_reshape.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Layer that reshapes inputs into the given shape. — layer_reshape","text":"","code":"layer_reshape(object, target_shape, ...)"},{"path":"https://keras3.posit.co/dev/reference/layer_reshape.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Layer that reshapes inputs into the given shape. — layer_reshape","text":"object Object compose layer . tensor, array, sequential model. target_shape Target shape. List integers, include samples dimension (batch size). ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_reshape.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Layer that reshapes inputs into the given shape. — layer_reshape","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_reshape.html","id":"input-shape","dir":"Reference","previous_headings":"","what":"Input Shape","title":"Layer that reshapes inputs into the given shape. — layer_reshape","text":"Arbitrary, although dimensions input shape must known/fixed. Use keyword argument input_shape (list integers, include samples/batch size axis) using layer first layer model.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_reshape.html","id":"output-shape","dir":"Reference","previous_headings":"","what":"Output Shape","title":"Layer that reshapes inputs into the given shape. — layer_reshape","text":"(batch_size, *target_shape)","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_reshape.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Layer that reshapes inputs into the given shape. — layer_reshape","text":"","code":"x <- layer_input(shape = 12) y <- layer_reshape(x, c(3, 4)) shape(y) ## shape(NA, 3, 4) # also supports shape inference using `-1` as dimension y <- layer_reshape(x, c(-1, 2, 2)) shape(y) ## shape(NA, 3, 2, 2)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_resizing.html","id":null,"dir":"Reference","previous_headings":"","what":"A preprocessing layer which resizes images. — layer_resizing","title":"A preprocessing layer which resizes images. — layer_resizing","text":"layer resizes image input target height width. input 4D (batched) 3D (unbatched) tensor \"channels_last\" format. Input pixel values can range (e.g. [0., 1.) [0, 255]).","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_resizing.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"A preprocessing layer which resizes images. — layer_resizing","text":"","code":"layer_resizing(   object,   height,   width,   interpolation = \"bilinear\",   crop_to_aspect_ratio = FALSE,   pad_to_aspect_ratio = FALSE,   fill_mode = \"constant\",   fill_value = 0,   data_format = NULL,   ...,   antialias = FALSE )"},{"path":"https://keras3.posit.co/dev/reference/layer_resizing.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"A preprocessing layer which resizes images. — layer_resizing","text":"object Object compose layer . tensor, array, sequential model. height Integer, height output shape. width Integer, width output shape. interpolation String, interpolation method. Supports \"bilinear\", \"nearest\", \"bicubic\", \"lanczos3\", \"lanczos5\". Defaults \"bilinear\". crop_to_aspect_ratio TRUE, resize images without aspect ratio distortion. original aspect ratio differs target aspect ratio, output image cropped return largest possible window image (size (height, width)) matches target aspect ratio. default (crop_to_aspect_ratio=FALSE), aspect ratio may preserved. pad_to_aspect_ratio TRUE, pad images without aspect ratio distortion. original aspect ratio differs target aspect ratio, output image evenly padded short side. fill_mode using pad_to_aspect_ratio=TRUE, padded areas filled according given mode. \"constant\" supported time (fill constant value, equal fill_value). fill_value Float. Padding value use pad_to_aspect_ratio=TRUE. data_format string, either \"channels_last\" \"channels_first\". ordering dimensions inputs. \"channels_last\" corresponds inputs shape (batch, height, width, channels) \"channels_first\" corresponds inputs shape (batch, channels, height, width). defaults image_data_format value found Keras config file ~/.keras/keras.json. never set , \"channels_last\". ... Base layer keyword arguments, name dtype. antialias bool","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_resizing.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"A preprocessing layer which resizes images. — layer_resizing","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_resizing.html","id":"input-shape","dir":"Reference","previous_headings":"","what":"Input Shape","title":"A preprocessing layer which resizes images. — layer_resizing","text":"3D (unbatched) 4D (batched) tensor shape: (..., height, width, channels), \"channels_last\" format, (..., channels, height, width), \"channels_first\" format.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_resizing.html","id":"output-shape","dir":"Reference","previous_headings":"","what":"Output Shape","title":"A preprocessing layer which resizes images. — layer_resizing","text":"3D (unbatched) 4D (batched) tensor shape: (..., target_height, target_width, channels), (..., channels, target_height, target_width), \"channels_first\" format. Note: layer safe use inside tf.data pipeline (independently backend using).","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_rms_normalization.html","id":null,"dir":"Reference","previous_headings":"","what":"Root Mean Square (RMS) Normalization layer. — layer_rms_normalization","title":"Root Mean Square (RMS) Normalization layer. — layer_rms_normalization","text":"layer normalizes input tensor based RMS value. Keras layer performs operation described Root Mean Square Layer Normalization Biao Zhang et al. scale enabled, layer scale normalized outputs via learnable scaling factor. , scaling enabled, normalization equations follows: Let intermediate activations mini-batch inputs.   example:","code":"rms_normalization(x) = x * rsqrt(mean(square(x))) * scale layer <- layer_rms_normalization() layer$build(shape(5, 20, 30, 10)) op_shape(layer$scale$shape) ## shape(1)  op_shape(layer(op_array(runif(10)))) ## shape(10)"},{"path":"https://keras3.posit.co/dev/reference/layer_rms_normalization.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Root Mean Square (RMS) Normalization layer. — layer_rms_normalization","text":"","code":"layer_rms_normalization(object, axis = -1L, epsilon = 1e-06, ...)"},{"path":"https://keras3.posit.co/dev/reference/layer_rms_normalization.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Root Mean Square (RMS) Normalization layer. — layer_rms_normalization","text":"object Object compose layer . tensor, array, sequential model. axis int. axis perform normalization. epsilon float. small number add avoid division zero. ... forward/backward compatability.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_rnn.html","id":null,"dir":"Reference","previous_headings":"","what":"Base class for recurrent layers — layer_rnn","title":"Base class for recurrent layers — layer_rnn","text":"Base class recurrent layers","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_rnn.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Base class for recurrent layers — layer_rnn","text":"","code":"layer_rnn(   object,   cell,   return_sequences = FALSE,   return_state = FALSE,   go_backwards = FALSE,   stateful = FALSE,   unroll = FALSE,   zero_output_for_mask = FALSE,   ... )"},{"path":"https://keras3.posit.co/dev/reference/layer_rnn.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Base class for recurrent layers — layer_rnn","text":"object Object compose layer . tensor, array, sequential model. cell RNN cell instance list RNN cell instances. RNN cell class : call(input_at_t, states_at_t) method, returning (output_at_t, states_at_t_plus_1). call method cell can also take optional argument constants, see section \"Note passing external constants\" . state_size attribute. can single integer (single state) case size recurrent state. can also list integers (one size per state). output_size attribute, single integer. get_initial_state(batch_size=NULL) method creates tensor meant fed call() initial state, user specify initial state via means. returned initial state shape (batch_size, cell.state_size). cell might choose create tensor full zeros, values based cell's implementation. inputs input tensor RNN layer, shape (batch_size, timesteps, features). method implemented cell, RNN layer create zero filled tensor shape (batch_size, cell$state_size). case cell list RNN cell instances, cells stacked top RNN, resulting efficient stacked RNN. return_sequences Boolean (default FALSE). Whether return last output output sequence, full sequence. return_state Boolean (default FALSE). Whether return last state addition output. go_backwards Boolean (default FALSE). TRUE, process input sequence backwards return reversed sequence. stateful Boolean (default FALSE). TRUE, last state sample index batch used initial state sample index following batch. unroll Boolean (default FALSE). TRUE, network unrolled, else symbolic loop used. Unrolling can speed-RNN, although tends memory-intensive. Unrolling suitable short sequences. zero_output_for_mask Boolean (default FALSE). Whether output use zeros masked timesteps. Note field used return_sequences TRUE mask provided. can useful want reuse raw output sequence RNN without interference masked timesteps, e.g., merging bidirectional RNNs. ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_rnn.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Base class for recurrent layers — layer_rnn","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_rnn.html","id":"call-arguments","dir":"Reference","previous_headings":"","what":"Call Arguments","title":"Base class for recurrent layers — layer_rnn","text":"sequences: 3-D tensor shape (batch_size, timesteps, features). initial_state: List initial state tensors passed first call cell. mask: Binary tensor shape [batch_size, timesteps] indicating whether given timestep masked. individual TRUE entry indicates corresponding timestep utilized, FALSE entry indicates corresponding timestep ignored. training: Python boolean indicating whether layer behave training mode inference mode. argument passed cell calling . use cells use dropout.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_rnn.html","id":"input-shape","dir":"Reference","previous_headings":"","what":"Input Shape","title":"Base class for recurrent layers — layer_rnn","text":"3-D tensor shape (batch_size, timesteps, features).","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_rnn.html","id":"output-shape","dir":"Reference","previous_headings":"","what":"Output Shape","title":"Base class for recurrent layers — layer_rnn","text":"return_state: list tensors. first tensor output. remaining tensors last states, shape (batch_size, state_size), state_size high dimension tensor shape. return_sequences: 3D tensor shape (batch_size, timesteps, output_size).","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_rnn.html","id":"masking-","dir":"Reference","previous_headings":"","what":"Masking:","title":"Base class for recurrent layers — layer_rnn","text":"layer supports masking input data variable number timesteps. introduce masks data, use layer_embedding() layer mask_zero parameter set TRUE. Note using statefulness RNNs: can set RNN layers 'stateful', means states computed samples one batch reused initial states samples next batch. assumes one--one mapping samples different successive batches. enable statefulness: Specify stateful=TRUE layer constructor. Specify fixed batch size model, passing batch_size=... layer_input() layer(s) model. Remember also specify batch_size=... calling fit(), otherwise use generator-like data source like tf.data.Dataset. Specify shuffle=FALSE calling fit(), since batches expected temporally ordered. reset states model, call reset_state() either specific layer, entire model. Note specifying initial state RNNs: can specify initial state RNN layers symbolically passing named argument initial_state layer reset_state(). value initial_state tensor list tensors representing initial state RNN layer.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_rnn.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Base class for recurrent layers — layer_rnn","text":"First, define RNN Cell, layer subclass.   use cell RNN layer:","code":"rnn_cell_minimal <- Layer(   \"MinimalRNNCell\",    initialize = function(units, ...) {     super$initialize(...)     self$units <- as.integer(units)     self$state_size <- as.integer(units)   },    build = function(input_shape) {     self$kernel <- self$add_weight(       shape = shape(tail(input_shape, 1), self$units),       initializer = 'uniform',       name = 'kernel'     )     self$recurrent_kernel <- self$add_weight(       shape = shape(self$units, self$units),       initializer = 'uniform',       name = 'recurrent_kernel'     )   },    call = function(inputs, states) {     prev_output <- states[[1]]     h <- op_matmul(inputs, self$kernel)     output <- h + op_matmul(prev_output, self$recurrent_kernel)     list(output, list(output))   } ) cell <- rnn_cell_minimal(units = 32) x <- layer_input(shape = shape(NULL, 5)) layer <- layer_rnn(cell = cell) y <- layer(x) cells <- list(rnn_cell_minimal(units = 32), rnn_cell_minimal(units = 4)) x <- layer_input(shape = shape(NULL, 5)) layer <- layer_rnn(cell = cells) y <- layer(x)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_separable_conv_1d.html","id":null,"dir":"Reference","previous_headings":"","what":"1D separable convolution layer. — layer_separable_conv_1d","title":"1D separable convolution layer. — layer_separable_conv_1d","text":"layer performs depthwise convolution acts separately channels, followed pointwise convolution mixes channels. use_bias TRUE bias initializer provided, adds bias vector output. optionally applies activation function produce final output.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_separable_conv_1d.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"1D separable convolution layer. — layer_separable_conv_1d","text":"","code":"layer_separable_conv_1d(   object,   filters,   kernel_size,   strides = 1L,   padding = \"valid\",   data_format = NULL,   dilation_rate = 1L,   depth_multiplier = 1L,   activation = NULL,   use_bias = TRUE,   depthwise_initializer = \"glorot_uniform\",   pointwise_initializer = \"glorot_uniform\",   bias_initializer = \"zeros\",   depthwise_regularizer = NULL,   pointwise_regularizer = NULL,   bias_regularizer = NULL,   activity_regularizer = NULL,   depthwise_constraint = NULL,   pointwise_constraint = NULL,   bias_constraint = NULL,   ... )"},{"path":"https://keras3.posit.co/dev/reference/layer_separable_conv_1d.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"1D separable convolution layer. — layer_separable_conv_1d","text":"object Object compose layer . tensor, array, sequential model. filters int, dimensionality output space (.e. number filters pointwise convolution). kernel_size int list 1 integers, specifying size depthwise convolution window. strides int list 1 integers, specifying stride length depthwise convolution. one int specified, stride size used dimensions. strides > 1 incompatible dilation_rate > 1. padding string, either \"valid\" \"\" (case-insensitive). \"valid\" means padding. \"\" results padding evenly left/right /input. padding=\"\" strides=1, output size input. data_format string, either \"channels_last\" \"channels_first\". ordering dimensions inputs. \"channels_last\" corresponds inputs shape (batch, steps, features) \"channels_first\" corresponds inputs shape (batch, features, steps). defaults image_data_format value found Keras config file ~/.keras/keras.json. never set , \"channels_last\". dilation_rate int list 1 integers, specifying dilation rate use dilated convolution. one int specified, dilation rate used dimensions. depth_multiplier number depthwise convolution output channels input channel. total number depthwise convolution output channels equal input_channel * depth_multiplier. activation Activation function. NULL, activation applied. use_bias bool, TRUE, bias added output. depthwise_initializer initializer depthwise convolution kernel. NULL, default initializer (\"glorot_uniform\") used. pointwise_initializer initializer pointwise convolution kernel. NULL, default initializer (\"glorot_uniform\") used. bias_initializer initializer bias vector. NULL, default initializer ('\"zeros\"') used. depthwise_regularizer Optional regularizer depthwise convolution kernel. pointwise_regularizer Optional regularizer pointwise convolution kernel. bias_regularizer Optional regularizer bias vector. activity_regularizer Optional regularizer function output. depthwise_constraint Optional projection function applied depthwise kernel updated Optimizer (e.g. used norm constraints value constraints layer weights). function must take input unprojected variable must return projected variable (must shape). pointwise_constraint Optional projection function applied pointwise kernel updated Optimizer. bias_constraint Optional projection function applied bias updated Optimizer. ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_separable_conv_1d.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"1D separable convolution layer. — layer_separable_conv_1d","text":"3D tensor representing activation(separable_conv1d(inputs, kernel) + bias).","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_separable_conv_1d.html","id":"input-shape","dir":"Reference","previous_headings":"","what":"Input Shape","title":"1D separable convolution layer. — layer_separable_conv_1d","text":"data_format=\"channels_last\": 3D tensor shape: (batch_shape, steps, channels) data_format=\"channels_first\": 3D tensor shape: (batch_shape, channels, steps)","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_separable_conv_1d.html","id":"output-shape","dir":"Reference","previous_headings":"","what":"Output Shape","title":"1D separable convolution layer. — layer_separable_conv_1d","text":"data_format=\"channels_last\": 3D tensor shape: (batch_shape, new_steps, filters) data_format=\"channels_first\": 3D tensor shape: (batch_shape, filters, new_steps)","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_separable_conv_1d.html","id":"example","dir":"Reference","previous_headings":"","what":"Example","title":"1D separable convolution layer. — layer_separable_conv_1d","text":"","code":"x <- random_uniform(c(4, 10, 12)) y <- layer_separable_conv_1d(x, 3, 2, 2, activation='relu') shape(y) ## shape(4, 5, 3)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_separable_conv_2d.html","id":null,"dir":"Reference","previous_headings":"","what":"2D separable convolution layer. — layer_separable_conv_2d","title":"2D separable convolution layer. — layer_separable_conv_2d","text":"layer performs depthwise convolution acts separately channels, followed pointwise convolution mixes channels. use_bias TRUE bias initializer provided, adds bias vector output. optionally applies activation function produce final output.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_separable_conv_2d.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"2D separable convolution layer. — layer_separable_conv_2d","text":"","code":"layer_separable_conv_2d(   object,   filters,   kernel_size,   strides = list(1L, 1L),   padding = \"valid\",   data_format = NULL,   dilation_rate = list(1L, 1L),   depth_multiplier = 1L,   activation = NULL,   use_bias = TRUE,   depthwise_initializer = \"glorot_uniform\",   pointwise_initializer = \"glorot_uniform\",   bias_initializer = \"zeros\",   depthwise_regularizer = NULL,   pointwise_regularizer = NULL,   bias_regularizer = NULL,   activity_regularizer = NULL,   depthwise_constraint = NULL,   pointwise_constraint = NULL,   bias_constraint = NULL,   ... )"},{"path":"https://keras3.posit.co/dev/reference/layer_separable_conv_2d.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"2D separable convolution layer. — layer_separable_conv_2d","text":"object Object compose layer . tensor, array, sequential model. filters int, dimensionality output space (.e. number filters pointwise convolution). kernel_size int list 2 integers, specifying size depthwise convolution window. strides int list 2 integers, specifying stride length depthwise convolution. one int specified, stride size used dimensions. strides > 1 incompatible dilation_rate > 1. padding string, either \"valid\" \"\" (case-insensitive). \"valid\" means padding. \"\" results padding evenly left/right /input. padding=\"\" strides=1, output size input. data_format string, either \"channels_last\" \"channels_first\". ordering dimensions inputs. \"channels_last\" corresponds inputs shape (batch, height, width, channels) \"channels_first\" corresponds inputs shape (batch, channels, height, width). defaults image_data_format value found Keras config file ~/.keras/keras.json. never set , \"channels_last\". dilation_rate int list 2 integers, specifying dilation rate use dilated convolution. one int specified, dilation rate used dimensions. depth_multiplier number depthwise convolution output channels input channel. total number depthwise convolution output channels equal input_channel * depth_multiplier. activation Activation function. NULL, activation applied. use_bias bool, TRUE, bias added output. depthwise_initializer initializer depthwise convolution kernel. NULL, default initializer (\"glorot_uniform\") used. pointwise_initializer initializer pointwise convolution kernel. NULL, default initializer (\"glorot_uniform\") used. bias_initializer initializer bias vector. NULL, default initializer ('\"zeros\"') used. depthwise_regularizer Optional regularizer depthwise convolution kernel. pointwise_regularizer Optional regularizer pointwise convolution kernel. bias_regularizer Optional regularizer bias vector. activity_regularizer Optional regularizer function output. depthwise_constraint Optional projection function applied depthwise kernel updated Optimizer (e.g. used norm constraints value constraints layer weights). function must take input unprojected variable must return projected variable (must shape). pointwise_constraint Optional projection function applied pointwise kernel updated Optimizer. bias_constraint Optional projection function applied bias updated Optimizer. ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_separable_conv_2d.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"2D separable convolution layer. — layer_separable_conv_2d","text":"4D tensor representing activation(separable_conv2d(inputs, kernel) + bias).","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_separable_conv_2d.html","id":"input-shape","dir":"Reference","previous_headings":"","what":"Input Shape","title":"2D separable convolution layer. — layer_separable_conv_2d","text":"data_format=\"channels_last\": 4D tensor shape: (batch_size, height, width, channels) data_format=\"channels_first\": 4D tensor shape: (batch_size, channels, height, width)","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_separable_conv_2d.html","id":"output-shape","dir":"Reference","previous_headings":"","what":"Output Shape","title":"2D separable convolution layer. — layer_separable_conv_2d","text":"data_format=\"channels_last\": 4D tensor shape: (batch_size, new_height, new_width, filters) data_format=\"channels_first\": 4D tensor shape: (batch_size, filters, new_height, new_width)","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_separable_conv_2d.html","id":"example","dir":"Reference","previous_headings":"","what":"Example","title":"2D separable convolution layer. — layer_separable_conv_2d","text":"","code":"x <- random_uniform(c(4, 10, 10, 12)) y <- layer_separable_conv_2d(x, 3, c(4, 3), 2, activation='relu') shape(y) ## shape(4, 4, 4, 3)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_simple_rnn.html","id":null,"dir":"Reference","previous_headings":"","what":"Fully-connected RNN where the output is to be fed back as the new input. — layer_simple_rnn","title":"Fully-connected RNN where the output is to be fed back as the new input. — layer_simple_rnn","text":"Fully-connected RNN output fed back new input.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_simple_rnn.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Fully-connected RNN where the output is to be fed back as the new input. — layer_simple_rnn","text":"","code":"layer_simple_rnn(   object,   units,   activation = \"tanh\",   use_bias = TRUE,   kernel_initializer = \"glorot_uniform\",   recurrent_initializer = \"orthogonal\",   bias_initializer = \"zeros\",   kernel_regularizer = NULL,   recurrent_regularizer = NULL,   bias_regularizer = NULL,   activity_regularizer = NULL,   kernel_constraint = NULL,   recurrent_constraint = NULL,   bias_constraint = NULL,   dropout = 0,   recurrent_dropout = 0,   return_sequences = FALSE,   return_state = FALSE,   go_backwards = FALSE,   stateful = FALSE,   unroll = FALSE,   seed = NULL,   ... )"},{"path":"https://keras3.posit.co/dev/reference/layer_simple_rnn.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Fully-connected RNN where the output is to be fed back as the new input. — layer_simple_rnn","text":"object Object compose layer . tensor, array, sequential model. units Positive integer, dimensionality output space. activation Activation function use. Default: hyperbolic tangent (tanh). pass NULL, activation applied (ie. \"linear\" activation: (x) = x). use_bias Boolean, (default TRUE), whether layer uses bias vector. kernel_initializer Initializer kernel weights matrix, used linear transformation inputs. Default: \"glorot_uniform\". recurrent_initializer Initializer recurrent_kernel weights matrix, used linear transformation recurrent state.  Default: \"orthogonal\". bias_initializer Initializer bias vector. Default: \"zeros\". kernel_regularizer Regularizer function applied kernel weights matrix. Default: NULL. recurrent_regularizer Regularizer function applied recurrent_kernel weights matrix. Default: NULL. bias_regularizer Regularizer function applied bias vector. Default: NULL. activity_regularizer Regularizer function applied output layer (\"activation\"). Default: NULL. kernel_constraint Constraint function applied kernel weights matrix. Default: NULL. recurrent_constraint Constraint function applied recurrent_kernel weights matrix.  Default: NULL. bias_constraint Constraint function applied bias vector. Default: NULL. dropout Float 0 1. Fraction units drop linear transformation inputs. Default: 0. recurrent_dropout Float 0 1. Fraction units drop linear transformation recurrent state. Default: 0. return_sequences Boolean. Whether return last output output sequence, full sequence. Default: FALSE. return_state Boolean. Whether return last state addition output. Default: FALSE. go_backwards Boolean (default: FALSE). TRUE, process input sequence backwards return reversed sequence. stateful Boolean (default: FALSE). TRUE, last state sample index batch used initial state sample index following batch. unroll Boolean (default: FALSE). TRUE, network unrolled, else symbolic loop used. Unrolling can speed-RNN, although tends memory-intensive. Unrolling suitable short sequences. seed Initial seed random number generator ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_simple_rnn.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Fully-connected RNN where the output is to be fed back as the new input. — layer_simple_rnn","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_simple_rnn.html","id":"call-arguments","dir":"Reference","previous_headings":"","what":"Call Arguments","title":"Fully-connected RNN where the output is to be fed back as the new input. — layer_simple_rnn","text":"sequence: 3D tensor, shape [batch, timesteps, feature]. mask: Binary tensor shape [batch, timesteps] indicating whether given timestep masked. individual TRUE entry indicates corresponding timestep utilized, FALSE entry indicates corresponding timestep ignored. training: Python boolean indicating whether layer behave training mode inference mode. argument passed cell calling . relevant dropout recurrent_dropout used. initial_state: List initial state tensors passed first call cell.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_simple_rnn.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Fully-connected RNN where the output is to be fed back as the new input. — layer_simple_rnn","text":"","code":"inputs <- random_uniform(c(32, 10, 8)) simple_rnn <- layer_simple_rnn(units = 4) output <- simple_rnn(inputs)  # The output has shape `(32, 4)`. simple_rnn <- layer_simple_rnn(     units = 4, return_sequences=TRUE, return_state=TRUE ) # whole_sequence_output has shape `(32, 10, 4)`. # final_state has shape `(32, 4)`. c(whole_sequence_output, final_state) %<-% simple_rnn(inputs)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_solarization.html","id":null,"dir":"Reference","previous_headings":"","what":"Applies (max_value - pixel + min_value) for each pixel in the image. — layer_solarization","title":"Applies (max_value - pixel + min_value) for each pixel in the image. — layer_solarization","text":"created without threshold parameter, layer performs solarization values. created specified threshold layer augments pixels threshold value.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_solarization.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Applies (max_value - pixel + min_value) for each pixel in the image. — layer_solarization","text":"","code":"layer_solarization(   object,   addition_factor = 0,   threshold_factor = 0,   value_range = tuple(0L, 255L),   seed = NULL,   ... )"},{"path":"https://keras3.posit.co/dev/reference/layer_solarization.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Applies (max_value - pixel + min_value) for each pixel in the image. — layer_solarization","text":"object Object compose layer . tensor, array, sequential model. addition_factor (Optional)  tuple two floats single float, 0 1. augmented image value sampled provided range. float passed, range interpreted (0, addition_factor). specified, value (times value range input images, e.g. 255), added pixel solarization thresholding. Defaults 0.0. threshold_factor (Optional)  tuple two floats single float. augmented image value sampled provided range. float passed, range interpreted (0, threshold_factor). specified, pixel values threshold solarized. value_range tuple list two elements. first value represents lower bound values input images, second represents upper bound. Images passed layer values within value_range. Typical values pass (0, 255) (RGB image) (0., 1.) (scaled image). seed Integer. Used create random seed. ... Base layer keyword arguments, name dtype.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_solarization.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Applies (max_value - pixel + min_value) for each pixel in the image. — layer_solarization","text":"","code":"c(c(images, labels), .) %<-% dataset_cifar10() str(images) ##  int [1:50000, 1:32, 1:32, 1:3] 59 154 255 28 170 159 164 28 134 125 ... str(images[1, 1, 1, ]) ##  int [1:3] 59 62 63 # Note that images are Tensor with values in the range [0, 255] solarization <- layer_solarization(value_range = c(0, 255)) images <- solarization(images) |> as.array() str(images[1, 1, 1, ]) ##  num [1:3] 196 193 192"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_spatial_dropout_1d.html","id":null,"dir":"Reference","previous_headings":"","what":"Spatial 1D version of Dropout. — layer_spatial_dropout_1d","title":"Spatial 1D version of Dropout. — layer_spatial_dropout_1d","text":"layer performs function Dropout, however, drops entire 1D feature maps instead individual elements. adjacent frames within feature maps strongly correlated (normally case early convolution layers) regular dropout regularize activations otherwise just result effective learning rate decrease. case, SpatialDropout1D help promote independence feature maps used instead.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_spatial_dropout_1d.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Spatial 1D version of Dropout. — layer_spatial_dropout_1d","text":"","code":"layer_spatial_dropout_1d(object, rate, seed = NULL, name = NULL, dtype = NULL)"},{"path":"https://keras3.posit.co/dev/reference/layer_spatial_dropout_1d.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Spatial 1D version of Dropout. — layer_spatial_dropout_1d","text":"object Object compose layer . tensor, array, sequential model. rate Float 0 1. Fraction input units drop. seed Initial seed random number generator name String, name object dtype datatype (e.g., \"float32\").","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_spatial_dropout_1d.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Spatial 1D version of Dropout. — layer_spatial_dropout_1d","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_spatial_dropout_1d.html","id":"call-arguments","dir":"Reference","previous_headings":"","what":"Call Arguments","title":"Spatial 1D version of Dropout. — layer_spatial_dropout_1d","text":"inputs: 3D tensor. training: Python boolean indicating whether layer behave training mode (applying dropout) inference mode (pass-).","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_spatial_dropout_1d.html","id":"input-shape","dir":"Reference","previous_headings":"","what":"Input Shape","title":"Spatial 1D version of Dropout. — layer_spatial_dropout_1d","text":"3D tensor shape: (samples, timesteps, channels)","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_spatial_dropout_1d.html","id":"output-shape","dir":"Reference","previous_headings":"","what":"Output Shape","title":"Spatial 1D version of Dropout. — layer_spatial_dropout_1d","text":"input.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_spatial_dropout_1d.html","id":"reference","dir":"Reference","previous_headings":"","what":"Reference","title":"Spatial 1D version of Dropout. — layer_spatial_dropout_1d","text":"Tompson et al., 2014","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_spatial_dropout_2d.html","id":null,"dir":"Reference","previous_headings":"","what":"Spatial 2D version of Dropout. — layer_spatial_dropout_2d","title":"Spatial 2D version of Dropout. — layer_spatial_dropout_2d","text":"version performs function Dropout, however, drops entire 2D feature maps instead individual elements. adjacent pixels within feature maps strongly correlated (normally case early convolution layers) regular dropout regularize activations otherwise just result effective learning rate decrease. case, SpatialDropout2D help promote independence feature maps used instead.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_spatial_dropout_2d.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Spatial 2D version of Dropout. — layer_spatial_dropout_2d","text":"","code":"layer_spatial_dropout_2d(   object,   rate,   data_format = NULL,   seed = NULL,   name = NULL,   dtype = NULL )"},{"path":"https://keras3.posit.co/dev/reference/layer_spatial_dropout_2d.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Spatial 2D version of Dropout. — layer_spatial_dropout_2d","text":"object Object compose layer . tensor, array, sequential model. rate Float 0 1. Fraction input units drop. data_format \"channels_first\" \"channels_last\". \"channels_first\" mode, channels dimension (depth) index 1, \"channels_last\" mode index 3. defaults image_data_format value found Keras config file ~/.keras/keras.json. never set , \"channels_last\". seed Initial seed random number generator name String, name object dtype datatype (e.g., \"float32\").","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_spatial_dropout_2d.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Spatial 2D version of Dropout. — layer_spatial_dropout_2d","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_spatial_dropout_2d.html","id":"call-arguments","dir":"Reference","previous_headings":"","what":"Call Arguments","title":"Spatial 2D version of Dropout. — layer_spatial_dropout_2d","text":"inputs: 4D tensor. training: Python boolean indicating whether layer behave training mode (applying dropout) inference mode (pass-).","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_spatial_dropout_2d.html","id":"input-shape","dir":"Reference","previous_headings":"","what":"Input Shape","title":"Spatial 2D version of Dropout. — layer_spatial_dropout_2d","text":"4D tensor shape: (samples, channels, rows, cols) data_format='channels_first' 4D tensor shape: (samples, rows, cols, channels) data_format='channels_last'.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_spatial_dropout_2d.html","id":"output-shape","dir":"Reference","previous_headings":"","what":"Output Shape","title":"Spatial 2D version of Dropout. — layer_spatial_dropout_2d","text":"input.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_spatial_dropout_2d.html","id":"reference","dir":"Reference","previous_headings":"","what":"Reference","title":"Spatial 2D version of Dropout. — layer_spatial_dropout_2d","text":"Tompson et al., 2014","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_spatial_dropout_3d.html","id":null,"dir":"Reference","previous_headings":"","what":"Spatial 3D version of Dropout. — layer_spatial_dropout_3d","title":"Spatial 3D version of Dropout. — layer_spatial_dropout_3d","text":"version performs function Dropout, however, drops entire 3D feature maps instead individual elements. adjacent voxels within feature maps strongly correlated (normally case early convolution layers) regular dropout regularize activations otherwise just result effective learning rate decrease. case, SpatialDropout3D help promote independence feature maps used instead.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_spatial_dropout_3d.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Spatial 3D version of Dropout. — layer_spatial_dropout_3d","text":"","code":"layer_spatial_dropout_3d(   object,   rate,   data_format = NULL,   seed = NULL,   name = NULL,   dtype = NULL )"},{"path":"https://keras3.posit.co/dev/reference/layer_spatial_dropout_3d.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Spatial 3D version of Dropout. — layer_spatial_dropout_3d","text":"object Object compose layer . tensor, array, sequential model. rate Float 0 1. Fraction input units drop. data_format \"channels_first\" \"channels_last\". \"channels_first\" mode, channels dimension (depth) index 1, \"channels_last\" mode index 4. defaults image_data_format value found Keras config file ~/.keras/keras.json. never set , \"channels_last\". seed Initial seed random number generator name String, name object dtype datatype (e.g., \"float32\").","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_spatial_dropout_3d.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Spatial 3D version of Dropout. — layer_spatial_dropout_3d","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_spatial_dropout_3d.html","id":"call-arguments","dir":"Reference","previous_headings":"","what":"Call Arguments","title":"Spatial 3D version of Dropout. — layer_spatial_dropout_3d","text":"inputs: 5D tensor. training: Python boolean indicating whether layer behave training mode (applying dropout) inference mode (pass-).","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_spatial_dropout_3d.html","id":"input-shape","dir":"Reference","previous_headings":"","what":"Input Shape","title":"Spatial 3D version of Dropout. — layer_spatial_dropout_3d","text":"5D tensor shape: (samples, channels, dim1, dim2, dim3) data_format='channels_first' 5D tensor shape: (samples, dim1, dim2, dim3, channels) data_format='channels_last'.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_spatial_dropout_3d.html","id":"output-shape","dir":"Reference","previous_headings":"","what":"Output Shape","title":"Spatial 3D version of Dropout. — layer_spatial_dropout_3d","text":"input.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_spatial_dropout_3d.html","id":"reference","dir":"Reference","previous_headings":"","what":"Reference","title":"Spatial 3D version of Dropout. — layer_spatial_dropout_3d","text":"Tompson et al., 2014","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_spectral_normalization.html","id":null,"dir":"Reference","previous_headings":"","what":"Performs spectral normalization on the weights of a target layer. — layer_spectral_normalization","title":"Performs spectral normalization on the weights of a target layer. — layer_spectral_normalization","text":"wrapper controls Lipschitz constant weights layer constraining spectral norm, can stabilize training GANs.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_spectral_normalization.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Performs spectral normalization on the weights of a target layer. — layer_spectral_normalization","text":"","code":"layer_spectral_normalization(object, layer, power_iterations = 1L, ...)"},{"path":"https://keras3.posit.co/dev/reference/layer_spectral_normalization.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Performs spectral normalization on the weights of a target layer. — layer_spectral_normalization","text":"object Object compose layer . tensor, array, sequential model. layer Layer instance either kernel (e.g. layer_conv_2d, layer_dense...) embeddings attribute (layer_embedding layer). power_iterations int, number iterations normalization. ... Base wrapper keyword arguments.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_spectral_normalization.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Performs spectral normalization on the weights of a target layer. — layer_spectral_normalization","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_spectral_normalization.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Performs spectral normalization on the weights of a target layer. — layer_spectral_normalization","text":"Wrap layer_conv_2d:     Wrap layer_dense:","code":"x <- random_uniform(c(1, 10, 10, 1)) conv2d <- layer_spectral_normalization(   layer = layer_conv_2d(filters = 2, kernel_size = 2) ) y <- conv2d(x) shape(y) ## shape(1, 9, 9, 2) x <- random_uniform(c(1, 10, 10, 1)) dense <- layer_spectral_normalization(layer = layer_dense(units = 10)) y <- dense(x) shape(y) ## shape(1, 10, 10, 10)"},{"path":"https://keras3.posit.co/dev/reference/layer_spectral_normalization.html","id":"reference","dir":"Reference","previous_headings":"","what":"Reference","title":"Performs spectral normalization on the weights of a target layer. — layer_spectral_normalization","text":"Spectral Normalization GAN.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_stft_spectrogram.html","id":null,"dir":"Reference","previous_headings":"","what":"Layer to compute the Short-Time Fourier Transform (STFT) on a 1D signal. — layer_stft_spectrogram","title":"Layer to compute the Short-Time Fourier Transform (STFT) on a 1D signal. — layer_stft_spectrogram","text":"layer computes Spectrograms input signal produce spectrogram. layers layer computes Spectrograms based Short-Time Fourier Transform (STFT) utilizing convolution kernels, allows parallelization GPUs trainable kernels fine-tuning support. layer allows different modes output (e.g., log-scaled magnitude, phase, power spectral density, etc.) provides flexibility windowing, padding, scaling options STFT calculation.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_stft_spectrogram.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Layer to compute the Short-Time Fourier Transform (STFT) on a 1D signal. — layer_stft_spectrogram","text":"","code":"layer_stft_spectrogram(   object,   mode = \"log\",   frame_length = 256L,   frame_step = NULL,   fft_length = NULL,   window = \"hann\",   periodic = FALSE,   scaling = \"density\",   padding = \"valid\",   expand_dims = FALSE,   data_format = NULL,   ... )"},{"path":"https://keras3.posit.co/dev/reference/layer_stft_spectrogram.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Layer to compute the Short-Time Fourier Transform (STFT) on a 1D signal. — layer_stft_spectrogram","text":"object Object compose layer . tensor, array, sequential model. mode String, output type spectrogram. Can one \"log\", \"magnitude\", \"psd\", \"real\", \"imag\", \"angle\", \"stft\". Defaults \"log\". frame_length Integer, length frame (window) STFT samples. Defaults 256. frame_step Integer, step size (hop length) consecutive frames. provided, defaults half frame_length. Defaults frame_length %/% 2. fft_length Integer, size frequency bins used Fast-Fourier Transform (FFT) apply frame. greater equal frame_length.  Recommended power two. Defaults smallest power two greater equal frame_length. window (String array_like), windowing function apply frame. Can \"hann\" (default), \"hamming\", custom window provided array_like. periodic Boolean, TRUE, window function treated periodic. Defaults FALSE. scaling String, type scaling applied window. Can \"density\", \"spectrum\", None. Default \"density\". padding String, padding strategy. Can \"valid\" \"\". Defaults \"valid\". expand_dims Boolean, TRUE, expand output spectrograms two dimensions compatible image models. Defaults FALSE. data_format String, either \"channels_last\" \"channels_first\". ordering dimensions inputs. \"channels_last\" corresponds inputs shape (batch, height, width, channels) \"channels_first\" corresponds inputs shape (batch, channels, height, weight). Defaults \"channels_last\". ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_stft_spectrogram.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Layer to compute the Short-Time Fourier Transform (STFT) on a 1D signal. — layer_stft_spectrogram","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_stft_spectrogram.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Layer to compute the Short-Time Fourier Transform (STFT) on a 1D signal. — layer_stft_spectrogram","text":"Apply non-trainable preprocessing layer 3 audio tracks 1 channel, 10 seconds sampled 16 kHz.     Apply trainable processing layer 3 stereo audio tracks 2 channels, 10 seconds sampled 16 kHz. initialized non-trainable layer, can trained jointly within model.     Similar last example, add extra dimension output image used image models. apply signal 3 input channels output image tensor, hence directly applicable image model.","code":"layer <- layer_stft_spectrogram(   mode = 'log',   frame_length = 256,   frame_step = 128, # 50% overlap   fft_length = 512,   window = \"hann\",   padding = \"valid\",   trainable = FALSE # non-trainable, preprocessing only) ) random_uniform(shape=c(3, 160000, 1)) |> layer() |> op_shape() ## shape(3, 1249, 257) layer <- layer_stft_spectrogram(   mode = 'log',   frame_length = 256,   frame_step = 128,   # 50% overlap   fft_length = 512,   window = \"hamming\", # hamming windowing function   padding = \"same\",   # padding to preserve the time dimension   trainable = TRUE,   # trainable, this is the default in keras ) random_uniform(shape=c(3, 160000, 2)) |> layer() |> op_shape() ## shape(3, 1250, 514) layer <- layer_stft_spectrogram(   mode = 'log',   frame_length = 256,   frame_step = 128,   fft_length = 512,   padding = \"same\",   expand_dims = TRUE  # this adds the extra dimension ) random_uniform(shape=c(3, 160000, 3)) |> layer() |> op_shape() ## shape(3, 1250, 257, 3)"},{"path":"https://keras3.posit.co/dev/reference/layer_stft_spectrogram.html","id":"raises","dir":"Reference","previous_headings":"","what":"Raises","title":"Layer to compute the Short-Time Fourier Transform (STFT) on a 1D signal. — layer_stft_spectrogram","text":"ValueError: invalid value provided \"mode\", \"scaling\", \"padding\", input arguments. TypeError: input data type one \"float16\", \"float32\", \"float64\".","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_stft_spectrogram.html","id":"input-shape","dir":"Reference","previous_headings":"","what":"Input Shape","title":"Layer to compute the Short-Time Fourier Transform (STFT) on a 1D signal. — layer_stft_spectrogram","text":"3D tensor shape (batch_size, time_length, input_channels), data_format==\"channels_last\", shape (batch_size, input_channels, time_length) data_format==\"channels_first\", time_length length input signal, input_channels number input channels. kernels applied channel independently.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_stft_spectrogram.html","id":"output-shape","dir":"Reference","previous_headings":"","what":"Output Shape","title":"Layer to compute the Short-Time Fourier Transform (STFT) on a 1D signal. — layer_stft_spectrogram","text":"data_format==\"channels_first\" && !expand_dims, 3D tensor: (batch_size, input_channels * freq_channels, new_time_length) data_format==\"channels_last\" && !expand_dims, 3D tensor: (batch_size, new_time_length, input_channels * freq_channels) data_format==\"channels_first\" && expand_dims, 4D tensor: (batch_size, input_channels, new_time_length, freq_channels) data_format==\"channels_last\" && expand_dims, 4D tensor: (batch_size, new_time_length, freq_channels, input_channels) new_time_length depends padding, freq_channels number FFT bins (fft_length %/% 2 + 1).","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_string_lookup.html","id":null,"dir":"Reference","previous_headings":"","what":"A preprocessing layer that maps strings to (possibly encoded) indices. — layer_string_lookup","title":"A preprocessing layer that maps strings to (possibly encoded) indices. — layer_string_lookup","text":"layer translates set arbitrary strings integer output via table-based vocabulary lookup. layer perform splitting transformation input strings. layer can split tokenize natural language, see layer_text_vectorization. vocabulary layer must either supplied construction learned via adapt(). adapt(), layer analyze data set, determine frequency individual strings tokens, create vocabulary . vocabulary capped size, frequent tokens used create vocabulary others treated --vocabulary (OOV). two possible output modes layer. output_mode \"int\", input strings converted (0-based) index vocabulary (integer). output_mode \"multi_hot\", \"count\", \"tf_idf\", input strings encoded array dimension corresponds element vocabulary. vocabulary can optionally contain mask token well OOV token (can optionally occupy multiple indices vocabulary, set num_oov_indices). position tokens vocabulary fixed. output_mode \"int\", vocabulary begin mask token (set), followed OOV indices, followed rest vocabulary. output_mode \"multi_hot\", \"count\", \"tf_idf\" vocabulary begin OOV indices instances mask token dropped. Note: layer uses TensorFlow internally. used part compiled computation graph model backend TensorFlow. can however used backend running eagerly. can also always used part input preprocessing pipeline backend (outside model ), recommend use layer. Note: layer safe use inside tf.data pipeline (independently backend using). Note: working layer outputs directly (e.g., passing outputs another layer, using lower-level operations like [ op_*): returned indices 0-based. However, default settings, first (0) index OOV token, returned indices offset 1 may appear 1-based.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_string_lookup.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"A preprocessing layer that maps strings to (possibly encoded) indices. — layer_string_lookup","text":"","code":"layer_string_lookup(   object,   max_tokens = NULL,   num_oov_indices = 1L,   mask_token = NULL,   oov_token = \"[UNK]\",   vocabulary = NULL,   idf_weights = NULL,   invert = FALSE,   output_mode = \"int\",   pad_to_max_tokens = FALSE,   sparse = FALSE,   encoding = \"utf-8\",   name = NULL,   ...,   vocabulary_dtype = NULL )"},{"path":"https://keras3.posit.co/dev/reference/layer_string_lookup.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"A preprocessing layer that maps strings to (possibly encoded) indices. — layer_string_lookup","text":"object Object compose layer . tensor, array, sequential model. max_tokens Maximum size vocabulary layer. specified adapting vocabulary setting pad_to_max_tokens=TRUE. NULL, cap size vocabulary. Note size includes OOV mask tokens. Defaults NULL. num_oov_indices number --vocabulary tokens use. value 1, OOV inputs modulated determine OOV value. value 0, OOV inputs cause error calling layer. Defaults 1. mask_token token represents masked inputs. output_mode \"int\", token included vocabulary mapped index 0. output modes, token appear vocabulary instances mask token input dropped. set NULL, mask term added. Defaults NULL. oov_token used invert TRUE. token return OOV indices. Defaults \"[UNK]\". vocabulary Optional. Either array integers string path text file. passing array, can pass list, list, 1D NumPy array, 1D tensor containing integer vocabulary terms. passing file path, file contain one line per term vocabulary. argument set, need adapt() layer. idf_weights valid output_mode \"tf_idf\". list, list, 1D NumPy array, 1D tensor length vocabulary, containing floating point inverse document frequency weights, multiplied per sample term counts final TF-IDF weight. vocabulary argument set, output_mode \"tf_idf\", argument must supplied. invert valid output_mode \"int\". TRUE, layer map indices vocabulary items instead mapping vocabulary items indices. Defaults FALSE. output_mode Specification output layer. Values can \"int\", \"one_hot\", \"multi_hot\", \"count\", \"tf_idf\" configuring layer follows: \"int\": Return vocabulary indices input tokens. \"one_hot\": Encodes individual element input array size vocabulary, containing 1 element index. last dimension size 1, encode dimension. last dimension size 1, append new dimension encoded output. \"multi_hot\": Encodes sample input single array size vocabulary, containing 1 vocabulary term present sample. Treats last dimension sample dimension, input shape (..., sample_length), output shape (..., num_tokens). \"count\": \"multi_hot\", int array contains count number times token index appeared sample. \"tf_idf\": \"multi_hot\", TF-IDF algorithm applied find value token slot. \"int\" output, shape input output supported. output modes, currently output rank 2 supported. Defaults \"int\". pad_to_max_tokens applicable output_mode \"multi_hot\", \"count\", \"tf_idf\". TRUE, output feature axis padded max_tokens even number unique tokens vocabulary less max_tokens, resulting tensor shape (batch_size, max_tokens) regardless vocabulary size. Defaults FALSE. sparse Boolean. applicable \"multi_hot\", \"count\", \"tf_idf\" output modes. supported TensorFlow backend. TRUE, returns SparseTensor instead dense Tensor. Defaults FALSE. encoding Optional. text encoding use interpret input strings. Defaults \"utf-8\". name String, name object ... forward/backward compatability. vocabulary_dtype dtype vocabulary terms, example \"int64\" \"int32\". Defaults \"int64\".","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_string_lookup.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"A preprocessing layer that maps strings to (possibly encoded) indices. — layer_string_lookup","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_string_lookup.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"A preprocessing layer that maps strings to (possibly encoded) indices. — layer_string_lookup","text":"Creating lookup layer known vocabulary example creates lookup layer pre-existing vocabulary.     Creating lookup layer adapted vocabulary example creates lookup layer generates vocabulary analyzing dataset.     Note OOV token \"[UNK]\" added vocabulary. remaining tokens sorted frequency (\"d\", 2 occurrences, first) inverse sort order.     Lookups multiple OOV indices example demonstrates use lookup layer multiple OOV indices.  layer created one OOV index, OOV values hashed number OOV buckets, distributing OOV values deterministic fashion across set.     Note output OOV value 'm' 0, output OOV value \"z\" 1. -vocab terms output index increased 1 earlier examples (maps 2, etc) order make space extra OOV value. One-hot output Configure layer output_mode='one_hot'. Note first num_oov_indices dimensions ont_hot encoding represent OOV values.     Multi-hot output Configure layer output_mode='multi_hot'. Note first num_oov_indices dimensions multi_hot encoding represent OOV values.     Token count output Configure layer output_mode='count'. multi_hot output, first num_oov_indices dimensions output represent OOV values.     TF-IDF output Configure layer output_mode=\"tf_idf\". multi_hot output, first num_oov_indices dimensions output represent OOV values. token bin output token_count * idf_weight, idf weights inverse document frequency weights per token. provided along vocabulary. Note idf_weight OOV values default average idf weights passed .     specify idf weights oov values, need pass entire vocabulary including leading oov token.     adapting layer \"tf_idf\" mode, input sample considered document, IDF weight per token calculated log(1 + num_documents / (1 + token_document_count)). Inverse lookup example demonstrates map indices strings using layer. (can also use adapt() inverse=TRUE, simplicity pass vocab example.)     Note first index correspond oov token default. Forward inverse lookup pairs example demonstrates use vocabulary standard lookup layer create inverse lookup layer.     example, input value \"z\" resulted output \"[UNK]\", since 1000 vocabulary - got represented OOV, OOV values returned \"[UNK]\" inverse layer. Also, note inverse work, must already set forward layer vocabulary either directly via adapt() calling get_vocabulary().","code":"vocab <- c(\"a\", \"b\", \"c\", \"d\") data <- rbind(c(\"a\", \"c\", \"d\"), c(\"d\", \"z\", \"b\")) layer <- layer_string_lookup(vocabulary=vocab) layer(data) ## tf.Tensor( ## [[1 3 4] ##  [4 0 2]], shape=(2, 3), dtype=int64) data <- rbind(c(\"a\", \"c\", \"d\"), c(\"d\", \"z\", \"b\")) layer <- layer_string_lookup() layer %>% adapt(data) get_vocabulary(layer) ## [1] \"[UNK]\" \"d\"     \"z\"     \"c\"     \"b\"     \"a\" data <- rbind(c(\"a\", \"c\", \"d\"), c(\"d\", \"z\", \"b\")) layer <- layer_string_lookup() layer %>% adapt(data) layer(data) ## tf.Tensor( ## [[5 3 1] ##  [1 2 4]], shape=(2, 3), dtype=int64) vocab <- c(\"a\", \"b\", \"c\", \"d\") data <- rbind(c(\"a\", \"c\", \"d\"), c(\"m\", \"z\", \"b\")) layer <- layer_string_lookup(vocabulary = vocab, num_oov_indices = 2) layer(data) ## tf.Tensor( ## [[2 4 5] ##  [0 1 3]], shape=(2, 3), dtype=int64) vocab <- c(\"a\", \"b\", \"c\", \"d\") data <- c(\"a\", \"b\", \"c\", \"d\", \"z\") layer <- layer_string_lookup(vocabulary = vocab, output_mode = 'one_hot') layer(data) ## tf.Tensor( ## [[0 1 0 0 0] ##  [0 0 1 0 0] ##  [0 0 0 1 0] ##  [0 0 0 0 1] ##  [1 0 0 0 0]], shape=(5, 5), dtype=int64) vocab <- c(\"a\", \"b\", \"c\", \"d\") data <- rbind(c(\"a\", \"c\", \"d\", \"d\"), c(\"d\", \"z\", \"b\", \"z\")) layer <- layer_string_lookup(vocabulary = vocab, output_mode = 'multi_hot') layer(data) ## tf.Tensor( ## [[0 1 0 1 1] ##  [1 0 1 0 1]], shape=(2, 5), dtype=int64) vocab <- c(\"a\", \"b\", \"c\", \"d\") data <- rbind(c(\"a\", \"c\", \"d\", \"d\"), c(\"d\", \"z\", \"b\", \"z\")) layer <- layer_string_lookup(vocabulary = vocab, output_mode = 'count') layer(data) ## tf.Tensor( ## [[0 1 0 1 2] ##  [2 0 1 0 1]], shape=(2, 5), dtype=int64) vocab <- c(\"a\", \"b\", \"c\", \"d\") idf_weights <- c(0.25, 0.75, 0.6, 0.4) data <- rbind(c(\"a\", \"c\", \"d\", \"d\"), c(\"d\", \"z\", \"b\", \"z\")) layer <- layer_string_lookup(output_mode = \"tf_idf\") layer %>% set_vocabulary(vocab, idf_weights=idf_weights) layer(data) ## tf.Tensor( ## [[0.   0.25 0.   0.6  0.8 ] ##  [1.   0.   0.75 0.   0.4 ]], shape=(2, 5), dtype=float32) vocab <- c(\"[UNK]\", \"a\", \"b\", \"c\", \"d\") idf_weights <- c(0.9, 0.25, 0.75, 0.6, 0.4) data <- rbind(c(\"a\", \"c\", \"d\", \"d\"), c(\"d\", \"z\", \"b\", \"z\")) layer <- layer_string_lookup(output_mode = \"tf_idf\") layer %>% set_vocabulary(vocab, idf_weights=idf_weights) layer(data) ## tf.Tensor( ## [[0.   0.25 0.   0.6  0.8 ] ##  [1.8  0.   0.75 0.   0.4 ]], shape=(2, 5), dtype=float32) vocab <- c(\"a\", \"b\", \"c\", \"d\") data <- rbind(c(1, 3, 4), c(4, 0, 2)) layer <- layer_string_lookup(vocabulary = vocab, invert = TRUE) layer(data) ## tf.Tensor( ## [[b'a' b'c' b'd'] ##  [b'd' b'[UNK]' b'b']], shape=(2, 3), dtype=string) vocab <- c(\"a\", \"b\", \"c\", \"d\") data <- rbind(c(\"a\", \"c\", \"d\"), c(\"d\", \"z\", \"b\")) layer <- layer_string_lookup(vocabulary = vocab) i_layer <- layer_string_lookup(vocabulary = vocab, invert = TRUE) int_data <- layer(data) i_layer(int_data) ## tf.Tensor( ## [[b'a' b'c' b'd'] ##  [b'd' b'[UNK]' b'b']], shape=(2, 3), dtype=string)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_subtract.html","id":null,"dir":"Reference","previous_headings":"","what":"Performs elementwise subtraction. — layer_subtract","title":"Performs elementwise subtraction. — layer_subtract","text":"takes input list tensors size 2 shape, returns single tensor (inputs[0] - inputs[1)) shape.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_subtract.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Performs elementwise subtraction. — layer_subtract","text":"","code":"layer_subtract(inputs, ...)"},{"path":"https://keras3.posit.co/dev/reference/layer_subtract.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Performs elementwise subtraction. — layer_subtract","text":"inputs layers combine ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_subtract.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Performs elementwise subtraction. — layer_subtract","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_subtract.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Performs elementwise subtraction. — layer_subtract","text":"Usage Keras model:","code":"input_shape <- c(2, 3, 4) x1 <- random_uniform(input_shape) x2 <- random_uniform(input_shape) y <- layer_subtract(list(x1, x2)) input1 <- layer_input(shape = 16) x1 <- layer_dense(input1, units = 8, activation = 'relu') input2 <- layer_input(shape = 32) x2 <- layer_dense(input2, units = 8, activation = 'relu') subtracted <- layer_subtract(list(x1, x2)) out <- layer_dense(subtracted, units = 4) model <- keras_model(inputs = list(input1, input2), outputs = out)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_text_vectorization.html","id":null,"dir":"Reference","previous_headings":"","what":"A preprocessing layer which maps text features to integer sequences. — layer_text_vectorization","title":"A preprocessing layer which maps text features to integer sequences. — layer_text_vectorization","text":"layer basic options managing text Keras model. transforms batch strings (one example = one string) either list token indices (one example = 1D tensor integer token indices) dense representation (one example = 1D tensor float values representing data example's tokens). layer meant handle natural language inputs. handle simple string inputs (categorical strings pre-tokenized strings) see layer_string_lookup(). vocabulary layer must either supplied construction learned via adapt(). layer adapted, analyze dataset, determine frequency individual string values, create vocabulary . vocabulary can unlimited size capped, depending configuration options layer; unique values input maximum vocabulary size, frequent terms used create vocabulary. processing example contains following steps: Standardize example (usually lowercasing + punctuation stripping) Split example substrings (usually words) Recombine substrings tokens (usually ngrams) Index tokens (associate unique int value token) Transform example using index, either vector ints dense float vector. notes passing callables customize splitting normalization layer: callable can passed Layer, want serialize object pass functions registered Keras serializables (see register_keras_serializable() details). using custom callable standardize, data received callable exactly passed layer. callable return tensor shape input. using custom callable split, data received callable 1st dimension squeezed - instead list(\"string split\", \"another string split\"), Callable see c(\"string split\", \"another string split\"). callable return tf.Tensor dtype string first dimension containing split tokens - example, see something like list(c(\"string\", \"\", \"split\"), c(\"another\", \"string\", \"\", \"split\")). Note: layer uses TensorFlow internally. used part compiled computation graph model backend TensorFlow. can however used backend running eagerly. can also always used part input preprocessing pipeline backend (outside model ), recommend use layer. Note: layer safe use inside tf.data pipeline (independently backend using).","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_text_vectorization.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"A preprocessing layer which maps text features to integer sequences. — layer_text_vectorization","text":"","code":"layer_text_vectorization(   object,   max_tokens = NULL,   standardize = \"lower_and_strip_punctuation\",   split = \"whitespace\",   ngrams = NULL,   output_mode = \"int\",   output_sequence_length = NULL,   pad_to_max_tokens = FALSE,   vocabulary = NULL,   idf_weights = NULL,   sparse = FALSE,   ragged = FALSE,   encoding = \"utf-8\",   name = NULL,   ... )  get_vocabulary(object, include_special_tokens = TRUE)  set_vocabulary(object, vocabulary, idf_weights = NULL, ...)"},{"path":"https://keras3.posit.co/dev/reference/layer_text_vectorization.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"A preprocessing layer which maps text features to integer sequences. — layer_text_vectorization","text":"object Object compose layer . tensor, array, sequential model. max_tokens Maximum size vocabulary layer. specified adapting vocabulary setting pad_to_max_tokens=TRUE. Note vocabulary contains 1 OOV token, effective number tokens (max_tokens - 1 - (1 output_mode == \"int\" else 0)). standardize Optional specification standardization apply input text. Values can : NULL: standardization. \"lower_and_strip_punctuation\": Text lowercased punctuation removed. \"lower\": Text lowercased. \"strip_punctuation\": punctuation removed. Callable: Inputs passed callable function, standardized returned. split Optional specification splitting input text. Values can : NULL: splitting. \"whitespace\": Split whitespace. \"character\": Split unicode character. Callable: Standardized inputs passed callable function, split returned. ngrams Optional specification ngrams create possibly-split input text. Values can NULL, integer list integers; passing integer create ngrams integer, passing list integers create ngrams specified values list. Passing NULL means ngrams created. output_mode Optional specification output layer. Values can \"int\", \"multi_hot\", \"count\" \"tf_idf\", configuring layer follows: \"int\": Outputs integer indices, one integer index per split string token. output_mode == \"int\", 0 reserved masked locations; reduces vocab size max_tokens - 2 instead max_tokens - 1. \"multi_hot\": Outputs single int array per batch, either vocab_size max_tokens size, containing 1s elements token mapped index exists least batch item. \"count\": Like \"multi_hot\", int array contains count number times token index appeared batch item. \"tf_idf\": Like \"multi_hot\", TF-IDF algorithm applied find value token slot. \"int\" output, shape input output supported. output modes, currently rank 1 inputs (rank 2 outputs splitting) supported. output_sequence_length valid INT mode. set, output time dimension padded truncated exactly output_sequence_length values, resulting tensor shape (batch_size, output_sequence_length) regardless many tokens resulted splitting step. Defaults NULL. ragged TRUE output_sequence_length may still truncate output. pad_to_max_tokens valid  \"multi_hot\", \"count\", \"tf_idf\" modes. TRUE, output feature axis padded max_tokens even number unique tokens vocabulary less max_tokens, resulting tensor shape (batch_size, max_tokens) regardless vocabulary size. Defaults FALSE. vocabulary Optional. Either array strings string path text file. passing array, can pass list, list, 1D NumPy array, 1D tensor containing string vocabulary terms. passing file path, file contain one line per term vocabulary. argument set, need adapt() layer. idf_weights R vector, 1D numpy array, 1D tensor inverse document frequency weights equal length vocabulary. Must set output_mode \"tf_idf\". set otherwise. sparse Boolean. applicable \"multi_hot\", \"count\", \"tf_idf\" output modes. supported TensorFlow backend. TRUE, returns SparseTensor instead dense Tensor. Defaults FALSE. ragged Boolean. applicable \"int\" output mode. supported TensorFlow backend. TRUE, returns RaggedTensor instead dense Tensor, sequence may different length string splitting. Defaults FALSE. encoding Optional. text encoding use interpret input strings. Defaults \"utf-8\". name String, name object ... forward/backward compatability. include_special_tokens TRUE, returned vocabulary include padding OOV tokens, term's index vocabulary equal term's index calling layer. FALSE, returned vocabulary include padding OOV tokens.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_text_vectorization.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"A preprocessing layer which maps text features to integer sequences. — layer_text_vectorization","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_text_vectorization.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"A preprocessing layer which maps text features to integer sequences. — layer_text_vectorization","text":"example instantiates TextVectorization layer lowercases text, splits whitespace, strips punctuation, outputs integer vocab indices.         example instantiates TextVectorization layer passing list vocabulary terms layer's initialize() method.","code":"max_tokens <- 5000  # Maximum vocab size. max_len <- 4  # Sequence length to pad the outputs to. # Create the layer. vectorize_layer <- layer_text_vectorization(     max_tokens = max_tokens,     output_mode = 'int',     output_sequence_length = max_len) # Now that the vocab layer has been created, call `adapt` on the # list of strings to create the vocabulary. vectorize_layer %>% adapt(c(\"foo bar\", \"bar baz\", \"baz bada boom\")) # Now, the layer can map strings to integers -- you can use an # embedding layer to map these integers to learned embeddings. input_data <- rbind(\"foo qux bar\", \"qux baz\") vectorize_layer(input_data) ## tf.Tensor( ## [[4 1 3 0] ##  [1 2 0 0]], shape=(2, 4), dtype=int64) vocab_data <- c(\"earth\", \"wind\", \"and\", \"fire\") max_len <- 4  # Sequence length to pad the outputs to. # Create the layer, passing the vocab directly. You can also pass the # vocabulary arg a path to a file containing one vocabulary word per # line. vectorize_layer <- layer_text_vectorization(     max_tokens = max_tokens,     output_mode = 'int',     output_sequence_length = max_len,     vocabulary = vocab_data) # Because we've passed the vocabulary directly, we don't need to adapt # the layer - the vocabulary is already set. The vocabulary contains the # padding token ('') and OOV token ('[UNK]') # as well as the passed tokens. vectorize_layer %>% get_vocabulary() ## [1] \"\"      \"[UNK]\" \"earth\" \"wind\"  \"and\"   \"fire\" # ['', '[UNK]', 'earth', 'wind', 'and', 'fire']"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_tfsm.html","id":null,"dir":"Reference","previous_headings":"","what":"Reload a Keras model/layer that was saved via export_savedmodel(). — layer_tfsm","title":"Reload a Keras model/layer that was saved via export_savedmodel(). — layer_tfsm","text":"Reload Keras model/layer saved via export_savedmodel().","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_tfsm.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Reload a Keras model/layer that was saved via export_savedmodel(). — layer_tfsm","text":"","code":"layer_tfsm(   object,   filepath,   call_endpoint = \"serve\",   call_training_endpoint = NULL,   trainable = TRUE,   name = NULL,   dtype = NULL )"},{"path":"https://keras3.posit.co/dev/reference/layer_tfsm.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Reload a Keras model/layer that was saved via export_savedmodel(). — layer_tfsm","text":"object Object compose layer . tensor, array, sequential model. filepath string, path SavedModel. call_endpoint Name endpoint use call() method reloaded layer. SavedModel created via export_savedmodel(), default endpoint name 'serve'. cases may named 'serving_default'. call_training_endpoint see description trainable see description name String, name object dtype datatype (e.g., \"float32\").","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_tfsm.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Reload a Keras model/layer that was saved via export_savedmodel(). — layer_tfsm","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_tfsm.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Reload a Keras model/layer that was saved via export_savedmodel(). — layer_tfsm","text":"reloaded object can used like regular Keras layer, supports training/fine-tuning trainable weights. Note reloaded object retains none internal structure custom methods original object – brand new layer created around saved function. Limitations: call endpoints single inputs tensor argument (may optionally named list/list tensors) supported. endpoints multiple separate input tensor arguments, consider subclassing layer_tfsm implementing call() method custom signature. need training-time behavior differ inference-time behavior (.e. need reloaded object support training=TRUE argument __call__()), make sure training-time call function saved standalone endpoint artifact, provide name layer_tfsm via call_training_endpoint argument.","code":"model <- keras_model_sequential(input_shape = c(784)) |> layer_dense(10) model |> export_savedmodel(\"path/to/artifact\") ## Saved artifact at 'path/to/artifact'. The following endpoints are available: ## ## * Endpoint 'serve' ##   args_0 (POSITIONAL_ONLY): TensorSpec(shape=(None, 784), dtype=tf.float32, name='keras_tensor') ## Output Type: ##   TensorSpec(shape=(None, 10), dtype=tf.float32, name=None) ## Captures: ##   133630129211152: TensorSpec(shape=(), dtype=tf.resource, name=None) ##   133630129217872: TensorSpec(shape=(), dtype=tf.resource, name=None) reloaded_layer <- layer_tfsm(filepath = \"path/to/artifact\") input <- random_normal(c(2, 784)) output <- reloaded_layer(input) stopifnot(all.equal(as.array(output), as.array(model(input))))"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_time_distributed.html","id":null,"dir":"Reference","previous_headings":"","what":"This wrapper allows to apply a layer to every temporal slice of an input. — layer_time_distributed","title":"This wrapper allows to apply a layer to every temporal slice of an input. — layer_time_distributed","text":"Every input least 3D, dimension index one first input considered temporal dimension. Consider batch 32 video samples, sample 128x128 RGB image channels_last data format, across 10 timesteps. batch input shape (32, 10, 128, 128, 3). can use TimeDistributed apply Conv2D layer 10 timesteps, independently:     layer_time_distributed applies instance layer_conv2d timestamps, set weights used timestamp.","code":"inputs <- keras_input(shape = c(10, 128, 128, 3), batch_size = 32) conv_2d_layer <- layer_conv_2d(filters = 64, kernel_size = c(3, 3)) outputs <- layer_time_distributed(inputs, layer = conv_2d_layer) shape(outputs) ## shape(32, 10, 126, 126, 64)"},{"path":"https://keras3.posit.co/dev/reference/layer_time_distributed.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"This wrapper allows to apply a layer to every temporal slice of an input. — layer_time_distributed","text":"","code":"layer_time_distributed(object, layer, ...)"},{"path":"https://keras3.posit.co/dev/reference/layer_time_distributed.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"This wrapper allows to apply a layer to every temporal slice of an input. — layer_time_distributed","text":"object Object compose layer . tensor, array, sequential model. layer Layer instance. ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_time_distributed.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"This wrapper allows to apply a layer to every temporal slice of an input. — layer_time_distributed","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_time_distributed.html","id":"call-arguments","dir":"Reference","previous_headings":"","what":"Call Arguments","title":"This wrapper allows to apply a layer to every temporal slice of an input. — layer_time_distributed","text":"inputs: Input tensor shape (batch, time, ...) nested tensors, shape (batch, time, ...). training: Boolean indicating whether layer behave training mode inference mode. argument passed wrapped layer (layer supports argument). mask: Binary tensor shape (samples, timesteps) indicating whether given timestep masked. argument passed wrapped layer (layer supports argument).","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_torch_module_wrapper.html","id":null,"dir":"Reference","previous_headings":"","what":"Torch module wrapper layer. — layer_torch_module_wrapper","title":"Torch module wrapper layer. — layer_torch_module_wrapper","text":"layer_torch_module_wrapper wrapper class can turn torch.nn.Module Keras layer, particular making parameters trackable Keras. layer_torch_module_wrapper() compatible PyTorch backend used TensorFlow JAX backends.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_torch_module_wrapper.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Torch module wrapper layer. — layer_torch_module_wrapper","text":"","code":"layer_torch_module_wrapper(   object,   module,   output_shape = NULL,   name = NULL,   ... )"},{"path":"https://keras3.posit.co/dev/reference/layer_torch_module_wrapper.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Torch module wrapper layer. — layer_torch_module_wrapper","text":"object Object compose layer . tensor, array, sequential model. module torch.nn.Module instance. LazyModule instance, parameters must initialized passing instance layer_torch_module_wrapper (e.g. calling ). output_shape Shape output layer. Helps Keras infer shapes. name name layer (string). ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_torch_module_wrapper.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Torch module wrapper layer. — layer_torch_module_wrapper","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_torch_module_wrapper.html","id":"example","dir":"Reference","previous_headings":"","what":"Example","title":"Torch module wrapper layer. — layer_torch_module_wrapper","text":"example layer_torch_module_wrapper() can used vanilla PyTorch modules.","code":"# reticulate::py_install( #   packages = c(\"torch\", \"torchvision\", \"torchaudio\"), #   envname = \"r-keras\", #   pip_options = c(\"--index-url https://download.pytorch.org/whl/cpu\") # ) library(keras3) use_backend(\"torch\") torch <- reticulate::import(\"torch\") nn <- reticulate::import(\"torch.nn\") nnf <- reticulate::import(\"torch.nn.functional\")  Classifier(keras$Model) \\%py_class\\% {   initialize <- function(...) {     super$initialize(...)      self$conv1 <- layer_torch_module_wrapper(module = nn$Conv2d(       in_channels = 1L,       out_channels = 32L,       kernel_size = tuple(3L, 3L)     ))     self$conv2 <- layer_torch_module_wrapper(module = nn$Conv2d(       in_channels = 32L,       out_channels = 64L,       kernel_size = tuple(3L, 3L)     ))     self$pool <- nn$MaxPool2d(kernel_size = tuple(2L, 2L))     self$flatten <- nn$Flatten()     self$dropout <- nn$Dropout(p = 0.5)     self$fc <-       layer_torch_module_wrapper(module = nn$Linear(1600L, 10L))   }    call <- function(inputs) {     x <- nnf$relu(self$conv1(inputs))     x <- self$pool(x)     x <- nnf$relu(self$conv2(x))     x <- self$pool(x)     x <- self$flatten(x)     x <- self$dropout(x)     x <- self$fc(x)     nnf$softmax(x, dim = 1L)   } } model <- Classifier() model$build(shape(1, 28, 28)) cat(\"Output shape:\", format(shape(model(torch$ones(1L, 1L, 28L, 28L)))))  model |> compile(loss = \"sparse_categorical_crossentropy\",                  optimizer = \"adam\",                  metrics = \"accuracy\") model |> fit(train_loader, epochs = 5)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_unit_normalization.html","id":null,"dir":"Reference","previous_headings":"","what":"Unit normalization layer. — layer_unit_normalization","title":"Unit normalization layer. — layer_unit_normalization","text":"Normalize batch inputs input batch L2 norm equal 1 (across axes specified axis).","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_unit_normalization.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Unit normalization layer. — layer_unit_normalization","text":"","code":"layer_unit_normalization(object, axis = -1L, ...)"},{"path":"https://keras3.posit.co/dev/reference/layer_unit_normalization.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Unit normalization layer. — layer_unit_normalization","text":"object Object compose layer . tensor, array, sequential model. axis Integer list. axis axes normalize across. Typically, features axis axes. left-axes typically batch axis axes. -1 last dimension input. Defaults -1. ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_unit_normalization.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Unit normalization layer. — layer_unit_normalization","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_unit_normalization.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Unit normalization layer. — layer_unit_normalization","text":"","code":"data <- 1:6 |> op_array(\"float32\") |> op_reshape(c(2, 3)) normalized_data <- layer_unit_normalization(data) op_sum(normalized_data[1, ]^2) ## tf.Tensor(0.9999999, shape=(), dtype=float32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_upsampling_1d.html","id":null,"dir":"Reference","previous_headings":"","what":"Upsampling layer for 1D inputs. — layer_upsampling_1d","title":"Upsampling layer for 1D inputs. — layer_upsampling_1d","text":"Repeats temporal step size times along time axis.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_upsampling_1d.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Upsampling layer for 1D inputs. — layer_upsampling_1d","text":"","code":"layer_upsampling_1d(object, size = 2L, ...)"},{"path":"https://keras3.posit.co/dev/reference/layer_upsampling_1d.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Upsampling layer for 1D inputs. — layer_upsampling_1d","text":"object Object compose layer . tensor, array, sequential model. size Integer. Upsampling factor. ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_upsampling_1d.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Upsampling layer for 1D inputs. — layer_upsampling_1d","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_upsampling_1d.html","id":"example","dir":"Reference","previous_headings":"","what":"Example","title":"Upsampling layer for 1D inputs. — layer_upsampling_1d","text":"","code":"input_shape <- c(2, 2, 3) x <- seq_len(prod(input_shape)) %>% op_reshape(input_shape) x ## tf.Tensor( ## [[[ 1  2  3] ##   [ 4  5  6]] ## ##  [[ 7  8  9] ##   [10 11 12]]], shape=(2, 2, 3), dtype=int32) y <- layer_upsampling_1d(x, size = 2) y ## tf.Tensor( ## [[[ 1  2  3] ##   [ 1  2  3] ##   [ 4  5  6] ##   [ 4  5  6]] ## ##  [[ 7  8  9] ##   [ 7  8  9] ##   [10 11 12] ##   [10 11 12]]], shape=(2, 4, 3), dtype=int32)"},{"path":"https://keras3.posit.co/dev/reference/layer_upsampling_1d.html","id":"input-shape","dir":"Reference","previous_headings":"","what":"Input Shape","title":"Upsampling layer for 1D inputs. — layer_upsampling_1d","text":"3D tensor shape: (batch_size, steps, features).","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_upsampling_1d.html","id":"output-shape","dir":"Reference","previous_headings":"","what":"Output Shape","title":"Upsampling layer for 1D inputs. — layer_upsampling_1d","text":"3D tensor shape: (batch_size, upsampled_steps, features).","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_upsampling_2d.html","id":null,"dir":"Reference","previous_headings":"","what":"Upsampling layer for 2D inputs. — layer_upsampling_2d","title":"Upsampling layer for 2D inputs. — layer_upsampling_2d","text":"implementation uses interpolative resizing, given resize method (specified interpolation argument). Use interpolation=nearest repeat rows columns data.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_upsampling_2d.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Upsampling layer for 2D inputs. — layer_upsampling_2d","text":"","code":"layer_upsampling_2d(   object,   size = list(2L, 2L),   data_format = NULL,   interpolation = \"nearest\",   ... )"},{"path":"https://keras3.posit.co/dev/reference/layer_upsampling_2d.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Upsampling layer for 2D inputs. — layer_upsampling_2d","text":"object Object compose layer . tensor, array, sequential model. size Int, list 2 integers. upsampling factors rows columns. data_format string, one \"channels_last\" (default) \"channels_first\". ordering dimensions inputs. \"channels_last\" corresponds inputs shape (batch_size, height, width, channels) \"channels_first\" corresponds inputs shape (batch_size, channels, height, width). unspecified, uses image_data_format value found Keras config file ~/.keras/keras.json (exists) else \"channels_last\". Defaults \"channels_last\". interpolation string, one \"bicubic\", \"bilinear\", \"lanczos3\", \"lanczos5\", \"nearest\". ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_upsampling_2d.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Upsampling layer for 2D inputs. — layer_upsampling_2d","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_upsampling_2d.html","id":"example","dir":"Reference","previous_headings":"","what":"Example","title":"Upsampling layer for 2D inputs. — layer_upsampling_2d","text":"","code":"input_shape <- c(2, 2, 1, 3) x <- op_reshape(seq_len(prod(input_shape)), input_shape) print(x) ## tf.Tensor( ## [[[[ 1  2  3]] ## ##   [[ 4  5  6]]] ## ## ##  [[[ 7  8  9]] ## ##   [[10 11 12]]]], shape=(2, 2, 1, 3), dtype=int32) y <- layer_upsampling_2d(x, size = c(1, 2)) print(y) ## tf.Tensor( ## [[[[ 1  2  3] ##    [ 1  2  3]] ## ##   [[ 4  5  6] ##    [ 4  5  6]]] ## ## ##  [[[ 7  8  9] ##    [ 7  8  9]] ## ##   [[10 11 12] ##    [10 11 12]]]], shape=(2, 2, 2, 3), dtype=int32)"},{"path":"https://keras3.posit.co/dev/reference/layer_upsampling_2d.html","id":"input-shape","dir":"Reference","previous_headings":"","what":"Input Shape","title":"Upsampling layer for 2D inputs. — layer_upsampling_2d","text":"4D tensor shape: data_format \"channels_last\": (batch_size, rows, cols, channels) data_format \"channels_first\": (batch_size, channels, rows, cols)","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_upsampling_2d.html","id":"output-shape","dir":"Reference","previous_headings":"","what":"Output Shape","title":"Upsampling layer for 2D inputs. — layer_upsampling_2d","text":"4D tensor shape: data_format \"channels_last\": (batch_size, upsampled_rows, upsampled_cols, channels) data_format \"channels_first\": (batch_size, channels, upsampled_rows, upsampled_cols)","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_upsampling_3d.html","id":null,"dir":"Reference","previous_headings":"","what":"Upsampling layer for 3D inputs. — layer_upsampling_3d","title":"Upsampling layer for 3D inputs. — layer_upsampling_3d","text":"Repeats 1st, 2nd 3rd dimensions data size[0], size[1] size[2] respectively.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_upsampling_3d.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Upsampling layer for 3D inputs. — layer_upsampling_3d","text":"","code":"layer_upsampling_3d(object, size = list(2L, 2L, 2L), data_format = NULL, ...)"},{"path":"https://keras3.posit.co/dev/reference/layer_upsampling_3d.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Upsampling layer for 3D inputs. — layer_upsampling_3d","text":"object Object compose layer . tensor, array, sequential model. size Int, list 3 integers. upsampling factors dim1, dim2 dim3. data_format string, one \"channels_last\" (default) \"channels_first\". ordering dimensions inputs. \"channels_last\" corresponds inputs shape (batch_size, spatial_dim1, spatial_dim2, spatial_dim3, channels) \"channels_first\" corresponds inputs shape (batch_size, channels, spatial_dim1, spatial_dim2, spatial_dim3). unspecified, uses image_data_format value found Keras config file ~/.keras/keras.json (exists) else \"channels_last\". Defaults \"channels_last\". ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_upsampling_3d.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Upsampling layer for 3D inputs. — layer_upsampling_3d","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_upsampling_3d.html","id":"example","dir":"Reference","previous_headings":"","what":"Example","title":"Upsampling layer for 3D inputs. — layer_upsampling_3d","text":"","code":"input_shape <- c(2, 1, 2, 1, 3) x <- array(1, dim = input_shape) y <- layer_upsampling_3d(x, size = c(2, 2, 2)) shape(y) ## shape(2, 2, 4, 2, 3)"},{"path":"https://keras3.posit.co/dev/reference/layer_upsampling_3d.html","id":"input-shape","dir":"Reference","previous_headings":"","what":"Input Shape","title":"Upsampling layer for 3D inputs. — layer_upsampling_3d","text":"5D tensor shape: data_format \"channels_last\": (batch_size, dim1, dim2, dim3, channels) data_format \"channels_first\": (batch_size, channels, dim1, dim2, dim3)","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_upsampling_3d.html","id":"output-shape","dir":"Reference","previous_headings":"","what":"Output Shape","title":"Upsampling layer for 3D inputs. — layer_upsampling_3d","text":"5D tensor shape: data_format \"channels_last\": (batch_size, upsampled_dim1, upsampled_dim2, upsampled_dim3, channels) data_format \"channels_first\": (batch_size, channels, upsampled_dim1, upsampled_dim2, upsampled_dim3)","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_zero_padding_1d.html","id":null,"dir":"Reference","previous_headings":"","what":"Zero-padding layer for 1D input (e.g. temporal sequence). — layer_zero_padding_1d","title":"Zero-padding layer for 1D input (e.g. temporal sequence). — layer_zero_padding_1d","text":"Zero-padding layer 1D input (e.g. temporal sequence).","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_zero_padding_1d.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Zero-padding layer for 1D input (e.g. temporal sequence). — layer_zero_padding_1d","text":"","code":"layer_zero_padding_1d(object, padding = 1L, data_format = NULL, ...)"},{"path":"https://keras3.posit.co/dev/reference/layer_zero_padding_1d.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Zero-padding layer for 1D input (e.g. temporal sequence). — layer_zero_padding_1d","text":"object Object compose layer . tensor, array, sequential model. padding Int, list int (length 2). int: many zeros add beginning end padding dimension (axis 2). list 2 ints: many zeros add beginning end padding dimension ((left_pad, right_pad)). data_format string, one \"channels_last\" (default) \"channels_first\". ordering dimensions inputs. \"channels_last\" corresponds inputs shape (batch_size, axis_to_pad, channels) \"channels_first\" corresponds inputs shape (batch_size, channels, axis_to_pad). unspecified, uses image_data_format value found Keras config file ~/.keras/keras.json (exists). Defaults \"channels_last\". ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_zero_padding_1d.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Zero-padding layer for 1D input (e.g. temporal sequence). — layer_zero_padding_1d","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_zero_padding_1d.html","id":"example","dir":"Reference","previous_headings":"","what":"Example","title":"Zero-padding layer for 1D input (e.g. temporal sequence). — layer_zero_padding_1d","text":"","code":"input_shape <- c(2, 2, 3) x <- op_reshape(seq_len(prod(input_shape)), input_shape) x ## tf.Tensor( ## [[[ 1  2  3] ##   [ 4  5  6]] ## ##  [[ 7  8  9] ##   [10 11 12]]], shape=(2, 2, 3), dtype=int32) y <- layer_zero_padding_1d(x, padding = 2) y ## tf.Tensor( ## [[[ 0  0  0] ##   [ 0  0  0] ##   [ 1  2  3] ##   [ 4  5  6] ##   [ 0  0  0] ##   [ 0  0  0]] ## ##  [[ 0  0  0] ##   [ 0  0  0] ##   [ 7  8  9] ##   [10 11 12] ##   [ 0  0  0] ##   [ 0  0  0]]], shape=(2, 6, 3), dtype=int32)"},{"path":"https://keras3.posit.co/dev/reference/layer_zero_padding_1d.html","id":"input-shape","dir":"Reference","previous_headings":"","what":"Input Shape","title":"Zero-padding layer for 1D input (e.g. temporal sequence). — layer_zero_padding_1d","text":"3D tensor shape: data_format \"channels_last\": (batch_size, axis_to_pad, features) data_format \"channels_first\": (batch_size, features, axis_to_pad)","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_zero_padding_1d.html","id":"output-shape","dir":"Reference","previous_headings":"","what":"Output Shape","title":"Zero-padding layer for 1D input (e.g. temporal sequence). — layer_zero_padding_1d","text":"3D tensor shape: data_format \"channels_last\": (batch_size, padded_axis, features) data_format \"channels_first\": (batch_size, features, padded_axis)","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_zero_padding_2d.html","id":null,"dir":"Reference","previous_headings":"","what":"Zero-padding layer for 2D input (e.g. picture). — layer_zero_padding_2d","title":"Zero-padding layer for 2D input (e.g. picture). — layer_zero_padding_2d","text":"layer can add rows columns zeros top, bottom, left right side image tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_zero_padding_2d.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Zero-padding layer for 2D input (e.g. picture). — layer_zero_padding_2d","text":"","code":"layer_zero_padding_2d(object, padding = list(1L, 1L), data_format = NULL, ...)"},{"path":"https://keras3.posit.co/dev/reference/layer_zero_padding_2d.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Zero-padding layer for 2D input (e.g. picture). — layer_zero_padding_2d","text":"object Object compose layer . tensor, array, sequential model. padding Int, list 2 ints, list 2 lists 2 ints. int: symmetric padding applied height width. list 2 ints: interpreted two different symmetric padding values height width: (symmetric_height_pad, symmetric_width_pad). list 2 lists 2 ints: interpreted ((top_pad, bottom_pad), (left_pad, right_pad)). data_format string, one \"channels_last\" (default) \"channels_first\". ordering dimensions inputs. \"channels_last\" corresponds inputs shape (batch_size, height, width, channels) \"channels_first\" corresponds inputs shape (batch_size, channels, height, width). unspecified, uses image_data_format value found Keras config file ~/.keras/keras.json (exists). Defaults \"channels_last\". ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_zero_padding_2d.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Zero-padding layer for 2D input (e.g. picture). — layer_zero_padding_2d","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_zero_padding_2d.html","id":"example","dir":"Reference","previous_headings":"","what":"Example","title":"Zero-padding layer for 2D input (e.g. picture). — layer_zero_padding_2d","text":"","code":"input_shape <- c(1, 1, 2, 2) x <- op_reshape(seq_len(prod(input_shape)), input_shape) x ## tf.Tensor( ## [[[[1 2] ##    [3 4]]]], shape=(1, 1, 2, 2), dtype=int32) y <- layer_zero_padding_2d(x, padding = 1) y ## tf.Tensor( ## [[[[0 0] ##    [0 0] ##    [0 0] ##    [0 0]] ## ##   [[0 0] ##    [1 2] ##    [3 4] ##    [0 0]] ## ##   [[0 0] ##    [0 0] ##    [0 0] ##    [0 0]]]], shape=(1, 3, 4, 2), dtype=int32)"},{"path":"https://keras3.posit.co/dev/reference/layer_zero_padding_2d.html","id":"input-shape","dir":"Reference","previous_headings":"","what":"Input Shape","title":"Zero-padding layer for 2D input (e.g. picture). — layer_zero_padding_2d","text":"4D tensor shape: data_format \"channels_last\": (batch_size, height, width, channels) data_format \"channels_first\": (batch_size, channels, height, width)","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_zero_padding_2d.html","id":"output-shape","dir":"Reference","previous_headings":"","what":"Output Shape","title":"Zero-padding layer for 2D input (e.g. picture). — layer_zero_padding_2d","text":"4D tensor shape: data_format \"channels_last\": (batch_size, padded_height, padded_width, channels) data_format \"channels_first\": (batch_size, channels, padded_height, padded_width)","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/layer_zero_padding_3d.html","id":null,"dir":"Reference","previous_headings":"","what":"Zero-padding layer for 3D data (spatial or spatio-temporal). — layer_zero_padding_3d","title":"Zero-padding layer for 3D data (spatial or spatio-temporal). — layer_zero_padding_3d","text":"Zero-padding layer 3D data (spatial spatio-temporal).","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_zero_padding_3d.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Zero-padding layer for 3D data (spatial or spatio-temporal). — layer_zero_padding_3d","text":"","code":"layer_zero_padding_3d(   object,   padding = list(list(1L, 1L), list(1L, 1L), list(1L, 1L)),   data_format = NULL,   ... )"},{"path":"https://keras3.posit.co/dev/reference/layer_zero_padding_3d.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Zero-padding layer for 3D data (spatial or spatio-temporal). — layer_zero_padding_3d","text":"object Object compose layer . tensor, array, sequential model. padding Int, list 3 ints, list 3 lists 2 ints. int: symmetric padding applied depth, height, width. list 3 ints: interpreted three different symmetric padding values depth, height, width: (symmetric_dim1_pad, symmetric_dim2_pad, symmetric_dim3_pad). list 3 lists 2 ints: interpreted ((left_dim1_pad, right_dim1_pad), (left_dim2_pad, right_dim2_pad), (left_dim3_pad, right_dim3_pad)). data_format string, one \"channels_last\" (default) \"channels_first\". ordering dimensions inputs. \"channels_last\" corresponds inputs shape (batch_size, spatial_dim1, spatial_dim2, spatial_dim3, channels) \"channels_first\" corresponds inputs shape (batch_size, channels, spatial_dim1, spatial_dim2, spatial_dim3). unspecified, uses image_data_format value found Keras config file ~/.keras/keras.json (exists). Defaults \"channels_last\". ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_zero_padding_3d.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Zero-padding layer for 3D data (spatial or spatio-temporal). — layer_zero_padding_3d","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_zero_padding_3d.html","id":"example","dir":"Reference","previous_headings":"","what":"Example","title":"Zero-padding layer for 3D data (spatial or spatio-temporal). — layer_zero_padding_3d","text":"","code":"input_shape <- c(1, 1, 2, 2, 3) x <- op_reshape(seq_len(prod(input_shape)), input_shape) x ## tf.Tensor( ## [[[[[ 1  2  3] ##     [ 4  5  6]] ## ##    [[ 7  8  9] ##     [10 11 12]]]]], shape=(1, 1, 2, 2, 3), dtype=int32) y <- layer_zero_padding_3d(x, padding = 2) shape(y) ## shape(1, 5, 6, 6, 3)"},{"path":"https://keras3.posit.co/dev/reference/layer_zero_padding_3d.html","id":"input-shape","dir":"Reference","previous_headings":"","what":"Input Shape","title":"Zero-padding layer for 3D data (spatial or spatio-temporal). — layer_zero_padding_3d","text":"5D tensor shape: data_format \"channels_last\": (batch_size, first_axis_to_pad, second_axis_to_pad, third_axis_to_pad, depth) data_format \"channels_first\": (batch_size, depth, first_axis_to_pad, second_axis_to_pad, third_axis_to_pad)","code":""},{"path":"https://keras3.posit.co/dev/reference/layer_zero_padding_3d.html","id":"output-shape","dir":"Reference","previous_headings":"","what":"Output Shape","title":"Zero-padding layer for 3D data (spatial or spatio-temporal). — layer_zero_padding_3d","text":"5D tensor shape: data_format \"channels_last\": (batch_size, first_padded_axis, second_padded_axis, third_axis_to_pad, depth) data_format \"channels_first\": (batch_size, depth, first_padded_axis, second_padded_axis, third_axis_to_pad)","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/learning_rate_schedule_cosine_decay.html","id":null,"dir":"Reference","previous_headings":"","what":"A LearningRateSchedule that uses a cosine decay with optional warmup. — learning_rate_schedule_cosine_decay","title":"A LearningRateSchedule that uses a cosine decay with optional warmup. — learning_rate_schedule_cosine_decay","text":"See Loshchilov & Hutter, ICLR2016, SGDR: Stochastic Gradient Descent Warm Restarts. idea linear warmup learning rate, see Goyal et al.. begin training model, often want initial increase learning rate followed decay. warmup_target int, schedule applies linear increase per optimizer step learning rate initial_learning_rate warmup_target duration warmup_steps. Afterwards, applies cosine decay function taking learning rate warmup_target alpha duration decay_steps. warmup_target NULL skip warmup decay take learning rate initial_learning_rate alpha. requires step value  compute learning rate. can just pass backend variable increment training step. schedule 1-arg callable produces warmup followed decayed learning rate passed current optimizer step. can useful changing learning rate value across different invocations optimizer functions. warmup computed :   decay computed :   Example usage without warmup:   Example usage warmup:   can pass schedule directly optimizer learning rate. learning rate schedule also serializable deserializable using keras$optimizers$schedules$serialize keras$optimizers$schedules$deserialize.","code":"warmup_learning_rate <- function(step) {   completed_fraction <- step / warmup_steps   total_delta <- target_warmup - initial_learning_rate   completed_fraction * total_delta } if (is.null(warmup_target)) {   initial_decay_lr <- initial_learning_rate } else {   initial_decay_lr <- warmup_target }  decayed_learning_rate <- function(step) {   step <- min(step, decay_steps)   cosine_decay <- 0.5 * (1 + cos(pi * step / decay_steps))   decayed <- (1 - alpha) * cosine_decay + alpha   initial_decay_lr * decayed } decay_steps <- 1000 initial_learning_rate <- 0.1 lr_decayed_fn <- learning_rate_schedule_cosine_decay(     initial_learning_rate, decay_steps) decay_steps <- 1000 initial_learning_rate <- 0 warmup_steps <- 1000 target_learning_rate <- 0.1 lr_warmup_decayed_fn <- learning_rate_schedule_cosine_decay(     initial_learning_rate, decay_steps, warmup_target = target_learning_rate,     warmup_steps = warmup_steps )"},{"path":"https://keras3.posit.co/dev/reference/learning_rate_schedule_cosine_decay.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"A LearningRateSchedule that uses a cosine decay with optional warmup. — learning_rate_schedule_cosine_decay","text":"","code":"learning_rate_schedule_cosine_decay(   initial_learning_rate,   decay_steps,   alpha = 0,   name = \"CosineDecay\",   warmup_target = NULL,   warmup_steps = 0L )"},{"path":"https://keras3.posit.co/dev/reference/learning_rate_schedule_cosine_decay.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"A LearningRateSchedule that uses a cosine decay with optional warmup. — learning_rate_schedule_cosine_decay","text":"initial_learning_rate float. initial learning rate. decay_steps int. Number steps decay . alpha float. Minimum learning rate value decay fraction initial_learning_rate. name String. Optional name operation.  Defaults \"CosineDecay\". warmup_target float. target learning rate warmup phase. cast initial_learning_rate datatype. Setting NULL skip warmup begins decay phase initial_learning_rate. Otherwise scheduler warmup initial_learning_rate warmup_target. warmup_steps int. Number steps warmup .","code":""},{"path":"https://keras3.posit.co/dev/reference/learning_rate_schedule_cosine_decay.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"A LearningRateSchedule that uses a cosine decay with optional warmup. — learning_rate_schedule_cosine_decay","text":"1-arg callable learning rate schedule takes current optimizer step outputs decayed learning rate, scalar tensor type initial_learning_rate.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/learning_rate_schedule_cosine_decay_restarts.html","id":null,"dir":"Reference","previous_headings":"","what":"A LearningRateSchedule that uses a cosine decay schedule with restarts. — learning_rate_schedule_cosine_decay_restarts","title":"A LearningRateSchedule that uses a cosine decay schedule with restarts. — learning_rate_schedule_cosine_decay_restarts","text":"See Loshchilov & Hutter, ICLR2016, SGDR: Stochastic Gradient Descent Warm Restarts. training model, often useful lower learning rate training progresses. schedule applies cosine decay function restarts optimizer step, given provided initial learning rate. requires step value compute decayed learning rate. can just pass backend variable increment training step. schedule 1-arg callable produces decayed learning rate passed current optimizer step. can useful changing learning rate value across different invocations optimizer functions. learning rate multiplier first decays 1 alpha first_decay_steps steps. , warm restart performed. new warm restart runs t_mul times steps m_mul times initial learning rate new learning rate.","code":""},{"path":"https://keras3.posit.co/dev/reference/learning_rate_schedule_cosine_decay_restarts.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"A LearningRateSchedule that uses a cosine decay schedule with restarts. — learning_rate_schedule_cosine_decay_restarts","text":"","code":"learning_rate_schedule_cosine_decay_restarts(   initial_learning_rate,   first_decay_steps,   t_mul = 2,   m_mul = 1,   alpha = 0,   name = \"SGDRDecay\" )"},{"path":"https://keras3.posit.co/dev/reference/learning_rate_schedule_cosine_decay_restarts.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"A LearningRateSchedule that uses a cosine decay schedule with restarts. — learning_rate_schedule_cosine_decay_restarts","text":"initial_learning_rate float. initial learning rate. first_decay_steps integer. Number steps decay . t_mul float. Used derive number iterations -th period. m_mul float. Used derive initial learning rate -th period. alpha float. Minimum learning rate value fraction initial_learning_rate. name String. Optional name operation. Defaults \"SGDRDecay\".","code":""},{"path":"https://keras3.posit.co/dev/reference/learning_rate_schedule_cosine_decay_restarts.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"A LearningRateSchedule that uses a cosine decay schedule with restarts. — learning_rate_schedule_cosine_decay_restarts","text":"1-arg callable learning rate schedule takes current optimizer step outputs decayed learning rate, scalar tensor type initial_learning_rate.","code":""},{"path":"https://keras3.posit.co/dev/reference/learning_rate_schedule_cosine_decay_restarts.html","id":"example","dir":"Reference","previous_headings":"","what":"Example","title":"A LearningRateSchedule that uses a cosine decay schedule with restarts. — learning_rate_schedule_cosine_decay_restarts","text":"can pass schedule directly optimizer learning rate. learning rate schedule also serializable deserializable using keras$optimizers$schedules$serialize keras$optimizers$schedules$deserialize.","code":"first_decay_steps <- 1000 lr_decayed_fn <- learning_rate_schedule_cosine_decay_restarts(         0.001,         first_decay_steps)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/learning_rate_schedule_exponential_decay.html","id":null,"dir":"Reference","previous_headings":"","what":"A LearningRateSchedule that uses an exponential decay schedule. — learning_rate_schedule_exponential_decay","title":"A LearningRateSchedule that uses an exponential decay schedule. — learning_rate_schedule_exponential_decay","text":"training model, often useful lower learning rate training progresses. schedule applies exponential decay function optimizer step, given provided initial learning rate. schedule 1-arg callable produces decayed learning rate passed current optimizer step. can useful changing learning rate value across different invocations optimizer functions. computed :   argument staircase TRUE, step / decay_steps integer division decayed learning rate follows staircase function. can pass schedule directly optimizer learning rate.","code":"decayed_learning_rate <- function(step) {   initial_learning_rate * decay_rate ^ (step / decay_steps) }"},{"path":"https://keras3.posit.co/dev/reference/learning_rate_schedule_exponential_decay.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"A LearningRateSchedule that uses an exponential decay schedule. — learning_rate_schedule_exponential_decay","text":"","code":"learning_rate_schedule_exponential_decay(   initial_learning_rate,   decay_steps,   decay_rate,   staircase = FALSE,   name = \"ExponentialDecay\" )"},{"path":"https://keras3.posit.co/dev/reference/learning_rate_schedule_exponential_decay.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"A LearningRateSchedule that uses an exponential decay schedule. — learning_rate_schedule_exponential_decay","text":"initial_learning_rate float. initial learning rate. decay_steps integer. Must positive. See decay computation . decay_rate float. decay rate. staircase Boolean.  TRUE decay learning rate discrete intervals. name String.  Optional name operation.  Defaults \"ExponentialDecay\".","code":""},{"path":"https://keras3.posit.co/dev/reference/learning_rate_schedule_exponential_decay.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"A LearningRateSchedule that uses an exponential decay schedule. — learning_rate_schedule_exponential_decay","text":"1-arg callable learning rate schedule takes current optimizer step outputs decayed learning rate, scalar tensor type initial_learning_rate.","code":""},{"path":"https://keras3.posit.co/dev/reference/learning_rate_schedule_exponential_decay.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"A LearningRateSchedule that uses an exponential decay schedule. — learning_rate_schedule_exponential_decay","text":"fitting Keras model, decay every 100000 steps base 0.96:   learning rate schedule also serializable deserializable using keras$optimizers$schedules$serialize keras$optimizers$schedules$deserialize.","code":"initial_learning_rate <- 0.1 lr_schedule <- learning_rate_schedule_exponential_decay(     initial_learning_rate,     decay_steps=100000,     decay_rate=0.96,     staircase=TRUE)  model %>% compile(   optimizer = optimizer_sgd(learning_rate = lr_schedule),   loss = 'sparse_categorical_crossentropy',   metrics = c('accuracy'))  model %>% fit(data, labels, epochs=5)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/learning_rate_schedule_inverse_time_decay.html","id":null,"dir":"Reference","previous_headings":"","what":"A LearningRateSchedule that uses an inverse time decay schedule. — learning_rate_schedule_inverse_time_decay","title":"A LearningRateSchedule that uses an inverse time decay schedule. — learning_rate_schedule_inverse_time_decay","text":"training model, often useful lower learning rate training progresses. schedule applies inverse decay function optimizer step, given provided initial learning rate. requires step value compute decayed learning rate. can just pass backend variable increment training step. schedule 1-arg callable produces decayed learning rate passed current optimizer step. can useful changing learning rate value across different invocations optimizer functions. computed :   , staircase TRUE, :   can pass schedule directly optimizer_* learning rate.","code":"decayed_learning_rate <- function(step) {   initial_learning_rate / (1 + decay_rate * step / decay_step) } decayed_learning_rate <- function(step) {   initial_learning_rate /            (1 + decay_rate * floor(step / decay_step)) }"},{"path":"https://keras3.posit.co/dev/reference/learning_rate_schedule_inverse_time_decay.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"A LearningRateSchedule that uses an inverse time decay schedule. — learning_rate_schedule_inverse_time_decay","text":"","code":"learning_rate_schedule_inverse_time_decay(   initial_learning_rate,   decay_steps,   decay_rate,   staircase = FALSE,   name = \"InverseTimeDecay\" )"},{"path":"https://keras3.posit.co/dev/reference/learning_rate_schedule_inverse_time_decay.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"A LearningRateSchedule that uses an inverse time decay schedule. — learning_rate_schedule_inverse_time_decay","text":"initial_learning_rate float. initial learning rate. decay_steps often apply decay. decay_rate number.  decay rate. staircase Whether apply decay discrete staircase, o pposed continuous, fashion. name String.  Optional name operation.  Defaults \"InverseTimeDecay\".","code":""},{"path":"https://keras3.posit.co/dev/reference/learning_rate_schedule_inverse_time_decay.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"A LearningRateSchedule that uses an inverse time decay schedule. — learning_rate_schedule_inverse_time_decay","text":"1-arg callable learning rate schedule takes current optimizer step outputs decayed learning rate, scalar tensor type initial_learning_rate.","code":""},{"path":"https://keras3.posit.co/dev/reference/learning_rate_schedule_inverse_time_decay.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"A LearningRateSchedule that uses an inverse time decay schedule. — learning_rate_schedule_inverse_time_decay","text":"Fit Keras model decaying 1/t rate 0.5:","code":"... initial_learning_rate <- 0.1 decay_steps <- 1.0 decay_rate <- 0.5 learning_rate_fn <- learning_rate_schedule_inverse_time_decay(     initial_learning_rate, decay_steps, decay_rate)  model %>% compile(   optimizer = optimizer_sgd(learning_rate=learning_rate_fn),   loss = 'sparse_categorical_crossentropy',   metrics = 'accuracy') )  model %>% fit(data, labels, epochs=5)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/learning_rate_schedule_piecewise_constant_decay.html","id":null,"dir":"Reference","previous_headings":"","what":"A LearningRateSchedule that uses a piecewise constant decay schedule. — learning_rate_schedule_piecewise_constant_decay","title":"A LearningRateSchedule that uses a piecewise constant decay schedule. — learning_rate_schedule_piecewise_constant_decay","text":"function returns 1-arg callable compute piecewise constant passed current optimizer step. can useful changing learning rate value across different invocations optimizer functions.","code":""},{"path":"https://keras3.posit.co/dev/reference/learning_rate_schedule_piecewise_constant_decay.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"A LearningRateSchedule that uses a piecewise constant decay schedule. — learning_rate_schedule_piecewise_constant_decay","text":"","code":"learning_rate_schedule_piecewise_constant_decay(   boundaries,   values,   name = \"PiecewiseConstant\" )"},{"path":"https://keras3.posit.co/dev/reference/learning_rate_schedule_piecewise_constant_decay.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"A LearningRateSchedule that uses a piecewise constant decay schedule. — learning_rate_schedule_piecewise_constant_decay","text":"boundaries list Python numbers strictly increasing entries, elements type optimizer step. values list Python numbers specifies values intervals defined boundaries. one element boundaries, elements type. name string. Optional name operation. Defaults \"PiecewiseConstant\".","code":""},{"path":"https://keras3.posit.co/dev/reference/learning_rate_schedule_piecewise_constant_decay.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"A LearningRateSchedule that uses a piecewise constant decay schedule. — learning_rate_schedule_piecewise_constant_decay","text":"1-arg callable learning rate schedule takes current optimizer step outputs decayed learning rate, scalar tensor type boundary tensors. output 1-arg function takes step values[0] step <= boundaries[0], values[1] step > boundaries[0] step <= boundaries[1], ..., values[-1] step > boundaries[-1].","code":""},{"path":"https://keras3.posit.co/dev/reference/learning_rate_schedule_piecewise_constant_decay.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"A LearningRateSchedule that uses a piecewise constant decay schedule. — learning_rate_schedule_piecewise_constant_decay","text":"use learning rate 1.0 first 100001 steps, 0.5 next 10000 steps, 0.1 additional steps.   can pass schedule directly optimizer learning rate. learning rate schedule also serializable deserializable using keras$optimizers$schedules$serialize keras$optimizers$schedules$deserialize.","code":"step <- 0 boundaries <- c(100000, 110000) values <- c(1.0, 0.5, 0.1) learning_rate_fn <- learning_rate_schedule_piecewise_constant_decay(   boundaries, values)  # Later, whenever we perform an optimization step, we pass in the step. learning_rate <- learning_rate_fn(step)"},{"path":"https://keras3.posit.co/dev/reference/learning_rate_schedule_piecewise_constant_decay.html","id":"raises","dir":"Reference","previous_headings":"","what":"Raises","title":"A LearningRateSchedule that uses a piecewise constant decay schedule. — learning_rate_schedule_piecewise_constant_decay","text":"ValueError: number elements boundaries values lists match.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/learning_rate_schedule_polynomial_decay.html","id":null,"dir":"Reference","previous_headings":"","what":"A LearningRateSchedule that uses a polynomial decay schedule. — learning_rate_schedule_polynomial_decay","title":"A LearningRateSchedule that uses a polynomial decay schedule. — learning_rate_schedule_polynomial_decay","text":"commonly observed monotonically decreasing learning rate, whose degree change carefully chosen, results better performing model. schedule applies polynomial decay function optimizer step, given provided initial_learning_rate, reach end_learning_rate given decay_steps. requires step value compute decayed learning rate. can just pass backend variable increment training step. schedule 1-arg callable produces decayed learning rate passed current optimizer step. can useful changing learning rate value across different invocations optimizer functions. computed :   cycle TRUE multiple decay_steps used, first one bigger step.   can pass schedule directly Optimizer learning rate.","code":"decayed_learning_rate <- function(step) {   step = min(step, decay_steps)   ((initial_learning_rate - end_learning_rate) *     (1 - step / decay_steps) ^ (power)) +     end_learning_rate } decayed_learning_rate <- function(step) {   decay_steps = decay_steps * ceil(step / decay_steps)   ((initial_learning_rate - end_learning_rate) *       (1 - step / decay_steps) ^ (power)) +     end_learning_rate }"},{"path":"https://keras3.posit.co/dev/reference/learning_rate_schedule_polynomial_decay.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"A LearningRateSchedule that uses a polynomial decay schedule. — learning_rate_schedule_polynomial_decay","text":"","code":"learning_rate_schedule_polynomial_decay(   initial_learning_rate,   decay_steps,   end_learning_rate = 1e-04,   power = 1,   cycle = FALSE,   name = \"PolynomialDecay\" )"},{"path":"https://keras3.posit.co/dev/reference/learning_rate_schedule_polynomial_decay.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"A LearningRateSchedule that uses a polynomial decay schedule. — learning_rate_schedule_polynomial_decay","text":"initial_learning_rate float. initial learning rate. decay_steps integer. Must positive. See decay computation . end_learning_rate float. minimal end learning rate. power float. power polynomial. Defaults 1.0. cycle boolean, whether cycle beyond decay_steps. name String.  Optional name operation. Defaults \"PolynomialDecay\".","code":""},{"path":"https://keras3.posit.co/dev/reference/learning_rate_schedule_polynomial_decay.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"A LearningRateSchedule that uses a polynomial decay schedule. — learning_rate_schedule_polynomial_decay","text":"1-arg callable learning rate schedule takes current optimizer step outputs decayed learning rate, scalar tensor type initial_learning_rate.","code":""},{"path":"https://keras3.posit.co/dev/reference/learning_rate_schedule_polynomial_decay.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"A LearningRateSchedule that uses a polynomial decay schedule. — learning_rate_schedule_polynomial_decay","text":"Fit model decaying 0.1 0.01 10000 steps using sqrt (.e. power=0.5):   learning rate schedule also serializable deserializable using keras$optimizers$schedules$serialize keras$optimizers$schedules$deserialize.","code":"... starter_learning_rate <- 0.1 end_learning_rate <- 0.01 decay_steps <- 10000 learning_rate_fn <- learning_rate_schedule_polynomial_decay(     starter_learning_rate,     decay_steps,     end_learning_rate,     power=0.5)  model %>% compile(   optimizer = optimizer_sgd(learning_rate=learning_rate_fn),   loss = 'sparse_categorical_crossentropy',   metrics = 'accuracy' )  model %>% fit(data, labels, epochs=5)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/load_model.html","id":null,"dir":"Reference","previous_headings":"","what":"Loads a model saved via save_model(). — load_model","title":"Loads a model saved via save_model(). — load_model","text":"Loads model saved via save_model().","code":""},{"path":"https://keras3.posit.co/dev/reference/load_model.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Loads a model saved via save_model(). — load_model","text":"","code":"load_model(model, custom_objects = NULL, compile = TRUE, safe_mode = TRUE)"},{"path":"https://keras3.posit.co/dev/reference/load_model.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Loads a model saved via save_model(). — load_model","text":"model string, path saved model file, raw vector, returned save_model(filepath = NULL) custom_objects Optional named list mapping names custom classes functions considered deserialization. compile Boolean, whether compile model loading. safe_mode Boolean, whether disallow unsafe lambda deserialization. safe_mode=FALSE, loading object potential trigger arbitrary code execution. argument applicable Keras v3 model format. Defaults TRUE.","code":""},{"path":"https://keras3.posit.co/dev/reference/load_model.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Loads a model saved via save_model(). — load_model","text":"Keras model instance. original model compiled, argument compile = TRUE set, returned model compiled. Otherwise, model left uncompiled.","code":""},{"path":"https://keras3.posit.co/dev/reference/load_model.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Loads a model saved via save_model(). — load_model","text":"Note model variables may different name values (var$name property, e.g. \"dense_1/kernel:0\") reloaded. recommended use layer attributes access specific variables, e.g. model |> get_layer(\"dense_1\") |> _$kernel.","code":"model <- keras_model_sequential(input_shape = c(3)) |>   layer_dense(5) |>   layer_activation_softmax()  model |> save_model(\"model.keras\") loaded_model <- load_model(\"model.keras\") x <- random_uniform(c(10, 3)) stopifnot(all.equal(   model |> predict(x),   loaded_model |> predict(x) ))"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/load_model_weights.html","id":null,"dir":"Reference","previous_headings":"","what":"Load the weights from a single file or sharded files. — load_model_weights","title":"Load the weights from a single file or sharded files. — load_model_weights","text":"Weights loaded based network's topology. means architecture weights saved. Note layers weights taken account topological ordering, adding removing layers fine long weights. Sharding loading sharded weights, specify filepath ending \".weights.json\" (configuration file), corresponding shard files (*_xxxxx.weights.h5) located alongside . Partial weight loading modified model, instance adding new layer (weights) changing shape weights layer, can choose ignore errors continue loading setting skip_mismatch=TRUE. case layer mismatching weights skipped. warning displayed skipped layer.","code":""},{"path":"https://keras3.posit.co/dev/reference/load_model_weights.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Load the weights from a single file or sharded files. — load_model_weights","text":"","code":"load_model_weights(model, filepath, skip_mismatch = FALSE, ...)"},{"path":"https://keras3.posit.co/dev/reference/load_model_weights.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Load the weights from a single file or sharded files. — load_model_weights","text":"model keras model. filepath Path path-like object weights. Accepts .weights.h5, legacy .h5, sharded weights .weights.json manifest sitting alongside shard files (*_xxxxx.weights.h5). skip_mismatch Boolean, whether skip loading layers mismatch number weights, mismatch shape weights. ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/load_model_weights.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Load the weights from a single file or sharded files. — load_model_weights","text":"called primarily side effects. model returned, invisibly, enable usage pipe.","code":""},{"path":"https://keras3.posit.co/dev/reference/load_model_weights.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Load the weights from a single file or sharded files. — load_model_weights","text":"","code":"model |> load_model_weights(\"model.weights.h5\") model |> load_model_weights(\"model.weights.json\")"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/loss_binary_crossentropy.html","id":null,"dir":"Reference","previous_headings":"","what":"Computes the cross-entropy loss between true labels and predicted labels. — loss_binary_crossentropy","title":"Computes the cross-entropy loss between true labels and predicted labels. — loss_binary_crossentropy","text":"Use cross-entropy loss binary (0 1) classification applications. loss function requires following inputs: y_true (true label): either 0 1. y_pred (predicted value): model's prediction, .e, single floating-point value either represents logit, (.e, value [-inf, inf] from_logits=TRUE) probability (.e, value [0., 1.] from_logits=FALSE).","code":""},{"path":"https://keras3.posit.co/dev/reference/loss_binary_crossentropy.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes the cross-entropy loss between true labels and predicted labels. — loss_binary_crossentropy","text":"","code":"loss_binary_crossentropy(   y_true,   y_pred,   from_logits = FALSE,   label_smoothing = 0,   axis = -1L,   ...,   reduction = \"sum_over_batch_size\",   name = \"binary_crossentropy\",   dtype = NULL )"},{"path":"https://keras3.posit.co/dev/reference/loss_binary_crossentropy.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Computes the cross-entropy loss between true labels and predicted labels. — loss_binary_crossentropy","text":"y_true Ground truth values. shape = [batch_size, d0, .. dN]. y_pred predicted values. shape = [batch_size, d0, .. dN]. from_logits Whether interpret y_pred tensor logit values. default, assume y_pred probabilities (.e., values [0, 1)). label_smoothing Float range [0, 1]. 0, smoothing occurs. > 0, compute loss predicted labels smoothed version true labels, smoothing squeezes labels towards 0.5. Larger values label_smoothing correspond heavier smoothing. axis axis along compute crossentropy (features axis). Defaults -1. ... forward/backward compatability. reduction Type reduction apply loss. almost cases \"sum_over_batch_size\". Supported options \"sum\", \"sum_over_batch_size\", \"mean\", \"mean_with_sample_weight\" NULL. \"sum\" sums loss, \"sum_over_batch_size\" \"mean\" sum loss divide sample size, \"mean_with_sample_weight\" sums loss divides sum sample weights. \"none\" NULL perform aggregation. Defaults \"sum_over_batch_size\". name Optional name loss instance. dtype dtype loss's computations. Defaults NULL, means using config_floatx(). config_floatx() \"float32\" unless set different value (via config_set_floatx()). keras$DTypePolicy provided, compute_dtype utilized.","code":""},{"path":"https://keras3.posit.co/dev/reference/loss_binary_crossentropy.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Computes the cross-entropy loss between true labels and predicted labels. — loss_binary_crossentropy","text":"Binary crossentropy loss value. shape = [batch_size, d0, .. dN-1].","code":""},{"path":"https://keras3.posit.co/dev/reference/loss_binary_crossentropy.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Computes the cross-entropy loss between true labels and predicted labels. — loss_binary_crossentropy","text":"Recommended Usage: (set from_logits=TRUE) compile() API:   standalone function:                     Default Usage: (set from_logits=FALSE)","code":"y_true <- rbind(c(0, 1), c(0, 0)) y_pred <- rbind(c(0.6, 0.4), c(0.4, 0.6)) loss <- loss_binary_crossentropy(y_true, y_pred) loss ## tf.Tensor([0.91629076 0.7135582 ], shape=(2), dtype=float32) model %>% compile(     loss = loss_binary_crossentropy(from_logits=TRUE),     ... ) # Example 1: (batch_size = 1, number of samples = 4) y_true <- op_array(c(0, 1, 0, 0)) y_pred <- op_array(c(-18.6, 0.51, 2.94, -12.8)) bce <- loss_binary_crossentropy(from_logits = TRUE) bce(y_true, y_pred) ## tf.Tensor(0.865458, shape=(), dtype=float32) # Example 2: (batch_size = 2, number of samples = 4) y_true <- rbind(c(0, 1), c(0, 0)) y_pred <- rbind(c(-18.6, 0.51), c(2.94, -12.8)) # Using default 'auto'/'sum_over_batch_size' reduction type. bce <- loss_binary_crossentropy(from_logits = TRUE) bce(y_true, y_pred) ## tf.Tensor(0.865458, shape=(), dtype=float32) # Using 'sample_weight' attribute bce(y_true, y_pred, sample_weight = c(0.8, 0.2)) ## tf.Tensor(0.2436386, shape=(), dtype=float32) # 0.243 # Using 'sum' reduction` type. bce <- loss_binary_crossentropy(from_logits = TRUE, reduction = \"sum\") bce(y_true, y_pred) ## tf.Tensor(1.730916, shape=(), dtype=float32) # Using 'none' reduction type. bce <- loss_binary_crossentropy(from_logits = TRUE, reduction = NULL) bce(y_true, y_pred) ## tf.Tensor([0.23515666 1.4957594 ], shape=(2), dtype=float32) # Make the following updates to the above \"Recommended Usage\" section # 1. Set `from_logits=FALSE` loss_binary_crossentropy() # OR ...('from_logits=FALSE') ## <LossFunctionWrapper(<function binary_crossentropy at 0x0>, kwargs={'from_logits': False, 'label_smoothing': 0.0, 'axis': -1})> ##  signature: (y_true, y_pred, sample_weight=None) # 2. Update `y_pred` to use probabilities instead of logits y_pred <- c(0.6, 0.3, 0.2, 0.8) # OR [[0.6, 0.3], [0.2, 0.8]]"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/loss_binary_focal_crossentropy.html","id":null,"dir":"Reference","previous_headings":"","what":"Computes focal cross-entropy loss between true labels and predictions. — loss_binary_focal_crossentropy","title":"Computes focal cross-entropy loss between true labels and predictions. — loss_binary_focal_crossentropy","text":"According Lin et al., 2018, helps apply focal factor -weight easy examples focus hard examples. default, focal tensor computed follows: focal_factor = (1 - output)^gamma class 1 focal_factor = output^gamma class 0 gamma focusing parameter. gamma = 0, focal effect binary crossentropy loss. apply_class_balancing == TRUE, function also takes account weight balancing factor binary classes 0 1 follows: weight = alpha class 1 (target == 1) weight = 1 - alpha class 0 alpha float range [0, 1]. Binary cross-entropy loss often used binary (0 1) classification tasks. loss function requires following inputs: y_true (true label): either 0 1. y_pred (predicted value): model's prediction, .e, single floating-point value either represents logit, (.e, value [-inf, inf] from_logits=TRUE) probability (.e, value [0., 1.] from_logits=FALSE). According Lin et al., 2018, helps apply \"focal factor\" -weight easy examples focus hard examples. default, focal tensor computed follows: focal_factor = (1 - output) ** gamma class 1 focal_factor = output ** gamma class 0 gamma focusing parameter. gamma=0, function equivalent binary crossentropy loss.","code":""},{"path":"https://keras3.posit.co/dev/reference/loss_binary_focal_crossentropy.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes focal cross-entropy loss between true labels and predictions. — loss_binary_focal_crossentropy","text":"","code":"loss_binary_focal_crossentropy(   y_true,   y_pred,   apply_class_balancing = FALSE,   alpha = 0.25,   gamma = 2,   from_logits = FALSE,   label_smoothing = 0,   axis = -1L,   ...,   reduction = \"sum_over_batch_size\",   name = \"binary_focal_crossentropy\",   dtype = NULL )"},{"path":"https://keras3.posit.co/dev/reference/loss_binary_focal_crossentropy.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Computes focal cross-entropy loss between true labels and predictions. — loss_binary_focal_crossentropy","text":"y_true Ground truth values, shape (batch_size, d0, .. dN). y_pred predicted values, shape (batch_size, d0, .. dN). apply_class_balancing bool, whether apply weight balancing binary classes 0 1. alpha weight balancing factor class 1, default 0.25 mentioned reference Lin et al., 2018.  weight class 0 1.0 - alpha. gamma focusing parameter used compute focal factor, default 2.0 mentioned reference Lin et al., 2018. from_logits Whether interpret y_pred tensor logit values. default, assume y_pred probabilities (.e., values [0, 1]). label_smoothing Float [0, 1]. 0, smoothing occurs. > 0, compute loss predicted labels smoothed version true labels, smoothing squeezes labels towards 0.5. Larger values label_smoothing correspond heavier smoothing. axis axis along compute crossentropy (features axis). Defaults -1. ... forward/backward compatability. reduction Type reduction apply loss. almost cases \"sum_over_batch_size\". Supported options \"sum\", \"sum_over_batch_size\", \"mean\", \"mean_with_sample_weight\" NULL. \"sum\" sums loss, \"sum_over_batch_size\" \"mean\" sum loss divide sample size, \"mean_with_sample_weight\" sums loss divides sum sample weights. \"none\" NULL perform aggregation. Defaults \"sum_over_batch_size\". name Optional name loss instance. dtype dtype loss's computations. Defaults NULL, means using config_floatx(). config_floatx() \"float32\" unless set different value (via config_set_floatx()). keras$DTypePolicy provided, compute_dtype utilized.","code":""},{"path":"https://keras3.posit.co/dev/reference/loss_binary_focal_crossentropy.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Computes focal cross-entropy loss between true labels and predictions. — loss_binary_focal_crossentropy","text":"Binary focal crossentropy loss value shape = [batch_size, d0, .. dN-1].","code":""},{"path":"https://keras3.posit.co/dev/reference/loss_binary_focal_crossentropy.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Computes focal cross-entropy loss between true labels and predictions. — loss_binary_focal_crossentropy","text":"compile() API:   standalone function:","code":"y_true <- rbind(c(0, 1), c(0, 0)) y_pred <- rbind(c(0.6, 0.4), c(0.4, 0.6)) loss <- loss_binary_focal_crossentropy(y_true, y_pred, gamma = 2) loss ## tf.Tensor([0.32986468 0.20579839], shape=(2), dtype=float32) model %>% compile(     loss = loss_binary_focal_crossentropy(         gamma = 2.0, from_logits = TRUE),     ... ) # Example 1: (batch_size = 1, number of samples = 4) y_true <- op_array(c(0, 1, 0, 0)) y_pred <- op_array(c(-18.6, 0.51, 2.94, -12.8)) loss <- loss_binary_focal_crossentropy(gamma = 2, from_logits = TRUE) loss(y_true, y_pred) ## tf.Tensor(0.6912122, shape=(), dtype=float32) # Apply class weight loss <- loss_binary_focal_crossentropy(   apply_class_balancing = TRUE, gamma = 2, from_logits = TRUE) loss(y_true, y_pred) ## tf.Tensor(0.5101333, shape=(), dtype=float32) # Example 2: (batch_size = 2, number of samples = 4) y_true <- rbind(c(0, 1), c(0, 0)) y_pred <- rbind(c(-18.6, 0.51), c(2.94, -12.8)) # Using default 'auto'/'sum_over_batch_size' reduction type. loss <- loss_binary_focal_crossentropy(     gamma = 3, from_logits = TRUE) loss(y_true, y_pred) ## tf.Tensor(0.6469951, shape=(), dtype=float32) # Apply class weight loss <- loss_binary_focal_crossentropy(      apply_class_balancing = TRUE, gamma = 3, from_logits = TRUE) loss(y_true, y_pred) ## tf.Tensor(0.48214132, shape=(), dtype=float32) # Using 'sample_weight' attribute with focal effect loss <- loss_binary_focal_crossentropy(     gamma = 3, from_logits = TRUE) loss(y_true, y_pred, sample_weight = c(0.8, 0.2)) ## tf.Tensor(0.13312504, shape=(), dtype=float32) # Apply class weight loss <- loss_binary_focal_crossentropy(      apply_class_balancing = TRUE, gamma = 3, from_logits = TRUE) loss(y_true, y_pred, sample_weight = c(0.8, 0.2)) ## tf.Tensor(0.09735977, shape=(), dtype=float32) # Using 'sum' reduction` type. loss <- loss_binary_focal_crossentropy(     gamma = 4, from_logits = TRUE,     reduction = \"sum\") loss(y_true, y_pred) ## tf.Tensor(1.2218808, shape=(), dtype=float32) # Apply class weight loss <- loss_binary_focal_crossentropy(     apply_class_balancing = TRUE, gamma = 4, from_logits = TRUE,     reduction = \"sum\") loss(y_true, y_pred) ## tf.Tensor(0.9140807, shape=(), dtype=float32) # Using 'none' reduction type. loss <- loss_binary_focal_crossentropy(     gamma = 5, from_logits = TRUE,     reduction = NULL) loss(y_true, y_pred) ## tf.Tensor([0.00174837 1.1561027 ], shape=(2), dtype=float32) # Apply class weight loss <- loss_binary_focal_crossentropy(     apply_class_balancing = TRUE, gamma = 5, from_logits = TRUE,     reduction = NULL) loss(y_true, y_pred) ## tf.Tensor([4.3709317e-04 8.6707699e-01], shape=(2), dtype=float32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/loss_categorical_crossentropy.html","id":null,"dir":"Reference","previous_headings":"","what":"Computes the crossentropy loss between the labels and predictions. — loss_categorical_crossentropy","title":"Computes the crossentropy loss between the labels and predictions. — loss_categorical_crossentropy","text":"Use crossentropy loss function two label classes. expect labels provided one_hot representation. want provide labels integers, please use SparseCategoricalCrossentropy loss. num_classes floating point values per feature, .e., shape y_pred y_true [batch_size, num_classes].","code":""},{"path":"https://keras3.posit.co/dev/reference/loss_categorical_crossentropy.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes the crossentropy loss between the labels and predictions. — loss_categorical_crossentropy","text":"","code":"loss_categorical_crossentropy(   y_true,   y_pred,   from_logits = FALSE,   label_smoothing = 0,   axis = -1L,   ...,   reduction = \"sum_over_batch_size\",   name = \"categorical_crossentropy\",   dtype = NULL )"},{"path":"https://keras3.posit.co/dev/reference/loss_categorical_crossentropy.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Computes the crossentropy loss between the labels and predictions. — loss_categorical_crossentropy","text":"y_true Tensor one-hot true targets. y_pred Tensor predicted targets. from_logits Whether y_pred expected logits tensor. default, assume y_pred encodes probability distribution. label_smoothing Float [0, 1]. > 0, label values smoothed, meaning confidence label values relaxed. example, 0.1, use 0.1 / num_classes non-target labels 0.9 + 0.1 / num_classes target labels. axis axis along compute crossentropy (features axis). Defaults -1. ... forward/backward compatability. reduction Type reduction apply loss. almost cases \"sum_over_batch_size\". Supported options \"sum\", \"sum_over_batch_size\", \"mean\", \"mean_with_sample_weight\" NULL. \"sum\" sums loss, \"sum_over_batch_size\" \"mean\" sum loss divide sample size, \"mean_with_sample_weight\" sums loss divides sum sample weights. \"none\" NULL perform aggregation. Defaults \"sum_over_batch_size\". name Optional name loss instance. dtype dtype loss's computations. Defaults NULL, means using config_floatx(). config_floatx() \"float32\" unless set different value (via config_set_floatx()). keras$DTypePolicy provided, compute_dtype utilized.","code":""},{"path":"https://keras3.posit.co/dev/reference/loss_categorical_crossentropy.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Computes the crossentropy loss between the labels and predictions. — loss_categorical_crossentropy","text":"Categorical crossentropy loss value.","code":""},{"path":"https://keras3.posit.co/dev/reference/loss_categorical_crossentropy.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Computes the crossentropy loss between the labels and predictions. — loss_categorical_crossentropy","text":"Standalone usage:                 Usage compile() API:","code":"y_true <- rbind(c(0, 1, 0), c(0, 0, 1)) y_pred <- rbind(c(0.05, 0.95, 0), c(0.1, 0.8, 0.1)) loss <- loss_categorical_crossentropy(y_true, y_pred) loss ## tf.Tensor([0.05129331 2.3025851 ], shape=(2), dtype=float32) y_true <- rbind(c(0, 1, 0), c(0, 0, 1)) y_pred <- rbind(c(0.05, 0.95, 0), c(0.1, 0.8, 0.1)) # Using 'auto'/'sum_over_batch_size' reduction type. cce <- loss_categorical_crossentropy() cce(y_true, y_pred) ## tf.Tensor(1.1769392, shape=(), dtype=float32) # Calling with 'sample_weight'. cce(y_true, y_pred, sample_weight = op_array(c(0.3, 0.7))) ## tf.Tensor(0.8135988, shape=(), dtype=float32) # Using 'sum' reduction type. cce <- loss_categorical_crossentropy(reduction = \"sum\") cce(y_true, y_pred) ## tf.Tensor(2.3538785, shape=(), dtype=float32) # Using 'none' reduction type. cce <- loss_categorical_crossentropy(reduction = NULL) cce(y_true, y_pred) ## tf.Tensor([0.05129331 2.3025851 ], shape=(2), dtype=float32) model %>% compile(optimizer = 'sgd',               loss=loss_categorical_crossentropy())"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/loss_categorical_focal_crossentropy.html","id":null,"dir":"Reference","previous_headings":"","what":"Computes the alpha balanced focal crossentropy loss. — loss_categorical_focal_crossentropy","title":"Computes the alpha balanced focal crossentropy loss. — loss_categorical_focal_crossentropy","text":"Use crossentropy loss function two label classes want handle class imbalance without using class_weights. expect labels provided one_hot representation. According Lin et al., 2018, helps apply focal factor -weight easy examples focus hard examples. general formula focal loss (FL) follows: FL(p_t) = (1 - p_t)^gamma * log(p_t) p_t defined follows: p_t = output y_true == 1, else 1 - output (1 - p_t)^gamma modulating_factor, gamma focusing parameter. gamma = 0, focal effect cross entropy. gamma reduces importance given simple examples smooth manner. authors use alpha-balanced variant focal loss (FL) paper: FL(p_t) = -alpha * (1 - p_t)^gamma * log(p_t) alpha weight factor classes. alpha = 1, loss able handle class imbalance properly classes weight. can constant list constants. alpha list, must length number classes. formula can generalized : FL(p_t) = alpha * (1 - p_t)^gamma * CrossEntropy(y_true, y_pred) minus comes CrossEntropy(y_true, y_pred) (CE). Extending multi-class case straightforward: FL(p_t) = alpha * (1 - p_t) ** gamma * CategoricalCE(y_true, y_pred) snippet , num_classes floating pointing values per example. shape y_pred y_true (batch_size, num_classes).","code":""},{"path":"https://keras3.posit.co/dev/reference/loss_categorical_focal_crossentropy.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes the alpha balanced focal crossentropy loss. — loss_categorical_focal_crossentropy","text":"","code":"loss_categorical_focal_crossentropy(   y_true,   y_pred,   alpha = 0.25,   gamma = 2,   from_logits = FALSE,   label_smoothing = 0,   axis = -1L,   ...,   reduction = \"sum_over_batch_size\",   name = \"categorical_focal_crossentropy\",   dtype = NULL )"},{"path":"https://keras3.posit.co/dev/reference/loss_categorical_focal_crossentropy.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Computes the alpha balanced focal crossentropy loss. — loss_categorical_focal_crossentropy","text":"y_true Tensor one-hot true targets. y_pred Tensor predicted targets. alpha weight balancing factor classes, default 0.25 mentioned reference. can list floats scalar. multi-class case, alpha may set inverse class frequency using compute_class_weight sklearn.utils. gamma focusing parameter, default 2.0 mentioned reference. helps gradually reduce importance given simple examples smooth manner. gamma = 0, focal effect categorical crossentropy. from_logits Whether output expected logits tensor. default, consider output encodes probability distribution. label_smoothing Float [0, 1]. > 0, label values smoothed, meaning confidence label values relaxed. example, 0.1, use 0.1 / num_classes non-target labels 0.9 + 0.1 / num_classes target labels. axis axis along compute crossentropy (features axis). Defaults -1. ... forward/backward compatability. reduction Type reduction apply loss. almost cases \"sum_over_batch_size\". Supported options \"sum\", \"sum_over_batch_size\", \"mean\", \"mean_with_sample_weight\" NULL. \"sum\" sums loss, \"sum_over_batch_size\" \"mean\" sum loss divide sample size, \"mean_with_sample_weight\" sums loss divides sum sample weights. \"none\" NULL perform aggregation. Defaults \"sum_over_batch_size\". name Optional name loss instance. dtype dtype loss's computations. Defaults NULL, means using config_floatx(). config_floatx() \"float32\" unless set different value (via config_set_floatx()). keras$DTypePolicy provided, compute_dtype utilized.","code":""},{"path":"https://keras3.posit.co/dev/reference/loss_categorical_focal_crossentropy.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Computes the alpha balanced focal crossentropy loss. — loss_categorical_focal_crossentropy","text":"Categorical focal crossentropy loss value.","code":""},{"path":"https://keras3.posit.co/dev/reference/loss_categorical_focal_crossentropy.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Computes the alpha balanced focal crossentropy loss. — loss_categorical_focal_crossentropy","text":"Standalone usage:                 Usage compile() API:","code":"y_true <- rbind(c(0, 1, 0), c(0, 0, 1)) y_pred <- rbind(c(0.05, 0.95, 0), c(0.1, 0.8, 0.1)) loss <- loss_categorical_focal_crossentropy(y_true, y_pred) loss ## tf.Tensor([3.2058331e-05 4.6627346e-01], shape=(2), dtype=float32) y_true <- rbind(c(0, 1, 0), c(0, 0, 1)) y_pred <- rbind(c(0.05, 0.95, 0), c(0.1, 0.8, 0.1)) # Using 'auto'/'sum_over_batch_size' reduction type. cce <- loss_categorical_focal_crossentropy() cce(y_true, y_pred) ## tf.Tensor(0.23315276, shape=(), dtype=float32) # Calling with 'sample_weight'. cce(y_true, y_pred, sample_weight = op_array(c(0.3, 0.7))) ## tf.Tensor(0.16320053, shape=(), dtype=float32) # Using 'sum' reduction type. cce <- loss_categorical_focal_crossentropy(reduction = \"sum\") cce(y_true, y_pred) ## tf.Tensor(0.46630552, shape=(), dtype=float32) # Using 'none' reduction type. cce <- loss_categorical_focal_crossentropy(reduction = NULL) cce(y_true, y_pred) ## tf.Tensor([3.2058331e-05 4.6627346e-01], shape=(2), dtype=float32) model %>% compile(   optimizer = 'adam',   loss = loss_categorical_focal_crossentropy())"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/loss_categorical_generalized_cross_entropy.html","id":null,"dir":"Reference","previous_headings":"","what":"Computes the generalized cross entropy loss. — loss_categorical_generalized_cross_entropy","title":"Computes the generalized cross entropy loss. — loss_categorical_generalized_cross_entropy","text":"generalized cross entropy (GCE) loss offers robustness noisy labels interpolating categorical cross entropy (q -> 0) mean absolute error (q -> 1). true-class probability p noise parameter q, loss loss = (1 - p^q) / q.","code":""},{"path":"https://keras3.posit.co/dev/reference/loss_categorical_generalized_cross_entropy.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes the generalized cross entropy loss. — loss_categorical_generalized_cross_entropy","text":"","code":"loss_categorical_generalized_cross_entropy(   y_true,   y_pred,   q = 0.5,   ...,   reduction = \"sum_over_batch_size\",   name = \"categorical_generalized_cross_entropy\",   dtype = NULL )"},{"path":"https://keras3.posit.co/dev/reference/loss_categorical_generalized_cross_entropy.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Computes the generalized cross entropy loss. — loss_categorical_generalized_cross_entropy","text":"y_true Integer class indices shape (batch_size) (batch_size, 1). y_pred Predicted class probabilities shape (batch_size, num_classes). q Float (0, 1). Controls transition cross entropy mean absolute error. Defaults 0.5. q approaches 0: behaves like categorical cross entropy. q approaches 1: behaves like mean absolute error. ... forward/backward compatibility. reduction Type reduction apply loss. almost cases \"sum_over_batch_size\". Supported options \"sum\", \"sum_over_batch_size\", \"mean\", \"mean_with_sample_weight\" NULL. \"sum\" sums loss, \"sum_over_batch_size\" \"mean\" sum loss divide sample size, \"mean_with_sample_weight\" sums loss divides sum sample weights. \"none\" NULL perform aggregation. Defaults \"sum_over_batch_size\". name Optional name loss instance. dtype Dtype used loss computations. Defaults config_floatx() (global float type).","code":""},{"path":"https://keras3.posit.co/dev/reference/loss_categorical_generalized_cross_entropy.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Computes the generalized cross entropy loss. — loss_categorical_generalized_cross_entropy","text":"Generalized cross entropy loss value(s).","code":""},{"path":"https://keras3.posit.co/dev/reference/loss_categorical_generalized_cross_entropy.html","id":"references","dir":"Reference","previous_headings":"","what":"References","title":"Computes the generalized cross entropy loss. — loss_categorical_generalized_cross_entropy","text":"Zhang & Sabuncu (2018), \"Generalized Cross Entropy Loss Training Deep Neural Networks Noisy Labels\"","code":""},{"path":"https://keras3.posit.co/dev/reference/loss_categorical_generalized_cross_entropy.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Computes the generalized cross entropy loss. — loss_categorical_generalized_cross_entropy","text":"","code":"y_true <- c(0L, 1L, 0L, 1L) y_pred <- rbind(   c(0.7, 0.3),   c(0.2, 0.8),   c(0.6, 0.4),   c(0.4, 0.6) ) gce <- loss_categorical_generalized_cross_entropy(q = 0.7) gce(y_true, y_pred) ## tf.Tensor(0.34529287, shape=(), dtype=float32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/loss_categorical_hinge.html","id":null,"dir":"Reference","previous_headings":"","what":"Computes the categorical hinge loss between y_true & y_pred. — loss_categorical_hinge","title":"Computes the categorical hinge loss between y_true & y_pred. — loss_categorical_hinge","text":"Formula:   neg=maximum((1-y_true)*y_pred) pos=sum(y_true*y_pred)","code":"loss <- maximum(neg - pos + 1, 0)"},{"path":"https://keras3.posit.co/dev/reference/loss_categorical_hinge.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes the categorical hinge loss between y_true & y_pred. — loss_categorical_hinge","text":"","code":"loss_categorical_hinge(   y_true,   y_pred,   ...,   reduction = \"sum_over_batch_size\",   name = \"categorical_hinge\",   dtype = NULL )"},{"path":"https://keras3.posit.co/dev/reference/loss_categorical_hinge.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Computes the categorical hinge loss between y_true & y_pred. — loss_categorical_hinge","text":"y_true ground truth values. y_true values expected either {-1, +1} {0, 1} (.e. one-hot-encoded tensor) shape <- [batch_size, d0, .. dN]. y_pred predicted values shape = [batch_size, d0, .. dN]. ... forward/backward compatability. reduction Type reduction apply loss. almost cases \"sum_over_batch_size\". Supported options \"sum\", \"sum_over_batch_size\", \"mean\", \"mean_with_sample_weight\" NULL. \"sum\" sums loss, \"sum_over_batch_size\" \"mean\" sum loss divide sample size, \"mean_with_sample_weight\" sums loss divides sum sample weights. \"none\" NULL perform aggregation. Defaults \"sum_over_batch_size\". name Optional name loss instance. dtype dtype loss's computations. Defaults NULL, means using config_floatx(). config_floatx() \"float32\" unless set different value (via config_set_floatx()). keras$DTypePolicy provided, compute_dtype utilized.","code":""},{"path":"https://keras3.posit.co/dev/reference/loss_categorical_hinge.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Computes the categorical hinge loss between y_true & y_pred. — loss_categorical_hinge","text":"Categorical hinge loss values shape = [batch_size, d0, .. dN-1].","code":""},{"path":"https://keras3.posit.co/dev/reference/loss_categorical_hinge.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Computes the categorical hinge loss between y_true & y_pred. — loss_categorical_hinge","text":"","code":"y_true <- rbind(c(0, 1), c(0, 0)) y_pred <- rbind(c(0.6, 0.4), c(0.4, 0.6)) loss <- loss_categorical_hinge(y_true, y_pred)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/loss_circle.html","id":null,"dir":"Reference","previous_headings":"","what":"Computes Circle Loss between integer labels and L2-normalized embeddings. — loss_circle","title":"Computes Circle Loss between integer labels and L2-normalized embeddings. — loss_circle","text":"designed minimize within-class distances maximize -class distances L2 normalized embedding space. metric learning loss designed minimize within-class distance maximize -class distance flexible manner dynamically adjusting penalty strength based optimization status similarity score. use Circle Loss effectively, model output embeddings without activation function (Dense layer activation=NULL) followed UnitNormalization layer ensure unit-norm embeddings.","code":""},{"path":"https://keras3.posit.co/dev/reference/loss_circle.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes Circle Loss between integer labels and L2-normalized embeddings. — loss_circle","text":"","code":"loss_circle(   y_true,   y_pred,   ref_labels = NULL,   ref_embeddings = NULL,   remove_diagonal = TRUE,   gamma = 80L,   margin = 0.4,   ...,   reduction = \"sum_over_batch_size\",   name = \"circle\",   dtype = NULL )"},{"path":"https://keras3.posit.co/dev/reference/loss_circle.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Computes Circle Loss between integer labels and L2-normalized embeddings. — loss_circle","text":"y_true Tensor ground truth labels integer format. y_pred Tensor predicted L2 normalized embeddings. ref_labels Optional integer tensor labels reference embeddings. NULL, defaults y_true. ref_embeddings Optional tensor L2 normalized reference embeddings. NULL, defaults y_pred. remove_diagonal Boolean, whether remove self-similarities positive mask. Defaults TRUE. gamma Scaling factor determines largest scale similarity score. Defaults 80. margin relaxation factor, distance, negatives weighted positives weighted. Similarly, distance negatives weighted positive weighted. Defaults 0.4. ... forward/backward compatability. reduction Type reduction apply loss. almost cases \"sum_over_batch_size\". Supported options \"sum\", \"sum_over_batch_size\", \"mean\", \"mean_with_sample_weight\" NULL. \"sum\" sums loss, \"sum_over_batch_size\" \"mean\" sum loss divide sample size, \"mean_with_sample_weight\" sums loss divides sum sample weights. \"none\" NULL perform aggregation. Defaults \"sum_over_batch_size\". name Optional name loss instance. dtype dtype loss's computations. Defaults NULL, means using config_floatx(). config_floatx() \"float32\" unless set different value (via config_set_floatx()). keras.DTypePolicy provided, compute_dtype utilized.","code":""},{"path":"https://keras3.posit.co/dev/reference/loss_circle.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Computes Circle Loss between integer labels and L2-normalized embeddings. — loss_circle","text":"Circle loss value.","code":""},{"path":"https://keras3.posit.co/dev/reference/loss_circle.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Computes Circle Loss between integer labels and L2-normalized embeddings. — loss_circle","text":"Usage compile() API:","code":"model <- keras_model_sequential(input_shape = c(224, 224, 3)) |>   layer_conv_2d(16, c(3, 3), activation = 'relu') |>   layer_flatten() |>   layer_dense(64, activation = NULL) |>   # No activation   layer_unit_normalization()  # L2 normalization  model |>   compile(optimizer = \"adam\", loss = loss_circle())"},{"path":"https://keras3.posit.co/dev/reference/loss_circle.html","id":"reference","dir":"Reference","previous_headings":"","what":"Reference","title":"Computes Circle Loss between integer labels and L2-normalized embeddings. — loss_circle","text":"Yifan Sun et al., 2020","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/loss_cosine_similarity.html","id":null,"dir":"Reference","previous_headings":"","what":"Computes the cosine similarity between y_true & y_pred. — loss_cosine_similarity","title":"Computes the cosine similarity between y_true & y_pred. — loss_cosine_similarity","text":"Formula:   Note number -1 1. negative number -1 0, 0 indicates orthogonality values closer -1 indicate greater similarity. makes usable loss function setting try maximize proximity predictions targets. either y_true y_pred zero vector, cosine similarity 0 regardless proximity predictions targets.","code":"loss <- -sum(l2_norm(y_true) * l2_norm(y_pred))"},{"path":"https://keras3.posit.co/dev/reference/loss_cosine_similarity.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes the cosine similarity between y_true & y_pred. — loss_cosine_similarity","text":"","code":"loss_cosine_similarity(   y_true,   y_pred,   axis = -1L,   ...,   reduction = \"sum_over_batch_size\",   name = \"cosine_similarity\",   dtype = NULL )"},{"path":"https://keras3.posit.co/dev/reference/loss_cosine_similarity.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Computes the cosine similarity between y_true & y_pred. — loss_cosine_similarity","text":"y_true Tensor true targets. y_pred Tensor predicted targets. axis axis along cosine similarity computed (features axis). Defaults -1. ... forward/backward compatability. reduction Type reduction apply loss. almost cases \"sum_over_batch_size\". Supported options \"sum\", \"sum_over_batch_size\", \"mean\", \"mean_with_sample_weight\" NULL. \"sum\" sums loss, \"sum_over_batch_size\" \"mean\" sum loss divide sample size, \"mean_with_sample_weight\" sums loss divides sum sample weights. \"none\" NULL perform aggregation. Defaults \"sum_over_batch_size\". name Optional name loss instance. dtype dtype loss's computations. Defaults NULL, means using config_floatx(). config_floatx() \"float32\" unless set different value (via config_set_floatx()). keras$DTypePolicy provided, compute_dtype utilized.","code":""},{"path":"https://keras3.posit.co/dev/reference/loss_cosine_similarity.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Computes the cosine similarity between y_true & y_pred. — loss_cosine_similarity","text":"Cosine similarity tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/loss_cosine_similarity.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Computes the cosine similarity between y_true & y_pred. — loss_cosine_similarity","text":"","code":"y_true <- rbind(c(0., 1.), c(1., 1.), c(1., 1.)) y_pred <- rbind(c(1., 0.), c(1., 1.), c(-1., -1.)) loss <- loss_cosine_similarity(y_true, y_pred, axis=-1) loss ## tf.Tensor([-0.         -0.99999994  0.99999994], shape=(3), dtype=float32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/loss_ctc.html","id":null,"dir":"Reference","previous_headings":"","what":"CTC (Connectionist Temporal Classification) loss. — loss_ctc","title":"CTC (Connectionist Temporal Classification) loss. — loss_ctc","text":"CTC (Connectionist Temporal Classification) loss.","code":""},{"path":"https://keras3.posit.co/dev/reference/loss_ctc.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"CTC (Connectionist Temporal Classification) loss. — loss_ctc","text":"","code":"loss_ctc(   y_true,   y_pred,   ...,   reduction = \"sum_over_batch_size\",   name = \"ctc\",   dtype = NULL )"},{"path":"https://keras3.posit.co/dev/reference/loss_ctc.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"CTC (Connectionist Temporal Classification) loss. — loss_ctc","text":"y_true tensor shape (batch_size, target_max_length) containing true labels integer format. 0 always represents blank/mask index used classes. y_pred tensor shape (batch_size, output_max_length, num_classes) containing logits (output model). normalized via softmax. ... forward/backward compatability. reduction Type reduction apply loss. almost cases \"sum_over_batch_size\". Supported options \"sum\", \"sum_over_batch_size\", \"mean\", \"mean_with_sample_weight\" NULL. \"sum\" sums loss, \"sum_over_batch_size\" \"mean\" sum loss divide sample size, \"mean_with_sample_weight\" sums loss divides sum sample weights. \"none\" NULL perform aggregation. Defaults \"sum_over_batch_size\". name Optional name loss instance. dtype dtype loss's computations. Defaults NULL, means using config_floatx(). config_floatx() \"float32\" unless set different value (via config_set_floatx()). keras$DTypePolicy provided, compute_dtype utilized.","code":""},{"path":"https://keras3.posit.co/dev/reference/loss_ctc.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"CTC (Connectionist Temporal Classification) loss. — loss_ctc","text":"CTC loss value.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/loss_dice.html","id":null,"dir":"Reference","previous_headings":"","what":"Computes the Dice loss value between y_true and y_pred. — loss_dice","title":"Computes the Dice loss value between y_true and y_pred. — loss_dice","text":"Formula:   Formula:","code":"loss = 1 - (2 * sum(y_true * y_pred)) / (sum(y_true) + sum(y_pred)) loss = 1 - (2 * sum(y_true * y_pred)) / (sum(y_true) + sum(y_pred))"},{"path":"https://keras3.posit.co/dev/reference/loss_dice.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes the Dice loss value between y_true and y_pred. — loss_dice","text":"","code":"loss_dice(   y_true,   y_pred,   ...,   reduction = \"sum_over_batch_size\",   name = \"dice\",   axis = NULL,   dtype = NULL )"},{"path":"https://keras3.posit.co/dev/reference/loss_dice.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Computes the Dice loss value between y_true and y_pred. — loss_dice","text":"y_true Tensor true targets. y_pred Tensor predicted targets. ... forward/backward compatability. reduction Type reduction apply loss. almost cases \"sum_over_batch_size\". Supported options \"sum\", \"sum_over_batch_size\", \"mean\", \"mean_with_sample_weight\" NULL. \"sum\" sums loss, \"sum_over_batch_size\" \"mean\" sum loss divide sample size, \"mean_with_sample_weight\" sums loss divides sum sample weights. \"none\" NULL perform aggregation. Defaults \"sum_over_batch_size\". name String, name object axis List dimensions loss calculated. Defaults NULL. dtype dtype loss's computations. Defaults NULL, means using config_floatx(). config_floatx() \"float32\" unless set different value (via config_set_floatx()). keras$DTypePolicy provided, compute_dtype utilized.","code":""},{"path":"https://keras3.posit.co/dev/reference/loss_dice.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Computes the Dice loss value between y_true and y_pred. — loss_dice","text":"y_true y_pred provided, Dice loss value. Otherwise, Loss() instance.","code":""},{"path":"https://keras3.posit.co/dev/reference/loss_dice.html","id":"example","dir":"Reference","previous_headings":"","what":"Example","title":"Computes the Dice loss value between y_true and y_pred. — loss_dice","text":"","code":"y_true <- array(c(1, 1, 0, 0,                   1, 1, 0, 0), dim = c(2, 2, 2, 1)) y_pred <- array(c(0, 0.4, 0,   0,                   1,   0, 1, 0.9), dim = c(2, 2, 2, 1))  axis <- c(2, 3, 4) loss_fn <- loss_dice(axis = axis, reduction = NULL) loss <- loss_fn(y_true, y_pred) stopifnot(shape(loss) == shape(2)) loss ## tf.Tensor([0.5        0.75757575], shape=(2), dtype=float32) loss_fn <- loss_dice() loss <- loss_fn(y_true, y_pred) stopifnot(shape(loss) == shape()) loss ## tf.Tensor(0.6164384, shape=(), dtype=float32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/loss_hinge.html","id":null,"dir":"Reference","previous_headings":"","what":"Computes the hinge loss between y_true & y_pred. — loss_hinge","title":"Computes the hinge loss between y_true & y_pred. — loss_hinge","text":"Formula:   y_true values expected -1 1. binary (0 1) labels provided convert -1 1.","code":"loss <- mean(maximum(1 - y_true * y_pred, 0), axis=-1)"},{"path":"https://keras3.posit.co/dev/reference/loss_hinge.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes the hinge loss between y_true & y_pred. — loss_hinge","text":"","code":"loss_hinge(   y_true,   y_pred,   ...,   reduction = \"sum_over_batch_size\",   name = \"hinge\",   dtype = NULL )"},{"path":"https://keras3.posit.co/dev/reference/loss_hinge.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Computes the hinge loss between y_true & y_pred. — loss_hinge","text":"y_true ground truth values. y_true values expected -1 1. binary (0 1) labels provided converted -1 1 shape = [batch_size, d0, .. dN]. y_pred predicted values shape = [batch_size, d0, .. dN]. ... forward/backward compatability. reduction Type reduction apply loss. almost cases \"sum_over_batch_size\". Supported options \"sum\", \"sum_over_batch_size\", \"mean\", \"mean_with_sample_weight\" NULL. \"sum\" sums loss, \"sum_over_batch_size\" \"mean\" sum loss divide sample size, \"mean_with_sample_weight\" sums loss divides sum sample weights. \"none\" NULL perform aggregation. Defaults \"sum_over_batch_size\". name Optional name loss instance. dtype dtype loss's computations. Defaults NULL, means using config_floatx(). config_floatx() \"float32\" unless set different value (via config_set_floatx()). keras$DTypePolicy provided, compute_dtype utilized.","code":""},{"path":"https://keras3.posit.co/dev/reference/loss_hinge.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Computes the hinge loss between y_true & y_pred. — loss_hinge","text":"Hinge loss values shape = [batch_size, d0, .. dN-1].","code":""},{"path":"https://keras3.posit.co/dev/reference/loss_hinge.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Computes the hinge loss between y_true & y_pred. — loss_hinge","text":"","code":"y_true <- array(sample(c(-1,1), 6, replace = TRUE), dim = c(2, 3)) y_pred <- random_uniform(c(2, 3)) loss <- loss_hinge(y_true, y_pred) loss ## tf.Tensor([1.0610152  0.93285507], shape=(2), dtype=float32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/loss_huber.html","id":null,"dir":"Reference","previous_headings":"","what":"Computes the Huber loss between y_true & y_pred. — loss_huber","title":"Computes the Huber loss between y_true & y_pred. — loss_huber","text":"Formula:   See: Huber loss.","code":"for (x in error) {   if (abs(x) <= delta){     loss <- c(loss, (0.5 * x^2))   } else if (abs(x) > delta) {     loss <- c(loss, (delta * abs(x) - 0.5 * delta^2))   } } loss <- mean(loss)"},{"path":"https://keras3.posit.co/dev/reference/loss_huber.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes the Huber loss between y_true & y_pred. — loss_huber","text":"","code":"loss_huber(   y_true,   y_pred,   delta = 1,   ...,   reduction = \"sum_over_batch_size\",   name = \"huber_loss\",   dtype = NULL )"},{"path":"https://keras3.posit.co/dev/reference/loss_huber.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Computes the Huber loss between y_true & y_pred. — loss_huber","text":"y_true tensor true targets. y_pred tensor predicted targets. delta float, point Huber loss function changes quadratic linear. Defaults 1.0. ... forward/backward compatability. reduction Type reduction apply loss. Options \"sum\", \"sum_over_batch_size\" NULL. Defaults \"sum_over_batch_size\". name Optional name instance. dtype dtype loss's computations. Defaults NULL, means using config_floatx(). config_floatx() \"float32\" unless set different value (via config_set_floatx()). keras$DTypePolicy provided, compute_dtype utilized.","code":""},{"path":"https://keras3.posit.co/dev/reference/loss_huber.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Computes the Huber loss between y_true & y_pred. — loss_huber","text":"Tensor one scalar loss entry per sample.","code":""},{"path":"https://keras3.posit.co/dev/reference/loss_huber.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Computes the Huber loss between y_true & y_pred. — loss_huber","text":"","code":"y_true <- rbind(c(0, 1), c(0, 0)) y_pred <- rbind(c(0.6, 0.4), c(0.4, 0.6)) loss <- loss_huber(y_true, y_pred)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/loss_kl_divergence.html","id":null,"dir":"Reference","previous_headings":"","what":"Computes Kullback-Leibler divergence loss between y_true & y_pred. — loss_kl_divergence","title":"Computes Kullback-Leibler divergence loss between y_true & y_pred. — loss_kl_divergence","text":"Formula:   y_true y_pred expected probability distributions, values 0 1. get clipped [0, 1] range.","code":"loss <- y_true * log(y_true / y_pred)"},{"path":"https://keras3.posit.co/dev/reference/loss_kl_divergence.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes Kullback-Leibler divergence loss between y_true & y_pred. — loss_kl_divergence","text":"","code":"loss_kl_divergence(   y_true,   y_pred,   ...,   reduction = \"sum_over_batch_size\",   name = \"kl_divergence\",   dtype = NULL )"},{"path":"https://keras3.posit.co/dev/reference/loss_kl_divergence.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Computes Kullback-Leibler divergence loss between y_true & y_pred. — loss_kl_divergence","text":"y_true Tensor true targets. y_pred Tensor predicted targets. ... forward/backward compatability. reduction Type reduction apply loss. almost cases \"sum_over_batch_size\". Supported options \"sum\", \"sum_over_batch_size\", \"mean\", \"mean_with_sample_weight\" NULL. \"sum\" sums loss, \"sum_over_batch_size\" \"mean\" sum loss divide sample size, \"mean_with_sample_weight\" sums loss divides sum sample weights. \"none\" NULL perform aggregation. Defaults \"sum_over_batch_size\". name Optional name loss instance. dtype dtype loss's computations. Defaults NULL, means using config_floatx(). config_floatx() \"float32\" unless set different value (via config_set_floatx()). keras$DTypePolicy provided, compute_dtype utilized.","code":""},{"path":"https://keras3.posit.co/dev/reference/loss_kl_divergence.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Computes Kullback-Leibler divergence loss between y_true & y_pred. — loss_kl_divergence","text":"KL Divergence loss values shape = [batch_size, d0, .. dN-1].","code":""},{"path":"https://keras3.posit.co/dev/reference/loss_kl_divergence.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Computes Kullback-Leibler divergence loss between y_true & y_pred. — loss_kl_divergence","text":"","code":"y_true <- random_uniform(c(2, 3), 0, 2) y_pred <- random_uniform(c(2,3)) loss <- loss_kl_divergence(y_true, y_pred) loss ## tf.Tensor([ 2.4290292 -0.6284211], shape=(2), dtype=float32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/loss_log_cosh.html","id":null,"dir":"Reference","previous_headings":"","what":"Computes the logarithm of the hyperbolic cosine of the prediction error. — loss_log_cosh","title":"Computes the logarithm of the hyperbolic cosine of the prediction error. — loss_log_cosh","text":"Formula:   Note log(cosh(x)) approximately equal (x ** 2) / 2 small x abs(x) - log(2) large x. means 'logcosh' works mostly like mean squared error, strongly affected occasional wildly incorrect prediction.","code":"loss <- mean(log(cosh(y_pred - y_true)), axis=-1)"},{"path":"https://keras3.posit.co/dev/reference/loss_log_cosh.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes the logarithm of the hyperbolic cosine of the prediction error. — loss_log_cosh","text":"","code":"loss_log_cosh(   y_true,   y_pred,   ...,   reduction = \"sum_over_batch_size\",   name = \"log_cosh\",   dtype = NULL )"},{"path":"https://keras3.posit.co/dev/reference/loss_log_cosh.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Computes the logarithm of the hyperbolic cosine of the prediction error. — loss_log_cosh","text":"y_true Ground truth values shape = [batch_size, d0, .. dN]. y_pred predicted values shape = [batch_size, d0, .. dN]. ... forward/backward compatability. reduction Type reduction apply loss. Options \"sum\", \"sum_over_batch_size\" NULL. Defaults \"sum_over_batch_size\". name Optional name instance. dtype dtype loss's computations. Defaults NULL, means using config_floatx(). config_floatx() \"float32\" unless set different value (via config_set_floatx()). keras$DTypePolicy provided, compute_dtype utilized.","code":""},{"path":"https://keras3.posit.co/dev/reference/loss_log_cosh.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Computes the logarithm of the hyperbolic cosine of the prediction error. — loss_log_cosh","text":"Logcosh error values shape = [batch_size, d0, .. dN-1].","code":""},{"path":"https://keras3.posit.co/dev/reference/loss_log_cosh.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Computes the logarithm of the hyperbolic cosine of the prediction error. — loss_log_cosh","text":"","code":"y_true <- rbind(c(0., 1.), c(0., 0.)) y_pred <- rbind(c(1., 1.), c(0., 0.)) loss <- loss_log_cosh(y_true, y_pred) # 0.108"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/loss_mean_absolute_error.html","id":null,"dir":"Reference","previous_headings":"","what":"Computes the mean of absolute difference between labels and predictions. — loss_mean_absolute_error","title":"Computes the mean of absolute difference between labels and predictions. — loss_mean_absolute_error","text":"Formula:","code":"loss <- mean(abs(y_true - y_pred))"},{"path":"https://keras3.posit.co/dev/reference/loss_mean_absolute_error.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes the mean of absolute difference between labels and predictions. — loss_mean_absolute_error","text":"","code":"loss_mean_absolute_error(   y_true,   y_pred,   ...,   reduction = \"sum_over_batch_size\",   name = \"mean_absolute_error\",   dtype = NULL )"},{"path":"https://keras3.posit.co/dev/reference/loss_mean_absolute_error.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Computes the mean of absolute difference between labels and predictions. — loss_mean_absolute_error","text":"y_true Ground truth values shape = [batch_size, d0, .. dN]. y_pred predicted values shape = [batch_size, d0, .. dN]. ... forward/backward compatability. reduction Type reduction apply loss. almost cases \"sum_over_batch_size\". Supported options \"sum\", \"sum_over_batch_size\", \"mean\", \"mean_with_sample_weight\" NULL. \"sum\" sums loss, \"sum_over_batch_size\" \"mean\" sum loss divide sample size, \"mean_with_sample_weight\" sums loss divides sum sample weights. \"none\" NULL perform aggregation. Defaults \"sum_over_batch_size\". name Optional name loss instance. dtype dtype loss's computations. Defaults NULL, means using config_floatx(). config_floatx() \"float32\" unless set different value (via config_set_floatx()). keras$DTypePolicy provided, compute_dtype utilized.","code":""},{"path":"https://keras3.posit.co/dev/reference/loss_mean_absolute_error.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Computes the mean of absolute difference between labels and predictions. — loss_mean_absolute_error","text":"Mean absolute error values shape = [batch_size, d0, .. dN-1].","code":""},{"path":"https://keras3.posit.co/dev/reference/loss_mean_absolute_error.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Computes the mean of absolute difference between labels and predictions. — loss_mean_absolute_error","text":"","code":"y_true <- random_uniform(c(2, 3), 0, 2) y_pred <- random_uniform(c(2, 3)) loss <- loss_mean_absolute_error(y_true, y_pred)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/loss_mean_absolute_percentage_error.html","id":null,"dir":"Reference","previous_headings":"","what":"Computes the mean absolute percentage error between y_true and y_pred. — loss_mean_absolute_percentage_error","title":"Computes the mean absolute percentage error between y_true and y_pred. — loss_mean_absolute_percentage_error","text":"Formula:   Division zero prevented dividing max(y_true, epsilon) epsilon = config_epsilon() (default 1e-7).","code":"loss <- 100 * op_mean(op_abs((y_true - y_pred) / y_true),                       axis=-1)"},{"path":"https://keras3.posit.co/dev/reference/loss_mean_absolute_percentage_error.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes the mean absolute percentage error between y_true and y_pred. — loss_mean_absolute_percentage_error","text":"","code":"loss_mean_absolute_percentage_error(   y_true,   y_pred,   ...,   reduction = \"sum_over_batch_size\",   name = \"mean_absolute_percentage_error\",   dtype = NULL )"},{"path":"https://keras3.posit.co/dev/reference/loss_mean_absolute_percentage_error.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Computes the mean absolute percentage error between y_true and y_pred. — loss_mean_absolute_percentage_error","text":"y_true Ground truth values shape = [batch_size, d0, .. dN]. y_pred predicted values shape = [batch_size, d0, .. dN]. ... forward/backward compatability. reduction Type reduction apply loss. almost cases \"sum_over_batch_size\". Supported options \"sum\", \"sum_over_batch_size\", \"mean\", \"mean_with_sample_weight\" NULL. \"sum\" sums loss, \"sum_over_batch_size\" \"mean\" sum loss divide sample size, \"mean_with_sample_weight\" sums loss divides sum sample weights. \"none\" NULL perform aggregation. Defaults \"sum_over_batch_size\". name Optional name loss instance. dtype dtype loss's computations. Defaults NULL, means using config_floatx(). config_floatx() \"float32\" unless set different value (via config_set_floatx()). keras$DTypePolicy provided, compute_dtype utilized.","code":""},{"path":"https://keras3.posit.co/dev/reference/loss_mean_absolute_percentage_error.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Computes the mean absolute percentage error between y_true and y_pred. — loss_mean_absolute_percentage_error","text":"Mean absolute percentage error values shape = [batch_size, d0, ..dN-1].","code":""},{"path":"https://keras3.posit.co/dev/reference/loss_mean_absolute_percentage_error.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Computes the mean absolute percentage error between y_true and y_pred. — loss_mean_absolute_percentage_error","text":"","code":"y_true <- random_uniform(c(2, 3)) y_pred <- random_uniform(c(2, 3)) loss <- loss_mean_absolute_percentage_error(y_true, y_pred)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/loss_mean_squared_error.html","id":null,"dir":"Reference","previous_headings":"","what":"Computes the mean of squares of errors between labels and predictions. — loss_mean_squared_error","title":"Computes the mean of squares of errors between labels and predictions. — loss_mean_squared_error","text":"Formula:","code":"loss <- mean(square(y_true - y_pred))"},{"path":"https://keras3.posit.co/dev/reference/loss_mean_squared_error.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes the mean of squares of errors between labels and predictions. — loss_mean_squared_error","text":"","code":"loss_mean_squared_error(   y_true,   y_pred,   ...,   reduction = \"sum_over_batch_size\",   name = \"mean_squared_error\",   dtype = NULL )"},{"path":"https://keras3.posit.co/dev/reference/loss_mean_squared_error.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Computes the mean of squares of errors between labels and predictions. — loss_mean_squared_error","text":"y_true Ground truth values shape = [batch_size, d0, .. dN]. y_pred predicted values shape = [batch_size, d0, .. dN]. ... forward/backward compatability. reduction Type reduction apply loss. almost cases \"sum_over_batch_size\". Supported options \"sum\", \"sum_over_batch_size\", \"mean\", \"mean_with_sample_weight\" NULL. \"sum\" sums loss, \"sum_over_batch_size\" \"mean\" sum loss divide sample size, \"mean_with_sample_weight\" sums loss divides sum sample weights. \"none\" NULL perform aggregation. Defaults \"sum_over_batch_size\". name Optional name loss instance. dtype dtype loss's computations. Defaults NULL, means using config_floatx(). config_floatx() \"float32\" unless set different value (via config_set_floatx()). keras$DTypePolicy provided, compute_dtype utilized.","code":""},{"path":"https://keras3.posit.co/dev/reference/loss_mean_squared_error.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Computes the mean of squares of errors between labels and predictions. — loss_mean_squared_error","text":"Mean squared error values shape = [batch_size, d0, .. dN-1].","code":""},{"path":"https://keras3.posit.co/dev/reference/loss_mean_squared_error.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Computes the mean of squares of errors between labels and predictions. — loss_mean_squared_error","text":"","code":"y_true <- random_uniform(c(2, 3), 0, 2) y_pred <- random_uniform(c(2, 3)) loss <- loss_mean_squared_error(y_true, y_pred)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/loss_mean_squared_logarithmic_error.html","id":null,"dir":"Reference","previous_headings":"","what":"Computes the mean squared logarithmic error between y_true and y_pred. — loss_mean_squared_logarithmic_error","title":"Computes the mean squared logarithmic error between y_true and y_pred. — loss_mean_squared_logarithmic_error","text":"Note y_pred y_true less equal 0. Negative values 0 values replaced config_epsilon() (default 1e-7). Formula:","code":"loss <- mean(square(log(y_true + 1) - log(y_pred + 1)))"},{"path":"https://keras3.posit.co/dev/reference/loss_mean_squared_logarithmic_error.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes the mean squared logarithmic error between y_true and y_pred. — loss_mean_squared_logarithmic_error","text":"","code":"loss_mean_squared_logarithmic_error(   y_true,   y_pred,   ...,   reduction = \"sum_over_batch_size\",   name = \"mean_squared_logarithmic_error\",   dtype = NULL )"},{"path":"https://keras3.posit.co/dev/reference/loss_mean_squared_logarithmic_error.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Computes the mean squared logarithmic error between y_true and y_pred. — loss_mean_squared_logarithmic_error","text":"y_true Ground truth values shape = [batch_size, d0, .. dN]. y_pred predicted values shape = [batch_size, d0, .. dN]. ... forward/backward compatability. reduction Type reduction apply loss. almost cases \"sum_over_batch_size\". Supported options \"sum\", \"sum_over_batch_size\", \"mean\", \"mean_with_sample_weight\" NULL. \"sum\" sums loss, \"sum_over_batch_size\" \"mean\" sum loss divide sample size, \"mean_with_sample_weight\" sums loss divides sum sample weights. \"none\" NULL perform aggregation. Defaults \"sum_over_batch_size\". name Optional name loss instance. dtype dtype loss's computations. Defaults NULL, means using config_floatx(). config_floatx() \"float32\" unless set different value (via config_set_floatx()). keras$DTypePolicy provided, compute_dtype utilized.","code":""},{"path":"https://keras3.posit.co/dev/reference/loss_mean_squared_logarithmic_error.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Computes the mean squared logarithmic error between y_true and y_pred. — loss_mean_squared_logarithmic_error","text":"Mean squared logarithmic error values shape = [batch_size, d0, .. dN-1].","code":""},{"path":"https://keras3.posit.co/dev/reference/loss_mean_squared_logarithmic_error.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Computes the mean squared logarithmic error between y_true and y_pred. — loss_mean_squared_logarithmic_error","text":"","code":"y_true <- random_uniform(c(2, 3), 0, 2) y_pred <- random_uniform(c(2, 3)) loss <- loss_mean_squared_logarithmic_error(y_true, y_pred)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/loss_poisson.html","id":null,"dir":"Reference","previous_headings":"","what":"Computes the Poisson loss between y_true & y_pred. — loss_poisson","title":"Computes the Poisson loss between y_true & y_pred. — loss_poisson","text":"Formula:","code":"loss <- y_pred - y_true * log(y_pred)"},{"path":"https://keras3.posit.co/dev/reference/loss_poisson.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes the Poisson loss between y_true & y_pred. — loss_poisson","text":"","code":"loss_poisson(   y_true,   y_pred,   ...,   reduction = \"sum_over_batch_size\",   name = \"poisson\",   dtype = NULL )"},{"path":"https://keras3.posit.co/dev/reference/loss_poisson.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Computes the Poisson loss between y_true & y_pred. — loss_poisson","text":"y_true Ground truth values. shape = [batch_size, d0, .. dN]. y_pred predicted values. shape = [batch_size, d0, .. dN]. ... forward/backward compatability. reduction Type reduction apply loss. almost cases \"sum_over_batch_size\". Supported options \"sum\", \"sum_over_batch_size\", \"mean\", \"mean_with_sample_weight\" NULL. \"sum\" sums loss, \"sum_over_batch_size\" \"mean\" sum loss divide sample size, \"mean_with_sample_weight\" sums loss divides sum sample weights. \"none\" NULL perform aggregation. Defaults \"sum_over_batch_size\". name Optional name loss instance. dtype dtype loss's computations. Defaults NULL, means using config_floatx(). config_floatx() \"float32\" unless set different value (via config_set_floatx()). keras$DTypePolicy provided, compute_dtype utilized.","code":""},{"path":"https://keras3.posit.co/dev/reference/loss_poisson.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Computes the Poisson loss between y_true & y_pred. — loss_poisson","text":"Poisson loss values shape = [batch_size, d0, .. dN-1].","code":""},{"path":"https://keras3.posit.co/dev/reference/loss_poisson.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Computes the Poisson loss between y_true & y_pred. — loss_poisson","text":"","code":"y_true <- random_uniform(c(2, 3), 0, 2) y_pred <- random_uniform(c(2, 3)) loss <- loss_poisson(y_true, y_pred) loss ## tf.Tensor([1.6422468  0.81166863], shape=(2), dtype=float32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/loss_sparse_categorical_crossentropy.html","id":null,"dir":"Reference","previous_headings":"","what":"Computes the crossentropy loss between the labels and predictions. — loss_sparse_categorical_crossentropy","title":"Computes the crossentropy loss between the labels and predictions. — loss_sparse_categorical_crossentropy","text":"Use crossentropy loss function two label classes.  expect labels provided integers. want provide labels using one-hot representation, please use CategoricalCrossentropy loss.  # classes floating point values per feature y_pred single floating point value per feature y_true. snippet , single floating point value per example y_true num_classes floating pointing values per example y_pred. shape y_true [batch_size] shape y_pred [batch_size, num_classes].","code":""},{"path":"https://keras3.posit.co/dev/reference/loss_sparse_categorical_crossentropy.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes the crossentropy loss between the labels and predictions. — loss_sparse_categorical_crossentropy","text":"","code":"loss_sparse_categorical_crossentropy(   y_true,   y_pred,   from_logits = FALSE,   ignore_class = NULL,   axis = -1L,   ...,   reduction = \"sum_over_batch_size\",   name = \"sparse_categorical_crossentropy\",   dtype = NULL )"},{"path":"https://keras3.posit.co/dev/reference/loss_sparse_categorical_crossentropy.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Computes the crossentropy loss between the labels and predictions. — loss_sparse_categorical_crossentropy","text":"y_true Ground truth values. y_pred predicted values. from_logits Whether y_pred expected logits tensor. default, assume y_pred encodes probability distribution. ignore_class Optional integer. ID class ignored loss computation. useful, example, segmentation problems featuring \"void\" class (commonly -1 255) segmentation maps. default (ignore_class=NULL), classes considered. axis Defaults -1. dimension along entropy computed. ... forward/backward compatability. reduction Type reduction apply loss. almost cases \"sum_over_batch_size\". Supported options \"sum\", \"sum_over_batch_size\", \"mean\", \"mean_with_sample_weight\" NULL. \"sum\" sums loss, \"sum_over_batch_size\" \"mean\" sum loss divide sample size, \"mean_with_sample_weight\" sums loss divides sum sample weights. \"none\" NULL perform aggregation. Defaults \"sum_over_batch_size\". name Optional name loss instance. dtype dtype loss's computations. Defaults NULL, means using config_floatx(). config_floatx() \"float32\" unless set different value (via config_set_floatx()). keras$DTypePolicy provided, compute_dtype utilized.","code":""},{"path":"https://keras3.posit.co/dev/reference/loss_sparse_categorical_crossentropy.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Computes the crossentropy loss between the labels and predictions. — loss_sparse_categorical_crossentropy","text":"Sparse categorical crossentropy loss value.","code":""},{"path":"https://keras3.posit.co/dev/reference/loss_sparse_categorical_crossentropy.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Computes the crossentropy loss between the labels and predictions. — loss_sparse_categorical_crossentropy","text":"Usage compile() API:","code":"y_true <- op_array(c(1L, 2L)) y_pred <- op_array(rbind(c(0.05, 0.95, 0), c(0.1, 0.8, 0.1))) loss <- loss_sparse_categorical_crossentropy(y_true, y_pred) loss ## tf.Tensor([0.05129339 2.30258509], shape=(2), dtype=float64) y_true <- op_array(c(1L, 2L)) y_pred <- op_array(rbind(c(0.05, 0.95, 0), c(0.1, 0.8, 0.1))) # Using 'auto'/'sum_over_batch_size' reduction type. scce <- loss_sparse_categorical_crossentropy() scce(y_true, y_pred) ## tf.Tensor(1.1769392, shape=(), dtype=float32) # Calling with 'sample_weight'. scce(y_true, y_pred, sample_weight = op_array(c(0.3, 0.7))) ## tf.Tensor(0.8135988, shape=(), dtype=float32) # Using 'sum' reduction type. scce <- loss_sparse_categorical_crossentropy(reduction=\"sum\") scce(y_true, y_pred) ## tf.Tensor(2.3538785, shape=(), dtype=float32) # Using 'none' reduction type. scce <- loss_sparse_categorical_crossentropy(reduction=NULL) scce(y_true, y_pred) ## tf.Tensor([0.05129344 2.3025851 ], shape=(2), dtype=float32) model %>% compile(optimizer = 'sgd',                   loss = loss_sparse_categorical_crossentropy())"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/loss_squared_hinge.html","id":null,"dir":"Reference","previous_headings":"","what":"Computes the squared hinge loss between y_true & y_pred. — loss_squared_hinge","title":"Computes the squared hinge loss between y_true & y_pred. — loss_squared_hinge","text":"Formula:   y_true values expected -1 1. binary (0 1) labels provided convert -1 1.","code":"loss <- square(maximum(1 - y_true * y_pred, 0))"},{"path":"https://keras3.posit.co/dev/reference/loss_squared_hinge.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes the squared hinge loss between y_true & y_pred. — loss_squared_hinge","text":"","code":"loss_squared_hinge(   y_true,   y_pred,   ...,   reduction = \"sum_over_batch_size\",   name = \"squared_hinge\",   dtype = NULL )"},{"path":"https://keras3.posit.co/dev/reference/loss_squared_hinge.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Computes the squared hinge loss between y_true & y_pred. — loss_squared_hinge","text":"y_true ground truth values. y_true values expected -1 1. binary (0 1) labels provided convert -1 1 shape = [batch_size, d0, .. dN]. y_pred predicted values shape = [batch_size, d0, .. dN]. ... forward/backward compatability. reduction Type reduction apply loss. almost cases \"sum_over_batch_size\". Supported options \"sum\", \"sum_over_batch_size\", \"mean\", \"mean_with_sample_weight\" NULL. \"sum\" sums loss, \"sum_over_batch_size\" \"mean\" sum loss divide sample size, \"mean_with_sample_weight\" sums loss divides sum sample weights. \"none\" NULL perform aggregation. Defaults \"sum_over_batch_size\". name Optional name loss instance. dtype dtype loss's computations. Defaults NULL, means using config_floatx(). config_floatx() \"float32\" unless set different value (via config_set_floatx()). keras$DTypePolicy provided, compute_dtype utilized.","code":""},{"path":"https://keras3.posit.co/dev/reference/loss_squared_hinge.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Computes the squared hinge loss between y_true & y_pred. — loss_squared_hinge","text":"Squared hinge loss values shape = [batch_size, d0, .. dN-1].","code":""},{"path":"https://keras3.posit.co/dev/reference/loss_squared_hinge.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Computes the squared hinge loss between y_true & y_pred. — loss_squared_hinge","text":"","code":"y_true <- array(sample(c(-1,1), 6, replace = TRUE), dim = c(2, 3)) y_pred <- random_uniform(c(2, 3)) loss <- loss_squared_hinge(y_true, y_pred)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/loss_tversky.html","id":null,"dir":"Reference","previous_headings":"","what":"Computes the Tversky loss value between y_true and y_pred. — loss_tversky","title":"Computes the Tversky loss value between y_true and y_pred. — loss_tversky","text":"loss function weighted alpha beta coefficients penalize false positives false negatives. alpha=0.5 beta=0.5, loss value becomes equivalent Dice Loss. loss function weighted alpha beta coefficients penalize false positives false negatives. alpha=0.5 beta=0.5, loss value becomes equivalent Dice Loss.","code":""},{"path":"https://keras3.posit.co/dev/reference/loss_tversky.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes the Tversky loss value between y_true and y_pred. — loss_tversky","text":"","code":"loss_tversky(   y_true,   y_pred,   ...,   alpha = 0.5,   beta = 0.5,   reduction = \"sum_over_batch_size\",   name = \"tversky\",   axis = NULL,   dtype = NULL )"},{"path":"https://keras3.posit.co/dev/reference/loss_tversky.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Computes the Tversky loss value between y_true and y_pred. — loss_tversky","text":"y_true tensor true targets. y_pred tensor predicted targets. ... forward/backward compatability. alpha coefficient controlling incidence false positives. Defaults 0.5. beta coefficient controlling incidence false negatives. Defaults 0.5. reduction Type reduction apply loss. almost cases \"sum_over_batch_size\". Supported options \"sum\", \"sum_over_batch_size\", \"mean\", \"mean_with_sample_weight\" NULL. \"sum\" sums loss, \"sum_over_batch_size\" \"mean\" sum loss divide sample size, \"mean_with_sample_weight\" sums loss divides sum sample weights. \"none\" NULL perform aggregation. Defaults \"sum_over_batch_size\". name Optional name loss instance. (string) axis Axis (1-based) dtype dtype loss's computations. Defaults NULL, means using config_floatx(). config_floatx() \"float32\" unless set different value (via config_set_floatx()). keras$DTypePolicy provided, compute_dtype utilized.","code":""},{"path":"https://keras3.posit.co/dev/reference/loss_tversky.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Computes the Tversky loss value between y_true and y_pred. — loss_tversky","text":"Tversky loss value.","code":""},{"path":"https://keras3.posit.co/dev/reference/loss_tversky.html","id":"reference","dir":"Reference","previous_headings":"","what":"Reference","title":"Computes the Tversky loss value between y_true and y_pred. — loss_tversky","text":"Salehi et al., 2017","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/mark_active.html","id":null,"dir":"Reference","previous_headings":"","what":"active_property — mark_active","title":"active_property — mark_active","text":"mark_active() alias active_property(). See ?active_property() full documentation.","code":""},{"path":"https://keras3.posit.co/dev/reference/mark_active.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"active_property — mark_active","text":"","code":"mark_active(fn)"},{"path":"https://keras3.posit.co/dev/reference/mark_active.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"active_property — mark_active","text":"fn R function","code":""},{"path":"https://keras3.posit.co/dev/reference/mark_active.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"active_property — mark_active","text":"fn, additional R attribute cause fn converted active property converted method custom subclass.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_auc.html","id":null,"dir":"Reference","previous_headings":"","what":"Approximates the AUC (Area under the curve) of the ROC or PR curves. — metric_auc","title":"Approximates the AUC (Area under the curve) of the ROC or PR curves. — metric_auc","text":"AUC (Area curve) ROC (Receiver operating characteristic; default) PR (Precision Recall) curves quality measures binary classifiers. Unlike accuracy, like cross-entropy losses, ROC-AUC PR-AUC evaluate operational points model. class approximates AUCs using Riemann sum. metric accumulation phrase, predictions accumulated within predefined buckets value. AUC computed interpolating per-bucket averages. buckets define evaluated operational points. metric creates four local variables, true_positives, true_negatives, false_positives false_negatives used compute AUC.  discretize AUC curve, linearly spaced set thresholds used compute pairs recall precision values. area ROC-curve therefore computed using height recall values false positive rate, area PR-curve computed using height precision values recall. value ultimately returned auc, idempotent operation computes area discretized curve precision versus recall values (computed using aforementioned variables). num_thresholds variable controls degree discretization larger numbers thresholds closely approximating true AUC. quality approximation may vary dramatically depending num_thresholds. thresholds parameter can used manually specify thresholds split predictions evenly. best approximation real AUC, predictions distributed approximately uniformly range [0, 1] (from_logits=FALSE). quality AUC approximation may poor case. Setting summation_method 'minoring' 'majoring' can help quantify error approximation providing lower upper bound estimate AUC. sample_weight NULL, weights default 1. Use sample_weight 0 mask values.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_auc.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Approximates the AUC (Area under the curve) of the ROC or PR curves. — metric_auc","text":"","code":"metric_auc(   ...,   num_thresholds = 200L,   curve = \"ROC\",   summation_method = \"interpolation\",   name = NULL,   dtype = NULL,   thresholds = NULL,   multi_label = FALSE,   num_labels = NULL,   label_weights = NULL,   from_logits = FALSE )"},{"path":"https://keras3.posit.co/dev/reference/metric_auc.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Approximates the AUC (Area under the curve) of the ROC or PR curves. — metric_auc","text":"... forward/backward compatability. num_thresholds (Optional) number thresholds use discretizing roc curve. Values must > 1. Defaults 200. curve (Optional) Specifies name curve computed, 'ROC' (default) 'PR' Precision-Recall-curve. summation_method (Optional) Specifies Riemann summation method used. 'interpolation' (default) applies mid-point summation scheme ROC.  PR-AUC, interpolates (true/false) positives ratio precision (see Davis & Goadrich 2006 details); 'minoring' applies left summation increasing intervals right summation decreasing intervals; 'majoring' opposite. name (Optional) string name metric instance. dtype (Optional) data type metric result. thresholds (Optional) list floating point values use thresholds discretizing curve. set, num_thresholds parameter ignored. Values [0, 1]. Endpoint thresholds equal {-epsilon, 1+epsilon} small positive epsilon value automatically included correctly handle predictions equal exactly 0 1. multi_label boolean indicating whether multilabel data treated , wherein AUC computed separately label averaged across labels, (FALSE) data flattened single label AUC computation. latter case, multilabel data passed AUC, label-prediction pair treated individual data point. set `FALSE“ multi-class data. num_labels (Optional) number labels, used multi_label TRUE. num_labels specified, state variables get created first call update_state. label_weights (Optional) list, array, tensor non-negative weights used compute AUCs multilabel data. multi_label TRUE, weights applied individual label AUCs averaged produce multi-label AUC. FALSE, used weight individual label predictions computing confusion matrix flattened data. Note unlike class_weights class_weights weights example depending value label, whereas label_weights depends index label flattening; therefore label_weights used multi-class data. from_logits boolean indicating whether predictions (y_pred update_state) probabilities sigmoid logits. rule thumb, using keras loss, from_logits constructor argument loss match AUC from_logits constructor argument.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_auc.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Approximates the AUC (Area under the curve) of the ROC or PR curves. — metric_auc","text":"Metric instance returned. Metric instance can passed directly compile(metrics = ), used standalone object. See ?Metric example usage.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_auc.html","id":"usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Approximates the AUC (Area under the curve) of the ROC or PR curves. — metric_auc","text":"Standalone usage:         Usage compile() API:","code":"m <- metric_auc(num_thresholds = 3) m$update_state(c(0,   0,   1,   1),                c(0, 0.5, 0.3, 0.9)) # threshold values are [0 - 1e-7, 0.5, 1 + 1e-7] # tp = [2, 1, 0], fp = [2, 0, 0], fn = [0, 1, 2], tn = [0, 2, 2] # tp_rate = recall = [1, 0.5, 0], fp_rate = [1, 0, 0] # auc = ((((1 + 0.5) / 2) * (1 - 0)) + (((0.5 + 0) / 2) * (0 - 0))) #     = 0.75 m$result() ## tf.Tensor(0.75, shape=(), dtype=float32) m$reset_state() m$update_state(c(0,   0,   1,   1),                c(0, 0.5, 0.3, 0.9),                sample_weight=c(1, 0, 0, 1)) m$result() ## tf.Tensor(1.0, shape=(), dtype=float32) # Reports the AUC of a model outputting a probability. model |> compile(   optimizer = 'sgd',   loss = loss_binary_crossentropy(),   metrics = list(metric_auc()) )  # Reports the AUC of a model outputting a logit. model |> compile(   optimizer = 'sgd',   loss = loss_binary_crossentropy(from_logits = TRUE),   metrics = list(metric_auc(from_logits = TRUE)) )"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/metric_binary_accuracy.html","id":null,"dir":"Reference","previous_headings":"","what":"Calculates how often predictions match binary labels. — metric_binary_accuracy","title":"Calculates how often predictions match binary labels. — metric_binary_accuracy","text":"metric creates two local variables, total count used compute frequency y_pred matches y_true. frequency ultimately returned binary accuracy: idempotent operation simply divides total count. sample_weight NULL, weights default 1. Use sample_weight 0 mask values.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_binary_accuracy.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Calculates how often predictions match binary labels. — metric_binary_accuracy","text":"","code":"metric_binary_accuracy(   y_true,   y_pred,   threshold = 0.5,   ...,   name = \"binary_accuracy\",   dtype = NULL )"},{"path":"https://keras3.posit.co/dev/reference/metric_binary_accuracy.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Calculates how often predictions match binary labels. — metric_binary_accuracy","text":"y_true Tensor true targets. y_pred Tensor predicted targets. threshold (Optional) Float representing threshold deciding whether prediction values 1 0. ... forward/backward compatability. name (Optional) string name metric instance. dtype (Optional) data type metric result.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_binary_accuracy.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Calculates how often predictions match binary labels. — metric_binary_accuracy","text":"y_true y_pred missing, Metric instance returned. Metric instance can passed directly compile(metrics = ), used standalone object. See ?Metric example usage. y_true y_pred provided, tensor computed value returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_binary_accuracy.html","id":"usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Calculates how often predictions match binary labels. — metric_binary_accuracy","text":"Standalone usage:             Usage compile() API:","code":"m <- metric_binary_accuracy() m$update_state(rbind(1, 1, 0, 0), rbind(0.98, 1, 0, 0.6)) m$result() ## tf.Tensor(0.75, shape=(), dtype=float32) # 0.75 m$reset_state() m$update_state(rbind(1, 1, 0, 0), rbind(0.98, 1, 0, 0.6),                sample_weight = c(1, 0, 0, 1)) m$result() ## tf.Tensor(0.5, shape=(), dtype=float32) # 0.5 model %>% compile(optimizer='sgd',                   loss='binary_crossentropy',                   metrics=list(metric_binary_accuracy()))"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/metric_binary_crossentropy.html","id":null,"dir":"Reference","previous_headings":"","what":"Computes the crossentropy metric between the labels and predictions. — metric_binary_crossentropy","title":"Computes the crossentropy metric between the labels and predictions. — metric_binary_crossentropy","text":"crossentropy metric class used two label classes (0 1).","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_binary_crossentropy.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes the crossentropy metric between the labels and predictions. — metric_binary_crossentropy","text":"","code":"metric_binary_crossentropy(   y_true,   y_pred,   from_logits = FALSE,   label_smoothing = 0,   axis = -1L,   ...,   name = \"binary_crossentropy\",   dtype = NULL )"},{"path":"https://keras3.posit.co/dev/reference/metric_binary_crossentropy.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Computes the crossentropy metric between the labels and predictions. — metric_binary_crossentropy","text":"y_true Ground truth values. shape = [batch_size, d0, .. dN]. y_pred predicted values. shape = [batch_size, d0, .. dN]. from_logits (Optional) Whether output expected logits tensor. default, consider output encodes probability distribution. label_smoothing (Optional) Float [0, 1]. > 0, label values smoothed, meaning confidence label values relaxed. e.g. label_smoothing=0.2 means use value 0.1 label \"0\" 0.9 label \"1\". axis axis along mean computed. Defaults -1. ... forward/backward compatability. name (Optional) string name metric instance. dtype (Optional) data type metric result.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_binary_crossentropy.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Computes the crossentropy metric between the labels and predictions. — metric_binary_crossentropy","text":"y_true y_pred missing, Metric instance returned. Metric instance can passed directly compile(metrics = ), used standalone object. See ?Metric example usage. y_true y_pred provided, tensor computed value returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_binary_crossentropy.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Computes the crossentropy metric between the labels and predictions. — metric_binary_crossentropy","text":"Standalone usage:         Usage compile() API:","code":"m <- metric_binary_crossentropy() m$update_state(rbind(c(0, 1), c(0, 0)), rbind(c(0.6, 0.4), c(0.4, 0.6))) m$result() ## tf.Tensor(0.8149245, shape=(), dtype=float32) m$reset_state() m$update_state(rbind(c(0, 1), c(0, 0)), rbind(c(0.6, 0.4), c(0.4, 0.6)),                sample_weight=c(1, 0)) m$result() ## tf.Tensor(0.91629076, shape=(), dtype=float32) model %>% compile(     optimizer = 'sgd',     loss = 'mse',     metrics = list(metric_binary_crossentropy()))"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/metric_binary_focal_crossentropy.html","id":null,"dir":"Reference","previous_headings":"","what":"Computes the binary focal crossentropy loss. — metric_binary_focal_crossentropy","title":"Computes the binary focal crossentropy loss. — metric_binary_focal_crossentropy","text":"According Lin et al., 2018, helps apply focal factor -weight easy examples focus hard examples. default, focal tensor computed follows: focal_factor = (1 - output)^gamma class 1 focal_factor = output^gamma class 0 gamma focusing parameter. gamma = 0, focal effect binary crossentropy loss. apply_class_balancing == TRUE, function also takes account weight balancing factor binary classes 0 1 follows: weight = alpha class 1 (target == 1) weight = 1 - alpha class 0 alpha float range [0, 1].","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_binary_focal_crossentropy.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes the binary focal crossentropy loss. — metric_binary_focal_crossentropy","text":"","code":"metric_binary_focal_crossentropy(   y_true,   y_pred,   apply_class_balancing = FALSE,   alpha = 0.25,   gamma = 2,   from_logits = FALSE,   label_smoothing = 0,   axis = -1L )"},{"path":"https://keras3.posit.co/dev/reference/metric_binary_focal_crossentropy.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Computes the binary focal crossentropy loss. — metric_binary_focal_crossentropy","text":"y_true Ground truth values, shape (batch_size, d0, .. dN). y_pred predicted values, shape (batch_size, d0, .. dN). apply_class_balancing bool, whether apply weight balancing binary classes 0 1. alpha weight balancing factor class 1, default 0.25 mentioned reference. weight class 0 1.0 - alpha. gamma focusing parameter, default 2.0 mentioned reference. from_logits Whether y_pred expected logits tensor. default, assume y_pred encodes probability distribution. label_smoothing Float [0, 1]. > 0 smooth labels squeezing towards 0.5, , using 1. - 0.5 * label_smoothing target class 0.5 * label_smoothing non-target class. axis axis along mean computed. Defaults -1.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_binary_focal_crossentropy.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Computes the binary focal crossentropy loss. — metric_binary_focal_crossentropy","text":"Binary focal crossentropy loss value shape = [batch_size, d0, .. dN-1].","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_binary_focal_crossentropy.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Computes the binary focal crossentropy loss. — metric_binary_focal_crossentropy","text":"","code":"y_true <- rbind(c(0, 1), c(0, 0)) y_pred <- rbind(c(0.6, 0.4), c(0.4, 0.6)) focal_loss <- loss_binary_focal_crossentropy(y_true, y_pred, gamma = 2) focal_loss ## tf.Tensor([0.32986468 0.20579839], shape=(2), dtype=float32) # Compare with binary crossentropy. # Binary focal crossentropy emphasizes harder examples, yielding a larger # relative loss where the model struggles. bce_loss <- loss_binary_crossentropy(y_true, y_pred) cbind(focal_loss = as.array(focal_loss),       bce_loss = as.array(bce_loss),       ratio = as.array(focal_loss / bce_loss)) ##      focal_loss  bce_loss     ratio ## [1,]  0.3298647 0.9162908 0.3600000 ## [2,]  0.2057984 0.7135582 0.2884115"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/metric_binary_iou.html","id":null,"dir":"Reference","previous_headings":"","what":"Computes the Intersection-Over-Union metric for class 0 and/or 1. — metric_binary_iou","title":"Computes the Intersection-Over-Union metric for class 0 and/or 1. — metric_binary_iou","text":"Formula:   Intersection--Union common evaluation metric semantic image segmentation. compute IoUs, predictions accumulated confusion matrix, weighted sample_weight metric calculated . sample_weight NULL, weights default 1. Use sample_weight 0 mask values. class can used compute IoUs binary classification task predictions provided logits. First threshold applied predicted values threshold converted class 0 threshold converted class 1. IoUs classes 0 1 computed, mean IoUs classes specified target_class_ids returned.","code":"iou <- true_positives / (true_positives + false_positives + false_negatives)"},{"path":"https://keras3.posit.co/dev/reference/metric_binary_iou.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes the Intersection-Over-Union metric for class 0 and/or 1. — metric_binary_iou","text":"","code":"metric_binary_iou(   ...,   target_class_ids = list(0L, 1L),   threshold = 0.5,   name = NULL,   dtype = NULL )"},{"path":"https://keras3.posit.co/dev/reference/metric_binary_iou.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Computes the Intersection-Over-Union metric for class 0 and/or 1. — metric_binary_iou","text":"... forward/backward compatability. target_class_ids list list target class ids metric returned. Options 0, 1, c(0, 1). 0 (1), IoU metric class 0 (class 1, respectively) returned. c(0, 1), mean IoUs two classes returned. threshold threshold applies prediction logits convert either predicted class 0 logit threshold predicted class 1 logit threshold. name (Optional) string name metric instance. dtype (Optional) data type metric result.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_binary_iou.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Computes the Intersection-Over-Union metric for class 0 and/or 1. — metric_binary_iou","text":"Metric instance returned. Metric instance can passed directly compile(metrics = ), used standalone object. See ?Metric example usage.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_binary_iou.html","id":"note","dir":"Reference","previous_headings":"","what":"Note","title":"Computes the Intersection-Over-Union metric for class 0 and/or 1. — metric_binary_iou","text":"threshold=0, metric behavior IoU.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_binary_iou.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Computes the Intersection-Over-Union metric for class 0 and/or 1. — metric_binary_iou","text":"Standalone usage:                 Usage compile() API:","code":"m <- metric_binary_iou(target_class_ids=c(0L, 1L), threshold = 0.3) m$update_state(c(0, 1, 0, 1), c(0.1, 0.2, 0.4, 0.7)) ## tf.Tensor( ## [[1 1] ##  [1 1]], shape=(2, 2), dtype=int64) m$result() ## tf.Tensor(0.33333334, shape=(), dtype=float32) m$reset_state() m$update_state(c(0, 1, 0, 1), c(0.1, 0.2, 0.4, 0.7),                sample_weight = 10 * c(0.2, 0.3, 0.4, 0.1)) ## tf.Tensor( ## [[2 4] ##  [3 1]], shape=(2, 2), dtype=int64) m$result() ## tf.Tensor(0.1736111, shape=(), dtype=float32) model %>% compile(     optimizer = 'sgd',     loss = 'mse',     metrics = list(metric_binary_iou(         target_class_ids = 0L,         threshold = 0.5     )) )"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/metric_categorical_accuracy.html","id":null,"dir":"Reference","previous_headings":"","what":"Calculates how often predictions match one-hot labels. — metric_categorical_accuracy","title":"Calculates how often predictions match one-hot labels. — metric_categorical_accuracy","text":"can provide logits classes y_pred, since argmax logits probabilities . metric creates two local variables, total count used compute frequency y_pred matches y_true. frequency ultimately returned categorical accuracy: idempotent operation simply divides total count. y_pred y_true passed vectors probabilities, rather labels. necessary, use op_one_hot expand y_true vector. sample_weight NULL, weights default 1. Use sample_weight 0 mask values.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_categorical_accuracy.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Calculates how often predictions match one-hot labels. — metric_categorical_accuracy","text":"","code":"metric_categorical_accuracy(   y_true,   y_pred,   ...,   name = \"categorical_accuracy\",   dtype = NULL )"},{"path":"https://keras3.posit.co/dev/reference/metric_categorical_accuracy.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Calculates how often predictions match one-hot labels. — metric_categorical_accuracy","text":"y_true Tensor true targets. y_pred Tensor predicted targets. ... forward/backward compatability. name (Optional) string name metric instance. dtype (Optional) data type metric result.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_categorical_accuracy.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Calculates how often predictions match one-hot labels. — metric_categorical_accuracy","text":"y_true y_pred missing, Metric instance returned. Metric instance can passed directly compile(metrics = ), used standalone object. See ?Metric example usage. y_true y_pred provided, tensor computed value returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_categorical_accuracy.html","id":"usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Calculates how often predictions match one-hot labels. — metric_categorical_accuracy","text":"Standalone usage:           Usage compile() API:","code":"m <- metric_categorical_accuracy() m$update_state(rbind(c(0, 0, 1), c(0, 1, 0)), rbind(c(0.1, 0.9, 0.8),                 c(0.05, 0.95, 0))) m$result() ## tf.Tensor(0.5, shape=(), dtype=float32) m$reset_state() m$update_state(rbind(c(0, 0, 1), c(0, 1, 0)), rbind(c(0.1, 0.9, 0.8),                c(0.05, 0.95, 0)),                sample_weight = c(0.7, 0.3)) m$result() ## tf.Tensor(0.3, shape=(), dtype=float32) # 0.3 model %>% compile(optimizer = 'sgd',                   loss = 'categorical_crossentropy',                   metrics = list(metric_categorical_accuracy()))"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/metric_categorical_crossentropy.html","id":null,"dir":"Reference","previous_headings":"","what":"Computes the crossentropy metric between the labels and predictions. — metric_categorical_crossentropy","title":"Computes the crossentropy metric between the labels and predictions. — metric_categorical_crossentropy","text":"crossentropy metric class used multiple label classes (2 ). assumes labels one-hot encoded, e.g., labels values c(2, 0, 1), y_true rbind(c([0, 0, 1), c(1, 0, 0), c(0, 1, 0)).","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_categorical_crossentropy.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes the crossentropy metric between the labels and predictions. — metric_categorical_crossentropy","text":"","code":"metric_categorical_crossentropy(   y_true,   y_pred,   from_logits = FALSE,   label_smoothing = 0,   axis = -1L,   ...,   name = \"categorical_crossentropy\",   dtype = NULL )"},{"path":"https://keras3.posit.co/dev/reference/metric_categorical_crossentropy.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Computes the crossentropy metric between the labels and predictions. — metric_categorical_crossentropy","text":"y_true Tensor one-hot true targets. y_pred Tensor predicted targets. from_logits (Optional) Whether output expected logits tensor. default, consider output encodes probability distribution. label_smoothing (Optional) Float [0, 1]. > 0, label values smoothed, meaning confidence label values relaxed. e.g. label_smoothing=0.2 means use value 0.1 label \"0\" 0.9 label \"1\". axis (Optional) Defaults -1. dimension along entropy computed. ... forward/backward compatability. name (Optional) string name metric instance. dtype (Optional) data type metric result.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_categorical_crossentropy.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Computes the crossentropy metric between the labels and predictions. — metric_categorical_crossentropy","text":"y_true y_pred missing, Metric instance returned. Metric instance can passed directly compile(metrics = ), used standalone object. See ?Metric example usage. y_true y_pred provided, tensor computed value returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_categorical_crossentropy.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Computes the crossentropy metric between the labels and predictions. — metric_categorical_crossentropy","text":"Standalone usage:           Usage compile() API:","code":"# EPSILON = 1e-7, y = y_true, y` = y_pred # y` = clip_op_clip_by_value(output, EPSILON, 1. - EPSILON) # y` = rbind(c(0.05, 0.95, EPSILON), c(0.1, 0.8, 0.1)) # xent = -sum(y * log(y'), axis = -1) #      = -((log 0.95), (log 0.1)) #      = [0.051, 2.302] # Reduced xent = (0.051 + 2.302) / 2  m <- metric_categorical_crossentropy() m$update_state(rbind(c(0, 1, 0), c(0, 0, 1)),                rbind(c(0.05, 0.95, 0), c(0.1, 0.8, 0.1))) m$result() ## tf.Tensor(1.1769392, shape=(), dtype=float32) # 1.1769392 m$reset_state() m$update_state(rbind(c(0, 1, 0), c(0, 0, 1)),                rbind(c(0.05, 0.95, 0), c(0.1, 0.8, 0.1)),                sample_weight = c(0.3, 0.7)) m$result() ## tf.Tensor(1.6271976, shape=(), dtype=float32) model %>% compile(   optimizer = 'sgd',   loss = 'mse',   metrics = list(metric_categorical_crossentropy()))"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/metric_categorical_focal_crossentropy.html","id":null,"dir":"Reference","previous_headings":"","what":"Computes the categorical focal crossentropy loss. — metric_categorical_focal_crossentropy","title":"Computes the categorical focal crossentropy loss. — metric_categorical_focal_crossentropy","text":"Computes categorical focal crossentropy loss.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_categorical_focal_crossentropy.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes the categorical focal crossentropy loss. — metric_categorical_focal_crossentropy","text":"","code":"metric_categorical_focal_crossentropy(   y_true,   y_pred,   alpha = 0.25,   gamma = 2,   from_logits = FALSE,   label_smoothing = 0,   axis = -1L )"},{"path":"https://keras3.posit.co/dev/reference/metric_categorical_focal_crossentropy.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Computes the categorical focal crossentropy loss. — metric_categorical_focal_crossentropy","text":"y_true Tensor one-hot true targets. y_pred Tensor predicted targets. alpha weight balancing factor classes, default 0.25 mentioned reference. can list floats scalar. multi-class case, alpha may set inverse class frequency using compute_class_weight sklearn.utils. gamma focusing parameter, default 2.0 mentioned reference. helps gradually reduce importance given simple examples smooth manner. gamma = 0, focal effect categorical crossentropy. from_logits Whether y_pred expected logits tensor. default, assume y_pred encodes probability distribution. label_smoothing Float [0, 1]. > 0 smooth labels. example, 0.1, use 0.1 / num_classes non-target labels 0.9 + 0.1 / num_classes target labels. axis Defaults -1. dimension along entropy computed.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_categorical_focal_crossentropy.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Computes the categorical focal crossentropy loss. — metric_categorical_focal_crossentropy","text":"Categorical focal crossentropy loss value.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_categorical_focal_crossentropy.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Computes the categorical focal crossentropy loss. — metric_categorical_focal_crossentropy","text":"","code":"y_true <- rbind(c(0, 1, 0), c(0, 0, 1)) y_pred <- rbind(c(0.05, 0.9, 0.05), c(0.1, 0.85, 0.05)) loss <- loss_categorical_focal_crossentropy(y_true, y_pred) loss ## tf.Tensor([2.6340148e-04 6.7591208e-01], shape=(2), dtype=float32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/metric_categorical_hinge.html","id":null,"dir":"Reference","previous_headings":"","what":"Computes the categorical hinge metric between y_true and y_pred. — metric_categorical_hinge","title":"Computes the categorical hinge metric between y_true and y_pred. — metric_categorical_hinge","text":"Formula:   neg=maximum((1-y_true)*y_pred) pos=sum(y_true*y_pred)","code":"loss <- maximum(neg - pos + 1, 0)"},{"path":"https://keras3.posit.co/dev/reference/metric_categorical_hinge.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes the categorical hinge metric between y_true and y_pred. — metric_categorical_hinge","text":"","code":"metric_categorical_hinge(   y_true,   y_pred,   ...,   name = \"categorical_hinge\",   dtype = NULL )"},{"path":"https://keras3.posit.co/dev/reference/metric_categorical_hinge.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Computes the categorical hinge metric between y_true and y_pred. — metric_categorical_hinge","text":"y_true ground truth values. y_true values expected either {-1, +1} {0, 1} (.e. one-hot-encoded tensor) shape = [batch_size, d0, .. dN]. y_pred predicted values shape = [batch_size, d0, .. dN]. ... forward/backward compatability. name (Optional) string name metric instance. dtype (Optional) data type metric result.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_categorical_hinge.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Computes the categorical hinge metric between y_true and y_pred. — metric_categorical_hinge","text":"Categorical hinge loss values shape = [batch_size, d0, .. dN-1].","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_categorical_hinge.html","id":"usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes the categorical hinge metric between y_true and y_pred. — metric_categorical_hinge","text":"Standalone usage:","code":"m <- metric_categorical_hinge() m$update_state(rbind(c(0, 1), c(0, 0)), rbind(c(0.6, 0.4), c(0.4, 0.6))) m$result() ## tf.Tensor(1.4000001, shape=(), dtype=float32) m$reset_state() m$update_state(rbind(c(0, 1), c(0, 0)), rbind(c(0.6, 0.4), c(0.4, 0.6)),                sample_weight = c(1, 0)) m$result() ## tf.Tensor(1.2, shape=(), dtype=float32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/metric_concordance_correlation.html","id":null,"dir":"Reference","previous_headings":"","what":"Calculates the Concordance Correlation Coefficient (CCC). — metric_concordance_correlation","title":"Calculates the Concordance Correlation Coefficient (CCC). — metric_concordance_correlation","text":"Formula:   CCC evaluates agreement true values (y_true) predicted values (y_pred) considering precision accuracy. coefficient ranges -1 1, value 1 indicates perfect agreement. metric useful regression tasks important assess well predictions match true values, taking account correlation proximity 45-degree line perfect concordance.","code":"loss <- mean(   2 * (y_true - mean(y_true)) * (y_pred - mean(y_pred)) /     (var(y_true) + var(y_pred) + (mean(y_true) - mean(y_pred))^2) )"},{"path":"https://keras3.posit.co/dev/reference/metric_concordance_correlation.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Calculates the Concordance Correlation Coefficient (CCC). — metric_concordance_correlation","text":"","code":"metric_concordance_correlation(   y_true,   y_pred,   axis = -1L,   ...,   name = \"concordance_correlation\",   dtype = NULL )"},{"path":"https://keras3.posit.co/dev/reference/metric_concordance_correlation.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Calculates the Concordance Correlation Coefficient (CCC). — metric_concordance_correlation","text":"y_true Tensor true targets. y_pred Tensor predicted targets. axis (Optional) integer tuple integers axis/axes along compute metric. Defaults -1. ... forward/backward compatability. name (Optional) string name metric instance. dtype (Optional) data type metric result.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_concordance_correlation.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Calculates the Concordance Correlation Coefficient (CCC). — metric_concordance_correlation","text":"Usage compile() API:","code":"ccc <- metric_concordance_correlation(axis=-1) y_true <- rbind(c(0, 1, 0.5),                 c(1, 1, 0.2)) y_pred <- rbind(c(0.1, 0.9, 0.5),                 c(1, 0.9, 0.2)) ccc$update_state(y_true, y_pred) ccc$result() ## tf.Tensor(0.9816317, shape=(), dtype=float32) model |> compile(   optimizer = 'sgd',   loss = 'mean_squared_error',   metrics = c(metric_concordance_correlation()) )"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/metric_cosine_similarity.html","id":null,"dir":"Reference","previous_headings":"","what":"Computes the cosine similarity between the labels and predictions. — metric_cosine_similarity","title":"Computes the cosine similarity between the labels and predictions. — metric_cosine_similarity","text":"Formula:   See: Cosine Similarity. metric keeps average cosine similarity predictions labels stream data.","code":"loss <- sum(l2_norm(y_true) * l2_norm(y_pred))"},{"path":"https://keras3.posit.co/dev/reference/metric_cosine_similarity.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes the cosine similarity between the labels and predictions. — metric_cosine_similarity","text":"","code":"metric_cosine_similarity(   ...,   name = \"cosine_similarity\",   dtype = NULL,   axis = -1L )"},{"path":"https://keras3.posit.co/dev/reference/metric_cosine_similarity.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Computes the cosine similarity between the labels and predictions. — metric_cosine_similarity","text":"... forward/backward compatability. name (Optional) string name metric instance. dtype (Optional) data type metric result. axis (Optional) Defaults -1. dimension along cosine similarity computed.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_cosine_similarity.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Computes the cosine similarity between the labels and predictions. — metric_cosine_similarity","text":"Metric instance returned. Metric instance can passed directly compile(metrics = ), used standalone object. See ?Metric example usage.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_cosine_similarity.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Computes the cosine similarity between the labels and predictions. — metric_cosine_similarity","text":"Standalone usage:         Usage compile() API:","code":"m <- metric_cosine_similarity(axis=2) m$update_state(rbind(c(0., 1.), c(1., 1.)), rbind(c(1., 0.), c(1., 1.))) m$result() ## tf.Tensor(0.5, shape=(), dtype=float32) m$reset_state() m$update_state(rbind(c(0., 1.), c(1., 1.)), rbind(c(1., 0.), c(1., 1.)),                sample_weight = c(0.3, 0.7)) m$result() ## tf.Tensor(0.7, shape=(), dtype=float32) model %>% compile(   optimizer = 'sgd',   loss = 'mse',   metrics = list(metric_cosine_similarity(axis=2)))"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/metric_f1_score.html","id":null,"dir":"Reference","previous_headings":"","what":"Computes F-1 Score. — metric_f1_score","title":"Computes F-1 Score. — metric_f1_score","text":"Formula:   harmonic mean precision recall. output range [0, 1]. works multi-class multi-label classification.","code":"f1_score <- 2 * (precision * recall) / (precision + recall)"},{"path":"https://keras3.posit.co/dev/reference/metric_f1_score.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes F-1 Score. — metric_f1_score","text":"","code":"metric_f1_score(   ...,   average = NULL,   threshold = NULL,   name = \"f1_score\",   dtype = NULL )"},{"path":"https://keras3.posit.co/dev/reference/metric_f1_score.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Computes F-1 Score. — metric_f1_score","text":"... forward/backward compatability. average Type averaging performed data. Acceptable values NULL, \"micro\", \"macro\" \"weighted\". Defaults NULL. NULL, averaging performed result() return score class. \"micro\", compute metrics globally counting total true positives, false negatives false positives. \"macro\", compute metrics label, return unweighted mean. take label imbalance account. \"weighted\", compute metrics label, return average weighted support (number true instances label). alters \"macro\" account label imbalance. can result score precision recall. threshold Elements y_pred greater threshold converted 1, rest 0. threshold NULL, argmax y_pred converted 1, rest 0. name Optional. String name metric instance. dtype Optional. Data type metric result.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_f1_score.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Computes F-1 Score. — metric_f1_score","text":"Metric instance returned. Metric instance can passed directly compile(metrics = ), used standalone object. See ?Metric example usage.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_f1_score.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Computes F-1 Score. — metric_f1_score","text":"","code":"metric <- metric_f1_score(threshold = 0.5) y_true <- rbind(c(1, 1, 1),                 c(1, 0, 0),                 c(1, 1, 0)) y_pred <- rbind(c(0.2, 0.6, 0.7),                 c(0.2, 0.6, 0.6),                 c(0.6, 0.8, 0.0)) metric$update_state(y_true, y_pred) result <- metric$result() result ## tf.Tensor([0.49999997 0.79999995 0.66666657], shape=(3), dtype=float32)"},{"path":"https://keras3.posit.co/dev/reference/metric_f1_score.html","id":"returns","dir":"Reference","previous_headings":"","what":"Returns","title":"Computes F-1 Score. — metric_f1_score","text":"F-1 Score: float.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/metric_false_negatives.html","id":null,"dir":"Reference","previous_headings":"","what":"Calculates the number of false negatives. — metric_false_negatives","title":"Calculates the number of false negatives. — metric_false_negatives","text":"sample_weight given, calculates sum weights false negatives. metric creates one local variable, accumulator used keep track number false negatives. sample_weight NULL, weights default 1. Use sample_weight 0 mask values.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_false_negatives.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Calculates the number of false negatives. — metric_false_negatives","text":"","code":"metric_false_negatives(..., thresholds = NULL, name = NULL, dtype = NULL)"},{"path":"https://keras3.posit.co/dev/reference/metric_false_negatives.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Calculates the number of false negatives. — metric_false_negatives","text":"... forward/backward compatability. thresholds (Optional) Defaults 0.5. float value, Python list float threshold values [0, 1]. threshold compared prediction values determine truth value predictions (.e., threshold TRUE, FALSE). used loss function sets from_logits=TRUE (.e. sigmoid applied predictions), thresholds set 0. One metric value generated threshold value. name (Optional) string name metric instance. dtype (Optional) data type metric result.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_false_negatives.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Calculates the number of false negatives. — metric_false_negatives","text":"Metric instance returned. Metric instance can passed directly compile(metrics = ), used standalone object. See ?Metric example usage.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_false_negatives.html","id":"usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Calculates the number of false negatives. — metric_false_negatives","text":"Standalone usage:","code":"m <- metric_false_negatives() m$update_state(c(0, 1, 1, 1), c(0, 1, 0, 0)) m$result() ## tf.Tensor(2.0, shape=(), dtype=float32) m$reset_state() m$update_state(c(0, 1, 1, 1), c(0, 1, 0, 0), sample_weight=c(0, 0, 1, 0)) m$result() ## tf.Tensor(1.0, shape=(), dtype=float32) # 1.0"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/metric_false_positives.html","id":null,"dir":"Reference","previous_headings":"","what":"Calculates the number of false positives. — metric_false_positives","title":"Calculates the number of false positives. — metric_false_positives","text":"sample_weight given, calculates sum weights false positives. metric creates one local variable, accumulator used keep track number false positives. sample_weight NULL, weights default 1. Use sample_weight 0 mask values.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_false_positives.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Calculates the number of false positives. — metric_false_positives","text":"","code":"metric_false_positives(..., thresholds = NULL, name = NULL, dtype = NULL)"},{"path":"https://keras3.posit.co/dev/reference/metric_false_positives.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Calculates the number of false positives. — metric_false_positives","text":"... forward/backward compatability. thresholds (Optional) Defaults 0.5. float value, Python list float threshold values [0, 1]. threshold compared prediction values determine truth value predictions (.e., threshold TRUE, FALSE). used loss function sets from_logits=TRUE (.e. sigmoid applied predictions), thresholds set 0. One metric value generated threshold value. name (Optional) string name metric instance. dtype (Optional) data type metric result.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_false_positives.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Calculates the number of false positives. — metric_false_positives","text":"Metric instance returned. Metric instance can passed directly compile(metrics = ), used standalone object. See ?Metric example usage.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_false_positives.html","id":"usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Calculates the number of false positives. — metric_false_positives","text":"Standalone usage:","code":"m <- metric_false_positives() m$update_state(c(0, 1, 0, 0), c(0, 0, 1, 1)) m$result() ## tf.Tensor(2.0, shape=(), dtype=float32) m$reset_state() m$update_state(c(0, 1, 0, 0), c(0, 0, 1, 1), sample_weight = c(0, 0, 1, 0)) m$result() ## tf.Tensor(1.0, shape=(), dtype=float32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/metric_fbeta_score.html","id":null,"dir":"Reference","previous_headings":"","what":"Computes F-Beta score. — metric_fbeta_score","title":"Computes F-Beta score. — metric_fbeta_score","text":"Formula:   weighted harmonic mean precision recall. output range [0, 1]. works multi-class multi-label classification.","code":"b2 <- beta^2 f_beta_score <- (1 + b2) * (precision * recall) / (precision * b2 + recall)"},{"path":"https://keras3.posit.co/dev/reference/metric_fbeta_score.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes F-Beta score. — metric_fbeta_score","text":"","code":"metric_fbeta_score(   ...,   average = NULL,   beta = 1,   threshold = NULL,   name = \"fbeta_score\",   dtype = NULL )"},{"path":"https://keras3.posit.co/dev/reference/metric_fbeta_score.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Computes F-Beta score. — metric_fbeta_score","text":"... forward/backward compatability. average Type averaging performed across per-class results multi-class case. Acceptable values NULL, \"micro\", \"macro\" \"weighted\". Defaults NULL. NULL, averaging performed result() return score class. \"micro\", compute metrics globally counting total true positives, false negatives false positives. \"macro\", compute metrics label, return unweighted mean. take label imbalance account. \"weighted\", compute metrics label, return average weighted support (number true instances label). alters \"macro\" account label imbalance. can result score precision recall. beta Determines weight given recall harmonic mean precision recall (see pseudocode equation ). Defaults 1. threshold Elements y_pred greater threshold converted 1, rest 0. threshold NULL, argmax y_pred converted 1, rest 0. name Optional. String name metric instance. dtype Optional. Data type metric result.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_fbeta_score.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Computes F-Beta score. — metric_fbeta_score","text":"Metric instance returned. Metric instance can passed directly compile(metrics = ), used standalone object. See ?Metric example usage.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_fbeta_score.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Computes F-Beta score. — metric_fbeta_score","text":"","code":"metric <- metric_fbeta_score(beta = 2.0, threshold = 0.5) y_true <- rbind(c(1, 1, 1),                 c(1, 0, 0),                 c(1, 1, 0)) y_pred <- rbind(c(0.2, 0.6, 0.7),                 c(0.2, 0.6, 0.6),                 c(0.6, 0.8, 0.0)) metric$update_state(y_true, y_pred) metric$result() ## tf.Tensor([0.3846154  0.90909094 0.8333332 ], shape=(3), dtype=float32)"},{"path":"https://keras3.posit.co/dev/reference/metric_fbeta_score.html","id":"returns","dir":"Reference","previous_headings":"","what":"Returns","title":"Computes F-Beta score. — metric_fbeta_score","text":"F-Beta Score: float.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/metric_hinge.html","id":null,"dir":"Reference","previous_headings":"","what":"Computes the hinge metric between y_true and y_pred. — metric_hinge","title":"Computes the hinge metric between y_true and y_pred. — metric_hinge","text":"Formula:   y_true values expected -1 1. binary (0 1) labels provided convert -1 1.","code":"loss <- mean(maximum(1 - y_true * y_pred, 0), axis=-1)"},{"path":"https://keras3.posit.co/dev/reference/metric_hinge.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes the hinge metric between y_true and y_pred. — metric_hinge","text":"","code":"metric_hinge(y_true, y_pred, ..., name = \"hinge\", dtype = NULL)"},{"path":"https://keras3.posit.co/dev/reference/metric_hinge.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Computes the hinge metric between y_true and y_pred. — metric_hinge","text":"y_true ground truth values. y_true values expected -1 1. binary (0 1) labels provided converted -1 1 shape = [batch_size, d0, .. dN]. y_pred predicted values shape = [batch_size, d0, .. dN]. ... forward/backward compatability. name (Optional) string name metric instance. dtype (Optional) data type metric result.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_hinge.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Computes the hinge metric between y_true and y_pred. — metric_hinge","text":"y_true y_pred missing, Metric instance returned. Metric instance can passed directly compile(metrics = ), used standalone object. See ?Metric example usage. y_true y_pred provided, tensor computed value returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_hinge.html","id":"usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes the hinge metric between y_true and y_pred. — metric_hinge","text":"Standalone usage:","code":"m <- metric_hinge() m$update_state(rbind(c(0, 1), c(0, 0)), rbind(c(0.6, 0.4), c(0.4, 0.6))) m$result() ## tf.Tensor(1.3, shape=(), dtype=float32) m$reset_state() m$update_state(rbind(c(0, 1), c(0, 0)), rbind(c(0.6, 0.4), c(0.4, 0.6)),                sample_weight = c(1, 0)) m$result() ## tf.Tensor(1.1, shape=(), dtype=float32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/metric_huber.html","id":null,"dir":"Reference","previous_headings":"","what":"Computes Huber loss value. — metric_huber","title":"Computes Huber loss value. — metric_huber","text":"Formula:   See: Huber loss.","code":"for (x in error) {   if (abs(x) <= delta){     loss <- c(loss, (0.5 * x^2))   } else if (abs(x) > delta) {     loss <- c(loss, (delta * abs(x) - 0.5 * delta^2))   } } loss <- mean(loss)"},{"path":"https://keras3.posit.co/dev/reference/metric_huber.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes Huber loss value. — metric_huber","text":"","code":"metric_huber(y_true, y_pred, delta = 1)"},{"path":"https://keras3.posit.co/dev/reference/metric_huber.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Computes Huber loss value. — metric_huber","text":"y_true tensor true targets. y_pred tensor predicted targets. delta float, point Huber loss function changes quadratic linear. Defaults 1.0.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_huber.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Computes Huber loss value. — metric_huber","text":"Tensor one scalar loss entry per sample.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_huber.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Computes Huber loss value. — metric_huber","text":"","code":"y_true <- rbind(c(0, 1), c(0, 0)) y_pred <- rbind(c(0.6, 0.4), c(0.4, 0.6)) loss <- loss_huber(y_true, y_pred)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/metric_iou.html","id":null,"dir":"Reference","previous_headings":"","what":"Computes the Intersection-Over-Union metric for specific target classes. — metric_iou","title":"Computes the Intersection-Over-Union metric for specific target classes. — metric_iou","text":"Formula:   Intersection--Union common evaluation metric semantic image segmentation. compute IoUs, predictions accumulated confusion matrix, weighted sample_weight metric calculated . sample_weight NULL, weights default 1. Use sample_weight 0 mask values. Note, class first computes IoUs individual classes, returns mean IoUs classes specified target_class_ids. target_class_ids one id value, IoU specific class returned.","code":"iou <- true_positives / (true_positives + false_positives + false_negatives)"},{"path":"https://keras3.posit.co/dev/reference/metric_iou.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes the Intersection-Over-Union metric for specific target classes. — metric_iou","text":"","code":"metric_iou(   ...,   num_classes,   target_class_ids,   name = NULL,   dtype = NULL,   ignore_class = NULL,   sparse_y_true = TRUE,   sparse_y_pred = TRUE,   axis = -1L )"},{"path":"https://keras3.posit.co/dev/reference/metric_iou.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Computes the Intersection-Over-Union metric for specific target classes. — metric_iou","text":"... forward/backward compatability. num_classes possible number labels prediction task can . target_class_ids list target class ids metric returned. compute IoU specific class, list single id value provided. name (Optional) string name metric instance. dtype (Optional) data type metric result. ignore_class Optional integer. ID class ignored metric computation. useful, example, segmentation problems featuring \"void\" class (commonly -1 255) segmentation maps. default (ignore_class=NULL), classes considered. sparse_y_true Whether labels encoded using integers dense floating point vectors. FALSE, argmax function used determine sample's likely associated label. sparse_y_pred Whether predictions encoded using integers dense floating point vectors. FALSE, argmax function used determine sample's likely associated label. axis (Optional) -1 dimension containing logits. Defaults -1.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_iou.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Computes the Intersection-Over-Union metric for specific target classes. — metric_iou","text":"Metric instance returned. Metric instance can passed directly compile(metrics = ), used standalone object. See ?Metric example usage.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_iou.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Computes the Intersection-Over-Union metric for specific target classes. — metric_iou","text":"Standalone usage:                 Usage compile() API:","code":"m <- metric_iou(num_classes = 2L, target_class_ids = list(0L)) m$update_state(c(0, 0, 1, 1), c(0, 1, 0, 1)) ## tf.Tensor( ## [[1 1] ##  [1 1]], shape=(2, 2), dtype=int64) m$result() ## tf.Tensor(0.3333333, shape=(), dtype=float32) m$reset_state() m$update_state(c(0, 0, 1, 1), c(0, 1, 0, 1),                sample_weight = 10 * c(0.3, 0.3, 0.3, 0.1)) ## tf.Tensor( ## [[3 3] ##  [3 1]], shape=(2, 2), dtype=int64) m$result() ## tf.Tensor(0.3333333, shape=(), dtype=float32) model %>% compile(   optimizer = 'sgd',   loss = 'mse',   metrics = list(metric_iou(num_classes = 2L, target_class_ids = list(0L))))"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/metric_kl_divergence.html","id":null,"dir":"Reference","previous_headings":"","what":"Computes Kullback-Leibler divergence metric between y_true and — metric_kl_divergence","title":"Computes Kullback-Leibler divergence metric between y_true and — metric_kl_divergence","text":"Formula:   y_true y_pred expected probability distributions, values 0 1. get clipped [0, 1] range.","code":"loss <- y_true * log(y_true / y_pred)"},{"path":"https://keras3.posit.co/dev/reference/metric_kl_divergence.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes Kullback-Leibler divergence metric between y_true and — metric_kl_divergence","text":"","code":"metric_kl_divergence(y_true, y_pred, ..., name = \"kl_divergence\", dtype = NULL)"},{"path":"https://keras3.posit.co/dev/reference/metric_kl_divergence.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Computes Kullback-Leibler divergence metric between y_true and — metric_kl_divergence","text":"y_true Tensor true targets. y_pred Tensor predicted targets. ... forward/backward compatability. name (Optional) string name metric instance. dtype (Optional) data type metric result.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_kl_divergence.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Computes Kullback-Leibler divergence metric between y_true and — metric_kl_divergence","text":"y_true y_pred missing, Metric instance returned. Metric instance can passed directly compile(metrics = ), used standalone object. See ?Metric example usage. y_true y_pred provided, tensor computed value returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_kl_divergence.html","id":"usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes Kullback-Leibler divergence metric between y_true and — metric_kl_divergence","text":"Standalone usage:         Usage compile() API:","code":"m <- metric_kl_divergence() m$update_state(rbind(c(0, 1), c(0, 0)), rbind(c(0.6, 0.4), c(0.4, 0.6))) m$result() ## tf.Tensor(0.45814303, shape=(), dtype=float32) m$reset_state() m$update_state(rbind(c(0, 1), c(0, 0)), rbind(c(0.6, 0.4), c(0.4, 0.6)),                sample_weight = c(1, 0)) m$result() ## tf.Tensor(0.91628915, shape=(), dtype=float32) model %>% compile(optimizer = 'sgd',                   loss = 'mse',                   metrics = list(metric_kl_divergence()))"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/metric_log_cosh.html","id":null,"dir":"Reference","previous_headings":"","what":"Logarithm of the hyperbolic cosine of the prediction error. — metric_log_cosh","title":"Logarithm of the hyperbolic cosine of the prediction error. — metric_log_cosh","text":"Formula:   Note log(cosh(x)) approximately equal (x ** 2) / 2 small x abs(x) - log(2) large x. means 'logcosh' works mostly like mean squared error, strongly affected occasional wildly incorrect prediction.","code":"loss <- mean(log(cosh(y_pred - y_true)), axis=-1)"},{"path":"https://keras3.posit.co/dev/reference/metric_log_cosh.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Logarithm of the hyperbolic cosine of the prediction error. — metric_log_cosh","text":"","code":"metric_log_cosh(y_true, y_pred)"},{"path":"https://keras3.posit.co/dev/reference/metric_log_cosh.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Logarithm of the hyperbolic cosine of the prediction error. — metric_log_cosh","text":"y_true Ground truth values shape = [batch_size, d0, .. dN]. y_pred predicted values shape = [batch_size, d0, .. dN].","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_log_cosh.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Logarithm of the hyperbolic cosine of the prediction error. — metric_log_cosh","text":"Logcosh error values shape = [batch_size, d0, .. dN-1].","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_log_cosh.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Logarithm of the hyperbolic cosine of the prediction error. — metric_log_cosh","text":"","code":"y_true <- rbind(c(0., 1.), c(0., 0.)) y_pred <- rbind(c(1., 1.), c(0., 0.)) loss <- metric_log_cosh(y_true, y_pred) loss ## tf.Tensor([0.2168904 0.       ], shape=(2), dtype=float32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/metric_log_cosh_error.html","id":null,"dir":"Reference","previous_headings":"","what":"Computes the logarithm of the hyperbolic cosine of the prediction error. — metric_log_cosh_error","title":"Computes the logarithm of the hyperbolic cosine of the prediction error. — metric_log_cosh_error","text":"Formula:","code":"error <- y_pred - y_true logcosh <- mean(log((exp(error) + exp(-error))/2), axis=-1)"},{"path":"https://keras3.posit.co/dev/reference/metric_log_cosh_error.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes the logarithm of the hyperbolic cosine of the prediction error. — metric_log_cosh_error","text":"","code":"metric_log_cosh_error(..., name = \"logcosh\", dtype = NULL)"},{"path":"https://keras3.posit.co/dev/reference/metric_log_cosh_error.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Computes the logarithm of the hyperbolic cosine of the prediction error. — metric_log_cosh_error","text":"... forward/backward compatability. name (Optional) string name metric instance. dtype (Optional) data type metric result.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_log_cosh_error.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Computes the logarithm of the hyperbolic cosine of the prediction error. — metric_log_cosh_error","text":"Metric instance returned. Metric instance can passed directly compile(metrics = ), used standalone object. See ?Metric example usage.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_log_cosh_error.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Computes the logarithm of the hyperbolic cosine of the prediction error. — metric_log_cosh_error","text":"Standalone usage:         Usage compile() API:","code":"m <- metric_log_cosh_error() m$update_state(rbind(c(0, 1), c(0, 0)), rbind(c(1, 1), c(0, 0))) m$result() ## tf.Tensor(0.108445205, shape=(), dtype=float32) m$reset_state() m$update_state(rbind(c(0, 1), c(0, 0)), rbind(c(1, 1), c(0, 0)),                sample_weight = c(1, 0)) m$result() ## tf.Tensor(0.21689041, shape=(), dtype=float32) model %>% compile(optimizer = 'sgd',                   loss = 'mse',                   metrics = list(metric_log_cosh_error()))"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/metric_mean.html","id":null,"dir":"Reference","previous_headings":"","what":"Compute the (weighted) mean of the given values. — metric_mean","title":"Compute the (weighted) mean of the given values. — metric_mean","text":"example, values c(1, 3, 5, 7) mean 4. sample_weight specified c(1, 1, 0, 0) mean 2. metric creates two variables, total count. mean value returned simply total divided count.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_mean.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Compute the (weighted) mean of the given values. — metric_mean","text":"","code":"metric_mean(..., name = \"mean\", dtype = NULL)"},{"path":"https://keras3.posit.co/dev/reference/metric_mean.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Compute the (weighted) mean of the given values. — metric_mean","text":"... forward/backward compatability. name (Optional) string name metric instance. dtype (Optional) data type metric result.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_mean.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Compute the (weighted) mean of the given values. — metric_mean","text":"Metric instance returned. Metric instance can passed directly compile(metrics = ), used standalone object. See ?Metric example usage.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_mean.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Compute the (weighted) mean of the given values. — metric_mean","text":"","code":"m <- metric_mean() m$update_state(c(1, 3, 5, 7)) m$result() ## tf.Tensor(4.0, shape=(), dtype=float32) # calling a metric directly is equivalent to calling # m$update_state(); m$result() m <- metric_mean() m(c(1, 3, 5, 7)) ## tf.Tensor(4.0, shape=(), dtype=float32) m$reset_state() m$update_state(c(1, 3, 5, 7), sample_weight = c(1, 1, 0, 0)) m$result() ## tf.Tensor(2.0, shape=(), dtype=float32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/metric_mean_absolute_error.html","id":null,"dir":"Reference","previous_headings":"","what":"Computes the mean absolute error between the labels and predictions. — metric_mean_absolute_error","title":"Computes the mean absolute error between the labels and predictions. — metric_mean_absolute_error","text":"Formula:","code":"loss <- mean(abs(y_true - y_pred))"},{"path":"https://keras3.posit.co/dev/reference/metric_mean_absolute_error.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes the mean absolute error between the labels and predictions. — metric_mean_absolute_error","text":"","code":"metric_mean_absolute_error(   y_true,   y_pred,   ...,   name = \"mean_absolute_error\",   dtype = NULL )"},{"path":"https://keras3.posit.co/dev/reference/metric_mean_absolute_error.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Computes the mean absolute error between the labels and predictions. — metric_mean_absolute_error","text":"y_true Ground truth values shape = [batch_size, d0, .. dN]. y_pred predicted values shape = [batch_size, d0, .. dN]. ... forward/backward compatability. name (Optional) string name metric instance. dtype (Optional) data type metric result.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_mean_absolute_error.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Computes the mean absolute error between the labels and predictions. — metric_mean_absolute_error","text":"y_true y_pred missing, Metric instance returned. Metric instance can passed directly compile(metrics = ), used standalone object. See ?Metric example usage. y_true y_pred provided, tensor computed value returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_mean_absolute_error.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Computes the mean absolute error between the labels and predictions. — metric_mean_absolute_error","text":"Standalone usage:         Usage compile() API:","code":"m <- metric_mean_absolute_error() m$update_state(rbind(c(0, 1), c(0, 0)), rbind(c(1, 1), c(0, 0))) m$result() ## tf.Tensor(0.25, shape=(), dtype=float32) m$reset_state() m$update_state(rbind(c(0, 1), c(0, 0)), rbind(c(1, 1), c(0, 0)),                sample_weight = c(1, 0)) m$result() ## tf.Tensor(0.5, shape=(), dtype=float32) model %>% compile(     optimizer = 'sgd',     loss = 'mse',     metrics = list(metric_mean_absolute_error()))"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/metric_mean_absolute_percentage_error.html","id":null,"dir":"Reference","previous_headings":"","what":"Computes mean absolute percentage error between y_true and y_pred. — metric_mean_absolute_percentage_error","title":"Computes mean absolute percentage error between y_true and y_pred. — metric_mean_absolute_percentage_error","text":"Formula:   Division zero prevented dividing maximum(y_true, epsilon) epsilon = keras$backend$epsilon() (default 1e-7).","code":"loss <- 100 * mean(abs((y_true - y_pred) / y_true), axis=-1)"},{"path":"https://keras3.posit.co/dev/reference/metric_mean_absolute_percentage_error.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes mean absolute percentage error between y_true and y_pred. — metric_mean_absolute_percentage_error","text":"","code":"metric_mean_absolute_percentage_error(   y_true,   y_pred,   ...,   name = \"mean_absolute_percentage_error\",   dtype = NULL )"},{"path":"https://keras3.posit.co/dev/reference/metric_mean_absolute_percentage_error.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Computes mean absolute percentage error between y_true and y_pred. — metric_mean_absolute_percentage_error","text":"y_true Ground truth values shape = [batch_size, d0, .. dN]. y_pred predicted values shape = [batch_size, d0, .. dN]. ... forward/backward compatability. name (Optional) string name metric instance. dtype (Optional) data type metric result.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_mean_absolute_percentage_error.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Computes mean absolute percentage error between y_true and y_pred. — metric_mean_absolute_percentage_error","text":"y_true y_pred missing, Metric instance returned. Metric instance can passed directly compile(metrics = ), used standalone object. See ?Metric example usage. y_true y_pred provided, tensor computed value returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_mean_absolute_percentage_error.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Computes mean absolute percentage error between y_true and y_pred. — metric_mean_absolute_percentage_error","text":"Standalone usage:         Usage compile() API:","code":"m <- metric_mean_absolute_percentage_error() m$update_state(rbind(c(0, 1), c(0, 0)), rbind(c(1, 1), c(0, 0))) m$result() ## tf.Tensor(2.5e+08, shape=(), dtype=float32) m$reset_state() m$update_state(rbind(c(0, 1), c(0, 0)), rbind(c(1, 1), c(0, 0)),                sample_weight = c(1, 0)) m$result() ## tf.Tensor(5e+08, shape=(), dtype=float32) model %>% compile(     optimizer = 'sgd',     loss = 'mse',     metrics = list(metric_mean_absolute_percentage_error()))"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/metric_mean_iou.html","id":null,"dir":"Reference","previous_headings":"","what":"Computes the mean Intersection-Over-Union metric. — metric_mean_iou","title":"Computes the mean Intersection-Over-Union metric. — metric_mean_iou","text":"Formula:   Intersection--Union common evaluation metric semantic image segmentation. compute IoUs, predictions accumulated confusion matrix, weighted sample_weight metric calculated . sample_weight NULL, weights default 1. Use sample_weight 0 mask values. Note class first computes IoUs individual classes, returns mean values.","code":"iou <- true_positives / (true_positives + false_positives + false_negatives)"},{"path":"https://keras3.posit.co/dev/reference/metric_mean_iou.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes the mean Intersection-Over-Union metric. — metric_mean_iou","text":"","code":"metric_mean_iou(   ...,   num_classes,   name = NULL,   dtype = NULL,   ignore_class = NULL,   sparse_y_true = TRUE,   sparse_y_pred = TRUE,   axis = -1L )"},{"path":"https://keras3.posit.co/dev/reference/metric_mean_iou.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Computes the mean Intersection-Over-Union metric. — metric_mean_iou","text":"... forward/backward compatability. num_classes possible number labels prediction task can . value must provided, since confusion matrix dimension = [num_classes, num_classes] allocated. name (Optional) string name metric instance. dtype (Optional) data type metric result. ignore_class Optional integer. ID class ignored metric computation. useful, example, segmentation problems featuring \"void\" class (commonly -1 255) segmentation maps. default (ignore_class=NULL), classes considered. sparse_y_true Whether labels encoded using integers dense floating point vectors. FALSE, argmax function used determine sample's likely associated label. sparse_y_pred Whether predictions encoded using integers dense floating point vectors. FALSE, argmax function used determine sample's likely associated label. axis (Optional) dimension containing logits. Defaults -1.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_mean_iou.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Computes the mean Intersection-Over-Union metric. — metric_mean_iou","text":"Metric instance returned. Metric instance can passed directly compile(metrics = ), used standalone object. See ?Metric example usage.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_mean_iou.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Computes the mean Intersection-Over-Union metric. — metric_mean_iou","text":"Standalone usage:                 Usage compile() API:","code":"# cm = [[1, 1], #       [1, 1]] # sum_row = [2, 2], sum_col = [2, 2], true_positives = [1, 1] # iou = true_positives / (sum_row + sum_col - true_positives)) # result = (1 / (2 + 2 - 1) + 1 / (2 + 2 - 1)) / 2 = 0.33 m <- metric_mean_iou(num_classes = 2) m$update_state(c(0, 0, 1, 1), c(0, 1, 0, 1)) ## tf.Tensor( ## [[1 1] ##  [1 1]], shape=(2, 2), dtype=int64) m$result() ## tf.Tensor(0.33333334, shape=(), dtype=float32) m$reset_state() m$update_state(c(0, 0, 1, 1), c(0, 1, 0, 1),                sample_weight= 10 * c(0.3, 0.3, 0.3, 0.1)) ## tf.Tensor( ## [[3 3] ##  [3 1]], shape=(2, 2), dtype=int64) m$result() ## tf.Tensor(0.23809525, shape=(), dtype=float32) model %>% compile(   optimizer = 'sgd',   loss = 'mse',   metrics = list(metric_mean_iou(num_classes=2)))"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/metric_mean_squared_error.html","id":null,"dir":"Reference","previous_headings":"","what":"Computes the mean squared error between y_true and y_pred. — metric_mean_squared_error","title":"Computes the mean squared error between y_true and y_pred. — metric_mean_squared_error","text":"Formula:","code":"loss <- mean(square(y_true - y_pred))"},{"path":"https://keras3.posit.co/dev/reference/metric_mean_squared_error.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes the mean squared error between y_true and y_pred. — metric_mean_squared_error","text":"","code":"metric_mean_squared_error(   y_true,   y_pred,   ...,   name = \"mean_squared_error\",   dtype = NULL )"},{"path":"https://keras3.posit.co/dev/reference/metric_mean_squared_error.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Computes the mean squared error between y_true and y_pred. — metric_mean_squared_error","text":"y_true Ground truth values shape = [batch_size, d0, .. dN]. y_pred predicted values shape = [batch_size, d0, .. dN]. ... forward/backward compatability. name (Optional) string name metric instance. dtype (Optional) data type metric result.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_mean_squared_error.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Computes the mean squared error between y_true and y_pred. — metric_mean_squared_error","text":"y_true y_pred missing, Metric instance returned. Metric instance can passed directly compile(metrics = ), used standalone object. See ?Metric example usage. y_true y_pred provided, tensor computed value returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_mean_squared_error.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Computes the mean squared error between y_true and y_pred. — metric_mean_squared_error","text":"","code":"m <- metric_mean_squared_error() m$update_state(rbind(c(0, 1), c(0, 0)), rbind(c(1, 1), c(0, 0))) m$result() ## tf.Tensor(0.25, shape=(), dtype=float32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/metric_mean_squared_logarithmic_error.html","id":null,"dir":"Reference","previous_headings":"","what":"Computes mean squared logarithmic error between y_true and y_pred. — metric_mean_squared_logarithmic_error","title":"Computes mean squared logarithmic error between y_true and y_pred. — metric_mean_squared_logarithmic_error","text":"Formula:   Note y_pred y_true less equal 0. Negative values 0 values replaced keras$backend$epsilon() (default 1e-7).","code":"loss <- mean(square(log(y_true + 1) - log(y_pred + 1)), axis=-1)"},{"path":"https://keras3.posit.co/dev/reference/metric_mean_squared_logarithmic_error.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes mean squared logarithmic error between y_true and y_pred. — metric_mean_squared_logarithmic_error","text":"","code":"metric_mean_squared_logarithmic_error(   y_true,   y_pred,   ...,   name = \"mean_squared_logarithmic_error\",   dtype = NULL )"},{"path":"https://keras3.posit.co/dev/reference/metric_mean_squared_logarithmic_error.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Computes mean squared logarithmic error between y_true and y_pred. — metric_mean_squared_logarithmic_error","text":"y_true Ground truth values shape = [batch_size, d0, .. dN]. y_pred predicted values shape = [batch_size, d0, .. dN]. ... forward/backward compatability. name (Optional) string name metric instance. dtype (Optional) data type metric result.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_mean_squared_logarithmic_error.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Computes mean squared logarithmic error between y_true and y_pred. — metric_mean_squared_logarithmic_error","text":"y_true y_pred missing, Metric instance returned. Metric instance can passed directly compile(metrics = ), used standalone object. See ?Metric example usage. y_true y_pred provided, tensor computed value returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_mean_squared_logarithmic_error.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Computes mean squared logarithmic error between y_true and y_pred. — metric_mean_squared_logarithmic_error","text":"Standalone usage:         Usage compile() API:","code":"m <- metric_mean_squared_logarithmic_error() m$update_state(rbind(c(0, 1), c(0, 0)), rbind(c(1, 1), c(0, 0))) m$result() ## tf.Tensor(0.12011322, shape=(), dtype=float32) m$reset_state() m$update_state(rbind(c(0, 1), c(0, 0)), rbind(c(1, 1), c(0, 0)),                sample_weight = c(1, 0)) m$result() ## tf.Tensor(0.24022643, shape=(), dtype=float32) model %>% compile(   optimizer = 'sgd',   loss = 'mse',   metrics = list(metric_mean_squared_logarithmic_error()))"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/metric_mean_wrapper.html","id":null,"dir":"Reference","previous_headings":"","what":"Wrap a stateless metric function with the Mean metric. — metric_mean_wrapper","title":"Wrap a stateless metric function with the Mean metric. — metric_mean_wrapper","text":"use class quickly build mean metric function. function needs signature fn(y_true, y_pred) return per-sample loss array. metric_mean_wrapper$result() return average metric value across samples seen far. example:","code":"mse <- function(y_true, y_pred) {   (y_true - y_pred)^2 }  mse_metric <- metric_mean_wrapper(fn = mse) mse_metric$update_state(c(0, 1), c(1, 1)) mse_metric$result() ## tf.Tensor(0.5, shape=(), dtype=float32)"},{"path":"https://keras3.posit.co/dev/reference/metric_mean_wrapper.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Wrap a stateless metric function with the Mean metric. — metric_mean_wrapper","text":"","code":"metric_mean_wrapper(..., fn, name = NULL, dtype = NULL)"},{"path":"https://keras3.posit.co/dev/reference/metric_mean_wrapper.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Wrap a stateless metric function with the Mean metric. — metric_mean_wrapper","text":"... Keyword arguments pass fn. fn metric function wrap, signature fn(y_true, y_pred). name (Optional) string name metric instance. dtype (Optional) data type metric result.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_mean_wrapper.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Wrap a stateless metric function with the Mean metric. — metric_mean_wrapper","text":"Metric instance returned. Metric instance can passed directly compile(metrics = ), used standalone object. See ?Metric example usage.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/metric_one_hot_iou.html","id":null,"dir":"Reference","previous_headings":"","what":"Computes the Intersection-Over-Union metric for one-hot encoded labels. — metric_one_hot_iou","title":"Computes the Intersection-Over-Union metric for one-hot encoded labels. — metric_one_hot_iou","text":"Formula:   Intersection--Union common evaluation metric semantic image segmentation. compute IoUs, predictions accumulated confusion matrix, weighted sample_weight metric calculated . sample_weight NULL, weights default 1. Use sample_weight 0 mask values. class can used compute IoU multi-class classification tasks labels one-hot encoded (last axis one dimension per class). Note predictions also shape. compute IoU, first labels predictions converted back integer format taking argmax class axis. computation steps base IoU class apply. Note, one channel labels predictions, class class IoU. case, use IoU instead. Also, make sure num_classes equal number classes data, avoid \"labels bound\" error confusion matrix computed.","code":"iou <- true_positives / (true_positives + false_positives + false_negatives)"},{"path":"https://keras3.posit.co/dev/reference/metric_one_hot_iou.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes the Intersection-Over-Union metric for one-hot encoded labels. — metric_one_hot_iou","text":"","code":"metric_one_hot_iou(   ...,   num_classes,   target_class_ids,   name = NULL,   dtype = NULL,   ignore_class = NULL,   sparse_y_pred = FALSE,   axis = -1L )"},{"path":"https://keras3.posit.co/dev/reference/metric_one_hot_iou.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Computes the Intersection-Over-Union metric for one-hot encoded labels. — metric_one_hot_iou","text":"... forward/backward compatability. num_classes possible number labels prediction task can . target_class_ids list list target class ids metric returned. compute IoU specific class, list (list) single id value provided. name (Optional) string name metric instance. dtype (Optional) data type metric result. ignore_class Optional integer. ID class ignored metric computation. useful, example, segmentation problems featuring \"void\" class (commonly -1 255) segmentation maps. default (ignore_class=NULL), classes considered. sparse_y_pred Whether predictions encoded using integers dense floating point vectors. FALSE, argmax function used determine sample's likely associated label. axis (Optional) dimension containing logits. Defaults -1.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_one_hot_iou.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Computes the Intersection-Over-Union metric for one-hot encoded labels. — metric_one_hot_iou","text":"Metric instance returned. Metric instance can passed directly compile(metrics = ), used standalone object. See ?Metric example usage.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_one_hot_iou.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Computes the Intersection-Over-Union metric for one-hot encoded labels. — metric_one_hot_iou","text":"Standalone usage:         Usage compile() API:","code":"y_true <- rbind(c(0, 0, 1),                 c(1, 0, 0),                 c(0, 1, 0),                 c(1, 0, 0)) y_pred <- rbind(c(0.2, 0.3, 0.5),                 c(0.1, 0.2, 0.7),                 c(0.5, 0.3, 0.1),                 c(0.1, 0.4, 0.5)) sample_weight <- c(0.1, 0.2, 0.3, 0.4)  m <- metric_one_hot_iou(num_classes = 3, target_class_ids = c(0, 2)) m$update_state(y_true = y_true,                y_pred = y_pred,                sample_weight = 10*sample_weight) ## tf.Tensor( ## [[0 0 6] ##  [3 0 0] ##  [0 0 1]], shape=(3, 3), dtype=int64) m$result() ## tf.Tensor(0.071428575, shape=(), dtype=float32) model %>% compile(   optimizer = 'sgd',   loss = 'mse',   metrics = list(metric_one_hot_iou(     num_classes = 3L,     target_class_id = list(1L)   )) )"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/metric_one_hot_mean_iou.html","id":null,"dir":"Reference","previous_headings":"","what":"Computes mean Intersection-Over-Union metric for one-hot encoded labels. — metric_one_hot_mean_iou","title":"Computes mean Intersection-Over-Union metric for one-hot encoded labels. — metric_one_hot_mean_iou","text":"Formula:   Intersection--Union common evaluation metric semantic image segmentation. compute IoUs, predictions accumulated confusion matrix, weighted sample_weight metric calculated . sample_weight NULL, weights default 1. Use sample_weight 0 mask values. class can used compute mean IoU multi-class classification tasks labels one-hot encoded (last axis one dimension per class). Note predictions also shape. compute mean IoU, first labels predictions converted back integer format taking argmax class axis. computation steps base MeanIoU class apply. Note, one channel labels predictions, class class metric_mean_iou. case, use metric_mean_iou instead. Also, make sure num_classes equal number classes data, avoid \"labels bound\" error confusion matrix computed.","code":"iou <- true_positives / (true_positives + false_positives + false_negatives)"},{"path":"https://keras3.posit.co/dev/reference/metric_one_hot_mean_iou.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes mean Intersection-Over-Union metric for one-hot encoded labels. — metric_one_hot_mean_iou","text":"","code":"metric_one_hot_mean_iou(   ...,   num_classes,   name = NULL,   dtype = NULL,   ignore_class = NULL,   sparse_y_pred = FALSE,   axis = -1L )"},{"path":"https://keras3.posit.co/dev/reference/metric_one_hot_mean_iou.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Computes mean Intersection-Over-Union metric for one-hot encoded labels. — metric_one_hot_mean_iou","text":"... forward/backward compatability. num_classes possible number labels prediction task can . name (Optional) string name metric instance. dtype (Optional) data type metric result. ignore_class Optional integer. ID class ignored metric computation. useful, example, segmentation problems featuring \"void\" class (commonly -1 255) segmentation maps. default (ignore_class=NULL), classes considered. sparse_y_pred Whether predictions encoded using natural numbers probability distribution vectors. FALSE, argmax function used determine sample's likely associated label. axis (Optional) dimension containing logits. Defaults -1.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_one_hot_mean_iou.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Computes mean Intersection-Over-Union metric for one-hot encoded labels. — metric_one_hot_mean_iou","text":"Metric instance returned. Metric instance can passed directly compile(metrics = ), used standalone object. See ?Metric example usage.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_one_hot_mean_iou.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Computes mean Intersection-Over-Union metric for one-hot encoded labels. — metric_one_hot_mean_iou","text":"Standalone usage:         Usage compile() API:","code":"y_true <- rbind(c(0, 0, 1), c(1, 0, 0), c(0, 1, 0), c(1, 0, 0)) y_pred <- rbind(c(0.2, 0.3, 0.5), c(0.1, 0.2, 0.7), c(0.5, 0.3, 0.1),                 c(0.1, 0.4, 0.5)) sample_weight <- 10 * c(0.1, 0.2, 0.3, 0.4) m <- metric_one_hot_mean_iou(num_classes = 3L) m$update_state(     y_true = y_true, y_pred = y_pred, sample_weight = sample_weight) ## tf.Tensor( ## [[0 0 6] ##  [3 0 0] ##  [0 0 1]], shape=(3, 3), dtype=int64) m$result() ## tf.Tensor(0.04761905, shape=(), dtype=float32) model %>% compile(     optimizer = 'sgd',     loss = 'mse',     metrics = list(metric_one_hot_mean_iou(num_classes = 3L)))"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/metric_pearson_correlation.html","id":null,"dir":"Reference","previous_headings":"","what":"Calculates the Pearson Correlation Coefficient (PCC). — metric_pearson_correlation","title":"Calculates the Pearson Correlation Coefficient (PCC). — metric_pearson_correlation","text":"Formula:   PCC measures linear relationship true values (y_true) predicted values (y_pred). coefficient ranges -1 1, value 1 implies perfect positive linear correlation, 0 indicates linear correlation, -1 indicates perfect negative linear correlation. metric widely used regression tasks strength linear relationship predictions true labels important evaluation criterion.","code":"loss = mean(l2norm(y_true - mean(y_true) * l2norm(y_pred - mean(y_pred)))"},{"path":"https://keras3.posit.co/dev/reference/metric_pearson_correlation.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Calculates the Pearson Correlation Coefficient (PCC). — metric_pearson_correlation","text":"","code":"metric_pearson_correlation(   y_true,   y_pred,   axis = -1L,   ...,   name = \"pearson_correlation\",   dtype = NULL )"},{"path":"https://keras3.posit.co/dev/reference/metric_pearson_correlation.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Calculates the Pearson Correlation Coefficient (PCC). — metric_pearson_correlation","text":"y_true Tensor true targets. y_pred Tensor predicted targets. axis (Optional) integer tuple integers axis/axes along compute metric. Defaults -1. ... forward/backward compatability. name (Optional) string name metric instance. dtype (Optional) data type metric result.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_pearson_correlation.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Calculates the Pearson Correlation Coefficient (PCC). — metric_pearson_correlation","text":"Usage compile() API:","code":"pcc <- metric_pearson_correlation(axis = -1) y_true <- rbind(c(0, 1, 0.5),                 c(1, 1, 0.2)) y_pred <- rbind(c(0.1, 0.9, 0.5),                 c(1, 0.9, 0.2)) pcc$update_state(y_true, y_pred) pcc$result() ## tf.Tensor(0.99669963, shape=(), dtype=float32) # equivalent operation using R's stats::cor() mean(sapply(1:nrow(y_true), function(i) {   cor(y_true[i, ], y_pred[i, ]) })) ## [1] 0.9966996 model |> compile(   optimizer = 'sgd',   loss = 'mean_squared_error',   metrics = c(keras.metrics.PearsonCorrelation()) )"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/metric_poisson.html","id":null,"dir":"Reference","previous_headings":"","what":"Computes the Poisson metric between y_true and y_pred. — metric_poisson","title":"Computes the Poisson metric between y_true and y_pred. — metric_poisson","text":"Formula:","code":"metric <- y_pred - y_true * log(y_pred)"},{"path":"https://keras3.posit.co/dev/reference/metric_poisson.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes the Poisson metric between y_true and y_pred. — metric_poisson","text":"","code":"metric_poisson(y_true, y_pred, ..., name = \"poisson\", dtype = NULL)"},{"path":"https://keras3.posit.co/dev/reference/metric_poisson.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Computes the Poisson metric between y_true and y_pred. — metric_poisson","text":"y_true Ground truth values. shape = [batch_size, d0, .. dN]. y_pred predicted values. shape = [batch_size, d0, .. dN]. ... forward/backward compatability. name (Optional) string name metric instance. dtype (Optional) data type metric result.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_poisson.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Computes the Poisson metric between y_true and y_pred. — metric_poisson","text":"y_true y_pred missing, Metric instance returned. Metric instance can passed directly compile(metrics = ), used standalone object. See ?Metric example usage. y_true y_pred provided, tensor computed value returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_poisson.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Computes the Poisson metric between y_true and y_pred. — metric_poisson","text":"Standalone usage:         Usage compile() API:","code":"m <- metric_poisson() m$update_state(rbind(c(0, 1), c(0, 0)), rbind(c(1, 1), c(0, 0))) m$result() ## tf.Tensor(0.49999997, shape=(), dtype=float32) m$reset_state() m$update_state(rbind(c(0, 1), c(0, 0)), rbind(c(1, 1), c(0, 0)),                sample_weight = c(1, 0)) m$result() ## tf.Tensor(0.99999994, shape=(), dtype=float32) model %>% compile(   optimizer = 'sgd',   loss = 'mse',   metrics = list(metric_poisson()) )"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/metric_precision.html","id":null,"dir":"Reference","previous_headings":"","what":"Computes the precision of the predictions with respect to the labels. — metric_precision","title":"Computes the precision of the predictions with respect to the labels. — metric_precision","text":"metric creates two local variables, true_positives false_positives used compute precision. value ultimately returned precision, idempotent operation simply divides true_positives sum true_positives false_positives. sample_weight NULL, weights default 1. Use sample_weight 0 mask values. top_k set, calculate precision often average class among top-k classes highest predicted values batch entry correct can found label entry. class_id specified, calculate precision considering entries batch class_id threshold /top-k highest predictions, computing fraction class_id indeed correct label.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_precision.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes the precision of the predictions with respect to the labels. — metric_precision","text":"","code":"metric_precision(   ...,   thresholds = NULL,   top_k = NULL,   class_id = NULL,   name = NULL,   dtype = NULL )"},{"path":"https://keras3.posit.co/dev/reference/metric_precision.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Computes the precision of the predictions with respect to the labels. — metric_precision","text":"... forward/backward compatability. thresholds (Optional) float value, Python list float threshold values [0, 1]. threshold compared prediction values determine truth value predictions (.e., threshold TRUE, FALSE). used loss function sets from_logits=TRUE (.e. sigmoid applied predictions), thresholds set 0. One metric value generated threshold value. neither thresholds top_k set, default calculate precision thresholds=0.5. top_k (Optional) Unset default. int value specifying top-k predictions consider calculating precision. class_id (Optional) Integer class ID want binary metrics. must half-open interval [0, num_classes), num_classes last dimension predictions. name (Optional) string name metric instance. dtype (Optional) data type metric result.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_precision.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Computes the precision of the predictions with respect to the labels. — metric_precision","text":"Metric instance returned. Metric instance can passed directly compile(metrics = ), used standalone object. See ?Metric example usage.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_precision.html","id":"usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes the precision of the predictions with respect to the labels. — metric_precision","text":"Standalone usage:                 Usage compile() API:   Usage loss from_logits=TRUE:","code":"m <- metric_precision() m$update_state(c(0, 1, 1, 1),                c(1, 0, 1, 1)) m$result() |> as.double() |> signif() ## [1] 0.666667 m$reset_state() m$update_state(c(0, 1, 1, 1),                c(1, 0, 1, 1),                sample_weight = c(0, 0, 1, 0)) m$result() |> as.double() |> signif() ## [1] 1 # With top_k=2, it will calculate precision over y_true[1:2] # and y_pred[1:2] m <- metric_precision(top_k = 2) m$update_state(c(0, 0, 1, 1), c(1, 1, 1, 1)) m$result() ## tf.Tensor(0.0, shape=(), dtype=float32) # With top_k=4, it will calculate precision over y_true[1:4] # and y_pred[1:4] m <- metric_precision(top_k = 4) m$update_state(c(0, 0, 1, 1), c(1, 1, 1, 1)) m$result() ## tf.Tensor(0.5, shape=(), dtype=float32) model |> compile(   optimizer = 'sgd',   loss = 'binary_crossentropy',   metrics = list(metric_precision()) ) model |> compile(   optimizer = 'adam',   loss = loss_binary_crossentropy(from_logits = TRUE),   metrics = list(metric_precision(thresholds = 0)) )"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/metric_precision_at_recall.html","id":null,"dir":"Reference","previous_headings":"","what":"Computes best precision where recall is >= specified value. — metric_precision_at_recall","title":"Computes best precision where recall is >= specified value. — metric_precision_at_recall","text":"metric creates four local variables, true_positives, true_negatives, false_positives false_negatives used compute precision given recall. threshold given recall value computed used evaluate corresponding precision. sample_weight NULL, weights default 1. Use sample_weight 0 mask values. class_id specified, calculate precision considering entries batch class_id threshold predictions, computing fraction class_id indeed correct label.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_precision_at_recall.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes best precision where recall is >= specified value. — metric_precision_at_recall","text":"","code":"metric_precision_at_recall(   ...,   recall,   num_thresholds = 200L,   class_id = NULL,   name = NULL,   dtype = NULL )"},{"path":"https://keras3.posit.co/dev/reference/metric_precision_at_recall.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Computes best precision where recall is >= specified value. — metric_precision_at_recall","text":"... forward/backward compatability. recall scalar value range [0, 1]. num_thresholds (Optional) Defaults 200. number thresholds use matching given recall. class_id (Optional) Integer class ID want binary metrics. must half-open interval [0, num_classes), num_classes last dimension predictions. name (Optional) string name metric instance. dtype (Optional) data type metric result.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_precision_at_recall.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Computes best precision where recall is >= specified value. — metric_precision_at_recall","text":"Metric instance returned. Metric instance can passed directly compile(metrics = ), used standalone object. See ?Metric example usage.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_precision_at_recall.html","id":"usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes best precision where recall is >= specified value. — metric_precision_at_recall","text":"Standalone usage:         Usage compile() API:","code":"m <- metric_precision_at_recall(recall = 0.5) m$update_state(c(0,   0,   0,   1,   1),                c(0, 0.3, 0.8, 0.3, 0.8)) m$result() ## tf.Tensor(0.5, shape=(), dtype=float32) m$reset_state() m$update_state(c(0,   0,   0,   1,   1),                c(0, 0.3, 0.8, 0.3, 0.8),                sample_weight = c(2, 2, 2, 1, 1)) m$result() ## tf.Tensor(0.33333334, shape=(), dtype=float32) model |> compile(   optimizer = 'sgd',   loss = 'binary_crossentropy',   metrics = list(metric_precision_at_recall(recall = 0.8)) )"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/metric_r2_score.html","id":null,"dir":"Reference","previous_headings":"","what":"Computes R2 score. — metric_r2_score","title":"Computes R2 score. — metric_r2_score","text":"Formula:   also called coefficient determination. indicates close fitted regression line ground-truth data. highest score possible 1.0. indicates predictors perfectly accounts variation target. score 0.0 indicates predictors account variation target. can also negative model worse random. metric can also compute \"Adjusted R2\" score.","code":"sum_squares_residuals <- sum((y_true - y_pred) ** 2) sum_squares <- sum((y_true - mean(y_true)) ** 2) R2 <- 1 - sum_squares_residuals / sum_squares"},{"path":"https://keras3.posit.co/dev/reference/metric_r2_score.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes R2 score. — metric_r2_score","text":"","code":"metric_r2_score(   ...,   class_aggregation = \"uniform_average\",   num_regressors = 0L,   name = \"r2_score\",   dtype = NULL )"},{"path":"https://keras3.posit.co/dev/reference/metric_r2_score.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Computes R2 score. — metric_r2_score","text":"... forward/backward compatability. class_aggregation Specifies aggregate scores corresponding different output classes (target dimensions), .e. different dimensions last axis predictions. Equivalent multioutput argument Scikit-Learn. one NULL (aggregation), \"uniform_average\", \"variance_weighted_average\". num_regressors Number independent regressors used (\"Adjusted R2\" score). 0 standard R2 score. Defaults 0. name Optional. string name metric instance. dtype Optional. data type metric result.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_r2_score.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Computes R2 score. — metric_r2_score","text":"Metric instance returned. Metric instance can passed directly compile(metrics = ), used standalone object. See ?Metric example usage.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_r2_score.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Computes R2 score. — metric_r2_score","text":"","code":"y_true <- rbind(1, 4, 3) y_pred <- rbind(2, 4, 4) metric <- metric_r2_score() metric$update_state(y_true, y_pred) metric$result() ## tf.Tensor(0.57142854, shape=(), dtype=float32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/metric_recall.html","id":null,"dir":"Reference","previous_headings":"","what":"Computes the recall of the predictions with respect to the labels. — metric_recall","title":"Computes the recall of the predictions with respect to the labels. — metric_recall","text":"metric creates two local variables, true_positives false_negatives, used compute recall. value ultimately returned recall, idempotent operation simply divides true_positives sum true_positives false_negatives. sample_weight NULL, weights default 1. Use sample_weight 0 mask values. top_k set, recall computed often average class among labels batch entry top-k predictions. class_id specified, calculate recall considering entries batch class_id label, computing fraction class_id threshold /top-k predictions.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_recall.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes the recall of the predictions with respect to the labels. — metric_recall","text":"","code":"metric_recall(   ...,   thresholds = NULL,   top_k = NULL,   class_id = NULL,   name = NULL,   dtype = NULL )"},{"path":"https://keras3.posit.co/dev/reference/metric_recall.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Computes the recall of the predictions with respect to the labels. — metric_recall","text":"... forward/backward compatability. thresholds (Optional) float value, Python list float threshold values [0, 1]. threshold compared prediction values determine truth value predictions (.e., threshold TRUE, FALSE). used loss function sets from_logits=TRUE (.e. sigmoid applied predictions), thresholds set 0. One metric value generated threshold value. neither thresholds top_k set, default calculate recall thresholds=0.5. top_k (Optional) Unset default. int value specifying top-k predictions consider calculating recall. class_id (Optional) Integer class ID want binary metrics. must half-open interval [0, num_classes), num_classes last dimension predictions. name (Optional) string name metric instance. dtype (Optional) data type metric result.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_recall.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Computes the recall of the predictions with respect to the labels. — metric_recall","text":"Metric instance returned. Metric instance can passed directly compile(metrics = ), used standalone object. See ?Metric example usage.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_recall.html","id":"usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes the recall of the predictions with respect to the labels. — metric_recall","text":"Standalone usage:         Usage compile() API:   Usage loss from_logits=TRUE:","code":"m <- metric_recall() m$update_state(c(0, 1, 1, 1),                c(1, 0, 1, 1)) m$result() ## tf.Tensor(0.6666667, shape=(), dtype=float32) m$reset_state() m$update_state(c(0, 1, 1, 1),                c(1, 0, 1, 1),                sample_weight = c(0, 0, 1, 0)) m$result() ## tf.Tensor(1.0, shape=(), dtype=float32) model |> compile(   optimizer = 'sgd',   loss = 'binary_crossentropy',   metrics = list(metric_recall()) ) model |> compile(   optimizer = 'adam',   loss = loss_binary_crossentropy(from_logits = TRUE),   metrics = list(metric_recall(thresholds = 0)) )"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/metric_recall_at_precision.html","id":null,"dir":"Reference","previous_headings":"","what":"Computes best recall where precision is >= specified value. — metric_recall_at_precision","title":"Computes best recall where precision is >= specified value. — metric_recall_at_precision","text":"given score-label-distribution required precision might achievable, case 0.0 returned recall. metric creates four local variables, true_positives, true_negatives, false_positives false_negatives used compute recall given precision. threshold given precision value computed used evaluate corresponding recall. sample_weight NULL, weights default 1. Use sample_weight 0 mask values. class_id specified, calculate precision considering entries batch class_id threshold predictions, computing fraction class_id indeed correct label.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_recall_at_precision.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes best recall where precision is >= specified value. — metric_recall_at_precision","text":"","code":"metric_recall_at_precision(   ...,   precision,   num_thresholds = 200L,   class_id = NULL,   name = NULL,   dtype = NULL )"},{"path":"https://keras3.posit.co/dev/reference/metric_recall_at_precision.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Computes best recall where precision is >= specified value. — metric_recall_at_precision","text":"... forward/backward compatability. precision scalar value range [0, 1]. num_thresholds (Optional) Defaults 200. number thresholds use matching given precision. class_id (Optional) Integer class ID want binary metrics. must half-open interval [0, num_classes), num_classes last dimension predictions. name (Optional) string name metric instance. dtype (Optional) data type metric result.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_recall_at_precision.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Computes best recall where precision is >= specified value. — metric_recall_at_precision","text":"Metric instance returned. Metric instance can passed directly compile(metrics = ), used standalone object. See ?Metric example usage.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_recall_at_precision.html","id":"usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes best recall where precision is >= specified value. — metric_recall_at_precision","text":"Standalone usage:         Usage compile() API:","code":"m <- metric_recall_at_precision(precision = 0.8) m$update_state(c(0,   0,   1,   1),                c(0, 0.5, 0.3, 0.9)) m$result() ## tf.Tensor(0.5, shape=(), dtype=float32) m$reset_state() m$update_state(c(0,   0,   1,   1),                c(0, 0.5, 0.3, 0.9),                sample_weight = c(1, 0, 0, 1)) m$result() ## tf.Tensor(1.0, shape=(), dtype=float32) model |> compile(   optimizer = 'sgd',   loss = 'binary_crossentropy',   metrics = list(metric_recall_at_precision(precision = 0.8)) )"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/metric_root_mean_squared_error.html","id":null,"dir":"Reference","previous_headings":"","what":"Computes root mean squared error metric between y_true and y_pred. — metric_root_mean_squared_error","title":"Computes root mean squared error metric between y_true and y_pred. — metric_root_mean_squared_error","text":"Formula:","code":"loss <- sqrt(mean((y_pred - y_true) ^ 2))"},{"path":"https://keras3.posit.co/dev/reference/metric_root_mean_squared_error.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes root mean squared error metric between y_true and y_pred. — metric_root_mean_squared_error","text":"","code":"metric_root_mean_squared_error(   ...,   name = \"root_mean_squared_error\",   dtype = NULL )"},{"path":"https://keras3.posit.co/dev/reference/metric_root_mean_squared_error.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Computes root mean squared error metric between y_true and y_pred. — metric_root_mean_squared_error","text":"... forward/backward compatability. name (Optional) string name metric instance. dtype (Optional) data type metric result.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_root_mean_squared_error.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Computes root mean squared error metric between y_true and y_pred. — metric_root_mean_squared_error","text":"Metric instance returned. Metric instance can passed directly compile(metrics = ), used standalone object. See ?Metric example usage.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_root_mean_squared_error.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Computes root mean squared error metric between y_true and y_pred. — metric_root_mean_squared_error","text":"Standalone usage:         Usage compile() API:","code":"m <- metric_root_mean_squared_error() m$update_state(rbind(c(0, 1), c(0, 0)), rbind(c(1, 1), c(0, 0))) m$result() ## tf.Tensor(0.5, shape=(), dtype=float32) m$reset_state() m$update_state(rbind(c(0, 1), c(0, 0)), rbind(c(1, 1), c(0, 0)),                sample_weight = c(1, 0)) m$result() ## tf.Tensor(0.70710677, shape=(), dtype=float32) model %>% compile(   optimizer = 'sgd',   loss = 'mse',   metrics = list(metric_root_mean_squared_error()))"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/metric_sensitivity_at_specificity.html","id":null,"dir":"Reference","previous_headings":"","what":"Computes best sensitivity where specificity is >= specified value. — metric_sensitivity_at_specificity","title":"Computes best sensitivity where specificity is >= specified value. — metric_sensitivity_at_specificity","text":"Sensitivity measures proportion actual positives correctly identified (tp / (tp + fn)). Specificity measures proportion actual negatives correctly identified (tn / (tn + fp)). metric creates four local variables, true_positives, true_negatives, false_positives false_negatives used compute sensitivity given specificity. threshold given specificity value computed used evaluate corresponding sensitivity. sample_weight NULL, weights default 1. Use sample_weight 0 mask values. class_id specified, calculate precision considering entries batch class_id threshold predictions, computing fraction class_id indeed correct label. additional information specificity sensitivity, see following.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_sensitivity_at_specificity.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes best sensitivity where specificity is >= specified value. — metric_sensitivity_at_specificity","text":"","code":"metric_sensitivity_at_specificity(   ...,   specificity,   num_thresholds = 200L,   class_id = NULL,   name = NULL,   dtype = NULL )"},{"path":"https://keras3.posit.co/dev/reference/metric_sensitivity_at_specificity.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Computes best sensitivity where specificity is >= specified value. — metric_sensitivity_at_specificity","text":"... forward/backward compatability. specificity scalar value range [0, 1]. num_thresholds (Optional) Defaults 200. number thresholds use matching given specificity. class_id (Optional) Integer class ID want binary metrics. must half-open interval [0, num_classes), num_classes last dimension predictions. name (Optional) string name metric instance. dtype (Optional) data type metric result.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_sensitivity_at_specificity.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Computes best sensitivity where specificity is >= specified value. — metric_sensitivity_at_specificity","text":"Metric instance returned. Metric instance can passed directly compile(metrics = ), used standalone object. See ?Metric example usage.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_sensitivity_at_specificity.html","id":"usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes best sensitivity where specificity is >= specified value. — metric_sensitivity_at_specificity","text":"Standalone usage:         Usage compile() API:","code":"m <- metric_sensitivity_at_specificity(specificity = 0.5) m$update_state(c(0,   0,   0,   1,   1),                c(0, 0.3, 0.8, 0.3, 0.8)) m$result() ## tf.Tensor(0.5, shape=(), dtype=float32) m$reset_state() m$update_state(c(0,   0,   0,   1,   1),                c(0, 0.3, 0.8, 0.3, 0.8),                sample_weight = c(1, 1, 2, 2, 1)) m$result() ## tf.Tensor(0.33333334, shape=(), dtype=float32) model |> compile(   optimizer = 'sgd',   loss = 'binary_crossentropy',   metrics = list(metric_sensitivity_at_specificity()) )"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/metric_sparse_categorical_accuracy.html","id":null,"dir":"Reference","previous_headings":"","what":"Calculates how often predictions match integer labels. — metric_sparse_categorical_accuracy","title":"Calculates how often predictions match integer labels. — metric_sparse_categorical_accuracy","text":"can provide logits classes y_pred, since argmax logits probabilities . metric creates two local variables, total count used compute frequency y_pred matches y_true. frequency ultimately returned sparse categorical accuracy: idempotent operation simply divides total count. sample_weight NULL, weights default 1. Use sample_weight 0 mask values.","code":"acc <- sample_weight %*% (y_true == which.max(y_pred))"},{"path":"https://keras3.posit.co/dev/reference/metric_sparse_categorical_accuracy.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Calculates how often predictions match integer labels. — metric_sparse_categorical_accuracy","text":"","code":"metric_sparse_categorical_accuracy(   y_true,   y_pred,   ...,   name = \"sparse_categorical_accuracy\",   dtype = NULL )"},{"path":"https://keras3.posit.co/dev/reference/metric_sparse_categorical_accuracy.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Calculates how often predictions match integer labels. — metric_sparse_categorical_accuracy","text":"y_true Tensor true targets. y_pred Tensor predicted targets. ... forward/backward compatability. name (Optional) string name metric instance. dtype (Optional) data type metric result.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_sparse_categorical_accuracy.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Calculates how often predictions match integer labels. — metric_sparse_categorical_accuracy","text":"y_true y_pred missing, Metric instance returned. Metric instance can passed directly compile(metrics = ), used standalone object. See ?Metric example usage. y_true y_pred provided, tensor computed value returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_sparse_categorical_accuracy.html","id":"usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Calculates how often predictions match integer labels. — metric_sparse_categorical_accuracy","text":"Standalone usage:         Usage compile() API:","code":"m <- metric_sparse_categorical_accuracy() m$update_state(rbind(2L, 1L), rbind(c(0.1, 0.6, 0.3), c(0.05, 0.95, 0))) m$result() ## tf.Tensor(0.5, shape=(), dtype=float32) m$reset_state() m$update_state(rbind(2L, 1L), rbind(c(0.1, 0.6, 0.3), c(0.05, 0.95, 0)),                sample_weight = c(0.7, 0.3)) m$result() ## tf.Tensor(0.3, shape=(), dtype=float32) model |> compile(optimizer = 'sgd',                  loss = 'sparse_categorical_crossentropy',                  metrics = list(metric_sparse_categorical_accuracy()))"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/metric_sparse_categorical_crossentropy.html","id":null,"dir":"Reference","previous_headings":"","what":"Computes the crossentropy metric between the labels and predictions. — metric_sparse_categorical_crossentropy","title":"Computes the crossentropy metric between the labels and predictions. — metric_sparse_categorical_crossentropy","text":"Use crossentropy metric two label classes. expects labels provided integers. want provide labels one-hot encoded, please use metric_categorical_crossentropy() metric instead. num_classes floating point values per feature y_pred single floating point value per feature y_true.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_sparse_categorical_crossentropy.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes the crossentropy metric between the labels and predictions. — metric_sparse_categorical_crossentropy","text":"","code":"metric_sparse_categorical_crossentropy(   y_true,   y_pred,   from_logits = FALSE,   ignore_class = NULL,   axis = -1L,   ...,   name = \"sparse_categorical_crossentropy\",   dtype = NULL )"},{"path":"https://keras3.posit.co/dev/reference/metric_sparse_categorical_crossentropy.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Computes the crossentropy metric between the labels and predictions. — metric_sparse_categorical_crossentropy","text":"y_true Ground truth values. y_pred predicted values. from_logits (Optional) Whether output expected logits tensor. default, consider output encodes probability distribution. ignore_class Optional integer. ID class ignored loss computation. useful, example, segmentation problems featuring \"void\" class (commonly -1 255) segmentation maps. default (ignore_class=NULL), classes considered. axis (Optional) Defaults -1. dimension along entropy computed. ... forward/backward compatability. name (Optional) string name metric instance. dtype (Optional) data type metric result.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_sparse_categorical_crossentropy.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Computes the crossentropy metric between the labels and predictions. — metric_sparse_categorical_crossentropy","text":"y_true y_pred missing, Metric instance returned. Metric instance can passed directly compile(metrics = ), used standalone object. See ?Metric example usage. y_true y_pred provided, tensor computed value returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_sparse_categorical_crossentropy.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Computes the crossentropy metric between the labels and predictions. — metric_sparse_categorical_crossentropy","text":"Standalone usage:         Usage compile() API:","code":"m <- metric_sparse_categorical_crossentropy() m$update_state(array(c(1, 2)),                rbind(c(0.05, 0.95, 0), c(0.1, 0.8, 0.1))) m$result() ## tf.Tensor(1.1769392, shape=(), dtype=float32) m$reset_state() m$update_state(array(c(1, 2)),                rbind(c(0.05, 0.95, 0), c(0.1, 0.8, 0.1)),                sample_weight = c(0.3, 0.7)) m$result() ## tf.Tensor(1.6271976, shape=(), dtype=float32) model |> compile(   optimizer = 'sgd',   loss = 'mse',   metrics = list(metric_sparse_categorical_crossentropy()) )"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/metric_sparse_top_k_categorical_accuracy.html","id":null,"dir":"Reference","previous_headings":"","what":"Computes how often integer targets are in the top K predictions. — metric_sparse_top_k_categorical_accuracy","title":"Computes how often integer targets are in the top K predictions. — metric_sparse_top_k_categorical_accuracy","text":"Computes often integer targets top K predictions. default, arguments expected update_state() : y_true: tensor shape (batch_size) representing indices true categories. y_pred: tensor shape (batch_size, num_categories) containing scores sample possible categories. from_sorted_ids=TRUE, arguments expected update_state : y_true: tensor shape (batch_size) representing indices IDs true categories. y_pred: tensor shape (batch_size, N) containing indices IDs top N categories sorted order highest score lowest score. N must greater equal k. from_sorted_ids=TRUE option can efficient set categories large model optimized way retrieve top ones either without scoring without maintaining scores possible categories.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_sparse_top_k_categorical_accuracy.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes how often integer targets are in the top K predictions. — metric_sparse_top_k_categorical_accuracy","text":"","code":"metric_sparse_top_k_categorical_accuracy(   y_true,   y_pred,   k = 5L,   ...,   name = \"sparse_top_k_categorical_accuracy\",   dtype = NULL,   from_sorted_ids = FALSE )"},{"path":"https://keras3.posit.co/dev/reference/metric_sparse_top_k_categorical_accuracy.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Computes how often integer targets are in the top K predictions. — metric_sparse_top_k_categorical_accuracy","text":"y_true Tensor true targets. y_pred Tensor predicted targets. k (Optional) Number top elements look computing accuracy. Defaults 5. ... forward/backward compatability. name (Optional) string name metric instance. dtype (Optional) data type metric result. from_sorted_ids (Optional) FALSE, default, tensor passed y_pred contains unsorted scores possible categories. TRUE, y_pred contains indices IDs top categories.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_sparse_top_k_categorical_accuracy.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Computes how often integer targets are in the top K predictions. — metric_sparse_top_k_categorical_accuracy","text":"y_true y_pred missing, Metric instance returned. Metric instance can passed directly compile(metrics = ), used standalone object. See ?Metric example usage. y_true y_pred provided, tensor computed value returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_sparse_top_k_categorical_accuracy.html","id":"usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes how often integer targets are in the top K predictions. — metric_sparse_top_k_categorical_accuracy","text":"Standalone usage:             Usage compile() API:","code":"m <- metric_sparse_top_k_categorical_accuracy(k = 1L) m$update_state(   rbind(2, 1),   op_array(rbind(c(0.1, 0.9, 0.8), c(0.05, 0.95, 0)), dtype = \"float32\") ) m$result() ## tf.Tensor(0.5, shape=(), dtype=float32) m$reset_state() m$update_state(   rbind(2, 1),   op_array(rbind(c(0.1, 0.9, 0.8), c(0.05, 0.95, 0)), dtype = \"float32\"),   sample_weight = c(0.7, 0.3) ) m$result() ## tf.Tensor(0.3, shape=(), dtype=float32) m <- metric_sparse_top_k_categorical_accuracy(k = 1, from_sorted_ids = TRUE) m$update_state(array(c(2, 1)), rbind(c(1, 0, 3),                                      c(1, 2, 3))) m$result() ## tf.Tensor(0.5, shape=(), dtype=float32) model %>% compile(optimizer = 'sgd',                   loss = 'sparse_categorical_crossentropy',                   metrics = list(metric_sparse_top_k_categorical_accuracy()))"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/metric_specificity_at_sensitivity.html","id":null,"dir":"Reference","previous_headings":"","what":"Computes best specificity where sensitivity is >= specified value. — metric_specificity_at_sensitivity","title":"Computes best specificity where sensitivity is >= specified value. — metric_specificity_at_sensitivity","text":"Sensitivity measures proportion actual positives correctly identified (tp / (tp + fn)). Specificity measures proportion actual negatives correctly identified (tn / (tn + fp)). metric creates four local variables, true_positives, true_negatives, false_positives false_negatives used compute specificity given sensitivity. threshold given sensitivity value computed used evaluate corresponding specificity. sample_weight NULL, weights default 1. Use sample_weight 0 mask values. class_id specified, calculate precision considering entries batch class_id threshold predictions, computing fraction class_id indeed correct label. additional information specificity sensitivity, see following.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_specificity_at_sensitivity.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes best specificity where sensitivity is >= specified value. — metric_specificity_at_sensitivity","text":"","code":"metric_specificity_at_sensitivity(   ...,   sensitivity,   num_thresholds = 200L,   class_id = NULL,   name = NULL,   dtype = NULL )"},{"path":"https://keras3.posit.co/dev/reference/metric_specificity_at_sensitivity.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Computes best specificity where sensitivity is >= specified value. — metric_specificity_at_sensitivity","text":"... forward/backward compatability. sensitivity scalar value range [0, 1]. num_thresholds (Optional) Defaults 200. number thresholds use matching given sensitivity. class_id (Optional) Integer class ID want binary metrics. must half-open interval [0, num_classes), num_classes last dimension predictions. name (Optional) string name metric instance. dtype (Optional) data type metric result.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_specificity_at_sensitivity.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Computes best specificity where sensitivity is >= specified value. — metric_specificity_at_sensitivity","text":"Metric instance returned. Metric instance can passed directly compile(metrics = ), used standalone object. See ?Metric example usage.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_specificity_at_sensitivity.html","id":"usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes best specificity where sensitivity is >= specified value. — metric_specificity_at_sensitivity","text":"Standalone usage:         Usage compile() API:","code":"m <- metric_specificity_at_sensitivity(sensitivity = 0.5) m$update_state(c(0,   0,   0,   1,   1),                c(0, 0.3, 0.8, 0.3, 0.8)) m$result() ## tf.Tensor(0.6666667, shape=(), dtype=float32) m$reset_state() m$update_state(c(0,   0,   0,   1,   1),                c(0, 0.3, 0.8, 0.3, 0.8),                sample_weight = c(1, 1, 2, 2, 2)) m$result() ## tf.Tensor(0.5, shape=(), dtype=float32) model |> compile(   optimizer = 'sgd',   loss = 'binary_crossentropy',   metrics = list(metric_sensitivity_at_specificity()) )"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/metric_squared_hinge.html","id":null,"dir":"Reference","previous_headings":"","what":"Computes the hinge metric between y_true and y_pred. — metric_squared_hinge","title":"Computes the hinge metric between y_true and y_pred. — metric_squared_hinge","text":"Formula:   y_true values expected -1 1. binary (0 1) labels provided convert -1 1.","code":"loss <- mean(square(maximum(1 - y_true * y_pred, 0)))"},{"path":"https://keras3.posit.co/dev/reference/metric_squared_hinge.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes the hinge metric between y_true and y_pred. — metric_squared_hinge","text":"","code":"metric_squared_hinge(y_true, y_pred, ..., name = \"squared_hinge\", dtype = NULL)"},{"path":"https://keras3.posit.co/dev/reference/metric_squared_hinge.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Computes the hinge metric between y_true and y_pred. — metric_squared_hinge","text":"y_true ground truth values. y_true values expected -1 1. binary (0 1) labels provided convert -1 1 shape = [batch_size, d0, .. dN]. y_pred predicted values shape = [batch_size, d0, .. dN]. ... forward/backward compatability. name (Optional) string name metric instance. dtype (Optional) data type metric result.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_squared_hinge.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Computes the hinge metric between y_true and y_pred. — metric_squared_hinge","text":"y_true y_pred missing, Metric instance returned. Metric instance can passed directly compile(metrics = ), used standalone object. See ?Metric example usage. y_true y_pred provided, tensor computed value returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_squared_hinge.html","id":"usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes the hinge metric between y_true and y_pred. — metric_squared_hinge","text":"Standalone usage:","code":"m <- metric_squared_hinge() m$update_state(rbind(c(0, 1), c(0, 0)), rbind(c(0.6, 0.4), c(0.4, 0.6))) m$result() ## tf.Tensor(1.86, shape=(), dtype=float32) m$reset_state() m$update_state(rbind(c(0, 1), c(0, 0)), rbind(c(0.6, 0.4), c(0.4, 0.6)),                sample_weight = c(1, 0)) m$result() ## tf.Tensor(1.46, shape=(), dtype=float32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/metric_sum.html","id":null,"dir":"Reference","previous_headings":"","what":"Compute the (weighted) sum of the given values. — metric_sum","title":"Compute the (weighted) sum of the given values. — metric_sum","text":"example, values [1, 3, 5, 7] sum 16. sample_weight specified [1, 1, 0, 0] sum 4. metric creates one variable, total. ultimately returned sum value.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_sum.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Compute the (weighted) sum of the given values. — metric_sum","text":"","code":"metric_sum(..., name = \"sum\", dtype = NULL)"},{"path":"https://keras3.posit.co/dev/reference/metric_sum.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Compute the (weighted) sum of the given values. — metric_sum","text":"... forward/backward compatability. name (Optional) string name metric instance. dtype (Optional) data type metric result.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_sum.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Compute the (weighted) sum of the given values. — metric_sum","text":"Metric instance returned. Metric instance can passed directly compile(metrics = ), used standalone object. See ?Metric example usage.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_sum.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Compute the (weighted) sum of the given values. — metric_sum","text":"","code":"m <- metric_sum() m$update_state(c(1, 3, 5, 7)) m$result() ## tf.Tensor(16.0, shape=(), dtype=float32) m <- metric_sum() m$update_state(c(1, 3, 5, 7), sample_weight = c(1, 1, 0, 0)) m$result() ## tf.Tensor(4.0, shape=(), dtype=float32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/metric_top_k_categorical_accuracy.html","id":null,"dir":"Reference","previous_headings":"","what":"Computes how often targets are in the top K predictions. — metric_top_k_categorical_accuracy","title":"Computes how often targets are in the top K predictions. — metric_top_k_categorical_accuracy","text":"Computes often targets top K predictions.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_top_k_categorical_accuracy.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes how often targets are in the top K predictions. — metric_top_k_categorical_accuracy","text":"","code":"metric_top_k_categorical_accuracy(   y_true,   y_pred,   k = 5L,   ...,   name = \"top_k_categorical_accuracy\",   dtype = NULL )"},{"path":"https://keras3.posit.co/dev/reference/metric_top_k_categorical_accuracy.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Computes how often targets are in the top K predictions. — metric_top_k_categorical_accuracy","text":"y_true Tensor true targets. y_pred Tensor predicted targets. k (Optional) Number top elements look computing accuracy. Defaults 5. ... forward/backward compatability. name (Optional) string name metric instance. dtype (Optional) data type metric result.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_top_k_categorical_accuracy.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Computes how often targets are in the top K predictions. — metric_top_k_categorical_accuracy","text":"y_true y_pred missing, Metric instance returned. Metric instance can passed directly compile(metrics = ), used standalone object. See ?Metric example usage. y_true y_pred provided, tensor computed value returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_top_k_categorical_accuracy.html","id":"usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes how often targets are in the top K predictions. — metric_top_k_categorical_accuracy","text":"Standalone usage:         Usage compile() API:","code":"m <- metric_top_k_categorical_accuracy(k = 1) m$update_state(   rbind(c(0, 0, 1), c(0, 1, 0)),   op_array(rbind(c(0.1, 0.9, 0.8), c(0.05, 0.95, 0)), dtype = \"float32\") ) m$result() ## tf.Tensor(0.5, shape=(), dtype=float32) m$reset_state() m$update_state(   rbind(c(0, 0, 1), c(0, 1, 0)),   op_array(rbind(c(0.1, 0.9, 0.8), c(0.05, 0.95, 0)), dtype = \"float32\"),   sample_weight = c(0.7, 0.3)) m$result() ## tf.Tensor(0.3, shape=(), dtype=float32) model |> compile(optimizer = 'sgd',                  loss = 'categorical_crossentropy',                  metrics = list(metric_top_k_categorical_accuracy()))"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/metric_true_negatives.html","id":null,"dir":"Reference","previous_headings":"","what":"Calculates the number of true negatives. — metric_true_negatives","title":"Calculates the number of true negatives. — metric_true_negatives","text":"sample_weight given, calculates sum weights true negatives. metric creates one local variable, accumulator used keep track number true negatives. sample_weight NULL, weights default 1. Use sample_weight 0 mask values.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_true_negatives.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Calculates the number of true negatives. — metric_true_negatives","text":"","code":"metric_true_negatives(..., thresholds = NULL, name = NULL, dtype = NULL)"},{"path":"https://keras3.posit.co/dev/reference/metric_true_negatives.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Calculates the number of true negatives. — metric_true_negatives","text":"... forward/backward compatability. thresholds (Optional) Defaults 0.5. float value, Python list float threshold values [0, 1]. threshold compared prediction values determine truth value predictions (.e., threshold TRUE, FALSE). used loss function sets from_logits=TRUE (.e. sigmoid applied predictions), thresholds set 0. One metric value generated threshold value. name (Optional) string name metric instance. dtype (Optional) data type metric result.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_true_negatives.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Calculates the number of true negatives. — metric_true_negatives","text":"Metric instance returned. Metric instance can passed directly compile(metrics = ), used standalone object. See ?Metric example usage.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_true_negatives.html","id":"usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Calculates the number of true negatives. — metric_true_negatives","text":"Standalone usage:","code":"m <- metric_true_negatives() m$update_state(c(0, 1, 0, 0), c(1, 1, 0, 0)) m$result() ## tf.Tensor(2.0, shape=(), dtype=float32) m$reset_state() m$update_state(c(0, 1, 0, 0), c(1, 1, 0, 0), sample_weight = c(0, 0, 1, 0)) m$result() ## tf.Tensor(1.0, shape=(), dtype=float32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/metric_true_positives.html","id":null,"dir":"Reference","previous_headings":"","what":"Calculates the number of true positives. — metric_true_positives","title":"Calculates the number of true positives. — metric_true_positives","text":"sample_weight given, calculates sum weights true positives. metric creates one local variable, true_positives used keep track number true positives. sample_weight NULL, weights default 1. Use sample_weight 0 mask values.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_true_positives.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Calculates the number of true positives. — metric_true_positives","text":"","code":"metric_true_positives(..., thresholds = NULL, name = NULL, dtype = NULL)"},{"path":"https://keras3.posit.co/dev/reference/metric_true_positives.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Calculates the number of true positives. — metric_true_positives","text":"... forward/backward compatability. thresholds (Optional) Defaults 0.5. float value, Python list float threshold values [0, 1]. threshold compared prediction values determine truth value predictions (.e., threshold TRUE, FALSE). used loss function sets from_logits=TRUE (.e. sigmoid applied predictions), thresholds set 0. One metric value generated threshold value. name (Optional) string name metric instance. dtype (Optional) data type metric result.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_true_positives.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Calculates the number of true positives. — metric_true_positives","text":"Metric instance returned. Metric instance can passed directly compile(metrics = ), used standalone object. See ?Metric example usage.","code":""},{"path":"https://keras3.posit.co/dev/reference/metric_true_positives.html","id":"usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Calculates the number of true positives. — metric_true_positives","text":"Standalone usage:","code":"m <- metric_true_positives() m$update_state(c(0, 1, 1, 1), c(1, 0, 1, 1)) m$result() ## tf.Tensor(2.0, shape=(), dtype=float32) m$reset_state() m$update_state(c(0, 1, 1, 1), c(1, 0, 1, 1), sample_weight = c(0, 0, 1, 0)) m$result() ## tf.Tensor(1.0, shape=(), dtype=float32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/multi-assign.html","id":null,"dir":"Reference","previous_headings":"","what":"Assign values to names — %<-%","title":"Assign values to names — %<-%","text":"See %<-% details.","code":""},{"path":"https://keras3.posit.co/dev/reference/multi-assign.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Assign values to names — %<-%","text":"","code":"x %<-% value"},{"path":"https://keras3.posit.co/dev/reference/multi-assign.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Assign values to names — %<-%","text":"right-hand-side argument, value, invisibly. called primarily side-effect assigning symbols current frame.","code":""},{"path":"https://keras3.posit.co/dev/reference/new_callback_class.html","id":null,"dir":"Reference","previous_headings":"","what":"Callback — new_callback_class","title":"Callback — new_callback_class","text":"new_callback_class() alias Callback(). See ?Callback() full documentation.","code":""},{"path":"https://keras3.posit.co/dev/reference/new_callback_class.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Callback — new_callback_class","text":"","code":"new_callback_class(   classname,   on_epoch_begin = NULL,   on_epoch_end = NULL,   on_train_begin = NULL,   on_train_end = NULL,   on_train_batch_begin = NULL,   on_train_batch_end = NULL,   on_test_begin = NULL,   on_test_end = NULL,   on_test_batch_begin = NULL,   on_test_batch_end = NULL,   on_predict_begin = NULL,   on_predict_end = NULL,   on_predict_batch_begin = NULL,   on_predict_batch_end = NULL,   ...,   public = list(),   private = list(),   inherit = NULL,   parent_env = parent.frame() )"},{"path":"https://keras3.posit.co/dev/reference/new_callback_class.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Callback — new_callback_class","text":"classname String, name custom class. (Conventionally, CamelCase). on_epoch_begin   Called start epoch. Subclasses override actions run. function called TRAIN mode. Args: epoch: Integer, index epoch. logs: Named List. Currently data passed argument method may change future. on_epoch_end   Called end epoch. Subclasses override actions run. function called TRAIN mode. Args: epoch: Integer, index epoch. logs: Named List, metric results training epoch, validation epoch validation performed. Validation result keys prefixed val_. training epoch, values Model's metrics returned. Example: list(loss = 0.2, accuracy = 0.7). on_train_begin   Called beginning training. Subclasses override actions run. Args: logs: Named list. Currently data passed argument method may change future. on_train_end   Called end training. Subclasses override actions run. Args: logs: Named list. Currently output last call on_epoch_end() passed argument method may change future. on_train_batch_begin   Called beginning training batch fit() methods. Subclasses override actions run. Note steps_per_execution argument compile Model set N, method called every N batches. Args: batch: Integer, index batch within current epoch. logs: Named list. Currently data passed argument method may change future. on_train_batch_end   Called end training batch fit() methods. Subclasses override actions run. Note steps_per_execution argument compile Model set N, method called every N batches. Args: batch: Integer, index batch within current epoch. logs: Named list. Aggregated metric results batch. on_test_begin   Called beginning evaluation validation. Subclasses override actions run. Args: logs: Named list. Currently data passed argument method may change future. on_test_end   Called end evaluation validation. Subclasses override actions run. Args: logs: Named list. Currently output last call on_test_batch_end() passed argument method may change future. on_test_batch_begin   Called beginning batch evaluate() methods. Also called beginning validation batch fit() methods, validation data provided. Subclasses override actions run. Note steps_per_execution argument compile() Model set N, method called every N batches. Args: batch: Integer, index batch within current epoch. logs: Named list. Currently data passed argument method may change future. on_test_batch_end   Called end batch evaluate() methods. Also called end validation batch fit() methods, validation data provided. Subclasses override actions run. Note steps_per_execution argument compile() Model set N, method called every N batches. Args: batch: Integer, index batch within current epoch. logs: Named list. Aggregated metric results batch. on_predict_begin   Called beginning prediction. Subclasses override actions run. Args: logs: Named list. Currently data passed argument method may change future. on_predict_end   Called end prediction. Subclasses override actions run. Args: logs: Named list. Currently data passed argument method may change future. on_predict_batch_begin   Called beginning batch predict() methods. Subclasses override actions run. Note steps_per_execution argument compile() Model set N, method called every N batches. Args: batch: Integer, index batch within current epoch. logs: Named list. Currently data passed argument method may change future. on_predict_batch_end   Called end batch predict() methods. Subclasses override actions run. Note steps_per_execution argument compile Model set N, method called every N batches. Args: batch: Integer, index batch within current epoch. logs: Named list. Aggregated metric results batch. ..., public Additional methods public members custom class. private Named list R objects (typically, functions) include instance private environments. private methods symbols scope public methods (See section \"Symbols Scope\"). instance private environment. objects private invisible Keras framework Python runtime. inherit custom class subclass. default, base keras class. parent_env R environment class methods grandparent.","code":"\\(epoch, logs = NULL) \\(epoch, logs = NULL) \\(logs = NULL) \\(logs = NULL) \\(batch, logs = NULL) \\(batch, logs=NULL) \\(logs = NULL) \\(logs = NULL) \\(batch, logs = NULL) \\(batch, logs = NULL) \\(logs = NULL) \\(logs = NULL) \\(batch, logs = NULL) \\(batch, logs = NULL)"},{"path":"https://keras3.posit.co/dev/reference/new_callback_class.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Callback — new_callback_class","text":"function returns custom Callback instances, similar builtin callback functions.","code":""},{"path":"https://keras3.posit.co/dev/reference/new_layer_class.html","id":null,"dir":"Reference","previous_headings":"","what":"Layer — new_layer_class","title":"Layer — new_layer_class","text":"new_layer_class() alias Layer(). See ?Layer() full documentation.","code":""},{"path":"https://keras3.posit.co/dev/reference/new_layer_class.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Layer — new_layer_class","text":"","code":"new_layer_class(   classname,   initialize = NULL,   call = NULL,   build = NULL,   get_config = NULL,   ...,   public = list(),   private = list(),   inherit = NULL,   parent_env = parent.frame() )"},{"path":"https://keras3.posit.co/dev/reference/new_layer_class.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Layer — new_layer_class","text":"classname String, name custom class. (Conventionally, CamelCase). initialize, call, build, get_config Recommended methods implement. See description details sections. ..., public Additional methods public members custom class. private Named list R objects (typically, functions) include instance private environments. private methods symbols scope public methods (See section \"Symbols Scope\"). instance private environment. objects private invisible Keras framework Python runtime. inherit custom class subclass. default, base keras class. parent_env R environment class methods grandparent.","code":""},{"path":"https://keras3.posit.co/dev/reference/new_layer_class.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Layer — new_layer_class","text":"composing layer constructor, similar behavior layer functions like layer_dense(). first argument returned function object, enabling initialize()ing call() layer one step composing layer pipe, like   initialize() layer instance call() , pass missing NULL value object, pass arguments initialize() name.","code":"layer_foo <- Layer(\"Foo\", ....) output <- inputs |> layer_foo() layer <- layer_dense(units = 2, activation = \"relu\") layer <- layer_dense(NULL, 2, activation = \"relu\") layer <- layer_dense(, 2, activation = \"relu\")  # then you can call() the layer in a separate step outputs <- inputs |> layer()"},{"path":"https://keras3.posit.co/dev/reference/new_learning_rate_schedule_class.html","id":null,"dir":"Reference","previous_headings":"","what":"LearningRateSchedule — new_learning_rate_schedule_class","title":"LearningRateSchedule — new_learning_rate_schedule_class","text":"new_learning_rate_schedule_class() alias LearningRateSchedule(). See ?LearningRateSchedule() full documentation.","code":""},{"path":"https://keras3.posit.co/dev/reference/new_learning_rate_schedule_class.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"LearningRateSchedule — new_learning_rate_schedule_class","text":"","code":"new_learning_rate_schedule_class(   classname,   call = NULL,   initialize = NULL,   get_config = NULL,   ...,   public = list(),   private = list(),   inherit = NULL,   parent_env = parent.frame() )"},{"path":"https://keras3.posit.co/dev/reference/new_learning_rate_schedule_class.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"LearningRateSchedule — new_learning_rate_schedule_class","text":"classname String, name custom class. (Conventionally, CamelCase). call, initialize, get_config Recommended methods implement. See description details sections. ..., public Additional methods public members custom class. private Named list R objects (typically, functions) include instance private environments. private methods symbols scope public methods (See section \"Symbols Scope\"). instance private environment. objects private invisible Keras framework Python runtime. inherit custom class subclass. default, base keras class. parent_env R environment class methods grandparent.","code":""},{"path":"https://keras3.posit.co/dev/reference/new_learning_rate_schedule_class.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"LearningRateSchedule — new_learning_rate_schedule_class","text":"function returns LearningRateSchedule instances, similar built-learning_rate_schedule_* family functions.","code":""},{"path":"https://keras3.posit.co/dev/reference/new_loss_class.html","id":null,"dir":"Reference","previous_headings":"","what":"Loss — new_loss_class","title":"Loss — new_loss_class","text":"new_loss_class() alias Loss(). See ?Loss() full documentation.","code":""},{"path":"https://keras3.posit.co/dev/reference/new_loss_class.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Loss — new_loss_class","text":"","code":"new_loss_class(   classname,   call = NULL,   ...,   public = list(),   private = list(),   inherit = NULL,   parent_env = parent.frame() )"},{"path":"https://keras3.posit.co/dev/reference/new_loss_class.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Loss — new_loss_class","text":"classname String, name custom class. (Conventionally, CamelCase). call   Method implemented subclasses: Function contains logic loss calculation using y_true, y_pred. ..., public Additional methods public members custom class. private Named list R objects (typically, functions) include instance private environments. private methods symbols scope public methods (See section \"Symbols Scope\"). instance private environment. objects private invisible Keras framework Python runtime. inherit custom class subclass. default, base keras class. parent_env R environment class methods grandparent.","code":"function(y_true, y_pred)"},{"path":"https://keras3.posit.co/dev/reference/new_loss_class.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Loss — new_loss_class","text":"function returns Loss instances, similar builtin loss functions.","code":""},{"path":"https://keras3.posit.co/dev/reference/new_metric_class.html","id":null,"dir":"Reference","previous_headings":"","what":"Metric — new_metric_class","title":"Metric — new_metric_class","text":"new_metric_class() alias Metric(). See ?Metric() full documentation.","code":""},{"path":"https://keras3.posit.co/dev/reference/new_metric_class.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Metric — new_metric_class","text":"","code":"new_metric_class(   classname,   initialize = NULL,   update_state = NULL,   result = NULL,   ...,   public = list(),   private = list(),   inherit = NULL,   parent_env = parent.frame() )"},{"path":"https://keras3.posit.co/dev/reference/new_metric_class.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Metric — new_metric_class","text":"classname String, name custom class. (Conventionally, CamelCase). initialize, update_state, result Recommended methods implement. See description section. ..., public Additional methods public members custom class. private Named list R objects (typically, functions) include instance private environments. private methods symbols scope public methods (See section \"Symbols Scope\"). instance private environment. objects private invisible Keras framework Python runtime. inherit custom class subclass. default, base keras class. parent_env R environment class methods grandparent.","code":""},{"path":"https://keras3.posit.co/dev/reference/new_metric_class.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Metric — new_metric_class","text":"function returns Metric instances, similar builtin metric functions.","code":""},{"path":"https://keras3.posit.co/dev/reference/new_model_class.html","id":null,"dir":"Reference","previous_headings":"","what":"Model — new_model_class","title":"Model — new_model_class","text":"new_model_class() alias Model(). See ?Model() full documentation.","code":""},{"path":"https://keras3.posit.co/dev/reference/new_model_class.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Model — new_model_class","text":"","code":"new_model_class(   classname,   initialize = NULL,   call = NULL,   train_step = NULL,   predict_step = NULL,   test_step = NULL,   compute_loss = NULL,   compute_metrics = NULL,   ...,   public = list(),   private = list(),   inherit = NULL,   parent_env = parent.frame() )"},{"path":"https://keras3.posit.co/dev/reference/new_model_class.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Model — new_model_class","text":"classname String, name custom class. (Conventionally, CamelCase). initialize, call, train_step, predict_step, test_step, compute_loss, compute_metrics Optional methods can overridden. ..., public Additional methods public members custom class. private Named list R objects (typically, functions) include instance private environments. private methods symbols scope public methods (See section \"Symbols Scope\"). instance private environment. objects private invisible Keras framework Python runtime. inherit custom class subclass. default, base keras class. parent_env R environment class methods grandparent.","code":""},{"path":"https://keras3.posit.co/dev/reference/new_model_class.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Model — new_model_class","text":"model constructor function, can call create instance new model type.","code":""},{"path":"https://keras3.posit.co/dev/reference/newaxis.html","id":null,"dir":"Reference","previous_headings":"","what":"New axis — newaxis","title":"New axis — newaxis","text":"alias NULL. meant used [ tensors, expand dimensions tensor","code":""},{"path":"https://keras3.posit.co/dev/reference/newaxis.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"New axis — newaxis","text":"","code":"newaxis"},{"path":"https://keras3.posit.co/dev/reference/newaxis.html","id":"format","dir":"Reference","previous_headings":"","what":"Format","title":"New axis — newaxis","text":"object class NULL length 0.","code":""},{"path":"https://keras3.posit.co/dev/reference/newaxis.html","id":"details","dir":"Reference","previous_headings":"","what":"Details","title":"New axis — newaxis","text":"","code":"x <- op_convert_to_tensor(1:10)  op_shape(x) op_shape(x[]) op_shape(x[newaxis]) op_shape(x@py[newaxis]) op_shape(x@r[newaxis])  op_shape(x[newaxis, .., newaxis]) op_shape(x@py[newaxis, .., newaxis]) op_shape(x@r[newaxis, .., newaxis])"},{"path":"https://keras3.posit.co/dev/reference/normalize.html","id":null,"dir":"Reference","previous_headings":"","what":"Normalizes an array. — normalize","title":"Normalizes an array. — normalize","text":"input R array, R array returned. backend tensor, backend tensor returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/normalize.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Normalizes an array. — normalize","text":"","code":"normalize(x, axis = -1L, order = 2L)"},{"path":"https://keras3.posit.co/dev/reference/normalize.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Normalizes an array. — normalize","text":"x Array normalize. axis axis along normalize. order Normalization order (e.g. order=2 L2 norm).","code":""},{"path":"https://keras3.posit.co/dev/reference/normalize.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Normalizes an array. — normalize","text":"normalized copy array.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_abs.html","id":null,"dir":"Reference","previous_headings":"","what":"Compute the absolute value element-wise. — op_abs","title":"Compute the absolute value element-wise. — op_abs","text":"Compute absolute value element-wise.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_abs.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Compute the absolute value element-wise. — op_abs","text":"","code":"op_abs(x)"},{"path":"https://keras3.posit.co/dev/reference/op_abs.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Compute the absolute value element-wise. — op_abs","text":"x Input tensor","code":""},{"path":"https://keras3.posit.co/dev/reference/op_abs.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Compute the absolute value element-wise. — op_abs","text":"array containing absolute value element x.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_abs.html","id":"example","dir":"Reference","previous_headings":"","what":"Example","title":"Compute the absolute value element-wise. — op_abs","text":"","code":"x <- op_convert_to_tensor(c(-1.2, 1.2)) op_abs(x) ## tf.Tensor([1.2 1.2], shape=(2), dtype=float64)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_add.html","id":null,"dir":"Reference","previous_headings":"","what":"Add arguments element-wise. — op_add","title":"Add arguments element-wise. — op_add","text":"Add arguments element-wise.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_add.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Add arguments element-wise. — op_add","text":"","code":"op_add(x1, x2)"},{"path":"https://keras3.posit.co/dev/reference/op_add.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Add arguments element-wise. — op_add","text":"x1 First input tensor. x2 Second input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_add.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Add arguments element-wise. — op_add","text":"tensor containing element-wise sum x1 x2.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_add.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Add arguments element-wise. — op_add","text":"op_add also broadcasts shapes:     Note function automatically called using R operator + tensors.","code":"x1 <- op_convert_to_tensor(c(1, 4)) x2 <- op_convert_to_tensor(c(5, 6)) op_add(x1, x2) ## tf.Tensor([ 6. 10.], shape=(2), dtype=float64) # alias for x1 + x2 x1 + x2 ## tf.Tensor([ 6. 10.], shape=(2), dtype=float64) x1 <- op_convert_to_tensor(array(c(5, 5, 4, 6), dim =c(2, 2))) x2 <- op_convert_to_tensor(c(5, 6)) op_add(x1, x2) ## tf.Tensor( ## [[10. 10.] ##  [10. 12.]], shape=(2, 2), dtype=float64) x <- op_ones(c(3)) op_add(x, x) ## tf.Tensor([2. 2. 2.], shape=(3), dtype=float32) x + x ## tf.Tensor([2. 2. 2.], shape=(3), dtype=float32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_all.html","id":null,"dir":"Reference","previous_headings":"","what":"Test whether all array elements along a given axis evaluate to TRUE. — op_all","title":"Test whether all array elements along a given axis evaluate to TRUE. — op_all","text":"Test whether array elements along given axis evaluate TRUE.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_all.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Test whether all array elements along a given axis evaluate to TRUE. — op_all","text":"","code":"op_all(x, axis = NULL, keepdims = FALSE)"},{"path":"https://keras3.posit.co/dev/reference/op_all.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Test whether all array elements along a given axis evaluate to TRUE. — op_all","text":"x Input tensor. axis integer tuple integers represent axis along logical reduction performed. default (axis = NULL) perform logical dimensions input array. axis may negative, case counts last first axis. keepdims TRUE, axes reduced left result dimensions size one. option, result broadcast correctly input array. Defaults FALSE.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_all.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Test whether all array elements along a given axis evaluate to TRUE. — op_all","text":"tensor containing logical reduction axis.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_all.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Test whether all array elements along a given axis evaluate to TRUE. — op_all","text":"keepdims = TRUE outputs tensor dimensions reduced one.","code":"x <- op_convert_to_tensor(c(TRUE, FALSE)) op_all(x) ## tf.Tensor(False, shape=(), dtype=bool) (x <- op_convert_to_tensor(array(c(TRUE, FALSE, TRUE, TRUE, TRUE, TRUE), dim = c(3, 2)))) ## tf.Tensor( ## [[ True  True] ##  [False  True] ##  [ True  True]], shape=(3, 2), dtype=bool) op_all(x, axis = 1) ## tf.Tensor([False  True], shape=(2), dtype=bool) op_all(x, keepdims = TRUE) ## tf.Tensor([[False]], shape=(1, 1), dtype=bool)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_angle.html","id":null,"dir":"Reference","previous_headings":"","what":"Element-wise angle of a complex tensor. — op_angle","title":"Element-wise angle of a complex tensor. — op_angle","text":"Returns phase angle (radians) element x.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_angle.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Element-wise angle of a complex tensor. — op_angle","text":"","code":"op_angle(x)"},{"path":"https://keras3.posit.co/dev/reference/op_angle.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Element-wise angle of a complex tensor. — op_angle","text":"x Input tensor. Can real complex.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_angle.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Element-wise angle of a complex tensor. — op_angle","text":"Tensor shape x, containing element-wise angles.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_angle.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Element-wise angle of a complex tensor. — op_angle","text":"","code":"x <- op_convert_to_tensor(matrix(c(1 + 3i, 2 - 5i, 4 - 3i, 3 + 2i), nrow = 2)) op_angle(x) ## tf.Tensor( ## [[ 1.24904577 -0.64350111] ##  [-1.19028995  0.5880026 ]], shape=(2, 2), dtype=float64)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_any.html","id":null,"dir":"Reference","previous_headings":"","what":"Test whether any array element along a given axis evaluates to TRUE. — op_any","title":"Test whether any array element along a given axis evaluates to TRUE. — op_any","text":"Test whether array element along given axis evaluates TRUE.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_any.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Test whether any array element along a given axis evaluates to TRUE. — op_any","text":"","code":"op_any(x, axis = NULL, keepdims = FALSE)"},{"path":"https://keras3.posit.co/dev/reference/op_any.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Test whether any array element along a given axis evaluates to TRUE. — op_any","text":"x Input tensor. axis integer tuple integers represent axis along logical reduction performed. default (axis = NULL) perform logical dimensions input array. axis may negative, case counts last first axis. keepdims TRUE, axes reduced left result dimensions size one. option, result broadcast correctly input array. Defaults FALSE.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_any.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Test whether any array element along a given axis evaluates to TRUE. — op_any","text":"tensor containing logical reduction axis.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_any.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Test whether any array element along a given axis evaluates to TRUE. — op_any","text":"keepdims = TRUE outputs tensor dimensions reduced one.","code":"x <- op_array(c(TRUE, FALSE)) op_any(x) ## tf.Tensor(True, shape=(), dtype=bool) (x <- op_reshape(c(FALSE, FALSE, FALSE,                    TRUE, FALSE, FALSE),                  c(2, 3))) ## tf.Tensor( ## [[False False False] ##  [ True False False]], shape=(2, 3), dtype=bool) op_any(x, axis = 1) ## tf.Tensor([ True False False], shape=(3), dtype=bool) op_any(x, axis = 2) ## tf.Tensor([False  True], shape=(2), dtype=bool) op_any(x, axis = -1) ## tf.Tensor([False  True], shape=(2), dtype=bool) op_any(x, keepdims = TRUE) ## tf.Tensor([[ True]], shape=(1, 1), dtype=bool) op_any(x, 1, keepdims = TRUE) ## tf.Tensor([[ True False False]], shape=(1, 3), dtype=bool)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_append.html","id":null,"dir":"Reference","previous_headings":"","what":"Append tensor x2 to the end of tensor x1. — op_append","title":"Append tensor x2 to the end of tensor x1. — op_append","text":"Append tensor x2 end tensor x1.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_append.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Append tensor x2 to the end of tensor x1. — op_append","text":"","code":"op_append(x1, x2, axis = NULL)"},{"path":"https://keras3.posit.co/dev/reference/op_append.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Append tensor x2 to the end of tensor x1. — op_append","text":"x1 First input tensor. x2 Second input tensor. axis Axis along tensor x2 appended tensor x1. NULL, tensors flattened use.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_append.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Append tensor x2 to the end of tensor x1. — op_append","text":"tensor values x2 appended x1.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_append.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Append tensor x2 to the end of tensor x1. — op_append","text":"axis specified, x1 x2 must compatible shapes.","code":"x1 <- op_convert_to_tensor(c(1, 2, 3)) x2 <- op_convert_to_tensor(rbind(c(4, 5, 6), c(7, 8, 9))) op_append(x1, x2) ## tf.Tensor([1. 2. 3. 4. 5. 6. 7. 8. 9.], shape=(9), dtype=float64) x1 <- op_convert_to_tensor(rbind(c(1, 2, 3), c(4, 5, 6))) x2 <- op_convert_to_tensor(rbind(c(7, 8, 9))) op_append(x1, x2, axis = 1) ## tf.Tensor( ## [[1. 2. 3.] ##  [4. 5. 6.] ##  [7. 8. 9.]], shape=(3, 3), dtype=float64) x3 <- op_convert_to_tensor(c(7, 8, 9)) try(op_append(x1, x3, axis = 1)) ## Error in py_call_impl(callable, call_args$unnamed, call_args$named) : ##   tensorflow.python.framework.errors_impl.InvalidArgumentError: {{function_node __wrapped__ConcatV2_N_2_device_/job:localhost/replica:0/task:0/device:CPU:0}} ConcatOp : Ranks of all input tensors should match: shape[0] = [2,3] vs. shape[1] = [3] [Op:ConcatV2] name: concat"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_arange.html","id":null,"dir":"Reference","previous_headings":"","what":"Return evenly spaced values within a given interval. — op_arange","title":"Return evenly spaced values within a given interval. — op_arange","text":"arange can called varying number positional arguments: arange(end): Values generated within half-open interval [1, end] (words, interval including start end). arange(start, end): Values generated within interval [start, end]. arange(start, end, step): Values generated within interval [start, end], spacing values given step.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_arange.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Return evenly spaced values within a given interval. — op_arange","text":"","code":"op_arange(start, end, step = 1L, dtype = NULL, include_end = TRUE)"},{"path":"https://keras3.posit.co/dev/reference/op_arange.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Return evenly spaced values within a given interval. — op_arange","text":"start Integer real, representing start interval. interval includes value. end Integer real, representing end interval. include_end FALSE, interval include value, except cases step integer floating point round-affects length . Defaults NULL. step Integer real, represent spacing values. output , distance two adjacent values, [+1] - []. default step size 1. step specified position argument, start must also given. dtype type output array. dtype given, infer data type input arguments. include_end TRUE FALSE. FALSE, end included output sequence.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_arange.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Return evenly spaced values within a given interval. — op_arange","text":"Tensor evenly spaced values. floating point arguments, length result ceiling((end - start)/step). floating point overflow, rule may result last element greater end.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_arange.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Return evenly spaced values within a given interval. — op_arange","text":"","code":"op_arange(3L) ## tf.Tensor([1 2 3], shape=(3), dtype=int32) op_arange(3) # float ## tf.Tensor([1. 2. 3.], shape=(3), dtype=float32) op_arange(3, dtype = 'int32') #int ## tf.Tensor([1 2 3], shape=(3), dtype=int32) op_arange(3L, 7L) ## tf.Tensor([3 4 5 6 7], shape=(5), dtype=int32) op_arange(3L, 7L, 2L) ## tf.Tensor([3 5 7], shape=(3), dtype=int32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_arccos.html","id":null,"dir":"Reference","previous_headings":"","what":"Trigonometric inverse cosine, element-wise. — op_arccos","title":"Trigonometric inverse cosine, element-wise. — op_arccos","text":"inverse cos , y = cos(x), x = arccos(y).","code":""},{"path":"https://keras3.posit.co/dev/reference/op_arccos.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Trigonometric inverse cosine, element-wise. — op_arccos","text":"","code":"op_arccos(x)"},{"path":"https://keras3.posit.co/dev/reference/op_arccos.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Trigonometric inverse cosine, element-wise. — op_arccos","text":"x Input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_arccos.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Trigonometric inverse cosine, element-wise. — op_arccos","text":"Tensor angle ray intersecting unit circle given x-coordinate radians [0, pi].","code":""},{"path":"https://keras3.posit.co/dev/reference/op_arccos.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Trigonometric inverse cosine, element-wise. — op_arccos","text":"","code":"x <- op_convert_to_tensor(c(1, -1)) op_arccos(x) ## tf.Tensor([0.         3.14159265], shape=(2), dtype=float64)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_arccosh.html","id":null,"dir":"Reference","previous_headings":"","what":"Inverse hyperbolic cosine, element-wise. — op_arccosh","title":"Inverse hyperbolic cosine, element-wise. — op_arccosh","text":"Inverse hyperbolic cosine, element-wise.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_arccosh.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Inverse hyperbolic cosine, element-wise. — op_arccosh","text":"","code":"op_arccosh(x)"},{"path":"https://keras3.posit.co/dev/reference/op_arccosh.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Inverse hyperbolic cosine, element-wise. — op_arccosh","text":"x Input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_arccosh.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Inverse hyperbolic cosine, element-wise. — op_arccosh","text":"Output tensor shape x.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_arccosh.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Inverse hyperbolic cosine, element-wise. — op_arccosh","text":"","code":"x <- op_convert_to_tensor(c(10, 100)) op_arccosh(x) ## tf.Tensor([2.99322285 5.29829237], shape=(2), dtype=float64)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_arcsin.html","id":null,"dir":"Reference","previous_headings":"","what":"Inverse sine, element-wise. — op_arcsin","title":"Inverse sine, element-wise. — op_arcsin","text":"Inverse sine, element-wise.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_arcsin.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Inverse sine, element-wise. — op_arcsin","text":"","code":"op_arcsin(x)"},{"path":"https://keras3.posit.co/dev/reference/op_arcsin.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Inverse sine, element-wise. — op_arcsin","text":"x Input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_arcsin.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Inverse sine, element-wise. — op_arcsin","text":"Tensor inverse sine element x, radians closed interval [-pi/2, pi/2].","code":""},{"path":"https://keras3.posit.co/dev/reference/op_arcsin.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Inverse sine, element-wise. — op_arcsin","text":"","code":"x <- op_convert_to_tensor(c(1, -1, 0)) op_arcsin(x) ## tf.Tensor([ 1.57079633 -1.57079633  0.        ], shape=(3), dtype=float64)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_arcsinh.html","id":null,"dir":"Reference","previous_headings":"","what":"Inverse hyperbolic sine, element-wise. — op_arcsinh","title":"Inverse hyperbolic sine, element-wise. — op_arcsinh","text":"Inverse hyperbolic sine, element-wise.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_arcsinh.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Inverse hyperbolic sine, element-wise. — op_arcsinh","text":"","code":"op_arcsinh(x)"},{"path":"https://keras3.posit.co/dev/reference/op_arcsinh.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Inverse hyperbolic sine, element-wise. — op_arcsinh","text":"x Input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_arcsinh.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Inverse hyperbolic sine, element-wise. — op_arcsinh","text":"Output tensor shape x.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_arcsinh.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Inverse hyperbolic sine, element-wise. — op_arcsinh","text":"","code":"x <- op_convert_to_tensor(c(1, -1, 0)) op_arcsinh(x) ## tf.Tensor([ 0.88137359 -0.88137359  0.        ], shape=(3), dtype=float64)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_arctan.html","id":null,"dir":"Reference","previous_headings":"","what":"Trigonometric inverse tangent, element-wise. — op_arctan","title":"Trigonometric inverse tangent, element-wise. — op_arctan","text":"Trigonometric inverse tangent, element-wise.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_arctan.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Trigonometric inverse tangent, element-wise. — op_arctan","text":"","code":"op_arctan(x)"},{"path":"https://keras3.posit.co/dev/reference/op_arctan.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Trigonometric inverse tangent, element-wise. — op_arctan","text":"x Input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_arctan.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Trigonometric inverse tangent, element-wise. — op_arctan","text":"Tensor inverse tangent element x, interval [-pi/2, pi/2].","code":""},{"path":"https://keras3.posit.co/dev/reference/op_arctan.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Trigonometric inverse tangent, element-wise. — op_arctan","text":"","code":"x <- op_convert_to_tensor(c(0, 1)) op_arctan(x) ## tf.Tensor([0.         0.78539816], shape=(2), dtype=float64)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_arctan2.html","id":null,"dir":"Reference","previous_headings":"","what":"Element-wise arc tangent of x1/x2 choosing the quadrant correctly. — op_arctan2","title":"Element-wise arc tangent of x1/x2 choosing the quadrant correctly. — op_arctan2","text":"quadrant (.e., branch) chosen arctan2(x1, x2) signed angle radians ray ending origin passing point (1, 0), ray ending origin passing point (x2, x1). (Note role reversal: \"y-coordinate\" first function parameter, \"x-coordinate\" second.) IEEE convention, function defined x2 = +/-0 either x1 x2 = +/-inf.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_arctan2.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Element-wise arc tangent of x1/x2 choosing the quadrant correctly. — op_arctan2","text":"","code":"op_arctan2(x1, x2)"},{"path":"https://keras3.posit.co/dev/reference/op_arctan2.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Element-wise arc tangent of x1/x2 choosing the quadrant correctly. — op_arctan2","text":"x1 First input tensor. x2 Second input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_arctan2.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Element-wise arc tangent of x1/x2 choosing the quadrant correctly. — op_arctan2","text":"Tensor angles radians, range [-pi, pi].","code":""},{"path":"https://keras3.posit.co/dev/reference/op_arctan2.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Element-wise arc tangent of x1/x2 choosing the quadrant correctly. — op_arctan2","text":"Consider four points different quadrants:     Note order parameters. arctan2 defined also x2 = 0 several points, obtaining values range [-pi, pi]:","code":"x <- op_array(c(-1, 1, 1, -1)) y <- op_array(c(-1, -1, 1, 1)) op_arctan2(y, x) * 180 / pi ## tf.Tensor([-135.  -45.   45.  135.], shape=(4), dtype=float32) op_arctan2(     op_array(c(1, -1)),     op_array(c(0, 0)) ) ## tf.Tensor([ 1.5707964 -1.5707964], shape=(2), dtype=float32) op_arctan2(     op_array(c(0, 0, Inf)),     op_array(c(+0, -0, Inf)) ) ## tf.Tensor([0.        3.1415927 0.7853982], shape=(3), dtype=float32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_arctanh.html","id":null,"dir":"Reference","previous_headings":"","what":"Inverse hyperbolic tangent, element-wise. — op_arctanh","title":"Inverse hyperbolic tangent, element-wise. — op_arctanh","text":"Inverse hyperbolic tangent, element-wise.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_arctanh.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Inverse hyperbolic tangent, element-wise. — op_arctanh","text":"","code":"op_arctanh(x)"},{"path":"https://keras3.posit.co/dev/reference/op_arctanh.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Inverse hyperbolic tangent, element-wise. — op_arctanh","text":"x Input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_arctanh.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Inverse hyperbolic tangent, element-wise. — op_arctanh","text":"Output tensor shape x.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_argmax.html","id":null,"dir":"Reference","previous_headings":"","what":"Returns the indices of the maximum values along an axis. — op_argmax","title":"Returns the indices of the maximum values along an axis. — op_argmax","text":"Returns indices maximum values along axis.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_argmax.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Returns the indices of the maximum values along an axis. — op_argmax","text":"","code":"op_argmax(x, axis = NULL, keepdims = FALSE, zero_indexed = FALSE)"},{"path":"https://keras3.posit.co/dev/reference/op_argmax.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Returns the indices of the maximum values along an axis. — op_argmax","text":"x Input tensor. axis default, index flattened tensor, otherwise along specified axis. keepdims set TRUE, axes reduced left result dimensions size one. Defaults FALSE. zero_indexed TRUE, returned indices zero-based (0 encodes first position); FALSE (default), returned indices one-based (1 encodes first position).","code":""},{"path":"https://keras3.posit.co/dev/reference/op_argmax.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Returns the indices of the maximum values along an axis. — op_argmax","text":"Tensor indices. shape x, dimension along axis removed. Note returned integer 1-based (.e., argmax first index position, returned value 1)","code":""},{"path":"https://keras3.posit.co/dev/reference/op_argmax.html","id":"note","dir":"Reference","previous_headings":"","what":"Note","title":"Returns the indices of the maximum values along an axis. — op_argmax","text":"similar R max.col(x) case 2-d array (matrix), nd-array, apply(x, axis, .max)","code":""},{"path":"https://keras3.posit.co/dev/reference/op_argmax.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Returns the indices of the maximum values along an axis. — op_argmax","text":"","code":"x <- op_array(rbind(c(11, 12, 10),                     c(12, 11, 10))) x ## tf.Tensor( ## [[11. 12. 10.] ##  [12. 11. 10.]], shape=(2, 3), dtype=float64) op_argmax(x) ## tf.Tensor(2, shape=(), dtype=int32) op_argmax(x, axis = 1) ## tf.Tensor([2 1 1], shape=(3), dtype=int32) op_argmax(x, axis = 2) ## tf.Tensor([2 1], shape=(2), dtype=int32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_argmin.html","id":null,"dir":"Reference","previous_headings":"","what":"Returns the indices of the minimum values along an axis. — op_argmin","title":"Returns the indices of the minimum values along an axis. — op_argmin","text":"Returns indices minimum values along axis.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_argmin.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Returns the indices of the minimum values along an axis. — op_argmin","text":"","code":"op_argmin(x, axis = NULL, keepdims = FALSE, zero_indexed = FALSE)"},{"path":"https://keras3.posit.co/dev/reference/op_argmin.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Returns the indices of the minimum values along an axis. — op_argmin","text":"x Input tensor. axis default, index flattened tensor, otherwise along specified axis. keepdims set TRUE, axes reduced left result dimensions size one. Defaults FALSE. zero_indexed TRUE, returned indices zero-based (0 encodes first position); FALSE (default), returned indices one-based (1 encodes first position).","code":""},{"path":"https://keras3.posit.co/dev/reference/op_argmin.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Returns the indices of the minimum values along an axis. — op_argmin","text":"Tensor indices. shape x, dimension along axis removed.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_argmin.html","id":"note","dir":"Reference","previous_headings":"","what":"Note","title":"Returns the indices of the minimum values along an axis. — op_argmin","text":"similar R expression apply(x, axis, .min), x R array.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_argmin.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Returns the indices of the minimum values along an axis. — op_argmin","text":"","code":"x <- op_arange(6L) |> op_reshape(c(2, 3)) |> op_add(10) x ## tf.Tensor( ## [[11. 12. 13.] ##  [14. 15. 16.]], shape=(2, 3), dtype=float32) op_argmin(x) ## tf.Tensor(1, shape=(), dtype=int32) op_argmin(x, axis = 1) ## tf.Tensor([1 1 1], shape=(3), dtype=int32) op_argmin(x, axis = 2) ## tf.Tensor([1 1], shape=(2), dtype=int32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_argpartition.html","id":null,"dir":"Reference","previous_headings":"","what":"Performs an indirect partition along the given axis. — op_argpartition","title":"Performs an indirect partition along the given axis. — op_argpartition","text":"returns array indices shape x index data along given axis partitioned order.","code":"x <- op_convert_to_tensor(c(9, 3, 6, 2, 8, 5, 7, 1, 10, 4)) x@r[op_argpartition(x, 3)] ## tf.Tensor([ 1.  2.  3. 10.  9.  8.  7.  6.  5.  4.], shape=(10), dtype=float64) x@r[op_argpartition(x, 5)] ## tf.Tensor([ 1.  2.  3.  4.  5. 10.  9.  8.  7.  6.], shape=(10), dtype=float64) x@r[op_argpartition(x, 7)] ## tf.Tensor([ 1.  2.  3.  4.  5.  6.  7. 10.  9.  8.], shape=(10), dtype=float64)"},{"path":"https://keras3.posit.co/dev/reference/op_argpartition.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Performs an indirect partition along the given axis. — op_argpartition","text":"","code":"op_argpartition(x, kth, axis = -1L, zero_indexed = FALSE)"},{"path":"https://keras3.posit.co/dev/reference/op_argpartition.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Performs an indirect partition along the given axis. — op_argpartition","text":"x Array sort. kth Element index partition . k-th element final sorted position smaller elements moved larger elements behind . order elements partitions undefined. provided sequence k-th partition sorted position . axis Axis along sort. default -1 (last axis). NULL, flattened array used. zero_indexed TRUE, returned indices zero-based (0 encodes first position); FALSE (default), returned indices one-based (1 encodes first position).","code":""},{"path":"https://keras3.posit.co/dev/reference/op_argpartition.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Performs an indirect partition along the given axis. — op_argpartition","text":"Array indices partition x along specified axis.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_argsort.html","id":null,"dir":"Reference","previous_headings":"","what":"Returns the indices that would sort a tensor. — op_argsort","title":"Returns the indices that would sort a tensor. — op_argsort","text":"Returns indices sort tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_argsort.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Returns the indices that would sort a tensor. — op_argsort","text":"","code":"op_argsort(x, axis = -1L, zero_indexed = FALSE)"},{"path":"https://keras3.posit.co/dev/reference/op_argsort.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Returns the indices that would sort a tensor. — op_argsort","text":"x Input tensor. axis Axis along sort. Defaults -1 (last axis). NULL, flattened tensor used. zero_indexed TRUE, returned indices zero-based (0 encodes first position); FALSE (default), returned indices one-based (1 encodes first position).","code":""},{"path":"https://keras3.posit.co/dev/reference/op_argsort.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Returns the indices that would sort a tensor. — op_argsort","text":"Tensor indices sort x along specified axis.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_argsort.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Returns the indices that would sort a tensor. — op_argsort","text":"One dimensional array:         Two-dimensional array:","code":"x <- op_array(c(3, 1, 2)) + .1 op_argsort(x) ## tf.Tensor([2 3 1], shape=(3), dtype=int32) x@r[op_argsort(x)] == op_sort(x) ## tf.Tensor([ True  True  True], shape=(3), dtype=bool) x <- op_array(rbind(c(0, 3),                     c(3, 2),                     c(4, 5))) + .1 (i <- op_argsort(x, axis = 1)) ## tf.Tensor( ## [[1 2] ##  [2 1] ##  [3 3]], shape=(3, 2), dtype=int32) x@r[i@r[, 1], ] # sort x-rows using first col of x ## tf.Tensor( ## [[0.1 3.1] ##  [3.1 2.1] ##  [4.1 5.1]], shape=(3, 2), dtype=float64) x@r[i@r[, 2], ] # sort x-rows using second col of x ## tf.Tensor( ## [[3.1 2.1] ##  [0.1 3.1] ##  [4.1 5.1]], shape=(3, 2), dtype=float64) (i <- op_argsort(x, axis = 2)) ## tf.Tensor( ## [[1 2] ##  [2 1] ##  [1 2]], shape=(3, 2), dtype=int32) x@r[, i@r[2,]] # sort x-cols using second row of x ## tf.Tensor( ## [[3.1 0.1] ##  [2.1 3.1] ##  [5.1 4.1]], shape=(3, 2), dtype=float64)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_array.html","id":null,"dir":"Reference","previous_headings":"","what":"Create a tensor. — op_array","title":"Create a tensor. — op_array","text":"Create tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_array.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Create a tensor. — op_array","text":"","code":"op_array(x, dtype = NULL)"},{"path":"https://keras3.posit.co/dev/reference/op_array.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Create a tensor. — op_array","text":"x Input tensor. dtype desired data-type tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_array.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Create a tensor. — op_array","text":"tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_array.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Create a tensor. — op_array","text":"","code":"op_array(c(1, 2, 3)) ## tf.Tensor([1. 2. 3.], shape=(3), dtype=float32) op_array(c(1, 2, 3), dtype = \"float32\") ## tf.Tensor([1. 2. 3.], shape=(3), dtype=float32) op_array(c(1, 2, 3), dtype = \"int32\") ## tf.Tensor([1 2 3], shape=(3), dtype=int32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_associative_scan.html","id":null,"dir":"Reference","previous_headings":"","what":"Performs a scan with an associative binary operation, in parallel. — op_associative_scan","title":"Performs a scan with an associative binary operation, in parallel. — op_associative_scan","text":"operation similar op_scan(), key difference op_associative_scan() parallel implementation potentially significant performance benefits, especially jit compiled. catch can used f binary associative operation (.e. must verify f(, f(b, c)) == f(f(, b), c)). introduction associative scans, refer paper: Blelloch, Guy E. 1990. Prefix Sums Applications.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_associative_scan.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Performs a scan with an associative binary operation, in parallel. — op_associative_scan","text":"","code":"op_associative_scan(f, elems, reverse = FALSE, axis = 1L)"},{"path":"https://keras3.posit.co/dev/reference/op_associative_scan.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Performs a scan with an associative binary operation, in parallel. — op_associative_scan","text":"f callable implementing associative binary operation signature  r = f(, b). Function f must associative, .e., must satisfy equation f(, f(b, c)) == f(f(, b), c). inputs result (possibly nested tree structures ) array(s) matching elems. array dimension place axis dimension. f applied elementwise axis dimension. result r shape (structure) two inputs b. elems (possibly nested tree structure ) array(s), axis dimension size num_elems. reverse boolean stating scan reversed respect axis dimension. axis integer identifying axis scan occur.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_associative_scan.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Performs a scan with an associative binary operation, in parallel. — op_associative_scan","text":"(possibly nested tree structure ) array(s) shape structure elems, k'th element axis result recursively applying f combine first k elements elems along axis. example, given elems = list(, b, c, ...), result list(, f(, b), f(f(, b), c), ...).","code":""},{"path":"https://keras3.posit.co/dev/reference/op_associative_scan.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Performs a scan with an associative binary operation, in parallel. — op_associative_scan","text":"","code":"sum_fn <- function(x, y) x + y xs <- op_arange(5L) op_associative_scan(sum_fn, xs) ## tf.Tensor([ 1  3  6 10 15], shape=(5), dtype=int32) sum_fn <- function(x, y) {   str(list(x = x, y = y))   map2(x, y, \\(.x, .y) .x + .y) }  xs <- list(op_array(1:2),            op_array(1:2),            op_array(1:2)) ys <- op_associative_scan(sum_fn, xs, axis = 1) ## List of 2 ##  $ x:List of 3 ##   ..$ :<tf.Tensor: shape=(1), dtype=int32, numpy=array([1], dtype=int32)> ##   ..$ :<tf.Tensor: shape=(1), dtype=int32, numpy=array([1], dtype=int32)> ##   ..$ :<tf.Tensor: shape=(1), dtype=int32, numpy=array([1], dtype=int32)> ##  $ y:List of 3 ##   ..$ :<tf.Tensor: shape=(1), dtype=int32, numpy=array([2], dtype=int32)> ##   ..$ :<tf.Tensor: shape=(1), dtype=int32, numpy=array([2], dtype=int32)> ##   ..$ :<tf.Tensor: shape=(1), dtype=int32, numpy=array([2], dtype=int32)> ys ## [[1]] ## tf.Tensor([1 3], shape=(2), dtype=int32) ## ## [[2]] ## tf.Tensor([1 3], shape=(2), dtype=int32) ## ## [[3]] ## tf.Tensor([1 3], shape=(2), dtype=int32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_average.html","id":null,"dir":"Reference","previous_headings":"","what":"Compute the weighted average along the specified axis. — op_average","title":"Compute the weighted average along the specified axis. — op_average","text":"Compute weighted average along specified axis.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_average.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Compute the weighted average along the specified axis. — op_average","text":"","code":"op_average(x, axis = NULL, weights = NULL)"},{"path":"https://keras3.posit.co/dev/reference/op_average.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Compute the weighted average along the specified axis. — op_average","text":"x Input tensor. axis Integer along average x. default, axis = NULL, average elements input tensor. axis negative counts last first axis. weights Tensor weights associated values x. value x contributes average according associated weight. weights array can either 1-D (case length must size along given axis) shape x. weights = NULL (default), data x assumed weight equal one. 1-D calculation : avg = sum(* weights) / sum(weights). constraint weights sum(weights) must 0.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_average.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Compute the weighted average along the specified axis. — op_average","text":"Return average along specified axis.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_average.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Compute the weighted average along the specified axis. — op_average","text":"","code":"data <- op_arange(1, 5, dtype = \"int32\") data ## tf.Tensor([1 2 3 4 5], shape=(5), dtype=int32) op_average(data) ## tf.Tensor(3.0, shape=(), dtype=float32) op_average(   op_arange(1, 11),   weights = op_arange(10, 0, -1) ) ## tf.Tensor(4.0, shape=(), dtype=float32) data <- op_arange(6) |> op_reshape(c(3, 2)) data ## tf.Tensor( ## [[1. 2.] ##  [3. 4.] ##  [5. 6.]], shape=(3, 2), dtype=float32) op_average(   data,   axis = 2,   weights = op_array(c(1/4, 3/4)) ) ## tf.Tensor([1.75 3.75 5.75], shape=(3), dtype=float32) # Error: Axis must be specified when shapes of x and weights differ. try(op_average(   data,   weights = op_array(c(1/4, 3/4)) )) ## Error in op_average(data, weights = op_array(c(1/4, 3/4))) : ##   Axis must be specified when shapes of x and weights differ."},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_average_pool.html","id":null,"dir":"Reference","previous_headings":"","what":"Average pooling operation. — op_average_pool","title":"Average pooling operation. — op_average_pool","text":"Average pooling operation.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_average_pool.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Average pooling operation. — op_average_pool","text":"","code":"op_average_pool(   inputs,   pool_size,   strides = NULL,   padding = \"valid\",   data_format = NULL )"},{"path":"https://keras3.posit.co/dev/reference/op_average_pool.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Average pooling operation. — op_average_pool","text":"inputs Tensor rank N+2. inputs shape (batch_size,) + inputs_spatial_shape + (num_channels,) data_format = \"channels_last\", (batch_size, num_channels) + inputs_spatial_shape data_format = \"channels_first\". Pooling happens spatial dimensions . pool_size int tuple/list integers size len(inputs_spatial_shape), specifying size pooling window spatial dimension input tensor. pool_size int, every spatial dimension shares pool_size. strides int tuple/list integers size len(inputs_spatial_shape). stride sliding window spatial dimension input tensor. strides int, every spatial dimension shares strides. padding string, either \"valid\" \"\". \"valid\" means padding applied, \"\" results padding evenly left/right /input output height/width dimension input strides = 1. data_format string, either \"channels_last\" \"channels_first\". data_format determines ordering dimensions inputs. data_format = \"channels_last\", inputs shape (batch_size, ..., channels) data_format = \"channels_first\", inputs shape (batch_size, channels, ...).","code":""},{"path":"https://keras3.posit.co/dev/reference/op_average_pool.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Average pooling operation. — op_average_pool","text":"tensor rank N+2, result average pooling operation.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_bartlett.html","id":null,"dir":"Reference","previous_headings":"","what":"Bartlett window function. — op_bartlett","title":"Bartlett window function. — op_bartlett","text":"Bartlett window triangular window rises falls linearly.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_bartlett.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Bartlett window function. — op_bartlett","text":"","code":"op_bartlett(x)"},{"path":"https://keras3.posit.co/dev/reference/op_bartlett.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Bartlett window function. — op_bartlett","text":"x Length window. Must positive integer.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_bartlett.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Bartlett window function. — op_bartlett","text":"1D tensor containing window values.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_bartlett.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Bartlett window function. — op_bartlett","text":"","code":"op_bartlett(5) ## tf.Tensor([0.  0.5 1.  0.5 0. ], shape=(5), dtype=float32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_batch_normalization.html","id":null,"dir":"Reference","previous_headings":"","what":"Normalizes x by mean and variance. — op_batch_normalization","title":"Normalizes x by mean and variance. — op_batch_normalization","text":"op typically used batch normalization step neural network. normalizes input tensor along given axis.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_batch_normalization.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Normalizes x by mean and variance. — op_batch_normalization","text":"","code":"op_batch_normalization(   x,   mean,   variance,   axis,   offset = NULL,   scale = NULL,   epsilon = 0.001 )"},{"path":"https://keras3.posit.co/dev/reference/op_batch_normalization.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Normalizes x by mean and variance. — op_batch_normalization","text":"x Input tensor. mean mean vector length axis dimension input thensor. variance variance vector length axis dimension input tensor. axis Integer, axis normalized. offset offset vector length axis dimension input tensor. NULL, offset added normalized tensor. Defaults NULL. scale scale vector length axis dimension input tensor. NULL, normalized tensor multiplied scale. Defaults NULL. epsilon Small float added variance avoid dividing zero. Defaults 1e-3.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_batch_normalization.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Normalizes x by mean and variance. — op_batch_normalization","text":"normalized tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_batch_normalization.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Normalizes x by mean and variance. — op_batch_normalization","text":"","code":"x <- op_convert_to_tensor(rbind(c(0.1, 0.2, 0.3),                                 c(0.4, 0.5, 0.6),                                 c(0.7, 0.8, 0.9))) op_batch_normalization(   x,   mean = c(0.4, 0.5, 0.6),   variance = c(0.67, 0.67, 0.67),   axis = -1 ) ## tf.Tensor( ## [[-0.36623513 -0.36623513 -0.36623513] ##  [ 0.          0.          0.        ] ##  [ 0.36623513  0.36623513  0.36623513]], shape=(3, 3), dtype=float64)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_binary_crossentropy.html","id":null,"dir":"Reference","previous_headings":"","what":"Computes binary cross-entropy loss between target and output tensor. — op_binary_crossentropy","title":"Computes binary cross-entropy loss between target and output tensor. — op_binary_crossentropy","text":"binary cross-entropy loss commonly used binary classification tasks input sample belongs one two classes. measures dissimilarity target output probabilities logits.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_binary_crossentropy.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes binary cross-entropy loss between target and output tensor. — op_binary_crossentropy","text":"","code":"op_binary_crossentropy(target, output, from_logits = FALSE)"},{"path":"https://keras3.posit.co/dev/reference/op_binary_crossentropy.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Computes binary cross-entropy loss between target and output tensor. — op_binary_crossentropy","text":"target target tensor representing true binary labels. shape match shape output tensor. output output tensor representing predicted probabilities logits. shape match shape target tensor. from_logits (optional) Whether output tensor logits probabilities. Set TRUE output represents logits; otherwise, set FALSE output represents probabilities. Defaults FALSE.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_binary_crossentropy.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Computes binary cross-entropy loss between target and output tensor. — op_binary_crossentropy","text":"Integer tensor: computed binary cross-entropy loss target output.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_binary_crossentropy.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Computes binary cross-entropy loss between target and output tensor. — op_binary_crossentropy","text":"","code":"target <- op_array(c(0, 1, 1, 0)) output <- op_array(c(0.1, 0.9, 0.8, 0.2)) op_binary_crossentropy(target, output) ## tf.Tensor([0.10536055 0.10536055 0.22314353 0.22314353], shape=(4), dtype=float32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_bincount.html","id":null,"dir":"Reference","previous_headings":"","what":"Count the number of occurrences of each value in a tensor of integers. — op_bincount","title":"Count the number of occurrences of each value in a tensor of integers. — op_bincount","text":"Count number occurrences value tensor integers.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_bincount.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Count the number of occurrences of each value in a tensor of integers. — op_bincount","text":"","code":"op_bincount(x, weights = NULL, minlength = 0L, sparse = FALSE)"},{"path":"https://keras3.posit.co/dev/reference/op_bincount.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Count the number of occurrences of each value in a tensor of integers. — op_bincount","text":"x Input tensor. must dimension 1, must contain non-negative integer(s). weights Weight tensor. must length x. default value NULL. specified, x weighted , .e. n = x[], [n] += weight[] instead default behavior [n] += 1. minlength integer. default value 0. specified, least number bins output tensor. greater max(x) + 1, value output index higher max(x) set 0. sparse Whether return sparse tensor; backends support sparse tensors.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_bincount.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Count the number of occurrences of each value in a tensor of integers. — op_bincount","text":"1D tensor element gives number occurrence(s) index value x. length maximum max(x) + 1 minlength.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_bincount.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Count the number of occurrences of each value in a tensor of integers. — op_bincount","text":"","code":"(x <- op_array(c(1, 2, 2, 3), dtype = \"uint8\")) ## tf.Tensor([1 2 2 3], shape=(4), dtype=uint8) op_bincount(x) ## tf.Tensor([0 1 2 1], shape=(4), dtype=int32) (weights <- x / 2) ## tf.Tensor([0.5 1.  1.  1.5], shape=(4), dtype=float32) op_bincount(x, weights = weights) ## tf.Tensor([0.  0.5 2.  1.5], shape=(4), dtype=float32) minlength <- as.integer(op_max(x) + 1 + 2) # 6 op_bincount(x, minlength = minlength) ## tf.Tensor([0 1 2 1 0 0], shape=(6), dtype=int32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_bitwise_and.html","id":null,"dir":"Reference","previous_headings":"","what":"Compute the bit-wise AND of two arrays element-wise. — op_bitwise_and","title":"Compute the bit-wise AND of two arrays element-wise. — op_bitwise_and","text":"Computes bit-wise underlying binary representation integers input arrays. ufunc implements C/Python operator &.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_bitwise_and.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Compute the bit-wise AND of two arrays element-wise. — op_bitwise_and","text":"","code":"op_bitwise_and(x, y)"},{"path":"https://keras3.posit.co/dev/reference/op_bitwise_and.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Compute the bit-wise AND of two arrays element-wise. — op_bitwise_and","text":"x Input integer tensor. y Input integer tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_bitwise_and.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Compute the bit-wise AND of two arrays element-wise. — op_bitwise_and","text":"Result tensor.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_bitwise_invert.html","id":null,"dir":"Reference","previous_headings":"","what":"Compute bit-wise inversion, or bit-wise NOT, element-wise. — op_bitwise_invert","title":"Compute bit-wise inversion, or bit-wise NOT, element-wise. — op_bitwise_invert","text":"Computes bit-wise underlying binary representation integers input arrays. ufunc implements C/Python operator ~.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_bitwise_invert.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Compute bit-wise inversion, or bit-wise NOT, element-wise. — op_bitwise_invert","text":"","code":"op_bitwise_invert(x)"},{"path":"https://keras3.posit.co/dev/reference/op_bitwise_invert.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Compute bit-wise inversion, or bit-wise NOT, element-wise. — op_bitwise_invert","text":"x Input integer tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_bitwise_invert.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Compute bit-wise inversion, or bit-wise NOT, element-wise. — op_bitwise_invert","text":"Result tensor.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_bitwise_left_shift.html","id":null,"dir":"Reference","previous_headings":"","what":"Shift the bits of an integer to the left. — op_bitwise_left_shift","title":"Shift the bits of an integer to the left. — op_bitwise_left_shift","text":"Bits shifted left appending y 0s right x. Since internal representation numbers binary format, operation equivalent multiplying x 2**y.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_bitwise_left_shift.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Shift the bits of an integer to the left. — op_bitwise_left_shift","text":"","code":"op_bitwise_left_shift(x, y)"},{"path":"https://keras3.posit.co/dev/reference/op_bitwise_left_shift.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Shift the bits of an integer to the left. — op_bitwise_left_shift","text":"x Input integer tensor. y Input integer tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_bitwise_left_shift.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Shift the bits of an integer to the left. — op_bitwise_left_shift","text":"Result tensor.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_bitwise_not.html","id":null,"dir":"Reference","previous_headings":"","what":"Compute bit-wise inversion, or bit-wise NOT, element-wise. — op_bitwise_not","title":"Compute bit-wise inversion, or bit-wise NOT, element-wise. — op_bitwise_not","text":"Computes bit-wise underlying binary representation integers input arrays. ufunc implements C/Python operator ~.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_bitwise_not.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Compute bit-wise inversion, or bit-wise NOT, element-wise. — op_bitwise_not","text":"","code":"op_bitwise_not(x)"},{"path":"https://keras3.posit.co/dev/reference/op_bitwise_not.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Compute bit-wise inversion, or bit-wise NOT, element-wise. — op_bitwise_not","text":"x Input integer tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_bitwise_not.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Compute bit-wise inversion, or bit-wise NOT, element-wise. — op_bitwise_not","text":"Result tensor.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_bitwise_or.html","id":null,"dir":"Reference","previous_headings":"","what":"Compute the bit-wise OR of two arrays element-wise. — op_bitwise_or","title":"Compute the bit-wise OR of two arrays element-wise. — op_bitwise_or","text":"Computes bit-wise underlying binary representation integers input arrays. ufunc implements C/Python operator |.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_bitwise_or.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Compute the bit-wise OR of two arrays element-wise. — op_bitwise_or","text":"","code":"op_bitwise_or(x, y)"},{"path":"https://keras3.posit.co/dev/reference/op_bitwise_or.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Compute the bit-wise OR of two arrays element-wise. — op_bitwise_or","text":"x Input integer tensor. y Input integer tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_bitwise_or.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Compute the bit-wise OR of two arrays element-wise. — op_bitwise_or","text":"Result tensor.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_bitwise_right_shift.html","id":null,"dir":"Reference","previous_headings":"","what":"Shift the bits of an integer to the right. — op_bitwise_right_shift","title":"Shift the bits of an integer to the right. — op_bitwise_right_shift","text":"Bits shifted right y. internal representation numbers binary format, operation equivalent dividing x 2**y.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_bitwise_right_shift.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Shift the bits of an integer to the right. — op_bitwise_right_shift","text":"","code":"op_bitwise_right_shift(x, y)"},{"path":"https://keras3.posit.co/dev/reference/op_bitwise_right_shift.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Shift the bits of an integer to the right. — op_bitwise_right_shift","text":"x Input integer tensor. y Input integer tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_bitwise_right_shift.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Shift the bits of an integer to the right. — op_bitwise_right_shift","text":"Result tensor.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_bitwise_xor.html","id":null,"dir":"Reference","previous_headings":"","what":"Compute the bit-wise XOR of two arrays element-wise. — op_bitwise_xor","title":"Compute the bit-wise XOR of two arrays element-wise. — op_bitwise_xor","text":"Computes bit-wise XOR underlying binary representation integers input arrays. ufunc implements C/Python operator ^.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_bitwise_xor.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Compute the bit-wise XOR of two arrays element-wise. — op_bitwise_xor","text":"","code":"op_bitwise_xor(x, y)"},{"path":"https://keras3.posit.co/dev/reference/op_bitwise_xor.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Compute the bit-wise XOR of two arrays element-wise. — op_bitwise_xor","text":"x Input integer tensor. y Input integer tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_bitwise_xor.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Compute the bit-wise XOR of two arrays element-wise. — op_bitwise_xor","text":"Result tensor.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_blackman.html","id":null,"dir":"Reference","previous_headings":"","what":"Blackman window function. — op_blackman","title":"Blackman window function. — op_blackman","text":"Blackman window taper formed using weighted cosine.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_blackman.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Blackman window function. — op_blackman","text":"","code":"op_blackman(x)"},{"path":"https://keras3.posit.co/dev/reference/op_blackman.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Blackman window function. — op_blackman","text":"x Length window. Must positive integer.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_blackman.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Blackman window function. — op_blackman","text":"1D tensor containing window values.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_broadcast_to.html","id":null,"dir":"Reference","previous_headings":"","what":"Broadcast a tensor to a new shape. — op_broadcast_to","title":"Broadcast a tensor to a new shape. — op_broadcast_to","text":"Broadcast tensor new shape.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_broadcast_to.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Broadcast a tensor to a new shape. — op_broadcast_to","text":"","code":"op_broadcast_to(x, shape)"},{"path":"https://keras3.posit.co/dev/reference/op_broadcast_to.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Broadcast a tensor to a new shape. — op_broadcast_to","text":"x tensor broadcast. shape shape desired tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_broadcast_to.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Broadcast a tensor to a new shape. — op_broadcast_to","text":"tensor desired shape.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_broadcast_to.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Broadcast a tensor to a new shape. — op_broadcast_to","text":"","code":"x <- op_array(c(1, 2, 3)) op_broadcast_to(x, shape = c(3, 3)) ## tf.Tensor( ## [[1. 2. 3.] ##  [1. 2. 3.] ##  [1. 2. 3.]], shape=(3, 3), dtype=float32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_cast.html","id":null,"dir":"Reference","previous_headings":"","what":"Cast a tensor to the desired dtype. — op_cast","title":"Cast a tensor to the desired dtype. — op_cast","text":"Cast tensor desired dtype.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_cast.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Cast a tensor to the desired dtype. — op_cast","text":"","code":"op_cast(x, dtype)"},{"path":"https://keras3.posit.co/dev/reference/op_cast.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Cast a tensor to the desired dtype. — op_cast","text":"x tensor variable. dtype target type.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_cast.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Cast a tensor to the desired dtype. — op_cast","text":"tensor specified dtype.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_cast.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Cast a tensor to the desired dtype. — op_cast","text":"","code":"(x <- op_arange(4)) ## tf.Tensor([1. 2. 3. 4.], shape=(4), dtype=float32) op_cast(x, dtype = \"float16\") ## tf.Tensor([1. 2. 3. 4.], shape=(4), dtype=float16)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_categorical_crossentropy.html","id":null,"dir":"Reference","previous_headings":"","what":"Computes categorical cross-entropy loss between target and output tensor. — op_categorical_crossentropy","title":"Computes categorical cross-entropy loss between target and output tensor. — op_categorical_crossentropy","text":"categorical cross-entropy loss commonly used multi-class classification tasks input sample can belong one multiple classes. measures dissimilarity target output probabilities logits.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_categorical_crossentropy.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes categorical cross-entropy loss between target and output tensor. — op_categorical_crossentropy","text":"","code":"op_categorical_crossentropy(target, output, from_logits = FALSE, axis = -1L)"},{"path":"https://keras3.posit.co/dev/reference/op_categorical_crossentropy.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Computes categorical cross-entropy loss between target and output tensor. — op_categorical_crossentropy","text":"target target tensor representing true categorical labels. shape match shape output tensor except last dimension. output output tensor representing predicted probabilities logits. shape match shape target tensor except last dimension. from_logits (optional) Whether output tensor logits probabilities. Set TRUE output represents logits; otherwise, set FALSE output represents probabilities. Defaults FALSE. axis (optional) axis along categorical cross-entropy computed. Defaults -1, corresponds last dimension tensors.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_categorical_crossentropy.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Computes categorical cross-entropy loss between target and output tensor. — op_categorical_crossentropy","text":"Integer tensor: computed categorical cross-entropy loss target output.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_categorical_crossentropy.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Computes categorical cross-entropy loss between target and output tensor. — op_categorical_crossentropy","text":"","code":"target <- op_array(rbind(c(1, 0, 0),                         c(0, 1, 0),                         c(0, 0, 1))) output <- op_array(rbind(c(0.9, 0.05, 0.05),                         c(0.1, 0.8, 0.1),                         c(0.2, 0.3, 0.5))) op_categorical_crossentropy(target, output) ## tf.Tensor([0.10536052 0.22314355 0.69314718], shape=(3), dtype=float64)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_cbrt.html","id":null,"dir":"Reference","previous_headings":"","what":"Computes the cube root of the input tensor, element-wise. — op_cbrt","title":"Computes the cube root of the input tensor, element-wise. — op_cbrt","text":"Returns real-valued cube root x, handling negative inputs real domain.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_cbrt.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes the cube root of the input tensor, element-wise. — op_cbrt","text":"","code":"op_cbrt(x)"},{"path":"https://keras3.posit.co/dev/reference/op_cbrt.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Computes the cube root of the input tensor, element-wise. — op_cbrt","text":"x Input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_cbrt.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Computes the cube root of the input tensor, element-wise. — op_cbrt","text":"tensor containing cube root element x.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_cbrt.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Computes the cube root of the input tensor, element-wise. — op_cbrt","text":"","code":"op_cbrt(c(-8, 0, 8)) ## tf.Tensor([-2.  0.  2.], shape=(3), dtype=float32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_ceil.html","id":null,"dir":"Reference","previous_headings":"","what":"Return the ceiling of the input, element-wise. — op_ceil","title":"Return the ceiling of the input, element-wise. — op_ceil","text":"ceil scalar x smallest integer , >= x.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_ceil.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Return the ceiling of the input, element-wise. — op_ceil","text":"","code":"op_ceil(x)"},{"path":"https://keras3.posit.co/dev/reference/op_ceil.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Return the ceiling of the input, element-wise. — op_ceil","text":"x Input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_ceil.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Return the ceiling of the input, element-wise. — op_ceil","text":"ceiling element x, float dtype.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_celu.html","id":null,"dir":"Reference","previous_headings":"","what":"Continuously-differentiable exponential linear unit. — op_celu","title":"Continuously-differentiable exponential linear unit. — op_celu","text":"defined : f(x) =  alpha * (exp(x / alpha) - 1) x < 0, f(x) = x x >= 0.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_celu.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Continuously-differentiable exponential linear unit. — op_celu","text":"","code":"op_celu(x, alpha = 1)"},{"path":"https://keras3.posit.co/dev/reference/op_celu.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Continuously-differentiable exponential linear unit. — op_celu","text":"x Input tensor. alpha value CELU formulation. Defaults 1.0.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_celu.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Continuously-differentiable exponential linear unit. — op_celu","text":"tensor shape x.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_celu.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Continuously-differentiable exponential linear unit. — op_celu","text":"","code":"x <- op_array(c(-1., 0., 1.)) op_celu(x) ## tf.Tensor([-0.63212055  0.          1.        ], shape=(3), dtype=float32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_cholesky.html","id":null,"dir":"Reference","previous_headings":"","what":"Computes the Cholesky decomposition of a positive semi-definite matrix. — op_cholesky","title":"Computes the Cholesky decomposition of a positive semi-definite matrix. — op_cholesky","text":"Computes Cholesky decomposition positive semi-definite matrix.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_cholesky.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes the Cholesky decomposition of a positive semi-definite matrix. — op_cholesky","text":"","code":"op_cholesky(x)"},{"path":"https://keras3.posit.co/dev/reference/op_cholesky.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Computes the Cholesky decomposition of a positive semi-definite matrix. — op_cholesky","text":"x Input tensor shape (..., M, M).","code":""},{"path":"https://keras3.posit.co/dev/reference/op_cholesky.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Computes the Cholesky decomposition of a positive semi-definite matrix. — op_cholesky","text":"tensor shape (..., M, M) representing lower triangular Cholesky factor x.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_clip.html","id":null,"dir":"Reference","previous_headings":"","what":"Clip (limit) the values in a tensor. — op_clip","title":"Clip (limit) the values in a tensor. — op_clip","text":"Given interval, values outside interval clipped interval edges. example, interval [0, 1] specified, values smaller 0 become 0, values larger 1 become 1.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_clip.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Clip (limit) the values in a tensor. — op_clip","text":"","code":"op_clip(x, x_min, x_max)"},{"path":"https://keras3.posit.co/dev/reference/op_clip.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Clip (limit) the values in a tensor. — op_clip","text":"x Input tensor. x_min Minimum value. x_max Maximum value.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_clip.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Clip (limit) the values in a tensor. — op_clip","text":"clipped tensor.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_concatenate.html","id":null,"dir":"Reference","previous_headings":"","what":"Join a sequence of tensors along an existing axis. — op_concatenate","title":"Join a sequence of tensors along an existing axis. — op_concatenate","text":"Join sequence tensors along existing axis.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_concatenate.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Join a sequence of tensors along an existing axis. — op_concatenate","text":"","code":"op_concatenate(xs, axis = 1L)"},{"path":"https://keras3.posit.co/dev/reference/op_concatenate.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Join a sequence of tensors along an existing axis. — op_concatenate","text":"xs sequence tensors concatenate. axis axis along tensors joined. Defaults 0.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_concatenate.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Join a sequence of tensors along an existing axis. — op_concatenate","text":"concatenated tensor.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_cond.html","id":null,"dir":"Reference","previous_headings":"","what":"Conditionally applies true_fn or false_fn. — op_cond","title":"Conditionally applies true_fn or false_fn. — op_cond","text":"Conditionally applies true_fn false_fn.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_cond.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Conditionally applies true_fn or false_fn. — op_cond","text":"","code":"op_cond(pred, true_fn, false_fn)"},{"path":"https://keras3.posit.co/dev/reference/op_cond.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Conditionally applies true_fn or false_fn. — op_cond","text":"pred Boolean scalar type true_fn Callable returning output pred == TRUE case. false_fn Callable returning output pred == FALSE case.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_cond.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Conditionally applies true_fn or false_fn. — op_cond","text":"output either true_fn false_fn depending pred.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_cond.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Conditionally applies true_fn or false_fn. — op_cond","text":"","code":"fn <- tensorflow::tf_function(function(x) {   op_cond(x > 0,     true_fn = \\() x + 1,     false_fn = \\() x - 1) })  fn(tensorflow::as_tensor(1)) ## tf.Tensor(2.0, shape=(), dtype=float64) fn(tensorflow::as_tensor(-1)) ## tf.Tensor(-2.0, shape=(), dtype=float64) # # Conditional side-effect (print only, no return value). file <- tempfile(fileext = \".txt\") fn <- tensorflow::tf_function(function(epochs) {   op_fori_loop(     0, epochs,     body_fun = \\(epoch, state) {       op_cond(epoch %% 20 == 0,               \\() {                 tensorflow::tf$print(                   \"epoch:\", epoch,                   output_stream = paste0(\"file://\", file))                 NULL               },               \\() {NULL})       state     },     init_val = tensorflow::as_tensor(0, \"float32\")) })  fn(tensorflow::as_tensor(100, \"float32\")) ## tf.Tensor(0.0, shape=(), dtype=float32) readLines(file) ## [1] \"epoch: 0\"  \"epoch: 20\" \"epoch: 40\" \"epoch: 60\" \"epoch: 80\" # cleanup unlink(file)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_conj.html","id":null,"dir":"Reference","previous_headings":"","what":"Returns the complex conjugate, element-wise. — op_conj","title":"Returns the complex conjugate, element-wise. — op_conj","text":"complex conjugate complex number obtained changing sign imaginary part.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_conj.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Returns the complex conjugate, element-wise. — op_conj","text":"","code":"op_conj(x)"},{"path":"https://keras3.posit.co/dev/reference/op_conj.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Returns the complex conjugate, element-wise. — op_conj","text":"x Input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_conj.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Returns the complex conjugate, element-wise. — op_conj","text":"complex conjugate element x.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_conv.html","id":null,"dir":"Reference","previous_headings":"","what":"General N-D convolution. — op_conv","title":"General N-D convolution. — op_conv","text":"ops supports 1D, 2D 3D convolution.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_conv.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"General N-D convolution. — op_conv","text":"","code":"op_conv(   inputs,   kernel,   strides = 1L,   padding = \"valid\",   data_format = NULL,   dilation_rate = 1L )"},{"path":"https://keras3.posit.co/dev/reference/op_conv.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"General N-D convolution. — op_conv","text":"inputs Tensor rank N+2. inputs shape (batch_size,) + inputs_spatial_shape + (num_channels,) data_format = \"channels_last\", (batch_size, num_channels) + inputs_spatial_shape data_format = \"channels_first\". kernel Tensor rank N+2. kernel shape (kernel_spatial_shape, num_input_channels, num_output_channels). num_input_channels match number channels inputs. strides int int tuple/list len(inputs_spatial_shape), specifying strides convolution along spatial dimension. strides int, every spatial dimension shares strides. Defaults 1. padding string, either \"valid\" \"\". \"valid\" means padding applied, \"\" results padding evenly left/right /input output height/width dimension input strides = 1. data_format string, either \"channels_last\" \"channels_first\". data_format determines ordering dimensions inputs. data_format = \"channels_last\", inputs shape (batch_size, ..., channels) data_format = \"channels_first\", inputs shape (batch_size, channels, ...). dilation_rate int int tuple/list len(inputs_spatial_shape), specifying dilation rate use dilated convolution. dilation_rate int, every spatial dimension shares dilation_rate.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_conv.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"General N-D convolution. — op_conv","text":"tensor rank N+2, result conv operation.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_conv_transpose.html","id":null,"dir":"Reference","previous_headings":"","what":"General N-D convolution transpose. — op_conv_transpose","title":"General N-D convolution transpose. — op_conv_transpose","text":"Also known de-convolution. ops supports 1D, 2D 3D convolution.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_conv_transpose.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"General N-D convolution transpose. — op_conv_transpose","text":"","code":"op_conv_transpose(   inputs,   kernel,   strides = 1L,   padding = \"valid\",   output_padding = NULL,   data_format = NULL,   dilation_rate = 1L )"},{"path":"https://keras3.posit.co/dev/reference/op_conv_transpose.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"General N-D convolution transpose. — op_conv_transpose","text":"inputs Tensor rank N+2. inputs shape (batch_size,) + inputs_spatial_shape + (num_channels,) data_format = \"channels_last\", (batch_size, num_channels) + inputs_spatial_shape data_format = \"channels_first\". kernel Tensor rank N+2. kernel shape [kernel_spatial_shape, num_output_channels, num_input_channels], num_input_channels match number channels inputs. strides int int tuple/list len(inputs_spatial_shape), specifying strides convolution along spatial dimension. strides int, every spatial dimension shares strides. padding string, either \"valid\" \"\". \"valid\" means padding applied, \"\" results padding evenly left/right /input output height/width dimension input strides = 1. output_padding int int tuple/list len(inputs_spatial_shape), specifying amount padding along height width output tensor. Can single integer specify value spatial dimensions. amount output padding along given dimension must lower stride along dimension. set NULL (default), output shape inferred. data_format string, either \"channels_last\" \"channels_first\". data_format determines ordering dimensions inputs. data_format = \"channels_last\", inputs shape (batch_size, ..., channels) data_format = \"channels_first\", inputs shape (batch_size, channels, ...). dilation_rate int int tuple/list len(inputs_spatial_shape), specifying dilation rate use dilated convolution. dilation_rate int, every spatial dimension shares dilation_rate.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_conv_transpose.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"General N-D convolution transpose. — op_conv_transpose","text":"tensor rank N+2, result conv operation.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_convert_to_numpy.html","id":null,"dir":"Reference","previous_headings":"","what":"Convert a tensor to an R or NumPy array. — op_convert_to_numpy","title":"Convert a tensor to an R or NumPy array. — op_convert_to_numpy","text":"Convert tensor R NumPy array.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_convert_to_numpy.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Convert a tensor to an R or NumPy array. — op_convert_to_numpy","text":"","code":"op_convert_to_numpy(x)  op_convert_to_array(x)"},{"path":"https://keras3.posit.co/dev/reference/op_convert_to_numpy.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Convert a tensor to an R or NumPy array. — op_convert_to_numpy","text":"x tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_convert_to_numpy.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Convert a tensor to an R or NumPy array. — op_convert_to_numpy","text":"NumPy array R array.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_convert_to_tensor.html","id":null,"dir":"Reference","previous_headings":"","what":"Convert an array to a tensor. — op_convert_to_tensor","title":"Convert an array to a tensor. — op_convert_to_tensor","text":"Native tensors current backend left unchanged unless dtype, sparse ragged arguments set.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_convert_to_tensor.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Convert an array to a tensor. — op_convert_to_tensor","text":"","code":"op_convert_to_tensor(x, dtype = NULL, sparse = NULL, ragged = NULL)"},{"path":"https://keras3.posit.co/dev/reference/op_convert_to_tensor.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Convert an array to a tensor. — op_convert_to_tensor","text":"x R NumPy array (can nested) backend tensor. dtype target type. NULL, type x used. sparse Whether keep sparse tensors. FALSE cause sparse tensors densified. default value NULL means sparse tensors kept backend supports . ragged Whether keep ragged tensors. FALSE cause ragged tensors densified. default value NULL means ragged tensors kept backend supports .","code":""},{"path":"https://keras3.posit.co/dev/reference/op_convert_to_tensor.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Convert an array to a tensor. — op_convert_to_tensor","text":"backend tensor specified dtype sparseness.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_convert_to_tensor.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Convert an array to a tensor. — op_convert_to_tensor","text":"","code":"x <- array(c(1, 2, 3)) y <- op_convert_to_tensor(x) y ## tf.Tensor([1. 2. 3.], shape=(3), dtype=float64) op_convert_to_tensor(c(1, 3, 2, 0), \"int32\") ## tf.Tensor([1 3 2 0], shape=(4), dtype=int32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_copy.html","id":null,"dir":"Reference","previous_headings":"","what":"Returns a copy of x. — op_copy","title":"Returns a copy of x. — op_copy","text":"Returns copy x.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_copy.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Returns a copy of x. — op_copy","text":"","code":"op_copy(x)"},{"path":"https://keras3.posit.co/dev/reference/op_copy.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Returns a copy of x. — op_copy","text":"x Input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_copy.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Returns a copy of x. — op_copy","text":"copy x.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_corrcoef.html","id":null,"dir":"Reference","previous_headings":"","what":"Compute the Pearson correlation coefficient matrix. — op_corrcoef","title":"Compute the Pearson correlation coefficient matrix. — op_corrcoef","text":"Compute Pearson correlation coefficient matrix.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_corrcoef.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Compute the Pearson correlation coefficient matrix. — op_corrcoef","text":"","code":"op_corrcoef(x)"},{"path":"https://keras3.posit.co/dev/reference/op_corrcoef.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Compute the Pearson correlation coefficient matrix. — op_corrcoef","text":"x 2D tensor shape (N, D), N number variables D number observations.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_corrcoef.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Compute the Pearson correlation coefficient matrix. — op_corrcoef","text":"tensor shape (N, N) representing correlation matrix.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_corrcoef.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Compute the Pearson correlation coefficient matrix. — op_corrcoef","text":"","code":"x <- op_array(matrix(c(1, 2, 3,                        2, 3, 4), nrow = 2, byrow = TRUE)) op_corrcoef(x) ## tf.Tensor( ## [[1. 1.] ##  [1. 1.]], shape=(2, 2), dtype=float64)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_correlate.html","id":null,"dir":"Reference","previous_headings":"","what":"Compute the cross-correlation of two 1-dimensional tensors. — op_correlate","title":"Compute the cross-correlation of two 1-dimensional tensors. — op_correlate","text":"Compute cross-correlation two 1-dimensional tensors.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_correlate.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Compute the cross-correlation of two 1-dimensional tensors. — op_correlate","text":"","code":"op_correlate(x1, x2, mode = \"valid\")"},{"path":"https://keras3.posit.co/dev/reference/op_correlate.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Compute the cross-correlation of two 1-dimensional tensors. — op_correlate","text":"x1 First 1-dimensional input tensor length M. x2 Second 1-dimensional input tensor length N. mode Either \"valid\", \"\" \"full\". default mode set \"valid\", returns output length max(M, N) - min(M, N) + 1. \"\" returns output length max(M, N). \"full\" mode returns convolution point overlap, output length N+M-1.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_correlate.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Compute the cross-correlation of two 1-dimensional tensors. — op_correlate","text":"Output tensor, cross-correlation x1 x2.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_cos.html","id":null,"dir":"Reference","previous_headings":"","what":"Cosine, element-wise. — op_cos","title":"Cosine, element-wise. — op_cos","text":"Cosine, element-wise.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_cos.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Cosine, element-wise. — op_cos","text":"","code":"op_cos(x)"},{"path":"https://keras3.posit.co/dev/reference/op_cos.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Cosine, element-wise. — op_cos","text":"x Input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_cos.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Cosine, element-wise. — op_cos","text":"corresponding cosine values.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_cosh.html","id":null,"dir":"Reference","previous_headings":"","what":"Hyperbolic cosine, element-wise. — op_cosh","title":"Hyperbolic cosine, element-wise. — op_cosh","text":"Hyperbolic cosine, element-wise.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_cosh.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Hyperbolic cosine, element-wise. — op_cosh","text":"","code":"op_cosh(x)"},{"path":"https://keras3.posit.co/dev/reference/op_cosh.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Hyperbolic cosine, element-wise. — op_cosh","text":"x Input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_cosh.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Hyperbolic cosine, element-wise. — op_cosh","text":"Output tensor shape x.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_count_nonzero.html","id":null,"dir":"Reference","previous_headings":"","what":"Counts the number of non-zero values in x along the given axis. — op_count_nonzero","title":"Counts the number of non-zero values in x along the given axis. — op_count_nonzero","text":"axis specified non-zeros tensor counted.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_count_nonzero.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Counts the number of non-zero values in x along the given axis. — op_count_nonzero","text":"","code":"op_count_nonzero(x, axis = NULL)"},{"path":"https://keras3.posit.co/dev/reference/op_count_nonzero.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Counts the number of non-zero values in x along the given axis. — op_count_nonzero","text":"x Input tensor. axis Axis tuple axes along count number non-zeros. Defaults NULL.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_count_nonzero.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Counts the number of non-zero values in x along the given axis. — op_count_nonzero","text":"integer tensor integers.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_count_nonzero.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Counts the number of non-zero values in x along the given axis. — op_count_nonzero","text":"","code":"x <- op_array(rbind(c(0, 1, 7, 0),                    c(3, 0, 2, 19))) op_count_nonzero(x) ## tf.Tensor(5, shape=(), dtype=int32) op_count_nonzero(x, axis = 1) ## tf.Tensor([1 1 2 1], shape=(4), dtype=int32) op_count_nonzero(x, axis = 2) ## tf.Tensor([2 3], shape=(2), dtype=int32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_cross.html","id":null,"dir":"Reference","previous_headings":"","what":"Returns the cross product of two (arrays of) vectors. — op_cross","title":"Returns the cross product of two (arrays of) vectors. — op_cross","text":"cross product x1 x2 R^3 vector perpendicular x1 x2. x1 x2 arrays vectors, vectors defined last axis x1 x2 default, axes can dimensions 2 3. dimension either x1 x2 2, third component input vector assumed zero cross product calculated accordingly. cases input vectors dimension 2, z-component cross product returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_cross.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Returns the cross product of two (arrays of) vectors. — op_cross","text":"","code":"op_cross(x1, x2, axisa = -1L, axisb = -1L, axisc = -1L, axis = NULL)"},{"path":"https://keras3.posit.co/dev/reference/op_cross.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Returns the cross product of two (arrays of) vectors. — op_cross","text":"x1 Components first vector(s). x2 Components second vector(s). axisa Axis x1 defines vector(s). Defaults -1. axisb Axis x2 defines vector(s). Defaults -1. axisc Axis result containing cross product vector(s). Ignored input vectors dimension 2, return scalar. default, last axis. axis defined, axis x1, x2 result defines vector(s) cross product(s). Overrides axisa, axisb axisc.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_cross.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Returns the cross product of two (arrays of) vectors. — op_cross","text":"Vector cross product(s).","code":""},{"path":"https://keras3.posit.co/dev/reference/op_cross.html","id":"note","dir":"Reference","previous_headings":"","what":"Note","title":"Returns the cross product of two (arrays of) vectors. — op_cross","text":"Torch backend support two dimensional vectors, arguments axisa, axisb axisc. Use axis instead.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_ctc_decode.html","id":null,"dir":"Reference","previous_headings":"","what":"Decodes the output of a CTC model. — op_ctc_decode","title":"Decodes the output of a CTC model. — op_ctc_decode","text":"Decodes output CTC model.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_ctc_decode.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Decodes the output of a CTC model. — op_ctc_decode","text":"","code":"op_ctc_decode(   inputs,   sequence_lengths,   strategy = \"greedy\",   beam_width = 100L,   top_paths = 1L,   merge_repeated = TRUE,   mask_index = 1L )"},{"path":"https://keras3.posit.co/dev/reference/op_ctc_decode.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Decodes the output of a CTC model. — op_ctc_decode","text":"inputs tensor shape (batch_size, max_length, num_classes) containing logits (output model). normalized via softmax. sequence_lengths tensor shape (batch_size) containing sequence lengths batch. strategy string decoding strategy. Supported values \"greedy\" \"beam_search\". beam_width integer scalar beam width used beam search. Defaults 100. top_paths integer scalar, number top paths return. Defaults 1. merge_repeated boolean scalar, whether merge repeated labels output. Defaults TRUE. mask_index integer scalar, (1-based) index mask character vocabulary. Defaults 1.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_ctc_decode.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Decodes the output of a CTC model. — op_ctc_decode","text":"list containing: tensor representing list decoded sequences. strategy=\"greedy\", shape (1, batch_size, max_length). strategy=\"beam_search\", shape (top_paths, batch_size, max_length). Note : -1 indicates blank label. strategy=\"greedy\", tensor shape (batch_size, 1) representing negative sum probability logits sequence. strategy=\"beam_search\", tensor shape (batch_size, top_paths) representing log probability sequence.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_ctc_loss.html","id":null,"dir":"Reference","previous_headings":"","what":"CTC (Connectionist Temporal Classification) loss. — op_ctc_loss","title":"CTC (Connectionist Temporal Classification) loss. — op_ctc_loss","text":"CTC (Connectionist Temporal Classification) loss.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_ctc_loss.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"CTC (Connectionist Temporal Classification) loss. — op_ctc_loss","text":"","code":"op_ctc_loss(target, output, target_length, output_length, mask_index = 1L)"},{"path":"https://keras3.posit.co/dev/reference/op_ctc_loss.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"CTC (Connectionist Temporal Classification) loss. — op_ctc_loss","text":"target tensor shape (batch_size, max_length) containing true labels integer format. output tensor shape (batch_size, max_length, num_classes) containing logits (output model). target_length tensor shape (batch_size) containing true label lengths. output_length tensor shape (batch_size) containing output lengths. mask_index index mask character vocabulary. Defaults 1 (first element vocabulary).","code":""},{"path":"https://keras3.posit.co/dev/reference/op_ctc_loss.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"CTC (Connectionist Temporal Classification) loss. — op_ctc_loss","text":"tensor, shape (batch_size), loss values.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_cumprod.html","id":null,"dir":"Reference","previous_headings":"","what":"Return the cumulative product of elements along a given axis. — op_cumprod","title":"Return the cumulative product of elements along a given axis. — op_cumprod","text":"Return cumulative product elements along given axis.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_cumprod.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Return the cumulative product of elements along a given axis. — op_cumprod","text":"","code":"op_cumprod(x, axis = NULL, dtype = NULL)"},{"path":"https://keras3.posit.co/dev/reference/op_cumprod.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Return the cumulative product of elements along a given axis. — op_cumprod","text":"x Input tensor. axis Axis along cumulative product computed. default input flattened. dtype dtype returned tensor. Defaults x$dtype.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_cumprod.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Return the cumulative product of elements along a given axis. — op_cumprod","text":"Output tensor.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_cumsum.html","id":null,"dir":"Reference","previous_headings":"","what":"Returns the cumulative sum of elements along a given axis. — op_cumsum","title":"Returns the cumulative sum of elements along a given axis. — op_cumsum","text":"Returns cumulative sum elements along given axis.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_cumsum.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Returns the cumulative sum of elements along a given axis. — op_cumsum","text":"","code":"op_cumsum(x, axis = NULL, dtype = NULL)"},{"path":"https://keras3.posit.co/dev/reference/op_cumsum.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Returns the cumulative sum of elements along a given axis. — op_cumsum","text":"x Input tensor. axis Axis along cumulative sum computed. default input flattened. dtype dtype returned tensor. Defaults x$dtype.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_cumsum.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Returns the cumulative sum of elements along a given axis. — op_cumsum","text":"Output tensor.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_custom_gradient.html","id":null,"dir":"Reference","previous_headings":"","what":"Decorator to define a function with a custom gradient. — op_custom_gradient","title":"Decorator to define a function with a custom gradient. — op_custom_gradient","text":"decorator allows fine grained control gradients sequence operations. may useful multiple reasons, including providing efficient numerically stable gradient sequence operations.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_custom_gradient.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Decorator to define a function with a custom gradient. — op_custom_gradient","text":"","code":"op_custom_gradient(f)"},{"path":"https://keras3.posit.co/dev/reference/op_custom_gradient.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Decorator to define a function with a custom gradient. — op_custom_gradient","text":"f Function f(...) returns tuple (output, grad_fn) : ... sequence unnamed arguments, tensor input nested structure tensor inputs function. output (potentially nested structure ) tensor outputs applying operations forward_fn f() .... grad_fn function signature grad_fn(..., upstream) returns list tensors size (flattened) ...: derivatives tensors output respect tensors .... upstream tensor sequence tensors holding initial value gradients tensor output.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_custom_gradient.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Decorator to define a function with a custom gradient. — op_custom_gradient","text":"function h(...) returns value f(...)[[1]] whose gradient determined f(...)[[2]].","code":""},{"path":"https://keras3.posit.co/dev/reference/op_custom_gradient.html","id":"note","dir":"Reference","previous_headings":"","what":"Note","title":"Decorator to define a function with a custom gradient. — op_custom_gradient","text":"Note grad function returns gradient computation requires ... well upstream named argument, depending backend set. JAX TensorFlow backends, requires one argument, whereas might use upstream argument case PyTorch backend. working TensorFlow/JAX backend, grad(upstream) sufficient. PyTorch, grad function requires ... well upstream, e.g. grad <- \\(..., upstream). Follow example use op_custom_gradient() way compatible backends.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_custom_gradient.html","id":"example","dir":"Reference","previous_headings":"","what":"Example","title":"Decorator to define a function with a custom gradient. — op_custom_gradient","text":"Backend-agnostic example.","code":"log1pexp <- op_custom_gradient(\\(x) {      e <- op_exp(x)      grad <- function(..., upstream = NULL) {       upstream <- upstream %||% ..1       op_multiply(upstream, 1.0 - 1.0 / op_add(1, e))     }      tuple(op_log(1 + e), grad)  })  if(config_backend() == \"tensorflow\") {   tf <- tensorflow::tf   x <- op_convert_to_tensor(100.0)   with(tf$GradientTape() %as% tape, {     tape$watch(x)     y <- log1pexp(x)   })   dy_dx <- tape$gradient(y, x)   stopifnot(as.numeric(dy_dx) == 1) }"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_deg2rad.html","id":null,"dir":"Reference","previous_headings":"","what":"Convert angles from degrees to radians. — op_deg2rad","title":"Convert angles from degrees to radians. — op_deg2rad","text":"conversion defined : rad = deg * (pi / 180).","code":""},{"path":"https://keras3.posit.co/dev/reference/op_deg2rad.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Convert angles from degrees to radians. — op_deg2rad","text":"","code":"op_deg2rad(x)"},{"path":"https://keras3.posit.co/dev/reference/op_deg2rad.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Convert angles from degrees to radians. — op_deg2rad","text":"x Input tensor angles degrees.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_deg2rad.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Convert angles from degrees to radians. — op_deg2rad","text":"tensor containing angles converted radians.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_deg2rad.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Convert angles from degrees to radians. — op_deg2rad","text":"","code":"op_deg2rad(c(0, 90, 180)) ## tf.Tensor([0.        1.5707964 3.1415927], shape=(3), dtype=float32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_depthwise_conv.html","id":null,"dir":"Reference","previous_headings":"","what":"General N-D depthwise convolution. — op_depthwise_conv","title":"General N-D depthwise convolution. — op_depthwise_conv","text":"ops supports 1D 2D depthwise convolution.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_depthwise_conv.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"General N-D depthwise convolution. — op_depthwise_conv","text":"","code":"op_depthwise_conv(   inputs,   kernel,   strides = 1L,   padding = \"valid\",   data_format = NULL,   dilation_rate = 1L )"},{"path":"https://keras3.posit.co/dev/reference/op_depthwise_conv.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"General N-D depthwise convolution. — op_depthwise_conv","text":"inputs Tensor rank N+2. inputs shape (batch_size,) + inputs_spatial_shape + (num_channels,) data_format = \"channels_last\", (batch_size, num_channels) + inputs_spatial_shape data_format = \"channels_first\". kernel Tensor rank N+2. kernel shape [kernel_spatial_shape, num_input_channels, num_channels_multiplier], num_input_channels match number channels inputs. strides int int tuple/list len(inputs_spatial_shape), specifying strides convolution along spatial dimension. strides int, every spatial dimension shares strides. padding string, either \"valid\" \"\". \"valid\" means padding applied, \"\" results padding evenly left/right /input output height/width dimension input strides = 1. data_format string, either \"channels_last\" \"channels_first\". data_format determines ordering dimensions inputs. data_format = \"channels_last\", inputs shape (batch_size, ..., channels) data_format = \"channels_first\", inputs shape (batch_size, channels, ...). dilation_rate int int tuple/list len(inputs_spatial_shape), specifying dilation rate use dilated convolution. dilation_rate int, every spatial dimension shares dilation_rate.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_depthwise_conv.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"General N-D depthwise convolution. — op_depthwise_conv","text":"tensor rank N+2, result depthwise conv operation.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_det.html","id":null,"dir":"Reference","previous_headings":"","what":"Computes the determinant of a square tensor. — op_det","title":"Computes the determinant of a square tensor. — op_det","text":"Computes determinant square tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_det.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes the determinant of a square tensor. — op_det","text":"","code":"op_det(x)"},{"path":"https://keras3.posit.co/dev/reference/op_det.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Computes the determinant of a square tensor. — op_det","text":"x Input tensor shape (..., M, M).","code":""},{"path":"https://keras3.posit.co/dev/reference/op_det.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Computes the determinant of a square tensor. — op_det","text":"tensor shape (...) representing determinant x.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_diag.html","id":null,"dir":"Reference","previous_headings":"","what":"Extract a diagonal or construct a diagonal array. — op_diag","title":"Extract a diagonal or construct a diagonal array. — op_diag","text":"Extract diagonal construct diagonal array.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_diag.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Extract a diagonal or construct a diagonal array. — op_diag","text":"","code":"op_diag(x, k = 0L)"},{"path":"https://keras3.posit.co/dev/reference/op_diag.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Extract a diagonal or construct a diagonal array. — op_diag","text":"x Input tensor. x 2-D, returns k-th diagonal x. x 1-D, return 2-D tensor x k-th diagonal. k diagonal consider. Defaults 0. Use k > 0 diagonals main diagonal, k < 0 diagonals main diagonal.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_diag.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Extract a diagonal or construct a diagonal array. — op_diag","text":"extracted diagonal constructed diagonal tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_diag.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Extract a diagonal or construct a diagonal array. — op_diag","text":"","code":"x <- op_arange(9L) |> op_reshape(c(3, 3)) x ## tf.Tensor( ## [[1 2 3] ##  [4 5 6] ##  [7 8 9]], shape=(3, 3), dtype=int32) op_diag(x) ## tf.Tensor([1 5 9], shape=(3), dtype=int32) op_diag(x, k = 1) ## tf.Tensor([2 6], shape=(2), dtype=int32) op_diag(x, k = -1) ## tf.Tensor([4 8], shape=(2), dtype=int32) op_diag(op_diag(x)) ## tf.Tensor( ## [[1 0 0] ##  [0 5 0] ##  [0 0 9]], shape=(3, 3), dtype=int32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_diagflat.html","id":null,"dir":"Reference","previous_headings":"","what":"Create a two-dimensional array with the flattened input diagonal. — op_diagflat","title":"Create a two-dimensional array with the flattened input diagonal. — op_diagflat","text":"k-th diagonal.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_diagflat.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Create a two-dimensional array with the flattened input diagonal. — op_diagflat","text":"","code":"op_diagflat(x, k = 0L)"},{"path":"https://keras3.posit.co/dev/reference/op_diagflat.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Create a two-dimensional array with the flattened input diagonal. — op_diagflat","text":"x Input tensor flattened placed diagonal. k diagonal place flattened input. Defaults 0. Use k > 0 diagonals main diagonal, k < 0 diagonals main diagonal.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_diagflat.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Create a two-dimensional array with the flattened input diagonal. — op_diagflat","text":"2-D tensor flattened input specified diagonal.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_diagonal.html","id":null,"dir":"Reference","previous_headings":"","what":"Return specified diagonals. — op_diagonal","title":"Return specified diagonals. — op_diagonal","text":"x 2-D, returns diagonal x given offset, .e., collection elements form x[, +offset]. x two dimensions, axes specified axis1 axis2 used determine 2-D sub-array whose diagonal returned. shape resulting array can determined removing axis1 axis2 appending index right equal size resulting diagonals.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_diagonal.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Return specified diagonals. — op_diagonal","text":"","code":"op_diagonal(x, offset = 0L, axis1 = 1L, axis2 = 2L)"},{"path":"https://keras3.posit.co/dev/reference/op_diagonal.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Return specified diagonals. — op_diagonal","text":"x Input tensor. offset Offset diagonal main diagonal. Can positive negative. Defaults 0 (main diagonal). axis1 Axis used first axis 2-D sub-arrays. Defaults 1 (first axis). axis2 Axis used second axis 2-D sub-arrays. Defaults 2 (second axis).","code":""},{"path":"https://keras3.posit.co/dev/reference/op_diagonal.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Return specified diagonals. — op_diagonal","text":"Tensor diagonals.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_diagonal.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Return specified diagonals. — op_diagonal","text":"","code":"x <- op_arange(4L) |> op_reshape(c(2, 2)) x ## tf.Tensor( ## [[1 2] ##  [3 4]], shape=(2, 2), dtype=int32) op_diagonal(x) ## tf.Tensor([1 4], shape=(2), dtype=int32) op_diagonal(x, offset = 1) ## tf.Tensor([2], shape=(1), dtype=int32) x <- op_array(1:8) |> op_reshape(c(2, 2, 2)) x ## tf.Tensor( ## [[[1 2] ##   [3 4]] ## ##  [[5 6] ##   [7 8]]], shape=(2, 2, 2), dtype=int32) x |> op_diagonal(0) ## tf.Tensor( ## [[1 7] ##  [2 8]], shape=(2, 2), dtype=int32) x |> op_diagonal(0, 1, 2) # same as above, the default ## tf.Tensor( ## [[1 7] ##  [2 8]], shape=(2, 2), dtype=int32) x |> op_diagonal(0, 2, 3) ## tf.Tensor( ## [[1 4] ##  [5 8]], shape=(2, 2), dtype=int32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_diff.html","id":null,"dir":"Reference","previous_headings":"","what":"Calculate the n-th discrete difference along the given axis. — op_diff","title":"Calculate the n-th discrete difference along the given axis. — op_diff","text":"first difference given [] = [+1] - [] along given axis, higher differences calculated using diff recursively.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_diff.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Calculate the n-th discrete difference along the given axis. — op_diff","text":"","code":"op_diff(a, n = 1L, axis = -1L)"},{"path":"https://keras3.posit.co/dev/reference/op_diff.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Calculate the n-th discrete difference along the given axis. — op_diff","text":"Input tensor. n number times values differenced. Defaults 1. axis Axis compute discrete difference(s) along. Defaults -1 (last axis).","code":""},{"path":"https://keras3.posit.co/dev/reference/op_diff.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Calculate the n-th discrete difference along the given axis. — op_diff","text":"Tensor diagonals.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_diff.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Calculate the n-th discrete difference along the given axis. — op_diff","text":"","code":"x <- op_array(c(1, 2, 4, 7, 0)) op_diff(x) ## tf.Tensor([ 1.  2.  3. -7.], shape=(4), dtype=float32) op_diff(x, n = 2) ## tf.Tensor([  1.   1. -10.], shape=(3), dtype=float32) x <- op_array(rbind(c(1, 3, 6, 10),                     c(0, 5, 6, 8))) op_diff(x) ## tf.Tensor( ## [[2. 3. 4.] ##  [5. 1. 2.]], shape=(2, 3), dtype=float64) op_diff(x, axis = 1) ## tf.Tensor([[-1.  2.  0. -2.]], shape=(1, 4), dtype=float64)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_digitize.html","id":null,"dir":"Reference","previous_headings":"","what":"Returns the indices of the bins to which each value in x belongs. — op_digitize","title":"Returns the indices of the bins to which each value in x belongs. — op_digitize","text":"Returns indices bins value x belongs.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_digitize.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Returns the indices of the bins to which each value in x belongs. — op_digitize","text":"","code":"op_digitize(x, bins, zero_indexed = FALSE)"},{"path":"https://keras3.posit.co/dev/reference/op_digitize.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Returns the indices of the bins to which each value in x belongs. — op_digitize","text":"x Input array binned. bins Array bins. one-dimensional monotonically increasing. zero_indexed TRUE, returned indices zero-based (0 encodes first bin); FALSE (default), returned indices one-based (1 encodes first bin).","code":""},{"path":"https://keras3.posit.co/dev/reference/op_digitize.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Returns the indices of the bins to which each value in x belongs. — op_digitize","text":"Output array indices, shape x.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_digitize.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Returns the indices of the bins to which each value in x belongs. — op_digitize","text":"","code":"x <- op_array(c(-1,    0, 1,    3, 3.5,      4.5,  1.6,    8)) bins <- array(c(    0,       3,         4.5,            7)) op_digitize(x, bins) ## tf.Tensor([1 2 2 3 3 4 2 5], shape=(8), dtype=int32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_divide.html","id":null,"dir":"Reference","previous_headings":"","what":"Divide arguments element-wise. — op_divide","title":"Divide arguments element-wise. — op_divide","text":"Note function automatically called using R operator * tensor.","code":"(x <- op_arange(4)) ## tf.Tensor([1. 2. 3. 4.], shape=(4), dtype=float32) op_divide(x, 2) ## tf.Tensor([0.5 1.  1.5 2. ], shape=(4), dtype=float32) x / 2 ## tf.Tensor([0.5 1.  1.5 2. ], shape=(4), dtype=float32)"},{"path":"https://keras3.posit.co/dev/reference/op_divide.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Divide arguments element-wise. — op_divide","text":"","code":"op_divide(x1, x2)"},{"path":"https://keras3.posit.co/dev/reference/op_divide.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Divide arguments element-wise. — op_divide","text":"x1 First input tensor. x2 Second input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_divide.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Divide arguments element-wise. — op_divide","text":"Output tensor, quotient x1/x2, element-wise.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_divide.html","id":"example","dir":"Reference","previous_headings":"","what":"Example","title":"Divide arguments element-wise. — op_divide","text":"","code":"op_divide(3, 2) ## tf.Tensor(1.5, shape=(), dtype=float32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_divide_no_nan.html","id":null,"dir":"Reference","previous_headings":"","what":"Safe element-wise division which returns 0 where the denominator is 0. — op_divide_no_nan","title":"Safe element-wise division which returns 0 where the denominator is 0. — op_divide_no_nan","text":"Safe element-wise division returns 0 denominator 0.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_divide_no_nan.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Safe element-wise division which returns 0 where the denominator is 0. — op_divide_no_nan","text":"","code":"op_divide_no_nan(x1, x2)"},{"path":"https://keras3.posit.co/dev/reference/op_divide_no_nan.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Safe element-wise division which returns 0 where the denominator is 0. — op_divide_no_nan","text":"x1 First input tensor. x2 Second input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_divide_no_nan.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Safe element-wise division which returns 0 where the denominator is 0. — op_divide_no_nan","text":"quotient x1/x2, element-wise, zero x2 zero.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_dot.html","id":null,"dir":"Reference","previous_headings":"","what":"Dot product of two tensors. — op_dot","title":"Dot product of two tensors. — op_dot","text":"x1 x2 1-D tensors, inner product vectors (without complex conjugation). x1 x2 2-D tensors, matrix multiplication. either x1 x2 0-D (scalar), equivalent x1 * x2. x1 N-D tensor x2 1-D tensor, sum product last axis x1 x2. x1 N-D tensor x2 M-D tensor (M >= 2), sum product last axis x1 second--last axis x2: dot(x1, x2)[,j,k,m] = sum([,j,:] * b[k,:,m]).","code":""},{"path":"https://keras3.posit.co/dev/reference/op_dot.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Dot product of two tensors. — op_dot","text":"","code":"op_dot(x1, x2)"},{"path":"https://keras3.posit.co/dev/reference/op_dot.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Dot product of two tensors. — op_dot","text":"x1 First argument. x2 Second argument.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_dot.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Dot product of two tensors. — op_dot","text":"Dot product x1 x2.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_dot.html","id":"note","dir":"Reference","previous_headings":"","what":"Note","title":"Dot product of two tensors. — op_dot","text":"Torch backend accept 0-D tensors arguments.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_dot_product_attention.html","id":null,"dir":"Reference","previous_headings":"","what":"Scaled dot product attention function. — op_dot_product_attention","title":"Scaled dot product attention function. — op_dot_product_attention","text":"Computes attention function Q (query), K (key), V(value): attention(Q, K, V) = softmax(Q * K / sqrt(d)) * V. define logits output Q * K probs output softmax. Throughout function, utilize following notation represent shape array: B: batch size S: length key/value T: length query N: number attention heads H: dimensions attention head K: number key/value heads G: number groups, equals N // K","code":""},{"path":"https://keras3.posit.co/dev/reference/op_dot_product_attention.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Scaled dot product attention function. — op_dot_product_attention","text":"","code":"op_dot_product_attention(   query,   key,   value,   bias = NULL,   mask = NULL,   scale = NULL,   is_causal = FALSE,   flash_attention = NULL,   attn_logits_soft_cap = NULL )"},{"path":"https://keras3.posit.co/dev/reference/op_dot_product_attention.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Scaled dot product attention function. — op_dot_product_attention","text":"query query array shape (B, T, N, H). key key array shape (B, S, K, H). K equals N, multi-headed attention (MHA) performed. Otherwise, grouped query attention (GQA) performed N multiple K. multi-query attention (MQA) performed K==1 (special case GQA). value value array shape key. bias Optional bias array added logits. shape must broadcastable (B, N, T, S). mask Optional mask array used filter logits. boolean mask TRUE indicates element take part attention. additive mask, users pass bias. shape must broadcastable (B, N, T, S). scale Optional scale logits. NULL, scale set 1.0 / sqrt(H). is_causal Whether apply causal mask. flash_attention Whether use flash attention. NULL, attempt use flash attention required conditions met. Typically, inputs must float16 bfloat16 dtype input layout requirements may vary depending backend. attn_logits_soft_cap Optional numeric cap attention logits softmax. supported JAX TPU backend.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_dot_product_attention.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Scaled dot product attention function. — op_dot_product_attention","text":"array attention output shape query.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_dot_product_attention.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Scaled dot product attention function. — op_dot_product_attention","text":"","code":"query = random_normal(c(2, 4, 8, 16)) key = random_normal(c(2, 6, 8, 16)) value = random_normal(c(2, 6, 8, 16)) op_dot_product_attention(query, key, value) |> op_shape() ## shape(2, 4, 8, 16)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_dtype.html","id":null,"dir":"Reference","previous_headings":"","what":"Return the dtype of the tensor input as a standardized string. — op_dtype","title":"Return the dtype of the tensor input as a standardized string. — op_dtype","text":"Note due standardization, dtype compare equal backend-specific version dtype.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_dtype.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Return the dtype of the tensor input as a standardized string. — op_dtype","text":"","code":"op_dtype(x)"},{"path":"https://keras3.posit.co/dev/reference/op_dtype.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Return the dtype of the tensor input as a standardized string. — op_dtype","text":"x tensor. function try access dtype attribute input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_dtype.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Return the dtype of the tensor input as a standardized string. — op_dtype","text":"string indicating dtype input tensor, e.g. \"float32\".","code":""},{"path":"https://keras3.posit.co/dev/reference/op_dtype.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Return the dtype of the tensor input as a standardized string. — op_dtype","text":"","code":"x <- op_zeros(c(8, 12)) op_dtype(x) ## [1] \"float32\""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_eig.html","id":null,"dir":"Reference","previous_headings":"","what":"Computes the eigenvalues and eigenvectors of a square matrix. — op_eig","title":"Computes the eigenvalues and eigenvectors of a square matrix. — op_eig","text":"Computes eigenvalues eigenvectors square matrix.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_eig.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes the eigenvalues and eigenvectors of a square matrix. — op_eig","text":"","code":"op_eig(x)"},{"path":"https://keras3.posit.co/dev/reference/op_eig.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Computes the eigenvalues and eigenvectors of a square matrix. — op_eig","text":"x Input tensor shape (..., M, M).","code":""},{"path":"https://keras3.posit.co/dev/reference/op_eig.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Computes the eigenvalues and eigenvectors of a square matrix. — op_eig","text":"list two tensors: tensor shape (..., M) containing eigenvalues tensor shape (..., M, M) containing eigenvectors.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_eigh.html","id":null,"dir":"Reference","previous_headings":"","what":"Computes the eigenvalues and eigenvectors of a complex Hermitian. — op_eigh","title":"Computes the eigenvalues and eigenvectors of a complex Hermitian. — op_eigh","text":"Computes eigenvalues eigenvectors complex Hermitian.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_eigh.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes the eigenvalues and eigenvectors of a complex Hermitian. — op_eigh","text":"","code":"op_eigh(x)"},{"path":"https://keras3.posit.co/dev/reference/op_eigh.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Computes the eigenvalues and eigenvectors of a complex Hermitian. — op_eigh","text":"x Input tensor shape (..., M, M).","code":""},{"path":"https://keras3.posit.co/dev/reference/op_eigh.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Computes the eigenvalues and eigenvectors of a complex Hermitian. — op_eigh","text":"list two tensors: tensor shape (..., M) containing eigenvalues tensor shape (..., M, M) containing eigenvectors.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_einsum.html","id":null,"dir":"Reference","previous_headings":"","what":"Evaluates the Einstein summation convention on the operands. — op_einsum","title":"Evaluates the Einstein summation convention on the operands. — op_einsum","text":"Evaluates Einstein summation convention operands.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_einsum.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Evaluates the Einstein summation convention on the operands. — op_einsum","text":"","code":"op_einsum(subscripts, ...)"},{"path":"https://keras3.posit.co/dev/reference/op_einsum.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Evaluates the Einstein summation convention on the operands. — op_einsum","text":"subscripts Specifies subscripts summation comma separated list subscript labels. implicit (classical Einstein summation) calculation performed unless explicit indicator -> included well subscript labels precise output form. ... operands compute Einstein sum unnamed arguments. Additional named arguments forwarded underlying backend.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_einsum.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Evaluates the Einstein summation convention on the operands. — op_einsum","text":"calculation based Einstein summation convention.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_einsum.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Evaluates the Einstein summation convention on the operands. — op_einsum","text":"Trace matrix:     Extract diagonal:     Sum axis:     higher dimensional tensors summing single axis can done ellipsis:     Compute matrix transpose reorder number axes:         Matrix vector multiplication:","code":"a <- op_arange(25) |> op_reshape(c(5, 5)) b <- op_arange(5) c <- op_arange(6) |> op_reshape(c(2, 3)) op_einsum(\"ii\", a) op_trace(a) ## tf.Tensor(65.0, shape=(), dtype=float32) ## tf.Tensor(65.0, shape=(), dtype=float32) op_einsum(\"ii -> i\", a) op_diag(a) ## tf.Tensor([ 1.  7. 13. 19. 25.], shape=(5), dtype=float32) ## tf.Tensor([ 1.  7. 13. 19. 25.], shape=(5), dtype=float32) op_einsum(\"ij -> i\", a) op_sum(a, axis = 2) ## tf.Tensor([ 15.  40.  65.  90. 115.], shape=(5), dtype=float32) ## tf.Tensor([ 15.  40.  65.  90. 115.], shape=(5), dtype=float32) op_einsum(\"...j -> ...\", a) op_sum(a, axis = -1) ## tf.Tensor([ 15.  40.  65.  90. 115.], shape=(5), dtype=float32) ## tf.Tensor([ 15.  40.  65.  90. 115.], shape=(5), dtype=float32) op_einsum(\"ji\", c) # return c unchanged ## tf.Tensor( ## [[1. 2. 3.] ##  [4. 5. 6.]], shape=(2, 3), dtype=float32) op_einsum(\"ij -> ji\", c) # transpose op_transpose(c)          # same as above ## tf.Tensor( ## [[1. 4.] ##  [2. 5.] ##  [3. 6.]], shape=(3, 2), dtype=float32) ## tf.Tensor( ## [[1. 4.] ##  [2. 5.] ##  [3. 6.]], shape=(3, 2), dtype=float32) op_einsum(\"ij, j\", a, b) op_einsum(\"...j, j\", a, b) a %*% b op_matmul(a, b) ## tf.Tensor([ 55. 130. 205. 280. 355.], shape=(5), dtype=float32) ## tf.Tensor([ 55. 130. 205. 280. 355.], shape=(5), dtype=float32) ## tf.Tensor([ 55. 130. 205. 280. 355.], shape=(5), dtype=float32) ## tf.Tensor([ 55. 130. 205. 280. 355.], shape=(5), dtype=float32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_elu.html","id":null,"dir":"Reference","previous_headings":"","what":"Exponential Linear Unit activation function. — op_elu","title":"Exponential Linear Unit activation function. — op_elu","text":"defined : f(x) =  alpha * (exp(x) - 1.) x < 0, f(x) = x x >= 0.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_elu.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Exponential Linear Unit activation function. — op_elu","text":"","code":"op_elu(x, alpha = 1)"},{"path":"https://keras3.posit.co/dev/reference/op_elu.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Exponential Linear Unit activation function. — op_elu","text":"x Input tensor. alpha scalar, slope positive section. Defaults 1.0.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_elu.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Exponential Linear Unit activation function. — op_elu","text":"tensor shape x.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_elu.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Exponential Linear Unit activation function. — op_elu","text":"","code":"x <- op_array(c(-1., 0., 1.)) op_elu(x) ## tf.Tensor([-0.63212055  0.          1.        ], shape=(3), dtype=float32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_empty.html","id":null,"dir":"Reference","previous_headings":"","what":"Return a tensor of given shape and type filled with uninitialized data. — op_empty","title":"Return a tensor of given shape and type filled with uninitialized data. — op_empty","text":"Return tensor given shape type filled uninitialized data.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_empty.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Return a tensor of given shape and type filled with uninitialized data. — op_empty","text":"","code":"op_empty(shape, dtype = NULL)"},{"path":"https://keras3.posit.co/dev/reference/op_empty.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Return a tensor of given shape and type filled with uninitialized data. — op_empty","text":"shape Shape empty tensor. dtype Desired data type empty tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_empty.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Return a tensor of given shape and type filled with uninitialized data. — op_empty","text":"empty tensor.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_equal.html","id":null,"dir":"Reference","previous_headings":"","what":"Returns (x1 == x2) element-wise. — op_equal","title":"Returns (x1 == x2) element-wise. — op_equal","text":"Note function automatically called using R operator == tensor.","code":"(x <- op_arange(4)) ## tf.Tensor([1. 2. 3. 4.], shape=(4), dtype=float32) op_equal(x, 2) ## tf.Tensor([False  True False False], shape=(4), dtype=bool) x == 2 ## tf.Tensor([False  True False False], shape=(4), dtype=bool)"},{"path":"https://keras3.posit.co/dev/reference/op_equal.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Returns (x1 == x2) element-wise. — op_equal","text":"","code":"op_equal(x1, x2)"},{"path":"https://keras3.posit.co/dev/reference/op_equal.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Returns (x1 == x2) element-wise. — op_equal","text":"x1 Tensor compare. x2 Tensor compare.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_equal.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Returns (x1 == x2) element-wise. — op_equal","text":"Output tensor, element-wise comparison x1 x2.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_erf.html","id":null,"dir":"Reference","previous_headings":"","what":"Computes the error function of x, element-wise. — op_erf","title":"Computes the error function of x, element-wise. — op_erf","text":"Computes error function x, element-wise.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_erf.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes the error function of x, element-wise. — op_erf","text":"","code":"op_erf(x)"},{"path":"https://keras3.posit.co/dev/reference/op_erf.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Computes the error function of x, element-wise. — op_erf","text":"x Input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_erf.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Computes the error function of x, element-wise. — op_erf","text":"tensor dtype x.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_erf.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Computes the error function of x, element-wise. — op_erf","text":"","code":"x <- op_array(c(-3, -2, -1, 0, 1)) op_erf(x) ## tf.Tensor([-0.999978   -0.99532235 -0.84270084  0.          0.84270084], shape=(5), dtype=float32) # array([-0.99998 , -0.99532, -0.842701,  0.,  0.842701], dtype=float32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_erfinv.html","id":null,"dir":"Reference","previous_headings":"","what":"Computes the inverse error function of x, element-wise. — op_erfinv","title":"Computes the inverse error function of x, element-wise. — op_erfinv","text":"Computes inverse error function x, element-wise.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_erfinv.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes the inverse error function of x, element-wise. — op_erfinv","text":"","code":"op_erfinv(x)"},{"path":"https://keras3.posit.co/dev/reference/op_erfinv.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Computes the inverse error function of x, element-wise. — op_erfinv","text":"x Input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_erfinv.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Computes the inverse error function of x, element-wise. — op_erfinv","text":"tensor dtype x.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_erfinv.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Computes the inverse error function of x, element-wise. — op_erfinv","text":"","code":"x <- op_array(c(-0.5, -0.2, -0.1, 0.0, 0.3)) op_erfinv(x) ## tf.Tensor([-0.4769363  -0.17914344 -0.088856    0.          0.27246267], shape=(5), dtype=float32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_exp.html","id":null,"dir":"Reference","previous_headings":"","what":"Calculate the exponential of all elements in the input tensor. — op_exp","title":"Calculate the exponential of all elements in the input tensor. — op_exp","text":"Calculate exponential elements input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_exp.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Calculate the exponential of all elements in the input tensor. — op_exp","text":"","code":"op_exp(x)"},{"path":"https://keras3.posit.co/dev/reference/op_exp.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Calculate the exponential of all elements in the input tensor. — op_exp","text":"x Input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_exp.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Calculate the exponential of all elements in the input tensor. — op_exp","text":"Output tensor, element-wise exponential x.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_exp2.html","id":null,"dir":"Reference","previous_headings":"","what":"Calculate the base-2 exponential of all elements in the input tensor. — op_exp2","title":"Calculate the base-2 exponential of all elements in the input tensor. — op_exp2","text":"Calculate base-2 exponential elements input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_exp2.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Calculate the base-2 exponential of all elements in the input tensor. — op_exp2","text":"","code":"op_exp2(x)"},{"path":"https://keras3.posit.co/dev/reference/op_exp2.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Calculate the base-2 exponential of all elements in the input tensor. — op_exp2","text":"x Input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_exp2.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Calculate the base-2 exponential of all elements in the input tensor. — op_exp2","text":"Output tensor, element-wise base-2 exponential x.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_expand_dims.html","id":null,"dir":"Reference","previous_headings":"","what":"Expand the shape of a tensor. — op_expand_dims","title":"Expand the shape of a tensor. — op_expand_dims","text":"Insert new axis axis position expanded tensor shape.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_expand_dims.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Expand the shape of a tensor. — op_expand_dims","text":"","code":"op_expand_dims(x, axis)"},{"path":"https://keras3.posit.co/dev/reference/op_expand_dims.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Expand the shape of a tensor. — op_expand_dims","text":"x Input tensor. axis Position expanded axes new axis (axes) placed.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_expand_dims.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Expand the shape of a tensor. — op_expand_dims","text":"Output tensor number dimensions increased.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_expm1.html","id":null,"dir":"Reference","previous_headings":"","what":"Calculate exp(x) - 1 for all elements in the tensor. — op_expm1","title":"Calculate exp(x) - 1 for all elements in the tensor. — op_expm1","text":"Calculate exp(x) - 1 elements tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_expm1.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Calculate exp(x) - 1 for all elements in the tensor. — op_expm1","text":"","code":"op_expm1(x)"},{"path":"https://keras3.posit.co/dev/reference/op_expm1.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Calculate exp(x) - 1 for all elements in the tensor. — op_expm1","text":"x Input values.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_expm1.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Calculate exp(x) - 1 for all elements in the tensor. — op_expm1","text":"Output tensor, element-wise exponential minus one.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_extract_sequences.html","id":null,"dir":"Reference","previous_headings":"","what":"Expands the dimension of last axis into sequences of sequence_length. — op_extract_sequences","title":"Expands the dimension of last axis into sequences of sequence_length. — op_extract_sequences","text":"Slides window size sequence_length last axis input stride sequence_stride, replacing last axis [num_sequences, sequence_length] sequences. dimension along last axis N, number sequences can computed : num_sequences = 1 + (N - sequence_length) // sequence_stride","code":""},{"path":"https://keras3.posit.co/dev/reference/op_extract_sequences.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Expands the dimension of last axis into sequences of sequence_length. — op_extract_sequences","text":"","code":"op_extract_sequences(x, sequence_length, sequence_stride)"},{"path":"https://keras3.posit.co/dev/reference/op_extract_sequences.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Expands the dimension of last axis into sequences of sequence_length. — op_extract_sequences","text":"x Input tensor. sequence_length integer representing sequences length. sequence_stride integer representing sequences hop size.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_extract_sequences.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Expands the dimension of last axis into sequences of sequence_length. — op_extract_sequences","text":"tensor sequences shape [..., num_sequences, sequence_length].","code":""},{"path":"https://keras3.posit.co/dev/reference/op_extract_sequences.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Expands the dimension of last axis into sequences of sequence_length. — op_extract_sequences","text":"","code":"x <- op_convert_to_tensor(1:6) op_extract_sequences(x, 3, 2) ## tf.Tensor( ## [[1 2 3] ##  [3 4 5]], shape=(2, 3), dtype=int32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_eye.html","id":null,"dir":"Reference","previous_headings":"","what":"Return a 2-D tensor with ones on the diagonal and zeros elsewhere. — op_eye","title":"Return a 2-D tensor with ones on the diagonal and zeros elsewhere. — op_eye","text":"Return 2-D tensor ones diagonal zeros elsewhere.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_eye.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Return a 2-D tensor with ones on the diagonal and zeros elsewhere. — op_eye","text":"","code":"op_eye(N, M = NULL, k = 0L, dtype = NULL)"},{"path":"https://keras3.posit.co/dev/reference/op_eye.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Return a 2-D tensor with ones on the diagonal and zeros elsewhere. — op_eye","text":"N Number rows output. M Number columns output. NULL, defaults N. k Index diagonal: 0 (default) refers main diagonal, positive value refers upper diagonal, negative value lower diagonal. dtype Data type returned tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_eye.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Return a 2-D tensor with ones on the diagonal and zeros elsewhere. — op_eye","text":"Tensor ones k-th diagonal zeros elsewhere.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_fft.html","id":null,"dir":"Reference","previous_headings":"","what":"Computes the Fast Fourier Transform along last axis of input. — op_fft","title":"Computes the Fast Fourier Transform along last axis of input. — op_fft","text":"Computes Fast Fourier Transform along last axis input.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_fft.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes the Fast Fourier Transform along last axis of input. — op_fft","text":"","code":"op_fft(x)"},{"path":"https://keras3.posit.co/dev/reference/op_fft.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Computes the Fast Fourier Transform along last axis of input. — op_fft","text":"x list real imaginary parts input tensor. tensors provided floating type.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_fft.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Computes the Fast Fourier Transform along last axis of input. — op_fft","text":"list containing two tensors - real imaginary parts output tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_fft.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Computes the Fast Fourier Transform along last axis of input. — op_fft","text":"","code":"x = c(op_array(c(1., 2.)),       op_array(c(0., 1.))) op_fft(x) ## [[1]] ## tf.Tensor([ 3. -1.], shape=(2), dtype=float32) ## ## [[2]] ## tf.Tensor([ 1. -1.], shape=(2), dtype=float32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_fft2.html","id":null,"dir":"Reference","previous_headings":"","what":"Computes the 2D Fast Fourier Transform along the last two axes of input. — op_fft2","title":"Computes the 2D Fast Fourier Transform along the last two axes of input. — op_fft2","text":"Computes 2D Fast Fourier Transform along last two axes input.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_fft2.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes the 2D Fast Fourier Transform along the last two axes of input. — op_fft2","text":"","code":"op_fft2(x)"},{"path":"https://keras3.posit.co/dev/reference/op_fft2.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Computes the 2D Fast Fourier Transform along the last two axes of input. — op_fft2","text":"x list real imaginary parts input tensor. tensors provided floating type.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_fft2.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Computes the 2D Fast Fourier Transform along the last two axes of input. — op_fft2","text":"list containing two tensors - real imaginary parts output.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_fft2.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Computes the 2D Fast Fourier Transform along the last two axes of input. — op_fft2","text":"","code":"x <- c(op_array(rbind(c(1, 2),                      c(2, 1))),        op_array(rbind(c(0, 1),                      c(1, 0)))) op_fft2(x) ## [[1]] ## tf.Tensor( ## [[ 6.  0.] ##  [ 0. -2.]], shape=(2, 2), dtype=float64) ## ## [[2]] ## tf.Tensor( ## [[ 2.  0.] ##  [ 0. -2.]], shape=(2, 2), dtype=float64)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_flip.html","id":null,"dir":"Reference","previous_headings":"","what":"Reverse the order of elements in the tensor along the given axis. — op_flip","title":"Reverse the order of elements in the tensor along the given axis. — op_flip","text":"shape tensor preserved, elements reordered.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_flip.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Reverse the order of elements in the tensor along the given axis. — op_flip","text":"","code":"op_flip(x, axis = NULL)"},{"path":"https://keras3.posit.co/dev/reference/op_flip.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Reverse the order of elements in the tensor along the given axis. — op_flip","text":"x Input tensor. axis Axis axes along flip tensor. default, axis = NULL, flip axes input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_flip.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Reverse the order of elements in the tensor along the given axis. — op_flip","text":"Output tensor entries axis reversed.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_floor.html","id":null,"dir":"Reference","previous_headings":"","what":"Return the floor of the input, element-wise. — op_floor","title":"Return the floor of the input, element-wise. — op_floor","text":"floor scalar x largest integer , <= x.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_floor.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Return the floor of the input, element-wise. — op_floor","text":"","code":"op_floor(x)"},{"path":"https://keras3.posit.co/dev/reference/op_floor.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Return the floor of the input, element-wise. — op_floor","text":"x Input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_floor.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Return the floor of the input, element-wise. — op_floor","text":"Output tensor, element-wise floor x.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_floor_divide.html","id":null,"dir":"Reference","previous_headings":"","what":"Returns the largest integer smaller or equal to the division of inputs. — op_floor_divide","title":"Returns the largest integer smaller or equal to the division of inputs. — op_floor_divide","text":"Note function automatically called using R operator %/% tensor.","code":"(x <- op_arange(10)) ## tf.Tensor([ 1.  2.  3.  4.  5.  6.  7.  8.  9. 10.], shape=(10), dtype=float32) op_floor_divide(x, 2) ## tf.Tensor([0. 1. 1. 2. 2. 3. 3. 4. 4. 5.], shape=(10), dtype=float32) x %/% 2 ## tf.Tensor([0. 1. 1. 2. 2. 3. 3. 4. 4. 5.], shape=(10), dtype=float32)"},{"path":"https://keras3.posit.co/dev/reference/op_floor_divide.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Returns the largest integer smaller or equal to the division of inputs. — op_floor_divide","text":"","code":"op_floor_divide(x1, x2)"},{"path":"https://keras3.posit.co/dev/reference/op_floor_divide.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Returns the largest integer smaller or equal to the division of inputs. — op_floor_divide","text":"x1 Numerator. x2 Denominator.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_floor_divide.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Returns the largest integer smaller or equal to the division of inputs. — op_floor_divide","text":"Output tensor, y <- floor(x1/x2)","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_fori_loop.html","id":null,"dir":"Reference","previous_headings":"","what":"For loop implementation. — op_fori_loop","title":"For loop implementation. — op_fori_loop","text":"loop implementation.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_fori_loop.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"For loop implementation. — op_fori_loop","text":"","code":"op_fori_loop(lower, upper, body_fun, init_val)"},{"path":"https://keras3.posit.co/dev/reference/op_fori_loop.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"For loop implementation. — op_fori_loop","text":"lower initial value loop variable. upper upper bound loop variable. body_fun callable represents loop body. Must take two arguments: loop variable loop state. loop state updated returned function. init_val initial value loop state.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_fori_loop.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"For loop implementation. — op_fori_loop","text":"final state loop.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_fori_loop.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"For loop implementation. — op_fori_loop","text":"","code":"lower <- 0L upper <- 10L body_fun <- function(i, state) state + i init_state <- 0L final_state <- op_fori_loop(lower, upper, body_fun, init_state) final_state ## tf.Tensor(45, shape=(), dtype=int32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_full.html","id":null,"dir":"Reference","previous_headings":"","what":"Return a new tensor of given shape and type, filled with fill_value. — op_full","title":"Return a new tensor of given shape and type, filled with fill_value. — op_full","text":"Return new tensor given shape type, filled fill_value.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_full.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Return a new tensor of given shape and type, filled with fill_value. — op_full","text":"","code":"op_full(shape, fill_value, dtype = NULL)"},{"path":"https://keras3.posit.co/dev/reference/op_full.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Return a new tensor of given shape and type, filled with fill_value. — op_full","text":"shape Shape new tensor. fill_value Fill value. dtype Desired data type tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_full.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Return a new tensor of given shape and type, filled with fill_value. — op_full","text":"Output tensor.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_full_like.html","id":null,"dir":"Reference","previous_headings":"","what":"Return a full tensor with the same shape and type as the given tensor. — op_full_like","title":"Return a full tensor with the same shape and type as the given tensor. — op_full_like","text":"Return full tensor shape type given tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_full_like.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Return a full tensor with the same shape and type as the given tensor. — op_full_like","text":"","code":"op_full_like(x, fill_value, dtype = NULL)"},{"path":"https://keras3.posit.co/dev/reference/op_full_like.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Return a full tensor with the same shape and type as the given tensor. — op_full_like","text":"x Input tensor. fill_value Fill value. dtype Overrides data type result.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_full_like.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Return a full tensor with the same shape and type as the given tensor. — op_full_like","text":"Tensor fill_value shape type x.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_gelu.html","id":null,"dir":"Reference","previous_headings":"","what":"Gaussian Error Linear Unit (GELU) activation function. — op_gelu","title":"Gaussian Error Linear Unit (GELU) activation function. — op_gelu","text":"approximate TRUE, defined : f(x) = 0.5 * x * (1 + tanh(sqrt(2 / pi) * (x + 0.044715 * x^3))) approximate FALSE, defined : f(x) = x * P(X <= x) = 0.5 * x * (1 + erf(x / sqrt(2))), P(X) ~ N(0, 1).","code":""},{"path":"https://keras3.posit.co/dev/reference/op_gelu.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Gaussian Error Linear Unit (GELU) activation function. — op_gelu","text":"","code":"op_gelu(x, approximate = TRUE)"},{"path":"https://keras3.posit.co/dev/reference/op_gelu.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Gaussian Error Linear Unit (GELU) activation function. — op_gelu","text":"x Input tensor. approximate Approximate version GELU activation. Defaults TRUE.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_gelu.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Gaussian Error Linear Unit (GELU) activation function. — op_gelu","text":"tensor shape x.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_gelu.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Gaussian Error Linear Unit (GELU) activation function. — op_gelu","text":"","code":"x <- op_array(c(-1., 0., 1.)) op_gelu(x) ## tf.Tensor([-0.158808  0.        0.841192], shape=(3), dtype=float32) op_gelu(x, FALSE) ## tf.Tensor([-0.15865526  0.          0.8413447 ], shape=(3), dtype=float32) x <- seq(-5, 5, .1) plot(x, op_gelu(x),      type = \"l\", #, frame.plot = FALSE,      panel.first = grid())"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_get_item.html","id":null,"dir":"Reference","previous_headings":"","what":"Return x[key]. — op_get_item","title":"Return x[key]. — op_get_item","text":"Return x[key].","code":""},{"path":"https://keras3.posit.co/dev/reference/op_get_item.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Return x[key]. — op_get_item","text":"","code":"op_get_item(x, key)"},{"path":"https://keras3.posit.co/dev/reference/op_get_item.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Return x[key]. — op_get_item","text":"x dictionary-like object key Generally, string, object __hash__ method acceptable.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_get_item.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Return x[key]. — op_get_item","text":"key.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_get_item.html","id":"note","dir":"Reference","previous_headings":"","what":"Note","title":"Return x[key]. — op_get_item","text":"Generally, calling x[[key]] x$key preferable.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_glu.html","id":null,"dir":"Reference","previous_headings":"","what":"Gated Linear Unit (GLU) activation function. — op_glu","title":"Gated Linear Unit (GLU) activation function. — op_glu","text":"defined : f(x) = * sigmoid(b) x split b along given axis.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_glu.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Gated Linear Unit (GLU) activation function. — op_glu","text":"","code":"op_glu(x, axis = -1L)"},{"path":"https://keras3.posit.co/dev/reference/op_glu.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Gated Linear Unit (GLU) activation function. — op_glu","text":"x Input tensor. axis axis along split input tensor. Defaults -1.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_glu.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Gated Linear Unit (GLU) activation function. — op_glu","text":"tensor shape half input.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_glu.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Gated Linear Unit (GLU) activation function. — op_glu","text":"","code":"x <- op_array(c(-1., 0., 1. , 1.)) op_glu(x) ## tf.Tensor([-0.7310586  0.       ], shape=(2), dtype=float32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_greater.html","id":null,"dir":"Reference","previous_headings":"","what":"Return the truth value of x1 > x2 element-wise. — op_greater","title":"Return the truth value of x1 > x2 element-wise. — op_greater","text":"Note function automatically called using R operator > tensor.","code":"(x <- op_arange(4)) ## tf.Tensor([1. 2. 3. 4.], shape=(4), dtype=float32) op_greater(x, 2) ## tf.Tensor([False False  True  True], shape=(4), dtype=bool) x > 2 ## tf.Tensor([False False  True  True], shape=(4), dtype=bool)"},{"path":"https://keras3.posit.co/dev/reference/op_greater.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Return the truth value of x1 > x2 element-wise. — op_greater","text":"","code":"op_greater(x1, x2)"},{"path":"https://keras3.posit.co/dev/reference/op_greater.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Return the truth value of x1 > x2 element-wise. — op_greater","text":"x1 First input tensor. x2 Second input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_greater.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Return the truth value of x1 > x2 element-wise. — op_greater","text":"Output tensor, element-wise comparison x1 x2.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_greater_equal.html","id":null,"dir":"Reference","previous_headings":"","what":"Return the truth value of x1 >= x2 element-wise. — op_greater_equal","title":"Return the truth value of x1 >= x2 element-wise. — op_greater_equal","text":"Note function automatically called using R operator >= tensor.","code":"(x <- op_arange(4)) ## tf.Tensor([1. 2. 3. 4.], shape=(4), dtype=float32) op_greater_equal(x, 2) ## tf.Tensor([False  True  True  True], shape=(4), dtype=bool) x >= 2 ## tf.Tensor([False  True  True  True], shape=(4), dtype=bool)"},{"path":"https://keras3.posit.co/dev/reference/op_greater_equal.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Return the truth value of x1 >= x2 element-wise. — op_greater_equal","text":"","code":"op_greater_equal(x1, x2)"},{"path":"https://keras3.posit.co/dev/reference/op_greater_equal.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Return the truth value of x1 >= x2 element-wise. — op_greater_equal","text":"x1 First input tensor. x2 Second input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_greater_equal.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Return the truth value of x1 >= x2 element-wise. — op_greater_equal","text":"Output tensor, element-wise comparison x1 x2.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_hamming.html","id":null,"dir":"Reference","previous_headings":"","what":"Hamming window function. — op_hamming","title":"Hamming window function. — op_hamming","text":"Hamming window defined : w[n] = 0.54 - 0.46 * cos(2 * pi * n / (N - 1)) 0 <= n <= N - 1.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_hamming.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Hamming window function. — op_hamming","text":"","code":"op_hamming(x)"},{"path":"https://keras3.posit.co/dev/reference/op_hamming.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Hamming window function. — op_hamming","text":"x Length window. Must positive integer.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_hamming.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Hamming window function. — op_hamming","text":"1D tensor containing window values.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_hamming.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Hamming window function. — op_hamming","text":"","code":"op_hamming(5) ## tf.Tensor([0.08000001 0.54       1.         0.54       0.08000001], shape=(5), dtype=float32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_hanning.html","id":null,"dir":"Reference","previous_headings":"","what":"Hanning window function. — op_hanning","title":"Hanning window function. — op_hanning","text":"Hanning window defined : w[n] = 0.5 - 0.5 * cos(2 * pi * n / (N - 1)) 0 <= n <= N - 1.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_hanning.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Hanning window function. — op_hanning","text":"","code":"op_hanning(x)"},{"path":"https://keras3.posit.co/dev/reference/op_hanning.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Hanning window function. — op_hanning","text":"x Length window. Must positive integer.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_hanning.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Hanning window function. — op_hanning","text":"1D tensor containing window values.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_hanning.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Hanning window function. — op_hanning","text":"","code":"op_hanning(5) ## tf.Tensor([0.  0.5 1.  0.5 0. ], shape=(5), dtype=float32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_hard_shrink.html","id":null,"dir":"Reference","previous_headings":"","what":"Hard Shrink activation function. — op_hard_shrink","title":"Hard Shrink activation function. — op_hard_shrink","text":"Hard Shrink function thresholding operation defined : f(x) = x |x| > threshold, f(x) = 0 otherwise.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_hard_shrink.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Hard Shrink activation function. — op_hard_shrink","text":"","code":"op_hard_shrink(x, threshold = 0.5)"},{"path":"https://keras3.posit.co/dev/reference/op_hard_shrink.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Hard Shrink activation function. — op_hard_shrink","text":"x Input tensor. threshold Threshold value. Defaults 0.5.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_hard_shrink.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Hard Shrink activation function. — op_hard_shrink","text":"tensor shape x.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_hard_shrink.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Hard Shrink activation function. — op_hard_shrink","text":"","code":"x <- op_array(c(-0.5, 0., 1.)) op_hard_shrink(x) ## tf.Tensor([0. 0. 1.], shape=(3), dtype=float32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_hard_sigmoid.html","id":null,"dir":"Reference","previous_headings":"","what":"Hard sigmoid activation function. — op_hard_sigmoid","title":"Hard sigmoid activation function. — op_hard_sigmoid","text":"defined : 0 x < -2.5, 1 x > 2.5, (0.2 * x) + 0.5 -2.5 <= x <= 2.5.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_hard_sigmoid.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Hard sigmoid activation function. — op_hard_sigmoid","text":"","code":"op_hard_sigmoid(x)"},{"path":"https://keras3.posit.co/dev/reference/op_hard_sigmoid.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Hard sigmoid activation function. — op_hard_sigmoid","text":"x Input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_hard_sigmoid.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Hard sigmoid activation function. — op_hard_sigmoid","text":"tensor shape x.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_hard_sigmoid.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Hard sigmoid activation function. — op_hard_sigmoid","text":"","code":"x <- op_array(c(-1., 0., 1.)) op_hard_sigmoid(x) ## tf.Tensor([0.33333334 0.5        0.6666667 ], shape=(3), dtype=float32) x <- as.array(seq(-5, 5, .1)) plot(x, op_hard_sigmoid(x),      type = 'l', panel.first = grid(), frame.plot = FALSE)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_hard_silu.html","id":null,"dir":"Reference","previous_headings":"","what":"Hard SiLU activation function, also known as Hard Swish. — op_hard_silu","title":"Hard SiLU activation function, also known as Hard Swish. — op_hard_silu","text":"defined : 0 x < -3 x x > 3 x * (x + 3) / 6 -3 <= x <= 3 faster, piecewise linear approximation silu activation.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_hard_silu.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Hard SiLU activation function, also known as Hard Swish. — op_hard_silu","text":"","code":"op_hard_silu(x)  op_hard_swish(x)"},{"path":"https://keras3.posit.co/dev/reference/op_hard_silu.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Hard SiLU activation function, also known as Hard Swish. — op_hard_silu","text":"x Input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_hard_silu.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Hard SiLU activation function, also known as Hard Swish. — op_hard_silu","text":"tensor shape x.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_hard_silu.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Hard SiLU activation function, also known as Hard Swish. — op_hard_silu","text":"","code":"x <- op_convert_to_tensor(c(-3.0, -1.0, 0.0, 1.0, 3.0)) op_hard_silu(x) ## tf.Tensor([-0.         -0.33333333  0.          0.66666667  3.        ], shape=(5), dtype=float64)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_hard_tanh.html","id":null,"dir":"Reference","previous_headings":"","what":"Applies the HardTanh function element-wise. — op_hard_tanh","title":"Applies the HardTanh function element-wise. — op_hard_tanh","text":"defined : f(x) = -1 x < -1, f(x) = x -1 <= x <= 1, f(x) = 1 x > 1.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_hard_tanh.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Applies the HardTanh function element-wise. — op_hard_tanh","text":"","code":"op_hard_tanh(x)"},{"path":"https://keras3.posit.co/dev/reference/op_hard_tanh.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Applies the HardTanh function element-wise. — op_hard_tanh","text":"x Input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_hard_tanh.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Applies the HardTanh function element-wise. — op_hard_tanh","text":"Output tensor shape x values clamped -1 1.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_hard_tanh.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Applies the HardTanh function element-wise. — op_hard_tanh","text":"","code":"x <- op_array(c(-2., -1., 0., 1., 2.)) op_hard_tanh(x) ## tf.Tensor([-1. -1.  0.  1.  1.], shape=(5), dtype=float32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_heaviside.html","id":null,"dir":"Reference","previous_headings":"","what":"Heaviside step function. — op_heaviside","title":"Heaviside step function. — op_heaviside","text":"Heaviside step function defined : heaviside(x1, x2) = 0 x1 < 0, heaviside(x1, x2) = 1 x1 > 0, heaviside(x1, x2) = x2 x1 == 0.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_heaviside.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Heaviside step function. — op_heaviside","text":"","code":"op_heaviside(x1, x2)"},{"path":"https://keras3.posit.co/dev/reference/op_heaviside.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Heaviside step function. — op_heaviside","text":"x1 Tensor input. x2 Value use x1 == 0.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_heaviside.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Heaviside step function. — op_heaviside","text":"tensor broadcast x1 x2 containing 0, 1, x2.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_heaviside.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Heaviside step function. — op_heaviside","text":"","code":"x1 <- op_array(c(-2, 0, 3)) op_heaviside(x1, 0.5) ## tf.Tensor([0.  0.5 1. ], shape=(3), dtype=float32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_histogram.html","id":null,"dir":"Reference","previous_headings":"","what":"Computes a histogram of the data tensor x. — op_histogram","title":"Computes a histogram of the data tensor x. — op_histogram","text":"Computes histogram data tensor x.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_histogram.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes a histogram of the data tensor x. — op_histogram","text":"","code":"op_histogram(x, bins = 10L, range = NULL)"},{"path":"https://keras3.posit.co/dev/reference/op_histogram.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Computes a histogram of the data tensor x. — op_histogram","text":"x Input tensor. bins integer representing number histogram bins. Defaults 10. range pair numbers representing lower upper range bins. specified, use min max x.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_histogram.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Computes a histogram of the data tensor x. — op_histogram","text":"list two tensors containing: tensor representing counts elements bin. tensor representing bin edges.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_histogram.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Computes a histogram of the data tensor x. — op_histogram","text":"","code":"input_tensor <- random_uniform(8) c(counts, edges) %<-% op_histogram(input_tensor)  counts #> tf.Tensor([2. 1. 1. 1. 0. 1. 0. 0. 0. 2.], shape=(10), dtype=float32) edges #> tf.Tensor( #> [0.01647806 0.10867032 0.20086257 0.29305482 0.38524708 0.47743934 #>  0.5696316  0.66182387 0.7540161  0.84620833 0.9384006 ], shape=(11), dtype=float32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_hstack.html","id":null,"dir":"Reference","previous_headings":"","what":"Stack tensors in sequence horizontally (column wise). — op_hstack","title":"Stack tensors in sequence horizontally (column wise). — op_hstack","text":"equivalent concatenation along first axis 1-D tensors, along second axis tensors.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_hstack.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Stack tensors in sequence horizontally (column wise). — op_hstack","text":"","code":"op_hstack(xs, ...)"},{"path":"https://keras3.posit.co/dev/reference/op_hstack.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Stack tensors in sequence horizontally (column wise). — op_hstack","text":"xs, ... list tensors.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_hstack.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Stack tensors in sequence horizontally (column wise). — op_hstack","text":"tensor formed stacking given tensors.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_identity.html","id":null,"dir":"Reference","previous_headings":"","what":"Return the identity tensor. — op_identity","title":"Return the identity tensor. — op_identity","text":"identity tensor square tensor ones main diagonal zeros elsewhere.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_identity.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Return the identity tensor. — op_identity","text":"","code":"op_identity(n, dtype = NULL)"},{"path":"https://keras3.posit.co/dev/reference/op_identity.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Return the identity tensor. — op_identity","text":"n Number rows (columns) n x n output tensor. dtype Data type output tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_identity.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Return the identity tensor. — op_identity","text":"identity tensor.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_ifft2.html","id":null,"dir":"Reference","previous_headings":"","what":"Computes the 2D Inverse Fast Fourier Transform along the last two axes of — op_ifft2","title":"Computes the 2D Inverse Fast Fourier Transform along the last two axes of — op_ifft2","text":"input.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_ifft2.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes the 2D Inverse Fast Fourier Transform along the last two axes of — op_ifft2","text":"","code":"op_ifft2(x)"},{"path":"https://keras3.posit.co/dev/reference/op_ifft2.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Computes the 2D Inverse Fast Fourier Transform along the last two axes of — op_ifft2","text":"x Tuple real imaginary parts input tensor. tensors tuple floating type.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_ifft2.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Computes the 2D Inverse Fast Fourier Transform along the last two axes of — op_ifft2","text":"tuple containing two tensors - real imaginary parts output.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_ifft2.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Computes the 2D Inverse Fast Fourier Transform along the last two axes of — op_ifft2","text":"","code":"x <- list(op_array(rbind(c(1, 2), c(2, 1))),           op_array(rbind(c(0, 1), c(1, 0)))) op_ifft2(x) ## [[1]] ## tf.Tensor( ## [[ 1.5  0. ] ##  [ 0.  -0.5]], shape=(2, 2), dtype=float64) ## ## [[2]] ## tf.Tensor( ## [[ 0.5 -0. ] ##  [-0.  -0.5]], shape=(2, 2), dtype=float64)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_imag.html","id":null,"dir":"Reference","previous_headings":"","what":"Return the imaginary part of the complex argument. — op_imag","title":"Return the imaginary part of the complex argument. — op_imag","text":"Return imaginary part complex argument.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_imag.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Return the imaginary part of the complex argument. — op_imag","text":"","code":"op_imag(x)"},{"path":"https://keras3.posit.co/dev/reference/op_imag.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Return the imaginary part of the complex argument. — op_imag","text":"x Input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_imag.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Return the imaginary part of the complex argument. — op_imag","text":"imaginary component complex argument.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_image_affine_transform.html","id":null,"dir":"Reference","previous_headings":"","what":"Applies the given transform(s) to the image(s). — op_image_affine_transform","title":"Applies the given transform(s) to the image(s). — op_image_affine_transform","text":"Applies given transform(s) image(s).","code":""},{"path":"https://keras3.posit.co/dev/reference/op_image_affine_transform.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Applies the given transform(s) to the image(s). — op_image_affine_transform","text":"","code":"op_image_affine_transform(   images,   transform,   interpolation = \"bilinear\",   fill_mode = \"constant\",   fill_value = 0L,   data_format = NULL )"},{"path":"https://keras3.posit.co/dev/reference/op_image_affine_transform.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Applies the given transform(s) to the image(s). — op_image_affine_transform","text":"images Input image batch images. Must 3D 4D. transform Projective transform matrix/matrices. vector length 8 tensor size N x 8. one row transform [a0, a1, a2, b0, b1, b2, c0, c1], maps output point (x, y) transformed input point (x', y') = ((a0 x + a1 y + a2) / k, (b0 x + b1 y + b2) / k), k = c0 x + c1 y + 1. transform inverted compared transform mapping input points output points. Note gradients backpropagated transformation parameters. Note c0 c1 effective using TensorFlow backend considered 0 using backends. interpolation Interpolation method. Available methods \"nearest\", \"bilinear\". Defaults \"bilinear\". fill_mode Points outside boundaries input filled according given mode. Available methods \"constant\", \"nearest\", \"wrap\" \"reflect\". Defaults \"constant\". \"reflect\": (d c b | b c d | d c b ) input extended reflecting edge last pixel. \"constant\": (k k k k | b c d | k k k k) input extended filling values beyond edge constant value k specified fill_value. \"wrap\": (b c d | b c d | b c d) input extended wrapping around opposite edge. \"nearest\": (| b c d | d d d d) input extended nearest pixel. fill_value Value used points outside boundaries input fill_mode = \"constant\". Defaults 0. data_format string specifying data format input tensor. can either \"channels_last\" \"channels_first\". \"channels_last\" corresponds inputs shape (batch, height, width, channels), \"channels_first\" corresponds inputs shape (batch, channels, height, width). specified, value default config_image_data_format().","code":""},{"path":"https://keras3.posit.co/dev/reference/op_image_affine_transform.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Applies the given transform(s) to the image(s). — op_image_affine_transform","text":"Applied affine transform image batch images.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_image_affine_transform.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Applies the given transform(s) to the image(s). — op_image_affine_transform","text":"","code":"x <- random_uniform(c(2, 64, 80, 3)) # batch of 2 RGB images transform <- op_array(rbind(c(1.5, 0, -20, 0, 1.5, -16, 0, 0),  # zoom                            c(1, 0, -20, 0, 1, -16, 0, 0)))  # translation)) y <- op_image_affine_transform(x, transform) shape(y) ## shape(2, 64, 80, 3) # (2, 64, 80, 3) x <- random_uniform(c(64, 80, 3)) # single RGB image transform <- op_array(c(1.0, 0.5, -20, 0.5, 1.0, -16, 0, 0))  # shear y <- op_image_affine_transform(x, transform) shape(y) ## shape(64, 80, 3) # (64, 80, 3) x <- random_uniform(c(2, 3, 64, 80)) # batch of 2 RGB images transform <- op_array(rbind(   c(1.5, 0,-20, 0, 1.5,-16, 0, 0),  # zoom   c(1, 0,-20, 0, 1,-16, 0, 0)  # translation )) y <- op_image_affine_transform(x, transform, data_format = \"channels_first\") shape(y) ## shape(2, 3, 64, 80) # (2, 3, 64, 80)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_image_crop.html","id":null,"dir":"Reference","previous_headings":"","what":"Crop images to a specified height and width. — op_image_crop","title":"Crop images to a specified height and width. — op_image_crop","text":"Crop images specified height width.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_image_crop.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Crop images to a specified height and width. — op_image_crop","text":"","code":"op_image_crop(   images,   top_cropping = NULL,   left_cropping = NULL,   bottom_cropping = NULL,   right_cropping = NULL,   target_height = NULL,   target_width = NULL,   data_format = NULL )"},{"path":"https://keras3.posit.co/dev/reference/op_image_crop.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Crop images to a specified height and width. — op_image_crop","text":"images Input image batch images. Must 3D 4D. top_cropping Number columns crop top. left_cropping Number columns crop left. bottom_cropping Number columns crop bottom. right_cropping Number columns crop right. target_height Height output images. target_width Width output images. data_format string specifying data format input tensor. can either \"channels_last\" \"channels_first\". \"channels_last\" corresponds inputs shape (batch, height, width, channels), \"channels_first\" corresponds inputs shape (batch, channels, height, width). specified, value default config_image_data_format().","code":""},{"path":"https://keras3.posit.co/dev/reference/op_image_crop.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Crop images to a specified height and width. — op_image_crop","text":"Cropped image batch images.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_image_crop.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Crop images to a specified height and width. — op_image_crop","text":"","code":"images <- op_reshape(op_arange(1, 28, dtype=\"float32\"), c(3, 3, 3)) images[, , 1] # print the first channel of the images  cropped_images <- op_image_crop(images, 0, 0, 2, 2) cropped_images[, , 1] # print the first channel of the cropped images"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_image_elastic_transform.html","id":null,"dir":"Reference","previous_headings":"","what":"Applies elastic deformation to the image(s). — op_image_elastic_transform","title":"Applies elastic deformation to the image(s). — op_image_elastic_transform","text":"Apply random elastic deformation 3D 4D image tensors.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_image_elastic_transform.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Applies elastic deformation to the image(s). — op_image_elastic_transform","text":"","code":"op_image_elastic_transform(   images,   alpha = 20,   sigma = 5,   interpolation = \"bilinear\",   fill_mode = \"reflect\",   fill_value = 0,   seed = NULL,   data_format = NULL )"},{"path":"https://keras3.posit.co/dev/reference/op_image_elastic_transform.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Applies elastic deformation to the image(s). — op_image_elastic_transform","text":"images Input image batch images. Must 3D 4D. alpha Scaling factor controls intensity deformation. sigma Standard deviation Gaussian filter used smoothing displacement fields. interpolation Interpolation method. Available methods \"nearest\", \"bilinear\". Defaults \"bilinear\". fill_mode Points outside boundaries input filled according given mode. Available methods \"constant\", \"nearest\", \"wrap\" \"reflect\". Defaults \"reflect\". \"reflect\": (d c b | b c d | d c b ) input extended reflecting edge last pixel. \"constant\": (k k k k | b c d | k k k k) input extended filling values beyond edge constant value k specified fill_value. \"wrap\": (b c d | b c d | b c d) input extended wrapping around opposite edge. \"nearest\": (| b c d | d d d d) input extended nearest pixel. fill_value Value used points outside boundaries input fill_mode=\"constant\". Defaults 0. seed Optional integer seed random number generator. data_format string specifying data format input tensor. can either \"channels_last\" \"channels_first\". \"channels_last\" corresponds inputs shape (batch, height, width, channels), \"channels_first\" corresponds inputs shape (batch, channels, height, width). specified, value default keras.config.image_data_format.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_image_elastic_transform.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Applies elastic deformation to the image(s). — op_image_elastic_transform","text":"Transformed image batch images elastic deformation.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_image_elastic_transform.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Applies elastic deformation to the image(s). — op_image_elastic_transform","text":"","code":"x <- random_uniform(c(2, 64, 80, 3))  # batch of 2 RGB images y <- op_image_elastic_transform(x) op_shape(y) ## shape(2, 64, 80, 3) x <- random_uniform(c(64, 80, 3))  # single RGB image y <- op_image_elastic_transform(x) op_shape(y) ## shape(64, 80, 3) x <- random_uniform(c(2, 3, 64, 80))  # batch of 2 RGB images y <- op_image_elastic_transform(   x,   data_format = \"channels_first\",   seed = 123 ) op_shape(y) ## shape(2, 3, 64, 80)"},{"path":"https://keras3.posit.co/dev/reference/op_image_extract_patches.html","id":null,"dir":"Reference","previous_headings":"","what":"Extracts patches from the image(s). — op_image_extract_patches","title":"Extracts patches from the image(s). — op_image_extract_patches","text":"Extracts patches image(s).","code":""},{"path":"https://keras3.posit.co/dev/reference/op_image_extract_patches.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Extracts patches from the image(s). — op_image_extract_patches","text":"","code":"op_image_extract_patches(   images,   size,   strides = NULL,   dilation_rate = 1L,   padding = \"valid\",   data_format = \"channels_last\" )"},{"path":"https://keras3.posit.co/dev/reference/op_image_extract_patches.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Extracts patches from the image(s). — op_image_extract_patches","text":"images Input image batch images. Must 3D 4D. size Patch size int list (patch_height, patch_width) strides strides along height width. specified, NULL, defaults value size. dilation_rate input stride, specifying far two consecutive patch samples input. value 1, strides must 1. NOTE: strides > 1 supported conjunction dilation_rate > 1 padding type padding algorithm use: \"\" \"valid\". data_format string specifying data format input tensor. can either \"channels_last\" \"channels_first\". \"channels_last\" corresponds inputs shape (batch, height, width, channels), \"channels_first\" corresponds inputs shape (batch, channels, height, width). specified, value default config_image_data_format().","code":""},{"path":"https://keras3.posit.co/dev/reference/op_image_extract_patches.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Extracts patches from the image(s). — op_image_extract_patches","text":"Extracted patches 3D (batched) 4D (batched)","code":""},{"path":"https://keras3.posit.co/dev/reference/op_image_extract_patches.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Extracts patches from the image(s). — op_image_extract_patches","text":"","code":"image <- random_uniform(c(2, 20, 20, 3), dtype = \"float32\") # batch of 2 RGB images patches <- op_image_extract_patches(image, c(5, 5)) shape(patches) ## shape(2, 4, 4, 75) # (2, 4, 4, 75) image <- random_uniform(c(20, 20, 3), dtype = \"float32\") # 1 RGB image patches <- op_image_extract_patches(image, c(3, 3), c(1, 1)) shape(patches) ## shape(18, 18, 27) # (18, 18, 27)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_image_gaussian_blur.html","id":null,"dir":"Reference","previous_headings":"","what":"Applies a Gaussian blur to the image(s). — op_image_gaussian_blur","title":"Applies a Gaussian blur to the image(s). — op_image_gaussian_blur","text":"Applies Gaussian blur image(s).","code":""},{"path":"https://keras3.posit.co/dev/reference/op_image_gaussian_blur.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Applies a Gaussian blur to the image(s). — op_image_gaussian_blur","text":"","code":"op_image_gaussian_blur(   images,   kernel_size = list(3L, 3L),   sigma = list(1, 1),   data_format = NULL )"},{"path":"https://keras3.posit.co/dev/reference/op_image_gaussian_blur.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Applies a Gaussian blur to the image(s). — op_image_gaussian_blur","text":"images Input image batch images. Must 3D 4D. kernel_size tuple two integers, specifying height width Gaussian kernel. sigma tuple two floats, specifying standard deviation Gaussian kernel along height width. data_format string specifying data format input tensor. can either \"channels_last\" \"channels_first\". \"channels_last\" corresponds inputs shape (batch, height, width, channels), \"channels_first\" corresponds inputs shape (batch, channels, height, width). specified, value default keras.config.image_data_format.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_image_gaussian_blur.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Applies a Gaussian blur to the image(s). — op_image_gaussian_blur","text":"Blurred image batch images.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_image_gaussian_blur.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Applies a Gaussian blur to the image(s). — op_image_gaussian_blur","text":"","code":"x <- op_ones(c(2, 64, 80, 3))  # batch of 2 RGB images y <- op_image_gaussian_blur(x) op_shape(y) ## shape(2, 64, 80, 3) x <- op_ones(c(64, 80, 3))  # single RGB image y <- op_image_gaussian_blur(x) op_shape(y) ## shape(64, 80, 3) x <- op_ones(c(2, 3, 64, 80))  # batch of 2 RGB images, channels_first y <- op_image_gaussian_blur(x, data_format = \"channels_first\") op_shape(y) ## shape(2, 3, 64, 80)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_image_hsv_to_rgb.html","id":null,"dir":"Reference","previous_headings":"","what":"Convert HSV images to RGB. — op_image_hsv_to_rgb","title":"Convert HSV images to RGB. — op_image_hsv_to_rgb","text":"images must float dtype, output well defined values images [0, 1].","code":""},{"path":"https://keras3.posit.co/dev/reference/op_image_hsv_to_rgb.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Convert HSV images to RGB. — op_image_hsv_to_rgb","text":"","code":"op_image_hsv_to_rgb(images, data_format = NULL)"},{"path":"https://keras3.posit.co/dev/reference/op_image_hsv_to_rgb.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Convert HSV images to RGB. — op_image_hsv_to_rgb","text":"images Input image batch images. Must 3D 4D. data_format string specifying data format input tensor. can either \"channels_last\" \"channels_first\". \"channels_last\" corresponds inputs shape (batch, height, width, channels), \"channels_first\" corresponds inputs shape (batch, channels, height, width). specified, value default config_image_data_format().","code":""},{"path":"https://keras3.posit.co/dev/reference/op_image_hsv_to_rgb.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Convert HSV images to RGB. — op_image_hsv_to_rgb","text":"RGB image batch RGB images.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_image_hsv_to_rgb.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Convert HSV images to RGB. — op_image_hsv_to_rgb","text":"","code":"x <- random_uniform(c(2, 4, 4, 3)) y <- op_image_hsv_to_rgb(x) shape(y) ## shape(2, 4, 4, 3) x <- random_uniform(c(4, 4, 3)) # Single HSV image y <- op_image_hsv_to_rgb(x) shape(y) ## shape(4, 4, 3) x <- random_uniform(c(2, 3, 4, 4)) y <- op_image_hsv_to_rgb(x, data_format=\"channels_first\") shape(y) ## shape(2, 3, 4, 4)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_image_map_coordinates.html","id":null,"dir":"Reference","previous_headings":"","what":"Map the input array to new coordinates by interpolation. — op_image_map_coordinates","title":"Map the input array to new coordinates by interpolation. — op_image_map_coordinates","text":"Note interpolation near boundaries differs scipy function, fixed outstanding bug scipy/issues/2640.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_image_map_coordinates.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Map the input array to new coordinates by interpolation. — op_image_map_coordinates","text":"","code":"op_image_map_coordinates(   inputs,   coordinates,   order,   fill_mode = \"constant\",   fill_value = 0L )"},{"path":"https://keras3.posit.co/dev/reference/op_image_map_coordinates.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Map the input array to new coordinates by interpolation. — op_image_map_coordinates","text":"inputs input array. coordinates coordinates inputs evaluated. order order spline interpolation. order must 0 1. 0 indicates nearest neighbor 1 indicates linear interpolation. fill_mode Points outside boundaries inputs filled according given mode. Available methods \"constant\", \"nearest\", \"wrap\" \"mirror\" \"reflect\". Defaults \"constant\". \"constant\": (k k k k | b c d | k k k k) inputs extended filling values beyond edge constant value k specified fill_value. \"nearest\": (| b c d | d d d d) inputs extended nearest pixel. \"wrap\": (b c d | b c d | b c d) inputs extended wrapping around opposite edge. \"mirror\": (c d c b | b c d | c b b) inputs extended mirroring edge. \"reflect\": (d c b | b c d | d c b ) inputs extended reflecting edge last pixel. fill_value Value used points outside boundaries inputs fill_mode = \"constant\". Defaults 0.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_image_map_coordinates.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Map the input array to new coordinates by interpolation. — op_image_map_coordinates","text":"Output input batch inputs.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_image_pad.html","id":null,"dir":"Reference","previous_headings":"","what":"Pad images with zeros to the specified height and width. — op_image_pad","title":"Pad images with zeros to the specified height and width. — op_image_pad","text":"Pad images zeros specified height width.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_image_pad.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Pad images with zeros to the specified height and width. — op_image_pad","text":"","code":"op_image_pad(   images,   top_padding = NULL,   left_padding = NULL,   bottom_padding = NULL,   right_padding = NULL,   target_height = NULL,   target_width = NULL,   data_format = NULL )"},{"path":"https://keras3.posit.co/dev/reference/op_image_pad.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Pad images with zeros to the specified height and width. — op_image_pad","text":"images Input image batch images. Must 3D 4D. top_padding Number rows zeros add top. left_padding Number columns zeros add left. bottom_padding Number rows zeros add bottom. right_padding Number columns zeros add right. target_height Height output images. target_width Width output images. data_format string specifying data format input tensor. can either \"channels_last\" \"channels_first\". \"channels_last\" corresponds inputs shape (batch, height, width, channels), \"channels_first\" corresponds inputs shape (batch, channels, height, width). specified, value default config_image_data_format().","code":""},{"path":"https://keras3.posit.co/dev/reference/op_image_pad.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Pad images with zeros to the specified height and width. — op_image_pad","text":"Padded image batch images.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_image_pad.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Pad images with zeros to the specified height and width. — op_image_pad","text":"","code":"images <- random_uniform(c(15, 25, 3)) padded_images <- op_image_pad(     images, 2, 3, target_height = 20, target_width = 30 ) shape(padded_images) ## shape(20, 30, 3) batch_images <- random_uniform(c(2, 15, 25, 3)) padded_batch <- op_image_pad(batch_images, 2, 3,                              target_height = 20,                              target_width = 30) shape(padded_batch) ## shape(2, 20, 30, 3)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_image_perspective_transform.html","id":null,"dir":"Reference","previous_headings":"","what":"Applies a perspective transformation to the image(s). — op_image_perspective_transform","title":"Applies a perspective transformation to the image(s). — op_image_perspective_transform","text":"Applies perspective transformation image(s).","code":""},{"path":"https://keras3.posit.co/dev/reference/op_image_perspective_transform.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Applies a perspective transformation to the image(s). — op_image_perspective_transform","text":"","code":"op_image_perspective_transform(   images,   start_points,   end_points,   interpolation = \"bilinear\",   fill_value = 0L,   data_format = NULL )"},{"path":"https://keras3.posit.co/dev/reference/op_image_perspective_transform.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Applies a perspective transformation to the image(s). — op_image_perspective_transform","text":"images Input image batch images. Must 3D 4D. start_points tensor shape (N, 4, 2) (4, 2), representing source points original image define transformation. end_points tensor shape (N, 4, 2) (4, 2), representing target points output image transformation. interpolation Interpolation method. Available methods \"nearest\", \"bilinear\". Defaults \"bilinear\". fill_value Value used points outside boundaries input extrapolation needed. Defaults 0. data_format string specifying data format input tensor. can either \"channels_last\" \"channels_first\". \"channels_last\" corresponds inputs shape (batch, height, width, channels), \"channels_first\" corresponds inputs shape (batch, channels, height, width). specified, value default keras.config.image_data_format.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_image_perspective_transform.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Applies a perspective transformation to the image(s). — op_image_perspective_transform","text":"Applied perspective transform image batch images.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_image_perspective_transform.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Applies a perspective transformation to the image(s). — op_image_perspective_transform","text":"","code":"# Batch of 2 RGB images (channels_last) x <- op_ones(c(2, 64, 80, 3)) start_points <- op_stack(list(   rbind(c(0, 0), c(0, 64), c(80, 0), c(80, 64)),   rbind(c(0, 0), c(0, 64), c(80, 0), c(80, 64)) )) end_points <- op_stack(list(   rbind(c(3, 5), c(7, 64), c(76, -10), c(84, 61)),   rbind(c(8, 10), c(10, 61), c(65, 3), c(88, 43)) )) y <- op_image_perspective_transform(x, start_points, end_points) op_shape(y) ## shape(2, 64, 80, 3) # Single RGB image (channels_last) x <- op_ones(c(64, 80, 3)) start_points <- rbind(c(0, 0), c(0, 64), c(80, 0), c(80, 64)) end_points <- rbind(c(3, 5), c(7, 64), c(76, -10), c(84, 61)) y <- op_image_perspective_transform(x, start_points, end_points) op_shape(y) ## shape(64, 80, 3) # Batch of 2 RGB images (channels_first) x <- op_ones(c(2, 3, 64, 80)) start_points <- op_stack(list(   rbind(c(0, 0), c(0, 64), c(80, 0), c(80, 64)),   rbind(c(0, 0), c(0, 64), c(80, 0), c(80, 64)) )) end_points <- op_stack(list(   rbind(c(3, 5), c(7, 64), c(76, -10), c(84, 61)),   rbind(c(8, 10), c(10, 61), c(65, 3), c(88, 43)) )) y <- op_image_perspective_transform(   x, start_points, end_points,   data_format = \"channels_first\" ) op_shape(y) ## shape(2, 3, 64, 80)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_image_resize.html","id":null,"dir":"Reference","previous_headings":"","what":"Resize images to size using the specified interpolation method. — op_image_resize","title":"Resize images to size using the specified interpolation method. — op_image_resize","text":"Resize images size using specified interpolation method.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_image_resize.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Resize images to size using the specified interpolation method. — op_image_resize","text":"","code":"op_image_resize(   images,   size,   interpolation = \"bilinear\",   antialias = FALSE,   crop_to_aspect_ratio = FALSE,   pad_to_aspect_ratio = FALSE,   fill_mode = \"constant\",   fill_value = 0,   data_format = NULL )"},{"path":"https://keras3.posit.co/dev/reference/op_image_resize.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Resize images to size using the specified interpolation method. — op_image_resize","text":"images Input image batch images. Must 3D 4D. size Size output image (height, width) format. interpolation Interpolation method. Available methods \"nearest\", \"bilinear\", \"bicubic\". Defaults \"bilinear\". antialias Whether use antialiasing filter downsampling image. Defaults FALSE. crop_to_aspect_ratio TRUE, resize images without aspect ratio distortion. original aspect ratio differs target aspect ratio, output image cropped return largest possible window image (size (height, width)) matches target aspect ratio. default (crop_to_aspect_ratio=FALSE), aspect ratio may preserved. pad_to_aspect_ratio TRUE, pad images without aspect ratio distortion. original aspect ratio differs target aspect ratio, output image evenly padded short side. fill_mode using pad_to_aspect_ratio=TRUE, padded areas filled according given mode. \"constant\" supported time (fill constant value, equal fill_value). fill_value Float. Padding value use pad_to_aspect_ratio=TRUE. data_format string specifying data format input tensor. can either \"channels_last\" \"channels_first\". \"channels_last\" corresponds inputs shape (batch, height, width, channels), \"channels_first\" corresponds inputs shape (batch, channels, height, width). specified, value default config_image_data_format().","code":""},{"path":"https://keras3.posit.co/dev/reference/op_image_resize.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Resize images to size using the specified interpolation method. — op_image_resize","text":"Resized image batch images.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_image_resize.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Resize images to size using the specified interpolation method. — op_image_resize","text":"","code":"x <- random_uniform(c(2, 4, 4, 3)) # batch of 2 RGB images y <- op_image_resize(x, c(2, 2)) shape(y) ## shape(2, 2, 2, 3) x <- random_uniform(c(4, 4, 3)) # single RGB image y <- op_image_resize(x, c(2, 2)) shape(y) ## shape(2, 2, 3) x <- random_uniform(c(2, 3, 4, 4)) # batch of 2 RGB images y <- op_image_resize(x, c(2, 2), data_format = \"channels_first\") shape(y) ## shape(2, 3, 2, 2)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_image_rgb_to_grayscale.html","id":null,"dir":"Reference","previous_headings":"","what":"Convert RGB images to grayscale. — op_image_rgb_to_grayscale","title":"Convert RGB images to grayscale. — op_image_rgb_to_grayscale","text":"function converts RGB images grayscale images. supports 3D 4D tensors, last dimension represents channels.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_image_rgb_to_grayscale.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Convert RGB images to grayscale. — op_image_rgb_to_grayscale","text":"","code":"op_image_rgb_to_grayscale(images, data_format = NULL)"},{"path":"https://keras3.posit.co/dev/reference/op_image_rgb_to_grayscale.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Convert RGB images to grayscale. — op_image_rgb_to_grayscale","text":"images Input image batch images. Must 3D 4D. data_format string specifying data format input tensor. can either \"channels_last\" \"channels_first\". \"channels_last\" corresponds inputs shape (batch, height, width, channels), \"channels_first\" corresponds inputs shape (batch, channels, height, width). specified, value default config_image_data_format().","code":""},{"path":"https://keras3.posit.co/dev/reference/op_image_rgb_to_grayscale.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Convert RGB images to grayscale. — op_image_rgb_to_grayscale","text":"Grayscale image batch grayscale images.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_image_rgb_to_grayscale.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Convert RGB images to grayscale. — op_image_rgb_to_grayscale","text":"","code":"x <- random_uniform(c(2, 4, 4, 3)) y <- op_image_rgb_to_grayscale(x) shape(y) ## shape(2, 4, 4, 1) x <- random_uniform(c(4, 4, 3)) # Single RGB image y = op_image_rgb_to_grayscale(x) shape(y) ## shape(4, 4, 1) x <- random_uniform(c(2, 3, 4, 4)) y <- op_image_rgb_to_grayscale(x, data_format=\"channels_first\") shape(y) ## shape(2, 1, 4, 4)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_image_rgb_to_hsv.html","id":null,"dir":"Reference","previous_headings":"","what":"Convert RGB images to HSV. — op_image_rgb_to_hsv","title":"Convert RGB images to HSV. — op_image_rgb_to_hsv","text":"images must float dtype, output well defined values images [0, 1]. HSV values [0, 1]. hue 0 corresponds pure red, 1/3 pure green, 2/3 pure blue.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_image_rgb_to_hsv.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Convert RGB images to HSV. — op_image_rgb_to_hsv","text":"","code":"op_image_rgb_to_hsv(images, data_format = NULL)"},{"path":"https://keras3.posit.co/dev/reference/op_image_rgb_to_hsv.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Convert RGB images to HSV. — op_image_rgb_to_hsv","text":"images Input image batch images. Must 3D 4D. data_format string specifying data format input tensor. can either \"channels_last\" \"channels_first\". \"channels_last\" corresponds inputs shape (batch, height, width, channels), \"channels_first\" corresponds inputs shape (batch, channels, height, width). specified, value default config_image_data_format().","code":""},{"path":"https://keras3.posit.co/dev/reference/op_image_rgb_to_hsv.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Convert RGB images to HSV. — op_image_rgb_to_hsv","text":"HSV image batch HSV images.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_image_rgb_to_hsv.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Convert RGB images to HSV. — op_image_rgb_to_hsv","text":"","code":"x <- random_uniform(c(2, 4, 4, 3)) y <- op_image_rgb_to_hsv(x) shape(y) ## shape(2, 4, 4, 3) x <- random_uniform(c(4, 4, 3)) # Single RGB image y <- op_image_rgb_to_hsv(x) shape(y) ## shape(4, 4, 3) x <- random_uniform(c(2, 3, 4, 4)) y <- op_image_rgb_to_hsv(x, data_format = \"channels_first\") shape(y) ## shape(2, 3, 4, 4)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_in_top_k.html","id":null,"dir":"Reference","previous_headings":"","what":"Checks if the targets are in the top-k predictions. — op_in_top_k","title":"Checks if the targets are in the top-k predictions. — op_in_top_k","text":"Checks targets top-k predictions.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_in_top_k.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Checks if the targets are in the top-k predictions. — op_in_top_k","text":"","code":"op_in_top_k(targets, predictions, k)"},{"path":"https://keras3.posit.co/dev/reference/op_in_top_k.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Checks if the targets are in the top-k predictions. — op_in_top_k","text":"targets tensor true labels. predictions tensor predicted labels. k integer representing number predictions consider.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_in_top_k.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Checks if the targets are in the top-k predictions. — op_in_top_k","text":"boolean tensor shape targets, element indicates whether corresponding target top-k predictions.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_in_top_k.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Checks if the targets are in the top-k predictions. — op_in_top_k","text":"","code":"targets <- op_array(c(2, 5, 3), \"int32\") predictions <- op_array(dtype = \"float32\", rbind(   c(0.1, 0.4, 0.6, 0.9, 0.5),   c(0.1, 0.7, 0.9, 0.8, 0.3),   c(0.1, 0.6, 0.9, 0.9, 0.5) )) op_in_top_k(targets, predictions, k = 3L) ## tf.Tensor([ True False  True], shape=(3), dtype=bool)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_inner.html","id":null,"dir":"Reference","previous_headings":"","what":"Return the inner product of two tensors. — op_inner","title":"Return the inner product of two tensors. — op_inner","text":"Ordinary inner product vectors 1-D tensors (without complex conjugation), higher dimensions sum product last axes. Multidimensional arrays treated vectors flattening last axes. resulting dot product performed last axes.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_inner.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Return the inner product of two tensors. — op_inner","text":"","code":"op_inner(x1, x2)"},{"path":"https://keras3.posit.co/dev/reference/op_inner.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Return the inner product of two tensors. — op_inner","text":"x1 First input tensor. x2 Second input tensor. last dimension x1 x2 must match.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_inner.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Return the inner product of two tensors. — op_inner","text":"Output tensor. shape output determined broadcasting shapes x1 x2 removing last axes.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_inv.html","id":null,"dir":"Reference","previous_headings":"","what":"Computes the inverse of a square tensor. — op_inv","title":"Computes the inverse of a square tensor. — op_inv","text":"Computes inverse square tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_inv.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes the inverse of a square tensor. — op_inv","text":"","code":"op_inv(x)"},{"path":"https://keras3.posit.co/dev/reference/op_inv.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Computes the inverse of a square tensor. — op_inv","text":"x Input tensor shape (..., M, M).","code":""},{"path":"https://keras3.posit.co/dev/reference/op_inv.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Computes the inverse of a square tensor. — op_inv","text":"tensor shape (..., M, M) representing inverse x.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_irfft.html","id":null,"dir":"Reference","previous_headings":"","what":"Inverse real-valued Fast Fourier transform along the last axis. — op_irfft","title":"Inverse real-valued Fast Fourier transform along the last axis. — op_irfft","text":"Computes inverse 1D Discrete Fourier Transform real-valued signal inner-dimension input. inner-dimension input assumed result RFFT: fft_length / 2 + 1 unique components DFT real-valued signal. fft_length provided, computed size inner-dimension input (fft_length = 2 * (inner - 1)). FFT length used compute odd, provided since inferred properly. Along axis IRFFT computed , fft_length / 2 + 1 smaller corresponding dimension input, dimension cropped. larger, dimension padded zeros.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_irfft.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Inverse real-valued Fast Fourier transform along the last axis. — op_irfft","text":"","code":"op_irfft(x, fft_length = NULL)"},{"path":"https://keras3.posit.co/dev/reference/op_irfft.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Inverse real-valued Fast Fourier transform along the last axis. — op_irfft","text":"x List real imaginary parts input tensor. tensors list floating type. fft_length integer representing number fft length. specified, inferred length last axis x. Defaults NULL.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_irfft.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Inverse real-valued Fast Fourier transform along the last axis. — op_irfft","text":"tensor containing inverse real-valued Fast Fourier Transform along last axis x.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_irfft.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Inverse real-valued Fast Fourier transform along the last axis. — op_irfft","text":"","code":"real <- op_array(c(0, 1, 2, 3, 4)) imag <- op_array(c(0, 1, 2, 3, 4)) op_irfft(c(real, imag)) #> tf.Tensor( #> [ 2.         -2.0606601   0.5        -0.35355338  0.          0.06066012 #>  -0.5         0.35355338], shape=(8), dtype=float32) all.equal(op_irfft(op_rfft(real, 5), 5), real) #> [1] TRUE"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_is_tensor.html","id":null,"dir":"Reference","previous_headings":"","what":"Check whether the given object is a tensor. — op_is_tensor","title":"Check whether the given object is a tensor. — op_is_tensor","text":"Check whether given object tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_is_tensor.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Check whether the given object is a tensor. — op_is_tensor","text":"","code":"op_is_tensor(x)"},{"path":"https://keras3.posit.co/dev/reference/op_is_tensor.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Check whether the given object is a tensor. — op_is_tensor","text":"x variable.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_is_tensor.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Check whether the given object is a tensor. — op_is_tensor","text":"TRUE x tensor, otherwise FALSE.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_is_tensor.html","id":"note","dir":"Reference","previous_headings":"","what":"Note","title":"Check whether the given object is a tensor. — op_is_tensor","text":"checks backend specific tensors passing TensorFlow tensor return FALSE backend PyTorch JAX.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_isclose.html","id":null,"dir":"Reference","previous_headings":"","what":"Return whether two tensors are element-wise almost equal. — op_isclose","title":"Return whether two tensors are element-wise almost equal. — op_isclose","text":"Return whether two tensors element-wise almost equal.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_isclose.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Return whether two tensors are element-wise almost equal. — op_isclose","text":"","code":"op_isclose(x1, x2, rtol = 1e-05, atol = 1e-08, equal_nan = FALSE)"},{"path":"https://keras3.posit.co/dev/reference/op_isclose.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Return whether two tensors are element-wise almost equal. — op_isclose","text":"x1 First input tensor. x2 Second input tensor. rtol Relative tolerance. atol Absolute tolerance. equal_nan TRUE, element-wise NaNs considered equal.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_isclose.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Return whether two tensors are element-wise almost equal. — op_isclose","text":"Output boolean tensor.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_isfinite.html","id":null,"dir":"Reference","previous_headings":"","what":"Return whether a tensor is finite, element-wise. — op_isfinite","title":"Return whether a tensor is finite, element-wise. — op_isfinite","text":"Real values finite NaN, positive infinity, negative infinity. Complex values finite real imaginary parts finite.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_isfinite.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Return whether a tensor is finite, element-wise. — op_isfinite","text":"","code":"op_isfinite(x)"},{"path":"https://keras3.posit.co/dev/reference/op_isfinite.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Return whether a tensor is finite, element-wise. — op_isfinite","text":"x Input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_isfinite.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Return whether a tensor is finite, element-wise. — op_isfinite","text":"Output boolean tensor.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_isinf.html","id":null,"dir":"Reference","previous_headings":"","what":"Test element-wise for positive or negative infinity. — op_isinf","title":"Test element-wise for positive or negative infinity. — op_isinf","text":"Test element-wise positive negative infinity.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_isinf.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Test element-wise for positive or negative infinity. — op_isinf","text":"","code":"op_isinf(x)"},{"path":"https://keras3.posit.co/dev/reference/op_isinf.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Test element-wise for positive or negative infinity. — op_isinf","text":"x Input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_isinf.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Test element-wise for positive or negative infinity. — op_isinf","text":"Output boolean tensor.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_isnan.html","id":null,"dir":"Reference","previous_headings":"","what":"Test element-wise for NaN and return result as a boolean tensor. — op_isnan","title":"Test element-wise for NaN and return result as a boolean tensor. — op_isnan","text":"Test element-wise NaN return result boolean tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_isnan.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Test element-wise for NaN and return result as a boolean tensor. — op_isnan","text":"","code":"op_isnan(x)"},{"path":"https://keras3.posit.co/dev/reference/op_isnan.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Test element-wise for NaN and return result as a boolean tensor. — op_isnan","text":"x Input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_isnan.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Test element-wise for NaN and return result as a boolean tensor. — op_isnan","text":"Output boolean tensor.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_istft.html","id":null,"dir":"Reference","previous_headings":"","what":"Inverse Short-Time Fourier Transform along the last axis of the input. — op_istft","title":"Inverse Short-Time Fourier Transform along the last axis of the input. — op_istft","text":"reconstruct original waveform, parameters stft.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_istft.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Inverse Short-Time Fourier Transform along the last axis of the input. — op_istft","text":"","code":"op_istft(   x,   sequence_length,   sequence_stride,   fft_length,   length = NULL,   window = \"hann\",   center = TRUE )"},{"path":"https://keras3.posit.co/dev/reference/op_istft.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Inverse Short-Time Fourier Transform along the last axis of the input. — op_istft","text":"x Tuple real imaginary parts input tensor. tensors list floating type. sequence_length integer representing sequence length. sequence_stride integer representing sequence hop size. fft_length integer representing size FFT produced stft. type int32. length integer representing output clipped exactly length. specified, padding clipping take place. Defaults NULL. window string, tensor window NULL. window string, available values \"hann\" \"hamming\". window tensor, used directly window length must sequence_length. window NULL, windowing used. Defaults \"hann\". center Whether x padded sides t-th sequence centered time t * sequence_stride. Defaults TRUE.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_istft.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Inverse Short-Time Fourier Transform along the last axis of the input. — op_istft","text":"tensor containing inverse Short-Time Fourier Transform along last axis x.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_istft.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Inverse Short-Time Fourier Transform along the last axis of the input. — op_istft","text":"","code":"x <- op_convert_to_tensor(c(0, 1, 2, 3, 4)) op_istft(op_stft(x, 1, 1, 1), 1, 1, 1) ## tf.Tensor([], shape=(0), dtype=float64) # array([0.0, 1.0, 2.0, 3.0, 4.0])"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_kaiser.html","id":null,"dir":"Reference","previous_headings":"","what":"Kaiser window function. — op_kaiser","title":"Kaiser window function. — op_kaiser","text":"Kaiser window defined : w[n] = I0(beta * sqrt(1 - (2 * n / (N - 1) - 1)^2)) / I0(beta) I0 modified zeroth-order Bessel function first kind.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_kaiser.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Kaiser window function. — op_kaiser","text":"","code":"op_kaiser(x, beta)"},{"path":"https://keras3.posit.co/dev/reference/op_kaiser.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Kaiser window function. — op_kaiser","text":"x Length window. Must positive integer. beta Shape parameter window.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_kaiser.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Kaiser window function. — op_kaiser","text":"1D tensor containing window values.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_kaiser.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Kaiser window function. — op_kaiser","text":"","code":"op_kaiser(5, beta = 14) ## tf.Tensor([7.7268669e-06 1.6493215e-01 1.0000000e+00 1.6493215e-01 7.7268669e-06], shape=(5), dtype=float32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_layer_normalization.html","id":null,"dir":"Reference","previous_headings":"","what":"Layer normalization (Ba et al., 2016). — op_layer_normalization","title":"Layer normalization (Ba et al., 2016). — op_layer_normalization","text":"Normalizes activations x example independently centering mean 0 scaling unit variance along specified axis.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_layer_normalization.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Layer normalization (Ba et al., 2016). — op_layer_normalization","text":"","code":"op_layer_normalization(   x,   gamma = NULL,   beta = NULL,   axis = -1L,   epsilon = NULL,   ... )"},{"path":"https://keras3.posit.co/dev/reference/op_layer_normalization.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Layer normalization (Ba et al., 2016). — op_layer_normalization","text":"x Input tensor. gamma Optional scaling factor applied normalized output. beta Optional offset added normalized output. axis Axis axes along compute statistics. Defaults -1. epsilon Small constant added variance numerical stability. ... forward/backward compatibility.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_layer_normalization.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Layer normalization (Ba et al., 2016). — op_layer_normalization","text":"Tensor shape x containing normalized values.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_layer_normalization.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Layer normalization (Ba et al., 2016). — op_layer_normalization","text":"","code":"x <- op_arange(5, dtype = \"float32\") op_layer_normalization(x) ## tf.Tensor([-1.4142134 -0.7071067  0.         0.7071068  1.4142137], shape=(5), dtype=float32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_leaky_relu.html","id":null,"dir":"Reference","previous_headings":"","what":"Leaky version of a Rectified Linear Unit activation function. — op_leaky_relu","title":"Leaky version of a Rectified Linear Unit activation function. — op_leaky_relu","text":"allows small gradient unit active, defined : f(x) = alpha * x x < 0 f(x) = x x >= 0.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_leaky_relu.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Leaky version of a Rectified Linear Unit activation function. — op_leaky_relu","text":"","code":"op_leaky_relu(x, negative_slope = 0.2)"},{"path":"https://keras3.posit.co/dev/reference/op_leaky_relu.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Leaky version of a Rectified Linear Unit activation function. — op_leaky_relu","text":"x Input tensor. negative_slope Slope activation function x < 0. Defaults 0.2.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_leaky_relu.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Leaky version of a Rectified Linear Unit activation function. — op_leaky_relu","text":"tensor shape x.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_leaky_relu.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Leaky version of a Rectified Linear Unit activation function. — op_leaky_relu","text":"","code":"x <- op_array(c(-1., 0., 1.)) op_leaky_relu(x) ## tf.Tensor([-0.2  0.   1. ], shape=(3), dtype=float32) # array([-0.2,  0. ,  1. ], shape=(3,), dtype=float64) x <- seq(-5, 5, .1) plot(x, op_leaky_relu(x),      type = 'l', panel.first = grid())"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_left_shift.html","id":null,"dir":"Reference","previous_headings":"","what":"Shift the bits of an integer to the left. — op_left_shift","title":"Shift the bits of an integer to the left. — op_left_shift","text":"Bits shifted left appending y 0s right x. Since internal representation numbers binary format, operation equivalent multiplying x 2^y.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_left_shift.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Shift the bits of an integer to the left. — op_left_shift","text":"","code":"op_left_shift(x, y)"},{"path":"https://keras3.posit.co/dev/reference/op_left_shift.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Shift the bits of an integer to the left. — op_left_shift","text":"x Input integer tensor. y Input integer tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_left_shift.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Shift the bits of an integer to the left. — op_left_shift","text":"Result tensor.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_less.html","id":null,"dir":"Reference","previous_headings":"","what":"Return the truth value of x1 < x2 element-wise. — op_less","title":"Return the truth value of x1 < x2 element-wise. — op_less","text":"Note function automatically called using R operator < tensor.","code":"(x <- op_arange(4)) ## tf.Tensor([1. 2. 3. 4.], shape=(4), dtype=float32) op_less(x, 2) ## tf.Tensor([ True False False False], shape=(4), dtype=bool) x < 2 ## tf.Tensor([ True False False False], shape=(4), dtype=bool)"},{"path":"https://keras3.posit.co/dev/reference/op_less.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Return the truth value of x1 < x2 element-wise. — op_less","text":"","code":"op_less(x1, x2)"},{"path":"https://keras3.posit.co/dev/reference/op_less.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Return the truth value of x1 < x2 element-wise. — op_less","text":"x1 First input tensor. x2 Second input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_less.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Return the truth value of x1 < x2 element-wise. — op_less","text":"Output tensor, element-wise comparison x1 x2.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_less_equal.html","id":null,"dir":"Reference","previous_headings":"","what":"Return the truth value of x1 <= x2 element-wise. — op_less_equal","title":"Return the truth value of x1 <= x2 element-wise. — op_less_equal","text":"Note function automatically called using R operator <= tensor.","code":"(x <- op_arange(4)) ## tf.Tensor([1. 2. 3. 4.], shape=(4), dtype=float32) op_less_equal(x, 2) ## tf.Tensor([ True  True False False], shape=(4), dtype=bool) x <= 2 ## tf.Tensor([ True  True False False], shape=(4), dtype=bool)"},{"path":"https://keras3.posit.co/dev/reference/op_less_equal.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Return the truth value of x1 <= x2 element-wise. — op_less_equal","text":"","code":"op_less_equal(x1, x2)"},{"path":"https://keras3.posit.co/dev/reference/op_less_equal.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Return the truth value of x1 <= x2 element-wise. — op_less_equal","text":"x1 First input tensor. x2 Second input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_less_equal.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Return the truth value of x1 <= x2 element-wise. — op_less_equal","text":"Output tensor, element-wise comparison x1 x2.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_linspace.html","id":null,"dir":"Reference","previous_headings":"","what":"Return evenly spaced numbers over a specified interval. — op_linspace","title":"Return evenly spaced numbers over a specified interval. — op_linspace","text":"Returns num evenly spaced samples, calculated interval [start, stop]. endpoint interval can optionally excluded.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_linspace.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Return evenly spaced numbers over a specified interval. — op_linspace","text":"","code":"op_linspace(   start,   stop,   num = 50L,   endpoint = TRUE,   retstep = FALSE,   dtype = NULL,   axis = 1L )"},{"path":"https://keras3.posit.co/dev/reference/op_linspace.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Return evenly spaced numbers over a specified interval. — op_linspace","text":"start starting value sequence. stop end value sequence, unless endpoint set FALSE. case, sequence consists last num + 1 evenly spaced samples, stop excluded. Note step size changes endpoint FALSE. num Number samples generate. Defaults 50. Must non-negative. endpoint TRUE, stop last sample. Otherwise, included. Defaults TRUE. retstep TRUE, return (samples, step), step spacing samples. dtype type output tensor. axis axis result store samples. Relevant start stop array-like. Defaults 1, first axis.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_linspace.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Return evenly spaced numbers over a specified interval. — op_linspace","text":"tensor evenly spaced numbers. retstep TRUE, returns (samples, step)","code":""},{"path":"https://keras3.posit.co/dev/reference/op_linspace.html","id":"note","dir":"Reference","previous_headings":"","what":"Note","title":"Return evenly spaced numbers over a specified interval. — op_linspace","text":"Torch backend support axis argument.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_log.html","id":null,"dir":"Reference","previous_headings":"","what":"Natural logarithm, element-wise. — op_log","title":"Natural logarithm, element-wise. — op_log","text":"Natural logarithm, element-wise.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_log.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Natural logarithm, element-wise. — op_log","text":"","code":"op_log(x)"},{"path":"https://keras3.posit.co/dev/reference/op_log.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Natural logarithm, element-wise. — op_log","text":"x Input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_log.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Natural logarithm, element-wise. — op_log","text":"Output tensor, element-wise natural logarithm x.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_log10.html","id":null,"dir":"Reference","previous_headings":"","what":"Return the base 10 logarithm of the input tensor, element-wise. — op_log10","title":"Return the base 10 logarithm of the input tensor, element-wise. — op_log10","text":"Return base 10 logarithm input tensor, element-wise.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_log10.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Return the base 10 logarithm of the input tensor, element-wise. — op_log10","text":"","code":"op_log10(x)"},{"path":"https://keras3.posit.co/dev/reference/op_log10.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Return the base 10 logarithm of the input tensor, element-wise. — op_log10","text":"x Input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_log10.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Return the base 10 logarithm of the input tensor, element-wise. — op_log10","text":"Output tensor, element-wise base 10 logarithm x.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_log1p.html","id":null,"dir":"Reference","previous_headings":"","what":"Returns the natural logarithm of one plus the x, element-wise. — op_log1p","title":"Returns the natural logarithm of one plus the x, element-wise. — op_log1p","text":"Calculates log(1 + x).","code":""},{"path":"https://keras3.posit.co/dev/reference/op_log1p.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Returns the natural logarithm of one plus the x, element-wise. — op_log1p","text":"","code":"op_log1p(x)"},{"path":"https://keras3.posit.co/dev/reference/op_log1p.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Returns the natural logarithm of one plus the x, element-wise. — op_log1p","text":"x Input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_log1p.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Returns the natural logarithm of one plus the x, element-wise. — op_log1p","text":"Output tensor, element-wise natural logarithm 1 + x.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_log2.html","id":null,"dir":"Reference","previous_headings":"","what":"Base-2 logarithm of x, element-wise. — op_log2","title":"Base-2 logarithm of x, element-wise. — op_log2","text":"Base-2 logarithm x, element-wise.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_log2.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Base-2 logarithm of x, element-wise. — op_log2","text":"","code":"op_log2(x)"},{"path":"https://keras3.posit.co/dev/reference/op_log2.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Base-2 logarithm of x, element-wise. — op_log2","text":"x Input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_log2.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Base-2 logarithm of x, element-wise. — op_log2","text":"Output tensor, element-wise base-2 logarithm x.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_log_sigmoid.html","id":null,"dir":"Reference","previous_headings":"","what":"Logarithm of the sigmoid activation function. — op_log_sigmoid","title":"Logarithm of the sigmoid activation function. — op_log_sigmoid","text":"defined f(x) = log(1 / (1 + exp(-x))).","code":""},{"path":"https://keras3.posit.co/dev/reference/op_log_sigmoid.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Logarithm of the sigmoid activation function. — op_log_sigmoid","text":"","code":"op_log_sigmoid(x)"},{"path":"https://keras3.posit.co/dev/reference/op_log_sigmoid.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Logarithm of the sigmoid activation function. — op_log_sigmoid","text":"x Input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_log_sigmoid.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Logarithm of the sigmoid activation function. — op_log_sigmoid","text":"tensor shape x.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_log_sigmoid.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Logarithm of the sigmoid activation function. — op_log_sigmoid","text":"","code":"x <- op_convert_to_tensor(c(-0.541391, 0.0, 0.50, 5.0)) op_log_sigmoid(x) ## tf.Tensor([-1.00004181 -0.69314718 -0.47407698 -0.00671535], shape=(4), dtype=float64)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_log_softmax.html","id":null,"dir":"Reference","previous_headings":"","what":"Log-softmax activation function. — op_log_softmax","title":"Log-softmax activation function. — op_log_softmax","text":"defined : f(x) = x - max(x) - log(sum(exp(x - max(x))))","code":""},{"path":"https://keras3.posit.co/dev/reference/op_log_softmax.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Log-softmax activation function. — op_log_softmax","text":"","code":"op_log_softmax(x, axis = -1L)"},{"path":"https://keras3.posit.co/dev/reference/op_log_softmax.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Log-softmax activation function. — op_log_softmax","text":"x Input tensor. axis Integer, axis along log-softmax applied. Defaults -1.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_log_softmax.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Log-softmax activation function. — op_log_softmax","text":"tensor shape x.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_log_softmax.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Log-softmax activation function. — op_log_softmax","text":"","code":"x <- op_array(c(-1., 0., 1.)) op_log_softmax(x) ## tf.Tensor([-2.407606   -1.4076059  -0.40760595], shape=(3), dtype=float32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_logaddexp.html","id":null,"dir":"Reference","previous_headings":"","what":"Logarithm of the sum of exponentiations of the inputs. — op_logaddexp","title":"Logarithm of the sum of exponentiations of the inputs. — op_logaddexp","text":"Calculates log(exp(x1) + exp(x2)).","code":""},{"path":"https://keras3.posit.co/dev/reference/op_logaddexp.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Logarithm of the sum of exponentiations of the inputs. — op_logaddexp","text":"","code":"op_logaddexp(x1, x2)"},{"path":"https://keras3.posit.co/dev/reference/op_logaddexp.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Logarithm of the sum of exponentiations of the inputs. — op_logaddexp","text":"x1 Input tensor. x2 Input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_logaddexp.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Logarithm of the sum of exponentiations of the inputs. — op_logaddexp","text":"Output tensor, element-wise logarithm sum exponentiations inputs.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_logdet.html","id":null,"dir":"Reference","previous_headings":"","what":"Computes log of the determinant of a hermitian positive definite matrix. — op_logdet","title":"Computes log of the determinant of a hermitian positive definite matrix. — op_logdet","text":"Computes log determinant hermitian positive definite matrix.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_logdet.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes log of the determinant of a hermitian positive definite matrix. — op_logdet","text":"","code":"op_logdet(x)"},{"path":"https://keras3.posit.co/dev/reference/op_logdet.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Computes log of the determinant of a hermitian positive definite matrix. — op_logdet","text":"x Input matrix. must 2D square.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_logdet.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Computes log of the determinant of a hermitian positive definite matrix. — op_logdet","text":"natural log determinant matrix.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_logical_and.html","id":null,"dir":"Reference","previous_headings":"","what":"Computes the element-wise logical AND of the given input tensors. — op_logical_and","title":"Computes the element-wise logical AND of the given input tensors. — op_logical_and","text":"Zeros treated FALSE non-zeros treated TRUE.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_logical_and.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes the element-wise logical AND of the given input tensors. — op_logical_and","text":"","code":"op_logical_and(x1, x2)"},{"path":"https://keras3.posit.co/dev/reference/op_logical_and.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Computes the element-wise logical AND of the given input tensors. — op_logical_and","text":"x1 Input tensor. x2 Input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_logical_and.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Computes the element-wise logical AND of the given input tensors. — op_logical_and","text":"Output tensor, element-wise logical inputs.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_logical_and.html","id":"details","dir":"Reference","previous_headings":"","what":"Details","title":"Computes the element-wise logical AND of the given input tensors. — op_logical_and","text":"Note function automatically called using R operator & tensor.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_logical_not.html","id":null,"dir":"Reference","previous_headings":"","what":"Computes the element-wise NOT of the given input tensor. — op_logical_not","title":"Computes the element-wise NOT of the given input tensor. — op_logical_not","text":"Zeros treated FALSE non-zeros treated TRUE. Note function automatically called using R operator ! tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_logical_not.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes the element-wise NOT of the given input tensor. — op_logical_not","text":"","code":"op_logical_not(x)"},{"path":"https://keras3.posit.co/dev/reference/op_logical_not.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Computes the element-wise NOT of the given input tensor. — op_logical_not","text":"x Input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_logical_not.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Computes the element-wise NOT of the given input tensor. — op_logical_not","text":"Output tensor, element-wise logical input.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_logical_or.html","id":null,"dir":"Reference","previous_headings":"","what":"Computes the element-wise logical OR of the given input tensors. — op_logical_or","title":"Computes the element-wise logical OR of the given input tensors. — op_logical_or","text":"Zeros treated FALSE non-zeros treated TRUE. Note function automatically called using R operator | tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_logical_or.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes the element-wise logical OR of the given input tensors. — op_logical_or","text":"","code":"op_logical_or(x1, x2)"},{"path":"https://keras3.posit.co/dev/reference/op_logical_or.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Computes the element-wise logical OR of the given input tensors. — op_logical_or","text":"x1 Input tensor. x2 Input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_logical_or.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Computes the element-wise logical OR of the given input tensors. — op_logical_or","text":"Output tensor, element-wise logical inputs.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_logical_xor.html","id":null,"dir":"Reference","previous_headings":"","what":"Compute the truth value of x1 XOR x2, element-wise. — op_logical_xor","title":"Compute the truth value of x1 XOR x2, element-wise. — op_logical_xor","text":"Compute truth value x1 XOR x2, element-wise.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_logical_xor.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Compute the truth value of x1 XOR x2, element-wise. — op_logical_xor","text":"","code":"op_logical_xor(x1, x2)"},{"path":"https://keras3.posit.co/dev/reference/op_logical_xor.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Compute the truth value of x1 XOR x2, element-wise. — op_logical_xor","text":"x1 First input tensor. x2 Second input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_logical_xor.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Compute the truth value of x1 XOR x2, element-wise. — op_logical_xor","text":"Output boolean tensor.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_logspace.html","id":null,"dir":"Reference","previous_headings":"","what":"Returns numbers spaced evenly on a log scale. — op_logspace","title":"Returns numbers spaced evenly on a log scale. — op_logspace","text":"linear space, sequence starts base ** start ends base ** stop (see endpoint ).","code":""},{"path":"https://keras3.posit.co/dev/reference/op_logspace.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Returns numbers spaced evenly on a log scale. — op_logspace","text":"","code":"op_logspace(   start,   stop,   num = 50L,   endpoint = TRUE,   base = 10L,   dtype = NULL,   axis = 1L )"},{"path":"https://keras3.posit.co/dev/reference/op_logspace.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Returns numbers spaced evenly on a log scale. — op_logspace","text":"start starting value sequence. stop final value sequence, unless endpoint FALSE. case, num + 1 values spaced interval log-space, last (sequence length num) returned. num Number samples generate. Defaults 50. endpoint TRUE, stop last sample. Otherwise, included. Defaults TRUE. base base log space. Defaults 10. dtype type output tensor. axis axis result store samples. Relevant start stop array-like.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_logspace.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Returns numbers spaced evenly on a log scale. — op_logspace","text":"tensor evenly spaced samples log scale.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_logspace.html","id":"note","dir":"Reference","previous_headings":"","what":"Note","title":"Returns numbers spaced evenly on a log scale. — op_logspace","text":"Torch backend support axis argument.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_logsumexp.html","id":null,"dir":"Reference","previous_headings":"","what":"Computes the logarithm of sum of exponentials of elements in a tensor. — op_logsumexp","title":"Computes the logarithm of sum of exponentials of elements in a tensor. — op_logsumexp","text":"Computes logarithm sum exponentials elements tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_logsumexp.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes the logarithm of sum of exponentials of elements in a tensor. — op_logsumexp","text":"","code":"op_logsumexp(x, axis = NULL, keepdims = FALSE)"},{"path":"https://keras3.posit.co/dev/reference/op_logsumexp.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Computes the logarithm of sum of exponentials of elements in a tensor. — op_logsumexp","text":"x Input tensor. axis integer list integers specifying axis/axes along compute sum. NULL, sum computed elements. Defaults NULL. keepdims boolean indicating whether keep dimensions input tensor computing sum. Defaults FALSE.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_logsumexp.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Computes the logarithm of sum of exponentials of elements in a tensor. — op_logsumexp","text":"tensor containing logarithm sum exponentials elements x.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_logsumexp.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Computes the logarithm of sum of exponentials of elements in a tensor. — op_logsumexp","text":"","code":"x <- op_convert_to_tensor(c(1, 2, 3)) op_logsumexp(x) ## tf.Tensor(3.4076059644443806, shape=(), dtype=float64)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_lstsq.html","id":null,"dir":"Reference","previous_headings":"","what":"Return the least-squares solution to a linear matrix equation. — op_lstsq","title":"Return the least-squares solution to a linear matrix equation. — op_lstsq","text":"Computes vector x approximately solves equation %*% x = b. equation may -, well-, -determined (.e., number linearly independent rows can less , equal , greater number linearly independent columns). square full rank, x (round-error) exact solution equation. Else, x minimizes L2 norm b - %*% x. multiple minimizing solutions, one smallest L2 norm  returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_lstsq.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Return the least-squares solution to a linear matrix equation. — op_lstsq","text":"","code":"op_lstsq(a, b, rcond = NULL)"},{"path":"https://keras3.posit.co/dev/reference/op_lstsq.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Return the least-squares solution to a linear matrix equation. — op_lstsq","text":"\"Coefficient\" matrix shape (M, N). b Ordinate \"dependent variable\" values, shape (M) (M, K). b two-dimensional, least-squares solution calculated K columns b. rcond Cut-ratio small singular values . purposes rank determination, singular values treated zero smaller rcond times largest singular value .","code":""},{"path":"https://keras3.posit.co/dev/reference/op_lstsq.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Return the least-squares solution to a linear matrix equation. — op_lstsq","text":"Tensor shape (N) (N, K) containing least-squares solutions. NOTE: output differs numpy.linalg.lstsq(). NumPy returns tuple four elements, first least-squares solutions others essentially never used. Keras returns first value. done ensure consistency across backends (achieved values) simplify API.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_lu_factor.html","id":null,"dir":"Reference","previous_headings":"","what":"Computes the lower-upper decomposition of a square matrix. — op_lu_factor","title":"Computes the lower-upper decomposition of a square matrix. — op_lu_factor","text":"Computes lower-upper decomposition square matrix.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_lu_factor.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes the lower-upper decomposition of a square matrix. — op_lu_factor","text":"","code":"op_lu_factor(x)"},{"path":"https://keras3.posit.co/dev/reference/op_lu_factor.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Computes the lower-upper decomposition of a square matrix. — op_lu_factor","text":"x tensor shape (..., M, M).","code":""},{"path":"https://keras3.posit.co/dev/reference/op_lu_factor.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Computes the lower-upper decomposition of a square matrix. — op_lu_factor","text":"tuple two tensors: tensor shape (..., M, M) containing lower upper triangular matrices tensor shape (..., M) containing pivots.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_map.html","id":null,"dir":"Reference","previous_headings":"","what":"Map a function over leading array axes. — op_map","title":"Map a function over leading array axes. — op_map","text":"Like purrr::map() base::lapply(), except inputs outputs form stacked arrays. Consider using op_vectorized_map() transform instead, unless need apply function element element reduced memory usage heterogeneous computation control flow primitives. xs array type, semantics op_map() match implementation:","code":"op_map <- function(xs, f) {   xs |>     op_unstack() |>     lapply(f) |>     op_stack() }"},{"path":"https://keras3.posit.co/dev/reference/op_map.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Map a function over leading array axes. — op_map","text":"","code":"op_map(xs, f)"},{"path":"https://keras3.posit.co/dev/reference/op_map.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Map a function over leading array axes. — op_map","text":"xs Values map along leading axis. f Callable defines function apply element-wise first axis axes xs.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_map.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Map a function over leading array axes. — op_map","text":"Mapped values.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_map.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Map a function over leading array axes. — op_map","text":"","code":"f <- function(x) x^2 xs <- op_arange(10) ys <- op_map(xs, f) ys ## tf.Tensor([  1.   4.   9.  16.  25.  36.  49.  64.  81. 100.], shape=(10), dtype=float32) f <- function(x) list(y1 = x^2, y2 = x * 10)  # Can have nested outputs ys <- op_map(xs, f) ys$y1 ## tf.Tensor([  1.   4.   9.  16.  25.  36.  49.  64.  81. 100.], shape=(10), dtype=float32) ys$y2 ## tf.Tensor([ 10.  20.  30.  40.  50.  60.  70.  80.  90. 100.], shape=(10), dtype=float32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_matmul.html","id":null,"dir":"Reference","previous_headings":"","what":"Matrix product of two tensors. — op_matmul","title":"Matrix product of two tensors. — op_matmul","text":"tensors 1-dimensional, dot product (scalar) returned. either tensor N-D, N > 2, treated stack matrices residing last two indexes broadcast accordingly. first tensor 1-D, promoted matrix prepending 1 dimensions. matrix multiplication prepended 1 removed. second tensor 1-D, promoted matrix appending 1 dimensions. matrix multiplication appended 1 removed.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_matmul.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Matrix product of two tensors. — op_matmul","text":"","code":"op_matmul(x1, x2)"},{"path":"https://keras3.posit.co/dev/reference/op_matmul.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Matrix product of two tensors. — op_matmul","text":"x1 First tensor. x2 Second tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_matmul.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Matrix product of two tensors. — op_matmul","text":"Output tensor, matrix product inputs.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_max.html","id":null,"dir":"Reference","previous_headings":"","what":"Return the maximum of a tensor or maximum along an axis. — op_max","title":"Return the maximum of a tensor or maximum along an axis. — op_max","text":"Return maximum tensor maximum along axis.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_max.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Return the maximum of a tensor or maximum along an axis. — op_max","text":"","code":"op_max(x, axis = NULL, keepdims = FALSE, initial = NULL)"},{"path":"https://keras3.posit.co/dev/reference/op_max.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Return the maximum of a tensor or maximum along an axis. — op_max","text":"x Input tensor. axis Axis axes along operate. default, flattened input used. keepdims set TRUE, axes reduced left result dimensions size one. Defaults FALSE. initial minimum value output element. Defaults NULL.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_max.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Return the maximum of a tensor or maximum along an axis. — op_max","text":"Maximum x.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_max.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Return the maximum of a tensor or maximum along an axis. — op_max","text":"","code":"(x <- op_convert_to_tensor(rbind(c(1, 3, 5), c(1, 5, 2)))) ## tf.Tensor( ## [[1. 3. 5.] ##  [1. 5. 2.]], shape=(2, 3), dtype=float64) op_max(x) ## tf.Tensor(5.0, shape=(), dtype=float64) op_max(x, axis = 1) ## tf.Tensor([1. 5. 5.], shape=(3), dtype=float64) op_max(x, axis = 1, keepdims = TRUE) ## tf.Tensor([[1. 5. 5.]], shape=(1, 3), dtype=float64)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_max_pool.html","id":null,"dir":"Reference","previous_headings":"","what":"Max pooling operation. — op_max_pool","title":"Max pooling operation. — op_max_pool","text":"Max pooling operation.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_max_pool.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Max pooling operation. — op_max_pool","text":"","code":"op_max_pool(   inputs,   pool_size,   strides = NULL,   padding = \"valid\",   data_format = NULL )"},{"path":"https://keras3.posit.co/dev/reference/op_max_pool.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Max pooling operation. — op_max_pool","text":"inputs Tensor rank N+2. inputs shape (batch_size,) + inputs_spatial_shape + (num_channels,) data_format = \"channels_last\", (batch_size, num_channels) + inputs_spatial_shape data_format = \"channels_first\". Pooling happens spatial dimensions . pool_size int tuple/list integers size len(inputs_spatial_shape), specifying size pooling window spatial dimension input tensor. pool_size int, every spatial dimension shares pool_size. strides int tuple/list integers size len(inputs_spatial_shape). stride sliding window spatial dimension input tensor. strides int, every spatial dimension shares strides. padding string, either \"valid\" \"\". \"valid\" means padding applied, \"\" results padding evenly left/right /input output height/width dimension input strides = 1. data_format string, either \"channels_last\" \"channels_first\". data_format determines ordering dimensions inputs. data_format = \"channels_last\", inputs shape (batch_size, ..., channels) data_format = \"channels_first\", inputs shape (batch_size, channels, ...).","code":""},{"path":"https://keras3.posit.co/dev/reference/op_max_pool.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Max pooling operation. — op_max_pool","text":"tensor rank N+2, result max pooling operation.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_maximum.html","id":null,"dir":"Reference","previous_headings":"","what":"Element-wise maximum of x1 and x2. — op_maximum","title":"Element-wise maximum of x1 and x2. — op_maximum","text":"Element-wise maximum x1 x2.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_maximum.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Element-wise maximum of x1 and x2. — op_maximum","text":"","code":"op_maximum(x1, x2)  op_pmax(x1, x2)"},{"path":"https://keras3.posit.co/dev/reference/op_maximum.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Element-wise maximum of x1 and x2. — op_maximum","text":"x1 First tensor. x2 Second tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_maximum.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Element-wise maximum of x1 and x2. — op_maximum","text":"Output tensor, element-wise maximum x1 x2.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_mean.html","id":null,"dir":"Reference","previous_headings":"","what":"Compute the arithmetic mean along the specified axes. — op_mean","title":"Compute the arithmetic mean along the specified axes. — op_mean","text":"Compute arithmetic mean along specified axes.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_mean.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Compute the arithmetic mean along the specified axes. — op_mean","text":"","code":"op_mean(x, axis = NULL, keepdims = FALSE)"},{"path":"https://keras3.posit.co/dev/reference/op_mean.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Compute the arithmetic mean along the specified axes. — op_mean","text":"x Input tensor. axis Axis axes along means computed. default compute mean flattened tensor. keepdims set TRUE, axes reduced left result dimensions size one.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_mean.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Compute the arithmetic mean along the specified axes. — op_mean","text":"Output tensor containing mean values.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_median.html","id":null,"dir":"Reference","previous_headings":"","what":"Compute the median along the specified axis. — op_median","title":"Compute the median along the specified axis. — op_median","text":"Compute median along specified axis.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_median.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Compute the median along the specified axis. — op_median","text":"","code":"op_median(x, axis = NULL, keepdims = FALSE)"},{"path":"https://keras3.posit.co/dev/reference/op_median.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Compute the median along the specified axis. — op_median","text":"x Input tensor. axis Axis axes along medians computed. Defaults axis = NULL compute median(s) along flattened version array. keepdims set TRUE, axes reduce left result dimensions size one.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_median.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Compute the median along the specified axis. — op_median","text":"output tensor.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_meshgrid.html","id":null,"dir":"Reference","previous_headings":"","what":"Creates grids of coordinates from coordinate vectors. — op_meshgrid","title":"Creates grids of coordinates from coordinate vectors. — op_meshgrid","text":"Given N 1-D tensors T0, T1, ..., TN-1 inputs corresponding lengths S0, S1, ..., SN-1, creates N N-dimensional tensors G0, G1, ..., GN-1 shape (S0, ..., SN-1) output Gi constructed expanding Ti result shape.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_meshgrid.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Creates grids of coordinates from coordinate vectors. — op_meshgrid","text":"","code":"op_meshgrid(..., indexing = \"xy\")"},{"path":"https://keras3.posit.co/dev/reference/op_meshgrid.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Creates grids of coordinates from coordinate vectors. — op_meshgrid","text":"... 1-D tensors representing coordinates grid. indexing \"xy\" \"ij\". \"xy\" cartesian; \"ij\" matrix indexing output. Defaults \"xy\".","code":""},{"path":"https://keras3.posit.co/dev/reference/op_meshgrid.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Creates grids of coordinates from coordinate vectors. — op_meshgrid","text":"Sequence N tensors.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_meshgrid.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Creates grids of coordinates from coordinate vectors. — op_meshgrid","text":"","code":"x <- op_array(c(1, 2, 3), \"int32\") y <- op_array(c(4, 5, 6), \"int32\") c(grid_x, grid_y) %<-% op_meshgrid(x, y, indexing = \"ij\") grid_x ## tf.Tensor( ## [[1 1 1] ##  [2 2 2] ##  [3 3 3]], shape=(3, 3), dtype=int32) # array([[1, 1, 1], #        [2, 2, 2], #        [3, 3, 3])) grid_y ## tf.Tensor( ## [[4 5 6] ##  [4 5 6] ##  [4 5 6]], shape=(3, 3), dtype=int32) # array([[4, 5, 6], #        [4, 5, 6], #        [4, 5, 6]))"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_min.html","id":null,"dir":"Reference","previous_headings":"","what":"Return the minimum of a tensor or minimum along an axis. — op_min","title":"Return the minimum of a tensor or minimum along an axis. — op_min","text":"Return minimum tensor minimum along axis.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_min.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Return the minimum of a tensor or minimum along an axis. — op_min","text":"","code":"op_min(x, axis = NULL, keepdims = FALSE, initial = NULL)"},{"path":"https://keras3.posit.co/dev/reference/op_min.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Return the minimum of a tensor or minimum along an axis. — op_min","text":"x Input tensor. axis Axis axes along operate. default, flattened input used. keepdims set TRUE, axes reduced left result dimensions size one. Defaults FALSE. initial maximum value output element. Defaults NULL.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_min.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Return the minimum of a tensor or minimum along an axis. — op_min","text":"Minimum x.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_min.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Return the minimum of a tensor or minimum along an axis. — op_min","text":"","code":"(x <- op_convert_to_tensor(rbind(c(1, 3, 5), c(1, 5, 2)))) ## tf.Tensor( ## [[1. 3. 5.] ##  [1. 5. 2.]], shape=(2, 3), dtype=float64) op_min(x) ## tf.Tensor(1.0, shape=(), dtype=float64) op_min(x, axis = 1) ## tf.Tensor([1. 3. 2.], shape=(3), dtype=float64) op_min(x, axis = 1, keepdims = TRUE) ## tf.Tensor([[1. 3. 2.]], shape=(1, 3), dtype=float64)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_minimum.html","id":null,"dir":"Reference","previous_headings":"","what":"Element-wise minimum of x1 and x2. — op_minimum","title":"Element-wise minimum of x1 and x2. — op_minimum","text":"Element-wise minimum x1 x2.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_minimum.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Element-wise minimum of x1 and x2. — op_minimum","text":"","code":"op_minimum(x1, x2)  op_pmin(x1, x2)"},{"path":"https://keras3.posit.co/dev/reference/op_minimum.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Element-wise minimum of x1 and x2. — op_minimum","text":"x1 First tensor. x2 Second tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_minimum.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Element-wise minimum of x1 and x2. — op_minimum","text":"Output tensor, element-wise minimum x1 x2.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_mod.html","id":null,"dir":"Reference","previous_headings":"","what":"Returns the element-wise remainder of division. — op_mod","title":"Returns the element-wise remainder of division. — op_mod","text":"Note function automatically called using R operator %% tensor.","code":"(x <- op_arange(10)) ## tf.Tensor([ 1.  2.  3.  4.  5.  6.  7.  8.  9. 10.], shape=(10), dtype=float32) op_mod(x, 3) ## tf.Tensor([1. 2. 0. 1. 2. 0. 1. 2. 0. 1.], shape=(10), dtype=float32) x %% 3 ## tf.Tensor([1. 2. 0. 1. 2. 0. 1. 2. 0. 1.], shape=(10), dtype=float32)"},{"path":"https://keras3.posit.co/dev/reference/op_mod.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Returns the element-wise remainder of division. — op_mod","text":"","code":"op_mod(x1, x2)"},{"path":"https://keras3.posit.co/dev/reference/op_mod.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Returns the element-wise remainder of division. — op_mod","text":"x1 First tensor. x2 Second tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_mod.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Returns the element-wise remainder of division. — op_mod","text":"Output tensor, element-wise remainder division.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_moments.html","id":null,"dir":"Reference","previous_headings":"","what":"Calculates the mean and variance of x. — op_moments","title":"Calculates the mean and variance of x. — op_moments","text":"mean variance calculated aggregating contents x across axes. x 1-D axes = c(1) just mean variance vector.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_moments.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Calculates the mean and variance of x. — op_moments","text":"","code":"op_moments(x, axes, keepdims = FALSE, synchronized = FALSE)"},{"path":"https://keras3.posit.co/dev/reference/op_moments.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Calculates the mean and variance of x. — op_moments","text":"x Input tensor. axes list axes compute mean variance. keepdims set TRUE, axes reduced left result dimensions size one. synchronized applicable TensorFlow backend. TRUE, synchronizes global batch statistics (mean variance) across devices training step distributed training strategy. FALSE, replica uses local batch statistics.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_moments.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Calculates the mean and variance of x. — op_moments","text":"list containing two tensors - mean variance.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_moments.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Calculates the mean and variance of x. — op_moments","text":"","code":"x <- op_convert_to_tensor(c(0, 1, 2, 3, 100), dtype = \"float32\") op_moments(x, axes = c(1)) ## [[1]] ## tf.Tensor(21.2, shape=(), dtype=float32) ## ## [[2]] ## tf.Tensor(1553.3601, shape=(), dtype=float32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_moveaxis.html","id":null,"dir":"Reference","previous_headings":"","what":"Move axes of a tensor to new positions. — op_moveaxis","title":"Move axes of a tensor to new positions. — op_moveaxis","text":"axes remain original order.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_moveaxis.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Move axes of a tensor to new positions. — op_moveaxis","text":"","code":"op_moveaxis(x, source, destination)"},{"path":"https://keras3.posit.co/dev/reference/op_moveaxis.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Move axes of a tensor to new positions. — op_moveaxis","text":"x Tensor whose axes reordered. source Original positions axes move. must unique. destination Destinations positions original axes. must also unique.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_moveaxis.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Move axes of a tensor to new positions. — op_moveaxis","text":"Tensor moved axes.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_multi_hot.html","id":null,"dir":"Reference","previous_headings":"","what":"Encodes integer labels as multi-hot vectors. — op_multi_hot","title":"Encodes integer labels as multi-hot vectors. — op_multi_hot","text":"function encodes integer labels multi-hot vectors, label mapped binary value resulting vector.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_multi_hot.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Encodes integer labels as multi-hot vectors. — op_multi_hot","text":"","code":"op_multi_hot(   inputs,   num_classes,   axis = -1L,   dtype = NULL,   sparse = FALSE,   ... )"},{"path":"https://keras3.posit.co/dev/reference/op_multi_hot.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Encodes integer labels as multi-hot vectors. — op_multi_hot","text":"inputs Tensor integer labels converted multi-hot vectors. num_classes Integer, total number unique classes. axis (optional) Axis along multi-hot encoding added. Defaults -1, corresponds last dimension. dtype (optional) data type resulting tensor. Default backend's float type. sparse Whether return sparse tensor; backends support sparse tensors. ... forward/backwards compatability","code":""},{"path":"https://keras3.posit.co/dev/reference/op_multi_hot.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Encodes integer labels as multi-hot vectors. — op_multi_hot","text":"Tensor: multi-hot encoded tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_multi_hot.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Encodes integer labels as multi-hot vectors. — op_multi_hot","text":"","code":"data <- op_convert_to_tensor(c(0, 4)) op_multi_hot(data, num_classes = 5) ## tf.Tensor([1. 0. 0. 0. 1.], shape=(5), dtype=float32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_multiply.html","id":null,"dir":"Reference","previous_headings":"","what":"Multiply arguments element-wise. — op_multiply","title":"Multiply arguments element-wise. — op_multiply","text":"Note function automatically called using R operator * tensor.","code":"(x <- op_arange(4)) ## tf.Tensor([1. 2. 3. 4.], shape=(4), dtype=float32) op_multiply(x, x) ## tf.Tensor([ 1.  4.  9. 16.], shape=(4), dtype=float32) x * x ## tf.Tensor([ 1.  4.  9. 16.], shape=(4), dtype=float32)"},{"path":"https://keras3.posit.co/dev/reference/op_multiply.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Multiply arguments element-wise. — op_multiply","text":"","code":"op_multiply(x1, x2)"},{"path":"https://keras3.posit.co/dev/reference/op_multiply.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Multiply arguments element-wise. — op_multiply","text":"x1 First input tensor. x2 Second input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_multiply.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Multiply arguments element-wise. — op_multiply","text":"Output tensor, element-wise product x1 x2.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_nan_to_num.html","id":null,"dir":"Reference","previous_headings":"","what":"Replace NaN with zero and infinity with large finite numbers. — op_nan_to_num","title":"Replace NaN with zero and infinity with large finite numbers. — op_nan_to_num","text":"Replace NaN zero infinity large finite numbers.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_nan_to_num.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Replace NaN with zero and infinity with large finite numbers. — op_nan_to_num","text":"","code":"op_nan_to_num(x, nan = 0, posinf = NULL, neginf = NULL)"},{"path":"https://keras3.posit.co/dev/reference/op_nan_to_num.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Replace NaN with zero and infinity with large finite numbers. — op_nan_to_num","text":"x Input data. nan Optional float int. Value replace NaN entries . posinf Optional float int. Value replace positive infinity . neginf Optional float int. Value replace negative infinity .","code":""},{"path":"https://keras3.posit.co/dev/reference/op_nan_to_num.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Replace NaN with zero and infinity with large finite numbers. — op_nan_to_num","text":"x, non-finite values replaced.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_nan_to_num.html","id":"example","dir":"Reference","previous_headings":"","what":"Example","title":"Replace NaN with zero and infinity with large finite numbers. — op_nan_to_num","text":"","code":"(x <- op_convert_to_tensor(c(1, NaN, -Inf, Inf))) ## tf.Tensor([  1.  nan -inf  inf], shape=(4), dtype=float64) op_nan_to_num(x) ## tf.Tensor([ 1.00000000e+000  0.00000000e+000 -1.79769313e+308  1.79769313e+308], shape=(4), dtype=float64) op_nan_to_num(x, nan = -1, posinf = 2, neginf = -2) ## tf.Tensor([ 1. -1. -2.  2.], shape=(4), dtype=float64)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_ndim.html","id":null,"dir":"Reference","previous_headings":"","what":"Return the number of dimensions of a tensor. — op_ndim","title":"Return the number of dimensions of a tensor. — op_ndim","text":"Return number dimensions tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_ndim.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Return the number of dimensions of a tensor. — op_ndim","text":"","code":"op_ndim(x)"},{"path":"https://keras3.posit.co/dev/reference/op_ndim.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Return the number of dimensions of a tensor. — op_ndim","text":"x Input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_ndim.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Return the number of dimensions of a tensor. — op_ndim","text":"number dimensions x.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_negative.html","id":null,"dir":"Reference","previous_headings":"","what":"Numerical negative, element-wise. — op_negative","title":"Numerical negative, element-wise. — op_negative","text":"Note function automatically called using unary R operator - tensor.","code":"(x <- op_arange(4)) ## tf.Tensor([1. 2. 3. 4.], shape=(4), dtype=float32) op_negative(x) ## tf.Tensor([-1. -2. -3. -4.], shape=(4), dtype=float32) -x ## tf.Tensor([-1. -2. -3. -4.], shape=(4), dtype=float32)"},{"path":"https://keras3.posit.co/dev/reference/op_negative.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Numerical negative, element-wise. — op_negative","text":"","code":"op_negative(x)"},{"path":"https://keras3.posit.co/dev/reference/op_negative.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Numerical negative, element-wise. — op_negative","text":"x Input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_negative.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Numerical negative, element-wise. — op_negative","text":"Output tensor, y = -x.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_nonzero.html","id":null,"dir":"Reference","previous_headings":"","what":"Return the indices of the elements that are non-zero. — op_nonzero","title":"Return the indices of the elements that are non-zero. — op_nonzero","text":"Return indices elements non-zero.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_nonzero.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Return the indices of the elements that are non-zero. — op_nonzero","text":"","code":"op_nonzero(x)"},{"path":"https://keras3.posit.co/dev/reference/op_nonzero.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Return the indices of the elements that are non-zero. — op_nonzero","text":"x Input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_nonzero.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Return the indices of the elements that are non-zero. — op_nonzero","text":"Indices elements non-zero.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_nonzero.html","id":"example","dir":"Reference","previous_headings":"","what":"Example","title":"Return the indices of the elements that are non-zero. — op_nonzero","text":"op_nonzero() indices can used <tensor>@r[","code":"(x <- op_scatter(indices = rbind(1, 5, 10), values = c(1, 2, 3), shape = c(10))) ## tf.Tensor([1. 0. 0. 0. 2. 0. 0. 0. 0. 3.], shape=(10), dtype=float64) (nz <- op_nonzero(x)) ## tf.Tensor( ## [[ 1] ##  [ 5] ##  [10]], shape=(3, 1), dtype=int32) x@r[nz] ## tf.Tensor([1. 2. 3.], shape=(3), dtype=float64) # same as `x@r[nz]` x@r[x != 0] ## tf.Tensor([1. 2. 3.], shape=(3), dtype=float64) x@r[op_cast(x, \"bool\")] ## tf.Tensor([1. 2. 3.], shape=(3), dtype=float64) # 2d example (x2 <- op_stack(c(x, op_roll(x, 1), op_roll(x, 2)))) ## tf.Tensor( ## [[1. 0. 0. 0. 2. 0. 0. 0. 0. 3.] ##  [3. 1. 0. 0. 0. 2. 0. 0. 0. 0.] ##  [0. 3. 1. 0. 0. 0. 2. 0. 0. 0.]], shape=(3, 10), dtype=float64) x2@r[op_nonzero(x2)] ## tf.Tensor([1. 2. 3. 3. 1. 2. 3. 1. 2.], shape=(9), dtype=float64) x3 <- op_stack(c(x2, x2*1.1)) x3@r[op_nonzero(x3)] ## tf.Tensor([1.  2.  3.  3.  1.  2.  3.  1.  2.  1.1 2.2 3.3 3.3 1.1 2.2 3.3 1.1 2.2], shape=(18), dtype=float64)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_norm.html","id":null,"dir":"Reference","previous_headings":"","what":"Matrix or vector norm. — op_norm","title":"Matrix or vector norm. — op_norm","text":"function able return one eight different matrix norms, one infinite number vector norms (described ), depending value ord parameter.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_norm.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Matrix or vector norm. — op_norm","text":"","code":"op_norm(x, ord = NULL, axis = NULL, keepdims = FALSE)"},{"path":"https://keras3.posit.co/dev/reference/op_norm.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Matrix or vector norm. — op_norm","text":"x Input tensor. ord Order norm (see table Notes). default NULL. axis axis integer, specifies axis x along compute vector norms. axis length 2 vector, specifies axes hold 2-D matrices, matrix norms matrices computed. keepdims set TRUE, axes reduced left result dimensions size one.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_norm.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Matrix or vector norm. — op_norm","text":"Norm matrix vector(s).","code":""},{"path":"https://keras3.posit.co/dev/reference/op_norm.html","id":"note","dir":"Reference","previous_headings":"","what":"Note","title":"Matrix or vector norm. — op_norm","text":"values ord < 1, result , strictly speaking, mathematical 'norm', may still useful various numerical purposes. following norms can calculated: matrices: ord=NULL: Frobenius norm ord=\"fro\": Frobenius norm ord=\"nuc\": nuclear norm ord=Inf: max(sum(abs(x), axis=2)) ord=-Inf: min(sum(abs(x), axis=2)) ord=0: supported ord=1: max(sum(abs(x), axis=1)) ord=-1: min(sum(abs(x), axis=1)) ord=2: 2-norm (largest sing. value) ord=-2: smallest singular value : supported vectors: ord=NULL: 2-norm ord=\"fro\": supported ord=\"nuc\": supported ord=Inf: max(abs(x)) ord=-Inf: min(abs(x)) ord=0: sum(x != 0) ord=1: ord=-1: ord=2: ord=-2: : sum(abs(x)^ord)^(1/ord)","code":""},{"path":"https://keras3.posit.co/dev/reference/op_norm.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Matrix or vector norm. — op_norm","text":"","code":"x <- op_reshape(op_arange(9, dtype=\"float32\") - 4, c(3, 3)) op_norm(x) ## tf.Tensor(8.306623, shape=(), dtype=float32) # 7.7459664"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_normalize.html","id":null,"dir":"Reference","previous_headings":"","what":"Normalizes x over the specified axis. — op_normalize","title":"Normalizes x over the specified axis. — op_normalize","text":"defined : op_normalize(x) = x / max(norm(x), epsilon).","code":""},{"path":"https://keras3.posit.co/dev/reference/op_normalize.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Normalizes x over the specified axis. — op_normalize","text":"","code":"op_normalize(x, axis = -1L, order = 2L, epsilon = NULL)"},{"path":"https://keras3.posit.co/dev/reference/op_normalize.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Normalizes x over the specified axis. — op_normalize","text":"x Input tensor. axis axis axes along perform normalization. Default -1. order exponent value norm formulation. Defaults 2. epsilon lower bound value norm. Defaults config_epsilon().","code":""},{"path":"https://keras3.posit.co/dev/reference/op_normalize.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Normalizes x over the specified axis. — op_normalize","text":"normalized array.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_normalize.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Normalizes x over the specified axis. — op_normalize","text":"","code":"x <- op_convert_to_tensor(rbind(c(1, 2, 3), c(4, 5, 6))) x_norm <- op_normalize(x) x_norm ## tf.Tensor( ## [[0.26726124 0.53452248 0.80178373] ##  [0.45584231 0.56980288 0.68376346]], shape=(2, 3), dtype=float64)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_not_equal.html","id":null,"dir":"Reference","previous_headings":"","what":"Return (x1 != x2) element-wise. — op_not_equal","title":"Return (x1 != x2) element-wise. — op_not_equal","text":"Note function automatically called using R operator != tensor.","code":"(x <- op_arange(4)) ## tf.Tensor([1. 2. 3. 4.], shape=(4), dtype=float32) op_not_equal(x, 2) ## tf.Tensor([ True False  True  True], shape=(4), dtype=bool) x != 2 ## tf.Tensor([ True False  True  True], shape=(4), dtype=bool)"},{"path":"https://keras3.posit.co/dev/reference/op_not_equal.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Return (x1 != x2) element-wise. — op_not_equal","text":"","code":"op_not_equal(x1, x2)"},{"path":"https://keras3.posit.co/dev/reference/op_not_equal.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Return (x1 != x2) element-wise. — op_not_equal","text":"x1 First input tensor. x2 Second input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_not_equal.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Return (x1 != x2) element-wise. — op_not_equal","text":"Output tensor, element-wise comparison x1 x2.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_one_hot.html","id":null,"dir":"Reference","previous_headings":"","what":"Converts integer tensor x into a one-hot tensor. — op_one_hot","title":"Converts integer tensor x into a one-hot tensor. — op_one_hot","text":"one-hot encoding representation integer value converted binary vector length equal num_classes, index corresponding integer value marked 1, indices marked 0.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_one_hot.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Converts integer tensor x into a one-hot tensor. — op_one_hot","text":"","code":"op_one_hot(   x,   num_classes,   axis = -1L,   dtype = NULL,   sparse = FALSE,   zero_indexed = FALSE )"},{"path":"https://keras3.posit.co/dev/reference/op_one_hot.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Converts integer tensor x into a one-hot tensor. — op_one_hot","text":"x Integer tensor encoded. shape can arbitrary, dtype integer. R factors coerced integer. num_classes Number classes one-hot encoding. x factor num_classes NULL missing, levels(x) used. axis Axis along encoding performed. -1 represents last axis. Defaults -1. dtype (Optional) Data type output tensor. provided, defaults default data type backend. sparse Whether return sparse tensor; backends support sparse tensors. zero_indexed TRUE, treats indices zero-based (0 encodes first position); FALSE (default), treats indices one-based (1 encodes first position).","code":""},{"path":"https://keras3.posit.co/dev/reference/op_one_hot.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Converts integer tensor x into a one-hot tensor. — op_one_hot","text":"Integer tensor: One-hot encoded tensor shape x except specified axis dimension, length num_classes. dtype output tensor determined dtype default data type backend.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_one_hot.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Converts integer tensor x into a one-hot tensor. — op_one_hot","text":"","code":"x <- op_array(c(2, 4, 3, 1), \"int32\") op_one_hot(x, num_classes = 4) ## tf.Tensor( ## [[0. 1. 0. 0.] ##  [0. 0. 0. 1.] ##  [0. 0. 1. 0.] ##  [1. 0. 0. 0.]], shape=(4, 4), dtype=float32) op_one_hot(x - 1, num_classes = 4, zero_indexed = TRUE) ## tf.Tensor( ## [[0. 1. 0. 0.] ##  [0. 0. 0. 1.] ##  [0. 0. 1. 0.] ##  [1. 0. 0. 0.]], shape=(4, 4), dtype=float32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_ones.html","id":null,"dir":"Reference","previous_headings":"","what":"Return a new tensor of given shape and type, filled with ones. — op_ones","title":"Return a new tensor of given shape and type, filled with ones. — op_ones","text":"Return new tensor given shape type, filled ones.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_ones.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Return a new tensor of given shape and type, filled with ones. — op_ones","text":"","code":"op_ones(shape, dtype = NULL)"},{"path":"https://keras3.posit.co/dev/reference/op_ones.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Return a new tensor of given shape and type, filled with ones. — op_ones","text":"shape Shape new tensor. dtype Desired data type tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_ones.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Return a new tensor of given shape and type, filled with ones. — op_ones","text":"Tensor ones given shape dtype.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_ones_like.html","id":null,"dir":"Reference","previous_headings":"","what":"Return a tensor of ones with the same shape and type of x. — op_ones_like","title":"Return a tensor of ones with the same shape and type of x. — op_ones_like","text":"Return tensor ones shape type x.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_ones_like.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Return a tensor of ones with the same shape and type of x. — op_ones_like","text":"","code":"op_ones_like(x, dtype = NULL)"},{"path":"https://keras3.posit.co/dev/reference/op_ones_like.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Return a tensor of ones with the same shape and type of x. — op_ones_like","text":"x Input tensor. dtype Overrides data type result.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_ones_like.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Return a tensor of ones with the same shape and type of x. — op_ones_like","text":"tensor ones shape type x.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_outer.html","id":null,"dir":"Reference","previous_headings":"","what":"Compute the outer product of two vectors. — op_outer","title":"Compute the outer product of two vectors. — op_outer","text":"Given two vectors x1 x2, outer product :","code":"out[i, j] = x1[i] * x2[j]"},{"path":"https://keras3.posit.co/dev/reference/op_outer.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Compute the outer product of two vectors. — op_outer","text":"","code":"op_outer(x1, x2)"},{"path":"https://keras3.posit.co/dev/reference/op_outer.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Compute the outer product of two vectors. — op_outer","text":"x1 First input tensor. x2 Second input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_outer.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Compute the outer product of two vectors. — op_outer","text":"Outer product x1 x2.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_pad.html","id":null,"dir":"Reference","previous_headings":"","what":"Pad a tensor. — op_pad","title":"Pad a tensor. — op_pad","text":"Pad tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_pad.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Pad a tensor. — op_pad","text":"","code":"op_pad(x, pad_width, mode = \"constant\", constant_values = NULL)"},{"path":"https://keras3.posit.co/dev/reference/op_pad.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Pad a tensor. — op_pad","text":"x Tensor pad. pad_width Number values padded edges axis. ((before_1, after_1), ...(before_N, after_N)) unique pad widths axis. ((, ),) yields pad axis. (pad,) int shortcut = = pad width axes. mode One \"constant\", \"edge\", \"linear_ramp\", \"maximum\", \"mean\", \"median\", \"minimum\", \"reflect\", \"symmetric\", \"wrap\", \"empty\", \"circular\". Defaults \"constant\". constant_values Value pad mode == \"constant\". Defaults 0. ValueError raised NULL mode != \"constant\".","code":""},{"path":"https://keras3.posit.co/dev/reference/op_pad.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Pad a tensor. — op_pad","text":"Padded tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_pad.html","id":"note","dir":"Reference","previous_headings":"","what":"Note","title":"Pad a tensor. — op_pad","text":"Torch backend supports modes \"constant\", \"reflect\", \"symmetric\" \"circular\". Torch backend supports \"circular\" mode. Note: Tensorflow backend supports modes \"constant\", \"reflect\" \"symmetric\".","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_polar.html","id":null,"dir":"Reference","previous_headings":"","what":"Constructs a complex tensor whose elements are Cartesian — op_polar","title":"Constructs a complex tensor whose elements are Cartesian — op_polar","text":"coordinates corresponding polar coordinates absolute value abs angle angle. operation numerically equivalent torch.polar(). equivalent scipy.lingalg.polar() performs Singular Value Decomposition. Given magnitude (abs_) angle (angle), function computes corresponding complex number form real + imaginary * 1i, : real = abs_ * cos(angle) imaginary = abs_ * sin(angle)","code":""},{"path":"https://keras3.posit.co/dev/reference/op_polar.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Constructs a complex tensor whose elements are Cartesian — op_polar","text":"","code":"op_polar(abs_, angle)"},{"path":"https://keras3.posit.co/dev/reference/op_polar.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Constructs a complex tensor whose elements are Cartesian — op_polar","text":"abs_ magnitude (absolute value) complex number. angle angle (radians) complex number.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_polar.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Constructs a complex tensor whose elements are Cartesian — op_polar","text":"complex number (array complex numbers) shape abs_ angle.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_polar.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Constructs a complex tensor whose elements are Cartesian — op_polar","text":"","code":"abs_ <- random_normal(c(1, 2)) angle <- random_normal(c(1, 2)) op_shape(op_polar(abs_, angle)) ## shape(1, 2) op_polar(abs_, angle) ## tf.Tensor([[-0.10252866-1.2460487j -0.2586355 -1.0352845j]], shape=(1, 2), dtype=complex64)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_power.html","id":null,"dir":"Reference","previous_headings":"","what":"First tensor elements raised to powers from second tensor, element-wise. — op_power","title":"First tensor elements raised to powers from second tensor, element-wise. — op_power","text":"Note function automatically called using R operator ^ tensor.","code":"(x <- op_arange(4)) ## tf.Tensor([1. 2. 3. 4.], shape=(4), dtype=float32) op_power(2, x) ## tf.Tensor([ 2.  4.  8. 16.], shape=(4), dtype=float32) 2 ^ x ## tf.Tensor([ 2.  4.  8. 16.], shape=(4), dtype=float32)"},{"path":"https://keras3.posit.co/dev/reference/op_power.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"First tensor elements raised to powers from second tensor, element-wise. — op_power","text":"","code":"op_power(x1, x2)"},{"path":"https://keras3.posit.co/dev/reference/op_power.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"First tensor elements raised to powers from second tensor, element-wise. — op_power","text":"x1 bases. x2 exponents.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_power.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"First tensor elements raised to powers from second tensor, element-wise. — op_power","text":"Output tensor, bases x1 raised exponents x2.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_prod.html","id":null,"dir":"Reference","previous_headings":"","what":"Return the product of tensor elements over a given axis. — op_prod","title":"Return the product of tensor elements over a given axis. — op_prod","text":"Return product tensor elements given axis.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_prod.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Return the product of tensor elements over a given axis. — op_prod","text":"","code":"op_prod(x, axis = NULL, keepdims = FALSE, dtype = NULL)"},{"path":"https://keras3.posit.co/dev/reference/op_prod.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Return the product of tensor elements over a given axis. — op_prod","text":"x Input tensor. axis Axis axes along product performed. default, axis = NULL, compute product elements input tensor. keepdims set TRUE, axes reduce left result dimensions size one. dtype Data type returned tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_prod.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Return the product of tensor elements over a given axis. — op_prod","text":"Product elements x given axis axes.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_psnr.html","id":null,"dir":"Reference","previous_headings":"","what":"Peak Signal-to-Noise Ratio (PSNR) function. — op_psnr","title":"Peak Signal-to-Noise Ratio (PSNR) function. — op_psnr","text":"function computes Peak Signal--Noise Ratio two signals, x1 x2. PSNR measure quality reconstructed signal. higher PSNR, closer reconstructed signal original signal. Note can become negative signal power smaller noise power.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_psnr.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Peak Signal-to-Noise Ratio (PSNR) function. — op_psnr","text":"","code":"op_psnr(x1, x2, max_val)"},{"path":"https://keras3.posit.co/dev/reference/op_psnr.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Peak Signal-to-Noise Ratio (PSNR) function. — op_psnr","text":"x1 first input signal. x2 second input signal. Must shape x1. max_val maximum possible value signals.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_psnr.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Peak Signal-to-Noise Ratio (PSNR) function. — op_psnr","text":"float: PSNR value x1 x2.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_psnr.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Peak Signal-to-Noise Ratio (PSNR) function. — op_psnr","text":"","code":"x1 <- random_normal(c(2, 4, 4, 3)) x2 <- random_normal(c(2, 4, 4, 3)) max_val <- 1.0 op_psnr(x1, x2, max_val) ## tf.Tensor(-3.5293808, shape=(), dtype=float32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_qr.html","id":null,"dir":"Reference","previous_headings":"","what":"Computes the QR decomposition of a tensor. — op_qr","title":"Computes the QR decomposition of a tensor. — op_qr","text":"Computes QR decomposition tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_qr.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes the QR decomposition of a tensor. — op_qr","text":"","code":"op_qr(x, mode = \"reduced\")"},{"path":"https://keras3.posit.co/dev/reference/op_qr.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Computes the QR decomposition of a tensor. — op_qr","text":"x Input tensor shape (..., M, N). mode string specifying mode QR decomposition. 'reduced': Returns reduced QR decomposition. (default) 'complete': Returns complete QR decomposition.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_qr.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Computes the QR decomposition of a tensor. — op_qr","text":"list containing two tensors. first tensor shape (..., M, K) orthogonal matrix q second tensor shape (..., K, N)upper triangular matrixr, K = min(M, N)`.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_qr.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Computes the QR decomposition of a tensor. — op_qr","text":"","code":"x <- op_convert_to_tensor(rbind(c(1, 2), c(3, 4), c(5, 6))) op_qr(x) ## $q ## tf.Tensor( ## [[-0.16903085  0.89708523] ##  [-0.50709255  0.27602622] ##  [-0.84515425 -0.34503278]], shape=(3, 2), dtype=float64) ## ## $r ## tf.Tensor( ## [[-5.91607978 -7.43735744] ##  [ 0.          0.82807867]], shape=(2, 2), dtype=float64) c(q, r) %<-% op_qr(x)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_quantile.html","id":null,"dir":"Reference","previous_headings":"","what":"Compute the q-th quantile(s) of the data along the specified axis. — op_quantile","title":"Compute the q-th quantile(s) of the data along the specified axis. — op_quantile","text":"Compute q-th quantile(s) data along specified axis.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_quantile.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Compute the q-th quantile(s) of the data along the specified axis. — op_quantile","text":"","code":"op_quantile(x, q, axis = NULL, method = \"linear\", keepdims = FALSE)"},{"path":"https://keras3.posit.co/dev/reference/op_quantile.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Compute the q-th quantile(s) of the data along the specified axis. — op_quantile","text":"x Input tensor. q Probability sequence probabilities quantiles compute. Values must 0 1 inclusive. axis Axis axes along quantiles computed. Defaults axis=NULL compute quantile(s) along flattened version array. method string specifies method use estimating quantile. Available methods \"linear\", \"lower\", \"higher\", \"midpoint\", \"nearest\". Defaults \"linear\". desired quantile lies two data points < j: \"linear\": + (j - ) * fraction, fraction fractional part index surrounded j. \"lower\": . \"higher\": j. \"midpoint\": (+ j) / 2 \"nearest\": j, whichever nearest. keepdims set TRUE, axes reduce left result dimensions size one.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_quantile.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Compute the q-th quantile(s) of the data along the specified axis. — op_quantile","text":"quantile(s). q single probability axis = NULL, result scalar. multiple probabilities levels given, first axis result corresponds quantiles. axes axes remain reduction x.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_ravel.html","id":null,"dir":"Reference","previous_headings":"","what":"Return a contiguous flattened tensor. — op_ravel","title":"Return a contiguous flattened tensor. — op_ravel","text":"1-D tensor, containing elements input, returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_ravel.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Return a contiguous flattened tensor. — op_ravel","text":"","code":"op_ravel(x)"},{"path":"https://keras3.posit.co/dev/reference/op_ravel.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Return a contiguous flattened tensor. — op_ravel","text":"x Input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_ravel.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Return a contiguous flattened tensor. — op_ravel","text":"Output tensor.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_real.html","id":null,"dir":"Reference","previous_headings":"","what":"Return the real part of the complex argument. — op_real","title":"Return the real part of the complex argument. — op_real","text":"Return real part complex argument.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_real.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Return the real part of the complex argument. — op_real","text":"","code":"op_real(x)"},{"path":"https://keras3.posit.co/dev/reference/op_real.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Return the real part of the complex argument. — op_real","text":"x Input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_real.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Return the real part of the complex argument. — op_real","text":"real component complex argument.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_rearrange.html","id":null,"dir":"Reference","previous_headings":"","what":"Rearranges the axes of a Keras tensor according to a specified pattern, — op_rearrange","title":"Rearranges the axes of a Keras tensor according to a specified pattern, — op_rearrange","text":"einops-style.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_rearrange.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Rearranges the axes of a Keras tensor according to a specified pattern, — op_rearrange","text":"","code":"op_rearrange(tensor, pattern, ...)"},{"path":"https://keras3.posit.co/dev/reference/op_rearrange.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Rearranges the axes of a Keras tensor according to a specified pattern, — op_rearrange","text":"tensor Input Keras tensor. pattern String describing rearrangement einops notation. ... axes_lengths, named arguments specifying lengths axes axes decomposition used.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_rearrange.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Rearranges the axes of a Keras tensor according to a specified pattern, — op_rearrange","text":"Tensor: Keras tensor rearranged axes. Follows logic : decomposition needed, reshape match decomposed dimensions. Permute known inferred axes match form output. Reshape match desired output shape. Example Usage:","code":"images <- op_ones(c(32, 30, 40, 3)) # BHWC format  # Reordering to BCHW op_rearrange(images, 'b h w c -> b c h w') |> op_shape() ## shape(32, 3, 30, 40) # \"Merge\" along first axis - concat images from a batch op_rearrange(images, 'b h w c -> (b h) w c') |> op_shape() ## shape(960, 40, 3) # \"Merge\" along second axis - concat images horizontally op_rearrange(images, 'b h w c -> h (b w) c') |> op_shape() ## shape(30, 1280, 3) # Flatten images into a CHW vector op_rearrange(images, 'b h w c -> b (c h w)') |> op_shape() ## shape(32, 3600) # Decompose H and W axes into 4 smaller patches op_rearrange(images, 'b (h1 h) (w1 w) c -> (b h1 w1) h w c', h1 = 2, w1 = 2) |> op_shape() ## shape(128, 15, 20, 3) # Space-to-depth decomposition of input axes op_rearrange(images, 'b (h h1) (w w1) c -> b h w (c h1 w1)', h1 = 2, w1 = 2) |> op_shape() ## shape(32, 15, 20, 12)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_reciprocal.html","id":null,"dir":"Reference","previous_headings":"","what":"Return the reciprocal of the argument, element-wise. — op_reciprocal","title":"Return the reciprocal of the argument, element-wise. — op_reciprocal","text":"Calculates 1/x.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_reciprocal.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Return the reciprocal of the argument, element-wise. — op_reciprocal","text":"","code":"op_reciprocal(x)"},{"path":"https://keras3.posit.co/dev/reference/op_reciprocal.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Return the reciprocal of the argument, element-wise. — op_reciprocal","text":"x Input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_reciprocal.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Return the reciprocal of the argument, element-wise. — op_reciprocal","text":"Output tensor, element-wise reciprocal x.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_relu.html","id":null,"dir":"Reference","previous_headings":"","what":"Rectified linear unit activation function. — op_relu","title":"Rectified linear unit activation function. — op_relu","text":"defined f(x) = max(0, x).","code":""},{"path":"https://keras3.posit.co/dev/reference/op_relu.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Rectified linear unit activation function. — op_relu","text":"","code":"op_relu(x)"},{"path":"https://keras3.posit.co/dev/reference/op_relu.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Rectified linear unit activation function. — op_relu","text":"x Input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_relu.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Rectified linear unit activation function. — op_relu","text":"tensor shape x.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_relu.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Rectified linear unit activation function. — op_relu","text":"","code":"x1 <- op_convert_to_tensor(c(-1, 0, 1, 0.2)) op_relu(x1) ## tf.Tensor([0.  0.  1.  0.2], shape=(4), dtype=float64) x <- seq(-10, 10, .1) plot(x, op_relu(x))"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_relu6.html","id":null,"dir":"Reference","previous_headings":"","what":"Rectified linear unit activation function with upper bound of 6. — op_relu6","title":"Rectified linear unit activation function with upper bound of 6. — op_relu6","text":"defined f(x) = op_clip(x, 0, 6).","code":""},{"path":"https://keras3.posit.co/dev/reference/op_relu6.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Rectified linear unit activation function with upper bound of 6. — op_relu6","text":"","code":"op_relu6(x)"},{"path":"https://keras3.posit.co/dev/reference/op_relu6.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Rectified linear unit activation function with upper bound of 6. — op_relu6","text":"x Input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_relu6.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Rectified linear unit activation function with upper bound of 6. — op_relu6","text":"tensor shape x.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_relu6.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Rectified linear unit activation function with upper bound of 6. — op_relu6","text":"","code":"x <- op_convert_to_tensor(c(-3, -2, 0.1, 0.2, 6, 8)) op_relu6(x) ## tf.Tensor([0.  0.  0.1 0.2 6.  6. ], shape=(6), dtype=float64) x <- seq(-10, 10, .1) plot(x, op_relu6(x))"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_repeat.html","id":null,"dir":"Reference","previous_headings":"","what":"Repeat each element of a tensor after themselves. — op_repeat","title":"Repeat each element of a tensor after themselves. — op_repeat","text":"Repeat element tensor .","code":""},{"path":"https://keras3.posit.co/dev/reference/op_repeat.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Repeat each element of a tensor after themselves. — op_repeat","text":"","code":"op_repeat(x, repeats, axis = NULL)"},{"path":"https://keras3.posit.co/dev/reference/op_repeat.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Repeat each element of a tensor after themselves. — op_repeat","text":"x Input tensor. repeats number repetitions element. axis axis along repeat values. default, use flattened input array, return flat output array.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_repeat.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Repeat each element of a tensor after themselves. — op_repeat","text":"Output tensor.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_reshape.html","id":null,"dir":"Reference","previous_headings":"","what":"Gives a new shape to a tensor without changing its data. — op_reshape","title":"Gives a new shape to a tensor without changing its data. — op_reshape","text":"Gives new shape tensor without changing data.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_reshape.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Gives a new shape to a tensor without changing its data. — op_reshape","text":"","code":"op_reshape(x, newshape)"},{"path":"https://keras3.posit.co/dev/reference/op_reshape.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Gives a new shape to a tensor without changing its data. — op_reshape","text":"x Input tensor. newshape new shape compatible original shape. One shape dimension can -1 case value inferred length array remaining dimensions.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_reshape.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Gives a new shape to a tensor without changing its data. — op_reshape","text":"reshaped tensor.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_rfft.html","id":null,"dir":"Reference","previous_headings":"","what":"Real-valued Fast Fourier Transform along the last axis of the input. — op_rfft","title":"Real-valued Fast Fourier Transform along the last axis of the input. — op_rfft","text":"Computes 1D Discrete Fourier Transform real-valued signal inner-dimension input. Since Discrete Fourier Transform real-valued signal Hermitian-symmetric, RFFT returns fft_length / 2 + 1 unique components FFT: zero-frequency term, followed fft_length / 2 positive-frequency terms. Along axis RFFT computed , fft_length smaller corresponding dimension input, dimension cropped. larger, dimension padded zeros.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_rfft.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Real-valued Fast Fourier Transform along the last axis of the input. — op_rfft","text":"","code":"op_rfft(x, fft_length = NULL)"},{"path":"https://keras3.posit.co/dev/reference/op_rfft.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Real-valued Fast Fourier Transform along the last axis of the input. — op_rfft","text":"x Input tensor. fft_length integer representing number fft length. specified, inferred length last axis x. Defaults NULL.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_rfft.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Real-valued Fast Fourier Transform along the last axis of the input. — op_rfft","text":"list containing two tensors - real imaginary parts output.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_rfft.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Real-valued Fast Fourier Transform along the last axis of the input. — op_rfft","text":"","code":"x <- op_convert_to_tensor(c(0, 1, 2, 3, 4)) op_rfft(x) ## [[1]] ## tf.Tensor([10.  -2.5 -2.5], shape=(3), dtype=float64) ## ## [[2]] ## tf.Tensor([0.         3.4409548  0.81229924], shape=(3), dtype=float64) op_rfft(x, 3) ## [[1]] ## tf.Tensor([ 3.  -1.5], shape=(2), dtype=float64) ## ## [[2]] ## tf.Tensor([0.        0.8660254], shape=(2), dtype=float64)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_right_shift.html","id":null,"dir":"Reference","previous_headings":"","what":"Shift the bits of an integer to the right. — op_right_shift","title":"Shift the bits of an integer to the right. — op_right_shift","text":"Bits shifted right y. internal representation numbers binary format, operation equivalent dividing x 2^y.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_right_shift.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Shift the bits of an integer to the right. — op_right_shift","text":"","code":"op_right_shift(x, y)"},{"path":"https://keras3.posit.co/dev/reference/op_right_shift.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Shift the bits of an integer to the right. — op_right_shift","text":"x Input integer tensor. y Input integer tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_right_shift.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Shift the bits of an integer to the right. — op_right_shift","text":"Result tensor.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_rms_normalization.html","id":null,"dir":"Reference","previous_headings":"","what":"Performs Root Mean Square (RMS) normalization on x. — op_rms_normalization","title":"Performs Root Mean Square (RMS) normalization on x. — op_rms_normalization","text":"Keras operation implements operation described Root Mean Square Layer Normalization Biao Zhang et al. operation different LayerNormalization RMS scaling. defined rms_normalization(x) = x * rsqrt(mean(square(x))) * scale","code":""},{"path":"https://keras3.posit.co/dev/reference/op_rms_normalization.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Performs Root Mean Square (RMS) normalization on x. — op_rms_normalization","text":"","code":"op_rms_normalization(x, scale = NULL, axis = -1L, epsilon = NULL)"},{"path":"https://keras3.posit.co/dev/reference/op_rms_normalization.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Performs Root Mean Square (RMS) normalization on x. — op_rms_normalization","text":"x Input tensor. scale Optional scaling factor normalization. axis axis axes along perform normalization. Defaults -1. epsilon lower bound value norm. Defaults config_epsilon().","code":""},{"path":"https://keras3.posit.co/dev/reference/op_rms_normalization.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Performs Root Mean Square (RMS) normalization on x. — op_rms_normalization","text":"normalized array.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_rms_normalization.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Performs Root Mean Square (RMS) normalization on x. — op_rms_normalization","text":"","code":"x <- random_normal(c(1, 10)) op_rms_normalization(x) op_rms_normalization(x, scale = 10)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_roll.html","id":null,"dir":"Reference","previous_headings":"","what":"Roll tensor elements along a given axis. — op_roll","title":"Roll tensor elements along a given axis. — op_roll","text":"Elements roll beyond last position re-introduced first.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_roll.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Roll tensor elements along a given axis. — op_roll","text":"","code":"op_roll(x, shift, axis = NULL)"},{"path":"https://keras3.posit.co/dev/reference/op_roll.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Roll tensor elements along a given axis. — op_roll","text":"x Input tensor. shift number places elements shifted. axis axis along elements shifted. default, array flattened shifting, original shape restored.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_roll.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Roll tensor elements along a given axis. — op_roll","text":"Output tensor.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_rot90.html","id":null,"dir":"Reference","previous_headings":"","what":"Rotate an array by 90 degrees in the plane specified by axes. — op_rot90","title":"Rotate an array by 90 degrees in the plane specified by axes. — op_rot90","text":"function rotates array counterclockwise 90 degrees k times plane specified axes. Supports arrays two dimensions.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_rot90.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Rotate an array by 90 degrees in the plane specified by axes. — op_rot90","text":"","code":"op_rot90(array, k = 1L, axes = list(1L, 2L))"},{"path":"https://keras3.posit.co/dev/reference/op_rot90.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Rotate an array by 90 degrees in the plane specified by axes. — op_rot90","text":"array Input array rotate. k Number times array rotated 90 degrees. axes tuple two integers specifying plane rotation (defaults (1, 2)).","code":""},{"path":"https://keras3.posit.co/dev/reference/op_rot90.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Rotate an array by 90 degrees in the plane specified by axes. — op_rot90","text":"Rotated array.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_rot90.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Rotate an array by 90 degrees in the plane specified by axes. — op_rot90","text":"","code":"m <- 1:4 |> op_reshape(c(2, 2)) m ## tf.Tensor( ## [[1 2] ##  [3 4]], shape=(2, 2), dtype=int32) op_rot90(m) ## tf.Tensor( ## [[2 4] ##  [1 3]], shape=(2, 2), dtype=int32) m <- 1:8 |> op_reshape(c(2, 2, 2)) m ## tf.Tensor( ## [[[1 2] ##   [3 4]] ## ##  [[5 6] ##   [7 8]]], shape=(2, 2, 2), dtype=int32) op_rot90(m, k = 1, axes = c(2, 3)) ## tf.Tensor( ## [[[2 4] ##   [1 3]] ## ##  [[6 8] ##   [5 7]]], shape=(2, 2, 2), dtype=int32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_round.html","id":null,"dir":"Reference","previous_headings":"","what":"Evenly round to the given number of decimals. — op_round","title":"Evenly round to the given number of decimals. — op_round","text":"Evenly round given number decimals.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_round.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Evenly round to the given number of decimals. — op_round","text":"","code":"op_round(x, decimals = 0L)"},{"path":"https://keras3.posit.co/dev/reference/op_round.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Evenly round to the given number of decimals. — op_round","text":"x Input tensor. decimals Number decimal places round . Defaults 0.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_round.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Evenly round to the given number of decimals. — op_round","text":"Output tensor.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_rsqrt.html","id":null,"dir":"Reference","previous_headings":"","what":"Computes reciprocal of square root of x element-wise. — op_rsqrt","title":"Computes reciprocal of square root of x element-wise. — op_rsqrt","text":"Computes reciprocal square root x element-wise.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_rsqrt.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes reciprocal of square root of x element-wise. — op_rsqrt","text":"","code":"op_rsqrt(x)"},{"path":"https://keras3.posit.co/dev/reference/op_rsqrt.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Computes reciprocal of square root of x element-wise. — op_rsqrt","text":"x input tensor","code":""},{"path":"https://keras3.posit.co/dev/reference/op_rsqrt.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Computes reciprocal of square root of x element-wise. — op_rsqrt","text":"tensor dtype x.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_rsqrt.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Computes reciprocal of square root of x element-wise. — op_rsqrt","text":"","code":"x <- op_convert_to_tensor(c(1, 10, 100)) op_rsqrt(x) ## tf.Tensor([1.         0.31622777 0.1       ], shape=(3), dtype=float64) # array([1, 0.31622776, 0.1], dtype=float32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_saturate_cast.html","id":null,"dir":"Reference","previous_headings":"","what":"Performs a safe saturating cast to the desired dtype. — op_saturate_cast","title":"Performs a safe saturating cast to the desired dtype. — op_saturate_cast","text":"Saturating cast prevents data type overflow casting dtype smaller values range. E.g. op_cast(c(-1, 256), \"float32\") |> op_cast(\"uint8\") returns c(255, 0), op_cast(c(-1, 256), \"float32\") |> op_saturate_cast(\"uint8\") returns c(0, 255).","code":""},{"path":"https://keras3.posit.co/dev/reference/op_saturate_cast.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Performs a safe saturating cast to the desired dtype. — op_saturate_cast","text":"","code":"op_saturate_cast(x, dtype)"},{"path":"https://keras3.posit.co/dev/reference/op_saturate_cast.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Performs a safe saturating cast to the desired dtype. — op_saturate_cast","text":"x tensor variable. dtype target type.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_saturate_cast.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Performs a safe saturating cast to the desired dtype. — op_saturate_cast","text":"safely casted tensor specified dtype.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_saturate_cast.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Performs a safe saturating cast to the desired dtype. — op_saturate_cast","text":"Image resizing bicubic interpolation may produce values outside original range.     Casting resized image back uint8 cause overflow.     Saturate casting uint8 clip values uint8 range casting cause overflow.","code":"image2x2 <- np_array(as.integer(c(0, 1, 254, 255)), \"uint8\") |>   array_reshape(c(1, 2, 2, 1)) image4x4 <- image2x2 |>   tensorflow::tf$image$resize(shape(4, 4), method=\"bicubic\") image4x4 |> as.array() |> drop() ##           [,1]      [,2]      [,3]      [,4] ## [1,] -22.50000 -22.20462 -21.61891 -21.32353 ## [2,]  52.52605  52.82143  53.40715  53.70253 ## [3,] 201.29752 201.59288 202.17859 202.47395 ## [4,] 276.32355 276.61893 277.20465 277.50006 image4x4_casted <- op_cast(image4x4, \"uint8\") image4x4_casted |> as.array() |> drop() ##      [,1] [,2] [,3] [,4] ## [1,]  234  234  235  235 ## [2,]   52   52   53   53 ## [3,]  201  201  202  202 ## [4,]   20   20   21   21 image4x4_saturate_casted <- image4x4 |> op_saturate_cast(\"uint8\") image4x4_saturate_casted |> as.array() |> drop() ##      [,1] [,2] [,3] [,4] ## [1,]    0    0    0    0 ## [2,]   52   52   53   53 ## [3,]  201  201  202  202 ## [4,]  255  255  255  255"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_scan.html","id":null,"dir":"Reference","previous_headings":"","what":"Scan a function over leading array axes while carrying along state. — op_scan","title":"Scan a function over leading array axes while carrying along state. — op_scan","text":"type xs array type NULL, type ys array type, semantics op_scan() given roughly implementation:   loop-carried value carry (init) must hold fixed shape dtype across iterations. TensorFlow, y must match carry shape dtype. required backends.","code":"op_scan <- function(f, init, xs = NULL, length = NULL) {   xs <- xs %||% vector(\"list\", length)   if(!is.list(xs))     xs <- op_unstack(xs)   ys <- vector(\"list\", length(xs))   carry <- init   for (i in seq_along(xs)) {     c(carry, y) %<-% f(carry, xs[[i]])     ys[[i]] <- y   }   list(carry, op_stack(ys)) }"},{"path":"https://keras3.posit.co/dev/reference/op_scan.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Scan a function over leading array axes while carrying along state. — op_scan","text":"","code":"op_scan(f, init, xs = NULL, length = NULL, reverse = FALSE, unroll = 1L)"},{"path":"https://keras3.posit.co/dev/reference/op_scan.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Scan a function over leading array axes while carrying along state. — op_scan","text":"f Callable defines logic loop iteration. accepts two arguments first value loop carry second slice xs along leading axis. callable returns pair first represents new value loop carry second represents slice output. init initial loop carry value. can scalar, tensor, nested structure. must match structure first element returned f. xs Optional value scan along leading axis. can tensor nested structure. xs provided, must specify length define number loop iterations. Defaults NULL. length Optional integer specifying number loop iterations. length provided, defaults sizes leading axis arrays xs. Defaults NULL. reverse Optional boolean specifying whether run scan iteration forward reverse, equivalent reversing leading axes arrays xs ys. unroll Optional positive integer boolean specifying many scan iterations unroll within single iteration loop. integer provided, determines many unrolled loop iterations run within single rolled iteration loop. boolean provided, determine loop completely unrolled (unroll=TRUE) left completely unrolled (unroll=FALSE). Note unrolling supported JAX TensorFlow backends.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_scan.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Scan a function over leading array axes while carrying along state. — op_scan","text":"pair first element represents final loop carry value second element represents stacked outputs f scanned leading axis inputs.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_scan.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Scan a function over leading array axes while carrying along state. — op_scan","text":"","code":"sum_fn <- function(c, x) list(c + x, c + x) init <- op_array(0L) xs <- op_array(1:5) c(carry, result) %<-% op_scan(sum_fn, init, xs) carry ## tf.Tensor(15, shape=(), dtype=int32) result ## tf.Tensor([ 1  3  6 10 15], shape=(5), dtype=int32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_scatter.html","id":null,"dir":"Reference","previous_headings":"","what":"Returns a tensor of shape shape where indices are set to values. — op_scatter","title":"Returns a tensor of shape shape where indices are set to values. — op_scatter","text":"high level, operation zeros[indices] = updates returns output. equivalent :","code":"output <- op_scatter_update(op_zeros(shape), indices, values)"},{"path":"https://keras3.posit.co/dev/reference/op_scatter.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Returns a tensor of shape shape where indices are set to values. — op_scatter","text":"","code":"op_scatter(indices, values, shape)"},{"path":"https://keras3.posit.co/dev/reference/op_scatter.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Returns a tensor of shape shape where indices are set to values. — op_scatter","text":"indices tensor list specifying indices values values. values tensor, values set indices. shape Shape output tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_scatter.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Returns a tensor of shape shape where indices are set to values. — op_scatter","text":"tensor shape shape indices set values.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_scatter.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Returns a tensor of shape shape where indices are set to values. — op_scatter","text":"","code":"op_scatter(indices = rbind(1), values = 1, shape = c(5)) ## tf.Tensor([1. 0. 0. 0. 0.], shape=(5), dtype=float64) op_scatter(indices = rbind(1, 3), values = 1:2, shape = c(5)) ## tf.Tensor([1 0 2 0 0], shape=(5), dtype=int32) indices <- rbind(c(1, 2), c(2, 2)) values <- op_array(c(1, 1)) op_scatter(indices, values, shape = c(2, 2)) ## tf.Tensor( ## [[0. 1.] ##  [0. 1.]], shape=(2, 2), dtype=float32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_scatter_update.html","id":null,"dir":"Reference","previous_headings":"","what":"Update inputs via updates at scattered (sparse) indices. — op_scatter_update","title":"Update inputs via updates at scattered (sparse) indices. — op_scatter_update","text":"high level, operation inputs[indices] <- updates. Assume inputs tensor shape (D1, D2, ..., Dn), 2 main usages scatter_update. indices 2D tensor shape (num_updates, n), num_updates number updates perform, updates 1D tensor shape (num_updates). example, inputs op_zeros(c(4, 4, 4)), want update inputs[2, 3, 4] inputs[1, 2, 4] 1, can use:     2 indices 2D tensor shape (num_updates, k), num_updates number updates perform, k (k <= n) size index indices. updates n - k-D tensor shape (num_updates, shape(inputs)[-(1:k)]). example, inputs <- op_zeros(c(4, 4, 4)), want update inputs[1, 2, ] inputs[2, 3, ] [1, 1, 1, 1], indices shape (num_updates, 2) (k = 2), updates shape (num_updates, 4) (shape(inputs)[3:4] == 4). See code :","code":"inputs <- op_zeros(c(4, 4, 4)) indices <- rbind(c(2, 3, 4),                  c(1, 2, 4)) updates <- op_array(c(1, 1), \"float32\") op_scatter_update(inputs, indices, updates) ## tf.Tensor( ## [[[0. 0. 0. 0.] ##   [0. 0. 0. 1.] ##   [0. 0. 0. 0.] ##   [0. 0. 0. 0.]] ## ##  [[0. 0. 0. 0.] ##   [0. 0. 0. 0.] ##   [0. 0. 0. 1.] ##   [0. 0. 0. 0.]] ## ##  [[0. 0. 0. 0.] ##   [0. 0. 0. 0.] ##   [0. 0. 0. 0.] ##   [0. 0. 0. 0.]] ## ##  [[0. 0. 0. 0.] ##   [0. 0. 0. 0.] ##   [0. 0. 0. 0.] ##   [0. 0. 0. 0.]]], shape=(4, 4, 4), dtype=float32) inputs <- op_zeros(c(4, 4, 4)) indices <- rbind(c(2, 3),                  c(3, 4)) updates <- op_array(rbind(c(1, 1, 1, 1),                           c(1, 1, 1, 1)),                     \"float32\") op_scatter_update(inputs, indices, updates) ## tf.Tensor( ## [[[0. 0. 0. 0.] ##   [0. 0. 0. 0.] ##   [0. 0. 0. 0.] ##   [0. 0. 0. 0.]] ## ##  [[0. 0. 0. 0.] ##   [0. 0. 0. 0.] ##   [1. 1. 1. 1.] ##   [0. 0. 0. 0.]] ## ##  [[0. 0. 0. 0.] ##   [0. 0. 0. 0.] ##   [0. 0. 0. 0.] ##   [1. 1. 1. 1.]] ## ##  [[0. 0. 0. 0.] ##   [0. 0. 0. 0.] ##   [0. 0. 0. 0.] ##   [0. 0. 0. 0.]]], shape=(4, 4, 4), dtype=float32)"},{"path":"https://keras3.posit.co/dev/reference/op_scatter_update.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Update inputs via updates at scattered (sparse) indices. — op_scatter_update","text":"","code":"op_scatter_update(inputs, indices, updates)"},{"path":"https://keras3.posit.co/dev/reference/op_scatter_update.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Update inputs via updates at scattered (sparse) indices. — op_scatter_update","text":"inputs tensor, tensor updated. indices tensor list shape (N, inputs$ndim), specifying indices update. N number indices update, must equal first dimension updates. updates tensor, new values put inputs indices.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_scatter_update.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Update inputs via updates at scattered (sparse) indices. — op_scatter_update","text":"tensor, shape dtype inputs.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_searchsorted.html","id":null,"dir":"Reference","previous_headings":"","what":"Perform a binary search. — op_searchsorted","title":"Perform a binary search. — op_searchsorted","text":"Return insertion indices keep values sorted order placed sorted_sequence.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_searchsorted.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Perform a binary search. — op_searchsorted","text":"","code":"op_searchsorted(sorted_sequence, values, side = \"left\", zero_indexed = FALSE)"},{"path":"https://keras3.posit.co/dev/reference/op_searchsorted.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Perform a binary search. — op_searchsorted","text":"sorted_sequence 1-D input tensor sorted along innermost dimension. values N-D tensor query insertion values. side Either \"left\" \"right\", choosing side insert ties. zero_indexed TRUE, returned indices zero-based (0 encodes first position); FALSE (default), returned indices one-based (1 encodes first position).","code":""},{"path":"https://keras3.posit.co/dev/reference/op_searchsorted.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Perform a binary search. — op_searchsorted","text":"Tensor insertion indices shape values.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_searchsorted.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Perform a binary search. — op_searchsorted","text":"","code":"sorted <- op_array(c(-1, 0, 2, 4)) queries <- c(-2, 0, 3, 7) op_searchsorted(sorted, queries) ## tf.Tensor([1 2 4 5], shape=(4), dtype=int32) op_searchsorted(sorted, queries, side = \"right\", zero_indexed = TRUE) ## tf.Tensor([0 2 3 4], shape=(4), dtype=int32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_segment_max.html","id":null,"dir":"Reference","previous_headings":"","what":"Computes the max of segments in a tensor. — op_segment_max","title":"Computes the max of segments in a tensor. — op_segment_max","text":"Computes max segments tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_segment_max.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes the max of segments in a tensor. — op_segment_max","text":"","code":"op_segment_max(data, segment_ids, num_segments = NULL, sorted = FALSE)"},{"path":"https://keras3.posit.co/dev/reference/op_segment_max.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Computes the max of segments in a tensor. — op_segment_max","text":"data Input tensor. segment_ids N-D tensor containing segment indices element data. head(op_shape(data), length(op_shape(segment_ids))) match op_shape(segment_ids) num_segments integer representing total number segments. specified, inferred maximum value segment_ids. sorted boolean indicating whether segment_ids sorted. Defaults FALSE.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_segment_max.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Computes the max of segments in a tensor. — op_segment_max","text":"tensor containing max segments, element represents max corresponding segment data.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_segment_max.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Computes the max of segments in a tensor. — op_segment_max","text":"","code":"data <- op_convert_to_tensor(c(1, 2, 10, 20, 100, 200)) segment_ids <- op_array(c(1, 1, 2, 2, 3, 3), \"int32\") num_segments <- 3 op_segment_max(data, segment_ids, num_segments) ## tf.Tensor([  2.  20. 200.], shape=(3), dtype=float64) # array([2, 20, 200], dtype=int32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_segment_sum.html","id":null,"dir":"Reference","previous_headings":"","what":"Computes the sum of segments in a tensor. — op_segment_sum","title":"Computes the sum of segments in a tensor. — op_segment_sum","text":"Computes sum segments tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_segment_sum.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes the sum of segments in a tensor. — op_segment_sum","text":"","code":"op_segment_sum(data, segment_ids, num_segments = NULL, sorted = FALSE)"},{"path":"https://keras3.posit.co/dev/reference/op_segment_sum.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Computes the sum of segments in a tensor. — op_segment_sum","text":"data Input tensor. segment_ids N-D tensor containing segment indices element data. Num dims segment ids strictly smaller equal number dims data. num_segments integer representing total number segments. specified, inferred maximum value segment_ids. sorted boolean indicating whether segment_ids sorted. Defaults FALSE.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_segment_sum.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Computes the sum of segments in a tensor. — op_segment_sum","text":"tensor containing sum segments, element represents sum corresponding segment data.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_segment_sum.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Computes the sum of segments in a tensor. — op_segment_sum","text":"","code":"data <- op_array(c(1, 2, 10, 20, 100, 200)) segment_ids <- op_array(c(1, 1, 2, 2, 3, 3), \"int32\") num_segments <- 3 op_segment_sum(data, segment_ids, num_segments) ## tf.Tensor([  3.  30. 300.], shape=(3), dtype=float32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_select.html","id":null,"dir":"Reference","previous_headings":"","what":"Return elements from choicelist, based on conditions in condlist. — op_select","title":"Return elements from choicelist, based on conditions in condlist. — op_select","text":"Return elements choicelist, based conditions condlist.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_select.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Return elements from choicelist, based on conditions in condlist. — op_select","text":"","code":"op_select(condlist, choicelist, default = 0L)"},{"path":"https://keras3.posit.co/dev/reference/op_select.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Return elements from choicelist, based on conditions in condlist. — op_select","text":"condlist List boolean tensors. list conditions determine array choicelist output elements taken. multiple conditions satisfied, first one encountered condlist used. choicelist List tensors. list tensors output elements taken. list length condlist. default Optional scalar value. element inserted output conditions evaluate FALSE.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_select.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Return elements from choicelist, based on conditions in condlist. — op_select","text":"Tensor output position m m-th element tensor choicelist m-th element corresponding tensor condlist TRUE.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_select.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Return elements from choicelist, based on conditions in condlist. — op_select","text":"","code":"x <- op_arange(6L) condlist <- list(x < 3, x > 3) choicelist <- list(x, x^2) op_select(condlist, choicelist, 42) ## tf.Tensor([ 1  2 42 16 25 36], shape=(6), dtype=int32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_selu.html","id":null,"dir":"Reference","previous_headings":"","what":"Scaled Exponential Linear Unit (SELU) activation function. — op_selu","title":"Scaled Exponential Linear Unit (SELU) activation function. — op_selu","text":"defined : f(x) =  scale * alpha * (exp(x) - 1.) x < 0, f(x) = scale * x x >= 0.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_selu.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Scaled Exponential Linear Unit (SELU) activation function. — op_selu","text":"","code":"op_selu(x)"},{"path":"https://keras3.posit.co/dev/reference/op_selu.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Scaled Exponential Linear Unit (SELU) activation function. — op_selu","text":"x Input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_selu.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Scaled Exponential Linear Unit (SELU) activation function. — op_selu","text":"tensor shape x.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_selu.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Scaled Exponential Linear Unit (SELU) activation function. — op_selu","text":"","code":"x <- op_array(c(-1, 0, 1)) op_selu(x) ## tf.Tensor([-1.1113307  0.         1.050701 ], shape=(3), dtype=float32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_separable_conv.html","id":null,"dir":"Reference","previous_headings":"","what":"General N-D separable convolution. — op_separable_conv","title":"General N-D separable convolution. — op_separable_conv","text":"ops supports 1D 2D separable convolution. separable_conv depthwise conv followed pointwise conv.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_separable_conv.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"General N-D separable convolution. — op_separable_conv","text":"","code":"op_separable_conv(   inputs,   depthwise_kernel,   pointwise_kernel,   strides = 1L,   padding = \"valid\",   data_format = NULL,   dilation_rate = 1L )"},{"path":"https://keras3.posit.co/dev/reference/op_separable_conv.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"General N-D separable convolution. — op_separable_conv","text":"inputs Tensor rank N+2. inputs shape (batch_size,) + inputs_spatial_shape + (num_channels,) data_format=\"channels_last\", (batch_size, num_channels) + inputs_spatial_shape data_format=\"channels_first\". depthwise_kernel Tensor rank N+2. depthwise_kernel shape [kernel_spatial_shape, num_input_channels, num_channels_multiplier], num_input_channels match number channels inputs. pointwise_kernel Tensor rank N+2. pointwise_kernel shape (*ones_like(kernel_spatial_shape), num_input_channels * num_channels_multiplier, num_output_channels). strides int int tuple/list len(inputs_spatial_shape), specifying strides convolution along spatial dimension. strides int, every spatial dimension shares strides. padding string, either \"valid\" \"\". \"valid\" means padding applied, \"\" results padding evenly left/right /input output height/width dimension input strides=1. data_format string, either \"channels_last\" \"channels_first\". data_format determines ordering dimensions inputs. data_format=\"channels_last\", inputs shape (batch_size, ..., channels) data_format=\"channels_first\", inputs shape (batch_size, channels, ...). dilation_rate int int tuple/list len(inputs_spatial_shape), specifying dilation rate use dilated convolution. dilation_rate int, every spatial dimension shares dilation_rate.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_separable_conv.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"General N-D separable convolution. — op_separable_conv","text":"tensor rank N+2, result depthwise conv operation.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_shape.html","id":null,"dir":"Reference","previous_headings":"","what":"Gets the shape of the tensor input. — op_shape","title":"Gets the shape of the tensor input. — op_shape","text":"Gets shape tensor input.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_shape.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Gets the shape of the tensor input. — op_shape","text":"","code":"op_shape(x)"},{"path":"https://keras3.posit.co/dev/reference/op_shape.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Gets the shape of the tensor input. — op_shape","text":"x tensor. function try access shape attribute input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_shape.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Gets the shape of the tensor input. — op_shape","text":"list integers NULL values, indicating shape input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_shape.html","id":"note","dir":"Reference","previous_headings":"","what":"Note","title":"Gets the shape of the tensor input. — op_shape","text":"TensorFlow backend, x tf.Tensor dynamic shape, dimensions dynamic context compiled function tf.Tensor value instead static integer value.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_shape.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Gets the shape of the tensor input. — op_shape","text":"","code":"x <- op_zeros(c(8, 12)) op_shape(x) ## shape(8, 12)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_sigmoid.html","id":null,"dir":"Reference","previous_headings":"","what":"Sigmoid activation function. — op_sigmoid","title":"Sigmoid activation function. — op_sigmoid","text":"defined f(x) = 1 / (1 + exp(-x)).","code":""},{"path":"https://keras3.posit.co/dev/reference/op_sigmoid.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Sigmoid activation function. — op_sigmoid","text":"","code":"op_sigmoid(x)"},{"path":"https://keras3.posit.co/dev/reference/op_sigmoid.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Sigmoid activation function. — op_sigmoid","text":"x Input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_sigmoid.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Sigmoid activation function. — op_sigmoid","text":"tensor shape x.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_sigmoid.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Sigmoid activation function. — op_sigmoid","text":"","code":"x <- op_convert_to_tensor(c(-6, 1, 0, 1, 6)) op_sigmoid(x) ## tf.Tensor([0.00247262 0.73105858 0.5        0.73105858 0.99752738], shape=(5), dtype=float64)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_sign.html","id":null,"dir":"Reference","previous_headings":"","what":"Returns a tensor with the signs of the elements of x. — op_sign","title":"Returns a tensor with the signs of the elements of x. — op_sign","text":"Returns tensor signs elements x.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_sign.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Returns a tensor with the signs of the elements of x. — op_sign","text":"","code":"op_sign(x)"},{"path":"https://keras3.posit.co/dev/reference/op_sign.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Returns a tensor with the signs of the elements of x. — op_sign","text":"x Input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_sign.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Returns a tensor with the signs of the elements of x. — op_sign","text":"Output tensor shape x.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_signbit.html","id":null,"dir":"Reference","previous_headings":"","what":"Return the sign bit of the elements of x. — op_signbit","title":"Return the sign bit of the elements of x. — op_signbit","text":"output boolean tensor contains TRUE sign x negative, FALSE otherwise.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_signbit.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Return the sign bit of the elements of x. — op_signbit","text":"","code":"op_signbit(x)"},{"path":"https://keras3.posit.co/dev/reference/op_signbit.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Return the sign bit of the elements of x. — op_signbit","text":"x Input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_signbit.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Return the sign bit of the elements of x. — op_signbit","text":"Output boolean tensor shape x.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_silu.html","id":null,"dir":"Reference","previous_headings":"","what":"Sigmoid Linear Unit (SiLU) activation function, also known as Swish. — op_silu","title":"Sigmoid Linear Unit (SiLU) activation function, also known as Swish. — op_silu","text":"SiLU activation function computed sigmoid function multiplied input. defined f(x) = x * sigmoid(x).","code":""},{"path":"https://keras3.posit.co/dev/reference/op_silu.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Sigmoid Linear Unit (SiLU) activation function, also known as Swish. — op_silu","text":"","code":"op_silu(x)"},{"path":"https://keras3.posit.co/dev/reference/op_silu.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Sigmoid Linear Unit (SiLU) activation function, also known as Swish. — op_silu","text":"x Input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_silu.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Sigmoid Linear Unit (SiLU) activation function, also known as Swish. — op_silu","text":"tensor shape x.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_silu.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Sigmoid Linear Unit (SiLU) activation function, also known as Swish. — op_silu","text":"","code":"x <- op_convert_to_tensor(c(-6, 1, 0, 1, 6)) op_sigmoid(x) ## tf.Tensor([0.00247262 0.73105858 0.5        0.73105858 0.99752738], shape=(5), dtype=float64) op_silu(x) ## tf.Tensor([-0.01483574  0.73105858  0.          0.73105858  5.98516426], shape=(5), dtype=float64)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_sin.html","id":null,"dir":"Reference","previous_headings":"","what":"Trigonometric sine, element-wise. — op_sin","title":"Trigonometric sine, element-wise. — op_sin","text":"Trigonometric sine, element-wise.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_sin.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Trigonometric sine, element-wise. — op_sin","text":"","code":"op_sin(x)"},{"path":"https://keras3.posit.co/dev/reference/op_sin.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Trigonometric sine, element-wise. — op_sin","text":"x Input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_sin.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Trigonometric sine, element-wise. — op_sin","text":"Output tensor shape x.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_sinh.html","id":null,"dir":"Reference","previous_headings":"","what":"Hyperbolic sine, element-wise. — op_sinh","title":"Hyperbolic sine, element-wise. — op_sinh","text":"Hyperbolic sine, element-wise.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_sinh.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Hyperbolic sine, element-wise. — op_sinh","text":"","code":"op_sinh(x)"},{"path":"https://keras3.posit.co/dev/reference/op_sinh.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Hyperbolic sine, element-wise. — op_sinh","text":"x Input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_sinh.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Hyperbolic sine, element-wise. — op_sinh","text":"Output tensor shape x.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_size.html","id":null,"dir":"Reference","previous_headings":"","what":"Return the number of elements in a tensor. — op_size","title":"Return the number of elements in a tensor. — op_size","text":"Return number elements tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_size.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Return the number of elements in a tensor. — op_size","text":"","code":"op_size(x)"},{"path":"https://keras3.posit.co/dev/reference/op_size.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Return the number of elements in a tensor. — op_size","text":"x Input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_size.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Return the number of elements in a tensor. — op_size","text":"Number elements x.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_slice.html","id":null,"dir":"Reference","previous_headings":"","what":"Return a slice of an input tensor. — op_slice","title":"Return a slice of an input tensor. — op_slice","text":"high level, operation explicit replacement array slicing e.g. inputs[start_indices:(start_indices + shape)]. Unlike slicing via brackets, operation accept tensor start indices backends, useful indices dynamically computed via tensor operations.","code":"(inputs <- op_arange(5*5) |> op_reshape(c(5, 5))) ## tf.Tensor( ## [[ 1.  2.  3.  4.  5.] ##  [ 6.  7.  8.  9. 10.] ##  [11. 12. 13. 14. 15.] ##  [16. 17. 18. 19. 20.] ##  [21. 22. 23. 24. 25.]], shape=(5, 5), dtype=float32) start_indices <- c(3, 3) shape <- c(2, 2) op_slice(inputs, start_indices, shape) ## tf.Tensor( ## [[13. 14.] ##  [18. 19.]], shape=(2, 2), dtype=float32)"},{"path":"https://keras3.posit.co/dev/reference/op_slice.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Return a slice of an input tensor. — op_slice","text":"","code":"op_slice(inputs, start_indices, shape)"},{"path":"https://keras3.posit.co/dev/reference/op_slice.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Return a slice of an input tensor. — op_slice","text":"inputs tensor, tensor sliced. start_indices list length inputs$ndim, specifying starting indices updating. shape full shape returned slice.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_slice.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Return a slice of an input tensor. — op_slice","text":"tensor, shape dtype inputs.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_slice_update.html","id":null,"dir":"Reference","previous_headings":"","what":"Update an input by slicing in a tensor of updated values. — op_slice_update","title":"Update an input by slicing in a tensor of updated values. — op_slice_update","text":"high level, operation inputs[start_indices: start_indices + updates.shape] = updates. Assume inputs tensor shape (D1, D2, ..., Dn), start_indices must list n integers, specifying starting indices. updates must rank inputs, size dim must exceed Di - start_indices[]. example, 2D inputs inputs = op_zeros(c(5, 5)), want update intersection last 2 rows last 2 columns 1, .e., inputs[4:5, 4:5] = op_ones(c(2, 2)), can use code :","code":"inputs <- op_zeros(c(5, 5)) start_indices <- c(3, 3) updates <- op_ones(c(2, 2)) op_slice_update(inputs, start_indices, updates) ## tf.Tensor( ## [[0. 0. 0. 0. 0.] ##  [0. 0. 0. 0. 0.] ##  [0. 0. 1. 1. 0.] ##  [0. 0. 1. 1. 0.] ##  [0. 0. 0. 0. 0.]], shape=(5, 5), dtype=float32)"},{"path":"https://keras3.posit.co/dev/reference/op_slice_update.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Update an input by slicing in a tensor of updated values. — op_slice_update","text":"","code":"op_slice_update(inputs, start_indices, updates)"},{"path":"https://keras3.posit.co/dev/reference/op_slice_update.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Update an input by slicing in a tensor of updated values. — op_slice_update","text":"inputs tensor, tensor updated. start_indices list length inputs$ndim, specifying starting indices updating. updates tensor, new values put inputs indices. updates must rank inputs.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_slice_update.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Update an input by slicing in a tensor of updated values. — op_slice_update","text":"tensor, shape dtype inputs.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_slogdet.html","id":null,"dir":"Reference","previous_headings":"","what":"Compute the sign and natural logarithm of the determinant of a matrix. — op_slogdet","title":"Compute the sign and natural logarithm of the determinant of a matrix. — op_slogdet","text":"Compute sign natural logarithm determinant matrix.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_slogdet.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Compute the sign and natural logarithm of the determinant of a matrix. — op_slogdet","text":"","code":"op_slogdet(x)"},{"path":"https://keras3.posit.co/dev/reference/op_slogdet.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Compute the sign and natural logarithm of the determinant of a matrix. — op_slogdet","text":"x Input matrix. must 2D square.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_slogdet.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Compute the sign and natural logarithm of the determinant of a matrix. — op_slogdet","text":"list: (sign, logabsdet). sign number representing sign determinant. real matrix, 1, 0, -1. complex matrix, complex number absolute value 1 (.e., unit circle), else 0. logabsdet natural log absolute value determinant.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_soft_shrink.html","id":null,"dir":"Reference","previous_headings":"","what":"Soft Shrink activation function. — op_soft_shrink","title":"Soft Shrink activation function. — op_soft_shrink","text":"defined f(x) = x - threshold x > threshold, f(x) = x + threshold x < -threshold, f(x) = 0 otherwise.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_soft_shrink.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Soft Shrink activation function. — op_soft_shrink","text":"","code":"op_soft_shrink(x, threshold = 0.5)"},{"path":"https://keras3.posit.co/dev/reference/op_soft_shrink.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Soft Shrink activation function. — op_soft_shrink","text":"x Input tensor. threshold Threshold value. Defaults 0.5.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_soft_shrink.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Soft Shrink activation function. — op_soft_shrink","text":"tensor shape x.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_soft_shrink.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Soft Shrink activation function. — op_soft_shrink","text":"","code":"x <- op_array(c(-1, 0, 1)) op_soft_shrink(x) ## tf.Tensor([-0.5  0.   0.5], shape=(3), dtype=float32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_softmax.html","id":null,"dir":"Reference","previous_headings":"","what":"Softmax activation function. — op_softmax","title":"Softmax activation function. — op_softmax","text":"elements output vector lie within range (0, 1), total sum exactly 1 (excluding floating point rounding error). vector processed independently. axis argument specifies axis along function applied within input. defined : f(x) = exp(x) / sum(exp(x))","code":""},{"path":"https://keras3.posit.co/dev/reference/op_softmax.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Softmax activation function. — op_softmax","text":"","code":"op_softmax(x, axis = -1L)"},{"path":"https://keras3.posit.co/dev/reference/op_softmax.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Softmax activation function. — op_softmax","text":"x Input tensor. axis Integer, axis along softmax applied.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_softmax.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Softmax activation function. — op_softmax","text":"tensor shape x.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_softmax.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Softmax activation function. — op_softmax","text":"","code":"x <- op_array(c(-1, 0, 1)) op_softmax(x) ## tf.Tensor([0.09003057 0.24472848 0.66524094], shape=(3), dtype=float32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_softplus.html","id":null,"dir":"Reference","previous_headings":"","what":"Softplus activation function. — op_softplus","title":"Softplus activation function. — op_softplus","text":"defined f(x) = log(exp(x) + 1), log natural logarithm exp exponential function.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_softplus.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Softplus activation function. — op_softplus","text":"","code":"op_softplus(x)"},{"path":"https://keras3.posit.co/dev/reference/op_softplus.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Softplus activation function. — op_softplus","text":"x Input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_softplus.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Softplus activation function. — op_softplus","text":"tensor shape x.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_softplus.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Softplus activation function. — op_softplus","text":"","code":"x <- op_convert_to_tensor(c(-0.555, 0, 0.555)) op_softplus(x) ## tf.Tensor([0.45366606 0.69314718 1.00866606], shape=(3), dtype=float64) x <- seq(-10, 10, .1) plot(x, op_softplus(x))"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_softsign.html","id":null,"dir":"Reference","previous_headings":"","what":"Softsign activation function. — op_softsign","title":"Softsign activation function. — op_softsign","text":"defined f(x) = x / (abs(x) + 1).","code":""},{"path":"https://keras3.posit.co/dev/reference/op_softsign.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Softsign activation function. — op_softsign","text":"","code":"op_softsign(x)"},{"path":"https://keras3.posit.co/dev/reference/op_softsign.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Softsign activation function. — op_softsign","text":"x Input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_softsign.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Softsign activation function. — op_softsign","text":"tensor shape x.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_softsign.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Softsign activation function. — op_softsign","text":"","code":"x <- op_convert_to_tensor(c(-0.100, -10.0, 1.0, 0.0, 100.0)) op_softsign(x) ## tf.Tensor([-0.09090909 -0.90909091  0.5         0.          0.99009901], shape=(5), dtype=float64) x <- seq(-10, 10, .1) plot(x, op_softsign(x), ylim = c(-1, 1))"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_solve.html","id":null,"dir":"Reference","previous_headings":"","what":"Solves a linear system of equations given by a x = b. — op_solve","title":"Solves a linear system of equations given by a x = b. — op_solve","text":"Solves x equation %*% x == b.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_solve.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Solves a linear system of equations given by a x = b. — op_solve","text":"","code":"op_solve(a, b)"},{"path":"https://keras3.posit.co/dev/reference/op_solve.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Solves a linear system of equations given by a x = b. — op_solve","text":"tensor shape (..., M, M) representing coefficients matrix. b tensor shape (..., M) (..., M, N) representing right-hand side \"dependent variable\" matrix.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_solve.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Solves a linear system of equations given by a x = b. — op_solve","text":"tensor shape (..., M) (..., M, N) representing solution linear system. Returned shape identical b.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_solve.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Solves a linear system of equations given by a x = b. — op_solve","text":"","code":"a <- op_array(c(1, 2, 4, 5), dtype=\"float32\") |> op_reshape(c(2, 2)) b <- op_array(c(2, 4, 8, 10), dtype=\"float32\") |> op_reshape(c(2, 2)) op_solve(a, b) ## tf.Tensor( ## [[2. 0.] ##  [0. 2.]], shape=(2, 2), dtype=float32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_solve_triangular.html","id":null,"dir":"Reference","previous_headings":"","what":"Solves a linear system of equations given by a %*% x = b. — op_solve_triangular","title":"Solves a linear system of equations given by a %*% x = b. — op_solve_triangular","text":"Solves linear system equations given %*% x = b.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_solve_triangular.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Solves a linear system of equations given by a %*% x = b. — op_solve_triangular","text":"","code":"op_solve_triangular(a, b, lower = FALSE)"},{"path":"https://keras3.posit.co/dev/reference/op_solve_triangular.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Solves a linear system of equations given by a %*% x = b. — op_solve_triangular","text":"tensor shape (..., M, M) representing coefficients matrix. b tensor shape (..., M) (..., M, N) representing right-hand side \"dependent variable\" matrix. lower logical. Use data contained lower triangle . Default use upper triangle.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_solve_triangular.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Solves a linear system of equations given by a %*% x = b. — op_solve_triangular","text":"tensor shape (..., M) (..., M, N) representing solution linear system. Returned shape identical b.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_sort.html","id":null,"dir":"Reference","previous_headings":"","what":"Sorts the elements of x along a given axis in ascending order. — op_sort","title":"Sorts the elements of x along a given axis in ascending order. — op_sort","text":"Sorts elements x along given axis ascending order.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_sort.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Sorts the elements of x along a given axis in ascending order. — op_sort","text":"","code":"op_sort(x, axis = -1L)"},{"path":"https://keras3.posit.co/dev/reference/op_sort.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Sorts the elements of x along a given axis in ascending order. — op_sort","text":"x Input tensor. axis Axis along sort. NULL, tensor flattened sorting. Defaults -1; last axis.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_sort.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Sorts the elements of x along a given axis in ascending order. — op_sort","text":"Sorted tensor.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_sparse_categorical_crossentropy.html","id":null,"dir":"Reference","previous_headings":"","what":"Computes sparse categorical cross-entropy loss. — op_sparse_categorical_crossentropy","title":"Computes sparse categorical cross-entropy loss. — op_sparse_categorical_crossentropy","text":"sparse categorical cross-entropy loss similar categorical cross-entropy, used target tensor contains integer class labels instead one-hot encoded vectors. measures dissimilarity target output probabilities logits.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_sparse_categorical_crossentropy.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes sparse categorical cross-entropy loss. — op_sparse_categorical_crossentropy","text":"","code":"op_sparse_categorical_crossentropy(   target,   output,   from_logits = FALSE,   axis = -1L )"},{"path":"https://keras3.posit.co/dev/reference/op_sparse_categorical_crossentropy.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Computes sparse categorical cross-entropy loss. — op_sparse_categorical_crossentropy","text":"target target tensor representing true class labels integers. shape match shape output tensor except last dimension. output output tensor representing predicted probabilities logits. shape match shape target tensor except last dimension. from_logits (optional) Whether output tensor logits probabilities. Set TRUE output represents logits; otherwise, set FALSE output represents probabilities. Defaults FALSE. axis (optional) axis along sparse categorical cross-entropy computed. Defaults -1, corresponds last dimension tensors.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_sparse_categorical_crossentropy.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Computes sparse categorical cross-entropy loss. — op_sparse_categorical_crossentropy","text":"Integer tensor: computed sparse categorical cross-entropy loss target output.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_sparse_categorical_crossentropy.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Computes sparse categorical cross-entropy loss. — op_sparse_categorical_crossentropy","text":"","code":"target <- op_array(c(0, 1, 2), dtype=\"int32\") output <- op_array(rbind(c(0.9, 0.05, 0.05),                         c(0.1, 0.8,  0.1),                         c(0.2, 0.3,  0.5))) op_sparse_categorical_crossentropy(target, output) ## tf.Tensor([0.10536052 0.22314355 0.69314718], shape=(3), dtype=float64)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_sparse_plus.html","id":null,"dir":"Reference","previous_headings":"","what":"SparsePlus activation function. — op_sparse_plus","title":"SparsePlus activation function. — op_sparse_plus","text":"defined f(x) = 0 x <= -1. f(x) = (1/4) * (x + 1)^2 -1 < x < 1. f(x) = x x >= 1.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_sparse_plus.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"SparsePlus activation function. — op_sparse_plus","text":"","code":"op_sparse_plus(x)"},{"path":"https://keras3.posit.co/dev/reference/op_sparse_plus.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"SparsePlus activation function. — op_sparse_plus","text":"x Input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_sparse_plus.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"SparsePlus activation function. — op_sparse_plus","text":"tensor shape x.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_sparse_plus.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"SparsePlus activation function. — op_sparse_plus","text":"","code":"x <- op_array(c(-1.0, 0.0, 1.0)) op_sparse_plus(x) ## tf.Tensor([0.   0.25 1.  ], shape=(3), dtype=float32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_sparse_sigmoid.html","id":null,"dir":"Reference","previous_headings":"","what":"Sparse sigmoid activation function. — op_sparse_sigmoid","title":"Sparse sigmoid activation function. — op_sparse_sigmoid","text":"defined f(x) = 0 x <= -1, f(x) = 0.5 * (x + 1) -1 < x < 1, f(x) = 1 x >= 1.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_sparse_sigmoid.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Sparse sigmoid activation function. — op_sparse_sigmoid","text":"","code":"op_sparse_sigmoid(x)"},{"path":"https://keras3.posit.co/dev/reference/op_sparse_sigmoid.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Sparse sigmoid activation function. — op_sparse_sigmoid","text":"x Input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_sparse_sigmoid.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Sparse sigmoid activation function. — op_sparse_sigmoid","text":"tensor shape x.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_sparse_sigmoid.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Sparse sigmoid activation function. — op_sparse_sigmoid","text":"","code":"x <- op_array(c(-1.0, 0.0, 1.0)) op_sparse_sigmoid(x) ## tf.Tensor([0.  0.5 1. ], shape=(3), dtype=float32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_sparsemax.html","id":null,"dir":"Reference","previous_headings":"","what":"Sparsemax activation function. — op_sparsemax","title":"Sparsemax activation function. — op_sparsemax","text":"batch , class j, sparsemax activation function defined : sparsemax(x)[, j] = max(x[, j] - (x[, :]), 0).","code":""},{"path":"https://keras3.posit.co/dev/reference/op_sparsemax.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Sparsemax activation function. — op_sparsemax","text":"","code":"op_sparsemax(x, axis = -1L)"},{"path":"https://keras3.posit.co/dev/reference/op_sparsemax.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Sparsemax activation function. — op_sparsemax","text":"x Input tensor. axis int, axis along sparsemax operation applied.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_sparsemax.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Sparsemax activation function. — op_sparsemax","text":"tensor, output sparsemax transformation. type shape x.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_sparsemax.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Sparsemax activation function. — op_sparsemax","text":"","code":"x <- op_array(c(-1., 0., 1.)) op_sparsemax(x) ## tf.Tensor([0. 0. 1.], shape=(3), dtype=float32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_split.html","id":null,"dir":"Reference","previous_headings":"","what":"Split a tensor into chunks. — op_split","title":"Split a tensor into chunks. — op_split","text":"Split tensor chunks.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_split.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Split a tensor into chunks. — op_split","text":"","code":"op_split(x, indices_or_sections, axis = 1L)"},{"path":"https://keras3.posit.co/dev/reference/op_split.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Split a tensor into chunks. — op_split","text":"x Input tensor. indices_or_sections integer, N, tensor split N equal sections along axis. 1-D array sorted integers, entries indicate indices tensor split along axis. axis Axis along split. Defaults 1, first axis.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_split.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Split a tensor into chunks. — op_split","text":"list tensors.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_split.html","id":"note","dir":"Reference","previous_headings":"","what":"Note","title":"Split a tensor into chunks. — op_split","text":"split result equal division using Torch backend.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_split.html","id":"example","dir":"Reference","previous_headings":"","what":"Example","title":"Split a tensor into chunks. — op_split","text":"","code":"x <- op_arange(12)  # pass a scalar integer for n sections op_split(x, 2) ## [[1]] ## tf.Tensor([1. 2. 3. 4. 5. 6.], shape=(6), dtype=float32) ## ## [[2]] ## tf.Tensor([ 7.  8.  9. 10. 11. 12.], shape=(6), dtype=float32) op_split(x, 3) ## [[1]] ## tf.Tensor([1. 2. 3. 4.], shape=(4), dtype=float32) ## ## [[2]] ## tf.Tensor([5. 6. 7. 8.], shape=(4), dtype=float32) ## ## [[3]] ## tf.Tensor([ 9. 10. 11. 12.], shape=(4), dtype=float32) # 1-d array/tensor for indices op_split(x, array(c(3, 8))) ## [[1]] ## tf.Tensor([1. 2.], shape=(2), dtype=float32) ## ## [[2]] ## tf.Tensor([3. 4. 5. 6. 7.], shape=(5), dtype=float32) ## ## [[3]] ## tf.Tensor([ 8.  9. 10. 11. 12.], shape=(5), dtype=float32) op_split(x, array(c(3))) ## [[1]] ## tf.Tensor([1. 2.], shape=(2), dtype=float32) ## ## [[2]] ## tf.Tensor([ 3.  4.  5.  6.  7.  8.  9. 10. 11. 12.], shape=(10), dtype=float32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_sqrt.html","id":null,"dir":"Reference","previous_headings":"","what":"Return the non-negative square root of a tensor, element-wise. — op_sqrt","title":"Return the non-negative square root of a tensor, element-wise. — op_sqrt","text":"Return non-negative square root tensor, element-wise.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_sqrt.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Return the non-negative square root of a tensor, element-wise. — op_sqrt","text":"","code":"op_sqrt(x)"},{"path":"https://keras3.posit.co/dev/reference/op_sqrt.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Return the non-negative square root of a tensor, element-wise. — op_sqrt","text":"x Input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_sqrt.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Return the non-negative square root of a tensor, element-wise. — op_sqrt","text":"Output tensor, non-negative square root x.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_square.html","id":null,"dir":"Reference","previous_headings":"","what":"Return the element-wise square of the input. — op_square","title":"Return the element-wise square of the input. — op_square","text":"Return element-wise square input.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_square.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Return the element-wise square of the input. — op_square","text":"","code":"op_square(x)"},{"path":"https://keras3.posit.co/dev/reference/op_square.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Return the element-wise square of the input. — op_square","text":"x Input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_square.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Return the element-wise square of the input. — op_square","text":"Output tensor, square x.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_squareplus.html","id":null,"dir":"Reference","previous_headings":"","what":"Squareplus activation function. — op_squareplus","title":"Squareplus activation function. — op_squareplus","text":"Squareplus activation function defined : f(x) = (x + sqrt(x^2 + b)) / 2","code":""},{"path":"https://keras3.posit.co/dev/reference/op_squareplus.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Squareplus activation function. — op_squareplus","text":"","code":"op_squareplus(x, b = 4L)"},{"path":"https://keras3.posit.co/dev/reference/op_squareplus.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Squareplus activation function. — op_squareplus","text":"x Input tensor. b Smoothness parameter. Defaults 4.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_squareplus.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Squareplus activation function. — op_squareplus","text":"tensor shape x.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_squareplus.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Squareplus activation function. — op_squareplus","text":"","code":"x <- op_array(c(-1.0, 0.0, 1.0)) op_squareplus(x) ## tf.Tensor([0.618034 1.       1.618034], shape=(3), dtype=float32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_squeeze.html","id":null,"dir":"Reference","previous_headings":"","what":"Remove axes of length one from x. — op_squeeze","title":"Remove axes of length one from x. — op_squeeze","text":"Remove axes length one x.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_squeeze.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Remove axes of length one from x. — op_squeeze","text":"","code":"op_squeeze(x, axis = NULL)"},{"path":"https://keras3.posit.co/dev/reference/op_squeeze.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Remove axes of length one from x. — op_squeeze","text":"x Input tensor. axis Select subset entries length one shape.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_squeeze.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Remove axes of length one from x. — op_squeeze","text":"input tensor subset dimensions length 1 removed.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_stack.html","id":null,"dir":"Reference","previous_headings":"","what":"Join a sequence of tensors along a new axis. — op_stack","title":"Join a sequence of tensors along a new axis. — op_stack","text":"axis parameter specifies index new axis dimensions result.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_stack.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Join a sequence of tensors along a new axis. — op_stack","text":"","code":"op_stack(x, axis = 1L)"},{"path":"https://keras3.posit.co/dev/reference/op_stack.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Join a sequence of tensors along a new axis. — op_stack","text":"x sequence tensors. axis Axis along stack. Defaults 1, first axis.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_stack.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Join a sequence of tensors along a new axis. — op_stack","text":"stacked tensor.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_std.html","id":null,"dir":"Reference","previous_headings":"","what":"Compute the standard deviation along the specified axis. — op_std","title":"Compute the standard deviation along the specified axis. — op_std","text":"Compute standard deviation along specified axis.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_std.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Compute the standard deviation along the specified axis. — op_std","text":"","code":"op_std(x, axis = NULL, keepdims = FALSE)"},{"path":"https://keras3.posit.co/dev/reference/op_std.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Compute the standard deviation along the specified axis. — op_std","text":"x Input tensor. axis Axis along compute standard deviation. Default compute standard deviation flattened tensor. keepdims set TRUE, axes reduced left result dimensions size one.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_std.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Compute the standard deviation along the specified axis. — op_std","text":"Output tensor containing standard deviation values.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_stft.html","id":null,"dir":"Reference","previous_headings":"","what":"Short-Time Fourier Transform along the last axis of the input. — op_stft","title":"Short-Time Fourier Transform along the last axis of the input. — op_stft","text":"STFT computes Fourier transform short overlapping windows input. giving frequency components signal change time.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_stft.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Short-Time Fourier Transform along the last axis of the input. — op_stft","text":"","code":"op_stft(   x,   sequence_length,   sequence_stride,   fft_length,   window = \"hann\",   center = TRUE )"},{"path":"https://keras3.posit.co/dev/reference/op_stft.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Short-Time Fourier Transform along the last axis of the input. — op_stft","text":"x Input tensor. sequence_length integer representing sequence length. sequence_stride integer representing sequence hop size. fft_length integer representing size FFT apply. specified, uses smallest power 2 enclosing sequence_length. window string, tensor window NULL. window string, available values \"hann\" \"hamming\". window tensor, used directly window length must sequence_length. window NULL, windowing used. Defaults \"hann\". center Whether pad x sides t-th sequence centered time t * sequence_stride. Otherwise, t-th sequence begins time t * sequence_stride. Defaults TRUE.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_stft.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Short-Time Fourier Transform along the last axis of the input. — op_stft","text":"list containing two tensors - real imaginary parts STFT output.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_stft.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Short-Time Fourier Transform along the last axis of the input. — op_stft","text":"","code":"x <- op_array(c(0, 1, 2, 3, 4)) op_stft(x, 3, 2, 3) ## [[1]] ## tf.Tensor( ## [[ 0.  0.] ##  [ 2. -1.] ##  [ 4. -2.]], shape=(3, 2), dtype=float32) ## ## [[2]] ## tf.Tensor( ## [[ 0.         0.       ] ##  [ 0.        -1.7320508] ##  [ 0.        -3.4641016]], shape=(3, 2), dtype=float32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_stop_gradient.html","id":null,"dir":"Reference","previous_headings":"","what":"Stops gradient computation. — op_stop_gradient","title":"Stops gradient computation. — op_stop_gradient","text":"Stops gradient computation.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_stop_gradient.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Stops gradient computation. — op_stop_gradient","text":"","code":"op_stop_gradient(variable)"},{"path":"https://keras3.posit.co/dev/reference/op_stop_gradient.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Stops gradient computation. — op_stop_gradient","text":"variable tensor variable gradient computation disabled.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_stop_gradient.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Stops gradient computation. — op_stop_gradient","text":"variable gradient computation disabled.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_stop_gradient.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Stops gradient computation. — op_stop_gradient","text":"","code":"var <- op_convert_to_tensor(c(1, 2, 3), dtype=\"float32\") var <- op_stop_gradient(var)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_subset.html","id":null,"dir":"Reference","previous_headings":"","what":"Subset elements from a tensor — op_subset","title":"Subset elements from a tensor — op_subset","text":"Extract elements tensor using common R-style [ indexing idioms. function can also conveniently accessed via syntax tensor@r[...].","code":""},{"path":"https://keras3.posit.co/dev/reference/op_subset.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Subset elements from a tensor — op_subset","text":"","code":"op_subset(x, ...)  op_subset(x, ...) <- value  op_subset_set(x, ..., value)"},{"path":"https://keras3.posit.co/dev/reference/op_subset.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Subset elements from a tensor — op_subset","text":"x Input tensor. ... Indices specifying elements extract. argument ... can : integer scalar 1-d integer logical vector NULL newaxis .. symbol slice expression using : single argument supplied ..., ..1 can also : logical array shape x integer matrix ncol(..1) == op_rank(x) value new value replace selected subset .","code":""},{"path":"https://keras3.posit.co/dev/reference/op_subset.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Subset elements from a tensor — op_subset","text":"tensor containing subset elements.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_subset.html","id":"details","dir":"Reference","previous_headings":"","what":"Details","title":"Subset elements from a tensor — op_subset","text":"semantics similar R's [, differences:","code":""},{"path":"https://keras3.posit.co/dev/reference/op_subset.html","id":"differences-from-r-s-","dir":"Reference","previous_headings":"","what":"Differences from R's [:","title":"Subset elements from a tensor — op_subset","text":"Negative indices follow Python-style indexing, counting end array. NULL newaxis adds new dimension (equivalent op_expand_dims()). fewer indices dimensions (op_rank(x)) provided, missing dimensions implicitly filled. example, x matrix, x[1] returns first row. .. all_dims() expands include unspecified dimensions (see examples). Extended slicing syntax (:) supported, including: Strided steps: x@r[start:end:step] NA values start end. NA start defaults 1, NA end defaults axis size. logical array matching shape x selects elements row-wise order.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_subset.html","id":"similarities-with-r-s-","dir":"Reference","previous_headings":"","what":"Similarities with R's [:","title":"Subset elements from a tensor — op_subset","text":"Similarities R's [ (differences Python's [): Positive indices 1-based. Slices (x[start:end]) inclusive end. 1-d logical/integer arrays subset along respective axis. Multiple vectors provided different axes return intersected subsets. single integer matrix ncol() == op_rank(x) selects elements coordinates. row matrix specifies location one value, column corresponds axis tensor subsetted.  means use 2-column matrix subset matrix, 3-column matrix subset 3d array, .","code":""},{"path":"https://keras3.posit.co/dev/reference/op_subset.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Subset elements from a tensor — op_subset","text":"","code":"(x <- op_arange(5L) + 10L) ## tf.Tensor([11 12 13 14 15], shape=(5), dtype=int32) # Basic example, get first element op_subset(x, 1) ## tf.Tensor(11, shape=(), dtype=int32) # Use `@r[` syntax x@r[1]           # same as `op_subset(x, 1)` ## tf.Tensor(11, shape=(), dtype=int32) x@r[1:2]         # get the first 2 elements ## tf.Tensor([11 12], shape=(2), dtype=int32) x@r[c(1, 3)]     # first and third element ## tf.Tensor([11 13], shape=(2), dtype=int32) # Negative indices x@r[-1]          # last element ## tf.Tensor(15, shape=(), dtype=int32) x@r[-2]          # second to last element ## tf.Tensor(14, shape=(), dtype=int32) x@r[c(-1, -2)]   # last and second to last elements ## tf.Tensor([15 14], shape=(2), dtype=int32) x@r[c(-2, -1)]   # second to last and last elements ## tf.Tensor([14 15], shape=(2), dtype=int32) x@r[c(1, -1)]    # first and last elements ## tf.Tensor([11 15], shape=(2), dtype=int32) # Slices x@r[1:3]          # first 3 elements ## tf.Tensor([11 12 13], shape=(3), dtype=int32) x@r[NA:3]         # first 3 elements ## tf.Tensor([11 12 13], shape=(3), dtype=int32) x@r[1:5]          # all elements ## tf.Tensor([11 12 13 14 15], shape=(5), dtype=int32) x@r[1:-1]         # all elements ## tf.Tensor([11 12 13 14 15], shape=(5), dtype=int32) x@r[NA:NA]        # all elements ## tf.Tensor([11 12 13 14 15], shape=(5), dtype=int32) x@r[]             # all elements ## tf.Tensor([11 12 13 14 15], shape=(5), dtype=int32) x@r[1:-2]         # drop last element ## tf.Tensor([11 12 13 14], shape=(4), dtype=int32) x@r[NA:-2]        # drop last element ## tf.Tensor([11 12 13 14], shape=(4), dtype=int32) x@r[2:NA]         # drop first element ## tf.Tensor([12 13 14 15], shape=(4), dtype=int32) # 2D array examples xr <- array(1:12, c(3, 4)) x <- op_convert_to_tensor(xr)  # Basic subsetting x@r[1, ]      # first row ## tf.Tensor([ 1  4  7 10], shape=(4), dtype=int32) x@r[1]        # also first row! Missing axes are implicitly inserted ## tf.Tensor([ 1  4  7 10], shape=(4), dtype=int32) x@r[-1]       # last row ## tf.Tensor([ 3  6  9 12], shape=(4), dtype=int32) x@r[, 2]      # second column ## tf.Tensor([4 5 6], shape=(3), dtype=int32) x@r[, 2:2]    # second column, but shape preserved (like [, drop=FALSE]) ## tf.Tensor( ## [[4] ##  [5] ##  [6]], shape=(3, 1), dtype=int32) # Subsetting with a boolean array # Note: extracted elements are selected row-wise, not column-wise mask <- x >= 6 x@r[mask]             # returns a 1D tensor ## tf.Tensor([ 7 10  8 11  6  9 12], shape=(7), dtype=int32) x.r <- as.array(x) mask.r <- as.array(mask) # as.array(x)[mask] selects column-wise. Use `aperm()` to reverse search order. all(aperm(x.r)[aperm(mask.r)] == as.array(x@r[mask])) ## [1] TRUE # Subsetting with a matrix of index positions indices <- rbind(c(1, 1), c(2, 2), c(3, 3)) x@r[indices] # get diagonal elements ## tf.Tensor([1 5 9], shape=(3), dtype=int32) x.r[indices] # same as subsetting an R array ## [1] 1 5 9 # 3D array examples # Image: 4x4 pixels, 3 colors (RGB) # Tensor shape: (img_height, img_width, img_color_channels) shp <- shape(4, 4, 3) x <- op_arange(prod(shp)) |> op_reshape(shp)  # Convert to a batch of images by inserting a new axis # New shape: (batch_size, img_height, img_width, img_color_channels) x@r[newaxis, , , ] |> op_shape() ## shape(1, 4, 4, 3) x@r[newaxis] |> op_shape()  # same as above ## shape(1, 4, 4, 3) x@r[NULL] |> op_shape()     # same as above ## shape(1, 4, 4, 3) x <- x@r[newaxis] # Extract color channels x@r[, , , 1]          # red channel ## tf.Tensor( ## [[[ 1.  4.  7. 10.] ##   [13. 16. 19. 22.] ##   [25. 28. 31. 34.] ##   [37. 40. 43. 46.]]], shape=(1, 4, 4), dtype=float32) x@r[.., 1]            # red channel, same as above using .. shorthand ## tf.Tensor( ## [[[ 1.  4.  7. 10.] ##   [13. 16. 19. 22.] ##   [25. 28. 31. 34.] ##   [37. 40. 43. 46.]]], shape=(1, 4, 4), dtype=float32) x@r[.., 2]            # green channel ## tf.Tensor( ## [[[ 2.  5.  8. 11.] ##   [14. 17. 20. 23.] ##   [26. 29. 32. 35.] ##   [38. 41. 44. 47.]]], shape=(1, 4, 4), dtype=float32) x@r[.., 3]            # blue channel ## tf.Tensor( ## [[[ 3.  6.  9. 12.] ##   [15. 18. 21. 24.] ##   [27. 30. 33. 36.] ##   [39. 42. 45. 48.]]], shape=(1, 4, 4), dtype=float32) # .. expands to all unspecified axes. op_shape(x@r[]) ## shape(1, 4, 4, 3) op_shape(x@r[..]) ## shape(1, 4, 4, 3) op_shape(x@r[1, ..]) ## shape(4, 4, 3) op_shape(x@r[1, .., 1, 1]) ## shape(4) op_shape(x@r[1, 1, 1, .., 1]) ## shape() # op_subset<- uses the same semantics, but note that not all tensors # support modification. E.g., TensorFlow constant tensors cannot be modified, # while TensorFlow Variables can be.  (x <- tensorflow::tf$Variable(matrix(1, nrow = 2, ncol = 3))) ## <tf.Variable 'Variable:0' shape=(2, 3) dtype=float64, numpy= ## array([[1., 1., 1.], ##        [1., 1., 1.]])> op_subset(x, 1) <- 9 x ## <tf.Variable 'UnreadVariable' shape=(2, 3) dtype=float64, numpy= ## array([[9., 9., 9.], ##        [1., 1., 1.]])> x@r[1,1] <- 33 x ## <tf.Variable 'UnreadVariable' shape=(2, 3) dtype=float64, numpy= ## array([[33.,  9.,  9.], ##        [ 1.,  1.,  1.]])>"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_subtract.html","id":null,"dir":"Reference","previous_headings":"","what":"Subtract arguments element-wise. — op_subtract","title":"Subtract arguments element-wise. — op_subtract","text":"Note function automatically called using R operator - tensor.","code":"x <- op_ones(c(3)) op_subtract(x, x) ## tf.Tensor([0. 0. 0.], shape=(3), dtype=float32) x - x ## tf.Tensor([0. 0. 0.], shape=(3), dtype=float32)"},{"path":"https://keras3.posit.co/dev/reference/op_subtract.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Subtract arguments element-wise. — op_subtract","text":"","code":"op_subtract(x1, x2)"},{"path":"https://keras3.posit.co/dev/reference/op_subtract.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Subtract arguments element-wise. — op_subtract","text":"x1 First input tensor. x2 Second input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_subtract.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Subtract arguments element-wise. — op_subtract","text":"Output tensor, element-wise difference x1 x2.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_sum.html","id":null,"dir":"Reference","previous_headings":"","what":"Sum of a tensor over the given axes. — op_sum","title":"Sum of a tensor over the given axes. — op_sum","text":"Sum tensor given axes.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_sum.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Sum of a tensor over the given axes. — op_sum","text":"","code":"op_sum(x, axis = NULL, keepdims = FALSE)"},{"path":"https://keras3.posit.co/dev/reference/op_sum.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Sum of a tensor over the given axes. — op_sum","text":"x Input tensor. axis Axis axes along sum computed. default compute sum flattened tensor. keepdims set TRUE, axes reduced left result dimensions size one.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_sum.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Sum of a tensor over the given axes. — op_sum","text":"Output tensor containing sum.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_svd.html","id":null,"dir":"Reference","previous_headings":"","what":"Computes the singular value decomposition of a matrix. — op_svd","title":"Computes the singular value decomposition of a matrix. — op_svd","text":"Computes singular value decomposition matrix.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_svd.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Computes the singular value decomposition of a matrix. — op_svd","text":"","code":"op_svd(x, full_matrices = TRUE, compute_uv = TRUE)"},{"path":"https://keras3.posit.co/dev/reference/op_svd.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Computes the singular value decomposition of a matrix. — op_svd","text":"x Input tensor shape (..., M, N). full_matrices Logical compute_uv Logical","code":""},{"path":"https://keras3.posit.co/dev/reference/op_svd.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Computes the singular value decomposition of a matrix. — op_svd","text":"list three tensors: tensor shape (..., M, M) containing left singular vectors, tensor shape (..., M, N) containing singular values tensor shape (..., N, N) containing right singular vectors.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_swapaxes.html","id":null,"dir":"Reference","previous_headings":"","what":"Interchange two axes of a tensor. — op_swapaxes","title":"Interchange two axes of a tensor. — op_swapaxes","text":"Interchange two axes tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_swapaxes.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Interchange two axes of a tensor. — op_swapaxes","text":"","code":"op_swapaxes(x, axis1, axis2)"},{"path":"https://keras3.posit.co/dev/reference/op_swapaxes.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Interchange two axes of a tensor. — op_swapaxes","text":"x Input tensor. axis1 First axis. axis2 Second axis.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_swapaxes.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Interchange two axes of a tensor. — op_swapaxes","text":"tensor axes swapped.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_switch.html","id":null,"dir":"Reference","previous_headings":"","what":"Apply exactly one of the branches given by index. — op_switch","title":"Apply exactly one of the branches given by index. — op_switch","text":"index bounds, clamped within bounds. semantics switch given roughly implementation:","code":"op_switch <- function(index, branches, ...) {   index <- op_clip(index, 1, length(branches))   branches[[index]](...) }"},{"path":"https://keras3.posit.co/dev/reference/op_switch.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Apply exactly one of the branches given by index. — op_switch","text":"","code":"op_switch(index, branches, ...)"},{"path":"https://keras3.posit.co/dev/reference/op_switch.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Apply exactly one of the branches given by index. — op_switch","text":"index integer scalar indicating branch function apply (1-based). branches list functions applied based index. ... Inputs whichever branch applied.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_switch.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Apply exactly one of the branches given by index. — op_switch","text":"outputs branch(...) branch selected based index.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_switch.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Apply exactly one of the branches given by index. — op_switch","text":"","code":"add_fn <- function(x, y) x + y + 100 subtract_fn <- function(x, y) x - y - 100 x <- op_array(2) y <- op_array(0.5) branches <- list(add_fn, subtract_fn) op_switch(0, branches, x, y) # + ## tf.Tensor(102.5, shape=(), dtype=float32) op_switch(1, branches, x, y) # + ## tf.Tensor(102.5, shape=(), dtype=float32) op_switch(2, branches, x, y) # - ## tf.Tensor(-98.5, shape=(), dtype=float32) op_switch(3, branches, x, y) # - ## tf.Tensor(-98.5, shape=(), dtype=float32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_take.html","id":null,"dir":"Reference","previous_headings":"","what":"Take elements from a tensor along an axis. — op_take","title":"Take elements from a tensor along an axis. — op_take","text":"Take elements tensor along axis.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_take.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Take elements from a tensor along an axis. — op_take","text":"","code":"op_take(x, indices, axis = NULL, zero_indexed = FALSE)"},{"path":"https://keras3.posit.co/dev/reference/op_take.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Take elements from a tensor along an axis. — op_take","text":"x Source tensor. indices indices values extract. axis axis select values. default, flattened input tensor used. zero_indexed TRUE, treats indices zero-based (0 encodes first position); FALSE (default), treats indices one-based (1 encodes first position).","code":""},{"path":"https://keras3.posit.co/dev/reference/op_take.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Take elements from a tensor along an axis. — op_take","text":"corresponding tensor values.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_take_along_axis.html","id":null,"dir":"Reference","previous_headings":"","what":"Select values from x at the 1-D indices along the given axis. — op_take_along_axis","title":"Select values from x at the 1-D indices along the given axis. — op_take_along_axis","text":"Select values x 1-D indices along given axis.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_take_along_axis.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Select values from x at the 1-D indices along the given axis. — op_take_along_axis","text":"","code":"op_take_along_axis(x, indices, axis = NULL, zero_indexed = FALSE)"},{"path":"https://keras3.posit.co/dev/reference/op_take_along_axis.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Select values from x at the 1-D indices along the given axis. — op_take_along_axis","text":"x Source tensor. indices indices values extract. axis axis select values. default, flattened input tensor used. zero_indexed TRUE, treats indices zero-based (0 encodes first position); FALSE (default), treats indices one-based (1 encodes first position).","code":""},{"path":"https://keras3.posit.co/dev/reference/op_take_along_axis.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Select values from x at the 1-D indices along the given axis. — op_take_along_axis","text":"corresponding tensor values.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_tan.html","id":null,"dir":"Reference","previous_headings":"","what":"Compute tangent, element-wise. — op_tan","title":"Compute tangent, element-wise. — op_tan","text":"Compute tangent, element-wise.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_tan.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Compute tangent, element-wise. — op_tan","text":"","code":"op_tan(x)"},{"path":"https://keras3.posit.co/dev/reference/op_tan.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Compute tangent, element-wise. — op_tan","text":"x Input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_tan.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Compute tangent, element-wise. — op_tan","text":"Output tensor shape x.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_tanh.html","id":null,"dir":"Reference","previous_headings":"","what":"Hyperbolic tangent, element-wise. — op_tanh","title":"Hyperbolic tangent, element-wise. — op_tanh","text":"Hyperbolic tangent, element-wise.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_tanh.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Hyperbolic tangent, element-wise. — op_tanh","text":"","code":"op_tanh(x)"},{"path":"https://keras3.posit.co/dev/reference/op_tanh.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Hyperbolic tangent, element-wise. — op_tanh","text":"x Input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_tanh.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Hyperbolic tangent, element-wise. — op_tanh","text":"Output tensor shape x.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_tanh_shrink.html","id":null,"dir":"Reference","previous_headings":"","what":"Applies the tanh shrink function element-wise. — op_tanh_shrink","title":"Applies the tanh shrink function element-wise. — op_tanh_shrink","text":"defined : f(x) = x - tanh(x).","code":""},{"path":"https://keras3.posit.co/dev/reference/op_tanh_shrink.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Applies the tanh shrink function element-wise. — op_tanh_shrink","text":"","code":"op_tanh_shrink(x)"},{"path":"https://keras3.posit.co/dev/reference/op_tanh_shrink.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Applies the tanh shrink function element-wise. — op_tanh_shrink","text":"x Input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_tanh_shrink.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Applies the tanh shrink function element-wise. — op_tanh_shrink","text":"Output tensor shape x, element transformed according tanh shrink operation.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_tanh_shrink.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Applies the tanh shrink function element-wise. — op_tanh_shrink","text":"","code":"x <- op_array(c(-1., 0., 1.)) op_tanh_shrink(x) ## tf.Tensor([-0.23840582  0.          0.23840582], shape=(3), dtype=float32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_tensordot.html","id":null,"dir":"Reference","previous_headings":"","what":"Compute the tensor dot product along specified axes. — op_tensordot","title":"Compute the tensor dot product along specified axes. — op_tensordot","text":"Compute tensor dot product along specified axes.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_tensordot.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Compute the tensor dot product along specified axes. — op_tensordot","text":"","code":"op_tensordot(x1, x2, axes = 3L)"},{"path":"https://keras3.posit.co/dev/reference/op_tensordot.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Compute the tensor dot product along specified axes. — op_tensordot","text":"x1 First tensor. x2 Second tensor. axes integer, N, sum last N axes x1 first N axes x2 order. sizes corresponding axes must match. , list axes summed , first sequence applying x1, second x2. sequences must length.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_tensordot.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Compute the tensor dot product along specified axes. — op_tensordot","text":"tensor dot product inputs.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_threshold.html","id":null,"dir":"Reference","previous_headings":"","what":"Threshold activation function. — op_threshold","title":"Threshold activation function. — op_threshold","text":"function thresholds input x follows: f(x) = x x > threshold, f(x) = default_value otherwise.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_threshold.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Threshold activation function. — op_threshold","text":"","code":"op_threshold(x, threshold, default_value)"},{"path":"https://keras3.posit.co/dev/reference/op_threshold.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Threshold activation function. — op_threshold","text":"x Input tensor. threshold value decides retain replace x. default_value Value assign x <= threshold.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_threshold.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Threshold activation function. — op_threshold","text":"tensor shape x.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_threshold.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Threshold activation function. — op_threshold","text":"","code":"x <- op_array(c(-1.0, 0.0, 1.0, 2.0)) op_threshold(x, 1, 0) ## tf.Tensor([0. 0. 0. 2.], shape=(4), dtype=float32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_tile.html","id":null,"dir":"Reference","previous_headings":"","what":"Repeat x the number of times given by repeats. — op_tile","title":"Repeat x the number of times given by repeats. — op_tile","text":"repeats length d, result dimension max(d, x.ndim). x.ndim < d, x promoted d-dimensional prepending new axes. x.ndim > d, repeats promoted x.ndim prepending 1's .","code":""},{"path":"https://keras3.posit.co/dev/reference/op_tile.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Repeat x the number of times given by repeats. — op_tile","text":"","code":"op_tile(x, repeats)"},{"path":"https://keras3.posit.co/dev/reference/op_tile.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Repeat x the number of times given by repeats. — op_tile","text":"x Input tensor. repeats number repetitions x along axis.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_tile.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Repeat x the number of times given by repeats. — op_tile","text":"tiled output tensor.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_top_k.html","id":null,"dir":"Reference","previous_headings":"","what":"Finds the top-k values and their indices in a tensor. — op_top_k","title":"Finds the top-k values and their indices in a tensor. — op_top_k","text":"Finds top-k values indices tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_top_k.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Finds the top-k values and their indices in a tensor. — op_top_k","text":"","code":"op_top_k(x, k, sorted = TRUE, zero_indexed = FALSE)"},{"path":"https://keras3.posit.co/dev/reference/op_top_k.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Finds the top-k values and their indices in a tensor. — op_top_k","text":"x Input tensor. k integer representing number top elements retrieve. sorted boolean indicating whether sort output descending order. Defaults TRUE. zero_indexed TRUE, returned indices zero-based (0 encodes first position); FALSE (default), returned indices one-based (1 encodes first position).","code":""},{"path":"https://keras3.posit.co/dev/reference/op_top_k.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Finds the top-k values and their indices in a tensor. — op_top_k","text":"list containing two tensors. first tensor contains top-k values, second tensor contains indices top-k values input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_top_k.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Finds the top-k values and their indices in a tensor. — op_top_k","text":"","code":"x <- op_array(c(5, 2, 7, 1, 9, 3), \"int32\") op_top_k(x, k = 3) ## $values ## tf.Tensor([9 7 5], shape=(3), dtype=int32) ## ## $indices ## tf.Tensor([5 3 1], shape=(3), dtype=int32) c(values, indices) %<-% op_top_k(x, k = 3) values ## tf.Tensor([9 7 5], shape=(3), dtype=int32) indices ## tf.Tensor([5 3 1], shape=(3), dtype=int32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_trace.html","id":null,"dir":"Reference","previous_headings":"","what":"Return the sum along diagonals of the tensor. — op_trace","title":"Return the sum along diagonals of the tensor. — op_trace","text":"x 2-D, sum along diagonal given offset returned, .e., sum elements x[, +offset] . two dimensions, axes specified axis1 axis2 used determine 2-D sub-arrays whose traces returned. shape resulting tensor x axis1 axis2 removed.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_trace.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Return the sum along diagonals of the tensor. — op_trace","text":"","code":"op_trace(x, offset = 0L, axis1 = 1L, axis2 = 2L)"},{"path":"https://keras3.posit.co/dev/reference/op_trace.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Return the sum along diagonals of the tensor. — op_trace","text":"x Input tensor. offset Offset diagonal main diagonal. Can positive negative. Defaults 0. axis1 Axis used first axis 2-D sub-arrays. Defaults 1. (first axis). axis2 Axis used second axis 2-D sub-arrays. Defaults 2. (second axis).","code":""},{"path":"https://keras3.posit.co/dev/reference/op_trace.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Return the sum along diagonals of the tensor. — op_trace","text":"x 2-D, sum diagonal returned. x larger dimensions, tensor sums along diagonals returned.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_transpose.html","id":null,"dir":"Reference","previous_headings":"","what":"Returns a tensor with axes transposed. — op_transpose","title":"Returns a tensor with axes transposed. — op_transpose","text":"Returns tensor axes transposed.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_transpose.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Returns a tensor with axes transposed. — op_transpose","text":"","code":"op_transpose(x, axes = NULL)"},{"path":"https://keras3.posit.co/dev/reference/op_transpose.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Returns a tensor with axes transposed. — op_transpose","text":"x Input tensor. axes Sequence integers. Permutation dimensions x. default, order axes reversed.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_transpose.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Returns a tensor with axes transposed. — op_transpose","text":"x axes permuted.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_tri.html","id":null,"dir":"Reference","previous_headings":"","what":"Return a tensor with ones at and below a diagonal and zeros elsewhere. — op_tri","title":"Return a tensor with ones at and below a diagonal and zeros elsewhere. — op_tri","text":"Return tensor ones diagonal zeros elsewhere.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_tri.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Return a tensor with ones at and below a diagonal and zeros elsewhere. — op_tri","text":"","code":"op_tri(N, M = NULL, k = 0L, dtype = NULL)"},{"path":"https://keras3.posit.co/dev/reference/op_tri.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Return a tensor with ones at and below a diagonal and zeros elsewhere. — op_tri","text":"N Number rows tensor. M Number columns tensor. k sub-diagonal array filled. k = 0 main diagonal, k < 0 , k > 0 . default 0. dtype Data type returned tensor. default \"float32\".","code":""},{"path":"https://keras3.posit.co/dev/reference/op_tri.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Return a tensor with ones at and below a diagonal and zeros elsewhere. — op_tri","text":"Tensor lower triangle filled ones zeros elsewhere. T[, j] == 1 j <= + k, 0 otherwise.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_tril.html","id":null,"dir":"Reference","previous_headings":"","what":"Return lower triangle of a tensor. — op_tril","title":"Return lower triangle of a tensor. — op_tril","text":"tensors ndim exceeding 2, tril apply final two axes.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_tril.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Return lower triangle of a tensor. — op_tril","text":"","code":"op_tril(x, k = 0L)"},{"path":"https://keras3.posit.co/dev/reference/op_tril.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Return lower triangle of a tensor. — op_tril","text":"x Input tensor. k Diagonal zero elements. Defaults 0. main diagonal. k < 0 , k > 0 .","code":""},{"path":"https://keras3.posit.co/dev/reference/op_tril.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Return lower triangle of a tensor. — op_tril","text":"Lower triangle x, shape data type x.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_triu.html","id":null,"dir":"Reference","previous_headings":"","what":"Return upper triangle of a tensor. — op_triu","title":"Return upper triangle of a tensor. — op_triu","text":"tensors ndim exceeding 2, triu apply final two axes.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_triu.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Return upper triangle of a tensor. — op_triu","text":"","code":"op_triu(x, k = 0L)"},{"path":"https://keras3.posit.co/dev/reference/op_triu.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Return upper triangle of a tensor. — op_triu","text":"x Input tensor. k Diagonal zero elements. Defaults 0. main diagonal. k < 0 , k > 0 .","code":""},{"path":"https://keras3.posit.co/dev/reference/op_triu.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Return upper triangle of a tensor. — op_triu","text":"Upper triangle x, shape data type x.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_trunc.html","id":null,"dir":"Reference","previous_headings":"","what":"Return the truncated value of the input, element-wise. — op_trunc","title":"Return the truncated value of the input, element-wise. — op_trunc","text":"truncated value scalar x nearest integer closer zero x . short, fractional part signed number x discarded.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_trunc.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Return the truncated value of the input, element-wise. — op_trunc","text":"","code":"op_trunc(x)"},{"path":"https://keras3.posit.co/dev/reference/op_trunc.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Return the truncated value of the input, element-wise. — op_trunc","text":"x Input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_trunc.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Return the truncated value of the input, element-wise. — op_trunc","text":"truncated value element x.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_trunc.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Return the truncated value of the input, element-wise. — op_trunc","text":"","code":"x <- op_array(c(-1.7, -1.5, -0.2, 0.2, 1.5, 1.7, 2.0)) op_trunc(x) ## tf.Tensor([-1. -1. -0.  0.  1.  1.  2.], shape=(7), dtype=float32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_unravel_index.html","id":null,"dir":"Reference","previous_headings":"","what":"Convert flat indices to coordinate arrays in a given array shape. — op_unravel_index","title":"Convert flat indices to coordinate arrays in a given array shape. — op_unravel_index","text":"Convert flat indices coordinate arrays given array shape.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_unravel_index.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Convert flat indices to coordinate arrays in a given array shape. — op_unravel_index","text":"","code":"op_unravel_index(indices, shape)"},{"path":"https://keras3.posit.co/dev/reference/op_unravel_index.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Convert flat indices to coordinate arrays in a given array shape. — op_unravel_index","text":"indices integer array integers representing flat indices. shape shape array unravel .","code":""},{"path":"https://keras3.posit.co/dev/reference/op_unravel_index.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Convert flat indices to coordinate arrays in a given array shape. — op_unravel_index","text":"Tuple arrays dimension unraveled indices.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_unravel_index.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Convert flat indices to coordinate arrays in a given array shape. — op_unravel_index","text":"","code":"indices <- c(1, 5) shape <- array(c(3, 3)) op_unravel_index(indices, shape) ## [[1]] ## tf.Tensor([0 1], shape=(2), dtype=int32) ## ## [[2]] ## tf.Tensor([0 1], shape=(2), dtype=int32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_unstack.html","id":null,"dir":"Reference","previous_headings":"","what":"Unpacks the given dimension of a rank-R tensor into rank-(R-1) tensors. — op_unstack","title":"Unpacks the given dimension of a rank-R tensor into rank-(R-1) tensors. — op_unstack","text":"Unpacks given dimension rank-R tensor rank-(R-1) tensors.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_unstack.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Unpacks the given dimension of a rank-R tensor into rank-(R-1) tensors. — op_unstack","text":"","code":"op_unstack(x, num = NULL, axis = 1L)"},{"path":"https://keras3.posit.co/dev/reference/op_unstack.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Unpacks the given dimension of a rank-R tensor into rank-(R-1) tensors. — op_unstack","text":"x input tensor. num length dimension axis. Automatically inferred NULL. axis axis along unpack.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_unstack.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Unpacks the given dimension of a rank-R tensor into rank-(R-1) tensors. — op_unstack","text":"list tensors unpacked along given axis.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_unstack.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Unpacks the given dimension of a rank-R tensor into rank-(R-1) tensors. — op_unstack","text":"[3, 4))]: R:3,%204))","code":"x <- op_array(rbind(c(1, 2),                    c(3, 4))) op_unstack(x, axis=1) ## [[1]] ## tf.Tensor([1. 2.], shape=(2), dtype=float64) ## ## [[2]] ## tf.Tensor([3. 4.], shape=(2), dtype=float64) op_unstack(x, axis=2) ## [[1]] ## tf.Tensor([1. 3.], shape=(2), dtype=float64) ## ## [[2]] ## tf.Tensor([2. 4.], shape=(2), dtype=float64) all.equal(op_unstack(x),           op_unstack(x, axis = 1)) ## [1] TRUE all.equal(op_unstack(x, axis = -1),           op_unstack(x, axis = 2)) ## [1] TRUE # [array([1, 2)), array([3, 4))]"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_var.html","id":null,"dir":"Reference","previous_headings":"","what":"Compute the variance along the specified axes. — op_var","title":"Compute the variance along the specified axes. — op_var","text":"Compute variance along specified axes.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_var.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Compute the variance along the specified axes. — op_var","text":"","code":"op_var(x, axis = NULL, keepdims = FALSE)"},{"path":"https://keras3.posit.co/dev/reference/op_var.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Compute the variance along the specified axes. — op_var","text":"x Input tensor. axis Axis axes along variance computed. default compute variance flattened tensor. keepdims set TRUE, axes reduced left result dimensions size one.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_var.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Compute the variance along the specified axes. — op_var","text":"Output tensor containing variance.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_vdot.html","id":null,"dir":"Reference","previous_headings":"","what":"Return the dot product of two vectors. — op_vdot","title":"Return the dot product of two vectors. — op_vdot","text":"first argument complex, complex conjugate first argument used calculation dot product. Multidimensional tensors flattened dot product taken.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_vdot.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Return the dot product of two vectors. — op_vdot","text":"","code":"op_vdot(x1, x2)"},{"path":"https://keras3.posit.co/dev/reference/op_vdot.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Return the dot product of two vectors. — op_vdot","text":"x1 First input tensor. complex, complex conjugate taken calculation dot product. x2 Second input tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_vdot.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Return the dot product of two vectors. — op_vdot","text":"Output tensor.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_vectorize.html","id":null,"dir":"Reference","previous_headings":"","what":"Turn a function into a vectorized function. — op_vectorize","title":"Turn a function into a vectorized function. — op_vectorize","text":"Turn function vectorized function.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_vectorize.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Turn a function into a vectorized function. — op_vectorize","text":"","code":"op_vectorize(func, ..., excluded = NULL, signature = NULL)"},{"path":"https://keras3.posit.co/dev/reference/op_vectorize.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Turn a function into a vectorized function. — op_vectorize","text":"func Callable single tensor argument. ... forward/backward compatability. excluded Optional set integers representing positional arguments function vectorized. passed directly func unmodified. signature Optional generalized universal function signature, e.g., \"(m,n),(n)->(m)\" vectorized matrix-vector multiplication. provided, func called (expected return) arrays shapes given size corresponding core dimensions. default, func assumed take scalar tensors input output.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_vectorize.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Turn a function into a vectorized function. — op_vectorize","text":"new function applies func every element input along axis 1 (batch axis, first axis).","code":""},{"path":"https://keras3.posit.co/dev/reference/op_vectorize.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Turn a function into a vectorized function. — op_vectorize","text":"","code":"# currently does not work w/ tensorflow backend if(config_backend() != \"tensorflow\") {    myfunc <- function(a, b) a + b    vfunc <- op_vectorize(myfunc)   y <- vfunc(c(1, 2, 3, 4), 2)   print(y)   # with Jax backend, y is:   # Array([3., 4., 5., 6.], dtype=float32) }"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_vectorized_map.html","id":null,"dir":"Reference","previous_headings":"","what":"Parallel map of function f on the first axis of tensor(s) elements. — op_vectorized_map","title":"Parallel map of function f on the first axis of tensor(s) elements. — op_vectorized_map","text":"Schematically, op_vectorized_map() maps first dimension provided tensors. elements list tensors, tensors required size first dimension, iterated together.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_vectorized_map.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Parallel map of function f on the first axis of tensor(s) elements. — op_vectorized_map","text":"","code":"op_vectorized_map(elements, f)"},{"path":"https://keras3.posit.co/dev/reference/op_vectorized_map.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Parallel map of function f on the first axis of tensor(s) elements. — op_vectorized_map","text":"elements see description f function taking either tensor, list tensors.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_vectorized_map.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Parallel map of function f on the first axis of tensor(s) elements. — op_vectorized_map","text":"tensor list tensors, result mapping f across elements.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_vectorized_map.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Parallel map of function f on the first axis of tensor(s) elements. — op_vectorized_map","text":"Note f may traced compiled. Meaning, R function may evaluated symbolic tensors using Jax TensorFlow backends, eager tensors. See output str() examples:","code":"(x <- op_arange(12L) |> op_reshape(c(3, 4))) ## tf.Tensor( ## [[ 1  2  3  4] ##  [ 5  6  7  8] ##  [ 9 10 11 12]], shape=(3, 4), dtype=int32) x |> op_vectorized_map(\\(row) {row + 10}) ## tf.Tensor( ## [[11 12 13 14] ##  [15 16 17 18] ##  [19 20 21 22]], shape=(3, 4), dtype=int32) list(x, x, x) |> op_vectorized_map(\\(rows) Reduce(`+`, rows)) ## tf.Tensor( ## [[ 3  6  9 12] ##  [15 18 21 24] ##  [27 30 33 36]], shape=(3, 4), dtype=int32) # simplest case, map f over rows of x, # where .x is 1 row of x input <- x output <- op_vectorized_map(input, function(.x) {   str(.x)   .x + 10 }) ## <tf.Tensor 'loop_body/GatherV2:0' shape=(4) dtype=int32> output ## tf.Tensor( ## [[11 12 13 14] ##  [15 16 17 18] ##  [19 20 21 22]], shape=(3, 4), dtype=int32) # map f over two tensors simultaneously. Here, # `.x` is a list of two # tensors. The return values from each call of `f(row)` are stacked to form the # final output input <- list(x, x) output <- op_vectorized_map(input, function(.x) {   str(.x)   .x[[1]] + 10 }) ## List of 2 ##  $ :<tf.Tensor 'loop_body/GatherV2:0' shape=(4) dtype=int32> ##  $ :<tf.Tensor 'loop_body/GatherV2_1:0' shape=(4) dtype=int32> output ## tf.Tensor( ## [[11 12 13 14] ##  [15 16 17 18] ##  [19 20 21 22]], shape=(3, 4), dtype=int32) # same as above, but now returning two tensors in the final output output <- op_vectorized_map(input, function(.x) {   str(.x)   c(.x1, .x2) %<-% .x   list(.x1+10, .x2+20) }) ## List of 2 ##  $ :<tf.Tensor 'loop_body/GatherV2:0' shape=(4) dtype=int32> ##  $ :<tf.Tensor 'loop_body/GatherV2_1:0' shape=(4) dtype=int32> output ## [[1]] ## tf.Tensor( ## [[11 12 13 14] ##  [15 16 17 18] ##  [19 20 21 22]], shape=(3, 4), dtype=int32) ## ## [[2]] ## tf.Tensor( ## [[21 22 23 24] ##  [25 26 27 28] ##  [29 30 31 32]], shape=(3, 4), dtype=int32) # passing named lists. # WARNING: if passing a named list, the order of elements of `.x` supplied # to `f` is not stable. Only retrieve elements by name. input <- list(name1 = x, name2 = x) output <- op_vectorized_map(input, function(.x) {   str(.x)   list(outname1 = .x$name1 + 10,        outname2 = .x$name2 + 20) }) ## List of 2 ##  $ name1:<tf.Tensor 'loop_body/GatherV2:0' shape=(4) dtype=int32> ##  $ name2:<tf.Tensor 'loop_body/GatherV2_1:0' shape=(4) dtype=int32> output ## $outname1 ## tf.Tensor( ## [[11 12 13 14] ##  [15 16 17 18] ##  [19 20 21 22]], shape=(3, 4), dtype=int32) ## ## $outname2 ## tf.Tensor( ## [[21 22 23 24] ##  [25 26 27 28] ##  [29 30 31 32]], shape=(3, 4), dtype=int32) # passing a tuple() is equivalent to passing an unnamed list() input <- tuple(x, x) output <- op_vectorized_map(input, function(.x) {   str(.x)   list(.x[[1]] + 10) }) ## List of 2 ##  $ :<tf.Tensor 'loop_body/GatherV2:0' shape=(4) dtype=int32> ##  $ :<tf.Tensor 'loop_body/GatherV2_1:0' shape=(4) dtype=int32> output ## [[1]] ## tf.Tensor( ## [[11 12 13 14] ##  [15 16 17 18] ##  [19 20 21 22]], shape=(3, 4), dtype=int32)"},{"path":"https://keras3.posit.co/dev/reference/op_vectorized_map.html","id":"debugging-f","dir":"Reference","previous_headings":"","what":"Debugging f","title":"Parallel map of function f on the first axis of tensor(s) elements. — op_vectorized_map","text":"Even eager contexts, op_vectorized_map() may trace f. case, want eagerly debug f (e.g., browser()), can swap manual (slow) implementation op_vectorized_map(). Note example debug implementation handle edge cases op_vectorized_map(), particular, f returns structure multiple tensors.","code":"op_vectorized_map_debug <- function(elements, fn) {    if (!is.list(elements)) {     # `elements` is a single tensor     batch_size <- op_shape(elements)[[1]]     out <- elements |>       op_split(batch_size) |>       lapply(fn) |>       op_stack()     return(out)   }    # `elements` is a list of tensors   batch_size <- elements[[1]] |> op_shape() |> _[[1]]   elements |>     lapply(\\(e) op_split(e, batch_size)) |>     zip_lists() |>     lapply(fn) |>     op_stack()  }"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_view_as_complex.html","id":null,"dir":"Reference","previous_headings":"","what":"Convert a real tensor with two channels into a complex tensor. — op_view_as_complex","title":"Convert a real tensor with two channels into a complex tensor. — op_view_as_complex","text":"Expects real-valued tensor whose last dimension size 2, holding real imaginary parts. Returns corresponding complex tensor last dimension removed.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_view_as_complex.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Convert a real tensor with two channels into a complex tensor. — op_view_as_complex","text":"","code":"op_view_as_complex(x)"},{"path":"https://keras3.posit.co/dev/reference/op_view_as_complex.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Convert a real tensor with two channels into a complex tensor. — op_view_as_complex","text":"x Real-valued tensor whose trailing dimension encodes complex components.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_view_as_complex.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Convert a real tensor with two channels into a complex tensor. — op_view_as_complex","text":"complex tensor shape op_shape(x)[-length(op_shape(x))].","code":""},{"path":"https://keras3.posit.co/dev/reference/op_view_as_complex.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Convert a real tensor with two channels into a complex tensor. — op_view_as_complex","text":"","code":"x <- op_array(matrix(c(1, 2, 3, 4), nrow = 2, byrow = TRUE)) op_view_as_complex(x) ## tf.Tensor([1.+2.j 3.+4.j], shape=(2), dtype=complex64)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_view_as_real.html","id":null,"dir":"Reference","previous_headings":"","what":"Convert a complex tensor into a stacked real representation. — op_view_as_real","title":"Convert a complex tensor into a stacked real representation. — op_view_as_real","text":"Produces real-valued tensor last dimension gathers real imaginary parts complex input.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_view_as_real.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Convert a complex tensor into a stacked real representation. — op_view_as_real","text":"","code":"op_view_as_real(x)"},{"path":"https://keras3.posit.co/dev/reference/op_view_as_real.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Convert a complex tensor into a stacked real representation. — op_view_as_real","text":"x Complex-valued tensor converted.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_view_as_real.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Convert a complex tensor into a stacked real representation. — op_view_as_real","text":"real tensor shape c(op_shape(x), 2) containing real imaginary parts.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_view_as_real.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Convert a complex tensor into a stacked real representation. — op_view_as_real","text":"","code":"x <- op_array(matrix(c(1, 2, 3, 4), nrow = 2, byrow = TRUE)) z <- op_view_as_complex(x) op_view_as_real(z) ## tf.Tensor( ## [[1. 2.] ##  [3. 4.]], shape=(2, 2), dtype=float32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_vstack.html","id":null,"dir":"Reference","previous_headings":"","what":"Stack tensors in sequence vertically (row wise). — op_vstack","title":"Stack tensors in sequence vertically (row wise). — op_vstack","text":"Stack tensors sequence vertically (row wise).","code":""},{"path":"https://keras3.posit.co/dev/reference/op_vstack.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Stack tensors in sequence vertically (row wise). — op_vstack","text":"","code":"op_vstack(xs, ...)"},{"path":"https://keras3.posit.co/dev/reference/op_vstack.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Stack tensors in sequence vertically (row wise). — op_vstack","text":"xs, ... Sequence tensors.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_vstack.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Stack tensors in sequence vertically (row wise). — op_vstack","text":"Tensor formed stacking given tensors.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_where.html","id":null,"dir":"Reference","previous_headings":"","what":"Return elements chosen from x1 or x2 depending on condition. — op_where","title":"Return elements chosen from x1 or x2 depending on condition. — op_where","text":"Return elements chosen x1 x2 depending condition.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_where.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Return elements chosen from x1 or x2 depending on condition. — op_where","text":"","code":"op_where(condition, x1 = NULL, x2 = NULL)"},{"path":"https://keras3.posit.co/dev/reference/op_where.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Return elements chosen from x1 or x2 depending on condition. — op_where","text":"condition TRUE, yield x1, otherwise yield x2. x1 Values choose condition TRUE. x2 Values choose condition FALSE.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_where.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Return elements chosen from x1 or x2 depending on condition. — op_where","text":"tensor elements x1 condition TRUE, elements x2 condition FALSE.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_while_loop.html","id":null,"dir":"Reference","previous_headings":"","what":"While loop implementation. — op_while_loop","title":"While loop implementation. — op_while_loop","text":"loop implementation.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_while_loop.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"While loop implementation. — op_while_loop","text":"","code":"op_while_loop(cond, body, loop_vars, maximum_iterations = NULL)"},{"path":"https://keras3.posit.co/dev/reference/op_while_loop.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"While loop implementation. — op_while_loop","text":"cond callable represents termination condition loop. Must accept loop_vars like structure argument. loop_vars tuple unnamed list, element loop_vars passed positionally callable. body callable represents loop body. Must accept loop_vars like structure argument, return update value structure. loop_vars tuple unnamed list, element loop_vars passed positionally callable. loop_vars arbitrary nested structure tensor state persist across loop iterations. maximum_iterations Optional maximum number iterations loop run. provided, cond output -ed additional condition ensuring number iterations executed greater maximum_iterations.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_while_loop.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"While loop implementation. — op_while_loop","text":"list tensors, shape dtype loop_vars.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_while_loop.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"While loop implementation. — op_while_loop","text":"","code":"i <- 0 loop_vars <- list(i)  # cond() must return a scalar bool cond <- function(i) i < 10L  # body must return same shape as loop_vars body <- function(i) list(i + 1L)  op_while_loop(cond, body, loop_vars) ## [[1]] ## tf.Tensor(10.0, shape=(), dtype=float32) x <- 0; y <- 1 cond <- \\(x, y) x < 10 body <- \\(x, y) list(x+1, y+1) op_while_loop(cond, body, list(x, y)) ## [[1]] ## tf.Tensor(10.0, shape=(), dtype=float32) ## ## [[2]] ## tf.Tensor(11.0, shape=(), dtype=float32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_zeros.html","id":null,"dir":"Reference","previous_headings":"","what":"Return a new tensor of given shape and type, filled with zeros. — op_zeros","title":"Return a new tensor of given shape and type, filled with zeros. — op_zeros","text":"Return new tensor given shape type, filled zeros.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_zeros.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Return a new tensor of given shape and type, filled with zeros. — op_zeros","text":"","code":"op_zeros(shape, dtype = NULL)"},{"path":"https://keras3.posit.co/dev/reference/op_zeros.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Return a new tensor of given shape and type, filled with zeros. — op_zeros","text":"shape Shape new tensor. dtype Desired data type tensor.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_zeros.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Return a new tensor of given shape and type, filled with zeros. — op_zeros","text":"Tensor zeros given shape dtype.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/op_zeros_like.html","id":null,"dir":"Reference","previous_headings":"","what":"Return a tensor of zeros with the same shape and type as x. — op_zeros_like","title":"Return a tensor of zeros with the same shape and type as x. — op_zeros_like","text":"Return tensor zeros shape type x.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_zeros_like.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Return a tensor of zeros with the same shape and type as x. — op_zeros_like","text":"","code":"op_zeros_like(x, dtype = NULL)"},{"path":"https://keras3.posit.co/dev/reference/op_zeros_like.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Return a tensor of zeros with the same shape and type as x. — op_zeros_like","text":"x Input tensor. dtype Overrides data type result.","code":""},{"path":"https://keras3.posit.co/dev/reference/op_zeros_like.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Return a tensor of zeros with the same shape and type as x. — op_zeros_like","text":"tensor zeros shape type x.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/optimizer_adadelta.html","id":null,"dir":"Reference","previous_headings":"","what":"Optimizer that implements the Adadelta algorithm. — optimizer_adadelta","title":"Optimizer that implements the Adadelta algorithm. — optimizer_adadelta","text":"Adadelta optimization stochastic gradient descent method based adaptive learning rate per dimension address two drawbacks: continual decay learning rates throughout training. need manually selected global learning rate. Adadelta robust extension Adagrad adapts learning rates based moving window gradient updates, instead accumulating past gradients. way, Adadelta continues learning even many updates done. Compared Adagrad, original version Adadelta set initial learning rate. version, initial learning rate can set, Keras optimizers.","code":""},{"path":"https://keras3.posit.co/dev/reference/optimizer_adadelta.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Optimizer that implements the Adadelta algorithm. — optimizer_adadelta","text":"","code":"optimizer_adadelta(   learning_rate = 0.001,   rho = 0.95,   epsilon = 1e-07,   weight_decay = NULL,   clipnorm = NULL,   clipvalue = NULL,   global_clipnorm = NULL,   use_ema = FALSE,   ema_momentum = 0.99,   ema_overwrite_frequency = NULL,   name = \"adadelta\",   ...,   loss_scale_factor = NULL,   gradient_accumulation_steps = NULL )"},{"path":"https://keras3.posit.co/dev/reference/optimizer_adadelta.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Optimizer that implements the Adadelta algorithm. — optimizer_adadelta","text":"learning_rate float, [LearningRateSchedule()] instance, callable takes arguments returns actual value use. learning rate. Defaults 0.001. Note Adadelta tends benefit higher initial learning rate values compared optimizers. match exact form original paper, use 1.0. rho floating point value. decay rate. Defaults 0.95. epsilon Small floating point value maintaining numerical stability. weight_decay Float. set, weight decay applied. clipnorm Float. set, gradient weight individually clipped norm higher value. clipvalue Float. set, gradient weight clipped higher value. global_clipnorm Float. set, gradient weights clipped global norm higher value. use_ema Boolean, defaults FALSE. TRUE, exponential moving average (EMA) applied. EMA consists computing exponential moving average weights model (weight values change training batch), periodically overwriting weights moving average. ema_momentum Float, defaults 0.99. used use_ema = TRUE. momentum use computing EMA model's weights: new_average = ema_momentum * old_average + (1 - ema_momentum) * current_variable_value. ema_overwrite_frequency Int NULL, defaults NULL. used use_ema = TRUE. Every ema_overwrite_frequency steps iterations, overwrite model variable moving average. NULL, optimizer overwrite model variables middle training, need explicitly overwrite variables end training calling optimizer$finalize_variable_values() (updates model variables -place). using built-fit() training loop, happens automatically last epoch, need anything. name String. name use momentum accumulator weights created optimizer. ... forward/backward compatability. loss_scale_factor Float NULL. float, scale factor multiplied loss computing gradients, inverse scale factor multiplied gradients updating variables. Useful preventing underflow mixed precision training. Alternately, optimizer_loss_scale() automatically set loss scale factor. gradient_accumulation_steps Int NULL. int, model optimizer variables updated every step; instead updated every gradient_accumulation_steps steps, using average value gradients since last update. known \"gradient accumulation\". can useful batch size small, order reduce gradient noise update step. EMA frequency look \"accumulated\" iterations value (optimizer steps // gradient_accumulation_steps). Learning rate schedules look \"real\" iterations value (optimizer steps).","code":""},{"path":"https://keras3.posit.co/dev/reference/optimizer_adadelta.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Optimizer that implements the Adadelta algorithm. — optimizer_adadelta","text":"Optimizer instance","code":""},{"path":"https://keras3.posit.co/dev/reference/optimizer_adadelta.html","id":"reference","dir":"Reference","previous_headings":"","what":"Reference","title":"Optimizer that implements the Adadelta algorithm. — optimizer_adadelta","text":"Zeiler, 2012","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/optimizer_adafactor.html","id":null,"dir":"Reference","previous_headings":"","what":"Optimizer that implements the Adafactor algorithm. — optimizer_adafactor","title":"Optimizer that implements the Adafactor algorithm. — optimizer_adafactor","text":"Adafactor commonly used NLP tasks, advantage taking less memory saves partial information previous gradients. default argument setup based original paper (see reference). gradients dimension > 2, Adafactor optimizer delete last 2 dimensions separately accumulator variables.","code":""},{"path":"https://keras3.posit.co/dev/reference/optimizer_adafactor.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Optimizer that implements the Adafactor algorithm. — optimizer_adafactor","text":"","code":"optimizer_adafactor(   learning_rate = 0.001,   beta_2_decay = -0.8,   epsilon_1 = 1e-30,   epsilon_2 = 0.001,   clip_threshold = 1,   relative_step = TRUE,   weight_decay = NULL,   clipnorm = NULL,   clipvalue = NULL,   global_clipnorm = NULL,   use_ema = FALSE,   ema_momentum = 0.99,   ema_overwrite_frequency = NULL,   name = \"adafactor\",   ...,   loss_scale_factor = NULL,   gradient_accumulation_steps = NULL )"},{"path":"https://keras3.posit.co/dev/reference/optimizer_adafactor.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Optimizer that implements the Adafactor algorithm. — optimizer_adafactor","text":"learning_rate float, LearningRateSchedule() instance, callable takes arguments returns actual value use. learning rate. Defaults 0.001. beta_2_decay float, defaults -0.8. decay rate beta_2. epsilon_1 float, defaults 1e-30. small offset keep denominator away 0. epsilon_2 float, defaults 1e-3. small offset avoid learning rate becoming small time. clip_threshold float, defaults 1.0. Clipping threshold. part Adafactor algorithm, independent clipnorm, clipvalue, global_clipnorm. relative_step bool, defaults TRUE. learning_rate constant relative_step=TRUE, learning rate adjusted based current iterations. default learning rate decay Adafactor. weight_decay Float. set, weight decay applied. clipnorm Float. set, gradient weight individually clipped norm higher value. clipvalue Float. set, gradient weight clipped higher value. global_clipnorm Float. set, gradient weights clipped global norm higher value. use_ema Boolean, defaults FALSE. TRUE, exponential moving average (EMA) applied. EMA consists computing exponential moving average weights model (weight values change training batch), periodically overwriting weights moving average. ema_momentum Float, defaults 0.99. used use_ema = TRUE. momentum use computing EMA model's weights: new_average = ema_momentum * old_average + (1 - ema_momentum) * current_variable_value. ema_overwrite_frequency Int NULL, defaults NULL. used use_ema=TRUE. Every ema_overwrite_frequency steps iterations, overwrite model variable moving average. NULL, optimizer overwrite model variables middle training, need explicitly overwrite variables end training calling optimizer$finalize_variable_values() (updates model variables -place). using built-fit() training loop, happens automatically last epoch, need anything. name String. name use momentum accumulator weights created optimizer. ... forward/backward compatability. loss_scale_factor Float NULL. float, scale factor multiplied loss computing gradients, inverse scale factor multiplied gradients updating variables. Useful preventing underflow mixed precision training. Alternately, optimizer_loss_scale() automatically set loss scale factor. gradient_accumulation_steps Int NULL. int, model optimizer variables updated every step; instead updated every gradient_accumulation_steps steps, using average value gradients since last update. known \"gradient accumulation\". can useful batch size small, order reduce gradient noise update step. EMA frequency look \"accumulated\" iterations value (optimizer steps // gradient_accumulation_steps). Learning rate schedules look \"real\" iterations value (optimizer steps).","code":""},{"path":"https://keras3.posit.co/dev/reference/optimizer_adafactor.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Optimizer that implements the Adafactor algorithm. — optimizer_adafactor","text":"Optimizer instance","code":""},{"path":"https://keras3.posit.co/dev/reference/optimizer_adafactor.html","id":"reference","dir":"Reference","previous_headings":"","what":"Reference","title":"Optimizer that implements the Adafactor algorithm. — optimizer_adafactor","text":"Shazeer, Noam et al., 2018.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/optimizer_adagrad.html","id":null,"dir":"Reference","previous_headings":"","what":"Optimizer that implements the Adagrad algorithm. — optimizer_adagrad","title":"Optimizer that implements the Adagrad algorithm. — optimizer_adagrad","text":"Adagrad optimizer parameter-specific learning rates, adapted relative frequently parameter gets updated training. updates parameter receives, smaller updates.","code":""},{"path":"https://keras3.posit.co/dev/reference/optimizer_adagrad.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Optimizer that implements the Adagrad algorithm. — optimizer_adagrad","text":"","code":"optimizer_adagrad(   learning_rate = 0.001,   initial_accumulator_value = 0.1,   epsilon = 1e-07,   weight_decay = NULL,   clipnorm = NULL,   clipvalue = NULL,   global_clipnorm = NULL,   use_ema = FALSE,   ema_momentum = 0.99,   ema_overwrite_frequency = NULL,   name = \"adagrad\",   ...,   loss_scale_factor = NULL,   gradient_accumulation_steps = NULL )"},{"path":"https://keras3.posit.co/dev/reference/optimizer_adagrad.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Optimizer that implements the Adagrad algorithm. — optimizer_adagrad","text":"learning_rate float, LearningRateSchedule() instance, callable takes arguments returns actual value use. learning rate. Defaults 0.001. Note Adagrad tends benefit higher initial learning rate values compared optimizers. match exact form original paper, use 1.0. initial_accumulator_value Floating point value. Starting value accumulators (per-parameter momentum values). Must non-negative. epsilon Small floating point value maintaining numerical stability. weight_decay Float. set, weight decay applied. clipnorm Float. set, gradient weight individually clipped norm higher value. clipvalue Float. set, gradient weight clipped higher value. global_clipnorm Float. set, gradient weights clipped global norm higher value. use_ema Boolean, defaults FALSE. TRUE, exponential moving average (EMA) applied. EMA consists computing exponential moving average weights model (weight values change training batch), periodically overwriting weights moving average. ema_momentum Float, defaults 0.99. used use_ema=TRUE. momentum use computing EMA model's weights: new_average = ema_momentum * old_average + (1 - ema_momentum) * current_variable_value. ema_overwrite_frequency Int NULL, defaults NULL. used use_ema=TRUE. Every ema_overwrite_frequency steps iterations, overwrite model variable moving average. NULL, optimizer overwrite model variables middle training, need explicitly overwrite variables end training calling optimizer$finalize_variable_values() (updates model variables -place). using built-fit() training loop, happens automatically last epoch, need anything. name String. name use momentum accumulator weights created optimizer. ... forward/backward compatability. loss_scale_factor Float NULL. float, scale factor multiplied loss computing gradients, inverse scale factor multiplied gradients updating variables. Useful preventing underflow mixed precision training. Alternately, optimizer_loss_scale() automatically set loss scale factor. gradient_accumulation_steps Int NULL. int, model optimizer variables updated every step; instead updated every gradient_accumulation_steps steps, using average value gradients since last update. known \"gradient accumulation\". can useful batch size small, order reduce gradient noise update step. EMA frequency look \"accumulated\" iterations value (optimizer steps // gradient_accumulation_steps). Learning rate schedules look \"real\" iterations value (optimizer steps).","code":""},{"path":"https://keras3.posit.co/dev/reference/optimizer_adagrad.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Optimizer that implements the Adagrad algorithm. — optimizer_adagrad","text":"Optimizer instance","code":""},{"path":"https://keras3.posit.co/dev/reference/optimizer_adagrad.html","id":"reference","dir":"Reference","previous_headings":"","what":"Reference","title":"Optimizer that implements the Adagrad algorithm. — optimizer_adagrad","text":"Duchi et al., 2011.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/optimizer_adam.html","id":null,"dir":"Reference","previous_headings":"","what":"Optimizer that implements the Adam algorithm. — optimizer_adam","title":"Optimizer that implements the Adam algorithm. — optimizer_adam","text":"Adam optimization stochastic gradient descent method based adaptive estimation first-order second-order moments. According Kingma et al., 2014, method \"computationally efficient, little memory requirement, invariant diagonal rescaling gradients, well suited problems large terms data/parameters\".","code":""},{"path":"https://keras3.posit.co/dev/reference/optimizer_adam.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Optimizer that implements the Adam algorithm. — optimizer_adam","text":"","code":"optimizer_adam(   learning_rate = 0.001,   beta_1 = 0.9,   beta_2 = 0.999,   epsilon = 1e-07,   amsgrad = FALSE,   weight_decay = NULL,   clipnorm = NULL,   clipvalue = NULL,   global_clipnorm = NULL,   use_ema = FALSE,   ema_momentum = 0.99,   ema_overwrite_frequency = NULL,   name = \"adam\",   ...,   loss_scale_factor = NULL,   gradient_accumulation_steps = NULL )"},{"path":"https://keras3.posit.co/dev/reference/optimizer_adam.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Optimizer that implements the Adam algorithm. — optimizer_adam","text":"learning_rate float, LearningRateSchedule() instance, callable takes arguments returns actual value use. learning rate. Defaults 0.001. beta_1 float value constant float tensor, callable takes arguments returns actual value use. exponential decay rate 1st moment estimates. Defaults 0.9. beta_2 float value constant float tensor, callable takes arguments returns actual value use. exponential decay rate 2nd moment estimates. Defaults 0.999. epsilon small constant numerical stability. epsilon \"epsilon hat\" Kingma Ba paper (formula just Section 2.1), epsilon Algorithm 1 paper. Defaults 1e-7. amsgrad Boolean. Whether apply AMSGrad variant algorithm paper \"Convergence Adam beyond\". Defaults FALSE. weight_decay Float. set, weight decay applied. clipnorm Float. set, gradient weight individually clipped norm higher value. clipvalue Float. set, gradient weight clipped higher value. global_clipnorm Float. set, gradient weights clipped global norm higher value. use_ema Boolean, defaults FALSE. TRUE, exponential moving average (EMA) applied. EMA consists computing exponential moving average weights model (weight values change training batch), periodically overwriting weights moving average. ema_momentum Float, defaults 0.99. used use_ema=TRUE. momentum use computing EMA model's weights: new_average = ema_momentum * old_average + (1 - ema_momentum) * current_variable_value. ema_overwrite_frequency Int NULL, defaults NULL. used use_ema=TRUE. Every ema_overwrite_frequency steps iterations, overwrite model variable moving average. NULL, optimizer overwrite model variables middle training, need explicitly overwrite variables end training calling optimizer$finalize_variable_values() (updates model variables -place). using built-fit() training loop, happens automatically last epoch, need anything. name String. name use momentum accumulator weights created optimizer. ... forward/backward compatability. loss_scale_factor Float NULL. float, scale factor multiplied loss computing gradients, inverse scale factor multiplied gradients updating variables. Useful preventing underflow mixed precision training. Alternately, optimizer_loss_scale() automatically set loss scale factor. gradient_accumulation_steps Int NULL. int, model optimizer variables updated every step; instead updated every gradient_accumulation_steps steps, using average value gradients since last update. known \"gradient accumulation\". can useful batch size small, order reduce gradient noise update step. EMA frequency look \"accumulated\" iterations value (optimizer steps // gradient_accumulation_steps). Learning rate schedules look \"real\" iterations value (optimizer steps).","code":""},{"path":"https://keras3.posit.co/dev/reference/optimizer_adam.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Optimizer that implements the Adam algorithm. — optimizer_adam","text":"Optimizer instance","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/optimizer_adam_w.html","id":null,"dir":"Reference","previous_headings":"","what":"Optimizer that implements the AdamW algorithm. — optimizer_adam_w","title":"Optimizer that implements the AdamW algorithm. — optimizer_adam_w","text":"AdamW optimization stochastic gradient descent method based adaptive estimation first-order second-order moments added method decay weights per techniques discussed paper, 'Decoupled Weight Decay Regularization' Loshchilov, Hutter et al., 2019. According Kingma et al., 2014, underlying Adam method \"computationally efficient, little memory requirement, invariant diagonal rescaling gradients, well suited problems large terms data/parameters\".","code":""},{"path":"https://keras3.posit.co/dev/reference/optimizer_adam_w.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Optimizer that implements the AdamW algorithm. — optimizer_adam_w","text":"","code":"optimizer_adam_w(   learning_rate = 0.001,   weight_decay = 0.004,   beta_1 = 0.9,   beta_2 = 0.999,   epsilon = 1e-07,   amsgrad = FALSE,   clipnorm = NULL,   clipvalue = NULL,   global_clipnorm = NULL,   use_ema = FALSE,   ema_momentum = 0.99,   ema_overwrite_frequency = NULL,   name = \"adamw\",   ...,   loss_scale_factor = NULL,   gradient_accumulation_steps = NULL )"},{"path":"https://keras3.posit.co/dev/reference/optimizer_adam_w.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Optimizer that implements the AdamW algorithm. — optimizer_adam_w","text":"learning_rate float, LearningRateSchedule() instance, callable takes arguments returns actual value use. learning rate. Defaults 0.001. weight_decay Float. set, weight decay applied. beta_1 float value constant float tensor, callable takes arguments returns actual value use. exponential decay rate 1st moment estimates. Defaults 0.9. beta_2 float value constant float tensor, callable takes arguments returns actual value use. exponential decay rate 2nd moment estimates. Defaults 0.999. epsilon small constant numerical stability. epsilon \"epsilon hat\" Kingma Ba paper (formula just Section 2.1), epsilon Algorithm 1 paper. Defaults 1e-7. amsgrad Boolean. Whether apply AMSGrad variant algorithm paper \"Convergence Adam beyond\". Defaults FALSE. clipnorm Float. set, gradient weight individually clipped norm higher value. clipvalue Float. set, gradient weight clipped higher value. global_clipnorm Float. set, gradient weights clipped global norm higher value. use_ema Boolean, defaults FALSE. TRUE, exponential moving average (EMA) applied. EMA consists computing exponential moving average weights model (weight values change training batch), periodically overwriting weights moving average. ema_momentum Float, defaults 0.99. used use_ema=TRUE. momentum use computing EMA model's weights: new_average = ema_momentum * old_average + (1 - ema_momentum) * current_variable_value. ema_overwrite_frequency Int NULL, defaults NULL. used use_ema=TRUE. Every ema_overwrite_frequency steps iterations, overwrite model variable moving average. NULL, optimizer overwrite model variables middle training, need explicitly overwrite variables end training calling optimizer$finalize_variable_values() (updates model variables -place). using built-fit() training loop, happens automatically last epoch, need anything. name String. name use momentum accumulator weights created optimizer. ... forward/backward compatability. loss_scale_factor Float NULL. float, scale factor multiplied loss computing gradients, inverse scale factor multiplied gradients updating variables. Useful preventing underflow mixed precision training. Alternately, optimizer_loss_scale() automatically set loss scale factor. gradient_accumulation_steps Int NULL. int, model optimizer variables updated every step; instead updated every gradient_accumulation_steps steps, using average value gradients since last update. known \"gradient accumulation\". can useful batch size small, order reduce gradient noise update step. EMA frequency look \"accumulated\" iterations value (optimizer steps // gradient_accumulation_steps). Learning rate schedules look \"real\" iterations value (optimizer steps).","code":""},{"path":"https://keras3.posit.co/dev/reference/optimizer_adam_w.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Optimizer that implements the AdamW algorithm. — optimizer_adam_w","text":"Optimizer instance","code":""},{"path":"https://keras3.posit.co/dev/reference/optimizer_adam_w.html","id":"references","dir":"Reference","previous_headings":"","what":"References","title":"Optimizer that implements the AdamW algorithm. — optimizer_adam_w","text":"Loshchilov et al., 2019 Kingma et al., 2014 adam Reddi et al., 2018 amsgrad.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/optimizer_adamax.html","id":null,"dir":"Reference","previous_headings":"","what":"Optimizer that implements the Adamax algorithm. — optimizer_adamax","title":"Optimizer that implements the Adamax algorithm. — optimizer_adamax","text":"Adamax, variant Adam based infinity norm, first-order gradient-based optimization method. Due capability adjusting learning rate based data characteristics, suited learn time-variant process, e.g., speech data dynamically changed noise conditions. Default parameters follow provided paper (see references ). Initialization:   update rule parameter w gradient g described end section 7.1 paper (see reference section):","code":"m <- 0  # Initialize initial 1st moment vector u <- 0  # Initialize the exponentially weighted infinity norm t <- 0  # Initialize timestep t <-  t + 1 m <- beta1 * m + (1 - beta) * g u <- max(beta2 * u, abs(g)) current_lr <- learning_rate / (1 - beta1 ** t) w <- w - current_lr * m / (u + epsilon)"},{"path":"https://keras3.posit.co/dev/reference/optimizer_adamax.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Optimizer that implements the Adamax algorithm. — optimizer_adamax","text":"","code":"optimizer_adamax(   learning_rate = 0.001,   beta_1 = 0.9,   beta_2 = 0.999,   epsilon = 1e-07,   weight_decay = NULL,   clipnorm = NULL,   clipvalue = NULL,   global_clipnorm = NULL,   use_ema = FALSE,   ema_momentum = 0.99,   ema_overwrite_frequency = NULL,   name = \"adamax\",   ...,   loss_scale_factor = NULL,   gradient_accumulation_steps = NULL )"},{"path":"https://keras3.posit.co/dev/reference/optimizer_adamax.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Optimizer that implements the Adamax algorithm. — optimizer_adamax","text":"learning_rate float, LearningRateSchedule() instance, callable takes arguments returns actual value use. learning rate. Defaults 0.001. beta_1 float value constant float tensor. exponential decay rate 1st moment estimates. beta_2 float value constant float tensor. exponential decay rate exponentially weighted infinity norm. epsilon small constant numerical stability. name: String. name use momentum accumulator weights created optimizer. weight_decay Float. set, weight decay applied. clipnorm Float. set, gradient weight individually clipped norm higher value. clipvalue Float. set, gradient weight clipped higher value. global_clipnorm Float. set, gradient weights clipped global norm higher value. use_ema Boolean, defaults FALSE. TRUE, exponential moving average (EMA) applied. EMA consists computing exponential moving average weights model (weight values change training batch), periodically overwriting weights moving average. ema_momentum Float, defaults 0.99. used use_ema=TRUE. momentum use computing EMA model's weights: new_average = ema_momentum * old_average + (1 - ema_momentum) * current_variable_value. ema_overwrite_frequency Int NULL, defaults NULL. used use_ema=TRUE. Every ema_overwrite_frequency steps iterations, overwrite model variable moving average. NULL, optimizer overwrite model variables middle training, need explicitly overwrite variables end training calling optimizer$finalize_variable_values() (updates model variables -place). using built-fit() training loop, happens automatically last epoch, need anything. name String, name object ... forward/backward compatability. loss_scale_factor Float NULL. float, scale factor multiplied loss computing gradients, inverse scale factor multiplied gradients updating variables. Useful preventing underflow mixed precision training. Alternately, optimizer_loss_scale() automatically set loss scale factor. gradient_accumulation_steps Int NULL. int, model optimizer variables updated every step; instead updated every gradient_accumulation_steps steps, using average value gradients since last update. known \"gradient accumulation\". can useful batch size small, order reduce gradient noise update step. EMA frequency look \"accumulated\" iterations value (optimizer steps // gradient_accumulation_steps). Learning rate schedules look \"real\" iterations value (optimizer steps).","code":""},{"path":"https://keras3.posit.co/dev/reference/optimizer_adamax.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Optimizer that implements the Adamax algorithm. — optimizer_adamax","text":"Optimizer instance","code":""},{"path":"https://keras3.posit.co/dev/reference/optimizer_adamax.html","id":"reference","dir":"Reference","previous_headings":"","what":"Reference","title":"Optimizer that implements the Adamax algorithm. — optimizer_adamax","text":"Kingma et al., 2014","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/optimizer_ftrl.html","id":null,"dir":"Reference","previous_headings":"","what":"Optimizer that implements the FTRL algorithm. — optimizer_ftrl","title":"Optimizer that implements the FTRL algorithm. — optimizer_ftrl","text":"\"Follow Regularized Leader\" (FTRL) optimization algorithm developed Google click-rate prediction early 2010s. suitable shallow models large sparse feature spaces. algorithm described McMahan et al., 2013. Keras version support online L2 regularization (L2 regularization described paper ) shrinkage-type L2 regularization (addition L2 penalty loss function). Initialization:   Update rule one variable w:   Notation: lr learning rate g gradient variable lambda_1 L1 regularization strength lambda_2 L2 regularization strength lr_power power scale n. Check documentation l2_shrinkage_regularization_strength parameter details shrinkage enabled, case gradient replaced gradient shrinkage.","code":"n <- 0 sigma <- 0 z <- 0 prev_n <- n n <- n + g^2 sigma <- (n^(-lr_power) - prev_n^(-lr_power)) / lr z <- z + g - sigma * w if (abs(z) < lambda_1) {   w <- 0 } else {   w <- (sgn(z) * lambda_1 - z) / ((beta + sqrt(n)) / alpha + lambda_2) }"},{"path":"https://keras3.posit.co/dev/reference/optimizer_ftrl.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Optimizer that implements the FTRL algorithm. — optimizer_ftrl","text":"","code":"optimizer_ftrl(   learning_rate = 0.001,   learning_rate_power = -0.5,   initial_accumulator_value = 0.1,   l1_regularization_strength = 0,   l2_regularization_strength = 0,   l2_shrinkage_regularization_strength = 0,   beta = 0,   weight_decay = NULL,   clipnorm = NULL,   clipvalue = NULL,   global_clipnorm = NULL,   use_ema = FALSE,   ema_momentum = 0.99,   ema_overwrite_frequency = NULL,   name = \"ftrl\",   ...,   loss_scale_factor = NULL,   gradient_accumulation_steps = NULL )"},{"path":"https://keras3.posit.co/dev/reference/optimizer_ftrl.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Optimizer that implements the FTRL algorithm. — optimizer_ftrl","text":"learning_rate float, LearningRateSchedule() instance, callable takes arguments returns actual value use. learning rate. Defaults 0.001. learning_rate_power float value, must less equal zero. Controls learning rate decreases training. Use zero fixed learning rate. initial_accumulator_value starting value accumulators. zero positive values allowed. l1_regularization_strength float value, must greater equal zero. Defaults 0.0. l2_regularization_strength float value, must greater equal zero. Defaults 0.0. l2_shrinkage_regularization_strength float value, must greater equal zero. differs L2 L2 stabilization penalty, whereas L2 shrinkage magnitude penalty. input sparse shrinkage happen active weights. beta float value, representing beta value paper. Defaults 0.0. weight_decay Float. set, weight decay applied. clipnorm Float. set, gradient weight individually clipped norm higher value. clipvalue Float. set, gradient weight clipped higher value. global_clipnorm Float. set, gradient weights clipped global norm higher value. use_ema Boolean, defaults FALSE. TRUE, exponential moving average (EMA) applied. EMA consists computing exponential moving average weights model (weight values change training batch), periodically overwriting weights moving average. ema_momentum Float, defaults 0.99. used use_ema=TRUE. momentum use computing EMA model's weights: new_average = ema_momentum * old_average + (1 - ema_momentum) * current_variable_value. ema_overwrite_frequency Int NULL, defaults NULL. used use_ema=TRUE. Every ema_overwrite_frequency steps iterations, overwrite model variable moving average. NULL, optimizer overwrite model variables middle training, need explicitly overwrite variables end training calling optimizer$finalize_variable_values() (updates model variables -place). using built-fit() training loop, happens automatically last epoch, need anything. name String. name use momentum accumulator weights created optimizer. ... forward/backward compatability. loss_scale_factor Float NULL. float, scale factor multiplied loss computing gradients, inverse scale factor multiplied gradients updating variables. Useful preventing underflow mixed precision training. Alternately, optimizer_loss_scale automatically set loss scale factor. gradient_accumulation_steps Int NULL. int, model optimizer variables updated every step; instead updated every gradient_accumulation_steps steps, using average value gradients since last update. known \"gradient accumulation\". can useful batch size small, order reduce gradient noise update step. EMA frequency look \"accumulated\" iterations value (optimizer steps // gradient_accumulation_steps). Learning rate schedules look \"real\" iterations value (optimizer steps).","code":""},{"path":"https://keras3.posit.co/dev/reference/optimizer_ftrl.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Optimizer that implements the FTRL algorithm. — optimizer_ftrl","text":"Optimizer instance","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/optimizer_lamb.html","id":null,"dir":"Reference","previous_headings":"","what":"Optimizer that implements the Lamb algorithm. — optimizer_lamb","title":"Optimizer that implements the Lamb algorithm. — optimizer_lamb","text":"Lamb stochastic gradient descent method uses layer-wise adaptive moments adjusts learning rate parameter based ratio norm weight norm gradient helps stabilize training process improves convergence especially large batch sizes.","code":""},{"path":"https://keras3.posit.co/dev/reference/optimizer_lamb.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Optimizer that implements the Lamb algorithm. — optimizer_lamb","text":"","code":"optimizer_lamb(   learning_rate = 0.001,   beta_1 = 0.9,   beta_2 = 0.999,   epsilon = 1e-07,   weight_decay = NULL,   clipnorm = NULL,   clipvalue = NULL,   global_clipnorm = NULL,   use_ema = FALSE,   ema_momentum = 0.99,   ema_overwrite_frequency = NULL,   loss_scale_factor = NULL,   gradient_accumulation_steps = NULL,   name = \"lamb\",   ... )"},{"path":"https://keras3.posit.co/dev/reference/optimizer_lamb.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Optimizer that implements the Lamb algorithm. — optimizer_lamb","text":"learning_rate float, LearningRateSchedule() instance, callable takes arguments returns actual value use. learning rate. Defaults 0.001. beta_1 float value constant float tensor, callable takes arguments returns actual value use. exponential decay rate 1st moment estimates. Defaults 0.9. beta_2 float value constant float tensor, callable takes arguments returns actual value use. exponential decay rate 2nd moment estimates. Defaults 0.999. epsilon small constant numerical stability. Defaults 1e-7. weight_decay Float. set, weight decay applied. clipnorm Float. set, gradient weight individually clipped norm higher value. clipvalue Float. set, gradient weight clipped higher value. global_clipnorm Float. set, gradient weights clipped global norm higher value. use_ema Boolean, defaults FALSE. TRUE, exponential moving average (EMA) applied. EMA consists computing exponential moving average weights model (weight values change training batch), periodically overwriting weights moving average. ema_momentum Float, defaults 0.99. used use_ema = TRUE. momentum use computing EMA model's weights: new_average = ema_momentum * old_average + (1 - ema_momentum) * current_variable_value. ema_overwrite_frequency Int NULL, defaults NULL. used use_ema = TRUE. Every ema_overwrite_frequency steps iterations, overwrite model variable moving average. NULL, optimizer overwrite model variables middle training, need explicitly overwrite variables end training calling optimizer$finalize_variable_values() (updates model variables -place). using built-fit() training loop, happens automatically last epoch, need anything. loss_scale_factor Float NULL. float, scale factor multiplied loss computing gradients, inverse scale factor multiplied gradients updating variables. Useful preventing underflow mixed precision training. Alternately, optimizer_loss_scale() automatically set loss scale factor. gradient_accumulation_steps Int NULL. int, model optimizer variables updated every step; instead updated every gradient_accumulation_steps steps, using average value gradients since last update. known \"gradient accumulation\". can useful batch size small, order reduce gradient noise update step. EMA frequency look \"accumulated\" iterations value (optimizer steps // gradient_accumulation_steps). Learning rate schedules look \"real\" iterations value (optimizer steps). name String. name use momentum accumulator weights created optimizer. ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/optimizer_lamb.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Optimizer that implements the Lamb algorithm. — optimizer_lamb","text":"Optimizer instance","code":""},{"path":"https://keras3.posit.co/dev/reference/optimizer_lamb.html","id":"references","dir":"Reference","previous_headings":"","what":"References","title":"Optimizer that implements the Lamb algorithm. — optimizer_lamb","text":"Yang et al.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/optimizer_lion.html","id":null,"dir":"Reference","previous_headings":"","what":"Optimizer that implements the Lion algorithm. — optimizer_lion","title":"Optimizer that implements the Lion algorithm. — optimizer_lion","text":"Lion optimizer stochastic-gradient-descent method uses sign operator control magnitude update, unlike adaptive optimizers Adam rely second-order moments. makes Lion memory-efficient keeps track momentum. According authors (see reference), performance gain Adam grows batch size. update Lion produced sign operation, resulting larger norm, suitable learning rate Lion typically 3-10x smaller AdamW. weight decay Lion turn 3-10x larger AdamW maintain similar strength (lr * wd).","code":""},{"path":"https://keras3.posit.co/dev/reference/optimizer_lion.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Optimizer that implements the Lion algorithm. — optimizer_lion","text":"","code":"optimizer_lion(   learning_rate = 0.001,   beta_1 = 0.9,   beta_2 = 0.99,   weight_decay = NULL,   clipnorm = NULL,   clipvalue = NULL,   global_clipnorm = NULL,   use_ema = FALSE,   ema_momentum = 0.99,   ema_overwrite_frequency = NULL,   name = \"lion\",   ...,   loss_scale_factor = NULL,   gradient_accumulation_steps = NULL )"},{"path":"https://keras3.posit.co/dev/reference/optimizer_lion.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Optimizer that implements the Lion algorithm. — optimizer_lion","text":"learning_rate float, LearningRateSchedule() instance, callable takes arguments returns actual value use. learning rate. Defaults 0.001. beta_1 float value constant float tensor, callable takes arguments returns actual value use. rate combine current gradient 1st moment estimate. Defaults 0.9. beta_2 float value constant float tensor, callable takes arguments returns actual value use. exponential decay rate 1st moment estimate. Defaults 0.99. weight_decay Float. set, weight decay applied. clipnorm Float. set, gradient weight individually clipped norm higher value. clipvalue Float. set, gradient weight clipped higher value. global_clipnorm Float. set, gradient weights clipped global norm higher value. use_ema Boolean, defaults FALSE. TRUE, exponential moving average (EMA) applied. EMA consists computing exponential moving average weights model (weight values change training batch), periodically overwriting weights moving average. ema_momentum Float, defaults 0.99. used use_ema=TRUE. momentum use computing EMA model's weights: new_average = ema_momentum * old_average + (1 - ema_momentum) * current_variable_value. ema_overwrite_frequency Int NULL, defaults NULL. used use_ema=TRUE. Every ema_overwrite_frequency steps iterations, overwrite model variable moving average. NULL, optimizer overwrite model variables middle training, need explicitly overwrite variables end training calling optimizer$finalize_variable_values() (updates model variables -place). using built-fit() training loop, happens automatically last epoch, need anything. name String. name use momentum accumulator weights created optimizer. ... forward/backward compatability. loss_scale_factor Float NULL. float, scale factor multiplied loss computing gradients, inverse scale factor multiplied gradients updating variables. Useful preventing underflow mixed precision training. Alternately, optimizer_loss_scale() automatically set loss scale factor. gradient_accumulation_steps Int NULL. int, model optimizer variables updated every step; instead updated every gradient_accumulation_steps steps, using average value gradients since last update. known \"gradient accumulation\". can useful batch size small, order reduce gradient noise update step. EMA frequency look \"accumulated\" iterations value (optimizer steps // gradient_accumulation_steps). Learning rate schedules look \"real\" iterations value (optimizer steps).","code":""},{"path":"https://keras3.posit.co/dev/reference/optimizer_lion.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Optimizer that implements the Lion algorithm. — optimizer_lion","text":"Optimizer instance","code":""},{"path":"https://keras3.posit.co/dev/reference/optimizer_lion.html","id":"references","dir":"Reference","previous_headings":"","what":"References","title":"Optimizer that implements the Lion algorithm. — optimizer_lion","text":"Chen et al., 2023 Authors' implementation","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/optimizer_loss_scale.html","id":null,"dir":"Reference","previous_headings":"","what":"An optimizer that dynamically scales the loss to prevent underflow. — optimizer_loss_scale","title":"An optimizer that dynamically scales the loss to prevent underflow. — optimizer_loss_scale","text":"Loss scaling technique prevent numeric underflow intermediate gradients float16 used. prevent underflow, loss multiplied (\"scaled\") certain factor called \"loss scale\", causes intermediate gradients scaled loss scale well. final gradients divided (\"unscaled\") loss scale bring back original value. LossScaleOptimizer wraps another optimizer applies dynamic loss scaling . loss scale dynamically updated time follows: train step, nonfinite gradient encountered, loss scale halved, train step skipped. dynamic_growth_steps occurred since last time loss scale updated, nonfinite gradients occurred, loss scale doubled.","code":""},{"path":"https://keras3.posit.co/dev/reference/optimizer_loss_scale.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"An optimizer that dynamically scales the loss to prevent underflow. — optimizer_loss_scale","text":"","code":"optimizer_loss_scale(   inner_optimizer,   initial_scale = 32768,   dynamic_growth_steps = 2000L,   ...,   name = NULL,   weight_decay = NULL,   clipnorm = NULL,   clipvalue = NULL,   global_clipnorm = NULL,   use_ema = NULL,   ema_momentum = NULL,   ema_overwrite_frequency = NULL,   loss_scale_factor = NULL,   gradient_accumulation_steps = NULL )"},{"path":"https://keras3.posit.co/dev/reference/optimizer_loss_scale.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"An optimizer that dynamically scales the loss to prevent underflow. — optimizer_loss_scale","text":"inner_optimizer keras Optimizer instance wrap. initial_scale Float. initial loss scale. scale updated training. recommended high number, loss scale high gets lowered far quickly loss scale low gets raised. dynamic_growth_steps Int. often update scale upwards. every dynamic_growth_steps steps finite gradients, loss scale doubled. ... forward/backward compatability. name String. name use momentum accumulator weights created optimizer. weight_decay Float. set, weight decay applied. clipnorm Float. set, gradient weight individually clipped norm higher value. clipvalue Float. set, gradient weight clipped higher value. global_clipnorm Float. set, gradient weights clipped global norm higher value. use_ema Boolean, defaults FALSE. TRUE, exponential moving average (EMA) applied. EMA consists computing exponential moving average weights model (weight values change training batch), periodically overwriting weights moving average. ema_momentum Float, defaults 0.99. used use_ema=TRUE. momentum use computing EMA model's weights: new_average = ema_momentum * old_average + (1 - ema_momentum) * current_variable_value. ema_overwrite_frequency Int NULL, defaults NULL. used use_ema=TRUE. Every ema_overwrite_frequency steps iterations, overwrite model variable moving average. NULL, optimizer overwrite model variables middle training, need explicitly overwrite variables end training calling optimizer$finalize_variable_values() (updates model variables -place). using built-fit() training loop, happens automatically last epoch, need anything. loss_scale_factor Float NULL. float, scale factor multiplied loss computing gradients, inverse scale factor multiplied gradients updating variables. Useful preventing underflow mixed precision training. Alternately, optimizer_loss_scale() automatically set loss scale factor. gradient_accumulation_steps Int NULL. int, model optimizer variables updated every step; instead updated every gradient_accumulation_steps steps, using average value gradients since last update. known \"gradient accumulation\". can useful batch size small, order reduce gradient noise update step. EMA frequency look \"accumulated\" iterations value (optimizer steps // gradient_accumulation_steps). Learning rate schedules look \"real\" iterations value (optimizer steps).","code":""},{"path":"https://keras3.posit.co/dev/reference/optimizer_loss_scale.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"An optimizer that dynamically scales the loss to prevent underflow. — optimizer_loss_scale","text":"Optimizer instance","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/optimizer_muon.html","id":null,"dir":"Reference","previous_headings":"","what":"Optimizer that implements the Muon algorithm. — optimizer_muon","title":"Optimizer that implements the Muon algorithm. — optimizer_muon","text":"Note optimizer used following layers: Embedding layer Final output fully connected layer 0- 1-D variables optimized using AdamW. Muon optimizer can use Muon update step AdamW update step based following: variable 2D, 3D 4D, AdamW step used. configurable. argument exclude_embeddings (defaults TRUE) set TRUE, AdamW step used. variable name matches expression listed argument exclude_layers (list), AdamW step used. variable uses Muon step. Typically, need pass name densely-connected output layer exclude_layers, e.g. exclude_layers = \"output_dense\".","code":""},{"path":"https://keras3.posit.co/dev/reference/optimizer_muon.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Optimizer that implements the Muon algorithm. — optimizer_muon","text":"","code":"optimizer_muon(   learning_rate = 0.001,   adam_beta_1 = 0.9,   adam_beta_2 = 0.999,   epsilon = 1e-07,   weight_decay = 0.1,   clipnorm = NULL,   clipvalue = NULL,   global_clipnorm = NULL,   use_ema = FALSE,   ema_momentum = 0.99,   ema_overwrite_frequency = NULL,   loss_scale_factor = NULL,   gradient_accumulation_steps = NULL,   name = \"muon\",   exclude_layers = NULL,   exclude_embeddings = TRUE,   muon_a = 3.4445,   muon_b = -4.775,   muon_c = 2.0315,   adam_lr_ratio = 0.1,   momentum = 0.95,   ns_steps = 6L,   nesterov = TRUE,   ... )"},{"path":"https://keras3.posit.co/dev/reference/optimizer_muon.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Optimizer that implements the Muon algorithm. — optimizer_muon","text":"learning_rate float, LearningRateSchedule() instance, callable takes arguments returns actual value use. learning rate. Defaults 0.001. adam_beta_1 float value constant float tensor, callable takes arguments returns actual value use. exponential decay rate 1st moment estimates. Defaults 0.9. adam_beta_2 float value constant float tensor, ora callable takes arguments returns actual value use. exponential decay rate 2nd moment estimates. Defaults 0.999. epsilon small constant numerical stability. \"epsilon hat\" Kingma Ba paper (formula just Section 2.1), epsilon Algorithm 1 paper. used AdamW. Defaults 1e-7. weight_decay Float. set, weight decay applied. clipnorm Float. set, gradient weight individually clipped norm higher value. clipvalue Float. set, gradient weight clipped higher value. global_clipnorm Float. set, gradient weights clipped global norm higher value. use_ema Boolean, defaults FALSE. TRUE, exponential moving average (EMA) applied. EMA consists computing exponential moving average weights model (weight values change training batch), periodically overwriting weights moving average. ema_momentum Float, defaults 0.99. used use_ema = TRUE. momentum use computing EMA model's weights: new_average = ema_momentum * old_average + (1 - ema_momentum) * current_variable_value. ema_overwrite_frequency Int NULL, defaults NULL. used use_ema = TRUE. Every ema_overwrite_frequency steps iterations, overwrite model variable moving average. NULL, optimizer overwrite model variables middle training, need explicitly overwrite variables end training calling optimizer$finalize_variable_values() (updates model variables -place). using built-fit() training loop, happens automatically last epoch, need anything. loss_scale_factor Float NULL. float, scale factor multiplied loss computing gradients, inverse scale factor multiplied gradients updating variables. Useful preventing underflow mixed precision training. Alternately, optimizer_loss_scale() automatically set loss scale factor. gradient_accumulation_steps Int NULL. int, model optimizer variables updated every step; instead updated every gradient_accumulation_steps steps, using average value gradients since last update. known \"gradient accumulation\". can useful batch size small, order reduce gradient noise update step. EMA frequency look \"accumulated\" iterations value (optimizer steps // gradient_accumulation_steps). Learning rate schedules look \"real\" iterations value (optimizer steps). name String, name object exclude_layers List strings, keywords layer names exclude. layers keywords path use AdamW. exclude_embeddings Boolean value. TRUE, embedding layers use AdamW. muon_a Float, parameter muon algorithm. recommended use default value. muon_b Float, parameter b muon algorithm. recommended use default value. muon_c Float, parameter c muon algorithm. recommended use default value. adam_lr_ratio Float, ratio learning rate using Adam main learning rate. recommended set 0.1. momentum Float, momentum used internal SGD. ns_steps Integer, number Newton-Schulz iterations run. nesterov Boolean, whether use Nesterov-style momentum. ... forward/backward compatibility.","code":""},{"path":"https://keras3.posit.co/dev/reference/optimizer_muon.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Optimizer that implements the Muon algorithm. — optimizer_muon","text":"Optimizer instance","code":""},{"path":"https://keras3.posit.co/dev/reference/optimizer_muon.html","id":"references","dir":"Reference","previous_headings":"","what":"References","title":"Optimizer that implements the Muon algorithm. — optimizer_muon","text":"Original implementation Liu et al, 2025","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/optimizer_nadam.html","id":null,"dir":"Reference","previous_headings":"","what":"Optimizer that implements the Nadam algorithm. — optimizer_nadam","title":"Optimizer that implements the Nadam algorithm. — optimizer_nadam","text":"Much like Adam essentially RMSprop momentum, Nadam Adam Nesterov momentum.","code":""},{"path":"https://keras3.posit.co/dev/reference/optimizer_nadam.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Optimizer that implements the Nadam algorithm. — optimizer_nadam","text":"","code":"optimizer_nadam(   learning_rate = 0.001,   beta_1 = 0.9,   beta_2 = 0.999,   epsilon = 1e-07,   weight_decay = NULL,   clipnorm = NULL,   clipvalue = NULL,   global_clipnorm = NULL,   use_ema = FALSE,   ema_momentum = 0.99,   ema_overwrite_frequency = NULL,   name = \"nadam\",   ...,   loss_scale_factor = NULL,   gradient_accumulation_steps = NULL )"},{"path":"https://keras3.posit.co/dev/reference/optimizer_nadam.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Optimizer that implements the Nadam algorithm. — optimizer_nadam","text":"learning_rate float, LearningRateSchedule() instance, callable takes arguments returns actual value use. learning rate. Defaults 0.001. beta_1 float value constant float tensor, callable takes arguments returns actual value use. exponential decay rate 1st moment estimates. Defaults 0.9. beta_2 float value constant float tensor, callable takes arguments returns actual value use. exponential decay rate 2nd moment estimates. Defaults 0.999. epsilon small constant numerical stability. epsilon \"epsilon hat\" Kingma Ba paper (formula just Section 2.1), epsilon Algorithm 1 paper. Defaults 1e-7. weight_decay Float. set, weight decay applied. clipnorm Float. set, gradient weight individually clipped norm higher value. clipvalue Float. set, gradient weight clipped higher value. global_clipnorm Float. set, gradient weights clipped global norm higher value. use_ema Boolean, defaults FALSE. TRUE, exponential moving average (EMA) applied. EMA consists computing exponential moving average weights model (weight values change training batch), periodically overwriting weights moving average. ema_momentum Float, defaults 0.99. used use_ema=TRUE. momentum use computing EMA model's weights: new_average = ema_momentum * old_average + (1 - ema_momentum) * current_variable_value. ema_overwrite_frequency Int NULL, defaults NULL. used use_ema=TRUE. Every ema_overwrite_frequency steps iterations, overwrite model variable moving average. NULL, optimizer overwrite model variables middle training, need explicitly overwrite variables end training calling optimizer$finalize_variable_values() (updates model variables -place). using built-fit() training loop, happens automatically last epoch, need anything. name String. name use momentum accumulator weights created optimizer. ... forward/backward compatability. loss_scale_factor Float NULL. float, scale factor multiplied loss computing gradients, inverse scale factor multiplied gradients updating variables. Useful preventing underflow mixed precision training. Alternately, optimizer_loss_scale() automatically set loss scale factor. gradient_accumulation_steps Int NULL. int, model optimizer variables updated every step; instead updated every gradient_accumulation_steps steps, using average value gradients since last update. known \"gradient accumulation\". can useful batch size small, order reduce gradient noise update step. EMA frequency look \"accumulated\" iterations value (optimizer steps // gradient_accumulation_steps). Learning rate schedules look \"real\" iterations value (optimizer steps).","code":""},{"path":"https://keras3.posit.co/dev/reference/optimizer_nadam.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Optimizer that implements the Nadam algorithm. — optimizer_nadam","text":"Optimizer instance","code":""},{"path":"https://keras3.posit.co/dev/reference/optimizer_nadam.html","id":"reference","dir":"Reference","previous_headings":"","what":"Reference","title":"Optimizer that implements the Nadam algorithm. — optimizer_nadam","text":"Dozat, 2015.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/optimizer_rmsprop.html","id":null,"dir":"Reference","previous_headings":"","what":"Optimizer that implements the RMSprop algorithm. — optimizer_rmsprop","title":"Optimizer that implements the RMSprop algorithm. — optimizer_rmsprop","text":"gist RMSprop : Maintain moving (discounted) average square gradients Divide gradient root average implementation RMSprop uses plain momentum, Nesterov momentum. centered version additionally maintains moving average gradients, uses average estimate variance.","code":""},{"path":"https://keras3.posit.co/dev/reference/optimizer_rmsprop.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Optimizer that implements the RMSprop algorithm. — optimizer_rmsprop","text":"","code":"optimizer_rmsprop(   learning_rate = 0.001,   rho = 0.9,   momentum = 0,   epsilon = 1e-07,   centered = FALSE,   weight_decay = NULL,   clipnorm = NULL,   clipvalue = NULL,   global_clipnorm = NULL,   use_ema = FALSE,   ema_momentum = 0.99,   ema_overwrite_frequency = NULL,   name = \"rmsprop\",   ...,   loss_scale_factor = NULL,   gradient_accumulation_steps = NULL )"},{"path":"https://keras3.posit.co/dev/reference/optimizer_rmsprop.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Optimizer that implements the RMSprop algorithm. — optimizer_rmsprop","text":"learning_rate float, learning_rate_schedule_* instance, callable takes arguments returns actual value use. learning rate. Defaults 0.001. rho float, defaults 0.9. Discounting factor old gradients. momentum float, defaults 0.0. 0.0., optimizer tracks momentum value, decay rate equals 1 - momentum. epsilon small constant numerical stability. epsilon \"epsilon hat\" Kingma Ba paper (formula just Section 2.1), epsilon Algorithm 1 paper. Defaults 1e-7. centered Boolean. TRUE, gradients normalized estimated variance gradient; FALSE, uncentered second moment. Setting TRUE may help training, slightly expensive terms computation memory. Defaults FALSE. weight_decay Float. set, weight decay applied. clipnorm Float. set, gradient weight individually clipped norm higher value. clipvalue Float. set, gradient weight clipped higher value. global_clipnorm Float. set, gradient weights clipped global norm higher value. use_ema Boolean, defaults FALSE. TRUE, exponential moving average (EMA) applied. EMA consists computing exponential moving average weights model (weight values change training batch), periodically overwriting weights moving average. ema_momentum Float, defaults 0.99. used use_ema=TRUE. momentum use computing EMA model's weights: new_average = ema_momentum * old_average + (1 - ema_momentum) * current_variable_value. ema_overwrite_frequency Int NULL, defaults NULL. used use_ema=TRUE. Every ema_overwrite_frequency steps iterations, overwrite model variable moving average. NULL, optimizer overwrite model variables middle training, need explicitly overwrite variables end training calling optimizer$finalize_variable_values() (updates model variables -place). using built-fit() training loop, happens automatically last epoch, need anything. name String. name use momentum accumulator weights created optimizer. ... forward/backward compatability. loss_scale_factor Float NULL. float, scale factor multiplied loss computing gradients, inverse scale factor multiplied gradients updating variables. Useful preventing underflow mixed precision training. Alternately, optimizer_loss_scale() automatically set loss scale factor. gradient_accumulation_steps Int NULL. int, model optimizer variables updated every step; instead updated every gradient_accumulation_steps steps, using average value gradients since last update. known \"gradient accumulation\". can useful batch size small, order reduce gradient noise update step. EMA frequency look \"accumulated\" iterations value (optimizer steps // gradient_accumulation_steps). Learning rate schedules look \"real\" iterations value (optimizer steps).","code":""},{"path":"https://keras3.posit.co/dev/reference/optimizer_rmsprop.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Optimizer that implements the RMSprop algorithm. — optimizer_rmsprop","text":"Optimizer instance","code":""},{"path":"https://keras3.posit.co/dev/reference/optimizer_rmsprop.html","id":"usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Optimizer that implements the RMSprop algorithm. — optimizer_rmsprop","text":"","code":"opt <- optimizer_rmsprop(learning_rate=0.1)"},{"path":"https://keras3.posit.co/dev/reference/optimizer_rmsprop.html","id":"reference","dir":"Reference","previous_headings":"","what":"Reference","title":"Optimizer that implements the RMSprop algorithm. — optimizer_rmsprop","text":"Hinton, 2012","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/optimizer_sgd.html","id":null,"dir":"Reference","previous_headings":"","what":"Gradient descent (with momentum) optimizer. — optimizer_sgd","title":"Gradient descent (with momentum) optimizer. — optimizer_sgd","text":"Update rule parameter w gradient g momentum 0:   Update rule momentum larger 0:   nesterov=TRUE, rule becomes:","code":"w <- w - learning_rate * g velocity <- momentum * velocity - learning_rate * g w <- w + velocity velocity <- momentum * velocity - learning_rate * g w <- w + momentum * velocity - learning_rate * g"},{"path":"https://keras3.posit.co/dev/reference/optimizer_sgd.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Gradient descent (with momentum) optimizer. — optimizer_sgd","text":"","code":"optimizer_sgd(   learning_rate = 0.01,   momentum = 0,   nesterov = FALSE,   weight_decay = NULL,   clipnorm = NULL,   clipvalue = NULL,   global_clipnorm = NULL,   use_ema = FALSE,   ema_momentum = 0.99,   ema_overwrite_frequency = NULL,   name = \"SGD\",   ...,   loss_scale_factor = NULL,   gradient_accumulation_steps = NULL )"},{"path":"https://keras3.posit.co/dev/reference/optimizer_sgd.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Gradient descent (with momentum) optimizer. — optimizer_sgd","text":"learning_rate float, learning_rate_schedule_* instance, callable takes arguments returns actual value use. learning rate. Defaults 0.01. momentum float hyperparameter >= 0 accelerates gradient descent relevant direction dampens oscillations. 0 vanilla gradient descent. Defaults 0.0. nesterov boolean. Whether apply Nesterov momentum. Defaults FALSE. weight_decay Float. set, weight decay applied. clipnorm Float. set, gradient weight individually clipped norm higher value. clipvalue Float. set, gradient weight clipped higher value. global_clipnorm Float. set, gradient weights clipped global norm higher value. use_ema Boolean, defaults FALSE. TRUE, exponential moving average (EMA) applied. EMA consists computing exponential moving average weights model (weight values change training batch), periodically overwriting weights moving average. ema_momentum Float, defaults 0.99. used use_ema=TRUE. momentum use computing EMA model's weights: new_average = ema_momentum * old_average + (1 - ema_momentum) * current_variable_value. ema_overwrite_frequency Int NULL, defaults NULL. used use_ema=TRUE. Every ema_overwrite_frequency steps iterations, overwrite model variable moving average. NULL, optimizer overwrite model variables middle training, need explicitly overwrite variables end training calling optimizer$finalize_variable_values() (updates model variables -place). using built-fit() training loop, happens automatically last epoch, need anything. name String. name use momentum accumulator weights created optimizer. ... forward/backward compatability. loss_scale_factor Float NULL. float, scale factor multiplied loss computing gradients, inverse scale factor multiplied gradients updating variables. Useful preventing underflow mixed precision training. Alternately, optimizer_loss_scale() automatically set loss scale factor. gradient_accumulation_steps Int NULL. int, model optimizer variables updated every step; instead updated every gradient_accumulation_steps steps, using average value gradients since last update. known \"gradient accumulation\". can useful batch size small, order reduce gradient noise update step. EMA frequency look \"accumulated\" iterations value (optimizer steps // gradient_accumulation_steps). Learning rate schedules look \"real\" iterations value (optimizer steps).","code":""},{"path":"https://keras3.posit.co/dev/reference/optimizer_sgd.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Gradient descent (with momentum) optimizer. — optimizer_sgd","text":"Optimizer instance","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/pad_sequences.html","id":null,"dir":"Reference","previous_headings":"","what":"Pads sequences to the same length. — pad_sequences","title":"Pads sequences to the same length. — pad_sequences","text":"function transforms list (length num_samples) sequences (lists integers) 2D NumPy array shape (num_samples, num_timesteps). num_timesteps either maxlen argument provided, length longest sequence list. Sequences shorter num_timesteps padded value num_timesteps long. Sequences longer num_timesteps truncated fit desired length. position padding truncation happens determined arguments padding truncating, respectively. Pre-padding removing values beginning sequence default.","code":"sequence <- list(c(1), c(2, 3), c(4, 5, 6)) pad_sequences(sequence) ##      [,1] [,2] [,3] ## [1,]    0    0    1 ## [2,]    0    2    3 ## [3,]    4    5    6 pad_sequences(sequence, value=-1) ##      [,1] [,2] [,3] ## [1,]   -1   -1    1 ## [2,]   -1    2    3 ## [3,]    4    5    6 pad_sequences(sequence, padding='post') ##      [,1] [,2] [,3] ## [1,]    1    0    0 ## [2,]    2    3    0 ## [3,]    4    5    6 pad_sequences(sequence, maxlen=2) ##      [,1] [,2] ## [1,]    0    1 ## [2,]    2    3 ## [3,]    5    6"},{"path":"https://keras3.posit.co/dev/reference/pad_sequences.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Pads sequences to the same length. — pad_sequences","text":"","code":"pad_sequences(   sequences,   maxlen = NULL,   dtype = \"int32\",   padding = \"pre\",   truncating = \"pre\",   value = 0 )"},{"path":"https://keras3.posit.co/dev/reference/pad_sequences.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Pads sequences to the same length. — pad_sequences","text":"sequences List sequences (sequence list integers). maxlen Optional Int, maximum length sequences. provided, sequences padded length longest individual sequence. dtype (Optional, defaults \"int32\"). Type output sequences. pad sequences variable length strings, can use object. padding String, \"pre\" \"post\" (optional, defaults \"pre\"): pad either sequence. truncating String, \"pre\" \"post\" (optional, defaults \"pre\"): remove values sequences larger maxlen, either beginning end sequences. value Float String, padding value. (Optional, defaults 0)","code":""},{"path":"https://keras3.posit.co/dev/reference/pad_sequences.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Pads sequences to the same length. — pad_sequences","text":"Array shape (len(sequences), maxlen)","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/pipe.html","id":null,"dir":"Reference","previous_headings":"","what":"Pipe operator — %>%","title":"Pipe operator — %>%","text":"See %>% details.","code":""},{"path":"https://keras3.posit.co/dev/reference/pipe.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Pipe operator — %>%","text":"","code":"lhs %>% rhs"},{"path":"https://keras3.posit.co/dev/reference/pipe.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Pipe operator — %>%","text":"commonly, result calling right hand side left hand side argument: rhs(lhs). See magritter vignette , advanced, usages.","code":""},{"path":"https://keras3.posit.co/dev/reference/plot.keras.src.models.model.Model.html","id":null,"dir":"Reference","previous_headings":"","what":"Plot a Keras model — plot.keras.src.models.model.Model","title":"Plot a Keras model — plot.keras.src.models.model.Model","text":"Plot Keras model","code":""},{"path":"https://keras3.posit.co/dev/reference/plot.keras.src.models.model.Model.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Plot a Keras model — plot.keras.src.models.model.Model","text":"","code":"# S3 method for class 'keras.src.models.model.Model' plot(   x,   show_shapes = FALSE,   show_dtype = FALSE,   show_layer_names = FALSE,   ...,   rankdir = \"TB\",   expand_nested = FALSE,   dpi = 200,   layer_range = NULL,   show_layer_activations = FALSE,   show_trainable = NA,   to_file = NULL )"},{"path":"https://keras3.posit.co/dev/reference/plot.keras.src.models.model.Model.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Plot a Keras model — plot.keras.src.models.model.Model","text":"x Keras model instance show_shapes whether display shape information. show_dtype whether display layer dtypes. show_layer_names whether display layer names. ... passed Python keras.utils.model_to_dot(). Used forward backward compatibility. rankdir string specifying format plot: 'TB' creates vertical plot; 'LR' creates horizontal plot. (argument passed PyDot) expand_nested Whether expand nested models clusters. dpi Dots per inch. Increase value image text appears excessively pixelated. layer_range list containing two character strings, starting layer name ending layer name (inclusive) indicating range layers plot generated. also accepts regex patterns instead exact name. case, start predicate first element matches layer_range[1] end predicate last element matches layer_range[2]. default NULL considers layers model. Note must pass range resultant subgraph must complete. show_layer_activations Display layer activations (layers activation property). show_trainable whether display layer trainable. to_file File name plot image. NULL (default), model drawn default graphics device. Otherwise, file saved.","code":""},{"path":"https://keras3.posit.co/dev/reference/plot.keras.src.models.model.Model.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Plot a Keras model — plot.keras.src.models.model.Model","text":"Nothing, called side effects.","code":""},{"path":"https://keras3.posit.co/dev/reference/plot.keras.src.models.model.Model.html","id":"raises","dir":"Reference","previous_headings":"","what":"Raises","title":"Plot a Keras model — plot.keras.src.models.model.Model","text":"ValueError: plot(model) called model built, unless input_shape =  argument supplied keras_model_sequential().","code":""},{"path":"https://keras3.posit.co/dev/reference/plot.keras.src.models.model.Model.html","id":"requirements","dir":"Reference","previous_headings":"","what":"Requirements","title":"Plot a Keras model — plot.keras.src.models.model.Model","text":"function requires pydot graphviz. pydot default installed install_keras(), installed Keras means, can install pydot directly :   can install graphviz directly : https://graphviz.gitlab.io/download/ Linux platforms, can install graphviz via package manager. example, Ubuntu/Debian can install   macOS can install graphviz using brew:   conda environment, can install graphviz :","code":"reticulate::py_install(\"pydot\", pip = TRUE) sudo apt install graphviz brew install graphviz reticulate::conda_install(packages = \"graphviz\") # Restart the R session after install."},{"path":"https://keras3.posit.co/dev/reference/plot.keras_training_history.html","id":null,"dir":"Reference","previous_headings":"","what":"Plot training history — plot.keras_training_history","title":"Plot training history — plot.keras_training_history","text":"Plots metrics recorded training.","code":""},{"path":"https://keras3.posit.co/dev/reference/plot.keras_training_history.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Plot training history — plot.keras_training_history","text":"","code":"# S3 method for class 'keras_training_history' plot(   x,   y,   metrics = NULL,   method = c(\"auto\", \"ggplot2\", \"base\"),   smooth = getOption(\"keras.plot.history.smooth\", TRUE),   theme_bw = getOption(\"keras.plot.history.theme_bw\", FALSE),   ... )"},{"path":"https://keras3.posit.co/dev/reference/plot.keras_training_history.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Plot training history — plot.keras_training_history","text":"x Training history object returned fit.keras.src.models.model.Model(). y Unused. metrics One metrics plot (e.g. c('loss', 'accuracy')). Defaults plotting captured metrics. method Method use plotting. default \"auto\" use ggplot2 available, otherwise use base graphics. smooth Whether loess smooth added plot, available ggplot2 method. number epochs smaller ten, forced false. theme_bw Use ggplot2::theme_bw() plot history black white. ... Additional parameters pass plot() method.","code":""},{"path":"https://keras3.posit.co/dev/reference/plot.keras_training_history.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Plot training history — plot.keras_training_history","text":"method == \"ggplot2\", ggplot object returned. method == \"base\", function draw graphics device return NULL, invisibly.","code":""},{"path":"https://keras3.posit.co/dev/reference/pop_layer.html","id":null,"dir":"Reference","previous_headings":"","what":"Remove the last layer in a Sequential model — pop_layer","title":"Remove the last layer in a Sequential model — pop_layer","text":"Remove last layer Sequential model","code":""},{"path":"https://keras3.posit.co/dev/reference/pop_layer.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Remove the last layer in a Sequential model — pop_layer","text":"","code":"pop_layer(object, rebuild = TRUE)"},{"path":"https://keras3.posit.co/dev/reference/pop_layer.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Remove the last layer in a Sequential model — pop_layer","text":"object Sequential keras model object rebuild bool. Whether rebuild model removing layer. Defaults TRUE.","code":""},{"path":"https://keras3.posit.co/dev/reference/pop_layer.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Remove the last layer in a Sequential model — pop_layer","text":"removed layer.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/predict.keras.src.models.model.Model.html","id":null,"dir":"Reference","previous_headings":"","what":"Generates output predictions for the input samples. — predict.keras.src.models.model.Model","title":"Generates output predictions for the input samples. — predict.keras.src.models.model.Model","text":"Generates output predictions input samples.","code":""},{"path":"https://keras3.posit.co/dev/reference/predict.keras.src.models.model.Model.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Generates output predictions for the input samples. — predict.keras.src.models.model.Model","text":"","code":"# S3 method for class 'keras.src.models.model.Model' predict(   object,   x,   ...,   batch_size = NULL,   verbose = getOption(\"keras.verbose\", default = \"auto\"),   steps = NULL,   callbacks = NULL )"},{"path":"https://keras3.posit.co/dev/reference/predict.keras.src.models.model.Model.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Generates output predictions for the input samples. — predict.keras.src.models.model.Model","text":"object Keras model object x Input samples. can : array (array-like), list arrays (case model multiple inputs). backend-native tensor, list tensors (case model multiple inputs). TF Dataset. Python generator function. ... forward/backward compatability. batch_size Integer NULL. Number samples per batch computation. unspecified, batch_size default 32. specify batch_size input data x form TF Dataset generator (since generate batches). verbose \"auto\", 0, 1, 2. Verbosity mode. 0 = silent, 1 = progress bar, 2 = one line per epoch. \"auto\" becomes 1 cases, 2 knitr render running distributed training server. Note progress bar particularly useful logged file, verbose=2 recommended running interactively (e.g., production environment). Defaults \"auto\". steps Total number steps (batches samples) draw declaring prediction round finished. steps NULL, predict() run x exhausted. case infinitely repeating dataset, predict() run indefinitely. callbacks List Callback instances. List callbacks apply prediction.","code":""},{"path":"https://keras3.posit.co/dev/reference/predict.keras.src.models.model.Model.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Generates output predictions for the input samples. — predict.keras.src.models.model.Model","text":"R array(s) predictions.","code":""},{"path":"https://keras3.posit.co/dev/reference/predict.keras.src.models.model.Model.html","id":"details","dir":"Reference","previous_headings":"","what":"Details","title":"Generates output predictions for the input samples. — predict.keras.src.models.model.Model","text":"Computation done batches. method designed batch processing large numbers inputs. intended use inside loops iterate data process small numbers inputs time. small numbers inputs fit one batch, directly call model model$call faster execution, e.g., model(x), model(x, training = FALSE) layers BatchNormalization behave differently inference.","code":""},{"path":"https://keras3.posit.co/dev/reference/predict.keras.src.models.model.Model.html","id":"note","dir":"Reference","previous_headings":"","what":"Note","title":"Generates output predictions for the input samples. — predict.keras.src.models.model.Model","text":"See FAQ entry details difference Model methods predict() call().","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/predict_on_batch.html","id":null,"dir":"Reference","previous_headings":"","what":"Returns predictions for a single batch of samples. — predict_on_batch","title":"Returns predictions for a single batch of samples. — predict_on_batch","text":"Returns predictions single batch samples.","code":""},{"path":"https://keras3.posit.co/dev/reference/predict_on_batch.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Returns predictions for a single batch of samples. — predict_on_batch","text":"","code":"predict_on_batch(object, x)"},{"path":"https://keras3.posit.co/dev/reference/predict_on_batch.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Returns predictions for a single batch of samples. — predict_on_batch","text":"object Keras model object x Input data. must array-like.","code":""},{"path":"https://keras3.posit.co/dev/reference/predict_on_batch.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Returns predictions for a single batch of samples. — predict_on_batch","text":"Array(s) predictions.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/process_utils.html","id":null,"dir":"Reference","previous_headings":"","what":"Preprocessing and postprocessing utilities — process_utils","title":"Preprocessing and postprocessing utilities — process_utils","text":"functions used preprocess postprocess inputs outputs Keras applications.","code":""},{"path":"https://keras3.posit.co/dev/reference/process_utils.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Preprocessing and postprocessing utilities — process_utils","text":"","code":"application_preprocess_inputs(model, x, ..., data_format = NULL)  application_decode_predictions(model, preds, top = 5L, ...)"},{"path":"https://keras3.posit.co/dev/reference/process_utils.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Preprocessing and postprocessing utilities — process_utils","text":"model Keras model initialized using application_ function. x batch inputs model. x missing, preprocess_input function appropriate model returned. ... Additional arguments passed preprocessing decoding function. data_format Optional data format image tensor/array. NULL means global setting config_image_data_format() used (unless changed , uses \"channels_last\"). Defaults NULL. preds batch outputs model. top number top predictions return.","code":""},{"path":"https://keras3.posit.co/dev/reference/process_utils.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Preprocessing and postprocessing utilities — process_utils","text":"list decoded predictions case application_decode_predictions(). batch preprocessed inputs case application_preprocess_inputs().","code":""},{"path":"https://keras3.posit.co/dev/reference/process_utils.html","id":"functions","dir":"Reference","previous_headings":"","what":"Functions","title":"Preprocessing and postprocessing utilities — process_utils","text":"application_preprocess_inputs(): Pre-process inputs used model application_decode_predictions(): Decode predictions model","code":""},{"path":"https://keras3.posit.co/dev/reference/process_utils.html","id":"ref-examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Preprocessing and postprocessing utilities — process_utils","text":"","code":"if (FALSE) { # \\dontrun{ model <- application_convnext_tiny()  inputs <- random_normal(c(32, 224, 224, 3)) processed_inputs <- application_preprocess_inputs(model, inputs)  preds <- random_normal(c(32, 1000)) decoded_preds <- application_decode_predictions(model, preds)  } # }"},{"path":"https://keras3.posit.co/dev/reference/quantize_weights.html","id":null,"dir":"Reference","previous_headings":"","what":"Quantize the weights of a model. — quantize_weights","title":"Quantize the weights of a model. — quantize_weights","text":"Note model must built first calling method. quantize_weights() recursively call layer$quantize(mode) layers skipped layer implement function. Currently Dense EinsumDense layers support quantization.","code":""},{"path":"https://keras3.posit.co/dev/reference/quantize_weights.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Quantize the weights of a model. — quantize_weights","text":"","code":"quantize_weights(object, mode, ...)"},{"path":"https://keras3.posit.co/dev/reference/quantize_weights.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Quantize the weights of a model. — quantize_weights","text":"object Keras Model Layer. mode mode quantization. 'int8' supported time. ... Passed object quantization method.","code":""},{"path":"https://keras3.posit.co/dev/reference/quantize_weights.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Quantize the weights of a model. — quantize_weights","text":"model, invisibly. Note just convenience usage |>, model modified -place.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/random_beta.html","id":null,"dir":"Reference","previous_headings":"","what":"Draw samples from a Beta distribution. — random_beta","title":"Draw samples from a Beta distribution. — random_beta","text":"values drawn Beta distribution parametrized alpha beta.","code":""},{"path":"https://keras3.posit.co/dev/reference/random_beta.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Draw samples from a Beta distribution. — random_beta","text":"","code":"random_beta(shape, alpha, beta, dtype = NULL, seed = NULL)"},{"path":"https://keras3.posit.co/dev/reference/random_beta.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Draw samples from a Beta distribution. — random_beta","text":"shape shape random values generate. alpha Float array floats representing first parameter alpha. Must broadcastable beta shape. beta Float array floats representing second parameter beta. Must broadcastable alpha shape. dtype Optional dtype tensor. floating point types supported. specified, config_floatx() used, defaults \"float32\" unless configured otherwise (via config_set_floatx(float_dtype)). seed Optional R integer instance random_seed_generator(). default, seed argument NULL, internal global random_seed_generator() used. seed argument can used ensure deterministic (repeatable) random number generation. Note passing integer seed value produce random values call. generate different random values repeated calls, instance random_seed_generator() must provided seed value. Remark concerning JAX backend: tracing functions JAX backend global random_seed_generator() supported. Therefore, tracing default value seed=NULL produce error, seed argument must provided.","code":""},{"path":"https://keras3.posit.co/dev/reference/random_beta.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Draw samples from a Beta distribution. — random_beta","text":"tensor random values.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/random_binomial.html","id":null,"dir":"Reference","previous_headings":"","what":"Draw samples from a Binomial distribution. — random_binomial","title":"Draw samples from a Binomial distribution. — random_binomial","text":"values drawn Binomial distribution specified trial count probability success.","code":""},{"path":"https://keras3.posit.co/dev/reference/random_binomial.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Draw samples from a Binomial distribution. — random_binomial","text":"","code":"random_binomial(shape, counts, probabilities, dtype = NULL, seed = NULL)"},{"path":"https://keras3.posit.co/dev/reference/random_binomial.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Draw samples from a Binomial distribution. — random_binomial","text":"shape shape random values generate. counts number array numbers representing number trials. must broadcastable probabilities. probabilities float array floats representing probability success individual event. must broadcastable counts. dtype Optional dtype tensor. floating point types supported. specified, config_floatx() used, defaults \"float32\" unless configured otherwise (via config_set_floatx(float_dtype)). seed Optional R integer instance random_seed_generator(). default, seed argument NULL, internal global random_seed_generator() used. seed argument can used ensure deterministic (repeatable) random number generation. Note passing integer seed value produce random values call. generate different random values repeated calls, instance random_seed_generator() must provided seed value. Remark concerning JAX backend: tracing functions JAX backend global random_seed_generator() supported. Therefore, tracing default value seed=NULL produce error, seed argument must provided.","code":""},{"path":"https://keras3.posit.co/dev/reference/random_binomial.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Draw samples from a Binomial distribution. — random_binomial","text":"tensor random values.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/random_categorical.html","id":null,"dir":"Reference","previous_headings":"","what":"Draws samples from a categorical distribution. — random_categorical","title":"Draws samples from a categorical distribution. — random_categorical","text":"function takes input logits, 2-D input tensor shape (batch_size, num_classes). row input represents categorical distribution, column index containing log-probability given class. function output 2-D tensor shape (batch_size, num_samples), row contains samples corresponding row logits. column index contains independent samples drawn input distribution.","code":"x <- matrix(c(100, .1, 99), nrow = 1) random_categorical(x, num_samples = 5, seed = 1234) ## tf.Tensor([[3 1 1 3 3]], shape=(1, 5), dtype=int32) random_categorical(x, num_samples = 5, seed = 1234,                    zero_indexed = TRUE) ## tf.Tensor([[2 0 0 2 2]], shape=(1, 5), dtype=int32) op_take(x, random_categorical(x, num_samples = 5, seed = 1234)) ## tf.Tensor([[ 99. 100. 100.  99.  99.]], shape=(1, 5), dtype=float64) op_take(x, random_categorical(x, num_samples = 5, seed = 1234,                               zero_indexed = TRUE),         zero_indexed = TRUE) ## tf.Tensor([[ 99. 100. 100.  99.  99.]], shape=(1, 5), dtype=float64)"},{"path":"https://keras3.posit.co/dev/reference/random_categorical.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Draws samples from a categorical distribution. — random_categorical","text":"","code":"random_categorical(   logits,   num_samples,   dtype = \"int32\",   seed = NULL,   zero_indexed = FALSE )"},{"path":"https://keras3.posit.co/dev/reference/random_categorical.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Draws samples from a categorical distribution. — random_categorical","text":"logits 2-D Tensor shape (batch_size, num_classes). row define categorical distribution unnormalized log-probabilities classes. num_samples Int, number independent samples draw row input. second dimension output tensor's shape. dtype Optional dtype output tensor. seed Optional R integer instance random_seed_generator(). default, seed argument NULL, internal global random_seed_generator() used. seed argument can used ensure deterministic (repeatable) random number generation. Note passing integer seed value produce random values call. generate different random values repeated calls, instance random_seed_generator() must provided seed value. Remark concerning JAX backend: tracing functions JAX backend global random_seed_generator() supported. Therefore, tracing default value seed=NULL produce error, seed argument must provided. zero_indexed TRUE, returned indices zero-based (0 encodes first position); FALSE (default), returned indices one-based (1 encodes first position).","code":""},{"path":"https://keras3.posit.co/dev/reference/random_categorical.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Draws samples from a categorical distribution. — random_categorical","text":"2-D tensor (batch_size, num_samples).","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/random_dropout.html","id":null,"dir":"Reference","previous_headings":"","what":"Randomly set some values in a tensor to 0. — random_dropout","title":"Randomly set some values in a tensor to 0. — random_dropout","text":"Randomly set portion values tensor 0.","code":""},{"path":"https://keras3.posit.co/dev/reference/random_dropout.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Randomly set some values in a tensor to 0. — random_dropout","text":"","code":"random_dropout(inputs, rate, noise_shape = NULL, seed = NULL)"},{"path":"https://keras3.posit.co/dev/reference/random_dropout.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Randomly set some values in a tensor to 0. — random_dropout","text":"inputs tensor rate numeric noise_shape shape() value seed Initial seed random number generator","code":""},{"path":"https://keras3.posit.co/dev/reference/random_dropout.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Randomly set some values in a tensor to 0. — random_dropout","text":"tensor copy inputs values set 0.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/random_gamma.html","id":null,"dir":"Reference","previous_headings":"","what":"Draw random samples from the Gamma distribution. — random_gamma","title":"Draw random samples from the Gamma distribution. — random_gamma","text":"Draw random samples Gamma distribution.","code":""},{"path":"https://keras3.posit.co/dev/reference/random_gamma.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Draw random samples from the Gamma distribution. — random_gamma","text":"","code":"random_gamma(shape, alpha, dtype = NULL, seed = NULL)"},{"path":"https://keras3.posit.co/dev/reference/random_gamma.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Draw random samples from the Gamma distribution. — random_gamma","text":"shape shape random values generate. alpha Float, parameter distribution. dtype Optional dtype tensor. floating point types supported. specified, config_floatx() used, defaults float32 unless configured otherwise (via config_set_floatx(float_dtype)). seed Optional R integer instance random_seed_generator(). default, seed argument NULL, internal global random_seed_generator() used. seed argument can used ensure deterministic (repeatable) random number generation. Note passing integer seed value produce random values call. generate different random values repeated calls, instance random_seed_generator() must provided seed value. Remark concerning JAX backend: tracing functions JAX backend global random_seed_generator() supported. Therefore, tracing default value seed=NULL produce error, seed argument must provided.","code":""},{"path":"https://keras3.posit.co/dev/reference/random_gamma.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Draw random samples from the Gamma distribution. — random_gamma","text":"tensor random values.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/random_integer.html","id":null,"dir":"Reference","previous_headings":"","what":"Draw random integers from a uniform distribution. — random_integer","title":"Draw random integers from a uniform distribution. — random_integer","text":"generated values follow uniform distribution range [minval, maxval). lower bound minval included range, upper bound maxval excluded. dtype must integer type.","code":""},{"path":"https://keras3.posit.co/dev/reference/random_integer.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Draw random integers from a uniform distribution. — random_integer","text":"","code":"random_integer(shape, minval, maxval, dtype = \"int32\", seed = NULL)"},{"path":"https://keras3.posit.co/dev/reference/random_integer.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Draw random integers from a uniform distribution. — random_integer","text":"shape shape random values generate. minval integer, lower bound range random values generate (inclusive). maxval integer, upper bound range random values generate (exclusive). dtype Optional dtype tensor. integer types supported. specified, \"int32\" used. seed Optional R integer instance random_seed_generator(). default, seed argument NULL, internal global random_seed_generator() used. seed argument can used ensure deterministic (repeatable) random number generation. Note passing integer seed value produce random values call. generate different random values repeated calls, instance random_seed_generator() must provided seed value. Remark concerning JAX backend: tracing functions JAX backend global random_seed_generator() supported. Therefore, tracing default value seed=NULL produce error, seed argument must provided.","code":""},{"path":"https://keras3.posit.co/dev/reference/random_integer.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Draw random integers from a uniform distribution. — random_integer","text":"tensor random values.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/random_normal.html","id":null,"dir":"Reference","previous_headings":"","what":"Draw random samples from a normal (Gaussian) distribution. — random_normal","title":"Draw random samples from a normal (Gaussian) distribution. — random_normal","text":"Draw random samples normal (Gaussian) distribution.","code":""},{"path":"https://keras3.posit.co/dev/reference/random_normal.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Draw random samples from a normal (Gaussian) distribution. — random_normal","text":"","code":"random_normal(shape, mean = 0, stddev = 1, dtype = NULL, seed = NULL)"},{"path":"https://keras3.posit.co/dev/reference/random_normal.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Draw random samples from a normal (Gaussian) distribution. — random_normal","text":"shape shape random values generate. mean Float, defaults 0. Mean random values generate. stddev Float, defaults 1. Standard deviation random values generate. dtype Optional dtype tensor. floating point types supported. specified, config_floatx() used, defaults float32 unless configured otherwise (via config_set_floatx(float_dtype)). seed Optional R integer instance random_seed_generator(). default, seed argument NULL, internal global random_seed_generator() used. seed argument can used ensure deterministic (repeatable) random number generation. Note passing integer seed value produce random values call. generate different random values repeated calls, instance random_seed_generator() must provided seed value. Remark concerning JAX backend: tracing functions JAX backend global random_seed_generator() supported. Therefore, tracing default value seed=NULL produce error, seed argument must provided.","code":""},{"path":"https://keras3.posit.co/dev/reference/random_normal.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Draw random samples from a normal (Gaussian) distribution. — random_normal","text":"tensor random values.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/random_seed_generator.html","id":null,"dir":"Reference","previous_headings":"","what":"Generates variable seeds upon each call to a function generating random numbers. — random_seed_generator","title":"Generates variable seeds upon each call to a function generating random numbers. — random_seed_generator","text":"Keras, random number generators (random_normal()) stateless, meaning pass integer seed (seed=42), return values repeated calls. get different values call, SeedGenerator providing state random generator used. Note random number generators default seed NULL, implies internal global SeedGenerator used. need decouple RNG global state can provide local StateGenerator either deterministic random initial state. Remark concerning JAX backen: Note use local StateGenerator seed argument required JIT compilation RNG JAX backend, use global state supported.","code":""},{"path":"https://keras3.posit.co/dev/reference/random_seed_generator.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Generates variable seeds upon each call to a function generating random numbers. — random_seed_generator","text":"","code":"random_seed_generator(seed = NULL, name = NULL, ...)"},{"path":"https://keras3.posit.co/dev/reference/random_seed_generator.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Generates variable seeds upon each call to a function generating random numbers. — random_seed_generator","text":"seed Initial seed random number generator name String, name object ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/random_seed_generator.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Generates variable seeds upon each call to a function generating random numbers. — random_seed_generator","text":"SeedGenerator instance, can passed seed =  argument random tensor generators.","code":""},{"path":"https://keras3.posit.co/dev/reference/random_seed_generator.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Generates variable seeds upon each call to a function generating random numbers. — random_seed_generator","text":"Usage layer:","code":"seed_gen <- random_seed_generator(seed = 42) values <- random_normal(shape = c(2, 3), seed = seed_gen) new_values <- random_normal(shape = c(2, 3), seed = seed_gen) layer_dropout2 <- new_layer_class(   \"dropout2\",   initialize = function(...) {     super$initialize(...)     self$seed_generator <- random_seed_generator(seed = 1337)   },   call = function(x, training = FALSE) {     if (training) {       return(random_dropout(x, rate = 0.5, seed = self$seed_generator))     }     return(x)   } )  out <- layer_dropout(rate = 0.8) out(op_ones(10), training = TRUE) ## tf.Tensor([0. 5. 5. 0. 0. 0. 0. 0. 0. 0.], shape=(10), dtype=float32)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/random_shuffle.html","id":null,"dir":"Reference","previous_headings":"","what":"Shuffle the elements of a tensor uniformly at random along an axis. — random_shuffle","title":"Shuffle the elements of a tensor uniformly at random along an axis. — random_shuffle","text":"Shuffle elements tensor uniformly random along axis.","code":""},{"path":"https://keras3.posit.co/dev/reference/random_shuffle.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Shuffle the elements of a tensor uniformly at random along an axis. — random_shuffle","text":"","code":"random_shuffle(x, axis = 1L, seed = NULL)"},{"path":"https://keras3.posit.co/dev/reference/random_shuffle.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Shuffle the elements of a tensor uniformly at random along an axis. — random_shuffle","text":"x tensor shuffled. axis integer specifying axis along shuffle. Defaults 0. seed Optional R integer instance random_seed_generator(). default, seed argument NULL, internal global random_seed_generator() used. seed argument can used ensure deterministic (repeatable) random number generation. Note passing integer seed value produce random values call. generate different random values repeated calls, instance random_seed_generator() must provided seed value. Remark concerning JAX backend: tracing functions JAX backend global random_seed_generator() supported. Therefore, tracing default value seed=NULL produce error, seed argument must provided.","code":""},{"path":"https://keras3.posit.co/dev/reference/random_shuffle.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Shuffle the elements of a tensor uniformly at random along an axis. — random_shuffle","text":"tensor, copy x axis axis shuffled.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/random_truncated_normal.html","id":null,"dir":"Reference","previous_headings":"","what":"Draw samples from a truncated normal distribution. — random_truncated_normal","title":"Draw samples from a truncated normal distribution. — random_truncated_normal","text":"values drawn normal distribution specified mean standard deviation, discarding re-drawing samples two standard deviations mean.","code":""},{"path":"https://keras3.posit.co/dev/reference/random_truncated_normal.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Draw samples from a truncated normal distribution. — random_truncated_normal","text":"","code":"random_truncated_normal(shape, mean = 0, stddev = 1, dtype = NULL, seed = NULL)"},{"path":"https://keras3.posit.co/dev/reference/random_truncated_normal.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Draw samples from a truncated normal distribution. — random_truncated_normal","text":"shape shape random values generate. mean Float, defaults 0. Mean random values generate. stddev Float, defaults 1. Standard deviation random values generate. dtype Optional dtype tensor. floating point types supported. specified, config_floatx() used, defaults float32 unless configured otherwise (via config_set_floatx(float_dtype)) seed Optional R integer instance random_seed_generator(). default, seed argument NULL, internal global random_seed_generator() used. seed argument can used ensure deterministic (repeatable) random number generation. Note passing integer seed value produce random values call. generate different random values repeated calls, instance random_seed_generator() must provided seed value. Remark concerning JAX backend: tracing functions JAX backend global random_seed_generator() supported. Therefore, tracing default value seed=NULL produce error, seed argument must provided.","code":""},{"path":"https://keras3.posit.co/dev/reference/random_truncated_normal.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Draw samples from a truncated normal distribution. — random_truncated_normal","text":"tensor random values.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/random_uniform.html","id":null,"dir":"Reference","previous_headings":"","what":"Draw samples from a uniform distribution. — random_uniform","title":"Draw samples from a uniform distribution. — random_uniform","text":"generated values follow uniform distribution range [minval, maxval). lower bound minval included range, upper bound maxval excluded. dtype must floating point type, default range [0, 1).","code":""},{"path":"https://keras3.posit.co/dev/reference/random_uniform.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Draw samples from a uniform distribution. — random_uniform","text":"","code":"random_uniform(shape, minval = 0, maxval = 1, dtype = NULL, seed = NULL)"},{"path":"https://keras3.posit.co/dev/reference/random_uniform.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Draw samples from a uniform distribution. — random_uniform","text":"shape shape random values generate. minval Float, defaults 0. Lower bound range random values generate (inclusive). maxval Float, defaults 1. Upper bound range random values generate (exclusive). dtype Optional dtype tensor. floating point types supported. specified, config_floatx() used, defaults float32 unless configured otherwise (via config_set_floatx(float_dtype)) seed Optional R integer instance random_seed_generator(). default, seed argument NULL, internal global random_seed_generator() used. seed argument can used ensure deterministic (repeatable) random number generation. Note passing integer seed value produce random values call. generate different random values repeated calls, instance random_seed_generator() must provided seed value. Remark concerning JAX backend: tracing functions JAX backend global random_seed_generator() supported. Therefore, tracing default value seed=NULL produce error, seed argument must provided.","code":""},{"path":"https://keras3.posit.co/dev/reference/random_uniform.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Draw samples from a uniform distribution. — random_uniform","text":"tensor random values.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/reexports.html","id":null,"dir":"Reference","previous_headings":"","what":"Objects exported from other packages — reexports","title":"Objects exported from other packages — reexports","text":"objects imported packages. Follow links see documentation. dotty . generics compile, fit, generate magrittr %<>% reticulate %%, array_reshape, as_iterator, iter_next, iterate, np_array, tuple, use_python, use_virtualenv tensorflow all_dims, as_tensor, evaluate, export_savedmodel, tensorboard tfruns flag_boolean, flag_integer, flag_numeric, flag_string, flags, run_dir","code":""},{"path":"https://keras3.posit.co/dev/reference/register_keras_serializable.html","id":null,"dir":"Reference","previous_headings":"","what":"Registers a custom object with the Keras serialization framework. — register_keras_serializable","title":"Registers a custom object with the Keras serialization framework. — register_keras_serializable","text":"function registers custom class function Keras custom object registry, can serialized deserialized without needing entry user-provided custom_objects argument. also injects function Keras call get object's serializable string key. Note serialized deserialized, classes must implement get_config() method. Functions requirement. object registered key 'package>name' name, defaults object name passed.","code":""},{"path":"https://keras3.posit.co/dev/reference/register_keras_serializable.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Registers a custom object with the Keras serialization framework. — register_keras_serializable","text":"","code":"register_keras_serializable(object, name = NULL, package = NULL)"},{"path":"https://keras3.posit.co/dev/reference/register_keras_serializable.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Registers a custom object with the Keras serialization framework. — register_keras_serializable","text":"object keras object. name name serialize class package. package package class belongs . used key (\"package>name\") identify class. Defaults current package name, \"Custom\" outside package.","code":""},{"path":"https://keras3.posit.co/dev/reference/register_keras_serializable.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Registers a custom object with the Keras serialization framework. — register_keras_serializable","text":"object returned invisibly, convenient piping. primarily called side effects.","code":""},{"path":"https://keras3.posit.co/dev/reference/register_keras_serializable.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Registers a custom object with the Keras serialization framework. — register_keras_serializable","text":"","code":"# Note that `'my_package'` is used as the `package` argument here, and since # the `name` argument is not provided, `'MyDense'` is used as the `name`. layer_my_dense <- Layer(\"MyDense\") register_keras_serializable(layer_my_dense, package = \"my_package\")  MyDense <- environment(layer_my_dense)$`__class__` # the python class obj stopifnot(exprs = {   get_registered_object('my_package>MyDense') == MyDense   get_registered_name(MyDense) == 'my_package>MyDense' })"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/regularizer_l1.html","id":null,"dir":"Reference","previous_headings":"","what":"A regularizer that applies a L1 regularization penalty. — regularizer_l1","title":"A regularizer that applies a L1 regularization penalty. — regularizer_l1","text":"L1 regularization penalty computed : loss = l1 * reduce_sum(abs(x)) L1 may passed layer string identifier:   case, default value used l1=0.01.","code":"dense <- layer_dense(units = 3, kernel_regularizer = 'l1')"},{"path":"https://keras3.posit.co/dev/reference/regularizer_l1.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"A regularizer that applies a L1 regularization penalty. — regularizer_l1","text":"","code":"regularizer_l1(l1 = 0.01)"},{"path":"https://keras3.posit.co/dev/reference/regularizer_l1.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"A regularizer that applies a L1 regularization penalty. — regularizer_l1","text":"l1 float, L1 regularization factor.","code":""},{"path":"https://keras3.posit.co/dev/reference/regularizer_l1.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"A regularizer that applies a L1 regularization penalty. — regularizer_l1","text":"Regularizer instance can passed layer constructors used standalone object.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/regularizer_l1_l2.html","id":null,"dir":"Reference","previous_headings":"","what":"A regularizer that applies both L1 and L2 regularization penalties. — regularizer_l1_l2","title":"A regularizer that applies both L1 and L2 regularization penalties. — regularizer_l1_l2","text":"L1 regularization penalty computed : loss = l1 * reduce_sum(abs(x)) L2 regularization penalty computed loss = l2 * reduce_sum(square(x)) L1L2 may passed layer string identifier:   case, default values used l1=0.01 l2=0.01.","code":"dense <- layer_dense(units = 3, kernel_regularizer = 'L1L2')"},{"path":"https://keras3.posit.co/dev/reference/regularizer_l1_l2.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"A regularizer that applies both L1 and L2 regularization penalties. — regularizer_l1_l2","text":"","code":"regularizer_l1_l2(l1 = 0, l2 = 0)"},{"path":"https://keras3.posit.co/dev/reference/regularizer_l1_l2.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"A regularizer that applies both L1 and L2 regularization penalties. — regularizer_l1_l2","text":"l1 float, L1 regularization factor. l2 float, L2 regularization factor.","code":""},{"path":"https://keras3.posit.co/dev/reference/regularizer_l1_l2.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"A regularizer that applies both L1 and L2 regularization penalties. — regularizer_l1_l2","text":"Regularizer instance can passed layer constructors used standalone object.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/regularizer_l2.html","id":null,"dir":"Reference","previous_headings":"","what":"A regularizer that applies a L2 regularization penalty. — regularizer_l2","title":"A regularizer that applies a L2 regularization penalty. — regularizer_l2","text":"L2 regularization penalty computed : loss = l2 * reduce_sum(square(x)) L2 may passed layer string identifier:   case, default value used l2=0.01.","code":"dense <- layer_dense(units = 3, kernel_regularizer='l2')"},{"path":"https://keras3.posit.co/dev/reference/regularizer_l2.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"A regularizer that applies a L2 regularization penalty. — regularizer_l2","text":"","code":"regularizer_l2(l2 = 0.01)"},{"path":"https://keras3.posit.co/dev/reference/regularizer_l2.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"A regularizer that applies a L2 regularization penalty. — regularizer_l2","text":"l2 float, L2 regularization factor.","code":""},{"path":"https://keras3.posit.co/dev/reference/regularizer_l2.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"A regularizer that applies a L2 regularization penalty. — regularizer_l2","text":"Regularizer instance can passed layer constructors used standalone object.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/regularizer_orthogonal.html","id":null,"dir":"Reference","previous_headings":"","what":"Regularizer that encourages input vectors to be orthogonal to each other. — regularizer_orthogonal","title":"Regularizer that encourages input vectors to be orthogonal to each other. — regularizer_orthogonal","text":"can applied either rows matrix (mode=\"rows\") columns (mode=\"columns\"). applied Dense kernel shape (input_dim, units), rows mode seek make feature vectors (.e. basis output space) orthogonal .","code":""},{"path":"https://keras3.posit.co/dev/reference/regularizer_orthogonal.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Regularizer that encourages input vectors to be orthogonal to each other. — regularizer_orthogonal","text":"","code":"regularizer_orthogonal(factor = 0.01, mode = \"rows\")"},{"path":"https://keras3.posit.co/dev/reference/regularizer_orthogonal.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Regularizer that encourages input vectors to be orthogonal to each other. — regularizer_orthogonal","text":"factor Float. regularization factor. regularization penalty proportional factor times mean dot products L2-normalized rows (mode=\"rows\", columns mode=\"columns\") inputs, excluding product row/column .  Defaults 0.01. mode String, one {\"rows\", \"columns\"}. Defaults \"rows\". rows mode, regularization effect seeks make rows input orthogonal . columns mode, seeks make columns input orthogonal .","code":""},{"path":"https://keras3.posit.co/dev/reference/regularizer_orthogonal.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Regularizer that encourages input vectors to be orthogonal to each other. — regularizer_orthogonal","text":"Regularizer instance can passed layer constructors used standalone object.","code":""},{"path":"https://keras3.posit.co/dev/reference/regularizer_orthogonal.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Regularizer that encourages input vectors to be orthogonal to each other. — regularizer_orthogonal","text":"","code":"regularizer <- regularizer_orthogonal(factor=0.01) layer <- layer_dense(units=4, kernel_regularizer=regularizer)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/reset_state.html","id":null,"dir":"Reference","previous_headings":"","what":"Reset the state for a model, layer or metric. — reset_state","title":"Reset the state for a model, layer or metric. — reset_state","text":"Reset state model, layer metric.","code":""},{"path":"https://keras3.posit.co/dev/reference/reset_state.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Reset the state for a model, layer or metric. — reset_state","text":"","code":"reset_state(object)"},{"path":"https://keras3.posit.co/dev/reference/reset_state.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Reset the state for a model, layer or metric. — reset_state","text":"object Model, Layer, Metric instance Layers resettable state (E.g., adapt()-able preprocessing layers rnn layers resettable state, layer_dense() ). Calling Layer instance without resettable-state error.","code":""},{"path":"https://keras3.posit.co/dev/reference/reset_state.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Reset the state for a model, layer or metric. — reset_state","text":"object, invisibly.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/rnn_cell_gru.html","id":null,"dir":"Reference","previous_headings":"","what":"Cell class for the GRU layer. — rnn_cell_gru","title":"Cell class for the GRU layer. — rnn_cell_gru","text":"class processes one step within whole time sequence input, whereas layer_gru() processes whole sequence.","code":""},{"path":"https://keras3.posit.co/dev/reference/rnn_cell_gru.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Cell class for the GRU layer. — rnn_cell_gru","text":"","code":"rnn_cell_gru(   units,   activation = \"tanh\",   recurrent_activation = \"sigmoid\",   use_bias = TRUE,   kernel_initializer = \"glorot_uniform\",   recurrent_initializer = \"orthogonal\",   bias_initializer = \"zeros\",   kernel_regularizer = NULL,   recurrent_regularizer = NULL,   bias_regularizer = NULL,   kernel_constraint = NULL,   recurrent_constraint = NULL,   bias_constraint = NULL,   dropout = 0,   recurrent_dropout = 0,   reset_after = TRUE,   seed = NULL,   ... )"},{"path":"https://keras3.posit.co/dev/reference/rnn_cell_gru.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Cell class for the GRU layer. — rnn_cell_gru","text":"units Positive integer, dimensionality output space. activation Activation function use. Default: hyperbolic tangent (tanh). pass NULL, activation applied (ie. \"linear\" activation: (x) = x). recurrent_activation Activation function use recurrent step. Default: sigmoid (sigmoid). pass NULL, activation applied (ie. \"linear\" activation: (x) = x). use_bias Boolean, (default TRUE), whether layer use bias vector. kernel_initializer Initializer kernel weights matrix, used linear transformation inputs. Default: \"glorot_uniform\". recurrent_initializer Initializer recurrent_kernel weights matrix, used linear transformation recurrent state. Default: \"orthogonal\". bias_initializer Initializer bias vector. Default: \"zeros\". kernel_regularizer Regularizer function applied kernel weights matrix. Default: NULL. recurrent_regularizer Regularizer function applied recurrent_kernel weights matrix. Default: NULL. bias_regularizer Regularizer function applied bias vector. Default: NULL. kernel_constraint Constraint function applied kernel weights matrix. Default: NULL. recurrent_constraint Constraint function applied recurrent_kernel weights matrix. Default: NULL. bias_constraint Constraint function applied bias vector. Default: NULL. dropout Float 0 1. Fraction units drop linear transformation inputs. Default: 0. recurrent_dropout Float 0 1. Fraction units drop linear transformation recurrent state. Default: 0. reset_after GRU convention (whether apply reset gate matrix multiplication). FALSE = \"\", TRUE = \"\" (default cuDNN compatible). seed Random seed dropout. ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/rnn_cell_gru.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Cell class for the GRU layer. — rnn_cell_gru","text":"Layer instance, intended used layer_rnn().","code":""},{"path":"https://keras3.posit.co/dev/reference/rnn_cell_gru.html","id":"call-arguments","dir":"Reference","previous_headings":"","what":"Call Arguments","title":"Cell class for the GRU layer. — rnn_cell_gru","text":"inputs: 2D tensor, shape (batch, features). states: 2D tensor shape (batch, units), state previous time step. training: Python boolean indicating whether layer behave training mode inference mode. relevant dropout recurrent_dropout used.","code":""},{"path":"https://keras3.posit.co/dev/reference/rnn_cell_gru.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Cell class for the GRU layer. — rnn_cell_gru","text":"","code":"inputs <- random_uniform(c(32, 10, 8)) outputs <- inputs |> layer_rnn(rnn_cell_gru(4)) shape(outputs) ## shape(32, 4) rnn <- layer_rnn(    cell = rnn_cell_gru(4),    return_sequences=TRUE,    return_state=TRUE) c(whole_sequence_output, final_state) %<-% rnn(inputs) shape(whole_sequence_output) ## shape(32, 10, 4) shape(final_state) ## shape(32, 4)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/rnn_cell_lstm.html","id":null,"dir":"Reference","previous_headings":"","what":"Cell class for the LSTM layer. — rnn_cell_lstm","title":"Cell class for the LSTM layer. — rnn_cell_lstm","text":"class processes one step within whole time sequence input, whereas layer_lstm() processes whole sequence.","code":""},{"path":"https://keras3.posit.co/dev/reference/rnn_cell_lstm.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Cell class for the LSTM layer. — rnn_cell_lstm","text":"","code":"rnn_cell_lstm(   units,   activation = \"tanh\",   recurrent_activation = \"sigmoid\",   use_bias = TRUE,   kernel_initializer = \"glorot_uniform\",   recurrent_initializer = \"orthogonal\",   bias_initializer = \"zeros\",   unit_forget_bias = TRUE,   kernel_regularizer = NULL,   recurrent_regularizer = NULL,   bias_regularizer = NULL,   kernel_constraint = NULL,   recurrent_constraint = NULL,   bias_constraint = NULL,   dropout = 0,   recurrent_dropout = 0,   seed = NULL,   ... )"},{"path":"https://keras3.posit.co/dev/reference/rnn_cell_lstm.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Cell class for the LSTM layer. — rnn_cell_lstm","text":"units Positive integer, dimensionality output space. activation Activation function use. Default: hyperbolic tangent (tanh). pass NULL, activation applied (ie. \"linear\" activation: (x) = x). recurrent_activation Activation function use recurrent step. Default: sigmoid (sigmoid). pass NULL, activation applied (ie. \"linear\" activation: (x) = x). use_bias Boolean, (default TRUE), whether layer use bias vector. kernel_initializer Initializer kernel weights matrix, used linear transformation inputs. Default: \"glorot_uniform\". recurrent_initializer Initializer recurrent_kernel weights matrix, used linear transformation recurrent state. Default: \"orthogonal\". bias_initializer Initializer bias vector. Default: \"zeros\". unit_forget_bias Boolean (default TRUE). TRUE, add 1 bias forget gate initialization. Setting TRUE also force bias_initializer=\"zeros\". recommended Jozefowicz et al. kernel_regularizer Regularizer function applied kernel weights matrix. Default: NULL. recurrent_regularizer Regularizer function applied recurrent_kernel weights matrix. Default: NULL. bias_regularizer Regularizer function applied bias vector. Default: NULL. kernel_constraint Constraint function applied kernel weights matrix. Default: NULL. recurrent_constraint Constraint function applied recurrent_kernel weights matrix. Default: NULL. bias_constraint Constraint function applied bias vector. Default: NULL. dropout Float 0 1. Fraction units drop linear transformation inputs. Default: 0. recurrent_dropout Float 0 1. Fraction units drop linear transformation recurrent state. Default: 0. seed Random seed dropout. ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/rnn_cell_lstm.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Cell class for the LSTM layer. — rnn_cell_lstm","text":"Layer instance, intended used layer_rnn().","code":""},{"path":"https://keras3.posit.co/dev/reference/rnn_cell_lstm.html","id":"call-arguments","dir":"Reference","previous_headings":"","what":"Call Arguments","title":"Cell class for the LSTM layer. — rnn_cell_lstm","text":"inputs: 2D tensor, shape (batch, features). states: 2D tensor shape (batch, units), state previous time step. training: Boolean indicating whether layer behave training mode inference mode. relevant dropout recurrent_dropout used.","code":""},{"path":"https://keras3.posit.co/dev/reference/rnn_cell_lstm.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Cell class for the LSTM layer. — rnn_cell_lstm","text":"","code":"inputs <- random_uniform(c(32, 10, 8)) output <- inputs |>   layer_rnn(cell = rnn_cell_lstm(4)) shape(output) ## shape(32, 4) rnn <- layer_rnn(cell = rnn_cell_lstm(4),                  return_sequences = T,                  return_state = T) c(whole_sequence_output, ..final_state) %<-% rnn(inputs) str(whole_sequence_output) ## <tf.Tensor: shape=(32, 10, 4), dtype=float32, numpy=…> str(final_state) ## List of 2 ##  $ :<tf.Tensor: shape=(32, 4), dtype=float32, numpy=…> ##  $ :<tf.Tensor: shape=(32, 4), dtype=float32, numpy=…>"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/rnn_cell_simple.html","id":null,"dir":"Reference","previous_headings":"","what":"Cell class for SimpleRNN. — rnn_cell_simple","title":"Cell class for SimpleRNN. — rnn_cell_simple","text":"class processes one step within whole time sequence input, whereas layer_simple_rnn() processes whole sequence.","code":""},{"path":"https://keras3.posit.co/dev/reference/rnn_cell_simple.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Cell class for SimpleRNN. — rnn_cell_simple","text":"","code":"rnn_cell_simple(   units,   activation = \"tanh\",   use_bias = TRUE,   kernel_initializer = \"glorot_uniform\",   recurrent_initializer = \"orthogonal\",   bias_initializer = \"zeros\",   kernel_regularizer = NULL,   recurrent_regularizer = NULL,   bias_regularizer = NULL,   kernel_constraint = NULL,   recurrent_constraint = NULL,   bias_constraint = NULL,   dropout = 0,   recurrent_dropout = 0,   seed = NULL,   ... )"},{"path":"https://keras3.posit.co/dev/reference/rnn_cell_simple.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Cell class for SimpleRNN. — rnn_cell_simple","text":"units Positive integer, dimensionality output space. activation Activation function use. Default: hyperbolic tangent (tanh). pass NULL, activation applied (ie. \"linear\" activation: (x) = x). use_bias Boolean, (default TRUE), whether layer use bias vector. kernel_initializer Initializer kernel weights matrix, used linear transformation inputs. Default: \"glorot_uniform\". recurrent_initializer Initializer recurrent_kernel weights matrix, used linear transformation recurrent state. Default: \"orthogonal\". bias_initializer Initializer bias vector. Default: \"zeros\". kernel_regularizer Regularizer function applied kernel weights matrix. Default: NULL. recurrent_regularizer Regularizer function applied recurrent_kernel weights matrix. Default: NULL. bias_regularizer Regularizer function applied bias vector. Default: NULL. kernel_constraint Constraint function applied kernel weights matrix. Default: NULL. recurrent_constraint Constraint function applied recurrent_kernel weights matrix. Default: NULL. bias_constraint Constraint function applied bias vector. Default: NULL. dropout Float 0 1. Fraction units drop linear transformation inputs. Default: 0. recurrent_dropout Float 0 1. Fraction units drop linear transformation recurrent state. Default: 0. seed Random seed dropout. ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/rnn_cell_simple.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Cell class for SimpleRNN. — rnn_cell_simple","text":"Layer instance, intended used layer_rnn().","code":""},{"path":"https://keras3.posit.co/dev/reference/rnn_cell_simple.html","id":"call-arguments","dir":"Reference","previous_headings":"","what":"Call Arguments","title":"Cell class for SimpleRNN. — rnn_cell_simple","text":"sequence: 2D tensor, shape (batch, features). states: 2D tensor shape (batch, units), state previous time step. training: Python boolean indicating whether layer behave training mode inference mode. relevant dropout recurrent_dropout used.","code":""},{"path":"https://keras3.posit.co/dev/reference/rnn_cell_simple.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Cell class for SimpleRNN. — rnn_cell_simple","text":"","code":"inputs <- random_uniform(c(32, 10, 8)) rnn <- layer_rnn(cell = rnn_cell_simple(units = 4)) output <- rnn(inputs)  # The output has shape `(32, 4)`. rnn <- layer_rnn(     cell = rnn_cell_simple(units = 4),     return_sequences=TRUE,     return_state=TRUE ) # whole_sequence_output has shape `(32, 10, 4)`. # final_state has shape `(32, 4)`. c(whole_sequence_output, final_state) %<-% rnn(inputs)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/rnn_cells_stack.html","id":null,"dir":"Reference","previous_headings":"","what":"Wrapper allowing a stack of RNN cells to behave as a single cell. — rnn_cells_stack","title":"Wrapper allowing a stack of RNN cells to behave as a single cell. — rnn_cells_stack","text":"Used implement efficient stacked RNNs.","code":""},{"path":"https://keras3.posit.co/dev/reference/rnn_cells_stack.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Wrapper allowing a stack of RNN cells to behave as a single cell. — rnn_cells_stack","text":"","code":"rnn_cells_stack(cells, ...)"},{"path":"https://keras3.posit.co/dev/reference/rnn_cells_stack.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Wrapper allowing a stack of RNN cells to behave as a single cell. — rnn_cells_stack","text":"cells List RNN cell instances. ... Unnamed arguments treated additional cells. Named arguments passed underlying layer.","code":""},{"path":"https://keras3.posit.co/dev/reference/rnn_cells_stack.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Wrapper allowing a stack of RNN cells to behave as a single cell. — rnn_cells_stack","text":"Layer instance, intended used layer_rnn().","code":""},{"path":"https://keras3.posit.co/dev/reference/rnn_cells_stack.html","id":"example","dir":"Reference","previous_headings":"","what":"Example","title":"Wrapper allowing a stack of RNN cells to behave as a single cell. — rnn_cells_stack","text":"","code":"batch_size <- 3 sentence_length <- 5 num_features <- 2 new_shape <- c(batch_size, sentence_length, num_features) x <- array(1:30, dim = new_shape)  rnn_cells <- lapply(1:2, function(x) rnn_cell_lstm(units = 128)) stacked_lstm <- rnn_cells_stack(rnn_cells) lstm_layer <- layer_rnn(cell = stacked_lstm)  result <- lstm_layer(x) str(result) ## <tf.Tensor: shape=(3, 128), dtype=float32, numpy=…>"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/save_model.html","id":null,"dir":"Reference","previous_headings":"","what":"Saves a model as a .keras file. — save_model","title":"Saves a model as a .keras file. — save_model","text":"Saves model .keras file.","code":""},{"path":"https://keras3.posit.co/dev/reference/save_model.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Saves a model as a .keras file. — save_model","text":"","code":"save_model(model, filepath = NULL, overwrite = FALSE, zipped = NULL, ...)"},{"path":"https://keras3.posit.co/dev/reference/save_model.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Saves a model as a .keras file. — save_model","text":"model keras model. filepath string, Path save model. Must end .keras. overwrite Whether overwrite existing model target location, instead ask user via interactive prompt. zipped Whether save model zipped .keras archive (default saving locally), unzipped directory (default saving Hugging Face Hub). ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/save_model.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Saves a model as a .keras file. — save_model","text":"filepath provided, function called primarily side effects, model returned invisibly. filepath provided NULL, serialized model returned R raw vector.","code":""},{"path":"https://keras3.posit.co/dev/reference/save_model.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Saves a model as a .keras file. — save_model","text":"saved .keras file zip archive contains: model's configuration (architecture) model's weights model's optimizer's state () Thus models can reinstantiated exact state.","code":"model <- keras_model_sequential(input_shape = c(3)) |>   layer_dense(5) |>   layer_activation_softmax()  model |> save_model(\"model.keras\") loaded_model <- load_model(\"model.keras\") x <- random_uniform(c(10, 3)) stopifnot(all.equal(   model |> predict(x),   loaded_model |> predict(x) )) zip::zip_list(\"model.keras\")[, \"filename\"] ## [1] \"metadata.json\"    \"config.json\"      \"model.weights.h5\""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/save_model_config.html","id":null,"dir":"Reference","previous_headings":"","what":"Save and load model configuration as JSON — save_model_config","title":"Save and load model configuration as JSON — save_model_config","text":"Save re-load models configurations JSON. Note representation include weights, architecture.","code":""},{"path":"https://keras3.posit.co/dev/reference/save_model_config.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Save and load model configuration as JSON — save_model_config","text":"","code":"save_model_config(model, filepath = NULL, overwrite = FALSE)  load_model_config(filepath, custom_objects = NULL)"},{"path":"https://keras3.posit.co/dev/reference/save_model_config.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Save and load model configuration as JSON — save_model_config","text":"model Model object save filepath path json file model config. overwrite Whether overwrite existing model configuration json filepath, instead ask user via interactive prompt. custom_objects Optional named list mapping names custom classes functions considered deserialization.","code":""},{"path":"https://keras3.posit.co/dev/reference/save_model_config.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Save and load model configuration as JSON — save_model_config","text":"called primarily side effects. model returned, invisibly, enable usage pipe.","code":""},{"path":"https://keras3.posit.co/dev/reference/save_model_config.html","id":"details","dir":"Reference","previous_headings":"","what":"Details","title":"Save and load model configuration as JSON — save_model_config","text":"Note: save_model_config() serializes model JSON using serialize_keras_object(), get_config(). serialize_keras_object() returns superset get_config(), additional information needed create class object needed restore model. See example extract get_config() value saved model.","code":""},{"path":"https://keras3.posit.co/dev/reference/save_model_config.html","id":"example","dir":"Reference","previous_headings":"","what":"Example","title":"Save and load model configuration as JSON — save_model_config","text":"","code":"model <- keras_model_sequential(input_shape = 10) |> layer_dense(10) file <- tempfile(\"model-config-\", fileext = \".json\") save_model_config(model, file)  # load a new model instance with the same architecture but different weights model2 <- load_model_config(file)  stopifnot(exprs = {   all.equal(get_config(model), get_config(model2))    # To extract the `get_config()` value from a saved model config:   all.equal(       get_config(model),       structure(jsonlite::read_json(file)$config,                 \"__class__\" = keras_model_sequential()$`__class__`)   ) })"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/save_model_weights.html","id":null,"dir":"Reference","previous_headings":"","what":"Saves all weights to a single file or sharded files. — save_model_weights","title":"Saves all weights to a single file or sharded files. — save_model_weights","text":"default, weights saved single .weights.h5 file. Enable sharding via max_shard_size split weights across multiple files (GB) produce .weights.json manifest tracks shard metadata. saved sharded files contain: *.weights.json: configuration file containing metadata weight_map entries. *_xxxxxx.weights.h5: weight shards limited max_shard_size.","code":"model <-   keras_model_sequential(input_shape = 2) |>   layer_dense(4)  path_h5 <- tempfile(fileext = \".weights.h5\") path_json <- tempfile(fileext = \".weights.json\")  model |> save_model_weights(path_h5) model |> save_model_weights(path_json, max_shard_size = 0.01)  model |> load_model_weights(path_h5) model |> load_model_weights(path_json)"},{"path":"https://keras3.posit.co/dev/reference/save_model_weights.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Saves all weights to a single file or sharded files. — save_model_weights","text":"","code":"save_model_weights(model, filepath, overwrite = FALSE, max_shard_size = NULL)"},{"path":"https://keras3.posit.co/dev/reference/save_model_weights.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Saves all weights to a single file or sharded files. — save_model_weights","text":"model keras Model object. filepath Path weights saved. Accepts .weights.h5, sharding enabled, .weights.json manifest path. .weights.h5 provided sharding, filename overridden end .weights.json. overwrite Whether overwrite existing weights target location, instead ask user via interactive prompt. max_shard_size Numeric size GB sharded file. Use NULL disable sharding.","code":""},{"path":"https://keras3.posit.co/dev/reference/save_model_weights.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Saves all weights to a single file or sharded files. — save_model_weights","text":"called primarily side effects. model returned, invisibly, enable usage pipe.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/serialize_keras_object.html","id":null,"dir":"Reference","previous_headings":"","what":"Retrieve the full config by serializing the Keras object. — serialize_keras_object","title":"Retrieve the full config by serializing the Keras object. — serialize_keras_object","text":"serialize_keras_object() serializes Keras object named list represents object, reciprocal function deserialize_keras_object(). See deserialize_keras_object() information full config format.","code":""},{"path":"https://keras3.posit.co/dev/reference/serialize_keras_object.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Retrieve the full config by serializing the Keras object. — serialize_keras_object","text":"","code":"serialize_keras_object(obj)"},{"path":"https://keras3.posit.co/dev/reference/serialize_keras_object.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Retrieve the full config by serializing the Keras object. — serialize_keras_object","text":"obj Keras object serialize.","code":""},{"path":"https://keras3.posit.co/dev/reference/serialize_keras_object.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Retrieve the full config by serializing the Keras object. — serialize_keras_object","text":"named list represents object config. config expected contain simple types , can saved json. object can deserialized config via deserialize_keras_object().","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/set_random_seed.html","id":null,"dir":"Reference","previous_headings":"","what":"Sets all random seeds (Python, NumPy, and backend framework, e.g. TF). — set_random_seed","title":"Sets all random seeds (Python, NumPy, and backend framework, e.g. TF). — set_random_seed","text":"can use utility make almost Keras program fully deterministic. limitations apply cases network communications involved (e.g. parameter server distribution), creates additional sources randomness, certain non-deterministic cuDNN ops involved. sets: R session seed: set.seed() Python session seed: import random; random.seed(seed) Python NumPy seed: import numpy; numpy.random.seed(seed) TensorFlow seed: tf$random$set_seed(seed) (TF installed) Torch seed: import(\"torch\")$manual_seed(seed) (backend torch) disables Python hash randomization. Note TensorFlow seed set even using TensorFlow backend framework, since many workflows leverage tf$data pipelines (feature random shuffling). Likewise many workflows might leverage NumPy APIs.","code":""},{"path":"https://keras3.posit.co/dev/reference/set_random_seed.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Sets all random seeds (Python, NumPy, and backend framework, e.g. TF). — set_random_seed","text":"","code":"set_random_seed(seed)"},{"path":"https://keras3.posit.co/dev/reference/set_random_seed.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Sets all random seeds (Python, NumPy, and backend framework, e.g. TF). — set_random_seed","text":"seed Integer, random seed use.","code":""},{"path":"https://keras3.posit.co/dev/reference/set_random_seed.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Sets all random seeds (Python, NumPy, and backend framework, e.g. TF). — set_random_seed","text":"return value, called side effects.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/set_state_tree.html","id":null,"dir":"Reference","previous_headings":"","what":"Assigns values to variables of the model. — set_state_tree","title":"Assigns values to variables of the model. — set_state_tree","text":"method takes dictionary (named list) nested variable values, represents state tree model, assigns corresponding variables model. keys (list names) represent variable names (e.g., 'trainable_variables', 'optimizer_variables'), values nested dictionaries containing variable paths corresponding values.","code":""},{"path":"https://keras3.posit.co/dev/reference/set_state_tree.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Assigns values to variables of the model. — set_state_tree","text":"","code":"set_state_tree(object, state_tree)"},{"path":"https://keras3.posit.co/dev/reference/set_state_tree.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Assigns values to variables of the model. — set_state_tree","text":"object keras model. state_tree dictionary representing state tree model. keys variable names, values nested dictionaries representing variable paths values.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/shape.html","id":null,"dir":"Reference","previous_headings":"","what":"Tensor shape utility — shape","title":"Tensor shape utility — shape","text":"function can used get create tensor shape.","code":""},{"path":"https://keras3.posit.co/dev/reference/shape.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Tensor shape utility — shape","text":"","code":"shape(...)  # S3 method for class 'keras_shape' format(x, ..., prefix = TRUE)  # S3 method for class 'keras_shape' print(x, ...)  # S3 method for class 'keras_shape' x[...]  # S3 method for class 'keras_shape' as.integer(x, ...)  # S3 method for class 'keras_shape' Summary(..., na.rm = FALSE)  # S3 method for class 'keras_shape' as.list(x, ...)  # S3 method for class 'keras_shape' x == y  # S3 method for class 'keras_shape' x != y"},{"path":"https://keras3.posit.co/dev/reference/shape.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Tensor shape utility — shape","text":"... shape specification. Numerics, NULL tensors valid. NULL, NA, -1L can used specify unspecified dim size. Tensors dispatched op_shape() extract tensor shape. Values wrapped () used asis (see examples). objects coerced via .integer(). x, y keras_shape object. prefix Whether format shape object prefix. Defaults \"shape\". na.rm passed Summary group generics like prod(). Unknown axes treated NA.","code":""},{"path":"https://keras3.posit.co/dev/reference/shape.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Tensor shape utility — shape","text":"list \"keras_shape\" class attribute. element list either ) NULL, b) R integer c) scalar integer tensor (e.g., supplied TF tensor unspecified dimension function traced).","code":""},{"path":"https://keras3.posit.co/dev/reference/shape.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Tensor shape utility — shape","text":"3 ways specify unknown dimension     functions take 'shape' argument also coerce shape()     can also use shape() get shape tensor (excepting scalar integer tensors).             Combine expand shapes         Scalar integer tensors treated axis values. commonly encountered tracing function graph mode, axis size might unknown.     useful pattern unpack shape() %<-%, like :                     want resolve shape tensor can potentially scalar integer, can wrap tensor (), use op_shape().","code":"shape(1, 2, 3) ## shape(1, 2, 3) shape(NA,   2, 3) shape(NULL, 2, 3) shape(-1,   2, 3) ## shape(NA, 2, 3) ## shape(NA, 2, 3) ## shape(NA, 2, 3) layer_input(c(1, 2, 3)) layer_input(shape(1, 2, 3)) ## <KerasTensor shape=(None, 1, 2, 3), dtype=float32, sparse=False, ragged=False, name=keras_tensor> ## <KerasTensor shape=(None, 1, 2, 3), dtype=float32, sparse=False, ragged=False, name=keras_tensor_1> symbolic_tensor <- layer_input(shape(1, 2, 3)) shape(symbolic_tensor) ## shape(NA, 1, 2, 3) eager_tensor <- op_ones(c(1,2,3)) shape(eager_tensor) ## shape(1, 2, 3) op_shape(eager_tensor) ## shape(1, 2, 3) shape(symbolic_tensor, 4) ## shape(NA, 1, 2, 3, 4) shape(5, symbolic_tensor, 4) ## shape(5, NA, 1, 2, 3, 4) tfn <- tensorflow::tf_function(function(x) {   print(op_shape(x))   x }, input_signature = list(tensorflow::tf$TensorSpec(shape(1, NA, 3)))) invisible(tfn(op_ones(shape(1, 2, 3)))) ## shape(1, Tensor(\"strided_slice:0\", shape=(), dtype=int32), 3) c(batch_size, seq_len, channels) %<-% shape(x)  # `%<-%` also has support for skipping values # during unpacking with `.` and `...`. For example, # To retrieve just the first and/or last dim: c(batch_size, ...) %<-% shape(x) c(batch_size, ., .) %<-% shape(x) c(..., channels) %<-% shape(x) c(batch_size, ..., channels) %<-% shape(x) c(batch_size, ., channels) %<-% shape(x) echo_print <- function(x) {   message(\"> \", deparse(substitute(x)));   if(!is.null(x)) print(x) } tfn <- tensorflow::tf_function(function(x) {   c(axis1, axis2, axis3) %<-% shape(x)   echo_print(str(list(axis1 = axis1, axis2 = axis2, axis3 = axis3)))    echo_print(shape(axis1))               # use axis1 tensor as axis value   echo_print(shape(axis1, axis2, axis3)) # use axis1 tensor as axis value    # use shape() to compose a new shape, e.g., in multihead attention   n_heads <- 4   echo_print(shape(axis1, axis2, n_heads, axis3/n_heads))    x }, input_signature = list(tensorflow::tf$TensorSpec(shape(NA, 4, 16)))) invisible(tfn(op_ones(shape(2, 4, 16)))) ## > str(list(axis1 = axis1, axis2 = axis2, axis3 = axis3)) ## List of 3 ##  $ axis1:<tf.Tensor 'strided_slice:0' shape=() dtype=int32> ##  $ axis2: int 4 ##  $ axis3: int 16 ## > shape(axis1) ## shape(Tensor(\"strided_slice:0\", shape=(), dtype=int32)) ## > shape(axis1, axis2, axis3) ## shape(Tensor(\"strided_slice:0\", shape=(), dtype=int32), 4, 16) ## > shape(axis1, axis2, n_heads, axis3/n_heads) ## shape(Tensor(\"strided_slice:0\", shape=(), dtype=int32), 4, 4, 4) (x <- op_convert_to_tensor(2L)) ## tf.Tensor(2, shape=(), dtype=int32) # by default, shape() treats scalar integer tensors as axis values shape(x) ## shape(tf.Tensor(2, shape=(), dtype=int32)) # to access the shape of a scalar integer, # call `op_shape()`, or protect with `I()` op_shape(x) ## shape() shape(I(x)) ## shape()"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/split_dataset.html","id":null,"dir":"Reference","previous_headings":"","what":"Splits a dataset into a left half and a right half (e.g. train / test). — split_dataset","title":"Splits a dataset into a left half and a right half (e.g. train / test). — split_dataset","text":"Splits dataset left half right half (e.g. train / test).","code":""},{"path":"https://keras3.posit.co/dev/reference/split_dataset.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Splits a dataset into a left half and a right half (e.g. train / test). — split_dataset","text":"","code":"split_dataset(   dataset,   left_size = NULL,   right_size = NULL,   shuffle = FALSE,   seed = NULL )"},{"path":"https://keras3.posit.co/dev/reference/split_dataset.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Splits a dataset into a left half and a right half (e.g. train / test). — split_dataset","text":"dataset tf$data$Dataset, torch$utils$data$Dataset object, list arrays length. left_size float (range [0, 1]), signifies fraction data pack left dataset. integer, signifies number samples pack left dataset. NULL, defaults complement right_size. Defaults NULL. right_size float (range [0, 1]), signifies fraction data pack right dataset. integer, signifies number samples pack right dataset. NULL, defaults complement left_size. Defaults NULL. shuffle Boolean, whether shuffle data splitting . seed random seed shuffling.","code":""},{"path":"https://keras3.posit.co/dev/reference/split_dataset.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Splits a dataset into a left half and a right half (e.g. train / test). — split_dataset","text":"list two tf$data$Dataset objects: left right splits.","code":""},{"path":"https://keras3.posit.co/dev/reference/split_dataset.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Splits a dataset into a left half and a right half (e.g. train / test). — split_dataset","text":"","code":"data <- random_uniform(c(1000, 4)) c(left_ds, right_ds) %<-% split_dataset(list(data$numpy()), left_size = 0.8) left_ds$cardinality() ## tf.Tensor(800, shape=(), dtype=int64) right_ds$cardinality() ## tf.Tensor(200, shape=(), dtype=int64)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/summary.keras.src.models.model.Model.html","id":null,"dir":"Reference","previous_headings":"","what":"Print a summary of a Keras Model — summary.keras.src.models.model.Model","title":"Print a summary of a Keras Model — summary.keras.src.models.model.Model","text":"Print summary Keras Model","code":""},{"path":"https://keras3.posit.co/dev/reference/summary.keras.src.models.model.Model.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Print a summary of a Keras Model — summary.keras.src.models.model.Model","text":"","code":"# S3 method for class 'keras.src.models.model.Model' summary(object, ...)  # S3 method for class 'keras.src.models.model.Model' format(   x,   line_length = min(getOption(\"width\"), 180),   positions = NULL,   expand_nested = FALSE,   show_trainable = NA,   ...,   layer_range = NULL,   compact = TRUE )  # S3 method for class 'keras.src.models.model.Model' print(x, ...)"},{"path":"https://keras3.posit.co/dev/reference/summary.keras.src.models.model.Model.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Print a summary of a Keras Model — summary.keras.src.models.model.Model","text":"object, x Keras model instance ... summary() print(), passed format(). format(), passed model$summary(). line_length Total length printed lines positions Relative absolute positions log elements line. provided, defaults c(0.33, 0.55, 0.67, 1.0). expand_nested Whether expand nested models. provided, defaults FALSE. show_trainable Whether show layer trainable. provided, defaults FALSE. layer_range list, tuple, vector 2 strings, starting layer name ending layer name (inclusive) indicating range layers printed summary. also accepts regex patterns instead exact name. case, start predicate first element matches layer_range[[1]] end predicate last element matches layer_range[[1]]. default NULL considers layers model. compact Whether remove white-space lines model summary. (Default TRUE)","code":""},{"path":"https://keras3.posit.co/dev/reference/summary.keras.src.models.model.Model.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Print a summary of a Keras Model — summary.keras.src.models.model.Model","text":"format() returns length 1 character vector. print() returns model object invisibly. summary() returns output format() invisibly printing .","code":""},{"path":"https://keras3.posit.co/dev/reference/summary.keras.src.models.model.Model.html","id":"enabling-color-output-in-knitr-rmarkdown-quarto-","dir":"Reference","previous_headings":"","what":"Enabling color output in Knitr (RMarkdown, Quarto)","title":"Print a summary of a Keras Model — summary.keras.src.models.model.Model","text":"order enable color output quarto rmarkdown document html output format (include revealjs presentations), need following setup chunk:","code":"```{r setup, include = FALSE}  options(cli.num_colors = 256)  fansi::set_knit_hooks(knitr::knit_hooks)  options(width = 75) # adjust as needed for format  ```"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/test_on_batch.html","id":null,"dir":"Reference","previous_headings":"","what":"Test the model on a single batch of samples. — test_on_batch","title":"Test the model on a single batch of samples. — test_on_batch","text":"Test model single batch samples.","code":""},{"path":"https://keras3.posit.co/dev/reference/test_on_batch.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Test the model on a single batch of samples. — test_on_batch","text":"","code":"test_on_batch(object, x, y = NULL, sample_weight = NULL, ...)"},{"path":"https://keras3.posit.co/dev/reference/test_on_batch.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Test the model on a single batch of samples. — test_on_batch","text":"object Keras model object x Input data. Must array-like. y Target data. Must array-like. sample_weight Optional array length x, containing weights apply model's loss sample. case temporal data, can pass 2D array shape (samples, sequence_length), apply different weight every timestep every sample. ... forward/backward compatability","code":""},{"path":"https://keras3.posit.co/dev/reference/test_on_batch.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Test the model on a single batch of samples. — test_on_batch","text":"scalar loss value (metrics), named list loss metric values (metrics).","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/text_dataset_from_directory.html","id":null,"dir":"Reference","previous_headings":"","what":"Generates a tf.data.Dataset from text files in a directory. — text_dataset_from_directory","title":"Generates a tf.data.Dataset from text files in a directory. — text_dataset_from_directory","text":"directory structure :   calling text_dataset_from_directory(main_directory, labels='inferred') return tf.data.Dataset yields batches texts subdirectories class_a class_b, together labels 0 1 (0 corresponding class_a 1 corresponding class_b). .txt files supported time.","code":"main_directory/ ...class_a/ ......a_text_1.txt ......a_text_2.txt ...class_b/ ......b_text_1.txt ......b_text_2.txt"},{"path":"https://keras3.posit.co/dev/reference/text_dataset_from_directory.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Generates a tf.data.Dataset from text files in a directory. — text_dataset_from_directory","text":"","code":"text_dataset_from_directory(   directory,   labels = \"inferred\",   label_mode = \"int\",   class_names = NULL,   batch_size = 32L,   max_length = NULL,   shuffle = TRUE,   seed = NULL,   validation_split = NULL,   subset = NULL,   follow_links = FALSE,   verbose = TRUE )"},{"path":"https://keras3.posit.co/dev/reference/text_dataset_from_directory.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Generates a tf.data.Dataset from text files in a directory. — text_dataset_from_directory","text":"directory Directory data located. labels \"inferred\", contain subdirectories, containing text files class. Otherwise, directory structure ignored. labels Either \"inferred\" (labels generated directory structure), NULL (labels), list/tuple integer labels size number text files found directory. Labels sorted according alphanumeric order text file paths (obtained via os.walk(directory) Python). label_mode String describing encoding labels. Options : \"int\": means labels encoded integers (e.g. sparse_categorical_crossentropy loss). \"categorical\" means labels encoded categorical vector (e.g. categorical_crossentropy loss). \"binary\" means labels (can 2) encoded float32 scalars values 0 1 (e.g. binary_crossentropy). NULL (labels). class_names valid \"labels\" \"inferred\". explicit list class names (must match names subdirectories). Used control order classes (otherwise alphanumerical order used). batch_size Size batches data. NULL, data batched (dataset yield individual samples). Defaults 32. max_length Maximum size text string. Texts longer truncated max_length. shuffle Whether shuffle data. set FALSE, sorts data alphanumeric order. Defaults TRUE. seed Optional random seed shuffling transformations. validation_split Optional float 0 1, fraction data reserve validation. subset Subset data return. One \"training\", \"validation\" \"\". used validation_split set. subset=\"\", utility returns tuple two datasets (training validation datasets respectively). follow_links Whether visits subdirectories pointed symlinks. Defaults FALSE. verbose Whether display number information classes number files found. Defaults TRUE.","code":""},{"path":"https://keras3.posit.co/dev/reference/text_dataset_from_directory.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Generates a tf.data.Dataset from text files in a directory. — text_dataset_from_directory","text":"tf.data.Dataset object. label_mode NULL, yields string tensors shape (batch_size,), containing contents batch text files. Otherwise, yields tuple (texts, labels), texts shape (batch_size,) labels follows format described . Rules regarding labels format: label_mode int, labels int32 tensor shape (batch_size,). label_mode binary, labels float32 tensor 1s 0s shape (batch_size, 1). label_mode categorical, labels float32 tensor shape (batch_size, num_classes), representing one-hot encoding class index.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/time_distributed.html","id":null,"dir":"Reference","previous_headings":"","what":"layer_time_distributed — time_distributed","title":"layer_time_distributed — time_distributed","text":"time_distributed() alias layer_time_distributed(). See ?layer_time_distributed() full documentation.","code":""},{"path":"https://keras3.posit.co/dev/reference/time_distributed.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"layer_time_distributed — time_distributed","text":"","code":"time_distributed(object, layer, ...)"},{"path":"https://keras3.posit.co/dev/reference/time_distributed.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"layer_time_distributed — time_distributed","text":"object Object compose layer . tensor, array, sequential model. layer Layer instance. ... forward/backward compatability.","code":""},{"path":"https://keras3.posit.co/dev/reference/time_distributed.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"layer_time_distributed — time_distributed","text":"return value depends value provided first argument.  object : keras_model_sequential(), layer added sequential model (modified place). enable piping, sequential model also returned, invisibly. keras_input(), output tensor calling layer(input) returned. NULL missing, Layer instance returned.","code":""},{"path":"https://keras3.posit.co/dev/reference/timeseries_dataset_from_array.html","id":null,"dir":"Reference","previous_headings":"","what":"Creates a dataset of sliding windows over a timeseries provided as array. — timeseries_dataset_from_array","title":"Creates a dataset of sliding windows over a timeseries provided as array. — timeseries_dataset_from_array","text":"function takes sequence data-points gathered equal intervals, along time series parameters length sequences/windows, spacing two sequence/windows, etc., produce batches timeseries inputs targets.","code":""},{"path":"https://keras3.posit.co/dev/reference/timeseries_dataset_from_array.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Creates a dataset of sliding windows over a timeseries provided as array. — timeseries_dataset_from_array","text":"","code":"timeseries_dataset_from_array(   data,   targets,   sequence_length,   sequence_stride = 1L,   sampling_rate = 1L,   batch_size = 128L,   shuffle = FALSE,   seed = NULL,   start_index = NULL,   end_index = NULL )"},{"path":"https://keras3.posit.co/dev/reference/timeseries_dataset_from_array.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Creates a dataset of sliding windows over a timeseries provided as array. — timeseries_dataset_from_array","text":"data array eager tensor containing consecutive data points (timesteps). first dimension expected time dimension. targets Targets corresponding timesteps data. targets[] target corresponding window starts index (see example 2 ). Pass NULL target data (case dataset yield input data). sequence_length Length output sequences (number timesteps). sequence_stride Period successive output sequences. stride s, output samples start index data[], data[+ s], data[+ 2 * s], etc. sampling_rate Period successive individual timesteps within sequences. rate r, timesteps data[], data[+ r], ... data[+ sequence_length] used creating sample sequence. batch_size Number timeseries samples batch (except maybe last one). NULL, data batched (dataset yield individual samples). shuffle Whether shuffle output samples, instead draw chronological order. seed Optional int; random seed shuffling. start_index Optional int; data points earlier (exclusive) start_index used output sequences. useful reserve part data test validation. end_index Optional int; data points later (exclusive) end_index used output sequences. useful reserve part data test validation.","code":""},{"path":"https://keras3.posit.co/dev/reference/timeseries_dataset_from_array.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Creates a dataset of sliding windows over a timeseries provided as array. — timeseries_dataset_from_array","text":"tf$data$Dataset instance. targets passed, dataset yields list (batch_of_sequences, batch_of_targets). , dataset yields batch_of_sequences. Example 1: Consider indices [0, 1, ... 98]. sequence_length=10,  sampling_rate=2, sequence_stride=3, shuffle=FALSE, dataset yield batches sequences composed following indices:   case last 2 data points discarded since full sequence can generated include (next sequence started index 81, thus last step gone 98). Example 2: Temporal regression. Consider array data scalar values, shape (steps,). generate dataset uses past 10 timesteps predict next timestep, use:     Example 3: Temporal regression many--many architectures. Consider two arrays scalar values X Y, shape (100,). resulting dataset consist samples 20 timestamps . samples overlap. generate dataset uses current timestamp predict corresponding target timestep, use:","code":"First sequence:  [0  2  4  6  8 10 12 14 16 18] Second sequence: [3  5  7  9 11 13 15 17 19 21] Third sequence:  [6  8 10 12 14 16 18 20 22 24] ... Last sequence:   [78 80 82 84 86 88 90 92 94 96] data <- op_array(1:20) input_data <- data[1:10] targets <- data[11:20] dataset <- timeseries_dataset_from_array(   input_data, targets, sequence_length=10) iter <- reticulate::as_iterator(dataset) reticulate::iter_next(iter) ## [[1]] ## tf.Tensor([[ 1  2  3  4  5  6  7  8  9 10]], shape=(1, 10), dtype=int32) ## ## [[2]] ## tf.Tensor([11], shape=(1), dtype=int32) X <- op_array(1:100) Y <- X*2  sample_length <- 20 input_dataset <- timeseries_dataset_from_array(     X, NULL, sequence_length=sample_length, sequence_stride=sample_length) target_dataset <- timeseries_dataset_from_array(     Y, NULL, sequence_length=sample_length, sequence_stride=sample_length)   inputs <- reticulate::as_iterator(input_dataset) %>% reticulate::iter_next() targets <- reticulate::as_iterator(target_dataset) %>% reticulate::iter_next()"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/to_categorical.html","id":null,"dir":"Reference","previous_headings":"","what":"Converts a class vector (integers) to binary class matrix. — to_categorical","title":"Converts a class vector (integers) to binary class matrix. — to_categorical","text":"E.g. use loss_categorical_crossentropy().","code":""},{"path":"https://keras3.posit.co/dev/reference/to_categorical.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Converts a class vector (integers) to binary class matrix. — to_categorical","text":"","code":"to_categorical(x, num_classes = NULL)"},{"path":"https://keras3.posit.co/dev/reference/to_categorical.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Converts a class vector (integers) to binary class matrix. — to_categorical","text":"x Array-like class values converted matrix (integers 0 num_classes - 1). R factors coerced integer offset 0-based, .e., .integer(x) - 1L. num_classes Total number classes. NULL, inferred max(x) + 1. Defaults NULL.","code":""},{"path":"https://keras3.posit.co/dev/reference/to_categorical.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Converts a class vector (integers) to binary class matrix. — to_categorical","text":"binary matrix representation input R array. class axis placed last.","code":""},{"path":"https://keras3.posit.co/dev/reference/to_categorical.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Converts a class vector (integers) to binary class matrix. — to_categorical","text":"","code":"a <- to_categorical(c(0, 1, 2, 3), num_classes=4) print(a) ##      [,1] [,2] [,3] [,4] ## [1,]    1    0    0    0 ## [2,]    0    1    0    0 ## [3,]    0    0    1    0 ## [4,]    0    0    0    1 b <- array(c(.9, .04, .03, .03,               .3, .45, .15, .13,               .04, .01, .94, .05,               .12, .21, .5, .17),               dim = c(4, 4)) loss <- op_categorical_crossentropy(a, b) loss ## tf.Tensor([0.41284522 0.45601739 0.54430155 0.80437282], shape=(4), dtype=float64) loss <- op_categorical_crossentropy(a, a) loss ## tf.Tensor([1.00000005e-07 1.00000005e-07 1.00000005e-07 1.00000005e-07], shape=(4), dtype=float64)"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/train_on_batch.html","id":null,"dir":"Reference","previous_headings":"","what":"Runs a single gradient update on a single batch of data. — train_on_batch","title":"Runs a single gradient update on a single batch of data. — train_on_batch","text":"Runs single gradient update single batch data.","code":""},{"path":"https://keras3.posit.co/dev/reference/train_on_batch.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Runs a single gradient update on a single batch of data. — train_on_batch","text":"","code":"train_on_batch(object, x, y = NULL, sample_weight = NULL, class_weight = NULL)"},{"path":"https://keras3.posit.co/dev/reference/train_on_batch.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Runs a single gradient update on a single batch of data. — train_on_batch","text":"object Keras model object x Input data. Must array-like. y Target data. Must array-like. sample_weight Optional array length x, containing weights apply model's loss sample. case temporal data, can pass 2D array shape (samples, sequence_length), apply different weight every timestep every sample. class_weight Optional named list mapping class indices (integers, 0-based) weight (float) apply model's loss samples class training. can useful tell model \"pay attention\" samples -represented class. class_weight specified targets rank 2 greater, either y must one-hot encoded, explicit final dimension 1 must included sparse class labels.","code":""},{"path":"https://keras3.posit.co/dev/reference/train_on_batch.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Runs a single gradient update on a single batch of data. — train_on_batch","text":"scalar loss value (metrics), named list loss metric values (metrics). property model$metrics_names give display labels scalar outputs.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/use_backend.html","id":null,"dir":"Reference","previous_headings":"","what":"Configure a Keras backend — use_backend","title":"Configure a Keras backend — use_backend","text":"Configure Keras backend","code":""},{"path":"https://keras3.posit.co/dev/reference/use_backend.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Configure a Keras backend — use_backend","text":"","code":"use_backend(backend, gpu = NA)"},{"path":"https://keras3.posit.co/dev/reference/use_backend.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Configure a Keras backend — use_backend","text":"backend string, can \"tensorflow\", \"jax\", \"numpy\", \"torch\". gpu bool, whether use GPU. NA (default), attempt detect GPU availability Linux. M-series Macs, defaults FALSE TensorFlow TRUE JAX. Windows, defaults FALSE.","code":""},{"path":"https://keras3.posit.co/dev/reference/use_backend.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Configure a Keras backend — use_backend","text":"Called primarily side effects. Returns provided backend, invisibly.","code":""},{"path":"https://keras3.posit.co/dev/reference/use_backend.html","id":"details","dir":"Reference","previous_headings":"","what":"Details","title":"Configure a Keras backend — use_backend","text":"functions allow configuring backend keras use. Note one backend can configured time. function called library(keras3) calling functions within package (see example). experimental support changing backend keras initialized. using config_set_backend().","code":"library(keras3) use_backend(\"tensorflow\")"},{"path":"https://keras3.posit.co/dev/reference/with_custom_object_scope.html","id":null,"dir":"Reference","previous_headings":"","what":"Provide a scope with mappings of names to custom objects — with_custom_object_scope","title":"Provide a scope with mappings of names to custom objects — with_custom_object_scope","text":"Provide scope mappings names custom objects","code":""},{"path":"https://keras3.posit.co/dev/reference/with_custom_object_scope.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Provide a scope with mappings of names to custom objects — with_custom_object_scope","text":"","code":"with_custom_object_scope(objects, expr)"},{"path":"https://keras3.posit.co/dev/reference/with_custom_object_scope.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Provide a scope with mappings of names to custom objects — with_custom_object_scope","text":"objects Named list objects expr Expression evaluate","code":""},{"path":"https://keras3.posit.co/dev/reference/with_custom_object_scope.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Provide a scope with mappings of names to custom objects — with_custom_object_scope","text":"result evaluating expr within custom object scope.","code":""},{"path":"https://keras3.posit.co/dev/reference/with_custom_object_scope.html","id":"details","dir":"Reference","previous_headings":"","what":"Details","title":"Provide a scope with mappings of names to custom objects — with_custom_object_scope","text":"many elements Keras models can customized user objects (e.g. losses, metrics, regularizers, etc.). loading saved models use functions typically need explicitly map names user objects via custom_objects parameter. with_custom_object_scope() function provides alternative lets create named alias user object applies entire block code, automatically recognized loading saved models.","code":""},{"path":"https://keras3.posit.co/dev/reference/with_custom_object_scope.html","id":"examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Provide a scope with mappings of names to custom objects — with_custom_object_scope","text":"","code":"# define custom metric metric_top_3_categorical_accuracy <-   custom_metric(\"top_3_categorical_accuracy\", function(y_true, y_pred) {     metric_top_k_categorical_accuracy(y_true, y_pred, k = 3)   })  with_custom_object_scope(c(top_k_acc = sparse_top_k_cat_acc), {    # ...define model...    # compile model (refer to \"top_k_acc\" by name)   model |> compile(     loss = \"binary_crossentropy\",     optimizer = optimizer_nadam(),     metrics = c(\"top_k_acc\")   )    # save the model   model |> save_model(\"my_model.keras\")    # loading the model within the custom object scope doesn't   # require explicitly providing the custom_object   reloaded_model <- load_model(\"my_model.keras\") })"},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/zip_lists.html","id":null,"dir":"Reference","previous_headings":"","what":"Zip lists — zip_lists","title":"Zip lists — zip_lists","text":"conceptually similar zip() Python, R functions purrr::transpose() data.table::transpose() (albeit, accepting elements ... instead single list), one crucial difference: provided objects named, matching done names, positions.","code":""},{"path":"https://keras3.posit.co/dev/reference/zip_lists.html","id":"ref-usage","dir":"Reference","previous_headings":"","what":"Usage","title":"Zip lists — zip_lists","text":"","code":"zip_lists(...)"},{"path":"https://keras3.posit.co/dev/reference/zip_lists.html","id":"arguments","dir":"Reference","previous_headings":"","what":"Arguments","title":"Zip lists — zip_lists","text":"... R lists atomic vectors, optionally named.","code":""},{"path":"https://keras3.posit.co/dev/reference/zip_lists.html","id":"value","dir":"Reference","previous_headings":"","what":"Value","title":"Zip lists — zip_lists","text":"inverted list","code":""},{"path":"https://keras3.posit.co/dev/reference/zip_lists.html","id":"details","dir":"Reference","previous_headings":"","what":"Details","title":"Zip lists — zip_lists","text":"arguments supplied must length. positional matching required, arguments provided must unnamed. matching names, arguments must set names, can different orders.","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/reference/zip_lists.html","id":"ref-examples","dir":"Reference","previous_headings":"","what":"Examples","title":"Zip lists — zip_lists","text":"","code":"gradients <- list(\"grad_for_wt_1\", \"grad_for_wt_2\", \"grad_for_wt_3\") weights <- list(\"weight_1\", \"weight_2\", \"weight_3\") str(zip_lists(gradients, weights)) #> List of 3 #>  $ :List of 2 #>   ..$ : chr \"grad_for_wt_1\" #>   ..$ : chr \"weight_1\" #>  $ :List of 2 #>   ..$ : chr \"grad_for_wt_2\" #>   ..$ : chr \"weight_2\" #>  $ :List of 2 #>   ..$ : chr \"grad_for_wt_3\" #>   ..$ : chr \"weight_3\" str(zip_lists(gradient = gradients, weight = weights)) #> List of 3 #>  $ :List of 2 #>   ..$ gradient: chr \"grad_for_wt_1\" #>   ..$ weight  : chr \"weight_1\" #>  $ :List of 2 #>   ..$ gradient: chr \"grad_for_wt_2\" #>   ..$ weight  : chr \"weight_2\" #>  $ :List of 2 #>   ..$ gradient: chr \"grad_for_wt_3\" #>   ..$ weight  : chr \"weight_3\"  names(gradients) <- names(weights) <- paste0(\"layer_\", 1:3) str(zip_lists(gradients, weights[c(3, 1, 2)])) #> List of 3 #>  $ layer_1:List of 2 #>   ..$ : chr \"grad_for_wt_1\" #>   ..$ : chr \"weight_1\" #>  $ layer_2:List of 2 #>   ..$ : chr \"grad_for_wt_2\" #>   ..$ : chr \"weight_2\" #>  $ layer_3:List of 2 #>   ..$ : chr \"grad_for_wt_3\" #>   ..$ : chr \"weight_3\"  names(gradients) <- paste0(\"gradient_\", 1:3) try(zip_lists(gradients, weights)) # error, names don't match #> Error in zip_lists(gradients, weights) :  #>   All names of arguments provided to `zip_lists()` must match. Call `unname()` on each argument if you want positional matching # call unname directly for positional matching str(zip_lists(unname(gradients), unname(weights))) #> List of 3 #>  $ :List of 2 #>   ..$ : chr \"grad_for_wt_1\" #>   ..$ : chr \"weight_1\" #>  $ :List of 2 #>   ..$ : chr \"grad_for_wt_2\" #>   ..$ : chr \"weight_2\" #>  $ :List of 2 #>   ..$ : chr \"grad_for_wt_3\" #>   ..$ : chr \"weight_3\""},{"path":"https://keras3.posit.co/dev/news/index.html","id":"keras3-development-version","dir":"Changelog","previous_headings":"","what":"keras3 (development version)","title":"keras3 (development version)","text":"Added S3 methods JAX array: str, .array, .double, .integer, .numeric. Added str S3 method Keras Variables. layer_reshape() can now accept -1 sentinel automatically calculated axis size. Updated dependencies declared use_backend(\"jax\", gpu=TRUE) compatability keras-hub. Added training loop configuration helpers: config_max_epochs(), config_set_max_epochs(), config_max_steps_per_epoch(), config_set_max_steps_per_epoch(). caps can also set via KERAS_MAX_EPOCHS KERAS_MAX_STEPS_PER_EPOCH environment variables. Added config_is_nnx_enabled() check whether JAX NNX features enabled. LoRA-enabled layers (layer_dense(), layer_embedding(), layer_einsum_dense()) gain lora_alpha argument scale adaptation delta independently chosen rank. keras_variable() now accepts synchronization argument distributed strategies. Layer$add_weight() gains overwrite_with_gradient option layers now provide symbolic_call() method. Transposed convolution utilities now follow latest Keras API: op_conv_transpose() defaults strides = 1 layer_conv_*_transpose() layers expose output_padding precise shape control. layer_torch_module_wrapper() gains output_shape argument help Keras infer shapes wrapping PyTorch modules. save_model_weights() adds max_shard_size argument split large weight files manageable shards. Added elastic deformation utilities images: layer_random_elastic_transform() lower-level op_image_elastic_transform(). Added loss_categorical_generalized_cross_entropy() training noisy labels. Added Muon optimizer via optimizer_muon(). Added complex-valued helpers: S3 Arg() methods tensors, op_angle(), conversions op_view_as_real() / op_view_as_complex(). Added signal window operations: op_bartlett(), op_blackman(), op_hamming(), op_hanning(), op_kaiser(). Expanded numeric operations op_layer_normalization(), op_cbrt(), op_corrcoef(), op_deg2rad(), op_heaviside(), new op_sparse_sigmoid() plus matching activation_sparse_sigmoid(), attn_logits_soft_cap argument op_dot_product_attention(). layer_layer_normalization() removes rms_scaling argument.","code":""},{"path":"https://keras3.posit.co/dev/news/index.html","id":"keras3-140","dir":"Changelog","previous_headings":"","what":"keras3 1.4.0","title":"keras3 1.4.0","text":"CRAN release: 2025-05-04 New op_subset() x@r[...] methods enable tensor subsetting using R’s [ semantics idioms. New subset assignment methods implemented tensors: op_subset(x, ...) <- value x@r[...] <- value Breaking changes: operations prefixed op_ now return 1-based indices default. following functions return consume indices changed: op_argmax(), op_argmin(), op_top_k(), op_argpartition(), op_searchsorted(), op_argsort(), op_digitize(), op_nonzero(), op_split(), op_trace(), op_swapaxes(), op_ctc_decode(), op_ctc_loss(), op_one_hot(), op_arange() op_arange() now matches semantics base::seq(). default starts, includes end value, automatically infers step direction. op_one_hot() now infers num_classes supplied factor. op_hstack() op_vstack() now accept arguments passed via .... application_decode_predictions() now returns processed data frame default decoder function predictions missing. application_preprocess_inputs() returns preprocessor function inputs missing. Various new examples added documentation, including op_scatter(), op_switch(), op_nonzero(). New x@py[...] accessor introduced Python-style 0-based indexing tensors. New Summary group generic method keras_shape, enabling usage like prod(shape(3, 4)) KERAS_HOME now set tools::R_user_dir(\"keras3\", \"cache\") ~/.keras exist KERAS_HOME unset. new op_convert_to_array() convert tensor R array. Added compatibility Keras v3.9.2. New operations added: op_rot90() op_rearrange() (Einops-style) op_signbit() op_polar() op_image_perspective_transform() op_image_gaussian_blur() New layers introduced: layer_rms_normalization() layer_aug_mix() layer_cut_mix() layer_random_invert() layer_random_erasing() layer_random_gaussian_blur() layer_random_perspective() layer_resizing() gains antialias argument. keras_input(), keras_model_sequential(), op_convert_to_tensor() gain ragged argument. layer$pop_layer() gains rebuild argument now returns removed layer. New rematerialized_call() method added Layer objects. Documentation improvements minor fixes. Fixed issue op_shape() sometimes return TensorFlow TensorShape Fixes metric_iou(), op_top_k(), op_eye() called R atomic doubles","code":""},{"path":"https://keras3.posit.co/dev/news/index.html","id":"keras3-130","dir":"Changelog","previous_headings":"","what":"keras3 1.3.0","title":"keras3 1.3.0","text":"CRAN release: 2025-03-03 Keras now uses reticulate::py_require() resolve Python dependencies. Calling install_keras() longer required (still supported). use_backend() gains gpu argument, specify GPU-capable set dependencies resolved py_require(). progress bar fit(), evaluate() predict() now defaults presenting testthat tests. dotty::. now reexported. %*% now dispatches op_matmul() tensorflow tensors, relaxed shape constraints compared tf$matmul(). Fixed issue calling Metric Loss object unnamed arguments error.","code":""},{"path":"https://keras3.posit.co/dev/news/index.html","id":"added-compatibility-with-keras-v380-user-facing-changes-1-3-0","dir":"Changelog","previous_headings":"","what":"Added compatibility with Keras v3.8.0. User-facing changes:","title":"keras3 1.3.0","text":"activation_sparse_plus() activation_sparsemax() activation_threshold() layer_equalization() layer_mix_up() layer_rand_augment() layer_random_color_degeneration() layer_random_color_jitter() layer_random_grayscale() layer_random_hue() layer_random_posterization() layer_random_saturation() layer_random_sharpness() layer_random_shear() op_diagflat() op_sparse_plus() op_sparsemax() op_threshold() op_unravel_index() Add argument axis tversky loss New: ONNX model export export_savedmodel() Doc improvements bug fixes. JAX specific changes: Add support JAX named scope TensorFlow specific changes: Make random_shuffle() XLA compilable","code":""},{"path":[]},{"path":[]},{"path":"https://keras3.posit.co/dev/news/index.html","id":"activations-1-3-0","dir":"Changelog","previous_headings":"Added compatibility with Keras v3.7.0. User-facing changes: > New functions","what":"Activations","title":"keras3 1.3.0","text":"activation_celu() activation_glu() activation_hard_shrink() activation_hard_tanh() activation_log_sigmoid() activation_soft_shrink() activation_squareplus() activation_tanh_shrink()","code":""},{"path":"https://keras3.posit.co/dev/news/index.html","id":"configuration-1-3-0","dir":"Changelog","previous_headings":"Added compatibility with Keras v3.7.0. User-facing changes: > New functions","what":"Configuration","title":"keras3 1.3.0","text":"config_disable_flash_attention() config_enable_flash_attention() config_is_flash_attention_enabled()","code":""},{"path":"https://keras3.posit.co/dev/news/index.html","id":"layers-and-initializers-1-3-0","dir":"Changelog","previous_headings":"Added compatibility with Keras v3.7.0. User-facing changes: > New functions","what":"Layers and Initializers","title":"keras3 1.3.0","text":"initializer_stft() layer_max_num_bounding_boxes() layer_stft_spectrogram()","code":""},{"path":"https://keras3.posit.co/dev/news/index.html","id":"losses-and-metrics-1-3-0","dir":"Changelog","previous_headings":"Added compatibility with Keras v3.7.0. User-facing changes: > New functions","what":"Losses and Metrics","title":"keras3 1.3.0","text":"loss_circle() metric_concordance_correlation() metric_pearson_correlation()","code":""},{"path":"https://keras3.posit.co/dev/news/index.html","id":"operations-1-3-0","dir":"Changelog","previous_headings":"Added compatibility with Keras v3.7.0. User-facing changes: > New functions","what":"Operations","title":"keras3 1.3.0","text":"op_celu() op_exp2() op_glu() op_hard_shrink() op_hard_tanh() op_ifft2() op_inner() op_soft_shrink() op_squareplus() op_tanh_shrink()","code":""},{"path":"https://keras3.posit.co/dev/news/index.html","id":"new-arguments-1-3-0","dir":"Changelog","previous_headings":"Added compatibility with Keras v3.7.0. User-facing changes: > New functions","what":"New arguments","title":"keras3 1.3.0","text":"callback_backup_and_restore(): Added double_checkpoint argument save fallback checkpoint callback_tensorboard(): Added support profile_batch argument layer_group_query_attention(): Added flash_attention seed arguments layer_multi_head_attention(): Added flash_attention argument metric_sparse_top_k_categorical_accuracy(): Added from_sorted_ids argument","code":""},{"path":"https://keras3.posit.co/dev/news/index.html","id":"performance-improvements-1-3-0","dir":"Changelog","previous_headings":"Added compatibility with Keras v3.7.0. User-facing changes:","what":"Performance improvements","title":"keras3 1.3.0","text":"Added native Flash Attention support GPU (via cuDNN) TPU (via Pallas kernel) JAX backend Added opt-native Flash Attention support GPU PyTorch backend Enabled additional kernel fusion via bias_add TensorFlow backend Added support Intel XPU devices PyTorch backend install_keras() changes: GPU available, default now install CPU build TensorFlow GPU build JAX. use GPU current session, call use_backend(\"jax\").","code":""},{"path":[]},{"path":"https://keras3.posit.co/dev/news/index.html","id":"breaking-changes-1-3-0","dir":"Changelog","previous_headings":"Added compatibility with Keras v3.6.0. User-facing changes:","what":"Breaking changes:","title":"keras3 1.3.0","text":"using get_file() extract = TRUE untar = TRUE, return value now path extracted directory, rather path archive.","code":""},{"path":"https://keras3.posit.co/dev/news/index.html","id":"other-changes-and-additions-1-3-0","dir":"Changelog","previous_headings":"Added compatibility with Keras v3.6.0. User-facing changes:","what":"Other changes and additions:","title":"keras3 1.3.0","text":"Logging now asynchronous fit(), evaluate(), predict(). enables 100% compact stacking train_step calls accelerators (e.g. running small models TPU). using custom callbacks rely on_batch_end, disable async logging. can re-enable adding self$async_safe <- TRUE callbacks. Note TensorBoard callback considered async-safe default. Default callbacks like progress bar async-safe. New bitwise operations: op_bitwise_and() op_bitwise_invert() op_bitwise_left_shift() op_bitwise_not() op_bitwise_or() op_bitwise_right_shift() op_bitwise_xor() New math operations: op_logdet() op_trunc() op_histogram() New neural network operation: op_dot_product_attention() New image preprocessing layers: layer_auto_contrast() layer_solarization() New Model functions get_state_tree() set_state_tree(), retrieving model variables, including trainable, non-trainable, optimizer variables, metric variables. New layer_pipeline() composing sequence layers. class useful building preprocessing pipeline. Compared keras_model_sequential(), layer_pipeline() key differences: ’s Model, just plain layer. layers pipeline compatible tf.data, pipeline also remain tf.data compatible, regardless backend use. New argument: export_savedmodel(verbose = ) New argument: op_normalize(epsilon = ) Various documentation improvements bug fixes.","code":""},{"path":"https://keras3.posit.co/dev/news/index.html","id":"keras3-120","dir":"Changelog","previous_headings":"","what":"keras3 1.2.0","title":"keras3 1.2.0","text":"CRAN release: 2024-09-05 Added compatibility Keras v3.5.0. User facing changes: op_associative_scan() op_searchsorted() optimizer_lamb() keras$DTypePolicy instances can now supplied dtype argument losses, metrics, layers. Add integration Hugging Face Hub. can now save models Hugging Face Hub directly save_model() load .keras models directly Hugging Face Hub load_model(). Added compatibility NumPy 2.0. Improved keras$distribution API support large models. Bug fixes performance improvements. Add data_format argument layer_zero_padding_1d() layer. Miscellaneous documentation improvements. Bug fixes performance improvements.","code":""},{"path":"https://keras3.posit.co/dev/news/index.html","id":"keras3-110","dir":"Changelog","previous_headings":"","what":"keras3 1.1.0","title":"keras3 1.1.0","text":"CRAN release: 2024-07-17 Fixed issue GPUs found running Windows WSL Linux. (reported #1456, fixed #1459) keras_shape objects (returned keras3::shape()) gain == != methods. Fixed warning tfruns::training_run() unable log optimizer learning rate. Added compatibility Keras v3.4.1 (R user facing changes). Added compatibility Keras v3.4.0. User facing changes: op_argpartition() op_map() op_scan() op_switch() op_dtype() op_lstsq() op_image_hsv_to_rgb() op_image_rgb_to_hsv() Added support arbitrary, deeply nested input/output structures Functional models (e.g. lists lists lists inputs outputs…) keras_input() gains optional argument. keras_model_sequential() gains input_optional argument. Add support float8 inference Dense EinsumDense layers. Enable layer_feature_space() used tfdatasets pipeline even backend isn’t TensorFlow. layer_string_lookup() can now take tf$SparseTensor() input. layer_string_lookup() returns \"int64\" dtype default modes now. Layer() instances gain attributes path quantization_mode. Metric()$variables now recursive. Add training argument Model$compute_loss(). split_dataset() now supports nested structures dataset. applications gain name argument, accept custom name. layer_multi_head_attention() gains seed argument. losses gain dtype argument. loss_dice() gains axis argument. op_ctc_decode(), new default mask_index = 0 op_image_* functions now use default data_format value config_image_data_format() op_isclose() gains arguments rtol, atol, equal_nan. save_model() gains argument zipped. Bugs fixes performance improvements.","code":""},{"path":"https://keras3.posit.co/dev/news/index.html","id":"keras3-100","dir":"Changelog","previous_headings":"","what":"keras3 1.0.0","title":"keras3 1.0.0","text":"CRAN release: 2024-05-21 Chains layer_* calls |> now instantiate layers order %>% pipe chains: left-hand-side first (#1440). iterate(), iter_next() as_iterator() now reexported reticulate. User facing changes upstream Keras v3.3.3: new functions: op_slogdet(), op_psnr() clone_model() gains new args: call_function, recursive Updated example usage. op_ctc_decode() strategy argument new default: \"greedy\". Updated docs. loss_ctc() default name fixed, changed \"ctc\" User facing changes upstream Keras v3.3.2: new function: op_ctc_decode() new function: op_eigh() new function: op_select() new function: op_vectorize() new function: op_image_rgb_to_grayscale() new function: loss_tversky() new args: layer_resizing(pad_to_aspect_ratio, fill_mode, fill_value) new arg: layer_embedding(weights) providing initial weights matrix new args: op_nan_to_num(nan, posinf, neginf) new args: op_image_resize(crop_to_aspect_ratio, pad_to_aspect_ratio, fill_mode, fill_value) new args: op_argmax(keepdims) op_argmin(keepdims) new arg: clear_session(free_memory) clearing without invoking garbage collector. metric_kl_divergence() loss_kl_divergence() clip inputs (y_true y_pred) [0, 1] range. new Layer() attributes: metrics, dtype_policy Added initial support float8 training layer_conv_*d() layers now support LoRa op_digitize() now supports sparse tensors. Models layers now return owned metrics recursively. Add pickling support Keras models. (e.g., via reticulate::py_save_object()) Note pickling recommended, prefer using Keras saving APIs.","code":""},{"path":"https://keras3.posit.co/dev/news/index.html","id":"keras3-020","dir":"Changelog","previous_headings":"","what":"keras3 0.2.0","title":"keras3 0.2.0","text":"CRAN release: 2024-04-18 New functions: quantize_weights(): quantize model layer weights -place. Currently, Dense, EinsumDense, Embedding layers supported (enough cover majority transformers today) layer_mel_spectrogram() layer_flax_module_wrapper() layer_jax_model_wrapper() loss_dice() random_beta() random_binomial() config_set_backend(): change backend Keras initialized. config_dtype_policy() config_set_dtype_policy() New Ops op_custom_gradient() op_batch_normalization() op_image_crop() op_divide_no_nan() op_normalize() op_correlate() ` New family linear algebra ops op_cholesky() op_det() op_eig() op_inv() op_lu_factor() op_norm() op_erfinv() op_solve_triangular() op_svd() audio_dataset_from_directory(), image_dataset_from_directory() text_dataset_from_directory() gain verbose argument (default TRUE) image_dataset_from_directory() gains pad_to_aspect_ratio argument (default FALSE) to_categorical(), op_one_hot(), fit() can now accept R factors, offset 0-based (reported #1055). op_convert_to_numpy() now returns unconverted NumPy arrays. op_array() op_convert_to_tensor() longer error casting R doubles integer types. export_savedmodel() now works Jax backend. Metric()$add_variable() method gains arg: aggregration. Layer()$add_weight() method gains args: autocast, regularizer, aggregation. op_bincount(), op_multi_hot(), op_one_hot(), layer_category_encoding() now support sparse tensors. op_custom_gradient() now supports PyTorch backend layer_lstm() layer_gru() gain arg use_cudnn, default 'auto'. Fixed issue application_preprocess_inputs() error supplied R array input. Doc improvements.","code":""},{"path":"https://keras3.posit.co/dev/news/index.html","id":"keras3-010","dir":"Changelog","previous_headings":"","what":"keras3 0.1.0","title":"keras3 0.1.0","text":"CRAN release: 2024-02-17 package rebuilt Keras 3.0. Refer https://blogs.rstudio.com/ai/posts/2024-05-21-keras3/ overview https://keras3.posit.co current --date documentation.","code":""}]
diff --git a/man/layer_tfsm.Rd b/man/layer_tfsm.Rd
index 2abe0b1e2..caef73e34 100644
--- a/man/layer_tfsm.Rd
+++ b/man/layer_tfsm.Rd
@@ -59,8 +59,8 @@ model |> export_savedmodel("path/to/artifact")
 ## Output Type:
 ##   TensorSpec(shape=(None, 10), dtype=tf.float32, name=None)
 ## Captures:
-##   140648635143312: TensorSpec(shape=(), dtype=tf.resource, name=None)
-##   140648304538064: TensorSpec(shape=(), dtype=tf.resource, name=None)
+##   133630129211152: TensorSpec(shape=(), dtype=tf.resource, name=None)
+##   133630129217872: TensorSpec(shape=(), dtype=tf.resource, name=None)
 
 }\if{html}{\out{</div>}}
 
diff --git a/man/op_rms_normalization.Rd b/man/op_rms_normalization.Rd
index f2e15a9b5..b351ea258 100644
--- a/man/op_rms_normalization.Rd
+++ b/man/op_rms_normalization.Rd
@@ -31,31 +31,8 @@ It is defined as \code{rms_normalization(x) = x * rsqrt(mean(square(x))) * scale
 \section{Examples}{
 \if{html}{\out{<div class="sourceCode r">}}\preformatted{x <- random_normal(c(1, 10))
 op_rms_normalization(x)
+op_rms_normalization(x, scale = 10)
 }\if{html}{\out{</div>}}
-
-\if{html}{\out{<div class="sourceCode">}}\preformatted{## tf.Tensor(
-## [[-1.1701705   0.9987454   0.27809578 -0.18452731  1.5731977   1.8934134
-##    0.10656697  1.0256069   0.6185074  -0.12678868]], shape=(1, 10), dtype=float32)
-
-}\if{html}{\out{</div>}}
-
-\if{html}{\out{<div class="sourceCode r">}}\preformatted{op_rms_normalization(x, scale = 10)
-}\if{html}{\out{</div>}}
-
-\if{html}{\out{<div class="sourceCode">}}\preformatted{## tf.Tensor(
-## [[-11.701706    9.987453    2.7809577  -1.845273   15.731977   18.934134
-##     1.0656698  10.256069    6.185074   -1.2678868]], shape=(1, 10), dtype=float32)
-
-}\if{html}{\out{</div>}}
-
-[-1.1701705   0.9987454   0.27809578 -0.18452731  1.5731977   1.8934134
-\subsection{0.10656697  1.0256069   0.6185074  -0.12678868]: R:-1.1701705\%20\%20\%200.9987454\%20\%20\%200.27809578\%20-0.18452731\%20\%201.5731977\%20\%20\%201.8934134\%0A##\%20\%20\%20\%200.10656697\%20\%201.0256069\%20\%20\%200.6185074\%20\%20-0.12678868}{
-
-[-11.701706    9.987453    2.7809577  -1.845273   15.731977   18.934134
-}
-
-\subsection{1.0656698  10.256069    6.185074   -1.2678868]: R:-11.701706\%20\%20\%20\%209.987453\%20\%20\%20\%202.7809577\%20\%20-1.845273\%20\%20\%2015.731977\%20\%20\%2018.934134\%0A##\%20\%20\%20\%20\%201.0656698\%20\%2010.256069\%20\%20\%20\%206.185074\%20\%20\%20-1.2678868}{
-}
 }
 
 \seealso{
diff --git a/tools/make-man.R b/tools/make-man.R
index 0e315eb20..693cd1fb8 100755
--- a/tools/make-man.R
+++ b/tools/make-man.R
@@ -1,11 +1,7 @@
 #!/usr/bin/env Rscript
-writeLines(format(Sys.getpid()), "pidfile")
-Sys.setenv(
-  "KERAS_BACKEND" = "tensorflow-cpu"
-  # "KERAS_BACKEND_CONFIGURED" = "yes"
-)
-# options(warn = 2, error = browser)
-# keras3:::op_blackman(5)
+
+Sys.setenv("KERAS_BACKEND" = "tensorflow-cpu")
+
 devtools::document()
 
 # reticulate::py_require(c(
@@ -20,7 +16,6 @@ envir::import_from(purrr, walk)
 str_prefix <- function(x, prefix, ...) str_c(prefix, x, ...)
 
 
-
 # rd_file <- "man/activation_elu.Rd"
 itemize_family <- function(rd_file) {
   x <- rd_file |> readLines() |> trimws(which = "right")